Scribd Logo
Upload

Welcome to Scribd!

  • 11 views

    (Frank W. Warner) Foundations of Differentiable Ma (BookFi) PDF

    Uploaded by

    Junaid Aftab

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content

    You might also like

    (Frank W. Warner) Foundations of Differentiable Ma (BookFi) PDF

    Uploaded by

    Junaid Aftab
    11 views139 pages

    Original Title

    [Frank_W._Warner]_Foundations_of_Differentiable_Ma(BookFi).pdf

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    Download as pdf
    11 views139 pages

    (Frank W. Warner) Foundations of Differentiable Ma (BookFi) PDF

    Uploaded by

    Junaid Aftab

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    Download as pdf
    of 139
    FOUNDATIONS of DIFFERENTIABLE MANIFOLDS and LIEGROUPS FRANK W. WARNER University of Peonsyivania ‘Scot, Foresman and Company Glemvicnsllincis London. Editors Preface is textbook fills a long-standing gap. The beginning graduate student Tei ee Sanh Mae a or cndguniyenees ones sl a aed aa steely a den so ope of cr a ietrltngege Noor pon end Seine nin ay aoe he Al re ste ne en une a sean oso sete Na oi ite, teehee Te Sy eae ne Preface ‘This book provides the necessary foundation for students interested in any of the diverse areas of mathematics which require the notion of a differeatiable ‘manifold. Ttis designed as a beginning graduate-level textbook and presumes ‘good undergraduate training in algebra and analysis plus some knowledge of point set topology, covering spaces, and the fundamental group. It is also intended for use as a reference book since it includes a numberof items ‘which are difficult to ferret out ofthe literature, in particular, thecompleteand self-contained proofs of the fundamental theorems of Hodge and de Rham. ‘The core material it contained in Chapters 1, 2, and 4, This includes differentiable manifolds, tangent vectors, submanifolds, implicit function theorems, vector fields, distributions and the Frobenius theorem, differential forms, integration, Stokes’ theorem, and de Rham cohomology. Chapter 3 treats the foundations of Lie group theory, including the relationship between Lic groups and their Lie algebras, the exponential ‘map, the adjoint representation, and the closed subgroup theorem. Many xamples are given, and many properties of the classical groups are derived. ‘The chapter concludes with a discussion of homogeneous manifolds. ‘The standard reference for Lie group theory for over two decades has been Chovalley's Theory of Lie Groups, to which T am greatly indebted. For the de Rham theorem, which is the main goal of Chapter 5, axiomatic sheaf cohomology theory is developed. In addition to a proof of the strong form of the de Rham theorem—the de Rham homomorphism given by integration isa ring isomorphism from the de Rham cohomology ring to the B means that fis of the set into the set B. When describing a mapping by describing its effect on individual ‘emeots, we use the special arrow +7. thus “the mapping mi+(m) of {nto B” means that fis a mapping of the et A into the set B taking the element ‘mof A into the element f(m) of B. IFU < A, then | U denotes the restriction Of to U, and f(U) = (EB: f(@) = for wme se U). If Ce B, thea PAC) = ee A:f@)EC). A mapping fis one-to-one (so denoted 1:1), {rr injecrce, if whenever a and b are distinct elements of A, then f(a) x f(0)- ‘A mapping f's onto, oF surjective, if f(A) = B. Iff, A> B and g: C—» D, then the composition g «fis the map Ef fIBno+D defined by g+f(@) = s(/(@) for every aef-4B MC). For notational convenience, we sball not exclude the case In which f(BOC) = 2. ‘Thats, given any two mappings fandg, we shall consider their composition ff a being debaed, with the understanding thatthe domain of g » f may ‘well be the empty set. “The cartesion product A x B of two sets A and B is the set ofall pairs (a,8) of points ge and DEB. If f:A—C and g:B—~ D, thea the cartesian product f X g of the maps f and gis the map (a, B+ (f(a), (0)) of Ax B into Cx D. “We shall denote the idenrity map on any set by “id.” A diagram of maps such a8 A L \ : B——e is called commutative if go f= h. ‘We shall always use the term fincrion to mean a mapping into the real umber 4 Manifolds Let d 2 1 be an integer, and let Rem (0: a (61... 5.44) where the a, are real numbers. ‘Then R¢ isthe dimensional Euclidean space. In the case d = 1, we denote the real line B® simply by R. The origin (,... , 0) in Euclidean space of any dimension wl Be denoted 0, The notations ja] and (at dnote as le ine ot 0. 1.3 Definitions A locally Euclidean space M of dimension dis & Hausdort topological space M for which each point has a neighborhood Homeomorphec te un open subset of Euclidean space P4. If pis homen- torphitm of a connected open set Uc M onto an open subset of Fé, is called a coordinate map, the functions x, = r, ° 9 are called the coordinate inctons, avd the pair (0,9) (sometimes denoted by (U, ~) {alled a coordinate system. “A coordinate system (U, 9) is called a exbic scondinate system if 9(U) san open cube about the origin in RY. If me U tnd p(m) — 0, then the coordinate system is said to be centered atm. 14 Definitions A differentiable sirucre F of class C* (1 Sk S &) fon a locally Euclidean space M is a collection of coordinate systems {(U,, 90): A} satstying the following three properties: © UW=M. ) is C¥for alla, pe A. (©) The collection F is maximal with respect to (b); that (Wig) is a coordinate system such that ge grt and 9,°¢7 are CF for all 24, then (U,9) «F. 6 Manifolds Mt F,= (Ws.p):%64) is any collection of coordinate systems sntiafying propertce (a) and (b), then there is a unique differentiable struc- i Namely, let F = (U9): 9° 9e* and 9+ 9 tre C* for all EF). ‘Then F contains F, clearly satistis (a), and it is easly checked that F satisfies (b). Now ¥ is maximal by construction, and so Fis differentiable structure containing F. Clearly ¥ is the unique such structure, ‘We mention two other fundamental types of differentiable structures on locally Euctidean spacey, types that we ahall not treat in this text, namely, the structure of class C= and the complex analytic structure. “For differentiable structure of class C*, one requires that the compositions in (©) are locally given by convergent power series, For a complex analytic structure on 8 2d-dimensional locally Euclidean space, one requires that the ‘coordinate systems have range in complex dspace C* and overlap holo- morphically. ‘A d-dimensional differentiable manifold of class C* (imilatty C* or complex analytic) is a pair (M,F) consisting of a ¢-dimensional, second countable, locally Euclidean space M together with a differentiable structure F of class C*, We shall usually denote the differentiable manifold (M,#) simply by Af, with the understanding that when we speak of the “differentiable manifold M” we are considering the locally Euclidean space M with some siven differentiable structure. Our attention will be restricted solely to the case of class C°, 40 by difereniable we will always mean diferentiable ‘of class C°. We also use the terminology smooth to indicate diferentiablity ‘of class C°, We often refer to differentiable manifolds simply as manifolds, with differentiabtity of class C always implicitly assumed. A manifold can be viewed as a triple consisting of an underlying point set, a second countable locally Euclidean topology for this set, and a diferentible structure. If is a set, by a manifold structure on X we shall mean a choice of both a second countable locally Euclidean topology for X and a difer- entiable structure. Even though we shall restrict our attention to the C= case, many of our theorems do, however, have C* versions for k < co, which are ementally ‘20 more complicated than the ones we shall obtain. They simply require that one keep track of degrees of difereatiabilty, for differentiating a C* function may only yield a function of class