This action might not be possible to undo. Are you sure you want to continue?
Comprehensive Introduction to
DIFFERENTIAL GEOMETRY
'A
VOLUME ONE Third Edition
MICHAEL SPIVAK
PUBLISH OR PERISH, INC.
Houston. Texas
1999
ACKNOWLEDGEMENTS
I am greatly indebted to
Richard S. Palais
without his encouragement these volumes would have remained a short set of mimeographed notes
and
Donald E. Knuth
without his TEX program they would never have become typeset books
PREFACE
these "notes" truly have become a book. At one time I had optimistically planned to completely revise all this material for the momentous occasion, but I soon realized the futility of such an under taking As I examined these five volumes, written so many years ago, I could . scarcely believe that I had once had the energy to learn so much material, or even recall how I had unearthed some of it. So I have contented myself with the correction of errors brought to my atten tion by diligent readers, together with a few expository ameliorations; among these is the inclusion of a translation of Gauss' paper in Volume Aside from that, this third and final edition differs from the previous ones only in being typeset, and with figures redrawn. I have merely endeavored to typeset these books in a manner befitting a subject of such importance and beauty. As a final note, it should be pointed out that since the first volumes of this series made their appearance in 1 970, references in the text to "recent" results should be placed in context
T
he preface to the first edition, reprinted on the succeeding pages, excused this book's deficiencies on grounds that can hardly be justified now that
2.
Priface to the First Edition
HOW THESE NOTES CAME TO BE and how they did not come to be a book For many years I have wanted to wri te the Great American Differential Today a dilemma confronts any one intent on penetratOn the one hand, one can
Geometry book.
ing the mysteries of differential geometry.
consult numerous classical treatments of the sUbject in an attempt to form some idea how the concepts within it developed. Unfortunately,
a modern mathematical education tends to make classical mathematical works inaccessible, particularly those in differential geometry. On the
other hand, one can now find texts as modern in spiri t, and as clean in exposition, as Bourbaki's Algebra. But a thorough study of these books
usually leaves one unprepared to consult classical works, and entirely ignorant of the relationship between elegant modern constructions and their classical counterparts. Most students eventually find that this
ignorance of the roots of the subject has its price  no one denies that modern defini tions are clear, elegant, and precise; it's just that it's impossible to comprehend how any one ever thought of them. And even after
one does master a modern treatment of differential geometry, other modern treatments often appear simply to be about totally different subjects. Of course, these remarks merely mean that no matter how well some of the present day texts achieve their objective, I nevertheless feel that an
introduction to differential geometry ought to have qUite different aims.
There are two main premises on which these notes are based. The first
premise is that it is absurdly inefficient to eschew the modern language of manifolds, bundles, forms, etc. , which was developed precisely in order to rigorize the concepts of classical differential geometry. Rephrasing everything in more elementary terms involves incredible
x
Priface to the First Edition
Tbe work
contortions which are not only unnecessary, but misleading.
of Gauss, for example, which uses infinitesimals throughout, is most naturally rephrased in terms of differentials, even if it is possible to rewrite it in terms of derivatives. For this reason, the entire first volume of these notes is devoted to the theory of differentiable manifolds, the basic language of modern differential geometry. This
language is compared whenever possible with the classical language. so that classical works can then be read. Tbe second premise for these notes is that in order for an introduction to differential geometry to expose the geometric aspect of the subject, an historical approach is necessary; there is no point in introducing the curvature tensor without explaining how it was invented and what it has to do with curvature. I personally felt that I could never acquire
a satisfactory understanding of differentiable geometry until I read the original works. The second volume of these notes gives a detailed exposition of the fundamental papers of Gauss and Riemann. Gauss I work is now available in English (General Investigations of Curved Surfaces; Raven Press). There are also two English translations of Riemann I s work, but I have provided a (very free) translation in the second volume. Of course, I do not think that one should follow all the intricacies of the historical process, with its inevitable duplications and false leads What is intended, rather, is a presentation of the sUbject along the lines which its development might have followed; as Bernard Morin said to me, there is no reason, in mathematics any more than in biology, why ontogeny must recapitulate phylogeny. When modern terminology finally is introduced. it should be as an outgrowth of this (mythical) historical development. And all the major approaches have to be presented. for they were all related to each other, and all still play an important role.
Priface to the First Edition
At this point I am reminded of a paper described in Littlewood I S Mathematician I s Miscellany. The paper began "The aim of this paper is to prove ... " and it transpired only much later that this aim was not achieved (the author hadn It claimed that it was). What I have outlined
xz
above is the content of a book the realization of whose basic plan and the incorporation of whose details would perhaps be impossible; what I have written is a second or third draft of a preliminary version of this book. I have had to restrict myself to what I could write and learn about within the present academic year, and all revisions and corrections have had to be made within this same period of time. Although I may some day be able
to devote to its completion the time which such an undertaking deserves, at present I have no plans for this. Consequently, I would like to make these notes available now, despite their deficiencies, and with all the compromises I learned to make in the early hours of the morning. Tbese notes were written while I was teaching a year course in dif ferential geometry at Brandeis University, during the academic year
a few graduate students. Most of them were familiar with the material in Calculus � Manifolds, which is essentially regarded as a prerequisite. An acquaintance
196970.
The course was taken by six juniors and seniors, and audited by
More precisely, the complete prerequisi tes are advanced calculus using linear algebra and a basic knowledge of metric spaces.
with topological spaces is even better, since it allows one to avoid the technical troubles which are sometimes relegated to the Problems, but I tried hard to make everything work wi thout it. The material in the present volume was covered in the first term, except for Chapter 10. which occupied the first couple of weeks of the second
necessary to take rest cures of nearly a week after completing Chapters 2,
3, and 7.
term, and Chapter 11, which was not covered in class at all. We found it
The same material could eaSily be expanded to a full year course
whose expert knowledge saved me innumerable hours of labor. f or otherwise the second half of these notes would not have been written. grateful to the class for keeping up wi th my accelerated pace.XlI Priface to the First Edition in manifold theory with a pace that few would describe as excessively I am leisurely. I am also extremely grateful to Richard Palais. .
. . . 6 20 2. Orientation . . Vector bundles . . and derivations Vector fields . . MANIFOLDS Elementary properties of manifolds Examples of manifolds Problems CHAPTER . . . . .. . . . CHAPTER I. . . . . . . . . . . . and on what pages. . . . Immersion theorems Partitions of unity Problems CHAPTER 27 31 35 40 42 50 53 3. . Xlll 63 67 71 75 77 82 84 89 95 . . . . THE TANGENT BUNDLE The tangent space of �n The tangent space of an imbedded manifold . The tangent bundle of a manifold Equivalence classes of curves. the listing for each chapter gives some indication which topics are treated. .. . . . I .TABLE OF CONTENTS Although the chapters are not divided into sections. Equivalence of Tangent Bundles Problems . Coo functions Partial derivatives Critical points . . DIFF ERENTIAL STRUCTURES COO structures . Addendum. .
1 35 1 39 143 148 150 1 53 1 64 1 67 1 69 CHAPTER 6. Brackets . 2. Differential Equations Parameter Curves in Two Dimensions .. TENSORS The dual bundle The differential of a function Classical Multilinear functions Covariant and contravariant tensors Mixed tensors. . . The local flow Oneparameter groups of diffeomorphisms Lie derivatives . INTEGRAL MANIFOLDS Prologue. DIFF ERENTIAL FORMS Alternating functions The wedge product . Frobenius integrability theorem Global Theory Problems 179 190 194 198 CHAPTER 7. Addendum Addendum Problems I. . classical integrability theorems Local Theory.CHAPTER XlV Contents 4. . Problems 20 1 203 207 210 215 218 225 227 . V ECTOR FIELDS AND DIFF ERENTIAL EQUATIONS Integral curves Existence and uniqueness theorems . . Differential of a form Frobenius integrability theorem (second version) Closed and exact forms The Poincare Lemma . . . . Forms . . . and contraction Problems versus 1 07 109 III modern terminology 1 15 1 17 121 127 CHAPTER 5. . . .
30 1 308 312 316 323 334 341 344 348 CHAPTER 1 0. Stokes' Theorem . . . . . INTE GRATION Classical line and surface integrals Integrals over singular kcubes The boundary of a chain Stokes' Theorem . . . . Integrals over manifolds Volume elements . . Geodesic completeness .. de Rham cohomology Problems . LIE GRO UPS Lie groups .Contents CHAPTER xv 8. Left invariant vector fields Lie algebras .. . .. . . RIEMANNIAN METRICS Inner products . .. . . . . 239 246 248 253 256 258 261 263 283 CHAPTER 9. . The equations of structure . . Riemannian metrics Length of curves The calculus of variations The First Variation Formula and geodesics The exponential map . . . . . Oneparameter subgroups The exponential map Closed subgroups Left invariant forms Biinvariant metrics . . Problems 37 1 374 376 379 380 382 384 39 1 394 400 402 406 . .. Tubular Neighborhoods Problems . . . Addendum. Subgroups and subalgebras Homomorphisms. . .
· · To Chapters Problem . M aye rVietoris sequence for compact supports The exact sequence of a pair Poincare Duality . . . 9. . .. 477 · 481 . . 10 . . . . PoincareHopf Theorem Problems APPENDIX A To Chapter Problems Problems To Chapter To Chapter 419 424 426 428 430 432 439 442 446 450 453 I . 459 467 47 1 472 473 474 475 NOTATION INDEX . . The Thorn class · · · · · · · · · . The Euler characteristic. .CHAPTER XV! Contents I I. . . . . . . . 2 6 7. . . . . . . . INDEX . . EXCURSION IN THE REALM OF ALGEBRAIC TOPOLOGY Complexes and exact sequences The MayerVietoris sequence Triangulations . . . Index of a vector field .
Comprehensive Introduction to DIFFERENTIAL GEOMETRY A VOLUME ONE .
.
y) = �)yi .xn) with each E �. Clearly. and we shall almost never U to be the entire open ball since an open ball in �n is homeomorphic to �n. [If you know anything about topological spaces. This example immediately suggests the next: any open . we shall assume that it has the "usual metric" T all ntuples he nicest example of a metric space is Eu�lidean n�space ffi. then there is some neighborhood U of U is homeomorphic to �n .xi)2 . supplementing various chapters. x' unless another metric is explicitly suggested.CHAPTER 1 MANIFOLDS numbers. x and some integer lent metric (one which makes it homeomorphic to �n with the usual metric). which should be consulted if one allows a manifold to be nonmetrizable. ric with which mention it. .n as a metric space. in this case we can take Indeed. . where � is the set of real x = (xl. for each � n we can take U to be all of �n. =0 [=1 we will interpret �o as M with the following n=::O such that x E M. �n supplied with an equiva M is endowed plays almost no role. you can replace "metric space" by "topological space" in our definition. d(x. For n the single point A manifold is supposed to be "locally" like one of these exemplary metric spaces �n. is also a manifold. To be precise. . consisting of . of course. just �n itself. this new definition allows some pathological creatures which are not metrizable and which fail to have other properties one might carelessly assume must be possessed by spaces which are locally so nice Appendix A contains remarks. Whenever we speak of ffi. ] The second simplest example of a manifold is an open ball in �n.n. a hasty recollection of the definition shows that anything homeomorphic to a manifold is also a manifoldthe specific xE The simplest example of a manifold is. a manifold is a metric space property: If 0 E �. .
and thus to �n. and U is a neighborhood of x (U contains V with x E V) which is homeomorphic to �n by a homeomor phism ¢: U � �n. of course. there is an open ball W with ¢(x) E We ¢(V). then ¢(V) C �n is an open set containing ¢(x). Before proceeding to examine other examples. quite naturally. ConseIf is a point of a manifold some open set x quently. homeomorphic to W. though by no means all. which constitute mOst of this chapter. Exercising a mathematician's penchant for generalization.2 subset Chapter 1 V of �n is a manifoldfor each x E V we can choose U to be some open E U C V. . some preliminary remarks need to be made. ball with x T u we immediately announce a proposition whose proof is left to the reader: An open subset of a manifold is also a manifold (called. it is. and thus open in M. the set ¢1 (W) is open in V. an open submanifold of the original manifold). Thus x E ¢l(W) C V C U. The open subsets of �n already provide many different examples of manifolds (just how many is the subject of Problem 24). This compli cated little argument just shows that we can always choose the neighborhood U in our definition to be an open neighborhood. Since ¢: V � �n is continuous. M.
by the way.y. Theorem 1 is called "Invariance of Domain" . then V C �n is open and f: V � �n is oneone and continu f(V) c �n is open. left to the reader as an that n may depend on the point We next turn our attention to the integer 11 appearing in OUf definition. given can first replace each all x.Manifolds With a little thought. (It follows that f(V) is open for any open V C V.y E Mi.) 1. if M C �3 is O} U {(x. treatment may be found in Newman. we di with an equivalent metric Ji such that Ji(x . I ). is an unnecessarily complicated device for producing one manifold from two.z) : x = M={(x. then there would x. and f is a homeomorphism. U M2 .z) : z = M. An oldfashioned. Homology Tlteor)1 and Massey. stated here without proof' ous. THEOREM If . But to prove this. and M2. with metrics d. 0 and z=I} then we can choose n=2 for points in M. y) < 1 for = min(di. . 3 V must be open. Singular Homology Them]. for example. Notice V in our definition was not open). sO fI is continuous. The proof that the neighborhood easy exercise (it is also easy to see that if Theorem be an example where the V in our definition must be open is a simple deduction from Invariance of Domain. For example. we need the following theorem. Topology ojPla ne Sets. for it implies that the property of being a "domain" (a connected open set) is invariant under oneone continuous maps into �n. example.y. and d2 . The quickest routes arc probably pro vided by Vick. in fact. * All proofs require some amount of machinery. it begins to appear that. we can define di or J di= i 1 + di M. = 1 were false. In general. and 11 = 1 for points in M2 • This M. but pleasantly geometric.
In particular.". uncountable discrete space is not acompact (it cannot be written as a countable ::I 111. This construction can be applied disjoint union of the metric spaces Mi. and M2 are disjoint. which is a Odimensional manifold. sO is any discrete space {o ifx=y if x # y. they can be replaced by new sets M. The further deduction. Although different Il'S may be required at different points of a manifold it would seem that only one M. n can work at a given point x E M. that the n of our definition x E M. We will often need to know that this is the only way in which acompactness can fail to hold. the resulting metric space is called the is a manifold. however. In the new space M. A disjoint union of manifolds M.y)= I = {�i(X'Y) d on M = M. M is clearly a manifold also. It is convenient to refer to the manifold M as Mn when we want to indicate that M has dimension n. In these examples. For the proof �n is not homeomorphic to �m when of this intuitively obvious assertion we have recourse once again to Invariance of Domain. If M. A manifold has dimension n or is ndimensional or is an nmanifold if it has dimension 11 at each point. then X is acompact. as we see by taking a disjoint union of uncountably many manifolds homeomorphic to ffi. if not. for if n > 111.m into 2. we note that 11 �n. is left to the reader. an union of compact subsets). This unique n is called the dimension of M at x. Consequently. the manifold is not connected. locally compact metric space.4 Then we can define a metric Chapter 1 d(x. both M. As a first step. Y E M otherwise i to any number of spaceseven uncountably many. If X is a connected. then there is a oneone continuous map from ffi. a nonopen subset of is unique at each Consider once more a discrete space. THEOREM.y) (we assume that which are). The same phenomenon occurs with higherdimen sional manifolds.y):::r} . and M2 are open sets. U M2 by if there is some i such that x. and M2 are manifolds. since a space with one point is a manifold. defined by the metric d(x. PROOF For each x E X consider those numbers ball r > ° such that the closed {Y E X:d(x. The only compact subsets of such a space are finite subsets.
Now the closed ball B of radius �r(y) and Let that . .y) sn} . � Z1.Y) S r + d(x"x2)}. then X is acompact.y) r}. converges to some point Y E A. > 0 is an interval. Let A' be the union of all closed balls of radius r(y) and center y. 00 The triangle inequality implies that {y E X: d(xj.d(xj..Manifolds 5 is a compact set (there is at least one such r > 0. and X2 gives (2) � is continuous. some subsequence of the Yi. then for each all such x E X define rex) to be onehalf the least upper bound of X= r. If. The set of all such. proof is as follows.x 2)} C {y EX: d(x2 .. which we might as well assume is the sequence itself. The so the function r: X consequence Suppose A C . this set includes all r > 0. This has the following important X is compact.Z3. Then A' is also compact. since If not. {y E X: d(x"y) S r .Y) S r} C {y E X:d(x2. for some x. For each i there is a Yi E A such Zi is in the ball of radius r(Yi) with center Yi. Since A is compact.Z2. be a sequence in A'. . so that which implies that S (I) Interchanging x. for all y E A. since X is locally compact). n=1 U {y E X:d(x.
:. It is also closed. so x E A. Yi) r (Yi)} are contained in B. r(y) =:: . Since eventually the closed balls Chapter 1 ? Yi Y and since the function r is continuous. the limit point is actually in the closed ball of radius r (y) and center Y (Problem 10). . The only connected Imanifolds are the line � and the circle. and consequently some subsequence converges. Thus A' is compact. To see this. � I = = < By (I). y) �r(x). After this hassle with point·set topology. Their union A is clearly open.2d(x y) > rex) � . and A # 0 is ope and closed. or I ·dimensional sphere. . we present the longpromised exam ples of manifolds. Since X is connected. This shows that if y E A n ) then x E A n'.6 center y is compact. {y E X : dey. So the sequence Zi is eventually in the compact set B. Then there is some Y E A with d(x. �r(x) �r(x) 2 > d (x. Moreover.y). I rex) . it must be that X A. S I.O) = I }.  3 = 3 II which i s acompact. Now let Xo E X and consider the compact sets A {xo} An+1 An'. defined by Sl = = {x E �2: d(x. suppose that x is a point in the closure of A.
The point (0.2Jrl. and it is also homeomorphic to a certain subset of �3 which is what most people have in mind when they speak of . then MI x M2 is an (nl +n2) manifold. . on [0. an appropriate rotation of �2).) The function g: (Jr. There is another way to prove this. is our first example of a compact 2manifold or surface. 2Jr) � Sl defined by J(e) = (cose. is also a homeomorphism. I) may be taken care of similar ly. . .. We will often denote the point (cose. it is even continuous. It is ob viously homeomorphic to a subset of �4.sine) is a home omorphism. or it suf fices to note that Sf is "homogeneous"there is a homeomorphism taking point into any other (namely.!2. while Sl x Sl is commonly called "the torus". Considerations similar to these now show that the nsphere S" = is an nmanifold. I)} � � x {I} explicitly. I) onto the line � x {I} C � x �.(Of course. is a homeomorphism of Sl . I)} onto � x {I}: this is proved most simply by calculating P: Sl . by projecting onto � x {I}. The 2sphere S2.•   the above diagram.. defined by the same formula.{CO. commonly known as "the sphere". though not oneone. From these few manifolds we can already construct many others by noting that if Mi are manifolds of dimensionlli (i = 1.sine) E Sl simply by e E [0.{CO.. it is always necessary to check that use of this notation is valid. I) .Manifolds 7 The function J: (0. O) = I} Sl x . 2).. In particular .Jr) � Sl. together with J it shows that Sl is indeed a manifold.. The pro jection P from the point (0. illustrated in (0. X Sl is called the ntorus.!<. better suited to generalization.2]'[]. n times {x E �n+1 : d(x.
). The next simplest compact 2manifold is the 2holed torus. a spare homeomorphic to a torus with a hole ('ut out.y. the latter of which is also homeomorphic to the annulus. To provide a mOl'c explicit description of the 2holed torus. The resulting surface.. has cOlnponents homeomorphic either to the torus or to the cylinder S' x �. it is easiest to begin with a "handle". the region of the plane contained between two concentric circles. we throw .8 Chapter 1 "the torus": This subset may be obtained by revolving the circle {(O. more precisely.:) E �3 : J' > OJ.z) E �3 : (y_I)2 + Z2 al"Ound = 1/4} contained in {CO. called a surface of revolution.
. . _____ .. C3EE1. : : . . which remains in OUf handle. The nholed torus may be obtained by repeated applications of this proce dure. and then identifying points on the boundary of the i. : :. The 2holed torus may be obtained by piecing two of these together.9 away all the points on one side of a certain circle. " I . " .h handle with corresponding points on the i. " . I ' . and which will be referred to as the boundary of the handle.h boundary piece of the sphere. '83 .. " I . _____ .. _____ . ' " :: '=' There is one 2manifold of which most budding mathematicians make the acquaintance when they still know more about paper and paste than about . It is homeomorphic to the space obtained by starting with the disjoint union of 11 handles and a sphere with n holes. . :: . it is also described as the disjoint union of two handles with corresponding points on the boundaries "identified".
I.. .1) � �3 defined I(e.+ 1 cos � sine. .2".) x ( . With our recently intro duced terminology. and ( I . . . as investigation of the paper model will show.�cose.) x [I . this boundary is homeomorphic to a circle. I ) x (.2".( 2sine . cos � sine . I) I) ° 1  � • . " . If we define Ion [0. 1 sin � ) .1. we obtain the Mobius strip with a boundary.l) = 2 cos e + 1 cos circle of radius 2 . I) with (0. which you "make" by giving a strip of paper a half twist before pasting its ends together. This can be described by analytically as the image in �3 of the function I: [0. "identified".10 � Chapter 1 metric spacesthe famous Mobius strip. I) instead . We wish to identify each point x E S 2 with .  o � . y ''''e have not yet had to make precise this notion of "identification". but our next example will force the issue. the Mobius strip can also be described as [0.:e) 2cosB � . not to two disjoint circles.
This just means that the open sets of 11'2 are of the form J(V) where V C S2 is an open set with the additional important property that if it contains p it also contains . but we can easily say what the open sets will turn out to be (and this is all you need to know in order to check that 11'2 is a surface). so that [p) = [p). for which J(p) = J(q) implies p = ±q. The points of 11'2 are defined to be the sets {p. p} for p E S2.Manifidds 11 its antipodal point x E S2. We thus have a map J: S2 � 11'2 given by J(p) = [p). indeed. 11'2.p. . there is no subset of �3 which represents it adequately. We will denote this set by [p) E 11'2. The space which results. the projective plane. We will postpone for a while the problem of defining the metric giving the distance between two points [p) and [q). is a lot harder to visualize than previous examples. The precise definition of 11'2 uses the same trick that mathematicians always use when they want two things which are not equal to be equal. A subset V C 11'2 will be open if and only if JI (V) C S2 is open.
l) (0. J: [0. [(0. which is homeomorphic to the disc D2 = x E �2 : d(x.I). { 0) ::: I}. this is the same as identifying points on the sides of a square according to the scheme shown below (points on sides with the same label are identified in such a way that the heads of the arrows are identified with each other).1)).I)) { res. y)plane anyway.1) (1. since they are identified with points above the (x.(I. I) (s. so that the open sets of M are of the form J(V) where V is open and contains whenever it contains for = or 1. The dotted lines in this picture are the key to understanding 1l'2 If we distort the  .I)I)) ifs=OorI.12 Owpter 1 In exactly the same way. we can make things easier for ourselves by first throwing away all points of S2 below the (x.1) (1.1)) [(1. V (s.1) s 0 all points E x denoted by or x � M given by if = V To get an idea of what 1l' 2 looks like.I)}.1) (s. JI (V) [0. 1) (s. we could have defined the points of the Mobius strip M to be together with all sets There is a map and C M is open if and only if c x is open. Squaring things off a bit.1) {(O. and we must identify each P E Sl with p E Sl . 1 J«s.1) (1. This leaves the upper hemisphere (including the bounding circle). )')plane. s # 0.
in reverse direction. which is a single circle). at the upper left.. in other words. we are left with two pieces which can be rearranged to form something homeomorphic to a disc. is to be identified with the same thing at the lower right. a little experimentation wil1 convince you that this cannot be done (without having the two pieces of doth pass through each other).. the dotted line.ffi.. Th us to make a model of p2 we just have to sew a circular piece of cloth and a cloth Mobius strip together along their edges. by identifying points on the boundary and points on the boundary of the disc. al though not homeomorphic to 11'2 . can still be described mathematically in terms ( B 1 1 I . we obtain a Mobius strip with B l " " A " a boundary (namely. 3 whose image is this subset. If this Mobius strip is removed. moreovel. The projective plane is thus obtained from . There is dearly a continuous function f: p2 ?.Manifolds 13 region between them a bit w e see that the front part o f B followed by the back part of A. � " ' A B the disjoint union of a disc and a Mobius strip with a boundary. Such a function J . The subset of �3 obtained as the union of the Mobius strip and a disc. Unfortunately. every point P E 11'2 has a neighborhood U on which J is oneone. although J is not oneone. that is.. both of which are circles. it is locally oneone. : u of p2 .
with an extra dimension to play around v·. (This is also obvious from the fact that the Mobius strip is the projective plane with a disc removed. Another topological immersion of 1l'2 in �3 can be obtained by first immers ing the Mobius strip so that its boundary circle lies in a plane...... explained in Chapter 2).) This inner circle can be replaced by a quadrilateral. . . however. this Can be done in the following way. The figures below show that the Mobius strip may be ob tained from an annulus by identifyjng opposite points of the inner circle.. .. In �4. ..rith. the disc can be added so as not to intersect the Mobius strip.. . .. . . The crosscap together with the disc at the bottom is a topologically immersed 1l'2 I��\ D'\jC � .. . We can thus say that 1l'2 can be topologically immersed in �l.. . .. . When the resulting figure is drawn up into 3space and the appropriate identifications are made we obtain the "crosscap". .. . .14 Chapter 1 is caned a topological immersion (the single word "immersion" has a more spe cialized meaning. .. . . .... although not topologically imbedded (there is no homeomorphism J from 1l'2 to a subset of �3). . .. .
whidI ought to be manifolds.Manifold. b e .b.z) = J(a.+.a+b+c=±x 1+. we wi)) obtain our metric by a trick that simultaneously provides an imbedding of ll'2 in �4.:.e yz= be xz=ac xy= abo #0. so we also have hence (3) Using (2). first of all (I) II' a. it shows that things like ll'2 .z) = (yz. are. Clearly J(p) = J( pl· We maintain that J(p) = J(q) implies that p = ±q. this gives a+ b+c= ±(x+y+z).x2 +2y2+3z2). 'Ie) prove this. =±x a+b+e ' a ( ) ( ) .b.e).xz. which will Iatcr be used quite often. (fhose who know about topological spaces will recognize it as a disguised case of the Urysohn Metrization Theorem. and which the reader should peruse sometime before J't·acling Chapter 3. this leads to (2) Now bx y= a ex Z= a (x +y+z)2 = x2 +y2+z2 + 2(xy+xz+yz) = 1+2(xy+xz+yz). how('ver. We have. The missing metric can be supplied by an appeal to Problem 31. 15 'J'he one gap in the preceding discussion is the definition of a metric for p2 .y.y. Consider the function J: S2 � �4 defined by J(x.. Roughly speaking.:.) For the present. suppose that J(x.xy.
_ b2 (8) z (this holds even if y = b = since then z. so b (a. (x. = 0.O. Now suppose a = o.z) But y = z = 0 implies (by (I) again) that be = (±I. and we have (5) But (6) implies that yz = be 2y2 + 3 z2 2)·2 + 3z2 = 2b 2 + 3c2 (6) y2 + z2 b 2 + c2 and similarly for band e .c = ±I).O) or (O. Similarly. we obtain y = ±b. Y Now (4) gives = = 3 ±b .y. = ±c Since J(p) J('l) precisely when p ±'l. Clearly. we can define j: ll'2 � �4 b. so (5) gives = 3 . 0.' j([p) ) J(p).c) These equations clearly contradict = (O. b . = 0 or e = 0 and Thus x = (4) 0 also. 0.O). ['IJ) d (j([p)). Z = ±b. =3 _ _ y2 ) = = I I.16 Chapter 1 so x = ±a. In this case we have proved our contention without even using the fourth coordinate of f.±I. (4) also shows that the same sign holds in (7) and (8). If x then (I) would immediately give y = z = so that #0. which completes the proof. ". J('l))' = = = = . = This I11ap is oneone and we can LIse it to define the metric in p2 : ci( [p].y2 (7) 2y2 + 3 (1 y2 .±I)."ith the same sign (which comes from (3)) holding for all three equations.O. j(['l))) d (j(p).
Manifolds 17 Then one can check that the open sets are indeed the ones described above. or. For example. a !\16bius strip. The closest we can come to picturing this is by drawing a cross cap sticking on a torus. Although this can be pictured as two crosscaps joined together.xy) is a topological immersion of p2 in R 3 . it may also be obtained from the cylinder where I) x SI by identifying E I) X SI with is the reflection of x through a fixed diameter of the circle.x) [0.xz.z)) (yz. Notice that the [0. = V"ith the new surface p2 at our disposal. g([x. to a handle we can attach a projective space with a hole cut out. what amounts to the same thing. and famous. the map ll'2 � �3 defined by the first 3 components of g: j. which amounts to sewing two Mobius strips together along their boundary. x' . it has a nicer. representation. we can create other surfaces in t he same way as the nholed torus. By the way. The image in R3 is Steiner's "Roman surface". We can also join together a pair of projective planes with holes cut out. Consider the surface obtained from the square with identifications indicated below. (I.x'). (O.y.
x) �  .18 Chapter 1 identifications on the square force PI. O B B A P' P4 Rearrangement of the two parts shows that this surface is precisely two Mobius strips with corresponding points on their boundary identified. P2. (O. P3. below. one of which is shaded. so that the set {PI. form a single circle. and P4 to be identified. The description ill terms of [0. dividing the sides into thirds. P3. Turning one end of the cylinder around and pushing it through itself orients the lefthand boundary so that is directly opposite (I. to which il can then be joined. The dotted lines Pl P. P4} is a single point of our new space. forming the "Klein bottle". which separates the surface into two parts. P2. I) x Sl immediately suggests an immersion of the surface. Xl).
.rill see later that the spaces pll for even n differ in a very important way from the same spaces for odd n. We have already discussed some spaces which are not manifolds only because they have a "boundary". The set of all such x is called the boundary of M and is denoted by aM.. but they do have neighborhoods homeomorphic to an important subset of�". p} for p E S". . 11 A point in a manifoldwIthboundary cannot have a neighborhood homeo morphic to both�" and IHI" (Invariance of Domain again). but \\. A manifold in our sense is then called "non bounded". Notice that if M is a subset of�". The description of the open sets in Il'" is precisely analogous to the description for 1l'2 Although these spaces seem to It)rln a family as regular as the family sn. then there is some neighborhood U of x and some integer . then aM = 0. the Mobius strip and the disc. we will mention the t'('bted family of "projective spaces". .. we Can therefore distinguish those points x E M having a neighborhood homeomorphic to IHI". 19 Examples of higherdimensional manifolds will not be treated in nearly such d"tail.Manifold. then aM is not necessarily the same as the boundary of M in the old sense (defined for any subset of�"). Points no these "boundaries" do not have neighborhoods homeomorphic to ffi. We wi11 stick to the other terminology. we v. One further definition is needed to complete this introduction to manifolds. Projective nspace Il'" is defined as the rol1ection of all sets {p. for example..::O}.1I. but. Often. in addition to the family of nmanifolds S". indeed. the word "manifold" is used lor "manifoldwithboundary". If M is actually a manifold.:: 0 such that U is homeomorphic to either�" or IHI". a nonbounded compact manifold is calJed a "dosed manifold". If manifoldswithboundary are studied as frequently as manifolds. it becomes bothersome to use this long designation. The (closed) halfspace IHI" is defined by !\ manifoldwithboundary is a metric space M with the following property: IHI" = {(xt. If x E M.j1J sometimes use "bounded manifold" instead of "manifoldwithboundary". then all points of M will be boundary points of M. if M is a manifoldwithboundary of dimension < 11.x") E�" :x" .
d) (or any space homeomorphic to it) is called the disjoint union of the spaces X. (e) Every manifold is locally connected. and Y is homeomorphic to X if and only if Y = U. is homeomorphic to X. (b) If M is connected.11 but an arc is a oneone continuous image. 7.y E X. and X. while d(x. d) are also metrics and that they are equivalent to d (i.d) is a homeomorphism). '" 2. (a) Every manifold is locally compact. (c) X is locally connected if and only if components of open sets are open. y)== 3. A difficult theorem states that every path contains an arc between its end points. (a) If U c � is an inten'al and J: U � � is continuous and oneone.20 Chapter 1 PROBLEMS 1. Y. . then both d d!(I + d) and d min(I. then (X.. (d) A locally connected space is homeomorphic to the &. A space X is called locally connected if for each x E X it is the case that every neighborhood of x contains a connected neighborhood. The space (X. where the Y. d.. (a) A subset of an nmanifold is an nmanifold if and only if it is open.(x.d) � (X. (a) Connectedness does not imply local connectedness..) are metric spaces.e. (b) The integer n in our definition is unique for each x. with metrics d. (a) The neighborhood U in our definition of a manifold is always open.d) is a metric space. for some i.. If (X. and consequently homeomorphic to the di�joint union of its components. < I. 5. the identity map I: (X. so every neighborhood of a point in a locally connected space contains an open connected neighborhood.) 4. and d(x. Each X. where X = U. (c) A connected manifold is arcwise connected.y) = 1 otherwise. 6. are disjoint open sets and Y. 0') An open subset of a locally connected space is locally connected. and a connected manifold is pathwise connected. n Xj for i # j.y) if x. which are open submanifolds. for each i. then I is either increasing or decreasing.joint union of its com ponents. then the dimension of M at x is the same for all x E M. (b) Every manifold is locally pathwise connected. Show that if d is a metric on X. is an open subset of X.d. X. for i E I. but a direct proof of arcwiseconnectedness can be given for manifolds. (A path is a continuous image of [0.
and A is the boundary of each.{(O. then V = V V 10. (c) The map I is a homeomorphism. = {x E �n : d(x. (b) The image I( ) is open. 12. (2) If ffi. 21 V 8. I)} �t x {I} � �t explicitly for the map P on page 7.A has 2 components. (a) Give an elementary proof that �t is not homeomorphic to �n for n > L (h) Prove directly from the GeneralizedJordan Curve Theorem that �m is not homeomorphic to ffi.A (the "outside of A") is unbounded.Manifold. Every connected manifold (which is a metric space) has a countable base its topology. .1I for m ::I n. A C is homeomorphic to snI and I: � IR:II is oneone and continuous (so that J is a homeomorphism on A).. then /(inside of A) inside of I( A) .) (e) Pl'Ove Invariance of Domain. For this problem..p whenever it contains p. . . and the other (til<' "inside of A") is bounded. (b) Do the same for I: snt . and show that it is a homeomorphism.. (a) Check that the metric defined for 1l'2 gives the open sets described in the text. assume (I) (The GeneralizedJordan Cun'e Theorem) If A C �n is homeomorphic to snt. Explain how the two cases differ. 0') The analogous condition is necessary for the Mobius strip. (h) I f c �n is open. and a countable dense subset. 0') Check that 1l'2 is a surface.I)} � �nt.1I  B B is connected. (First prove c. 11. then �n . In the proof of Tlleorem 2. (a) Compute the composition I = s' . 14. Ie)!' 9. c �n is homeomorphic to Dn (a) One component of �n .O) S I}. (a) The text describes the open subsets of 1l'2 as sets of the form I(V). where V c S2 is open and contains . Show that this last condition is actual1y unnecessary. � 13.O. show that the limit of a convergent subsequence or the Zi is actually in the closed ball of radius r(y) and center y.{CO. which is discussed immediately afterwards.
and that all compact surfaceswithboundary are obtained from these by cutting out a finite number of discs. Let C c � C �2 be the Cantor set. (b) Since we can consider snl C S n. Show that �2 to the surface shown at the top of the next page. and since antipodal points in snI are still antipodal when considered as points in sn. A classical theorem of topology states that every compact surface other than S 2 is obtained by gluing together a certain number of tori and projective spaces. a hole in a hole in a hole 17.O) < J}. (a) Show that pI is homeomorphic to Sl . To which of these "standard" surfaces are the following homeomorphic? 15. we can consider pnI C pn in an obvious way. Show that pn .22 Chapter 1 16.pnl is homeomorphic to interior Dn = {x E �n: d(x. C is homeomorphic .
it will be used in Problem 24.C. A locally compact (but noncompact) space X "has one end" if for every rompact C C X there is a compact K such that C e K e X and X .K is connected. c Cz implies e(Cz) C c(CIl.{OJ does not "have one end" so�n . in such a way that C. X has exactly one end c if and only if X "has one end" in the sense of Problem 18.1 23 (a) If C c� is compact. (b) Show that�n has only one end c for n > I. (b) �n .C. the "right" one containing all numbers> some N. Let 8(X) be the set of all ends of a connected. An end of X is a function c which assigns to each compact subset C c X a lionempty component c(C) of X .{OJ is not homeomorphic to�m. If c is an end of�. or always the "right" one. Show that Xu 8(X) is a compact Hausdorff space. locally compact Hausdorff space X. then� . co(C) U {ends c: c(C) 20. 18. (c) This part requires some knowledge of topological spaces. More generally. This problem is a sequel to the previous one.C has exactly 2 unbounded components. (A) The infiniteholed torus: for all compact C. Define a topology on Xu 8(X) by dlOosing as neighborhoods Nc(co) of an end cO the sets Nc(co) = 19. � . but not if n = I. locally connected. the "left" component containing all numbers < some N. show that c(C) is either always the "left" component of� .Manifold. What is� U 8(�). (a) �n has one end if n > I. Consider the following three surfaces. Thus� has 2 ends. and�n U 8(�n) for n > I? = co(C)}.
Now draw in two more lines enclosing more holes. (b) Surfaces (A) and (C) are homeomorphic! Hint: The region cut out by the lines in the picture below is a cylinder. Jll10l0' _ �JlllUlll �Il\ fiJllL � __ I _ e Ie .24 Chapter 1 (B) The doubly infiniteholed torus: • • • (C) The infinite jail cell window: � ••• l�\ r lti r lti r lti r= JJ!11'L :]1\I0110l10l1[ :]1101 01 Ok[ �!l t J!l t J!l t _ _ _ (a) Surfaces (A) and (C) have one end. while surface (B) does not. which occurs at the left of (A). lH 1H 1\( 1\1 �1 ]l11'. and consider the region between the two pairs.
.. ...A) is homeomorphic to A. ) by goes to co at co. () () : i i. J j: M J: M x J(x ) n g: M(g(x). . and a � � which = � � x (� x � x .A and �2 . .. C)··" 24.. . . j(x) . . (b) If A c �2 is closed and totally disconnected (the only components of A are points).j. (a) If is a connected noncompact manifold. . . (a) It is possible for �2 . 25 ···· • • • C) C) () () . then e(�2 .j C) 22.e. (a) Every open subset of � is homeomorphic to the disjoint union of inter vals.A and �2 .. then (Compare with Problem 230. (b) The points inside the three surfaces of Problem 20 are homeomorphic.J(x». which is eventually in the complement of every compact set. For each n there is a subset An of � such that A/n ) consists of one point. . . (c) There are at most C nonhomeomorphic connected manifolds (where c = 2No is the cardinality of �). 23. i. � () () () () ..• • infinite swiss cheese . (b) There are only countably many nonhomeomorphic open subsets of �. that for any connected manifold there is a homeo morphism from to a subset of the countable product � x � x . . • • C') () () . We define A(n) inductively by A(I) = A' and A(n+l) = (A(n».Bare nonhomeomorphic if A and B are nonhomeomorphic closed totally discon nected sets. Hence �2 . J M M J: M j(M) M.. define Show that is closed. (c) The derived set A' of A is the set of all nonisolated points. if { n} is a sequence � co.Manifolds 21. (a) The three open subsets of �2 shown below are homeomorphic. then there is a continuous function � � such that "goes to co at co". For the purposes of this problem we will use a consequence of the Urysohn Metrization Theorem.) (b) Given a homeomorphism � � x � x .Bto be homeomorphic even though A and B are nonhomeomorphic closed subsets.
(b) If M is a manifoldwithboundary.1/2) and (I /2. then M .co) of each axis. If M C �n is a closed set and an ndimensional manifoldwithboundary. b. (e) There are c nonhomeomorphic connected open subsets of �2 25. (a.26 Chapter 1 '(d) There are c nonhomeomorphic closed totally disconnected subsets of �'. a sequence of points in C. 27.eI' C.'h derived set is {cil. and I' c J. 26. (c) The set {(a.U. . Hint: Let C be the Cantor set. .e) satisfies this equation and 0 oft D = Jb'e'+a'e'+a'b'. one can add a set A i such that its n l1. (c) If C. < e3 < . .' . Hint: Let x = be/D. is aM. is a manifoldwithboundary..b. e) E �3 : b'e'+ a'e'+alb' = abe} is the union of the Steiner surface and of the portions (co. then the topological boundary of M.e) is on Steiner's surface. . c) on Steiner's surface satisfies b'e2+a'e'+alb' (b) If = . etc. abe. i E J are the components of aM. This is not necessarily true if M is not a closed subset. (a) A manifoldwithboundary could be defined as a metric space M with the property that for each x E M there is a neighborhood U of x and an integer 11 =:: 0 such that U is homeomorphic to an open subset of lHIn.aM are manifolds. then (a. For each sequence n) < nz < . then aM is a closed subset of M and aM and M . as a subset of �n . (a) Every point (a. and e) < e.b.b.
notably 29. but the notion of a differentiable function J : M � � does not. it would seem reasonable to define J to be differentiable on U if J 0 </>1 : �n � � is differentiable. 229. We will use freely the notation and results of these chapters. including some problems. which the reader should bring along e are now ready to apply analysis to the study of manifolds. 215. The neces (x) we can expect J 0 1/1 1 to be differentiable for all J which make J 0 </>.I : �n � � is also differentiable. This is the case despite the fact that M is locally like �n. then it is not necessarily true that J 0 ". If U C M is an open set and we choose a homeomorphism </> : U � �n. 226. rather than by :n: (which will be used often enough in other contexts).CHAPTER 2 DIFFERENTIABLE STRUCTURES freshly sharpened. since Wsary tools of "advanced calculus". so that I i = Xi . if "' : V � �n is another homeomorphism. are contained in Chapters 2 and 3 of Calculus on Manifolds. where differentia bility of functions can be defined. however. and 335. 225. This is certainly not always 27 + .I : �n � �n is differentiable. we will denote the identity map from �n to �n by I.' differ entiable only if </> 0 ".. and U n V # 0. Unfortunately. On a general manifold M the notion of a continuous function J : M � � makes sense. Indeed. . 332.
�n If we insist on defining differentiable functions on any manifold. by considering the are Coo. two homeomorphisms x : V � xCV ) C �n and J' : V � reV) c �n are called COOrelated if the maps x ° y I : y(V n V) yo [ I : x(V n V) � � y(V n V) x(V n V) A family of mutually Coorelated homeomorphisms whose domains cover M is called an atlas for M. for the obvious reason that it provides a way of assigning "coordinates" to points on U. . it will suffice to consider homeomorphisms x: V � xCV) C �n onto open subsets of ffi. This make sense. VVe can even imagine a mesh of coordinate lines on U. This is preCisely what we shall do. . . It is necessary to adorn our manifolds with a little additional structure. just to provide a convenient name for the domain of x . for these homeomorphisms. and is automatically true. V ). xn (p ) to the point p E U.. we will be interested almost exclusively in functions J : �" � �n which are Coo (that is. where h : �n is a homeomorphism that is not diff erentiable. 1/! are in the collection. each component function Ji possesses continuous partial derivatives of all orders). V ) of an atlas A is called a chart (for the atlas A). We will often mention the pair (x. it makes sense. since x( V n V) and y( V n V) are open subsets of �" . the coordinates xl(p). which has "coordinates" xl(p).28 Chapter 2 � the case. sometimes we will use the words "differentiable': or "smooth" to mean Coo. instead of considering homeomorphisms from open subsets U of M onto �n. there is no way out of this impasse. we wish to select a certain collection with the property that </>01/!1 is diff erentiable whenever </>. . henceforth ad hered to almost religiousl)� is meant to encourage the casual confusion ofa point p E M with x(p) E �n. one need merely choose </> to be h o 1/!. First of all. for example. but a few refinements will be introduced along the way. if V n V = 0. Among all possible homeomorphisms from V C M onto �n. namely. . Moreover. xn (p). The use of the letters X. . A particular member (x . . where it is hard not to lapse back into the practice of denoting points by x and )'. instead of x alone./. . If V and V are open subsets of M. )" etc. . or a coordinate system on V. The only time this notation will be confusing (and it will be) is when we are referring to the manifold �n. Also. the precise nature of which is suggested by the previous discussion.
4. where 'U (the "usual COOstructure for �n") is the maximal atlas containing {I}.. .4.. Indeed. 'U differs only superficially from . we can adjoin as many such x's as we likeit is easy to check that they are all COOrelated to each other.' be thc set of all charts y which are Coorelated to all charts x E A It is easy to check that all charts in . 29 The simplest example of a manifold together with an atlas consists of �n with an atlas . then . Although (�..4.. LEMMA.. We can easily make the atlas bigger. The term actually used is x E 'U if and only if x 0 J E V.4. The advantage of this bigger atlas 'U is that the single word "chart". so .4. there is a oneone onto function J : � � � such that namely.4.n parallel to one of the axes. if U and V are homeomorphic open subsets of �n.\"3 . (and one would be foolish not to do so once and for all). one can easily construct 'U from . and it is clearly the unique maximal atlas containing A . If . about the simplest example of a Coo manifold is (�n.' is an atlas. •:.4. 'U).4.. Thus. where . V) where V contains the homeomorphism x t? . is a maximal atlas for M. PROOF.). .. V) are not the same. Another example is (�. Let . we can adjoin any homeomorphism x : U � V with the property that x and XI are COO.' are Coorelated.. when applied to this atlas.. What has just been said for the atlas {I} applies to any atlas: I . the obvious map J(x ) = x 3 • Thus (�.Differentiable 5lructureJ inverse images under x of lines in ffi.4. denotes something which must be described in cumbersome language if one can refer only to A Aside from this. V) are the sort of structures one would want to call "isomorphic". the identity J : �n � �n. or smooth manifold) to be a pair (M.. whose inverse is not Coo. together with all charts Coorelated to it. 'U) and (�.4. We now define a Coo manifold (or differentiable manifold.. 'U) and (�. is an atlas of COOrelated charts on M. is contained in a unique maximal atlas A' for M. of only one map.
. which are part of the field called "diff erential topology". . .I n {x E �n : x . It is easy to see that a diffeomorphism must be continuous. . . xn). . unique differentiable structures I on sn for 11 ::: 6. However.30 Chapter 2 "diff eomorphic": two Coo manifolds (M.I g. and speak elliptically of "the diff erentiable manifold M". conversely. we shall not come close to proving these assertions.(x) = (Xl.the "usual Coo structure for snb. 'Li). and II is clearly a diffeomorphism also. the proof for �3 would ccrtainly be too difficult for inclusion here. of course. This raises the natural question whether. Ifwe had not required our atlases to be maximal. : snI n {x E �n : x . . the projections PI and P2 from the points (0. . 0.+ I .(x) = g. < O} � �nI . . It will always be understood that �n refers to the pair (�n. but we can already describe a diff crentiable structure . This atlas may also be described in terms of the 2n homeomorphisms defined by f. the atlas for M is sometimes referred to as the diff erentiable structure for M. its inverse must also be continuous: so that a diff eomorphism is automatically a homeomorphism. 'Li). 2 A proof of the corresponding assertion for � is much harder. . which are Coorelated to P and P2.. . There are. Later (Problem 924) we will be able to prove easily that � with any atlas is diffeomorphic to (�. 0. They therefore determine an atlac. : sn . two homeomorphic manifolds are necessarily diff eomorphic. up to diff eomorphism. . we will suppress mention of the atlas for a differentiable manifold. X.A) and (N. . (perhaps most astonishing of all 4 has a diff is the quite recent discovery that � erentiable structure that is not diffeomorphic to the usual differentiable structure!) Other examples of differentiable manifolds will be given soon.::: 5 requires very difficult techniques from topology. and over 16 million on S 3 1 . In the case of spheres. . . rather then differential geometry. > O} � �n. Consequently. :B) are diffeomorphic if there is a oneone onto function I : M � N such that x E :B if and only if x 0 I E A The map I is called a diffeomorphism. . Normally. But there are 28 diffeomorphism classes of differentiable structures on S7.I ) of snI are easily seen to be Coorelated. x. .I . . the definition of diffeo morphism would have had to be more complicated.>4' on any open submanifold N of f. I) and (0. and the proof of the essential uniqueness of Coo structures on ffi.n for n .
V ) and (y.Differentiable Structure.' 31 a differentiable manifold (M. the atlas .4. If this is true for one pair of coordinate systems. Just as diffeomorphisms are analogues for Coo manifolds of homeomor phisms.4.' consists of all (x. . V ) is a diffeomorphism from V to xCV ). and hence J is differentiable if and only if J 0 XI is diff erentiable for each chart x. J y JR:m is called differentiable at p E M if y o J 0 XI is diff erentiable at x(p ) for coordinate systems (x . eren A differentiable function J : M � � refers. of course... x M2 of two diff eren . More particularly. V) with p E V and J(p) E V. a diff erentiable map is continuous. The diff erentiable structures on many manifolds are designed to make certain functions differentiable.). to the usual diff tiable structure on �. (4) a function J: M � N is differentiable if and only if each y i diff erentiable for each coordinate system y of N. It is easy to see that (2) a function J : M � �m is diff erentiable ifand only ifeach Ji : M � �m is diff eren tiabl e.. with V C N. it is easily seen to be true for any other pair. We can thus define differentiability of J on any open subset M' e M. the map y 0 J 0 XI : �n � �m is differentiable.4. 0 J is (5) a diff erentiable function J: M � N is a diffeomorphism if and only if J is oneone onto and JI : N � M is diff erentiable. Consider first the product M. there are analogues of continuous maps. as one would suspect. Clearly. V ) for M and (y. A function J : M � N is called differentiable if for every coordinate system (x . (3) a coordinate system (x. this coincides with differentiability of the restricted map JIM' : M' � N. V) for N. V) in . (I) a function J : �n � �m is differentiable as a map between Coo manifolds if and only if it is differentiable in the usual sense.
A "handle" is then a Coo manifoldwithboundary. where U does not contain . so that f l U is oneone. one needs to know that there are lots of I hem. We briefly recall here the necessary facts about such Coo functions (c. It is easy to define a diff erentiable structure on M. we first note that a Coo manifoldwithboundary can be defined in an obvious way. The collection of these homeomorphisms can then be extended to a maximal atlas. The details involved in this process are reserved for Problem 1 4. For each pair (XI. To deal with Coo functions effectively. Ca!. It is only necessary to know when a map f : lHln � �n is to be considered differentiable. Consider any coordinate system (x. UI) of coordinate systems on Mi) we construct the homeomorphism = defined by Then we extend this atlas to a maximal one. 29). The map X 0 (f I U)' is a homeomorphism on leU) C ll'n. and any two such are COOrelated. . .f. U) for sn. x M2 which makes each 7[1 diff erentiable. The existence of Coo functions on a manifold depends on the existence of Coo functions on �n which are 0 oUlside of a compact set. x M2 � Mi defined by 7[i(p" P2) Pi . there is a differentiable structure on ll'n which makes the map f: sn � ll'n (defined by f(p) [p) {p. and the two "projections" 7[i : M. pg.culus on Malfif olds. = = To obtain differentiable structures on other surfaces. A differentiable structure on the 2holed torus can be obtained by "matching" the diff erentiable structure on two handles.p if it con tains p. Similarly.pi) diff erentiable. we cali f differentiable when it can be extended to a differentiable function on an open neighborhood of lHln.32 Chapter 2 tiable manifolds MI.
e(x+I>' x E (I. e On a Coo manifold M we can now produce many nonconstant Coo func tions. e) x . j(xn ) = . il is 0 for x (4) The function g: �n � g(x) j(XI ) .I) .1) I � � (I) The function = � defined by 0 o =0 n. Similarly.Differentiable Structures 33 is Coo.. " x (e. Then there is a Coo function f : M � [0. increasing on (0. there is a Coo function k : elsewhere. The closure is called the support of f. e) and 0 elsewhere. and denoted simply by support f (or sometimes supp f)· {x : /(x) oft O} 2.:::: 0. such that f = on C and support f e U. Let C c U e M with C compact and U open. = 0 for all � defined by = o � which is positive on (0.) 1) I .. 8 > .8). it is positive on (e. 8) and 0 (3) The function I : � � � defined by is Coo. h: � � hex) { eI /x' x oft x h(n>(O) j: � � { e(xI)' . ]i[ +0 0= 1 x for E 2: 8. and � defined by /e je is Coo. LEMMA. J(x) x ¢ (1. and (2) The function i s Coo. (Compare Case 2 of the proof of Theorem 15.
so a CO manifold is just a manifold in the sense of Chapter I. where I is defined in (3). of course. Clearly it remains Coo if we extend � U� it to be 0 outside of V./) which converges in some neighborhood of a).. . and is positive on a neighborhood of p whose closure is contained in U. On C it is positive. e ) x . we could have defined C' manifolds for each . and it is convenient to agree that r < 00 < (J) for each integer . the smaller structure is the "stronger" one. The function g o x (where g is defined in (4)) is Coo on V. .. For each p E C. emellta1JI Diff ential Topolog)'_ 7 er . choose a coordinate system (x. The function J can be P P constructed for each p. . e ) for some e > o. the CO structUl"e (consisting of all homeomorphisms x: U � �n) being the largest. not just for 00" . as in Lemma I. . 2: I. . If ". x (e.34 Chapter 2 PROOF. so on C it is 2: 8 for some 8 > o. . Thus. < {3. This will not be proved here. but this atlas can always be extended to a bigger atlas of COrelated charts. Co manifolds for ". Since C is compact. P p. The converse of this trivial remark is a h ard theorem: For Q' � 1.1unkres. Of course. Let J be the extended function. J l + . •: " r = By the way.. . l:. a C� structure on M can always be extended to a Ccx structure in a unique way. A "C o function" isjust a continuous function. Then xC V) ::l (e. ). # co will hardly ever be mentioned again. Q' . One remark is in order now. the proof of Lemma 2 produces an appropriate Co filllction J OIl a Ccx lnanifold. then the charts of a maximal C� atlas are all COrelated..:::: 00. Let J = I o (f + . We can also define analytic manifolds (a function J : �n � � is analytic at a E �n if J can be expressed as a power series in the (xi . but it is unique up to diffeomot·phism. of the corresponding functions has support C U.* In f act.. The symbol ew stands for analytic. for Q' = (J) the proof * For a proof see 1\. for 0 :s. V) with V c U and x(p ) = o. 2: o. finitely many such neighborhoods cover C. + J Pl p. is Cr ifit has continuous partial derivatives up to order r ) .n ?. it is not unique. (A function J : ffi.ffi. and the sum. every Ccx structure contains a CfJ structure for each f3 > 0' . + I .
in fact it is just Notice that ifi = j ax' .. O lim Now. At this point classical notation for partial derivatives is systematically introduced.(h» . .J(a) the number J(a' . . e) � M by (or simply aI aXl = D. I = ' as an equation between functions).U).(h) = X. . � (p) = 8}� = 0 if i # j. with respect to a coordinate system (x. . then h. If we . ax' (p) = ax' p c. . ...Diff erentiable Structures 35 fails completely (and the result is falsean analytic function which is 0 on an open set is 0 everywhere).J(g (a» .(f 0 x. 0» x 0 0 x. then is the classical symbol for this partial derivative. With differentiable functions now at our disposal. i= I X. . so it is worth recalling a logical notation for the partial derivatives of a function J: �n � �. h.J(a) . . We denote by D. then this partial derivative is just If x happens to be the identity map of �n.".I ) define the curve Cj : (e. What we shall define are the partial derivatives of a dif ferentiable function J: M � �. .U) we define aJ aJ D.)(x(p» . .J(p) h ' {I (Joct!' (0). ax} h� O r J(c.(J J: M � � and a coordinate system (x. . an) . for a function Dj (f o g) (a) = L D. so it measures the rate change of J along the curve c.J(p) = aJlax' (p). . it is fitting that we begin differentiating them. which . . .. Dd (a)..I (x(p) + (0.a' + h. D. .. h The Chain Rule states that if g : �m � �n and J: �n � �.
yJ This really is a coordinate system on A in our sense..ffi. is the use of the symbols alar and alae in connection with "polar coordinates".y� . with its image being the set {r : r > O} x (0. y) is restricted to lie in the appropriate interval (eo . and this sometimes justifies the sloppiness involved in the definition of the polar coordinate system.. . y) E �2 : y = 0 and x ::: O} where r ex. 2rr). �.. e) = (r cos e . r sin e). eo + 2". y)..� .2 by p(x. often not completely clarified. many results are essentially independent of which line is deleted. y) is the unique number in (0. e (x. y) cos e (x ..).. y» . y) y = r ex. (Of course. OJ "Oaxis" � +.36 Chapter 2 Another classical instance of this notation.) with x = r ex. Jx2 + y2 and e(x. On the subset A of �2 defined by A = �2 = �2 L _ {(x . whose inverse p .I is defined simply by p .. y) = (r(x."raxis" 2" not restricted to the set A. the polar coordinate system is often + (r.% (x . One Can delete any ray other than L if e(x.I (r. y) = we can introduce a "coordinate system" P : A ?. y).2".) We havc really defined P as an inverse function. y) sin e(x.
The vector w = ( .:.. y) [r(x. y) + Dzf(x. e) = = 37 J(r cose. [r')' (p(x.y) sin e(x. of the curve C2 v·. and thus the (x.:.. y). y» ( This formula just gives the value of the directional derivative of J at (x.(x. A similar computation gives C l _ _ cos8(x. y).). .y) /8(x. y» is perpendicular to v. . y) % direction. .Y) = DI /(x. y) = DI /(r' (p(x. y). y») + Dz f(x. y). along a unit vector v = (cos e(x. y)) pointing outwards from the origin to (x. at the point (x. is just a line in this direction.I ) ( P(x. ..� / Xl' \ C2 .y» + Dzf(P' (p(X. y)[r(x . DI [ rl )2 (p(x.y» · D.e along the "eaxis".y)) .'hich is the inverse image under P of a cun. This is to be expected. sin e(x.sin e(x. y» by the Chain Rule DI /(x.) /1. "v' _ = ofa curve along the "raxis". The factor " (x. y) .". y). y) cos e(x. cos e(x. r sin e). the inverse image under P (x' Y). .DifJerentiable Structures From this formula we can compute aJlar explicitly: (f 0 p .'n9(X. sin e(x. so aJ a. y).I )(r.. DI J 0 P . y»). y) · cose(x. because CI . y) appears because this curve .
are. Note that aJ/ae is independent of which line is deleted from the plane in order to define the function e unambiguously. y) denotes the identity coordinate system of �2 We have ax lar = COs e. then on U n V we have (I) PROOF It's the Chain Rule. (J 0 y . if you just keep your cool: ar (p) = D. y) everyvvhere (thus writing an equation about functions). PROPOSITION. where (x.+ ar ax ar ay ar aJ aJ ax aJ ay = + ae ax ae ay ali ' I n classical notation. the Chain Rule would always b e written i n this way. If (x. aJ ay sin e aJ cos e.I )(y(p» a I I = D ([J 0 [ ) 0 [x 0 y. D.))(y(p)) i > = j =l I: Dj (f 0 xI )([x 0 yI )(y(p» ) . a ax ( rsm e) + / In particular. and J: M � � is differentiable. V) are coordinate systems on M. Using the notation aJlax for D1 !. U) and (y..y) times as fast as it should go in order to be used to compute the directional derivative of J in the direction w. and suppressing the argument (x. of course.38 Chapter 2 goes around a circle of that radius as e goes from 0 to 27[. so it is going r(x. etc.. so Qur formulas can be put in the form  aJ aJ ax aJ ay = . we can write the above equations as �= ae ar aJ = aJ ax aJ cos e + . [x 0 yI )j (y(p» . etc.. I t is a pleasure to report that henceforth this may always be done: 3. these formulas also tell us what axlar etc.
the nature of things. If (x. U) with p E U. g : M � �.(pl· . one can avoid what Elie Cartan has called the "debauch of indices". often with hoards of indices being summed over. •:. ay' P J= l ayi n I I: For later use we record a property of f = alaxilp: it is a "pointderivation".DifJerentiable Structures 39 = j= = :L Dj (J 0 [ I)(x(p» 1 1 j= aJ ax} :L � ( p ) . Left to the reader. the n matrix . almost always occur above and below) being summed over. a)'i p ' I thus a ax j a = (p ) ax j P . I won't use this notation because whenever I do. and the convention is to omit the L sign completelydouble indices (which by luck. and felicitous choice of notation. I 4. n x PROOF. U') are two coordinate systems on M. J= I n axj a here alay' is considered as an operator taking the function J to aJlay' . and any coordinate system (x. and because by do ing things "right". The operator taking J to aJlay'(p) is denoted by a . the operator f = alaxi lp satisfies f(Jg ) = J(p)f(g) + f(J)g(p ). ax) ayi n · Di [Xj 0 y. U) and (x'. We will often write formula (I) in the form a ay i = I: ayi axj . For any differentiable J. There are scads of for mulas in which this happens. No tice that the summation in this formula occurs for the index j. I soon forget I'm supposed to be summing. PROPOSITION.I) (y(p» At this point we could introduce the "Einstein summation convention". which appears both "above" (in axj lay') and "below" (in aJlaxj ). .:.
40 Chapter 2 is just the Jacobian matrix of x' 0 inverse is clearly XI at x(p). where v(Bn ) is the volume of Bn . is still a "small" subset of �'. b). In this case. It is nonsingular. . If I : � � �. The point p is called a critical point of I if the rank of I at p is < 111 (the dimension of the image N). it is called a regular point of f. the rank of the m x 11 matrix (::. if q ¢ I(M) .1l has "measure zero" if for every e > 0 there is a sequence B" B" B3. Recall that a set A C ffi. If p is a critical point of f. the image I(�') = � x to} C �.a nonvalue of I is still a "regular value". U) or (y. may have all points as critical points without being constant. for example. we will need some terminology. The most important theorem about critica1 points generalizes this fact. its Now if I : M n � N m is Coo and (y. I(x. however. � �.� (P») clearly does not depend on the coordinate system (x. if p is not a critical point of I.a1 [a. then x is a critical point of I if and only if j'(x) = O. . although this can happen only if I is constant On [a. V). then D I = D. of (closed Or open) rectangles with I and n=1 L v(Bn) < e. This is true. in particular. � � has all points as critical values. To state it. Other points in N are regular values. On the other hand. so I is again constant. the value I(p) is called a critical value of f. V) is a coordinate system around I(p). We want to define the same concept for a subset of a manifold. If I : �. in fact. It is possible for all points of the inten. a function I: �. thus q E N is a regular value if and only if p is a regular point of I for every p E lI (q). I = 0 everywhere. y) = x. b) to be critical points. . . which in turn depends on a lemma from Calculus on Manif olds. which we merely state. It is called the rank of I at p. To do this we need a lemma.
y l for all X. then J(A) n V = Ui J(A n Vi) n v. We can assume that A is contained in a compact set C (since �n is a countable union of compact sets). and A consists of one point from each component. (. . then x(A n V) C �n has measure ° for any coordinate system (x. LEMMA.:0 n2 Klx . (But if M is the disjoint union of uncountably many copies of �. by Lemma 6. V ) is a chart on N. Since J (Ui Vi ) is contained in the Ullion of at most countably many components of N. There is a sequence of charts (Xi . COROLLARY. This clearly implies that J(A) has measure ° if A does . A subset A of a Coo nmanifold M has measure zero if there is a sequence of charts (XI. Thus J takes rectangles of diameter d into sets of diameter ::. If (y. Lemma 5 implies that there is some K such that IJ(x) . with A c Ui Vi.J(y ) 1 ::. or has only countably many components. PROOF. it is easy to see that if A C M has measure 0. If J : M � N is a C I function between two nmanifolds and A e M has measure 0. V). j = 1 . then it follows easily from Theorem 12 that A has measure 0. y E A. PROOF. Thus y(J(A) n V ) has measure 0. it follows that J(A) has measure 0. Using Lemma 6. = y 0 J o [l(x(A n Vi» . Each set y(J(A n Vi) n V ) has measure 0. Vi). Conversely.Differentiable Structures 41 5 . y E C. Then IJ(x) . Let A C �n b e a rectangle and let J : A � � n b e a function such that I DjJi l ::. . 6. If J : �n � �n is C1 and A C �n has measure 0. then J(A) has measure o.J(y ) 1 . if this condition is satisfied and M is connected. Lemma 6 thus implies another result: 7. then A does not have measure 0). such that each set xi (A n Vi) c �" has measure 0.yl for all x. LEMMA. . then J(A) C N has measure 0. . n2 K d. •:. n 2 Klx . K on A for i. Vi) with A c Ui Vi and each set xi (A n Vi) of measure 0. n .
(2) If J has rank k in a neighborhood of p. . and its image could. . we can insure that J keeps the first k components of a point fixed. . .In. SC'T 1\. . because some k x k submatrix of (a(yi 0 J)/axi) has nonzero determinant at p . If y is the identity of �m . .42 Chapter 2 8. . then there is some coor dinate system (x. V) such that y o J o x. in Problem 824). . in terms of the rank k of J at p E M. Remark: The special case M = � n . and the case m > n is the trivial case (Problem 20). But this case is just Theorem 3. given any coordinate system )" the appropriate coordinate system on N can be obtained merely by permuting the component functions of y. XI k (a t . . and hence in a neighborhood of p. PROOF It clearly suffices to consider the case where M and N are �n . . and M has at most countably many components. Lectures 011 f'l" Diff erential Geomel1)'. for example. The stronger version ofSard 's Theorem . 9. m (a» . . states' that the critical values of a C k map J: M n � N"' are a set of measure 0 if k 2: 1 + max (n . Theorem 8 is the easy case. which we will never use (except once . a . then there are coordinate systems (x. for the present we are more interested in knowing what the image of J : M � N looks like lo�ally. contain only points with first coordinate o. O. U) around p and some coordinate system (y. These diffcomorphisms on �n and �m are clearly necessary. O) . . . (I) If J: M n � N m has rank k at p. .n. Although Theorem 8 will be very important later on. . . and then permuting the coordi nates in �m. then the critical values of J form a set of measure 0 in N. THEOREM (SARD 'S THEOREM).. 0). THEOREM. Topolog)' From the D{ff miiabtc Viewpoint or Sternberg. a n ) = (a t . . ".1i1IIor. . * For a proof. ".. . .k + t (a). ak . . U) and (y. More exact information Can be given when J actually has rank k in a neighborhood of p. If J : M � N is a C1 map between nmanifolds. which gives only local results. .14 of Calculus on Manif olds. . V) around J(p) with y 0 J 0 x t in the form yoJ0  I\1oreover. part (I) says that by first performing a diffeomorphism on �n. a n ) = (a t . since J may not even be oneone on ffi. N = �m i s equivalent t o the general theo rem. . .k x to} c ffi. It should be noted that J must have rank 2: k in some neighborhood of p.t (a t . .
. ....an) = (a'. 43 fORn) JRn.ak. . . __ ( a(v. . . .. . (I) Choose some coordinate system u around p. . k hence ur (q) = a r = k + I ... . k r = k + I. (I) det C au Define xa = ya 0 f xT = uT a = I." o a.. fl (p») . . n.an). . (a'.DifJerelltiabLe Structures In part (2) we must clearly allow more leeway in the choice of Y. . 0 /» ) = ax} . .... . . k . has the form in (I). q = x'(a'. . the lower square in the matrix . ).(q) = a. x0 u (p). . . Condition (I) implies that (2) This shows that i s a coordinate system i n some neighbor hood of since (2) and the Inverse Function Theorem show that u ' is a diff eomorphism in a neighborhood of Now x = (x 0 u') 0 u p. hence x. so y o f 0 x. .. By a permutation of the coordinate functions ui and yi we can arrange that (ya .an) = y 0 f(q) for q = x'(a'. . . (2) Choose coordinate systems x and v so that v 0 f 0 x .n. . since may not be contained in any kdimensional subspace of PROOF. . . Since rank f = k in a neighborhood of p. { ya 0 f(q) = aar a = I. . .. � = I ..an) means x(q) = (a'. .
. . . 1/rk+1 (a).ak.. 0 V I . 0) .. . . m . . .. v(q) = (bl. . Thus we can write r = k + l.44 Chapter 2 must vanish in a neighborhood of p.ak : Theorem 9 acquires a special form when the rank of f is n or 1 0.. . THEOREM. ..an) = y (al.. around f(p).. . bm) the Jacobian matrix y is a coordinate system in a neighborhood f(p yofox'(a'. ak» by (3) = (a 1 . so of ). . m: V) U) . . . Nm has rank m at p. 1/rm(a» = (al. . . .. . .. " " 1/rm(a) _ 1fm(al. vk). . . Define ya = va yr = vr _ 1fr for Since 0 (V i . . (1) If m S n and f: Mn .an) = y o vI o v o j oxl(al.. • . Moreover.... . . . .. ak). 1/rk+1 (a) _ 1fk+1 (al. .. . . • . . 0. . . ak. around p with has nonzero determinant.. then for any coordinate system (y. there is some coordinate system (x.
(CPI (bI .O.. fl (p») . . � = l . .bm) '" (X(¢I (bI ... y = 0 y. . V) = . to permute the yi in order to arrange that det f: Mn Nm has rank at p. bn)).. . .. there is a coordinate system around f(p) with y o f ox'(a' .O. .bn+'. . 45 (I) This is practically a special case o f (I) i n Theorem 9. 0). it is easily checked that A ljf is now the desired •:.. 1/1 f o¢'(a'.. . C �)�(1/r:(7nIT{(f »:12 0 f(a'. 0 Then A 1/1 0 0 . then for any coordinate p.bn) for (b I . .O) (a I . . . 0) .. .an..an) = A 1/1 f o¢'(b'. m. . . U) around PROOF. Define A by x f 0 <� f x = ".60 f p.. the rank of equals n in some neighborhood of It is convenient to think of the case and N '" ]Rm and produce the coordinate system y for ]Rm when we are given the identity coordinate system for Part (2) of Theorem 9 yields coordinate systems for ]Rn and 1/1 for ]Rm such that u. . .O.bn) '" ¢(a) '" A(b'. .O. ..'" . .. C(Yaau.. only the need be permuted. .. ... it is clearly unnecessary.O). . a .. the map f still takes ]Rn into the subset ¢ f(]Rn) which 1/1 takes to ]Rn {OJ ]Rmthe points of ]Rn just get moved to the wrong place in ]Rn {OJ... Even if we clo not perform ¢' first... bm). 0) 0 y. . > II (y. .an.. . . . in the proof of this case. . ..O..bn. A(bI . f M '" ]Rn ]Rn..bn)..an.. . This can be conected by another map on ]Rm. . so A 1/1 is the desired If we are given a coordinate system x on ]Rn other than the identif we just define y.. i t i s only necessary to observe that when k = In. . . . . (2) Since the rank of at any point must be :s n. ..an) (a'.. .Differentiable Structures (2) If n :s m and system (x. = = ..an) (a' . . .
On the other hand.(0) = o.46 Chapter 2 Although p is a regular point of f in case (I) of Theorem 10 and a critical point in case (2) (if 11 < 111 ). Another example is IR IR x>O x < x=O o.0) . I g (x)l). the dimension of the domain M. One can easily define a similar curve whose image looks like the picture below. the curve itself manages to be differentiable by slowing down to velocity 0 at the point (0. their images are not homeomorphic IR . A more illuminating example is the function h : IR > 1R2 defined by h(x) = ( g (x). The simplest example is the function f : > defined by f(x) = x3. with /. it is necessary that 111 ?: 11. Of course. Although the second and third immersions f31 and fh are oneone. A differentiable function f : M n > N m is called an immersion if the rank of f is 11. as defined in Chapter I). it is case (2) which most interests us. and it is clear from Theorem 10(2) that an immersion is locally oneone (so it is a topological immersion. Three immersions of in 1R 2 are shown below. at all points of M. a differentiable map f need not be an immersion even if it is globally oneone. although its image is the graph of a nondiff erentiable function.
Coo �+ f Coo p V C p 0 i Vn . Of course. r i. if N is the manifold neigh borhood of a point E M" then there is a function g on a neighborhood M of such that g = g on MI we can define ya (q' = = . MJ Mn VI MI E M" (y. MI M MI M IR M M). such that nV OJ. where { yr(q')) = 0ya (q) a = kl. I I. if is a k dimensional immersed submanifold of and is a neighborhood in of a point P then there is a coordinate system V) of around p.n. In general. > �!r! u. M VI = {q E M : yk+' (q) = .+. . . there is certainly some metric and some differentiable structure on f(P) which makes the inclusion map i : f(P) > an immer sion. .. = yn(q) = this consequence Theorem = g : MisI an immediate(considered as aoffunction on 10(2). a subset C with a differentiable structure (not nec essarily compatible with the metric inherits as a subset of is called an immersed submanifoId of if the inclusion map i : > is an immersion.DifferenliabLe SlTucluTes 47 to 1R. . second time around shows that may even be a dense subset of Despite these complications.k. The following picture. g(q) = g(q'). indicating the image of an immersion fJ. . . MI M. even if the oneone immersion f : P 7 is not a homeo morphism onto its image. with Mr) in aThus. : > S X S M MI M. l.
C M is an immersed submanifold. .. = {q E V : yk+' (q) is a neighborhood of f(p) in M. Let i : M. > M he the inclusion map. If M. The following figure shows that f might not even be continuous IR. with its Coo structure.. Actually. For example. even if g is Coo on all of M. on U. f : P > M is a Coo function with f(P) e M" and f is continuous considered as a map into M] . this is the only thing that can go wrong: 1 1 .' : fJ. . Given P E P. then f is also Coo considered as a map into Mj • PROOF. J is a coordinate system of M. this cannot be done if g is one of the functions rf. If M. . yk l U. c M is an immersed manifold.48 Chapter 2 On the other hand. as a map into M. = . choose a coordinate system (y. = yn (q) = OJ . and f : P > M is a Coo function with f(P) e M" it is not necessarily true that f is Coo when considered as a map into MI . we may not be able to define g on M. . PROPOSITION. . and (y ' I U" . We want to show that j.' 0 f is Coo ifit is continuous. V) for M around f(p) such that U.(M) > One other complication arises with immersed submanifolds.
. is open in M" this means that j.DifJerentiabLe Structures By assumption. If j : M" > N has constant rank k on a neighborhood of j' (y). In particular.I 0 49 f is continuous. i. C M is called simply a (COO) submanifold of M if the inclusion map i : M. . •:. > M is an imbedding. and gives the sphere S" . . There is one way of getting submanifolds which is very important. so that )' manifolds can be pictured as subsets of Euclidean space (though this picture is not always the most useful one).' 12.m)dimensional submanifold of M (or is empty). Since all y j 0 j are Coo. PROOF. .y which would be used in . Such an immersion f\1ost of these difficulties disappear when we consider oneone immersions is called an imbedding ("embedding" for the English).:. f : P 7 M which are homeomorphisms onto their image. Since U. as a special case.k (or is empty). it is called a closed submanifold of M if M. but we first develop some of the machine. then j . Thus j takes some neighborhood of P E P into U.' ( U. so 0 j' i (open set) is an open set. = Left to the reader. . yk are a coordinate system on VI) the function f is Coo considered as a map into MI . if y is a regular value of j : M" > N m . and y ' . submanifolds of )R N will seem like fairly concrete objects. \� � • •• % submanifold . It is to be hoped that however abstract the notion of Coo manifolds may appear.' C IR" .{OJ c )R".' (y) is a closed submanifold of M of dimension n . ) C P is open. lS also a closed subset of M. PROPOSITION. . defined as {x : Ixl 2 I}. then j . We will prove this fact only for compact manifolds.' (y) is an (II . . Now it turns out that ever (connected) Coo manifold can be imbedded in some )RN . An immersed subman ifold M.
there are compact sets C" C2 . Unfortunately. we can obviously cover Ai by a finite number of open sets. a cover (9' of M is a refinement of (9 (or "refines (9 ") iffor every V in (9' there is some V in (9 with V C V (the sets of (9' are "smaller" than those of (9)a subcover is a very special case of a refining cover. has an open neighborhood V. Moreover. and each contained in Vi = Ui+ 1 . C3 . with compact closure. . . PROOF We can obviously assume that M is connected.Ui. we can choose all members of (9' to be open sets diffeomorphic to IRn.2. and hence is M. <. Now M is the union for i > I of the "annular" regions Ai Vi . since we will need it later on anyway. there are many definitions and theorems involved.. Then V. THEOREM. then there is a n open cover (9' of M which is locally finite and which refines (9. 1 3. C Uj+1 . I f (9 i s a n open cover o f a manifold M .2 . j = IRn. • • • with M C.lhich refines (9 and which is locally finite. Let V_. = Vo = 0. If (9 is a cover of a space M. Continuing in this way. We can also choose these open sets to be diffeomorphic to In this way we obtain a cover (9' v. . we obtain open sets Vi. with U. each contained in some member of (9. whose union contains all Ci . compact and U. Clearly C.50 Chapter 2 the general case. A cover (9 is called locally finite if every p E M has a neighborhood W which intersects only finitely many sets in (9 . Since each Ai is compact. . U C2 U C3 U .Ui . U C2 has an open = neighborhood U2 with compact closure.l . By Theorem 1. since a point in Vi is not in V for j 2: 2 + i.
.(U2 U V3 U . . such that Coo V (1) support <Pu C V for each V. by (I)). THEOREM.+2 U··· )... [0. C] V{ UJV]. Un+i U�+j P E by cannot possibly eliminate p. . C] U V2 U V3 U = ) = V'V {V]. <. V' V' V PROOF. THEOREM (THE SHRINKING LEMMA). For any p E M there is a largest n with p E because (9 is locally finite. We can clearly assume that M is connected. for each in (9.. V3 . 15. (2) "L <Pu (p) u neighborhood of p.. one for each in (9.. U V� U (Vn+] U V. open set with c c c Continue in this way. U V3 U n V{ U V2 U·· . ' . Then it is possible to choose. This shows that an open locally finite cover of a connected manifold must be countable Qike the cover constructed in the proof of Theorem 13). . Let (9 be an open locally finite cover of a manifold M. Now P E it follows that since replacing C2 V2 V2V2. }. Let (9 Then is a closed set contained in and M = open set with C C C Now C] V] . V2. . Let W be an is a closed set contained in and M = i. Let (9 be an open locally finite cover of a manifold M. .V2. an open set with is also an C in such a way that the collection of all open cover of M. = 1 for all P E M (this sum is really a finite sum in some .. Let be an . Then there is a collection of functions <pu : M . C C 14.V]. then intersects only finitely many members of (9. V{ . V{ UVC2.DijjewztiabLe Structures 51 Notice that i f (9 i s an open locally finite cover o f a space M and C M is compact. V{UV2U " ' . I ] . .
I]} satisfying (1) and (2) is called a partition of u nity. (A collection {¢. provided that Lemma 2 is true for c U e M with closed (but not necessarily compact) and U open. C U. Let C C C C C. where since it is a finite sum in a neighborhood of each point. Chapter 2 Case 1. But this is a consequence of Case 1: For each p E choose an open set Up C U with compact closure. • . General case. 1 6. (3) for each i there is a U E (9 such that support ¢. jRN Coo manifold. COROLLARY. 1 ] 14.) It is now fairly easy to prove the last theorem of this chapter. it is easy to see that support f e u. I] such that (I) the collection of sets {p o ¢.(p) # O} is locally finite. it is called subordinate to (9. if it satisfies (3). VeC}' L ¢U. This case can be proved in the same way.(p) 1 for all p E M. clearly ! (p) = I for all p E Using the fact that (9 is locally finite. then there is a collection of functions ¢. clearly Coo Define Ue(') L 1/Iu > 0 ¢U = everywhere. = Coo (2) L ¢. Ue(') L 1/Iu Case 2. Since the U ' cover M . If M" is a compact cling ! : M > for some N. != L u ¢u(P) = I for all p. Cover Mwith open sets Va having compact closure and contained in M The open cover {Up. C. : This sum is Coo. and ¢u(p) = 0 when U C Va . �.52 PROOF. If (9 is any open cover of a manifold M. 1 7 . Each U in (9 has compact closure. V } has an open locally finite refinement (9 to which Case 1 a applies. then there is an imbed . Choose the U' as in Theorem function 1/Iu : M > [0. Since (9' = { U E (9 : U C Up for some pl. : M > [0. THEOREM. : M > [0. Apply Lemma 2 to U' c U e M to obtain a which i s I o n U ' and has support C U.
Di/Ji:renliabLe S'lruclures 53 PROOF.. and on Vi.. Moreover.1. . show that the 11 at each point is unique wi/houl using Illvariance of Domain.Ai are COOrelated. •:. so p = q) since Xi is oneone on Vi ... There is some i such that p E Vj.. This is an immersion. the N x II Jacobian matrix It is also oneone. Vkl with M = Vl U · · · U Vk . How many distinct Coo structures are there on 1R? (There is only one up to diffeomorphism. (b) Show similarly that aM is welldefined for a Coo manifoldwithboundary M. where N = lIk + k. Then Vri(p) = I. (a) All Coo functions are continuous. show that all charts in .. (a) Show that being COOrelated is not an equivalence relation. that is not the question being asked. . For suppose that f(p) = f(q). . .. and functions Vri : M . N is Coo if and only if g 0 f is Coo for every Coo function g : N . Choose V: as in Theorem 14. Vri ' Xi (P) = Vri ' Xi (q) . There are a finite number of coordinate systems (Xl . in fact.. This shows that we must have q E Vi . where VIi = I. jRn 3. by JRN . Vl ). 1] which are 1 on Vj and have support C Vi.. JR. because any poiIlt p is in V! for some i. we can always choose N = ar ( axf ) contains the 11 x 11 matrix (ax�) axf ' . as claimed. so also Vri(q) = I . 2. Problem 333 shows that. and the composition of Coo functions is Coo.. (Xk..) . Define f : M . 4. PROBLEMS 1. 211 + I .. (a) If M is a metric space together with a collection of homeomorphisms x: U 7 whose domains cover M and v·"hich are Coorelated. (b) In the proof of Lemma I. (b) A function f : M . [0.
. ( PI .. defined by Ji x h(Pl .. (Obvious. U) in . P2) of Ml ... a map / : N .. are Coo (i = 1 . M... Ml x M2.. show thaI h( P2) ).4. Let g : .... (b) The diffcrentiable structure on Ml x M2 makes the "slice" maps 8.. p snl p are Coo related to the 2n p2 p sn snIP'n IRn lHIn IR IR lHIn.. q E M.. M2 are Coo.. can one determine the rank of (fJ . be the map ...... M2 .. V) in A (c) Show that the inclusion i : N . .. M is Coo..... Compare the rank of / and the rank of / 0 g .4. P2 ) = (Ji is Coo and determine its rank in terms ofthe ranks of ji.. Moreover. h E M2.' is maximal for N if . h) : N ... p? (pt}. . and that . the Coo structure we have defined for Ml x M2 is the only one with this property. / can be extended to a Coo function on a neighborhood of (Not so obvious. Ml x M2 differentiable for all fit E Ml ..) IR lHIn.' is the unique atlas with this property. Ml x M2 is Coo if and only if the compo sitions 1fl 0 / : N .. then / is Coo.4. /1 x 12 : Nl x N2 ... M with c (O) = and c ( I ) = q ..54 Chapter 2 5. If / : 1HI" . h) .. is locally Coo. 1 0 ... (a) If N C M is open and . M is Coo if and . Ml x M2 at in terms of the ranks of Ji at For Ji : N.. M is an imbedding.. is maximal for M.Ai consists of all (x... (d) If Ji : N . with U C N. (b) Show that .. (a) Show that (MI 7.4...) (b) If / : .. (a) If M is a connected Coo manifold and .. (b) It is even possible to choose c to be oneone.. show that . 6.' can also be described as the set of all (x I V n N. If M is a Coo manifoldwith boundary. Check that the two projections PI and P2 on homeomorphisms Ji and g. M..4. 1 ] .. i s locally C oo (every point has a neighborhood on which / is COO). 2).. then / is Coo... i. Show that / : only if / 0 g : .. IRn1 1 2.. (c) More generally. PI . x M2) X M3 is diffeomorphic to MI x (M2 x M3) and that MI x M2 is diffeomorphic to M2 x MI . 9. [pl...... h to Coo functions in a neighborhood of then Djg and Djh are the same at points of x {OJ (so we can speak of Dj / at these points)..... (a) If U C p IP'n i s open and / : U .. (iit ...e. then there is a Coo curve c: [0. has two extensions g. then there is a unique Coo structure on aM such that the inclusion map i : aM . V n N) for (x. MI and 1f2 0 / : N . M is Coo.. 1 1 .
then U is a manifoldwithboundary. (3) f(x. (a) Let C be an open set such that boundary is an (n . y) c 1HJ 2 for y :0: 0.y) C ]R 2 _ 1HJ 2 for y = (x.O) < 2). The closure S of the final resulting figure is known as Alexander� Homed S plwe. O) < I or I < d(x. 0..Differentiable Structures 55 13. so part (a) is true only for differentiable submanifolds.S. Show that U is an IIdimensional manifold withboundary.) V if V smaller copies of the two parts of S' into the regions indicated by arrows. (a) There is a map f : ]R2 . ]R2 such that (I) f(x.. This figure may be extended by putting V Mn V. and then repeating this construction indefinitely. 0) for all x. O) (2) f(x. < ..I)dimen sional (differentiable) submanifold. but au # boundary (b) Consider the figure shown below. (It is well to bear in mind the following example: = {x E ]R" : d(x. Show that S is homeomorphic to S 2 (Hint: The additional points in the closure are homeomorphic to the Cantor seLl If is the unbounded component of ]R3 . 14. V = V. then S boundary but U is not a 2dimensional manifoldwithboundary.
aN aM a(p) V (p. Show that there is a unique Coo structure a: V .. (d) By using two different pairs define two diff erent Coo structures on considered as the union of two copies of with corresponding points on identified. Let P = be obtained from the disjoint union of and by identif ying E with E If is a coordinate system around E and (y. . (We will be able to prove later that such diffeomorphisms always exist). aN M N p aM x aM f(x) aN. I) induced by and is a diff eomorphism. (b) Suppose and are Coo manifoldswithboundary and is a diff eomorphism. Show that the resulting Coo manifolds are diffeomorphic.O) Mp aM. such that = for alI E and a similar diffeomorphism V x [0. V U aM M a N fJ (a. but itself is not Coo. nn2 JR ann22. f(V aM) aN. (x. (c) Now suppose that there is a neighborhood of in and a diffeo morphism x [0..I . . I). . fJ). V) f(p). x x f define a Coo structure on P. I). Show that this procedure does not f M N M UJ N f: aM . V) a coordinate system around with and (y 0 n = n we can define a n = V n homeomorphism from V C P to by sending to 1HI" by and V to the lower halfplane by the reflection of y.. aMfJ: . .V U fllVJR" aM xi VV aM. on P such that the inclusions of and are Coo and such that the map from V to x ( .56 (4) Chapter 2 restricted to the upper halfplane or the lower halfplane is Coo. but that the diff eomorphism ca"not be chosen arbitrarily close to the identity map.
I = �A I L a2i i=l 18. I: JR JR I(x) = { �1/x x > 0o x > CooCoo. where each Vi is of the form < E I}  onto ° on Let Vi . . Can the inclusion be an imbedding? Are the projections on each factor maps? (b) If and are manifoldswithboundary. = X .I 17. The proof turns out to be quite easy. > aJ. I] be a function with ° on Vi and Functions like J.: M will be called mixed partials of ai = Show that sup of all mixed partials of Ji . for any closed C (b) Let {V. I for some coordinate system taking an open subset of > [0. construct a structure on x sucl1 that all the "slice maps" (defined in Problem 8) are 16.. Coo J. Cl = C1 C2 CM. Let is Coo. we can even find with = 11 (1). Lemma 2 (as addended by the proof of Theorem 15) shows that if and C2 are disjoint closed subsets of then there is a function > [0. . I] such that C 11 (0) and C 11 (I). a2J. Coo I C = ({a JR": lal ) x MJ. Consider the coordinate system (yl. . + for JR2 defined by . of order 1 . . Actually. . .} be a countable cover of V. (a) Find a structure on 1HI1 x 1HI1 which makes the inclusion into JR2 a map.b) = a b.Differentiable Structures 57 15. once you 11 (0) and know the trick. axj ' axj axk ' J. C M M. =C M JR". defined by :s (a) It suffices to find.  a function with 11 (0). 2 Coo > x C1 I: M ° for < 0. . 2.C. .1 /x' is used just to get a function which is and e. C = and 11 (0). Show that the function Coo Coo Coo M NM N Coo is (the formula e. y2) yl(a.b) = a y2 (a.I /1xl could be used just as well). Ji of all orders :::: i.
