NJI©QDERZNKU

ELEMENTS OF MATHEMATICS

Lie Groups and Lie Algebras

Chapters 1-3

$2
Springer-Vedag
 NICOLAS BOURBAKI

ELEMENTS OF MATHEMATICS

Lie Groups
and Lie Algebras
Chapters 1—3

Springer-Verlag
Berlin Heidelberg NewYork
London Paris Tokyo
 Originally published as
ELEMENTS DE MATHEMATIQUE,
GROUPES ET ALGEBRES DE LIE
© Bourhaki, 1971, 1972

Mathematics Subject Classification (1980): 17B05, 22E

Distribution lights worldwide:

Springer-Verlag Berlin Heidelberg NewYork London Paris Tokyo
ISBN 3-540-50218-1 Spfinger-Verlag Berlin Heidelberg NewYork
ISBN 0-387-50218-1 Springer-Verlag NewYork Berlin Heidelberg
2nd printing 1989

Library of Congress Cataloging-in»l’ublicalion Data

Bonrbaki. Niwlns. [Groupcs et algflxes de Lie. English] Lie youps and Lie algebras I Nicolas Bonrbaki.
p. arm-(Elements oi mathematics/ Nicolas Bourbaki) Translalion of: Groupes el algebres de Lie,
Bibliography: p. includes index.
ISBN 0387-5ml8w1 (U.S.)
l. Lie groups. 2‘ Lie algebras. 1, Title. II. Series: Bourbaki. Nicolas, Elements de mthémllique. English.
QA38736513 I983 512'.55—d: l9 83-31203
0 N. Bourbaki, 1989
All lights reserved
Printed in Great Brillin
Bookbinding: J. Schiffer. Gn'insladl
2141f3l40-543210
 TO THE READER

1. This series of volumes, a list of which is given on page: ix and 2:, takes up
mathematics at the L ' ' and gives r' proofs. In ‘ ' '_ ' , it
requires no particular knowledge ofmathematics on the readers’ part, but only
a certain familiarity with mathematical reasoning and a certain capacity for
abstract thought. Nevertheless, it is directed especially to those who have a
good knowledge of at least the content of the first year or two of a university
mathematics course.
2. The method of exposition we have chosen is axiomatic and abstract, and
normally proceeds from the general to the particular. This choice has been
dictated by the main purpose of the treatise, which is to provide a solid
foundation for the whole body of modern mathematics. For this it is indis-
pensable to become familiar with a rather large number of very general ideas
and principles. Moreover, the demands of proofimpose a rigorously fixed order
on the subject matter. It follows that the utility of certain considerations will
not be immediately apparent to the reader unlas he has already a fairly
extended knowledge of mathematics; otherwise he must have the patience to
suspend judgment until the occasion arises.
3. In order to mitigate this disadvantage we have frequently inserted examples
in the text which refer to facts the reader may already know but which have
not yet been discussed in the series. Such examples are always placed between
two asterisks: * . . .*. Most readers will undoubtedly find that these examples
will help them to understand the text, and will prefer not to leave them out,
even at a first reading. Their omission would of course have no disadvantage,
from a purely logical point of view.
4. This series is divided into volumes (here called “Books“). The first six
Books are numbered and, in general, every statement in the text assumes as
known only those results which have already been discussed in the preceding

v
 TOTEEREADER

volumes. This rule holds good within each Book, but for convenience of expo-
sition these Books are no longer arranged in a consecutive order. At the begin-
ning of each of these Books (or of these chapters), the reader will find a precise
indication of its logical relationship to the other Books and he will thus be
able to satisfy himself of the absence of any vicious circle.
5. The logical framework of each chapter consists of the definitions, the axionu,
and the theorem; of the chapter. These are the parts that have mainly to be
borne in mind for subsequent use. Las important results and those which can
easily be deduced from the theorems are labelled as “propositions”, “lemmas”, _
“corollaries”, “remarks”, etc. Those which may be omitted at a first reading
are printed in small type. A commentary on a particularly important theorem
appears occasionally under the name of “scholium”.
To avoid tedious repetitions it is sometimes convenient to introduce nota-
tions or abbreviations which are in force only within a certain chapter or a
certain section of a chapter (for example, in a chapter which is concerned only
with commutative rings, the word “ring” would always signify “commutative
ring”). Such conventions are always explicitly mentioned, generally at the
beginning of the chapter in which they occur.
6. Some passages in the text are designed to forewarn the reader against
serious errors. These passages are signposted in the margin with the sign
2 ("dangerous bend”) .

7. The Exercises are designed both to enable the reader to satisfy himself that
he has digested the text and to bring to his notice results which have no place
in the text but which are nonetheless of interest. The most difficult exercises
bear the sign 1}.
8. In general, we have adhered to the commonly accepted terminology,
except where there appeared to be good reasons for deviating from it.
9. We have made a particular effort always to use rigorously correct language,
without sacrificing simplicity. As far as possible we have drawn attention in the
text to alum: qf language, without which any mathematical text runs the risk of
pedantry, not to say unreadability.
10. Since in principle the text consists of the dogmatic exposition of a theory,
it contains in general no references to the literature. Bibliographical references
are gathered together in Historical Nam, usually at the end of each chapter.
These notes also contain indications, where appropriate, of the unsolved
problems of the theory.
The bibliography which follows each historical note contains in general
only those books and original memoirs which have been of the greatest impor-
tance in the evolution of the theory under discussion. It makes no sort of pre—

vi
 TO THE READER

Deuce to completeness; in particular, references which serve only to determine

questions of priority are almost always omitted.
As to the exercises, we have not thought it worthwhile in general to indicate
their origins, since they have been taken from many difl‘erent sources (original
papers, textbooks, collections of exercises).
l 1. References to a part of this series are given as follows:
a) If reference is made to theorems, axioms, or definitions presented in the Mme
section, they are quoted by their number.
b) If they occur in another reclaim qfthe same chapter, this section is also quoted in
the reference.
0) If they occur in another chapter in the Jame Book, the chapter and section are
quoted.
d) If they occur in another Bonk, this Book is first quoted by in title.
The Summaries 9/ Result: are quoted by the letter R; thus Set Theory, R
signifies “Sumerian qesult: qf the Theory qm".

vii
 CX)NTENTS
OF
THE ELEMENTS OF MATHEMATICS SERIES

I. THEORY or SE18
1. Description of formal mathematics. 2. Theory of sets. 3. Ordered sets;
cardinals; natural numbers. 4. Structures.

ll. ADGEBRA
l. Algebraic structures. 2. Linear algebra. 3. Tensor algebras, exterior
algebras, symmetric algebras. 4. Polynomials and rational fractions.
5. Fields. 6. Ordered groups and fields. 7. Modules over principal ideal
rings. 8. Semi-simple modules and rings. 9. Sesquilinear and quadratic
forms.

Ill. GENERAL rommcv

1. Topological structures. 2. Uniform structures. 3. Topological groups.
4-. Real numbers. 5. One-parameter groups. 6. Real number spaces,
affine and projective spaces. 7. The additive groups R". 8. Complex
numbers. 9. Use of real numbers in general topology. 10. Function
spaces.

IV. Functions or A REAL VARIABLE

1. Derivatives. 2. Primitives and integrals. 3. Elementary functions.
4. Differential equations. 5. Local study of functions. 6. Generalized
Taylor expansions. The Euler—Madaurin summation formula. 7. The gam-
mafunction. Dictionary.

V. TOPOLOGICAL vscron sums

l. Topological vector spaces over a valued field. 2. Convex sets and locally
convex spaces. 3. Spaces ofcontinuous linear mappings. 4. Duality in topo-
logical vector spaces. 5. Hilbert spaces: elementary theory. Dictionary.

VI. Imomnon

1. Convexity inequalities. 2. Riesz spaces. 3. Measures on locally com-

pact spaces. 4. Extension of a measure. L' spaces. 5. Integration of mea-
sures. 6. Vectorial integration. 7. Haar measure. 8. Convolution and
representation. 9. Integration on Hausdorfi‘ topological spaces.
 CONTENTS OF THE RLEMENI’S OF MATHEMATI‘S SERIES

Lm GROUPS AND LIE ALGEBRAS

l. Lie algebras. 2. Free Lie algebras. 3. Lie groups. 4, Coxeter groups
and Tils systems. 5. Groups generated by reflections. 6. Root systems.

COMMUTATIVE ALGEBRA
1. Flat modules. 2. Localization. 3. Graduations, filmtions, and topo-
logies. 4. Amociated prime ideals and primary decomposition. 5. Integers.
6. Valuations. 7. Divison.

SPECTRAL menu
1. Nomad algebras. 2. Locally compact groups.

DIFFERENTIABLE AND ANALYTIC MANIFOLDS

Summary of results.
 CONTENTS

To THEREADER ...............................................

Comm or um ELEMENIS or MAmuAncs Seams ...............

CHAPTER I. LIE Amnnm ......................................

Hmfl®mUIUImHU—I-I
§ 1. Definition of Lie algebras ......... .... ..
Algebras .........
wwvwwrwpr

Lie algebms .............. .

Commutative Lie algebras ..........................
Ideals ............................................
Derived series, lower central series . .
Upper central series ..........
. Extensions .........
. Semi-direct products . . ..
. Change of base ring ................................ .—

§ 2. Enveloping algebra of a Lie algebra ......................

. Definition of the enveloping algebra
mmymgu+mn~

i Enveloping algebra of a product, . . .

. Enveloping algebra of a subalgebra .....
. Enveloping algebra of the opposite algebra.
Symmetric algebra of a module ........
. Filtration of the enveloping algebra . .
The Poincaré-BirkholT-Witt Theorem .
. Extension of derivations ............ ..
. Extension of the base ring ...........................

§ 3. Representations .....
l. Representations ...................................
 CONTENTS

. Tensor product of representations ....................

esaweww
. Representations on homomorphism modules .
Examples .............................
Invariant elements . . .
Invariant bilinear forms .
Casimir element ......... .
Extension of the base ring ...........................
§4. Nilpotent Lie algebras ..................................
l. Definition of nilpotent Lie algebras
2. Engel’s Theorem .........................
3. The largest nilpotency ideal of a representation .
4. The largest nilpotent ideal in a Lie algebra . . . . ...
5. Extension of the base field ...........................
§ 5. Solvable Lie algebras ..................................
1. Definition of solvable Lie algebras .
2. Radical of a Lie algebra .........
3. Nilpotent radical of a Lie algebra .
4. A criterion for solvability .........
5. Further properties of the radical ...
6. Extension of the base field ...........................
§ 6. Semi-simple Lie algebras ................................
1. Definition of semi-simple Lie algebras
2. Semi-simplicity of representations ....................
3. Semi-simple elements and nilpotent elements in semi-
simple Lie algebras ...............................
4. Reductive Lie algebras ......
5. Application: a criterion for semi Simplicity of representa-
tions .............
. Subalgebras reductive m a Lie algebra . .
8 lbfluai

. Examples of semi-simple Lie algebras . . .

. The Levi—Maloev Theorem ..........
. The invariants theorem .
. Change of base field . .
§ 7. Ado’s Theorem ........
l. Coefficients of a represenmtion
2. The extension theorem ....... ..
3. Ado’s Theorem ....................................
Exercises for§ l ...........................................
Exercises for § 2 ...........................................
Exercises for § 3 ...........................................

xii
 CONTENTS

Exercisesfor§4 ..........
Exercisesfor§5 ...................................
Exercises for§6 ........................................... 102
Exercisesfor§7 ........................................... 109

CHAPTER 11. FREE Luz Animus ................................ 111

§ 1. Enveloping bigebra of a Lie algebra ..................... 111
1. Primitive elements of a. cogebra . .. 111
2. Primitive elements of a bigebra. 113
3. Filtered bigebras .................. 114
4-. Enveloping bigebra of a Lie algebra . . . 115
5. Structure of the oogebra U(g) in characteristic 0. 116
6. Structure of filtered bigebras 1n characteristic 0 . . . . . . 119
§2. Free Lie algebras ..................................... 122
1. Revision of flee algebras ...... . 122
2. Construction of the free Lie algebra . 122
3. Presentation of a Lie algebra. . . 124 '
4. Lie polynomials and substitutions 124
5. Functorial properties ......... 125
6. Graduations ........ 126
74 Lower central series ........ 128
8. Derivations of free Lie algebras. 129
9. Elimination theorem ....... 130
10.Hallsetsinafreemagma ..... . 132
11. Hall bases ofa free Lie algebra ...................... 134
§ . Enveloping algebra of the free Lie algebra .................
b:

1. Enveloping algebra of L(X) ..... .

2. Projector of A*(X) onto L(X) ......................
3. Dimension of the homogeneous components of L(X) . . . .
§4. Central filtrations .....................................
1. Real filtrations . ...
2. Order function. . . . .........................
3. Graded algebra assocrated with a filtered algebra . .
4. Central filtrations on a group ...................
5. An example of a central filtration.
6. Integral central filtrations .............
§ 5. Magnus algebras .....................................
1. Magnus algebras . . ...
2. Magnus group .......................
 CDNTEN“

3. Magnus group and free group ....... 151

4. Lower central scrim ofa. free group . . .. 152
5. p—filtration of free groups ............................ 154
§6. The Hausdorff series ................................... 155
l. Exponential and logarithm in filmed algebras. 155
2. Hausdorff group ........................ 157
3. Lie formal power series . .. 158
4. The Hausdorfi selis ............ .. 160
5. Substitutions in the Hausdorff series .................. 171 ,
§ 7. Convergence of the Hausdorfi‘ series (real or complar case) . . 164-
1. Continuous-polynomials with values in g ............... 164-
2. Group germ defined by a complete normed Lie algebra . 165
3. Exponential in complete normed asociative algebras . . . . 169
§8. Convergence of the Hausdorff series (ultrametric case) ...... 170
l. p—adic upper bounds of the series up, log and H. .. 171
2. Normcd Lie algebras ........................ .. 172
3. Group defined by a oomplene normed Lie algebra ....... 172
4. Exponential in complete normed associative algebras . . . . 174»

Appendix. Mbbius function................................ 176

Exercises for § 1 ........................................... 178
Exercises for § 2 .......................................... 181
Exercises for § 3 . ............. 186
Exercises for § 4 ............................. 187
Exercises for § 5 ........................................... 193
Exercises for § 6 ........................................... 201
Exercises for § 7 ........................................... 204
Exercises for § 8 .......................................... 205
Exercises for the Appendix ...... 207
CHAPTER III. Lu: GROUPS ...................................... 209
§ 1. Lie groups ............................................ 209
1. Definition of a Lie group . .. 209
2. Morphisms of Lie groups. . 213
3. Lie subgroups .................... .... 214
4. Semi-direct products of Lie groups ................... 215
 GONTEN'IS

5. Quotient of a manifold by a Lie group ................ 217

6. Homogeneous spaces and quotient groups. . . . 219
7. Orbits .......................... 222
8. Vector bundles with operators . 223
9. Local definition of a Lie group . 226
10. Group germs ................ . 228
11. Law chunks of operation .............. 231
§ 2. Group of tangent vectors to a Lie group. 233
1. Tangent laws of composition ............ 233
2. Group of tangent vectors to a Lie group . . 235
3. Case ofgroup germs .................. 237
§ 3. Passage from a Lie group to its Lie algebra ......... 238
l. Convolution of point distributions on a Lie group . . 238
2. Functorial properties ......... 241
3. Case of a group operating on a man old ....... . 244
4. Convolution of point distributions and functions ........ 24-5
5. Fields ofpoint distributions defined by the action ofa group
on a manifold ................................... 248
6. Invariant fields of point distributions on a Lie group . 249
7. Lie algebra of a Lie group . . .. 251
8. Functorial propertiu of the Lie algebra ............... 254
9. Lie algebra of the group of invertible elements of an alge-
bra ........................................... 257
10 Lie algebras of certain linear groups 258
11. Linear representations . . 259
12. Adjoint representation ...... 264
13. Tensors and invariant forms . 268
14. Maurer—Cartan formulae ..... 269
15. Construction of invariant difi‘erential forms . 271
16. Haar measure on a Lie group ............. 271
17. Left differential .............. 274
18. Lie algebra of a Lie group germ . . 276
§4. Passage from Lie algebras to Lie groups. 279
1 Passage from Lie algebra morphisms to Lie group mor-
phisms ........................................ 279
2. Passage from Lie algebra: to Lie groups . . .. 281
3. Exponential mappings ................. 234
4. Functon’ality of exponential mappings ....... 288
5. Structure induced on a sub-group ............... 289
6. Primitives ofdlfierenual forms with values m a Lie algebra 291
7 Passage from law: of infinitesimal operation to laws of
operation ....................................... 294
 CONTEN'IS

§5. Formal calculations in Lie groups .......................

l. The coefficients a,“ ....... ..
2. Bracket in the Lie algebra
3. Powers ................. ..
4. Exponential ......................................

§s. Real and complex Lie groups ............................

l. Passage from Lie algebra morphisms to Lie group mor-
phisms .................
2. Integral subgroups ......... 306'
3. Passage from Lie algebras to Lie groups . . 310
4-4 Exponential mapping .................. 311
5 . Application to linear representations . . . .. 215
6. Normal integral subgroups .......................... 316
7 . Primitives of differential forms with values in a Lie algebra 318
8. Passage from laws of infinitesimal operation to laws of
operation ....................................... 318
9. Exponential mapping in the linear group .............. 320
0. Complexification of a finite-dimensional real Lie group. . 322
_.

§7. Lie groups over an ultrametric field ...................... 326

. Passage from Lie algebras to Lie groups . . 327
+0910»-

. Exponential mappings ................ 328

. Standard groups .......... 328
. Filtration of standard groups . 330
. Powers in standard groups . . ..
010‘

. Logarithmic mapping ..............................

§8. Lie groups over R or Q, ................................

1. Continuous morphisms. .
24 Closed subgroups ........................

§9. Commutators, centralizers and normalizers in a Lie grou . . .

. Commutators in a topological group ..................
«flamenco»—

. Commutators in a Lie group .......

. Centralizers .............
Normalizers .........
. Nilpotent Lie groups. .
. Solvable Lie groups . . . .
. Radical of a Lie group . . ..
. Semi-simple Lie groups ............................ 355
§10. The automorphism group of a Lie group .................. 359
l. lnfinitesimal automorphisms ......................... 359
 CONTENTS

2. The automorphism group of a Lie group (real or complex

case ................................ . . . 362
3. The automorphism group ofa C group (ultrametric case) 367

Appendix. Operations on linear represenmtions ................ 368

Exercises for § 1 ........................................... 370
Exercises for § 3 ........................................... 372
Exercises for §4 ..................................... 376
Exercises for § 5 . . 382
Exercises for § 6 ......................................... 383
Exercises for § 7 ........................................... 39]
Exercise for § 8 ........................................... 395
Exercises for § 9 ....................................... .. . . 397
Exercises for§ 10 .......................................... 409

HISTORICAL N011: (Um I TO III) ........................... 410

Bmuoomnv ................................................. 431

INDEX or Non-non ............................................ 435

INDExorTERMmOLocY ............ ................ 4-39

Summary of certain properties of finitedimcnsional Lie algebras over a

field of characteristic 0 .......................................... 449

xvii
 CHAPTERI

Lie Algebras

In paragraphs 1, 2 and 3, K denote: a commutative ring with uni! element. In paragraph

4, K denote: a wmmutativefield. In paragraphs 5, 6 and 7, K denote: a cumulative
field qf tharactuirtic 0.1’

§l. DEFINITION OF LIE ALGEBRAS

l. ALGEBRAS
Let M be a unitary module over K with a bilinear mapping (x,y) >—>xy of
M x M into M. All the axioms for algebras are satisfied except associativity
of multiplication. By an abuse of language, M is called a not mterrarily asso-
ciative algebra over K, or sometimes, when no confusion can arise, an algebra
over K. In this no. we shall use the latter notation.
If the K-module M is given the multiplication (x,y) n an algebra is
obtained called the opposite of the above algebra.
A sub-K-module N of M which is stable under multiplication is given the
structure of an algebra over K in an obvious way. N is called a .mbalgebra of M.
N is called a left (resp. right) ideal of M if the conditions x e N, y e M imply
ya: 6 N (resp. zy e N). If N is both a left ideal and a right ideal of M, N is called a
two-sided ideal of M. In this case the multiplication on M enables us to define,
on passing to the quotient, a bilinear multiplication on the quotient module
M/N such that M/N has an algebra structure. M/N is called the quotient '
algebra of M by N.

1' The propositions proved in this chapter depend only on the properties estab-
lished in Books I to V1 and on certain results of Camrmualive Algzbva, Chapter III,
§ 2.
 I LIE ALGEBRAS

Let M1 and M2 be two algebras over K and (I; a mapping of M1 into M,.
4) is called a homomorphism ifo is K-linear and (Mary) = ¢(x)d>(y) for x 6 M,,
y 6 M1. The kernel N of (1) is a two-sided ideal of M1 and the image of d) is a
subalgebra of M2. On passing to the quotient, 4) defines an isomorphism of the
algebra MllN onto the algebra ¢(Ml).
Let M be an algebra over Ki A mapping D of M into M is called a dm'mztz'nn
of M if it is K-linear and D(xy) = (Dx)y + x(Dy) for all x e M and y E M.
This definition generalizes Definition 3 of Algebra, Chapter IV, §4, no. 3. The
kernel of a derivation of M is a subalgebra of M. If D1 and D2 are derivations
of M, then Dn — D2D1 is a derivation of M (of. Algzbnz, Chapter IV, §4v,'
no. 3, Proposition 5: the proof of this proposition does not use the associativity
of the algebra).
Let M1 and M, be two algebras over K. On the product K-module
M = M, x M, we define a multiplication by writing

(“1: *2) (1/1: ya) = (“syn 32112),

for all :41, y1 in M1, x2, y; in M3. The algebra thus defined is called the pruduct
algebra of M1 and M2. The mapping x, »—> (x), 0) (resp. x2 >—> (0, x,» is an
isomorphism of M1 (resp. M2) onto a. two-sided ideal of M. Under these iso-
morphisms MI and M2 are identified with two-sided ideals of M. The K-
module M is then the direct sum of M1 and MT Conversely, let M be an
algebra over K and M,, M, two two-sided ideals of M such that M is the direct
sum of M1 and M,. Then M,M3 C Ml n M2 = {0); then, ifzh _l/l belong to
M, and :3, y, to M2, then (x; + x¢)(y1 + ya) = x1y1 + x21”, so that M is
identified with the product algebra M1 x M,. Every left (resp. right, two-
sided) ideal of M1 is a. lefi (reap. right, two-sided) ideal of M. We leave to the
reader the task of formulating the analogous results in the case of an arbitrary
finite family of algebras.
Let M be an algebra over K and suppose that the K-module M admits a
basis (aflnh. There exists a unique system (YMV)0.,D.V)EL x L XL of demerits of K
such that aha“ = v nwav for all A, y. in L. The y,“ are mlled the constants bf
structure of M with respect to the basis (0,).
Let M be an algebra over K, K,3 a commutative ring with unit element and
p a homomorphism of K0 into K mapping unit element to unit element. Then
M can be considered as an algebra over K0 by writing out = p(ot) .x
for a: E K0, 1: e M. This is the case in particular when K0 is a subring of K
containing the unit element and p is taken to be the inclusion mapping of K0
into K.
Let M be an algebra over K, Kl a commutative ring with unit element and
a a homomorphism of K into K1 mapping unit element to unit element. Let
114““ u) = Mam be the Kl-module derived from M by extending the ring of

2
 LIE “023m § 1.2

scalars to K1 (Algebra, Chapter II, § 5). The product on M defines canonically

a Kl-bilinear mapping of Mom x Mm, into Mam (Algebra, Chapter IX, § 1,
no. 4) such that Mom is given the structure of an algebra over K, (which is
said to be derivedfiam M by extending the ring afrcalar: tn K1). This is the case in
particular when K is a subn'ng of K1 containing the unit element and a- the
inclusion mapping of K into K1.

2. LE ALGEBRAS
DEFINITION 1. An algebra 5 over K i: called a Lie algebra over K {fits multiplication
(denoted by (x, y) I—> [x, y]) satisfies the identities:

(1) PM] = 0
(2) [’8 [% zll + [1b [1, *1] + [1: [#41] = 0
for all x, y, z in g.
The product [x, y] is called the bracket of x and y. Identity (2) is called the
Jacobi identity.
The bracket [x, y] is an alternating bilinear function ofx and y. We have the
identity:

(3) [by] = -L% r]

so that the Jacobi identity can be written:

(4) [*3 [11: z]] = [[x, y], z] + U, [’6 ll]-

Every subalgebra and every quotient algebra of a Lie algebra is a Lie
algebra. Every product of Lie algebras is a Lie algebra. If g is a Lie algebra,
the opposite algebra 9° is a Lie algebra and the mapping x» —x an isomor-
phism of 9 onto 9°, by virtue of identity (3).
Exampl: 1. Let L be an associative algebra over K. The bracket [x, y] = xy —yx
is a bilinear function OD: and g. It is easily verified that the law of composition
(x, y) ~—> [55, g] on the K-module L makes L into a Lie algebra over K.

Examplz 2. In Example 1 choose L to be the associative algebra of endomor-

phisms of a K-module E. We obtain the Lie algebra «J mdmrphirmr bf E,
denoted by 91(E). (If E = K", the Lie algebra gI(E) is denoted by gl(n, K).)
Every Lie subalgebra of 9|(E) is a Lie algebra over K. In particular:
(1) If E is given a (not necessarily associative) algebra structure, the
derivations of E form a Lie algebra over K.
(2) If E admits a finite basis, the endomorphism of E of zero trace form a
Lie algebra over K denoted by 51(E) (or 91(n, K) if E = K").
(3) The set M,(K) of square matrices of order n can be considered as a Lie

3
I 113W

algebra over K canonically isomorphic to 31(11, K). Let (Eu) be the canonical
basis ofM,(K) (Algebra, Chapter II, § 10, no. 3). It follows easily that:
[EmEu]=O ifjaélc and i761
(5) [Em En] = Eu iff # I
[Em EM] = ‘51:; 1f! 9‘ k
[Em Eu] = Eu ‘ E11
The Lie subalgebra of M,(K) consisting of the triangular matrices (resp.
triangular matrices ofzero trace, mp. triangular matrices of zero diagonal) is
denoted by 1(n, K) (resp. “(11, K), reap. n(n, K)) (Algzbm, Chapter II, § 10,‘
no. 7).
*Example 3. Let V be an infinitely difl‘erentiable real manifold. The differen-
tial operators with infinitely differentiable real coefficients constitute an asso-
ciative algebra over R and hence, by Example 1, a Lie algebra A over R. The
bracket of two infinitely diflerentiable vector fields on V is an infinitely
difl'erentiable vector field and hence the infinitely difl'erentiable vector fields
on V constitute a Lie subalgebra f of A. IfV is a real Lie gimp, the left invariant
vector fields constitute a Lie subalgebra 9 of f called the Lie algebra of V. The
vector space 9 is identified with the tangent space to V at z (the identity element
of V). Let V’ be another real Lie group, e’ its identity element and 9' its Lie
algebra. Every analytic homomorphism of V into V’ defines a linear mapping
of the tangent space to V at 2 into the tangent space toV' at z’ ; this mapping is
a homomorphism of the Lie algebra 9 into the Lie algebra 9’. If V is the linear
group of a finite-dimensional real vector space E there exists a canonical iso-
morphism of 91(E) onto the Lie algebra g ofV, under which 9 is identified with
91(13):
DEmou 2. Let g b: a Lie algebra and x an element of 9. Th; linear "upping
y» [x,g] q into 9 isculledtheafioinllineaimappingqfx audit denotedby ad“): or
ad x.
PROPOSITION 1. Let 9 In a Lie algebra. For all x e 9, ad x it a derivation. 771: mapping
a: >—> ad at is a hmmorphirm ifthe Lie algebra 9 into the Lie algzbm b qfdm'ualions (J
g. IfD e b ands E 9, [D, ad x] = ad(Dx).
Identity (4) can be written:
(adXHy; 2] = [(34904) z] + U,(3dx)-21
or:
(adl-z = (adx).<(ady>.z) — (ady).<(ad~>.z)
whence the first two assertions. On the other hand, if D e b, x E 9, y e 9, then
[D, ad x] . y = D([x,y]) — [x, Dy] = [Dx,y] = (ad Dx) .y, whence the last
assertion.

The mapping ad I: is also called the inn! dermal!" "an defined by x.

4
 ooumrrA'nvz LIE mums § 1.3
3. COMMUTATIVE LIE ALGEBRAS
DEFINITION 3. Two element: x, 31 cf a Lie algebra are said to be permutable if
[x, y] = 0. g is said ta be enmmutatiue ifany two qfils elements are pmrmtable.
Example I. Let L be an associative algebra and g the Lie algebra defined by
it (no. 2, Example 1). Two elemenm x, y are permutable in 9 if and only if
xy = get in L.
'Example 2. Ifa real Lie group G is commutative, its Lie algebra is commuta-
five,
Every K—module can obviously be given a unique commutative Lie algebra
structure over K.
If g is a Lie algebra, every monogenous submodule of g is a commutative Lie
subalgebra of g.

44 IDEALS
It follows from identity (3) that in a Lie algebra 9 there is no distinction
between left ideals and right ideals, every ideal being two-sided. We therefore
speak simply of ideals.
‘Example. Let G be a Lie group, 5 its Lie algebra and H a Lie subgroup of G.
Every left invariant vector field on H defines canonically a left invariant
vector field on G, whence there is a canonical injection of the Lie algebra I; of
H into 9; b is identified with a Lie subalgebra of 9 under this injection. If H is
normal in G, the canonical image of b in g is an ideal of 9.,
An ideal of g is a submodule of g which is stable under the inner derivations
of 9.
DEFINITION 4. A submadule 9f 9 which is stable under every derivation of 5 i: called
a characteristic ideal cf 9.
PROPOSITION 2. Let g be a Lie algebra, a an ideal (reap. a characteristic ideal) of g
and b a characteristic ideal Lfa. 77121: b is an ideal (resp. a chaiaetm'slic ideal) ofg.
Every inner derivation (resp. every derivation) of g lava a stable and
induces on a a derivation and hence leaves b stable.
LetgbeaLiealgebra. Ifaandbareidealsofg, a + bandan bareideals
of 9.
Let n and b be two submodule: ofg. By an abuse of notation, the submodule
of 9 generated by the elements of the form [x,y](x€a, ye b) is denoted by
[0, b]. We have [a, b] = [b, a] by identity (3). If 2 E g, [2, a], or [a, 2],
denotes the submodule [K1, (1] = (ad z) (a).
PROPOSITION 3. If a and b are ideals (resp. characteristic ideals) if g, [a, b] is an
ideal (resp. a characteristic ideal) cf 9.
l LIE AwEBRAS

Let D be an inner derivation (resp. a derivation) ofg. If x e :- andy e b, then

D([x,y]) = [Day] + [a Dy] 6 [-1. b]~
Hence the proposition.
[fa is a submodule ofg, the set ofx e 5 such that (ad x) .a C a is a subalgebra
n of g called the narmalizer of a in g. If further a is a subalgebra
ofg,thenaCnandaisanidealofn.

5. DERIVED SERIES, LOWER CENTRAL SERIES _

The characteristic ideal [9, g] is called the derived ideal of a Lie algebra g and
denoted by 95.
Every submodule of 9 containing 99 is an ideal of 9.
The derived min ofg is the decreasing sequence 9°g, 9‘9, . . . of characteristic
ideals ofg defined inductivelyasfollows: (l) 9°g = 5; (2) 9""9 = [9’g,9'g].
The lower central min of g is the decreasing sequence V‘s, V1,. . . of charac-
teristic ideals of 5 defined inductively as follows: (1) g‘g = g; (2)
«“19 = [9, «95]. Then 2’39 = 99 and WHQ 3 9’9 for all p, as is imme-
diately seen by induction on p.
PROPOSITION 4. Let g and b be twa Lie algebras om K and f a homomorphism
Ufa onto b- T71¢")"(9‘10 = Q’bJW’Q) = “3%.
If a and b are submodules of g, it follows immediately that

f([", b]) = [f(°):f(b)]-

The proposition is then immediate by induction on p.
COROLLARY. Let g be a Lie algebra and a an ideal qf 9. F0! the Lie algebra g/n to
be mrmnutatiwJ it is necessary and W that a D 99.
To say that g/a is commutative amounts to saying that 9(g/a) = {0}.
But 9(g/a) is, by Proposition 4, the canonical image of 99 in g/n.

5. UPPER CENTRAL SERIES

Let 9 be a. Lie algebra and P a subset of g. The cenhaliw of P in g is the set of
elements of g which are permutable with those of P. This centralizer is the
intersection of the kernels of the ad y, where y runs through P ; it is therefore a
subalgebra of g.
PROPOSITION 5. Let 9 be a Lie algebra and a an ideal (resp. a chatactzrislic ideal) q.
The :entralizer n’ of a in g 1': an ideal (resp. a chaiamristic ideal) (ff 9.
Let D be an inner derivation (resp. a derivation) ol'g. Ifx e n' and y E a, then
[DH] = D([x,y]) — [’8 Dy] = 0;
hence D: e (1’. Hence the proposition.

6
 muons § 1.7

Let g be a Lie algebra. The centralizer of g in g is called the centre of 9,

that is the characteristic ideal of x E 9 such that [x, y] = 0 for all y e g. The
centre of g is the kernel of the homomorphism x' >—> ad x.
The upper eentral series of g is the increasing sequence «cg, 3’19, . . . of charac-
teristic ideals of 9 defined inductivelyas follows: (1) fig = {0}; (2) V, Hg is the
inverse image under the canonical mappingofg onto 9M,g ofthe centre of9Mg;
The ideal G’lg is the centre of gt
7. EXTENSIONS
DEFINITION 5. Let a and b be two Lie algebra: over K. An exmm'on b by a is a
sequence:
1. II
a —’ g —> b
wheregxkaLiealgebmavnK,pamzjectiuehmnomphimofgonlabandlan
infective hamamarphirm 0f a 07110 the kernel qf (1..
The kernel n of y. is called the kernel of the extension. The homomorphism A
is an isomorphism of 0 onto n and the homomorphism it defines an isomorphism
of g/n onto b when passing to the quotient.
By an abuse of language, 9 is also called an extension of b by a.
Two extensions:
1 u x , w
a —> g —> b, n ——> g ——> b
are said to be equivalent if there exists a homomorphismf of 9 into 9’ such that
the following diagram:

ayix‘,
l
X9,/

is commutative (that is such thatfn )t = l', p.’ of = It). We show that such a
homomorphism is necasarily bxjeefivei First f is injective. For if x e g is such
thatf(x) = 0, then p(x) = y.’(f(x)) = O and hence x = My) for some y e a;
then A’(y) =f(x(y)) =f(x) = 0,hencey = Oand hence: = 0. On the other
hand, f is surjective. For p.’ of = p. is suljective and hencef(9) + A’(n) = 5’;
on the other handf(g) Df()\(a)) = A’(a).
It follows from this that the relation just defined between two extensions
of b by a is an equivalence relation.
Pxoposmou 6. Let
A II
a —> g —> b
be an extensionofbbyaandnitxkemel.
I LIE AmEBRAS

(a) Ifthere exist: a .mbalgebra m ofg supplementary to n in g, the restriction of p. to m

is an isomorphism ofm onto b. (/1: denotes the inverse ixornorphinn ofthu restriction, v is a
homomorphism of b into 9 and p. a v i: the identity automorphism of b.
(b) Conversely, if there exists a homomorphism v of b into 5 such that p. a v i: the
identity automorphism of b, then v(b) i: a supplementary :ubalgebm of n in g.
The assertions of (a) are immediate. 0n the other hand, let v be a homo-
morphism of 1: into 9 such that y. o v is the identity automorphism of b. Then
v'(b) 1s a subalgebra of g and g is the direct sum of v(b) and y. (0) = n (Algebra,
Chapter VIII, § 1, no. 1)
DEFINITION 6. Let
A II
a -—> g —> b
be an extension of b by a and n it: kernel. This extension is called inessential (resp.
trivial) if there exist: a subalgehm (rap. an ideal) ofg supplementary to n in g. This
extension is called eentral if n is contained in the tentre of 5.
If the extension is trivial, let m be an ideal ofg supplementary to n in 9. Then
(cf. no. 1) g is canonically identified with the Lie algebra m x n and hence
with the Lie algebra a x b. Conversely, let a and b be two Lie algebras;
then a X b is a trivial extension of a by b.
An inasential central extension is trivial. For let 9 be a. Lie algebra, I1 an
ideal of 9 contained in the centre of g and m a subalgebra ofg supplementary to
n in 9. Then [m, g] = [m, m] + [m, n] = [m, m] c m and hence m is an
ideal of g.

8. SEMI-DIRECT PRODUCTS
Let a and b be two Lie algebras over K. It is not easy to construct all the
extensions of b by 0. But we shall describe quite simply all the inmential
extensions of b by a.
Let g be an inessential extension of b by a. We identify a with an ideal of 9, b
with a subalgebra of g supplementary to a and the module g with the module
a x b. For all h e b, let (I), be the restriction to a ofad, b; this is a derivation ofn
and the mapping h » it» is a homomorphism of b into the Lie algebra of deriva-
tions of a. On the other hand, for a, a' in u and b, b' in b, we have:
(6) [(a, b), (a', b’)] = [a + h, a’ + b’]
= [4, 4’] + [41, ’1'] + lb, '1’] + [5, 1"]
= (la, 0'] + W“ - {Mb [5, 5'1)-
Conversely, let a and b be Lie algebras over K and h H 4:. a homomorphism
of 6 into the Lie algebra of derivations of a. On the product 9 ofthe K—module: a
and b we define the bracket of two elements by writing:
[(11,12). (a’,b')] = ([a, a’] + ¢.a’ — w, [1), b’])
 sum-imma- pkonucis § 1.8

for all a, a’ in a, b, b' in b. It is immediate that this bracket is an alternating

bilinear function of (11,11), (a', b’); we show that, given 3 elements (a, b),
(a’, b’), (a', b’) ofn x b: '
(7) [(B, b), [(u', I"). (0', b')]] + [(4’) I"), [(0') I"), (a, b)]]
+ [(a', 1")» [(11, 5), («Cl/)1] = 0-
As the left hand side of (7) is an alternating trilinm function of (a, b), (a’, b'),
(a', b'), it suffices to make the verification when this system of elements takes
one of the following forms:
(8) (a, 0), (a’. 0). ('1', 0)
(9) (a, 0), (a’, 0). (0, b')
(10) (a, 0), (o, 1'), (o, b')
(11) (0,1), (0,1'), (0, b’).
In cases (8) and (l l), relation (7) is an immediate consequence of the Jacobi
identity in a and b. In case (9), we have
[(a, 0), [(030), (0, 1")1] = [(11, 0), (filo-E3 0)] = (-[4; halo)
[(a': 0)) [(0, b”): (as 0)]1 = [(a', 0). (than, 0)] = ([43 W0]: 0)
[(0) ’1'), [(HJ 0): (R3 0)]1 = [(0, I"): ([0, 01,0)] = (¢o~([a, 0'1),0)
and relation (7) follows fiom the equation:
lira“: “ID = [in-o ‘1‘] + [‘1’ ‘l’b'a'l-
In case (10), we have:
[(1110): [(0:5'), (0, ”1] = [(010), (0: [’73 VD] = (-<l>w.o~14,0)
[(0, b'), [(0, I"), (a, 0)]1 = [(0, I"): (W11, 0)] = (MW, 0)
[(0, I"), [(a. 0), (0, II')]] = [(0, 5'), (filo/'1, 0)] = (—¢w¢ra,0)
and relation (7) follows from the equation:
imam = their — l’o'i’nu
Hence a Lie algebra structure has been defined on g. The mapping (a, b) 1—» b
of 9 onto b is a homomorphism 1:. whose kernel n is the ideal of elements of g of
the form (a, 0). The mapping a» (a, 0) is an isomorphism )1 of a onto n.
Hence:
(12) a —*> g '—"> b
is an extension of b by a of kernel 11, which is said to be canonically defined by a,
b, 4» The mapping b >—> (0, b) is an isomorphism v of 5 onto a subalgebra of g
supplementary to n in 5; hence the extension is inessential.
! Ln: mums

If n is identified with n undo“ I and b with v(b) under v, then, for a E n and
b E b:
(ad l0J1 = [(0, b), (a, 0)] = (4M, 0) = 4M-
tn (b = 0, g is the product Lie algebra of b and a. In the general case, 9 is
called the semi—direct product of b by a (corresponding to the homomorphism
b r—> 4)” of b into the Lie algebra of derivations of a).
We have therefore established the following proposition:
Pnorosmon 7. Let a and b be two Lie algebra: over K,
L II
n ——> g ——> b

an inmential ulna-ion 12/ b by a, v an {307710111t [fit onto a :ubalgebra 0f 5 such that

uevisthaidauityautonmphimqmd¢themmdinghommwphhmqinto
the Lie algebra 9f derivation: af :1. let
cl)%1>b
be the inwenh'al anemia» q by a canonically dqiywd by 4:. Then the map-
ping (a, b) >—> Ma) + v(b) i: an isomorphirmfqf go onto 9 and the/"allowing diagram

«Wm
x9 1
i: cammutatiue, so that the two ekiom are equivalent.
Example 1. Let g be a Lie algebm over K and D a derivation of 3. Let I; be the
commutative Lie algebra K; The mapping x w ADO e K) is a homomorphism of
1) into the Lie algebra of derivations of g. \N’e form the corresponding semi-
direct product I ofb by 9. Let x0 be the element (0, 1) of I. For all x s g,
Dx = [x.,, x].
Example 2. Let g be a Lie algebra over K, M a K-module and p a homomor-
phism of 9 into gl(M). If M is considered as a commutative Lie algebra, the Lie
algebra of derivations of M is gl(M). We can therefore form the semi-direct
product [9 of 9 by M corresponding to p.
In particular, let 9 = 51(M) and p be the identity mapping of gl(M). The
semi-directproductofg byM is then denoted by af(M) (or af(n, K) ifM = K") .
An element of af(M) is an ordered pair (m, u), where m e M, u E 9|(M); and
the bracket is defined by
Him"). ("1" u')] = (“(M') - "'(M), [I‘d/l)-
 CHANGE or BASE RING § 1.9

‘When M is a finite-dimensional vector space over R, nf(M) is canonically

identified with the Lie algebra of the afim group of M1
Let t be a Lie algebra over K. A linear mapping 0 of t into ai(M) can be
written x >—> ((§(x), n(x)), where I is a linear mapping 0“ into M and '4 a
linear mapping of t into 5|(M). We examine the conditions that C and n must
satisfy for 0 to be a homomorphism. For x E t, y e t, we must have

OHM/J) = [0(x), 9M]

that is
(t([x,y]), 7)([*:y])) = [(300, 110)): (my), mm
= (710‘) H!) — 1M) I“), [11(3), 7'1(y)])~
Hence for 9 to be a homomorphism oft into nf(M), it is necusary and sufficient
that 11 be a homomorphism oft into 91(M) and that C satisfy the relation:
(13) «FWD = 71(1) J4M - WI) 320‘).
Let N be the K-module M x K. We make t to be the subalgebra ofgl(N) con-
sisting ofthe w e 91(N) such that w(N) C M. For all w 6!, let 7)(w) E 91(M) be
the restriction ofw to M and let t(w) = w(O, l) e M. For In, E t, w, at,
C([wu wal) = w1(t(w2)) - wz(C(wx)) = n(w1).§(wa) — 71(14):) -§(wx)»
Hence the mapping :1) >—> (C(10)), 1)(w)) is a. homomorphism 0 0“ into af(M).
Clearly G is bijecliw. Let 4) = 0“. If (m, u) e af(M), (Mm, u) is the element w oft
defined by
w(m’, X) = (u(m’) + m, 0).
af(M) is often identified with the subalgebra t of51(N) under the isomorphism

*When M is a finitedimensional vector space over R, the homomorphism 4;

of nf(M) into gI(N) corresponds to a canonical homomorphism q, of the afline
group A of M into the group GL(N); if a e A, Ma) is the unique element g of
GL(N) such that g(m, 1) = (a(m), l) for all In E M. This homomorphism is
injective and MA) is the set of automorphisms of N which leave invariant all
the linear varieties of N parallel to M.*

9. CHANGE OF BASE RING

Let K0 be a commutative ring with unit element and p a homomorphism of
K0 into K mapping unit element to unit element. Let 9 be a Lie algebra over
K. Let g’ be the algebra obtained by considering g as an algebra over Ko by
means of p (cfi no. 1). Then 9’ is a Lie algebra. The subalgebras (resp. ideals) of
g are subalgebras (rap. ideals) of 9’. Ifn and b are submodules ofg, the bracket
[0, b] is the same in g and in 9'; for [a, b] is the set of elements of the form

Z [x,, yJ where x, E a, y. E b. It follows that 9’9 = 9'9’, W9 = @’g' for allp.

4-1

ll
I ma Mam

The centralize: ofa subset is the same in g and 9’. Hence fps = 7,9' for all

Let Kl be a commutative ring with unit element and a a homomorphism of

K into K1 mapping unit element to unit element. Let g be a Lie algebra over
K. Let 90h) be the algebm over Kl derived from g by extending the base ring
(cf. no. l). Then gm) is a Lie algebra. Ifa is a subalgebra (resp. an ideal) of g,
the canonical image of “am in 9m) is a subalgebra (resp. an ideal) 01' gm). If a
and b are submodules ofg, the canonical image in gm) of [0, bl, is equal to the
bracket of the canonical images of am) and ban. It follows that 9‘19“”) is the
canonical image of (994mm, and that @‘(gmg is the canonical image of
w’is )l‘
If?‘ is a field, K1 an extension field of K and a the canonical injection of K
into K,, then with the usual identifications we have

ll“: blah) = [“(Km [’11)]: 9496.)) = (9'90“):

g"(90(9) = («99)(K1r
These results are completed in §2, no. 9.
If M is a finitedimensional vector space over the field K, Mm) is a finite-
dimensional vector space over K, and the associative algebra yiMam) is
canonically identified with the associative algebra $(M)m,. Hence the Lie
algebra gl(M(x‘,) is canonically identified with the Lie algebra gl(M)(xl,.

§2. ENVELOPING ALGEBRA OF A LIE ALGEBRA

1. DEFINITION OF THE ENVELOPING ALGEBRA

Let 9 be a Lie algebra over K‘ For any associative algebra with unit element L
over K, an nit-mapping ofg into L is a K-linear mapping r: ofg into L such that
a(its yl) = UMEU) - “@600 (3) y in 9)
(in other words a homomorphism of 9 into the Lie algebra associated with L).
If U is another associative algebra with unit element over K and 'r a homo-
morphism of L into L’ mapping 1 to 1, then 1 o a is an m—mapping of 9 into L'.
We shall look for an associative algebra with unit element and an a-mapping
of g into this algebm which are universal (Set Theary, Chapter IV, § 3, no. 1).
DEFINITION 1. Let g be a Lie algebra aver K, T the temar algebra if the K-rnadule g
and J the two-sided ideal 0f T generated by the tensor: x ® y — y ® a: — [x, y] where
x E g, y e g. The associative algebra U = TI] is called the enveloping algebra qf'g. The
restriction to g qf the emnical mapping (JT ante U i: called the :arumical mapping q
into U.
Let T + be the tWD—Sidcd ideal ofT consisting ofthe tensors whose component

12
 mwpmc ALGEBRA or A nooucr § 2.2

oforder 0 is zero. Let T0 = K. I be the set ofelemenm ofT oforder 0. Let U+

and UL) be the canonical images ofT+ and To in U. As] C TH the decom-
position into a. direct sum T = To + T, implies a decomposition into a direct
.sum U = Uo + U +- The algebra U therefore has a unit element distinct from
0 and U0 = K. I. For all x e U, the component of x in U0 is called the constant
term of x. The elements with constant term zero form a two-sided ideal of U,
namely the two-sided ideal U” generated by the canonical image of g in U.
The associative algebra U is generated by l and the canonical image of g in

Ifxsg andyeg,x®y —y®xand [23y] are congruent inTmoduloJ;

hence, if <1u denotes the canonical mapping of 9 into U,
”o(")°o(y) — “o(!/)°'o(") = “0“":91)
in U. In other words, no is an m—mapping of 9 into U.
PROPOSITION 1. Let a be an nt-nmpfiing of 9 into the associative algzbm L with unit
element. There exist: am and only one hommaqohirm 1- aU into L, mapping 1 to i, such
that a = 1 c on, when an denotes the canonical mapping q into U.
Let 1" be the unique homomorphism of T into L which extends a and maps
1 to 1. Then, for any in g,
40! 8w - y ® I? - DMD = 6(x)<r(y) - awak) - a([Ml) = 0;
hence -.' is zero on] and defines on passing to the quotient a homomorphism
r of U into L, mapping 1 to I, such that a = 1'0 do. The uniqueness of 1 is
immediate since a°(g) and I generate the algebra U.
Let 9’ be another Lie algebra over K, U’ its enveloping algebra and 0;, the
canonical mapping of 9' into 0'. Let (b be a homomorphism of 9 into 9’. Then
all, a {a is an m—mapping ofg into U’; hence there exists one and only one homo-
morphism }; of U into U’ mapping 1 to 1 and such that the diagram
9—0) 9'

°olU <—. U1°:

is commutative. This homomorphism maps the elements of U whose constant
term is zero to elements of U' whose constant term is zero. If 9' is another Lie
algebra over K and (b' is a homomorphism ofg' into 9', then (tb’ o (1))” = ii? 0 (b.

2. ENVELOPING ALGEBRA OF A PRODUCT 0P LIE ALGEBRAS

Let 9,, 92 be two Lie algebras over K, U. the enveloping algebra of g. and o" the
canonical mapping of 9, into UI (i = l, 2), Let g = g, x 9,,Ube its enveloping
algebra and a the canonical mapping ofg into U. The canonical injections of 91
and 92 into 9 define canonical homomorphisms of U1 and U2 into U whose

13
I LIE ALGEBRAS

images commute and hence a homomorphism (I) of the algebra UI ®‘ U2 into

the algebra U, mapping 1 to 1.
PROPOSITION 2. The homomorphism d) i: an algebra ismrpllicm
The mapping “'1 (*1, ‘2) H Uii‘i) ® 1 + l 8 020‘s) (‘1 5 Hi: :52 E 82) is 3-“
tit-mapping of 9 into UI ®K U, and hence there exists (no. 1, Proposition 1) a
unique homomorphism 1- of U into Ul ®x U, mapping 1 to 1, such that:
( l) a" = 1 o a.

Now¢aeea=¢oa’=aandro¢oa’= too-=0”, hence¢otand1=¢

are the identity mappings of U and U1 @x U, respectively. Hence the prop-
ositioni
U, ®x U, is identified with U under the isomorphism (I). Then the canonical
mapping of 9 into U is identified by (l) with the mapping:
(:1, x3) D-D 610:1) ® 1 + 1 ® a,(x,).
Analogously, Kg”. . . , g, are Lie algebras over K with enveloping algebras
U1, 4 . ., U“, the envelopingalgebraU of91 x - - - x 9,, iscanonically identified
with U; ‘83- - -®xU,, and the canonical mapping of 9‘ X - - - x 9,. into U is
identified with the mapping:
(x,,...,x,,)>—>ul(xl) ® 1 ®---®l +...+ l ®~~~® l ®on(x,,)
(a, denoting the canonical mapping of 9‘ into U,).
3. ENVELOP'ING ALGERIA OF A LIE SUBALGEBRA
Let g be a Lie algebra over K, b a subalgebra of g and o, c’ the canonical
mappings of g, 9 into their enveloping algebras U, V. Then the canonical
injection 1‘ of i) into 9 defines a homomorphism i, called canonical, of V into U
such that o' oi = 1” u o’. The algebra x‘ (V) is generated by l and a(b). We shall
see (no. 7, Corollary to Theorem 5) that i is injective in important cases.
If i) is an ideal of g, the left ideal of U generated by o(b) coincides with the
right ideal generated by a(b), in other words it is a two-sided ideal R. This
follows since, for x e b and x’ e g,

60060?) = amok) + 6([x,x’])

and [x, x’] e l).
PROPOSITION 3. Le! b be an ideal afg, p the canonical hamonwrphim afg onto 5/b and
W the enveloping algebra of 54/1). The homomorphism:
fizU »> W
defined canonically byp is rurjeclive and it: kernel is the ideal R of U generated by a(b).

l4
 ENVELOPING Amman or THE OPPOSITE ALGEBRA §2.4

Let a” be the canonical mapping of 3/!) into W. The commutative diagram:

9—‘—>94>slb
1.15 if
v—L>U——>w
proves that i is zero on 6(b) and hence on R. Let 4) be the canonical homomor-
phism of U onto U/R. Them exists a homomorphism 4) of U/R

9—'—>sb

. 4*
U4» U/R $W

into W such that f = 4) o 4:. The mapping 4; o a of 9 into U/R is an oz-mapping

and is zero on b and hence defines an u-mapping 6 of 9/!) into U/R such that
00;) = 4106'. Then 4)o6op = 4)o4;oa= 0'01). Hence (1)0 9 = 6'. There
exists (no. 1, Proposition 1) one and only one homomorphism 4)’ ofW into U/R
mapping 1 to l andsuchthate = 4)’oo'.Then4)’o4)o0 =4)’oa’ = Band
4) o 4)’ o a” = 4) o e = a” and hence 4)’ o 4) and 4) o 4)’ are the identity mappings
of U/R and W respectively. This completes the proof.
U/R is identified with W under the isomorphism 4). Then the canonical
mapping er” of 9/!) into W is identified with 0, that is with the mapping of 9/!)
into U/R derived from o by taking quotients.

t ENVELOPING ALGEBRA OF THE OPPOSITE LIE ALGEBRA

Let g be a Lie algebra over K, 9" the opposite Lie algebra and o‘ and Go the
canonical mappings of g and 9° into their enveloping algebras U and V. Then
a is an a-mapping of 9° into the associative algebra U" opposite to the asso-
ciative algebra U. Hence there exists one and only one homomorphism 4) of V
into U° mapping 1 to l and such that a = 4) 0 0'0.
PROPOSITION 4. The hmnmnarphinn 4) it an isomorphism U” V onto U°.
There exists a homomorphism 4)’ of U into V° mapping 1 to l and such that
do = 4)’ 0 cr. 4)’ can be considered as a homomorphism of U” into V. Then
co = 4)'04)ua°ando' = 4)o4)’ooandhence4)’o4)and4)o4)’aretheidentity
mappings of V and U. Hence the proposition.
V is identified with U° under the isomorphism 4). Then (1., is identified with a.
With this identification, the isomorphism 0: x »—> —-x of 9 onto 9" defines an
isomorphism 5 of U onto V = U°. This isomorphism can be considered as an

15
I LIE ALGEBRAS

antiautomorphism of U. It is called the primipal antiquiamarphirm of U. If

x1, . . ., x" are in 9, then:

(2) M.) . . .co.» = was.» . Mo.» = (-a<x.)>...<—a(x1>)

= (—1)"a(x.>. . -a(x1)-
5. SYMNIETRIC ALGEBRA OF A MODULE
Let V be a K-module. V can be considered in a unique way as a commutative
Lie algebra. The enveloping algebra of V can then be obtained as follows: let
T be the tensor algebra ofV; let I be the two-sided ideal of T generated by the
tensors x ®y — y (8 x (xEV,yEV); then form the algebra S = T/I.
Recalling (Algebra, Chapter III, § 6) that S is called the symmetric algebra of
V, we summarize briefly the properties needed in this chapter, the proofs of
which are immediate. Let T” be the set of homogeneous tensors of order n in T.
Then I = (I n T“) + (I n T“) + ~ . ~ and hence S is the direct sum of the
canonical images S” of the T". The elements of S" are called homogeneous of
degree n. S0 = K. 1 , S1 is identified with V and S"S" C 5"". The algebra S is
generated by l and S1 = V. Clearly any two elements of 51 are permutable
and hence S is commutative. If V is a fm K-module with basis (1,)“ A, the
canonical homomorphismfof the polynomial algebra K[X,‘],‘E A onto S which
maps 1 to l and Xx to xx for all k e A is an isomorphism: for by the universal
property of S (no. 1, Proposition 1) there exists a homomorphism g of S into
K[X,~],~M which maps 1 to l and x,‘ to X,~ for all A e A andfand g are inverse
homomorphism of one another.
Let S’" C T" be the set of homogeneous symmetric tensors of order 7:
(Algebra, Chapter III, § 5, no. 1, Definition 2). If K is a field ofcharacteristic 0,
S’" and I n T" are supplementary in T". For let (:r,.),‘E A be a basis of V. We
give A a total ordering (Set Theory, Chapter III, § 2, no. 3, Theorem 1). Let A,‘
be the set ofincreasing sequences ofn demerits ofA. For M = (M, . . . , 1,.) EA,”
let
1
ll)! = :i ;_ ”Mr-n) ®' ' ' ® ”Mar

The y“ for M E A,I form a system of generators of the vector K-space 5'". Now
their canonical images in S" constitute, by the above paragraph, a basis of S".
Hence (yM)ME A" is a basis of a supplementary subspace of I h T" in T”
(Algebra, Chapter II, § 1, no. 6, Proposition 4), which establishes our assertion.
Thus, when K is a field ofcharacteristic 0, the restriction to S”I of the canoni-
cal mapping T'| ~> S" is an isomorphism of the space 5'" onto the space S'| and
therefore has an inverse isomorphism. The inverse isomorphisms thus obtained
for each 71 define a canonical isomorphism of the space S onto the space
S’ = 120 S’" of symmetric tensors.

16
 FILTRATION or THE ENVELOPING ALGEBRA §2.6

G. FILTRATION OF THE ENVELOPING ALGEBRA

Let g be a Lie algebra over K and T the tensor algebra of the K-module 9.
Let T’l be the submodule ofT consisting of the homogeneous tensors of order n
andTn = ‘2‘ T‘.ThenT,I C T,,.,1,T0 = K.l,T_l = {O}andT,,T,, C Tn”.
Let U,, be the canonical image of T,l in the enveloping algebra U of 9. Then
U" C Una: U0 = K.l, U_1 = {0} and UnUp C Ufipg hence U can be
described as an algebra filtered by the U” (Commutative Algebra, Chapter III,
§ 2, no. 1); the elements of Un will be said to be nffiltmliun s it.
Let G’I be the K-module U,,,/U,l_1 and let G be the K—module the direct sum
of the G". Multiplication on U defines, on taking quotients, a bilinear mapping
of G” x G'" into G'| H" and hence a bilinear mapping of G X G into G, which
is associative. Thus G is given an associative K—algebra structure. Then
G"G"‘ C GM”. The elements of G’I are said to be of degree n. The graded alge-
bra thus obtained is just the graded algebra associated with the filtered algebra
U (Cumutaliue Algebra, Chapter III, §2, not 3).
Let 4),, be the composition of the canonical K-linear mappings
T” —> U,l —> G".
As T“ is supplementary to Th1 in T”, 42,. is surjective. The 42,, define a K-
linear mapping (1) of n T'I = T onto n G" = G.
PROPOSITION 5. The mapping (b 9/T onto G is an algebra homomhim mapping 1 to
l and {1 zero on the twovsided ideal generated by the Immune: ® 3/ — 31 ® a (x e g, y e g).
If te T" and t’ e T’, then ({>(t)<l>(t') = ¢(lt’) by definition of the multi-
plication on G. Hence 45 is an algebra homomorphism and clearly d>(l) = I. If
x, y are in 9, then x 8 y — y ® a E T2 and the canonical image ofthis element in
U, is equal to that of [x, y] and therefore belongs to U1. Hence
(Ma: ® 3/ — y 6; x) = 0, which proves the proposition.
Let S be the symmetric algebra of the K-module g and ‘r the canonical
homomorphism of T onto S. Proposition 5 proves that there exists a unique
homomorphism (-1, called rannm‘ml, of the algebra S onto the algebra G,
mapping 1 to I, such that 45 = a) o 1-. We have (”(5") = (“T”) = G". Let 1,, be
the restriction of r to T", 0),, the restriction ofto to S", 4;" the canonical mapping
of T" into U" and 0,1 the canonical mapping of U” onto G”. The definition of
4a,, proves that the following diagram is commutative:

. U"
T"/
Uu \Gn
0.

kAsn/ml
I LIE ALGEBRAS

PROPOSITION 6. If K is Noethen'aa and g i: a finitely generated module, the ring U is

right and left Noetherian.
S is a finitely generated algebra over K and hence a Noetherian ring
(Commutative Algebra, Chapter 111, §2, no. 10, Corollary 3 to Theorem 2).
Hence G, which is isomorphic to a quotient ring of S, is Noetherian. Hence U
is right and left Noetherian (Commutative Algebra, Chapter III, §2, no. 10,
Remark 2).
COROLLARY. Suppose that K is a field and that g is finitedimendonal over K. Let _
Ii, . . ., 1,, be right (resp. lefl) ideals offinite cadimerLrion in U. Then the product ideal
111,. . .I,,‘ i: rffinite eodimemion.
By induction on m it suflices to consider the case of, for example, two right
ideals. The right U-module II is generated by a finite number of elements
u1,...,u,, (Proposition 6). Let 111,” .,vq be elements of U whose classes
modulo In generate the vector space U/Ia. Then the canonical images in
11/111: of the 11,0, generate the vector space Ill/Illa, which is therefore finite-
dimensional. Hence dimx(U/Illg) = dimK(U/Il) + dimK(11/ILIR) < +00.

Remark. Let g’ be another Lie algebra over the ring K, U’ its enveloping
algebra, U,’, the set of elements of U’ of filtration $7; and U" (resp. U’") the
set of canonical images in U (resp. U’) of the homogeneous symmetric
tensor: of g (resp. g’) of order 11. Let n be a homomorphism of 3 into 9' and
let i be the corresponding homomorphism of U into U’. Then

71(Un) C U?» i(U") C U’"-

In particular, the principal antiautomorphism of U leaves U,I and U’|
stable. The K-linea: mapping of T" onto itself which maps

fli®*2®"‘®flu 1° xu®xn-1®‘ "@1’1

for all at}, . . ., x“ in 5 is a symmetry operator and hence leaves fixed the
homogeneous symmetric tensors of order 7:. Hence the principal antiauto-
morphism of U induces on each U" the homothety of ratio (— l)".

7. THE POINGARE—BIRKHOFF-WITT THEOREM

THEOREM 1. Let g be a Lie K—algebra, U its enveloping algebra, G the graded algebra
associated with thefiltered algebra U and S the symmetrit algebra of the K-module g. If g
is a free K-module, the canonical homomorphism o): S —> G is an isomorphism.
Let (19.),6A be a basis of the K»module 9; we give A a total ordering (Set
Theory, Chapter 111, §2, no. 3, Theorem 1). Let P be the polynomial algebra
K[Zh]hs A in indeterminates 2,, in one-to—one correspondence with the at»
For every sequence M = (M, 12,. . ., ) of elements of A, let 2M denote the
monomial zMzM. . .2M and x" the tensor "M (8 x“ @. ~ - ® 1,“.
The 2M, for M increasing, form a basis of the K—rnodule P (we make the
convention that D is an increasing sequence and that 2,, : 1). Let P1, be the

18
 THE Pomonm‘sqammOmm THEOREM § 2.7

submodule of polynomials of degree <1). We shall first prove several lemmas.

(To abbreviate, we write A < M if A < P. for every index [1. of the sequence
M.)
Lemma 1. For every integer p 2 0, there exist: a unique homomorphism f,, of the K-
mmiule g ®x P, into the K—nmdule P mtisfiing the following canditiom:
(An) fix”). ® 2M) = ZAZM f” A < M: Zneppi
(B1,) f,(x,‘ 8) 2M) — zkzM GP for 2M 6 Pa: 4 <p
(Cs) p‘A @fni’fln ® Zn» =fi1(xu ®fiz(xh ® Zn» +f..([m) tn] ® ZN)
far 1" e Pp_1. (The term: appearing in (C,) are meaning/ill by(B”).)
Moreover, the mtrietian qff, to 5 ® Pr 1 coincide: withfk 1.
The last assertion follows from the when since the restriction of f, to
g ®K P1,,1 satisfies conditions (A,_,), (3,,1) and (CPI). We shall prove
the existence and. uniqueness off, by induction on [2. For 12 = 0, condition
(AD) givesffix,‘ ® 1) = z,‘ and conditions (Bo) and (Co) are then obviously
satisfied. Suppose now that the existence and uniqueness off,,_1 are proved.
We show thatf,,_1 admits a unique extensionf, to g ®x P,7 satisfying conditions
(A,), (13.) and (0.).
We must define flog ® 2“) for an incrasing sequence M of}: elements.
If A S M, the value is given by condition (A,). Otherwise, M can be
written uniquely in the form (n, N), where y. < A, p. S N. Then
zM = zuzu =f,,_1(.1r,.l ® 2")
by (Au-l): so that the left hand side of (0,) is f,(zA ® 1“). Now the right
hand side of (C,) is already defined: for (B,_1) allows us to write:
fp(7‘t 8 ZN) =fp-1(‘;. ® ZN) = ZAZN + 7"
where w e P,_1; hence the right hand side of (0,) becomes:
tuZN +fp—1(*u ® w) +fv—1([*xa "ul ® ZN)‘
Thusf, is defined uniquely and obviously satisfies conditions (A,) and (3,).
Condition (0,) is satisfied if p. < A, p. g N. As [xm xi] = —[x,‘, xu], condition
(0,) is also satisfied for A < (1., A S N. As (0,) is trivially satisfied for A = u,
((3,) is therefore satisfied if A < N or p. S N. If none of these inequalities holds,
= (v, Q), where v g Q, v < A, v < in. Writing henceforth to abbreviate
f,,(x ® 1) = xz for x e g and z 6 Pp, we have by the induction hypothesis:
"uZN = Minn) = *va) + [*mxvllm
Now z”;a is of the form zuz.2 + w, where w E P,_,. (C?) can be applied to
xh(xv(zuzq)), since v s Qand v < p, and to xA(s) by the induction hypothe-
sis, and hence to x,‘(zv(xuzq)). Hence:
Jami") = xv(xk(xuzfi)) + [16» 4105.10) + [xmxvlmld
+ [*A: [*mxvlllu-
I LIE ALGEBRAS

Exchanging 1 and y. and subtracting:

”LOWZN) — xui‘ilfl) = "v("h("uzc) " "u("hza))
+ [3» [11» "VHZIJ — [Kw [a xviila
"v(["h1 #11110) + ["L: [3’1” *3:a + [xiv [v xfl1z0
[#1, xulmzu) + (PM ["m M]
+ ["m [16.1, xvi] + [Kw [MMdDza
and hence by the Jacobi identity
50.911111) — *ul‘xzu) =[1n ”nil"
which completes the proof of Lemma 1.
Lemma 2 There :11:i an a—mapping a 9/ 9 into 3141’) such that:
(I) 110:,s = lufi" 7‘ <M
(2) a(x,‘)zu-
= zAzM (mod P,) y" M has]: elemmli.
By Lemma 1 there exists a homomorphismf of the K-module g 8,; P, into
P satisfying, for all p, conditions (A,), (3,), (0,) (wheref, is replaced byf),
This homomorphism defines a homomorphism a of the K-moduie 9 into the
K-module 91((1’) and a is an a—mapping because of condition (C,). Finally, :1
satisfies properties (l) and (2) of the lemma because of conditions (A,) and
, .
Lemma 3. Let t be a [may in T,I n]. The hwwgenem component tn qfl Iy'order n {.1
in the kernel I 43/ the canonical homomorphism T —> S,

We write I in the form ”21 XM" where the M. are sequences ofn elements of
A. The mapping 6 extends to a homomorphism of the algebra T into the
algebra $14?) (which we shall also denote by a), which is zero on J By Lemma

,cr.(t) 1 is a polynomial whose terms of highest degree are 2 111411 As I EJ:

[=1

c(t ) = 0 and hence ”21 z”, = 0 in P. Now P 1s canonically identified with S,

since 9 has basis (1;). Hence the canonical image of t,l in S is zero, that is
t,‘ e I.
We can now prove Theorem I It 13 necessary to prove that the canonical
homomorphism ofS onto C 1s injective In olher'words, if! E T” and 1}; denotes
the canonical homomorphism of '1 onto U, it is 11ccessary to show that the
condition 1];(t) e U,._1 implies 1E1. Now 1H!) 6 U"- 1 means that there exists
a tensor t’ e Tn_1 such that t — 1’ GJ. The tensor t — t’ admits t as homo-
geneous component of order n and hence is I by Lemma 3.
COROLLARY l. Supprm that 9 i: afm K—madulz. Let W h: a mb—K—modul: qf T".

20
 'n-nz rommkfiqmmomwm THEOREM § 2.7

.6”, in the notation ofdiagram (3), the restriction cf 1-,. to W i: an iramorphirm qfW onto
5", then the rerhictian af 4),, ta W ir an immorphirm qf W onto a supplement of Un_1
in U”.
The restriction to W often a 1',I is a bijection ofW onto C"; so is the restriction
6,, o 41,, to W. Hence the corollary.
COROLLARY 2. If g is aflee K-module, the cammieal mapping of 9 into iLr enuelaping
algebra ir injective.
This follows from Corollary 1 taking W = T1.
When 9 is a free K-module (in particular when K is a field), 9 is identified
with a submodule of U under the canonical mapping of 3 into U. This con-
vention is adopted from the following corollary onwards,
COROLLARY 3. If g admit: a tatally entered barir 09.)“ A, the element: xxlxh. . .xh
4f the enveloping algebra U, where (M, . . . , An) 1': an arbitrary increasingfinitc sequence
of element: qf A, firm a basis qf the K-module U.
Let A,‘ be the set of increasing sequences of n elements of A. For
M = (1b.. ., A") EA,” let yM = xx] ®"M ®~~~®xw Let W be the sub-
module of T’‘ with basis (yulus 1w Corollary 1 shows that the restriction of a
to W is an isomorphism of W onto a supplement of U,._l in U”. But
‘i’n(!lM) = "xfih- - J‘s“,
whence the corollary.
COROLLARY 4. Let S’” C T" be the set of homogeneous rymmetn'e tensor: qf arder n.
Suppose that K it afield characterim'c 0. Then the cmfiarite mapping If the canonical
mapping:
8" —> S"I —> U”
is an iramorphirm qf the vector space S“ onto a .mpplement q".I in U,,.
This follows from Corollary 1 taking W = 8’”.
Suppose henceforth that K is a field of characteristic 0. Let n, be the
mapping of S’| into U,I just defined. Let U’I = 7],,(S"). The vector space U is
the direct sum of the U". The n, define an isomorphism 'q of the vector space
S = Z S” onto the vector space U = 2: U", called the cammical iramnrlzhirm of
S antu U; this is nut an algebra isomorphism. We have the commutative
diagram:
un
4b- (in
S'n/ 7: \Gn
N sand

21
I LIE AwEBRAS

where each arrow represents a vector space isomorphism. If x1, x2, . . ., x" are
ing,-q,,mapsther ’ xlxa...x, ' ' rdinS,tothe '

31025:“ ”am’amr - Jam

calculated in U.
COROLLARY 5. Let b be a mbalgebra ofthe Lie algebra g and U’ it: enveloping algebra.
Suppose that the K-moduler 9 and g/b arefree (for example IfK is afield). Let (xu), E L, be
a basis (fl) and (when afamily q/element: 1J3 whose mmmiml image: in g/bfarm (1
km of 9/!»
(a) The eanam'ml homomorphism of U' into U i: injective.
(b) [fM is totally ardered, the element: ynl. . 'v where {31 < ~ ' ' € (3.,f0rm a
bad: «J U wandered as a left or right module aver U’.
We give L U M a total ordering such that every element of L is less than
every element of M. The elements at, I". . “calculated in U (where
m1< < a,) form a basis of U' (Corollary 3).xThe elements
x l . .xu'yal. . .yh
calculated in U (where 11 g - - - S a, < [3, S - ~ - S flu) similarly form a basis
of U. Hence the canonical homomorphism ofU' into U maps the elements of a.
basis of U' to linearly independent elements of U and is therefore injective.
It is moreover seen that the ya}. . .ynq (where (i, S ‘ - - g (3") form a basis ofU
considered as a left U’-module. Ordering L U M so that every element of M
is less than every element of L, it is similarly seen that the Ila,» . .q (where
{31 < - - - s {30) form a basis of U considered as a right U’-module.
Under the conditions ofCorollary 5, U’ is identified with the subalgebra of U
generated by I) by means of the canonical homomorphism of U' into U.
COROLLARY 6. Suppose that the K-module 5 is the direct sum qfsubalgebrar 9,, 92, . . . ,
g” and that each 5‘ is afree K-module. Let U. be the enveloping algebra ofg‘ (l S i S a) .
Let 4) be the K-linear mapping ofthe K-madule Ul ®x‘ - - ®K U,I into U defined by the
multilinear mapping (141,. . ., u") >—> a1. . .u,I «fUl x - - - x U,l into U. Then (1) i: a
K-module isomorphirm.
Let (4%“. be a basis of 9.. We totally order Ll U- - -U L,I so that every
element of Li exceeds every element of L, for i 2 j. Then the elements:
(x1342. . .xlw) ®~ - -® (3-3333. . .xzq),
where 11$ kgs-usMs-ugvlsvas-HSVQ, constitute a basis of
Ul ®x' - . (8),; U”. They are mapped by (5 to the elements:

we. . m:
which constitute a basis of U. Hence the corollary.

22
 EXTENSION OF DERIVATIONS § 2.8

COROLLARY 7. If K is an integral domain and g i: a free K—madule, the algebra U

has no divisors of zero.
G is isomorphic to a polynomial algebra over K (Theorem 1) and is there-
fore an integral domain (Algebra, Chapter IV, § 1, no. 4, Theorem 1). Hence
the corollary (Commutative Algebra, Chapter 111, § 2, no. 3, Proposition 1).

B. EXTENSION OF DERIVATIONS

Lemma 42 Let V be a K-modale and T the tensor algebra of V. Let a be an endomor-

phism ofV. There exists one and only one derivation ofT which extend: u. This denvatian
commute: with the symmetry operator: on T.
LetF = V x V x -- - x V (n factors). The mapping
(”Is~~~s"n)"’""1®”2®"'®xn
+31®Wa®”'®xn+“‘+"1®#2®"'®w’n
n n
of F into ® V is multilinear. Hence there exists an endomorphism an of ® V
such that:
u..(x1®~~®xn) =u~1®~~~®xn +~~-+x1 ®~--®uxn
for all x), . . ., x” in V. Then a; = a. Let u be the endomorphism of the K-
a
module T which coincides with an on each T" = ® V and which is zero on
To = K. 1. We show that a is a derivation ofT. Ifxn . . ., x,” y” . . .,y, are
elements of V, then
”((5‘1 ®-"®"n) ®(y1®--~®yp))

= '21”: ®“'®"I—1®uxc®xt+1®"‘®xn®!/1 ®'”®!Ip

F

+ ,2, x1 ‘8" -®xu®y1 ®"'®!l/-1 ®W1®yl+l ®"'®!/v

= ”(”1 ®" ’®"n) ® 0/: ®~~~®yp) + (11 @“’®’%) ®v(y, ®"‘®h)-
By linearity it follows that v is a derivation. The uniqueness of a is obvious.
it
Finally, clearly 14,. commutes with all the symmetry operators on 8’ V, whence
the last assertion.
PROPOSITION 7. Let g be a Lie algebra, U it; enveloping algebra, u the canonical
mapping of 9 into U and D a derivation of g.
(a) There exists an: and only one derivation DU y" U such that a o D = D" a a
(that is :ueh that DU extends D, when 9 can be idenlyied with a .mbmodula ofU under a).
(b) Dl7 leave: stable Uu and the set U” of image: in U of the homogeneous :ymmetrie
tensors aforder a over 9.

23
1 LIE menus

(c) Du commutes with the principal antiautommphism of U. - .

(d) If D it the inner derivation If 9 defined by an element x Qf g, DU :5 the tuner
derivation of U defined by a(x).
Let DT be the derivation of the tensor algebra T of g which extends D
(Lemma 4). The two-sided ideal j of T generated by the
x®y-y®x- [m]
(x, y in g) is stable under Dr. For:
Dw(x ®y —y ®x — [MD = Dr 6% -y <8: Dx - [Dw]
+x®Dy~Dy®x — [x,D_1/].
On passing to the quotient, DT defines a derivation DU of U such that
u- o D = Du o a. The uniqueness of Du is immediate since 1 and 6(9) generate
the algebra U. Assertion (b) is obvious. Let A be the principal antiautomor-
phism of U. We now prove (c). Ifxl, . . ., x, are in g, then
DUA(°’("1) > - $05)) = Du((-1)"6(~n)- . -°("1))
n
= (.1)n l-l
2 60:“) . . .Du(a(x,)). . .a(x,)
= (—- 1)" 2:1 a(x,.) . . .a(Dx,) . . .a(x1)

= A(2
i=1
(10:1). . .c(Dx,) . . 43(4))
= ADu(c(x,) . . .o(x,.)).
Finally, let xe 9. Let A be the inner derivation y >—> c(x)y — yo(x) of U
(Algebra, Chapter IV, §4, no. 3, Exampb 2). Then, for z’ e g,
(A o c)(x’) = o(x)u'(z') — o‘(x’)a(x) = c([x, x’]) = (a o ad x)(x‘),
whence A a a = a o ad x. This completes the proof.
Applying Proposition 7 to the case of a commutative Lie algebra, it is seen
that every endomorphism u of a K-module can be extended uniquely to a
derivation of the symmetric algebra of this module; this derivation is derived
on passing to the quotient from the derivation of the tensor algebra which
extends it.
We again take a Lie algebra 3 over K and let D be a derivation of 9. We use
the earlier notation T, S, U, G. Let D1, D3 be the derivations of T, S which
extend D and let D" be the unique derivation of U such that a o D = D0 0 0'.
Since DU leaves the U" stable, Du defines on taking quotients a derivation
D5 of G. Since DH and D3 are derived from Dr when passing to quotients, the
commutative diagram (3) proves that DG can also be derived from Ds by the

24
 REPREENTATIONS § 3.1

homomorphism a) defined in no. 6. Iffurther K is a field ofcharacteristic 0, the

isomorphism of diagram (4) map one into another the restrictions of D1, D5,
Du, Do to S", S", U”, G". Hence the canonical irmrphiwn if S auto U map: D5
to Du.

9. EXTENSION OF THE BASE RING

Let g be a Lie algebra over K, T its tensor algebra, J the two-sided ideal of T
generatedbythex ®y — y ® a: —— [x,y] (x,ying) andU = TU. LetKlbea
commutative ring with unit element and a a homomorphism of K into K1
mapping 1 to 1. Then the tensor algebra of 9m) is canonically identified with
Tan)- Let J’ be the two-sided ideal om, generated by the
x! ®yl —y' ®xt _ [x',y']

(x', y' in gm”). Clearly the canonical image ofjm, in Tan) is contained in J’.
To see that it is equal to J', it suffices to show that, ifx' and 3/ denote two ele-
ments of 9m): x’ 8) y' — y’ 8) x’ — [1', y’] belongs to this image. New

X' =t®lu '=;%®W (x.,y,ing,>..,u,inK.);

whence
"I ®lll ‘11, ‘8" " [‘311'] = g: (”I ®!II ‘91 @‘t _ [‘uyll) 8 Nil!
which proves our assertion. Then it can be seen that Um, = (T/Jhxj) is
canonically identified with Taco/ll: the enveloping algebra of 90(1) is canonically
identified with Um) and the canonical mapping of Sam into is enveloping
algebra is identified with n- ® 1 (where 6 denotes the canonical mapping of g
into U).

§ 3. REPRESENTATIONS

l. REPRESENTATIONS
DEFINITION 1. Let g be a Lie algebra over K and M a K-module. A homomorphirm of
9 into the Lie Algebra 91(M) i: called a representation of g on the module M. An
injective representation is calledfail/$11. IfK i: afield, the dimion (finite or White) of
M aver K it called the dimmion qflhe representation. The momenta/ion x >—> ad 7: (ft;
on the K-madule 9 it called the adjoin! reprermlalion 1J9.
A representation of g on M is thus a K-linear mapping 9 of 9 into the
endomorphism module of M such that
P([x,y])-m = P(*)P(!)-'" — 9(y)9(x)~m
forallxeg,yeg, meM.

25
I LE ALGEBRAS

*Example. Let G be a real Lie group, 9 its Lie algebra and 0 an analytic
representation of G on a finite-dimensional real vector space E. Then the
corresponding homomorphism of 9 into 91(E) is a representation of g on E.,,
Let U be the enveloping algebra of9. Proposition 1 of§ 2, no. 1 defines a one-
to—one correspondence between the set of representations of g on M and the set
of representations of U on M. On the other hand we know (Algebra, Chapter
VIII, § 13, no. 1) that there is an equivalence between the notion of repre-
sentation of the associative algebra U and that of left U-module.
DEFINITION 2. Let g be a Lie algebra aver K and U it: enveloping algebra. A unitary
left module aver U i: called a left g-rrwdule, or simply a 9-madule‘
If M is a g-module and x e U, X” will denote the homothety of M defined
by x (cf. Algebra, Chapter VIII, § 1, no. 2).
A unitary right module over U is called a right g-module. Such a module is
identified with a left U‘l-module, that is (§ 2, no. 4) with a left g°-modu1e.
Let 4) be the principal antiautomorphism of U. If M is a right g-module, a
left g-module structure is defined on M by writing 4.”: = m.¢(a) for m E M
and a e U.
The notions and results of the theory of modules can be translated into the
language of representations:
(1) Two representations 9 and p' of g on M and M' are called similar or
isomorphic if the g-modules M and M’ are isomorphic. For this it is necessary
and sufficient that there exist an isomorphism u of the K-module M onto the
K-module M' such that
9'(~) = u . m) ., w
for all x e g.
(2) For all i E I, let p; be a representation of g on M,. Let M be the g-module
the direct sum of the 9-modules M,. There is a corresponding representation p
ofg on M, called the dim! sum of the p. and denoted by g p‘ (or p1 + - - - + 9,,
in the case of n representations 91, . . ., 9”). If m = ("10.5, is an element of M
and x E 9, then PM” = (91(3) we.“
(3) A representation 9 of g on M is called simple or irreducible if the asso-
ciated g-module is simple. It amounts to the same to say that there exists no
sub-K-module of M (other than {0) and M) stable under all the p(x), x E g. A
class of simple g-modules (Algebra, Chapter VIII, § 3, no. 2) defines a clan Qf
simple representatian: of g.
(4) A representation 9 of g on M is called semi-simple or completely reducible if
the associated g-module is semi-simple. It amounm to the same to say that p is
similar to a direct sum ofsimple representations or that every sub«K-module of

26
 REPRESENTATIONS § 3.1

M stable under the p(x) (x e g) has a supplement stable under the p(x) (x E g)
(cf. Algebra, Chapter VIII, § 3, no. 3).
(5) Let 8 be a class of simple representations of 9 corresponding to a class C
of simple g-modules. On the other hand let p be a representation of g on M.
The isotypical component MC ofspecies C of the g—module M (Algebra, Chapter
VIII, § 3, no. 4-) is also called the iralypical component of M afrpm'er 8. This com-
ponent is the sum of the sub-K-modules of M stable under the 9(x) and on
which the p(x) induce a representation of class 8; it is the direct sum of certain
of these submodules; if M0 is of length n, p is said to contain 8 n times. The sum
of the different M,J is direct; it is equal to M if and only if p is semi-simple.
(6) Let p, p’ be two representations of g. p’ is milled a mbnpmentatian (resp.
quotient reprerenlation) of p if the module of p' is a submodule (resp. quotient
module) of the module of 9.
Let M be a K-module. The zero representation of g on M defines on M a
g-module structure. With this structure M is called a trivial g-module.
Let M be a g-module. The quotient g—modules of the sub-g-modules of M are
also the sub-g—modules of the quotient modules of M: they are obtained by
considering two sub-g-modules U, U’ of M such that U D U’ and forming the
g-module U/U’. Then if all the simple modules of the above type are isomor-
phic to a given simple g-module N, M is called a pure g-module of specie: N.
If p and a are the representations of 9 corresponding to M and N, we also say
that p is pure cfrpeeies a.
Let M’ be a sub-g-module of M. For M to be pure ofspecies N, it is necessary
and sufficient that M’ and M/M’ be pure of species N. For the condition is
obviously necessary. Suppose that it holds and let U, U’ be sub- g-modules of
M such that U' C U and U/U’ is simple; let <1) be the canonical homomorphism
ofM onto M/M’; if MU) sé <1)(U’), U/U’ is isomorphism to d)(U)/¢(U’) and
hence isomorphic to N; if¢(U) = ¢(U’), then U C U' + M’, hence U/U’ is
isomorphic to a simple submodule of (U' + M’)/U’ and the latter module is
itself isomorphic to M’/(U’ n M’); hence U/U' is again isomorphic to N, so
that M is pure of species Ni
Henceforth let M be a g-module and suppose that the set of sub-g-modules
of M which are pure of species N admits a maximal element M’. Then every
submodule M” of M which is pure of species N is contained in M’. For
M’/(M’ n M”) and M’ are pure of species N, hence M' + M” is pure of
species N by the above and hence M' + M” C M'.
Suppose that the g-module M admits a Jordan-Holder series (Mao‘i‘n.
For M to be pure of species N, it is necessary and sufficient that Mn/Ml,
M1/M2, . . ., Mn_,/M,, be isomorphic to N; for the condition is obviously
necessary and its sufficiency follows immediately by induction on n from what
we have seen above.
PROPOSITION 1. Let g be a Lie algebra aver K and a an ideal q/g. Let M be a g-madule

27
I L“! ALGEBILAS

and N a simple n-module. Consider M a: an a-module and mppose that the set qfsuh—a—
modulu of M which an [21m qfrperies N admits a maximal damn: M’. Then M’ u a
mb-g-madule of M.
Let yeg. Let 4) be the canonical mapping of M onto M/M’ andf the
mapping m >—> (lg/Wm) of M’ into M/M’. It suffices to show thatf(M’) = {0}.
Let x e (I. Then, for In 6 M,
”mm-f0") = “India-m) = ‘l’lyuxu-m) + Ml’rsylu-ml'
Now [24, y] e a, whence ¢([x, y],..m) = 0; on the other hand,
¢(yuxfl'm) =f("u-'")-
Hence xM,Mr.f(m) = f(xMJn). It follows that f(M’) is a sub-o—module of
M/M’ isomorphic to a quotient of MI and hence pure of species N; hence
f(M’) = {0}-
COROLLARY. Let g be a Li: algebra over K and a an ideal of 9. Let M be a simple
g-maa'ule, qffinite length a: a K-madulo. Tiler: exist: :1 mph n-module N such that M
i: a pure a-madule qfspetie: N.
Since the a-module M is of finite length, there exists a minimal element
N in the set of sub-a-modules of M: it is a simple sub-a-module of M. The
largat sub—a-module of M which is pure of species N is therefore #{0} and
is a sub-g-module of M (Proposition 1) and is therefore identical with M.

2. TENSOR PRODUCT 0F REPRESENTATIONS

We have defined in no. 1 the direct sum of a family of reprsentations of 9.
We shall now define other operations on representations.
Let g), g, be two Lie algebras over K and M. a g,-module (i = l, 2). Let U.
be the enveloping algebra of g‘ and o‘ the canonical mapping of 94 into U4.
Then M‘ is a left Upmodule and hence M1 ‘81: M, has a canonical left
(U1 ®x UQ-module structure. Now U1 ®x U, is the enveloping algebra of
g x g, and the mapping (xv x2) >—> 61(x1) ® 1 + 1 ® 630:2) is the canonical
mapping of 91 X 9; into this enveloping algebra (§ 2, no. 2). Hence there
exists a (g1 X 92)-m0dUIe stmcture on M = M1 ®x M2 such that:
(l) (*1: x2)w(m1 ®mnl = (°1("1) ® 1 + 1 ® 52(xa))~(’”1 ® "12)
= «film-“1) ® ”'2 + ml ® (“Qua-ma)-
This structure defines a representation of 9, x g, on M.
If now 91 = g, = g, the homomorphism xr—> (x, x) of 9 into 5 x 9, com-
posed with the above representation, defines a representation of g on M and
hence a 9-module structure on M such that:
(2) x“. (ml ® ma) = (xMrrnl) (8) ma + m1 ® (xurma).

28
 REPRESENTATIONS on nonomonrmsm MODULES § 3.3

By an analogous argument we see that:

PROPOSITION 2. Let g be a Lie algebra over K and IVL a g-modulz (l S is it). On
the two! product M1 ®x M2 ®- - -® M," there exists one and only one g—module
mutiny: null that

(3) xM-(m1®”'®mn)=._Z,mi®"'®(xupmi) ®-~®mn

forallxeg, mleM,,...,mneM,..
The corresponding representation is called the tensor product of the given
representations of g on the M.
In particular, if M is a g-module, Proposition 2 defines a g-module structure
9

on each M, = Q9 M and hence on the tensor algebra T of M.

Formula (3) shows that, for all x e g, x, is the unique derivation of the algebra
T which extends x“. We know (§ 2, no. 8) that x; defines on passing to the
quotient a derivation of the symmetric algebra S of Mt Hence S can be con-
sidered as a quotient g-module of T and the xs are derivations of S.
Still more particularly, consider 9 as a g-module by means of the adjoint
representation of 9. Let U be the enveloping algebra of 3. By Proposition 7 of
§2, xM defines on passing to the quotients a. derivation of U which is just the
inner derivation defined by o(x) (a denoting the canonical mapping of 9 into
U). Then U can be considered as a quotient g-module of T. If K is a. field of
characteristic 0, the canonical isomorphism of S onto U is a g-module isomor-
phism (§ 2, no. 8).

3. REPRESENTATIONS 0N HOMOMORPHISM MOEULES

Again let 91 and g, be two Lie algebras over K and M‘ a gi-module (i = I , 2).
Let Ui be the enveloping algebra of g. and a‘ the canonical mapping of 9‘ into
Ul. Then M‘ is a left UI-module and hence 23(Ml, M,) has a canonical left
(U2 ® U2)-module structure. Now U‘} ®K U2 is the enveloping algebra of
g? x g, and the mapping
(*1) 1%) H 010%) ® 1 + 1 8 5202)
is the canonical mapping of g? X 92 into this enveloping algebra. Hence there
exists a (9‘: x gag-module structure on M = S’AMD M2) such that
(4) ((3‘): x2)M-") -m1 = ((°1("1) ® I + 1 ® 02(x2)).u).m1
= "((x1)Mn'ml) + (”aluyuimil
for all u E $K(M1, M2), in, G M1. This structure defina a representation of
g‘,’ >< g, on M.
If now g1 = 9, = g, the homomorphism x»—>(—x,x) of 9 into 9" x g,

29
I LIE ALGEBRAS

composed with the above representation, defines a representation of g on M

and hence a g-module structure on M such that
(5) (xM.u).m1 = xua.u(m1) — 1101.41.71“)

or
(6) xwu = In," — uxul.
Combining these results with Proposition 2, we see that:
Pnorosmon 3. Let g be a Lie algebra aver K and MK a g-moduk (l S i S n + l).
n
Let N be the K-madul: 2AM”. . ., Mn; Mn“) ufmultilitwar mapping: #1:]; M.
into Mn“. The" exist: on: and only one g—module drum”: on N such that
n
(7) (z...u)<m.,...,m.) = —.Zlu(m.,....x....m.,...,m.)
+ xMMI.u(m,,...,m,,)
forallxeg,ueNtzmtlm‘eMl (l S {g n).
In particular, let g be a Lie algebra over K and M a g—module and consider
K as a trivial g-module. Proposition 3 defines a g—module structure on
S’K(M, K) = M“. The corresponding repmentation is called the dual repre-
sentation of the representation at >—> x“. We have:

(3) (I‘m-fxm) = —f(*u‘”‘)

for all x E g,fE M“, m e M. In other words:
(9) x“. = —‘xM.
When K is a field and M is finite-dimensional, the g-module M is simple
(resp. semi-simple) if and only if the g-module M" is simple (rcsp. semi-
simple).
PROPOSITION 4. Le! M1, M2 be two g-modules. The canonical K-linzar mapping:
(Algebra, Chapter II, §4, no. 2, Proposition 2 and no. 1, Proposition 1):

M; 93),. M2 l) yK(M,,M.), YK(M1,M;‘) —*) (M1 8),. M,)*

(where the strand is bijectiue) are g—module homomorphism.
We write
N = M: ® M2, P = 5!’(M1, M3), Q = $(Ml, M;), R = (M1 ® M,)*.

30
 EXAMPLE § 3.4

Then, for x e 9,fe Mf, m1 5 M], m2 6 Ma,

((¢*n)(f® 1:12)) J”: = (¢(xu:f® "'2 +f® ”Mama” M1
= (“Mi/i ”11>”: + (f) "hymn“:
((‘r‘l’)(f® m2))~”‘1 = "Ma(‘l’(f® m2)-”‘1) " <l’(f® ”’2)("Mlmx)
= (f; “Dining " <f: "uimi>”‘2
and hence ox" = xpdz. On the other hand, for x e g, u e $(Mh M:), m1 6 M1,
ma 6 M,:
(4%")("11 ® m3) = (“do-"‘1, "12) = 9315””: — “thumb in“)
(*RWX'": ® m2) = —<'~i"‘a ”Mimi ® ”‘2 + m1 ® Kayne)
= ‘(w’ulmn m2) — (um!) ”MIWD
and hence 4m; = ataxia, which completes the proof.
The g-modules $(M1, Mg) and (M1 ® M,)‘ are identified under the iso-
morphism 44. If M1 and Mn have finite bases, 4) is an isomorphism (Algebra,
Chapter II, § 4-, no. 2, Proposition 2), which allows us to identify the g-modules
Mr ® M2 and 5.4M], M2); in that case, we can therefore identify the g-
modulu M: ® M}, 2’(M,, Mg) and (M1 (8 M2)*.

4. EXAMPLES
Exampl: I. Let g be a Lie algebra over K and M a 9-module. The g-module
structure on M and the trivial g-module structure on K define a g-module
structure on the K-module N = 2’(M, M; K) of bilinear forms on M. Then
(10) (In-BXMM') = ~B(xn-ma m') + lwm’)
forallxe g, m, m' in M,BEN. IfflisagivenelementofN, theset ofxegsnch
that xwfl = O is a subalgebra of 9.
Let M be a K-module and B a bilinear form on M. By the above, the set of
x e gI(M) such that
B(x.m, m’) + fl(m,x.m’) = 0
for all m E M and m' e M is a Lie subalgebra of gl(M). Suppose that K is a
field, M is finite-dimensional and B is non-degenerate. Then every xe gl(M)
admits a left adjoint x* (relative to p) which is everywhere defined and the
subalgebra in question is the set of x E gl(M) such that x“ = —x. By this pro-
cess we can construct two important examples of Lie algebras:
(3) Take M = K" and
WE], - . ~, Eu): (7)» - - and) = 5m + ~ ~ ~ + 5m..-
We canonically identify gl(K") with Mn(K). Then the Lie algebra obtained is
the Lie algebra of skew-symmetric matrices. *(When K = R, this algebra is
the Lie algebra of the orthogonal group 0(n, R)).“

31
l LIE ALGEBRAS

(b) Take M = K“ and

“(an - ~ -: Em): (’41: - - -: 712m» : 51min — Thin-+1 + ' ' ' + Em‘flzm — 'flmEm-
The matrix of p with respect to the canonical basis of K2" is the matrix
0 1,,l _ A B)
be the matrix with respect to the canonical
(—1", 0)‘ I“ U‘ (c D
basis of K” of an element u of 91(M) (A, B, C, D lying in MM(K)). By formula
(50) of Algebra, Chapter IX, § 1, no. 10, u* has with respect to the same basis V
the matrix
0 —-I,,l A ‘C)( 0 1,.) _( ’D -—'B).
(In, 0) ‘B ‘D —I,,I 0 _ —‘C ‘A
The condition u* = —u is therefore equivalent to the conditions
= -—‘A B = ‘B C = 'C.
*When K = R, the Lie algebra obtained is the Lie algebra of the symplectic
group Sp(2m, R).‘
Example 2. We preserve the notation of Example 1.
The g-module structure on M defines on the K-module P = JAM, M) of
endomorphism of M a g—module structure. By (6), for all x E g and u e P:
(11) 569.14 = [Jew u] = (adxM).u
where ad xM denota the image of It“ under the adjoint representation of
gl(M). In other words:
(12) x1, = ad It“
in $(.Z’(M, M)) = 3’(gl(M)).
5. INVARIANT ELEMENTS
DEFINmON 3. Let g be a Lie algzbra and M a g-madule. An element in e M it pulled
invariant (with mpett to the g-madule :tnuhm an M or with respect to the tonerfionding
itfimntatian ofg) zfJn = Ofir all x e 9.
*Let G be a connected real Lie group, 9 its Lie algebra, 9 an analytic
representation of G on a finite-dimensional real vector space E and p the
corresponding representation of g on E. Let 7/: e E. The element 7» is invariant
with respect to p ifand only if0(g) .m = m for all g e G. This justifies the use of
the word “invariant’ia:
Example 1. Let M, N be two g-modules and P : $14M, N). For an elementf
of P to be invariant, it is necessary and sufficient, by (6), thatf be a homomor-
phism of the g-modulc M into the g-module N. In particular, if M = N and
x“ = x" for all x6 9, fis invariant if and only iff is permutable with the
am.

32
 lNVARIANT BILIN'EAR FORMS §3.6

Example 2. Let M be a K—module with a finite basis. If M has a g—module

structure, 3(M, M) and M* (8) M have g-modnle structures and the canonical
mapping of M* ® M into E(M, M) is a g~module isomorphism (Proposition
4). As 1 e $(M, M) is obviously an invariant (of. Example 1), the correspond-
ing element a of M* ® M is an invariant. If (e,)1‘.‘,, is a basis of M and
n
. . _ *
(4)1‘,“ 15 the dual basxs, we have u — 2; el ® ep

Example 3. Let M be a g—module. Let [S be a bilinear form on M and f the

corresponding element of E(M, M*). For 13 to be invariant, it is necessary and
sufl'icient thatf be a g-module homomorphism (Proposition 4- and Example 1).
Suppose that K is a field and that dimK M < +00. A non-degenerate invariant
bilinear form {5 on M defines an ixamnrphikm of the g-module M onto the g-
module M* and hence an isomorphism of the 9-module M ® M onto the g-
module M“ (8 M. Thus, by Example 2, giving 9 defines canonically an invariant
element a in the g-module M 8 M, which can be constructed as follows:
let (e.)1“‘,l be a basis of M and 991‘.“ the basis of M such that

@(el, 4) = 8,,; then e = 2; e‘ ® ez.

PROPOSITION 5. Let g be a Lie K-algehra, b an ideal q, p a representation (ya on M
and p’ the restriction 0f 9 to b. Then the re! N If element: of M invariant with rexpeet
ta 9' i: Jtable under 9(9).
Let n EN andy e 9; for all x e b, [x,y] e b and hence

WNW)" = 9([x,y])n + 9(y)o(t)n = 0;

hence p(y)n EN.
PROPOSITION 6. Let M be a :emi-Jimfle g-module. Then the rubmadule Mo Qf in-
variant element: of M admitx one and anly one supplement .ttable under the xM, namely
the .mbmaa'ule M1 generated by the aunt (x e g, m e M).
Let M’ be a submodule of M which is stable under the xM and a supplement
of MD in M. For all m e M, m = mo + m' with mo 5 Mo, m’ s M’, and hence
m = m’ e M’. Hence M1 C M'. Let M, be a submodule of M’ stable
under the am and supplementary to M1 in M’. For all m e M2,
atMrnEM2 n M1 = {0}
for all z e 9, hence m E M“ and hence m = 0. Hence M, = {0}, which proves
that M1 = M'.

6. INVARIANT BILINEAR FORMS

Let g be a Lie algebra over K. The adjoint representation of g on g and the
zero representation of g on K define a 9-module structure on the K—module

33
 I LIE ALGEBRAS

N = 3’ (g, 3; K) ofbilinear forms on g. Briefly we say that a bilinear form p on g

is invariant if it is invariant under the representation at n—> or“. By formula (10)
the necessary and sufficient condition for this to be so is that:

(13) MPH], 2) = 90‘, [.1]: 2])

for all x, y, z in 9.
Now let b be the Lie algebra of derivations of g. The identity representation
of b and the zero representation of b on K define a representation D >—> DN of b
on N. Briefly we say that a bilinear form on g is completely invariant if it is in-'
variant under the representation D ~—> D". A completely invariant bilinear
form is invariant. For a bilinear form [5 on g to be completely invariant, it is
necessary and sufl'icient that:

(14) B(Dx,y) + {$061M = 0

for all x, y in g and D e b.
PRorosmou 7. Let g be a Lie algebra, {3 an invariant symmetric bilinearfirm on g
and a an ideal q.
(a) The orthogonal a’ of a with respect to B is an ideal of g.
(b)1fais" "andBis "' ' ,a’isl istic.
(c) If a is nan-degenerate, a n a’ is tomnudatiue.
Let D be a derivation of 9. Suppose that a is stable under D and that
{3(Dx, y) + {3(x, Dy) = 0 for x, y in 9. Then 1 e a' implies Dz e a’, since, for all
t5 a, Dt e a and hence [3(Dz, t) = —B(z, Dt) = 0. Thus a’ is stable under D.
This establishes (a) and (1)).
Now let b be an ideal of g and suppose that the restriction of B to b is zero.
For x, y in b and 269, fl([x,y], z) = (30:, [g], 2]) = 0, for [y, 2] Sb. Thus
[17, b] is orthogonal to g. If p is non-degenerate, b is therefore commutative.
This result applied to a n a' proves (c).
DEFINITION 4-. Let g be a Lie K-algebra and M a g-module. Suppose that M, con-
sidered as a K-module, admits a finite basis. The bilinearfirm associated with the g-
module M (or with the eamsponding representation) is the symmetric bilinear form
(x, g) >—> Tr(xMyM) on g. If the representation in question is the adjoint representation,
the associated bilinearfurm is called the Killingfirm of g.

PROPOSITION 8. Let 9 be a Lie algebra and M a g-module. Sapose that M, considered

as a K-rnodale, admits a finite basis. The bilinearfonn associated with M is invariant.
For a, y, z in 9, we have:

Tr<[x)y]MZM = Tr("MyMZM) — Tr(!lM"MZM) : Tr("M!/MZM) — TT(*MZMyM)

= Tr(xM[% Z]u)~
PROPORTION 9. Suppose that K is afield and that the Lie algebra g isfinite—dimearional

3 4»
 CASIMIR ELEMENT § 37

over K. Let a be an ideal q, B the Killingform ofg and p’ the Killingform qf n. Then
3' is the restriction (f [3 to a.
Let a be an endomorphism of the vector space g which leaves a stable. Let u
be the ratriction of a to u and w the endomorphism of the vector space g/u
derived from a when passing to the quotient. Then Tr a = Tr v + Tr w as is
seen by taking a basis (xi, . . . , x") of g ofwhich the firstp demerits form a basis
of G. Then let x E a, y e a and apply the above formula to the case where
u = (adv x)(adiy). Thenu = (add x)(aduy) and w = O. Henccfl(x,y) =fl'(x,y).

PROPOSITION 10. Suppose that K is afield and that the Lie algebra g isfinite-dimen-
:ianal over K. The Killingfirm {5 q is tompletely invariant.
Let D be a derivation of 5. There exists a Lie algebra {5' containing g as an
ideal of codimension 1 and an element no of 9' such that Dx = [5%, x] for all
xsg (§1, no. 8, Example I). Let B’ be the Killing form of 9'. For x, y in 9,
[We may) = w, DMD, that is MUM) + rm, Dy) = 0. Now the res-
triction of {1’ to g is [3 (Proposition 9). Hence the proposition.

7. CASIMIR ELEMENT
PROPOSITION 11. Let g be a Lie algebra over afield K, U its enveloping algebra, b a
finite-dimenrianal ideal of g and B an invariant bilinearfirm on 9, whose restriction to b
is nan-degenerate. Let 00“.“, (4)191“ be two base: of I) such that we” e;) = 8”.
n
Then the element 5 = ”:1 ele,’ of U belong: to the tentre of U and is independent to the
choice of basis (e,).
For x e 9 let x» be the restriction to I) ofad. 1. Then a: r-> in, is a representation
of g on the vector space I) and the ratriction [5' of B to b is invariant under this

representation. By no. 5, Example 3, the tensor ”:1 e‘ «8 e.’ is independent of the

choice of basis (e‘) and is an invariant element of the tensor algebra of I). It is
also an element of the tensor algebra T of 9, which is invariant for the repre-
sentation derived from the adjoint representation of 9. Its canonical image in
U, that is c, is therefore independent of the choice of basis (e.) and is an in-
variant for the representation of g on U considered at the end of no. 2. This
element is therefore per-mutable with every element of g and therefore belongs
to the centre of U.
When 13 is the bilinear form associated with a g-module M, the element 4‘ of
Proposition 1 l is called the Carimir element associated with M (or with the cor-
responding representation). This element exists if the restriction of 9 to I) is non-
degenerate.
PROPOSITION 12. Let g be a Lie algebra over afield K, b an ideal q 1y"finite dimmiau

35
1 LE AmEBRAS

n and M a g—module offinite dimnm'on 0m K. Le! a b: the Crm’mir elemenl ((1:d to

exit!) mmduted with M and b.
(a) Tr(cM) = n. .
(b) IfM is simple and n is not divisible by the characteristic afK, an {J an auto-
nwrphirm if M.
In the notation of Proposition 11,

Tran) = ,Z, T.((.,),,(.;),,) = Z we. a0 = n.

i-1 .
Hence, if n is not divisible by the characteristic of K, c“ aé 0. On the other
hand, as 6 belongs to the centre of U, cm is permutable with all the x”, z E g.
If further M is simple, a“ is therefore invertible in 2’(M) (Algtbm, Chapter
VIII, §4, no. 3, Proposition 2).

8. EXTENSION 01’ m BASE RING

Let Kl be a commutative ring with unit element and ({> a homomorphism of K
into K1 mapping 1 to 1. Let g be a Lie K—algebra, U its enveloping algebra and
M a. left g—module, that is a. left U-module. Then Mm) has a canonical left
Umv-module structure and hence a left gmrmodule structure. Let p and gum
be the representations of g and gun) corresponding to M and MK“: pm, is said
to be derived from p by extending the bare ring and the results ofAlgebra, Chapter
VIII, § 13, no. 4 can be applied. If xeg, pm,(x) is just the endomorphism
9(1) ® 1 of Man) = M ®x K1.
Suppose that K is a field, that K1 is an extension of K and that 4) is the
canonical injection of K into K1. Let V and V' be vector subspaces of M. Let a
be the vector subspace of 9 consisting of the x e 9 such that p(x) (V) C V'. Let
a’ be the vector subspace of Sun) consisting of the x’ 6 Gun) such that
9(K‘)(x’)(V(x,)) C Vin). Then a’ = am). For clearly am, c (1’. Now let
u
x’ e a’. We may write x’ = ‘21 Am, where the x, are in g and the A. are elements
of K1 linearly independent over K. For all u e V, p(x’) .u e Vim: that is
..
2; 1.90:.) .14 6V5“), whence p(::.).u eV’, hence x, e a and x' E “<e This
shows that a' = “urn- In particular, the mztrz of 901;) is derived from the
centre of g by extending K to K,: it suffices to apply the above to the adjoint
representation of g. It follows that grains.» = (’gpgxxl, for all p. Similarly, let I)
be a subalgebra of g and II the mmaliur of b in 9. Then the normalizer of
Dan) in Sun) is “(Kn-
Let K, K1, 9, p, M be as in the last paragraph. Let b be a vector subspace ofg
and W a vector subspace of M. Let V be the vector subspace of M consisting
of the m e M such that p(b) .m C W. Let V’ be the vector subspace of Mom
consisting of the m' e Mix.) such that 9(Ki)(blxn) .m' C Wm). As above it is seen

36
 EXTENSION 01? THE BASE RING § 3.8

that V’ = Vam- In particular, the vector subspace of invariant: of Man) is

derived from the vector subspace of invariants of M by extending the base
field from K to K1.
Let K, K1 and <1) be as at the beginning of this no. Let g be a Lie K-algehra
and M and N g-modula. If M and N are isomorphic 9-modules, Mm) and
Non) are isomorphic gum-modules. Conversely:
PROPOSITION 13. Let K be afield, K1 an extension 13fK, g a Lie K—algebm and M, N
twa g-madule: bffinite dimension mm K. If Man and Nm, are isomorphic 9m)"
modules, M and N are isomorphic 9-modules.
The proof is in two steps.
(1) Suppose first that K] is an extension of K offinite degree 11. Let U be the
enveloping algebra of 9, so that the enveloping algebra of 9m) is
U0“) = U ®K K1 (§ 2, no. 9). As Mai], and Nm) are isomorphic as Um)‘
modules they are a fom'on' isomorphic as U-modules; but as U-modules they
are respectively isomorphic to M” and N". Now M and N are U-modules offinite
length; M (resp. N) is therefore the direct sum of a family (Pi‘)1<c<p (resp.
(Q?)1‘,‘a) of submodules such that the P. (resp. Q) are indecomposable and
two P4 (resp. (2,) of different indices are not isomorphic (Algebra, Chapter
VIII, §2, no. 2, Theorem 1). Then M” (resp. N") is isomorphic to the direct
sum of the PI"! (resp. QE'I); it follows (lac. tit.) that p = q and that after per-
rnuting the Q, if necessary nr. = m. and P, is isomorphic to Q,‘ for 1 < i < p,
hence M is isomorphic to N.
(2) General care. Let P be the g-module $4M, N) and ce subspace of
invariants of P that 1s the set of homomorphisms of the g—module M into the
g-moclule N. In the 90:1,-module 2111(Mmfi, Nam) = (3AM, N))(K‘,, the
subspace of mvariants is Q“ ). The hypothesis that Mm ) and Nor.) are iso-
morphic implies that M and N have the same dimension over K and that there
exists in Q51, an element g which is an isomorphism of Man) onto Nmr Let
(fh . . .,f¢) be a basis oover K and choose bases ofM and N over K. If
A!

7w 5 K1 for 1 S k S d, the matrix off = 1:21 )1]:fk with respect to these bases
has determinant which is a polynomial D0,, . . ., Ad) with coefficients in K.
Whenf = g, this determinant is non-zero and hence the coefficients of D are
not all non-zero. Therefore, if Q is the algebraic closure ofK, there exists (since
{I is infinite) elements W E Q (l < k < 11) such that D(u1, . . ., M) at 0
(Algebra, Chapter IV, § 2, no. 5, Proposition 8). If K2 is the algebraic extension
11

of K generated by the [1,, (1 S I: S d), it follows that g1 nkf,‘ is an isomor-

phism of Mam onto Nam; but K3 is of finite degree over K (Algebra, Chapter
V, § 3, no. 2, Proposition 5) and hence M and N are isomorphic by the first
part of the mment.
Again let K, K1 and <1: be as the beginning of this no. Let p be a representation

37
 I LIE ALGEBRAS

of g on a K-module M with a finite basis (x1, . . . , it“). Then the bilinear

form on Ham associated with pm) is derived from the bilineax fonn associated
with p by extending the base ring to K1 (for, if u e $K(M), a has the same
matrixwithrespect to (1:1, . . ., 1:“) 3514 ® lwithrespect to (x1 ® 1, . . ., at“ ® 1)
and hence a and n ® 1 have the same trace). In particular, if the K-module g
has a finite basis, the Killingform of gum is derived from that of g by extending
the base ring to K1.

§4. NILPOTENT LIE ALGEBRAS

Recall that henceforth K denotes a commativefield. In the rest of the chapter the Lie
algebra: are assumed It: befinite-dimensional over K.

1. DEFINITION OF NILPOTENT LEE ALGEBRAS

DEFINITION l. A Lie algebra 9 is called nilpatent éf there exists a damning finite
sequent: afideal: (80mm: fife with 90 = 9, 9, = {0}, Inch that [9, 9.] C safer
0 g i < p.

A commutative Lie algebra is nilpotent.

PROPOSITION 1. Let g be a Lie algebra. Thefnlluwing condition: are equivalent :

(a) g i: nilpotent;
(b) “*9 = {0}f0r WWW/17g: 1:;
(c) «kg = gfor mfieiently large k;
((1) there exist: an integer k such that aclxl mad):2 0- - -u ad xk = 0 for all
elements x1, x2, . . ., xk in g;
(e) there exist: a detreasing sequence ofideal: (900 g“ q with go = g, g,| = {0},
null that [9, 94] C 3,“ anddim gi/gH1 =1for0 g i < 71.
If W‘g = {0} (rap. K; = 5), clearly the sequence ((19, . . ., W‘g (resp.
figmfi‘dg, . . .,%’ug) has the properties of Definition 1 and hence g is
nil-
potent. Conversely, suppose that there exists a sequence (mafia with the
properties of Definition 1. It is seen by induction on n that g‘ D ?'*’g and
g,_, C ‘tfig. Hence K“‘g : {0} and 9,9 = 9. We have thus proved that con-
ditions (a), (b) and (c) are equivalent. 0n the other hand, fi‘g is the set of
linear combinations of elements of the form

["n [*2: - ~ -, [xi—2) [*4—1: xtll- - -ll

where :41,- x3, . . . , atl run through 9. Hence conditions (b) and (d) are equivalent.
Finally, If there eXIsts a sequence (91)“: 9 of idmls with the properties of

38
 ENGEL’s THEOREM §4.2

Definition 1, there exists a decreasing sequence (Menu ofvector subspaces of

g ofdimensions n, n — l, n —- 2, l . ., 0 and a sequence ofindices
io<i1<-~-<i,
with 90 = bin, 91 = b.3- - -, 9., = be; then as [9, 34.} C him the b: are ideals
and [g, h] C b,“ for all 1'. Hence conditions (a) and (e) are equivalent.
COROLLARY 1. The centre y'a non-zero nilpateni Lie algebra i: rum-zero.
COROLLARY 2‘ The Killingform qf a nilpalent Lie algebra ix zero.
For all x and y in a nilpotent Lie algebra ad 1: 0 ad y is nilpotent and hence of
zero trace. '
PROPOSITION 2t Subalgebms, quotient algebra: and central extemiom of a nilpntent
Lie algebra are nilpotmt. A finite product of nilpotent Lie algebrar i: a nilpotent Lie
algebra.
Let g be a Lie algebra, 9’ a subalgebra ofg, I; an ideal of g, I = 9/!) and b the
canonical mapping ofg onto 2. Hg is nilpotent, then W‘s = {0) for some integer
1:, hence W‘g' C W‘s = {0) and W‘l = 4%n) = {0) and hence 9' and t are
nilpotent. If I is nilpotent and b is contained in the centre ofg, then ‘6’”! = {0}
for some integer k, hence W‘g C I) and therefore %’**15 C [1), g] = {0}, so
that g is nilpotent. Finally, the assertion concerning products follows for
example from the assertion (a) a (d) of Proposition 1.

Definition 1 and Proposition 2 show that nilpotent Lie algebras are pre-
cisely the ' L " ' J from ive Lie ' L bya
of central extensions.

PROPOSITION 3. Let g be a nilpotent Lie algebra and b a .mbalgebra q dirtinetfiam g.

The narmalizer of b in g i: distinclfiom b.
Let k be the greatest integer such that W‘s + 1) ye I). Then
[m + M] c ‘5’”“9 + b Cb
and hence the normalizer of b in 9 contains “(kg + b.

2. ENGEL’S THEOREM
Lemma 1. Lo! V be a vector xpare over K. If x i: a nilpalent :ndumorphirm qf V, [he
mapping y H [1, y] If 3’(V) inlu E (V) ix nilpalzut.
Iff denotes this mapping, f"(y) is a sum of terms of the form 1 x‘yx’ with
i+j = m. Ifx" = 0, thenf“"(y) = Ofm' ally.
THEOREM 1 (Engel). Let V be a vector space over K and g a finite-dimensianal rub-
algebra 13f 91(V) whore elements are nilpatenl endumorphiamr af V. If V aé (0}, there
exirh'u aé 0inVruchthalx.u = Oforallxeg.

39
l LIE ALGEB‘RAS

The proof proceeds by induction on the dimension n of g. The theorem is

obvious if n = 0, Suppose that it is true {or algebras of dimension < a.
Let b be a Lie subalgebra of g of dimension m < a. If): E I), and ad, 5: maps
1) into itself and defines on passing to the quotient an endomorphism a(x) of
the space g/b. By Lemma 1 ad,' at is nilpotent and hence 6(x) is nilpotent. By
the induction hypothesis there cxism a non—zero element of 51/!) which is
annihilated by all the 6(a), 1: e I). It follows that b is an ideal in a certain
(m + l)-dimensional subalgebra of 9.
We conclude (by iteration starting with b = {0}) that g has an ideal I) of '
dimension n — 1. Let aeg, a $1). We again use the induction hypothesis:
the u E V such that x.u = 0 for all x e 1) form a non-zero vector subspace U of
V. This subspace is stable under a (§ 3, no. 5, Proposition 5). Since a is a nil-
potent endomorphism of V, there exists a non-zero element of U which is
annihilated by a and hence by every element of g.

COROLLARY 1. Fur a Lie algebra g to be nilpateat, it it neeexxary and .tuflieieat that,far

all x e 3, ad x be nilpotent.
The condition is necessary (Proposition 1). Suppose that its sufficiency has
been proved for Lie algebras of dimension <7: (I: aé 0). Let 5 be an n-dimen-
sional Lie algebra such that, for all x e 9, ad x is nilpotent. Theorem I, applied
to the set of ad to (x e g), proves that the centre c of g is non-zero. Then g is a
central extension of the Lie algebra g/c, which is nilpotent by our induction
hypothesis. The proof is completed by applying Proposition 2.
COROLLARY 2. Let g be a Lie algebra and I) an ideal of 9. Suppose that g/b i: nilpbten!
and that, for all x E g, the restrietion afad x to I} i: ailpaleat. Then 9 is nil/intent.
Let x E 9. As 9/!) is nilpotcnt, there exism an integer k such that (ad x)"(g) C b.
By hypothesis there exists an integer Ic’ such that (ad x)”'(b) = {0}. Hence
(ad x)“*"’(g) = {0}. Corollary 2 is hence a consequence of Corollary 1.
COROLLARY 3. Let V be a vector space and g afinile-dimeasiaaal wbalgebra of gl(V)
whose elements are ailpbtent endomrphirm: of V. Tllea g i: a nilpolent Lie algebra.
This follows immediately from Lemma 1 and Corollary 1.
Example. The algebra "(71, K) (§ 1, no. 2, Example 2 (3)) is nilpotent.

3. THE LARGEST NILPOTENGY IDEAL OF A REPRESENTATION

Lemma 2. Let g be a Lie algebra, a an ideal (y g and M a Jimple g-module. If, for all
x e a, Jr” is nilpbtent, than at” = Ofor all x E a.
Let N be the subspace of M consisting of the m e M such that x“. m = 0 for
all z E a. By Theorem 1, N aé (0). On the other hand, for ally E g, N is stable
under y“ (§ 3, no. 5, Proposition 5). Hence N = M, which proves the lemma.
Lemma 3. Let g be a Lie algebra, a an ideal «f9, M a g—madule affiuile dimension aver

40
 THE. mom mum-nanny IDEAL or A REPRESENTATION § 4.3

K and (M00 <1 ‘n a Jordan-Hitlder series of the g-rnodule M. Theflllowing conditions

are equivalent:
(a) fir all x e a, x“ is nilpotent;
(b) for all x E a, at“ is in theJaeabron radical ofthe assotiatioe algebra A generated
by l and they” where]; E g;
(c) for all x e a,
“M(Mo) c Mn "M(M1) c M2: - - ~) xM(Mn—1) C Mm
If these conditions arefalfilled, a is orthogonal to g with respect to the bilinearfann
associated with the g-module M.
(b) :> (a): as A is finite-dimensional over K, the Jacobson radical of A is a
nilpotent ideal (Algebra, Chapter VIII, § 6, no. 4, Theorem 3) and hence every
element of this radical is nilpotent.
(a) :- (c) 2 each QI = Mi/M.+ 1 (0 S i < n) is asimple g-module. For allx E a,
the endomorphism xq‘ (which is derived from x“ by restricting to M. and
passing to the quotient) is nilpotent if condition (3) holds and hence zero by
Lemma 2; in other words, xu(M‘) C MI+1~
(c) => (b): suppose condition (c) holds; let x 6 a and z s A. Then
z(M,) C M, (0 S i < n) and hence (#M)“(M) = {0}; thus An:M is a left
nilideal of A and hence is contained in the Jacobson radical of A (Algebra,
Chapter VIII, §6, no. 3, Corollary 3 to Theorem 1).
Finally, suppose conditions (a), (b) and (c) hold. Let xe a and y e 9. We
have just seen that nM is nilpotent and hence Tr(nM) = O, which proves
the last assertion of the lemma.
PROPOSITION 4-. Let g be a Lie algebra, M a g—module offinite dimension over K and A
the assotiatiue algebra generated by l and the set ofxu (x e g).
(a) The ideals a of 9 Mb that x” is nilpotentfor all x e a are all eontained in one of
them, n.
(b) The ideal n is the set ofx e 9 such that it“ belongs to the Jacobson radical of A.
(c) Let (Mou‘fl be a Jordan-Holder series of the g-module M; then I1 is also the
set ofx E 9 such that ("lunMHI = Ofir all 1'.
(d) n is orthogonal to g with respect to the bilinearfonn associated with p.
The set oh: 6 9 such that 3M belongs to thejawobson radical ofA is obviously
an ideal of g. The proposition then follows immediately from Lemma 3.
DEFINITION 2. The ideal n of Proposition 4 is rolled the largest nilpolenty idralfor the
g-module M or the largest nilpotenoy ideal If the corresponding representation.
Clearly n contains the kernel of this repracntation. It equals it when M is
semi-simple (Proposition 4 (c)), but not in general. It should be noted that an
N

element of x of 9 such that arm is nilpotent does not necessarily belong to n.

We also note that a particular case of Lemma 3 immediately gives the
following result:

41
I LIE AmEBRAS

PROPOSITION 5. Let V b: a motor spice offinite dimension n over K and g a Lie rab-
algebra of gl(V) whose elements an nilpotmt mdomorphimu cf V. Then there exists a
decreasing some of vector subspaces V0, V1, . i . , V" of V, of dimensions
a,» — l,...,0,.mchthalx(V.) C V,+,j‘orallxegandi = 0,1,...,n — 1.

4. THE LARGEST NILPOTENT IDEAL OF A LIE ALGEBRA

Let g be a Lie algebra and a an ideal of 9. For a to be nilpotent, it is necessary
and sufficient that, for all x E 0, ad. 1: be nilpotent; the condition is obviously.
sufiicient and is necessary, for, if a is nilpotent and x e a, ad, x is nilpotent and
ad. x maps g into 0, hence ad, x is nilpotent. Then Proposition 4 applied to the
adjoint representation of 9 gives the following result:
Pnoposmon 6. Let g be a Li: algzbra and E the associative :ubalgchm of 3’ (9) gm:-
mled by l and the ad, 1: (x E 9). Let R be the Jacobson radical «f E.
(a) Th: m n tfy e 9 such that ad, y e R is the larger! nilpolmt ideal of 5.
(b) It is orthogonal to 9 under the Killingfarm.
It should be noted that 9/" can have non-zero nilpotent ideals.

5. EXTENSION OF THE BASE FIELD

Let g be a Lie K-algebra, KI an extension of K and 9’ = gm“. As
W‘g’ = (W‘gxxn, g is nilpotent if and only ifg' is nilpotent.
Let M be a g-module of finite dimension over K, n the largest nilpotency,
ideal for M and M’ = Mum. Let (Mdaum be aJordan-Hfilder series of the
g-modulc M‘ Then xu(M‘) C M“,1 for all i and all x e n, hence

xir((Mi)<x.)) c (Ml+1)(Kx)
for all i and all x’ e um); hence xfil is nilpotent for x' e "am so that “am is con-
tained in the largest nilpotency ideal n’ for M'. We shall now see that, if K1
1‘: separable over K, then n' = "(Kw Let E be the associative K-algebra generated
by l and the cm (x E g), E' the associative K-algebra generated by l and the
x“: (x' e g') and R and R' the jacobson radicals of E and E’. The algebra E’ is
canonically identified with Eam- Then R’ = Run) (Algebra, Chapter VIII,
§ 7, no. 2, Corollary 2 (c) to Proposition 3). Then let y' E n’ and write

y’ = I; My” where the y. are in g and the A, 6 K1 are linearly independent

.-
over K. Then yin: = l; Mm)“, and yh. e R' = Run)- Hence (y‘)M e R and
therefore m e n for all i. It follows that y’ e n(Km whence n' C "am-
In particular, lf KI is separable over K, the largest nilpotent ideal of Sam is
derived from that of g by extending the base field from K to K1.

42
 DEFINITION or SOLVABLE LIE menus § 5.1

§5. SOLVABLE LIE ALGEBRAS

Recall that K henceforth denotes afield of characteristic 0 and that all Lie algebras are
assumed to befinite—dimensional over K.T

!. DEFINITION OF SOLVABLE LIE ADGEBRAS

DEFINITION 1. A Lie algebra g is called solvable if its kth derived algebra 9*9 is zero
fir saficiently large k.
A nilpotent Lie algebra is solvable.

PROPOSITION 1. Subalgebras and quotient algebras ofa solvable Lie algebra are solvable.
Every extension ofa solvable algebra by a solvable algebra is solvable. Eueryfinile product
of solvable algebras is solvable.
Let g be a Lie algebra, 9' a subalgebra, b an ideal of g, I = g/b and b the
canonical mapping of 9 onto !. If g is solvable then 9kg = {0} for some integer
k, hence 9’? C 9kg = {0}, 9"! = Mflkg) = {0} and hence 5’ and I are
solvable. If I) and I are solvable there exist integers s, t such that

9‘0 = 9‘! = (0};

then Q‘g c 9, hence 9“”g = 9'(9‘g) C 9'!) = {0} and g is solvable. The last
assertion follows from the second by induction on the number of factors.
PROPOSITION 2. Let g be a Lie algebra. Thefollowing conditions are equivalent:
(a) g is solvable;
(b) there exists a decreasing sequence 9 = go 9 g, D - - - D 9,, = {0} qfideals ofg
such that the algebras gl/gHI are commutative (i = 0, l, . . ., n — l);
(c) there exists a decreasing sequence 9 = go 3 9’13-3 9,: {0} ofsubalgebras
qsuch that 9"“ isanidealofg; and 91/9,H iscornrnutative30's;= 0, l,.
(d) there exists a decreasing sequence 9—= 53 D 913 (=0) ofsubalgebras
q such that g!“ is an ideal ofgqcodirnension l (i = 0,1,9“ ., q — l).
(a) => (b): it suflices to consider the sequence of derived ideals of 9.
(b) :> (c): this is obvious.
(c) => ((1) : suppose that condition (c) holds; every vector subspace of g; con-
taining 91“ is an ideal of 91, whence immediately (d).
(d) => (a): this follows immediately from the fact that an extension of a
solvable algebra by a solvable algebra is solvable.

1L The reader will note that the hypothesis on the characteristic of K is not
used in nos. 1 and 2 of this paragraph.

43
 LIE ALGEBRAS

Example; afrolvahle Lie algebra:

I. Let 5 be a 2—dimensional vector space over K and (:1, ea) 2. basis of 9. There
exists one and only one alternating bilinear multiplication (x, y) »—> [x, g] on 9
such that [en e,] = :2. It is easily verified that g is thus given a solvable Lie
algebra structure. Now let I) be a non—commutative Lie algebra of dimension 2
over K. We show that b is isomorphic to 9. Let (f1, f,) be a basis of b. The ele-
ment If“ f,] is non-zero (otherwise 9 would be commutative) and hence it
generates a 1-dimensional subspace l of 9. Then [1), b] = 2. Let (2;, 0’2) be a
basis of 9 such that 0'2 E t. Then [4, :5] = he; with A aé 0. Replacing e; by
A'lel, it can be assumed that A = l, whence our assertion.
II. Formulae (5) of§ l prove that 9t(n, K) = n(n, K). As n(n, K) is nil-
potent and hence solvable, 1(n, K) is solvable. Therefore “(11, K) is solvable. In
particular, “(2, K) is isomorphic to the algebra of Example I.

2. RADICAL OF A LIE ALGEBRA

Let a, b be two solvable ideals of a Lie algebra g. The algebra (on + b) [b is
isomorphic to a/(a n b) and hence is solvable and a + b, which is an utension
of (a + b)/b by b, is also solvable (Proposition 1). It follows that a maximal
solvable ideal of g contains every solvable ideal of g and hence g has a largest
solvable ideal. This enables us to make the following definition:
DEFINITION 2. The radical of a Lie algebra is it: largest solvable ideal.
Pnoposmou 3. The radical r 9fa Lie algebra g i: the smallest ideal cfg such that 9/:
has radical {0}.
Let a be an ideal ofg and (I) the canonical mapping of9 onto g/a. If the radical
of g/a is zero, then (Mr), which is a solvable ideal of g/a, is zero; hence I: C a.
-1
On the other hand, the inverse image cl) (t’) of the radical t’ ofg/r is an ideal ofg
which is solvable by Proposition l and hence is equal to r; therefore t’ = {0}.
PROPOSITION 4-. Let 9,, . . ., g, be Lie algebras. The radical I If the product Q)“ the
9, it the product 43/ the radical: t‘ of the 9,.
The product r' of the rl is a solvable ideal (Proposition 1) and hence :' C t.
The canonical image of r in 9‘ is a solvable ideal of 9. and hence is contained
in q; hence t C r’.

B. NILPOTENT RADICAL OF A LIE ALGEBRA

Dermmon 3. Let 9 be a Lie algebra. The nilpntmt radical y“ g is the intermtian
if the kernels y" thefinile—dimemiarml simpk representation: of 9.
Remarks. (1) Let 3 be the nilpotent radical of 0. As every decreasing sequence
of vector subspaces of 9 is stationary, there exists a finite number of finite-
dimensional simple representations of 9 whose kernels have intersection s.

44
 NILPOTENT RADICAL or A LIE ALGEBRA § 5.3

The direct sum of those representations is semi-simple and has kernel 9. It

follows that the set ofkelnels offinite—dimensional semi-simple representations
of g has a least element, namely a.
(2) By Proposition 4 (c) of § 4, no. 3, e is also the intersection of the largest
nilpateney ideal: of the finite—dimensional representations of g. In particular,
9 is contained in the largest nilpotent ideal of g and is therefore a nilpotent
ideal of g.

(3) Every linear form A on g which is zero on 99 is a simple representation

(with space K) of g, whence 1(9) = {0}. It follows that a C 99. On the
other hand, a is contained in the radical t of g by Remark 2. We shall prove
that a = r n 99.
Lemma 1. Let V be a finite—dimianal vector space over K, g a subalgebra of gl(V)
such that V is a :z'mple g-mmiale and a a commutative ideal if g. 77m: a n 99 = {0}.
Let S be the subalgebra of 3’ (V) generated by l and a.
If I! is an ideal ofg contained in a such that Tr b: = 0 for all I: e b and all
:65, then in particular, by definition of S, Tr(b") = 0 for every integer
n > 0 and hence I: is nilpotent (Algebra, Chapter VII, §5, no. 5, Corollary 4
to Proposition 13); as the elements of b are all nilpotent, b = {0} (§ 4-, no. 3,
Lemma 2). We first apply this to the ideal [g, a] of g. Ifxe g, a e a, 36 S,
then Tr[x, a}v = Tr(xa: —- axs) = Tr x(a: — ta) = 0 since a: = :a; hence
[9, a] = {0}. Hence the elements of g commute with those of a and hence
also with those of S. Ifx, y belong to g and s e S, then
Tr[z,y]s = Tr(xys — yxx) = Tr x(y: '— A‘y) = 0
since y: = :3]; then taking b to be the ideal 95 n a, it follows that 99 n a = {0}.
THEOREM 1. Le! g be a Lie algebra, t it: radical and a it: nilpatent radical. Then
3 = 99 n r.
We already know that a C 99 n 1'. Hence it will suflice to show that if p
is a finite-dimensional simple representation of Q then 9(99 n r) = {0}. Let
k be the least integer 20 such that p(9““r) = {0); we write 9’ = 9(3),
(1' = p(D“r) ; as 9": is an ideal ofg, a' is an ideal of 9’; this ideal is commutative
since p(9"“r) = {0}. IfV is the space of p, g' C gl(V) and V is a simple
g’-module. Then 9(99 n 9%) C 99’ n a’ = {0). Ifk > 0, then 9": C 9:; and
9(95) = {0), contrary to the definition of 1:. Hence I: = 0, that is

9(99 n r) = {0}.
COROLLARY 1. Let g be a :alvable Lie algebra. The nilpaten! radical of g i: 99.
If p i: a finite—dimenxianal :imfle reprerentaliaa y" 5, 9(9) is commutative and the
axsaeiative algebra L generated by l and 9(9) 1': afield 1yrfinite degree over K.

45
l LIE AIDEBRAS

Herer = g,whence5 = 99. Hence 9(99) = (0}, which shows that g' = p(g)
is commutative. Every element #0 of L is invertible by Schur’s Lemma;
hence L is a field.
COROLLARY 2 (Lie’s Theorem). Let g be a solvable Lie algebra; suppose that K
is algebraically closed. Let M be a g-module 4y" finite dimension aver K and let
(MQD‘K, be a Jordan-Holder series LfM. Then M‘-,/M, is (y’dimension 1 over
Kfor l S i S rand,far all x59, 2“,t = A.(x).l, where). is a linearfonn on
g whieh is zero on 99. In particular, every simple g-module offinite dimension over K
is infaet of dimension 1.
Let p, be the representation of g on M‘d/M‘. The associative algebra
L1 generated by 1 and p,(g) is a field, a finite extension of K and therefore equal
to K; and M‘d/M. is a simple Isl-module, whence dim M._1/M. = 1. The
rest of the corollary is obvious.
Remarks. (1) If (MONK, is replaced by another Jordan-Holder series of M,
the sequence (M, . . ., 1,) is replaced bya sequence of the fom‘i (Am), . . ., Mm):
where 1: is a permutation of {1, . . ., r}, as follows from the Jordan-Holder
Theorem.
(2) Let (en . . ., e,) be a basis ofM such that e, e M,_1, el (7? M, (l S i g r).
If x e g the endomorphism of M corresponding to x is represented with respect
to this basis by a triangular matrix whose diagonal coefficients are

110‘), - - -, 7‘1(")~

COROLLARY 3 Suppose that K is algebraically closed. If g is an r-dimensional solv-

able Lie algebra, every ideal of g is a term of a decreasing sequence of ideal: of dimen-
sion: 1, r — l,. ,0.
Every ideal 15 part of a Jordan-Holder series of 9, considered as the space
of the adjoint representation (Algebra, Chapter I, §6, no. 14, Corollary to
Theorem 8); then it suffices to apply Corollary 2.
COROLLARY 4. Suppose that K = R. Let g be a solvable Lie algebra. Every simple
represmtation of g is of dimension <2. Every ideal y g is a term of a demasing se-
quence (91)o<xsm‘lfid‘“l5m‘hlh“t90 = 9.5... = WNW at. 1/9. < 2 (1 < i < M)-
This IS proved m a similar way to that for Corollaries 2 and 3, using the
fact that every algebraic extension of R is of degree <2
COROLLARY 5. For a Lie algebra g to be solvable, it is necessary and mfieient that
99 be nilpotent.
The condition is necessary by Corollary 1. It is sufficient since 9/99 is
commutative.
COROLLARY 6. Let p be a finite—dimensional representation of a Lie algebra 9. Let

46
 A CRITERION FOR SOLVABILITY § 5.4

t he the radical of 9. Every element at 61‘ such that p(x) is nilpotent belongs to the
largest nilpoleary ideal n of 9.
Let V be the space of 9; let (VOar be a Jordan-Holder series for the
t—module structure on V and let p, be the representation of r with space
V./V,.l (l < i S r). If p(x) is nilpotent, so is 9.0:); as for all i the algebra
generated by p,(x) is a field, 9.0:) = 0. Conversely, if p,(x) = 0 for all i,
p(x) = 0. This shows that the set a of are: such that p(x) is nilpotent is an
ideal of r. On the other hand, [9, a] C 99 n t C n n t C a and hence a is
an ideal of g. This proves that a C n.

COROLLARY 7. Let g be a Lie algebra and t its radical. The following four sets are
identical: (a) the largest nilpoteat ideal of g; (b) the largest nilpatent ideal y” t;
(c) the set If x e r such that ad. 2: is ailpatent; (d) the set of x e r such that ad, 2:
is ailpateat.
Let these sets be denoted by a, b, c, b. The inclusions n C b C b C c are
clear. c C a by Corollary 6 applied to the adjoint representation of 9.

4. A CRITERION FOR SOLVABILI’I'Y

Lemma 2. Let x be an endomorphism of a film-dimensional vector space V and s
(rap. n) its semi-simple (resp. nilpotent) component (cf. Algebra, Chapter VIII,
§9, no. 4-, Definition 4-). Let ad 1:, ads, ad 1: be the respective images of x, s, rt
in the adjaint representation of gl(V). Then ads (resp. ad a) is the semi-simple
(reap. nilpateat) component of adx and is equal to a polynomial in ad x with coefi-
cients in K and aa constant term.
We know that adx = ad 3 + ad rt, [ad s, ad a] = 0 and ad a is nilpotent
(§ 4-, Lemma 1). We show that ad s is semi-simple. It suflicec to do this when
K is algebraically closed (cf. Algebra, Chapter VIII, §9, no. 2, Proposition 3).
Then let 0mg“ be a basis of V such that s(e,) = he, (I, E K). Let (Eu)
be the canonical basis of Mn(K) = g[(V). By formulae (5) of § l,

(ad ‘)-Eu = (N — 301511

and hence ad 5 is semi-simple. The last assertion of the lemma follows from
Algebra, Chapter VIII, §9, no. 4, Proposition 8.

Lemma 3. Let M be a finite—dimensional vector space, A and B two vector subspaces

of gl(M) such that B C A and T the set oftegl(M) such that [I, A] C B. If
2 e T is such that Tr(zu) = 0 for all a e T, then 2 is ailpolent.
It suffices to prove this when K is algebraically closed, which we shall
assume henceforth. Let s and n be the semi-simple and nilpotent components
of z and let (2,) be a basis of M such that s(e,) = he, ()4 e K). Let V C K
be the vector space over Q generated by the M. We hood to prove that V = {0).
Letf be a Q—linear form on V and let I be the endomorphism of M such that

47
I LIE ALGEBRAS

lei = f (1,)“. If (Eu) is the canonical basis of gl(M) defined by Ene,‘ = 81kt”
then
(ad ‘)Eu = (’1 — ADE"
(ad’)Eu = (f0!) '—f(7v))Eu~
There exists a polynomial P with no constant term and with coefficients in
Ksuch thatPOq —— 1,) =f()..) —f(A,) foralliandj(forif)q -— A, = In — A,”
thenfO‘t) —f(7v) =f(7w) -f(7w) and, int - A, = 0,1111) —f(7v) = 0)-e
Then ad t = P (ad J). On the other hand, ad .r is a polynomial with no constant
term in ad 1. Now (ad 2) (A) C B, whence also (ad t)(A) C B. By the
hypothesis 0 = Tr(zt) = 2 MO“), whence 0 =f(Tr(zt)) = 21%).)?
Since the f(N) are rational numbers, f = 0, which completes the proof.
THEOREM 2 (Cartan’s criterion). Let g be a Lie algebra, M a finite-dimensional
vector xpaee, g a representation of g on M and B the bilinearfarm on g awaciated with
9. Then 9(9) 1': tall/able If and only ifflg is orthngtmal to g with respect to {3.
It can obviously be reduced to the case where g is a Lie subalgebm of
gl(M) and p is the identity mapping. If g is solvable, 99 is contained in the
largest nilpotency ideal of the identity representation of 3 (Theorem 1) and
hence is orthogonal to g with respect to [3 (§ 4, Proposition 4 (11)). Suppose
that 99 is orthogonal to g with mpect to B. We prove that g is solvable.
Let T be the set of t e 91(M) such that [1, g] C 99. If [ET and x, y belong
to 9, then [13, x] e 99 and hence

Tr(t[x:yl) = MUM) = 0
whence by linearity Tr(tu) = 0 for all u 695. Also, clearly 95 C T. Hence
(Lemma 3) every element of 99 is nilpotent. It follows that 99 is nilpotent
(§ 4, Corollary 3 to Theorem 1) and hence that g is solvable (no. 3, Corollary
5 to Theorem 1).

