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

0, January 2003

Allen Hatcher

Copyright c (2003 by Allen Hatcher

Paper or electronic copies for noncommercial use may be made freely without explicit permission from the author.

All other rights reserved.

Table of Contents

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Chapter 1. Vector Bundles . . . . . . . . . . . . . . . . . . . . 4

1.1. Basic Deﬁnitions and Constructions . . . . . . . . . . . . 6

Sections 7. Direct Sums 9. Inner Products 11. Tensor Products 13.

Associated Fiber Bundles 15.

1.2. Classifying Vector Bundles . . . . . . . . . . . . . . . . . 18

Pullback Bundles 18. Clutching Functions 22. The Universal Bundle 27.

Cell Structures on Grassmannians 31. Appendix: Paracompactness 35.

Chapter 2. K–Theory . . . . . . . . . . . . . . . . . . . . . . . 38

2.1. The Functor K(X) . . . . . . . . . . . . . . . . . . . . . . . 39

Ring Structure 40. The Fundamental Product Theorem 41.

2.2. Bott Periodicity . . . . . . . . . . . . . . . . . . . . . . . . 51

Exact Sequences 51. Deducing Periodicity from the Product Theorem 53.

Extending to a Cohomology Theory 55. Elementary Applications 58.

2.3. Division Algebras and Parallelizable Spheres . . . . . . 59

H–Spaces 59. Adams Operations 62. The Splitting Principle 65.

2.4. Bott Periodicity in the Real Case [not yet written]

2.5. Vector Fields on Spheres [not yet written]

Chapter 3. Characteristic Classes . . . . . . . . . . . . . . 73

3.1. Stiefel-Whitney and Chern Classes . . . . . . . . . . . . 77

Axioms and Constructions 77. Cohomology of Grassmannians 84.

3.2. Euler and Pontryagin Classes . . . . . . . . . . . . . . . . 88

The Euler Class 91. Pontryagin Classes 94.

Chapter 4. The J–Homomorphism . . . . . . . . . . . . . . 98

4.1. Lower Bounds on Im J . . . . . . . . . . . . . . . . . . . . 99

The Chern Character 100. The e Invariant 102. Thom Spaces 103.

Bernoulli Denominators 106.

4.2. Upper Bounds on Im J [not yet written]

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

Preface

Topological K–theory, the ﬁrst generalized cohomology theory to be studied thor-

oughly, was introduced around 1960 by Atiyah and Hirzebruch, based on the Periodic-

ity Theorem of Bott proved just a few years earlier. In some respects K–theory is more

elementary than classical homology and cohomology, and it is also more powerful for

certain purposes. Some of the best-known applications of algebraic topology in the

twentieth century, such as the theorem of Bott and Milnor that there are no division

algebras after the Cayley octonions, or Adams’ theorem determining the maximum

number of linearly independent tangent vector ﬁelds on a sphere of arbitrary dimen-

sion, have relatively elementary proofs using K–theory, much simpler than the original

proofs using ordinary homology and cohomology.

The ﬁrst portion of this book takes these theorems as its goals, with an exposition

that should be accessible to bright undergraduates familiar with standard material in

linear algebra, abstract algebra, and point-set topology. Later chapters of the book

assume more, approximately the contents of a standard graduate course in algebraic

topology. A concrete goal of the later chapters is to tell the full story on the stable

J–homomorphism, which gives the ﬁrst level of depth in the stable homotopy groups

of spheres. Along the way various other topics related to vector bundles that are of

interest independent of K–theory are also developed, such as the characteristic classes

associated to the names Stiefel and Whitney, Chern, and Pontryagin.

Introduction

Everyone is familiar with the M¨obius band, the twisted product of a circle and a

line, as contrasted with an annulus which is the actual product of a circle and a line.

Vector bundles are the natural generalization of the M¨obius band and annulus, with

the circle replaced by an arbitrary topological space, called the base space of the vector

bundle, and the line replaced by a vector space of arbitrary ﬁnite dimension, called the

ﬁber of the vector bundle. Vector bundles thus combine topology with linear algebra,

and the study of vector bundles could be called Linear Algebraic Topology.

The only two vector bundles with base space a circle and one-dimensional ﬁber

are the M¨obius band and the annulus, but the classiﬁcation of all the different vector

bundles over a given base space with ﬁber of a given dimension is quite difﬁcult

in general. For example, when the base space is a high-dimensional sphere and the

dimension of the ﬁber is at least three, then the classiﬁcation is of the same order

of difﬁculty as the fundamental but still largely unsolved problem of computing the

homotopy groups of spheres.

In the absence of a full classiﬁcation of all the different vector bundles over a

given base space, there are two directions one can take to make some partial progress

on the problem. One can either look for invariants to distinguish at least some of the

different vector bundles, or one can look for a cruder classiﬁcation, using a weaker

equivalence relation than the natural notion of isomorphism for vector bundles. As it

happens, the latter approach is more elementary in terms of prerequisites, so let us

discuss this ﬁrst.

There is a natural direct sum operation for vector bundles over a ﬁxed base space

X, which in each ﬁber reduces just to direct sum of vector spaces. Using this, one can

obtain a weaker notion of isomorphism of vector bundles by deﬁning two vector bun-

dles over the same base space X to be stably isomorphic if they become isomorphic

after direct sum with product vector bundles X×R

n

for some n, perhaps different

n’s for the two given vector bundles. Then it turns out that the set of stable isomor-

phism classes of vector bundles over X forms an abelian group under the direct sum

operation, at least if X is compact Hausdorff. The traditional notation for this group

is

¯

KO(X). In the case of spheres the groups

¯

KO(S

n

) have the quite unexpected prop-

erty of being periodic in n. This is called Bott Periodicity, and the values of

¯

KO(S

n

)

are given by the following table:

n mod 8 1 2 3 4 5 6 7 8

¯

KO(S

n

) Z

2

Z

2

0 Z 0 0 0 Z

For example,

¯

KO(S

1

) is Z

2

, a cyclic group of order two, and a generator for this group

is the M¨obius bundle. This has order two since the direct sum of two copies of the

2 Introduction

M¨obius bundle is the product S

1

×R

2

, as one can see by embedding two M¨obius bands

in a solid torus so that they intersect orthogonally along the common core circle of

both bands, which is also the core circle of the solid torus.

Things become simpler if one passes fromthe real vector spaces to complex vector

spaces. The complex version of

¯

KO(X), called

`

K(X), is constructed in the same way

as

¯

KO(X) but using vector bundles whose ﬁbers are vector spaces over C rather than

R. The complex form of Bott Periodicity asserts simply that

`

K(S

n

) is Z for n even

and 0 for n odd, so the period is two rather than eight.

The groups

`

K(X) and

¯

KO(X) for varying X share certain formal properties with

the cohomology groups studied in classical algebraic topology. Using a more general

form of Bott periodicity, it is in fact possible to extend the groups

`

K(X) and

¯

KO(X)

to a full cohomology theory, families of abelian groups

`

K

n

(X) and

¯

KO

n

(X) for n ∈ Z

that are periodic in n of period two and eight, respectively. There is more algebraic

structure here than just the additive group structure, however. Tensor products of

vector spaces give rise to tensor products of vector bundles, which in turn give prod-

uct operations in both real and complex K–theory similar to cup product in ordinary

cohomology. Furthermore, exterior powers of vector spaces give natural operations

within K–theory.

With all this extra structure, K–theory becomes a powerful tool, in some ways

more powerful even than ordinary cohomology. The prime example of this is the very

simple proof, once the basic machinery of complex K–theory has been set up, of the

theorem that there are no ﬁnite dimensional division algebras over R in dimensions

other than 1, 2, 4, and 8, the dimensions of the classical examples of the real and

complex numbers, the quaternions, and the Cayley octonions. The same proof shows

also that the only spheres whose tangent bundles are product bundles are S

1

, S

3

, and

S

7

, the unit spheres in the complex numbers, quaternions, and octonions.

Another classical problem that can be solved more easily using K–theory than

ordinary cohomology is to ﬁnd the maximum number of linearly independent tangent

vector ﬁelds on the sphere S

n

. In this case complex K–theory is not enough, and the

added subtlety of real K–theory is needed. There is an algebraic construction of the

requisite number of vector ﬁelds using Clifford algebras, and the harder part is to

show there can be no more than this construction provides. Clifford algebras also

provide a nice explanation for the mysterious sequence of groups appearing the in

real form of Bott periodicity.

Now let us return to the original classiﬁcation problem for vector bundles over a

given base space and the question of ﬁnding invariants to distinguish different vector

bundles. The ﬁrst such invariant is orientability, the question of whether all the ﬁbers

can be coherently oriented. For example, the M¨obius bundle is not orientable since

as one goes all the way around the base circle, the orientation of the ﬁber lines is

reversed. This does not happen for the annulus, which is an orientable vector bundle.

Introduction 3

Orientability is measured by the ﬁrst of a sequence of cohomology classes associ-

ated to a vector bundle, called Stiefel-Whitney classes. The next Stiefel-Whitney class

measures a more reﬁned sort of orientability called a spin structure, and the higher

Stiefel-Whitney classes measure whether the vector bundle looks more and more like

a product vector bundle over succesively higher dimensional subspaces of the base

space. Cohomological invariants of vector bundles such as these with nice general

properties are known as characteristic classes. It turns out that Stiefel-Whitney classes

generate all characteristic classes for ordinary cohomology with Z

2

coefﬁcients, but

with Z coefﬁcients there are others, called Pontryagin and Euler classes, the latter

being related to the Euler characteristic. Although characteristic classes do not come

close to distinguishing all the different vector bundles over a given base space, except

in a few low dimensional cases, they have still proved themselves to be quite useful

objects.

To motivate the deﬁnition of a vector bundle let us consider tangent vectors to

the unit 2 sphere S

2

in R

3

. At each point x ∈ S

2

there is a tangent plane P

x

. This

is a 2 dimensional vector space with the point

x as its zero vector 0

x

. Vectors v

x

∈ P

x

are

thought of as arrows with their tail at x. If

we regard a vector v

x

in P

x

as a vector in R

3

,

then the standard convention in linear algebra

would be to identify v

x

with all its parallel

translates, and in particular with the unique

translate τ(v

x

) having its tail at the origin in

R

3

. The association v

x

τ(v

x

) deﬁnes a

τ

x

v

P

x

x

v

x

( )

function τ : TS

2

→R

3

where TS

2

is the set of all tangent vectors v

x

as x ranges

over S

2

. This function τ is surjective but certainly not injective, as every nonzero

vector in R

3

occurs as τ(v

x

) for inﬁnitely many x, in fact for all x in a great circle

in S

2

. Moreover τ(0

x

) = 0 for all x ∈ S

2

, so τ

−1

(0) is a whole sphere. On the other

hand, the function TS

2

→S

2

×R

3

, v

x

(x, τ(v

x

)), is injective, and can be used to

topologize TS

2

as a subspace of S

2

×R

3

, namely the subspace consisting of pairs

(x, v) with v orthogonal to x.

Thus TS

2

is ﬁrst of all a topological space, and secondly it is the disjoint union

of all the vector spaces P

x

for x ∈ S

2

. One can think of TS

2

as a continuous family

of vector spaces parametrized by points of S

2

.

The simplest continuous family of 2 dimensional vector spaces parametrized by

points of S

2

is of course the product S

2

×R

2

. Is that what TS

2

really is? More

precisely we can ask whether there is a homeomorphism h: TS

2

→S

2

×R

2

that takes

each plane P

x

to the plane ¦x¦×R

2

by a vector space isomorphism. If we had such

an h, then for each ﬁxed nonzero vector v ∈ R

2

the family of vectors v

x

= h

−1

(x, v)

would be a continuous ﬁeld of nonzero tangent vectors to S

2

. It is a classical theorem

in algebraic topology that no such vector ﬁeld exists. (See §2.2 for a proof using

Basic Deﬁnitions and Constructions Section 1.1 5

techniques from this book.) So TS

2

is genuinely twisted, and is not just a disguised

form of the product S

2

×R

2

.

Dropping down a dimension, one could consider in similar fashion the space TS

1

of tangent vectors to the unit circle S

1

in R

2

. In this case there is a continuous

ﬁeld v

x

of nonzero tangent vectors to S

1

, obtained by regarding

points x ∈ S

1

as unit complex numbers and letting v

x

be the

translation of the vector ix that has its tail at x. This leads to a

x

v

x

homeomorphism S

1

×R→TS

1

taking (x, t) to tv

x

, with ¦x¦×R

going to the tangent line at x by a linear isomorphism. Thus TS

1

really is equivalent to the product S

1

×R

1

.

