This action might not be possible to undo. Are you sure you want to continue?

# GAUGE

THEORY

FOR

FIBER

BUNDLES

Peter W. Michor

Mailing address: Peter W. Michor,

Institut f¨ ur Mathematik der Universit¨at Wien,

Strudlhofgasse 4, A-1090 Wien, Austria.

E-mail michor@awirap.bitnet

1980 Mathematics subject classiﬁcation 53-02, 53C05, 53C10, 58-02, 58D05, 58A35

Extended version of a series of lectures held at the Institute of Physics

of the University of Napoli, March 28 — April 1, 1988

Typeset by A

M

S-T

E

X

1

Table of contents

Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2

1. Notations and conventions . . . . . . . . . . . . . . . . 5

2. Calculus of smooth mappings . . . . . . . . . . . . . . . 6

3. Calculus of holomorphic mappings . . . . . . . . . . . . . 10

4. Calculus of real analytic mappings . . . . . . . . . . . . . 12

5. Inﬁnite dimensional manifolds . . . . . . . . . . . . . . . 14

6. Manifolds of mappings . . . . . . . . . . . . . . . . . . 17

7. Diﬀeomorphism groups . . . . . . . . . . . . . . . . . . 22

8. The Fr¨olicher-Nijenhuis Bracket . . . . . . . . . . . . . . 27

9. Fiber Bundles and Connections . . . . . . . . . . . . . . 36

10. Principal Fiber Bundles and G-Bundles . . . . . . . . . . 44

11. Principal and Induced Connections . . . . . . . . . . . . 60

12. Holonomy . . . . . . . . . . . . . . . . . . . . . . . 76

13. The nonlinear frame bundle of a ﬁber bundle . . . . . . . . 81

14. Gauge theory for ﬁber bundles . . . . . . . . . . . . . . 83

15. A classifying space for the diﬀeomorphism group . . . . . . 86

16. A characteristic class . . . . . . . . . . . . . . . . . . 90

17. Self duality . . . . . . . . . . . . . . . . . . . . . . 98

References . . . . . . . . . . . . . . . . . . . . . . . . 101

List of Symbols . . . . . . . . . . . . . . . . . . . . . 102

Index . . . . . . . . . . . . . . . . . . . . . . . . . . 103

2

Introduction

Gauge theory usually investigates the space of principal connections

on a principal ﬁber bundle (P, p, M, G) and its orbit space under the

action of the gauge group (called the moduli space), which is the group

of all principal bundle automorphisms of P which cover the identity

on the base space M. It is the arena for the Yang-Mills-Higgs equa-

tions which allows (with structure group U(1) SU(2)) a satisfactory

uniﬁed description of electromagnetic and weak interactions, which was

developed by Glashow, Salam, and Weinberg. This electro-weak the-

ory predicted the existence of massive vector particles (the intermediate

bosons W

+

, W

−

, and Z), whose experimental veriﬁcation renewed the

interest of physicists in gauge theories.

On the mathematical side the investigation of self dual and anti self

SU(2)-connections on 4-manifolds and of their image in the orbit space

led to stunning topological results in the topology of 4-manifolds by

Donaldson. The moduli space of anti self dual connections can be com-

pleted and reworked into a 5 dimensional manifold which is a bordism

between the base manifold of the bundle and a simple manifold which

depends only on the Poincar´e duality form in the second dimensional

homology space. Combined with results of Freedman this led to the

discovery of exotic diﬀerential structures on R

4

and on (supposedly all

but S

4

) compact algebraic surfaces.

In his codiﬁcation of a principal connection [Ehresmann, 1951] be-

gan with a more general notion of connection on a general ﬁber bundle

(E, p, M, S) with standard ﬁber S. This was called an Ehresmann con-

nection by some authors, we will call it just a connection. It consists

of the speciﬁcation of a complement to the vertical bundle in a diﬀeren-

tiable way, called the horizontal distribution.

One can conveniently describe such a connection as a one form on

the total space E with values in the vertical bundle, whose kernel is

the horizontal distribution. When combined with another venerable no-

tion, the Fr¨olicher-Nijenhuis bracket for vector valued diﬀerential forms

(see [Fr¨olicher-Nijenhuis, 1956]) one obtains a very convenient way to

describe curvature and Bianchi identity for such connections. Parallel

transport along curves in the base space is deﬁned only locally, but one

may show that each bundle admits complete connections, whose paral-

lel transport is globally deﬁned (theorem 9.10). For such connections

one can deﬁne holonomy groups and holonomy Lie algebras and one

may prove a far-reaching generalization of the Ambrose Singer theorem:

A complete connection tells us whether it is induced from a principal

connection on a principal ﬁber bundle (theorem 12.4).

Introduction 3

A bundle (E, p, M, S) without structure group can also be viewed as

having the whole diﬀeomorphism group Diﬀ(S) of the standard ﬁber

S as structure group, as least when S is compact. We deﬁne the non

linear frame bundle for E which is a principle ﬁber bundle over M with

structure group Diﬀ(S) and we show that the theory of connections on

E corresponds exactly to the theory of principal connections on this non

linear frame bundle (see section 13). The gauge group of the non lin-

ear frame bundle turns out to be just the group of all ﬁber respecting

diﬀeomorphisms of E which cover the identity on M, and one may con-

sider the moduli space Conn(E)/ Gau(E). There is hope that it can be

stratiﬁed into smooth manifolds corresponding to conjugacy classes of

holonomy groups in Diﬀ(S), but this will be treated in another paper.

The diﬀeomorphism group Diﬀ(S) for a compact manifold S admits

a smooth classifying space and a classifying connection: the space of all

embeddings of S into a Hilbert space

2

, say, is the total space of a princi-

pal bundle with structure group Diﬀ(S), whose base manifold is the non

linear Grassmanian of all submanifolds of

2

of type (diﬀeomorphic to)

S. The action of Diﬀ(S) on S leads to the (universal or classifying) as-

sociated bundle, which admits a classifying connection: Every S-bundle

over M can be realized as the pull back of the classifying bundle by a

classifying smooth mapping from M into the non linear Grassmanian,

which can be arranged in such a way that it pulls back the classifying

connection to a given one on E — this is the Narasimhan-Ramadas

procedure for Diﬀ(S).

Characteristic classes owe their existence to invariant polynomials on

the Lie algebra of the structure group like characteristic coeﬃcients of

matrices. The Lie algebra of Diﬀ(S) for compact S is the Lie algebra

X(S) of vector ﬁelds on S. Unfortunately it does not admit any invari-

ants. But there are equivariants like X →L

X

ω for some closed form ω

on S which can be used to give a sort of Chern-Weil construction of char-

acteristic classes in the cohomology of the base M with local coeﬃcients

(see sections 16 and 17).

Finally we discuss some self duality and anti self duality conditions

which depend on some ﬁberwise structure on the bundle like a ﬁberwise

symplectic structure.

This booklet starts with a short description of analysis in inﬁnite

dimensions along the lines of Fr¨olicher and Kriegl which makes inﬁnite

dimensional diﬀerential geometry much simpler than it used to be. We

treat smoothness, real analyticity and holomorphy. This part is based on

[Kriegl-Michor, 1990b] and the forthcoming book [Kriegl-Michor]. Then

we give a more detailed exposition of the theory of manifolds of mappings

and the diﬀeomorphism group. This part is adapted from [Michor, 1980]

4 Introduction

to the use of the Fr¨olicher-Kriegl calculus. Then we present a careful

introduction to the Fr¨olicher-Nijenhuis bracket, to ﬁber bundles and

connections, G-structures and principal connection which emphasizes

the construction and recognition of induced connections.

The material in the rest of the book, from section 12 onwards, has

been published in [Michor, 1988]. The version here is much more detailed

and contains more results.

The material in this booklet is a much extended version of a series

of lectures held at the Institute of Physics of the University of Napoli,

March 28 – April 1, 1988. I want to thank G. Marmo for his hospitality,

his interest in this subject, and for the suggestion to publish this booklet.

Wien, August 6, 1990 P. Michor

5

1. Notations and conventions

1.1. Deﬁnition. A Lie group G is a smooth manifold and a group

such that the multiplication µ : G G → G is smooth. Then also the

inversion ν : G →G turns out to be smooth.

We shall use the following notation:

µ : GG →G, multiplication, µ(x, y) = x.y.

λ

a

: G →G, left translation, λ

a

(x) = a.x.

ρ

a

: G →G, right translation, ρ

a

(x) = x.a.

ν : G →G, inversion, ν(x) = x

−1

.

e ∈ G, the unit element.

If g = T

e

G is the Lie algebra of G, we use the following notation:

Ad

G

: G →Aut

Lie

(g) and so on.

1.2. Let : G M → M be a left action, so

ˇ

: G → Diﬀ(M) is a

group homomorphism. Then we have partial mappings

a

: M → M

and

x

: G →M, given by

a

(x) =

x

(a) = (a, x) = a.x.

For any X ∈ g we deﬁne the fundamental vector ﬁeld ζ

X

= ζ

M

X

∈

X(M) by ζ

X

(x) = T

e

(

x

).X = T

(e,x)

.(X, 0

x

).

Lemma. In this situation the following assertions hold:

(1) ζ : g →X(M) is a linear mapping.

(2) T

x

(

a

).ζ

X

(x) = ζ

Ad(a)X

(a.x).

(3) R

X

0

M

∈ X(GM) is -related to ζ

X

∈ X(M).

(4) [ζ

X

, ζ

Y

] = −ζ

[X,Y ]

.

1.3. Let r : M G → M be a right action, so ˇ r : G → Diff(M)

is a group anti homomorphism. We will use the following notation:

r

a

: M →M and r

x

: G →M, given by r

x

(a) = r

a

(x) = r(x, a) = x.a.

For any X ∈ g we deﬁne the fundamental vector ﬁeld ζ

X

= ζ

M

X

∈

X(M) by ζ

X

(x) = T

e

(r

x

).X = T

(x,e)

r.(0

x

, X).

Lemma. In this situation the following assertions hold:

(1) ζ : g →X(M) is a linear mapping.

(2) T

x

(r

a

).ζ

X

(x) = ζ

Ad(a

−1

)X

(x.a).

(3) 0

M

L

X

∈ X(M G) is r-related to ζ

X

∈ X(M).

(4) [ζ

X

, ζ

Y

] = ζ

[X,Y ]

.

6

2. Calculus of smooth mappings

2.1. The traditional diﬀerential calculus works well for ﬁnite dimen-

sional vector spaces and for Banach spaces. For more general locally

convex spaces a whole ﬂock of diﬀerent theories were developed, each of

them rather complicated and none really convincing. The main diﬃculty

is that the composition of linear mappings stops to be jointly continuous

at the level of Banach spaces, for any compatible topology. This was

the original motivation for the development of a whole new ﬁeld within

general topology, convergence spaces.

Then in 1982, Alfred Fr¨olicher and Andreas Kriegl presented inde-

pendently the solution to the question for the right diﬀerential calculus

in inﬁnite dimensions. They joined forces in the further development

of the theory and the (up to now) ﬁnal outcome is the book [Fr¨olicher-

Kriegl, 1988], which is the general reference for this section. See also the

forthcoming book [Kriegl-Michor].

In this section I will sketch the basic deﬁnitions and the most impor-

tant results of the Fr¨olicher-Kriegl calculus.

2.2. The c

∞

-topology. Let E be a locally convex vector space. A

curve c : R → E is called smooth or C

∞

if all derivatives exist and

are continuous - this is a concept without problems. Let C

∞

(R, E) be

the space of smooth functions. It can be shown that C

∞

(R, E) does

not depend on the locally convex topology of E, only on its associated

bornology (system of bounded sets).

The ﬁnal topologies with respect to the following sets of mappings

into E coincide:

(1) C

∞

(R, E).

(2) Lipschitz curves (so that ¦

c(t)−c(s)

t−s

: t ,= s¦ is bounded in E).

(3) ¦E

B

→ E : B bounded absolutely convex in E¦, where E

B

is

the linear span of B equipped with the Minkowski functional

p

B

(x) := inf¦λ > 0 : x ∈ λB¦.

(4) Mackey-convergent sequences x

n

→ x (there exists a sequence

0 < λ

n

¸∞ with λ

n

(x

n

−x) bounded).

This topology is called the c

∞

-topology on E and we write c

∞

E for the

resulting topological space. In general (on the space T of test functions

for example) it is ﬁner than the given locally convex topology, it is not

a vector space topology, since scalar multiplication is no longer jointly

continuous. The ﬁnest among all locally convex topologies on E which

are coarser than c

∞

E is the bornologiﬁcation of the given locally convex

topology. If E is a Fr´echet space, then c

∞

E = E.

2. Calculus of smooth mappings 7

2.3. Convenient vector spaces. Let E be a locally convex vector

space. E is said to be a convenient vector space if one of the following

equivalent (completeness) conditions is satisﬁed:

(1) Any Mackey-Cauchy-sequence (so that (x

n

−x

m

) is Mackey con-

vergent to 0) converges. This is also called c

∞

-complete.

(2) If B is bounded closed absolutely convex, then E

B

is a Banach

space.

(3) Any Lipschitz curve in E is locally Riemann integrable.

(4) For any c

1

∈ C

∞

(R, E) there is c

2

∈ C

∞

(R, E) with c

1

= c

2

(existence of antiderivative).

2.4. Lemma. Let E be a locally convex space. Then the following

properties are equivalent:

(1) E is c

∞

-complete.

(2) If f : R

k

→E is scalarwise Lip

k

, then f is Lip

k

, for k > 1.

(3) If f : R →E is scalarwise C

∞

then f is diﬀerentiable at 0.

(4) If f : R →E is scalarwise C

∞

then f is C

∞

.

Here a mapping f : R

k

→ E is called Lip

k

if all partial derivatives

up to order k exist and are Lipschitz, locally on R

n

. f scalarwise C

∞

means that λ ◦ f is C

∞

for all continuous linear functionals on E.

This lemma says that a convenient vector space one can recognize

smooth curves by investigating compositions with continuous linear func-

tionals.

2.5. Smooth mappings. Let E and F be locally convex vector spaces.

A mapping f : E → F is called smooth or C

∞

, if f ◦ c ∈ C

∞

(R, F) for

all c ∈ C

∞

(R, E); so f

∗

: C

∞

(R, E) → C

∞

(R, F) makes sense. Let

C

∞

(E, F) denote the space of all smooth mapping from E to F.

For E and F ﬁnite dimensional this gives the usual notion of smooth

mappings: this has been ﬁrst proved in [Boman, 1967]. Constant map-

pings are smooth. Multilinear mappings are smooth if and only if they

are bounded. Therefore we denote by L(E, F) the space of all bounded

linear mappings from E to F.

2.6. Structure on C

∞

(E, F). We equip the space C

∞

(R, E) with the

bornologiﬁcation of the topology of uniform convergence on compact

sets, in all derivatives separately. Then we equip the space C

∞

(E, F)

with the bornologiﬁcation of the initial topology with respect to all map-

pings c

∗

: C

∞

(E, F) →C

∞

(R, F), c

∗

(f) := f ◦ c, for all c ∈ C

∞

(R, E).

2.7. Lemma. For locally convex spaces E and F we have:

(1) If F is convenient, then also C

∞

(E, F) is convenient, for any

8 2. Calculus of smooth mappings

E. The space L(E, F) is a closed linear subspace of C

∞

(E, F),

so it also convenient.

(2) If E is convenient, then a curve c : R → L(E, F) is smooth if

and only if t →c(t)(x) is a smooth curve in F for all x ∈ E.

2.8. Theorem. Cartesian closedness. The category of convenient

vector spaces and smooth mappings is cartesian closed. So we have a

natural bijection

C

∞

(E F, G)

∼

= C

∞

(E, C

∞

(F, G)),

which is even a diﬀeomorphism.

Of coarse this statement is also true for c

∞

-open subsets of convenient

vector spaces.

2.9. Corollary. Let all spaces be convenient vector spaces. Then the

following canonical mappings are smooth.

ev : C

∞

(E, F) E →F, ev(f, x) = f(x)

ins : E →C

∞

(F, E F), ins(x)(y) = (x, y)

( )

∧

: C

∞

(E, C

∞

(F, G)) →C

∞

(E F, G)

( )

∨

: C

∞

(E F, G) →C

∞

(E, C

∞

(F, G))

comp : C

∞

(F, G) C

∞

(E, F) →C

∞

(E, G)

C

∞

( , ) : C

∞

(F, F

) C

∞

(E

, E) →C

∞

(C

∞

(E, F), C

∞

(E

, F

))

(f, g) →(h →f ◦ h ◦ g)

:

C

∞

(E

i

, F

i

) →C

∞

(

E

i

,

F

i

)

2.10. Theorem. Let E and F be convenient vector spaces. Then the

diﬀerential operator

d : C

∞

(E, F) →C

∞

(E, L(E, F)),

df(x)v := lim

t→0

f(x +tv) −f(x)

t

,

exists and is linear and bounded (smooth). Also the chain rule holds:

d(f ◦ g)(x)v = df(g(x))dg(x)v.

2. Calculus of smooth mappings 9

2.11. Remarks. Note that the conclusion of theorem 2.8 is the starting

point of the classical calculus of variations, where a smooth curve in a

space of functions was assumed to be just a smooth function in one

variable more.

If one wants theorem 2.8 to be true and assumes some other obvious

properties, then the calculus of smooth functions is already uniquely

determined.

There are, however, smooth mappings which are not continuous. This

is unavoidable and not so horrible as it might appear at ﬁrst sight. For

example the evaluation E E

**→ R is jointly continuous if and only if
**

E is normable, but it is always smooth. Clearly smooth mappings are

continuous for the c

∞

-topology.

For Fr´echet spaces smoothness in the sense described here coincides

with the notion C

∞

c

of [Keller, 1974]. This is the diﬀerential calculus

used by [Michor, 1980], [Milnor, 1984], and [Pressley-Segal, 1986].

A prevalent opinion in contemporary mathematics is, that for inﬁnite

dimensional calculus each serious application needs its own foundation.

By a serious application one obviously means some application of a hard

inverse function theorem. These theorems can be proved, if by assuming

enough a priori estimates one creates enough Banach space situation for

some modiﬁed iteration procedure to converge. Many authors try to

build their platonic idea of an a priori estimate into their diﬀerential

calculus. I think that this makes the calculus inapplicable and hides the

origin of the a priori estimates. I believe, that the calculus itself should

be as easy to use as possible, and that all further assumptions (which

most often come from ellipticity of some nonlinear partial diﬀerential

equation of geometric origin) should be treated separately, in a setting

depending on the speciﬁc problem. I am sure that in this sense the

Fr¨olicher-Kriegl calculus as presented here and its holomorphic and real

analytic oﬀsprings in sections 2 and 3 below are universally usable for

most applications.

10

3. Calculus of holomorphic mappings

3.1. Along the lines of thought of the Fr¨olicher-Kriegl calculus of smooth

mappings, in [Kriegl-Nel, 1985] the cartesian closed setting for holomor-

phic mappings was developed. The right deﬁnition of this calculus was

already given by [Fantappi´e, 1930 and 1933]. I will now sketch the ba-

sics and the main results. It can be shown that again convenient vector

spaces are the right ones to consider. Here we will start with them for

the sake of shortness.

3.2. Let E be a complex locally convex vector space whose underlying

real space is convenient – this will be called convenient in the sequel.

Let D ⊂ C be the open unit disk and let us denote by C

ω

(D, E) the

space of all mappings c : D →E such that λ ◦ c : D →C is holomorphic

for each continuous complex-linear functional λ on E. Its elements will

be called the holomorphic curves.

If E and F are convenient complex vector spaces (or c

∞

-open sets

therein), a mapping f : E → F is called holomorphic if f ◦ c is a holo-

morphic curve in F for each holomorphic curve c in E. Obviously f is

holomorphic if and only if λ◦f : E →C is holomorphic for each complex

linear continuous functional λ on F. Let C

ω

(E, F) denote the space of

all holomorphic mappings from E to F.

3.3. Theorem of Hartogs. Let E

k

for k = 1, 2 and F be complex

convenient vector spaces and let U

k

⊂ E

k

be c

∞

-open. A mapping f :

U

1

U

2

→ F is holomorphic if and only if it is separably holomorphic

(i. e. f( , y) and f(x, ) are holomorphic for all x ∈ U

1

and y ∈ U

2

).

This implies also that in ﬁnite dimensions we have recovered the usual

deﬁnition.

3.4 Lemma. If f : E ⊃ U → F is holomorphic then df : U E → F

exists, is holomorphic and C-linear in the second variable.

A multilinear mapping is holomorphic if and only if it is bounded.

3.5 Lemma. If E and F are Banach spaces and U is open in E, then

for a mapping f : U →F the following conditions are equivalent:

(1) f is holomorphic.

(2) f is locally a convergent series of homogeneous continuous poly-

nomials.

(3) f is C-diﬀerentiable in the sense of Fr´echet.

3. Calculus of holomorphic mappings 11

3.6 Lemma. Let E and F be convenient vector spaces. A mapping

f : E → F is holomorphic if and only if it is smooth and its derivative

is everywhere C-linear.

An immediate consequence of this result is that C

ω

(E, F) is a closed

linear subspace of C

∞

(E

R

, F

R

) and so it is a convenient vector space if F

is one, by 2.7. The chain rule follows from 2.10. The following theorem

is an easy consequence of 2.8.

3.7 Theorem. Cartesian closedness. The category of convenient

complex vector spaces and holomorphic mappings between them is carte-

sian closed, i. e.

C

ω

(E F, G)

∼

= C

ω

(E, C

ω

(F, G)).

An immediate consequence of this is again that all canonical struc-

tural mappings as in 2.9 are holomorphic.

12

4. Calculus of real analytic mappings

4.1. In this section I sketch the cartesian closed setting to real ana-

lytic mappings in inﬁnite dimension following the lines of the Fr¨olicher-

Kriegl calculus, as it is presented in [Kriegl-Michor, 1990a]. Surprisingly

enough one has to deviate from the most obvious notion of real analytic

curves in order to get a meaningful theory, but again convenient vector

spaces turn out to be the right kind of spaces.

4.2. Real analytic curves. Let E be a real convenient vector space

with dual E

**. A curve c : R →E is called real analytic if λ ◦ c : R →R
**

is real analytic for each λ ∈ E

**. It turns out that the set of these curves
**

depends only on the bornology of E.

In contrast a curve is called topologically real analytic if it is locally

given by power series which converge in the topology of E. They can be

extended to germs of holomorphic curves along R in the complexiﬁcation

E

C

of E. If the dual E

**of E admits a Baire topology which is compatible
**

with the duality, then each real analytic curve in E is in fact topological

real analytic for the bornological topology on E.

4.3. Real analytic mappings. Let E and F be convenient vector

spaces. Let U be a c

∞

-open set in E. A mapping f : U → F is called

real analytic if and only if it is smooth (maps smooth curves to smooth

curves) and maps real analytic curves to real analytic curves.

Let C

ω

(U, F) denote the space of all real analytic mappings. We

equip the space C

ω

(U, R) of all real analytic functions with the initial

topology with respect to the families of mappings

C

ω

(U, R)

c

∗

−→C

ω

(R, R), for all c ∈ C

ω

(R, U)

C

ω

(U, R)

c

∗

−→C

∞

(R, R), for all c ∈ C

∞

(R, U),

where C

∞

(R, R) carries the topology of compact convergence in each de-

rivative separately as in section 2, and where C

ω

(R, R) is equipped with

the ﬁnal topology with respect to the embeddings (restriction mappings)

of all spaces of holomorphic mappings from a neighborhood V of R in

C mapping R to R, and each of these spaces carries the topology of

compact convergence.

Furthermore we equip the space C

ω

(U, F) with the initial topology

with respect to the family of mappings

C

ω

(U, F)

λ

∗

−→C

ω

(U, R), for all λ ∈ F

.

It turns out that this is again a convenient space.

4. Calculus of real analytic mappings 13

4.4. Theorem. In the setting of 4.3 a mapping f : U → F is real

analytic if and only if it is smooth and is real analytic along each aﬃne

line in E.

4.5. Theorem Cartesian closedness. The category of convenient

spaces and real analytic mappings is cartesian closed. So the equation

C

ω

(U, C

ω

(V ), F)

∼

= C

ω

(U V, F)

is valid for all c

∞

-open sets U in E and V in F, where E, F, and G

are convenient vector spaces.

This implies again that all structure mappings as in 2.9 are real ana-

lytic. Furthermore the diﬀerential operator

d : C

ω

(U, F) →C

ω

(U, L(E, F))

exists, is unique and real analytic. Multilinear mappings are real an-

alytic if and only if they are bounded. Powerful real analytic uniform

boundedness principles are available.

14

5. Inﬁnite dimensional manifolds

5.1. In this section I will concentrate on two topics: Smooth partitions

of unity, and several kinds of tangent vectors. The material is taken

from [Kriegl-Michor, 1990b].

5.2. In the usual way we deﬁne manifolds by gluing c

∞

-open sets in

convenient vector spaces via smooth (holomorphic, real analytic) diﬀeo-

morphisms. Then we equip them with the identiﬁcation topology with

respect to the c

∞

-topologies on all modeling spaces. We require some

properties from this topology, like Hausdorﬀ and regular (which here is

not a consequence of Hausdorﬀ).

Mappings between manifolds are smooth (holomorphic, real analytic),

if they have this property when composed which any chart mappings.

5.3. Lemma. A manifold M is metrizable if and only if it is paracom-

pact and modeled on Fr´echet spaces.

5.4. Lemma. For a convenient vector space E the set C

∞

(M, E) of

smooth E-valued functions on a manifold M is again a convenient vector

space. Likewise for the real analytic and holomorphic case.

5.5. Theorem. Smooth partitions of unity. If M is a smooth

manifold modeled on convenient vector spaces admitting smooth bump

functions and | is a locally ﬁnite open cover of M, then there exists a

smooth partition of unity ¦ϕ

U

: U ∈ |¦ with carr/supp(ϕ

U

) ⊂ U for all

U ∈ |.

If M is in addition paracompact, then this is true for every open cover

| of M.

Convenient vector spaces which are nuclear admit smooth bump func-

tions.

5.6. The tangent spaces of a convenient vector space E. Let

a ∈ E. A kinematic tangent vector with foot point a is simply a pair

(a, X) with X ∈ E. Let T

a

E = E be the space of all kinematic tangent

vectors with foot point a. It consists of all derivatives c

(0) at 0 of

smooth curves c : R → E with c(0) = a, which explains the choice of

the name kinematic.

For each open neighborhood U of a in E (a, X) induces a linear map-

ping X

a

: C

∞

(U, R) → R by X

a

(f) := df(a)(X

a

), which is continu-

ous for the convenient vector space topology on C

∞

(U, R), and satisﬁes

X

a

(f g) = X

a

(f) g(a) + f(a) X

a

(g), so it is a continuous derivation

over ev

a

. The value X

a

(f) depends only on the germ of f at a.

5. Inﬁnite dimensional manifolds 15

An operational tangent vector of E with foot point a is a bounded

derivation ∂ : C

∞

a

(E, R) → R over ev

a

. Let D

a

E be the vector space

of all these derivations. Any ∂ ∈ D

a

E induces a bounded derivation

C

∞

(U, R) →R over ev

a

for each open neighborhood U of a in E. So the

vector space D

a

E is a closed linear subspace of the convenient vector

space

U

L(C

∞

(U, R), R). We equip D

a

E with the induced convenient

vector space structure. Note that the spaces D

a

E are isomorphic for all

a ∈ E.

Example. Let Y ∈ E

**be an element in the bidual of E. Then for
**

each a ∈ E we have an operational tangent vector Y

a

∈ D

a

E, given by

Y

a

(f) := Y (df(a)). So we have a canonical injection E

→D

a

E.

Example. Let : L

2

(E; R) →R be a bounded linear functional which

vanishes on the subset E

⊗ E

**. Then for each a ∈ E we have an
**

operational tangent vector ∂

2

[

a

∈ D

a

E given by ∂

2

[

a

(f) := (d

2

f(a)),

since

(d

2

(fg)(a)) =

= (d

2

f(a)g(a) +df(a) ⊗dg(a) +dg(a) ⊗df(a) +f(a)d

2

g(a))

= (d

2

f(a))g(a) + 0 +f(a)(d

2

g(a)).

5.7. Lemma. Let ∈ L

k

sym

(E; R)

**be a bounded linear functional which
**

vanishes on the subspace

k−1

i=1

L

i

sym

(E; R) ∨ L

k−i

sym

(E; R)

of decomposable elements of L

k

sym

(E; R). Then deﬁnes an operational

tangent vector ∂

k

[

a

∈ D

a

E for each a ∈ E by

∂

k

[

a

(f) := (d

k

f(a)).

The linear mapping → ∂

k

[

a

is an embedding onto a topological direct

summand D

(k)

a

E of D

a

E. Its left inverse is given by ∂ → (Φ → ∂((Φ ◦

diag)(a+ ))). The sum

k>0

D

(k)

a

E in D

a

E is a direct one.

5.8. Lemma. If E is an inﬁnite dimensional Hilbert space, all opera-

tional tangent space summands D

(k)

0

E are not zero.

5.9. Deﬁnition. A convenient vector space is said to have the (borno-

logical) approximation property if E

⊗ E is dense in L(E, E) in the

bornological locally convex topology.

The following spaces have the bornological approximation property:

R

(N)

, nuclear Fr´echet spaces, nuclear (LF)spaces.

16 5. Inﬁnite dimensional manifolds

5.10 Theorem. Tangent vectors as derivations. Let E be a conve-

nient vector space which has the approximation property. Then we have

D

a

E = D

(1)

a

E

∼

= E

**. So if E is in addition reﬂexive, each operational
**

tangent vector is a kinematic one.

5.11. The kinematic tangent bundle TM of a manifold M is constructed

by gluing all the kinematic tangent bundles of charts with the help of the

kinematic tangent mappings (derivatives) of the chart changes. TM →

M is a vector bundle and T : C

∞

(M, N) →C

∞

(TM, TN) is well deﬁned

and has the usual properties.

5.12. The operational tangent bundle DM of a manifold M is con-

structed by gluing all operational tangent spaces of charts. Then π

M

:

DM →M is again a vector bundle which contains the kinematic tangent

bundle TM embeds as a splitting subbundle. Also for each k ∈ N the

same gluing construction as above gives us tangent bundles D

(k)

M which

are splitting subbundles of DM. The mappings D

(k)

: C

∞

(M, N) →

C

∞

(D

(k)

M, D

(k)

N) are well deﬁned for all k including ∞ and have the

usual properties.

Note that for manifolds modeled on reﬂexive spaces having the borno-

logical approximation property the operational and the kinematic tan-

gent bundles coincide.

17

6. Manifolds of mappings

6.1. Theorem. Manifold structure of C

∞

(M, N). Let M and N be

smooth ﬁnite dimensional manifolds. Then the space C

∞

(M, N) of all

smooth mappings from M to N is a smooth manifold, modeled on spaces

C

∞

(f

∗

TN) of smooth sections of pullback bundles along f : M → N

over M.

A careful description of this theorem (but without the help of the

Fr¨olicher-Kriegl calculus) can be found in [Michor, 1980]. I include a

proof of this result here because the result is important and the proof is

much simpler now with the help of the Fr¨olicher Kriegl-calculus.

Proof. Choose a smooth Riemannian metric on N. Let exp : TN ⊇

U →N be the smooth exponential mapping of this Riemannian metric,

deﬁned on a suitable open neighborhood of the zero section. We may

assume that U is chosen in such a way that (π

N

, exp) : U →N N is a

smooth diﬀeomorphism onto an open neighborhood V of the diagonal.

For f ∈ C

∞

(M, N) we consider the pullback vector bundle

M

N

TN f

∗

TN

π

∗

N

f

−−−−→ TN

f

∗

π

N

¸

¸

_

¸

¸

_

π

N

M

f

−−−−→ N.

Then C

∞

(f

∗

TN) is canonically isomorphic to the space C

∞

f

(M, TN) :=

¦h ∈ C

∞

(M, TN) : π

N

◦ h = f¦ via s → (π

∗

N

f) ◦ s and (Id

M

, h) ← h.

We consider the space C

∞

c

(f

∗

TN) of all smooth sections with compact

support and equip it with the inductive limit topology

C

∞

c

(f

∗

TN) = inj lim

K

C

∞

K

(f

∗

TN),

where K runs through all compact sets in M and each of the spaces

C

∞

K

(f

∗

TN) is equipped with the topology of uniform convergence (on

K) in all derivatives separately. For f, g ∈ C

∞

(M, N) we write f ∼ g

if f and g agree oﬀ some compact subset in M. Now let

U

f

:= ¦g ∈ C

∞

(M, N) : (f(x), g(x)) ∈ V for all x ∈ M, g ∼ f¦,

u

f

: U

f

→C

∞

c

(f

∗

TN),

u

f

(g)(x) = (x, exp

−1

f(x)

(g(x))) = (x, ((π

N

, exp)

−1

◦ (f, g))(x)).

18 6. Manifolds of mappings

Then u

f

is a bijective mapping from U

f

onto the set ¦s ∈ C

∞

c

(f

∗

TN) :

s(M) ⊆ f

∗

U¦, whose inverse is given by u

−1

f

(s) = exp ◦(π

∗

N

f) ◦s, where

we view U →N as ﬁber bundle. The set u

f

(U

f

) is open in C

∞

c

(f

∗

TN)

for the topology described above.

Now we consider the atlas (U

f

, u

f

)

f∈C

∞

(M,N)

for C

∞

(M, N). Its

chart change mappings are given for s ∈ u

g

(U

f

∩ U

g

) ⊆ C

∞

c

(g

∗

TN) by

(u

f

◦ u

−1

g

)(s) = (Id

M

, (π

N

, exp)

−1

◦ (f, exp ◦(π

∗

N

g) ◦ s))

= (τ

−1

f

◦ τ

g

)

∗

(s),

where τ

g

(x, Y

g(x)

) := (x, exp

g(x)

(Y

g(x)

))) is a smooth diﬀeomorphism

τ

g

: g

∗

TN ⊇ g

∗

U →(g Id

N

)

−1

(V ) ⊆ MN which is ﬁber respecting

over M.

Smooth curves in C

∞

c

(f

∗

TN) are just smooth sections of the bundle

pr

∗

2

f

∗

TN → R M, which have compact support in M locally in R.

The chart change u

f

◦ u

−1

g

= (τ

−1

f

◦ τ

g

)

∗

is deﬁned on an open subset

and obviously maps smooth curves to smooth curves, therefore it is also

smooth.

Finally we put the identiﬁcation topology from this atlas on the space

C

∞

(M, N), which is obviously ﬁner than the compact open topology and

thus Hausdorﬀ.

The equation u

f

◦ u

−1

g

= (τ

−1

f

◦ τ

g

)

∗

shows that the smooth structure

does not depend on the choice of the smooth Riemannian metric on

N.

6.2. Lemma. Smooth curves in C

∞

(M, N). Let M and N be

smooth ﬁnite dimensional manifolds.

Then the smooth curves c in C

∞

(M, N) correspond exactly to the

smooth mappings ˆ c ∈ C

∞

(RM, N) which satisfy the following property:

(1) For each compact interval [a, b] ⊂ R there is a compact subset

K ⊂ M such that the restriction ˆ c[([a, b] (M¸ K)) is constant.

Proof. Since R is locally compact property (1) is equivalent to

(2) For each t ∈ R there is an open neighborhood U of t in R and

a compact K ⊂ M such that the restriction ˆ c[(U (M ¸ K)) is

constant.

Since this is a local condition on R and since smooth curves in C

∞

(M, N)

locally take values in charts as in the proof of theorem 6.1 it suﬃces to

describe the smooth curves in the space C

∞

c

(E) of sections with compact

support of a vector bundle (E, p, M, V ). It is equipped with the inductive

limit topology

C

∞

c

(E) = inj lim

K

C

∞

K

(E),

6. Manifolds of mappings 19

where K runs through all compact subsets of M, and where C

∞

K

(E)

denotes the space of sections of E with support in K, equipped with the

compact C

∞

-topology (the topology of uniform convergence on compact

subsets, in all derivatives separately). Since this injective limit is strict

(each bounded subset is contained and bounded in some step) smooth

curves in C

∞

c

(E) locally factor to smooth maps in C

∞

K

(E) for some K.

Let (U

α

, ψ

α

: E →U

α

V ) be a vector bundle atlas for E. Then

C

∞

K

(E) ÷ s →(pr

2

◦ ψ

α

◦ s[U

α

)

α

∈

α

C

∞

(U

α

, V )

is a linear embedding onto a closed linear subspace of the product, where

the factor spaces on the right hand side carry the topology used in 2.6.

So a curve in C

∞

K

(E) is smooth if and only if each its ”coordinates” in

C

∞

(U

α

, V ) is smooth. By cartesian closedness 2.8 this is the case if and

only if their canonically associates are smooth R U

α

→ V . these ﬁt

together to a smooth section of the pullback bundle pr

∗

2

E →RM with

support in R K.

6.3. Theorem. C

ω

-manifold structure of C

ω

(M, N). Let M and N

be real analytic manifolds, let M be compact. Then the space C

ω

(M, N)

of all real analytic mappings from M to N is a real analytic manifold,

modeled on spaces C

ω

(f

∗

TN) of real analytic sections of pullback bun-

dles along f : M →N over M.

The proof is a variant of the proof of 6.1, using a real analytic Rie-

mannian metric. It can also be found in [Kriegl-Michor, 1990a].

6.4. Theorem. C

ω

-manifold structure on C

∞

(M, N). Let M and

N be real analytic manifolds, with M compact. Then the smooth mani-

fold C

∞

(M, N) with the structure from 6.1 is even a real analytic man-

ifold.

Proof. For a ﬁxed real analytic exponential mapping on N the charts

(U

f

, u

f

) from 6.1 for f ∈ C

ω

(M, N) form a smooth atlas for C

∞

(M, N),

since C

ω

(M, N) is dense in C

∞

(M, N) by [Grauert, 1958, Proposition

8]. The chart changings u

f

◦ u

−1

g

= (τ

−1

f

◦ τ

g

)

∗

are real analytic: this

follows from a careful description of the set of real analytic curves into

C

∞

(f

∗

TN). See again [Kriegl-Michor, 1990a] for more details.

6.5 Remark. If M is not compact, C

ω

(M, N) is dense in C

∞

(M, N)

for the Whitney-C

∞

-topology by [Grauert, 1958, Prop. 8]. This is not

the case for the T-topology from [Michor, 1980], in which C

∞

(M, N)

is a smooth manifold. The charts U

f

for f ∈ C

∞

(M, N) do not cover

C

∞

(M, N).

20 6. Manifolds of mappings

6.6. Theorem. Smoothness of composition. If M, N are ﬁ-

nite dimensional smooth manifolds, then the evaluation mapping ev :

C

∞

(M, N) M →N is smooth.

If P is another smooth manifold, then the composition mapping

comp : C

∞

(M, N) C

∞

prop

(P, M) →C

∞

(P, N)

is smooth, where C

∞

prop

(P, M) denote the space of all proper smooth map-

pings P →M (i.e. compact sets have compact inverse images). This is

open in C

∞

(P, M).

In particular f

∗

: C

∞

(M, N) → C

∞

(M, N

) and g

∗

: C

∞

(M, N) →

C

∞

(P, N) are smooth for f ∈ C

∞

(N, N

) and g ∈ C

∞

prop

(P, M).

The corresponding statement for real analytic mappings is also true,

but P and M have to be compact.

Proof. Using the description of smooth curves in C

∞

(M, N) given in 6.2

we immediately see that (ev ◦(c

1

, c

2

)) = ˆ c

1

(t, c

1

(t)) is smooth for each

smooth (c

1

, c

2

) : R →C

∞

(M, N) M, so ev is smooth as claimed.

The space of proper mappings C

∞

prop

(P, M) is open in the manifold

C

∞

(P, M) since changing a mapping only on a compact set does not

change its property of being proper. Let (c

1

, c

2

) : R → C

∞

(M, N)

C

∞

prop

(P, M) be a smooth curve. Then we have (comp ◦(c

1

, c

2

))(t)(p) =

ˆ c

1

(t, ˆ c

2

(t, p)) and one may check that this has again property (1), so it

is a smooth curve in C

∞

(P, N). Thus comp is smooth.

The proof for real analytic manifolds is similar.

6.7. Theorem. Let M and N be smooth manifolds. Then the inﬁnite

dimensional smooth vector bundles TC

∞

(M, N) and C

∞

c

(M, TN) ⊂

C

∞

(M, TN) over C

∞

(M, N) are canonically isomorphic. The same

assertion is true for C

ω

(M, N), if M is compact.

Proof. Here by C

∞

c

(M, TN) we denote the space of all smooth mappings

f : M → TN such that f(x) = 0

π

M

f(x)

for x / ∈ K

f

, a suitable compact

subset of M (equivalently, such that the associated section of the pull

back bundle (π

M

◦ f)

∗

TN →M has compact support).

One can check directly that the atlas from 6.1 for C

∞

(M, N) induces

an atlas for TC

∞

(M, N) which is equivalent to that for C

∞

(M, TN) via

some natural identiﬁcations in TTN. This is carried out in great detail

in [Michor, 1980, 10.19].

6.8. Theorem. Exponential law. Let / be a (possibly inﬁnite

dimensional) smooth manifold, and let M and N be ﬁnite dimensional

smooth manifolds.

6. Manifolds of mappings 21

Then we have a canonical embedding

C

∞

(/, C

∞

(M, N)) ⊆ C

∞

(/M →N),

where the image in the right hand side consists of all smooth mappings

f : /M →N which satisfy the following property

(1) For each point in / there is an open neighborhood | and a

compact set K ⊂ M such that the restriction f[(| (M¸ K)) is

constant.

We have equality if and only if M is compact.

If M and N are real analytic manifolds with M compact we have

C

ω

(/, C

ω

(M, N)) = C

ω

(/M, N)

for each real analytic (possibly inﬁnite dimensional) manifold.

Proof. This follows directly from the description of all smooth curves in

C

∞

(M, N) given in the proof of 6.6.

22

7. Diﬀeomorphism groups

7.1. Theorem. Diﬀeomorphism group. For a smooth manifold

M the group Diﬀ(M) of all smooth diﬀeomorphisms of M is an open

submanifold of C

∞

(M, M), composition and inversion are smooth.

The Lie algebra of the smooth inﬁnite dimensional Lie group Diﬀ(M)

is the convenient vector space C

∞

c

(TM) of all smooth vector ﬁelds on

M with compact support, equipped with the negative of the usual Lie

bracket. The exponential mapping Exp : C

∞

c

(TM) → Diﬀ

∞

(M) is the

ﬂow mapping to time 1, and it is smooth.

For a compact real analytic manifold M the group Diﬀ

ω

(M) of all real

analytic diﬀeomorphisms is a real analytic Lie group with Lie algebra

C

ω

(TM) and with real analytic exponential mapping.

Proof. It is well known that the space Diﬀ(M) of all diﬀeomorphisms

of M is open in C

∞

(M, M) for the Whitney C

∞

-topology. Since the

topology we use on C

∞

(M, M) is ﬁner, Diﬀ(M) is an open submani-

fold of C

∞

prop

(M, M). The composition is smooth by theorem 6.6. To

show that the inversion inv is smooth, we consider a smooth curve

c : R →Diﬀ(M) ⊂ C

∞

(M, M). Then the mapping ˆ c : RM →M sat-

isﬁes 6.6.(1) and (inv ◦c)´fulﬁlls the ﬁnite dimensional implicit equation

ˆ c(t, (inv ◦c)´(t, m)) = m for all t ∈ R and m ∈ M. By the ﬁnite dimen-

sional implicit function theorem (inv ◦c)´is smooth in (t, m). Property

6.6.(1) is obvious. So inv maps smooth curves to smooth curves and is

thus smooth.

This proof is by far simpler than the original one, see [Michor, 1980],

and shows the power of the Fr¨olicher-Kriegl calculus.

By the chart structure from 6.1, or directly from theorem 6.7we see

that the tangent space T

e

Diﬀ(M) equals the space C

∞

c

(TM) of all vec-

tor ﬁelds with compact support. Likewise T

f

Diﬀ(M) = C

∞

c

(f

∗

TM)

which we identify with the space of all vector ﬁelds with compact support

along the diﬀeomorphism f. Right translation ρ

f

is given by ρ

f

(g) =

f

∗

(g) = g ◦ f, thus T(ρ

f

).X = X ◦ f and for the ﬂow Fl

X

t

of the vector

ﬁeld with compact support X we have

d

dt

Fl

X

t

= X ◦ Fl

X

t

= T(ρ

Fl

X

t

).X.

So the one parameter group t →Fl

X

t

∈ Diﬀ(M) is the integral curve of

the right invariant vector ﬁeld R

X

: f →T(ρ

f

).X = X ◦ f on Diﬀ(M).

Thus the exponential mapping of the diﬀeomorphism group is given by

Exp = Fl

1

: X

c

(M) → Diﬀ(M). To show that is smooth we consider a

smooth curve in X

c

(M), i. e. a time dependent vector ﬁeld with com-

pact support X

t

. We may view it as a complete vector ﬁeld (0

t

, X

t

) on

R M whose smooth ﬂow respects the level surfaces ¦t¦ M and is

smooth. Thus Exp maps smooth curves to smooth curves and is smooth

7. Diﬀeomorphism groups 23

itself. Again one may compare this simple proof with the original one

in [Michor, 1983, section 4].

Let us ﬁnally compute the Lie bracket on X

c

(M) viewed as the Lie

algebra of Diﬀ(M).

Ad(Exp(sX))Y =

∂

∂t

¸

¸

0

Exp(sX) ◦ Exp(tY ) ◦ Exp(−sX)

= T(Fl

X

t

) ◦ Y ◦ Fl

X

−t

= (Fl

X

−t

)

∗

Y, thus

∂

∂t

¸

¸

0

Ad(Exp(tX))Y =

∂

∂t

¸

¸

0

(Fl

X

−t

)

∗

Y

= −[X, Y ],

the negative of the usual Lie bracket on X

c

(M). To understand this

contemplate 1.2.(4).

7.2. Remarks. The group Diﬀ(M) of smooth diﬀeomorphisms does

not carry any real analytic Lie group structure by [Milnor, 1984, 9.2],

and it has no complexiﬁcation in general, see [Pressley-Segal, 1986, 3.3].

The mapping

Ad ◦ Exp : C

∞

c

(TM) →Diﬀ(M) →L(C

∞

(TM), C

∞

(TM))

is not real analytic, see [Michor, 1983, 4.11].

For x ∈ M the mapping ev

x

◦ Exp : C

∞

c

(TM) →Diﬀ(M) →M is not

real analytic since (ev

x

◦ Exp)(tX) = Fl

X

t

(x), which is not real analytic

in t for general smooth X.

The exponential mapping Exp : C

∞

c

(TM) → Diﬀ(M) is in a very

strong sense not surjective: In [Grabowski, 1988] it is shown, that

Diﬀ(M) contains an arcwise connected free subgroup on 2

ℵ

0

generators

which meets the image of Exp only at the identity.

The real analytic Lie group Diﬀ

ω

(M) is regular in the sense of [Milnor,

1984. 7.6], who weakened the original concept of [Omori, 1982]. This

condition means that the mapping associating the evolution operator

to each time dependent vector ﬁeld on M is smooth. It is even real

analytic, compare the proof of theorem 7.2.

By theorem 6.6 the left action ev : Diﬀ(M) M → M is smooth.

If dimM ≥ 2 this action is n-transitive for each n ∈ N, i. e. for two

ordered ﬁnite n-tupels (x

1

, . . . , x

n

) and (y

1

, . . . , y

n

) of pair wise diﬀerent

elements of M there is a diﬀeomorphism f ∈ Diﬀ(M) with f(x

i

) = y

i

for each i. To see this connect x

i

with y

i

by a simple smooth curve

c

i

: [0, 1] →M which is an embedding, such that the images are pair wise

disjoint. Using tubular neighborhoods or submanifold charts one can

24 7. Diﬀeomorphism groups

show that then there is a vector ﬁeld with compact support X ∈ X

c

(M)

which restricts to c

i

(t) along c

i

(t). The diﬀeomorphism Exp(X) = Fl

X

1

then maps x

i

to y

i

.

In contrast to this one knows from [Omori, 1978] that a Banach Lie

group acting eﬀectively on a ﬁnite dimensional manifold is necessarily

ﬁnite dimensional. So there is no way to model the diﬀeomorphism

group on Banach spaces as a manifold. There is, however, the pos-

sibility to view Diﬀ(M) as an ILH-group (i. e. inverse limit of Hilbert

manifolds) which sometimes permits to use an implicit function theorem.

See [Omori, 1974] for this.

7.3. Theorem. Principal bundle of embeddings. Let M and N

be smooth manifolds.

Then the set Emb(M, N) of all smooth embeddings M →N is an open

submanifold of C

∞

(M, N). It is the total space of a smooth principal

ﬁber bundle with structure group Diﬀ(M), whose smooth base manifold

is the space B(M, N) of all submanifolds of N of type M.

The open subset Emb

prop

(M, N) of proper (equivalently closed) em-

beddings is saturated under the Diﬀ(M)-action and is thus the total

space of the restriction of the principal bundle to the open submanifold

B

closed

(M, N) of B(M, N) consisting of all closed submanifolds of N of

type M.

In the real analytic case this theorem remains true provided that M

is supposed to be compact, see [Kriegl-Michor, 1990a], section 6.

Proof. Let us ﬁx an embedding i ∈ Emb(M, N). Let g be a ﬁxed Rie-

mannian metric on N and let exp

N

be its exponential mapping. Then let

p : ^(i) → M be the normal bundle of i, deﬁned in the following way:

For x ∈ M let ^(i)

x

:= (T

x

i(T

x

M))

⊥

⊂ T

i(x)

N be the g-orthogonal

complement in T

i(x)

N. Then

^(i)

¯

i

−−−−→ TN

p

¸

¸

_

¸

¸

_

π

N

M −−−−→

i

N

is an injective vector bundle homomorphism over i.

Now let U

i

= U be an open neighborhood of the zero section of ^(i)

which is so small that (exp

N

◦

¯

i)[U : U →N is a diﬀeomorphism onto its

image which describes a tubular neighborhood of the submanifold i(M).

Let us consider the mapping

τ = τ

i

:= (exp

N

◦

¯

i)[U : ^(i) ⊃ U →N,

7. Diﬀeomorphism groups 25

a diﬀeomorphism onto its image, and the open set in Emb(M, N) which

will serve us as a saturated chart,

|(i) := ¦j ∈ Emb(M, N) : j(M) ⊆ τ

i

(U

i

), j ∼ i¦,

where j ∼ i means that j = i oﬀ some compact set in M. Then by

[Michor, 1980, section 4] the set |(i) is an open neighborhood of i in

Emb(M, N). For each j ∈ |(i) we deﬁne

ϕ

i

(j) : M →U

i

⊆ ^(i),

ϕ

i

(j)(x) := (τ

i

)

−1

(j(x))

Then ϕ

i

= ((τ

i

)

−1

)

∗

: |(i) →C

∞

(M, ^(i)) is a smooth mapping which

is bijective onto the open set

1(i) := ¦h ∈ C

∞

(M, ^(i)) : h(M) ⊆ U

i

, h ∼ 0¦

in C

∞

(M, ^(i)). Its inverse is given by the smooth mapping τ

i

∗

: h →

τ

i

◦ h.

We have τ

i

∗

(h ◦ f) = τ

i

∗

(h) ◦ f for those f ∈ Diﬀ(M) which are near

enough to the identity so that h ◦ f ∈ 1(i). We consider now the open

set

¦h ◦ f : h ∈ 1(i), f ∈ Diﬀ(M)¦ ⊆ C

∞

((M, U

i

)).

Obviously we have a smooth mapping from it into C

∞

c

(U

i

) Diﬀ(M)

given by h →(h◦(p◦h)

−1

, p◦h), where C

∞

c

(U

i

) is the space of sections

with compact support of U

i

→ M. So if we let Q(i) := τ

i

∗

(C

∞

c

(U

i

) ∩

1(i)) ⊂ Emb(M, N) we have

J(i) := |(i) ◦Diﬀ(M)

∼

= Q(i) Diﬀ(M)

∼

= (C

∞

c

(U

i

) ∩1(i)) Diﬀ(M),

since the action of Diﬀ(M) on i is free. Furthermore π[Q(i) : Q(i) →

Emb(M, N)/ Diﬀ(M) is bijective onto an open set in the quotient.

We now consider ϕ

i

◦ (π[Q(i))

−1

: π(Q(i)) → C

∞

(U

i

) as a chart

for the quotient space. In order to investigate the chart change let

j ∈ Emb(M, N) be such that π(Q(i)) ∩ π(Q(j)) ,= ∅. Then there is

an immersion h ∈ J(i) ∩ Q(j), so there exists a unique f

0

∈ Diﬀ(M)

(given by f

0

= p◦ϕ

i

(h)) such that h◦f

−1

0

∈ Q(i). If we consider j ◦f

−1

0

instead of j and call it again j, we have Q(i)∩Q(j) ,= ∅ and consequently

|(i) ∩ |(j) ,= ∅. Then the chart change is given as follows:

ϕ

i

◦ (π[Q(i))

−1

◦ π ◦ (τ

j

)

∗

: C

∞

c

(U

j

) →C

∞

c

(U

i

)

s →τ

j

◦ s →ϕ

i

(τ

j

◦ s) ◦ (p

i

◦ ϕ

i

(τ

j

◦ s))

−1

.

26 7. Diﬀeomorphism groups

This is of the form s → β ◦ s for a locally deﬁned diﬀeomorphism β :

^(j) →^(i) which is not ﬁber respecting, followed by h →h◦(p

i

◦h)

−1

.

Both composants are smooth by the general properties of manifolds of

mappings. So the chart change is smooth.

We show that the quotient space B(M, N) = Emb(M, N)/ Diﬀ(M) is

Hausdorﬀ. Let i, j ∈ Emb(M, N) with π(i) ,= π(j). Then i(M) ,= j(M)

in N for otherwise put i(M) = j(M) =: L, a submanifold of N; the

mapping i

−1

◦ j : M → L → M is then a diﬀeomorphism of M and

j = i◦(i

−1

◦j) ∈ i◦Diﬀ(M), so π(i) = π(j), contrary to the assumption.

Now we distinguish two cases.

Case 1. We may ﬁnd a point y

0

∈ i(M) ¸ j(M), say, which is not a

cluster point of j(M). We choose an open neighborhood V of y

0

in N

and an open neighborhood W of j(M) in N such that V ∩ W = ∅. Let

1 := ¦k ∈ Emb(M, N) : k(M) ∩ V ¦ J := ¦k ∈ Emb(M, N) : k(M) ⊂

W¦. Then 1 is obviously open in Emb(M, N) and 1 is even open in the

coarser compact open topology. Both 1 and J are Diﬀ(M) saturated,

i ∈ J, j ∈ 1, and 1 ∩ J = ∅. So π(1) and π(J) separate π(i) and

π(j) in B(M, N).

Case 2 Let i(M) ⊂ j(M) and j(M) ⊂ i(M). Let y ∈ i(X), say. Let

(V, v) be a chart of N centered at y which maps i(M) ∩ V into a linear

subspace, v(i(M)∩V ) ⊆ R

m

∩v(V ) ⊂ R

n

, where n = dimM, n = dimN.

Since j(M) ⊆ i(M) we conclude that we have also v((i(M) ∪ j(M)) ∩

V ) ⊆ R

m

∩v(V ). So we see that L := i(M)∪j(M) is a submanifold of N

of the same dimension as N. Let (W

L

, p

L

, L) be a tubular neighborhood

of L. Then W

L

[i(M) is a tubular neighborhood of i(M) and W

L

[j(M)

is one of j(M).

7.4. Theorem. Let M and N be smooth manifolds. Then the diﬀeo-

morphism group Diﬀ(M) acts smoothly from the right on the smooth

manifold Imm

prop

(M, N) of all smooth proper immersions M → N,

which is an open subset of C

∞

(M, N).

Then the space of orbits Imm

prop

(M, N)/ Diﬀ(M) is Hausdorﬀ in the

quotient topology.

Let Imm

free, prop

(M, N) be set of all proper immersions, on which

Diﬀ(M) acts freely. Then this is open in C

∞

(M, N) and is the total

space of a smooth principal ﬁber bundle

Imm

free,prop

(M, N) →Imm

free,prop

(M, N)/ Diﬀ(M).

This is proved in [Cervera-Mascaro-Michor, 1989], where also the exis-

tence of smooth transversals to each orbit is shown and the stratiﬁcation

of the orbit space into smooth manifolds is given.

27

8. The Fr¨olicher-Nijenhuis Bracket

8.1. In this section let M be a smooth manifold. We consider the graded

commutative algebra Ω(M) =

dimM

k=0

Ω

k

(M) of diﬀerential forms on

M. The space Der

k

Ω(M) consists of all (graded) derivations of degree k,

i.e. all linear mappings D : Ω(M) →Ω(M) with D(Ω

(M)) ⊂ Ω

k+

(M)

and D(ϕ ∧ ψ) = D(ϕ) ∧ ψ + (−1)

k

ϕ ∧ D(ψ) for ϕ ∈ Ω

(M).

Lemma. Then the space Der Ω(M) =

k

Der

k

Ω(M) is a graded Lie

algebra with the graded commutator [D

1

, D

2

] := D

1

◦ D

2

−(−1)

k

1

k

2

D

2

◦

D

1

as bracket. This means that the bracket is graded anticommutative,

[D

1

, D

2

] = −(−1)

k

1

k

2

[D

2

, D

1

], and satisﬁes the graded Jacobi identity

[D

1

, [D

2

, D

3

]] = [[D

1

, D

2

], D

3

] + (−1)

[

D

2

, [D

1

, D

3

]] (so that ad(D

1

) =

[D

1

, ] is itself a derivation).

Proof. Plug in the deﬁnition of the graded commutator and compute.

From the basics of diﬀerential geometry one is already familiar with

some graded derivations: for a vector ﬁeld X on M the derivation i

X

is

of degree −1, L

X

is of degree 0, and d is of degree 1. Note also that the

important formula L

X

= d i

X

+i

X

d translates to L

X

= [i

X

, d].

8.2. A derivation D ∈ Der

k

Ω(M) is called algebraic if D [ Ω

0

(M) = 0.

Then D(f.ω) = f.D(ω) for f ∈ C

∞

(M, R), so D is of tensorial character

and induces a derivation D

x

∈ Der

k

ΛT

∗

x

M for each x ∈ M. It is

uniquely determined by its restriction to 1-forms D

x

[T

∗

x

M : T

∗

x

M →

Λ

k+1

T

∗

M which we may view as an element K

x

∈ Λ

k+1

T

∗

x

M ⊗ T

x

M

depending smoothly on x ∈ M. To express this dependence we write

D = i

K

= i(K), where K ∈ C

∞

(Λ

k+1

T

∗

M ⊗ TM) =: Ω

k+1

(M; TM).

We call Ω(M, TM) =

dimM

k=0

Ω

k

(M, TM) the space of all vector valued

diﬀerential forms.

Theorem. (1) For K ∈ Ω

k+1

(M, TM) the formula

(i

K

ω)(X

1

, . . . , X

k+

) =

=

1

(k+1)! (−1)!

σ∈§

k+

sign σω(K(X

σ1

, . . . , X

σ(k+1)

), X

σ(k+2)

, . . . )

for ω ∈ Ω

(M), X

i

∈ X(M) (or T

x

M) deﬁnes an algebraic graded

derivation i

K

∈ Der

k

Ω(M) and any algebraic derivation is of this form.

(2) By i([K, L]´) := [i

K

, i

L

] we get a bracket [ , ]´ on the space

Ω

∗−1

(M, TM) which deﬁnes a graded Lie algebra structure with the

grading as indicated, and for K ∈ Ω

k+1

(M, TM), L ∈ Ω

+1

(M, TM)

we have

[K, L]´= i

K

L −(−1)

k

i

L

K.

28 8. Fr¨olicher-Nijenhuis bracket

[ , ]´is called the algebraic bracket or also the Nijenhuis-Richard-

son bracket, see [Nijenhuis-Richardson, 1967].

Proof. Since Λ

∗

T

∗

x

M is the free graded commutative algebra generated

by the vector space T

∗

x

M any K ∈ Ω

k+1

(M, TM) extends to a graded

derivation. By applying it to an exterior product of 1-forms one can

derive the formula in (1). The graded commutator of two algebraic

derivations is again algebraic, so the injection i : Ω

∗+1

(M, TM) →

Der

k

(Ω(M)) induces a graded Lie bracket on Ω

∗+1

(M, TM) whose form

can be seen by applying it to a 1-form.

8.3. The exterior derivative d is an element of Der

1

Ω(M). In view of

the formula L

X

= [i

X

, d] = i

X

d + d i

X

for vector ﬁelds X, we deﬁne

for K ∈ Ω

k

(M; TM) the Lie derivation L

K

= L(K) ∈ Der

k

Ω(M) by

L

K

:= [i

k

, d].

Then the mapping L : Ω(M, TM) → Der Ω(M) is injective, since

L

K

f = i

K

df = df ◦ K for f ∈ (

∞

(M, R).

Theorem. For any graded derivation D ∈ Der

k

Ω(M) there are unique

K ∈ Ω

k

(M; TM) and L ∈ Ω

k+1

(M; TM) such that

D = L

K

+i

L

.

We have L = 0 if and only if [D, d] = 0. D is algebraic if and only if

K = 0.

Proof. Let X

i

∈ X(M) be vector ﬁelds. Then f →(Df)(X

1

, . . . , X

k

) is

a derivation C

∞

(M, R) → C

∞

(M, R), so there is a unique vector ﬁeld

K(X

1

, . . . , X

k

) ∈ X(M) such that

(Df)(X

1

, . . . , X

k

) = K(X

1

, . . . , X

k

)f = df(K(X

1

, . . . , X

k

)).

Clearly K(X

1

, . . . , X

k

) is C

∞

(M, R)-linear in each X

i

and alternating,

so K is tensorial, K ∈ Ω

k

(M; TM).

The deﬁning equation for K is Df = df ◦ K = i

K

df = L

K

f for

f ∈ C

∞

(M, R). Thus D−L

K

is an algebraic derivation, so D−L

K

= i

L

by 8.2 for unique L ∈ Ω

k+1

(M; TM).

Since we have [d, d] = 2d

2

= 0, by the graded Jacobi identity we

obtain 0 = [i

K

, [d, d]] = [[i

K

, d], d] + (−1)

k−1

[d, [i

K

, d]] = 2[L

K

, d]. The

mapping L →[i

L

, d] = L

K

is injective, so the last assertions follow.

8.4. Applying i(Id

TM

) on a k-fold exterior product of 1-forms we see

that i(Id

TM

)ω = kω for ω ∈ Ω

k

(M). Thus we have L(Id

TM

) = ω =

i(Id

TM

)dω −d i(Id

TM

)ω = (k +1)dω −kdω = dω. Thus L(Id

TM

) = d.

8. Fr¨olicher-Nijenhuis bracket 29

8.5. Let K ∈ Ω

k

(M; TM) and L ∈ Ω

(M; TM). Then obviously

[[L

K

, L

L

], d] = 0, so we have

[L(K), L(L)] = L([K, L])

for a uniquely deﬁned [K, L] ∈ Ω

k+

(M; TM). This vector valued form

[K, L] is called the Fr¨olicher-Nijenhuis bracket of K and L.

Theorem. The space Ω(M; TM) =

dimM

k=0

Ω

k

(M; TM) with its usual

grading is a graded Lie algebra. Id

TM

∈ Ω

1

(M; TM) is in the center,

i.e. [K, Id

TM

] = 0 for all K.

L : (Ω(M; TM), [ , ]) →Der Ω(M) is an injective homomorphism

of graded Lie algebras. For vector ﬁelds the Fr¨olicher-Nijenhuis bracket

coincides with the Lie bracket.

Proof. df ◦ [X, Y ] = L([X, Y ])f = [L

X

, L

Y

]f. The rest is clear.

8.6. Lemma. For K ∈ Ω

k

(M; TM) and L ∈ Ω

+1

(M; TM) we have

[L

K

, i

L

] = i([K, L]) −(−1)

k

L(i

L

K), or

[i

L

, L

K

] = L(i

L

K) −(−1)

k

i([L, K]).

This generalizes the usual formula for vector ﬁelds.

Proof. For f ∈ C

∞

(M, R) we have [i

L

, L

K

]f = i

L

i

K

df − 0 = i

L

(df ◦

K) = df ◦ (i

L

K) = L(i

L

K)f. So [i

L

, L

K

] − L(i

L

K) is an algebraic

derivation.

[[i

L

, L

K

], d] = [i

L

, [L

K

, d]] −(−1)

k

[L

K

, [i

L

, d]] =

= 0 −(−1)

k

L([K, L]) = (−1)

k

[i([L, K]), d]).

Since [ , d] kills the ”L’s” and is injective on the ”i’s”, the algebraic

part of [i

L

, L

K

] is (−1)

k

i([L, K]).