5. FURTHER PROPERTIFS OF THE RADICAL

PROPOSITION 5. Let g be a Lie algebra and t it: radical.
(a) If p is a finite-dimensional representation of g and (3 is the mweiated bilinear
farm, 2 and 99 are orthogonal with rupee! to B.
(b) r i: the orthugonal nfgg with respect la the Killingform.
Let x,y be in g, zet. Then [y, z] 599 n r and hence

MPH]. 2) = We [yt 2]) = 0

(Theorem 1), Hence (a).
Let t’ be the orthogonal ofgg with respect to the Killing form. It is an ideal
of 9 (§ 3, no. 6, Proposition 7 (a)) which contains r by the above. On the other

48
 EXTENSION OF THE BASE FIELD § 5.6

hand, the image 5 of t' under the adjoint representation of g is solvable

(Theorem 2) and hence r' is solvable being a central extension of 3. Hence
r' C r.
COROLLARY 1. Let g be a Lie algebra. Then 9 is salt/able gfaml only 5793 is arllw-
ganal to g with rerpect to the Killiagftmn.
This is an immediate consequence of Proposition 5 (b).
COROLLARY 2. The radical r qf a Lie algebra g 1': a characteristic ideal.
93 is a characteristic ideal and the Killing form is completely invariant
(§ 3, no. 6, Proposition 10). Hence the orthogonal of 99 with respect to the
Killing form is a characteristic ideal (§ 3, no. 6, Proposition 7 (b)).
COROLLARY 3. Let g be a Lie algebra, t itx radical and a an ideal of g. Then the
radical (If a is equal to t n a.
r n a is a solvable ideal of a and hence is contained in the radical r’ of a.
Conversely, t’ is an ideal of g (Corollary 2 and § 1, no. 4-, Proposition 2)
and hence r’ C r.
Corollary 2 can be made more precise as follows:
PROPOSITION 6. Let g be a Lie algebra, t it: radical and n i1: laigest nilpatent ideal.
Every derivation af 9 map: r into n.
Let D be a derivation of 9. Let 5’ = g + Kxo be a Lie algebra in which
9 is an ideal of codimension 1 such that D): = [30, x] for all x E 9 (§ 1, no. 8,
Example 1). By Corollary 3 to Proposition 5, r is contained in the radical t’
ofg'. Then D(t) = [x0, I] C [g’, 9'] n r’ = e’. For all 1: e 5’, ad“. .2: is nilpotent
(Theorem 1). Hence, for all xes' fig, adax is nilpotent. Hence D(t) is
contained in the nilpotent ideal e' n g of g.

COROLLARY. The largest nilpotcnt ideal of a Lie algebra is a characteristic ideal.

Remark. To summarize some of the above results, note that, if t, n, 5, 1 denote
respectively the radical of 9, the largest nilpotent ideal of g, the nilpotent
radical of g and the orthogonal of 9 with respect to the Killing form, then

t D I D n D 5.

The inclusion r 3 ! follows from Proposition 5 (b). The inclusion t: n

follows fi‘om §4, no. 4, Proposition 6 (b). The inclusion n D a has been
pointed out in Remark 2 of no. 3.

6. EXTENSION OF 11-11: BASE FELD

Let g be a Lie K—algebra and K1 an extension of K. Clearly gm, is solvable
if and only if g is solvable, since 9'19“”) = (9'?t

49
I LIE MEBRAS

Let r be the radical of 9. Then rm) 2‘: the radical (J Eam- For let; [3 be the
Killing form of 9. As t is the orthogonal of Fig with respect to (3 (Proposition
5 (b)), for.) is the orthogonal of (99%“, = 9(Sam) with respect to the form
derived from [3 by extension from K to K,, that is the Killing form of 90(1)
(§ 3, no. 8). Our assertion then follows from a further application of Proposi-
tion 5 (b).

§6. SEMI-SIMPLE LIE ALGEBRAS

Reeall that K denote: afield afehamcteristie 0 and that all Lie algebra: are assumed
to be finite-dimensional over K.

1. DEFINITION OF SEMI-SIMPLE LIE ALGEBRAS

DEFINITION I. let a be a Lie algebra. g is called semi-simple if and only if the only
commutative ideal qf 9 is (0}.
Remarks. (1) The algebra {0} is semi-simple. An algebra of dimension 1 or
2 is not semi-simple (cf. § 5, no. 1, Example 1). There exist semi-simple algebras
of dimension 3 (cf. no. 7).
(2) A semi-simple algebra has zero centre and hence its adjoint representa-
tion is faithful.
(3) If 91,. . ., 9,, are semi-simple, g = 91 x - - - x g,I is semi-simple: for
if a is a commutative ideal of g, the projections of 0 onto 91, . . ., 9,, reduce to
(0}.
THEOREM 1. Let 9 be a Lie algebra. Thefillawing conditions are equivalent:
(a) g is semi-simple.
(b) The radical t nfg it zero.
(c) The Killingform {3 of g i: non-degenerate.
Mureover, a semi-simple Lie algebra is equal In its derived ideal.
(a) a (b): for if I' 9e (0}, the last non—zero derived algebra of I: is a com-
mutative ideal of g.
(b) => (c): this follows from Proposition 5 (b) of § 5, no. 5 (which at the
same time proves the last assertion of the theorem).
(c) 2 (a): this follows from Proposition 6 (b) of § 4, no. 4.
COROLLARY. Let g be a :emidimfle Lie algebra and p a representation of 9 an a
finite-dimensional veelar :pate V. Then 9(5) C 5l(V).
The linear form x>—>Tr 9(x) (x e g) is zero when x is of the form [y, z]
(1/69, 25 g) and hence on 99 = g.

50
 SEMI-SIMPLICITY or REPRESENTATIONS § 6.2

PROPOSITION 1. Let g be a remi-rimple Lie algebra and p afinite—dimensionalfaitlyiel

representation of 5. Then the bilinearfinn on 3 associated with p it nan—degenerate.
The orthogonal of g with respect to this form is a solvable ideal (§ 5, no. 4,
Theorem 2) and hence is zero.

COROLLARY 1. Let g be a Lie algebra, {i it: Killing firm and o a semi-ample .mb-
algebra of g. The orthogonal b of a with respect to B is a supplementary subspace of
a in g and [(1, 1)] C I). (f a it an ideal of 9, so is I), which i: then the eentralizer of a
1n 9.
Let B’ be the restriction of B to a: it is the bilinmr form associated with the
representation at» ad,x of a on the space g. This representation is faithful
and hence fl' is non-degenerate (Proposition 1). Hence I) is supplementary
to uin 9. 0n theotherhand, ifx,yarein aand 26!), then

130?, U, z]) = 130:" y], z) = 0.

since [x,y] e a, and hence [g], z] e b, which proves that [a, b] C I). If a is an
ideal of 9, we know that b is an ideal of 5 (§ 3, Proposition 7) and g is identi-
fied with a x I). As the centre of a is zero, the centralizer of a in g is l).

COROLLARY 2. Every extenrian 13/ a semi-simple Lie algebra by a semi-simple Lie

algebra is semi-simple and trivial‘
This follows immediately from Corollary 1.
COROLLARY 3. If g is semi-simple, every derivation If g is inner.
ad g is isomorphic to g and hence semi-simple and IS an ideal of the Lie
algebra b of derivations of g (§ 1, Proposition 1). If D e D commutes with the
elements of ad 5, then, for all reg, ad D(x) = [D, ad x] = 0, whence
D(x) = 0; hence D = 0. Corollary 3 then follows from Corollary 1.

2. SEW-SIMPLE)!“ 0F REPRESENTATIONS
Lemma 1. Let g be a “mi-simple Lie algebra. The adjoin! representativn vf g is remi-
rimple. Every ideal and every quatien! algebra {If g is semi-simple.
Let a be an ideal of g. The orthogonal b of a in g with respect to the Killing
form is an ideal of g and a n b is a commutative ideal (§ 3, no. 6, Proposition
7) and hence zero. Hence b is supplementary to a in 9. Moreover, as the
Killing form of g is non-degenerate, so are its restrictions to a and 1: (Algebra,
Chapter IX, §4, no. 1, Corollary to Proposition 1) and hence a and b are
semi-simple (no. 1, Theorem 1 and § 3, no. 6, Proposition 9).

Lemma 2. Let g be a Lie algebra. Then thefollowing two conditivnr are equivalent:
(a) All 5‘ ‘ " ' 'linear , ' «#9 are ' p’.
(b) Given a linear representation p qf g on a finite-dimmmnal vector space V and

51
I LIE ALGEBRAS

a vector mbrpaee W bf eodimemian 1 such that p(x)(V) C W for all x-E-g, there
exists a supplementary line of W which it stable under 9(9) (and heme annihilated by
g .
Méleatly (a) implies (b). Suppose that (b) holds. Let a be a finite—dimen-
sional representation ofg on a vector space M and N a vector subspace which
is stable under 6(9). Let u be the representation of g on J (M) canonically
derived from a (§ 3, no. 3): recall that pL(x) = adgunfire). Let V (resp. W)
be the subspace of $(M) consisting of the linear mappings of M into N
whose restriction to N is a homothety (rap. zero); then W is of codimension '
l in V and Ma) (V) C W for all xe 9. By condition (b), there exists a EV
which is annihilated by [1(a) for all xeg and whose restricu'on to N is_ a
non-zero homothety. By multiplying u by a suitable scalar, it can be as-
sumed that u is a projector of M onto N. To say that y.(x) .u = 0 means that
u is permutable with 6(a). Hence the kernel of a is a supplement of N in M
which is stable under a(x) for all x e 5. Hence a is semi-simple.

Lemma 3. Let g be a semi-simple Lie algebra, p a linear representation bf 9 an a

finite-dimensional better space V and W a subspace of V of codimenxion 1 such that
9(a) (V) C W for all x e 9. Then there exists a supplementary line of W whieh is
stable under 9(5).
For all x s 9 let a(x) be the restriction of p(x) to W, Suppose first that o'
issimple. Ifo = 0, then p(z)p(y) = 0forallx,yin 9, hence 9(9) = 9(99) = {0}
and our assertion is obvious. If 0' 9e 0, let I! be the kernel of c' and let in be
a supplementary ideal of n in 9 (Lemma 1); then m aé {0} and the restriction
of o' to m is faithful; the restriction to m of the bilinear form associated with
o‘ is non-degenerate (Proposition 1) and hence the Cwimir element a associ-
ated with m and a can be formed. By Proposition 12 of § 3, no. 7, 6(6) is an
automorphism of W. On the other hand, 9(a) (V) C W. Hence the kernel
Z of p(e) is a supplementary line of W; since e belongs to the centre of the
enveloping algebra of 9, [3(0) is permutable with p(x) for all x E g and hence
Z is stable under 9(9).
In the general case we argue by induction on the dimension of V. Let T
be a minimal non-zero stable subspace of W. Let 9' be the quotient representa-
tion on V’ = V/T. Then, for all xeg, p’(x)(V’) C W', where W’ = W/T
is of codimension 1 in V'. By the induction hypothesis there exists a line 2’
which is supplementary to W' and stable under p’(g). Its inverse image 2
in V is stable under p(g), contains T as subspace ofcodimension 1, Z n W = T,
and hence p(a) (Z) C T for all x e 9. By what was proved above, there exists
a supplementary line of T in Z which is stable under 9(9) ; this line is sup-
plementary to W in V, which completes the proof.

THEOREM 2 (H, Weyl). Every finite—dinmuianal linear representatibn «y a semi-

simple Lie algebra is empletely redatible.

52
 SEMI-SIMPLICH‘Y or REPRESENTATIONS § 6.2

This follows fiom Lemmas 2 and 3.

DEFINITION 2. A Lie algebra g is called simple if the only ideals of g are {0} and g
and iffinther g is non-commutative.
A simple Lie algebra is semi-simple. The algebra (0} is not simple.
PROPOSITION 2. For a Lie algebra g to be semi-simple, it is necessary and mfieieat
that it he a product Ifsimple algebras.
The condition is suflicient (no. 1, Remark 3). Conversely, suppose that 3
is semi-simple. Since the adjoint representation of g is semi-simple, g is the
direct sum of minimal non-zero ideals «11, i . ., am. Then 9 is identified with
the product algebra of the in (§ 1, no. 1). Every ideal of a‘ is then an ideal
of g and hence zero or equal to 0.. On the other hand a, is non-commutative.
Hence the q are simple Lie algebras.
COROLLARY l. A semi-simple Lie algebra is the product ofit: simple ideal: 9‘. Every
ideal of g i: the product of certain of the 9,.
g = a, x ‘ - . x a,” where the a, are simple. As the centre of al is zero,
the centralizer of ai in g is the product of the a, for j 95 2'. Then let a be an
ideal of g. If it does not contain 0., then a n a( = {0}, hence [a, a,] = {O}
and a is contained in the product of the a, for j ye i. It follows that a is the
product of certain of the a.. Hence the simple ideals of g are precisely the
a‘.

The simple idmls of a semi-simple Lie algebra are called the simple cam-
[uments of g.
COROLLARY 2. Let g, 9’ be two Lie algebras, v and r’ their radical: andf a homo-
morphixm q onto 9'. Then I’ =f(r).
As f(r) is solvable, f(r) C t". On the other hand, g/t is semi-simple (§ 5,
no. 2, Proposition 3), hence g’/f(r), which is isomorphic to a quotient of
g/t, is semi-simple (Lemma 1) and hence f(t) D I’ (§ 5, no. 2, Proposition
3).
Remarks. (1) Theorem 2 admits a converse: if every finite-dimensional repre-
sentation of 9 is semi-simple, g is semi-simple. For since the adjoint representa-
tion is semi-simple, every ideal of g admits a supplementary ideal and hence
can be considered as a quotient of g. If g is not semi-simple then 3 admits a
non-zero commutative quotient and therefore a quotient of dimension 1.
Now the Lie algebra K of dimension 1 admits non-semi-simple representa-
tions, for example

A 0

(2) Let 9 be a Lie algebra over K and a a representation of g on a vector

53
I LIE ALGEBRAS

space M. On the other hand let f be a K-linear mapping of 9 into M such

that:

(1) f([x,y]) = c(x).f(y) - UMJO‘)

for all x, y in 9. By § 1, no. 8, Example 2, being given a and f is equivalent
to being given a homomorphism x >—> (f(x), a(x)) of g into af(M). On the
other hand we have seen (lac. cit.) that the element (f(x), c(x)) of af(M) is
canonically identified with the element p(x) of gl(N) (where N = M x K) .
which induces e(x) on M and maps the element (0, l) of N to f(x). And 9
is then a representation of 5 on N such that p(x) (N) C M for all x E 9.
Then, if g is semi-simple, there exists (Lemma 3) a line Z which is supple-
mentary to M in N and annihilated by 9(3). In other words, they: exist: an
element moeM such that (—mo, 1) EN is annihilated by 9(x), that is such
that

(2) f(IV) = 00:) ~”‘o

for all x e 5.

'Suppose that K = R. Let G be a connected Lie group with Lie algebra

9. Consider an analytic homomorphism 4: of G into the affine group A
of M corresponding to a homomorphism x >—-> (f(x), e(x)) of 5 into af(M).
The above results can be interpreted by saying that if g is semi-simple
¢(G) leaves a point of M fixed. For let H be the set of elements of GL(N)
which leave stable all the linear varieties of N parallel to M. There exists
(§ 1, no. 8, Example 2) a canonical isomorphism 4; of A onto H. Let Z be
a supplementary line of M in N. To say that 9(9) annihilates Z amounts
to saying that (t1; 9 42) (G) leaves the points of Z fixed and hence (taking
account of the definition of 4:) that (MG) leaves fixed the projection onto
M of the point of intersection of Z and M x {1).

3. SEMI-SIMPLE ELEMENTS AND NILPOTENT ELEMENTS 1N SEW-SIMPLE

LIE ALGEBRAS
PROPOSITION 3. Let M b: u finile-dimsional vector rpm-e aver K and g a semi-
simpla mbalgebm uf 91(M). Then 3 contain: the :eml-simplz and m'lpotmt component:
affix elements.
If K1 is an extension of K, the Killing form of Slum is the extension to
Sum of that of 9 (§ 3, no. 8) and hence is non-degenerate; therefore gm, is
semi-simple. It therefore sullices Io prove Proposition 3 when the base field
is algebraically closed, which we shall henceforth assume to be the case.
For every subspace N of M, let gN be the subalgebra of gl(M) consisting of
the elements which leave N stable and whose restriction to N has trace zero.
As 3 = 99, g C 9,. if N is stable under 9. Then let 9" be the intersection
of the normalizer of g in gl(M) and the algebras 9" where N runs through the

54
 SEMI-SIMPLE 12m AND mwo'nmr stamens § 6.3

set of subspaces of M which are stable under 9. As the semi-simple (resp.

nilpotent) component .t (resp. n) of xegl(M) is a polynomial in x with no
constant term and ads (resp. ad n) is the semi-simple (resp. nilpotent) part
ofad x (§ 5, no. 4, Lemma 2), clearly x e 9‘ implies .t E 9* and n e 9*; it there-
fore suffices to show that 9* = 9. Since 9 is a semi-simple ideal ofg*, 9* = a x :5
(no. 1, Corollary 1 to Proposition 1). Let a e a and let N be a. subspace which
is minimal among the non-zero subspaces of M which are stable under 9. The
restriction of a to N is a scalar multiple of the identity by Burnside’s Theorem,
has trace zero by construction and hence is zero since K is of characteristic 0.
As M is the direct sum of subspaces such as N, it follows that a = 0 and
hence 9* = g.

COROLLARY. An element x of g is a semi-simple (rap. nilpotent) endomorphism qf

M if and only ifad, x is a semi-simple (resp. nilpalent) endomarphism 12/ 9.
Let 5 (resp. n) be the semi-simple (resp. nilpotemt) component of x69.
Then :eg and nag (Proposition 3). Then ad” (resp. ada n) is the semi-
simple (resp. nilpotent) component of ad. x, by Lemma 2 of § 5, no. 4-. If x
is semi-simple (resp. nilpotent) so then is ad, :r. If now ad“): is semi-simple
(rap. nilpotent), it is equal to ad‘s (mp. ad, 7:) and hence x = : (resp.
x = n) since the adjoint reprwentation of g is faithful.

DEFINITION 3. Let g be a semi-ample Lie algebra. An element x 4/ 9 it called mm"-

simple (resp. nilpotent) if, fir every g-moa'ule M affinite dimenrian aver K, Ma 1': a
semi-simple (resp nilpotent) endomorphum qf M.

PROPOSITION 4. Let g, 9’ be semi—simple Lie algebras and f a homomorphism of 5

into 9’. Ifx e g 1': semi-simple (resp. nilpntent), :0 isf(x). Iff is :wjettiue, every remi-
simfile (resp. nilpatent) element uf 9' it the image under f of a semi-Jimple (resp.
nilpotent) element of g.
If g is a representation of 9’, p of is a representation of g, whence the first
assertion. Iff is surjective, there exists a homomorphism g of 9’ into 9 such
that f o g is the identity homomorphism of 9' (no. 1, Corollary 2 to Proposi-
tion 1) and the second assertion then follows from the first.

THEOREM 3. Let g be a semi—simple Lie algebra.

(a) Let x E g. If there exirt: a fizillful representatian p cf 5 ruth that p(x) is a
:emi-rlmple (rcsp. nilpotent) endomurphirm, then x is mni—ximple (resp. nilpbtent).
(b) Every element Qf 9 can be written uniquely a: the sum (f a Jami-simple element
and a nil/intent element which cnmmute with one another.
Suppose that the hypothesis of (a) holds. Let a be a representation of g,
b the supplementary ideal of the kernel of a and a: the projection of 3 onto
b. Then adfl x is semi-simple (resp. nilpotent) by the Corollary to Proposition
3 and hence ad, 1(x) is semi-simple (rap. nilpotent). As e(x) = o'(at(x)),

55
I LIE ALGEBRAS

the first assertion follows from the Corollary to Proposition 3. Thevsecond

result then follows from Proposition 3 applied to a faithful representation.

4. REDUCTIVE LIE ALGEBRAS

DEFINITION 4. A Lie algebra it called reductive if it: adjoin! representation is Jemi-
simple.
PROPOSITION 5. Let 9 he a Lie algebra and t in radical. The following condition:
are equivalent: 4
(a) g is reductive.
(b) 99 i: semi-simple.
(c) g is the product nf a send-simple algebra and a commutative algebra.
(d) g has a finite-dimerm'aaal representation such that the associated bilinear farm
1': nan-degenerate.
(e) g has a faithful smi-sirnplefinite—donensiaaal representation.
(f) The allbatent radical of g is zero.
(g) r is the centre afg.
(a) a (b): if the adjoint representation of g is semi-simple, 5 is a direct
sum of minimal non-zero ideals a, and hence g is isomorphic to the product of
the a‘; and a. has no ideals other than {0} and a, and hence is simple or corn-
mutative of dimension 1. Therefore 99 is equal to the product of those a,
which are simple and hence is semi-simple.
(b) = (c): if 99 is semi-simple, g is isomorphic to the product of 99 by a
Lie algebra b (no. 1, Corollary 1 to Proposition 1); I) is isomorphic to 9/95
and hence is commutative.
(:2) => (d): let 9‘ and 92 be two Lie algebras, p‘ a finite-dimensional repre-
sentation of g. and B, the bilinear form on 9, associated with p‘ (i = l, 2);
p1 and p, can be considered as representations of g = g, x 92; let p be their
direct sum. Clearly the bilinear form on 3 associated with p is the direct sum
of ‘31 and {3, and hence is non-degenerate if [31 and (32 are non-degenerate
Then to prove the implication (c) o (d) it suffices to consider the 2 following
cases: (1) g is semi-simple; then the adjoint representation admits as associ-
ated form the Killing form, which is non-degenerate; (2) g = K; then the
identity representation of g on K has an associated bilinear form which is
non-degenerate.
(d) => (e): let p be a finite-dimensional representation of g and B the
associated bilinear form; by Proposition 4 of §4, no. 3, there exists a finite‘
dimensional semi—simple representation a of 9 such that the kernel in of o
is orthogonal to g with respect to B. If B is non—degenerate, then it = {0} and
hence a is faithful.
(e) => (f): this is obvious.
(f) :> (g): if the nilpotent radical of g is zero, 99 n t is zero (§ 5, no. 3,
Theorem 1); as [9, r] C 99 n r, r is the centre of g.

56
 imnmmvn Lm menus § 6.4-

(g) s (a): if r is the centre of g, the adjoint reprmentation of 9 is identified

with a representation of g/r, which is a semi—simple Lie algebra (§ 5, no. 2,
Proposition 3); this representation is therefore semi-simple (Theorem 2).
Remark. If a Lie algebra 9 can be decomposed as a product a x b of a com-
mutative Lie algebra a and a semi-simple Lie algebra I), this decomposition
is unique. More precisely, the centre of g is equal to the product of the centres
ofa and b and is hence equal to a. And 99 = 9a x 9!) = b.
COROLLARY. (a) Everyfinite product ifreductive algebras it a reductive algebra.
(b) If g is a reductive Lie algebra of centre c, every ideal if g i: a directfactor, the
product (fit: intersectiaru with c and 99, and is a reductive Lie algebra.
(c) Every quutient if a reductive Lie algebra it a reductive Lie algebra.
Assertion (a) follows for example from condition ((2) of Proposition 5.
Suppose that g is reductive. Let a be an ideal of 5. Since the adjoint repre-
sentation of g is semi-simple, a has a supplementary ideal b and g is identified
with a x b. For all x E 9, let p(x) be the restriction of ad, 2: to n. Then 9 is a
semi-simple representation of g which is zero on b and defines on passing to
the quotient the adjoint representation on a. Hence a is reductive. Similarly,
g/a and b, which are isomorphic, are reductive. Finally, let i), b’ be the centres
ofaandb;thena=b><.@a, =D'x9b, bxh’=c,9ax9b=99;
hence a = (a n c) + (a n 95).
PROPOSITION 6. Let g be a Lie algebra, t it: radical and 6 its nilpatcnt radical.
(a) s = [g,r] = 99hr.
(b) a it the interaction qf the arthaganal: 1y" 9 with respect to the biliaear form:
associated with thefinite-dimeruional representations of g.
Clearly [9, r] C 95; n 1'. Now 9g hr = a by Theorem 1 of § 5, no. 3.
Let g’ = g/[g, t] and f be the canonical homomorphism of 9 onto 9’; then
f(r) is the radical r’ of 9’ (Corollary 3 to Proposition 2, no. 2), hence
[g', r'] = {0) and r’ is the centre of 9'; therefore (Proposition 5) g’ has a
finite-dimensional faithful semi-simple representation, whence 5 C [9, r].
This proves (a).
Let t be the intersection of the orthogonals of g with respect to the bilinear
forms associated with the finite-dimensional representations of 9. Then
5 C t (§ 4, no. 3, Proposition 4 (d)). On the other hand, g/s has a finite-
dimensional faithful semi-simple representation and hence (Proposition 5)
a finite-dimensional representation [3 such that the associated bilinear form
is non-degenerate; considered as a representation of g, F has an associated
bilinear form [3 on g and the orthogonal of g with respect to (5 is a, whence
t C s. Hencet = 0.
Even if a 36 {0} there may exist non-degenerate symmetric bilineur forms
on 9 (Exercise 18 (c)). Such forms, of course, are not associated with any
representation of 9.

57
I LIE ALGEBRAS

COROLLARY. Let 9, 9’ be Lie algebras, b (resp. 9') the nilpatent radical of 9 (resp.
g’) arqa homomorphism ofg onto g'.
(3) Then 5’ = f(s).
(b) 9’ ix reductive if and only if the kernel off contains 9.

Ifr,r’arethe radicalsofg.9',fllen 6’ = [83"] = [f(9),f(0] =f([9,|']) =f(!)~

Assertion (b) is an immediate consequence of (a).

5. APPLICATION: A CRITERION FOR SEMI-SMUCITY 0F REPRE-

SENTATIONS
THEOREM 4. Let g be a Lie algebra, t it: radical, p afinite—dimenrional representa-
tion g” g, 9’ = p(g) and r’ = 9(1). Then thefollowing condition: are equivalent.-
(a) p is semi-simple;
(b) 9’ i: reduetive and its centre eomirt: ofsemi-simple endomorphism;
(c) t’ consist: of semi-simple endomorphirms;
(d) the restriction of p to r is semi-simple.
(a) = (b): if p is semi-simple, g’ is reductive (Proposition 5); the associative
algebra generated by 1 and g' is semi-simple (Algebra, Chapter VIII, §5,
no. 1, Proposition 3), hence its centre is semi-simple (loo. cit, §5, no. 4,
Proposition 12) and hence the elements of this centre are semi-simple (lot.
rit., §9, no. 1, Proposition 2).
(b) a (c): if g’ is reductive, its centre is equal to its radical, that is t’,
whence the implication (b) => (c).
(c) = (d): suppose that r' consists of semi-simple endomorphisms. As
[g’, t’] consists ofnilpotent endomorphism: (no. 4, Proposition 6), [g’, r'] = (0}.
Then the implication (c) => ((1) follows from Algebra, Chapter VIII, §9,
no. 2, Theorem 1.
(d) a (a): let a be the nilpotent radical of g and p' the restriction of p
to r. The elements of [3(9) are nilpotent and hence a is contained in the largest
nilpotency ideal of (9’. As 9' is semi-simple, p'(o) = {O} and g’ is reductive
(Corollary to Proposition 6), so that g’ = a' x r’ with a’ semi-simple (Proposi-
tion 5). Let A’ (resp. R') be the associative algebra generated by l and u’
(resp. r’). It is semi-simple (Algebra, Chapter VIII, §5, no. 1, Proposition 3),
hence A' (8,; R' is semi-simple (lot. ei!., §7, no. 6, Corollary 4 to Theorem 3)
and hence the associative algebra generated by l and 9’, which is a quotient
of A’ ®K R', is semi-simple, which proves that p is semi-simple.

COROLLARY 1. Let g be a Lie algebra and p and 9' two finite-dimensional remi-
simple representations of 9. Then the tensor product of p and p' is semi-simple.
Let r be the radical of 9. For x e r, p(x) and p’(x) are semi-simple (Theorem
4), hence p(x) ® 1 + 1 ® p’(x) is semi-simple (Algebra, Chapter VIII, §9,
Corollary to Theorem 1) and hence the tensor product of p and 9’ is semi:
simple (Theorem 4).

58
 sumwsnm REDUCTIVE IN A LIE ALGBBRA § 6.6

COROLLARY 2. Let g be a Lie algebra, p a semi-simple representation (y g on a finite-

dimensional vector space V, T and S the [may and symmetric algebra: (j V and or,
as the representations of g on T and S canonically derived from 9. Then UT and as
are semi-simple and, more precisely, direct sum offinite~dimem~ianal simple ”presenta-
lions.
Let T’I be the subspace of T consisting of the homogeneous tensors of order n.
This subspace is stable under 0’1 and the representation defined by 0'1 on
T’I is semi-simple (Corollary 1). Hence the corollary for a... and therefore
for as, which is a quotient representation of UT.

COROLLARY 3. Let g be a Lie algebra and p and p’ Iwofinile—dimemional semi-simple

representations qf 9 an space: M and M'. Then the representation of g on 9AM, M’)
canonically derivedflow 9 and p' is :mi-simplet
The g—module .YflM, M’) is canonically identified with the g-module
M“ ®K M’ (§ 3, no. 3, Proposition 4), so that Corollary 3 follows from Corol-
lary l.
COROLLARY 4. Let g be a Lie algebra, a an ideal q and p a mni-n'mple representa-
tion of g.
(a) The restriction p' af p la a is semi-Wk.
(b) If p is simple, 9’ is a mm qfsimple represenlaliam isomorphic to one another.
Passing to the quotient by the kernel of p, p can be assumed to be faithful.
Then 9 is reductive. Let 5 = 9, X 9,, Where 91 is the centre of g and 55 is
semi-simple. Then a = a, x a2, where «I C 91, «2 C 9, and n1 is the centre
of a. The demerits of 9(91), and in particular those of 90“), are semi-simple
(Theorem 4) and hence p' is semi-simple (Theorem 4). Hence (a). Assertion
(b) follows from (a), using § 3, no. 1, Corollary to Proposition 1.

6. SUBALGEBRAS REDUCTIVE IN A LIE ALGHRA

DEFINITION 5. Let g be a Lie algebra and b a Lie .mbalgebra of g. b is called reductive
in 9 if the representatian x >—> ad, x of b is semi-simple.
This representation admits as subrepresentation the adjoint representation
of I). Hence, if I) is reductive in g, b is reductive. On the other hand, to say that
a Lie algebra is reductive in itself is equivalent to saying that it is reductive.
PROPOSITION 7. Le! g be a Lie algebra, b a subalgebra reductive in g, p a representa-
tion (If g on a vector space V and W lhe sum of Ihefinile—dimeminnal subspaces of V
which are simple b-mudules. Then W is stable under 9(9).
Let W0 be a finite-dimensional simple sub-b-module of V. We need to prove
that p(x)(Wo) C W for all x e 9. Let M denote the vector space 9 considered
as an b-module by means of the representation x>—> adll x of I) on 9. Then
M ®x W0 is a semi-simple b-module (Corollary 1 to Theorem 4-). Let 0 be

59
l LIE ALGEBRAS

the K—lincar mapping of M ®K Wo into V defined by 6(x ® w) = p(x)w.

This is an b—module homomorphism, for ify e!) then:

9([1/. x] 8 w + x 63 MW) = p([% X])w + 900904)!”

= 9(y)p(X)w = 9(y)0(x 69 w).
Hence 0(M ®x W0) is a finite-dimensional semi-simple b<module. Hence
0(M ®KW9) C W, that is p(x)(Wa) C W for all x E g.
COROLLARY 1. Let g be a Lie algebra, I) a .mbalgebra reductive in g and p a finite-
dimemianal ”mi-simple representation If 9. Then the mln'etion of p to b i: semi-
simple.
It suffices to consider the case where p is simple. We adopt the notation
V, W of Proposition 4. Let Wl be a subspace of V minimal among the non-
zero subspaces stable under p(b)i Then Wl C W, hence W ;é {0} and hence

COROLLARY 2. Let g be a Lie algebra, b a :ubalgebm reductive in g and I a :ubalgebra

of b reduelive in 1). Then [is reductive in g.
The representation at >—> ad, x of b on g is semi-simple and hence its restric-
tion to t is semi-simple (Corollary 1).

7. EXAMPLES OF SEMI-81mm LIE ALGEBRAS

PROPOSITION 8. Let V be a finite-dimerm'onal veetar space. Then 51(V) i: reductive.
it: centre it the It! nf bomathetie: if V, it: derived algebra is 51(V) and the latter is
semi-simple.
The identity representation of gl(V) is simple, hence 91(V) is reductive
and therefore gl(V) is the direct sum of its centre c and its derived algebra
9(gl(V)). The centre c is the set of homotheties (Algebra, Chapter II, §2,
no. 5, Corollary 1 to Proposition 5). Clearly 9(gl(V)) C sl(V). As
sl(V) n c = {0}, 9(gl(V)) = 51(V). Hence 91(V) is semi-simple.
Example We identify 51(K’) with the Lie algebra of matrices of order 2 and
trace zero. We write

X
=(3 a) he 3) H=<s -3)‘
Then X, Y, H form a basis ofsl(K‘) and

[H,X1= 2X [H, Y] = —2Y [X,Y] =H.

As an algebra of dimension 1 or 2 is not semi-simple (no. 1, Remark 1), 51(K’)
is simple; In fact, al(V) is simple for dimV 2 2, as we shall see later (of.
also Exercises 2] and 24).

60
 EXAMPLES or SEMI-SIMPLE LIE awaam § 6.7

PROPOSITION 9. Lot V be a water rpm: 11/finite dimmian a 00:7 K and {5 a arm-

dzgmmltt symmetric (resp. alternating) bilinearform 0n VA Let 9 be the Lie algebra
[omitting afthe xe gI(V) such that B(xm, m’) + [$(m, xm’) = Ofor all m, m' in
l‘11".dTllat g is mluctive; g is even ”mi-simple except in the ms: where [5 is symmetric
n = 2.
For all a e gl(V) let a“ denote its adjoint relative to B; then Tr(a) = Tr(u*)
by Proposition 7 of Algebra, Chapter IX, § 1, no. 8. The condition

Mum, m’) + B(m, am') = 0

for all m, m' in V means that u + 11* = 0. In particular, if veg!(V) then
(u —- v*)* = 12* — v and hence I; — 11* 69. Then let u be an element of 5
orthogonal to g with respect to the bilinear form ()5 associated with the identity
representation of g. For all a e 91(V), Tr 11(1) — 1/") = 0, hence
Tr(av) = Tr(uu*) = Tr(1w*)* = Tr(vu*) = —Tr(1)u) = —Tr(m/)
and hence Tr(uu) = 0. It follows that a = 0, so that 4) is non-degenerate.
Hence 9 is reductive (Proposition 5). It remains to show that the centre of
g is zero (except when [3 is symmetric and n = 2). By extending the base
field, we can assume that K is algebraically closed.
(a) When [i is symmetric, it can be identified with the bilinear form on
K" with matrix 1,1 with respect to the canonical basis (Algebra, Chapter IX,
§ 6, Corollary 1 to Theorem 1). Under these conditions 9 is identified with the
Lie algebra of skew-symmetric matrices (§ 3, no. 4, Example 1). Let
U = (“11) s 9; we use the fact that U commutes with the matrix (14,) e g all
of whose elements are zero except are," and 11,0.” (1’o aé jo) which are equal
respectively to l and —1. We find that um = um = “no = a,“ = O for
i aé imjo andj # imjo. Ifa > 2, there exist, for all distinct indices in andj,
distinct indices i and jo such thati 56 in, jo aé j, jo aé 1],; hence uh, = 0. This
proves that an element of the centre of g is zero.
(b) When [i is alternating and a = 2m, [3 can be identified with the bi-
linear form on K’" with matrix ( ‘2’» 3’") with respect to the canonical
basis (Algebra, Chapter IX, §5, Corollary 1 to Theorem 1). Under these
conditions 3 is identified with the Lie algebra of matrices of the form
U = (g 11:) where D = —‘A, B and Care symmetric (A, B, c, D in MM(K))
(§ 3, no. 4-, Example 1). We use first the fact that U commutes with the matrix
(3" gr),whereXeMm(K).Then AX = XA,CX = —'X0, X3 = —B.‘X;
as these equalities must hold for all X, it follows that A is a scalar matrix
Mm. We now use the fact that U commutes with the matrix (3 (1;), where

61
I LIE ALGEBRAS

Y is a symmetric matrix of Mm(K). Then AY = Y0 = CY = 0. This proves

first that A = 0. Moreover, for all X 6 M,"(K), X + ‘X is symmetric and
hence X : —‘XC. Using the equation CX = —‘XC obtained above, we
see that C commutes with every element of Mm(K) and hence that C is a
scalar matrix, necessarily zero since Y0 = 0. It is similarly shown that B = 0.

For B symmetric and n = 2, g is of dimension 1 and hence commutative.

For the other cases cf. Exercises 25 and 26.

8. THE LEVI—MALCEV THEOREM

Let E be a complete normed vector space over R and u a continuous endo-
morphism of B. We have seen (Funclim y” a real variable, Chapter IV, §2,
u
no. 6) that the sequence :1 is summable in .5.”(E) and we wrote

fi'
ell = exp 11 = “a 1‘”

Now let E be a vector space over the field K and u a m'lpotml endomorph-
. . u" .
mm of E. The series "20 '7 has only a finite number of non-zero terms and we
can therefore write
"n
e“ = exp" = Zn
This definition agrees with the above if K = R and if E is complete and
normed. If v is another nilpotent endomorphism of E which commutes with
u, then:

»»=(i"“)ti),i—
n-ofi
a,
«217! mp!
m
4—. (Z. ("W = Z; <u + W =
l q ) I

In particular, e":"‘ = rue" = a“ = l and hence :“ is always an automorph-

ism of E.
if further E is a (not necessarily associative) algebra and u is a (nilpotent)
derwahmz of E, then 5" is an aulumamhirm of the algebra E. For if x, y e E then

WW) = ”5:7
2 (f)u'(x)u‘(y)
62
 'n-IE LEVI-MALGEV THEOREM § 6.8

for every integer p 2 0 (Leibniz’s formula). It follows that

e"(zy)= Zfiu’ayw Z Z #"M

pan p>or+r=p r! .5!

= ”2,0 # # = mm)
m u’ x u‘

whence our assertion.

Now let 9 be a Lie algebra. If It belongs to the nilpotent radical of g, the
derivation ad' a: of g is nilpotent. We can therefore make the following defini-
tion:

DEFINITION 6. A special autvnmphitm if g is an automorphism qf g 13/ the farm

a“ ", when x is in the nilpolmt radical q.
Clearly a special automorphism lavas every ideal of g stable.

DEFINITION 7. Let 9 be a Lia algebra and r it: radical. A Levi .mbalgebm 4y“ 9 is
any rubalgebm a g mpplmwntary to r.

A Levi subalgebra is isomorphic to g/r and hence is semi-simple. As a

semi-simple subalgebra has only 0 in common with t, every semi-simple
subalgebra D such that g = t + I) is a. Levi suhalgebra; consequently the
image of a Levi subalgcbra under a surjective homomorphism is a Levi
subalgebra.

THEOREM 5 (Levi—Malcev). A Lie algebra 9 always has a Levi mbalgebra a.

Every Levi .mhalgebra vf g is the image ofs under a special autanwrphirm.
Let r denote the radical of Q. We first treat two special cases.
(a) [3, r] = {0}.
By Proposition 5, g is then the product of its centre r by 99 which is semi-
simple. Hence 99 is a Levi subalgebra. Moreover, if 5’ is a. semi-simple sub-
algebra, then a’ = 96’ (Theorem 1), hence E’ C 99 and 9;} is the unique
Levi subalgehra of g.
(b) [9, I] 9e {0} and the only ideals of 9 contained in t are {0} and t.
Then [9, t] = r, [r, r] = {0} and the centre of g is zero. Let M (resp. N)
be the subspace of 3(3) consisting of the linear mappings of 9 into t whose
restriction to r is a homothety (rap. zero); N is therefore of codimension
l in M. For m e M, let Mm) denote the ratio of the homothety of t defined
by m. Let cr be the representation of g on 59(9) canonically derived from
the adjoint representation; recall that a(x).u = [adI 1,11] for all .169 and
all u 6 3(5).

63
I LIE ALDEBRAS

2(3) Clearly a(z)(M) C N for all x e 9. Moreover, if x e r, y e g and

U u e M, then
M (4) (swam) = [Mon — u([x,y1) = —A(u)[x,y]
U since [1-, r] = {O}; and (4-) can be written:
N (5) (x).u = —-ad().(u).x).
U
P
As the centre of g is zero, the mapping x>—> adfl x defines a bijection d) of t
onto a subspace P of $(g). This subspace is stable under 5(9) and contained
in N since t is a commutative ideal and (5) shows that 6(x) (M) C P for
x e r. The representation of g on M/P = V derived from a is therefore zero
on r and defines a representation 5’ of the semi-simple algebra g]: on V. For
all y e g/r, the space a’(y) (V) is contained in N/P, which is of codimension
l in V. Consequently (no. 2, Lemma 3) there exists uoEM such that
- 1
Mun) = —1 and a(x).uoeP for all xeg. The mapping x>—> 11> (600.140) is
a linear mapping of 9 into I. By (5) its restriction to t: is the identity mapping
of 1:. Hence its kernel is a subspace a of g supplementary to r in :5. As a is the
set of x e 9 such that 5(1) .u0 = 0, a is a subalgebra of g and therefore a Levi
subalgebra of 9.
Let 6’ be another Levi subalgebra. For all x e 5’, let h(x) be the unique
element of r such that x + h(x) e 5. Since a is a subalgebra and r is com-
mutative, for x, y in 9':
[x + 11(3),!I + h(y)l = [my] + mm] + [h(X)ay] 6°
hence
h([x,y]) = (adx)~h(y) - (“w-hot)-
By Remark 2 of no. 2, there exists a E t such that h(x) = — [x, a] for all x e 5’.
Then:
(6) x+h(x)=x+[a,z]=(l+ada).x.
As I is commutative, (ad a)2 = 0 and hence I + ada = 2“”. As I = [g, r],
5““ is a special automorphism of {5. By (6) this special automorphism maps
a’ to e.
(c) General use:
We argue by induction on the dimension n of the radical. There is nothing
to prove if n = 0 and hence it can be assumed that the theorem holds for
Lie algebras whose radical is of dimension <dim r. By (a) it sufiices to con-
sider the case where [9, r] # {0}. As [9, r] is nilpotent (no. 4, Proposition 6),
its centre c is #{O}. Let m be a minimal non-zero ideal of 9 contained in c.
If m = r, we have case (b). Therefore let m # t: and let f be the canonical
mapping of g onto 9' = g/m. The radical of g’ is r’ = r/m. By the induction
hypothesis, 9' has a Levi subalgebra 9’. Then I; = fw') is a subalgebra of

64
 THE LEVI-minev THEOREM § 6.8

9 containing m such that b/m = b’ is semi-simple and hence having m as radical.

By the induction hypothesis 1) = m + 5 where s is a. semi-simple subalgebra.
Thentheequalityg’=r’+Vimpliesg=t+b=r+m+s=r+sand
hence a is a Levi subalgebra of 9.
Let 5’ be another Levi subalgebra of 9. Then f(a) and f(8’) are two Levi
subalgehras of g’ and there exists by the induction hypothesis 4’ e [g’,t’]
such that (“Wf(9')) = f(a). If as [g, r] is such that f(a) = a’, it follows
that:
ox=e‘“(s’)cm+e=b.

Then 91 and 9 are two Levi subalgebras of b and by the induction hypothesis
there exists b e m such that e‘”(91) = 6‘ Hence a = ew”.e““'(o’). Finally, as
m is in the centre of [g,r], :“”’.¢“" = cam”) and b + ae[g,t], which
completes the proof.
COROLLARY 1. Le! a be a Levi subalgebm 9/ g and b a semi-simple :ubalgebm of g.
(3) There exist: a :pea'al automorphism of g mapping 1} onto a .mbalgebm 9f 5.
(b) b is contained in a Levi subalgebra (jg.
Let r be the radical of g and a = b + t, which is a subalgebra of g. Then
a]: is semi-simple and r is solvable, hence t is the radical of a and I; is a Levi
subalgebra of a. On the other hand, a n 5 = b' is a supplementary subalgebra
to r in a and hence also a Levi subalgebra of 0. Then there exists (Theorem 5)
a e [a, r] such that Nu" maps 13 onto b’. Now a e [9, t]; Nu" maps 1; onto
a subalgebra of e and e"°u"(5) is a Levi subalgebra of 9 containing I).

COROLLARY 2. For a :ubalgebm b of g to be a Levi .mbalgebm af 9, it is necertary

and .mficient that I) be a maximal :mi-rimfile tubalgcbra of g.
This follows immediately from Corollary 1.

COROLLARY 3. Le! g be a Lie algebra and m an ideal qf 9 such that g/m is semi-
:imple. Then 5 contains a .mbalgebm supplementary to m in g. In uthcr words, every
extension If a semi-simple Lie algebra is inmential.
Let a be a Levi subalgebra of 9 (Theorem 5). Its canonical image in g/m
is a Levi subalgebra and therefore equal to g/m, hence g = a + m. Then an
ideal of a supplementary in s to the ideal in n a is a subalgebra of g supple-
mentary to m in g.
COROLLARY 4. Let g be a Lie algebra, r it: radical, a a Levi :ubalgebm y” g and m
an ideal cfg. Then m is the direct mm q n t which is its radical and m n 5 which
i: a Levi .mbalgebm of m.
We know that m n e is the radical of in (§ 5, no. 5, Corollary 3 to Proposi-
tion 5). Let b be a Levi subalgebra of m and 5' a Levi subalgebra of g contain-
ing 9 (Corollary 1). The algebra m n 3’ is an ideal of 9', is therefore semi-
simple, and contains I) and is therefore equal to 1). Hence m is the direct sum

65
1 LIE menus

of m n r and m n 3’. There exists a special automorphism mapping 9’ onto

a; this automorphism leaves t and m invariant; hence m is the direct sum
ofm n t and m n s and m n s is a Levi subalgebra of m.

9. THE WVARIANTS THEOREM

Let g be a Lie algebra and p a representation of g on a vector space M. For
every class 8 of simple representations of g let M5 be the isotypical compon-
ent of M of species 8. The subspace M0 of invariant elements of M is just .
Mao, where 80 denotes the class of the zero representation of g on a space of
dimension 1.
Lemma 4. Let p, a, 'r be representations of g on vector space: M, N, P. Suppoxe that
we are given a K-bilinear mapping (m, n) »—> met If M X N into P such that

(poem + atom») = «(noun

farallmEM, neN,zeg.
(a) If mo 6 M0, the mapping n »—> ma . n is a g-module homomorphism.
(b) Ifn 6 N,, then mum 5P5.
(c) If M ii a (not necessarily associative) algebra and the 9(x) are derivation; bf
M, M0 is a .mbalgebra of M and each M5 i: a right and left Mg-madule.
For mo 5M0, neN and x69,

«comm = Ma~(6(~)n),
whence (a). Assertion (b) follows from (a) (Algebra, Chapter VIII, §3, no.
4-, Proposition 10). IfN = P = M and a = r = p, assertion (b) gives asser-
tion (0) as a special case.
Lemma 5. Suppose further that a and 1' are semi-simple and hence N (resp. P) 1'; the
direct rum 4/ the N5 (resp. P5). For all n E N (resp. p E P), let nll (resp. p“) be it:
component in No (resp. Po). Let ma E MD. Thenfor all n EN, (rno.n)'q = mom“.
By linearity it suffices to consider the case where n E No. If 8 75 80, n"‘ = 0
and momePb (Lemma 4), hence (7710.71)“ = 0 = mom". If 8 = 80, n" = n
and mmneli'o (Lemma 4), hence (mom)?! = mom = mom".
THEOREM 6. Let g be a Lie algebra, V a :emi-n'mfle g-modale qffinite dimension
over K, S the symmetric algebra of V and res the derivatian of S which extend: xv
(III that it )—> A's is a representative! (if g on S).
(a) The algebra So of invariant: af S is generated by a finite number (if elements.
(b) For every class 8 If simple reprexentation: (f g offinite dimension aver K, let
Sa be the iratypical component «J S of species 8. Then S5 is a finitely generated So-
module.
Let S C S be the ideal of elements of S with no constant term. Let I be the
ideal of S generated by S“ n S and let (x1, :2, . . ., :p) be a finite system of

66
 THE INVARXANTS THEOREM § 6.9

generators of the ideal I (Commutative Algebra, Chapter III, §3). It can be

assumed that the :. belong to SD n S and are homogeneous (the x3 preserve
degrees and hence each 55 is a graded submodule). Let S1 be the subalgebra
of S generated by 1 and the 5,. Then Sl C So. We show that S1 = So. For
this we prove that every homogeneous element .r of So is in S1, arguing by
induction on the degree n of r. If n = 0, our assertion is obvious. Suppose
therefore I: > 0 and that our assertion has been proved when the degree of
P

.t is <n. As I e I, s = 4;) 5,51, where the J,’ are elements of S which can be
assumed to be homogeneous, with deg(s") = deg(:) — deg(.r.) < 11. Lemma 5
can be applied, as the g-module S is semi-simple (no. 5, Corollary 2 to Theorem
4-); in the notation of this lemma,
1’ 9

s = s“ = ‘21 (5.11)“ = (213ml.

The 5;“ are demerits of So which are homogenq and of degree <n (since
each Sn is a graded submodule). Hence theyare in 81 by the induction hypoth-
esis. Hence I 6 SI, which proves (a).
We now consider a simple representation of g of class 8 on a finite-dimen-
sional space M. Let L = 544M, S). For all 5 ES and all L, let {fbe the
element of L defined by (if/Mm) = :.f(m) (m E M); an S-module structure
is thus defined on L; as M is finite-dimensional over K, clearly L is a finitely
generated S-module and hence a Noetherian S-module since the ring S is
Noetherian. On the other hand, L has a canonical g-module structure. For
every integer n 2 0 let S" be the set of homogeneous elements of S of degree
n; then the g-module 3AM, S“) is semi-simple (no. 5, Corollary 3 to Theorem
4) and hence the g-module L is semi-simple. Moreover, for .t E S, f e L,
x e g and m e M,

(xt(5f))('n) = *a((#')(m)) —- (flown)

"s(‘~f(m)) - sforum)
(“53) -f("1) + I-(xsf('n)) - :.f(x,,m)
= ((xssm ("1) + (50:11))(”I)
and hence zLGf) = (xsx)f + J(x,_f). We can therefore apply Lemma 5.
The subset L, of invariant elements of L is just the set of homomorphisms
of the g-rnodule M into the g-module S. Hence, if dz denotes the canonical
homomorphism of M ®x L onto S, (MM ®K L0) = So. As (I) is obviously an
S-module homomorphism, it suffices to show that L0 is a finitely generated
So-module. Let J be the sub-S-module of L generated by Lo. Since L is a
Noetherian S-module, there exists a finite sequence (fl, . . .,fq) of elements
of L0 generating the S-module J. Let LI be the So-module generated by

67
 1 Lu! ALGEBRAS

q
fn...,f¢. Then Ll C Lo. On the other hand, iffeLo, thenf= .2141];
with I. E S for all i and hence by Lemma 5 whose notation we adopt: 7
‘1 H a

f=f~ = (2.50?) = Z. s'm 6 L1-

Hence L0 = L1, so that L0 is a finite generated So-module.
10. CHANGE OF BASE FmLD
Let K1 be a commutative attension of K. For a Lie algebra 9 over K to
be semi-simple, it is necessary and sufficient that But.) be semi-simple; for
the Killing form pl of gm], is derived from the Killing form (3 of g by extend-
ing the base field from K to K1; hence [51 is non-degenerate if and only if B
is non-degenerate (Algebra, Chapter IX,§ 1, no. 4, Corollary to Proposition 3).
If Sam is simpleJ g is semi-simple by the above and cannot be a product of
two non-zero ideals, hence g is simple. On the other hand if g is simple 9m)
(which is semi-simple) may be not simple (Exercises 17 and 26 (b)).
Let g be a Lie algebra and 2‘ its radical. Then rm) is the radical of Ban
(§ 5, no. 6). Therefore, if 3 denotes the nilpotent radical of g, the nilpotent
radical of Sun) is [gagm ram] = [9, ’10:» = sum. It follows that g is reductive
if and only if Sam is reductive.
Let g be a Lie algebra and I; a subalgebra. Recall that a representation of
b is semi-simple if and only if the representation of {70(1) derived by extend-
ing the base field to K1 is semi-simple. Hence 1) is reductive in g if and only
if bug” is reductive in gm).
Now let K0 be a subfield of K such that [K2Ko] is finite. Let g be a Lie
algebra and go the (finite-dimensional) Lie algebra derived fi-om g by restrict-
ing the base field from K to K). Every commutative ideal ofg is a commutative
ideal of 9a; conversely, if a0 is a commutative ideal of go the smallest vector
subspace over K of 9 containing no is a commutative ideal of 3; hence g is
semi-simple if and only if 9., is semi-simple. If 90 is simple, clearly g is simple.
Conversely, suppose that g is simple. We show that go is simple. Let an be a
simple component of go. For all A e K‘, Ana is an ideal of go and
[00, Mo] = Alum “0] = Mo 9‘ {0},
hence Mo D an and therefore 10., = an since dimxooao) = dimKnao. Now the
vector subspace of 9 generated by no is a non-zero ideal of g and hence is the
whole of 9. Hence 9 2 no, which proves our assertion.

§7. ADO’S THEOREM

Rem” that K denote: a field 9/ characteristic 0 and that all Lie algebras are a:-
rumm’ In befiniIe—dimmional aver K.

68
 THE EXTENSION THEOREM § 7.2

l. COEFFICIENTS OF A REPRESENTATION
Let U be an associative algebra with unit element over K, U“ the dual of
the vector space U and p a representation of U on a vector space E. For
2 e E and c’ e E‘, let 0(2, 3’) E U* be the corresponding mgfinlmt of 9 (Algebra,
Chapter VIII, § 13, no. 3). Recall that 0(e, :’)(x) = (900:, z’) and that the
mapping 9 >—> 0(2, 2') is for fixed 6’ a homomorphism of the U-module E into
the U-module U“ of the coregular representation of U ([00. 011., Proposition
1); therefore the vector subspace of U“ generated by the coefficients of p
(a subspace which we shall denote by C(p) in this paragraph) is a sub-U-
module of U*. If (51),“ is a family of elements generating E“ over K, the
mapping 2 ~—> (0(9, 22)) is an injertive U-homomorphism of E into C(p)‘, for
0(:, 2;) = 0 for all 1' implies (0,4) = <p(l)e,efi> = 6(5, 90(1) = 0 for all i
and hence a = 0.
In particular, if U is the enveloping algebra of a Lie algebra g and p is a
representation of 9 (identified with a representation of U) on an n-dimen-
sional vector space E, the g-module E is isomorphic to a sub—g-module of
(0(9))"-
2. TH}: EXTENSION THEOREM
Let g = I; + g' be a Lie algebra which is the direct sum of an ideal g’ and
a subalgebra I), U the enveloping algebra of g and U’ C U the enveloping
algebra of 9’. There exists one and only one 9-module structure on U’ such
that: (a) for x s 9' and u eU', xwu = —ux; (fl) for x5!) and u e U’,
xuru = 1m — wt (the latter element is certainly in U’ since the inner deriva-
tion of U defined by x leaves 9’ and hence U’ stable). For conditions ((7.) and
((3) define uniquely a linear mapping max", of g into S’K(U’). It therefore
sufiices to Verify that [x, 31h. = [arm guy]; it is only necessary to consider the
following cases:
(1) x E g’,y e 9’: then

[Ic’yluiu = 4W - w) = (”M/tr - yaw)“;

(2) xel),yeg’: then
Mylo" = -"(xy - W) = x(-"y) - (”1.10” + (W — WW
= (“u'yu' — WWW)“;
(3) x e b, y E b: then [1, fl". and [xm 31“.] are two derivations of U’ whose
restrictions to 9' coincide with those of adg[x,y] and [ad°x,ad¢y]; hence
these derivations are equal.
We shall also consider the dual representation x >—> —‘xu. of g on U'*. For
x e g’, —'xl,« is the transpose of right multiplication by x in U’; the corre-
sponding representation of U' is therefore the coregular representation of U'.

69
I LIE ADGEBRAS

DEFINITION 1. Let g be a Lie algebra, g' a subalgebra of g and p' a representation

q’ on V’. A representation p 9/9 on V is called an extension qf p’ to 9 if there exists
an injectiue homomorphism if the g’-modale into the g’-module V. We also say that
the g-module V is an extension of the 9’-module V’.

If p’ is finite-dimensional and g’ is a solvable ideal of 9, it is necessary for

the existence of a finite-dimensional extension that [9, 5’] be contained in the
largest nilpotency ideal of 9’ (§ 5, no. 3, Theorem 1).
THEOREM 1 (Zassenhaus). Let g = g’ + I) be a Lie algebra which is the direct
sum 1y” An ideal g’ and a subulgebra I) and p' a finite—dimnsional representation of 9'
whose largest nilpotency ideal contains [b, 9’].
(a) There exists a finite-dimensional extension p {If p’ to 9 whose largest nilpotertey
ideal eantains that of p’.
(b) Iffor all x El) the restriction to g’ y” adnx is uilpotertt, p can be ehosert sueh
that moreover the largest nilpoteruy ideal of p eontains I).
Let U’ be the enveloping algebra of 9'. Suppose that U’ and U’* have the
g-module structures defined at the beginning of this no.
U” Let I C U’ be the kernel of 9' (identified with a representation
U of U’). It is a two-sided ideal of U’ of finite codimension. The
S subspace C(p’) of U" (cf. no. 1) is orthogonal to I. Let S be the
U sub—g-module of U" generated by C(p').
C(P')
We now show that S isfinite-dimensional over K. Let V’ be the space on which
9’ operates and V’ = V3 D V’l D . ~ ~ 3 VI, = {0} a Jordan-Holder series
of the g’-modu1e V’. Let 9" be the representation of g' on V,’_1/V{ derived
from p’ (1 S is d). Let I' C U’ be the intersection of the kernels of the
9" (identified with representations of U'). Then
I'dCICI’

and I’ n g’ is the largest nilpotency ideal of p'. By §2, no. 6, Corollary to

Proposition 6, I" is of finite codimension in U’. If xeb, the derivation
u >—>xu — we of U’ maps 9’ into [9, g'] C 1’, hence U’ into I’ and hence 1’“
into 1’“. On the other hand, clearly I'“ is a mow-module of U’. Hence
I" is a sub-g-module of U'. The orthogonal of 1’“ in U" is a finite-dimensional
sub-g-module which contains C(p') and therefore S. This shows that S is
finite-dimensional over K. For M: l' n g', x“ is obviously contained in the
annihilator of the g-module U’II" and hence also in the annihilator of the
g-module S.
We saw in no. 1 that the g’—module V' is isomorphic to a sub-g’-module
of a product (C(p'))". Hence the g-module S" provides a finite-dimensional
extension 9 of p’ to 9. Moreover, p(x) is nilpotent for x E I’ n g'; as I' n g'

70
 Aoo’s THEOREM § 7.3

is an ideal of g (for it contains {I}, g’] by hypothsis), we see that I’ n g’ is

contained in the largest nilpotency ideal of 9. Thus (a) is proved.
Suppose finally that for all are!) the restriction to g' of ad“): is nilpotent.
As the elements of I) operate on U’ by derivations, there exists for all u s U’
and all x e!) an integer e such that (1”,)‘Jz = 0; the endomorphisms derived
from xv. on U’II'“ and on S (which are finite-dimensional spaces) are there-
fore nilpotent. Thus p(x) is nilpotent for all x e I). We have seen earlier that
this is also true for x e I’ n 9'. As I’ n g' is an ideal of 9’ containing [1}, 9’],
the sum 1) + (I’ n 9') is also an ideal of g. Assertion (b) of Theorem 1 then
results from the following lemma:

Lemma L Let g = 9’ + l) be a Lie algebra which i: the rum qfan ideal g’ and a
mbalgebm I). Let a be a finite—dimensional representation cy‘ g. Suppose that u-(x) i:
nilpotentfor all x e g’ and all x E 9. Then 6(2) is nilpotentflzr all x e g.
Passing to the quotient by the kernel of a, a may be assumed to be faithful.
Then 5’ and l) are nilpotent and hence g, which is an extension of a. quotient
of b by g', is solvable. Then 1) and g’ are contained in the latgect nilpotency
ideal ofa‘ (§ 5, no. 3, Corollary 6 to Theorem 1).

For an improvement of Theorem 1, cf. Exercise 4.

3. ADO’S THEOREM
PROPOSITION 1. Let g be a Lie algebra, n its largest nilpatent ideal, a a nilpatent
ideal of g and p afinite—dimerm‘mal representation of a mh that every element of p(a)
Lt nilpotent. Then p admit: afinite-dimensional extension 0" to g such that every element
of 6(a) is nilpotent.
Let a = no C n, C - - - C n, = n be a sequence of subalgebras of“ such
that "(.1 is an ideal of n. of codimension l for 1 S i S [1 (§ 4, no. 1, Proposi-
tion 1 (e)). The algebra n, is therefore the direct sum of 1a-, and a l-dimen-
sional sub-algebra. As adnx is nilpotent for all x e II, it is possible (Theorem l)
to find one by one finite-dimensional extensions 9;, pg, . . ., pp = p’ of p to
n1, n2, . . ., II, = n such that every element of p’(n) is nilpotent.
Let r be the radical ofg and let tt = to C r, C - ~ ~ C I. = r be a sequence
of subalgebras of r such that 1}_1 is an ideal of I; of codimension 1 for
l s 1' $ q (§ 5, no. 1, Proposition 2 (d)). The algebra II is thus the direct
sum of r1_l and a l—dimensional subalgebra. As [r,t] C II, it is possible
(Theorem 1) to find one by one finite-dimensional extensions
9’1, pg, . . ., p; = p” of p' to r,, 12,. . ., ta = r such that every element of
p”(n) is nilpotent.
Finally 9 is the direct sum of r and a subalgebra 3 (§ 6, no. 8, Theorem 5).
As [3, r] c n, it is possible (Theorem I) to find a finite-dimensional exten-
sion 6 of p” to 9 such that every element of o(n) is nilpotent.

71
! LIE AMEBRAS

THEOREM 2. Every Lie algebra has a finite—dimerm'onal faitlgfixl linear represenla-

lion.
More precisely:
THEOREM 3 (Ado). Let g be a Lie algebra and n its- largest nilpotenl ideal. There
exixls a finite-dimmtiomlfaitlgfid representation 9 nf 9 Huh that every element qf p(n)
i: nilpatent.
The l-dimensional Lie algebra K admits finitdimensional faithful repre-
sentations 1' such that every element of 1(K) is nilpotent, for example the '
representation
A» 0 0 '
A 0
It is easily deduced that the centre c of g, which is a product of l-dimensional
algebras, admits a finite-dimensional faithful representation a such that
every element of a(c) is nilpotent. Let a, be a finite-dimensional extension
of a to 9 such that every element of u1(n) is nilpotent (Proposition 1); if I
denotes the kernel of 61, then In t = {0}. On the other hand, let 0'; be the
adjoint representation of 9, whose kernel is c; every element of 52(n) is nil-
potent. The direct sum 9 of 03 and a“; is finitedimensional; every element of
p(n) is nilpotent; and the kernel of p, contained in t and in c, is zero, so that
p is faithful.

72
 EXERCISES‘

§1
1. Let g be a Lie algebra and a and b ideals (rap. characteristic ideals)
of g. Then the set of: E 9 such that [3, b] C a is an ideal (resp. a characteristic
ideal) of 9 called the hampwter of b into a. Show that «“19 is the transporter
of g into «,9.
2. If g is an n-dimensional Lie algebra over a field K and the centre of
g is of dimension 2n — l, g is commutative.
3. Let M be a not necessarily associative algebra over a field K, (9‘)“,
a basis of the vector space M and cm: the constants of structure of M relative
to the basis (2‘). For M to be a Lie algebra, it is necessary and sufficient that
the ow satisfy the following conditions: (a) ”me = 0; (b) 5% = —c,.,‘; (c)

re! (’m‘rkx + cikrcru + fkir‘m) = 0 {01‘ all 1', j, k, [in I.

4-. (a) Suppose that 2 is invertible in K. Let a be a Lie algebra over K.
On the K-module a’ = a x a we define a bracket by the formula:
[0% 7‘s): (gnyzll = ([xrwi] + [*2,y2],[x1.y2] + [rad/ill-
Then a’ is a Lie K-algebra and the mappings in» fix, at), an» fix, —x) are
isomorphisms of 0 onto ideals b, c of a’ of which a’ is the direct sum. (Consider
the quadratic extension K’ of K with basis 1 and I: where k2 = 1. Form
am) and then restrict the ring of scalars to K. Then observe that -}(1 + k),
<}(l — k) are idempotents of K' and that (l + k)(l — k) = 0.)
(b) Let a be a real Lie algebra, g = am and b the real Lie algebra derived
from g by restricting the field of scalars to R, Show that 6(a) is isomorphic
to afc), where a' is the algebra introduced in (a). (Consider V = C ®3 C as
a vector space over C by (z® z')z” = 1 ® z'z'. Observe that V is the
complexification of the real vector subspace of V generated by 1 ® 1 and
1' ® 1' and that this subspace is identified with the quadratic extension of R
with basis I and k where II“ = l.)

73
I LIE ALGEBRAS

5. Let V and W be vector spaces of dimensions n + l and 71 over a field

K. Show that 91(W) is isomorphic to a subalgebra of s[(V)i (W can be as-
sumed to be a hyperplane of V. Let a EV, 2 ¢ W. For x E 91(W), let flat) be
the element of gl(V) which extends x and is such that y(e) = —Tr(x).e.
Show that the mapping 2: >—>y(x) is anisomorphism of gl(W) onto a sub-
algebra of 51(V).
6. For a Lie algebra g to be associative, it is necessary and sufficient that
99 be contained in the centre of g. -
fl 7. Let A be a ring (possibly without unit element) such that the relation
2a = 0 in A implies a = 0. Let U be a subring of A and an ideal of the Lie
Z-algebra associated with A.
(a) Ifx, y belong to U, then (:13; — yx)A C U (write:
W - W): = W!) - (wlx - W“ - 515))
and A(xy —yx)A C U (write:
r(xy - w): = (x1! — 11mr + v[(xy — ms} - [(xy — yx):]r)~
(17) Suppose that U is commutative. Let e, 56A and y = x5 — Ix.
Show that (yt)3 = 0 for all teA. (The element x commutes with x: — 5x
and xx” — r’x; deduce that 2(x: — 5:03 = 0, whence y“ = 0. Similarly
(y: —- ty)’ = 0 for all 16A.)
(5) Suppose henceforth that the only two-sided ideals of A are {0} and A
and that for every non-zero element y of A there exists [EA such that yt is
not nilpotent. Show that, if U 95 A, U is contained in the centre of A. (Show,
using (a), that U is commutative and then use (b).)
(d) Let V be an ideal of the Lie Z-algebta associated with A Then either
V is contained in the centre of A or V D [A, A]. (Apply (5) to the set U of
tsA such that [t, A] C V; use the identity [33], z] = [x, ya] + [g], zx].)
8. Ifxl, x,, x3) x, are 4- elements of a Lie algebra, then
“131: "Rla 7‘3]: x4] + [[[xza J‘1]: *4]: *3]
+ [[["5: x4], x1], x3] + [[[x,, ”all: x,], xi] = 0-
9. Let g be a Lie algebra in which [[x, y], y] = 0 for all x and y.
(a) Show that 3[[x,y], z] = 0 for all x, y, z in 9. (Observe that [[x,y], z]
is an alternating trilinear function of (x, y, z) and apply the Jacobi identity.)
(b) Show similarly, using (a) and Exercise 8, that [[[x,y], z], I] : O for
all x, y, z, t in g.

10. Let L be an associative algebra and g the associated Lie algebra.

Every derivation of L is a derivation of 9, but the converse is not true. (Con-
sider the identity mapping of a commutative associative algebra.)

74
 EmCISFJ

11. Let g be a Lie algebra and I) an ideal of 9 such that 5% = b. Show that
I) is a characteristic ideal of g.
12. Let g be a Lie algebra.
(a) For a derivation D of g to commute with all the inner derivations of
9, it is necessary and sufficient that D map 9 into its centre.
(b) Suppose henceforth that g has centre zero, so that g is identified with
an ideal of the Lie algebra 9 of all derivations of 91 Show that the centralizer
of g in $9 is zero (use (11)). In particular, the centre of 9 is zero.
(0) Show that a derivation A of 9 such that A(g)= {0} is zero.
E[)A§@), 9] C Aug, 9]) + [9 MED] C A(£1) = {0} and hence A(‘9)= {0} by
((1) Show that, if further 9—
= 99, every derivation A of 9 is inner (use
(5) and Exercise ll).
134 Let g be a Lie algebra and a and b two submodules of 9. For every in-
teger i 2 0 we define the submodules m. = m‘(a, b) and m = "‘01, b) as
follows: m1 = b, rnH1 = [m‘, b], 111 = a, n“r1 = [11” b]. Show that
[0, m] C "m
For i = l, 2, . . . (argue by induction on i, observing that
"Ml“, 5], b) = "110‘. 1’)
and ME“, b]: b) = mm, b))-
1T 14-. Let a be a Lie algebra and b a subalgebra of a. A campadtion sm'u
joining a to b is a decreasing sequence (“Quin of subalgebras of a such that
:1n = a, c1,l = b, a‘+1 is an ideal of a, for O < i < n. b is called a :ubinvuriant
subalgebra of u if there exists a composition series joining a to b.
(a) Let b be a subinvariant subalgebra of a and 010“,“ a composition
series joining a to b. Deduce from Exercise 13 that [W‘wlg a] C ’fl’b by ob-
serving that [an b] c «”1 for l s i < m
(b) Deduce from (a) that the intersection @“b of the if"? (p = 0, l, 2, . . .)
is an ideal of a.
(o) Deduce from (b) that a subinvariant subalgebra c of a such that c = 9:
is an ideal of a.
(d) When K is a field and dimKn < +00, show that the intersection
9“!) of the 9% (p = 0, 1, 2, r r .) is an ideal of a. (Show that this intersection
is a subinvariant subalgebra of a and apply (5).)
11 15. (a) Let a be a Lie algebra, b a subalgebra of n, c an ideal of b and
3 the centralizer of c in n. Then [3, b] C 3. Deduce that b + A is a subalgebra
of a in which 3 is an ideal.
(b) Let g be a Lie algebra, a a subinvariant subalgebra of g and b an
ideal of a. If the centralizer of b in a and the centralizer of a in g are zero,

75
1 LE ALGEBRAS

the centralizer a of b in g is zero. (Let I) = a + 3, which is a subalgebra by

(a) ; a is subinvariant in I); if a aé (0), then 0 ¢ 9, hence a is an ideal in a’,
witha ¢ a’ C b; let a’eu’, a’¢a and a' :11 + z, aea, 163, 2 ye 0; show
that [2, a] is contained in a and commutes with b, whence [2, a] = (0) giving
a contradiction.)
(a) Let g be a Lie algebra with centre zero, 91 the Lie algebra of deriva-
tions of g, 93 the Lie algebra of derivations of $1, . . . g is identified with an
ideal of @1, 591 with an ideal of 9,, . . . (cf. Exercise 12 (11)) Using (b) and
Exercise 12 (b), show that the centralizer of g in S). is zero for all i. (For a ‘
sequel to this exercise, cf. §4, Exercise 15.)
16. Let g be a Lie algebra and Q the Lie algebra of derivations of g. The
identity mapping of 9 defines a semi—direct product b of S) by 3 called the
holomorph of g. g is identified with an ideal and 9 with a subalgebra of I).
(a) Show that the centralizer 9* of g in I) is the set of ad, x — x(x e g).
(b) For every element x + :1 of!) (x e 9, de 9), we write
6(x+d) =adflx—x+d.
Show that 0 is an automorphism of b of order 2 which maps 9 onto 3*, so
that b can be considered as the holomorph of g“.
(t) For every ideal 2 ye {O} of b, In (9 + 9*) 95 {0}. (If tn 9 = {0),
I C 9'“)
(d) If g is commutative, every ideal t 96 {0} of b contains g. If K is a field
and dimxg < +m, deduce that, if t aé g, I) cannot be the holomorph of I
(use (b)).
17. Suppose that K is a field. Let T be an indeterminate and L = K((T));
L has canonically a valuation (order of the formal power series) which makes
it a complete valued field.
(a) Let V be a finite-dimensional vector space over K. Then Vm has
canonically a complete metrizable topological vector space structure (Tapa-
logiml Vector Spares, Chapter I, §2, no. 3, Theorem 2). Let (x,),ez be a family
of elements of V such that x, = 0 for [7 less than some rational integer. Then
+m
the series .2» xpT’ converga in Va.) and every element of Va“, can be ex-
pressed uniquely in this way. (Take a basis of V.)
(b) The vector space BMW”) = $K(V))m has a canonical complete
metrizable topological vector space structure. Every endomorphism of Va.)
+ u)
can be expressed uniquely in the form 2; up ”, where u, e ZK(V) and the
up are zero for p less than some rational integer.
(r) If K is of characteristic 0 and u e .‘Z’K(V), we write
1 l
5'“: l +fiuT+fiu2T2 + ---e.‘Z’L(Va,,).

76
 EXERCISES

11'e is such that ux = 0, then cT".x = x.

(d) If W is another finite-dimensional vector space over K and v e .YAW),
then eT(u®1+1®u) = en (8 a“.
(e) Deduce from (c) and (d) that, if V has a (not necessarily associative)
algebra structure and u is a derivation of V, then 9'“ is an automorphism of
the algebra Vu.)~
1T 18. Let 8 be a complex Hilbert space, Y and Z continuous Hennitian
operators on 3 and B = [2, Y]. Suppose that Y and B are permutable.
(a) Show that [Z, Y"] = nBY"‘l (argue by induction on n).
(b) Let f be a continuously difi‘erentiable function of a real variable.
Deduce from (a) that [Z,f(Y)] = Bf’(Y).
(o) Deduce from (b) that B = 0. (Show that, if B ye 0, f can be chosen
such that llf(Y)|| g l and ||Bf’(Y)]| is arbitrarily large.)
((1) Let g be the Lie algebra of continuous operators T on a such that
T" = — T. Deduce from (c) that the conditions Ye 3, Z 6 9, [[Z, Y], Y] = 0
imply [2, Y] = 0.
19. (a) Let L be an associative algebra over K. Given 1: elements
an, . . ., x» ofL, we write:

fk(x1) - - ., xx) = ‘2' "am - - - ”am-

In the algebra L ‘81: K[T,, . . ., Tk], where the T, are indeterminates,
110:1, . . ., am) is the coefficient of T1 . . .T,‘ in (xlT, + - - - + kk)"; it is
also the coefficient of TJ . . . T,‘ in:
k-l ,
’20 (MT: + ' ' ' + xk—lTk—l),kk(xlTl + ' ‘ ‘ + xlc—lTk-lyc-l—l'

Given two elements at, y in L, we define g._,,_1(x,y) as the coefficient of

T‘lT'g‘" in (2T, + yTz)”. Show that
i!(k — i)!gx.k—¢(x,y) =fie(’1, - - -, ‘1‘)
wheret,‘ =xforh< iandlh =yforh>i+ l.
(b) If L is considered as a. Lie algebra over K, then in $K(L):

(ad:r)'I = {Z (_I)VI-l(n) L:‘R:-l’

1'
where L,‘ (resp. R,) is the multiplication y »—> xy (rap. g n).
Deduce that, if K is of characteristic 1: (p prime), then:
(1) (adxw = (ad m
(2) (ad x)"1.y = [:20 x'yzv-l-I.
77
I LIE ALGEBRAS

Deduce from (2) that:

fin—10m x1) - - u ad xp-l) -!l =.fn(x1:‘ - -, xp—lry)
and: _
g,_,_1_.(adx, ady)._1/ = (P “ ”gm—«(fill)-
Conclude that for any two elements 5:, 3/ of L:

(3) (x + y)” = 1'” +y” + Army)

where:

A,(x,y>= 2, (p — 0- g. .-1--.(adx.ady) y
belongs to the Lie subalgebra of L generated by x and y (“Ja€ab:on’:finnulaz”).
20. Let g be a Lie algebra over a ring K such that pK = {0} (p prime).
A mapping maxi" of 9 into itself is called a p—mapping if it satisfies the rela-
tions
ad 1“" = (ad x)’
(MOM = NW”

(at + 30”" = x“ + y'”1 + A,(x,y) (cf. Exercise 19 (b))

for x, y in g and A e K. A Lie p-algebra over K is a set with the structure de-
fined by giving it a Lie algebra structure over K and a p—mapping. Every
associative algebra over K is a Lie p—algebra with [x,y] = xy — yx and
x"! = x” (Exercise 19). A mapping a of a Lie p-algebra 9 into a Lie p-algebra
g' is called a p-horrwmorphism if u is a Lie algebra homomorphism and
u(x“") = (u(x))m. Show that in a. Lie p-algebra 9 every p-mapping is of the
form act—”cl“ +f(x), where f is a mapping of 9 into its centre, which is
semi-linear for the endomorphism A ~—> A” of K. More generally, if u is a homo-
morphism of 5 onto a Lie p-algebra g’, u(x‘") —— (u(x))m belongs to the
centre of g’.
21. (a) Let K be a field of characteristic [1 > 0 and L a not necessarily
associative algebra over K. Show that the derivations of L form a Lie 1:-
algebra with the p-mapping D >—> D”.
(b) Let L be the algebra K[X]/(X") and m the p-algebra of derivations
of L. Let D( (0 < i S p — 1) denote the derivation of L such that D‘(X)= X‘.
Show that the D1 form a basis ofw over K and that [D,, D,]: (j — i)DHP 1
ifi+j<p, [D.,D,] =0 ifi+j>[1, and D.’=O, except for i: 1, in
which case D” = D.
(0) Show that, ifp> 3, the Lie algebra in only admits the ideals {0} and
w.(Using the multiplication table of the D,, show that every non-zero ideal
of to contains a non-zero multiple of D,_,.)
(11) Let V,, be the subspace of to with basis the D1 of index 2k. Show that,

78
 EXERNSE

if p 2 5, V,‘ is identical with the set of Z 6 to whose centralizer is of dimen-

sion 2]: + l. Deduce that, ifp 2 5, the automorphism group of!» is solvable.
22. Let g be a Lie p—group over a ring K ofchar‘acteristic p (p prime).
Let x >—> x” denote the p—mapping. For every subset E of 9, let E” denote the
.mbmodule tgf 9 generated by the x”, where xe E. An ideal a C 5 is called a
p-idzul if a” C a.
(a) For an ideal a C Q to be a p—ideal, it is necasary and sufficient that it
be the kernel of a p—homomorphism (Exercise 20).
(b) The sum of two fi—idenls is a p—ideal. The sum of a p—ideal and a p-
subalgebra is a p-subalgebia.
(a) Let I) be a Lie subalgebra of g. The smallest Lie p—subalgebra contain-
ingbisb+b’+b"+ =$;ifwewrivcbr=b+b’+ +b",b.is
an ideal of 5 and 5/!) is commutative. If D is an ideal of 9, so are the b, and 5
and the latter is the smallest p—ideal containing 1).
(d) The normalizer and centralizer of an arbitrary submodule of g are
p—subalgehras of g.
23. Let g be a finite-dimensional commutative Lie p—algebra over a perfect
field K of characteristic [1 > 0.
(a) Show that g is in a unique way the direct sum of two p—subalgebras
I}, t such that on I) the p-mapping is bijective and on t it is nilpotent (con-
sider the iterations of the p—mapping on 5; cf. Algebra, Chapter VIII, §2,
no. 2, Lemma 2 and Chapter VII, §5, Exercise 20 (a)). b is called the p-
:or: of g.
(b) Suppose that K is algebraically closed. Show that there exists a basis
(6,) of I) such that of = e‘ for all i (consider a p-subal'gebra of [3 generated by
a single element and show that it contains some at such that x” = x aé 0; then
argue by induction on the dimension of b). Deduce that the [z-subalgebras
of b are finite in number and are in one—to-one correspondence with the
vector subspaces of the vector space over the prime field F, generated by
the e‘.
(0) Show that there exists a basis (12,) of! (I S i S k, l s j S st for all i)
where the sequence of 5‘ is decreasing and which is such that ff, = 0 for
lsj S ruff, =f._u for 2 g is k, l sj S .r‘ (method ofAlgebra, Chap-
ter VII, §5, Exercise 20 (b)).
1} 24. Let K be a (commutative) field of (arbitrary) characteristic p and
Xb Y“ Z‘ 37: distinct indeterminates; to abbreviate we shall denote the sys-
tems (Xi)1<l<n) (Y()1<l£nr (zt)rsl<n by x, Y: 1 rapectively- Let Ax,y,x be
the algebra of formal power series in the 371 indeterminates X1, Y" Z; with
coefficients in K; we shall denote by A,” (resp. Ax) the subalgebra of Ax,y,z
consisting of the power series not containing 1 (resp. y, I). An element of
Ar... (resp. Aw) will be denoted by “(1,352) (rap. u<x,y>,u(x>);
79
I LIE ALGEBRAS

a system (u,(x, y, 2)) I g, Q, ofn elements of.” will be denoted by u(x, y, z);
analogous notation will be used for formal power series containing only one
or two of the three systems x, y, 1. If u(x, y, z) EA,_,,_z and f = 0;), g =
(g), h = (11,) are three systems of n elements of Ax,” which are formal
power series with no constant term, we denote by u(f(x,y, z), g(x,y, z),
h(x, y, 1)) the {urinal power series obtained by substituting fi for X1, g, for
Y” )1, for Z, (1 < i < n) in It. For every system at = (on, . . ., at“) ofn integers
20, let x“ denote the monomial X}: . . .i; y“ and z" are defined simi-
larly‘ Let 2‘ denote the system a: for which at, = 0 ifj ye i, on = 1. We denote
by e the system (0, . . . , 0) of n elements of Am”.
(a) A foimal gnu}? [am over K (or, by an abuse of language, a formal group
over K) of dimension n is a system G = f(x, y) of n elements of A,” with
the following properties: (1) {(x, f(y, 1)) = f(f(x,y),z); (2) He, y) = y,
f(x, (5) = x. Show that necessarily fi(x, y) = XI + Yi + g.(x, y), where g.
contains only monomials of degree >2, each of which contains at least one
X, and at least one Y,-. Show that there exists one and only one system h(x)
of n elements of A, such that f(x, h(x)) = f(h(x), x) = e (Algzbra, Chapter
IV, § 5, no. 9, Proposition 10). G is called commutative if f(y, x) = f(x, y).
(11) For all u 5A,, let L,u denote the element u(f(y, x)) of Ax”. A deriva-
tion D of can be canonically extended to a derivation MA,” (also denoted
D by an abuse of notation) by the condition that D(Yi) = 0 for all 2' (Algebra,
Chapter IV, §5, no. 8, Proposition 6). D is said to be left invariant for the
formal group G in question if L,.D = BL," We denote by D“) the linear
mapping of Amy into A, defined by D‘°’(u) = (Du) (e, y); it maps A.x into
K and is determined by its restriction Ax. Show that, for D to be left invariant,
it is necessary and sufficient that for all u E A,, D‘°’(L,v) = (Du)( y).
Let D. be the derivation a/ax. of Ax (1 g 1' < u); there exists one and
only one left invariant derivation T‘ such that T5” = DP. Show that the
T, are linearly independent over K and that every left invariant derivation
is a linear combination of the T. with coeflicients in K. Deduce that, for
the bracket [D, D’] and the p-mapping D |—> D” (when [J > 0), the set 5; of
left invariant derivations is a Lie algzbra (a Lie p-ulgebnz if 1; > 0) called the
Lie algzbra 11f theformal Lie group G.
(6) Show that for every formal power series 14 2 AK:

(1) mm» = um) + Z Ymu + Mm)

where in the series I), all the monomials are of degree 22 in the Y}. Deduce
that

(2) u(f(x,f(y, 2») = aw) + Z (Y. + 2m

+ siurirjxu» + oak, y, z)
 EXERCISES

where in the series as all the monomials are of degree 23 in the set of Y‘ and
2‘. Deduce that:

(3) "(£08m - "(f(Y,=)) = [<1

2 (XrY: - X:Y«)([TuT/]‘°’(u)) + 06(x, y)
where all the terms ofoé, are of total degree 2 3. Show that, ifG is commutative,
g is commutative.
(0') Let G’ be another formal group of dimension m, whose group law is f'.
A formal hamomorphism of G into G' is a system F = (F,(x)),‘,‘,,l of elements
of such that f’(F(x), F(y)) = F(f(x, y)). For every left invariant derivation
D E 9, show that u r—> D(u o F) is a left invariant derivation belonging to the
Lie algebra g' of G’. If it is denoted by l“ (D), F“ is a homomorphism (a p-
homomorphism forp > 0) ofg into 9’. If (T,)K,‘,., (TDKKM are the bases of
g and 9’ such that Ti" = D?" and T?” = D§°’, show that the matrix of 1"“
relative to these bases is the matrix (BE/(9X00 (constant term of amax”
where i is the index of the rows and j that of the columns).
25. The fimrml linear gruup in n variables over the field K, denoted by
GL(n, K) (when no confusion can arise), is the formal group of dimension 7:“

whose group law is defined byf,,(u, v) = —8‘, + 2:1 (8”,: + Um)(8k, + V”)
(8,, the Kronecker index; the fin2 indeterminate: of the general definition of
Exercise 24- are here denoted by U”, V”, Wu with 2 indices varying from 1 to
n). Show that, if D” = 6/6Uu, the left invariant derivations X” such that
Xi;’ = D}? are given by

Xi; = (1 + UulDu + ,; UMDkI'

The Lie algebra ofGL(n, K) is identified with gl(n, K) by identifying Xi, with
the element Eu of the canonical basis (use formula (3) of Exercise 24). If K is of
characteristic [2 > 0, X3 = X“ and X" = 0 fori aé j.
11' 26. (a) Let K be a field of characteristic #2, E an n-dimensional vector
space over K and d) a non—degenerate symmetric bilinear form on E. Suppose
that E has an orthogonal basis with respect to II) and let R denote the (diagonal)
matrix of (I) with respect to this basis. We know (Algebra, Chapter IX, §4,
Exercise 1 1) that every matrix (relative to the basis in question ) U of the ortho-
gonal group G((D) such that det(I + U) aé O can be written
U = (I— R‘IS)“(I+ R'1S),
where S = R(U — I)(U + I)" is an alternating matrix such that
det(I — R‘JS) aé 0; conversely, for every matrix S satisfying these con-
ditions, U = (I — R"S)“(I + R‘1S) is a matrix of G(<D) such that
det(I + U) 96 0. Let S”, S.’,, SQbe3n(n —— l)/2 indetenninates (l S i < j < n)

81
I LIE ALGEBRAS

and let L be a field containing the ring of formal power series K[[Si,, S,’,, SM].
If we denote by S, S’, S” the matrices

2 sum, — En),
I<l «<12 Sam — fl), 2, saw“ — Efl),
then det(I —— R"S) aé 0 and the analogues for S’ and S” hold; ifwe write
U = (I — R“S)"(I + R'IS), U' = (I — R“S’)“(1 + R'ls’),
then also det(I + UU') aé 0. We write
F(S,S') = R(UU’ — I)(UU' + I)’1 = (fi,(S,S’));
the fi,(S, S’) belong to the ring of formal power series K[[S‘,, S{,]] and, by
consideringF (S, S’) as a. formal power series with coefficients in the algebra of
matrices of order n over K, we can write:

F<s, 5') = s + S' + ms, S')

where in the elements of the matrix 02(S, S’) the terms are all of degree 21
in the S‘, and of degree >1 in the 51,; then similarly:
U = 1 + 212—13 + am)
the elements of a;(S) being formal power series of order 22. Show that
F (S, S’) is a formal group law over K of dimension n(n — l)/2; this formal
group is called the formal orthogonal gnu]: (corresponding to (D) and is denoted
by GOD) by an abuse of notation.
(17) If we write M(S) = (rm,(S)) = (I— R"S)"(I + R"S), M is a
formal homomorphism of (2(0) into GL(n, K). Let 0(0) be the Lie algebra of
G(<D) and let (T‘,),<, denote the basis of this algebra such that T? = Di?
(cf. Exercise 24 (d)); if an element D = 2; cuT‘, of 001)) is identified with the
matrix D = «<21 r‘,(Eu — E,.), show that (in the notation of Exercise 24- (d)
with the identification made in Exercise 25 of the Lie algebra of GL(n, K)
with gl(n, K)):
M*(D) = 2R'1D.
Deduce that M‘ is an isomorphism of 0(EI)) onto the subalgebra of 91(n, K)
consisting of the matrices X such that ‘XR + RX = 0 (for every ordered pair
(a, b) of elements of E, identified with matrices with one column, consider
(I)(a + M(S).a, b + M(S).b) = ‘a.(I + ‘M(S))R(I+ M(S)).b as a formal
power series reduced to its constant term and use the fact that the subalgebra
ofXE gl(n, K) such that ‘XR + RX = 0 is ofdimension n(n — l)/2).
(5) Define similarly the formal .tymplmir group in 27; variables over an arbi-

82
 EXERCISES

trary (commutative) field and show that its Lie algebra is identified with the
subalgebra of 91(2n, K) consisting of the matrices X such that

0 1».
‘XA+AX=0,whe1-e A:( —I,| 0
27. Let K be a field of characteristic p > 0 and G the 2—dimensional formal
group defined b1(xx Y) = xi + Y: + X1Y1:f2(x) Y) = X: + Yaa + Xl')
Show that G is not commutative but that its Lie algebra is commutative.

§2
1. We adopt the notation T, J, T,I of Definition 1 and no. 6. Let t be a
homogeneous tensor of T of order p and a a permutation of{l, 2, . . . , [1}. Then
I — at EJ + T,_1. (Reduce it to the case where a is a transposition of two
consecutive integers.)
2. Suppose that K is a field. Let g be a Lie K-algebra and U its enveloping
algebra.
(a) Let u, v be in U. Ifu is of filtration > n, and u of filtration >11, then uv is
of filtration >71 + 1) (use Theorem 1).
(b) Deduce from (a) that the only invertible elemenm of U are the scalars.
(c) Deduce from (b) that the Jacobson radical of U is zero.
3. Suppose that K is a field. Let g be a Lie K-algebra and U its enveloping
algebra. The left regular representation of U corresponds to a representation p
of g on the space U. Show that the set U+ of elements of U with no constant
term is stable under p and that U+ has no supplement in U stable under 9. In
particular, 9 is not semi-simple. (This will be considered again in Theorem 2 of

4. Suppose that K is a field.

(:1) Verify that there exists a 3-dimensional Lie algebra 9 with basis (x, y, a)
such that [x, y] = z, [x, z] = [y, z] = 0. LetU be the enveloping algebra ofg.
Show that the centre of U is the subalgebra generated by 1 and 2. (Consider,
for all 56 g, the derivation of the symmetric algebra S of g which extends
adog and look for the elements of S annihilated by these derivations; then
apply the final remark of no. 8.)
(1:) Verify that there exists a 3-dimensional Lie algebra g with basis (1, y, 2)
such that [x, y] = y, [x, z] = z, [y, z] = 0. Show that the centre of the envelop-
ing algebra of 9 reduces to the scalars. (Same method.)
5. Suppose that K is ofcbaracteristicp > 0 (p prime). Let g be a Lie algebra
over K with a basis, canonically identified with a submodule of its enveloping
algebra U. For an endomorphism <1; of the K-module g to be a p-mapping, it is
necessary and sufficient that x >—> x” — Mac) be a semi-linear mapping (for

83
1 LE ALGEBRAS

)t r—> A”) of 9 into the centre of U. Deduce that if (bA) is a basis of g, for there to
exist a p—mapping on 9, it is necessary and sufficient that, for all A, there exist
t,_ e 9 such that (ad bA 9 = ad 01; if this is so, there then exists one and only one
[I—mapping x >—> x!“ such that bi" = c. for all )t.
6. Let g be a Lie p—algebra over a ring K such that 11K = (0} (fl prime), U
its enveloping algebra and o' the canonical mapping of 9 into U. LetJ be the
two-sided ideal of U generated by the elements (c(x))" —< a-(xm) where 5: runs
through 9. The associative algebra U = U/_] is called the ”mined enveloping
algebra of g. The mapping cr defines when passing to the quotient a mapping 6-
(called canonical) of 9 into U, which is a p—homomorphism (when U is con-
sidered as a Lie p-algebra).
(a) The algebra U and the mapping 3 are solutions of a universal mapping
problem: for every p-homomorphismf of 5 into an associative algebra B over
K (considered as a Lie p-algebra), there exists one and only one K-homomor-
phism f' of U into B (with their associative algebra structures) such that
= ’ c 6-,
(b) Show that, if 01),,“ is a basis of 9 (where A is totally ordered), 5 is
injective and that, if x and 5(3) are identified for x e g, the elements l'lAbX‘
(where the A are in increasing order, the vi are zero except for a finite number
and 0 S v,_ < p for all A) form a basis of U. (Canonimlly identifying g with a
submodule of U, we write (Hz) = x’ —— x1" for all x 5 9. For every composite
index at = (05,) EN“), let at,‘ = 9,, + [In with 0 $ 5» < [I and let
Ta = (HibX*)(Ux(<l>(bx))“)~
Show that the Ta form a basis of U and the Ta such that Y = (7,.) 96 0 form a
basis of]; observe that the Mb») belong to the centre of U.)
7. Let g be a Lie p-algebra over a ring K such that [K = {0) (p prime). A
derivation D of g is called a p-derivalion if D(x’) = (ad x)"1. Dx for all x e 9.
Every inner derivation is a p—derivation.
(a) If L is an associative K-algebra, every derivation of L is a p—derivation
when L is considered as a Lie p-algebra. (Use formula. (2) of Exercise 19 of
§ 1.)
(1;) Suppose that g has a basis. For a derivation of 9 to be a fi-derivation, it
is necessary and sufficient that it can be extended to a derivation of the res-
tricted enveloping algebra of 9. Deduce that the p—derivations of 9 form a Lie
p-subalgebra of the p-algebra of derivations of g.
(c) If D is a p—derivation of g, the kernel of D is a p-subalgebra of g.
(d) For every derivation D of g, D(x’) — (ad x)”“. D): belongs to the
centre of g for all M; 9 (use formula (2) of Exercise 19 of § 1, applied to
L = 3%))-
8. Show that Theorem 1 remains valid if the module 9 is a direct sum of

84
 EXERCISES

monogenous modules. (Replace the module P in the proof by the symmetric

algebra of g.)
9. Let k be the field with two elements, V the vector space It“, (11, x2, x5) its
canoniatl basis and K the exterior algebra of V, which is an 8-dimensional
conunutative algebra over k. Let g be the Lie K—algebra admitting the basis
(51, 52, ts: 912; has ‘23) such that
[91: ‘2] = [‘2’ ‘1] = ‘12 [91: 33] = [53:21] = ‘15: [92:93] = [33,92] = ‘28
and the other brackets are zero. Let I) be the ideal of 9 generated by
u = #191 + 13:, + x395. As a module, 9 is generated by u, [11, :1] = 1:3!“ + x3e”,
["3 92] = J‘1‘1: + "a’aar [14» ’3] = 5151:: + xa‘aa- LCt
1’ = “1":‘12 + ”1"3‘13 + *fia‘za-
(a) Show that v é b. (Consider the K—linear form (I; on 5 such that
<M?) = (“92): (H‘s) = 0: <l’i‘iz) = In, “912) = x2, “‘23) = ”1-)
(b) Letf be a linear mapping of 9 into an associative K—algebra such that
f[(x,yl)= f(1)f(y) —f(y)f(*) for 311* y in 9 Show thatf(0) f('4)2
(r) Deduce from (a) and (b) that the canonical mapping of the Lie algebra
9/!) into its enveloping algebra is not injective.
10. Let g be a finite-dimensional Lie algebra over a field, U its enveloping
algebra and U,| the set of elements of U of filtration $71.
(a) Let x 6U“, y e U,I be two non-zero elements. Show that there exists
u e U, 1) eV such that w: = 9y. (Compare dim(U,x) = dim U,,
dim(U,y) = dim U, and dim U,.,,,. Deduce that p n Upy aé {0} for p
sufficiently large.)
(b) Show that U admits a field of left quotients (Algebra, Chapter I, §9,
Exercise 8) which is at the same time a field of right quotients.

§3
I. Let g be a Lie algebra, p a representation of g on a K-module M and u-
the associated representation of g on the tensor algebra of M. Show that the
submodule of symmetric tensors and the submodule of skew-symmetric
tensors are stable under a.
2. Let g be a Lie algebra, p a representation ofg on a K—module M and a the
associated representation of g on Q = .9’(M, M; M). Forf e Qto be invariant
under a, it is necessary and sufficient that the 9(x) be derivations of M with the
multiplication defined byf.
3. Let V be a finite-dimensional vector space over a perfect field K.
(a) The identity representation of gl(V) defines canonically representations
9 a

of gl(V) on ® V and ® V“ and hence a representation :4 »—> u: of 9l(V) on

85
I ”E AWEBRAS

V: = (® V) ® (® V“). Show that u: is semi-simple if u is. (Reduce it, by

extending the base field, to the case where K is algebraically closed.) Show that
u: is nilpotent if u is. Deduce that, ifs and n are the semi-simple and nilpotent
components of u, s: and n: are the semi-simple and nilpotent components of uf.
(b) Deduce from (a) that, if an element of V: is annihilated by 14:, it is
annihilated by r: and 712.
(r) Deduee from (b) and Exercise 2 that, if V has a not necessarily asso-
ciative algebra stmcture and u is a derivation of V, then s and n are derivations .
of V.
1] 4. Suppose that K is a field. Let 9 be a Lie K-algebra.
(a) Let M and N be g-modules, where N is finite-dimensional over K. Let
(fl)1‘1§n be a basis ofN over K. Ifthere are elements :1, . . r, 3,, ofM, not all

zero such that 2; ti ®fi is invariant in M ® N, then the subspace M1 of M

generated by the e‘ is stable under the x”; if further N is simple the representa-
tion of g on M1 is dual to the representation of g on N.
(b) Let Ml, M,, N, N"I be g-modules of finite dimension over K. Suppose
that M1 and M21 are simple and that the representations of g on N and N" are
dual. For the g-module M1 to be isomorphic to a sub-g-module of M2 ® N, it
is necessary and suflicient that Ma be isomorphic to a sub-g-module of
M1 8) N“. (Use Proposition 4; consider the representation of g on
M: 8) M2 ® N and apply (a) no the dual representation of the latter.)
1] 5. Suppose that K is a field of characteristic 0. Let V be a finite-dimen-
sional vector K-space, g = el(V), U the enveloping algebra of 9, U" the set of
elements of filtration <11 in U, U’I the set of images in U of homogeneous
symmetric tensors of order n over 9 and (1,, . . t, mm) a basis of the vector space
5
9. Let W, = ® V. The identity representation of 3 defines a representation
of g on W, which extends to a homomorphism 1:, of U into the algebra
M, = $(W,).
(a) Let M; be the subspace of M, generated by the elements of the form
[3, @321 ®- - -® [3” where is, E$(V) and [34 = l for at least one i. Let 2 be
an element of U3 and 20 = 1‘“
..... ' . gm “amt.“u- . .og. its component in U‘,
where the an”... are symmetric with respect to permutation of indices. Show
that
1r,(z) E 4‘! 101,224.“: a:,...r.°‘u ®v - ~® (1" (mod. MD
(b) Deduce from (u) that for all zeU there exists a finite-dimensional
representation 1: of U such that 1r(z) 9e 0. (For a sequel to this exercise,
cf. § 7, Exercise 3.)

86
 mom

6. Suppose that K is a field. Let g be a Lie K-algebra, rent“ a finite-

dimensional representation of g, (:1, . . ., on) a basis of M and (of, . . . , 2;)
the dual basis.
.
(a) Iffis a linear form on ® M, then:
f= KHZJNJUH ®---® «Me: ®~~~®e:).
(11) Let [5 be a non-degenerate bilinear form on M which is invariant under
5. Let (:1, . . ., 3:.) be the basis of M such that 5003;) = 8”. Letf be an
,
invariant linear form on ® M. Deduce from (a) that

1:1,.....£,‘n f9“ 8. O ' ® a") (5"! ®' . ‘® '1')

is an invariant element of ® M independent of the choice of basis (2,).

(0) Let U be the enveloping algebra of g and U" the dual space of U, The
adjoint representation of 9 can be extended to a representation at »—> xv of g on
U defined by xuu = xu — w: for all 14 EU. HenceU‘ has a g-module structure.
For an element f of U* to be invariant, it is necessary and sufficient that
f(1411) = f(on) for all u, 1/ in U.
(11) Let b be a finite-dimensional ideal of g and Y an invariant bilinear
form on 5 whose restriction to l) is non-degenerate. Let (yolfl‘n, (y;)1‘,‘,,
be two bases of D such that y(y,, y;) = 8”. Let r be an integer >1. Letf be a
linear form on U such thatf(w) = f(w) for all u, u in U. Deduce from (b) and
(c) that .
“unzwfwtyt. . 4mm,- . .14,
is an element of the centre of U independent of the choice of basis (m). In
particular recover the Casimir elements.
(.9) Let 0 be a permutation of {1, . . ., r}. Arguing similarly, show that

l<‘l',znh‘nf(yli(l)y‘l(2)' ' ’y‘Im)!/;i!/ila' ' 'y‘lr

is an element of the centre of U independent of the choice of basis (gt).
7. Let g be a complex Lie algebra, [3 its Killing form and an the Lie algebra
derived from g by restricting the base field to R. Show that the Killing form
of go is twice the real part of fl.
8. Let 5 be a real Lie algebra. The multilinear form
(x), . . ., x”) >—>Tr(ad oadx, 0- - wadxn)
on g” is invariant.

87
I LIE ALGEBRAS

9. Let g be a Lie K-algebra, p and p’ semi-simple representations of g on K-

modules V, V’ and 4) a homomorphism of the g-module V onto the g-mcdule
V’. Show that the image under (I) of the set of invafiants of V is the set of
invariants of V’ (use Proposition 6).
10. Suppose that K is a field. Let g be a Lie K—algebra and M1, M2 two
non-isomorphic simple g-modules of finite dimension over K. If K1 is a separ-
able extension of K, show that there exists no simple gum-module isomorphic
both to a sub-gmn-module of Mmm and to a sub-gmrmodule of M2(K))' (Use
Proposition 4- and note that the existence of an invariant element 9&0 '
in Mfum @111 Mm“, implies the existence of an invariant element #0 in
M1“ ®K Ma (Algebra, Chapter II, §5, no. 3, Theorem 1.)
11. Let It! be the Lie p-algebra defined in Exercise 2l of § 1. Show that the
Killing form of m is zero.
1} 12. Suppose that K is a field. Let g be a Lie K—algebra and M a g-module.
Let (We, M) denote the space of alternating linear mappings of 9’ into M.
We write C°(g, M) = M and, forp < 0, C”(g, M) = {0). Let C"(g, M) be the
direct sum of the C’(g, M). The elements of C“(g, M) are called the cachairu
q with values in M; those of C’(g, M) are said to be of degree ,0. For all y E 9,
we denote by i(y) the endomorphism of 0*(9, M) which maps each subspace
C’(g, M) into C"1(g, M) and which, for p > 0, is given by the formula:

(1) (i(y)f)(*n~~uxp-1) =f(y,In--~Mp—1)~

We have i(y)2 = 0.
(a) The adjoint representation of g and the representation of g on M define
a representation of g on the space of multilinear mappings of 9" into M. Show
that C’(s, M) is stable under this representation. Let 0 be the representation
of g on 0*(9, M) thus defined. Show that:
9(x)i(y) —- i(!)9(x) = “FWD
for all x, y in g.
(b) Show that there exists one and only one endomorphism d of 0"(g, M),
mapping C”(9, M) into C’“(g, M), such that

(5) My) + Ky)" = 90)

for all y E g. (Argue by induction on the degree of the cochains.) Show that, for
f E C”(9, M :
#(xn’tzw‘w‘pn) = “2:, (—1)l+{f([xb lxh"-’;‘lr-"1£1t~--txp+l)

+ Z (—1>”1(x.)..f<x., . . ., . . ., x...)
88
 WISHES

(where the symbol “ over a letter means that it is omitted).

(0) Show that, for all y e g,
(4) 11W) = WM
(Show first, usmg (2) and (3), that d0(y) — 6(y)d commutes with every i(x).
Then argue by induction on the degree of the cochains.)
(d) Show that d2 = 0. (Show first, using (3) and (4), that d2 commutes with
every 1' (x). The argue by induction on the degree of the cochajns.)
(z) The restriction of d to C"(g, M) has a kernel Z’(g, M) whose elements
are called wag/ole; ofdegree [I with values in M. The restriction OH to C” ' 1(g, M)
has image B’(g, M), whose elements are called tabmmdaric: of degree [a with
values in M. Then 139(9), M) C Z”(g, M). The quotient space
2”(95, M)/B’(9, M) = H’(a, M)
is called the mlwmology space of g of degree [a with values in M. The direct sum
of the H’(g, M) is denoted by H"(g, M). Show that H°(9, M) is identified
with the set of invariants of M. Let at be a homomorphism of the g-module M
into the Gmodule N. For allfe C’(g, M), <1; ofe CP(g, N), so that 4) can be
extended to a K-linear mapping (12’: C“ (5, M) —> C“ (9, N). Show that
(V o d = d a (V. Deduce that 4% defines a homomorphism

$1H"(9a M) —>H*(9,N)
which is said to be associated with (I).
(f) Let L be a sub-g-module of M and N the quotient g-module M/L. The
canonical homomorphisms L —‘> M —p> N define homomorphisms

0*(3, L) i» m9, M) 43 cm, N),

H“(g, L) L) H“(g, M) J1» H"(g, N).
Show that i ' is injective and has image the kernel of[1’ and that [1’ is surjective.
For all 5 e H"(g, N), let 2 e Z”(g, N) be a representative of: and a E C”(g, M)
be such thatp’(a) = 2; show that da 6 Z’I+ 1(g, L) and that its class in H’” 1(3, L)
depends only on :3 denote this class by SM Show that the sequence of homo-
morphisms:
o —) H°(g, L) ‘—°) H°(g, M) 1) H°(g, N) L")
F.) H‘(g, L) _"—> H‘(g, M) —"'—> H‘(9, N) L . ..
is an exact sequence.
(g) With the notation of (f), the exact sequence:
0 —> Y(N, L) —> .‘2’(N, M) —> $(N, N) —> o
89
 I LIE ALGEBRAS

defines an exact sequence

H°(9, 50(N, M)) ~> H°(9, Y(N: N)) —> H101. $(N, U)-
The identity mapping of N is an invariant u of .5?(N, N) and hence an element
of H°(g, 2’(N, N)); let a be its image in H‘(g, $(N, L)). Then for the exis-
tence in M of a sub-g-module supplementary to L it is necessary and sufficient
that a = 0. (This condition means that u is the image of an element of
H°(g, $(N, M)), that is a homomorphism u of the g-module N into the g-
module M; then u(N) is the desired supplement.) ,
(/1) Show that 21(9, 3) (where g is considered as a g-module by means of the
adjoint representation) is identified with the vector space of derivations of g
and that B1(g, g) is identified with the vector space of inner derivations of g.
(i) Let a and b be two Lie K-algebras with b commutative and b 1» g —u> a
an attension ofa by b. For all x e g, the restriction ot‘ado x to 1: depends only on
the class ofx modulo b, whence there is an a-module structure on b. Let v be a
K-linear mapping of a into 9 such that y. o v is the identity mapping of a. For
my in a, we writef(x,y) = [we vy] — v([x,y]). Show thatfs Z“(n, b) and that
the class 0 off in H’(a, b) does not depend on the choice ofv. For two extensions
of a by b, which define the same a-module structure on b, to be equivalent, it is
necessary and sufficient that the corresponding class 5 6 H101, b) be the same.
For the extension to be inessential, it is necessary and sufficient that c = 0, If
B is an a-module and a e H3(a, B), there exists an extension of a by B (consi-
dered as a commutative Lie algebra) defining the given a-module structure on B
and the given element 0 of H’(a, B).
(j) Let g be a finite-dimensional Lie algebra over K, U its enveloping
algebra, I) an ideal of g, B an invariant bilinear form on g whose restriction to I)
is non-degenerate, (y.),“‘,, and (yfllfl‘n two bases of 9 such that
n
W!!!) 1/» = 8.], I = ”21 my; 6U, M a g-module and p the endomorphism of
0*(9, M) which maps each C"(g, M) into C"‘(g, M) and which, for [a > 0,
is defined by:

(much. ..,~,,-1) = Z (mum, . . ., as.)

Finally let 1" be the endomorphism ofC‘ (g, M) extending t“ which maps each
cochainfof degree >0 onto the cocbain [M of. Show that pd + d9 = I‘. (If
for x E 5; we write:
n n
[m] = Z may” [l = Z am
-show first that a“ = —afk.) Deduce that I‘d = (11‘. If M is simple and finite-
dimensional over K, B is the associated bilinear form and cla 1) is not divisible

90
 EXERCISES

by the characteristic of K, show that H”(g, M) = {0} for all 1;. (Using Proposi-
tion 12, show that I‘ is an automorphism of C‘ (9, M) and hence induces auto-
morphisms of Z’(g, M) and B’(g, M). For

f6 21m, M), I‘f= (-19 + pd)f= dE 3% M),

whence Z’(g, M) = B’(9x M)-)

§4
The conventions of § 4 remain valid unless otherwise mentioned.
1. Let g be a nilpotent Lie algebra and p (resp. q) the smallest integer such
that V’g = (0} (resp. Veg = 9). Show that}: = q + l and that fig 3 VV‘Q.
(Use the argument of Proposition 1.)
2. Let g be a semi-direct product of an algebra b of dimension 1 and a com-
mutative ideal 9’. Let x e I), x aé O, and u be the restriction ofad, x to 5’.
(a) For g to be nilpotent, it is necessary and sufficient that u be nilpotent.
(b) For the Killing form of g to be zero, it is necessary and sufficient that
Tr(u2 = O.
(c) )Deduce from (a) and (b) that there exist non-nilpotent Lie algebras
whose Killing forms are zero.
(:1) Deduce from (a) that in a nilpotent Lie algebra such that 9”"9 $5 (0},
(Pg = {0) it is possible that «19 96 V’"g.
3. (a) Let g be a nilpotent Lie algebra, 3 its centre and b a non—zero ideal
of 9. Show that 3 n b aé {0}. (Consider b as a 5-module by means of the adjoint
representation.)
(b) If in a Lie algebra g an ideal b is contained in (Sing but not in fig,
show that the ideals b n 9kg are all distinct for 0 s k g i + 1. (Apply
(0-)
4. Let g be a nilpotent Lie algebra.
(a) Every subalgebra of g is subinvariant. (Use Proposition 3.)
(17) Let b be a vector subspace of 9 such that b + 99 = 9. Show that the
subalgebra of 9 generated by b is 9. (Apply (a) to this subalgebra. Reduce it
thus to the case where b is an ideal of g and use Proposition 4 of § 1.) Deduce
that the minimum number of generators of g is dim 9/99.

5. (a) Show that a Lie algebra g in which every subalgebra is subinvariant

is nilpotent. (Show that if dim 9 > 1 every x E 9 belongs to an ideal 1) ye g of g
which has the same property as g and hence is nilpotent by induction on the
dimension of 9; hence adfi x is nilpotent.)
(b) Show that if in a Lie algebra 9 every subalgebra distinct from g is distinct
from its normalizer, 9 is nilpotent. (Reduce it to (41).)

91
[ LIE ALGEBRAS

6. Let 9 be a nilpotent Lie algebra and a a commutative ideal of g. The

following conditions are equivalent: (a) a is a maximal commutative ideal of g ;
(b) a is a maximal commutative subalgebra of 9: (r) a is equal to its centralizer
n’ in 9. (The implications not (a) => not ([7) => not (5) are obvious. If a’ ;é a,
there exists by Proposition 1 an idal a' of 9 such that a C a” C a’ and
dim a”/a = 1. Then a' = a + Kx, hence [a', 9"] C [a, a] + [x, a] = {0} and
hence (a) is false.)
7. (a) Let V be a finite-dimensional vector space over K, 9 a Lie algebra v
of nilpotent endomorphisms of V and (V,)o(,‘,, a decreasing sequence of
vector subspaces ofV, with V0 = V, V,| = (0} and 9(V,) C VM for 0 < r < 71.
Show by induction on i that (9‘9) (V,) C VHzt. If dim V 2 2‘, the subspace
of elements of V annihilated by 9‘9 is of dimension 22‘. If dimV < 2‘,
9‘9 = {0}.
(12) Let 9 be a nilpotent Lie algebra and i an integer 2 0. Ifdim 9‘9 > 2‘ + l
the centre of 9‘9 is of dimension 22‘. If dim 9‘9 < 2‘ + 1, 9‘9 is commuta-
tive. (Apply (a) no the restrictions to 9‘9 of the ad, x, x e 9.)
(0) Let 9 be a nilpotent Lie algebra and i an integer 20. If 9‘9 is not
commutative, 9‘9/9‘“9 is of dimension 22‘ + l. (Reduce it, by passing to
the quotient, to the case where dim 9‘“g = 1 and use ((7).)
8. Let g be a Lie algebra, m an ideal ofcodimension l, x an element of 9 not
belonging to m and z e 9.
(a) Show that the linear mapping D which reduce to 0 on m and maps x to z
is a derivation if 1 belongs to the centralizer a of m in 9.
(b) Let q be the largest integer such that a C (€49. Show that if further
2 ¢ ‘6’“‘9, D is not an inner derivation of 9‘
(c) Deduce from (a) and (b) and Exercise 7 (c) that if 9 is a nilpotent Lie
algebra of dimension >1, the vector space of inner derivations of g is of co-
dimension 22 in the space of all derivations of 9.
9. (a) Verify that the following multiplication tables define two nilpotent
Lie algebras 93, 9, of dimensions 3 and 4:
93: [11:59] = 7‘3: ["1; *3] = ["s; 1‘3] = 0
Bl: [*n’ral = “a: ["ia’fal = *4: ["n 14] = [*mxa] = [”2“] = [3334] = 0»
(b) Show that 93 is isomorphic to “(3, K) and also to sl(2, K) if K is of
characteristic 2.
(a) Let 91 be the unique Lie algebra of dimension 1. Show that the nil-
potent Lie algebras of dimension <4- are given by the following table:
dimension 1: 91; dimension 2: (91)“;
dimension 3: (9,)“, 93; dimension 4: (90‘, 9;, x 91, 94.
(Use Exercise 7 (c). If 9 is nilpotent and 3—dimensional and dim 99 = l,
observe that 99 is contained in the centre of 9 (Exercise 3 (11)), whence

92
 EXERCISES

g = 53. If dim 9 = 4, dim 95 = I, observe that the bracket on 3 defines an

alternating bilinear form on 5/99, which is necessarily degenerate and hence
that there exists a subalgebra b of9 contained in the centre ofg with dim b = l,
I) n 99 = (0}, whence g = b x 93. Ifdim g = 4, dim 9g = 2, 9g is commu-
tative; applying Theorem 1 to the restrictions to 99 of the adI x (x e 9), show
that there exists a commutative ideal I) ofg with b D 99, dim b = 3; letx e g,
z ¢ 9; choose a basis of 9 such that the restriction of adl x to b has a Jordan
matrix with respect to this basis; then 9 = 54.)
10. Let 9 be the Lie algebra of derivations of the algebra 9, (Exercise 9).
Show that 9 is of dimension 7 and of zero centre and that the ideal 9' C 9 of
inner derivations has no supplementary suhalgebra in 9.
{I l 1. Let L be a commutative ring, A a left Artinian algebra over L and y a
mapping of A x A into L. Define on A the internal law of composition
(a, b) >—>a at: b = ab + 7(a, b)ba. For every subset E of A, let E denote the
subring (with no um'l element) of A generated by E. Show that if E consists of
nilpotent elements and is stable under the law :, then E is nilpotent (that is,
there exists 7; > 0 such that E” = {0}). Proceed as follows:
1. Suppose first that A is a simple ring and hence isomorphic to JD(T),
where T is a left vector space of finite dimension m over a field D. Argue by
induction on m. Let (D be the set of subsets F C E which are stable under a: and
such that f“ is nilpotent. Show that d) admits a maximal element M (note that
F“ = (0} for all F E (1)). Suppose M yé E. Show that there exists a e E such
that a e M and t s a e M for all Is M (observe that there can be no infinite
sequence (an) such that an e E, an ¢ M, an = In.1 as a,,_1 with t,,_1 E M). Let
S be the subspace of T the direct sum of the u(T) with u E M; show that
S aé {0} and S aé T and that 0(5) C 5. Let N be the set of DEE such that
11(5) C S. By using the induction hypothesis and considering the elements of N
as operating on S and on T/S, show that N e 1D, which implies a contradic-
tion.
2. In the general case, use the fact that the Jacobson radical of A is nil-
potent. Deduce from this result a new proof of Engel’s Theorem.
12. Let g be a Lie algebra and ?”g the intersection of the W9.
(a) The Lie algebra g/Vmg is nilpotent.
(b) Show that there exists a nilpotent subalgebra I) of 9 such that
g = I) + (fag. (Argue by induction on dim 3. Suppose that g is not nilpotent.
Let x e g be such that I = 0 (ad x)“(g) aé {0}. Let n be the union of the ker-
nels of the (ad 2:)". Show that this is a subalgebra of g and that g is the direct
sum of I and II. By the induction hypothesis tt is the sum of gwtt and a nil-
potent subalgebra b. Finally, V0" C ?“9 and I C 9‘9.)

93
I LIE ALGEBRAS

(0) Verify that the following multiplication table defines a (solvable)

Lie algebra g of dimension 5:
[11, x3] = ”5: [In ‘3] = ”a, ["n *4] = —*4. [’53, x4] = x5
and the other [:q, :9] are zero. Show that for this Lie algebra
gm9=K"3'l‘1<1‘4"l‘I<-Vs
but that there exists no supplementary subalgebra ofVEg in g.
1] 13. Let g be a Lie algebra and A the centralizer ofifmg in g. If; e): ?”g,
g has non-zero centre. (Write g = gag + I, where I is a nilpotent subalgebra
of 9 (Exercise 12). Let 91 = 3 + 2. Let I) be a nilpotent subalgebra such that
91 = «‘91 +1). Show thatg = @09 + band magma, C angmg. Lety be
an element of; not belonging to @“55. Write y = x + x' where x e I), x' E W‘s“;
then x e a and x aé 0 and hence 3 n b is a non-zero ideal of I); using Exercise
3 (a), deduce that there exists a non-zero element of g permutable with
¥”g and b.)
11 14. Let g be a Lie algebra and b a subinvariant subalgebra of 9 such that
the centralizer 3(1)) of b in g is zero.
(a) The centralizer a(¥"‘b) of «”1, in g is contained in W”). (If
366%) :1: @mb, a(‘6’”b) d: b by Exercise 13;?”6 is an ideal ofg (§ 1, Exercise
14) ; let c = b + 360%); c is a subalgebra, b is subinvariant in c; b is an ideal
of hi C c, where b 9E bl; let yEbl, y¢b; y = x + x’, where x53(?"’b),
x’ Sb; then xe “9%) n It), x¢ b; b = Kx + b is a subalgebra; f”) = m
because a n 303%) C €’°°b, whence W‘bC W‘b for all k; 3(¥“’b) n b c): w
and hence the centre of b is non-zero by Exercise 13; a fortiori, 3(b) aé {0},
which is absurd.)
(b) Dcduce from (a) that, if 9 denotes the Lie algebra of derivations of
war], and a the centre ofgwb, then dim g g dim 59 + dim a (let 5 operate on
fab by the adjoint representation).
15. Let I be a Lie algebra with zero centre, 91 the Lie algebra of derivations
of l, 192 the Lie algebra of derivations of 91, . . . . Then I is an ideal of $1, 91
is an ideal of ‘92, etc. (§ 1, Exercise 15 (5)). Let 9 be the Lie algebra of deriva-
tions of ‘6’”1 and a the centre of 9‘1.
(:1) Then dim 9, S dim 9 + dim 0. (Use Exercise 14 above and Exercise
15 of § 1.)
(b) Deducc from (a) that for i sufficiently large all the derivations of 9.
are mncr.
16. Let 91 and g, be Lie K-algebras and I11 and "2 their largest nilpotent
ideals. Show that the largest nilpotent ideal of 91 x g, is nl x 11,.
17. We reconsider the Lie algebra as of Exercise 9 (a) with its basis
(“1, ”a: ’53)-

94-
 EXERCISES

(a) Let V be the vector space K[X]. Let D be the differential operator
with respect to X on V and M the operator of multiplimtion by X on V.
Show that if K is of characteristic 0 the mapping

M1+s+ixaHaD+PM+Y
is an infinite-dimensional simple representation 9 of g on V.
(b) If K is of characteristic [1 > 0, the ideal (X?) of K[X] is stable under
p(g). Under the quotient representation of g on K[X] /(X1’), no line is stable.
18. Let g be a 7—dimensional vector space over K with a basis (2,)“.fl.
An alternating bracket is defined on g by the fonnulae:

(1) [51:91] =¢u€s+1 (1<i<j<7ai+j<7)

and all the other brackem [ch 2,} are zero for i < j.
(a) For the Jacobi identity to hold, it is necessary and sufficient that:

(2) -a28a15 + “mass = 0

(3) “120‘s; — 0(24083 + “14‘25 = 0-
(b) Suppose henceforth that all the 01., are #0. Show that the ideals Wg,
V33, '99, 3’59, ‘K‘g admit the following bass: (:3, e4, :5, 33, e7), (e4, :5, es, 2-,),
(e5, :5, e7), (:5, :7), (:7). Show that the centralizer I) of @55 admits the basis
(’2: ea: 94: :5, (a, ‘7)-
(6) Let (2})1‘“7 be another basis of 5 such that W35 admits the basis (:4),
3’59 admits the basis (:3, 91,), . . ., I) admits the basis (1;, 25, 2;, 23, 2;, :4). Show
that
[4,4] = 0414+! + 34+!“ + Winn + - ~- + ”i:
where
“5.4436061; jail l = “is“zsuiaias-A“

(d) Deduce that there wrists, if K is infinite, an infinity of non-isomorphic

nilpotent Lie algebras of dimension 7 over K and that there exist complot
nilpotent Lie algebras which cannot be derived from real nilpotent Lie algebras
by extending the base field from R to C.
19. (a) Let g be a Lie algebra. Let 5) be the Lie algebra of derivations of
9. Characteristic ideals 9"” of g are defined inductively as follows: gt") = 9
and 9‘3“] is the subspace of 9 generated by the Dx (D 69,16 9'“). The
following conditions are equivalent: (1) every derivation of g is nilpotent:
(2) 5"“ = {0} for k sufiiciently large; (3) the holomorph of g is nilpotent.
If they are fulfilled, g is said to be characteristically nilpatml. Such an algebra
is nilpotent.

95
I LIE ALGEBRAS

(b) Verify that an 8-dimensional Lie algebra is defined by die following

multiplimtion table:
[‘1’ ‘2] = ”a: ["n "al = *4) [‘1’ *4] = 5‘6) [X1: x5] = x6
["n ‘8] = "a: [‘11 x7] = ”s, [‘2’ 7‘s] = "s, ["2: "Al = 1‘6
[*2335] = 5‘7; [‘21 ”3] = 2‘s; ["aixsl = —"7 + “a: Day‘s] = —x5
and [x., x,] = 0 fori + j > 8. Show that
5 3 5 8

«fig = 2; Km, 3’39 = 2 K1” ““9 = 2 Kim as = Z Kt“

‘55:; = Kara, We = {0}, We, We] = M + Kxa,
8

and that the transporter ofWg into @‘g is :22 p Deduce that every deriva-
8

tion D of g is defined by formulae Dx‘ = z, uvxp Then show that u“ = 0

for all i if the characteristic of K differs from 2, so that g is characteristically
nilpotent.
(a) If the characteristic of K is #2, show that the Lie algebra g of (b) is
not the derived Lie algebra of any Lie algebra. (If g = 91), show first that
b is nilpotent. Observe that dim 9/99 = 2, which with Exercise 7 (5) leads
to a contradiction.)
fl 20. Let g be a Lie algebra and U its enveloping algebra.
(a) Let 9’ be an ideal of 9, U’ C U its enveloping algebra, xeg and
4,, . . ., a,| in U. Suppose that m = x for the indicesjl, . . -;j,, and a‘ e U’ for
the other indices (let k1, . . ., Icq be these indicm, with k) < k2 < > . - < k4).
Then 111a2 . . . afl — flaking.2 . . . ak‘ is a sum of terms of the form x”), where
b e U’ and p' < p. (Argue by induction on [1.)
(1;) Suppose henceforth that g is nilpotent and K is of characteristic 0.
Let g’ be an ideal of codimension l in 9, U' its enveloping algebra, x an
element of 9 not belonging to g' and U0 (resp. U6) the subalgebra of U (resp.
U’) annihilated by a set it of derivations of g (which extend to derivations
of U) mapping 9 into 9’. Suppose that U0 is contained in the centre of U and
U0 4: U’. Show that there exist «1 6 U3, a2 6 U’ such that :21 aé 0 and
a = xal + aze. (Let ambm + arm—lbw, + + bao with 11m,“ ., bo
in U', m > 0, [1m aé 0. Show, using (a), that, for every derivation D E b of
g, Db"| = 0, m(Dx)b,, + Dbmd = 0 and hence D(mxbm + bmd) = 0.) Show
that U0 is contained in the algebra K[u, u,“ ’, U3] generated by a, a; l, U",
in the field of fractions ofUo (which can be formed by Corollary 7 to Theorem
1 of § 2). Deduce that the field of fractions of U0 is the field generated by a
and U3. Show that a is transcendental over K(U{,).

96
 EXERCISE?

(5) Let (0) 2 3., C 51 C C g" = g be a sequence of ideals of g of

dimensions 0, l, . . .,n. Let x, be an element of a, not belonging to g,_,.
Let j1 < j, < - - - < j, be the indices j such that there exists in U1 (envelop-
ing algebra of 9,) an element of the centre Uo of U not belonging to U"1.
By (1;) there exists ajlenU’ and a,36U"1 such that a“ aé 0 and
a, = x,a,1 + a” 6 U0 n U’. Show by induction on n that
-i —1
U0 C K[a,l,. . ., an, an“. . ., am]

and that the field of fiacfions of U“ is generated by the algebraically inde-

pendent elements a,l,...,a,a. In particular, the field of fractions of the
centre of U is a pure transcendental extension of K.
1i 21. Let g be a. Lie algebra and a an automorphism of 9.
(4) Suppose first that K is algebraically closed; for all A e K let LA be the
set of x e g which are annihilated by a power of o — AI; 9 is the direct sum
of the LA. Show that [LN L,,] C L”. (Note that
(a — MIXEMD = [(a — mx, cry] + [M (a — PIN/l)
(b) With an arbitrary field K, suppose that none of the eigenvalues of a'
(in an algebraically closed extension of K) is a root of unity. Show that g
is nilpotent. (Reduce it to the case where K is algebraically closed. If )q, . . . , )‘m
are the distinct eigenvalues of 0-, the NM, 103, All}, . . . are not all eigenvalues
and hence ad“): is nilpotent for x 6 LM Conclude using Exercise 11 applied
to the set of adwz, where it runs through the union of the L1,.)
(c) With an arbitrary field K, suppose that a” = I, where q is a prime
number and that no eigenvalue of a is equal to 1. Show that g is nilpotent.
(Same method as in (b), observing that, if 9 ye {0}, q is not equal to the
characteristic of K and that for every ordered pair of eigenvalues N, )t, of a
there exists an integer k such that MA} = l.)
22. (a) Let u, u be two endomorphisrns of a vector space E of finite dimen-
sion over an algebraically closed field K. For every eigenvalue )t of 14 let
EA be the subspace of E consisting of the vectors annihilated by a power of
u — AI. Show that, if (ad u)"v = 0, the subspaces EA are stable under 1;.
(b) Deduce from (a) that, if g is a nilpotent Lie algebra of endomorphism
of E, E is the direct sum of subspaces F, (l < j S m) which are stable under
g and such that on each F, the restriction of every element u e 9 can be writ-
ten },(ufl + 11,, where A,(u) E K and u, is nilpotent.
(r) If K is of characteristic 2, E = K“ and g is nilpotent algebra 51(2, K),
show that m = l and that powibly Mu + v) 96 Mu) + 1(1)) for two elements
u, v of 9. (For a sequel to this exercise, cf. §5, Exercise 12.)
23. Let 9 be a Lie p—algebra over a perfect field K of characteristic [7 > 0.
g is called p—um'jwlm! if for all x e 9 there exists In such that x" = 0.

97
l LIE ALGEBRAS

(a) Show that every p—unipotent Lie p-algebra is nilpotent.

(b) Suppose that g is nilpotent and let b denote the p—core of the centre
of 9 (§ 1, Exercise 23 (11)). Show that 9/!) is p-unipotent.
(5) Let g be the nilpotent Lie [J-algebra with a basis of three elements
91, 62, ts such that [en ex] = [61,63] = 0, [42,23] = :1, ti = 9,, £5 = 95 = 0
Then b = KAI, but 9 is not the direct sum of b and a p-unipotentp-subalgebra.
(d) If g is p—unipotent, show that in the restricted enveloping algebra of
g the two-sided ideal generated by g is nilpotent. (Argue by induction on the
dimension of g.)

24. Suppose that the field K is of characteristic 2. Show that in the nil-
potent Lie algebra g, of Exercise 9 there mtists no 2-mapping.
25. Let g be a Lie p—algebra. Show that the largest nilpotcnt ideal of g
is a p—ideal (cf. § 1, Exercise 22).
26. Let G be a group and let (H,,),,>1 be a decreasing sequence of sub-
groups of G; suppose that H1 = G and that ifwe write (x,y) = xyx"y",
the relations at E H‘, g e H, imply (x, y) e H”).
(a) Let G, = Ill/HM“ show that G‘ is commutative and that the mapping
1, y1—> (x, 1/) defines on passing to the quotient a Z-bilinear mapping of
G‘ X G, into G,”

(b) Let gr(G) = “21 G; and extend by linearity the mappings

G. x G,—>GH, defined in (a) to a Z—bilinear mapping of gr(G) >< gr(G)
into gr(G). Show that gr(G) is a Lie Z-algcbra with this mapping (to verify
the Jacobi identity, use the following formula:

((1%): 7-")-((1I» 1), x‘)-((za fly") = t ay, 2 in G.

where x” denotes yxy“ and 9 denotes the identity element of G).
(c) Suppose that there exists 1: such that H” = (2}. Show that gr(G) 1s a
nilpotent Lie Z-algebra
27. Let A be an associative algebra with unit element 1 and let
A0 = A 2 A13- - - D A" D - ~ ~ be a decreasing sequence of two-sided
ideals of A such that A“ A: C A1.” Let G be a group with identity element
e and letf: G—>A be a mapping such thatf(¢) = l, f(xy) =f(x) .f(y)
and 1 —f(x) eA1 for all xeG Let H denote the set of xeG such that
l — f(x) e A". Show that the H,I satisfy the conditions of Exercise 26. Show
that the mapping x»—>f(x) — 1 defines on taking quotients an injective
homomorphism of the Lie algebra gr(G) into the Lie algebra associated
with the graded ring gr(A)= 2 A,./A,,+l (cf. Cummutative Algebra, Chapter
III).

98
 EXERCISES

§5
The conventions of § 5 remain valid unless otherwise mentioned.
1. Let g be the 2-dimensional non—commutative solvable Lie algebra. Show
that the Killing form of g is non-zero, that every invariant bilinear form on g
is degenerate and that every derivation of g is inner.
2. (a) Show that in the 3-dimensional solvable Lie algebra over R defined
by the multiplication table [x,y] = z, [x, z] = —y, [3], z] = 0, there exists
no decreasing sequence of ideals of dimensions 3, 2, l, 0.
(b) Show that in the non-commutative 2-dimensional solvable Lie algebra
9 there exists a sequence of ideals of dimensions 2, l, 0 but that g is not nil-
potent.
3. Let g be a solvable Lie algebra such that the conditions x e g, y e g,
[[x,y], y] = 0 imply [94,31] = 0. Show that g is commutative. (Let k be the
largest integer such that 9""5 yé {0}, 9kg = {0}. Assuming k 2 2, show
first that [9"‘25, 9"“9] = {O} and then that [9"‘29, 9“”5] = {0}, whence
a contradiction.)
4-. Show that the centre of st(n, K) is zero and that that of n(n, K) is of
dimension 1.
5. Let g be a Lie algebra and (9°55, Q‘g, . . ., 9‘9) the sequence of derived
algebras of g (n 2 0, 9"“9 95 9’5). Then dim 9‘5/9‘“g 2 2"1 + l for
1 g i < n — 2. (Taking quotients by 9‘9, reduce it to the case where g
is solvable. Then use the fact that 99 is nilpotent and Exercise 7 (c) of § 4-.)
6. (a) Verify that the following multiplication table defines a 5-dimen-
sional solvable Lie algebra g:
["1: x2] = x5, ["n "al = 7‘s: [32: ”4] = *4
["n’fil = [xmxa] = [huh] = [xv 9] = 0~
(11) Show that the orthogonal of g with respect to the Killing form is
99 = K33 + Kx, + s. Deduce that 99 is the largest nilpotent ideal of g.
(2‘) Show that there exists no supplementary subalgebra of 99 in g. Deduce
that g is not the semi-direct product of a commutative algebra and a nil-
potent ideal. (Show that this nilpotent ideal would necessarily be 99.)
7. Let 9 be the 3-dimensional solvable Lie algebra with basis (at, y, z) such '
that [x,y] = z, [x, z] = y, [y, z] = 0. Show that the linear mapping which
maps x to —x, y to —z, z to y is an automorphism of g of order 4. Compare
this result with Exercise 2i (r) of § 4.
8. (a) Let go be a 3-dimensional solvable real Lie algebra such that Ego is
commutative and 2-dimensional. Let g be the algebra derived from 90 by

99
l LIE amen“

extending the base field from R to C. For x6 9, let u,‘ be the restriction of
adfix to 95;. Show that the eigenvalues of 11,, either have the same absolute
value or are linearly dependent over R. (We have x = M + z with z E 9g,
y E 90, A E C and hence ux = My. Now an is the C—Iinear extension to 95 of
an R-linear endomorphism of 990.)
(b) Show that there exism a 3—dimensional solvable complat Lie algebra
g, with 99 commutative and 2-dimensional, and an element x of g such that
the restriction of ad"): to 99 has eigenvalues which neither have the same
absolute value nor are linearly dependent over R. (Construct g as a semi-
direct product of a I-dimensional algebra and a 2~dimensional commutative
algebra.)
(a) Show that the algebra constructed in (1)) cannot be derived from a
real Lie algebra by extending the base field from R to C.

9. Let g be a Lie algebra, r is radical, n its largest nilpotent ideal and D

a derivation of 9. Show that 99 n r c n. (Let i? be the Lie algebra of deriva-
tions ofg and r' its radical. Ifx e 9 is such that Dxet, then

ad(Dx) = [D, ad x] 59b n t’

(Corollary 2 to Proposition 5), hence ad(Dx) is nilpotent (Theorem 1) and

hence D: 5 n.)
10. Let g be a Lie algebra, 1: its radical and a a subinvariant subalgebra
of 9. Show that the radical of a is a n r. (Apply Corollary 3 to Proposition 5
several times.)

11. Let 3 be a Lie algebra, p a. finite-dimensional representation of g and

E the associative endomorphism algebra generated by 1 and 9(g). Show that
the largest nilpotency ideal n of p is equal to the set n' 01‘n such that
Tr(p(x)u) = 0 for all u E E. (To show that n’ C n, show that n' is an ideal
and that for all x e n’ the semi-simple component of 9(a) is zero, by noting
that Tr((p(x))") = 0 for every integer n > 0.)

12. Suppose that K is algebraically closed. Let g be a nilpotent Lie K-

algebra. Let p be a finite-dimensional representation of g on a vector space V.
For every linear form A on 9, let V" be the vector subspace of V consisting
of the EEV such that, for all xeg, (9(a) —- A(x)I)"E = O for sufliciently
large n.
(a) The V" are stable under 9(9) and the sum of the V" is direct. (Use
Exercise 22 of § 4-.)
(b) We have V = 2V“. (If each p(x) has a single eigenvalue, V = V‘a
by Corollary 2 to Theorem 1. If p(x0) has at least two distinct eigenvalues, V

100
 EXERCISES

is the direct sum of two non-trivial subspaces which are stable under 9(9).
Then argue by induction on the dimension of V.)
13. Let g be a Lie Algebra and 9 the Lie algebra of derivations of 3. For g
to be characteristically nilpotent, it is necessary and sufficient that 9 be
nilpotent and that dim g > 1. (To see that the condition is sufficient, write
g as a sum of subspaces 9", applying Exercise 12 to the identity mapping of
9. Show that [93", 9“] C 3““ and that each 3“ is an ideal of g. Deduce that
9" is commutative for A aé 0. Using again the fact that Q is nilpotent, show
that g = 9” if dimg > 1. Otherwise g = 9° x b, where b is commutative.
Show first that dim I) < 1. If dim b = 1, note that there would exist a deriva-
tion D of 9 such that D(g°) = {0) and D(b) is contained in the centre of 3‘7
(§ 4-, Exercise 8 (a)))
14. Let V be a finite-dimensional vector space over K and z an endo-
morphism of V. We adopt the notation z: of Exercise 3 of § 3. An endo-
morphism z’ of V is called a ”Mica of z if for all [I and 41 every zero of z: is a
zero of 2;”. Show that if Tr(zz') = 0 for every replica z’ of z, then 7. is nil-
potent. (Use the proof of Lemma 3. In the notation of this proof, prove in
particular that tis a replica of z.)
15. Let K be a field of characteristic 2. The identity representation of the
nilpotent Lie algebra 91(2, K) on K2 defines a semi—direct product is of
51(2, K) by K“. Show that b is solvable but that 90 is not nilpotent. Deduce
that 9 admin no faithful linear representation by triangular matrices. Show
also that the conclusions of Exercise 5 are false.
16. Let g be a Lie algebra over an arbitraIy field K, A a commutative
associative algebra over K and 9’ = g ®K A, which can be considered as a
Lie algebra over K.
(a) If D is a derivation ofA, show that there exists one and only one deriva-
tion D’ of 9’ such that D’(z ® a) = x ® Du for xeg, a GA.
(b) Let p be a prime number, G a cyclic group of order p and s a generator
of G. Suppose that K is of characteristic p and henceforth let A be the algebra
of G over K. Show that there exists one and only one derivation D of A such
that D0") = k:"'1 for all 1:52. Show that the K-linear combinations of
the x® (: — l)” (k =1,2,...,p —— l,xeg) form a solvable ideal I of 9’
and that 9’]: is isomorphic to g.
(0) Take 9 to be simple (cf. §6, no. 2, Definition 2). Then i: is the radical
ofg' butis not a characteristic ideal. (Observe thatD(x (8) (.r —- 1)) = 1 ® x.)1‘
17. Let g be a Lie algebra. Suppose that for every simple g-module M of
finite dimension over K the x” are permulable with one another. Show that
g is solvable. (Observe that 99 is contained in the nilpotent radical and hence
is solvable.)

T This result, " " was ' ’ to us by N. ' L

101
l LIE ALGEBRAS

§6
The conventions of § 6 remain valid unless otherwise mentioned.
fl1.Letgbea " l.‘ Lie ‘3‘“ andpafi" " ' ‘repre-
sentation of g on M.
(a) If p is simple and non-zero, H’(g, M) = (0) for all [1. (Use Proposition
1 and Exercise 12 (j) of§ 3.)
(b) For all p, H’(g, M) = {0). (If p is simple and non-zero, apply (a).
If p is zero, use the fact that g = 95. In the general case, argue by induction
on the dimension of 9: if N is a sub—g-module of M distinct from {0} and M,
use the exact sequence:
H‘(5, N) -> H‘(9, M) -> H‘(9, MIN)
established in Exercise 12 (f) of § 3.) Hence recover Remark 2 of no. 2.
(c) Deduce from (b) a proof of Theorem 2. (Use Exercise 12 (g) of § 3.)
Deduce also from (b) that every derivation of g is inner. (Use Exercise 12 (h)
of § 3.)
(d) For all p, H’(g, M) = {0). (Arguing as for ([7), it suffices to consider
the case where p = 0. Let 56H2(9, M). Consider, following Exercise 12 (i)
of § 3, the Central extension b of g by M defined by r. The adjoint representa-
tion of 9 defines a representation of g on b. By Theorem 2, M admits in l)
a supplement which is stable under 5 and hence the extension is trivial.
Hence 0 = 0 by Exercise 12 (i) of§ 3.)
(e) Deduce from (b) and (d) a proof of Theorem 5. (As in the text, it can
be reduced to the case where the radical is commutative.)
2. Let g be a Lie algebra, r its radical and (a0, a,, . . .) a sequence of ideals
of 9 defined as follows: (1) “a = {0); (2) «(H/ui is a maximal commutative
ideal of 5/4.. Letp be the smallest integer such that a, = a,“ = - . ., Show
that r = a,. (Show that 9/0,, is semi-simple.)
3. For a Lie algebra g to be semi-simple, it is necessary and sufficient that
it be reductive in every Lie algebra containing 9 as a subalgebra. (If g satis-
fies this condition, let M be a g-module of finite dimension over K. Consider-
ing M as a commutative algebra, form the semi-direct product of g and M,
in which 9 is reductive. Dcduce that M is semi-simple.)
4. Let g be a semi-simple Lie algebra and p a non-zero simple representation
of g on a finite-dimensional space M. Let b be the corresponding semi-direct
product. Show that I) = 91), that the centre of b is zero and that I) is not
the product of a semi-simple algebra and a solvable algebra.
5. Let n be a Lie algebra and tits radical. If t has a decreasing sequence of
characteristic ideals r = to D t, D - - - D tn = {0} such that dim rl/r,+1 = 1
for 0 < i < n, g is the product ofa semi-simple algebra and a solvable algebra.

102
 EXERCISES

(Let a be a Levi subalgebra of 9. For all x e 5, let 9(x) be the restriction of

21d to r. Then p is a direct sum of l-dimensional representations, which are
zero because a = 95.)
6. Let g be a Lie algebra and 9% the intersection ofthe 9% ([2 = l, 2,. . .).
Show that g/QQQ is solvable and that 9(9mg) = 9mg. For g to be isomorphic
to the product of a semi-simple algebra and a solvable algebra, it is necessary
and sufficient that 9mg be semi—simple.