Moving up to S

3

, the unit sphere in R

4

, the space TS

3

of tangent vectors is again

equivalent to the product S

3

×R

3

. Regarding R

4

as the quaternions, an equivalence is

the homeomorphism S

3

×R

3

→TS

3

sending (x, (t

1

, t

2

, t

3

)) to the translation of the

vector t

1

ix + t

2

jx + t

3

kx having its tail at x. A similar construction using Cayley

octonions shows that TS

7

is equivalent to S

7

×R

7

. It is a rather deep theorem, proved

in §2.3, that S

1

, S

3

, and S

7

are the only spheres whose tangent bundle is equivalent

to a product.

Although TS

n

is not usually equivalent to the product S

n

×R

n

, there is a sense

in which this is true locally. Take the case of the 2 sphere for example. For a point

x ∈ S

2

let P be the translate of the tangent plane P

x

that passes through the ori-

gin. For points y ∈ S

2

that are sufﬁciently close to x the map π

y

: P

y

→P sending

a tangent vector v

y

to the orthogonal projection of τ(v

y

) onto P is a linear iso-

morphism. This is true in fact for any y on the same side of P as x. Thus for

y in a suitable neighborhood U of x in S

2

the map (y, v

y

)

(y, π

y

(v

y

)) is a

homeomorphism with domain the subspace of TS

2

consisting of tangent vectors at

points of U and with range the product U×P . Furthermore this homeomorphism

has the key property of restricting to a linear isomorphism from P

y

onto P for each

y ∈ U. A convenient way of rephrasing this situation, having the virtue of easily gen-

eralizing, is to let p: TS

2

→S

2

be the map (x, v

x

)

**x, and then we have a homeo-
**

morphism p

−1

(U)→U×P that restricts to a linear isomorphism p

−1

(y)→¦y¦×P

for each y ∈ U.

6 Chapter 1 Vector Bundles

1.1 Basic Deﬁnitions and Constructions

Throughout the book we use the word map to mean a continuous function.

An n dimensional vector bundle is a map p: E→B together with a real vector

space structure on p

−1

(b) for each b ∈ B, such that the following local triviality

condition is satisﬁed: There is a cover of B by open sets U

α

for each of which there

exists a homeomorphism h

α

: p

−1

(U

α

)→U

α

×R

n

taking p

−1

(b) to ¦b¦×R

n

by a vec-

tor space isomorphism for each b ∈ U

α

. Such an h

α

is called a local trivialization of

the vector bundle. The space B is called the base space, E is the total space, and the

vector spaces p

−1

(b) are the ﬁbers. Often one abbreviates terminology by just calling

the vector bundle E, letting the rest of the data be implicit.

We could equally well take C in place of R as the scalar ﬁeld, obtaining the notion

of a complex vector bundle. We will focus on real vector bundles in this chapter.

Usually the complex case is entirely analogous. In the next chapter complex vector

bundles will play the larger role, however.

Here are some examples of vector bundles:

(1) The product or trivial bundle E = B×R

n

with p the projection onto the ﬁrst

factor.

(2) If we let E be the quotient space of I ×R under the identiﬁcations (0, t) ∼ (1, −t),

then the projection I ×R→I induces a map p: E→S

1

which is a 1 dimensional vector

bundle, or line bundle. Since E is homeomorphic to a M¨obius band with its boundary

circle deleted, we call this bundle the M¨ obius bundle.

(3) The tangent bundle of the unit sphere S

n

in R

n+1

, a vector bundle p: E→S

n

where E = ¦ (x, v) ∈ S

n

×R

n+1

¦ x ⊥ v ¦ and we think of v as a tangent vector to

S

n

by translating it so that its tail is at the head of x, on S

n

. The map p: E→S

n

sends (x, v) to x. To construct local trivializations, choose any point x ∈ S

n

and

let U

x

⊂ S

n

be the open hemisphere containing x and bounded by the hyperplane

through the origin orthogonal to x. Deﬁne h

x

: p

−1

(U

x

)→U

x

×p

−1

(x) ≈ U

x

×R

n

by h

x

(y, v) = (y, π

x

(v)) where π

x

is orthogonal projection onto the hyperplane

p

−1

(x). Then h

x

is a local trivialization since π

x

restricts to an isomorphism of

p

−1

(y) onto p

−1

(x) for each y ∈ U

b

.

(4) The normal bundle to S

n

in R

n+1

, a line bundle p: E→S

n

with E consisting of

pairs (x, v) ∈ S

n

×R

n+1

such that v is perpendicular to the tangent plane to S

n

at

x, or in other words, v = tx for some t ∈ R. The map p: E→S

n

is again given by

p(x, v) = x. As in the previous example, local trivializations h

x

: p

−1

(U

x

)→U

x

×R

can be obtained by orthogonal projection of the ﬁbers p

−1

(y) onto p

−1

(x) for y ∈

U

x

.

(5) Real projective n space RP

n

is the space of lines in R

n+1

through the origin.

Since each such line intersects the unit sphere S

n

in a pair of antipodal points, we can

Basic Deﬁnitions and Constructions Section 1.1 7

also regard RP

n

as the quotient space of S

n

in which antipodal pairs of points are

identiﬁed. The canonical line bundle p: E→RP

n

has as its total space E the subspace

of RP

n

×R

n+1

consisting of pairs (, v) with v ∈ , and p(, v) = . Again local

trivializations can be deﬁned by orthogonal projection.

There is also an inﬁnite-dimensional projective space RP

∞

which is the union

of the ﬁnite-dimensional projective spaces RP

n

under the inclusions RP

n

⊂ RP

n+1

coming from the natural inclusions R

n+1

⊂ R

n+2

. The topology we use on RP

∞

is

the weak or direct limit topology, for which a set in RP

∞

is open iff it intersects each

RP

n

in an open set. The inclusions RP

n

⊂ RP

n+1

induce corresponding inclusions

of canonical line bundles, and the union of all these is a canonical line bundle over

RP

∞

, again with the direct limit topology. Local trivializations work just as in the

ﬁnite-dimensional case.

(6) The canonical line bundle over RP

n

has an orthogonal complement, the space

E

⊥

= ¦ (, v) ∈ RP

n

×R

n+1

¦ v ⊥ ¦. The projection p: E

⊥

→RP

n

, p(, v) = ,

is a vector bundle with ﬁbers the orthogonal subspaces

⊥

, of dimension n. Local

trivializations can be obtained once more by orthogonal projection.

A natural generalization of RP

n

is the so-called Grassmann manifold G

k

(R

n

), the

space of all k dimenional planes through the origin in R

n

. The topology on this space

will be deﬁned precisely in §1.2, along with a canonical k dimensional vector bundle

over it consisting of pairs (, v) where is a point in G

k

(R

n

) and v is a vector in

. This too has an orthogonal complement, an (n − k) dimensional vector bundle

consisting of pairs (, v) with v orthogonal to .

An isomorphism between vector bundles p

1

: E

1

→B and p

2

: E

2

→B over the same

base space B is a homeomorphism h: E

1

→E

2

taking each ﬁber p

−1

1

(b) to the cor-

responding ﬁber p

−1

2

(b) by a linear isomorphism. Thus an isomorphism preserves

all the structure of a vector bundle, so isomorphic bundles are often regarded as the

same. We use the notation E

1

≈ E

2

to indicate that E

1

and E

2

are isomorphic.

For example, the normal bundle of S

n

in R

n+1

is isomorphic to the product bun-

dle S

n

×R by the map (x, tx)

**(x, t). The tangent bundle to S
**

1

is also isomorphic

to the trivial bundle S

1

×R, via (e

iθ

, ite

iθ

)

(e

iθ

, t), for e

iθ

∈ S

1

and t ∈ R.

As a further example, the M¨obius bundle in (2) above is isomorphic to the canon-

ical line bundle over RP

1

≈ S

1

. Namely, RP

1

is swept out by a line rotating through

an angle of π, so the vectors in these lines sweep out a rectangle [0, π]×R with the

two ends ¦0¦×R and ¦π¦×R identiﬁed. The identiﬁcation is (0, x) ∼ (π, −x) since

rotating a vector through an angle of π produces its negative.

Sections

A section of a vector bundle p: E→B is a map s : B→E assigning to each b ∈ B a

vector s(b) in the ﬁber p

−1

(b). The condition s(b) ∈ p

−1

(b) can also be written as

8 Chapter 1 Vector Bundles

ps = 11, the identity map of B. Every vector bundle has a canonical section, the zero

section whose value is the zero vector in each ﬁber. We often identity the zero section

with its image, a subspace of E which projects homeomorphically onto B by p.

One can sometimes distinguish nonisomorphic bundles by looking at the com-

plement of the zero section since any vector bundle isomorphism h: E

1

→E

2

must

take the zero section of E

1

onto the zero section of E

2

, so the complements of the

zero sections in E

1

and E

2

must be homeomorphic. For example, we can see that the

M¨obius bundle is not isomorphic to the product bundle S

1

×R since the complement

of the zero section is connected for the M¨obius bundle but not for the product bundle.

At the other extreme from the zero section would be a section whose values are

all nonzero. Not all vector bundles have such a section. Consider for example the

tangent bundle to S

n

. Here a section is just a tangent vector ﬁeld to S

n

. As we shall

show in §2.2, S

n

has a nonvanishing vector ﬁeld iff n is odd. From this it follows

that the tangent bundle of S

n

is not isomorphic to the trivial bundle if n is even

and nonzero, since the trivial bundle obviously has a nonvanishing section, and an

isomorphism between vector bundles takes nonvanishing sections to nonvanishing

sections.

In fact, an n dimensional bundle p: E→B is isomorphic to the trivial bundle iff

it has n sections s

1

, ··· , s

n

such that the vectors s

1

(b), ··· , s

n

(b) are linearly inde-

pendent in each ﬁber p

−1

(b). In one direction this is evident since the trivial bundle

certainly has such sections and an isomorphism of vector bundles takes linearly inde-

pendent sections to linearly independent sections. Conversely, if one has n linearly

independent sections s

i

, the map h: B×R

n

→E given by h(b, t

1

, ··· , t

n

) =

¸

i

t

i

s

i

(b)

is a linear isomorphism in each ﬁber, and is continuous since its composition with

a local trivialization p

−1

(U)→U×R

n

is continuous. Hence h is an isomorphism by

the following useful technical result:

Lemma 1.1. A continuous map h: E

1

→E

2

between vector bundles over the same

base space B is an isomorphism if it takes each ﬁber p

−1

1

(b) to the corresponding

ﬁber p

−1

2

(b) by a linear isomorphism.

Proof: The hypothesis implies that h is one-to-one and onto. What must be checked

is that h

−1

is continuous. This is a local question, so we may restrict to an open set

U ⊂ B over which E

1

and E

2

are trivial. Composing with local trivializations reduces

to the case of an isomorphism h: U×R

n

→U×R

n

of the form h(x, v) = (x, g

x

(v)).

Here g

x

is an element of the group GL

n

(R) of invertible linear transformations of

R

n

, and g

x

depends continuously on x. This means that if g

x

is regarded as an

n×n matrix, its n

2

entries depend continuously on x. The inverse matrix g

−1

x

also

depends continuously on x since its entries can be expressed algebraically in terms

of the entries of g

x

, namely, g

−1

x

is 1/(det g

x

) times the classical adjoint matrix of

g

x

. Therefore h

−1

(x, v) = (x, g

−1

x

(v)) is continuous. L¦

Basic Deﬁnitions and Constructions Section 1.1 9

As an example, the tangent bundle to S

1

is trivial because it has the section

(x

1

, x

2

)

(−x

2

, x

1

) for (x

1

, x

2

) ∈ S

1

. In terms of complex numbers, if we set

z = x

1

+ix

2

then this section is z

iz since iz = −x

2

+ix

1

.

There is an analogous construction using quaternions instead of complex num-

bers. Quaternions have the form z = x

1

+ix

2

+jx

3

+kx

4

, and form a division algebra

H via the multiplication rules i

2

= j

2

= k

2

= −1, ij = k, jk = i, ki = j , ji = −k,

kj = −i, and ik = −j . If we identify H with R

4

via the coordinates (x

1

, x

2

, x

3

, x

4

),

then the unit sphere is S

3

and we can deﬁne three sections of its tangent bundle by

the formulas

z

iz or (x

1

, x

2

, x

3

, x

4

)

(−x

2

, x

1

, −x

4

, x

3

)

z

jz or (x

1

, x

2

, x

3

, x

4

)

(−x

3

, x

4

, x

1

, −x

2

)

z

kz or (x

1

, x

2

, x

3

, x

4

)

(−x

4

, −x

3

, x

2

, x

1

)

It is easy to check that the three vectors in the last column are orthogonal to each other

and to (x

1

, x

2

, x

3

, x

4

), so we have three linearly independent nonvanishing tangent

vector ﬁelds on S

3

, and hence the tangent bundle to S

3

is trivial.