:) + fe (Hii ) ]' a/ax in terms of alar alae. 1] [0. k and An . (b) Also compute it from Proposition 3 (to find of and y 2) . . A n .. x into 22n squares. the operator alai depends in terms of polar coordinates..k is labeled simply k. I] [0. . /ayj . Answer: then compute 20.l+ l . then f(M) has measure that M has only countably many components). write each 1.. is and m > II. [0. (a) The lower left square is A n .k for all 2""2'1I � I :S 2k2n · Show that f is continuous. �[f. k +l have a common side. 1] (d) Squares An •41+ 1 . A n •41 +2 . square A n .. The numbering is determined by the following conditions: [0 . (b) The upper left square is A n• 22n. 2"'.1] 1..: (rf.b) yI I y af/ayl af/aI /. and 0 (provided [0 2 1 . not just on even though af/ayl (a. onto [0. The following pictures show. 1 ]. If M n . Notice that # on and i. I .2. . for II = and 3.. Define x []J [EJ 16 15 2 13 3 14 12 9 8 10 6 7 6 7 10 5 8 9 4 3 II I 4 5 I 2 II 12 13 E x . a2/ax2 f: Nm Cl aI. and not oneone. f: [0. I . . 1] by the condition kf(l) An.58 Chapter 2 (a) Compute from the definition. A n . Compute the "Laplacian" = 1 1 . (First compute from this). 1] . A n •41+3) An •41+4 are contained in A n l . a subdivision of . (c) Squares A n. in terms yl 19.
. p E V) is a submanifold of ]Rn. (b) Consider the homeomorphic image of S ' obtained by adding. Let c . can be made into a closed submanif if and only such old coordinate systems exist around every point of M. [0. x E ]R) is not the image of any immersion of ]R into ]R2 ]Rk is open and f. What does the inside of this curve look like? 23. n V = {p : x k+ ' (p) = . "Ld(C(li). You will need . . a semicircle with diameter the line segment AB. C M can be made into a kdimen sional submanifold of M if and only if around each point in M. For p/2n E [0. (Neither Theorem 9 nor l O is quite strong enough. Show that g is oneone. and that its image will not have ]Rn be continuous. i=1 k C 24. a set M. so that it has a continuous extension ]R2. Ixl) . f(p» E ]R" . dinates. = xn (p) = 0). (b) The subset M. Coo if {(x. For each partition P "" {Io. 1] [0. (b) Every submanifold of ]Rn is locally of tllis form. 1 ] '"' B = f(l) (a) Show that > measure 0 if the shaded triangles are chosen correctly.Differentiable Structures 22.C(li. below the image of g. P » ) is bounded above (with length equal to sup{i(c . . define > f is uniformly continuous. [0. V) on M such that M. The set 26. . P» )). (a) If V C Coo. after renumbering coor 25. . . 1]. A = f(O)�"" g . Id of i(c. define f(p/2n) E 59 ]R2 as shown below. there is a coordinate system (x. V > ]Rnk is then the graph of f = { (p. P) = The curve is rectifiable if {i(c.) . (a) If M is a manifold. 1]. Show that the image of a rectifiable curve has measure O.
3 1 7321. Let I : IPz . then II (y) is a (closed) submanifold of M of dimension n ... known as Boy's Surface... the function I can be made if M is a manif old. there is a proper map I : M .. (f) If M is a manifold. A continuous function I: .. (b) Prove the Shrinking Lemma when the cover (9 is pointfinite and countable (notice that localfiniteness is not really used). I] is in only finitely many members of the cover. whose image is the Steiner surface. • • • E with no convergent z subsequence..... X 3 . M is proper. y.xz..... The limit set L (f) of I is the set of all y E Y such that y = lim I(xn) for some sequence x" X . (b) I( ) c Y is closed if and only if L(f) C (c) There is a continuous I: IR . Theorem l O is es sentially Theorem 213 of Calculus on Manifolds. N (a) L(f) = 0 if and only if I is proper. (a) An immersion from one nmanifold to another is an open map (the image of an open set is open).. N N N 28... C X (C) X X X f(X). xy) defined in Chapter I.k (or is empty). pp. There is a way of immersing IPz in IR 3 .. pg.. Y is a homeomorphism (onto its image) if and only if L(f) n I(y) = 0. 29.. Y is proper if II is compact for every compact C Y. and I : M . Z]) = (yz. consider collections e . (e) A submanifold M. 1R.60 Chapter 2 the implicit function theorem (Calculus on Manifolds. (a) Find a cover of [0. Prove Proposition 12: If I : Mn . 1R 3 be the map g([x . 30. IRz with I(IR) closed. (d) A oneone continuous function I: . 41)..) 27. . comparison with the implicit function theorem will show how some information has been allowed to escape. C M is a closed submanifold if and only if the inclusion map i : M. (You will need Zorn's Lemma.. (c) Prove the Shrinking Lemma when (9 is a (not necessarily countable) point finite cover of any space. See Hilbert and CohnVossen. I] which is not locally finite but which is "point finite": every point of [0.. then I is onto. Show that g fails to be an immersion at 6 points (the image points are the points at distance ± 1 /2 on each axis). but L(f) # 0. Geometry and the Imagination. Coo Coo 31. is an immersion. (b) If M and are nmanifolds with M compact and connected.. has constant rank k on a neigh borhood of I'(Y).
or unimodular group. and that 1/1 0 RA = 1/1 for all A E O(n). I). U' C U. in this case. Using the formula for D(det) in Calculus on Manifolds . show that S L(n . show that for A E O(n ) the matrix 32. = (0. (a) The set of all nonsingular n x /1 matrices with real entries is called GL(/1. pg. and use partitions of unity). (a) If M. Coo ( a1/lU (A) ax kl ) has the same rank as ( a1/lij I axkl ( ) ). together with all other U E 19 covers the space. (Here xkl are the coordinate functions in IRn 2 . By applying the chain rule. The subgroup 1/1'(1) of GL(II. since it is an open subset of IRn'. U :J M. 1R) is a closed submanifold of GL(n . IR). 1R).) Conclude from Proposition 12 that O(n) is a submanifold of GL(n. 1R) of dimension n 2 . . (c) This is false if IR is replaced by a disconnected manifold of pairs (U. (b) This is f alse if M = IR and M. f has no continuous extension to a map from ]R2 to but the proof requires some topology. S I .Differentiable Structures 61 Remark: It is also false if M = 1R2 .) Coo. (extend locally. 1R) > (symmetric matrices) by 1/I(A) = A . Coo N S'. C M is a closed submanifold. U ') where U E 19 . (e) Using the formula 1/Iij (A) = L. A" where At is the transpose of A. f can always be extended to a function in a neighborhood of M. and 1/Iij the n(l1+ 1)/2 component functions of 1/1. is any neighborhood. > IR is then there is a function J: M > IR with J = f on M" and with support J c U. U') E e. (b) The symmetric n x n matrices may be thought of as IRn (n + ' ) /2 Define 1/1 : GL(n . (d) For any A E GL(/1. in fact. and the union of all U' for (U . 33. 1R).IR) is called the orthogonal group 0(11). Show that RA is a diffeomorphism. = = and f = identity. 24. (c) Show that O(n) is compact. 1R) by RA(B) = BA . define RA : GL(II. Coo N. aik ajk k (A = (a/j »). 1R) > GL(n . the general linear group. and f : M. M. However. Show that A E O(n ) ifand only if the rows [or columns] of A are orthonormal. IR). is the subgroup of all matrices with det = I. It is a manifold.I. The special linear group SL(n .
) Conclude that O(n) has dimension n (n . x 1 aj l au k=i#J k=j#i k =i=j 2aj/ o otherwise.62 Chapter 2 show that a 1fr ij . 1R) is called the special orthogonal group SO(n). The group O(n)n SL(n.Ao are < e . . 34. Let M(m . (f) Show that det A = ±I for all A E O(n).Ao are < e. k) C M(m. l1. k ) there are permutation matrices P and Q such that PXoQ = ( AO Co BO Do (b) There is some e > 0 such that A is nonsingular whenever all entries of A . then X has rank k if and only i f D = CAJ B. then ( � �) (d) M(m .:s m.(A) = axkl Show that the rank of this matrix is n(n + 1)/2 at I (and hence at A for all A E O(n). or the rotation group R(n). Hint: If h denotes the k x k identity matrix. n matrices. n.1)/2. n . k) the set (a) For every Xo E M(m . n) denote the set of all m of all m x n matrices of rank k . n . where the entries of A . (c) If PXQ = ) ' where Ao is k x k and nonsingular.k) for all k .n) is a submanifold of dimension k(m + n . and M(m .
and the "velocity vector" or "tangent vector" c'(t) of the curve c is customarily pictured as the arrow from crt) to crt) + c'(t). Then c'(t) = (cl'(t). But there are Amany situations where we would like to picture this same arrow as starting at a different point P E JR": For example. .CHAPTER 3 THE TANGENT BUNDLE point v E JR" is frequently pictured as an arrow from 0 to v. suppose c: JR > JR" is a differentiable curve. c"'(t» is just a point of JR". . but the line between crt) and c(t) + c'(t) is tangent to the curve. . . crt) + c'(t ) 63 .
64 Chapter 3 To give this picture mathematical substance. To recover the first member of a pair v E TIR". the one visualizable case occurring when 11 = 1. v). the "tangent space of 1R"". This picture gives rise to some TJRn JRn _ _ ]( '_ terminologywe call ](1 (p) the fibre over l  p p. b) = a.. we will denote the set of all (p. For any tangent vector v. like v. which we will also denote by TIR". the point Jr ( v) is "where it's at". we define the "projection" map 1r: 1R" x JR" . The set ]( . in conformity with this notation. v) E TIR" by vp ("the vector V at p"). The set of all such pairs is just 1R" x 1R". we simply describe the "arrow" from p to p + V by the pair (p. Alternately.. This fibre can be made into a .1 (p) may be pictured as all arrows starting at p. elements of TIR" are called "tangent vectors" of 1R". 1R" by ]( a . it can be pictured more geometJ·ically as a particular subset of 1R" x 1R". We will often denote (p. it is more convenient to denote a member of TIR" by a single Jetter.. At times. v) for v E 1R" by 1R"p .
(I) By our definition. v) (j) (p. . is de noted by f . should really b e thought o fa s defined on and JR TJRn. Suppose that g : JRm > JRk is another differentiable function.' (p) 1( . v + w) a o (p. by the chain rule. J. W) = (p. ( [ Df(p)(v)lJ(p) } = (Dg(f(p)}(Df(p)(v» }g(f(p))' This looks horribly complicated. however. has the map f. the symbol f. D (g 0 f)(p) = Dg(f(p» 0 Df(p). . then the linear trans formation Df(p) : JRn > JRm may be used to produce a linear map from JRn > JRmJ( ) defined by (j) ) p p Vp "'" [ Df(p)(v)lJ(p) ' This map. = f 0 1(. x respectively. so that. peJRfI x (j) and 0 (p. g.p p p ing diagram "commutes" (the two possible compositions from TlRn to jRm are equal). and 0 . v) = (p. Thus. This is not the only reason for defining f in this particular way.' (p). and p E JRn. denotes the map f : TJRn > TJRm which is the .The Tangent Bundle 65 vector space in an obvious way: we define (The operations U 1( . whose apparently anomalous features will soon b e justified. but. Since f (v) is defined to be a vector E JRmJ( ) .a · v). it can be written . using (I). the follow . 1( 0 f. .p union of all f. built into it. as well as all maps Df(p). Usually we will just use ordinary + and · instead of If f : JR" > JRm is a differentiable map.
J ) is the "unit" tangent vector of 1R at t . JR : 1'(1) then C'(I)C(I) = (I .. 0 I.66 Chapter 3 thus we have This relation would clearly fall apart completely if I.(vp) were not in JRmJ(p) . The tangent .. . JR" can be defined in terms of this concept.. c. c"'(I» c(I) . If g: JR" .. . . .. The tangent vector of c at 1 may be defined as g. = (g 0 fl •. in terms of I.] Notice that the tangent vector of c at 1 is the same as where I t = (t ... /(1 » . The "tangent vector" of a curve c : JR . this vector lies along the tangent line to the graph of 1 at (1. I. then g 0 c is a curve in JRm. like rank or singularity.(I. with our present definition of I. also. c'(I)c(I) E JR"C(I) ' [If c happens t o be of the form c(l ) = (1. JRm is differentiable. Henceforth. we will state almost all concepts about Jacobian matrices.) = [DC(I)(I)]c(l) = (c" (I)... rather than Df. /'(I» c(I).. 1(1 )) for I: JR . it is merely an elegant restatement of the chain rule..
(i 0 x ..I ) .X(p) . Consequently. Consider now an ndimensional manifold M and an imbedding i : M . Suppose we take a coordinate system (x..I ) .I ) . = (i o x .(p). 0 (x 0 y .I o x o y. n n 'y(p) : IR y(p) .(c. (lRnx(p) is an ndimen sional subspace of IR N .I is a map from IRn to IR N with rank n.(I..The Tangent Bundle 67 vector of g 0 C at t is (g 0 C). then (i o y I ).I ). This subspace doesn't depend on the coordinate system x) for if y is another coordinate system.) = g.. (tangent vector of c at I) .(I..) . IR N . (x 0 y.. Then i 0 X .» v crt) = g. I = (i 0 x .I ) � �� � �. and is an isomorphism (with inverse (y o x . U ) around p. IR x( p) .
then a = .I (p) has a vector space structure also. If c : (e. For each p = (cos e. Now so the tangent vector of every a is in (i o x. cose)p (it clearly doesn't matter which of the infinitely many possible e's we choose). i)p con n(v) = p if V E (M. each "fibre" 7f . Consider first the manifold M = S ' and the inclusion i : S ' > jR 2 .I 0 c is a curve in jR N which lies in . Beyond this we have to look a little more carefully at some specific examples. and every diff erentiable curve in i(M) is of this form (Proof?). e) > jRn is a curve with c(O) = x(p). our ndimensional subspace is just the set of all tangent vectors at i(p) to differentiable curves in i(M). The Curve c(e) = (cos e.i) = p EM U (M.J ) * (lRnX(p) ) ' Moreover. i)p' We now want to look at the (disjoint) union T(M. We will denote this ndimensional subspace by (M. 0 x .(M). cos e) # O. let up = (. sin e) E S ' .i)p. and c'(e) = (sine.sine) passes through every point of S ' .68 Chapter 3 There is another way to see this.i) > M by As in the case of TlRn . every vector in this subspace is the tangent vector of some a) since every vector in jRnx(p) is the tangent vector of some curve c.sin e. which justifies the picture we have drawn. We can define a "projection" map n : T(M. Thus. i)p C i(M) x jR N C TjR N . . Then (S ' .
then. i) .The Tangent Bundle 69 sists of all multiples of the vector up . S' x ( ) T(S ' .S ' x IR' � S. then each fibre has a vector space structure in a natural way.b) a] = T(S2. SS22 2 ]R2. which makes the following dia· [rr'(a. S2 fi. clearly I. for a fixed vector v #. one at each point of which varied continuously. restricted to a fibre is a linear isomorphism on to the image. It is a wellknown (hard) theorem of topology that this is impossible (you can't comb the hair on a sphere).). . and the inclusion i : Now consider the manifold M = C 1R 3 • In this case there is 1/0 map fz : X 1R with the properties of the map If there were. up = (p.O in the set of vectors T(S'. ... . ). : i) gram commute. would be a collection of nonzero tangent vectors. . We Can therefore define a homeomorphism I. i) fi . If we define the "fibres" of to be the sets (p). S2. Commutativity of the diagram means that fi takes fibres into fibres. ft rr' rr'' 1R ' by fi ).
i) > M X 1R 2 exists. 2 sin 8 + 1 cos � sin 8.0) ) = * (21r. 21r] x (I.1 .0. 0): Ifwe could. I) > 1R 3 defined by 1(8. without appealing to a hard theorem of topology. 21r] > 1R.0) . 1 )(2rr. This function would have to be nonzero everywhere and also satisfy A(21r) == A(O).0) (2.0) = ( 1 . 0 . which it can't (by an easy theorem of topology). shown * as dashed arrows in the picture. The impossibility of choosing nonzero dashed vectors continuously clearly shows that there is no way to map T(M. 0)(0. 1 sin n .O)J [ aal l [ DI(O. 0)(2.0) . 21r I r�� is a tangent vector.0.70 Chapter 3 There is another example where we can prove that no appropriate homeo morphism T(M. 0) (2. At each point p = (2cos 8.  .0. 0) i . 2 sin 8. The same is true for all multiples of 1 ( 0. Notice that 1 ( 0. 1 ( 0.0. 0 . = while = [� (O. 2 sin8. the particular subset of IR 3 defined in Chapter 1.0.0) ) . 1 ) (0. 0) of M. to be precise. for some continuous function A : [0. fibre by fibre. then each vector would be J (2. f). 1 )(8. 0) This means that we can never pick nonzero dashed vectors continuouslY on the set of all points (2cos 8 ..M is the image of the map I: [0.0) = ( . %2.1) = (2 cos8 + 1 cos � cos8.0) ). The map f will just be the inclusion M > 1R 3 where M is a Mobius strip. the vector (8.
. with Ell (n:I (p) x n:I (p») c n:I (p) and 0 (JR x n:I (p») c n:I (p). for each p E V. we can therefore define by f: f (mp(vx(p) ) n:I (V) .I (V). for all q E U... B is a continuous map onto B.. (2) n: : E . T(M. In standard jargon.. 0) .The Tangent Bundle 71 homeomorphically onto M x JR2• We thus have another case where T(M.. V x JR" which is a vector space isomorphism from each n:I (q) onto q x JR".. homeomorphically onto V x JR".. In fact.i) is "locally trivial". i) is always simple locally: if (x.. � = (E. V) is a coordinate system on M. v) . Ell .. fibre by fibre. B . the part of T(M. however.. i) to be included among an extremely important class of structures: An IIdimensional vector bundle (or nplane bundle) is a fivetuple where (I) E and B are spaces (the "total space" and "base space" of t respectively). This additional feature qualifies T(M. For any imbedding i: M .. V = x JR" (p..... 0: JR x E . E. n:. there is a neighborhood V of p and a homeomor phism I : n:I (V) . (3) Ell and 0 are maps Ell : p EB U n:I (p) x n:I (p) .. the structure of T(M. the fibre where the abbreviation mp has been introduced temporarily. JRN. i) over V. such that the following "local triviality" condition is satisfied: For each p E B. then n: .. can always be mapped.. i) does not "look like" a product M x JR". E. which make each fibre n:I (p) into an ndimensional vector space over lR.
v) by v + w. I] x IR by identifying (O. For vectors v. we will denote (j)(v. and a .' (p) and a E 1R. i) and T(M.. .. : E > B.a) with ( I .. B. w) and 0(a. and we shall usually refer simply to a bundle . . a). Notation as cumbersome as all this invites abuse. :::: �z) if there is a homeomorphism h : E. W E .n(x). . : X x 1R" > X the projection on the first factor. 0) automatically gives rise to a bundle �IA over any subset A C B.'(S'). to be precise. = " . i) are not trivial. (The local triviality condition for a bundle � just says that �IU is trivial for some neighborhood U of p. Equivalence is here a technical term: Two vector bundles �.) The bundles T(S z . or even denote the bundle by E alone.' (p) isomorphically onto "z' (p).n(B) is called trivial. .72 Chapter 3 Because this local triviality condition really is a local condition."nJ1sidered as a Idimensional vector bundle over S l .. This is called the trivial Ilplane bundle over X and will be denoted by . I } . for M can be obtained !i·om [0. The bundle T(S ' .. and the obvious vector space structure on each fibre.n(IRn). while S' can be obtained from {O. i») can be . : E. each bundle � = (E. i) considered before is equivalent to . respectively. The map h is called an equivalence. > Ez which takes each fibre ". but there is an even simpler example of a nontrivial bundle. (j). The "tangent bundle" TlRn is just . The Mobius strip itself (not T(M. The simplest example of an Ilplane bundle is juSt X x IRn with . > B and �z = "z : Ez > B are equivalent (�. v or av. A bundle equivalent to .
The diagram above illustrates local triviality near the point {O. > 1( . a)}) = {O. (I. EI BI � Bz . Since s must be ° somewhere. I }. Z jRk k l jRl TjR TjR jRk * There are actually several possible choices. The relations between some of these cases are considered in the problems. such that (I) the following diagram commutes jR 1(. I] > with S(O) = S(I). f: jRk jRjRll . s(8) E 1(1 (8) must be the ° vector for some 8 E S').' A bundle map from �. fl is a bundle map fi'om . i : 'M > and Nm C and j : N > are the inclusions. the map 1( is defined by 1( I. The analogue of a homomorphism is the following. > If M" c are submanifolds. I] by identifying ° with I . with j : E.a) = I for 0 < I < I and 1( {(0. > Bz. fl. Such a map corresponds to a continuous function s : [0.I U(p)) is a linear map. This surely shows that M is not a trivial bundle. a). depending on whether one is consider ing all bundles at once. Thus f may be restricted to be an isomorphism on fibres and f to be the identity. . An equivalence is obviously the analogue of an isomorphism. fixed bundles over various spaces. (2) j : 1(. ' (p) 1 � Ez l1(Z to for any differentiable The pair U . and the map f satisfies f(M) C N. > Ez and f : B. or a homeomorphism. then f .The Tangent Bundle 73 [0. or a fixed base space with varying bundles. I } of Sl Suppose that s : S' > M is a continuous function with 1( 0 s = identity of M (such a function is called a section). to �z is a pair of continuous maps (j. the section s must be ° somewhere (that is.
we could abbreviate to if we agreed that really denotes an equivalence class of bundles. j)p are of the form (j 0 x 1). T(M. rather than one bundle. In this way we obtain a bundle f o e in and consequently f.j). That is the T(M.i) TM. i) TMi . M. then (M. V) is another coordinate system. (w) for we obtain a bundle map from to which is obviously an equivalence. M N. In fact. j). while and are two imbeddings of in and respectively. i)p j (i xI M for WE WE IRnx( IRnx(p)' p). (w) T(M. (v) E map from to Actually. while elements of If we map T(M. (M. The map > induced on fibres is independent of the coordinate system for if (y. Elements of are of the form 0 ).i) T(N. i) T(N.74 Chapter 3 takes to to see this. we just consider the function j 0 f 0 > and extend it locally to The case which we want to examine most carefully is the simplest: where = N and f is the identity. just remember that v E is the tangent vector of a curve c in so f (v) is the tangent vector of the curve . j). IRk IRk i j M i.i)p k MIRIRI. T(M.j)IV. to the case where and are merely imbeddings of two abstract manifolds and and f : > is Coo. T(M. T(N.j)p x.I : i(M) N. I I We can therefore put all these maps together.j).i)p T(M. the dependence of on is al most illusory. N. and obtain an equivalence from to In other words.i(N) M N i T(M.i) T(M. T(M.i)I V T(M. it would have sufficed to begin with a Coo function f : > since f can be extended to locally. this construction could be generalized much further..
though quite natural. we can. Can we do this? Yes. where i : U > M is the inclusion. .The T angent Bundle 75 sort of thing an algebraist might do. en (JRn»). (I) If 1 : M Coo TJRn � TJRm 1m In en (JRn) old f. N) where f : M > N.. THEOREM. then (g 0 f) . precisely we will write I'old J/' when necessary. What we would like to do is to get a single bundle for each M. so in stating our result . such that: Coo > M is the identity. (2) There are equivalences In : TJRn > en (JRn) such that for every function f : JRn > JRm the following commutes. TJRn will be different from our old defini tion (namely.. M. TN PROOF The construction of TM is an ingenious. If • g : N > P. sub terfuge. then TU is equivalent to (TM)IU. : TU > TM are certainly different (they map TU into two different sets).. since the maps 1 * : TV + TV and i. em (JRm) 1 1 (3) If U c M is an open submanifold. �TM/f. it must be understood that the symbol "/" r lly refers to a triple (f. in some natural way. 0 f .. The identity map I of U to itself and the inclusion map i : U + M have to be considered as different. I . It is possible to assign to each nmanifold M an nplane bun dle TM over M. = g. then 1. i) has. We will obtain a single bundle for TM. : TM > TM is the identity. ea . but the elements of TM will each be large equivalence classes. * vVhen using the notation h. and for f: M > N the map (flU). TM TU (flU). and it is undoubtedly ugly. _ _ _ . f). More precisely. . : TU > TN is just the restriction of f•.' TU � TM)IU/c ( _ _ _ '" i... When we do. and so will f for f : JRn > JRm. and to each map f : M > N a bundle map (f. . there is an equivalence TU :::: (TM)IU such that the following diagrams commute. which has all the properties any one of these particular bundles T(M.
v]q � (q. : TV > T(x(V» .. since x. the equivalence class of (x. These equivalence classes will be called tangent vectors at p.(e) would be (y(p). JRn. Here v is just an element of (and every v would occur. but it is a little ticklish to produce. since (x(p). I (V) > V x JRn. because D(y o xI)(x(p» is an isomorphism from to Our definition of TM provides a oneone onto map (b) JRn JRn . v).76 Chapter 3 The construction is much easier to understand if we first imagine that we al rearfy had our bundles TM. namely [x. V) is a coordinate system. If JRn. v + w]p a · [x. w]p = [x.. v) p (y. and this would be an equivalence (with inverse (xI). w) by y. w) w = D(y o x. and TM is defined to be the set of all tangent vectors at all points p E M. IRn. maps . a . w) for some w E We can easily figure out what the relationship between v and w would be. v) is taken to (y(p). Since TV should be essentially (TM)IV. It is the clue which enables us to now define TM. this definition is independent of the particular coordinate system x or y.I (p) by Mp. Ix Ix : . W E we define if (a) is satisfied. we would have if JRn. If x and y are coordinate systems whose domains contain p.. v). though TMp might be better.). so we want I (A) to be open for every open A C V x and thus we want any union of such sets to be open. we would have a map x. in conformity with the notation p.. (x. v]p = [x. then y.I (p) isomorphicaIly onto {p) x If y is another coordinate system. We now have a bundle . We want this to be a homeomorphism. v) will be denoted by [x. I = (y 0 xI).. We will denote the fibre . and v..I (p) would be taken by x. JRn . 0 x.. v]p.I ) (x (p»)(v). v]p + [x. is supposed to be the old (y 0 XI). the map " takes p equivalence classes to p. : TM > M. Then (x. v]p. There is a metric with exactly these sets as open sets. We define a vector space structure on . (a) It is easy to check (using the chain rule) that p is an equivalence relation.I (p) by the formulas [x. This condition makes perfect sense without any mention of bundles. and T(x(U» should essentially be x(V) x a point e E .. to some (x(p). JRn JRn). and (y 0 xI). so we leave this one part of the proof to Problem l.
(y. In II I = by (c) CJ '1 C2 if and only if: The equivalence classes. then TM is equivalent to T(M. which is equivalent to T(M. [y. v]p) = O f course. > is an imbedding. i) will play no further role in this storythe abstract substitute TM will always be used instead. e) > M. But T(M. i). we next ask how fortuitous this was. If (x . : M IRk . For the first example. and I is the identity coordinate system of then If . it is trivial. and we proceed to define two different such bundles. V) is a coordinate system around p. D (y 0 I 0 xJ )(x(p») (v)]J(p)' I. T'M. V) and respectively.The Tangent Bundle 77 I : M > and (x. In II. each defined on some interval around 0. D(i 0 x J ) (x( p»)(V)] i (p) the composition is easily seen to be an equivalence. Having succeeded in producing a bundle over each M. mapping to have the same derivative at O. •:. IR IRn. V) is a coordinate system around p.([x . IRk i. V) are coordinate systems around p and I(p) . Can one find other bundles with the same properties? The answer is yes.. i lp . if (x. since in M there are more coordinate systems around p than there are in U C M. ( i ( p) . though perhaps confusing to the novice. To prove (2). For I : M > there is a map h taking the '1 equivalence class N x 0 CI and x a C2. will be the elements of our new bun dle. V]p) = [I. with c (o) = p. i ). D (i 0 x ')(x( p») (v)) E ( M . . In fact. for all p E M. IRn Ix Henceforth. it must b e checked that this definition is independent o f x and y (the chain rule again). the only difference is that each equivalence class for M contains some extra members. In fact. to prove commutativity of the diagram. we define (c) N. the fibre of TV over p E V is almost exactly the same as the fibre of TM over p. we define In. Condition (3) is pI"actically obvious also. the bundle 1C : TM > M will be called the tangent bundle of M. ([x . we consider curves c : (e. we define to be where I is the identity map of and is defined in (b). Condition (I) of our theorem is obvious.
Then n J(x) = J(x) .J(O) = Jo I hx'(t)dt = l0 I L Di J(tX) . In the second example. but it is not a priori clear what its dimension is. suppose that J = 0 in a neighborhood of p. There is a function h : M > with h (p) = 1 and support h C (0). and define i(f) as i(fh). .) of c to the J equivalence class of J 0 c. For.f(O). clearly i (f) = i (g) if J = g in a neighborhood of p. . let hx(t) J(tx).. 2. = :s .g) = i (f) . h corresponds to J•. This comes out of the following. Coo i (fg) = J(p)i(g) + g (p)i(f). The set of all such operators is a vector space. If J is defined only in a neighborhood of p. y' (0) = V.) For x E U. this is defined for 0 :s t I. then O = i(f . Thus. . . Then there are Coo functions gi Coo function in a convex open neighborhood U of 0 : U > with JR . This condition is actually true for any derivation i. vlp corresponds to: the 'J' equivalence class of where is a curve in with y JRn under this correspondence. Let J be a in (I) J(x l . we may use this trick to define i (f): choose h to be 1 on a neighborhood of p. if J = g in a neighborhood of O. with support h C JI (0).. We have already seen that the operators i = have this property.i(g). • JRn . Without bothering to check details. . (2) gi (O) = D. We define a tangent vector at p to be a linear operator i which operates on all functions J and which is a "derivation at p": xl o y. with J(O) = O. since U is convex. For these operators. r i=1 Therefore we can let g (x) = J� Di J(tx) dt :PROOF. . LEMMA. we can already see that this example is "really the same" as TM [x. (The second condition actually follows from the first. Then a/axi lp JI Coo JR 0 = i (O) = i (fh) = J(O)i(h) + h(O)i(f) = 0 + i(f).xn) = L:7= l xigi (X I . xi dt. things are not so simple.xn) for x E U.78 Chapter 3 (/.
a]p. l) = l · i(I) + l · i(l). choose gi as in Lemma 2. From Theorem 3 we can see that. U) is a coordinate system around p. a bundle constructed from all derivations at all points of is "really the same" as We can let correspond to the formula a/a Ii lo IRn M x M TM./(0). Consider the case where M = JRn and p = o. •:. they are clearly linearly inde pendent. It is a simple exercise to use the coordinate system to transfer this result from to . if (x. Assume U is convex. then Mn span this vector space. for the function / . . a�n Ip (so i is determined by the numbers i(xi »). This shows that span the vector space. . Notice that i(l) = i(l . Given / on U. THEOREM. once again. i(l) = 0 for any constant function c on U./(0)) = e(£) igi ) coordinate function) 1= 1 = I )i (Ii )gi (0) + I i (O)i(gi)] 1= 1 ith PROOF. In fact. Hence i(c) = c .e can be written a�I Ip ... and any derivation . The set of all linear derivations at p E is an ndimen sional vector space. Then (Ii denotes the i(f) = i(f . so i(l) = o. [x.The Tangent Bundle 79 3. .
TJRn 0 f. then Notice that if denotes the identity coordi� ate system on P corresponds to when we IdenllfY wIth We will usualJy make no distinction whatsoever between a tangent vector and and the linear derivation it corresponds to. In fact. the map which corresponds to can be defined as follows: (x.. . b ) . � . and the map f. that is.. If e : JR M is a differentiable curve.=1 a a!. and to is calJed the tangent vector to c at 10.IP 1= ax consequently. (i)](g) = i(g fl.v)(g) = v(g fl.a]p most easily analyzed from the relation (f. en (JRn) .80 derived in Chapter 2.a) 1=1 axi n a L1 a1. a tangent vector is often most easily described by telling what derivation it corresponds to. 0 It is customary to denote the identity coordinate system on write JRI by this is a basis for JRlo. ayj a' (p). JRn. and this is precisely the equation which says that p (y . t i . is often j . we wilJ not hesitate to write v(f) for a differentiable function f defined in a neighborhood of p. shows that Chapter 3 if and only if bJ = L.. We wilJ denote it by the suggestive symbol dt 10 de l · . between E v Mp x ap [f. It is easily checked that under this correspondence. then > dt 10 d I for I. . I [x.
an ) E ]R2n . xn (v)) E ]R2n is a homeomorphism from (TM)IV to x(V) ]Rn . V) is another coordinate system. This map.I (I). This expression shows that y. a natural way to put a structure on If > then every element E is is a chart on uniquely of the form "I" dl will often stand for de dl I ' I ]R. Then the map v � (x l (Jr(v» .xl(v). . Di (y1. 1=1 1=1 This shows that if (I. . L.I )(t)). . oxI )(x(p)).. V ]Rn. .a) = (I I . as we have already seen. . but it is a little delicate to state. x: V ]Rn CooTM M. when we identifY TV with V ]Rn in the standard way.a) (y ox .). As you might well expect. and quite a mess to prove.xn (Jr(v)). Since locally looks like x clearly is itself a manifold.. 0 = 0 0 0 then.a l . . . . The tangent bundle of a manifold has a little more structure than an arbitrary nplane bundle. . ax L.. If (y.I (l. is simply the map x. . (x Jr. TM TM. L:7=1 ai Di (yn x.x).I is Coo. moreover. (x. .). I TM. so it has been quarantined in an Addendum to this chapter. = V 0 b1. L:7=1 ai Di(yl xI )(I). .The Tangent Bundle 81 This symbol will be subjected to the standard abuses one finds (unexplained) in calculus textbooks: the symbol de the subscript now denoting a particular number E as well as the identity coordinate system. . then y. (x. and Let us denote x x p Jr( ). . . . I n . . = � a' ay}i (p) = � a'. . TM Coo(TM) I V v a i by Xi (v) . it is no accident that our second and third examples turned out to be "really the same" as There is a general theorem that all "reasonable" examples will have this property. there is.
82 Chapter 3 We thus have a collection of COOrelated charts on which can be ex tended to a maximal atlas. In general. The functions ai are continuous or Coo if and only if X: U > uous or Coo. If X and Y are two vector fields. U). . A section of is called a vector field on for submanifolds of a vector field may be pictured as a continuous selection ofarrows tangent to The theorem that you can't comb the hair on a sphere just states that TM. for Coo vector bundles we can also speak of Coo sections. if f: M JR. IRn. then for any coordinate system (x. (fX ) (p) = f(p)X(p). the local trivializations x* are Coo. in some If we think of as the set of derivations. Y. Recall that a section of a bundle 1C : E > B is a continuous function B > E such that 1C o S = identity of B. s: M there is no vector field on S2 which is everywhere nonzero. TM for all pE U.. we define a new vector field TM is contin X+Y by Similarly. With this Coo structure. Vector fields are customarily denoted by symbols like X. we have Mp) . We have shown that there do not exist two vector fields on the Mobius strip which are everywhere linearly independent. M TM M. or Z. a vector bundle 1C : E > B is called a Coo vector bundle if E and B are Coo manifolds and there are Coo local trivializations in a neighborhood of each point. It follows that 1C : E > B is Coo. and the vector X(p) is of ten denoted by Xp (sometimes X may be used to denote a single vector. we define the vector field fX by > (X + Y)(p) = X(p) + Y (p) .
Of ten. and X is a vector field. The reason for this is that if A : !F !F is any deriva tion. X(f + g) = X (f) + X(g) X(fg) = IX(g) + gX(f). In fact. if X (f) is Coo for every Coo function I. then X (f) is Coo for every Coo function I. Clearly. We have just seen that a Coo vector field X gives rise to a function X: !F !F. indeed. and the operator Xp thus defined is a derivation at p . we clearly must define Xp(f) = A (f)(p) . and I are Coo. iflocally n a X(p) = L a'. then A = X for a unique Coo vector field X.The Tangent Bundle 83 Clearly X + Y and IX are Coo if X. al X(f) = � a ' axi' 1= ) which is a sum of products of Coo functions. thus X is a "derivation" of the ring !F. (p) axi I ' P 1= 1 then . If > a newf unction > > > . a Coo vector field X is identified with the derivation X . then X is a Coo vector field (since X (xi ) = a i l. On U we can write a/axi now denoting the vector field the symbol I: M IR is a Coo function. Conversely. It is not hard to check that if X is a Coo vector field. Let !F denote the set of all Coo functions on M. Y.L. then we can define X (f): M IR by letting X operate on I at each point: X(f)(p) = Xp(f).
We call . . . For the present.. v'n) equally oriented if det A > 0 (i. Its sense is nicely conveyed by the definition in The American Heritage DictiollaT ': } "To understand profoundly through intuition or empathy". The class to which . Vn) and ( 'l . in particular on the tangen t bundle of a manifold.e. .) = v'. A simple example of the latter is the map I : IRn > IRn defined by I (x ) = (x l . To provide this extra structure. V (VI. . vn] . if I is orientation preserving) and oppositely oriented if det A < o. . purportedly as a word from the Martian language. The nonsingular linear maps I : V > V from a finite dimensional vector space to itselffall into two groups. The relation of being equally oriented is clearly an equivalence relation. If JL denotes one (VI. since we have not yet defined anything called "orientation". we will discuss just one new concept about manifolds. which is being preserved. The terminology "orientation preserving" is a bit strange. There is no way to pass continuously between these two groups: if we identify linear maps IRn 7 IRn with n x n matrices. . Either of these two equivalence classes is called an orientation for V. . . . . . . so that if JL is an orientation of V. Vn) E JL if and only if . the matrix A = (aij) of I is given by the equations V. which arises in this very way from the notion of "orientation" in a vector space. . . . . _ xn ) (reflection in the hyperplane xn = 0). divid ing the colJection of all ordered bases into just two equivalence classes.84 Chapter 3 The tangent bundle is the true beginning of the study of differentiable mani folds. those with det I > 0. . . [VI . . and you should not read further until you grok it. [VI . .l . . . x n . . .. One basic theme in all these chap ters is that any structure one can put on a vector space leads to a structure on any vector bundle. j=J n = * A cult word of the sixties) "grok" was coined. this clearly makes no sense unless we supply V and W with more structure. Vii L Qji Vj. (VI. by Robert A. we note that two ordered bases . vn) and (V'l .' The next few chapters constitute a detailed study of this bundle. . 'n ) for V determine an isomorphism I : V > V with I(v. . The problem becomes more acute if we want to define orientation preserving isomorphisms between two diff erent (but isomorphic) vector spaces V and W. (VI . vn) belongs will be denoted by . . then . and thus with IRn \ then the orientation preserving and orientation reversing maps are disjoint open subsets of the set of all nonsingular maps (those with det # 0). . Heinlein in his pop science fiction novel Stranger in a Strange Land. linear transformations in the first group are called orientation preserving and the others are called orientation reversing. and those with det I < 0. . vn] = JL. .
then (' automatically satisfies the same condition. v) are two ndimensional vector spaces. 1R2.I (U) U x IRn is an equivalence. en (X) X x IRn [(x. and (W. en] then det (aij) : > is continuous and never o. . (x. ej ).) = I:>ji (X) . = For the trivial bundle we can put the "standard orientation" on each fibre If > is an equiva lence. (x. . then is either orientation preserving or orientation reversing on each fibre. since rr. . and is connected. f: en (X) en (X) X f aij X IR n f(x. fL X IR fLp rr. for if we define the functions : > by f: [f(VI )' .I IRn fL) fL. If 1r: E > a non trivial nplane bundle. and the fibres of U x IRn are fL B is Notice that if this condition is satisfied for a certain t. vn ). (VI . . . then ( is either orientation preserving or orientation reversing on all fibres. and t' : > is another equivalence. . an orientation of E is defined to be a collection of orientations (p) which satisfy the following "compatibility condition" for for any open connected set If t : > given the standard orientation. . and 1R' orientation of V. . j=1 U C B: rr. an isomorphism V > W is called orientation preserving (with respect to and vI if = v whenever = if this holds for any one it clearly holds for all. f(vn)] [VI . ed. Vn ] fL. . . [eI . .I (U) U x lRn . . .e. en )] {x} x IRn .. the other will be denoted by and the orientation for will be called the "standard orientation". ··. . Now if (V .The Tangent Bundle 85 VI WI Examples of equally orien ted ordered bases in 1R.. together with orientations.
TM. fLp fL fL (O. we call itself orientable or nonorientable de pending on whether is orientable or nonorientable. I] x [. To see this we standard orientation. If I'01. I]. For... sn.a) a).} but not every bundle has an orientation. considered as a Idimensional bundle over SI . has no orientation. then A just looks like the boundary of the Mobius strip obtained from [0. since on which we have the C also orientable. we can pick two vectors from each fibre so that the totality A looks like two sections.fL) .I. For example. an oriented bundle is just a pair (�.86 Chapter 3 > : x x is an equivalence. in particular. to the tangent bundle of a Coo manifold In this case.I (p) with = A bundle is called orientable ifit has an orientation. the Mobius strip. I } with identified with ( I . I] x { .I IRn is M fL s: M [s(p)] fLp. an orientation of is also called an orientation of and an oriented manifold is a pair where an orientation for The manifold is orient able. TM IRn M. s(p) fL is M.1 U IRn U IRn U B. and nonorientable oth erwise. The sphere s(p) fLp . it has another orientation = { . This shows that the orientations define an orientation of E if the compatibility condition holds for a collection of sets which cover If a bundle E has orientation = {fLp }. TIRn :::: en(IRn). we can let A be [0. fLp . TM TM (M. although the Mobius strip has no nonzero section.fL) where is an orientation for �. This definition can be applied. we had compatible orientations we could define a section Sl > by choosing to be the unique vector E A n .I . For example.
The Tangent Bundle 87 note that for each p E Sn. V2] JLp . . 2 sin e.. v) p I IR'A Vi (p.O) } c there are continuously varying vectors but that it is impossible to choose continuously from among the dashed vectors for = 1. JLp P E S.i. (A(p). which also discusses the tangent bundle of a manifoldwithboundary.VI . A(v» . . 1)(0.I } thus defined is called the "standard arien tarian" of Sn. This can be seen by noting that for any two manifolds MI and M2 the fibre (MI M2 )p of T(MI M2) can be written as VI p V2p where (Jri). ·· .. so for VI .(vI l. This shows that if and E E then [vt. If we had orientations = and wp otherwise. w ] JL 11"2 p p p p vp .I . " Vn. since it contains the E B. .0») and their negatives. S2p 11"2 A : S2 > S2 A: IR3 > IR3 A.V2] JLA(p)' Thus the map A : S2 > S2 is .I ) E JLp if and only if (w. A (p) p. (vnI l) is in the standard orientation of IRnp . Vn. x = . so if = '' I t i 2" wp S M w [v . 0) i . Since TS I is trivial. The map ( is just � when is identified with a subspace of { x The map Uj) E Pl the bases is orientation reversing. the restriction �IB' Mobius strip (for any orientable bundle � = to any subset B' C B is also orientable). The torus Sl S I is another example of an orientable manifold. and consequently orientable.Ip we can define (VI .I E sn.i. by considering the "antipodal map" defined by = This map is just the restriction of a linear map defined by the same formula. A subset = { (2 cos e. . x x x Any nholed torus is also orientablethe proof is presented in Problem 1 6. For tlle imbedding of considered pre\�ously we have already seen that on the Ell M M (0. «0.(snIp) E TIRnp (Problem 21). if then we could simply choose The projective plane must be nonorientable also. A. The orientation JL {JLp : P E sn . [A. S2 JL is the standard orientation of S2 Jr : > are oppositely oriented. this shows that T(S I S I ) is also trivial.I the vector w = Pp E en (IRn ) :::: TIRn is not in i.: Vip > (Mi )p is an isomorphism and the subspaces Vip vary continuously (Problem 26).: S2p > S2A p) (p. Nonorientability of can be seen in another way. The Mobius strip is the simplest example of a nonorientable 2manifold.
. Condition (2) holds. For projective 3space the situation is just the opposite. In general...4. E ( ayi ) 0 U n v. If g : map � we obviously can define orientations vp for by requiring g to be orientation preserving. . and obtain an orientation of Although our original definition is easier to picture geometrically.'. 11"3 A : S3 S 3 > S3 11"3 11"3 > pn M An orientation JL of allows us to distinguish the subset . is orient able if there is a subset . (I) the domains of all (2) for all and V) .. which is impossible. these same arguments show that is orientable for n odd and nonorientable for n even. U) x(U) x. U) (y. U) E M... the map would then Vip } on be orientation preserving with respect to jj. = JL or JL. because it is just the condition that 0 is orientation preserving.4. which does not use the tangent bundle of at all. axj > TM (x.' as the collection of all for whieh x is orientation preserving (when x is given the standard orientation).4. (x. the antipodal map is the is orientation preserving. the determinant condition will be very important later on.4. for sueh that 11"2 S2n 11"2 s > f: M N > [p].4. given A' we can orient the fibres of in such a way that is orientation preserving. There is a more "elementary" definition of orientability.88 Chapter 3 "orientation reversing" (the notion of an orientation preserving or orientation reversing map makes sense for any imbedding of one oriented manifold into another oriented manifold of the same dimension). 11"2 f A p [pl.' of the atlas . > > (x. Conversely. In this case.' cover M M det on . From this fact it follows easily that is not orientable: If had an orientation v = {V[pl } and g : is the map P � then we could define an orientation by requiring g to be orientation preserving.:IRnTMI U T(x(U)) :::: x(U) IRn (y xI ). since jj.: T(x(U)) T(x (U)) TMI U x* TM. According to this definition.
. for certain equivalences 1m. THEOREM'... is. Ifwe have a bundle T'M over M for each M. Nevertheless. Those who have been through this sort of rigamarole before know (i. have faith) that it's going to work out. JRn.. 4.. the welcome symbol ... N. old f ) on * Euclidean spaces. for certain equivalences (3) of Theorem I. T'U '" (T'M)IU. * Functorites will notice that Theorems I and 4 say that there . T'M TM � TN f' M . eMl PROOF. What we have to hope is that our conditions on TM and T'M make them "look alike" in a sufficiently strong way for this idea to really work out. up to natural equiv alence. those for whom this sort of proof is a new experience should find it painful and instructive.. (2) of Theorem I.The Tangent Bundle 89 ADDENDUM EQ UIVALENCE OF TANGENT BUNDLES The fact that all reasonable candidates for the tangent bundle of M turn out to be essentially the same is stated precisely as follows. the idea behind the proof is simple enough. then both (TM)IU and (T'M)IU "look like" x (U) x so there ought to be a map taking the fibres of one to the fibres of the other. then there are equivalences such that the following diagram commutes for every Coo map eM ' TM . occurs quite a ways on. T'M � T'N leN The details of this proof are so horrible that you should probably skip it (and you should definitely quit when you get bogged down). N satisfying f) f' (I) of Theorem I.. If (x.e. a Ullique functor from the category of Coo manifolds and Coo maps to the category of bundles and bundle maps which is naturally equivalent to (en. and a bundle map (jil> for each Coo map M . U) is a chart on M.:. and to the restriction of the functor on open submanifolds..
TV � T(y(V)) =. (TlR")ly(V) '" IY l. Our main task is to show that this isomorphism between the fibres over p is independent of the coordinate system (x. e"(IR")ly(V) Ie <D Ij. e"(IR")lx(U) (1) (V (TM)I V =. which are denoted by the same symbol '" . U) be a coordinate system on M. (I) Suppose V maps C U is open and y = xIV We will need to name all the inclusion i : U > M I : V > M j : V > U k : y(V) > x(U) . consider the following diagram. are the equivalences mentioned in condition (3).TU � T(x(U)) � (TIR")lx(U) ' "lx(Ul.' for T'. To compare ax and ay.90 Chapter 3 ax = Let (x.).1 Similarly. ® Ik' ® Ie 0 Ie . Then fJx. we can define fJx . Two of them. so it takes the fibre of TM over p isomorphically to the fibre of T'M over p for each p E U. (TM)IU ""'=". 0 (". U).1 0 ax : (TM)IU > (T'M)IU is an equivalence. using equivalence ". This will be done in three stages. Let ax denote the composition (/nlx (U» 0 '" 0 x. Then we have the following string of equivalences.
> > i. using only properties (I). (2). JRn JRn This means that for P E V.1 oax ay fJx fJy. The two triangles on the left commute by con dition (3) for TM. ax. Again. as shown below. i e� TV Square ® commutes because k ° y = x ° Square ® commutes for the same reason as square (D. (2). we enlarge it. and the one on the right commutes because ° j I. Chasing through diagram (1) now shows that the following commutes. it will be proved for TM (where it is actually obvious). the isomorphism between the fibres over p is the same as Clearly the same is true for and since our proof used only properties (I).1 oay = fJx. To see this for square (D. fJy. . Thus on the fibres over p . not the explicit construction of TM. so that it is also true for T'M.The Tangent Bundle 91 Each ofthe four squares in this diagram commutes. i = (TM)IU (TM) I V 1e �Ti "Z ™� I . and (3). the inclusions x(U) and y(V) come into play. Square @ obviously commutes. (II) We now need a Lemma whieh applies to both TM and T'M. and (3). for every p E V.
where every :::: comes from the fact that some set is an open submanifold of another.92 Chapter 3 LEMMA. This implies that the two compositions TA =. Everything in this diagram obviously commutes. There is a map j : JRn JRm with j f on A . then the > TA =. p E A' C A . A C JRn f." are equal on Case 2. en (JRn) laid j. em (JRm) and are equal and this proves the Lemma in Case I. p. following diagram. j/' and JRn JRm > . p E A. B C JRm are open. and f : A B is Coo. A. General case. A'. Consider the JRn and j: B JRm are the inclusion maps. since the maps "old "old f. . (TJRn) IA � en (JR)n IA TB =. (TJRn ) I A � en (JRn) I A � en (JRn ) old j. If and following diagram commutes. For each we want to show that two maps are the with j = f on an same on the fibre over Now there is a map j : open set where We then have the following diagram. where i : A > > > 1 101d f• = . (TJRm) IB � em(JR)mI B PROOF Case 1.
@. To see that square @ (which has a triangle within it) commutes. (TlRn)IA �8n(lR)nIA > i. . and @ commutes by Case 1. it suffices to prove that v �l TA' TA y A 7 IRn T lRn :::: (A) (/C) � (V) . . and other maps have also been named. � . (J O i To prove that A 0 i. . for ease of reference. which amounts to proving commutativity of the following diagram. (2) / / . this does commute. (T lRn) I A o A o i* = v O IL O K./ � / (TA)IA' '" @ e @ e Boxes (D. we imbed it in a larger diagram. and ® obviously commute. in which j: is the inclusion map. . = � 0 /C . � . TA +. :::: (T lRn) IA' Je (�) Since j 0 i is just the inclusion of Y T lRn A' in IRn . since v is oneone. . Thus it suffices to prove J* 0 i* = v 0 JL 0 K.The Tangent Bundle 93 and i: A' A is the inclusion map.
Diagram (3) thus shows that fJy = old (y 0 X. Now that we have a welldefined bundle map TM T'M (the union of all fJx. and the rectangle commutes by part (II). which proves the Lemma. as well as to y and ylV n v.. Vn TA � TB =. with the composition TA =.1 0 ax follows immediately. = f� oeM is > Exactly the same result holds for T': left as a masochistic exercise for the reader. on the subset (TA) I A'.l o ay fJxl o ay (3) (TM)IU "":::. The proof that eN 0 f./ (y o [I). •:. e n (JRn)lx (U) T(x(U)) =. because part (I) applies to x and xiV n V. we can assume without loss of generality that V = V. 0 fJx. (TJRn)lx(U) v. T(y(U)) =. we have the following diagram.I). V) are any two coordinate systems with p E To prove that and induce the same isomorphism un the fibre of at p.I o ax). it is clearly an equivalence eM. and on A' we can replace "old h" by "old J/'.TU � . .94 Chapter 3 positions are equal in a neighborhood of any p E A. (TJRn) I A � en (JRn) I A old J. (TlRn)ly(U) / nly(U). The desired result fJyI 0 ay = fJx . em (JRm)IB. Assuming V = V. e n (JRn)ly(u) The triangle obviously commutes. and are thus equal. V) and (y. old (y o [I). TM fJy. l nlx(U). the two com Commutativity of diagram (2) shows that the composition (III) Now suppose (x. In other words. (T JRm)1 B � em (JRm)1 B coincides.
and define p on M by 00 1 p(p.The Tangent Bundle 95 M be any set. b e such a sequence for each open set Xi(Vi) . Show that p is actually a metric on M. (e) Suppose that for every p . . 1] by _ f g. a # 0.j in a single sequence GJ. j 1 . Gi(q»). this is not quite true: (a) Let M = JR U {*}. Define gj.q) = iI: 2i d (Gi(P). and Xi' ( Bi) nXjI (Bj) 0. 3. .j. q E xj' ( Bj).* X2 (*) = 0.j (P) P E Vii· P¢V Arrange all gi.q) may be ° for p # q) such that p is a metric on each Vi and each Xi is a homeomorphism: (b) If A C JRn is open. . and { (Xi.{O} U {*}. A 2 . then there is some Ji with Ji ( p ) ¢ fi(A nC) . G2 . Hint: First arrange in a sequence all pairs (Ai. Vi)} a sequence ofoneone functions Xi : Vi JRn Vi C M and X (Vi) open in JRn.{ ° i. A3. by showing that every neighborhood of ° would have to intersect every neighborhood of * .Aj) of part (b) with A. (P) . such that each Xj Xi' : Xi(Vj n Vj) Xj(Vi n Vj) is continuous.. Let with > PROBLEMS 0 > = = > > we x > > > = • • • = . q E M there is a Vi and Vj with E Vi and q E Vj and open sets Bj C Xi(Vi) and Bj C Xj(Uj) so that p PE XiI (Bi). which "separates points and closed sets": if C is closed and p E A . with X 2 : V2 JR defined by x2 (a) = a. can find on M a pseudometric p (a function p: M M JR with all properties for a metric except that p(p. 2. It would seem that M ought to have a metric which makes each Vi open and each Xi a homeomorphism. then there is a sequence A I . with support Ji c A. 1]. and let V2 JR . C Aj. (c) There is a sequence of continuous functions Ji : A [0. G3. Never theless. .). 1. Show that there is no metric on M of the required sort.C.. . Let VI JR and XI : VI JR be the identity. . . =1 Show that p is the required pseudometric. (d) Let fi. let d be a bounded metric on JR. where * ¢ JR.j : M [0. of open subsets of A such that every open subset of A is a union of certain Ai 'S. Actually.
... Iy : . Sn (P) (x... Show that in the definition of an equivalence it suffices to assume that the map E I 7 E2 is continuous.. then all Ix are homeomorphisms.... . continuity of follows automatically from continuity of ] : E I .I (p) . v) .. Show that in the definition of a bundle map. [y . but weakly equivalent. is a homeomorphism for a collection (Xi . V IRn). where ] is an isomorphism on each fibre. 3. are linearly independent for all E B. E with the vector of is a section. [x. (b) Show that an nplane bundle � is trivial if and only if there are II sections E SI . f. · . fl 5. . on cach fibre. V IRn . (c) Check the remaining details in Theorem l.I (p) s: s(p) p SJ (p).e. maps on TM. B. Given a bundle map (]. bundles over the following base spaces: (i) the disjoin t union of twO circles. (b) Show that if there is a metric on TM such that Ix. f: BI . . V IRn .. 7. • 8. .I (A) for A C V IRn open are exactly the sets of the form Iy.. Find two inequivalent.. w]q � (q..I (V) ... (Ii) a figure eight c::><::::) .. and is a homeomorphism of B ontO itself.. B . v). (To prove the inverse continuous. (c) Conclude from Problem I that there is a metric on TM which makes each Ix 2. (a) Check that p is an equivalence relation on the set of pairs (b) Check that the definition of is independent of the coordinate systems x and y which are used. V) are two coordinate systems. x x x x a homeomorphism . (iii) the torus. v]q � (q.I (B) for B C V IRn open. (c) Show that locally every IIplane bundle has 11 linearly independent sections. show that ] = g 0 h where g and h are contin uous maps such that h takes fibres linearly to fibres.. (a) Show that for any bundle . SII which are everywhere linearly independent.. E2.I ( V n V) the sets of the form Ix. the map 0 . A weak equivalence between two bundles over the same base space B is a bundle map (].. (a) Suppose (x..I ( V) . Show that in . w). note that locally it is just a map x x V IRn . Vi ) with M = Ui Vi .. f B2 fl .. i.. ..96 Chapter 3 V) and (y... giving rise to two Ix : . .. while g is an isomorphism 4. : E .. . 6.
If f: M N and f* is the 0 map on each fibre.a/ax.g E :Fp.a.The Tangent Bundle 97 xl . space (:F / W) * . 1 3. If g : JR JR is Coo show that g(x) g(O) + g' (O)x + x2h(x) for some Coo function h: JR R 1 2. (x) = 0 1 x .. (a) Let V be the vector space ofall C1 functions f: JR JR with frO) = 0. and p JR be a linear operator with eUg) 0 for all f.:: o. L. JR" [x. If V is a finite dimensional vector space over JR. (a) Show that the correspondence between and equivalence classes of curves under which p corresponds to the "j equivalence class of . = x 9. x ::: Show that all f. define a Coo structure on V and a homeomorphism from V V to TV which is independent of choice of bases.lp ' f* [f* (il](g) erg 0 fl.. for v.. W E V we will denote by E Vw the vector corresponding to (w. 10. (b) Let W be the vector subspace of :Fp generated by all products fg for f. .e has a unique extension to a derivation. f* [x. show that x l + W. If x is a coordinate system with X(p) 0. . the same must be true p of :F / W.o y for y a curve in with y ' = makes correspond to In. 1 1. p x"I + W is a basis for :Fp/ W (use Lemma 2). As in the case of JR". v). (c) Conclude that (V/W) * has dimension ce = 2 e• 14. as the next problem shows. (b) For O < e < I .a]p . p (c) Since (:F / W) * has dimension n dimension of M. The situation is quite different for C functions. . (a) Let :Fp be the set of all Coo functions f: M JR with f(p) 0. and let W be the subspace generated by all products. Show Jet e: :F that . v] (0) v. let { x +E O f. g E :F Show that the vector space of all derivations at p is isomorphic to the dual p. and that they represent linearly independent elements of V/W. Show that x f (x) /x 2 lim +o exists for all f E W. then f is constant on each component of M. are in V. (b) Show that under the correspondence the map can be defined by TM Vw > = > > = > = = = > > . .
n n g(p.t ). TM U TN > . TlHIn i: M M lHIna en(lHIn). then the rank of equals the rank of at g(a).98 Chapter 3 15.. Vn.vn_. Mp v) . i* (8M)p > v) lHIn E points inward. (Compare with Problem 233(d). More generally. If E a and x : > > f. p M lHIn. > is a subspace. Show that this subspace does not depend on the choice of X. v Mp p M U lHIn p. .] N x [0. (a) If is a manifoldwithboundary. fog g f M M. (b) Let a E IRn1 {OJ C A tangent vector in is said to point "in ward" if. (e) Show that if has an orientation JL.l ] ( )p Mp. A vector which is not in E is said to point "inward" if x. i. TP g: aM x [0. under the identification of with the vector is (a. i. so is a coordinate system around then directions". the tangent bundle TM is defined exactly as for elen.: Mp NJ(p). . the rank of at E is the rank of the linear transformation (b) If at a = where is a diffeomorphism. the still has tangent vectors "pointing in all vectors still run through IRn .ents of are p equivalence classes of pairs (x. then aM has a unique orientation aJL such that = JLp for = aJL if and only if every outward pointing W E (d) If JL is the usual orientation of show that aJL is (I)" times the usual orientation of IR 1 = a lHl . it is where a is the inclusion. . f o g f.) (e) Suppose we are in the setup of Problem 214.\IRnIX(p» C Mp vn v Mp x lHIn. lHIn .(V) lHInx(p) M [v" .t) '" (/(p). i. [w. (a) A map is an immersion if and only if is oneone on each fibre of TM. Show that this definition does not depend on the coordinate system x. in fact. . Although x takes a neIghborhood of E a onto rather than IRn. where > O. Define I) a Show that is obtained from I) by x. f: M N f p M f..) 16. (The reason for this choice will become clear in Chapter 8.v" .(aM)p .
a · b (a l + a2) b a l b + a2 . and there are no zero divisors: . Show that such a vector field can be picked so that near (0.g.. and v on M C P and N C P. .i) by € {to.(v) E (aN)J(p) .av) is orientationreversing.0. • • (a. and N is IJ X Let f be a diffeomorphism from M to N which is orientation preserving on one copy of and orientation reversing on the other. ..0.. v) E (S2.0) a = • = • = . a JL) (aN. b a (Ab) for A E ]R a · (1 .i)_p. b) = (Aa) .. b) . > (f) If M and N have orientations JL and v and f: (aM . show that P has an orientation which agrees with J1.a.B. b a (bl + b2) = a . Suppose we have a "multiplication" map from ]Rn x ]Rn to ]Rn that makes ]Rn into a (nonassociative) division algebra. which is diffeomorphic to ]R2. Show that identifying  TIP'2 is homeomorphic to the space obtained from T(S2. bl + a b2 A(a . What is the resulting manifold P? Sl [0. 18. (p. That is. (a "magnetic dipole"): 19. there is one on S2 0. v) (S2. I) the vector field looks like the foIlowing picture 17. (g) Suppose M is with two holes cut out.The Tangent Bundle 99 by identifYing v E (aM)p S2 with (.i)p with (p. . I)}.I ). Sl . Although there is no everywhere nonzero vector field on S 2 .
4.(S 2p) by showing that the inner product (p .100 Chapter 3 (For example. Show that for p E S2 .. el .) Let eJ . the quaternions are not commutative and the Cayley numbers are not even associative.ad + be). Conversely. that n = I. Multiplications with the required properties on ]R4 and ]Rs are provided by the "quaternions" and "Cayley numbers") respectively. Suppose that (TM)IA is trivial whenever A C M is homeomorphic to S l .0). for I. otherwise it suffices to assume the existence of a unique b with a . b) · (e. en are linearly independent. (d) Multiplication by a is continuous. = /l = 2 = 0.0. and for we can use "complex multiplication". the vector pp E ]R3p is not in i. see Cai£ulus on Manifolds. e'(O» ) = ° for all curves e with e ( O) = p and 1c (t)1 = for all t . el for a unique a E ]Rn. this condition implies that there are no zero divisors jf the multiplication is associative. a . and quaternions are the only associative examples.. Let M be a manifold. Arner. a · en on (snl. S..g '( t J'). complexes. = ( 1. . . ' " . .0. 0) = 2. nonexistence of zero divisors immediately implies that for a # ° there is some b with ab and b' with b'a = If the multiplication is associative it follows easily that b b'. . (c) If p a · el E snI. g)'(t) = (f'(t) '. Palais. It is a classical theorem that the reals. Monthly 75 (1968). > (1. i)p are linearly independent. or 8.. For a simple proof. where T : ]Rn ]Rn is a vector space isomorphism. 20. . Adams has proved. = (1. using methods of algebraic topology. . f Math. g(t» ) + (j(t). (0. (e) TSn1 is trivial. . (b) If a # then a . The tangent bundles TS 3 and TS7 are both trivial. 23. [Incidentally. (f) TlP'n1 is trivial.d) (ae . [0. where ' denotes the transpose. Tv).bd. see R. (a) Consider the space obtained from I] x ]Rn by identifying with ( I . J. . (a. pg. then the projection of a . v) 21. The Classification o Real Division Algebras. en be the standard basis of ]Rn.". e2 .0. (a) Every point in snI is a . . we can use ordinary multiplication. (b) Show that the resulting bundle is orientabJe if and only if T is orientation preserving. (Recall that 1 (f. so that we always have multiplicative inverses.) 22. b b . Hint: An arc e from Coo . Show that this can be made into the total space of a vector bundle over Sl (a generalized Mobius strip). 366368.J " /l = . . Show that M is orientabJe. a = O). F.
the fibre of � Ell ry over p is the direct sum 1C1 (p) Ell 1C'I (p). then 1C*(�) is not orientable. Y. (g) Let � = 1C : E be a vector bundle. Show that if � is not orientable. Let E' c Y x E be the set of all (Y. Since 1C : E is a continuous map from a space to the base space of �. This bundle is denoted by f*(�). Remark: Using results from the Addendum to Chapter 9. and define f : E' E by f(y. First consider pairs c.The Tangent Bundle 101 po E M to p E M is contained in some such A so (TM)lc is trivial. f) from �" to � which is an isomorph�sm on each fibre. (e) If � is orientable. Show that �" '" �' = f*(�). (f) Give an example where � is nonorientable. (c) If g : Z (d) If A c and i : A is the inclusion map. show that f*(� Ell ry) '" f*(�) Ell f*( ry). Show that 1C': E' Y is a bundle. then f*(�) is also orientable. (b) Suppose we have another bundle f' = 1C" : E" Y and a bundle map > > X > > a X > B B > B 24. e) = y. e') = 1C(e) = 1C'(e').e) with /(y) = 1C(e). then (f g )*(�) '" g*(f* (m . define 1C' : E' Y by 1C' (y. (a) Given an Ilplane bundle � = 1C : E and an mplane bundle ry = 1C' : E' let E" c E x E' be the set of all pairs (e. f(e)) E E'. and ( j. The next two problems deal with important constructions associated with vector bundles. then i*(�) '" �IA. Hint : Map e E E" to (1C"(e). Thus one can "transport" the orientation of Mpo to Mp . but f*(�) is orientable. (a) Suppose � = 1C : E is a bundle and f : Y is a continu ous map.e) = e. > B > B > B. e') with 1C(e) = 1C'(e'). It must be checked that this is independent of the choice of c. 23. and is called the bundle induced (from �) by f. A vector space structure can be defined on > > > > X X (f. The general. we can conclude that a neighborhood of some Sl C M is nonorientable if M is nonorientable. by using the vector space structure on 1C1 (f(y» . possibly quite messy. c' which meet only at po and p. It is called the Whitney sum � Ell ry of � and ry. the symbol 1C*(�) makes sense. . f) a bundle map which is an isomorphism on each fibre. together with Prob lem 29. Show that 1C" : E" is an (11 + m)plane bundle. (b) If f: Y > B. Let 1C"(e. case can be treated by breaking up c into small pieces contained in coordinate neighborhoods.
f. (a) If 1C : E > M is a COO vector bundle. find two nonorientable Iplane bundles and ry over such that Ell ry is also nonorientable. and 1CM [or 1CN] : M x N > M lor N] is the projection on M [or N]. (f) If � is orientable. (c) If M x N is orientable. then both M and N are orientable.x). (g) Define a "natural" orientation on V EEl V for any vector space V.e. show that � ry is also non = X � � � X Ell Ell Ell � 25. and each fibre 1C 1 is a Coo submanifold of E. the bundle T(TM) is (lrientable.s orientable. and use this to show that Ell is always orientable.) A different proof that TM is orientable is given in Problem 29. (b) The Osection of E is a submanifold.. i. . (c) Given bundles �i 1Ci : Ei > Bi. (a) Let be = [a()01C) /" . (e) If � and ry are orientable. then M x N is orientable.. and ry is nonorientable. 28. a ()o1C) /" a�I /" . a�n /J the form of Y* 0 (x*). the orientation for (TM)v can be defined as of x. carried diffeomorphically onto B by 1C. 27. (d) If Ll: B > B x B is the "diagonal map". Problem 5).!). U) be a coordinate system on M with X (p) 0 and let v E Mp I:7= 1 ai a/axi ip · Consider the curve c in TM defined by c rt ) = v + a�i /p . (b) If M and N are orientable.. show that � ry is orientable.1C2 (e2))· Show that this is a bundleI �I x �22 over BI x B2 . show that � ry :::: Ll* (� x ry). I . (Here is a more conceptual formulation: for v E TM. . then 1C* has maximal rank at .I shows that this orientation is independent of the choice (x.102 = Chapter 3 orientable. (h) If is a "figure eight" (c. Show that the Jacobian matrix of *( ) y* 0 (x*) I is of the form This shows that the manifold TM is alw. then T(M x N) :::: 1CM*(TM) Ell 1CN TN . define 1C : E x E > BI X B2 by 1C(eJ.·arh point.e2) (1CI (eI l. . Ll(X) = (x. . . (a) If M and N are Coo manifolds.(p) 26.
n) i v ' 29.. (a) Let Po E snI be the point (0. A sequence of bundle maps EI E2 l.. Using part (a). Tw) (for each v. . this informa tion will all be important in Chapter 10.. (a) If T : IRn IRn is a linear transformation. Show that I (po) is homeomorphic to SO (II . 30. prove that SO(II ) is connected for all II 0::. show that the matrix of T* is the transpose A t .. T* : IRn IRn. I v ' de a L (a) If � = n : E sequence > B is a o > Hint: (I) An element of the IOtal space of n*(�) is a pair of points in the same fibre. and then show that f.. so it is w . . > > > an exact (e. so it is connected. 0. = a di (O) = ax. E3 with f g = identity of B is exact if at each fibre it is exact as a sequence of vector space maps. is orientable if the other two are.I). (c) Show that 0 (11) has exactly two components. the adjoint of T. the map w .( Chapter I I). . show that there is n*(�) TE n*(TB) O.. (T*v.I) for all p E SnI . then each bundle E.. and induction on II.. (This is why the proof that the Mobius strip is a nonorientable manifold is so similar to the proof that the Mobius bundle over Sl is not orientable. which determines a tangent vector of the fibre. I. then the manifold E is not orientable. (c) T(TM) is always orientable. I). In addition to being used in Problem 32. 0:: 2 (b) SO(l) is a point.. .The Tangent Bundle 103 Show that (b) Find a curve whose tangent vector at 0 is a/a(X .. This problem requires some familiarity with the notion of exact sequences (c. w) for a unique T*v). (v. (2) Map X E (TE) e to E2 (b) If 0 EI E3 0 is exact. .n* X). > The next two Problems contain mOre information about the groups intro duced in Problem 233. w) = (v. For II define f: SO (n) > snI by f(A) = A (po) .I (p) is home fomorphic to SO (II . > > 31. Show that f is continuous and open. (d) If n : E M is not orientable. Tw) is linear.) > > > > Coo vector bundle. is defined by (T*v. If A is the matrix of T with respect to the usual basis.
show that H is continuous. means that f : X > Y and f( A ) c B. pg. A) > (Y. . Ji : (X. and use part (e). (Notice that this also gives us another way of finding the dimension of O(n).104 Chapter 3 standard basis. a continuous function H : X x [0. Two continuous functions f Ji : X > Y are called homotopic if there is a. A 2 where A. At = A. v) = 32. show that f : (IR" . A is always positive semidefinite. and A2 are continuous functions of A. A) to (Y. t ) may be thought o fa s a path of functions from Ho = fa to H. 1R) is homeomorphic to O(n) x IRn (n +' /2 and has exactly two com ponents. The map H is called a homotopy between fa and Ji . Conclude that a nonsingular linear transformation (IR" . Hint: If A (n) > A and A (n ) = A (" ) .IR" . It is a standard theorem that a symmetric A can be written as CDC' for some diagonal matrix D (for an analytic proof. w) (v. see Calculus on ManijiJlds. 1 . Hint: Consider At . A (n)2 . . A ) > (y. Show that C can be chosen orthogonal. We call /0.{O J . so that Ho and H. (a) If A: [0.IR" {OJ ) to (IRn. I] > Y such that Ji(x) = H (x . The notation f : (X. T: T T T* . 1R) is continuous and H: 1R" x [0. ) 01) GL(Il. ) converges. The functions H. (g) The matrices A.{OJ ) > (IRn. (c) A selfadjoint (or the corresponding symmetric A) is called positive semi definite if ?: 0 for all E IRn.) (f) Show that every A E GL(Il . IRn . IRn {OJ ). then some subsequence of {A (n) . by showing that eigenvectors for distinct eigenvalues are orthogonal. and positive definite if > 0 for all Show that a positive definite A is nonsingular. are homotopic as maps from (IRn. : (X. for A C X and B e Y.) (b) A linear transformation IRn > IRn is selfadjoint if so that = = for all W E IRn. B).{OJ ) with det > 0 is homotopic to the identity map. B)) if there is an H as above such that each H. B). (b) Suppose f : IRn > IRn is Coo and /(0) = 0. B) homotopic (as maps from (X. t ) = A(I)(X ) . (e) Show that a positive semidefinite A can be written as A B2 for some B. 122). A ) > (Y. : X > Y defined by H. A. If Df(O) is nonsingular. v # o. then is selfadjoint if and only if A is symmetric. i) i = 0. E O(n) and A2 is positive definite. Hint: Use the Schwarz inequality. (Remember that A is symmetric. IRn . (Tv. I] > GL(Il. while f(IRn .{OJ ) > (IRn .{OJ ) is T:   T . 1 ] > IRn is defined by H(x. v) T v (Tv. {A : det A > OJ and {A : det A < OJ. (d) Show that At . If A is the matrix of with respect to the (Tv.{OJ) C 1R" . IRn . = Ji . (x) = H(x. Tw) T v. IR) can be written uniquely as A = A.
]Rn . Hint: Define H(x. Let Mil C ]RN be a Coo ndimensional submanifold. We will say that f is orientation preserving at 0 if f 0 h is homotopic to the identity map I : (]Rn . then there is a collection e ofcharts whose domains cover M such that for every (x. (e) Notice that the condition on y o x. B. _x n ). if M is any (not necessarily differentiable) manifold. and let be the homeomorphism h : ]Rn > B. ]Rn . we will say that f is orientation preserving at p if T_ J(p ) 0 f 0 Tp is orientation preserving at O. . V C ]Rn are open. the map y o x . t ) = f{lx) for 0 < t S 1 and H(x . show that if M has such a collection e of homeomorphisms. . (d) For p E ]Rn . Check that this does not depend on the choice of C V.q for p.{OJ). let T ]Rn > ]Rn be Tp (q) = p + q. ]Rn . To prove that this definition agrees with the old one we need a fact from algebraic topology: If f : ]Rn > ]Rn is a homeomorphism with /(0) = 0 and T : IRn 7 IRn is T(x 1 . ]Rn . . U) and (y. f o /z : (]Rn . Show that if M is orientable. we can define M to be orientab!e if there is a collection e of homeomorphisms x : U 7 whose domains cover M. x n. Hint: Consider certain maps from appropriate open subsets of TM to S N . where U. then for any Coo structure on M the tangent bundle TM is orientable.The Tangent Bundle 105 homotopic to D/(O) : (]Rn . To prove continuity at points (x. Let C V be the open ball with center 0 and radius r. xn ) = .l . . Assuming this result.I is orientation preserving at X(p) for all p E U n V. O) = Df(O) (x). such that e satisfies the condition in part (d).{O)) > (]Rn . . If f: U > V is a p homeomorphism. V) in e. ]Rn _ {Oj) . JRII (Xl .I in part (d) makes sense even if y 0 x. (a) Prove that if N (i) no chord of M is parallel to v. (ii) no tangent plane Mp contains v.I . Thus.I is not differentiable. . : B. use Lemma 2. h(x) = then (� arctan IXI) x. 33. ]Rn _ {O)) > (]Rn .I such that M x M and .{O)). then precisely one of f and T 0 f is orientation preserving at O. q E M. . (c) Let U be a neighborhood of 0 E ]Rn and f: U > ]Rn a homeomorphism with frO) = O. > 2n + I.O). By a chord of M we mean a point of ]RN of the form p .{OJ) > (]Rn . then there is a vector v E S N .
Ann. 645680). of Math. of Math. together with the exis tence of a proper map f : M > jR. 17ze se/finl£rsections ofa smooth llmanifold in 2ns pace. Whitney.J the corresponding projection. jRN jRN. :nd Pollack. Elementary Diif lltial Topo/lJgy. In Munkres. Proofs may be found in Auslander and MacKenzie. if M is compact. Then we may show that M imbeds in jR2n +J using essentially the argument above. ff Ann. Di erentiable manifolds. given by Problem ·230 (compare Guillemin . then 1C I M is an imbedding. Show that 1C I M is a oneone immersion. with N = (n + 1)2). which gives the same result even for noncompact manifolds (H.106 Chapter 3 > (b) Let jRN. (c) Every compact Coo Ildimensional manifold can be imbedded in jR2n + J . A much harder J·esult of Whitney shows that f Mn can actually be imbedded in jR2n (H. 37 (1935). .J C jRN be the subspace perpendicular to V. Lectures on Di ff f f erential Geometo'. 45 (1944). In particular. Whitney. Di ferential Topology). lntmduction IJJ Di erentiable Manifolds and Sternberg. there is a different m sort of argument to prove that a notnecessarilycompact nmanifold M can be imbedded in some jRN (in fact. Note: This is the easy case of "''bitney's classical theorem. and 1C . 220246).
then I v· is the identity map of V' and if U V. It is clear that if I v : V V is the identity. Vn is a basis for V. we can define v" E V" = (V')' unambigu ously by for every A E V'. The linear function V'i depends on the entire set Vl . On the other hand. . The simplest case arises when we replace each fibre V by its dual space V'.l W' > W > JR . Vn . and the isomorphism from V to V* obtained by sending Vi to V'i is not independent ofthe choice of basis (consider what happens if Vl is replaced by 2 vl l. > W a are easily checked to be a basis for V'. . for finite dimensional V. = If V" (A) = 0 for every A E V'. . not just on Vi alone. defined by (f a f. . (j'A)(V) = A(jV). and then fit all these new vector spaces together to form a new vector bundle over the same base space. then there is a linear transformation f' : V· defined by rr. then (f = f ' .l ) . Recall that V' denotes the vector space of all linear functions A : V If f : V 7 is a linear transformation.CHAPTER 4 TENSORS Acommon feature. = I w· . then the elements V *i E V*. These simple remarks already show that f' is an isomorphism if f : V is. we replace each fibre (p) by some other vector space. . for (f' fl ' = I v' and > a a g: > g)' g' The dimension of V' is the same as that of V. . if VI. . then A(V) 107 0 for all A E V'. which implies that . In fact. if v E V. ll the constructions on vector bundles carried out in this chapter have a In each case. .
we can prove that there is no natural isomorphism from V to V'. (Problem 6 gives a precise meaning to the word "natural". The lack of a natural isomorphism from V to V' prevents us from constructing an equivalence between �. In contrast.) Now let � = 1f : E + B be any vector bundle. readers may ponder this question for themselves. of �. * (�')' is alwa equivalent to �. we will see later that in "most" cases �* is equivalent to �. the dual bundle �.108 v = O. If U c B.) + 1f'1 (p). If you B into a vector bundle. We construct the equivalence by mapping the fibre V y's of � over p to the fibre V" of � •• over p by the natural isomorphism. Let E' = B to take each [1f1 (p)]* to p. the bundle �. It is called the natural isomorphism from V to V'*. is an isomorphism. formulated only after the term had long been in use. and �. Once the meaning is made precise. and define the function 1f' : E ' and I : 1f1 (U) U X �n is a trivialization. it gives us an isomorphism At first it might appear that �. '" �. Chapter 4 Thus the map v >+ v " is an isomorphism from V (0 V" . Actually. for the present. (We first pick an isomorphism from (�n ) . then we can define a function + + peB U [1fI (p)]'. = . once and for all. by We can make 1f': E' requiring that all such [ ' be local tnvializations. to �n. in the obvious way: since the map [ restricted to a fibre. since each 1f1 (p) is isomorphic to However. this is true merely because the two vector spaces have the same dimension.
)I = y 0 (X )I ). Yo ' 0 (X . sections of T OM. the cotangent bundle T*M is actually a Coo vector bundle: since two trivializations x and y* of TM are Coorelated. �* . * * * We can thus define Coo. ' (in fact. ' and y.Tensors 109 think about how is constructed. The seclion df is called the differential of f. If w is a Coo section of T*M and X is a Coo vector field. Even if can be pictured geometrically (e. that dc/dI ll". is called the cotangent bundle of M. Recall that X is This means that df (� IJ (�I J t fl = CO (f) = = (f 0 C)' (IO) or to (f o e) d(f(e(t» ) dl I. is a section of then we can define a function from B to � by � �* �* . there is seldom a operates on If s is a section of and (J geometric picture for Rather. to . the re sulting bundle. where e(lo) = p.. * the same is clearly true for x. the fibre of T*M over p is (Mp)*. denoted by T*M. then a Coo section df of T*M can be + defined by df(p)(X) = X(f) for X E Mp.g. in particular. Suppose. then w(X) is the Coo function p >+ w(p)(X(p» . it will appear obvious that this map is indeed an equivalence. �* �: � � P >+ (J(p)(s(p» s(p) E 1C 1 (p) (J(p) E 1CH (p) = 1C 1 (p)* . This function will be denoted simply by (J(s). Like TM. if is TM). If f: M � is a Coo function. When this construction is applied to the tangent bundle TM of M. as well as continuous.
dxn(p) isjust the basis of M/ dual to the basis alax' lp. ..=1 if c define sums of sections and products of functions and sections in the obvious way ("pointwise" addition and multiplication). \. .=1 n W = L w. . . .. The section w is continuous or Coo if and only if the functions Wi are. this equation takes the nice form df dl (�) If (x. The section df must have some such expression.. We can also write W(p) = L Wi (p) dx i (p). then the dxi are sections of Applying the definition.. In fact. alaxnlp of Mp.IrO Chapter 4 Adopting the elliptical notations de dl for dt r ' de l dg(l) dl for g' (I). dt T *M over U. . . This means that every section w can be expressed ulliquely on as V for certain functions Wi 011 U. U) is a coordinate system. dxi . we obtain a classical formula: . we see that = d(f(e(t» ) . . . . Thus dx'(p).
df must be a function of this change. df(p)(Xp) = Xp(f) = " a' . which became tangent vectors. THEOREM. no matter how obscurely expressed. af . The dx i themselves then metamorphosed into functions. a tangent vector. i. like dc/dt . it was only a matter of making new definitions. =L axi (p) dx 1=1 •:.e.just as Leibnitz had.=1 X.. If (X. except for quotients of infinitely small quantities. then on U we have P ROOF If Xp E Mp is then Thus n . Classical differential geometers (and classical analysts) did not hesitate to talk about "infinitely small" changes dx i of the coordinates x i . . ' (p)(Xp) . No one wanted to admit that this was nonsense. one can usually see that. all classical notions involving infinitely small quantities became functions on tangent vectors. Looking back at the classical works fi·om OUf modern vantage point. Since df is supposed to be the infinitesimal change of f under an infinitesimal change of the point. and waiting for everybody to catch up.TeIlS01J 111 I . Eventually it was realized that the closest one can come to describing an infinitely small change is to describe a direction in which this change is supposed to occur. which means that df should be a function on tangent vectors.(p) L a . which preserved the old notation. because true results were ob tained when these infinitely small quantities were divided into each other (pro vided one did it in the right way). like df. and it became clear that they must be distinguished from the tangent vectors a/ax" Once this realization came. U) is a coordinate system and f is a Coo function. In short. this point of view was in n af .
) = L: � (c(I» dl ax' . . the diff erential d was usually introduced in the following way: l Let I be a function of the X l .. =j = I 0 c : � + �.. . . Then I becomes a function of I.=1 . = j = l oxI). say I I(XI. Then 1 0 C(I) = j(x l 0 e(I). x"(I» . is still used by physicists. . no matter what tilef unctions X i (I) are... . which suppresses the curve C. . In fact. We now have We now have dl = � af dX i .](x(e(I» ) . 1 (1) l(x l (I). . .==1 " al = " �(c(I» . . . (Xi 0 e)'(I) L 1=1 axl af = � ax' (C(I» L. . as we shall point out once again in Chapter 7.1 12 Chapter 4 some sense the one always taken by classical geometers.x"..) or = L: D. . c: � + j lffi. " ( ) Since every tangent vector at of the form dc/dl. ...) de al dl (.=1 (f 0 C)' (I) ax' dl (The classical notation.1I . where M be a curve. .. dl L. x").x" 0 C(I» .=1 Consequently. It is the closest approach in classical analysis to the realiza tion of as a function on tangent vectors.(This equation signifies that true results are obtained by dividing by dt again. . and x a coordinate system (so that I oX for some function on namely Let MODERN FORMULATION Let X i be functions of t. dx i (e(I» dl dC · dx i dl dl gives 1=1 af dl = � ax' dx" L. we have C(I) is dl af dl = � axl dx" L .=1 d(f(c(I» ) dl Multiplying by . say X i X i (I). CLASSICAL FORMULATION = Let I be a function on M. . (Xi 0 e)' (I) .
Despite these diff erences. we have a map J p : Mp + Nf(p) . Classical terminology used these same words. and there is no reason why J p. After a while. the same q E N may be /(Pt) for more than one Pi E M. we will continually examine the classical way of expressing all concepts which we introduce. . the CCtranslation" of classical terminology becomes only a little more difficult than the translation of the German in which it was written. in f act.e. and it just happens to have reversed this: a vector field is called a contravariant vector field. it gives rise to a map Strict notational propriety would dictate that this map be denoted by but everyone denotes it simply by (/*p)* .) This section � is * denoted. Suppose w is a section of T as follows: �(p) = w(/(p)) 0 J*p. we can * * do something with the cotangent bundle that we could not do with the tangent *N. and then operate on it by w. by J *w.. naturally enough. then there is a map J : TM + TN. i. * for each P E M. should equal J P2 ' On the other hand. and which contravariantit's just the opposite of what it logically ought to be. (The complex symbolism tends to hide the simple idea: to operate on a vector.Tensors 113 In preparation for our reading of Gauss and Riemann. Then we can define a section � of T *M bundle. Recall that if J : M + N is Coo. Nowadays such situations are always distinguished by calling the things which go in the same direction cccovariant" and the things which go in the opposite direction "con travariant". we can say. Since J p is a linear * * transformation between two vector spaces. that a map J: M + N pro duces a map J going in the same direction on the tangent bundle and a map * J* going in the opposite direction on the cotangent bundle. And no one has had the gall or authority to reverse terminology so sanctified by years of usage. we push it over to N by J . Notice that we cannot put all J* together to obtain a bundle map from T *N p to T *M. roughly. There is no corresponding way of trans ferring a vector field X on M over to a vector field on N. So it's very easy to remember which kind of vector field is covariant. while a section of T *M is called a covariant vector field.
b) If is another such coordinate system.=1 n " W = L WIi dxII . Notice that if we also have . so the x coordinate system is just an "oblique Cartesian coordinate system". x'j = L7=1 aijx" for certain au .1 14 Chapter 4 The rationale behind the classical terminology can be seen by considering coordinate systems on which are linear transformations. . In this case. then Clearly so x' aij = ax'j laxl . any combination xi. . " aax. Comparing (*) with D (x' = x' n I dxIj = L � dXI .=1 is also called "covariant". (ax'j laxl) is the constant I h is can be seen directly from the fact that the matrix matrix 0 X. . x(a l vl + . + anvn) = (a l . n W = L Wi dxi . . . . COllsequently. . bv.) = el. if then x lffi.n x(v. an). we see that the diff the coordinates hence they are cccovariant". _  1' P I x(p) = (a.I ) o XI .j 1=1 dxi erentials "change in the same way" as from Theorem I.
l lj = '"" i aaxlj . Substituting 115 into the first expression for w and comparing coefficients with the second.a X for a vector field. . . . respectively. Vrn.=1 L.) Covariant and contravariant vector fields. . . Vk_l . a function is multilinear if is linear for each choice of is clearly a vector space. the functions a'j must satisf y a n . . . . If V t+ T(VI . V. XII used to be denoted by XI . . Vk _l . Af ter this was firmly established. . .. sections of T *M and TM. We begin with some worse algebra.. Vm = V. Vk+ h " " Vrn) VI . . If Vb " " Vm are vector spaces. Xn . .e. . .. Vk+I . . . this vector space will be denoted . 1=1 These expressions can always be remembered by noting that indices which are summed Over always appear once "above" and once "below" . Wi ax" j L ax'j 1=1 La On the other hand. (Coordinate func tions X l . are also called covariant and contravariant tensors (or tensor fields) of order I . This suggested subscripts Wi for covariant vector fields and superscripts at for contravariant vector fields. . . which is a warning that worse things are to COme. . . .Tensors then we can express the W'. . . we find that n I W = . the indices on the x's were shifted upstairs again to make the summation convention work out. = � 'I axIl . i. given two expressions � a1 L 1=1 a a ax.. The set of all such T VI . i n terms o f the Wi .
v *n is the dual basis for V* = r'(V). T ® S is not S ® T. S. (S ® T) ® U = S ® (T ® U). · · . so we can define nfold tensor products unambiguously. . ® T. and S E r i (V) we can define the "tensor product" T ® S E rk+i (V) by T ® S(v" . these maps can 1C'' (U) + U X �n k . (S. . this tensor product operation is itself multilinear. . then the isomorphisms Ip : I: 1C' (U) 1C' (p) + + {pI U X �n X �n yield isomorphisms Ifwe choose an isomorphism r k ( �n) be put together to give a map ' I : + �nk once and for all. Vk . etc. . Vk ) . . . Vn is a basis for V and v *I . . ® T S. S(Vk+ " . Vk+i ). . . U c B and is a trivialization. O n the other hand. Vk +' . . defined completely analogously to the case m = I: are easily scen to be a basis for r k (V). . .) ® T = S. if VI . . then there is a linear transformation f* : rm(w) + rm (V). Notice that r ' ( V) = V*. . Vk +i ) = T(v" . .116 Chapter 4 by rm(v). In particular. If f: V + W is a linear transfor mation. . . + + and let If = peB U rk (1C' (p» . . We let E' O fcourse. . then the elements For T E r k (V). which thus has dimension nk We can use this new algebraic construction to obtain a new bundle from any vector bundle � = 1C : E + B.