7. (a) Let g be a Lie algebra, b a semi-simple ideal ofg and a the centralizer
of I; in 9, so that g is identified with I) x a. Show that, for every ideal I of 9,
then I = (I n b) X (1 n a)‘ (let ll be the canonical projection of 1 onto 1;;
it is an ideal of b and hence 9h = I]; deduce that II C t.)
(b) Let [3 be an invariant bilinear form on 9. Show that b and a are ortho-
gonal with respect to [3 (use the fact that I) = 9b and that {5 = (31 + {35,
where B, (resp. {52) is an invariant bilinear form whose restriction to a (resp.
b) is zero.
(a) Deduce from (a) that there mists in g a largest semi-simple ideal.
(Consider a maximal semi-simple ideal of g.)
8. Let g be a Lie algebra. An ideal I) of g is called minimal if b aé {0} and
every ideal of 5 contained in b is equal to {0} or b.
(a) Every simple ideal of g is minimal.
(b) Let b be a minimal ideal of g and t the radical of 9. Then either I) C r,
in which case b is abelian, or b n r = {0}, in which case I) is simple. (Use the
fact that the derived ideal of a Lie algebra and the simple components of a
semi-simple Lie algebra are characteristic ideals.)
9. For a Lie algebra g to be reductive, it is necessary and sufficient that its
centre c be equal to its largest nilpotent ideal. (If the condition holds, let I:
be the radical of 9; 9r is contained in the centre of t, hence I: is nilpotent
and hence r = c.)
10. Let g be a Lie algebra such that the conditions at e g, y e 9, [[1, y], y] = 0
imply Deg] = 0. Show that g is reductive. (Show that the radical r of 5 is
commutative using Exercise 3 of § 5. Then show that [9, r] = 0.)
ii 11. (a) Let g be a Lie algebra, 1: its radical, o a Levi subalgebra of g
and m an ideal of g containing r. There exists an idea] i of a, supplementary
to m in 9, such that [111,1] C r. '
([7) Let g be a Lie algebra and a a subinvariant subalgebra of 9. Then
there exists a composition series 9 = 9., :> 51 3 ~ - - D 9,: 2 n such that g, is
the direct sum of g”, and a subalgebra I)i which is either l-dimensional and
contained in the radical r(g,) ofg. or simple and such that [9,, g‘H] C t(g.H).
(Reduce it to the case where u is an ideal of 9. Let 9’ = g/a and r’ be the
radical of 9’. The algebra g'lr’ is the product of its simple ideals n1, «2, . . ., etc.

103
! LIB ALGEFRAS

Take a composition series of 9’ consisting of a composition series of r' whose

successive quotients are 1-dimensional and the inverse images of the iduls
aha, x (12,...”1 x a; x x ac. Then take the inverse image in g of
this composition series.)
12. Let g be a Lie algebra, r its radical and D a derivation of g.
(a) If D(r) = (0}, D is inner. (Let c be the centralizer of r in g. The adjoint
representation of 9 defines a representation x" r—> p(x*) of g" = g/r on c. On
the other hand, [D(g), r] C D([g, r]) + [9, D(r)] = {0}, hence D(g) C t and
hence D defines a linear mapping D*: g‘ —> c. Show that
D*([~*,y‘]) = 9(x‘)D‘y* - 9(y‘)D*X*
for all x“, 11* in 9*. Deduce that there exists me: such that D*x* = p(:f’)a
for all x“ e 9* (cf. no. 2, Remark 2), whence D}: = [x, a] for all x e g.)
(b) If D coincides on r with an inner derivation of g, D is inner. (Apply
(3).)
13. Let g be a Lie algebra over an algebraically closed field, I: in radical
and p a finitedimensional simple representation of 9. Then 9(2) is scalar for
z e r. (Reduce it to the case where g is reductive. In this case, r is the centre
of g.)
14-. Let X be a nilpotent matrix in 51(n, K). Show that there exist in gl(n, K)
two other matrices Y, Hsuch that [H,X] = X, [H, Y] = —Y, [X, Y] = H.
(Argue by induction on n, using the Jordan form of X; thus reduce it to the
case where X = Em + E23 + -- - + EFL”; then take Y to be a combina-
tion of the Ek, k-) and H a combination of the Eu.)
fl 15. (a) Let X, Y be two matrices of gl(n, K). Show that, if [X, Y] = X,
X is nilpotent. (For every polynomial f6 Km, observe that
[lL Y] = Xf’(X)-)
([1) Let 9 be a Lie algebra and h, a: two elements of 9 such that [h, x] = x
and there exists 2 e g with [x, z] = In. Show that there exists y e 9 such that
[x, y] = h and [h, g] = —y. (Observe first that [2, h] + 2 belongs to the
centralizer n of x in 9. Then show that n is stable under ad h and that the
restriction to n of I — ad h is bijective; to do this, note that ad x is nilpotent,
using (a), and prove that, if g, is the image of 5 under (ad x)‘, I — ad It gives
on taking quotients a bijection of (n n g,_1)/(n n 9,); for this use the relation
[ad 2, (adz)"] = —k(adx)""(adh + k g 11).)
(a) Let g be a subalgebra of 91(n, K); suppose that, for every nilpotent
matrix X E 9, there exist two other matrices H, Y in 9 such that [H, X] = X,
[H, Y] = — Y, [X, Y] = H. Let b be a subalgebra of 9 such that there exists

104-
 EXERCISES

a subspace m supplementary to b in g and such that [9, m] C In. Show that

I) has the same property as g for all its nilpotent matrices (use (b)).
16. Let g be a subalgebra of gl(n, K) such that the identity representation
of g is semi-simple. Show that every nilpotent matrix X e g is contained in
a 3-dimensional simple Lie subalgebra of 9. (Show that g is reductive in
gl(n, K) ; then use Exercise: 14 and 15 (c).)
17. Show that the algebra derived from a real simple Lie algebra by ex-
tending the base field from R to C is not always simple. (Let g be a real
simple Lie algebra. If g' = 9(a) is simple, let I) be the real Lie algebra derived
from g' by restricting the field of scalars to R. We know that I) is simple.
Then he) is, by Exercise 4 of § 1, the product of two algebras isomorphic
to 9" Cf. also Exercise 26 (b).)
18. (a) Let g be a simple Lie algebra. Every invariant bilinear form on
g is either zero or non-degenerate. If K is algebraically closed, every in-
variant bilinear form [3 on 9 is proportional to the Killing form flu. (Consider
the endomorphism 5 Of the vector space 9 defined by fix, 1/) = (30(5):, 1/) and
show that a commutes with adgx and is hence scalar.) Show that this result
may be false if K is not algebraically closed. (Use Exercise 17 and the fact
that the dimension of the space of invariant bilinear forms does not change
when the base field is extended.)
(1:) Let g be a semi-simple Lie algebra over an algebraically closed field.
Deduce from (a) and Exercise 7 (b) that the dimension of the space of in-
variant bilinear forms on g is equal to the number of simple components of
g and that all these forms are symmetric.
(a) Let g be a simple Lie algebra, M the dual space of the vector space 9,
with the dual representation of the adjoint repraentation, and I) the semi-
direct product of 9 and M defined by x >—> at“. For g, z in 5 and y', z' in M,
we write My + y', z + z’) = (y, z’) + (z,y’>. Show that L3 is a non-de-
generate invariant symmetric bilinear form on b, which is associated with
no representation of b. (Observe that the radical and nilpotent radical of b
are equal to M.)
19. Let g be a reductive Lie algebra, U its enveloping algebra, Z the centre
of U and V the subspace of U generated by the elements of the form an — 1214
(u e U, u e U).
(a) U is the direct sum of Z and V. (Apply Proposition 6 of §3 to the
representation x»—>adux of g on U, observing that, in the decomposition
U = 2: U" of § 2, no. 7, Corollary 4 to Theorem 1, the U" are stable under
this representation.)
(b) Let u»—>u“ be the projector of U onto Z parallel to V. Show that
Ow)" = (014)", (zu)ll = zul1 for all u E U, u E U, z E Z.

105
I LIE ALGEBRAS

(5) Let R be a two-sided ideal of U. Show that R = (R n 2) + (R n V).

(To show that the component in V of an element 7 of R belongs to R, reduce
it to the case where K is algebraically closed; observe that R is stable under
the representation 9 of U on U which extends xn—> adux; decompose 1 into
its components in the U"; finally, apply Corollary 2 to Theorem 1 of Algebra,
Chapter VIII, §4, no. 2 to the restriction of p to U".)
(ii) Let R’ be an ideal of Z. Let R1 be the two-sided ideal of U generated
by R’. Show that Rl n Z = R’. (Use (b).)

20. Suppose that K is algebraically closed. Let g be a Lie algebra and c

its centre.
(a) If g admits a. finite-dimensional faithful simple representation, 9 is
reductive and dim c < l.
([1) Conversely, if g is reductive and dim c < 1, g admits a finite-dimen-
sional faithful simple representation. (If g is simple, use the adjoint repre-
sentation. If 5 = 91 x 9, and g‘, g, admit finite-dimensional faithful simple
representations 9;, p, on spaces M1, M2, show that

(x1, *2) H P1("1) ® 1 + 1 ® 920(2)

is a semi-simple representation p of 9, then that the centralizer of the g-module

M1 ® M2 reduces to the scalars and hence that p is simple. When 9, and g,
are semi-simple or g, is semi-simple and g, is commutative, show that p is
faithful by considering the kernel of p which is an ideal of g, x 92.)
21. Let V be a vector space over K of finite dimension n > 1. Show that
eI(V) is simple. (Reduce it to the case where K is algebraically closed. Sup-
pose that bI(V) = a x b, with dima = a > 0, climb = b > 0. Let A (resp.
B) be the associative algebra‘generated by l and a (rap. h). Then V can be
considered as a simple (A ®x B)-module. Hence there exists an A-modulc P
of finite dimension 1) over K and a B—module Qof finite dimension q over K
such that V is (A ®x B)-isomorphic to P ®x Q (Algebra, Chapter VIII, § 7,
no. 7, Proposition 8 and no. 4, Theorem 2); P and Qare faithful and hence
4 <1)“, b < 9”. On the other hand, a + b = n’ —l and fig = n, whence
([72 — l)(q2 — l) < 2, whichis contradictory.)
22. (a) Let g be a nilpotent Lie algebra over an algebraically closed field
K of arbitrary characteristic. Let a be its centre, p a finite-dimensional repre-
sentation of g and 6 the associated bilinear form. Show that z n 99 is ortho-
gonal to g with respect to 1%. (Reduce it, using ajordan-Holder series, to the
case where p is simple. Then note that, for z e 3 n 9g, 9(2) is a scalar matrix
with trace zero. Then use Exercise 22 (b) of § 4.) Deduce that, if (3 is non-
degenerate, g is commutative. (Use Exercise 3 (a) of § 4-.)
(b) Suppose that K is of characteristic 2. Show that for the solvable Lie

106
 EXERCISES

algebra 9K2, K) = g the bilinear form associated with the identity repre-
sentation is non-degenerate, but that the centre 3 is contained in 99 and is
95 {0}-
(0) Let g be the 6<dimensional Lie algebra over K with basis (a, b, c, d, e,f)
and multiplication table [11, b] = —[b,a] = d, [11,0] = —[c, a] = 2,
[b, t] = — [5, b] = f and the other brackets are 0. Let B be the bilinear form
onasuchthatmd)=p(d,a)=1,3(a,f)=a(fia)=1,a(b,e)=p(a,b)= —1
and the other values of p at ordered pairs of the basis of g in question are 0.
Show that B is invariant, g nilpotent, a z 99 aé (0} and fl non-degenerate.
23. Let g be a 3«dimensional non-commutative Lie algebra over a field
K of arbitrary characteristic.
(a) Ifg has centres yé (0}, then dim 3 = 1 (§ ],Exercise 2) and dim 9g = 1.
If 3 # 9g, 5 is the product of 3 and the 2-dimensional non-commutative
algebra. If a = 95, g is the 3-dimensional non-commutative nilpotent algebra
(§ 4, Exercise 9 (a)).
(b) If a = {0} but there exists in g a 2-dimensional commutative sub-
algebra, this subalgebra is unique and is equal to 99; for all x¢99, the
restriction u of ad x to 99 is a bijection of this vector space, determined to
within a multiplicative constant. Conversely, every automorphism u of a
2-dimensional vector space Ka + Kb determines a Lie algebra structure on
g = Ka + Kb + K: by the conditions [a, I7] = 0, [0, a] = u(a), [2, b] = u(b).
For two Lie algebras thus defined by automorphisms ul, un to be isomorphic,
it is necessary and sufficient that the matrices of 141 and 112 be similar to within
a scalar factor.
(0) Suppose that there exists in 9 no commutative subalgebra of dimension
>1. Show that there then exists a E 9 such that a $‘(ad (1)9 (assuming that
an element 2: does not have this property, show, using the jacobi identity,
that another element has the desired property). There then exists a basis
(a, b, c) of 9 such that [a, b] = c, [a, t] = {517, [b, c] = 7a, with (3 ye 0 and
Y aé 0, and g is simple. Let 45 be a canonical isomorphism of 9 onto the ex-
2

terior power Ag“ of the dual space of 9, determined to within an invertible

constant factor (Algebra, Chapter III, §8, no. 5); let u be the linear mapping
of g into 9* defined by the condition <[x,y],u(z)> = (x /\ y,d>(l)>; the
bilinear form (D(x, y) = (x, u(y)> on g x g is symmetric and non-degenerate
and 2(1) is the Killing form of g to within a factor #0. For two 3-dimensional
simple Lie algebras over K to be isomorphic, it is necessary and suflicient that
the corresponding bilinear forms be equivalent to within a constant factor
aé 0.
(d) If K is not of characteristic 2, the simple algebra 9 defined in (r) is
isomorphic to the Lie algebra o((l>) of the formal orthogonal group GOD)
(§ 1, Exercise 26 (11)). For the existence in g of vectors x such that ad x admits
eigenvectors not proportional to x, it is necessary and sufficient that d) be

107
l LIE ALGEBRAB

of index >0; 9 then admits a basis (a, b, t) such that [4, b] = b, [a, c] = —c,
[b, o] = a.
(e) If K is of characteristic 2, show that there is no 2-mapping on g and
therefore that g is not the Lie algebra of a formal group. Show that g admits
derivations which are not inner.
1[ 24-. Let K be a field of arbitrary characteristic p. We again adopt Defini-
tion 1.
(a) Show that, unless p = n = 2, the only non-trivial ideals of 91(71, K)
n
are the 1-dimensional centre 3 with basis ‘21 E“ and the Lie algebra 91(71, K)
consisting of the matrices of trace 0. (Except in the exceptional case indicated,
note that if an ideal 0 contains one of the Eu, it necessarily contains 51(n, K).
If a contains an element not belonging to a, by multiplying this element by
at most four suitably chosen elements E", a non-zero multiple of one of the
E” is obtained.) If n is not a multiple ofp, show that gl(n, K) is the direct
sum of a and sl(n, K) and sl(n, K) is simple. If on the contrary I: is a multiple
of}: and n > 2, show that al(n, K) [3 is simple (same methods).
(b) Show that, for n a multiple ofp and n > 2, gl(n, K)/3 has radical {0},
but admits the quotient gl(n, K) /e[(n, K) which is Abelian.
fl 25. Let K be a field of arbitrary characteristic [1. Let ap(2n, K) denote
the Lie algebra of the formal symplectic group in 2» variables over K (§ 1,
Exercise 26 (0)). Show that this Lie algebra, which is identified with a sub-
algebra of gl(2n, K), has basis consisting of the elements
Ht = Eu — Ewan” (1 $ iS ’1):
Ft] = Eu — E1+n.l+n (1 < if g "1" #j):
Gil = Eun- + Em“: Gil; = Et+n.; + Ei+ml (1<i<j < n),
Emu and Eli-ml (1 < i< ’0’
(a) Show that, if 1; ye 2 and n 2 l, the algebra np(2n, K) is simple (for
n = 1, 619(2, K) = 91(2, K)), (Same method as in Exercise 24.)
(b) If]: = 2 and n 2 3, show that the H” E), G}, and GI, form an ideal
a of dimension n(2n — l); the elements of a of the form

2 NH: + m
1-1 Z “41E; + R]
Z $11G]! 'l' «1793;;
2

such that 4;: A. = 0 form an ideal b C a of dimension 71(2): — l) — l and

the multiples of 2:1 H. an ideal c, the centre of sp(2n, K); b, c are the only

108
 EXERCISES

ideals of a, b = 9(sp(2n,K)) and b/c is simple. (Same method.) What are

the ideals of sp(4, K) when K is of characteristic 2?
ll 26. Let K be a field of characteristic #2 and (I) a non~degenemte sym-
metric bilinear form on an n—dimensional vector space over K.
(a) Show that, if n 2 5, the Lie algebra 0(4)) is simple. (By extending
the base field, it can be reduced to the case where the index of (D is m = [fl/2].
If for example n = 2m is even, 0(0) has basis consisting of the elements
Hi = Ea, — El+m.H-ma F41: Eu — El+m.l+n (1<i:j S ”1:1. 99].):
64’] = Emu.- — Em”; and G"; = Eum.’ — Elwm (1<i<j $ m);
then argue as in Exercises 24 and 25. Proceed similarly when n = 2m + l
is odd.)
([2) Let A be the discriminant of (I) with respect to a. basis. Suppose that
n = 4. Show that, if A is not a square in K, 0(a)) is simple. If A is a square,
0(4)) is the product of two isomorphic 3-dimensional simple Lie algebras.
(Use the structure of GOD) described in Algzbra, Chapter IX, §9, Exercise
16.) Deduce an example of a simple Lie algebra which becomes non-simple
when extending the base field.
27. (a) Show that, in a finite-dimensional Lie p—algebra, the radical is a
p-ideal. (Use Exercise 22 (6) of § 1.)
(b) For a Lie p-algebra g to have zero radical, it is necessary and sufficient
that 9 contain no commutative p—ideal #{0}. (Use Exercise 22 (r) of § 1.)

§7
The conventions of § 7 remain valid unless otherwise mentioned.
1. (a) For a Lie K-algebra to be nilpotent, it is necessary and suflicient
that it be isomorphic to a subalgebra of an algebra n(n, K).
(b) Suppose that K is algebraically closed. For a Lie K-algebra to be
solvable, it is necessary and sufficient that it be isomorphic to a subalgebra
of an algebra t(n, K).
2‘ Let g be a Lie algebra and 11 its largest nilpotent ideal. There exist a
finite-dimensional vector space V and an isomorphism of 9 onto a subalgebra
of 91(V), which maps every element of n to a. nilpotent endomorphism of V.
(Use Ado’s Theorem and Exercise 5 of § 1.)
3. Let g be a Lie algebra and U its enveloping algebra.
(a) Show that for all u EU (u aé 0) there exists a finite-dimensional repre-
sentation 1: of U such that 1r(u) aé 0. (Use Exercise 2 and Exercise 5 (b) of

(b) Deduce from (a) that if g is send-simple there exists an infinity of in-

109
I LIE ADGEBRAS

equivalent finite-dimensional simple representations of 9. (Let pl, , . ., p" be

finite-dimensional simple representations of 9, N1, . i ., N” the annihilators of
n
the corresponding U—modules and N = ‘Q N‘. Show that N is of finite co-
dimension in U. Let n e N, n gé 0. Apply (a) to n.)
1T4. Let g be a Lie algebra, a a subinvariant subalgebra of g, r(a) the
radical of a and p a finitedimensional representation of a. For 9 to admit
a finite-dimensional extension to 9, it is necmsary and suflicient that 99 n r(a)
be contained in the largest nilpotency ideal of 9. (For the sufficiency proceed '
as follows; let 9 = go 3 gl 3 3 9,, = a be a decomposition series of g
with the properties of Exercise ll (b) of § 6. Show by induction the existence
of an extension 91 of p to 9,, whose largest nilpotency ideal contains 99 fl r(g,).
In the notation of Exercise I] (b) of § 6, the passage from Pm to 9‘ follows
from Theorem 1 when 9; is simple, or when I); is l-dimensional and
99 n mm) = 99 n «9.).
When 99 A rm“) # 95 n r(g.), it can be assumed that b: C 99 n r(g,).
Then r(g() = t(9) n 9. (§ 5, Exercise 10) and Theorem 1 can be applied)
1] 5. Let K be a field of characteristic [2 > 0, g a Lie algebra of finite
dimension n over K and U the enveloping algebra of g.
(a) A p-palynamiat (in one indeterminate) over K is any polynomial in
K[X] whose only terms #0 are of degree a power ofp. Show that for every
polynomial f(X) aé 0 in K[X] there exists g(X) e K[X] such that f(X) g(X2
is a p-polynomial (consider the remainders in Euclidean division of the X"
wag).
(b) Show that for every element 2 E 5 there exists a non-zero polynomial
f(X) eK[X] such that f(7.) belongs to the centre C of U. (Consider the
minimal polynomial of the endomorphism ad 2 and apply (a) to this poly-
nomial, and also formula (1) of Exercise 19 of § 1.)
(5) Let 00“,“ be a basis of 9. For each 1' letf. aé 0 be a polynomial in
K[X] such that f,(e.) s 0; let 41‘ be its degree and L the two-sided ideal of U
generated by the y, = f‘(e¢). Show that the classes mod. L of the elements
2:1 . . 1;", where 0 g at, < :4 form a basis ofU/L.
(:1) Show that g admits a finite-dimensional faithful linmr representation
(choose the fi so that d‘ > 1 for all i).
(e) Show that if g aé {0} there exists a non-semi—simple finite-dimensional
linear representation of 9. (Assuming that thef, are of degrees d, > 1, replace
the f. by f? in the definition of L and note that there then exist nilpotent
elements 960 in the centre of U/L and hence that U/L is not semi-simple.)

llO
 CHAPTER II

Free Lie Algebras

In this chaptnd’ the letter K denotes a rum-zero commutative ring. 771: unit alemm! 0f K
i: deflated by l. Unlm otherwise mlioned, all algebra, algebra: and bigebras, all
modules and all tmrm products are over K.
From §6, K will be maimed to be afield qf characteristic 0.

§ 1. ENVELOPING BIGEBRA OF A LIE ALGEBRA

Throughout this paragraph, 9 will denote a Lie algebra over K, U(g) or simply
U its enveloping algebra (Chapter I, § 2, no. 1), a the canonical mapping of 9
into U(g) (lac. oil.) and (U,),,,., the canonical filtration of U ([00. 012., no. 6).

l. PRIMITIVE ELEMENTS OF A COGEBRA

Throughout this no. we consider a cogebra E (Algebra, Chapter III, § ll, no. l)
with copmduct
c: E -—> E ® E
and munit a (lot. cit, no. 2). Recall that a is a linear form on the K-module E
such that (with the canonical identification of E ® K and K ® E with E):
Ida = (:43; Id!) up = (IdE (3):) ac.
Let E“ denote the kernel of s and let u be an element of B such that
0(a) = 11 ® 14 and 3(a) = I.

1' The results of Chapters II and III depend on the first six books, on Lie group:
and Lie Algebras, Chapter I, on Commutative Algebia and on Dtfirmfiable and Anulytir
Manifolds, Summary of Results; no. 9 of § 6 of Chapter III depends also on Spmml
Theories, Chapter I.

111
 11 FREE LIE ALGEBRAS

The K-module E is the direct sum of E‘L and the submodule K. u which is free
with basis 11; let nu: E —> E + and 1;“: E —> Kru denote me projectors associated
with this decomposition. Then
(I) m) = x — 6(x).u. mm = swu-
DEFINITION 1. An element x of E i: called u-primitiw if
(2) c(x)=x®u+u®x.
The u-pn'mitive elements of E form a submodule of E, denoted by Pu(E).
Pnoposmon 1. Every uwprimz'tiue dawn! of E belong: In E”.
(2) impliesx = :(x).u + 501).): = e(x).u + x, whence e(x) = 0.
Remark. Ife and c(x) = x’ 80 u + u ®x’, where x’, x' are in E+, then
x = e(x’) .u + :(u) .z” = x"; similarlyx = x’ and x is u-pfimitive.
For all :6 E“, we write
(3) £:(z)=c(x)—x®u—u®x.
PROPOSITION 2. W: have
(4) (1n, ® 1:”) co = c: o 1:".
Let x be in E; then
("u ° "0000) = ((1 - m) 8) (1 - 7)u))(€(x))
= 0(3) - (1 ®m)(‘(*)) — (m ® 1)(€(*)) + (m ® 71000))-
As z is counit of E,
(1 ® m)(‘(x)) = ’6 ® '1’ (m ®1)(F(*)) = u ® 3
whence
(m ® 71000)) = (m ® 1)((1 ® 140900)) = 50‘) -'4 ® 11;
fiom this we conclude
(in ® m)(v(x)) = v(~) — x ® u — u 8:: + a(x).u ® a.
On the other hand,
mum) = 0(z) — x®u — 1403:: + s(x).u®u,
whence formula (4).
As E+ is a direct factor submodule of E, E"® E+ can be identified with a
direct factor submodule of E 8) E. With this identification, 12,, ® 1:“ is a pro-
jector ofE ® E onto E*® E+. Byformula (4), 5,: maps E+ into E*® E’r and
1vu is a mam/tint: of the cogebm (E, 5) int!) the algebra (E+, a: ).

112
 PRIMJTIVE ELEMENTS OF A BIGEBRA § 1.2

PROPOSITION 3. If the eogebra (E, e) is evmsocialive (resp. aocammutative) (Algebra,

Chapter III, § 11, no. 2), .m is the eagebm (Eh 5:).
This follows from the following lemma:
Lemma 1. Let 7:: E —> E’ be a .mfiedive eagebm morphism. [f E is avwaeiatiue (resp.
eoeammutative), so is E'.
Let B be an associative K-algebra; the mapping f1—»fa 1: is an izy'eetive
algebra homomorphism of HomK(E', B) into Homx(E, B). It then suffices to
apply Proposition 1 (resp. Proposition 2) of Algebra, Chapter III, § 11, no. 2.

Z. PRIMITIVE ELEMENTS OF A BIGEBRA

Let E be a bigebra (Algebra, Chapter III, § 11, no. 4), 0 its coproduct, 2 its
counit and 1 its unit element. As e(l) = 1 and 2(1) = 1 ® 1, the results ofthe
previous no. can be applied with u = I. The l-primitive elements of E (no. 1,
Definition 1) are simply called primitive (cf. Algebra, Chapter III; § 11, no. 8),
that is the elements x of E such that
(5) e(x)=x®l+l®x.
We write simply 1r, 1;, P(E), 6* instead of 1:1, 1],, P1(E), of .
PROPOSITION 4. The at P(E) of primitive element: 9/ E is a Lie subalgebm qf
E
I”, y are in P(E), then

0W) = ¢(x)€(y) = (18” + 1 ®~)(y®l + 1 ®y)

=xy®l + l ®xy+x®y+y®x,
whence
t([~,y]) = [m] ® 1 + 1 69 [mil
PROPOSITION 5. Letf: E —> E’ be a bigebm morphism. (fr is a primitive elemen! afE,
then f(x) is a primitive element of E' and the restriction off to P(E) is a Lie algebra
homamorphism P(f): P(E) —> P(E').
Let c (resp. c’) be the coproduct ofE (resp. E'). Sincefis a cogebra morphism
0’ cf: (f®f) or, whence

0'(f(*)) = (f®f)(0(x)) = (f®f)(x®1 + 18")

=f(")®l +1®f(x),
for x primitive. Hence f maps P(E) into P(E') and f([x, 31]) = [f(x), f(y)]
since f is an algebra homomorphism.

Remarks. (1) Let p be a prime number such that pJ = 0 in K. The binomial

formula and the congruences (‘0) E 0 (mod. p) for l < 1' $ [1 — 1 imply that

113
11 FREE uE menus

P(E) is stable under the mapping x H x”.

(2) By definition, the diagram
0 ._> P(E) _> E+ ”—3 E+®E+
is an exact sequence. If K’ is a commutative ring and p: K—>K’ a ring
homomorphism, p“ (E) = E ®K K’ is a K’-bigebra and the inclusion P(E) —> E
defines a homomorphism of Lie K'—algebras
at: P03) ®K K’ —> P(E ®K K').

If K’ is flat over K (Commutative Algebra, Chapter I, §2, no. 3, Definition 2),

it follows from lac. cit. that the diagram
ct Exmx'
O -—> P(E) ®x K’ —> E+®KK 1 —> (E“®K K’) ®Kt (E"®x K’)
is an exact sequence, which implies that a: is an iromafphism.

3. FILTERED BIGEBRAS
DEFINITION 2. Let E be a bigebra with coproduct e. A filtration compatible with the
higebm :hueture an E is an iiu‘reasing sequence (E,,),Do ofsubmadule: of E such that
E°=K.l, E=nL)J°E,,
(6) EWE" C Em” for m 2 0,71 2 0
0(En) C H," Im(Ei ® E,) for n 2 0.1'
A bigehra with a filtration eampatible with its bigebra structure is called a filtered
bigebra.
Example. Let E be a graded bigebm (Algebra, Chapter III, § 11, no. 4, Defini-
1;
tion 3) and (E"),,,° its graduation. We write E" = l; E‘. The sequence (13,.) is
a filtration compatible with the bigebra structure on E.
PROPOSITION 6. Let E be afiltered bigebm and (E,),>0 it;filtration. For every integer
n2 0,letE,f = EnnE". TIME; = {0}aml
n— l

(7) e+(E,f) C 2; Im(Ef® Enfll) for n 2 0.T

As En = K.l, E; = 0. Ife," 1r(x) = x — e(x) .l (formula (1)), whence
"(1) 6E: and «(E,) C E:. It follows that 1r (8) 1: maps Im(El ® E,) into

1’ If A and B are two submodules of E, we denote by Im(A ® 3) the image of

the canonical mapping A ® B —> E ® E.

114
 ENVELOPING BIGEBRA or A Ln: ALGEBRA § 1.4

Im(E,+® E?) fori 2 O,j 2 0. As 0* = (1: ® 1:) u t in E.+ (no. 1, Proposition

2), by (5)
n n—l

owl-1:) c 2 Im<Er® 13.2.) = ,Z more Em.

COROLLARY. The element; if E; are primitive.
Ifx E Ef, then #0:) = 0 by (7), whence (5).

4. ENVELOPING BIGEBRA OF A LIE ALGEBRA

Recall that 5; denotes a Lie algebra and U its enveloping algebra, with its
canonical filtration (U”)n>o.
PROPOSITION 7. There exirtr an the algebra U (me and only one toproduet t which
make: U iata a bigelmz ruch that the element: of 6(9) are primitive. The bigebra (U, 5)
ix totammutatiue; its want! it the linearform a sueh that the mutant term (Chapter I,
§ 2, no. 1) afevery element x iy‘U is e(x) . l. The canonical/filtration (Un)n>0 QfU 1':
compatible with this higebm :trueture.
(a) Letxeg;wewrite 090:) = 6(1) ®1 +1® 6(x) EU (8 U. Ifx,y are in
g, then
ammo)
whence
= (6006(0) e! + 1 as («Mac») + «(:0 ® aw) + «(11) ® cor),
[‘oix)» to(y)] = €o([x, 5/1)-
By the universal property of U (Chapter I, §2, no. 1, Proposition 1), there
exists one and only one unital algebra homomorphism
t:U—>U®U

such that c(o(x)) = c(x) ® 1 + 1 ® 60:) for x e g. This proves the uniqueness
assertion of Proposition 7.
(b) We shew that t‘ it coastatialive. The linear mappings e’ and c” of U into
U ® U (8 U defined by
e' = (e®Id,_,) at and t” = (Idu®e) or
are unital algebra homomorphism which coincide on u(g) since, for a E 5(9),
c’(a)=a®1®l+1®a®l+l®l®a=e'(a),
whence the result.
(c) We show that e is toeammutatiae. Let 'r be the automorphism of U ® U
such that 1(a ® b) = h ® a for a, h in U. The mappings r o c and e ofU into
U (8 U are unital algebra homomorphisms which coincide on 0(9), whence the
result.
(d) We thaw that e 1': a taunt! for e. The mappings (IdU ® a) a e and

115
11 FREE LIE ALGEBRAS

(5 ® Idu) o r, of U into U are unital algebra homomorphism which coincide

with Idu on 6(9).
(e) We know that U" = K. 1, U“ c Um, U = .90 U,, and U,.U,,, c Um,
(Chapter I, §2, no. 6). Let a1, . . ., an be in a(g). Then

(8) 5(a1...a,,)=]‘.:I;t(a‘)=u(a,®l+l®a¢)

= "nag” (“am- - 410)) ® (“raun- ~ Jam):

where I(i) denotes the set of permutations of [1, n] which are increasing in
each of the intervals [1, i] and [i + l, a]. As U,I is the K-module generated by
the products ofat most n elements ofa(g), formula (8) implies that the filtration
(Un) is compatible with the bigebra structure of (U, o).
Damn-non 3. The bigebra (U, c) i: called the enveloping bigebra of the Lie
algebra g.

PROPOSITION 8. Let E be a bigebra with toproduet denoted by cm and let h be a Lie

algebra homomorphism g" 9 into P(E) (no. 2, Proposition 4-). The unital algebra
hommarphimf: U —> E such thatf(a(x)) = h(z)for all x 6 9 i: a bigebra morphiym.
We show that (f ®f) o c = 05 of. These are two unital algebra homomor-
phisms of U into E ® E and, for a 6 6(9),

(f®f)(t(a)) =f(a) ® 1 + 1 ®f(a) = 02(f(a))

since f(a) 6 P03). Similarly if :E is the counit of E, an of is a unital algebra
homomorphism U —> K which is zero on 0(9) (no. 1, Proposition 1) and there-
fore ooineides with a.
It follows from Propositions 5 and 8 that the mappingf»—>f o 0 defines a one-
to—one correspondence between bigebra homomorphisms U(g) —> E and Lie
algebra homomorphisms g —> P(E).
COROLLARY. Let gl (1' = l, 2) be a Lie algebra, U(g,) it: enveloping bigebra and
«3,: g, —> U(g.) the canonical mapping. For everyLie algebra homomorphism h: g, —> gm
the unital algebra homomorphixm U(h): U(g,) —> U(g,) such that U01) 0 el = 62 a In
(Chapter I, §2, no. 1) i: a bigebra morphism.

5. STRUCTURE OF THE COGEBRA U(9) 1N CHARACTERISTIC 0

In this no., K will be assumed to be afield oftharacterixlie 0.
Let 5(9) be the symmetric algebra of the vector space 9, es its eoproduet
(Algebra, Chapter 111, § 11, no. 1, Example 6) and n the canonical isomorphism

116
 STRUCTURE on THE COGEBRA U(5) IN CHARAm'ERxsnc 0 § 1.5

of the vector space 5(9) onto the vector space U (Chapter I, § 2, no. 7). Recall
thatit...,xnareing, then
1
(9) 7)("1- - J») = Ergo. “(find . - 59100).

In particular, for x e 9 and n 2 0

(10) not") = 60‘)"-
Note that by Algebra, Chapter III, §6, no. I, Remark 3, n is the unique [inal-
mapping of 5(9) into U satisfying condition (10).
PROPOSITION 9. For every integer n> 0, let U" be the vector subspace qf U generated
by the a-(x)"fi:r x e g.
(a) The :equeme (U"),,>0 i: a graduation oflhe vector space U compatible with 61x
cagebra :tmcture.
Let U be given the graduation (Un).
(b) The canonieal mapping 1): 5(a) —> U is an iwmorphina qfgraded aogebrar.
Let aces and neN. Then

(11) w") = cs(x)" = (m1 + 1 ®x)" = ‘20 (grow-t

since is is an algebra homomorphism. Similarly, by (10),

(12) v(n(’t")) = 00106)") = 001(k))" = (6(x) ® 1 + 1 ® 000)"

=26).()x)‘®c(x-‘=§(?)n(¢)®n(x”"),
whence
(n ® ’2)(‘s("")) = MW»-
As the x", forx e g and n e N, generate the vector space 5(9), (74 ® 71) 0 cs = c a '4
and 1) is a cogebra isomorphism.
On the other hand, formula (10) shows thatn($"(g)) = U", which completes
the proof of (a) and (b) taking account of the fact that the graduation of 5(g)
is compatible with its cogebra structure.
The graduation (U"),,;., of U is called the mamfeal graduation.
COROLLARY. The canonical mapping 6 define: an immatphirm afg onto the Lie algebra
P(U) qn'mitiI/e element: of U.
As 6" is a graded homomorphism of degree 0,

P(U) = n; (P(U) n U").

117
11 FREE LIE AIIGEBRAS

It suffices to prove that if n > 1 and a EU" is primitive, then a = 0. Now

a can be written as 2‘: Mar, where A, GK, :1‘ 5 0(9). By (12), the term of hi-

degree (1, n — 1) in ”(-1) is n; he. ® Rf”. Hence 2 Am ® a?" = 0. If

a 8 U —> U is the linear mapping defined by multiplication on U, then

a = (EM? = M(Z?«ag®ai'") = 0-
9:
Remarks. (1) Un = 420 U‘ (Chapter I, §2, no. 7, Corollary 4 to Theorem 1).
(2) The mapping n is the unique morphism of graded cogebras of 5(5) into
U such that 11(1) = l and not) = a(x) for x55. For if 11’ is a. morphism
satisfying these conditions, we prove by induction on n that n'(x") = TKX") for
n-l

x E g and n >1. As 030:") = 4:

21(2):! 8) x"“ by (3) and (11),

(n ® we; (9)) = (71' ® 71') (0s? (*"))

by the induction hypothesis. It follows that 5*(7)(x")) = 6*(n'(x")); it follows
that not") — n’(x") is a primitive element of degree n and hence is zero (Corol-
lary to Proposition 9).
(3) Let \IJ be the canonical isomorphism of the bigebra TS(9) onto the
bigebra S(g) (Algebra, Chapter IV, §5, Corollary 1 to Proposition 12). The
mapping
71 ° in T5(9) -> U
is called canonical. It is the unique morphism 11' of graded cogebras of
TS(g) into U such that 7/(1) = 1 and n’(x) = 6(X) for all x E g.
(4.) Let V be a vector space. The primitive elements of the bigebra S(V)
are the elements of degree 1. This follows from the Corollary to Proposition 9
applied to the commutative Lie algebra V.
Let (filter be a basis of the vector K-space 3, where the indexing set I is
totally ordered. For all a e N“), we write

(13) em = H “Wm.
is! m(i)!

The e“, for [at] < 1:, form a basis of the vector K-space U,I (Chapter I, §2,
no. 7, Corollary 3 to Theorem 1). Then

20 = 1, ‘2. = 6(0.) forieI.

118
 STRUCTURE or FILTERED BIGEBRAS 1N CHARACTERISTIC 0 § 1.6

As the graded algebra associated with the filtered algebra U is commutative

(lot. 011., Theorem 1), for at, (-3 in N‘D,
(l4) ewe” E ((m, fl)).zu+fl mod. UIuIHfll—l-
where
(a(') + N»!
«a, a» = H “ W
On the other hand, we have immediately

(15) 2(90) = l, 50,) = 0 for M 2 1.

Finally, formula (12) implies that, for a EN“),

(16) cc.) = “2:38 a a.

This formula allows us to determine the algebra U’ = Hom(U, K) dual to the
cogebra U (Algebra, Chapter III, § 1!, no. 2). For let K[[X1]]‘er be the
algebra of formal power sefiesinindeterminates (X1), 61 (cf. Algebra, Chapter III,
§ 2, no. 11); if A E U’, letfi denote the formal power series

)1 = Z (A, 29X“, where X“ = 1;; X3“)

and the summation index 0: runs through N“).

Pnoposmorx 10. The mapping A r»+f,_ is an immarphism of the algebra U’ onto the
algzbra affizrmal pawn mic: K[[X,]],EI. '
Because (cu) is a. basis of U, the mapping A Hfi is K-linear and bijective.
On the other hand, for A, p. in U’,

A“ = Z 0;», e.>x“ = Z 0 ® 51,694»)?

=2<A®wfl +v=u 2 Q, o ox“ (by (16))
= ,2, <1, em», ev>X°+v =M,
which shows that A Hfi is an algebra isomorphism and completes the proof.

6. STRUCTURE OF FILTERED BIGEBRAS 1N CHARACI‘ERISTIG 0

In this no. we continue to assume that K is afield afchammrixtic 0.
If E is a bigebra, the canonical injection P(E) —> E can be extended to a
bigebra morphism f3: U(P(E)) —> E (no. 4, Proposition 3).

119
[1 FREE LIE ALGEBRAS

THEOREM 1. Let E be a commutative bigebra.

(a) The bigebm morphirmfi: U(P(E)) —> E is irg'eetz've.
(b) If there existx on E a filtration compatible with its bigebm structure (no. 3,
Definition 2), the morphism f2 if an immorphism.
(In case (b), the bigebra E is therefore identified with the enveloping bigebra
of the Lie algebra of its primitive elements.)
Let eE (resp. em) be the coproduct (resp. counit) of E. We write g = P(E) ;
let (e046, be a basis of the vector K-space 9, where the indexing set I is totally
ordered, and let (“1),s be the basis introduced in the preceding no. We'
write X“ =fE(e¢) for at 6N“). By (15) and (16), we have:

(17) =z(Xo)=l, eg(&)=o for [0491,

(18) ‘I(Xu) = “2
~1=u
X, ‘81 X, for a 6N“),
sincefl; is a cogebra morphism.
We show thatfm i: iry'ecfive. This results from the following lemma:

Lemma 2. Let V be a vector space, E a mgebm andf: S(V) —> E a algebra morphirm.
If the restrirtian Iff to S°(V) + S‘(V) is injective, then f is irgiettive.
Let n 2 0; we write S,I = ‘2. S‘(V) and es for the coproduct of S(V) and
show by induction on n thatf | 5,, is injective. Since the assertion is trivial for
n = Oandn = l,wcassumethatn 2 Zandletuesnbe such thatf(u) = 0.
Then
0 = €n(f(“)) = (f®f7(€s(u))
=f('4) 81 + l ®f(“) + (f®fl(5§("))
= (f®f)(03 (1‘))
As e; (u) E 5,._] ® 5,.” by (11) the induction hypothesis shows that u is a
primitive element of S(V), hence is of degree I (no. 5, Remark 4) and hence
is zero, sincef ] S‘(V) is injective.
It follows in particular that the family (X‘,) isfree.
We .rlww that fE i: .mrjeatz'oe if E has a filtration compatible with its bigebra
structure. Let (E030 be such a filtration and write E: = E,| fl Ker(:z).
We show by induction on n that E; is contained in the image offs. As
E = K.1 + “go E}, this will imply the surjectivity offE. The assertion is
trivial for n = O and follows from the Corollary to Proposition 6 of no. 3 for
n = 1 ; suppose henceforth that n 2 2 and let x e E; . By Proposition 6 of no. 3,
"-1

cm) e 2 E: o EL.
l20
 mum“ or FILTERED BIGEBRAS m cmmmsuc 0 § 1.6

and by the induction hypothesis there exist scalars I“, 5, where a, 3 are in NW,
which are zero except for a finite number, such that

(19) um) = not!”

2 x...X.®x...
Hence by formula. (18) ‘-

(’;®Ide)(€;("))= Z u....xa®x5®x~
u. D: ’1 ¢ 0

(m. e cm; (x)) = “Z“, A.. MX. 69 X. e x.

By Proposition 3 ofno. l and the linear independence of the X, it follows that

(20) 7mm = kn" for a; B; Y inN‘” -{0}~

On the other hand, the coproduct c, is commutative; the same argument as
above implies
(21) 7.4.5 = A,“ for at, BinNfl’ — {0}.
Suppose that there exist: afamily qfscalar: (14‘) with Inc] > 2, such that
(22) (.L.+,=)\,J fora,flinNm—{0}.
Then
40:) = 2 Mn ex“ = m2, mag)
by formula (18), hence y = x — ”22 p..,X., is primitive and hence belongs to
P(E) C Im(f3). Hence

x = y + "E, Mac.) 5 1mm).

The proof will therefore be complete when we have proved the following
lemma:
Lemma 3. [f a family nfmzlar: (Ac-,3) lffim’l: support (with a, B in N“) — {0})
satisfies relation; (20) and (21), there existr afamily (91),“); qffiru't: supper! such
that pm.” = A" “fir a, B nan-um.
It suffices to prove that
(23) at + {5 = Y + 8
implies A“. a = A“, for a, B, 7, 8 non-zero. By Riesz’s Decomposition Lemma.
(Algebra, Chapter VI, § 1, no. 10, Theorem 1) there exist 1:, p, a, r in N“) such
that
u=1r+a, fl=p+'r, y=1r+p, 8=<r+'r.

121
11 FREE LIE ALGEBRAS

Suppose 1: ye O; as a + [3 = p + 8, relation (20) implies

NJ = 7mm,“ = Amn+u = A«.n-Hs = Mum = My
Ifontheotherhandn=0,rhen|3=y + rand8= a: + r,whence
)‘uJ = Au,v+t = )‘a+1.Y = 16.1
by (20), but also 15,, = A“, by (2]),whence A“, = 11...

§2. FREE LIE ALGEBRAS

l. REVISION OF FREE ALGEBRAS

Let X be a set. Recall the construction of the free magma M(X) constructed
on X (Algebra, Chapter I, § 7, no. I). By induction on the integer n 2 l, we
define the sets X" by writing X1 = X and taking X" to be the sum set of the
sets Xp x XVp where/1 =1, 2, . . ., n — l; ifX is finite, so is each X,“ The
sum set of the family (Xn),,3 1 is denoted by M(X); each of the sets X" (and in
particular X) is identified with a subset of M(X). Let w and w’ be in M(X);
let p and 4 denote the integers such that w e X, and w’ e X” and let u = p + q;
the image of the ordered pair (to, 10’) under the canonical injection of
X, x X,,_, into X,I is denoted by w.w’ and called the product of w and w’.
Every mapping of X into a magma M can be extended in a unique way to a
magma homomorphism of M(X) into M.
Let w be in M(X) 5 the unique integer n such that w E X,l is called the length
ofw and denoted by l(w). Then I(w.w’) = [(w) + 1(w’) for w, w’ in M(X).
The set X is the subset of M(X) consisting of the elements of length 1. Every
element w of length 22 can be written uniquely in the form w = w’.w'.
The algebra of the magma M(X) with coefficients in the ring K is denoted
by Lib(X), or LibK(X) when it is necessary Io indicate the ring K. The set
M(X) is a basis of the K-module Lib(X) and X will therefore be identified
with a subset of Lib(X). If A is an algebra, every mapping of X into A can be
extended uniquely to a homomorphism of Lib(X) into A (Algebra, Chapter 111,
§2, no. 7, Proposition 7).
2. CONSTRUCTION OF THE FREE LE ALGEBRA
DEFINITION l. Thefree Lie algebra over [he set X i: the quotient algebra
L(X) = Lib(X)/a,
where a i: the two-sided ideal qib(X) generated by the element: afbne aftheflrm:
(1) Q(u) = (1.41 firainLib(X),
(2) J(a,b,c) = a.(b.t) + b.(£.a) + c.(a.b)
fer a, b, a in Lib(X).

122
 CONSTRUCTION OF THE FREE LEE ALGEBRA §2.2

Clearly L(X) is a Lie K—lagehra; the composition of two elements u, u of

L(X) will be denoted by [u, 1)]. When it is necessary to indicate the ring K, we
write LK(X) for L(X).
The following proposition justifies the name free Lie algebra given to
L(X).
PROPOSITION 1. Let i]; be the canonical mapping of Lib(X) onto L(X) and (I) the
restriction of q, to X. For every mappingf qf X into a Lie algebra 3, there exists one and
only one homomorphism F: L(X) —> 9 such thatf = F o o.
(a) Existence of F: let h be the homomorphism of Lib(X) into g extending
f (no. 1). For alla in Lib(X), h(Q_(a)) = h(a.u) = [h(a), h(a)] = 0; similarly,
the Jacobi identity satisfied by 9 implies that It(_](a, b, c)) = 0 for a, b, c in
Lib(X). It follows that h(a) = 0, whence there is a homomorphism F ofL(X)
into 9 such that h = F o 4;. By restricting to X, we obtain f = F a o.
(b) Uniquenets of F : let F’: L(X) —> g be a. homomorphism such that
f = F’ o (b. The homomorphisms F 0 ill and F' u q, of Lib(X) into 9 coincide on
X and hence are equal; as i]; is surjective, F = F’.

COROLLARY 1. Thefamily (4700),“ irfree over K in L(X).

Let x1, x2, . . ., x" be distinct elements ofX and A” . . ., 1,, be elements ofK
such that

(3) ow.) +~~+ mm) = 0.

Let g be the commutative Lie algebra with K as underlying module. For
i = 1, 2, . . ., 71, there exists a homomorphism FI of L(X) into 9 such that
F,(¢(x()) = 1 and F,(¢(x)) = 0 for x aé z. (Proposition 1); applying F. to
relation (3), we obtain 1, = 0.
COROLLARY 2. Let a be a Lie algebra. Every extension ofL(X) by a i: inemntial.
Let cl) g—“>L(X) be such an extension (Chapter I, § 1, no. 7). As p. is
surjective, there exists a. mappingfof X into 9 such that ‘l> = p. of. Let F be the
homomorphism of L(X) into 9 such that f = F or!) (Proposition 1). Then
(p. o F) 0 <1) = y. of = d) and Proposition 1 shows that p. o Fis the identity auto-
morphism of L(X). The given extension is therefore inessential (Chapter I,
§ 1, no. 7, Proposition 6 and Definition 6).
As the ring K is non-zero, Corollary 1 to Proposition 1 shows that (la is
injective. Heme the set X can be identified by means of d) with its image in L(X) ; with
this convention, X generates L(X) and every mapping of X into a Lie algebra
9 can be extended to a Lie algebra homomorphism of L(X) into 9.
Remark. When X is empty, M(X) is empty and hence L(X) = {0}. If X con-
sists of a single element x, the submodule K.x of L(X) is a Lie subalgebra of

I23
1] FREE LIE AIJGEBRAS

L(X); as X generates L(X), Corollary 1 to Proposition 1 shows that L(X)

is a free module with basis {at}.

3. PRESENTATIONS OF A LIE ALGEBRA

Let g be a Lie algebra and a = (al)IE1 a family ofelements of 9. Letf" be the
homomorphism of L(I) into g mapping each is I to a‘. The image of]; is the
subalgebra of 9 generated by a; the elements of the kernel offi, are called the _
relatnr: of the family a. The family a is called generating (resp. free, basic) iff,
is surjective (resp. injective, bijective).
Let g be a Lie algebra. A pmmtatian of 9 is an ordered pair (a, 1‘) consisting
of a generating family a = (HoleI and a family r = (1,)“, of rclators of a
generating the ideal of L(I) the kernel offi. We also say that g is presented by
the kmi a related by the relators r, (j ej).
Let I be a set and r = (13),E 5 be a family of elements of the free Lie algebra
L(I); let a, be the ideal of L(I) generated by r. The quotient algebra
L(I, r) = L(I)/a, is called the Lie algebra defined by I and the family of
relators (744),“; we also say that L(I, r) is defined by the presentation (I, r),
or also by (I; (I, = 0),“). When the family r is empty, L(I, r) = L(I).
Let I and r be as above; let El denote the image ofi in L(I, r) , The generating
family E = (5‘)"! and the family of relators r constitute a presentation of
L(I, r). Conversely, if g is a Lie algebra and (a, r), where u = (17.),61, is a
presentation of 9, there exists a unique isomorphism u: L(I, r) —> 9 such that
“(51) = a. for all is I.

4. LIE POLYNOMIALS AND SUBSTITUTIONS

Let I be a set. Let T, denote the canonical image of the element i of l in L(I)
(which is also sometimes denoted by L((Tl),el)) ; the elements of L(I) are called
Lie polynomial: in the indeterminates (Tater
Let g be a Lie algebra. If t = 0.),“ is a family of elements of 9, letf. denote
the homomorphism of L(I) into 9 such that f,(T‘) = t, for i E I (no. 2, Propo-
sition 1). The image underf. of the element P of L(l) is denoted by P((t,). H).
In particular, P((T.),E1) = P; the above element P((t,)l61) is sometimes called
the element of 9 obtained by substituting the t; for the Ti in the Lie polynomial
mm. x).
Let a: g —> 9' be a Lie algebra homomorphism. For every family if = (tom;
of elements ofg and all P E L(I),

(4') «(P((h)s=:)) = P((a(t.)),fl),

for o ”f. maps T. to a(t.) forie I.
Let (Q,),E, be a family ofelt-mcnts ofL(I) and let P 6 LG). By substituting

124-
 FUNL'I‘ORIAL PROPERTIES § 2.5
the Q, for the T, in P, we obtain a Lie polynomial R = P((Q,),EJ) e L(I).
Then
(5) R((ti)isl) = P((QI((ti)lel))leJ)
for every family 1‘ = 0.),“ of elements ofa Lie algebra g, as is seen by operat-
ing by the homomorphism f, on the equation R = P((Q,),EJ) and using
(4).
Let g be a Lie algebra, I a finite set and P e L(I). Suppose that g is a free
K-module. The mapping
13: 9I —> 5
defined by 13%).“) = P((t‘)‘fl) is then palynomiaLT For the set F of mappings
of g‘ into 9 is a Lie algebra with the bracket defined by

(5) [45, Wt) = Wt), M0];

the set F’ of polynomial mappings of 5‘ into g is a Lie subalgebra of F by the
bilinearity of the bracket. Our assertion then follows from the fact that the
mapping P >—> P is a Lie algebra homomorphism and T. = pr‘ 6 F’ for all
i.

5. FUNCTORIAL PROPERTIES
PROPOSITION 2. Let X and Y be two rm. Every mapping a: X—>Y can be extended
uniquely to a Lie algebra hammarphixm L(u): L(X) —>L(Y). For every mapping
v: Y—> Z, L(u o u) = L(v) o L(u).
The existence and uniqueness of L(u) follow from Proposition 1 of no. 2.
The homomorphisms L(v o u) and L(v) o L(u) have the same restriction to X
and hence are equal (Proposition 1).
COROLLARY. If u i: irg'zctz'vz (rap. mgjzth've, bijettiue), w it L(u).
Since the assertion is trivial for X = a, we assume X yé 125 . If u is injective
there exists a mapping 1; of Y into X such that u o u is the identity mapping of
X; by Proposition 2, L(v) o L(u) is the identity of automorphism of L(X) and
hence L(u) is injective. When u is suijective, there exists a mapping w of Y into
X such that u a w is the identity mapping ofY ; then L(u) o L(w) is the identity
mapping of L(Y), which proves that L(u) is suijective.

1’ Recall (Algsbya, Chapter IV, § 5, no. 10) the definition of polynomial mappings
of a free module M into a module N: if q is an integer 20, a mappingf: M —> N is
called harmgarwau: polynamial qf dag”: q if there exists a. multilinear mapping :4 of
M' into N such that
for) = u(x, . . ., x) for all xe M.
A mapping of M into N is called pnlgpmmial if it is a finite sum of homogeneous
polynomial mappings of suitable degrees.

125
 1! FREE LIB AwEBRAS

Let X be a set and S a subset of X. The above corollary shows that the
canonical injection of S into X can be extended to an isomorphism on of L(S)
onto the Lie subalgebm L’(S) ofL(X) generated by S; we shall idmtyy L(S) and
L’(S) by means of on.
Let (Su)mel be a right directedfamily of subsets ofX with union S. The relation
S“ C SB implies L(Su) C L(SB) and hence the family of Lie subalgebras L(Su)
of L(X) is right directed. Therefore 9 = “Lg! L(Su) is a Lie subalgebra of
L(X); then S C 9, whence L(S) C 9, and, as L(S,) C L(S) for all eel,
g C L(S). Hence

<7) Ltgsu) =52.L<su>

for every right directed family (saw of subsets of X.
Applying the above to the family of finite subsets of X, we see that every
element of L(X) is of the form P(x1, . . ., x") where P is a Lie polynomial in n
indeterminates and x1, . . A, x" are elements of X.
PkoposmON 3. Let K’ be a rum-zero commutative ring and n: K —> K’ a ring homo-
morphism. For every set X there exists one and tmly one Lie K’-algebra hammarphixm
u: LAX) 03 K' —> LAX)
such that 1/(x ® 1) = x for x e X. Furlher, v is an isomorphism.
Applying Proposition 1 to g = LK«(X) considered as a Lie K-algebra and the
mapping x :—> x of X into 9, we obtain a K—homomorphism LK(X) —) Lx. (X),
whence there is a K’-homomorphism u: LK(X) ® K' —> LKV(X). The fact that
v is unique and is an isomorphism follows from the flact that the ordered pair
(LK(X) (8 K’, x ~—> x ® 1) is a solution of the same universal mapping problem
as the ordered pair (LK.(X), x »—> x).
Remark. Let b' be a Lie K’-algebra and I) the Lie K-algehra derived from b’ by
restricting the ring of scalars. If P e LK(X), we can define l3: bx -—> I; (no. 4).
We see immediately that
P = (0(P (81))".
6. GRADUATIONS
Let A be a commutative monoid, written additively. Let (bu denote a mapping
of X into A and (b the homomorphism of the free magma M(X) into A which
extends $0. For all 8 E A, let Lib°(X) be the submodule of Lib(X) with basis
the subset ¢“(8) of M(X). The family (Lib“(X))56A is a graduation of the
algebra Lib(X), that is
(8) Lib(X) = Q Lib"(X)
(9) Lib°(X) .LibV(X) c Lib““’(X) for s, s' in A
126
 GRADUATIONS § 2.6

(Algebra, Chapter III, §3, no. 1, Example 3).

Lemma 1. The ideal a qtfim'lion l is graded.
For a, b in Lib(X), let B(a, I7) = 11.11 + b.a. The formulae

(10) ma, b) = Q.(a + b) - Q(a) — cab)

(11) tm + + A. w.) = 2mm.) + “ZI 1111300“ w!)
for w1,.., w" in M(X) and 11,“, A“ in K, show that the families
(Q(a))ashlh(x) mid (Q.(w): BO”: 111 ))w m anon generate the same submodulc 0f
Lib (X). As] 15 trilinea: the ideal a is generated by the homogeneous elements
Q(w), B(w, w’) and J(w, w’, w"), where w, w’, w" are in M(X), and hence 13
graded (Algebra, Chapter III, §3, no. 3, Proposition 1).
Let the Lie algebra L(X)= Lib(X)/a be given the quotient graduation.
The homogeneous component ofL(X) ofdegree 8 1s denoted by L‘5 (X); it is the
submodule of L(X) generated by the images of the elements w e M(X) such
that ¢(w) = 3.
We shall make special use of the following two cases:
(a) Total graduation: we take A = N and (1)0(x) = l for all I: e X, whence
q)(w) = [(10) for w in M(X). The K-module L"(X) is generated by the images
of the elements of length n in M(X), which we shall call alternant: of degree n.
We shall see later that the module L"(X) is free and admits a basis consisting
of alternants of degree n (no. [1, Theorem 1). Then L((X)= Q L"(X
L1(X) admits X as basis (no 2, Corollary 1 to Proposition l)> 1By the) con-
struction of M(X),

(12) L”(X> = ,2 MX), L~-v(X)1

and in particular
(13) [L"(X),L"(X)1 c ”M(X).
(b) Multigraduation: we take A to be the free commutative monoid N‘x’ con-
structed on X. The mapping (b0 of X into A is defined by (¢o(x))(x’) = 8“»,
where 8,”. is the Kronecker symbol. For m e M(X) and x EX, the integer
(¢(w)) (x) is “the number of occurrences of the letter z in w”. For ac in N‘”, we
write |oc| = a(x), whence |<I>(w)] = l((w) for all win M(X). It follows that
as!

(14) L"<X)= .51?" L (X);

obviously
(15) mm. mxn c L~+'<X> for ., s in NW.
127
II FREE LEE ADGEBRAS

PROPOSITION 4. Let S be a Mb!!! q/X. IfN‘s) is identified with it: canonical image in
NO“ (Algebra, Chapter I, § 7, no. 7), then L(S) = “Hm L“(X). Further, for all
u 2 NS), the homogeneous component of degree a under the multigmdualian an L(S) is
equal ta L“(X).
Let a e N‘fi’. The module L“(S) is generated by the images in L(X) of the
elements 10 in M(S) such that (Mm) = a, that is (Algebra, § 7, no. 9, formulae
(23) and (24)) the set ofw in M(X) such that ¢(w) = 0:. Hence L°‘(S) = L“(X).
The proposition follows from this and the relation L(S) = L‘(S).
ash“)

COROLLARY. For everyfamily (8.)”x nfsubm: {If X,

(16) L(fl
(u
fl L(S‘).
s.) = (a!
This follows from Proposition 4 and the obvious formula
(17) N“) = I0
Gl
we
where we have written S = 'Q S‘.

7. LOWER CENTRAL SERIES

PROPOSITION 5. Let g be a Lie algebra and P a .mbmodule 91‘ 9. We define the .rub-
module: P,I of g by [heformulae P1 = P and PM1 = [P, Pu] fir n 2 1. Then
(13) [Pm Pu] C PM,»
15—1

(19) P" = ’21 (1),, P,_,,] fern 2 2.

We prove (18) by induction on m. The case in = I is obvious. By the Jacobi
identity,
[[P: PM], Pu] C [Pm KP, Pd] + [1” [Pm PH],
that is
[Pun-l: Pn] C [Pun Em] + [1” [PW PnH'
The induction hypothesis implies that [Pm PHI] C Pun-u and
[PW Pu] c PM, whence
[Pu-+1: Pu] C Pup-n+1 + in Palm] = Pawn-r
Il—l

By formula (18), P" D ”2; [P,, P”_,] 3 [P1, Pnd] = P," whence (19).

When we take P = g, the sequence (Pu) is the lawer central series (Wm) of 9
(Chapter I, § 1, no, 5). Hence:

128
 DERNA'HONS or me LIE mum §2.8

P12209110“ 6. Iztgbealieulgebmmd(?'g)_ntlulamermbalmi¢xq.

[@599] C (E‘HQ form> landl>

Definition 1 ofChapter I, §4, no.1, we shallsny thataLie

algebmgism'lpotmliff'g = (0}{hrnsuflicientlylarge'l‘heni1paknqdasofa
nilpotent Lie algebra g is the smallest integer n such that @“g = {0}.
PROPOSITION 7. Itbeaxdmdnminteger 21.
(a) L'“(X)= [13(34): L"(79]-
(b) The module L"(X) 1: generated by the damn: [x], [xhn ., [x‘_,, XI] . . .]]
when (3b.. ,x,,) mlhmughthemQ/"uqmaofnzmk a.
(c) The lawn central sent: s(X) “given by «“(LOQ) = ’2' L’(X).
(a) We apply Proposition 5 with g = L(X) and P = L‘(X). By induction
on n, we deduce from (12) (no. 6) and (19) the equality P. = L'(X). The
desired relatinn is then equivalent to the definition [P, P_] = PHI.
(b) This follows from (a) by induction on n.
(c) Letg=L(X) ands. =,;L,(X).'Ihmg=glandformula (13) of
no. 6 implia [9,, 9.] C 9“. and in particular [9, 5.] C 9",. By induction
on n, «“9 C 9‘. On the other hand, Rom (a) we deduce that L'(X) C «‘9 by
induction on In. As ’5'}; is an ideal ofg, the relation L’(X) C “implies that
L'“(X) = [1:‘(X),L'(X)] c m
by (a). Hence L’(X) C 9'5 for; 2 n, whence 9. C V'g.
COROLLARY. La! 9 be a Lie algebra and (Ame, a gmeiah'ngfmn'ly q. The Hit
mqfflubwercmlralsnmq‘girlhemfidegmaudbylhtflnahdhm-kn:
["Ip [‘mu -: [19,. p ‘51-- -W> ": W11.” '1: "I I
Letb the homomorphism ofL(I) into a such thatf(i) = I, fix all x e I.
AS (KO-ex 8611t 9, u- f(M1)), Whmoe “"9 f(91110)» 5} l’WPOSi'i‘m
4 of Chapter I, § 1, no. 5. The corollary then followsfiom asenions (b) and (c)
of Pmposition 7.

I. DERIVATIONS OF FREE IE mm
PROPosn'KONS. LetXbeam,MMbtmL(X)-mMe¢miI:ldb¢amppfigqf
XintoM. Warhkmaudmlymlinmmapptq(X)intoMmaflfig
dandmtigiringthenlafion:
(20) D([a, a']) = a.D(a') — a'.D(a) fa, a, u' in L(X).
We define a lie algebra g with underlying module M x L(X) by means of
the bracket
(21) [(M. a), (MC I1')] = (“'1' - B'Jn, [1.111).
ll FREE LIE ALGEBRAS

for a, a' in L(X) and m, m' in M (Chapter I, § 1, no. 8), Letfbe the homomor-
phism of L(X) into 9 such that f(x) = (d (x), x) for all x in X; let
f(a) = (D(a), u(u)) for all a in L(X). By formula (21), u is a homomorphism of
L(X) into itself; as u(x) = 1 for x in X, u(a) = a for all a in L(X), whence

(22) f(a) = (13(0), 0)-

By (21) and (22), relation (20) then follows from f([a, a']) = [f(a),f(a’)].
Conversely, let D’ be a mapping of L(X) into M satisfying relation (20’)
analogous to (20) and extending d. Let f’(a) = (D'(a), a) for aEL(X); by
(20') and (21), f’ is a. homomorphism of L(X) into g, coinciding withf on X,
whencef' =fand D’ = D.
COROLLARY. Every mapping qf X into L(X) can be extended uniquely ta a derivation
af L(X).
When M is equal to L(X) with the adjoint representation, relation (20)
means that D is a derivation.

9. ELIlVflNA’l'lON THEOREM
PROPOSITION 9. Let S1 and S, be [we disjoint sets and d a mapping of S1 x S2 into
L(S,). Let g be the quotient Lie algebra 9f L(S1 U Si) by the ideal generated by the
elements [51, :5] — d (3,, 5,) with :1 6 SJ, :2 E 8,; let 4/ be the canonical mapping qf
L(51 U 5,) onto 9.
(a) For i = l, 2, the restriction (In. 13/ q; to S, can be extended to an isomorphism
of L(Si) onto a subalgebra a. of g.
(b) g = a1 + a2,a1 ha, = {0}andu2isanidealafg.
For i = 1, 2, let 4;, denote the homomorphism of L(Si) into 5 which extends
(b, and 45 its image. Clearly ems.) generates «I.
Let :1 e 8,; we write D = ad (1)1(51). The derivation D ofg maps (1)2(82) into
an by the relation
Hid-‘1): ti>2(-'2)] = ‘i’nWUn 52)) {01” 52 5 52;
as the subalgebra a, of g is generated by 452(52), therefore D012) C n,. The set
of x e 9 such that ad 2: leaves «12 invariant is a Lie subalgebra. of g which con-
tains 471(51) by the above and hence also a, Hence
(23) [an “2] c “2‘
Therefore a1 + a3 is a Lie subalgebra of g and, as it contains the generating
53': ¢1(Si) U $262);
(24) a, + a2 = 9.
For all :1 E S there exists a derivation D,1 of L(Sg) such that

13:1(52) = 4(51, :2)

130
 ELMNATION mom § 2.9

for all :1 in S2 (no. 8, Corollary to Proposition 8) The mapping :1 1—» D.l can
be extended to a homomorphism D ofL(Sl) into the Lie algebra of der1vations
ofL(S,) Let b be the semi-direct product ofL(S ) by L(Sa) corresponding to D
(tter I, § 1, no. 8). As a module 9 is equal to L(Sl) x L(Sz) and in par-
men

(25) [(31, 0), (01%)] = (0,d(!1a¢2))

for 1'1 E 51 and :5 E 8;.
From (25) we deduce the existence ofa homomorphismfof9 into [3 such that
f(‘l’1(‘1))=(‘1: 0) andf(<[>2(:3)) = (0) 52) for ‘1 5 S1 Md 32 5 Sr We deduce
immediately the relation

(26) f(411011) + 412012)) = (”1: 42)

for a; e L(S,) and 112 e L(Sn).
Relation (26) shows that 4;, and 4;, are injective and that a, n a; = {0}.
Formulae (23) and (24) then imply the proposition.
PROPOSITION 10 (elimination theorem); Let X be a rel, S a subset if X and T
the set afsequmee: (11,1. .,' .1," :1) with 11> 0, $1,. ., 3,, in S under an X — 5.1
(a) The module L(X) is the direct mm of the rubalgebra L(S) q/ L(X) and the
ideal a q(X) generated by X — S.
(b) There exist: a Lie algebra isomorphic": 4) q/ L(T) onto a which maps
(:1, . . ., r,“ x) ta (ads1 o . - - «7 adrn)(x).
Let g be the Lie algebra constructed as in Proposition 9 given
51 = 5, s2 = T, .1(:, 1) = (s, s.,...,, :,., 3) er c L(T)
for t = (:1, . . ., 3”, x) in T and s 6 51. We identify L(S) and L(T) with their
canonical images in 9 (Proposition 9 (a)).
Let 4) be the mapping (:1, . 1 1, r", x) 1—» (ad :1 o . - 1 0 ad r")(x) of T into
L(X). Obviously M40, 1)) = [5, 41(1)] for :65 and IET and hence there
exists a homomorphism on: g—>L(X) whose restriction to S is the identity
and whose restriction to T is 1.11. Now X — S C T, whence there is a homo-
morphism fl: L(X) —> 9 whose restriction to X = S U (X — S) is the identity.
We show that a is an isomorphism and p the inve1se isomorphism1 As
M11) = x for x in X — S, we see that a. o B coincides with the identity on X,
whence a 0‘3 = Idwo. On the other hand, [5,!] = d(:, t) in 9 for :ES,
16 T by construction; it follows that t = (:1, . . ., r", x) is equal in g to
(ad :10 - - 0 ad r3)(x) whence t = 501(1)). As 1301(5)) = 3 forse S and S U T
generates g, 9 0 at = Id.
As at is an isomorphism of 9 onto L(X), Proposition 9 shows that the restric-

T For n = 0, we obtain the elements of X— S, whence X— S C T.

131
I! FREE LIE ALGEBRAS

tion of m to L(T) is an isomorphism 4) of L(T) onto an ideal I) of L(X) such

that the module L(X) is the direct sum of L(S) and b. Obviously
M11, - - -’ 5m x) = (ad 51 ° ' ' ‘ ° “15.000
for (:1, . . ., s,” x) in T.
Hence (MT) C (1, whereas I) C «1 since MT) generates the subalgebra b of
L(X). But I: is an ideal and X — S C (NT) C b, whence a C b.
COROLLARY. Let y e X. The free Lie algebra L(X) is the dim! mm 0f the free
.mbmadule K.y and the Lie subalgebm admitting a: baxic fizmilg the family qf
((ady)”.z) where n z 0 and z EX — {g}.
It suflices to put S = {y} in Proposition 10.
10. HALL SETS !N A FREE MAGMA
Let X be a set, M(X) the free magma constructed on X and M"(X), where
71 EN“, the set of elements of M(X) of length n (no. 1). If w E M(X) and
l (w) 2 2, let «(w) and [3(w) denote the elements of M(X) determined by
the relation w = a(w)fl(w); then l(u(w)) < [(w), l(B(w)) < [(w). Finally,
for u, 1) in M(X), let u“); denote the element defined by induction on the
integer m 2 0 by u°v = v and u’"“v = u(u"'v).
DEFINITION 2. A Hall set relative to X i: any Wally ordered subset H qf M(X)
satisfi/ing thefollowing conditions:
(A) IfueH, ”EH andl(u) < [(11), thenu < a.
(B) X C H and H n M’(X) consult 0f the pradwt: xy with x, y in X and
at <3].
(C) An element w q(X) oflength 25 belong: to H ifand only {fit is qfthe
forma(bc) witha, h,cin H, beeH,b < a < btmldb < c.
PRorosn‘mN 11. There exist: a Hall .tel relative to X.
We shall construct by induction on the integer It sets l-I,I C M"(X) and a.
total ordering on these sets:
(a) We write HI = X and give it a total ordering.
(b) The set H2 consists of the products xy with x, y in X and x < y. We
give it a total ordering.
(c) Let n 2 3 be such that the totally ordered sets H,,...,H,,_1 are
already defined. The set Hgd = Hl U - - - U H,”l has a total ordering
which induces the given relations on H1, . . ., HP, and is such that w < w’
if I (w) < l(w’). We define H,. to be the set of products a(bc) e M”(X) with
a, b, c in H;_1 satisfying the relations bteHfi,_l, b < a < be, 17 < c and give
H" a total ordering.
We write H = "Lg: H"; we give H the total ordering defined thus: w < w’
if and only if [(10) < l(u.r’) or [(w) = l(w’) = n and w s w’ in the set H".
It is immediate that H is a Hall set relative to X.

132
 HALL sen [N A FREE MAGMA § 2.10

For every subset S of X we identify the free magma M(S) with its canonical
image in M(X).

PROPOSITION 12. Let H be a Hall rel relative to X and let x, y be in X.

(a) H n Max» = {x}.
(b) Supfiore that x < y and let d, be the honmmorphirm qf M(X) intv N ruch that
d,(y) = l 11711111,,(2) = Ofor z EX, 1 yé y. The rat ofelements w EH n M({x,y})
such that 11,,(w) = l emrirt: qf the elmnts fly with 71 an integer 20.
By Definition 2 (B), e and HnM“({x}) = 25. If wEH hM({x}),
where n = l(w) 2 3, the elements m(w) and (3(w) also belong to H n M({x})
by Definition 2 (C). It immediately follows by induction on n that
H n M”((x}) = 21 for n 2 2, whence (a).
We now prove (b). By Definition 2 (B), yEH and tags H. We show by
induction on n that fly 6 H for u an integer 22. Now x"y = x(x(x"'2y)) and
the induction hypothesis implies that x”"yeH. Now [(14) < l(x""y) for
n > 2 and x < y, whence x < 1""y in any case; condition (C) of Definition
2 shows that My 5 H. On the other hand, certainly d,(x"y) = 1. Conversely,
let w EH h M((x,y}), with dy(w) = l. Ifl(w) :1, then 10 =]/; iIl(w) = 2,
then to = 1:]; by Definition 2 (B). If I (w) 2 3, then w = a(bc), with a, b, t,
I): in H n M((x,y}) (Definition 2 (0)). dv(bt) = 0 is impossible, since this
would imply bee M((x}), which is impossible by (a). Hence dubs) = l and
dy(a) = 0, whence a = x by (a). It follows immediately by induction on
n = l(w) that w = x""y, which completes the proof of (b).

COROLLARY. IfCard X 2 2, then H n M”(X) ;é [afar every integer n 2 l.

PROPOSITION 13. Let X be a finite set with at Iezut two elements. Let H deflate a
Hall ret relative to X. Then there exist a strictly Wearing bijectim p» to, if N
unto H and a sequence (P,),eN ofruhretr af H with thefallawing propertiet:
(a) P0 = X.
(b) For every integer p 2 0, w, e P,.
(c) For every integer n 2 I, there exists an integer [:(n) ruth that every element qf
P, 1': qflength > nfor all!) 2p(n).
(d) Far every integer]: 2 0, the re! P,“ with“: qfthe elements qftheform wpw,
wherei2 0, weP, andwgéwp.
As X is finite, each of the sets M"(X) is finite. Let H" = H n M"(X) for
all It 2 l. The Corollary to Proposition 12 shows that the finite set H,I is
non-empty. Let u" be the cardinal of H"; let v‘, = O and 0,, = 141 + - - - + u"
for n 2 1. As H" is a totally ordered finite set, there exists a strictly increasing
bijection p »—> w, of the interval (”n—1) v" — l] of N onto H... It is immediate
that p ~—> w, is a strictly increasing bijection of N onto H.
Let I“D = X and for every integer p 2 1 let P, be the set of elements w of
H such that w 2 w,, and, either w e X, or a:(w) < 10,, (note that if w is of

133
ll FREE LIE AIIGEBRAS

length 22 the relation w EH implies a(w) e H by condition (C) of Defini-

tion 2). Then w, E P1,; this is clear if w, EX and follows from the inequality
l(uc(w,)) < l(w,) and condition (A) of Definition 2 when 11/, ¢ X.
Hence conditions (a) and (b) are satisfied.
Let n be an integer 2 l and let p, > 17,. For all w e P,” l (w) l(w,) > n by
the very definition of the mapping p »—> w, This establishes (c).
We now show that every element of the form u = wLw with 12 0, w E P,
and w as w, belongs to Pn+1 If 1 9e 0, then [(14) > l(wp), whence u > top
and u 2 10,“; then u¢X and 1(a) = w, < 20,“, whence ue P1,”. If
1': 0, then ueP, and u aé 10,; then u > w,, whence it 2 w,+1; if a does
not belong to X, then «(10) < 10,, whence at(w) < 10p“; then again 14 5 lg”.
Conversely, let u 6 Pp“. We distinguish two cases:
(at) There exists no element 1/ of M(X) such that u = 10,». By definition of
Pg“, 14 > w,. Further, if u e X, then u(u) aé w, by the given hypothesis and
«(14) < w,+1 since u eh“; hence «(11) < w,. Hence :4 e P, and u yé w,.
(6) There exists a in M(X) such that u = 10,11. By Definition 2, of necessity,
either w, EX, 0 EX and w, < v, or 1/ ex and u(11) $ w, < 11. In either case,
7’ 5 Pan-
Then there exist an integer 1'20 and an element w of M(X) such that
u— —- (0,111, and either wax or wéXand «(10) 9e wp. Ifi = 0, we have case
(at) above, whence w e P, and w sé w,. Ifi > 0, the proof of ([3) above estab-
lishes, by induction on i, the relations w e P,” and. w aé w,. Suppose w ¢ X;
from w e Pn+1 it follows that «(10) < w, and as a(w) ;é 10,, we conclude that
w E P,. This completes the proof of (d).
Example. Suppose that X has two elemenm x, y; let X be ordered such that
x < y. The construction given in the proof of Proposition 11 gives a set H
with 14- elements of length <5 given in the following table:
H1 "’1 = x 102 = 1/
H2 "’3 = (”y)
Ha W. = Mtg» "’6 = W»
H1 1% = “(Wt/D) = (1104110)) we = WWW)»
H5 ws = (x(x(x(xy)))) win — (y(*(x(#y)))) wu = (y(y(x(*y))))
W12 = (y(y(y(xy)))) “'1: = (WWWD) w“ = (“NEW”)-
(The demerits of H have been numbered according to the total ordering
chosen on each H.

11. HALL BASES OF A FREE LIE ALGEBRA

We preserve the notation of the preceding no.
THEOREM 1. Let H be a Hall :2! nlaliw la X and ‘l’ the canonical mapping a(X)
inta thefm Lie algebra L(X). The restriction (ff ‘1’ to H is a ham if the module
L(X)-
134
 HALL BASES or A FREE LIE ALGEERA § 2.11

For every element 10 of H we write 27) = ‘P(w).

(A) The my: when X isfinite.
If X is empty, so are M(X) and therefore H and L(X) is zero. If X has a
single element x, H n M"(X) is empty for n 2 2 (Proposition 12 (3.)). There-
fore, H = {x}; we know also (no. 2, Remark) that the module L(X) is free
and has basis {:3}. The theorem is therefore true when X has at most one
element.
Suppose henceforth that X has at least two elements; choose sequences
(11),) and (P,) with the properties stated in Proposition 13. For every integer
p 2 0, let L, denote the submodule of L(X) generated by the elements LT);
with 0 < i < p and g, the Lie subalgebra of L(X) generated by the family
('2)uEP,'
Lemma 2. Far wary integer p 2 0, the module L, admit: the family (II/0““, a:
basis, the Lie algebra g, admit; (12%“, a: basic family and the modul: L(X) is the
direct mm 9/ L, and 9,.
Ln = {0} and Q“ = L(X) and the lemma is true for [I = 0. We argue by
induction on [1. Suppose then that the lemma is true for some integer p 2 0.
Let um, = (ad 111,)‘ldr = ‘l’(w;w) fori 2 0, w 6?,” to ye w,. By the Corollary
to Proposition 10 of no. 9, the free Lie algebra g, is the direct sum of the
module T, of basis {11),} and a Lie subalgebra 9,, admitting

y = (“1.w)l)0.wEP,,w¢wp

as basic family. By Proposition 13 (d), the family (12)“1,” H is equal to 9" and
hence is a basic family of I), = g“). Hence L(X) = L, @T, Q gfil and,
35 Lp+l = L1: + T1» L(X) = Lp+1 @Spn and(u7fi,i1n..., Li’s—1, £71,) is a
basis of the module L1,”.
Let n be a positive integer. By Proposition 13 (0) there exists an integer
pot) such that Pp has only elements of length >n for p 2 [7(a) For [7 2 [7(n),
the Lie subalgebra g, of L(X) is generated by elements ofdegree > n and hence
L"(X) n g, = (0}. On the other hand, the elements a”), of L(X) are homo-
geneous and the family (10,)“,(1, is a basis of a supplementary module of 9,.
It follows immediately that the family of elements Lia of degree n is a basis
of the module L"(X) and that the sequence (12000 is a basis of the module
L(X).
(B) General case.
If S is a subset of X, recall that M(S) is identified with the submagna of
M (X) generated by S and L(S) is identified with the Lie subalgebra. of L(X)
generated by S; we have seen that ifw e M(S) is oflength 2 2 then ot(w) E M(S)
and {5(w) e M(S). It follows immediately that H n M(S) is a Hall set relative
to S.
For every finite subset (I) of H there exists a finite subset S of X such that

135
ll FREE LIE ALGEBRAS

(l) C M(S). Case (A) then shows that the elements :71 with w e 1D are linearly
independent in L(S) and hence in L(X). Therefore the family ("7)“): is free.
For every element a of L(X) there exists a finite subset S of X such that
a E L(S). By case (A), the subset ‘I’(H n M(S)) of‘l’(H) generates the module
L(S) and hence a is a linear combination of elements of ‘I’(H). Hence ‘I’(H)
generates the module L(X), which completes the proof.
COROLLARY. The module L(X) is free and :0 is each of the submadule; L“(X)
where a e No” and L"(X) where n E N. The modules L“(X) are offinite rank and :0
are the module: L"(X) if X isfinite.
There exists a Hall set H relative to X (Proposition 11). For all to E H, the
element ‘l’(w) of L(X) belongs to one of the modules L“(X) (with a 51“”)
and the module L(X) is the direct sum of the submodules L“(X). Further,
for all at ENG”, the set of elements of M(X) whose canonical image in N“)
is equal to a is finite; this shows that each of the modules L“(X) is free and
of finite rank and that L(X) is free. Now L“(X) = M“ L“(X) and hence
L“(X) is free; when X is finite, the set of ateN‘x’ such that Ia] = n is finite
and hence L"(X) is then of finite rank.
DEFINITION 3. A Hall basis qf a free Lie algebra L(X) it any basir of L(X) which
is the canonical image of a Hall Jet relative to X.
Remark. Suppose that X consists of two distinct elements 2: and y and let
L‘ '1’ be the submoclule of L(X) the sum of the L“(X) where at ENX and
a(y) = 1. It follows immediately from Theorem I and Proposition 12 of
no. 10 that the elements of (ad x)".y where n is an integer 20 form a basis
of the submodule L‘ "’. It follows that the restriction to L‘ "l of the mapping
ad 1 ix infective.

§3. ENVELOPING ALGEBRA OF THE FREE LIE

ALGEBRA

In this paragraph, A(X) = AK(X) denote: the fiee aubciative algebra Libas(X)
of the set X over the ring K (Algebra, Chapter III, § 2, no. 7, Definition 2). X is
identified with its canonical image in A(X) ; recall that the K—madule A(X) admit:
a: bad: the free manoid M0(X) derived from X; A*(X) deflate: the .mbrnadule of
A(X) generated by the nan-empty wordy.

l. ENVELOPE“? ALGEBRA 0F L(X)

THEOREM 1. Let a: L(X) —>A(X) be the unique Lie algebra hamamarphirm ex-
tending tht canonical injection of X into A(X) (§ 2, no. 2, Proposition 1). Let

136
 ENVELOPING ALGEBRA or L(X) § 3.1

a: L(X) —> U(L(X)) he the canonical mapping of L(X) into it melaping algebra
and let p: U(L(X)) —>A(X) be the unique unital algebra hamomnrphixm mh that
[i 0 cr = at (Chapter I, §2, no. 1, Proposition 1). Then:
(a) at i: injective and a(L(X)) i: a direttfactar .mhmmtuk 11fA(X).
(b) {3 is biieetive.
Let B be a uniml K—algebra and (b a mapping of X into B; by Proposition
1 of § 2, no. 2, there exists a Lie algebra homomorphism Q}: L(X) —> B such
that L1] | X = (b; by Proposition 1 of Chapter 1, § 2, no. I, there exists a unital
algebra homomorphism 6: U(L(X)) ——>B such that 0 o a = q; and hence
such that (0 o a) [X = 4). As 6(X) generates the unital algebra U(L(X)),
the homomorphism 0 is the unique unital algebra homomorphism satisfying
the latter condition. This shows that the ordered pair (U(L(X)), o | X) is
a solution of the same universal mapping problem as A(X) ; taking (b to be
the canonical injection of X into A(X), we deduce that (5 is an isomorphism,
which proves (b).
Finally, as L(X) is a free K—module (§ 2, no. 11, Corollary to Theorem I),
u- is injective and U(L(X)) is a direct factor submodule of U(L(X)) (Chapter I,
§ 2, no. 7, Corollary 3 to Theorem 1). By (b), this proves (a).
COROLLARY 1. There exists on the algebra A(X) a unique taprodutt making A(X)
into a higebm mph that the elements {If X are primitive. Further, fl is an isomorphirm
of the higebra U(L(X)) unto A(X) with thir bigebra rtmtture.
This follows from assertion (b) of the theorem and the fact that X generates
the unital algebra A(X).
Henceforth A(X) is given this bigebra structure and L(X) is identified with
its image under at, that is with the Lie mbalgebra of A(X) generated by X.
COROLLARY 2. If K is afield ofcharatleristie 0, L(X) is the Lie algebra afprimitiue
element: q(X).
This follows fiom Corollary 1 and the Corollary to Proposition 9 of § 1l
no. 5.
Remarks. (1) Let K' be a commutative ring containing K. [I A(X), L(X)
and Lxr(X) are identified with subsets of AK»(X), we deduce from part (a)
of Theorem 1 the relation
(1) L(X) = LK.(X) n A(X).
(2) Corollary 2 to Theorem 1 remains valid if it only assumes that the
additive group of the ring K is torsion-free. For suppose first that K = Z;
every primitive element of A(X) is a primitive element of AQ(X) and hence
is in LQ(X) fl A(X) = L(X) (Corollary 2 and formula (1)). In the general
case, K is flat over Z and we apply Remark 2 of § 1, no. 2 and Proposition 3 of
§ 2, no. 5.

137
II FREE LIE ALGEBRAS

(3) Let A be a commutative monoid, d)“ a mapping of X into A and

¢:Mo(X) —>A the homomorphism of the associated monoid; if A(X) is
given the graduation (A°(X)),,EA defined in Algebra, Chapter III, §3, no. 1,
Example 3 and L(X) the graduation (L‘5(X))MA defined in §2, no. 6, we have
immediately, for 86A, L5(X) C L(X) nA°(X). As L is the sum of the
L”(X) for 8 E A, and the sum of the L(X) n A°(X) for 8 e A is direct, this
implies
(2) L°(X) = L(X) n A°(X).
(4) Let A be a unital associative algebra and t = (t. .61 a family ofelements
of A. We have a diagram

LI)—f‘—>A
: 1:

A(I)
where i is the canonical injection,f, is the Lie algebra homomorphism defined
by t and g, is the unital algebra homomorphism such that g,(i) = t, for i e I.
The diagram is commutative for g, a i andf, coincide on I. It follows that if
P EL(I), the element P((t,),“) defined in §2, no. 4, coincides with the ele-
ment Palms!) defined in Algtbra, Chapter III, §2, no. 8, Example 2.

2. PROJECTOR OF A" (X) ONTO L(X)

Let 1: be the linear mapping of A” (X) into L(X) defined by
(3) ' "(x1 - - J») = (31105:) ° °ad(~n-1))(xn)
{cm > 0,x,,.. .,x,,inX.
PROPOSITION l. (a) 771: mln'clion no of 1!: to L(X) i: a derivation qf L(X).
(b) Fur every inkgn n 2 1 and all u in L”(X), 1t(u) = 11.11.
(a) Let E be the endomorphism algebra of the module L(X) and e the
homomorphism of A(X) into B such that 0(x) = adx for all xeX. The
restriction of 0 to L(X) is a Lie algebra homomorphism of L(X) into E,
which coincides on X with the adjoint representation of L(X), whence
(4-) 0(u) .v = [u, v] for u, v in L(X).
Let a be in A(X) and b in A+(X); then
(5) 1r(a.b) = 0(a) . u(b).
It sufiices to consider the case a = x, .. .x,” b = x9“. ”x,” with p > 0,
q 2 l and x” . . ”at," in X; but then (5) follows immediately from (3)
since 9(x) = ad 7: for x E X.

138
 PROJECTOR or A" (X) ONTO L(X) §3.2

Let u and v be in L(X); by (4-) and (5),

«0([u, 0]) = «(up — m) = 0(a).1r(u) — 0(1)).1r(u)
= [14, "(PH - ['6 1:(14)] = [11, MM] + PM"), 0L
hence 1:0 is a derivation of L(X).
(b) Let 1:, be the endomorphism of the module L(X) which coincides on
L"(X) with multiplication by the integer n 2 l. The formula
[L"(X), L”‘(X)] C L"+’"(X) shows that 1:; is a derivation (Algebra, Chapter
III, § 10, no. 3, Example 6). The derivation 1:1 -— no of L(X) is zero on X
and, as X generates L(X), no = 1:1, whence (b).
COROLLARY. Supper: that K i: a Q—algzbm. Let P be the linear mapping zy’A*(X)
into itselfmh that

(6) P(:r1 . . .x,) = % (ad :1 a . . . o ad x,._l)(x.,)

for n 2landx1, . . ., 2,, in X. 77m: P i: aprojmor afA*(X) anta L(X).

The image of P is contained in L(X). Further, for all n 2 l and all u in
L"(X), P(u) = £1411), whence P(u) = u by Proposition 1. As

L(X) = ’21 L"(X),

we see that the restriction of P to L(X) is the identity.
Remark. Suppose that K is a field of characteristic zen) and let Qbe the pro-
jector of A(X) = U(L(X)) onto L(X) associated with the canonical gradua-
tion of U(L(X)), cf. § 1, no. 5. For a: eN‘x’, Q(A“‘(X)) C L“(X). For it
suffices to verify that the image and the kernel of Q are graded submodules
ofA(X) with the graduation of type N”). This is obvious for the image, which
is equal to L(X). On the other hand, let u be an integer 21. The vector sub-
space of A(X) generated by the y", where y eL(X), is equal to the vector
subspace ofA(X) generated by the a“). yang“, . . . yam, where yl, ya, . . ., y"
are homogeneous elements of L(X) ; then this subspace is a graded sub-
module of A(X).
(Note that, if Card (X) 2 2, the projectors P and Qda not coincide on A* (X).
For let x, y be in X with x aé y and write

2 = *[JW] + {my}: = *2! - W”-

Then Q(z) = 0 and P(z) = §[x, [1811]] $5 0, cf. §2, no. 10, Example and
no. 11, Theorem 1.)

139
11 FREE LIE ALGEBRAS

3. DIMENSION OF THE HOMOGENEOUS COMPONENTS OF L(X)

Let X be a set, a: an element of NO” and d an integer >0. We write 11] u.

if there exists {5 ENG" such that a = dB. The element 13, which is unique,
is then denoted by a/d.
Lemma 1. Let n I]: an inkgzr >0, T1,. . ., Tfl indeterminate: and uh . . ., un :12-
mmt: af Z. Let (c(u)),eN~_(o, be afamily qfelemenlx of Z such that

(7) 1 _ ”21.41". = £10 (1 _ mm.

For all a e N" — {0},
(8) 5(a): ml?” ”(47) M "am

when 11. i: the M61151“filmlitm (Appendlx).

Formula (7) is equivalent, on taking logarithms on both sides (Algebra,
Chapter IV, §6, no. 9) to:

(9) log (1 — "2‘ WT.) = goda) log (1 — T“).

Now
I 1 'I !

—log(l _ (Zlu‘T‘) =1>17(£=1u“T)

00> =m3.1 firm

l3

l_L
=m>a l—B-lg
—L!u"T“.

0n the other hand

—§01(a)1og(1 — 1‘93“”);m PM?“

(11) 1
[3
=>In| gm]: (5'),”
Hence (7) is equivalent to

(12) “25 EI: 4%) =%u3 forallpeN"—{0).

Let A be the set of (1,, 12.. ., 1,.) EN" — {0} such that the g.c. d. of
A“ A2,. ., )1” 15 equal to 1. Every element ofN" - {0} can be written uniquely

14-0
 DIMENSION or me HouooEN'Eous couponem or L(X) §3.3

in the form ml, where m is an integer >1 and AeA. Condition (12) is
is equivalent to

(13) 2 ”1—1 ml (m)!

4?): (n! 11"" forallAeAandallmzl.
RIM I:

By the Mfibius inversion formula (Appendix), condition (13) is equivalent to

"'_7‘ u
(14) Imam) = “2... W) (éyfi
d .
forallkeAandallm) l.

THEOREM 2. Let X be a finite .ret mi 11 = Card(X).

(a) For every integer r 2 l, the K~module L'(X) irfree of rank

(15) co) =12uwn'“,

r an
where y. is the Mb‘biuxfimcfion.
(b) For all an e N" — {0}, the K-madule L"(X) (§ 2, no. 6) irfiea of rank

(16) em = Th ; Md) “(3%

We already know that the modules L“(X), where r E N, and L“(X), where
u ENX, are free (§ 2, no. 11, Corollary to Theorem 1). Consider the multi-
graduation (A°‘(X))“Nx of A(X) defined by the canonical homomorphism d:
of Mo(X) into Nx (Algebra, Chapter III, § 3, no. 1, Example 3); then
A“(X) n L(X) = L‘(X) by Remark 3 of no. I. For aeNx, the K-module
A“(X) admits as basis the set of words in which each letter a: of X appears
«(:0 times. Let d(a) be the number of these words, that is the rank of A”(X);
we shall calculate in two different ways the formal power series
P((T:e)xex) E z[[(Tx)xsX]]
defined by

(17) P(T) = a; mm.

(1) We have m m

Pm = 2.x W” = Z. Z Tan - - . 1,, = .2 (2. n)’

whence

<18) Pm = (1 — 219‘“
I41
I! FREE LIE ALGEBRAS

(2) For all a ENx — (0), let (0,7,)lsygcm be a basis of L“(X) and give the
set I of ordered pairs (agj) such that or E Nx — (0) and l <j $ 5(1):) 3 total
ordering. By Theorem 1 of no. 1 and the Poincaré—Birkhoff—Witt Theorem
(Chapter I, § 2, no. 7, Corollary 3 to Theorem 1), the elements

y... = (all: (’“»’)M(“'D’

where the index m runs through N“), form a basis of A(X). Each ym is of
multidegree (~%51m(u,j)u. Let u(m) denote this multidegree. It follows '
that
PiT) = 2 Tuna) = 2 H Tmmdn
menu) meum (a. 1):!

= (mnerrZo = («EL (1 _ TV)”,

whence finally
_ - (a)
(19) HT) «em—w) (l T“) c ’
Comparing (18) and (19), we obtain
(20) 1 — x; T,‘ = ML“) (1 _ '1“) cm .
Lemma 1 then gives (b).
If we now substitute the same indeterminate U for the T, for xeX in
formula (20), we obtain
1 _ nU __ ”11(0) (1 _ U I a I )m) __ 1—1 (1 _ mm).
Applying Lemma 1 again, we deduce (a).
Examplu. We have '
[(1) = fl: 5(2) = 10'“ - fl), 0(3) = 1("3 - n),
44) = W — n=), «5) = W — n). 2(6) = W — na — na + n).
Remark. Let X be a set and let a GNU”; the rank of the free K-module LEl (X)
is also given by formula (l6). This follows immediately from Theorem 2 (b)
and Proposition 4 of § 2, no. 6.

§4. CENTRAL FILTRATIONS

1. REAL FILTRATIONS
DEFrNrnON 1. Le! G be a group. A realfiltmiian on G {.1- a family (Gm as“ 0fmb-
group: 11f G such that
(1) (3,, = ”C10, fmallaER.
I42
 ORDER mNcnON § 4.2

Formula (1) implies Ga C G‘3 for (3 < a: and hence the family (Cu) is de-
Hearing. The filtration (GE) is called :efiamted if 0 G.I reduces to the identity
element and is called exhawtive ifG = g G“.
Remark. Let (G,l),,Ez be a decreasing sequence of subgroups of G. It is a
decreasing filtration in the sense ofCummumiue Algebra, Chapter III, § 2, no. 1,
Definition 1. For every integer n and all or in the interval )n — l, n] of R, we
write H“ = G”, in particular H,I = G”. It is immediate that we thus obtain
a real filtration (Hagan on G; such a filtration will be called an integral
filtration. Hence decreasing filtrations in the sense of Commutative Algebra,
Chapter III, § 2, can be identified with integral filtrations.
Let A be an algebra; a real filtration (Au) on the additive group of A is
called compatible with the algebra structure KAWAB C A“, for a, B in R and
K.Afl c A“ for an ER. If the filtration is exhaustive, (A0,) is a fundamental
system of neighbourhoods of 0 under the topology on A which is compatible
with the algebra structure. Let B be a unital algebra; a real filtration (Ba) on
the additive group of B is called compatible with the uniral algebra structure if
it is compatible with the algebra structure and 1 6 Ba.

2. ORDER FUNCTION
Let G be a group with identity element e. Let (Ga) be a. real filtration on G.
For all z in G let I, denote the set ofreal numbersa such that x E G,. Ifot e I,
and B < a, then (-1 E I, and hence 1,, is an interval (General Topalagy, Chapter
IV, §2, no. 4-, Proposition 1). Using relation (1), we see that I,, contains is
least upper bound when this is finite. Therefore Ix is ofthe form ] — co, v(x)] ran
with 0(x) e R; we have 0(a) = sup{u | x e G.)
The mapping v of G into R is called the mderfimlicm associated with the real
filtration (Ga) and u(x) is called the order of x. This mapping has the following
properties:
(a) Far I: e G and u E R, the relation: x e G and v(x)2 a: are equivalent
(b) For x, y in G,
(2) 00:”) = 11(x), 11(2) = +00.
(3) "(m > i“("0"): ”(14))-
Further, we have equality m (3) if 11(2) > u(y).
G(c) For all 016R, let G; denote the set Iy‘xeG mh that 0(x) > on. Then
=09“ GB and in particular G: i: a subgroup q.
GConversely, let u be a mapping of G into R satisfying relations (2) and (3).
For all an ER, let Gu be the set ofx e G such that u(x) 2 on. Then (Gm)asR is a
real filtration of G and v is the order function associated with this filtration.

143
11 FREE LIE ALGEBRAS

For the filtration (Ga) to be integral, it is necessary and sufficient that 1; map G
into Z U (+00, —oo}. For it to be exhaustive (resp. separated), it is necessary
and sufficient that 3/1(—oo) = :21 (resp. _v1(+oo) = (2)).
Let A be a K-algebra (resp. unital K-algebra). By the above, the relation
“xeAu ¢u(;r) 2 (l forxeA and ask”
definm a bijection of the set of exhaustive real filtrations (Aafin compatible
with the algebra (resp. unital algebra) structure on A onto the set of mappings .
v: A —> 1—! not taking the value —00 and satisfying axioms (4) to (7) (resp. (4-)
to (8)) below:
(4) ”(x + y) > him“): 001)) (fly in A)
(m m—x)=vu) (xeA)
(6) 00.1) 2 11(1) (A E K, x EA)
(7) ”(‘11) > ”(Z) + ”(y) (m in A)
(8) 0(1) 2 0.
Remark. If ”(x) is not everywhere equal to +00, conditions (7) and (8) imply
0(1) = 0.

3. GRADE!) ALGEBRA ASSOCIATED WITH A FILTERED ALGEBRA

Let G be a mmutatiue group with a real filtration (GQNR. As before we write

m Q=H%
clearly G: is a subgroup of G“. We write gr,(G) = Cut/G;r and

g®=®mm men

The gradsd group associated with the filtered gum]: G is the group g'r(G) with its
natural graduation of type R.
Ramark. When the filtration (G8) is integral, gr¢(G) = {0) for non-integral at
and grn(G) = Gn/G".l for every integer n. The definition of the associated
graded group therefore coincides essentially with that of Commutative Algebra,
Chapter III, §2, no. 3.
Let A be an algebra (resp. unital algebra) and (Anna a real filtration com-
patible with the algebra (resp. unital algebra) structure (no. I). Then
Au'AB C A1141: Act-An + AuA; C ALB;
and the bilinear mapping of A“ x Afl into A“, the restriction of multiplica-
tion on A defines on taking quotients a bilinear mapping
gram x EMA) —>grm(A)-
144
 comm. rum-mom ON A GROUP §4.4

We derive a bilinear mapping of gr(A) x gr(A) into gr(A) which makes it a

graded algebra (resp. unital graded algebra) of type R. If A is an associative
(rap. commutative, resp. Lie) algebra, so is gr(A).

4. CENTRAL FILTRATIONS ON A GROUP

DEFINITION 2. Let G be a group. A real filtration (Ga) on G i: called eentral 1f
G =“L>Jo G"l and the mmmutalar (x,g) = x‘1y‘1xy if an element x uf GE and an
element y nfGfl belong: to Cu”.
In terms of the order function u, the above definition translates into the
relations
(l0) 11(x) > 0, 0((x, 11)) 2 v(x) + 0(y) for all x, g in G.
We deduce that v((x,g)) > 0(x) if 0(a) aé +00; if we write x” = 3/"xy (cf-
Algebm, Chapter I, § 6, no. 2), then x" = x.(x, y), whence

(ll) v(x”) = v(x).

This relation expresses the fact that each of the subgroups G“ of G is normal.
The Ga form a fundamental system of neighbourhoods of e for a tapnlngy com-
patible with the group structure on G (General Topology, Chapter III, § 1 , no. 2,
Example) said to be defined by the filtration (G¢).
In the rest of this no., G denots a group with a central filtration (Ga). For
all at E R, we define the subgroup G; of G by

(12) G: = age: G‘“

In particular G; = G“ = G for a S 0. Recall that if A and B are two sub-
groups of G, (A, B) denotes the subgroup of G generated by the commutators
(a, b) with a e A and b E B. With this notation we have the formulae
(13) (Ga) Ga) c Gnu
(13’) (Gui) G!) c Gun»
(14) (G, G.) c C:
By (14), G: is a normal subgroup oq for all at 6 R and the quotient group
gr,(G) = Gu/G; is commutative. We write gr(G) = g gr¢(G) and give this
group the graduation of type R in which gru(G) consists of elements of degree
at. Then gra(G) = {0} for a: s 0.
PROPOSITION 1. (i) Let a, B be in R. There exists a biadditive mapping

‘l’an: gra(G) X gra(G) -> gra+n(G)

145
II FREE LIE AMERICAS

whim 7MP: (#0; , yGE’ ) onto (x. r)GI+ n- .

(ii) Lat 42 be the biadditiw mapping afgr(G) x gr(G) inlu gr(G) who” restricmn
’0 @349) X ENG) 1'! tins/'0' “my ddlwi' (at, {5). TIM Milling 4> gim ET(G)
a Lie Z-algzbm ”mature.
(i) Recall the identity

(15) (10/, y) = (x, y)"'(x', 1/)

for x, x’, y in G (Algebra, Chapter I, §6, no. 2, formula (4 bis)).
For x 6 Q1 and y e G“, the class modulo GUT” of the element (x, y) of GM;
will be denoted byf(x,y). For a in G¢+5 andx’ in G, a“.a‘" = (a, x’) E GL3;
in particularf(at, 3/) is equal to the class modulo G12, 3 of (x, y)”. Formula (15)
therefore implies

(15) fWCll) =f(x:y)f(x',!l)'

Now (y,x) = bay)“, whence

(17) f(%*) =f("ay)—l'

From (16) and (17) we deduce

(13) f(my) =f(*»y)f(xall').

We have to prove that the mapping f: G“ X G,—> nB(G) defines on
raking quotients a mapping than: me) x gram) —> gram). By (16) and
(18) it suffices to prove thatf(x, y) = 0 ifx e G: org 6 G; , which follows from
13' .
( (ii) As (1:, x) = c, it follows from (17) that <1) is an alternating Z-bilinear
mapping. Hence it remains to prove that, for uegrm(G), uegr9(G) and
w E 31.1 (G):
(19) <l>(",<1>(v, w» + <1>(v,<l>(w, 14)) + <1>(w, 1101,10) = 0.
Let x e G“, g e Ga and z e GY be elements representing respectively u, v and w.
We know that x” and z are two elements ofG, which are congruent modulo G;
and hence x" is a representative of u in G“; as (y, z) is a representative of
(My, w) in Gun: we see that (x", (y, z)) is a representative of (Mu, (My, 10)) in
G“ M1. By cyclic permutation, we see that (y’, (z, 1)) and (2", (x, y)) represent
respectively My, ¢(w, 14)) and «Mm, ({>(u, u)) in Gan”. Relation (19) is then a
consequence of the following identity (Algebra, Chapter I, §6, no. 2, formula
(15)):
(20) (3’10, Z))~(y‘, (I: *))-(z": w» = £-
The Lie algebra gr(G) over Z defined in Proposition 1 is called the grudgd
Lie algebra muddled with thefiltzred group G.

146
 AN EXAMPLE OF A (ENTRAL FILTRATION §4.5

5. AN EXAJMPLE OF A CENTRAL FILTRATION

Let A be a unital associative algebra with a. unitzl algebra filtration (A) such
that An = A ; then A“ is a two—sided ideal of A for all at e R. Let A“ denote the
multiplicative group of invertible elements of A. For all a > 0, let I“ denote
the set ofxEA“ such that x — 15A,; we write I‘ = “L!“ I“ and I“, = I‘ for
{5 s 0.
PROPOSITION 2. The set I‘ i: a subgroup of A‘ and (1“) i: a central filtruh'm on
I‘.
r = “go I“ by construction and the relation I; = Q“ 1",, follows from
A" = new A“'
We show that I‘m is a subgroup ofA‘. Now 1 e I“; let x, y be in F“, whence
z — l EAwy — leArAsAflisatwo-sidedideal ofA, theformulae
(21) xy—l=(x—1)<y—1)+(x—1)+<y—I),
(22) x" — l = —x"(x — 1),
imply xy — leA‘x and x" — 16A,, whencexye F“ and x“ e 1‘“.
As 1" = ago I‘., this is a subgroup ofA“.
Finallyletai > 0, l‘) > 0,e‘,andyeI‘B. Leta: — l = Kandy — l = n.
Then
(23) (w) — 1 = arty-1(a) — oil;
by hypothesis, E EA“ and 1? GAB, whence En — nEeAflfl. As AM“ is a two-
sided ideal of A, (x, y) — l 6 AH“, whence (x, y) 6 Fa”.
Remark. Let a: 2 0,3 9 Oandxe Fwy e 1"”. By formulae (2]), (22) and (23).
(24») x" — l E —(x —- 1) mod. A3“
(25) W —15(* — 1) + (y -1) mOdvAoH-D
(26) (fly) — l E [(x — 1),(y _ 1)] mOd-Auwnnumm-
We prove for example (26). Ifx — 1 = Eandy — l = n, (23) gives:
(ziy)—1—[E,n]=((x“ - 1) + (31" - I) + (W1 -1)(y" -1))[Er 11]-
Now [5, 71] eA..,,, (x‘l — 1) EA“, (31" — l) 5A5, whence we obtain (26).
Let G be a group and p: G —> 1" a homomorphism. For all real 1, we write
G“ = 9-101“). As (1“,) is a central filtration on r, it is immediate that (Ga) is
a central filtration on G.
Pnoposmow 3. (i) For all a e R, the]: exist: a unique group homommphirm
glz gru(G) —> 314A) which maps the dim modulo G: qf an element a e Gu In the
class module A; of p(u) — l.

147
II FREELIE ALGEBRAS

(ii) Let g be tlu group homomorphism of gr(G) into gr(A) what: restriction to
grm(G) is gufm' all a. The mapping g i: an irg'utiue hammphism qie Z-algtbmr.
(i) Let at > 0. By hypothesis, for all a in G“, 9(a) — l 6 Au; let A,I (a) denote
the class of 9(a) — 1 modulo A; 4 As A,“ C A: , relation (25) implies
[11(ab) = p¢(a) + paw). Then 11 EC; if and only if 9(a) — l 6A3; therefore
G; is the kernel of the homomorphism pm of G“ into gru(A). On passing to the
quotient, p, then defines an injective homomorphism gal of gra(G) into gru(A).
For at < 0, grm(G) : (0) and file only choice is gm : 0.
(ii) As g“ is injective for all real :1, g is injective. We show that g is a Lie
algebra homomorphism. As gra(G) = {0} for I: < 0, it suffices to establish the
formula
(27) mafia, 17)) = L0.(a),h(b)l
for a. > 0, [3 > 0, —> e G“ and b 6G5, which follows from (26).

6. INTEGRAL m FILTRATIONS
Recall (no. 1, Remark) that a filtration (G9!) on the group G is called integral
if Gm = G,‘ for every integer n and all on e]n — l, a]. To be given an integral
central filtration on a group G is equivalent to being given a sequence (0%)”)1
of subgroups of G satisfying the conditions
(i) G1 = G
(ii) G,,DGM1 foralIn>l
(iii) (Gm,G,,) C Gm" form 2 l andn 2 1.
For every integer n 2 l, G, is a normal subgroup of G and the quotient
grn(G) = Gn/Gnu is commutative. On taking quotients, the mapping
(x,y) »—>(x,y) = x'ly'lxy of G", x G,. into Gm” allows us to define on
gr(G) = Q grn(G) a graded Lie algebra structure of type N over the ring Z.
Recall (Algebra, Chapter I, §6, no. 3, Definition 5) that the lower antral
mic: of the group G is defined by
(28) (PC = G, C"+1 = (G, C”G) for a 2 1.
The corresponding filtration is called the lawn amtmlfiltratian of G.
PROPOSITION 4-. (i) The lawn central series of G is an integral central filtration on
G
(ii) If (Gang. is an integral central filtration on G, then C"G C CH for all
n e N*.
Assertion (i) has been proved in Algebra, Chapter I, §6, no. 3, formula
(7)
We prove (ii) by induction on n; C‘G = G = Cl; for n > 1,
CG = (G, C"“G) C (G,G,.-1) C G".

148
 MAGNUS ALGEBRAS § 5.1

PROPOSITION 5. Let G be a group and gr(G) the graded Lie Z—algebm axsoeialed with
the lower centralfiltration on G. Then gr(G) it generated by grl(G) = G/(G, G).
Let L be the Lie subalgcbra of gr(G) generated by gr1(G) ; we show that
L D gr"(G) by induction on n, the assertion being trivial for n = 1. Suppose
that n > 1 and L 3 gr,_l(G). As C"G = (G, C““G), the construction ofthe
Lie algebra law on gr(G) shows immediately that
gr..(G) = mt(G),grn~1(G)] C L-
The above proof shows that the lower central series of the Lie algebra
gr(G) (§ 2, no. 7) is given by

(29) Warm» = ”23.40).

Remark. Let k be a ring, a an integer >0 and A the set of lower triangular
matricu with 11 rows and 7: columns and elements in k. For p 2 0, let A, be the
set of(x.,) EAsuch thatx” = Ofori —j < p. ThenA0 = AandApAa C Arm.
Let I‘F = l + A,. Then I‘] is a subgroup of GL(n, k) called the min lower
triangular group of order a over It. By Proposition 2 of no. 5, (I‘,) is an integral
filtration on Pl. A5 l",I = {l}, we see that the group l"1 i: nilpaten! (Algebra,
Chapter I, § 6, no. 3, Definition 6).

§5. MAGNUS ALGEBRAS

In this paragraph, X denotes a rel, F(X) thefree group commuted on X (Algebra,

Chapter I, § 7, no. 5) and A(X) thefree associative algebra comlmted on X with it:
total graduation (A"(X)),.,o (cf. Algebra, Chapter III, §3, no. 1, Example 3). X
is identified with its image: in F(X) and A(X).

l. MAGNUS ALGEBRAS

Let A(X) be the product module 1:; A”(X). We define on A(X) a multipli-
cation by the rule

(I) («.12). = Z m-
where a = (11,.) and b = (b,) are in A(X). We know (Commutative Algebra,
Chapter III, §2, no. 12, Example 1) that A(X) is an associative algebra and
that A(X) is identified with the subalgebra ofA(X) consisting of the sequence
all of whose terms are zero except for a finite number.
A(X) is given the product topology of the discrete topologies on the factors

149
11 FREE LIE ALGEBRAS

A"(X) ; this topology makes A(X) into a complete Hausdorfi‘ topological alge-
bra, when the ring K has the discrete topology, and A(X) is dense in A(X)
Let a = (a,,) e A(X) ; the family (a,,),,)n is summable and a = “2:0 an.
For every integer m 2 0, let AM(X) denote the ideal consisting of the series
a = ”m afl such that a,I eA”(X) for all n 2 m. This sequence of ideals is a
fundamental system of neighbourhoods ofO in A(X) and an integral filtration
on A(X). The order function associated with the above filtration is denoted by .
a); then 0(0) = +00 and m(a) = m il‘a = “Z“ an with a,eA"(X) for all
n 2 mandam 9e 0 (§4, nos. 1 and2).
A(X) is called the Magnu: algebra of the set X with coefficients in K. If there
is any ambiguity over K we write AK(X).
PROPOSITION I. Let B be a unilal associative algebra with a realfiltration (3,)“3 such
that B is Hausdozfl and complete (§ 4, nos. 1 and 2). Letf be a mapping uf X into B
such that there exist: A > Of»! whichf(X) C BX. Then f can be extended in one and
anly one way to a continuous unital homomorphismfofA(X) into B.
Letf’ be the unique unital algebra homomorphism ofA(X) into B extending
f (Algebra, Chapter 111, § 2, no. 7, Proposition 7). We show that f’ is continuous:
f’(A"(X)) C BM whence f'(A,,(X) n A(X)) C B”. Therefore f' can be
extended in one and only one way by continuity to a homomorphism
f: A(X) —> B.
We preserve the hypotheses and notation of Proposition 1 and let u E A(X).
The elementf(u) is denoted by u((f(x)) fix) and called the result ofsubstituting
thef(:r)far the z in a. In particular, u((x),,ex) = u. Now let u = (11,)e be a
family of elements ofA1(X) and let a E A(Y). The above allows us to define the
element a((u,,),,ey) e A(X) It is denoted by u o u. As u,((f(x))) E BA, the
elements uy((f(x))) can be substituted for the y in v. The mappings
v~—> (v ou)((f(x))) and a>—>v((uy((f(x))))) are then two continuous homo-
morphisms of unital algebras of A(X) into B taking the same value u,,((f(x)))
at the element 3/ 6 Y. Therefore (Proposition 1)
(2) (0 ° u)((f("))) = ”((MUO‘D)»
for all u E A(Y).

2. MAGNUS GROUP
For all a = (110,.” in A(X), the element ao of K will be called the constant Ienn
of a and denoted by 5(a). Formula (I) shows that z is an algebra homomorphism
ofA(X) into K.
Lemma 1. For an element a 13/A(X) (0 be invertible, it is necessary and mficienl that its
emtant term be invertible in K.

150
 MAGNUS GROUP AND FREE GROUP §5.3

If a is invertible in A(X), e(a) is invertible in K. Conversely, if 2(a) is

invertible in K, there exists u 6A1(X) such that a = a(a)(l — u); we write
h = (2:0 u")a(a)"1. Then ab = ba = l and u is invertible.

The set of elements ofA(X) of constant term 1 is therefore a subgroup of the

multiplicative monoid A(X), called the Magnu: gruup constructed on X (rela-
tive to K). In this chapter it will be denoted by F(X) or simply I‘. For every
integer 71> 1, we denote by 1" the set of a6 I‘| such that m(a — l) > n. By
Proposition 2 of § 4, no. 5, the sequence (F001 is an integral centralfiltration an
I‘.

3. MAGNUS GROUP AND FREE GROUP

THEOREM 1. Let r be a mapping If X into A(X) such that m(r(x)) 2 2 for all
x EX. The unique homomorphism g of theflee group F(X) into the Magma: grim;
F(X) such that g(x) = 1 + x + r(x)far all x EX is injectine.
We first prove three lemmas.
Lemma 2. Let n he a non-zero rational integer. In the ring offormal power eerie:
K[[t]] we write (I + t)" = gamt’. There exist: an integerj) 1 such that
cm, aé 0.
If n > 0, then ‘1». n = I by the binomial formula.
Suppose that u < 0 and let in: —n. If 1:, ”=0 for allj> 1, then
(1 + t)" = l, whence, taking the mvexse, (l + t)" = l, which contradicts the
formula em m = 1
Lemma 3. Let 5:1,. ,1, be element: «jX such that .t> >1and x, yen” for
1 < i S .t — 1; let n1,.,n, be yum-zeta rational integers. Then the element
1
£1 (1 + mm ofA(X) is #1.
Let m be a maximal ideal ofK and k the field K/m; letp: AK(X) —> Ak(X)
be the unique continuous homomorphism of unital K-algebras such that
fix) =x for xeX (no. 1, Proposition 1). It suffices to prove that
[:(n (l + 251)“) sé l and the problem is reduced to the case where K is a
field.
In the notation of Lemma 2:
s
110 + x0"! = ”go ”only - Jnun. *i1~-"’i'-
By Lemma 2, there exist integers u‘ > 0 such that cab,” ¢ 0 (1 g i < 5). By
Algebra, Chapter I, § 7, no. 4, Proposition 6, no monomial xi". . .94" such that

151
n ma LIE menus

bI 2 0 and (b1, . . ., h.) 9e (a1, . . ., a.) can be equal tox‘l“. . .4“. It follows that

the coefficient of x‘;k..xf' in EU +4)": is ‘a..u~-‘a..u. 96 0, which

implies the result.
Lemma 4. Let a he the eontinuoux endomorphism q(X) such that a(x) = x + 1(x)
font 9 X (no. 1, Proposition 1). Then a- is an Mmephirm amia(Am(X)) = A",(X)
for all m E N.
6(a) E 2: mod. A,(X) for x e X, whence, for n 2 l and x1, . . ., x" in X,
6(x,). . .o(x,.) E :1. . .x, mod. An+,(X);
it follows by linearity that 0(a) E a modulo A“,(X) for all a e A"(X) and in
particular c(A“(X)) C A,,(X). It follows that o-(A“(X)) C Afl(X) for m 2 n,
whence a(A,,(X)) C AJX). In other words, a is compatible with the filtration
(AM(X)) on A(X) and its restriction to the associated graded ring is the
identity. Hence a is bijective (Commutative Algebra, Chapter III, §2, no. 8,
Corollary 3 to Theorem 1).
Finally we prove Theorem 1. Let to ye 1 be an element of F(X). By Algebra,
Chapter I, § 7, no. 5, Proposition 7, there exist x1, . . ., x, in X and non-zero
rational integers n1, . . .,n, such that: 21, x‘ # x”, (1 g i S 3 —— l) and
' w = :31". . .11“.
In the notation of Lemma 4,

:0») = H (1 + «crow = «(1—1 (1 + arm),

hence g(w) #- 1 by Lemmas 3 and 4.

4. LOWER CENTRAL SERIES OF A FREE GROUP

We shall prove the two following theorems:
THEOREM 2. Sufipose that in the ring K the relation 71.1 = 0 implies n = 0 for every
integer 7:. Let r be a mapping of X into A(X) such that ro(r(x)) 2 2 for x e X and
let g be the homomorphism of F(X) into the Magma group F(X) .tueh that
got) = 1 + x + r(x)fi11 x e X. For all n 2 l, C"F(X) i: the irwem image underg
of the subgroup l + A,(X) 1y“ F(X).
THEOREM 3. For allx E X, [4215(1) be the mnoru'eal image of): in F(X)/(F(X), F(X)).
Let 9 he the graded Lie Z-algehm associated with thefiltration (C"F(X)),,,l of F(X)
(§ 4, no. 6). The unique homomorphism of thefree Lie Z-algehra Lz(X) into 9 which
extend: e i: an isomorphism.
loosely speaking, the graded Lie Z—algebra associated with the free group
F(X) (with the lower central series) is the free Lie Z-algebra Lz(X).

152
 rowan CENTRAL sums or A FREE GROUP §5.4

We write F(X) = F, I‘(X) = I‘, A(X) = A, Az(X) = A1, C"F(X) = C",

I‘,‘ = l + An(X) and let 1111(X) —> g be the homomorphism introduced in
the statement of Theorem 3.
(A) Preliminary reductions.
Let Y denote the homomorphism of F into I‘ defined by 7(x) = l + x for
x E X. By Lemma 4- there exists an automorphism a of the algebra A compatible
with the filtration on A and such that 6(1 + x) = g(x) for all x e X; then
0-(I‘n) = F” for all 71‘ As the homomorphisms g and a a Y of F into 1" coincide on
_I _
X, g = a o y and hence g (I‘n) = -;(I‘,.). Under the hypotheses of Theorem 2,
Z can be identified with a subring of K; the Magnus algebra A: is therefore
identified with a subring of A and the filtration on A, is induced by that on A.
As Y maps F into AZ, we see that it suffices to prove Theorems 2 and 3 under the
supplementary hypotheses K = Z, r = 0 and hence g = 7, hypotheses which we
shall hung/07th make.
(3) Swjectivily of at.
As X generates the group F = C1, the set 0(X) generates the Z-module
91 = Cl/C’. But 9‘ generates the Lie Z—algebra g ('9‘ 4, no. 6, Proposition 5)
and hence c(X) generates g, which proves that at is surjective.
(C) We identify the graded algebra gr(A) with A(X) under the canonical
isomorphisms A"(X) —> An/AMI. For every integer n 2 l, we write
F" =2; (I‘,); we know (§ 4, no. 5) that (F"),,>1 is an integral central filtration
on F. Let 9' denote the associated graded Lie Z—algebra (§ 4, no. 4). Letfbe the
Lie algebra homomorphism of 9’ into A(X) associated with 7 (§ 4-, no. 5,
Proposition 3). New C" C F" for every integer n 2 l (§_4, no. 6, Proposition 4)
and hence there is a canonical homomorphism e of g = "(‘2 0'70"“ into
9' = g Fn/Fn+l

LAX) ——:—> g —‘> 9' —’> A(X).

We write fl = f o e; we give [5 explicitly as follows: if u is the class modulo
0"“ of an element w of C", then y(w) — l is of order an in A and {5(a) is the
homogeneous component of 1(a)) — l of degree n. In particular,
(3) B(c(x)) = z for all x e X.
(D) Praafcy‘" Theorem: 2 and 3.
The Lie algebra homomorphism [3 o a: LZ(X) —> A(X) restricted to X is the
identity by (3) and hence is the canonical injection (§ 3, no. I). Therefore at is
injective and hence bijective by (B) ; this proves Theorem 3. As [5 o a: = f a c o a
is injective and a: is bijective, e is injective. For all n 2 1,
en: C"/C"” —> F"/F’”1
is injective and hence
an n Fun = Cin+1.

153
11 FREE LIE ADGEBRAS

C‘ = F = F1; if C” = F", then C"r\F"+1 = F'”1 whence 0"“ = F"+1

which prove: Theorem 2 by induction on n 2 l.

COROLLARY. “n
)1 C"F(X) = {5}-

Applying Theorem 2 with K = Z and r = 0,

01 Wm = “Q'g‘a + MX» ='g’(fl (1 + MK») = 2(1) = {a}.

n3 3I

Remark. Let H be a Hall set relative to X (§ 2, no. 10). Let M be the magma
defined by the law of composition (x, y) »—> (x, y) = x'ly‘lxg/ on F(X) and let
(I) be the homomorphism of M(X) into M whose restriction to X is the
identity. The elements of ¢(H) are called the basic commutator: of F(X) asso-
ciated with the Hall set H. For every integer n 2 1, let H,| be the subset of H
consisting of the elements of length n; we know (§ 2, no. 11, Theorem 1) that
the canonical mapping «3a into LZ(X) is a basis of the Abelian group L; (X).
Moreover, 42(Hn) C C“; for all m E H,., let 4),.(m) denote the class mod. 0‘”
of (Mm) 6 C". Theorem 3 then shows that (bu i: a bijeetion oa onto a basis of the
Abelian grim]: C"/C"”. It follows immediately that, for all w E F(X) and all
i 2 1, there exists a unique element on; of 1(a) such that, for n 2 l,
n
(4) w = (1:! "1:; ¢(m)"4("" mod. CV“,

where the product is calculated according to the total ordering given on H.

Examfile. Suppose that X is a set with two elements x, y and let H, = (x, y},
Ha = {#51}. Every element 11) of F(X) can therefore be written
w E z‘y”(x,y)‘ mod. C3 with a, b, e in Z.
For w = Qty)", 11 = b = n and e = n(l — n)/2 (cf. Exercise 9), whence

M)" a W"(x,y)”“"‘”’ mod. 05.

5. p-mmTION OF FREE GROUPS
In this no., p denotes a prime number and we assume that K = Pp. Let y be
the homomorphism ofF(X) into F(X) defined by y(x) = l + x for x in X ; we
_ —1
write FYKX) = Y (1 + An(X)). The sequence (F¥)(X))n,1 is an integral
central filtration on F(X), which is xepumled since Y is injective (no. 3, Theorem
I). It is called the p—filtratian on F(X).
PROPOSITION 2. Suppose that X i: finile. For every integer n 2 l, the group
F(X)/F§,”’(X) is ufinilap-gwup qfnilpaleney clan Sn.

154
 EXPON'ENTIAL AND LOGARm-IM IN FILTERED MEBRAS § 6.1

Arguing by induction on n, it suffices to prove that K.”(X)/F;’11(X) is a

finite commutative p—group for all n 2 1. For all weFffKX), the element
-{(w) — l of A(X) is of order 2n; we denote by 8,,(w) the homogeneous com-
ponent of 7(w) — l of degree n. The mapping 8": FEI”(X) 4> A"(X) is a homo-
morphism with kemel Fifi l(X) (§ 4, no. 5, Proposition 3) and hence
Fff’(X) /Ff,’2,1(X) is isomorphic to a subgroup of A"(X). Since X is finite,
A” (X) is a finite—dimensional vector space over F, and hence a finite commuta-
tive p—group and so is Fn")(X)/F§.’11(X).
PROPOSITION 3. Fur all to 95 l in F(X), there exist a finite p—group G and a homo-
morphiqfF(X) into C such ihatf(w) yé 1.
There exist elements 11,. ..,x, of X and integers n1,..i,n, such that
w = x15. . .xZ‘v. Let Y = {xv . . ., x,}. The canonical injection of Y into X
extends to a homomorphism on: F(Y) —> F(X); on the other hand, let {3 be the
homomorphism of F(X) into F(Y) whose restriction to Y is the identity and
which maps X — Y to {I}. Then B(a(y)) = y fory EY and hence Be a: is the
identity automorphism of F(Y). There obviously exists w' in F(Y) such that
w = at(w’); then {3(w) = w’ aé 1; now no) Ff.”(Y) = {l} and there therefore
exists an integer n 2 1 such that B(w) ¢ FE.” (Y) By Proposition 2, the group
G = F(Y) /F;’)(Y) is a finite p-group. Iffis the composition of B and the canoni-
cal homomorphism of F (Y) onto G, thenf(w) a6 l.
COROLLARY. The intersection of the normal subgroup offinite index in F(X) it
(1}-
§6. THE HAUSDORFF SERIES

In this paragraph we assume that K i: afield nfcharactnixtic 0.

l. EXPONENTKAL AND LOGARITI'M m mTERED ALGEBRAS

Let A be a unital associative algebra which is Hausdorff and complete under a
real filtration (Au) We write m = A; = “L!“ A“.
For x e m, the family (JP/710“" is summable. We write

(1) e" = expat = "Elf/rd.

Then exp(:r) e l + m and the mapping exp: m —> 1 + m is called the apartm-
tial mapping ofA.
For all ye] + m, the family ((—l)"“(y — 1)"/")n>1 is summable. We
write
(2) logy = Z(—1>"-*(y — 1w».
155
11 FREE LIE menus

Then log yem and the mapping logzl + m—>m is called the logarithmic
mapping of A.
PROPOSITION l. The exponential mapping ir a homenmarphirm of III min 1 + m and
the logarithmic mapping is the invent: homeomorphism.
For x E A“, :15: EA”. It follows that the series defining the exponential con-
verges uniformly on each of the sets A“ for a > 0; as A“ is open in m and
m = “U0 A“, the exponential mapping is continuous. It can be similarly shown '
that the logarithmic mapping is continuous.
Let 2 and I be the formal power series with no constant term
XE
:(X) = an H' l(X)= n>1
Z(— I)“ ‘X"/n
We know (Algebra, Chapter IV, §G, no 9) that e(l(X)) = l (e(X)) = X on
A({X})= K[[X]]. By substitution (§ 5, no. 1), we deduce that

0001)) = 1(444)) = x
for x e m; as
exp): = :(x) +1, log(l + x) = l(x)
it follows immediately that
logexpx=x, explog(l+x)=l+x
for x in m, whence the proposition.
Remarks. (1) If x e m, y e III and x and y commute, then

aw + y) = expor) exm),
1
since the family (#311! ) is summable (cf. Algebra, Chapter IV, §6, no. 9,
LIEN
Proposition 11)
(2) As the series a and l are without constant term and A“ is a closed ideal
ofA, ep C l + A andlog(l + A“) r: Auwhenceepu = l + A‘and
log(1+ A“): Aviora > 0.
(3) Let B be a complete Hausdorfl' filtered unital associative algebra and
II C U" B. Letf be a continuous unital homomorphism of A into B such that
f(m) C0n. Then f(exp x) = expf(x) for xem and f(log:y)= logf(y) for
y E l + I11; we show for example the first of these formulae

f(mp1)=,,;,f(x”/n'n;f(x)"ln!=expf(x)-
156
 HAUSDORFF GROUP § 6.2

(4-) Let E be a unital associative algebra. If a is a nilpotent element of E,

n
the family (%) has finite support and we write exp a = n; a“/n!. An
‘ ”EN
element b is unipotent if b — l is nilpotent; then we write

10gb = "Z, (~1)"‘1(b — l)“/n.

We deduce from the relations e(l(X)) = l(e(X)) = X that the mapping

a r—> exp 4 is a bijection of the set of nilpotent elements of E onto the set of
unipotent elements of E and that b »-> log I) is the inverse mapping.

1. HAUSDORFF GROUP
Let X be a set. We use the notation of § 5, nos. 1 and 2. The free Lie algebra
L(X) is identified with its canonical image in A(X) (§ 3, no. 1, Theorem I). We
denote by L(X) the closure of L(X) in A(X), that is the set of elements
of A(X) of the form a = 2:1 a,I such that aneL"(X) for all n 2 0; this is
filtered Lie subalgebra “A(X).
THEOREM 1. The mtriction qf Ihe exponential mapping y” A(X) to L(X) is a
bifmian 1y” L(X) arm: a closed subgroup affine Magma" graup I‘(X).
We write A(X) = A, A"(X) = A", A(X) = A, L"(X) = L", L(X) = L,
I‘(X) = I‘. Let B be the algebra A ® A with the graduation of type N defined
by B" = ”Z“ A‘ ® A’. Let B = I); B" be the associated complete filtered
algebra (Commutaliw Algtbra, Chapter III, §2, no. 12, Example 1). The co-
product 0: A—>A ®A defined in §3, no. 1, Corollary 1 to Theorem 1 is
graded of degree 0 and hence extends by continuity to a homomorphism
I: A —> 3 given by
it: an) = Z c(a,.) for a" 6A“.
)0 you

We also define continuous homomorphism 8' and 8" of A into B by

Z (an® 1),
812;) 1,.) = man 8120.1"): 2 (l ®a,,) for unsA".
u) nan
By Corollary 2 to Theorem 1 of § 3, no. 1, L” is the set of a,‘ e A" such that
:01") = 1,, ® I + 1 ® an. It follows that L is the set ofa EA such that
(a) 5(a) = 3'01) + 8'(a).
Let A be the set of 1: EA of constant term equal to l and satisfying the
relation
(4) 5(5) = 8’(b)~8'(b),
157
ll FREE LIE ALGEBRAS

in other words, the set of]; = 20 bn such that b" E A" for all n 2 0, £0 = I and
4b,) = ”/2" b, ® b, for n 2 0. The latter characterization shows that A is a
closed subset of I‘; as i, 8’ and 8" are ring homomorphism and every element of
SKA) commutes with every element of 8"(A), the restrictions to 1" of the
mappings o and 8’8” are group homomorphisms and A is a subgroup of I‘.
By Proposition 1 of no. 1, the exponential mapping ofA is a bijection of the
set A* of elements of A with no constant term onto I‘. Let aEA* and
b = exp a. As 5 is a continuous ring homomorphism,
3(1)) = 5(2):" o"/n!) = ”Z, £(a)"/n! = exp 5(a).
The relations .
8'(l’) = exp 3’01), 3'0) = exp 3'01)
are proved similarly and, as 8’(a) commutes with 8"(a) (no. 1, Remark 1),
8’(b) 8"(17) = exp (8'01) + 8'(a)).
Therefore 4 satisfies (3) ifand only if b satisfia (4) , which proves the theorem.
Remark. The above proof shows that expd.) is the subgroup A of I‘ consisting
of the b satisfying (4).
Hence the group law of A can be transported by the exponential mapping to
L. In other words, L is a complete topological group with the law of com-
position (a, b) >—> a H 1) given by
an I; = log(exp a.exp b).
The topological group thus obtained is called the Hausdorflgroup (derived from
X relative to K).
Let g be the homomorphism' of the free group F = F(X) into I‘ such that
g(x) = expx for xeX. As expx — l — x = unfl/n! is of order 22, g is
injective by Theorem 1 of § 5, no. 3. Therefore the mapping log a g is an injectiue
homomfirphfirn of F into the Hausdorfl" group whieh extend: the canonical injection
X —> .
For every integer 7n 2 l we denote by L," the set of elements of order 2 m in
L and by I‘m the set of u e I‘ such that u — 1 is of order 2m. Then
- l
L," = exp(I‘,,.) by Remark 2 of no. 1; as (1",!)m“ is an integral central
filtration on I‘ (§ 4, no. 5 Proposition 2), (I'm)m>1 it an integral central filtration
on the group L.
3. LIE FORMAL POWER serum
Lemma 1. Let g be a filtered Lie algebra (§ 4, no. 1), (5a)“: it: filtration and let
a: E R. Let P be a homogeneoux Lie polynomial of degree n in the indeterminate: (TI).E‘
(§ 2, no. 4). Then P((a,)) E g,“for everyfamily (Q‘s, Ly'elements of g“.

158
 LIE FORMAL rowan SERIES §6.3

Every Lie polynomial of degree n 2 2 is a finite sum of terms of the form

[Q,R] where Qand R are of degree <n and the sum their degrees is equal to n
(§ 2, no. 7, Proposition 7). The lemma follows by induction on n.
A Lie formal power :erier‘l’ (with coefficients in K) in the indeknninam (T‘ (E;
is any element of the Lie algebra L((T,)‘E,) = L(I). Such an element 11 can be
written uniquely as the sum of a summable family (uv),,enm where 11V 6 L"(I).
Suppose that I is finite. Let g be a complete Hausdorff filtered Lie algebra
such that g = m>0
U gm; let 1‘ = (4).;I be a family Ofelements of g.
PROPOSITION 2. The honwmamhismfi: L(I) —> g ml: thatf‘(Tt) = t, (§ 2, no. 4)
Mn be extended by continuity to one and only one continuous homomorphismfl q(I) inta
g.
There exists a > 0 such that t‘ E g“ for all ie I; hencef.(L"(I)) C llu for
all v (Lemma 1), which implies the continuity Off}.
If u e L(I), we write u((t,)) =fi(u). In particular, taking 9 = LU),
u = u((Ti)); in the general case, u((!.)) is called the result of substituting the t,
for the Ti in the Lie formal power scrim u((T‘)). If u = V ”a, my, where
uv E L"(X), the family (uv((t‘)))venm is summable and

(5) :40.» = 2,
EN mo.»
Let a be a continuous homomorphism of 9 into a complete Hausdorff
filtered Lie algebra 9’ such that 9' = “go 9;. For everyfinite family t = 00‘“
ofelements ofg and all u E La),
(6) «(an)» = «(an)»,
for the homomorphism e of, is continuous and maps T, to 0-0.) for i e I.
Let u = (u,),u be afinile family of elements of L(I) and let ve LG); by
substituting the u, for the T, in v, we obtain an element w = I/((u,),e,) of L(I)
denoted by u o 14. Then
(7) WWW“) = ”(("I((1A)Iel))1ea)
for everyfinite family t = (1‘).“ of elements of g, as is seen by operating with
the continuous homomorphismfi on the equation w = v((u,),e,).
Let u = ‘12“ qLU), where qL"(l). The mapping :7 (li) >—> u((ti)) of V
g‘ into 9 is continuous: for in each of the open sets 9‘, with a. > 0 the family of
fin is uniformly summable and it suffices to prove that each 17,, is continuous,
which is immediate by induction on M.

T A Lie formal power series is not in general a formal power series in the sense
of Algebra, Chapter IV, §6.

159
XI FREE LIE ALGEBRAS

4. THE HAUSDORFF SERIES

Let (U, V) be a set with two elements.
DEFINITION 1. The zlcmmt H = U H V = log(exp U.exp V) (no. 2) If the
Lie algebra L.({U, V}) is called the Hmdmflsnie: in the indelmninates U and V.
We denote by H" (mp. Hunt) the homogeneous component of H of total
degree n (resp.mu1tidegree (1, .r)). Then

(a) H: ZHH= 2°11”, H=+ZIHH

71)0

>1, Ihm
THEOREM? [f 1 and J an two positive integer: .mch that r +s>
H“ = H.’ + H1,” when
(9) (7 + M15.» =
z (—1)“ dU v,
m
l+u>1....,r:.-_‘x:'.:_x>1
((E —‘ n) —‘ .8 Vanni)“
”—1 adU '4 adV n

<10) <r+s)H¢..=
2‘;— I)" ’
m>1
Z
u+~~+r.—1=r-1
(“1:1<adrf)~(adsx>~)w).
Y‘+I];llt:::-Vll-l+’m— [)1

r .
In A.((U,V}), ep.cxpV =1+ w, where w = +2.11%; whence
H = ”‘21 (—1)’"“W"‘/m (no. 2), that is:

(11> H,,.=Z<_-1>““
own In
Z
In+~~ n."
fimfl
In"! A“!
m n. =3
u+s>1."r...+;..>1

The linear mapping Pu, defined by Pn(zl,.. ., x") = -:(fi (ad x0) (x,,) for
2 l and x1, . . ., x” in {U, V}, is a projector ofA°((U, V}) onto L”({U V})
(§3, no. 2, Corollary m Proposition 1); as 1-1,, belongs to w’aufvn,
H” = r+l<Hn.) NOW
U'V'
(12) Pm;( ‘ —‘)
r_! r!

_ 1 ((110-1d (saw) (adU)'—(ad_V>l"‘1)(v)

‘=‘ n! 1.. in! 5,...
 SUBSTITUTIONS IN THE HAUSDORFF SERIES § 6.5

when I," 2 l and

(13) 1%(1—11"
£=1 n! E)
5,! = 7 + 5((1—1—(“111!U)" (—I—adv)")(—adm!
Ur” )(U)
when r,,l 2 l and I". = 0. Moreover, obviously (ad t)"1.t = 0 ifp 2 2 and
(ad t)°.t = t. It follows that the two sides of (12) are zero when :m 2 2 and
those of (13) are zero when 7,, 2 2. The theorem then follows since HL , is the
sum of the terms of type (12) and H1, is the sum of the terms of type (13).
Remarks. (1) We have defined (§ 3, no. 2, Remark) a projector QofA(X) onto
L(X) such that Q(n‘”) = 0 for aeL(X) and n1 2 2 and QC) = 0. Then
H = Q(exp H) = Q(exp U.exp V), whence immediately

(14) H” = (QC—(11;) forr + s 21.