8.7. The space Der Ω(M) is a graded module over the graded algebra

Ω(M) with the action (ω ∧ D)ϕ = ω ∧ D(ϕ).

Theorem. Let the degree of ω be q, of ϕ be k, and of ψ be . Let the

other degrees be as indicated. Then we have:

[ω ∧ D

1

, D

2

] = ω ∧ [D

1

, D

2

] −(−1)

(q+k

1

)k

2

d

2

(ω) ∧ D

1

. (1)

i(ω ∧ L) = ω ∧ i(L) (2)

ω ∧ L

K

= L(ω ∧ K) + (−1)

q+k−1

i(dω ∧ K). (3)

[ω ∧ L

1

, L

2

]´= ω ∧ [L

1

, L

2

]´− (4)

30 8. Fr¨olicher-Nijenhuis bracket

−(−1)

(q+

1

−1)(

2

−1)

i(L

2

)ω ∧ L

1

.

[ω ∧ K

1

, K

2

] = ω ∧ [K

1

, K

2

] −(−1)

(q+k

1

)k

2

L(K

2

)ω ∧ K

1

(5)

+ (−1)

q+k

1

dω ∧ i(K

1

)K

2

.

[ϕ ⊗X, ψ ⊗Y ] = ϕ ∧ ψ ⊗[X, Y ] (6)

−

_

i

Y

dϕ ∧ ψ ⊗X −(−1)

k

i

X

dψ ∧ ϕ ⊗Y

_

−

_

d(i

Y

ϕ ∧ ψ) ⊗X −(−1)

k

d(i

X

ψ ∧ ϕ) ⊗Y

_

= ϕ ∧ ψ ⊗[X, Y ] +ϕ ∧ L

X

ψ ⊗Y −L

Y

ϕ ∧ ψ ⊗X

+ (−1)

k

(dϕ ∧ i

X

ψ ⊗Y +i

Y

ϕ ∧ dψ ⊗X) .

Proof. For (1) , (2) , (3) write out the deﬁnitions. For (4) compute

i([ω ∧ L

1

, L

2

]´). For (5) compute L([ω ∧ K

1

, K

2

]). For (6) use (5) .

8.8. Theorem. For K ∈ Ω

k

(M; TM) and ω ∈ Ω

**(M) the Lie deriv-
**

ative of ω along K is given by the following formula, where the X

i

are

vector ﬁelds on M.

(L

K

ω)(X

1

, . . . , X

k+

) =

=

1

k! !

σ

sign σ L(K(X

σ1

, . . . , X

σk

))(ω(X

σ(k+1)

, . . . , X

σ(k+)

))

+

−1

k! (−1)!

σ

sign σ ω([K(X

σ1

, . . . , X

σk

), X

σ(k+1)

], X

σ(k+2)

, . . . )

+

(−1)

k−1

(k−1)! (−1)! 2!

σ

sign σ ω(K([X

σ1

, X

σ2

], X

σ3

, . . . ), X

σ(k+2)

, . . . ).

Proof. It suﬃces to consider K = ϕ⊗X. Then by 8.7.3 we have L(ϕ⊗

X) = ϕ ∧ L

X

−(−1)

k−1

dϕ ∧ i

X

. Now use the usual global formulas to

expand this.

8.9. Theorem. Global formula for the Fr¨olicher-Nijenhuis

bracket. For K ∈ Ω

k

(M; TM) and L ∈ Ω

**(M; TM) we have for the
**

Fr¨olicher-Nijenhuis bracket [K, L] the following formula, where the X

i

are vector ﬁelds on M.

[K, L](X

1

, . . . , X

k+

) =

=

1

k! !

σ

sign σ [K(X

σ1

, . . . , X

σk

), L(X

σ(k+1)

, . . . , X

σ(k+)

)]

+

−1

k! (−1)!

σ

sign σ L([K(X

σ1

, . . . , X

σk

), X

σ(k+1)

], X

σ(k+2)

, . . . )

8. Fr¨olicher-Nijenhuis bracket 31

+

(−1)

k

(k−1)! !

σ

sign σ K([L(X

σ1

, . . . , X

σ

), X

σ(+1)

], X

σ(+2)

, . . . )

+

(−1)

k−1

(k−1)! (−1)! 2!

σ

sign σ L(K([X

σ1

, X

σ2

], X

σ3

, . . . ), X

σ(k+2)

, . . . )

+

(−1)

(k−1)

(k−1)! (−1)! 2!

σ

sign σ K(L([X

σ1

, X

σ2

], X

σ3

, . . . ), X

σ(+2)

, . . . ).

Proof. It suﬃces to consider K = ϕ ⊗ X and L = ψ ⊗ Y , then for

[ϕ ⊗ X, ψ ⊗ Y ] we may use 8.7.6 and evaluate that at (X

1

, . . . , X

k+

).

After some combinatorial computation we get the right hand side of the

above formula for K = ϕ ⊗X and L = ψ ⊗Y .

There are more illuminating ways to prove this formula, see [Michor,

1987].

8.10. Local formulas. In a local chart (U, u) on the manifold M we

put K [ U =

K

i

α

d

α

⊗∂

i

, L [ U =

L

j

β

d

β

⊗∂

j

, and ω [ U =

ω

γ

d

γ

,

where α = (1 ≤ α

1

< α

2

< < α

k

≤ dimM) is a form index,

d

α

= du

α

1

∧ . . . ∧ du

α

k

, ∂

i

=

∂

∂u

i

and so on.

Plugging X

j

= ∂

i

j

into the global formulas 8.2, 8.8, and 8.9, we get

the following local formulas:

i

K

ω [ U =

K

i

α

1

...α

k

ω

iα

k+1

...α

k+−1

d

α

[K, L]´[ U =

_

K

i

α

1

...α

k

L

j

iα

k+1

...α

k+

−(−1)

(k−1)(−1)

L

i

α

1

...α

K

j

iα

+1

...α

k+

_

d

α

⊗∂

j

L

K

ω [ U =

_

K

i

α

1

...α

k

∂

i

ω

α

k+1

...α

k+

+ (−1)

k

(∂

α

1

K

i

α

1

...α

k

) ω

iα

k+2

...α

k+

_

d

α

[K, L] [ U =

_

K

i

α

1

...α

k

∂

i

L

j

α

k+1

...α

k+

−(−1)

k

L

i

α

1

...α

∂

i

K

j

α

+1

...α

k+

−kK

j

α

1

...α

k−1

i

∂

α

k

L

i

α

k+1

...α

k+

+ (−1)

k

L

j

α

1

...α

−1

i

∂

α

K

i

α

+1

...α

k+

_

d

α

⊗∂

j

8.11. Theorem. For K

i

∈ Ω

k

i

(M; TM) and L

i

∈ Ω

k

i

+1

(M; TM) we

have

[L

K

1

+i

L

1

, L

K

2

+i

L

2

] = (1)

= L

_

[K

1

, K

2

] +i

L

1

K

2

−(−1)

k

1

k

2

i

L

2

K

1

_

+i

_

[L

1

, L

2

]´+ [K

1

, L

2

] −(−1)

k

1

k

2

[K

2

, L

1

]

_

.

32 8. Fr¨olicher-Nijenhuis bracket

Each summand of this formula looks like a semidirect product of graded

Lie algebras, but the mappings

i : Ω(M; TM) →End(Ω(M; TM), [ , ])

ad : Ω(M; TM) →End(Ω(M; TM), [ , ]´)

do not take values in the subspaces of graded derivations. We have in-

stead for K ∈ Ω

k

(M; TM) and L ∈ Ω

+1

(M; TM) the following rela-

tions:

i

L

[K

1

, K

2

] = [i

L

K

1

, K

2

] + (−1)

k

1

[K

1

, i

L

K

2

] (2)

−

_

(−1)

k

1

i([K

1

, L])K

2

−(−1)

(k

1

+)k

2

i([K

2

, L])K

1

_

[K, [L

1

, L

2

]´] = [[K, L

1

], L

2

]´+ (−1)

kk

1

[L

1

, [K, L

2

]]´− (3)

−

_

(−1)

kk

1

[i(L

1

)K, L

2

] −(−1)

(k+k

1

)k

2

[i(L

2

)K, L

1

]

_

The algebraic meaning of the relations of this theorem and its conse-

quences in group theory have been investigated in [Michor, 1989b]. The

corresponding product of groups is well known to algebraists under the

name ”Zappa-Szep”-product.

Proof. Equation (1) is an immediate consequence of 8.6. Equations (2)

and (3) follow from (1) by writing out the graded Jacobi identity, or as

follows: Consider L(i

L

[K

1

, K

2

]) and use 8.6 repeatedly to obtain L of

the right hand side of (2). Then consider i([K, [L

1

, L

2

]´]) and use again

8.6 several times to obtain i of the right hand side of (3).

8.12. Corollary (of 8.9). For K, L ∈ Ω

1

(M; TM) we have

[K, L](X, Y ) = [KX, LY ] −[KY, LX]

−L([KX, Y ] −[KY, X])

−K([LX, Y ] −[LY, X])

+ (LK +KL)[X, Y ].

8.13. Curvature. Let P ∈ Ω

1

(M; TM) be a ﬁber projection, i.e.

P ◦ P = P. This is the most general case of a (ﬁrst order) connection.

We may call ker P the horizontal space and imP the vertical space of the

connection. If P is of constant rank, then both are sub vector bundles

of TM. If imP is some primarily ﬁxed sub vector bundle or (tangent

bundle of) a foliation, P can be called a connection for it. Special cases

of this will be treated extensively later on. For the general theory see

[Michor, 1989a] The following result is immediate from 8.12.

8. Fr¨olicher-Nijenhuis bracket 33

Lemma. We have

[P, P] = 2R + 2

¯

R,

where R,

¯

R ∈ Ω

2

(M; TM) are given by R(X, Y ) = P[(Id −P)X, (Id −

P)Y ] and

¯

R(X, Y ) = (Id −P)[PX, PY ].

If P has constant rank, then R is the obstruction against integra-

bility of the horizontal bundle ker P, and

¯

R is the obstruction against

integrability of the vertical bundle imP. Thus we call R the curvature

and

¯

R the cocurvature of the connection P. We will see later, that for a

principal ﬁber bundle R is just the negative of the usual curvature.

8.14. Lemma. General Bianchi identity. If P ∈ Ω

1

(M; TM) is a

connection (ﬁber projection) with curvature R and cocurvature

¯

R, then

we have

[P, R +

¯

R] = 0

[R, P] = i

R

¯

R +i ¯

R

R.

Proof. We have [P, P] = 2R + 2

¯

R by 8.13 and [P, [P, P]] = 0 by the

graded Jacobi identity. So the ﬁrst formula follows. We have 2R =

P ◦ [P, P] = i

[P,P]

P. By 8.11.2 we get i

[P,P]

[P, P] = 2[i

[P,P]

P, P] − 0 =

4[R, P]. Therefore [R, P] =

1

4

i

[P,P]

[P, P] = i(R+

¯

R)(R+

¯

R) = i

R

¯

R+i ¯

R

R

since R has vertical values and kills vertical vectors, so i

R

R = 0; likewise

for

¯

R.

8.15. Naturality of the Fr¨olicher-Nijenhuis bracket. Let f :

M → N be a smooth mapping between manifolds. Two vector val-

ued forms K ∈ Ω

k

(M; TM) and K

∈ Ω

k

(N; TN) are called f-related

or f-dependent, if for all X

i

∈ T

x

M we have

(1) K

f(x)

(T

x

f X

1

, . . . , T

x

f X

k

) = T

x

f K

x

(X

1

, . . . , X

k

).

Theorem.

(2) If K and K

**as above are f-related then i
**

K

◦ f

∗

= f

∗

◦ i

K

:

Ω(N) →Ω(M).

(3) If i

K

◦ f

∗

[ B

1

(N) = f

∗

◦ i

K

[ B

1

(N), then K and K

are

f-related, where B

1

denotes the space of exact 1-forms.

(4) If K

j

and K

j

are f-related for j = 1, 2, then i

K

1

K

2

and i

K

1

K

2

are f-related, and also [K

1

, K

2

]´and [K

1

, K

2

]´are f-related.

(5) If K and K

**are f-related then L
**

K

◦ f

∗

= f

∗

◦ L

K

: Ω(N) →

Ω(M).

34 8. Fr¨olicher-Nijenhuis bracket

(6) If L

K

◦ f

∗

[ Ω

0

(N) = f

∗

◦ L

K

[ Ω

0

(N), then K and K

are

f-related.

(7) If K

j

and K

j

are f-related for j = 1, 2, then their Fr¨olicher-

Nijenhuis brackets [K

1

, K

2

] and [K

1

, K

2

] are also f-related.

Proof. (2) By 8.2 we have for ω ∈ Ω

q

(N) and X

i

∈ T

x

M:

(i

K

f

∗

ω)

x

(X

1

, . . . , X

q+k−1

) =

=

1

k! (q−1)!

σ

sign σ (f

∗

ω)

x

(K

x

(X

σ1

, . . . , X

σk

), X

σ(k+1)

, . . . )

=

1

k! (q−1)!

σ

sign σ ω

f(x)

(T

x

f K

x

(X

σ1

, . . . ), T

x

f X

σ(k+1)

, . . . )

=

1

k! (q−1)!

σ

sign σ ω

f(x)

(K

f(x)

(T

x

f X

σ1

, . . . ), T

x

f X

σ(k+1)

, . . . )

= (f

∗

i

K

ω)

x

(X

1

, . . . , X

q+k−1

)

(3) follows from this computation, since the df, f ∈ C

∞

(M, R) sep-

arate points. (4) follows from the same computation for K

2

instead of

ω, the result for the bracket then follows 8.2.2.

(5) The algebra homomorphism f

∗

intertwines the operators i

K

and

i

K

by (2), and f

∗

commutes with the exterior derivative d. Thus f

∗

intertwines the commutators [i

K

, d] = L

K

and [i

K

, d] = L

K

.

(6) For g ∈ Ω

0

(N) we have L

K

f

∗

g = i

K

d f

∗

g = i

K

f

∗

dg and

f

∗

L

K

g = f

∗

i

K

dg. By (3) the result follows.

(6) The algebra homomorphism f

∗

intertwines L

K

j

and L

K

j

, so also

their graded commutators which equal L([K

1

, K

2

]) and L([K

1

, K

2

]), re-

spectively. Now use (6) .

8.16. Let f : M →N be a local diﬀeomorphism. Then we can consider

the pullback operator f

∗

: Ω(M; TM) →Ω(M; TM), given by

(1) (F

∗

K)

x

(X

1

, . . . , X

k

) = (T

x

f)

−1

K

f(x)

(T

x

f X

1

, . . . , T

x

f X

k

).

Clearly K and f

∗

K are then f-related.

Theorem. In this situation we have:

(2) f

∗

[K, L] = [f

∗

K, f

∗

L].

(3) f

∗

i

K

L = i

f

∗

K

f

∗

L.

(4) f

∗

[K, L]´= [f

∗

K, f

∗

L]´.

(5) For a vector ﬁeld X ∈ X(M) and K ∈ Ω(M; TM) the Lie

derivative L

X

K =

∂

∂t

¸

¸

0

(Fl

X

t

)

∗

K satisﬁes L

X

K = [X, K], the

Fr¨olicher-Nijenhuis-bracket.

8. Fr¨olicher-Nijenhuis bracket 35

We may say that the Fr¨olicher-Nijenhuis bracket, [ , ]´, etc. are

natural bilinear concomitants.

Proof. (2) – (4) are obvious from 8.16. (5) Obviously L

X

is R-linear,

so it suﬃces to check this formula for K = ψ ⊗ Y , ψ ∈ Ω(M) and

Y ∈ X(M). But then

L

X

(ψ ⊗Y ) = L

X

ψ ⊗Y +ψ ⊗L

X

Y

= L

X

ψ ⊗Y +ψ ⊗[X, Y ]

= [X, ψ ⊗Y ] by 8.7.6.

8.17. Remark. At last we mention the best known application of the

Fr¨olicher-Nijenhuis bracket, which also led to its discovery. A vector

valued 1-form J ∈ Ω

1

(M; TM) with J ◦ J = −Id is called a nearly

complex structure; if it exists, dimM is even and J can be viewed as a

ﬁber multiplication with

√

−1 on TM. By 8.12 we have

[J, J](X, Y ) = 2([JX, JY ] −[X, Y ] −J[X, JY ] −J[JX, Y ]).

The vector valued form

1

2

[J, J] is also called the Nijenhuis tensor of J.

For it the following result is true:

A manifold M with a nearly complex structure J is a complex man-

ifold (i.e., there exists an atlas for M with holomorphic chart-change

mappings) if and only if [J, J] = 0. See [Newlander-Nirenberg, 1957].

36

9. Fiber Bundles and Connections

9.1. Deﬁnition. A (ﬁber) bundle (E, p, M, S) consists of manifolds E,

M, S, and a smooth mapping p : E →M; furthermore each x ∈ M has

an open neighborhood U such that E [ U := p

−1

(U) is diﬀeomorphic to

U S via a ﬁber respecting diﬀeomorphism:

E [ U

ψ

−−−−→ U S

p

¸

¸

_

pr

1

¸

¸

_

U U

E is called the total space, M is called the base space or basis, p is a

surjective submersion, called the projection, and S is called standard

ﬁber. (U, ψ) as above is called a ﬁber chart.

A collection of ﬁber charts (U

α

, ψ

α

), such that (U

α

) is an open cover

of M, is called a (ﬁber) bundle atlas. If we ﬁx such an atlas, then

ψ

α

◦ψ

β

−1

(x, s) = (x, ψ

αβ

(x, s)), where ψ

αβ

: (U

α

∩U

β

)S →S is smooth

and ψ

αβ

(x, ) is a diﬀeomorphism of S for each x ∈ U

αβ

:= U

α

∩ U

β

.

We may thus consider the mappings ψ

αβ

: U

αβ

→ Diﬀ(S) with values

in the group Diﬀ(S) of all diﬀeomorphisms of S; their diﬀerentiability is

a subtle question, which will be discussed in 13.1 below. In either form

these mappings ψ

αβ

are called the transition functions of the bundle.

They satisfy the cocycle condition: ψ

αβ

(x) ◦ ψ

βγ

(x) = ψ

αγ

(x) for x ∈

U

αβγ

and ψ

αα

(x) = Id

S

for x ∈ U

α

. Therefore the collection (ψ

αβ

) is

called a cocycle of transition functions.

Given an open cover (U

α

) of a manifold M and a cocycle of transition

functions (ψ

αβ

) we may construct a ﬁber bundle (E, p, M, S) similarly as

in the following way: On the disjoint union

α

¦α¦U

α

S we consider

the equivalence relation:

(α, x, s) ∼ (β, y, t) if and only if x = y and ψ

αβ

(y, t) = s.

The quotient space then turns out to be a ﬁber bundle, and we may use

the mappings ψ

α

([(α, x, s)]) := (x, s) as a bundle atlas whose cocycle of

transition functions is exactly the one we started from.

9.2. Lemma. Let p : N → M be a surjective submersion (a ﬁbred

manifold) such that p

−1

(x) is compact for each x ∈ M and let M be

connected. Then (N, p, M) is a ﬁber bundle.

Proof. We have to produce a ﬁber chart at each x

0

∈ M. So let (U, u) be

a chart centered at x

0

on M such that u(U)

∼

= R

m

. For each x ∈ U let

ξ

x

(y) := (T

y

u)

−1

.u(x), then ξ

x

∈ X(U), depending smoothly on x ∈ U,

9. Connections on general ﬁber bundles 37

such that u(Fl

ξ

x

t

u

−1

(z)) = z + t.u(x), so each ξ

x

is a complete vector

ﬁeld on U. Since p is a submersion, with the help of a partition of unity

on p

−1

(U) we may construct vector ﬁelds η

x

∈ X(p

−1

(U)) which depend

smoothly on x ∈ U and are p-related to ξ

x

: Tp.η

x

= ξ

x

◦ p. Thus

p ◦ Fl

η

x

t

= Fl

ξ

x

t

◦p and Fl

η

x

t

is ﬁber respecting, and since each ﬁber is

compact and ξ

x

is complete, η

x

has a global ﬂow too. Denote p

−1

(x

0

)

by S. Then ϕ : U S → p

−1

(U), deﬁned by ϕ(x, y) = Fl

η

x

1

(y), is

a diﬀeomorphism and is ﬁber respecting, so (U, ϕ

−1

) is a ﬁber chart.

Since M is connected, the ﬁbers p

−1

(x) are all diﬀeomorphic.

9.3. Let (E, p, M, S) be a ﬁber bundle; we consider the ﬁber linear

mapping Tp : TE → TM and its kernel ker Tp =: V E which is called

the vertical bundle of E. The following is special case of 8.13.

Deﬁnition A connection on the ﬁber bundle (E, p, M, S) is a vector

valued 1-form Φ ∈ Ω

1

(E; V E) with values in the vertical bundle VE such

that Φ ◦ Φ = Φ and ImΦ = V E; so Φ is just a projection TE →V E.

Then ker Φ is of constant rank, so ker Φ is a sub vector bundle of TE,

it is called the space of horizontal vectors or the horizontal bundle and

it is denoted by HE Clearly TE = HE ⊕ V E and T

u

E = H

u

E ⊕ V

u

E

for u ∈ E.

Now we consider the mapping (Tp, π

E

) : TE → TM

M

E. Then

(Tp, π

E

)

−1

(0

p(u)

, u) = V

u

E by deﬁnition, so (Tp, π

E

) [ HE : HE →

TM

M

E is ﬁber linear over E and injective, so by reason of dimensions

it is a ﬁber linear isomorphism: Its inverse is denoted by

C := ((Tp, π

E

) [ HE)

−1

: TM

M

E →HE →TE.

So C : TM

M

E →TE is ﬁber linear over E and is a right inverse for

(Tp, π

E

). C is called the horizontal lift associated to the connection Φ.

Note the formula Φ(ξ

u

) = ξ

u

−C(Tp.ξ

u

, u) for ξ

u

∈ T

u

E. So we can

equally well describe a connection Φ by specifying C. Then we call Φ

the vertical projection and χ := id

TE

− Φ = C ◦ (Tp, π

E

) will be called

the horizontal projection.

9.4. Curvature. If Φ : TE → V E is a connection on a ﬁber bundle

(E, p, M, S), then as in 8.13 the curvature R of Φ is given by

2R = [Φ, Φ] = [Id −Φ, Id −Φ] = [χ, χ] ∈ Ω

2

(E; V E).

We have R(X, Y ) =

1

2

[Φ, Φ](X, Y ) = Φ[χX, χY ], so R is an obstruc-

tion against integrability of the horizontal subbundle. Since the vertical

bundle V E is integrable, by 8.14 we have the Bianchi identity [Φ, R] = 0.

38 9. Connections on general ﬁber bundles

9.5. Pullback. Let (E, p, M, S) be a ﬁber bundle and consider a

smooth mapping f : N → M. Since p is a submersion, f and p are

transversal and thus the pullback N

(f,M,p)

E exists. It will be called

the pullback of the ﬁber bundle E by F and we will denote it by f

∗

E.

The following diagram sets up some further notation for it:

f

∗

E

p

∗

f

−−−−→ E

f

∗

¸

¸

_

p

¸

¸

_

N −−−−→

f

M

Proposition. In the situation above we have:

(1) (f

∗

E, f

∗

p, N, S) is again a ﬁber bundle, and p

∗

f is a ﬁber wise

diﬀeomorphism.

(2) If Φ ∈ Ω

1

(E; TE) is a connection on the bundle E, the vector val-

ued form f

∗

Φ, given by (f

∗

Φ)

u

(X) := T

u

(p

∗

f)

−1

.Φ.T

u

(p

∗

f).X

for X ∈ T

u

E, is again a connection on the bundle f

∗

E. The

forms f

∗

Φ and Φ are p

∗

f-related in the sense of 8.15.

(3) The curvatures of f

∗

Φ and Φ are also p

∗

f-related: R(f

∗

Φ) =

f

∗

R(Φ).

Proof. (1). If (U

α

, ψ

α

) is a ﬁber bundle atlas of (E, p, M, S) in the sense

of 9.1, then (f

−1

(U

α

), (f

∗

p, pr

2

◦ψ

α

◦p

∗

f)) is visibly a ﬁber bundle atlas

for (f

∗

E, f

∗

p, N, S), by the formal universal properties of a pullback.

(2) is obvious. (3) follows from (2) and 8.15.6.

9.6. Flat connections. Let us suppose that a connection Φ on the

bundle (E, p, M, S) has zero curvature. Then by 9.4 the horizontal bun-

dle is integrable and gives rise to the horizontal foliation. Each point

u ∈ E lies on a unique leaf L(u) such that T

v

L(u) = H

v

E for each

v ∈ L(u). The restriction p [ L(u) is locally a diﬀeomorphism, but in

general it is neither surjective nor is it a covering onto its image. This

is seen by devising suitable horizontal foliations on the trivial bundle

pr

2

: RS

1

→S

1

.

9.7. Local description of connections. Let Φ be a connection on

(E, p, M, S). Let us ﬁx a ﬁber bundle atlas (U

α

) with transition functions

(ψ

αβ

), and let us consider the connection (ψ

α

)

−1

)

∗

Φ ∈ Ω

1

(U

α

S; U

α

**TS), which may be written in the form
**

((ψ

α

)

−1

)

∗

Φ)(ξ

x

, η

y

) =: −Γ

α

(ξ

x

, y) +η

y

for ξ

x

∈ T

x

U

α

and η

y

∈ T

y

S,

9. Connections on general ﬁber bundles 39

since it reproduces vertical vectors. The Γ

α

are given by

(0

x

, Γ

α

(ξ

x

, y)) := −T(ψ

α

).Φ.T(ψ

α

)

−1

.(ξ

x

, 0

y

).

We consider Γ

α

as an element of the space Ω

1

(U

α

; X(S)), a 1-form on

U

α

with values in the inﬁnite dimensional Lie algebra X(S) of all vector

ﬁelds on the standard ﬁber. The Γ

α

are called the Christoﬀel forms of

the connection Φ with respect to the bundle atlas (U

α

, ψ

α

).

Lemma. The transformation law for the Christoﬀel forms is

T

y

(ψ

αβ

(x, )).Γ

β

(ξ

x

, y) = Γ

α

(ξ

x

, ψ

αβ

(x, y)) −T

x

(ψ

αβ

( , y)).ξ

x

.

The curvature R of Φ satisﬁes

(ψ

−1

α

)

∗

R = dΓ

α

+ [Γ

α

, Γ

α

]

X(S)

.

The formula for the curvature is the Maurer-Cartan formula which

in this general setting appears only in the level of local description.

Proof. From ψ

α

◦ (ψ

β

)

−1

(x, y) = (x, ψ

αβ

(x, y)) we get that

T(ψ

α

◦ (ψ

β

)

−1

).(ξ

x

, η

y

) = (ξ

x

, T

(x,y)

(ψ

αβ

).(ξ

x

, η

y

)) and thus:

T(ψ

−1

β

).(0

x

, Γ

β

(ξ

x

, y)) = −Φ(T(ψ

−1

β

)(ξ

x

, 0

y

)) =

= −Φ(T(ψ

−1

α

).T(ψ

α

◦ ψ

−1

β

).(ξ

x

, 0

y

)) =

= −Φ(T(ψ

−1

α

)(ξ

x

, T

(x,y)

(ψ

αβ

)(ξ

x

, 0

y

))) =

= −Φ(T(ψ

−1

α

)(ξ

x

, 0

y

)) −Φ(T(ψ

−1

α

)(0

x

, T

(x,y)

ψ

αβ

(ξ

x

, 0

y

)) =

= T(ψ

−1

α

).(0

x

, Γ

α

(ξ

x

, ψ

αβ

(x, y))) −T(ψ

−1

α

)(0

x

, T

x

(ψ

αβ

( , y)).ξ

x

).

This implies the transformation law.

For the curvature R of Φ we have by 9.4 and 9.5.3

(ψ

−1

α

)

∗

R((ξ

1

, η

1

), (ξ

2

, η

2

)) =

= (ψ

−1

α

)

∗

R[(Id −(ψ

−1

α

)

∗

Φ)(ξ

1

, η

1

), (Id −(ψ

−1

α

)

∗

Φ)(ξ

2

, η

2

)] =

= (ψ

−1

α

)

∗

Φ[(ξ

1

, Γ

α

(ξ

1

)), (ξ

2

, Γ

α

(ξ

2

))] =

= (ψ

−1

α

)

∗

Φ

_

[ξ

1

, ξ

2

], ξ

1

Γ

α

(ξ

2

) −ξ

2

Γ

α

(ξ

1

) + [Γ

α

(ξ

1

), Γ

α

(ξ

2

)]

_

=

= −Γ

α

([ξ

1

, ξ

2

]) +ξ

1

Γ

α

(ξ

2

) −ξ

2

Γ

α

(ξ

1

) + [Γ

α

(ξ

1

), Γ

α

(ξ

2

)] =

= dΓ

α

(ξ

1

, ξ

2

) + [Γ

α

(ξ

1

), Γ

α

(ξ

2

)]

X(S)

.

40 9. Connections on general ﬁber bundles

9.8. Theorem. Parallel transport. Let Φ be a connection on a

bundle (E, p, M, S) and let c : (a, b) → M be a smooth curve with 0 ∈

(a, b), c(0) = x.

Then there is a neighborhood U of E

x

¦0¦ in E

x

(a, b) and a

smooth mapping Pt

c

: U →E such that:

(1) p(Pt(c, t, u

x

)) = c(t) if deﬁned, and Pt(c, u

x

, 0) = u

x

.

(2) Φ(

d

dt

Pt(c, t, u

x

)) = 0

c(t)

if deﬁned.

(3) Reparametrisation invariance: If f : (a

, b

) → (a, b) is smooth

with 0 ∈ (a

, b

), then Pt(c, f(t), u

x

) = Pt(c ◦ f, t, Pt(c, u

x

, f(0)))

if deﬁned.

(4) U is maximal for properties (1) and (2).

(5) The parallel transport is smooth as a mapping

C

∞

(R, M)

(ev

0

,M,p)

E R ⊃ U

Pt

−→E,

where U is its domain of deﬁnition.

(6) If X ∈ X(M) is a vector ﬁeld on the base then the parallel trans-

port along the ﬂow lines of X is given by the ﬂow of the horizontal

lift CX of X:

Pt(Fl

X

(x), t, u

x

) = Fl

CX

t

(u

x

).

Proof. We ﬁrst give three diﬀerent proofs of assertions (1) – (4).

First proof. In local bundle coordinates Φ(

d

dt

Pt(c, u

x

, t)) = 0 is an ordi-

nary diﬀerential equation of ﬁrst order, nonlinear, with initial condition

Pt(c, u

x

, 0) = u

x

. So there is a maximally deﬁned local solution curve

which is unique. All further properties are consequences of uniqueness.

Second proof. Consider the pullback bundle (c

∗

E, c

∗

p, (a, b), S) and the

pullback connection c

∗

Φ on it. It has zero curvature, since the horizontal

bundle is 1-dimensional. By 9.6 the horizontal foliation exists and the

parallel transport just follows a leaf and we may map it back to E, in

detail: Pt(c, u

x

, t) = p

∗

c((c

∗

p [ L(u

x

))

−1

(t)).

Third proof. Consider a ﬁber bundle atlas (U

α

, ψ

α

) as in 9.7. Then

ψ

α

(Pt(c, ψ

−1

α

(x, y), t)) = (c(t), γ(y, t)), where

0 =

_

(ψ

−1

α

)

∗

Φ

_ _

d

dt

c(t),

d

dt

γ(y, t)

_

= −Γ

α

_

d

dt

c(t), γ(y, t)

_

+

d

dt

γ(y, t),

so γ(y, t) is the integral curve (evolution line) through y ∈ S of the

time dependent vector ﬁeld Γ

α

_

d

dt

c(t)

_

on S. This vector ﬁeld visibly

9. Connections on general ﬁber bundles 41

depends smoothly on c. Clearly local solutions exist and all properties

follow.

(5). By the considerations of section 6 it suﬃces to show that Pt

maps a smooth curve

(c

1

, c

2

, c

3

) : R →U ⊂ C

∞

(R, M)

(ev

0

,M,p)

E R

to a smooth curve in E. We use the description of the smooth curves

in C

∞

(R, M) from 6.2 and see that the ordinary diﬀerential equation

considered in the ﬁrst proof then just depends on some parameters more.

By the theory of ordinary diﬀerential equations the solution is smooth

in this parameters also, which we then set equal.

(6) is obvious from the deﬁnition.

9.9. Lemma. Let Φ be a connection on a bundle (E, p, M, S) with

curvature R and horizontal lift C. Let X ∈ X(M) be a vector ﬁeld on

the base.

Then for the horizontal lift CX ∈ X(E) we have

L

CX

Φ =

∂

∂t

¸

¸

0

(Fl

CX

t

)

∗

Φ = [CX, Φ] = −

1

2

i

CX

R.

Proof. From 8.16.(5) we get L

CX

Φ =

∂

∂t

¸

¸

0

(Fl

CX

t

)

∗

Φ = [CX, Φ]. From

8.11.(2) we have

i

CX

R = i

CX

[Φ, Φ]

= [i

CX

Φ, Φ] −[Φ, i

CX

Φ] + 2i

[Φ,CX]

Φ

= −2Φ[CX, Φ]

The vector ﬁeld CX is p-related to X and Φ ∈ Ω

1

(E; TE) is p-related

to 0 ∈ Ω

1

(M; TM), so by 8.15.(7) the form [CX, Φ] ∈ Ω

1

(E; TE) is also

p-related to 0 = [X, 0] ∈ Ω

1

(M; TM). So Tp.[CX, Φ] = 0, [CX, Φ] has

vertical values, and [CX, Φ] = Φ[CX, Φ].

9.10. A connection Φ on (E, p, M, S) is called a complete connection,

if the parallel transport Pt

c

along any smooth curve c : (a, b) → M is

deﬁned on the whole of E

c(0)

(a, b). The third proof of theorem 9.8

shows that on a ﬁber bundle with compact standard ﬁber any connection

is complete.

The following is a suﬃcient condition for a connection Φ to be com-

plete:

There exists a ﬁber bundle atlas (U

α

, ψ

α

) and complete Riemann-

ian metrics g

α

on the standard ﬁber S such that each Christoﬀel

form Γ

α

∈ Ω

1

(U

α

, X(S)) takes values in the linear subspace of

g

α

-bounded vector ﬁelds on S

42 9. Connections on general ﬁber bundles

For in the third proof of theorem 9.8 above the time dependent vector

ﬁeld Γ

α

(

d

dt

c(t)) on S is g

α

-bounded for compact time intervals. So by

continuation the solution exists over c

−1

(U

α

), and thus globally.

A complete connection is called an Ehresmann connection in [Greub

- Halperin - Vanstone I, p 314], where it is also indicated how to prove

the following result.

Theorem. Each ﬁber bundle admits complete connections.

Proof. Let dimM = m. Let (U

α

, ψ

α

) be a ﬁber bundle atlas as in

9.1. By topological dimension theory [Nagata, 1965] the open cover

(U

α

) of M admits a reﬁnement such that any m + 2 members have

empty intersection. Let U

α

) itself have this property. Choose a smooth

partition of unity (f

α

) subordinated to (U

α

). Then the sets V

α

:= ¦ x :

f

α

(x) >

1

m+2

¦ ⊂ U

α

form still an open cover of M since

f

α

(x) = 1

and at most m + 1 of the f

α

(x) can be nonzero. By renaming assume

that each V

α

is connected. Then we choose an open cover (W

α

) of M

such that W

α

⊂ V

α

.

Now let g

1

and g

2

be complete Riemannian metrics on M and S,

respectively (see [Nomizu - Ozeki, 1961] or [Morrow, 1970]). For not

connected Riemannian manifolds complete means that each connected

component is complete. Then g

1

[U

α

g

2

is a Riemannian metric on

U

α

S and we consider the metric g :=

f

α

ψ

∗

α

(g

1

[U

α

g

2

) on E.

Obviously p : E →M is a Riemannian submersion for the metrics g and

g

1

. We choose now the connection Φ : TE → V E as the orthonormal

projection with respect to the Riemannian metric g.

Claim: Φ is a complete connection on E.

Let c : [0, 1] → M be a smooth curve. We choose a partition 0 = t

0

<

t

1

< < t

k

= 1 such that c([t

i

, t

i+1

]) ⊂ V

α

i

for suitable α

i

. It suﬃces

to show that Pt(c(t

i

+ ), t, u

c(t

i

)

) exists for all 0 ≤ t ≤ t

i+1

− t

i

and

all u

c(t

i

)

∈ E

c(t

i

)

, for all i — then we may piece them together. So

we may assume that c : [0, 1] → V

α

for some α. Let us now assume

that for some (x, y) ∈ V

α

S the parallel transport Pt(c, t, ψ

α

(x, y)) is

deﬁned only for t ∈ [0, t

) for some 0 < t

**< 1. By the third proof of 9.8
**

we have Pt(c, t, ψ

α

(x, y)) = ψ

α

(c(t), γ(t)), where γ : [0, t

) → S is the

maximally deﬁned integral curve through y ∈ S of the time dependent

vector ﬁeld Γ

α

(

d

dt

c(t), ) on S. We put g

α

:= (ψ

−1

α

)

∗

g, then (g

α

)

(x,y)

=

(g

1

)

x

(

β

f

β

(x)ψ

βα

(x, )

∗

g

2

)

y

. Since pr

1

: (V

α

S, g

α

) →(V

α

, g

1

[V

α

)

is a Riemannian submersion and since the connection (ψ

−1

α

)

∗

Φ is also

given by orthonormal projection onto the vertical bundle, we get

∞> g

1

-length

t

0

(c) = g

α

-length(c, γ) =

_

t

0

[(c

(t),

d

dt

γ(t))[

g

α

dt =

9. Connections on general ﬁber bundles 43

=

_

t

0

_

[c

(t)[

2

g

1

+

β

f

β

(c(t))(ψ

αβ

(c(t), −)

∗

g

2

)(

d

dt

γ(t),

d

dt

γ(t)) dt ≥

≥

_

t

0

_

f

α

(c(t)) [

d

dt

γ(t)[

g

2

dt ≥

√

m+ 2

_

t

0

[

d

dt

γ(t)[

g

2

dt.

So g

2

-lenght(γ) is ﬁnite and since the Riemannian metric g

2

on S is

complete, lim

t→t

γ(t) =: γ(t

**) exists in S and the integral curve γ can
**

be continued.

44

10. Principal Fiber Bundles and G-Bundles

10.1. Deﬁnition. Let G be a Lie group and let (E, p, M, S) be a ﬁber

bundle as in 9.1. A G-bundle structure on the ﬁber bundle consists of

the following data:

(1) A left action : G S → S of the Lie group on the standard

ﬁber.

(2) A ﬁber bundle atlas (U

α

, ψ

α

) whose transition functions (ψ

αβ

)

act on S via the G-action: There is a family of smooth map-

pings (ϕ

αβ

: U

αβ

→ G) which satisﬁes the cocycle condition

ϕ

αβ

(x)ϕ

βγ

(x) = ϕ

αγ

(x) for x ∈ U

αβγ

and ϕ

αα

(x) = e, the unit

in the group, such that ψ

αβ

(x, s) = (ϕ

αβ

(x), s) = ϕ

αβ

(x).s.

A ﬁber bundle with a G-bundle structure is called a G-bundle. A ﬁber

bundle atlas as in (2) is called a G-atlas and the family (ϕ

αβ

) is also

called a cocycle of transition functions, but now for the G-bundle.

To be more precise, two G-atlases are said to be equivalent (to de-

scribe the same G-bundle), if their union is also a G-atlas. This trans-

lates as follows to the two cocycles of transition functions, where we

assume that the two coverings of M are the same (by passing to the

common reﬁnement, if necessary): (ϕ

αβ

) and (ϕ

αβ

) are called coho-

mologous if there is a family (τ

α

: U

α

→ G) such that ϕ

αβ

(x) =

τ

α

(x)

−1

.ϕ

αβ

(x).τ

β

(x) holds for all x ∈ U

αβ

.

In (2) one should specify only an equivalence class of G-bundle struc-

tures or only a cohomology class of cocycles of G-valued transition func-

tions. From any open cover (U

α

) of M, some cocycle of transition func-

tions (ϕ

αβ

: U

αβ

→ G) for it, and a left G-action on a manifold S,

we may construct a G-bundle by gluing, which depends only on the

cohomology class of the cocycle. By some abuse of notation we write

(E, p, M, S, G) for a ﬁber bundle with speciﬁed G-bundle structure.

Examples: The tangent bundle of a manifold M is a ﬁber bundle

with structure group GL(n). More general a vector bundle (E, p, M, V )

is a ﬁber bundle with standard ﬁber the vector space V and with GL(V )-

structure.

10.2. Deﬁnition. A principal (ﬁber) bundle (P, p, M, G) is a G-bundle

with typical ﬁber a Lie group G, where the left action of G on G is just

the left translation.

So by 10.1 we are given a bundle atlas (U

α

, ϕ

α

: P[U

α

→U

α

G) such

that we have ϕ

α

ϕ

−1

β

(x, a) = (x, ϕ

αβ

(x).a) for the cocycle of transition

functions (ϕ

αβ

: U

αβ

→G). This is now called a principal bundle atlas.

Clearly the principal bundle is uniquely speciﬁed by the cohomology

class of its cocycle of transition functions.

10. Principal Fiber Bundles and G-Bundles 45

Each principal bundle admits a unique right action r : P G → P,

called the principal right action, given by ϕ

α

(r(ϕ

−1

α

(x, a), g)) = (x, ag).

Since left and right translation on G commute, this is well deﬁned. As

in 1.3 we write r(u, g) = u.g when the meaning is clear. The principal

right action is visibly free and for any u

x

∈ P

x

the partial mapping

r

u

x

= r(u

x

, ) : G → P

x

is a diﬀeomorphism onto the ﬁber through

u

x

, whose inverse is denoted by τ

u

x

: P

x

→ G. These inverses together

give a smooth mapping τ : P

M

P → G, whose local expression is

τ(ϕ

−1

α

(x, a), ϕ

−1

α

(x, b)) = a

−1

.b. This mapping is also uniquely deter-

mined by the implicit equation r(u

x

, τ(u

x

, v

x

)) = v

x

, thus we also have

τ(u

x

.g, u

x

.g

) = g

−1

.τ(u

x

, u

x

).g

and τ(u

x

, u

x

) = e.

10.3. Lemma. Let p : P → M be a surjective submersion (a ﬁbred

manifold), and let G be a Lie group which acts freely on P such that the

orbits of the action are exactly the ﬁbers p

−1

(x) of p. Then (P, p, M, G)

is a principal ﬁber bundle.

Proof. Let the action be a right one by using the group inversion if nec-

essary. Let s

α

: U

α

→P be local sections (right inverses) for p : P →M

such that (U

α

) is an open cover of M. Let ϕ

−1

α

: U

α

G → P[U

α

be given by ϕ

−1

α

(x, a) = s

α

(x).a, which is obviously injective with in-

vertible tangent mapping, so its inverse ϕ

α

: P[U

α

→ U

α

G is a

ﬁber respecting diﬀeomorphism. So (U

α

, ϕ

α

) is already a ﬁber bun-

dle atlas. Let τ : P

M

P → G be given by the implicit equation

r(u

x

, τ(u

x

, u

x

)) = u

x

, where r is the right G-action. τ is smooth by

the implicit function theorem and clearly τ(u

x

, u

x

.g) = τ(u

x

, u

x

).g.

Thus we have ϕ

α

ϕ

−1

β

(x, g) = ϕ

α

(s

β

(x).g) = (x, τ(s

α

(x), s

β

(x).g)) =

(x, τ(s

α

(x), s

β

(x)).g) and (U

α

, ϕ

α

) is a principal bundle atlas.

10.4. Remarks. In the proof of Lemma 10.3 we have seen, that a

principal bundle atlas of a principal ﬁber bundle (P, p, M, G) is already

determined if we specify a family of smooth sections of P, whose domains

of deﬁnition cover the base M.

Lemma 10.3 can serve as an equivalent deﬁnition for a principal bun-

dle. But this is true only if an implicit function theorem is available, so

in topology or in inﬁnite dimensional diﬀerential geometry one should

stick to our original deﬁnition.

From the Lemma itself it follows, that the pullback f

∗

P over a smooth

mapping f : M

**→M is again a principal ﬁber bundle.
**

10.5. Homogeneous spaces. Let G be a Lie group with Lie algebra g.

Let K be a closed subgroup of G, then by a well known theorem K is a

closed Lie subgroup whose Lie algebra will be denoted by /. There is a

46 10. Principal Fiber Bundles and G-Bundles

unique structure of a smooth manifold on the quotient space G/K such

that the projection p : G → G/K is a submersion, so by the implicit

function theorem p admits local sections.

Theorem. (G, p, G/K, K) is a principal ﬁber bundle.

Proof. The group multiplication of G restricts to a free right action

µ : GK →G, whose orbits are exactly the ﬁbers of p. By lemma 10.3

the result follows.

For the convenience of the reader we discuss now the best know ho-

mogeneous spaces.

The group SO(n) acts transitively on S

n−1

⊂ R

n

. The isotropy

group of the ”north pole” (0, . . . , 0, 1) is the subgroup

_

1 0

0 SO(n −1)

_

which we identify with SO(n − 1). So S

n−1

= SO(n)/SO(n − 1) and

(SO(n), p, S

n−1

, SO(n −1)) is a principal ﬁber bundle. Likewise

(O(n), p, S

n−1

, O(n −1)),

(SU(n), p, S

2n−1

, SU(n −1)),

(U(n), p, S

2n−1

, U(n −1)), and

(Sp(n), p, S

4n−1

, Sp(n −1)) are principal ﬁber bundles.

The Grassmann manifold G(k, n; R) is the space of all k-planes con-

taining 0 in R

n

. The group O(n) acts transitively on it and the isotropy

group of the k-plane R

k

¦0¦ is the subgroup

_

O(k) 0

0 O(n −k)

_

,

therefore G(k, n; R) = O(n)/O(k) O(n−k) is a compact manifold and

we get the principal ﬁber bundle (O(n), p, G(k, n; R), O(k) O(n −k)).

Likewise

(SO(n), p,

˜

G(k, n, R), SO(k) SO(n −k)),

(U(n), p, G(k, n, C), U(k) U(n −k)), and

(Sp(n), p, G(k, n, H), Sp(k) Sp(n −k)) are principal ﬁber bundles.

The Stiefel manifold V (k, n, R) is the space of all orthonormal k-

frames in R

n

. Clearly the group O(n) acts transitively on V (k, n, R)

and the isotropy subgroup of (e

1

, . . . , e

k

) is 1O(n−k), so V (k, n, R) =

O(n)/O(n −k) is a compact manifold and we get a principal ﬁber bun-

dle (O(n), p, V (k, n, R), O(n − k)) But O(k) also acts from the right

on V (k, n, R), its orbits are exactly the ﬁbers of the projection p :

V (k, n, R) → G(k, n, R). So by lemma 10.3 we get a principal ﬁber

10. Principal Fiber Bundles and G-Bundles 47

bundle (V (k, n, R), p, G(k, n, R), O(k)). Indeed we have the following

diagram where all arrows are projections of principal ﬁber bundles, and

where the respective structure groups are written on the arrows:

(a)

O(n)

O(n−k)

−−−−−→ V (k, n, R)

O(k)

¸

¸

_

O(k)

¸

¸

_

V (n −k, n, R) −−−−−→

O(n−k)

G(k, n, R),

It is easy to see that V (k, n) is also diﬀeomorphic to the space ¦ A ∈

L(R

k

, R

n

) : A

t

.A = 1

k

¦, i.e. the space of all linear isometries R

k

→

R

n

. There are furthermore complex and quaternionic versions of the

Stiefel manifolds.

More examples will be given in sections on jets below.

10.6. Homomorphisms. Let χ : (P, p, M, G) → (P

, p

, M

, G) be

a principal ﬁber bundle homomorphism, i.e. a smooth G-equivariant

mapping χ : P →P

**. Then obviously the diagram
**

(a)

P

χ

−−−−→ P

p

¸

¸

_ p

¸

¸

_

M −−−−→

¯ χ

M

commutes for a uniquely determined smooth mapping ¯ χ : M →M

. For

each x ∈ M the mapping χ

x

:= χ[P

x

: P

x

→P

¯ χ(x)

is G-equivariant and

therefore a diﬀeomorphism, so diagram (a) is a pullback diagram.

But the most general notion of a homomorphism of principal bundles

is the following. Let Φ : G → G

**be a homomorphism of Lie groups.
**

χ : (P, p, M, G) → (P

, p

, M

, G

**) is called a homomorphism over Φ of
**

principal bundles, if χ : P →P

**is smooth and χ(u.g) = χ(u).Φ(g) holds
**

in general. Then χ is ﬁber respecting, so diagram (a) makes again sense,

but it is no longer a pullback diagram in general.

If χ covers the identity on the base, it is called a reduction of the

structure group G

to G for the principal bundle (P

, p

, M

, G

) — the

name comes from the case, when Φ is the embedding of a subgroup.

By the universal property of the pullback any general homomorphism

χ of principal ﬁber bundles over a group homomorphism can be written

as the composition of a reduction of structure groups and a pullback

48 10. Principal Fiber Bundles and G-Bundles

homomorphism as follows, where we also indicate the structure groups:

(b)

(P, G) −−−−→ ( ¯ χ

∗

P

, G

) −−−−→ (P

, G

)

p

¸

¸

_

¸

¸

_ p

¸

¸

_

M M −−−−→

¯ χ

M

**10.7. Associated bundles. Let (P, p, M, G) be a principal bundle and
**

let : GS →S be a left action of the structure group G on a manifold

S. We consider the right action R : (P S) G → P S, given by

R((u, s), g) = (u.g, g

−1

.s).

Theorem. In this situation we have:

(1) The space P

G

S of orbits of the action R carries a unique

smooth manifold structure such that the quotient map q : PS →

P

G

S is a submersion.

(2) (P

G

S, ¯ p, M, S, G) is a G-bundle, where ¯ p : P

G

S → M is

given by

(a)

P S

q

−−−−→ P

G

S

pr

1

¸

¸

_

¯ p

¸

¸

_

P −−−−→

p

M.

In this diagram q

u

: ¦u¦S →(P

G

S)

p(u)

is a diﬀeomorphism

for each u ∈ P.

(3) (P S, q, P

G

S, G) is a principal ﬁber bundle with principal

action R.

(4) If (U

α

, ϕ

α

: P[U

α

→ U

α

G) is a principal bundle atlas with

cocycle of transition functions (ϕ

αβ

: U

αβ

→ G), then together

with the left action : GS →S this cocycle is also one for the

G-bundle (P

G

S, ¯ p, M, S, G).

Notation: (P

G

S, ¯ p, M, S, G) is called the associated bundle for the

action : GS →S. We will also denote it by P[S, ] or simply P[S] and

we will write p for ¯ p if no confusion is possible. We also deﬁne the smooth

mapping τ : P

M

P[S, ] → S by τ(u

x

, v

x

) := q

−1

u

x

(v

x

). It satisﬁes

τ(u, q(u, s)) = s, q(u

x

, τ(u

x

, v

x

)) = v

x

, and τ(u

x

.g, v

x

) = g

−1

.τ(u

x

, v

x

).

In the special situation, where S = G and the action is left translation,

so that P[G] = P, this mapping coincides with τ considered in 10.2.

Proof. In the setting of the diagram in (2) the mapping p◦pr

1

is constant

on the R-orbits, so ¯ p exists as a mapping. Let (U

α

, ϕ

α

: P[U

α

→U

α

G)

10. Principal Fiber Bundles and G-Bundles 49

be a principal bundle atlas with transition functions (ϕ

αβ

: U

αβ

→

G). We deﬁne ψ

−1

α

: U

α

S → ¯ p

−1

(U

α

) ⊂ P

G

S by ψ

−1

α

(x, s) =

q(ϕ

−1

α

(x, e), s), which is ﬁber respecting. For each orbit in ¯ p

−1

(x) ⊂

P

G

S there is exactly one s ∈ S such that this orbit passes through

(ϕ

−1

α

(x, e), s), since the principal right action is free. Thus ψ

−1

α

(x. ) :

S → ¯ p

−1

(x) is bijective. Furthermore

ψ

−1

β

(x, s) = q(ϕ

−1

β

(x, e), s) =

= q(ϕ

−1

α

(x, ϕ

αβ

(x).e), s) = q(ϕ

−1

α

(x, e).ϕ

αβ

(x), s) =

= q(ϕ

−1

α

(x, e), ϕ

αβ

(x).s) = ψ

−1

α

(x, ϕ

αβ

(x).s),

so ψ

α

ψ

−1

β

(x, s) = (x, ϕ

αβ

(x).s) So (U

α

, ψ

α

) is a G-atlas for P

G

S and

makes it into a smooth manifold and a G-bundle. The deﬁning equation

for ψ

α

shows that q is smooth and a submersion and consequently the

smooth structure on P

G

S is uniquely deﬁned, and ¯ p is smooth by the

universal properties of a submersion.

By the deﬁnition of ψ

α

the diagram

(b)

p

−1

(U

α

) S

ϕ

α

×Id

−−−−→ U

α

GS

q

¸

¸

_

Id×

¸

¸

_

¯ p

−1

(U

α

) U

α

S

commutes; since its lines are diﬀeomorphisms we conclude that q

u

: ¦u¦

S → ¯ p

−1

(p(u)) is a diﬀeomorphism. So (1), (2), and (4) are checked.

(3) We rewrite the last diagram in the following form:

(c)

p

−1

(U

α

) S q

−1

(V

α

)

λ

α

−−−−→ V

α

G

q

¸

¸

_

pr

1

¸

¸

_

¯ p

−1

(U

α

) V

α

Here V

α

:= ¯ p

−1

(U

α

) ⊂ P

G

S and the diﬀeomorphism λ

α

is deﬁned

by λ

−1

α

(ψ

−1

α

(x, s), g) := (ϕ

α

(x, g), g

−1

.s). Then we have

λ

−1

β

(ψ

−1

α

(x, s), g) = λ

−1

β

(ψ

−1

β

(x, , ϕ

βα

(x).s), g) =

= (ϕ

−1

β

(x, g), g

−1

.ϕ

βα

(x).s) = (ϕ

−1

α

(x, ϕ

αβ

(x).g), g

−1

.ϕ

αβ

(x)

−1

.s) =

= (λ

−1

α

(ψ

−1

α

(x, s), ϕ

αβ

(x).g),

so λ

α

λ

−1

β

(ψ

−1

α

(x, s), g) = (ψ

−1

α

(x, s), ϕ

αβ

(x).g) and we conclude that

(P S, q, P

G

S, G) is a principal bundle with structure group G and

the same cocycle (ϕ

αβ

) we started with.

50 10. Principal Fiber Bundles and G-Bundles

10.8. Corollary. Let (E, p, M, S, G) be a G-bundle, speciﬁed by a co-

cycle of transition functions (ϕ

αβ

) with values in G and a left action

of G on S. Then from the cocycle of transition functions we may glue a

unique principal bundle (P, p, M, G) such that E = P[S, ].

This is the usual way a diﬀerential geometer thinks of an associated

bundle. He is given a bundle E, a principal bundle P, and the G-bundle

structure then is described with the help of the mappings τ and q.

10.9. Equivariant mappings and associated bundles.

1. Let (P, p, M, G) be a principal ﬁber bundle and consider two left

actions of G, : G S → S and

: G S

→ S

. Let furthermore

f : S → S

**be a G-equivariant smooth mapping, so f(g.s) = g.f(s) or
**

f ◦

g

=

g

◦ f. Then Id

P

f : P S → P S

**is equivariant for the
**

actions R : (P S) G → P S and R

: (P S

) G → P S

, so

there is an induced mapping

(a)

P S

Id×f

−−−−→ P S

q

¸

¸

_ q

¸

¸

_

P

G

S

Id×

G

f

−−−−→ P

G

S

,

which is ﬁber respecting over M, and a homomorphism of G-bundles in

the sense of the deﬁnition 10.10 below.

2. Let χ : (P, p, M, G) → (P

, p

, M

**, G) be a principal ﬁber bundle
**

homomorphism as in 10.6. Furthermore we consider a smooth left action

: G S → S. Then χ Id

S

: P S → P

S is G-equivariant and

induces a mapping χ

G

Id

S

: P

G

S →P

G

S, which is ﬁber respecting

over M, ﬁber wise a diﬀeomorphism, and again a homomorphism of G-

bundles in the sense of deﬁnition 10.10 below.

3. Now we consider the situation of 1 and 2 at the same time. We

have two associated bundles P[S, ] and P

[S

,

]. Let χ : (P, p, M, G) →

(P

, p

, M

**, G) be a principal ﬁber bundle homomorphism and let f :
**

S →S

be an G-equivariant mapping. Then χ f : P S →P

S

is

clearly G-equivariant and therefore induces a mapping χ

G

f : P[S, ] →

P

[S

,

**] which again is a homomorphism of G-bundles.
**

4. Let S be a point. Then P[S] = P

G

S = M. Furthermore let

y ∈ S

be a ﬁxed point of the action

: GS

→S

**, then the inclusion
**

i : ¦y¦ → S

is G-equivariant, thus Id

P

i induces Id

P

G

i : M =

P[¦y¦] →P[S

], which is a global section of the associated bundle P[S

].

If the action of G on S is trivial, so g.s = s for all s ∈ S, then the

associated bundle is trivial: P[S] = M S. For a trivial principal ﬁber

bundle any associated bundle is trivial.

10. Principal Fiber Bundles and G-Bundles 51

10.10. Deﬁnition. In the situation of 10.8, a smooth ﬁber respecting

mapping γ : P[S, ] →P

[S

,

] covering a smooth mapping ¯ γ : M →M

**of the bases is called a homomorphism of G-bundles, if the following con-
**

ditions are satisﬁed: P is isomorphic to the pullback ¯ γ

∗

P

**, and the local
**

representations of γ in pullback-related ﬁber bundle atlases belonging to

the two G-bundles are ﬁber wise G-equivariant.

Let us describe this in more detail now. Let (U

α

, ψ

α

) be a G-atlas

for P

[S

,

] with cocycle of transition functions (ϕ

αβ

), belonging to

the principal ﬁber bundle atlas (U

α

, ϕ

α) of (P

, p

, M

, G). Then the

pullback-related principal ﬁber bundle atlas (U

α

= ¯ γ

−1

(U

α

), γ

α

) for

P = ¯ γ

∗

P

**as described in the proof of 9.5 has the cocycle of transition
**

functions (ϕ

αβ

= ϕ

αβ

◦ ¯ γ); it induces the G-atlas (U

α

, ψ

α

) for P[S, ].

Then (ψ

α

◦γ◦ψ

−1

α

)(x, s) = (¯ γ(x), γ

α

(x, s)) and γ

α

(x, ) : S →S

should

be G-equivariant for all α and all x ∈ U

α

.

Lemma. Let γ : P[S, ] → P

[S

,

] be a homomorphism of associated

bundles and G-bundles. Then there is a principal bundle homomorphism

χ : (P, p, M, G) →(P

, p

, M

**, G) and a G-equivariant mapping f : S →
**

S

such that γ = χ

G

f : P[S, ] →P

[S

,

].

Proof. The homomorphism χ : (P, p, M, G) → (P

, p

, M

, G) of princi-

pal ﬁber bundles is already determined by the requirement that P =

¯ γ

∗

P

, and we have ¯ γ = ¯ χ. The G-equivariant mapping f : S → S

can

be read oﬀ the following diagram

(a)

P

M

P[S]

τ

−−−−→ S

χ×

M

γ

¸

¸

_

f

¸

¸

_

P

M

P

[S

] −−−−→

τ

S

,

which by the assumptions is seen to be well deﬁned in the right col-

umn.

So a homomorphism of associated bundles is described by the whole

triple (χ : P → P

, f : S → S

(G-equivariant) , γ : E → E

), such that

diagram (a) commutes.

10.11. Associated vector bundles. Let (P, p, M, G) be a principal

ﬁber bundle, and consider a representation ρ : G → GL(V ) of G on a

ﬁnite dimensional vector space V . Then P[V, ρ] is an associated ﬁber

bundle with structure group G, but also with structure group GL(V ), for

in the canonically associated ﬁber bundle atlas the transition functions

have also values in GL(V ). So by section 6 P[V, ρ] is a vector bundle.

52 10. Principal Fiber Bundles and G-Bundles

If (E, p, M) is a vector bundle with n-dimensional ﬁbers we may con-

sider the open subset GL(R

n

, E) ⊂ L(M R

n

, E), consisting of all

invertible linear mappings, which is a ﬁber bundle over the base M.

Composition from the right by elements of GL(n) gives a free right ac-

tion on GL(R

n

, E) whose orbits are exactly the ﬁbers, so by lemma 10.3

we have a principal ﬁber bundle (GL(R

n

, E), p, M, GL(n)). The asso-

ciated bundle GL(R

n

, E)[R

n

] for the banal representation of GL(n) on

R

n

is isomorphic to the vector bundle (E, p, M) we started with, for

the evaluation mapping ev : GL(R

n

, E) R

n

→ E is invariant un-

der the right action R of GL(n), and locally in the image there are

smooth sections to it, so it factors to a ﬁber linear diﬀeomorphism

GL(R

n

, E)[R

n

] = GL(R

n

, E)

GL(n)

R

n

→ E. The principal bun-

dle GL(R

n

, E) is called the linear frame bundle of E. Note that local

sections of GL(R

n

) are exactly the local frame ﬁelds of the vector bundle

E.

To illustrate the notion of reduction of structure group, we consider

now a vector bundle (E, p, M, R

n

) equipped with a Riemannian met-

ric g, that is a section g ∈ C

∞

(S

2

E

∗

) such that g

x

is a positive def-

inite inner product on E

x

for each x ∈ M. Any vector bundle ad-

mits Riemannian metrics: local existence is clear and we may glue

with the help of a partition of unity on M, since the positive deﬁ-

nite sections form an open convex subset. Now let s

= (s

1

, . . . , s

n

) ∈

C

∞

(GL(R

n

, E)[U) be a local frame ﬁeld of the bundle E over U ⊂ M.

Now we may apply the Gram-Schmidt orthonormalization procedure to

the basis (s

1

(x), . . . , s

n

(x)) of E

x

for each x ∈ U. Since this proce-

dure is real analytic, we obtain a frame ﬁeld s = (s

1

, . . . , s

n

) of E over

U which is orthonormal with respect to g. We call it an orthonormal

frame ﬁeld. Now let (U

α

) be an open cover of M with orthonormal frame

ﬁelds s

α

= (s

α

1

, . . . , s

α

n

), where s

α

is deﬁned on U

α

. We consider the vec-

tor bundle charts (U

α

, ψ

α

: E[U

α

→ U

α

R

n

) given the orthonormal

frame ﬁelds: ψ

−1

α

(x, v

1

, . . . , v

n

) =

s

α

i

(x).v

i

=: s

α

(x).v. For x ∈ U

αβ

we have s

α

i

(x) =

s

β

j

(x).g

βα

j

i

(x) for C

∞

-functions g

αβ

: U

αβ

→ R.

Since s

α

(x) and s

β

(x) are both orthonormal bases of E

x

, the matrix

g

αβ

(x) = (g

αβ

j

i

(x)) is an element of O(n, R). We write s

α

= s

β

.g

βα

for short. Then we have ψ

−1

β

(x, v) = s

β

(x).v = s

α

(x).g

αβ

(x).v =

ψ

−1

α

(x, g

αβ

(x).v) and consequently ψ

α

ψ

−1

β

(x, v) = (x, g

αβ

(x).v). So the

(g

αβ

: U

αβ

→O(n, R)) are the cocycle of transition functions for the vec-

tor bundle atlas (U

α

, ψ

α

). So we have constructed an O(n, R)-structure

on E. The corresponding principal ﬁber bundle will be denoted by

O(R

n

, (E, g)); it is usually called the orthonormal frame bundle of E. It

is derived from the linear frame bundle GL(R

n

, E) by reduction of the

10. Principal Fiber Bundles and G-Bundles 53

structure group from GL(n) to O(n). The phenomenon discussed here

plays a prominent role in the theory of classifying spaces.

10.12. Sections of associated bundles. Let (P, p, M, G) be a princi-

pal ﬁber bundle and : GS →S a left action. Let C

∞

(P, S)

G

denote

the space of all smooth mappings f : P →S which are G-equivariant in

the sense that f(u.g) = g

−1

.f(u) holds for g ∈ G and u ∈ P.

Theorem. The sections of the associated bundle P[S, ] correspond ex-

actly to the G-equivariant mappings P →S; we have a bijection

C

∞

(P, S)

G

∼

= C

∞

(P[S]).

