Документ взят из кэша поисковой машины. Адрес оригинального документа : http://higeom.math.msu.su/~asmish/Lichnaja-2010/Version2010-11-20/UchProcess-2015-16/Literature/CitedPapers/%5B4%5D-Milnor-1965.pdf Дата изменения: Sun May 29 19:09:42 2011 Дата индексирования: Sun Apr 10 01:53:10 2016 Кодировка: Поисковые слова: south pole

TOPOLOGY

FROM THE
DIFFERENTIABLE VIEWPOINT
By John W. Milnor
Princeton University

Based on notes by David W. Weaver

The University Press of Virginia Charlottesville


PREFACE

THESE lectures were delivered at the University of Virginia in December 1963 under the sponsorship of the Page-Barbour Lecture Foundation. They present some topics from the beginnings of topology, centering about L. E. J. Brouwer's definition, in 1912, of the degree of a mapping. The methods used, however, are those of differential topology, rather than the combinatorial methods of Brouwer. The concept of regular value and the theorem of Sard and Brown, which asserts that every smooth mapping has regular values, play a central role. To simplify the presentation, all manifolds are taken to be infinitely differentiable and to be explicitly embedded in euclidean space. A small amount of point-set topology and of real variable theory is taken for granted. I would like here to express my gratitude to David Weaver, whose untimely death has saddened us all. His excellent set of notes made this manuscript possible.

J. W. M.
Princeton, New Jersey March 1965

vii


CONTENTS

Preface
1. Smooth manifolds and smooth maps Tangent spaces and derivatives Regular values The fundamental theorem of algebra

v11
1 2 7 8

2. The theorem of Sard and Brown Manifolds with boundary The Brouwer fixed point theorem
3. Proof of Sard's theorem
4. The degree modulo 2 of a mapping Smooth homotopy and smooth isotopy

10 12 13
16

I

20 20 26 27 32
42 50

J

5. Oriented manifolds The Brouwer degree
6. Vector fields and the Euler number 7. Framed cobordism; the Pontryagin construction The Hopf theorem
8. Exercises

52
55
59

Appendix: Classifying l-manifolds Bibliography Index

63

ix


Q1. SMOOTH MANIFOLDS AND SMOOTH MAPS

FIRST let us explain some of our terms. Rk denotes the k-dimensional euclidean space; thus a point x E Rk is an k-tuple x = (xi,. . . , xk) of

real numbers. Let U C Rk and V C R' be open sets. A mapping f from U to V (written f : U .+ V) is called smooth if all of the partial derivatives ay/ax,, . ax, exist and are continuous. More generally let X C Rk and Y C R' be arbitrary subsets of euclidean spaces. A map f : X Y is called smooth if for each x E X there exist an open set U C Rk containing x and a smooth mapping F : U --f R1 that coincides with f throughout U A X. If f : X + Y and g : Y -+ 2 are smooth, note that the composition g o f : X + 2 is also smooth. The identity map of any set X is auto matically smooth.

--

---f

DEFINITION. A map f : X Y is called a di$eomorphisnz if f carries X homeomorphically onto Y and if both f and f-' are smooth. We can now indicate roughly what diferential topology is about by saying that it studies those properties of a set X C Rhwhich are invariant under diffeomorphism. We do not, however, want to look at completely arbitrary sets X. The following definition singles out a particularly attractive and useful class.
--f

DEFINITION. A subset ilf C Rkis called a smooth inanifold of dimension m if each x E Af has a neighborhood W A M that is diffeomorphic to an open subset U of the euclidean space R". Any particular diffeomorphism g : U --f W A Af is called a parametrixation of the region W A M. (The inverse diffeomorphism W A AI + U is called a system of coordinates on W A M.)


Q 1. Smooth manifolds

Tangent spaces

3

1

Rm

t
Figure 1. Parametrization of a region in IIf

origin that is parallel to this. (Compare Figures 1 and 2.) Similarly one thinks of the nonhomogeneous linear mapping from the tangent hyperplane at x to the tangent hyperplane at y which best approximates f . Translating both hyperplanes to the origin, one obtains dfz. Before giving the actual definition, we must study the special case of mappings between open sets. For any open set U C Rkthe tangent space TU, is defined to be the entire vector space Rk. For any smooth map f : U + V the derivatiue
dfz
;

Rk+ R'

is defined by the formula
df,(h)
=

Sometimes we will need to look at manifolds of dimension zero. By definition, M is a manifold of dimension zero if each x E M has a neighborhood W n M consisting of x alone.

lim (f(x
t-0

+

th) - f(x))/t

EXAMPLES. The unit sphere S2,consisting of all (2, y, z) E R3 with x2 y2 z2 = 1 is a smooth manifold of dimension 2. In fact the diff eomorphism

++ +

for x E U, h E Rk.Clearly df,(h) is a linear function of h. (In fact df, is just that linear mapping which corresponds to the 1 X k matrix (af Jaxi>, of first partial derivatives, evaluated at 2.) Here are two fundamental properties of the derivative operation:

(x, Y)

+

(x, Y, d1 - x2 - Y",

for x2 y2 < 1, parametrizes the region z > 0 of S2.By iritcrchanging the roles of x, y, z, and changing the signs of the variables, we obtain similar parametrizations of the regions x > 0, y > 0, x < 0, y < 0, and z < 0. Since these cover S2,it follows that S2 is a smooth manifold. More generally the sphere Sn-' C R" consisting of all (xl, . , x,,) with x: = 1 is a smooth manifold of dimension n - 1. For example SoC R' is a manifold consisting of just two points. A somewhat wilder example of a smooth manifold is given by the set of all (x, y) E R2 with x # 0 and y = sin(l/x).

f(x)

1 (Chain rule). If f : U + V and g : V + W are smooth maps, with = y, then d(g
0

f&

=

dg,

0

dfz.

In other words, to every commutative triangle

U w
9
Of

r/\

rv\

TANGENT SPACES AND DERIVATIVES
To define the notion of derivative df, for a smooth map f : M + N of smooth manifolds, we first associate with each x E M C Rk a linear subspace TM, Rkof dimension m called the tangent space of dd at x. Then df, will be a linear mapping from TM, to TN,, where y = f(x). Elements of the vector space TM, are called tangent vectors to llil at x. Intuitively one thinks of the m-dimensional hyperplane in Rk which best approximates 114 near x; then TM, is the hyperplane through the

of smooth maps between open subsets of Rk, Rnkthere corresponds R', a commutative triangle of linear maps

Rf Rd 2
49

R

Y

c

f)=

2. If I is the identity map of U, then dI, is the identity map of Rk. More generally, if U C U' are open sets and

i:U--+U'

I):


6

51. Smooth maniJolds

Regular values

7

smooth map

transformation by going around the bottom of the diagram. That is:

f:M+N
with f(x)
=

df,

= dh,

0

d(h-'

0

f 0 g)u 0 (dgJ*.
-+

y. The derivative

This completes the proof that
3

dfz : TM,

TN,

df, : TM,

TN,

is defined as follows. Since J is smooth there exist an open set W containing x and a smooth map F:W4R1
that coincides with f on W (7 M. De&e df,(v) to be equal to dF,(v) for all v e TM,. To justify this definition we must prove that dF,(v) belongs to TN, and that it does not depend on the particular choice of F. Choose parametri~t' ,a ions

is a well-defined linear mapping. *4s before, the derivative operation has two fundamental properties:
1. (Chain rule). If f : M + N and g : N --+ P are smooth, with f(s)= y, then d(g 0 f>r = dgu 0 dfz. 2. If I is the identity map of M, then dIz is the identity map of TM,. More generally, if M C N with inclusion map i, then TM, C Thrz with irrclusion mup di,. (Compare Figure 2.)

g:U+MCRE

and h : V - + N C R '

for neighborhoods g(U) of x and h(V) of y. Replacing U by a smaller set if necessary, we may assume that g(U) C W and that f maps g(U) into h(V). It follows that h-' o f o g : U -+ V

is a well-defined smooth mapping. Consider the commutative diagram

9

f

P
h-1

Figure 2. 7'he tangent space of a submanifold

fo

U

gJh V

The proofs are straightforward. As before, these two properties lead to the following:
ASSERTION. f : M If N is a difleonzoyphism, then df, : TAT, -+ TN, is an isomorphism of vector spaces. In payticular the dimension of ill must be equal to the dimension of N.
7

of smooth mappings between open sets. Taking derivatiws, we obtain a commutative diagram of linear mappings

RE
dgu

R m d(h-'

T

dFZ

>R1
dha,
0 g>u

0

f

REGULAR VALUES
Let f : ill .+ N be a smooth map between manifolds of the same dimension.* We say that x e n/r is a regular point of j if the derivative

>R"

where

U

= g-'(x), v = h-'(y).

It follows immediately that dF, carries TAT, = Image (dg,) into TN,, = Image (dh,). Furthermore the resulting map df, does not depend on the particular choice of F, for we can obtain the same linear

*

This restriction will be removed in $2.


8

0

1. Smooth manifolds

Fundamental theorem of algebra

9

df, is nonsingular. In this case it follows from the inverse function theorem that f maps a neighborhood of x in M diffeomorphically onto an open set in N. The point y E N is called a regular value if f-'(y) contains only regular points. If df, is singular, then x is called a critical point of f , and the image f(x) is called a critical value. Thus each y E N is either a critical value or a regular value according as f-'(y) does or does not contain a critical point. Observe that if M is compact and ?J E N is a regular value, then f-'(y) is afinite set (possibly empty). For f-'(y) is in any case compact, being a closed subset of the compact, space M; and f-'(y) is discrete, since f is one-one in a neighborhood of each x P f-'(y). For a smooth f : M --$ N, with M compact, and a regular value y E N, we define #f-'(y) to be the number of points in f-'(y). The first observation to be made about #f-'(y) is that it is locally constant as a function of y (where y ranges only through regular values!). I.e., there is a neighborhood V C N of y such that #f-'(y') = #f-'(y) for any y' E V. [Let xl, . . , xk be the points of f-'(y), and choose pairwise disjoint neighborhoods U,, , Ur of these which are mapped diffeomorphically onto neighborhoods V,, . * - , V, in N. We may then take

Figure 3. Stereographic projection

hood of the north pole. To see this we introduce the stereographic projection h- from the south pole (0, 0, -1) and set
&(z) = h-fhI'(z).

-

Kotc, by elementary geometry, that
Sow if P(z) = a,$ alzn-l computation shows that
&(z) =

--

+

h,hI'(z)

+

= z / 1 ~ 1= ~

.. .

+

1/Z.

a., with a, # 0, then a short
***

V

=

V, A V, A

A V, - f(M - U, -

- UiJ .I

z"/(a0

+ d,z +

+

an?).