(2) We have
(15) H(U, V) U + V + 5[U, V] + —‘1—,[U [U, V]]
+ .2[V, [V, U1] i‘s: W: [U, V]]]
modulo gs L"({U, V}).
(3) Ha,» = HM, = 0 for every integer n 55 l, whence
(16) H(U, 0) = H(O, U) =
On the other hand, as [U, —-U] = 0,
(17) H(U, —U) = o.
5. SUBSTITUTIONS [N THE HAUSDORl-‘F SERIES
As K is a. field containing Q, the Hausdorff series can be considered as a
Lie formal power series with coefficients in K. Therefore, if g is a complete
Hausdorff filtered Lie algebra with g = “90 9“, then, for a, b in g, a and b
can be substituted for U and V in H (cf. no. 3 and § 2, no. 5, Remmk).
In particular, let A be a complete Hausdorff filtered unital associative
algebra. We write m = “HA“ and mu = A“ n m for 116R; hence m“ = A“ ,
for a. > 0 and mu = m for a: S 0. With the bracket [a, b] = ab — ba, 111 is a
complete Hausdorfl‘ filtered Lie algebra, to which the above can be applied.
With this notation, we have the following result which completes Proposition
1 of no. 1.
Pnoposmou 3. If a e m, b e m, then exp H(a, b) = exp a.exp b.
Leta,bbeinm;thereexistsa > OsuchthataeAuandbeAwThen there

[61
II FREE LIE ALGEBRAS

exists a continuous homomorphism 6 of the Magnus algebra A({U, V}) into

A mapping U to a and V to b (§ 5, no. 1, Proposition 1).
The restriction of 0 to L({U,V}) is a continuous homomorphism of Lie
algebras of L({U, V}) into m which maps U (resp. V) to a (rap. 1;). By formula
(6) of no. 3, therefore 0(H) = H(a, h). It then suffices to apply the con-
tinuous homomorphism 6 to the two sidm of the relation
tarp H(U, V) = exp U.exp V
taking account of Remark 3 of no. 1.
Rnnark 1. Ifa and b commute, then H,_,(a, b) = 0 for r + .r 2 2, for every
homogeneous Lie polynomial of degree 22 is zero at (a, b). Then
H(a, b) = a + b and Proposition 3 recovers the formula
exp(a + b) = exp a.exp b.
Pkorosmon 4. Let g be a complete Ham-dam"filtered Lie algebra such that g = “go 3,.
The mapping (a, b) »—> H(a, b) i: a group law on 9 compatible with the topology an a
under which 0 is the identity element and —a i: the inverse of a for all a E g.
The mapping (a, b) »—>H(a, b) of g x 9 into 9 is continuous (no. 3); as
the mapping a ~—> —u is obviously continuous, it suffices to prove the relations
(13) H(H(a, 5), I) = H“: HUI. 6))
(19) H(a, —a) = 0
(20) H(a, 0) = H(0, a) = a
for a, b, o in 9. By formula (7) of no. 3, it suffices to prove these formulae when
a, b, c are three indeterminate: and g = L({a, b, 0}). Now the restriction of
the exponential mapping to L({a, h, c}) is an injection into the Magnus
algebra A({a, b, 6}) and by Proposition 3:
exp H(H(a, b), o) = exp H(a, b) .exp 0 = exp a.exp b.exp 0
exp H(a, HUI, 5)) = exp a.exp H(l7, c) = exp a.exp l7.exp :
ep(a, —a) = exp a.exp(—a) = exp(a — a) = expO
exp H(a, 0) = exp a.exp0 = exp a
ep(0, a) = exp 0.exp a = exp a.
This establishes relations (18) to (20)‘
Rzmarkr. (2) Take 9 to be the Lie algebra L(X). The group law introduced
in the above proposition coincides with the law defined in no. 2. In other
words,

(21) aHh = H(a,b) for a, bin E(X);

thus the Hausdorff group law is given by the Hausdorff series.

162
 suns’n'ru'nons IN THE HAUSDORFF SEW §5.5

(3) Let g be a Lie algebra with the integral filtration (W9) defined by the
lower central series. Suppose that there exist: m 2 1 such that W9 = {0}.
With the topology derived from the filtration («v-9),“, the Lie algebra g is
Hausdorff, complete and even discrete. Then P(al, . . ., 4,) = 0 for al, . . ., a,
in g and for every homogeneous Lie polynomial P of degree >m; in particular,
H',(a,
, b) = 0 for r + s> m and the series H(a, b): 2 H, .(a, b) has only
a finite number of non-zero terms. The group law (a, b) »—> H(a, b) on g is
then a polynomial mapping (§ 2, no 4-).
PROPOSITION 5. La Km be tlw mmponml qf H(U + V, —U) of mullidagree
(7,5). Then
K..1(U, V) = (—fil—l—f. (ad U)"(V) for» 2 0
We write K(U, V) = H(U + V, -U), K,(U, V) = n; Kn',(U, V). We
denote by L (reap. R) left (resp. right) multiplication by U on A({U, V}).
We can write
Rt' -U = ”Z.—
U’V
p —U ) ‘
_!V (q!

_ 21 (m——R)") V)
— no 11—! mum!

=21! (L— R)“.V

")0"

and therefore
1 .
(22) zUVe'-v = go—,7! (ad U)”V.
We now calculate modulo the ideal ”2:0 '22 A’" "({U, V}) of A({U, V}.) For
all 71>, l,

(U + V)" .=_ U" + g U'VU"'1"

whence
n-l

(3'd mm + V)" E ((L - R) .21 L‘R"‘)-V

a (L" — R") .V
E U"V — VU”.
Therefore
(23) (ad U).ev+v a JV — V.“
summing over n.

163
1! FREE LIE ALGEBRAS

On the other hand, K1(U, V) a K(U, V) and exl‘u‘v’ E l + K1(U, V)

and hence
KIEBK— l Eeu+Vr"—l
by Proposition 3. We deduce that
(ad U)K, s Ue‘HVJ‘" _ ¢U+v‘—UU E (Ugtnv _ ¢U+VU):-U
E (z‘TV — V:")e"’ by (23)
E eUVe'“ —

E 21 "1—, (ad U)”V by (22)

5 (ad U) (“a ($1111); v)-
It then suffices to apply the Remark of § 2, no. 11.

§7. CONVERGENCE OF THE HAUSDORFF SERIES

(REAL OR COMPLEX CASE)

In this paragraph we assume that K is one of the fields R or C with its usual
absolute value. Recall that a normable algebra over K is a (not necessarily
associative) algebra over K with a topology .7” with the following properties:
(1) .7 can be defined by a norm:
(2) the mapping (x, y) >—> xy of A x A into A is continuous
A normed algebra over K is an algebra A over K with a norm such that
nxynWe< enote
M Mbyforg 211mm A.
a complete normable Lie algebra over K. We choose a
norm on 9 and a number M > 0 such that

(l) ||[x,y]l| < Mlll llyll for oyin 9~

l. CONTINUOUS-POLYNOMIALS WITH VALUES IN a
Let I be a finite set and let P(g‘; 9) (resp P(g‘; 9)) be the vector space of
cantinuum—polynomialx (resp. formal pawn Jen's: with continuum components) on 51
with values in 9. Recall (Diflmntiable and Analytic Manifolds, R, Appendix)
that 13'“;I ; g) has a graduation of type N1 and that 13(91; 5) is identified with
the completion of the vector space P(g‘; 9) with the topology defined by the
filtration associated with the graduation of P(g‘; 9). Moreover, P(g‘; 9) is a
graded Lie algebra with the bracket defined by Lf, g](x) = [f(x), g(x)] for

164-
 GROUP GERM DEFINED BY A COMPLETE Homo LIE § 7.2

f, g in P(g‘; 9), x 59‘; this Lie algebra structure can be extended by con-
tinuity to 15(31; 3) and makes it into a complete Hausdorfffiltered Lie algebra.
By proposition 2 of § 6, no. 3, there exists one and only one continuous Lie
algebra homomorphism ([21:11 >—> a of L(I) into f’(9‘; g) mapping the indeter- ‘
minate of index i to pr‘ for all ieI, since prieP(g‘;g). It follows that
fieP(g‘; g) for ueL(I); more precisely, when uEL(I), I? is just the poly-
nomial mapping (a) »—> u((t,)) of §2, no. 4-. On the other hand, clearly 4:, is
compatible with the multigmduations of L(I) and P(g‘; 9). If u = Viz“ u,,,
where u., e L"(I) for v e N’, then
veN
a= veNI
Z wherefivemss)
Let u = (11,)”, be a finite family of elements of L(I), let ueLU) and let
w = v on (§ 6, no. 3). We write i2 = (17,)msfi. Then

(2) iiofi=(vou)".

This follows by extending by continuity formula (7) of §6, no. 3 and Earn
Dg‘femttiable and Analytic Manifolds R, Appendix, no. 6.

2. GROUP GERM DEFINED BY A COMPLETE NORMED LIE ALGEBRA

Let H = 2 HmeL(U, V) be the Hausdorff series (§6, no. 4, Defini-

tion 1). We sliall show that the corresponding formal power said

(3) fi = ”2011 ”em x 9, s)

is mnwrgmt (Dyfirmtiable and Analytic Manifolds, R, 3.1.1).
We introduce the following formal power series 11 e Q[[U, V]]

(4) n(U.V) = —Ios(2 — exp<U + V»

(5) = Z-,l” («pm + V) — 1)»
1 UV: V'x U'a V‘u

(6) =glmp... .Zy.?fifi"‘}:f‘

1.+ai>1
Hence

(7) mu, V) = ,2 7|...U'V'y

[65
1! FREE LIE ALGEHRAS

where

(8) nm= milm;1+::

21 13:71.._————. .1.
rm..|‘1.....r,n. .-
71‘9“}?

Now let u and u be two positive real numbers such that u + v < log 2;
then 0 S sip(u + v) — l < l; the series derived from (5) and (6) by substi-
tuting u for U and 1/ for V are convergent and the above calculations imply
that

(9) 1.1)!)
mJu'v‘ = —log(2 — exp(u + 11)) < +00.
Let r, s 2 0 and let "fin,“ be the norm of the continuous-polynomial Id”
(Dg‘firmtiable and Analytic Mamfllds, R, Appendix, no. 2).
Lemma 1‘ 1|fi,,||<Mm-l ..
Letihs‘be 1aorl i< m,w1th:m— l wewriter=2rbs=254
and consider the following element of L({U, V}):
m— l

= ((221 (ad U)’I(ad V)’l) (ad U)’n) (V).

Then 2 = fo p, where f is the following (7 + :)-linear mapping of 9'” into
9:
(1‘1: - . ') J‘nyn - - cg!) H
(“1(a) ° NEW") MMiG/1) ° ' " °ad(!h,) ° adorn“) ° “ad("r))(!/z)
and where p is the following mapping of 9’ into 9'”:
(my) ._. (mushy. - - -.y);
r .1
hence HZH< || f ||< M'+I 1 (Dg'fermtiablz and Analytic Many’oldx, R, Appen-
dix). Applying these inequalitia to the various terms on the right hand side
of formula (9) of § 6, no 4, we obtain:
(10) IKHL.) ll
MY§S-1 l 1

< r+x ”I; 7

r '+'+- +1 3—1-1 I I v
1 ....r,..sl....:,,,_,.1‘
71+“)lnn -.1--1§l.— ‘)l
A similar axgnment gives
(11) II(H’r'..) II
< MY-O-Sil 2 l— 1

7+; m>1m n+- -_+v.n 1=r_-1 r !...r.,,_1!:l!...:,,,_,l

-+:
”up:..... 1._1+_Im_1)l

166
 GROUP Gm DEFINED BY A COMPLETE NORMED LIE § 7.2

whence, by (8)
- M , 4S - 1

llHull < m. m < ’IvI-M'M’

which proves the lemma.
PROPOSITION l. The fm-mal power :m‘ex H i: a camergmt series (Difermtiable and
Analytic Manifolds, R, 3.1.1) ; it: domain of gbmlute convergence (Difl’enntiable and
Analytic Mufti/01b, R, 3.1.4) contain; the open set

n = {(m) 69 x al M + l yl < $41032}.

I

Letu,vbctworealnumbers >Osuchthatu +0 < filog 2; then (Lemma

1)
(12) M vuhu
2 "www-
s 1“,!)0
2 n,,M'*'u'v' = —log(2 — exp M(u + 0)) < +00
bY(9)
Let h: D —> 9 denote the analyticfltwtion (Diferentiable and Analytic Manifizlir,
R, 3.2.9) defined by H, that is by the formula

(13) h((x,y =20

' Emmy): m2” H,‘.((x,y) for (x, y)e D.
This function is called the Hawdorflfimtion of 9 relative to M (or simply the
Hausdorff function of 9 if no confusion can arise). Note that H,‘.(U, —U) = 0
ifr+3> 23ndhence

(14) h(::, —x) = 0 for ||x|| <—

M1.og2
Similaxly
(15) mg) = mo) = x for "at" < I‘Ldlog 2.
PROPOSITION 2. Let
1
n' = {04.4% x 9 x slllxll+||y||+l|1||<—M1°tzg}
Way. 2) e n: m
(16) (1,!)59, (h(x,y),z)efi, (11,2)69. (x,h(y,z))efl
and

(17) ”(MIJ), Z) = h(X. My. 2))-

I67
u mm In mums

Let (x, y, z) e 9'; clearly (at, y) e I] and (y, z) E Q. Moreover:

llh(x,y)|l < 2, "H..." llxllrllyllx,

and hence by (13)

llh(x,y)|l < “lama - expMlll + llyll))~

Now M(||*l| + M) < 103% — MIIZII; we write u = exp(MIIZII); then
x < u < 3‘ and
mum)" + llzll) < —log<2 — expaozi — Mllzll)) + an
3 211’
= ——log( — 2—") + logu = “SET—3

= log(2 + mix—W) g log 2.

We see similarly that (x, h(y, z)) E (I.

We now prove (17). In the Lie algebra £({U, V, W}),
H(H(U, V), W) = H(U, H(V. W))
by Proposition 4 of § 6, no. 5. By not 1, formula (2), we therefore have in
H9 x g x g, g) the relation
Ho(fi x Id.) = Quad, x H).
By Difemitiabls and Analytic Manfflzldr, R, 3.19, there exists a number s > 0
such that formula (17) is tnie when “x", || y“ and “z“ are <2. But the functions
(at, y, z) r—> h(h(x, y), z) and (x,y, z) >—> h(x, h(y, 1)) are analytic functions on
9' with values in g (Dzflmntiable and Analytic Mung‘cildr, R, 32.7). As 9’ is
connected and they coincide in a neighbourhood of 0, they are equal (Dlfirm-
liable and Analytic Mam'firltir, R, 3.2.5).
The above results imply:
Let a be a real number such that 0 < a: S fi log =11. Let

G = {15%| lll < a),

(E) = {(x,g)EG x G [h(x,y)EG} and m:®—>G be the restriction ofh to
. Then:
(1) G) is open in G X G and m is analytic.
(2) x E G implies (0, x) e G), (x, 0) e G) and m(0, x) = m(x, 0) = x.
(5) 7:50 implies —XEG, (at, —x) E 0, (—x, x) e 0 and
m(x, —x) = m(—~x,x) = 0.

168
 Exronanruu. m COMPLETE nomo ASSOCIATIVE ALGEBRAS §7.3

(4-) Let x, y, z be elements of G such that (x,y) e 0, (m(x,y), z) E 0,

(y, z) e e and (x. mo, 2)) e 9. men mow), 2) = m(~. mo. 2)).
*In other words (Chapter III, §l), if we write -—x = 6(a), the quadmple
(G, 0, a, m) is a Lie group germ om K.,

3. EXPONENTIAL IN COWLETE NORMED ASSOGATIVE ALGEBRAS

In this no. we denote by A a campletc normed unital associative algebra (Gonna!
Topology, Chapter IX, § 3, no. 7). Then “my" S “x" . "y“ for z, y in A.
Let I be a finite set and let 13(A‘; A) be the vector space offlrmal power
mic: with mntinam components on AI with values in A (Dtflemuiable and Analytic
Mani/”(114;J R, Appendix, no. 5) with the algebra structure obtained by writing

f.g = m . (f,g) forf, gin l"(A‘; A),

where m: A x A —> A denotes multiplication on A. Arguing as in no. 1 and
using Proposition 1 of §5, no. 1, we define a continuous homomorphism of
unital algebras 14912 of A(I) into f’(A’;A) mapping the indeterminate of
index i to pn; this homomorphism extends the Lie algebra homomorphism
ofL(I) into 15(A’; A) defined in no. 1. Ifu = 2v: av with uv E A"(I) for v E N‘,

then i = 2": (L, where 12‘, is the polynomial mapping (tom, H u,,((t‘)).
Let u = (11,)“; be a finite family of elements of A(I), let ueAfl) and
write w = you (§ 5, no.1). Then
(18) (uou)”=17°fi.
This follows by extending by continuity formula (2) of §5, no. I and from
Diflmztiable and Analytic Mani/our, R, Appendix, no. 6.
In particular we take I = {U}, identify A and Au" and consider the images
2‘ and i of the series 2(U) = “Z! U‘In! and [(U) = n; (—- l)""U"/n in
15(A; A). Then “13"" S l for "x, . . .xn“ S “a“ . . . “an“ for x1, . . .,x,, in A.
Therefore the radius (fabsolulz convergent: 4J5 (resp. I) i: itgfinitz (resp. 2 1).
We shall denote by Q (resp. IA) the analytic mapping of A into A (resp. of
B into A, where B is the open unit ball of A) defined by the convergent series
5 (resp. i) and we shall write ep(z) = l + :A(x) (for s) and

1084*) = M" — 1)
(forzEA, “x — I" < 1). Then

(19) 1"
exp = "Z”! (xeA)

169
1! FREE LIE ALGEBRAS

(20) (’t — 1)" (xeA, [|x _ 1" <1).

103,1: “EA—1) ._ 1—”—
As (zol)(U) = (lo:)(U) = U (cf. §6, no. 1), by (18) int: i=5: 1d,.
Therefore (Diffimliablc m Analytic Manifolds, R, 3.1.9)
(21) EXPAUOEAQD = ’0 (XEA, ll” “ 1H < 1)
(22) l°gA(ep(x)> = x (x EA; ”I“ < 108 2)
for ”z” < log2 implies ||ep(x) — 1|] S exp "at" — 1 < 1.
Finally we consider A as a complete normed Lie algebra. Then
|l[x.y]|| = llxy —yXIl < 2||Xll-|lyll~
Proposition 1 of no. 2 implies that the domain of absolute convergence of
the formal power series I“; contains the set
9 = {MINA X Al llxll + M < Hos?)-
Hence H defines an analytic function h: 9 —> A. Then h(x, y) = ’2. H,,,(x, y)
(cf. §3, no. 1, Remark 4). '
PROPOSITION 3. For ||x|] + |ly|| < {log 2,
(23) ep x-eXpAy = CXPA Mm)
It follows fiom (l8) and the relation 9"2" = (Hwy) that
17190 +51 +5): (1 +a)ofi
in HA X A; A). We therefore deduce from Dzjfirmliable and Analytic Mani-
folds, R, 3.1.9 that (23) is true for (x, y) sufficiently close to (0, 0), whence the
proposition follows by analytic continuation (Difcmttiable and Analytic Mani-
folds, R, 3.2.5).

§8. CONVERGENCE OF THE HAUSDORFF SERIES

(ULTRAMETRIC CASE)

In this paragraph we assume that K is a non—discrete tamfilcte valucdficld of

characteristic zero, with an ultrametric absolute value. We denote by p the charac-
teristic of the residue field of K (Commutative Algebra, Chapter VI, §3, no. 2).
If [7 ye 0, we write a = |p|; we know (Cummutative Algebra, Chapter VI,
§6, nos. 2 and 3) that 0 < a < l and that there exists one and only one
valuation v on K with values in R whose restriction to Qis the p—adic valua-
tion 12,, and which is such that [x] = 11"“ for all x E K. Also we write:

(I) o = —-
170
 p—Amc upmm neurons or 11-11: sums exp, log AND H § 8.1

Up = 0, we denote by a a real number such that 0 < a < land by v a

valuation on K with values in R such that [x] = aw" for all x e K ([00. oil)
Then 0(x) = 0 for x E Q)“. Also we write:
(2) 0 = 0.

I. p-ADIC UPPER BOUNDS OF THE SERIES exp, log AND H

In this no. we axmme that]; 96 0.

Lemma 1. Let n be an integer 20 and let n=no+n1p+ +n,,p", with

0<m<p—l,bethep—adic¢xpmsioan.LetS(n)=no+n1+ +nk.
Then

(3) _ n — so»
”14"!) — fl
1.
v,(n!) = "21 1/,(i) and the number of integers 1' between 1 and n for which
7),.(5) 2 j is equal to the integral part [n/p’] of n/p’. Then

W) = Egan/m — [n/p’*‘]) = ,2; W1.

2 my”, the lemma follows.
A3 [fl/fi’] = «>1
Lemma 2. 0(71) S v(n!) s (n — l)0 and ”(71) g (Iogn)/(logp) fir every integer
n 2 1.
1101!) = 0,,(71!) = (n — S(n))0 < (n — Do by Lemma 1.
On the other hand, 7: 2 11“”, whence 0(a) S (log 11)](log [1).
Let I = (U, V) be a set of two elements and let

H = ,2 MU. V) e M)
be the Hausdorff series (§ 6, no. 4, Definition 1). Let 20,, be the local ring of
Z relative to the prime ideal (p) and (5,)“, a basis of 1.4,,(1) over Z (§ 2,
no. 11, Theorem 1). It is also a basis oq(I) over Q.

PROPOSITION 1. Let r and .r be two integer: 2 0. If Hm = ”23 he," where N, e Q,

it the decomposition 91‘ H with respect to the basis (4),“, then
(4) 11,00 2 —(r+:— 1)!) farallbeB.
The ring Azmu) is identified with the sub-Lw-module of AQ(I) generated
by the words w E Mo(I). As 1,4”(1) is a direct factor of Alma),

(5) L7<n(I) = Alma) “ 110(1).

171
ll FREE LIE ALGEBRAS

Letf be the integer such thatfé (1 + s — 1)8 <f+ 1. Relation (4.) is

equivalent to v,()‘,,) 2 —f for all heB, that is H” ep"L,m. But this is
equivalent also by (5) to H", Ep"Azm(I).
By formula (11) of §6, no. 4, it suffices to show that, for every integer
m 2 landallintegers y,,...,r,,,sl,...,.r,,.such that
(6) 71+-~+rm=n :1+~~~+:...=:,
r.+s.>1 forl <i<m,
we have
(7) u,(m.r!,..rml.rx!...:m!) <f-
But by Lemma 2, 11,,(nls‘!) S (r. + s, — l)6 and v,(m) g 11,,(m!) S (m — US;
the left hand side of (7) is therefore bounded above by

6(m— l +‘2;(r.+s.— l)) =0(1+:—1);

as it is an integer, it is S]; which completes the proof.

2. NORMED LIE ALGEBRAS

DEFINITION 1. A "Md Lie algebra aver K i: a Lie algebra with a norm such that
(3) ll’‘ + yll s SuP(||"“: Hg")
(9) “PM" < llxll-llyll
fir all x, y in 9.
Throughout the rest of this paragraph, 9 denotes a eomplete normed Lie algebra.
For every finite set I we define as in §7, no. 1 a continuous Lie algebra
homomorphism u >—> ii of L(I) into 13(9‘; 9). We see as in § 7 that ifu = 2": av,
with uv e L"(I) for v e N‘, then it = 2 iv, where 12‘, is the polynomial mapping
(t,),E,I—>u,,((t‘)) defined in §2, no. 4, The composition formula (2) of § 7,
no. 1, remains valid.

3. GROUP DEFINED BY A COMPLETE NORMED LIE ALGERIA

Let H = 2:0 Hme L({U, V)) be the Hausdorff series (§ 6, no. 4-, Definition
1). We shall show that the corresponding formal power series with continuous
components

(10) H = ns>0
2 11461361 x 9,9)
is convergent (Difl'emm'ahle and Analytic Manifuldx, R, 4.1.1).

172
 GROUP DEFINED w A COMPLETE NORMED LIE ADGBBRA §8.3

Let 1 2 0, s 2 0 be such that r + .t 9e 0 and let "Hm“ be the norm ofthe

continuous polynomial H” (Difl'nenliable and Analytic Mmfiltir, R, Appendix,
no. 2)
Lemma 3. “H..." s a-"N-m-
Let B be a Hall set relative to I and let H,_, = 2:3 M, be the decomposi-
tion of Hm with respect to the corresponding basis of L({U, V}). Then
(11) ll‘bl < a—(1+x—1)6.
This is trivial for [I = 0, for ADEQ; and it follows from Proposition l of
no. 1 for p 96 0.
Moreover,
(l2) "5.“ S l for b EB.
We show more generally by induction on n that, for every altemant b ofdegree
n in the two indetenninatm U and V (§ 2, no. 6), “1;" < l. Ifn = 1,5is one
of the projections of g x 9 onto 5 and hence is of norm <1; if n > 1, there
exist two altemants b, and be of degrees <n such that b = [b], 172]. As the
mapping y: (x,y) >—> [my] of 5 X 9 into 9 is bilineax of norm <1, we have
(Difi’mntiable and Analytic Mani/Mix, R, Appendix, no. 4)

(l3) “5" = llY°(l;1)52)ll < ||51||.|152|| S 1-

Relations (11) and (12) imply the lemma.
PROPOSITION 2. The formal power mic: H i: a convergent series (Diflmntiable
and Analytic Manifolds, R, 4.l.l). If G i: the ball {at e g,| "x" < 11°}, til: domain
of abxalut: convergence of H (Difemuiable and Analytic Manifolds, R, 4.1.3) can-
taim G x G.
If u and v are two real numbers >0 such that u < a9 and v < a“, then
(Lemma 3)
(H) "Hall!” < 4°(M‘°)'(W'°)'
and |]H,',|[u'v' tends to 0 when r + .t tends to infinity.
We denote by II: G x G —> g the analytic finclion (Dlflmnliuble and Analytic
Mamfilds, R, 4.2.4) defined by H, that is by the formula

(15) how) = ,_,Z,.,H...(x,y> = ,,_2,,H...(~,y) for (w) 56 x G.

This function is called the Hmdarflftmclion of 9.
Let (my) EG x G. Then
(16) "114069)” S s“MIMI, M)
(17) llh(x,y)ll < supfllxll, Ill“)-
173
n mum LIE. AmEmus

(17) follows immediately from (16) and (16) is trivial for r = r = 0;if1 2 1,
then _
"17...“, y) [I S "Hull IIXII'Ilyll‘
X 7-1 I
g mg) (%)
< 1"";
we argue similarly if: > 1.
In particular, ]|h(x,y)|| < a“ for (my) 5G x G.
Pnoposmon 3. Let G be the ball {at e g] ”x” < a”). The analytic mapping
h: G x G —> G
makes G into a group in which 0 is the identity element and —x i: the inverse afxfw
alIzEG. Moreover, ifR i: a realnumhermh ”1010 < R < a”, the ball

{reel M < R}
(reap. {x 6 9| ||x]| S R}) i: an 0pm subgroup QfG.
As H(U, —U) = o and H(0, U) = H(U, 0) = U, h(x, —x) = o and
h(0, x) = h(x, 0) = x
for all x e G. It therefore remains to prove the associativity formula
(13) h<h<x,y),z) = h(x,h(y,2)) forx,y,zin G.
As
H(H(U,V),W) = H(U, H(V, W))
in L((U, V, W)) (§ 6, no. 5, Proposition 4), we have
(19) FIo(H >< Ida) =I.-Io(Id, x H)
in 15(9 x g x g; 9) (no. 2) and (19) implies (18) by (16) and Dgflerentiable
and Analytic Manyoldr, R, 4_1.5_
*In other words (Chapter III, §l), G with the Hausdorff function is a
Lie group.,,.

1. EXPONENTIAL IN COMPLETE NORMED ASSOCIATIVE ALGEBRAS

In this no. A will denote a unital associative algebra with a norm :0» "x"
satisfying the conditions:
llx + yll < suP(||’¢|l) “1“)
"W“ < llxII-Ilyll
lllll = 1
174
 EXPONBNTIAL m oomum: Homo Axsocmnvz menus § 8.4

for x, y in A, and ample” with this norm. The results of the second and third
paragraphs of § 7, no. 3, remain valid.
We take I = {U} and consider the images 2 and iof the series 2(U)= :— U
101 71—!
andl(U)= Z (—1)"‘1U—in P(A- A). Then

(20) (3%)" g a-(n-lw

mu
(21) (U7)~ S a'mp

by Lemma 2 of no. 1. Hence the radius if absolute cmergnwe of the series 5

(resp. i) is 211° (rap. 2]) (Difl'emitiahle and Analytic Manifilds, R, 4.1.3).
For R > 0, let G1‘ = {xeAl ”at" < R}; we write G = G9. The series 5
(resp. i) defines an analytic mapping 9A (reep. IA) of G (resp. G1) into A. We
write:
(22) ep(x) = l + eA(x) =n§ofi forxeG

(23) logA(x)=lA(x—l)= ”2p 1)n-1(‘—" )" four—led,

(we omit the index A when no confusion can axle). For x E Ga and n 2 l 5
x”
(24) g <
nl < R"a'("‘1’° = Mg)"—1
and hence :A(GR) C G“, IA(GR) C GB for R S a“.
PROPOSITION 4-. Let R be a real number such that 0 < R < a9. The mapping expA
define: an analytic isomorphism qf GR onto 1 + Ca and the imam bmmphiwn i:
the mtrictian rgtflogA to l + Ca.
¢(1(X)) = l(:(X)) = X. By (20), (21) and Dgfirmh‘able and Analytic Mum'-
foldl', R, 4.1.5, we deduce that eA(lA(x)) = IA(2A(x)) for x6 Ga. Then
ep(logA x) = x forarsl + GR
logA(ep x) = x for x e GR
which completes the proof.
If A 15 given the bracket [x, y]: y,x A becomes a complete normed
Lie algebra: for IW— W“ < SUPUWII l 11*") < Mill Hill Proposition 2 of
no. 3_implies that the domain of absolute convergence of contains G x G
and H therefore defines an analytic function h. G X G ~>A; then

(25) Iz(x.y)= 1.2no H. .(m)

175
II FREE LIB AmEBRAS

PROPOSITION 5. Far 1, y in G,
(26) exPA-CXPAy = ep m": .11)-
2%" = emu-V) and hence
ma(1 +€,1+E)=(l+i)afl
in FHA x A;A) (where m denotes multiplication on A). The proposition
then follows from Proposition 2, Lemma 3 and Dyfirmtiabl: and Analytic
Manifolds, R, 4.1.5. .

APPENDIX
MOBIUS FUNCTION
Let n be an integer >1. If n is divisible by the square of a prime number, we
write y.(n) = 0. Ifn is not divisible by the square ofa prime number, we write
p.01) = (—l)", where k is the number of prime divisors of n. The function
u: N“ —> (— l, 0, 1} thus defined is called the Mfibiuxfirwtian.
Recall that given two integers n1 2 l, 712 2 1, we write n,|n2 if 711 divides
n2.
Pnopoamou. (i) The fimm'on y. i: the unique mapping If N“ into Z such that
y.(1) = 1 and
(1) “2M = o
for way integer n > 1.
(ii) Let r and t be two mapping: JN‘ into a cammutative group written additiuely.
In order that
(2) :(n) = gm!) firevay integern 2 I,
it i: mommy and mgficient that
(a) t(n) = 1% “(0:6) for my integer n 2 1.
The uniqueness assertion in (i) is obvious, for (1) allows us to determine
Mn) by induction on n. We show that the function y. satisfies (1). Let It be an
integer >1. Let P be the set of prime divisors of n and let u = 1—; 11%“) be
us
the decomposition of 7: into prime factors. If at is a divisor of n, then y.(d) = 0
unless a' is of the form E1), where H is a subset of P. Then

Z W) = Z (—1)°-""*
Iiln 3:?

176
 APPENDIX

Let .r and I be two mapping of N‘ into a commutative group written addi-

tively. Let n e N*. If (2) holds, then

“,2, ”((0:6) = “2, 9(4) “(2,“, ,(3) = 2 mm)

= 6|»
22(8) [(2,901) = too.
Conversely, if (3) holds, then

2:01) = an»
am 229(3):“)
m E = m
2:01) dun/u)
2 “(8) = :(n),
which completes the proof.
Formula (3) is called the Miz'bimr itwem‘mflrmulm

177
 EXERCISES

§1
1. Suppose that K is a field of characteristic 0. Let E be a cocommutative
bigebra of finite rank over K. Show that P(E) = (0} (apply Theorem 1 of
no. 6).
2. Let E be a bigebra such that P(E) = {0} and let (En),.,,, be a filtration
of E compatible with its bigebra structure. Show by induction on n that
E; = {0} for all n 9 0 and deduce that E“ = {0}, Le. that E reduces to K.
3. Let G be a monoid and E = K[G] its algebra (Algebra, Chapter 111,
§2, no. 6).
(a) Show that there exists on E one and only one cogebra structure such
that :(g) = g ® g for all g e G; this structure is compatible with the algebra
structure on E and makes E into a cocommutative bigebra whose counit a
is such that e(g) = 1 for all g e G.
(b) Show that every primitive element of E is zero. Deduce that, if G eé {e},
the bigebra E admits no filtration compatible with its bigebra structure (apply
Exercise 2).
(6) Suppose that K is an integral domain. Show that the elements of G are
the only non-zero demerits x e E such that :(x) = x ® x. Show that E has an
inversion 1' (cf. Algebra, Chapter III, § ll, Exercise 4.) if and only if G is a
group and that in that case i(g) : g'1 for all g e G.
4. Let E be a cocommutative bigebra and let G be the set of g e E such
that 5(g) = l and 0(g) = g (8 g. Show that G is stable under multiplication
and that it is a group if E has an inversion. Show that, if K is a field, the
demerits of G are linearly independent over K.
5. Let E be a bigebm and let E' : Hom(E, K) be its dual with the algebra
structure derived by duality from the cogebra structure on E (Algebra, Chap-
ter III, §ll, no. I). Let m be the kernel of the homomorphism ut—>u(l) of

178
 EXERCISES

E’ onto K; it is an ideal ofE’. Show that, ifu e m2 and x e P(E), then u(x) = 0.
When K is a. field show that P(E) is the orthogonal of m“ in E"; deduce that
dim P(E) S dim m/m2 and that we have equality if E is of finite rank over K.

6. Let E be a bigebra and let u e E’ = Hom(E, K) (of, Exercise 5). Let

gEE be such that e(g) = l and £(g) = g®g. Show that the mapping
u~—> u(g) is an algebra homomorphism of E’ into K and that the mapping
y »—> gy is an endomorphism of the cogebra E; show that this endomorphism
is an automorphism if E has an inversion.

7. Suppose that K is a field. Let E be a bigebra satisfying the following

conditions:
(i) E is commutative;
(ii) E has an inversion (Algebra, Chapter III, § 11, Exercise 4);
(iii) P(E) = {0};
(iv) E is of finite rank over K.
Let E’ denote the dual K-algebra of the oogebra E. Let G denote the set of
elements geE such that e(g) = land c(g) = g ®g; it is a group (cf. Exer-
cise 4).
(a) Let g e G and let m, be the kernel of the homomorphism u l—b u(g) of
E’ into K. Show that m, = m: (reduce it to the case g = I using Exercise 6;
then apply Exercise 5).
(17) Suppose that K is algebraically closed. Show that the ideals m, are the
only maximal ideals of E’; deduce that their intersection is {0}, that E’ is
identified with the product KCl and that E is identified with the bigebra K[G]
of the finite group G (cf. Exercise 3).
8. Show that the enveloping bigebra of a Lie algebra g has an inversion i
and that i(x) = —x for all x s 17(9).
9. Suppose that K is a Q-algebra. Let g be a Lie algebra admitting a K-
basis and let U be its enveloping bigebra with the canonical filtration (Un),,,o.
Let x e U" and let it be an integer 21. Show that It belongs to U: if and only
if N (It) belongs to 4 2.. mm: to U7).
l 3/;
1

10. Suppose that K is a Q;algebra. Let E be a cooommutative K-bigebra

with a filtration compatible with its bigebra structure. Show that the morphism
fa: U(P(E)) ~> E
defined in no. 4, Proposition 8 is an isomorphism, if P(E) is assumed to be a
free K-module (same proof as for Theorem 1).

11 ll. Let E be a bigebra and let I be a finite set. We are given a basis (4.)

l 79
II FREE LIE ALGEBRAS

of E indexed by the elements at ofNI and such that:

(i) to = 1;
(ii) 5(a) = “25:“ em 8) 25 for all y E N“.
Condition (ii) implies that E is isomorphic as a cogebra to the cogebra TS (K’)
of symmetric tensors of KI (cf. Algebra, Chapter IV, §5, no. 7).
(a) Let E’ = Hom(E, K) be the dual algebra ofE and let A = K[[(X1),E,]]
be the algebra of formal power series in indeterminam X4 indexed by I. For
all u E E' let 6).. be the formal power series .

2 new,
where x“ = 1:! X7“).
Show that u H d)“ is an isomorphism of E’ onto A and that this isomorphism
maps the topology of simple convergence on E’ to the product topology on A.
(17) Let a“, be the constants of structure of the algebra E relative to the
basis (cu). Then
‘oh = Z ‘uov‘v-
We write x instead of (X05, and y instead of (Y.)m, where the Y‘ are new
indeterminates. Show that there exists a. family f(x, y) = (fi(x,y)).“ of
formal power series in the variables x, y such that

My v = game“ for all y em,

where we write, as above, at“ = H Xi“) and similarly for y” and f(x, y)”’.
Show thatf (x, y) is aformal group law over K of dimension n = Ca.rd(I), in
the sense of Chapter I, § 1, Exercise 241'; define a Lie algebra isomorphism of
this formal group law onto the Lie algebra P(E), which has basis the ca with
[at] = l.
(0) Conversely, show that every formal group law over K in the x and y
can be obtained by the above procedure, which is unique up to isomorphism.
(d) When K is a Qalgebra, apply the above to the enveloping bigebra of
a Lie algebra g with a finite basis over K. Deduce the existence (and unique-
ness, up to isomorphism) of a formal group law with g as Lie algebra.
(e) Give explicitly the bigebra corresponding to the formal group law in
one parameterf(x,y) = x + y (resp.f(x,y) = x + y + xy).
1} 12. Suppose that K is a field of characteristic 1; > 0.
(a) Let g be a Lie p-algebra (Chapter I, § 1, Exercise 20) and let [-1 be its

1‘ The exercise in question assumes that K is a field, but nothing is lost if this
hypothesis is suppressed.

180
 EXERCISES

restricted enveloping bigebra (Chapter I, §2, Exercise 6). Show that there
exiss on 0 one and only one bigebra structure which is compatible with its
algebra structure and for which the elements of g are primitive. Show that 1-} is
commutative and that Pa?) = g.
(b) If n is an integer 20, let C1,, denote the vector subspace of El generated
by the products x,. . .x” where x‘ e g for all 1'. Show that (73“),0‘, is a filtration
of (I compatible with its bigebra structure.
(a) Let E be a bigebra and let 5 = P(E) be the Lie algebra of its primitive
elements. The mapping x >—>x" leaves 9 stable and gives 9 a Lie p-algebra
structure. Show that the canonical injection 3 —> E an be extended uniquely
to a bigebra morphism U —> E and that this morphism is injective (use the
fact that, if (Mm is a basis of g with a total ordering, the monomials H2:
(0 S n. < 12) form a basis ofU (loo. oil.) and argue as in the proofofLemma 2).

§2
1. Let g be a free Lie algebra with basic family (x1, . . .,x,,. . .). Let 9,,
denote the subalgebra of 9 generated by x”. ..,x,, and let 5,, denote the
smallest ideal of g,‘ containing .7,"
(a) Show that [1,. is the submodule of 9,, generated by the

(aim) ° ... ° ad(#r.))(xn),

wherek20andihgnforallh. ,
(b) Show that 5,, = g,_1 e b"; deduce that g is the direct sum of the Ia,I for
n 2 1.
(6) Show, using (a) and (b), that the K-module g is generated by the ele-
ments [21,: [‘42) . . ., ["tk—v m]. . .]], where k 2 0 and 1",. g i,‘ for]; < 1:.

fl 2. Let X be a countable set with at least two elements and let 3 be the set
of subsets of M(X) which are Hall sets (CE Definition 2). Show that
Card(b) = 2‘0.
3. Let X be a set with a total ordering. Show that there exists a Hall set
relative to X such that H n M3(X) consists of the products z(yx) with y < x,
y S z and that H n M‘(X) consists of the products w(z(yz)) with w 2 z 2 y,
y < x and the products (ab)(cd) with a < b, c < dand either a < cora = c
and b < d.
4. (a) Show that the Lie algebra defined by the presentation

bray; [a [MD = H, [MD = 0}

l8l
I! FREE LIE ALGEBRAS

admits the basis (2:, y, [3, y]). Define an embedding of this algebra in a matrix
algebra.
(b) Do the same for the Lie algebra. defined by the presentation
(at: [:8 [m1] — 2x = [a [a y]] + 2y = 0}.
(1:) Show that the Lie algebra with presentation
{2,y,z;[x,y] -x = [mz] ~11 = [4*] - z = 0}
reduces to 0.
5. Let g be a free Lie algebra and let I be an ideal of 9. Suppose that
r = [g, I]. Show am 2 = {0} (use the fact that "Q“ m = {0}).
6. Let E be a module. Let ME be the flee algebra of E (Algebra, Chapter
111, §2, Exercise 13). Recall am ME = g; E," where
131:3, E,,=y£‘qD_nE,®Ea if ">2,
and the algebra structure on ME is defined by means ofthe canonical mappings
Eu 8 En -> BMW
(:1) LetJ he the smallest two-sided ideal of ME containing the elements of
the form xx and (IQ/)2 + (yz)x + (zx)y with x, y, z in E. Let LE = ME/J.
Show that] is a graded ideal of ME; deduce a graduation (L"E),k>1 of LE.
(b) Show that LE is a Lie algebra; it is called the free Li: algtbra of the
module E. Show that for every Lie algebra g and every linear mappingf: E —> g
there exists one and only one Lie algebra homomorphism F: LE —> g which
extends f.
(0) Define isomorphisms E —> LE and A 2E —> LE. Show that LSE is
identified with the quotient of E ® AaE by the submodule generated by
the elements
x®(y A z) +y®(z Ax) +z®(x Ay) forx,y,zinE.
(d) Show that, when E is a free module of basis X, LE is identified with the
free Lie algebra L(X) defined in no. 2.
(a) Show that the canonical mapping of E into the enveloping algebra
U(LE) of LE can be extended to an isomorphism of the tensor algebra TE
onto U(LE).
(f) Let 0' be the linear mapping of A 2E into TzE such that
a(xAy)=x®y—y®x.
Construct a module E for which a is not injective. Deduce that for this module
the canonical mapping of LE into U(LE) is not injective (compare with
Exercise 9 of Chapter I, § 2).

182
 EXERCISES

7. Suppose that K is a. field. Let I be a free Lie algebra with basic family
(mm! and let M be an I-module.
(a) Show that H2(l, M) = {0}, of. Chapter I, §3, Exercise 12. (Use Corol-
lary 2 to Proposition 1 and part (i) of the Exercise in question.)
(b) Show that, for every family (711,)”, of elements of M, there exists a
cocycle 4): I——>M of degree 1 such that 420:.) = m. and that this cocycle is
unique. Deduce an exact sequence:
0—>H°(l, M) -—>M—>M"—>H‘(l, M) —>0.
If I is finite of cardinal n and M is of finite rank over K, then
rg H1(I, M) —— rg H°(l, M) = (n — l) rgM.
8. Suppose that K is a field. Let I be a free Lie algebra with basic family
(2),“, let r be an ideal of 1 contained in [l, I] and let 9 = l/r; let x, denote the
image of 2‘ in g.
Show the equivalence of the following properties:
(i) g is a flee Lie algebra.
(ii) g admits (2;) as basic family (i.e. t = (0}).
(iii) For every g-module M, H38!) M) = {0}.
(iv) If g operates trivially on K, then H’(g, K) = {0).
(The implications (ii) => (i) and (iii) 3 (iv) are obvious and (i) => (iii)
follows from Exercise 7. To prove that (iv) => (ii) it suffices to show that
r = [L r], cf. Exercise 5; otherwise take a hyperplane b of 0: containing [1, r]
and note that the extension r/b —> [/9 —> 1/: = g is essential.)

1T 9. Let g = "(‘2 9,, be a graded Lie algebra. If D is a graded subalgebra of

9 such that g = I) + [9, 9], show that b = g. If (x.) is a family of homogeneous
elements of g, deduce that (xi) is a generating family if and only if the images
of the It in Ms 9] generate 9/[s. 9]-
IfK is a field on which 9 operates trivially and H‘(g, K) = {0), Show that the
family (X!) is basic if and only if the images of the x‘ in g/[g, 9] form a basis of
this vector space. (Use Exercise 8.)1'
10. Let [be a free Lie algebra admitting a finite basic family with n elements
and let u be a surjeetive endomorphism of I. Show that u is an isomorphism.
(Write gr"(l) = (WI/W'“! and denote by gr"(u) the endomorphism of gr”(l)
derived from u; note that gr"(u) is surjective; as gr”(l) is a free K-module of
finite rank, deduce that gr"(u) is bijective since the kemel of u is contained in
Q m, which is {0}.)

1 When 5 is not graded, we do not know whether the condition

“H’(g, M) = {0} for every 9-module M”
implies that 9 is a free Lie algebra.

183
[1 FREE LIE ALGEBRAS

If (y), . . . , y“) is a generating family of I, show that m 2 n and that we have

equality if and only if (yl, . . ., y”) is a basic family.
1i ll. Let Y and Z be two disjoint sets (with Y totally ordered), let
X = Y L) Z and I = L(X). Let aY be the submodule of I generated by Z and
[1, l] ; it is an ideal of the Lie algebra [. We propose to find explicitly a basic
family of (1,.
Let M be the set of positive integral functions on Y of finite support. If
m e M, we denote by 6" the endomorphism of 1 given by
on = "1:; (ad ”mm,

where the product is relative to the order relation given on Y. Let 6 denote the
subset of Y x M consisting of the ordered pairs (14, m) such that there exists
y e Y for which 3/ > u and m(g) 2 1. Show that the elements
0’"(u) for (u,m) 56 and 6’"(z) for (z, In) El x M
form a basic family of the Lie algebra fly.
(Reduce the problem to the case where Y is finite and argue by induction
on Card(Y). Use the Corollary to Proposition 10 applied to the greatest
element 3/ of Y and apply the induction hypothesis to Y - {y}.)
12. Let X = {1,31} be a set of two elements. Show that the derived algebra
of L(X) is a flee Lie algebra admitting as basic family the elements
((ady)“ a (ad z)’)(y), for]: 2 l, q 2 0. (Use Exercise ll.)
1[ 13. (In this exercise we assume that every projective module over K is
free; this is so for example if K is a principal ideal domain.)

Let I = "21 I,‘ be a graded Lie algebra admitting a basic family B consisting
of homogeneous elements and let 0 be a graded Lie subalgebra of l, which is a
direct factor of I considered as a module.
(a) For all i 2 0, let I“) be the graded subalgebra of I such that
1;" = b, ifj <21
1}”: I, ifj> 1'.
Then I = I“) D I“) D . - - D I). Show, arguing by induction on i, the existence
of a basic family B“) of [(0, consisting of homogeneous elements, such that the
elements of B“'” and of B“) of degree <i — l are the same. (Suppose that
B“‘” has been constructed Let m, be the intersection of I‘ = 6““ with the
subalgebra generated by the b, for j < i and let b; be the submodule of l,
generated by the elements of B““’ of degree 1'. The induction hypothesis
implies that I‘ = tnl ® 1).. As m, C bl, we can decompose has b‘ = 1). ® 3., so
that I). = m, ® 6:; by the hypothesis on K the modules 9‘ and a, are free;

184
 EXERCISES

changing the elements 013““) of degree i ifnecessary, we can assume that a, is

generated by a subset of B“‘”. Applying Exercise 11 to [("1’ and its ideal I“),
we obtain a basic family 8‘” of I“) with the daired properties)
(b) Deduce from (a) the fact that I) has a basic family consisting of homo-
geneous elements.
1i 14. Suppose that K is a field. Let X be a set and x, y two elements of
L(X) which are linearly independent over K. Show that the family (x, y) is
free in the Lie algebra L(X). (Let x, (resp. y,) be the non-zero homogeneous
component of x (resp. y) of highest degree. Adding if necessary a multiple of
y to x, we can assume that x, and y, are linearly independent. The subalgebra
of L(X) generated by x, and y, is graded and hence free, cf. Exercise 13.
Deduce that (29,, ya) is a. flee family and pass from this to (x, y).)
1T 15. Let X = {x, y} be a set with two elements and let a- be an automor-
phism of the Lie algebra L(X). Show that, ifK is a field, a preserves the gradua-
tion of L(X) (apply Exercise 14 w the non-zero homogeneous components of
6(x) and e(y) of highest degree); deduce an isomorphism of Aut(L(X)) onto
GL(2, K), Extend these results to the case where the ring K has no nilpotent
element #0. When K contains an element 2 eé 0 of zero square, show that
x 9):, gay + s[x,y] can be extended to an automorphism of L(X) which
does not preserve the graduation of L(X).
1f 16. Suppose that K is a Q—algebra. If g is a Lie algebra, we denote by
U(g) its enveloping algebra, by a the canonical mapping of 9 into U(g) and by
5(9) the symmetric algebra of the K—module g. There exists one and only one
linear mapping
11(9): 5(8) -> U(9)
such that 11(9) (x") = 6(1)" for all x E g and all u E N. Then
1
mm. . .x.) = 72,, cm. . arm), x. e s-
We pmpose to show that 11(9) is bijmive.
(a) Prove that 11(9) is surjective and that it is bijective when the K—module g
is free (use the Poincaré—Birkhoff—Witt Theorem).
(b) Let b be an ideal of 9. Let a, (resp. uh) denote the ideal of 5(9) (reap. the
two-sided ideal of U(g)) generated by b (resp. a(b)). The diagram
0 —> ’6 ——> 5(3) —> S(g/b) —> 0
no) lute/t)
0 —> u,—> U(g) -—> U(g/b) —> 0
is commutative and its rows are exact. Deduce that the image it, of 5,, under
71(9) is contained in u6 and that ft, = u, when fig) and 11(g/b) are injective
(in particular, when the module 5] and g/b are free).

185
II FREE LIE ALGEBRAS

(0) Take 5 to be a free Lie algebra with basic family (X1, . . . , X,u H),
where n > 0, and take b to be the ideal of 9 generated by H. Show that g and
all) are free modules. Deduce that, in that case, in, = 11.. In particular, there
exism x e ”a such that (n(g))(x) = 6(Xl) . . .a(X,.)a(H).
(d) We return to the general case. Show that u, is generated over K by the
elements of the form 60:1). . .a(x,)o(h), with n 2 0, x) e g and h e 9 (note that
“a coincides with the left ideal generated by 1)). Show, using (0), that such an
element belongs to it, (use a suitable homomorphism of a free Lie algebra into
9). Deduce that fla = uh. v
(a) Show that, if n(5) is bijective, so is 11(g/b). Deduce finally that, for every
Lie algebra 9,1;(9) is bijzttiva (write g as the quotient of a free Lie algebra) and
that the canonical homomorphism
(015(9) —> grU(g) (cf. Chapter I, § 2, no. 6)
is an isamorphirm (“Poincaré—Birkhofi—Witt Theorem” for Lie algebras over
Q—algebras).

§3
The letter X denotes a set.
I. (a) Show that every element u E A"(X) can be written uniquely in the
form it = “x uzx, where ux e A(X).
(b) Show that {0) is the only submodule ofA+ (X) which is stable under all
the mappings u ~—> u,‘ (x e X); (Ifa is such a submodule and a aé {0), consider a
non-zero element of a of minimal degree.)
2. Let g be a Lie algebra which is a free module; it is identified by means of
a: g —> Us with its image in the enveloping algebra U9, Let U *9 be the kernel
of the canonical homomorphism Ug —> K. Let i be a mapping of X into 9 such
that i (X) generates g as a Lie algebra.
(a) Show that U+g is generated by i(X) as a left Ug-module.
(b) Show the equivalence of the following properties:
(i) g is free with basic family i: X —> 9.
(ii) The family i is a basis of the left Ug-module U*g.
(The implication (i) 9 (ii) follows from Exercise 1 (4) using the isomor-
phism Us; —> A(X). To show (ii) => (i), apply Exercise 1 (b) to the kernel of
the homomorphism A(X) —> Ug defined by L)
3. Let x e X. Show that the centralizer ofx in A(X) is the subalgebra gener-
ated by z. Deduce that the only elements of L(X) which commute with x are
the multiples ofz. In particular, the centre ofL(X) is (0) if Card(X) 2 2 and
the centre of L(X)/(n; L"(X)) reduces to the canonical image of L”(X).

186
 EXERCISES

1i 4-. Suppose that K is a field of characteristic p > 0.

(a) Let H be a bam's of the Lie suhalgebra L(X) of A(X). Show (using the
isomorphism U(L(X)) —> A(X) and the Poincaré—BirkhofllWitt Theorem)
that the elements It" (h e H, n e N) are linearly independent over K.
(b) Ifm is an integer 2 0, let L,,I denote the submodule ofA(X) with basis the
If", where h e H, n < in. Show that L“ is a Lie subalgebra ofA(X) and that, if
a 6 Ln-“ then a” 5 LM (argue by induction on m and use the Jawbmnformulae,
cf. Chapter I, § 1, Exercise 19). Deduce that L, does not depend on the choice
of H and that it is the Lie subalgebra of A(X) generated by the x’", where
x e X, n < m.
(0) Let L(X, p) be the union of the Lm for m > 0. Show that L(X, p) is the
smallest Lie p—subalgebra of A(X) containing X (cf. Chapter I, § 1, Exercise
20); it admits as basis the family ofh’", where It 6 H, n E N.
(ii) Let CT(L(X, 17)) be the restricted enveloping algebra of L(X, 1;), cf.
Chapter I, § 2, Exercise 6. Show that the injection ofL(X, fl) into A(X) can be
extended to an isomorphism of fi(L(X, 9)) onto A(X). (Use the Poincare—
Birkhoff—Witt Theorem and the exercise quoted above.) Deduce that L(X,p)
is the set of primitive elements of A(X), of. § 1, Exercise 12.
(e) Letf be a mapping of X into a Lie p-algebra 9. Show thatf can be ex-
tended uniquely to a p—homomorphism F: L(X, p) —> 3. (Begin by extendingf
to an algebra homomorphism ofA(X) into Us.)
(The Lie p-algebra L(X, p) is called thefree Lie p—algebm on the set X.)

§4
In thefollowing exorcism the let!” G denote: a group.
1. Let (Gn) be an integral central filtration on G. Show that the Lie algebra
gr(G) is generated by gr1(G) if and only if G" = GM1.C”G for all n 2 1. In
that case, show that G" = 0”,.a for m > n and deduce that (0,.) = (C"G)
if there exists an integer m such that Gm = {2).
2. Let (Gm) be a real filtration on a group G and let u be the corresponding
order function. Let H be a subgroup of G.
(a) Let Hu = H n G“. Show that (H,) is a real filtration of H and that the
corresponding order function is the restriction of v to H. H: = H n G: and
gr(H) is identified with a graded subgroup of gr(G).
(1)) Suppose that H is normal and that ”(G) n R is a discrete subset ofR.We
write (G/H)“ = (GnH)/H. Show that ((G/HL) is a real filtration ofG/H and
that the corresponding order function 05,3 is given by the formula
”GIEO‘) = $3314!)-
Show that, for all at E R, there is an exact sequence
0 —> gram) -> grAG) —> gram/H) —> 0.
187
II FREE LIE ALGEBRAS

(6) With the hypotheses of (b), suppose that (G“) is a neutral filtration. Show
that so are the filtrations on H and G/H induced by (Ga) and that the Lie
algebra gr(G/H) is identified with the quotient of gr(G) by the ideal gr(H).
fl 3. Let (Hm) be a real filtration on a group H and let vH be the correspond-
ing order function. Suppose that G operates on H on the left and that, for all
g e G, the mapping h »—> g(}z) is an automorphism of the filtered group H.
(a) If a E R, we denote by Ga the set ofg E G such that
vn(h‘1g(h)) 2 1/301) + a for all h e H.
Show that (G) is a real filtration of G and that Gm = G if a: S 0. Let 0 denote
the corresponding order function.
(b) Suppose that the filtration (Ha) is central and that G; = G. Show that
(GE) is a central filtration of G. If E E grm(G) and n E grad-I), let g (resp. h) be a
representative of E (resp. 1;) in G“ (resp. Hg) 3 show that the image of h'1g(h) in
grm+ fl(H) is independent of the choice ofg and h; ifit is denoted by D301), show
that D: can be extended to a derivation of the Lie algebra g'r(H) of degree on
and that EH D: is a homomorphism of the Lie algebra gr(G) into the Lie
algebra of derivations of gr(H). If 123(H) n R is contained in a discrete sub-
group I‘ ofR, so is 0(G) n R and, for all a e R, the mapping EH Dz defined
above is injective.
4. Let H be a nilpotent group ofclass c and let G be the group of automor-
phisms of H which act trivially on H/(H, H). Show that G is nilpotent of class
<2 —- 1. (Apply Exercise 3 to the lower central series of H and note that G
operates trivially on gr(H).) Show that, if H is a finite p—group (p prime), so is
G (same method).
5. Let K be a commutative field and let L be a finite Galois extension of K
with Galois group G. Let u be a valuation of L with valud in Z (Commutative
Algebra, Chapter VI, § 3, no. 2) which is invariant under G. If g e G, we
write

"(9 = :23. “(W)-

Show that v is the order function of a separated integral filtration (G) on G
such that Go = G and that the restriction of this filtration to G1 is central
(apply Exercise 3, taking H to be the ideal of ”l Show that G1 is a p-group
(resp. reduces to the identity element) if the residue field ofL is of characteristic
1; > 0 (resp. of characteristic zero); when the fields L and K have the same
residue field, G1 is the kernel of the homomorphism (I) defined in Commutative
Algebra, Chapter VI, § 8, Exercise 11 ([7).
1i 6. Let K[G] be the algebra of G over K and let I be the kernel of the
canonical homomorphism K[G] —> K (“augmentation ideal”). Then

188
 MERGERS

K[G] = K («B I and I admits as basis the family ol'g — 1 wherege G-{z}.
(a) If n is an integer 2 0, we denote by I’I the ideal of K[G] the n-th power
of I. Let G" be the set ofg e G such that g — l e I". Show that (G) is an in-
tegral central filtration on G. In particular, G" 3 CG for all n.
(b) Show that if K = Z the mapping g >—> g — 1 defines on taking quotients
an isomorphism of G/(G, G) onto III“. Deduce that G, = C’G.‘|’
(5) Suppose that K is a field of characteristic zero and that G is finite.
Show that I" = I foralln 21.
(:1) Suppose that K is a field of characteristic p > 0 and that G is a p-group.
Show that I"(0} for n sufficiently large. (Show first, using Algebra, Chapter I,
§6, no. 5, Proposition 11, that every simple K[G]-module is isomorphic to K ;
deduce that I is the radial of K[G] and hence is nilpotent since K[G] is of
finite rank over K.)
(2) Suppose that K = Z and that G has the following property: for all
g e G such that g aé 1, there exists a prime number p, a p-group P and a homo-
morphisrnf: G—> P such that f(g) # a. Show that in that case 0 I” = {0}.
(Reduce it immediately to the case where G = P. By applying (d) to the field
17,, We see that there exists m such that I'" C p.Z[G] and, as I is a direct
factor in Z[G], this implia that I"l C p1, whence Q I'"" C Q p"I, which
reducu to 0 since I is a finitely generated Abelian group.)
7. Let G be given the filtration (C‘G) and suppose that gr, (G) = G/(G, G)
is cyclic. Show that grn(G)= {0} for n 2 2 (use Proposition 5) and deduce that
C"G= (G, G)forn2 2. '

s. LetG=SL2(Z)andletx=(01 1),y=(l
1 1 o1),w=(_lo 0).
l
(a) Verify the formulae w4= l, w = xy“x, warm" = y'.
I

(b) Ifg = (Z :) isanelement ofG, we writel((g): In] + |b[. Show that

l(g) = 1 if and only ifg is of the form ynw“, Where n 61 and O< S < 3. If
l(g) 2 2, show that there exists a. power h of: ory such that 1(gh) < [(g).
Deduce that G is generated by (x, y).
(5) Using (a) and (b), show that G/(G, G) is generated by the image E of x
and that E” = LI Deduce that C"G = (G, G) for n 2 2. (apply Exercise 7).

9. Let G be given the filtration (CG).

2 grq(G).
(4) Show that the ideal of gr(G) generated by gr,(G) is U>2

1' We do not know whether G, = C”G for all n. It is true in any case when G is a
file group, cf. §5, Exercise 1.
1 It can be shown that E is nfnrdn 12.

189
X! FREE LIE AMEBRAS

(b) Let (11):“ be a generating family of G and let Ill 2 1. Suppose that, for
all (i,j) e I”, (x,, x,)"' 5 (PG. Show that, {or all q 2 2 and all u 6 04G, 14'" E C“ 1G
("56 (3))
10. Let x, y be in G and let i, 5 be two integers 21. Suppose that (x', y‘) = e.
(a) Show that, for all n 2 l, (x, y)”I e C”*’G. (It can be assumed that G is
generated by {x, y}; then apply Exercise 9 (b), noting that (x', y') E (x, y)"
mod. 03G.)
(b) Suppose that z’ = y’ = a; let t denote the g.c.d. of r and .t. Show that _
(x, y)‘ e CaG and deduce that (x,y)“ E C"”G for all n 2 l (same method).
11. Let H be a subgroup ofG and let in be an integer 2 1. Suppose that G is
generated by a family on)“, such that x2" 6 H for all i.
(a) Let H be given the filtration induced by the filtration (C’G) on G and
let gr(H) be identified with a graded Lie subalgebra of gr(G), of. Exercise 2.
Show that, for all n 2 0,
m".gr,,(G) C gr,(H).
Deduce that, for all z e H. CG, 2’” e H. O‘HG.
(b) If G is nilpotent, show that there exiss an integer N 2 0 (depending
only on the nilpotency class of G) such that 2"" e H for all z e G.
12. Suppose that G is nilpotent. Let H be a subgroup of G and let m be an
integer 2 l and let x, y be elements ofG such thatx’" E H and y'" e H. Show that
there exists an integer N 2 0 such that (xi/W" e H. (Apply Exercise 11 to the
group generated by {x, y) and its intersection with H.)
13. (a) Let F be a Bee group with basic family (7:, y) of two elements, let
c be an integer 22 and let m be an integer 21. We write Fc = F/C°F and de-
note by x, y the images of}, y in F‘. Let Ff,l be the subgroup of F” generated by
{#311}. Show that there exists an integer N 2 0 such that 2“" E F; for all
z e F0 (use Exercise 11).
([1) Let Ic (reap. 1;) be the normal subgroup of Fc (resp. Ff") generated by y.
Show that, if N is chosen as above and :5 1‘, then 2“" e I; (note that FCIIc
is an infinite cyclic group with generator the image of z and deduce that
Fin/If, —> F°/Ic is injective and hence that I; = Ff, n I”). In particular
xy’""x‘1 e If".
(6) Suppose that G is nilpotent. Let H be a subgroup of G and let L be a
normal subgroup ofH. Let g E G and let in 2 1 be such that g'" E H. Show that,
if N is sufficiently large, glm" g” e L for all I e L. (If G is of class < 1:, choose N
as in (a) and use the homomorphismf: F9 —> G such thatf(x) = g, f (y) :1;
note thatf(a) C H, f(1,3,) C L and apply ([1) above.)
1} 14. Let P be a set of prime numbers. An integer n is called a P-integn if it
is #0 and all its prime factors belong to P. An element x e G is called a P-
tam'on element if there exists a P—integer n such that x“ = e; G is called a P-

190
 EXERCISES

tom'un (resp. P-tam'tmfm) group if every element of G is a P—tmsion element

(rap. if no element of G other than 2 is a P-toision element). G is called P-
dioisible if, for all x e G and every P-integer 12, there exists g EG such that
x = ".
Suppose that G is nilpatmt.
(a) Let H be a subgroup of G and let HP be the set of x e G such that there
exists a P-integer n for which x" e H. Show that Hp is a subgroup of G (use
Exercise 12) and that (Hp)P = Hp. The group HP is called the P—mtumtim of
H in G. If G is P-divisible, so is HP.
(b) Let L be a normal subgroup of H. Show that LP is normal in Hp. (Use
Exercise 13 (c) to prove that, ifg 6 HP and l E L, then gig“ E LP.)
In particular, if L is normal in G, so is LP and G/LP is P—torsion-free. Deduce
that the set of P-torsion elements of G is the smallest normal subgroup N of G
such that G/N is P-torsion-frce.
(6) Suppose that G is P-torsion-free. Let n be a P—integer and let x, y s G be
such that x" = 3/”. Show that x = 3/. (Apply Exercise 10 (a) with 7 = J = n.
Deduce that there exists a P-integer N such that (x, y)" = e, whence (x, y) = 5
since G is P-torsion-frce; then (x' y)“ = e, whence finally 1: = y.)
(d) Let H be a subgroup ofG, let L be a nilpotent group and letf: H —> L
be a homomorphism. Let I1 be the graph off in H x L and let I} be its P-
saturation in G x L. Show that I} is contained in HP x L and that
prlz I‘p—> Hp is surjective if L is P-divisible and injective if L is P-torsion-
free.
Suppose that L is P—divisible and P-torsion-free. Show thatfcan be extended
uniquely to a homomorphism f9: H, —> L and that the graph off; is I}.
{I 15. Preserving the notation of the above exercise, suppose that G is nil-
potent. Let i: G —> G be a homomorphism of G into a nilpotent group G.
(i, G) is called a P-ewelnpe of G if the following conditions hold:
(i) 6 is P-divisible and P-torsion-fiee.
(ii) The kernel of i is the set of P-torsion elements of G.
(iii) The P-saturation of i(G) in G is equal to G.
(a) Let (i, G) be a P—envelope of G and let L be a P-divisible P-torsion-free
nilpotent group. Show that, for every homomorphism f: G —> L, there exists
one and only one homomorphismf: G —> L such thatfa i = f (reduce it to the
case where G is P-torsion-free and use Exercise 14 (11)). Deduce that, if G has a
P-envelope,T this is unique up to isomorphism.
(b) Let (i, G) be a P—envelope of G, let H be a subgroup of G, let H be the
P-envelope of i (H) in G and let in: H —> H the homomorphism induced by 1'.
Show that (in, H) is a P—envelope of H.
1' In fact, every nilpotent group has a P—euvelopc, cf. §5, Exercise 6; see also
M. Lazard, Amie: E. N. 5., 71 (1954), pp. 101—190, Chapter 11, § 3.
191
II FREE LIE ALGEBRAS

Suppose that H is normal in G. Then 1:1 is normal in G, cf. Exercise l4 (b_) ; if

ism: G/H ~> G/fi is the homomorphism induced by i, show that (1.6:!!!) G/H) is
a P-envelope of G/H.
16. We preserve the notation of the two previous exercises
(a) Let Sp be the set of P-integels and let Zxi = S; 12 be the ring offi-actions
ofZ defined by SF (Commutative Algebra, Chapter II, § 2, no. 1). Suppose that G
is nilpotent, P-torsion-free and P-divisible and let IE 21,, g E G. Show that
there exists one and only one element h of G such that h' = g" for all r E SP V
such that .rt 6 Z. The element h is called the t-th pawn ofg and denoted by g‘.
The mapping tn—> g‘ is a homomorphism of ZF into G. If t is invertible in 2?,
g »-> g“ is bijective. (Use Exercise 14 (5)).
(17) Let A be a commutative group and let i be the canonical mapping of A
into AP = Z? x A. Show that (1', AP) is a P-envelope of A.
(0) Let G (rap. G) be the lower strict triangular group of order I: over a ring
I: (resp. over the ring kl, = 1: ® Z?) and let i be the homomorphism of G into
G defined by the canonical homomorphism of k into k? (Algebra, Chapter II,
§ 10, no. 4). Show that (i, G) is a P—envelope of G,
(:1) Let G be a finite nilpotent group and hence the direct product of its
Sylow p—groups GP (Algebra, Chapter I, §6, no. 7, Theorem 4). Let G be the
product of the G, for [7 ¢ P and let i be the canonical projection of G onto G.
Show that (i, G) is a P-envelope of G.
1T 17. Suppose that G is nilpotent. Let P be a set of prime numbers and let
fi, (G) be the set of normal subgroups of G of finite index whose index is a P-
integer (cf. Exercise 14-). Let 9;,(G) be the topology defined on G by the filter
base 93,,(G) (General Topology, Chapter III, § 1, no. 2, Example).
(42) Let N be a subgroup of G of finite index such that (G: N) is a P—integer
and let N’ be the intersection of the conjugates of N. Show that (G: N’) is a.
P-integer (use the fact that the group G/N' is the product of its Sylow groups) ;
deduce that N is open with respect to .7},(G).
(b) Let H be a subgroup of G of finite index. Show that .7;(H) coincides
with the topology induced on H by 9;;(G). (Same method.)
(0) Let H be a normal subgroup of G such that G/H is isomorphic to Z and
let x be a representative in G of a generator of G/H. Let N e EP(H) and let
N’ = a x"Nx". Show that N’ e‘JJP(H) and that x ’x‘1 = N'. If N” is the
subgroup generated by N' and x, show that (G: N”) =' (H: N’). Deduce that
.7}(H) is induced by 934G).
(d) Suppose that G is finite. Show that, if H is a subgroup of G, there
exists a sequence of subgroups
H=Hoccu~c=G
such that H‘ is normal in H‘H for 0 < i < I: and that I-LH/H. is finite or

192
 EXERCISES

isomorphic to Z. Deduce, using (b) and (c), that j}(H) is induced by £(G).
(0) With the hypotheses of (d), let P' denote the complement of P in the set
of prime numbers. Show that the closure of H under 9;,(G) coincides with the
P’-saturation of H in G (Exercise 14) ; in particular, G is Hausdorff if and only
if it is P’-torsion-fiee.
18. The upper central series (ZLG) of the group G is defined inductively as
follows:
(i) Z,G = (e}i.fi< 0;
(ii) ZiG/Z¢_1G is the centre of G/ZhlG.
Then {a} = ZoG C Zc and ZIG is the centre of G; the LG are
characteristic subgroups of G.
(a) Show that G is nilpotent of class <5 if and only if G = ZbG.
(b) Show that (C"G, ZmG) c Zm_nG.
(0) Suppose that G is nilpotent and P-torsion-fi‘ee (where P is a set of
prime numbers, cf. Exercise 14). Show that the Z.G are P-saturated. (It
suffices to verify this for ZIG; ifn is a P-integer and g e G is such that g" e 21G,
then xg"x'1 = g" for all x e G, whence xgx'1 = g by Exercise 14 (r) and it
follows that g belongs to 21G.)
§5
In the following exercises we assume the hypotheses and notation of §5.
F denotes the free group F(X) and g the unique homomorphism of F into
the Magnus group I‘(X) such that g(::) = l + x for all xEX (cf. Theorem
1).
1. Let the algebra K[F] of F be given the filtration (1") consisting of the
powers of the augmentation ideal I (§ 4, Exercise 6).
(a) Let E: K[F] —> A(X) be the unique algebra homomorphism extending
g: F —>A(X)*. Show that 3 maps I" to the ideal A,(X) (cf. no. 1) and defines
when taking quotients an irorrwrphism E” of K[F]/1'I onto A(X)/A(X). (Define
an inverse homomorphism of g, by means of the homomorphism of A(X)
into K[F] which maps 1: to x — l for all xeX.) Deduce that A(X) is im-
morphic tn the Hausdoyftamplefim a[F] under the topology defined by (1").
(11) Suppose that K = Z. Show that the filtration (1") is separated (use
Proposition 3 and Exercise 6 (e) of § 4-) and that the filtration of F defined by
it (Exercise 6 (a) of §4) coincides with the filtration (C‘F). Deduce that -
E: Z[F] —> AZ(X) is injective.
2. Let G be a group, let K[G] be its algebra over K and let I be its aug-
mentation ideal (§ 4-, Exercise 6). Let 5 denote the canonical homomorphism
of K[G] onto K; then Ker(e) = I.
(a) Let M be a left K[G]-module and let Z(G, M) be the group of crossed
hommorfihimu of G into M (Algebra, Chapter I, §6, Exercise 7). If

[93
11 FREE LIE ALGEBRAS

f e Homxma, M), let fl be the mapping g»f(g — l) of G into M. Show

thatfl is a crossed homomorphism (use the identity
gg’— 1 =g(g’— 1) +g—1
to show thatf1(gg') = g.fl(g') +f1(g)) and thatf»—>f1 is an isomorphirm of
Homxmfl, M) onto Z(G, M).
(b) Take G to be the free group F = F(X). Show that, for every mapping
a: X —> M, there exists one and only one elementfa o(M, F) which extends
0. (Use the interpretation of Z(F, M) in terms of the semi-direct product of
F by M, cf. Algebra, Chapter I, lot. cit.) Deduce, using (a), that the family
(I: —- 1):“ is a basis of] as a left K[F]-module.
(0) By (b) every element 11 E K[F] can be written uniquely in the form

n = so» + 2 MW — 1),
:EX

where D,(u) E K[F] and the D,,(u) are zero swept for a finite number. The
mapping u v—> D,(u) is called the partial derivative with respect to 1:. It is charac-
terized by the following properties:
(i) D, is a K—linear mapping of K[F] into K[F].
(ii) Doom) = u.Dx(II) + s(v).Dz(u) for u, v in K[F].
(iii) D,(x) = l and Dg(y) = 0 inX — {x}.
If n 2 1, then
D,(x") =1 +x+ +z"'1
D,‘(x"‘) ..__ —x"‘ _ x-IH-l _ . . . _ 3-1,

(d) Let eA be the canonical homomorphism of A(X) onto K. Show that

every element u of A(X) can be written uniquely in the form

u = mu) + g, d.(u) .x,

where d,(u) e A(X) and the 41,,(11) are zero except for a finite number. Show
that dz: :1 d(u) is an endomorphism of A(X) which can be extended by
continuity to A(X) and enjoys properties analogous to (i), (ii), (iii) above.
Show (1,”; = END,” where g is the homomorphism of K[F] into A(X)
which extends g (cf. Exercise l).‘]'
1T 3. We preserve the notation of Exercise 2.
(a) LetJ be a right ideal of K[F] contained in I and let u be an element
of I. Show that
ueJ.I oDAu) EJ for allX.
In particular, an element of I belongs to I" (n 2 1) if and only if all its partial
derivatives belong to 1"”.

1‘ For more details on the D3, sec R. Fox, Ann qf Math, 57 (1953), pp. 547—560.

194
 EXERCISES

(b) Let R be a normal subgroup ofF, let G = F/R and let] be the kernel
of the canonical homomorphism 7: K[F] ~> K[G]. Show thatj is generated
as a left (resp. right) ideal by the elements I ’— 1, where 16R. Prove the
exactness of the sequence

(*) o—>J/q.1)i>1/0.1)_”>K[G]-i>K—>o,
where at is derived by taking quotients from the inclusion of] in I and fl is
derived by taking quotients from the restriction ofy to I.
(c) For n e I, 9: ex, let 5,04) denge the image of Dx(u) in K[G] under
y. Show, using (a), that the family (Dmx defines on taking quotients an
immarfihirm of I/(J ‘1) onto K[G]“L
(d) For 76R, let 0(r) be the image off ~ 1 inJ/(JJ). Show that
9(r1’) = 9(1) + 9(7') and 9(y7y‘1) =y.0(r) forr,r’inR,yEF.
(Use the identities
rr’—l= (r— l)(r'-— 1) + (r—l) +(r’—l)
m" - 1 =y(r - 1)(y" - 1) +y(r - 1).)
Deduce that 6 defines a homomorphism of R/ (R, R) into J/(J .1) compatible
with the action of G and show that the image of this homomorphism generates
the K-module j/(JJ); when K = Z, show that an bamphiwn is thus ob-
tained of R/ (R, R) onto j/(JJ) (define the inverse homomorphism directly).
(5) Let (1a)“E A be a family of elements of R generating R as a normal
subgroup of F. Show that the 9(a) generate the K[G]-moduleJ/(J .I).
(f) The matrix (:I5,,,(rm)),‘exvc‘5A defines a homomorphism
p: K[G]‘A’ —> K[G]"n
of left K[G]-modules. Show that the sequence

(H) K[G](A) _”) K[G]“) —°> K[G] —‘> K -—> o

is exact, where 8 is the homomorphism (14,) »—> ,2; u,(y(x) — l).
(Transform the exact sequence (*) using (c), (d), (5) above.”

4-. For all n E N, let 6) denote the polynomial

T(T — l).,.(T— n + l)/n!.

T For more details on this exercise, see K. W. Gruenberg, 11mm Note; in Math,
no. 143, Chapter 3, and R. Swan, J. qlgcbm, 12 (1969), pp. 535—601.

[95
ll. FREE LIE ALGEBRAS

If f(T) is a polynomial, let Af denote the polynomial f(T + I) — f(T)

ThenA(T)
0
=Al =0andA(T)
n
=(n —T l )ir»; 1.
(a) Letf s Q[T]. Show the equivalence of the following properties:
(i) f maps Z into itself.
(ii) f is a linear combination with coefficients in Z of the (D-

(iii) f(0) e Z and Afmaps 2 into itself.

(Argue by induction on deg(f), noting that deg(Af) = deg(f) — 1 if _
deg(f) a 0.)
A polynomial f satisfying the above properties is called binamial. The sum,
product and composition of two binomial polynomials is a binomial poly-
nomial.
([7) Let [I be a prime number and letf be a binomial polynomial. Show that
f maps the ring Z, ofp-adic integers into imelf (use the continuity off and
the fact that Z is dense in 2,).
(a) Let P be a set of prime numbers, let S, be the set of P-integers (§ 4,
Exercise 14) and let Z, = S; 1Z. Show that iff is a binomial polynomial,
f maps 2,, into imelf (apply (17) to the prime numbers [a not belonging to P).

5. Let F = F(X) be the Magnus group on X and let (1}) be its natural
filtration (no. 2). Show that the graded Lie algebra gr(I‘) corresponding to
this filtration is isomorphic to the Lie subalgebra g An(X) ofA(X). IfX 96 0,
this Lie algebra is not generated by its elements of degree 1 ; deduce that
(I‘n) is not the lower central series of 1".

fl 6. Let P be a set of prime numbers, let 8,, be the set of P-integers (§ 4»,
Exercise l4) and let Z? = S; 1Z.
(a) Suppose that, for all 5681,, the mapping [Ct->5]: is a bijection of K
onto itself; this is equivalent to saying that K can be given a ZP-algebra
structure.
With the notation of Exercise 5, show that, for all 355?, the mapping
a n—> 11a of 1" into itself is bijective and that the same is true for each quotient
I‘ll",l for n 2 I. If! e ZP and a E F (resp. a e l‘/l",l , u‘ is defined as in Exercise
16 of § 4; writing a in the form l + a, where a EA,(X), show that

a‘ = <1 + «r = .2 0“"- I

(Note that the coefficients (fl) belong to 21,, cf. Exercise 4-.)
(b) Suppose now that K = 21,. Lets be an integer al. The group
F“ = F/CcF is identified with a subgroup of I‘ll"c by means of the homomorph-

196
 memes

ism derived from g by taking quotients (notation of Theorem 2). The group
F/I‘L. is a P-torsion-free P—divisible group of class <r. Let F; be the P-satura—
tion of F” in l"/I‘c (§ 4, Exercise 14) and let i be the injection of F” into Ff,
The ordered pair (1', F?) is a P—ermlape of Fc (§ 4, Exercise 15). Deduce that
«my m'lpatmt gum]; ha: :2 Rem/clap: (note that every nilpotent group of class
<6 is the quotient of a group F“ for suitable X and use part (b) of Exercise
15 of § 4).
(c) If n < a, let Fi>.n be the intersection of Ff, with I‘n/I‘,; if n 2 c, we
write FE,“ = {a}. Show that F11”, is the P-saturation of C”F° in F/I‘c. The
filtration (F;.,.) is an integral central filtration of Ffs; let gr(F§) be the graded
module associated with this filtration. Show that, if n < r, the image of
grn(F§) in gr,.([‘) = AQIJX) is S;‘.L;(X) = L;,(X). Deduce that the Lie
algebra gr(F§) is generated by its elements of degree 1 and hence that
Ff“, = C"(Ff=) for all 7: (§ 4, Exercise 1). Show that the group Ff, is generated
by the x1" for x E X, .r 6 5; (note that the imagm of these elements in
gr,(Ff=) 2 Z?" generate the group gr,(Ff,) and apply Corollary 3 to Proposi-
tion 8 of Algebra, Chapter I, §6, no. 3).
(d) Let H be a Hall set relative to X (§ 2, no. 10) and let H(c) be the
subset of H consisting of the elements of length <5. For meH(c), let ¢o(m)
denote the image in F” of the basic commutator Mm) defined by m (of. no. 4,
Remark). Show that, for all w e F; there exists a unique element a 529“”)
such that

H «(mm
=nemn)

(Use the determination ofgr(F$) obtained above.)

7. We preserve the notation of Exercise 6.
(a) Let G be a nilpotent group of class <5 and let (i, G) be a P-envelope
of G. Let p: F” —> G be a surjective homomorphism (such a homomorphism
exists if X is suitably chosen) and let 5 be the corresponding homomorphism
of F; into (—3 (§ 4-, Exercise 15). The homomorphism 5 is surjective and maps
cur; = F;_,, onto 0-5. Show that 0‘5 1‘: the P—envelape qfi(C"G) in G and
that the Lie algebra me) is imgsed with 2P 8, gr(G).
(b) Deduce that G is P—divisible and P-torsion—free if and only if the
C"G/C"“G are.
11 8. Suppose that K = Z and that X is finite. For every integer k 2 0,
let (ck, n.) denote the basis of Ak(X).
(a) Let (Wuhan be a sequence of elements of F. Let wk'.(n) be the coeffi-
cient ofekm in the term ofg(w,,) EA(X) of degree It. Then:

g(w,,) = g wk.u(n):,‘vu for all n e Z.

197
11 FREE LIE ALGEBRAS

The sequence (10,) is called typical if, for every ordered pair (It, a) the func-
tion wk“: 71 »—> wk'a(n) is a binomial polynomial nfdtgnz <15, cf. Exercise 4. This
condition is independent of the choice of the bases (61m)- Show that it is
equivalent to the existence of a sequence (49“,, of elements of Aq(X) such
that m(ak) 2 k for all k and

£00» = k; n"a,g for allneZ.

(b) Show that, if (w,) and (10;) are two typical sequences, so are (111,?)
and (mung).
(5) Let w be an element of C“F for k 2 l and let f be a binomial poly-
nomial of degree <k. Show that (10’0”) is a typical sequence. In particular
if z e F, the sequence (1") of powers of z is typical.
(:1) Let (111,.) be a sequence of elements of F. Show that (w,,) is typical if
and only ifthere exist ya,y,,...,y,,,... in F, where y‘EC‘F for all i2 1
and
n
w,I E gag/'1' . . .yg") mod C"*1F
for all n e Z and all k 2 1. (Determine the y. from (w,.) by setting successively
n = 0,1, . . .. For example we = 1/0, w, = 110111, . . ..)
(e) Let H be a Hall set relative to X and let (w,) be a sequence of elements
of F. Let («(m, 11))".53‘“, be the family of integers such that

w, E I; “mum-'0 mod c
for all n e Z and all 5 2 1 (cf. no. 4, Remark). Show that (run) is typical ifand
only if, for all m e H, the function m—> a(m, n) is a binomial polynomial of
degree sl(m), where l (m) is the length of m.1‘
(f) A sequence (wn) is called l-Iypical if the ccrraponding functions wk‘,
are binomial polynomials of degree <1: —- 1. Show that, if (wn) is l-typical,
there existy. E C‘“F fori = 0,1, . . ., such that

w" E yog’l'. . .319) mod. CkuF

{orallneZandallk 21.
11 9. Let X be the set with two elements {15, y}.
(a) Show that there exism a sequence w,, 11/3,. . . of elements of F such
thatw‘eC‘Fforalliand
(" . . .w“<")
(xy " E :r"_t/"w22 modC‘HF

TFor more details on typical sequences, see M. Lazard, Annals: E. N. S., 71

(1954), pp. 101—190, Chapter II, § 1, 2.
198
 EXERCISES

for all n E Z and all i > 1. (Apply Exercise 8 (d) to the sequence of x‘"(x_1/)".)
Show that w; = y‘1(y‘1, x‘1)y E (x,y)'1 mod C’F.
(b) Let G be a nilpotent group of class <0 and let x, y be in G. Deduce
from the above the following formula (“Hull’iformula”);
c
(xy " = WU mm)“.
(c) Letpbeaprimenumbenshowthatthereexismu.EC‘F(2 < i <1: — l)
and w 5 OT such that
Qty)” = x’y’ug r . . u:_1w.

(Use (a), noting that (p) is divisible by 1: if 1 < i < p.) Let u') be the image
1

of w in gr,(F) = LflX) and let 1?), be the image of Liv in LEAK). Show that
u'), = ,(x,_t/), cf. Chapter I, § 1, Exercise 19. (Use the extension g of F to
A(X) and compare the terms of degree [I in g((xy)’) and g(z"y’u§ . A , u,’;_,m).
The first is equal to (x + y)” and the second is congruent mod. p to
x” + y” + rm. Hence the result.)
Show that there exist a E 03F and w’ 6 OT such that
(zya " = x’y’w’.
(Use the above formula for (:gr)’.) Deduce that in a nilpotent group of class
1; the [z-th powers form a subgroup.
(:1) Show that there exists a sequence ya, 11;, . . . of elements of F, where
0‘ e C‘F for all i and
n n
(m) a (mun?) . . . ultl mod CMF
for all n E N and all i 2 1. (Apply Exercise 8 (f) to the sequence of (x“,y).)
Deduce that, if p is a prime number, there exism t.eC‘F for 3 < i s p
and z e CH‘F such that
(fly) = (’6n - . : '51-
Show that the image 2, of z in 1.2:1(X) is (ad x)’(y). (Same method as for
(PM
11 10. Let p be a prime number and let G be a gmup with a real central
filtration (GK). Suppose that the relation xeGa implies z” e G”, in which
use the filtration (Go!) is called mm'md‘
(a) Show that the Lie algebra gr(G) associated with (G,) is such that
p.gr(G) = O and hence can be given an algzbm structure over F,.
(17) Let Es gru(G) and let x be a representative of E in G“. Show that the
image ofx’ in n(G) does not depend on the choice ofx (use Exercise 9 (0)).
If this image is denoted by Em, prove that E» EU” is a linear mapping of
gr,(G) into grw(G) and that (g + a)!" = E'" + Em + A,(E, g'). (Same
method.)

199
11 me LIE menus

Show that, if E e gr_(G) and 1) E gr,(G), then

[5"], 1)] = (ad E)”(n)~
(Use Exercise 9 (4).)
(a) Show that there exists one and only one p-mapping of gr(G) into itself
(Chapter I, § 1, Exercise 20) which extends the mappings E »—> Em defined
on the gru(G). With this structure, gr(G) is a Lie Mgzbm, said to be associ-
ated with the restricted filtration (Ga).
{I 11. (a) Let A be a filtered algebra satisfying the conditions of § 4-, no. 5 V
and let 1" = A“ n (l + A3). I‘ is given the filtration induced by that on A
(lac. 51-h, Proposition 2). Show that, if K is a field of characteristic 9 > 0,
(F1) is a rertrimd filtration (Exercise l0) and the embedding of gr(l") in
gr(A) defined in too. cit, Proposition 3 is compatible with the Lie p—algebra
structures on gr(I‘) and gr(A).
(b) Suppose that K = F,. We apply the above to the filtered algebra
A(K), in which F = F(X) is embedded by means of the homomorphism g.
The filtration (1",,) of l" induces a filtration (R) on F, which is a restricted
filtration. The Lie p-algebra gr(F) is identified with a Lie p—subalgebra of
A(X) containing X. Deduce that g-r(F) contains the free Lie p-algebra L(X,p),
cf. § 3, Exercise 4- (e).
(a) Let H be a Hall set relative to X; for every integer i, let H. be the set
of elements of H of length i. Let w e F and let on e 2“" be such that

w a 1—1 H d>(m)°‘n("‘) mod 0"“F

4-1 men
for all 1;, cf. no. 4. For all m e H, let l(m) denote the length of m and q(m, w)
the largest power of!) which divides «l(m), where i = l(w) (if 14m) = 0, we
make the convention q(m, w) = +00). Let N = Ing (l (m) .q(m, w)). Show that
me
the ¢(m)°‘l‘"” belong to F" and even to Fun if I (w) .q(m, w) > N. Show that,
if l(w) .q(m, w) = N, the image of ¢(m)‘n"’“ in n(A(X)) = A"(X) is a non-
zero multiple of 17mm“, where I; is the element of L(X) defined by 771 (§ 2,
no. 11). Now these elements are linearly independent (§ 3, Exercise 4);
deduce that the image of w in n(F) is #0 and belongs to L(X, [2), whence
the fact that gr(F) = L(X,p).
12. Let g be a Lie algebra over a field I: of characteristic p > 0.
(a) Let h be an integer <17 — l. g is said to satisfy Engel’: h-th condition if
(ad x)”(y) = 0 for every ordered pair (flag) of elements of 9. Show that this
is equivalent to
0;” ad(xou,) ° ' ' ' ° 3d(*om)) = 0
for x1, . . ., an, in 9 (apply the formula (ad 1:)” = O to the linear combinations
of the x,).

200
 EXERCISES

(b) Let at, g be elements of 9 such that A,(ax, by) = 0 for all a, b in 1:.
Show that, if M,” is the bihomogeneous component of A” of bidegree (r, r),
then A;"(x, y) = 0 for every ordered pair (7, r). Deduce (taking r = [I — l,
.\‘ = I) that (ad x)"1(y) = 0. In particular, if A,(x,y) = O for x, y in g, the
Lie algebra 9 satisfies Engel’s (p — l)-th condition.
(it) Let c be the centre of 9. Suppose that (ad x)” = 0 for all xeg. Show
that g/c satisfies Engel’s (p —- l)-th condition. (Show that A,(ad x, ad y) = 0
for x, y in g and apply (b) to the Lie algebra ad 9 = g/c.)
(4) Let G be a group such that z” = I for all x e G and let (Gm) be a real
central filtration of G. The filtration (G2) is remitted (Exercise 10) and gr(G)
is a Lie p—algebra whose p—mapping is zero. Deduce that gr(G) satisfies Engel’s
([1 — l)-th conditionif

§6

1. Show that exp(U).exp(V).exp(——U)= ”(2° —l—' (adU)"(V)) De-

duce that

H(U,H<v,—U>)= 2 —.(adU) (V)= (exp<adU>)(V).

”=0 71,

2. Show that H(U, V) = go%(ad U)"(H(V,U)). (Apply Exercise 1,

noting that H(U, H(H(V, U), —U)) = H(U, V).)

3. Let M(U, V) = H(—U,U + V). Then .
CXP(M(U, V)) = exv(-U) ~wP(U + V)-
Let M1(U, V) (resp. H1(U, V)) denote the sum of the bihomogeneous terms
of the series M (resp. H) whose degree in V is l.
(a) Show that

M1<U, V) = 2(51‘};!(adU)n(V) = (f(ad U))(V>

wheref(T) = (l — e")/T.
(Same argument as for the proof of Proposition 5. It can also be reduced
to Proposition 5 by using the identity
exp(U).exp(M(U,V)).aIp(—U) = exp(H(U + V, -—U)),
together with Exercise I.)

1' For properties of Lie algebras satisfying an Engel condition and their applica-
tions to the “restricted Burnside problem”, see A. Kostrikin, Izu. Akad. Nauk SSSR,
23 (1959), pp. 3734.
201
II FREE LIE ALGBBRAS

(b) Show that H(U, M(U, V)) = U + v. Deduce that

HAU, M1(U V)) = V

(—2")! tz“, where the b,“ are

(c) Letg(T) = l /f (T) = 1 + T/2 + “21—

the Bemom'lli number: (Fumtiom'qf a Real Variable, Chapter VI, § 1, no. 4). Show
that

(—2“) . heed U) W)
H.(U, V) = (g(ad U))<V) = V + an V] + Z—
Derive the first few terms of the expansion of H1(U, V):

Hm, V) = V + had U)<V) + 1% (ad U)=(V)— 7—20 (ad U)‘(V)

+fi(adU)°(V) —
(d) Show that the sum of the bihomogeneous terms of H(U, V) whose
degree in U is l is the series:

U + HU, V] + 21%”). be..(ad V)”'(U)

(Apply (0) to H(V, U) and use Exercise 2 to pass from H(V, U) to H(U, V).)
(e) Let W be another indeterminate and let X(U, V, W) be the sum of
the trihomogeneous terms of H(U,V + W) whose degree in W is 1. Show
that
X(U, H1(V’ W» = H:(H(U, V), W)~
(Use the identity H(U, H(V, W)) = H(H(U, V), W).) Deduce that
X(U, V, W) = (g(ad H(U, V)) °f(ad WNW).
11 4. Let X be a finite set and let I. =L(X). We give 1. the Hausdorff law
of composition (a,b)v—>aub, which we denote simply by a.b. If 26K,
a E L, we write a‘ = In.
(a) Let .9? be a Hall set relative to X. If m e 3?, let mL denote the corre-
sponding element of the Lie algebra L(X) (§ 2, no. 11) and m! the basic
commutator of the group L defined by m (§ 5, no 4', Remark). Ifm is of length I,
show that 7”,: ml, + p, where u is of filtration > I + I (argue byinduction
on I). Deduce that every element w of the group I. can be written uniquely 1n
the form
_ nun)
w _ m1; m". ’

202
 mncrsm

where «(711) E K and the product is convergent in the topological group L,

(b) Let P be the set of prime numbers and let a be an integer 21. Let
K = Q. Show that the group L/Lc is identified with the P-envelope Ff. ofthe
group F“ = F (X) /C‘F(X), cf. §5, Exercise 6.
(0) Take X to be a set of two elements {U, V}. Show the existence of two
families «(715), Mm) of rational numbers such that
_ ()
U+V—ml;lfm:r”‘

[U, V] = flung").
For example
U + V = U.V.(U,V)"f...
mw=mwm
(d) Let 9 be a nilpotent Lie Q—algebra with the Hausdorff group law. Let
u, v be in 9. Show that
u + u = flmx(u, 9)“)

mo=£lmomwu
where at and 5 are the families of rational numbers defined above (“inversion
9/ the Hamdorflformula”). (Use the continuous homomorphism 45: L —> g such
that (MU) = u, ¢(V) = u and note that it is also a homomorphism for the
Hausdorff group law.)
(2) Let G be a torsion-free divisible nilpotent group (i.e. P-torsion-free and
P-divisible). If u e G, t E Q, 11‘ is defined (§ 4-, Exercise 16). If u e G, v e G, we
define it + n and [11, v] by the formulae of (4') above. Show that G thus has a
nilpotent Lie nzbra structure and that the corresponding Hausdorff law is
the given group law on G (verify these assertions when G is a group Ff», cf.
(1:), and pass from this to the general case using homomorphisms of F; into G).
Let f be a mapping of G into a torsion-free divisible nilpotent group G’.
Show thatfis a (group) homomorphism if and only if it is a Lie algebra homo-
morphism. (The Hausdorfl law therefore defines an isomorphism of the “cate-
gory” of nilpotent Lie Qalgebras onto that of torsion-free divisible nilpotent
groups.)
5. Let 9 be a nilpotent Lie Q—algebra which we give the Hausdorff group
law, denoted by (x, y) H my.
(a) Show that, ifx e g, y e 9, then

x.y.z'1 = “2:0 $ (ad x)“(y).

(Use Exercise 2.)

203
II FREE LIE ALGEBRAS

(b) Let I) be a subset ofg. Show that b is a Lie subalgebxa (reap. an ideal) of 9
if and only if it is a saturated subgroup (resp. a saturated normal subgroup) of
the group 9. (Use the formulae of Exercise 4 (d) to pass from the group law to
the Lie algebra law.)
(a) Let I) be a subgroup of the group 3. Show that the P-saturation of b is
.1).
(11) Let b be a Lie subalgebra of 9. Show that the centralizer (resp. nor-
malizer) of b in the group 9 is the set of x e 9 such that (ad x)(b) = 0 (rap.
(ad ~)(b) C b)- '
(2) Show that the lower central series ofthe Lie algebra g coincides with that
of the group 9 and that the associated graded Lie algebra g'r(g) is the same
from the “group” point of view as from the “Lie algebra” point of view.
1T 6. Let G be a finitely generated torsion-free nilpotent group. Show that, if
n is sufficiently large, G can be embedded in the lower strict triangular group
of order 71 over Z. (Let (i, C) be a P-envelope of G, cf. § 5, Exercise 6; with its
canonical Lie algebra structure (Exercise 4), G is a finite-dimensional nil-
potent Lie Q—algebra. Apply Ado’s Theorem (Chapter I, § 7) to G and deduce
an embedding of G in a strict triangular group over Q. Pass from Qto Z by
conjugating by a suitable integral diagonal matrix.)

§7
1. Let g be a complete normed Lie algebra over K such that
| [*5 y] I < “I‘ll M
for x, y in 9. Let 6 denote the set ofx e 9 such that ][x][ < § log in For x, y in G),
h(x,y) e 0, cf. no. 2.
(a) We writef(T) = (l — a‘T)/T and g(T) = l/f(T). The seriesfcon-
verges throughout the complex plane and the series g on the open disc of
radius 21: (cf. Function: qfa Real Variable, Chapter VI, § 2, no. 3). Deduce that,
if z s 9, f(ad z) and g(ad z) are defined and that they are elements of $(g; g)
and inverses of one another.
(II) Let Dzh(x,y) be the second partial derivative of h at a point (x,y) of
G) x 9 (DWmntiable and Analytic Mani/”0141:, R, 1.6.2). It is an element of
3(9; 9). Show that
1321106 y) = 3(ad h(~.y)) °f(ady)-
(Use Exercise 3 (z) of § 6 to show this formula when x and y are sufficiently
close to zero and pass from this to the general case by analytic continuation.)
Show that the formula
f(ad Maw) ° Dsh("').'l) =f(ady)
is valid at eve_ry point of the domain of absolute convergence of the formal
power series H (cf. Proposition 1).

204
 EXERCISES

§3
We suppose that the residual characteristic p of the field K is >0. We
denote by I); the ring of the valuation v of K.
1. (n) An element 1: ofK is called admisible ifu(1r) = 0 and 1:”-l E —1
(mod. wax).
(Note that 1:"1/p 5 DK since v(1:) = 9.)
Show that there exist fields K satisfying the conditions of the paragraph and
containing an admissible element (adjoin to Q, a (p — l)-th root of —p).
(b) Let 1: be an admissible element of K. Consider the following formal
power series with coefficients in K:

“(In = 11::(RU) = ”Z; 1r"'1U"/n!,

1,.(U) = 111mm) = 21 (—n)"'1U”/n.

Show that these series are inverse: of one another and that their coeflicienm
belong to 0K (use Lemma 2).
(a) Let {U, V} be a set with two elements and let H(U, V) be the Hausdorff
series. We write 1
H,<U, V) = ; Harv, M.
Then H..(U, V) e Law, V».
Show that
Haw, V) = lx(‘z(U) + 0::(V) + N!x(U)~':(V))‘
Deduce that H" e L,‘({U, V», i.e. that the coefficients of H, belong to 0,;
hence another proof of Proposition 1.
(d) Let in, I, and H, denote the series obtained by reducing en, l,K and H“
modulo 1:05; their coefficients belong to the ring oK/nox.
Show that i"(U) = U -— U’ (note that 0,,(n) < (n — l)8 if n aé l, p).
Deduce that
5,,(U) =U +U’ +°--+U" +...
and that
MU, V) = AU) + 5:(V) - (MU) + MW)”:
or also

H.(U, V) = U + v — AA“; U»: n20w“)

where A, is defined by the formula
Ap(U, V) = (U + V)’ — U’ — V", cf. Chapter I, § 1, Exercise 19.

205
n mus): LIE means

In particular, 1:1,‘(U, V) belongs to ipflflU, VD and its homogeneous compo-

nent of degree [7 is v—AP(U, V). Then
H.(U. V) = HAV: U)
H.(U. fiAV, W)) = 1341:1413, V), W) in ip,({U, V; W}),
where W is a third indeterminate.
(e) Let C(U, V) = H(—U, H(—V, H(U, V))) (“Hausdorfl'commutator”).
Then exp(C(U, V)) = exp(—U) exp(—V) exp(U) exp(V). We write

C,,(U, V) = icocu, RV).

Show that the coefl-iciems of C,(U, V) belong to noK (use the fitct that
HAU, V) = Hm, U))«T
2. We know (§ 6, Exercise 3) that if n 2 2 the component of H(U, V) of
bidegree (n, 1) is $17,.(ad U)“(V), where I),L is the n-th Bernoulli number.

Deduce from this and Proposition 1 the inequality

”pm/n!) < n/(P — 1)~

Recover this result by means of the Clausen-von Smudt Theorem (Functions «y’a
l Variable, Chapter VI, §2, Exercise 6) and show that there is equality if
and only ifS(n) =p — 1.
3. Suppose that K contains a primitive p-th root of unity w. We write
1: = w — 1. Using the formula
w‘— l =7:(l +w+-~-+w"1)
show that H(w‘ - l) =v(1c) forl < igp — 1 andthat
10—1
1
E 1' (mod. 1:).
713

7-1
Deduce, by means of the formula 1) = 1:! (w‘ —- 1), that 1: is adminiblz
(Exercise 1).
1] 4. Let a be an integer 2 l and let P be a set of prime numbers containing
all prime numbers $5. Let Z? = S; 1Z (§ 4, Exercise 16). Show that the terms
of degree $0 in the Hausdorfi series H(U, V) belong to LzF({IJ, V}). Deduce

TFcr more details on this exercise, see M. Lazard, Bull. 50:. Math. France,
91 (1963), pp. 435—451.

206
 EXERCISE

that, if g is a nilpotent Lie ZP-algebra of class $5, the law of composition

(u, u) >—> H(u, 0) makes 9 into a P-torsion-l'ree P—divisible nilpotent group of
class $6. Show that conversely every group with these properties can be ob-
tained by this process. (Use “the inversion of the Hausdorff formula”, cf. §6,
Exercise 4, and show that the exponents a:(m), Born) which appear there belong
to 2,, when 1(m) < a.)
In particular, every p—group of order [1" and class <11 can be obtained by
means of the Hausdorff law from a nilpotent Lie Z—algebra of class <p with
1)" elements. (Take P to be the set of prime numbers distinct from [1.)

APPENDIX
1. Let (I),l be the cyclotomic polynomial of index 1: (Algebra, Chapter V,
§ 11, no. 2). Using the formula.
x» — 1 = H (FAX),
“In
show that
(1),.(X) = 1;; (ad _ ”um.

(Apply the Mobius inversion formula to the multiplicative group of the field
(109-)
2. Let D be the total algebra of the monoid N‘ (Algebra, Chapter 111, § 2,
no. 10). Ifn E N“, let n“ denote its image in D, so that l” = land (7111:)“ = 71m
for n, m in N*.1' Every element f of D can be written uniquely as a series

”Z: ann", where a,I e K.

a
(a) Letf = "Z, and” be an element ofD. Show thatfis invertible in D ifand
only iflz1 is invertible in K. In particular, if K is a local ring, so is D.

(b) We write E = "2‘ n“ and y. = .121 y.(n)n“’. Show that C and y. are
inverses of one another. Deduce that, if 5 = 25(71):“ and t = 21007:" are
two elements of D, the relations .r = Li and t = y"; are equivalent (variant of
the Mobius inversion formula).
(6) Let P be the set of prime numbers. Show that the family of (l -— p“) for
p E P is multipliable in D and that

y.=Z(l—p")
per
and :JI ‘
use 1 _ 9°"
1 We often write —-.r instead of a; we then say that the elements of D are the
Dirichletfimal power rain with cocflicients in K.

207
 CHAPTER III

Lie Groups

Throughout the chapter, K denotes either the valuedfield R ofreal numbers, or the valued
field C ofcomplex numbers, or a nan—discrete conialete ultrametric commutative field. We
assume that K is ofcharacteristic Ofiom § 4 onwards, that K = R or C in § 6, that K
is ultrametric in § 7. Unless otherwise ”motioned, all the manifolds, all the algebras and
all the vector spaces considered are over K. Recall that, when we speak y” a manifold of
classC’,IeNK,thatirr=mifKaéRand1<r< mifK=R.
The conventions on norms, nonnable spaces and normed spaces are those
of Diferentiable and Analytic Manyblds, R.
Recall that a normable algebra over K is a (not necasarily associative) algebra
A over K, with a topology .7 which possesses the following properties:
(I) .7 can be defined by a norm;
(2) the mapping (:4, y) )—> any of A x A into A is continuous.
The group of bicontinuous automorphisms of A is denoted by Aut(A).
Every finite-dimensional algebra over K is a normable algebra with the cano-
nical topology. A normed algebra over K is an algebra A oyer K with a norm such
that “any“ < “x“ H y“ for all x, y in A ; the algebra A with the topology defined by
this norm is a normable algebra. If A is a normable algebra, there exists a
norm on A defining its topology and making A into a normed algebra,
If G is a group, ea, or simply e, denotes the identity element of G. For
geG, ~{(g), 8(g) and Int(g) denote the mappings g’ Hgg’, g’ Hg’g‘l and
g’ Hgg’g“ of G into G. Iff is a mapping ofG into a set E,fdenotes the
mapping g»f(g") ofG into E.

§1. LIE GROUPS

1. DEFINITION OF A LIE GROUP

Let G be a set. A group structure and an analytic K-manifold structure on G

are called compatible if the following condition holds:
(GL) The mapping (g, h) v—> gh" of G X G into C is analytic.

209
III LI'E GROUPS

DEFINITION l. A Lie group over K is a set C with a group simmer: and an analytic
K—manifold itmoture such that these two structures are compatible.
A Lie group over R (resp. C, Q12) is called a real (resp. complex, [z-adic) Lie
group. .
Let G be a group with an analytic manifold stmcture. For g, h, g”, ha in G,

(l) gh“ = (gob; ‘)ho((g6 g)(h3 111) “MS 1.

It follows that G is a Lie group ifand only if the following three conditions hold:
(GL,) for all go E G, the mapping g »—-r gag of G into G is analytic; '
(GL2) for all go 5 G, the mapping g -—> goggo' ‘ ofG into G is analytic in an open
neighbourhood of a;
(GL3) the mapping (g, h) n—> gh" of G x G into G is analytic in an open
neighbourhood of (e, e).
Let G be a Lie group. For all g e G, fig) and 8(g) are automorphisms of the
underlying manifold of C. It follows that this manifold is pure (Difl’emxtiable
and Analytic Manifoldr, R, 5.1.7). In particular, the dimension of G at g is equal
to dim G for all g e G (recall that dim G is an integer >0 or +00).
Since an analytic mapping is continuous, a Lie group is a topological group
with the underlying topology of its manifold structure. Let G be a set. A topo-
logical group structure and an analytic K—manifold structure on G are called
compatible if the group structure and the manifold structure are compatible and
the topology on G is the topology underlying the manifold structure.
Lemma 1. Let G be a Lie group, U an open neighbourhood of e, E a complete normed
space and ‘i’: U —> E a chart of the manifild G. There exists a neighbourhood W g" e
contained in U sueh that 4) | W is an isomorphism ofW (with {he righ! uniform xtructure)
onto MW) (with the uniform strurture induced by that an E).
It can beassumed that¢(e) = 0. Let U' = ¢(U)i Let 4;: U’ —> U be the inverse
mapping of 4). Let V be a symmetric open neighbourhood of e such that
V2 C U and letV' = ([>(V).We define mappings 01, 82 ofV’ x V' into V’ X U’
as follows:
010%) = (x, ¢(¢(x)¢(y)‘1))
920611) = (x,<l>(¢(y)"¢(*)))-
It is immediately verified that 92(9,(x,y)) = 0,(02(x,y)) = (x,y) for x, y
sufficiently close to 0. On the other hand, 6, and 02 are analytic and hence
strictly differentiable at (0, 0). Therefore (Diflerenliable and Analytic Manyoldr,
R, 1.2.2) there exist a neighbourhood W' of 0 in V' and constants a > 0,
b > 0 such that
a(“"1 — “all + |l¢(¢(xx)¢(y:)") - <l>(¢(xz)¢(ya)‘1)|l)
$ "*1 — “a" + “g: *h”
s b(Ila: — JWall + |l¢(¢(n)¢(yx)") — ¢(¢(12)¢(yz)“) ID
210
 DEFINITION OF A LIE cnoup § 1.1

for all :5, :2, y,, y2 in W’. Writing at, = It, = y,, we obtain

(2) all¢(¢(xl)¢(y:)“)ll < in -1I:|| < blli>(¢(*1)¢(y1)")llc

For 8 > 0, let No be the set of ordered pairs (1, y) EW’ x W’ such that
”x — y” < 8. The N, form a fundamental system of entourages in W’. We
write W = ¢(W'). Let M‘5 be the set of ordered pairs (14, 1)) 6W x W such
that "MW—l) II $ 3. The M5 form a fundamental system of entourages in W
with the right uniform structure. But relation (2) proves that

Na C (o X ti>)(M..-Its), (d> >< ¢)(Mo) c N»

and hence W has the property of the lemma
PROPOSITION l. A Lie group is a complete metrizablc topological group.
Since e admits an open neighbourhood homeomorphic to an open ball of a
normed space, e admits a countable fundamental system of neighbourhoods
whose intersection is {e}. Hence G is metrizable (General Topology, Chapter III,
§ 1, Corollary to Proposition 2 and Chapter IX, § 3, Proposition 1). By Lemma
1, there exists a neighbourt ofe which is complete under the right uniform
structure and hence G is complete (General Topology, Chapter III, § 3, Propo-
sition 4.
PROPOSH'ION 24 Let G be a Lie group.
(i) If K = R or C, G is locally connected.
(ii) [I K i: dixtinct from R and C, G i: zero-dinmional (General Topology,
Chapter IX, §6, Definition 5).
(iii) Suppose that K it locally compact. For G to be locally compact, it is necessary
and .tufia’ent that G befinile-dimemional.
(iv) If G is generated by a .mbspoce whore topology admits a countable base, then the
topology on G admit: a countable bare.
Let U be a neighbourhood etc. There exists an open neighbourhood U1 of e
contained in U and homeomorphic to an open ball of a normed space E over
K. If K = R or C, Ul is connected, which proves (i). Suppose that K is ultra-
metric. There exists a neighbourt U2 of e which is closed in G and such
that U2 C U). Then there exists a neighbourhood Ua ofc such that Us C U2
and U3 is open and closed relative to U1. Then U3 is closed relative to U2 and
hence G and open relative to U1 and hence G. This proves (ii). For G to be
locally compact, it is necessary and sufficient that E be locally compact; if K
is locally compact, this amounts to saying that E is finite-dimensional (Topo-
logical Vector Spaces, Chapter I, § 2, Theorem 3), whence (iii). Suppose that G is
generated by a subset V and let W = V U V“; then
G =WUW’UW3U...;
if there exists a sequence dense in V, we see that there exists a sequence dense

211
ll! LIE GROUPS

in G and, as G is metrizable (Proposition 1), the topology on G admits a

countable base.
COROLLARY. If K = R or C and G is tonnzttnl andfinite—dimenrional, then G 2';
locally connected and locally compact and its topology admit: a countable base.
Lemma 2‘ Let X In a manifold of class C', e a point of X, U and V open neighbour-
hood: of e and In a mapping of clan CT of U X U into X satisfying the following
conditions:
(a) m(e, x) = m(x, e) = xfor alle;
(h) V c U, m(V x V) C U and m(rn(x,y), z) = m(x, m(y, 2))fi77 all x, y, '
z in V.
Then thm exists on open neighbourhood W of e in V and an automorphixm 6 If
the manifold W such that 0(a) = e, 0(0(x)) = x and m(x, 0(x)) = m(6(x), x) = e
for all x e W‘
m(e, y) = y for all y e U and hence, by the implicit function theorem, there
exists an open neighbourhood W1 of z in V and a. mapping 9, of class C' of
W1 into V such that 0,(:) = e, m(x, 01(x)) = a for all x E W1. Similarly, there
exists an open neighbourhood W, of a in V and a mapping a, of class C’ of
W, into V such that 02(5) = z, m(0,(x), x) = z for all x 5 W5. For x 6 W1 n W3.

920) = "492(1), ‘) = “(92(3), mot, 9100))

= W(M(0z(z),~), 0100) = "IO: 91(1)) = 910‘)
Let 0(x) be the common value of 0,(x) and 02(x) for x (;W1 (1 W2. Let W be
the set of newl n W2 such that 0(x) SW1 (1 W5. The set W is open. For
x e W,

0(900) = "Wot: 9(1)), 903(3)» = ”'06, MW), 8(900») = “(A e) = x

and hence 0(x) eW. We see that 0 [W defines an automorphism of the
manifold W.
PROPOSmON 3. Let X be an analytic manifold and m an analytic associative law Qf
compoxition on X admitting on identity element. The set G of invertible elements 13f X i:
open in X and G is a Li: group with m | (G x G) and the nmny’old :trurture induced
by that on X.
By Lemma 2, G is a neighbourhood of the identity element. For all g E G,
the mapping x»—> m(g, x) is an automorphism of the manifold X. Hence the
image of G under this mapping is a neighbourhood ofg, obviously contained in
G. Therefore G is open in X. Clearly conditions (GL,) and (GL,) hold. Con-
dition (GL3) holds by Lemma 2.
Examfllcs (If Lie groups
(1) Let E be a complete normable space over K. The mapping
(my) H x — y of E X E into E is continuous and linear and hence analytic.

212
 uonpmsus or LIE GROUPS § 1.2

Hence E, with its additive group and analytic manifold structures, is a Lie
group.
In particular, K is a Lie group.
(2) Let A be a complete normable unital associative algebra over K. The
multiplication (x, y) r—>xy of A x A into A is bilinear and continuous and
hence analytic. Proposition 3 shows that the group A“ of invertible elements of
A is open in A (which also follows from General Topology, Chapter IX, §3,
Proposition 13) and that A* is a Lie group.
For example, let E be a complete normable space over K and let A = 3’(E)
(General Topology, Chapter IX, §3, Proposition 5). Then A* is the automor-
phism group GL(E) of E. This group therefore has canonically a Lie group structure
over K. More particularly, GL(n, K), with the manifold structure induced by
that on Mn(K), is a Lie group. For n = l, we see that the multiplicative group
K* is a Lie group with the manifold structure induced by that on K.
(3) Let G be a Lie group over K. Let K’ = R or C or a non-discrete com-
plete ultrametric field and a an isomorphism of the valued field K’ onto a
valued subfield of K. Then the group G, with the K'-manifold structure
obtained by restriction of scalars, is a Lie group over K', which is said to be
derivedfrom the Lie group G by rextriction ofsealar: (from K to K’ by means of a).
For example, every complex Lie group has canonically a. real Lie group
structure. Again, with every complex Lie group G is associated a complex Lie
group called the conjugate of G, derived from G by means of the automorphism
z >—> 2 of C.

2. MORPHISMS 0F LIE GROUPS

DEFINITION 2. Let G and H be Lie groups. A Lie group morphirm of G into H (or
simply a morphism of G into H if no eorjusion can arise) it a mapping of G into H
whr‘eh it a group homomorphism and 1': analytic. The automorphism group of G is
denoted by Aut(G).

The identity mapping of G is a morphism. The composition oftwo morphisms

is a morphism. Iff: G —> H andf’: H —> G are two inverse morphisms, f and
f’ are Lie group isomorphisms.

Examples. (1) Let G be a Lie group. For all x e G, Int(x) is an automorphism

of the Lie group G.
(2) Let G be a Lie group. Let G" denote the opposite group to G, with the
same manifold structure as C. It is immediate that G" is a Lie group (called
the opposite Lie group to G) and that the mapping g >—> g“ is an isomorphism of
the Lie group G onto the Lie group G".
(3) Let G be a Lie group and E a complete normable space. An analytic
linear representation of G on E (or simply a linear representation of G on E when

213
 In LIE GROUPS

no confusion can arise) is a morphism of the Lie group G into the Lie group
GL(E), in other words an analytic mapping 1: of G into GL(E) such that
7:(gg’) = 1:(g)1r(g’) for g, g’ in G. Suppose that E admits a. finite basis
(43,, e,, . . ., en) over K; let (cf, ta“, . . ., e:) be the dual basis; let p be a homo—
morphism of the group G into the group GL(E); then the following conditions
are equivalent:
(i) p is an analytic linear representation;
(ii) for all x e E and x’ e E’, the function g >—> (p(g)x, x’) on G is analytic;
(iii) for all 1' and j, the function g >—> <p(g)e., ef) on G is analytic. _
For the implications (i) 2 (ii) a (iii) are clear. On the other hand, the
functions u >—> (ueb cf) form a coordinate system on J(E); hence their res-
trictions to GL(E) form a coordinate system on GL(E), whence the implication
(iii) = (i).
Let G be a real Lie group, E a real complete normable space and p a
homomorphism of the group G into the group GL(E). We shall see in § 8,
Theorem 1 that, if p is continuous (when GL(E) has the topology derived
from the norm on L(E)), then 9 is analytic. But note that this notion of
‘ ‘ is "" from that ‘J ed in ' 6 ' , Chapter VIII,
§2, Definition l (ii) (Exercise 1).

(4-) Let G be a real Lie group and E a complex complete normable space. An
analytic linear represenlafian of G on E is a morphism of G into the underlying
real Lie group of GL(E).
PROPOSITION 4-. Let G and H be Lie group: andf a homomorphism 0f the group G
into the group H. Forf to be analytic, i! is necessary and :uficient that there exixl a non-
empty open subset U of G me}: thatf | U i: analytic.
The condition is obviously necessary. Suppose that it holds. For all x0 E G,
f(xox) = f(are) f(x) for all 2: EU and hence f | xoU is analytic. But the sets
xoU, where In E G, form an open covering of G.
Remark. Iff is an immersion at e (resp. a submersion at e), clearly f is an
immersion (resp. a submersion).

5. LEE SUHGROUPS
Let G be a Lie group and H a subgroup of G which is at the same time a sub—
manifold of G. Then the mapping (x, y) >—> xy" of H x H into G is analytic
and hence the mapping (:4, y) >—> xy“ of H x H into H is analytic. (Difl'eren-
liable and Analytic Manifolds, R, 5.8.5). Thus H, with the group and manifold
structures induced by those on G, is a Lie group.
DEFINITION 3. Let G be a Lie group. A subset H uf G is called a Lie rubgmup if
H is a rubgnmfi and a .mbmanifold of G.

214
 max-mum vacuum or [m onoups § 1.4

An open subgroup of G is a Lie subgroup of G. In particular, if G is a real

or complex Lie group, its identity component is a Lie subgroup of G.
PROPOSITION 5. Let G be a Lie group and H a Lie mbgmup of G.
(i) H is dosed in G.
(i) The mnom'eol injection g” H into G i: a Lie group morphism.
(iii) Let L be a Lie gnmp audfa mopping q into G such thatf(L) C H. For
f to be a morphism of L into H, it is necessary and mfieient thatf be a morphism of L
into G.
By Difl'erentz'uble and Analytic Manifoldr, R, 5.8.3, H is locally closed. Hence H
is closed (General Topology, Chapter III, § 2, Proposition 4). Assertion (ii) is
obvious. Assenion (iii) follows from Diflerentiuble and Analytic Mani/011k, R,
5.8.5.
PROPOSITION 6. Let G be a Lie group and H a subgroup of G. For H to be a Lie rub-
group of G, it it necessary and :uflieient that there exist a point h e H and an open neigh-
bourhood U of h in G such that H n U is a rubmunyold y’ G.
The condition is obviously necessary. Suppose that it holds. For all h’ E H,
the translation y(h’h'1) is an automorphism of the manifold G and maps the
submanifold H n U of U into the submanifold (h’h‘lH) fl (h’h“U) of
h’h'lU. As h’h‘1H = H and h’h"U is an open neighbourhood ofh’ in G, we
see that every point of H has an open neighbourhood V such that V n H is a
submanifold of G. Hence H is a submanifold of G.
Let G be a Lie group and H a Lie subgroup ofG. If L is a Lie subgroup ofH,
L is a Lie subgroup of G by Difl'erentiable and Analytic Manifolds, R, 5.8.6. Let
M be a Lie subgroup of G such that M C H. Then M is a Lie subgroup of H,
for the canonical injection of M into H is obviously an immersion.
Let I: be a non-discrete closed subfield of K. A Lie k-subgroup of G is a Lie
subgroup of the underlying Lie k-group of G.
Remark. If ‘ ' " “" is r‘ " b “1 i " ifold" in L‘ E ' '
3, we obtain the definition (3k quasi—subgroups of G. (For finite-dimensional
G, the Lie quasi-subgroups are just the Lie subgroups.) Suppose that K is of
characteristic 0. Proposition 5 remains valid with the same proof for Lie
quasi-subgroups. Proposition 6 remains valid with the same proof, replacing
“Lie subgroup" by “Lie quasi-subgroup” and “submanifold” by “quasi-
submanifold”.

4. SEW-DIRECT PRODUCTS 0F LIE GROUPS

Let I be afinite set and (1%),“ a family of Lie groups. The group and manifold
structures on L = 4 a I L‘ are compatible and L thus has a Lie group structure.
L is called the product Lie group of the family of Lie groups (L,),E,.
Let L and M be Lie groups and a a homomorphism ofL into the automor-
phism group of the group M. Let S be the external semi—direct product of L by
M relative to 6 (Algebra, Chapter I, §6, no. 1, Definition 2).

2l5
III LIE GROUPS

PROPOSITION 7. If the mapping (m, l) ~—>a(l)m of M x L into M is analytic, the

group S, with the product manifold .rtrurture of M and L, i: a Lie group.
For l, l' in L and m, m’ in M,
(m, l)(m', l’)'1 = mlt‘lm"1 : m(a(ll'‘1)m"‘)ll"1
= (m(6(”"‘)m"’)) ll"‘)
whence the proposition.
If the conditions of Proposition 7 hold, the Lie group S is called the (external)
semi-dim! produet Lie group if L by M relative to u. '
Clearly the canonical injection of L (resp. M) into S is an isomorphism of L
(resp. M) onto a Lie subgroup of S which we identify with L (resp. M). The
canonical mapping of S onto L is a Lie group morphism.
Conversely, let G be a Lie group and L, M two Lie subgroups such that
the group G is (algebraically) the semi—direct product of L by M(Algebm,
Chapter I, §6, no. I). We write 6(1)»: = lml'l for l e L and m E M. Then 0'
satisfies the conditions of Proposition 7. We can therefore form the semi-direct
product Lie group S of L by M relative to a. The mapping j: (m, l) ~—> ml of S
onto G is a group isomorphism and is analytic. Ifj is a Lie group isomorphism,
the Lie group G is called the (internal) semi—direetproduet ifL by M and S and G are
identified. For all g E G, we write g = p(g)q(g), wherep(g) E M and q(g) E L.
For the Lie group G to be the semi-direct product of L by M, it is necessary
and sufficient that one of the mappings p: G —> M and q: G -> L be analytic,
in which case both are analytic; or alternatively, it is necessary and sufficient
that T,(G) be the topological direct sum ofT,(M) and T,(L) (for, if this con-
dition holds, j is e'tale at es).
Example. Let E be a normable space, G = GL(E), T the translation group of E
and A the permutation group of E generated by G and T. The gmup A is
algebraically the semi-direct product of G by T. (If E is finite-dimensional, A
is the afiine group ofE, cf. Algebra, Chapter II, § 9, no. 4). Let a be the identity
linear representation of G on E and S the external semi-direct product of G
by E relative to at For all xe E, let 1,, be the translation of E defined by x.
The mapping (x, u) >—> t, o u is isomorphism (D of the group S onto the group A.
The mapping (x, u) r» 0(a)): = u(x) of E x $03) into E is continuous and
bilinear and hence analytic; its mtriction to E x G is therefore analytic.
Thus the group S, with the product manifold structure of E and G, is a Lie
group. We transport this structure to A by means of (I). Then A becomes a Lie
group, the internal semi~direct product of G by T as a Lie group.
PROPOSITION 8. Let G and H be Lie groups, p: G —> H and J: H —> G Lie group
marphirm: rue}! thatp o s = idH audN = Kerp. Then N is a Lie subgroup ofG, .v i:
an irnmorphim: g” H onto a Lie subgroup afG and the Lie group G 1': the internal semi-
direet produet qf:(H) by N.

216
 QUOTlENT or A MANIFOLD m: A LIE GROUP § 1.5

T,(p) a T,(s) = idma, and hence[I (recp. I) is a submersion (rcsp. an immer-

sion). By Difermtiable and Analytic Manifalds, R, 5.10.5, N is a Lie subgroup of
G. On the other hand, I is a homeomorphism of H onto :(H) and hence x is an
isomorphismofH ontoaLie LB ruofG (Di t “ " andAnalytic "’ "’1
R, 5.8.3). Finally, for all gEG, g = (sop)(g).n for some nEN; as so]: is
analytic, the Lie group G is the semi-direct product of :(H) by N.

5. QUOTE“ OF A MANIFOLD BY A LIE GROUP

Let G be a Lie group, X a manifold of class C' and (g, x) r—> gx a. law ofleft
operation (Algebra, Chapter I, § 5, no. 1) of class C" of G on X. For all g E G,
let 1( g) denote the automorphism x r—> gx of X defined by g. For all x e X, let
9(x) denote the orbital mapping g I» gx of G into X defined by x. Then

(3) 90¢) = 9(gx) 08(1:), 9(1) = 1(g)° 9(1) °Y(g")

forallgeGandxeX. Hence
(4) T.(P(*)) = T.(9(gx)) ° T.(3(g))
(5) T4900) = T~(r(g)) °T.(9(x)) ° T.(7(g"))
Pnorosmon 9. Let x e X amigo e G.
(i) If p(x) it an immersion (resp. a submersion, a mbirnmemon) at go, then,
for all g e G, p(gx) i: an immerrion (resp. a mbmminn, a mbtmem'an).
k(ii) If p(x) is qfrank I: at go, lhenfnr all g e G, p(gx) u gamma! rank equal to

kThis follows immediately from formulae (4) and (5) since T,,(8(g)) T,‘(-r(g))
and T,(y(g‘1)) are isomorphisms.
Comm“. Let x e x. IfK i: qfcharacterislic o and x irflnile—dinwnfianal, pm
i: a whimnmdon. [ffimher p(x) is injective, p(x) is an immem'an.
This follows from Proposition 9 and serenliable and Analytic Manifolds, R,
5.10.6.
Observe that, ifn denotes the mapping (3, x) H gr of G x X into X, then,
{or g E G, x s X, 14 e T,(G), v e T,‘(X),

T(F.z)(7l)(u) 1)) = Tw.x)("l)(") 0) + T<y.x)(n)(0. 3)

that is
(6) T:y.x)(n)(u, v) = T4900)" + Tx(1(g))v-
PROPOSITION 10. Let G I): a Lie group and X a manifiild 9f class C’ with a law if
12]} operatian 4y” elm: C' of G an X. Support that:
(a) the group G 0119mm [mpnly andfreely on X;
(b) for all x EX, p(x) i: an immersinn (which is a cameqmnce qf (a) if K it (f
characterixtit 0 and X isfinilz—dimmsiwml).

217
to ma GROUPS

Tlun the equivalence ”lotion defined by G on X irregular (Difiermtiabl: and Analytic

Manifolds, R, 5.9.5). There exist: on the quotient rel X/G am and only one manifold
rhuotare moh that the canonical mapping 1:: X —> X/G is a submersion. The andnlying
topology of thir manifold structure is the quotient topology of that on X; it i: Hawdorfi
Finally, (X, G, X/G, or) is a principal leflfihn bundle.T
Let 0 be the mapping (g, x) r—> (x, gx) of G x X into X x X. This mapping
is of class C’. We show that it is an immersion. For a eT,(G) and v E TAX),
by (6).
(7) T(a.x)(0)(u, 1') = (”,Tg(9(x))u + Tx(’r(£))v)- .
But T,(p(x)) is injective by hypothesis (b) and hence T”, J0(0) is injective. Its
image is the topological direct sum of the subspace Hm: Consisting of the
vectors (a, Tx(r(g))u) for o e T,(X) and the subspace
Inna: = {0} X Ty(9(*))(To(G))~
By hypothesis (b), T,(p(x))(T,(G)) admits a topological supplement J,_,, in
T,~(X). Hence the image of T“, MO) admits the topological supplement
{0} x J“? Hence we have proved that 0 is an immersion ofG x X in X x X.
As G operates freely on X, 0 is injective. Let C be the graph of the equiva-
lence relation R defined by G on X. As G operates properly, 6 is a homeo-
morphism of G x X onto C (General Topology, Chapter I, § 10, Proposition 2).
By Dtjfinntiahle and Analytic Manifolds, R, 5.8.3, it is a submanifold of X x X
and 6 is an isomorphism of the manifold G x X onto the manifold C. The
tangent space T0,. ",(C) is identified with
T(v.x>(e)(T(v.x)(G X X» = Home I“ C T(x.rx)(x X X)-
Let prl and pr, be the canonical projections of X X X onto the two factors.
It is immediate that T“, ,x,(pr,) maps Hm x onto Tx(X) and that the kernel of
T(,,‘,x)(pr1) |T(",,)(C) is 1",. Thus pr1 I C is a submersion of C onto X.
By Dzfinmiable and Analytic Manifolds, R, 5.9.5, R is regular. By definition,
there therefore exists on the quotient set X/G one and only one manifold
structure such that 1: is a submersion. The underlying topology of X/G is the
quotient topology of that of X (Dijfirmtiable and Analylio Mani/77bit, R, 5.9.4).
This topology is Hausdorff (Control Topology, Chapter III, § 4, Proposition 3).
For all h e X/G, there exist an open neighbourhood W of h and a morphism
o': W ——> X such that 1: a o = idw (Difnenliablr and Analytic Manifolds, R,
5.9.1). Let d: be the bijection (g, w) Hga(w) ofG x W onto 1:1(W). It is of
class C’. Then 1c(gc(w)) = w and
9"(6(W),g6(W)) = (g, 600))
T The principal fibre bundles defined in Diflfimm'able and Analytic Manifizldr, R,
6.2.1 are principal right fibre bundles. The definition of principal left fibre bundles
can be deduced from this in an obvious way.

218
 noncomous spams AND Qumran-r onoues § 1.6

and hence the inverse bijection old) is of class 0. Clearly (12(gg', w) = g¢(g', w)
for w e W, g e G, g' e G. Hence (X, G, X/G, 1:) is a principal left fibre
bundle.
Remark. With the above hypotheses, further let H be a manifold of class C’
and (x, h) >—> m(x, h) a mapping of class C’ of X x H into X such that
m(gx, h) = gm(x, h) for: e X,g e G, h e H. Letnbe themappingof(X/G) x H
into X/G derived from m by taking quotients. We show that n i: of class (3'.
Consider the diagram
x x H—">X
xx] x

(X/G) xH—"—->X/G
It is commutative, 1: o m is of class C' and 1: x l is a smjective submersion; it
then suffices to apply terenfiable and Analytic Mang'flzlds, R, 5.9.5.
Let G be a Lie group, X a manifold of class C' and (g, x) >—>zg a law of
right operation of class C' of G on X. Let 1:(g)x = p(x)g = xg for geG,
Jr 6 X. Then this time

(3’) 9(2) = We) no“), 90¢) = 1’(g) ° 906) ° 8(a)

and hence

(4’) T.(9(~)) = T.(9(xg)) °T.(Y(g"))

(5’) T.(P(~)) = Tx(1'(g)> °T:(9(~)) ° Ty(3(g)).
On the other hand, if 1; denotes the mapping (g, x) n—> xg of G x X into X,
formula (6) remains valid. Proposition 9, its Corollary and Proposition 10
remain equally true (with “principal left fibre bundle” replaced by “principal
right fibre bundle” in the last).

6. HOMOGENEOUS SPACES AND QUOTENT GROUPS

PROPOSITION 11. Let X be a Lie group and G a Lie subgroup of X.
(i) There exixt: on the homogeneou: set X/G one and only one analytio maul/721d
structure such that the tanonioalpny'eelion 1: g" X onto X/G it a submersion. The law of
operation of X on X/G ix analytic. For all x e X, the kernel qx(1t) is obtainedfrom
TAG) by Tow».
(ii) If G is normal in X, X/G it a Lie group with it: gvoup :lrueture and the
many’old structure defined in (i). The mapping 1: i: a Lie group morphism.
By General Topology, Chapter III, §4, no. 1, Example 1, G operates properly
and freely on X by right translation. Hence the first assertion of (i) follows from

219
I'll LIE GROUPS

Proposition 10 ofno. 5. The second follows from the Remark ofno. 5. Since 1: is a
submersion, the kernel of Tx(7r) is the tangent space at x to
-l
1! ("(X)) = ’60 = Y(X)(G)
and hence is obtained from T,(G) by T.(Y(x)).
Suppose that G is normal. Let m be the mapping (x, y) >—>xy" of
(X/G) x (X/G) into X/G. Then (m c (1: x 1r))(x,y) = «(z-y“) for all x, y in
X. Hence in o (1r x 1:) is analytic. As 1: x 1: is a surjective submersion, m is
analytic (Difierentioble and Analytie IbIonifoldx, R. 5.9.5), whence (ii). '
The homogeneous set X/G with the manifold structure defined in (i) is
called the quotient (left) Lie homogeneous spate of X by G. The (right) Lie homo-
geneous space G\X is defined analogously. When G is normal, the Lie group
X/G defined in (ii) is called the quotient Lie group Q/ X by G.
PROPOSITION 12. Let X be a Lie group and Y a non-empty analytic ”.11a with a
law of analytic left operation of X on Y. For ally e Y, let p(y) be the orbital mapping
by y and X” the stabilizer q in X. Thefollowing conditions are equivalent:
(i) there exist: y E Y such that 9(y) is a .rurjeetiue submersion;
(i’) for all y 6 Y, p(y) is a .rurieetive submersion;
(ii) there eximy E Y :t that X, is a Lie subgroup ofX and the canonical mapping
of XIX], into Y is a manifold isomyphiym;
(ii’) for all g e Y, X, i: a Lie subgroup of X and the canonical mapping of X/X,
into Y is a manifold isomorphism;
. (iii) the mapping (2, y) >—> (y, xy) of X x Y into Y x Y i: a rug'ectioe submer-
A10”.

As the canonical mapping of X onto XIX, is a submersion, the equivalences

(i) 4: (ii), (i') «9 (ii') are immediate. (i) o (i’) by Proposition 9 of no. 5.
The equivalence (i') c: (iii) follows from formula (7) of no. 5.
Under the conditions of Proposition 12, Y is called a (left) Lie homogeneour
space of X. Right Lie homogeneous spaces of X are defined analogously.
Example. Let G be a Lie group. We make C x G operate on G on the left by
(g,, g,)x = glxg; ‘. Let p be the orbital mapping ofe. Then the restrictions of
T(s.:)(P) ‘0 T(e.c)(G X (0}) = TAG) X {0} and t0
Tte,e)({‘l X G) = {0} X TAG)
are isomorphisms of these spaces onto T,(G). Hence TK,’.,(9) is surjective and
Ker T(,_,,(p) admits for example the topological supplement T.(G) x {0} in
T“, e,(G x G). Thus p is a submersion at (e, e). Hence G is a left Lie homo-
geneous space of G X G.
PROPOSITION 13. Let G be a Lie group, H a normal Lie subgroup of G, X a manifold
qfelar: C' and (g, x) v» gx a law 1y" lefl operation ofclass C’ q on X. Suppose that
conditions (a) and (b) If Proposition 10 hold.

220
 HouooENEous SPACES AND QUO‘l'IENT GROUPS § LG

(i) Th: low of left operation (h, x) s—> hx of H on X satisfies conditions (a) and (b)
of Proposition 10 (so that we can consider the quotient ”moi/old: X/G and X/H).
(ii) The low of left operation of G on X defines on taking quotients o low If left
opemlion of [lass C" of G/H on X/H; this law satisfies condition: (a) and (b) of
Proposition 10 (so that we can considzr the quotient mong’old (X/H)/(G/H)).
(iii) The canonical mapping of X onto X/H defines on taking quotient: a bijection
of X/G onto (X/H) /(G/H). This hijation i: on isomorphism of manifold: 1y”olms 0’.
Clearly H operates freely on X; it operates properly by Gmrol Topology,
Chapter III, § 4, no. 1, Example 1. The orbital mappings of H on X are immer-
sions since the canonical injection of H into G is an immersion. This proves
(1).
The law of left operation of G on X obviously defines, on taking quotients, a
law of left operation of G/H on X/H. This law is of class C' by Diflzmitiabl: and
Analytic Manifolrit, R, 5.9.6. Letg e G and x e X be such that (Hg) (Hz) = Hr;
then H(gx) = H): and hence gx e H: and g e H; this proves that G/H operates
freely on X/H. The mapping 0: (g, x) -> (x, gx) of G x X into X X X is
closed; on the other hand, 6(Hg x Hx) = Hx x H(gx) ; it follows immediate-
ly that the mapping
(Hg, HI) H (HI: I{(90)
of (G/H) >< (X/H) into (X/H) >< (X/H) is closed; as moreover G/H operates
freely on X/H, Theorem I (c) of Central Topology, Chapter I, § 10, no. 2 proves
that G/H operates properly on X/H.
Let 1: be the canonical mapping of X onto X/H, a the canonical mapping
ofG onto G/H, :: an element ofX andy = 1r(x).

Gflx
G/HflX/H
The“ 7' ° 90‘) = PM) ° 6 and hence
Tx(1=)° T.(P(x)) = T.(P(y)) o T.(a).
Let u e TAG/H) be such that T,(p(y))u = 0. There exists 1/ E T¢(G) such that
u = T,(a-)o. Then T,(1r)(T,(p(x))o) = 0, hence T,(p(x))v is tangent to Hz
(Dyermtiable and Analytic Manifolds, R, 5.10.5) and therefore is of the form
T,(p(x) | H)u’ for some 11’ eT.(H). As T,(p(z)) is injective, it follows that
o = o’, whence v E T,(H) and therefore 14 = 0. Thus T.(p(y)) is injective. The
image of T,(p(y)) is equal to that of T,(7r) oT,(p(x)); now the image of
Te(p(x)) admits a topological supplement in T,(X) and contains the kernel of
Tx(1r). It is therefore seen that p(y) is an immersion, which completes the proof
of (ii).
221
III LIE GROUPS

Assertion (iii) follows from the above and Diflerentiable and Analytic Alangflalds,
R, 5.9.7.
COROLLARY. Let G be a Lie group and H and L normal Lie subgroups of G with
L C H. Then H/L is a normal Lie subgroup (if G/L and the canonical bijection of
G/H onto (G/L)/(H/L) is a Lie group isomorphism.

7. ORBITS
PROPosmON 14. let G be a Lie group, X an analytic manifold and (g, x) t—> gx a,
law of analytic left operation of G on X. let x E X. Suppose that the corresponding
orbital mapping p(x) it a subimmersian (which is always the case if K is ofcharacteristic
0 and X irfinite-dimmsional (Corollary to Proposition 9)). Let G,‘ be the stabilizer
o x in G.
f(i) G, is a Lie subgroup and T,(Gx = Ker T,(p(x)).
(ii) The canonical mapping ix ofthe homogeneous space G/Gx into X is an immersion
with image Git.
(iii) {ffilrther the orbit Ga is locally closed and the topology on G admits a countable
base, then Gx is a submanifold of X, ix is an isomorphism of the manifold G/G, onto
the manifold Oz and Tx(Gx) = Im T,(p(x)).
The inverse image of at under p(x) is G”. As p(x) is a subimmersion, G; is a
submanifold and, for all g5 G, the tangent space J to n = p(;,’(gx) at g
is Ker T,(p(x)) (Difierentiable and Analytic Manifolds, R, 5.10.5), whence (i).
Let 1:: G —> G/G, be the canonical mapping. Then i, o 1: = 90:). As G/G, is a
quotient manifold of G, this equality proves that iJr is analytic. Further, the
kernels of T,(p(x)) and Tg(1r) are both equal to J. Hence Tn(,)(i,) is injective.
The image of Tm,(i,‘) is equal to the image of T,(p(x)) and hence admits a
topological supplement. This proves (ii).
Suppose that Gx is locally closed. Every point of Gx then has a neigh-
bourhood in G1: which is homeomorphic to a closed subspace of a complete
metric space and hence is a Baire space. Hence Gx is a Baire space (General
Topology, Chapter IX, '9' 5, Proposition 4). IfG has a countable base, ix is there-
fore a homeomorphism of G/Gx onto 0:: (General Topology, Chapter IX, §5).
Then by (ii) and Diflerentiable and Analytic Manifolds, R, 5.8.3, ix is an isomor-
phism of the manifolds G/G,‘ onto the manifold Gx and

max) = Im Two) = Im max».

Remark. Let G be a finite-dimensional Lie group, X a manifold of class C.’
and (g, x) >—> gx a law of left operation of class C’ of G on X. Then Propo-
sition 14 remains valid. The only point which needs a difierent proof is
the fact that G, is a Lie subgroup. But if r as m, K = R; as clearly G“ is
closed, G,‘ is a Lie subgroup by § 8, Theorem 2.

COROLLARY. Let G be a Lie group whose topology admits a countable base and X a

222
 vecron BUNDLES wrm OPERATORS § 1.8

nan-enWyfmite-dirmnsianal analyti: manifild with a law ofanalytic left alteration of G

on X. Supp”: that G operate: tramilivity an X and that K i: qfcharaclefixtio 0. Then
X i: a Lie homogenem spacefar G.
Let x e X. The orbit of x, equal to X, is closed and we can therefore apply
Proposition 14 (iii).
a. VECTOR BUNDLES WITH OPERATORS
Let G be a Lie group, X a manifold of class C" and (g, x) »—> gt a law of left
operation of class C' of G on X. Let E be a vector bundle of class C’, with base
space X and 1:: E —> X the projection of E onto X. For all x e X, let Ex be the
fibre of E at x. Let (g, u) >—> gu be a law of left operation of G on B such that
7: is compatible with the operations of G on X and on E. For all g e G and all
x e X, the restriction to Ex of the mapping :1 r—> gu is a bijection ¢,_ ,, of E,‘ onto
EM. We shall assume that, for all g E G and all x e X, LP“, is continuous and
linear and hence is an isomorphism of the normable space E, onto the norm-
ablc space E”.
Let <1) be the automorphism (g, x) »—> (g, gx) of the manifold G x X. Let p
be the canonical projection of G x X onto X and E' the inverse image of E
underp. Let 4;: E' —> E' be the mapping the sum of the slim): Efg'x, —> Elms)!»
DEFINITION 4-. If LII i: a iii-morphism cf nectar bundle: ofclm'r C’, E is called a vector
G~bundle of class C’.
In other words, E is a vector G-bundle of class C' if for all (go, x0) 5 G x X
the following condition holds: there wrists an open neighbourhood U of
(gmxo) in G x X such that, ifE’ | U (resp. E' | 4>(U)) is identified with a
trivial vector bundle of fibre M (resp. N) by means of a vector chart, the
mapping (g, x) ~—> 4/“, of U into $(M, N) is of class C’.
The mapping 4; is obviously bijective and it follows from the above local
criterion that £11“ is a o'1-morphism of vector bundles so that q; is a 45-12m-
morphism of vector bundles.
A trivial vector G—lmndle qf bare X is a vector bundle X x F (where F is a.
complete norrnable space) with the law of operation (g, (x,f)) >—> ( gx,f) of G
on X x F.
We again assume the hypotheses and notation preceding Definition 4 and
further take 1- to be a vector functor of class C' for isomorphisms (Difmntiable
and Analytic Mani/97111:, R, 7.6.6). Then TE is a vector bundle with base space X.
For all x E X, its fibre (1E) x is equal to 1,-(E,). For all normable spaces N1, N5,
let Isom(N1, N3) denote the set of isomorphisms of N1 onto N2. Ifg e G, then
101'”. x) E 150m((TE)n (113)”)-
The 1011,“) define a law of left operation (g, u) »—> gu of G on TE and the cano-
nical projection of1E onto X is compatible with the operations of G on X and
TE.