Proof. If f ∈ C

∞

(P, S)

G

we construct s

f

∈ C

∞

(P[S]) in the follow-

ing way: graph(f) = (Id, f) : P → P S is G-equivariant, since

(Id, f)(u.g) = (u.g, f(u.g)) = (u.g, g

−1

.f(u)) = ((Id, f)(u)).g. So it

induces a smooth section s

f

∈ C

∞

(P[S]) as seen from

(a)

P

(Id,f)

−−−−→ P S

p

¸

¸

_

q

¸

¸

_

M −−−−→

f

s

P[S]

If conversely s ∈ C

∞

(P[S]) we deﬁne f

s

∈ C

∞

(P, S)

G

by f

s

:=

τ ◦(Id

P

M

f) : P = P

M

M →P

m

P[S] →S. This is G-equivariant

since f

s

(u

x

.g) = τ(u

x

.g, f(x)) = g

−1

.τ(u

x

, f(x)) = g

−1

.f

s

(u

x

) by 10.4.

The two constructions are inverse to each other since we have f

s(f)

(u) =

τ(u, s

f

(p(u))) = τ(u, q(u, f(u))) = f(u) and s

s(f)

(p(u)) = q(u, f

s

(u)) =

q(u, τ(u, f(p(u))) = f(p(u)).

10.13. Theorem. Consider a principal ﬁber bundle (P, p, M, G) and

a closed subgroup K of G. Then the reductions of structure group from

G to K correspond bijectively to the global sections of the associated

bundle P[G/K,

¯

λ], where

¯

λ : GG/K →G/K is the left action on the

homogeneous space.

Proof. By theorem 10.12 the section s ∈ C

∞

(P[G/K]) corresponds to

f

s

∈ C

∞

(P, G/K)

G

, which is a surjective submersion since the action

¯

λ : GG/K →G/K is transitive. Thus P

s

:= f

−1

s

(¯ e) is a submanifold

of P which is stable under the right action of K on P. Furthermore

the K-orbits are exactly the ﬁbers of the mapping p : P

s

→ M, so by

lemma 10.3 we get a principal ﬁber bundle (P

s

, p, M, K). The embedding

P

s

→P is then a reduction of structure groups as required.

54 10. Principal Fiber Bundles and G-Bundles

If conversely we have a principal ﬁber bundle (P

, p

, M, K) and a

reduction of structure groups χ : P

→ P, then χ is an embedding

covering the identity of M and is K-equivariant, so we may view P

**as a sub ﬁber bundle of P which is stable under the right action of K.
**

Now we consider the mapping τ : P

M

P → G from 10.2 and restrict

it to P

M

P

. Since we have τ(u

x

, v

x

.k) = τ(u

x

, v

x

).k for k ∈ K this

restriction induces f : P →G/K by

P

M

P

τ

−−−−→ G

¸

¸

_

p

¸

¸

_

P = P

M

P

/K −−−−→

f

G/K;

and from τ(u

x

.g, v

x

) = g

−1

.τ(u

x

, v

x

) it follows that f is G-equivariant

as required. Finally f

−1

(¯ e) = ¦u ∈ P : τ(u, P

p(u)

) ⊆ K¦ = P

, so the

two constructions are inverse to each other.

10.14. The bundle of gauges. If (P, p, M, G) is a principal ﬁber bun-

dle we denote by Aut(P) the group of all G-equivariant diﬀeomorphisms

χ : P →P. Then p ◦ χ = ¯ χ ◦ p for a unique diﬀeomorphism ¯ χ of M, so

there is a group homomorphism from Aut(P) into the group Diﬀ(M) of

all diﬀeomorphisms of M. The kernel of this homomorphism is called

Gau(P), the group of gauge transformations. So Gau(P) is the space of

all χ : P →P which satisfy p ◦ χ = p and χ(u.g) = χ(u).g.

Theorem. The group Gau(P) of gauge transformations is equal to the

space C

∞

(P, (G, conj))

G ∼

= C

∞

(P[G, conj]).

Proof. We use again the mapping τ : P

M

P → G from 10.2. For

χ ∈ Gau(P) we deﬁne f

χ

∈ C

∞

(P, (G, conj))

G

by f

χ

:= τ ◦ (Id, χ).

Then f

χ

(u.g) = τ(u.g, χ(u.g)) = g

−1

.τ(u, χ(u)).g = conj

g

−1f

χ

(u), so f

χ

is indeed G-equivariant.

If conversely f ∈ C

∞

(P, (G, conj))

G

is given, we deﬁne χ

f

: P → P

by χ

f

(u) := u.f(u). It is easy to check that χ

f

is indeed in Gau(P) and

that the two constructions are inverse to each other.

10.15 The tangent bundles of homogeneous spaces. Let G be

a Lie group and K a closed subgroup, with Lie algebras g and /, re-

spectively. We recall the mapping Ad

G

: G → Aut

Lie

(g) from 1.1 and

put Ad

G,K

:= Ad

G

[K : K → Aut

Lie

(g). For X ∈ / and k ∈ K we

have Ad

G,K

(k)X = Ad

G

(k)X = Ad

K

(k)X ∈ /, so / is an invariant

10. Principal Fiber Bundles and G-Bundles 55

subspace for the representation Ad

G,K

of K in g,and we have the factor

representation Ad

⊥

: K →GL(g//. Then

(a) 0 →/ →g →g// →0

is short exact and K-equivariant.

Now we consider the principal ﬁber bundle (G, p, G/K, K) and the

associated vector bundles G[g//, Ad

⊥

] and G[/, Ad

G,K

].

Theorem. In these circumstances we have

T(G/K) = G[g//, Ad

⊥

] = (G

K

(//, p, G/K, g//).

The left action g →T(

¯

λ

g

) of G on T(G/K) coincides with the left action

of G on G

K

g//. Furthermore G[g//, Ad

⊥

] ⊕G[/, Ad

G,K

] is a trivial

vector bundle.

Proof. For p : G → G/K we consider the tangent mapping T

e

p : g →

T

¯ e

(G/K) which is linear and surjective and induces a linear isomorphism

T

e

p : g// → T

¯ e

(G/K). For k ∈ K we have p ◦ conj

k

= p ◦ λ

k

◦ ρ

k

−1 =

¯

λ

k

◦ p and consequently T

e

p ◦ Ad

G,K

(k) = T

e

p ◦ T

e

(conj

k

) = T

¯ e

¯

λ

k

◦ T

e

p.

Thus the isomorphism T

e

p : g// → T

¯ e

(G/K) is K-equivariant for the

representations Ad

⊥

and T

¯ e

¯

λ : k →T

¯ e

¯

λ

k

.

Now we consider the associated vector bundle G[T

¯ e

(G/K), T

¯ e

¯

λ] =

(G

K

T

¯ e

(G/K), p, G/K, T

¯ e

(G/K)), which is isomorphic to the vec-

tor bundle G[g//, Ad

⊥

], since the representation spaces are isomorphic.

The mapping T

¯

λ : GT

¯ e

(G/K) →T(G/K) is K-invariant and therefore

induces a mapping ψ as in the following diagram:

(b)

GT

¯ e

(G/K)

T

¯

λ

−−−−→ T(G/K)

q

¸

¸

_

_

_

_

G

K

T

¯ e

(G/K) −−−−→

ψ

T(G/K)

p

¸

¸

_

π

G/K

¸

¸

_

G/K G/K

This mapping ψ is an isomorphism of vector bundles.

It remains to show the last assertion. The short exact sequence (a)

induces a sequence of vector bundles over G/K:

G/K 0 →G[/, Ad

K

] →G[(, Ad

G,K

] →G[g//, Ad

⊥

] →G/K 0

This sequence splits ﬁber wise thus also locally over G/K, so we obtain

G[g//, Ad

⊥

] ⊕G[/, Ad

G,K

]

∼

= G[g, Ad

G,K

] and it remains to show that

G[g, Ad

G,K

] is a trivial vector bundle. Let ϕ : G g → G g be given

56 10. Principal Fiber Bundles and G-Bundles

by ϕ(g, X) = (g, Ad

G

(g)X). Then for k ∈ K we have ϕ((g, X).k) =

ϕ(gk, Ad

G,K

(g

−1

)X) = (ak, Ad

G

(g.k.k

−1

)X) = (gk, Ad

G

(g)X). So ϕ is

K-equivariant from the ”joint” K-action to the ”on the left” K-action

and therefore induces a mapping ¯ ϕ as in the diagram:

(c)

Gg

ϕ

−−−−→ Gg

q

¸

¸

_

¸

¸

_

G

K

g −−−−→

¯ ϕ

G/K g

p

¸

¸

_

pr

1

¸

¸

_

G/K G/K

The map ¯ ϕ is a vector bundle isomorphism.

10.16 Tangent bundles of Grassmann manifolds. From 10.5 we

know that (V (k, n) = O(n)/O(n − k), p, G(k, n), O(k)) is a principal

ﬁber bundle. Using the banal representation of O(k) we consider the

associated vector bundle (E

k

:= V (k, n)[R

k

], p, G(k, n)). It is called the

universal vector bundle over G(k, n) for reasons we will discuss below in

chapter 11. Recall from 10.5 the description of V (k, n) as the space of

all linear isometries R

k

→ R

n

; we get from it the evaluation mapping

ev : V (k, n) R

k

→R

n

. The mapping (p, ev) in the diagram

(a)

V (k, n) R

k

(p,ev)

−−−−→ G(kn) R

n

q

¸

¸

_

_

_

_

V (k, n)

O(k)

R

k

−−−−→

ψ

G(kn) R

n

is O(k)-invariant for the action R and factors therefore to an embedding

of vector bundles ψ : E

k

→ G(k, n) R

n

. So the ﬁber (E

k

)

W

over

the k-plane W in R

n

is just the linear subspace W. Note ﬁnally that

the ﬁber wise orthogonal complement E

k

⊥

of E

k

in the trivial vector

bundle G(k, n) R

n

with its standard Riemannian metric is isomorphic

to the universal vector bundle E

n−k

over G(n−k, n), where the isomor-

phism covers the diﬀeomorphism G(k, n) → G(n − k, n) given also by

the orthogonal complement mapping.

Corollary. The tangent bundle of the Grassmann manifold is

TG(k, n) = L(E

k

, E

k

⊥

).

10. Principal Fiber Bundles and G-Bundles 57

Proof. We have G(k, n) = O(n)/(O(k) O(n−k)), so by theorem 10.15

we get

TG(k, n) = O(n)

O(k)×O(n−k)

so(n)/(so(k) so(n −k)).

On the other hand we have V (k, n) = O(n)/O(n − k) and the right

action of O(k) commutes with the right action of O(n − k) on O(n),

therefore

V (k, n)[R

k

] = O(n)/O(n −k)

O(k)

R

k

= O(n)

O(k)×O(n−k)

R

k

,

where O(n −k) acts trivially on R

k

. Finally

L(E

k

, E

k

⊥

) = L

_

O(n)

O(k)×O(n−k)

R

k

, O(n)

O(k)×O(n−k)

R

n−k

_

= O(n)

O(k)×O(n−k)

L(R

k

, R

n−k

),

where the left action of O(k) O(n − k) on L(R

k

, R

n−k

) is given by

(A, B)(C) = B.C.A

−1

. Finally we have an O(k)O(n−k) - equivariant

linear isomorphism L(R

k

, R

n−k

) →so(n)/(so(k)so(n−k)), as follows:

so(n)/(so(k) so(n −k)) =

_

skew

_

_

skew 0

0 skew

_ =

__

0 A

−A 0

_

: A ∈ L(R

k

, R

n−k

)

_

10.17. The tangent group of a Lie group. Let G be a Lie group

with Lie algebra g. We will use the notation from 1.1. First note that

TG is also a Lie group with multiplication Tµ and inversion Tν, given

by the expressions T

(a,b)

µ.(ξ

a

, η

b

) = T

a

(ρ

b

).ξ

a

+T

b

(λ

a

).η

b

and T

a

ν.ξ

a

=

−T

e

(λ

a

−1).T

a

(ρ

a

−1).ξ

a

.

Lemma. Via the isomomorphism Tρ : g G → TG, Tρ.(X, g) =

T

e

(ρ

g

).X, the group structure on TG looks as follows: (X, a).(Y, b) =

(X+Ad(a)Y, a.b) and (X, a)

−1

= (−Ad(a

−1

)X, a

−1

). So TG is isomor-

phic to the semidirect product gG.

Proof. T

(a,b)

µ.(Tρ

a

.X, Tρ

b

.Y ) = Tρ

b

.Tρ

a

.X +Tλ

a

.Tρ

b

.Y =

= Tρ

ab

.X +Tρ

b

.Tρ

a

.Tρ

a

−1.Tλ

a

.Y = Tρ

ab

(X + Ad(a)Y ).

T

a

ν.Tρ

a

.X = −Tρ

a

−1.Tλ

a

−1.Tρ

a

.X = −Tρ

a

−1.Ad(a

−1

)X.

58 10. Principal Fiber Bundles and G-Bundles

Remark. In the left trivialization Tλ : G g → TG, Tλ.(g, X) =

T

e

(λ

g

).X, the semidirect product structure looks somewhat awkward:

(a, X).(b, Y ) = (ab, Ad(b

−1

)X +Y ) and (a, X)

−1

= (a

−1

, −Ad(a)X).

10.18. Tangent and vertical bundles.

For a ﬁber bundle (E, p, M, S) the subbundle V E = ¦ ξ ∈ TE : Tp.ξ =

0 ¦ of TE is called the vertical bundle and is denoted by (V E, π

E

, E).

Theorem. Let (P, p, M, G) be a principal ﬁber bundle with principal

right action r : P G → P. Let : G S → S be a left action. Then

the following assertions hold:

(1) (TP, Tp, TM, TG) is again a principal ﬁber bundle with principal

right action Tr : TP TG →TP.

(2) The vertical bundle (V P, π, P, g) of the principal bundle is trivial

as a vector bundle over P: V P = P g.

(3) The vertical bundle of the principal bundle as bundle over M is

again a principal bundle: (V P, p ◦ π, M, TG).

(4) The tangent bundle of the associated bundle P[S, ] is given by

T(P[S, ]) = TP[TS, T].

(5) The vertical bundle of the associated bundle P[S, ] is given by

V (P[S, ]) = P[TS, T

2

] = P

G

TS.

Proof. Let (U

α

, ϕ

α

: P[U

α

→ U

α

G) be a principal ﬁber bundle atlas

with cocycle of transition functions (ϕ

αβ

: U

αβ

→ G). Since T is a

functor which respects products, (TU

α

, Tϕ

α

: TP[TU

α

→ TU

α

TG)

is again a principal ﬁber bundle atlas with cocycle of transition func-

tions (Tϕ

αβ

: TU

αβ

→ TG), describing the principal ﬁber bundle

(TP, Tp, TM, TG). The assertion about the principal action is obvious.

So (1) follows. For completeness sake we include here the transition

formula for this atlas in the right trivialization of TG:

T(ϕ

α

◦ ϕ

−1

β

)(ξ

x

, T

e

(ρ

g

).X) =

= (ξ

x

, T

e

(ρ

ϕ

αβ

(x).g).(δϕ

αβ

(ξ

x

) + Ad(ϕ

αβ

(x))X)),

where δϕ

αβ

∈ Ω

1

(U

αβ

; g) is the right logarithmic derivative of ϕ

αβ

which

is given by δϕ

αβ

(ξ

x

) = T(ρ

ϕ

αβ

(x)

−1).T(ϕ

αβ

).ξ

x

.

(2) The mapping (u, X) → T

e

(r

u

).X = T

(u,e)

r.(0

u

, X) is a vector

bundle isomorphism P g →V P over P.

(3) Obviously Tr : TP TG → TP is a free right action which acts

transitive on the ﬁbers of Tp : TP → TM. Since V P = (Tp)

−1

(0

M

),

the bundle V P →M is isomorphic to TP[0

M

and Tr restricts to a free

right action, which is transitive on the ﬁbers, so by lemma 10.3 the result

follows.

10. Principal Fiber Bundles and G-Bundles 59

(4) The transition functions of the ﬁber bundle P[S, ] are given by the

expression ◦ (ϕ

αβ

Id

S

) : U

αβ

S →GS →S. Then the transition

functions of TP[S, ] are T( ◦ (ϕ

αβ

Id

S

)) = T ◦ (Tϕ

αβ

Id

TS

) :

TU

αβ

TS →TGTS →TS, from which the result follows.

(5) Vertical vectors in TP[S, ] have local representations (0

x

, η

s

) ∈

TU

αβ

TS. Under the transition functions of TP[S, ] they trans-

form as T( ◦ (ϕ

αβ

Id

S

)).(0

x

, η

s

) = T.(0

ϕ

αβ

(x)

, η

s

) = T(

ϕ

αβ

(x)

).η

s

=

T

2

.(x, η

s

) and this implies the result

60

11. Principal and Induced Connections

11.1. Principal connections. Let (P, p, M, G) be a principal ﬁber

bundle. Recall from 9.3 that a (general) connection on P is a ﬁber

projection Φ : TP → V P, viewed as a 1-form in Ω

1

(P; TP). Such a

connection Φ is called a principal connection if it is G-equivariant for

the principal right action r : P G → P, so that T(r

g

).Φ = Φ.T(r

g

)

and Φ is r

g

-related to itself, or (r

g

)

∗

Φ = Φ in the sense of 8.16, for

all g ∈ G. By theorem 8.15.6 the curvature R =

1

2

.[Φ, Φ] is then also

r

g

-related to itself for all g ∈ G.

Recall from 10.18.2 that the vertical bundle of P is trivialized as

a vector bundle over P by the principal right action. So ω(X

u

) :=

T

e

(R

u

)

−1

.Φ(X

u

) ∈ g and in this way we get a g-valued 1-form ω ∈

Ω

1

(P; g), which is called the (Lie algebra valued) connection form of the

connection Φ. Recall from 1.3. the fundamental vector ﬁeld mapping

ζ : ( →X(P) for the principal right action.

Lemma. If Φ ∈ Ω

1

(P; V P) is a principal connection on the principal

ﬁber bundle (P, p, M, G) then the connection form has the following two

properties:

(1) ω reproduces the generators of fundamental vector ﬁelds, so we

have ω(ζ

X

(u)) = X for all X ∈ g.

(2) ω is G-equivariant, so we have ((r

g

)

∗

ω)(X

u

) = ω(T

u

(r

g

).X

u

) =

Ad(g

−1

).ω(X

u

) for all g ∈ G and X

u

∈ T

u

P. Consequently we

have for the Lie derivative L

ζ

X

ω = −ad(X).ω.

Conversely a 1-form ω ∈ Ω

1

(P, g) satisfying (1) deﬁnes a connection Φ

on P by Φ(X

u

) = T

e

(r

u

).ω(X

u

), which is a principal connection if and

only if (2) is satisﬁed.

Proof. (1). T

e

(r

u

).ω(ζ

X

(u)) = Φ(ζ

X

(u)) = ζ

X

(u) = T

e

(r

u

).X. Since

T

e

(r

u

) : ( →V

u

P is an isomorphism, the result follows.

(2). From T

e

(r

ug

).ω(T

u

(r

g

).X

u

) = ζ

ω(T

u

(r

g

).X

u

)

(ug) = Φ(T

u

(r

g

).X

u

)

and T

e

(r

ug

).Ad(g

−1

).ω(X

u

) = ζ

Ad(g

−1

).ω(X

u

)

(ug) = T

u

(r

g

).ζ

ω(X

u

)

(u) =

T

u

(r

g

).Φ(X

u

) both directions follow.

11.2. Curvature. Let Φ be a principal connection on the principal

ﬁber bundle (P, p, M, G) with connection form ω ∈ Ω

1

(P; g). We already

noted in 11.1 that the curvature R =

1

2

[Φ, Φ] is then also G-invariant,

(r

g

)

∗

R = R for all g ∈ G. Since R has vertical values we may again deﬁne

a g-valued 2-form Ω ∈ Ω

2

(P; g) by Ω(X

u

, Y

u

) := −T

e

(r

u

)

−1

.R(X

u

, Y

u

),

which is called the (Lie algebra-valued) curvature form of the connection.

We take the negative sign here to get the usual connection form as in

[Kobayashi-Nomizu I, 1963].

11. Principal bundles and induced connections 61

We equip the space Ω(P; g) of all g-valued forms on P in a canonical

way with the structure of a graded Lie algebra by

[Ψ, Θ](X

1

, . . . , X

p+q

) =

=

1

p! q!

σ

signσ [Ψ(X

σ1

, . . . , X

σp

), Θ(X

σ(p+1)

, . . . , X

σ(p+q)

)]

g

or equivalently by [ψ ⊗ X, θ ⊗ Y ] := ψ ∧ θ ⊗ [X, Y ]. In particular for

ω ∈ Ω

1

(P; g) we have [ω, ω](X, Y ) = 2[ω(X), ω(Y )].

Theorem. The curvature form Ω of a principal connection with con-

nection form ω has the following properties:

(1) Ω is horizontal, i.e. it kills vertical vectors.

(2) Ω is G-equivariant in the following sense: (r

g

)

∗

Ω = Ad(g

−1

).Ω.

Consequently L

ζ

X

Ω = −ad(X).Ω.

(3) The Maurer-Cartan formula holds: Ω = dω +

1

2

[ω, ω]

g

.

Proof. (1) is true for R by 9.4. For (2) we compute as follows:

T

e

(r

ug

).((r

g

)

∗

Ω)(X

u

, Y

u

) = T

e

(r

ug

).Ω(T

u

(r

g

).X

u

, T

u

(r

g

).Y

u

) =

= −R

ug

(T

u

(r

g

).X

u

, T

u

(r

g

).Y

u

) = −T

u

(r

g

).((r

g

)

∗

R)(X

u

, Y

u

) =

= −T

u

(r

g

).R(X

u

, Y

u

) = T

u

(r

g

).ζ

Ω(X

u

,Y

u

)

(u) =

= ζ

Ad(g

−1

).Ω(X

u

,Y

u

)

(ug) =

= T

e

(r

ug

).Ad(g

−1

).Ω(X

u

, Y

u

), by 1.3.

(3). For X ∈ g we have i

ζ

X

R = 0 by (1) and

i

ζ

X

(dω +

1

2

.[ω, ω]

g

) = i

ζ

X

dω +

1

2

[i

ζ

X

ω, ω] −

1

2

[ω, i

ζ

X

ω] =

= L

ζ

X

ω + [X, ω] = −ad(X)ω +ad(X)ω = 0

So the formula holds for vertical vectors, and for horizontal vector ﬁelds

X, Y ∈ C

∞

(H(P)) we have

R(X, Y ) = Φ[X −ΦX, Y −ΦY ] = Φ[X, Y ] = ζ

ω([X,Y ])

(dω +

1

2

[ω, ω])(X, Y ) = Xω(Y ) −Y ω(X) −ω([X, Y ]) = −ω([X, Y ])

62 11. Principal bundles and induced connections

11.3. Lemma. Any principal ﬁber bundle (P, p, M, G) (with paracom-

pact basis) admits principal connections.

Proof. Let (U

α

, ϕ

α

: P[U

α

→U

α

G)

α

be a principal ﬁber bundle atlas.

Let us deﬁne γ

α

(Tϕ

−1

α

(ξ

x

, T

e

λ

g

.X)) := X for ξ

x

∈ T

x

U

α

and X ∈ g. An

easy computation involving lemma 1.3 shows that γ

α

∈ Ω

1

(P[U

α

; g) sat-

isﬁes the requirements of lemma 11.1 and thus is a principal connection

on P[U

α

. Now let (f

α

) be a smooth partition of unity on M which is

subordinated to the open cover (U

α

), and let ω :=

α

(f

α

◦ p)γ

α

. Since

both requirements of lemma 11.1 are invariant under convex linear com-

binations, ω is a principal connection on P.

11.4. Local descriptions of principal connections. We consider a

principal ﬁber bundle (P, p, M, G) with some principal ﬁber bundle atlas

(U

α

, ψ

α

: P[U

α

→U

α

G) and corresponding cocycle (ϕ

αβ

: U

αβ

→G)

of transition functions. We consider the sections s

α

∈ C

∞

(P[U

α

) which

are given by ϕ

α

(s

α

(x)) = (x, e) and satisfy s

α

.ϕ

αβ

= s

β

.

(1) Let Θ ∈ Ω

1

(G, g) be the left logarithmic derivative of the iden-

tity, i.e. Θ(η

g

) := T

g

(λ

g

−1).η

g

. We will use the forms Θ

αβ

:=

ϕ

αβ

∗

Θ ∈ Ω

1

(U

αβ

; g).

Let Φ = ζ ◦ω ∈ Ω

1

(P; V P) be a principal connection, ω ∈ Ω

1

(P; g). We

may associate the following local data to the connection:

(2) ω

α

:= s

α

∗

ω ∈ Ω

1

(U

α

; g), the physicists version of the connection.

(3) The Christoﬀel forms Γ

α

∈ Ω

1

(U

α

; X(G)) from 9.7, which are

given by (0

x

, Γ

α

(ξ

x

, g)) = −T(ϕ

α

).Φ.T(ϕ

α

)

−1

(ξ

x

, 0

g

).

(4) γ

α

:= (ϕ

−1

α

)

∗

ω ∈ Ω

1

(U

α

G; g), the local expressions of ω.

Lemma. These local data have the following properties and are related

by the following formulas.

(5) The forms ω

α

∈ Ω

1

(U

α

; g) satisfy the transition formulas

ω

α

= Ad(ϕ

−1

βα

)ω

β

+ Θ

βα

,

and any set of forms like that with this transition behavior de-

termines a unique principal connection.

(6) γ

α

(ξ

x

, Tλ

g

.X) = γ

α

(ξ

x

, 0

g

) +X = Ad(g

−1

)ω

α

(ξ

x

) +X.

(7) Γ

α

(ξ

x

, g) = −T

e

(λ

g

).γ

α

(ξ

x

, 0

g

) = −T

e

(λ

g

).Ad(g

−1

)ω

α

(ξ

x

).

Proof. From the deﬁnition of the Christoﬀel forms we have

Γ

α

(ξ

x

, g) = −T(ϕ

α

).Φ.T(ϕ

α

)

−1

(ξ

x

, 0

g

)

= −T(ϕ

α

).T

e

(r

ϕ

α

(x,g)

)ω.T(ϕ

α

)

−1

(ξ

x

, 0

g

)

= −T

e

(ϕ

α

◦ r

ϕ

α

(x,g)

)ω.T(ϕ

α

)

−1

(ξ

x

, 0

g

)

= −T

e

(λ

g

)ω.T(ϕ

α

)

−1

(ξ

x

, 0

g

) = −T

e

(λ

g

)γ

α

(ξ

x

, 0

g

).

11. Principal bundles and induced connections 63

This is the ﬁrst part of (7). The second part follows from (6).

γ

α

(ξ

x

, Tλ

g

.X) = ω(T(ϕ

α

)

−1

(ξ

x

, 0

g

)) =

= γ

α

(ξ

x

, 0

g

) +ω(ζ

X

(ϕ

−1

α

(x, g))) = γ

α

(ξ

x

, 0

g

) +X.

So the ﬁrst part of (6) holds. The second part is seen from

γ

α

(ξ

x

, 0

g

) = γ

α

(ξ

x

, T

e

(ρ

g

)0

e

) = (ω ◦ T(ϕ

α

)

−1

◦ T(ρ

g

))(ξ

x

, 0

e

) =

= (ω ◦ T(r

g

◦ ϕ

−1

α

))(ξ

x

, 0

e

) = Ad(g

−1

)ω(T(ϕ

−1

α

)(ξ

x

, 0

e

))

= Ad(g

−1

)(s

α

∗

ω)(ξ

x

) = Ad(g

−1

)ω

α

(ξ

x

).

Via (7) the transition formulas for the ω

α

are easily seen to be equivalent

to the transition formulas for the Christoﬀel forms in lemma 9.7.

11.5. The covariant derivative. Let (P, p, M, G) be a principal ﬁber

bundle with principal connection Φ = ζ ◦ ω. We consider the horizontal

projection χ = Id

TP

−Φ : TP →HP, cf. 9.3, which satisﬁes χ◦ χ = χ,

imχ = HP, ker χ = V P, and χ ◦ T(r

g

) = T(r

g

◦ χ) for all g ∈ G.

If W is a ﬁnite dimensional vector space, we consider the mapping

χ

∗

: Ω(P; W) →Ω(P; W) which is given by

(χ

∗

ϕ)

u

(X

1

, . . . , X

k

) = ϕ

u

(χ(X

1

), . . . , χ(X

k

)).

The mapping χ

∗

is a projection onto the subspace of horizontal diﬀer-

ential forms, i.e. the space Ω

hor

(P; W) := ¦ψ ∈ Ω(P; W) : i

X

ψ =

0 for X ∈ V P¦. The notion of horizontal form is independent of the

choice of a connection.

The projection χ

∗

has the following properties: χ

∗

(ϕ∧ψ) = χ

∗

ϕ∧χ

∗

ψ,

if one of the two forms has real values; χ

∗

◦χ

∗

= χ

∗

; χ

∗

◦(r

g

)

∗

= (r

g

)

∗

◦χ

∗

for all g ∈ G; χ

∗

ω = 0; and χ

∗

◦ L(ζ

X

) = L(ζ

X

) ◦ χ

∗

. They follow

easily from the corresponding properties of χ, the last property uses

that Fl

ζ(X)

t

= r

exp tX

.

Now we deﬁne the covariant exterior derivative d

ω

: Ω

k

(P; W) →

Ω

k+1

(P; W) by the prescription d

ω

:= χ

∗

◦ d.

Theorem. The covariant exterior derivative d

ω

has the following prop-

erties.

(1) d

ω

(ϕ ∧ ψ) = d

ω

(ϕ) ∧ χ

∗

ψ + (−1)

deg ϕ

χ

∗

ϕ ∧ d

ω

(ψ).

(2) L(ζ

X

) ◦ d

ω

= d

ω

◦ L(ζ

X

) for each X ∈ g.

(3) (r

g

)

∗

◦ d

ω

= d

ω

◦ (r

g

)

∗

for each g ∈ G.

(4) d

ω

◦ p

∗

= d ◦ p

∗

= p

∗

◦ d : Ω(M; W) →Ω

hor

(P; W).

(5) d

ω

ω = Ω, the curvature form.

(6) d

ω

Ω = 0, the Bianchi identity.

(7) d

ω

◦ χ

∗

−d

ω

= χ

∗

◦ i(R) ◦ d, where R is the curvature.

(8) d

ω

◦ d

ω

= χ

∗

◦ i(R) ◦ d.

64 11. Principal bundles and induced connections

Proof. (1) through (4) follow from the properties of χ

∗

.

(5). We have

(d

ω

ω)(ξ, η) = (χ

∗

dω)(ξ, η) = dω(χξ, χη)

= (χξ)ω(χη) −(χη)ω(chξ) −ω([χξ, χη])

= −ω([χξ, χη]) and

R(ξ, η) = −ζ(Ω(ξ, η)) = Φ[χξ, χη].

(6). We have

d

ω

Ω = d

ω

(dω +

1

2

[ω, ω])

= χ

∗

ddω +

1

2

χ

∗

d[ω, ω]

=

1

2

χ

∗

([dω, ω] −[ω, dω]) = χ

∗

[dω, ω]

= [χ

∗

dω, χ

∗

ω] = 0, since χ

∗

ω = 0.

(7). For ϕ ∈ Ω(P; W) we have

(d

ω

χ

∗

ϕ)(X

0

, . . . , X

k

) = (dχ

∗

ϕ)(χ(X

0

), . . . , χ(X

k

))

=

0≤i≤k

(−1)

i

χ(X

i

)((χ

∗

ϕ)(χ(X

0

), . . . ,

χ(X

i

), . . . , χ(X

k

)))

+

i<j

(−1)

i+j

(χ

∗

ϕ)([χ(X

i

), χ(X

j

)], χ(X

0

), . . .

. . . ,

χ(X

i

), . . . ,

χ(X

j

), . . . )

=

0≤i≤k

(−1)

i

χ(X

i

)(ϕ(χ(X

0

), . . . ,

χ(X

i

), . . . , χ(X

k

)))

+

i<j

(−1)

i+j

ϕ([χ(X

i

), χ(X

j

)] −Φ[χ(X

i

), χ(X

j

)], χ(X

0

), . . .

. . . ,

χ(X

i

), . . . ,

χ(X

j

), . . . )

= (dϕ)(χ(X

0

), . . . , χ(X

k

)) + (i

R

ϕ)(χ(X

0

), . . . , χ(X

k

))

= (d

ω

+χ

∗

iR)(ϕ)(X

0

, . . . , X

k

).

(8).

d

ω

d

ω

= χ

∗

dχ

∗

d = (χ

∗

i

R

+χ

∗

d)d by (7)

= χ

∗

i

R

.

11. Principal bundles and induced connections 65

11.6 Theorem. Let (P, p, M, G) be a principal ﬁber bundle with princi-

pal connection ω. Then the parallel transport for the principal connection

is globally deﬁned and g-equivariant.

In detail: For each smooth c : R → M there is a smooth mapping

Pt

c

: R P

c(0)

→P such that the following holds:

(1) Pt(c, t, u) ∈ P

c(t)

, Pt(c, 0) = Id

P

c(0)

, and ω(

d

dt

Pt(c, t, u)) = 0.

(2) Pt(c, t) : P

c(0)

→P

c(t)

is G-equivariant, i.e.

Pt(c, t, u.g) = Pt(c, t, u).g.

(3) For smooth f : R →R we have

Pt(c, f(t), u) = Pt(c ◦ f, t, Pt(c, f(0), u)).

Proof. The Christoﬀel forms Γ

α

∈ Ω

1

(U

α

, X(G)) of the connection ω

with respect to a principal ﬁber bundle atlas (U

α

, ϕ

α

) take values in

the Lie subalgebra X

L

(G) of all left invariant vector ﬁelds on G, which

are bounded with respect to any left invariant Riemannian metric on G.

Each left invariant metric on a Lie group is complete. So the connection

is complete by the remark in 9.10.

Properties (1) and (3) follow from theorem 9.8, and (2) is seen as

follows: ω(

d

dt

Pt(c, t, u).g) = Ad(g

−1

)ω(

d

dt

Pt(c, t, u)) = 0 implies that

Pt(c, t, u).g = Pt(c, t, u.g).

11.7. Holonomy groups. Let (P, p, M, G) be a principal ﬁber bundle

with principal connection Φ = ζ ◦ ω. We assume that M is connected

and we ﬁx x

0

∈ M.

In view of developments which we make later in section 12 we de-

ﬁne the holonomy group Hol(Φ, x

0

) ⊂ Diﬀ(P

x

0

) as the group of all

Pt(c, 1) : P

x

0

→P

x

0

for c any piecewise smooth closed loop through x

0

.

(Reparametrizing c by a function which is ﬂat at each corner of c we may

assume that any c is smooth.) If we consider only those curves c which

are nullhomotopic, we obtain the restricted holonomy group Hol

0

(Φ, x

0

).

Now let us ﬁx u

0

∈ P

x

0

. The elements τ(u

0

, Pt(c, t, u

0

)) ∈ G form

a subgroup of the structure group G which is isomorphic to Hol(Φ, x

0

);

we denote it by Hol(ω, u

0

) and we call it also the holonomy group of the

connection. Considering only nullhomotopic curves we get the restricted

holonomy group Hol

0

(ω, u

0

) a normal subgroup Hol(ω, u

0

).

Theorem. 1. We have Hol(ω, u

0

.g) = Ad(g

−1

) Hol(ω, u

0

) and

Hol

0

(ω, u

0

.g) = Ad(g

−1

) Hol

0

(ω, u

0

).

2. For each curve c in M we have Hol(ω, Pt(c, t, u

0

)) = Hol(ω, u

0

)

and Hol

0

(ω, Pt(c, t, u

0

)) = Hol

0

(ω, u

0

).

3. Hol

0

(ω, u

0

) is a connected Lie subgroup of G and the quotient group

Hol(ω, u

0

)/ Hol

0

(ω, u

0

) is at most countable, so Hol(ω, u

0

) is also a Lie

subgroup of G.

66 11. Principal bundles and induced connections

4. The Lie algebra hol(ω, u

0

) ⊂ g of Hol(ω, u

0

) is linearly generated

by ¦Ω(X

u

, Y

u

) : X

u

, Y

u

∈ T

u

P¦. It is isomorphic to the Lie algebra

hol(Φ, x

0

) we considered in 9.9.

5. For u

0

∈ P

x

0

let P(ω, u

0

) be the set of all Pt(c, t, u

0

) for c any

(piecewise) smooth curve in M with c(0) = x

0

and for t ∈ R. Then

P(ω, u

0

) is a sub ﬁber bundle of P which is invariant under the right

action of Hol(ω, u

0

); so it is itself a principal ﬁber bundle over M with

structure group Hol(ω, u

0

) and we have a reduction of structure group, cf.

10.6 and 10.13. The pullback of ω to P(ω, u

0

) is then again a principal

connection form i

∗

ω ∈ Ω

1

(P(ω, u

0

); hol(ω, u

0

)).

6. P is foliated by the leaves P(ω, u), u ∈ P

x

0

.

7. If the curvature Ω = 0 then Hol

0

(ω, u

0

) = ¦e¦ and each P(ω, u) is

a covering of M.

8. If one uses piecewise C

k

-curves for 1 ≤ k < ∞ in the deﬁnition,

one gets the same holonomy groups.

In view of assertion 5 a principal connection ω is called irreducible

if Hol(ω, u

0

) equals the structure group G for some (equivalently any)

u

0

∈ P

x

0

.

Proof. 1. This follows from the properties of the mapping τ from 10.2

and from the from the G-equivariancy of the parallel transport.

The rest of this theorem is a compilation of well known results, and

we refer to [Kobayashi-Nomizu I, 1963, p. 83ﬀ] for proofs.

11.8. Inducing principal connections on associated bundles.

Let (P, p, M, G) be a principal bundle with principal right action r :

P G → P and let : G S → S be a left action of the structure

group G on some manifold S. Then we consider the associated bundle

P[S] = P[S, ] = P

G

S, constructed in 10.7. Recall from 10.18 that

its tangent and vertical bundle are given by T(P[S, ]) = TP[TS, T] =

TP

TG

TS and V (P[S, ]) = P[TS, T

2

] = P

G

TS.

Let Φ = ζ ◦ω ∈ Ω

1

(P; TP) be a principal connection on the principal

bundle P. We construct the induced connection

¯

Φ ∈ Ω

1

(P[S], T(P[S]))

by the following diagram:

TP TS

Φ×Id

−−−−→ TP TS T(P S)

Tq=q

¸

¸

_ q

¸

¸

_

Tq

¸

¸

_

TP

TG

TS −−−−→

¯

Φ

TP

TG

TS T(P

G

S).

Let us ﬁrst check that the top mapping Φ Id is TG-equivariant. For

g ∈ G and X ∈ g the inverse of T

e

(λ

g

)X in the Lie group TG is denoted

11. Principal bundles and induced connections 67

by (T

e

(λ

g

)X)

−1

, see lemma 10.17. Furthermore by 1.3 we have

Tr(ξ

u

, T

e

(λ

g

)X) = T

u

(r

g

)ξ

u

+Tr((O

P

L

X

)(u, g))

= T

u

(r

g

)ξ

u

+T

g

(r

u

)(T

e

(λ

g

)X)

= T

u

(r

g

)ξ

u

+ζ

X

(ug).

We may compute

(Φ Id)(Tr(ξ

u

, T

e

(λ

g

)X), T((T

e

(λ

g

)X)

−1

, η

s

))

= (Φ(T

u

(r

g

)ξ

u

+ζ

X

(ug)), T((T

e

(λ

g

)X)

−1

, η

s

))

= (Φ(T

u

(r

g

)ξ

u

) + Φ(ζ

X

(ug)), T((T

e

(λ

g

)X)

−1

, η

s

))

= (Φ(T

u

(r

g

)ξ

u

) +ζ

X

(ug), T((T

e

(λ

g

)X)

−1

, η

s

))

= (Tr((Φ Id)ξ

u

, T

e

(λ

g

)X), T((T

e

(λ

g

)X)

−1

, η

s

))

So the mapping ΦId factors to

¯

Φ as indicated in the diagram, and we

have

¯

Φ ◦

¯

Φ =

¯

Φ from (Φ Id) ◦ (Φ Id) = Φ Id. The mapping

¯

Φ is

ﬁberwise linear, since Φ Id and q

**= Tq are. The image of
**

¯

Φ is

q

(V P TS) = q

(ker(Tp : TP TS →TM))

= ker(Tp : TP

TG

TS →TM) = V (P[S, ]).

Thus

¯

Φ is a connection on the associated bundle P[S]. We call it the

induced connection.

From the diagram it also follows, that the vector valued forms ΦId ∈

Ω

1

(P S; TP TS) and

¯

Φ ∈ Ω

1

(P[S]; T(P[S])) are (q : P S →P[S])-

related. So by 8.15 we have for the curvatures

R

Φ×Id

=

1

2

[Φ Id, Φ Id] =

1

2

[Φ, Φ] 0 = R

Φ

0,

R¯

Φ

=

1

2

[

¯

Φ,

¯

Φ],

that they are also q-related, i.e. Tq ◦ (R

Φ

0) = R¯

Φ

◦ (Tq

M

Tq).

By uniqueness of the solutions of the deﬁning diﬀerential equation we

also get that

Pt¯

Φ

(c, t, q(u, s)) = q(Pt

Φ

(c, t, u), s).

11.9. Recognizing induced connections. We consider again a prin-

cipal ﬁber bundle (P, p, M, G) and a left action : GS →S. Suppose

that Ψ ∈ Ω

1

(P[S]; T(P[S])) is a connection on the associated bundle

P[S] = P[S, ]. Then the following question arises: When is the connec-

tion Ψ induced from a principal connection on P? If this is the case, we

say that Ψ is compatible with the G-structure on P[S]. The answer is

given in the following

68 11. Principal bundles and induced connections

Theorem. Let Ψ be a (general) connection on the associated bundle

P[S]. Let us suppose that the action is inﬁnitesimally eﬀective, i.e.

the fundamental vector ﬁeld mapping ζ : g →X(S) is injective.

Then the connection Ψ is induced from a principal connection ω on

P if and only if the following condition is satisﬁed:

In some (equivalently any) ﬁber bundle atlas (U

α

, ψ

α

) of P[S] be-

longing to the G-structure of the associated bundle the Christof-

fel forms Γ

α

∈ Ω

1

(U

α

; X(S)) have values in the sub Lie algebra

X

fund

(S) of fundamental vector ﬁelds for the action .

Proof. Let (U

α

, ϕ

α

: P[U

α

→ U

α

G) be a principal ﬁber bundle atlas

for P. Then by the proof of theorem 10.7 the induced ﬁber bundle atlas

(U

α

, ψ

α

: P[S][U

α

→U

α

S) is given by

ψ

−1

α

(x, s) = q(ϕ

−1

α

(x, e), s), (1)

(ψ

α

◦ q)(ϕ

−1

α

(x, g), s) = (x, g.s). (2)

Let Φ = ζ ◦ ω be a principal connection on P and let

¯

Φ be the induced

connection on the associated bundle P[S]. By 9.7 its Christoﬀel symbols

are given by

Γ

α

¯

Φ

(ξ

x

, s) = −(T(ψ

α

) ◦

¯

Φ ◦ T(ψ

−1

α

))(ξ

x

, 0

s

)

= −(T(ψ

α

) ◦

¯

Φ ◦ Tq ◦ (T(ϕ

−1

α

) Id))(ξ

x

, 0

e

, 0

s

) by (1)

= −(T(ψ

α

) ◦ Tq ◦ (Φ Id))(T(ϕ

−1

α

)(ξ

x

, 0

e

), 0

s

) by 11.8

= −(T(ψ

α

) ◦ Tq)(Φ(T(ϕ

−1

α

)(ξ

x

, 0

e

)), 0

s

)

= (T(ψ

α

) ◦ Tq)(T(ϕ

−1

α

)(Γ

α

Φ

(ξ

x

, e)), 0

s

) by 11.4,(3)

= −T(ψ

α

◦ q ◦ (ϕ

−1

α

Id))(0

x

, ω

α

(ξ

x

), 0

s

) by 11.4,(7)

= −T

e

(

s

)ω

α

(ξ

x

) by (2)

= −ζ

ω

α

(ξ

x

)

(s).

So the condition is necessary. Now let us conversely suppose that a

connection Ψ on P[S] is given such that the Christoﬀel forms Γ

α

Ψ

with

respect to a ﬁber bundle atlas of the G-structure have values in X

fund

(S).

Then unique g-valued forms ω

α

∈ Ω

1

(U

α

; g) are given by the equation

Γ

α

Ψ

(ξ

x

) = ζ(ω

α

(ξ

x

)),

since the action is inﬁnitesimally eﬀective. From the transition formulas

9.7 for the Γ

α

Ψ

follow the transition formulas 11.4.(5) for the ω

α

, so that

they give a unique principal connection on P, which by the ﬁrst part of

the proof induces the given connection Ψ on P[S].

11. Principal bundles and induced connections 69

11.10. Inducing principal connections on associated vector

bundles. Let (P, p, M, G) be a principal ﬁber bundle and let ρ : G →

GL(W) be a representation of the structure group G on a ﬁnite di-

mensional vector space W. We consider the associated vector bundle

(E := P[W, ρ], p, M, W), which was treated in some detail in 10.11.

The tangent bundle T(E) = TP

TG

TW has two vector bundle

structures with the projections

π

E

: T(E) = TP

TG

TW →P

G

W = E,

Tp ◦ pr

1

: T(E) = TP

TG

TW →TM;

the ﬁrst one is the vector bundle structure of the tangent bundle, the

second one is the derivative of the vector bundle structure on E.

Now let Φ = ζ ◦ ω ∈ Ω

1

(P; TP) be a principal connection on P. We

consider the induced connection

¯

Φ ∈ Ω

1

(E; T(E)) from 11.8. Inserting

the projections of both vector bundle structures on T(E) into the di-

agram in 11.8 one easily sees that the induced connection is linear in

both vector bundle structures: the new aspect is that it is a linear endo-

morphism of the vector bundle (TE, Tp, TM). We say that it is a linear

connection on the associated bundle.

Recall now from 11.8 the vertical lift vl

E

: E

M

E → V E, which is

an isomorphism, pr

1

–π

E

–ﬁberwise linear and also p–Tp–ﬁberwise linear.

Now we deﬁne the connector K of the linear connection

¯

Φ by

K := pr

2

◦ (vl

E

)

−1

◦

¯

Φ : TE →V E →E

M

E →E.

Lemma. The connector K : TE → E is π

E

–p–ﬁberwise linear and

Tp–p–ﬁberwise linear and satisﬁes K◦vl

E

= pr

2

: E

M

E →TE →E.

Proof. This follows from the ﬁberwise linearity of the composants of K

and from its deﬁnition.

11.11. Linear connections. If (E, p, M) is a vector bundle, a con-

nection Ψ ∈ Ω

1

(E; TE) such that Ψ : TE →V E →TE is also Tp–Tp–

ﬁberwise linear is called a linear connection. An easy check with 11.9

or a direct construction shows that Ψ is then induced from a unique

principal connection on the linear frame bundle GL(R

n

, E) of E (where

n is the ﬁber dimension of E).

Equivalently a linear connection may be speciﬁed by a connector K :

TE → E with the three properties of lemma 11.10. For then HE :=

¦ξ

u

: K(ξ

u

) = 0

p(u)

¦ is a complement to V E in TE which is Tp–ﬁberwise

linearly chosen.

70 11. Principal bundles and induced connections

11.12. Covariant derivative on vector bundles. Let (E, p, M) be a

vector bundle with a linear connection, given by a connector K : TE →

E with the properties in lemma 11.10.

For any manifold N, smooth mapping s : N → E, and vector ﬁeld

X ∈ X(N) we deﬁne the covariant derivative of s along X by

(1) ∇

X

s := K ◦ Ts ◦ X : N →TN →TE →E.

If f : N → M is a ﬁxed smooth mapping, let us denote by C

∞

f

(N, E)

the vector space of all smooth mappings s : N → E with p ◦ s = f –

they are called sections of E along f. From the universal property of the

pullback it follows that the vector space C

∞

f

(N, E) is canonically linearly

isomorphic to the space C

∞

(f

∗

E) of sections of the pullback bundle.

Then the covariant derivative may be viewed as a bilinear mapping

(2) ∇ : X(N) C

∞

f

(N, E) →C

∞

f

(N, E).

Lemma. This covariant derivative has the following properties:

(3) ∇

X

s is C

∞

(N, R)-linear in X ∈ X(N).

(4) ∇

X

s is R-linear in s ∈ C

∞

f

(N, E). So for a tangent vector

X

x

∈ T

x

N the mapping ∇

X

x

: C

∞

f

(N, E) → E

f(x)

makes sense

and (∇

X

s)(x) = ∇

X(x)

s.

(5) ∇

X

(h.s) = dh(X).s + h.∇

X

s for h ∈ C

∞

(N, R), the derivation

property of ∇

X

.

(6) For any manifold Q and smooth mapping g : Q → N and Y

y

∈

T

y

Q we have ∇

Tg.Y

y

s = ∇

Y

y

(s ◦g). If Y ∈ X(Q) and X ∈ X(N)

are g-related, then we have ∇

Y

(s ◦ g) = (∇

X

s) ◦ g.

Proof. All these properties follow easily from the deﬁnition (1).

Remark. Property (6) is not well understood in some diﬀerential geo-

metric literature. See e.g. the clumsy and unclear treatment of it in

[Eells-Lemaire, 1983].

For vector ﬁelds X, Y ∈ X(M) and a section s ∈ C

∞

(E) an easy

computation shows that

R

E

(X, Y )s : = ∇

X

∇

Y

s −∇

Y

∇

X

s −∇

[X,Y ]

s

= ([∇

X

, ∇

Y

] −∇

[X,Y ]

)s

is C

∞

(M, R)-linear in X, Y , and s. By the method of 7.4 it follows

that R

E

is a 2-form on M with values in the vector bundle L(E, E),

11. Principal bundles and induced connections 71

i.e. R

E

∈ Ω

2

(M; L(E, E)). It is called the curvature of the covariant

derivative.

For f : N → M, vector ﬁelds X, Y ∈ X(N) and a section s ∈

C

∞

f

(N, E) along f we obtain

∇

X

∇

Y

s −∇

Y

∇

X

s −∇

[X,Y ]

s = (f

∗

R

E

)(X, Y )s.

11.13. Covariant exterior derivative. Let (E, p, M) be a vector

bundle with a linear connection, given by a connector K : TE →E.

For a smooth mapping f : N →M let Ω(N; f

∗

E) be the vector space

of all forms on N with values in the vector bundle f

∗

E. We can also view

them as forms on N with values along f in E, but we do not introduce

an extra notation for this.

The graded space Ω(N; f

∗

E) is a graded Ω(N)-module via

(ϕ ∧ Φ)(X

1

, . . . , X

p+q

) =

=

1

p! q!

σ

sign(σ) ϕ(X

σ1

, . . . , X

σp

)Φ(X

σ(p+1)

, . . . , X

σ(p+q)

).

It can easily be shown that the graded module homomorphisms H :

Ω(N; f

∗

E) →Ω(N; f

∗

E) (so that H(ϕ∧Φ) = (−1)

deg H. deg ϕ

ϕ∧H(Φ))

are exactly the mappings µ(K) for K ∈ Ω

q

(N; f

∗

L(E, E)), which are

given by

(µ(K)Φ)(X

1

, . . . , X

p+q

) =

=

1

p! q!

σ

sign(σ) K(X

σ1

, . . . , X

σp

)(Φ(X

σ(p+1)

, . . . , X

σ(p+q)

)).

The covariant exterior derivative d

∇

: Ω

p

(N : f

∗

E) →Ω

p+1

(N; f

∗

E) is

deﬁned by (where the X

i

are vector ﬁelds on N)

(d

∇

Φ)(X

0

, . . . , X

p

) =

p

i=0

(−1)

i

∇

X

i

Φ(X

0

, . . . ,

´

X

i

, . . . , X

p

)

+

0≤i<j≤p

(−1)

i+j

Φ([X

i

, X

j

], X

0

, . . . ,

´

X

i

, . . . ,

´

X

j

, . . . , X

p

).

Lemma. The covariant exterior derivative is well deﬁned and has the

following properties.

(1) For s ∈ C

∞

(f

∗

E) = Ω

0

(N; f

∗

E) we have (d

∇

s)(X) = ∇

X

s.

(2) d

∇

(ϕ ∧ Φ) = dϕ ∧ Φ + (−1)

deg ϕ

ϕ ∧ d

∇

Φ.

(3) For smooth g : Q →N and Φ ∈ Ω(N; f

∗

E) we have d

∇

(g

∗

Φ) =

g

∗

(d

∇

Φ).

(4) d

∇

d

∇

Φ = µ(f

∗

R

E

)Φ.

72 11. Principal bundles and induced connections

Proof. It suﬃces to investigate decomposable forms Φ = ϕ ⊗ s for ϕ ∈

Ω

p

(N) and s ∈ C

∞

(f

∗

E). Then from the deﬁnition we have d

∇

(ϕ⊗s) =

dϕ ⊗ s + (−1)

p

ϕ ∧ d

∇

s. Since by 11.12,(3) d

∇

s ∈ Ω

1

(N; f

∗

E), the

mapping d

∇

is well deﬁned. This formula also implies (2) immediately.

(3) follows from 11.12,(6). (4) is checked as follows:

d

∇

d

∇

(ϕ ⊗s) = d

∇

(dϕ ⊗s + (−1)

p

ϕ ∧ d

∇

s) by (2)

= 0 + (−1)

2p

ϕ ∧ d

∇

d

∇

s

= ϕ ∧ µ(f ∗ R

E

)s by the deﬁnition of R

E

= µ(f

∗

R

E

)(ϕ ∧ s).

11.14. Let (P, p, M, G) be a principal ﬁber bundle and let ρ : G →

GL(W) be a representation of the structure group G on a ﬁnite dimen-

sional vector space W.

Theorem. There is a canonical isomorphism from the space of P[W, ρ]-

valued diﬀerential forms on M onto the space of horizontal G-equivariant

W-valued diﬀerential forms on P:

q

: Ω(M; P[W, ρ]) →Ω

hor

(P; W)

G

= ¦ϕ ∈ Ω(P; W) : i

X

ϕ = 0

for all X ∈ V P, (r

g

)

∗

ϕ = ρ(g

−1

) ◦ ϕ for all g ∈ G¦.

In particular for W = R with trivial representation we see that

p

∗

: Ω(M) →Ω

hor

(P)

G

= ¦ϕ ∈ Ω

hor

(P) : (r

g

)

∗

ϕ = ϕ¦

is also an isomorphism. The isomorphism

q

: Ω

0

(M; P[W]) = C

∞

(P[W]) →Ω

0

hor

(P; W)

G

= C

∞

(P, W)

G

is a special case of the one from 10.12.

Proof. Recall the smooth mapping τ : P

M

P → G from 10.2, which

satisﬁes r(u

x

, τ(u

x

, v

x

)) = v

x

, τ(u

x

.g, u

x

.g

) = g

−1

.τ(u

x

, u

x

).g

, and

τ(u

x

, u

x

) = e.

Let ϕ ∈ Ω

k

hor

(P; W)

G

, X

1

, . . . , X

k

∈ T

u

P, and X

1

, . . . , X

k

∈ T

u

P

such that T

u

p.X

i

= T

u

p.X

i

for each i. Then we have for g = τ(u, u

),

so that ug = u

:

q(uϕ

u

(X

1

, . . . , X

k

)) = q(ug, ρ(g

−1

)ϕ

u

(X

1

, . . . , X

k

))

= q(u

, ((r

g

)

∗

ϕ)

u

(X

1

, . . . , X

k

))

= q(u

, ϕ

ug

(T

u

(r

g

).X

1

, . . . , T

u

(r

g

).X

k

))

= q(u

, ϕ

u

(X

1

, . . . , X

k

)), since T

u

(r

g

)X

i

−X

i

∈ V

u

P.

11. Principal bundles and induced connections 73

By this prescription a vector bundle valued form Φ ∈ Ω

k

(M; P[W]) is

uniquely determined.

For the converse recall the smooth mapping ¯ τ : P

M

P[W, ρ] → W

from 10.7, which satisﬁes ¯ τ(u, q(u, w)) = w, q(u

x

, ¯ τ(u

x

, v

x

)) = v

x

, and

¯ τ(u

x

g, v

x

) = ρ(g

−1

)¯ τ(u

x

, v

x

).

For Φ ∈ Ω

k

(M; P[W]) we deﬁne q

∈ Ω

k

(P; W) as follows. For

X

i

∈ T

u

P we put

(q

Φ)(X

1

, . . . , X

k

) := ¯ τ(u, Φ

p(u)

(T

u

p.X

1

, . . . , T

u

p.X

k

)).

Then q

**Φ is smooth and horizontal. For g ∈ G we have
**

((r

g

)

∗

(q

Φ))

u

(X

1

, . . . , X

k

) = (q

Φ)

ug

(T

u

(r

g

).X

1

, . . . , T

u

(r

g

).X

k

)

= ¯ τ(ug, Φ

p(ug)

(T

ug

p.T

u

(r

g

).X

1

, . . . , T

ug

p.T

u

(r

g

).X

k

))

= ρ(g

−1

)¯ τ(u, Φ

p(u)

(T

u

p.X

1

, . . . , T

u

p.X

k

))

= ρ(g

−1

)(q

Φ)

u

(X

1

, . . . , X

k

).

Clearly the two constructions are inverse to each other.

11.15. Let (P, p, M, G) be a principal ﬁber bundle with a principal con-

nection Φ = ζ ◦ ω, and let ρ : G → GL(W) be a representation of the

structure group G on a ﬁnite dimensional vector space W. We con-

sider the associated vector bundle (E := P[W, ρ], p, M, W), the induced

connection

¯

Φ on it and the corresponding covariant derivative.

Theorem. The covariant exterior derivative d

ω

from 11.5 on P and the

covariant exterior derivative for P[W]-valued forms on M are connected

by the mapping q

from 11.14, as follows:

q

◦ d

∇

= d

ω

◦ q

: Ω(M; P[W]) →Ω

hor

(P; W)

G

.

Proof. Let ﬁrst f ∈ Ω

0

hor

(P; W)

G

= C

∞

(P, W)

G

, then f = q

s for s ∈

C

∞

(P[W]) and we have f(u) = τ(u, s(p(u))) and s(p(u)) = q(u, f(u))

by 11.14 and 10.12. Therefore Ts.Tp.X

u

= Tq(X

u

, Tf.X

u

), where

Tf.X

u

= (f(u), df(X

u

)) ∈ TW = W W. If χ : TP → HP is the

horizontal projection as in 11.5, we have Ts.Tp.X

u

= Ts.Tp.χ.X

u

=

Tq(χ.X

u

, Tf.χ.X

u

). So we get

(q

d

∇

s)(X

u

) = τ(u, (d

∇

s)(Tp.X

u

))

= τ(u, ∇

Tp.X

u

s) by 11.13,(1)

= τ(u, K.Ts.Tp.X

u

) by 11.12.(1)

74 11. Principal bundles and induced connections

= τ(u, K.Tq(χ.X

u

, Tf.χ.X

u

)) from above

= τ(u, pr

2

.vl

−1

P[W]

.

¯

Φ.Tq(χ.X

u

, Tf.χ.X

u

)) by 11.10

= τ(u, pr

2

.vl

−1

P[W]

.Tq.(Φ Id)(χ.X

u

, Tf.χ.X

u

))) by 11.8

= τ(u, pr

2

.vl

−1

P[W]

.Tq(0

u

, Tf.χ.X

u

))) since Φ.χ = 0

= τ(u, q.pr

2

.vl

−1

P×W

.Tq(0

u

, Tf.χ.X

u

))) since q is ﬁber linear

= τ(u, q(u, df.χ.X

u

)) = (χ

∗

df)(X

u

)

= (d

ω

q

s)(X

u

).

Now we turn to the general case. It suﬃces to check the formula for

a decomposable P[W]-valued form Ψ = ψ ⊗ s ∈ Ω

k

(M, P[W]), where

ψ ∈ Ω

k

(M) and s ∈ C

∞

(P[W]). Then we have

d

ω

q

(ψ ⊗s) = d

ω

(p

∗

ψ q

s)

= d

ω

(p

∗

ψ) q

s + (−1)

k

χ

∗

p

∗

ψ ∧ d

ω

q

s by 11.5,(1)

= χ

∗

p

∗

dψ q

s + (−1)

k

p

∗

ψ ∧ q

d

∇

s from above

= p

∗

dψ q

s + (−1)

k

p

∗

ψ ∧ q

d

∇

s

= q

(dψ ⊗s + (−1)

k

ψ ∧ d

∇

s)

= q

d

∇

(ψ ⊗s).

11.16. Corollary. In the situation of theorem 11.15 above we have

for the curvature form Ω ∈ Ω

2

hor

(P; g) and the curvature R

P[W]

∈

Ω

2

(M; L(P[W], P[W])) the relation

q

L(P[W],P[W])

R

P[W]

= ρ

◦ Ω,

where ρ

= T

e

ρ : g →L(W, W) is the derivative of the representation ρ.

Proof. We use the notation of the proof of theorem 11.15. By this the-

orem we have for X, Y ∈ T

u

P

(d

ω

d

ω

q

P[W]

s)

u

(X, Y ) = (q

d

∇

d∇s)

u

(X, Y )

= (q

R

P[W]

s)

u

(X, Y )

= τ(u, R

P[W]

(T

u

p.X, T

u

p.Y )s(p(u)))

= (q

L(P[W],P[W])

R

P[W]

)

u

(X, Y )(q

P[W]

s)(u).

11. Principal bundles and induced connections 75

On the other hand we have by theorem 11.5.(8)

(d

ω

d

ω

q

s)

u

(X, Y ) = (χ

∗

i

R

dq

s)

u

(X, Y )

= (dq

s)

u

(R(X, Y )) since R is horizontal

= (dq

s)(−ζ

Ω(X,Y )

(u)) by 11.2

=

∂

∂t

¸

¸

0

(q

s)(Fl

ζ

Ω(X,Y )

−t

(u))

=

∂

∂t

¸

¸

0

τ(u. exp(−tΩ(X, Y )), s(p(u. exp(−tΩ(X, Y )))))

=

∂

∂t

¸

¸

0

τ(u. exp(−tΩ(X, Y )), s(p(u)))

=

∂

∂t

¸

¸

0

ρ(exp tΩ(X, Y ))τ(u, s(p(u))) by 11.7

= ρ

(Ω(X, Y ))(q

s)(u).

76

12. Holonomy

12.1. Holonomy groups. Let (E,p,M,S) be a ﬁber bundle with a

complete connection Φ, and let us assume that M is connected. We

choose a ﬁxed base point x

0

∈ M and we identify E

x

0

with the standard

ﬁber S. For each closed piecewise smooth curve c : [0, 1] → M through

x

0

the parallel transport Pt(c, , 1) =: Pt(c, 1) (pieced together over the

smooth parts of c) is a diﬀeomorphism of S. All these diﬀeomorphisms

form together the group Hol(Φ, x

0

), the holonomy group of Φ at x

0

, a

subgroup of the diﬀeomorphism group Diﬀ(S). If we consider only those

piecewise smooth curves which are homotopic to zero, we get a subgroup

Hol

0

(Φ, x

0

), called the restricted holonomy group of the connection Φ at

x

0

.

12.2. Holonomy Lie algebra. In the setting of 12.1 let C : TM

M

E → TE be the horizontal lifting as in 9.3, and let R be the curvature

(9.4) of the connection Φ. For any x ∈ M and X

x

∈ T

x

M the horizontal

lift C(X

x

) := C(X

x

, ) : E

x

→ TE is a vector ﬁeld along E

x

. For X

x

and Y

x

∈ T

x

M we consider R(CX

x

, CY

x

) ∈ X(E

x

). Now we choose any

piecewise smooth curve c from x

0

to x and consider the diﬀeomorphism

Pt(c, t) : S = E

x

0

→ E

x

and the pullback Pt(c, 1)

∗

R(CX

x

, CY

x

) ∈

X(S). Let us denote by hol(Φ, x

0

) the closed linear subspace, generated

by all these vector ﬁelds (for all x ∈ M, X

x

, Y

x

∈ T

x

M and curves c

from x

0

to x) in X(S) with respect to the compact C

∞

-topology, and

let us call it the holonomy Lie algebra of Φ at x

0

.

12.3. Lemma. hol(Φ, x

0

) is a Lie subalgebra of X(S).