C™* if 1 < k < oo. Unleat we indicate otherwise, we shall aluaye tee M and N to denote differentiable manifolds, and M* will indicate thet M is © manifold of dimension d. 1.5 Examples (@) The standard diferentiable structure on Euclidean space Ris obtained by taking F to be the maximal collection (with respect to 1.4(b)) containing (F,), where /: RA —~ Ris the identity map. o © © o © Difjerenniaoie stanyoms + Let ¥ be a finite dimensional real vector space. Then ¥ bas a natural ‘Maifold srvture. Indeed, if {e) isa basis of V, then the elements of the dual basis {r,) are the coordinate functions of a global coordinate Biicm on ¥. Such a global coordinate system uniquely determines & mereotiabls structure F on V. This differentiable structre is inde- fendent ofthe choice of basis, since differen bases give C> overlapping Penddinate syotems. ln fac, ihe change of coordinates in given simply by a constant non-singular matrix. Complex > in a real 2rdimensional vector space, and £0, by Example (b), has a natural structure as a 2n-dimensional real Py told” Ite) is the canonical complex basis in which e, is the ‘tuple consisting of zeros exept for a 1 in theft spot, then fe JTie,..-+V=Te} is a real basis for C*, and its dual basis is the canonical global coordinate system on C* “The dsphere is the set 7 Sta ent: Zat = 0. Let m= p.2+,0,1) ands (0,-.-,0,—1). Then the standard Uiterentiable sructire on S*is obtained by taking $F to be the maximal Salection containing (S* —n, p) and (S*— sp), where pa and Py re stereographic projections from m ands reepectvely. ‘Am open subset U of a diferentiable manifold (MF) is itself & ‘iferetiable manifold with differentiable structure Fy = (Up OU, G| Ue): Urs EF 10: Unless specified otherwise, open subsets of differentiable manifolds will siways be piven this natural differentiable structure, “The general linear group Gllx.P) isthe set of all n x m non-singular Teal Catrices, If we identify in the obvious way the points of IR™* with nxn real matrices, then the determinant becomes & continuous func- fon on R™._ Gi(n,R) receives & manifold structure as the open subset ‘of R™ where the determinant function does not vanish. Product manifols, Let (My F) and (Ms Hy) be diferentiable Fronifolds of dimensions dy\and dy respectively. ‘Then M; x Ms Tecomes a differentiable manifold of dimension dy + dy with diferen- table structure & the maximal collection containing Ug Mp Be v9 Ue KV Re x RAY: Us 90) Fa» Vo my EF: 1.6 Definitions Let Uc M be open. We say that f:U-—+R isa (C* function on U (denoted fe C™(U)) if f= 9-* is C™ for each coordinate map g on M. A continuous map y: MN is said to be diferenlable of class C™ (denoted y © Cm(M,N) or simply ye C")ifg © pine C> function yr Gonain of for aC futon dead on open st in W | inuous map » is C* if and if peyertis ean eet ors ee oy iteee ata ‘Clearly the composition of two differentiable mape is again differentiable. (Observe that « mapping y: AM — Nis C* if and only if for each me M there ‘exists an open neighborhood U of m such that y| U is C=, THE SECOND AXIOM OF COUNTABILITY The second axiom of countabilty has many co sanifo Aton then, manflds ae more mens, tos parcompact ar compactness implies the existence of partitions of unity, an ettremely useful tool for piecing together global functions and structures out of local ones, and conversely for representing global structures as locally Anite sums of local ones. After giving the necessary definitions, we shall give « simple direct proof of paracompactness for manifolds, and shall then derive the existence of partitions of unity. It is evident that manifolds are regular topological spaces and their normality follows easily from this and the paracompactness. We shal leave the proof that manifolds are normal as an enercie. For the fact that manifolds are metrizable, see (13). 1.7 Definitions A collection (U,) of subsets of M is a cover of WED, ctr cata ese Cae collection of the U, which stl covers is called a subcover. A refinement {¥,) of the cover (U,) isa cover such that for each f there is an « such that VE U,. A collection (4,) of subsets of M is locally finite if whenever me hr exe mighborhood Wa of m sth tht yt for a an open only aoe anament Pe nomPaet tery open cover 18 Defnion A partion of wy on M iso ealction (9: C® functions on M such that vey ‘ (a) The calles 61) of ©) Zep) forall pem, and g,(p) 2 0 for all pe M and ve, ‘A partition of unity {9,: £1) is subordinate to the cover (U,: « € A) if. cai (6 10 the coves (U2) if elas St an eta, We attr covet (Uf 101) wth te ie tse pain of ny ‘supp 9% © U, for each ie J. paniion of nity Second Axiom of Countabllity 9 1.9 Lemma Let X be a topological space which is locally compact (Gach point has ot least one compact neighborhood), Hausdorff, and second countable (manifolds, for example). Then X is paracompact. In fact, each ‘pen cover has a counable, locally nite refinement consisting of open set with compact closures. poor We prove fiat that there exists a sequence ( of open sets such that 13.03 Gris compact, Be Gas xebo, be a countable basis of the ot x Consnting of open sets with compact clonures Sich a basis can be Cbuined by sarng with any countable basis and selecting the sub- Callection Zomisng of bese sta with compact closure. The fact that X is Hausdorff and locally compact implies that this subcolletion is itself a basis. Now, let G, = Uy. Suppose that Ge UU UU. ‘Then let /yg be the smallest postive integer greater than, sch that Getu. ‘and define “ Gare v Us ‘This defines inductively a sequence (G,} satisfying (1). o. o..} at Gun J 10 Manifolds Let {U,: «€ A) be an arbitrary open cover. The vet G,—G.. is ‘compact and contained in the open vet Gxt — Gra. For each iz 3 choose « finite tubcover of the open cover (Us Gun — Gea): @€4) of Z,— Ges, and choose « finite subcover of the open cover {U,.A Gy: « € A} of the compact set G,. This collection of open sets is asl seen to be a countable, locally nite refinement of the open cover (Uy), and consists of open sets with compact closures, 1.10 Lemma There exists a non-negative C* function 9 on R* which ‘equals \ om the closed cube C(I) and zero on the complement of the open cube q). root We need only let g be the product o ea herd (herd, where h it a non-negative C® function on the real line which is 1 om [21,1] and zero outside of (—2,2). To construct such an fh, we start with the function few" 8 >0 ® wonl iz0 which is non-negative, C*, and postive for t > 0, Then the function 3 oO _ ® OTSA) is non-negative, C”, and takes the value 1 for ¢ 2 1 and the value zero for ¢:$0. We obtain the desired function h by setting o MO = £0 + Dg — 0. 1.11. Theorem (Existence of Partitions of Unity) Let M be a difer- ‘entiable manifold and (U,: a. A) an open cover of M. Then there exists @ countable partition of unity (9: i= 1,2,3,...} subordinate to the cover {U,) mith supp 9, compact for each 1. Ufone does not requre compact supports, then there is a partition of unity {9,) subordinate to the cover {U,) (that ts, supp 9. © U.) with at most countably many of the not identically zero. moor Let the sequence {G,} cover Mas in 1.9(I), and set Gy= Fae pe M, let he the lergeet integer wich that pe M — @, Choose fan g, such that p € U,, , and let (Vx) be & coordinate system centered at p such that V-< UL, © Gyae— G,,) and such that r(V) contains the closed cube C@). Define alert ony ® % A a Tangent Vectors and Diferentiais 11 ‘where @ isthe function 1.10(1). Thea y, is a C* function on M which East vale fon rm open niger 1, oa a opt support Iying in V < U,, Giax— G,)- For each i > 1, choose manos on Gee Oey, tape cover G; — G,. Order the corresponding y, functions in « sequence Wim 1,23... The supports of the y, form a locally finite family OF Subsets of AF. Thus the funetion @ vate is a well-defined C* function on M, and moreover (p) > 0 for each eM. For each im 1, 2,3,... define ® a ‘Then the functions (94: fm 1,2,3,...