223
m LIE (mom

PROPOSITION 15. If E is a vector G—bmulle of class 0, TE 1': a motor G-bumlle of

elasx 0’.
Let go, x“, U, M, N be as in the paragraph following Definition 4. Then the
mapping (g, x) |—>’|,'(ql",‘) of U into .‘l(-rM, 1N) is the composition of the
mapping (g, x)»—M]1,_x of U into $(M, N) and the mapping f>—>r(_f) of
Isom(M, N) into Isom(1-M, 1N) ; these two mappings are ofclass C’ and hence
so is their composition, whence the proposition.

PROPOSITION 16. Let G be a Lie group, X a manifold of class C’ (r 2 2) and

(g, x) »—> gx a law of left operation afelarr C' of G on X, whence, by trmpofling the '
structure, there is a law an" left operation t on TX. Under this law, TX is a vector G-
bundle ofclars 0‘1.
Let prl (resp. prg) be the canonical projection of G x X onto G (resp. X)
and let E, (resp. E3) be the inverse image of TG (resp. TX) relative to pr1
(resp. pra). Then the vector bundle T(G X X) is the direct sum of E1 and E2.
Let 1': E2—>T(G x X) and a: T(G x X) —> E be the canonical vector
bundle morphisms defined by this decomposition into a direct sum. Let (I) be
the mapping (g,x) »—> (g,gx) of G x X into G x X. Then the mapping
denoted by kl) in Definition 4 (where we put E = TX) isjust q o T(cb) n 1'. But
T(rb) is a o-morphisrn ofvector bundles ofdass C"1 (Difl'erentiable and Analytic
Manifillir, R, 8.1.2).

COROLLARY. {ft is a vectarflnetor of alas: C' for isomomhixms, r(TX) i: a ueetar

G—bandle of class C’ '1.
This follows from Propositions 15 and 16.

Remark 1. If r is a vector functor of class C" {or isomorphism in finite dimen-

sion and E is of finite rank, 1E is defined similarly and Proposition 15 remains
valid; the Corollary to Proposition 16 remains valid provided X is finite-
dimensional.

Examples. With the hypotheses and notation of Proposition 16, let F be a com-
plete normable space. Then .‘Z’((TX)’; F) is a vector G~bundle of class C"‘; so
is AIt"(TX; F) if K is of characteristic zero or X is finite-dimensional (cf.
Dzjfirmtiable and Analytic Mamfilds, R, 7.7, 7.8). if X is finite-dimensional,
(8)” (TX) ® (8)" (TX)* is a vector G-bundle of class 0‘1.
PROPOSITION 17. Let G be a Lie group, X a left Lie homogeneous space of G, are a
point of X, Go the stabilizer of x0 in G, E and 13' left vector Gbundlex of class Cr
and have space X, E, (rwp. E3) the fibre of E (resp. E’) at x0 andf an element of
$(Eo, E’D) such thatf(gu) = gf(u) for all a e E0 and g e G“. Then there exist: one
and only one morphism of E into E' compatible with the operation: of G and extendingf.
The uniqueness of this morphism is obvious. We prove its existence. Let g,

224
 - VECTOR BUNDLES WITH OPERATORS § 1.8

g’ elements of G and u EE0 be such that gu = g'u. Then g"1geGo and

g"‘gu = u and hence g"1gf(u) =f(u), that is gflu) = g'f(u). Hence a
mapping 4) is defined of E into E’ by writing ¢(gu) = gf(u). Clearly this
mapping extendsf and it is compatible with the operations of G. We show that
vb is a vector bundle morphism of class 0’. Let x1 E X. There exists an open
neighbourhood V oft:1 in X and a submanifold W of G such that the mapping
g »—> gxo is an isomorphism 0 of class C' of W onto V. By shrinking V and W it
can he assumed that:
(l) E I V (rap. E' I V) is identified with a trivial vector bundle of fibre M
(resp. M’);
(2) if 41, (rcsp. 44) denotes the mapping u t—> gu of E0 (reap. E’D) into E,“o
(mp. Erma), then the mappings g H 4'. and g e 4/? (rest g H 4»; and
g t—> 44—1) of W into $(Eo, M) and $(M, E0) (resp. $(E6, M’) and
$(M', 133)) are of class C’.
For e, let ‘l’x: M—>N be the restriction old) to E; = M. Then (I), is
obtained by composing the following mappings:
(1) the mapping (¢°'l:)_1 oi'M into E0;
(2) the mappingf of E0 into E0;
(3) the mapping (leg-1x of ED into M’.
Hence we see that the mapping 1: >—> 'i>x of V into .‘Z’(M, M’) is of class C'.
COROLLARY 1. Let Ego be the m qfelementx ao which are invariant under Go. For
all u 5 E30, let a“ be the mapping ofX into E defined by c“(gxo) = gufar all g E G.
(i) The G-immriant :ectimrl‘ qf E are qftlass 0'.
(ii) u >—> a“ is a bijem'on qgn onto the set of G-irwariant mtion: zy‘E.
Assertion (ii) is obvious. To prove (i) it is sufficient to prove that each
section a“ is of class 0’. Let E’ be the trivial G-bundle of base X and fibre Ego‘
Letfbe the canonical injection of Ego into E0. By Proposition 17 there exists a
morphism 4; of E’ into E compatible with the operations of G and extendingf.
Ifu E Ego and g E G, then

64.4%) = gu = gflu) = Mg") = 1W", 3%))

and hence 0.,(x) = 42((14, 1)) for all x E X, which proves our assertion.

*For example, let G be a finite-dimensional real Lie group, G, a compact

Lie subgroup of G and X the homogeneous space G/Gn. Let x9 denote the
canonical image of e in X. There exists a positive definite symmetric bilinear
form on T,U(X) invariant under Go (Integration, Chapter VII, §3, Propo-
sition 1). Applying the above to (TX)' 8) (TX)‘, we see that there exists
on X an analytic Riemannian metric invariant. under Gr

1' By a union of E we here mean a mapping a (not necessarily of class C’) of

X into E such that p n e = 1d,, where p denotes the projection of E onto X.

225
III LU! GROUPS

COROLLARY 2. Suppose that Go operates trivially on E0. Let E’ be the trivial G-

bundle with bare spate X ondfihre E0. There exist: one and only one iromorphism ofE
onto E’ compatible with the operatiom q and extending Idle.
This follows immediately from Proposition 17.
Remark 2. In this not, the laws of left operation can be replaced throughout
by laws of right operation.
9. LOCAL DEFINITION OF A LIE GROUP
PROPOSITION 18. Let G be a group and U and V two subset: of G containing e.
Suppose that U has an analytic manifold structure satigfying thefollowing eonditionr:
(i) V = V'1,V2 C U,VitopeninU;
(ii) the mapping (x,y) n‘l q x V into U is analytic;
(iii) for all g E G, then: exist: an open neighbourhood V' If e in V mh that
gV'g'il C U and such that the mapping 1: ng" ofV' into U i: analytie.
Then there exist: one and only one analytic manifold structure on G with the follow-
ing properties:
(at) G with this structure is a Lie group;
((3) V is open in G;
(7) the manifold structure: on G and U induce the some :trut'tuie on V.
(a) Let A be an open subset ofV and on an element ofV such that ooA C V.
Then voA is the set of o e V such that v; ‘o e A and hence is an open subset of
V (taking account of (ii)). Moreover, (ii) implies that the mappings
o r» you of A onto (IDA and o v» oo— 1o of 11A onto A are inverse analytic bijec-
tions and hence analytic isomorphisms.
(b) We choose an open neighbourt W of e in V such that W = W",
W3 C V and there exists a. chart (W, 4), E) of the manifold U with domain
W. For all g e G, let <12, be the mapping h »—> 45 (g'lh) ofgW into B. We show
that the charts (1), of G are analytically compatible. Let g1, g be elements of
G such that g‘W n gaw at 0, so that g; 1g, and gf 1g, belong to W“. By (a),
W n g; lgaW is an open subset of W and hence
11>..(glw 0 saw) = ¢(W n 31‘ EM)
is an open subset D of E. For d e D,
(4%. ° in?) (d) = <l>(sz' gxh‘Wl);
by (a) we see that (hm o 4);] is analytic.
(c) By (b) there exists on G an analytic manifold structure such that
“my“, is an atlas on G. For all goe G, the mapping g» gag (g6 G) leaves
this atlas invariant and hence is an automorphism of G with this manifold
structure. In particular condition (GL1) is satisfied.
(d) Let no EV. By (ii) there exists an open neighbourhood A of e in W
such that, 00A C V. This proves first that V is open in G. By (a) the mapping
0 >—> our; of A onto voA is an analytic isomorphism with the structures induced

226
 LOCAL DEFINITION or A LIE GROUP § 1.9

by U. By (c) we see that the manifold structures on G and U induce the

same structure on voA and hence finally on V.
(e) by (d), (ii) and (iii), we see that conditions (GL,) and (GL5) hold.
Hence G is a Lie group.
(f) If a manifold structure on G is compatible with the group structure
of G and is such that V is an open submanifold of G, then (4),),50, is an atlas
on G. Hence the uniqueness assertion of the proposition.
PROPOSITION 19. Let G be a topological group, H a Lie group andf a homomorphism
of the group G into the group H. Suppose that there exist an open neighbourhood [If eG
in G, a chart (V, (b, E) of the manifold H at eK and a closed vector subspace F If E
admitting a topological supplement, such that f(U) C V and 4) ef) I U is a homeo-
morphism of U onto (NV) n F. Then there exists a unique manifold structure on G
such thatf is an immersion; this structure is the inverse image under f of the manifold
structure on H. With this structure G is a Lie group.
As translations of G (resp. H) are homeomorphisms (resp. analytic iso-
morphisms), f satisfies condition (R) of Dzflerentiable and Analytic Manifolds,
R, 5.8.1. The first two assertions of the proposition then follow from Dtjfiren-
liable and Analytic Manifizlds, R, loc. cit. Consider the commutative diagram
GxG—LG
l I

HxH—">H
where m(x,y) = zy" (reap. n(x,y) = erg/'1) for x, y in G (resp. H). Then
71 e (f x f) is analytic, hence f o m is analytic and hence on is analytic since
f is an immersion. Therefore G is a Lie group.
The Lie group structure on G is allied the inverse image of the Lie group
structure on H underf.
COROLLARY. Let G be a topological group, N a discrete normal subgroup of G and
7: the canonical mapping of G onto G/N. Suppose that an analytic manifold structure
is given on G/N, compatible with the topolog'ual group structure on G/N. Then there
exists a unique manifold structure on G such that 1: is an immersion; this structure is
the inverse image under 1: {If the manifold structure on G/N. With this structure, 1: is
Itale, G is a Lie group and G/N is the quotient Lie group ofG by N.
Remark. Let H be a connected real or com lex Lie group, H its universal
oovering‘l‘ and re the canonical mapping of onto H. When we speak of H

TCf. General Topology, Chapter XI; whilst awaiting the publication of this
chapter, see for example L. S. Pontrjagin, Topological groups, 2nd edition trans-
lated from Russian, Gordon and Breach, 1966; or G. Hochschild, The structure of
Lie groups, Holden-Day, l965.

227
III LIE GROUPS

of
as a Lie group, we shall always mean with the inverse image structure
that on H under 1:.
10. GROUP GERMS
DEFINITION 5. A Lie group germ aver K is a system (G, e, 0, m) ratirfi/ing the
following conditions :
(i) G is an analyltk manifold over K;
(ii) e e G;
(iii) 6 Lr An analytic mapping if G into G;
(iv) m is an analytic mapping 0f an open subset Q q x G into G;
(V) fl" allge G, (M) e Q, (g, t) E 9, "1(2. 3) = “(3, I) = g;
(Vi) fordllge G, (g, 9(9) E 9’ (9(3),.g) G 9,"!(5', 9(3)) = m(°(g):£) = 6;
(vii) ifg, h, k are elements ofG such that (g, h) e (1, (h, k) e (2, (m(g, h), k) E Q,
(g: ”‘01: ’0) E 9, the“ m(m(£a ll): ’5) = "1(g, "‘0'; k)).
a is called the identity element of the group germ. We often write gh
instead of m(g, h) and (by an abuse of notation) g'l instead of 0(g).
A Lie group G is a Lie group gem with the obvious choice of e, 0, m.
Let G be a Lie group germ. Then :3" = I, that is
(8) e" = a.
For all g e G,
g = cg = ((g")"g'1)g = (g“)“(g‘ g) = (3")‘16,
that is
(9) (g'1)‘1 = g~
A subset of G invariant under the mapping g »—> g“ is called symmetric.
The manifold G, with the point e, the mapping g» g" and the mapping
(g, h) »—> hg is a Lie Group germ GV called the opposite of G.
The Lie group germ G is called commutative if, for all (g, It) 5G >< G
such that gh is defined, hg is defined and equal to git.
Let G be a Lie group germ. The set of (g,h) eG x G such that g}: is
defined is a neighbourhood of (e, 2). On the other hand, the mappings
(g, It) &—> gh and g ~—> g'1 are continuous. Hence (gh)k = g(hk) for g, h, k
sufficiently close to 2. Similarly, (h‘lg'l) (gh) = h"(eh) = h'1 h = t for g, h
sufiiciently close to c, whence multiplying on the right by (gh)",
(10) (3h) ‘ ‘ = It‘lg“ for g, h sufficiently close to e.
Pnorosrnou 20. Le! G b: a Lie group germ and g e G. Time zxist an open neigh-
bourhood U aft and an alien neighbourhwd V q with thefollowing properties:
(21) ug ir definedfor all u E U;
(h) vg'1 is defined/0r all u e V;
(c) the mappiflgi u H) 14.2. I! I~> 1g" 1 an inverse amtlylit isomorphisms of one
analher of U ml!) V and 0/ V unto U.

228
 GROUP GERMS § 1.10

As the set of definition of the product is open in G x G, there exist an

open neighbourhood U of 3 and an open neighbourhood V of g with proper-
ties (a) and (b). Let 11(u) = ug for 11 EU, n’(v) = rig—1 for 1) EV. By shrink-
ing U and V, it can be assumed that (u‘g)g'l = u and (ug‘1)g = v for u EU
and 1; EV. Then ‘4 and 1)’ are injections. By shrinking U further, it can be
assumed that n(U) C V. Then n’(V) D U and 7)(U) is the inverse image of
U under n’ and hence is an open neighbourhood of g in V. Replacing V by
n(U), we finally arrive at the situation where 1) and 71’ are analytic inverse
bijections.
Let G1, G2 be two Lie group germs with identity elements 91, 9,. A mapping
f of G1 into G2 is called a morphism iff satisfies the following conditions:
(i) f is analytic;
(ii) f(¢1) = ea;
(iii) if g, h are elements of G1 such that gh is defined, then f(g) f (h) is
defined and is equal tof(gh).
Let g 9 G1. As gg“ is defined and is equal to s,,f(g)f(g") is defined and
is equal to e, and hence
fig)“ =f(g)“(f(g)f(s“)) = (f(g)'1f(g))f(g")
that is
(11) fig)“ =f(£")-
The composition of two morphisms is a morphism.
Iff : G1 —> G, and f': G, —> G1 are two inverse morphisms of one another,
they are isomorphisms (using in particular formula (11)).
Let G1, G2 be two Lie group germs, and f a mapping of G1 into G2 satis-
fying conditions (ii) and (iii) above, which is analytic in an open neighbour-
hood of :1. Using Proposition 20, it can be proved as in Proposition 4 that f
is a morphism.
Let (G, e, 0, m) be a Lie group germ and 9 the set of definition of 17:.
Let H be a submanifold of G containing 2, which is stable under 0. Suppose
that the set 01 of (x,y) 6 Q n (H x H) such that m(x,y) EH is open in
H X H. Then (H, e, 0|H,m|Ql) is a Lie group germ. Such a Lie group
germ is called a Lie rubgmup germ g" G. The canonical injection of H into
G is a morphism. Iff : L—> G is a morphism of Lie group germs such that
f(L) c H, then f: L —> H is a morphism of Lie group germs.

Suppose that K is of characteristic 0. If we replace the hypothesis that H is

a submanifold of G by the hypothesis that H is a quasi-submanifold of G,
the results of the above paragraph remain true (cf. Diflnmh'able and Analytic
Mani/blah, R, 5.8.5). H is then called a Lie quasi-subgroup germ of G.

If G is a Lie group gem with identity element 1, every symmetric open

229
m LIE GROUPS

neighbourhood of e in G is a Lie subgroup gem of G. (This applies in par-

ticular when G is a Lie group.) Let H be a Lie subgroup germ of G; if H is
a neighbourhood of e in G, then H is open in G by Proposition 20. .
The product Lie group germ of a finite number of Lie group germs 15
defined in an obvious way.
PROPOSITION 2L Let G, H be two Lie group germs and 4) a morphism of G into H.
Thefollowing conditim are equivalent:
(i) (I) ixltuleute; '
(ii) there exist open Lie subgroup germ G’, H’ of G, H such that 11>[G' 14‘ an
iromorphirm of G' onto H’.
The implication (ii) 2 (i) is obvious. Suppose that (I; is étale at e. There
exists an open Lie subgroup germ G1 of G such that 11>(Gl) is open in H and
(HG, is an isomorphism of the manifold Gl onto the manifold {>(G1). Then
there exists an open Lie subgroup germ G' of G, such that the product in
G of two elements of G’ is always defined and belongs to G). If g, g’ are
elements of G' such that gg’eG’, then ¢(g)<]>(g') = ¢(gg') etMG'); if g, g’
are elements of G’ such that g’ 6 G1 — G', then

d>(g)4>(g’) = 4>(gg')E<l>(G1) - ¢(G')-

Hence (HG’ is an isomorphism of the Lie group germ G' onto the open Lie
subgroup germ (MG’) of H.
If the conditions of Proposition 21 hold, G and H are called locally iro-
morphic.
PROPOSITION 22. Let H he a Lie group, U a Lie mbgroup gem of H and N the set
ofg e H such that U and gUg‘1 have tiresome germ ate (General Topology, Chapter I,
§6, no. 10). Then N is a subgroup ifH containing U. There exist: one and only one
analytic manifold xtructure on N with thefollowing properties:
(i) N with this structure it a Lie group;
(ii) U i: an open rubmunifold of N;
(iii) the canonical injection of N into H i: an immersion.
Clearly N is a subgroup of H. Ifge U, then gee U and geg" EU, hence
gu e U and gug‘1 E U for u sufficiently close to e in U and hence the germ of
gUg'1 at e is contained in that of U; exchanging g and g“, we see that
the germs ong'l and U at e are equal. Hence U C N.
Let V be an open neighbourhmd ofe in U such that V = V“, V2 C U.
Conditions (i), (ii), (iii) ofProposition 13 ofno. 9 (where G is replaced by N)
are satisfied. Hence there exists an analytic manifold structure on N with the
following properties: (on) N with this structure is a Lie group; (B) V is open
in N; (-y) the manifold structures on N and U induce the same structure on
V. Since V is a submanil‘nld of H, the canonical injection ofN into H is an
immersion at e and hence at every point of N. Let ueU. There exists an

230
 LAW CHUNKS or OPERATION § 1.1!

open neighbourhood V' of e in V such that the mapping 0 v—> all is an analytic
isomorphism of V’ onto an open neighbourhood of u in U (Proposition 20)
and at the same time onto an open neighbourhood of u in N. Hence U is
open in N and the identity mapping of U is an isomorphism with the given
manifold structure on U and the open submanifold structure on N ; in other
words, U is an open submanifold of N.
Finally, we consider an analytic manifold structure on N with properties
(i) and (ii) of the proposition and let N“ be the Lie group thus obtained. Then
the identity mapping of N into N* is étale at e and hence a Lie group iso-
morphism. This proves the uniqueness assertion of the proposition.

Let H be a Lie group, U a Lie quasi-subgroup germ of H and N the set

ofg E H such that U and gUg" have the same germ at e. If K is of charac-
teristic 0, there exists on G one and only one manifold structure with
properties (i) and (ii) of Proposition 22. The proof is the same as for
Proposition 22.
COROLLARY. Preserving the notation of Proposition 22, let G be the subgroup of H
generated by U. Then G i: an open subgroup of N. There Mitts one and only one Lie
group structure on G such that U is an open .mbmanifold of G and the canonical injec-
tion ofG into H is an immersion.
Remark. Preserving the notation of Proposition 22 and its corollary, suppose
that K is of characteristic 0, that H is finite-dimensional and that the topology
on U admits a countable base. Even with all these hypotheses it is possible
that G is not closed in H (Exercise 3). But, if G is closed, G is a Lie subgroup of
H. For the mapping (g, h) H gh is a law of analytic left operation of G on H.
The orbit of e is G. Our assertion then follows from Propositions 2 (iv) and
14 (iii).

11. LAW CHUNKS OF OPERATION

Let (G, e, 0, rn) be a Lie group germ and X a manifold of class (3'.
DEFINITION 6. A law chunk of lefl operation of class C' of G on X is a mapping 4;
defined on an open subset (I of G x X containing {e} x X, with value: in X and
with thefollowing properties:
(i) ill 1': ofclar: C';
(ii) fizr all x E X, Me, x) = 2;
(iii) there exists a neighbourhood (21 of {e} x {e} X X in G X G x X such that,
fl" (g. g'; I0 6 91: ”'6 616mm“ 0' “(3, g'), ¢(m(g) g'): ’0: Mg, Mg’, J0) 4”
dtfim’ and Mg, 44!, ’0) = ¢(m(g,g'), 1)-
Law chunks of right operation of class C' are defined similarly.
We often write gx instead 013.14g, x).
Let G’ be a Lie subgroup germ of G and X’ a submanifold of X. Suppose

231
lIl LIE GROUPS

ls open in
that the set (1’ of (g, x) e Q n (G' x X’) such that 44g, 2:) EX’
.X)..Then
G’ x X’ (a condition which is always fulfilled if X’ is open in
is said
449’ is a law chunk of left operation of class C" of G' on X , Wth
to be derived from t]; by restriction to G’ and X’.
(3',
Pkoposmou 23. Let (G, e, 0, m) be a Lie group germ, X a manifold of clot:
x0 a point of X, 0 an open neighbourhood of (e, no) in G x X and q, a mapping of
9 into X with thefollowing properties:
(i) LIJ is ofolasx C';
(ii) Me, at) is equal to zfor x mfieiently close to x0;
(iii) ¢(m<g. g'). x) = Mg. Mg’, ow to g', x) mommy close to (e, ., n)-
Then there exist on open neighbourhood X’ of no in X and on open subset Q’ of
Q n (G X X') such that 449’ i: a low thunk qflefl operation ufclass C’ ofG
on X’.
There exist an open neighbourhood X’ of xn in X and an open neighbour-
hood G' of: in G such that Me, x) = x for all x e X, and

Mg, Mg’, 3)) = ¢(M(g.g’), X)

for (g,g’,x) eG' x G’ x X’. Let (1’ he the set of (g,x) 60 n (G’ x X’)
such that ¢(g,x) eX’. Then (2’ is open in G x X’ and X', 9' have the
properties of the proposition.
Lemma 3. Let X be a normal space and (Kl);EX a locally finite open oooering of X.
For all (i, j) e I X I and all :rEXl n X,, let V,,(x) be a neighbourhood of x [07!-
tained in X‘ n X,. Then we can associate with every :5 X a neighbourhood V(x)
of x .mch that thefizllowing conditions arefilled :
(a) the relation x e X‘ n X, implies V(x) C V,,(x) ;
(b) ifV(x) and V(y) meet, there exist: {6 I such that V(x) U V(y) C X,.
There exists an open covering (X1),$1 of X such that X} C Xi for all is I
(General Topology, Chapter IX, §4, Theorem 3). Let x e X. Let V1(x) be the
intersection of the V,,(:t) and the X}, which contain x; this is an open neigh-
bourhood of 2:. Let V,(x) be a neighbourhood of x contained in V,(x) and
mee_ting only a finite number of X‘. Then V2(x) meets only a finite number
of X,’ and hence the set

V(;r) = V301) n (net? (X _Y’)

is a neighbourhood of x. If x EX, n X,, then V1(x) C XI n X, and hence

V(;e) C X( n X,. Let x, y be in X and suppose that V(x) and V(y) meet.
There exists iE_I_such that 7: eXi’. Then V,(x) C X1, hence V(x) C X; and
hence V(y) n Xhé 0. Then y 6 2?; by definition of V(y), whence y EXl
and V(y) C X‘. Thus X‘ contains V(x) and V(y).

232
 TANGENT uws or composmorz § 2.1

PROPOSITION 24. Let G be a Lie group gem, X a manifold of elm: C’ and (dx
a locallyfinite open covering of X. For all {9 I, let vb, be a law chard: of left operation
of clay: C’ g" G on X. Suppose that the underlying topological :pace If X is normal
and thal,for all (i,j) E I X I and all x EX, r\ X,, (It; and 4;, coincide on a neigh-
bourhood of (e, x). There exist: a law chunk 4/ of left operation of clay: C' of G on X
Inch that, for all i E I and all x e X‘, 4;. and q; coincide on a neighbourhood of (e, x).
For all (i,j) 61 x I and all xeX1 n X, choose an open neighbourhood
Vl,(x) of x in X, n X, such that 4;, and vb, are defined and equal on a neigh-
bourhood of {e} x V.,(x) in G x X. For all x e X choose an open neighbour-
hood V(x) of x in X such that conditions (a) and (b) of Lemma 3 are ful-
filled. Let I, be the set of i e I such that XEX‘. This is a finite set. Let U,‘
be the set of (g,y) SO x V(x) such that the q“ for £51,: are defined and
coincide on a neighbourhood of (g,y). Then U, is open and (47:) EU,”
The 41, for ie 1,, all have the same restriction to Ux. Let x, y be in X. If U”
and U” meet, V(x) and V(y) meet and hence there exists is I such that

Var) u V(y) c X..

Then ie Ix, ie, 4;d = 41*, MU” = vb, and hence

444W: n Uy) = *l‘uKUx 0 Us)-

The 44,, therefore define a mapping 4; of U = $s Ux into X and U is an
open neighbourhood of {e} x X in G x X. Clearly K!) is of class C” and
Me, x) = x for all x e X. For all is I and all xeXl, q; coincides with 4;, and
hence with q“ in a neighbourhood of (e, x) and hence 4; satisfies condition
(iii) of Definition 6. .

§2. GROUP OF TANGENT VECTORS TO A LIE

GROUP

1. TANGENT LAWS OF CONlPOSITION

Let X and Y be manifolds of class C'. We know (Difl’erentiable and Analytic
Manifolds, R, 8.1.4) that X x Y is a manifold of class C’ and that the map-
ping (T(prl), T(pr2)), the product of the tangent mappings to the canonical
projections, is an isomorphism of class C" 1 ofT(X x Y) onto T(X) x T(Y).T
This isomorphism is compatible with the vector bundle structures with base
space X x Y and allows us to identify T(X X Y) with T(X) x T(Y). Let

T For 1 = 1,1his means that (T(prl), T(pr2)) is a homomorphism ofT(X x Y)

onto T(X) x T(Y),

233
m LIE (mom

a E X, b eY, u e T,(X), u eTh(Y); the above identification allows us to con-

sider (u, u) as an element of T(a‘,,(X x Y); then
('1: 0) = (14,0) + (0,”)
and (u, 0) (resp. (0, 11)) is the image of u (reap. 1;) under the tangent mapping
to the immersion x»—> (x,b) (resp. y» (a, 31)) of X (resp‘ Y) mto X x Y.
When it is necessary to be precise, we shall write 0‘I for the zero element of
T (x). .
“Now let X, Y, Z be manifolds of class C’ and f a mapping of class.C' of _
X x Y into Z. The tangent mapping is, using the above identification, a
mapping of class C"1 of T(X) X T(Y) into T(Z). For 11 e T,(X) and
v e Tow),
(1) T(f)(u, '1) = T(f)(u, 0») + T(f)(0.» 11):
(2) T(f)(0a, 0n) = 0m,»-
On the other hand, the mapping 3/»f(a, y) is the composition of the im-
mersion y .—> (11,31) andf; it follows that
(3) T(f) (0, v) is the image of a under the tangent mapping to y >->f(a, y).
Similarly
(4) T(f )(u, 0) is the image of 11 under the tangent mapping to x I—>f(x, b).
If the mappingf of X x Y into Z is denoted by (x, y) ._> xy, no is often used
to denote the element T(f)(u, u) for u e T(X), u E T(Y).
Let X be a manifold of class C' and m: X x X —> X a law of composition
of class C’ on X. Then T(m) is a law of composition of class C"1 on T(X).
It is called the law qf composition tangent to In. The canonical projection 11 of
T(X) onto X is compatible with the laws In and T(m); in other words,
(5) [1°T(m)=m°(fi><fl)-
It follows from (2) that
(5) T(m)(0x: 01) = 0m.»
for all x, y in X; in other words, the zero section 2: ~—> 0,, ofT(X) is compatible
with the laws m and T(m).
PROPOSITION 1. Let X be a manwd 0f class C' and m a law afwmporilion ofclm:
C' on X. If m is associative (rap. mmmutative), then T(m) is axxodative (resp.
commutative).
If m is associative, then m o (m X Idx) = m o (Idx x m), whence

T(M) ° (T(m) X Id‘rm) = T(M) 0 (”mo >< T(MD

234
 GROUP OF TANGENT VEL‘I‘ORS TO A LEE GROUP § 2.2

and hence T(m) is associative. Let .r be the mapping (x, y) r» (y, x) of X x X

into X x X. If m is commutative, then m o x = m and hence

T(m) em) = T(m).

But T(:) is the mapping (u, v) »—> (v, u) of T(X) x T(X) into T(X) x T(X).
Hence T(m) is commutative.
PROPOSITION 2. Let X be a manifvld (3/ class C’, rn a law qf composition qf claxs
C' on X and e an identity elementfir m.
(i) The vector 0, it an identity elementfir T(m).
(ii) T,(X) ix .rtable under T(m) and the law of composition induced on T,(X) by
T(m) i: the vector space addition on T,(X).
(iii) Let U be an apen mbxet of X and a a mapping (3)" class C' of U in!» X .vuch
that,for all x EU, a(x) is the inverse 11f: under rn. Then,for all uET(U), T(a)u
i: the inverxe of it under T(m).
Properties (3) and (4) show that T(m)(0‘, u) = T(m) (u, 0,) = u for all
u e T(X), whence (1'). For u, v in T,(X),

T0006 17) = T('")(“: 0.) + T(mXO” 1') = u + 0.

whence (ii). Finally the relations rn(x,ot(x)) = m(a(x),x) = e for all e
imply
T(m)(u,T(¢)(u)) = TWO"(1)“, u) = 0.
for all u e T(U), whence (iii).
PROPOSITION 3. Let X1, X3, . . ., X,,Y be manifold: qf class C’, i an integer qf
[1,p], m. (reap. n) a law qfcompoxition qfclas: C' an X, (reap. Y) and u a mapping
qfclaxs C” ol x X2 x x X, into Y. [[14 is dixtributive relative to the
variable of index i, then T(u) i: dixh-ibutive relative to the variable qfindex i.
The proof is analogous to that of Proposition 1.

1. GROUP OF TANGENT VECTORS TO A LIE GROUP

PROPOSITION 4. Let G be a Lie group. Then T(G), with the law If composition
tangent to the multiplication 9/ G, i: a Lie group. The identity element If T(G) i:
the vector 0,.
This follows from Propositions 1 and 2.
PROPOSITION 5. Let G and H be Lie group: andfa morphine: vfG into H. Then
T(f) i: a rrwrphixm afthe Lie group T(G) into the Lie group T(H).
We know that T(f ) is analytic. On the other hand, let m (tsp. n) denote
the multiplication on G (resp. H). Thenf o m = n o (f x f), whence

T(f) °T(m) = T(n) ° (T(f) X T(f)):

235
II! LIE GROUPS

which expresses the fact that T(f ) is a group homomorphism,

COROLLARY. Let G,,.. .,G,I he Lie groups. The canonical isomwphicm of the
manifold T(G1 x « x G") onto the manifold T(G,) x . - - x T(Gn) is a Lie
group isomorphism.
pr, is a morphism of G1 x - - - x G" into G‘ and hence T(pr‘) is a morphism
oi‘T(G1 x - - - x G") into T(G.).
PROPOSITION 6. Let G be a Lie group.
(i) The canonical prqioction p: T(G) —> G it a Lie group morphism. .
(ii) The kernel ofp is T,(G). It is a Lie .mhgroup cy’T(G). The Lie group :tmc-
lure induced on T,(G) by that on T(G) i: the Lie group structure of the complete norm-
ohle space T‘ (G).
(iii) The zero section x is an isomorphism of the Lie group G onto a Lie .ruhgroup
(G) of T(G) (which mbgvoup we identm/ with G).
(iv) The Lie group T(G) i: the semi-direct product of G by T,(G) .
Assertion (i) follows from (5). Assertion (ii) is obvious taking account of
Proposition 2 (ii). Assertions (iii) and (iv) follow from (6) and § 1, Proposi-
tion 8.
Let u E T(G) and gE G. By (3) and (4), the products ug, gu calculated in
the group T(G) are the images of u under T(8(g'1)) and T(~{(g)). It follows
from § 1, Corollary 2 to Proposition 17 that the mapping (g, u) >—> gu of
G x T,(G) intoT(G)isanisomorphismof the trivialvectorbundleG x T,(G)
with base space G onto the vector bundle T(G). The inverse isomorphism is
called the left trivializatz'on of T(G). By considering the mapping (g, u) )—> ug,
the right trioializotion of T(G) is defined similarly.
PROPOSITION 7. Let G be a Lie group, M a manifold of clan- C' andf and g map-
ping: of class C" of M into G, so thotfg it a mapping of elm: C' of M into G. Let
m 6M, 7: =f(M),y = g0"), u eTm(M)- Then
(Tfs)" = T(flu-y + x-T(g)"-
Let m be the multiplication of G. Then fg = m o (f, g). Now

TU? 90‘) = (T(f)",T(g)u),

hence T(fg)u = T(f)u.T(g)u. It then suffices to apply (1) withf replaced
by m.
COROLLARY. Le! 7162. The tangent mapping at e to the mapping gag" of G
into C i: the mapping 1 t—~> m: 4fT,(G) into TAG).
For n 2 0, this follows by induction on n from Proposition 7. On the other
hand, the tangent mapping at e to the mapping gHg“ is the mapping
xr—> -x (no. 1, Proposition 2).

236
 CASE OF GROUP GERMS §2.3

Let G be a Lie group, X a manifold of class C’ and (g, x) r—> gx a law of

left operation of class C’ of G on X. Arguing as for Proposifion l, we derive
a law of left operation of class C"‘ of T(G) on T(X), which we shall also
denote by (u, u) Hm}. Identifying G (resp. X) with the image of the zero
section of T(G) (resp. T(X)), we see by (6) that the law of left operation of
T(G) on T(X) extends the law of left operation of G on X. For all u e T,(G)
and veT.(X),by(1>,
(7) w = gv + we.
Hg 5 G and u e TAX), gu is, by (3), the image ofu under the tangent map-
ping at x to the mapping yHgy of X into X. This tangent mapping is an
isomorphism of Tx(X) onto T,,(X). In particular,
(3) g(v + v') = gv + gu', g(M) = Mgu) for a, 11' in T,(X), A E K.
If x e X and u e T,(G), w: is by (4—) the image of 14 under the tangent mapping
at g to the mapping h »—> hx of G into X. Hence
(9) (u + u')x = me + u’x, (M)x = Max) for u, u' in T,(G), A e K.
The above can be applied to the case of a Lie group operating on itself
by left (resp. right) translation. The corresponding law of operation of T(G)
on T(G) is defined by left (resp. right) translation of the Lie group T(G).
Formulae (7), (8) and (9) are therefore valid in T(G).
PROPOSITION 8. Let G1 and G2 be Lie groups, X1 and X2 mung/old: of elm: C'
andfi a law of lefl operation of class C’ q. 1m Xi (i = 1,2).Let4z be a morphism
qf Gl into C: and 44:1 o-morfihbm af X; into X,. Then T(qa) i: a T(¢)-mor[7hum
‘lfT(X1) "“0 T(l
f2 . (4: x ¢)= ma, whence
T(fz) ° (T(¢) >< T010) = T011) °T(fx)-
Let G be a Lie group, X a manifold of class C' and (g, x) »—> £1: a law of
left operation of class C' of G on X. Let I be an open subset of K containing
0 and 7: I —> G a mapping of class C’ such that 7(0) = e. Let
a = To(y)l eT.(G).
Let x E X. Using (4), ax is the image under the tangent mapping to )l »—> 7(1)::
of the tangent vector 1 to I at 0. Hence the vectorfield x >—> we on X i: the ueetar
field defined by the mapping (A, x) »—> 70.)): in the .tense of Diflerentiable and Analytic
Mam‘faldr, R, 8.4.5.

3. CASE OF GROUP GERMS

Let (G, s, 0, m) be a Lie group germ and Q the set of definition of m. Then
T(fl) is identified with an open subset of T(G) x T(G) and T(m) is an

237
III LIE GROUPS

analytic mapping of T(f).) into T(G). It can be verified as in no. 2 that

(T(G), 0.,T(0),T(rn)) is a Lie group germ. The products of G and T(G)
are often written multiplicatively. The canonical projection of T(G) onto G
is a morphism of Lie group germs. The restriction of T,(m) to TAG) is the
vector space addition of T,(G). The zero section of T(G) is an isomorphism
of the Lie group germ G onto a Lie subgroup germ of T(G) which we
identify with G. If f is a morphism of G into a Lie group germ H,
T(f) : T(G) —> T(H) is a morphism ofLie group germs.
The mapping 0): (g, u) »—> gu of G x T,(G) into T(G) is an isomorphism of .
the trivial vector bundle G X T,(G) with base space G onto the vector bundle
T(G); for (I; and d)” are analytic and are vector bundle morphisms, so that
it suffices to apply DWmntiable and Analytic Manifolds, R, 7.2.1. (The proof
of no. 2 could also be adapted.) The isomorphism 42“ is called the left trivial-
ization of T(G). The inverse isomorphism of the mapping (git) Hag is
called the right trivialization.
Let X be a manifold of class C' and d; a law chunk of left operation of class
Cf of G on X. Then T(LIJ) is a law chunk of left operation of class C"1 of
T(G) on T(X) extending 4;. Formulae (7), (8) and (9) remain valid if gx
is defined. If I is an open subset of K containing 0, if y: I —> G is a mapping
of class C1r such that y(0) = e and if a = To(~()1, the vector field mean
defined on X is the vector field defined by the mapping (A, x) ~—> 7(1):: in the
sense of Diferontiolzle and Analytic Manifolds, R, 8.4.5.

§3. PASSAGE FROM A LIE GROUP TO ITS LIE

ALGEBRA

1. CONVOLUTION 0F POINT DISTRIBUTIONS ON A LE GROUP

DenmEON 1. Let G be a Lie group, g and g’ two point: if G and let te T;"°)(G)
t’ e '1“, (G) be two point dimihationr at g and g’ on G (Diflerentiable and Analytic
Manmldy, R, 13.2.1). The convolution product of t and t', denoted by to t’, l: the
zmoge oft ® t' under the mapping (h, h') >—>Izh’ Qf G x G into G (Difwentiahle
and Analytic Manifolds, R, 13.23).
PROPOSITION l. (i) [ft e 'I‘;“(G) and t’ E 'I“,'."(G), then t as t’ e Tfi’”(G).
(ii) If t or 1 ha: no constant term, H: t' ha: no tomtant term.
(iii) e! * ‘v' = 5”,.
(iv) ILet tETfi‘KG), t’ eT‘fiKG) and let f be a funetion Qf elm: C”" in an
open nezghbourhood Iy’gg’ with value: in a Hamdorflpolynormed space. Then

(I * 13f) = (1” h’ '-> (I, tWl')»

= (I, h ’-> GB ’1' ”fall/D)-
238
 CONVOLWON OF POINT DISTRIBIH'IONS ON A LIE GROUP § 3.1

This follows from Dyfmntinble and Analytic Manifolds, R, 13.4.1, 13.2.3 and
13.4.4.

Suppose that K = R or C and that G is finite-dimensional. Then G is

locally compact. If t, t’ are point measures, the definition of t t t’ agrees
with that of Integration, Chapter VIII, § 1. We shall see later that the con-
volution product of measures and that of point distributions are two special
cases of the convolution product of distributions which are not necessarily
point distributions.

Let .7‘“”(G) be the direct sum of the T?”(G) for g E G (cf. Difi’mntiable
and Analytic Manifoldr, R, 13.6.1). We define the convolution product in
31“”(G) as the bilinear mapping of .7‘”’(G) x 3"”)(G) into .7'“’(G)
extending the convolution product of Definition 1. We also denote it by a.
Thus .7“’(G) has an algebra structure filtered by the .7‘”(G). The subalgebra
.7‘°’(G) = 9?“ T‘,°)(G) is identified with the algebra K“) of the group G
over K.
Pkorosmou 2. The algebra 3"”)(G) is madaliw It is commutative if and only
If G is cumulative.
Let tEY“’(G), t’ 6.7““(G), !' 6.7“”(G). Then 1*(1’ t I”) is the image
of t®t’ ® I” under the mapping (g,g’,g”) Hg(g'g”) of G x G x G
into G and (t a: t’) t t” is the image of t ® I' ® t" under the mapping
(g,g’,g") v—> (gg’)g" ofG x G x G into G. Hence (t*t') t t' = H (1’ Ir 1").
It is seen similarly that, if G is commutative, H t' = t’ no: t. If the convolution
product is commutative, G is commutative by Proposition 1 (iii)
PROPOSITION 3. If! E .7“’(G) andg e G, thenflg)“ = e, at t, 8(g)*t = H 9,—1,
(Int g)“ = s, a t m 5,-1. In parlicular, c. i: the unit element f‘”’(G).
Consider the diagram
G 34> G x G L» G
where (I) is the mapping 1: v—> (g, h) and 44 is the mapping (h', h) »—> h'h. Then
Y(£) = 4! ° ‘1 and hence fig):t = 41.0140) But 4>*(').= =. 8 I and hcnce
¢.(¢*(t)) = c, t- t. The argument is similar for 8(g)*t. Fmally,
Intg = Hg) ° 8(.3)
and hence (1m .9). = fig): ° Mg).-
It is therefore seen that, for teT(G), e, t l and It s, are equal to g!
and 3g calculated in the group T(G) (§ 2, no. 2). But it should be noted
that, for t, l’ in T(G), the product It’ in the sense of § 2 is in general dif-
ferent from t n t'.

239
m LIE GROUPS
af the
DEFINITION 2. Let G be a Lie group. TIIL‘ .mbalgehra nffl"“"(G) consisting
distributions with support contained in a is denoted by U(G).
This algebra is filtered by the subspaces
U.(G) = U(G) n y"’(G) = Tl"(G)-
We write U+(G) = T‘,"”(G), U: (G) = U*(G) n U,(G) (cf. Difi'mntiable
and Analytic Manififlds, R, 13.2.1). Recall that Uo(G) is identified with K and
U: (G) with the tangent space T,(G), In U(G), U" (G) is a two-sided ideal -
supplementary to U0(G).
Example. Let E be a complete normable space considered as a Lie group.
Then the vector space U(E) is canonically identified with the vector space
TS(E) (Difimntiable and Analytic Manifolds, R, 13.2.4). Let m: E x E—>E
be addition on E. Then
1a,: TS(E x E) —>TS(E)
is equal to TS(m) (tnzntiable and Analytic Mani/blah, R, 13.2.4). For t, t’ in
U(E) = TS(E), the image t a t’ of the symmetric tensor product t ® t' under
m... is therefore TS (m) (t ® t'). By Algebra, Chapter IV, §5, no. 6, Proposition
7, this image is just the product tt’ in the algebra TS(E). Thus the algebra
U(E) is identified with the algebra TS(E).
PROPOSITION 4. Consider the bilinear mapping (:4, u) D—> a It 1/ (resp. (u, v) ~—> v a u)
of U(G) ® K“) into 9"”)(G). The corresponding linear mapping qf U(G) ® K‘“’
into 9"”)(G) is a 1mm spa“ ismrphiwn.
K“) is the direct sum of the K5,, for x e G. On the other hand, the mapping
u l—> a It a, (resp. u >—> c, e u) is an isomorphism of the vector space
U(G) = fl”7(G) onto the vector space 9:”)(0) by Proposition 3. Finally,
.7"°‘”(G) is the direct sum of the T§”)(G) for g E G.
Let X be a manifold of class C' (r 2 co) and xEX. We have defined
(Difmntiahle and Analytic Mamfilds, R, 13.3.1) a canonical filtration on the
vector space 9';°°’(X) and a canonical isomorphism ix”, of the associated
graded vector space onto the graded vector space TS(T,‘(X)). In particular,
let T,(G) = L; then i9”, is an isomorphism of the graded vector space gr U(G)
onto the graded vector space TS(L). But U(G) is a filtered algebra, from which
we obtain a graded algebra structure on gr U(G).
PROPOSITION 5. The isomorphism £5.92 gr U(G) —» TS(L) is an algebra iso-
morphism.
Let p be the mapping (t, t') >—>t® t’ of U(G) x U(G) into U(G x G).
Let c be the mapping (1, t') »—> t It I’ of U(G) x U(G) into U(G). Let m be
the mapping (g, g') H gg' of G x G into G. Then by Definition 1
(1) c = mop.
240
 FUNCTORIAL “cram-ms § 3.2

Consider the diagram

grU(G) x grU(G)fl>grU(G x G) fligmo)
«mate... luxofl Io..
TS(L) x TS(L) —‘>TS(L x L) m» TS(L)
where q is the mapping derived from the canonical isomorphism of
TS(L) x TS(L) onto TS(L x L). By Diflmnliable mi Analytic Manifaldr, R,
13.4.6 and 13.3.5, the two squares of the diagram are commutative. Hence
by (l) the diagram
gr U(G) >< gr U(G) —>"“’ grU(G)
«Loam. 19,.
TS(L) x TS(L) _.“"“"”“ TS(L)
is commutative. Now T(m): L x L—>L maps (x,y) to x + g (§ 2, no. 1,
Proposition 2 (ii)). By Algebra, Chapter IV, §5, no. 6, Proposition 7,
TS(T(m)) a q is therefore the multiplication of the algebra TS(L).

2. FUNCTORIAL PROPERTIES
PROPOSITION 6. Let G, H b: Lie groups and 115 a morphism (If G into H. Fur t, t’
in 9"”(0): 4M! * V) = 4M!) * 4%")-
Consider the diagram
G x G —”‘> G
«no 0
H x H —"> H
where m(g, g') = g', 710:, h’) = hh’. This diagram is commutative. Hence

Mt't’) = ¢*(ms('®l')) = ":(Gb X ¢)*(‘®"))

= "*(M!) ® 4’30,» = M!) * dub”)-
The Lie groups G and Gv have the same underlying manifold and hence
the vector spaces 9"”’(G) and 37"”)(GV) are the same. Let 0 be the mapping
g>—> g“, which is an isomorphism of the Lie group G onto the Lie group
G". Then 0* is an automorphism of the vector space .7‘”’(G), which auto-
morphism we denote by l»—> IV . Then (1:4,)V = 5,-1. If I ETAG), then
(2) IV = —t (§ 2, Proposition 2).
Example. Suppose that G is the Lie group defined by a complete normable
space E. Then U(G) is identified with TS (E) and the restriction 0* to U(G)
is identified with TS(T,(0)) (Dgyemuiabze and Analytic Manifolds, R, 13.2.4).
Therefore, if t E TS’(E), Iv = (— l)‘L

241
 m 1m GROUPS

PROPOSITION 7. Let G be a Lie group. Let t, t’ be in 9"“’(G). I

(i) The product It t' mlmlaled relative ta G" Lt equal to the product t *t eul-
eulated relative to G.
(ii) (tu’)" = I" *t”.
Consider the diagram
(G1 XG2)X(G1 x (194’ G1." Ga

' P170:
(G1 x GI) x (Ga x Ga)

where :(g, g’) = (g', g). m(g, g’) = gg’, "(.43. g') = g’g for all g, g’ in G. This
diagram is commutative. Hence n.(! ® t') = m.(:*(t ® t')) = m.(t’ ® t).
This equality is precisely (i). Assertion (ii) follows fi‘om (i) and Proposition 6.
Pnoposmon 8. let G, H be Liegraupx andtb anwrphirm ofG in!» H. [ft 5 .7"“”(G),
thenLetMr) = (MW.
0 (resp. 0’) be the mapping gag—1 of G into C (rap. of H into H).
Then 4: ., e = e' . <1», whence MM!» = sum»-
PROPOSITION 9. Let G”- - -, G,I be Lie group: and G = GI x - - - x G". [fthe
vector spam 51"”(G) and 9"”)(G1) ® - ~ ~ ®9""”(G,,) are canonically identi-
fied, the algebraY‘“)(G) ixthetmmproduc!qfthealgebrax.7‘“’(G,), . . . . 3"”)(Gn).
{flief‘wKGwi = 1,. . .,n, then
(’1® ®tn V = li® @11-
It suffices m consider the case n = 2. Let I], I; be in 9"”)(Gl), I3, I; in
9"“)(69. We need to show that (11 ® :2) a: (t; ® 1;) = (t, : 1;) ® (ta I t,’.)
and that (11 ® 19V = l} ® 1;. Consider the diagram

G x G —‘> G x G

where m((x1, x2), (:4, x9) = (xlxi, 1,43),

"((xn #2), (xi, 1/2)) = ((#1) xi), (#2, x9),
plot” :4) = 213;, p,(x,, x;) = mg. This diagram is commutative. Hence
"MU; 812) 69 (II ® 14)) = (I); ®l12)s(":((11 @12) ® ('3 ® 19)),
that is
(t: 5912) * (Ii 8’ ti) = (P1 ®Pa).((li ® ti) ® ('2 ® '9)
= [Hoax ® ti) ®Pu(’2 ® ’4)
= (II t t;) Q; (t, * pg).

242
 FUNUI‘ORJAL vacuums § 3.2

It is seen analogously that (t1 ® [2) " = t} ® t}.

PROPOSITION 10. Let H b: a Lie subgroup of G and i: H —> G the canonical ity'zt-
h'on. Then 1', is an injecting hornonmphixm of the algebra 5"”)(H) into the algzbra
27"”)(H) and i,(tV) = (i‘(t)) " for all 15.7““)(H).
This follows from Propositions 6 and 8 and DWzmltiabLe and Analytic Mani-
fildr, R, [3.2.3.
31“”(H) is identified with a subalgebra of Y‘”’(G) by means of the iso-
morphism of Proposition 10.

Remark. Proposition 10 remains valid if H is a Lie quasi-subgroup.

We recall (Difzrmh'oble and Analytic Manifoldr, R, 13.5.1) that, if V is an

analytic manifold over K, 91°”(V) has canonically a cogebra structure over
K with a counit; the counit is the linear mapping of 9"“(G) into K which
associates with each element of ’l'S,”(V) is constant term.
PROPOSITION 11. Let G be a Lie group.
(i) The cogebm Y‘”)(G), with convolution, i: a bigebra (Algzbm, Chapter III,
§ l1, no. 4).
(ii) Let c be the ooproduot on 9"”)(G). Let :6 IMKG) and write

:0) = 1.21 t. ® :1.

n
Then 0(t") = ”21 p: ® :gV.
We prove (i). In the definition of bigebras referred to, condition (I) fol-
lows from Propositions 2 and 3 and condition (2) follows from Diflmnfiobls
and Analytic Manifolds, R, 13.5.1. Let d be the mapping g 1—» (g, g) of G into
G x G. Then 0 = d. and hence c is an algebra mo hism (Propositions 6
and 9), which is condition (3). Let te'l'?”(G), t’ e T? (G) have no constant
term and A, 7.’ be elements of K; then :, ® 11', t8) 2,, !® 2’ are without
constant term (Diyfirenlioble and Analytic Mamfildx, R, 13.4.1) and hence the
constant term of (key + t) no: (#2,, + t’) is NJ; hence condition (4) holds.
We prove (ii). By Propositions 8 and 9,
n
co”) = zw) = (4.0))” = (Z 1. ® :1) v = Z 2." o to.
PROPOSITION 12. Let G, H be two Lie group: and d) a morphiwn of G into H. Then
(I), i: o bigehra morphism o‘”)(G) into .7‘”’(H).
This follows from Proposition 6 and Diflnzntiable and Analytic Munifiltir, R,
l 3.5. l .

243
 in ma GROUPS

Let G be a. Lie group. The restrictions of the convolution and the coproduct
to U(G) define a. bigebra structure on U(G). We have U(G)" = U(G).- If
(h: G—>H is a Lie group morphism, we denote by .U(¢) the mapping
tI—>¢,‘(t) of U(G) into U(H); this is a bigebra morphism. If 4;: H—>L 15
another Lie group morphism, then U(tl; a (b) = U(IJI) a U(o). If 4) 15 an im-
mexsion (resp. a submersion), U(rlz) is injective (resp. surjectwe) by Dz mn-
tiable and Analytic Manifolds, R, 13.2.3. In particular, if H is a Lie subgroup
of G, U(H) is identified with a subalgebra of U(G), the coproduct on U(H)
being the restriction of the coproduct on U(G). If H is open in G, then
U(H) = U(G). If G1, G1 are Lie groups, U(G1 x 0,) is identified with
U(G,) >< U(Gz). The primitive elements of U(G) are those of T,(G) (Dif-
fermtiable and Analytic Manifialds, R, 13.5.3).
Again let (I): G—>H be a Lie group morphism. If gr U(G) is identified
TS(T.(G)) and grU(H) with T$(T.(H)), then grU(¢) is identified with
TS(T,(¢)) (Difirmtiable and Analytic Mang'fizlds, R, 13.3.5). We apply this to
the isomorphism g» g" of G onto G"; then T,(¢) = —l and hence

(3) teU,(G) => t" s (4):t U,_,(G).

3. CASE OF A GROUP OPERATING ON A MANIFOLD
Let G be a Lie group, X a manifold of class C" and f a law of left operation
of class C' of G on X. If!eT§"(G) and u eTfif'KX) and .r + s' S r, we de-
note by u u the image of t ® u under f, We extend the product It: to a.
bilinear mapping also denoted by tr, off"’(G) + .7""(X) into .7'"""’(X).
Proposition 1 of no. 1 can be extended with obvious modifications to the
present situation.
When G operates on itself by left translation, we recover Definition 1 of
no. 1.
Pkorosmcm 13. Let t e .7'"’(G), t’ E Y‘WG), u e .7'""(X), such that

s + a" + s' < 7.

Then (Ht') ant = t* (t'w).

This can be proved as is Proposition 2 of no. 1.
In particular, if 1 S 00, the vector space %”’(X) is a left module over the
algebra .7‘”’(G) with the product :.
PROPOSITION 14-. (i) Let g.7 e G and e(gu) be the mapping x H-f(g , x) of X into
X. _I.fu E 9""(X), then 1(go),u = en t u. 0
(11) Let x0 E X and 9(19) be the mapping g Hf(g, x") ts into X. If! E T")(G),
than p(xo)*t = ta: an.
This can be proved as is Proposition 3 of no. 1.

244
 convournou or pom-r ms-numn-Ions AND wucnons § 3.4

In particular, ifu eT(X) and IE T(G), e,” nu u and M em are equal to the
products gun and txo defined'In §2, no. 2.
PROPOSITION 15. Let G (resp. G’) be a Lie group and X (resp. X') a maniald of
clan C'. Suppox: that a law 12/ lefl 017mm,; qfclm: (3' cf G (rap. G’) on X (resp.
X’) is given. Let 21> be a morphixm If G into G' and (ll a tb—mmphism Qf X into X’.
Le! IEF‘WG), u eT‘"’(X) be such that: + s’ g 7. Then
‘Mt * U) = (Mt) * 4401)-
This can be proved as is Proposition 6 of no. 2.
Remark. Let f be a law of fight operation of class C' of G on X. 1ft 6 3"”(G)
and 1459'“”(X), with 3 + s’ < r, we denote by 11*! the image of u®t
underf,. Propositions l3, 14-, 15 go over to this situation in an obvious way.
PROPOSITION 16. Let G, G' be Lie groupx, X a mamfald 9)“ class C' and supper:
that G (resp. G’) operate: an X on Ike left (resp. right), with (gx)g’ = g(xg') fur
all xeX, geG, g’eG’. Let tefl'"’(G), I'ei"”(G’), t'eT‘"’(X), with
:+:’+:"<r. Then
(tu') u’ = 1* (t'u‘).
(t at t") t t’ (reap. H (t' H’)) is the image oft ® I” ® 1' under the map-
ping ( g, x, g’) »—> (gx)g’ (resp. g(xg’)) of G x X x G' into X.

4. CONVOLUTION OF POINT DISTRIBUTIONS AND FUNCI'IONS

Let G be a Lie group, X a manifold of class C' and (g, x) I-> gx a law of left
operation of class C' of G on X. For all apex, let p(x) denote the orbital
mapping of x.
DEFINITION 3. Let lei—“’(G) with x S r. Letf: X-—> F be afimctian qftlass C"
with values in a Hmdozflpolymrmed xpace (for exampk F = K). The convolution
aftandf, dmlzdbyltfiixlhefincfimoaith vdwsiacfirwdby
(”13(5) = <tv unfi-
Then
(“10(1) = (9(z)*("):f> (n0. 3, Proposition 14 00)
(4-) = (tV ,fo p(x)) (Difi @Anal. Mam, R, 13.2.3)
= (t, (fo p(x))") (Difll 62° Anal. Mam, R, [3.23).
Note that Definition 3 can also be written in the more symmetric form
(5) (Ex, ‘ *f) = (1" * emf)-
The function (g,x) »—>f(gx) = (fa p(x))(g) on G X X is of class C'. By
Difmntiabh and Analytic Manyblds, R, 13.4.4, the function 1:» (tv ,fo p(x))
is therefore of class C’ " if .r < 00. In other words, if .r < 00, la:f i: If plan
0".

245
m LIE GROUPS

Clearly t :0:f depends linearly on t andf.

Formula (4) implies in particular, for g E G,
(6) (s, tf)(x) =f(g“x)
that is
(7) a, *f = 7(5))?
Suppose that K = R or C, that G and X are finite-dimensional and
tf
that X has a positive measure invariant under G. The definition of s,
agrees with that of Integration, Chapter VIII, §4, no. 1 (cf. formula (2),
lac. tit).

PROPOSITION 17. Let te.7'"’(G), t’ 53"“(X) andf: X-—> F afimch'an nfclass

C'wiih:+:’< r. Then
<t', Hf) = (‘V '1',f>~
<13 Hf) = (2',x'-><t,g'->f(g"*)>> by (4)
= (2 ® t’, (g,x) »f(g'1x)> (Dif Ct? Anal. Mam, R, 13.4.4(
= (1" ® t’, (g, x) Hf(gx)) (D117. 6 Anal. Mam, R, 13.2.3)
= (tv # t’,f).
PROPOSITION l8. Le! IGWKG), t’ 5.7“"(G) andf: X—> F afmwtian nfclau-
C', with: + 3' S r. Then
(”I’) *f= tx(t'#f).
For all x e X,
<3» (H ") *f> = ((WI')v * zmf> by (5)
= 2"“ a (tV * ex),f> (Propositions 2 and 7)
= (3" at a,“ t’ if) (Proposition 17)
= (Ex, I * (t’ *f)) (Proposition 17).
If r 2 no, we see that the set of functions of class C" on X with values in
F is a left module over the algebra .7“”(G).
PROPOSITION 19. Le! ley‘”(G), with s g r. Lat f (resp. f’) b: a fumtian uf
clan C'on X with value: in a Hamdurfl pnlynarmed space F (resp. F’). Let
(u, u’) Huu' be a continuum bilinzar mapping qf F x F’ inlo a Huwdarfi' [July-
normed .rpace F', to that ff ’ i: afimm'on 9f clan C' on X with value: in F”. Let

2; t, ® 1; be u» imag: m in yum) ® yum) under the cnpmdud. mil

u (ff’) = Z, (t. *f)(h'~f’)-

24-6
 CONVOLUTION 0F POINT DISFRIBUTIONS AND FUNCTIONS §3.4‘

Let x E X and let 9(3) denote the orbiml mapping of x. Then

<2,” : : or» = «V. (m o p<~>> by (4)
= <22 (f° 9(X))(f’ o W)»
= ,2, «Mo 9(x)><!.".f‘ o For»
(Difli a Man. Am, R, 13.5.2)
= 2 <2. t. *f><sm t.’*f’> by (4).
Remark 1. Let G be a Lie group, X a manifold ofclass C' and (fig) >—>xg a
law of right operation of class C’ of G on X. If tefl’KG) with .r < r and
f: X—>F is a function ofclass C’ on X, we denote byftt the function on
X defined by
(5,!“ I) = <sx * 1" J)
(a) = <p<x).(r),f>
= <1v ’f“ 90‘»
= (t, (f° 9(3))”-
In particular
(9) (f* “0(5) =f(x£")
that is
(10) f* in = 3(3)")‘3
Propositions 17, 18, [9 become, in the obvious notation,
(11) <t’,f~ t> = <t' w,» '
(12) f*(t#t’)=(ftt)tt’

<13) (m u = 2 on too“ : :1).

PROPOSITION 20. Let G, G' be Lie gmups, X a manifold oftlass C' and (g, x) .—> gx
(mp. (x, g’) e—> xg’) a law If lgfl (resp. right) operation If class C’ of G (resp. G’)
on X. Suppou that (gx)g’ = g(xg’) for all x EX, geG, g’ e G’. Let :6 9""(G),
t' 6.7("’(G') amif: X—>F b: afimction afclasx C’such that: + s’ s r. len
(Hf) *I’ = [*(fakt').
For all x e X,
(t:,('*f)*"> = (Ex*l'v,t*f> bY(3)
= (1" =t (a, :u t”),f) (Proposition 17)
= (tV 1k emf: 2’) (Proposition 2 and (11))
= (5:! t‘ (f* 1’» by (5)-
247
m LIE GROUPS

In particular, consider G as operating on itself by left and right transla-

tions. Iff: G —> F is a function of clas 0" on G and t e .7‘"(G) (with s g r),
In:f and f t I are, if .r < co, functions of class C' " on G. Further, let
t' 59"”)(0), with 5 + .r' S 7. Then
(14) (uf)*z' =u(f»t’).
In particular, ?”(G) is a (.7(”’(G), y‘”’(G))-bimodule. Formulae (5) and
(8) admit as special cases
(15) (hf) = (55) 1" *f) = <¢.,f* 1")-
Remaik 2. Let (g, x) »—> gx be a law of left operation of class C' of G on X.
Let te U.(G) with x < 7, Q be an open subset ofX andf: O —> F a function
of class C'. tauf can also be defined by formula (4) or (5) ; it is a function
defined on Q with values in F, of class C"’ if 3 < so. The results of this no.
extend in an obvious way to this situation.

5. FIELDS OF POINT DISTRBUTIONS DEFINED BY THE ACTION OF A

GROUP ON A MANIFOLD
Let (g, x) >—>).(g, x) = gx be a law of lelt operation of class C’ of G on X.
Let .t S r and teU,(G). For all xeX, teexE'IgKX). The mapping
2: n—> I a 5,, is called tlufidd qaint distribution: defined by t and the anion qf G on
X and denoted sometimes by D: or simply D). Let Q be an open subset of
X and F a Hausdorfi‘ polynomed space. Iff2 (2 —> F is of class C' and .t < r,
the function 1" ef on Q is also denoted by DJ. Then
(16) (D,f)(x) = (Hemf).
If r < 00, then D‘fs 3""(Q, F) by no. 4. Thus fv—> tis a mapping of
V19, F) into V'"(fl, F) (often denoted by D, by an abuse of notation).
IfteU,(G), t' e U.»(G) and x + x’ < r, then, by Proposition 18 of no. 4,

(17) Deaf = Dr(Dcf)

and hence, using the abuse of notation indicated above,

(18) DM' = Dt' ° D:-

Suppose that G and X are finitehclimensional. The mapping (I, x) >—> t ® ex
of T“’(G) x X into the vector bundle 'I‘”(G x X) (cf. Diflmntiabls and
Analytic Maniflildr, R, 13.2.5) is of class C"'. Hence (Dzfirmtiable and Analytic
Manif‘alds, R, 13.2.5) the mapping (1‘, x) >—> t t 5x of T")(G) x X into the
vector bundle T‘"(X) is of class C"’. In particular, D¢ is a differential opera-
tor of order <3 and class C" " in the sense of Diflnentiable and Analytic Mani-
folds, R, 14.1.6. By formula (16), the function D,f is then the image off
under this difl‘erential operator (Dgfimifiabl: and Analytic Many'aldx, R, 14.1.4).

248
 mvnmm‘ 12mm OF POINT DISTRIBUTIONS § 3.6

We now no longer suppose that G and X are finite-dimensional. Let 4:

be an automorphism of the manifold X and A a field of point distfibutions
on X. Conforming with the general definitions, the transform of A under 4/
is the field ofpoint distributions on tose value at 440:) is ¢*(A(x)) ; we denote
this mapping by MA). Hg 6 G and 1(g) denotes the automorphism x >—> gx
of X, the transform of A under 1-(g) is also called the transform of A under g.
PROPOSITION 21. Let (I: be an autmnarphixm of X commuting with the operation: qf
G. Tim: D, is invariant under (It.
For all x E X,

(¢(D.))(¢(x)) = ¢*(D¢(*)) = ‘i’#(t*‘x)

= tt 4;.(ex) (Proposition 15)
= H We) = Dx(¢(x))~
Pnovosmou 22. Ifg e G, the transform q, under g if D,,.,.‘,_‘.
The value of this transform at gx is

1094110)) = 1(3):“. ts)

e, e (t as: ex) (Proposition 14 (i))
(a, 4: I s =,- I) t 5,, (Propositions l and 2)
= D.,....,-=(gx>.
Let (x, g) >—> p.(x, g) = xg be a law of right operation of class C' of G on X.
Let s < r and teU,(G). For all xeX, ex*te'1(;)(X). The mapping
x »—> e, an t is called the field of distributions defined by t and the action of
G on X and is sometimes denoted by D‘,‘ or simply D,. Let Q be an open subset
of X. Iff: 0 —> F is of class C", the functionfak 1" is denoted by D‘f. Then

(19) (DJW‘) = <51: * Lf>

and, in the obvious notation,

<20) Dwf = D:(Drf)

(21) BM = D, o Dy.

Proposition 21 remains valid. Let g e G. The transform of D, under g (that is

under the automorphism x 5—» xg of X) is Dfi—xmgr

6. INVARIANT FIELDS OF POINT DISTRIBUTIONS ON A Ll}: GROUP

DEFINI'I'ION 4-. Lg! G be a Lie group. A field afdixtribulim an G ix callzd left (resp.
right) invariant if it is invariant under left (resp. right) translation: If G.

249
III LIE GROUPS

if
In other words, a field of distributions g »—> A, on G is lefi invariant
A”, = Y(g)*l5n’ forg, g’ in G,
agam' if
or A", = e, e A, for g, g' in C.
It is right invariant if
Aw = 3(£"‘)*Ag forg,g'in G,
or A,,» = A, s em for g, g’ in G.
DEFINITION 5. Let G b: a Lie group and l e U(G). Let L, denote thefield of dirm'baL
lions-gas,ttanGandR,tfieldq/dirhibutiomgv—>t*e,MG.
In other words, L, (mp. R,) is the field of distributions defined by i and
G operating on G on the right (rap. left) by means of the mapping
(g, g’) >—> gg'. Let Q be an open subset of G and F a Hausdorff polynomed
space; iffe Wm, F), then L,f=f* t" s «”(9, F) and
R,f= 2V *fe@”(Q,F)
(no. 5). If G is finite-dimensional, the differential operators L, and R, are of
class C" (no. 5).
PRoposmON 23. (i) 7714 mapping tv—>L, (reap. I»R,) is an isomorphism qf
the vector rpacs U(G) onto the vector field qf lefl (resp. right) inunn'anl dimibuiim
an G
(ii) For t, t’ in U(G), L,.,, = L,o Ly, R, ,r = R, a R,, L, oRr = R,’ a L,
(with the abuse of notation firm. 5).
(iii) Ife is the mappinggHg‘l afG onto G, then e(L,) = Ry
(iv) 1ft sU<G> mtg e G, the» (L0. = (Rant-1)..
In G every right translation commutes with every left translation. By Proposi-
tion 21 of no. 5, L, is therefore left invariant, As (L,), = t, the mapping t>—> L,
is injective. Let A be a field of left invariant distributions on G; let 2 = A,;
then A and L, have the same value at e and are left invariant and hence
A = L,. This proves (i) for L, and the argument is similar for R,. The formulae
L,_,. = L,o Ly, RM = R, 0R, follow from (21) and (18). Let tsU,(G),
t' e U,,(G),fE W“), F), where Q is open in G and .t + .t' S r; then
mw= w m = (1” sf) u
=tW a(f*l") (PropositionQO)
= R1’L,f
and hence L, o Ry = Ru 0 L,. As 0 is an isomorphism of G onto GV, 0(L,)
is a field of right invariant distributions on G; its value at a is 0‘(t) = (V;
hence 9(L,) = R,.. Finally,
(L1); = en *t = (an * t * Ea") ‘ a, = (Rink-Iran-

250
 LIE Amen“. or A LIE snoop §3.7

Remark 1. It is the action of G on itself by right translation which defines the

fields of left invariant distributions.
Remark 2. Suppose that G is finite-dimensional. The mapping

(be) H (11,), = 1* E:
of U,(G) X G into T‘"(G) is an isomorphism of analytic vector bundles; for
this mapping is bijective, linear on each fibre and analytic (no, 5); on the
other hand, let (I): T“’(G) —> U,(G) x G be the inverse bijection; if t e Tf,” (G),
then (Ht) = (t*e,—1,g) and hence (I) is analytic. The isomorphism 4) is
called the right trivialization of T‘"(G). Similarly, consider the mapping
(t, g) ~—> (L,), = ey *t of U'(G) x G into T‘"(G); the inverse isomorphism
is called the left trivialization of T‘”(G). By restriction we recover the right
and left trivializations ofT(G) (§ 2, no. 2).

7. LE ALGEBRA OF A LIE GROUP

Let G be a Lie group. In U(G), as in any associative algebra, we write
[t, t'] = to t’ — t’ It t. As T,(G) is the set of primitive elemenm of U(G),
['I‘,(G), T‘(G)] c T,(G) (Chapter II, § 1, no. 2, Proposition 4). The restric-
tion of the bracket to T,(G) therefore defines on T.(G) a Lie algebra struc-
ture.
lemma 1. Let X and X' be complete nonnable rpaeer, X0 an open neighbourhwd of
0 in X and f an analytic mapping of X0 into X’ melt that f (0) = 0. Let
1‘”=fl +f2 +fa + . - - betheexpamionoffaranintegralrm'exabouto,wherefiisa
homogeneour continuous-polynomial of degree 1' an X with value: in X’, Let t be an element
y" TS“(X), eomidered as a point distribution on X with rapport contained in {0}. Let
t’ =f*(t) e TS(X’). The homogeneous component oft’ ofdegree l i: <f3, t).
Let t; be this component. Then, for every continuous linear mapping u
of X’ into a polynormcd space,
a“) = (1', 14) because u is continuous and linear
= (I, u of) (Diff Sf Anal. Mam, R, 13.2.3)
= (I, u of2) bemuse I e TS’(X)
= u(<t,f2>) (Dgfli 6? Anal. Mam, R, 13.2.2),
whence the lemma.
PROPOSITION 24-. Let G be a Lie group and (U, b, E) a chart on G rue]! that
Me) = 0. Let V be an open neighbourhood qfe melt that V2 C U. let m be the
analytic matting (a, b) H 4>(<l>“‘(a)<l>"(’1)) l1f<l>(V) X NV) into E- M
7": “I?“
2 "‘I.’

251
"I LIE GROUPS

be the expansion Qf m at an integral series about (0, 0)? where ml: , ts a bxhomogeneou:
continuum-polynomial of bidegree (i, j) on E x E math valuex m E.
(i) m“D = “0,1 = Oforalliaé l and] #1.
(ii) ml'a(a, b) = a and mo,1(a, b) = bfar all a GE, b EE. .
(iii) let 4;: T.(G) —> E be the deflerential qfd) at e. For all a, I) m T,(G),

M1601) = ”11.101410, 4201)) - "11.101410, M10)-

m(a, 0) = a, m(0, b) = b for all a, b in ¢(V), which proves (i).and (ii).
Let u, 11 be in TAG). Let T0(E) be identified with E and hence (ll WIfll T, (M.
The images of a and a under T,(<]>) are Ma) and 44(0). The tensor product
point distribution of these images is the symmetric product of (Ma), 0) and
(0, 4:01)) in TS(E x E) = TS(E) x TS(E), thatis
(WI). 0) ® (0’ 41(0)) + (0, W» 69 (Mu). 0)-
Hence 4> a (u * 1/) is the image of the above element under the mapping m
of ¢(V) x ¢(V) into E. In component of degree I in TS(E) is, by Lemma 1,

x = ("11.1. (Mu): 0) ® (0, 440)) + (0, 41(0)) 8 (Mil); 0)>-

We define a bflinear mapping a: (E x E)’ —> E by

"((4) b), (4', b')) = ӣ1,101, V)-

Then rt((a, b), (a, b)) = m,,1(o, b) and hence

a: = <n, (W), 0) o (o, w» + (o, w» to (an), 0»

= “1.10900: 4407)) + “1.1(0: 0) = ”1.10“”; 440’»
Similarly, (I; x: (o #14) admits ml,,(¢(v), 4504)) as component of degree 1 in
TS(E). As ¢([u, 12]) is of degree 1, this proves (iii).
COROLLARY. The normable space T,(G) together with the bracket, is o normoble
Lie algebra.
DEFINITION 6. The narmable :paee T,(G), together with the bran/eel, i: called the
nannable Lie algebra 13/ G, or simply the Lie algebra of G, and i: denoted by L(G).
PROPOSITION 25. Let G be a Lie group mid E(G) the enveloping algebra cf L(G).
The eanonical {ejection of L(G) into E(G) defim a homomorphism 0 of the algebra
E(G) into the algebra U(G). If K is of characteristic 0, 1) 1'3 a bigebra ixornarphixm.
The bigebra U(G) is cocommutative (Dmnentiable and Analytic Mangfoldx,
R, 13.5.1) and the filtration (U,(G)) is compatible with the bigebra structure.
The set of primitive elements of U(G) is L(G), It then suffices to apply chap-
ter II, § 1, no. 6, Theorem 1.
When K is of chamcteristic 0 we shall in future identify U(G) with the

252
 1.12 ALGEBRA or A us. GROUP § 3.7

enveloping algebra of L(G). By (2) and Proposition 7 (ii), the mapping

t1—>tv of U(G) into U(G) is then identified with the principal antiauto-
morphism of U(G) (Chapter I, §2, no. 4-).
PkorosrnoN 26, Suppose that K is of characteristic p > 0. For all aEL(G),
a” e L(G) and ad(a’) = (ad a)” (the power a” being cahulated in U(G)).
If a e L(G), a is primitive in U(G), hence a" is primitive in U(G) (Chapter
11, § 1, no. 2, Remark 1) and hence a” e L(G). Let a“ (resp. 1.3,) be the linear
mapping x 1—> a e x (resp. max 4: a) of U(G) into U(G). For all x E U(G),
(ad a)(x) = (an — ca)(x) and hence (ad a)” = (ca —— 13,)”. But “a and 7,,
commute and therefore (a, — Ta)" = (6.)” — (13,)” = 1,19— 6", whence
the second assertion.
DEmON 7. Let X be a mmfild of class C’ (r 2 2) and g a complete nonnable
Lie algebra. A law of irfinitesimal lg? (resp. right) operation of class C”1 of s
an X is a mapping a >—> D4 of 3 into the set of vector fields on X with the following
properties:
(a) the mapping (a, x) »—> D,(x) is a morphism of class C"1 of the trivial vector
bundle 9 x X into the vector bundle T(X) ;
('3) [D5, Do] = ‘Dxam (“3131 [Du Do] = Dmmllfé” a” a, b in Q-
In particular, each vector field DA is of class C"‘.
Remark. Let X be a manifold of class C', 9 a finite—dimensional Lie algebra
and a HD, 2 linear mapping of 9 into the vector space of vector fields of
class C"1 on X. Then condition (a) of Definition 7 holds. For, by consider-
ing a basis of g and applying Diferentiable and Analytic Maniflzlds, R, 7. 7.1
the problem 15 reduced to the case dim g = 1 and our assertion is then ob-
vious.
PROPOSITION 27. Let G be a Lie group and X a manrfild of class 0'. Suppose that
a law qflefl (resp. right) operation qfcloss C’ ofG on X is given. For all a E L(G),
let D, be thefield ofpoint distribution: defined by a on X.
(i) The mapping (am) HDu(x) is a morphism of class C"1 of the trivial
vector bundle L(G) x X into the vector bundle T(X).
(ii) Let I be an open subset :y' K containing 0 and y: I —> G a mapping of class
C’ such that Y(0) = e. Let a = To(y)l E L(G). {ff is afunction ofclass C’ on
an open subset of X, then
(D,f )(x) kelim o‘k ‘(f (y(lc)x) —f (x)) if G operates on the le ,
(Daf)(x) “15311:? ‘(f(*'r(k)) -f(*)) if G ”PM!“ 0'1”“ right
(iii) If r> 2, theKmapping a»—>D is a law of iryinitesimal lefl (resp. right)
operation Ifclass C.’ ‘ ofL(G) on X.
Suppose that G operates on X on the left. Let (I): G x X —> X be the law
of operation. Then T((b) is a (lb-morphism of class C"1 of the vector bundle
T(G) x T(X) into the vector bundle T(X) (Diflhentiable and Analytic

253
II! LIE GROUPS

X X) is
Manifaln's, R, 8.1.2). The induced Vector bundle (T(G) X T(X))[({e)
identified with E = L(G) x T(X). Hence T(¢)|E is a vector bundle morph-
ism of class 0"]. For (a, x) e L(G) x X, T(¢)(a, x) = Dn(x), whence (i). .
The formula giving (D,f) (it) follows from §2, the end of no. 2, and Dif-
fnmtiable and Analytic Manifolds, R, 8.4.5. .
Suppose that r 2 2. Let a, b be in L(G) and f be a function of class C'
on an open subset of X. Then
Dimalf = Dh(Daf) — Du(Dbf) by (17)
= [D,, D,]f (Dif {‘9‘ Anal. Mam, R, 8.5.3).
Let I: 5 X. By takingf to be a chart on an open neighbourhood of x, it follows
that Dlnlbfix) = [DM D3] (1:), whence (iii). The argument is similar if G
operates on X on the right.
When 7 2 2, the mapping a H D, is called the law of infinitesimal opera-
tion associated with the given law of operation.
8. FUNCTORIAL PROPERTIES OF THE LIE ALGEBRA
Let G and H be Lie groups and (I; a morphism of G into H. The restriction
of U(<]>) to L(G), which is just T,(<]>), is a continuous morphism of L(G) into
L(H), which we denote by L(tb). If 4; is a morphism of H into a. Lie group,
then LOP ° (i) = M4!) ° L(tb)‘
For (I) to be an immersion, it is necessary and suflicient that L(o) be an
isomorphism of L(G) onto a subalgebra of L(H) admitting a topological
supplement. In particular, if G is a Lie subgroup of H and 4’ is the canonical
injection, L(G) is identified with a Lie subalgebra of L(H) by means of
L((lz). More particularly, if G is an open subgroup of H, L(G) = L(H).
If G is a Lie quasi-subgroup of H, L(G) is also identified with a closed Lie
subgroup of L(H).
For <15 to be a submersion, it is necessary and suflicient that L(o) be sur-
jecfive and that its kernel admit a topological supplement. In that case, the
kernel N of cl) is a Lie subgroup of G and L(N) = Ker L(o). In particular,
if H is the quotient Lie group of G by a normal Lie subgroup P, L(P) is an
ideal of L(G) and, if d) is the canonical surjection of G onto H, L(G/P) is
identified with L(G)/L(P) by means of the morphism derived from L(¢)
when passing to the quotient.
Let I be a finite set, (0.)“ a family of Lie groups, G their product and p.
the canonical morphism of G onto G‘. Then (L(p,))m is a morphism of the
Lie algebra L(G) into the Lie algebra El L(G‘) and is an isomorphism of
normable spaces. L(G) is therefore identified with EilL(G.) by means of
(L(Nhsp
254
 FUNL'I‘ORIAL PROPERTIES OF THE LEE ALGEBRA § 3.8

PROPOSITION 28. Let G and H be Lie groups and <1; a morphism of G into H. Sup-
pose that K is qfeharaeteristie 0 and that H isfinite—dimensional.
(i) The kernel N oftb is a Lie subgroup ofG andL(N) = Ker L(zb).
. (ii) The morphism 4: of G/N into H derivedfrom ‘l’ when passing to the quotient
is an immersion.
(iii) If MG) is elosed in H and the topology of G has a countable base, MG) is
a Lie subgroup of H, vb is an isomorphism If the Lie group G/N onto the Lie group
<1>(G) andL(<l>(G)) = Im L(o).
Let G operate on H on the left by the mapping (g, h) I—> ¢(g)h. It suffices
to apply Proposition 14 of § 1, no. 7, to the orbit of e.
PROPOSITION 29. Let G and H be Lie groups and ti) a morphism (f G into H. Sup-
pose that K is of tharaeteristio 0 and that H itfinite—dirnemional. If H’ is a Lie sub-
group ofH, then G’ = q; "1(H') is a Lie subgroup q andL(G’) = L(o) '1(L(H')).
Let 1: be the canonical mapping ofH into the homogeneous space X = H(H’.
Let G operate on X on the left by the mapping (g, x) I» ¢(g)x. The stabilizer
oi'1:(e) is G’, which is therefore a Lie subgroup of G (§ 1, no. 7, Proposition
14). The orbital mapping of1r(e) is 7: o (I). By Proposition 14 of § 1, L(G’) is
the kernel of L(r: up) = T,(n) oL(<)>). The kernel of T,(n) is L(H') (§ 1,
not 6, Proposition 11 (i)) and hence Ker L(n o (b) = L(¢) “(L(H’)).
COROLLARY 1. Let G, H be Lie groups and 4); and (I), morphisms ofG into H.
Suppose that K is of characteristic 0 and that H is/inite-dimensional. The set ofg e G
such that ¢1(g) = ¢,(g) is a Lie subgroup G' ofG and L(G') is the set ofx e L(G)
such that L(¢,)x = L(¢,)x.
We write (Mg) = (¢,(g), ¢2(g)) for all geG, so that (I; is a morphism of
G into H x H. Let A be the diagonal subgroup ofH x H. Then G’ = ¢'1(A)
and L(¢)x = (L(¢1)x, L(¢2)x) for all x EL(G). It now suffices to apply
Proposition 29.
COROLLARY 2. Let G be a finite-dimensional Lie group and G, and G1 two Lie
subgroups efG. Suppose that K is qfoharaeteristio 0. Then G1 n G2 is a Lie subgroup
of G with Lie algebra L(Gl) n L(G,).
We apply Proposition 29 to the canonical injection of G1 into G and the
subgroup Ga.
COROLLARY 3. Let G, G’, H be Lie groups and (I): G—>H and o’: G’ ~>H Lie
group morphism. Suppose that K is Ifeharacteristio 0 and that H isfinite-dimensional.
Let F be the m qf(g,g') 50 x G’ such that (Hg) = (my). Then F is a Lie
subgroup of G x G' and L(F) is the set of (x,x') EL(G) X L(G') such that
L(¢)* = L(o')1'-
We apply Corollary 1 to the morphisms (g. e’) -> Mg) and (g, s’) H Ms’)
of G X G’ into H.
PROPOSITION 30. Let G be afinite—dirnem‘ionol Lie group with a countable base and

255
III LIB GROUPS

H and H' Li: subgroups hf G. Suppose that K it qfcllamcterim'c 0 and that HH' is
locally closed in G. I
(i) HH’ is a mbmmy'ald q/G andT,(HH') =. L(H) + L(H ). I I
(ii) Suppuxe that every element of H commutes with every element af H .I Then HH
1': a Lie subgroup 13/ G. Let 'i> he the mapping (h, h') r—>hh’ if H x H onto ‘HH .
The kernel of d; is the set if (m, m“) where m 6-H n H’ and the morphum af
(H X H’)/Ker 4) onto HH’ derivedfrom (I) by parxmg to the quotient u a Lie group
ismnorphirm.
Let H x H’ operate on G on the right by the mapping ( (h, h’), g) r—> hgh’ ’ 1.
The orbital mapping 9 of e is (h, h’) >—>hh"‘. By Proposition 14 (iii) of § 1,
no. 7, HH’ is a submanifold of G and T3(HH’) = Im T.(p). Now

Te(9)(L(H) X {0}) = L(H) and T.(9)({0} X L(H)) = L(H')

and hence T,(HH’) = L(H) + L(H'). Suppose that every element of H com-
mutes with every element of H'. Then HH' is a subgroup of G. By (i), it is
a. Lie subgroup of G. The rest. of the statement follows from Proposition 28.
PROPOSITION 31. Let G be afiniIe-dimemionat Lie group with countable bare, H a
normal Lie subgroup qf G and A a Lie subgroup af G. Suppose that K i: If charac-
teristic 0 and that AH is closed. Let 4) be the canonical morphism of G onto G/H.
Then the canonical mappings

A/(H n A) -> MA), AH/H -> #(A)

are Lie group isomorphimu.
By Proposition 30, AH is a Lie subgroup of G. By Corollary 2 to Proposi-
tion 29, H n A is a Lie subgroup of G. It is therefore meaningful to speak
of the groups AH/H and A/(H nA). On the other hand, MA), which is
the canonical image of AH in G/H, is closed and hence is a Lie subgroup of
G/H (Proposition 28 (iii)). Proposition 28, applied to the composite morphisms
A ——> G —> G/H and AH —> G —> G/H, proves that the canonical mappings
of the proposition are Lie group isomorphisms.
PROPOSITION 32. Let G and H be Lie groups, I: a nan-discrete closed .mbfield of K
and cl) a morphim of G into H fir the Lie group structure: over 1:. Suppose that K is
qf characteristic 0. If L(rb) i: K-linear, <1) i: a matphirm for the Lie group structure:
aver K.
For all g e G,
T,(¢) = T.(Y(<l>(g))) ° L(h) °T.(Y(g)")
and hence T,(¢) is K~linear. The proposition then follows from serentiahle
and Analytic Mani/77M; R, 5.l4.6.

256
 LIE ALGEBRA § 3.9

9. LE ALGEBRA 0F Tim GROUP OF INVERTIBLE ELEMENTS OF AN

ALGEBRA
Let A be a complete nonnable associative algebra with unit element e. Let
A" be the group of invertible elements of A. We have seen (§ 1, no. 1) that A“
is an open submanifold of A and is a Lie group. Let G be a Lie group andfa
morphism of the Lie group G into the Lie group A*. f can be considered
as an analytic mapping of G into the complete normable space A. Hence, if
is .97'(")(G), we can form <t,f>, which is an element ofA.
PROPOSITION 33. The mapping I» (t,f ) is a morphism of the algebra Y‘”’(G)
into the algebra A.
It suffices to verify that, if t and t’ are point distributions on G, then
(1* t':f> = <’:f><t':f>. But
(1* t'.f> = <1 69 t’, (5:!) Hf(££')>
= (l ‘8 t’, (g, g') Hf(g)f(£’)>
= <‘,f><",f>
(Difl: 6? Anal. Mara, R, 13.4.3).
The morphism of Proposition 33 is said to be associated withf.
Take G to be the group A“ imelf andf to be the identity mapping l. of A“.
We obtain a morphism, called canonical, of the algebra 3"”>(A*) into the
algebra A. The tangent space T,(A*) is canonimlly identified with A; and
if t e TG(A*), the definition of this identification is such that (1, I) = t. Then
Proposition 33 implies the following corollary:
COROLLARY. The canonical mapping 2; of L(A*) into A is an isomorphism 9f the
Lie algebra, L(A*) onto the Lie algebra A. In other worth,
{(Ea, 5]) = §(fl)C(”) - Wilma)
for all a, b in L(A*). IfK is q/characterixticp > 0, then C(a”) = l;(a)"for all
a e L(A*).
Henceforth L(A") and A are identified by means of the isomorphism t.
The canonical morphism of .7"(”’(A*) into A has been obtained as a special
case of the morphism of Proposition 33. But it is possible to argue in the
opposite direction:
PROPOSITION 34-. Let H be a Lie group, A a unital complete normable associative
algebra and 4): H —> A” a Lie group morphirm. The associated morphism (12’ of
57"”)(1-1) into A is obtained by composing (In. with the canonical morphism of.7(“WA“)
into A. In particular, (fix) = L(q>)(x) fir all at E L(H).
Let i be the identity mapping ofA’ into A. Then, for all [6 .7”(°“’(H),
W) = <!, <1>> = (Li 04>)
= (4.41),» (n es Anal. Mm, R, 13.2.3).
257
m LIE GROUPS

10. LIE ALGEBRAS OF CERTAIN LINEAR GROUPS

Let E be a complete normable space. Then 55’(E) is a unital completenorm-
able algebra and GL(E) is a Lie group. By the Corollary to Proposition 33,
no. 9, if T1(GL(E)) is canonically identified with f(E), the LR: algebra
structure on L(GL(E)) is given by the bracket (x,y) r—wty —'yx 01'" two ele—
ments of $(E). In particular, L(GL(n, K)) is canonically Identified wrth
gl(n, K) (Chapter I, § 1, no. 2).
PROPOSITION 35. Let E be afinits—dimensianal rector :paee. Le! <1; be the morphism
g» detg if the Lie group GL(E) into the Lie group K‘. The mapping L(b) of
$03) into K {r the mapping xs->Tr x. The kernel SL(E) of (I) it a Lie subgroup
qL(E) with Lie algebra el(E).
We choose a norm and a basis of E. The expansion of the determinant
proves that
det(l + u) 61 + Tr u + a(|[u||)
when u tends to 0 in Y(E). Hence, using Proposition 34, no. 9, for
x e 9(E) = L(GL(E)):
L(¢)(x) = (x, b) = TI” 3’
It follows that 42 is a submersion. Therefore, Ker <I> '= SL(E) is a Lie subgroup
of GL(E) whose Lie algebra is Ker L((b) = 51(E).
Let E,, . . . , E“ be complete normable spaces and E their direct sum. Every
1 e 3’(E) can be represented by a matrix “01“,“, where xi, E 3’03” E,).
PROPOSITION 36. Let I be a what qf{l, 2, . . ., n} and G the subgroup qL(E)
unwitting ‘lf’heg = (341)1‘5J‘n SCI-(E) 47”" ”W Eu = 0/0" <j audg“ = 1
for i e I. Then G i: a Lie subgroup of GL(E) and L(G) i: the set qf
J: = (Kuhghfinegm) such that x“ = Ofori <jamlxu = OfarieI.
Let S be the set of (am) e $(E) such that 2,, = 0 for i <j and x“ = 0 for
i e I. Then G is the intersection of GL(E) and the affine subspace l + S of
3(E) Hence G is a submanifold of 61.03) and the tangent space to G at l
is identified with S.
In particular, in GL(n, K), the total lower triangular subgroup and the
lower strict triangular subgroup, defined as in Integration, Chapter VII, §3,
no. 3, are Lie subgroups with Lie algebras ((11, K) and "(71, K) (Chapter I,
§ 1, no. 2).
PROPOSITION 37. Let A be a complete nannable unital associative algebra and x r—> x‘
a continuum linear mapping of A into A mh that (x‘)' = x and (xy)‘ = y‘a‘ for
all x, y in A. Suppose that K i: of churaetnirtie #2. Let G be the rubgmup ofA“
cann'stiug afthe xeA such that w = xx = 1. Then G ir a Lie subgroup ofA“
and L(G) is the u! n GA .mch thuty‘ = —y,

258
 LINEAR REPRESENTATIONS § 3.11

Let S (resp. S’) be the set ofy EA such thaty = y‘ (rap. y = —y‘). Then
S, S’ are closed vector subspaces of A. The formula

r=«1(r +y') +i(y-y‘)

proves that A is the topological direct sum of S and 8’. Letf be the mapping
of A into S defined by f (x) = xx‘. This mapping is analytic. For all y EA,
f(l +y) = 1 +1] +y‘ +yy‘; choose a norm on A compatible with its
algebra structure; then

f(1 +r) 51 +y +r‘ + 0(llyll) forytendingtoo.

Thus, T1(f )(y) = y + 31‘, so that f 15 a submelsion at 1. Therefore, there
exists an open neighbourt U of l m A such that U n G is a. submanifold
of U. Hence (§ 1, no. 3, Proposition 6) G 15 a Lie subgroup OfA*. Moreover,
L(G) = L(G) = Ker Tm.
COROLLARY 1. Suppose that K is ofrhoraeteristie #2. Let E be afinite—dimmsional
vector xpaee over K and (b a non-degenerate symmetric (reap. alternating) bilinear
form on E. For all u E 3’(E), let 11* be the adjoint ofu relative to 4). Let G be the
orthogonal (resp. :ymplettic) group of (b. Then G is a Lie subgroup of GL(E) and
L(G) i: the set of: e 3’(E) such that x“ = —x.
We apply Proposition 37 with A = .‘Z’(E) and x‘ = x“.
Remark. Let B be a basis of E and J the matrix of 1b with mpect to B. Then
L(G) is the set of elements of.‘?(E) whose matrix X with respect to B satisfies
the equation
‘X = —JXJ'1.
This follows fiom Algebra, Chapter IX, § 1, formulh (50).
COROLLARY 2. Let E be a complex (resp. real) Hilbert space and U the unitary group
of E. Then U i: a real subgroup ofGL(E) and L(U) is the set ofx e .‘2’(E) mph that
rt“ = ~x.
We apply Proposition 37 with A = A?(E) considered as an algebra over R
and x‘ = x*.
COROLLARY 3. Let E be afinfte—dimemional complex vector :pace, (I; o non-degenerate
Hermitiau :esquilineurfinn on E and U the unitary group oftb. Then U i: a real Lie
subgroup of GL(E) and L(U) i: the set qfx e .701) such that ix is Hermitian.
When E ¢ {0), U is not a Lie subgroup of the complex Lie group GL(E),
for L(U) is not a complex vector subspace of 3’(E).

11. LINEAR REPRESENTATIONS

Let G be a Lie group, E a complete normable space and 1: an analytic linear
representation of G on E (§ 1, no. 2). The associated morphism I» (t, 1:) of

259
III LIE GROUPS

33) and its

Y‘”)(G) into 2(E) is an algebra morphism (no. 9, Proposition
on E
restriction to L(G) is L(rr). Hence L(rr) is a representation of L(G)
(Chapter I, § 3, Definition 1).
mapping
Pnoposmou 38. Consider G as operating on E or: the left by the
(g, x) >—> 1r(g)x. Let b e E and p(b) be itr orbital mapping. Let T,,(E) be tammeally
identfied with E. For all t e L(G),
(Leonor) = <t. 901)) = pow = m».
In particular, the vectorfield defined by t on E is thefield b I» (L(re)t) (b).
L(rr)t = (t, 7:) (no. 9, Proposition 34-). As the mapping A HAb of 3’05)
into E is continuous and linear, it follows that
(L(n)t)(b) = <1, g H 1r(g)b>
= (t, Id; ., 9(b)>
= (900.1, Idg> (0:27. am. Mm, R, 13.2.3)
= 9(5):!-
Finally, 9(1)) 4 = u a. (no. 3, Proposition 14 (ii)).
PROPOSITION 39. Suppose that K i: qfaharaeterim'e 0. Let G be a Lie group, E a
finite-dimensional vector spate and 1t an analytic linear‘repmentation of G on E.
Let E1, E, be ueetor subxpaees afE such that E, C E1. The set C; ofg e G such that
r:(g)x E x (mod. E2)for all x5 E1 is a Lie subgroup ofG and L(G,) i: the set of
a e L(G) such that L(rr)a map: E, into E,.
This follows from Propositions 29 (no. 8) and 36 (no. 10).
COROLLARY I. In the notation of Proposition 39, the set of g e G such that
rc(g)(E1) C E1 it a Lie subgroup ofG and it: Lie algebra is the set ofa e L(G) meh
that L(1r)a map: E, into E,.
We apply Proposition 39 with E1 = E2.
COROLLARY 2. Let G, E, 1: be a: in Proposition 39. Let F be a .mbset of E. The :et
ofgeGsueh that1r(g)x = xforallxeFisaLiembgrouqGanditsLioalgebra
is the m (y‘ a e L(G) such that (L(rr)a)(:t) = Ojbr all x E F.
We apply Proposition 39 with E, = {0} and El the vector subspace of E
generated by F.
Let 1:1, 1:2, . . ., 7:, be analytic linear representations of G. Clearly the direct
sum 7: of the 1:. (Algebra, Chapter VIII, § 13, no. 1) is an analytic linear repre-
sentation of G and L(rr) is the direct sum ofL(1r1), L(Ka), . . ., L(rr”) (Chapter
I, §3, no. 1).
PROPOSITION 40. Let G be a Lie group, E a complete nor-enable space, 1: an analytic
linear reprexentation g” G on E and F a closed vector subspace if E xtable under 7c(G).
Suppose that K i: qfeharaeteriitit 0, or that F iv a direttfaetor of E.

260
 LINEAR REPRESENTATIONS § 3.11

(i) The mbrzprenntatiun 1:1 and th: quotient iqzrexznlation 1r, qf 1: defined by F are
analytic repm'mtatim.
(ii) F 1.: xtabl: under L(1r) (L(G)).
(iii) Let 91 and Pg b: the mbnpresentation and quotient representation qf L(1:)
dzfined by F. Than L(-n:,) = p], L(1:3)=
Let A be the set ofu e $(E) such that:2u(F) C F. Then A Is a closed vector
subspace of Y (E) and 1: take: us valua m A. By virtue of the hypotheses on K
and F, the mapping 3’: G —> A with the same graph as 1: is analytic (Dijfiren-
tiable and Analytin Manifolds, R, 5.8.5). The canonical mappings 91: A —> $(F)
and 62: A —> $(E/F) are continuous and linar and hence analytic. This
proves (i). The mappings T,(1r) and T,(1I:’) have the same graph and hence
L(1r)(L(G)) C A, which proves (ii). We have
TAM) = T49; W') = 91 °Ta("') = 91
Tim) = me. «- n') = 92 . .(n') = a.
PROPOSITION 41. Let G be a Lie group and 1:1, 1:2, . . ., 1:,” 7: analytic linear repre-
xmtatian cf G on complete mmnable spate: E1, E2, . . ., E,” E. Let
(x,, x,, . . ., x") HxIxz. . .x,
In a continuous mallilimar mapping @e1 x E, x ' - - x E" into E. Suppose that
"(5)0‘1’72- - '5‘») = ("1(3)”1)(1'2(g)”2) - - - ("n(£)‘n)
fatallgeG,x1eE,,...,xneE. Then

<L<r=>a) (w. x.) = Z x.-.<(L(m)a)x.)~..1

firallaeL(G),x,eE,,.. ,Jc,,eE,I '
As an example we perform the calculation for n = 2.
(L(1v)a(IX-Wk) = <0: 3 '—> fi(£)(*:*a)> (PIOPOSition 33)
= <11, (g H ":(g)~x)(g -—> "2(g)*2)>
= (a,g»n1(g)x1>.xa + #1-<a,g'—>7=z(g)x2> (Difl’. 5‘? Anal-
Man, R, 5.5.6)
= ((L(n1)a)x,).x, + x..((L(n,)a)x,) (Proposition 33).
COROLLARY 1. Let G be a Lie gruup, E1, . . ., E”, complete rwnnablc spam and
7:1, . . ., 7r...” analytic linear representatiunr ufG on E1, . . ., En”. Let
E = $(E1: - - u E»; En+l)
the sample" mnnable xpacz of continuous multilineai mapping: of E1 x - ~ - x En into
En+1 (Gama! Topalagy, Chapter X, §3, no. 2). For all g 6G, 1:! 1t(g) b: thz
automorphinn of E defined by
(”(3)")0‘1: - - -, I.) = "n+1(g)(u(1r1(g)“xn - - -, 1r»(g)"xn))-
261
Ill LE GROUPS

Then 7: is an analytic linear representation qf G an E and

((L(«)a)u>(x1,....xn) = — 2 not, . . ., n-1, (Loom, arm, . . ., x")

[=1
+ (L("n+1)“)(“("u - . u 56.))
forallaEL(G), ueE, xl 6E1, . . .,x,,eE,..
Every element (A1,. . .,A,+l) of $(E1) x ~ - - x 2’03”“) defines a con-
tinuous endomorphism 0(A1, . . ., AMI) of E by the fin'mula
(0041, - - '2 An+1)“)(‘¢n - ~ u 3n) = An+1(“(A1*n . - ') 24..q-
The mapping 6 of.‘Z’(E,) x i - . x 3(E“ 1) into 3’(E) is continuous and multi-
linear. Then, for all g e G,
7‘(g) = 9(fi1(g"),~ - -: W43"): "n+l(g))
and hence 1: is analytic. We apply Proposition 41 w the mapping
(:1, . . ., x,” u) v—>u(x,, . . ,, xn)
of'El ><~~>< E,l x EinwEHrThen
"n+1(g)("(*1, - ~ -, *0) = (W(£)")(K1(g)xn - - -, n..(g)*»)
and hence n ’

(L(nm>a)(u(x1, . . ., 2"» = Z um, . . .. (Loom . . ., x")

+ ((L(7=)a)")(xh , ~ ., :9.)-
When the E. are finite-dimensional, the representation L(n) of L(G) is
derived from the representations L(nl), . . ., L(nnu) under the procedure
of Chapter I, § 3, Proposition 3.

COROLLARY 2. Let G be a Lie group and 1: an analytic linear representation of G on

a complete normable spate E. Then g ~—> ‘1:(g) ‘1 is an analytic linear representation 9
qf G on the complete nmnabl: spat: $(E, K)T and L(p)a = —‘(L(1:)a) for all
a e L(G).
This is a special case of Corollary 1.
p is called the contragrzdient representation of 1:.

When E is finite-dimensional, L(p) is the dual representation of L(n)

in the sense of Chapter I, § 3, no. 3.

COROLLARY 3. Le! G be a Lie group and 1:1, . . ., 1r,I analytie linear representatians

T A: when K = R or C, the transpose ‘1:(g) considered here is the restriction to

.2’(E, K) of the transpose of 1r(g) in the purely algebraic sense.

262
 LINEAR upmm‘emous § 3.11

of G on finite—dimmrional vector space: E1, . . i , E... Then the reprexentation

n1 8- - -® 7:" if G (Appendix) it analytic aru1,L(1r1 ®~ - -® 1:“) is the tenor
product ofL(1r1), . . . , L(rr“).
The mapping (A1, . A ., A") HA, ®- - -® An ofSflEl) x - - - x $(En) into
.‘2’(E1 8‘ - - ® E“) is multilinear, whence the fact that 1: is analytic. Consider
the mapping (x1,....,x,,) v—ne, ®---®::fl ofE1 ><~c~ X E, into

E1 ® ‘ ‘ ‘ ® En-
By Proposition 41, we see that

(Lama. comm) = ,2,x. ®~~® (L(m)a)x« ®---®x.

for all a e L(G), x. e E. for 1 S i < n. Hence L(rr) is the tensor product of the
L(m).
COROLLARY 4. Let G be a Lie group and 1: an analytic linear representation of G on a
finite-dimensional vector space E. Then the representation T"(7c), S"(1t) and A"(rr) of
G (Appendue) are analytic and

L(T"(W)) = “(14(10): L(5"(1=)) = 5"(L(7r)), L(A"(7=)) = /\"(L(7=))-

This follows from Corollary 3 and Proposition 40.
COROLLARY 5. Let A be afinite—dimenfional algebra. Suppose that K it qfchamcterix-
tic 0. The automorphism group Aut(A) ofA i: a Lie subgroup ofGL(A) and L(Aut(A))
i: the Lie algebra of derivation ofA.
This follows from Corollary 1 (applied to E = 3’(A, A; A)) and Corollary
2 of Proposition 39 (applied to the subset of E consisting only of multiplication
on A).
Remark. We apply Corollary 1 with G = GL(F) (F a complete normable
space), 7:1 = n, = 1d,, and 1:: the trivial representation of G on K. We obtain
an analytic representation r: of GL(F) on 3’(F, F; K). We assume that F is
finite-dimensional and that K is of characteristic 0. Applying Corollary 2 to
Proposition 39 to 7:, we recover part of Corollary 1 to Proposition 37.
PROPOSITION 42. Let G be a Lie group, X an analytic manifold, (g, x) »—> gx (resp.
xg) a law ofanalytic lefl (resp. right) operation of G on X and so a point llf X which
it invariant under G. For all g e G, let -r(g) be the autornorphirrn x» gx (resp. xg)
of X And let n( g) be the automorphism oj"T,,n(X) tangent at no to r(g).
(i) n i: an analytic representation of G (resp. G") on T,0(X).
(ii) For all a E L(G) and all 50 e T,D(X), L(n)a. £0 can be calculated mfiJllowr:
let D“ be the vectorfield defined by a on X and E a uectorfield q" clam 01 in an open
neighbourhood lfxn ruch Ihat E(xo) = 50; then
L(n)o.E., = —[Da, Elfin)-

263
Ill LIE GROUPS

fig!) = T(3)1(g’) (rap- T(.e.")-r(g)) and hence "(3.6) = ”(g)1r(g') (rap.

7c(g’)1r(g)). On the other hand, since TX is a vector G~bundle of class 0"
(§ 1, no. 8, Proposition 16), 1c is analytic, whence (i).
To prove (ii), suppose that G operates on the left. There exists an open
neighbourhood I of 0 in K and an analytic mapping 7 of I into G such that
7(0) = 2, Ta (7)1 = n. Then D, is the vector field on X defined by the mapping
(1;: (Ax) e> y().)x of I x X into X (§ 2, no. 2). If 4). denotes the bijection
x >—> 70):: ofX into X, then

[D.,51(xo)= (4% (T.....(¢;‘)z(¢.(xo>)>)h o (Dar. Want. Mm, R.

‘ 8.4.5)
= (f; <T..(¢:’)z.)) A.»
d
= (gowmrlaa) 1-0
Asthemappings A n—> -y().)‘1 and A »y(—Z) are tangent atO, this is also equal
to

—(d—"i<w(m»§o)) AI!)

= —(d%(n°¥)(l)) x-o a.
= —L(1r)a.Eo.

12. ADJOINT REPRESENTATION

Let G be a Lie group. Consider the law of analytic left operation
(3, 3’) 933’s“ = (mg)!
of G into G. This law of opemtion defines, by no. 3, a bilinear mapping of
9"“)(G) x .9"°‘”(G) into .7"”(G), which we shall denote by T in this no.
By Proposition 13 of no. 3,
(22) (n: t’) T I" = t T (1’ T t”)
for all I, t’, t” in .7"”(G). By Proposition 14 (i) of no. 3,
(23) s, T I = (Intg)*t
for all g e G and t E 9"”)(G). In particular, the mapping 1‘ >—> 6,, T Io‘”’(G)
into .9""”(G) is an automorphism of the bigebra 7““(G). Its restrictions to
U(G), Ua(G): L(G) are denoted by Adu(G)(g): Adu,(o)(g)a Aduu)(£)~ We often
write Ad(g) instead of Adm},(g) when no confusion is possible. By (23),
Ad(g) i: the langmt mapping at e la Int(g). It is an automorphism of the normable

2G4
 ADJOINT REPRESENTATION § 3.12

Lie algebra L(G). When K 15 of characteristic 0, Adam“) is the unique auto-

morphism of U(G) which extends Ad(g)
If h'IS a morphism of the Lie group G into a Lie gmup H, then
(24) 1M! T V) = lid!) T 440)
l for all t, t' in 3"“)(G); this follows from Proposition 15 of no. 3.

Pnoposmou 43. Let r, u be in y<°°)(G). Let 1-2) I‘ Q) t,’ be the image 0/ t under the
mpmduot. Then
n
tTu=i-21I,*utt(v.
By definition, t T u is the image of! 8) u under the mapping (g, g') .—> gg’g-i
of G x G into G. Now this mapping is obtained by composing the following
mappings:
a: (in!) » (g,g,g') ofG x (:1a x G x G
B: (g,g',g')->(g,g"1,g’) ofG x G x GintoG x G x G
Y: (“3g”) »—>gg’g’ ofG x G x G into G.
On the other hand:

2 (tl®‘a) ®u= Zh®ti®;

“tl’ ® 1‘): I-l

l5#(‘2:1h®ti®u)=‘;h®tiv

Ye(‘:1’n®ti ®“)=‘:1h*“*tiv-u
COROLLARY]. Let u EL(G) andu’ ei‘"(G). Then a T u’ = u e u’ —- u’ n: u.
The image of 14 under the coproduet is u ® a, + 2, ® 14, whence
uTu' =Il*u'*€, +e,*u’*u" =usu’ — u’tu.
COROLLARY 2. Let tef‘mw) and geG. Then a, T I = ‘a*”%"‘ If
teL(G), then s, T! = glg'1 (when the latter product is evaluated in the group
T(G)).
The image of a, under the coproduct is 2, ® 5,.
COROLLARY 3. Le! a e L(G). The vector field defined by a and the lefl operation
gHIntgq onGisthefieldR, — L“.
The value of this field at g is
a Ta; = «*z, — even (Corollary 1)
: (Rn); — (L1,)? (Definition 5).

265
Ill LIE GROUPS

For allgeG andall IEL(G),

(25) (Adg)(t) = z, T t = s, *t*a,-i = gig—1.
Since Ad g = T,(Int g), Proposition 42 of no. l1 prova that Ad is an analytic
linear representation of G on the normable space L(G).
DEFINITION 8. 7713 reprumtah'm Ad qfG an L(G) is called [he adjoint reflexentalinn
qf G.
Pnorosmou 4-4. For all a e L(G),
(L(Ad))(“) = admin“-
Let b e L(G). By Proposition 4-2 (ii) of no. 11 and Corollary 8 to Proposition
43,
(L(Ad))(a)-b = _[Ra - L." 141(9)-
Now Ra o L, = L, a R“ (no. 6, Proposition 23 (ii)), whence [Rm L,] = 0;
then, using Proposition 23 (ii),

(L(Ad))(fl)-b = [Lax Lake) = Lm.ol(‘) = [0, b] = (“luau”-

PROPOSITION 45. Suppose that G irfinite-dimm-ional and that K i: afchamderixii: 0.
Let s be an integer 20. Then the ”naming 1:: gHAdWMg) i: an analytic linear
upmmmtion of G on U,(G) and L(1r)a = min“, afor all a e L(G).
s
The linear representation 1: is a quotient ofQ, T’(Ad) and is hence analytic.
For a e L(G) and x1, x” . . ., x, in L(G),

(L(n)a)(x1x,. . .3.) = ‘21 x1. . . (L(Ad)a.x‘) . . .x, (Proposition 41)

= 2 x1. . .([a, 1.1) . . .x,

II l
(Proposition 44)
= (AdU.(G) a)("1"2~ - 41:)-

PROPOSITION 46. Let hsG, xsTn(G) and aeL(G). Let d) b: the mapping
(g,g’) Hgg’g‘l afG x G into C. The imagey qf(a, x) eT.(G) x TAG) under
T(-.n)(¢) uy = x + h((Adh“)a — a).
We have
y “(c.h)¢)(a®3n + 9c®")
=aTsn+eeTx
=ateh—znta+x
= h((Ad h“)a) — ha + x.
266
 ADJOINT REPRESENTATION § 3.12

PROPOSITION 47. Let G be a Lie group, H and E Lie mbgt’aupx g"G and mppoie that
hEh‘1 = Efor all h EH. Then 3"”(H) T 3"”’(E) C 3"”)(E). In partieular,
Ad(H>(L(E)) c L(E) andIL<H),L(E)1 c L(E).
Iite 9"”)(H) and t' 6.7“”)(E), then t 87 t' 69"“‘(H x E) and the image
of H x E under the mapping (g, g’) l—> gg’g“ is contained in E.

PROPOSITION 48. Let G be a Lie group and H and E Lie .mbgroup: ifG Suppose that
G 1.3a: a Lie group, the term-direct produet ofH by E Let p be the linear representation
g r—> (Ad g) | L(E) of the Lu group G on L(E) (of Proposition 47) and let a be the
rextrietion of p to H. Then:
(i) L(G) i: the topological direct mm ofL(H) and L(E);
(ii) L(H) it a :ubalgebra ofL(G) and L(E) it an ideal of L(G) ;
(iii) L(o) it a linear reprexentation of L(H) on the Lie algebra of derivation: of

(iv; 13G) is the iemi-direet produet ofL(H) by L(E) Mined by L(o') (Chapter I,
§ 1, no.
(i) is obvious and (ii) follows li‘om Proposition 4-7. L(c): L(p) |L(H).
Now by Propositions 40 (no. 11) and 44 (no. 12), L(p) (t)1s, for all t E L(G),
the restriction ofadua, t to L(E). This proves (iii) Using (i) and (ii), this also
proves (iv).

COROLLARY. Let G be a Lie group. Let T,(G) be given it: unique commutative Lie
algebra structure. Let 1' be an adjoint representation of L(G). Then the Lie algebra of
T(G) is the .remi-direot product of L(G) by T.(G) defined by r. In other word:, for
x, x’ in L(G) andy,y’ in TAG),

[(Jw), (x’,y')] = ([3, 1’], [my’] + [113])

(where the bracket on the left is evaluated in L(T(G)) and the bracket: on the right in
L(G)).
This follows from Proposition 48 and Proposition 6 of § 2, no. 2.

PROPOSITION 49. Let A be a oomplete nonnable unital associative algebra. We

identify A with L(A“). Then, 1fg e A‘ andy GA, (Ad g)y = gyg".
Recall that Ad g = T1(Int g). Let u, be the mapping 1: 1—» gag" ofA into A.
The identity chart of A' into A transforms Int g into u, | A“. The tangent
mapping at each point of A‘ to this mapping is equal to u,, whence the prop-
osition.

COROLLARY. For all g e A‘, (at i(g) be the autarnarphiirn y 1—» gyg‘1 efA, to that
i is an analytic linear representation q’ an A. For all z e L(A’) = A, L(i)z i: the
inner derivation y ~—> zy — yz ofA.
This follows from Propositions 49 and 44.

267
m LIE GROUPS

13. TENSORS AND INVARIANT FORMS

Let G be a Lie group. We consider G as operating on itself by left (resp. right)
translation. Let A be a vector functor of class C") for isomorphisms. Then
MTG) is an analytic left (resp. right) vector G-bundle (§_ 1, no. 8, Corollary
to Proposition 16). The mapping (g, u) »—> ga (resp. 11g) at G x ML(G)) onto
MTG) is an isomorphism ti) (resp. \ll) ofvector G-bundles (§ 7, no. 8-, Corollary 2
to Proposition 17). Every G-invariant section of MTG) is analytic and deter-
mined by its value at 2 (§ 1, no. 8, Corollary 1 to Proposition 17). Such a.
section is called left (resp. right) invariant. Let a be a left invariant section of
MTG); the transform (1’ of a under a right translation 8(g) is delined by
u'(8(g)h) = MT,,(8(g)))a(h) for all hEG; it is also left invariant; it is also
derived from o by 7(g) o 8(g) = Int(g) and hence
(26) o’(e) = MAdg) .o'(z).
Similarly, let 1- be a right invariant section of MTG); the transform 1’ of 1
under a left translation 7(g) is also right variant and
(27) ‘r'(a) = MAdg) .1(e).
We now consider G x G as operating on G on the left by
((g, g’): a”) ”gig”-
Then G is a left Lie homogeneous space of G x G (§ 1, no. 6, Example). Hence
MTG) is an analytic left vector (G x G)-bundle. A section of MTG) is called
biinwzriant ifit is invariant under the action ofG x G on MTG), in other words
if it is invariant under left and right translations. Let ML(G))D be the set of
elements of ML(G)) invariant under MAd(G)). For all u E ML(G))0, let on be
the mapping of G into MTG) defined by 0-,,(g) = gu = ug. Then a r—> a", is a
bijection of ML(G))0 onto the set of biinvariant sections of MTG) (§ 1, no. 8,
Corollary 1 to Proposition 17).
PROPOSITION 50. Let G be a Lie Group (assumed ta be finite-dimenrional if K i: of
eharacterirtie > 0). Let E be the ueclur space tftnnlinuou: alternating multilinearfizrm:
qfdegree I: an Te (G). For all u e E, let a)" be the dyfirenlialfarm ofdegree k an G such
that (0)"), i: the multilinearfbrm an TAG) derivedfrom u by the translation 11 >—> gh
(resp. h r—> hg). Then a)" is analytic and left (resp. right) irwan'ant an G. The mapping
u »—> to" is an immarphism qf E onto the vector space If left (resp. right) invariant
diflerentialfarmx of degree I: an G.
This is a special case of what we have said above.
Let F be a complete normable space. Proposition 50 remains true ifdifl‘eren-
tial forms on G with values in K are replaced by differential fon'ns on G with
values in F. For every continuous linear mapping :4 of TAG) into F, there
exists a differential form a)" of degree 1 on G, with values in F, such that
(13“)! = u o T,(T(g)“). In particular, take F = T.(G) and a = Idrmm. We

268
 MAURER—CARTAN FORMULAE § 3.14-

then obtain the differential form won G such that m, = T, (y(g‘ 1)) ; this difieren-
tial form is left invariant and analytic; it is called the lefl canonical dijfirmtial
form ofG. m,(t) = g"! for all teT,(G).
If F is again an arbitrary complete normable space and u e Y(T,(G), F),
then a)“ = u a m. In particular (taking F = K), the mapping v>—> u a m is a
linear bijection of the dual ofT,(G) onto the vector space of differential forms
of degree 1 with values in K which are left invariant under G.
Similarly, the differential form (0' on G such that a); = T,(B(g)) is called the
right canonical diflrmtialform of G. There are analogous properties to those ofco,
which we leave to the reader to state. The mapping g >—> g" 1 of G onto G trans-
forms on into m’.

14. MAURER—CARTAN FORMULAE

Let X be a manifold of class C", of finite dimension if K is of characteristic
> 0, and let L be a complete normable Lie algebra. Let u be adifiermfial form
of degree 1 on X with valum in L of class C"1. Let x EX. The mapping
(“1: "2) H [110(1): “x(“2)]
of T,‘(X) x TAX) into L is a continuous alternating bilinear form on TAX)
with values in L. We shall denote it by [a]; so that [an]2 is a differential form of
degree 2 on X with values in L. Identifying an open neighbourhwd ofx in X
with an open subset of a Banach space, we see immediately that [an]2 is of class
C" 1. If X’ is a manifold of class C' andf: X’ —> X is a morphism, then

(23) [f"(w)]z =f([a]’)-

Let a, [i be two differential forms of degree 1 on X with values in L of class
C" 1. The exterior product a: A B of at and B (Diflnmiable and Analytic Mam'fildr,
R, 7.8.2) is a differential form of degree 2 on X with values in L of class C"1;
we have
(29) (a A BLAH» "2) = [¢x("1)xlax("2)] - [14149, 9:05)]
for 141, u, in T,,(X). It is immediate that

(30) [‘1 +15]2 = [all2 + [5]2 + a A 9

(31) at A at = 2[az]2.
PROPOSITION 51. Let G be a Li: group, tffiuil: dimmion if K i: of:haramrim‘c > 0,
and [5t 41, . . ., a, b: element: q(G), F a complete non/table spec: and a a dfinmial
form afdegm p — l on G with value: in F. If a ir lefl invariant, then

(4a),(a1,...,a,> = ,2, (-1)'+’«.([at, «,1, a1, . . a am, . . .,a,_1,

an,“ . . ., 0,).

269
III LIE GROUPS

If a is right invariant, then

(4004111, - . -) 11,)
= —'<§; (—1)'+’m.([au 11,], a1, - - was—1: ”(+1; - - .,a,_1,a,+1, - - n a,).

Suppose that a is left invariant. By termtiable and Analytic Manifaltir, R,

8.5.7, then
(dam... - - ., L...) = Z (—Ir-‘L... 4L... - - ., L.-. L... . - -. L...)
+ ;(—1)”’1([Lan L1,]! Lap - . ., Lm-z- L‘H-l’ ‘ ' -’ Lav-v Lam, - - H La.)-
But the functions 04L“, . . ., Laid, LN”, . . ., L“) on G are left invariant and
therefore constant. Hence
Lmu(Lax, . . ., Laid, Lam) . . ., Lay) = 0.
Moreover, [Lon L,,] = LEM” (Proposition 23), whence the first formula of
Proposition 5!. The second can be established analogously, this time using the
relation [Rap Ry] = ~11,“ “,1.
COROLLARY 1. Let G be a Lie group, offim’h dimmn'on if K i: qfcharacmislic > 0,
and u) and m’ the lgft and right canonical dzflmntialflm: q. Then
do) + [0.12 = 0, dm’ — [d 2 = o.
By Proposition 51,
(400.01» a2) = “"3([ala 02]) = -[h: '12] = -[m.(a1), 6.012)]
= —[‘1’]5(“n 02)
whence the first formula. The second can be established analogously.
COROLLARY 2. Suppose that G i: finite-dimensianal. Let (:1, . . ., 9,.) be a ban': af
L(G), (2f, . . ., 4:) th: dual basis, (54”.) the mutant: ofstructure ofL(G) rzlative tn the
hair (:1, . . ., an) and a). (rcsp. «32) the left (resp. right) invariant diflermh'alfizrm an
G with values in K mph that (on), = of (resp. (0);), = :2"). Then

dmk+z
4<1"”‘°"A to, = 0 (I: = l,2,...,n)

an; — gawk“; A m; = o (k =1,2,...,n).

Ifr<:,
(46k).(¢n¢s) = -(0x).([¢n (.1)
= —Zc...<w.).(e.)
= —-c,,,,

= figmflnu A my)¢(‘v‘r)-
The argument is similar for the w,’,.

270
 mmonnue GROUP §3,16
15. CONSTRUCTION OF INVARIANT DIFFERENTIAL FORMS
Lemma 2. Let G be a Lie group, U a Jyrnmetric open neighbourhood of e in G, E a
complete normable space and c): U2 —> E an analytic mapping. For all g e U, let u), be
the diferential at the point g of the mapping h v—> ¢(g“h). Then (a iv the restriction to
U of the left invariant diferentialform on G whose value at e i: 434%
Clearly m, = 11,21). For allgeU and all tET,(G),

(up T.('{(g))'> = (Mb ° Y(g)“), Te(Y(g))l>

= (44> ° a(Y(g)"):T-(Y(g))l) = (d. ,1)
and hence a), is derived fiom 4% by T,(y(g)).
PROPOSITION 52. Let n be an integer >0, G an n-dimemional Lie group, U a sym-
metric open neighbourhood ofe in G and 4;: U’ —> K” a chart ofG such that Me) = 0.
If (zl, . . ., x") are the coordinate: qe MU) and (y1, . . .,y,,) the coordinates of
y e MU), we denote by

”Men, - - ., "mllb ~ - ~r.'/n)r - . '5 "50% - - ~: any» ~ - ’klyn)

the coordinate: of ¢(¢'1(z)‘1¢1(y)). Then, if we wrzte,for l< <11,

(32) more - - '9 3n) = Dnélk‘h - - ~: "m :1, - - u Nu) 4": + - - -

+ Dmmk(xl, . . ., x,” x], . . ., x”) d1",
the dfirential form an, on ¢(U) are derived through kl! from kfl invariant afierential
firm: on G and are such that mk(0,. , 0) = dz,,.
We apply Lemma 2 with E= K, taking (Mg) to be the coordinate of Mg) of
index k. We obtain a differential form a)»; let on, be we transform under k]; The
value of 51,, at (:9, . . ., x") is the differential at (x,, . . ., 1..) of the function
y »—> mkorl, . . ., guy], . . .,y,,); this value is therefore given by formula (32).
It then suffices to use the conclusion to Lemma 2.
PROPOSITION 53. Let G be a Lie group, A a complete norrnable algebra and 2b a Lie
group morphism afG into A‘. For all g EG, let to, = ¢(g)'1.d,¢. Then a) i: the
left invariant dgfl'erentialfarm on G whose value at e is d,4>.
We apply Lemma 2 with E = A and U = G. The difl'erential at g of the
mapping ’1 *-> ¢(g"h) = <l>(g)'1<l>(") is Ms) ‘1'd9‘l’~
16. Km MEASURE ON A LIE GROUP
Let G be a Lie group of finite dimension n. Then A"(T,(G)) is ofdimension 1.
Hence (no. 13) the vector space S of left invariant difl‘erential forms of degree n
on G is of dimension 1. Let ((1),, . . ., m") be a basis of the space of left invariant
differential forms of degree 1 on G; then «)1 A a), A ‘ - - A 0),. is a basis of S.

271
Ill LIE GROUPS

PROPOSITION 54. Let G be a Lie group «y’finile dimension n, m a left Marion! dg‘fl'eren-
tiolfonn qfdegree n on G and 11> on endomorphmn ofG. Then
<l>*(0) = (‘1‘:t L0H)!”-
We write L(¢) = u, a). =fand ¢‘(m), = g. For all x” . . ., x, in L(G),
g(x1, . . ., x“) =f(ux,, . . ., we”) = (detu)f(x1, . . ., x“)
and hence ¢‘(o>). = det L(<[>) .0)” On the other hand, ifg e G,
11> ° Y(:) = Y(<l?(g)) ° (l
and hence 7(g)*<]>*(m) = 41*(0). Thus (19(9)) is left invariant, whence the
proposition.
COROLLARY. For all g e G,
8(g)*m = (detAdg)m.
S(g)‘u = 3(g)‘y(g)*w = (Int g)*m and L(Int g) = Ad g.
Let G be a locally compact group and 4) an endomorphism of G. Suppose
that there exist open neighbourhoods V, V' of e such that ¢(V) = V' and
(I) | V is a local isomorphism of G into G. Let y. be a left Haar measure on G.
By Integration, Chapter VII, §l, Corollary to Proposition 9, there exists a
unique number a > 0 such that (My. | V) = fly. I V’. Clearly a is indepen-
dent of the choice of V, V’ and x-L. It is called the modulus ofcl> and is denoted
by moda <1) or simply mod (I). When <1) is an automorphism of G, we recover
Definition 4 of Integration, Chapter VII, § 1.
Pkoposmou 55. Suppose that K is locally tompad. Let y. be a Haor measure on the
additive group of K. Let G be a Lie group offinite dimension n.
(i) Let a) be n non-zero le/l invariant dferenlial form of degree n on G. Then the
meme: mod(w)u (Dyfmntiable and Analytic Manifoldx, R, 10.1.6) i: a lefl Haar
measure on G. If K = R and G ho: the orientation defined by m, the measure defined by
a: (Difirentioble and Analytic Manifoldr, R, 10.4.3) is a left Haar memre on G.
(ii) Let (I: be an elole endomorphim of G. Then mod <1) = mod det L(¢).
(i) is obvious. Let V, V’ be open neighbourhoods of e such that (NV) = V'
and (b ] V is a local isomorphism of G into G. Then

o‘1(mod(¢o), l V') = mod(<[>*(m))u I V) by transport of structure

= mod(det L(¢)m I V)“ (Proposition 54)
= mod det L(<]>)(mod(m)n | V)
whence mod (I) = mod det L(¢) by definition of mod (I).

272
 HMR MEASURE ON A LEE onour § 3.16

COROLLARY. For all g E G, AG(g) = (mod det Ad g) ‘1. In partibular,fir G to be

unimmialar, it is necessary and ngflin'mt that mod dct Ad g = l for all g e G.
AG(g) = (mod Int g) ‘1 (Integration, Chapter VII, § 1, formula (33))
= (mod det L(Int g» ‘1 (Proposition 55)
= (mod det Ad g) '1.

Remark. Preserving the hypotheses and notation of Proposition 52, suppose

that K is locally compact. Let p. be the measure

mod det(DnHmu(’fn - ~ -5 “m *1: - - u *n))l<l.k(n dti - - . dxu

on MU). Then V101) is the restriction to U of a Haar measure on G.

PROPOSITION 56. Let G be a Lie group offinite dimension n, H a p—dimemiunal Lie

subgroup and X the Lie homogeneous spate G/H. Suppose that
det Adm» h = det Adm h
fir all h EH. Then:
(i) The difi’rentialform: of degree n — p on X which an invariant under G are
analytic.
(ii) The vector space 4Jthesefarm: is of dimension 1.
(iii) If m i: .ruch a non-zeroform and K it locally compact, mod(m)u i: a non-zero
measure on X which is invariant under G.
By § ], no 8, Examples, Alt“"(TX, K) is an analytic vector G-bundle.
Let at” be the canonical image of e in X; its stabilizer is H. The fibre of
Alt“"’(TX, K) at so is /\"_'T,¢u(X)‘ and Txu(X) is canonically identified
with L(G)/L(H). If h e H, the automorphism 1,, of X defined by h is derived
when passing to the quotient from the automorphism g »—> hgh'1 of G. Hence
the automorphism T,0(‘t,‘) is derived when passing to the quotient from
Ammo)- As det Adm” h = (det Adm h) . (det T,o(r,,)),

the hypothesis implies fliatdeto(h) = 1. Thus, everyelementofAn_pTxo(X)*

is invariant under H. Then (i) and (ii) follow from § 1, no. 8, Corollary 1 to
Proposition 17 and (iii) is obvious.

The existence of a non-zero positive measure on X invariant under G

follows from Integration, Chapter VII, §2, Corollary 2 to Theorem 3, for
the hypothesis of Proposition 56 implies AGIH = An (Corollary to Proposi-
tion 55).

PROPOSITION 57. Let G be a Lie group of finite dimemian 71. Choose a basis for
/\"T,(G)*; by mean: nfthe right (resp. left) trivializatian Qf /\"T(G)‘, we can
273
 m LIE onoups

identify this vector bundle with the trivial vector bundle G x K, .m that the transpose
of a scalar definenlial operator is ident'yied with a .realar dgfemifial operator.
Then, z'fu e U(G), the transpose qf L“ (rap. R1,) is L: (rap. RX).
We shall consider the case where A"T(G)“ has been trivialized using a right
invariant form to.
Suppose that the proposition has been proved for elemcnm ul, u, of U(G).
Then,
‘(qg = '(Lu, o Lug) (Proposition 23)
= ‘(L,,3) a ”(Lh) (Dz'fl. 69’ Anal. Mam, R, 14.3.3)
= Ll2 o L; by hypothesis
= Lin!“ (Proposition 23)
= n“), (Proposition 7)
and hence the proposition is true for u1 t 11;. It therefore suffices to prove the
proposition when u e T,(G). Now L,‘ is defined by G operating on G on the
right (no. 6) and hence Ohm = 0 since as is right invariant (Diflerentiabl: and
Analytic Manifolds, R, 8.4.5); therefore, iff is an analytic function in an open
neighbourhood of 2 with values in K, then 01m(ft») = (ea (Diflmntiable
and Analytic Manifolds, R, 8.4.8). Using the identifications made and Di mn-
liable and Analytic Mmg'fivldx, R, 14.4.1, the transpose of L" is —L,,, that is
L“.
COROLLARY. Le! G be a finitkdimemianal real Lie group, p. (resp. v) a left (resp.
right) Haar measure on G, I: an integer >0, ueUk(G) and f and g real-valued
flmdiam of class C" on G wilh men/Jae! support. Then

L (1mm = form.» 4'-

L (mm = ja/(n) dv.
This follows fi'om Proposition 57 and Dfiemltiable and Analytic Manifolds, R,
14.3.8.

17. LEFT DIFFERENTIAL

DEFINmON 9. Let G be a Lie graup, M a mangbld of [last 0 andf a mapping of
class C’ If M into C. The 115/? (resp. right) diferential aff 1': the defimxtialfom qf
degree 1 an M with values in L(G) whirh associate: with every veetar u e Tm(M) the
elemmflm) ‘1 - (Tnf) (“) (Few (Tmf) (’4) f0") '1)-
In this chapter we shall only consider the left difl'erential, which we shall

274-
 LBFI‘DD’FERENTIAL §3.l7

denote byf ‘1 . if, and leave to the reader the task of translating the results for
the right differential.
Iff is the identity mapping of G, 1’"d is the canoniwl left difl‘erential
form a) of G. Returning to the general case of Definition 8,
(Jr-14.1mm = ‘"10») °T..(.f)
and hencef'1.¢‘1f=f*(m). This implies thatf‘1.dfis ofclass 0'".
Examples. (1) If G is the additive group of a. complete normable space and
To(E) is canonically identified with E, f "1.4;” is the difl‘erential 11f defined in
DWermIiable and Analytic Mani/Eddy, R, 8.2.2.
(2) Suppose that G is the multiplicative group A‘ associated with a com-
plete normable algebra A. Thenfcan be considered as a mapping of M into A
and hence the differential dfin the sense ofDiflnzntiable and Analytic Manijbldr,
R, 8.2.2 is defined and the product f ' luff in the sense of iflrmh'able and
Analytic Mani/"0M1; R, 8.3.2 is defined. Clearly the latter form is identical with
the left differential off.
Psoposmon 58. Let G and H be two Lie groups, M a mmfild nf clas: 0’, f a
mapping qfclas: C' q into G and}! a mtphitm q into H. Then
(I: °f)“-d(h =f) = we o (run = (h'1.dh)o mm
For all 16M and u eT,(M),
(ll °f)"-4’(ll °f)(") = ((h °f)(x))"-T(ll °f)(")«
The latter expression is equal, on the one hand, to
T(")(f(*)"-T(f) (14)) (§ 2, PI‘OPOSifion 5)
= .(h)((f“.df)("))
and, on the other hand, to
h(f(")) "T(h)(T(f)")
= (Ix-l mama)
PROPOSITION 59. Let G be a Lie group, M a manifizld (folly: 0’ f and g mapping:
qfclass C" qf M into G and]: the canonical my'zction of TM nnIaM
(1) (f5) 1410%) = (Ad °§ ° 1") 1 ° (fa-'1!) + Edi:-
(ii) Writing Mm) = f(m) " fur all m s M,
h-uh = —(Adnfop) nor-1.4;).
Assertion (i) follows from § 2, no. 2, Proposition 7. Assertion (ii) follows from
(i) by putting g = h.
COROLLARY 1. Let :EG and 5g be the mapping x»:g(x) If M into G. That
(Xg)“.d(5g) = (Mg.
275
m LIE GROUPS

This follows from Proposition 59 (i) taking f to be the constant mapping

5: 1-) .r of M into G.
COROLLARY 2. Ifthe mappings/and g qf M into G ham the same lefl dfnential, the
tangent mapping to jig“ is everywhere zero. [ffimher K 1: of oharactemtw 0, then
fig“ is locally eomtant.
By Proposition 59,

Us")"-d(fg“) = (Adww) °(f"~df) - (Ad°g°fl) ° (g"-11£)-

Iff“.df= g‘1.llg, then Uri)“. dug") = 0, that is T,‘(fg‘1 =. 0 for all
x e M. This provm the first assertion. The second follows from it by Dyfimtttable
and Analytic Manifolds, R, 5.5.3.
PROPOSITION 60. Let G be a Lie group, offinite dimension if K it If characteristic
>0, M a manifialdqfclax: C',_fa mapping ty‘clax: C' 0fM into G and an the kfl
difimltial 9/”f. Then do: + [a]“ = 0.
Let 0: be the canonical Ielt difierential form of G. Using Corollary 1 to
Proposition 51 , no. 14, we have

44 dome» =f*(dc->) =f*(-[w]“)

-Lf*(m)]’ = —[«1*.
l8. LIE ALGEBRA OF A LIE GROUP GERM
In this no. (G, e, 0, m) denotes a Lie group germ. A large part of the results
of the § are still true with the same proof. We shall review those which we shall
find useful.
18.1. Let Q be the set of definition of m. Let (g, g’) (-39, teT§”)(G),
t’ E'I‘ffi'KG). As in no. 1, the convolution product of t and t', denoted by
N t’, is the image of t ® 1’ under in. We write U(G) = T2”)(G),
U.<G) = 1*."(G).U+<G) = T:°”*<G). Una) = TMG). For t. t’in U(G).
ta: t’ is defined and belongs to U(G). With the convolution product, U(G) is
an associative algebra with unit element 5,, filtered by the U,(G). The canoni-
cal isomorphism is» of gr U(G) onto TS(T,(G)) is an algebra isomorphism.
18.2. Let G, H be Lie group germs and ‘15: G —> H a morphism. 1ft 6 U(G),
the image U(¢)(t) oft under <1), is an element of U(H) and U(o) is a mor-
phism of the algebra U(G) into the algebra U(H). The mapping 0: z >—> x' 1 of
G into G defines amappingt» I" ofU(G) into U(G). For t, t’ in U(G), the
product 1 a: t’ evaluated relative to G" is equal to the product t’ a: t evaluated
relative to G and (I * I')v = t"’ x IV . Then U(¢)(t") = (U(¢)t)". If
G1, . . ., G,I are Lie group germs and G = G1 x - - - x G," the canonical
isomorphism ofU(G1) ®v - - ® U(Gn) onto U(G) is an algebra isomorphism;

276
 Lm AMEBRA or A LIE axon? GERM § 3.18

for t1,...,l,, in U(G), (tl ®m®tn V = 1; @...® '31- Let H be a Lie

subgroup germ of G and i: H —> G the canoniml injection. Then U(i) is an
injecfive homomorphism of the algebra U(H) into U(G) and
WIN") = (U(i)(‘))v
for all Is U(H). With the convolution product and the coproduct defined by
the manifold structure on G, U(G) is a bigebra and U((b) is a bigebra mor-
phism.
18.3. Let G be a Lie group germ, X a manifold of class C' and g]; a law
chunk oflefi operation ofclass C' ofG on X. Let Q be the set of definition ofip.
Ifte Tff’(G), u 5 TL“ ’(X), (g, x) E Q and .r + 5’ < 7, let t t- n denote the image
oft ® u under 4“. Let t e T‘,"(G), t’ E 'I‘I’(G), u s '13:"(X) ; if: + 1’ + 3” S r
and gg’, (gg’)x, g’x, g(g’x) are defined, then
(ti-1’)“: = ”(l'tu).
Let in, e X and p(xo) be the mapping g >—> gxo, which is defined in an open
neighbourt of e. If t e U,(G), then p(xo) .t = t a: 2‘“. Here and in the rest of
this no., we shall leave to the reader the task of translating the results for law
chunks of right operation.
18.4, Preserving the notation of 18.3, let teU,(G) with 5 $ r. Let f be a
function of class C’ on X with values in a Hausdorff polynormed space. Let
t *f denote the function on X defined by
(HfXx) = <!,g*->f(¢(9(g),*))>
= 031“1 PM) = (9(X)*(‘V),f> = (Iv*ex,f>~
If teU,(G), t’ EU,«(G) and .r + s’ S r, then <t’,t*f> = G" s t',f> and
(t s t’) if: H (I' sf). Let IeU.(G),fandf’ be functions ofclass C’ on X
with values in Hausdorfl‘ polynormed spaces F, F' and (u, u’) H u.u’ be a con-
tinuous bilinear mapping of F X F’ into a Hausdorff polynormed space; let

‘2 t‘ ® I,' be the image of I under the ooproduct; if s g r, then

m (m = Z (I. ma: *f’)~

18.5. Preserving the notation of 18.3, let I E U,(G) with s S I. The mapping
1: >—> t is ex is called the field of point distributions defined by I and the law
chunk of operation and is sometimes denoted by D,‘ or D‘. Iff: X —> F is a
function ofclass C’, the function tv sfon X is also denoted by D:f; it is ofclass
C"'if: < co. IftEU,(G), t’ E U,.(G) and: + .r' g r, then Dwf= D,V(D,f).
If G and X are finite-dimensional, D, is a difl‘erential operator on X of order
S: and of class C"' (if s < 00). The function DJ is then the transform off
under this difiermtial operator.