THE FUNDAMENTAL THEOREM OF ALGEBRA
As an application of these notions, we prove the fundamental theorem of algebra: every nonconstant complex polynomial P(z) must have a zero. For the proof it is first necessary to pass from the plane of complex numbers to a compact manifold. Consider the unit sphere X2 C R3 and the stereographic projection

h+ : Sz- ((0, 0, 1)) -+ R2 X 0 C R3
from the "north pole" (0, 0, 1) of S2.(See Figure 3.) We will identify R2 X 0 with the plane of complex numbers. The polynomial map P from R2 x 0 itself corresponds to a map f from S2to itself; where
f(x) f(0, 0, 1)
=
=

Thus Q is smooth in a neighborhood of 0, and it follows that f = hI'Qhis smooth in a neighborhood of (0, 0, 1). Next observe that f has only a finite number of critical points; for P fails to be a local diffeomorphism only at the zeros of the derivative polynomial P'(z) = a,-i jzl-', and there are only finitely many zeros since P' is not identically zero. The set of regular values of f , being a sphere with finitely many points removed, is therefore connected. Hence the locally constant function #f-'(y) must actually be constant on this set. Since #f-'(y) can't be zero everywhere, we conclude that it is zero nowhere. Thus f is an onto mapping, and the polynomial P must have a zero.

h;'Ph+(x)
(0, 0, 1).

for x # (0, 0, 1)

It is well known that this resulting map f is smooth, even in a neighbor-


Regular values

11

82. THE THEOREM

OF SARD AND BROWN

has rank less than n (i.e. is not onto). Then C will be called the set of critical points, f(C) the set of critical values, and the complement N - f (C) the set of repular values of f. (This agrees with our previous definitions in the case n~ = n.) Since M can be covered by a countable collection of neighborhoods each diffeomorphic to an open subset of R"', we have:
f :M

Corollary (A. B. Brown). The set of regular values of a smooth map --f N is everywhere dense in N.

In order to exploit this corollary we will need the following:

Lemma 1. If f : AT --+ N is a smooth map between manifolds of dimen sion m 2 n, and if y E N is a regular value, then the set f-'(y) C M is a smooth manifold of dimension m - n.

IN

GENERAL, it is too smooth map be finite. by the next theorem, earlier work by A. P.

much to hope that But this set will be which was proved Illorse. (References

the set of critical values of a LLsmall,'l the sense indicated in by A. Sard in 19-12 following [30], [HI.)

Theorem. Let f : U + R" be a smooth map, deJined on an open set U C Rm, and let C = { X E U I rankdf, Since a set of measure zero cannot contain any nonvacuous open set, it follows that the complement R" - f(C) must be everywhere denset in R". The proof will be given in $3. It is essential for the proof that f should have many derivatives. (Compare Whitney [38].) We will be mainly interested in the case m 2 n. If m < n, then clearly C = U; hence the theorem says simply that f(U) has measure zero. More generally consider a smooth map f : M -+ N, from a manifold of dimension m to a manifold of dimension n. Let C be the set of all x E M such that

P ROOF. Let x E f-'(y). Since y is a regular value, the derivative djl must map TAI, onto 2'". The null space 91 C TM, of dfL. therefore will be an (nz.- n)-dimensional vector space. If M Rkl choose a linear map I, : Rk -+ R'''-n that is nonsingular on this subspace (31 C TM, C Rk. Now define

c

F :M
by F(E)
=

-+

N X R"-"

(f(t),L(t)).Thc

derivative dF, is clearly given by the formula

dPz(v) = (dfdv), U V ) ) . Thus dF, is nonsingular. Hence F maps some neighborhood U of x diffeoinorphically onto a neighborhood I' of (y, L(x)).Kote that f-'(y) corresponds, under F, to the hyperplane y X R"'-". In fact F maps f-'(y) r\ U diffeomorphically onto (y X R"-") A V. This proves that f-'(y) is a smooth manifold of dimension m - n.
As an example we can give an easy proof that the unit sphere Sm-' is a smooth manifold. Consider the function f : R" --+ R defined by

f(.)

=

x:

+ x; +

***

+

xi.

Any y # 0 is a regular value, and the smooth manifold f-'(l) is the unit sphere.

df, : TM, + TN,(,,

* In other words, given any B > 0,it is possible to cover f(C) by a sequence of cubes in R" having total n-dimensional volume less than E. t Proved by Arthur B. Brown in 1935. This result was rediscovered by Dubovickii in 1953 and by Thom in 1954. (References [51, [S], [361.)

If M' is a manifold which is noted that TM: is a subspace of plement of TM: in TM, is then called the space of normal vectors In particular let M' = f-'(y)

contained in &I, it has already been TM, for x E M'. The orthogonal corna vector space of dimension na - m' to M' in M at x. for a regular value y of f : M -+ N.


16

Lemma 2. The null space of df, : TM, TN, is precisely equal to the tangent space TM: C TM, of the submanifold M' = f-'(y). Hence dfl maps the orthogonal complement of TM: isomorphically onto TN,. P
ROOF.

-

52. Sard - Brown theorem

Brouwer fixed point theorem

13

E

XAMPLE

. The unit disk Dm, consisting of all x
1-

cx: 2

E

R'" with

0,

From the diagram

M'

i

is a, smooth manifold, with boundary equal to AS"-'. Now consider a smooth map f : X -+N from an m-manifold with boundary to an n-manifold, where m > n.
f

we see that df, maps the subspace TA!!: C TM, to zero. Counting dimensions we see that df, maps the space oj normal vectors to M' isomorphically onto TN,.

Lemma 4. If y c N is a regular value, both for f and for the restriction then f-'(y) C X is a smooth (m - n) - manifold with boundary. Furthermore the boundary a(f-'(y)) is precisely equal to the intersection of f-'(y) with aX.

1 ax,

MANIFOLDS WITH BOUNDARY
The lemmas above can be sharpened so as to apply to a map defined on a smooth "manifold with boundary." Consider first the closed half-space

PROOF. Since we have to prove a local property, it suffices to consider the special case of a map f : H " + R", with regular value y E R". Let Z E f-'(y). If z is an interior point, then as before f-'(y) is a smooth manifold in the neighborhood of 2. Suppose that 3 is a boundary point. Choose a smooth map g : U + R" that is defined throughout a neighborhood of 3 in R" and coincides with f on U A H". Replacing U by a smaller neighborhood if necessary, we may assume that g has no critical points. Hence g-'(y) is a smooth manifold of dimension m - n. Let IT : g-l(y) --f R denote the coordinate projection,
n(x1, * . .

H" = {(x1,

e

.

0

,

x,) E R "

I

X,

2 O}.

,

2,)

=

x,.

The boundary dH" is defined to be the hyperplane R"-' X 0

C R".
~

DEFINITION. A subset X C Rk is called a smooth m - manifold with boundary if each x E X has a neighborhood U n X diffeomorphic to an open subset V n H" of H " . The boundary aX is the set of all points in X which correspond to points of dH" under such a diffeomorphism. It is not hard to show that dX is a well-defined smooth manifold of dimension nz - 1. The interior X - dX is a smooth manifold of dimension m. The tangent space TX, is defined just as in 51, so that TX, is a full m-dimensional vector space, even if x is a boundary point. Here is one method for generating examples. Let M be a manifold without boundary and let g : M ---f R have 0 as regular value.

We claim that IT has 0 as a regular value. For the tangent space of g-'(y) at a point x c r-'(O) is equal to the null space of

dg,

=

df, : R" + R";

but the hypothesis that f I aH" is regular at x guarantees that this null space cannot be completely contained in R"-' X 0. Therefore the set g-'(y) fl H" = f-'(y) A U, consisting of all x E g-'(y) with n(x) 2 0, is a smooth manifold, by Lemma 3; with boundary equal to n-'(O). This completes the proof.

THE BROUWER FIXED POINT THEOREM
We now apply this result to prove the key lemma leading to the classical Brouwer fixed point theorem. Let X be a compact manifold with boundary.

Lemma 3. The set of x in M with g(x) 2 0 is a smooth manifold, with boundary equal to g-'(O) .
The proof is just like the proof of Lemma 1.


14

$2. Sard-Brown theorem
+

Brouwer jixed point theorem

15

Lemma 5. There is no smooth map f : X
wise jixed.

a

X that leaves a X point-

P ROOF (following M. Hirsch). Suppose there were such a map f. Let y E ax be a regular value for f. Since y is certainly a regular value for the identity map f I aX also, it follows that f-'(y) is a smooth 1manifold, with boundary consisting of the single point
f-'(y)

n

ax

=

{v).
Fiyrtre

But f-'(y) is also compact, and the only compact 1-manifolds are finite disjoint unions of circles and segments,* so that aff'(y) must consist of an even number of points. This contradiction establishes the lemma. In particular the unit disk

4

D"= (xER"Ix:+

+z:<

1)

of D" into points outside of D". To correct this we set

is a compact manifold bounded by the unit sphere x"-'. Hence as a special case we have proved that the identity map of A"-' cannot be extended to a smooth map D" --f AY-'.
z
P

P(x>

=

P'(X)/(l

Lemma 6. Any smooth map g : D" D" with g(x) = x).

+

D" has aJixed point (i.e. a point
E

P ROOF. Suppose g has no fixed point. For x E D", let f(x) the point nearer x on the line through x and g(x). (See Figure f : D" --+ s"-' is a smooth map with f(x) = x for x E s"-', impossible by Lemma 5. (To see that f is smooth we make the explicit computation: f(x) = x tu, where

+

8'-' be 4.) Then which is following

Then clearly P maps D" into D" and I IP(x) - G(s)1 I < 2e for x E D". Suppose that G(x) # x for all x E D". Then the continuous function 1 IG(x) - $1 I must take on :L minimum p > 0 on W. Choosing 1' : D"+ I)" as above, with I IP(x) - G(x)I I < p for all x, we clearly have P(x) # x. Thus P is a smooth map from D" to itself without a fixed point. This contradicts Lemma 6, and completes the proof.

+

4.

I

U=

IIX

x - g(x> - g(4ll '

t

=

-x.u

+

d1 - x'x

+

(x.u)z,

The procedure employed here can frequently be applied in more general situations: to prove a proposition about continuous mappings, we first establish the result for smooth mappings and then try to use an approximation theorem to pass to the continuous case. (Compare 58, Problem 4.)

the expression under the square root sign being strictly positive. Here and subsequently llxll denotes the euclidean length dx; . . . x:.)

+

+

Brouwer Fixed Point Theorem. Any continuous function G : D"
has a Jixed point.

+

D"

PROOF. We reduce this theorem to the lemma by approximating G by a smooth mapping. Given t > 0, according to the Weierstrass approximation theorem,t there is a polynomial function P, : R" -+ R" with IIP1(x) - G(x)II < t for 2 E D". However, P, may send points
* A proof is given in the Appendix.
f See for example Dieudonn6 [7, p. 1331.


Step 1

17

the C, is vacuous unless f is constant on an entire component of U. Hcnce in this case it is sufficient to carry out Steps 1 and 2.)

Q3. PROOFOF

SARD'S THEOREM*

P ROOF O F STEP 1. This first step is perhaps the hardest. We may assume that p 2 2, since C = C, when p = 1. We will need the well known theorem of Fubini* which asserts that a measurable set
A

C R"

=

R` X R"-'
(constant) X Rp-'

iriust have measure xero if it intersects each hyperplane in a set of (p - 1)-dimensional measure zero.

For each f E C - C, we will find an open neighborhood V C R" so that f(V (7 C) has measure zero. Since C - C, is covered by countably many of these neighborhoods, this will prove that f (C - C,) has measure

FIRST let us recall the statement:

Theorem of Sard. Let f : U .+R" be a smooth map, with U open in Rn, and let C be the set of critical points; that is the set of all x E U with
rank dfz
Then f(C)

zero. Sincc f # C,, there is some partial derivative, say dfl/dzl, which is not zero at 3. Consider the map h : U --f R" defined by

<

p.

c

R" has measure zero.