The underlying reason why this works is that quaternion multiplication satisﬁes

¦zw¦ = ¦z¦¦w¦, where ¦·¦ is the usual norm of vectors in R

4

. Thus multiplication by a

quaternion in the unit sphere S

3

is an isometry of H. The quaternions 1, i, j, k form

the standard orthonormal basis for R

4

, so when we multiply themby an arbitrary unit

quaternion z ∈ S

3

we get a new orthonormal basis z, iz, jz, kz.

The same constructions work for the Cayley octonions, a division algebra struc-

ture on R

8

. Thinking of R

8

as H×H, multiplication of octonions is deﬁned by

(z

1

, z

2

)(w

1

, w

2

) = (z

1

w

1

−w

2

z

2

, z

2

w

1

+w

2

z

1

) and satisﬁes the key property ¦zw¦ =

¦z¦¦w¦. This leads to the construction of seven orthogonal tangent vector ﬁelds on

the unit sphere S

7

, so the tangent bundle to S

7

is also trivial. As we shall show in

§2.3, the only spheres with trivial tangent bundle are S

1

, S

3

, and S

7

.

Another way of characterizing the trivial bundle E ≈ B×R

n

is to say that there is a

continuous projection map E→R

n

which is a linear isomorphism on each ﬁber, since

such a projection together with the bundle projection E→B gives an isomorphism

E ≈ B×R

n

, by Lemma 1.1.

Direct Sums

Given two vector bundles p

1

: E

1

→B and p

2

: E

2

→B over the same base space B,

we would like to create a third vector bundle over B whose ﬁber over each point of B

is the direct sum of the ﬁbers of E

1

and E

2

over this point. This leads us to deﬁne

the direct sum of E

1

and E

2

as the space

E

1

⊕E

2

= ¦ (v

1

, v

2

) ∈ E

1

×E

2

¦ p

1

(v

1

) = p

2

(v

2

) ¦

There is then a projection E

1

⊕E

2

→B sending (v

1

, v

2

) to the point p

1

(v

1

) = p

2

(v

2

).

The ﬁbers of this projection are the direct sums of the ﬁbers of E

1

and E

2

, as we

10 Chapter 1 Vector Bundles

wanted. For a relatively painless veriﬁcation of the local triviality condition we make

two preliminary observations:

(a) Given a vector bundle p: E→B and a subspace A ⊂ B, then p: p

−1

(A)→A is

clearly a vector bundle. We call this the restriction of E over A .

(b) Given vector bundles p

1

: E

1

→B

1

and p

2

: E

2

→B

2

, then p

1

×p

2

: E

1

×E

2

→B

1

×B

2

is also a vector bundle, with ﬁbers the products p

−1

1

(b

1

)×p

−1

2

(b

2

). For if we have

local trivializations h

α

: p

−1

1

(U

α

)→U

α

×R

n

and h

β

: p

−1

2

(U

β

)→U

β

×R

m

for E

1

and

E

2

, then h

α

×h

β

is a local trivialization for E

1

×E

2

.

Then if E

1

and E

2

both have the same base space B, the restriction of the product

E

1

×E

2

over the diagonal B = ¦(b, b) ∈ B×B¦ is exactly E

1

⊕E

2

.

The direct sum of two trivial bundles is again a trivial bundle, clearly, but the

direct sum of nontrivial bundles can also be trivial. For example, the direct sum of

the tangent and normal bundles to S

n

in R

n+1

is the trivial bundle S

n

×R

n+1

since

elements of the direct sum are triples (x, v, tx) ∈ S

n

×R

n+1

×R

n+1

with x ⊥ v, and

the map (x, v, tx)

**(x, v +tx) gives an isomorphism of the direct sum bundle with
**

S

n

×R

n+1

. So the tangent bundle to S

n

is stably trivial: it becomes trivial after taking

the direct sum with a trivial bundle.

As another example, the direct sum E⊕E

⊥

of the canonical line bundle E→RP

n

with its orthogonal complement, deﬁned in example (6) above, is isomorphic to the

trivial bundle RP

n

×R

n+1

via the map (, v, w)

(, v + w) for v ∈ and w ⊥ .

Specializing to the case n = 1, the bundle E

⊥

is isomorphic to E itself by the map that

rotates each vector in the plane by 90 degrees. We noted earlier that E is isomorphic

to the M¨obius bundle over S

1

= RP

1

, so it follows that the direct sum of the M¨obius

bundle with itself is the trivial bundle. To see this geometrically, embed the M¨obius

bundle in the product bundle S

1

×R

2

by taking the line in the ﬁber ¦θ¦×R

2

that makes

an angle of θ/2 with the x axis, and then the orthogonal lines in the ﬁbers form a

second copy of the M¨obius bundle, giving a decomposition of the product S

1

×R

2

as

the direct sum of two M¨obius bundles.

Example: The tangent bundle of real projective space. Starting with the isomor-

phism S

n

×R

n+1

≈ TS

n

⊕NS

n

, where NS

n

is the normal bundle of S

n

in R

n+1

,

suppose we factor out by the identiﬁcations (x, v) ∼ (−x, −v) on both sides of this

isomorphism. Applied to TS

n

this identiﬁcation yields TRP

n

, the tangent bundle to

RP

n

. This is saying that a tangent vector to RP

n

is equivalent to a pair of antipodal

tangent vectors to S

n

. A moment’s reﬂection shows this to be entirely reasonable,

although a formal proof would require a signiﬁcant digression on what precisely tan-

gent vectors to a smooth manifold are, a digression we shall skip here. What we will

show is that even though the direct sum of TRP

n

with a trivial line bundle may not

be trivial as it is for a sphere, it does split in an interesting way as a direct sum of

nontrivial line bundles.

Basic Deﬁnitions and Constructions Section 1.1 11

In the normal bundle NS

n

the identiﬁcation (x, v) ∼ (−x, −v) can be written as

(x, tx) ∼ (−x, t(−x)). This identiﬁcation yields the product bundle RP

n

×R since

the section x

**(−x, −x) is well-deﬁned in the quotient. Now let us consider the
**

identiﬁcation (x, v) ∼ (−x, −v) in S

n

×R

n+1

. This identiﬁcation respects the co-

ordinate factors of R

n+1

, so the quotient is the direct sum of n + 1 copies of the

line bundle E over RP

n

obtained by making the identiﬁcations (x, t) ∼ (−x, −t)

in S

n

×R. The claim is that E is just the canonical line bundle over RP

n

. To see

this, let us identify S

n

×R with NS

n

by the isomorphism (x, t)

(x, tx). hence

(−x, −t)

**((−x, (−t)(−x)) = (−x, tx). Thus we have the identiﬁcation (x, tx) ∼
**

(−x, tx) in NS

n

. The quotient is the canonical line bundle over RP

n

since the iden-

tiﬁcations x ∼ −x in the ﬁrst coordinate give lines through the origin in R

n+1

, and

in the second coordinate there are no identiﬁcations so we have well-deﬁned vectors

tx in these lines.

Thus we have shown that the tangent bundle TRP

n

is stably isomorphic to the

direct sum of n + 1 copies of the canonical line bundle over RP

n

. When n = 3, for

example, TRP

3

is trivial since the three linearly independent tangent vector ﬁelds on

S

3

deﬁned earlier in terms of quaternions pass down to linearly independent tangent

vector ﬁelds on the quotient RP

3

. Hence the direct sumof four copies of the canonical

line bundle over RP

3

is trivial. Similarly using octonions we can see that the direct

sum of eight copies of the canonical line bundle over RP

7

is trivial. In §2.5 we will

determine when the sum of k copies of the canonical line bundle over RP

n

is at least

stably trivial. The answer turns out to be rather subtle: This happens exactly when k

is a multiple of 2

ϕ(n)

where ϕ(n) is the number of integers i in the range 0 < i ≤ n

with i congruent to 0, 1, 2, or 4 modulo 8. For n = 3, 7 this gives 2

ϕ(n)

= 4, 8,

the same numbers we obtained from the triviality of TRP

3

and TRP

7

. If there were

a 16 dimensional division algebra after the octonions then one might expect the sum

of 16 copies of the canonical line bundle over RP

15

to be trivial. However this is not

the case since 2

ϕ(15)

= 2

7

= 128.

Inner Products

An inner product on a vector bundle p: E→B is a map ¹ , ) : E⊕E→R which

restricts in each ﬁber to an inner product, a positive deﬁnite symmetric bilinear form.

Proposition 1.2. An inner product exists for a vector bundle p: E→B if B is compact

Hausdorff or more generally paracompact.

The deﬁnition of paracompactness we are using is that a space X is paracom-

pact if it is Hausdorff and every open cover has a subordinate partition of unity, a

collection of maps ϕ

β

: X→[0, 1] each supported in some set of the open cover, and

with

¸

β

ϕ

β

= 1, only ﬁnitely many of the ϕ

β

being nonzero near each point of X.

Constructing such functions is easy when X is compact Hausdorff, using Urysohn’s

12 Chapter 1 Vector Bundles

Lemma. This is done in the appendix to this chapter, where we also show that certain

classes of noncompact spaces are paracompact. Most spaces that arise naturally in

algebraic topology are paracompact.

Proof: An inner product for p: E→B can be constructed by ﬁrst using local trivial-

izations h

α

: p

−1

(U

α

)→U

α

×R

n

, to pull back the standard inner product in R

n

to an

inner product ¹·, ·)

α

on p

−1

(U

α

), then setting ¹v, w) =

¸

β

ϕ

β

p(v)¹v, w)

α(β)

where

¦ϕ

β

¦ is a partition of unity with the support of ϕ

β

contained in U

α(β)

. L¦

In the case of complex vector bundles one can construct Hermitian inner products

in the same way.

In linear algebra one can show that a vector subspace is always a direct summand

by taking its orthogonal complement. We will show now that the corresponding result

holds for vector bundles over a paracompact base. A vector subbundle of a vector

bundle p: E→B has the natural deﬁnition: a subspace E

0

⊂ E intersecting each ﬁber

of E in a vector subspace, such that the restriction p: E

0

→B is a vector bundle.

Proposition 1.3. If E→B is a vector bundle over a paracompact base B and E

0

⊂ E

is a vector subbundle, then there is a vector subbundle E

⊥

0

⊂ E such that E

0

⊕E

⊥

0

≈ E.

Proof: With respect to a chosen inner product on E, let E

⊥

0

be the subspace of E

which in each ﬁber consists of all vectors orthogonal to vectors in E

0

. We claim

that the natural projection E

⊥

0

→B is a vector bundle. If this is so, then E

0

⊕E

⊥

0

is

isomorphic to E via the map (v, w)

v +w, using Lemma 1.1.

To see that E

⊥

0

satisﬁes the local triviality condition for a vector bundle, note

ﬁrst that we may assume E is the product B×R

n

since the question is local in B.

Since E

0

is a vector bundle, of dimension m say, it has m independent local sections

b

(b, s

i

(b)) near each point b

0

∈ B. We may enlarge this set of m independent

local sections of E

0

to a set of n independent local sections b

(b, s

i

(b)) of E

by choosing s

m+1

, ··· , s

n

ﬁrst in the ﬁber p

−1

(b

0

), then taking the same vectors for

all nearby ﬁbers, since if s

1

, ··· , s

m

, s

m+1

, ··· , s

n

are independent at b

0

, they will

remain independent for nearby b by continuity of the determinant function. Apply

the Gram-Schmidt orthogonalization process to s

1

, ··· , s

m

, s

m+1

, ··· , s

n

in each ﬁber,

using the given inner product, to obtain new sections s

¹

i

. The explicit formulas for

the Gram-Schmidt process show the s

¹

i

’s are continuous, and the ﬁrst m of them are

a basis for E

0

in each ﬁber. The sections s

¹

i

allow us to deﬁne a local trivialization

h: p

−1

(U)→U×R

n

with h(b, s

¹

i

(b)) equal to the i

th

standard basis vector of R

n

.

This h carries E

0

to U×R

m

and E

⊥

0

to U×R

n−m

, so h¦

¦E

⊥

0

is a local trivialization of

E

⊥

0

. L¦

Note that when the subbundle E

0

is equal to E itself, the last part of the proof

shows that for any vector bundle with an inner product it is always possible to choose

Basic Deﬁnitions and Constructions Section 1.1 13

local trivializations that carry the inner product to the standard inner product, so the

local trivializations are by isometries.

We have seen several cases where the sum of two bundles, one or both of which

may be nontrivial, is the trivial bundle. Here is a general result result along these lines:

Proposition 1.4. For each vector bundle E→B with B compact Hausdorff there

exists a vector bundle E

¹

→B such that E⊕E

¹

is the trivial bundle.