277
 m LIE GROUPS

18.6. Let G be a Lie group germ and t E U(G). L. denotes the field of point
distributions g >—> :, a: ton G and R‘ the field ofpoint distributions g >7> t A: e, on
G. Iffe WKG, F), then L,fe @“(G, F) and RJE W‘KG, F). For 1,! in U(G),
Lm' = L: ° Ln RM’ = Rl’ ° R!) L: ° R:’ = Rt“ ° Ln 0(L:) = R:-

13.7. As T.(G) is the set of primitive elements of U(G),

[11(0), T9(G)] C TAG)

The normable space T5(G), together with the bracket, is a normable Lie '
algebra, mlled the normable Lie algebra of G (or Lie algebra of G) and de-
noted by L(G). Let E(G) be the enveloping algebra of L(G). The canonical
injection of L(G) into U(G) defines a homomorphism n of the algebra
E(G) into the algebra U(G); if K is of characteristic 0, 1) is a bigebra isomor-
phism, by means of which U(G) is identified with E(G). Using the notation of
18.3, for all a e L(G), let D, be the field of point distributions defined by a on
X. The mapping (a, x) >—> Da(x) is a morphism ofclass C" l of the trivial vector
bundle L(G) X X into the vector bundle T(X). Let I be an open subset of K
containing 0 and y: I —> G a mapping of class C’ such that 7(0) = 2. Let
a = To(y)l e L(G). Iff: X—>F is a function ofclass C’, then

(DJW) = fimksm,t~ok‘1£f(vik)x) -f(x))-

If r 2 2, the mapping a -—> D, is a law of left infinitesimal operation of class
C"1 ofL(G) on X.
18.8. Let G and H be Lie group germs and o a morphism of G into H. The
restriction of U(¢) to L(G), which is just T,(¢), is a continuous morphism of
L(G) into L(H), which we denote by L(zb). If 4; is a morphism ofH into a
Lie group germ, then L(sl; o 4)) = L(qa) a L(qs). For J} to be an immersion, it is
necessary and sufficient that L(tb) be an isomorphism of L(G) onto a Lie sub-
algebra of L(H) which admits a topological supplement. In particular, if
G is a Lie subgroup germ of H and d; is the canonical injection, L(G) is iden-
tified with a Lie subalgebra of L(H) by mans of L(o). If (Go‘s, is a finite
family of Lie group germs and G is their product, L(G) is canonically identi-
fied with H L(Gl).
18.9. Let G be a Lie group germ, of finite dimension if K is of characteristic
> 0. Let F be a complete nonnable space. Let at be a differential form of degree
I: on G with values in F. a is called left invariant on G ifot, is derived from at, by
the mapping I: I» gh of a neighbourt of: onto a neighbourhood ofg. If at is
left invariant, :1 is analytic. The mapping a: x» u, is a bijection of the set of left
invariant differential forms of degree k on G with values in F onto the set of
continuous alternating k-linear mappings of T,(G) into F. If at, = Id-mm, a: is

278
 PASSAGE FROM LIE ALGEBRA nonpmsms T0 LIE GROUP uoapmsms §4.l

called the left mmmieal dtflerentialform of G. The definitions of right invariant

differential forms and the right canonical difl‘erential form of G are analogous.
If m is the left canonical differential form of G, then dc) + [m]a = 0. Let M be
a manifold ofclass Cr andfa mapping ofclass C' of M into G. The left differen-
tial off, denoted by f ' 1.11;”, is the differential form of degree 1 on M with
values in L(G) which associates with each vector u e T,,,(M) the element
f(m)‘1.(T,,,f)(u). Thenf'1.df=f*(m) and do: + [at]2 = 0. Iftwo mappings
fand g of M into G have the same lefi differential and K is of characteristic 0,
thenfg' 1 is locally constant.

§4. PASSAGE FROM LIE ALGEBRAS TO LIE

GROUPS

Recall that, until the end of the chapter, K is assumed to be of characteristic

0.

l. PASSAGE FROM LIE ALGEBRA MORPHISMS TO LIE GROUP

MORPHISMS
Lemma 1. Let G be a Lie group gem and b a Lie subalgehra of L(G) admitting a
topological supplement. The union of the g1} (mp. bg) for geG i: an integrabls
vector mbbundle of T(G).
By considering the left trivialization of T(G) (§ 2, no. 3), it is seen im-
mediately that the gb, for g e G, are the fibres ofa vector suhbundle E of T(G).
Let g e G. The set of (L,),,, where a e b, is equal to gb. Now, ifa and b belong
to b, then [Lm L,] = Lu.“ and [a, b] e [)4 Hence E is integrable (Difl’renlioble
and Analytic Manifolds, R, 9.3.3 (iv)). The argument is similar for the bg.
The integral foliation (Dfirmtiable and Analytic Manifoldr, R, 9.3.2) of the
union of the gt) (resp. bg) is called the left (reap. right) foliation of G associated
with b.
THEOREM 1. Let G and H be Lie group germs andf a continuous morphism of L(G)
into L H .
(i) (Tiler: exil't an open Lie .mbgroup germ G’ of G and a morphism d) of G' into
H :uth thatf = L(¢).
(ii) Let G1, G2 be open Lie .mbgwup gem of G and d), a morphixm of G‘ into
H such thatf = L((ln) for i = I, 2. Then $1 and $2 coincide on a neighbourhood if e.
Let plzG x H—>G, pn X H—>H be the canonical projections. For
all (g, h) e G x H, let f,.,, be the mapping ga»—>l1f(a) of T,(G) = IgL(G)
into Th(H) = hL(H). By considering the left trivializalions of T(G) and
T(H), it is seen immediately that the fm ,, define a morphism ofp(G) into

279
III LE GROUPS

p;T(H). Let a be the graph off; it is a closed Lie subalgebra ofL(G) >< L(H)
which admits {0} x L(H) as topological supplement. For all (g, h) EG X H,
the graph off,‘ ,, is (g, h) .a, The union of these graphs is an integrable vector
subbundle of T(G x H) (Lemma 1). Then there exist (Dyfzrenttable and
Analytic Many’bldr, R, 9.3.7) an open neighbourt U of ‘5 in G and an
analytic mapping 4; of U into H such that (Meg) = on and T,(¢) = ft“,
for all g 5 U. In particular, T,o(d>) = f.
Let V be an open neighbourhood of ea in G such that, for (s, t) s V x V,
the products st and q>(:)¢(t) are defined and :16 U. Consider the mappings .
a“ an ofV x V into H defined by

051(5, t) = WI), «2(5) t) = MOM-V)-

Then a,(t, 2) = (M!) = 1,0, c). On the other hand, let I be fixed in V and
5, be the mapping J v—> no(:, t) ofV into H. Then, for all .t e V and all a e L(G),

T450011) = Tn(¢)(ka) =fu.¢(m(‘m)

= ¢(l~')f(a) =fi,n.m(m)
T.(52)(M) = ¢(¢)T.(<l>)(m) = i>(t)f..o(.)(m)
= i(‘)<l>(:)f(a) =f«,na<->(W)-
Hence (Dzflmntiable and Analytic Manifolds, R, 9.3.7) «1 and a3 coincide on a
neighbourhood of (:9, as). The restriction of 4) to a sufficiently small sym-
metric open neighbourhood oftG is therefore a morphism of Lie group germs,
whence (i).
Let G” G,, ([2,, 453 be as in (ii) and let us prove that 4),, d), coincide on a
neighbourhood of :6. There exists an open neighbourhood W of to such that
$104) = ¢1(t)4>1(:), $303) = ¢,(t)¢2(:) for all 5, tin W. Then, if: SW and
aeL (G) ,
T.(<l>x)(m) = ¢i(3)Te(4>t)(a) = i>«(5)f(a) =f..uw(M)
for i = l, 2. As (1)1035) = ‘3 = 41,05), it follows from Diflnmtiable and Analytic
Manifolds, R, 9.3.7 that <l>1 and 4’2 coincide on a neighbourhood of ca.
COROLLARY 1. Let G and H be two Lie group gmm. IfL(G) and L(H) are im-
morphic, G and H are locally ironwrphio.
This follows from Theorem 1 and § 1, no. 10, Proposition 21.
COROLLARY 2. Let G be a Lie group germ. If L(G) is commutative, G L? locally
isomorphic to the additive Lie group L(G).
The Lie algebra of the additive group L(G) is isomorphic to L(G).
Hence
it suflices to apply Corollary 1.
COROLLARY 3. Let G be a Lie group. If L(G) if commutative, G oontainx a com-
mutative open subgroup.

280
 PASSAGE mom LIE menus TO 1.11: (moms §4.2

There exists an open Lie subgroup germ U of G which is commutafive

(Corollary 2). Let V be a neighbourhood of e such that V“ C U. Then
xy = ya: for all x, y in V. Hence the subgroup of G generated by V is com-
mutative; it is obviously open.

2. PASSAGE FROM LIE ALGEBRAS T0 LIE GROUPS

We shall denote the Hausdorff series by H(X,Y) (Chapter II, §6, no. 4,
Definition 1).
Lemma 2. Let L be a complete normed Lie algebra aver R or G. Let G be the set of
zeL suth that "It” < }10g%. Let 0 be themappiag x:—> -x ofG into G. Let H
be the reshietioa to G x G of the Hausdoqffunetian of L (Chapter II, § 7, no. 2).
(i) (G, 0, 6, H) it a Lie group germ.
(ii) Let 4) be the identity mapping of G into L. The dlfl'erential if 4) at O is an
isomorphism of the normable Lie algebra L(G) onto L.
(i) follow: from Chapter II, § 7, no. 2.
As <1) is a chart on G, the diflerenfial vb 0N) at 0is an isomorphism of normable
spaces. On the other hand, the expansion as an integral series H = ”2):“ H”
of the mapping H is such that Hu(::, 3/) = fix, y]. By § 3, Proposition 24, for
all a, b in L(G),

¢([a, 17]) = Hn(¢(a): Mb» - H11(‘i’(b):‘l‘(“)) = [41(4): 445)]

which proves (ii).
G is called the Lie group germ defined by L.
Suppose that K is ultrametrie. Let p be the characteristic of the residue
field ofK. p # 0, let A = lplm’d’; if]: = 0, let A =1.
Lemma 3. Let L be a complete normed Lie algebra over K. Let G be the set of x e L
such that “x” < A. Let H: G x G —> G be the Hawdorffuaetion ofL (Chapter II,
8 no. 3 . '
§ (’i) With the law {if composition H, G is a Lie group in which 0 is identity element
and —-x the inverse 0f 2:for all x e G.
(ii) Let (b be the identity mapping cy’ G into L. The difeflntial of (b at 0 is an
isomorphism of the normable Lie algebra L(G) onto L.
(iii) For all p. ER:, let G, be the set qfxeL mth that “at“ < a. Thea the G“,
fir p. < A, form a fundamental system 1y” open and dosed neighbourhoods 9/ 0 and are
subgroup: of G.
Assertions (i) and (iii) follow from Chapter II, § 8, no. 3, Proposition 3 and
(ii) can be proved as in Lemma 2.
G is called the Lie group defined by L.

281
III LIE GROUPS

group
THEOREM 2. Let L be a complete normable Liz algzbra. Than exirtr a Lie
germ G rash that L(G) is irornarphia to L. Two such Lie group germs on locally
isomorphic. .
The first assertion follows from Lemmas 2 and 3. The second assertion
follows from Corollary 1 to Theorem 1 of no. 1.

COROLLARY 1. Let G bl a Lie group. There exirt: a neighbourhood if a which [an-

taim‘ no finite subgroup distinct/ion: {2}. If K = R or C, than exist: an open neigh-
bourhood of e which contains no mbgroufi dittinctfiom {e}.
We write L(G) = L. Choose a norm on L defining the topology on L and
such that "mu < M Hall foranwin L.
Suppose that K = R or G. Let G' be the Lie group germ defined by L.
There exist an open ball U’ of centre 0 in G’ and an isomorphism 45 of the
Lie group germ U' onto an open neighbourhood U of a in G. Let V' = iU’,
V = ¢(V’), H be a subgroup of G contained in V and h E H. Let
x = ¢'1(h) eV’. Ifx #- 0, there exists an integern > 0 such that at, 2x, . . ., M
are in V', (n + 1)e', (n + l)1¢V'. Then [1, hz, . . ., h’| are in V,
h“*IEU, hn+1¢v’

which is absurd. Hence H = {5}.

Suppose that K is ultrametric. It suffices to prove the corollary when G is
the Lie group associated with L. If g e G, the powers ofg evaluated in G are
the elements of 23 evaluated in L. These are distinct if g aé :. Hence G
contains no finite subgroup distinct from {2}.

COROLLARY 2. Let k be a non-discrete cloud rubfwld of K, G a Lie group over It

and L = L(G). Suppose that L ha: a normable Lie K-algabra mutton L', com-
patible with the normable Liz Ic-algeln'a rhwtur: and inuarianl under the adjoint repre-
rmtatian of G. Then there exists on G on: and only one Lie K-group :lructur: compatible
with the Lie k—group ”mature andfor which the Lie algebra is L'.
There exists a Lie group germ Gl over K such that L(G1) = L’ (Theorem
2). By Corollary 1 to Theorem 1 of no. 1, G and G,, considered as Lie I:-
group germs, are locally isomorphic. Hence there exist an open neighbour-
hood G' of a in G and a Lie Kvgroup germ structure on G’, with Lie algebra
L, which is compatible with the Lie group germ structure over k. Let V be
a symmetric open neighbourhmd of z in G such that v: C G’. Let g e G.
Then (I) = Int g is a k-isomorphism of a sufficiently small open Lie subgroup
germ of G’ onto an open Lie subgroup germ of G'; and T,(¢) is K-linear
and hence T,(¢) is K—linear for x sufficiently close to 5; therefore the restric-
tion of Int g to a sufficiently small open neighbourhood of 2 in V is K~analytic
(Dyfimm‘able and Analytic Manifolds, R, 5.14.6). By '9‘ 1, no. 9, Proposition 18,
there exists on G an analytic K-manifold structure with which G is a Lie

282
 PASSAGE mom LIE AUGEBRAS 'ro LIE GROUPS §4.2

K-gmup and V is an open sub-K-manil'old of G. By translation, it is seen

that the underlying k-manii‘old structure of G is the given structure. The Lie
algebra of the Lie K-group G is the same as that of the open Lie K—subgroup
germ V and hence is L’. Finally, the uniqueness stated in the corollary fol-
lows fiom § 3, no. 8, Proposition 32.

THEOREM 3. Let G be a Lie group gem and b a Lie suhalgebm of L(G) admitting
a Iapologicul supplement. There exist: a Lieruhgroup gem H of G such that L(H) = b.
If H1 and H, are Lie subgroup germs of G such that

L(H‘) = L(H,) = b,

thenH1 nH, ixapminHl andH,.

There exists a Lie group germ H' with Lie algebra isomorphic to I) (Theorem
2). Shrinking H' if necessary, it can be assumed that there exists a morphism
(b of H' into G such that L((b) is an isomorphism of L(H') onto I; (no. 1,
Theorem 1). As I) admits a topological supplement, (I) is an immersion at e.
Hence, shrinking H’ further, it can be assumed that (b is an isomorphism of
the manifold H’ onto a submanifold of G. This proves the existence of H.
The second assertion follows from the following proposition:

Pnoposmou 1. Let G be a Lie group germ and H and H’ two Lie subgroup gen/u.
In order that L(H) => L(H’), it is memory andsufia'ent that H n H’ be open in H'.
If H n H’ is open in H', then L(H’) = L(H n H’) C L(H). Suppose that
L(H) D L(H'). Let i, i’ be the canonical injections of H, H’ into C. By
shrinking H’ if necessary, it can be assumed that there exists a morphism lb
of H' into H such that L(qa) is the canonical injection of L(H’) into L(H)
(no. 1, Theorem 1). Then L(i o 4;) = L(i’) and hence there exists a neigh-
bourhood V of:1T in H’ such that i 0 vi; and 1" coincide on V (Theorem 1).
Therefore V C H, hence VC HnH’ and HnH' is open in H' (§1, no.
10).

Pnorosmon 2. Let G be a Lie group over K, I: a non—discrete closed subfleld of K

and H u Lie subgroup if the Lie k—gmup G. Suppose that L(H) is a vector sub-K-
space of L(G) which admit: a topological supplement. Then H is a Lie subgroup Qf
the Lie K-group G.
There exists a Lie subgroup germ H' of the Lie K-group G such that
L(H') = L(H) (Theorem 3). Consider G, H, H’ as Lie k-g'roup germs;
Theorem 3 then proves that H h H' is open in H and 11'. Hence there exists
an open neighbourhood U of e in G such that U n H is a submanifold of G
over K. Therefore, H is a Lie subgroup of the Lie K»group G (§ 1, no. 3,
Proposition 6).

283
I" LIE GROUPS

3. EXPONENTIAL MAPPINGS
open neighbourhood
THEOREM 4. Let G be a Lie group germ, L it: Lie algebra, V an
= IdL.
of 0 in L, <1; an analytic mapping of V into G me}: that M0) = 0 and To(¢)
Thefollowing eonditionr are equivalent: .
(i) For all b e L, “(A + A’)b) = d»(lb)¢(1’b)for |x| and DJ| .mfinenlly mall.
(ii) For all b e L and every integer n > 0, <1>*(b") is homogeneous of degree n 1n
U(G) (T1,”(L) is identfiied with TS(L) and b" it enoluated in TS(L)).
(iii) The mapping 4’: of TS(L) into U(G) is compatible with the graduation: of
-
TS(L) and U(G).
(iv) The mapping 4": of TS(L) into U(G) is the canonical mapping of TS(L)
into the enveloping algebra y“ L.
(v) There exist a norm on L defining the topology of L and such that
"Biz/ll] S M M
for all x, y in L and an open subgroup germ W C V of the Lie group germ defined
by L (no. 2) such that (MW 1': an isomorphism of W onto an open Lie subgroup germ
a G.
f(v) => (i): obvious, for (110.0(1)) = (A + k')b in W for m and [NI suffi-
ciently small.
(i) => (ii): suppose that condition (i) is satisfied. Let b EL. Let L}; be the
restriction of 4; to V n Kb. By hypothsis, there exists a symmetric neighbour-
hood T of 0 in the additive Lie group Kb such that '4! [T is a morphism of the
Lie group germ T into G. Hence
454”") = (“TM“) = ((4111345)? = “MUD",
so that ¢*(b") is homogeneous of degree n in U(G).
(ii) => (iii): this follows from the fact that TS"(L) is the vector subspace of
TS(L) generated by the n-th powers of the demerits of L (Algebra, Chapter IV,
§ 5, Proposition 5).
(iii) r» (iv): the canonical mapping of TS(L) into the enveloping algebra
of L is the unique morphism of graded cogebras mapping 1 to 1 and extending
IdL (Chapter II, 13‘ 1, no. 5, Remark 3). Now (I). is a cogebra morphism and
¢*|L = IdL by hypothesis. If condition (iii) holds, it is seen that condition
(iv) also holds.
(iv) => (v): suppose that condition (iv) is satisfied. Choose a norm on L
defining the topology of L and such that ||[x, y] I} s “x“ H y“ for all x, y in L.
Let H be the Lie group germ defined by the normed Lie algebra L. By
Theorem 1, there exist an open subgroup germ S C V of H and an iso-
morphism o’ of S onto an open subgroup germ of G. As we know already that
(v) => (iv), the mapping o; of TS(L) into U(G) is the canonical mapping of
TS (L) into the enveloping algebra of L. Thus ¢,(t) = (1);“) for all t E T5,”)(L).
As <12 and 4)’ are analytic, d} and l” coincide on a neighbourt of 0.

284
 EXPONENTIAL mpmcs § 4.3

DEFINITION 1. Let G be :1 Liz group gem and L its Lie algebra. An exponential
mapping if G is any analytic mapping 4: dqined on an open neighbourhood 1y” 0 in L,
with values in G and satisfying the conditions of Theamn 4.
Theorem 4 implies immediately that, for every Lie group germ G, there
exists an expzmential mapping of G and that two exponential mappings qf G mincidt
on a neighbourhood qfO.
Examples. (1) Let G be the additive group of a complete normable space E.
The canonical isomorphism of L(G) onto E satisfies condition (i) of Theorem
4- and is therefore an exponential mapping of G.
(2) Let A be a complete normed unital associative algebra. Let A* be the
Lie group consisting of the invertible demerits of A. We identify L(A*) with
A (§ 3, no. 9, Corollary to Proposition 33). If K = R or C, we know that
the mapping exp of A into A‘ defined in Chapter II, § 7, no. 3 satisfies condi-
tion (i) of Theorem 4- and hence is an exponential mapping. Now let K be
ultrametric. Let p be the characteristic of the residue field of K. If p 96 0,
let A = ”41””);i = 0, let A =1. Let U be the set ofxEA such that
[l < A. We know (Chapter II, §8, no. 4) that the mapping exp of U
into A‘ satisfies condition (i) ofTheorem4ahd henceis anexponential mapping.
Note that U is an additive subgroup of A.
This example explains the terminology adopted in Definition 1.
Let G be a Lie group germ and (1) an exponential mapping of G. Then
(l) is étale at 0 and hence there exists an open neighbourhood U of 0 in L(G)
such that (MU) is open in G and ¢|U is an isomorphism of the analytic mani-
fold U onto the analytic manifold ¢(U).
A cananical chart (qf thefirst species) on G is a chart qi'on the analytic mani-
fold G whose inverse mapping is an exponential mapping. If further G is
finite-dimensional and a basis ofL(G) is chosen, the coordinate system defined
by s]; and this basis in the domain of £1) is called a canonical coordinata system (of
thefirst species).
PnorosmON 3. Let G be a Lie group germ, L its Liz algebra and (I) an exponential
mapping of G. Let L1, . . ., L" b2 vector subspaces qf L suth that L is the tapologiml
dimt sum qf L1, . . ., L,,. The mapping
(b1) 11:, ~ ~ -, 17..) >-> 0(b1, 52. ~ ~ u bn) = #0044112) . . w.»
defintd an an open subset (y'L, >< Lz x - ~ ~ x Lm is analytic. The tangent mapping
at (0, 0, . . ., 0) to 0 is the canonical mapping (y‘L1 x - - - x L,I into L.
Let k. be the canonical injection of L, into L1 x L2 >< - ~ - x L... Then, for
all bELo (T<u.....u)e)(TokA)(b) = (Tol’xb) = b and hence (Tto.....a)e)lLi is
the canonical injection of Ll into L.
In particular, 0 is étale at (0, 0, . . ., 0). Its restriction to a sufficiently small

285
ll'l LIE GROUPS

open neighbourhood U of (0, 0, . . ., 0) has an open image in G and is 'an

isomorphism of the manifold U onto the manifold 0(U). The inverse mapping
7) of 0(U) onto U is called a canonical [hurl «fl/1e second spam of G, assoeiated
with the given decomposition of L as a direct sum. If further G is finite-
dimensional and each L3 is generated by a non-zero vector 1., the coordinate
system on 0(U) defined by n and the e, is filled a mymm'eal coordinate :yrtem if
the second species:
PaorosmoN 4. Let G be a Lie group gem and (I; an injectiue exponential mapping
ofG. For all x, y in L(G), .

a) x + y = ”33:3;303-14:-1(¢(m¢<xy))
(2) [w] = 3333:; r24»-I<¢<u)¢(3y)¢(—w¢(—Ay)>
(note am <1>“(d>(7»~)¢(7:y)) and ¢-‘(¢(M)¢<xy)¢<—m¢(—xy)) are defined
for |A| sufficiently small .
Let L = L(G) be given a norm defining the topology of L and such that
H [x, y] H < ”x” [| y[| for all x, y in L. Using Theorems 2 and 4:, it can be assumed
G is the Lie group germ defined by L and that 4) = Ida. Let
(x, y) >—> my denote the product in the group G. The formulae to be proved
can then be written
(a) x + y = uxlifiqflk"((kxl.(ky))
(4) [m1 = “333333-=<w>.w).(—m4-33)).
There exists an open neighbourhood V of 0 in K such that the function
lO) = (MHM)
is defined and analytic on V. By Chapter II, §6, no. 4, Remark 2, the expan-
sion off as an integral series about the origin is
W + y) + Why] +
and this proves (3). On the other hand, for u, v in G and flu", |[u|| sufficiently
small, 14.0 is an analytic function of (u, v) and the terms of degrees 1 and 2
in the expansion of this function as an integral series about the origin are
u + v + flu, 1)]. By Dgfirmtz'able and Analytic Manfildx, R, 3.2.7 and 4.2.3,
the terms of degrees 1 and 2 in the expansion of the function f (A) .f(—-)\)
as an integral series about the origin are the terms of degrees 1 and 2 in
, f0) +f(-7\) +l[f(7:):f(-7:)]
or also m
10' + y) + ‘Wlmyl - M” + y) + ll’Dw]
and this pm. (4), + W" + y): —x(~ + y)] = mm]
286
 EXPONENT‘AL MAPPINGS § 4.3

PROPOSlTlON 5. Let G be a Lie group, I: a non-diam: dated mbfield of K, G’ the

group G mnn'dered a: a Lie group over k and (I; (resp. (If) an exponential mapping of
G (resp. G'). Then :17 and (If coincide on a neighbourhood of 0.
(1) satisfies hypothesis (i) of Theorem 4 relative to G’ and is therefore an
exponential mapping of G'.
PROPOSITION 6. Let G be a Lie group germ, L itr Lie algebra and ti): V—> G an
exponential mapping of G. For all x e V, let T,(L) be identified with L, ID that the
right difi’erentiol H(X) oft!) at x i: a linear mapping of L into L. For x Wong close
to 0,
1
1110:) = no
2 (I: + 1)!
(ad x)u'

Let L be given a norm compatible with im topology and such that

]| [x, y] II S Hz“ fly” for all x, y EL. It suffices to consider the case where G is
the Lie group germ defined by L and (I) = Id“. By definition, 130:) is then the
tangent mapping at x to the mapping y»—>y.x“ of G into G. If H(X, Y)
denotes the Hausdorff series, m(::) is therefore, for ||x|| sufficiently small, the
tangent mapping at 0 to the mapping y» H(x,+ g, —x) of G into G. In
H(X + Y, —X), the sum of tem of the first degree in Y is

m; m (“1 X) Y
l I

(Chapter II, § 6, no. 5, Proposition 5). The proposition then follows fiom
Dmrmtioble and Analytic Monifirlds, R, 3.2.4 and 4.2.3.
Let G be a Lie group germ and teK. A t-th power mopping if G is any
mapping, defined and analytic on an open neighbourhood of e, with values
in G and coinciding on a neighbourhwd of e with a mapping

.4 H “14> "(9)
where 4) is an injective exponential mapping of G.
Pnoposmon 7. (i) If 1 e2, a t-th power mapping m'oeide: on a neighbourhood of
e with the mapping g »—> g’.
(ii) The tangent mapping at e to 11 Mb power mapping is the homothrty 12f ratio t.
(iii) If h it a t-th power mapping and h' a t'-th power mapping «f G, h o h’ i: o
(tt’)-th power mapping and g 2—» h(g)h’(g) is o (t + t')-th power mapping,
(iv) [f h is a t-th power mapping and u e U"(G), then h,.(u) = fire.
It suffices to prove the proposition when G is the Lie group germ defined
by a complete normed Lie algebra and when the t-th power mappings con-
sidered are constructed using the exponential mapping 4) = Ida. But in that
use everything is obvious.

287
 III [E GROUPS

4. FUNCTORIALITY OF EXPONENTIAL MAPPINGS

PROPOSITION 8. Let G and H be Lie group germ, h a morphism of G into H and
‘l’o and (1);, exponential mapping: of G and H. There exixtx a neighbourhood V of 0
in L(G) mh that h o (be and <l>u o L(h) ooiee on V.
Let L(G) and L(H) be given norms defining their topologies and such that
|l[x,y]fl S ”or” fly” for all x and y. It can be assumed that G (resp. H) is the
Lie group gem defined by L(G) (rcsp. L(H)), 50 that cl)“ (resp. ’i’n) coin-
cides with MG (resp. Ida) in a neighbourhood of 0. On the other hand, there
exists a symmetric open neighbourhood W of 0 in L(G) such that L(h) is a .
morphism of the Lie group germ W into H. By Theorem I, L(h) coincides
with h on a neighbourhood of 0, whence the proposition.

Loosely speaking, if G and H are identified in a. neighbourhood of the

identity element with L(G) and L(H) by means of exponential mappings,
every morphism of G into H is Iinoar in a neighbourhood of 0.

COROLLARY I. Let G be a Lie group germ, G’ a Lie subgroup germ of G and (b

an exponential mapping of G.
(i) There mil-t: an open neighbourhood V of 0 in L(G’) such that ¢|V i: an iso-
morphirm of the manifold V onto an open neighbourhood of e in 0'.
(ii) Let x e L(G). The fillowing conditions are equivalent: (a) x E L(G’) ;
(b) (box) 5 G’ for [1| mfioiently mall.
(i) is obtained by applying Proposition 8 to the canonical injection of G'
into G and (ii) follows from (i).
COROLLARY 2. Let G be a Lie group, p an analytic linear representation of G and
(I) an exponential mapping of G. There exist: a neighbourhood V of 0 in L(G) such
that

fir all x e V.
9%)) = mam»)
This follows from Proposition 8 and Example 2 of no. 3.
COROLLARY 3. Let G be a Lie group and (b an exponential mapping of G.
(i) There ӣi a neighbourhood V of 0 in L(G) such that

Ad(¢(x)) = exp ad x.
fir all x e V.
(ii) Ifg e G, there exist: a neighbourhood W of O in L(G) such that

g¢(*)g“ = (Ads-x)
fir all n SW.
(i) follows from Corollary 2 and § 3, no. 12, Proposition 44.
(ii) follows fiom Proposition 8 applied to Int g.

288
 smucruu maven) ON A summon-p §4.5

5. STRUCTURE INDUCED ON A SUBGROUP

Lemma 4. Let G be afinite-dimensional Lie group, 0 a symmetric open neighbourhood
of: in G and H a subset «f9 containing a such that the conditions e, yEH,
any" 5 0 imply xy" 5 H. Let re NK. For all x EH, let I), be the set ofn eTx(G)
with thefollowing property: there exist an open neighbourhood I ofO in K and a map-
pingqtltm C’ (#1 into Gmhthatf(0) = x,f(I) C H, (Tof)(l) = a.
(i) Let b, = I). Thm b is a Li: subalgebra 4y“ L(G) which is invariant under
Aduey(H)-
(ii) 1),, = xb = bxfor all x e H (when xb and 1):: are evaluated in T(G)).
(iii) Let V be a manybld oftlass C', 1/0 a point of V andf a mapping of class C'
q into G such thatf(vn) = : andf(V) C H. For every Lia subgroup gm» H'
of G with Li: algabm I}, f (o) e H’ fir 11 mfioimtly olosz to no.
(iv) For every Lie subgroup gem H’ MG with Lie algebra b, H’ n H i: a neigh-
bourhood 1f: in H’.
(v) For all x E H and all a 6 bx, there axis! on open neighbourhood I of 0 in K
andamppingfafcltms C“ofI intohthatf(0) = x,f(I) c H, (Tof)(l) = 11.
Clearly Kb = l) and Mn; = I)“, for x, y, 141/, xyz in H. This implies (ii) and
the fact that b is invariant under AdL(G,(H).
Let ax, a, he in 9. Let I be an open neighbourhood of O in K and f,, f,
mappings of class C’ of I into G such thatf,(0) = o,f,(I) C H, (Tofixl) = a,
(j = l, 2). We definef: I—>G byf(A) =f1(A)f3(Z). Thenfis of class C'
and f (0) = 2. By shrinking I if necessary, we have f (I) C H. On the other
hand, the mapping of T,(G) x T.(G) into T,(G) tangent to the mapping
(g,g') —>gg’ is addition; hence (Tof) = a; + an. Hence a1 + 42 El) and I)
is a vector subspace ofL(G). Since xbx" = b for all x e H, (At:lf1()‘)).a2 e b
for all A e I. The tangent mapping at 0 to the mapping A» Adf1()\) is, by
Proposition 44 of § 3, no. 12, the mapping A »—> 341011,); hence

[01,a2] = (ada,).a,el)

since I) is closed in L(G). Hence we have proved (i). In the rest of the proof
we fix a Lie subgroup germ H' of G with Lie algebra b.
Let V, v0, f be as in (iii). Let Y be the left foliation of G associated with I)
(no. 1). For all y eH’, T,(H’) = yl). On the other hand, for all 1) EV, the
image of T,(V) under T.,(f ) is contained in 9m) = f(v)l) (by definition of
hm»). By Diflerentiabl: and Analytic Manifolds, R, 9.3.2, f is a morphism of V
into Y. As H’ is a leaf of Y (Diferentiable and Analytic Manifolds, R, 92.8),
f (v) e H’ for v sufficiently close to 00.
Let (a1, . . ., a,) be a basis of I). There exist an open neighbourhwd I of 0
in K and mappings f“ . . .,f, of class C' of I into G such that f,(0) = e,
f,(I) C H, (Tf,)1 = a, for all j. By (iii), £0.) eH’ for [A] sufficiently small.
Hence thef,(>.,)f,(l¢) . . .f.(}\,) constitute, for It‘ll, lhzl: . . ., IM sufiiciently

289
11! ME GROUPS

small, a neighbourhood of e in H‘ ; and this neighbourhood is contained in H.

Hence (iv). .
If a e i), there exist an open neighbourhood I of O in K and a mappingf of
class 0’” of1 into G such thatf(0) = e,f(I) C H', ('I‘of)l = a. Tlus, With
(iv), implies (v).

DErmmON 2. b is called the mbalgehra tangent to H at e.

PROPOSITION 9. Let G be afinite—dimensional Lie group and H a subgroup of G. . .

(i) There exist: on H one and only one analytic maxi/Md structure with thefollawmg
property: fir all r between 1 and m, for every mamfild V of class C“ and flowery
mappingqVintaH,fi:¢fcla:sC'asamappingofVintoHifandonlg/gffu
qfclaxx C’ as a mapping cf V into G. . .
(ii) With this structure, H is a Lie group, the canonical injection 1' of H We G 1:
an immersion and L(i) (L(H)) is the Lie subalgehra tangent at e to H.
In (i) the uniqueness is obvious. We prove the existence. Let i) be the Lie
algebra tangent at e to H. Let H’ be a Lie subgroup germ of G with Lie
algebra I). By replacing H’ by an open subgroup germ of H’, it can be as-
sumed that H’ C H (Lemma 4 (iv)). For all x e H, :eH'Je'1 is a Lie subgroup
germ of G with Lie algebra xbr" = l). Hence H’ n (xH'x‘l) is open in H’
(no. 2, Theorem 3) and the mapping 3] axgjx" is an isomorphism of
H’ nx‘lH’x onto xH’x" n H’. Using Proposition l8 of § 1, no. 9, there
exist an open Lie subgroup germ W of H’ and a Lie group structure on H
with the following properties: W is open in H and the manifold structures
on H and H’ induce the same structure on W. It then follows that the canon-
ical injection i of H into G is an immersion and that L(i) (L(H)) = L(H’) = 1).
Moreover, let V andfbe as in (i). Ifs—>H is ofclass C', inf:V—>G
is of class 0'. Suppose that inf : V—> G is of class C'; then we prove that
f: V —> H is of class C’. By translation, it suffices to consider the case where
there exists voEV such thatf(vo) = e and to prove that f:V—>H is of
class C” in an open neighbourhood of on. Now, by Lemma 4 (iii), f (1;) EH’
for a sufficiently close to v“, whence our assertion. Thus we have proved (i)
and (ii) has been obtained on the way.

DEFINITION 3. The Lie group structure an H defined in Proposition 9 is called the

structure induced on H by the Lie gruup structure on G.

If H is a Lie subgroup of G, its Lie group structure is induced by that on G

(Diferentiahle and Analytic Manifolds, R, 5.3.5).

If G =. R and H = Q, then b = (0} and hence the structure induced on H

is the discrete Lie group structure. Similarly if G = C (considered as a
complex Lie group) and H = R.

290
 PRmmVPs 0P Dmum‘ml. FORMS §4.6

irmnmornmmmronmmmvummnm
ALGEBRA ,
Lemma 5. Let X be ammifizld 0felas: C', FandF' veetorbundlerofelar: C’ with
base:pmeXand¢amorphianq/FintoF’.FarallxeX,letbelhesetqf

(a, W» e F. (B F;
for a sh. Then the union 5 ofthe s,‘ is a veaormbbmdla ofF @F’.
Let 9 and 6’ be the mappings of F (-BF’ into itself defined as follows: if
(u, 11) 6F, ® F;, then
9(a) '1) = ("HI + Mu». 9’04?) = (14,” — $00)
By Dzfirentiable and Analytic Manifaldx, R, 7.7.1, 0 and 8’ are morphisms of
F (-BF’ into itself. Clearly 0 o 6' = 0' o 0 = Idfepr. Hence 0 and 0’ are
automorphisms of F G F’. Therefore, S = 0(F @(0}) is a vector subbundle
of F a F’.
Lemma 6. Let G be a Lie group germ, m the canonical lefl dzflerential form qf G
(§ 3, no. 18.9), M a mung/721:1 qfclm C’ (I 2 2) and a: a dg'fl’erenfialform ofelas:
G"1 and degree 1 on M with value: in L(G).
(i) The element: q(M x G) at which the dtferentialform
0 = prfa — lax-:0)
is zen; constitute a vector .mbbundle S ofT(M x G) qftlaxs 0'“.
(ii) For all (x,g) EM x G, T(pr1)|S(,,',, is an iromorphirm qf SUM, auto
T (M).
’‘(iii) If do: + [a 2 = 0 (cf. §3, no. 14-) the vector subbundle S is integrable.
If (2,5) 6 M x G and (u, u) eTx(M) X T,(G), then
e(x.n("x u) = 10‘) — g'l'“
Hence the kernel of 00",, is the set Sam, of (u, ga(u)) for u sTx(M), whence
(ii). We consider T(M x G) as the direct sum of two vector bundles F and
F' with E...) = TAM) x {0} and F2...) = {0} x T.(G) for an
(x, g) e M x G.
For it e Tx(M) x {0), we write (Mu) = (0, ga(u)). Using the left trivialization
of T(G), it is seen that (I; is a morphism of F into F', whence (i) (Lemma 5).
Finally, if act + [at]2 = 0, then
:16 prfldu) —— pram)
—%(pri‘ a A P’i a - Pr; 6 A Pr: a)
— —*l(Pri‘¢c — PT? 0’) A (Prl'a + Pr: 0*)
= —;e A (prra + pea)
and hence S is integrable (Diflererm'able and Analytic Manyaltir, R, 9.3.6).

291
m up. (moors
d¢
M5. LetGlzeuLic ou gem,Manmmfold¢folas:C' (r22)ari
in_L(G), such the:
33$;tial form if l‘lflfl‘ 0’5, ail degree 1 on M with values
Mined and
dot + [at]“ = 0. For all are M and all geG, there exutr a mappzngf,
that f (z) '= g
of elm: CY“ on an open neighbourhood of x, with values 1.". G, such
on a neigh-
and f ‘1.df = a. Two mopping: which satisfy there condition: coincide
.
ho hood 0 x.
"Itet x éfM and ge G. By Lemma 6 (whose nomtion we adopt) and Dif-
od
ferentiable and Analytic Manifolir, R, 9.3.7, there exist: an open neighbourho
U ofx in M and a mapping rm» Mm) = (rn,f(m)) ofclass C"1 ofU mto
M X G such thatf(x) = g and ¢*(0) = 0. Then

f".tlf=f‘(o) (§3,no.18.9)
= (P's ° 4’) ‘ (40) “Off = 9’2 ° i)
= win: a — 0) (Lem-a 6)
= (WW? 00 (for 4>*(°) = 0)
= at (for pr, 0 (l: = Id“).

Let f’ be a mapping of class C"1 of U into C such that f’(x) = g and

f"’df' = a. By §3, 18.9,fl'"‘ is locally constant and hence f' =fin a
neighbourhood of x.

PROPOSITION 10. Let M be an analytic manifold, 53 a complete normable Lie algebra

and at an analytic diflerentialform ofdegree 1 on M, with values in g, with thefollowing
properties:
(a) for all m e M, an, i: an isomorphism rm(M) onto 9;
(b) dot + [oi]2 = 0.
Then, for all m‘J E M, there exit! on open neighbourhood M’ y“ mo in M and a Lie
group germ structure an M’, compatible with the manifold structure on M', with
identity element mo and with thefillowing properties:
(i) am ii an isomorphinn 4y” L(M’) onto g;
(ii) the diflhantialform In r—> on}: 0 am is the left canonical dyferential form of M’.
[fMfl and M; are two such group germx, M; and M; have a common open xubgroup
germ.
There exists a Lie group germ G such that L(G) = 9, Let mo EM. By
Theorem 5, there exist an open neighbourhood M’ ofmn in M and an analytic
mapping f of M' into C such that f(rna) = e and f ".df = 11. Then
Tmn( f) = “no is an isomorphism om°(M) onto 9; hence, shrinking M' and
C, it can be assumed thalf is an isomorphism of the manifold M' onto the
manifold G. We transport to M’ the Lie group germ structure on G by means
of f '1. Then Tm,(f ) becomes an isomorphism of L(M’) onto L(G) = g,

292
 PRIMITIVES or DIFFERENTIAL FORMS §4.6

whence (i). On the other hand, if c.) denotes the left canonial differential
form of G, then
”Jim". = (Tmf)_1°(f_1‘4f)(m)
= (Tmf)“°m(f(m)) °Tmf
and hence m r—> am'al o a", is the left: canonical differential form of M’.
Let M” be an open neighbourhood of mo, with a Lie group germ structure,
with identity element Inn and with the analogous properties to properties (i)
and (ii). Then an,“ is an isomorphism of L(M') onto 9 and also of L(M”)
onto 9 and hence L(M’) = L(M"). Therefore, shrinking M’ and M”, it can
be assumed that there exists an isomorphism (I) of the group germ M’ onto
the group germ M” (no. 1, Corollary 1 to Theorem 1). Then ¢‘1.dd> is the
canonical left differential of M’. On the other hand, let 4; be the canonical
injection of the manifold M’ h M' into the Lie group germ M”; clearly
ill-1.44; is a restriction of the canonical left differential of M”. Hence
(¢‘1.d¢)(m) = 09,701 0 at", = (¢“.d¢)(m) for all meM’ n M”. Therefore (it
and t]; coincide on a neighbourhood of mo (§ 3, 18.9). This proves the last
assertion of the proposition.
COROLLARY. Let M be an analytic many’old offinite dimension n. Let (0,, . . ., a),
be analytic dfirentialfim: 4y” degree 1 on M, with scalar values, which are linearly
independent at each point of M and such that, for all k = l, t . t, n, dink i: a linear
combination with constant coeflicient: of the to. A 0,. Then, for all rn0 EM, there
exist: an open neighbourhood M' of mo in M and a Lie group germ structure an M'
compatible with the manifizld structure on M', with identity element mo and such that
mllM’, . . ., mn|M’fimn a ban": qfthe space oflcfl invariant a'tfl'erentialflrm: on
M' of degree 1 with scalar values.
if M; and M; are two such group gem, M; and M; have a common open sub—
group germ.
Let X1, . . ., X,l be the vector fields on M such that, at each point m of M,
the (X,),,, constitute the basis of T,"(M) dual to ((co,),,,,. t ., (9%)...)- These
fields are analytic. By hypothesis, there exist cw, GK (1 $ 1', j, k S n) such
that cmc = —c,,,, and do», = (<2, cake), A 0),. By DMerentiable and Analytic
Manifolds, R, 8.5.7, formula (1 l),

<[Xt,x,], up = —<dmt)(X.,x,) = {Zena A «Jog, x1) = w;

and hence [X,,X,] = —c,x,‘. It then follows that the —c,,,, are the
constants of structure of a Lie algebra 9 relative to a basis (eh . . r, en). For
all m E M, let a", be the linear mapping of T,,.(M) into 9 which maps (X,),,,
to e1, . . ., (X,),,, to e,,. Then a. is an analytic differential form on M of degree

293
III LIE GROUPS

1 with values in g and a”. is an isomorphism of T,"(M) onto a. On the other

7;
hand, a: = x21 cow: and hence
n
Z (2, 0,3,4», A 0,)0,‘
= k; (4%)!» = 16-1
and

(5) [uragtwuwzwo A (mm) (W formula (30))

(<1

2 (‘1’: A milk“ '1]

——Z 2 (4mm. /\ 0,)9,‘

k- 1 (<1

= —du.
It then suffices to apply Proposition 10.

7. PASSAGE FROM LAWS OF INFINITESIMAL OPERATION TO LAWS

OF OPERATION
PROPOSITION 11. Let G1 and G2 be Lie group gum: and XI and X2 manifoltir qf
clan C’ (7 > 2). Fori = 1, 2, let 44‘ b: a law chunk y’left upemtion qfclaxs C' qf
G‘ on X‘ and D‘ the associatzd law of irgfiniteximal operafiun. Let p: G1 —> G2 be a
marphirm and <1)l —>X2 a mapping of class C". Suppoxe that, for all u E L(G),
the vectarfield: (D1)“ and (Dzlum are (b-relatzd (Diflermtiuble and Analytic Mani-
folds, R, 8.2.6). Then there exists a neighbourhood Q qf{:} x X, in G, X XI
Mk ”WM‘lIfig, #0) = ¢2(u(g),¢(x))fir 4” (£31) 69.
Let A: Glx X,» 9G,, [22: G, x X,—>X2 be the canonical projections.
For all (g, x,) EGl x X2, letf“ x, be the mapping gin» (D,)uu,a(x2) of
T,X(G,) = g1L(G1) into Tx,(X,). Thcf,‘ ,3 define a morphism ofp(G1)
into p*T(Xa)
Let 1'0 5 X1. There exist an open neighbourhood G of e in G1 and an open
neighbourhood X ofxo in X1 such that 4:1(g, x) and q;,(p.(g), ¢(x)) are defined
for (g,x) 5G x X. Wewritc, for (g,x) EG x X,
043:“) = <l>(¢1(g,X)) 6X2, BMW) = ¢2(H(g),<l>(x)) 6X:-
If G and X are sufficiently small, then, for all (a, g, x) e L(G,) x G x X,
(Tax-2&0» = (T¢)((Dx)a(¢x(g:*)))
= (D2)um(<l>(¢x(g.1)))
(Dzluu)n°‘(.¥: 7‘):
(TBWK, 0x) = (T412) (“ma-Mg), <l>(X))
= (Dilumn(‘l’2(l‘(3)1¢("))l
: (Delhmmmga 5‘)~

294
 LAWS OF maximum arm-non §4~.7

Hence for 32X, the morphism gr—>a(g,x) and gr—tfl(g,x) are integrals
off; as ,
90, It) = M") = 1(6) x)
for all xeX, it follows from Dlflerentiable and Analytic Manifoldr, R, 9.3.7,
that a and fl coincide on a neighbourhood of (3, 1°). Hence the proposition.
COROLLARY. Let G be a Lie group gem and X a manifold of tlass C" Comider
two law chunks lfleft operation qfolass C' of G on X. Suppose that, for all a e L(G),
the corresponding vettorfield D,1 on X is the samefor both low chunks. Then these two
law thanks tointide on a neighbourhood of {e} x X.
THEOREM 6. Let G be a Lie group germ, X a mamfild of class C' (r 2 2) and x“
a point ofX. Let a >—>Dn be a law of left infinitesimal operation ofelass C"1 of
L(G) on X.
(i) There exists an open neighbourhood X’ of en in X and a law thunk of left
operation of class C"1 of G on X' such that the associated law of infinitesimal opera-
tion is o ~—> DQIX’.
(ii) Let there be two law [hunks of lefl operation of tlass C"1 of G on an open
neighbourhood X” of :0; if they admit a n—> D¢|X' as associated law of infinitesimal
operation, they coincide on a neighbourhood of (e, no).
Assertion (ii) follows from the Corollary to Proposition 11. We prove (i).
Ferall (g,x) eG x Xand all aeL(G),wewrite
0.441,") = (agyDa(~)) GTAG) X Tz(X)~
Let Sm,” be the set of QA(g, x) for a s L(G). By Lemma 5 of no. 6, the s“,
are the fibra of a vector subbundle S of T(G) x T(X)’. Let a, b be in L((}) ;
then
[Qm Qu](g, x) = “Ra: Rh] (5), [DD Do] (1‘))
= (-Rm.m(g), -Dm.u(’t)) (§ 3: 1&5)
= Q-la.o](g:*)
and hence S is integrable (Difl'ermtiahle and Analytic Manifolds, R, 9.3.3, (iv)).
By Dzfirentiable and Analytic Mam'fizlds, R, 9.3.7, there exist an open neigh-
bourhood G‘ of e in G, an open neighbourhood X1 ofxa in X and a mpping
(g,x) n of class C"1 ofGl x Xl into X such that ex = x for all xeX,
and
(6) (ag)x = D,(gx) for a e L(G), g e G], x 6 X1.
In particular
(7) ax = Duo).
Let G, be an open neighbourt ofe in GI and X2 an open neighbourhmd
of 3., in X1 such that g' is defined and belongs to CI for g, g’ in G2 and gx

295
III LIE GROUPS

is defined and belongs to Xl for (g, x) EGa X Xa- Consider the mappings
a1, :12 ofG, x (G, x X2) in“) X defined by

Mg» (II, 3)) = gl), (120:: (lb 10) = (shy.

They are ofclass 0"1. Then

we, (1%)) = hx = «2(9, (h, 3)).

0n the other hand
T(u:)(ag:Oa-,~>) = (ag)(h*)
= Da(g(h*)) by (5)
= D..(°u(g, (h, ’0)),
T(°‘2)(“£: Oman) = (“5’0”
= a((gh)~) by (6)
= Da(¢2(g. (h, x)))~
By Difl'erentiable and Analytic Mang'fallir, R, 9.3.7, at, and on, coincide on a
neighbourhood of (e, (e, xo)). Then (i) follows from (7) and Proposition 23
of § 1, no. 11. '
COROLLARY 1. Let G be a Lie group germ and X a paracampaet manifizld qf class
C" (r 2 2). Let a »—> D,‘ be a law qflefl infinimimal operation afclass C"l ofL(G)
on X.
(i) There exirt: a law chunk of lefl nperah'on of class C"1 qf G on X such that
the associated law Qf infinitesimal operation is a ~—> D,.
(ii) Two law: 9/ left operation of alas: C"l qf G on X which admit a >—> D, as
associated law of infinitesimal operation em‘neide an a neighbourhood of {e} x X.
Assertion (ii) follows from the Corollary to Proposition 11. By Theorem
6 (i) , there exist an open covering (X1)m of X and, for all i e I, a law chunk
of left operation 4;. of Class C"‘1 of G on X. such that the associated law of
infinitesimal operation is a v—> D3|X.,. As X is paracornpaet, the covering
(X0451 can be assumed to be locally finite. For all (i,j) EI x I and all
:4 ex‘ n X,, 4;. and \b, coincide on a neighbourhood of (e, x) (Corollary to
Proposition 11). As X is normal, we can apply Proposition 24 of § 1, no. ll ,
which prova (i). ‘
COROLLARY 2. Let X be a paracompaet manifold of alas: C' (r 2 2) and E a veetar
field qfclarr C"1 an X. There exist: a law chunk afoperatilm q, velar: C"1 t
an X such that for all x E X, €(x) is the image under ts—> (I, x) of the tangent veetar
l to K at 0. Tum law chunk: of operation with the above praperty caina'de an a neigh-
bourhaaa' of {0} x X.
This is a special case of Corollary 1.

296
 'n-m uoemcmms 5“,, § 5.1

Remark. Laws oflefl operation can of course be replaced, throughout this no.,
by laws of right operation.

§5. FORMAL CALCULATIONS IN LIE GROUPS

Letf, g be two formal power series with coefficients in K in the same indeter-
minates, let )2 (resp. g.) be the homogeneous component off (rexp. g) of de-
gree i. We shall write
f E g mod deg p
iffi =g.fori <p.
In this §, C denotes a Lie group gem of finite dimension n; the base field
K is assumed to be of characteristic zero. We identify once and for all, by
means of a chart, an open neighbourhood of e in G with an open neighbour-
hood U ofO in K", so that e is identified with 0. For x,g in U and n 52, any
denotes the product of x and y and x‘“ the m-th power of x in C (When they
are defined). The coordinates of x e U are denoted by x1, x,, . . ., x".

1. ms COEFFICIENTS c...
Let Q be the set of (nay) e U x U such that mg is defined and belongs to U.
Then 9 is open in U x U and the mapping (x,y) why of 0 into U is
analytic. The coordinates zl, . i ., z,I of z = my therefore admit expansions as
integral series about the origin in the powers of xi, . . ., x,” 3/1, . . . , 3/". There-
fore, there exist well defined constants a,”"u,”‘gbwgflvbuuvn E K, such that

v v _ v B B
(1) 251l ' ' ' Zn" — a" "when clt1_...,m~,9h...,8.,v1,"”1”l - ' - 3:"!11‘ ‘ . 41.1”
for 7‘, . . .,y,, in N. Adopting the conventions of Dgferenh'ablc and Analytic
Mani/Hit, R, we shall write these formulae more briefly:

(2) (W = ”2., a...” (Y5N")~

Since x.0 = 0.: = xfor e,
(3) [man = ‘o.a.v = 8M
where 8“ is the Kronecker index. In particular, writing henceforth I: instead
ofckfork = l,...,n,
(4) (M): = xx + y. + WEN“ awry“-
Writing 5” = (can, M32: . . ., 5.x”) 6 K", it then follows that

(5) any 2 x +y + lungmncflxuyn’

297
111 LIE GROUPS

component of
On the right hand side of (5), we consider the homogeneous
degree 2:

Bow) = ,2, am;

so that (x, y) 1—» B(x, y) is a bilinm mapping of K" x K'I into K". Then

(6) w a x + y + 130%) mod dog 3.

Formula (4) implies on the other hand that

(7) 01.5.1 = 0 ifI‘I‘I + lfll < I‘ll

and that the terms of total degree M in the expansion of z’ are also those of
(x; + mm + m. .. (:1 + y" 'n = W an, own (cf. Dmrenfiabt:
and Analytic Maniflzlds, R, Notation and conventions). Hence:

(8) 0am = 0 if M + [9! = M b1" 0‘ + l5 9* Y

(9) 0mm» = ((0%, 3))-
The associativity of the product implies the relation

=2 that: awry”) z"

0; 0min“(921 Enzy°l")‘
for all x, y, z sufficiently close to 0, whence

(10) Z ‘m’m = 25:11-19:13: (1: B, Y: 7) in N")-

The group germ G admits a commutative open subgroup germ if and only
if ram, = cm for all on, B, y in N“.

2. BRACKET IN THE LIE ALGEBRA

y
For an E N", let en, be the point distribution—
1 at the origin. In particular,
at_'. 3—2"
2,, = a“ = 3—2; The a“ form a basis of the vector space U(G). Iffis an analytic

function on an open neighbourhood ofO in G andf (x) = Z M is its expan-

sion as an integml series about the origin, then

<t«.f> = 7*:-
In particular,
(t,, x”) = 8a,.

298
 BRACKET m m up. ALGEBRA §5.2

Hence
(a, t ‘5: x') = (ea ® :5, (x.y)‘>

= (‘3 ® 0m “2 ‘u’a’fi‘'11")

= 2,, omen, am’><e.,y°'> = cm

and hence

(11) 2.. * a. = 2 cm.

(Formula (10) then expresses associativity on U(G).)
In particular, since L(G) is stable under the bracket.

(12) [5h ‘1] = 2 (‘uk — ‘Ilk)‘k~

The constants of structure ofL(G) relative to the basis (eh . . . , 9,.) are therefore
the cm, — cm. In other words, canonically identifying L(G) with K", we ob-
tain:
(13) [Jay] = BOW) - Emit)-
Pnorosmon l.
(i) xr-n—x + B(x, 1) mod deg 3
III

(ii) any/«(“1 y + [x, 3/] mod deg 3

III

(iii) y"“.x.y E x + [x,y] mod deg 3

(iv) x"”.y[‘“.x.y E [x,y] mod deg 3
(v) xey.x"“.y"“ E [x,y] moddegS.
(In (i), x“ u of course represents the expansion of the function x v—> 3"” as an
integral series about the origin; the other formulae can be interpreted ana-
logously.)
Let g1 and g: be the homogeneous components of x"“ of degrees 1 and 2.
Then
0 = x.x"“
E x + £10!) + 3083109)) mod deg 2 (by (6))
E x + g,(::) mod deg2

and hence g,(x) = —x. Then

0 = x.x““
E x + (—3 + 32(1)) + 30‘: -x + £200) mod (1683
E 320') “ 3(3) x) mod deg 3

299
I'll LIB GROUPS

and hence g,(x) = B(x, x). This proves (i). Then, using (i),
x.y.x"“ E (x + y + B(x,y)).(—x + B(x,x)) mod degB
E x + y + B(x,y) + (—x + B(x,x)) + B(x + y, -x) mod deg3
E y + B(x,y) — B(y, x) mod deg 3
—=— y + [16y] mod deg 3 (by (13))
whence (ii). The proof of (iii) is analogous. Combining (i) and (iii), we obtain
shag/["1114] E (—x + B(x,x)).(x + [x,y]) mod deg3
a —x + B(x,x) + x + [x,y] + B(—x,x) mod deg3
E [x, y] mod deg 3
whence (iv). The proof of (v) is analogous.

3. POWERS
Consider j points of G:
3(1) = (1(1):, “(1)2, . ~ ': 3(1),.)
’42) = (1(2):: 42):, ~ ~ -, 3(2)")

IO) = Mi)» 10):, . - -,'IU)n)~

The mapping (x(l),x(2), . . l, x(j)) »x(1).x(2) . . .x(j) admits an expansion
as an integral series about the origin:

(14) x(l).x(2). . .x(j) = amen)_____ «(new awhwmx(l)“‘“. . .x(j)°‘u’

where the “3(1)"...uu) are demerits ofK". We write, forj = 0, 1, 2, . . . ,

(15) 4;,(x) = “unmzmmfla¢<l>u~uw>fm+mww

where the right hand side is a convergent integral series in the variable
x E K". This series is obtained by suppressing in (14) the terms in which one of
the variables x(l), . . ., x(j) does not occur explicitly and then writing
1(1) =x(2) =--~=x(j) =x.
If t e K, all the t-th power mappings of C have the same expansion as an
integral series about the origin, since any two of them coincide on a neighbour-
hood of 0. We denote this expansion as an integral series by x‘“.
PROPOSITION 2. (i) 4;, E 0 mod degj.
(ii) If! e K, than
I
(16) a“ = Z 044*)
what theformal Pawn Ieriex an the right ir meaningfid buaure If (i).

300
 POWERS § 5.3

(We writs
(t) _ t(t — 1)...(t — i +,1)
i _ 1'!
far all t E K.)
Assertion (i) is obvious fi‘om the definition of the 49.
We prove (ii) for I an integer 20. By (14),

_ 1 #...+ I
(17) ”m — «(lwzwmm am).i...mu)"u“ "Q
For at = (1(1), . . ., «(0) e (N")’, let 6(a) denote the set ofje{l, 2, . . ., I}
such that «(1') aé 0. If, in the sum (17), we group together the terms for which
a(u) is the same, then

(18) gm = "Enh‘flm
where

(19) h,_n(x) = “1):!" “#1)”...03) fillinwltflfl)‘

Let a = {j,,j,, . . .,j,,}withj1 <j, < . . . < jg. In (14) (wherejis replaced by
t), we substitute 0 for x(k) for 1: ¢ 6; as 0 is the identity element ofG, we obtain
the expansion of AA) .x(j,) . . .x(j¢) as an integral series about the origin:

"(jil -"(j2)~ - "‘00 = 2C u

(1(a) am).....u(t)"(j1l“u‘)"(jzlw’)- - -"(jq)w')
and hence, by the definition clip“:

.).
(20) 4’4") = ,(;,a¢(i).....w> fU:)+---+w
By (19) and (20), we see that h,'°(x) = ‘lc-nl ,(x). Then (18) implim

,,.. = 2'; (3m = 2 (3...).

Z (i) q;,(::) for all 26 K. In the integral series a“ and
Then we write 3‘” - {‘0
zm', each coefficient is a polynomial function of I. For this is obvious for
x‘“’. As far as xm is concerned, it suflices to prove that, for all u E U(G), the
image of it under a: w» x") is a polynomial function of I. Now, for u e U"‘(G),
this image is t"‘u (§ 4, no. 3, Proposition 7 (iv)).
As all" = 3“" for I an integer 20, it then follows that 1“] = 2“" for all
t 5-: K.

301
Ill LIE GROUPS

Remarks. (1) We write condition (ii) of Proposition 2 {or t an integer >0:

0 = W")
X = 4100‘) + 4’10”)
1‘“ = 4m“) + 2411(1) + 4420:)

These formulae suflice to determine the 41p

(2) We see that 4400:) = 0, 4;,(x) = x, ¢3(x) = as!“ — 2x,

act-u = Z, (—woo.
(3) The above expremion for 4;, and formula (6) prove that
(21) ¢2(x) E B(x, 1) mod deg 3.
Using Proposition 2, (i) and (ii), we see that

(22) z‘“ E m + (2)13“, x) mod deg 3.

(4-) Let ¢,,,,,(:r) and h,,,,,(x) denote the homogeneous components of 4;,(x)
and xm ofdegree M. Then 4‘9.» = 0 for m < p. 011 the other hand, Proposition
2 (ii) gives

(23) Inner) = ZM—pwabm)

that is

(24) hm») = ,Z mime)

where the ‘i’nm are homogeneous polynomial mappings of degree In of K" into
K". In particular, by (23),

(25) me) = 2,, (":,’””¢,.m(x)

(26) mm = 5 some).
(5) If K is of characteristic >0, the results of nos. 1 and 2 remain true,
provided em in no. 2, is defined by (2“, 2: KM”) = A“. In no. 3, if the 44, are

defined as above, the argument again proves that 1‘“ = ‘21 (I) ¢.(x) if
= I
teN.

302
 EXPONBNTIAL § 5.4

4. EXPONENTIAL
Let E(x) be the expansion ofan uponential mapping of G as an integral series
about 0‘ Let L(x) be the expansion of the inverse mapping of an injective
exponential mapping of G as an integral series about 0. Since the tangent
mapping at 0 to any exponential mapping is the identity of L(G),
E(x) E x mod deg 2 and L(x) _=_ a: mod deg 2. Since E(L(x)) = L(E(x)) for x
sufficiently close to 0, the formal power series E and L are such that
E(L(X)) = L(E(X)) = X. An analogous argument shows that
EUX) = (1309)"), L(xm) = tL(X)
for t E K.
PROPOSmON 3.
__ -1
(27) L = §,( 2” 4a:

m l
(28) E = 21W“
(Recall that 41,. p is the homogmmu component (If 4;, qfdegm 1;.)
E(1") = (EGO)m
or, by (24-),
Em) = M; “Z“ Hemmer».
The two sides are formal power series in t and x. Equating the terms of first
degree in t, we obtain
(29) x = “2., 4mm».
Now the inversion of a system of formal power series, when it is possible, is
possible in only one way (Algebra, Chapter IV, § 6, Corollary Do Proposition 8).
Then
we = Z mot) by (29)
1100

= 2 flow) by (25)
Pull p

= Z ‘7)” m),
whence (i). Similarly, for 2 ye 0,
L0”) = 1L((Ix)“’1‘)
= IL(Z Z t'""<]>,_,,,(x)) by (24).
"we 10‘»:

303
in LIE GROUPS

Equating the terms of first degree in I, we obtain

a: = L(,?;:o (bu-"(6)
whence
E(x) 2 two)
m>o

Zofiivm.m(~) by (26).
= m;
PROPosmON 4. For Ihe chart of G mm! to b: omnital, it is nemrury and .mficimt
thattl;,=0forj2 2.
This is sufficient by Proposition 3. Suppose that the chart is canonical and
n
that 4,. = 0 for 2 g i < n. Then to: = gm = 20 (mix) = ux + Mg),
whence 41“ = 0. Hence 4;, = 0 for j 2 2 by induction on j.

§6. REAL AND COMPLEX, LIE GROUPS

In this paragraph, K is assumed to be equal to R or C.

l. PASSAGE FROM LIE ALGEBRA MORPHISMS TO LIE GROUP

MORPHISMS
Lemma 1. Let G be a simply [01:71:5d topological group, W a symmetric connected
open neighbourhood qfe, H a group andf a mapping of W3 into H such that
f(W1) =f(X)f(y)f(Z)
finny, z in W. Thmexima morphismf’ ofG into Hmch lhatf’ I W =f| W.
For (g, h) e G x H and U an open neighbourt of e in W, let A(g, h, U)
be the set of (gu,lf(u)) 6G x H for ueU. Then (g, h) eA(g, 11, U) and
A(g, )1, U1) n A(g, 11, U2) = A(g, 1:, U1 n U2). Let (s, t) eA(g, 1:, U); then
s = gu and t = ly’(u) for u E U; there exists an open neighbourhood U’ of e in
W such that uU' e U; then, for u' E U',

(374’, tf('l')) = (W3 UW'» EA(£, k) U)

1' Cf. Chapter XI of General Topology (to appear). It is proved in this chapter
that if G‘, G, are connected topological groups, 4; is an open continuous homo-
morphism of G. onto G, with discrete kernel and G4 is simply connected, then
4) is a homeomorphism. Recall on the other hand that a simply connected space is
connected.

304
 LIB ALGEBRA MORPHISMS T0 LIE GROUP MORPHISMS § 6.1

and hence A(.t, l, U') C A(g, h, U). It then follows that the A(g, h, U) form
the base of a topology on G x H. We shall denote by Y the set G X H with
this topology and denote by [z the canonical projection of Y onto G, which is
open. The restriction ofp to A(g, h, U) is a homeomorphism ofA(g, h, U) onto
gU. Hence (Y, p) is a covering space of G. Let Yo be the subgroup of Y gene-
rated by A(e, e, W) and let 93 be the set of A(e, e, U). Clearly % satisfies con-
ditions (GV{) and (GVfi) of General Topology, Chapter III, § 1, no. 2. The set
Y5 ofy 5 Yo such that the mappings z ny' 1 and z Hy‘1zy o0 into Yo are
continuous at (e, e) is a subgroup OfYa- Let w E W. The mapping w' n—> ww’w"
of W into G is continuous and hence the mapping
("13f(w')) H (ww'w'fif(ll/“1'10"”
ofA (e, e, W) into Y is continuous at (e, e) . Nowf(ww'w ' l) = f(w)f(w')f(w ' 1)
and therefore (ww'w",f(ww'w‘1)) = (w,f(w))(w’,f(W’))(w,f(w))“-
As w“ EW, we see that (w,f(w)) 6Y3. Thus, A(e, e, W) C Y3, so that
Y", = Yo. The group Yo, with the filter base 3 therefore satisfies condition
(GVl’n) of General Topology, Chapter III, § 1, no. 2. As
(g, h) ~A(€, 6, U) = Mg, 1!, U),
Yo is a topological group which is connected since A(e, e, W) is connected.
Then MYo) is an open subgroup of G, whence 17(Yo) = G since G is connected.
The kernel of[z I Yo is discrete. As G is simply connected, [1 | Yo is a homeo-
morphism of Y0 onto G. Therefore Yo is the graph of a morphismf’ of G into
H- For g E W, (g,f(g)) 6 AG: 6 W) C Yo, whenceflg) =f'(s’)-
THEOREM 1. Let G and H be Lie groups and h a continuous morphism of L(G) into
L(H). Suppose that G is simply connected. Then there exists one and only one Lie group
morphism (I) of G into H such that It = L(<]>).
The existence of 4) follows from Lemma 1 and §4, not 1, Theorem 1 (i).
The uniqueness of <1) follows from § 4-, no. 1, Theorem 1 (ii) and the fact that G
is connected.
COROLLARY. Let G be a finite-dimensional simply connected Lie group. There exists
afinite-dimemional analytic linear representation of G whose kernel is discrete.
There exist (Chapter I, § 7, Theorem 2) a finite-dimensional vector space E
and an injective morphism h ofL(G) into the Lie algebra End(E). By Theorem
1, there exists a morphism (I: ofG into GL(E) such that L(¢) = h. Hence cl: is an
immersion and therefore its kernel is discrete.
Remarks. (1) There exist finite-dimensional simply connected Lie groups
which have no finite-dimensional injective analytic linear representation
(Exercise 2).
(2) There exist finite-dimensional connected Lie groups G such that every
finite-dimensional analytic linear representation of G has non-discrete kernel
(Exercises 3 and 4).

305
Ill LIE GROUPS

2. WTEGRAL SUBGROUPS
DEFINITION 1. Let G b: a Lie group. An integral subgroup ofG is a subgroup H with a
connected Lie group structure such that the canonical injection if H mm G u an
immersion.
A one-parameter subgroup of G is a l-dimensional integral subgroup of

'Let H be an integral subgroup ofG and i the canonical injection of H into G.

Then L(i) defines an isomorphism of L(H) onto a Lie subalgebra of L(G)
admitting a topological supplement. L(H) is identified with its image under
L(i).
Examples. (1) A connected Lie subgroup of G is an integral subgroup of
G
(2) Suppose that G is finite-dimensional. Let H be a subgroup of G; we give
it the structure induced by the Lie group structure on G (§ 4, no. 5, Definition
3). Then its identity component H0 is an integral subgroup of G and the sub-
algebra tangent to H at e is L(Hu) (§ 4-, no. 5, Proposition 9 (ii)).
(3) Let G be a complex Lie group, H an integral subgroup of G and G1
(resp. H,) the underlying real Lie group of G (resp. H). Then H1 is an integral
subgroup of G1 and L(H,) is the underlying real "Lie algebra of L(H).
THEOREM 2. Let G be a Lie group.
(i) The mapping H >—> L(H) it a bijech'lm of the m of integral subgroup: of G 1min
thuetqz': L ’g ‘ q(G) ‘ ' 'ua , 'D' ' ”'
(ii) Let H be an integral subgmup if G. Every connnted Lie subgrnup germ If G
with Lie algebra L(H) i: an open mbmnnifizld «J H which generates H.
(a) Let I) be a Lie subalgebra of L(G) admitting a topological supplement.
Let H1 be a Lie subgroup germ of G such that L(H,) = I) (§ 4, Theorem 3),
H1 can be chosen so that it is connected. Let H be the subgroup of G generated
by H,. There exists (§ 1, Corollary to Proposition 22) a Lie group structure
on H such that H‘ is an open submanifold of H and the canonical injection of
H into G is an immersion. As H1 is connected, H is connected and is hence an
integral subgroup of G. Then L(H) = L(H‘) = b. This proves that the map-
ping considered in (i) is surjective.
(b) Let H be an integral subgroup of G and N1 a connected Lie subgroup
gem of G with Lie algebra L(H). As the canonical injection of H into G is an
immersion, there exists an open subgroup germ Hl of H which is at the same
time a submanifold of G and hence a Lie subgroup germ of G with Lie
algebra L(H). On the other hand, let N be the subgroup of G generated by
N,; by part (a) of the proof, it has the structure of an integral subgroup of G
such that N‘ is an open submanifold of N. By §4, Theorem 3, H1 fl N1 is
open in H1 and N1. Hence the subgroup of G generated by Hi n N1 is equal
on the one hand to H and on the other to N. Therefore the Lie groups H and N

306
 mam suBGRovJPs § 6.2

are equal. This proves (ii) and also proves that the mapping considered in (i)
is injective.
Remark 1. Let H be an integral subgroup of G. Let Y be the lelt foliation of G
associated with L(H). If g e G, let gH be given the manifold structure derived
from that on H by y(g). By Difierentiable and Analytic Manifolit, R, 9.3.2, the
canonical injection ofgH into Y is a morphism. This morphism is étale. Hence
the maximal connected leaves of Y are the left oosets modulo H.
PROPOSITION 1. Let G and M be Lie groups, H an integral :ubgrnup of G and 4) a
morphism of M inla G web that L(<]>) (L(M)) C L(H). Suppose that M it connected.
Then ¢(M) C H and (5, considered a: a mapping qfM into H, i: a Lie grnup murphism.
In the notation of Remark 1, d) is a. morphism of M into Y (Difierentiabls and
Analytic Manifiilds, R, 9.3.2) and hence (HM) C H since M is connected.
COROLLARY 1. Let G and H be Lie groups, 4) a Lie group morphism ifG into H, N the
kernel of d) and h = L(<])). Suppose that G is connected and that H isfinite-dimenrianal.
(i) N is a Lie subgroup q and L(N) = Ker h.
(ii) Let H’ be the integral rubgruup 13f H with Lie algebra Im h. Then (MG) = H’.
(iii) The mapping (f GIN inta H’ derivedfrom 4; when passing to the quotient i: a
Lie gran]: itamarphixm.
(i) has already been proved (§ 3, no. 8, Proposition 28).
Let 4; be the Lie group morphism of G/N into H derived from :1) when
passing to the quotient; it is an immersion (§ 3, no. 8, Proposition 28). By
Proposition 1, u]; is a Lie group morphism of G/N into H’. This morphism is
étale and hence ¢(G/N) = H' since H’ is connected; this proves (ii). Then
4;: G/N —> H' is bijective and is a Lie group isomorphism, which proves
(iii).
COROLLARY 2. Let G be a Lie group and H1 and H2 integral subgroup: g" G. If
L(H2) C L(H1), then H, is an integral subgroup al.
Let i1: Hl —> G, i2: H2 —> G be the canonical injections. Then
L(i2)(L(Ha)) = L(H2) C L(H1)-
By Proposition 1, i2 is an analytic mapping ofH2 into H1 and even an immer-
sion of H2 into H1 since L62) is an isomorphism ofL(H2) onto a subalgebra of
L(H1) admitting a topological supplement.
COROLLARY 3. Let G be a finite-dimem‘imal Lie group and (HI)IE r a family 9/" Lie
subgroup; q. Then H = .01 H, i: a Lie subgroup ofG and
fl L(H.).
L(H) = tel
There exists a finite subset J of I such that ‘0‘, L(H‘) is equal to the inter-

307
III LIE GROUPS

section M of all the L(H,). We know that H“ = ‘0] H‘ is a Lie subgroup such
that L(H“) = M (§ 3, no. 8, Corollary 2 to Proposition 29). Let HD be the
identity component of H*. It is a Lie subgroup of G and L(Hu) = M. By
Corollary 2, Ho C H. for all i and hence Ho C H C H*, whence the corollary.
COROLLARY 4-. Let G be afinite-dimenrional connected Lie group. Thefullnwing con-
ditian: are equivalent:
(i) G i: unirncdular (Integration, Chapter VII, § I, no. 3, Definition 3);
(ii) detAdg = lfarallgeG;
(iii) Trad a = Ofnr all as L(G).
The mapping g »—> det Ad g is a morphism cl: of G into K“. By § 3, Proposi-
tions 35 (no. 10) and 44 (no. 12), L(¢)a = Tr add for all aeL(G). Clearly
Im L((b) = (0} or K. In the first (resp. second) case, Im q5 = {l} (resp.
In) 42 = K*) by Corollary I and hence G is unimodular (resp. not unimodular)
by § 3, no. 16, Corollary to Proposition 55.
PROPOSITION 2. Let G be afinite-dimenxianal Lie group and H an integral .mbgroup 13/
G. Thefollowing conditions are equivalent:
(i) H ix closed;
(ii) the tapology on H L: induced by that on G;
(iii) H i: a Lie subgroup of G.
(i) => (iii): this follows from § 1, Propositions 2 (iv) (no. 1) and 14 (iii)
(no. 7).
(iii) 3 (ii): obvious.
(ii) => (i): if the topology on H is induced by that on G, H is closed because
H is complete (§ 1, no. 1, Proposition 1).
PROPOSITION 3. Let G be a Lie group, H an integral subgroup of G, M a nan-empty
connected analytic mam'fiild,fa mapping qfM into C and r E NK. Consider thefillnwing
conditions:
(i) fix afclaxs C' andf(M) C H;
(ii) f(M) C H and]; considered a: a mapping ofM into H, is qfclaxs C';
(111) f u of class 0’, f(M) meek H and the image of TM(M) is contained in
f(m) .L(H)for all m E M.
Then (ii) ¢> (iii) 2 (i). If the tapolagy an H admit: a countable bare, the three can-
dih'anr are equivalent.
(ii) 2 (i) and (ii) 3 (iii): obvious.
(iii) :> (ii): suppose that condition (iii) holds. By Deflemm'able and Analytic
Manifolds, R, 9.2.8,f15 a morphism of class C' of M into the left foliation asso-
ciated with L(H). As M is connected,f(M) C H.
Ifthe topology on H admits a countable base, condition (i) implies thatfis a
mapplng of class C' of M into H (Diflbentiable and Analytic Manifoldr, R,
9.2.8); hence (i) :> (ii).

308
 INTEGRAL summours § 6.2

COROLLARY 1. Let G be afinite—dimensional Lie group and H an integral subgroup of

G. Then the Lie subalgebra tangent to H at e (§ 4-, no. 5, Definitions 2 and 3) is
L(H) and the Lie group structure on H is the structure induced by that on G.
As H is connected and finite—dimensional, its topology admits a countable
base.
COROLLARY 2. Let G be a Lie group and H1 and H2 integral subgroups q, Suppose
that the topology on Hl admits a countable base. Then
Hz c Hl <9 L(H,) c L(H,)
and, if these conditions hold, H2 is an integral subgroup of H1.
Thc last assertion and the implication L(H,) C L(H,) 2.. H, C Hl follow
from Corollary 2 to Proposition l. The converse implication follows from
Proposition 3.
COROLLARY 3. Let G be a Lie group and H1 and H2 integral subgroups of G whose
topology admits a countable base. If H1 and H3 have the same underlying set, the Lie
group structures on H1 and H, are the same.
This follows from Corollary 2.
Remark 2. Let G be a finite-dimensional Lie group. Let H be a subgroup of G.
We shall say, by an abuse of language, that H is an integral subgroup of G if
there exists a Lie group structure S on H such that H, together with S, is an
integral subgroup of G. By Corollary 3 to Proposition 3, if S exists, S is unique.
Remark 3. Let V be a manifold of class (3'. Let M be a subset of V and x and y
elements of M. Consider the following property:
PMvz'y: there exist I, x0, 2:], l . ., xmfi, . . .,f;, such that: (a) I is a connected
open subset ofK; (b) so, . . .,xn are in M, no = z, x" :3]; (c) for! S i < n,
fl is a mapping of class C' of I into V which takes the values x44 and x‘ and
fi(I) C M.
We shall say that M is a C'eonnected subset of V if, for all elements at, y of M,
the property PM? 2:. , holds.
PROPOSITION 4. Let G be afinite-dimensional Lie group and H a subgroup if G. Let
r E Nx. Thefilluwing conditions are equivalent:
(i) H is an integral subgroup (JG;
(ii) with the Lie group structure induced by that on G, H is connected;
(iii) H is C'—connected.
(ii) => (i): obvious.
(i) 2 (iii): suppose that H has a Lie group structure such that H is an in-
tegral subgroup of G. Using the notation ofRemark 3, the set ofy e H such that
property P3,“, holds is an Open subgroup of H. As H is connected, this sub-
group is equal to H and hence condition (iii) is satisfied.

309
 I" LIE GROUPS

(iii) 2 (ii): suppose that condition (iii) is satisfied and let H be given the
structure induced by the Lie group structure on G. Let I) be the subalgebra
tangent to H at e. The identity component Hu of H is an integral subgroup of
G such that L(Ho) = I). We show that H = Ho. It suffices to prove the follow-
ing: let I be a connected open subset of K, f a mapping of class C’ of I into C
such thatf(I) C H and A and [1. two points 0“; ifffl) 9 Ho, thenf(p) EH0.
But, for all v e I, (v)(K) Cf(v)l} by definition of I), so that our assertion
follows from Proposition 3.
Remark 4-. If K = R, the integral subgroups of G can also be characterized as
the subgroups which, with the topology induced by that on G, are arcwise eon-
neoted (§ 8, Exercise 4). However, subgroups may be conneoted but not integral
(Commutative Algebra, Chapter VI, §9, Exercise 2).
COROLLARY. Let G be a [mite-dimensional Lie group and H1 and H, two integral
subgroups ofG. The subgroup of G generated by H1 and H2 and the subgroup (H1, H2)
of G are integral subgroups ofG.
The subgroup (G, G) of G is not always closed (§ 9, Exercise 6).
Recall (§ 3, no. 11, Corollary 5 to Proposition 41) that if a is a finite—dimen-
sional algebra, Aut(a) is a Lie subalgebra of GI.(a) and that L(Aut(n)) is the
Lie algebra of derivations of a.
DEFINITION 2. Let a be aflnite-dimemionol Lie algebra. Let Ad(a) or Int(a) denote
the integral subgroup ofAut(a) with Lie algebra ad(n). The elements of this group are
called inner automorphism of a.
By transport ofstructurc, ad(a) is invariant under Aut(a) and hence Int(a) is
normal in Aut(a). By § 4, no. 4, Corollary 1 to Proposition 8 and the fact that
Int(a) is connected, the elements of Int(a) are the finite products of auto-
morphisms of the form exp adx where tee a. In general, Int(a) is not a. Lie
subgroup ofAut(a) (Exercise 14).

3. PASSAGE FROM LIE ALGHRAS TO LIE GROUPS

THEOREM 3. (i) If L is afinite—dimensional Lie algebra, there exists a simply ton-
neoted Lie group G such that L(G) is isomorphic to L.
(ii) Let G1 and G2 be two connected Lie groups, with G1 simply connetted. Letf be
an isomorphism qf'L(G,) onto L(G,), (I; the morphism ofG, into G2 such that L(o) = f
and N the kernel of (I). Then N is a diserete subgroup of the centre of G, and the mor-
phism 'ZfG1/N into G2 derivedfrom 4’ is a Lie group isomorphism. If G2 is simply
eonneeted, <1) is an isomorphism.
Let L be a finite-dimensional Lie algebra. There exists a finite-dimensional
vector space E such that L can be identified with a Lie subalgebra. of End(E)
(Chapter I, §7, Theorem 2). Let H be the integral subgroup of GL(E) with

3l0
 EXPONENTIAL MAPPING § 6.4

Lie algebra L. Let H be its universal covering (§ 1, no. 9, Remark). Then

L( ) is isomorphic to L, whence (i). '
Let G1, G,,f, (b, N be as in (ii). Then ‘l’ is étale, hence (MGI) is an open sub-
group ofG2 and hence 4>(G1) = G,. On the other hand, N is discrete and hence
contained in the centre of G1 (Integration, Chapter VII,§ 3, Lemma 4'). Clearly
the morphism of G1 /N onto G2 derived from 4) is a Lie group isomorphism. If
G2 is simply connected, every étale mapping of G1 onto G2 is injective and
hence N = {e}.
PROPOSITION 5. Let G be u eonneeted real Lie group. Suppose that L(G) has a eon;-
plex normuble Lie algebra structure L' compatible with its real normable Lie algebra
strueture. There exists on G one and only one complex Lie group structure compatible with
the real Lie group structure and with Lie algebra L'.
By § 4, no. 2, Corollary 2 to Theorem 2, it suffices to prove that the structure
of L’ is invariant under Ad G. Let ()2 be an exponential mapping of C. By § 4,
no. 4, Corollary 3 (i) to Proposition 8, there exists a neighbourhood V of 0 in
L(G) such that the structure of L’ is invariant under Ad ¢(V). But Ad ¢(V)
generates G because G is connected.
The conclusion of Proposition 5 is not necessarily true if G is not assumed
to be connected (Exercise 7).

PROPOSITION 6. Let G be a connected complex Lie group. If G is rampart, G is com-

mutative.
The holomorphie mapping g n—> Ad g ofG into 3’(L(G)) is constant (Diflren—
tiable and Analytic Many'oldx, R, 3.3.7) and hence ad a = 0 for all a eL(G)
(§ 3, no. 12, Proposition 44). Hence G is cornmutativev(§ 4, Corollary 3 to
Theorem 1).

4. EXPONENT’AL MAPPING
THEOREM 4. Let G be a Lie group. There exist: one and only one exponential mapping
(JG defined an L(G).
There exist a convex open neighbourhood U ofO in L(G) and an exponential
mapping ofG defined on U. It can be assumed, by choosing U sufficiently small,
that
(1) NO + 75)“) = MMMOJ“)
for a eL(G), A, 7" in K, M, Na, (A + flu in U.
Let a e L(G). There exists an integer n > 0 such that; a e U. Ifm is another

integer such that id 6 U, then #5 a E U and relation (1) implies

«we»: «was»: 311

I" LIE GROUPS

hence (49611))" = (4,61 a))m. There exists an extension 4,; L(G) —> G ofd)

such that Ma) = ($61))“ for aeL(G) and n an integer >0 such that

; a E U. Clearly q; is analytic and is an exponential mapping of G. If

AI": L(G) —> G is an exponential mapping of G, 4: and 4" coincide on a neigh- _
bourhood of 0 and hence are equal since L(G) is connected.
Henceforth when we speak of the exponential mapping q, we shall mean the
mapping considered in Theorem 4. It will be denoted by expo or exp if there
is no risk of confusion.
Example. Let A be a complete normed unital associative algebra. Then ep.
is the exponential mapping defined in Chapter II, § 7, no. 3.
PROPOSITION 7. Let G be a Lie group and a an element qf L(G). The mapping
1 ~—> expora) qf K into G 1': the unique morphirm (l) Qflhe Lie group K into G such
that (T°¢)l = a.
The mappings (A, )i') >—>exp()‘a) expOi’a) and (k, N) I—> expo + A’)n of
K x K into G are analytic and coincide on a neighbourhood of (0, 0). As
K x K is connected, these mappings are equal. Hence (I): A >—> epa) is a
Lie group morphism of K into G. The tangent mapping at 0 to k >—> M is the
mapping A >—> M; and T,(cxp) = Idum; hence (To¢)l = a. The uniqueness
assertion of the proposition follows from Theorem I.
PROPOSITION 8. Let G be a Lie grnup. For all x, y in L(G) and n an integer,
(exP £10)»
(2) exP(x + y) = “hm” ((exp 5 x)

(()<)<)<))
By Proposition 7, this follows from Proposition 4 of § 4, no. 3, taking 1 = l-
n
PROPOSITION 9. Let G be a temple): Lie group and G' the underlying real Lie gimp.
Then expo = expuu
This follows from Proposition 5 of § 4, no. 3 and the analyticity ofexpa and
expuy.
PROPOSITION 10. Let G and H be Lie granp: and d; a morphism of G into H.
(9 it o em} ='eXPH . L(4>)-
(n) [f G I: an Integral subgroup of H, then exp} = expH | L(G).

312
 EXPONENTIAL MAPPING § 6.4-

The two sides of the equality (i) are analytic mappings of L(G) into H
which coincide in a neighbourt of 0 (§ 4, Proposition 8, no. 4) and hence
are equal. Assertion (ii) is a special case of (i).
COROLLARY 1. Let G be a Lie group, G’ a Lie .mbgmup of G and a e L(G). The
following condition: are equivalent:
(i) a e L(G’) ;
(ii) exp(M) E G’for all 16 K and IA] Wendy mall;
(iii) expoia) e Gfot all k E K.
The argument is as in § 4, no. 4, Corollary 1 to Proposition 8.
COROLLARY 2. Let G be a Lie group, H an integral subgroup of G and a E L(G).
Conflder thefollowing conditions:
(i) a e um;
(ii) expG(M) e Hfor all A e K.
Then (i) = (ii). If the topology ofH admit: a countable base, then (i) 4: (ii).
Let i be the canonical injection of H into G. If a E L(H), then
fiPnW) = (CXPG ° L(0)011) = (i ° exPH)(M) GH‘
Hence (i) = (ii). The converse when H has a countable base follows from
Proposition 3.

COROLLARY 3. Let G be a Lie group, p an analytic linear representation ofG, x e L(G)

and g e G.
(i) 9(CXP er) = exP L(PV;
(ii) Ad(exp z)=
(iii) £(cxp fig =OXFXNAdg. Jr).
The argument is as in § 4, no. 4-, Corollaries 2 and 3 to Proposition 8.
COROLLARY 4-. Let G be afinzte-dzmenxzonat connected Lie group.
(i) Int(L(G)) = Ad(G).
(ii) Let Z be the centre of G. Then Z is a Lie subgroup ofG whose Lie algebra is the
centre of L(G). The mapping of G/Z into Int L(G) derived from g~—>Ad g when
passing to the quotient i: a Lie group ironwrphirtn.
Assertion (i) follows from Corollary 3 (ii) and the remarks after Definition 2.
Let g E G. In order that Ad g = Idum, it is necessary and suflicient that Int g
coincide with Ido on a. neighbourhood of e (§ 4, no. 1, Theorem 1 (ii)) and
hence on the whole of G; in other words, it is necessary and sufficient that
g e Z. Then (ii) follows from Corollary 1 to Proposition 1.
DEmNmON 3. Let G be a finite—dimensional connected Lie group. The Lie group
Int(L(G)) = Ad(G) is called the odioiut group of G.
PROPOSITION 11. Let G be a connected commutative Lie group.
(i) exp i: an e'tale morphirm of the additive Lie group L(G) onto G.

313
"I LIE GROUPS

(ii) If K = R and G ixfinite—dimensional, G if iromovphic to a Lie group of the

form R” x T4 (p, q integer: 20).
By the Hausdorff formula, (exp x) (exp y) = exp(x + y) for x, y sufficiently
dose to 0 and hence for all x, y in L(G) by analytic continuation. Hence exp is
a group homomorphism and is étale since
T.(exp) = Iduo)‘
Hence (i). Assem'on (ii) follows from (i) and General Topology, Chapter VII,-
§ 1, Proposition 9.
PROPOSITION 12. Let G be a Lie group and L = L(G). For all x e L, let T,,(L) be
identified with L, In that the right diferenh'al m(x) of exp at x i: a linear mapping of L
into L. For all x E L,

w Z4
(x) —— 700 (n + l)! (ad x)" I

The two sides are analytic functions of: and are equal for x sufficiently close
to 0 (§ 4, no. 3, Proposition 5).
Remork. 117(1) . (ad 2:) = exp ad x - 1. We write, by an abuse of notation,
_ exp ad x — 1.
a“) — ad x
COROLLARY. Let G be a complex Lie group and x e L(G). The tangent mapping at x
to expfi has kernel neZ—(D)
® Ker(ad x — 2i1rn).
. . l e’ — l .
The Integral function z~—> 20 m 1", equal to z for z 95 0, admits

as zeros the points of 21112 — (0}, which are all simple zeros. The corollary then
follows from Proposition 12 and the following lemma:
Lemma 2. Let E be a complex Banach space, u an element of .‘2’(E), S the xpectrum ofu
in g (E) (Spectral Theories, Chapter I, § 1, no. 2) and f a holomorphie complex
function on on open neighbourhood Q ofS. Suppose tho!fadmit: in (I only afinite number
of dixtinctrzemx 1,, . . ., z," of multiplicities h“ . . ., h". Then Kerf(u) i: the direct
rum :fthe Ker(u — z,)"lforl S i g n.
(For the definition off(u), see Spectral Theories, Chapter I, §4, no. 8.)
There exists a holomorphie function g on Q, everywhere non-zero, such that
M = (z — zoo. . .(z 2 2mm. mm gong-1(a) = g-'(u>g<u) = 1 and
hence Kerf(u) = Ker ‘1]: (u — 1,)”. Considering Kerf(u) as a C[X]—
module by means of the exnernal law (h, x) n—> h(u)x for h e C[X], x e Kerf(u),

314
 APPLICATION TO LINEAR REPRESENTATIONS §6.5

we see that Kerf(a) Is the direct sum of the Ker(u — z,)"t, using Algebra,
Chapter VII, §2, no. 1, Proposition].

5. APPLICATION TO LINEAR REPRESENTATIONS

PROPOSITION 13. Let G be a connected Lie group and 9 an analytic linear representation
of G on a complete normable space E. Let E], E, be two closed vector subspaces of E
such that E2 C E. Thefollowing conditions are equivalent:
(i) p(g)x E x (mod E2)far ollgE G andall e,;
(ii) L(p)(L(G)) maps El into Ea.

p(g)x2x (modEa) {orallgeGandallel

o p(exp a): a 1 (mod E,) for all a e L(G) and all x e E1
e (exp L(p)a)x E a: (mod E2) for all a e L(G) and all x e E.

On the other hand, if u e 3’ (E), then

expOax): E x (mod E2) {or all A e K and all x e E1

¢u(E1) C E:

whence the proposition.

COROLLARY 1. For E1 to be stable under 9, it is necessary and Wm that El be

stable under L(p).
It suffices to take E1 = E2 in Proposition 13.

COROLLARY 2. Suppose that p isfinite-dimensianal. For 9 to be simple (resp. semi-

simple), it is necessary and W711 that L(p) be simple (resp. semi-simple).
This follows from Corollary 1.

COROLLARY 3. Let e. For x to be invariant under 9(0), it is necessary and

suficient that x be annihilated by L(p) (L(G)) (that is that x be invariant under L(p)
in the sense of Chapter I, § 3, Definition 3).
It suffices to take E, = Kx and E2 = 0 in Proposition 13.

COROLLARY 4. Let p’ be another analytic linear representation of G on a complete

non/table space E’. Let T E 2’ (E, E'). Thefillowing conditions are equivalent:
(0 TM) = P’(g)vr allgeG;
(ii) TL<p><n> = L(p'xoar all ”L(G).
Let a be the linear representation of G on f(E, E') derived from p and 9’
(§ 3, no. 11, Corollary 1 to Proposition 41). Condition (i) mains that T is
invariant under a(G) . Condition (ii) means that T is annihilated by L(G) (L(G) ) .
It then suffices to apply Corollary 3.

315
111 LEE GROUPS
p’ to be
COROLLARY 5. Suppose that p and p’ are finite-dimensionali [for p and
equivalent, it is necmary and .ruflicient that L(p) and L(p’) be equwalenl.
This is a special case of Corollary 4».

COROLLARY 6. Suppose that G 1':finite-dinumional. Let t e U(G). For L, (resp. R,)

to be right (resp. lefl) invariant, it is necessary and mficient that t belong to the centre of
U G .
(Fol L, (resp. R,) to be right (resp. left) invariant, it is necessary and suilicient
that e, * t = ta: e, for all g5 G, that is that (Int g)*t = I. There emsts an,
integer n such that t E U"(G). By Corollary 3 and Proposition 45 of § 3, no. l2,
(Intg)*t = tfor all g6 G ifand only if [a, t] = 0 for all a EL(G), that is if
and only if t commutes with U(G).

6. NORMAL INTEGRAL SUBGROUPS

Lemma 3. Let G be a Lie group, HI and H, integral subgroup: whore topology admits
a countable base and g e G. Then

gHig“ = H2 0 (Ad g)(L(Hi)) = L(Ha)~

Adg = T,(Int g). Hence, by transport of structure, (Int g) (H1) has Lie
algebra (Ad g) (L(H,)). As H1 and H2 have countable bases, to say that the
sets H2 and (Int g) (H1) are equal amounts to saying that the integral subgroups
Hi and (Int g)(H1) are equal (no. 2, Corollary 3 to Proposition 3). Then the
lemma follows from Theorem 2 (i).
PROPOSITION 14. Let G be a Lie group and H on Integral subgroup whore topology
admitsa “ base. The " '9 r" are , ' ' :
(i) H it normal in G;
(ii) L(H) 1': invariant under Ad(G).
Iffurther G is connected, these conditions are equivalent to thefillowing:
(iii) L(H) is an ideal q(G).
Iffurther G is simply connected and L(H) is Qffinite codimension in L(G), these con-
dition imply that H is a Lie subgroup of G and that G/H is simply connected.
The equivalence (i) ¢> (ii) follows from Lemma 3. If further G is con-
nected, condition (ii) is equivalent to saying that L(H) is stable under ad L(G)
(no. 5, Corollary 1 to Proposition 13 and § 3, no. 12, Proposition 44).
Suppose that G is simply connected and that L(H) is an ideal of finite
codimension in L(G). By Theorem 3 of no, 3, there exists a simply connected
Lie group G' such that L(G’) is isomorphic to L(G)/L(H). There exists a con-
tinuous morphismfof L(G) onto L(G’) with kernel L(H). By Theorem 1 of no.
1, there exists a morphism (I) ofG into 0’ such that L((l)) = f. This morphism is
a submersion and hence its kernel N is a Lie subgroup of G such that
L(N) = Kerf = L(H). Hence H is the identity component of N and is there-

316
 NORMAL 1mm suncnoues § 6.6

fore a Lie subgroup ofG. Let (I: be the morphism ofG/H into G’ derived from 4)
when passing to the quotient. This morphism is étale; since G is simply con<
nected, 4; is an isomorphism of G/H onto G'.
COROLLARY 1. Let G be afinite—dimemionol simply connected Lie group. Let m, I) be
Lie subalgebrax of L(G) such that L(G) is the semi-direct product of m by I). Let M, H
be the corresponding integral mbalgebrax of G. Then M and H are simply connected Lie
subgroups g"G and, as a Lie group, G i; the semi-direct product of M by H.
By Proposition 14-, H is a normal Lie subgroup of G and the Lie group G/H
is simply connected. Let 1: be the canonical morphism of G onto G/H. There
exists a morphism 6 ofG/H into M such that L(O) is the canonical isomorphism
ofL(G)/L(H) = L(G)/b onto L(M) = m. Then
L(u o 6) = L(n) o L(O) = Iduam,
and hence it o 0 = Ida/H. By no. 1, Corollary 1 to Proposition 1, 0(G/H) = M.
By Proposition 8 of § 1, no. 4, M is a Lie subgroup of G and the Lie group G is
the semi-direct product of M by H.
COROLLARY 2. Let G be a finite—dimensional simply connected Lie group, H a normal
connected Lie subgroup of G and 1c the canonical morphism of G onto G/H.
(i) There exists an analytic mapping 9 of G/H into G such that 1: a p = Ida/3.
(ii) For every mapping p with the properties of (i), the mapping (h, m) »—> hp(m) of
H x (G/H) into G is an isomorphism of analytic manzfoldx.
(iii) H and G/H are simply connected.
- Let n = dim G — dim H. The corollary is obvious for n = 0. We argue by
induction on n.
Suppose that there exists an ideal of L(G) containing L(H) distinct from
L(G) and L(H). Let H’ be the corresponding connected Lie subgroup of G.
Let 1:1: G —> G/H’ and 1:2: H’ —> H’IH be the canonical morphisms. By the
induction hypothesis, there exist analytic mappings pl: G/H’ —> G,
p]: H’/H —> H’ such that 1:, 9 (a1 = Idem], 1:, 0 p2 = Idflymi Let
1:5: G/H —> G/H'
be the canonical morphism. If x e G/H and y is a representative of x in C, y
and 91(n3(x)) have the same canonical image modulo H’ and hence
x'1n(pl(1ra(x))) e H'IH. Let
Pi“) = P;(Wa(x))92("(91("a(x)))"’0 5G-
Clearly p is an analytic mapping of G/H into C and
"(900) = "(tumor)”W2(92(W(Pi("a(x)))'50)
= "(Pi(1ra(*)))1r(91("s(x)))"I = x~
If new the only ideals of L(G) containing L(H) are L(G) and L(H), the Lie

317
 I" LIE GROUPS

algebra L(G)/L(H) is either l-dimensional or simple. In both cases, L(G) is

the semi-direct product of a subalgebra by L(H); this is obvious in the first
Case and in the second this follows from Chapter I, § 6, Corollary 3 to Theorem
5. Assertion (i) then follows from Corollary 1.
Assertion (ii) is obvious. Assertion (iii) follows from (i) and (ii).
The conclusions of Corollary 2 are no longer necessarily true when G
is infinite-dimensional or when H is not normal (Exercises 8 and 15).

COROLLARY 3. Let G be afinite-dimarional connected Lie group and H a normal eon- ‘

netted Lia mbgmup g” G. The canonical morphism if 1:1(H) into «1(G) it injective.
Let G1 be the universal covering of G and )t the canonical mapping of G1
onto G. The Lie algebra of G1 is identified with L(G). The Lie subgroup
Fla-l) of G1 is normal in G1 and its Lie algebra is L(H). Let H, be the iden-
tity component of A"(H) and A, = A | HI. Then L(H1) = L(H) and hence
k1 is an étale morphism of H1 onto H. On the other hand, H1 is simply con-
nected (Corollary 2) and hence is identified with the universal covering of
H. Then, by General Topolagy, Chapter XI, the canonical morphism of 1:1(H)
into «1(G) is identified with the canonical injection of Ker k1 into Ker A.

7. PRIMITIVES OF DIFFERENTIAL FORMS WT-I‘I'I VALUES IN A LIE

ALGEBRA
PROPOSITION 15. Let G be a Lie group, M a manifold of class C’ (r z 2) and a: a
dtferential form of clan C1"1 and degree 1 on M with value: in L(G), well that
do: + [at]’ = 0. Suppose that M is simply connected. For all x E M and all x E G,
there exixt: one and only one mappingf of class C"1 tf M into C rack thatf(x) = .r
andf '1.df = a.
The uniqueness off follows from §3, no. 17, Corollary 2 to Proposition 59
and the fact that M is connected. We prove the existence off There exist an
open covering (U,),E 1 of M and, for all is I, a mapping g.: U( —> G of class
0’” such that g,"‘.dg = at on U‘ (§ 4, no. 6, Theorem 5). By §3, no. 17,
Corollary 2 to Proposition 59, gg,“ is locally constant on U‘ AU, Let
g,g,' 1 = g”. Let G, be the group G with the discrete topology. The
g”: Ut n U,—> G, are continuous and gflgfl, 2 gm on U. n U, n Uk. Since
M is simply connected, there exist continuous mappings M: Ul —> G,, such that
g‘g,‘ 1 : Mk] ‘ on U, n U,. Let g be the mapping of M into G whose restriction
to U‘ is A" 1gI for all i E I. This mapping is of class C"1 and g“dg = 0:. The
mapping f of M into G defined by f = 3(g(x))‘1g satisfies the conditions
f‘kdf: atandf(x) = s.

B. PASSAGE FROM LAWS OF INFmITESIMAL OPERATION TO LAWS OF

OPERATION
Lemma 4-. Let G be a connected topological group, X a Hamdozfl" topological rpm: and
f1,f2 laws oflefi (resp. right) operation ofG on X meh that,for all x e X, the mapping:

318
 LAWS or INFINITESIMAI. OPERATION TO LAWS or opens-non §6.8

.n—>f1(:, x), :»—>f3(:,x) of G into X are continuous. Suppose that there exist: a
neighbourhood V of {e} X X in G x X Such that f1 andf, coincide on V. Then
f1 = f:-
Let x e X and A be the set ofg 6G such Ihatf,(g, x) =f,(g, x). Then A is
closed in G. On the other hand, let geA; we write y =f1(g, x) =f3(g, x).
There exists a neighbourhood U of e in G such thatf1 (t, y) = f,(t, y) for t e U,
in other words such thatf,(t', x) : f2(t’, x) for t’ E Ug (resp. gU). Hence A is
open in G and therefore A = G.

PROPOSITION 16. Let G be a connected Lie group, X a Hausdorflmanifold ofclam C’

and f1, f2 law: of left (resp. right) operation of clan C' of G on X. If the laws of
iryiniterimal operation associated withf, andf2 are equal, thenf1 = f2.
By § 4-, no. 7, Corollary to Proposition 11, there exists a neighbourhood V of
{e} x X in G x X such thatfl andfa coincide on V. Hencef1 = f, (Lemma
4).

Lemma 5. Let G be a simply connected topological group, X a Hausdotfl topological

space, U an open neighbourhood «y’ e in G and \II a continuous mapping of U x X into
X such that Me, x) = x and My, ¢(t,x)) = ¢(:t,x) for all xeX and s, t in U
such that it e U. Let W be a symmetn'a connected open neighbourhood g“ e mah that
W3 C U. There exists one and only one law of eontinaour [efl operation qr of G on
Xmahthatill'audxbcaincideonw >< X.IfGiraLiegroupanaman§/ntldof
class C' and t]; it of class C', then 41’ is of class C'.
The uniqueness on 4/ follows from Lemma 4. Let P be the permutation
group of X. For u 9W5, the mapping at» Mu, x) is an element f (u) of P
and
f(“1“2"a)=f(“x)f(“2)f(“a) '
for u), 14,, us in W. Applying Lemma 1 of no. 1, we obtain a morphism f ’ of
G into P which cxtendsW. Let ¢’(g, z) =f’(g)(x) for all (g, x) e G x X.
Then 4/ is a law of left operation of G on X which coincides with s}; on W X Xi
As 4/(g, \b'(g’, x)) = ¢’(gg',x) for (g, g’, x) 6G X G x X, the continuity of
4/ on W x X implies the continuity of h’fl' on gW x X for all g e G. Hence 4"
is continuous. If 44 is of class C', it is seen similarly that d." is of class C’.

THEOREM 5. Let G be a simply connected Lie group, X a compact manifold of clam

C' (r 2 2) and ar—>D‘I a law of left (resp. right) infinitesimal operation of class
C"1 of L(G) an X. There exist: one and only one law of left (resp. right) operation
ofclass C"—1 ofG on X such that the associated law ofinfinitesimal operation is a »—> D“.
The uniqueness follows from Proposition 16. By §4, no. 7, Corollary 1 to
Theorem 5, there exist a neighbourt V of {e} x X in G x X and a law
chunk of left (resp. right) operation of class 0'“ of G on X, defined on V,
such that the associated law of infinitesimal operation is ar—>D,,. As X is

319
Ill LIE GROUPS

an open
compact, V can be assumed to be of the form U x X, where U is
neighbourhood of e in G. It then suflices to apply Lemma 5.

9. EXPONENTIAL MAPPNG IN THE LINEAR GROUP

PROPOSITION 17. Let A be the set of 266 such that —1t < 1(1) < n and A’
the set ty’ 2 e C which are not real <0. let E be a complete normable space over C and
A (resp. A’) the set if): e Y(E) whose spectrum Sp(x) is contained in A (resp. A’). V
Then A (resp. A’) is an open subset of 2(E) (resp. GL(E)) and the mappings
expzA—>A’ and log:A'—>A (Spectral Theories, Chapter I, §4, no. 9) are
inverse isomorphisms of analytic manifolds.
This follows from Spectral Theories, Chapter I, §4, Proposition 10 and no. 9.
THEOREM 6. Let E be a real or complex Hilbert space and U the unitary group of E.
(i) The set H of Hermitian elements of 5?(E) is, with the real normed space struc-
ture, a closed oector subspace of 3(E) admitting a topological supplement.
(ii) The set H’ of elements 20 qL(E) is a real analytic submonifold of GL(E).
(iii) The restriction to H of the mapping exp is an isomorphism of real analytic
manifolds of H onto H’.
(iv) The mapping (h, u) »—> (exp h)u of H x U into GL(E) is an isomoephism
of real analytic manifolds.
Recall that, if x G Y(E), x“ denotes the adjoint of 1:. Let H1 be the set of
x6305) such that x“ = —x. The formula x = fix + x‘) + fix — 3*)
proves that, with its real normed space structure, 3’03.) is the topological
direct sum of H and H1, whence (i).
Suppose that K = C. In the notation of Proposition 17, H’ is the set of
h e H n A’ such that Sp h C Rt. As exp(R) = Rf“ (ii) and (iii) follow fiom
Proposition 17 and S , tral Theories, Chapter I, §4, Proposition 8 and §6,
no. 5. The mappinggh, u) >—> y = (exp h)u of H x U into GL(E) is bijec-
rive by Spectral Theories, Chapter I, §6, Proposition 15. It is real analytic by
the above. The mapping yr—>h = flog(yy") is real analytic and so also is
the mapping y i—> u = (exp h) “y. Hence (iv).
Suppose that K = R. Let E be the complexified Hilbert space of E and J
the mapping E + i-q H E — in (for E, 1; in E) ofE into E. Let H, H’, U de-
note the sets defined for E as were H, H', U for E. Then H (resp. H’, U) is
identified with the set of x E H (resp. H', U) such that ya“ = x. Properties
(ii), (iii) and (iv) then follow easily from (i) and the analogous properties in
the complex case.
PROPOSITION 18. Let E be a complete normable space over C, v e .Y(E) and g = exp 0.
Suppose that Sp(o) contains none of the points 2i1rn with n E Z — {0). Then, for all
x E E, the conditions on = 0 and gx = x are equivalent.
This follows from Lemma 2 of no. 4, applied to the function 2 »—> e‘ — l.

320
 EXPONENTIAL MAPPING [N THE LINEAR GROUP § 6.9

COROLLARY 1. Let E be a rumplete mmmble :[mec aver C and F the xpaee of can-
timwus n-linear mapping: 4/ E" into E. For all u e 2(E), let 0(a) be the element qf
$(F) defined by

(«morn . . m) = won, . . m» — 21ml, . . -,vx:,~~~, x”).

For all g e GL(E), let p(g) be the element afGL(F) defined by

(9(g)f)(xb - - -, x.) = g(f(g“xn ~ ~ ~.g"xn))-

Let mega?) be neck that every zeSpu :attkfiex |J(z)[ < "T! . Then,fl;rall
f6 F, the eondition: o(u)f = 0 and p(e)£p u)f =fare equivalent.
L(p) = a (§ 3, no. 11, Corollary 1 to Proposition 41) and hence

9(EXP '4) = EXP 601)

(no. 4, Corollary 3 to Proposition IO). By Proposition 18 it then sufiices to
prove that Sp 6(a) does not meet 2i1r(Z —{0}) But this follows from the
following lemma:
Lemma 6. ve.9(E), then Spa(v) C 5130 + Spa + + Spv, where the
mm comprixe: n + 1 terms.
We define elements on, 0,, . . ., v,I of .S?(F) by writing, for allfe F,

(”of)(x:,. > on) = "(1'04” > - um)

(vaf)(xnm, ”1-) = "f("1x~-~,Wu~~~,1‘n) for] $ {S n.

Then 11(1)) = ‘2; 1/, and the u‘ are pairwise permutable. Let A be the total
closed subalgebra of 2’(F) generated by the 11,; it is commutative (Spectral

Theories, Chapter I, §l, 1104 4) and Spgmu' = SpA v’ C ‘20 Sp 1), (Spectral
Theories, Chapter I, §3, Proposition 3 (ii)). Now, if A EC is such that u — A
is invertible, clearly the v. — A, are invertible and hence Sp 0‘ C Spy for
all 1'.

COROLLARY 2. Let E be a complete non/table algebra aver C and w E $(E). Sup-

pm‘e that every 1 e Sp w satixfiex [J(2)] < 23"- The following condition: are equiva-
lent:
(i) w is a derivation q;
(ii) exp w 2': an autonwrphirm if E.
This follows from Corollary 1 with fl = 2 andf the multiplication of E.

321
III LIE GROUPS

PROPOSITION 19. Let E be a tomplete normable space over C, v e 56’(E) and g = exp v.
Suppose that every zeSp v satisfies —n < 1(1) < 1:. Then, for every closed vector
subspace E1 of E, the conditions v(E’) C E’ and g(E’) = E’ are equivalent.
The condition o(E’) C E’ implies g(E') C E’ and g“(E’) C E’ and hence
g(E’) = E’. Suppose that g(E’) = E’. We use the notation A, A' of Proposi-
tion 17. Since Sp v is a compact subset of A, there exists a compact rectangle
Q = (a, b] x [a', b’] such that Sp o C QC A. The set A — Qis connected.
Hence Sp g C exp Q C A’, the set exp Qis compact and the set A’ — exp Q
is connected. The closure of the latter contains ]—00, 0] and hence
(A'—ep) U]—ao,0] : C—ep
is connected. Then exp Qis polynomially convex (Spectral Theories, Chapter I,
§ 3, Corollary 2 to Proposition 9) and hence the fiinction log, defined on A’,
is a limit in 0(exp Q) of polynomial functions (Spectral Theories, Chapter I,
§4, Proposition 3). Hence 1; = log g is the limit in $(E) of elements of the
form P(g), where P is a polynomial (Spectral Theories, Chapter I, §4, Theorem
3). As P(g)(E‘) C E', it follows that v(E') C E'.
COROLLARY. Let E be a complete normed space over C, o e $(E) and g = exp v.
Suppose that every 2 eSp v satisfies —1—2r < 1(2) < 7—5: Then, for every dosed vector
subspace M of 3 (E), the conditions gMg" = M and [17, M] C M are equivalent.
Let F = $(E), g' be the map}:oingfv—>gfg'l of F into F and v’ be the
mapping f l—> [v,f] of F into F. Then g’ = expv' (no. 4, Corollary 3 to
Proposition 10 and §3, no. 11, Corollary 1 to Proposition 41). Lemma 6
proves that — 1: < 1(1) < 1: for all z e Sp 11’. It then suffices to apply Proposi-
tion 19.

10. COMPLEXIFICATION OF A KNIFE-DIMENSIONAL REAL LIE GROUP

Lemma 7. Let B be a group, A a normal subgroup ofB, C the group B/A and i : A —> B
and p: B —> C the canonical morphisms. Let A’ be a group andf a homomorphism 9f
A into A’. Let a) be a morphism of B into the automorphism group of A'. Suppose that,
for aeA’, o’ eA’, beB,
f (1’05“) = (”WK“); “(4)4' =f(fl)a'f(a) ‘1~
Let B” be the semi—direct product of B by A’ relative to (a and q the canonical morphism
of B” onto 3.
(i) The mapping a>—> (f(a"), 1(a)) of A into B” is a morphism ofA onto a
normal subgroup D ofB". Let B’ = B'ID and i': A' —> E', g: B —> B' be the morphisms
WFMAI; and 1; into B’ derived when taking quotients fiom the canonical irg'ections of A’
a into ".
(ii) The morphism p o 1] 13/3” into 0 defines when passing to the quotient a morphism
of B’ into C.

322
 CDMPLEXIFICATION or A “NIKE-DIMENSIONAL REAL 1m GROUP § 6.10
. (iii) 1" it injecting, p’ i: .mjem’w, KerUz') = In_1(i’) and Ihefallowing diagram
15 mmmutatwa
A —‘> B .3.) C
(4-) r a in,
A’ LB’L’) 0.
(iv) [fbeBanda’eAfl [hm
.4:(’7)i'('1')g(b)‘1 = i'(‘»(”)a')-
(i) For an 12 in A, we have, in B”,
(f(01_ 1): z'(111))(f(42 ); 1.(012)) = (flail) (“(111)1'011», “405012”
=(f(flf )f (11“: an" ), “than”
= (f((”x”2)—l i(“1“2))
and hence a» (f(a“),”i(a)) is a homomorphism I: ofA into B” Let a GA,
a eA’, beB; then,inB’,
W4)b 1 = ”(VIM—1 = (“(b)f(a'1))(bab")
=f(ba“b‘1)(bab'1) = h(bab‘1)
a’h(a)a"1 = a’f(a")aa"‘ = aff(a")(m(a)a"‘)a
= «7(a“)f(u)a"V(a")a = 1'01)
and hence h(A) = D is normal in 13'.
(ii) For a e A,
(1’ ° 001(4)) = P(q(f(a")a)) = [1(0) = 6
and hence p o q is trivial on D.
(iii) Let a’ e A’ be such that i’(a’) = a; then a’ e D and hence there
exists a e A such that a' =f(a“)a; this implies a = 2, whence a’ = z; thus
1" is injective. As [1 and q are surjective, p’ is suzjectivc.
Let r denote the canonical morphism of B" onto B’. Let a' e A’, I; 53 and
b’ = r(a’b). If b’ e Im(i'), there exists a; e A’ such that b’ = r(a§); then there
exists a GA such that a’b = a;f(a“)a, whence b = :1 EA and
NV) =1’(q(a'b)) = [’(b) = I;
thus, Im(i') C Ker(p’). We preserve the notation a’, I], b’ but assume that
b’ e KcrUz’) ; then 2 = p’(b’) = p(q(a’b)) = p(b) and hence b E A, whence
0' = '(a'f(")f(b")”) = '(fl'f(b))€1m(i');
thus Ker([z') C Im(i’).
If a e A, then

i'(f(a)) = ’(f(“)) = '(f(")f(a“)fl) = 7(a) = NW)-

323
HI LIE GROUPS

If h E B, then
P'(g(b)) = M”)
and hence diagram (4) is commutative.
(iv) Let h e B, n' E A’. Then
g(b)i’(4’)t'(’>)‘1 = 700443701)" = ’(WV‘)
= '(NU’W) = 1"(@0003
PRoposmou 20 Let G be 11 finite—dimensional real Lie group.
(i) There exist a complex Lie group G and an R-nnnlytie morphism y of G into G
with the following properties: for every eomplex Lie group H and every R—onolyh'e
morphism 4) of G into H, there exists one and only one C-analylw morphirm q, of G
into H such that (11:4, 0 Y
(ii) If (day) has th_e some properties a: (G, 7), there exist: one and only one
isomorphism 6 ofG onto G’ such tha! 9 0 Y = y'
(iii) The C-linear mapping of L(G) ® G into L(G) which extend: L(Y) is our-
jeetioe; in particular dimc(G) < dimR(G).
Assertion (ii) is obvious. We prove the existence of an ordered pair (G, 7)
with properties (i) and (iii)
(21) Suppose first that G is connected. Let g =.L(G), g: = g ®R C be the
complexification of g, S (resp. S’) the simply connected real (resp. complex)
Lie group with Lie algebra g (rap. 9.) and o- the unique R—analytic morphism
of S into S’ such that L(o-) is the canonical injection of g into 99. Let 1: be the
unique R-analytic morphism of S onto G such that L(x) = Idua) and F =
Ker 1:4
1 -———>S

’1
1
6.
G ———> G¢

For every complex Lie group H and every R-analytic morphism (I) of G
into H, L(<Ii):g—>L(H) has a unique C-linmr extension to Sc and this
artension is of the form L(<{>*), where d)’ is a (fl-analytic morphism of S’ into
H. Then
L(i> ° 7') = 1-(4’) ° N") = L015) = L(<l>‘) ° 140) = LW‘ ° 6)
and hence 4’ o 11: = tb‘ n 6. Therefore ¢‘(5(F)) = d>(1r(F)) = {e}, whence
a(F) C Ker <1”.
Let P be the intersection of the Ker 42* for variable (I). This 15_a normal Lie
subgroup of S’ (no. 2, Corollary 3 to Proposition 1). LetG =S'[P and

324
 courmxrrrmrrou or “NM-DIMENSIONAL REAL LIE GROUP § 6.10

A: S’—>G be the canonical morphism. Then o(F) C P and hence there

exists one and only one R—analytie morphism y of G into G such that
yo 1: = Ana. If 4:: G—>H denotes the morphism derived from 42" when
passing to the quotient, then
(4>°Y)°1== ¢ (1...): roa=¢on
whence q, o 7— = 4». Clearly L(¢), and hence 4:, are determined uniquel
the equality s]; o Y = 4). Thus we have proved that the ordered pair((fibyéy
has properties (i) and (iii).
(b) We pass now to the general case. Let F be the identity component of
G, M = G/F and i: F——>G andp: G—>M be the canoniml morphisms. We
apply part (a) of the proof to F. We obtain an ordered pair (F, 8). For all
g E G, Intg|F = w'( g) is an automorphism of F. By the universal property
of F, there exists one and only one automorphism m(g) of the complat Lie
group F such that 8 o (n'(g) = m(g)o08. Clearly a) is a. morphism of G into
Aut(I:I). If g e G andfe F, then
Mm") = (3%“(3))(f) = (0(5) °3)(f) = w(£)(3(f))
Iffe F, then 8 o (Intpf)=(Int1~, 8(f))o08 and Inti 8(f) 13 an automorph-
ism of the complex Lie group F, hence Int; 8(f) = t.)(f )
Hence Lemma 7 can be applied, which gives a diagram
F-—>G—>M
.11.l
F——>G—p>M.
Let F be identified with a normal subgroup of G by means of 1'. The group G
is generated by F and 7(G); hence the automorphisms of F defined by the
elements of are automorphisms of the complex Lie group structure. By
§ 1, no. 9, Proposition 18, there exists one and only one complex Lie group
structure on G such that F is an open Lie subgroup of G. Henccforth G will
have this structure. As 8 is R-analytic, y is R-analytic.
The ordered pair (G,-y) has property (iii) of the proposition. We show
that it has property (i). Let H be a complex Lie group and 4; an R—analytic
morphism of G into H. There exists a C-analytic morphism 1] of F into H
such that n o 8 = ¢|F. LetgeG. The mappings
f” fl(¢¢(g)f)’ f” <l>(g)71(f)d>(g)"
of F into B are C—analytic morphisms; they coincide on 8(F), for, iff 'e F:

<1>(g)n(3(f'))¢(g)“ = <l>(g)¢(f’)<l>(g)" = Mgf'g")

= n(5(gf'g")) = n(w(g)3(f'));
325
III LIE GROUPS

therefore ‘4(¢o(g) f) = (I)(g)1‘(f)(h(g) -1 for all g e G and all fe i". If G’

denotes the semi-direct product of G by F relative to u, there then exists a
morphism C of the group G' into H which coincides with <1) on G and with
noni‘. ForfeF,
C(8(f'1)f) = 7'I(3(J"‘))<l>(f) = <l>(f")<l>(f) = 5-
Hence 2 defines on passing to the quotient a morphism 4; of G into H. Then
41 0 Y = 4: and q; o z‘ = n; the latter inequality implies that 4; is C-analytic.
Finally, let 4/ be a C—analytic morphism of G into H such that d) = d)’ a Y-
en
(1/0508 =¢'oyoi=¢°i= $0503
and hence q/ . z' = vb o i. As G is generated by id") and 740), 4/ = 4,.
DEFINITION 4. (G, 7), or simply G, i: called the universal tomplexficution 13/" G.
Remarks. (1) Let (G, y) be the universal copplexification of G. Let Go (resp.
Go) be the identity component of G (reap. G). By the proof of Proposition 20,
(Ga, ylGo) is the universal complexification of Go and the composite morph-
ism
G -> G —> G/Go '
defines on passing to the quotient an isomorphism of G/Go onto G/Go.
(2) Suppose that G is simply connected. Let g = L(G), go be the complexiti-
cation of g, S’ the simply connected complex Lie group with Lie algebra 9.;
and a- the morphism of G into S' such that L(¢) is the canonical injection of
9 into gc. We again use the notation of the proof of Proposition 20, part (a).
If H = S’ and (b = a, then ¢* = Idsl. Hence (5', a) is the universal com-
plexifimtion of G. Note that a is not in general injective (Exercise 16); how-
ever iAr kernel 1': dismle since L(o) is injective. On the other hand, let 8 be the
involution of 9c defined by g and let 7] be the corresponding automorphism
of the underlying real Lie group ofS’ ; let S’" be the set ofpoinls of S’ invariant
under 719 it is a real Lie subgroup of S' with Lie algebra 9 (§ 3, no. 8, Corol-
lary l to Proposition 29). By no. 1, Corollary 1 to Proposition 1, C(G) is a
real integral subgroup of S' with Lie algebra g and hence 0(0) 1': the identity
component qf S'"; in particular 6(G) is a real Lie subgroup of S'.

§7. LIE GROUPS OVER AN ULTRAMETRIC FIELD

In this paragraph, the valued field K is assumed to be ultrametric and of

characteristic 0. Let A denote the valuation ring of K, m the maximal ideal
of A and p the characteristic of the residue field A/m. If K is locally compact,
then 11 71- 0 (Cmmulative Algzbm, Chapter VI, §9, Theorem I).

326
 PmGE mom us ALGEBRAS TO LIE onoups § 7.1

l. PASSAGE FROM LIE ALGEBRAS TO LIE GROUPS

PROPOSITION 1. Let G be a Lie group gem with identity element e. There exists 4
fiendamental system of open neighbourhoods of e in G cam'sting of Lie subgroups of G.
Let L(G) be given a norm compatible with its topology and such that
‘|[(xG,y]|l < ”x” "y” for all x, y in L(G). Let G, be the Lie group defined by
).By §4, no. 2, Theorem 2, G and G, are locally isomorphic. Then it
sugces to apply § 4», no 2, Lemma 3 (iii)

THEOREM 1. Let L be a complete normable Lie algebra. There exists a Lie group G
such that L(G) is isomorphic to L. Two such groups are locally isomorphic.
The first assertion has been proved in §4, no. 2, Lemma 3, The second is
a. special case of § 4, no. 2, Theorem 2.

THEOREM 2. Let G be a Lie group and b a Lie subalgebra of L(G) admitting a topo-
logical supplement. There exists a Lie subgroup H of G such that L(H) = b. If H1
and H, are Lie subgroups y” C such that L(Hl) = L(Hz) = b, then H1 n H2 is
open in H1 and H2.
The first assertion follows from Proposition 1 and {$4, no. 2, Theorem 3.
The second is a special case of § 4, no. 2, Theorem 3.

THEOREM 3. Let G and H be Lie groups and h a continuous morphim of L(G) into
L(H).
(i) There exist an open subgroup G’ of G and a Lie group morphism b of G’ into
H such that h = L((b).
(ii) Let G“ G, be open subgroups ofG and (I). a morphism «MG, into H such that
h—- L((b‘). Then (I), and b2 coincide on an open subgroup ofG
By Proposition 1, this follows from § 4, no. 1, Theorem 1.

PROPOSITION 2. Let G be a Lie group and I) a Lie subalgebra of L(G) admitting a

The ‘ conditions are ,
M(i) There exist an open subgroup G of G and a normal Lie subgroup H of G'
such that L(H) = I)
(ii) b is an ideal afL(H).
If there exist G' and H with the properties of (i), then L(G') = L(G)
and L(H) is an ideal of L(G’) by § 3, no. 12, Proposition 47.
Suppose that b is an ideal of L(G). There exists a Lie group F such that
L(F) = L(G)/b (Theorem 1). Let h be the canonical morphism of L(G)
onto L(F). By Theorem 3 (i), there exist an open subgroup G’ of G and a
Lie group morphism ()2 of G' into F such that L((b) = h. By §3, no. 8, the
kernel H of b) is a Lie subgroup of G' and L(H): Ker L(tb)= Kerh = 1;.
Finally, H is normal in G' since H : Ker d).

327
III LIE GROUPS

2. EXPONENTIAL MAPPINGS
PROPOSITION 3. Let G be a Lie group. There exists an exponential mapping 4; of G
with the following properties: -.
(i) (I: is defined on an open subgroup U J the adahtwe group L(G) ; . _
(ii) ¢(U) is an open subgroup ofG and (I; air an uomorphum qfthe analytic manJold
U onto the analytic mom'fild MU);
(iii) ¢(nx) = 4>(x)"for ollx EU andall n 52.
Let L(G) be given a norm compatible with its topology and such that.
[I Dr, y] [I < ”x“ "y“ for x, y in L(G). Let G, be the Lie group defined by L(G).
Let 4; = IdG‘, which is an exponential mapping of G1. For all p. > 0, let L"
be the set of x e L(G) such that ||x|| < It. Then, for y. sufficiently small, L,
is an open subgroup of the additive group L(G), ML”) is an open subgroup
of G1 (§ 4-, no. 2, Lemma 3), ML“ is an isomorphism of analytic manifolds
of L“ onto ML”) and Mm) = 440:)” for all x e L“ and all n E Z. The L“ form
a fundamental system of neighbourhoods of O in L(G). By Theorem 1, there
exist p. and an open subgroup G' of G such that ML“) and G’ are isomorphic,
whence the proposition.
PROPOSITION 4. Let G be a Lie group and (I; an injective exponential mapping of G.
Suppose thatp > O. For all x, y in L(G),
(1) x + y = ”5-3“ [2 ”11>“(d>(o"x)¢(o"y))
(2) [w] = 11mm P""<l>"(¢(p"x)<l>(t’"y)<l>(-P'x)<l>(-!I"y))-
These are Special 0333 of Proposition 4 of § 4, no. 3.

3. STANDARD GROUPS?
If S(Xl, X2, . . ., X“) is a formal power series with coefficients In A, then,
for all x), . . .,x, in m, the series S(x,,x2,.. ,x,) is convergent. More pre-
cisely, m x m x ~ ~ - x m is contained'm the domain of absolute convergence
of S (Dferenttable and Analytic Manifolds, R, 4- l 3).
DEFINITION 1. Let r be an integer >0. A standard group of dimention 7 over K is
a Lie group G with thefollowing propertier:
(i) the underlying analytic manfold of G u m x m x - - - x m (rfaotorx) ,
(ii) there exists a formal power series F in 21 variables with coefioient: in A',
without ootant term, ouch that x._1/—= F(x, 1]) for all x, y in G.
Then 0.0 = 0 and hence the identity element of G is the origin of
m x m - - - x m.

T The results of nos. 3 and 4 and their proofs remain true when the character-
istic of K is > 0.

328
 STANDARD GROUPS § 7.3

L(G) will be identified with K'. By § 5, formula (13), the constants of struc-
ture of L(G) with respect to the canonical basis belong to A. We shall need,
in the same proof to consider the elements of m X - - - X m, now as elements
of G, now as elements of L(G).

Example. Let G = 1 + Mfl(m), which is an open subset oa(K). Ifx EC,

then detxel + mand henceG C GL(n,K). ClearlyGG C G. Ifx = 1 +1]
with y EMn(m), the calculation of the inverse of a matrix proves first that
x‘1 eMn(A); if we write x“ = l +y', then 3/ +y’ +yy’ = 0, hence
y’ eMn(m) and therefore at" E G. Thus G is an open subgroup of GL(n, K).
We identify G with 111"2 by means of the mapping (8‘, + y”) >—> (31”). Clearly
G is a standard group.

THEOREM 4. Let G be a finile—dimemional Lie group. There exirt: on open subgroup

of G isomorphic to a standard group.
By replacing G by a group isomorphic to an open subgroup of G, the prob-
lem is reduced to the case where G is an open subset of K', with identity
element 0 and where the coordinates of the product my and the inverse
act—11 are given by formulae

(3) (x-y); = M +!Is + ”DEN“ calf!!!” (i = 1: 2- ..,r)

(4) (“r—u)1= _z‘+|u§l dmx“ (i= l,2,...,r)

where the series on the right hand side are convergent for x, y in G (§ 5,
no. 1). Let leK" and let the group law be transported from G to G’ = XG
by the homothety of ratio A. For s“, y’ in G’, the product x’.y’ and the in-
verse x’l‘“ evaluated in G' have coordinates

(xhy'). = x: + y: + WE“), cow“ (i = 1,2,...»

(fill-11h: _x;+ 2 was Iu|>l
(i: 1, 2,...,r)
where
as.. = x-*“'-'°'“o.a., 4;. = x-Iu'wa.
As the series (3) and (4) are convergent, we see that, for |A| sufficiently large,
for all a, p, i,
l‘itail g 1; [dulul g I

thatiscgmeAanddéeA;andontheotherhandG' 3m x m x x m.
Then m x m x x m isanopen subgroup ofG'andisastandard group.

329
Ill LIE GROUPS

4. FILTRATION 0F STANDARD GROUPS

We again use the notation of Definition 1. We choose a number a > I and a
real valuation v of K such that Ix] = n‘""" for all x e K (Commutative Algebra,
Chapter VI, §6, Proposition 3). If a is a non-zero (and hence open) ideal of
A contained 1n m, let G(a) denote the set of elements of G whose coordinates
belong to a. If 716R, let 0,, (resp. a; ) denote the set of xeK such that
o(x) 2 71 (rap. 11(x) > A); then do = A, a; = m. For x = (xv. ..,x,) 6G,
we shall write _

(5) “(1) = inf(v(x1), - - ‘) 0%))-

PRoposmon 5. Let G be a standard group.
(i) If a is a non—zero ideal of A contained in m, G(a) i: on open normal subgroup
o G.
f(ii) The G(a,), for A > 0, form a findomental tyttem of neighbourhood: of e in
1.
(iii) Suppote that 0,. C afor A 2 1,, and let the G(u)/G(o,_),for )t > A0, be
given the discrete topology. Then the topological group G(a) i: the inverse limit of the
gm!” GOO/OW)-
(iv) Let a, b be non-zero ideals of A contained in m such that a =2 b D a”. The
mapping x1—> (1:, mod b, . . ., 1:, mod b) ofG(a) into (a/b) x - - - x (n/b) define:
on passing to the quotient an isomorphism of the group G(a)/G(b) onto the additive
group (a/b) x x (a/b).
If x e G and y E G(a), the coordinates of x and any are equal modulo 0.
Hence, for z’, x” in G and g', y" in C(a), the coordinates of x’.x’ and
(x’ .y') . (x' .y”) are equal modulo 0. This proves (i).
(ii) is obvious.
(iii) follows from the above and General Topology, Chapter III, § 7, Proposi-
tion 2.
If x eG(a) and 1/ 60(0), the coordinates of any are congruent to those of
x + y modulo G012) by formula (4-) of § 5. This proves (iv).
COROLLARY. Suppose that K is locally compact and let q = Card(A /m)
(Si) )[fa = m“ and b = m” with b > a > 1, G(a)/G(b) Lt ap-group tfcardinal
qr 1:
(ii) G(a) i: an inverse limit ofp—groups.
The number of elements 1n G(a)/G(b) 1s (Card(u/b))’; if b = a + 1, a/b
is a l-dimensional vector space over A/m, whence (i) in this case; the general
case follows by induction on b — a. Assertion (ii) follows from (i) and Proposi-
tion 5 (iii).
Pnorosmou 6. let a, b, c, c’ be non-zero ideals of A contained in m such that
c’ C c, ab C c, at)2 C c', aab C c’.

330
 POWERS 1N STANDARD GROUPS § 7.5

(fx 6 G(a) andy e C(b), then xi‘“.y['“.xiy, x.y.x““y"” and [x, y] baking to
G(c) and an congruent moduli) C(c').
By § 5, no. 2, Proposition 1, there exist cue A’ such that
-ll [—1] _ _
K‘ .1] .xy [x,y]—Mglncflmfl.
If x = 0 or y = 0, then xI'".y[‘“.x.y — [x,y] = 0; hence ”on = can = 0.
On the other hand, the conditions
xEGO‘): yEGU’), I'll > 1, [BI > 1, [“I + ”3| 9 3
imply
cmpx'y“ EG(a3b + ab“) c C(c’)
and hence xi'u.y“"x.y - [x,y] E G(c’). We see similarly that
x.y.x"“.y"“ -— [x,y] eG(c’).
Finally, by § 5, formula (13), [x,y] eG(ab) C C(c).
PROPOSITION 7. (i) The family (C(01)) i: a centralfiltmtian on G (Chapter II,
§4, no. 4-, Definition 2).
(ii) For AeRfi, G(a,_) = {xeG]m(x) 2 7r}, G(a,f)= {xeG|m(x) > )1}.
(ii) is obvious. We prove (i). Clearly cm): ('1 G(a )andG= AU Gm)
On the other hand, if x 6 0(1).) and y e G(au), then
zl‘n-yhu-“JEGU‘mQ
by Proposition 6 applied with a = ax, b = a“, c = c' = a“...
By Chapter II, §4, no. 4, we can form the group gr(G) associated with
the group G, with the central filtration (0(a)). Writing GA = G(a,.)/G(a,f)
for all 1 > 0, we obtain gr(G) = $190 Gk Recall (lac. n'L, Proposition 1)
that the commutator in G allows us to define a bracket in gr(G) with which
gr(G) is a Lie algebra, as follows: if? E G,‘ and? e G”, choose a representative
:1: of x in G(u,_) and a representative y of? in 0(0“); then [if] is the class of
x“ 11 .y“ 1‘ .x yeG(aMu) in GM". By Proposition 6, applied with a = a,”
b = a“, c = «Mu, c’ = ax”, we see that [fey] is also the class of [x,y] in
GM“. Thus, when G 15 considered as a Lie subalgebra of L(G)= K', filtered
by the G(ah), the associated graded Lie algebra (Chapter II, §4, no. 3) is
equal to gr(G).

5. POWERS 1N STANDARD GROUPS

We preserve the notation of no. 4-.
PROPOSITION 8. Let n e Z and h,I he the mapping max" ofG into G. Let a he a
nan-um ideal of A canlained in m, such that 71 ¢ 0. Then h,.|G(a) it an isanmphirm
qfthe analytic manifold G(a) unto the analyh'v twofold C(na).

331
'fll LIE GROUPS

By definition of standard groups, [1,, is equal on the whole of G to the sum

of an integral series with coefficients in A’. By §5, formula (4), this sefies is
of the form
W) = m: + M2; we
Hence, for s e G,
’50”) = “(K + Z a,n""’x“)
[an >2
= 71550:)

where we write 5(a) = at + “:22 nan'“"’x“. This series S(x) defines an

analytic mapping, also denoted by S, of G into G. By Algebra, Chapter IV,
§6, Proposition 8, there exists an integral series 5' in 1 variables with coeffi-
cients in A' such that S’(S(X)) = S(S’(X)) = X. Hence S is an isomorphism
of the analytic manifold G onto itself and, for every non-zero ideal b of A
contained in m, S(G(b)) C G(b), S'(G(b)) C G(b) and therefore
S(G(b)) = C(b).
As h,,(y) = n’SGy) for y EnG, we see that h,[nG(b) is an isomorphism of
the analytic manifold nG(b) onto the analytic manifold n2G(b). But, as n é a,
|n| > [Al for all 7‘ e 0, hence n’la C m and hence a is ofthe form nb where b
is a non-zero ideal of A contained in m.
COROLLAe (fn is invertible in A, II" is an isomorphism of the analytic manifold G
anla itself. For every nan-zero ideal a of A contained in m, h,,(G(a)) = C(11). For all
x e G, m(x") = 6(a).
This follows immediately from Proposition 3.
PROPOSITION 9. Suppose that p 56 0.
(i) Let a, b be nan-zero ideal: q such that b C a C m. In thegroup G(a)/G(b),
every demon! has order a power of1).
(ii) Suppose um 12(9) = 1. Ifxe G i: mh that m(x) > fl, then
m(x’) = m(x) + 1.
By § 5, formula (4), for all x E G,

x’=px+ 2 tax“
lulu
where c, E A' for all (X. Even for proving (i) it can be assumed that 110) = 1.
Then if m(x) 2 1, it follows that (”0:”) 2 m(x) + l and hence m(x"') tends
to +00 as It tends to +00; this proves (i). As (6) is divisible by p for

332
 LOGARITHMIC MAPPING § 7.6

l S 1' < I — 1, Proposition 2 of§5, no. 3, proves that on. epA' for
2 s |u| <12 — 1 '
and hence
More“) > toque) = «3(3) + 1 for2 < M sp — 1.
On the other hand, if [an] 2 p, «(59%) > pm(:t) and park) > 01(x) + 1 if
l
o)(x) > P _ l « This proves (ii).

G. LOGARITHMIC MAPPING
Lemma 1. Suppose that p aé 0. let G be a Lie group, Gl an open subgroup of G which
is isomorphic to a standard group and x e G. Thefillowing condition: are equivalent:
(i) there exist: a power tfx which belongs to G1;
(ii) there exist: a strictly increasing sequence (n.) of integer: such that x": tend: to
e a: i tend: to +00.
(ii) = (i): obvious.
(i) =- (ii): suppose that y = 5:"e By Proposition 9 (i) of no. 5, y"
tends to e as n tends to +00, in other words rm” tends to e as It tends to +00.
PROPOSITION 10. Suppose that p aé 0. Let G be a finite-dimensional Lie group. Let
G, be the set of x e Gfir which there exist: a strictly inereming sequence (71.) qfintegm'
ruch that x"! tend: to e a: i tend: to +00.
(i) G, Lv open in G.
(ii) There exists one and only one mapping 4; of G, into L(G) with the following
properties:
(a) ¢(x") = n¢(x)for all x e G, and all 7152;
(b) there exist: an open neighbourhood V of e in G Mh that 411V i: the inverse
mapping of an injective exponential mapping.
(iii) The mapping 4; ix analytic.
There exists an open subgroup of G which is isomorphic to a standard
group (no. 3, Theorem 4-). Assertion (i) then follows from Lemma 1.
Let U be an open subgroup of L(G) and <1):U—><]>(U) an exponential
mapping of G with the properties of Proposition 3 of no. 2. U can be assumed
to be small enough for MU) C G,. Let xeG,. There exists meZ —{0}
such that x’" E ¢(U). The element-”11¢”(xm) does not depend on the choice
of m. For let m’ e Z be such that a“ e (NU). Then an“ e (MU) and
WWW”) = <l>“(x"‘"") = WWW),
whence our assertion. Let Her) = %¢"(fl)- Then MMU) = d)". On the
other hand, if n e Z, then

W") = fizb-‘(m =54r1<m = nw).

333
Ill LIE GROUPS

Hence I]; has properties (a) and (b) of the proposition. In a neighbourhood of
l
x, L]; is composed of the mappings x I—>:c’“, y» ¢'1(y) and z I»; 2; hence kl;
is analytic on 6,.
Finally, let da’ be a mapping of G, into L(G) and V' a neighbourhood of
e in G, such that ¢’(z") = nil/(x) for xEG, and a El and such that ¢’|V’
is the inverse mapping of an injective exponential mapping. Then i]; and 4/
coincide on a neighbourhood W of e. If are G,, there exists n 52 such that
x" s W. Then
"4"(x) = WI") = 440‘") = MN")
and hence x]; = 44’.
DEFINITION 2. The mapping 4! qmpoxition 10 is called the logarithmic mapping of
of G and is denoted by logo or simply log.
PROPOSITION ll. Sufioxe that p ;é 0. Let x, g be two permutable element: of G,.
Then xy E G, and log(xy) = log 1: + logy.
The fact that any E G, follows from Lemma 1. Let U be an open subgroup
of the additive group L(G) and (I): U —>¢(U) an exponential mapping of G
with the properties of Proposition 3 of no. 2; U can be assumed to be small
enough for loglMU) to be the invexse mapping 'of (b. For n E Z — {O} suitably
chosen, 1" E¢(U), y"e<]>(U). Let u = log :6", v = logy”, whence x” = (Ma),
y” = (My). By formula (2), [u, v] = 0. The Hausdorff formula proves then
that (1)001 + 12)) = 4>(M)4>(M) for N sufficiently small; hence, for every
sufficiently large integer 1',
MN“ + 0)) = ll>(1"u)4>(#”)
that is
["008 at” +10%") = 1°g(x""y’“")
or
np‘aogx + logy) = "1" 1°g(ry)-
PROPOSITION 12. Suppose that p 56 0. Let xeG,. The following conditions are
equivalent.-
(i) logx = 0;
(ii) x is offinite order in G.
If there exists an integer n > 0 such that x" = e, it follows that
nlogx = logx" = 0,
whence logx = 0. If log): = 0, let V be a neighbourhood of e in G, such
that loglV is the inverse mapping of an injective exponential mapping. There
exists an integer n > 0 such that x"eV; the equality log x" = 0 implies
z“ = e.