} form a partition of unity subordinate to the cover {U,) with supp 9, compact for each f. If we Tet 9, be identically zero if no’ y, has support in U, , and otherwise let ‘9, be the sum of the 9, with support in U, then {9,) is a partition of Unity subordinate to the cover {U,) with at most countably many of the f, not identically zero, To sce that the support of 9, lies in U,, 0D- serve that if of isa lofally finite family of closed sets, then UJ A = (J A (Observe, however, that the support of is not necessarily compact. Corollary Lat G be open in M, and let A be closed in M, with A&G. Then there exists @.C® function 9: M— R. such that (@) 0< ofp) <1 forall peM. & e@alyped. © mppesG. + proor There is a partition of unity (9,y} subordinate to the cover {G,M— A) of Mf with supp 9 G and suppyc M— A. Then @ is the desired function. ‘TANGENT VECTORS AND DIFFERENTIALS 1.12 A vector o with components 6, 5... 24 at a point p in Buelidesn apace Ré can be thought of as an operator on differentiable functions, Specitcally, iff diferentiable on a neighborhood ofp, then v assigns to {ube real number o(/) which isthe directional derivative of fim the direction atp. Thats, o ange | + tazl 12 Manifolds ‘This operation of the vector » on difeentnble functions satisfies two important proper, ° HU + 4g) = 07) + 4, U8) =F) + PD whenever f and g are ifrentable near p, and is real umber. The rt property sys that » acts incaly on incon, and the cond says that vis aderication. This motivates our definition of tangent vectors on manifolds. ‘They willbe Geena! derivatives, that i, inear Gevauons on fancubns. The operation of taking derivatives depends only on local properties of fncons, propertien in arbitrarily email neighborhoods of the po which the derivative being taken. Ta order fo expres most conveaicatly this dependence ofthe derivative onthe local nature of functions, we intro duce the notion of germs of functions. 1.13 Definitions Let me M. Functions f and g defined on open sets containing m are said to have the same germ at m if they agree on some seighborhood of m. This introduces an equivalence relation on the C* functions defined on neighborhoods of m, two functions being equivalent if and only if they have the same germ. The equivalence classes are called _germs, and we denote the set of germs at m by A. Iffis 2 C* function on ‘neighborhood of m, thea f will denote its germ. ‘The operations of addition, scalar multiplication, and multiplication of functions induce on F, the structure ofan algebra over R.A germ f has a well-defined value fm) atm, namely, the value at m of aay representative of the germ. Let F, ¢ Fy be the set of germs which vanish at m. Then Fy is an idea! in Fy, and we let Fat denote its kth power. Fats the ideal of F., consisting ofa nite linear ‘combinttions of k-fold products of elements of F. These form a descending sequence of ideals F, > Fy > Fy > Fyt> + 1.14 DeBnition _ A tangent vector © atthe point m & M ie linear deive- tion of the algebra Fl, That is, forall, ge F, and 2 R, (©) 6 + 26) = 010) + de. ©) wt a) = flo) + aime. My denotes the set of tangent vectors to M at m and is called the sangent space to M at m. Observe that if we define (0 + w)(f) and (2o)(0) by w © + wXf) = off) + wf) Coy) = ao) whenever 0, we My and A R, then » + w and 20 again are tangent vectors at m. So in this way M, becomes a real vector space. The fundamental property of the vector space Mr , which we shall establish in 1.17, is that its dimension equals the dimension of M. This definition of tangent vector Tangent Vectors and Diferentials 13 is not suitable in the C¥ ease for 1 < k < oo. (We will discuss the C* case further in 1.21.) We give this definition of tangent vector for several reasons. ‘One reason is that itis intrinsic; that is, it does not depend on coordinate systems. Another reason is that it generalizes naturally to higher order tangent vectors, ax we shall seein 1.26. 1.15 Lfeis the germ of « function with the constant value ¢ on 8 neighbor- hood of m, and if» isa tangent vector at m, then 0(c) = 0, for 2(@) = en(t), (0) = (0 +1) = 10(1) + 10(H) = 2e(1). 116 Lemma Matsnaturally isomorphic with (Fq/Faf)*. (The eymbol * denotes dual vector space.) and roor If 0 € My, then is « linear function on Fa vanishing on Ft because of the derivation property. Conversely, if Ce (Fa/Fat)*, We define a tangent vector o-atm by setting vA) = “if — f(m))) forte Fa. (Here fam) denotes the germ of the function with the constant value fim, and {) is used to denote cosets in Fa Fat.) Linearity of von Fa js clear. Tt is derivation since ate) = ite — s(m)e(a) = A(t = He) (& = ste) + Heme — at) + (f— fm))a(—) = ACE — f=))(& — 6m) + FO"4C@ — eC) + Bome(it— fom) = fombee(s) + glnA). “Thus we obtain mappings of Mz into (FajFat)*, and vice vera. It is easily checked that these are Tnverses OF each other and thus are isomorphisms. 447 Theorem dim Fa/Fat) = dim M. ‘The proof is based on the following calculus emma (31) Lemma If g is of class C* (k 2 2) on a convex open set U about pin Rt, then for each 9 € U, - 5281 (4a) - : () ala) = e(0) + 35% | (4) — rP)) +E CMW — Myla) - raf oe at Gr, rs et te-o In particular, if g€C*, then the second summation in (1) determines an clement of F,? since the integral as a function of 9 is of class C*. ‘PROOF OF 1.17 Let (U,9) be s coordinate system about m with co- ordinate functions %,,...%, (d= dim M), Let feF,. Apply (1) to f+ ¢", and compose with @ to obtain FBO | = nln) +E (= noo Mey = nA on & neighborhood of m, where he C*. Thus 1 Z9L 29) a — x¢e)) mod Fe a) Hence (x — x(a}: f= 1... 5d FalFat. Consequently dim FjFa' < 4. We dim tat thee clement a Hocasl independent. For nuppote that data simyere. Now, (x1 — xk) 8 = 5 ales — rom). ns — (9(m))) © Fim - (Sara n(em))) = 0 +d, which implies that the a, must all be zero, Corollary dim My = dim M, 1.18 In practice we will treat tangent vectors 23 operating on functions rather than on their germs. If fis a differentiable function defined on a neighborhood of m, and ve Ma. we define © = (0. Thus o(/) = op) whenever fund g age on t neighborhood of m, and y WAS + ig) olf) + dnl) GER), OF 8) = Smeg) + sem), where f+ Ag and f-g are defined on the intersection of the domaine of definition of fand 5. @ 1.19 Definition Let (U,9) be a coordinate system with coordinate functions x,,....x¢, and let meU. For cach /€(1,...,d), we define Tangent Vectors and Diferentials 15 a tangent vector (@/2x a © Mm by setting a ayer) ” Goer Le or fon f which is C* on a neighborhood of m. We interpret (1) {Sg deeal State of fat min tes cordont decon. We io use the notation ® alr GO 1.20 Remarks on 149 (a) Clearly ((@/2x)]_)(/) depends only on the germ of fat m, and (0) and © Gar Ne leesutadeds to lbeDls iva tangent vector atm. Moreover, {@)Px)la: i= 1,....d} isa basis of M.,. Indeed, itis the basis of Ma ual to the basis (Gx, — x(a): 1 = 1,.7- vd} of Fa/Fa since a im) = by. ay, | 1 2A) = (b) Ive Ma, then . 8 em Zonda. | Simply check that Doth es give the sume resus when applied to the functions (x; — x,(m)). 6 Suppose that (U9) and (Vay) are coordinate systems bout m, © Sapper rinse unetion sso. and Joyo. s)ateapeatvely. Then {Lisowsfrom rman (0 that 2| -3%| 2 Bye FLO yy Im Bm J tat (2) depends on g and not only on xy. Tn particular, Rae eer ecass nat nena flow that 79x, equal ay, (@ If we apply Definition 1.19 to the canonical coordinate system riss+++ra on R¢, then the tangent vectors which we obtain are none ‘other than the ordinary partial derivative operators (2/2). 