Proof. For X ∈ X(M) we consider the local ﬂow Fl

CX

t

of the horizontal

lift of X. It restricts to parallel transport along any of the ﬂow lines of

X in M. Then for vector ﬁelds on M the expression

d

dt

[

0

(Fl

CX

s

)

∗

(Fl

CY

t

)

∗

(Fl

CX

−s

)

∗

(Fl

CZ

z

)

∗

R(CU, CV ) E

x

0

= (Fl

CX

s

)

∗

[CY, (Fl

CX

−s

)

∗

(Fl

CZ

z

)

∗

R(CU, CV )] E

x

0

= [(Fl

CX

s

)

∗

CY, (Fl

CZ

z

)

∗

R(CU, CV )] E

x

0

is in hol(Φ, x

0

), since it is closed in the compact C

∞

-topology and the

derivative can be written as a limit. Thus

[(Fl

CX

s

)

∗

[CY

1

, CY

2

], (Fl

CZ

z

)

∗

R(CU, CV )] E

x

0

∈ hol(Φ, x

0

)

by the Jacobi identity and

[(Fl

CX

s

)

∗

C[Y

1

, Y

2

], (Fl

CZ

z

)

∗

R(CU, CV )] E

x

0

∈ hol(Φ, x

0

),

12. Holonomy 77

so also their diﬀerence

[(Fl

CX

s

)

∗

R(CY

1

, CY

2

), (Fl

CZ

z

)

∗

R(CU, CV )] E

x

0

is in hol(Φ, x

0

).

12.4. The following theorem is a generalization of the theorem of Am-

brose and Singer on principal connections.

Theorem. Let Φ be a connection on the ﬁber bundle (E, p, M, S) and

let M be connected. Suppose that for some (hence any) x

0

∈ M the

holonomy Lie algebra hol(Φ, x

0

) is ﬁnite dimensional and consists of

complete vector ﬁelds on the ﬁber E

x

0

Then there is a principal bundle (P, p, M, G) with ﬁnite dimensional

structure group G, an irreducible connection ω on it and a smooth action

of G on S such that the Lie algebra g of G equals the holonomy Lie

algebra hol(Φ, x

0

), the ﬁber bundle E is isomorphic to the associated

bundle P[S], and Φ is the connection induced by ω. The structure group

G equals the holonomy group Hol(Φ, x

0

). P and ω are unique up to

isomorphism.

By a theorem of [Palais, 1957] a ﬁnite dimensional Lie subalgebra of

X(E

x

0

) like Hol(Φ, x

0

) consists of complete vector ﬁelds if and only if it

is generated by complete vector ﬁelds as a Lie algebra.

Proof. Let us again identify E

x

0

and S. Then g := hol(Φ, x

0

) is a ﬁnite

dimensional Lie subalgebra of X(S), and since each vector ﬁeld in it

is complete, there is a ﬁnite dimensional connected Lie group G

0

of

diﬀeomorphisms of S with Lie algebra g, see [Palais, 1957].

Claim 1. G

0

equals Hol

0

(Φ, x

0

), the restricted holonomy group.

Let f ∈ Hol

0

(Φ, x

0

), then f = Pt(c, 1) for a piecewise smooth closed

curve c through x

0

, which is nullhomotopic. Since the parallel transport

is essentially invariant under reparametrization, 9.8, we can replace c by

c ◦ g, where g is smooth and ﬂat at each corner of c. So we may assume

that c itself is smooth. Since c is homotopic to zero, by approximation

we may assume that there is a smooth homotopy H : R

2

→ M with

H

1

[[0, 1] = c and H

0

[[0, 1] = x

0

. Then f

t

:= Pt(H

t

, 1) is a curve in

Hol

0

(Φ, x

0

) which is smooth as a mapping R S →S.

Claim 2. (

d

dt

f

t

) ◦ f

−1

t

=: Z

t

is in g for all t.

To prove claim 2 we consider the pullback bundle H

∗

E → R

2

with the

induced connection H

Φ

. It is suﬃcient to prove claim 2 there. Let

X =

d

ds

and Y =

d

dt

be the constant vector ﬁelds on R

2

, so [X, Y ] = 0.

Then Pt(c, s) = Fl

CX

s

[S and so on. We put

f

t,s

= Fl

CX

−s

◦ Fl

CY

−t

◦ Fl

CX

s

◦ Fl

CY

t

: S →S,

78 12. Holonomy

so f

t,1

= f

t

. Then we have in the vector space X(S)

(

d

dt

f

t,s

) ◦ f

−1

t,s

= −(Fl

CX

s

)

∗

CY + (Fl

CX

s

)

∗

(Fl

CY

t

)

∗

(Fl

CX

−s

)

∗

CY,

(

d

dt

f

t,s

) ◦ f

−1

t,s

=

_

1

0

d

ds

_

(

d

dt

f

t,s

) ◦ f

−1

t,s

_

ds

=

_

1

0

_

−(Fl

CX

s

)

∗

[CX, CY ] + (Fl

CX

s

)

∗

[CX, (Fl

CY

t

)

∗

(Fl

CX

−s

)

∗

CY ]

−(Fl

CX

s

)

∗

(Fl

CY

t

)

∗

(Fl

CX

−s

)

∗

[CX, CY ]

_

ds.

Since [X, Y ] = 0 we have [CX, CY ] = Φ[CX, CY ] = R(CX, CY ) and

(Fl

CX

t

)

∗

CY = C

_

(Fl

X

t

)

∗

Y

_

+ Φ

_

(Fl

CX

t

)

∗

CY

_

= CY +

_

1

0

d

dt

Φ(Fl

CX

t

)

∗

CY dt = CY +

_

1

0

Φ(Fl

CX

t

)

∗

[CX, CY ] dt

= CY +

_

1

0

Φ(Fl

CX

t

)

∗

R(CX, CY ) dt.

Thus all parts of the integrand above are in g and so (

d

dt

f

t,s

) ◦ f

−1

t,s

is in

g for all t and claim 2 follows.

Now claim 1 can be shown as follows. There is a unique smooth curve

g(t) in G

0

satisfying T

e

(ρ

g(t)

)Z

t

= Z

t

.g(t) =

d

dt

g(t) and g(0) = e; via

the action of G

0

on S the curve g(t) is a curve of diﬀeomorphisms on S,

generated by the time dependent vector ﬁeld Z

t

, so g(t) = f

t

and f = f

1

is in G

0

. So we get Hol

0

(Φ, x

0

) ⊆ G

0

.

Claim 3. Hol

0

(Φ, x

0

) equals G

0

.

In the proof of claim 1 we have seen that Hol

0

(Φ, x

0

) is a smoothly

arcwise connected subgroup of G

0

, so it is a connected Lie subgroup by

the results cited in 5.6. It suﬃces thus to show that the Lie algebra

g of G

0

is contained in the Lie algebra of Hol

0

(Φ, x

0

), and for that

it is enough to show, that for each ξ ∈ g there is a smooth mapping

f : [−1, 1] S →S such that the associated curve

ˇ

f lies in Hol

0

(Φ, x

0

)

with

ˇ

f

(0) = 0 and

ˇ

f

(0) = ξ.

By deﬁnition we may assume ξ = Pt(c, 1)

∗

R(CX

x

, CY

x

) for X

x

, Y

x

∈

T

x

M and a smooth curve c in M from x

0

to x. We extend X

x

and Y

x

to vector ﬁelds X and Y ∈ X(M) with [X, Y ] = 0 near x. We may also

suppose that Z ∈ X(M) is a vector ﬁeld which extends c

(t) along c(t):

make c simple piecewise smooth by deleting any loop, reparametrize it

in such a way that it is again smooth (see the beginning of this proof)

12. Holonomy 79

and approximate it by an embedding. The vector ﬁelds ξ thus obtained

still generate the ﬁnite dimensional vector space g. So we have

ξ = (Fl

CZ

1

)

∗

R(CX, CY ) = (Fl

CZ

1

)

∗

[CX, CY ] since [X, Y ](x) = 0

= (Fl

CZ

1

)

∗ 1

2

d

2

dt

2

[

t=0

(Fl

CY

−t

◦ Fl

CX

−t

◦ Fl

CY

t

◦ Fl

CX

t

)

=

1

2

d

2

dt

2

[

t=0

(Fl

CZ

−1

Fl

CY

−t

◦ Fl

CX

−t

◦ Fl

CY

t

◦ Fl

CX

t

) Fl

CZ

1

),

where we used the well known formula expressing the Lie bracket of two

vector ﬁelds as the second derivative of the (group theoretic) commutator

of the ﬂows. The parallel transport in the last equation ﬁrst follows c

from x

0

to x, then follows a small closed parallelogram near x in M

(since [X, Y ] = 0 near x) and then follows c back to x

0

. This curve is

clearly nullhomotopic, so claim 3 follows.

Step 4. Now we make Hol(Φ, x

0

) into a Lie group which we call G, by

taking Hol

0

(Φ, x

0

) = G

0

as its connected component of the identity.

Then Hol(Φ, x

0

)/ Hol

0

(Φ, x

0

) is a countable group, since the fundamen-

tal group π

1

(M) is countable (by Morse Theory M is homotopy equiv-

alent to a countable CW-complex).

Step 5. Construction of a cocycle of transition functions with values in

G. Let (U

α

, u

α

: U

α

→R

m

) be a locally ﬁnite smooth atlas for M such

that each u

α

: U

α

→ R

m

) is surjective. Put x

α

:= u

−1

α

(0) and choose

smooth curves c

α

: [0, 1] → M with c

α

(0) = x

0

and c

α

(1) = x

α

. For

each x ∈ U

α

let c

x

α

: [0, 1] → M be the smooth curve t → u

−1

α

(t.u

α

(x)),

then c

x

α

connects x

α

and x and the mapping (x, t) → c

x

α

(t) is smooth

U

α

[0, 1] →M. Now we deﬁne a ﬁber bundle atlas (U

α

, ψ

α

: E[U

α

→

U

α

S) by ψ

−1

α

(x, s) = Pt(c

x

α

, 1) Pt(c

α

, 1) s. Then ψ

α

is smooth since

Pt(c

x

α

, 1) = Fl

CX

x

1

for a local vector ﬁeld X

x

depending smoothly on x.

Let us investigate the transition functions.

ψ

β

ψ

−1

α

(x, s) =

_

x, Pt(c

α

, 1)

−1

Pt(c

x

α

, 1)

−1

Pt(c

x

β

, 1) Pt(c

β

, 1) s

_

=

_

x, Pt(c

β

.C

x

β

.(c

x

α

)

−1

.(c

α

)

−1

, 4) s

_

=: (x, ψ

βα

(x) s), where ψ

βα

: U

βα

→G.

Clearly ψ

βα

: U

βα

S →S is smooth which implies that ψ

βα

: U

βα

→G

is also smooth. (ψ

αβ

) is a cocycle of transition functions and we use it

to glue a principal bundle with structure group G over M which we call

(P, p, M, G). From its construction it is clear that the associated bundle

P[S] = P

G

S equals (E, p, M, S).

Step 6. Lifting the connection Φ to P.

For this we have to compute the Christoﬀel symbols of Φ with respect

80 12. Holonomy

to the atlas of step 5. To do this directly is quite diﬃcult since we

have to diﬀerentiate the parallel transport with respect to the curve.

Fortunately there is another way. Let c : [0, 1] →U

α

be a smooth curve.

Then we have

ψ

α

(Pt(c, t)ψ

−1

α

(c(0), s)) =

=

_

c(t), Pt(c

−1

α

, 1) Pt((c

c(t)

α

)

−1

, 1) Pt(c, t) Pt(c

c(t)

α

, 1) Pt(c

α

, 1)s

_

= (c(t), γ(t).s),

where γ(t) is a smooth curve in the holonomy group G. Now let Γ

α

∈

Ω

1

(U

α

, X(S)) be the Christoﬀel symbol of the connection Φ with respect

to the chart (U

α

, ψ

α

). From the third proof of theorem 9.8 we have

ψ

α

(Pt(c, t)ψ

−1

α

(c(0), s)) = (c(t), ¯ γ(t, s),

where ¯ γ(t, s) is the integral curve through s of the time dependent vector

ﬁeld Γ

α

(

d

dt

c(t)) on S. But then we get

Γ

α

(

d

dt

c(t))(s) =

d

dt

¯ γ(t, s) =

d

dt

(γ(t).s) = (

d

dt

γ(t)).s,

where

d

dt

γ(t) ∈ g. So Γ

α

takes values in the Lie sub algebra of fun-

damental vector ﬁelds for the action of G on S. By theorem 11.9 the

connection Φ is thus induced by a principal connection ω on P. Since

by 11.8 the principal connection ω has the ”same” holonomy group as Φ

and since this is also the structure group of P, the principal connection

ω is irreducible, see 11.7.

81

13. The nonlinear frame bundle of a ﬁber bundle

13.1. Let now (E, p, M, S) be a ﬁber bundle and let us ﬁx a ﬁber bundle

atlas (U

α

) with transition functions ψ

αβ

: U

αβ

S →S. By 6.8 we have

C

∞

(U

αβ

, C

∞

(S, S)) ⊆ C

∞

(U

αβ

S, S) with equality if and only if S

is compact. Let us therefore assume from now on that S is compact.

Then we assume that the transition functions ψ

αβ

: U

αβ

→Diﬀ(S, S).

Now we deﬁne the nonlinear frame bundle of (E, p, M, S) as follows.

We consider the set Diﬀ¦S, E¦ :=

x∈M

Diﬀ(S, E

x

) and give it the

inﬁnite dimensional diﬀerentiable structure which one gets by applying

the functor Diﬀ(S, ) to the cocycle of transition functions (ψ

αβ

). Then

the resulting cocycle of transition functions for Diﬀ¦S, E¦ gives it the

structure of a smooth principal bundle over M with structure group

Diﬀ(M). The principal action is just composition from the right.

We can consider now the smooth action ev : Diﬀ(S) S → S and

the associated bundle Diﬀ¦S, E¦[S, ev] =

Diﬀ{S,E}×S

Diﬀ(S)

. The mapping

ev : Diﬀ¦S, E¦S →E is invariant under the Diﬀ(S)-action and factors

therefore to a smooth mapping Diﬀ¦S, E¦[S, ev] →E as in the following

diagram:

Diﬀ¦S, E¦ S

pr

−−−−→

Diﬀ{S,E}×S

Diﬀ(S)

ev

¸

¸

_

_

_

_

E ←−−−− Diﬀ¦S, E¦[S, ev].

The bottom mapping is easily seen to be a diﬀeomorphism. Thus the

bundle Diﬀ¦S, E¦ is in full right the (nonlinear) frame bundle of E.

13.2. Lemma. In the setting of 13.1 the inﬁnite dimensional smooth

manifold Diﬀ¦S, E¦ is a splitting smooth submanifold of Emb(S, E),

with the obvious embedding.

13.3. Let now Φ ∈ Ω

1

(E; TE) be a connection on E. We want to lift

Φ to a principal connection on Diﬀ¦S, E¦, and for this we need a good

description of the tangent space T Diﬀ¦S, E¦. With the methods of

[Michor, 1980] one can show that

T Diﬀ¦S, E¦ =

_

x∈M

¦f ∈ C

∞

(S, TE[E

x

) : Tp ◦ f = one point

in T

x

M and π

E

◦ f ∈ Diﬀ(S, E

x

)¦.

Starting from the connection Φ we can then consider ω(f) := T(π

E

◦

f)

−1

◦ Φ◦ f : S →TE →V E →TS for f ∈ T Diﬀ¦S, E¦. Then ω(f) is

a vector ﬁeld on S and we have

82 13. The nonlinear frame bundle of a ﬁber bundle

Lemma. ω ∈ Ω

1

(Diﬀ¦S, E¦; X(S)) is a principal connection and the

induced connection on E = Diﬀ¦S, E¦[S, ev] coincides with Φ.

Proof. This follows directly from 11.9. But we also give a direct proof.

The fundamental vector ﬁeld ζ

X

on Diﬀ¦S, E¦ for X ∈ X(S) is given

by ζ

X

(g) = Tg ◦ X. Then ω(ζ

X

(g)) = Tg

−1

◦ Φ ◦ Tg ◦ X = X since

Tg ◦ X has vertical values. So ω reproduces fundamental vector ﬁelds.

Now let h ∈ Diﬀ(S) and denote by r

h

the principal right action. Then

we have

((r

h

)

∗

ω)(f) = ω(T(r

h

)f) = ω(f ◦ h) = T(π

E

◦ f ◦ h)

−1

◦ Φf

.

◦ h

= Th

−1

◦ ω(f) ◦ h = Ad

Diﬀ(S)

(h

−1

)ω(f).

13.4 Theorem. Let (E, p, M, S) be a ﬁber bundle with compact stan-

dard ﬁber S. Then connections on E and principal connections on

Diﬀ¦S, E¦ correspond to each other bijectively, and their curvatures are

related as in 11.8. Each principal connection on Diﬀ¦S, E¦ admits a

global parallel transport. The holonomy groups and the restricted holo-

nomy groups are equal as subgroups of Diﬀ(S).

Proof. This follows directly from 11.8 and 11.9. Each connection on E

is complete since S is compact, and the lift to Diﬀ¦S, E¦ of its parallel

transport is the global parallel transport of the lift of the connection, so

the two last assertions follow.

13.5. Remark on the holonomy Lie algebra. Let M be connected,

let ρ = −dω−

1

2

[ω, ω]

X(S)

be the usual X(S)-valued curvature of the lifted

connection ω on Diﬀ¦S, E¦. Then we consider the R-linear span of all

elements ρ(ξ

f

, η

f

) in X(S), where ξ

f

, η

f

∈ T

f

Diﬀ¦S, E¦ are arbitrary

(horizontal) tangent vectors, and we call this span hol(ω). Then by the

Diﬀ(S)-equivariance of ρ the vector space hol(ω) is an ideal in the Lie

algebra X(S).

13.6. Lemma. Let f : S → E

x

0

be a diﬀeomorphism in Diﬀ¦S, E¦

x

0

.

Then f

∗

: X(S) → X(E

x

0

) induces an isomorphism between hol(ω) and

the R-linear span of all g

∗

R(CX, CY ), X, Y ∈ T

x

M, and g : E

x

)

→E

x

any diﬀeomorphism.

The proof is obvious. Note that the closure of f

∗

(hol(ω)) is (a priori)

larger than hol(Φ, x

0

) of 12.2. Which one is the right holonomy Lie

algebra?

83

14. Gauge theory for ﬁber bundles

We ﬁx the setting of section 13. In particular the standard ﬁber S is

supposed to be compact.

14.1. We consider the bundle Diﬀ¦E, E¦ :=

x∈M

Diﬀ(E

x

, E

x

) which

bears the smooth structure described by the cocycle of transition func-

tions Diﬀ(ψ

−1

αβ

, ψ

αβ

) = (ψ

αβ

)

∗

(ψ

βα

)

∗

, where (ψ

αβ

) is a cocycle of tran-

sition functions for the bundle (E, p, M, S).

14.2. Lemma. The associated bundle Diﬀ¦S, E¦[Diﬀ(S), conj] is iso-

morphic to the ﬁber bundle Diﬀ¦E, E¦.

Proof. The mapping A : Diﬀ¦S, E¦ Diﬀ(S) → Diﬀ¦E, E¦, given by

A(f, g) := f ◦g◦f

−1

: E

x

→S →S →E

x

for f ∈ Diﬀ(S, E

x

), is Diﬀ(S)-

invariant, so it factors to a smooth mapping Diﬀ¦S, E¦[Diﬀ(S)] →

Diﬀ¦E, E¦. It is bijective ans admits locally over M smooth inverses,

so it is a ﬁber respecting diﬀeomorphism.

14.3. The gauge group Gau(E) of the bundle (E, p, M, S) is by deﬁ-

nition the group of all principal bundle automorphisms of the Diﬀ(S)-

bundle (Diﬀ¦S, E¦ which cover the identity of M. The usual reasoning

gives that Gau(E) equals the space of all smooth sections of the asso-

ciated bundle Diﬀ¦S, E¦[Diﬀ(S), conj] which by 14.2 equals the space

of sections of the bundle Diﬀ¦E, E¦ →M. We equip it with the topol-

ogy and diﬀerentiable structure as a space of smooth sections, see 6.1.

Since only the image space is inﬁnite dimensional but admits exponen-

tial mappings (induced from the ﬁnite dimensional ones) this makes no

diﬃculties. Since the ﬁber S is compact we see from a local (on M) appli-

cation of the exponential law 6.8 that C

∞

(Diﬀ¦E, E¦ →M) →Diﬀ(E)

is an embedding of a splitting closed submanifold.

14.4. Theorem. The gauge group Gau(E) = C

∞

(Diﬀ¦E, E¦) is a

splitting closed subgroup of Diﬀ(E), if S is compact. It has an exponen-

tial mapping which is not surjective on any neighborhood of the identity.

Its Lie algebra consists of all vertical vector ﬁelds with compact support

on E, with the negative of the usual Lie bracket.

Proof. The ﬁrst statement has already been shown before the theo-

rem. A curve through the identity of principal bundle automorphisms

of Diﬀ¦S, E¦ → M is a smooth curve through the identity in Diﬀ(E)

consisting of ﬁber respecting mappings. the derivative of such a curve is

thus an arbitrary vertical vector ﬁeld with compact support. The space

of all these is therefore the Lie algebra of the gauge group, with the

negative of the usual Lie bracket.

84 14. Gauge theory for diﬀeomorphism groups

The exponential mapping is given by the ﬂow operator of such vector

ﬁelds. Since on each ﬁber it is just isomorphic the exponential mapping

of Diﬀ(S), it has all the properties of the latter.

14.5. Remark. If S is not compact we may circumvent the nonlinear

frame bundle, and we may deﬁne the gauge group Gau(E) directly as the

splitting closed subgroup of Diﬀ(E) which consists of all ﬁber respecting

diﬀeomorphisms which cover the identity of M. The Lei algebra of

Gau(E) consists then of all vertical vector ﬁelds on E with compact

support on E.

14.6 The space of connections. Let J

1

(E) →E be the aﬃne bundle

of 1-jets of sections of E → M. We have J

1

(E) = ¦ ∈ L(T

x

M, T

u

E) :

Tp ◦ = Id

T

x

M

, u ∈ E, p(u) = x¦. Then a section of J

1

(E) →E is just

a horizontal lift mapping TM

M

E →TE which is ﬁber linear over E,

so it describes a connection as treated in 9.2 and we may view the space

of sections C

∞

(J

1

(E) →E) as the space of all connections.

14.7. Theorem. The action of the gauge group Gau(E) on the space

of connections C

∞

(J

1

(E)) is smooth.

Proof. This follows from 6.6

14.8. We will now give a diﬀerent description of the action. We view

now a connection Φ again as a linear ﬁber wise projection TE → V E,

So the space of connections is now Conn(E) := ¦Φ ∈ Ω

1

(E; TE) : Φ ◦

Φ = Φ, Φ(TE) = V E¦. Since S is compact the canonical isomorphism

Conn(E) → C

∞

(J

1

(E)) is even a diﬀeomorphism. Then the action of

f ∈ Gau(E) ⊂ Diﬀ(E) on Φ ∈ Conn(E) is given by f

∗

Φ = (f

−1

)

∗

Φ =

Tf ◦ Φ ◦ Tf

−1

. Now it is very easy to describe the inﬁnitesimal action.

Let X be a vertical vector ﬁeld with compact support on E and consider

its global ﬂow Fl

X

t

.

Then we have

d

dt

[

0

(Fl

X

t

)

∗

Φ = L

X

Φ = [X, Φ], the Fr¨olicher Nijenhuis

bracket, by 8.16(5). The tangent space of Conn(E) at Φ is the space

T

Φ

Conn(E) = ¦Ψ ∈ Ω

1

(E; TE) : Ψ[V E = 0¦. The ”inﬁnitesimal orbit”

at Φ in T

Φ

Conn(E) is ¦[X, Φ] : X ∈ C

∞

c

(V E)¦.

The isotropy subgroup of a connection Φ is ¦f ∈ Gau(E) : f

∗

Φ = Φ¦.

Clearly this just the group of all those f which respect the horizontal

bundle HE = ker Φ. The most interesting object is of course the orbit

space Conn(E)/ Gau(E). If the base manifold M is compact then one

can show that the orbit space is stratiﬁed into smooth manifolds, each

one corresponding to a conjugacy class of holonomy groups in Diﬀ(S).

Those strata whose holonomy group groups are up to conjugacy con-

tained in a ﬁxed compact group (like SU(2)) acting on S are diﬀeo-

morphic to the strata of the usual ﬁnite dimensional gauge theory. The

14. Gauge theory for diﬀeomorphism groups 85

proof of this assertions is quite complicated and will be dealt with in

another paper, see [Gil-Medrano Michor, 1989] for the starting idea.

86

15. A classifying space

for the diﬀeomorphism group

15.1. Let

2

be the Hilbert space of square summable sequences and

let S be a compact manifold. By a slight generalization of theorem

7.3 (we use a Hilbert space instead of a Riemannian manifold N) the

space Emb(S,

2

) of all smooth embeddings is an open submanifold of

C

∞

(S,

2

) and it is also the total space of a smooth principal bundle

with structure group Diﬀ(S) acting from the right by composition. The

base space B(S,

2

) := Emb(S,

2

)/ Diﬀ(S) is a smooth manifold mod-

eled on Fr´echet spaces which are projective limits of Hilbert spaces.

B(S,

2

) is a Lindel¨of space in the quotient topology and the model

spaces admit bump functions, thus B(S,

2

) admits smooth partitions of

unity. We may view B(S,

2

) as the space of all submanifolds of

2

which

are diﬀeomorphic to S, a nonlinear analogy of the inﬁnite dimensional

Grassmanian.

15.2. Lemma. The total space Emb(S,

2

) is contractible.

So by the general theory of classifying spaces the base space B(S,

2

)

is a classifying space of Diﬀ(S). I will give a detailed description of the

classifying process in 15.4 below.

Proof. We consider the continuous homotopy A :

2

[0, 1] →

2

through

isometries which is given by A

0

= Id and by

A

t

(a

0

, a

1

, a

2

, . . . ) = (a

0

, . . . , a

n−2

, a

n−1

cos θ

n

(t), a

n−1

sin θ

n

(t),

a

n

cos θ

n

(t), a

n

sin θ

n

(t), a

n+1

cos θ

n

(t), a

n+1

sin θ

n

(t), . . . )

for

1

n+1

≤ t ≤

1

n

, where θ

n

(t) = ϕ(n((n + 1)t −1))

π

2

for a ﬁxed smooth

function ϕ : R →R which is 0 on (−∞, 0], grows monotonically to 1 in

[0, 1], and equals 1 on [1, ∞)].

Then A

1/2

(a

0

, a

1

, a

2

, . . . ) = (a

0

, 0, a

1

, 0, a

2

, 0, . . . ) is in

2

even

and on

the other hand A

1

(a

0

, a

1

, a

2

, . . . ) = (0, a

0

, 0, a

1

, 0, a

2

, 0, . . . ) is in

2

odd

.

The same homotopy makes sense as a mapping A : R

∞

R → R

∞

,

and here it is easily seen to be smooth: a smooth curve in R

∞

is locally

bounded and thus takes locally values in a ﬁnite dimensional subspace

R

N

⊂ R

∞

. The image under A then has values in R

2N

⊂ R

∞

and the

expression is clearly smooth as a mapping into R

2N

. This is a variant

of a homotopy constructed by [Ramadas, 1982].

Given two embeddings e

1

and e

2

∈ Emb(S,

2

) we ﬁrst deform e

1

through embeddings to e

1

∈ Emb(S,

2

even

), and e

2

to e

2

∈ Emb(S,

2

odd

);

then we connect them by te

1

+ (1 − t)e

2

which is a smooth embedding

for all t since the values are always orthogonal.

15. Classifying spaces for diﬀeomorphism groups 87

15.3. We consider the smooth action ev : Diﬀ(S) S → S and the

associated bundle Emb(S,

2

)[S, ev] = Emb(S,

2

)

Diﬀ(S)

S which we

call E(S,

2

), smooth ﬁber bundle over B(S,

2

) with standard ﬁber S. In

view of the interpretation of B(S,

2

) as the nonlinear Grassmannian we

may visualize E(S,

2

) as the ”universal S-bundle” as follows: E(S,

2

) =

¦(N, x) ∈ B(S,

2

)

2

: x ∈ N¦ with the diﬀerentiable structure from

the embedding into B(S,

2

)

2

.

The tangent bundle TE(S,

2

) is then the space of all (N, x, ξ, v) where

N ∈ B(S,

2

), x ∈ N, ξ is a vector ﬁeld along and normal to N in

2

, and

v ∈ T

x

2

such that the part of v normal to T

x

N equals ξ(x). This follows

from the description of the principal ﬁber bundle Emb(S,

2

) →B(S,

2

)

given in 7.3 combined with 6.7. Obviously the vertical bundle V E(S,

2

)

consists of all (N, x, v) with x ∈ N and v ∈ T

x

N. The orthonormal

projection p

(N,x)

:

2

→ T

x

N deﬁnes a connection Φ

class

: TE(S,

2

) →

V E(S,

2

) which is given by Φ

class

(N, x, ξ, v) = (N, x, p

(N,x)

v). It will

be called the classifying connection for reasons to be explained in the

next theorem.

15.4. Theorem. Classifying space for Diﬀ(S). The ﬁber bundle

(E(S,

2

), pr, B(S,

2

), S) is classifying for S-bundles and Φ

class

is a clas-

sifying connection:

For each ﬁnite dimensional bundle (E, p, M, S) and each connection

Φ on E there is a smooth (classifying) mapping f : M →B(S,

2

) such

that (E, Φ) is isomorphic to (f

∗

E(S,

2

), f

∗

Φ

class

). Homotopic maps pull

back isomorphic S-bundles and conversely (the homotopy can be chosen

smooth). The pulled back connection is invariant under a homotopy

H if and only if i(C

class

T

(x,t)

H.(0

x

,

d

dt

))R

class

= 0 where R

class

is the

curvature of Φ

class

.

Since S is compact the classifying connection Φ

class

is complete and

its parallel transport Pt

class

has the following classifying property:

˜

f ◦ Pt

f

∗

Φ

class

(c, t) = Pt

class

(f ◦ c, t) ◦

˜

f,

where

˜

f : E

∼

= F

∗

E(S,

2

) → E(S,

2

) is the ﬁberwise diﬀeomorphic

which covers the classifying mapping f : M →B(S,

2

).

Proof. We choose a Riemannian metric g

1

on the vector bundle V E →E

and a Riemannian metric g

2

on the manifold M. We can combine these

two into the Riemannian metric g := (Tp[ ker Φ)

∗

g

2

⊕g

1

on the manifold

E, for which the horizontal and vertical spaces are orthogonal. By the

theorem of Nash there is an isometric imbedding h : E → R

N

for N

large enough. We then embed R

N

into the Hilbert space

2

and consider

88 15. Classifying spaces for diﬀeomorphism groups

f : M →B(S,

2

), given by f(x) = h(E

x

). Then

E

˜

f=(f,h)

−−−−−→ E(S,

2

)

p

¸

¸

_

¸

¸

_

pr

M −−−−→

f

B(S,

2

)

is ﬁberwise a diﬀeomorphism, so this diagram is a pullback and we have

f

∗

E(S,

2

) = E. Since T(f, h) maps horizontal and vertical vectors to

orthogonal ones, (f, h)

∗

Φ

class

= Φ. If Pt denotes the parallel transport

of the connection Φ and c : [0, 1] →M is a (piecewise) smooth curve we

have for u ∈ E

c(0))

Φ

class ∂

∂t

¸

¸

0

˜

f(Pt(c, t, u)) = Φ

class

.T

˜

f.

∂

∂t

¸

¸

0

Pt(c, t, u)

= T

˜

f.Φ.

∂

∂t

¸

¸

0

Pt(c, t, u) = 0, so

˜

f(Pt(c, t, u)) = Pt

class

(f ◦ c, t,

˜

f(u)).

Now let H be a continuous homotopy M I → B(S,

2

). Then

we may approximate H by smooth mappings with the same H

0

and

H

1

, if they are smooth, see [Br¨ocker-J¨anich, 1973], where the infnite

dimensionality of B(S,

2

) does not disturb. Then we consider the bundle

H

∗

E(S,

2

) → M I, equipped with the connection H

∗

Φ

class

, whose

curvature is H

∗

R

class

. Let ∂

t

be the vector ﬁeld tangential to all ¦x¦I

on M I. Parallel transport along the lines t → (x, t) with respect

H

∗

Φ

class

is given by the ﬂow of the horizontal lift (H

∗

C

class

)(∂

t

) of ∂

t

.

Let us compute its action on the connection H

∗

Φ

class

whose curvature

is H

∗

R

class

by 9.5.(3). By lemma 9.9 we have

∂

∂t

_

Fl

(H

∗

C

class

)(∂

t

)

t

_

∗

H

∗

Φ

class

= −

1

2

i

(H

∗

C

class

)(∂

t

)

(H

∗

R

class

)

= −

1

2

H

∗

_

i(C

class

T

(x,t)

H.(0

x

,

d

dt

))R

class

_

which implies the result.

15.5. Theorem. [Ebin-Marsden, 1970] Let S be a compact orientable

manifold, let µ

0

be a positive volume form on S with total mass 1. Then

Diﬀ(S) splits smoothly into Diﬀ(S) = Diﬀ

µ

0

(S)Vol(S) where Diﬀ

µ

0

(S)

is the Lie group of all µ

0

-preserving diﬀeomorphisms and Vol(S) is the

space of all volume forms of total mass 1.

Proof. We show ﬁrst that there exists a smooth mapping τ : Vol(S) →

Diﬀ(S) such that τ(µ)

∗

µ

0

= µ.

15. Classifying spaces for diﬀeomorphism groups 89

We put µ

t

= µ

0

+t(µ−µ

0

). We want a smooth curve t →f

t

∈ Diﬀ(S)

with f

∗

t

µ

t

= µ

0

. We have

∂

∂t

f

t

= X

t

◦ f

t

for a time dependent vector

ﬁeld X

t

on S. Then 0 =

∂

∂t

f

∗

t

µ

t

= f

∗

t

L

X

t

µ

t

+f

∗

t

∂

∂t

µ

t

= f

∗

t

(L

X

t

µ

t

+(µ =

µ

0

)), so L

X

t

µ

t

= µ

0

− µ and consequently

_

S

L

X

t

µ

t

=

_

S

(µ

0

− µ) = 0.

So the cohomology class of L

X

t

µ

t

in H

dimS

(S) is zero, and we have

L

X

t

µ

t

= di

X

t

µ

t

+i

X

t

0 = dω for some ω ∈ Ω

dimS−1

(S). Now we choose

ω such that dω = µ

0

− µ and we choose it smoothly depending on µ

by the theorem of Hodge. Then the time dependent vector ﬁeld X

t

is

uniquely determined by i

X

t

µ

t

= ω since µ

t

is nowhere 0. Let f

t

be the

evolution operator of X

t

and put τ(µ) = f

−1

1

.

Now we may prove the theorem proper. We deﬁne a mapping Ψ :

Diﬀ(S) → Diﬀ

µ

0

(S) Vol(S) by Ψ(f) := (f ◦ τ(f

∗

µ

0

)

−1

, f

∗

µ

0

), which

is smooth by sections 6 and 7. An easy computation shows that the

inverse is given by the smooth mapping Ψ

−1

(g, µ) = g ◦ τ(µ).

15.6. A consequence of theorem 15.5 is that the classifying spaces of

Diﬀ(S) and Diﬀ

µ

0

(S) are homotopy equivalent. So their classifying

spaces are also homotopy equivalent.

We now sketch a smooth classifying space for Diﬀ

µ

0

. Consider the

space B

1

(S,

2

) of all submanifolds of

2

of type S and total volume

1 in the volume form induced from the inner product on

2

. It is a

closed splitting submanifold of codimension 1 of B(S,

2

) by the Nash-

Moser inverse function theorem, see [Hamilton, 1982]. This theorem is

applicable if we use

2

as image space, because the modeling spaces are

then tame Fr´echet spaces in his sense. It is not applicable directly for

R

∞

as image space.

15.7. Theorem. Classifying space for Diﬀ

ω

(S). Let S be a com-

pact real analytic manifold. Then the space Emb

ω

(S,

2

) of real analytic

embeddings of S into the Hilbert space

2

is the total space of a real

analytic principal ﬁber bundle with structure group Diﬀ

ω

(S) and real

analytic base manifold B

ω

(S,

2

), which is a classifying space for the Lie

group Diﬀ

ω

(S). It carries a universal Diﬀ

ω

(S)-connection.

In other words:

(Emb

ω

(S, N)

Diﬀ

ω

(S)

S →B

ω

(S,

2

)

classiﬁes ﬁber bundles with typical ﬁber S and carries a universal (gen-

eralized) connection.

The proof is similar to that of 15.4 with the appropriate changes to

C

ω

.

90

16. A characteristic class

16.1. Bundles with ﬁber volumes. Let (E, p, M, S) be a ﬁber bun-

dle and let µ be a smooth section of the line bundle Λ

n

V

∗

E →E where

n = dimS, such that for each x ∈ M the n-form µ

x

:= µ[E

x

is a positive

volume form of total mass 1. Such a form will be called a ﬁber volume on

E. Such forms exist on any oriented bundle with compact oriented ﬁber

type. We may plug n vertical vector ﬁelds X

i

∈ C

∞

(V E) into µ to get

a function µ(X

1

, . . . , X

n

) on E, but we cannot treat µ as a diﬀerential

form on E.

16.2. Lemma. Let µ

0

be a ﬁxed volume form on S and let µ be a ﬁber

volume on the bundle E. Then there is a bundle atlas (U

α

, ψ

α

: E[U

α

→

U

α

S) such that (ψ

α

[E

x

)

∗

µ

0

= µ

x

for all x ∈ U

α

.

Proof. We start with any bundle atlas (U

α

, ψ

α

: E[U

α

→ U

α

S).

By theorem 15.5 there is a diﬀeomorphism ϕ

α,x

∈ Diﬀ(S) such that

(ϕ

α,x

)

∗

((ψ

α

)

−1

[¦x¦ S)

∗

µ

x

= µ

0

, and ϕ

α,x

depends smoothly on x.

Then ψ

α

:= (Id

U

α

ϕ

−1

α,x

) ◦ ψ

α

: E[U

α

→ U

α

S is the desired bundle

chart.

16.3. Let now Φ be a connection on (E, p, M, S) and let C : TM

M

E → TE be its horizontal lifting. Then for X ∈ X(M) the ﬂow Fl

CX

t

of its horizontal lift CX respects ﬁbers, so for the ﬁber volume µ the

pullback (Fl

CX

t

)

∗

µ makes sense and is again a ﬁber volume on E. We

can now deﬁne a kind of covariant derivative by

(a) ∇

Φ

X

µ :=

d

dt

[

0

(Fl

CX

t

)

∗

µ.

In fact ∇

Φ

is a covariant derivative on the vector bundle Ω

n

(p) :=

Ω

n

ver

(E) :=

x∈M

Ω

n

(E

x

) with its obvious smooth vector bundle struc-

ture, whose standard ﬁber is the nuclear Fr´echet space Ω

n

(S).

We shall need a description of the convariant derivative ∇

Φ

in terms

of Lie derivatives. So let j : V E →TE be the embedding of the vertical

bundle and let us consider j

∗

: Ω

k

(E) → C

∞

(Λ

k

V

∗

E). Likewise we

consider Φ

∗

: C

∞

(Λ

k

V

∗

E) → Ω

k

(E). These are algebraic mappings,

algebra homomorphisms for the wedge product, and satisfy j

∗

◦ Φ

∗

=

Id

C

∞

(Λ

k

V

∗

E)

. The other composition Φ

∗

◦ j

∗

: Ω

k

(E) → Ω

k

(E) is a

projection onto the image of Φ

∗

.

If X ∈ X(M) is a vector ﬁeld on M, then the horizontal lift CX ∈

X(E) and X are p-related, i.e. Tp ◦ CX = X ◦ p. Thus the ﬂows are

p-conjugated: p ◦ Fl

CX

t

= Fl

X

t

◦p. So Fl

CX

t

respects the vertical bundle

16. Characteristic classes for general ﬁber bundles 91

V E and we have j ◦ Fl

CX

t

= Fl

CX

t

◦j. Therefore we have

∇

Φ

X

µ =

∂

∂t

¸

¸

0

(Fl

CX

t

)

∗

µ =

∂

∂t

¸

¸

0

(Fl

CX

t

)

∗

j

∗

Φ

∗

µ

(b)

= j

∗ ∂

∂t

¸

¸

0

(Fl

CX

t

)

∗

Φ

∗

µ = j

∗

L

CX

Φ

∗

µ = 0 +j

∗

i

CX

dΦ

∗

µ,

where L

CX

is the usual Lie derivative. This formula also shows that

∇

Φ

X

µ is C

∞

(M, R)-linear in X, as asserted.

Now let (U

α

, ψ

α

) be a ﬁber bundle atlas for (E, p, M, S). Put Φ

α

:=

(ψ

−1

α

)

∗

Φ, a connection on U

α

S → U

α

. Its horizontal lift is given by

C

α

(X) = (X, Γ

α

(X)). Let µ be a ﬁber k-form on U

α

S, so µ : U

α

→

Ω

k

(S) is just a smooth mapping. Thus we have for x ∈ U

α

(∇

Φ

α

X

µ) =

∂

∂t

¸

¸

0

((Fl

Γ

α

(X)

)

∗

(µ ◦ Fl

X

t

)) (c)

= d

U

α

µ(X) +L

S

Γ

α

(X)

µ

= X(µ) +L

S

Γ

α

(X)

µ

where d

U

α

is the exterior derivative of Ω

k

(S)-valued forms on U

α

, and

where L

S

is the Lie derivative on S.

16.4. Lemma. Let Φ be a connection on the bundle (E, p, M, S) with

curvature R ∈ Ω

2

(E; V E). Then the curvature of the linear covari-

ant derivative ∇

Φ

on the vector bundle (Ω

k

(p), M, Ω

k

(S)) is given by

R(∇

Φ

)(X, Y )µ = L

v

R(CX,CY )

µ, where L

v

denotes the vertical Lie deriv-

ative, on each ﬁber separately.

Proof. The easiest proof is local. Let (U

α

, ψ

α

) be a ﬁber bundle atlas for

(E, p, M, S). Put again Φ

α

:= (ψ

−1

α

)

∗

Φ, a connection on U

α

S → U

α

with horizontal lift is given by C

α

(X) = (X, Γ(X)). Let µ be a ﬁber

k-form on U

α

S, so µ : U

α

→ Ω

k

(S) is just a smooth mapping. Thus

we have for X, Y ∈ X(U

α

), where we write ∇

α

for ∇

Φ

α

,

R(∇

α

)(X, Y )µ = (∇

α

X

∇

α

Y

−∇

α

Y

∇

α

X

−∇

α

[X,Y ]

)µ

= ∇

α

X

(Y (µ) +L

S

Γ(Y )

µ) −∇

α

Y

(X(µ) +L

S

Γ(Y )

µ) −[X, Y ](µ) −L

S

Γ[X,Y ]

µ

= X(Y (µ)) +L

S

XΓ(Y )

µ +L

S

Γ(Y )

X(µ) +L

S

Γ(X)

Y (µ) +L

S

Γ(X)

L

S

Γ(Y )

µ

−Y (X(µ)) −L

S

Y Γ(X)

µ −L

S

Γ(X)

Y (µ) −L

S

Γ(Y )

X(µ) −L

S

Γ(Y )

L

S

Γ(X)

µ

−[X, Y ](µ) −L

S

Γ[X,Y ]

µ

= L

S

(XΓ(Y ) −Y Γ(X) −Γ([X, Y ])) µ + [L

S

Γ(X)

, L

S

Γ(Y )

]µ

= L

S

_

d

U

α

Γ(X, Y ) + [Γ(X), Γ(Y )]

_

µ

= L

S

(((ψ

−1

α

) ∗ R)(X, Y ))µ.

92 16. Characteristic classes for general ﬁber bundles

In this computation we used that L

S

: X(S)Ω

k

(S) →Ω

k

(S) is bilinear

and continuous so that the product rule of diﬀerential calculus applies.

The last formula comes from lemma 9.7.

16.5. Deﬁnition. The connection Φ on E is called µ-respecting con-

nection if ∇

Φ

X

µ = 0 for all X ∈ X(M).

Lemma.

(1) A connection Φ is µ-respecting if and only if for some (any) bun-

dle atlas (U

α

, ψ

α

) as in lemma 16.2 the corresponding Christoﬀel

forms Γ

α

take values in the Lie algebra X

µ

0

(S) of divergence free

vector ﬁelds on S (i.e. L

X

µ

0

= 0).

(2) There exist many µ-respecting connections.

(3) If Φ is µ-respecting and if R is its curvature, then for X

x

and

Y

x

∈ T

x

M the form i(R(CX

x

, CY

x

))µ is closed in ω

n−1

(E

x

).

Proof. 1. Put Φ

α

:= (ψ

−1

α

)

∗

Φ, a connection on U

α

S → U

α

. Its hor-

izontal lift is given by C

α

(X) = (X, Γ

α

(X)). Thus we have ∇

Φ

α

X

µ

0

=

L

Γ

α

(X)

µ

0

, and this is zero for all X ∈ X(U

α

) if and only if Γ

α

∈

Ω

1

(U

α

; X

µ

0

(S)).

2. By the ﬁrst part µ-respecting connections exist locally and since

∇

Φ

X

is C

∞

(M, R)-linear in X ∈ X(M) we may glue them via a partition

of unity on M.

3. We compute locally in a chart (U

α

, ψ

α

) belonging to a ﬁber bundle

atlas satisfying lemma 16.2. Then by 1. we have Γ

α

∈ Ω

1

(U

α

; X

µ

0

(S)),

so by lemma 9.7 the local expression of the curvature R

α

= (ψ

−1

α

)

∗

R =

d

U

α

Γ

α

+ [Γ

α

, Γ

α

] is an element of Ω

2

(U

α

; X

µ

0

(S)). But then

d

S

i

S

R

α

(X,Y )

µ

0

= L

S

R

α

(X,Y )

µ

0

= 0.

16.6. We consider the cohomology class [i(R(CX, CY ))µ] ∈ H

n−1

(E

x

)

for X, Y ∈ T

x

M and the ﬁnite dimensional vector bundle H

n−1

(p) :=

x∈M

H

n−1

(E

x

) → M. It is described by a cocycle of transition func-

tions U

αβ

→ GL(H

n−1

(S)), x → H

n−1

(ψ

βα

(x, )), which are locally

constant by the homotopy invariance of cohomology. So the vector bun-

dle H

n−1

(p) → M admits a unique ﬂat linear connection ∇ respecting

the resulting discrete structure group, and the induced covariant exterior

derivative d

∇

: Ω

(

M; H

n−1

(p)) → Ω

k+1

(M; H

n−1

(p)) satisﬁes d

2

∇

= 0

and deﬁnes the De Rham cohomology of M with twisted coeﬃcient do-

main H

n−1

(p).

We may view [i

R

µ] : X

x

, Y

x

→ [i(R(CX

x

, CY

x

))µ] as an element of

Ω

2

(M; H

n−1

(p)).

16. Characteristic classes for general ﬁber bundles 93

16.7. Lemma.

(1) If γ ∈ C

∞

(Λ

k

V

∗

E → E) induces a section [γ] : M → H

k

(p)

then we have ∇

X

[γ] = [∇

Φ

X

γ], for any connection Φ on E.

(2) If Φ is a µ-respecting connection on E then d

∇

[i

R

µ] = 0.

Proof. (1). For any bundle atlas (U

α

, ψα) the parallel sections of the

vector bundle H

k

(p) for the unique ﬂat connection respecting the dis-

crete structure group are exactly those which in each local bundle chart

U

α

H

k

(S) are given by the (locally) constant mappings: U

α

→H

k

(S).

Obviously ∇

Φ

X

[γ] := [∇

Φ

X

γ] deﬁnes a connection and in the local bun-

dle chart U

α

→H

k

(S) we have

∇

Φ

α

X

γ

α

=

d

dt

[

0

(Fl

(X,Γ

α

(X))

t

)

∗

γ

α

= d

U

α

γ

α

(X) +L

S

Γ

α

(X)

γ

α

,

where d

U

α

is the exterior derivative on U

α

of Ω

k

-valued mappings. But

if [γ] is parallel for the ﬂat connection respecting the discrete structure

group, so [γ

α

] is locally constant, then d

U

α

γ

α

takes values in the space

B

k

(S) of exact forms. And since γ

α

(x) is closed for all x ∈ U

α

the

second summand L

S

Γ

α

(X)

γ

α

= d

S

i

Γ

α

(X)

γ + i

Γ

α

(X)

0 takes values in the

space of exact forms anyhow. So the two connections have the same

local parallel sections, so they coincide.

(2). Let X

0

, X

1

, X

2

∈ X(M). Then we have

d

∇

[i

R

µ](X

0

, X

1

, X

2

) =

=

cyclic

(∇

X

0

([i

R

µ](X

1

, X

2

)) −[i

R

µ]([X

0

, X

1

], X

2

))

=

cyclic

_

∇

X

0

(i

R(CX

1

,CX

2

)

µ) −i

R(C[X

0

,X

1

],CX

2

)

µ

¸

=

cyclic

_

j

∗

L

CX

0

i

R(CX

1

,CX

2

)

Φ

∗

µ −i

R(C[X

0

,X

1

],CX

2

)

µ

¸

=

_

j

∗

i

[CX

0

,R(CX

1

,CX

2

)]

Φ

∗

µ+

+j

∗

i

R(CX

1

,CX

2

)

L

CX

0

Φ

∗

µ −i

R(C[X

0

,X

1

],CX

2

)

µ

¸

=

_

cyclic

j

∗

i

_

[CX

0

, R(CX

1

, CX

2

)] + 0 −R(C[X

0

, X

1

], CX

2

)

_

Φ

∗

µ

¸

,

where we used [i

CX

, L

Z

] = i

[CX,Z]

, that j

∗

Φ

∗

= Id, and that R has

vertical values, which implies

j

∗

i

R(CX

1

,CX

2

)

L

CX

0

Φ

∗

µ = i

R(CX

1

,CX

2

)

j

∗

L

CX

0

Φ

∗

µ

= i

R(CX

1

,CX

2

)

∇

Φ

X

0

µ = 0.

94 16. Characteristic classes for general ﬁber bundles

Now we need the Bianchi identity [R, Φ] = 0 from 9.4. Writing out the

global formula 8.9 for the horizontal vector ﬁelds CX

i

we get

0 = [R, Φ](CX

0

, CX

1

, CX

2

)

=

cyclic

(−Φ[CX

0

, R(CX

1

, CX

2

)] −R(C[X

0

, X

1

], CX

2

)).

From this it follows that d

∇

[i

R

µ] = 0.

16.8. Deﬁnition. So we may deﬁne a characteristic class k(E, µ) of

the bundle (E, p, M, S) with ﬁber volume µ as the class

k(E, µ) := [[i

R

µ]

H

n−1

(p)

]

H

2

(M;H

n−1

(p))

.

16.9. Lemma. The class k(E, µ) is independent of the choice of the

µ-respecting connection.

Proof. Let Φ and Φ

**be two µ preserving connections on (E, p, M), and
**

let C and C

be their horizontal liftings. We put D = C −C

: TM

M

E → V E, which is ﬁber linear over E, then C

t

= C + tD = (1 −

t)C + tC

**is a curve of horizontal lifts which preserve µ. In fact we
**

have L

D(X

x

)

µ[E

x

= 0 for each X

x

∈ T

x

M. Let Φ

t

be the connection

corresponding to C

t

and let R

t

be its curvature. Then we have

R

t

(C

t

X, C

t

Y ) = Φ

t

[C

t

X, C

t

Y ] = [C

t

X, C

t

Y ] −C

t

[X, Y ]

= R(CX, CY ) +t[CX, DY ] +t[DX, CY ] +t

2

[DX, DY ] −tD[X, Y ].

∂

∂t

¸

¸

0

R

t

(C

t

X, C

t

Y ) = [CX, DY ] + [DX, CY ] + 2t[DX, DY ] −D[X, Y ].

∂

∂t

¸

¸

0

i(R

t

(C

t

X, C

t

Y ))µ = j

∗

_

∂

∂t

¸

¸

0

i(R

t

(C

t

X, C

t

Y ))

_

Φ

∗

µ

= j

∗

_

i

[CX,DY ]

−i

[CY,DX]

+ 2ti

[DX,DY ]

−i

D[X,Y ]

_

Φ

∗

µ

= j

∗

_

[L

CX

, i

DY

] −[L

CY

, i

DX

] + 2t[L

DX

, i

DY

] −i

D[X,Y ]

_

Φ

∗

µ.

= j

∗

_

L

CX

i

DY

−0 −L

CY

i

DX

−i

D[X,Y ]

+ 2t(di

DX

i

DY

−0)

_

Φ

∗

µ.

Here we used that D has vertical values which implies j

∗

i

DY

L

CX

Φ

∗

µ =

i

DY

j

∗

L

CX

Φ

∗

µ = i

DY

∇

Φ

X

µ = 0.

Next we use lemma 16.7.1 to compute the diﬀerential of the of the

1-form [i

D

µ] ∈ Ω

1

(M; H

n−1

(p)).

d

∇

[i

D

µ](X, Y ) = ∇

X

[i

D(Y )

µ] −∇

Y

[i

D(X)

µ] −[i

D([X,Y ])

µ]

= [∇

Φ

X

i

D(Y )

µ −∇

Φ

Y

i

D(X)

µ −i

D([X,Y ])

µ]

= [j

∗

(L

CX

i

D(Y )

−L

CY

i

D(X)

−i

D([X,Y ])

)Φ

∗

µ].

So we get

∂

∂t

¸

¸

0

[i

R

t

µ] = d

∇

[i

D

µ] in the space Ω

2

(M; H

n−1

(p)).

16. Characteristic classes for general ﬁber bundles 95

16.10. Remarks. The idea of this class is originally due to [Kainz,

1985].

From the point of view of algebraic topology it is not very interest-

ing since it coincides with the second diﬀerential in the Serre spectral

sequence of the ﬁbration E →M of the class corresponding to µ.

16.11. Review of linear connections on vector bundles.

Let (V, p, M, R

n

) be a vector bundle equipped with a ﬁber metric g. Let

∇ be a linear covariant derivative on E which respects g. Let s = (s

i

)

be a local frame ﬁeld over U ⊂ M of E: so s

i

∈ C

∞

(E[U) and the s

i

(x)

are a linear basis of E

x

for each x ∈ U.

Then s deﬁnes a vector bundle chart ψ : E[U → U R

n

by the

prescription ψ(

s

i

u

i

) = (u

i

) or ψ(s.u) = u. If s

**is another frame ﬁeld,
**

we have s

i

=

j

s

j

h

j

i

or s

= s.h for a function h ∈ C

∞

(U ∩U

, GL(n)).

For the vector bundle charts we have then ψ

ψ

−1

(x, u) = (x, h(x).u).

We also have Tψ : T(E[U) →TU R

n

R

n

, and for the chart change

we have

T(ψ

◦ ψ

−1

) : T(U ∩ U

) R

n

R

n

→T(U ∩ U

) R

n

R

n

T(ψ

◦ ψ

−1

)(X

x

, v, w) = (X

x

, h(x).v, dh(X

x

).v +h(x).w)

The connection form is deﬁned by ∇

X

s

i

=

j

s

j

ω

j

i

(X) or ∇s = s.ω

for ω ∈ Ω

1

(U, gl(n)) in general and ω ∈ Ω

1

(U, o(n)) if ∇ respects g and

the frame ﬁeld s is orthonormal. If ∇s

= s

.ω

**is the connection form for
**

another frame s

= s.h, we have the transition formula h.ω

= ω.h +dh.

For a general section s.u =

j

s

j

u

j

we have ∇(s.u) = s.ω.u +s.du.

Now we want to express the connection Φ = Φ

∇

∈ Ω

1

(E; V E), the

horizontal lift C : TM

M

E → TE and the connector K = vpr ◦ Φ :

TE → E locally in terms of the connection form (compare with 11.10

and following). Since for a local section x → (x, u(x)) of U R

n

→ U

we have

∇

X

(s.u) = s.(ω(X).u +du(X))

= K ◦ T(s.u) ◦ X by 11.12,

(T(ψ) ◦ T(s.u))(X

x

) = (X

x

, u(x), du(X

x

))

we see that

(ψ ◦ K ◦ T(ψ

−1

))(X

x

, v, w) = (X

x

, ω(X

x

).v +w)

(Tψ ◦ Φ ◦ T(ψ

−1

))(X

x

, v, w) = (0

x

, v, w +ω(X

x

).v)

(Tψ ◦ C ◦ (Id ψ))(X

x

, v) = (X

x

, v, −ω(X

x

).v).

96 16. Characteristic classes for general ﬁber bundles

The Christoﬀel form Γ of Φ with respect to the bundle chart ψ is given

by 9.7

(0

x

, Γ(X

x

, v)) = −T(ψ)ΦT(ψ

−1

)(X

x

, v, 0)

= (0

x

, v, −ω(X

x

).v)

Γ(X

x

, v) = −ω(X

x

).v

Now the curvature R

∇

(X, Y ) = [∇

X

, ∇

Y

] −∇

[X,Y ]

. If we put R

∇

.s =

s.Ω, then we have Ω ∈ Ω

2

(U, gl(n)) and Ω ∈ Ω

2

(U, o(n)) if ∇ respects

g and if s is an orthonormal frame ﬁeld. Then the curvature form Ω is

given by Ω = dω +ω ∧ ω.

The curvature R =

1

2

[Φ, Φ] ∈ Ω

2

(E; V E) can be written in terms of

the Christoﬀel forms by 9.7

(ψ

−1

)

∗

R(X, Y ) = d

U

Γ(X, Y ) + [Γ(X), Γ(Y )]

X(R

n

)

= d(−ω)(X, Y ) −(−ω(X)) ∧ (−ω(Y ))

= −Ω(X, Y ),

since GL(n) acts from the left on R

n

, so the fundamental vector ﬁeld

mapping is a Lie algebra anti homomorphism.

Let us consider now the unit sphere bundle SE = ¦u ∈ E : g(u, u) =

1¦ in the vector bundle E. If we use an orthonormal frame then ψ :

SE[U →U S

n−1

is a ﬁber bundle chart. We have

T(U S

n−1

) = ¦(X

x

, v, w) ∈ TU R

n

R

n

: [v[ = 1, v⊥w = 0¦

Φ

local

(X

x

, v, w) = (0

x

, v, w +ω(X

x

).v)

¸v, w +ω(X

x

).v) = ¸v, ω(X

x

).v) = 0,

since ω(X

x

) ∈ o(n). So Φ induces a connection on SE with the same

Christoﬀel forms (acting on a submanifold now). If we assume that E is

ﬁber orientable and if we consider the ﬁber volume µ on SE induced by g,

then these Christoﬀel forms take values in the Lie algebra of divergence

free vector ﬁelds, so the induced connection Φ

SE

is µ-respecting. The

local expression of the curvature of Φ

SE

is again −Ω.

16.12. Lemma. 1. For a ﬁber orientable vector bundle (E, p, M, R

n

)

with ﬁber dimension n > 2 and a ﬁber Riemannian metric g we consider

the induced ﬁber volume µ on the unit sphere bundle SE. Then the

characteristic class k(SE, µ) = 0.

2. For a complex line bundle (E, p, M) with smooth hermitian metric

g. We consider again the induced ﬁber volume on the unit sphere bundle

(SE, p, S

1

). Then for the characteristic class we have

k(SE, µ) = −c

1

(E),

16. Characteristic classes for general ﬁber bundles 97

the negative of the ﬁrst Chern class of E.

Proof. 1. This follows from H

n−2

(S

n−1

) = 0.

2. This follows since H

0

(p) = MR, so from 16.8 the class k(SE, µ)

is induced by the negative of the curvature form Ω, which also deﬁnes

the ﬁrst Chern class.

98

17. Self duality

17.1. Let again (E, p, M, S) be a ﬁber bundle with compact standard

ﬁber S. By a ﬁberwise symplectic form ω on E we mean a smooth section

of the vector bundle Λ

2

V

∗

E →E such that ω

x

:= ω E

x

∈ Ω

2

(E

x

) is a

symplectic form on each ﬁber E

x

. So as in 16.1 we may plug two vertical

vector ﬁelds X and Y into ω to get a function ω(X, Y ) on E.

If dimS = 2n, let ω

n

x

= ω

x

∧ ∧ω

x

be the Liouville volume form of

ω

x

on E

x

. We will suppose that each ﬁber E

x

has total mass 1 for this

ﬁber volume: we may multiply ω by a suitable positive smooth function

on M.

I do not know whether lemma 16.2 remains true for ﬁberwise sym-

plectic structures.

17.2. Example. For a compact Lie group G let Ad

∗

: G → GL(g

∗

)

be the coadjoint action. Then the union of all orbits of principal type

(maximal dimension) is the total space of a ﬁber bundle which bears a

canonical ﬁberwise symplectic form.

17.3. Let now Φ be a connection on (E, p, M, S) and let ∇ be the as-

sociated covariant derivative on the (inﬁnite dimensional) vector bundle

Ω

2

ver

(E) :=

x∈M

Ω

2

(E

x

) over M which is given by

∇

X

ω :=

d

dt

[

0

(Fl

CX

t

)

∗

ω

as in 16.3.

Deﬁnition. Let ω be a ﬁberwise symplectic form on E. The connection

Φ on E is called ω-respecting if ∇

X

ω = 0 for all X ∈ X(M).

17.4. Lemma. If dimH

2

(S; R) = 1 then there are ω-respecting con-

nections on E and they form a convex set in the space of all connections.

Proof. Let us ﬁrst investigate the trivial situation. So e = M S and

the ﬁberwise symplectic form is given by a smooth mapping ω : M →

Z

2

(S) ⊂ Ω

2

(S) such that ω

x

is a symplectic form for each x. Since

_

S

ω

n

x

= 1 for each x and since α →α

n

is a covering mapping H

2

(S)¸0 =

R¸0 →H

2n

¸0 = R¸0, we see that the cohomology class [ω

x

] ∈ H

2

(S) is

locally constant on M, so ω : M →ω

x

0

+B

2

(S) if M is connected. Thus

d

M

ω : TM →B

2

(M) is a 1-form on M with values in the nuclear vector

space B

2

(S) of exact forms, since it is a closed subspace of Ω

2

(S) by

Hodge theory. Let G : Ω

k

(S) →Ω

k

(S) be the Green operator of Hodge

theory for some Riemannian metric on S and deﬁne Γ ∈ Ω

1

(M : X(S))

17. Self duality 99

by i(Γ(X

x

))ω

x

= −d

∗

G(d

M

ω)

x

(X

x

). Since ω

x

is non degenerate, Γ is

uniquely determined by this procedure and we have

L

Γ(X

x

)

ω

x

= d

S

i

Γ(X

x

)

ω

x

+ 0 = −dd

∗

G(d

M

ω)

x

(X

x

) = −(d

M

ω)

x

(X

x

),

since it is in B

2

(S). For the covariant derivative ∇ induced by the

connection with Christoﬀel form Γ we have

∇

X

ω =

d

dt

[

0

(Fl

(X,Γ(X))

t

)

∗

ω = (d

M

ω)(X) +L

Γ(X)

ω = 0,

so the Christoﬀel form Γ deﬁnes an ω-respecting connection.

For the general situation we know now that there are local solutions,

and we may use a partition of unity on M to glue the horizontal lifts to

get a global ω-respecting connection.

The second assertion is obvious.

17.5. Let now Φ be an ω-respecting connection on (E, p, M, S) where

ω is a ﬁberwise symplectic form on E. Let R = R(Φ) be the curvature

of Φ and let C be the horizontal lifting of Φ. Then i(R(CX

x

, CY

x

))ω

x

∈

Ω

1

(E

x

) and i(R(CX

x

, CY

x

))ω

n

x

∈ Ω

2n−1

(E

x

). As in lemma 16.5 we may

prove that both are closed forms. So [i(R(CX

x

, CY

x

))ω

x

)] ∈ H

1

(E

x

)

and [i(R(CX

x

, CY

x

))ω

n

x

] ∈ H

2n−1

(E

x

).

Let us also ﬁx an auxiliary Riemannian metric on the base manifold

M and let us consider the associated Hodge -∗-operator: Ω

k

(M) →

Ω

m−k

(M), where m = dimM. Let us suppose that the dimension

of M is 4 and let us consider the forms [i

R

ω] ∈ Ω

2

(M; H

1

(p)) and

∗[i

R

ω

n

] ∈ Ω

2

(M; H

2n−1

(p)) with values in the ﬂat vector bundles H

i

(p)

from 16.6. Both are d

∇

-closed forms (before applying ∗ to the sec-

ond one) since the proof of 16.7 applies without changes. Let D

x

:

H

2n−1

(E

x

) → H

1

(E

x

) be the Poincar´e duality operator for the com-

pact oriented manifold E

x

; these ﬁt together to a smooth vector bundle

homomorphism D : H

2n−1

(p) →H

1

(p).

Deﬁnition. An ω-respecting connection Φ is called self dual or anti self

dual if

D ∗ [i

R

ω

n

] = [i

R

ω] or

= −[i

R

ω] respectively.

This notion makes sense also for not ω -respecting connections.

100 17. Self duality

17.6. For the notion of self duality of 17.5 there is also a Yang-Mills

functional if M

4

is supposed to be compact. Let Φ be any connection.