This can fail when B is noncompact. An example is the canonical line bundle over

RP

∞

, as we shall see in Example 3.6.

Proof: To motivate the construction, suppose ﬁrst that the result holds and hence

that E is a subbundle of a trivial bundle B×R

N

. Composing the inclusion of E into

this product with the projection of the product onto R

N

yields a map E→R

N

that

is a linear injection on each ﬁber. Our strategy will be to reverse the logic here, ﬁrst

constructing a map E→R

N

that is a linear injection on each ﬁber, then showing that

this gives an embedding of E in B×R

N

as a direct summand.

Each point x ∈ B has a neighborhood U

x

over which E is trivial. By Urysohn’s

Lemma there is a map ϕ

x

: B→[0, 1] that is 0 outside U

x

and nonzero at x. Letting

x vary, the sets ϕ

−1

x

(0, 1] form an open cover of B. By compactness this has a ﬁnite

subcover. Let the corresponding U

x

’s and ϕ

x

’s be relabeled U

i

and ϕ

i

. Deﬁne

g

i

: E→R

n

by g

i

(v) = ϕ

i

(p(v))[π

i

h

i

(v)] where p is the projection E→B and π

i

h

i

is the composition of a local trivialization h

i

: p

−1

(U

i

)→U

i

×R

n

with the projection

π

i

to R

n

. Then g

i

is a linear injection on each ﬁber over ϕ

−1

i

(0, 1], so if we make

the various g

i

’s the coordinates of a map g : E→R

N

with R

N

a product of copies of

R

n

, then g is a linear injection on each ﬁber.

The map g is the second coordinate of a map f : E→B×R

N

with ﬁrst coordinate

p. The image of f is a subbundle of the product B×R

N

since projection of R

N

onto

the i

th

R

n

factor gives the second coordinate of a local trivialization over ϕ

−1

i

(0, 1].

Thus we have E isomorphic to a subbundle of B×R

N

so by preceding proposition

there is a complementary subbundle E

¹

with E⊕E

¹

isomorphic to B×R

N

. L¦

Tensor Products

In addition to direct sum, a number of other algebraic constructions with vec-

tor spaces can be extended to vector bundles. One which is particularly important

for K–theory is tensor product. For vector bundles p

1

: E

1

→B and p

2

: E

2

→B, let

E

1

⊗E

2

, as a set, be the disjoint union of the vector spaces p

−1

1

(x)⊗p

−1

2

(x) for

x ∈ B. The topology on this set is deﬁned in the following way. Choose isomorphisms

h

i

: p

−1

i

(U)→U×R

n

i

for each open set U ⊂ B over which E

1

and E

2

are trivial. Then

a topology T

U

on the set p

−1

1

(U)⊗p

−1

2

(U) is deﬁned by letting the ﬁberwise tensor

product map h

1

⊗h

2

: p

−1

1

(U)⊗p

−1

2

(U)→U×(R

n

1

⊗R

n

2

) be a homeomorphism. The

14 Chapter 1 Vector Bundles

topology T

U

is independent of the choice of the h

i

’s since any other choices are ob-

tained by composing with isomorphisms of U×R

n

i

of the form (x, v)

(x, g

i

(x)(v))

for continuous maps g

i

: U→GL

n

i

(R), hence h

1

⊗h

2

changes by composing with

analogous isomorphisms of U×(R

n

1

⊗R

n

2

) whose second coordinates g

1

⊗g

2

are

continuous maps U→GL

n

1

n

2

(R), since the entries of the matrices g

1

(x) ⊗g

2

(x) are

the products of the entries of g

1

(x) and g

2

(x). When we replace U by an open sub-

set V , the topology on p

−1

1

(V)⊗p

−1

2

(V) induced by T

U

is the same as the topology

T

V

since local trivializations over U restrict to local trivializations over V . Hence we

get a well-deﬁned topology on E

1

⊗E

2

making it a vector bundle over B.

There is another way to look at this construction that takes as its point of depar-

ture a general method for constructing vector bundles we have not mentioned previ-

ously. If we are given a vector bundle p: E→B and an open cover ¦U

α

¦ of B with lo-

cal trivializations h

α

: p

−1

(U

α

)→U

α

×R

n

, then we can reconstruct E as the quotient

space of the disjoint union

¸

α

(U

α

×R

n

) obtained by identifying (x, v) ∈ U

α

×R

n

with h

β

h

−1

α

(x, v) ∈ U

β

×R

n

whenever x ∈ U

α

∩ U

β

. The functions h

β

h

−1

α

can

be viewed as maps g

βα

: U

α

∩ U

β

→GL

n

(R). These satisfy the ‘cocycle condition’

g

γβ

g

βα

= g

γα

on U

α

∩ U

β

∩ U

γ

. Any collection of ‘gluing functions’ g

βα

satisfying

this condition can be used to construct a vector bundle E→B.

In the case of tensor products, suppose we have two vector bundles E

1

→B and

E

2

→B. We can choose an open cover ¦U

α

¦ with both E

1

and E

2

trivial over each U

α

,

and so obtain gluing functions g

i

βα

: U

α

∩U

β

→GL

n

i

(R) for each E

i

. Then the gluing

functions for the bundle E

1

⊗E

2

are the tensor product functions g

1

βα

⊗g

2

βα

assigning

to each x ∈ U

α

∩U

β

the tensor product of the two matrices g

1

βα

(x) and g

2

βα

(x).

It is routine to verify that the tensor product operation for vector bundles over a

ﬁxed base space is commutative, associative, and has an identity element, the trivial

line bundle. It is also distributive with respect to direct sum.

If we restrict attention to line bundles, then the set Vect

1

(B) of isomorphism

classes of one-dimensional vector bundles over B is an abelian group with respect

to the tensor product operation. The inverse of a line bundle E→B is obtained by

replacing its gluing matrices g

βα

(x) ∈ GL

1

(R) with their inverses. The cocycle con-

dition is preserved since 1×1 matrices commute. If we give E an inner product, we

may rescale local trivializations h

α

to be isometries, taking vectors in ﬁbers of E to

vectors in R

1

of the same length. Then all the values of the gluing functions g

βα

are

:1, being isometries of R. The gluing functions for E⊗E are the squares of these

g

βα

’s, hence are identically 1, so E⊗E is the trivial line bundle. Thus each element of

Vect

1

(B) is its own inverse. As we shall see in §3.1, the group Vect

1

(B) is isomorphic

to a standard object in algebraic topology, the cohomology group H

1

(B; Z

2

) when B

is homotopy equivalent to a CW complex.

These tensor product constructions work equally well for complex vector bundles.

Tensor product again makes the complex analog Vect

1

C

(B) of Vect

1

(B) into an abelian

Basic Deﬁnitions and Constructions Section 1.1 15

group, but after rescaling the gluing functions g

βα

for a complex line bundle E, the

values are complex numbers of norm 1, not necessarily :1, so we cannot expect E⊗E

to be trivial. In §3.1 we will show that the group Vect

1

C

(B) is isomorphic to H

2

(B; Z)

when B is homotopy equivalent to a CW complex.

We may as well mention here another general construction for complex vector

bundles E→B, the notion of the conjugate bundle E→B. As a topological space, E

is the same as E, but the vector space structure in the ﬁbers is modiﬁed by redeﬁning

scalar multiplication by the rule λ(v) = λv where the right side of this equation

means scalar multiplication in E and the left side means scalar multiplication in E.

This implies that local trivializations for E are obtained from local trivializations for

E by composing with the coordinatewise conjugation map C

n

→C

n

in each ﬁber. The

effect on the gluing maps g

βα

is to replace them by their complex conjugates as

well. Specializing to line bundles, we then have E⊗E isomorphic to the trivial line

bundle since its gluing maps have values zz = 1 for z a unit complex number. Thus

conjugate bundles provide inverses in Vect

1

C

(B).

Besides tensor product of vector bundles, another construction useful in K–theory

is the exterior power λ

k

(E) of a vector bundle E. Recall from linear algebra that

the exterior power λ

k

(V) of a vector space V is the quotient of the k fold tensor

product V ⊗ ··· ⊗V by the subspace generated by vectors of the form v

1

⊗ ··· ⊗v

k

−

sgn(σ)v

σ(1)

⊗ ··· ⊗v

σ(k)

where σ is a permutation of the subscripts and sgn(σ) =

:1 is its sign, +1 for an even permutation and −1 for an odd permutation. If V has

dimension n then λ

k

(V) has dimension

n

k

. Nowto deﬁne λ

k

(E) for a vector bundle

p: E→B the procedure follows closely what we did for tensor product. We ﬁrst form

the disjoint union of the exterior powers λ

k

(p

−1

(x)) of all the ﬁbers p

−1

(x), then we

deﬁne a topology on this set via local trivializations. The key fact about tensor product

which we needed before was that the tensor product ϕ⊗ψ of linear transformations

ϕ and ψ depends continuously on ϕ and ψ. For exterior powers the analogous fact

is that a linear map ϕ: R

n

→R

n

induces a linear map λ

k

(ϕ) : λ

k

(R

n

)→λ

k

(R

n

) which

depends continuously on ϕ. This holds since λ

k

(ϕ) is a quotient map of the k fold

tensor product of ϕ with itself.

Associated Fiber Bundles

If we modify the deﬁnition of a vector bundle by dropping all references to vector

spaces and replace the model ﬁber R

n

by an arbitrary space F , then we have the more

general notion of a ﬁber bundle with ﬁber F . This is a map p: E→B such that there

is a cover of B by open sets U

α

for each of which there exists a homeomorphism

h

α

: p

−1

(U

α

)→U

α

×F taking p

−1

(b) to ¦b¦×F for each b ∈ U

α

. We will describe

now several different ways of constructing ﬁber bundles from vector bundles.

Having an inner product on a vector bundle E, lengths of vectors are deﬁned, and

we can consider the subspace S(E) consisting of the unit spheres in all the ﬁbers. The

16 Chapter 1 Vector Bundles

natural projection S(E)→B is a ﬁber bundle with sphere ﬁbers since we have observed

that local trivializations for E can be chosen to be isometries in each ﬁber, so these

local trivializations restrict to local trivializations for S(E). Similarly there is a disk

bundle D(E)→B with ﬁbers the disks of vectors of length less than or equal to 1. It is

possible to describe S(E) without reference to an inner product, as the quotient of the

complement of the zero section in E obtained by identifying each nonzero vector with

all positive scalar multiples of itself. It follows that D(E) can also be deﬁned without

invoking a metric, namely as the mapping cylinder of the projection S(E)→B. This

is the quotient space of S(E)×[0, 1] obtained by identifying two points in S(E)×¦0¦

if they have the same image in B.

The canonical line bundle E→RP

n

has as its unit sphere bundle S(E) the space

of unit vectors in lines through the origin in R

n+1

. Since each unit vector uniquely

determines the line containing it, S(E) is the same as the space of unit vectors in

R

n+1

, i.e., S

n

.

Here are some more examples.

(1) Associated to a vector bundle E→B is the projective bundle P(E)→B, where P(E)

is the space of all lines through the origin in all the ﬁbers of E. We topologize P(E) as

the quotient of the sphere bundle S(E) obtained by factoring out scalar multiplication

in each ﬁber. Over a neighborhood U in B where E is a product U×R

n

, this quotient

is U×RP

n−1

, so P(E) is a ﬁber bundle over B with ﬁber RP

n−1

, with respect to the

projection P(E)→B which sends each line in the ﬁber of E over a point b ∈ B to b.

(2) For an n dimensional vector bundle E→B, the associated ﬂag bundle F(E)→B

has total space F(E) the subspace of the n fold product of P(E) with itself consisting

of n tuples of orthogonal lines in ﬁbers of E. The ﬁber of F(E) is thus the ﬂag

manifold F(R

n

) consisting of n tuples of orthogonal lines through the origin in R

n

.

Local triviality follows as in the preceding example. More generally, for any k ≤ n one

could take k tuples of orthogonal lines in ﬁbers of E and get a bundle F

k

(E)→B.

(3) As a reﬁnement of the last example, one could form the Stiefel bundle V

k

(E)→B,

where points of V

k

(E) are k tuples of orthogonal unit vectors in ﬁbers of E, so V

k

(E)

is a subspace of the product of k copies of S(E). The ﬁber of V

k

(E) is the Stiefel

manifold V

k

(R

n

) of orthonormal k frames in R

n

.

(4) Generalizing P(E), there is the Grassmann bundle G

k

(E)→B of k dimensional

linear subspaces of ﬁbers of E. This is the quotient space of V

k

(E) obtained by

identifying two k frames in a ﬁber if they span the same subspace of the ﬁber. The

ﬁber of G

k

(E) is the Grassmann manifold G

k

(R

n

) of k planes through the origin in

R

n

.