. ./. Xk Notice that A is multilinear on the set V of Coo vector fields: A(XI . and a section is called a covariant tensor field of order k. . . .XI . . . .i/.: where dX il ® . t" . . . . . . which yields a function: .Xk ) + A(XI . . . Thus. . . The section A is continuous or Coo if and only if the functions A i . .. n are a basis for Tk (Mp). . .. on U every covariant tensor field A of order k is a basis for can be written dx l (p). Xi + X'I .. ® dxlk (p).. . and multilinearity of ®. . Xk ). il. A covariant tensor field A of order k can just be thought of as an operation A on k vector fields XI . aXi. then the kfold tensor products dXil (p) ® . . il. .. so that TM. ® dX ik now denotes a section of Tk (TM) .. .lk (P) dxi' (p) ® . . • • . . . . ik are. .Tensors 117 We make H ' : E ' + B into a vector bundle by requiring that all such / ' be local trivializations. ® dX lk (p) E Tk (Mp) I ::. . Xk) = A(XI . .. To derive this equation.··. we just use equation (**) on page 1 1 4.. . . il . . .. . .: A = L A II . For the case of the bundle is called the bundle of covariant tensors of order k.ik ::. .. . . . .k or simply Ifwe also have then (the products are just ordinary products of functions). ® dXlk .ik dX11 ® . . . . . A(Xh . . . dxn (p) A (p) = L AI) . X'" . Tk (�) �*k T (TM) (Mp) *. The bundle is the special case k = I .···. Xk ) Xk ) = aA(XI . U) is a coordinate system. If (x. . .
. . 1X" . A : V x · · · x V + F ". i. 2. if U) is a coordinate system and is linear over F. . Xk)(P) = A (p)(Xd p). \vhere each at now denotes a constant function and f is a Coo function with I(p) = I and support I C V. .=1 n a I on U outside V. . . . Xd p» = I(p) · A(X" . then we can define v = L al axl p ' . XI . . Xt (p). then there is a unique tensor field A with A = A.e. because A is defined "pointwise". . THEOREM. . . (p) for each we claim that i. Xk)(P)· = l (p)A(p)(Xdp). Vk E Mp are extended to vector fields X" . then there is a vector Geld X E V with X(p) = v. . . If P ROOF Note first that if v E Mp is any tangent vector. (p). Xd p» We are finally ready for another theorem. . .1 18 Chapter 4 rvforeover. . . . .' k limes (x. . . . In fact. if I is Coo. . . then for we have A(X" . . . Now if v" .. . . · · · . . . it is actually linear over the Coo I Ullclions F. . one that is used over and over. Xk E V we clearly must define The problem is to prove that this is welldefined: If XI (p) = Y. . f(p)X..
b' a x ' V) be a coordinate system around p. A since bl (p) (g � ) a 1 = (p) o. the cotangent bundle: a function fi·om V + 'F which is linear over 'F comes fiom a covariant vector field w .. A(Y) (p). The proof that A(X)(p) = A(Y)(p) when X(p) = Y(p) is in two steps. If g is 1 in a neighborhood V of p. then is a welldefined Coo vector field on all of A(X)(p) = M which equals X on V.Tensors 1 19 (I) Suppose first that X = Y in a neighborhood V of p . Just as with covariant vector fields) a Coo map f : M + N gives a map f* taking covariant tensor fields A of order k on N to covariant tensor fields f* A of order k on M: . it obviously suffices to show that A(X)(p) X = L. Because of Theorem 2.) For simplicity. we will never distinguish between the tensor field A and the operation A.. so f A(X) evaluating at p gives X(p) = o. •:. AUX) = AUY) A(X)(p) (2) To prove the result. and support 1=1 � a . = 0 if g C V. Now A(Y)(p) = = 0. can write = (The map A "lives at points". = fA(Y). � bl (p) . Let f be a Coo function with f(p) = 1 and support f C V. so that by (I). nor will we use the symbol A any longer. so that on V we = A(Y)(p). in particular. where T k (TM) = T*M. to use the in terminology. to the case k = 1.. Then fX = fY. Let ( x . where all b' (p) = o. Note that Theorem 2 applies. take the case k = 1 (the general case is exactly analogous).
Recall that a contravariant vector field is a section X of TM. we just define V(A) to be A(V). . We could also use the notation '1k(TM). if only for the sake of completeness we should define contravariant tensor fields.. a A = L A. h. . . aX}k ax}' h..P" " Pk = L jl. if and are covariant tensor fields of orders k and I. In local coordinates we can write k + 1 times).: Ah . ._ ' aX'Jk aX'}1.jk (remember that each alax j Ip operates on Mp*)... each on Mp* . · ·}. Now an element v ofa vector space V can be thought of as a linear function v : V* + �. A A(p) a. and to identify contravariant tensor fields of order k with operators on k covariant vector fields that are linear over the Coo f unctions F . there is an analogue of Theorem 2. A contravariant tensor field of order k is just is a klinear function a section of the bundle T k (T*M). . j(. or simply a A = Lh. " <8> _a_.h� <8> . then we can define a new covariant tensor field of order k + I: A B (A <8> B)(p) = A(p) <8> B(p) Mp x . . ·· h __ <8> . that allows us to dispense with the notation A.. Wk into a function: A Naturally. " <8> . .k (p) . a A(p) = L A}.. thus. then we easily compute that A. <8> __ .IP <8> . if we use TdV) to denote all klinear functions on V*. x Mp Although covariant tensor fields will be our main concern. .jk A contravariant tensor field A of order k can be considered as an operator taking k covariant vector fields WI . j(.120 Chapter 4 A <8> B (operating on Moreover.IP aX}k ax}'. respectively. . . So each Xp E Mp.···.··.···. proved exactly the same way.j. . If we have another such expression.
' (�). obtained by replacing each fibre 1C . In particular. given by Generally speaking.' ( V) is linear and one one. The inverse. Since both End(V) and 7. however. . .Tensors 121 Finally. If v" . In particular. this map is an isomorphism.A) = A(V). for 5 = implies that A (S(V» = for all A.j .' ( V). sections of 7. Given S. our isomorphism can be used to transfer any operation from End( V) to 7/ (V). which implies that S(v) = for all v.' (1C . There are all sorts of algebraic tricks one can play with 7. the correspondence S >+ 5 from End( V) to 7. let 7. V is a basis of V and n T = L T/v·i <8> Vj . although they should be kept to a minimum. given a bilinear e(v. ' (V). To make the intro duction less painful.' (TM) are called tensor fields. Notice that each S E End( V) gives rise to a bilinear 5 E 7. i. 5: by the formula Vx v· + �.' (p» . covariant of order I and contravariant of order I . 5(V. for each v the vector S(v) E V is merely determined by describing the action ofa A on it according to (*).' (p) by 7. Moreover. Let End( V) denote the vector space of all linear transformations T : V + V ("endomor phisms" of V).A) = A (S(V» . certain ones are quite important. It is not hard to check that this isomorphism of End(V) and 7. . T: V x V · + �. If V is a vector space. is not so easy to describe.' ( V) denote all bilinear functions T: V x V· + R A vector bundle � = 1C : E + B gives rise to a vector bundle 7.' ( V) makes the identity map I : V + V in End( V) correspond to the CCevaluation" map e : V x V· + � in 7/ ( V) 0 0 0. we are ready to introduce "mixed" tensor fields.' ( V) have dimension n'. this number is called the contraction of T. we can take the trace of the corresponding S : V + V. we consider a special case first.
122 then we can find the matrix A = in terms of the T/. and we obtain a function "contraction of A". = Chapter 4 (aij) of S. aj/ v*j (S(v/» T(v/. J. given a tensor field A. in any fibre bundle 7i (�). In particular. i=1 = II � trace A (p) if we consider A (p) E End(rrl (p» . I L a axj ' II (contraction of A) = The general notion of a mixed tensor field is a straightforward generalization. v*j) T/ . obtained by replacing each fibre rrI (p) by End(rrI (p» . a section A of Ti (�) can just as well be considered as a section of the bundle Elld(�). Define to be the set of all (k + I)linear L A. covariant of order and contravariant of order 1 . each A (p) is an endomorphism of rr1 (p). <8> i i. each section A gives rise to af unction (contraction of A) : B defined by p r+ contraction of A (p) = + 2 T L T/. fibre by fibre.1 (TM). j=1 = II Thus contraction of = = (The term CCcontraction" comes from the fact that the number of indices is con tracted from to 0 by setting the upper and lower indices equal and summing. 1=1 1/ ( V) T: V 'v' k times <8> • • • <8> V x V* '. which is a section of we can consider each A (p) as an endomorphism of Mp. then 7. If in a coordinate system .' i limes <8> • • • <8> V* + R .) These identifications and operations can be carried out. in fact. Thus. . defined by S(vt) L aji Vj . In this case. A= A J dX '. More over.
. and if A' = L il...j � ax'p .. aX.a axj et L this is certainly true.'. i.j ax. Classical differential geometry books are filled with monstrosities like this equation.Tensors 123 Every bundle gives rise to a bundle Sections of are called tensor fJelds. ..···. ..e. nk+l functions in terms of the coordinate system and then checking that (*) holds. Here is an important example. or simply of type an abbreviation that also saves everybody embarrassment about the use of the words "covariant" and "contravariant".'\'o. J. Consequently. a tensor field of type can be expressed as � J. in classical diff erential geometry. it is asserted that if one chooses the same functions for each coordinate system.k (TM) (�). n' of of oP . Locally. Ai.. then (*) holds.·. h." In other words. all important tensors are actually defined by defining the nHI "HI x x'. ax'P.: � .k (�) ... ·. A (�) A= L il k ..al ' " ax'ctkJ.·.lh/. covariant of order k and contravariant of order I. for .ijr .. x. the classical definition of a tensor field is: an assignment of functions to every coordinate system so that (*) holds between the functions assigned to any two coordinate systems and (!) Or even.: jl. .i��.. i. ·. In every classical diff erential geometry book.ijr • .. cca set of functions which changes according to (*)".: then axil axl ax'PI A'g::: g� = L Af(. axh i( /. one will find the following assertion: "The Kronecker delta is a tensor. jl. In fact.
E Mp' with the expressions J. Mp x Mp' we consider = v E Mp and J. in a similar manner. The contraction of a tensor is defined.124 Chapter 4 a " A Le 0f.j j = L... a I " n Thus isjuSI the evaluation map Mp x Mp' + �. i=1 = J. independent of the choice of the coordinate system x To identify the mysterious map A(p). classically.. P=I = + �.) i. i.0..J is a certain tensor field.(v). sat isf ying A(p) AI. _ L.. dxl <8> axj = From our point of view.J.j j b i..J. then A (p)(v.) LO! dx'(p) <8> a!j IP (v..) a collection of functions one for each coordinate system. considered as an endo morphism of Mp.e. it isjust the identity map. Given a tensor. .. what this equation shows is that '. aib. L bp dXP (p).
then 2 E T. The revolution in the modern approach is tllat the set of all vectors is made into a bundle. is independent of the basis with respect to which the matrix is written. Incidentally. .A) .=1 n so that this sum is a welldefined function. . a n to each coordinate system numbers a l ) . . Ifwe consider each A(p) E T. A2) = contraction of: . (Mp) to be B(p)(v) .ja ) L L L: ax">: ax a=1 i. a tensor of type can be contracted with respect to any pair of upper and lower indices. For example... defined as the sum of the diagonal entries of its matrix. A) >+ A (p)(v. . This calculation tends to obscure the one part which is really necessaryverification of the fact that the trace of a linear transformation. V2. . A) . A2.3 (Mp ). a vector at a single point p is classically just an such that the assignment of n TZumbers a l .} n ax = L A !.Tensors we note that 125 a=1 � A'aa = � (" Aj' � aX. V2.yv = L AaPvya P " J1. a 'n assigned to satisfy x. . L. L ax'a. VI . so that vector fields can be defined as sections.j a=1 ax = L A{a) i. . a=1 "transform correctly" if the A�r do. .j = L AL . n . x' ' aJ (v. A). . the functions (�) we are laking B(p) n BJ1.. rather than as equivalence classes of sets offunctions. ax/ja . While a COlltravarianl vector field is classically a set of n functions which "transforms in a certain way". and that all other alp..(p).ax i . ax'j = ") a'.= This is precisel the definition we adopted when we defined tangent vectors as y equivalence classes [x.
The modern revolt against the classical point of view has been so complete in certain quarters that some mathematicians will give a three page proof that avoids coordinates in preference to a three line proof that uses them. We won't go quite that far. and hence must come from some A . instead we define a function A on vector fields. Unlike the "Kronecker delta)' and contractions. At the appropriate time we will discuss whether or not this is all a big cheat. For a long time. the return to the old definition was influenced by the "functorial" point of view of Theorems 31 and 34. which miraculously turns out to be linear over the Coo functions F.126 Chapter 4 types of tensors are constructed from this bundle. invariant definitions of all the important tensors in diff erential geometry are made by means of Theorem 2. or as equivalence classes of curves. We seldom define A (p) directly. As we shall see. the party line held thaI TM must be defined either as derivations. such invariant definitions are usually not so easy to come by. . but we will give an "invariant" definition (one that does not use a coordinate system) of any tensors that are defined. The tangent bundle itselfwas almost a victim of the excesses of revolutionary zeal.
(p) · ifJ If(p) ' L. . (c) Show that U*dyj)(p) = t a(y. respectively.(v) = df(v)f(p) E �f(p)· v*). . Let + dUg) = fdg+g dj. For v E Mp. show that f. express j�(L7=1 a i alax i l p ) in terms of the alay j lp .g: M f: M dx" f* ( . L. If f.) (d) Express in terms of the 2.. . show that 4 .x� f) (p) . Vn and WI . U) and (y. v* N are Coo. .j" dyh <8> <8>dyj. (b) Show that a bundle � is orientable if and only if �. . (a) If g: N �. 1=1 0 f = identity.Tensors PROBLEMS 127 f: Nm. Y j=1 and. (a) Show that i fthe ordered bases VI . V) are coordinate p and f(p). Jn • • • + 3. . . dxl (p). . . w*) . is orientable. � be Coo. then the same is true of the bases n and n for V*. ) w* . Let Mn systems around + + (Proposition 23 is the special case (b) Show that a a(y f) a f* ( axi Ip) = � ajXI.. more generally.···. . • • Wn for V are equally oriented. and suppose that (x. ah . .) } 1 . then 1 . . .
In each case. in particular.1 independent vector fields i Aali for Ci = . . . (vi) I f a j AiA j i s a n invariant for A i a n arbitrary vector. or b = c and ars is skewsymmetric. If aij and bkl are components of a tensor field. so are aiJ bki .} fJ. and by definition 5. the other sets are com ponents of a vector field. where i for i = I. then (alj . (iv) If aij = a ji for the components of a tensor field in one coordinate system. . n . . and then translate the problem and solution into modern terms.(Jj )�i� j = 0 is satisfied by every vector field �i. i (vii) If a ij AiJ. then aij + Qj = O. n the vector.�fJ. (v) If aij and bij are components of a tensor field. and consequently i 1 a ij + aj = (fJj!Jj + fJj (Ji ). where fJi is a given covariant vector.= 0 is expressible linearly in terms of II . . so are aij + b lj .1 which satisfy the equation. .128 Chapter 4 (i) If the quantity ).. show that if bars + casr = 0. using the classical methods. . An "invariant" is just a (welldefined) function. if Qij)J). i (iii) If fJ. after (xiii). . that is. Hints and answers are given at the end. i .. . if Vi is defined [cJ (iii)] by a ij Aali v j = 0.) = 0. Ci = 1. det(AaI') oJ 0.1 and fJi Vi oJ 0. Show that the rank is invariant under all transformations of coordinates. then any vectorfield Ai is expressible in the f orm where the a's are invariants. (ii) If Aali are the components of n vector fields [in an nmanifold] .j = 0 for all vectors Ai such that Ai fJi = 0. . then a ij + aj i are the i components of a tensor. . Il indicates the component and ex for ex = I. . 1. (ix) By definition the rallk of a tensor of the second order aij is the rank of the matrix (aU). . n . The following statements and problems are all taken from Eisenhart's clas sical work Riemallnian Geomet�).ij = a'ji for the coordinates in any other coordinate system. so Ai jLi means L7= 1 Ai JLi .: (viii) If ars are the components of a tensor and b and c are invariants. . check them. . then a. then either b = c and ars is symmetric. any vectorfield A satisfy i ing Ai fJ. . and these vectors are independent.is an invariant and either Ai or fJi are the components i of an arbitrary [covariant or contravariant] vector field.are the components of a given vectorfield. Remember that the summation convention is always used.
. . = ° 1 I 1 if ja = jp for some a oJ fJ Or ia = ip for SOme a oJ fJ or if U" . w) = T(w.) . The rank of T may be defined as the rank of T (consider the matrix of t with respect to bases vI . jp is a n even permutation o f i) . . (xii) Let V = p' . ip if j) . . v) + T (w.:·.i�p are the components of a tensor in all coordinate systems. can i be written in the form (a j . then (xiii) If j 8 h . (Begin by choosing Yo complementary to ker fJ. Vn and v* ) .l. ip} i f j) . If O"(Z) = T( Y. . jp is an odd permutation of iI. v) = for all v implies that T + T' = 0. .and writing v uniquely as vo+cYo for Vo E ker fJ. . .. . (vii) Given w [with w(p) oJ for all pj . Then T + T' is determined by S(v) = T(v. w) + T (w. and vice versa. . v* n) . v) + T(v. . (xi) Show that the tensor equation aij Ai = aAj. . T(v + w. . if the equation is to hold for an arbitrary vector Ai . For example.. . Similarly. . (This is true only locally. Show also that a ij = 8 ija. v) . .such that HINTS AND ANSWERS. v + w) = T(v. T(v. w)] . let T' (v. . . Z). 'I ···'p then Of. . . . on S' x � there is an w such that ker w (p.) (vi) For T: V x V + �. . . . is one. . ip (i) w is determined if w(X) is known for all X. then T(Z. Z) = w(Z)O"(Z)!w(y) for all vector fields Z. there is a y complementary to ker fJ. . where a is an invariant. For. which span ker w at each point. . jp) oJ {i" . (iii) Given w [with w (p) oJ for all pj . ° ° ° M T(v) = av + fJ(v)y for all v . (ix) T: V x V + � corresponds to T: V + V' [where T(v)(w) = T(v.E V * and T(v) = av for all v E kerfJ. Xn _. there are everywhere linearly indepen dent vector fields X" .is a given vector. . .a8lj)Ai = 0. . 1f T : V + V and fJ. I) consists of vectors tangent to S' x {I}. . where fJ. where ai and bj are the components of two vectors. v) . . choose Y complementary to kerw at all points. i (xii) I f a ij Ai = aAj holds for all vectors A i such that p iAi = 0 . w). . . show that for the symmetric tensor aibj + ajbi the rank is two.Tensors 129 (x) Show that the rank of the tensor of components aibj.
the following diagram commutes: V � V** f 1) Show that there do nOI exist isomorphisms i v : V lowing diagram always commutes. . (b) The "double dual functor". . . k F U) ( T)(AI> . Ap) = det(Ai (Vj» . !" . FU) = f'* is a functor.130 Chapter 4 0 : V x . F(V) = V. F U) = f * is a contravariant functor. + W is an isomorphism. except that FU) : F(W) + F(V) and F(g 0 f) = FU) 0 F(g). FU) = f is a functor. . . F(V) = 1k ( V) = Tk ( V*). . F(V) = T k ( V). may also be defined . . Vp . f l fr + �* which makes the (a) The "identity functor". 6. F(V) = V*. Show that for any linear transformation f : V + W. . . x V x V* p times x (xiii) Define p times . . V � V* 1** . · · . covariant in some and contravariant in others. . W � W** l 1 + V* such that the fol 7. . W � W* Hinl: There does not even exist an isomorphism i : � diagram commute for all linear f : � + �. such that F( I v) = I F(V) and F(g 0 f) = F(g) 0 F(v). A covarianlfunctor from (finite dimensional) vector spaces to vector spaces is a function F which assigns to every vector space V a vector space F( V) and to ev ery linear transformation f : V + W a linear transformation F(f) : F( V) + F(W). Ak 0 f) is a functor. (d) If F is any functor and f : V isomorphism. ' Ak ) = T(AI 0 f. . (c) The "T functor". . FU) = f * is a contravariant functor. F(V) = V*'. then FU) is an A conl>ava/ian/f ullclor is defined similarly. (a) Let iv : V + V** be the "natural isomorphism" iV (V)(A) = A(V). . (f) The "T k functor". . (e) The "dual functor". x V* + � by O(VI . Functors of morc than one argument. .
. . �) + GL(k . . 9. . Show that a different choice of bases leads to a homeomorphic metric on Hom(V. (a) Let F be a functor from Y". an isomorphism F(�n ) + �k Then F(A) can be considered as a map h(A) : �k + �k Show that Iz : GL(n. and show how to construct a bundle F(�). vn ) and w = (w" . once and for all. (The bundle T*M is a special case of the construction in (d). . . as well as an example of a functor which is not continuous.Tensors 131 corresponds a trivialization 8. Show that if � = 1C : E + B is any vector bundle.) Generally. �) we can consider it as a map A : �n + �n. . Choos ing a basis for V and W. show that the following diagram commutes . W ) denote all linear transformations f rom V t o W. W) to Hom(F(V). then there is a bundle F(�) = 1C' : E' + B for which 1C'1 (p) = F(1C.) (d) DefIne a continuous contravariant functor F.1 (p» . wn ) be ordered bases of V and let e = (e l . e ) be the standard basis of �n . (a) Let Hom(V. See the next two problems for other examples. and such that to every trivialization I : 1C1(U) + U X �n (c) The functor 7J (V) = 7 1 ( V*) = V** is continuous. F(W» . the same construction can be used when F is a functor of several arguments. (b) A functor F gives a map from Hom (V. . Call F continuous if this map is always continuous [using the metric in part (a)] . �) is a homomorphism. Then F(A) : F(�n) + F(�"). W). (The bundle Ti (TM) is just a case of the construction in (b). and F is continuous. . (e) The functor F(V) = V* is continuous. and consequently give it the metric of �nm . the class of ndimensional vector spaces. The bundles 'Jjk (M) are all special cases. we can identify Hom( V. . W) with the m x n matrices. to yk Given A E GL(n. Choose. . If e + v denotes the isomorphism n taking ei to Vi. (b) How does the homomorphism h depend on the initial choice of the isomor phism F(�") + �k ? (c) Let v = (V I .
q) by [v. q) if q = h(A)q' where W. and define yn F. j =1 . aq] � (w .) = L'j�. (f)[v .q . choose ordered bases V . This suggests a way of proving the following. �) = � be h (A) = Ci (det A). q].wj. If fl : GL(n. q) for a given v. h (A)(q)]. : yn yk (d) For q. that Fh is a functor... and f: V + W.�) + GL(k . Show that � is an equivalence relation.vj .j) is defined by Wi = L aj. and that F. Show that this is a welldefined linear transformation. Then Fh : + is a noncontinuous functor. define ( V . �) is any homomorphism. q2] = [v. q') are welldefined operations making the set of all equivalence classes into a kdimensional vector space Fh ( V). q] = [w . THEOREM. q] r+ q. W E f(v. (e) Show that the operations [v. yn y ' . F(V) n After identifying F(�n) with �k . + such that the homomorphism defined there is a functor Fh in part (a) is equal to h. and that everyequivalence class contains exactly one element (v. �) + GL( I . (g) Let Ci: � + � be a noncontinuous homomorphism (compare page 380). aj.Vj . + q2] a · [v. (A) = h(A) when we identify Fh (�n) with �k by [e. this means that �k h(A) �k [ // F(e + v) F(e + w) also commutes.q' E �k . W. and let h : GL(n. qd + [V . define A by (f) If V.132 where A = Chapter 4 (a. = L'j�. We will denote the equivalence class of (v.q] = [v. aj.
so every section on can be expressed as a . . Let Fh be the functor given by the Theorem. .. h takes A into multiplication by 1 det A I. I G:'� ) I . show that a nonzero element of F. in addition to mixed tensor fields. For example. The Theorem in Problem 9 allows us to con struct a bundle whose sections correspond to these classical entities Oater we will have a more illuminating way): (x.Tensors 133 10. instead. These new rules are of the form A' = A operated on by h ( axa ) ax'P . (b) If. Clx for a unique function Q . assignments of a single function a to each coordinate system such that the function a ' assigned to satisfies x' a ' = det ( ax' ) · a . Conclude that the bundle of odd scalar densities is not trivial if M is not orientable. In classical tensor analysis there are. V) is nonzero. other "quantities" which are defined as sets of functions which transform according to yet other rules.. If is another coordinate system and a · Clx = a t . and consider the I dimensional bundle Fh (TM) obtained by replacing each fibre Mp with F. assignments for which (a) Let h : GL(n. �) take A into multiplication by det A. (Xx '. x axl} are called (even) scalar densities. If is a coordinate system. · a.(Mp). show that the corre sponding equation is a ' = det (c) For this h. then are called odd scalar densities.(V) determines an orientation for V. �) + GL( I . show that x' V a ' = det ( ax' )  ax'. a.
. 1 . .k(A) w an integer. . . . . AI 0 f.. . ) I).k(A) of GL(IZk+I . For k = I = 0 we obtain the bundle of (even) relative scalars of weight w [the (even) scalar densities are the (even) relative scalars of weight 1]. . . �). n .ill are the com ponents in every coordinate system of a certain contravariant relative tensor of weight 1 . . . .1 with these compo nents in every coordinate system. . . is an even permutation of I . (See Problem 712 for a geometric interpretation of these relative tensors. . . . (e) Define if if Show that there i s a covariant relative tensor o f weight . we can consider it as a map A : �n + �n. Al 0 f). . . f(Vk ). · · · . Show that the transformation law for the components of sections of these bundles is = (det A)wr.i" = 8.. .134 Chapter 4 k+1 by taking (d) We can identify r/ (ll�n ) with �n Recall that if f : V + V. �). . ) " . . . . If (det A)W is replaced by I det A l w (w any real number). AI ) = T(/(vtl.j ) replaced by I det(ax i lax. �) be defined by h(A) The bundle F(TM) is called the bundle of (even) relative tensors of type and weight w. .k(�n) determines an element r. . . in is an odd permutation of = ip i). we define 'Ilk (f) : 'Ilk (V) + 'Ilk (V) by I .k(�n) + r. 'Ilk (f)(T)(V Given A E GL(IZ.i n for some Ci oJ {J .k (A ) : r. Then r. . �) + GL(n k +I . Also show that gi l . Let iz : GL(II. (�) (�) j (or the same formula with det(ax i lax. . n I. . we obtain the bundle of odd relative tensors oftype and weight w. Vk .) if ia iJ .
. we have a vector field x. . Cl* X does not make sense for Coo functions a : M + N.e. . There is a function f : x(U) + � n with i. (�IJ = �. e) P. if ex is a diffeomorphism. (x. In particular. then we define It is not hard to check (Problem I) that a. Consider the curve The condition c = x op. in general. .X is Coo on a(M). 135 . It = X(p(I» . We would like to know if there is a curve p : through p whose tangent vectors coincide with X. U) around p and transfer the vector field X to x(U) C �n . X)q has "components" P (q). and its Let X be a vector field defined in a neigh + M borhood of P E M. vector fields. means that P. fn (q).CHAPTER 5 VECTOR FIELDS AND DIFFERENTIAL EQ UATIONS e return to a more detailed study of the tangent bundle TM.X on x (U) C �n . that is. Since this a local question.. we wish to introduce a coordinate system (x.. (� IJ dp = X(p(I» dl = X(p(I» . Recall that. p(O) = P (e. i. a curve p with Wsections.e. However.
. (� IJ = x.136 hence Chapter 5 �It = x. . etc. solutions were consequently called "integrals" of the equation.e. . If we use c'(I) to denote the ordinary derivative of the �nvalued function then this equation finally becomes simply This is a simple example of a diff erential equation for a function c : � + �n. e V C'(I) = /(c(l» is called an integral curve for c(0) = x XEU / with initial condition c(O) = x. of course. . n . . Ch(l) = f ( C' (I). X)c(t) . will denote points of �n. If c �n is open and / : + �n. (X(p(l))) c'(I) = /(c(l» . . which may also be considered as a system of II diff erential equations for the functions ci . For quite sonle time. then a curve c : ) + M with V (e..X) x(p(t)) (x. Part of this terminology is still preserved. we will work entirely in Euclidean space. . e) p(O) = dl p <!£ = X(p(l» is called an integral curve for X with initial condition p(O) = p. . a form to which our particular equations never reduce). cn (l») We also want the "initial conditions" i = l. Similar ter minology is applied. c. A Curve p: + M with (. y.P. Solving a diff erential equation used to be described as "integrating" the equation (the process is integration when the equation has the special form C'(I) = /(1) for /: � + �. and for a while x. . = = (x. to the diff erential equations one obtains upon introducing a coordinate system.
the correct solution is for ali i C(I) = 0 . if F' = I If.) To obtain the initial conditions c(O) = a. we must take X I c(1) = This works in all cases except 1 + l /a ' a = O. This can be checked directly if you don't believe the above manipulations. In this case.[c(l)f. which would be written classically in terms of a function . c'(I) = . so C(I) = I + C = x + C· I I hence. is the special case J(a) = _a 2 is to write The standard method of solving this equation dy dx = y2 y: R + R as �2 = dx _y f �2 = f dX _y I . we consider some special cases..Vecto. the equation in question asserts that y' = J a y. then (F a y)' = I F(y(x» = + C for some C.=x+C Y Y Thus the curves are supposed to be solutions.Fields and Diff erential Equations 137 Before stating the main theorem about the existence and uniqueness of such integral curves. (They really do make sense. The equation for a curve C with range R.
. In terms of vector fields.y l for all x.Jla the curve escapes to infinity because the vector field gets big too f ast. but never reaches. We say that the function I satisfies a Lipschitz condition on V if there is some K such that I/(x) . even or though X is defined on all of � . except c(/) = 0.. the curve C (I) = J /(I + J la) is defined for all large I. but this has nothing to do with the case. in fact. i s not differentiable. but it can also be obtained with a rather Jess stringent condition. 0. (2) C(I) = � 1 3 27 In this case. . written classically as I . can be defined f all I . there is no K with I/(x) .y E U. On the other hand. or (J) for all I. Unique ness will always be insured when I : V + �n is C I .138 Chapter 5 y).1(0) 1 :5 Kl x l for x near 0. For a > 0. the curves c are (which we missed by dividing by the integral curves of Notice that no integral curve. . This will continue to be true even if we modif the vector field near ° so that it is never 0. Notice that I(a) a2/3 is not Lipschitz. . It might be thought that this somehow reflects the fact that X (0) = 0. namely . and as 1 + 00 it approaches. There are two different solutions with the initial condition c(o) dy = y2/3 dx C(I) = ° f all I. since = = 0. given by I(a) = a2/3 . y Another phenomenon is illustrated by the equation c' (/ ) = C(I) 2/3 .IU) I :5 Klx . as 1 + . the function I.
'"v' n times Thus P(Xn . f then shows that . Let (M. . . A Lipschitz function is also clearly bounded on any bounded set. .e. and Jet f : M + M be a "contraction". . y E M. lhat is. suppose there is some C < I such that p(f(x). .rector Fields and Dijferential Equations 139 A Lipschitz function is clearly continuous. + Cn Since C < ] . THEOREM (THE CONTRACTION LEMMA). p) be a non empty complete metric space.l )p(XO. it satisfies a Lipschitz condition in a neighborhood of each pointthis follows from Lemma 25. j(y» Then there i s a unique x E "fixed point"). so there is some x with Continuity of 0 as n + 00. f(x) = Ixl). +C n+k I + Thus the sequence {xn } is Cauchy. i. + p(Xn+k I . that is. but not necessarily diff erentiable (for example. XI). Xn+l) + . The basic existence and uniqueness theorem for differential equations de pends on a simple lemma about complete metric spaces. I . . P ROOF NOlice that f is clearly continuous.. � (C n + . Xn+k l +k. . Xn+1 Then an easy induction argument shows that = fn (xo) = f a f a . . a C l function is locally Lipschitz.y) for all x. a f(xo). Xn+k ) � P(Xn . the sum L::'o e n converges. Let Xo E {xn } inductively by M such that f(x) = x (the function f has a unique M and define a sequence Xn+1 = f(xn ). On the other hand. so C n + . � Cp (x.
It I(U) dUI :5 K(b . then the new metric space is also complete. b) of step functions. g) = sup p (f(x) . xeX (J (f. clearly satisfies V V V. In particular. where 11 / 11 = sup I/(x) l · xeX Our basic strategy in solving diff erential equations will be to replace differen tiable functions and derivatives by continuous functions and integrals.x = <>(/) . we need only one simple estimate. bJ + �n satisfies 1 /1 :5 K. and thus f continuous functions. then <> is diff erentiable. we note that it is true for constant functions. hence continuous. which are unif or orm limits on [a. if <> satisfies (J). then we do not even need X to be compact. Moreover. If a continuous function I: [a. then l' <>'(u) du = l' 1(<>('1» duo To prove this. then a continuous function <> : (b. thus af = f 0 ex is continuous. then the set of all continuous functions I : X + M is complete with the metric (J(f. plus the fact that each ). g) = II I . b) + �n and I : defined on some interval around 0. if M is complete. so l' 1(<> (u» du. (I) <>'(/) = I(<>(t)) <>(0) = X <>(/) = x + if it satisfies the integral equation (2) where the integral ofan � n valued function is defined by integrating each com ponent function separately.a). Conversely. I (x) exists since M is complete. .gil. p) is a metric space and X is compact.1 40 Chapter 5 We are going to apply the Contraction Lemma to certain spaces offunctions. then the set of all continuous functions I : X + M is a metric space if we define the metric (J by If M is bounded. this is basically just the theorem that the uniform limit of continuous functions is continuous.<>(0) = For the proof of the basic theorem. g(x» . hence for step functions. If c + �n is continuous. if M is a m n oo compact subset of �n . <>(/) .!. Recall that if (M.
Moreover.SfJ l = s�p < bK sup h<u<b If I(a(u») . b) we have l' I(a(u» du d bL by (I) E2a (xo)}.xo l :s a. Choose b > 0 so that (I) (2) (3) b :s aIL (4) b < 1 1K. Y E E2a (xo). define a curve Sa o n . where c �n is open. H Thus S: M M. fJ E M.b).Vector Fields and Diff erential Equations 141 2 .(u) 1 la(u fJ . < :s a Since Ix .y l for x. of radius 2a and center xo. Let I : + �n b e any function. For each (b. b). Then + li Sa . Choose x E Ea (xo).xo l 2a. for all I E (b.fJ lI · . which will be fixed for the remainder ofthe proof + Then M i s a complete metric space. b) V such that a/(I) = I(ax(l» ax(O) = x.b) + (the integral exists since I is continuous on E2a (xo» continuous. Let Xo E and let a > 0 be a number such that the closed ball E2a(xo).I(Y)I :s Klx . a E M. for any I E (b. is contained in U. Then for each x PROOF. Now suppose a. Suppose that V V V 1 / 1 :s L on E2a (xo) I/(x) . THEOREM. Let E Ea (xo) there is a unique ax : (b. it follows that ISa(l) .b) by M = { continuous a : Sa(l) = x + (b. Sa is clearly The curve I Sa(l) .x l = < If I(a(u» u l by (3) .l(fJ(U» dU I by (2) = bKlIa . so Sa(l) E B2a (xo) c E2a (xo) for i E (b.
b) + This. (J(I) = X + provided that f f({J (u» du B2a(xo) [the open ball] (J(u) E B2a(xo) s o certainly if for all u with O ::s u I. is in V satisfying Reason: We claim that any such (J actually lies in B2a(XO). by (**). This clearly implies that (J (a + E B2a(XO) f or sufficiently small > 0. Consider first numbers t > O. this shows that Hence S has a unique fixed point: There is a unique S : M + M is a contraction. We can now use a simple least upper bound argument.b) + B2a(xo) with a(l) = x + 10' f(a(u» du. We clearly have (J (u) E B2a(xo) for O ::s u < a. Let A < b and (J (u) E B2a (xo) for O ::s u I } . So it must be that sup A = b. < Let a = sup A . A similar argument works for b < 1 ::s O. s s) .1 42 Chapter 5 Since we chose bK < I (by (4» . a : (b. Having used the elegant Con traction Lemma. •:. is not quite what the theorem states. < (J (u) E B2a (XO) = {I : O ::s 1 for all u with 0 ::s u ::s t . we pay for it by finishing off with a finicky detail: The map a is the unique {J(I) = X + f f({J(u» du. the unique fixed point ax of the map S is the unique curve with the desired properties. which contradicts the fact that a = sup A. in B2a(xo). To sum up. So (J(a) E B2a(xo). {J : (b. in fact. Suppose a < b. We have already seen (statement H) that for each 1 with 0 :s 1 < b. alas.
Thus the set is open. alai or dIdl in this . then the functions {J i satisfy the same differential equation. (Jt'(I) = /((J. (J'(I) = a'(lo + I) = /(a(lo + I)) = /({J(I)). b) x Bp(xo ) + V [i. It is clearly also closed and nonempty.x). Let x E V and let a" a2 be two maps or on some open interval I with a l (/). Ifwe define PROOF.. THEOREl'v!. x) a(O.x» x» . (lo + I).X) = /(a(l. Hence {JI (I) = (J 2 (1) for sufficiently small I.(O) = x i = 1 . 2. •:.(/) a.a2 (/) c V and Then a l = a2 on I.e.x) = x d di a(I. Suppose /: V + �n is locally Lipschitz. and have the same initial condition (J. a/(I) = /(a. This remark allows us to extend the uniqueness part of Theorem 2. but we will frequently use /(a(l. Suppose a l (10) = a2 (10) for some 10 E I. (/) . that is. around each point there is a ball on which / satisfies condition (2) of Theorem 2 f some K or (and hence also condition (I) f some L). so it equals I.vecfor Fields and Dijferential Equations Notice that solutions of the differential equation 143 a' (I) = /(a(I» remain solutions under additive changes of parameter. (Ji (I) = a. by Theorem J. (O) = a l (10) = a2 (10) E V. We will write ax (I) as a(l. Dl a(l. that is. if iJ(l) then = a (10 + I). 3. We now revert to the situation in Theorem 2. so that we have a map {I E I : ad/) = a2 (1)) a: satisf ying (b.
X) [ = Cix(/)]. THEOREM. if we fix I. the whole flow Ci is continuous (as a function of both I and xl: 4. This map Ci is called a local flow for f in (b.b) x Ba(xo). the map 1 >+ Ci(I. Then . to indicate explicitly the role of x. This map <p. To focus attention on this map. then the integral curve Cix with the initial condition Cix(O) = x differs from the integral curve Ciy with initial condition Ciy(O) = Y only by a change of parameter.X) gives the result of pushing each x along the integral curve through it.1 44 Chapter 5 discussion] . b) x Ba(xo) + V given by Theorem 2 is continuous. To picture this map Ci. In fact. then the map x >+ Ci (I. PROOF. On the other hand. is always continuous. the best we can do is to draw the images of the integral cUIVes Cix. Let us denote the map S defined in the proof of Theorem 2 by Sx. so the two images overlap. for a time inteIVal of I. For each fixed x. If y = Cix (/O).: <Pdx) = Ci (I . If f : V + �n is locally Lipschitz. we denote it by <p. then the flow Ci : (b.X) for b < 1 < b lies along part of the curve through x.
and differential calculus in Banach spaces). when you're fin ished you'll also know about Banach spaces. cxx.S. and a recently discovered proof can be f ound in Lang.S. then tJlej/ow a : (b .axll S lIax .I ax .Syax ll + I Syax . If additional conditions are placed upon the map /. we have .b) x Ba (xo) + V satisfies a (O. In fact. + (b K n. then Recall also that in Theorem I the fixed point any Ct .). y)l. . Real and Functional Ana!Jsis (3rd ed.. Since the map a : (b. Hence Cl)' = /� S.x) = x.: I 1 _ bK l x .Y I · .) Ix . A clean exposition of the clas sical proof is given in Lang's Introduction to Diff erentiable Manifolds (2nd ed.a for lIax . If f : V + �n is Ck . and then read about diff erential calculus in Banach spaces. besides. this certainly proves continuity of a .ay ll S 1 Since Ilax ayll = sup la(l. Notice that the maps <P. are consequently Coo if f is Coo.). In order to read this highpowered proof. you must first learn the elements of Banach spaces. pp. then further smoothness conditions can be proved for a.yl s l _ bK l x . including the inverse and implicit function theorems (Real and Functional Ana!Jsis. . so we obtain � Ilax .T!ector Fields and Differential Equations Recall that 145 li Sa . .SI'l I s bKli a .S. 371379.I'l ll · If S. 360365). x) a(l.axll 1 ) I S (I + b K + . + IIS. denotes the nfold iterate of Sy. including the HahnBanach theorem. pp.YI · I ay of Sy is the limit of S. We will just accept this fact. but this is probably easier than reading the classical proof (and. b) x Ba(xo) + V is also C k Unfortunately. .axll + . this is a very hard theorem. .
that for ls i < e each <Ps is a diffeomorphism. can be resaid.X» = Ci(S + I. in particular.Ci (S. We have also noted that satisfies (J(I) = Ci(S + I. so we can also define y(l) = Ci(I.(x) = Ci (I. Roughly speaking. and x E Ba/2 (xo). then <Ps(<P' (x)) = <ps+. If we now let <p. defined f or Is + I I < e.X).. then the point Ci(S. <Ps = <Ps <p. x) f x E Ba/2 (XO). X» III < e. we can or ls i. (x) E Ba/2 (xo). This shows. Ci(S. or if I sl.e) x Ba/2 (XO) + Ba(xo). say: (J'(I) = f({J(I» (J (O) = Ci (S.Ci(S. Consequently. : Ba/2 (xo) + �n b e <p. In other words. \I\�thout requiring any more proof. [If x E Ba/2 (XO). <p. (x). Is + I I < e and x. since it is local. l s + l l < e.I = <ps.X» f 1 1 1 < e. II I. This satisfies y' (I) = f(Y(I» y(O) = Ci(S. then Ci(I. (J (I) = Ci(I.] So if l s i < e. then the integral curve with initial condition x stays in Ba (xo) for III < e. Everything we if a a have said.X) E Ba(xo).X). with inverse <Ps .X). III. X).1 46 Chapter 5 x Continuity of Ci and compactness of {OJ e > 0 such that lia/2 (Xo) imply that there is some Ci : (e. . on a manifold. <P'+s = <p.
(3) If The examples given previously show that we cannot expect to be defined for all I. and M. (3). then Xq is the tangent vector at I 0 of the curve I >+ <PI <Pdq). . Let X be a Coo vector field on M.Vector Fields and Dijferentiai Equations 1 47 5. iterated k times] iterated k times] for k 2: 0 for k < O. ./2/2 ' <P. p) = . d E V. The support of a vector field X is just the closure of {p E M : Xp oJ OJ. then there are diffeomorphisms M + M for all I E � with properties (I).. Notice that by uniqueness. . PROOF. . and I I" I < ej2.e) x M + M is Coo.0 0 0. and let p E M. or on all of M. III. 11 + s l < e. then q <P q) <P <Pdp).:. such that there is a unique col lection of diffeomorphisms V + (V) c M for II I < e with the following properties: <PI: <PI (J) <p : (2) If (e.0 0 = k (ej2) + I" with k an integer. = is Coo. . " " V given by n Theorem 5 with con'esponding el . Then there is an open set V containing p and an e > 0. . So j we can define if E Vi = . write or <PI <PI 1 q <Pdq) {<Pi(q) q q <pi (q) (q) q if <PI+S <PI o <ps = 11 1 ../2 <P. if '" support X. e) x V Isl. Let e = min(e l . THEOREl\![. Let It is easy to check that this is the desired {<PI }' . (2). . To define f II I 2: e. ls l .. = <Pi for E Vi n V .. . en and diff eomorphisms <p. . defined by (I. <P <PI { <P. q E V. If X has compact support (in particular.. ls + 1 1 + < e. <PI: Clearly <p : (e./2 = 0. en ). 6. and and each is a diffeomorphism. THEOREl\![. In one case however. Cover support X by a finite number of open sets VI . if M is compact)./2 <Pr <Pr [[<P./2 <P. this can be attained.
The vector field X is sometimes called the "infinitesimal generator" of {<P. and which also has important theoretical uses. Then there is a coordinate system (x. is called a Iparameter group of diffeomorphisms. U) around p such that X = f (Xf) (q) = Xq / = hlim (<Ph (q) . we can assume that The idea of the proof is that in a neighborhood of 0 there is a X(O) = unique integral cUl"e through each point (0. l\1oreover. (q) PROOF It is easy to see that we can assume M = �n (with the standard coor dinate system t I " . if q lies on the integral a ax ' on U. say).1 48 Chapter 5 The unique collection {<P. . .f(q) . In the local case of Theorem 5. .} (vector fields used to be called "infinitesimal transformations"). we obtain a "local I parameter group oflocal diff eomorphisms". Let X be a Coo vector field on M with X(p) oJ O. . . Recall that I r+ Thus. an ). (/) dl = df (c (I» dt = (/ o c)'(I). or more precisely. _ o � 0 of the curve t r+ <P. M + �. The first use is to derive a corollary of Theorem 5 which allows us to simplif many calculations involving vector y fields. a'. alal ' lo ' .a'f+_ _ _ . This equation will be used very frequently. THEOREM. and is said to be generated by X.} given by Theorem 6. . to say that Xq is the tangent vector at I = amounts to saying that '!!:.(O. 7.n . from � to the group of all diffeomorphisms of M. Condition (3) in Theorem 5 can be rephrased in terms of the action of Xq on a Coo function f. ( ". the map <P. and p = 0 E lffi.
curve through this point. = Jim _ [/( x (a' + 0 �l = Jim . . .[/(<PI> ( x (a» ) . This is the desh'ed coordinate system. X. ./(x (a» ] (Xf)( x (a)). ./(p)] . . . an) = <Pal (0. an). a2. f it is easy to see that the equation or X.I coordinates of q and the time interval it takes the curve to get to q as the first coordinate. . .I may be used as a coordinate system in a neighborhood of O. an as the last n . . . this shows that X. . . h . The fact that X/ can be defined totally in terms of the diffeomorphisms <PI> suggests that an action of X on other objects can be _ I . . . . . . O» . ./(x (a))J h l + h . . To do this. . Jet X generate <PI and consider the map X defined on a neighborhood of 0 in �n by x (a' .rector Fields alld Differential 1!. . . ) a l o (f o X Moreover. a2./(0)] �l hO h Since X(O) = a/al ' l o by assumption. we will use a2./(x (a» ] I h /t70 = = /1+0 lim � [/(<Pa l +h (0. h. . 0) ./(O)] iI+O II = lim � [/(O. (a/at ' ) = X o X . .uations q 1 49 We compute that for a = (a' . .O = I is nonsingulaJ: Hence x = X. a2. . The second use of the equation = �� I o' (Xf) (p) = lim . [/(<PI> (p» . . (�l ) a . an» . . . . . . . . a (f) = a ' a (f o X) /1 I . which we have just proved. •:. .1+ 0 l is more comprehensive. an). . . an» . . for i > X. . is equivalent to X = a/ax' . a . ( � IJ ] we can at least compute ( f) = = lim � [J ( X (O. .
we first introduce the notation for Xf. integral curve / . If Y is another vector field. the Lie derivative of w with respect to X.. then A fairly easy direct argument (Problem 8) shows that this limit always exists..150 Chapter 5 Lx / obtained in a similar way. but we will Soon compute this vector field explicitly in a coordinate system. whose value at p is denoted variously by (Lx f)( p) = Lx / ( p) = ( X/) ( p) = Xp (f). and these facts will then be obvious. Y)p = <Ph' (Y¢_h (P» is obtained by eValuating Y at <P.' (p) = <P. I l�o h [(<Ph' W) (P) .. <P.h (p). Y defined at the beginning of the chapter. . I (LxY)(p) = }..(<Ph . we can define the Lie derivative of Y with respect to X. If a : M + N is a diff eomorphism and Y is a . . Recall that if Xp E Mp. Y)p]. oh The vector field <Ph . we define a new covariant vector field. and then moving it back to p by <p". Y appearing here is a special case of the vector field a.w (p)]. Thus (<Ph. it is another function. for a : M + N a diffeomorphism and Y a vector field on M.. by (Lxw) (p) = .� [Yp . and that the newly defined covariant vector field Lx w is Coo. To emphasize the fundamental similarity of these notions. (p) of X through p j p The definition of Lx Y can be made to look more closely analogous to Lx / and Lx w in the following way. Now if w is a COO covariant vector field. We call Lx / the (Lie) derivative of / with respect to X . This is the limit of certain members of Mp' .
+ W2) = Lx WI + Lx W2 .!(<Ph (p» <Ph* Y¢_I. if w(Y) denotes the function p r+ w(p)(Yp) and Lxw and Lx Y exist.(<Ph h h� h [ ( ¥h )p . then Lx (Y.Y)p] = (LxY) (p)."ctor Fields and Dijferential Equations vector field on the range 151 N.[.(A.Il [f(p)Yp . PROOF (I) and (2) are tri\�al. Finally. If (I) (2) then (5) Lx (w(Y» = (Lx w)(y) + w(Lx Y). We now wish to compute Lx w and Lx Y in a coordinate system. then (3) Lx!Y = X! · Y + ! . 2. (P) · + h�O !(P) . The remaining equations are all proved by the same trick.(pl] [ ] . lim (Lx !y)p = 1170 I [(/Y)p .<Ph* (/Y).'!. (pl] lim = 1170 !(P). '¥ = k�O k [Yp .(<Pk.p = h�O !. 8. We will do number (3) here.h h _ /t". (4) Lx ! · w = X! · w + ! · Lx w.k* y)p] Y ] r Yp . a * (Y) is just * Y)p r I r I ' *y A. If Lx Y and Lx w exist.(<Ph* !Y)p]  = lim = h70 . Nevertheless. we will stick to the original (equivalent) definition. LX Yi and Lx Wi exist for i = 1 . then a vector field a * Y on M can b e defined by Now notice that Of course. PROPOSITION.I ) * y. [Yp . + Y2) = LXYI + Lx h Lx (w.!.<Ph* Y¢_h (pl] h !(<P (P» A. Y r 'f'!t* ¢I.O l lim _ h I . k7o k . [Yp = lim . The cal culation is made a lot easier by first observing (a . [f(p)Yp . LxY. the one used in finding (/g)'(x).
dx' (p)] = lim Now the coefficient of dx j (p) is iJ"'O I � a(x' O </lh ) ' h L aXl· /t7 0 .We first compute Lx (dx'). I Xl I � a(x' O . a .(p) dx (p) . Recall (Prob lem 41 ) that if f : M + N and y is a coordinate system on N.dx' (p) [ .f(</Idp»  k Xf(p). [ _ 8' J (*) ] = lim = It.. n 8ai .x' </1.) . Then Lx (dx')(p) = l�o h [(</Ih*)dx' (p) .. . I a( �x�f) dxi We can apply this to </Ih * . = To justif *) we note that the map A (h . then f* (dy') = t j =1 . .</Ih) (p) h ax' [ _ ] {this step will be justified in a moment} . lim � a(x' O </Ih) h ax' .P X( = . Suppose X = L:7= .] a(x' 0 . aXl. . It now follows that ( a aa' x') . but . a'a/ax'.  .p /t".(p). Yp .152 The first limit is clearly approaches Chapter 5 f(p) · Lx Y (p). O · a I lim .j=1 . thus a2 A /ahax j = a2 A /ax j ah . which is what the interchange of limits amounts to. • + : We are now ready to compute Lx in terms of a coordinate system (x. In the second limit. Lx dx' = I: axj dx ' . where y is x. j =1 We could now use (2) and (4) of Proposition 8 to compute Lxw in general. V) on M.o h [( 0 aXl. the term in brackets = while an easy argument shows that </IMY¢_h(P ) k7 0 lim f(p) ..</10) (P) ax' (x' 0 </10)] . q) = X ' (</Ih (q» is Coo from � x M y to �. .
YX. . (4).a The second partial derivatives which arise here cancel those in the expression for Y(Xj). M + N is a Coo function. also denoted by [X.= Using (3) we obtain a . a n aa' a Lx . Clearly x = L a'.L b' . The trick needed to deal with this complication has already been used to prove (3). If j . then Yj is a function.' .. axj ) = Lxb'. and (5) of Proposition 8. = .. axj + b'Lx ( axj ) n . ax' ax' thus.) .=1 a b � (� a'.and Differential Equations 153 we are really interested in computing LxY. YJ..· axj "1 axj ax' .. and we can now use (5) to get the answer immediately.=1 axI n . Summing over j and then interchanging and obtain '�I i j in the second double sum we '�I LXY = L j=1 . To compute Lx ( alax' ) we could imitate the calculations of Lx dx'. and we find that LxY = XY . so XYj = X(Yf) makes sense.aax. abj a n .. = . aaj ) axj ' Laxl Lx This somewhat complicated expression immediately leads to a much simpler coordinatefi·ee expression f or Y. but there would be a complication.b'.· ax' axj axj ax' 0= . T!ector Fields h* on vector fields involves one more composition than <Ph* on covariant vector fields. because <P Lx oj = Lx [dX' ( a!j )] (Lx dx') ( a!j ) + dx' (Lx a!j ) so aa' a dx' (Lx. a a Lx (b'.j .L. . aa' a = L a'.= .
p) = 10 as I(SI. so that [X. This sort of thing irks some people to no end.154 Chapter 5 [X.Yp (Xf). p) ds r' a . PROOF.(sl) ds. A straightforward verification shows that [X. Fortunately. in this case the coordinatefree proof is short. p) = Ig(l. Y] (which is called the "bracket" of X and Y) is just defined as XY . p) al & (0. p) = g(O.YX. and can theref be considered as a member ore of Mp. If I : ( e. though hardly obvious. LEMMA. namely g(l) = l' /. This has an immediate generalization.e) x M + � is Coo and I(O. note that this means = Xp (Yf) . Y]p(/). Y]p(/) Often. Y]p(g) + g(p)[X. Y] have both been defined independently of any coordinate system. 9.e) + � with 1(0) = 0 can be written 1(1) = lg(l) for a Coo function g : (e. We are now in a very strange situation. [X. e) + � with g(O) = /. : . pl . •. 1(1. but they have Gcen proved equal using a coordinate system.p) then there is a Coo function g : (e. (0). Two vector fields Lx Y and [X. e) x M + � with = 0 for all p E M. Y]p ( /g ) = l(p)[X. In Chapter 3 we proved a lemma which for the special case of � says that a Coo function I: (e. Define g(l. Y]p is a derivation at p.
X]. there is a f amily of Coo functions g. Lax. Yll + [Y. [Y.(<Ph* Y)P](f) = h70 � [(Yf)(p) . PROOF. + 1 0. Y. Ly D. Let f: M � be Coo. then LxY = [X. + bLx Y2.(YfH <Ph . *:. +bX. + bY2 ) = aLx Y. This equation is capable of two interpretations in terms of Lie derivatives: (a) Lx [Y.YI = Lx a Ly . = f + lg. (b) as ope. [Lx. Lx X = O.Yp(Xf). [X. Zll + [Z.l�o (YghH<Ph (p» = (LxYf) (p) . so I (p» ] Jim lim h /1 h70 . Y] = XY . Xll = o.YX reveals certain f acts about which are by no means obvious from the definition. Finally. Y = aLx. If X and By Lemma 9 Then (<Ph* Y)p(f) = <Ph* (Y¢_I. a straightforward calculation proves the 'Jacobi identity": [X.(P» (f) = Y¢_I.·ators on Coo functions.Ly a Lx (which might be written as . Lx Z].(YgO) (p) = Xp(Yf) . go = Xf. Lx Y The equality Lx Y = [X. X] = o. Consequently. Y + bLx. [Z.[."ctor Fields aud Differential Equations 155 Y are Coo vector fields. diately that L is also linear with respect to X: LxY = LyX. we have L[x. so sO [X. it follows imme Since we obviously have Lx (aY. (P) (f a <Ph) = Y¢_I. Z] = [Lx Y.[Yp . Y]. Let X generate <p" II I < e. on M such that f a <P. THEOREl\I[.(P)(f + hgh ). Y] = [Y. Clearly [X. Z] + [Y.
even if Xp = 0.X¢_. Y]p = Oin the formula [X. y(t) = <P.156 Chapter 5 Finally. it does not necessarily follow that [X. If X generates <Pt..* X¢_Ii (p) is the tangent vector. Thus </>. Recall that (<Ph. (p). and thus the tangent vector. for reasons that \vill become more and more apparent. Y]p(f) = Xp(Yf) Yp(Xf) the first term Xp(Yf) is zero. pops up in the definition of prac ticaJly all other tensors. YJ. or a simple calculation using the definition of [X. for Xf may have a nonzero derivative in the Yp direction even though (Xf) (p) = o. but on the vector fields X and Y. Y] + f(Xg)Y . ] is no t a tensorthat is.g(Yf)X. Y]. h(p). to the curve But this tangent vector is just Xp. not over the Coo functions F. [X. shows that [fX.. The bracket [X. at time I = 0. X)p = <Ph. Thus. In particular.X)p = Xp for all h .. Now X¢_. Y]p does not de pend only on Xp and Yp (which is not surprisingwhat can one do to two vectors in a vector space except take linear combinations of them?).(p) is just the tangent vector at p <Ph (P) time t = h to the curve t >+ to the curve <P. note that Lx Y is linear over constants only. gY] = fg[X. the bracket operation [ . Y].(p) . we will endeavor to become more at ease with the Lie derivative by taking time out to prove directly from the definition of Lx Y two f acts which are obvious from the defi nition of [X. but the second may not be. (I)  LxX = 0. at time I = 0. . it certainly suffices to show that «Pt. Bef ore proceeding to examine its geometric interpretation. Proposition 8.. In f act. although not a tensor.
If <PI 0 "'S '" "'s 0 <PI for all s . •. consider the curve c : (e. Y] we first prove two lemmas. !. LEMMA. Y] = 0 if 1 2.I (q ) (/ o a) = lim _ [(/ 0 a)(<p" (a .X)q (/) = [a.X.(/ 0 a) (a.l. Then a. If this by Corollary f all or Given q. X generates {a 0 <PI 0 a . or Conversely. iI70 1 = lim . LEMMA. To develop an interpretation of I I . suppose that [X. of course. Y [X. conversely. Y] = 0.I }. then clearly Lx Y = O. so that 1 3. then <p Y '" " is true f all t.0 h I . p E M.T4!ctor Fields and Differential Equations 157 Since Xp = 0. PROOF We have (a. Let X generate {<Pd and and only if <PI 0 "'S = "'s 0 <PI for all s. then LxY(p) = O. e) + Mp given by . the unique integral curve c with c (O) = p and dcldl = X(C(I» is simply C(I) = P (an integral curve starting at p can never get away.I (q)](f) = X.I (q))) . Let a : M + N be a diffeomorphism and X a vector field on M which generates {<PI}.I (q» ] . . + M. Then Yp = 0 and (2) If Xp and Yp are both 0. X = X ifand only if <PI oa = a o<PI Then PROOF. Y generate {"'d. . then a. so Lx Y(p) = O. If a : M for all I./(q)] : 1 2. I (q)) . COROLLARY. [X. an integral curve starting at some other point can never get to p). [ f(a 0 <Ph 0 a 11".
[c(/ For the derivative. . The remarkable f act is that the condition [X. . <p. .. by a linear change ofcoordinates. Y)¢_. (p) . of this map into the vector space Mp we have h_O + h) . THEOREl\t[. and [X. • . a short calculation immediately gives the result 0. so <p" Y = Y. . 1 2. [ a�' ' a�2 ] = a Y = ax2 • 14. <P. = 1jJ. that ex ]. (O) = <p {l�o X [(<Ph. everywhere linearly independent of X. that = I . k. l . Y] = ° is s'!!ficient. (<Ph. . I = l'!:'o h [<P. Xp] = ° for I � ex. (p)] " = <P. Y)¢. •. for the existence of the desired coordinate system.C(I)] . (p) . As in the proof of Theorem 7. V) around p such that a Xa = axa on V. If Y is another vector field. for a a We have already shown that if X(p) oJ 0. so there is no hope offinding a coordinate system satisf ying (*) unless [X. M = �n . . (P)] } using (*) with q = <Pd p) By Corollary = 0. then there is a coordinate system (x. and.158 Chapter 5 I h c' (/) = lim . Y] = 0. Xk are linearly independent Coo vector fields in a neighborhood of p. as well as necessary. Consequently c(/) = alJ s. • . ex :::::: PROOF. k. : c(O).Y¢.. If X" . c' (/).fJ � k.<p" Y¢_. . . . then there is a coordinate system x with X = a/ax'. we can assume that p = 0. then we might expect to find a coordinate system with H a X = ax " However. 1jJ.
. x . As in the proof of Theorem 7. . Ci = k + I . and less important result. an) . » . If X and Y are two vector fields in a neighborhood of p. . ax' •:. we ar appeal to Lemma 1 3. . . just as before we see that (� I ) � I 0 = 1 X. . in some sense. . more difficult to Pl"Ove. .( . al' Thus x = X I can be used as a coordinate system in a neighborhood of Moreover. . . . an) = 4>�. ( .k I. (4):. x ( a l . Xa = � . . .vector Fields alld Diff erential Equations If 159 X. an) = 4>� (4). . ak + . . . we can compute that . .0. . p = O..0. . which makes this assertion much more precise. . XI = Nothing said so f uses the hypothesis [X" Xp] = O. . . We thus see that the bracket [X. . . (0. . (0) = a . . . . it shows that f each between 1 and the map X can or also be written a ax l ' Ci k. . . o al' 0 � l Ci = . then for sufficiently small h we can (I) follow the integral curve of through p for time h. There is a more complicated. . Y] measures. To make use of it. X (2) starting from that point. the extent to which the integral curves of X and Y can be used to form the "coordinate lines" of a coordinate system. . an» . y X y X (4) then follow the integral curve of Y backwards for time h. . (0. . n. . . . » and our previous argument then shows that . ( 3) then follow the integral curve of backwards for time h. follow the integral curve of Y for time h. (4)�. ak +I . . . generales {4>f } ' define X by I x ( a l . .
. . . a2 (0. c'(O) = O. .I) = a2 (1. (a) a dl.(O.O. (I. . O) (4) (0. so that this "parallelogram" is always closed. then Moreover. PROPOSITION. . . h) = 1fJd<Ph (1fJd<Ph (p» » . 0) (h.0) (O. . Even when X and Y are Oinearly independent) vector fields with [X. . c is the constant curve p up to first order.1) a. h. If we define 1 5. Yj oJ 0. . . (1. .I). h) = <P. . Then the CUlve PROOF. O. the parallelogram is "closed up to first order". h .O. .1) = a.0).160 Chapter 5 If there happens to be a coordinate system x with x (p) = 0 and a x = ax " (J) (2) a Y = ax2 ' then these steps take us to points with coordinates (3) (h.1) (b) .d1fJh (<Ph (p» ) a. . which means that c' (O) = y'(O)] is the following.(I. . The meaning of this phrase [an extension of the terminology "c = y up to first order at 0".h) = 1fJd<Ph (P» a2 (I.0. Let c(h) be the point which step (4) ends up at.0. C (I) = a.O. that is.
0) = DJ (/ a a. 0) + [Ddf a (2)(0. .h) = Xf(adO.)(O. but clever.O)] a Consequently. 0)] = DJ (/ a. (d). 0) + DJ (/ a (2) (0. (c). a(/ o aJ ) at = Yf o aJ a(/ a (2) = Xf o a 2 t . repeated use of the chain rule gives using (b) using (a). using the assumption e' (0) = 0.) = Yf o a. Until we get to Lie groups it will not be clear how anyone ever thought of the next theorem. h (/ a c)' (0) = DJ (/ a a. 0) + D2(/ a (2) (0.h» . Thus.aa(/ a. •. which ends the chapter.Xf(p) + Yf(p) + Xf(p) = o e" (O) (/) = (/ a e)" (O). we can define a new vector e"(O) or d2e/dt 2 l o by A simple calculation shows. (A more general construction is presented in Prob lem 1 7. 0) + [Ddf a J)(O. t . but can easily be skipped. the bracket [X. that this operator e" (0) is a derivation. cal culation. (e).)(O. 0) + D2 (/ a a. e" ( 0) E Mp. It is followed by an addendum containing some additional important points about diff erential equations which are used later.)(O.aa (f) a(/. and a second addendum concerning linearly independent vector fields in dimension 2. : . Whenever we have a curve e: (e. Y]p is related to this "second order" derivation.e) + M with e(O) = p and e'(O) = 0 E Mp. and (f) give a (/ 0 (') ' (0) = Yf(p) . The proof. O) + D2 (/ a aJ) (O.)(O.(O.T4!ctor helds and Differential Equations and for any (c) (d) (e) while 161 Coo function f : M + �. is an horrendous.) It turns out that for the curve e defined previously.al l aa.
) = 2[DJ (Yf 0 CY.2(/ 0 CY.2) (0.2)(0.2XYf(p) by (e) by (b) and the chain rule by (d) by (a) and the chain rule by (c) and (f).2 D2 (Yf 0 CY.2[DJ (Yf 0 CY.2) (0..J) (O.2 (/ 0 CY. 2 (/ 0 CY.0) = 2XYf(p) .. since (/ 0 C) (I) we have H (/ 0 cJ" (0) = DJ.2(/ 0 CY. = = = .O) = 2DJ (Yf o CY.2) (0..)(O.162 J 6.)(O.2[DJ (Xf 0 CY. 0) + D2 (Yf 0 CY. O) + D2 (Yf 0 CY. Using the notation of the previous proof.)(O.0) DJ(Xf 0 CY.O) = Y Yf(p) Now (J) by (e) by (e).2)(0. = (/0 CY. Since (b) gives we have (3) D2.. 0) XXf(p) .)(O.J (/ 0 CY.2)(0.2(/ 0 CY.J (/ 0 CY.2)(0. 0) by (d) XXf(p) .df 0 CY.. ) (I. Chapter 5 c"(O) = 2[X. 0) + 2D2.0) + D2.Yf o CY.2YYf(p) .2YXf(p) . 0)] = 2XYf(p) . Y]p .df o CY.)(O. 0) = DJ.0) by (c) and (f).. 0)] by (d) and the chain rule + D2.)(O. 0) + 2D2. 0)] = 2XYf(p) .2 (/ 0 CY. (2) 2D2.) (O. 0) + D2.J (/ 0 CY. THEOREM.2)(0.J)(O.O) + D2. DJ...2XXf(p) + D2.2)(0.J)(O. O) = Dd.J)(O.2) (0. I) PROOF.O) + D2( Xf 0 CY.2 )(0.0) + 2D2 (Xf 0 CY. J (/ 0 CY.O)..
from 163 we have (4) D2.1 (/ 0 CY(1)(0. 0) + 2D2.T4!ctor Fields and Diff erential Equalions Finally. 0) = YYf(p) + 2XYf(p) + XXf(p) . . 0) == Substituting (J)(4) in H yields the theorem. D1 J (f 0 CY(1)(0.0) + D2. .:.2 (/ 0 CY(1)(0. by (c) and (f).2 (/ 0 CY(2)(0.
we would like to solve equations by where / : (c.x» wherever it occurs in the proof. For the first component function & 1 this means that .x» .x» a (O. In general. Then there is a flow I(s. X) = /(<>(I.x) = 1 a' (O.c) x V + �n+l i<> (I. one can replace 0 by 10 everywhere in the proof of Theorem 2.X) = X. for some to.<>(I.c) x V + �n . Define I: (c.164 Chapter 5 ADDENDUM 1 DIFFERENTIAL EQUATIONS Although we have always solved diff erential equations with the initial condition we could just as a at<> (I.x) = /(I. /(S.X» al <>(O.X). at<> (I.lo. 2) = a : (b.<>(I. x). b) x W + � X �n with (a' . x» by /(I. S.x) = x. a a ata(I.x) = x. To prove this.s. X) = /(a(l.X» <>(O.x) = s. or else just replace <> by 1 >+ <> (1 . well have required. x) = (s. s. x) = ( I . Another omission in our treatment of diff erential equations is more glaring: the diff erential equations <>'(1) = /(<>(1» do not even include simple equations of the form <>'(1) = g(I). One way t o d o this i s t o replace /(<> (1.s. a _. let alone equations like <>' (1) = 1<>(1). that <>(IO. There is also a clever trick. S.
r�ctar Field. Finally. alld Diff erential Equalion.x) = s + 1.& 2 (I. a 2 ii/& (I. x) '" (s. /(I. This remark allows us to prove an important property of linear diff erential equations. In this case If c is any n x n (constant) matrix. {J (I. If g is a continuous n x n matrixvalued function on (a.X» . we could also have arranged for {J (10. x) = x (by first finding & with & (lo. then the solutions of the equation a'(I ) = g(l) · a(l) can all be defined on (a. S. X) = g(I) ' x. .S.x). na t by considering the curve 1 r+ (J (I . x) = /(&(I. PROPOSITION. distinguishing them from general differential equations a'(I) = /(I.s. s.X» = /(& ' (I. where g is an 11 x 11 matrixvalued function on (a. 1 7.S.&2 (I. a(I). x). consider the special case of a linear diff erential equation a'(I) = g(I) . Then is the desired flow with a ii/{J (I. even if / : (a. b). then (c · a)' (l) = c · a'(I) = g(I) ' c · a (l) so c·a is also a solution of the same diff erential equation.X» (J (O.b). which may have solutions defined only on a small time interval.lo.!' thus 165 For the second component & 2 w e have &' (I.X) = X.X» '" /(s + 1. b).a(I)). b) x �n + �n is Coo. X» . Of course. S. x) = /(I..S.
X) = g(l) . Ci must be defined for all I with a < I :s 10 . Extend it as far as possible. a contradiction. (J'(I) = g (I ) . let I. �. .) "1 0. c . b) we can solve the equation. Similarly. •: . in a neighborhood of 10. 1 hus Ci may be extended past I) as c . f I near I. Hence there is c with (c · (J}(I *) = Ci(I*). (J(I) (J(I. Pick (J with Then {J (1 * ) "I 0 for 1 * < I. (J coincides with Ci on the interval where they are defined. So for any 10 E ( a. or close enough to I. be the least upper bound of the set of I 's for which it is defined. x is locally Lipschitz. If the extended solution Ci is not defined for all I with 10 :s I < b.166 Chapter 5 PROOF Notice that continuity of g implies that /(I. with any given initial condition. By uniqueness.
2 X l (q) The map q >+ (X l (q). Xn defined in a neighborhood of p E M and linearly independent at p . . . A simple example (Problem 20) shows that even this modest hope cannot be fulfilled in dimension 3.x2 (q» is Coo. Similarly. On the other hand. and that X. 0).. . is the required diff eomorphism . PROOF. whose parameter cUn"S in the i th direction are the integral curves of the X. Its inverse. . Given 11 vector fields Xl .. (O) = (e.. However. with Jacobian equal to I at 0 (these facts also follow from the proof of Theorem 7). for we might not have [X. in the special case of dimension 2. we know that there is usually no immersion f: V + M with p E f(U). . an ) is called a parameter curve in the j th direction. We can assume that p = 0 E �2. . Then there is an imbedding f : V + M. where V c �2 is open and p E f(U). the curve I t+ / (a} . .)o. . whose i th parameter lines lie along the integral curves of X. we might hope to find an immersion f for which the j parameter cUIVes in the i th direction lie along the integral curves of the X.T4!ctor Fields and Differential Equations ADDENDUM 2 PARAMETER CUR VES IN TWO DIMENSIONS 167 If f : V + M is an immersion from an open set V C �n into an l1dimen sional manifold M. l. such an imbedding can be found: 18. .. .. X 1 = o. •:. Let XI . q is on a unique integral curve of X through a point (X l (q). Every point q in a sufficiently small neighborhood of 0 is on a unique integral curve of XI thl"Ough a point (0. .. aj+l . . but have diff erent parameterizations.. . . . PROPOSITION.X2 (q» we proved precisely this fact in Theorem 7. aj_ l . in a sufficiently small neighborhood of 0. X2 be linearly independent vector fields in a neighbol·hood of a point p in a 2dimensional manifold M.
crt»�. then the map (x. and f which f(t.y) r+ (a(x). where V c �2 is open and p E f(V). whose i th parameter lines lie along the integral curves of Xi . for example. or c . g(y» takes the diagonal {(x. . which gives us considerable flexibility. Consequently.168 Chapter 5 We can always compose f with a map of the form (x. y) r+ (x. y) r+ (c. y). or X2 . x)} to C. t) = crt). by composing with (x. If. we can further arrange that crt) maps to (c(t). C C � 2 is the graph of a monotone function g. Then there is an imbedding f : V + M. tl(y» for diffeomorphisms a and tl of �. Moreover.' (x). and let c be a curve in M with c(O) = p and c'(t) never a multiple of X. we can state 1 9 . PROPOSITION. f any particular or parameterization c = (c" (2 ) : � + � 2 of C. Let X" X2 be linearly independent vector fields in a neighborhood of a point p in a 2dimensional manifold M.
3. the equation Ci'(I) = yCi(I) Ci(O) = X. 5. (b) If Ci : M + N is a diffeomorphism. VYe sometimes have to solve equations "depending on parameters") a ii/Ci(I. but f has no fixed point.x) : (b. p) and a function f : M + M such that p(f(x). Ci(I). y) (t) = Ci(I. y). then Ci. if we write Ci (x . (a) If Ci : M + N is Coo.b) and Ci"(I) = f(l. (c) If Ci : � + � is Ci(I) = 1 3 . y.b) x W + V is Coo. Yo) E V x V. x. y) for all X. X» Ci(O. b) + V with Ci'(I) E V for I E (b. V C �n are open. 4. Let f : (c.Ci' (I» Ci(O) = X Ci'(O) = y. . c) x V x V + �n . b) + V for each initial condition x and "parameter" y. Yo) and a num ber b > 0 such that for each (x. y) E W there is a unique Ci = Ci (x. Find an example ofa complete metric space (M. x) = f(l. y E M. y. and we are solving for Ci (y. Ci(I). where V. f(y» < p(x. then Ci: (b. then Ci. Prove that there is a neighborhood W of (xo. : TM + TN is Coo.y) : (b.[ector Fields and Differential Equalions 169 PROBLEMS 1. then there is a Coo vector field X on � such that Ci. where f: ( e. Ci(I. 2. and X is a Coo vector field on M. X is a Coo vector field on N. For example. {J (I» . y. y. Find a nowhere 0 vector field on � such that all integral curves can be defined only on some interval around O. X is not a Coo vector field. x) = x. { Hinl: Consider the system of equations Moreover. and let (xo. for open V c �n and V c �m. Ci' (I) = (J(I) (J ' (I) = f(I.c) x V x V + �n be Coo.
x» . x» j(t. x» · D2a(I. x) = !(a(l. x) = !(I. a (l. x) = (y. and that if the derivative with respect to these arguments is denoted by D2a.. (b) Show that equations of the form (**) a ai"a(l. a(l. x) = (O.x) = x can be reduced to equations of the form (*) (and thus to equations a ai"a(l. xo). ultimately). a(l.).x» . W + �m X �n is a flow for j in a neighborhood of ( c . y. j: (b. the hard part is to prove that if ! is C ' .a of the fOl"m (**).a(l. y.b) x (a' . so that a ai"a(l. (a) Define by If Chapter 5 a(l) = xeyt. D.j(I.a(l. a2) = a : (Yo.b) x W + V. y. y. if ! is C 2 Diff erentiability of class C k is then proved similarly. c) x V x V + � m X �n show that we can write for some a. X) (a result which follows directly from the original equation if ! is C 2 . by induction.! is C ' .x» . i. since D. y.x.170 with solution is such a case.x) = D2/(a(l.e. y. x» a(O. x) = (y. [When one proves that a C k function ! : V + �n has a C k flow a : (b. y.x). then a is diff erentiable with respect to the arguments in W.] D. = D2D.x) = !(I. D2a(l. x» a(O. and conclude that a satisfies (*). then D. x) = !(a(l. Since (***) is an equation for D. it follows that D2 a is differentiable if D.
A2 A' A4 j)th = BA.' ) = T (exp A)T' . where g: � + �. (b) Let A = (aij ) be an n x n matrix. we have I A I s i A l s n I A I.A) = I . is clearly differentiable (it is even analytic). then exp(A + B) Hint: Write and show that I R N I + 0 as N + 00.= p i L i L p. (e) If AB = eA "" I + A + 2! + 3! + 4! + . (f) (exp A) (exp . The remainder of this problem inves tigates the extent to which similar results hold for a system of linear diff erential equations. and let IA I denote the maximum of all l aij I. Show Ct(I) = ef g(r)dr . p=O =O 2N (A + BJ P ( = (exp A) (exp B). entry of the partial sums converge absolutely for each (i. (a) Consider a linear differential equation 171 Ct' (I) = g(I)Ct(I).[eclor Fields and Dijjerenl. where f g(l) dl denotes some function G with G'(I) = g (one can obtain all positive multiples simply by changing G). so exp A is always invertible. 2 2 (g) The map exp.) . considered as a map cxp : �n + �n . p. the usual norm of A E = B (= exp(O) . B).ons 6. (d) Show that exp(TA T. .. �n2 . so that we are solving for a realvalued function that all solutions are multiples of Ct. p=O p. Show that IA + BI s i A l + I B I IA B I s n I A I · I B I · (c) Conclude that the infinite series of n x n matrices exp A converges absolutely [in the sense that the (i.al Equal. N AP )( N BP ) + RN (Notice that for I A I. . ill and uniformly in any bounded sel. Show that exp' (O) (B) L .
If B'(t) denotes the matrix whose enllies are the derivatives of the entries of B. Check that if the coordinate system x is x = XI . �n + M.) (fo' g(s) dS ) 8. U) is a coordinate system on M. defined by al a(f( · . then N + � p >+ I( p . so every system of linear equations with constant co efficients can be solved explicitlythe exponential of J� g(s) ds = t A can be found by putting A in Jordan canonical form. show that B'(t) = A'(t) · exp (A (t» . for x : X= a/ax' is equivalent 10 X.) (j) Show that the linear differential equation et'(t) = g (t) · et (t) has the solution et(t) = exp prollided that g(s)g(t) = g(t)g (s) for all s. show that for every Coo function I : M + �. (This certainly happens when g(t) is a constant matrix A . (b) If <p : (B. (This is clealiy true if A(s)A(t) = A ( t ) A (s) for all S. ( a/at ') = X 0 x. t . let 1( · . show that the function al/axi. For a Coo function I : M x and q E N.B) x M + M is a I parameter group of diff eomorphisms.172 Chapter 5 (h) Use the limit established in pan (g) to show that exp' (A)(B) = exp(A) · B if A B = BA . (a) Let M and N be Coo manifolds. axi is a Coo function on M x N. t. 2 erentiable. and let (i) Let A � + �n b e diff : B(t) = exp(A (t» . q) denote the function from M to � defined by 7. If (x. q) = � (p). provided that A(t)A'(t) = A'(t)A(t). q » (P . the limit . q).
unnecessarily clumsy way. isjust a special case derived in an The formula for Lx dx . (b) How would Proposition as 8 have to be changed ifwe had defined (Lx Y )(p) (b) Using (a). In the next part we get a much simpler proof that Lx Y = [X. Lx that for Y E Mp. x TM + TM is defined by (c) If 4>* : 1 73 (e. (c) Let X and Y be vector fields on M. define Lx df = d(Lx f). Give the argument to show that tion 8. 9. the limit 4>* (1. using the technique which appeared in the proof of Proposi tion 15. (a) Prove that Lx(f · w) = Xf · w + f · Lxw Lx[w(Y)] = (Lx w)(Y) + w(LxY). show directly from the definition of II. (v) . (a) Show that [Lx df(p)](Yp) = Yp(Lx f).. e) show that 4>* is Coo. If X generates {4>.Hxtor Fields and Diff erential Equations exists. and conclude that 4> * (df)(Y) = Y(f 0 4» . Y]. i . 4>h* Y"'_h(P) + Yp in the proof of Proposi 10. and f: M + IR a Coo function. and defines a Coo function on M. v) = 4>. derived in the text. X and covari exists and defines a Coo function on (d) Treat Lx Y similarly. M. and conclude that for every Coo vector field ant vector field w on M. }.
Conclude that for c ( h ) = a(h. on M. If V I .* A)(p) . 1 . b .. </>* Vk . . lim . . (</>. . we define </> * A on M as follows. we define M generates {</>.+ x ax a X + b Y + cZ is an isomorphism (from a certain set of vector fields to and that [U. .x. . (r l ) *A[). Y.. . . A[ E Mp* . . . . 13. (b) Show that the flow of aX + b Y + cZ is a rotation of about some axis through o. Vk . . V] r+ the crossproduct of the images of U and V. Z=y ax ay (a) Show that the map a Y = z. . .1 74 Show that Chapter 5 Dl a(O. 0) = Yp(Xf). A I > . Vk E Mp. O) = . (b) If the vector field X on (a) Check that under the identification of a vector field [or covariant vector field) with a tensor field of type (�) [or type this agrees with our old </>* Y.c' (O) 1 2. . h) we have . c) E 1R3 1R3) 1R3 14. . and A is a tensor field of type (J) (�)J I (LxA)(p) = "_0 _ [(</>. On 1R3 let X. Check the Jacobi identity. Z be the vector fields a X = Zay a = Lx Y(p)(f) = [X. ). Yilz a az a a . and A I . . . If A is a tensor field of type on N and </> : M + N is a diffeomorphism. . then (�) [4>*A(P)](VI . .1 ) * A I > . . .Xp( Yf) D2a(0. Y]p(f).A(p)]. . r+ (a . A[) = A (</>(P»(</>* V I .
. (�). . h � (LxA)IJJ . . (CT)(v" Show that = contraction of . . Xk . . X cr LL 0'=1 . 'J / . . Xk . A . . .).. WI . . Ap+ " . . . k I . . . . WI ) can be obtained by applying contrac tions repeatedly to A ® X. A + Lx. W I . x . aai .lk . .. A) f+ T(V I . . Lx Wi . . . Xk . . A � J . . . V. J ' a i + " " Al J.. A = Lx. .=1 I . . Vk . +x.J l}cr+J . . . Lx(CA) = C(Lx A). . . Lx(A (XI .. V. . . LxXi. . use (d) to show that (LxA)(X" . .=1 . . WI).." A" . . Xk . . ® w[. . a axj n ai a/axi .. W I . = LxA + LxB Lx. .•laJ11a+J ·'·'k .=1 show that the coordinates of Lx A are given by k n . .! ! !cr A' J ···. . W" . j n . . Xk . . .. . A /_ ' ) (v ._ I .. (�). • . .' and Diflmlliiai Equations Show that 175 Lx (A ® B) = (LxA) ® B + A ® Lx B (so that in particular). Ap_" A.=1 n . .!t _ ! ··. . wI) '':':: a " ai� " " LL L ax' cr=l j=1 . . WI) + L A (XI . .=1 + L A ( X1 . V. . . WI») = (e) Noting that A (X" . W I . Hint: We already know that i t is true for A o f type (d) Let be any contraction (�). ® . . (f) If A has components Ai ' in a coordinate system and X = L . Vk_" A" . . . . . . .lleclor Field. . . . (c) Show that Lx (A + B) Lx{fA) = X{f)A + jLxA). .lk' = l . . . ... A/ _. . . . . . • . . . . ® Xk ® W I ® . .+ I .
2 (Use Taylor's Theorem .1 76 Chapter 5 15. then this unique extension is Lx . 1 (. (a) Show that D has a unique extemion to an operator taking tensor fields of type to themselves.h. DA X = A (X). Show that the tangent vector e"(O) defined by e"(O)(/) = (1 0 e)"(O) can also be described by e"(O) = 2y'(0). show that Generalize to tensors of type 16. (d) Show that LJx = ILx  DXfj!!dJ . 0») Let A be a tensor field oft)l)e G).g(O) h  _ /. (c) Show that ( DA W) ( P) = A ( p) * (w ( p» . so that we can consider A ( p ) E Elld(Mp). then DA has a un�'1ue extension satisfying (I). Y. (e) If T is of type (�). such that (J) (2) D ( A (I) D is linear over ® B) = DA ® B + A ® DB IR (3) for any contraction C.) (b) Given c : IR + M with c'(O) = 0 E Mp. DY + DI . Show . Define g (t) = that the righthand derivative (�). (0) . Hillt: Check this for functions and vector fields first. DC = CD. and the Coo vector fields V to V./1 ) for t ::: o. If we take DI = XI and D Y = Lx Y.. Show that if we define DAI = 0. (a) Let I : IR + IR satisfy 1'(0) = o./I ) for t ::: o. Let D be an operator taking the Coo functions F to F. and (3). _ r g+ (0) . then A (X) is a vector field for each vector field X. . (2).!� g + (h) . define y(t) = e ( . such that D : F + F and D : V + V are linear over IR and D U Y) = I .
be the curve of Theorems 15 and 16. Yp E M and g : N + I (X. = denotes a function such that 19. (c) The rank of (d) Let I : M + define Show that (a2fjaxi axj (p» N p have •• a2I axi aXj (p).L1 L j= "= a' . choose vector fields X. For Xp.leclor Fields and DijjerClltiai Equations 1 77 1 7. (a) Let vectors Define I : M + IR have as a critical point. and b J if} a' b J . (e) If IR + M has as a critical point. IR is independent of the coordinate system. Let around pe x (p) 0. Y with 51p = Xp and Yp = Yp.p = O. show that 1 (Xp. . (b) Show that I •• n n (LI axa i Ip ' j�1n . p Using the fact that [X. (a) If M is compact and is a regular value of I : M + IR.. Given Xp. : Mp Mp + NJ(p) x ( (( is a welldefined bilinear map. Yp E Mp. as a critical point. xa IP) = i. then there is a neighborhood U of E IR such that I. show that where [X. axi Ip ' a '. = 0. If with = and e"(O) defined by e"(O)(f) = (f 0 e)"(O). 0 0 11 (0) . Y]p = n a f..I (U) is diffeomorphic to x u. Yp) •• is symmetric. show that e: 0 c•• (0) : 1R0 x 1R0 + Me(o) takes ( 1 0 . 1 0) to the tangent vector 18. so that I. Y) g ) = 51p Y g 0 I»· I. x is a coordinate system 0(/2) xi (e(/» ai/ 2 + 0(/ 2). Y]p(f) conclude that it is welldefined.
and let X will be a vector field one of whose integral curves is the xaxis.and zaxes. as shown in the first part of the figure below.I (q2 ) are diffeomorphic for q l . does it follow that M is diff eomorphic to I.178 Cha pter 5 by a diff omorphism </> : 1 . while certain other integral curves arc parabolas in the planes y = constant. let Y and . 0») More generally. then II (qI l and I. (c) It follows f rom (b) that if all points of N are regular values. then there is a neighborhood U of q and a diffeomorphism </> : II (q) x U + II (U) with I(</>(p.q')) = q'. 20. If I is onto N.X = dldl. q2 sufficiently close. Hint: e Use Theorem 7 and a partition of unity to construct a vector field X on a neighborhood of II (0) such that I.I (q) x N? Z b e unit vector fields always pointing along the y . show that Proposition 18 does not hold in dimension 3 . I)) = I. Using the second part of the figure. In IR ' . respectively. if M is compact and q E N is a regular value of I: M + N.1 (0) x U + II (U) with 1(</>(p.