R EMARK. The cases where n _< p are comparatively easy. (Compare de Rham [29, p. 101.) We will, however, give a unified proof which makes these cases look just as bad as the others. The proof will be by induction on n. Note that the statement makes sense for n 2 0, p 2 1. (By definition Ro consists of a single point.) To start the induction, the theorem is certainly true for n = 0. Let C, C C denote the set of all x E U such that the first derivative dfz is zero. More generally let C, denote the set of x such that all partial derivatives of f of order
h(x) = (fl(X>, xz, . * * 12,). Sincc dh, is nonsingular, h maps some neighborhood V of phically onto an open set V'. The coniposition g = f o map V' into R". Note that the set C' of critical points of h(V n C); hence the set g(C') of critical values of g is equal

3 diffeomorh-' will then g is precisely to f(V n C).

c
S S S
TEP
TEP
TEP

3

c, 3 c,

3

c,

3 ..*

*

The proof will be divided into three steps as follows:
1. The image f(C - C,) has measure zero. 2. The image f(C, - C,,,) has measure zero, for i 2 1. 3. The image f(C,) has measure zero for k sufficiently large.

(R

EMARK

. If f happens to be real analytic, then the intersection of

I

Figure 5. Construction of the map g

Our proof is based on that given by Pontryagin [28]. The details are somewhat easier since we assume that f is infinitely differentiable.

*

* For an easy proof (as well as an alternative proof of Sard's theorem) see Sternberg [Xi, pp. 51-52]. Sternberg assumes that A is compact, but the general case follows easily from this special case.


4

18

$3. Proof of Surd's theorem

Step 3

19

For each (t, xpl -. . , 2,) the hyperplane t X R"-'
planes. Let
gt :

V' note that g(t, x2, . , x,J belongs to C R": thus g carries hyperplanes into hyperE
-+

(t X Rn-l) A V'

tX

R"-'
1

denote the restriction of g. Note that a point of t X R"-' is critical for gt if and only if it is critical for g; for the matrix of first derivatives of g has the form

PROOF S TEP 3. Let I " C U be a cube with edge 6. If k is sufficiently OF large (k > n/p - 1 to be precise) we will prove that f(Ck A I " ) has measure zero. Since c k can be covered by countably many such cubes, this will prove that f(C,) has measure zero. From Taylor's theorem, the compactness of I " , and the definition of ck, we see that

f(.
where
1)

+
ll&,

h)

=

f(x)

+

R(z1 h)

According to the induction hypothesis, the set of critical values of gt has measure zero in t X R"-'. Therefore the set of critical values of g intersects each hyperplane t x R"-' in a set of measure zero. This set g(C') is measurable, since it can be expressed as a countable union of compact subsets. Hence, by Fubini's theorem, the set

I

for x E C, A I " , x h E I " . Here c is a coiistant wliich depends oiily on f and I " . Now subdivide I " into Y* cubes of edge 6/r. Let I, be a cube of the subdivision which contains a point x of c k . Then any point of I1 can be written as x h, with

+

WlI I

c llhll"'

+

g(C'> = j(V
has measure zero, and Step 1 is complete.

C)

PROOFF S TEP 2. For each z E c k - C,+, there is some (Ic O I)-'' derivative ak*'f,/ax,,...az.,+,which is not zero. Thus the function
W(X) =

+

llhll I di(6h). I'rom 1) it follows that /(I,) lies in a cube of edge a / ~ ~ ccntered +' about f(z), where a = 2c (di 6)k+1 is constant. EIence f(ck I " ) is contained in a union of at most 1''' cubes having total volume
2)

akf,/ax8, . . . ax,*+,
=

vanishes at but aw/dx,, does not. Suppose for definiteness that s, Then the map h : U -+ R" defined by

1.

If k 1 > n/p, then evidently V tends to 0 as 1' -+ m ; so f(C, n In) must have measure zero. This completes the proof of Sard's theorem.

+

V 5 rn(a/l'k+l)n a =

D n-(k+l)p

1'

h(z>=

(W(Z>,

22,

*-*

1

z,>

carries some neighborhood V of diffeomorphically onto an open set V'. Note that h carries Ck A V into the hyperplane 0 X R"-'. Again we consider g = f o h-' : V' -+ R". Let

0 : (0 X Rn-')

n V'

-+

R"

denote the restriction of g. By induction, the set of critical values of 0 has measure zero in R". But each point in h(C, A V) is certainly a critical point of 0 (since all derivatives of order _
n V)

=

f(Ck

n

V) has measure zero.

Since c, - Ck+1 is covered by countably many such sets V, it follows that f(C, -. Ck+l) has measure zero.


Homotopy and isotopy

21
+

smooth map F : X X [O, I]

Y with
Fb, 1)
= 9(z)

F(x, 0)

=

fb),

84. THE DEGREE MODULO 2

I

OF A MAPPING

for all z E X. This map F is called a smooth homotopy between f and g. Note that the relation of smooth homotopy is an equivalence relation. To see that it is transitive we use the existence of a smooth function rp : [0, I] 3 [0, I] with
p(t) =

0 for 0 5 t 5
for $

5

p(t) = 1

5 t 5 1.

CONSIDER a smooth map f : S"

+ S". If y is a regular value, recall that #f-'(y) denotes the number of solutions x to the equation f(z) = y. We will prove that the residue class modulo 2 of #f-'(y) not depend on does the choice of the regular value y. This residue class is called the mod 2 degree of f. More generally this same definition works for any smooth

(For example, let p(t) = X(t - Q)/(X(t - 8) A($ - t)), where X(T) = 0 for T 5 0 and X(7) = exp( -.-I) for 7 > 0.) Given a smooth homotopy F between f and g, the formula G(x, t) = F(x, p(t)) defines a smooth homotopy G with
I

+

G(z, t) = f(x) for 0 5 t

f:M--+N
where M is compact without boundary, N is connected, and both manifolds have the same dimension. (We may as well assume also that N is compact without boundary, since otherwise the mod 2 degree would necessarily be zero.) For the proof we introduce two new concepts.

Now if f g and g h, then, with the aid of this construction, it is easy to prove that f h. If f and g happen to be diffeomorphisms from X to Y, we can also define the concept of a "smooth isotopy" between f and g. This also will be an equivalence relation.

-

G(z, t) = g(x)

-

5 for Q 5 t 5

+

1.

,

DEFINITION. The diffeomorphism f is smoothly isotopic to g if there exists a smooth homotopy F : X X [O, 11 -+ Y from f to g so that, for each t E [0, 11, the correspondence
z
-+

SMOOTH HOMOTOPY AND SMOOTH ISOTOPY
Given X C Rk, X X [0, I] denote the subset* of Rk" consisting let of all (x,t) with z E X and 0 5 t 5 1. Two mappings
f, g

F(z, t)

maps X diffeomorphically ont.0

Y.

It will turn out that the mod 2 degree of a map depends only on its smooth homotopy class:
Homotopy Lemma. Let f, g : M --+ N be smoothly homotopic maps between manifolds of the same dimension, where M is compact and without boundary. If y E N is a regular value for both f and g, then
#f-'(y> = #9-'(y>

:x+

Y

are called smoothly homotopic (abbreviated f

-

g) if there exists

a

manifold bounded by two "copies" of M. Boundary points of M will give rise to "corner" points of M X [O, 13.

* If M is a smooth manifold without boundary, then M X [0, 11 is a smooth

(mod 2).

PROOF. Let F : M X [0, 11 + N be a smooth homotopy between f and g. First suppose that. y is also a regular value for F. Then F-'(y)


22

$4. Degree modulo 2

Homotopy and isotopy

23

is a compact 1-manifold, with boundary equal to
~ - y y n (M )

x ov

~I/Ix 1) = f-'(y)

x ov

g-l(y)

x

1.

Thus the total number of boundary points of F-'(y) is equal to

#f-'(y)

+

#s-'(y).

(For the special case N = x" the proof is easy: simply choose h to be the rotation which carries y into z and leaves fixed all vectors orthogonal to the plane through y and z.) The proof in general proceeds as follows: We will first construct a smooth isotopy from R" to itself which
1) leaves all points outside of the unit ball fixed, and 2) slides the origin to any desired point of the open unit ball.

But we recall from $2 that a compact 1-manifold always has an even number of boundary points. Thus #j-'(y) #g-'(y) is even, and therefore

+

#f-'(y)

= #g-'(Y>T(mod 2).
C

Figure 7. Dejorming the unit ball Mx 0 Mxl

Figure 6. The number of boundary points on the left is congruent to the number on the right modulo 2

Let

cp

: R" --+

R be a smooth function which satisfies
&)
q(x)

>
=

0 .for 11x1 0 for

I<

1

Now suppose that y is not a regular value of F. Recall (from 51) that #f-'(y') and #g-'(y') are locally constant functions of y' (as long as we stay away from critical values). Thus there is a neighborhood V, C N of y, consisting of regular values of f , so that

11x11 2 1.

(For example let p(z) = A(l - ll~11~) where A(t) = 0 for t 5 0 and A(t) = exp(-t-') for t > 0.) Given any fixed unit vector c E S"", consider t,he differential equations dxi dt -- ci&, -

#f-'(Y'>
for all y' for all y'
E

=

#f-W

V,; and there is an analogous neighborhood V,
#g-'(u')
= #g-'(Y)

c

-.-,

z,);

i

=

1,

-**

, n.

N so that

For any 2 E R" these equations have a unique solution x = x(t), defined for all* real numbers which satisfies the initial condition
x(0)
= 2.

E

V,. Choose a regular value z of F within V, A V,. Then
#f-'(Y> = #f-'(z) = #g-%)
= #s-YY)),

We will use the notation x(t) = Ft(2) for this solution. Then clearly
1) F,(2) is defined for all t and 2 and depends smoothly on t and 2, 2) F"(3) = 3, 3) F.+,(3) = F, 0 F,(Z).

which completes the proof. We will also need the following:

Homogeneity Lemma. Let y and z be arbitrary interior points of the smooth, connected manifold N. Then there exists a diffeomorphism h: N +N that is smoothly isotopic to the identity and carries y into z.

I
~

*

Compare [22, 02.41.


24

$4. Degree modulo 2

Homotopy and isotopy

25

Therefore each F, is a diffeomorphism from R" onto itself. Letting t vary, we see that each F, is smoothly isotopic to the identity under an isotopy which leaves all points outside of the unit ball fixed. But clearly, with suitable choice of c and t, the diffeomorphism F, will carry the origin to any desired point in the open unit ball. Now consider a connected manifold N. Call two points of N "isotopic" if there exists a smooth isotopy carrying one to the other. This is clearly an equivalence relation. If y is an interior point, then it has a neighborhood diffeomorphic to R"; hence the above argument shows that every point sufficiently close to y is "isotopic" to y. In other words, each "isotopy class" of points in the interior of N is an open set, and the interior of N is partitioned into disjoint open isotopy classes. But the interior of N is connected; hence there can be only one such isotopy class. This completes the proof. We can now prove the main result of this section. Assume that M is compact and boundaryless, that N is connected, and that f : A4 -+ N is smooth.

which is a regular value for both f and g. The congruence (mod 2) deg, f = #f-'(y) = #g-'(y) = deg, g now shows that deg, f is a smooth homotopy invariant, and completes the proof.