For X

i

∈ T

x

M we consider the 4-form

ϕ

x

(Φ)(X

1

, . . . , X

4

) :=

1

4

Alt

_

E

x

(i

R

ω)

x

(X

1

, X

2

) ∧ ∗(i

R

ω

n

)

x

(X

3

, X

4

),

where Alt is the alternator of the indices of the X

i

. The Yang-Mills

functional is then

F(Φ) :=

_

M

ϕ(Φ).

17.7. A typical example of a ﬁber bundle with compact symplectic

ﬁbers is the union of coadjoint orbits of a ﬁxed type of a compact Lie

group. Closely related to this are open subsets of Poisson manifolds. ω-

respecting connections exist if the orbit type has second Betti number 1.

The Yang-Mills functional and the self duality notion exist even without

this requirement. It would be nice to investigate the characteristic class

of section 16 for the ﬁberwise symplectic structure instead of a ﬁber

volume form for examples of coadjoint actions.

17.8. Finally we sketch another notion of self duality. We consider

a ﬁberwise contact structure α on the ﬁber bundle (E, p, M, S) with

compact standard ﬁber S: So α

x

∈ Ω

1

(E

x

) depends smoothly on x and

is a contact structure on E

x

, i. e. α

x

∧(dα

x

)

n

is a positive volume form

on E

x

, where dimS = 2n + 1 now. A connection Φ on the bundle E is

now called α-respecting if in analogy with 16.2 we have

∇

Φ

X

α :=

∂

∂t

¸

¸

0

(Fl

CX

t

)

∗

α = 0

for each vector ﬁeld X on M. If R is the curvature of such a connec-

tion then iR(CX, CY )α is ﬁberwise locally constant by the analogue of

16.5.(3). If S is connected then (X, Z) →iR(CX, CY )α deﬁnes a closed

2-form on M with real coeﬃcients and it deﬁnes also a cohomology class

in H

2

(M; R). But it is not clear that α-respecting connections exist.

101

References

Br¨ocker, Theodor; J¨anich, Klaus, Einf¨ uhrung in die Diﬀerentialtopologie, Springer--

Verlag, Heidelberg, 1973.

Cervera, Vicente; Mascaro, Francisca; Michor, Peter W., The orbit structure of the

action of the diﬀeomorphism group on the space of immersions, Preprint (1989).

De La Harpe, Pierre; Omori, Hideki, About interaction between Banach Lie groups

and ﬁnite dimensional manifolds, J. Math. Kyoto Univ. 12 (1972).

Ebin, D.G.; Marsden, J.E., Groups of diﬀeomorphisms and the motion of an incom-

pressible ﬂuid, Ann. Math. 92 (1970), 102–163.

Eells, James, A setting for global analysis, Bull AMS 72 (1966), 571–807.

Ehresmann, Charles, Les connexions inﬁnit´esimales dans un espace ﬁbr´e diﬀeren-

tiable, Colloque Topologie, Bruxelles 1950, Liege, 1951, pp. 29–55.

Fr¨olicher, Alfred; Kriegl, Andreas, Linear spaces and diﬀerentiation theory, Pure and

Applied Mathematics, J. Wiley, Chichester, 1988.

Fr¨olicher, A.; Nijenhuis, A., Theory of vector valued diﬀerential forms. Part I., Inda-

gationes Math 18 (1956), 338–359.

Gil-Medrano, Olga; Peter W. Michor, The Riemannian manifold of all Riemannian

metrics, to appear, Quaterly Journal of Math. (Oxford).

Grabowski, Janusz, Free subgroups of diﬀeomorphism groups, Fundamenta Math.

131 (1988), 103–121.

Greub, Werner; Halperin, Steve; Vanstone, Ray, Connections, Curvature, and Co-

homology I, Academic Press, 1972.

Hamilton, Richard S., The inverse function theorem of Nash and Moser, Bull. Am.

Math. Soc. 7 (1982), 65–222.

Hirsch, Morris W., Diﬀerential topology, GTM 33, Springer-Verlag, New York, 1976.

Kainz, Gerd, Eine charakteristische Klasse f¨ ur F-Faserb¨ undel, handwritten manu-

script 1985.

Kobayashi, S.; Nomizu, K., Foundations of Diﬀerential Geometry. Vol. I., J. Wiley,

1963.

Kriegl, Andreas; Michor, Peter W., A convenient setting for real analytic mappings,

Acta Mathematica 165 (1990a), 105–159.

Kriegl, A.; Michor, P. W., Aspects of the theory of inﬁnite dimensional manifolds,

Diﬀerential Geometry and Applications 1(1) (1990b).

A. Kriegl, Michor, P. W., Foundations of Global Analysis, A book in the early stages

of preparation.

Lecomte, P. B. A.; Michor, P. W.; Schicketanz, H., The multigraded Nijenhuis-Ri-

chardson Algebra, its universal property and application, Preprint 1990, submit-

ted to J. Pure Applied Algebra. (1990).

Marathe, Kishore B.; Martucci, G., The geometry of gauge ﬁelds, J. Geometry

Physics 6 (1989), 1–106.

Michor, Peter W., Manifolds of diﬀerentiable mappings, Shiva Mathematics Series

3, Orpington, 1980.

Michor, P. W., Manifolds of smooth mappings IV: Theorem of De Rham, Cahiers

Top. Geo. Diﬀ. 24 (1983), 57–86.

Michor, Peter W., Remarks on the Fr¨olicher-Nijenhuis bracket, Proceedings of the

Conference on Diﬀerential Geometry and its Applications, Brno 1986, D. Reidel,

1987, pp. 197–220.

102 References

Michor, P. W., Gauge theory for diﬀeomorphism groups, Proceedings of the Confer-

ence on Diﬀerential Geometric Methods in Theoretical Physics, Como 1987, K.

Bleuler and M. Werner (eds.), Kluwer, Dordrecht, 1988, pp. 345–371.

Michor, P. W., Graded derivations of the algebra of diﬀerential forms associated

with a connection, Diﬀerential Geometry, Proceedings, Peniscola 1988,, Springer

Lecture Notes 1410, 1989a, pp. 249–261.

Michor, P. W., Knit products of graded Lie algebras and groups, Proceedings of the

Winter School on Geometry and Physics, Srni 1989, Suppl. Rendiconti Circolo

Matematico di Palermo, Ser. II 22 (1989b), 171–175.

Michor, Peter W., The moment mapping for a unitary representation, J. Global

Analysis and Geometry 8, No 3 (1990), 299–313.

Michor, Peter W.; Neuwirther, Martin, Pseudoriemannian metrics on spaces of bi-

linear structures, preprint 1990.

Modugno, Marco, Systems of vector valued forms on a ﬁbred manifold and appli-

cations to gauge theories,, Lecture Notes in Math., vol. 1251, Springer-Verlag,

1987.

Morrow, J., The denseness of complete Riemannian metrics, J. Diﬀ. Geo. 4 (1970),

225–226.

Nagata, J., Modern dimension theory, North Holland, 1965.

Newlander, A.; Nirenberg, L., Complex analytic coordinates in almost complex man-

ifolds, Ann. of Math. 65 (1957), 391–404.

Nijenhuis, A.; Richardson, R., Deformation of Lie algebra structures, J. Math. Mech.

17 (1967), 89–105.

Nomizu, K.; Ozeki, H., The existence of complete Riemannian metrics, Proc. AMS

12 (1961), 889–891.

Omori, Hideki, On Banach Lie groups acting on ﬁnite dimensional manifolds, To-

hoku Math. J. 30 (1978), 223–250.

Omori, Hideki, Inﬁnite dimensional Lie transformation groups, Lecture Notes in

Math. 427, Springer-Verlag, Berlin etc., 1974.

Palais, Richard S., A global formulation of the Lie theory of transformation groups,

Mem. AMS 22 (1957).

Ramadas, T. R., On the space of maps inducing isomorphic connections, Ann Inst.

Fourier 32 (263-276), 1982.

List of Symbols 103

List of Symbols

C

∞

(E), also C

∞

(E → M) the space of smooth sections of a ﬁbre

bundle

C

∞

(M, R) the space of smooth functions on M

d : Ω

p

(M) →Ω

p+1

(M) usually the exterior derivative

(E, p, M, S), also simply E usually a ﬁbre bundle with total space

E, base M, and standard ﬁbre S

Fl

X

t

the ﬂow of a vector ﬁeld X

I

k

, short for the k k-identity matrix Id

R

k.

L

X

Lie derivative

G usually a general Lie group with multiplication µ : GG →G,

left translation λ, and right translation ρ

: GS →S usually a left action

M usually a (base) manifold

N natural numbers

N

0

nonnegative integers

R real numbers

r : P G → P usually a right action, in particular the principal

right action of a principal bundle

TM the tangent bundle of a manifold M with projection π

M

:

TM →M

Z Integers

104

Index

α-respecting connection, 100

µ-respecting connection, 92

ω-respecting connection, 98

A

algebraic bracket, 28

algebraic derivation, 27

anti self dual, 99

approximation property (bornological) , 15

associated bundle, 48

associated vector bundles, 52

B

base space, 36

basis, 36

Bianchi identity, 37

bundle atlas, 36

bundle of gauges, 54

bundle, 36

C

cartesian closedness, 8, 11, 13

characteristic class, 94

Chern class, 97

Christoﬀel forms, 39

classifying connection, 87

classifying space, 53, 87, 89

cocurvature, 33

cocycle condition, 36

cocycle of transition functions, 36

cohomologous, 44

complete connection, 41

connection form (Lie algebra valued), 60

connection form, 95

connection, 32, 37, 92

connector, 69

continuous derivation over ev

a

, 14

convenient vector space, 7

covariant derivative on vector bundles, 70

covariant derivative, 63, 70

covariant exterior derivative, 63, 71

curvature form (Lie algebra valued), 60

curvature form, 96

curvature, 32, 33, 37, 70

c

∞

-topology, 6

C

ω

-manifold structure of C

ω

(M, N), 19

C

ω

-manifold structure on C

∞

(M, N), 19

D

De Rham cohomology of M with twisted

coeﬃcient domain H

n−1

(p), 92

derivations (graded) , 27

diﬀeomorphism group, 22

E

Ehresmann connection, 41

equivariant mappings and associated bundles,

50

exponential law, 20

F

f-dependent, 33

f-related, 33

ﬁber bundle atlas, 36

ﬁber bundle, 36

ﬁber chart, 36

ﬁber volume, 90

ﬁberwise symplectic form, 98

ﬂat connections, 38

frame bundle, 81

Fr¨olicher-Nijenhuis bracket, 29

fundamental vector ﬁeld, 5

fundamental vector ﬁeld, 5

G

G-atlas, 44

G-bundle structure, 44

G-bundle, 44

gauge group of a ﬁber bundle, 83

gauge transformations, 54

general Bianchi identity, 33

generalization of the theorem of Ambrose and

Singer, 77

global formula for the Fr¨olicher-Nijenhuis

bracket, 30

Grassmann manifold, 46

H

Hodge-∗-operator, 99

holomorphic curves, 10

holomorphic, 10

holonomy group, 65, 76

holonomy groups for principal connections, 65

holonomy Lie algebra, 76

homogeneous spaces, 45

homomorphism of G-bundles, 51

homomorphism over Φ of principal bundles, 47

horizontal bundle, 37

horizontal diﬀerential forms, 63

horizontal foliation, 38

horizontal lift, 37

horizontal projection, 37

horizontal space, 32

horizontal vectors, 37

I

induced connection, 66, 67

inducing principal connections on associated

vector bundles, 69

irreducible principle connection, 66

K

kinematic tangent vector, 14

L

Lie derivation, 28

Lie group, 5

linear connection, 69

linear frame bundle, 52

105

local description of connections, 38

local descriptions of principal connections, 62

local formulas for the Fr¨olicher-Nijenhuis

bracket, 31

M

manifold structure of C

∞

(M, N), 17

Maurer-Cartan formula, 39

N

n-transitive, 23

natural bilinear concomitants, 34

naturality of the Fr¨olicher-Nijenhuis bracket,

33

nearly complex structure, 35

Nijenhuis tensor, 35

Nijenhuis-Richardson bracket, 28

nonlinear frame bundle, 81

O

operational tangent vector, 14

orthonormal frame bundle, 53

orthonormal frame ﬁeld, 52

O

parallel transport, 39

principal (ﬁber) bundle, 44

principal bundle atlas, 44

principal bundle of embeddings, 24

principal connection, 60

principal ﬁber bundle homomorphism, 47

principal right action, 45

projection, 36

proper, 20

pullback operator, 34

pullback, 38

pullback of ﬁber bundles, 37

R

real analytic curves, 12

real analytic mappings, 12

real analytic, 12

recognizing induced connections, 67

reduction of the structure group, 47

regular, 23

restricted holonomy group, 65, 76

Riemannian metric , 52

right logarithmic derivative, 58

S

sections of associated bundles, 53

self dual, 99

smooth curves in C

∞

(M, N), 18

smooth mappings, 7

smooth partitions of unity, 14

smooth, 6, 7

smoothness of composition, 19

space of connections, 84

standard ﬁber, 36

Stiefel manifold, 46

T

tangent and vertical bundles of principal and

associated bundles, 58

tangent bundles of Grassmann manifolds, 56

tangent bundles of homogeneous spaces, 54

tangent group of a Lie group, 57

tangent vectors as derivations, 16

theorem of Hartogs, 10

topologically real analytic, 12

total space, 36

transformation law for the Christoﬀel forms,

39

transition functions, 36

U

universal vector bundle, 56

V

vector valued diﬀerential forms, 27

vertical bundle, 37

vertical bundle, 58

vertical projection, 37

vertical space, 32

Y

Yang-Mills functional, 100

1

Table of contents

Introduction . . . . . . . . . . . . . . . . . 1. Notations and conventions . . . . . . . . . 2. Calculus of smooth mappings . . . . . . . . 3. Calculus of holomorphic mappings . . . . . . 4. Calculus of real analytic mappings . . . . . . 5. Inﬁnite dimensional manifolds . . . . . . . . 6. Manifolds of mappings . . . . . . . . . . . 7. Diﬀeomorphism groups . . . . . . . . . . . 8. The Fr¨licher-Nijenhuis Bracket . . . . . . . o 9. Fiber Bundles and Connections . . . . . . . 10. Principal Fiber Bundles and G-Bundles . . . 11. Principal and Induced Connections . . . . . 12. Holonomy . . . . . . . . . . . . . . . . 13. The nonlinear frame bundle of a ﬁber bundle . 14. Gauge theory for ﬁber bundles . . . . . . . 15. A classifying space for the diﬀeomorphism group 16. A characteristic class . . . . . . . . . . . 17. Self duality . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . List of Symbols . . . . . . . . . . . . . . . Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 5 6 10 12 14 17 22 27 36 44 60 76 81 83 86 90 98 101 102 103

2

Introduction

Gauge theory usually investigates the space of principal connections on a principal ﬁber bundle (P, p, M, G) and its orbit space under the action of the gauge group (called the moduli space), which is the group of all principal bundle automorphisms of P which cover the identity on the base space M . It is the arena for the Yang-Mills-Higgs equations which allows (with structure group U (1) × SU (2)) a satisfactory uniﬁed description of electromagnetic and weak interactions, which was developed by Glashow, Salam, and Weinberg. This electro-weak theory predicted the existence of massive vector particles (the intermediate bosons W + , W − , and Z), whose experimental veriﬁcation renewed the interest of physicists in gauge theories. On the mathematical side the investigation of self dual and anti self SU (2)-connections on 4-manifolds and of their image in the orbit space led to stunning topological results in the topology of 4-manifolds by Donaldson. The moduli space of anti self dual connections can be completed and reworked into a 5 dimensional manifold which is a bordism between the base manifold of the bundle and a simple manifold which depends only on the Poincar´ duality form in the second dimensional e homology space. Combined with results of Freedman this led to the discovery of exotic diﬀerential structures on R4 and on (supposedly all but S 4 ) compact algebraic surfaces. In his codiﬁcation of a principal connection [Ehresmann, 1951] began with a more general notion of connection on a general ﬁber bundle (E, p, M, S) with standard ﬁber S. This was called an Ehresmann connection by some authors, we will call it just a connection. It consists of the speciﬁcation of a complement to the vertical bundle in a diﬀerentiable way, called the horizontal distribution. One can conveniently describe such a connection as a one form on the total space E with values in the vertical bundle, whose kernel is the horizontal distribution. When combined with another venerable notion, the Fr¨licher-Nijenhuis bracket for vector valued diﬀerential forms o (see [Fr¨licher-Nijenhuis, 1956]) one obtains a very convenient way to o describe curvature and Bianchi identity for such connections. Parallel transport along curves in the base space is deﬁned only locally, but one may show that each bundle admits complete connections, whose parallel transport is globally deﬁned (theorem 9.10). For such connections one can deﬁne holonomy groups and holonomy Lie algebras and one may prove a far-reaching generalization of the Ambrose Singer theorem: A complete connection tells us whether it is induced from a principal connection on a principal ﬁber bundle (theorem 12.4).

as least when S is compact. real analyticity and holomorphy. Finally we discuss some self duality and anti self duality conditions which depend on some ﬁberwise structure on the bundle like a ﬁberwise symplectic structure. There is hope that it can be stratiﬁed into smooth manifolds corresponding to conjugacy classes of holonomy groups in Diﬀ(S). 1990b] and the forthcoming book [Kriegl-Michor]. 1980] . We deﬁne the non linear frame bundle for E which is a principle ﬁber bundle over M with structure group Diﬀ(S) and we show that the theory of connections on E corresponds exactly to the theory of principal connections on this non linear frame bundle (see section 13). but this will be treated in another paper. p. The diﬀeomorphism group Diﬀ(S) for a compact manifold S admits a smooth classifying space and a classifying connection: the space of all embeddings of S into a Hilbert space 2 . whose base manifold is the non linear Grassmanian of all submanifolds of 2 of type (diﬀeomorphic to) S. We treat smoothness. Characteristic classes owe their existence to invariant polynomials on the Lie algebra of the structure group like characteristic coeﬃcients of matrices. Unfortunately it does not admit any invariants. and one may consider the moduli space Conn(E)/ Gau(E). But there are equivariants like X → LX ω for some closed form ω on S which can be used to give a sort of Chern-Weil construction of characteristic classes in the cohomology of the base M with local coeﬃcients (see sections 16 and 17). The Lie algebra of Diﬀ(S) for compact S is the Lie algebra X(S) of vector ﬁelds on S. M.Introduction 3 A bundle (E. Then we give a more detailed exposition of the theory of manifolds of mappings and the diﬀeomorphism group. S) without structure group can also be viewed as having the whole diﬀeomorphism group Diﬀ(S) of the standard ﬁber S as structure group. is the total space of a principal bundle with structure group Diﬀ(S). say. which admits a classifying connection: Every S-bundle over M can be realized as the pull back of the classifying bundle by a classifying smooth mapping from M into the non linear Grassmanian. which can be arranged in such a way that it pulls back the classifying connection to a given one on E — this is the Narasimhan-Ramadas procedure for Diﬀ(S). This part is based on [Kriegl-Michor. The action of Diﬀ(S) on S leads to the (universal or classifying) associated bundle. The gauge group of the non linear frame bundle turns out to be just the group of all ﬁber respecting diﬀeomorphisms of E which cover the identity on M . This booklet starts with a short description of analysis in inﬁnite dimensions along the lines of Fr¨licher and Kriegl which makes inﬁnite o dimensional diﬀerential geometry much simpler than it used to be. This part is adapted from [Michor.

his interest in this subject. Wien. Then we present a careful o introduction to the Fr¨licher-Nijenhuis bracket. 1988. I want to thank G. 1990 P. has been published in [Michor. Michor . from section 12 onwards. The material in the rest of the book. 1988]. The material in this booklet is a much extended version of a series of lectures held at the Institute of Physics of the University of Napoli. August 6. The version here is much more detailed and contains more results.4 Introduction to the use of the Fr¨licher-Kriegl calculus. March 28 – April 1. G-structures and principal connection which emphasizes the construction and recognition of induced connections. and for the suggestion to publish this booklet. Marmo for his hospitality. to ﬁber bundles and o connections.

X = T(e. λa : G → G. A Lie group G is a smooth manifold and a group such that the multiplication µ : G × G → G is smooth. We shall use the following notation: µ : G × G → G. X). (2) Tx (ra ). We will use the following notation: ra : M → M and rx : G → M . we use the following notation: AdG : G → AutLie (g) and so on.(0x .2.ζX (x) = ζAd(a−1 )X (x.a). Lemma. so ˇ : G → Diﬀ(M ) is a group homomorphism. (2) Tx ( a ). right translation. x) = a.x. M For any X ∈ g we deﬁne the fundamental vector ﬁeld ζX = ζX ∈ x X(M ) by ζX (x) = Te ( ). y) = x. ν(x) = x−1 . the unit element. 1.e) r. left translation. ρa : G → G.ζX (x) = ζAd(a)X (a. Then also the inversion ν : G → G turns out to be smooth. 0x ).a. ζY ] = −ζ[X. inversion. Deﬁnition. ν : G → G. Lemma. a) = x. µ(x.a. so r : G → Dif f (M ) ˇ is a group anti homomorphism.5 1. ζY ] = ζ[X. Then we have partial mappings a : M → M and x : G → M .x) . given by rx (a) = ra (x) = r(x.x. Notations and conventions 1.Y ] . (4) [ζX . (3) RX × 0M ∈ X(G × M ) is -related to ζX ∈ X(M ). λa (x) = a. In this situation the following assertions hold: (1) ζ : g → X(M ) is a linear mapping. given by a (x) = x (a) = (a. M For any X ∈ g we deﬁne the fundamental vector ﬁeld ζX = ζX ∈ X(M ) by ζX (x) = Te (rx ). In this situation the following assertions hold: (1) ζ : g → X(M ) is a linear mapping. Let r : M × G → M be a right action. (4) [ζX . (3) 0M × LX ∈ X(M × G) is r-related to ζX ∈ X(M ).(X. .x). e ∈ G. If g = Te G is the Lie algebra of G.Y ] . multiplication. ρa (x) = x. 1. Let : G × M → M be a left action.1.X = T(x.y.3.

t−s (3) {EB → E : B bounded absolutely convex in E}. The ﬁnest among all locally convex topologies on E which are coarser than c∞ E is the bornologiﬁcation of the given locally convex topology. 1988]. convergence spaces. Let E be a locally convex vector space. (2) Lipschitz curves (so that { c(t)−c(s) : t = s} is bounded in E). E). each of them rather complicated and none really convincing. They joined forces in the further development of the theory and the (up to now) ﬁnal outcome is the book [Fr¨lichero Kriegl. o 2. (4) Mackey-convergent sequences xn → x (there exists a sequence 0 < λn ∞ with λn (xn − x) bounded). The traditional diﬀerential calculus works well for ﬁnite dimensional vector spaces and for Banach spaces. Alfred Fr¨licher and Andreas Kriegl presented indeo pendently the solution to the question for the right diﬀerential calculus in inﬁnite dimensions.2. where EB is the linear span of B equipped with the Minkowski functional pB (x) := inf{λ > 0 : x ∈ λB}. then c∞ E = E.this is a concept without problems. The ﬁnal topologies with respect to the following sets of mappings into E coincide: (1) C ∞ (R. In general (on the space D of test functions for example) it is ﬁner than the given locally convex topology. since scalar multiplication is no longer jointly continuous. Then in 1982.1. If E is a Fr´chet space. E) be the space of smooth functions. for any compatible topology. For more general locally convex spaces a whole ﬂock of diﬀerent theories were developed. E) does not depend on the locally convex topology of E. which is the general reference for this section. It can be shown that C ∞ (R. The c∞ -topology. it is not a vector space topology. only on its associated bornology (system of bounded sets). Calculus of smooth mappings 2. See also the forthcoming book [Kriegl-Michor]. Let C ∞ (R. This was the original motivation for the development of a whole new ﬁeld within general topology.6 2. e . This topology is called the c∞ -topology on E and we write c∞ E for the resulting topological space. In this section I will sketch the basic deﬁnitions and the most important results of the Fr¨licher-Kriegl calculus. A curve c : R → E is called smooth or C ∞ if all derivatives exist and are continuous . The main diﬃculty is that the composition of linear mappings stops to be jointly continuous at the level of Banach spaces.

For locally convex spaces E and F we have: (1) If F is convenient. E). E) with the bornologiﬁcation of the topology of uniform convergence on compact sets. then EB is a Banach space. f : R → E is scalarwise C ∞ then f is C ∞ . so f∗ : C ∞ (R. Convenient vector spaces. E) there is c2 ∈ C ∞ (R. Constant mappings are smooth. 1967]. F ) makes sense. This is also called c∞ -complete. Structure on C ∞ (E. 2. (2) If B is bounded closed absolutely convex. locally on Rn . Lemma. (3) Any Lipschitz curve in E is locally Riemann integrable. f scalarwise C ∞ means that λ ◦ f is C ∞ for all continuous linear functionals on E. We equip the space C ∞ (R. F ) with the bornologiﬁcation of the initial topology with respect to all mappings c∗ : C ∞ (E. if f ◦ c ∈ C ∞ (R. Let E be a locally convex vector space. for k > 1. F ) denote the space of all smooth mapping from E to F . Therefore we denote by L(E. E).3. (4) For any c1 ∈ C ∞ (R. in all derivatives separately. E) → C ∞ (R. E) with c1 = c2 (existence of antiderivative).5. Then we equip the space C ∞ (E. Lemma. F ) the space of all bounded linear mappings from E to F . E is said to be a convenient vector space if one of the following equivalent (completeness) conditions is satisﬁed: (1) Any Mackey-Cauchy-sequence (so that (xn − xm ) is Mackey convergent to 0) converges. A mapping f : E → F is called smooth or C ∞ . Then the following properties are equivalent: (1) (2) (3) (4) E If If If is c∞ -complete.2. 2. c∗ (f ) := f ◦ c. F ). Let E be a locally convex space. Here a mapping f : Rk → E is called Lipk if all partial derivatives up to order k exist and are Lipschitz. Multilinear mappings are smooth if and only if they are bounded.6. then also C ∞ (E.7.4. For E and F ﬁnite dimensional this gives the usual notion of smooth mappings: this has been ﬁrst proved in [Boman. F ) is convenient. F ). Let E and F be locally convex vector spaces. F ) → C ∞ (R. for all c ∈ C ∞ (R. Calculus of smooth mappings 7 2. 2. f : R → E is scalarwise C ∞ then f is diﬀerentiable at 0. then f is Lipk . for any . Smooth mappings. f : Rk → E is scalarwise Lipk . Let C ∞ (E. This lemma says that a convenient vector space one can recognize smooth curves by investigating compositions with continuous linear functionals. 2. F ) for all c ∈ C ∞ (R.

F ) × C ∞ (E . F ) is a closed linear subspace of C ∞ (E. Calculus of smooth mappings E. df (x)v := lim t→0 f (x + tv) − f (x) . then a curve c : R → L(E. G) )∨ : C ∞ (E × F. x) = f (x) ins(x)(y) = (x. G) → C ∞ (E. F ). Also the chain rule holds: d(f ◦ g)(x)v = df (g(x))dg(x)v. F )). 2. F ) is smooth if and only if t → c(t)(x) is a smooth curve in F for all x ∈ E. G)) → C ∞ (E × F.9. G) × C ∞ (E. C ∞ (E . E × F ).8 2. Let all spaces be convenient vector spaces. E) → C ∞ (C ∞ (E. ) : C ∞ (F. Then the diﬀerential operator d : C ∞ (E. ev : C ∞ (E. The space L(E. (2) If E is convenient. 2. so it also convenient.8. G)) . y) ) : C (E. ins : E → C (F. So we have a natural bijection C ∞ (E × F. Theorem. Let E and F be convenient vector spaces. Fi ) 2. F ) → C ∞ (E. F )) C ∞ (Ei . Corollary.10. G) C ∞( (f. g) → (h → f ◦ h ◦ g) : Ei . L(E. C ∞ (F. F ). G) ∼ C ∞ (E. Theorem. . G)). C (F. ( ( ∧ ∞ ∞ ∞ ev(f. = which is even a diﬀeomorphism. F ) → C ∞ (E. Cartesian closedness. F ) × E → F. Then the following canonical mappings are smooth. C ∞ (F. t exists and is linear and bounded (smooth). The category of convenient vector spaces and smooth mappings is cartesian closed. Of coarse this statement is also true for c∞ -open subsets of convenient vector spaces. Fi ) → C ∞ ( comp : C ∞ (F.

Note that the conclusion of theorem 2. but it is always smooth. Many authors try to build their platonic idea of an a priori estimate into their diﬀerential calculus. Calculus of smooth mappings 9 2.8 to be true and assumes some other obvious properties. that the calculus itself should be as easy to use as possible.8 is the starting point of the classical calculus of variations. and [Pressley-Segal. If one wants theorem 2. For example the evaluation E × E → R is jointly continuous if and only if E is normable. 1974]. smooth mappings which are not continuous. A prevalent opinion in contemporary mathematics is. This is the diﬀerential calculus used by [Michor. There are.2. 1986]. and that all further assumptions (which most often come from ellipticity of some nonlinear partial diﬀerential equation of geometric origin) should be treated separately. [Milnor. that for inﬁnite dimensional calculus each serious application needs its own foundation. I think that this makes the calculus inapplicable and hides the origin of the a priori estimates. 1980]. however. Clearly smooth mappings are continuous for the c∞ -topology. I am sure that in this sense the Fr¨licher-Kriegl calculus as presented here and its holomorphic and real o analytic oﬀsprings in sections 2 and 3 below are universally usable for most applications. By a serious application one obviously means some application of a hard inverse function theorem. These theorems can be proved. 1984]. I believe. where a smooth curve in a space of functions was assumed to be just a smooth function in one variable more. Remarks. This is unavoidable and not so horrible as it might appear at ﬁrst sight.11. if by assuming enough a priori estimates one creates enough Banach space situation for some modiﬁed iteration procedure to converge. . in a setting depending on the speciﬁc problem. then the calculus of smooth functions is already uniquely determined. For Fr´chet spaces smoothness in the sense described here coincides e ∞ with the notion Cc of [Keller.

Along the lines of thought of the Fr¨licher-Kriegl calculus of smooth o mappings. (2) f is locally a convergent series of homogeneous continuous polynomials. Here we will start with them for the sake of shortness. 3. Let D ⊂ C be the open unit disk and let us denote by C ω (D. Let C ω (E. a mapping f : E → F is called holomorphic if f ◦ c is a holomorphic curve in F for each holomorphic curve c in E. y) and f (x. Obviously f is holomorphic if and only if λ◦f : E → C is holomorphic for each complex linear continuous functional λ on F . If f : E ⊃ U → F is holomorphic then df : U × E → F exists.10 3. If E and F are Banach spaces and U is open in E. Theorem of Hartogs. 1930 and 1933]. I will now sketch the bae sics and the main results. Calculus of holomorphic mappings 3. in [Kriegl-Nel. e . then for a mapping f : U → F the following conditions are equivalent: (1) f is holomorphic. 2 and F be complex convenient vector spaces and let Uk ⊂ Ek be c∞ -open. Its elements will be called the holomorphic curves.1. F ) denote the space of all holomorphic mappings from E to F . (3) f is C-diﬀerentiable in the sense of Fr´chet.5 Lemma. 3. e. This implies also that in ﬁnite dimensions we have recovered the usual deﬁnition. A mapping f : U1 × U2 → F is holomorphic if and only if it is separably holomorphic (i. 3. ) are holomorphic for all x ∈ U1 and y ∈ U2 ).3. E) the space of all mappings c : D → E such that λ ◦ c : D → C is holomorphic for each continuous complex-linear functional λ on E. 1985] the cartesian closed setting for holomorphic mappings was developed. If E and F are convenient complex vector spaces (or c∞ -open sets therein).2. Let Ek for k = 1. 3. f ( . The right deﬁnition of this calculus was already given by [Fantappi´. is holomorphic and C-linear in the second variable. Let E be a complex locally convex vector space whose underlying real space is convenient – this will be called convenient in the sequel. It can be shown that again convenient vector spaces are the right ones to consider.4 Lemma. A multilinear mapping is holomorphic if and only if it is bounded.

An immediate consequence of this result is that C ω (E. The category of convenient complex vector spaces and holomorphic mappings between them is cartesian closed. . F ) is a closed linear subspace of C ∞ (ER . Calculus of holomorphic mappings 11 3. e. G) ∼ C ω (E. = An immediate consequence of this is again that all canonical structural mappings as in 2.7 Theorem. 3.9 are holomorphic. Let E and F be convenient vector spaces.6 Lemma.7. by 2. Cartesian closedness.8. A mapping f : E → F is holomorphic if and only if it is smooth and its derivative is everywhere C-linear.3. The chain rule follows from 2. G)). C ω (F. The following theorem is an easy consequence of 2. FR ) and so it is a convenient vector space if F is one. C ω (E × F.10. i.

U ) → C ω (U. A curve c : R → E is called real analytic if λ ◦ c : R → R is real analytic for each λ ∈ E . If the dual E of E admits a Baire topology which is compatible with the duality. for all λ ∈ F . → where C ∞ (R. Real analytic mappings. R) of all real analytic functions with the initial topology with respect to the families of mappings C ω (U. and where C ω (R. In this section I sketch the cartesian closed setting to real analytic mappings in inﬁnite dimension following the lines of the Fr¨lichero Kriegl calculus. R) is equipped with the ﬁnal topology with respect to the embeddings (restriction mappings) of all spaces of holomorphic mappings from a neighborhood V of R in C mapping R to R. Let C ω (U. for all c ∈ C ∞ (R. Surprisingly enough one has to deviate from the most obvious notion of real analytic curves in order to get a meaningful theory. → It turns out that this is again a convenient space.12 4. F ) denote the space of all real analytic mappings. U ). but again convenient vector spaces turn out to be the right kind of spaces. then each real analytic curve in E is in fact topological real analytic for the bornological topology on E. A mapping f : U → F is called real analytic if and only if it is smooth (maps smooth curves to smooth curves) and maps real analytic curves to real analytic curves. In contrast a curve is called topologically real analytic if it is locally given by power series which converge in the topology of E. 4. R). Calculus of real analytic mappings 4. F ) with the initial topology with respect to the family of mappings C ω (U. We equip the space C ω (U. R) carries the topology of compact convergence in each derivative separately as in section 2. R) − C ∞ (R. F ) − ∗ C ω (U. Furthermore we equip the space C ω (U. R).3. as it is presented in [Kriegl-Michor. 1990a]. They can be extended to germs of holomorphic curves along R in the complexiﬁcation EC of E. Real analytic curves.1. R) − C ω (R. Let E and F be convenient vector spaces. Let U be a c∞ -open set in E. It turns out that the set of these curves depends only on the bornology of E.2. Let E be a real convenient vector space with dual E . R). λ c∗ c∗ . 4. for all c ∈ C ω (R. and each of these spaces carries the topology of compact convergence.

. Powerful real analytic uniform boundedness principles are available. Theorem Cartesian closedness. The category of convenient spaces and real analytic mappings is cartesian closed. F .9 are real analytic. and G are convenient vector spaces.4. where E. F ) ∼ C ω (U × V.5. Calculus of real analytic mappings 13 4. 4. L(E. F ) → C ω (U. F )) exists. Multilinear mappings are real analytic if and only if they are bounded. Theorem. F ) = is valid for all c∞ -open sets U in E and V in F . C ω (V ).3 a mapping f : U → F is real analytic if and only if it is smooth and is real analytic along each aﬃne line in E. Furthermore the diﬀerential operator d : C ω (U.4. This implies again that all structure mappings as in 2. So the equation C ω (U. is unique and real analytic. In the setting of 4.

if they have this property when composed which any chart mappings. Smooth partitions of unity. The material is taken from [Kriegl-Michor. 5. Mappings between manifolds are smooth (holomorphic.1. 5. The value Xa (f ) depends only on the germ of f at a. Lemma. 1990b]. For each open neighborhood U of a in E (a. Convenient vector spaces which are nuclear admit smooth bump functions. The tangent spaces of a convenient vector space E. Inﬁnite dimensional manifolds 5. real analytic) diﬀeomorphisms. E) of smooth E-valued functions on a manifold M is again a convenient vector space. Let a ∈ E. Lemma. which explains the choice of the name kinematic. e 5. then this is true for every open cover U of M .5. Likewise for the real analytic and holomorphic case. If M is a smooth manifold modeled on convenient vector spaces admitting smooth bump functions and U is a locally ﬁnite open cover of M . If M is in addition paracompact. In this section I will concentrate on two topics: Smooth partitions of unity. A manifold M is metrizable if and only if it is paracompact and modeled on Fr´chet spaces. then there exists a smooth partition of unity {ϕU : U ∈ U} with carr/supp(ϕU ) ⊂ U for all U ∈ U.14 5. and satisﬁes Xa (f · g) = Xa (f ) · g(a) + f (a) · Xa (g). which is continuous for the convenient vector space topology on C ∞ (U. X) induces a linear mapping Xa : C ∞ (U. In the usual way we deﬁne manifolds by gluing c∞ -open sets in convenient vector spaces via smooth (holomorphic.6. R). For a convenient vector space E the set C ∞ (M. Then we equip them with the identiﬁcation topology with respect to the c∞ -topologies on all modeling spaces. A kinematic tangent vector with foot point a is simply a pair (a. Let Ta E = E be the space of all kinematic tangent vectors with foot point a.2.3.4. . It consists of all derivatives c (0) at 0 of smooth curves c : R → E with c(0) = a. so it is a continuous derivation over eva . We require some properties from this topology. and several kinds of tangent vectors. like Hausdorﬀ and regular (which here is not a consequence of Hausdorﬀ). X) with X ∈ E. real analytic). 5. 5. Theorem. R) → R by Xa (f ) := df (a)(Xa ).

9. Any ∂ ∈ Da E induces a bounded derivation C ∞ (U. Let Da E be the vector space of all these derivations. nuclear Fr´chet spaces. Deﬁnition. 5. R) be a bounded linear functional which sym vanishes on the subspace k−1 Li (E. deﬁnes an operational The linear mapping → ∂ k |a is an embedding onto a topological direct (k) summand Da E of Da E. e . Let : L2 (E. R). The sum k>0 Da E in Da E is a direct one. 5.8. since (d2 (f g)(a)) = = (d2 f (a)g(a) + df (a) ⊗ dg(a) + dg(a) ⊗ df (a) + f (a)d2 g(a)) = (d2 f (a))g(a) + 0 + f (a) (d2 g(a)). Example. We equip Da E with the induced convenient vector space structure.5. So we have a canonical injection E → Da E. R) ∨ Lk−i (E. all opera(k) tional tangent space summands D0 E are not zero. Lemma. R) → R over eva . 5. The following spaces have the bornological approximation property: R(N) . If E is an inﬁnite dimensional Hilbert space. Its left inverse is given by ∂ → (Φ → ∂((Φ ◦ (k) diag)(a+ ))). A convenient vector space is said to have the (bornological) approximation property if E ⊗ E is dense in L(E. R) → R over eva for each open neighborhood U of a in E.7. Then for each a ∈ E we have an operational tangent vector ∂ 2 |a ∈ Da E given by ∂ 2 |a (f ) := (d2 f (a)). Then for each a ∈ E we have an operational tangent vector Ya ∈ Da E. Inﬁnite dimensional manifolds 15 An operational tangent vector of E with foot point a is a bounded ∞ derivation ∂ : Ca (E. R). nuclear (LF)spaces. So the vector space Da E is a closed linear subspace of the convenient vector space U L(C ∞ (U. Example. Then sym tangent vector ∂ k |a ∈ Da E for each a ∈ E by ∂ k |a (f ) := (dk f (a)). Let Y ∈ E be an element in the bidual of E. R). R) sym sym i=1 of decomposable elements of Lk (E. given by Ya (f ) := Y (df (a)). E) in the bornological locally convex topology. Note that the spaces Da E are isomorphic for all a ∈ E. Let ∈ Lk (E. Lemma. R) → R be a bounded linear functional which vanishes on the subset E ⊗ E .

So if E is in addition reﬂexive. . The kinematic tangent bundle T M of a manifold M is constructed by gluing all the kinematic tangent bundles of charts with the help of the kinematic tangent mappings (derivatives) of the chart changes. D(k) N ) are well deﬁned for all k including ∞ and have the usual properties. Let E be a convenient vector space which has the approximation property. Then πM : DM → M is again a vector bundle which contains the kinematic tangent bundle T M embeds as a splitting subbundle. Tangent vectors as derivations. T M → M is a vector bundle and T : C ∞ (M. 5. each operational = tangent vector is a kinematic one. Also for each k ∈ N the same gluing construction as above gives us tangent bundles D(k) M which are splitting subbundles of DM . N ) → C ∞ (D(k) M.16 5.11. Then we have (1) Da E = Da E ∼ E . T N ) is well deﬁned and has the usual properties.10 Theorem.12. N ) → C ∞ (T M. Note that for manifolds modeled on reﬂexive spaces having the bornological approximation property the operational and the kinematic tangent bundles coincide. Inﬁnite dimensional manifolds 5. The mappings D(k) : C ∞ (M. 5. The operational tangent bundle DM of a manifold M is constructed by gluing all operational tangent spaces of charts.

N ) : (f (x). N ) of all smooth mappings from M to N is a smooth manifold. N ) we write f ∼ g if f and g agree oﬀ some compact subset in M . uf (g)(x) = (x. −− f ∞ Then C ∞ (f ∗ T N ) is canonically isomorphic to the space Cf (M. g ∈ C ∞ (M. deﬁned on a suitable open neighborhood of the zero section. f (x) . Theorem. Manifold structure of C ∞ (M. T N ) : πN ◦ h = f } via s → (πN f ) ◦ s and (IdM . Then the space C ∞ (M. exp) : U → N × N is a smooth diﬀeomorphism onto an open neighborhood V of the diagonal. g ∼ f }. Now let Uf := {g ∈ C ∞ (M. Manifolds of mappings 6. For f ∈ C ∞ (M. o Proof. Let M and N be smooth ﬁnite dimensional manifolds. For f . N ) we consider the pullback vector bundle M ×N T N f ∗T N − − → T N −N− π f ∗π N π∗ f N M − − → N. Let exp : T N ⊇ U → N be the smooth exponential mapping of this Riemannian metric. ((πN . h) ← h. I include a o proof of this result here because the result is important and the proof is much simpler now with the help of the Fr¨licher Kriegl-calculus. A careful description of this theorem (but without the help of the Fr¨licher-Kriegl calculus) can be found in [Michor. g(x)) ∈ V for all x ∈ M. exp)−1 ◦ (f. K where K runs through all compact sets in M and each of the spaces ∞ CK (f ∗ T N ) is equipped with the topology of uniform convergence (on K) in all derivatives separately. modeled on spaces C ∞ (f ∗ T N ) of smooth sections of pullback bundles along f : M → N over M .17 6. ∞ uf : Uf → Cc (f ∗ T N ). Choose a smooth Riemannian metric on N . T N ) := ∞ ∗ {h ∈ C (M. We may assume that U is chosen in such a way that (πN .1. N ). 1980]. g))(x)). ∞ We consider the space Cc (f ∗ T N ) of all smooth sections with compact support and equip it with the inductive limit topology ∞ ∞ Cc (f ∗ T N ) = inj lim CK (f ∗ T N ). exp−1 (g(x))) = (x.

Since this is a local condition on R and since smooth curves in C ∞ (M. which have compact support in M locally in R. It is equipped with the inductive limit topology ∞ ∞ Cc (E) = inj lim CK (E). −1 The chart change uf ◦ u−1 = (τf ◦ τg )∗ is deﬁned on an open subset g and obviously maps smooth curves to smooth curves. K . whose inverse is given by uf (s) = exp ◦(πN f ) ◦ s. Let M and N be smooth ﬁnite dimensional manifolds. exp ◦(πN g) ◦ s)) g −1 = (τf ◦ τg )∗ (s). N ) locally take values in charts as in the proof of theorem 6. Yg(x) ) := (x. 6. (πN . where τg (x. where ∞ we view U → N as ﬁber bundle.N ) for C ∞ (M.1 it suﬃces to ∞ describe the smooth curves in the space Cc (E) of sections with compact support of a vector bundle (E. ∞ Smooth curves in Cc (f ∗ T N ) are just smooth sections of the bundle ∗ ∗ pr2 f T N → R × M . Finally we put the identiﬁcation topology from this atlas on the space C ∞ (M. The set uf (Uf ) is open in Cc (f ∗ T N ) for the topology described above. therefore it is also smooth. N ). N ). Its ∞ chart change mappings are given for s ∈ ug (Uf ∩ Ug ) ⊆ Cc (g ∗ T N ) by ∗ (uf ◦ u−1 )(s) = (IdM . N ) which satisfy the following property: ˆ (1) For each compact interval [a.18 6. ˆ Proof. uf )f ∈C ∞ (M. Since R is locally compact property (1) is equivalent to (2) For each t ∈ R there is an open neighborhood U of t in R and a compact K ⊂ M such that the restriction c|(U × (M \ K)) is ˆ constant. M. which is obviously ﬁner than the compact open topology and thus Hausdorﬀ. p. Now we consider the atlas (Uf . V ). b] × (M \ K)) is constant. N ). expg(x) (Yg(x) ))) is a smooth diﬀeomorphism τg : g ∗ T N ⊇ g ∗ U → (g × IdN )−1 (V ) ⊆ M × N which is ﬁber respecting over M .2. Lemma. −1 The equation uf ◦ u−1 = (τf ◦ τg )∗ shows that the smooth structure g does not depend on the choice of the smooth Riemannian metric on N. exp)−1 ◦ (f. Then the smooth curves c in C ∞ (M. b] ⊂ R there is a compact subset K ⊂ M such that the restriction c|([a. Smooth curves in C ∞ (M. N ) correspond exactly to the smooth mappings c ∈ C ∞ (R×M. Manifolds of mappings ∞ Then uf is a bijective mapping from Uf onto the set {s ∈ Cc (f ∗ T N ) : −1 ∗ ∗ s(M ) ⊆ f U }.

6. where the factor spaces on the right hand side carry the topology used in 2. V ) is a linear embedding onto a closed linear subspace of the product. The chart changings uf ◦ u−1 = (τf ◦ τg )∗ are real analytic: this g follows from a careful description of the set of real analytic curves into C ∞ (f ∗ T N ). N ) is a smooth manifold. N ). It can also be found in [Kriegl-Michor.3. since C ω (M. in all derivatives separately). N ). N ) do not cover C ∞ (M. using a real analytic Riemannian metric. uf ) from 6. 1990a] for more details. 8]. C ω -manifold structure on C ∞ (M. Then ∞ CK (E) s → (pr2 ◦ ψα ◦ s|Uα )α ∈ α C ∞ (Uα . and where CK (E) denotes the space of sections of E with support in K. 6. Then the smooth manifold C ∞ (M. modeled on spaces C ω (f ∗ T N ) of real analytic sections of pullback bundles along f : M → N over M . N ) of all real analytic mappings from M to N is a real analytic manifold.1 for f ∈ C ω (M. 1990a]. 1958. N ).8 this is the case if and only if their canonically associates are smooth R × Uα → V . Since this injective limit is strict (each bounded subset is contained and bounded in some step) smooth ∞ ∞ curves in Cc (E) locally factor to smooth maps in CK (E) for some K. Manifolds of mappings 19 ∞ where K runs through all compact subsets of M . This is not the case for the D-topology from [Michor. 6. ψα : E → Uα × V ) be a vector bundle atlas for E. N ) is dense in C ∞ (M.6. By cartesian closedness 2. these ﬁt ∗ together to a smooth section of the pullback bundle pr2 E → R×M with support in R × K. Let M and N be real analytic manifolds. . See again [Kriegl-Michor. Then the space C ω (M. Prop. C ω (M. C ω -manifold structure of C ω (M. 1980]. equipped with the compact C ∞ -topology (the topology of uniform convergence on compact subsets. N ) is dense in C ∞ (M. 6. ∞ So a curve in CK (E) is smooth if and only if each its ”coordinates” in ∞ C (Uα . N ) with the structure from 6. V ) is smooth. The charts Uf for f ∈ C ∞ (M. Let (Uα .5 Remark. Proof. The proof is a variant of the proof of 6. N ) for the Whitney-C ∞ -topology by [Grauert. N ) by [Grauert. For a ﬁxed real analytic exponential mapping on N the charts (Uf . let M be compact.1 is even a real analytic manifold. 1958.1. N ) form a smooth atlas for C ∞ (M. Proposition −1 8].4. Let M and N be real analytic manifolds. with M compact. Theorem. in which C ∞ (M. N ). Theorem. If M is not compact.

c2 )) = c1 (t. Then we have (comp ◦(c1 . N ) induces an atlas for T C ∞ (M. N ). Smoothness of composition.6. . then the composition mapping ∞ comp : C ∞ (M. N ) × Cprop (P. N ) and Cc (M. 10. then the evaluation mapping ev : C ∞ (M. T N ) ⊂ ∞ ∞ C (M. N are ﬁnite dimensional smooth manifolds. Theorem. M ) denote the space of all proper smooth mappings P → M (i. Using the description of smooth curves in C ∞ (M. Let M and N be smooth manifolds. N ) and g ∈ Cprop (P.8. This is carried out in great detail in [Michor. M ). N ) given in 6.7.1 for C ∞ (M. Let (c1 .e. N ) × M → N is smooth. N ) → ∞ ∞ C (P. a suitable compact / subset of M (equivalently. but P and M have to be compact. N ) × ∞ Cprop (P. M ) since changing a mapping only on a compact set does not change its property of being proper. M ) be a smooth curve. The proof for real analytic manifolds is similar. T N ) we denote the space of all smooth mappings f : M → T N such that f (x) = 0πM f (x) for x ∈ Kf . N ) ∞ is smooth. c2 ) : R → C ∞ (M. p)) and one may check that this has again property (1). so ev is smooth as claimed. N ) are canonically isomorphic. If P is another smooth manifold. c1 (t)) is smooth for each ˆ smooth (c1 . T N ) over C (M.20 6. This is open in C ∞ (P. One can check directly that the atlas from 6. N ) are smooth for f ∈ C ∞ (N. If M . 6. Exponential law. T N ) via some natural identiﬁcations in T T N . Then the inﬁnite ∞ dimensional smooth vector bundles T C ∞ (M. ∞ The space of proper mappings Cprop (P. N ) × M . c2 (t. compact sets have compact inverse images). ∞ Proof. N ) → C ∞ (M. c2 ) : R → C ∞ (M. M ) is open in the manifold ∞ C (P. M ) → C ∞ (P. where Cprop (P. so it ˆ ˆ is a smooth curve in C ∞ (P. Proof. N ) which is equivalent to that for C ∞ (M. such that the associated section of the pull back bundle (πM ◦ f )∗ T N → M has compact support). N ) and g ∗ : C ∞ (M. 6. if M is compact. c2 ))(t)(p) = c1 (t.19]. Theorem. N ). Manifolds of mappings 6.2 we immediately see that (ev ◦(c1 . 1980. Let M be a (possibly inﬁnite dimensional) smooth manifold. The same assertion is true for C ω (M. M ). Here by Cc (M. and let M and N be ﬁnite dimensional smooth manifolds. The corresponding statement for real analytic mappings is also true. In particular f∗ : C ∞ (M. Thus comp is smooth. Theorem.

where the image in the right hand side consists of all smooth mappings f : M × M → N which satisfy the following property (1) For each point in M there is an open neighborhood U and a compact set K ⊂ M such that the restriction f |(U × (M \ K)) is constant. This follows directly from the description of all smooth curves in C ∞ (M.6. C ∞ (M. We have equality if and only if M is compact. If M and N are real analytic manifolds with M compact we have C ω (M. N ) for each real analytic (possibly inﬁnite dimensional) manifold. N ) given in the proof of 6. N )) = C ω (M × M. Manifolds of mappings 21 Then we have a canonical embedding C ∞ (M. N )) ⊆ C ∞ (M × M → N ).6. C ω (M. . Proof.

To show that the inversion inv is smooth. e. Then the mapping c : R × M → M satˆ isﬁes 6. M ). Diﬀeomorphism group.X. For a smooth manifold M the group Diﬀ(M ) of all smooth diﬀeomorphisms of M is an open submanifold of C ∞ (M. o By the chart structure from 6. thus T (ρf ). Property 6. m). M ).1. The composition is smooth by theorem 6. Since the topology we use on C ∞ (M. Proof. M ) for the Whitney C ∞ -topology. Likewise Tf Diﬀ(M ) = Cc (f ∗ T M ) which we identify with the space of all vector ﬁelds with compact support along the diﬀeomorphism f .6.1. Thus the exponential mapping of the diﬀeomorphism group is given by Exp = Fl1 : Xc (M ) → Diﬀ(M ). This proof is by far simpler than the original one. 1980].22 7. M ). Diﬀ(M ) is an open submani∞ fold of Cprop (M. For a compact real analytic manifold M the group Diﬀ ω (M ) of all real analytic diﬀeomorphisms is a real analytic Lie group with Lie algebra C ω (T M ) and with real analytic exponential mapping. Diﬀeomorphism groups 7. t So the one parameter group t → FlX ∈ Diﬀ(M ) is the integral curve of t the right invariant vector ﬁeld RX : f → T (ρf ). Right translation ρf is given by ρf (g) = f ∗ (g) = g ◦ f . and shows the power of the Fr¨licher-Kriegl calculus. or directly from theorem 6.(1) is obvious.6. a time dependent vector ﬁeld with compact support Xt . The exponential mapping Exp : Cc (T M ) → Diﬀ ∞ (M ) is the ﬂow mapping to time 1.X = X ◦ f and for the ﬂow FlX of the vector t d X X ﬁeld with compact support X we have dt F lt = X ◦ F lt = T (ρFlX ). Thus Exp maps smooth curves to smooth curves and is smooth . We may view it as a complete vector ﬁeld (0t . So inv maps smooth curves to smooth curves and is thus smooth. Theorem. By the ﬁnite dimenˆ sional implicit function theorem (inv ◦c) is smooth in (t.6. see [Michor. equipped with the negative of the usual Lie ∞ bracket. To show that is smooth we consider a smooth curve in Xc (M ).(1) and (inv ◦c) fulﬁlls the ﬁnite dimensional implicit equation c(t.7we see ∞ that the tangent space Te Diﬀ(M ) equals the space Cc (T M ) of all vec∞ tor ﬁelds with compact support. and it is smooth.X = X ◦ f on Diﬀ(M ). It is well known that the space Diﬀ(M ) of all diﬀeomorphisms of M is open in C ∞ (M. we consider a smooth curve c : R → Diﬀ(M ) ⊂ C ∞ (M. Xt ) on R × M whose smooth ﬂow respects the level surfaces {t} × M and is smooth. m)) = m for all t ∈ R and m ∈ M . M ) is ﬁner. i. composition and inversion are smooth. (inv ◦c) (t. The Lie algebra of the smooth inﬁnite dimensional Lie group Diﬀ(M ) ∞ is the convenient vector space Cc (T M ) of all smooth vector ﬁelds on M with compact support.

the negative of the usual Lie bracket on Xc (M ). Using tubular neighborhoods or submanifold charts one can . . which is not real analytic in t for general smooth X. 1983. ∞ The exponential mapping Exp : Cc (T M ) → Diﬀ(M ) is in a very strong sense not surjective: In [Grabowski. It is even real analytic. Ad(Exp(sX))Y = = = ∂ ∂t 0 ∂ ∂t 0 Exp(sX) ◦ Exp(tY ) ◦ Exp(−sX) Ad(Exp(tX))Y = T (FlX ) ◦ Y ◦ FlX t −t (FlX )∗ Y.6 the left action ev : Diﬀ(M ) × M → M is smooth. The mapping ∞ Ad ◦ Exp : Cc (T M ) → Diﬀ(M ) → L(C ∞ (T M ).2. . C ∞ (T M )) is not real analytic. xn ) and (y1 . . 4. 3.2. . Y ]. 7. 1984. To understand this contemplate 1.2]. see [Pressley-Segal. 1986. The group Diﬀ(M ) of smooth diﬀeomorphisms does not carry any real analytic Lie group structure by [Milnor.3]. 9. see [Michor. Remarks. 1984. thus −t X ∗ ∂ ∂t 0 (Fl−t ) Y = −[X. . yn ) of pair wise diﬀerent elements of M there is a diﬀeomorphism f ∈ Diﬀ(M ) with f (xi ) = yi for each i. ∞ For x ∈ M the mapping evx ◦ Exp : Cc (T M ) → Diﬀ(M ) → M is not X real analytic since (evx ◦ Exp)(tX) = F lt (x). i.6]. 1] → M which is an embedding.7. compare the proof of theorem 7. who weakened the original concept of [Omori. . 1983. 7. Diﬀeomorphism groups 23 itself.(4). This condition means that the mapping associating the evolution operator to each time dependent vector ﬁeld on M is smooth. Let us ﬁnally compute the Lie bracket on Xc (M ) viewed as the Lie algebra of Diﬀ(M ).2. Again one may compare this simple proof with the original one in [Michor. for two ordered ﬁnite n-tupels (x1 . If dim M ≥ 2 this action is n-transitive for each n ∈ N. The real analytic Lie group Diﬀ ω (M ) is regular in the sense of [Milnor. .11]. section 4]. 1988] it is shown. and it has no complexiﬁcation in general. By theorem 6. that Diﬀ(M ) contains an arcwise connected free subgroup on 2ℵ0 generators which meets the image of Exp only at the identity. . e. To see this connect xi with yi by a simple smooth curve ci : [0. 1982]. such that the images are pair wise disjoint.

whose smooth base manifold is the space B(M. Theorem. see [Kriegl-Michor. Let us consider the mapping τ = τ i := (expN ◦¯ i)|U : N (i) ⊃ U → N. Then let p : N (i) → M be the normal bundle of i. Let M and N be smooth manifolds. Let g be a ﬁxed Riemannian metric on N and let expN be its exponential mapping. .3. N ) of B(M. Then N (i) − − → T N −− π p M −−→ N −− i ¯ i N is an injective vector bundle homomorphism over i. N ) consisting of all closed submanifolds of N of type M . N ) of all smooth embeddings M → N is an open submanifold of C ∞ (M. 1990a]. N ) of proper (equivalently closed) embeddings is saturated under the Diﬀ(M )-action and is thus the total space of the restriction of the principal bundle to the open submanifold Bclosed (M. N ). There is. Then the set Emb(M. Principal bundle of embeddings. In contrast to this one knows from [Omori. 7. inverse limit of Hilbert manifolds) which sometimes permits to use an implicit function theorem. 1974] for this. Proof. In the real analytic case this theorem remains true provided that M is supposed to be compact. It is the total space of a smooth principal ﬁber bundle with structure group Diﬀ(M ). e. however. The diﬀeomorphism Exp(X) = FlX 1 then maps xi to yi . N ) of all submanifolds of N of type M . The open subset Embprop (M. See [Omori. Let us ﬁx an embedding i ∈ Emb(M. the possibility to view Diﬀ(M ) as an ILH-group (i. 1978] that a Banach Lie group acting eﬀectively on a ﬁnite dimensional manifold is necessarily ﬁnite dimensional.24 7. N ). Now let U i = U be an open neighborhood of the zero section of N (i) which is so small that (expN ◦¯ i)|U : U → N is a diﬀeomorphism onto its image which describes a tubular neighborhood of the submanifold i(M ). Diﬀeomorphism groups show that then there is a vector ﬁeld with compact support X ∈ Xc (M ) which restricts to ci (t) along ci (t). deﬁned in the following way: For x ∈ M let N (i)x := (Tx i(Tx M ))⊥ ⊂ Ti(x) N be the g-orthogonal complement in Ti(x) N . section 6. So there is no way to model the diﬀeomorphism group on Banach spaces as a manifold.

N ) we have W(i) := U(i) ◦ Diﬀ(M ) ∼ Q(i) × Diﬀ(M ) ∼ (Cc (U i ) ∩ V(i)) × Diﬀ(M ). Furthermore π|Q(i) : Q(i) → Emb(M. Then there is an immersion h ∈ W(i) ∩ Q(j). N ) : j(M ) ⊆ τ i (U i ). We consider now the open set {h ◦ f : h ∈ V(i). N (i)) is a smooth mapping which is bijective onto the open set V(i) := {h ∈ C ∞ (M. N )/ Diﬀ(M ) is bijective onto an open set in the quotient. 1980. and the open set in Emb(M. In order to investigate the chart change let j ∈ Emb(M. N (i)). So if we let Q(i) := τ∗ (Cc (U i ) ∩ V(i)) ⊂ Emb(M. section 4] the set U(i) is an open neighborhood of i in Emb(M. ∞ Obviously we have a smooth mapping from it into Cc (U i ) × Diﬀ(M ) i −1 ∞ given by h → (h ◦ (p ◦ h) . N (i)) : h(M ) ⊆ U i . Its inverse is given by the smooth mapping τ∗ : h → i τ ◦ h. h ∼ 0} i in C ∞ (M. . we have Q(i)∩Q(j) = ∅ and consequently U(i) ∩ U(j) = ∅. N ) which will serve us as a saturated chart. p ◦ h). U i )). We now consider ϕi ◦ (π|Q(i))−1 : π(Q(i)) → C ∞ (U i ) as a chart for the quotient space. so there exists a unique f0 ∈ Diﬀ(M ) −1 −1 (given by f0 = p ◦ ϕi (h)) such that h ◦ f0 ∈ Q(i). For each j ∈ U(i) we deﬁne ϕi (j) : M → U i ⊆ N (i). N ). Diﬀeomorphism groups 25 a diﬀeomorphism onto its image. N ) be such that π(Q(i)) ∩ π(Q(j)) = ∅. where j ∼ i means that j = i oﬀ some compact set in M . ϕi (j)(x) := (τ i )−1 (j(x)) Then ϕi = ((τ i )−1 )∗ : U(i) → C ∞ (M. where Cc (U ) is the space of sections ∞ i with compact support of U i → M . If we consider j ◦ f0 instead of j and call it again j. f ∈ Diﬀ(M )} ⊆ C ∞ ((M.7. Then by [Michor. Then the chart change is given as follows: ∞ ∞ ϕi ◦ (π|Q(i))−1 ◦ π ◦ (τ j )∗ : Cc (U j ) → Cc (U i ) s → τ j ◦ s → ϕi (τ j ◦ s) ◦ (pi ◦ ϕi (τ j ◦ s))−1 . U(i) := {j ∈ Emb(M. j ∼ i}. i i We have τ∗ (h ◦ f ) = τ∗ (h) ◦ f for those f ∈ Diﬀ(M ) which are near enough to the identity so that h ◦ f ∈ V(i). = = ∞ since the action of Diﬀ(M ) on i is free.

a submanifold of N. Both V and W are Diﬀ(M ) saturated. N ) : k(M ) ⊂ W }. 7. n = dim N . N ) be set of all proper immersions. We choose an open neighborhood V of y0 in N and an open neighborhood W of j(M ) in N such that V ∩ W = ∅. Then the diﬀeomorphism group Diﬀ(M ) acts smoothly from the right on the smooth manifold Immprop (M. Then V is obviously open in Emb(M. N ) = Emb(M. Let (WL . j ∈ Emb(M. so π(i) = π(j). N )/ Diﬀ(M ). j ∈ V. which is an open subset of C ∞ (M. where n = dim M .prop (M. 1989]. N ) → Immfree. say. So π(V) and π(W) separate π(i) and π(j) in B(M. Case 2 Let i(M ) ⊂ j(M ) and j(M ) ⊂ i(M ).prop (M. Let (V.26 7. Since j(M ) ⊆ i(M ) we conclude that we have also v((i(M ) ∪ j(M )) ∩ V ) ⊆ Rm ∩v(V ). We show that the quotient space B(M. N ). N ) with π(i) = π(j). pL . contrary to the assumption. where also the existence of smooth transversals to each orbit is shown and the stratiﬁcation of the orbit space into smooth manifolds is given. on which Diﬀ(M ) acts freely. Now we distinguish two cases. N )/ Diﬀ(M ) is Hausdorﬀ in the quotient topology. the mapping i−1 ◦ j : M → L → M is then a diﬀeomorphism of M and j = i◦(i−1 ◦j) ∈ i◦Diﬀ(M ). Let Immfree.4. Theorem. Case 1. Let V := {k ∈ Emb(M. Both composants are smooth by the general properties of manifolds of mappings. Let M and N be smooth manifolds. Let y ∈ i(X). Then the space of orbits Immprop (M. We may ﬁnd a point y0 ∈ i(M ) \ j(M ). v) be a chart of N centered at y which maps i(M ) ∩ V into a linear subspace. N ) : k(M ) ∩ V } W := {k ∈ Emb(M. So we see that L := i(M )∪j(M ) is a submanifold of N of the same dimension as N . Diﬀeomorphism groups This is of the form s → β ◦ s for a locally deﬁned diﬀeomorphism β : N (j) → N (i) which is not ﬁber respecting. N ) and V is even open in the coarser compact open topology. prop (M. followed by h → h◦(pi ◦h)−1 . i ∈ W. v(i(M )∩V ) ⊆ Rm ∩v(V ) ⊂ Rn . This is proved in [Cervera-Mascaro-Michor. So the chart change is smooth. Then WL |i(M ) is a tubular neighborhood of i(M ) and WL |j(M ) is one of j(M ). N )/ Diﬀ(M ) is Hausdorﬀ. . say. and V ∩ W = ∅. N ) and is the total space of a smooth principal ﬁber bundle Immfree. L) be a tubular neighborhood of L. Let i. Then this is open in C ∞ (M. N ) of all smooth proper immersions M → N . Then i(M ) = j(M ) in N for otherwise put i(M ) = j(M ) =: L. N ). which is not a cluster point of j(M ).