S. is spanned by a vector field locally. that is.p p I on a manifold M may be definable only for some small time interval. A Mathematician's Apolog)' though the vector field is COO on all of M. sup pose that for each p E M we have a I dimensional subspace /. The function /. H. G. We will now vary our question a little. find an integral curve c of X with initial condition or e(O) = p.q f q in a neighborhood of p. 1 79 . is called a I dimensional distribution (this kind of distribution has nothing whatsoever to do with the distributions of analysis. . but we define a (Idimensional) submanifold N of M to be an integral manifold of /. Instead of a vector field. A Afatllcmaticia1l1 A1iscellan ). For a given p E M. . we have seen that the integral curves of a vector field where i : N + M is the inclusion map. . a Coo distribution if such a vector field X can be chosen to be Coo in a neighborhood of each point. by defining N to be {c(I)}. we just choose a vector field X with 0 # Xq E /. we can always find all integral manifold N of a Coo distribution /. quoted in]. which include such things as the "afunction"). . so that global results can be obtained. Besicovitch. Then /. [Pioneer work is clumsy] A.p C Mp . This al'gument actually shows that for every p E M there is a coordinate system (x. Littlewood. and then forget about the parameterization of e. even n the previous chapter. U) such that for each fixed set of numbers a 2 . the set {q E U : x2 (q) = a 2. . E. xn (q) an } = . • . We call /. Beauty is the first test: there is no permanent place in the world for ugly mathematics. with p E N.CHAPTER 6 INTEGRAL MANIFOLDS PROLOGUE A mathematician's repu tation rests on the number of bad proof" he has given . Hardy. if for every p E N we have i* ( N ) = /. a " . we can choose (in many possible ways) a vee to!' field X such that 0 # Xq E /.q for all q in some open set around p. For a l dimensional distribution the notion of an integral curve makes no sense.
The entire manif old M can be written as a disjoint union of connected integral submanifolds of !. but on the . there is a distribution 011 lhe torus whose integral manifolds all look like the dense I dimensional = 1 11 . there is a distribution the torus .1 1 ���l il l l l l l 011 submanif old pictured in Chapter 2.hkh has olle compact connected integral manif old.. rather than curves with a particular parameterization. This is still a local result: but because we are dealing with submanifolds. and all other integral man ifolds nOlicompact. On the other halld. It happens that the integral manif olds ofthesC' two distributions are also the integral curves for certain vector fields. For example. which locally look like  (rather than like or something even more complicated). we can join overlapping integral submanifolcls together.180 Cha pter 6 is an integral manifold of !. .. on U.. and that these are the only integral manifolds in U.
then /.b. for each q U.c) is spanned by E pEN i*(Np) = /. b(x � . s. .p = /.p = If we identify T� 3 with � 3 x � J .q. . Thus /. . a Coo distribution if it is possible to choose Coo vector fields X" . Xq with this property. Xk such that X. consider' the 2dimensional distribution D. ( q). S E } (r. in a neighborhood of each point p . C For any P there is a neighborhood U is a k dimensional subspace of and k vector fields X" . integral manif olds do llot exist. .br)p. We are leaving out the details involved in fitting together these local integral manif olds because we will eventually do this over again in the higher dimen sional casco For the. .p consists of all may be pictured as the plane with the equation z {r �Ip + s �lp + br� lp : r. . As the simplest example. . . Thus /. M E M N M i: N M + p Mp is the inclusion map.p where Mp.P l where D.Integral Alanifolds 181 ]\1obiu5 strip there is a distribution which is spanned by a vector field only locally. A kdimensionaI distribution on is a function p 1+ D. In general.a).(a. A (kdimensional) submanifold of is called an integral manifold of /. even locally. in �3 for which /. Xdq) are a basis for /.p c= . . if for every we have Although the definitions given so f all look the same as the I dimensional ar casC'. the results will look very different. . .moment we \'\'ill investigate higher dimensional cases only locally. We call /.
yox)}.z) and the (y. : ) } .z)} would be a curve y in the (y. b. b. with 0 E N. Suppose there were an integral manif old N of /. The only such vectors have third component 0. For example. s o i t must b e the linc { (x. f each fixed Yo.'ork. This will be a curve in the plane { (x. c) isjust parallel to the one through (a. a proof can be given as follows. z)plane.c) through (a. The intel'section of N and {( O.. 0. 0) contains vectors with third component nonzero.p for points p = (a. with all tangent vectors having slope Yo. you can probably see that it has no integral manifolds. The plane /. yo. 0). Yo. b.b.(a. yo. . so Y lllust be the yaxis. y.y.. z)} through (0. 0). its tangent space at ( 1 . � � � � � � � � � � If you can picture this distribution. Now consider.182 Chapter 6 The figure below shows /. or the intersection N n { (x. Our intcgral manifold would have to look like the following picture. O). yO.(O . :)plane through 0 whose tangent vectors would have to lie in the intersection of /. But this submanifold c10es not .
f.c) = D.a) +g(a.b)(x .p if and only if I I aa a a . : p Now the tangent space at aa f(a. b. Since /.b. the submanifold is given locally as the graph of a function: N = ((x. (a.a(a. the plane /.y.(a.b) . geometrically. aa g(a.Ip + .c) through (a.b) = ay (a. (a..c = f(a.Ip . is r· l � } We now ask when the distribution /.b) + sg(a. ay = g. y) plane. /.b) = a. has an integral manifold N through each point.p is the plane with the equation 0 .. b).b. + a.(a.b).b).p is never perpendicular to the (x. ax These tangent vectors are in /. b) if ' p p .s E IR .. since f and g depend only on a and b. y)}.Integral Manifolds 1 83 p /.b)) is spanned by a aa a a. As in the first example. S o we need t o find a function a: 1R2 > IR with aa � .b)] Ip : r.z) z = a(x.p = �{ + s � p + [rf(a. To see in greater detail what is happening here. consider the somewhat more general case where D.b) (y . c) will be parallel to the one through (a. 0). aay ay • = (a.b. b.(a.
= b. we find a necessary condition on f and g: In our previous example.O) = :0 + namely. 0) = :0 (I) a. we define foX 1(I._0' . O). aa (x. I 0'. It is also wellknown that the neces "JI'Y condition (**) is sl!lJirienl fOl' the existence of the function a satisfying (*) in a neighborhood of any point. By using the equality of mixed partial derivatives.. a(x. b ) g(a. 0) = l(x. If I. al = ay ax I. g : 1R2 + IR satisfy in a neighborhood of 0. and ':0 E neighborhood of 0 E 1R2 . '!§. then there is a function a(O. = 0' so this necessary condition is not satisfied. PROPOSITION. O) dl. such that �. O.O) = :0 �=I ax aa ay = g. 0) so that a(O. b) = 0.1 84 Clzapter 6 It is wellknown that this is not always possible. and u. defined in a PROOF We first define a(x. I(a .
we define a(x. = 0 by (I) . To prove that it equals 0 for all shO\\' that its derivative is o. Suppose that N = : E { (x. y) so that (2) aa a:. s �} . y. the function + r 1(1. namely.y) = a(x. consider. But its derivative at y is f+ aa (x. for each 185 x. for each fixed x. ax y. 0) + = Zo This construction does 1)ot use (**). g : /'. I) dl 1 y This is 0 for )./(x. we define a(x. 0 0 laY g(x. To prove this. = �3 + �.Integral Manifolds Then. y)} . y).. (x. 0) dl + 1(" g(x. z) : z = a(x.p {r a�{ + s � Ip + [rl(p) + sg(p)] � Ip r. y). y) . we just have to VVc are now ready to look at essentially the most general case of a 2dimen sional distribution in �3 : where I. y) = g(x. and always provide us with an a satisf y� ing (2). then also aajax = f. aajay = g. I) dl. We claim that if (**) holds.
once again. b. b) a: p . b. b» = = aa � (a . b). b . a(a . b . b» + ilz (a. which means that for each ) pair (a .a(a. b» + . b. b .b) p aa a a a p + a (a. b.(a. a (a.ver ' point. b) is. b) = .b. b) ay aY az ay a x II ag a 2a ag aa (a. b» . a .(a.a(a. ' + � p � (a. a(a.186 Chapter 6 is an integral manifold of /. a(a.a(a . b» . through . a (a . Thus we obtain finally the necessary condition .p if and only if I I I I I(a. but we can substitute from (*) to obtain al al .. In order to obtain necessary conditions for the existence of such a function a) we again use the equality of mixed partial derivatives. b» + ilz (a. b» g(a. b. since it still involves the unknown function a. b. . (a. a x ay This condition is not very useful. a (a. b. b» .a(a. b) = � (a. b» · . b).a(a . y y • These tangent vectors are in /. b.b» + ilz (a. b) these equations must hold no matter what a(a.a(a . a(a. Thus (*) and the chain' rule imply that al a2a al aa . b). b» g(a.(a . The tangent space of N at p = (a. I(a. a(a. b. by aa a a a. (a. aa a. b» is spanned. Now we are looking for conditions which will be satisfied by I and g when there is an integral manifold of /.(a. b» ag ag = � (a.
such a function exists (and is automatically COO) in some neighborhood W if and only if there is a neighborhood of (0. Let U x V C JRm x JR" be open.) Moreover. we can treat a system of partial differential equations for n functions on �m (i. there is at most one function a : W + V . Then for every x E 11. there is no need to restrict ourselves to equations f a single or function defined on JR2 . . with a different twist at the end. 0) so that (I) a(O.a(l. . THEOREM. Necessity of PROOF Uniqueness will b e obvious f the conditions (**) is left to the reader as a simple exercise. . 0) = 11 (1. . O. agree on the component of WI n W2 which contains 0. 0. . Dm+. . defined in a neighborhood W of ° in JRm. .l111egral Manifolds 187 In this more general case. 0». al j (I) = Jj (More precisely. 0.e. 0. J. . m . x) E U V on which x i. J. rom the proof o f existence. . where U is a neighborhood of ° E JRm. In the following theorem. for i = 1.0. . . . . . 0. We first want to define a(l. . . .. . 0) = x aa atT (l. . and we will concern ourselves with proving existence if these conditions do hold. 1. . any two such functions a l and a2 .a(l» for ali I E W. defined on WI and W2 . so for a function f : �m X �n + �k we use al aI' al ax' for for D. j = 1. . . the necessary condition again turns out to be suf ficient. satisfying a(O) = x aa (I. In fact. Ill. a partial diff erential equation for a function from �m to JR"). . we will use I to denote points in JRm and x for points in �n . . . The proofwill be like that of Proposition 0. . . : U x V + JR" be Coo functions. . and let j. .
. . < a(l ' . . f3dl» . O. 0.O. . 0. . 0. 0. This has a unique solution f sufficiently small I. . 0) f3/(I) = h (t ' . . . . . . O. Note first that g(O) = 0 by (I). . 0) and all expressions involving Ii are to be evaluated at (I'. .O. 0» a1 2 11 ' 1 ' " III '2· We claim that f each fixcd I' with 11 ' 1 or we also have. ./. . I. defined for III III a(l.1. .L. I. (t < < < " < '2. At this point the reader must or < < " . . . . . . 0. O. . . (3) O = g(I) = iJiT (I . .0. . We then define fh (O) = x f3. 0) = f3dt) Then (I) holds f III or Now for each fixed I' with 11'1 '" consider the equation f32 (0) = a (I ' . 1. . .O. . . In the f ollowing. 0. then for 11'1 < " the solutions of the equations f f32 with or the initial conditions f32 (0) = a(l ' . f32 (1» . all expressions involv ing a are to be evaluated at (t' . . . 1. . . Define This equation has a unique solution.a(1 . . O) . . and verif the following assertion: If we choose " ). . .1. . ' (I) = jj (I. . � a/. We have a2a a/.. 1. 1. . 0» . for ali I with III aa .a(l .. 0». 0) = f32 (1) Then a(O. O. . . 0.I. . 0. O) = x aa (1 . . . . . . . " . I.0) will each be defined for III < '2 for some 82 > o. .a(I ' . 0. 0.al 2 . . . aak g. . We now derive an equation for g'(I). . (I) = al 2 al' . . .0. axk a/2 ' k= ' (4) . . . . 0. (2) 0) = h(l ' . < " . . . . I. . . we consider the ordinary differential equation refer back to Theorem 52. O.188 Cha pter 6 To do this.. sufficiently small.0. . . . . .
aft2 al by (2) by (2) again _ _ _ by definition.(I) � aft k L. a distribution with integral submanifolds through every point) then certain othel' conditions hold. Our investigation of the problem so far illustrates one basic fact about theorems in differential geometry: Many of the fundamental theorems of diff rential geometry fall into e one of two classes. axk h ah + t ah aak aft t aft f: k = al l k=1 axk al l al 2 k=l axk 2 a a = h + t h [g k (l) + ft k J al l k= axk 1 aft _ t aft f: k al 2 k=1 axk 2 t ahk gk (l) = k=1 ax = ail a ( aa2 ) . e l ) x ' " x (en . en ) and satisfies t*) . The solution with initial condition g (O) = 0.g. illustrates an even more important fact about the theorems of differential geometry: There al'e always incredibly concise and elegant ways to state the in tegrability conditions. these Con ditions are obtained by setting mixed partials equal. (3) Now equation is a diff erential equation with a unique solution for each initial condition. in which we will essentially begin anew... (5) .. The second kind of theorem justifies this terminology. •:. and prove their sufficiency. So (3) is true. Theorem I essentially solves for us the problem of deciding which distributions have integral manifolds. by showing that the " integrability conditions" afe suffi cient for recovering the nice situation. without ever even mentioning partial derivatives. given by (4). al .k=1. is clearly g (t) = 0 for alI I .Integral Manifolds and thus 1 89 (5) g . and are called "integrability conditions". It is a simple exercise to continue the definition of a until it is eventually defined on (e l . The remaining parts of our investigation. The first kind of theorem says that if one has a certain nice situation (e.
I a L a' a. PROPOSITION. a n . ' L IY�. are Coo (problem 3) .= 1 .1 90 Chapter 6 LOCAL THEORY If I : M + N is a Coo function.= 1 implics that the functions I.I (a . then there is a unique Coo vector field X on PROOF. . . V) around I(p) E N such that ' t J' 0 l o X. axl Y= f( p) . In One case. we say that X and Y are Irelated if l p (Xp ) = Y p) for . Thus if where (i are Coo functions. . •:. (Yg) 0 I = XU o g) . the latter condition is fulfilled: 2. . . and Y. are Irelated. . This ( a�. n . fOI' i [Y" Y21 are Irelated. Xp . . a given vector field X may not be Irelated to any vector field Y. This is easily seen to imply that I p. PROPOSITION. a. f(p) E I (Mp ). . 0 I = I. . f( each P E M. Clearly we must define Xp to be the unique element of Mp with Y p) = I . respectively. Of course. If g : N + JR is a Coo function. M which is Irelated to Y.. we use Theorem 210(2): there are p f( coordinate systems ( x . 2. then Y p) (g) = l Xp (g) . i f this i s true for all Coo functions g : N + JR. then [X[ ' X21 and . To prove that X is Coo. nor must a given vector field Y be Irelated to any vector field on M. . U) around P E M and (y.. IJ = a�. Let I : M + N be a Coo function such that I is an immersion. then X and Y are Irelated. then where The most important property of Irelatedness for us is the following: 3. and X and Y are Coo vector fields on M and N. 0). a n ) = (a . . If Y is a Coo vector field on N with Y p. . If X. .p f( = Xp (g o I). y X= n = 1 . O. so Conversely. .
to have an integral manifold through every point. Y] also belongs (I) So 0 0 0 0 0 0 If g: N + IR i s Coo.X2 (X. •.X2 ([Ytg] f) by (I). and Proposition 3 then shows that [X.(g f) i = {[Y" Y2]g} f = {Y. f] p = [X.g and Ytg. 0 + to /. f]p E Np .. The vector fields X = � + f� ax az a Y = a:ay +gazbelong to /. ( [Y2 g] f) . if Xp E /. Suppose that N is an integral manifold of /. X and X are i related.g) f = X. 0 0 0 . this therefore shows that [X. ''''le will say that a vector field X belongs to /. Thus i. 2. fp E Np such that Xp = i. and i : N M is the inclusion map. 191 (Y. Consequently. Y]p. in 1R 3 given by [X. which is precisely the condition for /. S E IR } .IntegraL JvlanifoLds PROOF.p .� + f� _ g af ) � . (Y2g)} f .(g f» .Y] = (� . For a moment look back at the distribution /. p In other words. Yp = i. Using the f ormula on page 1 56. Y]p E /. /. then 1 . only when the expression in parentheses is 0.f .p = { I' f lp + s hip + [rf(p) +sg(p)] �Ip : I'. and Y and f are irelated. ax ay az az az .[X. we see that This belongs to D..p for all p. if there is an integral manifold of /. f] and [X.. respectively = XdX. (g f» by (I) = [X" X2](g f) : Now consider a k dimensional distribution Do. with g replaced by Y. then [X. through every point p. Proposition 2 shows that X and f are Coo vector fields on N.. Y] are irelated. then for all p E N there are unique Xp .{Y2 (Ytg)} f = X.Xp. Here [X. If X and Y are two vector fields which belong to /.
Xj] = L C/. . THEOREM ([HE FROBENIUS INTEGRABILITY THEOREM.. FIRST VERSION). For every p E M there is a coordinate system V) with x . . /'. is called integrable if belongs to /'. in a neighborhood V of p. Let /'.gj Xj] We arc now ready f the main theorem.. 5. . e) such that for each a k + I . is integrable. ... . x n (q) = a n } is an integral manifold of /'. if X. x(p) = 0 ( . Theorem I can be derived f rom it (Problem 7).e.gY] = fg[X. Y] X and Y 4.. [Xi. the set {q E V : xk + l (q) = a k + l . . Xj] is a linear combination k [Xi . Y and [X. It is equivalent to Theorem I. Xj ]q q X k X = L1 . . Y] [jX.. This condition can be checked fairly easily: [X. restricted t o V i s contained i n one of these sets.. x (x.gY] belongs to /'. .Xi.fiXi i= f or Coo functions C. Since we have clearly [X. ) a n with all fa i r < e.. . e).. . X.J ' Such functions clearly exist if /'.. X. If and belong to /'. Xk span /'. ..(q). [f. . Y] f(Xg)Y g(Yf)X. suppose such f unctions exist.. [Xi . then /'.. •. But the proof is quite diff el·ent.Y] do : + . . If is integrable on V if and only if each X" .. in or fact. it obviously suffices to treat each separately. we can clearly write PROOF Y To prove belongs to /'. x IV) = ( e... since E /'.1 92 Chapter 6 In general. whenever belong to /'. be a Coo integrable kdimensional distribution on M.. PROPOSITION.. Any connected integral manifold o f /'. l Iich is spanned by the Conversely. [jX.
. . Then 1'>0 + IRko is an isomorphism.0 C �no is spanned by Let IRn + IRk be projection onto the first k f actors. . of Nq we h ave d(xm i)(Xq) = Xq(xm i) = i. x" (q) = a n } are clearly integral man ifolds of 1'>. •. since i. ][* is oneone on f or near O. j j By Theorem 514. with p = O. . . By Proposition 3. So near we can choose unique 1[: D. k .Integral Manifolds 1 93 PROOF.Xq E I'>q. is oneone on I'>q. I. . (on a neighborhood of 0 E IRn ) and a/aI (on IRk ) are 1[related. But. If N is a connected in tegral manifold of I'> restricted to U. 0 . k . . = ax' i = I. . Moreover. . So [X" X ] = o. 1[. consider d(xm i) for k + I s. which implies that xm i is constant on the connected mani f old N : so that + 0 0 0 m S.X. k. which is spanned by the a/axj l q f j = I . . . We can clearly assume that we are in �n . . we can assume that .[X" Xj ]q = [ aaI" aaIj ] . 0 . . since their tangent spaces are spanned by the a/ax' = X. 11. . . The sets {q E U : Xk + l (q) = ak + I . . there is a coordinate system x such that a X. .6. (q) = aaI. .q q 1[.. ' Then the vector fields X. 1[. Thus or d(xl1l i) = 0. .(q) = 0. . : 0. and 1[. By continuity.(q) i = I. k . with inclusion map For any tangent vector Xq i: N U. [X" X ]q E I'>q by assumption. for i = I..Xq(xm) = 0.
). we see that we can cover M by a sequence of coordinate systems (Xi . ( () Using Theorem 12. Let !. {q E V : xk+l (q) ak +I .xn (q) an } = = NnV are the sets of the form Each component of N is called a folium or leaf of the foliation N. For such a coordinate system V). a (usually disconnected) k dimensional submani fold N of M is called a foliation of M if every point of M is in (some com ponent of) N. Then M is foliated by an integral manifold of !. . But S n Vj has at most countably many components. . let us call each set (x.. . we introduce the following terminology. PROOF. (each component is called a maximal integral manifold of !. . . It is possible for a single slice S of Vi to intersect Uj in more than one slice of Vj' as shown below. . THEOREM. Vi ) satisfying the conditions of Theorem 5. be a Coo kdimensional integrable distribution 011 M. . x (0.1 94 Chapter 6 GLOBAL THEORY In order to express the global results succinctly. If M is a Coo manifold. with xIV) = (0. V). Notice that two distinCI components of N nV might belong to the same leaf of the foliation. . xn(q) an } = = a slice of U.. such that the components of {q E V : Xk+l (q) ak+I . 0). 6. 0) x . . and if around every point P E M there is a coordinate system (x.
. will be called joined to p if there is a sequence 0 = io . Vo) with P E Vo. For q #. the first union. .1 . [If we are allowing nonmetrizable manifolds. and only countably many such sequences. A slice S of some V. M is foliated by the disjoint union of all such submanifolds. . . Since there are at most countably many such sequences of slices for each se quence io. sO S nVj is contained in at most countably many slices of Uj . . choose a coordinate system (xo. . and let So be the slice of Vo containing p. there are at most countably many slices joined to p. however. Appendix A describes a nonparacompact manifold which is foliated by a lowerdimensional connected submanifold. . . In this case. i/. . . Consequently. il = i and corresponding slices with a = 0.p.] . since we do not have to find a countable number of coordinate systems for each leaf. .. the proof is even easier. the discussion to follow will not be validin fact. the corresponding union is either equal to. Using Problem 31. . we see that the union of all such slices is a submanifold of M. i 1 . or totally disjoint from.InlegmL ManifoLds 1 95 and each component is contained in a single slice of Vj by Theorem 5. 1 . Given P E M. •:. . and can merely describe the topology of the foliation as the smallest one which makes each slice an open set. this di'!ioint union is clearly an integral manifold of /'.
choose a coordinate system f(p) such that the slices MI _ (x. otherwise CiV ' . this folium would contain an uncountable disjoint family of open sets. . PROOF... and a f olium of the f olia tion determined by some distribution /'. According to Proposition 21 1. . (x. Let be a Coo manifold.1 96 Chapter 6 is a coordinate system of the sort considered in the proof Notice that if of the theorem.U) around {q E U : Xk+l (q) = ak +I . Let be another Coo manifold and f: Then f is Coo considered as a a Coo function with C + map into as a map into Given P E P. P M MI . THEOREM. . then infinitely many slices of may belong to the same f olium. U) U However. . at most cOUil/ablJ' man slices can belong to the same f y olium.. Now f is continuous as a map into M. . . This allows us to apply a proposition from Chapter 2.. 7.xn(q) = an} are integral manifolds of /'. so f takes f(P) MI . it suffices to show that f is continuous M P MI . .
by continuity. and x . For k + I :s i :s n. (f(p'» #. Consequently. for any p' E W. •:. . 0 f would take on all values between a . contradicting the fact that f( W) e M. (f(p'».IntegraL ManifoLds 1 97 some neighborhood W of p into U. then x .a. if we had x . we can choose W to be connected. This makes it clear that f is continuous as a map into MI . for all p' E W. (f(p'» = a. This would mean that f(W) contained points of un countably many slices. x . . In other words. f( W) is contained in the single slice of U which contains p.
5. check the assertion about choosing ciently small. 3. (b) For the case of Coo bundles � and �' over a Coo manifold M. .. such that /'. (b) Supply the proof of the uniqueness part of the theorem. Let /'. U) around each point.. In the proof of Proposition 4. and �I k plane bundle such that E' c E. define a Coo subbundle. are n . . . (x. (a) In the proof of Theorem I. we say that �' is a subbundle of � if (i. If i : EI + E is the inclusion map. M. ' 1=1 f3. show that the functions C.1 98 PROBLEMS Chapter 6 1 1[ : E' + B a 1. 1 . . and IB : B + B the identity map. . x = L f3 ' ax. • . a Coo. then the functions 4. dt. . /'. (a) Let � = 1[ : E + B be an nplane bundle. . (a) In the proof of Proposition 2. of dimensions dl . . 0 f = f3. be integrable distributions on Suppose that for each P E M.} actually are Coo . 1. and show that a kdimensional distribution is Coo if and only ifit is a Coo subbundle. show that 6'1 suffi (b) Complete the proof of Proposition 2 by showing that if so that with a. IB) is a bundle map.. . = 2.. aj a xd" etc. 1 . • . Show that there is a coordinate system i s spanned by aj ax l . Show that a kdimensional distribution on M is just a subbundle of TM.
= j m Ll 11 . show af3 j iU (U. we can consequently assume that e = 1 . I) f3(0. I)) = . so that the equations are of the form aD! (I) = jj (l). jj (ul. This is connected with the classical technique of "introducing new independent variables". al j . by reducing x W + V. VI). by considering the distribution !. I) for some function f3 : [0. (b) Show that = x. al j ajj al i aft al j ' with the integrability conditions we nevertheless work in �m x �n ) rather than �m. 0) that f3 must satisfy the equation 7. f3(u. A similar technique will be very important in Chapter II. in JRm x JR" (with coordinates I. l) al j = V· af3 ( I . Prove Theorem I from Theorem 5. with the same initial condition. (c) Conclude that af3 (v. 7.) By shrinking W. since the equation depends on the "parameter" I E JRm). x) . (a) If we want D!(UI) = f3(u. defined by Notice that even when the jj do not depend on x. This problem outlines another method of proving Theorem I. (Show that both functions satisfy the same differential equation as functions of u. VI) = f3 (UV. 1) We know that we can solve such equations (we need Problem 55. the partial differential equations to ordinary equations along lines through the origin. I).integraL ManifoLds 1 99 6. One has to check that one 0 can be picked which works for all l E W. f3(U.
= XI . ax aa ' a. I ) al j and v · h ( VI.Y �2 in a neighborhood of 0 E differential equation = = � v(x. y). This problem is for those who know something about complex analysis. 2. y ) . (e) Define a(l) = f3 ( I . with any given initial condition 4>(zo) = Woo . y" X" y" then 1 = u + iv satisfies the CauchyRiemann equations Use Theorem I to prove that we can solve the equations au aXi au aYi av aYi av aXi i = 1 . Let I: <C x <C + <C be complex analytic. y. a ' (x. Noting that a(vl) = f3(V. a l (x. 8. f3(v. y)) = ay (or of any point </>'(z) = I(z. I).200 Chapter 6 (d) Use the integrability condition on 1 to show that af3 (v. I). show that a satisfies the desired equation. y . y)) = . If we denote the coordinate functions in <C x <C by ZI . a' (x. and conclude that the (in which f denotes the complex derivative) has a solution in a neighborhood of =0. a' (x. as functions of v. Use (c) to conclude that satisfy the same diff the two functions are equal.</>(z)) 20 E q.ay <C aa ' u(x. I)) erential equation. 2.
then T is also alternating. . An element T E :r k (V) is called alternating if Wwith a special kind of tensor field. " vk ! + T ( V I " " . . Vi. vk ! + T ( VI " .T(vJ • . Therefore. [This is not true in the special case of a vector space over a field where 1 + 1 = 0. • . vk ! = 0 + T ( Vj " " . . in a certain sense. Moreover. " Vj" " . . It is clear that I1 k (V) c :rk ( V) is a subspace of :r k (V). • . . Vi. . Uk ) = T(VI " . . . if f: V + W is a linear transformation. . Vi + Vj • • • . T(VI " " . . vk ! = 0 If T is alternating� then for any VI . if T is skewsymmetric.] We will denote by I1 k (V) the set of all alternating T E :r k ( V ) . Vi. . one is a multiple of the 201 .jl . .CHAPTER 7 DIFFERENTIAL FORMS morc algebraic prclimi. Of course. " vk ! + T(VI " " . then f* : :r k (W) + :r k (V) preserves these subspaces f* : I1 k (W) + I1 k (V). Vi" " . in this case. and the condition of being alternating is the stronger one. Vj • . The most familiar example of an alternating T is the determinant function det E T"(�n). Vi. skewsymmetry is the same as symmetry. . . . Let V be an /1dimensional vector space over JR. we have 0 = T(v) • . vk ! + 0. " . . " " Vj" . Notice that I1 I ( V ) = :r1 (V) = V* . . Uk ) = . . It is also convenient to set l1o(V) = :ro( V ) = JR. Vj . but one case is \vellknown. " Vi" " . so I1 I ( V) has dimension /1 . f\1osl discussions of the determinant begin by showing that of any two alternating nlinear functions on �n. ' " vk ! + T(vj" . Uk ). T is skewsymmetric: T(v) . ) Vj. . Vj" " . " Vi" " . Vi • • . . " Vj" . . Vi" " . the discussion of which requires some e turn OUf attention once more to tensor fields. . . . • . but we will be concerned if Vi = Vj (i #. . Uk . At the moment it is not clear what the dimension of I1 k (V) equals for k > I. Vi + Vj • . Vj" " . the most general alternating function. • . c. . " Vj " .onsidered as a function of the n rows of a matrix we shall soon see that this function is.naries.
vd) = ( pa ) . if these v's have indices running in some order other than 1. . . wd = (Wa(I) . ) is to rename things: V3 = WI ) V2 = Wz. vd. which is a special case of the definition to follow. ' " Va(k) ) . an element a E Sk is a function i r+ a (i). of course. Vp(k) = Wk) a · (p . . The simplest way to figure out something like a ' (V3 ) V2 . Then one proves dim 11"(IR") = I by actually constructing the nonzero function det (it follows. . . ' .. . . VI = W3 • . vd = ( Va(I ). Thus warned. (VI . . . Walk) ) = ( Vp(a(l)) ) .1 if a is odd. . . . . (p . \'\. . . . is the a ( l )" of the v's on the left side. . . . ' . . dim 11" (IR") :S I. If (v" .202 Chapter 7 other. (v" . T(va(l )" ' " Va(k) ). . . Vp(a(k)) ) since Wet = vp(a)· Thus a . the first element. . k. .here sgn a is + I if a is an even permutation and . . . Now for any T E :r k ( V ) we define the "alternation of T" I Alt T = k! aeSk L i. . (vJ . Vk ) is a k tuple (of any objects) we set a · (VI . . in other words. . . vk! = k! sgn a . . for example. vd ) = a · (wl . This definition has a builtin confusion. we compute by setting so that Vp(l) = W I • . . VI. . ueS" L sgn a . then the first element on the right is 1lot necessarily that v whose index is a ( 1 ) . . . Let Sk denote the set of all permutations of { l. . that dim 11" ( V ) = I if V is any n dimensional vector space). . . k}. . O n the right side.e. . . T 0 a. . . I Alt T(v" . . The construction of det is usually by a messy: explicit formula. . . .
The funny coefficient is not essential. then Alt(Alt(T)) = Alt(T). an element w/\ ry E I1 k +1 ( V ) . /\ ry w /\ (ry. + w. PROOF. by ( w /\ ry = k!/! Alt (w ® ry) . (I) If S E :r k (V) and T E :rl ( V) and Alt(S) = 0. •:. but it makes some things work out more nicely. for w E 11 k (V) and ry E 111 ( V ) . ry E 111 ( V). then Alt(S ® T) = Alt(T ® S) = o. (I) If T E :rk ( V). Finally. associativity of /\ is proved in the following way.) = W /\ ry. as we shall soon see. 2. It is clear that (I) /\ is bilinear: (w. (2) I f w E 11k( V ) . /\ ry + w. ) /\ ry = w. 203 PROPOSITION. (3) If w E I1 k ( V).Dijje"n liaI Forms I. THEOREM. In particular. (2) Alt(Alt (w ® ry) ® e ) = Alt(w ® ry ® e) = Alt(w ® Alt( ry ® e)). 7879 of Calculus on Manifolds) . the wedge product of w and ry. then Alt(T) E I1 k ( V ) . then Alt w = w. aw /\ ry = w /\ ary = a(w /\ ry) k + /)! (2) I* (w /\ ry) = I*w /\ f* ry. We now define. if k is odd then w /\ w = o. f\1oreover. then (w /\ 1]) /\ e = w /\ (ry /\ e) = (k + 1 + 111 ) ! Alt(w ® ry ® e). (3) If T E :rk ( V). + w /\ ry. e E I1m ( v ) . + ry. Left to the reader (or see pp. it is easy to see that (3) /\ is "anticommutative": w /\ ry = ( _ I) klry /\ w. k ! l!m! .
. Then L sgn a . • . (vl . Then L sgn a . :. k + I fixed. " " Va(k+l) ' Now let G C Sk+l consist of all a which leave k + I. the sum over each being o. . (v l . the other equality is proved similarly. .Alt(w ® ry ® e). Vk+/» ueuuG [ a'eSk L sgn a ' · S(va'(1) . . Vk+I» ) . a contradiction. (3) We have a'eG L sgn a' . Suppose now that ao <t G. ( Vb . .ry ® e) = Alt( ry ® e) .I E G. so (I) implies thaI 0 = Alt(w ® [Alt( ry ® e) . (ao ' (VI . • . . ( S ® T) . . T(Vk+l > " " Vk+I ) = sgn ao ' ''''le have just gjwwn that this is 0 (since ao . . (a . . . (S ® T)(a' . . . T(Va(k+l) . . (w /\ ry ) /\ e = The othel' equality is proved similarly. ."lay: breaking Sk+! up into disjoint subsets.). Let aoG = {aoa' : a' E G). then a = aoa' for some a' E G.ry ® eJ ) = AIt(w ® Alt( ry ® e» . S ( Va(I). Va'(k) ] .. so ao = a(a. " " Va(k) ' T(Va(k+I). . .Alt(1] ® e) = 0. \!\le can then continue in this . = ry ® (k + /)!m ! k!/ ! . . . Vk+l) = = aeSk+l aeSk+l (1) We have L sgn a . Va(k+l) aeG = = o. Vk+I » L sgn a . S(Va(I ) . . . for if a E G n aoG. ( S ® T)(a . . . . (2) Clearly Alt(Alt(1] ® e) . . . . .204 Chapter 7 PROOF (k + /)! AIt(S ® T)(Vl . Va(k) . . (k + l + m ) ! Alt((w ) ® e) /\ ry (k + / ) ! m ! (k + 1 + m)! (k + /)! Alt(w ® e). . Notice that G n aoG = 0. The relation Alt(T ® S) = 0 is proved similarly. by (*) . . Vk +l) is just some other (k + I )tuple of vectors). .
(</>I I\ . . " . we would not have Alt(Al t T) = Alt T. and the first equation in the proof of (2) would fail. . ."2 k ( V). then 1\ could be defined by w 1\ ry = 1 k!l! Alt(w @ ry). .. . I! aeS" L sgn a · (</>I @ · · @ </>n ) o a · . If V I .. the factor 1/ k! in the definition of Alt is essentialwithout it. VII is a basis of V. because each term in the sum Alt( w @ ry)(VI . . . · I\ </>n )(VI . ellen over a field a ffinite characteristic. The set of all 1 ::: i) < " is a basis for !. . . ( I + . > 11.. . iJ. 1\ </>n = = In particular. . !. but without the factor 1/ k!. . . and 4>1 .) A basis for !.. . 4>il ® ' " . . ."2k ( V) = {OJ for k PROOF. which therefore has dimension () 11 k : . . . 1\ </>n = det. L ail ."2 k ( V) can now be described.) ® 4JiJ. (So i f V I . Vk+l) occurs k ! / ! times (since w and ry are alternating). . . then </> 1 1\ .] The factor (k + I)!/ k ! / ! has been inserted into the definition of 1\ for the follO\ving rcason. < ik ::: n = 11! k! (I1 . 4>n is the dual basis. .k) ! ' (In particular. . we can write W = iJ ."2 k ( V) C :r k ( V) . This makes sense. @ </>n ) I!" . Vn ) = I . . . 3. If w E !. . · + I)! Alt(</>I @ " .Dijjcrenliul j'lmlH 205 Notice that (2) just states that /\ is associative even if we had omitted the factor (k + /)!/ k! I! in the definition. . THEOREM. then </>1 1\ " . Vn i s the standard basis for IRn . [If we had defined Alt just like Alt . . and 1/ k! I! can be interpreted as meaning that these k ! / ! terms are replaced by just one.iJ. On the other hand.
/\ 4>h for }. . . Uk • . . Wk E !:) ' ( V). . . . ® 4>ik ) · Each Alt(4)i. ® . . 4>k. .ik 4>iI !\ · · · /\ <PiJ. Wk are linearly independent. . Wk are linearly inde pendent if and only if w. < }k span !:) ( V). . . . if then uniquely expressible as k (i" . /\ · · · /\ 4>ik .·. . it is convenient to let L al 4>l . . /\ Wk #. ) PROOF. /\ . . . . . /\ Wk is a basis element of !:) k ( V) . . . ik!. . .206 Chapter 7 So W = Alt(w) = iI . x k minor of (:�). . . . . vk! f+ determinant of a k One more simple theorem is in ordel� before we proceed to apply our con struction to manifolds. . . . . COROLLARY. . . If WI " 4. /\ . .o. If 0= i l < " '<ik L ai! .h L ai. < . < }k. On the other hand. then applying both sides to (Vii " ' " Vik ) gives ail . . Then every element of !:) ( V ) is To abbreviate formulas. . ·. 4>n satisfy 4>i = Wi for I :s i :s k .ik Alt(4)i. . so the elements 4>h /\ . . .:.. . I j denote a typical "multiindex" Notice that Theorem 3 implies that every W E S'"2 k ( �n ) is a linear combination of the functions (v" .i}. ® . . Vn • . . . . . then W' . . < .. and let 4>1 denote 4>i . /\ 4>h for some k }. . . = o. ® 4>ik ) is either 0 or = ± ( I / k!) 4>h /\ . . Then w. there is a basis V I " of V such that the dual basis vectors 4>" . so it is not o. If W" . . .
. > O. Remember that covariant tensors actually map contravariantly: If /: M + N is Coo. . . we obtain a new bundle !.) /\ ry = WI /\ I/ + W. and c = ry( et . We can also define WI + w. sections of n k ( TM). which are just alternating covariant tensor fields of order k."2 k +/ (�) by (w /\ = /\ E In particular.talnVl). . . j=1 i PROOF. " (Unl. . . . . Let 207 Then VI . let W E !. . all forms will be understood to be Coo forms unless the contrary is explicitly stated. j=1 j=1 Then clearly ry E !.en) = w(Vt. vn ) : 6. . . then we can define a section W /\ ry of !. so ry = c · det for some c E JR. unique orientation [vt . .B is a vector bundle.1 (p» . . . Vn be a basis for V. THEOREM."2 k (�) with is a function with E for each E B. . . v ] = /L IL for V such that n if and only if W(VI. The following properties of kforms are obvious from the corresponding properties for !. and let n Wi = L CXjiVj = l . we are ready to apply it to vector bundles."2k (1[1 (p» !."2 k (1[. Define ry E :r n ( JR ) by n . ."2 k ( V): (WI + w."2 1 (�). · . .DifJemiliaL Forms 5."2n ( v) . ."2 k (�) by replacing each fibre 1[. " . . !. n. ann») = W(f>lIVl. . are called kforms on M. we can speak of Coo forms. COROLLARY."2 n ( JRn ). . A I form is just a covariant vector field.1 A section W of !. . . /\ ry W /\ (ryl + ry.) = w /\ ryl + W /\ ry. If � = J[ : E ?."2 n ( v) .ant!. ."2 k +1 (1[ 1 (p» . /w /\ ry = w /\ /ry = /(w /\ ry) w /\ ry = ( .ill E !. •. . . . p ry)(p) w(p) ry(p) . If ry is a section of !. then there is a ry ( (u l I . and W is a kform on N. If V is ndimensional and 0 #. . Vn) With our new algebraic construction at hand. and w /\ ry. ."2 k (TM) can obviously be made into a Coo vector bundle. . . Since !. then /* w is a kfol'ln on M.I/Iry /\ w /*(w /\ ry) = /*w /\ /*ry. (p) w(p) !.
THEOREM. V) a coordinate system around E N. . ay a � a(i o f) a 1 . J* a�n IJ n a = dy'(q) /\ .k(p) (i1 < . . If U) i s a coordinate system around P E M. o f» ) dx' dx". /\ dyn (q) (f* a�' Ip .x�f» ) dx' /\ .. /\ dX. PROOF. . .. . then the dx. d et � A · · · f\. . < ik) are a basis for !. ay i= l  ( a(y.. . L The problem of finding the relationship between the WI and the functions W'I when W = L WI dx I = L W'I dyl I I w Of. and (y. so the dx.. ..(p) . dy" ) = (g o f) . . L. U) is a coordinate system. .' (p) /\ . Wl dxl. . It suffices to show that f: (x. . . Now. then q = f(p) f*(gl/Y' /\ . . C(Y. /\ dyn (q) (L a(ix l f) (p) �l 1 . If (x. . f*(dy' /\ .. f*(dy' /\ . /\ dy")(p) ( a�' Ip .. /\ dxiJ.. 1=1 q q ) . ik).. /\ dxn . .· by dx l for the multiindex I .. a. •:. f\. . w = = (il. . . 0 by Theorem 5 .. 1 M + N is a Coo function between l1manifolds. . but we will do one special case here. a�n IJ = dy' (q) /\ .208 Chapter 7 is left to the reader (Problem 16). 7. Thus every kform can be \\'Titten uniquely as W= if we denote dX i l /\ . by Problem 41. . (p) are a basis for Mp '. . . . . /\ dyn ) = det .1k(p).
THEOREM. . Vn E Mp . Vn] = /Lp. . If (x. vn] = /Lp. . if there is a nowhere zero l1�form w on an nmanifold M.DijlerentiaL F orms 209 8. = g . . .. h ( ) [This corollary shows that nforms are the geometric objects corresponding to the "even scalar densities" defined in Problem 41 0.. Vn) :0:: 0. U»). . U) choose an nform Wu on U such that f VI > . so that /L = {/Lp} is an orientation of �. . Vn» Now let O if and only if [VI > . . . By Theorem 213 and 21 5. the nonzero w(p) E I1 n (. . then a nowh". W = L 4>uwu. U) and (y. and a partition of unity {4>u } subordinate to e. . Moreover. Vn E Mp satisfy (4)u Wu )(P)(VI .. Ue(9 for every Then w is a Coo nform. . . and strict inequality holds f at least one U. . The converse also holds: 9. zero section w of I1 n (�) has a special significance: For each p E B. then there is an nform on M which is nowhere o. . if VI . COROLLARY. If a Coo manifold M is orientable.I (p) by Corollary 6.. /\ dyn = h dxl /\ . then eadl p.:.I (p » determines an orientation /Lp of . or .o . V) are two coordinate systems on M and g dyl /\ . . . P E U we have or WU ( VI > . then M is orientable (i.] If � = " : E + B is an nplane bundle. the bundle TM is orientable). . . we can choose a cover e of M by a col lection ofcoordinate systems {(x. In particular. /\ dxn. . Let /L be an orientation of M.e. det a/ . w PROOF. .. [VI . ax) Apply the theorem with f = identity map . . Thus w(p) #. For each ( x. It is easy to see that the collection of orientations {/Lp} satisf the "compatability condition" set forth y in Chapter 3. then PROOF. . . . •:. .
(3) d(dw) = o. f W'l WI. by dw = L dwl dx l I " aW = LL1 l dx· l\ dx I . We begin by finding some properties of (still defined with respect to this particular coordinate system). then r:2 k (n is trivial if and only if � is orientable. which in a coordinate system U) is given by dw with the The second method is a lot sneakier.210 Chapter 7 It turns out that this definition does not depend on the coordinate system. PROPOSITION.) = dWI W. W = L W'l dx l I Notice that the bundle r:2"(TM) is Idimensional. [Generally. then r:2"(TM) has a nowhere 0 section. provided that the base space B is "paracompact" (every open cover has a locallyfinite refinement). WI is a kform. j= ax} If W is a kform W = L Wl dx l . a Oform on M will just mean a function on M (and 1\ will just mean For every Oform we have the Iform (recall that ( X) XU»). which implies that it is trivial. We have shown that if M is orientable.] Just as r:20( V) has been introduced as another name for 1R. and we can define a (k + I )form dw. The first way is to use a bruteforce computation: comparing the coefficients in the expression f W f . I a= ax· f (x. df df = " af df = L1 � dxj . + ( I /wl dw. if the bundle r:2"(TM) is trivial. (I) (2) If d(wl + w. then d(wl w. This can be proved in several ways..) = dWI + dw. then it certainly has a nowhere 0 section. d' = o. w). the differential of w. if � is a kplane bundle.. so M is orientable. Conversely. 1 0 . Briefly. 1\ 1\  1\ . I then each dWI is a Iform. of course.
. == == == == == == (I) is clear. (2) d'(wl /\ w. = fg dx l /\ dxJ and d(wl /\ w. To prove (2) we fIrst note that because o f (I) it suffices to consider only In this sum. (3) It clearly suffices to consider only kforms of the form w = fdxl . (3) d'(d'f) = o.dx· /\ dxl L.. + (I) k w I /\ d'w. (I) d'(wl + W2 ) = d'w l + d'w. so n ( n a'f d(dw) L L aX"ax· dxP /\ dx· /\ dx l ) .. Then n af dw ". \!\le next note that these properties characterize I I . Then WI /\ W.. PROPOSITION. WI = fdx l w. + (llwI /\ dw. •:. = . (4) d'f (the old) dJ: Then d' = d on U. Suppose for all k. the terms a=1 /3=1 a=l Q  and cancel in pairs . .. ax.Di/ ji<rential F orms 211 PROOF.) = d'wl /\ w. +(Ilfdxl /\ dg /\ dxJ dWI /\ W. =gdxJ. and satisfies d on U. d' takes kforms on U to (k + I )forms on U.) dUg) /\ dx l /\ dxJ g df /\ dx l /\ dxJ + f dg /\ dx l /\ dxJ dWI /\ w.
U) we have a unique du defined. and p E M. . There is a unique operator d from the kforms on M to the (k + I)forms on M. dw(p) = du(w [ U)(p) ) . /\ . •:. pick any U with P E U and define PROOF and agreeing with the old d on functions. satisf ying = 0 . . Given the form w.d'X i. . • The third way of proving that the definition of d does not depend on the coordinate gystem is to give an invariant definition.0. d(wj + w. . . /\ d'(d'xi. . = df /\ dxl + f /\ d'(dxl) dxl = dX il /\ . COROLLARY. /\ w. /\ w. ) /\ d'xiz /\ . .) = dw. ' /\ d'Xik . d(w. where by (4) . /\ d'Xik ) d'(d'xi. Assuming it f k .I /w. d' = 0. . /\ dw. + dw. /\ d'Xik ) = 1 2 .212 Chapter 7 It is clearly enough to show that d'w = dw when w = PROOF. .) = dw. f dx1 Now So it suffices to show that d'(dxl) d'(j dxl) = d'j /\ dxl + f /\ d'(dxl) by (4). ' /\ dXik = d 'xi. by (2) by (3) and the inductive hypothesis . + (. /\ . /\ . /\ d 'Xik = 0. For each coordinate system (x. We will uSe induction on k .I we have or d'(dxl) = d'(d'xi. by (2). for all k.
. . X}.. I f i s a kform o n M. . . .X. . . This (k + I )form agrees with dw as defined previously. /\ dx . . .=1 + L (.. .) . ." say) ...u . X" l::.YI · X [X." ' (X" . io<j a brief inspection then shows that }:. k Theorem 4�2 shows that there is a unique covariant tensor field dw satisf y ing (*) .X" " " X. . . To compute dw in a coordinate system (x. X" .. Y] I[X. so that it is a (k + I)form. . jY] = I[X.. it is easily seen that 2:2 becomes I }:.. .J)w(X. Moreove. . + I }:. . Xk+. .. .. . . . X. becomes IX.L(lin+j (Xjl)w(X.X". . Y] + XI .li+j w([x. if Xio is and using the formulas [IX.i<j. . . ..Differential F orms 213 1 3. + }:.. . ) Xk+1 . we might as well assume w = I dx' /\ . Xj ]. . + }:." .. by renumbering. + = L(li+'" (x.. is clearly linear Coo Junctions y.. Xk+. Y.. In fact. it is actually linear over lhe replaced by then }:. . Y] .. It is easy to check that dw is alternating.<ill . .. X}. Xk+') to }:. . X..::k+1 (= }:. . f\1oreover.) where over indicates that it is omitted. THEOREM. The operator which takes over � . . . .. PROOF. .Xk+')." ". . then there i s a unique (k + I)form dw on M such that for every set of vector fields we have w XI) . becomes I}:. . U) it clearly suffices to compute d(1 dx l ). + }:.
. . . j= � /\ /\ /\ . . we have dw = It is clear from H that dw(a/ax· ' . . . . Xi p. a� dw = L(I/ ax dx' j>k af = ". Since the a's are increasing. dw(p . Q'k +d is a permutation of ( I . . . . . •:. . iii . . But we can hardly hope f anything or does not depend on the values better. and our first use of Theorem 42. This term is what makes the operator dw(X" " " Xk+'). /\ /\ . . . . k ). . Xk+. dx . as for any form.)(p) w p . . . . directly. but first find where are vector fields extending and then evaluate this function at By some sort of magic. .214 Chapter 7 dw. X" . Vk+') Vi . . /\ /\ . . We do not find )(v" . this turns out to be independent of the extensions This may not seem to be much of an improvement over using a coordinate system and checking that the definition is independent of the coordinate system. . /\ This is our first real example of an invariant definition of an important tensor. Xk+l . although of except at it does depend on the values of at points other than this must enter into our f ormula somehow. . One other feature of our definition is common to most invariant definitions of tensorsthe presence of a term involving brackets ofvarious vector fields. p. After all. . .. ax} dxj dx' L. in which case so which is just the old definition . .. /\ dxk dxj dxk dxk . Xi dw(X" .a/ax·k+ ' ) = 0 unless some (0') . j>k " aj = L1 ax} dx}. For . this happens only if j > k. .
.It dxP.Xl belong to !.q. .q a = k + I . J(!. . . J(!. Define the ring \1 (M) to be the direct sum of the rings of Iforms on M. Y) X(w(Y» .) if w J(!. and that W J(!. 1 4 . but it disappears in computations in a coordinate system. . J(!. . . Theorem 13 gives the following formula. PROPOSITION (THE FROBENlUS INTEGRABILITY THEOREM. SECOND VERSION). . Locally. which by Corol lary 4 implies that dx1 (q).k independent Iforms Wk+ 1. A distribution !.n. is a kdimensional distribution on M. This enables us to state a second version of Theorem 65 (The Frobenius Inte grability Theorem) in terms of differential forms. By continuity. dx k (q) are linearly independent !. ."L.lt(q)dxP (q) restricted to !.» c J(!. the same is true for q sufficiently close t o p . . P=l We can therefore let k w" = dx" . W(XI . .) is an ideal in the ring \1(M)].p. Then d(J(!.X ) It is clear that WI + w. In fact. dxl (p) dxk (p) is nonzero on !. around any point M we can choose a coordinate system (x.Y(w(X» . .) is generated by ..Di/Jemlliai F orms 215 linear over the Coo functions.) w w whenever XI.).) if Wl . /\ in . . the ideal J(!.). In the particular case where is a Iform. Y]) I = E ry /\ E E n P E . there are Coo functions I such that t k dx"(q) = "L. wn . U) so that .) [thus.. . If !. . . then c \1 (M) will denote the subring generated by the set of all forms with the property that (if has degree I) I =O E w I dw(X. for ali i. P= l /\ .w([X. • .W' J(!. There f ore. on M is integrable if and only if .
) = ej�/XiU). wn which span Mq for each q wk+ 1. . k. .) as dw" = L e. Xk span 1'>. /\ wp . we have following. VVe therefore h ave the E .. if I'>.Xj) = Xi (w"(Xj» .. . . Xn be the vector fields with . so we can write the condition d(J(I'>. Xj]) = 0 [Xi .Xj(w"(Xi» . • • Then with Xl .. :+ Now P=l = L OJ /\ wj If a > i<j for certain forms OJ .. . But each if and only if each belongs to I'>. Xj) = 0 w"([Xi .. j :S k and a > k.» c J(I'>. 0 = dw" (Xi. we can always For I :S i. . the forms dXk+1. U) such that the slices l l {q U : Xk+ (q) = J + . j = I. io ::. Xj]) = W"([Xi. . (i.w" ([X" Xj]). k are distinct.. is integrable). wi (Xj) = oj. Locally we can choose Iforms w I .). • • k. write dw"(Xi .Xj. .wn must be linear combinations of them. the first two terms on the right vanish. Xj. dw"(Xi . . . ) = L(ej /\wj )(Xi. .. and io. .). Xj] dw"(Xi . So if and only if o. Xn (q) = a n } are integral submanifolds of 1'>. . . . So I'>.Xj] = L ctxp i.. ... . . wn generate J (1'>. . .). dxn are a basis for J(I'>. ..e. is integrable ifand only ifthere are functions ct k [Xi .Xj) = 0 dw" Notice that since the w i /\w j (i < j) span fl2 (Mq) for each q. . so wk + 1 . Let Xl .216 Chapter 7 such that PROOF. . while each if and only if E J(I'>. P>k Once we have introduced a coordinate system (x.
. . /\ ' . Problem 1 8 gives another invariant definition of using induction on the degree of which is much simpler. /\ dXik' ») /\ f*dxik by the inductive hyposthesis f*(dg /\ dxi. the cases k > I will hardly ever be used (a very significant exception occurs in the last chapter of Volume V). U) be a coordinate system around f(p). then there are Iforms (ex. . Assuming the formula for k . .wn e. •. /\ . then j*(dw) d(f*w). . .. f* g is to be interpreted as g o f). /\ . . /\ dXik' j*dXik) d(J*(g dxi. /\ dxik . If E = = = = j*(dg)(X) = dg(f* X) [f* X](g) = X(g 0 f) d(g o f)(X) (and. p dw· = L e. PROOF. 0 0 =0 = = = = . If are linearly independent If orms in a neighborhood of E M. 1 6. . . of course. /\ dXik' ») /\ j*dXik + since dj* dX ik d d(Xik f) j*(d(g dxi. The reader may reflect on the difficulties which would be involved in using the definition of Theorem 13 to prove the f ollowing important property of w. . let (x. d: f: M + N is Coo and w is a kform on N.. /\w P if and only if there are functions f . For p M. /\ . tracing through some defini tions. /\ dXik' /\ dXik ) j*(dw) : = ) = = /\ We will use induction on k. /\ dXik' ) /\ f*dxik f*(dg /\ dxi. For k = 0 we have. w' = L ft dgP • P p > 1 5 . /\ ..g P (ex. {3 k) with . . we have d(f*w) d((f*gdxi. COROLLARY.. dw. PROPOSITION. . {3 > k) with Although Theorem 13 warms the heart of many an invariant lover.I.orms DifJerential F 217 wk+1 . We can assume w g dxiJ /\ .
ay C The necessary condition. = Wj . af ax. If is a 2form on � 3 . = 0 is a necessary condition for solving = If is a If orm d w dry w d' dw d2 dw = w W d(Pdx + Qdy + Rdz) aR _ aQ A ay az �: . is aA + aB + ac = 0.ap = . a differential was then called "exact" ifit actually was the differential of something. w W L w. then the condition dw = 0.�� = B aQ .. i.=1 n One property of qualifies. The term "dosed" is based on an analogy with chains. For 2forms the situation is more complicated. A form is called closed if = 0 and exact if = for some form YJ.e. ax ay az if and only if = then = =    . i. (The terminology "exact" is classical diff erential f orms used to be called simply "differentials".218 Chapter 7 Now we know from Theorem 61 that these conditions are also sufficient.) Since = 0. a. every exact form is dosed. dxi .e. which will be discussed in the next chapter. aWi aWj axj axi is necessary for solving w = dj. There is another set of terminology for stating the same thing. w dry. as a basic theorem of differential geometry. In other words. = . dw 0.. however. The relation = 0 is just an elegant way of stating that mixed partial derivatives are equal. by the criterion of the previous chapter.
the situation is very different if w is defined only on IR' . however. Then e. we are dealing with a rather strange collection o f partial differential equations (carefully selected so that we can get integrability conditions). we can define a different function e. namely a(x. . this result is true only locally. = e in the region A. Like our results about solutions to differential equations. e) : IR ' is the inverse of the map Coo. . and e. agree on their common .o. + O} x (0. then Jxo defined in Chapter 2.a sin b).b) r+ {L {I" : r > fY41Y g(x.{O}. = e+21f in the region A . L (a. Consequently de and de. whose derivative at (a. It turns out that these necessary conditions are also sufficient: if w 1S dosed.DijJerential Forms 219 I n general. 00) x {O}. in fact. The reasons f restricting ourselves to local results are now or somewhat different. b) has determinant equal to a #. Recall that if L c IR ' is [0. with aj ay ag = ax We know how to find a function a on all of IR' with w = da. Consider the case ora closed I form w on �2: w= j dx + g dy. By deleting a different ray L. then it is exact. is (I". 21f) (a cos b.y) = r j(t. t) dl. yo ) dt + On the other hand.
) = 0 on lR' . if w J.L d(de.e) = 0 on lR' . we can define lRn lRn H(p. since lR. dw = 0 [the two relations d(de) = 0 on lR' . A manifold M is callcd (smoothly) contractible to a point Po E M if there is a Coo function H: M [0. but this is an abuse of notation. but not exact.L. (It is still exact i n a neighborhood of any point of lR' . 1] + by l'vlore generally.I) = lp .O) = Po H: x for p E M. Nevertheless. w de only on lR ' L.{O}. that determines whether or not a closed form is necessarily exact. which implies that afjax = aelax and afjay = ao/ay and hence J = e+ constant on lR' L. then = =d + so d(f . clearly imply that this i s so] .L. �" is smoothly contractible to E �n . This example shows that it is the shape of the region. rather than its size. U C lR " i s contractible t o Po E U if U h a s t h e property that H(p. so that together they define a If orm w on lR' . In fact. . A computation (Problem 20) shows that w x = x'yy' dx + x' + y' dy.L . 1] + M x such that For example. The Iform w is usually denoted by de. which is impossible. S o w i s closed. + dJ = de on lR ' . I) = p H(p.{O}.220 Chapter 7 domain. 0 [0.) Clearly w is also not exact in any small region containing O. w is not dJ for any C' function J : lR'{O} Indeed.
I] (for any manifold M). For I E [0. If we think of [0. then .po) E U for ° :s I :s starshaped with respect to po). I] with dw = 0. I] as representing time.{o} is 1I0t contractible to a point. many other regions are also contractible to a point.{o}. but we will not be in a position to prove this until the next chapter.) The trick in proving our result is to analyze M x [0. this result and our investigation of the form de prove the intuitively obvious fact that JR' . and pay hardly any attention at all to H. at time 1 this is just the identity map. (By the way.Dijje. the same result holds for JR" . I] we define i. then every closed form on M is exact. I] by We claim that if w is a form on M x [0."tial F orms P E U implies po + I (p . I) of M into itself. then for each time I we have a map p r+ H (p. and at time ° it is the constant map. We will show that if M is smoothly contractible to a point. 221 1 (such a region U is called Of course. : M + M x [0..
aWti dx'· dl + L. (x.. I) . Now for W = 2::7= 1 Wi dx i + f dl we have � � af dw [terms not involving dl] . We will denote nience. I) dl. I]. dl. Consider first a I form on M x [0.222 Chapter 7 we will see later (and you may try to convince yourself right now) that the theorem follows trivially from this. We will begin by working in a coordi nate system on M x [0. the projection IT on the second coordinate).* (L Wi dXi + fdl ) = L Wi ( · . axl dx'. (namely. . It is easy to check (or should be) that xi 0 ITM by xi .xn O J[M . i==] i=] Wi t . = so (I) Wi (p. t) is a coordinate system on U x [0. I].Wi (p. . . l)dxi .L.I)dl ) dxi .a) denotes the function p wi (p.(p.a).O)dxi = L] (In0 1 aafi (P.O) = 1r0 ' aaw · (p.. . i=1 a i=l So dw 0 implies that f+ =  where /\ =' /\ Consequently.a)dxi. then W I]... I] (Xl 0 1fM . while J[M is the projection on M.l)dl f ' af (p..l)dl. . and if U) is a coordinate system on M..Ll w. for Conve n n i. . There is an obvious function I on M x [0. = i= x i=] i lfwe define g : M + IR by g(p) = l f(p. = r 10 axi n n n L Wi (p.
then any where w] (v] + v. of two subspaces. Vk) = w ° ifsome V = w] + i E ker HM. . (V ] + v. = j In general. I]) (p. is really independent of the coordinate system.) w.. there is then a unique j with w. we write the I form w on M x [0. it is easy to see (Problem 22) that we can write w dt. ) = = w(v] ) w(v. EEl kerHM •..I)form with the (dt /\ ry) . t)dt. . Notice that for the tangent space of M x [0. the function j.11 ( V) can be written EEl V. for a kform w. Applying this to the decomposition (*).Diff erential Forms then (2) 223 ag t aj aXi (p) = 10 axi (p.t) = kerH. [0. g Equations (I) and (2) show that Now although we seem to be using a coordinate system. uniquely as where w] (v]. I] as w] + w. I] M w If a vector space V is a direct sum V = V] E !. I] we have (M x [0. . and ry is a (k . and hence also. .).
n . /\ d. so we just have to consider two cases. .i.. /\ dieik f diel . O)]dx l (p) i.224 Chapter 7 Iw on M as follows: Iw(p)(v" . 1 7. Define a (k .io*w that holds even when dw #. /\ . . . (2) w = / dl /\ (r�i. Actually. it is easier to find a f ormula for i. this proves the result in this case. .*w .)dl. *w io*w 0. . * w io*w d(Iw) if dw 0. .*w .l)(i.a=] axa dl /\ d'X· /\ diel ) (P) � a! L = = = = = it is easy to see that = = =  + = = = d(Iw) d ([ /(p.l) dl) dx l Xl =] o (p. . I ) dl) dxl (p) [f(p. we can just as well work in a coordinate system (X l . i. I) dl dx· /\ d . . Now I(dw)(p) 1(. I(dw)(p) ([ � (P. Since Iw 0. t ). ) PROOF. (i{' / ) Clearly I(dw) d(Iw) 0 •:. /\ .*w .io*w d(Iw) I(dw). j. Since Iw is already invariantly defined. THEOREM.io*w d(Iw).*Vk_..*v" . Then af dw at dl /\ dx l . (I) w f d'Xi. = = and + = . . at a!. We claim that dw ° implies that i. Then i. (Consequently.f(p.. . *w(p) .Vk_') = 10' ry(p. I] we have i. I) . The operator I is clearly linear.�ik' f dl /\ d'X l .0. analogous property.I)form = = x = = + = . . For any kform w on M [0.io*w(p).
while d' = 0 is called the Poincare Lemma by some (I don't even know whether Poincare had anything to do with it. Corollary 18 is called the Poincare Lemma by most geometers. it has always been something of a mystery to me just why d turns out to be so important. 92 ( 1 959).DijJerential Forms 225 1 8. COROLLARY. •:. Trans. If M is smoothly contractible to a point po E M.io* (H*w) by the Theorem .O) = Po Thus H 0 il : M + M H 0 io : M + M for all P E M. Amer. f evelY form w on U. atwal Operations An answer to this question is provided by a theorem of Palais. then every closed f orm w on M is exact. So But so w = (H 0 il )*(w) = i l * (H*w) 0 = (H 0 io)*(w) = io* (H*w). we can find (Problem 23) an explicit formula for J(H*w). or we can solve the system of partial differeIltial equations w = dry explicitly in terms of integrals. We are given H : M x [0. where we have an explicit formula for H. is the identity is the constant map po. I ) = p H(p.141 .) In the case of a starshaped open subset U of IRn . Math. such that the following diagram w . Even though the Poincare Lemma and its converse fit very nicely into our pattern for basic theorems about differential geometry. = d(l(H*w» .0 = il * (H*w) . Since the new form is given by an integral. 1 25. I] + M with H(p. There are classical theorems about vector fields in 1R 3 which can be derived "'om the Poincare Lemma and its converse (Problem 27). PROOF. Soc. d(H*w) = H* (dw) = 0. and originally d was introduced in order to obtain a uniform generalization of all these results. X on Difjermlial F orms. Suppose we have any operator D from kforms to Iforms.
In this case. then can only be some multiple of d. = o. Dl D D DD D lD . these excep tional cases are the following. since d ' makes the above diagram commute!) There is only one other case where a nonzero existswhen k is the dimension of M and I = o. d ' = 0.226 Chapter 7 commutes for every Coo map I: M + N [it actually suffices to assume that the diagram commutes only for diffeomorphisms 11 . with few exceptions. If k = I. can be a multiple of "integration". which we discuss in the next chapter. Roughly. then can be a multiple of the identity map. If I = k + I. but nothing else. kforms on M � kforms on N Iforms on M � If orms on M Palais' theorem says that. (As a corollary.
. Vk_1). Vk . .flle a rk ) and a rk + I ) < a rk + 2) < " .. Let S' C Sk+1 be the subgroup of all a which leave both sets { I . (c) Show that for W1 V . </>n. . .) = (v . .J w). ..l lw1 /\ (v .Difforenlial F orms 227 PROBLEMS 1.). < a rk + /) . E 111 (V) we have (Use (b) and linearity of everything. < . . 4.J W 1 ) /\ w. V1 . ... ry 2. . Show a • p . .J W)(V1.1 (V) by (v .. Vn is a basis of V with dual basis </>1 . k +/} invariant. w ® ry(va(l) .. Va(k+/) . . .J w. + ( . and w /\ ry = ( _ I ) klry /\ w. .J w E I1k. ... .) . Vk) = ap • (V1 . . . .any i..J w) = . For v E V and w E 11k (V). (b) Show that if V1. V . .. . . E 11 k (V) and w.) (a) Show that for any cross section K we have w /\ ry(v]... Vk +/) = aEK L sgn a . . Let Alt be Alt without the factor 1 / k!.J (W1 /\ w.J ( w . .. . Show that the set of all shuffle permutations is a cross section of S'. (a) Show that This is sometime also called the inner product and the notation ivw is also used. j #. k} and {k + I. .J ( v ..w . then if j = i• . .. · . A cross seclion of S' is a subset K c Sk+1 containing exactly one element from each left coset of S' ..) (c) A permutation a E Sk +1 is called a s"u permutalion if a ( l ) < a (2) < .. (b) Show from this definition that w /\ ry is alternating. . . Vk). and define W 7\ ry = Alt(w ® ry) . .1 ) = w (v. " . that 7\ is nol associative. E 11 1 ( V) and e E 11' ( V) . (Try w. . (Proving associativity is quite messy. . . This definition may be used even in a field of finite characteristic. we define the contraction v . . . Show that if we define then 3. . (V1 . · . .
Reformulate parts (a)(d) in terms of this /\ product. /\ l + Wl /\ w. (c) If W and are decomposable. n n.1 1 (V) . . Vk !) = [(Vl .) if and only if W2 Wj l/\ for some (d) I f Wi are decomposable. .I)form X.J w(p). Ann(wl ) + Ann(w. In this case.. n .1k (V*). Show that if Wl is a kform. /\ </>k for some </>i E V* = !. l (d) Formula (c) can be used to give a definition of by induction on k 1 (which works for vector spaces over any field): If is defined for forms of degree adding up to < k I. (a) Show that dim Ann(w ) ::.4 = YJ W2 YJ.J w.J w)(p) = X(p)..J w.) = Ann(wl /\ w.1' (V) is decomposable. . + n + E ::. /\ W l . is skewsymmetric (it is only necessary to check that interchanging V l and v. (f) If X is a vector field on M and W a kform on M we define a (k .228 Chapter 7 + W /\W. (f) Since Vi E V can be regarded as elements of V**. Hin!: Look at W /\ w .](v" .. i = 1 . we define + + . Show that functions fi. = ( _ I )k! w. .1 l (V) is decomposable.. we define the annihilator of W to be Ann(w) = {</> E V* : </> /\ W = OJ. Show that with this definition WI /\ W. and that equality holds if and only if W is decomposable. . /\ Vk E !.).. then Ann(wl ) Ann(w.J Wl ) /\ w. then every W E !. .J(Wl /\ w. then W = (</>1 /\ </>. 5. . 7. For any W E !. we can consider V /\ . An element W !. changes the sign of the right side). + (I/Wl /\ (X. I : M IR form a coordinate system in a n neighborhood of p E M ifand only if dfi /\ . /\ dl (p) #.)](v" .) = {OJ if and only if Wl /\ w.(v" . (a) If dim V 3.1 k ( V) is called decomposable if W = </>1 /\ . . . . . #. Vk !) .) + (</>3 /\ </>4) is not decomposable.o. which is unique up to a multiplicative constant.. then Ann(w l ) C Alln (w.1 k (V).) = (X. Vk !) + (.).. (e) If V has dimension then any W E !. k.I/[Wl /\ (Vl . . .o. (b) If </>i. then X . (b) Every subspace of V* is Ann(w) for some decomposable w. (e) Prove by induction that /\ is bilinear and that W l /\ W. are independent. . n 6.J w by (X. .J Wl) /\ w.
.. . where w ' does not involve 1/11 or 1/12. < hp and K = k l < ' . and that the (r + I)f wedge product is O.. /\ Wp = L B H VH H Wp+l /\ . . /\ W is nonzero and decompos able. ap. ' < kq. (c) If w = L i<j a ijVti /\ Vtj .h] ap. . . Vtn so that W = </>l /\ </>2 + w '. : al•hP ' . Let I :s p :s n be fixed.Diff erential Forms 229 8. .1 2 (V). j= ! Wt /\ . old it is called the rank of w. .k] (a) If v" . i<j choose </>l involving Vth Vt3. . show that B H = det ( al. Show that there is a basis </>" • • • . C K = det ( ap+ l 'k' : au. . . K . Vn is a basis of V and show that n Wi = L QjiVj. For H = hl < . 9. Vtn and </>2 involving Vt2. show that the rank of w is the rank of the ma trix ( aij). . . Thus r is welldetermined. . . . . let Vt. (b) Show that the rfold wedge product w /\ . " /\ Wn = L: C K VK . and let q = n .. (a) Let W E !.itp ) . </>n of V* such that Hint: If W = L aijVti /\ Vtj . . . If 10. . Let A = (aij) be an 11 X il matrix.p. . Vn is a basis for V and Wj = 'LJ=l (ljiVj. .
I (p)]*). 1/Ik E y satisf . . ' . a a (a) Show that if then ./\ . eH.H' VI /\ . . . • ". . ) h l . · /\ . .. . . . . . Show that VH /\ Vx = . . (I . (Cartan's Lemma) Let <PI . . . j= 1 k where bundle. . H E V* V* be independent and suppose that aji = aij ' Then =:: 1/Ii L Qjiq. if � = :n: : E + B is a vector 12. In addition to forms. . h2. the bundle whose fibre at Since we can regard as an element of ( Mp )** . H' is the sign of the permutation {o eH. where eH. . .w B H C H' . . For example.H (arranged in increasing order). . . (c) Prove "Laplace's expansion" det A 1/1" 1 1 .j . . hp .' ax l axil TM using 11 and other operations.. " } . aIlY section of I1 n (T*M) can be written locally as h . kq = " .H' K = H'. L. .. kl . we can consider 11 k (C). . we can consider sections of bundles constructed from p is 11k ([:n:. /\ Vn " K #. (/>k .230 Chapter 7 (b) Let H' = { I .
Show that sections o f :Tik [n.W ( V) . corresponds to the map Show that sections of :Tik [nl (TM) correspond to (even) relative tensors of type (1) and weight I .) (c) If :Tif ] ( V) is defined similarly. . Vn is a basis for V.I . k times I times .w] (V) b e defined like :Tik [n] . x V* + r:2n ( V) n </>1 /\ /\ </>n . /\ v*nJ . . . we can identify :Tik[m] (V) with :Tif ) ( V) with m I k Q9 V* ® Q9 V ® Am(v) I k Q9 V* ® Q9 V ® Am( V*) .. ..Differential F orms 231 This shows that sections of r:2n ( T * M) are the geometric objects corresponding to the (even) relative scalars of weight I in Problem 410. W (f) For those who know about tensor products V ® W and exterior algebras k ( V). x V + IR which are of for some W E r:2 n ( v) . . </>n ) f+ ponents e'I . (Notice that if VI > . (b) Let :Tik [m] ( V) denote the vector space of all multilinear functions show that sections of :Tit. . (e) Suppose r:2n .w\ TM) correspond t o (even) relative tensors of type (J) and weight w. (d) Show that the covariant relative tensor o f type e ) and weight 1 defined i n Problem 410.] (TM) correspond to (even) relative tensors of type (1) and weight . . . Similarly for :Tit. Interpret the relative tensor with com 'v' times V x .. � x � + r:2m (v) . except that r:2m(v) is replaced by r:2m( v*). Let :Tik[n . • . W ( V) denotes all functions ry : the form V* x . these results can all be restated. We can identify :Tik ( V) with A w an integer Since r:2m(v) '" Am(v*) '" [Am(v)]*. then elements of r:2n(v) can be represented by real numbers [times the element V*I /\ . ] . . with components e.] · ··ill . except that r:2n(v) i s replaced by r:2n . I times k limes m • • • given by (</>1> . .i'1 similarly.
Chapter 7 and weight w and k 7it .(trace A)A n . if 8 is as defined on page 1 30. .I + C2 An.232 Consider.· . and A is a tensor of type G).A) = An .2 + · .. For every n XI1 matrix A = (aij) we then have a number P(aij).. We will . V + V is X(A) = det(A! .1 ) for all A and all invertible B . (This may be used as a definition of det A. more generally. (a) Show that Ck = trace of A * : !. k times 15.  n trix (a! l (with respect to some basis).. show that sections of 7ik[n.'. (b) Conclude that cdA B) = cdBA ).1n(v) + !. a!k ()Jl ··iN ' lk .I) " det A = An C l An. m Noting that An(V) 0 · · · 0 An(v) is always Idimensional. I}. show that . . (c) Let 8/. . then the function p r+ cdA (p» can be defined as a (2k)fold contraction of � A 0 · · · 0 A 0 8. Cn defined in Problem 14. .1k (V) + !. This problem outlines a proof that any invariant P is a polynomial in the polynomials CI .1 k (V) . respectively. j].wl (TM) correspond to (even) relative tensors of type (1) n 7ik [m:wl ( V) = &/ V* 0 &/ V 0 0w A m ( V) 13. .t + . 14. .jk V + V has a ma Thus. Show that C = det A .::// be as defined in Problem 45 (xiii). . Recall that the characteristic polynomial of A : W. + (_ I)nc . Let P(Xij) be a polynomial in n 2 variables. If A : �l l l Ck (A) = � " a'1l a'2z .) (b) Conclude that det AB = (det A) (det B). · · + ( . then the map A * : !. (a) If V has dimension n and A : V + V is a linear transformation. Call P invariant if PtA) = P(BA B . .wl (V) = 0 V* 0 01 V 0 0 w A m ( V*) .wl (TM) and 7if . .· k •l L i] .1n(v) must be multiplication by some constant c. .'h. .
that the equation holds for all matrices A over <C. \( 375. compare pg. . .1 ) for all real A and real invertible B. .I )"an. Yn). Yn. " Yn) to be PtA) where A is the diagonal matrix Yl ( . . . Yn Then there is a polynomial R such that Q (Yb . . Recall that the at can be defined by the equation n i= 1 TI (y . . an. Conclude. . . Thus. up to sign. .(A b " " An). . Since the eigenvalues A t . . (d) Show that P t A ) R(el (A). en(A» for all diagonalizable A . (e) The same equation holds f complex A and complex invertible B. · · . . .. . . (b) PtA) = R(edA). . . . .0 ). + ( . since the set where D # 0 is Zariskidense. . . Yn». (a) Define Q(y l o . . . Yn) = R (adyl o . . .atyn. . . where Ai are the eigenvalues of A. Yn can be written as a polynomial in at . . . (Regard or the equation as polynomial equations in the aij and bij .) = n' . . . Yn . We will first consider matrices A over the complex numbers <C (the coefficients of P may also be complex). o .1 + . The polynomial R has real coefficients if P does. . . (This last conclu sion follows even if <C is replaced by some other field. by definition. . (c) The discriminant D ( A ) i s defined a s n ih (Ai . . Show that D(A) can be written as a polynomial in the entries of A . where ai is the ith elementary symmetric polynomial of Yb . . by continuity. . .Yt) = yn .Aj)'. an(Yl . they are the coefficients.) Now suppose that the coefficients of P are real and that PtA) = P(BA B . en(A» whenever D(A) # O. . the roots of the polynomial X(A). An of a matrix A are. Yn) in the variables Yl . . of the polynomial with roots Y b . this is "the principal of irrelevance of algebraic inequalities". it follows that ei(A) = a.DijJ rential FOWlS e 233 need the algebraic result that any symmetric polynomial Q( Yb ' .
. j=] Q'jiVj ... /\ j _ (b) Show that is (c) Using 514(e). Xk+l ) k+l l = " ( .·. . In Problem 514 we defined Lx A for any tensor field A . Wk ) = l=ilL a/ W(Vi" <···<lk where a{ is the determinant of the k x k submatrix of (aij ) obtained by selecting rows .... .. .. (b) Generalize Theorem 7 and Corollary 8 to kforms. . . 18.. . ik . .Xk ). then s o Lxw. .. . . X". • • or Vn be a basis f V. (a) Let Chapter 7 Vl . . Wk V be given by Wi = L n • • . . L. . Xk» = LxW(X1 .. ..Xi . Show that d( Li<j aij dxi dx j ) = 0 if and only if aai aai aaj axk ax}k + ax'k = 0 for all i < j < k .. (d) Deduce the following two expressions: dw(X" . . E w 11k (V) show that W(WI. . . . (a) Show that i f W i s a kf orm.. . . For E . .. and let WI. Xk+l ) L.Xi . show that X(w(X" . Xk ) k + " (. .XJ. Vik )' i].l)i+l w([X. 17. . . .» = Lx(w(XI. . . .234 1 6. (c) Check directly from (b) that the definition of d does not depend on the coordinate system..I ) i+ LXiw(Xl. . .Xi].
. . . sin 21fx). dw(Xt .)  21f dx . . . .I ) {Xi(w(Xt . 1R' .. .J w)(X" . If c : [0.. Compute that Hin!: First set up partial differential equations for the functions fik aUk laxj . 19. . == (At most places 0 = arctan y Ix [ + a constant].J dw = Lxw . . i.d(X . Xk+l ) (This may be used to give an inductive definition of d. l.axj ax! a'a/j a'a!k == axk axi . .A dx = dg for some function g with g (O) = g(I)... l . Un in a neighborhood of any point in IRn with ( ) it is necessary and sufficient that 20.. . and use Theorem 61. .. 1 ] . show that there is a unique number A such that W . == au) laxk "dO" 21.{Oj be the inclusion. . . . S l is c(x) = (cos 21fx. Xk +t » i=l 235 _ (e) Show that X . Xk+l ) k+l 1 i+1 Xi. and let a' = i*(dO). show that d(Lxw) == Lx(dw). Xk+ I ). . Let aij be ... . . . show that cOra') (c) If w is a closed I form on S I show that there is a unique number A such that w Aa' is exact. Show that in order for there to be functions U I . (b) Let i : S l . I aUi aU + j ai = '2 axj axi l == (LxI w)(X" . a'aij a'aik axk ax! . ' functions on IRn with aU = aji .'2 I ) . .J w). I ] to find A.A dx = dg on [0. . .d(XI . Xk+t ) . .) (f) Using (e).axj axi == x dy' + y dx x y' for all i. I] with /(0) = /( 1 ). . .e. (a) If w is a Iform / dx on [0. . k . . Hin!: Integrate the equation w . _ .erential Forms Diff dw(XI .
.U is connected. and 0 #. (a) Let U c IR ' be a bounded open set such that IR ' . Show that U is diffeomorphic to IR ' . and hence smoothly contractible to a point.(V\. and define H: U [0. . but U is not contractible to a point... . W. . . Vk ) = 0 if some Vi V. 1/ k .and Wt(Vt. . . If W= WI 22.) can be written as a sum of forms w.Ik /\ . where W I is a kform and w. . .ic (w. a= ' l < " '<'k l _ . show that 24.1 (1 I k . (The converse is proved in Problem 89.I Wi .) Obtain U as an increasing union of sets.*. /\ dXik • Hint: (b) Find a bounded open set U C 1R 3 such that 1R 3 . J(H*w) 1 (IX) dl) xi. dxi. vp) = 0 if some V. .I)form such that WI (VI .. . the kth set being a finite union of squares containing the set of points in U whose distance from boundary U is ::. is a (k . Vk I ) = 0 ifsome Vi V. ic). . .l) Ip. = L.U i s connected. then W can be written uniquely as WI + (b) If dim V. . V. . has degree f3 = k .236 Chapter 7 W 11k (VI V. . . where has degree 0' and WI EEl 0' E E E /\ E E x = on U. va) = 0 if some Vi V2 W. 1] + U by H(p. t( 1 ) . /\ d7a /\ . (a) Show that every E /\ W.(VI . VI . = I. . .. . 23. . Let U c IRn be an open set starshaped with respect to 0.
l o w) n = x .) 26.t (JRn). b) L>i bi. .t for all W E JRn. show that there is a unique vector Vt x . (First find all e. Let ( . the gradient of j.= 1 = (a) If Vt. since the length of rays from 0 to the boundary of the set could vary discontinuously. grad j = " . (a) If j : JRn + JR. L a X l. . . . ' x det (b) Show that x . . aXl. x E \1 n . Is U homeo morphic to JRn? (It would certainly appear so. ) be the usual inner product on JRn.Di erential Forms ff 237 25.) Intl"Oducing the formal symbolism i= 1 i=1 Dd . (a. x ej. define a vector field grad j.' . . . but the "obvious" proof does not work. Vn _l with (Vt E �n. Let U c JRn be an open set starshaped with respect to O. . on JRn by a n a n aj . 27. X Vn _l E �n V . using the expansion of a matrL by minors. . . x (c) For JR3 show that () � Vn . . and express it in terms of the e*i.= " L aXl.
(d) If X is a vector field on a starshaped open set U C IR n and curl X = 0.238 Chapter 7 we can write grad I = VI If (grad J)(p) = wp . for n = 3. Similarly. X). where Du/(p) denotes the directional derivative i n t h e direction v a t p (or simply vp (J). 2::7=1 a ' a/ax' is a vector field on IR n . Conclude that V/(p) is the direction in which I is changing fastest at p . we define the divergence (b) If X of X as (Symbolically. show that Du/(p) = ( v.) We also define. we can write div X = (V. then X = curl Y for some vector field Y on U. Define forms Wx = a l �x = a l Show that (c) Conclude that dl = Wgcad J d(wx) = �cudX d(�x) = (div X) dx /\ dy /\ dz. if div X = 0. if we regard vp E IR np). . dx + a2 dy + a3 dz dy /\ dz + a2 dz /\ dx + a3 dx /\ dy. curl grad I = 0 div curl X = o. w ) . then X = grad I for some fU l Iction I : U > IR.
as the maximum of Ii . .c' (l1')] + g(C(�I))[C2{11) . that is. �) = L J(C(�I» [C'(li) . then the limit is denoted by l Jdx+gdy.c2{1i_ l l.CHAPTER 8 INTEGRATION c: [0. then we can divide the curve c into pieces. denote the coordinate functions on JR2). �II) . 1] JR2 is a curve and w = J dx + g dy is a I form on JR2 (where J.lid. each such piece is approximately a straight segment. then the equation lim S(P. the piece going fi'om C(li. If we choose a partition 0 = 10 < . < In = 1 of [0.Ii_I are small.Ii_ I approadlCS 0.e' (til) horizontal projection C'{lI) . 1]. When the differences Ii . for example.C2(t1_' ) e(O) c(I[_. �n ).c2{11_l ll i=1 If these sums approach a limit as the "mesh" I PI I of P approaches 0. if IIPII = m!'x( ti . consider the sum " S(P.) . (This is a complicated limit. i.' ) to C(li). y.\C. C(tl) e(�I) :_ C2(t1) . and and. .h x he basic concept of this chapter generalizes line and surface integrals. To be precise. Suppose. We can choose points C(�I) on each piece by choosing points �I [IihIll For each partition P and each such choice � = (� . . with e(1) that > > n T which first arose fi'om very physical considerations. g: JR2 JR.�) = [ Jdx +gdy lc E IIPII� O 239 .C'{lI_ ' ) and vertical projection c2 (11) .
+g ax ay then is the "work" involved in moving a unit mass along the curve c in the case where c is actually a straight line between and and and are constant along these straight line segments. so . there is a Ii > 0 such that for all partitions P with II P II < Ii.C2(ti_d S(P. the integral is the "sum" of these infinitely small displacements.�) = LJ(�i. YO)(li .�) IiI Ii J g J dx + g dy dx. a a J. while C2(li) .) IR2. If we consider a "force field" on described by the vector field � P. (In classical terminology. the differential would be described as the work done by the force field on an "in fInitely small" displacement with components ely. = yo+� In this case.) Before worrying about how to compute this limit.IiI .240 means: for all E we have > Chapter 8 0. it has a natural physical interpretation. C(I) (I.Iid· i=1 . CI (li ) . the limit js the natural definition of the work done in the general case. Yo).C l (tid = Ii . = 0. consider the special case where S(P. for all choices for The limit which we have just defined is called a "line integral".
a) l ' J(xb + ( 1 x)a.yo).lil l· i= = b These sums approach A somewhat messy argument (problem I) shows that these sums approach what (b .yo)dx.li1l f3i E [Ii"" Ii]. then a c' (Ii) .C' (li' ) = C" (Cii)(li liIl Cii E [Ii.�) = L { J(C(�i» C" (Cii) + g(C(�i » C" (f3i)} (Ii liI). On the other hand. if Yo C(I) = (Ib + (1 . denotes l' [J(C(I» C" (I) + g(C(I»)c" (I)] dl. namely Physicists' notation (or abuse thereof) makes it easy to remember this result. so S(P. x c'.�) = (b .Ii] C' (li) .I)a. by the mean value theorem.li' ).C' (ti1l = C" (f3i)(li .c' (Ii' ) = (b . YO)dX. we have.yo)dx lb J(x.a)(ti . C' (li) .integration These sums approach 241 lJdx + gdy = l ' J(X.. In general. Yo)(li . c' C y x . .a) . So S(P.e.�i)a. of are denoted simply by and The components [i. L1 J(�ib + ( 1 . for any curve c.=1 it looks like they should approach.
li1l [=1 clearly also approach Consider the special case where 1 1 [j(e(l) e ll(l) + g(e(I» e21(1)] d/ . RecaIJ that denotes the = de/dl(�i) e(ti ) tangent vector of c at lime de dl (�i) �i . this is indicated c1assicaIJy by saying "let x = X(I). we Can in troduce a more geometrical interpretation. Ii). The above in tegral is then written 1 / dx +gdy [o I [j(x'Y) dx +g(x'Y) di] d/ . dY d1 J " In preference to this physical interpretation of "line integrals". (I) ] . e goes with constant velocity on each (IiI . Then the sums = L [j(e(�i» e ll(�i) + g(e(�i»)c21(�i)] ' (Ii .242 0 y y o Cha pter 8 y =y x e and denotes e. .
can be used as a w(C(I» w: = limit of the sums (*) C crt). dc/dl C C.integration 243 In this case.IiI) = length of segment from C(li1l to C(li). �i E (li' . dy. then dc dt (�i) = the constant speed on (Ii). kernel w (c(t» But if W arbitrarily). we choose W(C(I» E !1 ' (IR'c (I» so that W(c(t)) (� ) = 1 . and then extend W to IR'. (Ii . � (�i)] . the infinitely small piece at with components having length 1W c. Before pushing this geometrical interpretation too far. this is the modern counterpart of y the classical conception. Ii).li1l dl c. we should note that there is no Iform w on IR' such that 1 w = length of C for all curves c. (choosing the kernel of can be thought of as the "length" of when our ruler is changing contin uously in a way specified by Notice that the restriction of to the I dimensional subspace of IR'c(l) spanned by is a constant times "signed length"..Ii) length of the segment from C(li 1l t o C(li) Ii {I I so [length of � (�i)] . If we choose any length of  is the length of and the limit of such sums. t' [ i= length of dx. (Ii . /(C(I)) dx + g(c(I » dy. for a general definition of the length of The line integral C. C is . It is true that for a given oneone curve c we can produce a form w which works for c. The natural way to specify a continuously changing length along is to specif a length on its tangent vectors. . whereby the curve is divided into infinitely small parts.
