Chris Zin MTH 864 April 2008 THE HOPF DEGREE THEOREM

Abstract. This paper provides a proof of the Hopf Degree Theorem as stated in Chapter 3, Section 6 of [1]. The proof will be provided through a series of exercises and hints as laid out in this section, and as such, will we often make uncited reference to [1]. Also, the hints are riddled with errors so this paper will assist at clearing up these errors for the reader.

1. Introduction The Hopf Degree Theorem is quite ambitious. It fullls the goal of classifying all maps (up to homotopy) from any compact, connected, oriented k-manifold into S k by appealing to degree theory. That is to say, the homotopy class of any such map is given by a single integer. Remarkable, indeed! The theorem is quite fundamental to the subject of topology, and a version was proved by Hopf around 1925 [3, 2.2]. Hence, the theorem derives it name directly from the man who proved it, rather than just being named in honor of him. Such a powerful theorem undoubtedly comes in handy to all sorts of mathematicians. Of course, a theorem is mere conjecture without its proof, so to appease these mathematicians (and hopefully yourself!) let us plot a course through the turbulent ocean of discovery prior to docking at the harbor of the Hopf Degree Theorem. Before embarking, remember to bring your spyglass to catch all the details along the way, for as many a globetrotter would say, the voyage is more important than the destination. 1 First o, let us establish some useful preliminary results in a section which should be entitled, Better than Sams Preliminaries. 2. Preliminaries Exercise 1. Let f of Rk , and let x be closed ball centered the boundary of B. W (f, z) = 1 if f : U R be any smooth map dened on an open subset U a regular point, with f (x) = z. Let B be a suciently small at x, and dene f : B Rk to be the restriction of f to Prove that W (f, z) = +1 if f preserves orientation at x and reverses orientation at x.
k

Proof. For simplicity, take x = 0 = z, and set A = df 0 . By Taylors theorem we can write f (x) = Ax + (x), where (x)/|x| 0 as |x| 0. Dene ft (x) = Ax + t (x) for t [0, 1]. Then, ft is a homotopy of f0 (x) = Ax and f1 (x) = f (x). Since x is a regular point, A is surjective, and hence, an isomorphism. So, the image of the unit ball under A contains some closed ball of radius c > 0. Furthermore, the boundary of the unit ball (S k1 ) is mapped dieomorphically
1It should be noted that this part of the introduction is an homage to the writing style Guillemin and Pollack used throughout their book, as the author of this paper would not normally use such odd references. 1

THE HOPF DEGREE THEOREM

onto the boundary of the image, by Exercise 2, Chapter 2, Section 1. Thus, |Ax| > c x for all x S k1 . It then follows, from |A |x| | > c and by linearity, that |Ax| > c|x| for all x Rk \ {0}. Since (x)/|x| 0 as |x| 0, we can choose the radius of the ball B small (x)| c enough so that | |x| < 2 for all x B. Then ft (x) = 0 on B since 1 1 |ft (x)| |Ax| t| (x)| > c|x| c|x| = c|x|. 2 2 Therefore, we can dene Ft (x) =
Ax |Ax| f (x) |f (x)| . ft (x) |ft (x)|