D2 ]. . . The space Derk Ω(M ) consists of all (graded) derivations of degree k. Plug in the deﬁnition of the graded commutator and compute. . It is ∗ ∗ uniquely determined by its restriction to 1-forms Dx |Tx M : Tx M → ∗ Λk+1 T ∗ M which we may view as an element Kx ∈ Λk+1 Tx M ⊗ Tx M depending smoothly on x ∈ M . . (1) For K ∈ Ωk+1 (M. . Then D(f. L ∈ Ω +1 (M. L] = iK L − (−1)k iL K. A derivation D ∈ Derk Ω(M ) is called algebraic if D | Ω0 (M ) = 0. D2 ] = −(−1)k1 k2 [D2 . In this section let M be a smooth manifold. The Fr¨licher-Nijenhuis Bracket o 8. . . We consider the graded dim M k commutative algebra Ω(M ) = Ω (M ) of diﬀerential forms on k=0 M . T M ) the space of all vector valued diﬀerential forms. Note also that the important formula LX = d iX + iX d translates to LX = [iX . D2 ] := D1 ◦ D2 − (−1)k1 k2 D2 ◦ D1 as bracket. ] is itself a derivation). Xk+ ) = = 1 (k+1)! ( −1)! σ∈§k+ sign σω(K(Xσ1 . . . (2) By i([K. d]. D3 ]] = [[D1 . i.27 8. Xσ(k+2) .1. all linear mappings D : Ω(M ) → Ω(M ) with D(Ω (M )) ⊂ Ωk+ (M ) and D(ϕ ∧ ψ) = D(ϕ) ∧ ψ + (−1)k ϕ ∧ D(ψ) for ϕ ∈ Ω (M ). LX is of degree 0. ] on the space Ω∗−1 (M. iL ] we get a bracket [ . T M ). so D is of tensorial character ∗ and induces a derivation Dx ∈ Derk ΛTx M for each x ∈ M . L] ) := [iK . [D1 . T M ) we have [K. ) for ω ∈ Ω (M ). D3 ]] (so that ad(D1 ) = [D1 . D1 ].D(ω) for f ∈ C ∞ (M. Xi ∈ X(M ) (or Tx M ) deﬁnes an algebraic graded derivation iK ∈ Derk Ω(M ) and any algebraic derivation is of this form. From the basics of diﬀerential geometry one is already familiar with some graded derivations: for a vector ﬁeld X on M the derivation iX is of degree −1. dim M We call Ω(M. and for K ∈ Ωk+1 (M. Xσ(k+1) ). 8. T M ).2. This means that the bracket is graded anticommutative. Then the space Der Ω(M ) = k Derk Ω(M ) is a graded Lie algebra with the graded commutator [D1 . T M ) the formula (iK ω)(X1 . [D2 . Proof. T M ) which deﬁnes a graded Lie algebra structure with the grading as indicated.e. Theorem. D3 ] + (−1)[ D2 . and satisﬁes the graded Jacobi identity [D1 .ω) = f. T M ) = k=0 Ωk (M. R). Lemma. To express this dependence we write D = iK = i(K). where K ∈ C ∞ (Λk+1 T ∗ M ⊗ T M ) =: Ωk+1 (M . . . [D1 . and d is of degree 1. .

d] = 2d2 = 0. Xk ) ∈ X(M ) such that (Df )(X1 . Xk ) = K(X1 . . . . Then f → (Df )(X1 . . R) → C ∞ (M. 1967]. . Thus L(IdT M ) = d. Proof. Xk ) is a derivation C ∞ (M. T M ) the Lie derivation LK = L(K) ∈ Derk Ω(M ) by LK := [ik . Theorem. so K is tensorial. ∗ Proof. so the last assertions follow.3. Since we have [d. . D is algebraic if and only if K = 0.28 8. d] = iX d + d iX for vector ﬁelds X. T M ) such that D = LK + iL . so D −LK = iL by 8. Since Λ∗ Tx M is the free graded commutative algebra generated ∗ by the vector space Tx M any K ∈ Ωk+1 (M. The graded commutator of two algebraic derivations is again algebraic. d] = LK is injective. T M ) and L ∈ Ωk+1 (M . K ∈ Ωk (M . 8. T M ) → Derk (Ω(M )) induces a graded Lie bracket on Ω∗+1 (M. . We have L = 0 if and only if [D. . see [Nijenhuis-Richardson. . Xk )f = df (K(X1 . . . we deﬁne for K ∈ Ωk (M . . R). . by the graded Jacobi identity we obtain 0 = [iK . . Xk )). Applying i(IdT M ) on a k-fold exterior product of 1-forms we see that i(IdT M )ω = kω for ω ∈ Ωk (M ). . so the injection i : Ω∗+1 (M. Fr¨licher-Nijenhuis bracket o [ . so there is a unique vector ﬁeld K(X1 . . d]] = 2[LK . . d]. since LK f = iK df = df ◦ K for f ∈ C ∞ (M. Thus D −LK is an algebraic derivation. The deﬁning equation for K is Df = df ◦ K = iK df = LK f for f ∈ C ∞ (M. . The mapping L → [iL . ] is called the algebraic bracket or also the Nijenhuis-Richardson bracket. d]. d]] = [[iK . T M ) → Der Ω(M ) is injective. . Then the mapping L : Ω(M. 8. d] + (−1)k−1 [d. [iK . d] = 0. . T M ). . For any graded derivation D ∈ Derk Ω(M ) there are unique K ∈ Ωk (M . T M ). d]. .2 for unique L ∈ Ωk+1 (M . T M ) extends to a graded derivation. . . T M ) whose form can be seen by applying it to a 1-form.4. By applying it to an exterior product of 1-forms one can derive the formula in (1). R)-linear in each Xi and alternating. . In view of the formula LX = [iX . Clearly K(X1 . The exterior derivative d is an element of Der1 Ω(M ). Thus we have L(IdT M ) = ω = i(IdT M )dω − d i(IdT M )ω = (k + 1)dω − kdω = dω. [d. R). Let Xi ∈ X(M ) be vector ﬁelds. Xk ) is C ∞ (M. R).

LY ]f . Y ] = L([X. R) we have [iL . Proof. 8. df ◦ [X. the algebraic part of [iL . L : (Ω(M . T M ) and L ∈ Ω +1 dim M (M . IdT M ∈ Ω1 (M . Then we have: (1) (2) (3) (4) [ω ∧ D1 . The space Ω(M . Theorem. 8. d]] = = 0 − (−1)k L([K. [ . i(ω ∧ L) = ω ∧ i(L) ω ∧ LK = L(ω ∧ K) + (−1)q+k−1 i(dω ∧ K). [K.8. IdT M ] = 0 for all K. d] = [iL . [iL . T M ) we have [LK . LK ] = L(iL K) − (−1)k i([L. K]). T M ) with its usual grading is a graded Lie algebra. Proof. T M ). of ϕ be k. Let the degree of ω be q. [LK .e. For f ∈ C ∞ (M. L]) = (−1)k [i([L. The rest is clear. L] ∈ Ωk+ (M . T M ). For vector ﬁelds the Fr¨licher-Nijenhuis bracket o coincides with the Lie bracket. ]) → Der Ω(M ) is an injective homomorphism of graded Lie algebras. LK ] − L(iL K) is an algebraic derivation. and of ψ be . [[LK . Since [ . LK ]f = iL iK df − 0 = iL (df ◦ K) = df ◦ (iL K) = L(iL K)f . d] = 0. This generalizes the usual formula for vector ﬁelds. Lemma.5. Let the other degrees be as indicated. For K ∈ Ωk (M . d] kills the ”L’s” and is injective on the ”i’s”. L]) − (−1)k L(iL K).6. L2 ] = ω ∧ [L1 . LL ].7. Let K ∈ Ωk (M . [[iL . So [iL . o Theorem. K]). T M ) is in the center. T M ) = k=0 Ωk (M . This vector valued form [K. Y ])f = [LX . i. so we have [L(K). iL ] = i([K. K]). L]) Then obviously for a uniquely deﬁned [K. L(L)] = L([K. The space Der Ω(M ) is a graded module over the graded algebra Ω(M ) with the action (ω ∧ D)ϕ = ω ∧ D(ϕ). Fr¨licher-Nijenhuis bracket o 29 8. T M ). L] is called the Fr¨licher-Nijenhuis bracket of K and L. LK ] is (−1)k i([L. L2 ] − . [ω ∧ L1 . LK ]. D2 ] − (−1)(q+k1 )k2 d2 (ω) ∧ D1 . d]] − (−1)k [LK . d]). D2 ] = ω ∧ [D1 . T M ) and L ∈ Ω (M . or [iL .

. For (1) . . .7. For K ∈ Ωk (M . L2 ] ). L] the following formula. . Xσ(k+2) . . For K ∈ Ωk (M . It suﬃces to consider K = ϕ ⊗ X. T M ) and ω ∈ Ω (M ) the Lie derivative of ω along K is given by the following formula. . Xσk ). Proof. (2) . Xσ(k+1) ]. Now use the usual global formulas to expand this. . . For (4) compute i([ω ∧ L1 . . .8. 8. Xσ(k+1) ]. . . . . . . ). Xσk ). Theorem. Theorem. . . L](X1 . . Y ] + ϕ ∧ LX ψ ⊗ Y − LY ϕ ∧ ψ ⊗ X + (−1)k (dϕ ∧ iX ψ ⊗ Y + iY ϕ ∧ dψ ⊗ X) . For (6) use (5) . Xσ(k+ ) )) sign σ ω([K(Xσ1 . . . . L(Xσ(k+1) . . where the Xi are vector ﬁelds on M . Xk+ ) = = + + 1 k! ! σ −1 k! ( −1)! σ (−1)k−1 (k−1)! ( −1)! 2! σ sign σ L(K(Xσ1 . . (LK ω)(X1 . K2 ] = ω ∧ [K1 . [ω ∧ K1 . . T M ) and L ∈ Ω (M . .3 we have L(ϕ ⊗ X) = ϕ ∧ LX − (−1)k−1 dϕ ∧ iX .9. Xσ3 . [K. . ) . . Xσk ). Xσ(k+2) . . 8. Xσ(k+ ) )] sign σ L([K(Xσ1 . Xσ(k+2) . Then by 8.30 8. . ψ ⊗ Y ] = ϕ ∧ ψ ⊗ [X. Fr¨licher-Nijenhuis bracket o − (−1)(q+ (5) (6) 1 −1)( 2 −1) i(L2 )ω ∧ L1 . . . Y ] − iY dϕ ∧ ψ ⊗ X − (−1)k iX dψ ∧ ϕ ⊗ Y − d(iY ϕ ∧ ψ) ⊗ X − (−1)k d(iX ψ ∧ ϕ) ⊗ Y = ϕ ∧ ψ ⊗ [X. . . . ). T M ) we have for the Fr¨licher-Nijenhuis bracket [K. . . . ) sign σ ω(K([Xσ1 . Xσk ))(ω(Xσ(k+1) . K2 ] − (−1)(q+k1 )k2 L(K2 )ω ∧ K1 + (−1)q+k1 dω ∧ i(K1 )K2 . . For (5) compute L([ω ∧ K1 . . Proof. Xσ2 ]. where the Xi o are vector ﬁelds on M . . Global formula for the Fr¨licher-Nijenhuis o bracket. . [ϕ ⊗ X. Xk+ ) = = + 1 k! ! σ −1 k! ( −1)! σ sign σ [K(Xσ1 . K2 ]). . (3) write out the deﬁnitions. .

L2 ] − (−1)k1 k2 [K2 . . see [Michor..... Xσ(k+2) . It suﬃces to consider K = ϕ ⊗ X and L = ψ ⊗ Y ... .αk+ i + (−1)k (∂α1 Kα1 . .α ∂i Kα +1 . Xσ2 ]. Xσ( +2) .αk−1 i ∂αk Li k+1 . L | U = Lj dβ ⊗ ∂j .. . ). ...αk Lj k+1 .αk ) ωiαk+2 .. Xσ3 . . .11. Proof. we get the following local formulas: iK ω | U = [K. . . . . Xσ3 . .. For Ki ∈ Ωki (M .. . L2 ] + [K1 . Xσ( +2) . and 8.αk+ α i + (−1)k Lj 1 .. LK2 + iL2 ] = = L [K1 .. T M ) we have (1) [LK1 + iL1 . Xk+ ). 1987].8. There are more illuminating ways to prove this formula. After some combinatorial computation we get the right hand side of the above formula for K = ϕ ⊗ X and L = ψ ⊗ Y .. L] | U = i Kα1 .8. β where α = (1 ≤ α1 < α2 < · · · < αk ≤ dim M ) is a form index. then for [ϕ ⊗ X... Local formulas..αk ωiαk+1 . )..αk+ α dα ⊗ ∂j 8.α −1 i ∂α Kα +1 . T M ) and Li ∈ Ωki +1 (M . 8.αk+ α j − kKα1 . . Xσ2 ]. ...αk ∂i Lj k+1 . ) + + + (−1)k (k−1)! ! σ sign σ K([L(Xσ1 . .α Kiα +1 . ∂i = ∂ui and so on. Xσ( (−1)k−1 (k−1)! ( −1)! 2! σ (−1)(k−1) (k−1)! ( −1)! 2! σ sign σ L(K([Xσ1 . Plugging Xj = ∂ij into the global formulas 8. K2 ] + iL1 K2 − (−1)k1 k2 iL2 K1 + i [L1 . ) sign σ K(L([Xσ1 . . .αk+ dα [K. . u) on the manifold M we i put K | U = Kα dα ⊗ ∂i .αk+ −1 dα i Kα1 . Fr¨licher-Nijenhuis bracket o 31 +1) ].. 8.7... ψ ⊗ Y ] we may use 8. . ∂ dα = duα1 ∧ ..αk+ iα − (−1)(k−1)( LK ω | U = −1) j Li 1 .6 and evaluate that at (X1 .αk+ α dα ⊗ ∂j i Kα1 .2.. ∧ duαk .αk+ α j − (−1)k Li 1 ... .. Theorem. . ). .. ..10.. . L] | U = i Kα1 .... . L1 ] .. and ω | U = ωγ dγ . Xσ ).9.αk ∂i ωαk+1 . In a local chart (U.

but the mappings i : Ω(M . Curvature. [ . ]) ]) do not take values in the subspaces of graded derivations. X]) − K([LX. For the general theory see [Michor.12. LY ] − [KY. T M ) and L ∈ Ω +1 (M .12. Y ] − [LY. We may call ker P the horizontal space and im P the vertical space of the connection. then both are sub vector bundles of T M . i. Let P ∈ Ω1 (M . P can be called a connection for it. iL K2 ] − (−1)k1 i([K1 . LX] − L([KX.e. Y ]. Fr¨licher-Nijenhuis bracket o Each summand of this formula looks like a semidirect product of graded Lie algebras. The corresponding product of groups is well known to algebraists under the name ”Zappa-Szep”-product. T M ). T M ) be a ﬁber projection. L])K2 − (−1)(k1 + (3) )k2 i([K2 . Y ) = [KX. [ ad : Ω(M . If P is of constant rank. We have instead for K ∈ Ωk (M . [L1 .6 several times to obtain i of the right hand side of (3). Equation (1) is an immediate consequence of 8. L2 ] ]) and use again 8. 1989a] The following result is immediate from 8.6. L2 ] + (−1)kk1 [L1 . K2 ] = [iL K1 . L ∈ Ω1 (M . 1989b]. [K.32 8. X]) + (LK + KL)[X.9). L1 ]. T M ) the following relations: (2) iL [K1 . For K. L1 ] The algebraic meaning of the relations of this theorem and its consequences in group theory have been investigated in [Michor. T M ) → End(Ω(M . L2 ]] − − (−1)kk1 [i(L1 )K. 8. Y ] − [KY. This is the most general case of a (ﬁrst order) connection. L](X. Proof. K2 ]) and use 8. Equations (2) and (3) follow from (1) by writing out the graded Jacobi identity. Corollary (of 8. [L1 . L2 ] − (−1)(k+k1 )k2 [i(L2 )K. L2 ] ] = [[K. Special cases of this will be treated extensively later on. . If im P is some primarily ﬁxed sub vector bundle or (tangent bundle of) a foliation. K2 ] + (−1)k1 [K1 . . Then consider i([K. or as follows: Consider L(iL [K1 . L])K1 [K. T M ) → End(Ω(M . 8. T M ) we have [K. P ◦ P = P .13. T M ).6 repeatedly to obtain L of the right hand side of (2).

if for all Xi ∈ Tx M we have (1) Kf (x) (Tx f · X1 . By 8. (5) If K and K are f -related then LK ◦ f ∗ = f ∗ ◦ LK : Ω(N ) → Ω(M ). Y ) = P [(Id − P )X. . 8.P ] [P.P ] P . . and also [K1 . where B 1 denotes the space of exact 1-forms. Naturality of the Fr¨licher-Nijenhuis bracket. (4) If Kj and Kj are f -related for j = 1. [P. (Id − ¯ P )Y ] and R(X. likewise ¯ for R. K2 ] and [K1 . then K and K are f -related. Two vector valued forms K ∈ Ωk (M . Theorem. So the ﬁrst formula follows. . Xk ).8. We have 2R = P ◦ [P. . R ∈ Ω2 (M . then iK1 K2 and iK1 K2 are f -related. If P has constant rank. P ] = iR R + iR R. T M ) are given by R(X. T M ) is a ¯ connection (ﬁber projection) with curvature R and cocurvature R.P ] [P. (3) If iK ◦ f ∗ | B 1 (N ) = f ∗ ◦ iK | B 1 (N ). Therefore [R. Tx f · Xk ) = Tx f · Kx (X1 . P ] = 2R + 2R by 8. P ].P ] P. Let f : o M → N be a smooth mapping between manifolds. We have ¯ [P. We have [P. (2) If K and K as above are f -related then iK ◦ f ∗ = f ∗ ◦ iK : Ω(N ) → Ω(M ). P ] = 4 i[P.11. . T N ) are called f -related or f -dependent. then we have ¯ [P.14. P ] − 0 = 1 ¯ ¯ ¯ ¯ 4[R. P ] = 2R + 2R. that for a principal ﬁber bundle R is just the negative of the usual curvature. P ] = i(R+ R)(R+ R) = iR R+iR R since R has vertical values and kills vertical vectors. 2. R + R] = 0 ¯ [R. ¯ ¯ Proof. General Bianchi identity. 8. and R is the obstruction against integrability of the vertical bundle im P . Thus we call R the curvature ¯ and R the cocurvature of the connection P . so iR R = 0. T M ) and K ∈ Ωk (N . Lemma. P ] = i[P. P ] = 2[i[P. Fr¨licher-Nijenhuis bracket o 33 Lemma. . . . . ¯ where R. P Y ].13 and [P. K2 ] are f -related. P ]] = 0 by the graded Jacobi identity. We will see later. then R is the obstruction against integra¯ bility of the horizontal bundle ker P . Y ) = (Id − P )[P X.15. If P ∈ Ω1 (M .2 we get i[P.

K2 ] and [K1 . the t Fr¨licher-Nijenhuis-bracket. K2 ]). o . L] = [f ∗ K. Xσk ). . Xq+k−1 ) (3) follows from this computation. T M ) the Lie ∂ derivative LX K = ∂t 0 (FlX )∗ K satisﬁes LX K = [X. . . . . respectively. the result for the bracket then follows 8. (6) For g ∈ Ω0 (N ) we have LK f ∗ g = iK d f ∗ g = iK f ∗ dg and ∗ f LK g = f ∗ iK dg. Theorem. given by (1) (F ∗ K)x (X1 . K2 ] are also f -related. . ). . . . Xσ(k+1) . Now use (6) . so also their graded commutators which equal L([K1 . . f ∗ L]. then K and K are f -related. . . . (5) The algebra homomorphism f ∗ intertwines the operators iK and iK by (2).34 8. d] = LK . (6) The algebra homomorphism f ∗ intertwines LKj and LKj . . . Then we can consider the pullback operator f ∗ : Ω(M . R) separate points. Thus f ∗ intertwines the commutators [iK .2. L] = [f ∗ K. . Proof. . Tx f · Xσ(k+1) . (4) follows from the same computation for K2 instead of ω. In this situation we have: (2) f ∗ [K. . ) = (f ∗ iK ω)x (X1 . . K]. (2) By 8. . ). f ∗ L] . f ∈ C ∞ (M. . Xk ) = (Tx f )−1 Kf (x) (Tx f · X1 . d] = LK and [iK . ) sign σ ωf (x) (Tx f · Kx (Xσ1 . K2 ]) and L([K1 . (7) If Kj and Kj are f -related for j = 1. Tx f · Xk ).2 we have for ω ∈ Ωq (N ) and Xi ∈ Tx M : (iK f ∗ ω)x (X1 . . . .16. . Let f : M → N be a local diﬀeomorphism.2. . Clearly K and f ∗ K are then f -related. . . (3) f ∗ iK L = if ∗ K f ∗ L. (5) For a vector ﬁeld X ∈ X(M ) and K ∈ Ω(M . since the df . . . 2. T M ). Xq+k−1 ) = = = = 1 k! (q−1)! σ 1 k! (q−1)! σ 1 k! (q−1)! σ sign σ (f ∗ ω)x (Kx (Xσ1 . then their Fr¨lichero Nijenhuis brackets [K1 . Fr¨licher-Nijenhuis bracket o (6) If LK ◦ f ∗ | Ω0 (N ) = f ∗ ◦ LK | Ω0 (N ). Tx f · Xσ(k+1) . (4) f ∗ [K. . and f ∗ commutes with the exterior derivative d. . By (3) the result follows. T M ) → Ω(M . 8. . . . ) sign σ ωf (x) (Kf (x) (Tx f · Xσ1 .

(5) Obviously LX is R-linear.12 we have [J. Y ) = 2([JX. Y ]). etc. A vector o valued 1-form J ∈ Ω1 (M . ψ ∈ Ω(M ) and Y ∈ X(M ). J] = 0. Fr¨licher-Nijenhuis bracket o 35 We may say that the Fr¨licher-Nijenhuis bracket.7.17. T M ) with J ◦ J = −Id is called a nearly complex structure.e.8. But then LX (ψ ⊗ Y ) = LX ψ ⊗ Y + ψ ⊗ LX Y = LX ψ ⊗ Y + ψ ⊗ [X. . ] . o natural bilinear concomitants. At last we mention the best known application of the Fr¨licher-Nijenhuis bracket. For it the following result is true: A manifold M with a nearly complex structure J is a complex manifold (i. are Proof. 1957]. 8. so it suﬃces to check this formula for K = ψ ⊗ Y . J] is also called the Nijenhuis tensor of J. if it exists. By 8. Y ] = [X. J](X. 1 The vector valued form 2 [J. Remark. See [Newlander-Nirenberg. which also led to its discovery. ψ ⊗ Y ] by 8. dim M is even and J can be viewed as a √ ﬁber multiplication with −1 on T M .6. Y ] − J[X. JY ] − [X. there exists an atlas for M with holomorphic chart-change mappings) if and only if [J.. [ . (2) – (4) are obvious from 8. JY ] − J[JX.16.

Then (N. p.36 9. Fiber Bundles and Connections 9. and a smooth mapping p : E → M . depending smoothly on x ∈ U . and we may use the mappings ψα ([(α. ψα ).u(x). t) = s. They satisfy the cocycle condition: ψαβ (x) ◦ ψβγ (x) = ψαγ (x) for x ∈ Uαβγ and ψαα (x) = IdS for x ∈ Uα . M.1. So let (U. A (ﬁber) bundle (E. p. s) as a bundle atlas whose cocycle of transition functions is exactly the one we started from. S. where ψαβ : (Uα ∩Uβ )×S → S is smooth and ψαβ (x. such that (Uα ) is an open cover of M . We may thus consider the mappings ψαβ : Uαβ → Diﬀ(S) with values in the group Diﬀ(S) of all diﬀeomorphisms of S. p is a surjective submersion. Given an open cover (Uα ) of a manifold M and a cocycle of transition functions (ψαβ ) we may construct a ﬁber bundle (E. Let p : N → M be a surjective submersion (a ﬁbred manifold) such that p−1 (x) is compact for each x ∈ M and let M be connected. x. s) ∼ (β. p. Therefore the collection (ψαβ ) is called a cocycle of transition functions.2. . their diﬀerentiability is a subtle question. y. and S is called standard ﬁber. Proof. M. Deﬁnition. called the projection. s) = (x. ) is a diﬀeomorphism of S for each x ∈ Uαβ := Uα ∩ Uβ . A collection of ﬁber charts (Uα . For each x ∈ U let = ξx (y) := (Ty u)−1 . then ξx ∈ X(U ). then ψα ◦ψβ −1 (x. S) consists of manifolds E. M is called the base space or basis. M ) is a ﬁber bundle.1 below. x. (U. Lemma. t) if and only if x = y and ψαβ (y. ψ) as above is called a ﬁber chart. In either form these mappings ψαβ are called the transition functions of the bundle. We have to produce a ﬁber chart at each x0 ∈ M . u) be a chart centered at x0 on M such that u(U ) ∼ Rm . ψαβ (x. If we ﬁx such an atlas. which will be discussed in 13. 9. s)). The quotient space then turns out to be a ﬁber bundle. s)]) := (x. furthermore each x ∈ M has an open neighborhood U such that E | U := p−1 (U ) is diﬀeomorphic to U × S via a ﬁber respecting diﬀeomorphism: E |U −−→ U ×S −− p pr1 U U ψ E is called the total space. M . is called a (ﬁber) bundle atlas. S) similarly as in the following way: On the disjoint union α {α} × Uα × S we consider the equivalence relation: (α.

so R is an obstruction against integrability of the horizontal subbundle. . then as in 8. by 8. p. p. Deﬁnition A connection on the ﬁber bundle (E.9.13. Then ker Φ is of constant rank. Then we call Φ the vertical projection and χ := idT E − Φ = C ◦ (T p. M. M. πE ) | HE : HE → T M ×M E is ﬁber linear over E and injective. Then ϕ : U × S → p−1 (U ). V E). so by reason of dimensions it is a ﬁber linear isomorphism: Its inverse is denoted by C := ((T p. χY ]. is 1 a diﬀeomorphism and is ﬁber respecting.u(x). V E) with values in the vertical bundle VE such that Φ ◦ Φ = Φ and ImΦ = V E. and since each ﬁber is t t t compact and ξx is complete.ξu . πE ) | HE)−1 : T M ×M E → HE → T E. deﬁned by ϕ(x. ηx has a global ﬂow too. Denote p−1 (x0 ) by S. the ﬁbers p−1 (x) are all diﬀeomorphic. so Φ is just a projection T E → V E. with the help of a partition of unity on p−1 (U ) we may construct vector ﬁelds ηx ∈ X(p−1 (U )) which depend smoothly on x ∈ U and are p-related to ξx : T p. Then (T p. p. M. 9. Since M is connected. S) be a ﬁber bundle.4. Let (E. y) = Flηx (y). Id − Φ] = [χ. Since the vertical bundle V E is integrable. πE ). Note the formula Φ(ξu ) = ξu − C(T p. πE )−1 (0p(u) . R] = 0. S). so each ξx is a complete vector t ﬁeld on U . Y ) = Φ[χX. Connections on general ﬁber bundles 37 such that u(Flξx u−1 (z)) = z + t. 1 We have R(X. So we can equally well describe a connection Φ by specifying C. ϕ−1 ) is a ﬁber chart. So C : T M ×M E → T E is ﬁber linear over E and is a right inverse for (T p.14 we have the Bianchi identity [Φ.ηx = ξx ◦ p. Now we consider the mapping (T p. Thus p ◦ Flηx = Flξx ◦p and Flηx is ﬁber respecting. χ] ∈ Ω2 (E.3. so ker Φ is a sub vector bundle of T E. u) = Vu E by deﬁnition. If Φ : T E → V E is a connection on a ﬁber bundle (E. Since p is a submersion. it is called the space of horizontal vectors or the horizontal bundle and it is denoted by HE Clearly T E = HE ⊕ V E and Tu E = Hu E ⊕ Vu E for u ∈ E. C is called the horizontal lift associated to the connection Φ. Curvature. Φ](X. so (U. The following is special case of 8. Y ) = 2 [Φ. so (T p. πE ) : T E → T M ×M E. S) is a vector valued 1-form Φ ∈ Ω1 (E. 9.13 the curvature R of Φ is given by 2R = [Φ. u) for ξu ∈ Tu E. Φ] = [Id − Φ. we consider the ﬁber linear mapping T p : T E → T M and its kernel ker T p =: V E which is called the vertical bundle of E. πE ) will be called the horizontal projection.

Local description of connections. Since p is a submersion.6. Let us ﬁx a ﬁber bundle atlas (Uα ) with transition functions (ψαβ ). S) is again a ﬁber bundle.4 the horizontal bundle is integrable and gives rise to the horizontal foliation. The forms f ∗ Φ and Φ are p∗ f -related in the sense of 8. Let (E.5.6.p) E exists. p. p. If (Uα . Pullback. The following diagram sets up some further notation for it: f ∗E − − → E −− p f∗ N −−→ M −− f p∗ f Proposition. M. (3) The curvatures of f ∗ Φ and Φ are also p∗ f -related: R(f ∗ Φ) = f ∗ R(Φ). ψα ) is a ﬁber bundle atlas of (E.Tu (p∗ f ). f and p are transversal and thus the pullback N ×(f. (2) If Φ ∈ Ω1 (E. (f ∗ p.7. given by (f ∗ Φ)u (X) := Tu (p∗ f )−1 . p.38 9. S) be a ﬁber bundle and consider a smooth mapping f : N → M . M. M. Uα × T S). This is seen by devising suitable horizontal foliations on the trivial bundle pr2 : R × S 1 → S 1 . S) in the sense of 9. T E) is a connection on the bundle E. S) has zero curvature. and p∗ f is a ﬁber wise diﬀeomorphism. is again a connection on the bundle f ∗ E. S).15. S). the vector valued form f ∗ Φ. Let Φ be a connection on (E.Φ. Connections on general ﬁber bundles 9. f ∗ p. Each point u ∈ E lies on a unique leaf L(u) such that Tv L(u) = Hv E for each v ∈ L(u). 9. M. N.15. but in general it is neither surjective nor is it a covering onto its image. The restriction p | L(u) is locally a diﬀeomorphism. y) + ηy for ξx ∈ Tx Uα and ηy ∈ Ty S.X for X ∈ Tu E. pr2 ◦ ψα ◦ p∗ f )) is visibly a ﬁber bundle atlas for (f ∗ E. (3) follows from (2) and 8. p. ηy ) =: −Γα (ξx . In the situation above we have: (1) (f ∗ E. Let us suppose that a connection Φ on the bundle (E. and let us consider the connection (ψα )−1 )∗ Φ ∈ Ω1 (Uα × S. . N. 9.M. Proof. which may be written in the form ((ψα )−1 )∗ Φ)(ξx .1. by the formal universal properties of a pullback. It will be called the pullback of the ﬁber bundle E by F and we will denote it by f ∗ E. Flat connections. (1). Then by 9. (2) is obvious. f ∗ p. then (f −1 (Uα ).

Proof. 0y )) = −1 = −Φ(T (ψα )(ξx .9. We consider Γα as an element of the space Ω1 (Uα . y) = Γα (ξx . ψα ). η 2 )] = −1 = (ψα )∗ Φ[(ξ 1 . (Id − (ψα )∗ Φ)(ξ 2 .(ξx .(0x . y)) − Tx (ψαβ ( . ξ 1 Γα (ξ 2 ) − ξ 2 Γα (ξ 1 ) + [Γα (ξ 1 ). 0y )) = −1 −1 = T (ψα ). 0y )) = −1 −1 = −Φ(T (ψα ).(ξx . ηy ) = (ξx . y) = (x. The transformation law for the Christoﬀel forms is Ty (ψαβ (x. η 2 )) = −1 −1 −1 = (ψα )∗ R [(Id − (ψα )∗ Φ)(ξ 1 . y)) := −T (ψα ).4 and 9. This implies the transformation law. y)). y))) − T (ψα )(0x . ψαβ (x.T (ψα ◦ ψβ ). )). From ψα ◦ (ψβ )−1 (x. ξ 2 ) + [Γα (ξ 1 ). ξ 2 ]) + ξ 1 Γα (ξ 2 ) − ξ 2 Γα (ξ 1 ) + [Γα (ξ 1 ). Lemma.(ξx . y)) we get that T (ψα ◦ (ψβ )−1 ).Γβ (ξx . y)) = −Φ(T (ψβ )(ξx .y) ψαβ (ξx .(ξx . X(S)). 0y )) − Φ(T (ψα )(0x . Γα (ξ 2 )]X(S) . y)). a 1-form on U α with values in the inﬁnite dimensional Lie algebra X(S) of all vector ﬁelds on the standard ﬁber.ξx ).T (ψα )−1 . Γα (ξx . Γβ (ξx . Tx (ψαβ ( . T(x. The Γα are given by (0x . (ξ 2 . Γα (ξ 2 ))] = −1 = (ψα )∗ Φ [ξ 1 .ξx . η 1 ). ξ 2 ]. The Γα are called the Christoﬀel forms of the connection Φ with respect to the bundle atlas (Uα . ηy )) and thus: −1 −1 T (ψβ ). 0y ))) = −1 −1 = −Φ(T (ψα )(ξx . Γα ]X(S) .y) (ψαβ ). The curvature R of Φ satisﬁes −1 (ψα )∗ R = dΓα + [Γα . ψαβ (x. Γα (ξ 2 )] = = dΓα (ξ 1 . Γα (ξ 2 )] = = −Γα ([ξ 1 .Φ.3 −1 (ψα )∗ R ((ξ 1 .y) (ψαβ )(ξx . T(x. For the curvature R of Φ we have by 9. T(x. 0y ). (ξ 2 . ψαβ (x.5. .(0x . Γα (ξx . The formula for the curvature is the Maurer-Cartan formula which in this general setting appears only in the level of local description. Γα (ξ 1 )). Connections on general ﬁber bundles 39 since it reproduces vertical vectors. η 1 ).

S) and the pullback connection c∗ Φ on it. (6) If X ∈ X(M ) is a vector ﬁeld on the base then the parallel transport along the ﬂow lines of X is given by the ﬂow of the horizontal lift CX of X: Pt(FlX (x). and Pt(c. γ(y. t) + d dt γ(y. Pt Second proof. ψα ) as in 9. Parallel transport. where −1 0 = (ψα )∗ Φ d d dt c(t). t) = p∗ c((c∗ p | L(ux ))−1 (t)). Then there is a neighborhood U of Ex × {0} in Ex × (a. with initial condition Pt(c. (a. so γ(y. ux . t)) = (c(t). f (t). ux . b).7. All further properties are consequences of uniqueness. ux . It has zero curvature. p. ux . t)). f (0))) if deﬁned. c∗ p. M ) ×(ev0 . ux )) = 0c(t) if deﬁned. We ﬁrst give three diﬀerent proofs of assertions (1) – (4). → where U is its domain of deﬁnition. dt γ(y. Pt(c. 0) = ux . nonlinear. t. b) is smooth with 0 ∈ (a . S) and let c : (a.40 9. γ(y. So there is a maximally deﬁned local solution curve which is unique. By 9. Connections on general ﬁber bundles 9.M. t) is the integral curve (evolution line) through y ∈ S of the d time dependent vector ﬁeld Γα dt c(t) on S. ux ) = FlCX (ux ). d (2) Φ( dt Pt(c. ux . Then −1 ψα (Pt(c. t. Consider the pullback bundle (c∗ E. t. b ) → (a. b) → M be a smooth curve with 0 ∈ (a.p) E × R ⊃ U − E. t). (3) Reparametrisation invariance: If f : (a . t) = −Γα d dt c(t). b). since the horizontal bundle is 1-dimensional. ψα (x.8. b) and a smooth mapping Ptc : U → E such that: (1) p(Pt(c. in detail: Pt(c. 0) = ux . M. Consider a ﬁber bundle atlas (Uα . (5) The parallel transport is smooth as a mapping C ∞ (R.6 the horizontal foliation exists and the parallel transport just follows a leaf and we may map it back to E. d First proof. (4) U is maximal for properties (1) and (2). Let Φ be a connection on a bundle (E. t Proof. then Pt(c. t)) = 0 is an ordinary diﬀerential equation of ﬁrst order. ux ) = Pt(c ◦ f. This vector ﬁeld visibly . In local bundle coordinates Φ( dt Pt(c. ux )) = c(t) if deﬁned. y). c(0) = x. b ). t. Third proof. Theorem.

iCX Φ] + 2i[Φ. M ) ×(ev0 . M.16. Φ]. [CX. 9. So T p.11. Lemma.8 shows that on a ﬁber bundle with compact standard ﬁber any connection is complete. M ) from 6. b) → M is deﬁned on the whole of Ec(0) × (a. Φ] ∈ Ω1 (E.10. T M ). 9. which we then set equal. T E) is p-related to 0 ∈ Ω1 (M . Φ] The vector ﬁeld CX is p-related to X and Φ ∈ Ω1 (E. T M ). The following is a suﬃcient condition for a connection Φ to be complete: There exists a ﬁber bundle atlas (Uα . From t = [iCX Φ. Φ]. By the theory of ordinary diﬀerential equations the solution is smooth in this parameters also.M. S) with curvature R and horizontal lift C.p) E × R to a smooth curve in E. Φ] (FlCX )∗ Φ = [CX. Φ] = − 1 iCX R.[CX. 0] ∈ Ω1 (M . Let X ∈ X(M ) be a vector ﬁeld on the base.15. (5). p. Φ] = 0.2 and see that the ordinary diﬀerential equation considered in the ﬁrst proof then just depends on some parameters more. From 8. b). By the considerations of section 6 it suﬃces to show that Pt maps a smooth curve (c1 .(5) we get LCX Φ = 8. X(S)) takes values in the linear subspace of gα -bounded vector ﬁelds on S .CX] Φ = −2Φ[CX. T E) is also p-related to 0 = [X. M.9. c3 ) : R → U ⊂ C ∞ (R. c2 . The third proof of theorem 9. and [CX. (6) is obvious from the deﬁnition. Φ] − [Φ.(7) the form [CX. S) is called a complete connection. A connection Φ on (E. p. Φ] = Φ[CX. if the parallel transport Ptc along any smooth curve c : (a. We use the description of the smooth curves in C ∞ (R. t 2 ∂ ∂t 0 Proof.9. Let Φ be a connection on a bundle (E. Connections on general ﬁber bundles 41 depends smoothly on c. so by 8. ψα ) and complete Riemannian metrics gα on the standard ﬁber S such that each Christoﬀel form Γα ∈ Ω1 (Uα . Then for the horizontal lift CX ∈ X(E) we have LCX Φ = ∂ ∂t 0 (FlCX )∗ Φ = [CX.(2) we have iCX R = iCX [Φ. Clearly local solutions exist and all properties follow. Φ] has vertical values.

42

9. Connections on general ﬁber bundles

For in the third proof of theorem 9.8 above the time dependent vector d ﬁeld Γα ( dt c(t)) on S is gα -bounded for compact time intervals. So by continuation the solution exists over c−1 (Uα ), and thus globally. A complete connection is called an Ehresmann connection in [Greub - Halperin - Vanstone I, p 314], where it is also indicated how to prove the following result. Theorem. Each ﬁber bundle admits complete connections. Proof. Let dim M = m. Let (Uα , ψα ) be a ﬁber bundle atlas as in 9.1. By topological dimension theory [Nagata, 1965] the open cover (Uα ) of M admits a reﬁnement such that any m + 2 members have empty intersection. Let Uα ) itself have this property. Choose a smooth partition of unity (fα ) subordinated to (Uα ). Then the sets Vα := { x : 1 fα (x) > m+2 } ⊂ Uα form still an open cover of M since fα (x) = 1 and at most m + 1 of the fα (x) can be nonzero. By renaming assume that each Vα is connected. Then we choose an open cover (Wα ) of M such that Wα ⊂ Vα . Now let g1 and g2 be complete Riemannian metrics on M and S, respectively (see [Nomizu - Ozeki, 1961] or [Morrow, 1970]). For not connected Riemannian manifolds complete means that each connected component is complete. Then g1 |Uα × g2 is a Riemannian metric on ∗ Uα × S and we consider the metric g := fα ψα (g1 |Uα × g2 ) on E. Obviously p : E → M is a Riemannian submersion for the metrics g and g1 . We choose now the connection Φ : T E → V E as the orthonormal projection with respect to the Riemannian metric g. Claim: Φ is a complete connection on E. Let c : [0, 1] → M be a smooth curve. We choose a partition 0 = t0 < t1 < · · · < tk = 1 such that c([ti , ti+1 ]) ⊂ Vαi for suitable αi . It suﬃces to show that Pt(c(ti + ), t, uc(ti ) ) exists for all 0 ≤ t ≤ ti+1 − ti and all uc(ti ) ∈ Ec(ti ) , for all i — then we may piece them together. So we may assume that c : [0, 1] → Vα for some α. Let us now assume that for some (x, y) ∈ Vα × S the parallel transport Pt(c, t, ψα (x, y)) is deﬁned only for t ∈ [0, t ) for some 0 < t < 1. By the third proof of 9.8 we have Pt(c, t, ψα (x, y)) = ψα (c(t), γ(t)), where γ : [0, t ) → S is the maximally deﬁned integral curve through y ∈ S of the time dependent d −1 vector ﬁeld Γα ( dt c(t), ) on S. We put gα := (ψα )∗ g, then (gα )(x,y) = ∗ (g1 )x ×( β fβ (x)ψβα (x, ) g2 )y . Since pr1 : (Vα ×S, gα ) → (Vα , g1 |Vα ) −1 is a Riemannian submersion and since the connection (ψα )∗ Φ is also given by orthonormal projection onto the vertical bundle, we get

t

∞ > g1 -lengtht (c) = gα -length(c, γ) = 0

0

d |(c (t), dt γ(t))|gα dt =

**9. Connections on general ﬁber bundles
**

t

43

=

0

|c (t)|21 + g

t

d d ∗ β fβ (c(t))(ψαβ (c(t), −) g2 )( dt γ(t), dt γ(t)) dt d fα (c(t)) | dt γ(t)|g2 dt ≥

≥

≥

0

√

t

m+2

0

d | dt γ(t)|g2 dt.

So g2 -lenght(γ) is ﬁnite and since the Riemannian metric g2 on S is complete, limt→t γ(t) =: γ(t ) exists in S and the integral curve γ can be continued.

44

**10. Principal Fiber Bundles and G-Bundles
**

10.1. Deﬁnition. Let G be a Lie group and let (E, p, M, S) be a ﬁber bundle as in 9.1. A G-bundle structure on the ﬁber bundle consists of the following data: (1) A left action : G × S → S of the Lie group on the standard ﬁber. (2) A ﬁber bundle atlas (Uα , ψα ) whose transition functions (ψαβ ) act on S via the G-action: There is a family of smooth mappings (ϕαβ : Uαβ → G) which satisﬁes the cocycle condition ϕαβ (x)ϕβγ (x) = ϕαγ (x) for x ∈ Uαβγ and ϕαα (x) = e, the unit in the group, such that ψαβ (x, s) = (ϕαβ (x), s) = ϕαβ (x).s. A ﬁber bundle with a G-bundle structure is called a G-bundle. A ﬁber bundle atlas as in (2) is called a G-atlas and the family (ϕαβ ) is also called a cocycle of transition functions, but now for the G-bundle. To be more precise, two G-atlases are said to be equivalent (to describe the same G-bundle), if their union is also a G-atlas. This translates as follows to the two cocycles of transition functions, where we assume that the two coverings of M are the same (by passing to the common reﬁnement, if necessary): (ϕαβ ) and (ϕαβ ) are called cohomologous if there is a family (τα : Uα → G) such that ϕαβ (x) = τα (x)−1 .ϕαβ (x).τβ (x) holds for all x ∈ Uαβ . In (2) one should specify only an equivalence class of G-bundle structures or only a cohomology class of cocycles of G-valued transition functions. From any open cover (Uα ) of M , some cocycle of transition functions (ϕαβ : Uαβ → G) for it, and a left G-action on a manifold S, we may construct a G-bundle by gluing, which depends only on the cohomology class of the cocycle. By some abuse of notation we write (E, p, M, S, G) for a ﬁber bundle with speciﬁed G-bundle structure. Examples: The tangent bundle of a manifold M is a ﬁber bundle with structure group GL(n). More general a vector bundle (E, p, M, V ) is a ﬁber bundle with standard ﬁber the vector space V and with GL(V )structure. 10.2. Deﬁnition. A principal (ﬁber) bundle (P, p, M, G) is a G-bundle with typical ﬁber a Lie group G, where the left action of G on G is just the left translation. So by 10.1 we are given a bundle atlas (Uα , ϕα : P |Uα → Uα ×G) such that we have ϕα ϕ−1 (x, a) = (x, ϕαβ (x).a) for the cocycle of transition β functions (ϕαβ : Uαβ → G). This is now called a principal bundle atlas. Clearly the principal bundle is uniquely speciﬁed by the cohomology class of its cocycle of transition functions.

b.3 we write r(u.a.3 can serve as an equivalent deﬁnition for a principal bundle. Lemma. whose local expression is τ (ϕ−1 (x.g when the meaning is clear. which is obviously injective with inα vertible tangent mapping. p. As in 1.g ) = g −1 . From the Lemma itself it follows. called the principal right action. 10. ux .τ (ux . g) = u. this is well deﬁned. sβ (x). Proof. Lemma 10. that a principal bundle atlas of a principal ﬁber bundle (P. a) = sα (x). vx )) = vx . 10. whose inverse is denoted by τux : Px → G. Let G be a Lie group with Lie algebra g. Let p : P → M be a surjective submersion (a ﬁbred manifold).4. ϕ−1 (x. τ is smooth by the implicit function theorem and clearly τ (ux . a). Principal Fiber Bundles and G-Bundles 45 Each principal bundle admits a unique right action r : P × G → P . ux ). 10. so its inverse ϕα : P |Uα → Uα × G is a ﬁber respecting diﬀeomorphism. Then (P. Let ϕ−1 : Uα × G → P |Uα α be given by ϕ−1 (x. p. ux )) = ux . Let sα : Uα → P be local sections (right inverses) for p : P → M such that (Uα ) is an open cover of M . thus we also have τ (ux . ux ) = e. In the proof of Lemma 10. τ (ux . α Since left and right translation on G commute. The principal right action is visibly free and for any ux ∈ Px the partial mapping rux = r(ux .g.g) = (x. So (Uα . a). These inverses together give a smooth mapping τ : P ×M P → G.g. Remarks. Homogeneous spaces. M. Let K be a closed subgroup of G. G) is a principal ﬁber bundle. ϕα ) is a principal bundle atlas.g) and (Uα . given by ϕα (r(ϕ−1 (x. where r is the right G-action. This mapping is also uniquely deterα α mined by the implicit equation r(ux . that the pullback f ∗ P over a smooth mapping f : M → M is again a principal ﬁber bundle. g)) = (x. so in topology or in inﬁnite dimensional diﬀerential geometry one should stick to our original deﬁnition.3.10.g)) = β (x. τ (ux . ) : G → Px is a diﬀeomorphism onto the ﬁber through ux .g) = τ (ux . Let τ : P ×M P → G be given by the implicit equation r(ux . ϕα ) is already a ﬁber bundle atlas.5. G) is already determined if we specify a family of smooth sections of P . ux ). Let the action be a right one by using the group inversion if necessary. and let G be a Lie group which acts freely on P such that the orbits of the action are exactly the ﬁbers p−1 (x) of p. τ (sα (x).3 we have seen. sβ (x)). b)) = a−1 . ux . then by a well known theorem K is a closed Lie subgroup whose Lie algebra will be denoted by K. Thus we have ϕα ϕ−1 (x. But this is true only if an implicit function theorem is available.g and τ (ux . ag). M. τ (sα (x). There is a . g) = ϕα (sβ (x). whose domains of deﬁnition cover the base M .

p. O(n − k)) But O(k) also acts from the right on V (k. n. ek ) is 1×O(n−k). . Sp(k) × Sp(n − k)) are principal ﬁber bundles. . Proof. . p. n. . SO(n − 1)) is a principal ﬁber bundle. R) is the space of all k-planes containing 0 in Rn . By lemma 10. K) is a principal ﬁber bundle. p. R) → G(k. n. n.3 we get a principal ﬁber . p.46 10. p. G(k. p. . G/K. n. Principal Fiber Bundles and G-Bundles unique structure of a smooth manifold on the quotient space G/K such that the projection p : G → G/K is a submersion. and (Sp(n). . The Stiefel manifold V (k. The group O(n) acts transitively on it and the isotropy group of the k-plane Rk × {0} is the subgroup O(k) 0 0 O(n − k) . SU (n − 1)). U (k) × U (n − k)). n. R) is the space of all orthonormal kframes in Rn . so by the implicit function theorem p admits local sections. Likewise ˜ (SO(n). S 2n−1 . Theorem. n. p. S 4n−1 . The group SO(n) acts transitively on S n−1 ⊂ Rn . The isotropy group of the ”north pole” (0. p. and (Sp(n). p. 1) is the subgroup 1 0 0 SO(n − 1) which we identify with SO(n − 1). The Grassmann manifold G(k. R) and the isotropy subgroup of (e1 . (U (n). R) = O(n)/O(k) × O(n − k) is a compact manifold and we get the principal ﬁber bundle (O(n). R). n. U (n − 1)). C). . G(k. n. (SU (n). SO(k) × SO(n − k)). its orbits are exactly the ﬁbers of the projection p : V (k. whose orbits are exactly the ﬁbers of p. R). V (k. So S n−1 = SO(n)/SO(n − 1) and (SO(n). G(k. Likewise (O(n). O(n − 1)). . The group multiplication of G restricts to a free right action µ : G × K → G. so V (k. p. R). (G. p. S 2n−1 . For the convenience of the reader we discuss now the best know homogeneous spaces. 0. n. H). n. So by lemma 10. G(k. n.3 the result follows. R). R) = O(n)/O(n − k) is a compact manifold and we get a principal ﬁber bundle (O(n). (U (n). R). S n−1 . S n−1 . Clearly the group O(n) acts transitively on V (k. Sp(n − 1)) are principal ﬁber bundles. therefore G(k. O(k) × O(n − k)). n.

p . p .Φ(g) holds in general. n.10. There are furthermore complex and quaternionic versions of the Stiefel manifolds. M. but it is no longer a pullback diagram in general. More examples will be given in sections on jets below.6. p. By the universal property of the pullback any general homomorphism χ of principal ﬁber bundles over a group homomorphism can be written as the composition of a reduction of structure groups and a pullback . 10. M . Principal Fiber Bundles and G-Bundles 47 bundle (V (k. R). M . G) be a principal ﬁber bundle homomorphism. M. if χ : P → P is smooth and χ(u. it is called a reduction of the structure group G to G for the principal bundle (P .g) = χ(u). n.A = 1k }. G) → (P . Indeed we have the following diagram where all arrows are projections of principal ﬁber bundles. R). Homomorphisms. M . Then obviously the diagram P −−→ P −− p p χ (a) M −−→ M −− χ ¯ commutes for a uniquely determined smooth mapping χ : M → M . G ) is called a homomorphism over Φ of principal bundles. G(k. n) is also diﬀeomorphic to the space { A ∈ L(Rk . i. n. Rn ) : At . G) → (P . the space of all linear isometries Rk → Rn . G ) — the name comes from the case. O(k)). For ¯ each x ∈ M the mapping χx := χ|Px : Px → Pχ(x) is G-equivariant and ¯ therefore a diﬀeomorphism. n. R). and where the respective structure groups are written on the arrows: O(n) O(k) − − − V (k. i. when Φ is the embedding of a subgroup. Let Φ : G → G be a homomorphism of Lie groups. Then χ is ﬁber respecting. a smooth G-equivariant mapping χ : P → P . χ : (P.e. But the most general notion of a homomorphism of principal bundles is the following. n. p . Let χ : (P.e. so diagram (a) makes again sense. so diagram (a) is a pullback diagram. R) − −→ O(k) O(n−k) O(n−k) (a) V (n − k. p. p. If χ covers the identity on the base. R) − − − G(k. − −→ It is easy to see that V (k.

In this situation we have: (1) The space P ×G S of orbits of the action R carries a unique smooth manifold structure such that the quotient map q : P ×S → P ×G S is a submersion. ] or simply P [S] and we will write p for p if no confusion is possible. M.s). G ) − − → (P . G) is a G-bundle.7. We also deﬁne the smooth ¯ −1 mapping τ : P ×M P [S. G) − − → (χ∗ P . and τ (ux . p. G ) −− ¯ −− p p (b) M M −−→ −− χ ¯ M 10. We consider the right action R : (P × S) × G → P × S. M. (2) (P ×G S. M. (3) (P × S. ¯ Notation: (P ×G S. In the setting of the diagram in (2) the mapping p◦pr1 is constant on the R-orbits. Theorem. vx )) = vx . s).2. S. p. P ×G S. so that P [G] = P .τ (ux . Principal Fiber Bundles and G-Bundles homomorphism as follows. where p : P ×G S → M is ¯ ¯ given by P × S − − → P ×G S −− pr1 p ¯ P −−→ −− p q (a) M.g. where we also indicate the structure groups: (P. vx ) := qux (vx ). where S = G and the action is left translation. p.g. vx ). g −1 . Let (P. (4) If (Uα . vx ) = g −1 . then together with the left action : G × S → S this cocycle is also one for the G-bundle (P ×G S. M. this mapping coincides with τ considered in 10. It satisﬁes τ (u. q. Proof.48 10. p. Let (Uα . G). q(ux . so p exists as a mapping. In this diagram qu : {u} × S → (P ×G S)p(u) is a diﬀeomorphism for each u ∈ P . In the special situation. given by R((u. We will also denote it by P [S. G) is called the associated bundle for the ¯ action : G×S → S. τ (ux . q(u. S. s)) = s. ] → S by τ (ux . G) is a principal ﬁber bundle with principal action R. g) = (u. ϕα : P |Uα → Uα × G) is a principal bundle atlas with cocycle of transition functions (ϕαβ : Uαβ → G). S. G) be a principal bundle and let : G × S → S be a left action of the structure group G on a manifold S. Associated bundles. ϕα : P |Uα → Uα ×G) ¯ .

since the principal right action is free.ϕαβ (x)−1 . So (1).g). since its lines are diﬀeomorphisms we conclude that qu : {u}× S → p−1 (p(u)) is a diﬀeomorphism. α −1 −1 so λα λ−1 (ψα (x.s). P ×G S. e).s). ϕαβ (x).ϕβα (x). ϕαβ (x). and p is smooth by the ¯ universal properties of a submersion. g) = λ−1 (ψβ (x. and (4) are checked. s). s) = q(ϕ−1 (x.g). ¯ (3) We rewrite the last diagram in the following form: p−1 (Uα ) × S (c) q −1 (Vα ) − − → Vα × G −α − q pr1 p−1 (Uα ) ¯ −1 λ Vα Here Vα := p (Uα ) ⊂ P ×G S and the diﬀeomorphism λα is deﬁned ¯ −1 by λ−1 (ψα (x. s). ϕαβ (x). s) = α α −1 = q(ϕ−1 (x.ϕαβ (x). ϕαβ (x). For each orbit in p−1 (x) ⊂ ¯ P ×G S there is exactly one s ∈ S such that this orbit passes through −1 (ϕ−1 (x.s) = (ϕ−1 (x. s). g) = (ψα (x. Principal Fiber Bundles and G-Bundles 49 be a principal bundle atlas with transition functions (ϕαβ : Uαβ → −1 −1 G). s) = ¯ −1 q(ϕα (x. The deﬁning equation for ψα shows that q is smooth and a submersion and consequently the smooth structure on P ×G S is uniquely deﬁned. g). s) = q(ϕ−1 (x. g) = β β = (ϕ−1 (x. ψα ) is a G-atlas for P ×G S and makes it into a smooth manifold and a G-bundle.s). g) := (ϕα (x. ) : α −1 S → p (x) is bijective. α −1 so ψα ψβ (x. e). (2).s) So (Uα .e). g −1 . ϕαβ (x). s) = β = q(ϕ−1 (x. which is ﬁber respecting. ϕαβ (x). G) is a principal bundle with structure group G and the same cocycle (ϕαβ ) we started with. . e). ϕαβ (x). g −1 . e). By the deﬁnition of ψα the diagram (b) p−1 (Uα ) × S − − → Uα × G × S −− q Id× ϕα ×Id p−1 (Uα ) ¯ Uα × S commutes. s). g). We deﬁne ψα : Uα × S → p−1 (Uα ) ⊂ P ×G S by ψα (x.s) = α β −1 = (λ−1 (ψα (x. s). Thus ψα (x. Furthermore ¯ −1 ψβ (x. Then we have α −1 −1 λ−1 (ψα (x. . s) = (x.g) and we conclude that β (P × S. ϕβα (x). q.s) = ψα (x.10. s). e). s). g −1 .

1. thus IdP × i induces IdP ×G i : M = P [{y}] → P [S ]. Let S be a point. M. M . M. G) be a G-bundle.s = s for all s ∈ S. He is given a bundle E.50 10. 3. Let χ : (P. and a homomorphism of G-bundles in the sense of the deﬁnition 10. If the action of G on S is trivial. Let χ : (P. G) be a principal ﬁber bundle homomorphism as in 10. which is a global section of the associated bundle P [S ]. ] which again is a homomorphism of G-bundles. G) be a principal ﬁber bundle homomorphism and let f : S → S be an G-equivariant mapping. ] and P [S . p. G) such that E = P [S. This is the usual way a diﬀerential geometer thinks of an associated bundle. ]. Let (E. 4. G) → (P . Furthermore we consider a smooth left action : G × S → S. so g. Let (P. 10.6. so there is an induced mapping P ×S −−→ −− q Id×G f Id×f (a) P ×S q P ×G S − − → P ×G S . Now we consider the situation of 1 and 2 at the same time. Principal Fiber Bundles and G-Bundles 10. 2. p. Equivariant mappings and associated bundles.f (s) or f ◦ g = g ◦ f . and again a homomorphism of Gbundles in the sense of deﬁnition 10. For a trivial principal ﬁber bundle any associated bundle is trivial. M. p.8.s) = g. We have two associated bundles P [S. which is ﬁber respecting over M . G) → (P . Then χ × IdS : P × S → P × S is G-equivariant and induces a mapping χ×G IdS : P ×G S → P ×G S. p. Then from the cocycle of transition functions we may glue a unique principal bundle (P.10 below. M . : G × S → S and : G × S → S . ]. and the G-bundle structure then is described with the help of the mappings τ and q. then the inclusion i : {y} → S is G-equivariant. ﬁber wise a diﬀeomorphism. −− which is ﬁber respecting over M . S. Corollary. Then IdP × f : P × S → P × S is equivariant for the actions R : (P × S) × G → P × S and R : (P × S ) × G → P × S . then the associated bundle is trivial: P [S] = M × S. Furthermore let y ∈ S be a ﬁxed point of the action : G × S → S . speciﬁed by a cocycle of transition functions (ϕαβ ) with values in G and a left action of G on S. ] → P [S . M. Then χ × f : P × S → P × S is clearly G-equivariant and therefore induces a mapping χ×G f : P [S. p . Then P [S] = P ×G S = M . so f (g.10 below. . a principal bundle P . G) be a principal ﬁber bundle and consider two left actions of G. p. Let furthermore f : S → S be a G-equivariant smooth mapping.9. p . M.

10. The G-equivariant mapping f : S → S can ¯ ¯ ¯ be read oﬀ the following diagram P ×M P [S] χ× γ M (a) −−→ S −− f τ τ P ×M P [S ] − − → S . and we have γ = χ. ψα ) be a G-atlas for P [S . ].5 has the cocycle of transition ¯ functions (ϕαβ = ϕαβ ◦ γ ). M. such that diagram (a) commutes. Let (Uα . G). ] → P [S . p . p. M . ρ] is an associated ﬁber bundle with structure group G. if the following conditions are satisﬁed: P is isomorphic to the pullback γ ∗ P . ϕ α) of (P . So by section 6 P [V. it induces the G-atlas (Uα . and consider a representation ρ : G → GL(V ) of G on a ﬁnite dimensional vector space V . for in the canonically associated ﬁber bundle atlas the transition functions have also values in GL(V ). p. belonging to the principal ﬁber bundle atlas (Uα . ) : S → S should γ be G-equivariant for all α and all x ∈ Uα . G) → (P . Lemma. G) of principal ﬁber bundles is already determined by the requirement that P = γ ∗ P . Deﬁnition. ] be a homomorphism of associated bundles and G-bundles. −− which by the assumptions is seen to be well deﬁned in the right column. Let γ : P [S. Proof. M . In the situation of 10. Then the pullback-related principal ﬁber bundle atlas (Uα = γ −1 (Uα ). f : S → S (G-equivariant) .11. ρ] is a vector bundle. Let (P. ] → P [S . Associated vector bundles. 10. but also with structure group GL(V ). p . G) → (P . s) = (¯ (x). Let us describe this in more detail now. The homomorphism χ : (P. M.10. Then there is a principal bundle homomorphism χ : (P. p . G) be a principal ﬁber bundle. . Principal Fiber Bundles and G-Bundles 51 10. p. So a homomorphism of associated bundles is described by the whole triple (χ : P → P . ] covering a smooth mapping γ : M → M ¯ of the bases is called a homomorphism of G-bundles. and the local ¯ representations of γ in pullback-related ﬁber bundle atlases belonging to the two G-bundles are ﬁber wise G-equivariant. ] with cocycle of transition functions (ϕαβ ).8. γα ) for ¯ P = γ ∗ P as described in the proof of 9. s)) and γα (x. a smooth ﬁber respecting mapping γ : P [S. M . G) and a G-equivariant mapping f : S → S such that γ = χ ×G f : P [S. Then P [V. ¯ −1 Then (ψα ◦γ◦ψα )(x. ]. γα (x. M. γ : E → E ). ψα ) for P [S. ] → P [S .

. We call it an orthonormal frame ﬁeld. We write sα = sβ . GL(n)).v) and consequently ψα ψβ (x. sα ). Now let (Uα ) be an open cover of M with orthonormal frame ﬁelds sα = (sα . p. (E. . Now we may apply the Gram-Schmidt orthonormalization procedure to the basis (s1 (x). g)). ψα ). sn ) of E over U which is orthonormal with respect to g. Principal Fiber Bundles and G-Bundles If (E. Rn ) equipped with a Riemannian metric g. p. gαβ (x). so by lemma 10. . v 1 . M.gβα i −1 for short. E) by reduction of the . the matrix gαβ (x) = (gαβ j (x)) is an element of O(n. . it is usually called the orthonormal frame bundle of E. The corresponding principal ﬁber bundle will be denoted by O(Rn . p. . We consider the vecn 1 tor bundle charts (Uα .v i =: sα (x). The associated bundle GL(Rn . The principal bundle GL(Rn . . . . which is a ﬁber bundle over the base M . E)|U ) be a local frame ﬁeld of the bundle E over U ⊂ M . M ) is a vector bundle with n-dimensional ﬁbers we may consider the open subset GL(Rn . .gβα i (x) for C ∞ -functions gαβ : Uαβ → R. E)[Rn ] for the banal representation of GL(n) on Rn is isomorphic to the vector bundle (E. . gαβ (x). E) whose orbits are exactly the ﬁbers. M ) we started with. . v n ) = i β j α we have si (x) = sj (x). . ψα : E|Uα → Uα × Rn ) given the orthonormal −1 sα (x). Composition from the right by elements of GL(n) gives a free right action on GL(Rn . E) × Rn → E is invariant under the right action R of GL(n).52 10. . To illustrate the notion of reduction of structure group. E) is called the linear frame bundle of E. for the evaluation mapping ev : GL(Rn . and locally in the image there are smooth sections to it. we consider now a vector bundle (E. Any vector bundle admits Riemannian metrics: local existence is clear and we may glue with the help of a partition of unity on M . R)-structure on E.v = −1 −1 ψα (x. It is derived from the linear frame bundle GL(Rn . . Note that local sections of GL(Rn ) are exactly the local frame ﬁelds of the vector bundle E. Since sα (x) and sβ (x) are both orthonormal bases of Ex . . R). Now let s = (s1 . . .v.v = sα (x). E)[Rn ] = GL(Rn . so it factors to a ﬁber linear diﬀeomorphism GL(Rn . sn (x)) of Ex for each x ∈ U . v) = sβ (x). consisting of all invertible linear mappings. we obtain a frame ﬁeld s = (s1 . v) = (x.3 we have a principal ﬁber bundle (GL(Rn . E). Since this procedure is real analytic. M. . that is a section g ∈ C ∞ (S 2 E ∗ ) such that gx is a positive definite inner product on Ex for each x ∈ M . E).v). E) ×GL(n) Rn → E. For x ∈ Uαβ frame ﬁelds: ψα (x. sn ) ∈ C ∞ (GL(Rn . . since the positive deﬁnite sections form an open convex subset. . p.gαβ (x). So we have constructed an O(n. R)) are the cocycle of transition functions for the vector bundle atlas (Uα . where sα is deﬁned on Uα . Then we have ψβ (x. So the (gαβ : Uαβ → O(n. E) ⊂ L(M × Rn .