PROPOSITION I3. Sit/thaw that [z ¢ 0. If G ix eampart or standard, then G, = G.

334
 momma MAPPING § 7.6

If G is standard, it suffices to use Lemma 1. Suppose that G is compact.

Let x e G and V be a neighbourhood of e in G. Let y be a. limit point of the
sequence (x"),.,o. For all n > 0, there exist two integers n), n, such that
n1 > 2712 2 n and x”: eyV, x"2 eyV, whence aft—"2 eV'1V and n1 — na 2 n.
Hence x e G/.
COROLLARY. Suppose that K is locally congoact. Then G, is the union of the compact
subgroups of G.
Let x e G. If x belongs to a compact subgroup of G, then x e G, (Proposi-
tion 13). Suppose that x e G,. As K is locally compact, there exists an open
subgroup G1 of G which is compact. Then there exists an integer in > 0 such
that x'" 6 G1. The closed subgroup G, generated by :c’" is contained in GI
and is therefore compact. Then at commutes with the elements of Ga and
hence G, U xG, U - ~ - U z’"“Gz is a compact subgroup of Gwhich contains
x.
Example. Suppose that K is locally compact. Let U be the set of invertible
elements of A; it is a compact open subgroup of the Lie group K". Then
U c (K‘), by Proposition 13; on the other hand, if :reK“ is such that
x¢U, either 1" tends to 0 as n tends to +00, on" tends to 0 as n tends to
—00; hence U = (K*),. The function Iogx. is defined and analytic on U,
with values in L(K‘) = K, and such that logx.(xy) = logx.(x) + logK.(g/)
for all x, y in U; the elements a: of U such that logx.(x) = 0 are the roots of
unity of K.
We again use the notation of nos. 3, 4- and 5.
Pnoposmon 14-. Suppose that p $5 0 and that v is chasm such that v(p) = 1. Let
G be a standard group and E(X) (resp. L(X)) the expansion of the exponentialfunc-
tion of G (resp. the logarithmicfunction of G) as an integral series about 0.
(i) The domain of absolute convergence (Drflerentiable and Analytic Manifolds, R,
4.1.3) ofE contains the set A ofx e G such that m(x) > [fi- Let E’ denote the
mapping defined on A by this series. Then E’ is an exponential mapping of G and is an
isomorphism of the manifold A onto itself.
(ii) The domain If absolute convergence of L contains G. Let L’ denote the mapping
on G defined by this series. Then L' is the logarithmic mapping g” G and the restric-
tion of L’ to A is the inverse mapping of E’.
(iii) The mapping E' is an isomorphism If A, with the Hausdozfl law, onto the
subgroup A of G.
Using the notation of § 5, nos. 3 and 4, E = "12>; 3%" (§ 5, no. 4, Proposi-
tion 3). As the coefficients cw, belong to A, [|q;,,,,,,|| S l (Diflerentiable and
Analytic Manifolds, R, Appendix) K' is assumed to have the noun
“(Au-n, Ar)|| = sup(lhl,- - -: IND)-
335
"1 UR GROUPS

m — l 1
By Chapter II, §8, no. 1, Lemma 1, ”(m!) g p _ 1- If m(x) >§_—1: we
see that m(x) — 0(m!) tends to +00 with m, whence

““2!“ l xl " s W W
l

which tends to 0 as m tends to +00 and

m'm(x) m _m—l_ 1 >
m(—m!—) >P_"1 ——p _ l —p—-—_1 form ,1.

Therefore A is contained in the domain of absolute convergence of E and

E’(A) C A. Clearly I?" is an exponential mapping. 4 .
If L,” denotes the homogeneous component of L of degree In, Proposrtron 3
of § 5, no. 4, proves that each coefficient of L". is of the form
1
”1+éas‘l‘ +54»-

with a,,a,,...,ami.nA; but

inf(v(l), 0G), . . ., 4%)) = 0(log m)- as m tends to +00

M(,,(1),u(;),...,.(,;)) 2 ”(5) s _r;_;;.

Therefore, ifu>(z) > 0, ]|L,,,|[.||x|[’" tends to 0 as 7/: tends to +00, so that G
is contained in the domain of absolute convergence of L. On the other hand,
H W) >131, then m(L,,.(x)) >1141 — p”5—1 — —1— for m 21 and
— l " p —- 1
hence L’(A) C A.
As the formal power series L(E(X)) and E(L(X)) are equal to X, no. 4.1.5
of Drfimm'abl: and Analytic Many'oldr, R, proves that
L'(E'(x)) = E’(L'(")) = x
for x E A. Hence E' is an isomorphism of the manifold A onto itself and the
inverse isomorphism is the restriction of L' to A.
L(XW) = nL(X) for u an integer >0 (cf. §5, no. 4). As G is contained in
the domain of absolute convergence of L and XI"), therefore L’(x") = nL’(::)
for all IEG. The relation L’IA = E"1 implies that L'(x") = log at“ for n
sufficiently large. Hence L’(x) = log x. We have thus proved (i) and (ii).
Let H = 1 no H“ , be the Hausdorff formal power series and h the Haus-
dorff function relative to L(G). The domain of absolute convergence of 1:1

336
 CONTINUOUS uonpmsus § 8.!

contains A X A and h is defined on A x A (Chapter II, §8, Proposition 2).

Then
E'(1)E'(y) = E'(h(x,y))
for x, y sufficiently close to 0 (§ 4, Theorem 4 (v)). Hence, in the notation of
no, 3, Definition 1, the formal power series F(E(X), E(Y)) and E(H(X, Y))
are equal. Let x, y be elements of A. Then
supllm“’m "”(supnxn llyll)" < 1
min. 1 lu Ilyl ‘ < mw-n
by Chapter II, §8, formula (14). By Dgfnsnliable and Analytic Mani/31%, R,
4.1.5 13' (x)E’(y) 15 obtained by substituting x for X and y for Y 1n
F<E<X), E(Y))
and E'(}1(x, 31)) is obtained by substituting x for X and y for Y in E(H(X, Y))
Hence E’(x)E’(_y) = E’(h(x,j)).

§8. LIE GROUPS OVER R AND Q,

1. CONTINUOUS MORPHISMS
THEOREM 1. Let G and H b: two Lie group germ: aver R or Q,. Let f be a contin-
uum morphirm qf G intn H. Then f i: analytic.
We give L(G) and L(H) norms which define their topologies and such that
”[1,11] H S "x“ [I y” for all x, 1/, There exist an open ball V of centre 0 in L(G)
and an exponential mapping (I; of G defined on V such that: (l) (NV) is an
open neighbourhood of e in G; (2) ¢ is an isomorphism of the analytic mani-
fold V onto the analytic manifold ¢(V) ; (3) (Hm) = 420:)" for all x e V and
all n e Z such that m: E V. We define similarly W and q; for H By shrinking
V if necasary, it can be assumed thatf (¢(V)) C MW). Then g= qr1°fa
is a. continuous mapping of V mto W.
We show that
(1) (e, AeQandMeV) =>g(M) = Ag“).
1: can be assumed um i as 0. Let A = iwith y, qinZ — {0}. Lety = 31x.
IfK=R,weWritez=$=€EV.'Ihenx=qz,y=rz,whence

g(x) = ¢"(f(¢(lll))) = il“(f(¢(l)'))= ¢"(f(¢(l))“)-

We show that 4;" (f(¢(z))") =q¢'1(f(<]>(z)))=qg(z). It suffices to verify
that, ifu 6 MW) is such that u" 6 MW), then 41—104") =q¢‘1(u); but if 11:1);
(v), aka/(m. then m e w and («w/«new. hence ¢(v1/q)=u=¢(v) and
therefore v1=qv.
Similarly, g(y) = rg(z), whence (l).

337
In LIE GROUPS

If K = Q,” we write 2 = 1x = qe, whence g(z) = rg(x) = qg(y),

whence again (1).
As Qis dense in K, (1) implies that
(2) (at 6V, 7t E K and M e V) => goat) = Ag(x).
Let x e L(G) and A, N be elements of K‘ such that M E V, 1': EV. Then

gm), —_ g(;7" M) —_ 21'A AM)

by (2) and hence % gox) = % g(A’x). Thus an extension h of g to L(G) is

defined by writing h(x) = %g(M) for all A such that MeV. Clearly I: is
continuous. We show that
(3) (x E L(G) and A E K) => 1104:) = M(x).
Let k’eK“ be such that )1e and A'MEV. Then
1 , 1 o 1 ,
110%) = 730 M) = ,7 M“) = Riga x) = M0!)-
Let x, y be in L(G). Then, by Proposition 4 of § 4, no. 3,
ll(it) + My) =A lim A‘“I"1(it(>~h(x))klt(>~h(y)))
= A6i133“x-‘o-1(¢(h<m)¢(h(zy))).
For |A| sufficiently small, Ax eV and Ag 6 V and hence the above expression
is equal to
”35;.” rlt"(f(4>(kx))f(4>(ly)))
= “£10 PW“ °f)(4>(M)<I>(7\y))
= “£13” k“§(¢"(4>(>\~)4>(1y)))
= ”£132“,h(’\“¢“(¢(l~)¢(7‘y)))
= M‘siimw l"<l>"(¢(M)¢(7‘y)))
= h(x + y).
Thus [1 is continuous and linear, hence g = h|V is analytic, hencef is analytic
on ¢(V) and hencef is analytic (§ 1, no. 10).
COROLLARY 1. Let G be a topological group. There exist: on G at most one analytic
manifold :tructurz over R (rcsp. Qp) compatible with the topological groufl :tmclur:
on G.

338
 com-muons MORPnIsns § 8.1

This follows immediately from Theorem 1.

DEFINITION 1. A Iopologioal group G is called a real (rap. p-adio) Lie group if
there exists on G a real (resp. p—adio) Lie group :truotur: compatible with {tr topology.
That structure is then unique and we can speak of the dimmrion of such a
group. If G and H are two such groups, every continuous morphism of G
into H is analytic.
COROLLARY 2. Zzt G be a topological group and V on open neighbourhood of :.
Suppose that V ha: an analytic manifold structure which maker it into a real (rap.
p-adio) Lie group germ. Then G i: a real (resp. p—adio) Liz group.
Let geG. There exists an open neighbourhood V’ of c in G such that
V’ U gV’g‘1 C V. The mapping 1: r—> gog—1 of V' into V is a continuous and
hence analytic morphism of the Lie group germ V’ into the Lie group germ
V. It then sufiices to apply Proposition 18 of § 1, no. 9.
Remarks. (1) Theorem 1 and its corollaries are no longer true if R (resp. Q,) is
replaced by C (Exercise 1).
(2) Let G be a topological group. It can be shownT that the following con-
ditions are equivalent: (a) G is a finite-dimensional real Lie group; (b) G is
locally compact and there exists a neighbourhood of 5 containing no subgroup
distinct from {a}; (c) there exists an open neighbourhood of: homeomorphic to
an open ball of a space R”. (For a much less difficult result, cf. Exercise 6.)

PROPOSITION 1. Let G, G’ be topologicolgmup: andfa continuous morphixm ofG into

G’. Supper: that one of thefillowiag three mm holds:
(a) G is a real Lie group and G' is a p—adr'o Lie group;
(b) G is a p-odio Lie group and G’ is a real Lie group;
(c) G 1': a p-adio Li: group and G’ is a p’-adio Lie group with p aé p’.
Thenf is locally constant.
Care (a). Let Go be the identity component of G. Thenf(Go) is a connected
subgroup of G' and hencef(G0) = {e} and Go is open in G.
Care (b). Let V’ be a neighbourhood of o in G’ such that every subgroup of
G’ contained in V’ reduces to {z} (§ 4, no. 2, Corollary 1 to Theorem 2). There
exists a neighbourhood V of 2 in G such thatf(V) C V’. Then there exists an
open subgroup G1 of G such that Cl C V (§ 7, no. 1, Proposition 1). Then
G, = z .
f(Caie (c){.}By § 7, Theorem 4 and the Corollary to Proposition 8, there exists
a neighbourhood V' of: in G’ such that, for all x' e V’ — {z}, x”" does not tend

TSee for example D. Montgomery and L. Zippin, Topologioal lraurfizrmatr'on

groups, Interscience tracts in pure and applied mathematics, no. 1, Intencience
publishers, New York 1955 (in particular pp. 169 and 184-).

339
ll! LIE GROUPS

to e as n tends to +00. There exists a neighbourhood V of e in G such that

f(V) C V’. By § 7, rI'l‘teore'n'l 4- and Proposition 9, there exists an open sub-
group Gl of G such that G1 C V and such that, for all x 6 G1, x’” tends to e as
It tends to +00. Thenf(G,) = {e}.
2. CLOSED SUBGROUPS
THEOREM 2. Let G lie ofinile—dimeru-ional Lie group over R or Q,. Every closed sub-
group of G is a Lie subgroup ofG. More generally, let U be asymmetric open neighbour-
hood of r in G and H a non-empty closed .mbxpaee of U ml: that the condition: x e H, .
y e H undxy“ e U implyxy" EH. Then H i: a Lie .mbgmupgerm q.
Let b be the Lie subalgebra tangent to H at e (§ 4, no. 5, Definition 2).
There exists a Lie subgroup germ HD ofG with Lie algebra b and contained in
H. We prove that He is open in H with the topology induced by that on G.
This will prove that H is an analytic submanifold of G and will establish the
theorem.
There exist a vector subspace I supplementary to I; in L(G), symmetric open
neighbourhoods V,, V2 of zero in I) and t respectively and an exponential
mapping (1) of G defined on V1 + Va and possessing the following properties:
(a) the mapping (a1, a2) t—> ¢(al)¢(ag) is an analytic isomorphism of
V1 x V; onto an open subset V of G;
(b) <l’(V1) c Ho;
(c) V2 C U.
We shall prove (and this will achieve the proof) that there exists an open
neighbourhood V; of 0 in V2 such that H n (¢(V1)¢(V;)) = ¢(V1).
Suppose that the assertion is false. Then we can find a sequence (x,,) in V1
and a sequence (y,,) in V3 — {0} tending to 0, such that ¢(x,l)¢(y,,) E H for all
n. Then My.) e H by (c)-
If K = Q,” it can be assumed that V2 is an additive subgroup of t and that
(HIM) = Na)” for all a 6V2 and allpez. Then (“Ag/1) 5H for all A E Z and
hence by continuity for all A E 2,. The mappingfz A >—> (“Ag/1) of 2,, into G is
analytic and takes its values in H and (Tof)(l) = y}. Hence y, e b, which is
absurd. The theorem is therefore established in the case of Q,.
If K = R, it can be assumed that V2 is connected and that 3],, belongs to
iv, — {0}. By taking a subsequence of (3],) ifnecessary, we can find a sequence
(An) nfnon-zero scalals such that M‘yn tends to an element y ol'VZ — {0}. The
sequence (An) tends to 0. Let A e R be such that Ay e i; we prove that
exp(Ay) s H. It can be assumed that AA; 1y" 51V, for all n. Let If" e Z be such
that |A — knAnl tends to O. For n sufficiently large, (A — knAn)A;1y,, E iVa and
hence knyn e n. Hence exp(hy,,) E H for h an integer and 0 < |lx| S 1a (as is
seen by induction on M). Then
ExPW) — fig CXPWJ‘yn) = "11!: (“pm - mm‘y.) “Wing/0)
= lim exp k,,yn SH.

340
 CLOSED suncnoups § 8.2

Hence the mappingf: A »—> exp Ay, where Ag E }V,, mks its values in H and
(Tof) (l) = y. Hence y e I), which is absurd. The theorem is thus established in
the case of R.

Theorem 2 is no longer true if G is not assumed to be finite-dimensional

(Exercise 12).

COROLLARY 1. Let G' be a locally compact group, G a finite—dimensional Lie group

over R (resp. (1,) andfa continuous morphism ofG' into G. Ifthe kernel affix
discrete, G' is afinite-dimenrional real (resp. p-adic) Lie group.
There exists a compact neighbourhood V of e in G’ such that f | V is a
homeomorphism of V onto a compact subspace of G. IfU is a sufficiently small
open neighbourhood of e in G, the hypotheses of Theorem 2 are satisfied with
H = f(V) n U. Hence H is a Lie subgroup germ of G. Let W be the inverse
image of H underf | V. Then W is a neighbourhood ofe in G'. Let W be given
the analytic manifold structure transported from that on H by (f I W) ‘1. For
all z e G’, the mapping x »—>f(z):gf(z) " of G into C is analytic; hence there
exists an open neighbourhood W' ofe in W such that the mapping x' >—> zx’z‘1
ofW’ into W is analytic. By Proposition 18 of§ I, no. 9, there exists on G’ a Lie
group structure which induces on a sufliciently small open neighbourhood of e
the same analytic structure as W and hence the same topology as the initial
topology on G‘.

COROLLARY 2. Let G be a finite-dimensional Lie group over K, H a subgroup of G,

V on open neighbourhood of e in G and (Mthar afamily of analytic manifild: ooer K;
for all i e I, letfi be a K-analytic mapping ofV into M. such that

HnV={e[f.(x) =f,(e)firollieI}.
(i) IfK = C,Hi.raLiembgrouqG.
(ii) IfK i: afim'te extension #9,, and I ixfim'le, H i: a Lie subgroup ofG.
(i) Suppose that K = C. We consider G as a real Lie group. Then H is a
real Lie subgroup of G (Theorem 2). Let a e L(H). There exists a connected
open neighbourhood W of 0 in C such that exp Aa E V for all A E W. Let i e I.
Then fi(exp Aa) =fi(e) if AERn W. Hence fl(exp Aa) =f.(e) if AEW by
analytic continuation. Thus, exp Aa e H for A e W and therefore pa 6 L(H) for
all pt E C. Therefore H is a Lie subgroup of the complex Lie group G (§ 4, no. 2,
Proposition 2).
(ii) Suppose that K is a finite extension of 1,2,. We consider G as a Lie group
over Qp. It is finite-dimensional and Theorem 2 implies that H is a p-adic Lie
subgroup ofG. Since I is finite, 1:! MI is a manifold and it can be assumed that
the family (fi) reduces to a single mappingf. Let a e L(G). Let (I) be an expo-
nential mapping of G. Then f(¢(Ao)) = f(e) for A E Q, and [AI sufficiently

341
ll] LIE GROUPS

small. Since f is K—analytie, it follows that f((1) (NO) = f(e) for A e K and [M
sufficiently small. Hence (MM) EH for A e K and |).[ suffiicently small and
therefore pm 6 L(H) for all p. E K. The proof is completed as in (i).
Corollary 2 (ii) is no longer true if we omit the hypothesis that I is finite.

§9. COMMUTATORS, CENTRALIZERS AND

NORMALIZERS IN A LIE GROUP

In this paragraph, K is assumed to be of characteristic zero.

I. COMMUTATORS IN A TOPOLOGICAL GROUP

Let G be a topological group. We define the groups fiG,fiG, WC, . ..

and 610, 55G, 65G, . . . by the formulae
fie = G, D‘“G = (We, in)
6G = G, We = (G, 6G).
PROPOSITION 1. Let G be a topological group and A and B subgroup: of G. 77101
(H) = (fixfiLU—Afih (TA.
Let 4) be the continuous mapping (x,y) v—>x“y“xy of G X G into G.
Then <1:(A x B) c (A, B), hence (“K x E) c (K, 1?) and hence
(3, 1—5) C (K. E);
Ezoppfl't-e inclusion is obvious and therefore (A, B) = (K, IE). Clearly
D°A = D°A; assuming the equality DTA = m, it follows that

DMK = (17;, 13K) = (m,m) = (D‘A, D'A) = D‘“A

and hence fix = DTA for all i. The proof of the formula 52: = (fl is
analogous.
COROLLARY 1. {f G is Haurdarf thefollowing condition: are equivalent:
(i) G is solvable (resp. nilpotent) ;
(ii) D‘G_= {e} (resp._C‘G = {9})for .rzgficimtly large i.
D'G C D‘G, C‘G C (1‘0 and hence (ii) 2 (i). {1'} = {T} and hence (i) 3 (ii)
by Proposition 1.
COROLLARY 2. Let G be a Hmdorf topological group and A a .rubgroup If G. For
A to be solvable (resp. nilpotent, commutative), it i: necesxary and :uflia'ent that K
be :0.

342
 oomuu-m-rons IN A LIE amour § 9.2

This follows immediam from Proposition 1.

PROPOSITION 2. Let G be a topalugiml group and A and B .mbgnmpr qf G. If A is
cannected, (A, B) is permuted.
For fixed y in B, the set M, of (x, y) with x e A is connected (for the mapping
x ~—> (x,y) of A into G is continuous). 55M.” and hence the union R of the
My with y e B is connected. But (A, B) is the subgroup of G generated by R,
whence the proposition.

2. COMMUTATORS IN A LIE GROUP

PROPOSITION 3. let G b: a finite‘dimnsional Lie group and H1 and H2 subgroups
of G. Let b1, b2 and b be the Lie .mbalgebm: tangent at: to H1 H2 and (H1, H3)
respectively. Then [91, 92] C 9.
Let 116171, b Ebg- There exist an open neighbourhood I of 0 in K and
analytic mappingsf1, f3 of I into G such that

f1(0) =fa(0) = I, fz(1) C H1, fa(1) C H2,

(Tofu)l = 0: (Tofa)1 = 1’-
We write
f(7\, It) = (fi(7\)»fa(P-)) 6 (H1, H2) for h, it in I.
We identify an open neighbourhood of e in G with an open subset of K'
by means of a chart which maps a to 0. Then L(G) is identified with K'.
By §5, no. 2, Proposition 1, the expansion off (A, y.) as an integral series
about the origin is
m, u) = M4. b] + ””2”,” Wu
where a” 5K' (the terms in )t‘ and p.’ in the expansion off (A, p) are zero
becausefO, 0) =f(0, p.) = 0). We fix it. in I. Letting A tend to 0, we see
that
y.[a,b] + Z u’a,,eb.
I):
Since this is true for all p. e I, it follows that [a, b] E [7.
Remark. Even if H1 and H, are connected Lie subgroups of G, the Lie sub-
algebra of L(G) generated by [by [)2] is in general distinct from I).
PRorosrrmN 4. Let G be aflnile—dimmrional real or complex Lit gimp. Let A, B, C
be inlzgml subgroup: of G such that [L(A), L(G)] C L(C) and
[L(B): L(G)] C L(G)-
lf [L(A), 143)] C I1(0), ”W (A, B) C G. If [L(AL L(13)] = L(G): ”w"
(A, C.

343
m LIE GROUPS

Suppose that [L(A),L(B)] C L(C). The sum L(A) + L(B) + L(G) is a

Lie subalgebra of L(G). By considering an integral subgroup of G wnh Lie
algebra L(A) + L(B) + L(C), the problem is reduced to the case where
L(A) + L(B) + L(C) = L(G)
and G is connected. Then L(C) is an ideal of L(G). Suppose first that G is
simply connected. Then G is a normal Lie subgroup of G (§ 6, no. 6, Proposi-
tion 14). Let (b be the canonical morphism of G onto G/C. Then

[L(¢)(L(A)), L(¢)(L(B))] = {0}

and hence (MA) and MB) commute by the Hausdorff formula; therefore
(A, B) C C. In the general case, let G’ be the universal covering of G and
A’, B', C’ the integral subgroups ofG’ such that L(A') = L(A), L(B’) = L(B),
L(C’) = L(C). Then (A', B’) C C' and A, B, C are the canonical images of
A’, B’, C’ in G, whence (A, B) C C. On the other hand, (A, B) is the under-
lying set of an integral subgroup of G (§ 6, no. 2, Corollary to Proposition 4)
and its Lie algebra contains [L(A), L(B)] (Proposition 3). If

[L(A): L(B)] = L(C),

then (A, B) 2 c, whence (A, B) = c. '
COROLLARY. Let G be a finite-dimensional connected real or complex Lie group with
Lie algebra g. The subgroup: D‘G (resp. C‘G) are integral subgroup: with Lie
algebra: 9‘9 (resp. W9). If G i: simply connected, they are Lie subgroups.
The first assertion follows from Proposition 4- by induction on i. The second
follows from the first and § 6, no. 6, Proposition 14-.
PROPOSITION 5. Let G be a finite-dimemional real or complex Lie group and A an
integmLmbgmup of G. Then DA = DA. In particular, A i: a normal subgroup of
A and A/A is commutative.
Let a = L(A). Let Gl be the set ofg 6G such that
(Ad g)x E 1: (mod 9a) for all x e a.
Then G1 is a closed subgroup of G. Ify e a, then expy 5G1, by §6, no. 4-,
Corollary 3 (ii) to Proposition 10. Hence Gl contains A and therefore A.
Thus, for g e A, L(Int g) leaves a stable and therefore Int g leaves A stable;
more precisely, L(Int g) defines the identity automorphism of a/Qa and
hence Intg defines the identity automorphism of A/DA. This proves that
(A, A) C DA. With the real Lie group structure on G, A is a Lie subgroup
(§ 8, no. 2, Theorem 2); let I:- be its Lie algebra. Let G2 be the set of g e G
such that
(Ad g): E 1: (mod 90) for all x e b.

344
 comm'm'rons IN A LIE GROUP § 9.2

By the above, G, D A and hence Ga 3 K. Therefore, for g E K, Int g leaves

DA stable and defines the identity automorphism of K/DA. Hence DA 3 DK.

PROPOSITION 6. Support that K ix ultrametn'c. Let G b: a fizite-dimmrionol Lie

group. Let A, B, C be Lie .mbgvoup: of G such that [L(A),L(C)] C L(C),
[L(B) L(C)] C L(C). {f [L(A), L(B)] C L(C), time exist open subgroup: A’,
B’of,A B ml: that (A', B’) C C. If[L(A), L(B)] C L(C), then m}! 0pm rub-
groupx A’, B’,C of A B, C such that (A', B') =
Suppose that [L(A), L(B)] C L(C). As 1n the proof of Proposition 4, the
problem reduces to the case where L(C) is an ideal of L(G). Then, by re-
placing G by an open subgroup, it reduces to the case where C is normal in
G (§ 7, no. 1, Proposition 2). Let (b be the canonical morphism of G onto
G/C. Then
[L(<l>)(L(A)), 14¢) (L(Bm = {0}
By the Hausdorff formula, there exist open subgroups A’, B’ of A, B such
that ¢(A’) and ¢(B’) commute, whence (A’, B') C C. Suppose fi1rther that

[L(A),L(B)] = L(C)~
By Proposition 3, the Lie subalgebra tangent to (A’, B’) at 2 contains L(C).
Hence (A’, B') contains a Lie subgroup germ of G with Lie algebra L(C).
Therefore (A’, B’) is an open subgroup of C.
COROLLARY. Support that K it ultramelric. Let G b: a finite-dimensional Lie group
with Lie algzbm 9. There exist: an open :ubgroup G, of G such that, fir all i, D‘Go
(resp. CG“) {4' a Lie subgroup of G with Li: algzbm 9‘9 (resp. ($9).
(a) By Proposition 3 applied inductively, for every op'en subgroup G1 of
G and for all i, D‘Gl contains a Lie subgroup germ of G with Lie algebra
9‘9.
(b) Let G' be an open subgroup of G such that, for :< n, D‘G is a Lie
subgroup of G with Lie algebra 9‘9 By Proposition 6, there exist open sub-
groups H1, H2 of D"G’, such that (H1, H2) is a Lie subgroup with Lie algebra
9"“ 1g. Let G” be an open subgroup ofG' small enough for D"G' C H, A 1-1;.
Then D"“G” C (H1, Hz). The relations
DDG" C DUG’, D‘G” C D‘G’, . . . ,D"G" C D”G’, D"“G” C (H1, H,)
prove, using (a), that DO" 15, for l< n + l, a Lie subgroup of G with Lie
algebra 9‘9.
(c) There exists an integer p such that 9V9 = 9””9 : - - ~. By the above,
there exists an open subgroup Go of G such that D'Go is, for i < P, a Lie
subgroup of G with Lie algebra 9‘9. But, by (a), the same assertion remains
true fori > p since D’G0 D D‘Go fori > p.
(d) The argument is similar for the Cl.

345
Ill LIE GROUPS

3. CENTRALIZERS
Recall that two elements x, 3/ of a group are called permutable if (x, y) = e,
or (Int x)y = y, or (Int g)z = x; and that two elements a, II of a Lie algebra
are called permutable if [11, b] = 0, or (ad a) .b = 0, or (ad b) .a = 0, Let G
be a Lie group, x e G, a e L(G) ; x and a are called permutable if (Ad x) . a = a,
that is ifxa = are in T(G).
Let G be a Lie group, 9 its Lie algebra, A a subset of G and a a subset of 9.
Let ZG(A) (resp. ZG(a)) denote the set of elements of G which are permutable
with all the elements of A (resp. a). It is a closed subgroup of G. Let MA)
(rap. 3am) denote the set of elements of g which are permutable with all
the elements of A (resp. a). It is a closed Lie subalgebra of g.
PROPOSITION 7. Let G be a finik—dimemimal Lie group, 9 it: Lie algebra and a a
subset of 9. Then Za(a) i: a Lie subgroup qf G with Lie algebra 3,01).
This follows fmm § 3, Proposition 44 and Corollary 2 to Proposition 39.
PROPOSITION 8. Let G be a finite-dimensional real or complex Lie group, g it; Lie
algebra and A a subret bf G. Then ZG(A) i: a Lie subgroup qf G with Lie algebra
bn(A)'
Suppose that A consists of a single point a. Then 25(A) is the set of fixed
points of Int a; hence ZG(A) is a Lie subgroup of G and L(ZG(A)) is the set
of fixed points ofAd 0, that is MA) (§ 3, no. 8, Corollary 1 to Proposition 29).
The general case follows by means of § 6, no. 2, Corollary 3 to Proposition 1.
PROPOSITION 9. Le! G be a finite-dimenxiannl real 0r temple: Lie grbup, 9 it: Lie
algebra, A an integral subgroup of G and a = L(A). Then ZG(A) = Z501) a
MA) = M“) and ZG(A) is a Lie subgroup of G with Lie algebra Ma).
Let x e G. Then

x 5 26m) «> A c Zo({"})

o a c L(ZG({x})) (§ 6, Corollary 2 to Proposition 3)
¢> a C 3‘({x}) (Proposition 8)
er x e 2.3 (a)

and hence Z°(A) = Zu(a). Let u e 9. Then

’1 5 MA) ¢A C 29({u})
¢ a c L(Zu({"})) (§ 6, Corollary 2 to Proposition 3)
¢ ‘1 C a.((u}) (Proposition 7)
o u e 33(a)

and hence MA) = Ma). The last assertion then follows from Proposition 7
or Proposition 8.

346
 NILPOTENT LIE GROUPS §9.5
4. NORMALIZERS
Let G be a Lie group, 9 its Lie algebra, A a subset of G and a a subset of g.
In this section, NG(A) will denote the set ofgeG such that gAg‘1 = A. It
is a subgroup of G, which is closed if A is closed. nu(a) will denote the set of
xe 9 such that [x, a] C 0 (cf. Chapter I, § 1, no. 4). It is a subalgebra of g,
which is closed if a is closed. N501) will denote the set of geG such that
gag‘1 = a.
PROPOSITION 10. Let G be a flaw-dimensional Lie group, 9 its Lie algebra and a
a vector subspace of 9. 77m: NG(a) is a Lie subgroup of G M Lie algebra n,(a).
This follows from § 3, Proposition 44 and Corollary 1 to Proposition 39.
Pkovosmmz 11. Let G be a finite-dimensional real or complex Lie group, 3 its Lie
algebra, A an integral subgroup of G and a = L(A). Then NG(A) = N503) and
NG(A) is a Lie subgroup if G containing A, with Lie algebra n,(a).
The equality NG(A) = NG(a) follows from § 6, no. 2, Corollary 2 to Proposi-
tion 3. By Proposition 10, NG(A) is then a Lie subgroup of G with Lie algebra
“‘01). Hence NG(A) is closed. As Nu(A) 3 A, NG(A) D A.
COROLLARY. If a = n,(a), A is a Lie subgmup of G and is the identity component
afNo<A).
This identity component is a Lie subgroup with Lie algebra n‘(a) (Proposi-
tion 11) and hence is equal to A by § 6, no. 2, Theorem 2 (i).

5. NILPOTENT LE GROUPS
PROPOSITION 12. Let G be afinile-dimensional Lie gmup. For, L(G) to be nilpoteut,
it is necessary and sufieimt that G have a nilpateut open subgroup.
Suppose that G has a nilpotent open subgroup Go. By the Corollaries to
Propositions 4- and 6, no. 2, W‘L(G0) = {0} for sufficiently large 1'. Hence
L(Gn) = L(G) is nilpotent.
Suppose that L(G) is nilpotent. If K = R or C, the identity component
Go of G is nilpotent by the Corollary to Proposition 4, no. 2, and G0 is open
in G. If K is ultrametric, the Corollary to Proposition 6, no. 2, proves that
there exist an open subgroup G, of G, an integer i > 0 and a neighbourhood
V of e in G such that C‘Gl n V = {3}. Then, if G“ is a sufficiently small sub-
group of G1, C‘Gu C V, hence C‘Go = {e} and Go is nilpotent.
Let g be a nilpotent Lie algebra. The Hausdorfl series H(X,Y) corres-
ponding to g has only a finite number of non-zero terms and we know (Chap-
ter II, § 6, no. 5, Remark 3) that the law ofcomposition (x, y) »—> H(x, y) defines
a group structure on 5;. Suppose further that g is complete and normable.
Clearly the law H is a continuous-polynomial (Dah‘able and Analytic Mom'-
folds, R, Appendix). Hence 9, together with the law H, is a Lie group G,

34-7
 111 LEE onoups

said to be msooinled with 9. By §4, no. 2, Lemmas 2 and 3, L(G) = g. The

identity mapping 4: of 9 into G is an exponential mapping of.G such that
<b(Ax)zl>(A'x) = ¢((A + A')x) for all x e g, A e K, A' E K. Every Lie subalgebra
I) of g admitting a topological supplement is a Lie subgroup H of G and
L(H) = 9‘
PROPOSITION 13. Let G be afinite—dimemionul simply connected nilpotent Lie group
over R or C.
(i) expG is an isomorphixm of the Lie group associated with L(G) onto G.
(ii) Every integral subgroup of G L: a simply connected Lie subgroup of G.
Let g = L(G), which is nilpotent (Proposition 12). As two simply con-
nected Lie groups over R or C which have the same Lie algebra are iso-
morphic (§ 6, no. 3, Theorem 3 (ii)), it suffices to prove the proposition when
G is the group associated with 5. Then (i) and (i) follow from what was said
before the proposition.

PROPOSITION 14. Let G be afinite-dimensional connected Lie group over R or C,

(i) If G it nilpatent, exp“ i: ltale and :urjectioe.
(ii) I)" K = C and expo ix etale, then G1": nilpoteut.
Let G’ be the universal covering space ofG. Let (I) be the canonical morphism
of G’ onto G. Then exp; = (l; a expo. (§ 6, no.4, Proposition 10) and hence
(i) follows from Proposition 13 (i).
If K = C and exp is étale, then, for all x e L(G), and x has no eigenvalue
belonging to 2in(Z — (0}) (§ 6, no. 4, Corollary to Proposition 12). Applying
this to M, where A varies through C, it follows that all the eigenvalues of ad x
are zero and hence that ad x is nilpotent. Therefore L(G) is nilpotent (Chap-
ter I, § 4-, Corollary 1 to Theorem 1) and hence G is nilpotent (Proposition 12).

PROPosmoN 15. Let G be a finite-dimdonal connected uilpotent Lie group over

R or C and A an integral subgroup of G. Then ZG(A) ii the connected Lie subgroup
of G with Lie algebra h(L(A)).
By Proposition 9 of no. 3 it suflices to prove that ZG(A) is connected. Let
g e ZO(A). There exists x e L(G) such that g = exp x (Proposition 14). Then
Adg|L(A) = 1 (no. 3, Proposition 9), hence Adg”|L(A) = l for all n 62
and hence exp(ad we)|L(A) = 1 for all n 62. As the mapping

A >—> exp(ad Ax) |L(A)

of K into .‘Z’(L(A), L(G)) is polynomial, exp(ad Ax)|L(A) = 1 for all A 5K,
that is exp(Ax) E ZG(A) for all A e K.
PROPOSITION 16. Let G be u finite-dimmional nilpotent Lie group ooer R or C
and A an integral mbgroup g G distinct from G. Then NG(A) is a connected Lie
subgroup of G distinctfrom A.

348
 NILPOTENT LIE GROUPS § 9.5

N°(A) sé A (Algebra, Chapter I, §6, Corollary 1 to Proposition 8). By

Proposition 11 of no. 4, we need only prove that NG(A) is connected. Let
g eNG(A). There exists xeL(G) such that g = expx (Proposition 14). Let
E be the vector subspace of $(L(G)) consisting of the u e $(L(G)) such that
u(L(A)) C L(A). Then Adg"eE and hence exp(ad 11x) GB for all In 62.
Hence exp(ad M) e E for all A e K, that is exp(M) e N5(A) for all A E K.
PROPOSITION 17. Let g be a finite-diammianal nilpatent Lie algebra over K and
(so, 911-“5910 a Jamming Mm“ did“!!! 0f a such that so = a, 9.. = {0} and
[9, 9‘] C gulfar 0 S i < 7:. Let a1, 03,. . ., a, be vector subspaces ofg mix that
ear}! 9. is the direct sum of its intersection: with the a,. [at g be given the Hauxdmf
law of composition H. Let (I) be the mapping
(*15x2,-~~2"y)*"*1"*2" ""p
afa1 x a, x x aDintog.
(i) daixaby'eetioaa, x «3 x x a,onto 3;
(ii) 4) and ¢l> 1 are polynomial mappings;
(in) the mapping (m) mt» 1W) W) I) of (a: x “a x x «.92
:11 x a: x X a, irpalymmial.
The proposition is obvious for dimg = 0. Suppose that dimg > 0 and
that the proposition has been established for dimensions <dim g. It can be
assumed that 9,- 1 9e (0} and g,“ l is then a non-zero central ideal of 9. There
exists an indexj such that!) = a“, n a, yé {0}. Let 9’ = 9/1), 9 be the canonical
morphism of 9 onto 9’, g; = 0(9.) and a; = 0(a). Then (93, 9;, . . ., 9,1) is a
decreasing sequence of ideals of 9’ such that 93 = g', g; = {0},
[92 92] C 9"“
for 0 s i < n and each g: is the direct sum of its intersections with the a}.
Let (V be the mapping
(x§,x§,...,x,’,) HxQnfl Hz;
of a; X a; X -- - x a; into 9’. By the induction hypothesis, (b' is hijective
and ‘l”: ¢"1 are polynomial mappings.
Let x e 9. We write

(1) 4>"‘(9(l)) = (3":(*)sxi(*)w-,x£(x))~

Then
(2) 0(x) = x;(x) H 1:50;) H ~ . - H xxx).
Let [)1 be a vector subspace supplementary to I) in g, the sum of the a» with
k aé j and a supplement of l) in 0,. There exists a bijection n of 9' onto [)1
such that 0 a 1' = Id.“ For x e 9, we write

(3) C00 = 1106(3)) H WOW!» H ' - ’ H 7104100) e s.

(4) fl") = C(")‘1 "x = (-§(x))-*-
349
m LIE GROUPS

By (2) and (3), 0(§(x)) = 0(x) and hence y(x) e 9. Finally we write

(5) W‘) = (‘4(xi(5))ymsn(*l(#)) +y(x),---,n(x£(#r))) 6“: X X “11'

As y(x) is central in 9,

“41(0) = 100400) H "710400) H MM?» HW‘)

= t(x) H y(x) by (3)
=x by (4).

Hence ¢o¢=ld,. Nowlet (:r1,::,,...,::,,)ea1 x a, x x a, andwrite

x=rb(xl,x,,...,x,) =x1Hx,H--~Hx,.Then
9(t) = 0(a) H 902) H ~ -- H 90"),
hence x,’(x) = 6(a) for l S i < p and therefore
{(x) = xlnxzu H (n0(x,)) H Hx,
W) = x: - 119(31)-
Then by (5)
W) = (x1,...,7)0(x,) + X: - n9(*;).--.,xp) = Quorum)-
Hence 4; a (I) = Id“ x “p. This proves (i). As the Hausdorfl‘law is polynomial,
(I) is polynomial. By the induction hypothesis, tb"1 is polynomial; by fon‘nula
(1), the functions at; are polynomial, hence I: is polynomial (formula (3)),
y is polynomial (formula (4)) and kl} is polynomial (formula (5)). This proves
(ii). Assertion (iii) follows from (i) and (ii) and the fact that the Hausdorff
law is polynomial.
Example qf a nilpntml Lie group. Let G be the lower strict triangular subgroup
of GL(n, K). It is a Lie subgroup of GL(n, K) and L(G) C 91(11, K) is the
Lie algebra of lower triangular matrices with zero diagonal (§ 3, no. 10,
Proposition 36). By Chapter II, §4, no. 6, Remark, G is nilpotent. Suppose
henceforth that K = R or C. As G is homeomorphic to K”"““/2, G is simply
connected. The exponential mapping of L(G) into C is just the mapping

u»—>expu= 2—:

(§ 6, no. 4, Example). By Proposition 13, the exponential is an isomorphism

of the manifold L(G) onto the manifold G. Proposition 17 of § 6, no. 9 gives
the inverse bijection log. We give K’I a norm. By Spettral Them-its, Chapter I,
§4, no. 9, forgEG and Hg -— 1“ < 1,

log: = .2 “1,3“ (g —1)~

350
 NILPom-r LIB GROUPS §9.5

that is

(6) Iogg = Z “,2” (g — 1)».

But the two sides of (6) are analytic functions ofg for g e G and are therefore
equal for all g e G‘
PROPOSITION 18. Let k be a mmmulatiwfield. V a vector space qfifinitz dimension >0
over It and G a subgroup of GL(V) whose elements an unipotmt.
(i) Tim: exist: a non-zero element 17 ofV M}! that gv = vfor all g e G.
(ii) There exist: a basis B 0fV MIL that,fi1r allgEG, the matrix ofg with
respect to B 1': lower triangular and has all iLt diagonal demerits equal to 1.
(iii) The group G is nilpotent.
(:1) Suppose first that k is algebraically closed and that the identity repre-
sentation of G is simple. Let a, b be in G. Then

Tr(a(b—l))=Tr(ab—-l)—Tr(a—l)=0—0=O
for ab — 1 and a — l are nilpotent. As the vector subspace of $(V) gener-
ated by G is J(V) (Algebra, Chapter VIII, § 4, Corollary 1 to Proposition 2),
Tr(u(b — 1)) = 0 for all u e$(V) and hence b = 1. Thus G = {1}.
(b) We pass now to the general case. Let It be an algebraic closure of k,
V = V®k£ and G C GL(V) the set ofa® l for aeG. Let W (165?“ W’)
be the set of elements of V (resp. V) invariant under G (resp. G). Then
W’ =W®kk forW = 9Q}Ker(g— 1) and W' =‘Q}Ker(g— 1) @1. If
V1 denotes a minimal element in the set of non-zero vector subspaces of V
which are stable under 6, then V1 C W’ by part (a) of the proof; hence
W aé {0}, which proves (i).
(c) By induction on dim V, it follows from (i) that there exists an increas-
ing sequence (V1,V2, . . ., V") of vector subspaces of V which are stable
under G such that V,I = V and the automorphism group of V./V,_1 canonic-
ally derived from G reduces to {1} for all i (we make the convention that
V, = {0} for r < 0), This implies first (ii) and consequently (iii) (Chapter
II, §4, no. 6, Remark).

COROLLARY 1. Let G be a finite-dimensional connected real or complex Lie group.

F01 G to be m'lpatcnt, it is necessary and mflwimt that may element of Ad G be nil-
atent.
P If every element of Ad G is unipotent, Ad G is nilpotent (Proposition 18)
and hence G, which is a central extension of Ad G, is nilpotent. If G is nil-
potent, L(G) is nilpotent, hence ad 2: is nilpotent for all x e L(G) and hence
Ad(exp x) = exp ad is is unipotent; but every element of G is of the form
exp x for some x e L(G) (Proposition 14-).

351
ll'l LIE GROUPS

COROLLARY 2. Every analytic subgroup qL(n, K) consisting of unipnlenl elements

is a simply eonneeted Lie subgroup. ..
This follows from Propositions 13 (ii) and 18 (u) and the fact that the
lower strict triangular group is simply connected.

6. SOLVABLE LIE GROUPS

PROPOSITION 19. Let G be a finite-dimensional Lie group. For L(G) to be sollmble,
it is neeessary and sufieient that G possess a solvable open subgroup. _
The proof is analogous to that of Proposition 12 of no. 5.
PROPOSITION 20. Let G be a simply connected solvable Lie group qffinite-dimension
7: (mar R or C and 9 = L(G). Let (9,, 9P1, . . ., go) be a sequence of subalgebras
of g of dimensions n,n— l,...,0, sud: that 9‘“ is an ideal of 5. far
i = n, n — l, . . ., 1.1' Let G‘ be the integral subgroup of G corresponding to 9,. Let
x, be a vector of 5‘ not belonging to 5._,. Let it; be the mapping

(31. My - - -, N) H (“P 7‘1‘1) (exp 12%) - ~ - (”‘13 11"!)

of K‘ into G. 77151: (15,, is an isomorphism of analytic manifilds and ¢.(K‘) = G‘ for
all i. .
For n = 0 the proposition is obvious. We argue by induction on n. Let H
be the integral subgroup of G such that L(H) = Keen. By § 6, no. 6, Corollary
1 to Proposition 14, H and G,,_l are simply connected Lie subgroups of G
and, as a Lie group, G is the semi~dircct product of H by G,_1. Hence
1 ~—> expflxn) is an isomorphism of K onto H and by the induction hypothesis
the mapping
(Al, N, . . ., 1””) »—> (exp Am) (exp Mara) . . . (exp Append)

is an isomorphism of the analytic manifold K"‘1 onto the analytic manifold

G".l which maps K‘ X (0} to G‘ for i = 1, 2, . . ., n — 1. Hence the proposi-
tion.
PROPOSITION 21. Let G be a simply connected finite-dimensional solvable Lie group
over R or C and M an integral subgroup of G. Then M is a Lie subgroup of G and is
simply eonnected.
We continue to use the notation n, g, 91, x., (I: of Proposition 20 but impose
on the x! the following supplementary condition: let i, > i,_1 > - - - > i1
be the integers i such that L(M) n 9, 7e L(M) n 9.4; then we take

at.“ E L(M) n g“

1 Such a sequence exists by Chapter I, § 5, Proposition 2.

352
 sownu: LIE (mom's § 9.6

for I: = l, 2, . . .,p. By induction on n, it is easily seen that (x5, x9”, . . ., x“)

is a basis of L(M). Let N be a simply connected Lie group such that there
exists an isomorphism h of L(N) onto L(M). Let

y» = h"(xi, , - - 4:111 = h"‘(*«.)«

By Proposition 20, the mapping

0‘1, 12: . - -: An) H (exp 11%)(“13 A2118) - ~ ~ (exp 112%)

is an isomorphism of the manifold K” onto the manifold N. There exists a
Lie group morphism 1 of N into C such that h = L(r) and 1(N) = M (§ 6,
no. 2, Corollary 1 to Proposition 1). Hence M is the set of elements of G of
the form
7((CXP M1) . - (up My,» = fimhMi) . . . WWI-(flee)
= export“) . . . exp()\,,ar.’).
Thus M = MT) where T is a vector subspace of K”.
Pkorosmon 22. Suppose that K = R or C. let V be a finite-dimensional vector
space and G a [Wheeled solvable subgroup qf GL(V). Suppose that the identity repre-
xentation of G is simple.
(i) IfK = R, thendimV < 20ndGireomrnutalive.
(ii) IfK = C, then dimV =1.
(i) Suppose that K = R. Then the closure H of G in GL(V) is a solvable
connected Lie subgroup of GL(V) (no. 1, Corollary 2 to Proposition 1).
Hence L(H) is solvable (Proposition 19). The identity representation of
L(G) is simple (§ 6, no. 5, Corollary 2 to Proposition 13). Hence dimV s 2
and L(G) is commutative (Chapter I, § 5, Corollaries l and 4- to Theorem 1).
Hence G is commutative.
(ii) Suppose that K = G. Let W be a. minimal element among the non-
zero real vector subspaces of V which are stable under G. The complex
vector subspace of V generated by W is equal to V since the identity repre-
sentation of G is simple. By (i), GIW is commutative. Hence G is commutative.
Therefore every element of G is a homothety (Algebra, Chapter VIII, §4,
Corollary 1 to Proposition 2), so that dim V = 1.
COROLLARY. Let V be a complex uemr rpaee offinite dimension >0 and G a con-
neeled solvable rubgroup q/"GL(V).
(i) There exim a non-zero element 11 0f V such that gr 5 Cufbr all g e G.
(ii) There exists a bari: B of V ruch that, far all geG, the matrix if g with
respect to B it luwer triangular.
Let V1 be a minimal element among the non-zero vector subspaces of V
which are stable under G. By Proposition 22 (ii), dim V1 = 1. This proves
(i). By induction on dim V, it then follows that there exists an increasing

353
 m LIE GROUPS

sequence (V1, V3, ‘ . ., V,) of vector subspaces of V which are stable under
G such that dimVHllV. = 1fori< h and V,I = V; hence (it).

7. RADICAL OF A LIE GROUP

PROPOe‘ION 23. Let G be a finite-dimensinul rear or complex Lie gnmp, r the
radical of L(G) (Chapter I, §5, Definition 2) and n the largest nilhotent ideal
of L(G) (Chapter I, §4, no. 4). Let R (resp. N) be the integral rubgnmp qf G
with Lie algebra t (resp. n). Then R (resp. N) is a .mluahle (reap. nilpotent) Lie ‘
subgroup of G, which it invariant under every cantinuaus automorphirm of G. Every
raluable (resp. nilfiotent) connected normal subgroup af G is contained in R (resp. N).
The group R is solvable (§ 6, Proposition 19). Suppose that K = R. Let
G’ be a connected solvable normal subgroup of G. Then G7 is a connected
solvable (no. 1, Corollary 2 to Proposition 1) normal Lie subgroup of G (§ 8,
no. 2, Theorem 2). Hence L((T) is a solvable ideal ofL(G), whence L(E') c r
and G’ C R. In particular, R C R, hence R is closed and therefore is a Lie
subgroup of G. Suppose that K = G. Let H be the underlying real Lie
subgroup of G. If t' is the radical of L(H), ir' is a solvable ideal of L(H),
whence r’ = ir'; hence I C t’ C t and, by the above, R is closed in H and
therefore in G; thus R is a Lie subgroup of G. Every connected solvable
normal subgroup of G is a connected solvable normal subgroup of H and
therefore contained in R. Hence we have proved for K = C and for K = R
that R is the largest connected solvable normal subgroup of G; therefore R
is invariant under every continuous automorphism of G. The proof for N is
completely analogous.
DEFINITION 1. Let G be a finite-dimensional real or complex Lie group. The radical
qf G it the largest connected rolvable normal mbgroup of G.
Remark. Even if G is connected there may be solvable normal subgroups of
G not contained in the radical of G.

PROPOSITION 24. Suppnre that K = R 01 G. Let G” G2 be two finite-dimmional

connected Lie gum/1:, R1 and R2 their radical: and (I) a mijective mmphirm of G1
into G,. Then (MR1) = R3.
By§ 3, no. 8, Proposition 28, L(tb) is surjective. Hence L((b) (L(R1)) = L(Rg)
(Chapter I, §6, Corollary 3 to Proposition 2). Let i be the canonical injec-
tion of R1 into G,. Then the image of (l) oi is R2 (§ 5, no. 2, Corollary 1 to
Proposition 1).

PROPOSITION 25. Suppuse that K = R or C. Le! G1, G; be finite-dimenrimml con-

nected Lie groups and R1 and R, their radicals. The radical afGl x G; 1': R1 x R,.
This follows from Chapter I, §5, Proposition 4.

354
 SEMI-SIMPLE LIE GROUPS §9.8
8. SEMI-SIMPLE LIE GROUPS
PROPOSITION 26. Let G be a finite-dimemiml connected real or complex Lie group.
Thefillcwing condition: are equivalent:
(i) L(G) is semi-simple;
(ii) the radical of G is {e};
(iii) every normal commutative integral subgroup of G is equal to {e}.
Condition (ii) means that the radical of L(G) is {0} and hence (i) c (ii)
(Chapter I, § 6, Theorem l). The equivalence of (i) and (iii) follows from § 6,
no. 6, Proposition 14.
DEFINITION 2. A connected real or complex Lie group is called semi-simple if it is
finite-dimem'onal and satisfies the conditions of Proposition 26.
Remark 1. Let G be a finite-dimensional connected real or complex Lie group.
If G is not semi-simple, G has a commutative connected Lie subgroup G’
which is invariant under every continuous automorphism, such that G’ 75 {e}.
For let u be the largest uilpotcnt ideal of L(G); then u eé {0} and the corre-
sponding analytic subgroup N is a Lie subgroup which is invariant under
every continuous automorphism of G (no. 7, Proposition 23); the centre G’
of N has the desired properties.
PROPOSITION 27. Let G be a finite—dimensional connected real or complex Lie group.
Thefollowing conditions are equivalent:
(i) L(G) is simple;
(ii) the only normal integral subgroups ofG are {e} and G andfirther G is not com-
mutative.
This follows from § 6, no. 6, Proposition 14.

DEFINITION 3. A connected real or complex Lie group is called almost simple if it is

finite-dimensional and satisfies the condition: ofProposition 27.
PROPOSITION 28. Let G be a simply connected real or complex Lie group. Thefollow-
ing conditions are equivalent:
(i) G is semi-simple;
(ii) G is isomorphic to the product of afinite number of almost simple groups.
If G is a finite product of almost simple Lie groups, L(G) is a finite product
of simple Lie algebras and is hence semi-simple. If G is semi-simple, L(G) is
isomorphic to a product of simple Lie algebras Ll, . . ., Ln. Let G; be a simply
connected Lie group with Lie algebra L" which is therefore almost simple.
Then G and G1 X - - - X G" are simply connected and have isomorphic
Lie algebras and are therefore isomorphic.
Lemma 1. Let G be a connected topological group, Z in centre and Z’ a discrete sub-
gmup of Z. Then the centre ofG/Z' is Z/Z’.
Let y be an element of G whose class modulo 2' is a central element of

355
m LIE (moors

G/Z'. Let 115 be the mapping gHgyg'1y‘1 of G into G. Then ¢(G) is con-
nected and contained in Z' and hence {>(G) = (I)({e}) = {2}. Therefore 11 E Z.
PROPOSITION 29. Let G be a semi-simple connected real or complex Lie group.
(i) G = (G, G). .
(ii) The centre Z of G i." dunete.
(iii) The centre t/Z i: {e}.
Assertion (1) follows from the Corollary to Proposition 4, no. 2 and Chap-
ter I,§6, Theorem 1. _
Assertion (ii) follows from §6, no. 4, Corollary 4 to Proposition 10 and
Chapter I, §6, no. 1, Remark 2.
Assertion (iii) follows from (ii) and Lemma 1.
PROPOSITION 30. (i) Let g be a semi-simple real or complex Lie algebra. Then
Int 9 is the identity component ofAut 9.
(ii) Let G be a moi-simple connected real or complex Lie group. 7725 adjoin! group
of G i: the identity component qllt L(G). It: tentre reduces to the identity element.
Every derivation of g is inner (Chapter I, §6, Corollary 3 to Proposition 1)
and hence L(Int g) = L(Aut g), which proves (i). The first assertion of (ii)
follows from (i). The second follows from Proposition 29 (iii) and §6, no. 4,
Corollary 4 (ii) to Proposition 10. .
Remark 2. Let g be a complex semi-simple Lie algebra and 90 its underlying
real Lie algebra. Then Aut(g) is open in Aut(go), for Int(go) C Aut(9).
PROPOSITION 31. Let G be a Jimply connected finite-dimensional real or complex Lie
group and R it: radical. There exists a semi-:imple simply connected Lie subgroup S of G
rueh that G, a: a Lie group, it the semi—direct product of S by R. lfS’ i: a :emi-rirnple
integral rubgraup of G, there exirtr x in the nilpatent radical of L(G) such that
(Ad exp x)(S’) c S.
This follows from §6, no. 6, Corollary 1 to Proposition 14 and Chapter I,
§ 6, Theorem 5 and Corollary 1.
Lemma 2. Let G be a group (reap. a topological group), G’ a normal subgroup of G,
V a finite-dimenrional vector space over a commutative field k (resp. over K), p a
linear representation (resp. a cantinuou: linear reprexentation) (JG on V and p' = FIG'.
(i) {f p is semioimple, p' i: semi-simple.
(ii) If p’ i: semi—simple and every finite-dimennlonal linear k-reprerentation (resp.
cantinuoru linear K-reprexenlation) of G/G’ (resp. G/G’) is semi-simple, then (a i:
:emi-rimple.
Suppose that p is semi-simple; we prove that 9’ is semi-simple. It sufiices
to consider the case where p is simple. Let V’ be a minimal non-zero sub-G’-
module of V. For all g e G, p(G’)p(g)V’ = p(g)p(G')V' = p(g)V', in other

356
 SEMI-SIMPLE LIE GROUPS §9.8

words p(g)V’ is stable under 9(G’) ; if V” is a sub-G’-module of p(g)V’, fllen

p(g)‘1V" is a sub-G'—module of V' and hence V' is equal to {0} or [7(g)V’.
Thus, for all g e G, p(g)V’ is a simple G’-module. But I; p(g)V’ is a non-zero
sub-G—module of V, whence V = ’26 p(g)V’. Hence 9’ is semi-simple.
Suppose that p’ is semi-simple. Let W be a non-zero sub-G-module of V.
As 9’ is semi-simple, there exists a projectorf0 of V onto W which commutes
with p’(G). Let E be the set of f e J?(V, V) which commute with 9(G’),
which map V into W and whose restriction to W is a homothety; for fe E,
let ot(f) denote the ratio ofthe homothetyf|W. Thenfa e E and fife) = 1.
Clearly a is a linear form on E. Let F = Ker on, which is a hyperplane of E.
ForfeE and geG, we write a(g)f= 9(g) ofo Mg)"; then c(g)fmaps
V into W and its restriction to W is the homothety of ratio 0t(f ); if g' e G’,
then
0(3))“ 9(3’) = 9(3) °f° 9(3)" ° 9(g')
= 9(g) °f° 9(g"g’g) 0 Me")
= 9(g) ° p(s":’g) °f° 9(g“)
= 9(3') ° 9(g) °f° 90;")
= 9(5’) °c(g)f-
Hence a(g) fe E. Therefore a- is a linear k-representation (resp. continuous
linear K-representation) of G on E leaving F stable. Then a(g) = Id, for
g e G’ and hence for g e 5 in the topological case. Suppose that every finite-
dimensional k-linear representation (resp. continuous K-linear representation)
of G/G’ (resp. 0/5) is semi-simple. Then there exists in E a supplement of
F which is stable under G. In other words, there existsf e E'such that oc(f ) = l,
which is invariant under G. Then f is a. projector of V onto W, and, for
geG, 9(g) ofo 9(g“) =f, that isfcommutes with 9(0). Thus 9 is semi-
simple.
THEOREM 1. Let G be a finite-dimensional real or complex Lie group, G, itx identity
component, R it: radical and l.‘ the radical of L(G) ; suppose that GIG“ iyfim'le. Let
p be a finite-dimensional analytic linear representation of G. The following condition;
are equivalent:
(i) 9 ix semi-simple;
(ii) 9|Go is semi-simple;
(iii) pIR ix :emi-n'mple;
(iv) L(p) i: semi-rimple;
(v) L(p) It is “mi-simple
(i) 0 (ii) by Lemma 2 and Inkgmtian, Chapter VII, §3, Proposition 1.
(ii) ¢> (iv) and (iii) o (v) by §6, no. 5, Corollary 2 (0 Proposition 13.
(iv) <> (v) by Chapter I, §6, Theorem 4.
COROLLARY I. III p, 9,, p, befinitevdimmrional semi—simple analytic linear repre-

357
"I LIE GROUPS

are
irritation: of G and n an integer 20. Then 9, ® 9,, T“p, 5’9, /\"p (Appendix)
:emi—rimple.
The semi-simplicity of 9, ® 92 follows from Theorem 1 and Chapter I,
§6, Corollary 1 to Theorem 4. The semi-simplicity of T"p, 5"p, A”p follows
from the semi-simplicity of pl ® 9,.

We shall see later that, if I; is a. commutative field of characteristic 0,

I‘ a group and p, and 92 finite-dimensional semi-simple linear k-repre-
sentations of 1", then pl (8 p, is semi-simple.

COROLLARY 2. Let p be a finite-dimensional :emi-ximple analytic linear representa-

tion of G on a vector :pace V, S the symmetric algebra of V and 5° the Jubalgebra of
S comirting of the element: invariant under (Sp) (G). Then 5° i: a finitely generated
algebra.
This follows from Theorem 1, Chapter I, §6, Theorem 6 (a) and Corn-
mutatioe Algebra, Chapter V, § 1, Theorem 2.
COROLLARY 3. Let G be a real or complex Lie group and Go it: identity component.
Suppore that G" is semi—simple and that G/Go isfinile. 77ml eoeryfinite-dimeruional
analytic linear representation of G is semi-simple.
Pnorosmon 32. Let G be a finite—dimensional connected real Lie group. Suppose
that L(G) i: reductive. Thefollowing condition: are equivalent:
(i) G/fiG is compact;
(ii) (resp. (ii‘)) every finite-dimensional analytic linear representation of G on a
complex (reap. reaD vector space i: semi-simple.
(i) => (ii’): Suppose that G/B‘G is compact. Then every continuous
linear representation of G/fi‘G on a finite-dimensional real vector space is
semi-simple (Integration, Chapter VII, §3, Proposition 1). Let p be a finite-
dimensional analytic linear representation of G on a real vector space. Then
plD‘G is analytic, D‘G is semi-simple (Chapter I, §6, Proposition 5) and
hence plD‘G is semi-simple (Corollary 3 to Theorem 1). Hence 9 is semi-
simple (Lemma 2).
It can be seen similarly that (i) => (ii).
(i) => (ii): Suppose that G/D‘G is not compact and hence isomorphic to
a group of the form R” x T“ with p > 0 (§ 6, no. 4, Proposition ll (ii)).
Then there exists a surjectivc morphism of G/fiG onto R and hence a sur-
jective morphism of G onto R. The mapping

gm“) = (:(g) (1))

is an analytic linear representation of G on R2 which is not semi-simple, for

358
 INFINITESIMAL Amazonian-115m § 10.1

the only l-dimensional vector subspace of R2 which is stable under a-(G) is

R(0, 1).
It can be seen similarly that (ii) a (i).
PROPOSITION 33. Let G be afioite-dimem‘ionol complex Lie group whore number of
connected component: is finite, p ofinite—dimemional analytic linear representation of G
and G’ on integml rubgrlmp of the real Lie group G much that L(G') generates L(G)
over C. Then, for p to be semi-simple, it it nemsary and .mflieient that p|G’ be semi-
Jimple.
Let p’ = p|G'. For 9 (rap. (3') to be semi-simple, it is necessary and suffi-
cient that L(p) (rap. L(p’)) be semi-simple (Theorem 1). Let V be the space
of 9. For a vector subspace of V to be stable under L(p) (L(G)), it is necessary
and sufficient that it be stable under L(p')(L(G'))‘ Hence the proposition.

§10. THE AUTOMORPHISM GROUP OF A LIE

GROUP

In this paragraph, K is assumed to be of characteristic zero.

I. MINITESIMAL AUTOMORPIIISMS
Lemma 1. Let G be a Liep and a a oeetorfieldon G. For allg 6G, let
Mo) = Mg)!1 6 L(G).
Thefillowing condition: are equivalent:
(i) a i: o homomorphirm of the group G into the group T(G) ;
(ii) for «11g, 47 in G mtg!) = «(5)! + g’;a(g)
010]?” 4’13: g m G 9(a)) = He) + (Ad 5W!)-
Condition (i) means that, for all g, g’ in G, we have In the group T(G):
l5(g)gfl(g‘)g’ = News!
0!‘
5(g)((Ad g)fl(g'))g£' = 9(gg')gg'~
But the product of 6(g) and (Ad g)B(g’) in T(G) 15 just the sum of [3(g) and
(Ad g){3(g) in L(G) (§2, no 1, Proposition 2). Hence (i)¢> ¢>(iii) On the
other hand condition (ii) may be written B(gg’)g = [3(g)gg’ + gB(g ) g , or
15(gg'= 3(3) + (Adgmg')
and hence (ii) 9 (iii).
DEFINITION 1. Let G be a Lie group. An irfiniflesimal automrphism g" G is my
analytic vetlurfield on G satixfi/ing the condition: 4fLemmo 1.
Lemma 2. Let K' be a non-discrete olored rub/field of K, A a K'-manifold, B and C
K-manifaldy and f a K’dmulytio mapping of A X B into C. Suppoxe that, fir all

359
Ill LIE GROUPS

4 EA, the mapping b Hf(a, b) q into C is K-analytic. Then, for all lETA,
the mapping :4 >—> (Tf) (I, u) «f TB into TC is K-analym. . I _
We fix teTA and write g(u) = (Tf)(t,u). Clearly g 15 K-analytlc. By
Diflerentiable and Analytic Manifolds, R, 5.14.6, it suffices to prove that the
tangent mappings to g are K-linear. It can be assumed that A, B, C are open
neighbourhoods of O in complete nonnable spaces E, F, .G over K , K, K
and that t is tangent to A at 0. We identify TA, TB, TC, th A x E, B x F,
C x G and twith an element of E. Then for all (x,y) 6TB = B x F.

3061/) = (f (0: x), (l)(0: ’00) + (s)(0,X)(y))-

We identify T(B x F) with (B x F) x (F x F) and T(C x G) with
(G x G) x (G x G). Then, forall
((#4), (h,k)) eT(B x F) = (B x F) x (F x F),

(Tg)((x,y), (M0) = ((a; b), (v, d», where

'1 =f(0, x),
b = (Dxf)(0:x)(t) + (Daf)(0,x)(y). c = (Daf)(0:")(h>,
d = (132s)(0: x) (t, h) + (DasXOJmh) + (DaO, x)(k)-
We now fix (x, y) e B X F. We need to prove that the mapping (h, k) r—v (c, d)
of F x F into G x G is K-linear. As the mapping 5: »—>f (0, x) of B into C is
K-analytie, the mappings

U“: k) H (Daf) (0, *) (h), (h, k) H (DaDaf) (0, My, h).

(h: k) H (s)(0, x)(k)
are K-linear. On the other hand
(Daf)(0’*)U,h) =Aeg§§~o A"‘((Daf)()~!,x)(ll) * (Daf)(0)x)(h))
and, for fixed A, the mapping a: n—>f (M, x) is K-analytic, so that the mapping
h »—> (Dzf) (M, x) (h) is K-linear.

PROPOSITION 1. Let K’ be a non-discrete closed subfield if K, G a Lie graup aver

K, V a munlfuld over K’ and (1/, g) v—> ug a K'-analytie mapping of V x G inta G.
Suppose that,fizr all u E V, the mapping g »—> vg qf G into G 1': an automorphirm nfG.
Let a he an element 0fV mph that eg = gfor all g6 G and a ET,(V). Then Ilze
1mmfield g ~—> ag an G i: an bwnilesimal automorphism of G.
For I) e V, gI EG, g; e G, u(g1g2) 2 (lg!) (vgn). Hence, for 111 ETC,
u2 ETG, (101,142) = (aul)(au3) (§ 2, no. 1, Proposition 3). In particular, the
mapping gv—nzg of G into TG is a group homomorphism. On the other
hand, this mapping is analytic by Lemma 2.
PROPOSITION 2. Le! G be a real or complex Lie gmup and a an bfinimimal auto-

360
 mrmrrEsnnAL AUTOMORPr-nsux § 10.1

morphism 11f G. Time nix-t: a low of analytic operation (A, g) v—> (”(g) of K on G
with thoflllowing proportion
(1) if D i: the emaciated law of infinitesimal operation, than D0) = a;
(2) for all 7. e K, 4),, e Aut G.
(a) For all p. > 0, let KI, he the open ball of centre 0 and radius y. in K.
For all g e G, let i, be the set of analytic integral curves f of a defined in a
ball KI, and such that f (0) = g. By Diflmntiablc and Analytic Manifoldx, R,
9.1.3 and 9.1.5, Fig is non-empty and two elements of .97, coincide on the
intersection of their domains of definition; let p.(g) be the least upper bound
of the numbers it. such that there exists an element of 3“, defined in K“;
there exists a unique element 0119', defined in Km); we denote it byf,.
(b) Let g1, g2 be in G,fi 5.97,],f369’” withfi andf, defined on the
same ball Kn. Then f1fa: Ku—> G is analytic and (f1f2) (0) = g,g2. On the
other hand, for all A e K”,

(Th(flfz»l = (TxflllzfaO‘) +f1()‘l~(TAfa)l (§ 2, Proposition 7)

= a(fi()\))f,(l) +f1(7\)a(f2(l))
w((f1fz)(7\)) (Lmum 1)
and hencefzfaeym, The Proves that u(g:g2)> inf<1"(£1l:1-L(gzl)
(c) By Diflmntioblc and Analytic Manifiln, 9.14and 9.1.5, there exists
a neighbourhood V of e in G such that 5—
— infy.(g) > 0. Let It EG and C be its
connected component. For all [1’ EC, p.(li’)> inf(o', p.(h)) > O by (b).()
the other hand, the functions f” , where h' E C, take their values in C. By
Dlfirentiable and Analytic Manifolds, R, 9.1.4- and 9.1.5, p. = +00 in C and
finally p. = +00 in G. Then let fy()‘) = ¢‘(g) for all geG and all teK.
By Difl’mntiablc and Analytic Manifolds, R, 9.1.4 and 9.1.5, the mapping
(A, g) >-> (1)“ g) is a law of analytic operation of K on G. Clearly, if D is the
associated law of infinitesimal operation, D(1) = on. By (b),

(“($1.32) = ‘l’x(gil‘l>>.(£2)

forall AeK,g1 56,57, EG.

PROPOSITION 3. Suppose that K ix ultramctn'r. Let G be a compact Liz group and at

an infinltcrimal autornorphirm of C. There can}! an open subgroup I of K and a law of
analytic operation (A, g) >—> 4;“g) of I on G with thefollowing proportion-
(l) ifD is the arrociated law ofinfinitm‘mal operation, than D(l) = a;
(2) fir all )r e I, (it), 6 Au! G.
As G is compact, there exist an open subgroup I’ of K and a law ofanalytie
operation (t, g) »—> ¢A(g) of 1’ on G with property (1) of the proposition

361
III LIE GROUPS

(§ 4, no. 7, Corollary 2 to Theorem 6). We write ¢A(g) =fi,(k) for 161’

and g E G. Then, for g,, g, in G and l e I’,

(Tz(fi.fix))1 = (Tmahmx) +fn(l)-('I'xfi,)l

«(12.0))1220) +.fn()‘)“(fira(7‘))
a(fi1()‘)fi2()‘))
and (fufa.)(0) = gig: =fm(0). Hencefuagm) =fr.(k)fia(l) for
(gs, ea, 1)
in a neighbourhood of (g,,g.,0) (Dgfirmtiable and Analytie Manifolds, R,
9.1.8). As G is compact, there exists an open subgroup I of I' such that
fun”) = fi‘()\)f,,()x) for all gl 5 G, g, E G, A e I. In other words, 4),, 6 Am G
fork 1.
Lemma 3. Let G and G’ be Lie groups and (b a homomorphism ofG into Aut(G’).
Letf(g, g') = (d;(g))(g’) for g E G, g' E G’. Consider lhefillowing conditions:
(i) f is analytie;
(ii) f is analytic in a neighbourhood 4y” (ea, :54) ;
(iii) for all g’ e G', the mapping g »f (g, g’) is analytie.
Then (i) «1» ((ii) and (iii)). If G is eormuteal, (i) ‘9 (ii).
Clearly (i) implies (ii) and (iii). Let go 5 G,'g3 e G’. For all g e G, g’ E G’,

f (gee, g’gf») = (<1>(e)<l>(go))(g’g6) = ¢(g)(¢(go)g’)>¢(g)(¢(go)g6)-

This proves the implication ((ii) and (iii)) a (i). Finally, if G’ is connected,
G' is generated by every neighbourhood of ear and hence (ii) a (iii).

2. THE Auromonm-nsm amour or A m: GROUP (REAL on

COMPLEX CASE)
In this no., we assume that K = R or C.
Lemma 4. let H be afinite—dirnensional simply connected Lie group.
(i) For all u eAut L(H), let 6(a) be the unique automorphism of H such that
L(0(u)) = 11. Then the mapping (a, g) ~—>0(u)g of (Aut L(H)) x H into H is
analytic.
(ii) Let N be a Lie subgroup of H and Aut(H, N) the set of v E Aut H surh that
0(N) z: N. Then 0"(Aut(H, N)) is a Lie subgroup afAut L(H).
(iii) Suppose that N is discrete and normal, so that the Lie algebra y” C = H/N
is identified with L(H). For all w e Aut G, kt 1;(w) be the unique automorphirm ‘Zf
H such that L(n(w)) = L(w). Then the mapping n is an isomorphism of the group
Aut G onto the group Aut(H, N).
To prove (i), it suffices, by Lemma 3 of no. 1, to verify that the mapping
(a, g) »—> 0(a) g is analytic in a neighbourhood of (Idum, e). There exists an

362
 AUTOMORPHISM GROUP or A LIE GROUP (REAL on courmx CASE) § l0.2

open neighbourhood B of 0 in L(H) such that t]; = expn|B is an analytic

isomorphism of B onto an open neighbourhood of e in H. There exist an open
neighbourhood U of Idm, in Aut L(H) and an open neighbourhood B' of
0 in L(H) such that U(B') C B. Then the mapping (a, g) n—> 0(u)g ofU x 443’)
into H is composed of the following mappings:
the mapping (mg) n—> (n, ¢“(g)) of U x 4103’) into U x B';
the mapping (a, x) »—> u(x) of U x B' into B;
the mapping y ~—> t];(y) of B into G.
Hence this mapping is analytic.
Let p be the canonical mapping of H into the homogeneous space HjN.
Then 0'1(Aut(H, N)) is the set of u eAut L(H) such that
M0003) = 9(6): P(9(u“)g) = 9(6)
for all g e N. By §8, no. 2, Theorem 2 and Corollary 2 to Theorem 2, this
proves (ii).
Suppose that N is discrete and normal. Let w 5 Ant G. Then
L(fi WM) = L(YKW» = L(H)) = MW “1!)
hence p a 71(w) = w o p and therefore n(w) eAut(H, N). Clearly the mapping
1] ofAut G into Aut(H, N) is an injective homomorphism. This homomorphism
is surjective because 1:: H —> G is a submersion.
Let G be a lowlly compact group and I‘ the automorphism group of G.
Recall that a topology .7“ has been defined on 1" (General Topolagy, Chapter X,
§3, no. 5). It is the coarsest topology for which the mappings vx—> v and
v »—> v" of I‘ into ?C(G; G) (space of continuous mappings of G into G with
the compact convergence topology) are continuous. The topology 3', is com-
patible with the group structure on I‘ (lac. cit.). For every compact subset
L of G and every neighbourhood U ofea in G, let N(L, U) be the set of ‘i> E I‘
such that Mg) egU and o"(g) egU for all ge L; then the N(L, U) form a
fundamental system of neighbourhoods of en. If G is generated by a compact
subset C, the topology .7, is also the coarsest topology for which the mappings
a >—> v|C and v~—> v"|C of 1" into KAC; G) are continuous (for every com-
pact subset of G is contained in (C U 0“)" for sufficiently large n). If K is
locally compact and V is a finite-dimensional vector space over K, the topology
.7, on GL(V) is just the usual topology.
THEOREM 1. Let G be a finite—dimensional Lie gum]: and Go its identity component.
Suppnre that G is generated by Go and afinile number qfelementr.
(i) 77m: exists an Ant G one and only one analytic manifold structure ram—fling
thefollowing canditz'an:
(AUT) for every analytic manifold M and every mapping f If M into Aut G, f is
analytic if and only g'fthe mapping (m, g) »—>f (m) g (.f M X G into G is analytic.

363
m LIE GROUPS

Suppose in the rest of the statement that Aut G has this structure.
(ii) Aut G i: afinite—dimensional Lie group.
(iii) The morphism (b: u »—> L(u) ofAut G into Aut L(G) is analytic.
(iv) If G is connected, 4: is an isomorphism of theLie group Aut G onto a Lie sub-
group of Aut L(G) ; this Lie rubgroup is equal to Aut L(G) if G is simply connected.
(V) Let a be the set of infinitesimal automorphism of G. Then a ir a Lie algebra
of vector field: and the law of infinitesimal operation associated with the mapping
(a, g) >—> u(g) of (Aut G) x G into G is an isomorphism ty‘L(Aut G) onto 0.
(vi) The topology of the Lie group Aut G is the topology .7}. .
(a) The uniqueness of the analytic structure considered in (i) is obvious.
(b) Suppose that G is connected. Let H be the universal covering space
of G, p the canonical morphism of H onto G and N = Ker p. We introduce
the notation 6, 11 and Aut(H, N) of Lemma 4. We transport the Lie group
structure of Aut L(G) to Aut H by means of 6. Then Aut H becomes a finite-
dimensional Lie group and Aut(H, N) a Lie subgroup ofAut H (Lemma 4 (ii)).
We transport the Lie group structure of Aut(H, N) to Aut G by means of
n'l. Then AutG becomes a finite—dimensional Lie group. Properties (ii),
(iii) and (iv) of the theorem are satisfied and the mapping (a, g) >—> u(g) of
(Aut G) x G into G is analytic (Lemma 4 (i)). Let M be an analytic mani-
fold, f a mapping of M into Aut G and (I) the mapping (m, g) >—>f (m) g of
M x G into G. Clearly, iff is analytic, (I) is analytic. Suppose that (I: is ana-
lytic. Then the To: TM x TG —> TG is analytic; its restriction to M x L(G),
that is the mapping (m, x) >—> L(f (m))x of M x L(G) into L(G) is therefore
analytic; as L(G) is finite-dimensional, it follows that the mapping
m >—> L(f (m)) of M into Aut L(G) is analytic and hence that f is analytic.
Thus (1) holds.
Let L(G) be given a norm. For all x > 0, let B,‘ be the open ball of centre
0 and radius A in L(G). We choose A > O sufficiently small for \l; = expulliA
to be an isomorphism of the analytic manifold BA onto the open submanifold
¢(B,‘) of G. Let G) be a filter on Aut G. For (D to converge to Ida, in Aut G,
it is necessary and sufficient that L((D) converge to Idug, in Aut L(G) and
hence that L((l))|BMa and L((D)"Bm converge uniformly to Idaw This
condition implies that (DIMBW) and ¢“]¢(Bm) converge unifomly to
Idwamr Conversely, suppose that (DHKBM) converges uniformly to Idmam).
There exists Med) such that, if ueM, then u(¢(BN2)) C (L(n); then
L(u) (BN2) is a connected subset ofL(G) whose image under expG is contained in
“Banal: hence L(u)(Bm) does not meet B,L — Bem and therefore
L(u)(Bm) C 13,; then the hypothesis that ¢|¢(BA,,) converges uniformly to
Elma”) implies that L((D)|Bm converges unifomly to Idam. It then follows
at:
((1) converga to Ida in Aut G) a ((1) convergs to 1d,; under .71.).

This proves (vi).

364
 AUTOMORPH’ISM GROUP or A LIE GROUP (REAL on COMPLEX CASE) § 10.2

Let D be the law of infinitesimal operation associated with the law of left
operation ofAut(G) on G. By Propositions 1 and 2 of no. 1, D(L(Aut G)) = a.
Hence a is a Lie algebra of vector fields and D is a morphism of L(Aut G)
onto a. Let x1 and xa be elements of L(Aut G) such that D(x,) = D(x,).
Then the laws of operation (A, g) >—> (exp M1)g and (A, g) s—> (exp M2) g of
K on G have the same associated law of infinitesimal operation; hence, for
ll] sufficiently small, exp M1 and exp M, coincide on a neighbourhood of 3
(§ 4, no. 7, Theorem 6), whence exp M1 = exp 1x2. It follows that x1 = x,
and hence D is an isomorphism of L(Aut G) onto a.
The theorem has thus been completely proved for G connected.
(c) V\’e pass to the general case. By hypothesis, G is generated by Go and
a finite number of elements :1, :2, . . ., x”. Every 14 eAut G leaves Go stable.
Let Aut1G be the set of u eAutG which, on passing to the quotient, give
the identity automorphism of G/Go. This is a normal subgroup of Aut C.
By part (b) of the proof, Aut G0 has a canonical Lie group structure and the
mapping (£1; £2: - ~ -: gm u) H ("31: “£2: ' - u “321) °f Gii X Aut Go into G3
is analytic. Let P be the corresponding sernidirect product of Aut Go by G3;
it is a finite-dimensional Lie group (§ 1, no. 4-, Proposition 7).
Ifw E Aut1 G, we write
111., = lo eAut Go
w. = xi-lWO‘t) 5G0 (1 g ig n)
{(w) = «"11, - - u '0»): we) 5 P-
For all w, w’ in Aut,
C(wmw') = ((101: ~ - '3 wn)(w0(wl)’ . . -, "Mn/i), mow/6)
= ((wlw0(wi)a - - -. wnwiz(w5»)), Wolv'o)
= ((X'lw(xx)w(x1'IW’(xi))u - -, x; ’W(xn)W(x;‘W'(xu))), wawé)
= (((wu/h, - - ‘a (um/)n), (ww')o)
= {(ww’)
and hence C is a homomorphism of Aut,G into P. This homomorphism is
obviously injective.
We show that {(Aut) is closed in P. Let 0 be a filter on Aut such that
{(43) converges to a. point ((wh . . ., w”), mo) of P. Then 11> converges point-
wise to a mapping v of G into G. Clearly I) is an endomorphism of the group
G. Moreover, 11 leaves each coset modulo G0 stable and I)|G,J 2 mo. It follows
that v EAut. As {(v) = ((wl, . . ., w"), wg), we have shown that C(Aut)
is closed in P.
(d) In part (d) of the proof we assume that K = R. By § 8, no. 2, Theorem
2, {(Aut) is a Lie subgroup of P. We transport the real Lie group struc-
ture on E(Aut) to Aut by means of C“. Thus Aut becomes a finite-
dimensional Lie group.

365
III LIE GROUPS

Let M be an analytic manifold, f a mapping of M into Aut and (I) the

mapping (m, g) Hf (m)g of M x G into C. We have the follovnng equiva-
lences:
f analytic
¢> the mappings m» (f (m))., where 0< <71, are analytic
the mappings m »—>f (m)x. of M into G, for 1 < i < n, are analytic
¢> and
{the mapping (m, g) »—>f (m) g of M X G0 into G is analytic
¢ <1) is analytic.
For w eAut, L(w) = L(wa) and hence the morphism w -—> L(w) of Aut
into Aut L(G) is analytic. We see as in (b) that the law of infinitesimal opera-
tion associated with the law of operation of Aut on G is an isomorphism
ofL(Aut1G) onto a.
Let C be a compact subset of Go generating Go. For a filter (I) to converge
to Ida on Aut, it is necessary and sufficient that 0K0 U {x1} U - - ‘ U {193)
and 0‘1[(C U {:41} U - . . U (m) converge uniformly to
mate u {*1} u u m».
The topology ofAut is therefore the topology 9],.
Clearly Aut,G is open in Aut G with the topology 9;. There exists on
Aut G a Lie group structure compatible with this topology and inducing on
Aut the structure constructed above (§ 8, no. 1, Corollary 2 to Theorem 1).
The fact that the Lie group Aut G has the properties of the theorem follows
from the corresponding properties for Aut.
(e) In part (e) of the proof we assume that K = C. By (c) and Theorem 2
of § 8, no. 2, there wrists on Aut 2 real Lie group structure such that t is
an isomorphism of Aut onto a real Lie subgroup of P.
The law of operation (10, g) v—> wg of (Aut) x G on G is real analytic.
Let D be the associated .law of infinitesimal operation. By Propositions l and
2 ofno. 1, D(L(Aut,G)) = a.
For all at E a, let an denote the restriction of ac to CD; it is an infinitesimal
automorphism of G0 which we identify, because of part (b) of the proof,
with an element of L(Aut Go). For 1 S i S n, we write
IX. = x," lat(x,) e L(G) = L(Ga).
Finally, we write f (at) = ((1,, . . ., a”), 1°)eL(P). Then f is a C-linear
mapping of a into L(P).
On the other hand, clearly L(E) =fo D. Hence L(C)(L(Aut,G)) = f(a)
is a complex vector subspace of L(P). By Proposition 2 of § 4, no. 2, C(Aut)
is a complex Lie subgroup of P and we can proceed exactly as in (d): we
transport the complex Lie group structure on Z(Aut) to Aut,G by means
of C' ‘ and we see as in (d) that Aut has the properties analogous to proper-
ties (i), (ii), (iii), (v) and (vi) of the theorem.

366
 Amouomrsu cxoup or A LIE amour (mm-mic cm) § 10.3

Clearly Aut is open in AutG with the topology 57],. Let w eAut G.
Let a be the automorphism u»—>wow'1 of Aut,Gr It is real analytic (§ 8,
no. 1, Theorem 1), L(a) is an R—automorphism of L(Aut) and
D a L(Aut,G) . 1)-1
is an R—automorphism of a. This automorphism is also the automorphism of
a derived from w by transport of structure; as w is K—analytic, we see that
L(u) is K-linear. Hence a is K-analytic (§ 3, no. 8, Proposition 32). By § 1,
no. 9, Proposition 18, there exists on Aut G one and only one Lie K—group
structure such that Aut,G is an open Lie subgroup of Aut G. The fact that
this structure has the properties of the theorem follows from the correspond-
ing properties for Aut.
COROLLARY I. Let G be ofinile-dimenxionol real Lie group and G0 it: identity com-
ponent. Suppose that G is generated by Go and afinite number of elements. Then Aut G
has the topology 9:, and is a finite-dimensional real Lie gimp.
COROLLARY 2. Let G be a semi-simple connected real or complex Lie group. The
group Int G is the identity component ofAut G.
The mapping u>—>L(u) is an isomorphism of AutG onto a Lie subgroup
of Aut L(G) (Theorem 1). The image of IntG under this isomorphism is
Ad G. But Ad G is the identity component of Aut L(G) (§ 9, no. 8, Proposi-
tion 30 (ii)).

3. EgEfUTOMORPHJSM GROUP OF A LIE GROUP MTRAMETRIC

THEOREM 2. When K i: ultrametrio and locally compact and G i: a compact Lie

group, assertion: (i), (ii), (iii), (v) and (vi) of Theorem 1 are true.
(a) The uniqueness of the analytic structure considered in (i) is obvious.
(b) Suppose that G is the Lie group defined by the normable Lie algebra L.
Then G is an open and closed ball in L. Let U) eAut G. Then L(w) coincides
with w in a neighbourhood of 0. Let x e G. Let p be the characteristic of the
residue field. Then [We tends to 0 as It tends to +00. There therefore exists
n such that w(p"x) = L(w)(p"x). Therefore
P"w(*) = 100‘)" = "1(1‘“) = WON)
= L(WXP") = P110100?)
whence w(x) = L(w)(x). Thus, w = L(w)]G.
Let I‘ be the set ofy EAut L(G) such that 7(G) = G. As G is open and
compact in L(G), 1" is an open subgroup Of Aut L(G). By the above, Aut G
is identified with I‘, whence there is a Lie group structure on Aut G, with
which properties (i), (ii), (iii) and (vi) of Theorem 1 are obvious. Property
(v) follows from Propositions l and 3 of no. 1.

367
 m LE GROUPS

(c) We pass to the general case. By §7, no. 1, Proposition 1, there exists
an open compact subgroup Go of G which is of the type considered in (h).
Then G is generated by Go and a finite number of elements x1, x3, . . ., x".
Let Aut be the set of u E Aut G such that u(Go) = Go and u(x.GD) = #463,
for l S i S n. We define as in the proof of Theorem 1, part (c), a semi-direct
product P of Aut Gn by GE and an injective homomorphism Z of Aut1G into
P, whose image is closed in P.
(d), (e) : the argument is exacfly as in parts (d), (e) of the proof of Theorem
1 with R replaced by Q, and using Proposition 3 instead of Proposition 2. .
Remark. If K = Q” and the Lie group G is gmrated by a compact subset
(cf. Exercise 2), assertions (i), (ii), (iii) and (vi) of Theorem 1 are still true,
but not (v) (Exercise 3).

APPENDIX
OPERATIONS ON LINEAR REPRESENTATIONS
Let G be a group, It a commutative field, E1, Em. . ., E” vector spaces over
I: and 7': a linear representation of G on E. (l .< i < n). The mapping
g-—> "1(a) ® --~ ® Ms)
is a linear mapping of G into the vector space E, ® . - - ® Em called the
tensor pmduct of 1:1, . . ., 1:, and denoted by 1:, ® - - - ® 1:".
Let E be a vector space over k and 1: a linear representation of G on E.
For all g 6G, let T(g) (resp. 0(g), e(g)) be the unique automorphism of the
algebra T(E) (resp. S(E), /\(E)) which extends 1r(g) (Algebra, Chapter 111,
§5, no. 2, § 6, no. 2 and § 7, no. 2). Then 4.- (resp. a, e) is a linear representa-
tion of G on T(E) (resp. S(E), /\(E)) denoted by T(n) (resp. 5(rr), /\(1r)).
The subreprsentation of T(n) (resp. 5(a), /\(1r)) defined by T"(E) (resp.
5"(E), A"(E)) is called the n-th tensor (resp. symmetric, exterior) power of
1: and is denoted by T"(rr) (resp. S"(1r), A”(1:)). Then
T"(1r) = 1r®1r® ®1r
(7: factors). The representations 5(a), /\(1r) are quotient representations of
T(n) and hence S"(1r), A”(1r) are quotient representations of T"(7r).
Let g be a Lie algebra over It. The tensor product of a finite number of
representations of g has already been defined in Chapter I, §3, no. 2; it is
denoted by 1:1 ® - - - ® 1:”. Let E be a vector space over I: and 7: a representa-
tion of g on E. For all x e 9, let 1"(x) (resp. a’(x), s'(x)) be the unique deriva-
tion of the algebra T(E) (resp. 503), /\(E)) which extends 1v(x) (Algtbra,
Chapter 111, § 10, no. 9, Example I). Then 1’ (resp. a’, a') is a linear repre-
sentation of 9 on T(E) (resp. S(E), /\(E)) by Algebra, lac. cit., formula (35),

368
 APPENDIX

denoted by T(1r) (resp. S(1r), /\(n)). The subrepmentation of T(1r) (resp.

SUI), /\(7r)) defined by T"(E) (resp. $“(E), A"(E)) is denoted by T"(1v) (resp.
S"(1c), A”(1r)). The representation T”(1c) is the tensor product on n representa-
tions identical with 1r. The representations S(1:), A(1r) are quotient representa-
tions of T(7r) and hence 5"(1r), A"(n) are quotient representations of T"(1:).

369
 EXERCISES

§1
1. Let G be a finite-dimensional connected real or complex Lie group.
Let H and L be Lie subgroups of G such that HL = G. Show that the canon-
ical mappings G—>G/H and G—>G/L define an isomorphism of analytic
manifolds
G/(H n L) —> (G/H) x (G/L).
2. Let P = {z E C|J(z) > 0}. Let G be the group of bijections z l—> uz + b
of P (a > O, b e R). Then G operates on P simply transitively, which allows
us to transport the complex analytic manifold structure on P to G. Show that
the structure obtained is invariant under left translations of G but not under
right translations.
3. Let R; be the additive group of real numbers with the discrete manifold
structure. Let Rd operate on the analytic manifold R by the law of operation
(1,3,) ~—> x + y. Then R, operata transitively on R but R is not a Lie homo-
geneous space of Rd.
4-. Let H be a compact real Lie group operating on a finite-dimensional
real manifold V; suppose that V is of class C', where reNR, and that the
action of H on V is of class C’. Let p eV be invariant under H; the group H
operates linearly on the tangent space T to V at 1). Show that there exist an
open neighbourhood U of [a which is stable under H and a C'—morphism
f: U —>T such that:
(a) f (p) = 0 and the tangent mapping to f at p is the identity;
(b) f commutes with the action of H on U and T.
(Choose first an fl, satisfying (a) and then definef by the formula

fix) = f haw-w,
where dli is the Haar measure on H with total mass 1.)

370
 EXERCISES

Deduce that the H-spaces V and T are locally isomorphic to a neighbour-

hood of p, in other words that there exist systems of coordinates on V at [I
with respect to which H operates linearly (“Bochner’s Theorem”).
5. Perform Exercise 4» for manifolds over an ultrametric field K, assuming
that H is a finite group whose order n is such that 71.1 aé 0 in K.

1] 6. Let G be a real Lie group operating properly on a finite-dimensional

Hausdorfl” real analytic manifold V. Let 1; EV, let H be the stabilizer of [z
and let G]: be the orbit off. The group H is compact and the manifold G1)
is identified with the Lie homogeneous space G/H.
(a) Show that there exists a submanifold S of V, passing through [7, stable
under H and such that T,(V) is the direct sum of T,(Gp) and TF(S). (Choose
a supplement of T,(Gp) in T,(V) which is stable under H and apply Exer-
cise 4-.)
(b) Suppose that S is chosen as above. The group H operates properly and
freely on G x S by h.(g,r) = (gh, h’k); let E = (G x S)/H be the corre-
sponding quotient manifold (cf. Dgfermliable and Analytic Manifolds, R, 6.5.1).
The mapping (g, s) »—> g: defines when passing to the quotient a morphism
p: E —> V. Show that p commutes with the action of G on E and V and that
there exists an open submanifold S’ of S containing 1), stable under H and
such that p induces a manifold isomorphism of E’ = (G x S’)/H onto a
saturated neighbourhood of the orbit G1).
(a) Let N, = T,(V)/T,,(Gp) be the transversal space at p to G}; (Di mn-
tiabl: and Analytic Manifoldx, R, 5.8.8); the group H operates on N,. Deduce
from (b) that knowing H and the linear representation of H on N, determines
the G-space V in a neighbourhood of the orbit ofpu
(d) If x 5 N,” let H, be the stabilizer of x in H. Show that there exists a
saturated neighbourhood V’ of G1: with the following property:
For all p’eV’, there exists xEN, such that the stabilizer of p’ in G is
conjugate (in G) to H,.
(e) Show that the Hx are finite in number, up to conjugation in H. (Argue
by induction on dim N1" = 1:, using the action of H on a sphere of dimension
n — l which is stable under H.)1'
1] 7. Let E be a complete normable space over R and F a closed vector
subspace of E such that there exism no bioontinuous linear bijection of E onto
F x (E/F).I Let X be the analytic manifold R X E X F. Let G be the

1’ For more details, see R. S. PALAIS, On the existence of slices for actions of non-
compact Lie groups, Ann. 44' Math, 72 (1961), pp. 295—323.
1 For an example of such an ordered pair (E, F), cl‘. for example A. DOUADY, Le
probléme des modules pour les sous-espaces analytiques compacts d’un espace
analytique donné, Ann. Inst. Fourier, 16 (l966),:p. [6.

371
m LEE GROUPS

additive group of F, which operates on X by the mapping

(is (A, of» »-> (A, e + of + Ag)
of G X X into X. For all x e X, let Rx be the G—orbit of x, which is a quasi-
submanifold of X. Then condition (a) of Proposition 10 is satisfied, but there
exists no ordered pair (Y, 1:) with the following properties: Y is an analytic
manifold, 1c is a morphism of X into Y and, for all x E X,

TA") 2 TAX) -> m>(Y)

is surjective with kemel Tx(R,,).
(Suppose that (Y, 1:) exists. Let H be the tangent space to Y at n(0, 0, 0).
Then H is isomorphic to R X F x (E/F). On the other hand H is isomorphic
to TwaY for x sufficiently close to (0, 0, 0) and is hence isomorphic to R x E.)
8. Let T be given the normalized Haar and $(L’(T)) the normie topology.
Show that the regular represenmtion of T on L2(T) is not continuous. (For
all geT distinct from 2, construct a function fe L‘(T) such that ||f|| = l,
|1Y(£)f‘fll = V1)
9. Let I be the set of(::, Viz) e R2 for —§ < x < i. Let U be the canonical
image of I in the real Lie group H = T“. Show that U is a Lie subgroup
germ of H but that the subgroup H’ of H generated by U is dense in H and
not closed in H.
10. Let R4 be the group R with the discrete manifold structure. Let H be
the real Lie group R x R4. Let U be the set of elements of H of the form
(x, x) or of the form (x, ——x) where xER. Then U is a Lie subgroup germ
of H, U is discrete, the subgroup G of H generated by U is equal to H and
the Lie group structure on G defined by the Corollary to Proposition 22 is
the discrete group structure.

§3
1. Let G be a Lie group and G1 and G, Lie subgroups of G. Suppose that
G2 is of finite codimension in G and that K is of characteristic 0. Then
G1 fl G2 is a Lie subgroup of G with Lie algebra L(Gl) fl L(G2) (argue as
in Proposition 29 and its Corollary 2).
‘ 2. Suppose that K is locally compact. Let G be a Lie group of finite dimen-
Slon ’l.

(a) Let (I) be an étalo endomorphism of G with finite kernel. Let u be a

left Haar measure of G. Then 4’ is proper and
Card(Ker 4))
(NF) = “-FWW‘) where m = m-
(Use Proposition 55.)

372
 EXERCISE

(b) Suppose that G is compact. Let (b be an étale endomorphism of G. Then

Ker d) is finite, (MG) is of finite index in G and
Card(Ker 4))
= mod det L(¢).
C7"Fd(G/¢l>(G))
(Use (a) to transform y.(G).)
(5) Suppose that G is compact and commutative. Let rEZ be such that
1.1 gé 0 in K and let 4’ be the endomorphism x -—>x" of G. Then
Card(Ker ¢) = (mod 1.1)".
Ca:d(G/¢(G))
(Observe that L(<I>) is the homothety of ratio r by §2, Corollary to Proposi-
tion 7.)
3. Let E be the complex vector space of symmetric complex matrices with
2 rows and 2 columns. For all 3 e SL(2, C), we define an automorphism p(:)
ofE by p(:).m = x.m.':.
(a) Show that p is an analytic linear representation of SL(2, C). Identifying
b . . .
(Z 5) 6E With (a, 17,5) 6C3, the matrix ofpc 91s

at“ 24B 3
(“Y 118 + M an) ‘
7’ 278 8’
(b) Show that m >—> det(m) is the non-degenerate quadratic form
q(a, b, t) = no — b2 .
on E and that p is a morphism of SL(2, C) onto SO(q) with kernel (I, —I}.
(For the surjectivity, observe that, in the matrix of an element of 80(q), the
lst column can be expressed in the form (an’, «y, y“) and the 3rd column in
the form (3‘, B8, 8“), where «’82 + BRY‘ — Quyfls = 1. Then observe that
2 elements of SO (q) which coincide on a non-isotropic plane are equal.)
(4) Deduce that the complex Lie group 50(3, C) is isomorphic to the
complex Lie group SL(2, C)/{I, -I}.
4-. (a) Let q be a quadratic form of signature (1, 2) on R“. Let P be the
set of m e R:3 such that q(m) > 0. Then P is the union of two disjoint oonvec
cones C and —C. IfsESO(q), then either 5(0) = C and .y(-—C) = -—C or '
:(C) = -—C and :(—C) = C. Let 50*(q) be the set of:eSO(q) such that
:(C) = G. Then 80*(q) is an open normal subgroup of 80(q) of index 2
in 50(q).
(b) By imitating the method of Exercise 3, define a morphism of SL(2, R)
onto SO+(q) with kernel {1, —1}. Deduce that the real Lie group SO+(q)
is isomorphic to the real Lie group SL(2, R) [(1, —I}.

373
“I LIE GROUPS

(5) For (Em) EC“ and (E', 71') EC“, let f((€,n)s (5': n'))1= Egl' — 70771
Show that the inner automorphism of SL(2, C) defined by 7—5 (_1. 7)
maps SU(f) to SL(2, R).
5. Let E be the real vector space of Hermitian complex matrices with 2
rows and 2 columns of zero trace. For all .r ESU(2, C), we define an auto-
morphism p(r) of E by p(:) (m) = m‘. We identify the element
( x y + £2)
y — 1'2 ——x
of E with the element (x, y, z) of R3. By imitating the method of Exercise 3,
show that p is a morphism of the real Lie group SU(2, C) onto the real Lie
group 50(3, R) with kernel {1, —I} and that 80(3, R) is isomorphic to
5U(2, GNU, -1}
1T 6. Let F be the vector space of Hermitian complex matrices with 2 rows
and 2 columns. For all .r e SL(2, C), we define an automorphism 6(5) of F by
o-(:) (m) = m*. We identify the element t : 1: y‘ :- 11) of F with the
element (t, x, y, z) of R4 and write
q(t,x,y,z) =1“ - I" -y’ — 2’.
Let C be the set of (t,x,y,z) eR4 such that t2 — x2 —y2 — 7.2 > 0, t > 0.
Let SO+ (g) be the set ofg E 50(4) such that g(C) = C; it is an open normal
subgroup of SO(q) of index 2 in SO(q) (cf. Exercise 4). Show that o' is a
morphism of the underlying real Lie group of SL(2, C) onto the real Lie
group SO+(q) with kernel (I, —I}r (For the suijectivity, use Exercise 5.)
Deduce that the real Lie group 80+(q) is isomorphic to the underlying rail
Lie group of SL(2, C)/{I, —I}, that is (Exercise 3) the underlying real Lie
group of SO(3, C).
7. (a) Let G be the set of quaternions of norm 1. This is a Lie subgroup of
the real Lie group 11*, homeomorphic to 5,, and hence simply connected,
cf. General Typology, Chapter XI.
(b) Let E be the real vector space of pure quaternions, identified with R3
by
(x,y, z)t—>xi +w' + 11:.
For all g e G, we define an automorphism p(g) of E by p(g)q = gqg“. Show
that p is a morphism of the Lie group G onto the real Lie group 50(3, R)
with kernel {1, — I}. Deduce that the Lie group 80(3, R) is isomorphic to
the Lie group G/{l, — l).
(5) Let C be identified with the subfield R + Ri of H. The mapping

374
 EXERCISES

(x, q) F) q)‘ of C x H into H makes H into a vector space over C which is

identified with C2 by the choice of basis (1, j). For all g E G, let a(g) be the
automorphism of this vector space defined by a(g)q = gq. Show that a is an
isomorphism of the real Lie group G onto the real Lie group SU(2, C). The
latter is therefore simply connected by (a)
flu) For (q1, 112) e G x G, let r(q,, 92) be the mapping 1; »—> qq, of
= R4 into itself. Show that 1: is a morphism of the real Lie group G x G
onto the real Lie group 80(4, R) with kernel N— = {(1, l), (—1,—l)} and
hence that 50(4, R) is isomorphic to (G x G)[N.

8. Let G be a finite-dimensional connected real Lie group and M a finite-

dimensional connected manifold of class C“; let there be given a law of right
operation of class C‘” of G on M. For all x e M, let p(x) be the corresponding
orbital mapping.
(a) For G to operate transitively on M, it is necessary and sufficient that,
for all x e M, the mapping T.(p(x)) of L(G) into TAM) be surjective. (To
see that the condition is sufficient, observe that then every G—orbit in M is
open.)
(b) Suppose that G operates transitively on M. We identify M with a
homogeneous space H\G, where H is the stabilizer of a point x0 of M. The
action of G on M is called imprimifive if there exists a closed submanifold V
of M of class C" such that 0 < dim V < dim M and for which every trans-
form VJ (where r e G) is either equal to V or disjoint from V. For this to be
so, it is necessary and sufficient that there exist a Lie subgroup L of G such
that H C L C G and dim(H) < dim(L) < dim(G). If this is not so, the
action of G on M is called primitive; for this it suflices that there exist no
subalgebra of L(G) lying between L(G) and L(H) and distinct from L(G)
and L(H).
(5) Under the hypothesis of (b), let .S!’ be the image of L(G) under the
law of infinitesimal operation associated with the action of G. Let of be the
algebra of functions of class C" on M and m the maximal ideal of 6 con-
sisting of the functions in 6 which are zero at are; we denote by .Z’F for
p = —l,0, 1,2,... the set ofxeg’ such that 0(X)fem”” for allfeé’;
then .921 = .7 and $0 is the image of L(H) under the law of infinitesimal
operation. If (U, 42, R“) is a chart of M centered on are and X1, . . ., X,I are
the vector fields on U defined by the chart, the elements of .7, are the
alX1 + - ~ ~ + a,,X,I with an . , .,az,l in m”|U. Show that [23, 2"] C 2’,”
(with the convention 2-2 = E). If there exists YEY, (for p 2 0) such
that Y $2,“, there exists xEL such that [Y, X] $3,. The Y, are Lie
subalgebras of .7. For I) 2 0, .‘Z’, is an ideal of .‘Z’u. Let p be the canonical
linear representation of H on T,0(M). The Lie algebra of H/(Ker p) is iso-
morphic to YO/Yr
(d) Suppose further that the intersection of the 3’, is {0}. Then there

375
III LIE GROUPS

exists a greatest index I such that 1?, sé {0} and all the 2, of indices <7
are distinct. For [I 2 0,
. It +
dlm($p/$,H) $ (P +{)

(a) The hypothesis of (d) IS satisfied if M is analytic and G operam analytic-

ally onM
9. In the notation of Proposition 30, HH' may be an open subset of G _
dense in G and distinct from G (take G = GL(2, R) with H the upper tn-
angular subgroup and H’ the lower strict triangular subgroup).

§4
1. Let G be a Lie group, H a normal Lie quasi-subgroup of G and 1c the
canonical mapping of G onto GIH. There exism on GIH one and only one
Lie group structure with the following property: for a homomorphism 8
of G/H into a Lie group G' to be a Lie group morphism, it is necessary and
suflicient that 0 c 1: be a Lie group morphism. Moreover, L(G/H) is canonic-
ally isomorphic to L(G)/L(H). (Let Q be a Lie group germ such that
L(Q) = L(G)/L(H). Show that, by shrinking Q if necessary, Q can be
identified with an open neighbourhood of 0 in GIH and that Proposition 18
of § 1 can then be applied to G/H.)
H11 2. Let G and H be two Lie groups andf a Lie group morphism of G'into

H(i) The kernel N of f is a normal Lie quasi-subgroup of G and

L(N)= Ker L(f ) (Use exponential mappings of G and H).
(ii) Let g: G/N —>f (G) be the mapping derived from f when passing to
the quotient. Iff and L(f ) have closed images, and the topology of G admits
a countable base, then f (G) is a. Lie quasi-subgroup of H with Lie algebra
Im L(f ) and g is a Lie group isomorphism, where G/N has the structure
defined in Exercise 1. (Reduce it by means of (i) to the case where N = {4.)
3. Let G be a Lie group, U an open neighbourhood of 0 in L(G) and (I)
an analytic mapping of U into C such that “0) = e and To‘b = IdL(g). The
following conditions are equivalent:
(a) (I) is an exponential mapping;
(b) There exists mEZ distinct from 0, 1, —1 such that (11(mx) = fix)“ in
a neighbourhood of 0.
(If condition (b) holds, show that 4;" satisfies condition (iii) of Theorem 4.)
114-1 (a) Let K = R or C or an ultrametric field with residue field of
characteristic >0. Let G be a Lie group over K, U an open neighbourhood
of 0 in L(G) and <11: U —> G a mapping which is differentiable at 0 and such

376
 EXERCISES

that (NO) = e, To(¢l>) = Idua, and 42((1 + x')b) = <1>(M)<]>()t’b) for M, Nb,
(A + A’)b in U. Then :1) coincides on a neighbourhood of 0 with an aponen-
tial mapping.
(b) Let k be a field of characteristic 0. Let K be the valued field k((X)),
which is of characteristic 0. Let G be the additive Lie group k[[X]]. Let ()2
be the continuous k-linear mapping of G into G such that (“X") = X” + X’"
for every integer n 2 0. Then (1) satisfies the conditions of (a), but coincides
on no neighbourhood of 0 with an arponential mapping.
5. Let (:1, e2, :3) be the canonical basis of 113‘ We give R“ the nilpotent
Lie algebra structure 9 such that [51,52] = 23, [91,33] = [65,93] = 0. Let
0”,, denote the constants of structure of 9 relative to the basis (:1, :3, :3).
(H) Show that on R3 the difierendal forms
‘01 = 1‘1, 02 = dim 0’s = “”1412 + (in
satisfy the relations
dmk = fig, We), A a), (k = 1,2, 3)
and are linearly independent at each point of R”.
(b) Deduce that there exists, on an open neighbourhood of 0 in R“, a
Lie group germ structure G such that L(G) = g and
(l) (‘11: “2: a3) (’51: ”a: *3) = (“'1 + “1; 7‘2 + “a: “a + '13 + “112)
for (al, a2, a3), (x1, #2, x3) sufliciently close to 0. Show that in fact formulae
(1) define a nilpotent Lie group structure on R“.
11 6. Let G be a finite-dimensional Lie group, 9 its Lie algebra, g‘ the dual
vector space of g, a the adjoint representation of G on g and p the representa-
tion ofG on 9* defined by p(g) = ‘a(g) '1 for all ge G.
(a) L(p)(:<) = —‘(ad x) for all x E g.
([7) Let fe 9*. Let G, denote the stabilizer off in G; it is a Lie subgroup
ofG and g, = L(G,) is the set ofx EG such that ‘(ad x)f= 0.
(c) For x, y in 9, we write B,(x, y) = f ([x, y]). Show that B, is an alter-
nating bilinear form on 9, that the mapping 5, of 9 into 9* left associated with
B, is the mapping x >—> '(ad 2:) f and that the orthogonal of g with respect to
B, is 9,. Let {3, denote the non—degenerate alternating bilinear form on 9/9,
(which is therefore of even dimension) derived from B, when passing to the
quotient. We denote by {3; the inverse form of B, on (g/g,)*.
((1) Let 9 denote an orbit of G on 9*; we give it the manifold structure
derived from that on GIG, by transport of structure, where f is an arbitrary
point of 9. Then G operates analytically on Q on the left. Let D be the
associated law of infinitesimal operation, For all a e g and all fe 9,
Du.(f) = —(‘ada)f-
377
[ll LIE GROUPS

The tangent subspace T,(Q) is the image of 5,, that is the orthogonal of g,
in g*, and is canonically identified with (g/g,)*, so that 5, defines a canonical
isomorphism r, of T,(Q) onto its dual. The fieldfn—> p; is an analytic differen-
tial form to of degree 2 on K], which is invariant under G. For f E Q, a E 5,
b e g, «(Dam Dbifl) =f([a: b1).
(9) Show that do) = 0.
(f) If at is a difl‘erential form of degree 1 on Q, the isomorphisms 7, allow
us to identify at with a vector field. Let A be the set of analytic functions on
Q with values in K. For d), '1? in A, we write [<1>, 4/] = um, (14‘) e A. Show that
A has thus a Lie algebra structure.
(g) For all a e 9, let v.11“ be the function f ~—>f (a) on Q. Show that a >—> up“
is a homomorphism of 9 into the Lie algebra A and that the form do, is
identified, by means of the isomorphisms 1,, with the vector field D“.
(/1) Let U be an open subset of g“ and d) an analytic function on U. For all
fe U, the differential d,<]> of d) atf is identified with an element of 5. Suppose
that X4; = 0 for every vector field X defined by the action of G on 9*. Show
that then, for allfE U, d,<lz belongs to the centre of 55,.
(2') For all f e 9‘, we write 1, = dim 9,. Let r = inf3,. Show that the set
i 5 I
V off e {1* such that r, = r is open and dense in 9‘. Show that, for all f e V,
g, is commutative. (Construct r functions 4: satisfying the conditions of (h)
in a neighbourhood off and such that their differentials at f are linearly
independent.)
11 7. (a) Let (A, x) ~—> L): be a law of continuous left operation of R on T.
Then one of the two following holds:
(1) either there exim a fixed point and the stabilizer of every point is either
R or {0};
(2) or there exists a point whose smbilizer is an infinite discrete subgmup
of R and R operates transitively on T. (If {Aelix = x} = {0}, the orbit
of x is homeomorphic to R and has frontier points Alim Lx which are fixed.
+ on
-
If {heRILx = x} = In with a $5 0, the orbit of x is homeomorphic to T
and R is transitive.)
(b) Iff is a homeomorphism of T onto itself and k an integer 21, a point
xeT is called periodic of period I: for f if f*(x) = x and f "(x) ;é a: for
1 < h < k. In the notation of (a), let f,L be the homeomorphism x»—> 1.1:.
Then the set of periodic points of period It > 1 forfi is empty or equal to T.
(If there exists a periodic point of period k > 1 for f,” its stabilizer is distinct
from {0} and R, hence R operates transitively on T and all the points of T
are periodic of period I: forfi.)
(1:) Let Q be a neighbourhood of e in Difl"“'(T) (Dzfirmtiable and Analytic
Mani/7211A, R, 15.3.8, Nole (1)). There exist an integer It > i and 9 with-
out fixed point such that the set of periodic points of period k for f is neither
empty nor equal to T. (Let p: R —> T bc the canonical mapping. Let p be

378
 mm

the mapping p04) Hp(x + 2—1:) of T into ’1'. Let Q’ be a neighbourhood of

e in Difi‘“’ (T) such that [2'2 C 9. For sufficiently large k, p 60'. On the
other hand, let a be a mapping of T into T such that 6(y) = y for

11 $1100, 2—,: )
and mm) = p(x + m» for x 510, 3,? :where
0<).(x)<2TR—x;

can be chosen such that a- e 0'. Thenf = p a a has the required properties.)
Such an f cannot, by (b), be in the image of a continuous morphism of R
into Diff"u (T).
((1) Let Y be a compact difi‘erentiable manifold of dimension 21. Every
neighbourhood of z in Difi‘”(Y) contains an element It with the following
property: It belongs to the image of no continuous morphism ofR into Difl‘°(Y).
(The manifold Y contains an open submanifold of the form T x D, where
D is the open Eulcidean ball of radius 1 and centre 0 in R" (n = dimY — 1).
Let geDifl‘“(D) be such that g(0) = 0, ||g(x)]| > “x” for 0 < “x“ < 41- and
g(x) = x for ”x” 2 i and g very close to e in Difl‘”(D). Letf be as in (c),
and, for 0 < t < l,1etf,:T—>T be defined as follows: for 3 ET, 1/ E];"(x),
zep“(f(x)) and |z — y| < 1:, we write fi(x) =p(ty + (1 — t)z). Then
there exists h eDifl'“(Y) such that h(x,y) = (fzflyn(x),g(y)) for x ET, 3/ ED
and h(u) = u for u eY —- (T x D); moreover, we can choose f and g such
that h is arbitrarily close to e in Difl‘“(Y). The points of T X {0} are the
y e Y such that when It tends to +00, h""(x) tends to a point not fixed under h
(in fact it is periodic of period k). Therefore, every homeomorphism of Y
onto Y which commutes with It leaves T x (0) invariant. Then apply (0).)
(e) Let Y be a compact differentiable manifold of dimension 21. There
do not exist a Lie group G and continuous morphism Diff ”(Y) —v> G,
G —> Diff°(Y) whose composition is the canonical injection of Diff ”(Y) into
Difl‘°(Y). (Use (11).)
8. Let G be a finite-dimensional Lie group. Suppose that L(G) is simple.
Let A be a normal subgroup of G. If A is not open, A is discrete and its
centralizer in G is open. (Let n be the subalgebra tangent to A at a. Show that
a = {0). Let x be an element of A not belonging to the centre of G. Consider
the mapping y ny" of G into A. Deduce from the relation is = {0} that
the centralizer of x in G is open. Then, using an exponential mapping, show
that the centralize!" of x in G contains a neighbourhood of 1 independent
of x.)

379
 [II LIE GROUPS

1] 9. Let G be a compact Lie group over K of dimension n. Suppose that

the Lie algebra L(G) is simple. For all g e G, every integer m 2 l and every
neighbourhood V of z in G, let M(g, m, V) denote the set of elements of G
of the form flew” g);:,‘1 with x), . . ., x,” 3/1, . . .,y,,, in V.
(a) Show that if g is an element of G whose centralizer is not open, if m
is an integer 2n and if V is a neighbourhood of c, then M(g, m, V) is a
neighbourhood of 2. (There exists a E L(G) such that (Ad g)(u) aé a. Let 4)
be an exponential mapping of G. Let y be the image of 1 under the tangent .
mapping at 0 to A»(¢(M),g) (where A 5K). Then b = (Adg — l)(a) ;6 0.
There exist x1, . . ., x", in V such that {(Ad in) (b), . . ., (Ad xm)(b)} contains
a basis of L(G). Consider the mapping
.
<~;,...,x.,yl,...,y.)HEMybgm-l
of G2" into G. Show that the tangent mapping at (x1, . . . , an, t, . . ., e) is sur-
jective.)
(b) Let U be a neighbourhood of e in G. Let It: be an integer >1. There
exists an element g of G whose centralizer is not open, such that
M(g, m, G) C U. (Argue by ”dumb ad abmrdum using the compactness of G.)
(c) Let G’ be a compact Lie group over K whose Lie algebra is simple.
Let (I): G—>G’ be a bijective homomorphism of abstract groups. Then 4) is
continuous. (Apply (b) to G’ and (a) to G.)
10. Let G be a Lie group over K. Show that there exists a neighbourhood
V of 9 with the following property: for every sequence (xn) of elements of V,
if we define inductively 1/1 = :1, y, = (3", y,_,), the sequence (yfl) tends to :.
ll. (a) For x, y in S,,,_1 C C“, we define a(:r,y) e (0, 1:] by
WHOM) = “(Mi/D-
For :, t in U(n), we write d(r, t) = ssup «(:x, Ix). Show that d is a left and
are .._
right invariant distance on U(n). 2 l
([7) Let : eU(n). Let 9,, . . ., 0,. be the numbers in )—1r, 1:] such that the
5‘91 are the distinct eigenvalues ofr. Let 0(r) = sup |0,[. Then 11 (z, r) = 0(r).
l ‘ I ‘fll
(Let V, be the eigenspace of r corresponding to (ml. For x e San—1: find a
lower bound to .9?((z|:x)) by decomposing with respect to the V,.)
(5) Let s, I be in U(n) be such that 0(t) < 12: and r commutes with (r, t).
Then 5 and t commute. (In the notation of (1:), let V; be the orthogonal
supplement ofV, and W, = l(V,) ; show that W, = (W, n V,) + (W, n V;)
and then that W, n V; = {0}.)
(:1) Show that there exists a compact neighbourhmd V of e in U(n) which

380
 EXERCISES

has the property of Exercise 10, which is stable under inner automorphism
of U(n) and which is such that if x, y are non-permutable elements of V then
x and (x, y) are not permutable. (Use (c).)
(2) Let {3 be the normalized Haar measure of U(n). Let Vl be a symmetric
compact neighbourhood of e in U(n) such that V: C V. Let p be an integer
such that {3(V1) > Up. Show that, for every finite subgroup F of GL(n, C),
there exists a commutative normal subgroup A of F such that Card(F/A) S [7
(Jordan’s theorem). (Reduce it to the case where F C U(n). Take A to be the
subgroup of F generated by F n V and use (41).)
12. Let G be a finite-dimensional connected real or complex Lie group
and or a left invariant difi‘erential form of degree 9 on G. For a: to be right
invariant too, it is necessary and suflicicnt that do: = 0. (Observe that the
condition for right invariance can be written
/\”(‘Ad(I))-¢(¢) = w)
for all s e G and hence it is equivalent to the condition
I
,Z;<°‘(‘)’"l A A 14,.1 A [u,u,] A u,“ A A 11,) =0
for all u, 111,. . ., u, in L(G).) In particular, the left and right invariant dif-
ferential forms of degree 1 on G form a vector space of dimension
dim L(G) — dim[L(G), L(G)].
13. Let R" be given the usual scalar product ((5,), (111)) H‘; Em“ Let
I(n) be the group of isometric affine transformations of R". It is canonically
isomorphic to the Lie subgroup of GL(n + l, R) consisting of the matrices

S=(é’ f)
where U 6 C(71) and x is an arbitrary element of R" (matrix of type (It, 1)).
I(n) can be identified with the semi-direct product of 0(a) by R" defined by
the canonical injection of 0(a) into GL(n, R) (the matrix S is then denoted
by (U, x)). Let p: 1(a) ~> 0(n) be the canonical surjection, with kernel R".
(a) Let 1" be a discrete subgroup of 1(a). Consider the closure ATP) ofp(I‘)
in 0(a), Show that its identity component is commutative. (Let V be a neigh-
bourhood of I in 0(n) with the properties of Exercise 11 (d) and such that
further ||U — I H S i for all U EV (the norm on End(R”) being derived
from the Euclidean norm on R"). Argue by reductir ad abmrdum, assuming
that there exist SI = (U1, x1), S2 = (Umxz) such that U1 and (I; belong to
V and do not commute. Show that (SJ, S3) = ((U1, U1), y) with
M < Mllxill + “12“)-
38]
III LIE GROUPS

Then define the Sk inductively by S,‘ = (S,, Ska) for k 2 3. Using the choice
of V, show that this sequence has an infinity of distinct terms and is bounded
in I(n), which is absurd)
(b) l" is called crystallographic if I(n)/l" is compact. Show that if this is so,
for all x ER", the affine subspace L of R“ generated by D; is equal to R“.
(Otherwise, for all y e R", all the points of I‘y are the same distance from L
and, as this distance can be arbitrarily large I(n)/I‘ is not compact.)
(c) If I‘ is commutative and crystallographic, then I‘ C R“. (If
S = (U, x) E I‘ is such that the vector subspace V C R" consisting of the -
points invariant under U is not equal to R" and VJ- denotes the orthogonal
subspace of V, then, for all 5’ = ( U’, x’) e F, U' leaves V and Vi stable. On
the other hand, since UIVi has no non-zero invariant point, show that this
allows us to assume, by changing the origin, that e. Using (b), there
exists 5' = (U’, x’) e I‘ such that the orthogonal projection y’ of 1’ onto V‘-
is #0. By evaluating Sx’ by the formula S = S’SS"‘, deduce that U.y' = y’
contrary to the definition of V.)
(d) If I‘ is crystallographic, I‘ n R’‘ is a free commutative group of rank
n and I‘/(I‘ ft R“) is a finite group (Bieberbach’s Theorem). (Let W be the
vector subspace of R" generated by I‘ n R". The compact group W has
only a finite number of connected components.» If W = {0}, I‘ contains, by
(a), a commutative normal subgroup of finite index 1‘1; 1‘1 is then crystallo-
graphic, which implies a contradiction with (0). Hence W sé {0}. Show that
9(1") leaves W stable and that p(l")|W is finite; otherwise there would exist
a sequence (Sm) in I‘ such that if S", = (Um, x”), the U,,l are distinct and tend
to I; then form the (I, a,)S,,,(I, a,)‘1S;‘, where (a,) is a basis of I‘ n R”, and
obtain a contradiction with the hypothesis that l" is discrete. Finally, to see
that W = R”, show that otherwise the action of I‘ on R"/W would be that
of a crystallographic group containing no translation #0.)

§5
1. Suppose that K = R. Let r be an integer 21 or 00. A group of clam C'
is a set G with a group structure and a manifold structure of class C' such
that the mapping (my) ~—> xy“ ofG X G into C is of class C’.
(a) Suppose henceforth that r 2 2. A neighbourhood of e in G is identified
with a neighbourhood of 0 in a Banach space. Let xy = P(x, 3/), where P is
a mapping of class C2 on an open neighbourhood of (0, 0). Let

(D1132?) (0: 0) = B-
Show that

xy = x + y + Bins) + 1*! MW)

382
 EXERCISES

as (x, y) —> (0, 0). (Use an expansion of P of order 2 with integral remainder.)
(b) Show that
x“ = —x + B(x,z) + lx]’u(l)
m“ 1/ + Bony) - 13m”) + lxl IyW)
x‘ly‘ 1y = 130w) — B(y.x) + III [y|v(1)
as (my) tends to (0, 0).
2. Let G be a group ofclass C’ with r > 2.
(a) Let t 6.7"")(G), t’ 6.9"“)(G). Ifr + A" S r, define I: t’ EY"""’(G) as
in § 3, no. 1. If t and t’ are without constant term, It t' is without constant
term. The image of t under the mapping x »—> x“ of G into G is denoted by
t"; then If" 6.7'“‘(G). Iffis a function of class Cr on G with values in a
Hausdorff polynormed space, f): I denotes the function on G defined by
(f4: t)(x) = (5,, * t‘,f) for all xeG; this function is ofclass C"’ if: < 00.
Show, asin§3, no. 4, thatfa (tst') = (fst) *t'.
(b) If t e T,(G), the vector field x -> a, a: I on G is denoted by L‘. Show, as
in §3, no. 6, that, for t, t’ in T,(G), L”. = L, o L,. and hence that
Law—t": = [Lo Lu}
Therefore, H t' — t' t t is an element of T.(G) which we denote by [t, t'].
(0) Using the notation of Exercise 1 (u), if t eT,(G), then
L4") = (DaPOr; 0))(1)
for x sufficiently close to 0. Deduce that [t, I’] = BO, 1’) — B(t’, t).
(d) Show that T,(G), with the bracket (I, t’) >—> [t, t'], is a normable Lie
algebra. (To prove the Jacobi identity, use (c), Exercise 1 (b) and identity
(5) of Algebra, Chapter I, §6, no. 2.)
(For a sequel to this exercise, cf. §8, Exercise 6.)

§6
I. Let D be a set of elements of SL(2, C) of the form (3 a? 1) where
a > 0 and b E C. Show that D is a Lie subgroup of the underlying real Lie
group of SL(2, C) and that the mapping (u, d) »—> ml of SU(2, C) X D into
SL(2, C) is an isomorphism of real analytic manifolds. Deduce from this
and §3, Exercise 7 (a) that SL(2, C) is simply connected.
2. Let G be the universal covering of SL(2, R), 1: the canoncial morphism
of G onto SL(2, R) and N its kernel.
(:1) Arguing as in Exercise 1, show that there exists an isomorphism of the
real analytic manifold U x R2 onto the real analytic manifold SL(2, R).
Deduce that N is isomorphic to Z.

383
11! LE GROUPS

(b) If p is an analytic linear representation of G on a complex vector space,

then Ker p D N. (The representation L(p) of 31(2, R) defines, by complem-
fication, a representation of 91(2, C) and the latter is, by Exercrse l, of the
form L(o), where a is an analytic linear representation of SL(2, G). Then
L(o o 1:) = L(p) and hence a o 1: = p.)
3. Let G1 be the simply connected nilpotent real Lie group defined in
Exercise 5 of § 4-. Let Z be the centre of G1, N a non-trivial discrete subgroup
of Z and G = G1/N. If p is a finite-dimensional analytic linear representa- ,
tion of G, p(Z/N) is a semi-simple set of automorphisms, for Z/N is compact;
but Z = (G, G) and hence p(Z/N) is unipotent (Chapter I, §6, Proposi-
tion 6). Therefore p is trivial on ZIN.
4. Let G be a compact connected complex Lie group. Show that every
analytic linear representation of G is trivial.
5. In the notation of Exercise 9 of § 1, show that H’ is an integral subgroup
of H. Deduce that in the simply connected group SU(2, C) X SU(2, C) (§ 3,
Exercise 7 ((2)), there exist one-parameter subgroups which are not closed.
1T 6. Let I be the interval [1, 2] with the discrete topology. Let E be the
complete normed space of functionsf: I —> R such that Hf |[ = g; l f(1')] < +
00. For all x e I, let 5, be the element ofE such that :x(t) = 0 fort 56 x and
:x(x) = 1. Let P be the subgroup of E generated by the Max with x e I; it is a
discrete subgroup of E. Let G be the Lie group E/P. Let F be the hyperplane
of E consisting of the fa B such that Efli) = 0. Let H be the Lie group
F/ (F n P). Let (b be the canonical morphism of H into C. Show that d) is bijec~
tive and is an immersion but is not a Lie group isomorphism. (Letf be a linear
form on E with kernel F; show thatf(P) = R and hence that F + P = E.)
Deduce that in Proposition 3 the countability hypothesis cannot be omitted.
7. Let (I) he the morphism of 2/22 into the permutation group of C which
maps the non-identity element to z >—> 2. Let G be the semi-direct product
of 2/22 by the real Lie group C corresponding to 4). It is a real Lie group
with lie algebra R“. Show that there exists on G no complex Lie group
structure compatible with the real Lie group structure.
11 8. Let H be a complex Hilbert space of dimension N0 and G the unitary
group of H considered as a real Lie group. The group G is simply connected.T
(a) Let Z be the centre of G, which is isomorphic to T. Let u be an irrational

1’ Cf. N. H. KUIPER, The homotopy type of the unitary group of Hilbert space,
Tupulngy, 3 (1965), pp. 19—30. In this article it is even proved that G is con-
lractible.

384-
 EXERCISES

number. Let 3 be the Lie subalgebra of L(G) x L(G) consisting of the (x, ax)
where xEL(Z). Let S be the corresponding integral subgroup of G x G.
Then 6 is an ideal of L(G) x L(G); however S is not closed and is dense in
Z x Z.
(b) Let g = (L(G) X L(G)) [5. There exists no Lie group with Lie algebra
9. (Let H be such a group. As G is simply connected, there exists a morphism
i“ G X G——>H such that L(o) is the canonical mapping of L(G) x L(G)
onto 3. LetN = Ker (b. ThenL(N) D 9, henceN D S and therefore N D Z x Z
by (0). Then L(N) 3 L(Z) x L(Z), which is absurd.)
9. Let X be a non-empty connected compact complex manifold of dimen-
sion n. Suppose that there exist holomorphic vector fields E1, . . ., E" on X
which are linearly independent at each point of X. Show that there exist a
complex Lie group G and a discrete subgroup D ofG such that X is difieomor-
phic to G/D. (Let cm: be the holomorphic functions on X such that
[Eu 5,] = 2 tag,“ The ‘uk are constants because X is compact. Take G to be
the simply connected complex Lie group whose Lie algebra admits the a,” as
constants of structure and use Theorem 5.)
10. Let G be a. Lie group with a finite number of connected components.
For G to be unimodular, it is necessary and sufficient that Tr ad a = 0 for all
a e L(G) .
11. Let E be a complete normable space over C, v e!(E) andg = exp(v).
Suppose that Sp(v) n 2i1:(Z — (0}) = 0. Let E1, E2 be closed vector sub-
spaces of E which are stable under 0, such that E; C E1. Suppose that the auto-
morphism of El/Ea derived from g is the identity. Then 0(E1) C E,.
1} 12. Consider a real or complex complete normable space E and a closed
vector subspace F of E such that F0 admits no topological supplement in E'.1‘
(a) Let A (resp. B) be the complete normable space of continuous endo-
morphisms of E (resp. F). Let C be the set ofu e A such that u(F) C F. Let at be
the mapping a >—> (u, u | F) of G into A x B. Then at is an isomorphism of the
complete normable space C onto a closed vector subspace of A X B and ot(C)
admits no topological supplement in A X B. (Let x e E be such that ,1: ¢ F. Let

1‘ Let 5.; (rap. 1 1, I m) be the Banach space ofreal or complex sequences (1:, :2, . . .)
satisfying lim x“ = 0 (resp. Z [2“] < +00, resp. sup [a < +oo) with the norm
a“: n
“at” = sup [xnl (rcsp. Z |x,.|, rcsp. sup l‘nl)- There exists a continuous morphism
n
1: of l1 gnto :0; let F be its kernel. Then (:0)’ = 11 IS identified with F”. But a
vector subspace of countable type of 1“ cannot be a topological direct factor in 1‘”
(cf. A. GKO ur les
K Su ' ’ , d’espaces
du type C(K), Can. J. Math, 5 (1953), p 169).
385
 III LIE GROUPS

EE Fn be such that (x. Z) = 1. If 1) e 13', let i denote the element y H (y, 11)::
of A. Suppose that there eidsts a. projector 11: of A x B onto u(C). We define
17:: E’ —> E’ by 11(1)) = ‘(a‘1(n(fi, 0)))(E). Then 1': is a projector ofE’ onto F“,
which is absurd.)
(b) A unital real or complex complete normable algebra M is constructed

as follows: M = 46—90 M‘ is graded by the M‘; M0 is the set of scalar multiples of

l ; M1 is the set ofscalar multiples ofa non-zero element 11; M2 = {0} ; M3 = F ;
M4 = E; M‘ = {0} for i 2 5; wt = x for all xeMa. Let N be thecomplete -
uormable space of continuous endomorphisms of the complete normable
space M. Let Nl be the set of continuous derivations of M. Show, using (a),
that Nl has no topological supplement in N.
(L) Show that the group of bicontinuous automorphisms of M is a Lie
quasi-subgroup of GL(M), but not a Lie subgroup of GL(M). (Use Corollary
2 to Proposition 18 and (b).)
13. Let G be a real Lie group, L = L(G), and 4:: L —> G a mapping which
is differentiable at 0, such that T0012) = ML and such that (12(71):) = MK)” for
all x e L and n e Z. Then 4) = expg. (Let V be a neighbourhood ofO in L and
W a neighbourhood of: in G such that 0 = exp, I V is an analytic isomorphism
of V onto W. Let 41 = 9'1 o (11> I ¢‘1(W)). Then To(¢) = IdL, whence
4; = IdL using the equality 416x) = %¢(K) for x sulficiently close to 0 and
n e Z — (0}.
14. Let «1 be the commutative Lie algebra R‘, identified with C”. Let
aa be the commutative Lie algebra R. Let 4) be the homomorphism of 02 into
Der(a1) which maps 1 to the derivation (z,, 22) >—> (1‘21, i\/2_12). Let a be the
semi-direct product of a, by al corresponding to <1).
(4) Show that (Int 1:) I 111 contains, for all <1) 6 R, the automorphism

(21) 12) H (twin G‘Jiozzl-

(b) Show that the closure G of (Int 0) | a1 in GLO“) contains, for all
((1), o’) e R“, the automorphism (21, 12) i—> (eld’zl, 2‘0'22).
(a) Show that L(G) contains the endomorphism u: (2,, 2,) n—> (zl, 0) of a1
and that there exists no at e a such that u = (ad 2:) I a].
(d) Deduce that Int(a) is not closed in Aut(a).
15. Show that the finite-dimensional simply connected real Lie group of
§3, Exercise 7 ((1) contains non-simply connected Lie subgroups (it contains
in fact Lie groups isomorphic to U).
1] 16. (a) Let g be a complex Lie algebra. Let g be the complex Lie algebra
derived from 5 using the automorphism 7t i—> X of G. Let g“ be the real Lie alge-

386
 EXERCISES

bra derived fiom g by restriction of scalars. Let g’ 3 g» be the complex Lie

algebra go ®n C. For all x e 9, we write '
f0!) = i0: ®1-(ix)®i)es’, 30‘) = i(x®l + (it) 831’) 69'.
Show thatf (resp. g) is an isomorphism of g (rap. §) onto an ideal in (resp. n)
of 9’ and that the ideals m, n are supplementary in g'. This defines projectors of
9' onto m, It Show that, for all x e g,f(x) = fix), g(x) = q(x).
(b) Suppose that g is finite-dimensional. Let G be the simply connected
complex Lie group with Lie algebra 9. Let G be the conjugate complex Lie
group and G0 the underlying real Lie group. Then L(G_) = g and L(Go) = go.
Let M (resp. N) be the simply connected complex Lie group with Lie algebra
m (resp. n). Then f defines an isomorphism 4) of G onto M, g defines an iso-
morphism Y of G onto N and the simply connected complex Lie group G’ with
with Lie algebra 9’ is identified with M x N. Show that the real integral
subgroup of G' with Lie algebra go is closed and simply connected and is
identified with G0. The restriction to G C M x N of pr; (reap. prz) is the
isomorphism <1) (resp. y) of G (resp. G) onto M (resp. N).
(a) Deduce from (b) and Remark 2 of no. 10 that G', with the canonical
injection of Go into G', is the universal compladfication of Go.
(d) Let H be a connected complex Lie group with Lie algebra g, I-I the
conjugate complex Lie group and H, the underlying real Lie group, such that G
is the universal covering space of H. Let K be the (discrete) kernel of the can-
onical morphism of G onto H. Show that K is a central subgroup of G'. The
canonical injection of Go into G' therefore defines an injection i of Ho into
G’/N. Show that (G’IN, i) is the universal complexification of Ho. (Note that
this ordered pair has the universal property of Proposition 20.)
(e) We write G’/N = (Holc- Deduce from (d) a canonical morphism ii of
(Hole onto H X 171, which is a covering mapping. Show, by the example
H = C‘, that L]; is not in general an isomorphism.
17. (a) Let G, G, Go be as in Exercise 16 (I?) Let V be a finite-dimensional
complex vector space and p an irreducible linear representation of Go on V.
Show that there exist finite—dimensional complex vector spaces X, Y, an analy-
tic irreducible linear representation a (resp. 1') of G (resp. G) on X (resp. Y)
and an isomorphism of V onto X ® Y which transforms 9 into 1: ® 1. (Use
Exercise 16 (11), Proposition 2 ofChapter I, § 2, and the Corollary to Proposition
8 of Algebra, Chapter VIII, § 7.)
(b) Show that the conclusion of (a) is not necessarily true if we omit the
hypothesis that G is simply connected. (Consider the complex Lie group
0*.)
18. Let A be a finite-dimensional connected commutative complex Lie
group with Lie algebra a; let A be the kernel of exp,“ so that A is identified
with a/A.

387
Ill um GROUPS

(a) The following conditions are equivalent: _ . ‘

(al) The canonical mapping C ®z A ——> a is aCCte.
(a2) A is isomorphic to a Lie subgroup of some (0‘)“.
(a3) A is isomorphic to a group (0‘)” x C”. _ .
(a4) A has a finite-dimensional faithful complet hnear representation.
(a5) A has a finite-dimensional semi-simple faithful complex linear
representation with closed image.
(b) The following conditions are equivalent:
(bl) The canonical mapping C (831 A ——> a is surjective.
(b2) A is isomorphic to a quotient of a group (0‘)“.
(b3) No direct factor of A is isomorphic to 0.
(b4) Every complex linear reprsentation of A is semi-simple.
(c) The following conditions are equivalent:
(CI) The canonical mapping C®z A —> a is bijective.
(c2) A is isomorphic to a group (C*)".
(:1) Let F be a finite subgroup of A and let A' = A/F. Show that A satisfies
conditions (a‘) (reap. ((1.), (0.)) if and only if A’ does.
19. Let G be a real Lie group. Show that there exists a neighbourhood V of 0
in L(G) such that, for x, y in V:

«poo + y» _— “Lug-w
~
(exp ‘5u up ’1!) "
ll

. 1x ——tx — "’
eXPU’Dr. 11]) = ”1131” (61p ;~ap%~wp 7422:]: 71y)
uniformly for ta [0, 1].
20. Let G be a Lie group, H a Lie subgroup ofG and A an integral subgroup
of G such that L(H) n L(A) = {0}. Show that H n A is discrete in the Lie
group A.
21. Let G be a real Lie group and H a normal l-dimensional integral sub-
ST°UP~
(a) If H is not closed, Fl is compact (Spectral Theories, Chapter 11, § 2,
Lemma 1), hence isomorphic to T" and hence central in G if G is connected
(use General Topology, Chapter VII, §2, Proposition 5).
(b) Suppose that H is closed. Let a be an element of G and C(u) the set of
elements of G which commute with a. Suppose that H 4: (1(4).
IfH is isomorphic to R, then C(a) n H = {3). (Consider the automorphism
at: h>—>o"ha of H.) If H is isomorphic to T, C(a) n H has two elements.
(Consider on again and use General Topology, Chapter VII, § 2, Proposition 6.)
The second situation is impossible if G is connected (use (a)).
(a) With the hypotheses of (b), suppose further that G/H is commutative.

388
 EXERCISES

Then G = C(a) .H. (Observe that the mapping hHh‘laUl) of H into H is

surjective.)
22. Let G be a real Lie group, H a. normal 1-dimensional integral subgroup
and A a closed subgroup of G such that AH is not closed in G. Then H is
central in the identity component ofA—H. (Reduce it to the case where G = m
and G is connected. Then the set B of elements of A which commute with H is
a normal subgroup of G. Passing to the quotient by B, reduce it to the case
B = {e}. As every commutator in G commutes with H, A is then commutative.
By assuming that H is closed and using Exercise 21 (0), show that AH would be
closed, whence a. contradiction. Hence H is not closed and Exercise 21 (a) can
be used.)
11 23‘ Let G be a real Lie group and c = (U, 4), E) a chart on the manifold
G centred on the identity element .9. If x e U, let [5:], denote the norm of Mr)
in the Banach space E‘
(a) Show that, if c’ = (U’, (If, E') is another chart centred on e, there exist
constants A > 0 and u > 0 such that
”1:40 s [n < la
for all x e G sufficiently close to a.
(b) Show that, for all p > 0, there exists a neighbourhood U, of a con-
tained in U such that, for my in U,, (1,31) 6 U0 and
“All”: S P-infllxlcr Illlcl-
(0) Suppose that G is finite-dimensional. Let I‘I be a discrete subgroup of G.
Apply (b) with 0 < p < 1 and choose U, to be relatively compact. The set
Ufl fl 1" is finite. Let
{c = Ym'Ynu-m}
be its elements, numbered so that
lYolc < lYllc < - - - $ I‘rmlc-
Show that, if i S m and j s m, the commutator (7., 7,) is equal to one of the
7,, with k < inf(i, ]) Deduce that the subgroup generated by U, n I‘ is nil-
potent of class gm.
Deduce from the above the existence of a neighbourhood V of a such that,
for every discrete subgroup I‘ ofG, there exists a nilpotent integral subgroup N
of G containing V n I‘.
(d) Suppose that G is connected and finite-dimensional and contains an
increasing sequence of discrete subgroups Dn whose union is dense in G. Then
G is nilpotent. (Using (5), prove the existence of a central one-parameter sub—
group H which meets D” for sufficiently large n at a point distinct from :.
Argue by induction on dim G, distinguishing two cases according to whether
H is closed or relatively compact (Exercise 21 (a)).)