EXAMPLES. A constant map c : M -+ M has even mod 2 degree. The identity map I of M has odd degree. Hence the identity map of a compact boundaryless manifold is not homotopic to a constant. In the case M = s",this result implies the assertion that no smooth map f : Dn+' S" leaves the sphere pointwise fixed. (I.e., the sphere -+ is not a smooth ((retract" of the disk. Compare $2, Lemma 5.) For such a map f would give rise to a smooth homotopy

I

F : S" X [0, 11

-+

!

Sn, F(z, t)

=

f(tz),

between a constant map and the identity.

Theorem. If y and

z are regular values of f then

#f-'(y)

= #f-'(z)

(modulo 2).

This common residue class, which is called the mod 2 degree of f, depends only on the smooth homotopy class of f.

PROOF. Given regular values y and z, let h be a diffeomorphism from N to N which is isotopic to the identity and which carries y to z. Then z is a regular value of the composition h o f. Since h o f is homotopic to I, the Homotopy Lemma asserts that
#(h 0 f)-'(z) = #f-'(z)

(mod 2).
c

But
(h 0 f)-'(z) = f-'h-'(z)
=

f-'(y),

so that
#(h O f)-'(z)
=

#f-'(y>.

Therefore
#f-'(y) = #f-'(z)
(mod 21,

as required. Call this common residue class deg2(f).Now suppose that f is smoothly homotopic to g. By Sard's theorem, there exists an element y P N


The Rrouwer. degree

27

Q5. ORIENTED MANIFOLDS

Each orientation for M determines an orientation for dM as follows: For x E dM choose a positively oriented basis (ul, v2) . U,) for TM, in such a way that v2, * . . , U , are tangent to the boundary (assuming that m 2 2) and that v, is an `(outward" vector. Then (u2, .-., U,) determines the required orientation for dM at x. If the dimension of M is 1, then each boundary point x is assigned the orientation - 1 or 1 according as a positively oriented vector at x points inward or outward. (See Figure 8.)

--

)

+

IN ORDER to define the degree as an integer (rather than an integer modulo 2) we must introduce orientations.
DEFINITIONS. An orientation for a finite dimensional real vector space is an equivalence class of ordered bases as follows: the ordered basis (bll , b,) determines the same orientation as the basis (6;, . , 69 if b: = a,,b, with det(a,,) > 0. It determines the opposite orientation if det(a.,) < 0. Thus each positive dimensional vector space has precisely two orientations. The vector space R" has a standard orientation corresponding to the basis (1, 0, . , 0))(0, 1, 0, , 0), , (0, , 0, 1). In the case of the zero dimensional vector space it is convenient to define an `(orientation" as the symbol 1 or - 1. An oriented smooth manifold consists of a manifold M together with a choice of orientation for each tangent space TM,. If m 2 1, these are required to fit together as follows: For each point of M there should exist a neighborhood U C M and a diffeomorphism h mapping U onto an open subset of R" or H"' which is orientation preserving, in the sense that for each x E U the isomorphism dh, carries the specified orientation for TM, into the standard orientation for R". If M is connected and orientable, then it has precisely two orientations. If M has a boundary, we can distinguish three kinds of vectors in the tangent space TM, at a boundary point:

--.

1I
Figure 8. How to orient a boundary

--

--

-

+

As an example the unit sphere Sm-' R" can be oriented as the C boundary of the disk D".

THE BROUWER DEGREE
Now let M and N be oriented n-dimensional manifolds without boundary and let
f :M+N

1) there are the vectors tangent to the boundary, forming an (m - 1)dimensional subspace T(dM), C TM,; 2) there are the (`outward" vectors, forming an open half space bounded by T(dM),; 3) there are the (`inward" vectors forming a complementary half space.

be a smooth map. If M is compact and N is connected, then the degree of f is defined as follows: Let x E M be a regular point of f, so that df, : TM, -+ TNt(,, is a linear isomorphism between oriented vector spaces. Define the sign of df, to be 1 or - 1 according as df, preserves or reverses orientation. For any regular value y E N define

+

deg(f; y)

=
zrf-'(u)

sign df,.

As in $1, this integer deg(f; y) is a locally constant function of y. It is defined on a dense open subset of N.


28

$5. Oriented manifolds

The Brouwer degree

29

Theorem A. The integer deg(f; y) does not depend on the choice of regular value y.

It follows immediately that
signdf,
=

-1,

It will be called the degree of f (denoted deg f).
Theorem B. If f is smoothly homotopic to g, then deg f
=

signdf,

=

+l;

deg g.

The proof will be essentially the same as that in $4.It is only necessary to keep careful control of orientations. First consider the following situation: Suppose that M is the boundary of a compact oriented manifold X and that M is oriented as the boundary of

with sum zero. Adding up over all such arcs A, we have proved that deg(f; y) = 0. More generally, suppose that yo is a regular value for f, but not for F. The function deg(f; y) is constant within some neigliborhood U of y,,. Hence, as in $4, we can choose a regular value y for F within U and observe that d 4 f ; Yo) This proves Lemma 1.
=

x.

deg(f; Y) = 0.
F : [0, I] x Ai? -+ N g b ) = F(1, z).

Lemma 1. If f : M -+ N extends to a smooth map F deg(f; y) = 0 for every regular value y. P
ROOF

:

X

--f

N, then

. First suppose that y is a regular value for F, as well as for f = F I M. The compact. 1-manifold F-'(y) is a finite union of arcs and circles, with only the boundary points of the arcs lying on M = aX. Let A C F-'(y) be one of these arcs, with aA = {a) U (0). We will show that sign dfa sign dfb = 0,

,

Now consider a smooth homotopy two mappings fb) = F(O, X I ,

between

Lemma 2. The degree deg(g; y) is equal to deg(f; y) for any common regular value y. PROOF. manifold [0, I] X M" can be oriented as a product, The and will then have boundary consisting of 1 X M" (with the correct orientation) and 0 X M" (with the wrong orientation). Thus the degree of F I a([O, 11 X M") at a regular value y is equal to the difference
deg(g; Y) - deg(f; Y>* According to Lemma 1 this difference must be zero. The remainder of the proof of Theorems A and B is completely analogous to the argument in 54. If y and z are both regular values for f : M ---$ N, choose a diffeomorphism h : N -+ N that carries y to z and is isotopic to the identity. Then h will preserve orientation, and

+

and hence (summing over all such arcs) that deg(/ ; y)

=

0.

Figure 9. How to orient F-'(y)

deg(f; Y) = deg(h O f; h(d) by inspection. But f is homotopic to h o f; hence deg(h
0

The orientations for X and N determine an orientation for A as follows: Given z e A, let (vl, * * , v,,+J be a positively oriented basis for TX, with U, tangent to A. Then U, determines the required orientation for TA, if and only if dF, carries (vz, . . , U,,+') into a positively oriented basis for TN,. Let v1 (x) denote the positively oriented unit vector tangent to A at x. Clearly U, is a smooth function, and v,(x) points outward at one boundary point (say b) and inward at the other boundary point a.

f; z)
=

=

deg(f; z)

-

by Lemma 2. Therefore deg(f; y)

deg(f; z), which completes the proof.

EXAMPLES. The complex function z -+ zk,z # 0, maps the unit circle onto itself with degree k. (Here k may be positive, negative, or zero.) The degenerate mapping
f :M
--+

constant e N


30

$5. Oriented manifolds

The Brouwer degree
using the euclidean inner product. If v(z) is nonzero for all z, then we may as well suppose that
2)

31

has degree zero. A diffeomorphism f : Jl + N has dtgree 1 or - 1 according as f preserves or reverses orientation. Thus an orientation reversing difeoinorphism of a compact boundaryless manifold is not smoothly homotopic to the identity. One example of an orientation reversing difi'eomorphism is provided by the reflection r, : S" ---f x", where

+

v(z).2r(x)

=

1 for all z

E

S".

r,(zl,

..- ,

2n+1)

=

(~1,

*.. , -L,

a

-

.

, G+J.

For in any case 8(z) = v ( ~ ) / ~ ~ would)be ~ vector field which does v~~ ~ a satisfy this condition. Thus we can think of v as a smooth function from x" to itself. Now define a smooth homotopy

The antipodal map of S" has degree ( - l)"", as we can see by noting that the antipodal map is the composition of n 1 reflections:

+

F : S" X [0,
by the formula F(z, 6) and that
Y

=z

cos 0

+

7r]

+

S"
1

v(x) sin 6. Computation shows that
=

-z = r1 or2

o

0

.

.

OT,,+~(X).

F(X, e).F(z, e)

Thus if n is even, the antipodal map of x" is not smoothly homotopic to the identity, a fact not detected by the dcgree modulo 2.

As an application, following Brouwer, we show that S" admits a smooth field of nonzero tangent vectors if and only if n is odd. (Compare Figures 10 and 11.)

F(z, 0)
i

= z,

F(z,

7r)

= - 2.

Thus the antipodal map of S" is homotopic to the identity. Rut for n even we have seen that this is impossible. On the other hand, if n = 2k - 1, the explicit formula
U(X1,
*-*

,

L2k)

= (22,-.cl,

54,

-

23,

...

7 X2ki

-22k-1)

defines a nonzero tangent vector field on S". This completes the proof. It follows, incidentally, that the antipodal map of S" is homotopic to the identity for n odd. A famous theorem due to Heinz Hopf asserts that two mappings from a connected n-manifold to the n-sphere are smoothly homotopic if and only if they have the same degree. In 57 we will prove a more general result which implies Hopf's theorem.

Figure 10 (above). A nonzero vector jield on the 1-sphere Figure 11 (below). Attempts for n = B

DEFINITION. A smooth tangent vector jield on M C Rk is a smooth map v : M -+ Rksuch that ~ ( x E) TM, for each x E M. In the case of R"" this is clearly equivalent to the condition the sphere S"

c

1)

V(X).X =

0 for all

2E

S",


The index

33

86. VECTOR FIELDS AND

THE EULER NUMBER

As a further application of the concept of degree, we study vector fields on other manifolds. Consider first an open set U C R" and a smooth vector field v :U--+R"
with an isolated zero at the point z
E

t= I

t=O

U. The function

fib> =

U(.)/

I lv(4 I

I
I

maps a small sphere centered at z into the unit sphere.* The degree of this mapping is called the index L of v at the zero z. Some examples, with indices - 1, 0, 1, 2, are illustrated in Figure 12. (Intimately associated with v are the curves "tangent" to v which are obtained by solving the differential equations dx,/dt = v,(xl, . . , xJ. It is these curves which are actually sketched in Figure 12.) A zero with arbitrary index can be obtained as follows: In the plane of complex numbers the polynomial zk defines a smooth vector field with a zero of index k at the origin, and the function zk defines a vector field with a zero of index -k. We must prove that this concept of index is invariant under diffeomorphism of U. To explain what this means, let us consider the more general situation of a map f : M --+ N, with a vector field on each manifold.

C=+l

c =+2
Figure 12. Examples of plane vector fields

DEFINITION. vector fields v on A I and U' on N correspond under f The if the derivative df, carries V(X) into v'(f(x)) for each x E M.