S)G we construct sf ∈ C ∞ (P [S]) in the following way: graph(f ) = (Id. sf (p(u))) = τ (u.g) = g −1 . p. f ) : P → P × S is G-equivariant. So it induces a smooth section sf ∈ C ∞ (P [S]) as seen from P −−→ P ×S −− p q M − − → P [S] −− fs (Id. fs (u)) = q(u. so by lemma 10. p. G/K)G . M.3 we get a principal ﬁber bundle (Ps . Proof. This is G-equivariant since fs (ux .4. If f ∈ C ∞ (P.12 the section s ∈ C ∞ (P [G/K]) corresponds to fs ∈ C ∞ (P.g) = (u. The sections of the associated bundle P [S. M.f (u) holds for g ∈ G and u ∈ P .f (u)) = ((Id. The embedding Ps → P is then a reduction of structure groups as required. f )(u. Let (P.g. f (u))) = f (u) and ss(f ) (p(u)) = q(u. Theorem.13. Sections of associated bundles. f (u.g. S)G by fs := τ ◦ (IdP ×M f ) : P = P ×M M → P ×m P [S] → S. K).g. we have a bijection C ∞ (P.τ (ux . Principal Fiber Bundles and G-Bundles 53 structure group from GL(n) to O(n). Thus Ps := fs (¯) is a submanifold e of P which is stable under the right action of K on P .g. The phenomenon discussed here plays a prominent role in the theory of classifying spaces. G) and a closed subgroup K of G. 10. g −1 . 10. .12. λ]. f (p(u))) = f (p(u)). q(u. f (x)) = g −1 . S)G ∼ C ∞ (P [S]).g)) = (u. M.10. since (Id. Theorem. ] correspond exactly to the G-equivariant mappings P → S. G) be a principal ﬁber bundle and : G × S → S a left action.fs (ux ) by 10. f )(u)). where λ : G × G/K → G/K is the left action on the homogeneous space. Consider a principal ﬁber bundle (P. = Proof. p. The two constructions are inverse to each other since we have fs(f ) (u) = τ (u.f ) (a) If conversely s ∈ C ∞ (P [S]) we deﬁne fs ∈ C ∞ (P. Furthermore the K-orbits are exactly the ﬁbers of the mapping p : Ps → M .g) = τ (ux . τ (u. S)G denote the space of all smooth mappings f : P → S which are G-equivariant in the sense that f (u. Let C ∞ (P. which is a surjective submersion since the action −1 ¯ λ : G × G/K → G/K is transitive. Then the reductions of structure group from G to K correspond bijectively to the global sections of the associated ¯ ¯ bundle P [G/K. f (x)) = g −1 . By theorem 10.

k for k ∈ K this restriction induces f : P → G/K by P ×M P −−→ −− τ G p P = P ×M P /K − − → G/K. G) is a principal ﬁber bundle we denote by Aut(P ) the group of all G-equivariant diﬀeomorphisms χ : P → P .g) = τ (u.k) = τ (ux . We use again the mapping τ : P ×M P → G from 10. conj]). Pp(u) ) ⊆ K } = P . Then p ◦ χ = χ ◦ p for a unique diﬀeomorphism χ of M . so the e two constructions are inverse to each other.g. vx ) = g −1 .g = conjg−1 fχ (u). (G. If (P. with Lie algebras g and K.15 The tangent bundles of homogeneous spaces. χ). −− f and from τ (ux . vx .2.14. p. Theorem. we deﬁne χf : P → P by χf (u) := u. so we may view P as a sub ﬁber bundle of P which is stable under the right action of K.54 10.K (k)X = AdG (k)X = AdK (k)X ∈ K. vx ). Principal Fiber Bundles and G-Bundles If conversely we have a principal ﬁber bundle (P . The group Gau(P ) of gauge transformations is equal to the space C ∞ (P. χ(u. (G. so K is an invariant . For X ∈ K and k ∈ K we have AdG.1 and put AdG. conj))G ∼ C ∞ (P [G. The bundle of gauges. Let G be a Lie group and K a closed subgroup. p . vx ) it follows that f is G-equivariant as required. The kernel of this homomorphism is called Gau(P ). We recall the mapping AdG : G → AutLie (g) from 1. M.τ (ux . conj))G is given.g)) = g −1 . If conversely f ∈ C ∞ (P. 10.g. For χ ∈ Gau(P ) we deﬁne fχ ∈ C ∞ (P. 10.f (u). M.g.g) = χ(u). Now we consider the mapping τ : P ×M P → G from 10. χ(u)). (G.K := AdG |K : K → AutLie (g). Then fχ (u. so ¯ ¯ there is a group homomorphism from Aut(P ) into the group Diﬀ(M ) of all diﬀeomorphisms of M .τ (u. the group of gauge transformations. = Proof. conj))G by fχ := τ ◦ (Id. then χ is an embedding covering the identity of M and is K-equivariant. Finally f −1 (¯) = {u ∈ P : τ (u.2 and restrict it to P ×M P . K) and a reduction of structure groups χ : P → P . It is easy to check that χf is indeed in Gau(P ) and that the two constructions are inverse to each other. Since we have τ (ux . So Gau(P ) is the space of all χ : P → P which satisfy p ◦ χ = p and χ(u. respectively. so fχ is indeed G-equivariant.

Proof.K ]. AdG.10. Te (G/K)).K ] → G[g/K. Ad⊥ ] ⊕ G[K. Principal Fiber Bundles and G-Bundles 55 subspace for the representation AdG. AdG. For p : G → G/K we consider the tangent mapping Te p : g → Te (G/K) which is linear and surjective and induces a linear isomorphism ¯ Te p : g/K → Te (G/K). AdG. Let ϕ : G × g → G × g be given . Ad⊥ ] and G[K. AdK ] → G[G. Ad⊥ ].K ] is a trivial vector bundle. In these circumstances we have T (G/K) = G[g/K.K ] ∼ G[g. G/K. G/K. which is isomorphic to the vec¯ ¯ tor bundle G[g/K. Then (a) 0 → K → g → g/K → 0 is short exact and K-equivariant. since the representation spaces are isomorphic. AdG. p. Te λ] = ¯ ¯ (G ×K Te (G/K). Now we consider the principal ﬁber bundle (G.K ] is a trivial vector bundle. The short exact sequence (a) induces a sequence of vector bundles over G/K: G/K × 0 → G[K. For k ∈ K we have p ◦ conjk = p ◦ λk ◦ ρk−1 = ¯ ¯ ¯ λk ◦ p and consequently Te p ◦ AdG. G/K. Theorem. ¯ ¯ ¯ Now we consider the associated vector bundle G[Te (G/K). Furthermore G[g/K.K (k) = Te p ◦ Te (conjk ) = Te λk ◦ Te p. p. g/K).K of K in g.and we have the factor representation Ad⊥ : K → GL(g/K. ¯ The left action g → T (λg ) of G on T (G/K) coincides with the left action of G on G ×K g/K. AdG. Ad⊥ ] ⊕ G[K. AdG. It remains to show the last assertion. p. K) and the associated vector bundles G[g/K. so we obtain G[g/K. Ad⊥ ] = (G ×K G/K. Ad⊥ ] → G/K × 0 This sequence splits ﬁber wise thus also locally over G/K. ¯ Thus the isomorphism Te p : g/K → Te (G/K) is K-equivariant for the ¯ ¯ ¯ representations Ad⊥ and Te λ : k → Te λk .K ] and it remains to show that = G[g. ¯ The mapping T λ : G×Te (G/K) → T (G/K) is K-invariant and therefore ¯ induces a mapping ψ as in the following diagram: G × Te (G/K) − − → T (G/K) −− ¯ q (b) G ×K Te (G/K) − − → T (G/K) −− ¯ ψ πG/K p G/K G/K ¯ Tλ This mapping ψ is an isomorphism of vector bundles.

n) × Rk → Rn . n) given also by the orthogonal complement mapping. n) × Rn with its standard Riemannian metric is isomorphic to the universal vector bundle En−k over G(n − k. n) as the space of all linear isometries Rk → Rn .5 we know that (V (k.5 the description of V (k. p. n) → G(n − k. AdG. Ek ⊥ ). AdG (g)X). Then for k ∈ K we have ϕ((g.56 10. where the isomorphism covers the diﬀeomorphism G(k. ¯ 10. we get from it the evaluation mapping ev : V (k. Recall from 10. The tangent bundle of the Grassmann manifold is T G(k. Using the banal representation of O(k) we consider the associated vector bundle (Ek := V (k. n)). n) × Rn . AdG (g)X). From 10. n) = O(n)/O(n − k). G(k.16 Tangent bundles of Grassmann manifolds. Note ﬁnally that the ﬁber wise orthogonal complement Ek ⊥ of Ek in the trivial vector bundle G(k. G(k. Principal Fiber Bundles and G-Bundles by ϕ(g.k.k −1 )X) = (gk.ev) (a) V (k. The mapping (p. So the ﬁber (Ek )W over the k-plane W in Rn is just the linear subspace W . n) ×O(k) Rk − − → G(kn) × Rn −− ψ is O(k)-invariant for the action R and factors therefore to an embedding of vector bundles ψ : Ek → G(k. n).K (g −1 )X) = (ak. p. It is called the universal vector bundle over G(k. n) for reasons we will discuss below in chapter 11. O(k)) is a principal ﬁber bundle. ev) in the diagram V (k. n). AdG (g. n)[Rk ]. So ϕ is K-equivariant from the ”joint” K-action to the ”on the left” K-action and therefore induces a mapping ϕ as in the diagram: ¯ G×g −−→ −− q (c) ϕ G×g G ×K g − − → G/K × g −− ϕ ¯ p pr1 G/K G/K The map ϕ is a vector bundle isomorphism. Corollary.k) = ϕ(gk. n) × Rk q − − → G(kn) × Rn −− (p. . X) = (g. X). n) = L(Ek .

T ρ. T(a.X = −T ρa−1 .(Y.X + T ρb . .1. Rn−k ) is given by (A. a−1 ).X. Finally we have an O(k) × O(n − k) . g) = Te (ρg ). Ek ⊥ ) = L O(n) = O(n) × O(k)×O(n−k) × O(k)×O(n−k) Rk . The tangent group of a Lie group. We have G(k.T ρa .b) µ.T λa−1 . Finally L(Ek . Principal Fiber Bundles and G-Bundles 57 Proof. Ta ν.T λa . Rn−k ). × O(k)×O(n−k) Rn−k where the left action of O(k) × O(n − k) on L(Rk .Ad(a−1 )X. Let G be a Lie group with Lie algebra g. b) = (X + Ad(a)Y. the group structure on T G looks as follows: (X.b) µ.(X.X + T λa . We will use the notation from 1. So T G is isomorphic to the semidirect product g G.T ρb .A−1 . Via the isomomorphism T ρ : g × G → T G. a).15 we get T G(k.ξa = −Te (λa−1 ). n)[Rk ] = O(n)/O(n − k) × Rk = O(n) O(k) × O(k)×O(n−k) Rk .X. ηb ) = Ta (ρb ).(T ρa .Y = = T ρab .T ρa .10. First note that T G is also a Lie group with multiplication T µ and inversion T ν. O(n) L(Rk . given by the expressions T(a.ηb and Ta ν.T ρa−1 . n) = O(n)/(O(k) × O(n − k)). T ρb . n) = O(n) × O(k)×O(n−k) so(n)/(so(k) × so(n − k)). n) = O(n)/O(n − k) and the right action of O(k) commutes with the right action of O(n − k) on O(n).T ρa . therefore V (k. Lemma. Rn−k ) → so(n)/(so(k)×so(n−k)).b) and (X. a)−1 = (−Ad(a−1 )X.equivariant linear isomorphism L(Rk .ξa + Tb (λa ). On the other hand we have V (k.C. as follows: so(n)/(so(k) × so(n − k)) = skew skew 0 0 skew = 0 A −A 0 : A ∈ L(Rk .Y = T ρab (X + Ad(a)Y ). Rn−k ) 10.T ρa . so by theorem 10.17. where O(n − k) acts trivially on Rk . Proof.X = −T ρa−1 .Ta (ρa−1 ). a.(ξa .Y ) = T ρb . B)(C) = B.ξa .

p. g) is the right logarithmic derivative of ϕαβ which is given by δϕαβ (ξx ) = T (ρϕαβ (x)−1 ). M. the bundle V P → M is isomorphic to T P |0M and T r restricts to a free right action. M. πE . (T Uα .X) = β = (ξx . (3) Obviously T r : T P × T G → T P is a free right action which acts transitive on the ﬁbers of T p : T P → T M . T2 ] = P ×G T S.X.ξ = 0 } of T E is called the vertical bundle and is denoted by (V E. p. X). Principal Fiber Bundles and G-Bundles Remark. (2) The mapping (u.58 10. Since V P = (T p)−1 (0M ). T ]. Then the following assertions hold: (1) (T P. π. T M.(b.T (ϕαβ ). the semidirect product structure looks somewhat awkward: (a. T λ. T ϕα : T P |T Uα → T Uα × T G) is again a principal ﬁber bundle atlas with cocycle of transition functions (T ϕαβ : T Uαβ → T G). (3) The vertical bundle of the principal bundle as bundle over M is again a principal bundle: (V P. where δϕαβ ∈ Ω1 (Uαβ . Let (Uα . M. (5) The vertical bundle of the associated bundle P [S.(δϕαβ (ξx ) + Ad(ϕαβ (x))X)).(g. ] is given by V (P [S. For completeness sake we include here the transition formula for this atlas in the right trivialization of T G: T (ϕα ◦ ϕ−1 )(ξx . . E). In the left trivialization T λ : G × g → T G. T G). T p. So (1) follows. G) be a principal ﬁber bundle with principal right action r : P × G → P . T G) is again a principal ﬁber bundle with principal right action T r : T P × T G → T P . (2) The vertical bundle (V P. X) is a vector bundle isomorphism P × g → V P over P . For a ﬁber bundle (E. Ad(b−1 )X + Y ) and (a. X) = Te (λg ). Proof. S) the subbundle V E = { ξ ∈ T E : T p. (4) The tangent bundle of the associated bundle P [S. X) → Te (ru ).3 the result follows. ] is given by T (P [S. X)−1 = (a−1 . Tangent and vertical bundles. which is transitive on the ﬁbers. Theorem. T G).ξx . 10. Te (ρϕαβ (x).g). Since T is a functor which respects products. Te (ρg ). ]) = T P [T S. T M. The assertion about the principal action is obvious. so by lemma 10. p ◦ π. ϕα : P |Uα → Uα × G) be a principal ﬁber bundle atlas with cocycle of transition functions (ϕαβ : Uαβ → G). Let : G × S → S be a left action. T p. Y ) = (ab.e) r.X = T(u.(0u . P. g) of the principal bundle is trivial as a vector bundle over P : V P = P × g. ]) = P [T S. − Ad(a)X).18. describing the principal ﬁber bundle (T P. Let (P.

(x. ηs ) and this implies the result . Principal Fiber Bundles and G-Bundles 59 (4) The transition functions of the ﬁber bundle P [S.ηs = T2 .(0x . ηs ) = T . ηs ) ∈ T Uαβ × T S.10.(0ϕαβ (x) . ηs ) = T ( ϕαβ (x) ). ] they transform as T ( ◦ (ϕαβ × IdS )). (5) Vertical vectors in T P [S. Under the transition functions of T P [S. ] have local representations (0x . ] are T ( ◦ (ϕαβ × IdS )) = T ◦ (T ϕαβ × IdT S ) : T Uαβ × T S → T G × T S → T S. from which the result follows. Then the transition functions of T P [S. ] are given by the expression ◦ (ϕαβ × IdS ) : Uαβ × S → G × S → S.

G) be a principal ﬁber bundle. We take the negative sign here to get the usual connection form as in [Kobayashi-Nomizu I. M. 11. so we have ω(ζX (u)) = X for all X ∈ g. 2 (rg )∗ R = R for all g ∈ G. the result follows. Conversely a 1-form ω ∈ Ω1 (P.Ad(g −1 ).ω. p.Φ = Φ. which is a principal connection if and only if (2) is satisﬁed. 1963]. (2) ω is G-equivariant. From Te (rug ).Xu ) (ug) = Φ(Tu (rg ).Φ(Xu ) both directions follow.Xu ) = ζω(Tu (rg ).60 11. g) by Ω(Xu . M.X. g). G) then the connection form has the following two properties: (1) ω reproduces the generators of fundamental vector ﬁelds.Xu ) and Te (rug ). Recall from 9.1 that the curvature R = 1 [Φ. Principal and Induced Connections 11. So ω(Xu ) := Te (Ru )−1 . (2). for all g ∈ G. Φ] is then also 2 rg -related to itself for all g ∈ G. g).T (rg ) and Φ is rg -related to itself. so that T (rg ).2. Let (P. Recall from 10. Lemma.15. T P ).3 that a (general) connection on P is a ﬁber projection Φ : T P → V P . which is called the (Lie algebra-valued) curvature form of the connection. . We already noted in 11.R(Xu . so we have ((rg )∗ ω)(Xu ) = ω(Tu (rg ). p. Proof. Recall from 1.16.ω(Tu (rg ). Since Te (ru ) : G → Vu P is an isomorphism. Let Φ be a principal connection on the principal ﬁber bundle (P.Φ(Xu ) ∈ g and in this way we get a g-valued 1-form ω ∈ Ω1 (P . V P ) is a principal connection on the principal ﬁber bundle (P. Curvature.ω(Xu ) (ug) = Tu (rg ). Yu ).18. Te (ru ). Principal connections.1. If Φ ∈ Ω1 (P .ω(Xu ).[Φ.6 the curvature R = 1 . Φ] is then also G-invariant. Yu ) := −Te (ru )−1 .ω(Xu ) = ζAd(g−1 ). p. (1).ω(Xu ) for all g ∈ G and Xu ∈ Tu P . Such a connection Φ is called a principal connection if it is G-equivariant for the principal right action r : P × G → P . g) satisfying (1) deﬁnes a connection Φ on P by Φ(Xu ) = Te (ru ). G) with connection form ω ∈ Ω1 (P .Xu ) = Ad(g −1 ).ζω(Xu ) (u) = Tu (rg ).2 that the vertical bundle of P is trivialized as a vector bundle over P by the principal right action. M. Consequently we have for the Lie derivative LζX ω = −ad(X). which is called the (Lie algebra valued) connection form of the connection Φ. Since R has vertical values we may again deﬁne a g-valued 2-form Ω ∈ Ω2 (P .3. By theorem 8. viewed as a 1-form in Ω1 (P . or (rg )∗ Φ = Φ in the sense of 8.ω(ζX (u)) = Φ(ζX (u)) = ζX (u) = Te (ru ). the fundamental vector ﬁeld mapping ζ : G → X(P ) for the principal right action.

11. ω])(X. Xp+q ) = 1 = signσ [Ψ(Xσ1 . Y ) = Xω(Y ) − Y ω(X) − ω([X. Yu ) = Te (rug ). . .((rg )∗ Ω)(Xu . Y − ΦY ] = Φ[X. (2) Ω is G-equivariant in the following sense: (rg )∗ Ω = Ad(g −1 ).Ω(Xu . Y ∈ C ∞ (H(P )) we have R(X. Xσ(p+q) )]g p! q! σ or equivalently by [ψ ⊗ X. ω] − [ω.Ad(g −1 ).((rg )∗ R)(Xu .Y ]) 1 (dω + [ω.Ω. In particular for ω ∈ Ω1 (P . it kills vertical vectors. .Ω(Xu . . Y ]) 2 by 1. Y ) = 2[ω(X). .e. g) we have [ω. (3) The Maurer-Cartan formula holds: Ω = dω + 1 [ω. and for horizontal vector ﬁelds X. Theorem. ω] = −ad(X)ω + ad(X)ω = 0 So the formula holds for vertical vectors. Yu ) = Tu (rg ). ω](X. Y ] = ζω([X. 2 Proof.R(Xu . ω(Y )]. . θ ⊗ Y ] := ψ ∧ θ ⊗ [X. Xσp ). Y ]. The curvature form Ω of a principal connection with connection form ω has the following properties: (1) Ω is horizontal. g) of all g-valued forms on P in a canonical way with the structure of a graded Lie algebra by [Ψ. Θ](X1 . . ω]g ) = iζX dω + [iζX ω. Yu ) = = −Tu (rg ).[ω. Consequently LζX Ω = −ad(X). Y ]) = −ω([X.4. Θ(Xσ(p+1) . Yu ). Tu (rg ). i.Xu . Y ) = Φ[X − ΦX.Yu ) (u) = = ζAd(g−1 ). (1) is true for R by 9. . Tu (rg ). ω]g .Ω(Tu (rg ). For (2) we compute as follows: Te (rug ).Ω. . For X ∈ g we have iζX R = 0 by (1) and 1 1 1 iζX (dω + . .Xu .Yu ) = = −Rug (Tu (rg ).Yu ) = −Tu (rg ). . .3. (3).Yu ) (ug) = = Te (rug ). . iζX ω] = 2 2 2 = LζX ω + [X.ζΩ(Xu . Principal bundles and induced connections 61 We equip the space Ω(P .

Now let (fα ) be a smooth partition of unity on M which is subordinated to the open cover (Uα ). 0g ) + X = Ad(g −1 )ωα (ξx ) + X. the local expressions of ω. g) = −T (ϕα ). Lemma. g). G) (with paracompact basis) admits principal connections. We may associate the following local data to the connection: (2) ωα := sα ∗ ω ∈ Ω1 (Uα .4. M. 0g ).T (ϕα )−1 (ξx . We will use the forms Θαβ := ϕαβ ∗ Θ ∈ Ω1 (Uαβ .Te (rϕα (x. (7) Γα (ξx . Te λg .Ad(g −1 )ωα (ξx ).X)) := X for ξx ∈ Tx Uα and X ∈ g.g) )ω.T (ϕα )−1 (ξx . 0g ) = −T (ϕα ). Let Φ = ζ ◦ ω ∈ Ω1 (P .1 are invariant under convex linear combinations. .T (ϕα )−1 (ξx . (3) The Christoﬀel forms Γα ∈ Ω1 (Uα . Local descriptions of principal connections. Any principal ﬁber bundle (P. These local data have the following properties and are related by the following formulas. Proof. α Lemma. (1) Let Θ ∈ Ω1 (G. T λg . (6) γα (ξx . G) with some principal ﬁber bundle atlas (Uα . 0g ) = −Te (λg )ω. ω is a principal connection on P . From the deﬁnition of the Christoﬀel forms we have Γα (ξx . Let (Uα . (4) γα := (ϕ−1 )∗ ω ∈ Ω1 (Uα × G. g)) = −T (ϕα ). g) satisﬁes the requirements of lemma 11.7. 0g ) = −Te (λg )γα (ξx .Φ.e. which are given by (0x .3 shows that γα ∈ Ω1 (P |Uα .ηg .T (ϕα )−1 (ξx . We consider a principal ﬁber bundle (P. Γα (ξx . Θ(ηg ) := Tg (λg−1 ).62 11. M. Proof. 0g ) = −Te (λg ).ϕαβ = sβ . and let ω := α (fα ◦ p)γα .γα (ξx . g). i.1 and thus is a principal connection on P |Uα . V P ) be a principal connection.3. (5) The forms ωα ∈ Ω1 (Uα . g). ϕα : P |Uα → Uα × G)α be a principal ﬁber bundle atlas. Since both requirements of lemma 11. g) = −Te (λg ). g) satisfy the transition formulas ωα = Ad(ϕ−1 )ωβ + Θβα .Φ. 0g ) = −Te (ϕα ◦ rϕα (x. 11. Principal bundles and induced connections 11. X(G)) from 9. g). e) and satisfy sα .T (ϕα )−1 (ξx .g) )ω. g) be the left logarithmic derivative of the identity. We consider the sections sα ∈ C ∞ (P |Uα ) which are given by ϕα (sα (x)) = (x. ω ∈ Ω1 (P . p. 0g ). An α easy computation involving lemma 1. βα and any set of forms like that with this transition behavior determines a unique principal connection. ψα : P |Uα → Uα × G) and corresponding cocycle (ϕαβ : Uαβ → G) of transition functions. Let us deﬁne γα (T ϕ−1 (ξx .X) = γα (ξx . the physicists version of the connection. p.

χ∗ ◦(rg )∗ = (rg )∗ ◦χ∗ for all g ∈ G. 0e )) α α = Ad(g −1 )(sα ∗ ω)(ξx ) = Ad(g −1 )ωα (ξx ). 11. . that Flt Now we deﬁne the covariant exterior derivative dω : Ωk (P . χ∗ ◦χ∗ = χ∗ . if one of the two forms has real values. the curvature form. the space Ωhor (P . . The projection χ∗ has the following properties: χ∗ (ϕ∧ψ) = χ∗ ϕ∧χ∗ ψ. Xk ) = ϕu (χ(X1 ). . Via (7) the transition formulas for the ωα are easily seen to be equivalent to the transition formulas for the Christoﬀel forms in lemma 9. im χ = HP . . They follow easily from the corresponding properties of χ. The covariant derivative. The second part is seen from γα (ξx . Principal bundles and induced connections 63 This is the ﬁrst part of (7). (4) dω ◦ p∗ = d ◦ p∗ = p∗ ◦ d : Ω(M . W ) → k+1 Ω (P . The second part follows from (6).X) = ω(T (ϕα )−1 (ξx . The mapping χ∗ is a projection onto the subspace of horizontal diﬀerential forms. (3) (rg )∗ ◦ dω = dω ◦ (rg )∗ for each g ∈ G. (2) L(ζX ) ◦ dω = dω ◦ L(ζX ) for each X ∈ g. and χ∗ ◦ L(ζX ) = L(ζX ) ◦ χ∗ . G) be a principal ﬁber bundle with principal connection Φ = ζ ◦ ω. 9. α So the ﬁrst part of (6) holds. W ) := {ψ ∈ Ω(P . W ) : iX ψ = 0 for X ∈ V P }. . The covariant exterior derivative dω has the following properties. the Bianchi identity. and χ ◦ T (rg ) = T (rg ◦ χ) for all g ∈ G. γα (ξx . We consider the horizontal projection χ = IdT P − Φ : T P → HP . 0g ) + X. we consider the mapping χ∗ : Ω(P . W ). where R is the curvature.11. (7) dω ◦ χ∗ − dω = χ∗ ◦ i(R) ◦ d. Te (ρg )0e ) = (ω ◦ T (ϕα )−1 ◦ T (ρg ))(ξx . (8) dω ◦ dω = χ∗ ◦ i(R) ◦ d. 0g )) = = γα (ξx . 0g ) + ω(ζX (ϕ−1 (x. . If W is a ﬁnite dimensional vector space. i. . p. ker χ = V P .3.7. χ(Xk )). M. W ) → Ω(P . 0e ) = = (ω ◦ T (rg ◦ ϕ−1 ))(ξx . χ∗ ω = 0. (6) dω Ω = 0. T λg . Let (P. (1) dω (ϕ ∧ ψ) = dω (ϕ) ∧ χ∗ ψ + (−1)deg ϕ χ∗ ϕ ∧ dω (ψ). W ) which is given by (χ∗ ϕ)u (X1 . the last property uses ζ(X) = rexp tX . W ) by the prescription dω := χ∗ ◦ d.5. The notion of horizontal form is independent of the choice of a connection. g))) = γα (ξx . Theorem. which satisﬁes χ ◦ χ = χ. 0e ) = Ad(g −1 )ω(T (ϕ−1 )(ξx . cf. . 0g ) = γα (ξx . W ) → Ωhor (P .e. . (5) dω ω = Ω.

. . . ω]) 2 = χ∗ ddω + 1 χ∗ d[ω. Principal bundles and induced connections Proof. . We have (dω ω)(ξ. . ω] − [ω. . . . ω] 2 = 1 χ∗ ([dω. . . χ(X0 ). For ϕ ∈ Ω(P . . . χ(Xk ))) + i<j (−1)i+j ϕ([χ(Xi ). . Xk ) = (dχ∗ ϕ)(χ(X0 ). . χ(Xj )] − Φ[χ(Xi ). χ(Xi ). . Xk ). . . since χ∗ ω = 0. η)) = Φ[χξ. χ(Xj ). . (1) through (4) follow from the properties of χ∗ . (5). ) = (dϕ)(χ(X0 ). χ(Xk )) = 0≤i≤k (−1)i χ(Xi )((χ∗ ϕ)(χ(X0 ). χη) = (χξ)ω(χη) − (χη)ω(chξ) − ω([χξ. ) = 0≤i≤k (−1)i χ(Xi )(ϕ(χ(X0 ). η) = dω(χξ. . χη]) and R(ξ. χ(Xk )) + (iR ϕ)(χ(X0 ). . . χ(Xj )]. (6). . . W ) we have (dω χ∗ ϕ)(X0 . . . . . . ω] 2 = [χ∗ dω. χ(X0 ). χη]. . η) = (χ∗ dω)(ξ. . . . . . . by (7) . . . . χ(Xj ). . dω dω = χ∗ dχ∗ d = (χ∗ iR + χ∗ d)d = χ∗ iR . η) = −ζ(Ω(ξ. χη]) = −ω([χξ. . . . . . . χ(Xi ). . χ(Xj )]. We have dω Ω = dω (dω + 1 [ω. . . χ(Xk ))) + i<j (−1)i+j (χ∗ ϕ)([χ(Xi ). (8). . χ(Xi ).64 11. . . . χ(Xk )) = (dω + χ∗ iR)(ϕ)(X0 . . . . . . . (7). . χ∗ ω] = 0. . . . . . dω]) = χ∗ [dω. χ(Xi ).

and ω( dt Pt(c. Theorem. Let (P. u0 )/ Hol0 (ω.11. x0 ). (Reparametrizing c by a function which is ﬂat at each corner of c we may assume that any c is smooth. u0 )) = Hol0 (ω. M. t.g) = Ad(g −1 ) Hol(ω. x0 ).10. u0 ) a normal subgroup Hol(ω.g) = Pt(c. u0 ). 2. Pt(c. u0 ) and Hol0 (ω. p. u) ∈ Pc(t) . u. The elements τ (u0 . and (2) is seen as d d follows: ω( dt Pt(c. u0 )) = Hol(ω. t. Then the parallel transport for the principal connection is globally deﬁned and g-equivariant. For each curve c in M we have Hol(ω. u0 ) is at most countable. i.g) = Ad(g −1 ) Hol0 (ω. t.g) = Ad(g −1 )ω( dt Pt(c. M. t. Pt(c. Properties (1) and (3) follow from theorem 9. Hol0 (ω. which are bounded with respect to any left invariant Riemannian metric on G. so Hol(ω. We have Hol(ω. Let (P.g). t) : Pc(0) → Pc(t) is G-equivariant. x0 ) ⊂ Diﬀ(Px0 ) as the group of all Pt(c. u0 ).8. t. Each left invariant metric on a Lie group is complete. u). t. t. Pt(c. t. We assume that M is connected and we ﬁx x0 ∈ M .g = Pt(c. 0) = IdPc(0) . t. So the connection is complete by the remark in 9. u.7. u0 ) is a connected Lie subgroup of G and the quotient group Hol(ω. u0 )) ∈ G form a subgroup of the structure group G which is isomorphic to Hol(Φ. t. u)) = 0. G) be a principal ﬁber bundle with principal connection Φ = ζ ◦ ω. 11. 1. (3) For smooth f : R → R we have Pt(c. In view of developments which we make later in section 12 we deﬁne the holonomy group Hol(Φ. u0 ) and we call it also the holonomy group of the connection.) If we consider only those curves c which are nullhomotopic. f (t). Considering only nullhomotopic curves we get the restricted holonomy group Hol0 (ω. u0 ). u). G) be a principal ﬁber bundle with principal connection ω. u)) = 0 implies that Pt(c. we denote it by Hol(ω. X(G)) of the connection ω with respect to a principal ﬁber bundle atlas (Uα . p. ϕα ) take values in the Lie subalgebra XL (G) of all left invariant vector ﬁelds on G.g. u0 . u0 ) and Hol0 (ω. f (0). we obtain the restricted holonomy group Hol0 (Φ. Holonomy groups. u0 ) is also a Lie subgroup of G. 3. In detail: For each smooth c : R → M there is a smooth mapping Ptc : R × Pc(0) → P such that the following holds: d (1) Pt(c. The Christoﬀel forms Γα ∈ Ω1 (Uα . 1) : Px0 → Px0 for c any piecewise smooth closed loop through x0 . Now let us ﬁx u0 ∈ Px0 . u) = Pt(c ◦ f.e. t. Pt(c. . Pt(c. Pt(c. (2) Pt(c. u). Proof. u)). u0 . Principal bundles and induced connections 65 11.6 Theorem. t.

M. T (P [S])) by the following diagram: TP × TS −−→ −− Φ×Id T q=q TP × TS q T (P × S) Tq T (P ×G S). 1963. 7. u0 ) is then again a principal connection form i∗ ω ∈ Ω1 (P (ω. u0 ) equals the structure group G for some (equivalently any) u0 ∈ Px0 . Yu ) : Xu . It is isomorphic to the Lie algebra hol(Φ.66 11. so it is itself a principal ﬁber bundle over M with structure group Hol(ω. If the curvature Ω = 0 then Hol0 (ω. ]) = P [T S. u) is a covering of M .2 and from the from the G-equivariancy of the parallel transport.18 that its tangent and vertical bundle are given by T (P [S. If one uses piecewise C k -curves for 1 ≤ k < ∞ in the deﬁnition. Then P (ω.8. T2 ] = P ×G T S. In view of assertion 5 a principal connection ω is called irreducible if Hol(ω. Proof. constructed in 10. x0 ) we considered in 9. Yu ∈ Tu P }. 10. ]) = T P [T S. P is foliated by the leaves P (ω. one gets the same holonomy groups. t. u0 ). u0 ) is a sub ﬁber bundle of P which is invariant under the right action of Hol(ω.6 and 10. u ∈ Px0 . u0 ) ⊂ g of Hol(ω. The rest of this theorem is a compilation of well known results.7. hol(ω. cf. u0 )). u0 ) = {e} and each P (ω. Let (P. p. Then we consider the associated bundle P [S] = P [S. Recall from 10. u0 ) is linearly generated by {Ω(Xu . T P ×T G T S − − → T P ×T G T S −− ¯ Φ Let us ﬁrst check that the top mapping Φ × Id is T G-equivariant.13. We construct the induced connection Φ ∈ Ω1 (P [S]. For u0 ∈ Px0 let P (ω. u0 ) be the set of all Pt(c. u0 ). u). The pullback of ω to P (ω. ] = P ×G S. T P ) be a principal connection on the principal ¯ bundle P . u0 ) and we have a reduction of structure group. u0 ) for c any (piecewise) smooth curve in M with c(0) = x0 and for t ∈ R. The Lie algebra hol(ω. Let Φ = ζ ◦ ω ∈ Ω1 (P . This follows from the properties of the mapping τ from 10. G) be a principal bundle with principal right action r : P × G → P and let : G × S → S be a left action of the structure group G on some manifold S.9. T ] = T P ×T G T S and V (P [S. Principal bundles and induced connections 4. 1. 6. Inducing principal connections on associated bundles. and we refer to [Kobayashi-Nomizu I. For g ∈ G and X ∈ g the inverse of Te (λg )X in the Lie group T G is denoted . 5. 11. 83ﬀ] for proofs. 8. p.

The image of Φ q (V P × T S) = q (ker(T p : T P × T S → T M )) = ker(T p : T P ×T G T S → T M ) = V (P [S.9. since Φ × Id and q = T q are. Te (λg )X) = Tu (rg )ξu + T r((OP × LX )(u. s)) = q(PtΦ (c. Te (λg )X). We call it the induced connection. So by 8. T ((Te (λg )X)−1 . T ((Te (λg )X)−1 . T (P [S])) are (q : P × S → P [S])related. s). The mapping Φ is ¯ ¯ ¯ have Φ ¯ is ﬁberwise linear. The answer is given in the following . Φ] × 0 = RΦ × 0. ηs )) = (Φ(Tu (rg )ξu + ζX (ug)). g)) = Tu (rg )ξu + Tg (ru )(Te (λg )X) = Tu (rg )ξu + ζX (ug). u). T ((Te (λg )X)−1 . T ((Te (λg )X)−1 . T P × T S) and Φ ∈ Ω1 (P [S].17. ηs )) ¯ So the mapping Φ × Id factors to Φ as indicated in the diagram. T q ◦ (RΦ × 0) = RΦ ◦ (T q ×M T q). Recognizing induced connections. ]. G) and a left action : G × S → S. We consider again a principal ﬁber bundle (P.15 we have for the curvatures RΦ×Id = 1 [Φ × Id. p. and we ¯ ◦ Φ = Φ from (Φ × Id) ◦ (Φ × Id) = Φ × Id. From the diagram it also follows. Then the following question arises: When is the connection Ψ induced from a principal connection on P ? If this is the case. ηs )) = (T r((Φ × Id)ξu . ηs )) = (Φ(Tu (rg )ξu ) + Φ(ζX (ug)).e. ¯ Thus Φ is a connection on the associated bundle P [S]. ηs )) = (Φ(Tu (rg )ξu ) + ζX (ug).3 we have T r(ξu . Principal bundles and induced connections 67 by (Te (λg )X)−1 . ¯ 2 that they are also q-related. t. Φ]. t. Φ × Id] = 1 [Φ. that the vector valued forms Φ×Id ∈ ¯ Ω1 (P × S. ]). M.11. i. q(u. ¯ By uniqueness of the solutions of the deﬁning diﬀerential equation we also get that PtΦ (c. see lemma 10. T ((Te (λg )X)−1 . Suppose that Ψ ∈ Ω1 (P [S]. Furthermore by 1. We may compute (Φ × Id)(T r(ξu . Te (λg )X). T (P [S])) is a connection on the associated bundle P [S] = P [S. ¯ 11. 2 2 ¯ ¯ RΦ = 1 [Φ. we say that Ψ is compatible with the G-structure on P [S].

Let Ψ be a (general) connection on the associated bundle P [S]. 0s ) ¯ Φ ¯ = −(T (ψα ) ◦ Φ ◦ T q ◦ (T (ϕ−1 ) × Id))(ξx .7 its Christoﬀel symbols are given by −1 ¯ Γα (ξx . s) = −(T (ψα ) ◦ Φ ◦ T (ψα ))(ξx . which by the ﬁrst part of the proof induces the given connection Ψ on P [S]. i. Then by the proof of theorem 10. g). s) = q(ϕ−1 (x. Then the connection Ψ is induced from a principal connection ω on P if and only if the following condition is satisﬁed: In some (equivalently any) ﬁber bundle atlas (Uα . Now let us conversely suppose that a connection Ψ on P [S] is given such that the Christoﬀel forms Γα with Ψ respect to a ﬁber bundle atlas of the G-structure have values in Xf und (S). So the condition is necessary.(7) by (2) = −(T (ψα ) ◦ = −(T (ψα ) ◦ T q ◦ (Φ × Id))(T (ϕ−1 )(ξx .4. Principal bundles and induced connections Theorem. ωα (ξx ). 0e ). Then unique g-valued forms ωα ∈ Ω1 (Uα . 0s ) α T q)(Φ(T (ϕ−1 )(ξx .8 by 11. By 9. α ¯ Let Φ = ζ ◦ ω be a principal connection on P and let Φ be the induced connection on the associated bundle P [S]. e). 0s ) α Φ = −T (ψα ◦ q ◦ s = −Te ( )ωα (ξx ) = −ζωα (ξx ) (s).7 the induced ﬁber bundle atlas (Uα . 0e )). Proof. g) are given by the equation Γα (ξx ) = ζ(ωα (ξx )). ψα : P [S]|Uα → Uα × S) is given by (1) (2) −1 ψα (x. so that Ψ they give a unique principal connection on P . e)). Ψ since the action is inﬁnitesimally eﬀective. Let (Uα . .e.(3) by 11. Let us suppose that the action is inﬁnitesimally eﬀective.4. s). ϕα : P |Uα → Uα × G) be a principal ﬁber bundle atlas for P .4.s).(5) for the ω α . 0s ) = (T (ψα ) ◦ T q)(T (ϕ−1 )(Γα (ξx . s) = (x. X(S)) have values in the sub Lie algebra Xf und (S) of fundamental vector ﬁelds for the action .7 for the Γα follow the transition formulas 11. 0s ) α by (1) by 11. 0e . 0s ) α (ϕ−1 α × Id))(0x .68 11. From the transition formulas 9. the fundamental vector ﬁeld mapping ζ : g → X(S) is injective. ψα ) of P [S] belonging to the G-structure of the associated bundle the Christoffel forms Γα ∈ Ω1 (Uα . α (ψα ◦ q)(ϕ−1 (x. g.

This follows from the ﬁberwise linearity of the composants of K and from its deﬁnition. T (E)) from 11. pr1 –πE –ﬁberwise linear and also p–T p–ﬁberwise linear.11. T E) such that Ψ : T E → V E → T E is also T p–T p– ﬁberwise linear is called a linear connection.8. If (E. T P ) be a principal connection on P . p. Proof.10. Let (P. T M ). T p ◦ pr1 : T (E) = T P ×T G T W → T M . Principal bundles and induced connections 69 11. which was treated in some detail in 10. Inserting the projections of both vector bundle structures on T (E) into the diagram in 11. a connection Ψ ∈ Ω1 (E. Recall now from 11. M. . M. Inducing principal connections on associated vector bundles. The connector K : T E → E is πE –p–ﬁberwise linear and T p–p–ﬁberwise linear and satisﬁes K ◦ vlE = pr2 : E ×M E → T E → E. We ¯ consider the induced connection Φ ∈ Ω1 (E. M ) is a vector bundle. G) be a principal ﬁber bundle and let ρ : G → GL(W ) be a representation of the structure group G on a ﬁnite dimensional vector space W .10. the second one is the derivative of the vector bundle structure on E. T p. Equivalently a linear connection may be speciﬁed by a connector K : T E → E with the three properties of lemma 11. An easy check with 11. Linear connections.11. The tangent bundle T (E) = T P ×T G T W has two vector bundle structures with the projections πE : T (E) = T P ×T G T W → P ×G W = E. We say that it is a linear connection on the associated bundle. We consider the associated vector bundle (E := P [W. p. Now let Φ = ζ ◦ ω ∈ Ω1 (P .11. ¯ Now we deﬁne the connector K of the linear connection Φ by ¯ K := pr2 ◦ (vlE )−1 ◦ Φ : T E → V E → E ×M E → E. Lemma. ρ]. p. For then HE := {ξu : K(ξu ) = 0p(u) } is a complement to V E in T E which is T p–ﬁberwise linearly chosen. the ﬁrst one is the vector bundle structure of the tangent bundle. E) of E (where n is the ﬁber dimension of E). 11.9 or a direct construction shows that Ψ is then induced from a unique principal connection on the linear frame bundle GL(Rn . W ).8 one easily sees that the induced connection is linear in both vector bundle structures: the new aspect is that it is a linear endomorphism of the vector bundle (T E.8 the vertical lift vlE : E ×M E → V E. which is an isomorphism.

See e. E) → Ef (x) makes sense and ( X s)(x) = X(x) s. This covariant derivative has the following properties: is C ∞ (N. the clumsy and unclear treatment of it in [Eells-Lemaire. If Y ∈ X(Q) and X ∈ X(N ) are g-related. Lemma. 1983]. From the universal property of the ∞ pullback it follows that the vector space Cf (N. By the method of 7. E). R)-linear in X ∈ X(N ). For vector ﬁelds X. the derivation property of X . E) is canonically linearly ∞ ∗ isomorphic to the space C (f E) of sections of the pullback bundle. Y )s : = = ([ X Y X. smooth mapping s : N → E. given by a connector K : T E → E with the properties in lemma 11. s− Y]− Y Xs − [X. E) the vector space of all smooth mappings s : N → E with p ◦ s = f – they are called sections of E along f .s) = dh(X).s + h.12.4 it follows that RE is a 2-form on M with values in the vector bundle L(E. X s for h ∈ C ∞ (N. E). For any manifold N . p. So for a tangent vector X ∞ Xx ∈ Tx N the mapping Xx : Cf (N.10. Remark. Property (6) is not well understood in some diﬀerential geometric literature. E) → Cf (N. and s.Yy s = Yy (s ◦ g). ∞ s is R-linear in s ∈ Cf (N.70 11. ∞ If f : N → M is a ﬁxed smooth mapping. and vector ﬁeld X ∈ X(N ) we deﬁne the covariant derivative of s along X by (1) Xs := K ◦ T s ◦ X : N → T N → T E → E. (3) (4) Xs Proof.Y ] )s is C ∞ (M. M ) be a vector bundle with a linear connection. (6) For any manifold Q and smooth mapping g : Q → N and Yy ∈ Ty Q we have T g. E). (5) X (h. R). . let us denote by Cf (N. All these properties follow easily from the deﬁnition (1). Y ∈ X(M ) and a section s ∈ C ∞ (E) an easy computation shows that RE (X. Then the covariant derivative may be viewed as a bilinear mapping (2) ∞ ∞ : X(N ) × Cf (N. Let (E. R)-linear in X. then we have Y (s ◦ g) = ( X s) ◦ g.Y ] s [X.g. Y . Covariant derivative on vector bundles. Principal bundles and induced connections 11.

Y )s.13.e. Xσ(p+q) )). Xp ). .11. . f ∗ E) is deﬁned by (where the Xi are vector ﬁelds on N ) p (d Φ)(X0 . . Xσp )Φ(Xσ(p+1) . Xj . . . . Covariant exterior derivative. . . . . . . . Xp+q ) = = 1 p! q! σ sign(σ) ϕ(Xσ1 . . The covariant exterior derivative is well deﬁned and has the following properties. . . The graded space Ω(N . Xp ) = i=0 (−1)i Xi Φ(X0 . . Let (E. f ∗ E) we have d (g ∗ Φ) = g ∗ (d Φ). The covariant exterior derivative d : Ωp (N : f ∗ E) → Ωp+1 (N . . Xj ]. . p. . . Xp ) + (−1)i+j Φ([Xi . f ∗ E) we have (d s)(X) = X s. given by a connector K : T E → E. f ∗ E) → Ω(N . For f : N → M . . . It can easily be shown that the graded module homomorphisms H : Ω(N . Xσp )(Φ(Xσ(p+1) . RE ∈ Ω2 (M . We can also view them as forms on N with values along f in E. . f ∗ E) (so that H(ϕ ∧ Φ) = (−1)deg H. 0≤i<j≤p Lemma. which are given by (µ(K)Φ)(X1 . . . (3) For smooth g : Q → N and Φ ∈ Ω(N . Principal bundles and induced connections 71 i. . . Xσ(p+q) ). f ∗ L(E. but we do not introduce an extra notation for this. . X0 . . . (2) d (ϕ ∧ Φ) = dϕ ∧ Φ + (−1)deg ϕ ϕ ∧ d Φ. . f ∗ E) is a graded Ω(N )-module via (ϕ ∧ Φ)(X1 . For a smooth mapping f : N → M let Ω(N . . f ∗ E) be the vector space of all forms on N with values in the vector bundle f ∗ E. . E) along f we obtain X Y s− Y Xs − [X. Xi . E)). E)). . Y ∈ X(N ) and a section s ∈ ∞ Cf (N. . . . . Xp+q ) = = 1 p! q! σ sign(σ) K(Xσ1 . Xi . . . L(E. M ) be a vector bundle with a linear connection. (1) For s ∈ C ∞ (f ∗ E) = Ω0 (N . (4) d d Φ = µ(f ∗ RE )Φ.Y ] s = (f ∗ RE )(X. . . . . deg ϕ ϕ ∧ H(Φ)) are exactly the mappings µ(K) for K ∈ Ωq (N . vector ﬁelds X. It is called the curvature of the covariant derivative. . . . 11. .

Xi = Tu p. P [W ]) = C ∞ (P [W ]) → Ω0 (P . since Tu (rg )Xi − Xi ∈ Vu P.g ) = g −1 . the mapping d is well deﬁned. .τ (ux . Xk )) = q(u . ρ(g −1 )ϕu (X1 . ϕu (X1 .Xk )) = q(u . 11.Xi for each i. Xk )) = q(ug. . This formula also implies (2) immediately. Then we have for g = τ (u. .2. Let (P. ux ) = e. (rg )∗ ϕ = ρ(g −1 ) ◦ ϕ for all g ∈ G}. . Let ϕ ∈ Ωk (P . . ρ]valued diﬀerential forms on M onto the space of horizontal G-equivariant W -valued diﬀerential forms on P : q : Ω(M . f ∗ E). P [W. and X1 . . .12.g . .g. In particular for W = R with trivial representation we see that p∗ : Ω(M ) → Ωhor (P )G = {ϕ ∈ Ωhor (P ) : (rg )∗ ϕ = ϕ} is also an isomorphism. . Principal bundles and induced connections Proof. W )G = C ∞ (P. W )G .(6). Xk )) = q(u . W )G hor is a special case of the one from 10. . Xk ∈ Tu P . Proof. . p. . which satisﬁes r(ux . . . It suﬃces to investigate decomposable forms Φ = ϕ ⊗ s for ϕ ∈ Ωp (N ) and s ∈ C ∞ (f ∗ E). . . ϕug (Tu (rg ). ux . . so that ug = u : q(uϕu (X1 . Recall the smooth mapping τ : P ×M P → G from 10.(3) d s ∈ Ω1 (N . W )G = {ϕ ∈ Ω(P .X1 .12. There is a canonical isomorphism from the space of P [W. W ) : iX ϕ = 0 for all X ∈ V P. (3) follows from 11. G) be a principal ﬁber bundle and let ρ : G → GL(W ) be a representation of the structure group G on a ﬁnite dimensional vector space W . ux ). τ (ux . . . and τ (ux .14. u ). . . Xk )). Since by 11. ρ]) → Ωhor (P . ((rg )∗ ϕ)u (X1 . .72 11. X1 . . . Xk ∈ Tu P hor such that Tu p. τ (ux . . (4) is checked as follows: d d (ϕ ⊗ s) = d (dϕ ⊗ s + (−1)p ϕ ∧ d s) by (2) = 0 + (−1)2p ϕ ∧ d d s = ϕ ∧ µ(f ∗ RE )s by the deﬁnition of RE = µ(f ∗ RE )(ϕ ∧ s). Tu (rg ).12. . Theorem. M. . . . The isomorphism q : Ω0 (M . Then from the deﬁnition we have d (ϕ⊗s) = dϕ ⊗ s + (−1)p ϕ ∧ d s. vx )) = vx .

χ.X1 . . Φp(ug) (Tug p. as follows: q ◦ d = dω ◦ q : Ω(M . Φp(u) (Tu p. P [W ]) is uniquely determined. we have T s. The covariant exterior derivative dω from 11. W )G = C ∞ (P. s(p(u))) and s(p(u)) = q(u. P [W ]) we deﬁne q ∈ Ωk (P .T s. . q(ux . (d s)(T p. w)) = w.X1 . M. Therefore T s. q(u. 11. M.Xu ).Xk )). df (Xu )) ∈ T W = W × W . . We consider the associated vector bundle (E := P [W. Xk ). Proof. For the converse recall the smooth mapping τ : P ×M P [W. T p. .Xu ).13. W ). .Tu (rg ). ¯ Then q Φ is smooth and horizontal. . p. If χ : T P → HP is the horizontal projection as in 11. vx )) = vx . W )G .11. . vx ).Xu = (f (u). . For Xi ∈ Tu P we put (q Φ)(X1 .14. . T f.T p. .Xu = T s.Xk ) = τ (ug.Xu s) by 11. . . . Xk ) = (q Φ)ug (Tu (rg ).12. ¯ τ For Φ ∈ Ωk (M .X1 . Tu p.15. Theorem.X1 .Xk )) τ = ρ(g −1 )(q Φ)u (X1 . .Xu . Clearly the two constructions are inverse to each other.14 and 10. τ (ux . . and ¯ ¯ τ (ux g.(1) = τ (u.χ. . . Let ﬁrst f ∈ Ω0 (P . Let (P. which satisﬁes τ (u.5.T p.Xu = T q(χ.12.Xu ) . Xk ) := τ (u. For g ∈ G we have ((rg )∗ (q Φ))u (X1 . f (u)) by 11. Φp(u) (Tu p. So we get (q d s)(Xu ) = τ (u.T p. . Principal bundles and induced connections 73 By this prescription a vector bundle valued form Φ ∈ Ωk (M . . .7. P [W ]) → Ωhor (P . W )G . ρ] → W ¯ from 10. .Xu = T q(Xu . and let ρ : G → GL(W ) be a representation of the structure group G on a ﬁnite dimensional vector space W . Tu p. K.Xk )) ¯ = ρ(g −1 )¯(u. vx ) = ρ(g −1 )¯(ux . .Xu )) = τ (u.Tu (rg ). W ) as follows. ρ]. T f. Tu (rg ). the induced ¯ connection Φ on it and the corresponding covariant derivative.T p. p. . where T f. G) be a principal ﬁber bundle with a principal connection Φ = ζ ◦ ω. . . .5 on P and the covariant exterior derivative for P [W ]-valued forms on M are connected by the mapping q from 11. Tug p. then f = q s for s ∈ hor C ∞ (P [W ]) and we have f (u) = τ (u. . .(1) by 11.

Proof.pr2 .P [W ]) RP [W ] = ρ ◦ Ω.16. T f. q. g) and the curvature RP [W ] ∈ hor Ω2 (M . T f. Y ∈ Tu P (dω dω qP [W ] s)u (X. where ψ ∈ Ωk (M ) and s ∈ C ∞ (P [W ]).15.vlP ×W .(1) from above 11.Xu ))) −1 τ (u. df.Y )s(p(u))) = (qL(P [W ]. W ) is the derivative of the representation ρ. Y ) = τ (u. .T q(χ. RP [W ] (Tu p.Xu )) ¯ = τ (u.Xu )) = (χ∗ df )(Xu ) = (dω q s)(Xu ).T q. Y ) = (q RP [W ] s)u (X. L(P [W ]. By this theorem we have for X. where ρ = Te ρ : g → L(W. Then we have dω q (ψ ⊗ s) = dω (p∗ ψ · q s) = dω (p∗ ψ) · q s + (−1)k χ∗ p∗ ψ ∧ dω q s = χ∗ p∗ dψ · q s + (−1)k p∗ ψ ∧ q d s = p dψ · q s + (−1) p ψ ∧ q d s = q (dψ ⊗ s + (−1)k ψ ∧ d s) = q d (ψ ⊗ s). Y )(qP [W ] s)(u).χ. T f. Tu p.Xu ))) τ (u.χ.T q(0u . It suﬃces to check the formula for a decomposable P [W ]-valued form Ψ = ψ ⊗ s ∈ Ωk (M.Xu ))) −1 τ (u.P [W ]) RP [W ] )u (X.(Φ × Id)(χ.vlP [W ] . pr2 . K. P [W ])) the relation qL(P [W ]. P [W ]). T f.vl−1 . Y ) = (q d d s)u (X.5.X.Φ.χ. Now we turn to the general case. ∗ k ∗ by 11. Principal bundles and induced connections = τ (u.χ.Xu . In the situation of theorem 11.Xu .74 11.χ.T q(0u . Corollary.Xu )) = = = = from above by 11.10 by 11. T f. pr2 .8 since Φ. pr2 . q(u.15 above we have for the curvature form Ω ∈ Ω2 (P .T q(χ. We use the notation of the proof of theorem 11.vlP [W ] .χ.Xu .χ = 0 since q is ﬁber linear P [W ] −1 τ (u.

s(p(u))) ρ(exp tΩ(X.2 (q ζΩ(X. Principal bundles and induced connections 75 On the other hand we have by theorem 11. .Y ) (u)) = = = = ∂ ∂t ∂ ∂t ∂ ∂t ∂ ∂t 0 0 0 0 since R is horizontal by 11.Y ) s)(Fl−t (u)) τ (u. Y ))(q s)(u). Y ))))) τ (u. exp(−tΩ(X. Y ) = (χ∗ iR dq s)u (X. s(p(u.(8) (dω dω q s)u (X. Y )). s(p(u))) by 11. Y ) = (dq s)u (R(X. Y )) = (dq s)(−ζΩ(X. Y ))τ (u.7 = ρ (Ω(X. exp(−tΩ(X. Y )).5. exp(−tΩ(X.11.

Yx ∈ Tx M and curves c from x0 to x) in X(S) with respect to the compact C ∞ -topology. For Xx and Yx ∈ Tx M we consider R(CXx . CYx ) ∈ X(Ex ). CY2 ]. generated by all these vector ﬁelds (for all x ∈ M . CV )] Ex0 ∈ hol(Φ. called the restricted holonomy group of the connection Φ at x0 . ) : Ex → T E is a vector ﬁeld along Ex . Proof. Y2 ].3. (FlCX )∗ (FlCZ )∗ R(CU. x0 ) s z by the Jacobi identity and [(FlCX )∗ C[Y1 . t) : S = Ex0 → Ex and the pullback Pt(c. since it is closed in the compact C ∞ -topology and the derivative can be written as a limit. 1)∗ R(CXx .1 let C : T M ×M E → T E be the horizontal lifting as in 9. In the setting of 12. 12.76 12. Thus [(FlCX )∗ [CY1 . and let us call it the holonomy Lie algebra of Φ at x0 .2.1. 1] → M through x0 the parallel transport Pt(c. Xx . (FlCZ )∗ R(CU. CV ) = (FlCX )∗ [CY. CV )] Ex0 ∈ hol(Φ. For X ∈ X(M ) we consider the local ﬂow FlCX of the horizontal t lift of X. x0 ) the closed linear subspace.3. CYx ) ∈ X(S). and let us assume that M is connected. x0 ). It restricts to parallel transport along any of the ﬂow lines of X in M .4) of the connection Φ. Now we choose any piecewise smooth curve c from x0 to x and consider the diﬀeomorphism Pt(c. x0 ). . If we consider only those piecewise smooth curves which are homotopic to zero. x0 ).M. For any x ∈ M and Xx ∈ Tx M the horizontal lift C(Xx ) := C(Xx . Lemma. (FlCZ )∗ R(CU. we get a subgroup Hol0 (Φ. 12. Let (E. All these diﬀeomorphisms form together the group Hol(Φ. Then for vector ﬁelds on M the expression CX ∗ CY ∗ CX ∗ CZ ∗ d dt |0 (Fls ) (Flt ) (Fl−s ) (Flz ) R(CU. Holonomy Lie algebra. a subgroup of the diﬀeomorphism group Diﬀ(S). and let R be the curvature (9. CV )] Ex0 s z Ex0 Ex0 is in hol(Φ. Holonomy 12. CV )] s −s z = [(FlCX )∗ CY. 1) =: Pt(c. hol(Φ. x0 ) is a Lie subalgebra of X(S). (FlCZ )∗ R(CU. Holonomy groups. 1) (pieced together over the smooth parts of c) is a diﬀeomorphism of S. We choose a ﬁxed base point x0 ∈ M and we identify Ex0 with the standard ﬁber S. x0 ). For each closed piecewise smooth curve c : [0. s z .p. Let us denote by hol(Φ. the holonomy group of Φ at x0 .S) be a ﬁber bundle with a complete connection Φ.

x0 ) which is smooth as a mapping R × S → S. It is suﬃcient to prove claim 2 there. an irreducible connection ω on it and a smooth action of G on S such that the Lie algebra g of G equals the holonomy Lie algebra hol(Φ. x0 ) is a ﬁnite dimensional Lie subalgebra of X(S). where g is smooth and ﬂat at each corner of c. 1) for a piecewise smooth closed curve c through x0 . p. the ﬁber bundle E is isomorphic to the associated bundle P [S]. which is nullhomotopic. x0 ). To prove claim 2 we consider the pullback bundle H ∗ E → R2 with the induced connection H Φ . CV )] Ex0 s z is in hol(Φ. x0 ). (FlCZ )∗ R(CU. 9. x0 ). 1) is a curve in Hol0 (Φ. x0 ). The structure group G equals the holonomy group Hol(Φ. x0 ) is ﬁnite dimensional and consists of complete vector ﬁelds on the ﬁber Ex0 Then there is a principal bundle (P.s = FlCX ◦ FlCY ◦ FlCX ◦ FlCY : S → S. The following theorem is a generalization of the theorem of Ambrose and Singer on principal connections. Claim 1. Let f ∈ Hol0 (Φ. we can replace c by c ◦ g. 12. G0 equals Hol0 (Φ. Holonomy 77 so also their diﬀerence [(FlCX )∗ R(CY1 . the restricted holonomy group. x0 ). d Claim 2. Let Φ be a connection on the ﬁber bundle (E. Theorem.4. x0 ) consists of complete vector ﬁelds if and only if it is generated by complete vector ﬁelds as a Lie algebra. Proof.8. Then ft := Pt(Ht . 1] = x0 . 1957]. S) and let M be connected. Then g := hol(Φ. CY2 ). so [X. ( dt ft ) ◦ ft−1 =: Zt is in g for all t. CX Then Pt(c. Y ] = 0. then f = Pt(c. By a theorem of [Palais. P and ω are unique up to isomorphism. M. see [Palais. and Φ is the connection induced by ω. and since each vector ﬁeld in it is complete. there is a ﬁnite dimensional connected Lie group G0 of diﬀeomorphisms of S with Lie algebra g.12. So we may assume that c itself is smooth. Since the parallel transport is essentially invariant under reparametrization. M. G) with ﬁnite dimensional structure group G. −s −t s t . 1] = c and H0 |[0. We put ft. Since c is homotopic to zero. Let d d X = ds and Y = dt be the constant vector ﬁelds on R2 . Suppose that for some (hence any) x0 ∈ M the holonomy Lie algebra hol(Φ. s) = Fls |S and so on. p. 1957] a ﬁnite dimensional Lie subalgebra of X(Ex0 ) like Hol(Φ. by approximation we may assume that there is a smooth homotopy H : R2 → M with H1 |[0. Let us again identify Ex0 and S.

CY ] = Φ[CX. x0 ) equals G0 . and for that it is enough to show. Then we have in the vector space X(S) −1 d ( dt ft. CYx ) for Xx . In the proof of claim 1 we have seen that Hol0 (Φ. 1)∗ R(CXx . We extend Xx and Yx to vector ﬁelds X and Y ∈ X(M ) with [X.s = 1 0 d ds −1 d ( dt ft. so it is a connected Lie subgroup by the results cited in 5. generated by the time dependent vector ﬁeld Zt . x0 ).s ) ◦ ft.g(t) = dt g(t) and g(0) = e. Y ] = 0 near x. t = CY + 0 −1 d Thus all parts of the integrand above are in g and so ( dt ft.s = −(FlCX )∗ CY + (FlCX )∗ (FlCY )∗ (FlCX )∗ CY.s ) ◦ ft.78 12. CY ] s t −s ds. Y ] = 0 we have [CX. reparametrize it in such a way that it is again smooth (see the beginning of this proof) . There is a unique smooth curve d g(t) in G0 satisfying Te (ρg(t) )Zt = Zt . x0 ) is a smoothly arcwise connected subgroup of G0 . ˇ with f By deﬁnition we may assume ξ = Pt(c.6. via the action of G0 on S the curve g(t) is a curve of diﬀeomorphisms on S. that for each ξ ∈ g there is a smooth mapping ˇ f : [−1. Yx ∈ Tx M and a smooth curve c in M from x0 to x. Now claim 1 can be shown as follows.s ) ◦ ft.1 = ft . Claim 3. Hol0 (Φ. CY ] = R(CX.s ds = 0 −(FlCX )∗ [CX. x0 ) ⊆ G0 . 1] × S → S such that the associated curve f lies in Hol0 (Φ. CY ] dt t Φ(FlCX )∗ R(CX.s ) ◦ ft. CY ) and (FlCX )∗ CY = C (FlX )∗ Y t t 1 + Φ (FlCX )∗ CY t 1 = CY + 0 CX ∗ d ) CY dt Φ(Flt dt = CY + 0 1 Φ(FlCX )∗ [CX. x0 ) ˇ (0) = 0 and f (0) = ξ. Since [X. Holonomy so ft.s is in g for all t and claim 2 follows. CY ) dt. s s t −s 1 −1 d ( dt ft. CY ] + (FlCX )∗ [CX. (FlCY )∗ (FlCX )∗ CY ] s s t −s −(FlCX )∗ (FlCY )∗ (FlCX )∗ [CX. so g(t) = ft and f = f1 is in G0 . So we get Hol0 (Φ. We may also suppose that Z ∈ X(M ) is a vector ﬁeld which extends c (t) along c(t): make c simple piecewise smooth by deleting any loop. It suﬃces thus to show that the Lie algebra g of G0 is contained in the Lie algebra of Hol0 (Φ.