1.21 Our proof of the finite dimensionality of F/Fa? certainly fails in the CC case for k < co since the remainder term in the lemma of 1.17 will not tbe m sum of products of C* functions, and the lemma doesa't even make sense in the C! case. In fact, it turns out (see [21]) that F/Fq! is always infinite dimensional in the C¥ ease for 1 < k < co. There are various waye to define tangent vectors in the C* ease in order that dim My = dim M (All of which work in the C* case, too). One way isto define a tangent vector atm as x mapping which assigns to each function (defined and differentiable ‘of class C*on a neighborhood of m) a real number v(f) such that if (U.g) coordinate system on # neighborhood of m, then there exists a list of real numbers +44) (depending on ¢) such that $0289 Ore loins’ ‘Then the apace My of tangent vectors again turns out to be finite dimensional, with a basis ((2/8%)a} 41.22 The Differential Let y: MN be C*, and let meM. The differential of y at m is the linear map ® dy: Ma Nias defined as follows. If v€ M... then dy) is to be a tangent vector at yim), s0 we describe how it operates on functions. Let g be a C* function on a neighborhood of y(m). Define dy()(g) by setting @ 4 (OVg) = 28 + v). It is easily checked that dy is a linear map of M,. into Nya Strictly speaking, this map should be denoted dy | M, or simply dys. However, we ‘omit the subscript m when there is no possibility of confusion. The map y is called non-singular at m is non-singular, that i, if the kernel of (1) av, consists of 0 alone, The dua! map Nero o 892 Nay + ME, is defined as usual by requiring that © 4y(a(0) = o(dy(o)) whenever o and ne M,,. In the special cate of a C® function Se M—R, if 06M and f(m) = ry, then 4 o aoy— NF} Tangent Vectors and Differentials 17 In this case, we usually take df to mean the element of Mz, defined by © FOr: ‘That is, we identify df with 8/(«), where @ isthe basis of the 1-dimensional Space 5 dul to (al), Paroular wage wil be clear fom the context 1.23 Remarks on 1.22 (@) Let (U, x1,.-.49 and (V, 71»... 90) be coordinate systems about rm and v(m) respectively, Then it follows from 1.22(2) and 1.20(b) that (a £axew)| 2 ole |) 3 Lay, Las 1» y)/@x;} is called the Jacobian of the map y (with 1 coordinate system). For maps between Euclidean will always be taken with respect to the canonical coordinate systems. () If (Uyx45-+-5 4) 16 8 coordinate system on M, and mc U, then {dl isthe basis of ME dual to {@/2xq)- If: M+R is a C™ function, then (©) Chain Rule. Let y: M— Wand g: N—X be C* maps. Then 49° Vn deyims dV, ‘or simply d(o « y) = de ody. It isa useful exercise to check the form that this equation takes when the maps are expressed in terms of the ‘trices obtained by choosing coordinate systems. (@ Thy: M— Nand fi NR are C®, then Sy(dfins) = AS Was fOr Sydfimd(o) = df (dy(o)) = dCf* vue) whenever 2 € Ma, (ACH mapping o: (2,8) M is called a smooth eurce in M. Let ‘ea, 8). Then the tangent vector tothe curve vat ¢is the vector of) eM We shall denote the tangent vector to o att by 4). 18 Manifolds ‘Now, if » is any element of M, , then v is the tangent vector smooth curve in M. For ove cat simply choose » ordi sytem (0,9), centered at m, for which ven . Then» isthe tangent yor #010 the CUE 14 469420040) One should obvete tnt may cures ean have We sume ange eso, and that vo tmooth Sever td rie for ohn eth Nid ve th same tangent vstor tad oly 40] — Aen) dr dr |, fort funtions f which ae C* on ghbthood of m Ifo happens to be a curve in the Euclidean space R*, then a a . r leco ales It we deny ie angent veo Wi the dement a = S| peek of Rs, then we have 6) = tim t= a h . ‘Thus with this identification our notion of tangent vector coincides, in this special case, with the ic not 4a this speci ‘geometric notion of tangent to a curve 1.24 ‘Theorem Let y be 2 C* mapping of the connected manif into the manifold N. Suppose . fold M pation ‘Suppose that for each me M, dy, m0. Then y ts a rroor Let ney(M). y(n) is closed. We need only ~ We need only show that W's oe, "ror, le meio Chnow coma es pear Sad (Va unos) BbOOL mand apect thar pit) © F, Thenon 6,772? Shove mand respectively 90 oniy(2) 3292 Ge - ‘, 2y, bese ‘which implies that Wee Gm. Tangent Vectors and Digerensials 19 “Thus the functions y,« yare constant on U. This implies that y(U) =m: thence y-i(n) is open and consequently y(n) = M. in a natural way the collection of all tangent etme nares eee called the sangent bundle, We have a similar dual object called the cotangent ‘Fundle formed from the linear functionals on the tangent spaces. 125 Tangent and Cotangent Bundles Let M be a C* manifold ‘with differentiable structure F. Let TM) = U Ma o TM) = UMS. ‘There are natural projections: wT) ~M, (am if ceMe, ® at: TIM) —+M, tam if reMs. Let (U,9) €F with coordinate functions #15... ¢- Deline $: #°4U)—+ RM and §*: (wt) 4(U) + R™ by GO) = (x (C0), --- SeCWOD) das - x0) RO- (eo. -zlatind( 2), ” (2) ° andr € (w*)-(U). Note that § and §* are both one-to-one serpr pop bse Rm Tne owing ep ovine he sotrcon ‘of s topology and a differentiable structure on T(M). The construction for TCA pom timilarly. The proofs are left as exercises. (© WU sg) and (V,9) €F, then Fe Ft iB Cm ; “The collection (3-407): W open in R&, (Ug) €F} forms 2 © Tasis for a topology on 7(M) which makes TCM) into x 24- dimensional, second countable, locally Euclidean space. (© Lt F be the maximal collection, with respect to 1.4(6), containing {O).3: Ua EM). ‘Then F in a differentiable structure on T(M)- i le called respectively T(t) and T*(M) with these differentible structures are. vey tangent bundle and the cotangent bundle. It will sometimes be convenien esate pom of TO) x pais ms) where me M and n= My (snd similarly for T*(A0). @ If y: M-N is a C* map, then the differential of y defines « mapping of the tangent bundles o dy: TU) = TO), where dy(m,s) = dp,(0) whenever v€ My. It is easily checked that (4) in aC map. 1.26t Higher Order Tangent Vectors and Differentials It is useful to look at Mu, as (FA/F.!*, for this point of view allows an immediate seneralizagion to higher Order tangeot vectors. We digress for a moment to give these deBnitions, ‘Recall that F., isthe algebra of germs of functions at m. Fa < Fy is the ideal of germs vanishing at m, and F,® (k an integer > 1) i the ideal of F., consisting ofall finite linear combinations of K-fold products of elements of Fe. : ‘The vector space Fa/F is called the space of kth order differentials at m, and we denote it by *ifq. AS before, f denotes the germ of fat m, and { will denote cosets in Fa/Fa". Let/be a diferentiable function on aneighbor- Ihood of m. We define the kth order diferential df off at m by ® ay= (t-te). ‘A kth order tangent vector at m is areal linear function on F.,vaniahing con F2?? and vanishing also on the set of germs of functions constant on & neighborhood of m. ‘The real linear space of kth order tangent veciory at ‘mill be denoted by M,2. We have a natural ieatifcation of M,* with (CM,)* since any kth order tangent vector restricted to. yields a linear function on Fa vanishing on F, and hence yields an element of ('M,) ‘and conversely an element of (M,)* uniquely determines a linear function (00 Fa vanishing on F2%, and this extends uniquely to a kth order tangent ‘vector by requiring it (9 annihilate germs of constant functions. ‘We can tie up this notion of higher order tangent vector with the usual notion of higher order derivative in Euclidean space by looking atthe forms that these tangent vectors and differentials take in a coordinate system, Let (U,¢) be a coordinate system about m with coordinate functions ‘convex open set in Euclidean space Fé, of non-negative integers. In addition to 11, we let (x = xt = elmore of ay Letf be a C* function on U. Then it follows from the lemma of 1.17, that fm seo + F ale sod hele — aCmdY, we ie {The material ofthis section will not be ued eoewbere in the book, and soit may be sipped without los of continuty, Tangent Veetors and Diferentials 21 where the A, are C® functions on U and where eyes o an ‘Hence © apm Sade xm) Thus the coletion o K@-x@y):1s bs 4] spans *M,,.. The proof that these elements are linearly independent in Tita isthe obvious generalization of the proof for the ease k = 2 which was ‘treated in 1.17. ‘Thus the collection (5) forms a