: B I S k1 , a homotopy of F0 (x) =

and F1 (x) = Since they are homotopic, the degrees of these two maps are equal, which implies W (A, 0) = W (f, 0). Now we invoke the linear isotopy lemma of Chapter 3, Section 4, which is stated in the appendix. This tells us that A is homotopic to either the identity (if A preserves orientation) or the reection map (x1 , . . . , xk ) (x1 , x2 , . . . , xk ) (if A reverses orientation). So, if A preserves orientation, then W (A, 0) = +1, and if A reverses orientation, then W (A, 0) = 1. The result follows immediately. Essentially, Exercise 1 limits the way in which local dieomorphisms can wind around a point. In Exercise 2 we exploit this information to count preimages. Exercise 2. Let f : B Rk be a smooth map dened on some closed ball B in Rk . Suppose that z is a regular value of f that has no preimages on the boundary sphere B, and consider f : B Rk . Prove that the number of preimages of z, counted with our usual orientation convention, equals the winding number W (f, z). Proof. To begin we appeal to the stack of records theorem (Exercise 7, Chapter 1, Section 5, and stated in the appendix below). This tells us that f 1 (z) is a nite set {x1 , . . . , xn } so that we can circumscribe disjoint balls Bi around each xi . Since f 1 (z) is disjoint from B, we may also shrink these balls as necessary so that Bi B = and so that each Bi is suciently small in the sense of Exercise 1. Let fi = f |Bi . Then, note that by Exercise 1 the number of preimage points, n counted with our usual orientation convention, equals i=1 W (fi , z). n f (x)z Let B = B \ i=1 Bi and consider the map u : B S k1 , x |f (x)z| . This extends to all of B , since |f (x) z| = 0 on B . Thus, W (f |B , z) = deg(u) = 0, with the second inequality due to the proposition on page 110 of [1] (see appendix). n However, since B = B i=1 (Bi ) (where the indicates that each Bi inherits the opposite orientation it has when considered as a submanifold of B), we n conclude that W (f |B , z) = W (f, z) i=1 W (fi , z). Therefore W (f, z) = n i=1 W (fi , z). Exercise 3. Let B be a closed ball in Rk , and let f : Rk \Int(B) Y be any smooth map dened outside the open ball Int(B). Show that if the restriction f : B Y is homotopic to a constant, then f extends to a smooth map dened on all of Rk into Y . Proof. For simplicity, assume B is centered at 0. Note that this allows us to write every non-zero point x B uniquely as x = ty for some y B and some t [0, 1]. Let gt : B Y be a homotopy with g1 = f and g0 = constant. Now consider

THE HOPF DEGREE THEOREM

the function F : Rk Y dened by F (x) = f (x) if x Rk \ Int(B) . gt (y) if x B and x = ty for some y B and some t [0, 1]

F is well-dened on Rk \ Int(B), since f and gt agree on the overlap (namely, the points in B where we have f = f = g1 ). F is also well-dened on B \ {0} since each point in this set can be written uniquely as x = ty. Finally, F is well-dened at {0}, since g0 is a constant function. Furthermore, since the functions agree on the overlap, F is continuous by the gluing lemma (as it is often called). Now, using the standard trick of Exercise 1, Chapter 1, Section 6, we can perform a homotopy inside B, near B, in order to make F smooth on all of Rk (note that it is already smooth on Rk \ B). Rather than do this explicitly, we will speed things up and use a well-known approximation theorem which is not in Guillemin and Pollack (although this theorem is within the scope of the book). You may nd this as Whitney Approximation on Manifolds in [2, Theorem 10.21] and stated in the appendix below. It allows up to homotope F rel Rk \ Int(B) to a smooth function. Then F is a smooth extension of f from all of Rk to Y . 3. The Special Case Now that we have gotten some basic results out of the way, our next step will be to prove a special case of Hopfs theorem. The full theorem will derive from this special case. Special Case. Any smooth map f : S l S l having degree zero is homotopic to a constant map. Exercise 4. Check that the special case implies the following corollary. Corollary. Any smooth map f : S l Rl+1 \ {0} having winding number zero with respect to the origin is homotopic to a constant. Proof. By assumption, the degree of the map f /|f | is zero. So by the special case, f /|f | is homotopic to a constant. The map gt : S l Rl+1 \ {0}, x f tf (x) + (1 t) |f (x) , denes a homotopy with g0 = f /|f | and g1 = f . Since (x)| homotopy is transitive, f is homotopic to a constant. To prove the special case we will induct on dimension. Hopfs theorem, in the case of 1-dimensional manifolds, is established as Exercise 9, Chapter 3, Section 3, and this will suce as our base case. From here on, we will assume the special case is true in dimension l = k 1. In order to complete the proof we need the following exercise. Exercise 5. Let f : Rk Rk be a smooth map with 0 as a regular value. Suppose that f 1 (0) is nite and that the number of preimage points in f 1 (0) is zero when counted with the usual orientation convention. Assuming the special case in dimension k 1, prove that there exists a mapping g : Rk Rk \ {0} such that g = f outside a compact set. Proof. Since f 1 (0) is nite we can choose a ball B centered at the origin with f 1 (0) Int(B). The number of preimages is zero when counted with the usual orientation convention, by assumption, so Exercise 2 implies the map f : B