+ . I] c. if we are interested in obtaining the ordinary length of a curve. 1] 7 [ [/(c(l» C" (I) g(c(t))c2l(t)] cit.�) for c is dearly equal to a sum S ( P'. 1 2 c One property of line integrals should be mentioned now. the subspaces /'. l l c c: [0. will have "length" Later we will see a way of circumventing this difficulty.� there is no element of !1 ' (IR 2c(d) which has the value on all three vectors. so it is clear from our first definition thal for a curve we have [0. I].2. in the situation shown belov. ] [0. In general. we note that the sums t*j. and w is a I form on M.even for a curve but in this case we can proceed right to the integral these sums approach. because it is ob vious with our original definition and merely true for our new definition.2 0. J] 7 ffi. given any w on which is everywhere nonzero. namely c: [0. 1] 7 ffi. [0. �') for c o p."hen we consider the sums (*) f a curve or M.244 Chapter 8 not oneone this may be impossible. I] [0. c: [0. For the present. so we can define Ie W as the limit of these sums. make sense even if is a curve in a manifold M (where there is no notion of "length"). any curve contained in an integral submanifold of /'. If is a oneone increasing function from onto then the p: I > > M is called a reparameterization of cit has exactly the curve c o p : same image as but transverses it at a different rate. Every sum S(P. used to define this generalized "length". for example. and conversely.p = ker w(p) form a I dimensional distribution on IR .2 wc op w ("the integral of w over (' is independent of the parameterization"). This is no longer 50 dear . I] [0. ffi. nor is it deal.
< Notice that if ffi. We are being a bit sloppy about all this because we are about to introduce yet a third definition.integration 245 The result then follows from a calculation: the substitution + For a curve in ffi. for a general manifold M. e I is the standard coordinate system on JR. 1]2 JRn . . let � = e* (�) a ae ay = e* ( ay ) . to be precise. 1].. we can introduce a coordinate system for our calculations if lies in one coordinate system.n . < In of [0. and take the integral of h on [0. If x and y m·e the coordinate functions on JR 2. if we choose < . 1]. so that formally we just integrate e*(/ dx + g dy). and a l form there is a similar calcula 2:::7=1 tion. Consider once again the case of a Iform on where [ [/(e(I»)c" (I) g(e(I»)c2l(1)] dl = lP' ( I) [/(e(p(u» )e" (p(u)) g(e(p(u)))e2l(p(u))]p'(u) du = [ [/(e p(u))(e p)" (u) g(e p(u))(e p)21(u)] duo p ' (O) + 0 I = p(u) gives 0 + 0 0 e([O. I] > JR we have 0 0 0 > > So s'" �ffQS . which will eventually become our formal choice. 0 0 0 0 0 e: [0. .. I]) w= Wi dxi . 1 w = [ [/(e(I» e"(I) + g(e(l))e2l(1)] dl. 1] JRn could be generalized to functions e: [0. we write c*(J dx + g dy) = h cll (in the unique possible way). Everything we have said for curves e: [0. or break up into several pieces otherwise. then for the map 2 e*(J dx + g dy) = (J e)e*(dx) + (g e)e*(dy) = (J c) d(x c) + (g c) dry c) = (J e)e" dl + (g e)e 21 dl.2. For a pair of partitions and 10 < .
w X l . l lk [ l lk as logic permits If w is a k form on M. c [0. . [O. Sj1 [ljI . To make a long slOry short. ( = ) = . are the coordinate functions.lj_l ) is a "generalized area" of the parallelogram spanned by ac ac a. . The limit of sums of these terms can be thought of as a "generalized area" of c.(�fj ). /\ dxk We define = J ! (X I . then Cha pter 8 [0. .. Si_1 c: [0. . ° E IR. .. For k 0. A Coo function It > M is called a singular kcube in M (the word "singular" indicates that is not necessarily oneone). then can be written uniquely as Sj w(C(�ij)) (�: (�jj). We will Jet I Jo = = so that a singular Ocube is determined by the one point M. and c is a singular k cube in M. and . we have a special definition: a Oform is a function !. . . we define 1 w = f c'w. we now proceed with the formal definitions. mImIC as far [O. IRo ck c(O) E IR Jk : [0. llk in c1assical notation.f ! notatIOn attempts to w�ic? modern . . [0. f W lO be O. ay(�ij ).Sj_JJ(lj . The inclusion map of It in will be denoted by It > it is called the standard kcube. . * (�ij» ) (Sf . lj1 and w is a 2form on IRn. l lk where the right hand side has just been defined. . and for a singular Ocube c we define 1 ! !(c(0)).. dXk [O.Xk ) dX I .246 X �jj E [Sj_l . Xk w w = ! dx l /\ . IRk. If is a kform on It.
• . l w = f f.illlegmiion 247 > JRn be a oneone singular ncube with PROPOSITION. It [0. II") by the change of variable formula . /\ . . /\ dx" by Theorem 7 7 dx" by assumption > 2. •:. C{[O. I}" 0 det e([O. We have L w = f (e plow = f p*(e*w) p = f e*(w) 0 � I� � It [0.11" = [O. COROLLARY. Let det 2: on Let be the nform 1. by the Proposition.1]" p: [0. Let be oneone onto with det let be a singular k cube in M and let w be a kform on M. since p is onto . Jc leo p p' 2: 0.1)II) PROOF. /\ /\ .11" = 0 e) (det [0. 1]n e' 0 [0. By definition. = 1 w f e*(w) f (J e') dxl f (J e)1 e'l dxl f J [0. w w = J dxl /\ • • . /\ dx". 1]". Then e: [0. . Then e PROOF. It [ w = [ w.
' lk For example. I] M then l' I(c(t)) dt is generally i' l' I(c(p(t»)) cll. which is called the boundary of c. It [0. purely formally. > f l o c. diff rential forms are the things we integrate be e cause they transform correctly (i. in the obvious The reason for introducing k chains is that to every kchain c (which may be just a singular k cube) we wish to associate a (k . it is called orientation preserving or orientation reversing depending on whether det p' 0 or det p' < 0 everywhere. and which is supposed to be the sum of the various singular .248 > Chapter 8 > independence of parameterization. an ori entation reversing reparameterization clearly changes the sign of the integral.. A kehain is simply a formal (finite) sum of singular k cubes multiplied by integers. c: [0. /\ dx k ).g. e. Notice that there would be no such result if we tried to define the integral over of a Coo function M > )R by the formula The map c o p: [0. The k chain Ie. . It M is called a reparameterization of c if p: [0. We add kchains. . and multiply them by integers. so that the change of variable formula will pop up).g.e. w(' define the integral of w over a kchain way: c J2::: uiCf [ W = I>lCj W.2<3 + 6c4. if From a formal point of view. I I . + 3<4 ) + (2)« . The corollary thus shows > I: c [O. e. provided it is orientation preserving. Our definition of the integral of a k form w aver a singular k cube < can immediately be generalized. will also be denoted simply by Ct . 1\1orcovcr. functions On a manifold cannot be integrated (we can integrate a function on the manifold )Rk only because it gives us a form dx ' /\ . I t is a Coo oneone onto map with det p' i' 0 everywhere (so that p' is also COO). in accordance with Theorem 77. + <3 + C2) = 2C2 . j = Li ajC..I )chain a c. <. =I 2(c. .
.l) = (x l . . . . I) = (X l . xiI . . )..1) . . . .xn . Xi .O. The boundary of y the sum of the four singular I cubes indicated below on the left.I ).l ). xi. . 1.O) (x) = I" (x l . . . l . . o f the four singular I cubes shown o n the right.O) In) I]n.Integration 249 (k .xnI ) 1.xn. O. 1 . P I I 12.I)cubes around the boundary of each singular k cube in c.. . . i .x i . . In practice. (Notice that this will not change the integral of a I form over a/ 2 ) For each i with 1 :0 i :0 n we first define two singular (n .:. . . it for example.I . (x) = I" (xl .I . ._n . but tbe sum.0) . . then 1i). +1 +1 with the indicated coefficients. .xil . /11 .O)face and (i. . . will not be is convenient to modif this idea.xi . I )face of as follows: If X E [0. . . . .x i . .I .:. .l) /11 .I)cubes and (the (i. X Ii). . . .
which we will usually simply identif with the point P = c(O).S) .O) We also have.(S PH (Q = 0.. and for a Ochain Li aiCi we define c: [0. i=1 a=O.1) R .I) C(1. > M.P) p C(2.250 Chapter 8 The (i.1 = O. the boundary of all nchain Li aici is defined by aC = L L ( _ I )i+. C(i.Q) . 1]0 Notice that for a I cube c : [0. 1 . I] > M we have so a(aC) = ac = C(2.O) s  .1) a(aC) = (R . C(l. y we define a c to be the number 1 E JR. For the case of a Ocube > M. 1]2 + C(2 . 0)." )face of a singular ncube Now we define C(O) � C is defined by C(i..O) / C(I) C 0 Finally.).l These definitions all make sense only for n ::: I ..(R .) ' C(I .) = C (Iv. C(1. for a singular 2cube c: [0.
.L 1=1 j=1 (t a=O. .Pl = ( /'( + . and consider ( /D.l (I)'+·+j+p(C(" .P) ' For x E [0. Similarly. . but even ac = O . a .1 . . xn2).P) · L . .. ... from the definition ( /(�. . .) U. Pl)(" . In general. Therefore ' all terms cancel in pairs.Pp(�:. .) u. and a(ac) = O. . . .1 nLI P=O. PROOF Let i :0 j :0 n . a2 = o.P) and ( CU+I . .) u . xj . " ..xj.I .1 (I )'+. ...)) L . Thus ( /. . . xj.) for i :0 j :0 n I. /3. It follows easily for any j I singular ncube c that ( c( . . .. xi.) U.P) (x ) = J'J . Since the theorem is true for singular ncubes. it is clearly also true for singular nchains. . x .) for i :0 j ::S n . . . Xn2).1 . .) ( /. . Now i I a(ac) = a = L a=O. X i . Pl)(" . For example.) U.P) (i. we have: 3.) occur with opposite signs.I . PROPOSITION. .\ (x » = J{ + .) (X I . . .) = J.I. P) ( . xn2) = In (x l . . a. fj. . . Notice that for some nchains c we have not only a(ac) = 0. xi. . Xj .( P) (x » / = J(�. .•) U . xiI . we have. ( /U+I.xj. a good exercise because this involves figuring out just what the boundary of a 3cube looks like. ( C(i. C(" . xiI . xn2) j I i = In (x l . Briefly. C{ . xj .. p. .1 . then a(ac) = o.P) = (CU + . . . where CT and C2 are two I cubes  . . this is the case if c = c) e2 . If c is any nchain in M. . . . •:. 1]"2 .j+I .=1  In this sum.integration 251 From a picture it can be checked that this also happens for a singular 3cube.: .P) (XI.
e. this terminology has been purposely chosen to parallel the terminology for chains (on the other hand. a chain of the form ac is not desCllbed. If c is just a singular I cube itself.. we have seen that on IR ' . Recall that a differential form w with dw = 0 is also called "closed". There is also a I chain c which is closed but not a boundary. by the classical term of "exact". reciprocally. when c is a "closed" curve. For example. expressed by the relations d' = 0 and a ' = O. any k chain c is called closed if ac = O.{OJ there is a I form "de" which is closed but not exact. This parallel terminology \'\'as not chosen merely because of the formal similarities between d and a. then ac = 0 precisely when c(O) = c(I). The connection between forms and chains goes much deeper than that.252 with c. In general. a closed curve encircling . but is simply called "a boundary"). namely. i. (0) = Chapter 8 C2 (0) and CI (1) = c2 (1).
w is a sum of (k .k ) d. . . a little notation translation shows that J[o. chains. First. . 1 .l)k1 [ J(k . . . 1)" I(x. . . /\ dxk ) if j i' i = [0. dx·k � .k if j i. . . . . I . dxk /(. Jae PROOF Most of the proof involves the special case where w is a (k . If w is a (k . 4. dx. xk ) dxl . /\ = jof dxi /\ . a.1 ) 1+ 1 [O.l)k f l(xl .integra/lOll 253 the point ° once. . . .k ) dx. . . then Jc [ dw = [ w. O. • . . • • . .I)forms of the type and it suffices to prove the theorem for each of these. l . . Although it is intuitively clear that (" is not the boundary of a 2chain in IR 2 .x. . . x. . and a. .I )form on M and c is a k chain in M.x·1 . We now compute.I )form on IR k and c = J k In this case. .I)k + (.)* U dx l j. . /\ . the simplest proof uses the theorem which establishes the connection between forms. .{OJ. .1/ f [O. Therefore = ( . THEOREM (STOKES' THEOREM). . .x·1 . d.
. dxk . /\ dxk ) /\ . . . xk ) dx' . . . ) . . . chasing through the definitions shows that [ dw = [ W. . . . (1' D (x' . . . . . )k By Fubini's theorem and the fundamental theorem of calculus we have [ d(J dx ' 1tk = (_I)i. . dxk [O. . .' f DJ. . Xk ) . 1tk Jatk c'w = [ w . . . xk ) ] dx ' . )k + ( )' f l(x ' . .1 1 · · · 1 [/(x'. . xk ) dx ' . /\ . . . [O.1 f . . . )k Thus . . dxi . .' l' . . . . /\ dxi /\ . . . I ' I(x . •:. )k (_I )i . /\ dxi /\ " . /\ dxk ) J . . lac . [O. latH 1tk [ w = [ c·w. . lc dw = [ c'(dw) = [ d(c'w) = [ Jtl. O. xk) dXi) dx' . ) . 0. . . . /\ dxi /\ . . .. . . dxk ' ' = () 1 . . . . Chapter 8 [ d(J dx ' }lk /\ . . dxi . . dxk = (_) 1. . . For an arbitrary singular k cube.254 On the other hand. . /\ = = f Di 1 dx' dx. . . .. . .. . /\ dx k [O. . Jac Jatk Theref ore The theorem clearly follows for k chains also .(x ' . .
00) x {OJ) > jR: We have or x .x2 + y 1c de = 1c y2 dx + 2 dy x2 + y 1c Jac2 1 c 1 c2 "de" really c l (£. For. we could just as well have used it to show that w = "de" is not exact. then we would have d(de) = 0 = o.integration 255 Notice that Stokes' 1l1eorem not only uses the fundamental theorem of cal culus.  1c 1c dl w= = [ I= Jac 1 0 I = o. we wiII eventually obtain a 2form w on jR3 . although closed. and e(1 . Tfwe did have c = ac 2 .£) . using the fact that is de f e : jR 2 . then we would have 21f = \Ale were previously able to give a simpler argument to show that "de" is not exact. [There is also a noncomputational argument.{OJ. but Stokes' Theorem is the tool which will enable us to deal with forms on jRn {OJ.] Although we used this calculation to show that c is not a boundary.e(£) > 21f as £ > 0. if we had w = dl for some Coo function I : jR 2 .e(£) . but actually becomes that theorem when c = / 1 and w = f. I £) f de = e(l . 1] > jR 2 _ {OJ defined by C(I) = (cos 21f1.{OJ > jR. de = [ de = But a straightforward computation (which will be good f the soul) shows that or = 21f. As an application of Stokes' Theorem.£) .([0. is not ac 2 for any 2chain c 2 . sin 21f1). w= x dy /\ dl Y dx /\ dz + z dx /\ dy (x 2 + y2 + z2) 3/2  . we show that the curve c : [0. For example.
However. each contained in some ('([0. This is possible only when M is orientable. I]" ) . IJ". The common number 1 w=l C2 1 w .1 OC] ) w= .0 (' .lion jm!servillg (with respect to the orientation jJ. and ('. will be denoted by > 1 with sup If w is an arbitrary nform 011 M. If w is an nf orm 011 M such that support w c [0. I]") and c orientation preserving. In fact. I]" ) n c. Assume that c. I]") n c.l a c) )' 2: 0. then there is a cover I!J of M by open sets U. and the usual orientation on JR"). if <I> is a partition of unity subordinate to this cover. l C] w. but they all turn out to give the same result. where c is a singular ncube of this sort.{OJ. which is basic for our definition. phisms in a neighborhood of [0. because of the fact that support w C (' . since ('1 and ("2 are both orientation preserving). I ]" 5. I]") . is not defmed on all of [0. but a straightforward calculation shows that dw = 0. . : M be two singular ncubes which can be extended to be diffeomor > c. Let M be an nmanifold with an orientation {/" and let c" c. ([0. I]" M ('([0. we first want to describe a way of integrating nforms over nmanifolds. ([0. the reason will be clear from the next result. For the moment we are keeping the origin of w a secret. and write port w C The only problem is that c. THEOREM. then . for singular ncubes c : [0. are both orienlo. I]").([O. To prove that w is not exact we will \'vant to integrate it over a 2chain which "fills up" the 2sphere S' C JR 3 . 11" (it does satisfy det{c2 . C2 then PROOF We want to use Corollary 2. There are lots of ways of doing this. •:.256 Chapter 8 which is closed but not exact.([O. w..l W• CI C20{c2. a glance at the proof of Corollary 2 will show that the result still follows. on M.
) \t\'ith minor modifications we can define J w even if M is an nmanifold withboundary. of M we clearly have IM.!')W' However) we usually omit explicit mention of p.1M. .. E [ so that our definition does not depend on the partition. If M C IR n is an ndimensional manifoldwithboundary and J : M > IR has compact support. which form a locally finite collection. . We wish to define 257 We will adopt this definition only when w has compact support. Likewise.!') W M where the right hand side denotes the ordinary integral. /\ dX" = f f. and w is an nform with compact support on N. then if J is orientation preserving if J is orientation reversing. if J: M n > N n is a diffeomorphism onto.!') W . Ifwe have another partition of unity \jJ (subordinate to a cover <9'). fM Jdx' /\ . (We really should denote this sum by for the OIientation  {/. then IM. then these sums are all finite. and the last sum can clearly also be written as [ 1M W = ¢Lc'pJM ¢ · w. in which case the sum is actually finite.. This is a simple conse quence of Proposition 1.1nlegralion is defined for each ¢ E <1>. M . since support w can intersect only finitely many of the sets {p : ¢(p) i' OJ.
. (x. . there is a way of discussing integration on nonorientable manifolds. Suppose that w is a function on M such that for each p E M we have for some �p E g. pUlling i n absolute value signs in the right place . . det a.. . One way of obtaining a volume element is to begin with an IIform � and then define w(p) = 1 � p) l .. V) a coordinate system around q = E N. U) is a coordinate system. If (x. /\ . For example. But this cannot be true on all of M. Vn E Mp we have Sudl a function w is called a volume element on each vector space it deter mines a way of measuring IIdimensional volume (not signed volume).r · Idx ' /\ . imbedded in ]R 3 Since Mp can be considered as a subspace of ]R3p. •:. 10calIy. consider the Mobius strip M.258 Chapter 8 Although IIforms can be integrated only over orientable manifolds. Theorem 77 has an obvious modification for volume elements: w(p)(vp. i. wp) = area of paralIelogram spanned by v and w. . for any n vectors VI . Howeve.e. U) is a coordinate system around p E M. . If J : M > N is a Coo function between l1manifolds. . /\ dxnl J J for J ?: 0. . 77'. not every volume element arises in this waythe form 1}p may not vary con tinuously with p. then on U we can write we calI w a Coo volume element if is COO. . w is of the form w = I�I for an l1form � . . we can define w = Idx ' /\ . since there is no IIform � on M which is everywhere nonzero. THEOREM. then for nonnegative V > ]R we have J*{g Idy ' J(p) PROOF Go through the proof o f Theorem 77. and (y. n ( Mp) . /\ dx"l· . ( It is not hard to see that w is a volume element. /\ dyn J) = g: (g I ( a(y'o J) ) 1 0 f) .
integration
259
then
78'. COROLLARY. If (x, U) and (y, V) are two coordinate systems on M and g Idy l /\ . . . /\ dY"1 = h Idx l /\ . . . /\ dx" l g, h ? 0
[This corollary shows that volume elements are the geometric objects corre sponding to the "odd scalar densities" defined in Problem 41 0.]
It is now an easy matter to integrate a volume element w over any manifold. First we define
1[0,1}11
[
w=
[O,I}"
f
J for w = J Idxl
>
Then for an nchain c : [0, I]"
Theorem 77' shows that Proposition I holds for a volume element w = J Idxl /\ . . . /\ dx"l even if det c ' is not � O. Thus Corollary 2 holds f volume elements or even jf det p' is not � O. From this we conclude that Theorem 5 holds for volume elements w on any manifold M, without assuming Cl , C2 orientation preserving (or even Ulat M is orientable). Consequenuy we can define JM w for any volume element w with compact support. Of course, when M is orientable these considerations are unnecessary. For, there is a nowhere zero nform 1] on M, and consequently any volume element w can be vvritten w = JI� J, J ? O. If we choose an orientation jJ., for M such that W(VI , " " v,J positively oriented, then we can define
>
l
e
M we define
/\ . . . /\
dx"J, J ? O.
w=
J[O,1}rI
[
c'w .
0 for V I , . . . , V
n
1M
[ w= [
1(M,,,)
h
Volume clements will be important later, but for the remainder of this chapter we are concerned only with integrating forms over oriented manifolds. In fact) ) our main result about integrals of forms over manifolds ) an analogue of Stokes Theorem about the integral of forms over chains) does not work for volume elements.
260
Chapter 8
Recall from Problem 316 that if M is a manifoldwithboundary, and p E then certain vectors v E Mp can be distinguished by tbe fact that for any coordinate system x : U > IHJ n around p, the vector x.(v) E IHJ nf(p) points "outwards') . Vile call such vectors v E Mp "outward pointing". If M has an
aM,
orientation {/" we define the induced orientation a{/, f aM by the condition that or [vl , . . . , vn _d E (a{/,)p if and only if [W, V I , . . . , Vn _ l ] E {/,p for every outward pointing W E Mp. If {/, is the usual orientation of IHJ n , then for p = (a , O) E IHJ n we have
{/,p = [(el lp, . . . , (en )p]
=
=
(_1)"  1 [(en )p, (el lp, . . . , (en llp] (  I ) n [( en )p, (el lp, . . , , (en l lp].
Since (  en )p is an outward pointing vector, this shows that the induced orien tation on IRn 1 x {OJ = alHJ n is (_ I) n times the usual one. The reason for this choice is the following. Let c be an orientation preserving singular ncube in (M, {/,) such that aM n c([O, 1]") = C(�,o) ([O, 1]"1). Then c(n ,O) : [O, l]n1 >
(aM, ap.,) is orientation preserving for even n, and orientation reversing for odd n. If w is an (n  I )fol'm on M whose support is contained in the interior
llllegratio11
261
of the image of c (this interior contains points in the image of C(n ,O)), it follows that
Jell/.m
But
[
w = (l)" [ w .
JaM
n c(n,O) appears with coefficient (_1) in ac. So
Jae
[ w= [
J<I)II C(II ,1ll
n w = (_1) [
Jell/.Ill
w = [ w.
JaM
Ifit were not for this choice of in the following theorem.
aft we would have some unpleasant minus signs
6. THEOREM (STOKES' THEOREM). If M is an oriented ndimensional manifoldwithboundaIY, and aM is given the induced orientation, and w is an (/l  I )form on M with compact support, then
JM
[ dw = [ w . JaM
PROOF Suppose first that there is an orientation preserving singular ncube c in M  aM such that support w C interior of image c . Then
[ dw = [ dw = [ w 1M Jc Jae =0
while we clearly have
by Theorem 4 since support w C interior of image c ,
Suppose next that there is an orientation preserving singular ncube c in M such that aMn c ([O, I J ") C(n, O)([O, 1]" 1 ), and support w C interior ofimage c. = Then once again
lM w = O.
[ dw = [ dw = [ w = [ w by (*). 1M Jc Jae JaM
262
ChapI,," 8
In general, there is an open cover I!J of M and a partition of unity <I> sub ordinate to I!J such that for each ¢ E <I> the form ¢ " w is one of the two sorts already considered. We h ave
0 = d(l) = d I »
¢E 4>
so
(
=
¢E 4>
L d¢,
Since w has compact support, this is really a finite sum, and we conclude that
¢E4>
L d¢ /\ w = O.
Therefore
1M dw = L 1M ¢ . dw = L 1M d¢ /\ w
¢E 4> ¢E 4> ¢E 4> 1M ¢E 4> laM
+¢
. dw
= L [ d(¢ ' W) = L [ ¢ · w = [ w. ·:·
laM
One of the simplest applications of Stokes' Theorem occurs when the oriented Ilmanifold (M,!,) is compact (so that every form has compact support) and aM = 0. In this case, if � is any (/1 I )form, then
[ d� = [ � = O. 1M laM
Therefore we can fwd an nform w on M v.lhich is not exact (even though it must be closed, because all (Il + I )fOrms on M are 0), simply by finding an w with Such a form w always exists. Indeed we have seen that there is a form that for VI , . . , VII E Mp we have

.
L w i' O.
w such
w(v] , . . . , v,,» O if [v] , . . . , v,,] = !'p.
If c; [0. 1]" > ( M . ll ) is orientation preserving. then the form c'w on clearly for some g > 0 on [0, I] ".
[0, I]" is
inl£gralion
263
so Ie W > o. It follows that 1M W > o. There is, moreover, no need to choose a form W with H holding evelywherewe can allow the > sign to be replaced by 2: . Thus we can even obtain a nonexact nform on M which has support contained in a coordinate neighborhood. This seemingly minor result already proves a theorem: a compact oriented manifold is not smoothly contractible to a point. As we have already empha sized, it is the "shape" of M, rather than its "size", which determines whether or not every closed form on M is exact. Roughly speaking, we can obtain more information about the shape of M by analyzing more closely the extent to which closed forms are not necessarily exact. In particular, we would now like to ask just how many nonexact nforms there are on a compact oriented Ilmanifold M. Naturally, if w is not exact, then the same is true for w + d� for any (n  I )form �, so we really want to consider w and w + d� as equivalent. There is, of course, a standard way of doing this, by considering quotient spaces. We will apply this construction not only to Ilforms, but to forms of any degree. For each k, the collection Z k (M) of all closed kforms on M is a vector space. The space B k (M) of all exact kforms is a subspace (since d 2 = 0), so we can form the quotient vector space H k (M) = Z k (M)/B k (M); this vector space Hk (M) is called the kdimensional de Rham cohomology vector space of M. rde RJzam � T heorem states that this veClor space is isomorphic to a ccrtain vector space defined purely in terms of the topology of M (for any space M), called the "kdimensional cohomology group of M with real coef ficients"; the notation Z k , Bk is chosen to correspond to the notation used in algebraic topology, where these groups are defined.] An element of H k (M) is an equivalence class [w] of a closed kform w, two closed kforms w, and W2 being equivalent if and only if their difference is exact. In terms of these vector spaces, the Poincare Lemma says that H k (JRn ) = 0 (the vector space containing only 0) if k > 0, or more generally, Hk (M) = 0 if M is contractible and k > o. To compute HO (M) we 110te first that B O (M) = 0 (there are no nonzero exact Oforms, since there are no nonzero (  I )forms for them to be the dif ferential of). So H O (M) is the same as the vector space of all Coo functions I : M > JR with dl = o. If M is connected, the condition dl = 0 implies that I is constant, sO HO (M) '" K (In general, the dimension of H O (M) is the number of components of M.) Aside from these trivial remarks, we presently knO\"/ only one other fact about Hk (M)if M is compact and oriented, then Hn (M) has dimension 2: I. The further study of Hk (M) requires a careful look at spheres and Euclidean space.
264
On S,,l C JR"  {OJ ISIII (1' > 0: for (vJ)P ) "
Chapter 8
there is a natural choice of an (n  I ) form (J' with " (Vnl )p E snlp , we define
Clearly this is > 0 if (V l )p , . . . , (V )p is a positively oriented basis. In fact, ,,l we defined the orientation of sn l in precisely this waythis orientation is just the induced orientation \'vhen sn l is considered as the boundary of the unit ball {p E JR" : ipi :0 Ij with the usual orientation. Using the expansion of a determinant by minors along the top fOW we see that (1' is the restriction to s n l of the form (J on JR" defined by
The form (J' on S,,l will now be used to find an (n  I )form on JR"  {OJ which is closed but not exact (thus showing that H,,l (JR"  {OJ) i' 0). Consider the map r : JR"  {OJ > S,,l defined by
n (J = L (  I ) i l X i dx 1
i=]
/\
. • .
/\
dxi /\ . . . /\ dx".
r(p)
Clearly r(p) = inclusion, then
=
TPi
p
=
p v(p) '
> JR" 
p if p
E
S"l; otherwise said, if i : S,,l
r 0
{OJ
is the
(In general, if A C X and r : X > A satisfies r(a ) = a for a E A, then r is called a retraction of X onto A.) Clearly, r* (1' is closed:
i = identity of S"l .
d(r*(J') = r*d(J' = O.
However, it is not exact, for if "*(1' = d1], then
(1' = i*r*a' = di*1J;
but we know that (1' is not exact.
inlegmlion
It is a worthwhile exercise to compute by brute force that for n = 2, for n = 3,
265
 dx x dy  y dx r *(J' = x dy + y = = de x 2 y2 v2 r*(J' =
1
Since we will actually need to know r*(11 in general, we evaluate it in another way:
= ;;; [x dy A dz  y dx A dz + z dx A dy].
x dy A dz  y dx A dz + z dx A dy (x 2 + y 2 + z 2 ) 3 /2
7. LEMMA. If (J is the form on IF!.n defined by
(J =
L (I);I x; dx l A · · · A d;! A · · · A dxn,
i= ]
n
and (1' is the restriction i*a of (J to snl) then
r*(J'(p) =
L i=]
(J(p) . I pln
So
r*(J' = �
VII
"(I);IX; dxl A · · · A d;! A · · · A dxn. 
n
PROOF. At any point p E IF!.n {OJ, the tangent space IF!.np is spanned by Pp and the vectors up in the tangent space of the sphere sn l (l pl) of radius I pl . So it suffices to check that both sides of (*) give the same result when applied to 11  1 vectors each of which is one of these two sorts. Now Pp is the tangent vector of a curve y lying along the straight line through 0 and p; this curve is taken to the single point r(p) by r, so r (pp ) = o. On the other hand,
*
0(,> (" , ( ."" . . . , (�,),) = de.
(,fJ
= o.
So it suffices to apply both sides of (*) to vectors in the tangent space of snl (Ipl). Thus (problem 15), it suffices to show that for such vectors up we
266
Chapter 8
But this is almost obvious: since the vector vp is the tangent vector of a circle y lying in sn l (Ipl), and the curve .. 0 y lies in snl and goes I II pi as far in the Same time . •:.
J: B
and define
8. COROLLARY (INTEGRATION IN "POLAR COORDINATES"). Let > JR, where B = {p E JRn : Ipl :0: I),
g:
snl
>
JR b,'
g
(p) =
lo' unl J(u · p) duo
I]
>
[ J = [ J dx ' /\ . . . /\ dxn = [ g(J' . }SIII JB JB PROOF. Consider snl X [0, I] and the two projections
1f l : S,, l 1f2 : snl
x [0,
X
Then
[0, J]
>
sn l [0, I].
[0, I]
Let us use the abbreviation
I
....!2
�
1'" c=::>
snl
From this it is easy to see that if we define h : snl
If (y, U ) is a coordinate system on sn l ) \"ith a corresponding coordinate sys tem ()' , I) = (y 0 1f , 1f2) on snl X [0, I], and (J' = Ci dy ' /\ . . . /\ dyn ' , then l clearly , ' (J A dl = Ix 0 1f1 dy ' /\ . . . /\ dyn /\ dl.
h(p, u) = unl J(u . p),
then
X
[0, J]
>
JR by
Now we can denne a diffeomorphism ¢ : B  {OJ ¢ (p) =
}SIII g(J
[
'
=
/w' A dl . ( _ I) nI [ 1s"1 x[O,I J
>
snl
(,, (p), v(p»
=
(pll pl, Ipl).
X
(0, I] by
bzle.gralian
267
Then
¢*«J' A dt) = ¢*(1C , *(J' /\ 1Cz*dt) = ¢*1f ] *a' 1\ ¢*1f2*dt = (1CI O ¢)*(J' /\ (1C2 0 ¢)*dt = r *a' 1\ v*dt n Xi i I n i = ;;n L (  I ) i  J )C i dx I /\ · · · /\ dx /\ · · · /\ dxn /\ L ; dx ,= 1 1=] ( _ I )n ' n . 2 ' . . . = '"'(X') dx /\ /\ dx n Vn + 1 � i=]

(
)
Hence
¢*(lw ' A dt) = (h ¢)¢*«J ' A dt)
( 1 )"' = vn 1 J . ;,;::r dx 1
= °
( _ I ) n ' J dx ' /\ . . . /\ dxn.
1\ . . . 1\
dxn
So,
iB
[ J dx ' /\ . . . /\ dxn = ( _ 1 )"1 [
= ( _ I ) n' [
iB{a)
¢*(h(J' A dt) }l(J' A dt
(This last step requires some justification, which should be supplied by the reader, since the forms involved do not have compact support on the mani folds B  {OJ and sn ' x (0, I] where they are defined.) .:. We are about ready to compute H k (M) in a few more cases. We are going to reduce our calculations to calculations within coordinate neighborhoods, which are submanifolds of M, but not compact. It is therefore necessary to introduce another collection of vector spaces, which are interesting in their own right.
= L,,, g(J'.
)snIx{o,l}
268
Chapter 8
Z� (M) is the vector space of closed kforms with compact support, and B�(M) is the vector space of all kforms d� where � is a (k  I )form with compact support. Of course, if M is compact, then H;(M) H k (M). Notice that B� (M) is not the same as the set of all exact kforms with compact support. For example, on IRn, if J 2: 0 is a function with compact support, and J > 0
where
=
The de Rham cohomology vector spaces with compact supports defined as H;(M) = Z;(M)/B�(M),
Ii; (M) are
at some point, then
w = J dx'
/\ . . . /\ dxn
is exact (every closed form on IRn is) and has compact support, but w is not d� for any form � with compact support. Indeed, if w = d� where � has compact support, then by Stokes' Theorem
This example shows that H;(IRn) i' 0, and a similar argument shows that if M is any orientable manifold, then H;(M) i' o . We are now going to show that for any connected orientable manifold M we actually have
L w L d� faR" �
= =
= o.
H;(M) "" R
This means that if we choose a fixed w with f w i' 0, then for any nform w' with compact support there is a real number a such that w' aw is exact. The number a can be described easily: if
M
w'
then so


[ [ ' [ 1M w' 1M aw 1M d�
=
GW
=
d1].
=
0,
H;(M) '" IR is equivalent to the assertion that
[w] >>
the problem, of course, is showing that 1] exists. Notice that the assertion that
is an isomorphism of H;(M) with �, i.e.) to the assertion that a dosed form w with compact support is the differential of another form with compact support if f w = o.
Lw
M
then it is true for ffi. . For simplicity assume that support w e {p E JR" : ipi < I ) .c has compact support.J.n. 1"' 1(1 . �(p) " = {. p) dl xi dx' /\ . so 1 is a (I) The theorem is true for M = IF!. then H. Step I . oo).I)manifolds. /\ dx" be an nform with compact support on JR" such that III" w = o.I )form � o n JR " such that w = d�. .' . Therefore c] = Cz = c and we have w = d(J where 1 . N) and a constant Cz on (N. Since support w is compact. (2) If the theorem is true for (n . . Let w = 1 dx' /\ . Moreover. .(_ J)i' (10 . Let w be a I form on JR with compact support such that III w = o.. .c) Step 2. We wiII establish the theorem in three steps: = f N constant c] N on ( 00. then it is true for any connected oriented nmanifoId. We know that there is an ( n . There is some function 1 (not necessarily with compact support) such that w = dJ.i /\ .(M) '" IF!. /\ dx". /\ . ) . in particular for S". we have an explicit formula for �. . from Problem 723. In fact. If M is a connected orientabIe nmanifoId.269 9. .. (3) If the theorem is true for JR". THEOREM. dl 0 outside some interval [N. N]. PROOF.
X Xi by Lemma 7. . � ) dU ) . r*(J' = r*(g(J').270 Using the substitution Chapter 8 Define On the set or u = I pl l �(p) = ( ripi un1 f (u I pl ) dU) �n I pl Jo n (_J )il dx1 /\ .d(hr*A» : .2)form A on snl . /\ dxn L i=] = (t i r l f (u · � ) dU) .1 f ( u .II JB w o. I] be any Coo function with h = I on A and h = 0 in a neighborhood of o. Let h : [0. Then hr* A is a Coo form on and � = r*(dA) = d(r*A). A �(p) = (10' u n .I )form g(J' on 1s"1 r � = (g 0 r ) . by the hypothe. . . r f dx1 /\ . g(J' = dA for some Hence (11 . by Corollary 8 we have for the (11 . !!. g(J' = snl. r*(J'(p).. r*(J'( p) II g: snl IR by g(p) = 10' u n1 f(u· p) duo A = {p IRn : I pi f = 0.. /\ dxn = = r JR. . Thus. IR" > IRn w = d� = d(� . II this becomes .is for Stefl 2. . > E > I I we have so on we have Moreover. /\ dxi /\ . .
:. .CrWr. . since on A we have � . In other words. Using the connectedness of M.oll 27 1 the form � . . It obviously suffices to find c. and �.gral.l = d1Jr : where all �.) = o.CZWI = dI}2 w' . we want to show that there is a number c and a form � with compact support such that w' = cw + d�. 0. we can write where each ¢iW' has compact support contained in some open set Vi C M with U. Since we are assuming the theorem for ffi. From this we clearly obtain the desired result. If w' is any other nform with compact support. with Vi n Vi+ 1 =j:. Choose an IIform w such that JM w i' 0 and w has compact support contained in an open set U C M. Step 3.d(hr*).l1.w' = ("jW + d1Ji' for each i. we can assume w' has support contained in some open V C M which is diffeomorphic to IRn.n we h ave WI . with ¢.C1 W = dl}l Wz . . Using a partition of unity. have compact support (C V.) = � . with U diffeomorphic to IRn . it is easy to see that there is a sequence of open sets U = V" .lille. Vr = V diffeomorphic to ffi.).) has compact support. .d(r*). Choose forms Wi with support Wi C Vi n Vi+ l and JVi Wi =j:. o. diff eomorphic to IRn .d(hr*).
Xi) where each Xi 0 Xi+ l . such that fu w i' 0 (this integral makes sense. 1 0. .n orientation preserving. Choose an Ilform W with compact support contained in an open set U diff eomorphic to JRn .272 Chapter 8 The method used in the last step can be used to derive another result.1 is orientation preserving. then Hn(M) = O. : PROOF Consider first an Ilform w with support contained in a coordinate neighborhood U which is diffeomorphic to JRn . then PROOF. Consider a sequence of coordinate systems ( Vi. Now if M is unorientable. U4. If M is any connected nonorientable nmanifold. • . JVI [ Wi /J� [ Wi . Taking w ' = W. there is such a sequence where Vr = VI but Xr OX1 . •. since U is orientable).1 is orientation rel1ersillg. . THEOREM. O . we have W = cw + d� for for c  >0 J so (c  We can also compute Hn(M) for noncompact M. Consequently. Wi > 0. I t follows that Wi = cw + d� where c > O. we have Iv. If M is a connected noncompact nmanifold (orientable or not). UJ . Since M is not compact. the numbers Ci = H. U2. I I .(M) = o. there is an infinite sequence U = UJ . It obviously suffices to show that w = d� for some form � with compact support.l are positive. using the orientation of Vi which makes Xi : Vi 7 ffi. i)w = d� c =j:. THEOREM. then also !Vi+1 Wi > O. Choose the forms Wi in Step 3 so that.
. has support contained in V. Cj+lWi+l + d1]i+l i 2: J . The cover !9 = {V) is then locally finite.+. d�. . c2d�3 + c.Integra/ion 273 of such coordinate neighborhoods such that V.. V3 . and such that the sequence is eventually in the complement of any compact set.. V2 . Let {¢u) be a partition of unity subordinate to !9. d�2 + c. C V. + C.+.+2 Hence . . i' o. but the sequence is still eventually outside of any compact set). with compact support . There are constants C. V. . w. V.c2d�3 + C. whose union is all of M (repetitions are allowed. where �. n V. d�2 + C. and V. and forms �. + d�. = w = d�. C2C3W3 = where the right side makes sense since the V. If W is an nform 011 M. may intersect several Vj for j < i. are eventually outside of any compact set. . + c.C2C3d�4 + . . . then for each V.w. C2W2 = d�. Now it can be shown (Problem 20) that there is actually such a sequence V. we have W d�. U U U . . d�2 + c. + C. such that w Wi Then = = c. we have seen that ¢u. Since any point P E M is eventually in the complement of the V. Now choose nforms Wi with compact support contained in Vj n Uj+h such that Ju W.w = d�. i' 0. + C.'s.
{OJ) i' 0. 1* also takes B k (N) to B k (M). = . then = cp when I( M. If N is also disconnected. then H. '" IR. Then I*(c) = c o l is also a constant function.. since we will eventually improve it. If N is connected.274 Chapter 8 SUMMAR OF RESULTS Y (I) For IRn we have (2) If M is a connected nmanifold.EA E9 IRp > E9 IR. then 1* : HO(N) > HO(M) is just the identity map under the natural identification of HO(N) and HO(M) with IR. In order to proceed further with our computations we need to examine the behavior of the de Rham cohomology vector spaces under Coo maps I: M > N.1 (IRn . consider the case k o. . If w is a closed kform on N. I f M i s disconnected. This shows that 1* induces a map = also denoted by 1*: /* : Hk (N) > H k ( M ) . with Cith component equal to c. For example. Ci E A . We also know that Hn. so 1* takes Z k (N) to Z k (M). On the other hand.. then HO(N) is just the collection of constant functions c : N > IR. where c� /* : the map 1* takes C E IR into the element of EB IR. since I*(d�) = d(J*�).(M) '" {� Hn Hn(M) '" { c 0 HO(M) '" IR if M is orientable if M is nonorientable if M is compact if M is not compact. then HO( M ) i s isomorphic to the direct sum E9 IR. with components M. ctEA PEB takes the element {cp) of EBpEB IRp to {c�}. but we have not listed this result. If M is connected. with components Np. where each IR. fJ E B.) C Np. then I*w is also closed (dl*w I*dw = 0).
J. Then there is some deg which depends only o n i s called the degree o f I The number and N are not compact.Integral. THEOREM. Until one sees the proof of the next theorem.l. and the only one we are presently in a position to look at. let For each P E sign J: M.l.P for and Vq for Nq) is orientation reversing. it is almost unbelievable that this number is alwqys an integer. v). Let q E N be a regular value of I (q). ) and (N. > N be a proper map between two connected 1 2. J M) and JR (and similarly for N) it J. but is proper (the inverse image of any compact If set is compact). form we have w fM w w 1M J*wo = a · L woo with fN Wo Lw i' o. and a number deg such that for all forms on N with compact support. so we really want to compare M J*: H" (N) L J*w > H"(M) and Wo for an nform on N.p: Mp J pJ= J !*p w L J*w = (deg /) Lw { if + Nq is orientation preserving (using the orientations J. Let oriented nmanifolds ( J. Choose one number such that w a is an isomorphism of H " ( Since r> follows that for eVeI). if Mp . L J*w = a . There is when no natural way to make H"(M) isomorphic to JR. then we have a map M a = J.on 275 A more interesting case. is the map and N are both compact connected oriented nmanifolds. L w. M J1 !.
. y) around q such that J' (V ) = V. ii J*w k U · · · U Vk !.' (q) PROOF Notice first that regular values exist. first choose a compact neighborhood W of q. Moreover. I . . We want to choose a coordinate system (V. by Sard's Theorem. xi i around Pi such that all points in Vi are regular values of J. So Since J is a diff eomorphism from each Vi to V we have L J* w = {. . •:. and let deg J = U W' C M be the compact set w' = J' ( W) . Iv w if J is orientation reversing. redefine Vi to be Vi n J' ( V). and the Vi are disjoint. pd· Choose coordinate systems (Vi . so the sum above is a finite sum. · · · Vk . . since it is compact and consists of isolated points. u . PEj. To do this. This ensures that J' (V) c V. Then J(W') is a closed set which does not contain q.(V. . Let J' (q ) = { p " .I (q ) is finite. . . Now choose w on N to be w = g dy ' /\ . Then support J* w C V. . U U U [ J* w = [ w if J is orientation preserving 11' lUi = Since J is orientation preserving [or reversing] precisely when signp J = I lor .I] tIllS proves the theorem . Vk . We can therefore choose V C W .276 Then Chapter 8 L signp J (= 0 if r ' (p) = 0). · · ·UVk • Finally.J(W'). /\ dyn where g 2: 0 has compact support contained in V.
we conclude that deg A = (. We can HOW prove a more general result.*(H*w) . Then J = H o io g = H o i) .I )form /w on M such that i l*W .io*w = d(Iw) + /(dw). J] we defined a (k . Since A .io*(H*w) = d(IH*w) + /(dH*w) = d(IH*w) + 0.1 ) " .gmlioll 277 We can draw an interesting conclusion from this result. g: M H(p. but we need to intro duce another important concept first.inle. Two functions J. I ] > N the map H is called a (smooth) homotopy between J and g. N are smoothly homotopic. J* PROOF By assumption. Notice that M is smoothly contractible to a point po E M if and only if the identity map of M is homotopic to the constant map po . Any element of H k (N) i s the equivalence class [w] of some closed kform w on N.(H 0 io)*w = i. J] = g* . O) = J(p) H(p. I ) = g(p) for all p E M. then the maps /* : H k (N) H k (M) g* : Hk (N) H k (M) > > > > are equal. there is a smooth map H : M x [0. Recall that for every kform w on M x [0. we compute the degree of the "antipodal map" A : S" > S" defined by A (p) = po . We used this fact to show that all closed forms on a smoothly contractible man ifold are exact. •.l (p) consists of just one point.J*w = (H o i) *w . But this means that g*([w]) = J*([w]) : . We have already seen that A is orientation preserving or reversing at all points. If f. g : M > N between two Coo manifolds are called (smoothly) homotopic if there is a smooth function As an immediate application of the theorem. .1. depending on whether II is odd or even. THEOREM. N with g*w . H: M with x [0. 1 3.
then deg j = deg g.J)".{OJ is the inclusion. X"+l ) E S" we define = For n odd we can explicitly construct a nowhere zero vector field on S" . If M and N are compact oriented nmanifolds and the maps j. there is a unique great semicircle Yp from p to A (p) = p whose tangent vector at p is a multiple of X(p).{OJ > S". then r 0 i : sn. If vector field on S" . I ) = Yp (I ) : .l is the identity of S". For another application of TIleorem 13. Define S" is ( . consider the retraction r: JR" . this is perpendicular to p = (X l . g : M > N are homotopic.278 Chapter 8 1 5 .l > JR" . (On S l this gives the standard picture. Since the identity map has degree ) . I . 1 4. . . COROLLARY.) The vector field on S" can then be used to give a homotopy between A and the identity map. . X2 . Il is even. then there does not exist a nowhere zero PROOF We have already seen that the degree of the antipodal map A : S" > A is not homotopic to the identity for n even. . X"+ l ). . But if there is a nowhere zero vector field on S". p (Xl . For each p.l If i: S". r (p) = p/ ipi. . then we can construct a homotopy between A and the identity map as follows. and therefore in S"p .l .l 7 sn. For H (p. . . COROLLARY. •.l .
{OJ. M x {OJ c M x jR' is clearly a deformation retraction of M x jRt So Hk(M) "" Hk(M x jRl ) .I)r(p) E jRn . Whenever I" is a deformation retraction. we can define the homotopy H by . Thus In particular. We need one further observation. We are now going to compute Hk (jRn . of course.{OJ) for all k.integration 279 The map is.I ) .{OJ i 0 I" ( p) = p/ i p i H(p.{OJ > jR" . identity of Hk(jR"  {OJ) So i* and r* are inverses of each other.// ' i 0 1" : jRn . A generator of Hn l (jRn .{OJ) is the closed form r*(1'.{OJ. = "r( p) (t=O) p(/=1 1 A retraction with this property is called a deformation retraction. The manifold for all I . we have and 1" ' 0 i' = (i 0 1" ) ' i ' 0 1" ' = ( I" 0 i)' = = identity of Hk( sn. Thus.{OJ) "" jR . the maps (r o il' and (i 0 1" ) ' are the identity. we have Hnl (jRn . l ) Ip + ( J . not the identity. for the case of s nI C jRn . but it is homotopic to the identity.
O) B w. for on A on B A w = d(J . We claim H I (w.{ (O.O) x [O. THEOREM.1) Since A and B are both starshaped (with respect to the points (0.4. I) A = w.c) B) w = d(J and JA .0.I we have H k (w.{O) = 0. If w is a closed I form on ffi. For ° < k < /1 . 0) = x x (00. Th us w is exact.O. The first case where there is anything to prove is 3 .I is a constant on A n B.3 .3 . on A n B. using A = w.I).280 pter 8 Cha 1 6. Let w be a closed I form on w.3 Let A and B be the open sets (0. 0. Induction on n. there are Oforms JA and Is on A and B with w = dJA l w=dB Now d(JA on A on B. . . oo)}.0.3 .{OJ) = H k (snl) = 0.{(O. x w" B) I =° and A n B = [ w.c = Is on A n B. 0]) [0.{(O.0. If w is a closed 2form on w. respectively). 0]) B = w. then we obtain Iforms �A and �B with w = d�A w on A = d�B on B.n . 00») .4 . .4 . I) and (0. there is a similar argument. 0.4 .2 .{(O. 0) x (00 . n= PROOF. (0.{all r B so clearly JA .
<PBj for the cover { A .d(<PBA» = d(�B + d(<PAA» so w is exact.d<PB /\ A = �A .<PB dA .�B) = 0 and SO �A . w = d�A = d(�A . The general inductive step is similar. on A n B =) =0 Now.{OJ: <PA + <PB d<PA + d<PB support <PA support<PB CA C B. . = �A + (<PA . on A n B on A . •:. if <PBA denotes { <PBA O and similarly for <PAA. and �B + d(<PAA) on B. To circumvent this difficulty.integration 281 Now d(�A . Clearly. = d�B /\ A U B by letting it be �A .�B = dA for some Oform A on A n B. Unlike the previous case.(A n B). Bj of IR J .{OJ = A on A.d(<PBA) on A on B. since this is not defined on A . then <PBA <PA A On A n B we have �A .dA + d(<PAA) is a COO form on A is a Coo form on B.J) dA + d<PA = �B + d(<PAA). So we can define a Coo form on IR n .d A. note that there is a partition of unity {<PA .d(<PBA) = �A . we cannot simply consider �A .
. Let / : w. Let w be a kform on w..n) = o.n.n .. where 2B denotes the ball of twice the radius of B.n > [0... PROOF The proof that H�(w. Let B be a dosed ball containing support w. which we will need in Chap ter 1 1.I we have from Theorem 16 that � = dA for SOme (k .2)form A on A . •: ..n with compact support.{OJ and k  I < 11 . 0 < k < n. THEOREM. Then d(JA) makes sense on all of w. Since A is diffeomorphic to O = w = dry w.. I] be a Coo function with / = 0 in a neighborhood of B and / = I on w.2B.. For 0 � k < 11 we have H: (w.n) = 0 is left to the reader. We know that w = d� for some (k .B we have d� = O.n .I )form � on w. the form � ..282 Chapter 8 We end this chapter with one more calculation. 1 7 . Then on A = w.d(JA» .d (/A) dearly has compact support contained in 2B .n and w = d� = d(� .n .
< In} be two partitions of For each i = I. s. b] ei 111 S s [a.= ] S [a. b]. and define is an ntuple = similarly. . P. b] > IR be bounded. b]. S. .=1 S(J. [a. Let [a. .Ii.�n) �i M. t] shaded lengths . Pl._I . S J on J J 1I). this sup or inf is called the Darboux integral of J on We call J Riemann integrable if . letm Ii] = length of the limit is called the Riemann integral of (a) We can define even if is not bounded. .]'S cmltained in [Ii.sum of lengths of all [s. P) S(j. let = on [Ii_I. Show however. {so { [Ii. We call J Darboux integrable if the sup of all L (J. A choice for = with E [ I We define the "lower sum". P) equals the inf of all U (J. P. b]. li]. For a partition = { o < . The Riema1l1l integral UeJJUS the Darboux integral.lil ) . b]. . "upper sum". .I .�) = L J(�i)(I' IiI)· Clearly L(J. that cannot exist if is unbounded. P � n L(J..add up to e. ._] . (d) Let J M on Let P = < .�) (b) If J is continuous on then J is Riemann and Darboux integrable on and the two integrals are equal.�) [a. < s } and Q = Io < . .b]. < be the inf of of [a. P) = Ll11i(J) ' (Ii IiI) i=l J U(J.] which are contained in [li_] .i� o S(J. [a. b]. and "Riemann sum" for a partition and choice by J: P I 111i mi(J) In} P � (�I" I"i] . (Use uniform continuity of J on (c) If J is Riemann integrable on then J is Darboux integrable on and the two integrals are equal.�) U (J. P) = L Mi(J) ' (Ii . Mi(f) Ii].I . P. P. I � I I I. [a. . [S. S(J. [a. [a. b]. P). . I . b].lnlegralion 283 PROBLEMS I. I ti_l . .) [a. n. b].
Q) + iL(M . (h) Show that fe f dx + g dy. and CR. (g) (Osgood's Theorem).)e. =l :s U(J. (a) Show that there is a singular 2cube c : [0. (b) If c : [0.n (t ) = () = (cos 21f1. Q) + (M .b]) H ). R sin2n1fI).�' for P. P) = inf{ U (J.CI . i=l There is a similar result for lower sums. and deduce Darboux's Theorem: ). defined as a limit of sums.m) L ei.284 Chapter 8 Show that.il� o L(J. b lim L f(�i)g(�'i)(li . I] JR2 . b]. sin 2Jr/ ) on [0. Let f and g be integrable on [a. )1 :s MIJ(�'i )f(�.n : [0. I I c· de.11 . Show that fOJ choices �. show that there is some such that is a boundary in JR2 . then f is Riemann integrable on [a.n .1 ) = [ fg· Ja i=l Hint: If l g l :s M on [a . IF JR2 .11 = > CR"n  n > = C . b]. (f) If f is Darboux integrable on [a. b].{OJ such that ac. CR2.{OJ. Compute fe de = f[o. I] n JR2 . let CR. ° as I PI I 0.)g(�'.� U(J. For 11 an integer. b]. (R cos 2 1fl. (e) Show that .M. then U(J. Q) : Q a partition of [a. P) :s U(J. li].{OJ is any cUlve with c(O) c(I). I]. It is called the winding number of C around 0. if Mi denotes the sup of f on [If_I . > R > 0. )I. where C I 3. equals > > H O n UPU+O 2.{OJ be defined by (c) Show that n is unique.)f(�i)g(�'. Q) : Q a partition of [a. b]). then 1J(�'. P) = sup{ L (J.£7= 1 e.
I ) a.I .. 1] > <C by cR. a.x' . ! . Let /: <C > <C be a polynomial. As before. +an .c. . Let !J. I (I)] n .c = C o a.CR. then cR.. x'. Hint: Note that CR".(x) = (x l . Define cR. n C JRn be the set of all x E JRn such that A singular nsimplex in M is a Coo function c : 6.) .'£7. where n :0: I .J)'a. define (a) Describe geometrically the images (b) Show that a2 = 0. ! : [0. . n M. n (t) = [CR. I ' = 5. O. there are some advantages in using singular simplexes. /(2) zn +a l znI + . . ! = / O CR . . x l . . and an nchain is a formal sum of singular nsimplexes.I ) in " n . Some approaches to integration use singular simplexes instead of singular cubes. and write 4. as indicated in the next Problem. xn . Hint: If /(z) i' ° for all z with Izl :S R. aC = L (.+ . .. . xn . .{OJ.Integration 285 n l /(2) = Zn ( 1 + . . · + . x' ] . .I ) a. . : " n I > " n by 7 ao(x) = ([I . a a (a) Show that if R is large enough. n i s the boundary of a chain in <C . Define a. . . let r : " n > JRn be the inclusion map.(" n . Then we define and for singular nsimplexes c. then CR . . i=O 0 < i ::: 11. . ! is a boundary. zn z (b) Show that /(z) = ° for some z E <C ("Fundamental Theorem of Algebra"). ! .co. Although Stokes' Theorem becomes more complicated. .
) (d) Define Ie w for any kchain C in M and kform w on M.C2 is a boundary. /\ dxn Jalll is an (n . ? . Show that c is not the boundary of any sum u of singular 2cubes. (b) Let c : [0.I )form w. 6. using either simplexes or cubes. what can be said about c for chains c in the obvious way. define c: "' k +1 > IRn by IV!orcover.c. I] > IR 2 be a closed curve. for 0 :0 :0 I . I]) is a point. k 7 c(x) = I . Every x E "' k + 1 can be written as ' IX . c(x ' ). For any singular k simplex c : 6. and x ' E ao("' kl .a.. and prove that 1 /" dw = [ w.J )form on IRn . a. . dxi /\ . x' is unique except when t = O. We then define (c) Show that we do have c = au + c' where c' is degenerate. show that CJ . Ozapter 8 /\ (Imitate the proof for cubes. L. (a) Show that a c = 0 implies that c = at. • .286 (c) Show that if w = J dxl then /\ . [ dw = [ w lc Jac for any (k . that is. IRn . Hint: If au = L. (d) If C I (O) = C2 ( 0) and CI (l) = C2 (1). . c'([O.
(d) If M is simplyconnected and P E M. then M is simplyconnected. any space M (not necessarily a manifold) is called simplyconnected ifit is connected and any continuous I : S l M is (continuously) contractible to a point. (Converse of nV UV V Problem 724. partition S l into s a finite number of intervals each of which is taken into either U or V. > dl. A manifold M is called simplyconnected if M is connected and if every smooth map I : S l M is smoothly contractible to a point.U is not con nected. Suppose that Ie W '" 0 for every closed 8. then any smooth map I : Sl > M is smoothly contractible to p.] (a) If M is smoothly contractible to a point. Hint: Show that a smooth I : Sl > s n is not onto.U. It is > 1 w = l(c(1» . (f) If M is simplyconnected. (b) A bounded connected open set U C IR 2 is smoothly contractible to a point if and only if it is simplyconnected.) Hint: If p is in a bounded component of IR 2 . (This gives another proof that n is simplyconnected for n > J . Show that U is not smoothly contractible to a point. [Actually. (b) Sl is not simplyconnected. (c) sn is simplyconnected for n > J . (a) Let U C IR 2 be a bounded open set such that IR 2 .) 9.Integration 287 curve c in M. then H I (M) = O. Show that w is exact. (e) If M = U where U and are simplyconnected open subsets with U connected.l(c(O» . Hint: If we do have w = curve c we have 7. Let w be a J form on a manifold M. (c) This is false for open subsets of IR 3 . (See Problem 7. then for any not hard to show that for a manifold we may insert "smooth" at both places. show that there is a curve in U which " surrounds" p. then M is simplyconnected.) Hint: Given I : Sl > M.
and suppose that w I I . there are only finitely many " which are nonzero on support ¢ . " . (a) This implies that L¢ E¢> f ¢ .. "' to different values. Let <I> and lj! be two partitions of unity by functions with compact support. Find two partitions of unity <I> and such that L¢E ¢> f ¢ . · w . define geometric objects corresponding to odd relative tensors of type (�) and weight w (w any real number).. I dx converge absolutely Il. w = x dy. using of its boundary. Following Problem 712.I )"/n and support I C Un A n . (Note that for each ¢ . ¢E ¢> cpe¢ cpe¢ l/re\ll cpe¢ l/fe\ll > 1 0. and show the same result with w replaced by Iwl. Let be an nform on an oriented manifold M n . I dx and L E 'l' fIl. Show that w. (e) Examine a partition of unity for (0. Let I : IR IR be a Coo function with fAn 1 = ( . and that M L lM ¢ . (a) Let M be E IR 2 : < J }.) 0) Similarly. M (b) Show that L 1M ¢ .y)1 (No computations needednote that equality would hold if we had the entire boundary. and let o w . (d) Let A n C (n.) (c) Show that L"' E 'l' f ." laM Problem 1 0. even though both sides make sense. find a counterexample to Stokes' Theorem when M = (0. Iwl < 00. J) by functions with compact support to see just why the proof of Stokes' Theorem breaks down in this case. w = L L 1M " + w. together with a proper portion lj! « x.288 Chapter 8 L 1M ¢ . We define this common sum to be fM w. w converges absolutely.. · Iwl < 00. 1 2.y) l(x. L dw . w = L lM . J ) and i s a Oform whose support i s not compact. n + J) be closed sets.
1 4. . but this only accounts for a factor of 1 /Ipl" . " (p. Use the formula for r*dxl (problem 41) to compute r*(J'. Wm are positively oriented in Mp and Nq. .q) '" Mp Ell Nq if VI .1 vectors 1 6. (Note tbat the map i o r : JR" . Show that de is 0 at some point. . · · · . . 1 7. since there are n . on M and N. show that (b) If h : M x N > [ 1f1*W /\ 1fz*� = [ W · [ �.{OJ > JR" . respectively. Wm is positively oriented in (M x N )(p.d· 1 5.) r * (1' = r * i*(1 = (i 0 r ) *O". then where g(p) = (c) Every (m + n)form on M x IN h (p. . )�.aMI. respectively. VI . 1M IN JMxN JR is Coo.{OJ is just r. q). . . . and let w and � be an nform and an mform with compact support. and e is an (n . We will orient M x N by agreeing that V I . " " Un. ) = q 1+ h (p. N is h 1fI*W /\ 1fz*� for some w and �. . V" and W I . Let MI . vn. Suppose M is a compact orientable nmanifold (witb no boundary). . Account for the factor I/Ipl" in Lemma 7 (we have r* (vp) = (J /lpl)vr (p) . Show that for any closed (n . .I . WI . . .Integration 289 13.{OJ. If 1f1 : M x N > M or N is projection on the ith factor. . (a) Let M" and Nm be oriented manifolds. .J )form on M.J )form w on MI. M2 C JR" be compact ndimensional manifoldswithboundary with M2 c MI . considered as a map into JR" .
. zl. For this n 1 9.{OJ . Let WI . (c) Let M C JRn{O) be a compact (11 . C (Ml C(Ml n S' D·. .. (a) Let p E JR" . M Show that [ r *(J' = [ r *(J' . y . .I )manifoldwithboundary which is the union of segments of rays through O .�l E JR3 except those with x = 0. The latter integral is the measure of the solid angle sub tended by reason we often denote r * (J' by de . For aJJ (x. 1M 1c(M)ns2 M. r *a' (v. Show that JM r *(J' = O. and let C(M) = {Ap : p E M . y. �) to be the angle between the positive zaxis and the ray from 0 through (x. 0]. y = 0.J )manifoldwithboundary which inter sects every ray through 0 at most once. .{OJ be a compact (n . y. we define ¢(x. .290 Chapter 8 1 8. Wn2) = o. W I . Z E (00. . " " Wn2 some A E IR. A ::c OJ . Show that E JR"p and let v E JR"p be (Ap)p for (b) Let M C JR" .
z) arctan y/x. z) E S 2 (r).z) (x. (b) If v (p ) = Ipl. and e is considered as a function on ]R 3 . then d¢(v) = 1 .z) = arctan( vlx 2 + y 2 /z) (with appropriate conventions). then de(wp) = Y X2 + / (d) If e and ¢ are taken to mean the restrictions of e and of] S 2. (the minus sign comes from the orientation). 00) or with x = 0. e(x. Z E (00.y. y . y) (a) ¢(x. X E [0. (e) Conclude that (J' = d( .. y. y. ¢ to [certain portions h: S 2 > ]R is h(x. y. then where . y = 0. If w points along a meridian through p = (x. y.cos ¢ de ) . ... then (v. z) = vix2 + y2 (J' = h de /\ d¢..Integration 291 (x. e. ¢) is a coordinate system on the set of all points (x. z) in ]R3 except those with y = 0. (c) If v is a longitudinal unit tangent vector on the sphere S 2 (r) of radius r. ) . 0].
. = and Show Hint: Choose w to be a real (d) Conclude that av /'(x+iy) = aau (x. U U U .. for the form (J de IR3 is just 1C*de. S 2(1 p l). Vi r*de de .y).. {OJ. express 20. ( + nl l. so that is the projection.292 (f) Let {OJ on Show that If > for the form (Jzapter 8 > IR2 1"2 : IR2 1C : IR3 IRde2 Sl be the retraction. +an/'(. (b) Let J Show that we have i where varies over complex numbers. ih. given by = is proper � J).y).y) x aav (x. I/'(x iy )12 DI(x. Show that if N is connected. Let > Nn be a proper map between oriented IImanifolds such that p 7 Nf(p} is orientation preserving whenever p is a regular point. (a) Show that a polynomial map ce > ce. and the sequence is 2 1 .)a "n2I: . 0 on {OJ in terms of on IR2 r2*de = de . l w I(x+i ) u(x. dBn_1 IRn1 dBn IRn Vi n VjVI V2 V3 . with eventually outside of any compact set.. or else all points are critical points of J. v dB3 r*(J' d(cos(¢ r) de) = d(cos¢de). . ·+a _ . 22.y)+iv(x. Prove that a connected manifold is the union the are coordinate neighborhoods.y) == for realvalued functions I(z) zn+aI Zn1 + I'(z) u v. = J* : I: Mn M (e) Write that .y)+i ax (x. where i' 0.) = IIZn111 +(11 . then the form on [part of] on [part of] Use this to show that = r (g) Also prove this directly by using the result in part (c). and then to be == + det . de = r2*i*(J. then either 1 is onto N.To[/(" + w) yl(z)]lw.y) i ay (x.. . and the fact that p vr(p)/lpl for tangent to (h) Conclude that = = * (v ) = (i) Similarly. y au h.
Conclude that w is con stant on each component of JR" .M. For p E JR" . not using Problem 2 1 (which relies on many theorems of this chapter). Let M" .{p) > L: be the obvious retraction.M.I )plane. Let I"p : JR" . (a) Show that this definition agrees with that in Problem 3.I )sphere L: around p such that all points inside L: are in JR" .Integmtion 293 where J' is defined in part (b). Define the winding number w (p) of M around p to be the degree of rp lM. Let p and q be points . give another proof of the Fundamental Theorem of Al gebra. (b) Show that this definition does not depend on the choice of L:. (f) There is a still simpler argument. (c) Show that w is constant in a neighborhood of p. Show directly that if I : M > N is proper. then the number of points in I.I (a) is a locally constant function on the set of regular values of f Show that this set is connected for a polynomial I: ce > ce. (d) Suppose M contains a portion A of an (n . 23.M. while DI is the linear transformation defined for any differentiable I: JR 2 > JR2 (e) Using Problem 21.I c JR" be a compact oriented manifold. choose an (11 . and conclude that I takes on all values.
x w(q) w(p) M > N be smoothly denote the (finite) number 1 g.I .) (b) Show. Let M and N be compact nmanifolds. and let homotopic. 25. The next few Problems show how to prove the same result even if M is not orientable. Show that = ± 1.294 Chapter 8 (a) Let of points in is a compact I manifoldwithboundary.t) g(p) t g t.M has at least 2 com ponents. g: H: N. For two maps M > smoothly homotopic to = I. 1] > be a regular value of Show that Let (mod 2).A and which does not take any point of A onto the point in the figure. and 0 elsewhere. I] > N such that for in a neighborhood of 0. H.) Show that each 1+ is 1I diffeomorphism.) ¢(H(t. but on opposite sides. x) . for in a neighborhood of I . (Show that I"q l M is homotopic to a map which equals 'p l M on M . If / is smoothly homotopic to by a smooth homotopy such that 1+ is a diffeomorphism for each we say that is smoothly isotopic to g. = H': M x [0.) (e) Show that. = == x H(t.t) /(p) t H'( p. that this result holds so long as 'I is a regular value of both and I. (This is one place \·vhere we use the stronger form of Sard's Theorem. q E N I (q).x» · p x at H(O. 1 H p (a) Being smoothly isotopic is an equivalence relation. by a smooth homotopy M x [0. and leaves all points outside the unit ball fixed. let IR x IRn > IRn satisfy ¢: (Each solution is defined for all by TheOI·em 56. more gencrally. For E s n. g.g:g. More precise conclusions are drawn in Chapter I I. then IRn . The number of points in its boundary is clearly even. in general.x) x. which is smoothly isotopic to the identity. H(p. (b) Let IH:" > IH: be a Coo function which is positive on the interior of the unit ball. close to this plane. N we will write / :::e g to indicate that 1 is (a) If l :::e then there is a smooth homotopy Q)) 26. H: p aH(t. H'(p. #11('1) I#rl (q) '" #gI (q) Hil l: H1(q) 24. t.t) �is an equivalence relatiOll. if M is orientable.
and the argument which was used in the proof of Theorem 56.. I1 h..1 (q) is defined in Problem 24). we have �. and I : M > N.1 )Wo + IWI · Show that the family cp. = (l . then for regular values ql .WI · .J )manifold.:. which is generated by {X ' } . Wt = I " . O. (b) Let Wo and W I b e nowhere zero nforms o n a compact oriented nmani fold M. (f) If M and N are compact nmanifolds. of kforms on (where #1. it follows that there is a Coo family {cp. q2 E N we have 27. Let {X' ) be a Coo family of Coo vector fields on a compact manifold M. and define w. } of diffeomorphisms of M [not necessarily a J parameter group] .lnl£gmlion 295 (c) Show that by choosing suitable p and I we can make H (1. then JR" . of diffeomorphisms generated by {X'} satisfies for ali I if and only if LXI Wt = Wo . with <Po = identity. show that if M c JR" is a compact (n . o h� h (a) Show that for �(t) = cp. + w.* (Lx' w.Wt . i. suppose X is a Coo vector field on M x [0.) From the addendum to Chapter 5. = cp. then there is a diffeomorphism I: M > M such that I(p) = q and I is smoothly isotopic to the identity.M has at least 2 components. then XI (p) will denote 1CM * X(p .e.. (e) Use part (d) to give an alternate proof of Sl£p 3 of Theorem 9. (d) If M is connected and p .. M we define the kform Wt +h . 1 ] .. O For a family w.. (p» (Xlf)(p) = lim CP'+h P» .). q E M. This number is called the mod 2 degree of f..0) be any point in the interior of the unit ball. (To be more precise. for any Coo function I : M > JR we have I( ( I(cp. *w. (g) By replacing "degree" with "mod 2 degree" in Problem 23.
296 Chapter 8 (c) Using Problem 718. Show that there is a diffeomorphism II : M > M such that Wo = J. Define (d) Suppose that JM Wo = JM W I . J A (u v) 1C o o [r (u.g . where M x N is oriented as in Problem 18. Let I : M k > JRn and g : Nt > JRn be Coo maps. t ) : p E M} (K(q . where M and N are compact oriented manifolds.O) H(p. n = k + 1 + J.g : by M x N > s n. *W I ' . J) = g(q) j(p) (H(p. g) = (Jlt + li(g.W I = d A for some A. N x [0. t ) : q Show that n E N } = 0 for every I . J) such that = = I(p) K(q .I r (g(q) . g) = 4 = i(j. g) (c) For I. so that Wo . n Cif. J ] > (a) i(f.I(p» C JRn  (OJ  Cif. O) = g(q) K(q.l(p)I ' i(f. q ) = We define the linking number of I and g to be '" g(q) I(p) Ig (q) . 'v)]3 du dv. (b) Let H: M x [0.g (P. J ] > JRn and K: with JRn be smooth homotopies H(p. g) = deg Cif. g). J ). i(f.  1 1 10 10 . g : S l > JR3 show that i ( f. show that this holds if and only if 28. and I( M ) g( N) = 0.
b' .b. b. b. (a.hltegmtion where 297 1" (u. b.(a.b. g) = 0 if J and g both lie in the same plane (first do it for (x.p (u) g3 (V) . (a. {a.JI (u) g2 (V) . y)plane).b'. (U» (/1 )'(U) (g3 )'(V) (g 2)'(V) A (u.c) = 0 for (a. v) = Ig(v) . Show that (VI )p. c) rt N. g) without calculating. The next problem shows how to determine C(J.b'. (V2 )p is positively oriented in ]R 3p when oriented in Mp.c') p triple wp. (a) For (a. b. c') = 41C. v) = det (g l )' (V) g l (V) .c) and (a'. (V2)P is positively lim g.J(u) 1 (P)'(U) (/3 ). b.M g.Mp for which the Wp 1M dEl(a. b.g.c) . c) = Let (a. c) is on the same side as a vector wp E ]R 3p .b.b. b . (a' . b. then g.c) E ]R 3 define dEl(a.c')'J> P Hint: First show that if and M = aN.c) _ For a compact oriented 2manifoldwithboundary let  (x . c) = 41C for (a. M C ]R3 and (a.a) dy /\ dz (y .b'.b) dx /\ dz + (z c) dx /\ dy [(x _ a) 2 + (y _ b) 2 + (z _ C) 2]'/2  g.c) /(a'.  . b. on opposite sides of M. (a. (a.J'(u) ( ) (the factor ) /41C comes from the fact that Is' (J ' = 41C [Problem 914]). c) . c') be points close to p E M. (d) Show that C(/. c) rt M. .b. (V I )p. c) E N . 29. Suppose (a.c)'I> p (a'.
y.[ * y . . z) 1 3 X .+ be the number of inter sections where dgjdl points in the same direction as the vector wp of part (a). and .=  J 41T (c) Show that ag. . Show that hut the surface on the rightincJuding the hemisphere behind the pJane of the paper is orientabJe. (An M with this property always exists.a) dy . Let . c) = S J* . y. b.C) dy aa (a. pg.g). z) 13 ag.(Y .a) dZ ab (a. Compute i(f.).lSI J l(x. y.lSI J l(x . + .b) d. n = . s' ) ) ) (d) Show that n = i(f..298 Chapter 8 (b) Let J : Sl > IR 3 be an imbedding such that f( S I ) = a M for some com pact oriented 2manifoldwithboundary M..the number of other intersections.(a. .C) dx . 1 38. c) . that when g(l) = P E M we have dgjdl rt Mp. z) I' I _ 1 ( ( ( l g*(dg.g) for the pairs shown below. [ * Z . ac I(x.b) dX ag.. .) Let g : S I > IR 3 and suppose The figure on the Jeft shows a nonorientable surface whose boundary is the "trefoil" knot. c) . b . Topowgy qf 3Manijolds. See Fort. b .n.(X ..(Z .
If Ci E H k (M) and fJ E HI (M).(p.e.. q} so that A and B are diffeomorphic to JRn . then Ci V fJ = (J llfJ v Ci . q E JRn be distinct.(p . show that A B po oq Hk (JRn .. and that . w 1\ � is exact if w is exact and Show that v is bilinear. and A n B is diff eomorphic to JRn.. 31.. Hk+1 (M x N) is defined by Show that x is welldefined. (b) Find the de Rham cohomology vector spaces of JRn . then [w] v [�] = [w 1\ �].1 (JRn . We define the cup product v : Hk (M) x HI (M) .(OJ. If J: M . (a) Let p . q }) has dimension 2.. i. and that Hn.. N.. Using an argument similar to that in the proof of Theorem 16. J* (Ci V fJ ) = J*Ci V J*fJ· (e) The crossproduct x : H k (M) x HI (N) . and Ci E H k ( N ) .Integration 299 30.J . B C JRn . � is closed. Choose open sets A .F where F C JRn is a finite set. q } ) = 0 for 0 < k < n . Hk+l (M) by (a) (b) (c) (d) Show that v is welldefined...{p. fJ E HI ( N ) ..
denote 1Cj*de. /\ de ik represent diff erent elements of H k (Tn ). show f]). • .300 (f) If that .. Sl . Tn has degree O. where 1fj : Tn 7 S l is projection on the jth factor. by finding submanifolds of T n over which they have diff erent integrals. Hint: Use Problem 25. .. (b) Show that every map J: sn let . ) Equality is proved in the Problems for Chapter I I . : Chapter 8 M > M x M is the "diagonal map". . 32. (a) Show Ihat all de i l /\ . given by "(p) Ci V f] = "* (Ci X X = (p. > X n times . Hence dim H k (Tn ) � (. On the ndimensional torus Tn = S l de . p).
. ) is clearly nondegenerate. b) . . = "L."L.n . 11. . for all v i' o. a. the field F will always be JR. and conse 30) . then n ( a 1 . which is symmetric. r :0.=1 For this inner product we have (a. an ) . ) on (a. v) > 0 A positive definite bilinear function ( quently an inner product. . w) 1+ (v. a'b'. w). . but there has been one notable w) = (w. albl . a. I with vector spaces.CHAPTER 9 RIEMANNIAN METRICS exceptionwe have never mentioned inner products.+1 . v) i' o. . for r = 11 we obtain the "usual inner product". n previous chapters we have exploited nearly every construction associated and nondegenerate: if v i' 0. . v). . . ) is called positive definite if (v. In general. . The time has now come to make use of this neglected tool. For each r with 0 :0. we can define an inner product ( .=1 > 0. a'b'. (a 1 . on JRn by (a. denoted by (v. i=r+ l n this is nondegenerate because if a i' 0. b).=1 . . An inner product on a vector space V over a field F is a bilinear function from V x V to F. ffi. a symmetric bilinear function ( . . and thus with bundles. then there is some w i' 0 such that (w. . a) > 0 for any a i' o. = "L. a ») = " (a ')2 L n r In particular. ( . For us. (v. . ).
a). . For allY basis V I . Vn of V.a j > 0 L . nondegeneracy of ( . . o. bv <pv(w) = (v . meaning that . w) = a(v)(w). Clearly. Thus. Also. for all a l . W W. )= i. .' dimensional. . ) is clearly nondegenerate if J is an isomorphism OllW V. ) gives lIS an iSOITIOI"phism a : V + V*. gij = (Vi . Vj ) ' gij = gji. an inner product ( . ) is defined on IR 2 by (a. Since an inner product { . b ) = a I b l . ) implies that <pv i' 0 if v i' O. with (v. then J* ( . .. wI· Since ( .= 1 .g. if ( . if V is finil<. . and J: IR > IR2 is J(a) = (a. if ( . a n with a t least one a i =j:. v*u. (gij) is symmetric. .j=i L gijv*i 0 v*j . so if > V is a linear transformation. the map v !)o ¢v is a linear transfor mation fi'om V 10 V*. we can define a linear functional ¢v E V*. for each V E V. with corresponding dual basis v*} . Thus. Nondegeneracy of ( . ) is a symmetric bilinear function on This symmetric bilinear function may be degenerate even if J is oneone. . . so symmetry of ( . J. gij a.9 Notice that an inner product ( . then J* ( . . However. ) is positive definite if and only if J is oneone. ) on V is an element of 7 2 (V). J* ( . . ) is cquivalenl to the condition that (gij ) is nonsingular.. ) is positive definite. the matrix (gij) is just the matrix of a: V > V* with I 'espect to the bases {Vi } lor V and {V*i} for V*. ) corresponds to the more complicated condition that the matrix {gij } be "positive definite". we can write ( . . ) i!i linear in the first argument. . ) implies that the matrix The matrix {gij } has another important illl crpretation. n I n this expression. } is linear in the second argument. e. Positive definiteness of ( .302 ChajJle.a 2 b2 . det(gij) i' 0.
j=1 . THEOREM. If not. equality clearly holds.w ll 2 = (AV . E V we have E IR.)2 . Il t jvf t 1=1 i. (3) II v + w ll :5 II v ll + II w ll (Triangle inequality).:5 o. and its discriminant must be negative. I . •:. . a ) = L )a. w ) II v ll 2 + II w ll 2 + 2 11 v ll . i=l The principal properties of II II are the following. v + w) = II v ll 2 + II w ll 2 + 2 (v .W . IRn we denote the norm corresponding to ( laJ = . w) J :5 II v ll . by (2) The function II II has certain unpleasant propeJ1iesfor example. For all v . a j . AV .2A (v.Riemannian 1I1etri(. w) + II w ll 2 So the right side is a quadratic equation in A with no real solution. (2) J (v . This latter function is a "quadratic function" on V in terms of a basis {Vi } for V it can be written as a "homogeneous polynomial of degree 2" in the components: = g.l 303 norm In Given any positive difmite inner product ( . w PROOF. ) on V we define the associated II II by II v ll = j(V.4 11 v ll 11 w ll < (3) II v + w ll 2 = ( v + w. the func tion I I on JRII is not differentiable at 0 E JRIIwhich do not arise for the function II 11 2 . W) 2 . (I) is trivial. Thus 2 2 4(v.w ) = A 2 11 v ll 2 . II w lL with equality if and only if v and w are linearly dependent (Schwarz inequality). Il w ll = (II v ll + II w ll) 2 .w for all A (I) lIav ll = JaJ · II v ll . so 0 < II Av . ) simply by J(a.V) (the positive square root is to be taken). (2) If v and w are linearly dependent. then 0 # AV .ja.
we can define /* ( . "up to isomorphism".304 More succinctly. . PROOF Let WI . II/(v)1I = IIvll for all v E V. If ated to an inner product ( . (f(v). w" be any basis for V. } � k VI . If ( . Chapter 9 11 11 2 = L gijV*. . THEOREM. An invariant definition of a quadratic function can be obtained (Problem I) from the following observation. W) = Wv + w ll2 . • • . w) for all v. . . ) = ( . . then I is also inner product preserving. ) . ) . . b) = (f(a). Vk so that .) Consequentl.: there is an isomorphism f : JRII T V such that (a. w) = Wv + wll 2 . then there is a basis VI . ' V'j. . /(b» . if I: V + V is norm preserving. \Ve will now see that. a. I n other words.IIv . that is. ) is a positive definite inner product on an Ildimen sional vector spare V. there is only one positive definite inner product.b E IR". . then (I) " i. •:. Vi) = DU' (Such a basis is called orthonormal with respect to ( . w E V. Suppose that we have constructed 1 :'0 i. Theorem 2 shows that tv. 3.j=i II II is the norm associ (2) (v. Similarly. Vn for V such that (Vi . 2. PROOF Compute.'O inner products which induce the same norm are themselves equal. THEOREM (POLARIZATION IDENTITY). .IIw1l 2] (v.IIvll2 .w 1l2 ]. I(w» = (v. ) on V. ancJ clearly IIvl li = 1. We obtain the desired basis by applying the "GramSchmidt orthonormalization process" to this basis: Since W I # 0. that is.
Now. ..:::::: span Wj . . The "triangle inequality" (Theorem I (3)) shows that this is indeed a metric. w) = IIv . Let It is easy to see that .I 305 and span V I . ) can be expressed by the equation Ci(V)(W) = Ci(W)(V). . A(V). Then W k +1 is linearly independent of V I . .wli. w). . .ic. k . ) on V provides an isomorphism Ci : V + V* with Ci(V)(W) = (v. defined by i(V)(A) we obtain an isomorphism fJ : V* + = a. . . . : melric on A positive definite inner product ( . Vk . We also call IIvll the length of v. . . ) * on V* by + ( V*)*.Riemannian 1\1el. • • . . So we can define and continue inductively. . = I. We have only one more algebraic trick to play. Vk . This is because we obtain a metric p on V by defining p(V. ) on V is sometimes called a Euclidean V.I V We can now use fJ to define a bilinear function ( . Recall that an inner product ( . • . . i Using the natural isomorphism i: V + V**. W k . the symmetry of ( .
)= . ) * = L gij V**i ® V**j i. (gU). and let ( Then .} and {v·*. choose a basis {v.j= 1 L g ij Vi ® vj . n (gu) so is the matrix of is the matrix of is the matrix of a: V V* V V** with respect to with respect to with respect to {v.)=1 n n = .) (gU ). if we consider Vj E V**.1 V· + Thus. without the invariant definition. ® v*j ..} for V.. )* is also symmetric.1 So aI : V* fJ : + {v*. )* is an inner product on the dual space V* (in fact.} be the dual basis for V*. To see what this all means. if we let g U b e the entries of the inverse matrix.} and {v*.306 Letting this can be written which shows that Clzapw 9 a(v) = ft. a(w) = /1. (g U ) = (gU )I . . that this equation defines ( .. let {v*. the one which produces fJ). Consequently ( .) . ( . so that then ( .j v*.} and {v.) {v*.j=l + L g. ) * independently of the choice of basis. One can check directly (Problem 9).
we need only describe which two elements. for A # 0. letting a(v) = A.' 307 Notice that if ( . If we write n Wj = L CijjVj. will have length 1 . . which we will not describe in a completely invariant way. . . Vn is orthonormahvith respect to ( . . vn ) = ±w(w" . Wn b e two bases o f V which are orthonormal with respect to ( . we are interested in only one case. . • It clearly follows that v *1 /\ We have thus distinguished two elements of [I n (V). .Riemannian 1I1etric. wn ) . . )* is also positive definite. This can also be checked directly from the definition in terms of a basis. so ( . . TdV) = T k (V*). /\ . and [l k (V). . ). So to produce an inner product on it. The vector space [I n (V) is I dimensional. Let V I . we have a(v)(v) > ° for v # 0. . Vn and WI" . . /\ v*n for {v. )* if and only if v" . /\ W *1/. . . In the positive definite case. . . } an orthonormal basis of V. ) is positive definite. . 1'=1 then So the transpose matrix A t of A = (aU) satisfies A · A t = 1. . so that then. they are both of the form v * . which implies that det A = ± l . However. . the simplest way to describe ( ) )* is as follows: The basis V*l. . k�' " w(v" . . . . v*n of V* is orthonormal with respect to ( . . Similar tricks can b e used (Problem 4) t o produce a n inner product on all the vector spaces T k (V). . It follows Iiom Theorem 75 that for any W E [I"(V) we have = L Cik i Cikj . . ) . We will call these two elements /\ V*n = ± W * 1 /\ • . w and w.
. Vn and write Wi = j=l L Cijj Vj o n Problem 79 implies that Ifwe write ( . /\ · · · /\ w * " gu = (w. the function is also continuous. . If � = 1f : E T B i s a vector bundle. To express the elements of norm 1 in terms of an arbitrary basis W I .j )..s /Jositivf. )= i. \IVe now apply our new tool t o vector bundles.308 Chapw 9 the elements of norm 1 in n n (v). Will we choose an orthonormal basis v} .j=l L gU w*. then we can further distinguish the one which is positive when applied to any (VI . . we will call it the positive element of norm 1 in n n (v).. ) which assigns to each p E B a positive definite inner product ( .' (p). . Wj ) . )p on . . 0 w*j . . then In particular. the elements of norm in n"(V) arc * ± Jdet(gu) w . vn ] = /1. we define a Riemannian metric on � to be a function ( . Ifwe also have an orientation J1. . and which is continuous in the sense that for any two continuous sections S1 . . n then = k� ' L CikiCikj . " .. . . " so if A = (Ci. . det(gU) is a/wll). If � is a Coo vector bundle over a Coo manifold we can also speak of Coo Riemannian metrics. Consequently. vn ) with [v" . . • . S2 : B T E.
+ « p . 2: 0.Riemannian 1I1etricJ 309 [Another approach to the definition can be given. For v. there is an obvious Riemannian metric. w)� = (lu (V) .] Notice that the argument in the final step would not work if we had merely picked liondegenerate inner products ( .] 4. but an instance of a more general structure. note that (v. To show that it is positive definite.. Then ( ... UeC9 each ¢u(p)(v. the new object Elie(�) that we obtain is not a vector bundle at all. p eB then a Riemannian metric on � can be defined to be a section of EliC(�).1 (p» ..I (p) by ElIc(. : E + M be a [COO] k plane bundle over a Coo manifold M. a fibre bundle.v) � . THEOREM. and let Eue(�) = U ElIe(. Let � = . ) on TS 2 which gives a symmetric bilinear function on each S2p which is not positive definite or negative definite but is still nondegenerate. lu (w» p .. and for some U strict inequality holds. W E . v)� [The same argument shows that any vector bundle over a paracompact space has a Riemannian metric. Then there is a [COO] Riemannian metric on �.. w)� UeC9 (v. . ) by (v.b» p = (a. v. ) u i s a [COO] Riemannian metric for � I U. The only problem is that Eue(V) is not a vector space.I (p). ) is continuous [COO] and each ( . Let {¢u} b e a partition of unity subordinate to (9. W E .W)p = L ¢u (p) (v. Let ElIe(V) be the set of all positive definite inner products on V.. Ifwe replace each .I (p). b ) . )p is a symmetric bilinear function on . define Then ( . We define ( . (p. •:. v)p = L ¢u (p) (v. a ) .I (p).. PROOF There is an open locally finite cover (9 of M by sets U for which there exists [COO] trivializations On V x IU : . I(U) U x IRk IRk . ) u In fact (Problem 7).I (p» . there is no ( .