* Each

sphere is to be oriented as the boundary of the corresponding disk.

+

If f is a diffeomorphism, then clearly The notation
2)'

U'

is uniquely determined by

U.

=

df

0 2, 0 f-1

will be used.

Lemma 1. Suppose that the vector jield v on U corresponds to
V' =

df o

of-'

on U' under a diffeomorphisiii f : U + U'. Then the index of v at an isolated zero z is equal to the index of U' at f(z).


34

$6. Vector jields

The index sum
we construct a one-parameter family of embeddings
ft : U+R"

36

Assuming Lemma 1, we can define the concept of index for a vector field w on an arbitrary manifold M as follows: If g : U + M is a para metrization of a neighborhood of z in M, then the index L of w at x is defined to be equal to the index of the corresponding vector field dg-' 0 w o g on U at the zero g-'(z). It clearly will follow from Lemma 1 that L is well defined. The proof of Lemma 1 will be based on the proof of a quite different result:

Lemma 2. Any orientation preserving diffeomorphism f of R" is smoothly isotopic to the identity.
(In contrast, for many values of m there exists an orientation prcserving diffeomorphism of the sphere S'" which is not smoothly isotopic to the identity. See [20, p. 4041.)

with fo = identity, fl = f, and f,(O) = 0 for all t. Let U, denote the vector field dft o v of;' on f,(U), which corresponds to v on U. These vector fields are all defined and nonzero on a sufficiently small sphere centered at 0. Hence the index of v = vo at 0 must be equal to the index of U' = U, at 0. This proves Lemma 1 for orientation preserving diff eomorphisms. To consider diffeomorphisms which reverse orientation it is sufficient to consider the special case of a reflection p. Then
v' = p ov
0

p-l,

so the associated function e'(x)

=

v'(x)/l lv'(z)lj on the +sphere satisfies
0 p-1.

5' = p 0 0

P

R OO F

. We may assume that f(0)
dfo(x)
=

=

0. Since the derivative at 0

can be defined by lim f(tz)/t,
t-0

Evidently the degree of 8' equals the degree of fi, which completes the proof of Lemma 1. We will study the following classical result: Let M be A compact manifold and w a smooth vector field on M with isolated zeros. If M has a boundary, then w is required to point outward at all boundary points. of Poincare-Hopf Theorem. The sum CL the indices at the zeros of such a vector jield is equal to the Euler number*

it is natural to define an isotopy

F : R" X [O, 11

.+

R"

by the formula
F(x, t) F(x, 0) f(x)
= =

f(tx)/t for 0
dfo(x>.


1,

x(M)

m

=
,=O

(-1)'

rank H,(M).

To prove that F is smooth, even as t
where g,,

-.

=

x1g1(x)

,

+- +
**

.+

0, we write f in the form*
xmgm(x),

In particular this index sum is a topological invariant of M: it does not depend on the particular choice of vector jield.
(A 2-dimensional version of this theorem was proved by Poincare in 1885. The full theorem was proved by Hopf [14] in 1926 after earlier partial results by Brouwer and Hadamard.) We will prove part of this theorem, and sketch a proof of the rest. First consider the special case of a compact domain in R"'. Let X C R" be a compact m-manifold with boundary. The Gauss mapping
9:

gm are suitable smooth functions, and note that

F(xi t)

=

Xlgl(tx)

+

*.*

+

xmgm(tx)

for all values of 1. Thus f is isotopic to the linear mapping dfn, which is clearly isotoyi(* to the identity. This proves Lemma 2. LEMMA 1. We may assume that z = f(x) = 0 and that U is convex. If f preserves orientation, then, proceeding exactly as above,
ROOF O F

P

ax

.+

sm-'

assigns to each x

E

aX the outward unit normal vector at x.

* See for example

[22, p. 51.

* Here Hi(M) denotes the i - th homology group of M. This will be our first and last reference to homology theory.


36

$6. Vector .fields

The index sum

37

Lemma 3 (Hopf). zeros, and if v points is equal to the degree c does not depend

If v : X + R" is a smooth vector Jield with isolated out of X along the boundary, then the index sum L of the Gauss inapping from dX to S"-'. In particular, on the choice of U.

in terms of the derivatives of v at z. Consider first a vector field U on an open set U C R"' and think of v as a mapping U -+ R", so that dv, : R"' + R"' is defined.

For example, if a vector field on the disk D" points outward along the boundary, then L = +l. (Compare Figure 13.)

c

DEFINITION. The vector field v is nondegenerate at z if the linear transformation dv, is nonsingular. It follows that z is an isolated zero. Lemma 4. The index of v at a nondegenerate zero z is either according as the determinant of dv, is positive or negative.

+

1 or - 1

PROOF. Think of v as a diffeomorphism from some convex neighborhood U, of z into R". We may assume that z = 0. If U preserves orientation, we have seen that vlUo can be deformed smoothly into the identity without introducing any new zeros. (See Lemmas 1, 2.) Hence the index is certainly equal to +l. If v reverses orientation, then similarly v can be deformed into a reflection; hence L = -1.
More generally consider a zero z of a vector field w on a manifold A l C Rk. Think of w as a map from llf to Rk so that the dcrivative dw, : TM. -+ Rk is defined.

Figure IS. An example with index sum +1

PROOF. Removing an -ball around each zero, we obtain a new manifold with boundary. The function 8(x) = v(z)/ilv(x)ll maps this manifold into S"-'. Hence the sum of the degrees of 8 restricted to the various boundary components is zero. But 6 I aX is homotopic to 9, and the degrees on the other boundary components add up to -E L. (The minus sign occurs since each small sphere gets the wrong orientation.) Therefore
deg(g) as required. of dX, since it can be expressed as a constant times the integral over aX of the Gaussian curvature. This integer is of course equal to the Euler number of X . For tn odd it is equal to half the Euler number of aX. Before extending this result to other manifolds, some more preliminaries are needed. It is natural to try to compute the index of a vector field v at a zero z
L

Lemma 5. The derivative dw, actually carries TM, into the subspace TM, C Rk, hence can be cmsidered as a linear transformation from and TM, to itself. If this linear transformation has determinant D # 0 then z is an isolated zero of w with index equal to 1 or - 1 according as D is positive or negative.

+

PROOF. Let h U 4 A I be a parametrization of some neighborhood of z. Let e' denote the i-th basis vector of R" and let
so that the vectors i', * , t" form a basis for the tangent space TllIhc,) . We must compute the image of t' = t'(u) under the linear transformation dw,,(,,, note that First

=

0

-

tt

=

dh,(e')

=

ah/au,

R

EMARK

. The degree of g is also known as the "curvatura integra"

.

dwh(-)(t') = d(w 0 h),(e') = aw(h(u))/au,. 1) v,e' be the vector field on U which corresponds to the vector Let = field w on M. By definition U = dh-'~ w 0 h, SO that w(h(u)) = dh,(v) =
Therefore
2)

Vati-

aw(h(u))/au, =

E:

(av,/au,)t'

+ c:v,(at'/azc,).


38

$6. Vector jields
U,

The index sum

39

Combining 1) and 2), and then evaluating at the zero h-'(z) of obtain the formula
3)

we

We will also consider the squared distance function
p(x) =
112

dwz(t*)

=

xi(dv,/au,)t'.

- r(2)1j2.
p

Thus dw, maps TM, into itself, and the determinant L) of this linear transformation TM, + TAI, is equal to the determinant of the matrix (av,/au,). Together with Lemma 4 this completes the proof. Now consider a compact, boundaryless manifold M C Rk.Let N, denote the closed eneighborhood of M(i.e., the set of all x E Rk with 115 - yII 5 for some y E M). For E sufficiently small one can show that N. is a smooth manifold with boundary. (See $8, Problem 11.)

An easy computation shows that the gradient of grad
p=

is given by the outward

2(x - r(z)).
= p-'(cz),

Hence, for each point 2 of the level surface dN. unit normal vector is given by

g(z)
Extend
U

=

grad

p/l

lgrad

p[ =

I

(z - r(z))/e.

to a vector field w on the neighborhood N, by setting
w(z)
=

Theorem 1. For any vector field v on M with only nondegenerate zeros,
the index sum
L

(x - r(z))

+

u(44).

is equal to the degree of the Gauss mapping* g : dN,
-+

Sk-'.

In particular this sum does not depend on the choice of vector jeld.

Then w points outward along the boundary, since the inner product w(x)-g(x) is equal to e > 0. Kote that w can vanish only at the zeros of v in M: this is clear since the two summands (x - r(x)) arid u(r(x)) are mutually orthogonal. Computing the derivative of w :it a zero z E AI, we see that
dw,(h) dw,(h)
= =

PROOF. For x E N, let ~ ( x E)M denote the closest point of AI. (Compurc $8, Problem 12.) Note that the vector x - r(x) is perpendicular to the tangent space of ill at ~(x), otherwise ~(x) for would not be the closest point of M. If E is sufficiently small, then the function r(z) is smooth a11d well defined.

dv,(h) for all h E TA1, h

for h E TM:.

Thus the determinant of dw, is equal to the determinant of dv,. Hence the index of U' at the zero z is equal to the index L of v at x. Kow according to Lemma 3 the index sum L is equal to the degree of g. This proves Theorem 1.

EXAMPLES. On the sphere S" there exists a vector field v which points "north" at every point.* At the south pole the vectors radiate outward; hence the index is 1. At the north pole the vectors converge inward; hence the index is ( - l)m. Thus the invariant L is equal to 0 or 2 according as m is odd or even. This gives a new proof that every vector field on an even sphere has a zero. For any odd-dimensional, boundaryless manifold the invariant L is zero. For if the vector field v is replaced by -U, then each index is and multiplied by ( - l)m, the equality

+

Figure 14. The eneighborhood of M

c

L

=

(-1)"

c

1,

* A different interpretation of this degree has been given by Allendoerfer and Fenchel: the degree of g can be expressed as the integral over M of a suitable curvature scalar, thus yielding an m-dimensional version of the classical Gauss-Bonnet theorem. (References [l], [9]. See also Chern [S].)

for

172

odd, implies that

L=

0

* For example, v can be defined by the formula u(z) = p - (p.z)z, where p is the north pole. (See Figure 11.)


40

56. Vector fields

The index sum

41

L = 0 on a connected manifold M, then a theorem REMARK. If of Hopf asserts that there exists a vector field on M with no zeros at all.

In order to obtain the full strength of the PoincarBHopf theorem, three further steps are needed.
L with the Euler number x(M). STEP 1. Identification of the invariant It is sufficient to construct just one example of a nondegenerate vector i equal to x(M). The most pleasant way of doing field on M with

but only a C'-manifold. The extension w, if defined as before by w(z) = v(r(z)) z - r(z), will only be a continuous vector field near aM. The argument can nonetheless be carried out either by showing that our strong differentiability assumptions are not really necessary or by other methods.

+

this is the following: According to Marston Morse, it is always possible to find a real valued function on M whose ('gradient" is a nondegenerate vector field. Furthermore, Morse showed that the sum of indices associated with such a gradient field is equal to the Euler number of M. For details of this argument the reader is referred to Milnor [22, pp. 29, 361.