Basic Deﬁnitions and Constructions Section 1.1 17

Exercises

1. Show that a vector bundle E→X has k independent sections iff it has a trivial

k dimensional subbundle.

2. For a vector bundle E→X with a subbundle E

¹

⊂ E, construct a quotient vector

bundle E/E

¹

→X.

3. Show that the orthogonal complement of a subbundle is independent of the choice

of inner product, up to isomorphism.

18 Chapter 1 Vector Bundles

1.2 Classifying Vector Bundles

As was stated in the Introduction, it is usually a difﬁcult problem to classify all

the different vector bundles over a given base space. The goal of this section will be to

rephrase the problemin terms of a standard concept of algebraic topology, the idea of

homotopy classes of maps. This will allow the problem to be solved directly in a few

very simple cases. Using machinery of algebraic topology, other more difﬁcult cases

can be handled as well, as is explained in §??. The reformulation in terms of homotopy

also offers some conceptual enlightment about the structure of vector bundles.

For the reader who is unfamiliar with the notion of homotopy we give the basic

deﬁnitions in the Glossary [not yet written], and more details can be found in the

introductory chapter of the author’s book Algebraic Topology.

Pullback Bundles

We will denote the set of isomorphism classes of n dimensional real vector bun-

dles over B by Vect

n

(B). The complex analogue, when we need it, will be denoted

by Vect

n

C

(B). Our ﬁrst task is to show how a map f : A→B gives rise to a function

f

∗

: Vect(B)→Vect(A), in the reverse direction.

Proposition 1.5. Given a map f : A→B and a vector bundle p: E→B, then there

exists a vector bundle p

¹

: E

¹

→A with a map f

¹

: E

¹

→E taking the ﬁber of E

¹

over

each point a ∈ A isomorphically onto the ﬁber of E over f(a), and such a vector

bundle E

¹

is unique up to isomorphism.

From the uniqueness statement it follows that the isomorphism type of E

¹

de-

pends only on the isomorphism type of E since we can compose the map f

¹

with

an isomorphism of E with another vector bundle over B. Thus we have a function

f

∗

: Vect(B)→Vect(A) taking the isomorphism class of E to the isomorphism class

of E

¹

. Often the vector bundle E

¹

is written as f

∗

(E) and called the bundle induced

by f , or the pullback of E by f .

Proof: First we construct an explicit pullback by setting E

¹

= ¦ (a, v) ∈ A×E ¦ f(a) =

p(v) ¦. This subspace of A×E ﬁts into the diagram at the right where p

¹

(a, v) = a

−

−

→

−

−

→

−−−−−→

E E

−−−−−→

A B

f

p p

f

∼

and f

¹

(a, v) = v. Notice that the two compositions fp

¹

and pf

¹

from E

¹

to B are equal since they both since a pair (a, v) to f(a).

The formula fp

¹

= pf

¹

looks a bit like a commutativity relation,

which may explain why the word commutative is used to describe a

diagram like this one in which any two compositions of maps from one point in the

diagram to another are equal.

If we let Γ

f

denote the graph of f , all points (a, f(a)) in A×B, then p

¹

factors

as the composition E

¹

→Γ

f

→A, (a, v)

(a, p(v)) = (a, f(a))

a. The ﬁrst of

these two maps is the restriction of the vector bundle 11×p: A×E→A×B over the

Classifying Vector Bundles Section 1.2 19

graph Γ

f

, so it is a vector bundle, and the second map is a homeomorphism, so their

composition p

¹

: E

¹

→A is a vector bundle. The map f

¹

obviously takes the ﬁber E

¹

over a isomorphically onto the ﬁber of E over f(a).

For the uniqueness statement, we can construct an isomorphismfroman arbitrary

E

¹

satisfying the conditions in the proposition to the particular one just constructed by

sending v

¹

∈ E

¹

to the pair (p

¹

(v

¹

), f

¹

(v

¹

)). This map obviously takes each ﬁber of E

¹

to the corresponding ﬁber of f

∗

(E) by a vector space isomorphism, so by Lemma 1.1

it is an isomorphism of vector bundles. L¦

One can be more explicit about local trivializations in the pullback bundle f

∗

(E)

constructed above. If E is trivial over a subspace U ⊂ B then f

∗

(E) is trivial over

f

−1

(U) since linearly independent sections s

i

of E over U give rise to independent

sections a

(a, s

i

(f(a))) of f

∗

(E) over f

−1

(U). In particular, the pullback of a

trivial bundle is a trivial bundle. This can also be seen directly from the deﬁnition,

which in the case E = B×R

n

just says that f

∗

(E) consists of the triples (a, b, v) in

A×B×R

n

with b = f(a), so b is redundant and we have just the product A×R

n

.

Now let us give some examples of pullbacks.

(1) The restriction of a vector bundle p: E→B over a subspace A ⊂ B can be viewed

as a pullback with respect to the inclusion map A

**B since the inclusion p
**

−1

(A)

E

is certainly an isomorphism on each ﬁber.

(2) Another very special case is when the map f is a constant map, having image a

single point b ∈ B. Then f

∗

(E) is just the product A×p

−1

(b), a trivial bundle.

(3) The tangent bundle TS

n

is the pullback of the tangent bundle TRP

n

via the quo-

tient map S

n

→RP

n

. Indeed, TRP

n

was deﬁned as a quotient space of TS

n

and the

quotient map takes ﬁbers isomorphically to ﬁbers.

(4) An interesting example which is small enough to be visualized completely is the

pullback of the M¨obius bundle E→S

1

by the two-to-one map f : S

1

→S

1

given by

f(z) = z

2

in complex notation. One can regard the M¨obius bundle as the quotient

of S

1

×R under the identiﬁcations (z, t) ∼ (−z, −t), and the quotient map for this

identiﬁcation is the map f

¹

exhibiting the annulus S

1

×R as the pullback of the M¨obius

bundle. More concretely, the pullback bundle can be thought of as a coat of paint

applied to ‘both sides’ of the M¨obius bundle, and the map f

¹

sends each molecule

of paint to the point of the M¨obius band to which it adheres. Since E has one half-

twist, the pullback has two half-twists, hence is the trivial bundle. More generally, if

E

n

is the pullback of the M¨obius bundle by the map z

z

n

, then E

n

is the trivial

bundle for n even and the M¨obius bundle for n odd. This can be seen by viewing

the M¨obius bundle as the quotient of a strip [0, 1]×R obtained by identifying the

two edges ¦0¦×R and ¦1¦×R by a reﬂection of R, and then the bundle E

n

can be

constructed from n such strips by identifying the right edge of the i

th

strip to the

20 Chapter 1 Vector Bundles

left edge of the i + 1

st

strip by a reﬂection, the number i being taken modulo n so

that the last strip is glued back to the ﬁrst.

(5) At the end of the previous section we deﬁned the ﬂag bundle F(E) associated to

an n dimensional vector bundle E→B to be the space of orthogonal direct sum de-

compositions of ﬁbers of E into lines. The vectors in the i

th

line form a line bundle

L

i

→F(E), and the direct sum L

1

⊕··· ⊕L

n

is nothing but the pullback of E with re-

spect to the projection F(E)→B since a point in the pullback

−

−

→

−

−

→

−−−−−→

E

−−−−−→

E B F ( )

⊕ ⊕ L

n

L

1

. . .

consists of an n tuple of lines

1

⊥ ··· ⊥

n

in a ﬁber of E

together with a vector v in this ﬁber, and v can be expressed

uniquely as a sum v = v

1

+ ··· + v

n

with v

i

∈

i

. This construction for pulling

an arbitrary bundle back to a sum of line bundles is a key ingredient in the so-called

‘splitting principle’ which is important in §2.3 and §3.1.

Here are some elementary properties of pullbacks:

(i) (fg)

∗

(E) ≈ g

∗

(f

∗

(E)).

(ii) 11

∗

(E) ≈ E.

(iii) f

∗

(E

1

⊕E

2

) ≈ f

∗

(E

1

)⊕f

∗

(E

2

).

(iv) f

∗

(E

1

⊗E

2

) ≈ f

∗

(E

1

)⊗f

∗

(E

2

).

The proofs are easy applications of the preceding proposition. In each case one just

checks that the bundle on the right satisﬁes the characteristic property of a pullback.

For example in (iv) there is a natural map from f

∗

(E

1

)⊗f

∗

(E

2

) to E

1

⊗E

2

that is an

isomorphism on each ﬁber, so f

∗

(E

1

)⊗f

∗

(E

2

) satisﬁes the condition for being the

pullback f

∗

(E

1

⊗E

2

).

Now we come to the main technical result about pullbacks:

Theorem 1.6. Given a vector bundle p: E→B and homotopic maps f

0

, f

1

: A→B,

then the induced bundles f

∗

0

(E) and f

∗

1

(E) are isomorphic if A is compact Hausdorff

or more generally paracompact.

Proof: Let F : A×I→B be a homotopy from f

0

to f

1

. The restrictions of F

∗

(E) over

A×¦0¦ and A×¦1¦ are then f

∗

0

(E) and f

∗

1

(E). So the theorem will be an immediate

consequence of the following result:

Proposition 1.7. The restrictions of a vector bundle E→X×I over X×¦0¦ and

X×¦1¦ are isomorphic if X is paracompact.

Proof: We need two preliminary facts:

(1) A vector bundle p: E→X×[a, b] is trivial if its restrictions over X×[a, c] and

X×[c, b] are both trivial for some c ∈ (a, b). To see this, let these restrictions

be E

1

= p

−1

(X×[a, c]) and E

2

= p

−1

(X×[c, b]), and let h

1

: E

1

→X×[a, c]×R

n

and h

2

: E

2

→X×[c, b]×R

n

be isomorphisms. These isomorphisms may not agree on

p

−1

(X×¦c¦), but they can be made to agree by replacing h

2

by its composition with

Classifying Vector Bundles Section 1.2 21

the isomorphism X×[c, b]×R

n

→X×[c, b]×R

n

which on each slice X×¦x¦×R

n

is

given by h

1

h

−1

2

: X×¦c¦×R

n

→X×¦c¦×R

n

. Once h

1

and h

2

agree on E

1

∩E

2

, they

deﬁne a trivialization of E.

(2) For a vector bundle p: E→X×I , there exists an open cover ¦U

α

¦ of X so that each

restriction p

−1

(U

α

×I)→U

α

×I is trivial. This is because for each x ∈ X we can ﬁnd

open neighborhoods U

x,1

, ··· , U

x,k

in X and a partition 0 = t

0

< t

1

< ··· < t

k

= 1 of

[0, 1] such that the bundle is trivial over U

x,i

×[t

i−1

, t

i

], using compactness of [0, 1].

Then by (1) the bundle is trivial over U

α

×I where U

α

= U

x,1

∩··· ∩U

x,k

.

Now we prove the proposition. By (2), we can choose an open cover ¦U

α

¦ of X

so that E is trivial over each U

α

×I . Let us ﬁrst deal with the simpler case that X

is compact Hausdorff. Then a ﬁnite number of U

α

’s cover X. Relabel these as U

i

for i = 1, 2, ··· , m. As shown in Proposition 1.4 there is a corresponding partition

of unity by functions ϕ

i

with the support of ϕ

i

contained in U

i

. For i ≥ 0, let

ψ

i

= ϕ

1

+··· +ϕ

i

, so in particular ψ

0

= 0 and ψ

m

= 1. Let X

i

be the graph of ψ

i

,

the subspace of X×I consisting of points of the form (x, ψ

i

(x)), and let p

i

: E

i

→X

i

be the restriction of the bundle E over X

i

. Since E is trivial over U

i

×I , the nat-

ural projection homeomorphism X

i

→X

i−1

lifts to a homeomorphism h

i

: E

i

→E

i−1

which is the identity outside p

−1

i

(U

i

) and which takes each ﬁber of E

i

isomorphically

onto the corresponding ﬁber of E

i−1

. The composition h = h

1

h

2

··· h

m

is then an

isomorphism from the restriction of E over X×¦1¦ to the restriction over X×¦0¦.

In the general case that X is only paracompact, Lemma 1.21 asserts that there is

a countable cover ¦V

i

¦

i≥1

of X and a partition of unity ¦ϕ

i

¦ with ϕ

i

supported in

V

i

, such that each V

i

is a disjoint union of open sets each contained in some U

α

. This

means that E is trivial over each V

i

×I . As before we let ψ

i

= ϕ

1

+··· +ϕ

i

and let

p

i

: E

i

→X

i

be the restriction of E over the graph of ψ

i

. Also as before we construct

h

i

: E

i

→E

i−1

using the fact that E is trivial over V

i

×I . The inﬁnite composition h =

h

1

h

2

··· is then a well-deﬁned isomorphism from the restriction of E over X×¦1¦ to

the restriction over X×¦0¦ since near each point x ∈ X only ﬁnitely many ϕ

i

’s are

nonzero, so there is a neighborhood of x in which all but ﬁnitely many h

i

’s are the

identity. L¦

Corollary 1.8. A homotopy equivalence f : A→B of paracompact spaces induces a

bijection f

∗

: Vect

n

(B)→Vect

n

(A). In particular, every vector bundle over a con-

tractible paracompact base is trivial.