[S(p)] = /J. But � :::: �* . then there is a nowhere 0 section of [1 1 m = �*. which gives a positive definite inner product ( . COROLLARY. J. S(p» p = I . Clearly s is a section. then � "" �*.310 Chapter 9 As an application of Theorem 4. We "p (v)(w) = (v. Then for each I' E have an isomorphism defined by M is a k plane bundle.1 (p)] * . Let ( . Continuiry of ( . If � = 1[ : E + PROOF. ) implies that the union ofall "p is a homeomorphism from E to E' = UPEM[1[ . ) be a Riemannian metric for �. 6. A ) . If � has an orientation Riemannian metric on M then there is a unique and ( . a Coo Riemannian metric ( . W E 1[ . we then define an equivalence ! : E + !(A S(p» = (p. COROLLARY. All these considerations take on special significance when our bundle is the tangent bundle TM of a Coo manifold M. then � is trivial if /J. •:. we settle some questions which have till now remained unanswered. w) p v. •:. ) is a with (S(p) . M x IR by ALTERN ATIVE PROOF We know (see the discussion after Theorem 79) that if � is orientable. so that �* is trivial. ) p on . In this case. M.p . ) for TM. If � = 1[ : E and only if � is orientable. + M is a I plane bundle.1 (1'). PROOF The "only if" part is trivial.
is called a R emann an metric on M. then on V we can write our Riemannian metric ( . So for every Coo map f: N T M there is a covariant tensor f* ( .j ) Similarly. j= 1 \vherc the Coo functions gij satisfy gjj = gji. which ( . i. ) is symmetric. If (x. ) on M determines two clements of n n (Mp). and we can write it as Our discussion of inner products induced on V'" shows that for each matrix (gU(p» is the inverse of the matrix (gU(p» .. is a contravariant tensor of order 2. even though it is usually not d of anything (even when M is orientable and it can be considered to be an nform).Jdet(gu(p» dx 1 (p) /\ .Ib·emalUzian A1etric. the k=l L g. since ( . we obtain a "volume element" on in Chapter 8. n If M has an orientation /J. A Riemannian metric ( .j dx. M. of course. ® dxj . it is a Riemannian metric on N if and only if f is an immersion (f p is oneone for * all p E N). /\ dx" . the elements of norm I.Jdet(g. U) is a coordinate system on M. ) as ( . ) on N. ) * . . which is dearly symmetric. ) induces on the dual bundle T*M. We have seen that they can be written ± . for each p E M the Riemannian metric ( . then /J. and we obtain an nform on M. • • /\ dx" l · . .j ) Idx 1 /\ This volume element is denoted by dV. ) on M is. a covariant tensor of order 2.I 31 1 each Mp. then on V this form can be written .k g kj = 8( .j) > 0 since ( . . . ) is positive definite.p allows us to pick out the positive element of norm 1. The Riemannian metric ( .Jdet(g. i i ) = L g. dx 1 /\ . thus p. as defined Even if M is not oricntable. in a coordinate system (x. V) it can be written as . and det(g. /\ dx n (p). if x : V T JRn is orientation preserving.
. If M is an IIdimensional manifold (withboundary) in IRn . . ) = L dx. We can then define the This certainly makes sense if M is compact. we have tangent vectors y' (t ) = dt dy E My (t). in which case we say that M has "infinite volume".=1 ' /\ . b] such that y is smooth on each [/. There is an even more important construction associated with a Riemannian metric on M.] (with possibly different = = t I dr I dl Y from a to b. .. ® dx. and can therefore use ( . 1. ) to define their length Wc can then define the length of dl dl Il drll ( dY ' dY ) ( = ( dY dY ) = dt' dt y(t ) ' to be precise . dV = I dx . which will occupy us for the rest of the chapter. For every Coo curve y : [a . then gij = 8U. = ) (t lI y'(t) 1I d ) = . ). with the "usual Riemannian metric" 1M dV. and in the noncompact case (see Problem 810) it either converges to a definite number. /\ and ccvolume" becomes ordinary volume. dxn l . or becomes arbitrarily large over compact subsets of M._. meaning that there is a partition a to < ' " < b of [a. L� (y ) I" If y is merely jJuuwise smootli. b] + M.312 Clzapw 9 and is called the volume element determined by volume of M as ( . so .
We can reparameterize y to be a curve on [0.old s(a) If y satisfies s(t) = I we say that y is parameterized by arclength (and then often use s instead of I to denote the argument in the domain of V). s(b) = L�(y). (*) s'(t) = Consequently dyjdl has constant length I precisely when S(I) = I thus precisely when s(t) = I . . [ I �.]) . + constant. .l (y I [IH . We can also define a function s : [a.a. b .II . . y(1 . we can define the length of y by L�(Y) = I >:.Riemannian Metnes 313 left. Whenever there is no possibility of misunderstanding we will denote L� simply by L. s(t) = L�(y) = L�+a(y) = old s(t + a) . . .a).and righthand derivatives at 11 . Then b a = I �. bJ + JR.aJ by defining y(t) = For the new curve new y we have = 1. In . i=1 . with the usual Riemannian metric L dx.. 1. the "arclength function of y" by s(t) = L�(y) = Naturally. ® dx.=1 this definition agrees with the definition of length as the least upper bound of the lengths of inscribed polygonal curves. I dl. A little argument shows (Problem 1 5) that f piecewise smooth curves in or JRn .t}.
) Consequently. b] + M from p to q (there is even a smooth curve from p to q).j=1 and we would use the same symbol today.dxi n II 11 2 = The symbol dx.i dx. It i s clear that d(p. Define d(p. but what it always actually meant was i i.d(p.r) < t.) Consider now a Riemannian metric ( .j=l L gu dx. If p. p) = O. : [a.dxi at p should not be interpreted as a bilinear function at ail. but the inner product of two "infinitely small displacements" dx and 8x. q E M are any two points. in classical books one usually sees the equation ds2 = Nowadays. ® dxi the value (dx. The classical way of indicating one wrote L g. q) < t q to r with L(Y2) d(q. (Classically. i f r E M i s a third point. (This makes some Sort of sense even in modern notation. this is sometimes interpreted as being the equivalent of the modern equation ( . b] .j=1 L gu dx. the norm II II on M was denoted by ds.dxi appearing here is Tlol a classical substitute for dx. then for any £ > 0. then there is at least one piecewise smooth curve y : [a. .314 Chapw 9 Classically. b] + IR we have Ids l = Y* ( II II ) on [a.b] + [b. Moreover. ® dxi . we can choose piecewise smooth curves y. q) 2: 0 and d ( p.8xi n where dx and 8x are independent infinitesimals.dxi dxi ® dxj was very strange: i. ) . ) on a cOTlnected manifold M. ) = L7. �' gU dx.dxi)(p) of dx. equation (*) says that for each curve y and corresponding s : [a. q) = inf { L ( y) : y a piecewise smooth curve from p to q}. but as the quadratic function i. the Riemannian metric was not a function on tangent vectors.c] + Y2 : M M from from p to q with L(y.
M > 0 such that m . The function d : M " M + lR is a metric on M. THEOREM. m < G <M Now if P E B and 0 # bp E lR"p . Then mlbl < Ib IG(p . 7.l is compact there are num bers m. and if p : M " M + lR is the original metric on M (which makes M a manifold). ) be any other Riemannian metric. G(p. lI ap lip = lI<l b l a)p llp = II b llp. this gives the desi�ed inequality (which clearly also holds for b = 0) . Let U be an open neighborhood of the closed ball B = {p E lR" : I p l :::: I } . it follows that :::: d(p. PROOF Both parts of the theorem are obviously consequences of the following 7'.Riemannian Metric.(y). let ( . and II II be the corresponding norms. ( . a) = Ib l . be the "Euclidean" or usual Riemannian metric on U. •:. then Y is a piecewise smooth curve from p to r and L(y) = L(YI ) + L (Y2 ) < d(p. ).b] + B we have mL. q) + d(q. r) + 2e.c] + M to be YI on [a. L dxi ® dxi.(y) :::: L(y) :::: M L. I I :::: II II :::: M . let a E S". Since this is true for all £ > d(p. q ) > 0 for p # q . and consequently for any curve y : [a. M > 0 such that on B " S".] The function d : M " M + lR has all properties for a metric.I . q) + d(q. except that it is not so clear that d(p . PROOF Define G: B " S"I + lR by Then G is continuous and positive. . Since B x SIl . then (M. This is made clear in the following. d) is homeomorphic to (M. r) 0. LEMMA. a) < M lb l. r). i=1 " and let ( .1 315 If we define y: [a. Then there are numbers 111 . = [Ifwe did not allow piecewise smooth curves. Let I I = II II. p). but d would still turn out to be the same (Problem 1 7).a) = lIap lip.b] and Y2 on [b.I be a = b/lb l . ). I I on B. since Ib I G(p. there would be difficulties in fitting together YI and Y2 .c] .
f(t). Suppose we are given a (suitably differentiable) function F. I n order to investigate the question of shortest curves more thoroughly. and q might be p.. if d(p. the manifold M might be 1R2 {OJ. As an introduction to such techniques. x . .316 Chapter 9 Notice that the distance d(p . we have to employ techniques fi'om the "calculus of variations". we consider first a simple problem of this sort. if" F(t. e. then y is clearly a shortest piecewise smooth curve from p to q (there might be morc than one shortest curve. Of course. For example.f'(t» a b dl . For example. q ) defined by our metric need not be L(y) for any piecewise smooth curve from p to q.g. among all functions b' f: [a. y) = �. lR x lR x lR + R We seek. the two semicircles between the points p and p on S l ).b] + IR with f(a) = a' and f(b) = b' one a' which will maximize (or minimize) the quantity t F(t. q) = L(y) for some y.
l(t» between (a. y) E ]R2 for which (*) D.e.. or even a local maximum or minimum.b] which makes the curve (t. To solve this problem. As a second example. b') of shortest length t )1 + [f'(t)F dl. This is the same as saying that the curves I r> l(x + l.y) = 21f X JI+Y2 .y) = o. those points x for which I'(x) = O. i. y) t r> l(x. A critical point is not necessarily a maximum or minimum.Riemannian 1I1etrics 317 I r> then w e are looking for a function I on [a.y + l) . but critical points arc the only candidates for maxima or minima if I is everywhere differ entiable. /(x. Similarly. if F(I. then we are trying to minimize the area of the surface obtained by revolving the graph of I around the xaxis. which is given (Problem 1 2) by To approach this sort of problem we recall first the methods used for solv ing the much simpler problem of determining the maximum or minimum of a function I: ]R + IR. a') and (b. we examine the critical points of I. for a function I : ]R2 + ]R we consider points (x.y) = Dd(x. x.
= du u=o du I aa at I Ib F (l. For a variation Ci we now compute dJ(&(u» d . f'(I))dl a : (e. such that The functions I r> a(u. f(t). e) to the set offunctions f : [a.e ) which pass through f for u = O. We will denote this function by &(u).e) + 1R2 with e(O) = (x. then we call a a variation of I keeping endpoints fixed.b] + IR a(O.a(u. e). This can be done by considering a "variation" of f. I) = f(I). I). We might try to get more information by considering 0 = (f 0 e)' (O) for every curve e : (. because of the chain rule. e) x [a. Thus & is a function from (e. but it turns out that these conditions follow from (*). l) are then a family of functions on ( e. To find maxima and minima for funclions f: [a.318 Chapw 9 have derivative 0 at the condition O.b] + we wish to proceed in an analogous way.b) = b' b' a ' a b for all u E (e. by considering curves in the sel o all f IR.(U.e. b] + IR. If each &(u) satisfies &(u)(a) = a' . . I)) dl u=o a .y). in other words if a(u. that is) a function J (f) = t F(I. & (u)(b) = b'.a) = a' a(u.
1(1). f'(t)))] dl aa aF ' I + au (0. for some � dJ(&(u» � : [a.aF1.(1. and we obtain (**) � dJ(&(u» ax au I u�o = Ib aa (0. I) [aF (1. f(l). .. ofcourse. /(1) . f'(I))) ] dl. 1 (I» b 11 =0 = a X . (I» a aF aa a a I Ib U (0.dt a ' In classical treatments ofthe calculus ofvariations. thus obtaining (*)  dJ(&(u» du For variations Ci keeping endpoints fixed. 1(1). f'(t» )] dl. I). f(I). flu l u�o J(&(u» J aF d aF 8J ax dl ay Ib � [. Then we obtain a (u. (I» ( ay (I.Riemannian 1I1etl1cS 319 = f [ :u lu�o F (I. we can apply integration by parts to the second term in the integrand. t)) ] dl Since a 2 ajaual = a 2 ajalau. 1'(1» d ( aF I. a(U.. 1. I) ay (1.d1(1). �7 (U.] dl. 1(1). b] + IR with �(a) = �(b) = O. the variations Ci were taken to be of the special form .dt = a . the second term is 0.f (� (t. The derivative is called the "first variation" of and is denoted classically by a/ ax Iu�o Ib �(I) [ aF (1. essentially the same. l) = 1(I) + u�(t). = a The final result is.1) [.
b] with II(a) = 1)(b) = 0. If I is to maximize or minimize then must be 0 for every variation Ci of I keeping endpoints fixed.(t).(t» . l(t). J'(t» ) = ax di ay 1)(t) = aCi (0. We call I a critical point of (or an extremal for if =0 for all variations of I keeping endpoints fixed. consider the case where equation is = ¢(b) = 0. LEMMA. b] + IR satisfies l b 1)(I)g(t) dl = 0 for every Coo function 1) on [a. The particular form (**) into which we have put now allows us to deduce an important condition.f. . not only are the ar guments I and (t. J.320 Chapw 9 As is usual in classical notation. J(t). y) = !i+7. J Ci J) 8J(Ci) 8J 8. If a continuous function g : [a. Ci 8J(Ci) 8J(Ci) J. As in the case of I dimensional calculus. the arguments of functions are either put in indiscriminately or left out indiscriminatelyin this case. The Euler [/ (1)] . THEOREM (EULER'S EQUATION). which vanishes at a and b. 8J Ci J. The point of if and only if I satisfies J C2 function I is a critical PROOF Clearly I must make the integral in (**) vanish for every aF (I. then g = O. there is no reason to expect that the condition = 0 for all will imply that I is even a local maximum or minimum for and we emphasize this by introducing a definition. b) and ¢(a) As an example. x PROOF Choos C 1) to be ¢g where ¢ ispositive on (a. So the theorem is a consequence of the following simple 8'. but the dependence of on is not indicated (which can make things pretty confusing). /'(1» omitted (resulting in the disappearance of the function I for which we are solving). •:' 0= di d ( '/1 /'(1)' 2 ) ' + F(t. 1 ) au o.d ( aF (t. .
we use one of the �o standard tricks Ocaving justification of the details to the reader). Notice that we would have obtained the same result if we had considered the case for then the Euler equation is simply This is analogous to the situation in I dimensional calculus.Riemannian Metrics so 321 hence which implies that So is linear.f'2 I/" = 0.y) = 1 + y2.( )1l+ [f''(t) F ) .x.11 +1' 0 = + 1'2)/" I'!" = I' + 1'2)/". o = �(2/'(1» . We let Then . J7 I. = dx ' the To solve this. dy dx 1 + ()2 y=O. ( J7)' = L . I" = 0. 2J7 F(I.�"��� ) .11+1'2 ' / .x.I �2 o= �(.. dl (t) f (t) 1 + . where Euler equation is _ . where the critical points of are the same as those of since For the case of the surface of revolution. I F(I. (I (I this leads to the equation which we will also write in the classical form 0= )1 + [f'(t)F :!. dx d2y2 dy P = y.y) = xJI+Y2..
� dy p = dx = Jey2 . The graph of X +k = c cosh ( . e eX + ex x = 2 cosh is shown below. Pp2 1+ � log(l + p2) = log y + constant 2 y = constant .) . dy .322 Chapter 9 so our equation becomes 1 + p 2 . decreasing for increasing for � o.dp = � d)'. and cosh .1 dy 1 = dx Jey2 Y and thus (see Problem 20 for the definition and properties of the "hyperbolic cosine" function cosh and its inverse) cosh. we write this as Y = x +k. x x � 0. it is symmetric about the yaxis.l ey Replacing c e by 1/ e.yp dp = 0.
Riemannian 1\1etnes 323 So our surface must look like the one drawn below. 1(1). not trivial to decide whether there a7" constants k and c which will make the graph of (*) pass through (a. a 1= 1 U + Thus.. I) aayFl (1.t) x [a . b] + M J of J must satisfy the n equations a.(1» ) = . b] + M is a piecewise smooth curve. J(I). any critical point We are now going to apply these results to the problem of finding shortest paths in a manifold M. Problem 2 1 investigates the special case where a' = b' . : (t.di ayFl (I.. J'(I)) .. � a. 1'(/ » Ib n . b'). by the way.(I» I l a. 1(/ ).. (0. a . t ) x for F: IR x IRn x IRn + IR. J(I). 1' (1) ) ] dl 1 X o. f(l). f. /. with y (a ) = p and y (b) = q. /'(1» dl .1) [D (I. we define a variation of y to be a function aF d (a axl (/. and compute = L. aF u=o [a. If y : [a . : ( t .a') and (b.b] + IRn J(f) = In this case we consider that t F(I..b] + IRn with &(0) = J. J(I). It is easy to generalize these considerations to the case where and J : [a. (1. It is.f G.... dJ(&(u)) dU b � a1 (0.
.a) = p . We would like to find which for all variations Ci keeping endpoints fixed. 1. We can assume that each y l [1. If (u.. . I) = " U (u. _ 1 .. (2) there is a partition on each strip We call " a variation of y keeping endpoints fixed if (3) u E (t. I). such that a = 10 < II < . aftenvards we will consider the relation between the two integrals.b] so that " is Coo (t. we will take a hint from our first example and first find the critical points for the CCenergy" I u=o =° E ( y) = z a I which has a much nicer integrand. we let paths y satisfy & (u) be the path 1 dL(&(u» du r> .r) =" (u.324 Chapter 9 > for some t 0._ 1.. b] we write lb I dy 1 2 dl = Zl ib ( dY dY ) dl.. .1) = y (t). As before.t) x [a. However.r) . b) = q for all (I) .1. U) (othelwise wejust refine the partition). (Iaa I ) U ( U. < IN = b of [a. (0.]. (.I ) aa aa"I (u.] lies in some coordinate system (x.t) x [1. I) is the standard coordinate system in (t.(u. .t). l) . dt a dt ' dt aa" . (u.(u.
and consider the gij [.c. i yi i. t) = xi (et(U. (dY dY ) dl ' 1. i ih. then aa iii (U. I» . Ifwe adopt the abbreviations eti (u..t) ti_ l E(y! [li l .j = 1 1 tj _ l " Then and So . li J) = 2 1 = 2 11.l) .Riemannian 1\1elr. then we are considering IRn . F(y(l).\ 325 Then aet/al(U. I) is the tangent vector at time I to the curve &(u).j =l dl dl If we usc the coordinate system x to identify U with as functions on JRn .dt.=1 1 l1 [}e/ 0 1 .. ndI dI dyi dyj 1 L gij (y(l» .I) So = L ii/(U. y'(t)) dt where F(x. ti _ l i.y) = '2 L gij (x) . a(u.
i d i dy . we note that a little index juggling gives r. dyj dy' = �. ax' ) [ .. agi/ dyi dyj = � agjl dyi dyj L. ax' dt dt j= From (***) we now obtain Notice that the final sum in the above formula is simpl\" Remember that y is only piecewise Coo. j= l ax' dt dt So �..= left hand tangent vector of y at I (l i· dl i ) �(tn dl To abbreviate the integral somewhat we introduce the symbols [ij. i.326 Chapw 9 In order to obtain a symmetrical looking result.1.. /] a = � gi/ 2 ( ax' + agjl ax' _ agij . Let dyI(l + ) = nght hand tangent vector 0f y at I .L. j= 1 ax' dl dl i.. L.. agl..
From the first variation formula it is. For the time being we present: with apologies. i = 1. even though different coordinate systems may actually be involved (and hence different gij and V i ). glr (Y(t) dd2. Let � aa: (O. . .Riemannian 1I1etrics These depend on the coordinate system. 11=0 N dy .fi).� (aCi (O. t) !. this coordinate dependent approach. 1. It should be noted that are not the components of a tensor.l Then we obtain the following formula (where there is a convention being used in the integral). Consequently. N . Ci . 9. . . I] ._.. but the integral  327 iI. simple to obtain conditions for critical points of E. later on we will have an invariant interpretation of the first variation formula. but there it is. For any variation w e have dE(Et(u» du I Ci.I](y(t) d� d�j n I [ n r n i ] dl. Now we just have to add up these results. Nevertheless. Il" d/) a. THEOREM (FIRST VA RIATION FORMULA). clearly cannot. (In the case of a variation from 1 to N . Ii]. we will use the exact same expression for each [IiI . the sum can be written [ij.) This result is not very pretty. which appears in our result.1 . of course. leaving endpoints fixed.' \�I [ij.
I) to be 0.[i2 + .328 Chapw 9 1 0 . If .I to Ii is written in terms of (x. such that satisfies y: E y E y(O) dy (0) = v (ji d2yk � j dy' dyjt = . except one. ) for some i. which is 0 outside of (1'1 .. In order to put the equations of Corollary 10 in a standard form we introduce another set of symbols OUf equations Can now be written We know from the standard theorem about systems of second order differential equations (Problem 54). (y(l)) d2. I!) : n r I.1 .L [ij.] is in U. Now apply the method of proof in Lemma 8'.I. there is a unique (t. that for each P M and each v Mp. d j =l r=1 PROOF. U). but a positive function times the term in brackets on (IiI. Suppose y is a critical point. The final term in the formula vanishes since y is Coo. COROLLARY. .). U) we have y: [a b] 11 y i Lg/. Given / with y(l) E U. 1. then in the first variation formula we can assume that the part of the integral from Ii. choose a partition of [a. then is a critical point of E� if and only if for every coordinate system (x. for some t > 0.' + i.t) + M.. If Ci is a variation of y keeping endpoints fixed. I](y(l)) d� d�j = 0 for y (t) E U.b] with I E (1. . choosing all aCi' jau(O.. r/k ( y (I » dt d l. L.j=1 =P o. and such that y l [I'_I. •. + M is a Coo path.
we could replace 0 by any other I.e) + M and Y2 ' (e. and ifmoreover d y.» .cs 329 Moreover. we see that yl [I' _h l. are O.e). ._.] satisfies the equation. COROLLARY. ) = L dx ' ® dx'i=l Here g'j = D. (0) = Y2 (0) then y. N . 1. . ./' dt ' aCt dy i = 1 .Riemannian 1I1et. consider the Euclidean metric on * y JRn ) ( . O] + M are Coo functions satisf ying this equation. because the final term in the first variation formula still vanishes. Let y be a critical point. b] + M is a critical point Eg if and only if Y is actually Coo on [a. (0. We already know that the integral in the first variation formula vanishes. This last fact shows that if y. A piecewise Coo path for (x.1. ' [O. Choosing the same as before (all are 0 outside of (1. + d Y2 dl (0 ) = dl (0 ) y. By our previous remarks. So we obtain which implies that all !:1/..j. U) satisfies Y ' [a.b].) = !J. this means that is actually Coo on all of [a. PROOF.. Naturally. ' I I . . Now choose Ci so that c/ c/ a. b] and for every coordinate system for y(I) E U. As the simplest possible case. and rt = O. The critical points y for the . . so all ag'j /ax k = energy function satisfy 0.e).. this y is Coo on (e. and Y2 together give a Coo function on (e. .:. I. We now have the more precise result.
The situation is now quite different from the first variational problem we considered. +!+ y I dyldlll . (y(t » dt df2 ' =l i. then y is param Now w e have ��:{ = [i/ . This shows that there are critical points for length which definitely aren't critical points for energy. dy d i dy' dy dyj " d2y dyj " " L agij (y(t» dt dtdt + L g'j (y(t) df'2dt = . and hence must be parameterized proportionally to arc length. when we considered only curves of the form I r> Any reparameterization of is also a critical point for length. y y y [a. both terms in parentheses are 0 (CorollalY 10).L ax' l.�.j=l /=l r. TllllS the length is constant. i]. this can be written as di dt r n Since is a critical point for E. since length is independent ofparameterization (Problem 16). the component functions of must be linear. f(t» . that y: E. Replacing agij lax' by the value given above. If eterized proportionally to arclength.j=l dyi d2y' + L gi. }] + [}/. since we have just seen that for to be a critical point for energy. This situation always prevails. y y y (t. so is a critical point for the length function as well. from the definitions.330 Chapter 9 Thus lies along a straight line.b] + M is a critical point for 12. d dy 2 d " gij (y(t» dit dyjt) I 1 = di (i. PROOF Observe first. THEOREM.
m] + [mk. It will also be useful to know a f ormula for ag ij / ax k To derive one. For the moment we consider only paths + M with # 0 everywhere. . we have y: . Ii] y [a dy/dl (x. g glm ogmj L. we first differentiate = m=l to obtain Thus we have ogikj L...j] + [jk.Riemannian 111etries The formula 331 ·· ag't [ik. For the por tion I of contained in a coordinate system U).m aY il gmj ([lk. Considering our coordinate system as JRn) we are now dealing with the case F(x.y) = \ iL gij (x)yiyj . g gmj oglkm aY I.j=l .i] ax occurring in this proof will be used on several occasions later on..m aYk I.m _ '"'" if _  _ '"'" if =  or We can find the equations for critical points of the length function L in ex actly the same way as we treated the energy function...b] y [li1.. l]) by (*) Lg I.
so it must satisf the same diff rential equation. given a critical point y for L with y I # ° everywhere.2 ax dl After a little more calculation we finally obtain the equations for a critical point of L : dt ds d2s dyk dl2 = O' di It is clear from this that critical points of E are also critical points of L (since they satisf y = 0). and we can consider the reparameterized curve e This reparameterized curve is automatically also a critical point for L . b] + [0. Since it is now parameterized by y e arclength. Conceivably a critical point for L might have a kink. There is only one detail which remains unsettled.j �1 ds ' y ( ) . dl . L� (y)] + M. the function d /d d2s/dl2 s: [a. but be Coo because it has a zero tangent vector y oS : [0. We introduce the arclength function Then So we have " a · d id j L 4· (y (t» � .L ax dl dl aF ( I dy ) _ � i. the third term vanishes. so y o S 1 is a critical point for E.332 Clzapw 9 s (t) = L� (y) . . L� (y)] 1 is a diff omorphism. Conversely.
Then there is a neighborhood U of p and a number < > 0 such that for every q E U and every tangent vector v E Mq with II v ll < < there is a unique geodesic yv : (2. which is concerned with the measurement of the earth's surface. A geodesic on the earth 's surface is a segment of a great circle. Let p E M. In this case it would not be possible to reparame terize y by arclength. d v (0) = v. Henceforth we will call a critical point of E a geodesic on M (for the Rie mannian metric ( . which are so far merely known to be critical points f length. Observe that the equations f a geodesic. then I r> y(cI) is also clearly a geodesic. including surveying and the measurement of degrees of latitude and longitude. »). The fundamental existence and uniqueness theorem says that there or is a neighborhood U of p and <] . This name comes f rom the science of geodesy. 2) + M satisf ying Yv(O) = q. <2 > 0 so that f q E U and v E Mq with IIvll < <1 there is a unique geodesic . Before we can say whether this is true for geodesics in general.Riemannian 111el1ics 333 there. The most elementary properties of geodesics depend only on facts about differential equations. we must initiate or a local study of geodesics. as in the figure below. which is the shortest path between two points. THEOREM. Problem 37 shows that this situation cannot arise. This feature of the equation allows us to improve the result given by the basic existence and uniqueness theorems. 13. � PROOF. or have an important homogeneity property: if y is a geodesic.
(v)· In particular..I dl 1=0 eXpq (c(I» = v. Identif ying the tangent space (Mq). In fact.« 21) . to obtain a curve c in the manifold Mq with dc/dl(O) = v E Mq = (Mq)o. d"i (0) = v. we have an induced map (expq).) The geodesic y can thus be described as y(l) = expq (lv). Then expq 0 c(l) = expq (lv). If (9 c Mq is the set of all vectors v E Mq for which expq (v) is defined. at v E Mq with Mq itself. •:. y : [0.. there is a natural way to give it a Coo structure. Choose < < < 1 <2 : Then if Ivl < < and III < 2 we have So we can define yv(l) to be yv/. we claim that the map (expq)o.(v) = :!. y(l)..334 Chapter 9 with the required initial conditions.. : Mq + Mexp. Since Mq is an ndimensional vector space. 1] + M If v E Mq is a vector for which there is a geodesic satisf ying then we define the exponential of v to be dy y(O) = q. the geodesic with tangent vector v at time 0) So (expq)o. . : Mq + Mq is the identity. exp(v) = expq(v) = (The reason for this terminology will b e explained in the next chapter. then the map expq : (9 + M is Coo) since the solutions of the diff rential equations for geodesics have a Coo e flow. we can let crt) = IV.
xn ) = (xl. we recall some facts about the manifold TM. Then is a coordinate system on 1[ . For every P E M there is a neighborhood W and a number e > 0 such that (1) Any two points of W are joined by a unique geodesic in M of length < e. . xn ) v E Mq. . .. q') is a Coo function from W x W + .l (U). . xn o n. (2) Let expq(v) TM.=1 a L X'. THEOREM. a v = L a' ox. xn . . then for q E U we can express every vector v E Mq uniquely as " . For tangent vectors l (X 0 1f) O . . If (x.Riemannian 1I1el. . . . . x l . q') r> v(q. q') denote the unique vector v E Mq of length < e such that = q'. . Then (q. (v) axi n We will denote ai by Xi (v). q . .cJ 335 Before proving the next result. the map expq maps the open eball in Mq diffeomor phically onto an open set Uq :::I W. .=1 where 1[ : TM + M is the projection. x l . C TM. U) is a coordinate system on M. . . v(q.. (3) For each q E W. so that v= . I I ' rr(v) q E U we therefore have the vectors while the vectors a/axt are all in the tangent space of the submanifold Mq a/axi l v span a complimentary subspace. 1 4.
Theorem 13 says that the vector 0 E Mp has a neighborhood V in the manifold TM such that exp is defined on V. Now. : Mp is the identity..p) F 0E F. consider the geodesics through q of the form I r> eXpq (lv) f IIvll < e. The close analysis of or geodesics depends on the f ollowing.. for . then + M is projection on the i th is a coordinate system on V x U. V) be a coordinate system around p. : M x M factor. p) E M x M.336 C/zapw 9 PROOF.(v). p) = . so maps some neighborhood V' of 0 diffeomorphically onto some neighborhood of (p. •:. ( a�i IJ a 1 . ( � IJ a!.1 (V). Choose W to be a smaller neighborhood of p for which F(V') :J W x W. F.. We will use the coordinate system described above. These fill out Vq . = 0 E Mp we have (P' p) + Mp (p . using the fact that (expp)o. it is not hard to see that at Given a W as in the theorem. Let (x. . We may assume that V' consists of all vectors v E Mq with q in some neighborhood V' of p and IIvll < e. Define the Coo function F: V + M x M by F(v) = (. If ". I + a!i I F. exp(v» . is oneone at Mp . and q E W. a . !2 . Consequently.
Let = constant < e}. I) are omitted. LEMMA (GAUSS ' LEMMA). in which the arguments (u) I) and Ci(U. v: lR + Mq be a smooth curve with IIvU) II k < e for all I. In Uq .Riemannian Metrics 337 1 5. for convenience: The first term on the right is Similarly. Thus the second term on the right of(2) is also O. au al expq(O) = q. It follows that (� ' � ) = 0 for all (u. aCt aCt � (U. which is just twice the second term on the right of (1). But aCt/ au(u. vU» I <u< 1.I) ' at (U. I) = o. the geodesics through q are perpen dicular to the hypersurfaces { expq(v) . I) )=0 f all (u . IIvll FIRST PROOF. . and define Ct(U. I). we obtain 0 since each curve u r> Ct( u l) is a geodesic. I). vU» .I) = ( � ' � ) is independent of u. I) = a constant We are claiming that for every such Ct we have ( = expq (u . au al . so lIaCt/aull = k . where IIvU)1I = k. I) is just the tangent vector at time u to the geodesiC u r> expq ( u . So But Ct(O. so aCt/al(O. or A calculation precisely like that in the proofofTheorem 1 2 proves the following equation.
0). = (t) r dl =  (� O ( 1\ 2 dl = k2 . ddly ( I ») ('!t (0. •:.l) = Then fJ is a variation of the geodesic y(t) = expq(t .{q} be a piecewise smooth curve. _ so 1 6. and define expq(t . But each curve ('!t (0. COROLLARY. .. . I).( a.bj + Uq . � (u) has energy . Let c : [a. Tt(I) ) . (0. � u» 10' 1 0 = dE�(U» lu� o ( = dg �. By u). v(t» . c(t) = expq(u(t) . we have = dE(du(u» 1u=o � E(  = the integral vanishing since y is a geodesic.(1») . v(u» fJ(u. v(O» . I). the first variation formula. defined on [0. I )..338 SECO}fD PROOF Chapter 9 Let v: IR + Mq be any smooth curve with IIv(t) II = a constant (note carefully the roles played by I and k < e for ali I. I]. ddly (0») au au afJ dy .
We can assume that q' = expq(rv) E Uq . and hence 2: lu(b) .for ° < u(l) < e and II v (l ) 1 I = Riemannian 111ellics 1.' iii = ) = e(l) = Ct(u(l). PROOF with equality if and only if u is monotonic and v is constant.8.u(a)l. Then 339 L�e 2: lu(b) . a. u is monotonic and v is constant. aCt + iii ' we have with equality if and only if aCt/al with equality if and only if I �r = + 1 �1 t I�I t lu'(M dl 2: lu'(I) l dl I �II = I . ] + M be any piecewise Coo path from q to q'. then di de Since ( aCt aCt = 0.{q} (otherwise break e up into smaller pieces). so that c is a radial geodesic joining two concentric spherical shells around q . Let W and e be as in Theorem 15. and clearly e must be a reparameterization of y f equality or to hold. v(l» . I) and u au (I) aCt . So the length of e is 2: r. By Corollary 16. 1 7 . the path c must contain a segment which joins the spherical shell of radius 8 to the spherical shell of radius r. 2 2: lu'(I)12 . v'(I) = 0.u(a)l. Thus 0. •:. let y : [0. and lies between them. Then I L(y) ::: L(e). •:. For 8 > 0. I) = expq(u . with equality holding if and only if c is a reparameterization of PROOF y. I ] + M be the geodesic of length < e joining q. q' E W. and let e : [0. . COROLLARY. the length of this segment has length 2: r .. If Ct (u.
) ' ) are Coo manifolds with Riemannian metrics. Since there is a great circle through any point of sn in any given direction) these are all the geodesics. All other pairs of points have a unique geodesic of minimal length between them but an infinite family of nonminimal geodesics. )) and (M'. )'. reflection through a plane £2 C )Rn+l is an isometry I : sn + sn. For example. Then I(C') is a geodesic of the same length as C' between I(p) = p and I(q) = q. If (M. the sphere have a continuum of geodesics of minimal length between them.340 Chapw 9 We thus see that su fficient6' small pieces of geodesics are minimal paths for arc length. which implies that C' c C. then the length of c with respect to ( . so that C is a geodesic. even among nearb paths. and if c is a geodesic. if we first introduce a notion which will play a crucial role later. then f o e is likewise a geodesic. depending on how many time� the geodesic goes around the sphere and in which direction it starts. ). For the isometry I : sn + sn mentioned above. . ( . q E C be two points with a unique geodesic C' of minimal length between them. the fixed point set is the great circle C = s n £2 Let p. then a oneone Coo function f : M T M' is called an isometry of M into M' if f*( . 1] + M is a Coo curve. ) ' = ( . without any computations. Antipodal points on ). ) is the length of f o e with respect to ( . ( . It is clear that if c : [0. So C' = I(C'). We can use Corollary 17 to determine the geodesics on a few simple surfaces. n Notice that a portion of a great circle which is larger than a semicircle is defll1itely not of minimal length.
. b] can be extended to a geodesic defined on all of lR. a manifold M with a Riemannian metric ( .L + JR2 by rolling Z onto JR 2 .. .Riemannian 111etlicJ 341 The geodesics on a right circular cylinder Z are the generating lines. Notice that on both the sphere and the infinite cylinder every geodesic y defined on an interval [a. ) is called geodesically complete if evelY geodesic y : [a. a bounded portion of JR". or JR" . q) = inf{L(y) : y a piecewise smooth curve from p to q}... This is false on a cylinder of bounded height. ) and the metric d : M x M + JR it determines.{OJ.. .I of the straight lines in JR2 Two points on Z have infinitely many geodesics between them. the . d(p. The geodesics on Z are just the images under I. We are now in a position to wind up our discussion of Riemannian me\ rics on M by establishing an important connection between the Riemannian metric ( . b] + M can be extended to a geodesic from JR to M. In fact) if L is a generating line of Z) then we can set up an isometry I : Z .. In general. and the helices on Z . circles cut by planes perpendicular to the generating lines.
q) = r . q) = r . There is a point on PROOF.q). d(y(t). since every curve from .342 Chapw 9 18. q) for all s E S. So d(po. r] be the least upper bound of all I f which (**) holds. 8 < e. Then d(y(to). q) :. Suppose M is geodesically complete. q). We claim that expp (r v ) = q.(d(y(to). q) = Let S C Up be the spherical shell of radius po = expp 8v. To prove this result. ) is a Riemannian metric on M.q) = �ir (d(p s) + d(s. choose Up as in Theorem 14. II v ll = 1 this will show that the geodesic y(t) = expp (t v) is a geodesic of minimal length between p and q. q E M with d(p. we clearly have d(p. Moreover. drs.r]. .8. Given p. ). First of all. This proves that (**) holds for I = 8. Then or (**) holds for 10 also. we will prove that S such that d(po. old r > 0. Suppose 10 < r.I IE [8.. q») = 8' + d(po' . q) = ��. If ( . Now let 10 E [8. of radius 8' around y(to) and let po' E S' be a point closest to q. THEOREM (HOPFRINOWDE RHAM). s) + d(s . q») = 8 + d( po. Let S' be a spherical shell p to q must intersect S. then M is geodesically complete if and only if M is complete in the metric d determined by ( . any two points in a geodesi cally complete manif can be joined by a geodesic of minimal length. by continuity.
showing that (**) holds for 10 + 8'. so it must be that 10 = r. q) = 10 + 8'. Hence y(lo + 8') = po'. and p E A .Riemannian Metrics so 343 d( po'. •:. q) . But the path e obtained by following y from p to y(lo) and then the minimal geodesic from y(lo) to po' has length precisely 10 +8'. which means that it coincides with y. so it converges to Some point p E M. note that there is always a min imal geodesic joining any two points of a compact manifold. From this result. d(y(lo + 8'). Conversely. Hence d(p. Given any geodesic y : (a. q) = r . So e is a path of minimal length. suppose M is complete as a metric space. it is not difficult to show that y can be extended past b. q) = (r . if A C M has diameter D. III other words. by a least upper bound argument. bounded subsets of M have compact closure. any geodesic can be extended to ffi. fact.8'. In Hence (***) gives . which proves (*). Consequently. Clearly y(ln ) is a Cauchy sequence in M. choose In + b.d(po'. From this it is clear that Cauchy sequences converge. This contradicts the choice of 10. . Using Theorem 14. b) + M. it follows easily that M is complete with the metric d. (**) holds for I = r. I n other words. and must therefore be a geodesic. As a particular consequence of Theorem 18.10) . then the map expp : Mp + M maps the closed disc of radius D in Mp onto a compact set containing A. po') 2: d( p.(10 + 8').
On the other hand if M is the Mobius strip and S I C M is a circle around the center. i. For example.C Np as MpL = {v E Np : (v. the normal bundle of M in N. . E + M take MpL to p. then we can define M/. so that for every p E M we have i. then it is not hard to see that the normal bundle v will be isomorphic to the (nontlivial) bundle M + S If we consider S I C M C J1> 2 . so it too is nontrivial. then the normal bundle 1 of S I in J1> 2 is exactly the Same as the normal bundle of S I in M. for v has a section consistiIlg of unit outward normal vectors. ) is a Riemannian metric for N. It is not hard to see that v = w : E + M is a kplane bundle over M.w) = 0 f all w E Mp} . with i : M + N the inclusion map. the normal bundle v of sn 1 C IRn is the trivial I plane bundle. or p eM Let E= U Mp L and w .(Mp) C Np ' If ( .344 Chapw 9 ADDENDUM TUBULAR NEIGHBORHOODS Let M" C Nn+ k be a submanifold of N.
{OJ. or then D is a submanif oldwithboundary o f U. and for which the Osection s : M + U is just the inclusion of M into U. If 1f : U + M is a tubular neighborhood. and H k (U) "" Hk (M). so 1f is a deformation retraction. and the map 1f I D : D + M is also a def ormation retraction. Let X be a compact metric space and Xo C X a closed subset. then clearly 1f 0 S = identity of M. Before proving the existence of tubular neighborhoods.e) ::: I } . But in general the neighborhood cannot be all of N. for which the Osection s : M + U is the inclusion of M in U. In the case where N is the total space of a bundle over M.Riemannian Metric. we add some remarks and a Lemma. Then there is a neighborhood U of Xo such that IIU is oneone. For example. thus M has the same de Rham cohomology as an open neighborhood. 1 9. is smoothly homotopic to the identity of U. Let I : X + Y be a local homeomorphism such that IIXo is oneone. if we choose a Riemannian metric ( . is called a tubular neighborhood of M in N. ) f 1f : U + M and define D = {e E U : (e. So M also has the same de Rham cohomology as a closed neighborhood. A bundle 1f : U + M with U an open neighborhood of M in N. s 0 1f . LEMMA. 345 Our aim is to prove that for compact M the normal bundle of M in N is always equivalent to a bundle 1f : U + M for which U is an open neighborhood of M in N. Moreover. as an appropriate neighborhood of S l C )R2 we can choose )R2 . this open neighborhood can be taken to be the whole total space.
•:. which is equivalent to the normal bundle of M in N. since exp is oneone on M e V" it f ollows from Lemma 19 that for sufficiently small < the map exp is a diff omorphism on E. 1] y One of the illteresting features of Theorem 20 is that all the paraphernalia of Riemannian metrics and geodesics are used in its proof. = V E. there is e 0 such that g 2: 2< on C. Let V e E be the set of a noncritical points f expo Then V :J M (consid or ered as a subset of E via the Osection). where we will also need the following modification. while they do not even appear in the statement. y: [0. the map or or exp is a diff omorphism f e rom E. Xo) + dry. then g 0 on C. that exp is defined on E. norm I I . is compact. x + Choose a Riemannian metric ( . with the corresponding It f ollows easily from Theorem 13.. = {v E E : I v l < e ) U. Then M has a tubular neighborhood 1f : U + M in N. Let E = {v : v E Np and v E M/".) C U. f if (Xn l Yn) is a sequence in C with Xn T x and or T y. onto U. e It is clear also that exp(E.346 PROOF. Then f is oneone on the {(x. Since C is compact. •:. . x # y and f(x) = f(y») . Xo). This means that = expp / O) where E E. for some p E M } E . THEOREM. . If g: C JR is g(x. y) = d(x.y) E X > = X: Yn + > old 20. PROOF. . ) for N. then f(x ) = lim f(xn) lim f( Yn ) = fry). and compactness of M. Chapter 9 Let C C X x X be x <neighborhood of Xo. and metric d: N N JR. n q q y (0) y ( 1 ) q. M) < e). and also x # y since f is locally oneone. To prove that exp is onto U" choose any E U" and a point p E M closest to If + N is the geodesic of length < < with = p and = it is easy to see that is perpendicular to M at p (compare the second proof of Gauss' Lemma). = {q E N : d (q.. This will clearly prove the theorem. Theorem 20 will be needed only in Chapter I I . and V. Let M e N be a compact submanif of N. We claim that f sufficiently small e. dyjd ( dyjdl(O) q. f sufficiently small < > O. Then C is closed.
•:. PROOF Exactly the same as the proof ofTheorem 20. Then aN has (arbitrarily small) open [and closed] neighborhoods f or which there are deformation retractions onto aN. THEOREM. . with compact bound ary aN.Riemannian 111etrics 347 2 1 . Let N be a manif oldwithboundary. using only inward point ing normal vectors .
. v). · v*j n q ( . /\ ¢k = 1/11 /\ . Conclude that W = W". ) be a Euclidean metric for V*. then f some aij . 3. ) an inner product on V which is not necessarily positive definite. show that ¢1 /\ . . Vn for V is called orthonormal if ( Vi. If ¢i = Lj aj. /\ ¢k ¢ O. ab and . (The sign is + if ¢1 . . Then show that q(v) = h ( v. ¢ (b) Let U1 .v) = v. . show that this volume is the same for 1/11 . .q (u) 0. ' " ¢k is det(aij ). 1/Ii E V* satisfy ¢1 /\ . and ( .q(u ) . • /\ Vk ofvectors Vi E V. show that the signed kdimensional volume of the parallelepiped spanned by ¢1 .) (c) Using Problem 79. Let ( . and q(2u) 4q(u). v) (c) Suppose q : V + F satisfies q( . 1/Ik. . (a) Show that W E W if and only if W /\ ¢1 /\ . . A basis VI . . • . he deduces geometrically the COlTcsponding conditions on the coordinates of Vi. or (b) Show that F by q(v) = h (v. = .q(v) ] . 2.q(v) . ' . + Let V be a vector space over a field F of characteristic # 2.q( w ). Suppose ¢i. Show that q (u+ v + w ) . . . and let h : V x F be symmetric and bilinear.v) = q(v) h ( u . .q( u ) . and let W and W". /\ 1/Ik # 0. j =l L Qij v*. and that iJ(u. = = q(u + v) .q( u + w ) . . Let V be an ndimensional vector space. E. + (a) Define q : V for V*. Show that if ¢1 . /\ 1/Ik . " " ¢k has the same orientation as al l ' .q ( u ) . be the subspaces of V* ¢ spanned by the ¢i and 1/Ii. . = q(u+v) . so that we have a wedge product VI /\ . " " ¢n is a basis q= i. .348 Chapw 9 PROBLEMS V I.uj . • /\ ¢k = 1/11 /\ . Cartan uses this condition to define n k (V*) as formal sums of equivalence classes of k vectors. . . Uk be an orthonormal basis of W¢ = W". (d) Conversely. . if W = W"' . Ifwe identify V with V**. Vj ) = ±8ij. . /\ Wk In Lcfons sur La Ceometrie des Espaces de Rieman1l. . . .othe:rvvise. v) q(v) is bilinear. . and the signed volumes of the parallelepipeds are ¢ the same. Wi. then we have a geometl"ic condition for equality with WI /\ . v).q(u) . . .q (v + w) Conclude that '1(0) =H = q( u + v) . .
. then V = W Ell WL . there is an isomorphism f: IRn + V with f*( . Show that these inner products agree with the ones defined above. X Vn _ ' . w) = 0 for all W E W } . ) = ( . ) k on n k (V) by requiring that 1 � i1 < . define inner products on 0k V and A k V by using the isomorphism V + V* given by the inner product on V. Thus. (c) I f ( . . then there is a vector v E V with (v. ) has index i . x . . . /\ . . ) i s also nondegenerate on WL . < ik � n be an orthonormal basis. • . ). S how that the index is n . v) # O. then = . V n be an orthonormal basis (see Problem 3). (b) Show that lv. (d) For those who know about 0 and A k Using the isomorphisms 0 k V* "" (0k V) * and A k (V*) "" (A k V)*. . 1/Ij )* } = det( ¢i. is defined on page 301). Vk. /\ 1/Id (c) I f ( . v" 4. and ( .} is a basis for W. /\ ¢k o 1/1. ) is the largest dimension of a subspace W C V such that ( . x .  . 5. where gij = (Vi . x . . consider the linear functionals Ai : V + IR defined by Ai(v) = (v. ) i s nondegenerate o n W.Riemannian 1\1etries 349 (a) If V # {OJ. with (a) Show that ( . Hint: Apply the result on page 308 to a certain (II I )dimensional subspace of IRn . ) k is independent of the basis v" (b) Show that (¢' /\ . . x V. I = . .. Prove that dim WL 2: n . (b) For W C V._. X Vn _. in Problem 726. Hint: If {w. Recall the definition of v. Wi ) . .) det( ¢i. . (e) The index of ( . . for some r (the inner product ( . and let . V i ) = O. Vj ) . . (a) Show that (v. thus showing that r is unique ("Sylvester's Law of Inertia"). Define an inner product . 1/Ji}').Jdet(gij ). . ) I W is negative definite. . ) be a (possibly nonpositive definite) inner product on V. ).r. let WL = {v E V : (v. Let ( .dim W. . (d) V has an orthonormal basis. (Use Problem 716 .
Show by a computation that if the functions . i}�� ' i.. g C). )p on each 1[ ." with det(gij) # 0. )p is constant on each component of B . (a) There is no way of continuously choosing a I dimensional subspace of S 2p. . show that S is diffeomorphic to S'. Define a tensor B of type @ by gil a '· a .. and the functions gij . ) and ( . of course. ) be two Riemannian metrics on a vector bundle � 1[ : E B.."". If � is a smooth bundle over a manifold M. So If the expression for A in a coordinate system is . is the classical way of defining the tensor [having the compo nents] 10. '. ) be a Riemannian metric on M.350 Chapter 9 6.� 'r:t� _ .) (b) There is no Riemannian metric of index 1 on S 2 gij and g'ij are related by axi axj g'r:t� = Lgjj a '. + = ' X /J then g This. a . An indefinite metric on � is a continuous choice of a nonpositive definite inner product ( . Show that S is homeomorphic to S'. Let ( .j x '  8. Let S be the set of e E E with (e. and define S' similarly. g ij are defined by 9. This problem requires a little knowledge of simpleconnectedness and cov ering spaces. Show that the index of ( . for each p E S 2 (Consider the space consisting of the two unit vectors in each subspace. Let � '= 1[ : E + B be a vector bundle.j axi axj  L. and A a tensor of type that A (p) : Mp + Mp .1 (p). 7.e) 1. (a) Let ( .
gkI A{ . Yn with (YI. where n Blk = 'L. . Let M C IRn be an (n .. . S how that if we consider v(p) as an element of IRn . define a tensor C of type (�) by C (p)(AI . . Y ) j ±81j in a neighborhood of any point. ). Show that the GramSchmidt process can be applied to the vector fields all at once. . . then n ckj 'L. (Vn I)p is positively oriented in IRnp when (VI)p. (b) Let d VnI be the volume element of M determined by the Riemannian metric it acquires a submanifold of IRn . . A2) = (A(p) * (AIl. as () . (vnIlp is positively oriented in Mp. show that the area of the surface obtained by revolving the graph of 1 around the xaxis is t 21f1JI + (/')2 . (a) If M = aN for an ndimensional manifoldwithboundary N C IRn . .A2)' Show that if C has components C kJ . . . . so that we obtain n everywhere orthonor mal vector fields Yh . . then V (p) dVn_l(p)( VI)p. . i=1 The tensors B and C are said to be obtained from A by "raising and lowering indices" . . . . _I)p) = det �I . . Ai gjk · j=1 (b) Similarly. Yn . (a) If I: [a.Riemannian 111etr. (a) Let XI. then v(p) is outward pointing in the sense of Chapter 8.cs show that = 351 B LI. find Yh . 12.. (b) For the case of a nonpositive definite metric. Vnl = = E /1. (V. (b) Compute the area of S2 13..k Blk dxl ® dxk . . 1 1 ..I)dimensional submanifold with orientation The outward unit normal v (p) at p M is defined to be that vector in IRnp oflength I such that v (p). (VI)p.bj + IR is positive. . Xn be linearly independent vector fields on a manifold M with a Riemannian metric ( .
x . /\dxn (p).. Denote the volume element of M by dVn .) /\ . Show that for IRn we have n v(p)) . IRn . . x vnd. /\ dxn . Denote the volume element of M by dA. /\ dxn (p). . v) dVn1 JM JaM (the function div X is defined in Problem 727). . v ) dA = [ (X. C Slfikes' Theorem: J1.) /\ . V _1 = v(p) for some IR (by Problem 5).i /\ . with odentation and outward unit normal v..  . T) ds JaM JM x X is defined in Problem 727). Let X be a vector field on M. . (a) Let v" be the volume of the unit ball in (V [ Cv x X. . Prove the Divergence Theorem: [ div X dVn = [ (X. Prove (the original) W E (w . and that of aM by dVn _ l . i=1 (e) Let M 1R3 be a compact 2dimensional manifoldwithboundary. (d) Let M c IRn be a compact ndimensional manifoldwithboundary. W n w 14. dVn_1 (p) = restriction to Mp of (l)i'dx' (p) /\ . . ' . Hint: Consider the form on M defined by = L(. X Ci Ci E X . X (w. /\(fXi(J. /\ (fxi(J. with v the outward unit normal on aM.l )iI ai dx' /\ .x2)(nI )/2 Vn_ 1 dx. /\ d. (VI VnI> v(p)) = V.•. Let X = Li a i a/ax i be a vector field on M. Show that Vn = /'I (1 ... . Conclude that vi (p) .. and that of aM by ds. . (c) Note that VI x .352 Conclude that Chapter 9 dVn_1 (p) is the restriction to Mp of t(l)iI vi (p) dx' (p) /\ . Let T be the vector field on aM consisting of positively oriented unit vectors. . .
1fnf2 r(l + n/2) . ) = Li dx i ® dx i . Hint: If the inscribed polygonal curve is determined by the points (Ii . but reversing the order of integration. show that nI (c) Using V. I] l' i=1 L [(ci )'(I)j2 dl. c(lll) for . Show that I l ' rnI An1 dr = AnI . show that this + [/'(1)]2 dl .. n n even ' 2. In = . 15. l' jl + JR2 given by C(I) = (l. show that I (:��:! o 2 (n +' )/2. this can be written (d) Let A n. with X(p) = pp . n JRn has the usual L(c) = (b) For the special case length.In2 . (1 .I be the (11 . V2 = IT.) Vn = (e) Obtain this same result by applying the Divergence 111eorem (Problem 13). is the least upper bound of the lengths of inscribed polygonal curves. c: [0. f(I» . Using the method of proof in Corollary 88.(nI)/2 I · 3·5" 11 n odd. where Riemannian metric ( .x2)(nI)/2 dx = .Riemannian Metlies 353 (b) If In = ['. I] + JRn be a differentiable curve. show that Vn = (In terms of the r function.I )volume of S nI . (a) Let c: [0.
In this way. Ii ] . The topic of surface area for nondifferentiable sUlfaces is a complex one. while the area of each approaches hi /2. this least upper bound is infinite for a bounded portion of a cylinderl To illustrate Schwarz's example I have plagiarized the following picture from a book called Maml�Mamu't{eCKu. . Schwarz first observed.planes of the hexagons closer together.li_I l2 + (J(lt) . we maintain the hexagonal arrangement.li _I l 2 + f'(�i )(li . on a compact set. but as H. (c) Prove the same result in the general case. CmmaK. the least upper bound of the areas of inscribed polygonal surfaces. written by someone called M. bUl mow the. Top view To in(T("asc the number of triangles.li_I l 2 for some �i E [Ii _ I . . 1]. Hillt: Use the results of Prob = 10 0 < . which we will not go into here.c(liI ll = J(li . then we have IC(li ) . and uniform continuity of "'.J(ti_Il)2 J(li . similarly.354 a partition Chapter 9 1 [0.u A1fClJlUJ 'Ua MrIOZOo6jJa3URX. we can increase the number of triangles indefinitely. so that the triangles are more nearly in a plane parallel to the bases of the cylinder. < In = of = lem 81. . It is natural to suppose that the area of a surface is.
Show that the metric on M may be defined using Coo. /'(1» dl] dl. d d 19. (Show how to round offcorners of a piecewise Coo path so that the length increases by less than any given £ > 0.S or does no/ have to be disconnected. q) ? min (p . by showing that M . Use q!eS this fact and Lemma 7' to complete the proof of Theorem 7.e] dl = 0. then 1 still satisfies the Euler equations (which are not a priori meaningful if 1 is not C2). I) [a au a a Y this result makes sense even if 1 is only C'. (c) Conclude that if the C' function 1 is a critical point of J. f M . show that 17. remember that the f ormula for length involves only first derivatives. (In the theory of infinite dimensional manifolds. show that (p.S. . g o n [a. If p : [0. a 1rbg(u)du.S is disconnected. (b) D u Bois Re )'mond� Lemma.b] satisfies 0.b] with Hillt: TIle constant c must be lb �'(I)g(l) dl = ° __ l b a � (a) = � (b) g is a constant. I) + M be a curve in a manifold M with a Riemannian metric ( .S and q E M . lb �' (I)[g(l) . If a continuous function . ). then = (1.) d L (e) = L(e 0 pl.l' �� (1'/(1).Riemannian 1I1ettics 355 16.B.S are disjoint open subsets of M . 1 (I» . 1] + [0.e. Show that M . for all Coo functions � on [a. 1] is a diffeomorphism. (b) If p E B . e= We clearly have so we need to find a suitable � with �'(I) = g(l) . (a) By applying integration by parts to the equation on pages 3 1 8319. these details become quite important. show that (d& (u» I dJ U u=o = a2Ci aF lb t (0. q').) 18. Let e: [0. (a) Let B C M be homeomorphic to the ball {p E JR" : Ipl :s: I } and let S c M be the subset corresponding to {p E JR" : I p l = I}. instead of piece wise Coo curves. /(1). and Theorem 7 is false.B and B .
. situated.and tanh.denotes the inverse of cosh I [0. and hyperbolic tangent functions sinh. coshx (a) Graph sinh. and tanh are defined by sinhx = . ) I cosh x = cos ix. at a and a... hyperbolic cosine.x) = cosh(tanh. 00).356 Chapter 9 20. Show that sinh(coshI x) I I = cosh(sinh.x) = J 1 + x2 1 v1x R=I r:. (c) For those who know about complex power series: sinh x = sin ix .' eX + ex 2 sinh x tan h X = . I respectively. I I (d) The inverse functions of sinh and tanh are denoted by sinh.. 2 (cosh _ ' )'(x) = 21.' eX _ 2 eX cosh x = . cosh. and tanh. 2 _1 1 . (b) Show that 2 2 cosh . The hyperbolic sine. Consider the problem of finding a surface of revolution joining two circles of radius I . while cosh. cosh. We are looking for a function Vx �.sinh = 1 2 2 tanh + 1 / cosh = 1 sinh(x + y) = sinh x cosh y + cosh x sinh y cosh (x + y) = cosh x cosh y + sinh x sinh y sinh' = cosh cosh' = sinh . for convenience.
and sketch the graph.1 . then there are two such c.Riemannian Metrics of the 357 form where e is supposed to satisfY x /(x) = c cosh c (c > e cosh :!.. Examine the sign of 1/ y tanh y f y > O. Aa (e) = c cosh :!. or (b) Examine the sign of cosh y . namely c = a/yo = I / cosh yo. find the value of Aa there. show that � = JY02 . = I e 0). (a) There is a unique Yo > 0 with tanh Yo = 1/ Yo.] . 67 . (c) Let c > O.I � . then there is a unique such c. with CI < a/yo < c'}. If a < yo/ cosh yo. (d) There exists c with c cosha/c = I if and only if a :s: yo/ cosh Yo. It turns out that the surface for C has smaller area. cosh Yo [YO � 1 . 2 c. c Show that Aa has a minimum at a/yo.y sinh y for y > O. so JY02 . yo/ cosh Yo 0 a Yo/ cosh yo (e) Using Problem 20 (d).2. If a = yo/ cosh Yo.
so the envelope consists of the straight lines c cosh  x c = c cosh Yo. but the graph of /c.x . the graph of f is tangent to q cosh Yo :' y = � x/ . pp. Q E (a. .= ±Yo. . I) is tangent to the envelope at that point. cosh Yo y = ± . TIle point Q is called conjugate to P along t he extremal hi ) and it is shown in the calculus of variations that the existence of this conjugate point implies that the portion of /cl from P to (a. ftr a < )'0/ cosh )'0.a].[ Volume III.358 Chapter 9 [These phenomena can be pictured more easily if we use the notion of an envelopec. 176ff. y= = cosh :: _ c c :: sinh C :: (' . a). '" a Yo/ ('osh Yo the envelope at points P. The envelope of the I parameter family of curves x /c (x ) = c cosh is determined by solving the equations 0= We obtain x c a/c (x) oc . I) does no/ . Yo The unique member of the family through (yo / cosh Yo . is tangent to the envelope at points outside [a.
1 359 give a local minimum for conjugate points of a geodesic in Volume IV.� l. ax'� ax k L.)] 22.k=1 ax ax'Y ax'· ax'� n ax} a2x' ax' a2x) + " g'l' ox'ct . /' (1» ax (a) Show that for any and conclude that the extremals for our problem satisfy � (F f aF ) f [ aF �aF] dl ay ax dl ay _ .. aF = 0. All of our illustratiOlls of calculus of variations problems involved an F which does not involve t.dt ay (/(1).. and note the remark on pg. Q. _ f and F: 1R2 + IR we have ( ) = O. (a) Let x and x' be two coordinate systems.j. i.� ax'Y  x . y] ' = so that [ij. 'l ax'.. ag'.+ ax'/3 (Jx'l3ax'Y ox.) For the corresponding [a!'l. (c) Also show that i. so that the Euler equations are actually f f. i. ax) ox'Y " [}2xl ox'Y " rk.) ax. 11 n [ij.. = . ture. (Consider the arclength function on a geodesic for Some Riemannian metric on IR. /' (1» .Jl+7 to obtain directly the equation dy/dx = _ � ag'J �� ax} k L.. k] and [a!'l. . with corresponding gil and g 'i) for the expression of a Riemannian metric. ax'� ax'Y + L ax'Y .1J + (/') 2 . ax'· ax'� ax' . show that 24.ct(Jx'Y .396. Y]'.f (b) Apply this to F(x.. Show that � which we eventually obtained in our solution to the problem.Riemannian 1I1ebic.k] are not the components of a tensor.j. Chapter 8. k] ax'. k=1 1=1 Show that any Coo structure on IR is diffeomorphic to the usual Coo struc Y r '. + L. V. F. (Compare with the discussion of aF d aF (/(1). � = __ _ __ _ .k=1 11 n ax' a2xJ ax' ax} axk L [ij.=1 (   ) . y) = . ax'� ax' ' )=1 . ay 23.}.
sufficient t o solve for the ei .) + Bk). where u l . Ifwe consider ei as a vector field on JR". j. (c) To solve for I i t is. ). In Lcfons sur La Ciomitrie des Espaces de Riemann.i gij du i ® du J be another metric. directly from our equations for a geodesic. • • • ) un again denotes the . since they correspond to lines parallel to the x i axis.r A k) r=1 Bik . e) ) . ill theory a t least.) Bi). E. ) = Li dx i ® dX i be the usual Riemannian metric on JR". and let L.J standard coordinate system on JR" . (b) Show that gil = (ei. go about finding I? satisfied by the ei 's.k = [ij. and to solve for these we want to find differential equations 25. Let ( .) = r[. (f) Deduce the result A. How can we .360 Chapter 9 (a) Let al/au i = ei : JR" + JRn . Cartan uses j this approach to motivate the introduction of the ri� . . k]. deduce that so that A� = rD. Show that we must have (d) Show that = = " L g)r A �k + g. (Note that the curveS obtained by setting all but one Ii constant are geodesics. show that I (a/au i ) = ei . . k. Suppose we are told that there is a diffeo morphism I : JRn + JRn such that Li ) gij dui ® du) = I' ( .i· (e) By cyclically permuting i .
) on V = V' Ell V" by (V' E9 v".b) + M be a geodesic. ( . Show that for v E Mc (r ) with (v. we define ( . then y is a geodesic. (c) If e satisfies for Il' JR + JR. (a) Show that ( . then e is a reparameterization of a geodesic. de/dl) = constant.) . w")".) ds. Show that e = y o p satisfies p : [". (The equation p"(I) = can be solved explicitly: p(t) = J' e M (. w' E9 w") = (v'.Riemannian Metlies 361 26. (a) Let y : [a. n and (V". if e satisfies this equation. (Gauss' Lemma is the '�special case" where c is constant. where M'(s) = /1(5). Describe the geodesics on M x N for this metric. 27. ( . and consider the hy (v. and let diffeomorphism. ) ") are two vector spaces with inner products. v is perpendicular to these hypersurfaces. where v E Mc (r ) with = OJ.b) be a (b) Conversely. Let c be a curve in persurfaces {exPC (r ) v : Il v ll = M with dc/dl # ° everywhere.) p'(I)/1(I) 28. If (V'. I'l) + [a. de /dl) 0. it follows that there is a natural way to put a Riemannian metric on M x N. w')' + (v". ) is an inner product. the geodesic U r> eXPc(r) u . (b) Given Riemannian metrics on M and N.
y(b» . b] + M with d(y(a).362 Chapter 9 29. ba ba L� (e) = d(e(a). In particular. . sufficiently small pieces of a geodesic mil limize energy. Vn of MP l so that we have a �Crectangular" coordinate system X on Mp given by Li a i vi r> (a t . If ( . If era) = y(a) and e(b) = y(b).a)E� (y). a n ) .) 30. (c) Let y: [a . .. b] + M be a geodesic with y(a) = p. Let y: [a. . then a curve y : [a. . ) is a Riemannian metric on M and d : M x M + IR is the corre sponding metric. . (a) Prove Schwarz's inequality by imitating the proof of Theorem 1 (2).t . .5 E(e). 31. so that each y k is linear. .5 (b . show that [L�(Y)f . Hint: Recall the equations for a geodesic. b) + M be a geodesic with L � (y) = d(y(a). (Gauss' Lemma still works on exp«('J). y(b» = L(y) is a geodesic. let x be the coordinate system xo exp. .. Schwarz's inequality for continuous functions states that with equality if and only if f and g are linearly dependent (over IR). Conclude that E(y) < E(e) L(y)2 L(e) 2 E(y) = . and note that a geodesic y through p is just exp composed with a straight line through 0 in Mp.5 . defined in a neighborhood U of p. e(b» . among all curves in exp«('J) . Let p be a point of a manifold M with a Riemannian metric ( . and suppose that expp is a diff omorphism on a neighborhood ('J C Mp of {ly ' (O) : 0 ::s I ::s I ). unless e is also a geodesic with (a) Show that in this coordinate system we have r6(p) = O. Choose a basis VI . (b) For any curve y show that with equality if and only if y is parameterized proportionally to arclength. Show e that y is a curve of minimal 1ength between p and q = y(b). ) . 32. .
(g) Let f: U + IRn be a diffeomorphism of a neighborhood U of 0 E IRn into IRn. then 0 (e) Let q and '1' be two points with r (q) . d(r y)/dl o.). roy 2. the maximum of occurs at either q or q ' .q). = {v E Mp II v ll = d. u (O) .. 0 so that (d) Let B. 8)...Riemannian 111etricJ 363 = r(q) d(p.J d k d I is sufficiently small. '.ball is convex. then Using part (a). and this geodesic lies completely in U. i. So that r o y Lk (yk)2. dl .) for 0 # I E (8.J. Show that for sufficiently small . > O. Show that d2 ( r o d) = 2 [ . (f) A set U C M is geodesically convex if every pair q. r(q') < . Show that exp ( { v E Mp : IIvll < d) is geodesically convex for sufficiently small . Show that for sufficiently small " the image of the open . dl 2 L. y y o = tangent to exp(S. dl dl k (c) Note that l L ddyi dyl n2 L( dylk )2.) and such that y'(O) is tangent to exp(S. ( dyk )2 _ " Y2rk dyi dyJ ] .k iL.)..p B. q ' E U has a unique geodesic of minimum length between them. and let be the unique geodesic of length < joining them. then there is Ii > 0 (depending on such that >t exp(B. {v E Mp : IIvll :s: 'I and S. Hint: If ' ( ) is y(O) d(r y)2/dl = y) y(l) s.. conclude that if Il y '(O) I (b) Let r: U + IR be = 'i I :s: is strictly increasing in a neighborhood of O. Show that the following is true for all sufficiently small ' > 0: if is a geodesic such that E exp(S. y .
1 d�. < for all tangent vectors (c) Conclude that I d(u��(t» I [ h = = lu i . h�o Ic(h) .O) to (1. then there = q Hint: In Mq we clearly have Since expq is locally a diffeomorphism there are 0 < K.364 Chapter 9 33. f(I» in ]R2 such = Hint: Make c look something like the following picture. c (h» L (c I [0. h] # lim c(t) (I. Show that if is some K > 0 such that for all near 0 we have I I u'(t) I 1 5. 1. h]) .'=__"::::__"10 _ u'(t)2 + I '!.O) is ! (b) Consider the situation in Corollary except that > 0 for near o. (a) There is an everywhere differentiable curve that length of c 1 [0 ..ICi (U (I). ..:. I C(I) V(I) if and only if approaches a c is C'.c (O)1 1.I) 1 2 dl a u(h) q..O) to (!.t) I · K2 such that V at points near lim h�o lim h�o d(p. then = a.. If c is C'. and suppose limit as + 0 (even though v(O) is undefined). length from (!.0) is I length from ����+ (�.
Riemannian Metrics 365 (d) If c is C'.(y'(O» = d/(y(I» dl Show that II /'(X) II = II X II . uSe Problem 34 to show that II IX . o Let f : M + N be an isometry. W E MP l then L (exp o c) = l lim . then f 0 y is a geodesic. 34. ) and corresponding metric d. = l . Y ) II X II · II Y II = il X II · II Y II II IX II · II I Y II ' II X II 2 + II Y II 2 [d(exp IX. Y E Mp. and Yv. show that if c is the straight line joining v. exp w) . w ) = 1. Let M be a manifold with Riemannian metric ( . = p. 36. Show that f is an isometry of the met 35. lim v. (c) Given X.w is the unique geodesic joining exp(v) and exp(w).w = exp 0 cv. ric space structures determined on M and N by their respective Riemannian metrics.w ) (c) Conclude that d(exp v.w�O L (cv.w. (b) Define /' : Mp + M f(p) as follows: For y a geodesic with y(O) /. exp l y)] = hm II IX II · II I Y II II X II · II Y II _ HO I1=0 .w�O L(c) (b) Similarly. l�� II v _ w ll v. let . Let f : M + M be a map of M onto itself which preserves the metric d.(cX) = c/'(X). then L(yv. (a) Using the methods of Problem 33. and that /. v. if Yv.I Y II 2 II X II 2 + II Y II 2 2(X. show that L(c) is the least upper bound of inscribed piecewise geodesic curves. (a) If y is a geodesic.
and hence an isometlY. y(l) y(t. II v ll · cos Ii where Ii is the angle between u and v. a contradiction. # 0. and then that f'(X + Y) = IRn with w # 0. lI v + l w ll . Show that if II is ti'O hm . critical . Use this to show that f is itself a diff omorphism. Y)p = (d) Part (c) shows that 1' : Mp + Mf(p ) is a diffeomorphism. (/'(X). show that hO lim II v + l w ll . and suppose that y'(lo+ ) # y'(to) for some 10 E (0.) y(O) close enough to 10. The same result then holds in allY vector space with a Euclidean metric ( . ) to y(to + u).lIv ll . e 37. v) = lI u l i . Alternately. 1] + M be a piecewise C' critical point for length. and finally the rest of y. then Hint: (c) Let y : [0. If v : IRn + IR is the norm. Choose II < 10 and consider the variation Ci for which & (u) is obtained by following y up to I" then the unique geodesic from y(t. then dL(&(u» /dulu�o paths for length cannot have kinks. one can use the equation (u. ). w E f'(X) + f'(Y).lI v ll I = (v.366 Conclude that Chapter 9 (X. /'(Y» f(p) . (a) For v. then the limit is Dv (v)(w). TIlus. 1 ). (b) Conclude that if w is linearly independent of v. w ) II v ll .lIl w ll t # O.
and let L be a generating line.L + jR2 which is a local � P(� isometl)� but which is usually not oneone. Q» Show that evcry geodesic y : jR + pn is closed (that is. and that every two geodesiCS intersect exactly or once. . there is a number a such that y(t + a) = y(l) f aU I). y all other ri� = O. Consider a cylinder Z C jR3 of radius t. = .\' 367 38. Investigate the geodesics on a cone (the number of geodesics between two points depends on the angle of the cone.3. = r"2 = rl. and some geodesics may come back to their initial point). r�. ) = dx ® dx + dy ® dy . The sum of the angles in any triangle is > 1f. TIle Poincare upper half plane with the Riemannian metric Jf' is the manif old {(x. y) E jR' ( . )) is the usual Riemannian metric on sn (thc one that makes the inclusion of Sn into jRn+' an isometry). Unfolding C . 39. = .Ri£1nannian 1\1etfie. p }. Consider a cone C (without the vertex).. TIlese results show that pH provides a model for "elliptical" nonEuclidean geometry. (a) Show that there i s a unique Riemannian metric « . 41.y. 40.L onto jR 2 produces a map f: C .�. )) on pn such that g * « . (c) Show that there are isometries of P" onto itself taking any tangent vector at one point to any tangent vector at any other point. Let g : S" + pn be the map g(p) = [p] = {p. Y2 (a) Compute that : y > O} ! ri. Find the metric d induced by the Riemannian metric it acquires as a subset of ffi.
c. Conclude that if length A B length A ' B' and length A C length A ' C' and the angle between the tangent vectors of f3 and y at A equals the angle between the tangent vectors of f3' and y' at A'. . d E 'llI. and that we can take any tangent vector at one point to any tangent vector at any other point by some J.bc > O are isometries. show that all the geodesics in Jf2 are the (suitably pa rameterized) semicircles with center on the xaxis. c) and radius R. show that y'(t) 2 y(t) (c) Using Problem 27. A. Show that the maps z J( ) = az + b cz + d a. Considering (b) Let C be a semicircle in Jf2 with center at it as a curve t 1+ (t. then there are infinitely many geodesics through p which do not intersect y. together with the straight lines parallel to the yaxis.. then length B C = length B'C' and the angles at B and B' and at C and C ' are equal ("sideangleside"). (f) For those who know a little about conformal mapping (compare with Prob lem IV 76).. y(t)). (e) Show that if y is a geodesic and p ¢ y. b . (d) Show that these geodesics have infinite length in either direction.. B' .: J 3' YN. Consider the upper halfplane as a subset of the complex num bers <C. . The sum of the angles in any triangle is < 7[. ad . These results show that the Poincare = = A ti c : c . upper half plane is a model for Lobachevskian nonEuclidean geometry.368 Chapter 9 (0. so that the upper halfplane is complete.
and give N the Riemannian metric described in Problem 26.Riemanni(J}l Metric. ). and (M. more generally. where N is a Coo manifold. Does it necessarily follow that the Riemannian manifold M is complete? 43. show that E2 ".g'( . having the property that y is a minimal geodesic between any two of its points. b 1 > M with initial value y(a) = p can be extended to all of JR. Let g : M > N be a covering space. over a paracompact space]. Let p be a point in a complete noncompact Riemannian manifold M. Ell E3 . conclude that TE '" ]f*(�) Ell ]f*(TM) . (b) If � = ]f : E > M is a smooth bundle. . (a) Given an exact sequence of bundle maps as in Problem 328. 369 42. E. (b) If � = ]f : E > M is a smooth kplane bundle over M". 46. ) is a Riemannian metric on N. where the bundles are over a smooth manifold M [or. 48. If ( .. Then there is a unique Coo structure on M which makes g an immersion. Let is indeed a kplane bundle. show that the normal bundle of M C E is equivalent to �. Let M be a Riemannian manifold such that every two points of M can be joined by a unique geodesic of minimal length. )) is complete. 49.00) > M with the initial value y(O) = p. (a) Show that the normal bundles v" V2 of M n C N n +k defined for two diff erent Riemannian metrics are equivalent. 45. ) is a Riemannian metric on M. This problem presupposes knowledge of covering spaces. show that v M and N be geodesically complete Riemannian manifolds. (b) Using the notion of Whitney sum Ell introduced in Problem 352. 44. Show that the Rie mannian manifold M x N is also complete. and choose a fixed point P E M. then g'( . Suppose that every geodesic y : [a. ( . 47. Let M be a manifold with a Riemannian metric ( . Prove that there is a geodesic y : [0. )) is complete if and only if (N. Show that the Riemannian manifold M is geodesically complete. show that the normal bundle v M x Ell TM '" (TN)IM. (a) If M n C N n +k is a submanifold of N.
According to Problem 322 there is S' c M so that (TM)IS' is not orientable (the Problem deals with the case where (TM)IS' is always trivial. it is not hard to show that a bundle over S' is trivial if and only if it is orientable). (a) Let M be a nonorient able manifold. (Thus.) . (b) Use Problem 329 to conclude that there is a neighborhood of some S' C M which is not orientable.370 Chapter 9 50. Using Problem 47. show that the normal bundle v of S' C M is not orientable. in fact. any nonorientable manifold contains a "fairly small" nonorientable open submanifold. but the same conclusions will hold if each (TM) IS' is orientable.
' from S' x S' to S' is also = 1+ 1+ cos2Jf(x . which can be expressed in coordinates as one of the two maps (x.CHAPTER 1 0 LIE GROUPS his chapter uses. because in the study of these problems the groups of autom orph isms of various structures play a central role. and illuminates.Y 1+ xy. it even suffices to assume that the map (x. b E G being denoted by ab) such that the maps (a. y) is Coo.y) sin2Jf(x . y) (x. many of the results and concepts of the It will also play an important role in later Volumes. 371 . with the operation + . consequently the map (x. and these groups can be studied by the methods now at our disposal. where Z C JR denotes the subgroup of integers.y) = Coo . The functions x 1+ cos 2n: X and x 1+ sin 2n: x are Coo functions on JR/Z. y) 1+ X . The simplest example of a Lie group is JRn . ab aI from G x G to G from G to G are continuous.cos2JfX sin 2Jfy. b) a 1+ 1+ T preceding chapters.I In In In JR x JR > JR > S' = JR/Z. A Lie group is a group G which is also a manifold with a Coo structure such that (x. y) 1+ xy' is Coo. y) 1+ xy is Coo. It clearly suffices to aSSume instead that the single map (a. b) 1+ ab' is continuous. Thus the map (x. y) 1+ xy.I are Coo functions. We will mainly be interested in a very special kind of topological group. where we are concerned 'V'lith geometric problems. cos2Jfx cos2JfY + sin 2JfX sin 2JfY sin 2Jfx cos2JfY . y) 1+ xy X 1+ X . As a matter offact (Problem I). A topological group is a space G which also has a group structure (the product of a. and at each point at least one of them is a coordinate system. The circle S' is also a Lie group. One way to put a group structure on S' is to consider it as the quotient group JR/Z. It clearly suffices to assume that the map (x.
b') represent the same element of S' x S' if and only if a ' . that if H e G is a subgroup of G and also a submanifold of G.a E Z and b' . Problem 233 presents a proof that 0(11) is a closed submanifold of GL(II. Multiplication of matrices is Coo. 1R) is a homeomorphism (since O(n) is a submanifold of GL(II. the set GL(II. 1R). then G x H.. 1R)).' (1R . a manifold. is norm presen1ing. with the product Coo structure. The torus may also be described as the quotient group IR x 1R/(Z x Z). b) and (a'. f S' C 1R 2 can be considered as the group or . is easily seen to be a Lie group. Many imporlant Lie groups are matrix groups. A t = I. since the entries of A B are polynomials in the entries of A and B.' )F = det A ij / det A. is also Coo as a map from 0(11) x 0(11) to 0(11). but this is true because the inclusion of 0(11) > GL(n . 1R) with A .1)/2. considered as a 2 subset of IRn Since the function det : IRn' > IR is continuous (it is a polynomial map). where A t is the transpose of A. IR) = det. the matrix A represents a linear transformation which is an "isometry'" i. The argument in the previous paragraph shows. By Proposition 21 I .372 C7wjJler 10 If G and H are Lie groups.' which is Coo on GL(II. of dimension 11(11 . which is equivalent to the condition that. (This gives another proof that S' is a Lie group. it suffices to show that it is continuous. the torus S' x S' is a Lie group. IR). the pairs (a. then H is a Lie group. One of the most important examples of a Lie group is the orthogonal group 0(11). This condition is equivalent to the condition that the rOws [and columns] of A are orthonormal. generally. where A ij i s t h e matrix obtained from A by deleting row i and column j. Smoothness of the inverse map follows similarly fi'om Cramer's Rule: IIIIIIIII!! (A. cOJ"isting of all A E GL(n.b E Z. To show that 0(11) is a Lie group we must show that the map (x. 1R) is the group of all nonsingular real II x II matrices. and in particular. Later in the chapter we will havc another way of proving that 0(11) is a Lie group. and hence can be given the Coo structure which makes it an open submanifold of IRn' . with respect to the usual basis of jRn.e. The general linear group GL(II. and the direct product group structure. In particular. y) 1+ xy .{OJ) is open. and thus inner product presen1ing.
For any b E G. the component containing the identity I. we have K. Since e = aIa E aI K. which we have already seen is a Lie group. We define a Lie subgroup H of G to be a subset H of G which is a subgroup of G. Since e E bK bI . then K is also open. S 3 is the Lie group of quaternions of norm I. we have aI K C K.Lie Groups 373 of complex numbers of norm I. PROOF If a E K. Since this is true for all a E K. then K is an open Lie subgroup. we mention £(11). ex) for e irrational. but we will not need this fact.I b is a homeomor phism of K to itself. It even turns out. Hence K is a Lie subgroup.) It is possible for a subgroup H of G to be Lie group with respect to a COO structure that makes it merely an immersed submanifold. The group SO(2) is just S l . is a subgroup. respectively. the two components consist of all A E 0(11) with det A = +1 and det A = . then the component K con taining the identity e E G is a closed normal subgroup of G. then the immersion. since b 1+ a . Clearly SO(I1) = {A E 0(11) : det A = I } . and also a Lie group for some Coo structure which makes the inclusion map i : H > G an I l lnl l I . The group 0(11) is disconnected. •:. the group of all Euclidean . PROPOSITION. so K is normal. It is know that these are the only spheres which admit a Lie group structure. For example. then aI K is connected. that a subgroup which is an immersed submanifold is always a Lie subgroup. after some work (Problem 18). As we have seen.1 K C K. As a final example of a Lie group. image of L in S l x S l = lR x lR/(Z x Z) is a dense subgroup. we have bKb1 C K. If G is a Lie group. since G is locally connected. This is not accidental.I . Similarly. it follows similarly that bK b I is connected. a subgroup which is an (imbedded) submanif old is always a Lie subgl"Oup. If G is a topological group. if L C lR x lR is a subgroup consisting of all (x. K is dosed since components are always closed. If G is a Lie group. I\1oreover. which proves that K is a subgroup. so K is a submanifold and a subgroup of G.
G. given Xe E Geo there is a unique left invariant vector field X On G which has the value Xe at e . E G. Clearly the component of e E E(I1) is the subgroup of all A T with A E SO(I1). In fact. so A TaA. A Ta(BTb).374 Chapter 10 motions. the maps La* : Gb Ra * : Gb 7 Gab 7 Gba for all are isomorphisms.1 (x) = A(A. which shows that E(I1) is a Lie group. For any Lie group G. Vie can give E(I1) the C"" structure which makes it diff eomorphic to 0(11) x IR" . . T where A C 0(11).1 X + a) = x + A(a) = TA(a) (X) . i. since translations and orthogonal transformations do not generally commute. isometries of IR". respectively.1 = A TaTbB.1 = A B. La : G > G and Ra : G > G. b It is easy to see that this is t�ue if we merely hav(' for all a E II E G. and T is a translation: T(X) = Ta (X ) = X + a .l . by La(b) = ab Ra(b) = ba .X = X Recall this means that for all a . if a E G we define the left and right translations.e.l = A Ta_bB.1 TB(a_b). A vector field X on G is called left invariant if La. Now E(I1) is not the direct product 0(11) x IRn as a group.l and Ra. A little argument shows (Problem 5) that every element of E(I1) can be written uniquely as A .. COJlsequently. Consequently. with inverses La . Notice that La and Ra are both diff eomorphisms. Consequently.
. ..I 375 PROOF. XXi (a) = Then " Xe (xi = j= 1 = . PROOF. b E V implies ab ' E U. • : 3.Xe(xi) 0 = Xe (xi La) . A Lie group G always has a trivial tangent bundle (and is consequently orientable). Let (x. so we can define an equivalence G x lR" J (t J=I ciX (a) i ) = " (a. This implies that X is I e where Xe = j =l L cj D X a I e which shows that XXi is Coo . . La. .X around a (Prob lem 51). x(e))./f. Then XI . ' . . . c ) . since the diffeomorphism La then takes X to the C"" vector field La. . " " X"e for Ge. X" be the left invari ant vector fields v. Choose a neighbor hood V of e so that a.Lie Grau/}. Since the map (a. It suffices to prove that X is Coo in a neighborhood of e.rith these values at e. Coo . COROLLARY. Xn are clearly eveI)"Alhere J : TG > linearly independent. Every left invariant vector field X on a Lie group G is COO. . . Let X" . Then for a E V we have XXi (a) = 2. c ' . b) 1+ a b i s C oo o n V x V we can write for some Coo function Ji on x(V) x x(V). . •:. U) be a coordinate system around e. Choose a basis X' e . PROPOSITION. Cj a(xi 0 La) � ax} 0 La) j= 1 L c j D"+j Ji (x(a).
Y. etc. and X. Ell V2 (= V. is called the Lie algebra of G. or 0(11) In general. X]. The Lie algebra £(S') of s' is I dimensional. . Z] + [[Y. X] + [[Z . ] operation is assumed alternating. Y](e). ] operation. YJ = [X . Z E V. are Lie algebras with bracket operations [ .. Xl Consequently. 1R) is an open submanifold of IRn ' . etc. ] on Ge by [X. The structure of gf (11. to denote the left invariant vector fields with X(e) = X. Proposition 63 shows that [ X . Clearly £(IRn) is abelian. or The Lie algebra of IRn is isomorphic as a vector space to IRn. Consequently. YJ '= 0 f all X. are left invariant and [a/ax . a Lie algebra is a finite dimensional vector space V. Y. Consequently. This notation requires o some conventions f particular groups. Henceforth we will use X. the Lie algebra £(S' x . The vector space Ge . YJ = 0 'jacobi identity" f all X. x V as 2 a set) by It is easy to check that this makes V into a Lie algebra. 1R) is more complicated. it is also skewsymmetric. the tangent space of GL(n . and will be denoted by £(G). If If. with a bilinear operation [ . We can then define an operation [ .. and consequently must be abelian. and that £(G x H) is isomorphic to £(G) x £(H) with this bracket operation.) We will also use the more customary notation 9 (a German Fraktur g) f r £(G).376 Chapter 10 A left invariant vector field X is just one that is Larelated to itself for all a . X] = o [[X. together with this [ . 2. then or we can define an operation [ . Y. since the vector fields a/ax . Since GL(n. (Sometimes the Lie algebra of G is defined instead to be the set of left invariant vector fields. 1R) or f the Lie algebra of 0(11). X S') is also abelian. 1R) f the Lie algebra of GL(Il. Y. etc. YJ = [Y. YJ i s left invariant i f X and Y are. to denote elements of Ge. Z] . ] satisfying [X. or Since the [ . ] on the direct sum V = V. we call a Lie algebra abelian or commutative if [X. Y(e) = Y. we write or gl (ll. . IR) at the identity I can be . [X. ] i f i = 1 . . a/axj ] = O. YJ.