THE HOPF DEGREE THEOREM

Rk \ {0} has winding number zero. We know B is dieomorphic to S k1 , so f is a map from S k1 into Rk \ {0}. We are assuming the special case is true in dimension k 1, so the corollary applies. Thus, f is homotopic to a constant. At this point we wish to replace f with the map f |Rk \Int(B) : Rk \ Int(B) Rk \ {0} so that we can apply Exercise 3. Then f extends to a smooth map g : Rk Rk \ {0} with f = g outside the compact set B. Notice that f and g are homotopic via the linear homotopy tf + (1 t)g, and that outside of B this is constant. All of the diculty in the following proof will be showing that we can homotope f to miss a single point a of S k . Once we have done this, it is easy to see that f is homotopic to a constant map, for S k \ {a} is contractible. Exercise 6. Establish the special case in dimension k. Proof. Using Sards theorem, pick distinct regular values a and b for f . We use the stack of records theorem again to tell us that f 1 (a) = {a1 , . . . , an } and f 1 (b) = {b1 , . . . , bm } are nite. Thus, we can nd an open neighborhood U of a1 such that U is dieomorphic to Rk via a dieomorphism : Rk U and such that bi U for / all i. Note that we are assuming k > 1, since the base case is already established. This allows us to apply the corollary to the isotopy lemma of Chapter 3, Section 6 (stated in the appendix) to each ai f 1 (a) with i = 1 to move ai into U via a dieomorphism which is isotopic to the identity and compactly supported. Since we are only looking at f up to homotopy, this move does not concern us. Therefore, we have a neighborhood U of f 1 (a) such that U is dieomorphic to Rk and b f (U ). / Let : S k \ {b} Rk be a dieomorphism such that (a) = 0. Then f is a smooth map from Rk into Rk . Since a is a regular value of f , 0 is a regular value of f , and since f 1 (a) is nite so is ( f )1 (0). Since deg(f ) = 0, by assumption, the number of preimages of a is zero when counted with our usual orientation convention (as this is how we calculate degree when the image point is a regular value). It follows that the number of preimage points in ( f )1 (0) is zero when counted with the usual orientation convention. Thus, we can apply Exercise 5 to nd a map h : Rk Rk \ {0} such that h = f outside a compact set B and h is homotopic to f on all of Rk . Note then, that on U , f is homotopic to 1 h 1 , which is a map from U into S k \ {a, b} since ( 1 )1 {a} = {a} = {0} (h 1 )(U ). Also note that on U \ 1 (B) we have f = 1 h 1 , since h = f outside of B. This gives us a smooth map g : S k S k \ {a} that is homotopic to f (take g = f on S k \ 1 (B) and g = 1 h 1 on 1 (B)). Since S k \{a} is dieomorphic to Rk , it is contractible. Therefore, g is homotopic to a constant map, and it follows that f is as well. 4. The Hopf Degree Theorem With the special case in hand we may now set our sights toward the goal, which itself is essentially a special case of the following theorem. Extension Theorem. Let W be a compact, connected, oriented k + 1 dimensional manifold with boundary, and let f : W S k be a smooth map. Prove that f extends to a globally dened map F : W S k , with F = f , if and only if the degree of f is zero.