basis of *M,. Consequently 144, ia finite dimensional with dimension equal to the binomial coeficient bs (4 - ‘. ‘Asthe dual space of *M,,-M,his also finite dimensional ‘Grin ve sume dimension, Since My? is identified with (¢M/,)*, and these spaces are finite dimensional, we have a canonical isomorphism of *M,, with (04..5°, under which the element of d'fe*M,,, considered as an element of (29°, satisfies © af) =). Let o | Pure) Ox he arr btm Since the derivative is linear, and since the value of ffx" at m depends only on the germ of f at m and vanishes if fis constant on a neighborhood of m or if fis an [2] + I-fold product of functions which vanish at m, then (@/2x*)| ie an [alth order tangent vector at m, It follows that ® i. is the basis of M,¥ dual to the basis (5) of #M,,.. If » is a kth order tangent vector at m, then ° where o) n= (e209) In terms of the basis (8), equation (3) becomes sistas} ay ‘As in the case of fiat order tangent vectors, we customarily think of tangent vectors as operating on the functions themselves rather than their ‘erms; indeod, we define ay 1 =o whenever fis C* on a neighborhood of m and o ina tangent vector of any order at m. oY Finally, jut as there are natural mappings of tangeat vectors and dfferen- tials associated with a diferentiable map g: MN, #0 are there linear ‘mappings Bg Mal + New as Pe: Nga My defined by a eloig) = ole + Peep) = ago 9) whenever 2 € Mat and g is a C function on a neighborhood of 9(m). tis easly checked that (14) does indeed define the mappings (13) and that the mappings d¥e and d¥e are dual definition of a first order tangent vector in this section agrees with Definition 1.14 in view of Lemma 1.16. Moreover, we have seen three interpretation of the fret order differential df of a function fy the inter- pretation (13) agrees with our original definition 1.22(1), the interpretation (© agrees with 1.22(6), and we have the additional interpretation (1). SUBMANIFOLDS, DIFFEOMORPHISMS, AND THE INVERSE FUNCTION THEOREM 1.27 Definitions Let y: M+ N be C*. (©) is an immersion if dys ix non-singular for each me M. (©). The pit (My) is a submanifold of N if yis a one-to-one immersion. © _y is an imbedding if y ix a one-to-one immersion which is also a ‘homeomorphism into; that is, y is open as a map into y(™M) with the relative topology. © ys demain i y map M onetoone oto a y+ 1.28 Remarks on 1.27 One can, for example, immerse the real line into the plane, as ilustrated in the following figure, so thatthe frst case is an immersion which isnot a submanifold, the second i a submanifold ‘whichis not an imbedding, and the third is an imbedding. ‘Submanifols, Difeomorphisms, Inverse Pinction Theorem 23 ae Immersion, Sobeaniol, but ota Submanifold but bot an Imbeding ‘Observe that if (U,9) is a coordinate system, then g: U—+ 9(U) is 2 Aitfeomorphism. "The composition of diffeomorphisms is again a diffeomorphism. Thus the relation of being difeomorphic is an equivalence relation an the collec tion of differentiable manifolds. It is quite posible fora locally Euclidean pace to powers distinct difreatable structures which ae difeomorphic. (Gee Excise 2.) In a remarkable paper, Milnor showed the existence of {beally Euclidean spaces (S"is an example) which possess non-lifeomorphic {Gafercatnble structures [19], There are aluo locally Euclidean spaccs which ppotsess no differentiable structures at all [14] Wty is a. difeomorphism, then dy, is an isomorphism since both (dp = Ney and (dy-*ody)|y ate the identity transformations. The {Srerve function theorem gives Us a local converse of this—whenever d¥m ientn somorphism, y is 4 diffeomorphism on a neighborhood of m. Before wwe recall the precise statement of the inverse function theorem, we give & Ustnition whieh will be needed in the corollaries. 1.29 Definition A set yr,...,94 of C* functions defined on some neighborhood of m in Mf ts called un independent set at m if the differentials yy... dy, form an independent set in M3, . 1.30 Inverse Function Theorem Let UCR! be open, and ler fi U-> Re be C®. If the Jacobian matrix ed ary Ie 1s non-singular at v5 U, then there exists an open set ¥ with rye = 0 ‘such that {| V maps V one-to-one onto the open set f(V), and (f| VY *is C*. ‘This is one of the results we shall assume from advanced calculus, For 1 proof, we refer the reader, for example, to [31] or [6]. 4 Manifolds Coroltary (a) Assume that y: M+ N ls C*, that me M, and that 2p Ma Naas an nomorphan. Then thre @ netghborhood U of m such that y: U'=» y(U) tsa difeomorphiam onto the open set y(U) In N. ‘rnoor Observe that dim Mf = dim N, say d. Choose coordinate systems (a) about m and (We) about yin) with y(V) © W. Let oC) =p and r(y(m)) = 4. The differential of the map rope g72| 9(¥) it ‘non-singular at p. Thus the inverse function theorem yields a ditfeo- ‘morphism a: »a{0) on a neighborhood 0 of p with OC 9(¥), ‘Then roa. 9 is the required ditfeomorphiss veo phism on the neighborhood Corollary (©) Suppose that dim M=d and that y, Independent set of functions at mg € M. Then the functions J, 4 coordinate system an a neighborhood of my PROOF Suppose that the y, are defined containing me Dang pen Regs nes om 8 Pen vEm) = (ysCm),---4Jelm) (mew). Then y is C~. Now dy isan isomorphism on (Ffiq,) since by(dr) = dir, 9) = dy, which implies that dylyiq,, takes a basis to a basis. Consequently, the Siena dyn, (whEN the dual of By) i an omorphisn So ie oven en tere inp @at'y x ieomorpi on neighborhood V < U of m, and consequent ions Yields coonate ten then ed to see Corollary (¢) Suppose that dim M = d and that y,. Smee ©. Sr tet iM = dad ie mig tft Th thy fi a 1Ye 8 an Ye form Yay with I< d, Of @ coordinate ‘Submanifolds, Difeomorphisms, Inverse Function Theorem 25 mmcor Let (U,%,.+++%) be & coordinate system about m. Then Ton. du diese sdtd) spans Mt. Choose d—I of the 2, so that (dy, dz,.,) 8 basis of Mz. Then apply Coroltary (8). Corollary (8) Let yi MeN be C®, and asume that dip: My» Nyon Srajetioe. Let Xj. 1% form a coordinate system on some Nel Teeny Then see yscecimny form port of a coordinate sytem on some neighborhood of m. pour ‘The fact that dq i2 surjective implies that the dual map Foley is injective. Thus the functions (x,0p:iml,...41) ate indspendent atm since Sy(dx,) = d(x;« y). The claim now follows from Corollary (6). Corollary (6) Suppose that y,,..-5Je it a set of C® functions on a teighborhood of m such that their dferentals span Ms,» Then a subset of the forms a coordinate system on a neighborhood of m. voor Simply choose a subset whose dilfercatials form a basis of ‘Mz, , and apply Corollary (b) Corotiary (0) Let ys MeN be C°, and assume that dy: My Noon 1 Injectve. Let %,,-..4%, form a coordinate system on @ ne of svi). Then a subset of the functions (x,y) forms a coordinate system on a Telghborhood of m. In particular, y 1s one-to-one on a neighborhood of m. yroor The fact that dy is injective implies that Bylya is surjective. This implies that (d0%y- y) = Oplds): t= 1.4K) spans MS ‘This corollary then follows from Corollary (c) 1.31. The situation often arises that one has a C™ mapping, say y, of @ ‘manifold N into a manifold M factoring through a submanifold (P,9) of M. That is, y(N) © @(P), whence there is a uniquely defined mapping vo of N into P such that @ © ys = y. The problem is: When is ys of class C=? ‘This is certainly not always the case AS an example. let NV and P both be the real line, and let Mf be the plane. Let (Ry) and (Rg) both be figure-8 submanifolds with precisely the same image sets, but with the difference that as 1 2:00, p(t) approaches the intersection along, the horizontal direction, but a(t) approaches along the vertical. Suppose also that y(0) = 9(0) = 0. ‘Then yo is nat even continuous since vs"¥(—1,1) consists of the origin plus two open sets of the form (2,+20), (—20,—2) for some « > 0. 