0 1 (x kl 0 LA : GL(n .=) n 0'=1 n •. So if N is another 11 x a __ . NlI = � (MN . M/ (xkl ) = M/ (xkl LA) .kl ) matrix.. a Xl} a = L Nkj Mjl = (NM lkl· j=l From this we see that " . 0'=1 n with (constant) partial derivatives � (xkl ax" 0 LA) = A ki {0 j =I J # I. For every A E GL(n.I.[M.lith jRn 2 . N/ (Mx. Now the function xkl Mxkl (A) = L Mij :ij a i. then an 11 x 11 (possibly singular) matrix M = (Mij ) can be identified with M/ Let M be the left invariant vector field on GL(n.2 . If \ve use the standard cOOldinates xij on jRn. 1R) > GL(n. 1R) i s the linear function LA) (B) 0 = xkl (AB) = L A k . 1R) as f ollows. 1R) corresponding to M.j I = MA (xkl ) = LA.kl _ Mjl Mx {0 ax ij 11 .B..l . ax l / a k. 1R).IA k Thus. Wc compute the function M xk l on GL(n.Lie Groups 377 identified ".j = L Nij . k=i k # i.(Mx kl ) . we have i. So = L Mi/ A ki = L M.NM lkl rJ .
the bracket operation is just [M. so 0(111I must consist exactly of skewsymmetric matrices. then k�i differentiating give. Notice that i n any ring. :2 Thus 0(11)1 C jRn can contain only matrices M= ( M which are skewsymmetric. then [ . IR) we can consider 0 (11) / as a subspace ' of GL(I1. If A : (8.NM. j with i < j. . we could use the following line of reasoning.b] = ab . This subspace may be determined as follows. N] = MN . IR) with IRn' . Mi n M1 2 0 0 M1 3 .sin ! sin ! cos t J (rotation in the (i. ] satisfies the Jacobi identi�: Since 0(11) is a submanifold of GL(I1. If we did not know the dimension of 0(11).378 thus. if we identify Chapter 10 gl(I1.Aj/(O) .ba . we can define a cun'e A : IR > 0(11) by A (') � ( j ' cos t . j)plane) j . 0 M1 2 M1 3 .1 )/2. Mi n 0 ) This subspace has dimension 11(11 . which shows that L A ik (t)Ajk (t) = �ih A i/ (O) = . . For each i. and thus identify 0(11) with a certain subspace of IRn . which is exactly the dimension of 0(11). and we denote (A(t))ij by Aij (t). IR)/.8) > 0(11) is a curve with A(O) = I. if we define [a.
then Lb. = so Lb. moreover. The proof of uniqueness is left to the reader. It follows immediately that Lb permutes the various maximal integral manifolds of /j. ([X. Let PROOF. Y](e))..(/j. For each E H C G. among themselves. Using Theorem 67. j) and (j.Lie Graul" 379 with sin t and . we have left translations 7 X X a La : H and So > H. I 's on the diagonal except at (i. other words.sin t at (i. so Lb .a be the subspace of Ga consisting of all for X E I). i. The fact that I) is a subalgebra of � implies that /j. In fact. 1 operation. He with this induced . (a. invariant. Now this map is clearly Coo as a map into G. = La : G > G In. and are irelated. leaves the distribution /j. it follows that it is Coo as a map into H. if Y E and are irelated. and O's elsewhere. Yl [X. j). we can identify He with a subspace of Ge.a) = /j. X(a) X = i. Y](e) Thus. X La o f i 0 La. Hence O(n)/ must consist exactly of skewsymmetric matrices. We do not need any new calculations to determine the bracket operation in o(n).. •: a X(a) e. Consequently. = = La. Any X E He can be extended to a left invariant vector field on H and a left invariant vector field on G. Then there is a unique connected Lie subgroup H of G whose Lie algebra is I). To prove that it is a Lie subgroup we just need to show that 1+ is Coo . He is a subspace of � which is closed under the . This implies that H is a subgroup of G. takes H to the maximal integral manifold containing Lb. i) and (j. THEOREM. consider a Lie subgroup H of any Lie group G.ba. Since i* : He Ge is an isomorphism into. and O(n) must have dimension n(n . that is. [ G be a Lie group. He C Ge = � is a subalgebra of �. containing If b E G. if E H. La. then [X. which means that [X. (i. b) ab. (H) = H. and let i : H > G be the inclusion. X) = X(a). In particular. f] X [ i. let /j. Then the set of all A'(O) span the skewsymmetric ma trices.1 )/2. Let H be the maximal integral manifold of /j.1 . is an integrable distribution. 1 operation is juSt I) = £(H). and I) a subalgebra of �. For E G. i). This correspondence between Lie subgroups of G and subalgebras of � turns out to work in the other direction also. b (b) e. 4. He. then clearly Lb.
. It then follows from Theorem 4 that ever . For example. Having considered subgroups of Lie groups (and subalgebras of their Lie algebras). we next consider.. X = %(4)(a)). then 4>. 4>.aLa.380 Chapter 10 There is a very difficult theorem of Ado which states that every Lie algebra is isomorphic to a subalgebra of G L( N .. more generally. 1R) for some N. . simply by 4>. then . For any a E G we clearly have 4> 0 La = L¢(aj 0 4>.x is the left invariant vector field on H with 4>*a%(a) = 4>.. We continue the study of Lie groups along the same route used in the study of groups.4>. We will soon see to what extent the Lie algebra of G determines G. Usually. Later ) on we will be able to obtain a "local" version of this result. because IR is a vector space of uncountable dimension over Q. we will denote 4>. and % value 4>*eX at e. the map algebra homomorphism. 4>. Thus X and % are 4>related. Consequently. : Ge > He .. X. But if 4> is Coo. [X. : � > Ij. : � > Ij is a Lie 4>. There are an enormous number of homomorphisms 4> : IR > 1R. If 4> : G > H is a Coo homomorphism..P. and every linear transformation is a group homomorphism. Lie algebra is isomorphic to Ihe lie algebra ifsome Lie group. (aX + bY) = a4>. so if X E Ge.. It is not hard to see that even a continuous 4> must be of this form (one first shows that 4> is of this form on the evaluating at s 4>' (1) = 4>' (0).X = L¢(aj.. X + b4>. suppose that G = H = JR. homomorphisms between Lie groups. that is.. Y]... then the condition 4>(s + I) = 4>(s) + 4>(1) implies that d4>(1 + s) ds = 0 gives d4>(s) . ds which means that 4>(1) = cl for some c (= 4>'(0)). Yj = [4>. Y 4>.
Let G and H be Lie groups. 4>. X E g we have 4>. Notice that the only continuous homomorphism 4> : S' > IR is the 0 map (since {OJ is the only compact subgroup of 1R). THEOREM. we do have a local result. .b. <I> (X)) = X . then 4> = Vr. For X E g we have SO w* : K(e. If H. We can identify £(IR) with 1R .ab E U. then w : K > G is a Coo homomorphism.e} Ge is an isomorphism. 7 w. . I K. and G is connected.Lie Groups 381 rational numbers). If H2 : G x H > H is projection on the second factor. A neighborhood of the identity e E S' can be identified with a neighborhood of 0 E 1R. P ROOF.e) E K such that w takes V diff eomorphically onto an open neighborhood U of e E G. Vr: G > H with 4>. By Theorem 4. we Can define Moreover. but H = S' = IR/Z.. there is a unique connected Lie subgroup K of G x H whose Lie algebra is f. 4>. Consequently. : IR > IR is just multiplication by c. = <1>. Now suppose that G = 1R. Consequently. . for X E g. The continuous homomorphisms 4> : IR > S' are clearly of the form once again. However. giving rise to an identification of £(S') with 1R. X = <I>(X). and <1> : g > l) a Lie algebra homomorphism. and w = H. 5.(X. Let I (German Fraktur k) be the subset f C g x l) of all (X. Clearly the map 4>. Since <I> is a homomorphism. : IR > IR is multiplication by c. a Lie algebra ho momorphism g > I) may not come from any Coo homomorphism 4> : G > H. Then there is a neighborhood U of e E G and a Coo map 4> : U > H such that 4>(ab) = 4>(a)4>(b) and such that for every when a. : G x H > G is projection on the first factor. if there are two Coo homomorphisms Vr. <I>(X)). I is a subalgebra of g x l) = £(G x H). there is an open neigh borhood V of (e.
G is locally isomorphic to IR". Given an isomorphism <1> : A > I). •:Remark: For those who know about simplyconnected spaces it is fairly easy (Problem 8) to conclude that two simplyconnected Lie groups with isomorphic Lie algebras are actually isomorphic. then w. For every X E Ge. and that all connected Lie groups with a given Lie algebra are covered by the same simplyconnected Lie group. By Corollary 6. PROOF. COROLLARY.X•  .X Given 4>. define the oneone map = ][2. (X. <I> (X)) = X. <I> (X)). PROOF. A connected Lie group G with an abelian Lie algebra is itself abelian . which implies that 1jr (a) = 4>(a) for all a E G . 7. COROLLARY. •:. <I> (X)) = <I>(X). It follows that G is abelian. let 4> be the map given by Theo rem 5.382 Chapter 10 TIle first condition on 4> is obvious. so 4>. if X E g. since (Problem 4) mH' . Since 4>*e = <P is an isomorphism. so ab = ba for a. If two Lie groups G and H have isomorphic Lie algebras. 1jr (a)). Thus G' = K.�O . x so £(G') = f. 4> is a diffeomorphism in a neighbor hood of e E G . •:. b in a neighborhood of e. (X. = (X. 4> : IR > G such that 8. then they are locally isomorphic. e: G > G x H by e(a) = (a. As for the second. 6. there is a unique d4> dl Coo homomorphism 1 . neighborhood of e generates G . COROLLARY. 1jr: G > H. The image G' of e is a Lie subgroup of G x H and for X E A we clearly have e.
III. If homomorphism.I ) and define 4>(1) = { 4>(e/2) · · . which proves uniqueness.( ) du u=o j' = L¢(r»XU) = %(4)(I))U). We know that integral cun'es of % exist locally. I = O.0 dl ' /(4)(1)4>('')) . 4>(s + 1) = 4> (s)4> (I ) and Clearly <I> is a Lie algebra homomorphism. then [4>(e/2) appears k times] k '" 0 4>(e/2) · 4> (e/2) · 4>(r) [4>( e/2) appears k times] k /: G > IR is Coo. Conversely) if 4> : IR > G is an integral cun'e of %. < o. ls + 1 1 (:!. SECON ( D DIRECT) PROOF. .· 4>(e/2) · 4>(r) . Define <1> : IR > £(G) by 383 <I>(a) = aX. .Lie Groups F IRST PROOF. and < < e 4>: IR > G is a Coo d4> ( f) = lim /(4)(1 + ")) . then 1 1 is an integral curve of % which passes through 4>(s) at time clearly true for 1+ 1+ 4>(s) . Irl e/2 r �O ddl4> I Isl. they can be extended to all of IR using the method of the first proof. I 1 Thus 4> must be an integral cun'e of X.:. .X. By 111eorem 5. Uniqueness also follows from Theorem 5./(4)(1)) lim = h'l>O h = � / 0 L¢(l) 0 4> du u=o d4> = L ¢(l)' .4>* dl r�O To extend 4> to IR we write every I with II I '" e uniquely as 1 = k(e/2) + r k an integer. on some neighbor hood (e. e) of 0 E IR there is a map 4> : (e. 6") > G with . 4>(1) 4>(s + I)./(4)(1)) " 1. The same is so 4> is a homomorphism.
Ii 2! 3! hence I A lk I A N+K AN (n I A I) N (nI A I )N+K +". Consequently: IA + BI IA B I :s: :s: I A I + IB I n I A I ' IBI: I . then clearly A A2 A3 + + +. More interesting things happen when we take G to be IR .+ +. We have already examined the I parameter subgroups vi Iii . All Coo homomorphisms 4> : IR > must satisfy the analogous diff erential equation 4>' (1 ) = 4>'(0) . with 4> (S + I) = 4>(s)4>(I). 4>(1 )."+ > 0 as N > 00. IR) where · now denotes matrix multiplication. The solutions of these equations can be written formally in the same way 4> (1 ) = exp(I4> ' (O)). must satisfy 4>' (1 ) = 4> ' (0)4>(1 ) 4> (0) The solutions of this equation are 4>(1 ) = = I. IR). where exponentiation of matrices is defined by exp(A) = I+ This follows from the facts in Problem 56.. If A = (a'j ) and I A I = max laij I. . Then all Coo homomorphisms 4> : IR > IR . with multiplication as the group operation. GL(n. Notice that IR .{OJ.' I A lk :s: nk l A l k .nN! (N + K ) ! :s: nk. some of which will be briefly reca pitulated here.. (N + K ) ! :s: y. We thus see that there is a unique I parameter subgroup 4> of G with given tangent vertnr d4>/dl(O) E Ge. e¢'(O)r.{OJ.{OJ is just GL( I .384 Chapter 1 0 A homomorphism 4> : IR > G i s called a Iparameter sub group o f G . 4> (0) = I.
I] exp(I</>'(O)) +.+O lim Ii + h2 </>'(0) 2 h 2! exp(I</>'(O)) = </>'(0)</>(1). For any Lie group G. if </>(1) is defined by (* *). Given X E �. and convergence is absolute and uniform in any bounded set..I 385 so the series for exp(A) converges (the J) th entry of the partial sums converge). if A B = BA. then </>' (1) = = (i. we now define the "exponential map" exp : � > G as follows.exp(I</>'(O)) h [exp(h</> (O)) 1 = lim h'l>O h</>'(O) = . The map exp : Ge > G is Coo (note that Ge '" lRn has a natural Coo structure). PROPOSITION. X) (exp 12 X) exp(IX) = (exp IX). and 0 is a regular point. h.. then exp(A + B) Hence. let </> : lR > G be the unique with d</>/dl(O) = X. then Coo homomorphism We clearly have Ge exp 0 1fr. Then exp(X) = </> ( 1 ) . = 1fr 0 expo exp G�H l � He l exp . (exp A ) (exp B)..Lie Grau/}.I . 9. so </> does satisfy (*) . h_O lim exp(I</>'(O) + h</>'(O)) . If 1fr : G > H is any Coo homomorphism. so that exp takes a neighborhood of 0 E Ge diff eomorphically onto a neighborhood of e E G. exp(11 + 12 ) X = (exp I . Moreover (see Problem 56)..
. •: 1 1 1 0. dl r�O Consequently.. Since it follows that exp is Coo.a) = 0 Ell X (a).:. Every oneone Coo homomorphism 4>: G > H is an immersion (so 4>(G) is a Lie subgroup of H). (X) = But then e = exp 4>.386 Chapter 1 0 PROOF. COROLLARY.a) o f the Coo manifold G e x G a t the point (X.X. o. (tX) = 4>(exp(IX)). then also 4>. So exp.. 0 Ell X). We define a vector field Y on Ge x G by Y(X. a) can be identified with Ge Ell Ga. . then the curve C(I) = IV in Ge has tangent vector v at o. Given Vr : G > H. Then Y has a flow a: IR x (Ge x G) > Ge x G. dl r �O Then Vr 0 4> : IR > H is a homomorphism with d(Vr 0 4» = Vr.o is the identity. exp(Vr. and hence oneone. and X E Ge. let 4> : IR > G be a homomorphism with d4> = X. PROOF. which we know is exp X = projection on G of a ( l . The tangent space (Ge x G)(X. I 1=0 = d dt I 1=0 exp(lv) So exp. Coo. If we identify a vector V E (Ge) o with Ge.X) = Vr 0 4>(1) = Vr (exp X) . contradicting the fact that 4> is oneone.o(v) = = d exp(c(I)) dt v.p( X (p)) = 0 for some nOllzero X E �. Therefore exp is a diffeomorphism in a neighborhood of O. If 4>.
)m = [exp(X/2 n )]m = exp(m/2n . (exp In Yn ) . so X/4 = Y/2. By continuity. Let 4>(e) = exp X. X) . X" for G. for all s E [ . X E exp( ! U). PROOF Since X/2. The map 1 1+ 4>(exp IXi) is a continuous homomorphism of IR to H. Moreover. Then on V.. PROOF. COROLLARY. exp X /4 E exp( t U ) . ( � IJ . e) = 4>(e/2.) · · · (exp ln Xn ) is Coo and clearly (*) and (*) shows that 4> 0 Vr is Coo. (exp In X. . then S o a has a square root i n exp( b = exp(Y/2) for Y E U.I . Choose a basis X" . Every continuous homomorphism 4>: JR > G is Coo. . 4> (se) = expsX 1 2. .Lie Groups 387 I I . then 2 a = exp(X/2) = [exp(X/4)] . •:. In ) = (exp l. Now the map Vr : JRn > G given by Vr (l" . ) ) = (exp I. Vr* a i = X"� so Vr is a diffeomorphism of a neighborhood U of 0 E JR" onto a neighbor hood V of e E G. .. so there is Yj E He such that 4> (exp IX. •:. . Let U be a starshaped open neighborhood of 0 E G. Yt l · · . so 2 2 exp(X/2) = a = b = [exp(Y/2)] = exp Y. For any a E exp( ! U). 4> ( expI. it follows from the above that 4>(e/2) = exp(X /2). 4> (m/2n . Now choose e > 0 so that 4>(1) E exp( ! U) for I I I � e. . I] . So 4> is Coo at e. 4> = (4) o Vr ) o r ' . This shows that every a E exp( t U) has a unique square root in the set exp( t U). if a = exp(X/2) for X E U. Thus. i f a = b2 for b E exp( ! U). . on which exp is oneone. COROLLARY.) = exp I Y.. X. Y E U it follows that X/2 = Y. By induction we have 4> (e/2n ) = exp(X/2n ) . Every continuous homomorphism 4> : G > H is Hence Coo. and thus everywhere. Xtl · · . Since 2 [4>( e/2)] = 4>(e) = [exp X/2f . t U).
1)/2 dimensional submanifold of GL(n . At = (cxp M)(exp N). In general.1 = exp(M). y] + O(l3)} (3) exp lX exp IY exp(IX) = exp{IY + 1 2 [X. 14. hencc Mt = M. and in its proof. bounded for small I. O(n) is itself a submanifold of GL(n . we have = I. extent (0 which this equation fails to hold. Then I = A . Since O(n) is a subgroup. IR) exp(Mt) = ( exp M)t . If G and G' are Lie groups which are isomorphic as topo logical groups. IR). (2) (I) exp lX exp l Y = exr l (X + Y) + '2 [X. Just as in GL(n. If G is a Lie group and 1 exp(IX) exp(IY) exp lX exp l Y = exp { 1 2 [X. we will denote it by 0(13). C on sely. So exp Mt = At = exp(M). then they are isomorphic as Lie groups. then 12 . IR). the x ver (exp M)(exp M)t = I . In the following Theorem. the equation exp(X + Y) = exp X exp Y holds whenever [X. It follows that a neighborhood of I in O(n) is an n(n . IR). [X. any A E O(n) sufficiently close t o I can be written A = exp M for some M. Y] measures. so e p N = (exp M) . i.388 Chapter 10 1 3 .e.. to indicate that a function c : IR > Ge has the property that C(I)/13 i. THEOREM. since exp(M + N) (exp M)(exp M) So if M is skewsymmetric. y] = 0 (Problem 1 3). It is easy to see that (= �l(n . For sufficiently small M and N this implies that N = M. COROLLARY. up to first order. Y E Ge. there is a diffeo morphism between them which is also a group isomorphism. M = _Mt. Moreover. Let At = exp N. •:. then = (exp M) ( exp N) when MN = NM. y] + 0(t 3 )}. Thus 0(13) will denote differenl functions at different limes. exp M E O(n). Y] + 0(1 3 ) \ j X. 2 The properties of the particular exponential map exp : IRn may !lOW be used to show that O(n) is a Lie group. PROOF Apply Corollary 1 1 to the continuous isomorphism and its inverse . that is. IR ») > GL(n.
Then we have '�. I J(exp sX exp l Y exp uy) Applying (iii) to YJ instead of J gives (iv) </>" (1) = [Y ( YJ)](exp sX exp I Y). du u=o _ I For fixed s. Now Taylor's Theorem says that Suppose that J(e) = (v) J(expsX exp l Y ) = J(exp sX) + I ( YJ)(expsX) _ _ o. (ii) d XJ(a) = Xa (J) = La. (i) We have Similarly.Lie Graul)j 389 PROOF. du u=o d YJ(a) = J(a · exp uY). d d (iii) </> (I) = . 2 . d d. Similarly. exp uX).o 12 _ _ + "2 [Y ( YJ)](exp sX) + 0(1 3 ) .X(J) = X(J 0 La) = J(a . u u=o by (ii).J(exp sX exp l Y ) = dt d = ( YJ) (exp sX exp l Y ) . for any F. </>(1) = </>(0) + </>' (0)1 + </> ) 1 2 + 0(1 3 ). let Then l </>(1) = J(exp sX exp l y). F(expsX) = (XF)(expsX) d2 F(exp sX) = [X(XF)] (exp sX) ds 2 F(expsX) = F(e) + s( XF)(e) + � [X (XF)](e) + 0(S3 ) .
In particular.+ X Y + 2 · 2 2 which gives thus proving (I). Equation (2) follows immediately from (I). = Z for some Z" Z 2 E Ge. . and F = Y (YJ) gives (vi) J<exp sX exp t Y) + s2 _ = s ( XJ) (e) + t ( YJ)(e) t2 _ _ _ 2 [X (XJ)](e) + 2 [Y (YJ)] (e) + stX (YJ)(e) _ _ 2 2 + 0(S 3 ) + 0(1 3 ) + 0 (s t ) + 0(st ) . If J(e) = 0. (Zt /)](e) + 0(1 3 ) .f's t o be coordinate functions.  X + Y = 2. comparison o f (vii) and (viii) gives 2. Z{I) = J(A(t)) + Since w e can take the . (vii) J(exp tX exp t Y) = t [(X + YJ Jl(e) No\v for small t we can write +t2 [( t x = + XY + YY 2 ) J (e) + 0(1 3 ). xx YY . J)(e) + t 2(22 1)(e) t2 _ _ + 2 [Z. so by (vi) we have (viii) /(exp Z (t)) = J (exp(t Z . 2. Applying Taylor's formula to t Z . then clearly J(A(t) + 0(1 3 )) 3 O{l ).390 Chapter 10 Substituting in (v) for F = J. + t 2 Z2 )) + 0(1 3 ) = t ( 2. = X + Y.+ Z2 = . ] exp tX exp t Y for some gives exp Z (I ) Coo function Z with values in Ge. F = YJ.  Z. + t 2 Z2 + 0(1 3 ).
X)J](e) + 1 + . > 0 with each I. or Proqf We can assume I. Let 1) c Ge be the set of all X E Ge such that exp lX E H f all I . Assertion J . there is a Coo structure on H. + I S.).X.J)(e) = 1 (. (I) = largest integer � Then �I < I. + 0(1 )). > X and let I.) = (expl. THEOREI'vl. + I S. E H f all i. > 0.51 J)](e) + 0(1 ). J)(e) + 1 ' ( .)]  (e) ' 3 exp lX exp I Y exp(IX) = exp(lS. since exp(I. Ifwe write 0 ' (1 ) . We attempt to reconstruct the Lie algebra of H as f ollows. .Lie Groups 391 To prove (3). The work involved in proving Theorem 14 is justified by its role in the f ol lowing beautiful theorem. Then X E lj. . again choose J with J(e) = O.1 E H. then H is a Lie subgroup of G. I Notice that f ormula (2) is a special case of Theorem 516 (compare also with Problems 516 and 518).:.(t) ::: ::: I. If G is a Lie group and H e G is a closed subset which is also a subgroup (algebraically) . # O .)) + 0(1 ) 5. Suppose expl. k. or PROOF. with llle relative topologJ') that makes it a Lie subgroup of G. ' 3 J(exp lX exp l Y exp(IX)) = J ( exp(lS.. . . Let X. 3 [(XX 2  yy + ""2 + 2 + X Y  XX  xx . Then similar calculations give (ix) J(exp lX exp l Y exp(IX)) = I [(X + Y .YX . let k..51 ( . E Ge with X.X. 1 5. For I > 0. ' 1.X. 3 + [ . More precisely. then we also have (x) Comparing (ix) and (x) gives the desired result.5.
but we will not even use this fact. .392 so Now Olapter 1 0 I. > 0 and let X. t (l) E H. so X E I). QE. Clearly X E I) implie. ) = [ exp (t. .) k.Y 11 11 1 I )n = exp {I(X + Y) + [X. for fixed I.D. Then exp(1) n U) is a submanif of G. It dearly suffices to show thaI old if U is small enough. Then Assertion J implies that X + Y E 1). 12 211 f taking limits as n > 00 gives exp 1 (X + Y) E H. sX E I) for all s E IR. . . Xk . > G defined by t/>(X + X') = exp X exp X' Xl . .] Now let U be an open neighborhood of 0 E Ge on which exp is a diffeomor phism. is a dilleomorphism in some neighborhood of O. then H n exp(U) = exp(I) n U).X. Proq{ Choose a basis Then t/> is given by of G. Xk a basis for I). so that I) is a subalgebra. = I) Ell Ij'.(t)I. . so that G. Alternatively.X exp .k. exp(k. ). . with . The map t/> : G. Xn X E 1). using (2) of Theorem 1 4 we see that [X. If X. We now claim that I) c Ge is a vector subspace. . . Choose positive I.X. Y] E 1). (t) > I. Y E I). Assertion 2. .X. since H is closed and exp is continuous. Thus exp IX E H. we can write. We clearly also have exp lX E H for 1 < 0. > IX. Y] + O( l /n2) l . [Similarly. Choose a subspace Il' C Ge complementary to I). we can write by (I) of Theorem 1 4 exp lX exp l Y = exp{I(X + y) + IZ(t)} where Z(t) > 0 as 1 > O. . = X + Y + Z(t. ( exp . X' E I)' X" .(t)I.
Choose integers JI j 'V'lith Assertion 3. ax 1jr* (�' I ) 0 = X. . since QE.D. a contradiction. exp X E H we obtain exp X' E H. . so 0 = X'. and a E exp(1) n U) . 7 Choosing a subsequence jf necessary.X.\ 393 Since the map . To prove the reverse inclusion. QE. •:. > 0. we can assume Xi' I/n. )(11. let a E H n exp( U). If the assertion were false.Lie Grau!}.') E H.D. Prorif. Choose an inner product on �' and let K C I)' be the compact set of all X' E IJ' with I ::: IX'I ::: 2.L7= a. . We can now complete the proof of the theorem. X' E W'. and 4> of Assertion 2 is a diffeomor phism on W x W'.)0 X' E K. 1 it suffices to show thaI 1+ (a) .X. Since exp(I /I1. is a diffeomorphism in a neighborhood of 0 E ]R" . it follows fi'om Assertion 1 that X' E I). There is a neighborhood V' of 0 in I)' such that exp X' ¢ H if 0 # X' E V'. . This is clear. there would be X. . Clearly exp(1) 11' n U) CH n exp(U). Choose a neighborhood = W x W' of Ge on which exp is a diffeomorphism. . with U W a neighborhood of 0 E I) W' a neighborhood of 0 E such that W' is contained in V' of Assertion 3. an) is a diffeomorphism of Ge onto jRn .' E 1)' with X/ 0 and exp X/ E H. . Since a. Then a = exp X exp X' X E W.
If wi (e). H ence dw(e)(X. 9]) = w([xj ']). iJ. of Lie groups are better expressed in terms of f orms.··· <iJ. . then f a E G we have or CO:: . (J)n are left invariant I f orms such that cv I (e). /\ o/k = L Aj o/ J f certain COI/SWI/Is aj. It f oHows that any left invariant f orm is If (J) is left invariant. " " Xn E Ge.394 Olapte. A f orm w is called left invariant if La'w = w f all a E G. . .w([X. Y) = X(w(l7)) . La* dw = d(La*w) dw. . Clearly. but man� propertie. . . . = so dw i. . . then every left invariant kform i� i]<. .o} I /\ . The interplay between left im"ariant and right invariant vector fields is the subject of Problem I I..Y(w(X)) . H ere we consider the case of f orms. Y]). . . wn(e) is the dual basis to or then any Coo vector field X can be written Xl .. . The f ormula on page 215 implies that for a ieI'. the bracket being the operation in A. cvn (e) span G/'. fj Xj j=1 Then f or Coo functions /i so 0/ is Coo. 1 0 U p t o Ilo\V: w e have concentrated on the left invariant vector fields. a left invariant kform w is determined by its value wee) E Qk(Ge) ' Hence� i f cv l � . L ai1 . invariant Iform (J) and left invariant vector fields X and Y we have dw(X. This means that or web) = La * [w(ab)]. . also left invariant. X = 'L. . Y) = w(e)([X.
then 1jr * we = (3) (I) A f orm W is left invariant if and only if (4) If G is abelian. so The f orm But So dw = o. then all left invariant If orms w are also right invariant. this tangent vector is just x. (_J)k we. Y E A. P ROOF.! 395 16. 1jr* (we) = ( _ J )k We. Let 1jr : G > G be 1jr(a) = a. It follows fi·om (*) that [X.(X) = X for X E Ge.1 = exp( I X). . (4) If G is abelian. (3) If W is a left and right invariant kform. then dw = o. then 1jr*w is right invariant. then A is abelian (converse of Corollary 7). we have dw is also left and right invariant. then so o/*w is right invariant. Yj = 0 for all X. If W is left and right invariant. The converse is similar. So dw = 0 for all left invariant Iforms. PROPOSITION. (2) It clearly suffices to prove this for k = J. Since o/*w and (J) are both left invariant. (I) Clearly so If (J) is left invariant. (2) If We E Qk(Gel. Now X is the tangent vector at t = 0 of the curve 1 1+ explX.Lie Group. So it is enough to show that 1jr.. So 1jr"X is the tangent vector at 1 = 0 of 1 1+ (exptX).I .
. . X] + 0(1 3)/1 2 . . First choose XI .L Ci� wi /\ wi i. . Y] + 0(1 3)/1 2 = HY. There are constants Ci� n Since .j � . wn(e).:. . By Theorem 1 4. . In fact. . ] and the Jacobi identity we obtain (I) Ci) = C)� L( C! C/'k + cJk cL + Cl'iC/') ) = . = dw is left invariant for any left invariant w. " " Xn of � ). it foHows that for a basis w 1 . . (J)1I in a neighborhood of 0 E jRn such that .396 Ozapter /0 X. of of invariant Iforms we can express each such that [Xi . . . Letting 1 > 0. � j k=l n The numbers Ci) are called the constants of structure of G (with respect to the basis XI .L Ci� Wi /\ w) i<j n (2) From (*) on page 394 we obtain = o. we obtain [X. It turns out that (2) is exactly what we obtain fi'om the relation d 2 w k = O. then for have Hence HX. Con dition (2) is thus an integrability condition. ' " X E Ge dual to Wi (e). we can prove (Problem 30) that if Ci� are constants satisf ying (I) and (2). X)] = " Cik Xk . From skewsymmetry of [ . . Y E Ge we Allernale proqf qf (4). dwk in terms of the 0/ /\ wi . X)] = L Ci�Xk : k=1 dearly we also have . if G is abelian. [Xi. . then we can find evelywherc linearly independent I forms cv 1 . Y] [Y. ll=l dwk = . Xl .
iii ) . . .iJ n . far 1 7 . . so on J(p.[a'. . i<j By Proposition 714. . _ " > .i< Ci� ([iJi /\ iJj ] . . .. ei /\ ej de L i<j Then f every p or M there is a neighborhood U and a diffeomorphism J: U G such that ei J*wi PROOF. • . This means that r contains the graph of a diff eomorphism J fi·om a neighborhood U of p to a n eighborhood of = = Then "2 : > a E a). Hence . [i} M "2 : > a. . THEOREM.L Ci� [iJi /\ (iJj _ w j )+ (iJ i _ wi ) /\ wj ]. w'. wn are linearly independent everyvvhere. iiJn are each linearly inde pendent. Let '" : M x G M and M x G G be the projections. b) 1+ ab in a neighborhood of 0 which is a group as as it can be and which has the (J/ as left invariant I forms. a} . . . . . e n be everywhere linearly independent I forms on M satisf ying k . the sets 8 1 . = > cI(iJk . Now iJ'. wIn and constants of structure Ci� . for which we supply an independent proof. . ." Ck. Let ..wk Choose G and let r be the folium through (p. . : r M and r G are each diffeomorphisms in some neighborhood of (p. From this latter fact and (a suitable local version of) Theorem 5 we could immediately deduce the following Theorem. M x G is foliated by IIdimensional manifolds whose tangent spaces at each point are annihilated by all iJk . Let G be a Lie group with a basis of left invariant I forms w'. = E > IJ .a} where {jk iiJk = 0. . /\ wj J) L j . 397 Moreover. Let Mn be a differentiable manifold and let e . . a).. en and wi. which is the set of vectors in (M x G)(p. .Lie Group. the existence of such (J/ implies (Problem 29) that we can define a multiplication (a.
THEOREM. (I)y.j* H2* a/ (H. )* W (Yz (l )) ] . (I)) (�IJ IR and the fwo maps YI ' Y2 : IR > G satisji Y2· For every left invariant Iform w == (d. let G be a Lie group.:. (�I ) [Ly.l ) w (Y2(1)) ( [ Ly. that is.Ii : M > G be two Coo maps such that . . then it becomes an ordinary dificrclltial equation for dY dt = 2 (d.*(w) La o . we have 0 = j*( {i .Ii * (w) for all lefi im·ariant I forms w. 0 j)* e k . M C�2) == Y2* W == = == It follows thai U we regard Yl a� b riven. we have YI (O) == Y2(O) . and let fi . _ == == == It is also po"ible to say by how much any two such maps differ: 1 8 . == == == j.1 ) YI* w (�IJ * dI · Y2 (of the type considered .iii ) == (q. and write this equation out in a coordinate system.(I).(I)y.li . (I).(I)y. [ (L y. there is a (unique) a E G such that j. J(q)) c r. Then fi and .' ] d y.398 Let OUipter 10 j: U > M x G be t h e map j( q ) Since (jk . (I) _ ' J. Let M be a connected manifold. We must show that YI W (Y2(1)) Case 1.iiJk = 0 On J. == PEDESTRIAN PROOF. j* H tek . w (y.Ii differ by a left translation.(H2 ° i)* a/ ek J*w k .
b} = i=l n ker(Jr) *a/ . By assumption.(G).. Case 3. " Then t.(O) = y. . Case 2.. b E G with . there are a .lr2*(J/ ).*(w) = c* J 2(W) = c* . 399 in the Addendum to Chapter 5). (t) for ali I . it follows fi'om Case 1 that La 0 y. ELEGANTPROOF. 0 c. = y.. then Since La 0 y... ./i *(w) = y/(w) . let Ll(a. M = IR.(O) = a · ydO). w" for the left invariant Iforms. arc the left cosets of t./i (p). y.:* .. Choose a E G so that y. y. b) E G x G. By Case 2. Now define h : M > G x G bv h(p) = Uj (p). s o /2(1') = a · . theu the maximal integral manif olds of t. y. . it follows that heM) is contained in some left coset of t. Since M is connected. (0) = y. . ... so it has a unique solution with the initial condition y. . aJd p ) = bJ ( p) for all p E M . )*(w) = yt ( L:w) = y/(w) = y.. (t) = a . But this solution is clearly y..Lie Group. Choose a basis w ' . (0). Gelleral case. i n particular for t = I . we have For any p E M there is a Le t y. = f.*(w). Let 7[. are arbitrary'. If (J) is a left invariant If orm. In other words. In fact. Let po E M.(G) c G x G is the diagonal subgroup {(a. a) : a E G}.(G). /2 (p)). is an integrable distribution on G x G. if t. Choose a E G so that (La 0 y. Then Coo curve c: IR > M with dO) = Po and c ( I ) = p. : G x G > G be projection on th e ith factor. For (a.(O). but the maps y" y. = y.. .
= \\'hich proves the formula]. If G is a connected Lie group and J: G > G is a map preserving left invariant forms. = La While left im'ariant Iforms play a fundamental role in the study of G. So { G c IR is a compact connected subgroup of IR{O}. the form Ra*w is left invariant. 20. an left invariant nform� arc a constant multiple of any nonzero one. For example. •:. of c. th e left invariant nforms are also very important. == where I note that Lb is an orientation presenTing diffeomorphism. also consider right invariant l1forms. so there is a unique real number with J(a) = Ra'w = l(a)w.ourse. Clearly. Suppose w # O. PROOF. left invariancr of implies that a" in other words. . But in one case they coincide. so L Jan = L Lb*uan) = LU o Lb )Lb*a" = LU o Lb )aN• g(a) J(ba). then determines an orientation on G. COROLLARY. The latter notation has advantages in certain cases. ''\'e can. we can define Coo an an 7 Since L Ja '" a" is llsLlal1y kept fixed in any discussion. If is a left invarjant nform. Hence J G () l(ab) J(ba) = J(a) · J(b). this is often abbreviated to L J or fG J(a)dCl. ( ) = {I} . If G is compact and connected and w is a left invariant Jlform� then (J) is also right invariant. PROPOSITION. and if J : G IR is a Coo function v'lith compact support. These generally turn out to be quite diff erent from the left invariant IZforms (see the example in Problem 25). For each a E G. then J for a unique a E G. iG /(a) da L /(ba) da.400 Ozapter 10 1 9.
) is any Riemannian metric we can choose a biinvariant IIf orm an and define a biinvariant (( .I)) = y(t + u) = y(u + 21). Since for any a E G.) : it is dear that fa is an isometry reversing geodesic through (2) Let y : IR > G be a geodesic with y(O) Y(U) = = e. (I) For any a E G. then (2) The geodesics y with y(O) = e are precisely the I parameter subgroups of G. (I) Since the map 1. )) on G by We are finally ready to account for some terminology from Chapter 9.. In the case of a compact group G there is always a Riemannian metric on G which is both left and right invariant. a. so it is an isometry on Ge. In fact: if ( .I (see the proof of Proposi tion 16 (2)).. F fixed I. G. let or y (1 + u ). i. the map le* : Ga 7 Ga.e.e. Clearly Ie reverses geodesics through e. the map fa : G 2 1 .) (Y(u .)(y (u)) fy(. the maps I 1+ exp(tX) for some X E A. But also . : Ge > Ge is just multiplication by . Since fa = Ra1e Ra .l is also an isometry. 401 Vife can also consider Riemannian metrics on G. PROPOSITION.. if y > G given by 1a ( b) = ab' a is an isometry is a geodesic and y (O) = a. PROOF.Lie Graul). Let G be a Lie group with a biinvariant m etric. Then y is a geodesic and y(O) fy(ljfe( Y (u)) = = Y(I) . So = fy(. which reverses geodesics through fa (y(t)) = y (I). i.
' If VI . . . These are the only geodesics. By continuity.. Vi .. y(l1I) = y(t)n for any integer n.402 so Chapter 10 y(t)y(u)y(t) = y(u + 21). Xk )v. i=l d w = LO/ ' Vi o i=l a simple calculation shows that this definition does not depend on the choice of basis VI . '. we define a Vvalued kform On M to be a function w such that each w(p) is an alternating map w(p) : Mp x . . so y is a homomorphism on Q. . and geodesics through e are determined by their tangent vectors at / = o. . Xk E Mp we have or k times d w(p)(X" " " Xk ) = L w. y is a I parameter subgroup.: i=1 we will write simply ror anv Vvalued kform w we define a Vvalued (k + I)form dw by d dw = L do/ . . then y(t' + I") = y(l)n'+n" = y(t')y(t "). . Vd for V. . We conclude this chapter by introducing some neat f ormalism which allows us to write the expression f dwk in an invariant way that does not use the or constants of structure of G. It follows by induction that If t f = n'! and t ff = n"t for integers n' and n". since there are I parameter subgroups with any tange nt vector at I = 0. . . . . . Vd is a basis for V. . . If V is a ddimensional vector space. (p)(X" .. (J)d such that f X" . . •:. . . . . then there are ordinary kforms cv l . x Mp > V.
We will denote this Wvalued (k + 1) form by p ( w /\ ry). Uc and 403 p: V" . there iI a natural nvalued If orm on G . where V and V have . for any Avalued kform ry alld any Avalued Iform A on G. Now suppose that Xl " ' " Xn E Ge = A is a basis. j=l d and ry is a Vvalued Iform then c d L L w. ]: A x A > A. /\ ryj · p(u. . . Uc or VJ . These concepts have a natural place in the study of a Lie group G. Vd . . namely the form w defined by w(a)(X(a)) = X E A. • • • . a calculation shows that this does not depend on the choice of bases u" • . suppose bases UJ .Lie Groups Similarly. . • . a new Avalued (k + I)form fry /\ A] on G. . • . The form w defined by (*) can clearly be written Then (I) .. vf . Although there is no natural way to choose a basis ofleft invariant Iforms on G. . we have. Vj ) i=l j=l is a Wvalued (k+I)form. If w is a Vvalued kform c w = L {J/ ' Ui i= 1 ry = I >j . respectively. and that cvI . Using the bilinear map [ . • • V x V > W is a bilinear map. Vd. (J)n is a dual basis of lefi invariant I forms.
we obtain the equations of structure of G : The equations of structure of a Lie group will play a n important role in Volume III.0 ..w([V.V(w(V)) . On the other hand. then (Problem 20) we can define a �valued function V(f) on G. V) = V(w( V)) . For the present we merely wish to point out that the terms dw and [w /\ w] appearing in this equation can also be defined in an invariant way. Y] = [W(a)(Xa). For vector fields V and V we can then define dw(U. V = Y for some X. Yo) = 0 . Y]) = [X. Chapter 10 [X" Xj] = L C. w(a)(Ya)] It f ollows that for any vector fields since by definition of Problem 20 gives an invariant definition of p(w/\ry) and shows that this equation is equivalent to the equations of structure. Y] is left invariant [ . ] in G. V) = . h) = w(e)([X. V]). Recall that the value at a E G of the right side depends only on the values Va and Va of V and V at a. by definition of w. . Y]a) = w(e)([X.w(a)([X. Y E G" then dw(a)(Xa.: Xk . I dw(U.404 On the other hand. w(V) is a �valued function on G.[W(V). I f we choose V = X. k=1 n so (2) Comparing (I) and (2).W( V)] · I V and V we have } [X. For the term dw we just modif the formula in Theorem 71 3: If V is a vector field y on G and J is a �valued function on G.
Lie Gmu/i. /\ dx . The appearance of the factor 4 bere has nothing to do with the 4 in the other f orm of the structure equations. V) = . Tbis makes their A /\ ry equal to 4 of ours for Iforms A and ry. It comes about because some books do not use the factor ( k + /)!/ k! I! in the definition of /\ . . ) as L dOJ.\ 405 WARNING: In some books the equation which we have just deduced appears as dOJ(U. OJ(V)].HOJ(U). Then the definition of deL OJ. makes their dOJ equal to � of ours for Iforms OJ. dx .
and H e G a subgroup.ak ax k. E U. IR) we have where M= . (a) Show that U n +1 is a neighborhood of u n . with I(O) = o. .L Mxkl . Show that the tangent bWldle TG ofa Lie group G can always be made into a Lie group.) (c) If G is locally compact and connected. xy) . and hence an element of O(n). and suppose that (x.406 Chapter 10 PROBLEMS is I. an for a. We have computed that f or M E �l(n. then H is closed. (b) Show that I takes planes to planes. (b) Conclude that Un u n = G. (d) Show that any element of £(n) can be written A . Let G be a connected topological group. l . y) 1+ (a) Find I. Show that the closure ii of H is also a subgroup. (c) Show that I is a linear transformation.I when I : G x G > G x G is I(x. (a) Show that I takes straight lines to straight lines.f(c) Conclude that G is a Lie group. lRlA = . 4. 6. U a neighborhood of e E G. 3. and Let u n denote all products a l . r for A E O(n) and r a translation. then so is every coset g H. e) is a regular point of . Let G be a topological group. = (x. (Use Problem 3. . Coco Let G be a group which is also a Coo manifold.y) (b) Show that (e. xy 2. or LA since LA is linear. � a= 1 (a) Show that this means that (It is actually clear a priori that M defined in this way is left invariant. then G is acompact. M(A) = A · M E IRn ' = GL(n.1 n M xkl (A) = " MaI A ka . Let I : IRn > IRn be distance presen'ing. (b) If H is open. . (a) If H is open.) (b) Find the right invariant vector field with value M at 1 . 5. Let G be a topological group and H e G a subgroup. 7. f LA.
I I. and where ljr : V > H satisfies ljr(a) · ljr (b)' = 4>(ab . and hence to �o. (a) Use Lemma 513 to show that (exp sX)(exp t Y) = (exp t y)(expsX). b. provided that U is connected. where V e G is an open neighbor hood of c with V · V. Show that Corollary 7 is false if G is not assumed connected.Lie Groups 407 (a) For each C E G. 1 2.I C U. Let X. Y] (� �) = B= O. defined by a. Y]. Y E Ge with [X. (a) Show that exp QJ) Use the matrices A and B below to show that exp(A + B) is not generally equal to (exp A) (exp B). ]) is isomorphic to £( GO). on some smaller neighborhood of c. Then (A. Show that the set of all ? equivalence classes. ljr). use Theorem 5 to show that exp is a homomorphism on the subspace of Ge spanned by X and Y. Define ( VI . In particular. b E V. for all c E G.sin a sin a cos a 13.I . (b) AO is a Lie algebra.' ) for a. and X 1+ X is an isomorphism of A onto AO. ljrd ? ( V" ljr. ( �) O I ) .) if ljrl = ljr. If G is a group. Let G and H be topological groups and 4> : U > H a map on a connected open neighborhood U of e E G such that 4>(ab) = 4>(a)4>(b) when a. exp(X + Y) = (exp X)(exp Y). (b) More generally. consider pairs (V.· to a homomorphism of G into H . ab E U 9. 10. . b = b · a. (a) GO is a group. ] be the operation on Ge obtained by using right invariant vector fields instead oneft invariant ones. with operation [ . then ljr is an isomorphism from G to Go. we define the opposite group GO to be the same set with the multiplication . show that 4> and ljr are equal even if they are defined only on a neighborhood U of e E G. we define the opposite Lie algebra AO to be the same set with the operation [X. Y]0 = [X. then 4> can be extended unique!. ]. [ . can be made into a covering space of G. In Theorem 5. A= ( 0 a a 0 ) ( = cos a . (e) Use this to give another proof that A is abelian when G is abelian. (d) Let [ . If A is a Lie algebra. (b) Conclude that if G is simplyconnected. and if ljr : G > G is a 1+ a. 8. (c) £(GO) is isomorphic to [£(G)]O = AO.
) (g) Now usc this result and Proposition 9 to give a fancy proof of (c). there is a neighborhood U of e E G such that every element in U has a unique nth foot in U. 0) of S' (considered as a subsct of IR'). (c) For G = S' . witbout using (c). (p)). = <p. show that for the homomorphism det : GL(n. (a) Let U be a neighborhood of the identity ( 1 . show that det exp M = (b) Show that the same equation holds for all diagonalizable M with complex entries. {o/s}. IR) > IR . (The diagonalizable matrices are dense.) (0 Prove the result in part (d) directly.. the map det. Y] = 0. Let . = 0/. and let ry. and trace are homomorphisms. N]. Show tbat no matter how small U is.}.) (d) Using Proposition 9. (b) Show tbat for each n ::: I . ( e ) = O. show tbat there is no neighborbood U which has this property for all n. 0 0/. A more general result holds. compare Problem 715. (b) Using Corollary 51 2. (a) Let (x. Problem 1 3 implies that exp t (X + Y) = (exp tX)(exp t Y ) i f [X. there are elements a E U which have square roots outside U in addition to their square root in U. 1 6.NM) = trace[M. (p) = X(ry. (Look at trace(M N . � dry. etracc M . (e) Usc this fact to give a fancy proof that trace MN = trace NM. 1 5..{O}.{Oll = IR is just M 1+ trace M. Y j = O. (a) If M is a diagonal matrix with complex entries. V) be a coordinate system around e E G with x. Let X and Y be vector fields on a Coo manif old M with corresponding local I parameter families of local diffeomorphisms {<p. it suffices to look at matrices with only one nonzero entry. 1 7. {ryrl is generated by X + Y. ( Y(o/. show that In other words. (Since det. IR) > £(IR . (e) Conclude that it holds for all M with complex entries. : AI(n.(p)) + <p.408 Chapter 10 (a) Show that 14. Suppose that [X. 0 <p.
LaRa .' . 1314.'(O)) a basis f �. it suffices to know that exptX exp t Y = exp{t(X + y) + O(t)}. Thus we have a homomorphism Ad : G > Aut(�). For a E G. and show that the tangent vectors of c lie in the distribution constructed in the proof of Theorem 4. where AUI (A). . isomorphic to G L(n . Thus. consider the map b 1+ aba . (Not even the full strength of (I) is needed. H is a Lie subgroup of G. (a) Ad(ab) = Ad(a)oAd(b). The map is denoted by Ad(a). fJ : (01 ' fJ)'(O) (c) Also deduce this result from Theorem 1 4 (1).l = = 01 ' (0) + fJ'(O) .l (b).e) > G are differentiable. the automorphism group of A. usually Ad(a)(X) is denoted simply by Ad(a)X. Show that H C K. It is even true that H e G is a Lie subgroup if H is path connected (by not necessarily Coo paths). . . IR) if A has dimension n). is the set of all nonsingular linear transformations of the vector space A onto itself (thus. 19. . (c) Let CI ' ' ' ' . By considering or k the map J(t l . . t ) = cd t l ) . Osaka Math. The map Ad is called the adjoint representation. Hint: This f ollows immediately from one of our propositions. The argument of part (a) is essentially equivalent to the initial part of the deduction of (I).) 18. (a) Show that I) is a subalgebra of Ge• (Use Theorem 14. and let H C G be a subgroup of G (algebraically). (d) If H e G is a subgroup and an immersed submanifold.Lie Groups for 409 Coo functions J. Hint: Join any a E H to e by a Coo curve c . Ck be curves in H with {c. Let � c Ge be the set of tangent vectors to all Coo paths lying in H. . Ck (tk ) .) (b) Let K e G be the connected Lie subgroup of G with Lie algebra �. (b) Show thai exp(Ad(a)X) = a ( exp X )a . . such that every a E H can be joined to e by a Coo path lying in H. Show that (e. On an arcwise connected subgroup ifa lie group. Let G be a Lie group. show that (b) If Ol. see Yamabe. . 2 (1 950).]. show that K C H. then H is a Lie subgroup.
This tangent space is isomorphic to End(A).nt space of Aut(�) . (O)(Y) = dt dI 1=0 c(Y). Yj is denoted by ad X E End(�).) The map (ad X) ' Ig + ad X + . • . JR). . (A proof may also be given using the fact that [X . End(�) = n x n matrices. Show that H is a normal subgroup of G if and only if I) = £ ( H) is an ideal of � = £(G). Vd.) Use (d) to show that Ad. � (= Ge ) > . w e have the map Ad*. V J . where V is a finite dimensional vector space. Y E I).at the Jdenuty map 10 of � to Jtself. .410 Chapter 1 0 Ad(A)M = A Mr (c) For A E GL(n. (a) Let J : M > V.+ . Yj. then to regard c'(O) as an element of Aut(A). (f) Conclude that Ad(exp X) = exp(ad X) = c . Yj E � for all X E �. Vl for Jl : M > JR. LxY. . . (e) Since Ad : G > �. Y j Y 1+ [X. Y] + 0 (1 ' ) .) (d) Show that Ad(exp tX)Y = Y + t [X. . �ange. For Xp E Mp. Vd for V. (X)(y) = [X. where End(A) is the vector space of all linear transformations of A into itself: If c is a curve in Aut(�) with c(O) = 10. . (It suffices to show this for M in a neighborhood of 0. . that is. JR) show that J . Show that this definition is indepen dent of the choice of basis VJ . ifand only if [X. 2! (g) Let G be a connected Lie group and H e G a Lie subgroup. . where J = "L1=J Jl . define Xp (J) E V by d 20.JR) and M E � 1 (n . . . AlIt(�) = GL(n. we let it operate on Y E A by (Compare with the case A = JR". .. Vl. with basis X(J) = L Xp (Jl) i=] .
Lie GmU/>5
411
(b) If w is a Vvalued k form, show that dw may be defined invariantly by the formula in Theorem 713 (using the definition in part (a)). (c) For p : U x V > W show that p(w /\ ry) may be defined invariantly by
p(w /\ ry)( Xl " ' " Xk , Xk+ , . . . , Xk+l ) l I = L sgn a · p (w( Xq(l j , . . . , Xq(kj ) , ry( Xq(k+lj , ' " , Xq(k+/) )) . ! l! k q eSk+1
Conclude, in particular, that
[w /\ w](X, Y) = 2[w( X ), w( Y )] .
(d) Deduce the structure equations f rom (b) and (c).
21. (a) If w is a U valued kform and ry is a V valued Iform, and p : U x V
W, then
d(p(w /\ ry)) = p(dw /\ ry) + ( _ I ) k p(w /\ dry) . + [w /\ ry] = ( _ I ) k l l [ry /\ w] .
>
(b) For a �valued kform w and Iform ry we have
(c) Moreover, if A is a �valued m form, then
22. Let G c GL(n, IR) be a Lie subgroup. The inclusion map G > G L(n , IR) > ' ' IRn will be denoted by P (for "point"). Then dP is an IRn valued Iform (it corresponds to the identity map of the tangent space of G into itself). We can also consider dP as a matrix of I forms; it is just the matrix (dx;j ), where each 2 dxU is restricted to the tangent bundle of G. We also have the IRn valued I form (or matrix of I forms) pl . dP, where · denotes matrix multiplication, and pl denotes the map A l+ A1 on G. ' ' ' (a) pl ·dP = p(pl /\ dP), where p: IRn x IRn > IRn is matrix multiplication. (b) LA*dP = A . dP. (Use J*d = d .) J* (c) pl . dP is left invariant; and (dP) . pl is right invariant. (d) pl . dP is the natural �valued Iform w on G. (It suffices to check that pl . dP = w at I.) (e) Using dP = P . w , show that 0 = dP . w + P . dw, where the matrix of 2forms p . dw is computed by formally multiplying the matrices of I forms dP and w . Deduce that
dw + w · w = O .
412
If w i s the matrix o f Iforms w
=
ChapteT 10
(w u ), this says that
/\
d
wij = _ L Wik k
(J)kj .
nience, denote the coordinates x
23. Let G C GL(2, IR) consist of all matrices with a # ll and on GL(2, 1R) by and
Check that these equations are equivalent to the equations of structure (use the form dw(X, Yl = [w(X),w( Yl].)
(a) Show that for the natural �valued form w on G we have
w
so that and are left invariant I forms on G, and a len invariant 2form is /\ (b) Find the structure constants for these forms. (c) Show that
dxjx dy)jx2 (dx dyjx
=
xl' � ( dX dy ) x o '
0
(� � )
x
o.
For conve
y.
(dP) .
p l
= ; ( d;
y dxo+ Xdy )
and find the right invariant 2forms.
24. (a) Show that the natural �((n, IR)valued I form w on GL(n, IR) is given by
where
(b) Show that botb tbe left and right invariant n ' forms are multiples of
(det(xap) ) I
n (dxJ l
25. Tbe special linear group SL(n, IR) C GL(n, IR) is tbe set of all matrices of determinant I .
(a) Using Problem 1 5 . sbow tbat its Lie algebra ,:;((n, lR) consists o f all matrices v\·ith trace =
/\
• . •
/\
dxn l)
/\ . . . /\
(dx1n
/\
. . •
/\
dxnn) .
o.
Lie Grouj>.,
(b) For the case of SL(2, IR), show that
413
where we use for differentiating the equation = I. (c) Show that a left invariant 3form is
/\ /\
vdx ydu vdy J'dV PJ . dP= ( udx+xdu udy +xdv ) ' x, y, u, v xvxll,yXJ2, x2J, x22. u v dx du dy  y dx du dv.
=
Check that the trace is 0 by
26. For M, N E 0(11)
£ (0(11))
=
{M : M
=
/\
_ Mt}, define
/\
(N, M) =  trace M . Nt. (a) ( , ) is a positive definite inner product on 0(11). (b) If A E 0(11), then (Ad(A)M,Ad(A)N)
=
(M, N) . , ) at 0(11)/ is also right
(Ad(A) is defined in Problem 19.) (c) The left invariant metric on 0(11) with value invariant.
27. (a) If G is a compact Lie group, then exp : A > G is onto. Hint: Use Proposition 2 1 . (b) Let A E SL(2, IR ) . Recall that A satisfies its characteristic polynomial, so A'  (trace A)A I = 0. Conclude that trace A' � 2. (c) Show that the following element of SL(2, IR) is not A ' for any A . Conclude that it is not in the image of expo
+
(
_0 o
0  1 /2
)
(d) SL(2, IR) does not have a biinvariant metric.
28. Let be a coordinate system around e in a Lie group G, let Kj : G x G > G be the projections, and let be the coordinate system around (e, e) given by y i = Xi =i = 0 Define 4/ : G x G 7 IR by
x
O]l"J , xi ][(2y. ,:)
414
Chapter 1 0 X, b e the left invariant vector field o n G with X; (e) = a ;
and let
(a) Show that
n a X; = " 1frf aXl j= l
� I, .
.
L
. ,
where (b) Using
LaLb = Lab, show that l La.X; (b)](xl ) = [X,(ab)](xl ).
n / L Vr/ (b) . aqj (a, b) = 1fr!(ab). az j= ] (V,}) be the inverse matrix of 1fr (1fr/), we can write n aq/ _ j (a, b) L 1fr,I (ab) . 1frj; (b). az I=J
= =
. aq/ 1frf (a) = a ; (a, e). o
Deduce that and then that
Letting
V,
=
This equation (or any of numerous things equivalent to it) is known as lie's.firsl J undamental theorem . The as,ociativity of G is implicitly contained in it, since we used the fact that LaLb = Lab. (c) Prove the convene o lie's .first Jundamenwl theorem, which states the following. j ' Let </> == (</> ] , . • • ,</>n ) be a differentiable function in a neighborhood of 0 E lR n z [with standard coordinate system y l , . . . , yn, : I , . . . , n] such that
</>(a,O) = a 1frJ (O) 8j I _; a</> I j a.. (a, b) L 1fr; (</>(a, b)) · 1frj (b)
= _ 
for
a E }R1I .
Suppose there arc differentiable functions 1frJ in a neighborhood of 0 E IR" [with standard coordinate system x I , . . . , xn ] such that
II
i=l
for (a, b) in a neighborhood of O E IR'''.
Lie Grauj).,
415
Then (a, b) 1+ 4>(a, b) i s a local Lie group structure o n a neighborhood o f0 E JRn (it is associative and has inverses for points close enough to 0, which serves as the identity); the corresponding left invariant vector fields are
[To prove associativity, note that
=
n a Xi = L,Vrf axj ' j=1
.
and then show that 4>(a, 4>(b, z )) satisfies the same equation.]
a4>1 (4)�:; b), z) t Vr! (4)(4>(a, b), z)) . v,,} (z) 1=1 [Xi , Xj] L, Ci)Xk k=1
=
29 . Lie's secondJ undamental theorem states that the left invariant vector fields a Lie group G satisfy
Xi of
for certain constants  in other words, the bracket of two left invariant vector fields is left invariant. The aim of this problem is to prove the converse if Lie� second undamental theorem, which states the following: A Lie algebra of vector fields f on a neighborhood of 0 E jRn) which is of dimension 12 over IR and contains a basis for JR"o, is the set of left invariant vector fields for some local Lie group structure on a neigh borhood of 0 E JRn. (a) Choose
Ci�
n " ,'f'/,Ji dxj ) L j=l then the 0/ are the dual forms, and consequently dwk   " C�· wi Wj Ci� constants. L lJ i<j (b) Let Hj : JR" x JRn JRn be the projections. Then H2'Wj HI'Wj t(v,,/ OH2) [d(xl OH2)  t(Vr! OH2) ' HI'Wi] . ;=1 1=1
0/
=
If
X" . . . , Xn in the Lie algebra so that Xi (O) a/axi lo and set n a Xi L,Vrf. ax} ' j=1
= =
>
=
/\
416
Chapter 1 0
ndimensional manifolds on which the forms d ( xl 0 ][2)  ,£7=1 (1frf 0 ][2) . ][ 1 *Wi all vanish. (c) Conclude, as in the proof of Theorem 17, that for fixed G, there is a function <Po : JRn > JR" satisfying <Po(O) = G and or equivalentl:;
Consequently, the ideal generated by the forms d (xl 0][2) ,£7= 1 (1frf 0][2)'][I*W ' is the same as the ideal J generated by the forms ][/w j ][ I *wj . Using the faci that the ej are COllstants, show that d ( J ) C J. Hence JRn x JRn is foliated by
k
d<P� (b) = L 1frf (<Po(b)) . w' (b), ;=1
=
30. lie's thirdfimdamenlal Ilzearrm states that the ej satisfy equations (I) and (2) k on page 396, i.e., that the left invariant vector fields form a Lie algebra under [ , ]. The aim of this problem is to prove the callverse if LU's thirdf undamental :/;mrem: which states that any ndimensional Lie algebra is the Lie algebra for SOlDe local Lie group in a neighborhood of 0 E JR". Let ei) be constants satisfying equations (1) and (2) on page 396. We would like to find vector fields Xh . . . , Xn on a neighborhood of 0 E JR" such that [X,. Xj ] = '£;;=1 e,) Xk · Equivalently, we want to find forms w' with
Now set </>(a, b ) rem.
=
<Pa ( b), and use the converse ofLie's first fundamental theo
a cp� ax j (b )
_ L 1fr,I (<Po (b)) . 1frjj (b). ;=1
n
do}
=

Then the result will follow from the converse of Lie's second fundamental the orem. (a) Let h� be functions on
L C;1 o/ /\ wj . i< j
JR x JRn such that
a h� at
=
ok
r
_
'\' ck
i, j
X L IJ
h;(O,x)
=
O.
j 'IIr
These al'e equations "depending on the parameters x" (see Problem 55 (b)). Note that h;( t , O) = O�I, so that h� ( l , O) = Let ak be the I form on JR x JR" defined by
o�.
Lie
Grauj!.'
417
and write
where ),.k and C{k do not involve ),. k =
j (ik = dxk  L Ci� x ia j . i,)
(b) Show that
ahk L ( ah!.k  ', ) dx'. /\ dx!.. ax!. ax' l<
�
dak = ),.k + (dl
/\ C{k). dl. Show that
(c) Let
Show that
dek
=
dl /\  " eA x'),.j  " " cAe' x' a' /\ a j L lJ L L IJ rs i,} i,} r. � + terms not involving dt.
.  � " " e  2 L L ( ik rs j e' _
(
)
Using
" L ek er.� a /\ aj " Ij " L i,} T,S
i,} T,S i,) T,S
k r eis e'j ) a' /\ a j j
'  � " " (ek ers + ek e'T ) a' /\ aj , is  LL I
2
j
and equation (2) on page 396, show that
dek
=
" I 5J dl /\  " eA x'),.j + � " L dT e' x' a' /\ aj L IJ 2L i,} i,} T,S + terms not involving dl.
=
(
)
Finally deduce that
dek
dl /\  L ej� xj e/ + terms not involving dl. j,/
418
(d) We can write
Cizaj)ter J ()
where gt(O.X) = 0 (Why?). Using (c), show that
k ek = '" gIJ. dxi /\ dxi . L i<j
Conclude that (e) We now haw
ek
=
O.
), k =  I L. c.IJk·ai /\a . . dak
=
L Ci� a i /\ a) + (dl /\ Cik ) . i,j Show that the (arms ,,/ (x) = a k ( l,x ) satis(,' do}  � L C;} Wi /\wl i,j
=
�
2
I,)
j
CHAPTER 1 1 EXCURSION IN THE REALM OF ALGEBRAIC TOPOLOGY
de Rham cohomology, of fundamental properties of the ordinary cohomology which is studied in algebraic topology, Because we deal only with manifolds, many of the proofs become significantly easier, On the other hand, we will be using some of the main tools of algebraic topology, thus retaining much of the flavor of that sul�lect. Along the way we will deduce all sorts of interesting con sequences, including a theorem about the po"ibilitv of imbedding nmanifolds in JRlI + l . Let M be a manifold with M = U ]I for open sets U, ]I C M, Before examining the cohomology of M we will simply look at the vector space Ck(M) of kforms 01 1 M, Let
Tspaces of a manifold. Our main results will be restatements, in terms of the
his chapter explores further properties of the de Rham cohomologv vector
U
iu : U > M iu : U n ]l > U
iv : ]1 > M iv : U n ]l > ]I
Q'
be the inclusion!o). Then we have two linear maps
and {3.
defined by
O'(ev) = (iu"(ev), iv'(ev))
Here iu'(ev) is just the restriction of ev to U. etc. Clearly fJ 0 0' = (): In other wOI'ds, imagc O' C ker fJ, 1\101eovel, the converse holds: ker fJ C image O'_ For, if fJ O" , A 2 ) = 0, then A I = A 2 on U n ]l. so we can define ev on M to be A I o n U and A 2 o n jl and then O'(ev) = O" , A2) The equation image 0' = k e r fJ i s expressed by saying that the above diagram is exact a t the middle vector space. '>\Ie can extend this diagram by putting the vector space containing only 0 at the 419
420
C1wpter J J
ends: the al·ro"� at either cnd of the following sequence are the only possibk linear maps .
I . LEMMA. The sequence
0 +
is exact at all places.
Ck (M) � Ck (V) Ell Ck (V) � Ck (V n V) + 0
PROOF. It is clear that Cl is oneone. This is equivalent to exactness at Ck (M). since the image ofthe first map is {OJ C Ck (M). Similarly, exactness at Ck (V n V) is equivalent to {J bein!! onto. To prove that {J is onto, let {</>u , </>v } be a partition 0(" unity subordinate to {V, VJ. Then W E Ck(V n V ) is w = (J(</>vw, </>uw).
where
</>vw denotes the (orm equal to </>vw on V n V, and equal to 0 on V  (V n V ) :
. •.
By putting in the maps d.
\\"(:'
can expand our diagram as follows,
0  C k (M) � Ck (V) Ell Ck ( V) � C k(V
()
�
Ck + I (M) � Ck +I (V) Ell Ck+ I ( V ) L Ck + I (U n V) � 0
jd
j
j d Ell d j
j
jn
jd j
V) ..., 0
.
j
so that the roWS arc all exact. It is easy to check that this diagram commute� . that is, any tvo/O compositions {rom one vector spaee to another are equal:
_ Cl_ (d Ell d) O Cl = Cl o d fJ d o {J = fJ o (d Ell d)
J d Ell d jd
= d
l _Cl_ l _fJ_
d Ell d
b:cunioll ill the Realm 'IfA(�ehmi( 7ojJolog}
421
OUf first main theorem depends only on the simple algebraic structure in� h erent in this diagram. To isolate this purely algebraic structure, we make the following definitions. A complex C is a sequence of vector spaces C' . k = 0, 1 , 2, . . . , together with a sequence of linear maps
satisf ying dk+' 0 dA = 0, or briefly, d2 = O. A map "' ; C, > C2 b e tween complexes is a sequence of linear maps
such that the following diagram commutes for all k . C,k
d/
l Clk + l � cl+
j
. "''_. C:'
j d! r r
The most important examples of complexes are obtained by choosing C" = C" (M) for some manifold M, with dk the ope ato d on kforms. Another example, implicit in our di cu ssion, is the direct Sum C = Cl EB C2 of two complexes, defined by
s
For any complex
C' we
can define the cohomology vector spaces of C by H " (C)
=
ker d' image dk  '
·
Naturally, if C = l C' (M)), then Hk(C) is just Hk(M). If "' ; C, > C2 is a map between compl exes. then we have a map. also denoted by "'.
x E C,k with
that every dement of H" (Cil is determined by some = o. Commutativitv of the above diagram shows that d/(",k(x)) = ",k +' d, " (x) = 0, so ",k(x) determines an element of Hk (C2), which we define to be ",(the class determined bv x). This map is welldefined.
To
d e fi
ne ", we not
d,"CY)
e
422 GiW/)le.J J for if we change x to x + dl k . When Clk = Ck(M). C/ = Ck (N).I (y)) = which determines the same element of Hk ( C2 ). if V and V arc overlapping portions of S 2 fOJ which there is a deformation retraction of V n V into S l . �ow suppose that \·ve have an exact sequence of complexes fJ Ci o > C > C2 > C3 > O. and Ci : Ck (M) > Ck (N) is J* for J : N > M.I (y)). .I . Cik (x) + d/I (Cik .f* : Hk (M) > Hk(N).. ) > Ci u v .I (y)) = Cik (x) + Cik(dl k . This is nol true . I \·vhich re ally 1l1eans a vast c om mutative diagram in \·vhich all rows are exact. For example. ) > Hk(C2 ) and fJ: Hk(C2 ) > Hk (C3)? The nicest thing that could h appen would be for the following dia gram to be exact: fJ Hk(C2 ) > Hk (C3) > o. Whal does this imply about the maps Ci: Hk(C. then Cik (x) is chanl!ed to Cik (X + dl k . then this map is jus. then we have an exac i sequen lT 0 > Hk(C.I (y) for some y E C/ .
there is y E C. something very nice is true: o II o II IR: 2. Bv exactness of the middle row of (*) . we must check that the result does not depend on the choice of x E cl representing the element of Hk (C3). 0. this implies that dl k+1 (0) = 0.I .I Cr') as .I (x') = dll fJk . > Hk (Cr) > Hk (C2) > Hk(C3) > Hk+l (Cr ) > .)' 423 an exact sequence II 1'\ evertheless.k . Theil o = dl (x) = dl fJk (y) = fJk+1 dl (y) . Ho (Cr) > H0 (C2) > H0 (C3) > H I (Cr) > . THEOREM.EtCU/:r. .on in the Realm qfAlgebraic but not 70/l0/0t.1 (y') . .k+1 is oneone. > 0 is a short exact sequence of so that the following infinitelv long sequence is exact (everywhere): 0 > Since ". and hence z = o. . · E d/ .k+I (z) for some (unique) Z E C k+I . let x' = fJk . So dl(y) E ker fJk+1 = i lnage". Moreover. If 0 > CI > C2 > complexes) then there are linear maps '" fJ C. Let x fJ '" /.J". . I ci with = PROOF.k with fJk(y) = x. In order to prove that ok is welldefined. fJ '" 8 . .le obtain 0 E Hk+I(Cl ) if we start with an element of the form di .I (y'). In this case. thus dle)') ".I (x') for x' E C.k+ l . So we have to show that V. This means that dl (y) = sO we choose d/ . Throughout the proof. so z determines an ele ment of Hk+I (C ): this element is defined to b e ok of the element of Hk (C3) I determined by x. Then x = tli (x) = O.I (y') = fJkdl . diagram (*) should be kept at hand.
then we have an exact sequence (eventually ending in O's): o� U HO l M ) . onto circles. the cohomology of nearly everything can be computed by a suitable application of the MayerVietoris sequence.� corollarv of Lemma I and Theorem 2... . this is left to the reader.. ". together with the first part of the proof that 0 is welldefined. and suppose that O'k(x) E cl represents 0 E Hk(C2 )' This means that O'k(x) = dlI (y) lor some y E C/. The proof that the sequence is exact consists of 6 similar diagram chases. . If M = U ]I. THEOREM (THE I\1AYERVIETORIS SEQ UENCE)..424 Glza/Jler J J 1 t is also necessary to check that our definition is independent of the choice of y with fik(y) = x .'h ere V and V are open. and the open sets 1) and V illustrated below. ''\'(' consider the torus T = S l X S ' . Let x E q k satisfy dl k (X) = 0. It is a worthwhile exercise to check that the main step in the proof of Theo rem 81 6 is preei se lv the proof that kerO' C image 0. HA +1 (M) > . � As several of the Problems shO\. 3. All of Theorem 816 can be derived directly from the followin. '''Ie will supply the proof that kerO' C image o.I . As a simple example. the :Maver.·. the definition of 8 i mmed i ately shows that the image of this element under 0 is precisely the cla" represented by x . Moreover.I (y) represents alI element of HkI (C3). Now So fik. and a deformation retraction or V Viet oris sequence i� n 11 onto 2 circles. •:. Since there is a deformation retraction of V and r H A (M) � HA (U) Ell HA ( V) > HA (U n )1) 2. .
Thus the image of the map H (U) Ell H I (Y) > H I (U n V) is Idimensional.dim V3 + .H ' (U n VI � H' (T) � o. " + ( _ I ) k . So the kernel of this map is I dimensional. " III E1l IR 11 IR � H ' W) Ell H ' (V) � IR /I IR The In ap H I (U n V) > H 2 (T) is not 0 (it is onto H 2 (T)). By induction on k. +:+ 0 = dim V2 /a( Vd . The reasoning used here can fortunately be systematized. .J .. It follows that dim H I (T) = 2. 7 Vk l 7 Vk 7 0 0 = dim h . this map is oneone. then 112 7 . If the sequence o 7 VI + a is exact. Aswme the theorem for k . which implie> that VI = O . Since the map V2 > V3 has kernel a(Yd. 4.I dim Vk . . = PROOF.dim VI + dim V2 .Excursion in the Realm 'IfAlgelJ1'aic T j O JOlogT 425 � HO (TI � HO (U) Ell HO (VI � HO W n VI � H ' (TI .a( VI ) > V3 . .dim V3 + . and consequencly the map H I (T) > H I (U) Ell H I (V) has a Idimensional image. = . so its kernel is I I dimensional. it induces a map V'. So we have an exact sequence of k .J vector spac("� hence which proves the theorem for k. . Similar rea soning shows that this map also has a I dimensional kernel. PROPOSITION.dim V2 + dim V3 . Moreovel. For k I We have the sequenu Exactness means that {OJ C VI is the kernel of the map VI > 0.
Geometric Integration 1 . /j.vard 3simplcxc . .j The subsel of /j. . 1 t is a difficult I heorem that every ('OC manifold has a triangulation.. ." is defined as the set /j. .�+ I L.0.. 0 . . standard tflangulaooll b' In al id \Z!J"lria. m is called a k face of !:J." is defined to be a dillcrent. C M is a dif leomorphic image 0(' some /j. Now by a triangulation of a compact nmanifold M we mean a [mite coHection 1 allj} of diffeomorphic ima. If /j. J " " . although homeomorphic. then the image of a kface of /j.ft ' €SO . ." obtained by setting n . we will use the MaverVietoris sequence to relate the dimensions of H k (M) to an entirely difT�rent set of numbers.. 1 . m . ..n = Jx ) E < < jR'"+1 . and is called a kface of /j... ( 0 0 1 f S' (afiter SI ' �.X i . then for some k the intersection ani n an} is a k ·lacr of both alii and an}.. ��_< 0V ��':he . Elcmc77la�J' Dijjerf1lial Topology.I and ..::.!!cs of /:). The standard nsim pl ex /j.426 CizajJter J J Rather than compute the cohomologv of other manifolds. for a proof see !\1unkres. l>o \ (In Problem 8S .���:oI1S v � A !1\ � � � � � or 2 4 \f mters(. set. arising from a «triangulation" of M. k . n which cover M and which satisf the following condition: y If a n. �. n allj #.k of the coordinates xi equal to 0 is homeomorphic to /j. ''''hitney. .=l xi = Ji J.'cUom of � .". a ne\\' structur� which we will now define.
> Hk (U) Ell Hk (Vn_ I i II n II o 11 .1 where V n Vn _ 1 has the same cohomology as a di�oint union of Gi copies of sn. and let Vn_ 1 be the complement ofthe set consisting of the centers of these balls..I. n II 0 VnJ l > Hk ( U n Vn.I i (3 ) H n'(U II nn Vn. so that Vn_ 1 is a neighborhood of the union of all (n . The :MayerVietoris n sequence breaks into pieces: (2) For I < k � H I ( U ) Ell H I ( V J l � H I (U n II o < Hkleu II (I n Vnd . We let ctk be the number of these ksimplexes. any kface of any an.Excunioll ill the Realm rljAlgebraic TO/Jolog} 427 Theor Assuming that our manifold M has a triangulation {an. one within each nsimplex ani.I (U) Ell H I ( Vn_d � � H� I (U n Vn_d n � Hn(M) � H ( u ) Ell Hn ( Vn_ J l II . Then M = V U 1111. Now let U be the disjoint union of open balls.l . will be called a ksimplex of the triangulation .} we will call ).I i � H n I (M ) � H .I )simplexes of M.> Hk (M) . Consider first the case where 11 > 2. each ani an nsimplex of the triangulation.
. dim Hk ( Vn .dim H n (M) + an . The above equations then imply that X ( Vn .. 111 the manifold Vn._ t l = X(M) .] X ( M) = X ( Vnd + ( .2 This makes sense for any manifold in which all Hk (M) are finite dimensional: we anticipate here a later result that Hk (M) is finite dimensional whenever M is compact.I )k dim Hk(M ) or 5.l (M) .I )k dim H k ( V.'.I we define a new open set U which consists of a disjoint union of sets diff eomorphic to IR".i l = dim Hk ( M) O ::: k � n .2 L ( . � comp ( "eWUonents of (11 = 2) .l [dim H .I For the case 11 = 2 we easily obtain the same result without splitting up the sequence..t l = n. + (I) n an . + (I)" dim H (M). one for each (n .dim H I (M) + dim H'(M) . dim H n. .(. " k =o n n + ( _ I )n. . = k �O n.I ) "a ..al + a2 . We now introduce the Euler characteristic X(M) of M. THEOREM.I (M) . joining the balls of the old U. . .I )face. X ( M) = aO .I ) " an · L ( . For any triangulation of a compact manifold M we have PROOF. . defined bv n X (M) = dim Ho(M) .dim H (M) + a.428 Chapter J J Applying Proposition 4 to these pieces yield.l ( Vn _ l ) = dim H n.
+ ( .2 IS .E+F=2 .rt. . .1 ) 1 0'1 + . . . . ('omplf'ment of � I/n. .11 the Realm q[Algebraic 7opolog1 429 vVe wil1 let 1111 2 be the complement of arcs� in the new V.. . /<� . . . joining the centers of the balls in the old U.Lxcuniol1 . . and F faces. .. : V 6.' . thell V . the last of these is a disjoint union of 0'0 .. . we have X(M) = X(Vn!) + (I )"O'n = X(Vn . the. . .EULER).0'1 + . each of which is smoothly co Iltractil)le to a point. If a convex polyhedron has vertices. while in an other cases we have Combining these equations. .I )"O'n] = x(Vo) + [(. COROLLARY (DESCARTES.3.I ) Q'n . Hence X( Vol = 0'0. • .I )"O'n] n = 0'0 . Vo. E edges. + ( .I)" IO'n _1 + ( .2 ) + [(. . ' (11 = 3) An argument precisely like that which proves the equation X(M) = X( Vnl l + ( . . � .. .I )"O'n also shows thai Similal'I)� we introduce VIl.
let (J) b e a kf())'ffi with compact support on M. V}.) iu' Ell iv' is oneone.he converse. Then w t t = </>uw + ¢vw E C. On tho' �'"::. (U) EIl C.hat O" . we will denme this extended form by i u'(w). and let {</>u . ) is the i mage of At E C:(V n V ) . then w can be extended M by letting it be 0 am side V. each map iu' and i v' is oneone. If V c M is open. </>v} be a partition of unity for 'he cover { V ..</>vw) E C. If C: (M) den otes the vector space of kforms with compact suppOrt on M. if w is a form with compact support C U. A. .(V) satisfies iU'(A t ) + iv'(A2) = 0. suppose . A. I is clear h ac is exact.." w other hand. is clearly 'he image of (</>uw. @support w LEMMA. C V n V. 7. So (A. a form (J) with compact suppOrt C M may not restrict to a fOfm with compact suppOrt C V: the inclusion map of V into M is not proper.. C V. we encounter a very different situation. It is clear that image Uu' Ell iv') C ker ( i u' + iv'). The sequence PROOF. (V) EIl C. in fact.. This means that AI = i. •:.430 ChapteT Jl If we turn from Hk to H. Since support At C V and support A. To prove . To prove that iuT +iv' is OntO. ( V ). thi' shows that support i" C V n V and support A. we can define a new sequence. support w U . .
'" /. . . then so is the sequence of dual maps and space� For any A E W3* we have a* fJ* (A) = a*(A W.e. . for all sn x IR"'. = O. PROOF. > IR wilh A = fJ *( X). Apply Theorem 2 10 the short exaci sequence of complexes given by the Lemma .. COROLLARY. suppose we want to find H. whel'e (U x IR) n (V x IR) is diffeomorphic to snI x 1R2 The only way to use induction is to find H.* W3* � W]* 0 � WJ* .I x lR . (UnV)' > WI a W2 fJ PROOF. •:. + 8 H. Ihen sn x IR = (U x IR) U ( V x IR). . Now suppose A E satisfies a*(A) fJ ) = A 0 (fJ 0 a) = A 0 0 = o. which is diffeomorphic to S. we "vi}} merely record one further resu1t� for later use.on in the Realm qfAlgelnaic 1O/!% g) 431 8. i... If M = long exact sequence .. The details will be left to the reader. ( M)' > [H:(U)EIlH:(V)]* > H. . > H.. . For example.EtcU1. .+I (UnV)' > H. Ifwe write S" = UuV in the usual way. THEOREM (MAYERVIETORIS FOR COMPACT SUPPORTS). (U) Ell H. > H: (U n V) > H. 9. (V) > H. starting with S l X IRm. Then A WI � w2 L wJ A there is �: W. So a* 0 fJ * = O. We just have to show that if the sequence of linear map' )0 )0 W3 is cxaCi at W2. 0 a= O. then there is a dual .:' I. A IR j . then there is a long exact sequence . and then proceed to yet anO! her application of Theorem 2. We claim that = fJ 0 t Given a w E W3 which is . V open in M. (M) This sequence is much harder to work with Ihan the M ayerVietoris sequence.+I (U n V) > .I x IR.. for IR" {OJ.. so Ihat unv is diffeomorphic 1 0 S. If the manifold M = U U V for U. V open in M. U U V fol' U.
Moreover.k ( N ) . the "germs of kfonns in a neighborhood of M". we define a map of complex".w" = 0'(=) for some z.k (N) > gk+ I (N). for if fi (W' ) = fi(w"). + j" g. • • with ni Vi = 1\'. '>\'e now construct l1 sequence of such neighborhoods V = VI :::J V2 :::J V3 :::J . while the image of e contains all W E e: (M) which are 0 in a neighborhood of N. There is a compact neighborhood V of N and a map l[ : V > N such that V is a manifoldwith· boundary. sO that we obtain a complex g. and define X to be 0 on W. we define �(W) = AW). /I' + e C: (M) + i* Ck (N). vVc appeal first to a result fJ"Om the Addendum to Chapter 9." classes into a Vector space f'/ (N). •:. This defines X on fi( W2 ) C W3. C: (M) in the obvious way: (J) � the equivalence class o r any w i lli .A(W") = AO'(=) = O. j such that Wj E Ck(Vj). vVe now consider a rather different situation. Now chooSt W e W) with W3 = fi( W2) Ell W. We will call Wi and Wj wil Vl = Wj I VI · l t is clear that we can make the set o f all equivalence. Finally. This sequence is 1lot exact at C! (M): the kernel of i" contains all W E e: (M) which are 0 on N. To circumvent this difficulty. equivalenl if there is f > i. This makes sense. then l[ 0 j is the identity of 1\'. then w . and if j : N > V is the inclusion.. Let N C M be a compact sub manifold of M. while j 0 lf is smoothly homotopic to the identity of V. so A(W) .N) where e is "extension". Then M N is also a manifold. it is eas\" to define d : g. We therefore have the sequencL C:(M .432 of the form Chapter Jl fi (w'). Now consider two forms Wi E Ck(Vi ). we will have to use a technical device.
If W E C.N. the de Rham cohomology of the manifoldwith boundary V.N has compact support C M . The cohomology vector spaces are isomorphic to Hk (N) for all k. Then there is an exact sequence + o H.k (N)) PROOF. and consequently W = 0 on Vi . there is some i such that Vi C U. (M . This means that i*e(w) = O.f�). + e C. THEOREM. 1 0. B y definition o f fJ. THEOREM (THE EXACT SEQ UENCE OF A PAIR). and A = e(AIM . any element of fJ. Let I : M > [0. Hence AIM . PROOF. If compact submanifold of M. LEMMA. . then W '= 0 in some neighborhood U of N. •:. consequently this element is i*(.) of the complex {fJ. Let M be a manifoldwithboundary.N) is exact.N) > H. The sequence 433 0 > C. Since N i. > Hk (N) is an M is a 1 2 . Apply Theorem 2 to the exact sequence of complexes given by Lemma In the proof of this theorem. But in the next theorem. Finally.N) > .N). . This follows easily from the fact that j* : Hk ( Vi) isomorphism for each Vi . (M). +I (M . LEMMA. it is the object of primary interest . 1] be a Coo function which is 1 on v. (M) + i* fJ. > H. entered only as an intermediary (and we could have replaced the Vi by their interiors). PROOF. having support I C + interior Vi . this means that A Wi = 0 for some i . Hk (fJ. Details are left to the reader.k (N) as 1 ] . compact and n. and then use Lemma I I .ExcU):rion in the Realm ifAlgebraic Topolog). then there is an exact sequence N c . Clearly e is oneone. (M . (M) > Hk (N) 10. with compa�t bound ary aM. which we will need later. •:.k (N) > 0 I I . (M . . Conversely.N). Vi = N. 1 3 .k (N). Then l� E C. and l� represents the same element of fJ. •:. suppose A E C/ (M) satisfies i*(A) = O.k (N) is represented by a form � on some V. ! .
using tubular neighborhoods V. (�") from a knowledge of Hk (S" . then M is ori entable. M is the boundary of each component. Then Theo rem 13 may be used to compute the cohomology of S" X sm . •:. The reader may use Theorem 12 to compute O R O It follows thaI (*) number of components of �"+I . and ffi.M ? 2 . THEOREM. . U R n H. by choosing M to be the closed ball B in �". pact "hypersurface" of �"+I). Let M C �"+ I be a compact ndimensional submanifold of �n+ 1 (a com As a simple application of Theorem 1 3 .'+I (�"+I . {p) X �m). For our next application we wiIJ seek bigger game.M has exactly 2 components. with H. dim H"(M) + 1 ? 2 .'(�"+ I ) � H"(M) � H.M = dim H" (M) + 1 .I ) . we can rederive H. But we also know (Problem 825) thaI (**) number of components of�"+ 1 . ( B) "" H k ( B) = 0 for k '" o.M) � H:+I (�"+I ) � H"+I (M). If M c �"+I is a compact hypersurface. M) gives H. Just like the proof of Theorem 1 2. I\1oreover.n+1 . PROOF. Using Theorem 817. (S" x �m). by considering the pair (S" x �m. of aM in M. the sequence of the pair (�"+ I .434 C hapter J J PROOF. From (*) and (**) we obtain 14.I = a(S" x closed ball in �m).
If the radial function p of a starshaped open set U C jRn is Coo. Neither the projective plane nor the Klein bottle can be imbedded in jR3 OUf next main result will combine some of the theorems we already have. 1 7. This produces difficulties at 0.M.. COROLLARY (GENERALIZED [COOl JORDAN CURVE THEOREM). If M C jRn+1 is a submanifold homeomorphic to sn. LEMMA. there are a number of technicalities involved.M has exactly two components. so every point of M is in the boundal"\' of each of the h¥O components. . Consider a bounded open set U C jRn which is starshaped with respect to o. then we can prove that U is diff eomorphic to the open ball B of radius 1 in jRn. we conclude that dim Hn (M) = 1. then H shows that jRn+1 . The proof in Problem 825 shows that every point of M is arbitrarily close to points in different components of jRn+1 .t 7ojJO/ogy 435 Since dim Hn(M) is either 0 or 1. then jRn+1 . 1 5. We will call p the radial function of U. then U is diffeomorphic to the open baJJ B of radius 1 in jRn. •:.Excunillll ill the Realm ifAlgelna.I > JR . Then U can be described as U = 11x : x E f a certain function p: or Sn I  and 0 5 I < p(x)) S.. 16. COROLLARY. so M is orientable. The basic idea of the proof is to take Ix E B to p(X)1 .M has two components. If p is Coo. However. which we will have to dispose of first. x E U. so a modification is necessary. and M is the boundary of each.
J' E W we have p (y) 2':. Choose IX E U with p(x ) PROOF. 1 8 . .1)/(1) is a Coo function with strictly positive derivative.e for all Y E W. Let I < 1. 1 on sn. the radial function p of a starshaped open set V C ffi. at 0. = Clearly h is a oneone map of B onto U. Since U is open. with a nonzero Jacobian. without loss of generality. 1L 0 ::: 2':. so it is Coo. LEMMA. there is an open ball B with x E B C U. ° /(J) I . 1] be a Coo function with = Define h : B / ° in a neighborhood of ° /' 2':. This means that for . There is clearly a neighborhood W of x with the property that for . 1] > We can assume.I .11 is "lower semicontinuous": for every E: > 0 there is a neigh borhood W of x in sn. In general.I .e . •:. > U by x E sn. At any other point the same conclusion follows from the fact that I r+ 1 + (p(x) . it might not even be continuous. However� the discontinuities of p can be of a certain form only. I t is the identity i n a neighborhood of 0.1' E W the point IJ' is in B.I . •:. At each point x E sn. that p [0. I > p(x) . / : [0.I such that p()') > p(x) . I Coo. (he function p need 110t be h(lx) = [I + (p(x) .436 eha/ller J J PROOF. I < e.1 )/(I)]X. and hence in U.