1 Let us investigate the transition functions. The parallel transport in the last equation ﬁrst follows c from x0 to x. From its construction it is clear that the associated bundle P [S] = P ×G S equals (E.(cα )−1 . 1) Pt(cβ . S). ψβα (x) s). Pt(cβ . p.Cβ . CY ) = (FlCZ )∗ [CX. G). so claim 3 follows. For this we have to compute the Christoﬀel symbols of Φ with respect . Put xα := u−1 (0) and choose α smooth curves cα : [0. M. Y ](x) = 0 1 1 1 d = (FlCZ )∗ 2 dt2 |t=0 (FlCY ◦ FlCX ◦ FlCY ◦ FlCX ) −t −t t t 1 2 = CZ 1 d2 2 dt2 |t=0 (Fl−1 FlCY ◦ FlCX ◦ FlCY ◦ FlCX ) FlCZ ). since the fundamental group π1 (M ) is countable (by Morse Theory M is homotopy equivalent to a countable CW-complex). 1)−1 Pt(cx . x0 ) into a Lie group which we call G. p. uα : Uα → Rm ) be a locally ﬁnite smooth atlas for M such that each uα : Uα → Rm ) is surjective. Clearly ψβα : Uβα × S → S is smooth which implies that ψβα : Uβα → G is also smooth. 1) Pt(cα . Y ] = 0 near x) and then follows c back to x0 . 1] → M with cα (0) = x0 and cα (1) = xα . Construction of a cocycle of transition functions with values in G. ψα : E|Uα → −1 Uα × S) by ψα (x. α α then cx connects xα and x and the mapping (x. x0 ) is a countable group. by taking Hol0 (Φ. (ψαβ ) is a cocycle of transition functions and we use it to glue a principal bundle with structure group G over M which we call (P. then follows a small closed parallelogram near x in M (since [X. Then ψα is smooth since α x Pt(cα .(cx )−1 . Step 4. This curve is clearly nullhomotopic. 1) = FlCXx for a local vector ﬁeld Xx depending smoothly on x. Holonomy 79 and approximate it by an embedding. Lifting the connection Φ to P . Let (Uα . x0 )/ Hol0 (Φ. −t −t t t 1 where we used the well known formula expressing the Lie bracket of two vector ﬁelds as the second derivative of the (group theoretic) commutator of the ﬂows. 1] → M be the smooth curve t → u−1 (t. 1) s. So we have ξ = (FlCZ )∗ R(CX. CY ] since [X. 1] → M . x0 ) = G0 as its connected component of the identity. s) = x. Step 6.12. Pt(cα . Then Hol(Φ. M. 4) s α =: (x. s) = Pt(cx .uα (x)). t) → cx (t) is smooth α α Uα × [0. −1 ψβ ψα (x. Now we deﬁne a ﬁber bundle atlas (Uα . Step 5. where ψβα : Uβα → G. 1) s α β x = x. 1)−1 Pt(cx . For each x ∈ Uα let cx : [0. Now we make Hol(Φ. The vector ﬁelds ξ thus obtained still generate the ﬁnite dimensional vector space g.

ψα ). where γ(t) is a smooth curve in the holonomy group G. see 11.8 we have −1 ψα (Pt(c. s) is the integral curve through s of the time dependent vector ¯ d ﬁeld Γα ( dt c(t)) on S.7. By theorem 11.9 the connection Φ is thus induced by a principal connection ω on P .s). 1) Pt((cc(t) )−1 . But then we get d Γα ( dt c(t))(s) = d ¯ dt γ (t.8 the principal connection ω has the ”same” holonomy group as Φ and since this is also the structure group of P . Holonomy to the atlas of step 5. Let c : [0. s). Since by 11. Then we have −1 ψα (Pt(c. γ (t. Pt(c−1 . t)ψα (c(0). t) Pt(cc(t) . So Γα takes values in the Lie sub algebra of fundamental vector ﬁelds for the action of G on S. s)) = (c(t). 1) Pt(cα . the principal connection ω is irreducible. To do this directly is quite diﬃcult since we have to diﬀerentiate the parallel transport with respect to the curve. . Now let Γα ∈ Ω1 (Uα . ¯ where γ (t. t)ψα (c(0). 1] → Uα be a smooth curve.s. 1)s α α α = (c(t). d where dt γ(t) ∈ g. 1) Pt(c.80 12. X(S)) be the Christoﬀel symbol of the connection Φ with respect to the chart (Uα . Fortunately there is another way. From the third proof of theorem 9. γ(t). s)) = = c(t). s) = d dt (γ(t).s) d = ( dt γ(t)).

E}. T E) be a connection on E.3. S)) ⊆ C ∞ (Uαβ × S. The principal action is just composition from the right. The nonlinear frame bundle of a ﬁber bundle 13. ) to the cocycle of transition functions (ψαβ ). Thus the bundle Diﬀ{S. E}[S. p.1 the inﬁnite dimensional smooth manifold Diﬀ{S. Then we assume that the transition functions ψαβ : Uαβ → Diﬀ(S. 13. The mapping Diﬀ(S) ev : Diﬀ{S.1. Ex )}. With the methods of [Michor. T E|Ex ) : T p ◦ f = one point in Tx M and πE ◦ f ∈ Diﬀ(S. Let now Φ ∈ Ω1 (E. We can consider now the smooth action ev : Diﬀ(S) × S → S and the associated bundle Diﬀ{S. Ex ) and give it the inﬁnite dimensional diﬀerentiable structure which one gets by applying the functor Diﬀ(S. We want to lift Φ to a principal connection on Diﬀ{S. Starting from the connection Φ we can then consider ω(f ) := T (πE ◦ f )−1 ◦ Φ ◦ f : S → T E → V E → T S for f ∈ T Diﬀ{S. In the setting of 13. E}[S. S) be a ﬁber bundle and let us ﬁx a ﬁber bundle atlas (Uα ) with transition functions ψαβ : Uαβ × S → S.2. with the obvious embedding. E} := x∈M Diﬀ(S.E}×S Diﬀ{S. 13. Let us therefore assume from now on that S is compact. E). Then ω(f ) is a vector ﬁeld on S and we have . E} is a splitting smooth submanifold of Emb(S. S) as follows. Now we deﬁne the nonlinear frame bundle of (E. E}[S. E}.81 13. ev] → E as in the following diagram: pr Diﬀ{S. Let now (E. E}×S → E is invariant under the Diﬀ(S)-action and factors therefore to a smooth mapping Diﬀ{S. C ∞ (S. ev]. S) with equality if and only if S is compact. p. M. E} is in full right the (nonlinear) frame bundle of E. M. ev] = Diﬀ{S. E} = x∈M {f ∈ C ∞ (S. We consider the set Diﬀ{S. By 6. E}. E} gives it the structure of a smooth principal bundle over M with structure group Diﬀ(M ). Then the resulting cocycle of transition functions for Diﬀ{S.8 we have C ∞ (Uαβ . E} × S − − → −− Diﬀ(S) ev E ← − − Diﬀ{S. 1980] one can show that T Diﬀ{S. Lemma. S). and for this we need a good description of the tangent space T Diﬀ{S. −− The bottom mapping is easily seen to be a diﬀeomorphism.E}×S .

ηf ∈ Tf Diﬀ{S. So ω reproduces fundamental vector ﬁelds. so the two last assertions follow. Then we consider the R-linear span of all elements ρ(ξf . Let (E. ω ∈ Ω1 (Diﬀ{S.9. Lemma. But we also give a direct proof. The fundamental vector ﬁeld ζX on Diﬀ{S. S) be a ﬁber bundle with compact standard ﬁber S. Then ω(ζX (g)) = T g −1 ◦ Φ ◦ T g ◦ X = X since T g ◦ X has vertical values. E}[S.6. and their curvatures are related as in 11. Proof. This follows directly from 11. The proof is obvious. where ξf . E} admits a global parallel transport. ev] coincides with Φ. E} of its parallel transport is the global parallel transport of the lift of the connection. X.8 and 11. E} for X ∈ X(S) is given by ζX (g) = T g ◦ X. The holonomy groups and the restricted holonomy groups are equal as subgroups of Diﬀ(S). Then by the Diﬀ(S)-equivariance of ρ the vector space hol(ω) is an ideal in the Lie algebra X(S). Which one is the right holonomy Lie algebra? . ω]X(S) be the usual X(S)-valued curvature of the lifted 2 connection ω on Diﬀ{S. Each principal connection on Diﬀ{S. ηf ) in X(S). Then connections on E and principal connections on Diﬀ{S. Then we have ((rh )∗ ω)(f ) = ω(T (rh )f ) = ω(f ◦ h) = T (πE ◦ f ◦ h)−1 ◦ Φf ◦ h .8. = T h−1 ◦ ω(f ) ◦ h = AdDiﬀ(S) (h−1 )ω(f ). E}x0 . E} correspond to each other bijectively. X(S)) is a principal connection and the induced connection on E = Diﬀ{S. This follows directly from 11. and the lift to Diﬀ{S. Y ∈ Tx M . Now let h ∈ Diﬀ(S) and denote by rh the principal right action. x0 ) of 12.5. Remark on the holonomy Lie algebra.4 Theorem. Proof. The nonlinear frame bundle of a ﬁber bundle Lemma. E}. M. Note that the closure of f∗ (hol(ω)) is (a priori) larger than hol(Φ. CY ).82 13. and g : Ex) → Ex any diﬀeomorphism. E}. 13. Each connection on E is complete since S is compact. Let M be connected. let ρ = −dω− 1 [ω. 13. E} are arbitrary (horizontal) tangent vectors. Then f∗ : X(S) → X(Ex0 ) induces an isomorphism between hol(ω) and the R-linear span of all g ∗ R(CX. p.2. Let f : S → Ex0 be a diﬀeomorphism in Diﬀ{S. 13.9. and we call this span hol(ω).

is Diﬀ(S)invariant. E} which cover the identity of M .1. so it is a ﬁber respecting diﬀeomorphism. E} × Diﬀ(S) → Diﬀ{E.8 that C ∞ (Diﬀ{E. conj] which by 14. The associated bundle Diﬀ{S. The mapping A : Diﬀ{S. The ﬁrst statement has already been shown before the theorem. 14. We consider the bundle Diﬀ{E. The usual reasoning gives that Gau(E) equals the space of all smooth sections of the associated bundle Diﬀ{S. E}. E}[Diﬀ(S). with the negative of the usual Lie bracket. We equip it with the topology and diﬀerentiable structure as a space of smooth sections. Ex ). Its Lie algebra consists of all vertical vector ﬁelds with compact support on E. p. 14. p. ψαβ ) = (ψαβ )∗ (ψβα )∗ . Theorem. Gauge theory for ﬁber bundles We ﬁx the setting of section 13. with the negative of the usual Lie bracket. so it factors to a smooth mapping Diﬀ{S. In particular the standard ﬁber S is supposed to be compact. S). 14. Lemma. the derivative of such a curve is thus an arbitrary vertical vector ﬁeld with compact support. where (ψαβ ) is a cocycle of transition functions for the bundle (E. The gauge group Gau(E) = C ∞ (Diﬀ{E. Since the ﬁber S is compact we see from a local (on M ) application of the exponential law 6. E} → M . given by A(f. M. The gauge group Gau(E) of the bundle (E. see 6. if S is compact. E}) is a splitting closed subgroup of Diﬀ(E). S) is by deﬁnition the group of all principal bundle automorphisms of the Diﬀ(S)bundle (Diﬀ{S. Proof. E}[Diﬀ(S)] → Diﬀ{E. E}[Diﬀ(S). Proof.4. . E} → M ) → Diﬀ(E) is an embedding of a splitting closed submanifold. A curve through the identity of principal bundle automorphisms of Diﬀ{S. M. E} := x∈M Diﬀ(Ex .2. It has an exponential mapping which is not surjective on any neighborhood of the identity.83 14. Ex ) which bears the smooth structure described by the cocycle of transition func−1 tions Diﬀ(ψαβ .2 equals the space of sections of the bundle Diﬀ{E.1.3. g) := f ◦g◦f −1 : Ex → S → S → Ex for f ∈ Diﬀ(S. conj] is isomorphic to the ﬁber bundle Diﬀ{E. Since only the image space is inﬁnite dimensional but admits exponential mappings (induced from the ﬁnite dimensional ones) this makes no diﬃculties. E}. E}. E} → M is a smooth curve through the identity in Diﬀ(E) consisting of ﬁber respecting mappings. 14. The space of all these is therefore the Lie algebra of the gauge group. It is bijective ans admits locally over M smooth inverses.

t d Then we have dt |0 (FlX )∗ Φ = LX Φ = [X.2 and we may view the space of sections C ∞ (J 1 (E) → E) as the space of all connections. This follows from 6. Gauge theory for diﬀeomorphism groups The exponential mapping is given by the ﬂow operator of such vector ﬁelds. The action of the gauge group Gau(E) on the space of connections C ∞ (J 1 (E)) is smooth. the Fr¨licher Nijenhuis o t bracket.5.6 14. Then a section of J 1 (E) → E is just a horizontal lift mapping T M ×M E → T E which is ﬁber linear over E. Those strata whose holonomy group groups are up to conjugacy contained in a ﬁxed compact group (like SU (2)) acting on S are diﬀeomorphic to the strata of the usual ﬁnite dimensional gauge theory. Φ].84 14. Φ(T E) = V E}. it has all the properties of the latter. T E) : Ψ|V E = 0}. Remark. If S is not compact we may circumvent the nonlinear frame bundle. T E) : Φ ◦ Φ = Φ. Let J 1 (E) → E be the aﬃne bundle of 1-jets of sections of E → M .16(5).8. so it describes a connection as treated in 9.6 The space of connections. Then the action of f ∈ Gau(E) ⊂ Diﬀ(E) on Φ ∈ Conn(E) is given by f∗ Φ = (f −1 )∗ Φ = T f ◦ Φ ◦ T f −1 . The . So the space of connections is now Conn(E) := {Φ ∈ Ω1 (E. Now it is very easy to describe the inﬁnitesimal action. Theorem. Clearly this just the group of all those f which respect the horizontal bundle HE = ker Φ. and we may deﬁne the gauge group Gau(E) directly as the splitting closed subgroup of Diﬀ(E) which consists of all ﬁber respecting diﬀeomorphisms which cover the identity of M . u ∈ E. Since S is compact the canonical isomorphism Conn(E) → C ∞ (J 1 (E)) is even a diﬀeomorphism. 14. The most interesting object is of course the orbit space Conn(E)/ Gau(E). Since on each ﬁber it is just isomorphic the exponential mapping of Diﬀ(S). by 8. 14. The Lei algebra of Gau(E) consists then of all vertical vector ﬁelds on E with compact support on E. The tangent space of Conn(E) at Φ is the space TΦ Conn(E) = {Ψ ∈ Ω1 (E. We have J 1 (E) = { ∈ L(Tx M. Φ] : X ∈ Cc (V E)}. Tu E) : T p ◦ = IdTx M . We will now give a diﬀerent description of the action. We view now a connection Φ again as a linear ﬁber wise projection T E → V E. Proof.7. The isotropy subgroup of a connection Φ is {f ∈ Gau(E) : f ∗ Φ = Φ}. each one corresponding to a conjugacy class of holonomy groups in Diﬀ(S). The ”inﬁnitesimal orbit” ∞ at Φ in TΦ Conn(E) is {[X. If the base manifold M is compact then one can show that the orbit space is stratiﬁed into smooth manifolds. 14. p(u) = x}. Let X be a vertical vector ﬁeld with compact support on E and consider its global ﬂow FlX .

.14. 1989] for the starting idea. see [Gil-Medrano Michor. Gauge theory for diﬀeomorphism groups 85 proof of this assertions is quite complicated and will be dealt with in another paper.

. ) = (0. a2 . By a slight generalization of theorem 7. 0.3 (we use a Hilbert space instead of a Riemannian manifold N ) the space Emb(S. 15. 1982]. thus B(S.4 below. . 2 )/ Diﬀ(S) is a smooth manifold modeled on Fr´chet spaces which are projective limits of Hilbert spaces. Proof. 2 ) is a Lindel¨f space in the quotient topology and the model o spaces admit bump functions. 0]. . an sin θn (t). a nonlinear analogy of the inﬁnite dimensional Grassmanian. . . 2 ) is a classifying space of Diﬀ(S). The base space B(S. . a1 . . Then A1/2 (a0 . 2 ) is contractible. ) is in 2 . 2 ) of all smooth embeddings is an open submanifold of C ∞ (S. I will give a detailed description of the classifying process in 15. even odd then we connect them by te1 + (1 − t)e2 which is a smooth embedding for all t since the values are always orthogonal. 2 ) admits smooth partitions of unity. an−1 sin θn (t). e B(S. This is a variant of a homotopy constructed by [Ramadas. . . . The image under A then has values in R2N ⊂ R∞ and the expression is clearly smooth as a mapping into R2N . 2 ) we ﬁrst deform e1 through embeddings to e1 ∈ Emb(S. and equals 1 on [1. So by the general theory of classifying spaces the base space B(S. 2 ). . a1 . 2 ) := Emb(S. an cos θn (t). The total space Emb(S. 2 ) as the space of all submanifolds of 2 which are diﬀeomorphic to S.2. a0 . 1] → 2 through At (a0 . odd The same homotopy makes sense as a mapping A : R∞ × R → R∞ . . 0. . a2 . an−2 . an+1 sin θn (t). a2 . where θn (t) = ϕ(n((n + 1)t − 1)) π for a ﬁxed smooth for 2 function ϕ : R → R which is 0 on (−∞. ) 1 ≤ t ≤ n . 0. and here it is easily seen to be smooth: a smooth curve in R∞ is locally bounded and thus takes locally values in a ﬁnite dimensional subspace RN ⊂ R∞ . Given two embeddings e1 and e2 ∈ Emb(S. We consider the continuous homotopy A : isometries which is given by A0 = Id and by 2 ×[0. 1]. 0. an+1 cos θn (t). an−1 cos θn (t). . ) = (a0 . grows monotonically to 1 in [0. 2 ) and it is also the total space of a smooth principal bundle with structure group Diﬀ(S) acting from the right by composition. . ∞)]. Let 2 be the Hilbert space of square summable sequences and let S be a compact manifold. a2 . a1 . We may view B(S.1. a1 . A classifying space for the diﬀeomorphism group 15. 1 n+1 . 0. .86 15. ) is in 2 even and on the other hand A1 (a0 . . a2 . . 0. Lemma. . 2 ). ) = (a0 . . . and e2 to e2 ∈ Emb(S. a1 . .

2 ) such that (E. x. 2 ) × 2 : x ∈ N } with the diﬀerentiable structure from the embedding into B(S. x) ∈ B(S. x. The ﬁber bundle (E(S. 2 ). We choose a Riemannian metric g1 on the vector bundle V E → E and a Riemannian metric g2 on the manifold M . We can combine these two into the Riemannian metric g := (T p| ker Φ)∗ g2 ⊕ g1 on the manifold E. By the theorem of Nash there is an isometric imbedding h : E → RN for N large enough. 2 ) as the ”universal S-bundle” as follows: E(S. 2 ) → B(S. The orthonormal projection p(N. Φ) is isomorphic to (f ∗ E(S.(0x . S) and each connection Φ on E there is a smooth (classifying) mapping f : M → B(S. B(S.3 combined with 6. ξ is a vector ﬁeld along and normal to N in 2 .4. 2 ) consists of all (N. Classifying space for Diﬀ(S). dt ))Rclass = 0 where Rclass is the class curvature of Φ . 2 ) which is given by Φclass (N. 2 ) → E(S. ξ. 2 ). 2 ) = {(N. ev] = Emb(S. 2 ) ×Diﬀ(S) S which we call E(S. 2 ) with standard ﬁber S. 2 ). x. This follows from the description of the principal ﬁber bundle Emb(S. pr. 2 ) × 2 .15. 2 )[S. for which the horizontal and vertical spaces are orthogonal.t) H. We consider the smooth action ev : Diﬀ(S) × S → S and the associated bundle Emb(S. p. The tangent bundle T E(S. v) where N ∈ B(S. 2 ). f ∗ Φclass ). It will be called the classifying connection for reasons to be explained in the next theorem. 2 ) is the ﬁberwise diﬀeomorphic = which covers the classifying mapping f : M → B(S. 2 ) is then the space of all (N. M.7. ξ. S) is classifying for S-bundles and Φclass is a classifying connection: For each ﬁnite dimensional bundle (E. x ∈ N . t) = Ptclass (f ◦ c. ˜ where f : E ∼ F ∗ E(S. smooth ﬁber bundle over B(S. In view of the interpretation of B(S. 2 ) given in 7. x.3. Theorem. Since S is compact the classifying connection Φclass is complete and its parallel transport Ptclass has the following classifying property: ∗ class ˜ ˜ f ◦ Ptf Φ (c. 2 ). We then embed RN into the Hilbert space 2 and consider .x) : 2 → Tx N deﬁnes a connection Φclass : T E(S. 2 ) as the nonlinear Grassmannian we may visualize E(S. Homotopic maps pull back isomorphic S-bundles and conversely (the homotopy can be chosen smooth). 2 ). The pulled back connection is invariant under a homotopy d H if and only if i(C class T(x. p(N. t) ◦ f . Classifying spaces for diﬀeomorphism groups 87 15.x) v). and v ∈ Tx 2 such that the part of v normal to Tx N equals ξ(x). 15. v) with x ∈ N and v ∈ Tx N . v) = (N. Proof. Obviously the vertical bundle V E(S. 2 ) → V E(S.

(0x . u)) = Φclass . Then we may approximate H by smooth mappings with the same H0 and H1 .(3). 2 ) − −→ pr p M − − → B(S. ∂t 0 Pt(c. t. Then we consider the bundle H ∗ E(S. f (u)).9 we have ∂ ∂t Flt (H ∗ C class )(∂t ) ∗ H ∗ Φclass = − 1 i(H ∗ C class )(∂t ) (H ∗ Rclass ) 2 1 d = − H ∗ i(C class T(x. . t. (f. [Ebin-Marsden. Parallel transport along the lines t → (x. u) ∂ ˜ = T f .h) ) is ﬁberwise a diﬀeomorphism. 1] → M is a (piecewise) smooth curve we have for u ∈ Ec(0)) Φclass ∂ ∂t 0 ˜ ˜ ∂ f (Pt(c. equipped with the connection H ∗ Φclass .88 15. t. 15. t) with respect H ∗ Φclass is given by the ﬂow of the horizontal lift (H ∗ C class )(∂t ) of ∂t . dt ))Rclass 2 which implies the result. whose curvature is H ∗ Rclass . t. so this diagram is a pullback and we have f ∗ E(S. Then E − − − E(S. u)) = Ptclass (f ◦ c. By lemma 9. Theorem. −− f 2 ˜ f =(f. u) = 0. ∂t 0 Pt(c. ). We show ﬁrst that there exists a smooth mapping τ : Vol(S) → Diﬀ(S) such that τ (µ)∗ µ0 = µ.t) H. let µ0 be a positive volume form on S with total mass 1. 2 ) = E. where the infnite o a dimensionality of B(S. given by f (x) = h(Ex ). if they are smooth.Φ. 1970] Let S be a compact orientable manifold. h)∗ Φclass = Φ. see [Br¨cker-J¨nich.5. 2 ) does not disturb. Let us compute its action on the connection H ∗ Φclass whose curvature is H ∗ Rclass by 9. h) maps horizontal and vertical vectors to orthogonal ones. Let ∂t be the vector ﬁeld tangential to all {x} × I on M × I. 2 ). Classifying spaces for diﬀeomorphism groups 2 f : M → B(S.5. 1973]. t. If Pt denotes the parallel transport of the connection Φ and c : [0. ˜ ˜ f (Pt(c. 2 ) → M × I.T f . Proof. so Now let H be a continuous homotopy M × I → B(S. Since T (f. Then Diﬀ(S) splits smoothly into Diﬀ(S) = Diﬀ µ0 (S)×Vol(S) where Diﬀ µ0 (S) is the Lie group of all µ0 -preserving diﬀeomorphisms and Vol(S) is the space of all volume forms of total mass 1.

So the cohomology class of LXt µt in H dim S (S) is zero. Consider the space B1 (S. It carries a universal Diﬀ ω (S)-connection. Then the time dependent vector ﬁeld Xt is uniquely determined by iXt µt = ω since µt is nowhere 0. f ∗ µ0 ).7.4 with the appropriate changes to Cω. Let ft be the −1 evolution operator of Xt and put τ (µ) = f1 . It is not applicable directly for e R∞ as image space. 15.5 is that the classifying spaces of Diﬀ(S) and Diﬀ µ0 (S) are homotopy equivalent. In other words: (Embω (S. 2 ) classiﬁes ﬁber bundles with typical ﬁber S and carries a universal (generalized) connection. µ) = g ◦ τ (µ). 1982]. Classifying space for Diﬀ ω (S). which is a classifying space for the Lie group Diﬀ ω (S). 2 ) of all submanifolds of 2 of type S and total volume 1 in the volume form induced from the inner product on 2 . 2 ) by the NashMoser inverse function theorem. It is a closed splitting submanifold of codimension 1 of B(S. An easy computation shows that the inverse is given by the smooth mapping Ψ−1 (g. Theorem. because the modeling spaces are then tame Fr´chet spaces in his sense. Let S be a compact real analytic manifold. . and we have LXt µt = diXt µt + iXt 0 = dω for some ω ∈ Ωdim S−1 (S). This theorem is applicable if we use 2 as image space. 2 ) of real analytic embeddings of S into the Hilbert space 2 is the total space of a real analytic principal ﬁber bundle with structure group Diﬀ ω (S) and real analytic base manifold B ω (S. so LXt µt = µ0 − µ and consequently S LXt µt = S (µ0 − µ) = 0. So their classifying spaces are also homotopy equivalent. N ) ×Diﬀ ω (S) S → B ω (S. The proof is similar to that of 15.15. We now sketch a smooth classifying space for Diﬀ µ0 . Classifying spaces for diﬀeomorphism groups 89 We put µt = µ0 +t(µ−µ0 ). We have ∂t ft = Xt ◦ ft for a time dependent vector ∂ ∂ ∗ ﬁeld Xt on S. We want a smooth curve t → ft ∈ Diﬀ(S) ∂ with ft∗ µt = µ0 . 2 ). 15. see [Hamilton. A consequence of theorem 15. Now we may prove the theorem proper.6. which is smooth by sections 6 and 7. Now we choose ω such that dω = µ0 − µ and we choose it smoothly depending on µ by the theorem of Hodge. Then 0 = ∂t ft µt = ft∗ LXt µt +ft∗ ∂t µt = ft∗ (LXt µt +(µ = µ0 )). We deﬁne a mapping Ψ : Diﬀ(S) → Diﬀ µ0 (S) × Vol(S) by Ψ(f ) := (f ◦ τ (f ∗ µ0 )−1 . Then the space Embω (S.

algebra homomorphisms for the wedge product. M. 16. dt |0 (Flt In fact Φ is a covariant derivative on the vector bundle Ωn (p) := Ωn (E) := x∈M Ωn (Ex ) with its obvious smooth vector bundle strucver ture. Then there is a bundle atlas (Uα . We may plug n vertical vector ﬁelds Xi ∈ C ∞ (V E) into µ to get a function µ(X1 . Bundles with ﬁber volumes. p. whose standard ﬁber is the nuclear Fr´chet space Ωn (S). Proof. Let now Φ be a connection on (E. 16. e We shall need a description of the convariant derivative Φ in terms of Lie derivatives. M. If X ∈ X(M ) is a vector ﬁeld on M . and ϕα. so for the ﬁber volume µ the pullback (FlCX )∗ µ makes sense and is again a ﬁber volume on E. i. p. .x ∈ Diﬀ(S) such that (ϕα. We start with any bundle atlas (Uα .e. S) be a ﬁber bundle and let µ be a smooth section of the line bundle Λn V ∗ E → E where n = dim S. then the horizontal lift CX ∈ X(E) and X are p-related.1. . S) and let C : T M ×M E → T E be its horizontal lifting.3. Lemma. We t can now deﬁne a kind of covariant derivative by (a) Φ Xµ := CX ∗ d ) µ. Likewise we consider Φ∗ : C ∞ (Λk V ∗ E) → Ωk (E). So let j : V E → T E be the embedding of the vertical bundle and let us consider j ∗ : Ωk (E) → C ∞ (Λk V ∗ E). ψα : E|Uα → Uα × S). The other composition Φ∗ ◦ j ∗ : Ωk (E) → Ωk (E) is a projection onto the image of Φ∗ .x depends smoothly on x. .x )∗ ((ψα )−1 |{x} × S)∗ µx = µ0 . A characteristic class 16. These are algebraic mappings.5 there is a diﬀeomorphism ϕα.x chart. but we cannot treat µ as a diﬀerential form on E. T p ◦ CX = X ◦ p. So FlCX respects the vertical bundle t t t .2.90 16. such that for each x ∈ M the n-form µx := µ|Ex is a positive volume form of total mass 1. By theorem 15. Then for X ∈ X(M ) the ﬂow FlCX t of its horizontal lift CX respects ﬁbers. Xn ) on E. . and satisfy j ∗ ◦ Φ∗ = IdC ∞ (Λk V ∗ E) . Let (E. Such a form will be called a ﬁber volume on E. Let µ0 be a ﬁxed volume form on S and let µ be a ﬁber volume on the bundle E. Then ψα := (IdUα × ϕ−1 ) ◦ ψα : E|Uα → Uα × S is the desired bundle α. Thus the ﬂows are p-conjugated: p ◦ FlCX = FlX ◦p. ψα : E|Uα → Uα × S) such that (ψα |Ex )∗ µ0 = µx for all x ∈ Uα . Such forms exist on any oriented bundle with compact oriented ﬁber type.

16. a connection on Uα × S → Uα with horizontal lift is given by C α (X) = (X. Y ∈ X(Uα ). Y ](µ) − LS Γ[X. so µ : Uα → Ωk (S) is just a smooth mapping. Γ(Y )] µ −1 = LS (((ψα ) ∗ R)(X. LS ) ]µ Γ(X) Γ(Y = LS dUα Γ(X. Proof. V E). on each ﬁber separately. where we write α for Φ . S).Y ] µ = LS (XΓ(Y ) − Y Γ(X) − Γ([X. Let Φ be a connection on the bundle (E.16. Let µ be a ﬁber k-form on Uα × S. Lemma. p. Its horizontal lift is given by C α (X) = (X. Thus we have for x ∈ Uα (c) ( Φα X µ) = =d ∂ ∂t 0 Uα ((FlΓ α (X) ∗ ) (µ ◦ FlX )) t µ(X) + LSα (X) µ Γ = X(µ) + LSα (X) µ Γ where dUα is the exterior derivative of Ωk (S)-valued forms on Uα . Y ](µ) − LS Γ(Y Γ[X. M. p. . as asserted. Ωk (S)) is given by v R( Φ )(X.4. so µ : Uα → Ωk (S) is just a smooth mapping. ψα ) be a ﬁber bundle atlas for −1 (E.Y ] )µ α X (Y (µ) + LS ) µ) − Γ(Y α Y (X(µ) + LS ) µ) − [X. Y ])) µ + [LS . Then the curvature of the linear covariant derivative Φ on the vector bundle (Ωk (p). Γ(X)). M. R( = α )(X. This formula also shows that Φ ∞ X µ is C (M. Characteristic classes for general ﬁber bundles CX V E and we have j ◦ F lt = FlCX ◦j. Let (Uα . Let µ be a ﬁber k-form on Uα × S. R)-linear in X. and where LS is the Lie derivative on S. Therefore we have t 91 (b) Φ Xµ = = CX ∗ ∂ ∂ ) µ = ∂t 0 (FlCX )∗ j ∗ Φ∗ µ t ∂t 0 (Flt ∂ j ∗ ∂t 0 (FlCX )∗ Φ∗ µ = j ∗ LCX Φ∗ µ = 0 t + j ∗ iCX dΦ∗ µ. Now let (Uα . Put Φα := −1 ∗ (ψα ) Φ. ψα ) be a ﬁber bundle atlas for (E. Put again Φα := (ψα )∗ Φ. Y ))µ.Y ] µ S S S S = X(Y (µ)) + LS XΓ(Y ) µ + LΓ(Y ) X(µ) + LΓ(X) Y (µ) + LΓ(X) LΓ(Y ) µ − Y (X(µ)) − LS Γ(X) µ − LS Y (µ) − LS ) X(µ) − LS ) LS µ Y Γ(X) Γ(Y Γ(Y Γ(X) − [X. p. where L denotes the vertical Lie derivative.CY ) µ. M. Y )µ = ( α X α Y − α Y α X − α [X. Γα (X)). S) with curvature R ∈ Ω2 (E. Thus α we have for X. Y )µ = Lv R(CX. S). Y ) + [Γ(X). The easiest proof is local. a connection on Uα × S → Uα . M. where LCX is the usual Lie derivative.

5. The connection Φ on E is called µ-respecting connection if Φ µ = 0 for all X ∈ X(M ). LX µ0 = 0). ψα ) as in lemma 16.7. 16. Characteristic classes for general ﬁber bundles In this computation we used that LS : X(S)×Ωk (S) → Ωk (S) is bilinear and continuous so that the product rule of diﬀerential calculus applies. )). H n−1 (p)). Xµ0 (S)).2 the corresponding Christoﬀel forms Γα take values in the Lie algebra Xµ0 (S) of divergence free vector ﬁelds on S (i.e. We may view [iR µ] : Xx . R)-linear in X ∈ X(M ) we may glue them via a partition of unity on M . CY ))µ] ∈ H n−1 (Ex ) for X. ψα ) belonging to a ﬁber bundle atlas satisfying lemma 16. 1. 3. . Thus we have Φ µ0 = X LΓα (X) µ0 . It is described by a cocycle of transition funcx∈M H tions Uαβ → GL(H n−1 (S)).6.Y ) µ0 = 0. R R 16. Put Φα := (ψα )∗ Φ. So the vector bundle H n−1 (p) → M admits a unique ﬂat linear connection respecting the resulting discrete structure group. Deﬁnition. (1) A connection Φ is µ-respecting if and only if for some (any) bundle atlas (Uα . then for Xx and Yx ∈ Tx M the form i(R(CXx . a connection on Uα × S → Uα . But then dS iS α (X. Xµ0 (S)). We consider the cohomology class [i(R(CX. and the induced covariant exterior derivative d : Ω( M . Γ ] is an element of Ω (Uα . we have Γα ∈ Ω1 (Uα . −1 Proof. X Lemma. The last formula comes from lemma 9. Yx → [i(R(CXx . Y ∈ Tx M and the ﬁnite dimensional vector bundle H n−1 (p) := n−1 (Ex ) → M . H n−1 (p)) → Ωk+1 (M . By the ﬁrst part µ-respecting connections exist locally and since Φ ∞ X is C (M.Y ) µ0 = LS α (X. H n−1 (p)) satisﬁes d2 = 0 and deﬁnes the De Rham cohomology of M with twisted coeﬃcient domain H n−1 (p). 2. Then by 1. Its horα izontal lift is given by C α (X) = (X. which are locally constant by the homotopy invariance of cohomology.7 the local expression of the curvature Rα = (ψα )∗ R = Uα α α α 2 d Γ + [Γ . x → H n−1 (ψβα (x. −1 so by lemma 9. Xµ0 (S)). CYx ))µ] as an element of Ω2 (M . Γα (X)). (2) There exist many µ-respecting connections. CYx ))µ is closed in ω n−1 (Ex ). and this is zero for all X ∈ X(Uα ) if and only if Γα ∈ Ω1 (Uα . (3) If Φ is µ-respecting and if R is its curvature.2. We compute locally in a chart (Uα .92 16.

16. Characteristic classes for general ﬁber bundles

93

16.7. Lemma. (1) If γ ∈ C ∞ (Λk V ∗ E → E) induces a section [γ] : M → H k (p) then we have X [γ] = [ Φ γ], for any connection Φ on E. X (2) If Φ is a µ-respecting connection on E then d [iR µ] = 0. Proof. (1). For any bundle atlas (Uα , ψα) the parallel sections of the vector bundle H k (p) for the unique ﬂat connection respecting the discrete structure group are exactly those which in each local bundle chart Uα × H k (S) are given by the (locally) constant mappings: Uα → H k (S). Obviously Φ [γ] := [ Φ γ] deﬁnes a connection and in the local bunX X dle chart Uα → H k (S) we have

Φα α X γ

d (X,Γα (X)) ∗ α |0 (Flt ) γ dt = dUα γ α (X) + LSα (X) γ α , Γ =

where dUα is the exterior derivative on Uα of Ωk -valued mappings. But if [γ] is parallel for the ﬂat connection respecting the discrete structure group, so [γ α ] is locally constant, then dUα γ α takes values in the space B k (S) of exact forms. And since γ α (x) is closed for all x ∈ Uα the second summand LSα (X) γ α = dS iΓα (X) γ + iΓα (X) 0 takes values in the Γ space of exact forms anyhow. So the two connections have the same local parallel sections, so they coincide. (2). Let X0 , X1 , X2 ∈ X(M ). Then we have d [iR µ](X0 , X1 , X2 ) = =

cyclic

(

X0 ([iR µ](X1 , X2 ))

− [iR µ]([X0 , X1 ], X2 )) − iR(C[X0 ,X1 ],CX2 ) µ

=

cyclic

X0 (iR(CX1 ,CX2 ) µ)

=

cyclic

j ∗ LCX0 iR(CX1 ,CX2 ) Φ∗ µ − iR(C[X0 ,X1 ],CX2 ) µ j ∗ i[CX0 ,R(CX1 ,CX2 )] Φ∗ µ+ + j ∗ iR(CX1 ,CX2 ) LCX0 Φ∗ µ − iR(C[X0 ,X1 ],CX2 ) µ j ∗ i [CX0 , R(CX1 , CX2 )] + 0 − R(C[X0 , X1 ], CX2 ) Φ∗ µ ,

cyclic

=

=

where we used [iCX , LZ ] = i[CX,Z] , that j ∗ Φ∗ = Id, and that R has vertical values, which implies j ∗ iR(CX1 ,CX2 ) LCX0 Φ∗ µ = iR(CX1 ,CX2 ) j ∗ LCX0 Φ∗ µ = iR(CX1 ,CX2 )

Φ X0 µ

= 0.

94

16. Characteristic classes for general ﬁber bundles

Now we need the Bianchi identity [R, Φ] = 0 from 9.4. Writing out the global formula 8.9 for the horizontal vector ﬁelds CXi we get 0 = [R, Φ](CX0 , CX1 , CX2 ) = (−Φ[CX0 , R(CX1 , CX2 )] − R(C[X0 , X1 ], CX2 )).

cyclic

From this it follows that d [iR µ] = 0. 16.8. Deﬁnition. So we may deﬁne a characteristic class k(E, µ) of the bundle (E, p, M, S) with ﬁber volume µ as the class k(E, µ) := [[iR µ]H n−1 (p) ]H 2 (M ;H n−1 (p)) . 16.9. Lemma. The class k(E, µ) is independent of the choice of the µ-respecting connection. Proof. Let Φ and Φ be two µ preserving connections on (E, p, M ), and let C and C be their horizontal liftings. We put D = C − C : T M ×M E → V E, which is ﬁber linear over E, then Ct = C + tD = (1 − t)C + tC is a curve of horizontal lifts which preserve µ. In fact we have LD(Xx ) µ|Ex = 0 for each Xx ∈ Tx M . Let Φt be the connection corresponding to Ct and let Rt be its curvature. Then we have Rt (Ct X, Ct Y ) = Φt [Ct X, Ct Y ] = [Ct X, Ct Y ] − Ct [X, Y ] = R(CX, CY ) + t[CX, DY ] + t[DX, CY ] + t2 [DX, DY ] − tD[X, Y ].

∂ ∂t 0 ∂ ∂t 0

Rt (Ct X, Ct Y ) = [CX, DY ] + [DX, CY ] + 2t[DX, DY ] − D[X, Y ]. i(Rt (Ct X, Ct Y ))µ = j ∗

∂ ∂t 0

i(Rt (Ct X, Ct Y )) Φ∗ µ

= j ∗ i[CX,DY ] − i[CY,DX] + 2ti[DX,DY ] − iD[X,Y ] Φ∗ µ = j ∗ [LCX , iDY ] − [LCY , iDX ] + 2t[LDX , iDY ] − iD[X,Y ] Φ∗ µ. = j ∗ LCX iDY − 0 − LCY iDX − iD[X,Y ] + 2t(diDX iDY − 0) Φ∗ µ. Here we used that D has vertical values which implies j ∗ iDY LCX Φ∗ µ = iDY j ∗ LCX Φ∗ µ = iDY Φ µ = 0. X Next we use lemma 16.7.1 to compute the diﬀerential of the of the 1-form [iD µ] ∈ Ω1 (M ; H n−1 (p)). d [iD µ](X, Y ) = = = So we get

X [iD(Y ) µ]

−

Y

[iD(X) µ] − [iD([X,Y ]) µ]

[ Φ iD(Y ) µ − X [j ∗ (LCX iD(Y )

Φ Y iD(X) µ

− iD([X,Y ]) µ]

− LCY iD(X) − iD([X,Y ]) )Φ∗ µ].

∂ ∂t 0

[iRt µ] = d [iD µ] in the space Ω2 (M ; H n−1 (p)).

16. Characteristic classes for general ﬁber bundles

95

16.10. Remarks. The idea of this class is originally due to [Kainz, 1985]. From the point of view of algebraic topology it is not very interesting since it coincides with the second diﬀerential in the Serre spectral sequence of the ﬁbration E → M of the class corresponding to µ. 16.11. Review of linear connections on vector bundles. Let (V, p, M, Rn ) be a vector bundle equipped with a ﬁber metric g. Let be a linear covariant derivative on E which respects g. Let s = (si ) be a local frame ﬁeld over U ⊂ M of E: so si ∈ C ∞ (E|U ) and the si (x) are a linear basis of Ex for each x ∈ U . Then s deﬁnes a vector bundle chart ψ : E|U → U × Rn by the prescription ψ( si ui ) = (ui ) or ψ(s.u) = u. If s is another frame ﬁeld, we have si = j sj hj or s = s.h for a function h ∈ C ∞ (U ∩ U , GL(n)). i For the vector bundle charts we have then ψ ψ −1 (x, u) = (x, h(x).u). We also have T ψ : T (E|U ) → T U × Rn × Rn , and for the chart change we have T (ψ ◦ ψ −1 ) : T (U ∩ U ) × Rn × Rn → T (U ∩ U ) × Rn × Rn T (ψ ◦ ψ −1 )(Xx , v, w) = (Xx , h(x).v, dh(Xx ).v + h(x).w)

j The connection form is deﬁned by X si = j sj ωi (X) or s = s.ω 1 1 for ω ∈ Ω (U, gl(n)) in general and ω ∈ Ω (U, o(n)) if respects g and the frame ﬁeld s is orthonormal. If s = s .ω is the connection form for another frame s = s.h, we have the transition formula h.ω = ω.h + dh. For a general section s.u = j sj uj we have (s.u) = s.ω.u + s.du. Now we want to express the connection Φ = Φ ∈ Ω1 (E; V E), the horizontal lift C : T M ×M E → T E and the connector K = vpr ◦ Φ : T E → E locally in terms of the connection form (compare with 11.10 and following). Since for a local section x → (x, u(x)) of U × Rn → U we have X (s.u)

= s.(ω(X).u + du(X)) = K ◦ T (s.u) ◦ X by 11.12,

(T (ψ) ◦ T (s.u))(Xx ) = (Xx , u(x), du(Xx )) we see that (ψ ◦ K ◦ T (ψ −1 ))(Xx , v, w) = (Xx , ω(Xx ).v + w) (T ψ ◦ Φ ◦ T (ψ −1 ))(Xx , v, w) = (0x , v, w + ω(Xx ).v) (T ψ ◦ C ◦ (Id × ψ))(Xx , v) = (Xx , v, −ω(Xx ).v).

v Now the curvature R (X. If we assume that E is ﬁber orientable and if we consider the ﬁber volume µ on SE induced by g. 16. v. Then for the characteristic class we have k(SE.s = s. p. ω(Xx ). v. then these Christoﬀel forms take values in the Lie algebra of divergence free vector ﬁelds. We have T (U × S n−1 ) = {(Xx . The local expression of the curvature of ΦSE is again −Ω. Lemma.7 (0x .v) v. Y ) = [ X . Let us consider now the unit sphere bundle SE = {u ∈ E : g(u. Φ] ∈ Ω2 (E. If we use an orthonormal frame then ψ : SE|U → U × S n−1 is a ﬁber bundle chart. since GL(n) acts from the left on Rn . We consider again the induced ﬁber volume on the unit sphere bundle (SE. For a complex line bundle (E. Rn ) with ﬁber dimension n > 2 and a ﬁber Riemannian metric g we consider the induced ﬁber volume µ on the unit sphere bundle SE. If we put R . v) = −ω(Xx ).Y ] . M. Y ) = dU Γ(X. µ) = −c1 (E).v = 0. Characteristic classes for general ﬁber bundles The Christoﬀel form Γ of Φ with respect to the bundle chart ψ is given by 9. p. so the fundamental vector ﬁeld mapping is a Lie algebra anti homomorphism.Ω. v. S 1 ). µ) = 0. w + ω(Xx ). Y ) + [Γ(X). 1.96 16. Y ). p.v) Γ(Xx . v)) = −T (ψ)ΦT (ψ −1 )(Xx . Γ(Y )]X(Rn ) = d(−ω)(X. then we have Ω ∈ Ω2 (U. w + ω(Xx ). The curvature R = 1 [Φ. w) = (0x . u) = 1} in the vector bundle E. 0) = (0x . o(n)) if respects g and if s is an orthonormal frame ﬁeld. Y ]− [X.7 (ψ −1 )∗ R(X. gl(n)) and Ω ∈ Ω2 (U. Y ) − (−ω(X)) ∧ (−ω(Y )) = −Ω(X.v = v. . So Φ induces a connection on SE with the same Christoﬀel forms (acting on a submanifold now). w) ∈ T U × Rn × Rn : |v| = 1. 2. For a ﬁber orientable vector bundle (E.12. Then the characteristic class k(SE. Then the curvature form Ω is given by Ω = dω + ω ∧ ω. Γ(Xx . V E) can be written in terms of 2 the Christoﬀel forms by 9. since ω(Xx ) ∈ o(n). v⊥w = 0} Φlocal (Xx . −ω(Xx ). M ) with smooth hermitian metric g. v. so the induced connection ΦSE is µ-respecting. v.

This follows since H 0 (p) = M × R. This follows from H n−2 (S n−1 ) = 0. 2. Characteristic classes for general ﬁber bundles 97 the negative of the ﬁrst Chern class of E. µ) is induced by the negative of the curvature form Ω. 1. so from 16.8 the class k(SE. Proof. .16. which also deﬁnes the ﬁrst Chern class.

3. So as in 16. so ω : M → ωx0 +B 2 (S) if M is connected. 17.1. R) = 1 then there are ω-respecting connections on E and they form a convex set in the space of all connections. Proof.2. Let G : Ωk (S) → Ωk (S) be the Green operator of Hodge theory for some Riemannian metric on S and deﬁne Γ ∈ Ω1 (M : X(S)) . 17. Let again (E. I do not know whether lemma 16. For a compact Lie group G let Ad∗ : G → GL(g∗ ) be the coadjoint action. Deﬁnition.98 17. n If dim S = 2n. S) and let be the associated covariant derivative on the (inﬁnite dimensional) vector bundle Ω2 (E) := x∈M Ω2 (Ex ) over M which is given by ver Xω := CX ∗ d ) ω dt |0 (Flt as in 16. Let us ﬁrst investigate the trivial situation. The connection Φ on E is called ω-respecting if X ω = 0 for all X ∈ X(M ). By a ﬁberwise symplectic form ω on E we mean a smooth section of the vector bundle Λ2 V ∗ E → E such that ωx := ω Ex ∈ Ω2 (Ex ) is a symplectic form on each ﬁber Ex . If dim H 2 (S. We will suppose that each ﬁber Ex has total mass 1 for this ﬁber volume: we may multiply ω by a suitable positive smooth function on M . M. 17. let ωx = ωx ∧ · · · ∧ ωx be the Liouville volume form of ωx on Ex . M.1 we may plug two vertical vector ﬁelds X and Y into ω to get a function ω(X. Let ω be a ﬁberwise symplectic form on E. since it is a closed subspace of Ω2 (S) by Hodge theory. p. Lemma. Then the union of all orbits of principal type (maximal dimension) is the total space of a ﬁber bundle which bears a canonical ﬁberwise symplectic form. we see that the cohomology class [ωx ] ∈ H 2 (S) is locally constant on M . S) be a ﬁber bundle with compact standard ﬁber S. Example. So e = M × S and the ﬁberwise symplectic form is given by a smooth mapping ω : M → Z 2 (S) ⊂ Ω2 (S) such that ωx is a symplectic form for each x. Thus dM ω : T M → B 2 (M ) is a 1-form on M with values in the nuclear vector space B 2 (S) of exact forms. Since ω n = 1 for each x and since α → αn is a covering mapping H 2 (S)\0 = S x R\0 → H 2n \0 = R\0.2 remains true for ﬁberwise symplectic structures.4. Let now Φ be a connection on (E.3. Y ) on E. Self duality 17. p.

Γ(X)) ∗ d ) ω dt |0 (Flt = (dM ω)(X) + LΓ(X) ω = 0. CYx ))ωx ∈ Ω2n−1 (Ex ). CYx ))ωx ∈ n Ω1 (Ex ) and i(R(CXx . Both are d -closed forms (before applying ∗ to the second one) since the proof of 16. An ω-respecting connection Φ is called self dual or anti self dual if D ∗ [iR ω n ] = [iR ω] or = −[iR ω] respectively. Since ωx is non degenerate. since it is in B 2 (S).17. so the Christoﬀel form Γ deﬁnes an ω-respecting connection. Deﬁnition. For the general situation we know now that there are local solutions. So [i(R(CXx . For the covariant derivative connection with Christoﬀel form Γ we have Xω induced by the = (X. CYx ))ωx ] ∈ H 2n−1 (Ex ).5. 17. these ﬁt together to a smooth vector bundle homomorphism D : H 2n−1 (p) → H 1 (p). . Γ is uniquely determined by this procedure and we have LΓ(Xx ) ωx = dS iΓ(Xx ) ωx + 0 = −dd∗ G(dM ω)x (Xx ) = −(dM ω)x (Xx ). and we may use a partition of unity on M to glue the horizontal lifts to get a global ω-respecting connection. S) where ω is a ﬁberwise symplectic form on E. Let us also ﬁx an auxiliary Riemannian metric on the base manifold M and let us consider the associated Hodge -∗-operator: Ωk (M ) → Ωm−k (M ). p. where m = dim M . Let now Φ be an ω-respecting connection on (E. CYx ))ωx )] ∈ H 1 (Ex ) n and [i(R(CXx . M. Let R = R(Φ) be the curvature of Φ and let C be the horizontal lifting of Φ. Let us suppose that the dimension of M is 4 and let us consider the forms [iR ω] ∈ Ω2 (M . Let Dx : H 2n−1 (Ex ) → H 1 (Ex ) be the Poincar´ duality operator for the come pact oriented manifold Ex . As in lemma 16. The second assertion is obvious.5 we may prove that both are closed forms. Self duality 99 by i(Γ(Xx ))ωx = −d∗ G(dM ω)x (Xx ). Then i(R(CXx . H 1 (p)) and ∗[iR ω n ] ∈ Ω2 (M .6. H 2n−1 (p)) with values in the ﬂat vector bundles H i (p) from 16.7 applies without changes. This notion makes sense also for not ω -respecting connections.

CY )α is ﬁberwise locally constant by the analogue of 16. If R is the curvature of such a connection then iR(CX.7. 17. CY )α deﬁnes a closed 2-form on M with real coeﬃcients and it deﬁnes also a cohomology class in H 2 (M . X2 ) ∧ ∗(iR ω n )x (X3 . For the notion of self duality of 17.(3). Finally we sketch another notion of self duality. p. Let Φ be any connection. . R). It would be nice to investigate the characteristic class of section 16 for the ﬁberwise symplectic structure instead of a ﬁber volume form for examples of coadjoint actions. If S is connected then (X. Closely related to this are open subsets of Poisson manifolds. A typical example of a ﬁber bundle with compact symplectic ﬁbers is the union of coadjoint orbits of a ﬁxed type of a compact Lie group. i.6. .5 there is also a Yang-Mills functional if M 4 is supposed to be compact. A connection Φ on the bundle E is now called α-respecting if in analogy with 16.5. . X4 ). The Yang-Mills functional and the self duality notion exist even without this requirement. M. . e. We consider a ﬁberwise contact structure α on the ﬁber bundle (E. where dim S = 2n + 1 now. αx ∧ (dαx )n is a positive volume form on Ex .100 17. Self duality 17. where Alt is the alternator of the indices of the Xi . 17. ωrespecting connections exist if the orbit type has second Betti number 1. . But it is not clear that α-respecting connections exist. X4 ) := 1 4 Alt Ex (iR ω)x (X1 .2 we have Φ Xα := ∂ ∂t 0 (FlCX )∗ α = 0 t for each vector ﬁeld X on M .8. Z) → iR(CX. The Yang-Mills functional is then F (Φ) := M ϕ(Φ). For Xi ∈ Tx M we consider the 4-form ϕx (Φ)(X1 . S) with compact standard ﬁber S: So αx ∈ Ω1 (Ex ) depends smoothly on x and is a contact structure on Ex .

Nomizu.. Kriegl. P. . Kainz. Halperin. Foundations of Diﬀerential Geometry. 92 (1970). Peter W. Diﬀerential Geometry and Applications 1(1) (1990b). Michor. Pierre.. B. Michor. 102–163. Vicente. Free subgroups of diﬀeomorphism groups. Ehresmann. A. Lecomte. Curvature. Kriegl. The multigraded Nijenhuis-Richardson Algebra. Vanstone. Bruxelles 1950. 1987. 131 (1988). Olga.. pp. Peter W. Groups of diﬀeomorphisms and the motion of an incompressible ﬂuid. Diﬀ. Andreas. G.. Soc. Schicketanz. Pure and o Applied Mathematics.. A. The orbit structure of the action of the diﬀeomorphism group on the space of immersions. Richard S.. W. Reidel. 1951. Kriegl. 1988. Preprint (1989). P. Kriegl. A.101 References Br¨cker.. D. Michor. Gerd.. Michor. Nijenhuis. D. and Cohomology I. Springer-o a u Verlag. Acta Mathematica 165 (1990a).. Peter W. Theodor. 1–106.. Math.. A setting for global analysis. 1972. Marsden. De La Harpe. 103–121. Academic Press. J. 1980. 57–86. (1990). Springer-Verlag. 7 (1982). Charles. Kishore B. 12 (1972). Geometry Physics 6 (1989).. The inverse function theorem of Nash and Moser. Liege. 1963. Fr¨licher. Hideki. Andreas. Part I. H. Kyoto Univ. Pure Applied Algebra. Eine charakteristische Klasse f¨r F -Faserb¨ndel. Hirsch. Aspects of the theory of inﬁnite dimensional manifolds. J. Indao gationes Math 18 (1956). Marathe. The geometry of gauge ﬁelds. Janusz. P. Shiva Mathematics Series 3. Diﬀerential topology.. Quaterly Journal of Math.. A book in the early stages of preparation. Mascaro. Michor. Theory of vector valued diﬀerential forms. Colloque Topologie. A. Peter W. Heidelberg. Ebin. Francisca. GTM 33. Ray. Math. Hamilton. 338–359. handwritten manuu u script 1985. Fundamenta Math. 197–220. Remarks on the Fr¨licher-Nijenhuis bracket. Manifolds of diﬀerentiable mappings. P. Martucci. Bull AMS 72 (1966). 29–55. its universal property and application. Preprint 1990. Steve. pp. K. Foundations of Global Analysis. A. Michor. W. Connections. Linear spaces and diﬀerentiation theory. 24 (1983). Kobayashi. 571–807.. Alfred. to appear. Omori. W. Cervera. 65–222. Eells. Chichester. Wiley.. Les connexions inﬁnit´simales dans un espace ﬁbr´ diﬀerene e tiable. Morris W. Klaus. 1973.. A convenient setting for real analytic mappings. Geo. Michor. Greub. Einf¨hrung in die Diﬀerentialtopologie. Ann. J. J.E.. (Oxford). Fr¨licher. Vol... New York. J. Michor. Michor. 105–159. Peter W. Am... Werner. 1976. submitted to J. I. J¨nich. The Riemannian manifold of all Riemannian metrics. Proceedings of the o Conference on Diﬀerential Geometry and its Applications. Manifolds of smooth mappings IV: Theorem of De Rham. James. Cahiers Top. Grabowski. W. P. S.G. Math. Wiley. Bull. About interaction between Banach Lie groups and ﬁnite dimensional manifolds. Gil-Medrano. Brno 1986.. Orpington.

1987. Nijenhuis. vol. R.. 391–404. Suppl.. The existence of complete Riemannian metrics... Nomizu. Palais. Geo.. Richard S. Rendiconti Circolo Matematico di Palermo. P. Nirenberg. A global formulation of the Lie theory of transformation groups. 249–261.. Mech. Systems of vector valued forms on a ﬁbred manifold and applications to gauge theories. AMS 22 (1957). Michor. Omori.. W. 1965. Neuwirther. Berlin etc. K. Tohoku Math. Springer-Verlag. Como 1987. Newlander. J.102 References Michor. The denseness of complete Riemannian metrics. Marco. 225–226.. Dordrecht. Mem.). W. T. Modern dimension theory. Michor. Michor. 1988. J. 171–175. 1974. Springer Lecture Notes 1410. 30 (1978). Proceedings of the Conference on Diﬀerential Geometric Methods in Theoretical Physics. Lecture Notes in Math. 889–891. Martin. 427. Diﬀerential Geometry. R. Kluwer. Omori. 17 (1967). Ramadas. Proceedings of the Winter School on Geometry and Physics. Graded derivations of the algebra of diﬀerential forms associated with a connection. Werner (eds. Morrow. Deformation of Lie algebra structures. Proceedings. W. 299–313. Fourier 32 (263-276). Inﬁnite dimensional Lie transformation groups. L. Diﬀ. Ozeki. pp.. 4 (1970). J. The moment mapping for a unitary representation.. North Holland. Springer-Verlag. 89–105. A. No 3 (1990). Peter W. Hideki. Bleuler and M. preprint 1990.. H. Michor. 345–371. P.. 1982. of Math. On Banach Lie groups acting on ﬁnite dimensional manifolds. J.. II 22 (1989b). 223–250. 1251. Pseudoriemannian metrics on spaces of bilinear structures. On the space of maps inducing isomorphic connections.. K. . AMS 12 (1961). Ann Inst. Complex analytic coordinates in almost complex manifolds.. Modugno. Lecture Notes in Math. Global Analysis and Geometry 8. Gauge theory for diﬀeomorphism groups. Peniscola 1988. Knit products of graded Lie algebras and groups. 65 (1957). P. Ann. Math. A. Ser. pp.. Proc. Srni 1989. J.. Richardson. Hideki.. 1989a. Peter W. J. Nagata..

also simply E usually a ﬁbre bundle with total space E. and right translation ρ : G × S → S usually a left action M usually a (base) manifold N natural numbers N0 nonnegative integers R real numbers r : P × G → P usually a right action. short for the k × k-identity matrix IdRk . left translation λ. in particular the principal right action of a principal bundle TM the tangent bundle of a manifold M with projection πM : TM → M Z Integers . p. R) the space of smooth functions on M d : Ωp (M ) → Ωp+1 (M ) usually the exterior derivative (E. LX Lie derivative G usually a general Lie group with multiplication µ : G × G → G.List of Symbols 103 List of Symbols C ∞ (E). base M . and standard ﬁbre S FlX the ﬂow of a vector ﬁeld X t Ik . also C ∞ (E → M ) the space of smooth sections of a ﬁbre bundle C ∞ (M. S). M.

10 holomorphic. 89 cocurvature. 41 equivariant mappings and associated bundles. 96 curvature. 19 C ω -manifold structure on C ∞ (M. 5 fundamental vector ﬁeld. 13 characteristic class. 33 cocycle condition. 36 bundle of gauges. 44 G-bundle. 41 connection form (Lie algebra valued). 70 c∞ -topology. 5 linear connection. 28 Lie group. 69 linear frame bundle. 36 cohomologous. 14 convenient vector space. 50 exponential law. 36 ﬁber chart. 33 generalization of the theorem of Ambrose and Singer. 27 diﬀeomorphism group. 37 horizontal diﬀerential forms. 32. 20 F f -dependent. 52 B base space. 87. 87 classifying space. 15 associated bundle. 36 basis. 44 complete connection. 6 C ω -manifold structure of C ω (M. 47 horizontal bundle. 37. 98 A algebraic bracket. 66. 66 K kinematic tangent vector. 69 irreducible principle connection. 77 global formula for the Fr¨licher-Nijenhuis o bracket. 94 Chern class. 5 G G-atlas. 28 algebraic derivation. 10 holonomy group. 52 . 54 general Bianchi identity. 44 gauge group of a ﬁber bundle. 99 approximation property (bornological) . 98 ﬂat connections. 44 G-bundle structure. 63. 32. 83 gauge transformations. 29 o fundamental vector ﬁeld. 37 horizontal space. 81 Fr¨licher-Nijenhuis bracket. 45 homomorphism of G-bundles. 36 cocycle of transition functions. 38 frame bundle. 92 ω-respecting connection. 22 E Ehresmann connection. 48 associated vector bundles. 54 bundle. 100 µ-respecting connection. 33 f -related. 69 continuous derivation over eva . 36 ﬁber volume. 36 Bianchi identity. 70 covariant exterior derivative. 8. 65.104 Index α-respecting connection. 76 homogeneous spaces. 63 horizontal foliation. 90 ﬁberwise symplectic form. 76 holonomy groups for principal connections. 46 H Hodge-∗-operator. 71 curvature form (Lie algebra valued). 67 inducing principal connections on associated vector bundles. 38 horizontal lift. 36 C cartesian closedness. 39 classifying connection. 53. 36 ﬁber bundle. 37 horizontal projection. 37 I induced connection. 14 L Lie derivation. 19 D De Rham cohomology of M with twisted coeﬃcient domain H n−1 (p). 27 anti self dual. 11. 37 bundle atlas. 70 covariant derivative. 92 connector. 60 connection form. 65 holonomy Lie algebra. 37. N ). 97 Christoﬀel forms. 95 connection. 51 homomorphism over Φ of principal bundles. 63. 60 curvature form. 33. 92 derivations (graded) . 99 holomorphic curves. 32 horizontal vectors. N ). 7 covariant derivative on vector bundles. 33 ﬁber bundle atlas. 30 Grassmann manifold.

38 local descriptions of principal connections. 84 standard ﬁber. 60 principal ﬁber bundle homomorphism. 39 N n-transitive. 67 reduction of the structure group. 36 Stiefel manifold. 12 real analytic. 46 T tangent and vertical bundles of principal and associated bundles. 56 V vector valued diﬀerential forms. 47 regular. 37 vertical space. 62 local formulas for the Fr¨licher-Nijenhuis o bracket. 58 tangent bundles of Grassmann manifolds. 56 tangent bundles of homogeneous spaces. 12 recognizing induced connections. 58 vertical projection. 34 pullback. 52 right logarithmic derivative. N ). 36 transformation law for the Christoﬀel forms. 57 tangent vectors as derivations. 12 total space. 52 O parallel transport. 10 topologically real analytic. 39 principal (ﬁber) bundle. 36 proper. 99 smooth curves in C ∞ (M. 34 naturality of the Fr¨licher-Nijenhuis bracket. 37 R real analytic curves. 31 M manifold structure of C ∞ (M. 39 transition functions. 38 pullback of ﬁber bundles. 28 nonlinear frame bundle. 53 orthonormal frame ﬁeld. 76 Riemannian metric . 16 theorem of Hartogs. 36 U universal vector bundle. 47 principal right action. 6. 100 . 32 Y Yang-Mills functional. 18 smooth mappings. 54 tangent group of a Lie group. 44 principal bundle of embeddings.105 local description of connections. o 33 nearly complex structure. 27 vertical bundle. 58 S sections of associated bundles. 45 projection. 23 natural bilinear concomitants. N ). 17 Maurer-Cartan formula. 53 self dual. 37 vertical bundle. 24 principal connection. 20 pullback operator. 35 Nijenhuis-Richardson bracket. 23 restricted holonomy group. 35 Nijenhuis tensor. 7 smooth partitions of unity. 7 smoothness of composition. 19 space of connections. 12 real analytic mappings. 65. 81 O operational tangent vector. 44 principal bundle atlas. 14 smooth. 14 orthonormal frame bundle.