THE HOPF DEGREE THEOREM

Before proving the extension theorem we address the possibility of extending boundary maps whose codomains are Rk+1 . Exercise 7. Let W be any compact manifold with boundary, and let f : W Rk+1 be any smooth map whatsoever. Prove that f may be extended to all of W . Proof. By denition, W is a subset of RN for some N , and since W is compact, it is a closed subset. Similarly, W is a closed subset of RN . Since f is a smooth map dened on a closed subset of RN it may be locally extended to a smooth map on open sets. Since W is compact and boundaryless, we can use the -neighborhood theorem to extend f to a map F dened on a neighborhood U of W in RN . Choose any smooth function that is one on W and zero outside some compact subset of U . Then F can be dened on all of RN , with the stipulation that it is zero outside U . This is an extension of f to all of RN since on W we have F = 1 F = f . Then, F |W : W Rk+1 is an extension of f to all of W . Exercise 8. Prove the Extension Theorem. Proof. The sucient direction follows from an application of the proposition on page 110 of [1]. For the necessary direction we begin by extending f to a map F : W Rk+1 using Exercise 7. By the transversality extension theorem ([1, pg. 72], see appendix) we may assume 0 to be a regular value of F . For this, take Z = {0}, Y = Rk+1 , C = W , and X = W in the theorem and replace the f in the theorem with our F . The set F 1 (0) is nite (stack of records theorem), so we may use the corollary to the isotopy lemma to move F 1 (0) so that F 1 (0) Int(B) where B is a closed ball contained in Int(W ). Consider the map F = F |B : B Rk+1 \ {0}. Notice that F/|F | extends from W to W = W \ Int(B), since we moved F 1 (0) inside Int(B). Thus, W (F |W , 0) = deg(F/|F |) = 0. We also know W (F |W , 0) = W (f, 0) = deg(f ) = 0, by assumption. Since W = W (B), we have W (F |W , 0) = W (F |W , 0) W (F, 0) which implies W (F, 0) = 0. By the corollary to the special case, F is homotopic to a constant. Therefore, it follows from Exercise 3 that F |W extends to a map F : W Rk+1 \ {0}. Then F /|F | : W S k is the global extension of f we were looking for. Finally we have made it to the Hopf Degree Theorem. As important as it is, it is an almost trivial corollary of the extension theorem. Exercise 9. Conclude: The Hopf Degree Theorem. Two maps of a compact, connected, oriented kmanifold X into S k are homotopic if and only if they have the same degree. Proof. Let f0 and f1 be any two maps from X to S k and let W = X I. Dene a map f : W S k by setting f = f0 on X {0} and f = f1 on X {1}. By the extension theorem, f extends to a map on all of W if and only if deg(f ) = 0. By denition, such an extension would be a homotopy of f0 and f1 . Also, since W = (X {1}) (X {0}) it follows that deg(f ) = deg(f1 ) deg(f0 ). Thus, the extension theorem translates to: f0 and f1 are homotopic if and only if deg(f0 ) = deg(f1 ).

THE HOPF DEGREE THEOREM

5. Consequences Now that any doubt as to the validity of Hopfs theorem has been eliminated, we may explore some of its applications, the rst of which is accessible to anyone who has read far enough in [1] to understand the ideas presented above. This theorem is presented on page 146 of [1]. Existence of Nonvanishing Vector Fields. A compact, connected, oriented manifold X possesses a nonvanishing vector eld if and only if its Euler characteristic is zero. Proof. This can be proved using hints given in the book. An obvious application, which strays a bit from the subject of dierential topology (into the realm of algebraic topology), involves homotopy groups of spheres. These are, in general, notoriously dicult to compute, but Hopfs theorem gives us an easy way to do it in one very special case. The k th homotopy group of S k . k (S k ) = Z, generated by the identity map. Proof. Follows from Hopfs theorem by letting X = S k and xing basepoints. With a little knowledge of homotopy theory, this corollary can be used to provide an alternate proof (as opposed to the one on page 65 in [1]) that the unit k-disk does not retract onto its boundary which, in turn, can be used to prove the Brouwer xed-point theorem. Although the Hopf Degree Theorem is useful for applications, it is also interesting in its own right. You can examine the cases in dimensions one and two and you may nd, in fact, that these low dimensional cases are intuitively true. But as we have seen above, to actually prove it requires quite a bit of work, much like the JordanBrouwer separation theorem. Of course, with work comes insight, so perhaps it is not so terrible that these obvious statements are not so obvious afterall. 6. Appendix The following statements were used during the course of the above proof, and were added here to cater to the needs of the lazy reader who does not feel like opening a copy of Guillemin and Pollack (or more specically, to cater to the needs of the author who will be such a lazy reader if he decides to revisit this paper at some point in the future). Proofs of these statements can be found in [1], unless otherwise stated. Linear Isotopy Lemma. Suppose that E is a linear isomorphism of Rk that preserves orientation. Then there exists a homotopy Et consisting of linear isomorphisms, such that E0 = E and E1 is the identity. If E reverses orientation, then there exists such a homotopy with E1 equal to the reection map E1 (x1 , . . . , xk ) = (x1 , x2 , . . . , xk ). Proof. To be provided by the reader. See Exercise 1, Chapter 3, Section 4 of [1] for assistance. Stack of Records Theorem (Exercise 7, Chapter 1, Section 4). Suppose that y is a regular value of f : X Y , where X is compact and has the same dimension as Y . Then f 1 (y) is a nite set {x1 , . . . , xN } and there exists a neighborhood U