. lxl is < p(x). If U is an open starshaped set in IRn. (x) are o. Since 1>dx) + . 437 Even when p is discontinuous.11 the Realm ifAlgebmic ToJ)olog). . •:. Let w be a closed kform with compact support K C U. . Then PI+dx). say WX1 " ' " Wx/ for convenience. it follows that p(x) < p(x). We claim that there is a Coo function p : snI > IR such that p < p and This will prove the Lemma. there is a neighborhood Wx of x in sn. (U) = 0 for O :S k < Il. . 19. for then V is diffeomorphic to IRn . IVx.Oll . choose Ix < p(x) such that all points in K of the form ux for 0 :s u :s p(x) actually have u < Ix. be a partition of unity subordinate to this cover. . PROOF. . (U) "" H. For each x E sn. . it looks as if U should be diffeomorphic to IR". We also know that H. . continuous. Proving this turns out to be quite a f eat.I such that Ix may also be used as 1. Since K is closed and p is lower semi K C V = {IX : x E SnI and 0 :s I < p(x)}. 1>. .I . .1lx. and define Any point X E snI is in a ceI"tain sllbcollection of the J.1' for all Y E W. then Hk (U) "" H k (IR") and H. we just have to show that H. LEMMA. since U is smoothly contractible to a point. + 1>1 (x) = 1. (U ) "" IR "" H. . . p..EXCU1J. Similarly.. . and we will be content with proving the following. let 'PI . cover S. K C V. (IRn) for all k. . By Theorem 817.I . Each Ix" . . (lRn). and hence in U. . The prooffor Hk is clear. . and consequently w = d� where � has compact support contained in V. Let IVx" . .
.} such that each nonempty intersection has the same Hk and H: as IRn. and choose a Riemannian metric for M. 20. n· . we can also choose V so that for any p E V the map expp takes an open subset of Mp diffeomorphic ally onto V. } as a subset of 1R2. . . On the other hand. then V is clearly geodesically convex. 3. /II is not of finite type. . It is fairlv clear that if we consider /II = { I . V" V) of M. PROPOSITION. II 0 where M n V has the same H' as a disjoint union of infinitely many copies of S ' . Let M be a compaci manifold. this shows Ihat H' (M) is infinite dimensional (see Problem 7 for more informalion abOUI the cohomology of M). V. . } be a finite cover by such open sets. V. . . and . clear for PROOF B v induction o n the number o f open sets r in a nicc cover. Suppose i t i s true for a certain 1'. U · · · U V. Then the theorem is true for V = V. Ii i s r = 1. . . we first use the MayerVielDris sequence f 1R2 = M U V. If M has finite type. (M) are finite dimensional for all k . In general. 2. such a cover will be called nice. 2. a manifold M will be called of finite type if there is a finite cover {V" . . If p E V. . then M = 1R2 . then Hk (M) and H. Ii follows from Lemma 19 that V has the same Hk and H: as IRn. . . and consider a nice cover {V" . 3. . . Let {V" . where V is a disjoint union of balls or around 1 . . We obtain H' (1R2) II 0 � .:. every point has a neighborhood V which is geodesically convex. H' (M) (fJ H' ( V) II 0 � H ' (M n V) � H2(1R2). n Vi. According to Problem 932. If any V = Vi. . is nonempty. then expp establishes a diffeomorphism of V with an open starshaped set in Mp . .438 Chapter II We can applv this last Lemma in the following way. To prove this rigorously.
the "Poincare dual" of a. One of the fundamental theorems of manifold theory states that P is always D an isomorphism. There is then a unique element of H. ] . Now consider the MayerVietoris sequence . /J. . with orientation IJ. . . (a v f3). We are all set up to prove this fact. and the kernel of a is also finite dimensional. As with most big theorems of algebraic topology. (M) with It is convenient to also use /J.. to denote both this element of H. [�]) We can also define [a! I\ �].(M) represented by any � E C. a We denote this element of that we have a map H. The proof f H. so P D(a) (f3) = v f3 E H. So Hk (M) must be finite dimensional. . (M) + /J. •:. Now every a E Hk (M) determines an element of the dual space H. .l (V n V) Ii H k (M) + The map a maps Hk (M) onto a finite dimensional vector space. in order not to plague ourselves with additional technical details. and the theorem itself is a simple corollar\'.k (M)* by f3 r+ 1 (M. > Hk. the main part of the proof is called a Lemma. but we shall restrict the theorem to manif olds of finite type. 1R.k (M)* by PD(a) . . (M) is similar. or For any manif old + a Hk (V) (fJ Hk ( V) > ' " . . since this has the nice cover {V n V. by the same f ormula. M we can define (see Problem 831) the cup product map Hk (M) x HI (M) "::' Hk+/ (M) r+ by ( [a!].ExcuIJion in the Realm ifAlgebraic 7ojJolog} 439 for U.p. Now suppose that Mn is connected and oriented.) � = J. . (M) and the isomorphism H�'(M) > IR which takes this element to I E 1R. It is also true for V n V. since a! 1\ � has compact support if � does. V n V.
then V PROOF. 1>4 0 = 1>3(X) = 1>3"2(Y) = f321>2 V3. 1>3 the reader. in which the top row is the MayerVietoris sequence. LEMMA.+I(Vn. ffi n � ffi n 1>1 . o. V. and the bottom row is the dual of the Mayer Vietoris sequence for compact supports. except possibly the middle one. If M = U U for open sets U and and is also an isomorphism for all k on M."2(Y) .<VW H. n � is an isomorphism 2 1 . Vve now forget all about our manifold and use a purely algebraic resull. are isomor phisms. So is oneonc.+I(U V)+ H. 1>3 x= "2(Y) = "2 "dw)) = ( (Q o. It is not hard to check (Problem 8) that every square in this diagram commutes up to sign� so that by changing �ome of the vertical isomorphism� to their negatives: we obtain a commutative diagram.(M)· [H.<U v)· Ell � � Let I = 11 .440 C1zapter J J H'IW) H'I(V) H'IW V) H'(M) H'(U) H'(V) HkW V) j PD j PDEIlPD 1PD 1PDEIlPD j PD [H. and U n V.k. and that are isomorphisms. Hence since Thus (Y) · Hence with for some Z Moreover. Thc proof that is onto is similar. : for some Then which implies that y = "dw). 1>3(X) = 0 f331>3 (x) = 0.H. Consider the following diagram." y E 1>2(Y)V2 f3d:) x = E WI . Hence "THE FIVE LEMMA". 1>4. aU vertical maps. for all k on U. = = 1>dw) w E VI .(U) H. 1>4 3 = "3 (X) = 0. This proves the . x E V3. PD V PD � � Ell � � n By as�umption. Then is also an isomorphism. •:. Consider the following commutative diagram of vec tor spaces and linear maps. 1>3 VI � V2 � V3 � V4 � V5 j 1>1 f3I 11>2 11>3 11>4 j 1>5 WI __� W2 � W3 � W4 � W5 PROOF.+I(U) H. and is left original Lemma . 1>2. 1>5 Supposc f some or Then (x) so is an isomorphism. By exactness at there i. Suppose that the rows are exact.
then Hk (M) and Hnk (M) have the same dimension. . By the Lemma. If M is a compact orientable odddimensional manifold. COROLLARY. PROOF Use the Theorem and Proposition 19. Even though the Poincare Duality Theorem holds for manifolds which are not of finite type. Corollary 23 does not. COROLLARY. We begin by considering a smooth kdimensional orientabie vector bundle S = 1f : E > M over M. it is true for M. •:. U · · · U Vr . 23. V. Ell v for the (n + k )manif old E. THEOREM (THE POINCARE DUALITY THEOREM). In fact. = ( _ I ) k+ ' dim Hnk (M) A more involved use of Poincare duality will eventually allow us to say much more about the Euler characteristic of any compact connected oriented man if old M". If M is a compact connected orientable nmanifold. If M is a connected oriented nmanif old of finite type. then the map PD : Hk (M) > H. and for V n V (as in the proof of Proposition 19). V} of M. COROLLARY.k ( M) ' is an isomorphism f all k . •: . •:. the terms (l)k dim Hk(M) and ( _ I )"k dim Hnk (M) cancel in pairs. Suppose it is true for a certain r. This completes the induction step. The theorem is true for V. f M and v f S give an or or orientation /J./II) have different (infinite) dimensions. Orientations /J.ExeulJion in Ilze Realm ifAlgebraic Topolog). Problem 7 shows that H' (IR2 . . 25. and consider a nice cover {V" . or PROOF B y induction on the number r o f open sets in a nice cover o f M . . The theorem is clearly true for r = 1 . noting that V' is isomorphic to V if V is finite dimensional . Let V = V. PROOF In the expression for X(M). . If M is a connected oriented nmanifold of finite type. 441 22. 24.k (M) have the same dimension. then H k (M) and H./II) and H� (IR2 . since E is locally a product. Vr . Vr } is a nice cover of M by geodesically convex sets so small that each bundle S I Vi is trivial. . then a slight modification of the proof for Lemma 1 9 . then X(M) = O. . If { V" .
On the other hand. and let jp : Fp > E be the inclusion map. p. and hence an element vp E H�(Fp). and let � E Cn(M) be a form representing /L. 26. so that A i s smoothly contractible to any point p E A . there is an element jp' U E H� (Fp). . so that J( M. The Poincare duality theorem shows that there is a unique class U E H� (E) such that 0S = identity of M is smoothly homotopic to identity of E.442 Chapter J J shows that {1f ' (U. /L) be a compact connected oriented manifold. Our definition of U states that ( I) Let A C M be an open set which i s diffeomorphic to IRn. This class U is called the Thorn class of $. Pick some closed form w E c� (E) representing U. . Since jp is proper. L 1f'� l\ w = 1.vp) jp'w = l .). Let (M. Also choose A so that there is an equivalence (Fp. Then the Thorn class U is the unique element of H�(E) with the property that f or all P E M we have jp 'U = vp.) PROOF. (This condition means that f where U is the class of the closed form w. the orientation v for $ determines an orientation vp for Fp. THEOREM. . ) � = 1. Our first goal will be to find a simpler property to characterize U. . so E is a manifold of M 1f 'E we haw' 1f s O 1f so 1f' : H I (M) > H I (E) is an isomorphism for all I. Notice also that for the maps s = Osection E. . and $ = 1f : E > M an oriented kplane bundle over M with orientation v.' (Ur)} is a nice cover of finite type. Let Fp = 1f' (p) be the fibre of $ over any point p E M. 1f .
EtC/miDlI in lize Realm ifAl gebraic ToJ)ology 443 This equivalence allows us to identif 1f1 (A) with A x Fp. O) = e me. We will also use 1f2 : A x Fp > Fp to denote projection on the second factor.I (A) with A x Fp. the map jp : Fp > 1f1 (A) corresponds to the map e r+ (p. we can assume that there is some K > 0 such that. I) : lIell < K}. I) 'f. it is easy to see that there is a smooth homotopy H : (A x Fp) x [0. lleII < K ) . under the identification of 1f . wejust pull the fibres along the smooth homotopy which makes A contractible to e. By choosing a smaller A if necessary. Let II II be a norm on Fp. support w Using the fact that A is smoothly contractible to p. e. the form H ' w o n (A x Fp) x [0. Under this identifica y tion. 1] has support contained i n {(q. For the H constructed in this way it follows that m e. 1f2 (e)) = }i> (1f2 (e)). 1) = (p.e) : q E A . e) f e E Fp. Consequently. or which we will continue to denote by jp. support w if lie II ? K . the support of wl1f1 (A) is contained in {(q. A glance a t the definition o f J (page 224) shows that the . 1] > A x Fp such that H(e.
444 Chapter II lIell < K}. on the one hand we have (Problem 817) (4) On the other hand. Thm (2 ) So support A c {(q. [ Jr*/L I\ A bv Stokes' Theorem because the form Jr*/L I\ A is clearly 0 on aA' x Fp (since aA' is (n sional). we claim that the last integral in (3) is O.io*(H*w) ° = = d(lH*w) + I(dH*w) d(lH*w). e) : lIell < K J .*(H*w) . (5) we See tha. e) p rem 714 shows that (jp Jr2 ) 'W .  I ) dimen . Theo form IH*w on A x F has support contained in {(q. (4). Combining (3).w = i. JaA1XF/. Since we have where Jr* /L 1\ A has compact support on A' x Fp by (2) = = ± 0 . i. Now. To prove this. " learly suffices to prove that the integral is 0 over A' x Fp for any closed ball A' C A.
ExcUlJion in the Realm ifAlgebraic ToJ10Iogy
445
This shows that iFp is independent of p, for p E A . Using connectedness. it is easy to see that it is independent of p for all p E M, so we will denote il simply by iF Thus
ip*a!
i*a!.
11fI(A) ,,*� a!
1\
=
JA
[ ,,*� . [ i*a!.
Comparing with equation (I), and utilizing partitions of unity, we conclude thaI
which proves the first part of the theorem. Now suppose we have another class V' E H; (E). Since
it f ollows that V' = cV for some C E
1R.
Consequently.
Hence V' has the same property as V only if c =
1. .:.
The Thorn class V of S = : E > M can now be used to determine an element of Hk (M). Let s : M > E be any section; there always is one (namely, Ihe Osection) and anv two are clearly smoothly homotopic. We define the Euler class X(S ) E Hk (M) of s by
,,
X(s) = s V .
Notice that if S has a nonzero section s : M > E, and E C;(E) rep resents V, then a suitable multiple c . s of s takes M to the complement of suppOrt w. Hence, in this case
*
a!
X(s) = (c · s)*V = o.
The terminology "Euler class" is connected with the special case of the bundle
TM, whose sections are, of course, vector fields on M. If X is a vector field on M which has an isolaled 0 at some point p (that is, X (p) = 0, but X(q) '" 0 (or q '" p in a neighborhood of p), then, quite independently of our previous considerations, we can define an "index" of X at p. Consider first a vector
446
field X O n a n open set U C JR" with a n isolated zero a t 0 E U . We can define x x a function I : U {O} > S"I by I (p) = X(p)/IX(p) l. If i : S"I > U is i(p) = <p, mapping S"I into U, then the map I 0 i : S"I > S"I has a x certain degree; it is independent of FJ, for small e, since the maps i , i2 : snI I U corresponding to < I and <2 will be smoothly homotopic. This degree is called the index of X at O.

Chapter J J
7
,, ~ ~
index 0
index
0
dI/� )V
index
I
~ ~ 2
)�
index index 2
index 1
index
I
index
I
in
�n
index
(_ I )"
h.X is the vector field On V with
Now consider a diffeomorphism h : U
>
V C JR" with h(O)
in W
=
O. Recall thaI
Clearly 0 is also an isolated zero of h.%. 27. LEMMA. If h : U > V C JR" is a diff eomorphism with h(O) = 0, and X has an isolated 0 at 0, then the index of fl.X at 0 equals the index of X at O.
Excursion in the Realm ifAlgebraic Topology
447
PROOF.
by
Suppose first that h is orientation preserving. Define
H : IRn x [0, I J H (x , l ) 
{ h(IX)
>
Dh (O)(x)
O < I ::S I 1 = 0.
JR:"
This is a smooth homotopy; to prove that it is smooth at 0 we use Lemma 32 (compare Problem 332). Each map H, = x r+ H(x, I) is clearly a diffeomor phism, 0 ::s 1 ::s 1 . Note that H, E SO(n), since h is orientation preserving. There is also a smooth homotopy { H, } , I ::s 1 ::s 2 with each H, E SO(n) and H2 = identity, since SO(n) is connected. So (see Problem 825), the map h is smoothly homotopic to the identity, \�a maps which are diffeomorphisms. This shows that f . X is smoothly homotopic to f on a sufficiently small region of x h x IRn  {OJ. Hence the degree of f h.X 0 i is the same as the degree of f 0 i. To deal with nonorientation preserving h, it ob�ously suffices to check the theorem for h(x) = (x ' , . . . , xn' , _xn). In this case which shows that degree f h.X 0 i
As a consequence of Lemma 27, we can now define the index of a vector field on a manifold. If X is a vector field on a manif old M, with an isolated zero at p E M, we choose a coordinate system (x, U) with x(p) = 0, and define the index of X at p to be the index of x,X at O. 28. THEOREM. Let M be a compact connected manifold with an orien tation /J. , which is, by definition, also an orientation for the tangent bundle S = 1f : TM > M. Let X : M > TM be a vector field with only a finite number of zeros, and let a be the sum of the indices of X at these zeros. Then
fh.X = h 0 f 0 h' , x = degree f 0 i . •:. x
X( S) = a . /J. E H" (M).
PROOF.
Let Pl, . . . , p, be the zeros of X. Choose disjoint coordinate systems (U" x, ) , . . . , (U" x,) with x/ (P/) = 0, and let
B, = X, l ({p E IRn : Ipl
I f W E C; (E) i s a closed form representing the Thorn class U o f S, then we are trying to prove tha,
::s
I ll .
f(M,p.) X'(w) = a.
448
ChajJter J J
We can clearly suppose that X(q) " support a! for q " Ui B,. So
thus it suffices to prove that
L X*(a!) = t.l, X*(a!);
X*(a!)
lBi
[
=
index of X at Pi .
It will be convenient to drop the subscript i f rom now on. We can assume that TM is tri\�aJ over B, so that 1f1 (B) can be identified with B x Mp. Let ip and 1f2 have the same meaning as in the proof of 111e orem 26. Also choose a norm II II on Mp. We can assume that under the identification of 1f I ( B ) with B x Mp, the support of a! 11f' (B) is contained in {(q, v) : q E A, IIvll � I }. Recall f rom the proof of Theorem 26 that support J.. c {(q, v) : IIvll Since we can assume that X(q) " support J.. for q
( I)
l
X*(a!)
= =
= [ X*1f,*(j;a!) [
JB
l
X*1f,*(jp *a!)
l
E aB, we have
:::
I ).
X* (dJ..) X* (J..)
by Stokes' Theorem
On the manifold Mp we have
l

JaB
X*1f,* (i/a! )·
ip' a!
= dp
p an (11  1 )form on Mp (with noncompact support).
If D c Mp is the unil disc (with respect to the norm II III and aD c Mp , then
snI
denotes
(2
}SIII
[
p
d [ = laD p = 1 p D
= 1D jp*a! by 111eorem 26, the = 1. that support jp*a!andD. facI C
.
1,�'Cl/nioll ill the Realm ifAlgehm;( 7ojlOlol{1'
Now, for q and X: (3)
449
E B  {pI, we can define
X(q) = X(q)/IX(q)l,
aB
l
>
TM is smoothly homotopic to X: aB
X * 1f2 (J/w)
= =
=
l
[ [ [
>
TM. So
X * 1f2 dp X * 1f2 P X*1f2 P (1f2
0
JaB JaB JaB
by Stokes' 1l1eorem
=
X) *p.
From the definition of the index of a vector field, together with equation (2), it follows that (4)
JaB
[
(1f2
Equations (I), (3), (4) together imply (*) .
0
X*)p = index of X at p.
•:.
At the momcnt, we do not even know that there is a vector field on M with finitdy many zeros, nor do we know what this constant sum of the indices is (although our terminology certainly suggests a good guess). To resolve these questions: we consider once again a triangulation of M. \!I./e can then find a vector field X with just one zero in each ksimplex of the triangulation. We begin by drawing the integral curve. of X along the I simplexes, with a zero at cach Osimplex and at one point in each I simplex. We then extend this picture
29. COROLLARY. If X and Y are two vector fields with only finitely many zero. on a compact orientable manifold, then the sum of the indices of X equal. the sum of the indices of Y.
450
Chapter ]]
to include the integral curves of X on the 2simplexes, producing a zero at onf
point in each of them. We then continue similarly until the nsimplexes are filled.
30. THEOREM (POINCARE HOPF). The sum of the indices of this vector field (and hence of any vector field) on M is the Euler characteristic X(M). Thus, for S = 1f : TM > M we have X(s) = X(M) ' /J..
PROOF.
At each Osimplex of the triangulation, the vector field looks like
with index 1 . Now consider the vector field in a neighborhood of the place where it is zero on a I simplex. The vector field looks like a vector field on IRn = IR I X IRn 1 which points direc inwards on IR I x {O} and directly outwards on {O} x IR,, I . lly
(b) II = 3
E, xcuTsion in Ihe Realm ifAlgebraic 7opology
451
For 1 1 = 2, the index is clearly  I . To compute the index in general, we note that I takes the "north pole" N = (0, . . . , 0, 1 ) to itself and no other point x goes to N. By Theorem 812 we just have to compute signN I . Now at N we x can pick projection on jRn  1 X {O} as the coordinate system. Along the inverse l axis the vector field looks exactly like figure (a) above, where we image of the x already know the degree is  I , so I takes the subspace of sn I N consisting of x. tangent vectors to this curve into the same subspace, in an orientation reversing way. Along the inverse image of the x 2 _, . • • , xn I axes the vector field looks like
so ! takes the corresponding subspaces of snl N into themselves in an ori x* entation preserving way. Thus signN I =  I , which is theref x ore the index of the vector field. In general, near a zero within a ksimplex, X looks like a vector field on jRn = jRk x jRn k which points directly inwards on jRk x {O} and directly outwards on {O} x jRnk The same argument shows that the index is ( _ I ) k Consequently, the sum of the indices is
ao  a t + C¥2

. . •
=
X(M) . •:.
We end this chapter with one more observation, which we will need in the last chapter of Volume V ! Let � = 1f : E > M be a smooth oriented k plane bundle over a compact connected oriented l1manifold M, and let ( , ) be a Riemannian metric f �. Then we can form the "associated disc bundle" and or "associated sphere bundle"
D = {e :
(e, e)
S = {e : ( e , e) = I ) .
5
I)
I t i s easy t o see that D is a compact oriented (11 + k)manifold, with a D = S; moreover, the D constructed f any other Riemannian metric is diff or eomorphic to this one. We let 1fo : S > M be 1f I S.
452
3 1 . THEOREM. A class multiple of X($ ).
Chapter II
PROOF.
Consider the following picture. The top row is the exact sequence
H! (D  SI
�':'ir: D�
' Hk (M)
since HO
a E Hk (M) satisfies Ho'(a) = 0 if and only if a is a
H'(�
for (D, S) given by Theorem 13. The map s: M > D  S is the Osection, \vhile s: M 7 D is tlle same Osection. Note that everything commutes.
HO' = i' 0 (HI D)'
and that
since (H I D) 0 S is smoothly homotopic to the identity. Now let a E H ( M) satisfy Ho'(a) = O. Then i'(HI D)'a = 0, so (H I D)'a E image e. Since D  S is diffeomorphic to E, and every element of H; (D  S) is a multiple of the Thorn class U of � , we conclude that
k
s* = s* 0 e
.,' 0
since extending a form to D does not affect its value on s(M),
= (H I D) 0 i,
(HI D)' = identity of Hk (M),
(H I D)'a = c · e(U)
for some C
ER
Hence
a = s'(HI D)'a = c · s' (e(U)) = c · s'U = c · X(�).
The proof of the converse is similar. •:.
i'.'xclIniou iu flu' Reahn ofAlgebraic 7opolog)
PROBLEMS
L Find Hk(S' x . , . dim Hk = G; )'] X
453
S ' ) by induction on the number 11 offactors. [Answer:
2. (a) Use the MayerVietoris sequence to determine Hk (M  {pI) in terms of Hk (M), for a connected manifold M. (b) If M and N are two connected l1manifolds, let M # N be obtained by joining M and N as shown below. Find the cohomology of M # N in terms of that of M and N.
(c) Find X f the l1holed torus. [Answer: 2  211.] or 3. (a) Find Hk(Mobius strip). (b) Find Hk(JP'2 ) . (c) Find Hk (p"'). (Use Problem 115 (b); it is necessary to consider whether a neighbol'ilOod of 1P'''' in 1P''' is orientable or not.) [Answer: dim Hk (IP'n) = 1 if k even and ::: 11, = 0 otherwise.] (d) Find Hk (Klein bottle). (e) Find the cohomology of M # (Mobius strip) and M # (Klein bottle) if M is the nholed torus. indicated identifications of edges we do 1101 obtain a triangulation of the torus. Why not>
4. (a) The figure below is a triangulation of a rectangle. If we perform the
A
A
454
Chapter II
(b) The figure below does give a triangulation of the torus when sides are iden tified. Find "0, " " " for this triangulation; compare with Theorem 5 and 2 Problem I .
5. (a) For any triangulation o f a compact 2manifold M, show that
3"2 = 2" , '" = 3("0  X(M))  1) "0("0 2: C¥) 2
1
"0 '" 2 (7 + V49  24X ( M ) ) .
=
(b) Show that for triangulations of S2 and the torus r2
S'
x
S'
we have
'" 14. 2 Find triangulations for which these inequalities are all equalities.
r2 :
S2 :
0'0 2: 4 "0 :0: 7
Q' l :::: 6
'" '" 21
"
0' 2
2:
4
7. (a) The vector space H' (1R2  /II) may be described as the set of all se quences of real numbers. Using the exact sequence of the pair (1R2 , /II) , show that H� (1R2  /II) may be considered as the set of all real sequences {a,,} such that an = 0 for all but finitely many II. (b) Describe the map PD: H ' ( 1R2  /II) .... H) (1R2  /11) * in terms of these descriptions of H' (1R2  /II) and H� (1R2  /II) , and show that it is an isomorphism. (c) Clearly H) (1R2  /II) has a countable basis. Show that H' (1R2  /II) does no\. Hint: If Vi = {aij} E H' (1R2  /II) , choose (b" b2) E 1R2 linearly indepen dent of (a , ' , a , 2): then choose (b" b4,b,) E IR' linearly independent of both (a ,' , a ,4, a ,') and (a ' , a 4, a '); etc. 2 2 2
6. (a) Find H; (S" x IR"') by induction on 11, using the MayerVietoris sequence for compact suPPOrts. (b) Use the exact sequence of the pair (s n x IRm , {pI x IRm ) to compute the same vector spaces. (c) Compute H k (S" x sm'), using Theorem 13.
VI. (It will be necessary to recall how various maps are defined. Vo is u nion of shaded VI is union of un shaded C�·. . " ) with "i E H.) 9. (c) Show that if the Poincare duality theorem holds for each Mi.l l . . Show that H. ( Mi). (Mi) and all but finitely many "i = 0 E H. . (Mi). except for the square Hk. ) with "i E Hk (Mi). . "2 . this "direct sum" consisting of all sequences (" I . "3. and U2 . is union ofshaded & . xcwsioll ill lize Realm ifA lgebraic Tapolog). 455 8. (b) Show that Hk (M) "" n i Hk(Mi). V u. "3 .E. "2. this "direct product" consisting of ali sequences ("1 . Show that the squares in the diagram in the proof of Lemma 21 commute.I (u n V) H�+I (U n V)' lPD > Hk (M) > H� (M)' lPD which commutes up to the sign ( . which is a good exercise. (d) The figure below shows a decomposition of a triangulated 2manifold into three open sets Vo. then it holds for M. Use an analogous decomposition in n dimen sions to prove that Poincare duality holds for any triangulated manifold. the only slightly difficult maps are the ones involved in the above diagram. (M) "" EEli H. (a) Let M = MI U M2 U M3 U · · be a disjoint union of oriented IImanif olds.
(Using the notation of Problem 323. Conclude that U represents 0 E Hk (E). and push everything off to infinity. then there is a vector field X on M with only one singu larity.I > snI of degree 0 is smoothly ho motopic to a constant map. thaI X(�) = 0 when � = . It follows. (b) . E M. show that this is 0 for k odd.. . then there is a nowhere 0 vector field on M. (E') are the Thorn classes. and (j. (a) If U E H. This p rovides another proof that X(M) = 0 for odd 11. then there is a . providing anotllCl proof that X(M) = 0 in this case. so that X(�) = O. Let � = . (Begin with a triangulation to obtain a vector field with a discrete set of zeros. in particular.456 Chapter JJ 10.. Using this. (d) If M is connected and not compact. with Thorn class U. prove the Thorn Isomorphism Theorem: The map HI (E) > H:+k (E) given by " r+ " V U is an isomorphism for all I. Join these by a ray going to infinity. then j* ( U) = U' . show that if X(M) = 0. p. then there is a nowhere 0 vector field on M. aM '" 0. Using anticommutativity of /\. X(j*(�))·) (e) If M is a connected manifoldwithboundary. (a) Let � = . Using Problem 826. : TM > M fOl' M of odd dimension. .X at p is ( _ I )n times the index of X at p. (E) and U' E H. with nov\. show that there is a subsel D C M difTeomorphic to the closed ball. . over a compact oriented manifold M.. 0» If M is compact. Using Poincare duality.here zero vector field on M. such that all Pi E interior D.) 1 I . we can form U v U E H. enclose this ray in a cone. f) a bundle map from �' to I: which is an isomorphism on each fibre. If a vector field X has an isolated singulaI"ity at p E Mn . . : E > M and r = . show that the index of . (a) Let P I . Q») Since we can also consider U as being in Hk (E). : E > M be an oriented kplane bundle over an oriented manifold M. (c) it is a fact that a Coo map f : sn. ' : E' > M be oriented kplane bundles.k (E). 1 3. we have f*(x(�)) = 12..f*(X(n) = xW).
1R). This Problem proves de Rham's Theorem. we let S'k( M) denote the Coo singular kchains. We will denote the group of singular kchains of X by SdX). and U n V. (Using the method of Problem 9.) It follows that there is an induced map Rh from the de Rham cohomology of M to the singular cohomology of M. (b) Show that Rh is an isomorphism on a smoothly contractible manifold (Lem mas 17. to show that if Rh is an isomorphism for U. It is not hard to show that there is a chain map r : SdM) > S'k(M) so that r 0 i = identity of S'k (M). and 19 will not be necessary for this. V. r is approximation by a Coo chain]. . then it is an isomorphism for U U V. 1 w. while i 0 r is chain homotopic to the identity of Sd M) [basically. it follows that Rll is an isomorphism f any triangulated or manifold. IR)}. (Hillt: Stokes' Theorem. For a manifold M. (d) Conclude that Rh is an isomorphism if M is of finite type. (a) If w is a closed kform on M. This means that we obtain the correct singular cohomology of M if we consider the complex Hom(S'k (M). IR) be Rh(w)(c) = Show that Rh i s a chain map from { Ck (M)} t o {Hom(S'k ( M).EXCU1JiOl1 in tlze Realm ifA( �ebmic TopologJ' 457 1 4. using the MayerVietoris sequence for singular cohomology. Basic knowledge of singular cohomology is required.) (e) Check that the cup product defined using 1\ corresponds to the cup product defined in singular cohomology. and let i : S'k(M) > Sd M) be the inclusion.) (c) Imitate the proof of Theorem 2 1. let Rh(w) E Hom(S'k ( M). 1 8.
.
We bave just seen that the Hausdorff property is not a "local property". these need not be regular even if they are Hausdorff) On the other hand. a compact manifold is metrizable. so every manifold is locally compact. (I) U of x and an integer n Condition (I) is necessary. JR. we will not consider them. This space may also be obtained by identif ying all points except 0 in one copy of JR with the corresponding point in another copy of JR. (In particular. that almost all "nice': properties of a � anifold are equivalent. (c) M is metrizable. a Haus dorfflocally compact space is regular. then (U n JR) U {OJ is a neighborhood of 0 (in JR). Ev ery manifold is also clearly locally connected. we will now define a manifold to be a topological space M such that (I) M is Hausdorff. there are manifolds which are not normal (Pt·oblem 6).iust like neighborhoods of o. we first note . (b) Each component of M is second countable (has a countable base for the topology). :Moreover. so every component is open. (d) M is paracompact. Although nonHausdorff manifold. The following properties are equivalent for any manifold M: (a) Each component of M is a compact. THEOREM. which are lo or cally like Banach spaces. . this argument does not work f "infinite dimensional" manifolds. Before exhibiting nonmetrizable manifolds. so every manifold is regular. I t consists of JR U {*} where * 'f. are important in certain cases.) 459 U n JR is open. with the following topology: A set U is open if and only if (2) If :>: 0 Thus the neighborhoods of * look . and thus a manifold itself. (2) For each x E M there is a neighborhood such that U is homeomorphic to lR:".APPENDIX A CHAPTER I Following the suggestions in this chapter. for there is even a I dimensional "manifold" which is not Hausdorff. but local compactness is. (By the way. * E U.
.'hos('" union contains all Cj. Clearly C. b . U V". There is a locall\' finite COver of the space by open sets with compact clo sure. U ./. Gellem/ 7ojJo/agT. Un':!. . U · . . (b) =} (c) follows (rom the Urvsohn metrization theorem. or . (d) =} (a) is a consequence of the following. we obta1n open sets Vi with Vi compact and Vi C Vi+ l . and hence =} SECOND PROOF (0) (a) =} (b) =} (c) and (d) =} (a) as before. is clearly open.. . Let M = C. . (a I =} (b) follows immediateh' fi·om the simple proposition that a acompact locally second countable space is second countable. (cJ =} (d) because anv metric space is paracompact (Kelley. ha. U Un . Similarly Do U VI U . is (a) is Theorem 12. : and so On. (Ord1nal numbers will nol be needed for a 2d1111ensional ex ample 10 come later. ' .) oflhe others. for if x is in the closure. The unioll Va U " . . U . . · . c E A (tl"ansitivit\·. ' U U". 160). . \l\'1th compact closure_ Continu1ng in th1s way. U V'I I imersects only Un l+1 • . This pro\'es the Lemma and the Theorem. l. U . U . (a) =} (d). LEMMA.s countable union of C0111pact set�. It lurWi out thaI there are even l manifolds "dlich are not paracompact. Then VI U C2 has an open nl'i�hborhood U:!. If Vo is one ofthe�e. M.. is compact. which are briefly explained here. PlOOf.:. paracompact space is a compact.ith compact closure. where each C. . U [in . It is also dosed. \". A connected. .460 Appendix A FIRST PROOF. II is easy to show iiom th1s that M is paracompact. .. U C. an open neighborhood VI "·. pg. The serond proof does not rei v on this difficult theorem. then x must be in the closure of a finite union of these Vi. = Uo U . because x has a neighborhood \vhich intersects only finitely man}: Thus x is 1n the union. then Vo can intersect onlya finite number lJ1 • . Since the space 1S connected. The cOllSl ruct10n of J11Cse eXalnples requires the ordinal numbers. . it equals thi. locally compact. . . ORDINAL NUMBERS Recall thaI an ordering < on a set A is a ]e1ation such thai (I) a < b and b < C implies a < c f all a .
Some wellordered sets are illustrated below: in this scheme we do not list any of the < relations which are consequences of the ones ab'eady 1isted. . <) are order isomorphic jf there is a oneOne onto function / : A > B such that a < b implies lea) < feb): the map f itself is called an order isomorphism. . 1 .· <w·2 « ··<w·3 < . 2)\ 0 < 1 <2<3< . > An ordered set is just a pair (A .. <) where < is an ordering on A .firs/ clement.. 3 . < w < . < w < w + l < w + 2 < · · · < w · ] < w <w+1 <w+2 < . one and onlv One of the following holds: (trichotomy).. Two ordered sets (A. . < w . (i) a = h (ii) a < h (iii) b < a (also written a h' o {OJ 0<1 0 < 1 < :' 0<1 <2<3 (A (A = etc. An orderinf! < On A lS a wellordering jf every nonelnpty subset B C A has a. b 461 E A. ) (w + I is. . an element b such that b ::5 b' for all b' E B. I ) ) CO. . 3 < .J (2) For all a. that is. O < 1 < 'J < · · .··<w·2 <w·2+ 1 < . .· · < w <w+1 (w is some set . 0<1 <2< 0 < 1 < 2 < · · · < w < w + l <w+2 < ·· · < w · 2 < w · 2+ 1 < ··· < w O< 1 <2 < ··· < w < .· < · '] < . < . o< 1 < ' " < w 1.· <w<w+l <w+2< . O < l < '2 < · · · < w 0 < 1 <2 < . for the prese]]!. and II is easily seen to be an order isomorphism also. 1 . just a set distinct from those already mentionedl 0 < 1 < 2 < . 'J < . . <) and (B. 2. . .... < w . . .I' 0. = (O.Appendi..
Notice that each set on OUf list 1S <1 . the only (wder isomorphism f)'om B to a segmeni of A is 1111 ldentit:: > f(h) = b. It is easy to show that any tWo such order isomorphisms agrC'(.hen' PROOF. TI. If B '" A is a segment of A. just consider the first element of {b E B : feb) '" b): an outright contradiction appears almost immediately. But there lS a much more general propOSltlOll which will settle all cases at once: J. also a wenordered set with tbt same ordering. the "wth H . If (A. the second with the second. Proposition I has a com pall ion. ..B. PROPOSITION. order isomorphic to a segment of the other. In fact.)! The way we prove this ligorouslv is amazingly simple: If feb) '" b for some b E B.rith the "w th". a subset B of a wellordered set A is called al l (initial) segment if b E B and a < b imply a E B. where b' is the second element. If f : B B' C A is all order isomorphism and B' is a segment of A .c. which makes the study of wellordered set' simply delightful. .. •:. on the smaller of their two domains (just cOllsider the smaJJesl element \. <w and 0 < I < . then one i.srI. B = {a' E A : a' < a J : o < I < . It is easy to see that if B is " seglncm of A . For example.en feb') must be b'. ... then B is not order iSOlnol· phl( to A. . <) and (B. of course. etc. <w <w+ I arc nOl order isomorphic because the second has both a last and a next to la�l eien1cnt. It 1S not hard to sec that no two sets on our lis1 are order isomorphic. And so Oll . PROOF.. To do Ihls rigorously.. while the first does not. etc.462 AfJPclldix A Any �ub�et of a wellordered set is.. < ) are wellordered sets.. PROPOSI TION. consider order isomorphisms from segments of A onto segments of B. a i s the first element of A .c?ment of the succeeding ones. We match the first element of A with the first of B. then either B = A or else there 1S some a E A such that In fact. . ('ven for the "wth" eienlellt (the first one after the first: second: third. In particular... until we run out of one . then for the firs! clement b of B (and hence of A) we clearly mus! haw 2 .
due to von Neu mann . let ( A .gnore elements 2:. c¥. <)".oll1orphism class. Given a nonemptv set . represented by a wellordered set (A . I f il ' is noL then its range m� st be all of B (or we could easily extend it) and we arc stl1l done. v·:hkh can be f ound in (he Appendix to Kelley. ordi nal !lumbers..'ellordered set with just one more element. PROPOSITION. .Appendix A 463 thev don't)..'enordered sets. then the wellordered set of all ordinals fJ < " has a particularly simpk representation: it is order isomorphic to the set (A .. we will denote by . They are beautiful:' 3. <) be a well ordered set representing one of its elements ex. ''Almost'': because the condition H(A. instead of '''lith the wellordered sets themselves. which is c1earlv the largest of all.A \'\'C can obviously i.rl from each ordrr l:. To produce a smallest element of . Notice that some ordinals are not of the form " + I for any ". and Propositions I and 2 show that we al most have trichotomy. <) < (B. <) when (A . <) is order isomorphic to a proper segment 01 (B. If it is defined on all of A we are done. E\'ery element < ex is represented by an ordered set which 1S order iS01TI01phic to some pl"Oper segment of A : earh o f these is the segment consisting o f elements o f A less that some a E A .A. then " + I is represented by a \". It determines a segmen t wh1ch represents somt' fJ E . it would be much !llcer to choose one specific well o)'dered :. Each ordinal number ls a horribly large set.A. General Topolog). Suppose wc define a relation < between wellordered sets by stipulating thaI (A .ularly elegant way to do this. Consider the least of these a's. These older lsomorphism classes arc called ordinal numbers. <r! must be replaced by "(A . This fJ is the smallest element of . <).+ I the smallest ordinal after " (if " is represented by the wellordered set ( A . Transitivity of < is obvious. . To obviate this difficulty V'!e' need only '''lark with order isomorphism classes of .:. < ) ! Roughly speaking: An ordinal number is order isomorphic to the set of an ordinals less than 1t. while those of the form " + I are called Successor * Only one feature mars the beauty of the ordinal numbers as presented here.A of ordinal numbers. and dcfine thesr speclfic sets to be thr PROOF. <).. There is a panic. <). < is a wellordering of the ordinal numbers. these are called limit ordinals. larger than all member!' of A). If " is an ordinal numbe. •:.. <) order isomorphic to (B. So all such order isomorphisms can be put together to give another. <) = (B. I\otice that if ex is an o)dinal number.
w' . il would represent a countable ordinal " E SI. 1 . . etc."ts llWlll loned so fal" are countable.464 Appendix A ordinals. . After one comes to and this 1S only the beginnln�� A ll the wellordered Sf. Then SI is uncountable. With a little thought. ( SI . 2. (h is hopeless to tl'Y 10 "reach" n by C'o 'ltinuing the llsting of well ordered �ets begun above. . to a proper segment o f itsdL contradicting* Prop osition J . and ell counter �rts with an ullcountable nmnber of degree� of complexity.. a countable wellordered set).) Although the countable ordinals exhibit ul1countably many degrees of COIll pkxit�: they are each simple in one way: rihe PROOF By Proposition 3. and consequentlv has onlv countably many pre decessors. There are indeed an enormous nUlnber of countable v·. . Our sv example. one can see how the symbol> w3. Let SI bc the collection of aJ! countable ordinals (ordinal. If it were countable. For . 3 + w · 4 + 6 would be used sOlnewhere between w3 and w4 ) : after an these one would need and after all these the symbol <0 pops up. OUl· list of wellordered sets onlv begins to suggest the complexity which well ordered Sets can achieve. PROPOSITION. . and henct' is order isomorphic to an initial s<'glllt'nt of (). this ". is clearly the first uncountable ordinal. see Kelley's Appendix. would appear (svmbols like w3 + w2 . i.tI. . <) is a wellordered set. . We will also denote some ordinals by the symbols appearing before: 0. We have thus established the existence of an uncountable ordinal. •:. fO I· 011e would have to go uncountably far. w + 1. 4.e. any member of SI is countable. . 3 . re presented by n.wuld mean that n is order i�omorphi(' to the C'oJknion of ordinals < Q . A leap of faith is required. By the remark after Proposition 3. so it represents an ordinal ex E (htl.I resohllioll of this paradox. w . . .reDordered sets: represented by * By ddetinll the words cOlllllable ami uncolilltable in this proofolle obtains the "Burali Forti Parmlox'·: the set Ord of all ordinal nurubers is wellordered.
. i). 0) and a could bv represented by a subset of � . thus also by a subset of (00. 1) . such that every f3 < a satisfies f3 < f3n for some 11 (we sa�' that (f3nl is "cofinal" in a). COROLLARY. there is a sequence f3 . The Corollarv to Proposition 5 implies easilv' that the long ray and the long line are I dimensional manifolds. then there would be uncoul1\ablv mallY disjoint intervals in � . < a.0) (w. aside f)'om the line and the circ1e� there are 110 other connected Imanifolds..O) (W+I. . a contradiction. all its members can be listed (in notnecessarilv Let f3 t = Yt and let f3n + t be the first Y in the increasing order) y" Y2. 6. This is impossible . 1) with the order topology (a subbase consists of sets of the form {s : X < xo l and (x : x > . and hence a smallest. the names "halflong linel ' alld "long line" may also be used.0) . then there is a sequence f3 t < f3. •:. namely those bel\'I'eCll the points representln.g. . is represented by a subset of ( 00. Suppose there were one. To disllnguish L + and L. The first example of a nonmtU'izable manifold is defined in terms of n. It cannot happen that a = f3 + 1. then a is represented bv some wellordered subset of �. So by Proposition 5. 0)).{(O. This can be pictured as follows: (0.Q and a + 1 for all a E n. PROOF. ). If a E n is a limit ordinal. no subset of � is order isomorphic to n . < f3) < . YJ .o)  The set n x [0. . O)} is the (open) long ray. for then f3 would be represented by a subset of �. The union of all these sets would then represent ex. 1 ) .AJ)pelldix A 465 5. < f32 < f33 < .O) (w+2. with the order < defined a s follows: (a. PROOF.\'0 ) ) is called the closed long ray (with "origin" (0. 11S1 \vhich comes after fill ' (. If a E n. ). I ) if a < f3 or if a = f3 and s < I . Howevel. < a cofinal in a . PROPOSITION. (1.0) . and L + = n x [0. Then f3. Consider n x [0. Since a is countable. . ) tw·2. If a subset of � were order isomorphic to n. . and we can easily arrange that the subset representing f3i is a segment of the subset representing f3j for i < j. The disjoint union of two copies of the closed long ray with their origins identified is the long line L.0) )' (2. a E n not represented bv some subset of � . . s ) < (f3 .
L x L+ (big halfplane). + But consider now. Y ) o ) = o« (xy.. Deline j� : (�. homeomorphic �F='. with P E (��) and f (p) E JR � o + identified. in fact. we will denote this set by �6.:.) > �� b�' + j� «X. (big quadrant). + to JR2 . L x S ' (long cylinder). (halflong strip. This is a Hausdorff manifold.\".:. The manifold itself is. 8 ) in the product of the closed long ray and S' produces another 2manifold. the following diagram shows two open sets homeomorphic to �2. Consider the disjoint union of �� and ��. (big plancj. There is another way of producing a nonmcuizable 2manifold which doe� not use n at all. for each a E �. Identifying all points « 0. 0) by (x. .\' . another copy of �2. and denote the point (x.)')a) = (a + y. L x IR (long strip). We begin with the open upper halfplane �� = {(x. which might be called the "big disc". 0)." � /_.466 Appendix A Quite a few new 2manifolds can now be constructed: L+ X S ' L + x IR LxL L+ x L+ (halflong cvlinder). y)o. y.' f . y) E �2 : r > OJ and another copy �2 x {OJ of the plane. Define a map /0 : (��) > �� b." '" + q�.. which we will denote by �. y). we could have thro\vn away �� to begin with slnce it is ldcntified by a homeomorphism with (��) . y). say �2 x {a}.
for it has an uncountable discrete subset. (b) ]( we denote (a + 1 ) + 1 by a + 2. any sequence has a convergent subsequence (but L + i� not compact!). (a + 2) + 1 by a + 3. i.y+a = x'. then any a equals f3 + II for a unique limit ordinal f3 and integer II � O.Y)a and (X'. Let c be a "choice function". . etc. Consequently. We mav dispense with �� completely. (Zorn's Lemma may be deduced from this fact fairly easily.\. one is an extension of the oth. 3. are a space homeomorphic to ��� so we will consider �� a subset of the resulting space. then {x } converges to some (b) If Xl � X2 :s: . and c(A ) E A for all A. O)a ] . namely the set { (O. and in the disjoint union of all �� identify each (x. the Priifer manifold. a E IR we wish to identify each p E (��) + with f a(P) E ��. (a) Show that of any two distingulshed wellorderings.. (a) L + and L are not metrizable. (e) From (d). show that if X and Y are disjoint equivalent infinite sets. (d) Show that on any infinite set th"'e is a wellordering which represents a limit ordinal. Given a set X. PROBLEMS (a) A wenordered set cannot contain a decreasing infinite sequence X l > X3 > .{y' E Y : / < y}).e. then X U Y 1S equivalent to 1'. and Problem I. is a sequence in L +. developed in the problems. c(A ) is defined for each set A '" 0. y = e(Y . but it cannot be second countable.) (c) Given two sets. This manifold. The equivalence c1asses� of course..l'+b.. (Thus one can define even and odd ordinals.ApjJendix A 467 In the disjoint union of �� and all ��. show that one of them is equivalent to (can be put in oneone correspondence with) a subset of the other. and related man]folds� have some very strange properties. This space is still a Hausdorff manifold. a wellordering < on a subset Y of X will be called "distinguished" if for all Y E Y.\3 :s n polnt.y')b for which y = y' > 0 and .) L X2 > 2.: (b) Show that there is a wellordering on X.
Hint: Given H : L + x [0. Similarly for L + x �. (gJ If f : L + > � is continuous. and r > s. (d) L (e) More generally. then f is eventually constant. L + x L. &). I] L + with H (x. one can explicitly construct these Slone. Sillee IR: = Q U Un A . (e) Of the 2manifolds constructed from L + or L with HI = 0 and one para compact end. then n n both sequences converge to the same point. L+ x S l and L x S' are nOl homeomorphic.' ([r.. and the big disc arc all distinct. any order topology is normal (completely different proof!.s]) and . L x �. (b) L + x III and L x � are not homeomorphic. Hint: Let A n = {a : the wedge centered at a has width � I /n } . L x L. b) x (0. (a) L + is not contractible. Show that U include. only L + x � has the homotopy type of L +. 6. (a) L + and L are not homeomorphic. (f) If f : L + > � is continuous. defin. (b) HI(L +) Ht {L) O . (c) HI(L + x S') = = Ht {L = = > = 5. n . then one of the sets f' « oo. (d) The StoneCech compactif. L + x L +.O) for every irrational a.cations of L x L. X Sl ) = z.468 Appendix A (c) If {X } and {Y } are sequences in L + with Xn S Yn :5 Xn +! for all 11. (Using Problem 3(g).t ech compactificatlon�.r. a whole rectangle of the form J (a.I)} L +. " paracompact ends'·. (c) Let U be an open set in �� \vhich 1S the union of "wedges" centered a t (a. 0) x for all x. L x L +. (a) Show that the Proler manifold P is Hausdorff (b) P does not have a countable dense subset. 4. 00)) is countablt. some A n 1S not nowhere denst . + (and also L) are normal. L + x L" . Hint: Imitating Problem 119. show that for every I we have ( H (x. (Use (c)).
s ) = Show that H is welldefined and that H(p. (a) If A C M has cardinality c (the cardinality of �). and such thai f lO) = { p ) (Consider functions defined on inidal segments of n vvith these same propertle�! and apply Zorn's Lemma. is a manifold which is not metrizable. (e) Define H : P x [0. It is known that everY second countable contractible 2manifold is S 2 or � 2 Hence the result of ro �structing the Priifer manifold using only copies �� for ratlonal a must be hOlneomorphic to � 2 .s + sy a ) if y if y > 0 : � o. one can require f(rx) to be the result of applying the choice function to the set of all open neighborhoods of the c10sun f(rx) is an open neighborhood of the closure of Up <a f( f3 ) . whose boundary is a disjoint union of uncountably many copies of JR . then C has an open neighborhood with cardinality (. Let M be a connected Hausdorff manifold which is not a point. with con·esponding points o n tht boundary identified. . { (O. = 469 C2 = { (0. but that thev are not contained in dis open sets. and C2 are closed. ( g) The dis joint union o f two copies o f P' .O)aJ for all p E P. (c) Let p E M. O)a a irrational) : : joint Show that C. 1 ] > P by H « x . (b) If C e M is closed and has cardinality c. but which has a countable dense subset. (f) P . 1) E �i U {(O. � )a J'\ I 1 + sy l . y)a y < 0) is a manifoldwithboundary P'. Its fundamental group is uncountable. There is a function f: n > (set of subsets of M ) such that f(rx) has cardinality c for all rx E n.Appendix A (d) Let C" C2 C P be C. Conclude that P is contractible.{ (x . Describe a homeomorphism of this manifold onto �2 8. O)a a rational ) . 7. then the closure A ha. 1( x 1 s + sy �' y  (x �. cardinality r. Alternatively. Y)a.
then M is homeomorphic to S' . (Given p' E M. or the halflong line. the long line. 1 ] ) with the properties of tbe func tion in part (c) is eventually constant. consider an arc from p to p'.) 9.470 Appendix A order.) (d) A function f : n > (set of subsets of [0. (b) Every I manifold M contains a maximal open submanifold N whose topol ogy is tbe order topology for some ordelC (c) If M is connected and N '" M. This is an example of defining a function by "transfinite induction" . (a) A connected I manifold whose topology is the order topology for some . is homeomorphic to either the real line. (e) M has cardinality r. of UP<a f(f3) with cardinality C. Then there is a unique f with the required properties.
with f(xn } = Xn . If (!) is a CW structure for L +. the problem is that a complex analytic structure on �2 mav bc conformally equivalent to the disc.L +p. f(/(/(q ))) > f(/(q)). This implies that it is also diffeomorphic to (0. but it . but this implies that f(x) = x for all x. we obtain Xl > Xo with f(x} ) = X l . \".L+p is clearly homeomor phic to L +. has a limit point Xo E L + . So for some ql > Xo we have f(ql ) '" ql.L +p. A CW stmcture exists on the Prufer manifold. and consequently that the structure on M can be extended if M is a proper subset of L +. Continuing i n this wa)� we obtain Xo < Xl < X2 < . since we can consider fI in the cOlltrary case.L +p. and then it is easy to see P I that we must also have f(/(q)) > f(q). How ever. are all CW. Using the CW structure on L + . Now f cannot be the identity on all points > Xo (for then it would be the identity ever)'\'Vhere. in other words. To see this we need the result of Problem 924any Coo [or CW] structuff on a manifold M homeomorphic to � is diffeomorphic to � with the usual structure. we must have f(q) > q. and hence for L +. . The increasing sequence q. In fact. then L + . f(q). since it is CW).e can get a CW structure on L + x L +. . Reasoning as before. An easy appllcation of Zorn's Lemma then shows that Coo and CW structures eXlst on L+' I do not know whether all Coo structures on L + are difleomorphic. and hence extendable. and L+p denotes all points S p. I do not know whether every 2manifold has a Coo struCtUlT. we can assume f(q J ) > q l . . . and even a CW structure. This sequence has a limit in L + . this follows immediately from the fact that the maps f used for identifying points in various (��) + with a. then it yields a CW structure for L + . and I(xo) = "0 . 1 ). points in ��.Appendix A 471 CHAPTER 2 The long ray L + can be given a Coo structure. f(/(q)). etc. It is knOvvll that there are uncountably many inequivalent CW structures on L + . . a contradiction. If p E L +. These are all distinct. there is no CW map q > p with a CW inverse. the method used fOI" obtaining a CW structUl'e on L + will not yield a complex ana!wic Jlruc/ure on L + x L +.
Kneser. 335340. Let (Y. Fennicae Se ries A. 18. (a) Every twO points X. Arner. Proc. I 25115 (1958). Ann.hich can be immersed in which. E X are contained In a compact connected C C X. Soc.iihlbarkeil. On the other hand. ). Of the Yalious manifolds mentioned in the previous section. Prove that lor I I .L+q . AnaiJ'lische Slrucklur und A b<. pp. p) be a metric space and let f : X > Y be a continuous locall" oneone map. (b) Let d(x. and locall" compact.L +p > 1 2. y) be the greatest lower bound of the diameters of f(C) (in the pmetric) lor all compact connected C containing x and y. where X is H ausdorfl connected. . 10. it is a classical theorem of Rado that every Riemannian surface (2manifold with a complex analytic structure) is second countable. Complex AnaD'lic Manifolds wilhoul Counlable Base. Sic. PROBLEMS L+ . References to these matters are to be found 111 Calabi and Rosenlicht. and not extendable. locally connected. try to deter mine v·. pp.472 Appendix A may also be equivalent to the complex plane. In fact. Show that d is a metric on X which gives the same topology for X q > p there is no nonconstant CW map f : L + . H. 4 (1953). a mod ification of the Priifer manifold yields a nonmetrizable manifold of complex dimension 2. Acad. Math.
y. points in �� being denoted by (x. and these are identified with the set of points (x. In the di�oint union of A and all ��.. II (1960). We will now describe a 3dimensional version with a twist. and the sets z = constant form a foliation of M by a 2dimensional manifold N. y. Z )a (x . c) E A with y > 0 is identified with the set of points (x.a). O) : y < O} C A . Math. 280281. z . Kneser.z". Now the folium containing { (x. the set of points (x. For. The equivalence classes form a 3dimensional H ausdorff manifold M. c .ala E �� with Y > O. y. y. a E IR we identify (X. . y . c . c . y.ala) contains the points (x.· chell A bbildung "wischen iiberiib"ahlbaren Mannigfaltigkeilen. Archiv. Since we can choose a = c/2.al a with y < O. y. y. y. we see that all leaves of the foliation are the same as the ieaf containing { (x. manifold there is an obvious function ". pp. This example is due to M. z + a ) (a + yx. Beispiel eiller dimensio'nserhohenden anab'li. The remarkable fact about this 2dimensional manifold N is that it is connected. On thi. y . c . z) a for y > O for y < 0 with with (a + yx. and for each a E � let �� be a copy of �J. y .z) E �3 : y '" OJ. Let A = {(x. o) a . y.2a) E A with y < O.Appendix A 473 CHAPTER 6 Problem A6(g) describes a nonparacompact 2manifold in which two open halfplanes are a dense set.
) 4. contradicting the fact that the tangem bundle is not trivial. A Lie group is automatically paracompact. Notice also that if M is nonparacompact.474 CH APTERS 7. 10 I. the tangent bundle TM cannot be trivial. Since there is no Riemannian metric on a nonparacompact manifold M. if f: N M lS an immerslon. a basic result about bundles says that a bundle over a paracompact contractible space is trivial. 3. . O)a) of the Prufer manifold. 6. Thus. submanifolds. not necessarily compact. Y. Although the results in the Addendum to Chapter 9 can be extended to closed. 2 7 2. Thus the tangent bundle of the long line is not trivial. So there are at least two inequivalent nontrivial bundles over M. but a much easier proof is now available: Let ( . Compare pg. We have seen that any paracompact Coo manifold has a Riemannian metric. Problem AI I implies that a manifold N immersed in a paracompact mani fold M is paracompact. a locally compact connected topological group is acompacl (Problem 104. for the folium is a submanifold. then N has the Riemannian metric f* ( . since its tangent bundle is trivial. The converSe also holds: slnce a Riemannlan metric determines an ordinary metric. More generall\'. as can be seen by considering the Odimensional sub· manifold {(O. they cannot be extended to non· paracompact manifolds. and hence paracompact. (On the other hand. then TM lS definitely not equivalent to T* M. The tangent bundle of the long line L is clearly orientable.. a Riemannlan metric on M. 7 V'ie can now dispense with the argument in the proof of Theorem 66 which was used to show that each folium of a distribution on a metrizable manifold i. even though the Prufer manifold is contractible. Appendix A 9. 2. nor is the tangent bundle of the Pnjfer manifold. since an equivalence would determine a Riemannian metric. 5. so there can· no/ be a nowhere zero Iform w on L! for w and the orientation would de termine a nowhere zero "ector field. Theorem 79 fails for L. ) b. ). also metrizable.
Is there a nowhere zerO 2form on the various nonparacompact 2mani folds which have been described? . It i s not clear that a nonparacompact manifold cannot have an lndefllll](' metric (a nondegenerate inner product on each tangent space). proved in Volume II (Chapter 8.Ajijlclldix A 475 7. PROBLEM 13. Addendum 1). TIlis will b.
.
i) Till" T' Xp J . }. W ) Tk < v ) Tk(i' ) T V) d T.i)p np TM T(M. . VnJ IA � Ix. �.· dl de di t(l ax l a 76 68 64 75 68 64 103 82 83 81 76 64 80 84 72 101 102 81 80 83 l CHAPTER 4 Hom( V. . v p a .· ' CHAPTER 3 E"(X) J. 61 34 34 28 34 35 61 6] v" v { f) Iv) . p I'm F 65 72 83 65.'(V) Tt' (� ) 7ik ( V ) I' J p' End(V) df d. � Ell 1. 1 1 9 1 13 131 109 1 16 1 16 1 20 121 121 1 22 I ii . . a a ar' a8 DJ � Ip ip 62 29 61 62 28 35 39 36 d. 'U) SL(n. �) O (n) R(n) (�n .. �) SO(n ) (x. x �2 � Mp (M. U ) aJ ar I ax. (p). 1 1 6. 75 65 101 T'M T®S 109 1 10 1 21 107. ax .' X At C' CO COC CW DJ(a ) GL(n. i.NOTATION I NDEX CHAPTER ] CHAPTER 2 M p2n pn sn aM JRn 8(X) lHln 23 19 4 S' 11 19 1 6 19 .
(. 234 202 Ali curl X All I dx ' /\ .) / ')" r" .c) f' ' . II.. . g ) mj '[La) wa d f l(LI) i1·w Iw dw M. Vk ) IA I (/(1 2 ) exp A Lx A Lx f Lx ) " Lx w [X. . II P II Lx w S. 1'·* V' 1 23 10. .. 2 1 3. 8fi. Y] CHAPTER 8 263 268 250 284 275 258 252 290 297 292 274 263 268 246 249 296 283 283 239 264 275 293 263 Bk (M ) B� ( M ) cu. " . CHAPTER q" 7 202 205 238 238 2 19. . X I.h. 234 23. /\ dxn l de den d8ta . 5 171 161 171 1 74 1 5 (1 1 50 1 50 1 ii 1 53 1 43 1 35 1 44 7ik1ml (V) 7i�ml ( V ) 7ik1n.�:��/ .. x) (01. . 1 0. 224 215 22.il.n OIx(l) = 01 (1 .I/..wI 7i� ( V ) T°(V) 231 231 231 231 20 1 227 206 202 227 215 20 1 201 203 v.J w <PI a· /\ a· CHAPTER ("" Q(M) Q k (V) QO(l ') (v.wl (V) n.i l .f Hk (M ) H� ( M ) I' deg f div A" de t(f. Vk ) (v" . 1 29 1 34 1 34 1 08 1 09 1 0.«) ('R. 235 2 10. 10. 2 1 5.478 N olation Inde> T (I. w(X) Iv t' �j l . signp J w(p) Zk ( M ) ..
288 294 246 299 299 266 9 322. 257. I II II. �) g[(". 259. I Jo CHAPTER A ( ( ( ( ( ( ( « I II II rti . ).u E(II) Elld(nl exp O C O B C' 'J [ij. )' . ) (g.. } .J ) exp d. 248.' // / J dx + g dy /w Lw # (' (q) u [O. 313 356 356 349 335 335 318 319 314 353 328 30 1 315 308 301 349 305 301 367 303 303 315 10 384 372 409 410 409 396 402. 246. .N olalion il1dc. 285 285 260 246 239 243.!. 404 373 410 385 385 372 407 407 376 CHAPTER d(p. }/ ..\ 479 Z.) . 245. �) . q) ds d l' ("osh ("osh . 356 356 314 314 31 I 309 309 324 334 302 306 32G 312 344 sinb tanh x'( v) xi r . ). 291 264 264 263 248.1 IAI A i) Ad( ) ad X Aul( g) a Euc(V) Euc(�) E(y ) J' ( .. HI ii(u) oj 0. )r . )' .(M) 6n a' 0' [wJ al' di ( 268 285 29 1 264./J L� Mp" exp(A) GL(II.
N otation index 401 374 379 376 388 376 41 1 41 1 374 373 376 379 380 395 403.: <jJ. u . 410 403 376 403 400 400 400 403 CHAPTER I I L.1 R" SO(n) X .Ix 6" [' M # !I' p. 0(13 ) .I [ f(o)da " . 'II p(W A ry) [ . APPENDIX A k=d L W' U! to Q Old <> + 1 464 463 464 464 461 46 1 463 .] [ry A A] p x (M) x (�) 419 446 4 32 453 439 457 442 426 423 435 428 445 439 w (natural nvaluecl J f ormJ [ fa" [ . £(G) o(n) P Ck (M) 9!(N) PD RI.480 i. L.
19J Besicovitch. 348. 248. 7 J BuraliForti Paradox. ] ] I tangerlt. 7 i trivial. 1 54 in gl(n. 45 I sphere bundle. �). 28 Choice. 309 induced. 293 AJgebraic inequaJities. 466 plane. 391 submanifold. 20i multilinear function. 2� Auslander. 7 1 normal. R Antipodal map. 19.. 28 maximal. 46b half plane. generalized. 344 of contra\'ariant tensors. 218. 202 Analytic manifold. 248. 27�: point. 464 Banach space. 1 9 long ray. 376. 252 geodesic. 36 i halfspace. 7 ) Basi� dual. 380 Alexander's Horned Sphere. 228 Annulus. ] 20 of covariant tensors. 246 Associated disc bundle. L n. J. 106 quadrant. 20 subgroup of a Lie group. 409 Area. 378 in o(n. I CJ Boy's Surface. 200 Cayley numbers) 100 Chain. I I Arclength. 1 03 Ado. 451 Atlas. 60 Bracket. 360 Cartan's Lemma. 379 Bundle cotangent. 20S f g. Elie. 1 07 for Mp'. 285. 20J Alternation. F. 466 Biinvariant metric. 40J Boundary. A S" 1 7 9 Big disc.k(p). L. J 45 Base space. 35. 46() Calabi. 2 1 0 vector. 7 3 nplane. Fundamentru Theorem of. 23(l CauchyRiemann equations. 1 1 1 Chart. 252 Bounded manifold. 465 manifold.. 34 Annihilator. 1 9 subgroup of a Lie group. 3 9 . �). 285 Chain Rule. 3 1 6 Cartan. 6 Closed form. 208 or Belongs to a distribution. 3 1 3. 31 2 function.INDEX Abelian Lie algebra. 4!=l . 38 Change) infinitely small. principle of irrelevance of. J O� dual. 1 0 1 map. 46� Circle. 382. 7 2 . 55 AJgebra. 395 Adams. 3 3 2 Arcwise connected. 283 Choice function. £" 472 Calculus of variations. l Oll Adjoint T* of a linear transf ormation T. 108 fibre. 233 Alternatinc ("ovaria� t tensor field.
. 30 Dilleomorphism. 78 of a ring. 28. manifold. 421 analytic structure. 5(1 Cramer's Rule. 29 manifo]dwithboundary. 232 DiAeomorphk. invariant. 220. 1 20 vector field. 32 at a point. 429 Coordinate lines. ]7 9 form. 25 DescartesEuler Theorem. 82 Coorelated. 133 Derh<ation.10 pointfinite. 3i3 Conjugate. 123 Decomposable. 209 odd scalar. 259 relative scalar. 275 mod 2. 299 Cube. 236 Contraction. 24b index Cup product. 3 1 manifold. 23 J scalar. 32 manifold.482 Closed (contiuued. ] 2] . 34 CO manifold. 420 Commutative Lie algebra. J O �l Covariant functor. Uy. 283 integral. 263 group of M with real coefficient:. ] 30 tensor field. geodesic'. 308 structure on TM. 465 Cohomology. 1 39 Contravariant functor. 30 on the long line. J 6{1 Cofinal. 34 C� distribution. 4 i ) numhers of norm 1 . 130 tensor field. up to first order. 83 Derived set. 40 i ll the calculus of \·ariations. vector field. 284 Debauch of indices. 238 Cvlinder. 283 Darboux's Theorem. 320 Critical value. 207 function. 286 Degree.. 15� Coordinate system. I 1 3 Convex geodesicaU)" 363 polyhedron. 22i Lemma. 299. 148 DiAerentiable. 65. 39. 40 Cross section. 227 Crosscap. scalar. 32 Riemannian metric. 39. 4 19 d e Rham. 32 . 372 Critical poim. 60 refinement of. 1 1 3 COWl locally finite. 28 Darbom: integrable. 358 Constants of structure. 3 1. 42 [ Commutative diagram. 38i Contractible. 139. 47 1 on pn. 14 Crossproduct. 429 Determinant. 29 structure. 225. 39b Continuous homomorphism. 8 C. 27. 341 Complex. 279 Degenerate. 214 Deformation retraction. 1 33. 15H Coordinates. 228 Definition. 3ib Complete. 28 Cotangent bundle. 295 Densin· eve. 133. 28. I I . 439 Curl. 30 oneparameter group of. singular. 263 of a complex.
16S forms. 1 69 Equations of structure. 1 8ll Divergence. 64. 467 relative scalar. 190 Frobenius Integrability Theorem. 32. 207 differential. 7 1 Finitt> characteristic. 429 characteristic. 89 Fundamental Theorem of Algebra. Five Lemma. 72 weak. 215 Fubini's theorem. space. 358 Equations depending on parameter�. 96 Euclidean metric.ntion. 438 First element. 293 Fundamental Theorem of Calculus. 352 Domain. 324 Envelope. 1 92. 164 depending on parameters. 428 class. 30 Differential. 433 of vector bundles. 3 Du Bois Revmond's Lemma. 45 1 Discriminant. 4� End.. 3 19. 179. 422 of a pair. 355 . 468 Endomorphism. 374 nspace. on �n . 134. 419. 209 Exact form. 3 1 S motion. 201 of a function. 23 paracompact. 4 Direct sum. 169 linear. associated. 334. 194 Folium. 139 Foliation. 1 0 7 vector bundle. 39 Elements of norm 1 . Exponential of matrices. l OB Euler. 421 Disc bundJe. 238 Theorem. 254 . 308 Elliptical nonEuclidean geometry. 2 1 5 o n torus. 205 type. ponential map. Dual basis. I Faith. 194 Force field. 36i Embedding. 374 right invariant. 32(1 Even ordinal. 201 left invariant. 10. 1 03 E.' Differentiable (cOTlliTlurd (structure con1iTllled. 240 Form. 440 Fixed point.. 23 1 relative tensor. 384 Extension. 400 Irelated. 464 Fibre. 254 Functor. 1 36. 385 . 23 1 scalar density. 2 113 sequence. 305. 20 Distribution.. 109 Dimension. 1 30 Functorites.Inde. 285. 68. 404 Equivalence (or vector bundles). 320 483 Einstein summation conv(. 4. 233 Disjoint union. 29 on s n . 461 First variation. 1 8 1 ideal of. 432 Extremal. 2 1 0 equation. 133. leap of. J 2 1 Energy. 445 Euler's Equation.
101 orientation. 1 7 9 Ha s one end. 462 Inner product. 136 of integral ClJrve. 356 Ideal of a Lie algebra. 1 9 Handle. 363 ' Geodesy. 39. 3 33 Germs of kforms. 46 Implicit function theorem. 277 Homotopy. 227. 106 HahnBanach theorem. 194 Gradient.484 Gauss's Lemma. 342 Inde> Hyperboli( cosine. 3 1 4 Infinitesimal generator. 410 Identification. 351 Induced bundle. 46S strip. 104. 123 raising and lowering. 387 of Lie algebras. 304. 69 HalfIonp cylinder. 432 Global theory of integral manifold�. 3 7 J Guillemin. 46 lOpologi �al. H. 23i GramSclllnidt orthonormalization process. 349 Index of vector field on a manifold.. 283 Riemann. 30l preserving. 49 topological. 37� Generalized area. 3 1 4 Initial conditions. 459 Homogeneous. H. 350 Independent infiniteBimals. 44i on �n .. 301 Inside. principle of irrelevance of algebraic. l 36 Initial segment. 466 Halfspace. 372 usual. 104. 145 Hair. 277 Hopf. 372 topological. 380 Homotopic. 61. 341 Geodesicallv convex. V W. 304 Gmk. 3 1 2 Infinitely small cbange. 3 7 1 matrix. 337 General linear group. 246 Geodesic. 283 . 84 Group Lie. 192 Integrable functioll Darboux. 14. 356 sine. 23:. G. independent. 189 Integrable distribution. 260 Inequalities. 7 Homomorphism continuous. 356 tangent. 14 Immersed submanif old. 3 i 2 opposite. 349 Infinite volume. 40i orthogonaJ. 2 3 Heiukin. I nertia. 401 Geodesically complete. H Hardy. 466 line. 333 closed. 342. 1 0 Imbedding.. 60 Indefinite metric. I I I Infinitely small displacements. 36. 47 Immersion. Sylvester's Law of. 148 Infinitesimals. Robert A. 450 Hopf Ril1owde Rham Theorem. 2 1 Integrability conditions. 44 6 Indite:debauch of. 314 Index of inner product. reversing map. 84 Hausdorff.
409 closed subgroup of. 243. 103 contraction of. locally: 382 lsomorphism. 165 svstems of. principle of. 155. ]. 40 J ordan Curve Theorem. 388 Lie $ubgroup... 4 1 5 oneparameter group of local diffeo morphism. 410 topologically isomorphic. 376. 415 normal su bgloup of. 460. 407 Lie derivative. 4 1 6 Limit ordinal. 463 ' set. 3 76 abelian. 179 Lives at points. 463. 150 Lie group. 138 Littlewood. 380 ideal of. 60 Line integral. 136 Darboux. H. 179. 296 Lipschitz condition. 1 7 J Lin� ar transformation ac\joint of. 305. 128. 283 line. 1 8 1 maximal. 3 7 1 arcwise connected subgroup of. 395 commutative. 226. 194 Leap of f aith. Sylvester's. 373 Lie's fundamental theorem first.ket in any ring. 340 Isomorphic Lie groups. 294 Jacobi identity. 382. 391 local. 7 1 Local theory o f integral manifoIds� 190 Kelley. 239.]. 435 Kneser. 18. 415 third. 230 Laplacian. 2 J 4 Irrelevance of algebraic inegualitie�. 1 36 Riemann. 104 positive semidefinite.. 34 9 Leaf. 243 manif old. 374 Length. Integral curve. 121 positive definite. S. 368 Local flow. 148 spanned locally. 145 Laplace's expansion. 136. 283 surface. 414 second. 394 n form. 232 definition. . 243 Linear differential equations. 1 1 9 Lobachevskian nonEuclidean geome try. 376 homomorphism of. :M. 374 Left translation. 194 of a differential equation. 179 triviality. 378 Jacobian matrix.. 245 Integration. 58 Lav�' of ] nertia. 3 1 2 o f a curve) 59 Lie algebra. 43S 485 vector field.. E. 376 for the blac. l O P Isotopic. 464 Kink. 104 Linking number. 400 . 473 Lang. 21. 233 Isometry. 472 Kneser.inde. natural. J44 Lie group. 239. 366 Klein bottle.. 239 In\'ariance of Domain. 410 opposite. 464 Left iJwariant form.. Invariant.
73 rank of. 20 connected. 46] open. 424 for compact supports.]. E. 4 . 308. 401 Euclidean. S.]. 50 isomorphic Lie groups. 466 N onorientablc bundle. 2�1 dimension of. 2�1 diflerentiable.1 Mobius strip. 1�1 C' 34 Co: 34 Coo. 465. 304. l OCi Magic. M . 42 1 bundle. 20 finite cover. 208 Multilinear function. 43 J :Measure zero. 403 Natural isomorphi. J06 MacKenzie. 34. 4 6G orielltation of. 4 ( 1 Massey. J 9 closed. 3 1 2 spaces. 460 closed. tubular. 4 imbedding in �N. 19 Nondegenerate. 367 Lobachevskian.'. 9 nmanifold. I . i Natural gvalued Iform. 3 1 5 indefinite. 52 integral. 20 Milnor.486 Locall) compact. 2 8 boundary o C 1 �l ' bounded. 1 8 1 maximal. 283 index MayerVietoris Sequence. 194 ndimensional. 86 manifold. 1 39 oneone. 344 . 305. ](l generalized. left invariant. 37:2 Normal bundle. Coo. 466 line. W. :)7�' Maximal intcgral manifold. 30 I NonEuclidean geometr>" elliptical. 3 1 1 usual. 438 Nonbounded. 382 Lipschitz. J 15 Munkres. 7 1 nsphere.. 239 Metric biinvariant. A . 20 Long cylinder. 179. 4 nforms. 36H Nonmctrizable manifold. 3 I\'latrix grou ps. 2�1 Manifold. V. 3 Nice cover. 345 Newman. 303 preserving. 42 Mod 2 degree.. Ht) smooth. R.m. i ntorus. 108 Neighborhood.. 465. I�l . 2 1 4 Manifold. 46:} Lower sum. 8(1 Norm. 29. 4 J Mesh. 3 4 ' atlas for. 400 nholed torus. R.<.vitllbnundarv. Ion Multiindex. 46" ray. 1 9 4 nOllmctrizable. disjoint union of. 350 Riemannian. 4 5 9 analvtic. 4 nplane bundle. 3� Map betweell complexes. H . 40.. 1 3 path wise connected.
. 259 Onedimensional distribution. 102 tensor. 88. 450 Poincare dual. 3 1 2 Pig. 86 manif old. 20 Piecewise smooth. 21 Outward pointing. local. 167 Parameterized by arclength . 88 Proper map. 384 Opm long ray. 148 of local diff eomorphisms. 60 Polar coordinates. J06 Positive definite. isomorphic. 245 of unity. S. 407 Lie algebra. 104 Product of vector bundles. 351 Odd ordinal. 52 Path wise connected. 465 map. 148 Oneparameter subgroup. 468 Parameter curves. 467 relative tensor. 32 Projective plane. 459 end. Opposite group. 239. 84 preserving. 98 outward. 461 Ordering. l l6 Projection. H. 348 Orthonormalization process. special. 209 Outside. 459 subgroup of a Lie group. 266 Polarization. ] 34. 30. 104. 35 Partition. 100. 225 Poincare upper half plane. 260 Pointderivation. 441 PoincareHopf Theorem. 465 Ordered set. 60. 60 submanifold. I I. 105. 88. 260 Outward unit normal. 463 Orientablc:bundle. 19. 410 outward unit. 304.index 487 Normal (cantinued) space. 304 Pollack.. 288 scalar density. 275 Prufer manif old� 467 Pseudometric. 434 Poincare.. 248 Orthogonal group. 367 Point inward. 461 topology. 450 Poincare Lemma. yellow. 248 reversing. 435 space. A. of a bundle. 7. 308 Positive semidefinite. 86 Oriel1latim. R. 460 Ordinal numbers. 39 Pointfinite cover. GramSchmidt. 304 Osgood's Theorem. 61. 36 integration in. 351 Nowhere zero section. 210. 84. 284 Palais. 86 of a vector space. 85 of a manif old. 225 Paracompact. 95 . 439 Poincare Duality Theorem. 84. 85. 133. 372 Orthonormal. 301 Positive element of norm I . 98. 3 1 3 Partial derivatives. 461 isomorphism. 407 Orde. 17!=l Onedimensional sphere: 6 Oneparameter group of diff eomorphisms.
17'1 Special linear group. 54 Smooth. 22� of a map.. 231 Scalar density. J7. 134. G. 220 homotopic... 374 Rinow. 134. 104 Semidefinite. 263. 264 def ormation. 462 Selfadjoint linear transf ormation. 98 Rectifiable. 263 with compact supports. 100 of norm 1. 451 RadiaJ f unction. 9 6 Segment. 50 Regular point. 3 J J usual. 8. 231. 103 MayerVietoris. 3 1 2 Smoothly contractible. 1 90 Relative scalar. 457 RicmaJ1Jl integrable. 134. 40 Related vector fields. 7 3 zero. 277 manif old. 354 Schwarz inequality. 3 1 2 Right invariant nform. 435 Rado. 7 Sphele bundle. 3i 3 Sard's Theorem.connected. 58 Spanned locally. 342 Roman surf ac{'. 424 f compact supports . positive. 285 Simply. 95 Sequence exact. T. 28:. H. 26 Roscnlichl. 283 Riemannian metric. 29 piecewise. 6(J Shuffle permutation. associated. 283 integral. 104 Separate points and dosed sets. \fl. 433 Shrinking Lemma. 288 Reparameterizatiol1. 422 of vector bundles. 45!=l Section of a vector bundle. initial. 40. 294 Solid angle. 308. M. 279 Revolution. 294 Scalar. 22i Simplex of a triangulation. 4 7'2 Rotation group. 427 singular. 459 value. 287 Lie group. 277 isotopic. 6 J Special orthogonal group . 62 . 5 I Shrinking Lemma. 244. 3iB Slice. 20J. 246 simplex. 62 Sphere. 419. 43J or of a pair.488 Quaternions.. 400 Right translatioll.'. sum. 248 Retraction. 42. 133 Schwarz. 40 space. 472 Rank of a f orm. 362 Second countable.. 290 Space filling curve. 268 de Rham's Theorem. 2 3 1 tensor. 194 Slice maps. 303. 382 Singular cube. surface of. 59 Rdlnement of a cover.. 321 de Rham. 285 Skewsymmetric. 342 de Rham cohomology v{'ctor spaces. relative. 28 homotopy.
349 Symmetric bilinear f orm. 7. Standard nsimplex. 134. 46 Topologically isomorphic Lie groups. 9 Total space.\'hitney sum. H. 461 Trivial vector bundle. 285. 98.Inrie. 63. 42. 116 Thorn class. 101 Support. 253. 426 singular cube. 3 7� Subbundle. 76 of �". 120 covariant. :2 Su ccessor ordinal. 72 Tubular neighborhood. 120 covariant. 303 Triangulation. 345 Twoholed torus. 8 Tangent bundle. 426 simpJex of. 352 Stonee ech compactification. 4�) immersed. 261. 122 Tensor product. 106 '. 4 7 open. 374 Triangle inequ ality. 239 of revolution. 4 . 123 contravariant. 14. 33. 1 2 1 . 1 1 3 mixed.. 301 System of linear diff erential equations.. 231 odd relative. 66 Vick. 17. 442 Thorn lsomorphism Theorem. 354 integral. 64 outward pointing. 374 right. 1 4 immersion. 26 Sternberg. 288 Tensor field classical definition of. 171 acompact. Whitney. 221 Steiner's surf ace. 147 Surface. S. 388 Torus. 50 Subgroup Lie. 203 Whitney. 459 489 Tensor contravariant. 77 Tangent space of �Ii. 468 Structure constants. 106 Stokes' Theorem. 8 nholed. 49 closed. 7 area. 321 Sylvester's Law of Inertia. 7. 101 . 98 of a manif old.. 25 Transitivity. 384 Submanif old. 460 Translation left. 464 Sum of vector bundles. 7 1 Totally disconnected. 8. 260 to a curve. 42 7 Trichotomy. 64 Tangent vector inward pointing. 1 1 3 even relative. J W. 456 Topological group. 134. 198 Subcover. 49 Coo. 246 Starshaped. 3 7 1 imbedding. 396 Subalgebra of a Lie algebra. 373 oneparameter. 3 Wedge product.
.
The bold face supplied by Monotype� even the "semibold": is obtru sively extended. which was created in three essential sizes: 9. results from the linear scaling. The bold letters for mathematics come directly from the bold fonts of the Times families. were made for various pu rposes. painted by the author in his spare moments. including additional kerns and alterations of set widths. bold type was useful in special circumstancesmainly for indicating defined terms. The Adobe Mathematical Pi 2 font was used for the ordinary sized German Fraktur letters. and special parentheses and other punctuation symbols were also required. the tall initial letters beginning each chapter were designed specially. The proportions and weights for these were then used for the three sizes of the symbol font and the other mathematics fonts. . as well as the fact that the upper case Baskerville letters are of somewhat heavier weight than the lower case. A thicker set ofnumerals was constructed for use with the upper case lettering in chapter headings and statements of theorems. so a nonextended version was created. are loosely based on Samuel Tavlor Coleridge's poem TI. T together with Berthold H orn's The mathematics fonts are a variation of the Math TlIne fonts: now based on the Monotype Times New Roman family. The elegant swashes of the italic ). including bold symbols. Numerous other modifications of this sort. On the other hand.in. The text font is I I point Monotype Baskervillethough the emdash has been modified together with its italic. in superscripts.e Rime qfthe Ancient Ma.. special gg and gf ligatures were also required. and new or modified fonts were created using FontographeI. in 16 point type. so a special gy ligature was added. 7 and 5� point. . Knuth's 'lEX typesetting system. as well as for the extension font and its bold version. DVIPSONE PostScript driver. The italic fonts of these three families were used as the basis for creating the three separate " math italic" fontsfor use at ordinary size. while blackboard bold letters were made by hollowing out these bold letters. and in second order superscripts. together with the Monotype Times N R Seven and Times Small Text familiespresumably these three families are based on the original designs for Times New Roman. Although a Baskerville bold face is unhistorical. The covers. The figures were produced with Adobe Illustrator. since simple scaling would have made them u npleasantly heavy. and additional special symbols. script letters. with su itably modified versions used for su perscripts.hese books were typeset using Donald E. and J cause problems in words like topologl' and apology. The somewhat bold appearance of chapter headings..
6 1 . 43: Replace the last line and displayed equation with the f ollowing: Since rank f =k in a neighborhood of p. v(p )) for some v(p) E JR . then each dashed vector. the lower rectangle in the matrix ( a(Ui ) o f) ax} = pg. . Xk . + M is a Coo manif old. restate part (c) as f ollows: (c) This is f alse if J : (f) If M is a connected manifold. homeomorphically onto 2 T(M. A(X" . and we could continuously pick w (p) E criterion that w(p) should make a positive angle with v(p). JR is replaced with J : M. If there were a way to map + N for a disconnected manifold N. . 103. liP) is called the boundary of 4> : V + pg. then A is Coo. there is a proper map J : M + JR. corresponding to a M x JR2 . . then is Coo.� �\ B: :B A. . 70: Replace the last two lines of page 70 and the first two lines of page 7 1 with the f ollowing: correspond to ( p. Xdp)). Problem 29(d). pg. theorem of topology). line 3: change pg. 14: relabel the lower left part o f the central figure as di(x. add: M is orientable. . . . Equivalently. in the sense that A(X) . Xk)(p) If A pg. . . . . A is also.i). Problem 32: For clarity.y) < 1 to di (x. JR2 . 3. 22: Replace the top left figure with IHI" such that M and is denoted by 4>(x) = aM. '� . . the function J can be made Coo if (g) The same is true if M has at most countably many components. pg. �3 . Problem 30: Change part (f) and add part (g): pg. 78: the third display should read: 0 = e(O) pg. = pg. . Add the hypothesis that = e(fh) = J(p)e(h) + h(p)e(f) = 0 + e(f). x E aM if and only if there is a neighborhood of x and O. . 60.CORRECTIONS FOR VOLUME I pg. Xk) is a Coo f unction for all COO vector fields XI . fibre by fibre. and aM itself is always a manifold JR" (but only one homeomorphic V pg. . A. 1 1 7: After the next to last display. . A (p) (X. 1 1 8: Add the following to the statement of the theorem : If . 1 9 : replace the nexttolast paragraph with the f ollowing: to The set of points in a manifoldwithboundary that do not have a neighborhood homeomorphic to a homeomorphism (without boundary). ( p). M. . If M is actually a manif old.A is Coo. then aM = 0.y) ::: 1. by using the up would pg..
restricting densities to those of weight 1). what are here called "odd scalar densities" should probably simply be called uscalar densities". 233. . . pg. replace parts (b) and (c) with: I (b) Determine the . and then replace Problem 19. O. 198. For X Y E Mp and . p. pg. when the transformation rule involves (det (7) A)W. 2 C V. the condition V. End with End . p N". 237.pg. Problem (d) Let f: M pg. 143. 177. V375" refers to pg. Though there is considerable variation in terminology.. what are called "even scalar densities" might best be called "signed scalar densities". Ell /'. the lower right entry should be "Iforms on pg. . is . h component of VI x ··· X VnI in terms of the (n .] is integrable. A follows from the fact that the function Ail . 226. In Problem + . In part (g) of H is a connected Lie subgroup. 0<2 be two maps on some open interval 1 such that 0<1(1). Aut should be replaced with Aut . . pg. For consistency with standard usage. In Problem 20. pg. show that . In Problem 26. we should be considering the h of part (a). . 375 of V olume V pg. omitting the modifier «signed" when it involves I det A l w. 133.p N.. add the hypothesis that pg. and suppose that f 5. part (c) should read: . ). Problem 16 (b) should read: "For any Lie group G.. show that n Vj # 0 should be CJ n V. . . 408410. part (d) should begin: = for some to E I. 1 3 1 .1) x (n . " . we should probably speak of the bundle of "signed tensor densities of type and weight w " (though sometimes the term relative tensor is used instead. 4 1 1 . 134. must also assume that each /'. 292. 1.+I pp. 133. The hypothesis o f Theorem Let pg. # 0. . for 1R3 . scalar densities In part (c) of Problem 10. pg. Problem 9: Let F be a covariant functor from V.i.1) submatrices of the matrix In particular. In the comutative diagram. V. pg. 408. 1 1 9 : Add the f ollowing at the end of the proof: Smoothness of pg. 3 should b e changed s o that i t reads: x and E V and let 0<1 . . 0<2(/) 0<1 (10) = 0<2(10 ) 0<1 ' (1 ) = f(O<I(I)) i = And the first sentence of the proof should be deleted. . The display in Problem 21 . The reference "pg. we 17. Extending the changed terminology f rom pg.A(ajaxl" . . not the h of part (b)! Thus conclude that the bundle of signed (not the scalar densities) is not trivial if M is not orientable. .. ajaXI.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.