26 Manifolds 1.32 Theorem Suppose that: N-» M is C*, that (Pq) tsa mbmanifld Of M, and that factors through (Pq), the 1, (NY & QP). Since @ 1 injective, there is @ unique mapping yy of N into P such that @ * Ya ™ y. N—*—+e os, h wes, spe (© wis C* iit continuous, (©) is continuous if i an imbedding. ‘Another important case in which vis continuous occurs when (P,9) is seg mai id ofan ivalve datibution ov Af owe tal on ‘rroor Result (b) is obvious. So assume that yy is continuous. We prove tat i is C>. I afr to show that P canbe covered by eo ‘Stdinate systems (U7) such that the map 7+ yy resticed to the open set y¥(U) is C*. Let pe P, and let (V,7) be a coordinate system on & neighborhood of ¢(p)in Mt. ‘Then by Corollary (fof 1.30 there exsta projection w of + onto a suitable subspace (obtained by setting certain of the coordinate functions equal to 0) such that the map r= ey 9 Yields a coordinate system on w neighborhood U of p. Then revel) = reyeoom| WU) = reyey| we), which is C*. 1.33 Further Remarks on Submanifolds Submanifolds (N,9,) and (ig) of Mf wil be called equlalen if there exists a difeomorphi a: Ny—> Ny such that @ = gaee. use & Gonorien N—* N ab This isan equivalence relation on the collection of all submanifolds of M. Each equivalence class € has a unique representative of the form (4) where 4 isa subset of M with a manifold structure such tht the inclusion map i: A—+ Mis a C* immersion. Namely, if (N,¢) is any representative Submanifolds, Diffeomorphisms, Inverse Function Theorem 27 of &, then the subset 4 of M must be @(N). We induce a manifold structure ond by requiring ¢: N+ A to be a difleomorphism. With this manifold srvctue, (A) ina submanifold of Mf equivalent to (Np). This i the only manifold structure on A with the property that (4,f) is equivalent to ( ‘hus this is the unique such representative of &. "The conclusion of some theorems in the following sections state that there exist unique mbmanifolds satisfying certain conditions. Uniqueness means tp to equivalence ax defined above. In particular, ifthe submanifolds of WW are viewed as subsets Ac M with manifold strictures for which the inclusion maps are C™ immersions, then uniqueness means unique rubset ‘with unique second countable locally Euclidean topology and unique ‘iferentiable structure Tn the case of a rubmanifold (4,f) of M where {is the inclusion map, we shall often drop the f and simply speak of the submanifold 4< M. Let A be subset of M. Then generally there is not a unique manifold structure on 4 such that (4.f) is a submanifold of M, if there is one at all For example, the diagrams io 1.31 illustrate two distinct manifold structures con the figure- in the plane, each of which makes the figure-8 together with the inclusion map a submanifold of R'. However, we have the following ‘ovo uniqueness theorems which involve conditions on the topology on A. (@) Let Mbe a diferentiable manifold and A a subset of M. Fix a topology ‘on A. Then there is at most one differentiable structure on A such that (Ai) tsa submanifold of M, where iis the inclusion map. (0) Again let A be a subset of M. If the relative topology, A has a differen- tiable structure such that (A,f) 13 a submanifold of M, then A has a tmigue manifold structure (hats, nique second countable locally Euclidean topology together with a walgue diferentiable structure) such that (4) tsa submanifold of M. Welcave these to the reader as onereises, Rerut (a) follows from an applica. tion of Theorem 1.32. Result (b) depends strongly on our assumption that manifolds are second countable, and for its proof you will need to use the proposition in Exercise 6 in addition to Theoren 1.32. 9) is & coordinate system on M with 1.34 Slices Suppose that ( snd that ¢ is an integer, O< Sd. Let ‘coordinate functions x,... ty ae 9(U), and let o Sm GEV: 3G) =X, im e+ Ly. “The subspace S of M together with the coordinate system @ |S fe bed forms a manifold which isa submanifold of M called a slice of the coordinate system (0,9). 28 Menfolés 1.35 Proposition Let y: Mt» N¢ be an immersion. and let m¢ M. ‘Then there exists a euble-cenered coordinate system (V,¢) about yon) and a neighborhood U of m such that | U ts 1:1 and y(U) 18 slice of ( oor Let (Ws) be centered coordinate system sboxt wn) with coordinate functions y.,...,y4- By Corollary 130 w Sone o Femorey {is 8 coordinate map on a neighborhood ” of m where 7,: Ré—+ Re projection on the fin ¢ coordinates, Define functions {x) on Greer) M(H) by wetting, @ =f te n—yeyePioner (meth. The functions {x} are independent at ym), since at ym), fay, Gah’ ant Zeudy, Cnet teed) for some constants ay .. By Corollary (b) of 1.30 the (x, forms coordinate system on a neighborhood of w(m). Let V'be a neighborhood of y(m) ‘on which the x,,..., xg form a cubic coordinate system. Denote the corresponding coordinalz map by 9. Let Um yw HV) OV". Then U and (¥,9) are the required neighborhood and coordinate system. a ® ye ah eS ‘Submanifolds, Difeomorphisms, Inverse Function Theorem 29 ‘We emphasize that this proposition only says that there is a neighborhood U of me such that y(U) it 8 slice of the coordinate system (V.¢). Even if (Myy)isa submanifold of N, it may well be that y(M) Vis far from being a slice or even a union of slices. For an example, consider again the fgure-8 submanifold of the plane: ‘However, inthe case that (My) is an imbedded submanifold, the coordinate system (V,g) can be chosen so that all of y(M) 7 V is a single slice of ¥. Let us now consider the question ofthe extent to which the set of C™ functions ona manifold determines the set of C= functions ona submanifold Let (iy) be a submanifold of N. Then, of course, f fe C*(N), then {JIM isa C® function on M. (More precisely, f« is &C function on M.) ‘in general, however, the converse does not hold; thats, not all C® functions oni arse asthe resnctions to M of C~ functions on N. For the converse to hold, itis necessary and sufficient to assume that yis an imbedding and that pCi) is cloned. We prove the saficiency inthe folowing proposition, tnd leave the necesty as Exercise II below. 1.36 Proposition Let yi M—=N be an Imbedding such that y(M) is closed in N. If C™(M), then there exists fe C*(N) such that fo y = 8. ‘To simplify notation, we shall suppress the map y and consider M'< N. vnoor For each point p € M there exists an open set O, in W containing and an extension of g ftom 0, AM to x C® function £, on O,. ‘One simply has to take O, to be & cubic-centered coordinate neighbor- hhood of pfor which M (0, is singe slice, and then define gto be the Composition of the natural projection of 0, onto the alice followed by The collection (O,: p €M) together with W”— M forms an open over of N. By Theorem 11. there exists partition of unity {p), with j= 1,2... subordinate to this cover. Take the subsequence (which ‘we shall continue to denote by {g,)) such that supp 9, OM xf. For cach such J, we ean choose a point p, such that supp ¢, = O,,- Then $= ¥ gids, is & C* fonction on N, and f] Mm g. 30 Manifolds IMPLICIT FUNCTION THEOREMS Fm he ee fesion throm we sal hi oo my re Gre etn hsv wt ee folds. Under suitable conditions on » differentiable Spine st ane aiceeen eae cop, ena of bral of rng ra mbaeld en ten Teme ie tn ion first “implicit function” theorem (1.38) that we shall prove for manifolds. fe late a res 19) a lf 137 Spl Puaoe Teen et UE Re pnd Ba ae at eae RE RE BY Fay 5 Teas Say ++ Sd- Suppose that at the point (Py 44) €U se ‘and that the matrix ah, oleate 4s non-singular. ‘Then there exists an open neighborhood V of 14 in Re and tn open mighorhod W of ein BO oll than Vc We Ue hr ess 2 C* map g: V+ W such that for each (pag eV x W o Sg) =0 <> gms). Implicit Function Theorems 31 1.38 Theorem Assume that y: Mt» N® ts C*, that nts a point of N, that P= yi(a) is nonempty, and that dy: Mu —» Nein 3 surjective for all IeeP. Then has a unique manifold structure Such that (P,f) isa submanifold Of M, where ts the Inclusion map. -Moreooer, {:P—+ M is actualy an Inbedding, and the dimension of Pts — a. ‘moor According to result (b) of 1.33, its sufcient to prove that in The relative topology, has a diferatable structure such that P.) ft submanifold of M of dimension ¢ — 4. For ths itis suficient to prove that ifm € P, then there exists a coordinate system on a neighborhood U of min M for which P 7 U is a single slice ofthe correct dimension. Let sys. 4%, be 8 coordinate system centered at nin N. Then since Fe Ti Niki surjective follows from Corollary (2) of 1.30 that the collection of functions Ore seep ta hde forms part of a coordinate aystem about me M. Complete to a co- ‘ordinate system yi, <= +a Jens -. Je on a neighborhood U of m. ‘Then P 7 U is precisely the slice of this coordinate system given by Aa ee. In this theorem the inverse image ofa point is shown to be a submanifold as long as the diferential Is surjective at cach point of the inverse image ‘A point can be thought of asa O-dimensional submanifold. We now generalize {IEE theorem by proving that under suitable conditions the inverse images of higher dimensional submanifolds are themeves submanifolds. 