Proof: If g is a homotopy inverse of f then we have f

∗

g

∗

= 11

∗

= 11 and g

∗

f

∗

=

11

∗

= 11. L¦

We might remark that Theorem 1.6 holds for ﬁber bundles as well as vector bun-

dles, with the same proof.

22 Chapter 1 Vector Bundles

Clutching Functions

Let us describe a way to construct vector bundles E→S

k

with base space a sphere.

Write S

k

as the union of its upper and lower hemispheres D

k

+

and D

k

−

, with D

k

+

∩D

k

−

=

S

k−1

. Given a map f : S

k−1

→GL

n

(R), let E

f

be the quotient of the disjoint union

D

k

+

×R

n

I D

k

−

×R

n

obtained by identifying (x, v) ∈ ∂D

k

−

×R

n

with (x, f(x)(v)) ∈

∂D

k

+

×R

n

. There is then a natural projection E

f

→S

k

and this is an n dimensional

vector bundle, as one can most easily see by taking an equivalent deﬁnition in which

the two hemispheres of S

k

are enlarged slightly to open balls and the identiﬁcation

occurs over their intersection, a product S

k−1

×(−ε, ε), with the map f used in each

slice S

k−1

×¦t¦. From this viewpoint the construction of E

f

is a special case of the

general construction of vector bundles by gluing together products, described earlier

in the discussion of tensor products.

The map f is called a clutching function for E

f

, by analogy with the mechanical

clutch which engages and disengages gears in machinery. The same construction

works equally well with C in place of R, so froma map f : S

k−1

→GL

n

(C) one obtains

a complex vector bundle E

f

→S

k

.

Example 1.9. Let us see how the tangent bundle TS

2

to S

2

can be described in

these terms. Consider the vector ﬁeld v

+

on D

2

+

obtained by choosing a tangent

vector at the north pole and transporting this along each

meridian circle in such a way as to maintain a constant

angle with the meridian. By reﬂecting across the plane

of the equator we obtain a corresponding vector ﬁeld v

−

on D

2

−

. Rotating each vector v

+

and v

−

by 90 degrees

counterclockwise when viewed from outside the sphere,

we obtain vector ﬁelds w

+

and w

−

on D

2

+

and D

2

−

. The

vector ﬁelds v

:

and w

:

give trivializations of TS

2

over

D

2

:

, allowing us to identify these two halves of TS

2

with D

2

:

×R

2

. We can then recon-

struct TS

2

as a quotient of the disjoint union of D

2

+

×R

2

and D

2

−

×R

2

by identifying

along S

1

×R

2

using the clutching function f which reads off the coordinates of the

v

−

, w

−

vectors in the v

+

, w

+

coordinate system, since the deﬁnition of a clutching

function was that it identiﬁes (x, v) ∈ ∂D

k

−

×R

n

with (x, f(x)(v)) ∈ ∂D

k

+

×R

n

. Thus

f rotates (v

+

, w

+

) to (v

−

, w

−

). As we go around the equator S

1

in the counter-

clockwise direction when viewed from above, starting from a point where the two

trivializations agree, the angle of rotation increases from 0 to 4π. If we parametrize

S

1

by the angle θ from the starting point, then f(θ) is rotation by 2θ.

Example 1.10. Let us ﬁnd a clutching function for the canonical complex line bundle

over CP

1

= S

2

. The space CP

1

is the quotient of C

2

− ¦0¦ under the equivalence

relation (z

0

, z

1

) ∼ λ(z

0

, z

1

) for λ ∈ C−¦0¦. Denote the equivalence class of (z

0

, z

1

)

by [z

0

, z

1

]. We can also write points of CP

1

as ratios z = z

0

/z

1

∈ C ∪ ¦∞¦ = S

2

. If

Classifying Vector Bundles Section 1.2 23

we do this, then points in the disk D

2

0

inside the unit circle S

1

⊂ C can be expressed

uniquely in the form [z

0

/z

1

, 1] = [z, 1] with ¦z¦ ≤ 1, and points in the disk D

2

∞

outside S

1

can be written uniquely in the form [1, z

1

/z

0

] = [1, z

−1

] with ¦z

−1

¦ ≤ 1.

Over D

2

0

a section of the canonical line bundle is given by [z, 1]

(z, 1) and over

D

2

∞

a section is [1, z

−1

]

(1, z

−1

). These sections determine trivializations of the

canonical line bundle over these two disks, and over their common boundary S

1

we

pass from the D

2

∞

trivialization to the D

2

0

trivialization by multiplying by z. Thus if

we take D

2

∞

as D

2

+

and D

2

0

as D

2

−

then the canonical line bundle has the clutching

function f : S

1

→GL

1

(C) deﬁned by f(z) = (z). If we had taken D

2

+

to be D

2

0

rather

than D

2

∞

then the clutching function would have been f(z) = (z

−1

).

A basic property of the construction of bundles E

f

→S

k

via clutching functions

f : S

k−1

→GL

n

(R) is that E

f

≈ E

g

if f and g are homotopic. For if we have a homo-

topy F : S

k−1

×I→GL

n

(R) from f to g, then the same sort of clutching construction

can be used to produce a vector bundle E

F

→S

k

×I that restricts to E

f

over S

k

×¦0¦

and E

g

over S

k

×¦1¦. Hence E

f

and E

g

are isomorphic by Proposition 1.7. Thus if

we denote by [X, Y] the set of homotopy classes of maps X→Y , then the association

f

E

f

gives a well-deﬁned map Φ: [S

k−1

, GL

n

(R)] →Vect

n

(S

k

).

There is a similarly-deﬁned map Φ for complex vector bundles. This turns out to

have slightly better behavior than in the real case, so we will prove the following basic

result about the complex form of Φ before dealing with the real case.

Proposition 1.11. The map Φ: [S

k−1

, GL

n

(C)]→Vect

n

C

(S

k

) which sends a clutching

function f to the vector bundle E

f

is a bijection.

Proof: We construct an inverse Ψ to Φ. Given an n dimensional vector bundle

p: E→S

k

, its restrictions E

+

and E

−

over D

k

+

and D

k

−

are trivial since D

k

+

and D

k

−

are contractible. Choose trivializations h

:

: E

:

→D

k

:

×C

n

. Then h

+

h

−1

−

deﬁnes a map

S

k−1

→GL

n

(C), whose homotopy class is by deﬁnition Ψ(E) ∈ [S

k−1

, GL

n

(C)]. To

see that Ψ(E) is well-deﬁned, note ﬁrst that any two choices of h

:

differ by a map

D

k

:

→GL

n

(C). Since D

k

:

is contractible, such a map is homotopic to a constant map

by composing with a contraction of D

k

:

. Now we need the fact that GL

n

(C) is path-

connected. The elementary row operation of modifying a matrix in GL

n

(C) by adding

a scalar multiple of one row to another row can be realized by a path in GL

n

(C), just

by inserting a factor of t in front of the scalar multiple and letting t go from 0 to 1.

By such operations any matrix in GL

n

(C) can be diagonalized. The set of diagonal

matrices in GL

n

(C) is path-connected since it is homeomorphic to the product of n

copies of C −¦0¦.

From this we conclude that h

+

and h

−

are unique up to homotopy. Hence the

composition h

+

h

−1

−

: S

k−1

→GL

n

(C) is also unique up to homotopy, which means that

Ψ is a well-deﬁned map Vect

n

C

(S

k

)→[S

k−1

, GL

n

(C)]. It is clear that Ψ and Φ are

inverses of each other. L¦

24 Chapter 1 Vector Bundles

Example 1.12. Every complex vector bundle over S

1

is trivial, since by the proposi-

tion this is equivalent to saying that [S

0

, GL

n

(C)] has a single element, or in other

words that GL

n

(C) is path-connected, which is the case as we saw in the proof of the

proposition.

Example 1.13. Let us show that the canonical line bundle H→CP

1

satisﬁes the rela-

tion (H⊗H)⊕1 ≈ H⊕H where 1 is the trivial one-dimensional bundle. This can be

seen by looking at the clutching functions for these two bundles, which are the maps

S

1

→GL

2

(C) given by

z

z

2

0

0 1

¸

and z

z 0

0 z

¸

More generally, let us show the formula E

fg

⊕n ≈ E

f

⊕E

g

for n dimensional vec-

tor bundles E

f

and E

g

over S

k

with clutching functions f, g : S

k−1

→GL

n

(C), where

fg is the clutching function obtained by pointwise matrix multiplication, fg(x) =

f(x)g(x).

The bundle E

f

⊕E

g

has clutching function the map f ⊕g : S

k−1

→GL

2n

(C) having

the matrices f(x) in the upper left n×n block and the matrices g(x) in the lower

right n×n block, the other two blocks being zero. Since GL

2n

(C) is path-connected,

there is a path α

t

∈ GL

2n

(C) from the identity matrix to the matrix of the trans-

formation which interchanges the two factors of C

n

×C

n

. Then the matrix product

(f ⊕11)α

t

(11 ⊕g)α

t

gives a homotopy from f ⊕g to fg ⊕11, which is the clutching

function for E

fg

⊕n.

The preceding analysis does not quite work for real vector bundles since GL

n

(R)

is not path-connected. We can see that there are at least two path-components since

the determinant function is a continuous surjection GL

n

(R)→R − ¦0¦ to a space

with two path-components. In fact GL

n

(R) has exactly two path-components, as

we will now show. Just as in the complex case we can construct a path from an

arbitrary matrix in GL

n

(R) to a diagonal matrix. Then by a path of diagonal matrices

we can make all the diagonal entries +1 or −1. Two −1’s represent a 180 degree

rotation of a plane so they can be replaced by +1’s via a path in GL

n

(R). This shows

that the subgroup GL

+

n

(R) consisting of matrices of positive determinant is path-

connected. This subgroup has index 2, and GL

n

(R) is the disjoint union of the two

cosets GL

+

n

(R) and αGL

+

n

(R) for α a ﬁxed matrix of determinant −1. The two cosets

are homeomorphic since the map β

αβ is a homeomorphism of GL

n

(R) with

inverse the map β

α

−1

β. Thus both cosets are path-connected and so GL

n

(R) has

two path-components.

The closest analogy with the complex case is obtained by considering oriented

real vector bundles. Recall from linear algebra that an orientation of a real vector

space is an equivalence class of ordered bases, two ordered bases being equivalent

if the invertible matrix taking the ﬁrst basis to the second has positive determinant.

Classifying Vector Bundles Section 1.2 25

An orientation of a real vector bundle p: E→B is a function assigning an orientation

to each ﬁber in such a way that near each point of B there is a local trivialization

h: p

−1

(U)→U×R

n

carrying the orientations of ﬁbers in p

−1

(U) to the standard ori-

entation of R

n

in the ﬁbers of U×R

n

. Another way of stating this condition would

be to say that the orientations of ﬁbers of E can be deﬁned by ordered n tuples of

independent local sections. Not all vector bundles can be given an orientation. For

example the M¨obius bundle is not orientable. This is because an oriented line bundle

over a paracompact base is always trivial since it has a canonical section formed by

the unit vectors having positive orientation.

Let Vect

n

+

(B) denote the set of isomorphism classes of oriented n dimensional

real vector bundles over B, where isomorphisms are required to preserve orientations.

The clutching construction deﬁnes a map Φ: [S

k−1

, GL

+

n

(R)] →Vect

n

+

(S

k

), and since

GL

+

n

(R) is path-connected, the argument from the complex case shows:

Proposition 1.14. The map Φ: [S

k−1

, GL

+

n

(R)] →Vect

n

+

(S

k

) is a bijection. L¦

To analyze Vect

n

(S

k

) let us introduce an ad hoc hybrid object Vect

n

0

(S

k

), the

n dimensional vector bundles over S

k

with an orientation speciﬁed in the ﬁber over

one point x

0

∈ S

k−1

, with the equivalence relation of isomorphism preserving the

orientation of the ﬁber over x

0

. We can choose the trivializations h

:

over D

k

:

to

carry this orientation to a standard orientation of R

n

, and then h

:

is unique up to

homotopy as before, so we obtain a bijection of Vect

n

0

(S

k

) with the homotopy classes

of maps S

k−1

→GL

n

(R) taking x

0

to GL

+

n

(R).