STEP 2. Proving the theorem /or a vector Jield with degenerate zeros. Consider first a vector field v on an open set U with an isolated zero at z. If x : U--+ [O, 11
takes the value 1 on a small neighborhood N, of z and the value 0 outside a slightly larger neighborhood N, and if y is a sufficiently small regular value of U, then the vector field

d(z)

= U(.)

- X(x)y

is nondegenerate* within N. The sum of the indices at the zeros within N can be evaluated as the degree of the map
0 : aN + X"-', and hence does not change during this alteration. More generally consider vector fields on a compact manifold M. Applying this argument locally we see that any vector .field with isolated zeros can be replaced by a nondegenerate vector jield without altering the 1. integer

C

STEP 3. Manifolds with boundary. If M C Rk has a boundary, then any vector field v which points outward along aM can again be extended over the neighborhood N, so as to point outward along aN,. However, there is some difficulty with smoothness around the boundary of M. Thus N, is not a smooth (i.e. differentiable of class Cm) manifold,

* Clearly d is nondegenerate within NI. But if y is sufficiently small, then v' will have no zeros at all within N - NI.


The Pontryagin cmstruction

@

87. FRAMED COBORDISM

MxO

MX I

M X2

THE PONTRYAGIN CONSTRUCTION

Figure 15.:Pastiny together two eobordisms within M

Figure 16.) The pair (N, b) is called a framed submanifold of M. Two framed submanifolds (N, b) and (N', b) are framed cobordant if there exists a cobordism X C A1 x [0, 11 between N and N' and a framing U of X , so that

THE degree of a mapping M -+ M' is defined only when the manifolds M and M' are oriented and have the same dimension. We will study a generalization, due to Pontryagin, which is defined for a smooth map
f:M--+S"
from an arbitrary compact, boundaryless manifold to a sphere. First some definitions. Let N and N' be compact n-dimensional submanifolds of M with aN = aN' = aM = 0. difference of dimensions m - n is called The the codimension of the submanifolds.

u'(x, t)
uE(z,t)

= (U'(.), =

0) for

(2,

t) E N X [0,

E)

(w'(x), for (z, t) E N' X (1 0)

E,

I].

Again this is an equivalence relation. Now consider a smooth map f : AP + S" and a regular value y E S". The map f induces a framing of the manifold f-'(y) as follows: Choose a positively oriented basis b = (U', . . . , U") for the tangent space T(SY)v. For each x E f-'(y) recall from page 12 that

dfb : TAP,

-+

z'(sp)y

DEFINITION. N is cobordant to N' within M if the subset

maps the subspace Tf-'(y)=to zero and maps its orthogonal complement Tf-'(y)t isoinorphically onto T(Sv)II. Hence there is a unique vector

N X [0, E) U N' X (1 -

E,

I]

w'(x)

E

Tj-'(y):

c

TM,

of M X [0, 11 can be extended to a compact manifold

X C M X 10, 11
so that

that maps into U' under df,. It will be convenient to use the notation tm = f*b for the resulting framing w'(x), . . . , ~"(x) ff'(y). of

ax

=

N

x

OUN'

x

1,

and so that X does not intersect M X 0 U M X 1 except at the points of ax. Clearly cobordism is an equivalence relation. (See Figure 15.)

DEFINITION. This framed manifold (f-'(y), f*b) will be called the Pontryagin manifold associated with f . Of course f has many Pontryagin manifolds, corresponding to different choices of y and b, but they all belong to a single framed cobordism class:

DEFINITION. A framing of the submanifold N C M is a smooth function t, which assigns to each x E N a basis
b(x)
= (U'(.),

F

F

Theorem A. If y' is another regular value off and b' is a positively then the framed manifold (f-'(y'), f*d) is oriented basis for T(SI))III, framed cobordant to (f-'(y), f*b). Theorem B. Two mappings from M to s" are smoothly homotopic if and only if the associated Pontryagin inanifolds are framed cobordant.

... ,

vrn-'(x))

for the space TN: C TM, of normal vectors to N in M at x. (See


44

$7. Framed cobordisin

The Pontryagin construction

45

can be identified with the space GL'(p, R) of matrices with positive determinant, and hence is connected. Such a path gives rise to the required framing of the cobordism f-'(y) X [0, I]. By abuse of notation we will often delete reference to f*b and speak simply of " the framed manifold f-'(y)."

Lemma 2. If y is U regular value off, and z is suficiently close to y, then f-'(z) is framed cobordant to f-'(y). . Since the set f(C) of critical 0 so that the e-neighborhood of Given z with I[z - yI( < e, choose of rotations (i.e. an isotopy) rr : S" --f
e

P

>

ROOF

values is conipact, we can y contains only regular a smooth one-parameter S" so that rl(y) = z, and

choose values. family so that

1) rt is the identity for 0 5 t < E', 2) rt equals r1 for 1 - e' < 1 5 1, and 3) each r;'(z) lies on the great circle from y to z, and hence is a regular value of f.

Define the homotopy

F : M X [0, 11

--f

Sv

by F(z, 1) = r,f(z). For each t note that z is a regular value of the composition rt 0 \ : ill -+ S".

It follows a fortiori that
Figure 16. Framed submanifolds and a framed cobordism

is a regular value for the mapping F. Herice

F-'(z) C M X 10, 11
is a framed manifold and provides a framed cobordism between the = f-'(y). This framed manifolds f-'(z) and (rl o f)-'(z) = f-'rF'(z) proves Lemma 2.

Theorem C. Any compact /rained suhmanifold (N, h~) codimension p oj in M occurs as Pontryagin manifold for some smooth mapping f : M S".
--f

Thus the homotopy classes of maps are in one-one correspondence with the framed cobordism classes of submanifolds. The proof of Theorem A will be very similar to the arguments in $04 and 5. It will be based on three lemmas.

Lemma 3. If f and g are smoothly homotopic and y is a regular value for both, then f-'(y) is framed cobordant to g-l(y). P
ROOF

. Choose a homotopy F with
F(z, t)
=

Lemma 1. If b and b' are two dij'erent positively oriented bases at y, then the Pontryagin manifold (f-'(y), f*b) is framed cobordant to (f-'(y),
f*b').

PROOF. Choose a smooth path from b to b' in the space of all positively oriented bases for T(S"),. This is possible since this space of bases

is framed cobordant to f-'(y) and so that g-'(z) is framed cobordant to g-'(y). Then F-'(z) is a framed manifold and provides a framed cobordism between f-'(z) and g-'(z). This proves Lemma 3.

f(z) 0 5 t < e, 1 - e < t I 1. F(z, t) = g(z) Choose a regular value z for F which is close enough to y so that f-'(z)


46

$7. Framed cobordism

The Pontryagin construction

47

PROOF THEOREM A. Given any two regular values y and z for f, OF we can choose rotations rt :SP-+S"
so that ro is the identity and rl(y) hence f-'(z) is framed cobordant to
=

z. Thus f is homotopic to rl 0 f;
=

Since N is compact, we could choose a sequence of such pairs with x converging, say to xo, with x' converging to x;, and with U -+ 0 and U' --+ 0. Then clearly xo = x;, and we have contradicted the statement 0). that g is one-one in a neighborhood of (xo, Thus g maps N X U, dlffeomorphically onto an open set. But U, is diffeomorphic to the full euclidean space R" under the correspondence
u-+u/(l - llui12/e2).

(rl

0

f)-'(z)

=

f-'r;'(z)

f-'(y).

This completes the proof of Theorem A. The proof of Theorem C will be based on the following: Let N C M be a framed submanifold of codimension p with framing b. Assume that N is compact and that aN = aM = @.

REMARK. Product neighborhoods do not exist for arbitrary subnianifolds. (Compare Figure 17.)

The reader who is familiar with geodesics will have no difficulty in checking that g:NXU,-+M is well defined and smooth, for e sufficiently small. The remainder of the proof proceeds exactly as before.

PROOF OF THEOREM C. Let N C AI be a compact, boundaryless, framed submanifold. Choose a product representation

Figure 17. An unJramable submanifold

g:NXIZ"-+VCM for a neighborhood V of N, as above, and define the projection
T:

V-+R"

PROOF. First suppose that M is the euclidean space R"'". Consider the mapping g : N X R" -+ M, defined by
g(x; t,, . . * , t")
=

by s(g(x, y)) = y. (See Figure 18.) Clearly 0 is a regular value, and n-'(O)is precisely N with its given framing.

x

+

t1U1(X)

+

**

.

+

t"U"(X).

Clearly dg(z;o,... is nonsingular; hence g maps some neighborhood ,o) of (2, 0) E N X R" diffeomorphically onto an open set. We will prove that g is one-one on the entire neighborhood N X U, of N X 0, providing that e > 0 is sufficiently small; where U, denotes the e-neighborhood of 0 in R". For otherwise there would exist pairs (z, ) # (x', U') in N X R" with llull and I/u'I( U arbitrarily small and with
!&l

N
Figure 18. Constructing a map with given Pontryagin manifold

U> =

dx',

U').

1

f

f

Product Neighborhood Theorem. Some neighborhood of N in M is diffeomorphic to the product N x R". Furthermore the diffeomorphism can be chosen so that each x E N corresponds to (2, 0) E N x R" and so that each normal frame ~ ( x corresponds to the standard basis for R". )

Since g(x, 0) = x, and since dg,,,,, does what is expected of it, this proves the Product Neighborhood Theorem for the special case M = E'". For the general case it is necessary to replace straight lines in R"'" by geodesics in M. More precisely let g(z; t,, . . . , t,) be the endpoirit of the geodesic segment of length I ltld (x) *. t,u"(x)I I in M which starts at x with the initial velocity vector
tlul(x)

+

.

+

t,U"(X)/~

It#'(X)

+ +

+

**

+

tgV(x)I

1.


@

$?. Framed cobordisrn

The Pontryagin construction

49

Now choose a smooth map cp : R" 3 x" which maps every x with 11x1 2 1into a base point so, and maps the open unit ball in Rpdiffeomorphically* onto 8" - so. Define

I

g(V) do not contain the antipode fj of y. Identifying V with N X R" and identifying S" - 5 with R", we obtain corresponding mappings

f:M-+S"
by
f(z) =

F, G : N X R"
with

--f

R",

F-`(O)

=

G-'(0) = N X 0,

&(x))
so

for x

E

V

and with
dF,,,,,
=

f(x>

=

for x # V.

dG(,,o, = (projection to R")

Clearly f is smooth, and the point cp(0) is a regular value of f. Since the corresponding Pontryagin manifold
f-l(P(0))
=

for all 5 E N. We will first find a constant c so that

n-'(O>

is precisely equal to the framed manifold N, this completes the proof of Theorem C.
In order to prove Theorem B we must first show that the Pontryagin S" manifold of a map determines its homotopy class. Let f, g : M be smooth maps with a common regular value y.
--f

F(z,u).u

>

0,

G(z,u).u

>

0
U)

for z E N and 0 < llul[ < c. That is, the points F(z, U ) and G(x, belong to the same open half-space in R". So the honiotopy (1 - t)F(z,U)

+

tG(x,

U)

Lemma 4. If the framed manifold (f-'(y), f*b) is equal to (g-'(y), g*b), then f is smoothly homotopic to g.
PROOF. It will be that f*b = g*b means First suppose that neighborhood V of N. Then the homotopy
convenient to set N = f-l(y). The hypothesis that df, = dg, for all 2 E N. f actually coincides with g throughout an entire Let h : S" - y -+ R" be stereographic projection. for x
e

bet,weeiiF and G will not map any new points into 0, at least for I Iu[ I By Taylor's theorem
IIF(z,u)

<

c.