1.39 Theorem Assome that y: M-+NF is C* and that (O'.9) # a submanifold of N. Suppose that whenever m €y(9(0)), then ” Non = AVM) + de(Opreen) (not necessarily a direct sum). Then if P= y¥(9(0)) and is non-empty, Pear be gioen a manifold suctre so that (P,) ts @ submanifold of M, where jis the inchtion map, with @ dim M — dim P = dim N — dim O. Moreover, if (0,9) isan imbedded submanifold, then so is (Pi), and in this ate there if wnigue manifold structure on P such that (Pf) is @ sbmanifold eM. Tn general, if (0,9) is not an imbedding, P need not have the relative topology. and there no unique manifold structure on P such that (P.) spn submanifold "We leave itt the reader to supply examples. 32 Manifolds rnost_Th ti conf in i co te a nt ci of ay ing con an sigue ot 25 22 Perma 135 wea ee coordinate functions x,...., x, about %$(p) such that (5) is the slice o RyeO Yabidmo. Let o eREAREY, (a) = Gray So ® eX(9) = 0). Let © vey, and it o wm terey|U: UR, ‘Now, (3) through (7) imply that . VO = ¥(9(). "or each point m € v0), dy, | M is surjective since d(x » 7) Darn aero er MN aa) = dp(Mu) + (0). ‘Thus by 1.38, 9-1(9()) has a unique manifold structure (eg07),) iba sebmanifeld of hy and in thi manifold structure, Implicit Function Theorems 33 ¥y(9(W/)) has the relative topology and has dimension equal to Yim M — dim N+ dim. Cover O by a countable collection (Wei 1,2,3,...) of such sets, Then © P= Dv om) 11 tj, the submanifolds y-*((H/)) and (oC) intersect in open subset ofeach other, It Ollow thatthe union of the topologies on the ‘various (¢(W¥) forms a basi fora topology on P. With this topology, is focally Euclidean space with dimension equal to dim M — dim N+ dim0; and, moreover, P is second countable since () ‘P as a countable union of second countable open subse ‘The manifold structures on the various y*(¢(W¥)) are compatible on overlaps because of the uniqueness of the manifold structure on ‘y-(o(W.)) (or any open subset of w-H((W,))) such that (y-“(9(H/)).f) is a submanifold of M (by result (b) of 1.33). ‘Thus the of coordinate systems on P containing the coordinate systems on the Various p"(g(7)) and maximal with repect to 1.4(0) forms a diferen- table structure on P. That (P,) is a submanifold of M now follows immediately from the fact that the (y"X(9(W))/) ate submanifolds. IF (0,9)is an mbedding, the coordinate neighborhood ¥ can be chosen small enough s0 that (0) © ¥ consists only of the single slice oH), and thus UMP = yi(q(W)). Te follows in thie case that P has the relative topology; hence (P,i) is an imbedding. ‘The uniqueness of the ‘manifold structure ia this case is guaranteed by result (b) of 1.33. 1.40 Examples (@) The differential of the function /(p) = Srp)" on IRt is susjctive ‘except at the origin. Thus it follows from 1.38 that the sphere f-1(r9), fora constant r > 0, has « unique manifold structure for which it is & submanifold of R¢ under the inclusion map. In particular, this is the Same manifold structure as the one defined in Example 1.5(@). (©) We define a map y from the general linear group Gi(d,R) (Example 1.5(0) to the vector space of all real symmetric d x d matrices by vA) = Aa, where A! is the transpout of the matrix A. Let Old) = XM) Where Fis the x d identity matrix. O(é) is a subgroup of Gia.) lunder matrix multiplication called the orthogonal group. To apply 1.38 to conclude that O(d) has a unique manifold structure such that (O(@yi) isa submanifold of GUa,R), and that in this manifold struc- ture i is an imbedding and O(@) has dimension }(d(d—1)), one surjective at each #6 O(d), For thi surjective, since whenever 6 € O(d), ve whee re (ight manaion by oie the difeomorpism of GER Seton by ri) = We ere et tte eer tn nee, VECTOR FIELDS 1.41 Definitions Smooth curves o:(a,b)-+M and their tangent vectors (0) were defined in123(¢). We say that amappingo: (0, -+ Mise Imnooth curse in M ite exteods to be a Co mapping of (e — 0,9) 21 for some « > 0. ‘The cue o: [ob] + M is tid to be plecevse smooth if there exists a partition a w 04 Misa smooth curve ia M, then te tangent vector 49 = do(2]) eMan in well-defined for each 1 € [a,b] 142 Definitions A vector field X along a curve o: [0,b)—>M is a mapping X: (0.8) -> T(M) which lifts 0: that is, eX ea. A vector field m0 Vector Fields 35 is called a smooth (C°) vector field along o if the mapping X: [a,b] —~ aia eee feu X on on open ot U in Mie ifing of Uinta T(M), that is, a map X: U—+ T(M) such that eX identity map on U. Again, sor field X to be smooth (C™) means that X € C°(U,T(M)). Se Srtowt es roe U form ne chou wy a er Seah eg Cole) of C= hncbot oo en en yee ne TE ee ot a van ten MUD nthe ion oo se ane tw 8 ey 1.43 Proposition Let X'be @ vector field on M. Then the folowing are equivalent: i (8) Us z1s.2-+80 1 a coordinate system on M, and if (a) isthe collection of functions on U defined by $,2 x|unSax. then 0,€ C*(U). (©) Whenever Vis open in M and fe C°(V), then X(f) €C°(V). ynoor (a) = (b) ‘The fact that X is smooth implies that X| U is smooth; and since the composition of differentiable maps is again ‘ifferentible, it follows that dx,«X|U is smooth. (Recall that the dx, are coordinate functions on x-¥(U) © T(M), 1.25(3)) But drjoX|U= a. Hence the a, are C® functions on U. (©) It mffces to prove that XY)|VeCe(U) where OO. ee talead esate hom oo for sich UC V. Bat, by (), fod the right-hand side fs & > function on U. ici vis (= (@® To prove that X is C°, it suffices to prove that X| Cr where (Uszies soz) a an asbitrary coordinate system 00M. ‘To prove that x | Uis C>, we need only check that X | U composed wit the canonical coordinate functions 1.25(3) on 2-%(U) are C° functions. Now, en eX|U =x, and de,eX|U = X(x), all of which are Cm functions on U. 36 Manifolds 1.44 Le Bracket If X and ¥ are smooth vector flds on M, we define ‘vector field (X,Y] called the Lie bracker of X and Y by setting. © BOYD = Xa — Ya 1.43 Proposition (©) (GY) ts indeed a smooth vector field on M. ©) Wf. C*(A), then [FXg¥] — fgX.Y] + S(XH)Y — gINX.” © BM -1nm (© (WXYZ) + [FZ12] + [12.217] = 0for ell smooth vector fields X,Y, and Z on M. We leave the proof as an exercise. Part (8) is known as the Jacob dent. A pete space with a bilinear operation satisfying (c) and (4) is called 1.46 Definition Let X'be a smooth vector field on M. A smooth curve in M isan integral curve of X it a HO = X(o(8)) {for each t in the domain of @. 