The map Vect

n

0

(S

k

)→Vect

n

(S

k

) that forgets the orientation over x

0

is a sur-

jection that is two-to-one except on vector bundles that have an automorphism (an

isomorphism from the bundle to itself) reversing the orientation of the ﬁber over x

0

,

where it is one-to-one.

When k = 1 there are just two homotopy classes of maps S

0

→GL

n

(R) taking

x

0

to GL

+

n

(R), represented by maps taking x

0

to the identity and the other point

of S

0

to either the identity or a reﬂection. The corresponding bundles are the trivial

bundle and, when n = 1, the M¨obius bundle, or the direct sum of the M¨obius bundle

with a trivial bundle when n > 1. These bundles all have automorphisms reversing

orientations of ﬁbers. We conclude that Vect

n

(S

1

) has exactly two elements for each

n ≥ 1. The nontrivial bundle is nonorientable since an orientable bundle over S

1

is

trivial, any two clutching functions S

0

→GL

+

n

(R) being homotopic.

When k > 1 the sphere S

k−1

is path-connected, so maps S

k−1

→GL

n

(R) tak-

ing x

0

to GL

+

n

(R) take all of S

k−1

to GL

+

n

(R). This implies that the natural map

Vect

n

+

(S

k

)→Vect

n

0

(S

k

) is a bijection. Thus every vector bundle over S

k

is orientable

and has exactly two different orientations, determined by an orientation in one ﬁber.

It follows that the map Vect

n

+

(S

k

)→Vect

n

(S

k

) is surjective and at most two-to-one.

26 Chapter 1 Vector Bundles

It is one-to-one on bundles having an orientation-reversing automorphism, and two-

to-one otherwise.

Inside GL

n

(R) is the orthogonal group O(n), the subgroup consisting of isome-

tries, or in matrix terms, the n×n matrices A with AA

T

= I , that is, matrices whose

columns form an orthonormal basis. The Gram-Schmidt orthogonalization process

provides a deformation retraction of GL

n

(R) onto the subspace O(n). Each step of

the process is either normalizing the i

th

vector of a basis to have length 1 or sub-

tracting a linear combination of the preceding vectors from the i

th

vector in order to

make it orthogonal to them. Both operations are realizable by a homotopy. This is

obvious for rescaling, and for the second operation one can just insert a scalar factor

of t in front of the linear combination of vectors being subtracted and let t go from 0

to 1. The explicit algebraic formulas show that both operations depend continuously

on the initial basis and have no effect on an orthonormal basis. Hence the sequence

of operations provides a deformation retraction of GL

n

(R) onto O(n). Restricting

to the component GL

+

n

(R), we have a deformation retraction of this component onto

the special orthogonal group SO(n), the orthogonal matrices of determinant +1.

This is the path-component of O(n) containing the identity matrix since deformation

retractions preserve path-components. The other component consists of orthogonal

matrices of determinant −1.

In the complex case the same argument applied to GL

n

(C) gives a deformation

retraction onto the unitary group U(n), the n×n matrices over C whose columns

form an orthonormal basis using the standard Hermitian inner product.

Composing maps S

k−1

→SO(n) with the inclusion SO(n)

GL

+

n

(R) gives a

function [S

k−1

, SO(n)]→[S

k−1

, GL

+

n

(R)]. This is a bijection since the deformation

retraction of GL

+

n

(R) onto SO(n) gives a homotopy of any map S

k−1

→GL

+

n

(R) to

a map into SO(n), and if two such maps to SO(n) are homotopic in GL

+

n

(R) then

they are homotopic in SO(n) by composing each stage of the homotopy with the

retraction GL

+

n

(R)→SO(n) produced by the deformation retraction.

The advantage of SO(n) over GL

+

n

(R) is that it is considerably smaller. For

example, SO(2) is just a circle, since orientation-preserving isometries of R

2

are

rotations, determined by the angle of rotation, a point in S

1

. Taking k = 2 as well,

we have for each integer m∈ Z the map f

m

: S

1

→S

1

, z

z

m

, which is the clutching

function for an oriented 2 dimensional vector bundle E

m

→S

2

. We encountered E

2

in

Example 1.9 as TS

2

, and E

1

is the real vector bundle underlying the canonical complex

line bundle over CP

1

= S

2

, as we saw in Example 1.10. It is a fact proved in every

introductory algebraic topology course that an arbitrary map f : S

1

→S

1

is homotopic

to one of the maps z

z

m

for some m uniquely determined by f , called the degree

of f . Thus via the bundles E

m

we have a bijection Vect

2

+

(S

2

) ≈ Z. Reversing the

orientation of E

m

changes it to E

−m

, so elements of Vect

2

(S

2

) correspond bijectively

Classifying Vector Bundles Section 1.2 27

with integers m≥ 0, via the bundles E

m

. Since U(1) is also homeomorphic to S

1

,we

obtain a similar bijection Vect

2

C

(S

2

) ≈ Z.

Another fact proved near the beginning of algebraic topology is that any two maps

S

k−1

→S

1

are homotopic if k > 2. (Such a map lifts to the covering space R of S

1

,

and R is contractible.) Hence every 2 dimensional vector bundle over S

k

is trivial

when k > 2, as is any complex line bundle over S

k

for k > 2.

It is a fact that any two maps S

2

→SO(n) or S

2

→U(n) are homotopic, so all

vector bundles over S

3

are trivial. This is not too hard to show using a little basic

machinery of algebraic topology, as we showin §??, where we also compute howmany

homotopy classes of maps S

3

→SO(n) and S

3

→U(n) there are.

For any space X the set [X, SO(n)] has a natural group structure coming from

the group structure in SO(n). Namely, the product of two maps f, g : X→SO(n) is

the map x

**f(x)g(x). Similarly [X, U(n)] is a group. For example the bijection
**

[S

1

, SO(2)] ≈ Z is a group isomorphism since for maps S

1

→S

1

, the product of

z

z

m

and z

z

n

is z

z

m

z

n

, and z

m

z

n

= z

m+n

.

The groups [S

i

, SO(n)] and [S

i

, U(n)] are isomorphic to the homotopy groups

π

i

SO(n) and π

i

U(n), special cases of the homotopy groups π

i

X deﬁned for all

spaces X and central to algebraic topology. To deﬁne π

i

X requires choosing a base-

point x

0

∈ X, and then π

i

X is the set of homotopy classes of maps S

i

→X taking a

chosen basepoint s

0

∈ S

i

to x

0

, where homotopies are also required to take s

0

to x

0

at all times. This basepoint data is needed in order to deﬁne a group structure in π

i

X

when i > 0. (When i = 0 there is no natural group structure in general.) However, the

group structure on π

i

X turns out to be independent of the choice of x

0

when X is

path-connected. Taking X = SO(n), there is an evident map π

i

SO(n)→[S

i

, SO(n)]

obtained by ignoring basepoint data. To see that this is surjective, suppose we are

given a map f : S

i

→SO(n). Since SO(n) is path-connected, there is a path α

t

in

SO(n) from the identity matrix to f(s

0

), and then the matrix product α

−1

t

f deﬁnes

a homotopy from f to a map S

i

→SO(n) taking s

0

to the identity matrix, which we

choose as the basepoint of SO(n). For injectivity of the map π

i

SO(n)→[S

i

, SO(n)]

suppose that f

0

and f

1

are two maps S

i

→SO(n) taking s

0

to the identity matrix,

and suppose f

t

is a homotopy from f

0

to f

1

that may not take basepoint to base-

point at all times. Then [f

t

(s

0

)]

−1

f

t

is a new homotopy from f

0

to f

1

that does take

basepoint to basepoint for all t . The same argument also applies for U(n).

Note to the reader: The rest of this chapter will not be needed until Chapter 3.

The Universal Bundle

We will showthat there is a special n dimensional vector bundle E

n

→G

n

with the

property that all n dimensional bundles over paracompact base spaces are obtainable

as pullbacks of this single bundle. When n = 1 this bundle will be just the canonical

line bundle over RP

∞

, deﬁned earlier. The generalization to n > 1 will consist in

28 Chapter 1 Vector Bundles

replacing RP

∞

, the space of 1 dimensional vector subspaces of R

∞

, by the space of

n dimensional vector subspaces of R

∞

.

First we deﬁne the Grassmann manifold G

n

(R

k

) for nonnegative integers n ≤ k.

As a set this is the collection of all n dimensional vector subspaces of R

k

, that is,

n dimensional planes in R

k

passing through the origin. To deﬁne a topology on

G

n

(R

k

) we ﬁrst deﬁne the Stiefel manifold V

n

(R

k

) to be the space of orthonormal

n frames in R

k

, in other words, n tuples of orthonormal vectors in R

k

. This is a

subspace of the product of n copies of the unit sphere S

k−1

, namely, the subspace

of orthogonal n tuples. It is a closed subspace since orthogonality of two vectors can

be expressed by an algebraic equation. Hence V

n

(R

k

) is compact since the product

of spheres is compact. There is a natural surjection V

n

(R

k

)→G

n

(R

k

) sending an

n frame to the subspace it spans, and G

n

(R

k

) is topologized by giving it the quotient

topology with respect to this surjection. So G

n

(R

k

) is compact as well. Later in this

section we will construct a ﬁnite CW complex structure on G

n

(R

k

) and in the process

show that it is Hausdorff and a manifold of dimension n(k −n).

The inclusions R

k

⊂ R

k+1

⊂ ··· give inclusions G

n

(R

k

) ⊂ G

n

(R

k+1

) ⊂ ···, and

we let G

n

(R

∞

) =

k

G

n

(R

k

). This is the set of all n dimensional vector subspaces of

the vector space R

∞

since if we choose a basis for a ﬁnite-dimensional subspace of

R

∞

, each basis vector will lie in some R

k

, hence there will be an R

k

containing all the

basis vectors and therefore the whole subspace. We give G

n

(R

∞

) the weak or direct

limit topology, so a set in G

n

(R

∞

) is open iff it intersects each G

n

(R

k

) in an open set.

There are canonical n dimensional vector bundles over G

n

(R

k

) and G

n

(R

∞

).

Deﬁne E

n

(R

k

) = ¦ (, v) ∈ G

n

(R

k

)×R

k

¦ v ∈ ¦. The inclusions R

k

⊂ R

k+1

⊂ ···

give inclusions E

n

(R

k

) ⊂ E

n

(R

k+1

) ⊂ ··· and we set E

n

(R

∞

) =

k

E

n

(R

k

), again with

the weak or direct limit topology.

Lemma 1.15. The projection p: E

n

(R

k

)→G

n

(R

k

), p(, v) = , is a vector bundle.,

both for ﬁnite and inﬁnite k.

Proof: First suppose k is ﬁnite. For ∈ G

n

(R

k

), let π

: R

k

→ be orthogonal projec-

tion and let U

= ¦

¹

∈ G

n

(R

k

) ¦

¦π

(

¹

) has dimension n¦. In particular, ∈ U

. We

will show that U

is open in G

n

(R

k

) and that the map h: p

−1

(U

)→U

× ≈ U

×R

n

deﬁned by h(

¹

, v) = (

¹

, π

(v)) is a local trivialization of E

n

(R

k

).

For U

**to be open is equivalent to its preimage in V
**

n

(R

k

) being open. This

preimage consists of orthonormal frames v

1

, ··· , v

n

such that π

(v

1

), ··· , π

(v

n

)

are independent. Let A be the matrix of π

**with respect to the standard basis in
**

the domain R

k

and any ﬁxed basis in the range . The condition on v

1

, ··· , v

n

is

then that the n×n matrix with columns Av

1

, ··· , Av

n

have nonzero determinant.

Since the value of this determinant is obviously a continuous function of v

1

, ··· , v

n

,

it follows that the frames v

1

, ··· , v

n

yielding a nonzero determinant form an open

set in V

n

(R

k

).

Classifying Vector Bundles Section 1.2 29

It is clear that h is a bijection which is a linear isomorphism on each ﬁber. We

need to check that h and h

−1

are continuous. For

¹

∈ U

**there is a unique invertible
**

linear map L

¹ : R

k

→R

k

restricting to π

on

¹

and the identity on

⊥

= Ker π

. We

claim that L

**¹ , regarded as a k×k matrix, depends continuously on
**

¹

. Namely, we

can write L

¹ as a product AB

−1

where:

— B sends the standard basis to v

1

, ··· , v

n

, v

n+1

, ··· , v

k

with v

1

, ··· , v

n

an or-

thonormal basis for

¹

and v

n+1

, ··· , v

k

a ﬁxed basis for

⊥

.

— A sends the standard basis to π

(v

1

), ··· , π

(v

n

), v

n+1

, ··· , v

k

.