389
ll! LIE GROUPS

24. Let G, G’ be finite-dimensional real Lie groups,fa surjecn've morphism

of G into G’, N its kernel, H an integral subgroup of G and H’ = f(H).
(a) Suppose that N is finite. For H’ to be closed, it is necmsary and sufficient
that H be closed.
(b) Suppose that N is compact. If H is closed, H’ is closed.
{I 25‘ Let G be a finite-dimensional real or complex Lie group. Let
S C L(G). Suppose that L(G) is the Lie algebra generated by S and that S is
stable under homotheties.
(a) Let H be the subgroup of G generated by exp S. Show that H is open in '
G. (Let A be the set of x e L(G) such that exp(Kx) C H and B the vector sub-
space of L(G) generated by A. Show that [A, S] C A and then that [A, A] C B.
Deduce that B is a subalgebra ofL(G) and then use Proposition 3 of § 4.)
(b) Suppose further that
(x e S andg/ e S) 2 ((Ad exp x)(y) e S).
Then the vector subspace V of L(G) generated by S is equal to L(G). (Using
(a), show that [L(G), V] C V.)
26. Let G be an n-dimensional cemented compact complex Lie group, X a
finite-dimensional connected complex analytic manifold and (g, x) n—> gx a law
of analytic left operation of G on X. For all geG, let p(g) be the mapping
1: ~—> gx of X into X. Suppose, that, for all g e G distinct from 2, p(g) ye Idx.
Then every orbit of G in X is an n-dimensional closed submanifold of X. (Let
x e X, H be the stabilizer of x and H’ the identity component of H. For all
g e H’, let u(g) be the tangent mapping at x to 9(g). Arguing as in Proposition
6 of no. 3, show that u(g) is the identity for all g e H’. Using § 1, Exercise 4 and
the connectedness of X, deduce that H’ = {3}.)
11 27. Let G be a compact real Lie group, M a. compact manifold of class C}2
and J an open interval of R containing 0. Let (x, (x, 5)) 9—» (m;(:, x), E) be a
mapping of class C.“ of G X (M x J) into M x J under which G operates on
M x J on the left.
(a) Let X be a vector field of class C‘ on M X J such that, for all
(x, E) e M X J, the second projection of X0“, is the tangent vector 1 to J.
Translating X by G and integrating over G, deduce the existence of a field
X’ with the same properties as X, which is moreover invariant under G.
(b) Show that there exists a difleomorphixm (x, E) v—> (h:(x), E) of M x J
onto itself such that
(i) for all E EJ, h! is a diflmmorpfism of M onto M;
(ii) m;(:, x) :1 h5(mo(:, h{‘(x))) for 3 e G, x e M, EeJ.
(Use (a) and Theorem 5 of no. 8.)
28. Give an example of a finite-dimensional real Lie group G and two
integral subgroups A, B of G such that L(G) =L(A) +L(B) but AB is not a
subgroup of G (cf. Integration, Chapter VII, § 3, no. 3, Proposition 6).

390
 EXERCISE

29. Let G be a finite—dimensional complex connected Lie group, Go the

underlying real Lie group and H an integral subgroup of G0. Show that there
exists a smallest integral subgroup H* of G containing H. Give an example
where H is closed in Go but H‘ is not closed in G (take G = (JR/Z“).
30. Let G be a compact connected complex Lie group and hence of the form
C"/D, where D is a discrete subgroup of C“ of rank 21:.
(a) Show that every holomorphic differential l-form m on G is invariant.
(Let 7:: C“ —> G be the canonical morphism. Let (1, . . ., L. be the coordinate
functions on G“. Then 1t*(m) is of the form Z’Za/itfl where the a, are holo-
morphic functions on C'I which are invariant under D and hence constant.)
(1:) Let G' = C"/D', where D' is a discrete subgroup of C” of rank 2n.
Show that every analytic manifizld isomorphism u: G —> G’ is of the form
5 |—> (1(5) + a', where v is a Lie group isomorphism and a' e G'. (Let n': C” —> G'
be the canonical morphism. There exists an isomorphism of analytic manifolds
i: C" —> C" such that u o 1': = n’ o i. For every holomorphic differential l-form
m’ on G’, u*(1t’* (w’)) is a l-form on C" which is invariant under translations by
(a). Deduce that 12 is an affine mapping.)

§7
1. Let G be a Lie group and (b: U -> G an exponential mapping such that
ZU C U and ¢(1x)= ¢(x)'fora.lleandallreZ. If]; > 0, :1) is ananalytic
isomorphism of U onto (MU). (If (Mac) = (My), then ¢(p”x) = (ME?) for all
n E N and hence x = y. Let W be an open neighbourhood ofO in L(G) such that
¢_1 is analytic on (MW). For all r e (MU), there exist I: e N and a neighbour-
hood V cfr in ¢(U) such that :eV => I" etI>(W).)
2. Let G be a Lie group, (I): U—>G an exponential mapping with the
properties of Proposition 3, V a neighbourhood of 0 in U such that ZV C V
and 4/: V —> G a tangent mapping at 0 to (I) such that Mm) = Ms)" for all
neV and all 7162. Up > 0, then 4’ = 42 |V. (Give L(G) anorm. Lete.
Then 9’s tends to 0 as 71 tends to +00, hence there exists an" > 0 such that an
tends to o and I <4>-1 ., mu) — M s anllwrll- But Wx) = W)“,
whence M" a door) — :41 s aux .)
3. Let U be the set of invertible elements of A and U’ = 1 + m C U.
(a) Show that U' C U,. (ID: = I + y withy e m, then x”" tends to l as It
tends to +00 by the binomial theorem.)
(b) Show that U, is the set of elements of U whose image in A/m is a root
of unity. (Use (11).) Hence recover the fact that U = U, when K is locally
compact (A/m is then finite).

39l
I]! LIE GROUPS

4. Let n E N“, p be a prime number and G the set of matrices belonging to

GL(n, 2,) all of whose elements are congruent to 1 modulo [I ifp # 2 (resp.
modulo 4 if]; = 2). Then G is an open subgroup of GL(n, 21,).
(a) Show that G has no element of finite order #1.
(b) Show that every finite subgroup of GL(n, Z”) is isomorphic to a sub-
group ofGL(n, Z/pZ) ifp 94 2 (resp. GL(n, 2/42) if]; = 2).
1T 5‘ (a) Let I‘ be a compact subgroup of G = GLO], Q9). Show that there
exists a conjugate of l‘ which is contained in GL(n, Z”). (If T is a lattice of.
Q; with respect to Z, (Commutative Algebra, Chapter VII, §4, Definition 1),
show that the stabilizer of T in l" is open in I‘ and hence of finite index and
2l" 7T is a lattice which is stable under 1".)
1m(b) Deduce that G, is the union of the conjugates of GL(n, 2,).
(c) Deduce that every finite subgroup (I) of GL(n, Q,) has order a divisor
of (1,,(11), where (1,,(p) is defined by
W) = (pa —1)(p" —12) . . .(r _,,»-1) ifp as 2
an(p) = 2”"(2’| — l)(2" — 2). . .(2" —— 2"“) if]: = 2.
(Using (a), it can be assumed that (I) C GL(n, _Z,,). Then use Exercise 4-.)
(d) Show that every finite subgroup of SL(n, (1,) has order a. divisor of
my), where Mp) = :91 forp ,t 2 and 3,.(2) = #
1T 6. In this exercise, I denotes a prime number #2. If a is an integer #0,
let 111(4) denote the l—adic valuation of a, i.e. the largest integer : such that
a E 0 (mod 1‘).
(a) Let m be an integer 2 1. We write
6(1, m) = 0 ifm ;é 0 (mod (I — l))

3(1, m) = ”'(1 L" 1) +1 ifm a 0 (mod([ — 1)).

Show that, if x is an integer prime to I, then
v,(x"‘ — I) 2 5(1, m)
and that there is equality if the image of x in the cyclic group (Z/I’Z)* is a
generator of this group.
([1) Let n be an integer 2 1. We write
n
r(l, n) = "2:1 2(1, 7»).
Show that

392
[—1[—][—]
 EXERCISES

where the symbol [(1.] denotes the integral part of the real number at. Show
that, if x is an integer prime to I, then
0,((x" — l)(:c"‘1 — l). . .(x — 1)) 2 r(l,n)
and that there is equality if the image of x in (Z/I‘Z)* is a generator of this
group.
(5) In the notation of Exercise 5 (0), show that 170.») is the greatest power
of l which divides all the 111(0), for p prime #1 (or for all sufficiently large p,
which amounts to the same). (Use ([1) applied to x = p; then choose [a such
that its image in (Z/I’Z) “ is a generator of this group, which is possible by the
arithmetic progression theorem.1')
(11) Let I‘ be a finite subgroup of GL(n, Q) and let I ' be the greatest power of
I dividing the order of I‘. Show that e S r(l, n). (Use Exercise 5 to prove that
l' divides all the a,.([1) and then apply (c) above.)
(2) Conversely, show that there exists a finite l-subg‘roup I‘M of GL(n, Q)
whose order is 1"“). (Reduce it to the case where n is of the form l“(l — l),
with a 2 0. Decompose Q” as a sum of l“ oopia of OJ" and use this decom-
position to give an operation on Q" of the semi—direct product H,_,, of the
symmetric group 5” and the group (Z/IZ "‘. Take Pl.» to be a Sylow l-group
of H. ,,.
Show) that, if n is even, I‘M, is contained in a conjugate of the symplectic
group 51201, 2).
(f) *Let I" be a finite subgroup ofGL(n, Q) which is an l—group. Show that
l" is conjugate to a subgroup of l- (Show first, using a suitable reduction
modulo 1;, that the reduction of I‘ is a finite subgroup of the reduction of a con-
jugate of I‘M; then use the character of the representation of I‘ on Q”.) In
particular, every subgroup of GL(n, Q) of order l"""’ is conjugate to I‘M”
fl 7. In this exercise, let 122(a) denote the 2-adic valuation of an integer n.
(a) Let I‘ be a finite subgroup ofGL(n, Q). Show that there exists a positive
definite quadratic form with coefficients in Z, which is invariant under 1".
Deduce (by the same argument as in Exercises 4 and 5) that, for all sufficiently
large p, l" is isomorphic to a subgroup of an orthogonal group 0(a) over the
field F, (resp. a subgroup of 50(n) if I‘ is contained in SL(n, (2)).
(b) Suppose that I‘ is contained in SL(n, Q) and let 2‘ denote the greatest
power of 2 dividing the order of I‘. Show that, if n is odd, 2‘ divides all the
integers
bn(11)=(l”‘“ —1)(17”'° -1)---(202 - l)
for sufficiently large prime [7. (Use (a) and Exercise 13 of Algebra, Chapter IX,
w
1‘ For a proof of this theorem, see for example A. SELEERG, Ann. of Math, 50
(1949), pp. 297—304.

393
III LIE GROUPS

Show that, if n is even and p is a sufficiently large prime, 2’ divides the

l.c.rn. bum) of the integers
2 _1)/(pm +1)
(9“ ‘1)(P"‘a — l). ..(p
‘ 1)(i0"’2 — l)...(1;2 _1)/(pm _ 1).
(9"
(Same method as for n odd.)
(0) Under the hypotheses of (b), we write

r(2,n) =n+ [g] + [2] + =n+v,(n!).

Let A be an integer 23. Show that 2‘3”"l is the greatest power of 2 which
divides all the bn(p) for p 2 A. (Same method‘l‘ as in Exercise 6; use the
existence of a prime number p Z A such that p E 5 (mod 8).) Deduce the
inequality 5 < 7(2, 71) —— l.
(d) Conversely, let C,I be the subgroup of GL(n, Z) generated by the
permutation matrices and the diagonal matrices with coefficients 11. The
order of C" is 2"nl and 1/,(2‘71!) = r(2,n). If F3”. denotes the intersection of
a Sylow 2-group of C,I with SL(n, Z), deduce that the order of F3,” is
gram-r
8. Let n be an integer 21. We write

M(") = U 1r(l.n),

where the product is taken over all prime numbers I and the 7(l, n) are defined
as in Exercises 6 and 7.
Then M0) = 2, M(2) = 25.3 = 24, M(3) = 2‘.3 = 48,
M(4) = 27.325 = 5760.
Deduce from Exercises 6 and 7 that the ltc.m. of the orders of the finite
subgroups of GL(n, Q) (or GL(n, 2), which amounts to the same) is equal
to M(n). Consider the same question for SL(n, Q) with M(n) replaced by
§M(n).
9. Suppose that K is locally compact. Let G = GL(u, K). Let G1 be the
set of g e G which leaves stable a lattice of K'I with respect to A. Let G2 be
the set of g e G which generate a relatively compact subgroup in G. Let Ga
be the set of gEG whose eigenvalues in an algebraic closure of K are of

T For more details on this exercise and on the preceding one, see H. mowsxr,
Garamm. AWL, Leipzig-Berlin, Teubner, 1911 (Ed I, S. 212—218) and W. Burmsins,
Theory 41/ groups qffmilz order (2nd «1.), Cambridge Univ. Pras, 1911 (pp. 479—
484).
394
 mam

absolute value 1. Then G1 = G2 = G; = G,. (Use the argument of Exer-

cise 5 (a).) '
10. Suppose that K is locally compact. Let G be a standard group of
dimension n over K. Let y. be a Haar measure on the additive group K".
Show that uIG is a left and right Haar measure on G. (Use the fact that G
is the inverse limit of the G(a,,) and Integration, Chapter VII, § 1, Proposi-
tion 7.)
§8
1. The mapping z-—>2 of G into C is a non-analytic continuous auto-
morphism of the complex Lie group C.
2. Let G be the Hilbert space of sequences ()1, X2, . . .) of real numbers
such that Z A? < +00. Consider G as a real Lie group. Let G,. be the set

of (A1, A2,...) EG such that Neil for l S m g n. The G,| are closed Lie

subgroups of G and hence H = Q G, is a closed subgroup of G. But this

group is totally disconnected and not discrete and hence is not a Lie subgroup
of G.
1] 3. In Q, x Q,” every closed subset can be defined by a family of analytic
equations. Deduce that Corollary 2 (ii) of Theorem 2 is no longer true if the
hypothesis that I is finite is omitted.
fl 4-. Let G be a finite—dimensional real Lie group and A a subgroup of G.
We say that an element x of L(G) is A-aocessible if, for every neighbourhood
U of a, there exists a continuous mapping a: of [0, 1] into A such that «(0) = s
and «(1) e exp(tx) .U for 0 S t < 1. Let I) be the set of A-accessible elements
of L(G).
(a) Show that I) is a Lie subalgebra of L(G), (Use Exercise 19 of§ 6.)
(b) Let H be an integral subgroup of G such that L(H) = I). Show that
H C A. (Let I = [— l, 1] and let R’ be given the Euclidean norm. Let
(x1, . . ., x,) be a basis of 1). Construct continuous mappings a1, . , ., a, of I
into A and continuous mappings f1: . . .,fr from I’ to R, such that, for
t= (In. ..,t,) 61',
1101) ~ ~ 4'01) = exP(f1(t)4\‘1) ~ ~ ~ exP(fv(t)#r)
ll‘ — (fl(!),~ufy('))|| < f
Then apply the following theorem: let f be a continuous mapping of I“ into
R’ such that H f (x) — x" < a}— for all x e I'; then f (1’) contains a neighbour-
hood of 0 in R'.1')
1’ This follows from the Brouwcr fixed point theorem, which can be consulted,
for example, in N. DUNFORD and J. T. SCHWARTZ, Limar operators, part I (Inter-
science publishers, 1958), pp. 467—470.

395
Ill LIE GROUPS

(0) Deduce that b is the subalgebra tangent to A at t.

(d) Show that if A is arcwise connected, then A : H.1‘
11 5. Let G be a Hausdorff topological group, H a closed subgroup of G
and 7: the canonical mapping of G onto G/H. Suppose that H is a finite-
dimensional real Lie group. There exist a neighbourhood U of «(2) in G/H
and a continuous mapping a of U into G such that it o o- = Id“. (Let p be
an analytic linear representation of H on GL(n, R), which is locally a homeo-
morphism (§ 6, Corollary to Theorem 1). Let f be a continuous function >0
on G, equal to l at 2 and zero outside a sufficiently small neighbourhood of e.
Let dr be a left Haar measure of H. For x 5 G, we write

g(~) = if(xx)9(x)"dseMn(R)~
Then govt) = g(z)p(t) for x e G and 16 H. IfV is sufficiently small,
30‘) 6 GM": R)
for x sufl‘iciently close to 9. Finally use the fact that the theorem to be proved
is true locally for GL(n, R) and p(H).
6. Let G be a group of class 0' (§ 5, Exercise 1) with r 2 2. There exists
on G one and only one real Lie group structure S such that the underlying
manifold structure of class C" of S is the given structure. (The uniqueness of
S follows from Corollary 1 to Theorem 1. Let L(G) be the normable Lie
algebra associated with G in Exercise 2 of § 5. There exist a real Lie group
genn G' and an isomorphism h of L(G’) onto L(G) (§ 4, Theorem 3). Verify
as in §4, no. 1 that there exist a symmetric open neighbourhood G" of 26
in G’ and a mapping ([2 of class C' of G" into G such that Tabb) = h and
¢(g1gfl) = ¢(g1)¢(gg) for g1, g2 in 0”. By shrinking G”, it can be assumed
that V = MG") is open in G and that (i) is an isomorphism of class C' of the
manifold G” onto the manifold V. There therefore exists on V a real Lie
group germ structure such that the underlying manifold structure of class
C" is the given structure. For all g E G, Int g defines as analytic mapping of
Vn(g'1Vg) onto (n") nV (Theorem 1). By Proposition 18 of § 1,
there exists on G a real Lie group structure S inducing the same analytic
structure as V on an open neighbourhood of e. By translation, the underlying
manifold structure of class C' of S on G is the given structure.)
7. Let G be a real Lie group and H a closed subgroup of G.
(a) Let b be the set of x6 L(G) such that exp(tx) EH for all teR. Then
I) is a Lie subalgebra of L(G). (Use Proposition 8 of § 6.)

T For more details, cf. M. Gm, On an arcwise connected subgroup of a Lie

group, Pm. Am. Malh. Sm, 20 (1969), pp. 157—162.

396
 EXERUSES

(b) Suppose that H is locally compact. Show that H is a Lie subgroup of

G. (Show first that I) is finite-dimensional by proving the existence in b of a
precompact neighbourhood of 0. Then imitate the proof of Theorem 2,
choosing Va such that (exp V2) 0 H is relatively compact.)

§9
1. Let G = 80(3, R). Let A1, A2 be two orthogonal lines in R3 and H.
the subgroup of G consisting of the rotations about A, (i = 1,2). Then
[L(Hl), L(Hfl)] is a l—dimensional Lie subalgebra of G distinct from the
Lie subalgebra tangent to (H1, H2) at a.
2. Suppose that K is ultrametric. Let G be a finite—dimensional Lie group,
9 its Lie algebra and A a finite subset of G. Then ZG(A) is a Lie subgroup of
G with Lie algebra MA). (Argue as in Proposition 8.)
3. Let G be a connected real or complex Lie group. The centre Z of G is
a Lie quasi-subgroup of G and L(Z) is the centre of L(G).
4. Suppose that K is ultrametric and p > 0 (in the notation of § 7). Let
G be a finite-dimensional Lie group, A a group of automorphisms of G and
B the corresponding group of automorphisms of L(G). Let GA (resp. L(G)“)
be the set of elements of G (reap. L(G)) which are fixed under A (resp. B).
Then GA is a Lie subgroup of G with Lie algebra L(G)“. (Use the logarithmic
mapping)
5. Let G be a finite—dimensional connected real or complex Lie group.
Let (G0, G1, . . .) be the upper central series of G (Chapter II, §4, Exercise
18) and let (go, 9,, . . .) be the upper central series of the Lie algebra L(G)
(Chapter I, § 1, no. 6). Then, for all i, G. is a Lie subgroup of G such that
L(G.) = 9..
6. Let 1" be the 3-dimensional simply connected nilpotent real Lie group
defined in Exercise 5 (b) of §4. Let ueR be an irrational number. Let P
be the discrete subgroup of I‘ x R consisting of the ((0, 0, at), out), where
It 62. Let G = (I‘ x R)/P. Show that (G, G) is not closed in G.
7. (a) Let G be a semi-simple connected real Lie group, Z its centre and 9 a
continuous linear representation of G on a finite—dimensional complex vector
space. There exists an integer p such that, for all z e Z, 9(2) is diagonalizable
and all the eigenvalues of 9(2) are p—th roots of unity. (It can be assumed
that 9 is irreducible. Then 9(2) is scalar by Schur’s Lemma. On the other
hand, det 9(g) = l for all g EG because G = 9G.)
(b) Deduce that, if G admits an injective finite-dimensional continuous
linear representation, Z is finite.

397
11! LE GROUPS

1T8. Let G be a finite-dimensional connected real Lie group, 9 its Lie

algebra, n the largest nilpotent ideal org and b = [9, 9] + n.
(a) I) is a characteristic ideal of g; the radical of b is n; for all
x E I), Tr adbx = O; for every Levi section [of g, b = l + n.
(b) Suppose that G is simply connected. Let Z be its centre, H the Lie
subgroup of G with Lie algebra b and (b the canonical morphism of G onto
G/H. Then 4>(Z) is discrete in G/H. (The group G is the semi-direct product
of a semi-simple Lie group S and its radical R. Let 2 E2. Then 2 = y‘lx,
where x e R and y belongs to the centre of S. There exists an integer [I such
that the eigenvalues of Adgy, and hence also of Adnx, are p-th roots of unity
(Exercise 7). Deduce that, if N is the Lie subgroup of G with Lie algebra n,
there exists a neighbourhood U of e in R satisfying U = UN = NU,
Zn (sv) c SN = H.)
(c) We no longer assume that G is simply connected. Let H be the integral
subgroup of G with Lie algebra I). Show that H is a unimodular normal Lie
subgroup of G with nilpotent radical, such that G/H is commutative. (Use
(a)(d)andDeduce
(b)-) that, if (G, G) is dense in G, the radical of G is nilpotent.
9. Let G be the universal covering of SL(2, R). We identify the kernel
of the canonical morphism G —>SL(2, R) with Z (§ 6, Exercise 2). Let a be
an element of '1‘" whose powers are everywhere dense in '1'" (General Toyolngy,
Chapter VII, § 1, Corollary 2 to Proposition 7). Let D be the discrete sub-
group ofG x T" generated by (l, a). Let H = (G x T")/D. Then
L(H) = N2, R) x R"
and the integral subgroup H’ of H with Lie algebra 3K2, R) is isomorphic to
GanddenseinH.D"H' = H'foralln 2 0.

1] 10. Let G be a finite-dimensional real or complex Lie group. Let

p = dimtuc), L(G)].
Let (G, G) be given its structure as an integral subgroup of G. There exists a
neighbourhood V of z in (G, G) such that every element of V is the product
ofp commutators of elements of G. (Let x1, y1, . . ., x,” y, be elements of L(G)
such that the [x., 3].] form a basis of L(G). For x, t in R, we write
Ma 1) = (64» w.) "(eXP Ix.) “(EXP 5%) (CXP Ix.)-
Apply the implicit function theorem to the mapping
(‘1: 1'1, - - -7 5w ’p) H Pi(—“n ’1) - - - 9903» ’11)
oil!” into (G, G).)

398
 EXERCISES

11. Let G be the semi—direct product of B: 2/22 by K corrsponding to

the automorphism x>—> —x of the Lie group K. Then L(K) is an ideal of
L(G) and L(B) = {0), but (K, B):

12. Let (:1, 22, 9;) be the canonical basis of K3. Consider the nilpotent Lie
algebra structure on K3 such that [51, 3,] = 23, [01, c3] = [92, :3] = 0. With
the associated group law on K3 (no. 5),

(fly: Z)(""y’: 1') = (7‘ + fly + y', z + 1' + ‘}(W' - W»-

Thc mapping
(A, H, V) H (exp MXeXP peak-KP v01 + ‘3»
of K“ into G is neither surjective (show that (0, 1, l) is not in the image)
nor injective (show that (0, l, 0) and (1, l, —-l) have the same image).

13. (a) A multiplication is defined on R:5 as follows:

(x,y, z).(x’,y’, z') = (x + x'cosz —y'sinz,y + x'sinz +y'cosz,z + 2’).

Show that a solvable real Lie group G is thus obtained such that
DG = R“ x {0}. The centre Z of G is {0} x {0} x 21:1.
(1') For (x, y, z) E G, let n(x, y, 2) be the mapping
(x,p.)»—>(Acosz- psinz+x,ksinz+p.cosz+y)
of R2 into R“. Show that 1!: is a morphism of G onto a Lie subgroup G’ of the
affine group of R7, generated by the translations and rotations of R’. Show
that Ker r: = Z.
(a) We canonically identify L(G) = T(o.o.0)G with R”. Let (:1, 2,, ‘5) be
the canonical basis of R“. Show that [21, :2] = 0, [93, 21] = 2,, [em :9] = -—:,.
((1') Show that, for 5 9E 0,

expo(a, [7, c) = (10(asinc + b cost — 11),; (—acosc + bsinc + a),t:)~

Deduce that, for all u eL(G), Z C exp(Ru). Show that expo: L(G) —> G is
neither injective nor suijective.

14. (a) Let G = GL(n, C). Show that expa is surjective. (Use the holo-
morphic functional calculus of Spam! Theories, Chapter I, §4, no. 8.)
(b) Let G’ = SL(2,C). Show that, if uses, the element (‘01 :1) of
G’ does not belong to the image of expfiu

399
III LIE GROUPS

15. (a) m xeél(2,R) and A = detx Show that

=ch\/— A.I+S——h'_-A~Ax ifA<0

a: cos1/_.I+sm—‘:—/—Z—: “>0.
0 . .
(b) Let H = SL(2,R). Show that g = (0 )‘_1)EH is the image of exp];
if and only if 7‘ > 0 or A = -1. If A = 0, all the one-parameter subgroups
containing g are equal.
16. Let G be a finite-dimensional Lie group. Show that there exists a
basis (x1, . . ., x“) of L(G) such that, for all i, exp(Rx,) is a Lie subgroup of G.
(Use Spectral Theories, Chapter II, §2, Lemma 1.)
1[ 17. (a) Let c denote a real solvable Lie algebra with a basis (a, b, 0)
such that [a, b] = :, [a,c] = —b, [b,c] = 0. Let b denote a real solvable
Lie algebra with abasis (a, b, c, 11) such that [a, b] = c, [a, c] = —b, [17, t] = d,
[a, d] = [[7, d] = [0, d] = 0. Let g be a real solvable Lie algebra. If 9 con-
tains non-zero elements a, y, 1 such that [x,y] = z and [x, z] = —_i/, then
g contains a subalgebra isomorphic to c or D. (Write a, = y, b, = z, :1 = [y, z] ;
define do bu 0: Inductively by a‘= [‘11— 1: ‘I— 1]: bl = [53— 1) [I— 1]; 5i: [11,, [’41
Consider the smallest integer I: such that 6,, = 0.
(b) Let g and 5* be real solvable Lie algebras and (l) a homomorphism of
9 onto g“. If 3* contains a subalgebra isomorphic to c or b, g has the same
property.
(e) Let I) be a complex solvable Lie algebra. Consider a Jordan—Holder
series for the adjoint representation of I); the quotients of this series define
l-dimensional representations of I) and hence linear forms on [7. These linear
forms, which depend only on I), are called the root: of I). If I)’ is a real solvable
Lie algebra, the roots of l)’ are the restrictions to l)’ of the roots of b’ ®n C
Then let G be a finite—dimensional simply connected solvable real Lie
group and 9 its Lie algebra. Show that the following conditions are equiva-
lent: (u) expa is injective; (fl) expG is suijective; (7) e2t is bijective; (8) expG
is an isomorphism of the analytic manifold L(G) onto the analytic manifold
G; (e) L(G) contains no subalgebra isomorphic to c or b; (0 there exists no
quotient algebra of L(G) admitting a subalgebra isomorphic to c; (1;) every
root of g is of the form <1; + io’ where (l), 42’ are in 9* and di’ is proportional
to 4); (0) for all are 9, the only pure imaginary eigenvalue (in C) of adx
is 0.T
t For more details, of. M. SAITO, Sur certains groupes de Lie resolubles, Sci.
Paper: af the College nf General Educaliun, Univ. of Tokyo, 7 (1957), pp. 1—11 and
157468.

400
 EXERCISES

18. Let G be a connected real Lie group, Z its centre, 20 the identity
component of Z, 9 the Lie algebra of G and a the centre of 9. Then Z and Z0
are Lie quasi-subgroups of G with Lie algebra 3 (Exercise 3). An acceptable
norm on g is a norm defining the topology of g and making 9 into a normed
Lie algebra.
(n) For all r > O and z e 3, there exists an acceptable norm on 9 whose
value at z is <r. (Let g be an acceptable norm on 9. Show that, for suitably
chosen A > 0, the function x» |[x|| = Aq(x) + 1nl'q(x + y) has the required
properties.)
(b) Show that the following conditions are equivalent:
(i) 20 is simply connected;
(ii) for every acceptable norm on g, the restriction of expg to the open
ball of centre 0 and radius 1: is injective;
(iii) there exists 1 > 0 such that, for every acceptable norm on g, the
restriction of exp‘; to the open ball of centre 0 and radius r is injecfive.
(To prove (iii) => (i), use (a). To prove (i) => (ii), suppose that x, y are
in a, M < w, M < 7‘, x #y, expax = upay- Then up adx = EXP Edy
and hence ad x = ad y by Proposition 17 of § 6 applied to a. complexification
of 9. Hence there exists a non-zero z in a such that expuz = e; it follows
that (i) is false.)
(6) By considering the group G' of Exercise 13 (b), show that the con-
clusion of (b) is no longer one if u is replaced by a number > 1:. (In the nota-
tion of Exercise 13, use the norm at; + baa + as v—> (a2 + I?2 + £3)i on L(G’).)

19. Let G be a locally compact group and H a closed subgroup of G. Sup-

pose that H is a finitedimensional simply connected solvable Lie group and
that G/H is compact. Show that there exists a compact subgroup L of G
such that G is the semi-direct product of L and H. (Argue by induction on
dim H. Consider the last non-trivial derived group of H and use Integration,
Chapter VII, § 3, Proposition 3.)

20. Let G be a finite-dimensional solvable connected real Lie group. We

introduce the notation S, S’, F, e of § 6, no. 10, proof of Proposition 20.
(a) Using Proposition 21, show that a is an isomorphism of S onto a Lie
subgroup of the underlying real Lie group of S’.
(b) Deduce that the universal complexification G of G is identified with
S'/a(F) and that the canonical mapping of G rnto G rs an isomorphism of G
onto a Lie subgroup of the underlying real Lie group of G

{I 21. (a) Let G be a simply connected solvable real Lie group with the
following properties: (a) L(G), which is n-dimensional, has a commutative
ideal of dimension n — 1, corresponding to a subgroup A of G; (9) there
exists an element a of the centre of G which does not belong to A. Show that

401
m LIE GROUPS

there exists an element x of L(G) such that expat = cr. Show that L(G) is
the product of a commutative ideal and an ideal with a basis
(7‘; an bl! - - u a,” bk)

“Ch that "1» bn - - a ”In bl: belong to L(A); [’8 at] = 2“”tbo ll") bl] = ‘2‘"7‘4'1t
(n1 6 Z — (0}) for all i. Generalize the results of Exercise 13 (d) to G.
(b) Let G be a finite-dimensional simply connected solvable real Lie group.
Let D be a discrete subgroup of the centre of G. There exist a basis
(x1, x2, . . ., x") of L(G) and an integer r g n with the following properties: .
(at) every element of G can be written uniquely in the form
(exp tlxl) . . t (exp tux")
where :1, . . ., t,. are in R; (6) x1, . . ., x, are pairwise permutable and
(exp #1: . . ., exp x.)
is a basis of the commutative group D. (Argue by induction on the dimension
of G. Let a be maximal commutative ideal of L(G). Let A be the corre-
sponding integral subgroup. Then A is closed in G and DA/A is a discrete
subgroup of the centre of G/A, to which the induction hypothesis can be
applied, whence there are elements xi“, . . ., x; of L(G/A) and an integer :.
For 1 < i < r, let oi be an element of D whose class modulo A is exp xf.
Construct one by one, using (a), representatives x1,” .,x,,, of :4,” .,x;
such that exp x, = a, and [x,, :9] = 0 for l S i,j S r.)1'
(c) Deduce from (b) that every finite—dimensional connected solvable real
Lie group is homomorphie to a. space R” x T'“ (m, n integers 20).
22. Let G be a finite-dimensional connected real Lie group such that
L(G) is reductive. Then G = (V x S)/N where V is a finite-dimensional
real vector space, where S is a simply connected semi-simple real Lie group
and N is a central discrete subgroup of V X S. Then G/fiG is isomorphic
to V/prl N. Hence G/D‘N is compact if and only if pr, N generates the
vector space V.
23. (a) Let G be a finite-dimensional commutative real Lie group with
only a finite number of connected components. For G to be compact, it is
necessary and suflicient that every finite-dimensional analytic linear repre-
sentation of G on a complex vector space he semi-simple. (Use the proof of
Proposition 32.) ,
(b) Let G be a finite-dimensional commutative complex Lie group with
only a finite number of connected components. Let N be the kernel of exp“.

1‘ For more details, cf. C. CHEVALLEY, Topological structure of solvable groups,

Ann. 4mm, 42 (1941), pp. 668—675.
402
 MERGERS

For N to generate over C the vector space L(G), it is neceSary and sufficient
that every finite-dimensional analytic linear reprtsentation of G be semi-
simple. (Use (4), Lemma 1 and Proposition 32.)
24'. The Lie group SL(2,R) is connected and almost simple, but (I, —I)
is a normal commutative subgroup of SL(2, R).
25. Let G be an almost simple connected real or complex Lie group. Let
A be a normal subgroup of G. If A # G, A is discrete and central. (Use
Exercise 8 of § 4'.) Therefore the quotient of G by its centre is simple as an
abstract group.
1I26. Let G be a finite-dimensional connected real Lie group. Suppose
that G admits a finite-dimensional injective continuous linear representation
p. Then (G, G) is closed in G. (Let R be the radial of G and S a maximal
semi-simple integral subgroup of G. Using Exercise 7 (b), reduce it to the
case where G is the semi-direct product of S and R. By Chapter I, § 6, Proposi-
tion 6, p is unipotent on (G, R). Hence p((G. R)) is closed in the linear group
and therefore (G, R) is closed in G.) '
{[27. Let G be a finite-dimensional solvable simply connected real Lie
group and N the largest connected nilpotent normal subgroup of G. Then G
admits a finite-dimensional injective continuous linear representation which
is unipotent on N. (Argue by induction on the dimension of G. Use Proposi-
tion 20 and Chapter I, § 7, Theorem 1.)T
{I 28. (a) Let k be a commutative field of characteristic 0, L an n—dimen-
sional Lie algebra over k, L‘ an (n — l)-dimensional subalgebra containing
no non-zero ideal of L and a0 an element of L not belonging to L]. For
i = 2, 3, . . ., we define Li inductively to be the set of xeL._1 such that
[3, aa]eLi_1. We write L, = L for i < 0. Show that [L4,L,] C L”, (by
induction on i + j) and then that, for O < i S n, L‘ is a subalgebra of L of
codimension i (by induction on i).
([7) For 0 < 1' < 71, Choose in L. an element a, not belonging to L...1 such
that [anJ a.] E {111-1 (mod. L). Show, by induction on the ordered pairs
(1', j) ordered lexicographically, that, for 0 S i < j < n, i + j — l < n and
[0,, (1,] E (J ’ i)at+;—r(m°d-Ll+y)-
(c) Deduce that L is either of dimension 1, or of dimension 2 and non-
commutative, or isomorphic to 61(2, k).
(11) Let A be a finite-dimensional real Lie algebra. A subalgebra B of A
is called extendable if there exists a subalgebra B1 3 B such that

TFor more details concerning injective linear representations of Lie groups,

cf. G. HOGHSCHILD, The timing ofLr'a groups, Holden-Day, 1965, and in particular
Chapter XVIII.

403
ll! LIE GROUPS

dimB1 = dim B + 1. Let R'(A) denote the intersection of all the non-
extendable subalgebras of A. Let R(A) denote the largest ideal of A with a
composition series as an A-module all of whose quotients are of dimension 1.
Show that R'(A) is a characteristic ideal of A. (Observe that R’ (A) is stable
under Aut(A).)
(2) Show that R'(A) D R(A) and that R’(A]R(A)) = R’(A)/R(A).
(f) Show that, if B is a subalgebra of A, then R’(B) D R’ (A) n B.
(g) If A is solvable, R’(A) 2: R(A). (It can be assumed that R(A) = {0},
by (5). Suppose that R’(A) aé {0}. By (d), there exists a minimal non-zero V
ideal I of A contained in R’(A). Using (f) and Chapter I, §5, Corollary 1
to Theorem 1, show that dim I = l, whence a contradiction.)
()1) Let P be the radical of R'(A), B a. semi-simple subalgebra of A and
C = P + B which is a subalgebra. of A by ((1). Using (f), show that ad
can be expressed in a triangular form with respect to a suitable basis of P.
Conclude that [P, B] = {0}.
(k) R(A) is the radical of R’(A). (Use (c), (f), (g), (h) and a Levi de-
composition of A.) Let R’(A) = R(A) 9 T be a Levi decomposition of R'(A).
By (h), R'(A) = R(A) x T. Then R(T) = {0}. Applying (a) to the simple
factors of T, conclude that T = {0}. It has therefore been shown that
R(A) = R’(A).
(I) Let G be a connected real Lie group with Lie algebra A. Let .9’ (G) be
the set of subgroups N of G such that there exists a decreasing sequence
(NM, NM“. . ., No) of subgroups with the following properties: m = N,
N0 = {5), each N. is a normal connected Lie subgroup of G and
dim Nt/N,_l = 1 for all i > 0. Show that the integral subgroup R(G) of
G with Lie algebra R(A) is the largest element of 3’(G). (Argue by induc-
tion on dim R(A). Let I be a l-dimensional ideal of A and N the correspond-
ing integral subgroup of G. Pass to the quotient by N. If N is not closed, use
Exercise 21 (a) of § 6.)
(m) A Lie subgroup H of G is called extendable if there exists a Lie sub-
group H1| D H such that dim Hl = dimH + 1. Let R'(G) denote the inter-
section of all the non-extendable connected Lie subgroups of G.
Let B be a non—extendable subalgebra of A. Show that the corresponding
integral subgroup of G is closed. (Use Proposition 5.) Deduce that
L(R’(G)) C R(A).
whence R'(G) C R(G).
(n) Show that R(G) = R'(G). (Reduce it to the case where R’(G) = {9}.
Then use Exercise 22 of § 6.)?
11 29. A real Lie group G is said to be of type (N) if it is finite—dimensional,

1‘ For more details, cf. J. Tris, Sur une classe de groupes de Lie résolubles,
Bull. Soc. Mam. B:lg., 11 (1959), pp. 100—1 15 and 14 (1962), pp. 196—209.
404
 MERGERS

nilpotent, connected and simply connected. IfG is such a group and g is its Lie
algebra, the mapping exp: g—>G is an isomorphism when 9 is given the
group structure defined by the Hausdorff law (cf. Chapter II, §6, no. 5,
Remark 3). Let log: G —> g denote the 1nverse isomorphism.
(a) Let V be a vector Q—subspace ofg. Show the equivalence ofthe following
conditions.
(i) V is a Lie Q—subalgebra of 9;
(ii) exp(V) is a subgroup of G.
(Use Exercise 5 of § 6 of Chapter II.)
For a subgroup H of G to be of the form exp(V), where V is a vector Q-
suhspace of 5;, it is necessary and sufficient that H be saturated in G (Chapter
II, § 4-, Exercise 14), i.e. that the relations a: e G, x" e H, n sé 0 imply x e H (Inc.
at). Ifthis is so, show that H is an integral subgroup of G if and only iflog(H)
is a Lie R—subalgebra of g.
(12) Let V be a Lie qbalgebra of 9 of finite dimension m, let (tb. . ., cm)
be a basis of V over Qand let A be the subgroup of V generated by s,,. . ., 2",.
Using the fact that the Hausdorff law 15 polynomial show that there exists an
integer 11 > 1 such that, for every integer r which IS a non-zero multiple of J,
exp (rA) is a subgroup of G. Let :1 be such an integer and r a multiple ON. Show
that exp(1A) is discrete if and only if (:1, . . . , a...) is a free family over R, i.e. if
the canonical mapping of V ‘80 R into g is infective. Suppose that this is the
case; show that G/exp(rA) is compact if and only if V 80 R —> g is bijmive, i.e.
ifV i: a Qfarm 1yr 9 (To show that the condition is suflicient, argue by induc-
tion on the nilpotency class of g )
(6) Conversely, let I‘ be a discrete subgroup of G, let I‘ be 1ts saturation in
G (Chapter II, lac. cit.) and let 9“ = log(I‘)= Q. log([‘) be the corresponding
Lie Q;suba.lgebra. Suppose that R. g = g, i.e. that I‘ is Contained in no distinct
integral subgroup of G. Let 2 be an element 761 of the centre of I‘ and let
x = log (2). Show that it belongs to the centre ofg. If X = exp(Rx), show that
X/(I‘ n X) is compact and that the image of I‘ in G/X is discrete. Deduce,
arguing by induction on dim G, that G/I‘ is rampart and that g, is a G-fvrm ofg.
If A is a lattice of 9,, show that there exists an integer d sé 0 such that 1" con-
tains exp(dA) and that the index of exp(dA) in F is then finite.
Let H be an integral subgroup of G with Lie algebra b. Show that H/ (H n I‘)
is compact if and only if I) is rational with respect to the Qstructure g, (in
particular if I) is one of the terms of the lower—or upper—central series of g).
(41) Let I‘ be a discrete subgroup of G, let H be the smallest integral sub-
group of G containing I‘ and let I) be its Lie algebra. Show that H/I‘ is compact
and that g ®qR—>g is injective and has image I). (Apply (a) to the nil-
potent group H.)
(e) Let I‘ be a discrete subgroup of G. Show that there exists a basis
(x1, . . ., x“) of L(G) with the following properties:
(i) for alliE[l,q),Rx. +--- + Rx. is an ideal n, of._1 +~~~+ Rx,;

405
"1 LIE GROUPS
(ii) there exists [7 e [1, q] such that I‘ is the set of products

fiPWm) €XP(mp + 1x!) + 1) . ' ' CXPO’W‘U)

where ”1,, . . ., m, are in Z;
(iii) ifG/I‘ is compact, (n1, n2, . . ., “4} contains the lower central series ofg.
(Using (d) and Proposition 16, reduce it to the case where G/I‘ is compact.
Then argue by induction on dim G, using (c).)
(30) Give an example ol'a 7-dimcnsional real nilpotent Lie algebra with no,
basis with respect to which the constants of structure are rational (cf. Chapter
I, § 4, Exercise 18)‘ Deduce that the corresponding group of type (N) (Exercise
29) has no discrete subgroup with compact subgroup. (Use Exercise 29.)
31. Let G and G’ be two Lie groups of type (N) (Exercise 29) and let I" be a
discrete subgroup of G such that G/l" is compact. Show that every homomor-
phism f: I‘ —>G’ can be extended uniquely to a Lie group morphism of G
into G’ (begin by extendingf to the saturation P of l" and derive a homomor-
phism of the Lie Q;a.lgebra. 103(1") into the Lie algebra of G', cf. Exercise 29).
32. Let G be a Lie group of type (N) (Exercise 29) and let I‘ be a discrete
subgroup of G. Prove the equivalence of the following conditions:
(a) G/l" is compact; ’
(b) the measure of G/F is finite (relative to a non-zero G—invariant positive
measure);
(6) every integral subgroup of G containing 1" is equal to G.
(Use Exercise 29‘)
1T 33. Let 1" be a group Prove the equivalence of the following conditions:
(a) I‘ is nilpotent, torsion-free and finitely generated;
(b) there exists a Lie group of type (N) (Exercise 29) containing F as a dis-
crete subgroup;
(0) there exists a Lie group G of type (N) (Exercise 29) containing 1" as a
discrete subgroup and such that G/I‘ is compact.
(The equivalence of (b) and (0) follows from Exercise 29. The implication
(r) => (a) is proved by induction on dim(G) by the method of Exercise 29 (c).
To prove that (a) a (r),show first that the Lie Q—algebra attached to the satura-
tion of 1" (cf. Chapter II, §6, Exercise 4) is finite-dimensional; then take the
tensor product of this algebra with R and the corresponding group of type
(NM
34-. Let G be a finite-dimensional connected real or complex Lie group and
[.3 an analytic linear representation of G on a finite~dimensional complex
vector space V. Suppose that p is semi-simple. The group G operates by auto-
morphisms on the algebra S(V*) of polynomial functions on V.
(a) Let S(V*)° be the set of G—invariant elements of S(V*). There exists a

406
 EXERCISES

projector p of S(V*) onto S(V")G which commutes with the operations of G

and which leaves stable every G—stable vector subspace of S(V*).
(b) Let LJ be ideals of S(V‘) which are stable under G. Let A (resp. B) be
the set ofzeros ofI (resp.]) in V, Suppose thatA n B = 25. There then exisis
u e I such that u is G-invariant and that u = l in B. (By Commutative Algebra,
Chapter 7, § 3, no. 3, Proposition 2, there exists 0 E I, w e] such that u + w = 1
Let u = pa. Show that 14 has the required properties.)
35. Let G be a compact subgroup of GL(n, R).
(a) Let A, B be disjoint compact G-invariant subsets of R“. There exists a
G-invariant polynomial function u on R", such that u = l on A and u = 0 on
B. (There exists a continuous real-valued function u on R“ such that u = 1 on
A and v = 0 on B. By the Stone-Weierstrass Theorem. there exists a polyno-
mial function to on R“ such that I11 — w] s 4} on A U B. Derive u by an inte-
gration process with respect to the normalized Haar measure of G.)
(1:) Let x E R". Then Gx is the set of zeros in R" of a finite number of G-
invariant polynomials. (Use (a) and Corollary 2 to Theorem 1, no. 8.)
36. In SL(2, R), the elements which can be expressed in the form (a, b),
where a, 17 in SL(2, R), are the elements aé —I.
1] 37. Let G be a compact real Lie group and G' a finite-dimensional con-
nected real Lie group. Suppose that L(G) is simple. Let p: G—-> G’ be a
homomorphism of abstract groups. Suppose that there exists a neighbourhood
V of :5 in G such that p(V) is relatively compact. Then 9 is continuous.
(Let V’ be a neighbourhood of :5, in G’. Let V' C V’ be a neighbourhood of
to» such that
n
(x EV”, x, e p(V),_t/, e 9(V) for l sj S 7:) => (1:! My” arm—1 EV’)

where n = dim G. It can be assumed that p is non-trivial. Then Kerp is

finite by Exercise 25. Therefore 9(G) is not countable and hence not discrete.
Hence there exists g 5 G with non-open centralizer such that 9(g) E V".
Using Exercise 9 of § 4 and its notation, p(M(g, n, V)) C V' and M(g, n, V)
is a neighbourhood of ea in G.)
38. Let G be a finite-dimensional real Lie group and Go its identity com-
ponent Consider the following property:
(F) Ad(G) is closed in Am L(G).
Show that G has property (F) in each of the following cases:
(i) G is connected and nilpotent;
(ii) every derivation of L(G) is inner;
(iii) G0 has property (F) and G/GD is finite;
(iv) G is an upper triangular group;
(v) G0 is semi-simple.

407
m LIE GROUPS

39. Let G be a finite-dimensional connected real Lie Group, H an integral

subgroup of G and H its closure in G. Suppose that H has property (F)
(Exercise 38).
(a) Let xefi. Then Admcyc leaves L(H) stable (no. 2, Proposition. 5);
let u(x) be its restriction to L(H). Then 14H) = Adum(H). (Observe that
AdL®(H) is dense in «(l-l) and apply property (F).)
(b) Let C be the centre of H. Then H = C.H and C is the closure of the
centre of H. (By (a), if x E H, there exists y e H such that Adm”): = Admmy. .
Then Ad(x‘1y) is the identity on L(H), hence x‘ y e Zfi(H) and z E Z§(H) .H.
By continuity, Z§(H) is equal to C. The group C is closed in H and hence in
G; therefore C n H C C. Let 16 C. There exists a. sequence (x,) in H such
that x,l tends to it. Then Adm” tends to l and hence there exists a sequence
(31,.) in H such that y,l tends to e and Adumyn = Adumxn. Then 3],? 1x” E C n H
and y; 12,. tends to x.)
(a) H = H if and only if the centre of H is closed in G. (Use (b).)
40. Let H be a finite—dimensional real Lie group and Ho its identity
component.
(11) Suppose that Ho satisfies condition (F) of Exercise 33, that H/Ho is
finite and that the centre of HD is compact. Let G be a finite-dimensional real
Lie group and f: H —> G a continuous homomorphism with discrete kernel.
Thenf(H) is closed in G. (Reduce it to the oasewhere H is connected. As the
kernel off is discrete, the centre off(H) is the image underfof the centre of H
(Lemma 1) and hence is closed in G. Apply Exercise 39 (5).)
(b) Suppose that H0 is semi—simple and that H/Ho is finite. Then the image
of H under a finite-dimensional continuous linear representation is closed in
the linear group. (Apply (a) and Exercise 7 (b).)
41. Let G be a finite-dimensional connected real Lie group and
p: G —> GL(n, C) a continuous homomorphism with finite kernel. Then
9((G, G)) is closed in GL(n,C). (Using Exercises 7 (b) and 40 (:1), reduce it to
the case where G is the semi-direct product of a semi-simple group S and its
radical R. Every element of 9((G, R)) is unipotent and hence p((G, R)) is
closed. The vector subspace Vl of fixed points of p((G, R)) is #{0}. It is stable
under p(G). By considering C"/V1 arguing by induction on n and using the
complete reducibility of 9(5), it is seen that it is possible to write
C" = V) ©“‘®V¢
where each VI is stable under 9(5) and p((G, R))(V,) C V1 (-9- - ~63 V._1 for
for all i. On the other hand, 9(5) is closed (Exercise 40 (b)). Finally,
P((G, G)) = 9(5) . PKG, R))-)
42, Let G be a finite-dimensional connected rail Lie group. Suppose that

408
 EXERCISES

G admits an injective continuous linear representau'on p: G —> GL(n, G).

Then G admits a linear representation a: G —>GL(yL’, C) which is a homomor-
phism of G onto a closed subgroup of GL(n', C). (By Exercise 4-1, 9((G, G)) is
closed in GL(n, C) and hence (G, G) is closed in G. Let p: G —> (G, G) be the
canonical morphism and 1 an injective linear representation with closed image
G/(G, G) which is connected and commutative. Take 6 = p (9 (-r o p).)

§1o
1. Let G be a finite-dimensional connected real Lie group. Show that the
canonical mapping of Aut(G) into Aut(L(G)) is not in general sutjective
(take G = T).
2. Suppose that K = Q7. Let G be a finiteudimensional Lie group. Show
that the following conditions are equivalent:
(a) there exist x1, x,, . . . , x" in G such that the subgroup of G generated by
{2:1, . . ., x”) is dense in G;
(b) G is generated by a compact subset.
(To prove that (b) implies (a), observe that, if (cl, . . ., 9,.) is a basis of L(G),
(exp 2,51) (exp 2,22) . . . (exp 2,2,.) is a neighbourhood of e.)
3. Let G be the set of (1,3,) E Q, X Q, such that |y| s 1. This is an open
subgroup of Q,1 x Q,.ThenL(G) = Q, X Q,.Letotbetheinfinitesimalauto—
morphism (x, y) )—> (O, x). Then the conclusion of Proposition 3 is false.
(4) Let K be a quadratic extension of QF and let oi E K — Q,” Let
G = Q, + 2,0), considered as a Lie group over K. The only automorphisms
of G are the (x, y) r—> (M, y) where )t is invertible in Zp. The set of these auto-
morphisms cannot be given a Lie gmup structure over K with the properties
of Theorem 1.
 HISTORICAL NOTE

(Chapters I to III)

I. GENESIS
The theory, called for nearly a century “theory of Lie groups”, was essentially
developed by one mathematician: Sophus Lie.
Before embarking on the history, we summarize briefly various earlier
research which prepared the way.
(a) Tramflrmaliun group: (Klein-Lie, 1869—1872)
About 1860 the theory of permutation groups of afim‘tz set was developing and
beginning to be used (Semt, Kronecker, Mathieu, Jordan). On the other
hand, the theory of invariants, which was in full flight, was familiarizing
mathematicians with some infinite sets of geometric transformations stable
under composition (notably linear or projective transformations). But before
the work ofJordan [7] in 1368 on “groups ofmovements” (closed subgroups of
the group of displacements of 3-dimensional Euclidean space), no conscious
link seems to have been established between these two currents of ideas.
In 1869, the young Felix Klein (1849—1925), a pupil of Plficker, established
a friendship in Berlin with the Norwegian Sophus Lie (1842—1899), who was a
few years older, brought together by their common interest in Plficker’s “line
geometry” and in particular the theory of line complexes. It was about this
time that Lie conceived one of his most original ideas, the introduction of the
notion of invariant into Analysis and Differential Geometry; one of the sources
was his observation that the classical methods of integration “by quadratures”
of difl‘erential equations depended entirely on the fact that the equation is
invariant under a “continuous” family of transformations. 1869 is the date of
the first work (edited by Klein) where Lie used this idea ; there he studied the
“Reye complex” (set of four lines cutting the faces of a tetrahedron in four
points with a given cross—ratio) and the curves and surfaces which admit as
tangents lines of this complex [3 a]: his method depends on the invariance of
the Reye complex under the 3-parameter commutative group (maximal

410
 HISTORICAL NOTE

torus of PGL(4, (3)) which leaves the vertices of the tetrahedron invariant.
The same idea dominated the work Klein and,Lie wrote together when
they were in Paris in the spring of 1870 [1, a]; there they essentially deter-
mined the commutative connected subgroups of the projective group of the
plane PGL(3, C) and studied the geometric properties of their orbits (under
the name of curves or surfaces V); this gave them, by a uniform procedure,
properties ofvarious curves, algebraic'or transcendental, such as y = ox” or the
logarithmic spirals. Both works served to underline the profound impression
made on them by the theories of Galois and Jordan (jordan’s commentary on
Galois had appeared in Math. Annalm in 1869; on the other hand, Lie had
heard talk of Galois theory as early as 1863). Klein, who in 1871 became
interested in non-Euclidean geometries, saw there the start of his research on a
classification principle for all known geometries which was to lead him in 1872
to the “Erlangen Programme”. For his part, Lie, in a letter of 1873 to A.
Mayer ([3], vol. V, p. 584-), dated his ideas on transformation groups from his
sojourn in Paris and in a work of 187] ([3 b] p. 208) he was already using the term
“transformation group” and explicitly posed the problem of the determination
of all subgroups (“oontinuous or discontinuous”) of GL(n, C). To be truthful,
Klein and Lie had both experienced some difficulty in entering this new
mathematical universe and Klein spoke ofJordan’s newly appeared “Treatise”
as a “book realzd with room Halo” ([2], p. 51); he wrote moreover concerning
[1 a] and [l b]: “To Lie belong: all the omiilfor the heuristic idea of a continuum;
group of r ' of dflmnttal equationr
and partial derivatives. All the notion: he developed later in his theory of continuous
group: were already there in embryo, but were however so little elaborated, that only after
long conversation: could I oorwince him of many details, for example to brgin with the
wry existence ofth: auras: V” ([2], p. 415).
(b) Iryinitmmal transfirmalion:
The concept of an “infinitely small” transformation goes back at least to the
beginnings of lnfinitesimal Calculus; we know that Descartes discovered the
instantaneous centre of rotation when admitting that “in the infinitely small”
every plane movement can be likened to a rotation: the elaboration of Analy-
tical Mechanics in the 13th century is entirely founded on similar ideas. In
1851, Sylvester, seeking to form invariants of the linear group GL(3, C) and
some ofits subgroups, gave the parameters appearing in his matrices “infinitely
small” increases of the forms nod! and expressed the fact that a functionf ((2,))
was invariant by writing the equationf((z, + u,(dt)) = f((z,)); this gave him
forf the linear partial differential equation Xf = 0, where

(1) Xf — af
_ 2,: a, 5?.

where X is therefore a difimrtial operator, the “derivative in the direction of the

411
 HISTORICAL NOTE

direction parameters at,” ([5], vol. 3, pp. 326 and 327); Sylvester seemed to
think that here was a general principle of considerable importance but appears
never to have returned to the question. A little later, Cayley ([6], vol. II,
pp. 164—178) proceeded similarly for the invariants of SL(2, C) under certain
representations of this group and showed that they are the solutions of two first
order partial differential equations Xf = 0, Yf 2 0, where X and Y are ob-
tained as above from “infinitely small” transformations
0 0 d 0 dt
(.1: 0) a" (0 0)‘
In modern terms, this is expressed by the fact that X and Y generate the Lie
algebra al(2, C); moreover Cayley calculated the bracket XY — YX explicitly
and showed that it was also derived from an “infinitely small” transformation.
In his memoir of 1868 on groups of movements [7],Jordan used from begin—
ning to end the concept of “infinitely small transformation”, but exclusively
from a geometric point of view. He is no doubt responsible for the idea of a
one-parameter group “generated” by an infinitely small transformation: for
Jordan, it was the set of transformations obtained by “suitably repeating” the
infinitely small transformation (lac. cit, p. 243). Klein and Lie, in their memoir,
used the same expression “repeated iyiniwly small transfinnatiun” [1 b], but the
context shows that they understood by that an integration of a differential
system. If the one-parameter group they considered consists of transformations
x’ = f(x, y, t), y' = g(x, y, t), the corresponding “infinitely small transfor-
mation" is given by
11* =P(x,y)dt: dy = 90w)!“
3
where Mm) = gm, to), «(m = 2% (m, to) and to corresponds to the
identity transformation ofthe group. As Klein and Lie knew the functionsfand
g explicitly, they had no difliculty in verifying that the functions

”may, t) and 19w. y, I)

give in parametric form the integral curve of the differential equation

9(E. 7:) 45 = 9(5, *1) dn

passing through the point (x, y), but they give no general argument; moreover
they nowhere use this fact in the remainder of their memoir.

(c) Contact tramformatinnr

In the next two years, Lie seemed to abandon the theory of transformation
groups (although he remained in very close contact with Klein, who published
his “Programme” in 1872) to study contact transformations, integration of

412
 HISTORICAL NOTE

first order partial differential equations and the relations between these two
theories. We are not concerned here with the history ofthese questions and we
shall confine ourselves to mentioning a few points which seem to have played
an important role in the genesis of the theory of transformation groups.
The notion of contact transformation generalizes both point transformations
and inverse polar transformations. Roughly a contact transformation'l' on
C” is an isomorphism of an open subset Q of the manifold T'(C") of cotangent
vectors to C“ onto another open subset Q’ of T’(C") mapping the canonical
l-form of!) to that of Q’. In other words, if (x1, . . ., xmpl, . . .,p,,) denote the
canonical coordinates of T'(C"), a contact transformation is an isomorphism

(an, 9,) n—> (X‘, Pi) satisfying the relation ; 1’,X = ; p,dx.. Such transfor-
mations occur in the study of the integration ofpartial differential equations of
the form
31 32
(2) F(x1,x2,...,xu,a—n,...,wn) =0.

Lie became familiar during his research on these questions with the manipula-
tion of Poisson brackets

” afag age
(3) (fig)= giaTaE'a—méi
and the brackets: [X, Y] = XY — YX of differential operators of type (1) ; he
interprets the Poisson bracket (3) as the effect onf of a transformation of type
(1) associated with g and observes on this occasion that the Jacobi identity for
Poisson brackets means that the bracket of the differential operators corres-
ponding tofand g is associated with the bracket (g, h). Research into functions
g such that (F, g) = 0, which occurs in Jacobi’s method for integrating the

T Here we are concerned with “homogeneous” contact transformations, Earlier,

studying equations of type (2) but with z occurring in F had led Lie to consider
contact transformations in 211 + 1 variables 2, :1, . . .x, ,, p1, ., p“, where it is
required to find 271 + 2 functions Z, P0 X‘ (l < 1'< n) and p (the latter #0 at
every point) such that dz — Z PidX. = p(dz — 21min). This case which appears
to be more general can be easily reduced to the “homogeneous” case ([4], vol. 2,
pp. 135—146).
: These had already occurred in the Jaeobi—Clebsch theory of “complete sys-
tems” of first order partial difierential equations X,f = 0 (1 $ j s r), a notion
which is equivalent to Frobenius’s “completely integrable system”; the funda-
mental theorem (equivalent to “Frobenius’s theorem”) which characterizes these
systems is that the brackets [X., X,] must be linear combinations (with variable
coefficients) of the Xk.

413
 HIS'I‘OIUCAL NOTE

partial differential equation (2), became for Lie the study of infinitesimal con-
tact transformations which leave the given equation invariant. Finally,
Lie was led to study sets of functions (11,)1 “5,, of the x; and pl such that the
brackets (14,, uk) are functions of the uh and called these sets “groups” (they
had already essentially been studied by Jacobi) [3 c]i

I]. CONTINUOUS GROUPS AND W TRANSFORMATIONS

Suddenly, in the autumn of 1873, Lie again took up the study of transformation
groups and obtained decisive results. Insofar as it is possible to follow his line '
of thought from a few letters to AA Mayer written in 1873—1874 ([3], vol. 5,
pp. 584—608), he started from a “continuous group” of transformations on 11
variables
(4-) x;=fi(x1,...,x,,,al,...,a,) (1 Sign)
depending efi‘ectivelyT on r parameters up. . ,, a,; he observed that, if the
transformation (4) is the identity for the values as, . . ., a? of the parameters}
then the first order Taylor expansions of the an:
(5) fr("n~-a"m“i + Inna“? + Zr)
7
=xt+zlztxm<x4mm>+~- (Isisn)
give a “genetic” infinitely small transformation depending linearly on the r
parameters 1,

(6) dx. = (Z sm(x1,...,x,,))dt (1 s is u).

k=I

Proceeding as in his memoir with Klein, Lie integrated the differential system
(7) d at = ...= d in = dt,
2 extra”. . ., at) Z ztxmel, . . ., an)
1' By this Lie meant that the f, could not be expressed in terms of less than 9‘
functions of the n,, or also that the jacobian matrix (afi/Ba,) is “in general” of
rank 7.
1 In his first notes, Lie thought that he could prove a priori the existence of the
identity and the inverse in the whole set of transformations (4) stable under
composition; he recognized later that his proof was incorrect and Engcl provided
him with a counterexample reproduced in [4-], vol. 1, § 44-. However, Lie showed
how “continuous" systems (4) stable under composition could be reduced to group
germs of transformations: such a system is of the form G o h, where G is a group
germ of transformations and h a transformation of the system ([4], vol. 1, Theorem
26, p. 163 and vol. 3, Theorem 46, p. 572).
4-14-
 HISTORICAL NOTE

which gave him, for each point (:1, . . ., 2,) a one—parameter group
(8) “”‘i=gli‘nn-axmlnnuZn‘) (1 <i<nl
such that g‘(xl, . . ., x,“ 11, . . ., 2,, 0) = x, for all i. He showed by an ingenious
method, using the fact that the transformations (4) form a set which is stable
under composition, that the one-parameter group (8) is a subgroup of the given
group [3 d]. The new idea, the key to the whole theory, is to take the Taylor
expansions of the functions (4) to the second aid”. The progress of his argument
was quite confused and heuristic ([3 d] and [3], vol. 5, pp. 600—601) ; it can be
presented as follows. For sufficiently small 1,, let t = l in (8); thus new para-
meters zl, . . ., z, are obtained for the transformations of the group (this is in
fact the first appearance of the “canonical parameters”). Then by definition,
using (7),
flag—t. =2 zkxktofiau -: xvi)
whence
0’
3f;‘= gab—S:—“(A ,4) 3x’37’
= 22* 6X377% a4.)(zz»xuo4,...,xn))
7

which gives

1" = x. + (; km("1: . . ., 19.))!

+ <§~( knit!
Z zkzn—
5:“(run ., x,)X,,,(x,, . . ., x,,))ti + - - -,

whence, for t = 1, the Taylor expansions in the parameters 2,

1;“) + - .. (1 s i < n).

(9) x; = x.++(Z 4X“) + it; zmxh,aa
We abridge these relations to x’ = C(x, 2) relating the vectors
X= (xv-warn), x’ = (xi....,z.’.), z = (Inn-,4);
the fundamental stability property of the set of these transformations under
composition can be written

(10) G(G(x,u): 1/) = G09 H(u, v))

where H = (H1, . , ,, H,) is independent of x; it is immediate that H(u, 0) = u
and H(0, v) = v, whence the expansions

(11) Him) = u. + a. + igauunvk +---,

415
 HM'ORICAL NOTE

where the terms omitted are not linmr in u or 1;. Transforming (10), using (9)
and (11) and then comparing the terms in uhnk of the two sides. Lie obtained
the relations
" '9i 5X“) .
(12) ’21 (XM ax, — H ex, = 1;: cmxu (1 g h,k < 7,1 g 1 s 71).
His experience with the theory of partial differential equations led him to
write these conditions in a simpler form: following the pattern of (l), he
associated with each of the infinitely small transformations obtained by setting ‘
z): = l, 2,, = 0 for h aé k in (6), the differential operator
.
af
Arm = 2 xi. a7;
(13)
and rewrote conditions (12) in the form

(14) mm = Z c...A.,
the corner stone of his theory. Until then he used the terms “infinitely small
transformation" and “infinitesimal transformation” indiflermfly (e.g. [3 e]);
the simplicity of relations (14) led him to call the operator (13) the “symbol”
of the infinitesimal transformation dx‘ = dt (l < i S n) [3 11] and very
soon he called the operator (13) itself the “iry‘initesimal tramfirmafiun” ([3 d] and
[3], vol. 5, p. 589).
He then became aware of the close links which united the theory of “con-
tinuous groups” with his earlier research on contact transformations and par-
tial differential equations. This bringing together filled him with enthusiasm:
“My earlier workr were a: it were all ready there waiting to found the new theory 13f
transfuimatian groups” he wrote to Mayer in 1874- ([3], vol. 5, p. 586).
In the following years, Lie pursued the study of transformation groups.
Besides the general theorems summarized below (§ III), he obtained some more
special results: the determination of the transformation groups of the line and
plane, the subgroups of low oodimension in the projective groups, the groups
in at most 6 parameters, etc. He did not abandon diflerential equations for
long. In fact, it seems that, for him, the theory of transformation groups was a
tool for integrating differential equations, where the transformation group
played a role analogous to that of the Galois group of an algebraic equations!"

T This research had little influence on the general theory of differential equa-
tions, since the automorphism group of such an equation is usually trivial. As a
compensation, for certain types of equations (for example linear equations),
interesting results have been obtained later by Picard, Vasiot and, more recently,
Ritt and Kolchin.

416
 HISTORICAL NOTE

We note that this research led him equally to the introduction of certain trans-
formation sets in an infinity of parameters which he called “infinite continuous
groups”1; he reserved the name “finite continuous groups” for transformation
groups in a finite number of parameters of type (4) above.

THE LIE GROUPS-LIE ALGEBRAS “DICYIONARY”

The theory of “finite continuous” groups, developed by Lie in numerous
memoirs beginning in 1874», is expounded systematically in the imposing
treatise “Theorie der Tmmfonnatiamgmppen” ([4], 1888—1893]), written in
collaboration with F. Engel§; it is the object ofstudy in the first volume and the
last five chapters of the third, the second being devoted to contact transfor-
mations.
As the title indicates, this work is only concerned with transformation
groups in the sense of equations (4), where the space 0 “variables” x, and the
space of “parameters” a, play initially such important roles. On the other hand,
the concept of “abstract” group had not been clearly isolated at that period;
when in 1885 ([3f], §5) Lie noted that in the notation of (10) the equation
10 = H(u, u) which gives the parameters of the composition of two transfor-
mations of the group defines a new group, he considered it as a transformation
group on the space of parameters, thus obtaining what he called the “parameter
group” (he even obtained two, which arejust the group of left translations and
the group of right translations‘f).
The variables x, and the parameters a, in equations (4) were in principle
assumed to be complex (except in Chapters XIX—XXIV of volume 3) and the
functionsfl to be analytic; Lie and Engel were of course aware of the fact that
these functions are not in general defined for all complex values of the x; and a,
and that therefore the composition of such functions raises serious difficulties
([4], vol. 1, pp. 15—17, pp. 33—40 and passim); and although throughout they
almost always wrote as if the composition of the transformations they were
studying was possible without restriction, this was no doubt for the convenience

1 Today they are called “Lie pseudo-groups”; care should be taken not to con-
fuse them with the “Banach” Lie groups defined in this volume.
§ From 1886 to 1898, Lie occupied the chair at Leipzig left vacant by Klein
and had Engel as his assistant; this circumstance favoured the production of an
active mathematical school as well as the diHusion of Lie’s ideas, so little known
until then (notably because of the fact that his first memoirs were usually written
in Norwegian and published in the Comptes Rendus de l’Académie cle Christiania,
which was little used elsewhere). Thus at a time when it was unusual for young
French mathematicians to go to Germany for instruction, E. Vessiot and A. Tresse
spent a year studying at Leipzig with Sophus Lie.
1' The analogous notion for permutation groups had been introduced and
studied by Jordan in his “Truths”.

4-17
 HISTORICAL NOTE

of the results and they explicitly reaffirmed the “local” point ofview whenever
necessary (cf. loo. cit, p. 168 or 189 for example or flirt, vol. 3, p. 2, note at the
bottom of the page); in other words, the mathematical object they studied is
close to what we call in this treatise a law chunk of operation. They did not
refrain, on occasions, from considering global groups, for example the 4 series
of classical groups ([4], vol. 3, p. 682), but do not appear to have asked them-
selves the question of what in general constitutes a “global group”; they were
content to obtain, for the “parameters” of the classical groups (the “variables”
of these groups introduced no difficulty, since the transformations in question
are linear transformations ofC") “local” systems of parameters in a neighbour-
hood of the identity transformation, without worrying about the domain of
validity of the formulae they were writing clown. They however set themselves
a problem which arises neatly out of the local theoryi: the study of “mixed”
groups, that is groups with a finite number of connected components, such as
the orthogonal group ([4], vol. 1, p. 7). They presented this study as a study of
a set of transformations stable under composition and passage to the inverse
which is the union of sets H, each of which is described by systems of functions
(f‘m) as in (4-); the number of (essential) parameters of each H, is even a [mini
assumed to depend on j, but they showed that in fact this number is the same
for all the H,. Their principal result was then the existence ofa finite continuous
group G such that H, = G n It, for some 11, e H, and for all j; they also estab-
lished that G is normal in the mixed group and noted that the determination
of the invariants of the latter reduces to that of the invariants of G and a dis-
continuous group ([4], vol. 1, Chapter 18).
The general theory developed in [4] ended (without the authors saying so
very systematically) by achieving a “dictionary” translating the properties of
“finite continuous” groups into those of the set of their infinitesimal transfor-
mations. It is based on the “three theorems ofLie”, each ofwhich consists of an
assertion and its converse.
Thefirst theorem ([4], vol. 1, pp. 33 and 72 and vol. 3, p. 563) affirms in the
first place that if in (4) the parameters are effective, the functions f; satisfy a
system of partial differential equations of the form

<15) 3%,; = Z ancflaanwa (1 s is n)

where the matrix (in) is of maximum rank and det(i[;,‘,) sé 0; conversely, if the

1 Recall (Historical Note to Algebra, Chapter VIII) that following a Note of

H. Poincare ([14], vol. V, pp. 77—79) several authors studied the group of in-
vertible elements of a finite-dimensional associative algebra. It is interesting to
note on this subject that E. Study, in his works on this subject, introduced a
symbolism which essentially amounted to considering the abstract group defined
by the parameter group.

418
 IHS’I‘ORICAL NOTE

funetionsfi have this property, formulae (4) define a group germ of trans-
formations.
The second theorem ([4], vol. I, pp 149 and 158, and vol. 3, p. 590) gives rela-
tions between the E“ on the one hand and the 41,, on the other: the con-
ditions on the in can be written in the form

(16) “21" (Ema

3”: — at af‘—') = Z 452k. (1 < w
. .
s n1 s Is n)
where the r“ are constants (l< 1,J,< 1) skew-symmetric in i, j. The con-
ditions on the slag, in the form given by Maurer [10], are:

07) %— gal—gut 2 wait—nag...) (1< k1m< r)

1(hl<r
By introducing the contragradimt matrix (au) of (4m) and the infinitesimal
transformations
7»
a
(18) Xk=‘215u5;i’ Ak=;“kia—:(l <k<’)
(16) and (17) can be written respectively:

(19) [xt X1] = 2c. (1 s 231' < r).

(20) [ADM = Z czAk
Conversely, if r infinitesimal transformations X,‘ (l < k < r) are given
which are linearly independent and satisfy conditions (19), the one-parameter
subgroups generated by these transfonnations generate-a transformation group
in I essential parameters.
Finally, the third theorem ([4], vol. 1, pp. 170 and 297 and vol. 3, p. 597)
reduces the determination of the systems of infinitesimal transformations
(Xk)1 ‘ H, satisfying (19) to a purely algebraic problem: the following must
hold:
(21) ‘u + 5n = 0

(22) [Z (”(1911: + @141 + twin) = 0 (1< ‘aj: k) ”‘< 7)-

Conversely,1' if (21) and (22) are satisfied, there exists a system of infinitesimal
T This converse was not obtained without difficulty. The first proof given by
Lie [3 1] consisted in passing to the adjoint group and was in fact only valid if the
centre of the given Lie algebra was (0). He then gave two general proofs ([4],
vol. 2, Chapter 17 and vol. 3, pp. 599—604); it is very significant that the first
was based on contact transformations and that Lie found it more natural than
the second.

419
 HISTORICAL NOTE.

transformations satisfying relations (19), whence a transformation group with r

parameters (in other words, the linear combinations of the Xk with constant
coefficients form a Lie algebra and conversely every finite-dimensional Lie
algebra can be obtained in this way).
These results are completed by studying the questions of isomorphisms. Two
transformation groups are called similar if it is possible to pass from one to the
other by an invertible coordinate transformation on the variables and an
invertible coordinate transformation on the parameters: from the very
beginning of his research, Lie had been concerned naturally with this notion
when defining the “canonical parameters”. He showed that two groups are
similar if, by a transformation on the “variables”, it is possible to transform the
infinitesimal transformations of the one into those of the other ([4], vol. 1,
p. 329). A necessary condition for this to be the case, is that the Lie algebras of
these two groups be isomorphic, which Lie expressed by saying that the groups
are “gleichzummmmgmtzt”; but this condition is not sufficient and a whole
chapter ([4], vol. 1, Chapter 19) is devoted to obtaining supplementary con-
ditions assuring that the groups are “similar”. The theory of permutation
groups on the other hand provided the notion of “holohedric isomorphism”
of two such groups (isomorphism of the underlying “abstract” groups); Lie
transposed this notion to transformation groups and showed that two such
groups are “holohedrically isomorphic” if and only if their Lie algebras are
isomorphic ([4], vol. 1, p. 418). In particular, every transformation group is
holohedrically isomorphic to each of its parameter groups and this shows that,
when we wish to study the structure of the group, the “variables” on which it
operates are of little importance and that in fact all that matters is the Lie
algebra.1'
Always by analogy with the theory of permutation groups, Lie introduced
the notions of subgroups, normal subgroups and “merihedric isomorphisms”
(surjective homomorphisms) and showed that they correspond to those of
subalgebras, ideals and snrjective Lie algebra homomorphism: ; he had already
previously come across an especially important example of a “merihedric
isomorphism”, the adjoint representation, and had recognized its relations with
the centre of the group ( [3f], § 3, no. 9). For these results, as for the funda-
mental theorems, the essential tool is the Jacobi-Clebsch Theorem giving the
complete integrability of a differential system (one of the forms of the theorem
called “Frobenius’s”); he however gave a new proof using one-parameter
groups ([4], vol. 1, Chapter 6).

1' A similar evolution can be pointed out in the theory of “abstract” groups, in
particular finite groups. They were first defined as transformation groups, but
Gayley had already noted that what is essential is the way the transformations
are composed with one another and not the nature of the concrete representation
of the group as a permutation group of particular objects.

420
 HISTORICAL NOTE

The notions of transitivity and primitivity, so important for permutation

groups, occurred just as naturally for “finite continuous” transformation
groups and the Lie—Engel treatise made a detailed study of this ([4], vol. 1,
Chapter 13 and parrim); the relations with the stabilizer subgroups of a point
and the notion of homogeneous space were perceived (insofar as was possible
without taking a global point of view) ([4], vol. 1, p. 4-25).
Finally the “dictionary” was completed, in [4], by the introduction of the
notions of derived group and solvable group (called “integrable group” by
Lie ; this terminology, suggested by the theory of differential equations,
remained in use until the works of H. Weyl) ([4], vol. 1, p. 261 and vol. 3,
pp. 678—679); the relation between oommutators and brackets had elsewhere
been perceived by Lie in 1883 ([3], vol. 5, p. 358).

Other proqf: of thefimdammtal theorzm:

In [8] F. Schur showed that in canonical coordinates the 41* of (15) satisfy
the differential equations

(23) .7.d «Ma» = a.. + E stoma).

These are integrated and give a formula equivalent to the formula

(24) m<X> = «zomim (ad(X))"

of our Chapter III, §6, no. 4, Proposition 12; in particular, in canonical
coordinates, the L1)” can be extended to integral functions of the ak. F. Schur
deduced a result which made precise an earlier remark of Lie: if, in definition
(4-) of transformation groups, it is only assumed that thej; are of class C“, then
the group is holohedrically isomorphic to an analytic group.1’ Following his

T Lie had already stated without proof a result of this type ([3 g], no. 7). He had
been led there by his research on the foundations of geometry (“Helmholz’s prob-
lem"), where he had remarked that the analyticity hypotheses are not natural.
The result of F. Schur was to lead Hilbert in 1900 to ask if the same conclusion
remained true if it was only assumed that the f. were continuous (“Hilbert’s 5th
problem”). This problem has stimulated much research. The most complete
result along this line is the following theorem, proved by A. Gleason, D. Mont-
gomery and L. Zippin; every locally compact topological group has an open sub-
group which is an inverse Limit of Lie groups; it implies that every locally Euclidean
group is a Lie group. For more details on this question, cf. D. MONTGOMERY and
L. ZIPPIN [41].

421
 HISTORICAL NOTE

research on the integration of differential systems, El Cartan ([12], vol. Ila,

p. 371) introduced in 1904 Pfaflian forms

(25) w~=g¢um (1$i<r)

(in the notation of (15)), called later a-Carlanformr. The Maurer condi-
tions (17) could be written

'1‘”): = —-};c§w‘ A “I;

E. Cartan showed that the theory of finite continuous groups could be deve-
loped starting from the 0),, and established the equivalence of this point ofview
and that of Lie. But, for him, the interest of this method was above all that it
could be adapted to “infinite continuous groups” the theory ofwhich he pushed
much further than Lie had done and which allowed him to develop his theory
of the generalized “moving frame”.

IV. THE THEORY OF LIE Mm

Having once acquired the correspondence between transformation groups and
Lie algebras, the theory took a considerably more algebraic turn and became
centred on a deep study of Lie algebras.T .
A first, short period, from 1888 to 1894, marked by the works of Engel and
his pupil Umlauf and especially Killing and E. Cartan, achieved a series of
spectacular results on complex Lie algebras. We have seen earlier that the
notion of solvable Lie algebra was due to Lie himself, who had shown (in the
complex case) the theorem on the reduction of solvable linear Lie algebras to
triangular form ([4], vol. I, p. 270).: Killing observed [1 1] that there exists in a
Lie algebra a largest solvable ideal (which we now call the radical) and that the
quotient of the Lie algebra by its radical has zero radical; he called Lie alge-
bras with zero radical semi-simple and proved that they are products of simple
algebras (the latter notion having already been introduced by Lie, who had
proved the simplicity of the “classical” algebras ([4], vol. 3, p. 682))
On the other hand, Killing introduced in a Lie algebra the characteristic
equation det(ad(x) — 03.1) = 0, already encountered by Lie when studying
2-dimensional Lie subalgebras containing a given element of a Lie algebra.
We shall return in other Historical Notes to this Book to the analysis of the
methods by which Killing, whilst engaged on a penetrating study of the pro-

T The term “Lie algebra" was introduced by H. Weyl in 1934: in his work of
1925, he had used the expression “infinitesimal group”. Earlier, mathematicians had
spoken simply of the “infinitesimal transformations X, f, . , ., X, f ” of the group,
which Lie and Engel frequently abbreviated by saying “the group Xx]; . . ., Xff” !
3 Almost at the beginning of his research, Lie had encountered solvable linear
groups and even in fact nilpotent linear groups [3 h].

422
 HISTORICAL NOTE

perties of the roots of the “generic” characteristic equation for a semi-simple

algebra, achieved the most remarkable of his results, the template determination
of the (complex) simple Lie algebras.§
Killing proved that the derived algebra of a solvable algebra is “of rank 0”
(which means that adx is nilpotent for every element x of the algebra).
Almost immediately, Engel showed that algebras “of rank 0” are solvable
(this statement is essentially what we called Engel’s Theorem in Chapter I,
§4, no. 2). In his thesis, E. Cartan introduced, on the other hand, what we
now call the “Killing form” and established the two fundamental criteria
which characterize by means of this form solvable Lie algebras and semi
simple Lie algebras.
Killing had affirmed ([11], IV) that the derived algebra of a Lie algebra is
the sum of a semi-simple algebra and its radical, which is nilpotent, but his
proof was incomplete. A little later, E. Cartan announced without proof
([12], vol. 1,, p. 104-) that more generally every Lie algebra is the sum of in
radical and a semi-simple subalgebra; the only result in this direction estab-
lished indisputably at this period was a theorem of Engel aflirming the exis-
tence, in every non-solvable Lie algebra, of a simple Lie subalgebra of dimen-
sion 3. The first published proof (for complex Lie algebras) of Cartan’s state-
ment is due to E. E. Levi [18] ; another proof (equally valid in the real case) was
given by J. H. C. Whitehead in 1936 [26 a]. In 1942, A. Malcev completed
this rault by the uniquenss theorem for “Levi sections” up to conjugation.
From his very first works, Lie was concerned with the problem of the iso-
morphism between any Lie algebra and a linear Lie algebra. He had believed
that he had solved it aflirmatively by considering the adjoint representation
(and deduced from it a proof of his “third theorem”) [3 d]; he realized that
his proof was correct only for Lie algebras with zero centre, gave another
erroneous proofthereof ([3f], §3, no. 9) and then recognized ([3 g], p. 231)
that the question remained open. In fact, it remained so for a long time and
was only finally resolved affirmatively in 1935 by Ado [27]. On the other
hand, Lie had essentially posed the problem of determining linear representa-
tions of simple Lie algebras of minimal dimension and had solved it for the
classical groups; in his Thesis, Cartan also solved this problem for the excep-
tional simple algebras‘l'; the methods he used for this were to be generalized
by him twenty years later to obtain all the irreducible representations of real
or complex simple Lie algebras.

§ Up to the fact that he found two exceptional groups of dimension 52 but did
not notice that they were isomorphic. (He only considered complex simple Lie
algebras for the more general problem had not been considered at that time;
Killing’s methods are in fact valid for any algebraically closed field of charac-
teristic 0).
1‘ Cartan’s approach consisted of studying the non-trivial extension Lie algebras
of a simple Lie algebra and a (commutative) radical of minimal dimension.

423
 HISTORICAL NOTE

The property of complete reducibility of a linear representation seems to

have been encountered for the first time (in a geometric form) by Study. In an
unpublished manuscript, but quoted in [4], vol. 3, pp. 785—788, he proved this
property for linear representations of the Lie algebra SL(2, C) and obtained
partial results for SL(3, C) and SL(4-, C). Lie and Engel conjectured on this
occasion that the complete reducibility theorem was true for SL(n, C) for all
n. The complete reducibility of linear representations of semi—simple Lie
algebras was established by H. Weyl in 19251’ by a global type of argument
(see later). The first algebraic proof was obtained in 1935 by Casimir and van ,
der Waerden [32] ; other algebraic proofs have since been given by R. Brauer
[31] (this is the one we have reproduced) and J. H. C. Whitehead [26 b].
Finally, in the course of his research on the exponential mapping (cf. infm),
H. Poincare ([14], vol. 3) considered the associative algebra of differential
operators of all orders, generated by the operators of a Lie algebra; he showed
essentially that, if (Xmflg,I is a basis of the Lie algebra, the associative
algebra generated by the X1 has as basis certain symmetric functions of the
X, (sum of the non-commutative “monomials” derived from a. given monomial
by all permutations of factors). The essence of his proof was algebraic in nature
and allowed him to obtain the enveloping algebra structure which we define
abstractly in Chapter I. Analogous proofi were given in 1937 by G. Birkhofl‘
[29 b] and E. Witt [30].; ’
Most of the results cited above are limited to real or complex Lie algebras
which alone correspond to Lie groups in the usual sense. The study of Lie
algebras over a field other than R or C was embarked upon by Jacobson
[28 a] who showed that the great majority of the classical results (i.e. those of
Chapter I) remain true for any field of characteristic zero.

V. EXPONEN'I'IAL AND HAUSDORFF FORMULA

The first research on the exponential mapping was due to E. Study and F.
Engel; Engel [9 b] remarked that the exponential is not surjective for SL(2, C)
(for example (—3) _‘1l) is not an exponential if a aé 0), but that it is for
GL(n, C) and hence also for PGL(n, C) (the latter property had already been
noted by Study for n = 2); thus SL(2, C) and PGL(2, C) give an example of
two locally isomorphic groups, which are however very diflerent from a global

1' H. Weyl remarked on this occasion that the construction given by E. Cartan
for irreducible representations implicitly uses this property.
1 The first use of difi‘erential operators of higher order generated by the Xl
was no doubt the use of the “Casimir operator” for the proof of the complete
reducibility theorem. After 1950, the research of Gelfand and his school and
Harish-Chandra, on infinite-dimensional linear representations, has brought
these operators into the forefront.

424
 I‘USTDRICAL NOTE

point of view. Engel also showed that the exponential is surjective for the
other classical groups augmented by homothetis; this work was taken up and
pursued by Maurer, Study and others, without producing substantial new
results.
In 1899, H. Poincare ([14], vol. 3, pp. 169—172 and 173—212) embarked
upon the study of the exponential mapping from a difl‘erent point of view. His
memoirs seem to have been hastily edited, for in several places be affirmed that
every element of a connected group is an exponential, whereas he gave
examples to the contrary elsewhere. His results were mainly concerned with the
adjoint representation: he showed that a semi-simple element of such a group
G may be the exponential of an infinity of elements of the Lie algebra L(G),
whereas a non-semi-simple element may not be an exponential at all. If
ad(X) has no eigenvalue which is a non-zero multiple of 2113', then exp is étale
at X. He also proved that, if U and V describe loops in L(G) and W is defined
by continuity such that c".e" = W, it is possible not to return to the original
value of W. He used a residue formula which amounts essentially to
_1 ¢(E)dE
d>(adX)_2—1;f————E_adx
where ad(X) is a semi-simple element whose non-zero eigenvalues are ofmulti-
plicity l, (D is an integral series with sufficiently large radius ofconvergence and
the integral is taken over a contour surrounding the eigenvalues of ad X; he
also studied what happens as X tends to a transformation with multiple eigen-
values.
Research into expressions for W as a function of U and V in the formula
2".” = :W had already, just before Poincaré’s work, been the object of two
memoirs by Campbell [13]. As Baker wrote a little later “. . . Liz theury rugged:
in an abvious way that the product :"zV i: theform ,w where W is a series ry‘altemnznlr in
U and V . . .”. The later works on this subject aimed at making this assertion
precise and giving an explicit formula (or a method of construction) for W
(“Hausdorff formula”). After Campbell and Poincare, Pascal, Baker [15] and
Hausdorff [16] returned to the question ; each considered that the proofs of his
predecessors were not convincing; the principal difficulty resided in what is
meant by “alternants”: are they special elements of the Lie algebra in ques-
tion, or universal ”symbolic” expressions? Neither Campbell, nor Poincare,
nor Baker expressed himself clearly on this point. Hausdorff’s memoir, on the
other hand, is perfectly precise; he worked first on the algebra of (non-
commutative) associative formal power series in a finite number of indeter-
minates and considered U, V, W as elements of this algebra. He proved the
existence of W by a differential equation argument analogous to that of his
predecessors. He used the same argument to prove the convergence of the
series when the indeterminates are replaced by elements ofa finite—dimensional
Lie algebra. As Baker had remarked, and Poincaré independently, this result

425
 HMORICAL NOTE

could be used to give a proof of Lie’s third theorem; he clarified the correspon-
dence between Lie groups and Lie algebras, for example where the commuta-
tor subgroup is concerned. _ ‘
In 1947, Dynkin [39] again took up the question and obtained the exphcrt
coefficients of the Hausdorfl" formula, by considering from the outset a normed
Lie algebra (of finite dimension or otherwise, over R, C or an ultrametric
field).T

V1. LINEAR REPRESENTATIVES AND GLOBAL LIE GROUPS

None of the works which we have just mentioned made any honest attempt to
define and study global Lie groups. It was H. Weyl who made the first move in
this direction. He was inspired by two theories, which until then had developed
independently: that of linear representations of complex semi-simple Lie
algebras, due to E. Cartan, and that of linear representations of finite groups,
due to Frobenius and which had just been transposed to the orthogonal group
by I. Schur, using an idea of Hurwitz. The later had shown [17] how to form
invariants {or the orthogonal group or unitary group by replacing the operation
of the mean on a finite group by an integration with respect to an invariant
measure. He had also noted that, by applying this method to the unitary group,
invariants are obtained for the general linear group, the first example of the
“unimrian trick”. In 1924-, I. Schur [20] used this procedure to show the com-
plete reducibility of the representations of the orthogonal group C(72) and the
unitary group U(n) by constructing an invariant non-degenerate positive
definite Hermitian form; he deduced, by the “unitarian trick”, the complete
reducibility of the holomorphic representations of 0(n, C) and SL(n, C),
established orthogonality relations for the characters of 0(n) and U(n) and
determined the characters of0(n). H. Weyl immediately extended this method
to complex semi—simple Lie algebras [21]. Given such an algebra 9, he showed
that it has a “compact real form” (which amounts to saying that is obtained by
extension of scalars from R to C from an algebra go over R whose adjoint
group Go is compact). Further, he showed that the fundamental group of Go
is finite and hence that the universal covering: of Go is compact. He deduced,
by a suitable adaptation of Schur’s procedure, the complete reducibility of the
representations of 9 and also gave, by a global method, the determination of the

T In the ultrametric case, the classical method of upper bounds cannot be ex-
tended without precautions hecause of the asymptotic behaviour of the padic
absolute value of 1 In as It tends to infinity.
: H. Weyl did not define explicitly this nation, with which he had been familiar
since the editing of his course on Riemann surfaces (1913). It was 0. Sehreier
[22] who in 1926—1927 gave, for the first time, the definition of a topological group
and that of a “continuous” group (i.e. locally homeomorphic to a Euclidean
space), and the construction of the universal covering of such a group.

4-26
 HISTORICAL NOTE

characters of the representations of g. Ina letter to I. Schur (Sitzungsber. Berlin,

1924, pp. 338—343), H. Weyl summarized the results of Carlan not known to
Schur (cf. [20], p. 299, note at the bottom of the page) and compared the two
approaches: Cartan’s method gave all the holomorphic representations of the
simply connected group with Lie algebra g; in the case of the orthogonal
group, representations were thus obtained of a double covering (called later
the spinor group), which had escaped Schur; on the other hand, Sehur’s
method had the advantage of showing the complete reducibility and giving
the characters explicitly.
After the work of H. Weyl, E. Cartan adopted an openly global approach in
his research on symmetric spaces and Lie groups. This approach was the basis
of his exposition of 1930 ([12], vol. 1,, pp. 1165—1225) of the theory of “finite
continuous” groups. In particular we find here the first proof of the global
version of the 3rd fundamental theorem (the existence of a Lie group with
given Lie algebra); Cartan also showed that every closed subgroup of a real
Lie group is a Lie group (Chapter III, § 8, no. 2, Theorem 2) which generalized
a result 0U. von Neumann on closed subgroups of the linear group [23]. In this
Memoir, von Neumann showed also that every continuous representation of a
semi-simple group is real analytic.
After these works, the theory of Lie groups in the “classical” sense (that is in
finite dimension over R or C) was more or less complete in its essentials. The
first detailed exposition of it was given by Pontrjagin in his book on topological
groups [36]; there he followed an approach quite close to that of Lie, but care-
fully distinguished between the local and the global. This was followed by
Chevalley’s book [38] which also contains the first systematic discussion of the
theory of analytic manifolds and exterior differential calculus; the “infinitesi-
mal transformations” of Lie appear there as vector fields and the Lie algebra
of a Lie group G is identified with the space of left invariant fields on G. He
leaves out any discussion on “group germs” and “transformation groups”.

VII. EXTENSIONS OF THE NOTION 0F LIE GROUP

Nowadays the vitality of Lie theory is manifested by the diversity of its
applications (in topology, difl‘erential geometry, arithmetic, etc.) and by the
creation of parallel theories where the underlying manifold structure is re-
placed by a similar structure (p—adic manifold, algebraic variety, scheme,
formal scheme, . . .). We are not concerned here with the history of these
developments and we shall confine ourselves to those touched on in Chapter
III: Banach Lie groups and p-adic Lie groups.

(a) Banach Lie groups

Here we are concerned with “infinite-dimensional” Lie groups. From the
local point of view, a neighbourhood of 0 in Euclidean space is replaced by a
427
 HISTORICAL NOTE

neighbourhood of 0 in a Banach space. This is what G. Birkhoff did in 1936

[29 a], thus achieving the notion of a complete mmrml Lie algebra and its corres-
pondence with a “group germ” defined on an open set of a Banach space.
About 1950, Dynkin completed these results by extending the Hausdorff
formula to this case (cf. supra).
The definitions and results of Birkhofl‘ and Dynkin are local. Until recently,
it does not seem that any one has sought to make the corresponding global
theory explicit, no doubt because of the lack ofapplications.1‘

(b) p—adic Lie group:

Such groups were encountered for the first time in 1907 in the works of Hensel
[19] on [J-adic analytic functions (defined by expansions as integral series). He
studied in particular the exponential and the logarithm; in spite of the a prion"
surprising behaviour of the series which define them (for example the uponen-
tial series does not converge everywhere), their fundamental functional proper-
ties remain valid, which provides a local iramumhism between the additive group
and the multiplicative group of Q, (or, more generally, of any complete
ultrametric field of characteristic zero).
A. Weil [33] and E. Lutz [34] were equally concerned with commutative
(but this time non-linear) groups in their work on p-adic elliptic curves
(1936). Besides arithmetic applications, a local isomorphism is constructed
here between the group and the additive group, based on the integration of an
invariant differential form. This method applies equally to Abelian varieties,
as was noted by C. Chabauty soon afterwards, who used it without further
explanation to prove a particular case of the “Mordell conjecture” [35].
From then on, it was clear that the local theory ofLie groups could be applied
with scarcely any change to the p-adic case. The fundamental theorems of the
Lie groups—Lie algebras “dictionary” were established in 1942 in the thesis of
R. Hooke [37], a pupil of Chevalley; this work also contained the
p-adic analogue of E. Canan’s theorem on closed subgroups of real Lie
groups.
More recently, M. Lazard [42 b] developed a more precise forth of the
“dictionary” for compact analytic groups over QV He showed that the exis-
tence of a p-adic analytic structure on a compact group G is closely related to
that of certain filtrations on G and gives various applications of this (for
example to the cohomology of C). One of Lazard’s tools is an improvement of
Dynkin’s results on the convergence of the p—adic Hausdorff series [42 a].

1’ If, in spite of this lack of applications, we have mentioned “Banach" groups

in Chapter III, it is because Banach manifolds are being used more and more in
Analysis (and even for the study of finite-dimensional manifolds) and because,
moreover, this generalization offers no extra difficulty.

428
 HISTORICAL NOTE

VIILFREELIEALGEBRAS
It remains to speak of a series of works on Lie algebra; where the connection
with the theory of Lie group: is very tenuous; this research has on the other
hand important applications to the theory of “abstract” groups and more
especially nilpotent groups.
Its origin is the work of P. Hall [24], which appeared in 1932. However Lie
algebras are not discussed here: P. Hall had in mind the study of a certain
class ofp-groups, those which he called “regular”. But this led him to examine
in detail iterated commutators and the lower central series of a group; on this
subject he established a version oftheJacobi identity (cf. Chapter II, § 4-, no. 4,
formula (20)) and the “Hall formula”
(xi/yI = fly“(x, y)’“‘"‘”’. . . (cf. Chapter II, § 5, Exercise 9).
Almost immediately (in 1935—1937) appeared the fundamental works of W.
Magnus ([25 a] and [25 b]) and E. Witt [30]. In [25 a] Magnus used the same
algebra offormal power series A as Hausdorff (since called “Magnus algebra") ;
he embedded the free group F in it and used the natural filtration of A to
obtain a decreasing sequence (Fn) of subgroups of F; it is one of the first
examples of a filtratiun. He conjectured that the F“ coincide with the terms of
the lower central series of F. This conjecture was proved in his second memoir
[25 b]; also in this work he showed explicitly the close relation between his
ideas and those of P. Hall and defined the free Lie algebra L (as a subalgebra
of A) which he showed essentially to be identified with the graded algebra
of F. In [30], Witt completed this result on various poinm. He showed notably
that the enveloping algebra of L is a free associative algebra and deduced
immediately the rank of the homogeneous components ofL (“Witt formulae”).
As for the proof of the basis of L known as the “Hall basis” (cf. Chapter 11,
§ 2, no. 11), it seems that it only appeared in 1950 in a note of M. Hall [40],
although it was implicit in the works of P. Hall and W. Magnus quoted
above.

429
 BIBLIOGRAPHY

1. F. KLEIN and S. Luz: (a) Sur une certaine famille de courbes et surfaces,
C. R. Ami. Sci, 70 (1870), pp. [222—1226 and 1275—1279 (=[2], pp.
416—420 and [3], vol. 1, pp. 78—85); (b) Uber diejenigen ebenen Kurven,
welche dutch tin geschlossenes System von einfach unendlich vielen
vertauschbaren linearen Transformationen in sich fibergehen, Math.
Arm, 4 (1871), pp. 50—84 (=[2], pp. 424—459 and [3], vol. 1, Abh. XIV,
pp. 229—266).
2. F. KLEIN, Gemmmelle mathematiyclle Abhandlungen, Ed. I, Berlin (Springer),
1921. '
3. S. LIE, Gesammlte Abhandlungm, 7 VOL, Leipzig (Teubner): (a) Uber die
Reziprozitfitsverhfiltnisse des Reyeschen Komplexes, vol. I, Abh. V,
pp. 68—77 (=Gétl. Natl]. (1870), pp. 53—66); (b) Uber einc Klasse
geometrischer Transformationen, vol. I, Abh. XII, pp. 153—214
(=Chn‘5tiana For. (1871), pp. 182—245); (c) Uber partielle Differential-
gleichungen erster Ordnung, vol. III, Abh. VII, pp. 32—63 (=Chn'xlz'anu
Far. (1873), pp. 16—51); ((1) Theorie der Transformationsgruppen 11,
vol. V, Abh. III, pp. 42—75 (=Arthivf. Math., 1 (1876), pp. 152—193);
(J) Ueber Gruppen von Transformationen, v01. V, Abh. I, pp. 1—8
(=Gb‘tt. Nathr. (1874), pp. 529—542); (f) Allgcmcine Untersuchun-
gen fiber Difl'erentialgleichungen, die eine kontinuierliche endliche
Gruppe gestatten, vol. VI, Abh. III, pp. 139—223 (=Math. Ann., 25
(1885), pp. 71451); (g) Beitréige zur ailgemeincn Transformationens-
theorie, vol. VI, Abh. V, pp. 230436 (=Lez'pzign B”. (1888), pp,
14—21); (h) Theorie der Transformationsgruppen III, vol. V, Abh.
IV, pp. 78—133 (=Archiuf. Malh., vol. III, (1887), pp. 93—165).
4. S. LIE and F. ENGEL, Themie der Transfummlionsgruppm, 3 vol., Leipzig
(Teubner), 1888—1893.
5. J. J. SYLVESTER, Collated Mathematical Paperx, 4 vols., Cambridge, 1904—
1911.
6. A. CAYLEY, Callntm' Mathematical Papers, 1?} vols., Cambridge, 1889~
1898.

430
 BIBLIOGRAPHY

7 . C. JORDAN, Mémoire sur 13 groups de mouvements, Annali di Math, 11

(1868—1869), pp. 167—215 and 3324545 (=0euom‘, vol. IV, pp. 231—
302 .
8 . F. SGHUR: (a) Zur Theorie der aus Haupteinheiten gebildeten Komplexen,
Math. Arm, 33 (1889), pp. 49—60; (h) Neue Beg-rfindung der Theorie der
endiichen Transformationsgruppen, Math. Ann, 35 (1890), pp. 161—-
197; (c) Zur Them-1e der endlichen Transfounationsgwppen, Math.
Arm, 38 (1891), pp. 273—286; (d) Uber den analytischen Character der
cine cndliche continuierliche Transformationsgruppe damtellende
Funktionen, Math. Arm, 4] (1893), pp. 509—538.
9 . F. ENGEL: (a) Uber die Definitionsgleichung der oontinuierlichen Trans-
formtionsg‘ruppen, Math. Arm, 27 (1886), pp. 1—57; (b) Die Emeugung
der endlichen Transformationen einer projektiven Gruppe dutch die
infinitesimalen Transformationen der Gruppc, I, Leipzig” Ben, 44
(1892), pp. 279—296, II (mit Beitrégen von E. Study), ibid., 45 (1893),
pp. 659—696.
. L. MAURER, Uber allgemcinere Invariantcn-Systcme, Sitmgxbtr. Mfin-
chm, 18 (1888), pp. 103—150.
. W. KILLING, Die Zusammensetzung der stetigen endlichen Transfor-
mationsgruppen: (I) Math/11111., 31 (1883), pp. 252—290; (II) ibid., 33
(1889), pp. 1—4-8; (III) ibid., 34 (1889), pp. 57—122; (IV) ibid., 36
(1890), pp. 161—189.
. E. CARTAN, Oeuvres complétex, 6 VOL, Paris (Gautluer-Villars), 1952—54.
. J. E. CAMPBELL: (a) On a law of combination of operators bearing on the
theory of continuous transformation groups, Pm: London Math. Sea,
(1) 28 (1897), pp. 381—390; (b) On a law of combination of operators
(second paper), {17131, 29 (1898), pp. 14—32.
. H. POINCARE, Oeuvres, 11 vol., Paris (Gauthier-Villa“), 1916—1956.
. H. F. BAKER, Alternants and continuous groups, P100. London Math. Sm,
(2) 3 (1905), pp. 24—47.
. F. HAUSDORFF, Die symbolische Exponentialformel in der Gruppen-
theorie, Leipzig” Ben, 53 (1906), pp. 19—48.
. A. HURwrrz, Uber die Eneugung der Invarianten dumb Integration,
Gé'tt, Nachr. (1897), pp. 71—90 (=Math. Wake, vol. II, pp. 546—564).
. E. E. LEVI, Sulia struttura dei gruppi finiti e continui, Atli Act. Sci.
Torino, 40 (1905), pp. 551—565 (=Opm, vol. I, pp. 101—115).
. K. HENSEL, Uber die arithmetischen Eigenschaften der Zahlen, Jahmbn.
daD.M.V., 16 (1907), pp. 299—319, 388—393, 474-496.
. I. SCHUR, Neue Anwendungen der Integralrcchnung auf Problem: der
Invariantentheorie, Sitzungxbn. Berlin, 1924, pp. 189—208, 297—321,
346—355.
. H. WEYL, Theorie der Dmtellung kontinuierlicher halb—einfacher
Gruppen dutch linear: Transformationen, I, Math. Zeimhm fl, (1925),

431
 BIBLIOGRAPHY

pp. 271—309; I], ibid., 24 (1926), pp. 328—376; III, ibid., 24 (1926),
pp. 377—395 (= Wake, vol. II, pp. 543—647).
22. O. SCHREIER: (a) Abstmkte konlinuierliche Gruppcn, Abh. math. Sm.
Hamburg, 4 (1926), pp. 15—32; ([1) Die Verwandschafi: stetiger Gruppen
in grossen, MM, 5 (1927), pp. 233—244.
23. J. VON NEUMANN, Zur Theorie der Darstellung kontinuierlicher Gruppen,
Sitzungsber. Berlin, 1927, pp. 76—90 (=Callmed Works, vol. I, pp. 134—148).
24. P. HALL, A contribution to the theory of groups of prime power order,
Proc. Landon Math. Son, (3) 4 (1932), pp. 29—95. ‘
25. W. MAGNUS: (a) Beziehungen zwischen Gruppen und Idealen in einem
speziellen Ring, Math. Ann, 111 (1935), pp. 259—280; (1;) Uber Bezie-
hungen zwischen hfiheren Kommutatoren, J. Crelle, 177 (1937), pp.
105—1 15.
26. j. H. C. Wmuznmz (It) On the decomposition of an infinitesimal group,
Pm. Camb. Phil. Sea, 32 (1936), pp. 229—237 (=Mathematiml Works, I,
pp. 281—289); (b) Certain equations in the algebra of a semi-simple
infinitesimal group, Quart. Jaum. nfMaUL, (2) 3 (1937), pp. 220—237
(=Mathematz'ml War/cs, I, pp. 291—308).
27. I. ADO: (a) Note on the representation of finite continuous groups by
means of linear substitutions (in Russian), Bull. Phys. Math. Soc. Kazan,
7 (1935), pp. 3-43; (b) The representation of Lie algebras by matrices
(in Russian), Uspeln’ Mat. Nauk, 2 (1947), pp. 159—173 (English trans-
lation: Amer. Math. Sat. TramL, (1) 9, pp. 308-327).
28. N. JAoonsouz (a) Rational methods in the theory of Lie algebras, Ann. 13/
Math, 36 (1935), pp. 875—881; (b) Classes of restricted Lie algebras of
characteristic [1, II, Dulcz Math. Journal, 10 (1943), pp. 107-121.
29. G. Bum-mm: (a) Continuous groups and linear spaces; Rec. Math.
Moment, 1 (1936), pp. 635—642; (b) Representability of Lie algebras and
Lie groups by matrica, Arm. ty'Math., 38 (1937), pp. 526—532.
30. E. WITT, Treue Dmtellung Lieschen Ringe, J. Crelle, 177 (1937), pp.
152—160.
31. R. BRAUER, Eine Bcdingung fur vollstindige Reduzibiljtfit von Darstellun—
gen gewc'ihnlicher und infinitesimaler Gruppen, Math. Zeilschr., 41
(1936), pp. 330—339.
32. H. CASIMIR—B. L. VAN DER WAERDEN, Algebraischer Beweis der voll-
stindigen Reduzibilitit der Darstellungen halbeinfacher Liescher
Gruppen, Maih. Ann., 111 (1935), pp. 1—12.
33. A. WEIL, Sur les fonctions elliptiquesp—adiques, C. R. Ami. Sci, 203 (1935),
p. 22.
34-. E. LU'rz, Sur l’équation y‘ = x3 — Ax — B dans les corps [z-adiques, J.
Crelle, 177 (1937), pp. 23744-7.
35. C. CHABAUTY, Sur les points rationels des courbes algébriques de genre
supérieur a l’unité, C. R. Ami. Sal, 212 (1941), pp. 882—884.

432
 BIBLIOGRAPHY

36. L. S. PONTRJAGIN, Topolagiml groups, Princeton Univ. Press, 1939.

37. R. HOOKE, Linear p—adic groups and their Lie algebras, Ann. qalh., 43
(1942), pp. 641»655.
38. C. CHEVALLEY, Theory qie groups, Princeton Univ. Press, 1946.
39. E. DYNKIN: ((1) Evaluation of the coefficients of the Campbell—Hausdorff
formula (in Russian), Dakl. Akad. Naulc, 57 (1947), pp. 323—326; (17)
Nor-med Lie algebras and analytic groups (in Russian), Uspehi Mat.
Nauk, 5 (1950), pp. 135—186 (English translation: Amer. Math. Sac.
TramL, (1) 9, pp. 470—534).
40. M. HALL, A basis for free Lie rings and higher commutators in free groups,
Proc. Am”. Math. Son, 1 (1950), pp. 575-581.
4-1. D. MONTGOMERY-L. ZIPPIN, Topological Tmngfmation Groupr, New York
(Interscience), 1955.
42. M. LAZARDZ (a) Quclques calculs ooncemant la formule de Haudorfl',
Bull. Soc. Math. France, 9! (1963), pp. 435—451; (D) Groups analytiques
p—adiques, Publ. Math. LH.E.S., no. 26 (1965), pp. 389—603.

433
 INDEX OF NOTATION

The reference numbers indicate respectively the chapter, paragraph and

number (or, occasionally, exercise).

[x, y] (x, y elements of a Lie algebra): 1.1.2.

9° (9 a Lie algebra): 1.1.2.
9103.), gl(n, K), al(E), 91(71, K), 1(n, K), 91(71, K), n(n, K) (E a K-module):
1.1.2.
adrv, ad 1: (J: an element of a Lie algebra 9): 1.1.2.
[(1, b], [2, a], [11, z] (a, b submodules, 1 an element of a Lie algebra): 1.1.4.
99, 9kg, W‘s; (g a Lie algebra): 1.1.5.
fig (5; a Lie algebra): I l 6.
af(M) (M a K-module). .l .8
9031) (g a Lie algebra): I. l ..9
U, (U the enveloping algebra of a Lie algebra): I 2.1.
T", S", S'“: I.2. 5.
T," U," G": 1.2.6.
at“ (I: an element of a Lie algebra g, M a g—module): 1.3.1.
2", exp u (u a nilpotent endomorphism of a vector space over a field of charac-
teristic 0): 1.6.8.
C(p) (p a representation of a Lie algebra): 1.7.1.
@“9, 9‘9: I.1, Exercise 14.
2‘": 1.1, Exercise 20.
GL(n, K) (formal group): 1.1, Exercise 25.
o(<I>): I. 1, Exercise 26.
0(a) M): 0*(91M): 1U), 9(1)» d Z"(9, M); B’(9, M), H’(9. M). H‘(B. M)
g a Lie algebra, M a g-module): I.3, Exercise 12.
sp(2n, K). I.6, Exercise 25.
K: II.Conventions.
9, U = Us, a: g—>Ug: IL].
2, c, u, 1:“, 1;,” 53:11.1.1.

435
 INDEX OF NOTATION

E, 13*: 11.1.1.
P(E), 1r, 1‘, 6*: 11.1.6.
5(9), c,, n: 11.1.5.
fig: U(P(E)) —> E: 11.1.6.
M(X), [(w), Lib(X) = LibK(X): 11.2.1.
L(X) = LK(X): 11.2.2.
9: X )) L(X): 11.2.2.
(2, r):11.2.3.
L(u): 11.2.5.
Lib°(X), L5(X), L"(X): 11.2.6.
P": 11.2.7.
W19): 11.2.7.
H, dv: 11.2.10.
E = ‘1’(w): 11.2.11.
A(X) = AK(X), A*(X), Mo(X): 11.3.
n: 11.3.2.
(Ga), (GD: 11.4.1.
II: I .42.
gr(G), gra(G): 11.4.3.
F(X), A(X), A"(X): 11.5.
A(X), to: 11.5.1.
:(a): 11.5.2.
1(x), exp(x), 103(31): 11.6.1.
:(X), [(X): 11.6.1.
£(X): 11.6.2.
a H b: 116.2.
H, H," H": 11.6.4.
H, Q: 11.7.2.
A, ep, logA, 15(AI, A): 11.7.3.
0,6: I, _ 1’ 11.8.
8(a): 11.8.1.
h(;r,y): 11.8.3.
Gflz'IIflA.
{L(u): 11.App. Y
‘1 9G, fig), 3(g), Int(g),f: 111.Conventions.
GL(E), GL(n, K): 111.].l, 111.3.10.
GV: 111.1.2.
1(g), p(x): 11115.
(G, g, 0, m):111.1.10.
T(m): 111.2.1.
T(G), T(q:): 111.2.2.
Ht':111.3.1, 111.3.18.

436
 INDEX OF NOTATION

U(G), U+(G), U,(G), U: (G): 111.3.1, 111.3.18.

T;"(G), ’I‘g”’(G), 31"”(G): 111.3.1, 111.3.18.
Hf: 111.3.4, 111.3.18.
D,: 111.3.5, 111.3.18.
Li, R,: 111.3.6, 111.3.18.
L(G): 111.3.7, [11.3.18.
L(:p): 111.3.8, 111.3.18.
(t,f): 111.3.9, 111.3.18.
SL(E): 111.3.10.
Ad, Ad(g)z 111.3.12.
[44]": 111.3.14.
mod(to)u, mod q): III.3.16.
f‘hb’: III.3.17, 111.3.18.
H: 111.4.2.
g‘, 1110:): 111.4.3.
any, x‘“: 111.5.
can, B(x,y): 1115.1.
:3: 111.5.2.
¢,, 4;”, (5):111.5.3.
E(x), Lu): 111. 5.4.
M, x ,, 6-2-
Ad(a) = Int(a):111 6.2.
exp, expa:111.6.4.
Ad(G) = Int(L(G)): 111.6.4.
L(p): 111.65.
C: 1116.10.
A, m, p: 111.7.
G(a): 111.7.4.
1,: 111.7.5.
G,, logG, log: III.7.6.
fits, 56:11131.
Za(A), Za(n). MAM. (0511193.
NG(A): N501): n,(a) 111.
R, N r,n:111..79
«1 ® -® x... T01). 50'), /\(1r),T"(n),S"(n), A“(v=)= ULApp-

437
 INDEX OF TERMINOLOGY

The reference numbers indicate respectively the chapter, paragraph and num-
ber (or, occasionally, exercise)
Adjoint group of a real or complex Lie group: III.6.4-.
linear mapping of an element of a Lie algebra: 1.1.2.
representation of a Lie algebra: 1.3.1.
representation of a Lie group: 111.3.12.
Ado’s Theorem: 1.7.3.
Algebra (not necessarily associative): 1.1.1.
derived from an algebra by extending the scalars: 1.1.1.
enveloping, of a Lie algebra: 1.2.1.
Lie: 1.1.2.
opposite: 1.1.1.
product. 1.1.1.
quotient. 1.1 1
mtricted enveloping, of a Lie p—algebra: 1.2, Exercise 6.
symmetric, of a module. 1.2.5.
Almost simple Lie group: 111.9.8.
Alternants of degree 7:: 11.2.6.
Antiautomorphism, principal, of the enveloping algebra of a Lie algebra: 1.2.4.
Associated (bilinear form) with a g-rnodule (a representation): 1.3.6.
(law of infinitesimal operation) with a law of operation: 111.3.7.
Associative algebra, free: 1113.
Autonmr'phism, special, of a Lie algebra: 1.6.8.
Basic commutators: 11.5.4.
family ofa Lie algebra: 11.2.3.
Bieberbach’s Theorem: 111.4, Exercise 13.
Bigebra: 11.1.2.
Biinvariant section: 111.3.13.
Binomial polynomial: 11.5, Exercise 4.
Bracket: 1.1.2.

439
 INDEX OF TERMINOLOGY

Canonical difl'erential form, leit: 111.3.13, 111.3.18.

Carmn, criterion: 1.5.4.
Casimir element: 1. .7.
Central extension: 1.1.7.
filtration of a group: 11.4.4.
Centralizer: 1.1.6, 111.9.3.
Centre of a Lie algebra: 1.1.6.
Characteristic ideal: 1.1.4.
Class, nilpotency: 11.2.7.
of simple representations: 1.3.1.
Coboundaries, cochains, cocyeles with values in a g-module: 1.3, Exercise 12.
Cogebra: 11.1.1.
Commutative Lie algebra: 1.1.3.
Compatible group and manifold structures: 111.1.1.
Completely invariant bilinear form: 1.3.6.
reducible representation: 1.3.1.
Complex Lie group: 111.1.1, 111.8.1.
Complexification of a real Lie group: 111.6.10.
Component, isotypical, of a g-module: 1.3.1.
simple, of a semi-simple Lie algebra: 1.6.2. -
Conjugate Lie group of a complex Lie group: III.l.l.
Constant term of an element of the enveloping algebra of a Lie algebra: 1.2.1.
term of an element of the Magnus algebra: 11.5.2.
Constants of structure of an algebra with respect to a basis: 1.1.1.
Containing a simple representation In times (representation): 1.3.1.
Contragredient of an analytic representation: 111.3.11.
Convolution of a point distribution and a function: 111.3.4.
product: 111.3.1, 111.3.18.
Counit of a oogebra: 11.1.1.
C'-connected subset 01a Lie group: 111.6.2.
Criterion, Cartan: 1.5.4.

Derivation of an algebra: 1.1,1.

inner, of a Lie algebra: 1.1.2.
left invariant, relative to a formal group: 1.1, Exercise 24-.
Derivative, partial, in the algebra of a free group: 11.5, Exercise 2.
Derived (algebra) from an algebra by extending the scalars: 1.1.1.
(representation) from a representation by extending the scalars: 1.3.8.
ideal: 1.1.5.
series: 1.1.5.
Differential, left, of a mapping into a Lie group: 111.3.17, 111.3.18.
Dimension of a representation: 1.3.1.
Dual representation: 1.3.3.

440
 mDEX 0F TERMINOIDGY

Element, Casimir: 1.3.7.

nilpotent: 1.6.3.
of degree n in the graded algebra associated with the enveloping algebra
of a Lie algebra: 1.2.6.
of filtration <1: in the enveloping algebra of a Lie algebra: 1.2.6.
semi-simple: 1.6.3.
Elements, permutable, in a Lie algebra: 1.1.3.
Elimination theorem: 11.2.9.
Engel’s Theorem: 1.4.2.
Enveloping algebra of a Lie algebra: 1.2.1.
bigebra: 11.1.4.
Equivalent extensions of a Lie algebra: I.l.7.
Exhaustive filtration: 11.4.1.
Exponential: 11.6.1, 111.4.3, 111.6.4.
Extension, central: 1. 1.7.
inessential: 1.1.7.
of a g—module, of a repraenmtlon of 3: 1.7.2.
ofa Lie algebra by a Lie algebra: 1.1.7.
trivial: I.l.7.
Extensions, equivalent: I.l.7.

Faithful representations: 1.3.1.

Field ofpoint distributions: 111.3.5, 111.3.18.
Filtered bigebra: 11.1.3.
Filtration, real, on a group: 11.4.1.
First species, canonical chart of the: 111.4.3.
species, system of canonical coordinates of the: 111.4.3.
Foliation, left, associated with a Lie subalgebra: 111.4.1.
Form, bilinear, associated with a g—module (a representation of 9): 1.3.6.
bilinear, completely invariant: 1.3.6.
bilinear, invariant: 1.3.6.
Killing: 1.3.6.
Formulae, Jacobson: 1.1, Exercise 19.
Free Lie algebra: 11.2.2.
Lie p—algebra: 11.3, Exercise 4.
magma: 11.2.1.
Function, order, associated with a filtration: 11.4.2.

G-bundle, vector: 111.1.8.

g-module, left (right): 1.3.1.
pure, of species N: 1.3.1.
trivial: 1.3.1.

4-41
 INDEX OF TERMINOLOGY

Graded group associated with a filtered group: 11.4.3.

Lie algebra, associated: 11.4.4.
Graduation, total, of L(1): 11.2.6.
Group
commutative formal: 1.1, Exercise 24.
formal: 1.1, Exercise 24.
formal linear: 1.1, Exercise 25.
formal orthogonal: 1.1, Exercise 26.
formal symplectic: 1.1, Exercise 26.
germ, Lie: 111.1.10.
germ, Lie, defined by a Lie algebra: 111.4.2.

Hall basis: 11.2.11.

formula: 11.5, Exercise 9.
set: 11.2.10.
Hausdorff function: 11.7.2, 11.8.3.
group: 11.6.2.
inversion formula: 11.6, Exercise 4.
series: 11.6.4.
Holomorph of a Lie algebra: 1.1, Exercise 16. I
Homomorphism, algebra: 1.1.1.
canonical, of the enveloping algebra of a Lie subalgebra of 9 into the
enveloping algebra of 9: 1.2.3.
canoniml, of the symmetric algebra of 9 onto the graded algebra associated
with the enveloping algebra of 9: 1.2.6.
formal: 1.1, Exercise 24.

Ideal, characteristic: 1.1.4.

derived: 1.1.5.
left (right, two-sided) of an algebra: 1.1.1.
of a Lie algebra: 1.1.4.
Identity, Jacobi: 1.1.2.
Induced Lie group structure: 111.4.5.
Inessential extension: 1.1.7.
Infinitesimal automorphism: 111.10.].
Inner derivation: 1.1.2.
Integral filtration of a group: 11.4.1.
subgroup of a Lie group: 111.6.2.
Invariant bilinear form: 1.3.6.
left, point distributions, field of: 111.315.
of a g-module (a representation): 1.3.5.
section: “1.3.13.

442
 INDEX OF TERmNOLDGY

Inverse image Lie group structure: 111.1.9.

Irreducible representation: 1.3.1.
Isomorphie representations: 1.3.1.
Isomorphism, canonical, of the symmetric algebra of g onm the underlying
vector space of its enveloping algebra: 1.2.7.
Isotypical component of a g-module: 1.3.1.

Jacobi identity: 1.1.2.

Jordan’s Theorem: 111.4, Exercise 11.

Kernel of an extension: 1.1.7.

Killing form: 1.3.6.

Largest nilpotency ideal of a g-module (a representation): 1.4.3.

Largest nilpotent ideal of a Lie algebra: 1.4.4.
Law chunk ofoperation: 111.1.11.
formal group: 1.1, Exercise 24.
ofinfinitaimal operation: 111.3.7, 111.3.18.
Length of an element of a free magma: 11.2.1.
Levi subalgebra: 1.6.8.
Levi-Malcev Theorem: 1.6.8.
Lie algebra: 1.1.2.
algebra, commutative: 1.1.3.
algebra, characteristically nilpotent: 1.4, Exercise 19.
algebra, nilpotent: 1.4.1.
algebra of a formal group: 1.1, Exercise 24.
algebra of a Lie group: 111.3.7.
algebra of a Lie group germ:111.3.18.
algebra of endomorphisms of a module. I l.2.
algebra, reductive: 16 4
algebra, semi-simple. .éJ.
algebra, simple: 1.6.2.
algebra, solvable: 1.5.1 .
group: 111.1.1.
Lie’s Theorem: 1.5.3.
Locally isomorphic group germ: 111.1.10.
Logarithm: 11.6.1, 111.7.6,
Lower central series: 1.1.5.

Magnus algebra: 11.5.1.

group: 11.5.2.
Mapping, canonical, of a Lie algebra into its enveloping algebra: 1.2.1.
Maurer-Cartan formulae: 111.3.14, 111.3.18.

443
 INDEX OF TERMINOLOGY

Miibius function: 11, Appendix.

inversion formula: 11, Appendix.
Morphism, Lie group: 111.1.2.
Lie group germ: 111.1.10.
Multigraduation of L(I): 11.2.6.

Nilpotent Lie algebra: 1.4.1, 11.2.7.

radical of a Lie algebra: 1.5.3.
Normable algebra: 11.7.
Normalizcr: 1.1.4, 111.9.4.
Normed Lie algebra: 11.7, 11.8.2.

Opposite algebra: 1.1.1.

Order of an element in a filtered group: 11.4.2.

p—adic Lie group: 111.1.1, 111.8.1.

[I-algebra, Lie: 1.1, Exercise 20.
p-unipotent Lie: 1.4, Exercise 23.
p—core of a commutative Lie p—algebra: 1.1, Exercise 23.
p—derivation: 1.2, Exercise 7.
P-envelope of a nilpotent group: 11.4, Exercise 15.
p—homomorphism: 1.1, Exercise 20.
p-ideal: 1.1, Exercise 22.
P-integer: 11.4, Exercise 14.
p—mapping: 1.1, Exercise 20.
p-polynomial: 1.7, Exercise 5.
P—saturation of a subgroup of a nilpotent group: 11.4, Exercise 14.
P-torsion group, P-torsion-free group: 11.4, Exercise 14.
Permutable elements: 1.1.3, 111.9.3.
Poincaré-Birkhofi-Witt Theorem: 1.2.7.
Polynomial, Lie: 11.2.4.
mapping: 11.2.4.
Presentation of a Lie algebra: 11.2.3.
Primitive element of a bigebra: 11.1.2.
Product algebra: 1.1.1.
of Lie groups: 111.1.4.
semi-direct, of Lie algebras: 1.1.8.
tensor, of representations: 1.3.2, 111, Appendix.
Pure g-module of species N: 1.3.1.
representation: 1.3.1,

Quasi-subgroup, Lie: 111.13.

444
 INDEX OF TERMINOLOGY

Quotient algebra: 1.1.1.

ofa Lie group: 111.1.6.
representation: 1.3.1.
Radical, nilpotent, of a Lie algebra: 1.5.3.
ofa Lie algebra: 1.5.2.
ofa Lie group: 111.9.7.
Real Lie group: 111.1.1, 111.8.1, 111.8.2.
Reductive Lie algebra: 1.6.4.
(Lie subalgebra) in a Lie algebra: 1.6.6.
Relators: 11.2.3.
Replica of an endomorphism: 1.5, Exercise 14.
Representation, adjoint: 1.3.1.
analytic linear, of a Lie group: 111.1.2.
completely reducible: 1.3.] .
containing a simple representation n times: 1.3.1.
dual: 1.3.3.
faithful: 1.3.1.
irreducible: 1.3.1.
obtained by extending the ring of scalars: 1.3.8.
ofa Lie algebra: 1.3.1.
pure, of species a: 1.3.1.
quotient: 1.3.1.
semi-simple: 1.3.].
simple: 1.3.1.
Representations, isomorphic: 1.3.1.
similar: 1.3.1.
Restriction of scalars, Lie group derived from a Lie group by: III.l.1.
Roots of a solvable Lie algebra: 111.9, Exercise 17.
Second species, canonical chart of the: 111.4.3.
species, system of canonical coordinates of the: 111.4.3.
Section of a vector bundle: 111.1.8.
Semi—direct product of Lie algebras: 1.1.8.
product of Lie groups: 111.1.4.
Semi-simple Lie algebra: 1.6.1.
Lie group: 111.9.8.
representation: 1.3.1.
Separated filtration: 11.4.1.
Series, composition, joining two subalgebras: 1.1, Exercise 14-.
derived: 1.1.5.
Lie formal power: 11.6.3.
lower central, of a Lie algebra: 1.1.5, 11.2.7.
upper central, of a Lie algebra: 1.1.6.

4-45
 INDEX OF TERMINOLOGY

Similar representations: 1.3.1.

Simple component of a semi-simple Lie algebra: 1.6.2.
Lie algebra: 1.6.2.
representation: 1.3.1.
Solvable Lie algebra: 1.5.1.
Space, whomology, with values in a g-module: 1.3, Exercise 12.
Lie homogeneous: 111.1.6.
Special automorphism: 1.6.8.
Species of an isotypical component, of a pure g-module: 1.3.1.
Standard group: 111.7.3.
Subalgebra: 1.1.1.
Lie: 1.6.8.
reductive in a Lie algebra: 1.6.6.
subinvariant: 1.1, Exercise 14.
Subgroup germ, Lie: 111.1.10.
Lie: 111.1.3.
Subrepreientation: 1.3.1.
Sum, direct, of representations: 1.3.1.
Symmetric algebra of a module: 1.2.5.

t-rh power mapping: 111.4.3.

Tangent law of composition: 111.2.1.
Lie subgroup: 111.4.5.
Theorem, Ado’s: 1.7.3.
Engel’s: 1.4.2.
Levi-Malcev: 1.6.8.
Lie’s: 1.5.3.
PoincareBirkhoH-Witt: 1.2.7.
cl’s: 1.6.2.
Zassenhaus’s: 1.7.2.
Transporter: 1.1, Exercise 1.
Triangular group, upper strict: 11.4.6.
Trivial extension: 1.1.7.
g-module: 1.3.1.
vector G-bundle: III.l.8.
Trivialization, right (rcsp. left), ofT(G): 111.2.1, 111.2.2.
Two-sided ideal, 1.1.1.
Type (N), real Lie group of: 111.9, Exercise 29.

u-primitive elements of a. cogebra: 11.1.1.

Unipotent endomorphism: 111.9.5.
Universal covering of a connecied Lie group: 111.1.9.

44-6
 INDEX OF TERMINOLOGY

Upper central scrim of a group: 11.4, Exercise 18.

central sen'es of a Lie algebra: 1.1.6.

Weyl’s Theorem: 1.6.2.

Zassenhaus’s Theorem: 1.7.2.

447
 SUMMARY
Qfeerlain properties qffinite-dimemional Lie algebra: am afield qfeharaeteristir 0.

Let g be a Lie algebra, tits radical, 11 its largest nilpotent idea], 5 its nilpotent
radical and l the orthogonal of 9 relative to the Killing form. Then it, n, b, I are
characteristic ideals and t D I D n D a.

(I) Any (me of thefollowing propertier characterize: :emi—simpk Lie algebras:

(l) r = (0}; (2) n = {0); (3) l = {0); (4) every commutative ideal of g is
zero; (5) the algebra g is isomorphic to a product of simple Lie algebras;
(6) every finite-dimensional representation of g is sewn-simple.

(11) Any one qf thefillnwing properties characterize: reductive Lie algebras:

(l) a = {0}; (2) I: is the centre ofg; (3) 99 is semi-simple; (4-) g is the product of
a semi-simple algebra and a commutative algebra; (5) the adjoint representa-
tion of g is semi-simple; (6) g has a finite-dimensional representation such that
the associated bilinear form is non-degenerate; (7) g has a finite-dimensional
semi-simple faithful representation.

(III) Any one of thefollowing properties characterize: solvable Lie algebras:

(1) 9’9 = {O} for sufliciently large p; (2) there exists a decreasing sequence
9 = 90 D 91 D . - ~ 3 9,, = {0} of ideals ofg such that the algebras my,“ are
commutative; (3) there exists a decreasing sequence

9 =93 D 913-": 9;: ={0}

of subalgebras of g such that 9: H is an ideal of 5" and gflgf H is commutative;
(4-) there exists a decreasing sequence 9 = g; 3 g; 3 - ~ . 3 9;. = {0} of sub-
algebras of 9 such that 93'“ is an ideal of wdimension l in 9!; (5) t D 95;
(6) 9g is nilpotent.

44-9
 SUMIMARY

(IV) Any one qfthefvllawing properties chamelm'zex nilpotent Li: algebras:

(l) V’g = {0) for sufficiently large p; (2) fig = g for sufficiently large p;
(3) there exists a decreasing sequence 9 = 90:: g, 3 ~ - - D g, = {0) of ideals
of 9 such that [9, g] c 9. +1; (4:) there exists a decreasing sequence
9 = 56 3 91: . . . 3 9;, = (0) ofideals ofg such that [9, 92] C 91“ and the
g;/g{+1 are 1»dimensional; (5) there exists an integer i such that
(M31) ° (adxa) °- ‘ ~° (ad x4) = 0
for all :5, . . ., x; in 9; (6) for all xeg, adx is nilpotent.
(V) g commutative => 9 nilpotent : the Killing form ofg is zero => 9 solvable.
g commutative : g reductive.
g semi-simple => 9 reductive.
(VI) Characterization: qft:
(l) t is the largest solvable ideal of 9; (2) r is the smallest ideal such that 5/: is
semi-simple; (3) t is the only solvable idml such that g]: is semi-simple; (4») r is
the orthogonal 01‘99 relative to the Killing form.
(VII) Chamrtnization: a:
(1) n is the largest nilpotent ideal of 9; (2) n is the largest nilpotent ideal of r;
(3) n is the set of x E r such that ad“): is nilpotent; (4-) n is the set of x E t such
that ad; is nilpotent; (5) n is the largest ideal of 9 such that, for all x e n,
ad,x is nilpotent; (6) n is the set of x E 9 such that adg’belongs to the radical
of the associative algebra generated by l and the add (y s 9),
(VIII) Characterizations qfs:
(l) s is the intersection of the kernels of the finite-dimensional simple repre-
sentations of g; (2) a is the smallest of the kernels of the finite-dimensional semi-
simple representations of 9; (3) a is the intersection of the largest nilpotency
ideals of the finite-dimensional representations of g; (4) s is the smallest ideal of
a such that 9/9 is reductive; (5) a = r n 99; (6) e = [r, g]; (7) s is the inter-
section of the orthogonals of 9 relative to the bilinear forms associated with the
finite-dimensional representations of 5.

450
 NICOLAS BOURBAKI

Elements of Mathematics

English translations public/zed:

General Topology, Part I
General Topology, Part II
Theory of Sets
Commutative Algebra
Algebra, Part I
Lie Groups and Lie Algebras, Part I
 General Topology, Part I

CHAPTER I. TOPOLOGICAL Smucrumzs

§ 1. Open sets, neighbourhoods, closed sets
§ 2. Continuous functions
§ 3. Subspaces, quotient spaoes
§ 4-. Product of topological spam
§ 5. Open mappings and closed mappings
§6. Filters
§7. Limits
§ 8. Hausdorff spaces and regular spaces
§ 9. Compact spaces and locally compact spaces
§ 10. Proper mappings
§ ll. Connectedness
Exercises
CHAPTER II. UNIFORM Sumo-runs
§ 1. Uniform spaces
§ 2. Uniformly continuous functions
§ 3. Complete spaces
§ 4'. Relations between uniform spaces and compact spaca
Exercises
CHAPTER III. TOPOLOGICAL GROUPS
§ 1. Topologies on groups
§ 2. Subgroups, quotient groups, homomorphism, homogeneous spaces,
product groups '
§ 3. Uniform structures on groups
§ 4. Groups operating properly on a topological space; compactness in
topological groups and spaces with operators
§ 5. Infinite sums in commutative groups
§6. Topological groups with operators; topological rings, division rings and
fields
§ 74 Inverse limits of topological groups and rings
Exercises
CHAPTER IV. REAL NUMBERS
§ 1. Definition of real numbers
§ 2. Fundamental topological properties of the real line
§ 3. The field of real numbers
§4. The extended real line
§ 5. Real-valued functions
§6. C' ‘ and ' ' real-valued “
§ 7. Infinite sums and products of real numbels
§ 8. Usual expansions of real numbers; the power of R
Exercises
 General Topology, Part II

CHAPTER V. ONE-PARAMETER Guam

§ 1. Subgroups and quotient groups of R
§ 2. Measurement of magnitudes
§ 3. Topological characterization of the groups R and T
§4. Exponentials and logarithms
Exercises
CHAPTER VI. REAL NUMBER SPACES AND PROJECTIVB SPACE
§ 1. Real number space R“
§ 2. Euclidean distance, balls and spheres
§ 3. Real projective spaces
Exercises
CHAPTER VII. THE AnnmvE Gxours R"
§ 1. Subgroups and quotient groups of R"
§ 2. Continuous homomorphisms of R" and its quotient groups
§ 3. Infinite sums in the groups R”
Exercises
CHAPTER VIII. COMPLEX NUMBERS
§ 1. Complex numbers, quaternions
§ 2. Angular measure, trigonometric functions
§ 3. Infinite sums and products of complex numbers
§4. Complex number spaces and projective spaces
Exercises '
CHAPTER IX. UsE or REAL NUMBERS IN GENERAL TOPOLOOY
§ 1. Generation of a uniformity by a family of pseudometrim; uniformizable

§ 2. Metric spaces and metrizable spam

§ 3. Metrizable groups, valued fields, normed spaces and algebras
§ 4. Normal spaces
§ 5. Baire spaces
§ 6. Polish spaces, Souslin spaces, Borel sets
Appendix: Infinite products in normed algebras
Exercises
CHAPTER X. FUNCTION SPACE:
§ 1. The uniformity of G-convcrgence
§ 2. Equicontinuous sets
§ 3. Special function spaces
§ 4-. Approximation of continuous real-valued functions
Exercises
 Theory of Sets

CHAPTER I. Dzscrumou or FORMAL MAmuAncs

§ 1. Terms and relations
§2. Theorems
§ 3. Logical theories
§4. Quantified theories
§5. Equalitarian theories
A 4‘ F‘- i a-iuu of terms and
Exerdses
CHAPTER II. THEORY or SETS
1. Collectivizing relations
2. Ordered pairs
3. Corrapondcnces
4. Union and intersection of a family of sea
5. Product of a family of sets
6. Equivalence relations
Exercises
CHAPTER III. ORDERED Sis-rs, CARDINALS, INTEGERS
1. Order relations. Ordered sets
2. Well-ordered sea
3. Equipotent sets. Cardinals
4. Natural integers. Finite sets
5. Properties of integers
6. Infinite sets
7. Inverse limits and direct limits
Exercises
CHAPTER IV. S'rnuc'rumzs
1. Structures and isomorphisms
2. Morphisms and derived structures
3. Universal mappings
Exercises
SUMMARY or RESULTS
Introduction
§ 1. Elements and subsets of a set
§2. Functions
§ 3. Products of sets
§ 4. Union, intersection, product of a family of set:
§ 5. Equivalence relations and quotient sets
§ 6. Ordered sets
§ 7. Powers. Countable sets
§ 8. Scales of sets. Structures
 Commutative Algebra
CHAPTER I. FLAT Mommas
§ 1. Diagram and exact sequences
2. Flat modules
3. Faithfully flat modq
4. Flat modules and “Tor” functors
Exercises
CHAPTER II. LOCALIZATION
1. Prime ideals
2. Rings and modules of fractions
3. Local rings. Passage from the local to the global
4. Spectra of rings and supports of modules
5. Finitely generated projective modules. Invertible fractional ideals
Exercises
Cram: III. GmuA-rxons, Firm-nous AND Tovomcm
l. Finitely generated graded algebras
2. General results on filtered rings and modules
3. m-adic topologies on Noethetian rings
4-. Lifting in complete rings
5. Flatness properties of filtered modules
Exercises
CHAPTER IV. Assocmren PRIME Innis AND Pmumw Decourosmou
1. Prime ideals associated with a module
2. Primary decomposition
3. Primary decomposition in graded modules
Exercises
CHAPTER V. IN-reosks
§ 1. Notion of an integral element
§2. The [in of prime ideals
§ 3. Finitely generated algebras over a field
Exercises
CHAPTER VI. Vary/mews
§ 1. Valuation rings
§2. Places
§ 3. Valuations
§ 4. The height of a valuation
§ 5. The topology defined by a valuation
§ 6. Absolute values
§7. Approximation theorem
8. F i of a ' ' to an
§ 9. Application: locally compact fields
§ 10. ' of a ' ‘ to a ‘
Exercises
 COWUTATIVE ALGEHRA

CHAPTER VII. vxsoxs

§ 1. Krull domains
§ 2. Dedekind domains
§ 3. Factorial domains
§ 4. Modules over integrally closed Noetherian domains
Exercises
 Algebra, Part I

CHAPTER I. ALGEBRAIG STRUCTURES

1. Laws of composition; associativity; commutativity
2. Identity element; cancellable elements; invertible elements
3. Actions
4. Groups and groups with operators
5. Groups operating on a set
6. Extensions, solvable groups, nilpotent groups
7. Free monoids, free groups
8. Rings
9. Fields
10. Inverse and direct limits
Exercises
CHAPTER II. LINEAR ALGEERA
1. Modules
2. Modules of linear mappings. Duality
3. Tensor products
4. Relations between tensor products and homomorphism modules
5. Extension of the ring of scalars
6. Inverse and direct limits of modules
7. Vector spaces
8. Restruction of the field of scalars in vector spaces
9. Affine spaces and projective spam
10. Matrices
ll. Graded modules and ring:
Appendix. Pseudomodules
Exercises
CHAPTER III. Tsusox ALGEBRAS, Emmox Arcturus, SYMMETRIC ALGEBRAS
1. Algebras
Examples of algebras
pmusvewp

Graded algebras
Tensor products of algebras
Tensor algebra. Tensors
Symmetric algebras
. Exterior algebras
. Determinants
Norms and traces
§ 10. Derivations
§ 11. Cogebras, products of multilinear forms, inner products and duality
Appendix. Alternative algebras. Octonions
Exercises
IlADE AND PRINTED' IN GREAT BRITAIN BY
WILLIAII CLOWES & SONS, LIMITED
LONDON, 330cm AND cowl-mm
ISBN 3-540-50218-1
ISBN 0—387-50218-1