1.47 Let Xbe a C® vector field on M, and let me M. Let us now consider the question: Does there exist an integral curve of X through m, and if 10, is there a unique one? ‘A curve 7: (a,b) > M isan integral curve of X if and only if a (Z|) =x0) (@e@d). Let us interpret this in coordinates. Suppose that 0 € (a8) and 7(0) = m. Choose a coordinate system (U,9) with coordinate functions 2... x, shout m. By 143(6), +, 2 ® xve$nd where the f; are C® functions on U. Moreover, for each ¢ such that p(t) U, © ay(4|) = 3en| 2 lal) A de Lela” ‘Thus, in view of (2) and (3), equation (1) becomes <2] 2 a Lael S605, @ Vector Fields 37 ‘Thus 7 is an integral curve of X on 7-*(U) if and only if vd and 1eXU), ©) Dafoe OO.) Ont where y, 2,7. Equation (9) isa system of frst order ordinary differential ‘equations for which there exist fundamental existence and uniqueness theorems {11}. These theorems, when translated into manifold terminology, give the following. 1.48 Theorem Let X be a C™ vector field on a diferensiable manifold Mt. For each m € M there exists a(m) and b(mm) in RU {sk 0), and a smooth curve O} (20m) bom) > at such that (@) 06 (a(mm),b(om)) and y.4(0) = m. () y_ ts.an integral curve of X. (©) If: (ed) M tsa smooth curve satisfying conditions (a) and (b), then (ed) © (a(em),b(em)) and p= yu | (rd). We continue with the statement of the theorem after the followi Definition For each rR, we define a transformation X, with domain o fore Me 4 (alm) be)) by setting ® Xie) = 7a- (4) For each me M, there exists an open neighborhood V of m and ‘ane > 0 such that the map CO (ep) X00) is defined and is C* from (—e,¢) x V into M. (©) & is open for each © UMnM. (g) Xj: Q, +P» isa diffeomorphism with inverse X, (h) Let sand te real numbers. Then the domain of X, © X, is contained in but generally not equal 10 Dyy,. However, the domain of X,~ Xs is @,4,inthe case in which sand t both hare the same sign. Moreover, fom the domain of X, «X, we hace © Keke = Xe rRoor We let (a(m).b(m)) be the union of all the open intervals Which contain the origin and which are domains of integral curves of X satisfying the intial condition that the origin maps to m. That {eim),b(r0) x 2 (and hence part (1) holds) follows from an application of the fundamental existence theorem (11, THEOREM 4, p. 28) to the system 1.47(5). Now if « and 8 are integral curves of X with domains the open intervals A and B (with A 7 Bw ), and if« and have the same intial conditions a() = Ala) at some point f€ A 7B, then the rubset of 4 7 on which « and f agree is nonempty, open by the basic ‘uniqueness theorem [I1, THEOREM 3, p. 28), and closed by contiouit tnd hence this subset is equal to A 7 by the connectedness of 7 It follows that there exists a curve yq, defined on (a(m).b(m)) and ‘satisfying parts (2), (b), and (¢). The existence of an ¢ > 0 and a acighborhood ¥ of m such that the ‘map (4)is defined on (—,8) x Vis the content of THEOREM 7 on p. 29 of [11], That the map (6) ix smooth (and hence part (4) holds) follows from THEOREX 9 08 p. 29 of [11] on the differentiability of the solutions of 1.47(S) with respect to their initial values. Next we prove part (h). Let re (a(m),b(m)). Then sr ya(6-+ 3) is an integral curve of X with the initial condition O+ yq(") and with ‘maximal domain (a(m) — ¢, b(m) — t). Tt follows from part (e) that © (em) = 1, Bem) — #) = (@( 7060) HC a(0))» ‘and for s in the interval (6), o Yruol) = Yall +8). ‘Now let m belong to the domain of X,~ X;. Then 1€ (a(m),b(m)) and 16 (o(y(0) Her 9(0)), 40 by (0.8 + 1 (e(m).Nom). Thos m eB. and (5) follows from (7). Tt is easy to construct examples to show thet the domain of X,+X, is generally not equal to @,,,. (Consider, for example, the vector field 0/0r, on R* — {0} with s= —1 and t= 1.) If, however, s and ¢ both have the same sign, and if m€,,,, that is, it's + 1 (a(m),b(m)), then it follows that’ re (a(m),B(m)) and, by (©), 8 (e(ya(0)) Gra); ence mis in the domain of X, +X, Parts (¢) and (g) are trivial if ¢ = 0, so assume that ¢ > 0 and that me. (A similar argument will prove (¢) and (&) if 1<0) It follows from part (8) and the compactness of [0] that there exists pean wo SO such thet the map) fe defined and C* on (—1 W. Choose a positive integer m large ‘enough 40 that ne (—e), Let ay = X,, | Wand let Wi = 94H). ‘Then for f= 2,... we inductively define ac Xun | Mes and Wem as). Vector Fields 39° ais a C® map on the open set Waa W. It follows that Ws an ‘open subset of W, that 1, contains m (since if Xy, composed with Oe appli tom we obtain y(t), which Hie ia W), and that by part), © eqeoges sea a Xs | Mae Consequently. W. € 9; hence 9, is open, which proves part (2), Finally, X-is a 1:1 map of onto 7, with inverse Xs. That 1X, is C™ Gimilarly for X) follows from (8), which locally expreses wee composition of C® maps. Hence X, is diffeomorphism from B, to 9 ,, which proves part (g) and finishes Theorem 1.48. 149 Definitions A smooth vector fickd X 00 M is complete if @, = M Tor al (that i the domain of 7 is (—22,2) for each me M). To this wire, the trancformations X, form a group of transformations of M pars: ‘retried by the real numbers called the !-porameter group of X- It X isnot Complete, the transformations X, do not form a group tince their domains crit In this case, we shail refer to the collection of transformations depend 1X; as the local l-parameter group of X. 1.50 Remarks A simple example of a non-complete vector field is ‘biained by considering the vector field 2/@r, on the plane with the origin connie. if @ > 0, the domain of the maximal integral curve through (exo) is (a, <2). In the case in which the manifold M is compact, any eee ector field on Mis complete, We leave the proof as an exercise, 1.51 Definition Let y: M-+ be C*. A smooth vector field X slong WMloat is Xe C“(MGT(N)) and 2 X= 9) as Local C™ extensions in N Gren me M there exist a neighborhood U of m and a neighbortooy Wee a such that y(U) © V, and there also exists a C™ vector field © fon V such that ao Poy[URx|U. 1:52 Remark It is easy to prove that a C™ vector field X along an ae earls always has local C> extensions ia N. However, if vita immersion, such exensions generally do not exit Consider the Following example. We abll fist define a amooth vector fed X along ® Smooth eure e: Fir Fin the rel line, Let o a=, =r where ris the canonical coordinate function on FR), and et (dat is, ® xo = 09 = a(S). 7(R) —#_, Ry & z . aaccee ro RO a Since «is a homeomo i Since it homeomorphiom, herein inducd a vector fed on Ri 4 hat the above diagram commu, Now, X is smooth vector field a ‘smooth vector eld on R. To show this, leu = 1. Then ® Bem Fag XO, a(d 1) walaLo alo © Rasen ar’ ‘Thus and the function rs not differentia ‘ci site eon ot iene a he gin. we end tbe ° a) = (0), again we let X(0) = 4(), then X is toetorane C= curves te plane, ehck adnan at Gl atene te — which admits no local C* extension to 1.83. Proposition Let me M4 moo Such thar XOn) 960. Thon thre ‘etna « eons we eran om Wg) coordinate functions x;,... , xg 0n a neighborhood of m such that oy ajv-2 vee rR00F Ch ral” 10086 a coordinate system functions ys sya, i nate system (Vir) cetered at m with coordinate 2 x~-2| ante It follows from 1.48(4) tha : mm | there exists an ¢ a nei W of the origin in R* such that an ¢>0 and 2 neighhorhood . U0.) = X40. 4... 00) i well-defined and smooth for (t,,...,0,)€(—e) x We RE Distributions and the Frobenius Theorem 41 ‘Now, ¢ is non-singular at the origin since (21) axe to(2 b> 2) eof) -- aL et mal => “Thus by Corotiay (a) of 130, ¢ = © is w coordina map on vome neighborhood U of m. Let x, x, denote the coordinate functions of the coordinate system (U. we have 1.54 Definition Let ¢: M—+N be C=, Smooth vector fields X on M and Yon N are called g-relared if dg X= Yeo. 1.85 Proposition Let g: M—+N be C*. Let X and X, be smooth vector fields on M, and let Y and Y, be smooth vector fields on N. If X is prelated to ¥, and if X, is qrelated to ¥,, then (X,X,) it p-related 0 LY. Yi) vroor We must show that dps [XX] = [¥%i]+9. For this, let me Mand fe C*(N). Then we must show that o AA CXIID = Yaa. ‘We simply unwind the definitions: O AKIN = UM 9) = Xa(T(F> ) — File XUV* 9) = Xa((de ° XN) — Kan (49° ON) = XACT + 9) — Mla) ° 9) =A lXM HN) — del XN) = Yom) — Yum (FN) = Kbum(D- DISTRIBUTIONS AND THE FROBENIUS THEOREM 1.56 Definitions Let ¢ be an integer, 1Se

    You might also like