-

~ l I l l ~ l l ~ for l c1 ,
4u).uI I c1

llull I 1.

Hence
I(F(.,
I/U113

and

w, 0

F(z,u).u 2

llU/j2

-

c1

[lull3> 0
-+

=

f(4

F(x, t)

=

h-'[t*h(f(x)) (1 - t).h(g(x))] for x E M - N

+

V

proves that f is smoothly homotopic to g. Thus is suffices to deform f so that it coincides with g in some small neighborhood of N, being careful not to map any new points into y during the deformation. Choose a product representation

for 0 < \lull < Min(c;', l),with a similar inequality for G. To avoid moving distant points we select a smooth map X : Rp with
X(U) X(U)
= =

R

1 for

llull 5 c/2
ljull 2 c.

0 for

Now the honiotopy
Ft(z,
U) =

N X R"

+

V CM

[l - X(U)~]F(Z, U)

for a neighborhood V of N, where V is small enough so that f(V)and

+

X(U)~G(X, U)

* For example, p(z) = h-l(z/A(\\z1\2)),where h is the stereographic projection from SO and where X is a smooth monotone decreasing function with X(t) > 0 for t < 1 and X(t) = 0 for 1 2 1.

deforms F = Fo into a mapping F, that (1) coincides with G in the region \(U(( < c/2, (2) coincides with F for (/U\( 2 c, and (3) has no new zeros. Making a corresponding deformation of the original mapping f, this clearly completes the proof of Lemma 4.


50

$7. Framed cobordism

The Hopf theorem

61

PROOF F THEOREM B. If f and g are smoothly homotopic, then O Lemma 3 asserts that the Pontryagin manifolds ff'(y) and g-l(y) are framed cobordant. Conversely, given a framed cobordism (X, r~) between f-'(y) and 9-l (y), an argument completely analogous to the proof of Theorem C constructs a homotopy
F : M X [0,1]-+x"

Theorem of Hopf. If M is connected, oriented, and boundaryless, then two maps M -+ S"' are smoothly homotopic if and only if they have the same degree.
On the other hand, suppose that M is not orientable. Then given a basis for TM, we can slide x around M in a closed loop so as to transform the given basis into one of opposite orientation. An easy argument then proves the following:

whose Pontryagin manifold (F-'(y), F 4 ) is precisely equal to (X, m). Setting F,(z) = F(z, t), note that the maps F, and f have exactly the same Pontryagin manifold. Hence F, f by Lemma 4;and similarly F1 g. Therefore f g, which completes the proof of Theorem B.

-

-

-

Theorem. If M is cmnected but nonorientable, then two maps M are homotopic if and only if they have the same mod 2 degree.

-+

S'"

.4

REMARKS. Theorems A, B, and C can easily be generalized so as to apply to a manifold M with boundary. The essential idea is to consider only mappings which carry the boundary into a base point so. The homotopy classes of such mappings
(AI,

The theory of framed cobordisni was iiitroduced by Pontryagiii in order to study homotopy classes of mappings

S"

-+

x"

am
+

--$

(S",so)

are in one-one correspondence with the cobordisni classes of franicd submanifolds

N C Interior(M)
of codimension p. If p 2 am 1, then this set of homotopy classes can be given the structure of an abelian group, called the p-th cohomotopy group ?rP(Ml dM). The composition operation in ?rp(M, dM) corresponds to the union operation for disjoint framed submanifolds of Interior (M). (Compare $8, Problem 17.)

THE HOPF THEOREM

with m > p. For example if m = p 1 2 4, there are precisely two hoiiiotopy classes of mappings S"' + S". Pontryagin proved this result by classifying framed 1-manifolds in S"'. With considerably more difficulty he was able to show that there are just two homotopy classes also in the case nz = p 2 2 4, using framed 2-nianifolds. However, for m - p > 2 this approach to the problem runs into manifold difficulties. It has since turned out to be easier to eriunierate homotopy classes of mappings by quite different, more algebraic methods.* Pontryagin's construction is, however, a double-edged tool. It not only allows us to translate information about manifolds into homotopy theory; it conversely enables us to translate any information about homotopy into manifold theory. Some of the deepest work in modern topology has come from the interplay of these two theories. Ren6 Thorn's work on cobordism is an important example of this. (References [36], [all.)

+

+

*
As an example, let M be a connected and oriented manifold of dimension m = p. A framed submanifold of codimension p is just a finite set of points with a preferred basis at each. Let sgn(z) equal +1 or - 1 according as the preferred basis determines the right or wrong orientation. Then sgn(x) is clearly equal to the degree of the associated map M -+ S"'. But it is not difficult to see that the framed cobordisni class of the 0-manifold is uniquely determined by this integer sgn(z). Thus we have proved the following.

See for example S.-T. Hu, Homotopy Theory.


Exercises
manifold and that the tangent space

53

88.

EXERCISES

Tr(z,v, TK x TN, c is equal to the graph of the linear map df,.

P

ROBLEM

10. Given M

C Rk, show

that the tangent bundle space

TAI

=

{(x,U) E AI X Rk 1 v E Tillz)
-+

is also a smooth manifold. Show that any smooth map f : M gives rise to a smooth map
df : TM
-+

N

TN
(dg) o (df).

where

I

N CONCLUSION

here are some problems for the reader.

d(identity)

=

identity, d(g 0 f)

=

PROBLEM 1. Show that the degree of a composition g 0 f is equal to the product (degree g) (degree f). PROBLEM 2. Show that every complex polynomial of degree n gives rise to a smooth map from the Gauss sphere S2 to itself of degree n. PROBLEM 3. If two maps f and g from X to x"satisfy I If(x) - g(x)11 < 2 for all x, prove that f is homotopic to g, the homotopy being smooth if f and g are smooth. PROBLEM 4.If X is compact, show that every continuous map X -+ S" can be uniformly approximated by a smooth map. If two smooth maps X -+ x" are continuously homotopic, show that they are smoothly homotopic.
PROBLEM 5. If m to a constant.

P

ROBLEM

11. Similarly show that the normal bundle space

E= ((~,~~)rnIXRk~~~TTnI,) is a smooth manifold. If M is compact and boundaryless, show that the correspondence
(x,4 H z from E to Rkmaps the E-neighborhood of M X 0 in E diffeomorphically (Compare the Product Neighboronto the eneighborhood N, of M in Rk. hood Theorem in $7.)

+

<

p, show that every map fiI"

--$

S" is homotopic
S" with degree

PROBLEM 12. Define r : N, -+ M by r(x v) = 2. Show that r(z v) is closer to x U than any other point of M. Using this retraction r, prove the analogue of Problem 4 in which the sphere S" is replaced by a manifold M.

+

+

+

P
-+

ROBLEM

13. Given disjoint manifolds M, N
X :M X N - +

PROBLEM 6. (Brouwer). Show that any map S" different from ( - l)"+' must have a fixed point.

c

R'", the linking map

Sk

PROBLEM 7. Show that any map s" -+ S" of odd degree must carry some pair of antipodal points into a pair of antipodal points. PROBLEM 8. Given smooth manifolds M C Rk and N C R', show that the tangent space T(M x N)c,,,,is equal to TM, X TN,.

is defined by X(z, y) = (x - y)/llx - yII. If M and N are compact, n = k, then oriented, and boundaryless, with total dimension m the degree of X is called the linking number l(M, N). Prove that

+

l(N, M)

=

(-l)(m+l)(n+l)

w1

NI.

PROBLEM 9. The graph r of a smooth map f : M -+ N is defined to be the set of all (5, y) E M X N with f(z) = y. Show that I' is a smooth

If M bounds an oriented manifold X disjoint from N, prove that Z(M, N) = 0. Define the linking number for disjoint manifolds in the sphere X"+"".


54

$8. Exercises

PROBLEM THE HOPF NVARIANT. If y # z are regular values for a 14, I map f : S"-' 4 S", then the manifolds f-'(y), f-'(z) can be oriented as in $5; hence the linking number l(f-'(y), f-'(z)) is defined. a) Prove that this linking number is locally constant as a function of y. b) If y and z are regular values of g also, where

APPENDIX CLASSIFYING 1-MAN1FO LD S

Ilf(4 - 9(")1/ <
for all x, prove that
Kf1(Y), f-'(zN =
Vg-l(Y),

IIY

- 41
=

f-'W

G'(Y),

g-`(4).

c) Prove that l(f-'(y), f-'(z)) depends only on the honlotopy class of f, and does not depend on the choice of y and z. This integer H(f) = l(f-'(y), f-'(z)) is called the Hopf invariant of f. (Reference [15].)

PROBLEM 15. If the dimension p is odd, prove that H(f) a composition
S2P-l

=

0. For

W E WILL prove the following result, which has been assumed in the text. A brief discussion of the classification problem for higher dimensional manifolds will also be given.

1,8"4 8"

Theorem. Any smooth, connected 1 - dimensional manifold is difeomorphic either to the circle S` or to some interval of real numbers.
(An interval is a connected subset of R which is not a point. It may be finite or infinite; closed, open, or half-open.) Since any interval is diffeomorphic* either to [0, 11, (0, 11, or (0, 1), it follows that there are only four distinct connected 1-manifolds. The proof will make use of the concept of "arc-length." Let I be an interval. DEFINITION. A map f : I 4 M is a parametrization by arc-length if f maps I difleomorphically onto an open subset t of M, and if the "velocity vector" df,(l) E TM,,,, has unit length, for each s E I. Any given local parametrization I' -+ M can be transformed into a parametrization by arc-length by a straightforward change of variables.

prove that H(g 0 f) is equal to H(f) multiplied by the square of the degree of g. The Hopf Jibration a : S3--+ S2is defined by
a(zl,2
2, 2.3,

2,) =

h-'((.cl

+

ix2)/(x3

+

id)
i

where h denotes stereographic projection to the complex plane. Prove that H(a) = 1.

PROBLEM 16. Two submanifolds N and N` of Af are said to intersect transversally if, for each x E N f l N', the subspaces TN, and TNJ together generate TM,. (If n n' < in this means that N f l N' = @.) If N is a framed submanifold, prove that it can be deformed slightly so as to intersect a given N` transversally. Prove that the resulting intersection is a smooth manifold.

+

PROBLEM 17. Let IIp(M) denote the set of all framed cobordism classes of codimension p in M. Use the transverse intersection operation to define a correspondence If p 2 inz 1, use the disjoint union operation to make an abelian group. (Compare p. 50.)

Lemma. Let f : I M and g : J --$ M be parametrizations by arc-length. Then f(I) A g(J) has at most two components. Ifit has only one component, then f can be extended to a parametrization by arc-length of the union f(I) g(J). If it has two components, then M must be diffeomorphic to X'. U
--f

+

rIP(M)x rI*(M)3 rI"+"(M).

nP(M) into

* For example, use a diffeomorphism of the form
f(t) = a tanh (t)

+

b.

t

Thus I can have boundary points only if M has boundary points.