THE HOPF DEGREE THEOREM

of y in Y such that f 1 (U ) is a disjoint union V1 VN , where Vi is an open neighborhood of xi and f maps Vi dieomorphically onto U . Proof. To be provided by the reader. Proposition (on page 110 of [1]). Suppose that f : X Y is a smooth map of compact oriented manifolds having the same dimension and that X = W (W compact). If f can be extended to all of W , then deg(f ) = 0. Whitney Approximation on Manifolds. Let N and M be smooth manifolds, and let F : N M be a continuous map. Then F is homotopic to a smooth map F : N M . If F is smooth on a closed subset A N , then the homotopy can be taken to be relative to A. Proof. See theorem 10.21 in [2]. The 1-Dimensional Hopf Degree Theorem (Exercise 9, Chapter 3, Section 3). Prove that two maps of the circle S 1 into itself are homotopic if and only if they have the same degree. Proof. To be provided by the reader. Sards Theorem. If f : X Y is any smooth map of manifolds, then almost every point in Y is a regular value of f . (i.e. The set of critical values of a smooth map of manifolds f : X Y has measure zero). Isotopy Lemma. Given any two points y and z in the connected manifold Y , there exists a dieomorphism h : Y Y such that h(y) = z and h is isotopic to the identity. Moreover, the isotopy may be taken to be compactly supported. Corollary (to the Isotopy Lemma). Suppose that Y is a connected manifold of dimension greater than 1, and let y1 , . . . , yn and z1 , . . . , zn be two sets of distinct points in Y . Then there exists a dieomorphism h : Y Y , isotopic to the identity, with h(y1 ) = z1 , . . . , h(yn ) = zn . Moreover, the isotopy may be taken to be compactly supported. -Neighborhood Theorem. For a compact boundaryless manifold Y in RM and a positive number , let Y be the open set of points in RM with distance less than from Y . If is suciently small, then each point w Y possesses a unique closest point in Y , denoted (w). Moreover, the map : Y Y is a submersion. When Y is not compact, there still exists a submersion : Y Y that is the identity on Y , but now must be allowed to be a smooth positive function on Y , and Y is dened as {w RM | |w y| < (y) for some y Y }. Transversality Extension Theorem. Suppose that Z is a closed submanifold of Y , both boundaryless, and C is a closed subset of X. Let f : X Y be a smooth map with f Z on C and f Z on C X. Then there exists a smooth map g : X Y homotopic to f , such that g Z, g Z, and on a neighborhood of C we have g = f . References
[1] V. Guillemin, A. Pollack, Dierential Topology, Prentice-Hall, Englewood Clis, NJ, 1974. [2] J. Lee, Introduction to Smooth Manifolds, Springer, New York, NY, 2002. [3] A. Hatcher, Algebraic Topology, Cambridge University Press, New York, NY, 2001.