66

Appendix

Classifying I-manifolds

57

PROOF. Clearly 9-l 0 f maps some relatively open subset of I diffeomorphically onto a relatively open subset of J. Furthermore the derivative of g-' o f is equal to i l everywhere. Consider the graph r C I x J, consisting of all (s, t) with f(s) = g(t). Then J? is a closed subset of I X J made up of line segments of slope f1. Since r is closed and 9-l 0 f is locally a diffeoniorphism, these line segments cannot end in the interior of I x J, but must extend to the boundary. Since g-' o f is one-one and single valued, there can be at most one of these segments ending on each of the four edges of the rectangle I x J. Hence has at most two components. (See Figure 19.) Furthermore, if there are two components, the two must have the same slope.
'1

The image h(S'), being compact and open in AI, must be the entire manifold M. This proves the lemma.

PROOF O F CLASSIFICATION T length can be extended to one

HEOREM

. Any parametrization by arc-

f :I+M

which is maximal in the sense that f cannot be extended over any larger interval as a parametrization by arc-length: it is only necessary to extend f as far as possible to the left and then as far as possible to the right. If M is not diffeoniorphic to A", we will prove that f is onto, and hence is a diffeoniorphism. For if the open set f(I) were not all of M, there would be a limit point z of f(I) in M - f(I). Parametrizing a neighborhood of x by arc-length and applying the lemma, we would see that f can be extended over a larger interval. This contradicts the assumption that f is maximal and hence completes the proof. REMAHKS. For manifolds of higher dimension the classification problem becomes quite formidable. For 2-dimensional manifolds, a thorough exposition has been given by KerBkjArtb [17]. The study of %dimensional manifolds is very much a topic of current research. (See l'apakyrialtopoulos [as].) lcor coinpact nianifolds of dimension 2 4 the classification problem is actually unsolvable.* But for high dimensional simply connected manifolds there has been much progress in recent years, as exemplified by the work of Smale [31]and Wall [37].
* See Markov [19]; and also a forthcoming paper by Boone, Haken, and Poknaru in Fundamenta Mathematicae.

8

-I

A

U

Figure 19. Three of the possibilities for

r

If r is connected, then g-l o f extends to a linear map L : R Now f and g o L piece together to yield the required extension
F : I U L-'(J) + f(I) U g(J).

i
-+

R.

If has two components, with slope say 1, they must be arranged as in the left-hand rectangle of Figure 19. Translating the interval J = (y, p) if necessary, we may assume that y = c and 6 = d, so that
a

+

<

b l c < d < a h:S'-+M

Now setting 8 = 2?rt/(a - a), the required difieomorphism
is defined by the formula

h(cos 0, sin 0)

=
=

f(t)
g(t)

for a
for c

< t < d, < t < 8.


BIBLIOGRAPHY

THE

following is a nliscellaneous list consisting of original sources and of reconimended textbooks. For the reader who wishes to pursue the study of differential topology, let me recommend Milnor [22], Alunkres [25], and Pontryagin [as]. The survey articles [23] and [32] should also prove useful. For background knowledge in closely related fields, let me recommend Hilton and Wylie [ 1I], Hu [16], Lang [18], de Rham [29], Steenrod [34], and Sternberg [35]. illcndocrfcr, C. 13., "Tht. Ihlw numlw of a Ricmann m:inifold," .I nier. Jour. ilrlath. 62 (1!140), 243-248. ,4postol, T. M., Mathematical Analysis. Hcwling, Mass. : .itltlison-Wcslcy, 1057. Auslander, L., and R. MacKenzic, Introduction to DiJerentiable Jlanilolds. New York: McGraw-Hill, 1963. Brouwer, L. E. J., "Uber Abbildung von ?tlannigfaltigkeitcn", Jlath. Annalen 71 (1012), 97-115. Brown, A. U.,"Functional dependence," Trans. Amer. Math. Soc. 38 (1935), 379-394. (See Theorem 3-111.) Chern, S. S., "A simple intrinsic proof of the Gauss-Ronnet formula for closed Riemannian manifolds," Annals of ddath. 45 (1944), 747-752. Dieudonnb, J., Foundations of Modern Analysis. New York: Academic Press, 1960. Dubovickii, A. Ya., "On differentiable mappings of an n-dimensional cube into a k-dimensional cube," Mat. Sbornik N.S. 32 (74), (1953), 443-464. (In Russian.) Fenchel, W.,"On total curvatures of Riemannian manifolds," Jour. London Math. Soc. 15 (1940), 1.5-22. Goffman, C., Calculus of Several Variables. New York: Harper & Row, Hilton, P., and S.Wiylie, Hoinology Theory. Cambridge Univ. Press, 1960.
1965.


60

Bibliography

Bibliography

61

[12] Hirsch, M., "A proof of the nonretractibility of a cell onto its boundary," Proc. Amer. Math. SOC. 14 (1963), 364-365. [ 131 Hopf, H., "Abbildungsklassen n-dimensionaler Mannigfaltigkeiten," Math. Annalen 96 (1926), 209-224. ~ 4 1__ , "Vektorfelder in n-dimensionalen Mannigfaltigkeiten," illath. Annalen 96 (1926), 225-250. , ~ 5 1- "Uber die Abbildungen von Spharen auf Spharen niedrigerer Dimension," Fundamenta Mathematicae 25 (1935), 427-440. 1161 Hu, S.-T., Homotopy Theory. New York: Academic Press, 1959. 1171 KerBkjdrt6, U. v., Vorlesungen uber Topologie. Berlin: Springer, 1923. [IS] Lang, S., Introduction to Differentiable Manifolds. New York: Interscience, 1962. [19] Markov, A. A., ``Insolubility of the problem of homeomorphy," Proceedings Intern. Congress of Math. 1955, Cambridge Univ. Press, 1960, pp. 300-306. (In Russian.) [2O] Milnor, J., "On manifolds homeomorphic to the 7-sphere," dnnals of Math. 64 (1956), 399-405. [21] - "A survey of cobordism theory," L'Enseignement math. 8 (1962), , 16-23. 1221 - Morse Theory. (Annals Studies 51.) Princeton Univ. Press, 1963. , 1231 ----, "Differential topology," Lectures on Modern Mathematics, 11, etl. T. L. Saaty, New York: Wiley, 1964, pp. 165-183. [24] Morse, A. I'., "The behavior of a function on its critical set," Annals of Math. 40 (1939), 62-70. [25] Munkres, J. R., Elementary Differential Topology. (Annals Studies 54). Princeton Univ. Press, 1963. [26] Papakyriakopoulos, C. D., "The theory of three-dimensional manifolds since 1950," Proceedings Intern. Congress of Math. 1958, Cambridge Univ. Press, 1960, pp. 433-440. [27] Pontryagin, L. S., "A classification of continuous transformations of a complex into a sphere," Doklady Akad. Nauk. S.S.S.R. (Comptes Rendues) 19 (1938), 147-149. , [28] - "Smooth manifolds and their applications in homotopy theory," Amer. Math. Soc. Translations, Ser. 2, I1 (1959), 1-114. (Translated from Trudy Inst. Steklov 45 (1955).) [29] Rham, G. de, Varidte`s diffkrentiables. Paris: Hermann, 1955. [30] Sard, A., "The measure of the critical points of differentiable maps," Bull. Amer. Math. Soc. 48 (1942), 883-890. 1311 Smale, S., "Generalized PoincarB's conjecture in dimensions greater than four," Annals of Math. 74 (1961), 391-406. 1321 - "A survey of some recent developments in differential topology," , Bull. Amer. hlath. Soc. 69 (1963), 131-145. [33] Spivak, M., Calculus on Manifolds. New York: Benjamin, 1965. [34] Steenrod, N., The Topology of Fibre Bundles. Princeton Univ. Press, 1951.

[35] Sternberg, S., Lectures on Diflerential Geometry. New York: I'renticeHall, 1964. [36] Thom, R., "Quelques propriBtBs globales des vanet& difErentiables," Commentarii Math. ZeZvet. 28 (1954), 17-86. [37] Wall, C. T. C., "Classification of (n - 1)-connected 2n-manifolds," Annals of Math. 75 (1962), 163-189. [38] Whitney, H., "A function not constant on a connected set of critical points," Duke Math. Jour. 1 (1935), 514-517.


INDEX

PAGE-BARBOUR LECTURE SERIES

The Page-Barbour Lccture Foundation was foundcd in 1007 by a gift from Mrs. Thomas Nelson Page (n6e Barbour) and the Honorable Thomas Nelson Page for the purpose of bringing to the University of Virginia each session a series of lectures by an eminent person in some field of scholarly endeavor. The materials in this volume were presented by Professor John W. Milnor in December, 1963, as the forty-seventh series sponsored by the Foundation.

antipodal niap, 30, 52 boundary, 12 Brouwer, L. E. J., 35, 52 degree, 20, 28 fixed point theorem, 14 Brown, A. B., 10, 11 chain rule for derivatives, 3, 7 cobordism, 42, 43, 51 (dimension, 42 cohomotopy, 50 coordinate system, 1 corresponding vector fields, 32 critical point, 8, 11 critical value, 8, 11 degree of a map, 28 mod two, 20 derivative of a map, 2-7 diffeomorphism, 1 differential topology, 1, 59 dimension of a manifold, 1, 5, 7 disk, 13, 14 Euler number, 35, 36, 40 framed cobordism, 43 framed submanifold, 42, 44, 46 Fubini theorem, 17 fundamental theorem of algebra, 8 Gauss-Bonnet theorem, 38 Gauss mapping, 35, 38 half-space, 12 Hirsch, M., 14

homotopy, 20, 21, 52 Hopf, H., 31, 35, 36, 40, 51, 54 index (of a zcro of a vector field), 3234 index sum, 35-41 inverse function theorem, 4, 8 inward vector, 26 isotopy, 21, 22, 34 linking, 53 Morse, A. I>., 10 Morse, M., 40 nondegencrate zcro (of a vector field) 37, 40 normal bundle, 53 normal vectors, 11, 12, 42 orientation, 26 of a boundary, 27 of F-'(y), 2s outward vector, 26 parametrization, 1, 2 by arc-length, 55 Poincark, H., 35 Pontryagin, L., 16, 42 Pontryagin manifold, 43 product neighborhood theorem, 46 regular point, 7 regular value, 8, 11,13,14,20,27,40,43 Sard, A., 10, 16 smooth manifolds, 1 with boundary, 12


64
smooth manifolds (cont.) the classification problem, 57 of dimension zero, 2 of dimension one, 14, 55 oriented, 26 smooth maps (= smooth mappings), 1 sphere, 2

lnrrcz
stereographic projection, 9, 48 tangent bundle, 53 tangent space, 2-5 at a boundary point, 12, 26 tangent vector, 2 vector fields, 30, 32-41 Weierstrass approximation theorem, 13

-~
INDEX OF SYMBOLS
deg (f; Y), 27 df,, 2-7
so!,

ax,
!*a,

Dm,

14

12
43
8

#P(Y)l

12 1 *-1, 2 TM,, 2-5 IIXII, 14
Hm,
Rk,

1

TOPOLOGY FROM THE DIFFERENTIABLE VIEWPOINT was composed and printed by The William Byrd Press, Inc., Richmond, Virginia and bound by Russell Rutter Co., Inc., New York, N. Y. The types are Modern Number 8 and Century Schoolbook, and the paper is Oxford's Bookbuildcrs Plate. Design is by Edward Foss.