Both A and B depend continuously on v

1

, ··· , v

n

. Since matrix multiplication and

matrix inversion are continuous operations (think of the ‘classical adjoint’ formula for

the inverse of a matrix), it follows that the product L

¹ = AB

−1

depends continuously

on v

1

, ··· , v

n

. But since L

¹ depends only on

¹

, not on the basis v

1

, ··· , v

n

for

¹

, it

follows that L

¹ depends continuously on

¹

since G

n

(R

k

) has the quotient topology

from V

n

(R

k

). Since we have h(

¹

, v) = (

¹

, π

(v)) = (

¹

, L

¹ (v)), we see that h is

continuous. Similarly, h

−1

(

¹

, w) = (

¹

, L

−1

¹ (w)) and L

−1

¹ depends continuously on

¹

, matrix inversion being continuous, so h

−1

is continuous.

This ﬁnishes the proof for ﬁnite k. When k = ∞ one takes U

to be the union of

the U

**’s for increasing k. The local trivializations h constructed above for ﬁnite k
**

then ﬁt together to give a local trivialization over this U

**, continuity being automatic
**

since we use the weak topology. L¦

We will mainly be interested in the case k = ∞ now, and to simplify notation we

will write G

n

for G

n

(R

∞

) and E

n

for E

n

(R

∞

). As earlier in this section, we use the

notation [X, Y] for the set of homotopy classes of maps f : X→Y .

Theorem 1.16. For paracompact X, the map [X, G

n

]→Vect

n

(X), [f]

f

∗

(E

n

),

is a bijection.

Thus, vector bundles over a ﬁxed base space are classiﬁed by homotopy classes

of maps into G

n

. Because of this, G

n

is called the classifying space for n dimensional

vector bundles and E

n

→G

n

is called the universal bundle.

As an example of how a vector bundle could be isomorphic to a pullback f

∗

(E

n

),

consider the tangent bundle to S

n

. This is the vector bundle p: E→S

n

where E =

¦ (x, v) ∈ S

n

×R

n+1

¦ x ⊥ v ¦. Each ﬁber p

−1

(x) is a point in G

n

(R

n+1

), so we have

a map S

n

→G

n

(R

n+1

), x

p

−1

(x). Via the inclusion R

n+1

R

∞

we can view this

as a map f : S

n

→G

n

(R

∞

) = G

n

, and E is exactly the pullback f

∗

(E

n

).

Proof of 1.16: The key observation is the following: For an n dimensional vector

bundle p: E→X, an isomorphism E ≈ f

∗

(E

n

) is equivalent to a map g : E→R

∞

that

is a linear injection on each ﬁber. To see this, suppose ﬁrst that we have a map

f : X→G

n

and an isomorphism E ≈ f

∗

(E

n

). Then we have a commutative diagram

30 Chapter 1 Vector Bundles

−

−

→

−

−

→

−−−−−→ −−−−−→

E E f E

−−−−−→

−

−

−

−

−

→

X G

f

f

p

π ∗

∼

( )

n

n n

R ≈

∞

where π(, v) = v. The composition across the top row is a map g : E→R

∞

that is

a linear injection on each ﬁber, since both

`

f and π have this property. Conversely,

given a map g : E→R

∞

that is a linear injection on each ﬁber, deﬁne f : X→G

n

by

letting f(x) be the n plane g(p

−1

(x)). This clearly yields a commutative diagram

as above.

To show surjectivity of the map [X, G

n

] →Vect

n

(X), suppose p: E→X is an

n dimensional vector bundle. Let ¦U

α

¦ be an open cover of X such that E is trivial

over each U

α

. By Lemma 1.21 in the Appendix to this chapter there is a countable

open cover ¦U

i

¦ of X such that E is trivial over each U

i

, and there is a partition

of unity ¦ϕ

i

¦ with ϕ

i

supported in U

i

. Let g

i

: p

−1

(U

i

)→R

n

be the composition

of a trivialization p

−1

(U

i

)→U

i

×R

n

with projection onto R

n

. The map (ϕ

i

p)g

i

,

v

ϕ

i

(p(v))g

i

(v), extends to a map E→R

n

that is zero outside p

−1

(U

i

). Near

each point of X only ﬁnitely many ϕ

i

’s are nonzero, and at least one ϕ

i

is nonzero,

so these extended (ϕ

i

p)g

i

’s are the coordinates of a map g : E→(R

n

)

∞

= R

∞

that is

a linear injection on each ﬁber.

For injectivity, if we have isomorphisms E ≈ f

∗

0

(E

n

) and E ≈ f

∗

1

(E

n

) for two

maps f

0

, f

1

: X→G

n

, then these give maps g

0

, g

1

: E→R

∞

that are linear injections

on ﬁbers, as in the ﬁrst paragraph of the proof. We claim g

0

and g

1

are homotopic

through maps g

t

that are linear injections on ﬁbers. If this is so, then f

0

and f

1

will

be homotopic via f

t

(x) = g

t

(p

−1

(x)).

The ﬁrst step in constructing a homotopy g

t

is to compose g

0

with the homotopy

L

t

: R

∞

→R

∞

deﬁned by L

t

(x

1

, x

2

, ···) = (1 −t)(x

1

, x

2

, ···) +t(x

1

, 0, x

2

, 0, ···). For

each t this is a linear map whose kernel is easily computed to be 0, so L

t

is injective.

Composing the homotopy L

t

with g

0

moves the image of g

0

into the odd-numbered

coordinates. Similarly we can homotope g

1

into the even-numbered coordinates. Still

calling the new g’s g

0

and g

1

, let g

t

= (1 − t)g

0

+ tg

1

. This is linear and injective

on ﬁbers for each t since g

0

and g

1

are linear and injective on ﬁbers. L¦

An explicit calculation of [X, G

n

] is usually beyond the reach of what is possible

technically, so this theorem is of limited usefulness in enumerating all the different

vector bundles over a given base space. Its importance is due more to its theoretical

implications. Among other things, it can reduce the proof of a general statement to

the special case of the universal bundle. For example, it is easy to deduce that vector

bundles over a paracompact base have inner products, since the bundle E

n

→G

n

has

an obvious inner product obtained by restricting the standard inner product in R

∞

to each n plane, and this inner product on E

n

induces an inner product on every

pullback f

∗

(E

n

).

Classifying Vector Bundles Section 1.2 31

The preceding constructions and results hold equally well for vector bundles over

C, with G

n

(C

k

) the space of n dimensional C linear subspaces of C

k

, and so on. In

particular, the proof of Theorem 1.16 translates directly to complex vector bundles,

showing that Vect

n

C

(X) ≈ [X, G

n

(C

∞

)].

There is also a version of Theorem 1.16 for oriented vector bundles. Let

`

G

n

(R

k

)

be the space of oriented n planes in R

k

, the quotient space of V

n

(R

k

) obtained by

identifying two n frames when they determine the same oriented subspace of R

k

.

Forming the union over increasing k we then have the space

`

G

n

(R

∞

). The universal

oriented bundle

`

E

n

(R

∞

) over

`

G

n

(R

∞

) consists of pairs (, v) ∈

`

G

n

(R

∞

)×R

∞

with

v ∈ . In other words,

`

E

n

(R

∞

) is the pullback of E

n

(R

∞

) via the natural projection

`

G

n

(R

∞

)→G

n

(R

∞

). Small modiﬁcations in the proof that Vect

n

(X) ≈ [X, G

n

(R

∞

)]

show that Vect

n

+

(X) ≈ [X,

`

G

n

(R

∞

)].

Both G

n

(R

∞

) and

`

G

n

(R

∞

) are path-connected since Vect

n

(X) and Vect

n

+

(X)

have a single element when X is a point. The natural projection

`

G

n

(R

∞

)→G

n

(R

∞

)

obtained by ignoring orientations is two-to-one, and readers familiar with the no-

tion of a covering space will have no trouble in recognizing that this two-to-one

projection map is a covering space, using for example the local trivializations con-

structed in Lemma 1.15. In fact

`

G

n

(R

∞

) is the universal cover of G

n

(R

∞

) since it

is simply-connected, because of the triviality of Vect

n

+

(S

1

) ≈ [S

1

,

`

G

n

(R

∞

)]. One can

also observe that a vector bundle E ≈ f

∗

(E

n

(R

∞

) is orientable iff its classifying map

f : X→G

n

(R

∞

) lifts to a map

`

f : X→

`

G

n

(R

∞

), and in fact orientations of E corre-

spond bijectively with lifts

`

f .

Cell Structures on Grassmannians

Since Grassmann manifolds play a fundamental role in vector bundle theory, it

would be good to have a better grasp on their topology. Here we show that G

n

(R

∞

)

has the structure of a CWcomplex with each G

n

(R

k

) a ﬁnite subcomplex. We will also

see that G

n

(R

k

) is a closed manifold of dimension n(k−n). Similar statements hold

in the complex case as well, with G

n

(C

k

) a closed manifold of dimension 2n(k −n).

For a start let us show that G

n

(R

k

) is Hausdorff, since we will need this fact later

when we construct the CW structure. Given two n planes and

¹

in G

n

(R

k

), it

sufﬁces to ﬁnd a continuous map f : G

n

(R

k

)→R taking different values on and

¹

.

For a vector v ∈ R

k

let f

v

() be the length of the orthogonal projection of v onto .

This is a continuous function of since if we choose an orthonormal basis v

1

, ··· , v

n

for then f

v

() =

¸

(v · v

1

)

2

+ ··· + (v · v

n

)

2

1/2

, which is certainly continuous in

v

1

, ··· , v

n

hence in since G

n

(R

k

) has the quotient topology from V

n

(R

k

). Now for

an n plane

¹

≠ choose v ∈ −

¹

, and then f

v

() = ¦v¦ > f

v

(

¹

).

There is a nice description of the cells in the CW structure on G

n

(R

k

) in terms

of the familiar concept of echelon form for matrices. Recall that any n×k matrix A

can be put into an echelon form by a ﬁnite sequence of elementary row operations

32 Chapter 1 Vector Bundles

consisting either adding a multiple of one row to another, multiplying a row by a

nonzero scalar, or permuting two rows. In the standard version of echelon form one

strives to make zeros in the lower left corner of the matrix, but for our purposes it will

be more convenient to make zeros in the upper right corner instead. Thus a matrix in

echelon form will look like the following, with the asterisks denoting entries that are

arbitrary numbers:

∗ ∗

∗ ∗ ∗

1 0

0 1 0 0 0 0 0

∗ ∗ ∗ 0 0 1 0 0 0 0

∗ ∗ ∗ ∗ ∗ 0 0 0 1 0

0 0 0 0 0 0

We are assuming that our given n×k matrix A has rank n and that n ≤ k. The

shape of the echelon form is speciﬁed by which columns contain the special entries

1, say in the columns numbered σ

1

< ··· < σ

n

. The n tuple (σ

1

, ··· , σ

n

) is called

the Schubert symbol σ(A). To see that this depends only on the givne matrix A and

not any particular reduction of A to echelon form, consider the n plane

A

in R

k

spanned by the rows of A. This is the same as the n plane spanned by the rows of

an echelon form of A since elementary row operations do not change the subspace

spanned by the rows. Let p

i

: R

k

→R

n−i

be projection onto the last n−i coordinates.

As i goes from 0 to k the dimension of the subspace p

i

(

A

) of R

k

decreases from

n to 0 in n steps and the decreases occur precisely when i hits one of the values σ

j

,

as is apparent from the echelon form since p

i

(

A

) has basis the rows of the matrix

obtained by deleting the ﬁrst i columns of the echelon form. Thus the Schubert

symbol only depends on the n plane spanned by the rows of A.

In fact the echelon form itself depends only on the n plane spanned by the rows.

To see this, consider projection onto the n coordinates of R

k

given by the numbers

σ

i

. This projection is surjective when restricted to the n plane, hence is also injective,

and the n rows of the echelon form are the vectors in the n plane that project to the

standard basis vectors.

As noted above, the numbers σ

j

in the Schubert symbol of an n plane are the

numbers i for which the dimension of the image p

i

() decreases as i goes from 0

to n. Equivalently, these are the i’s for which the dimension of the kernel of the

restriction p

i

¦

¦ increases. This kernel is just ∩R

i

where R

i

is the kernel of p

i

, the

subspace of R

k

spanned by the ﬁrst i standard basis vectors. This gives an alternative

deﬁnition of the Schubert symbol σ().

Given a Schubert symbol σ one can consider the set e(σ) of all n planes in R

k

having σ as their Schubert symbol. In terms of echelon forms, the various n planes

in e(σ) are parametrized by the arbitrary entries in the echelon form. There are σ

i

−i

of these entries in the i

th

row, for a total of (ß

1

−1)+··· (σ

n

−n) entries. Thus e(σ)

is homeomorphic to a euclidean space of this dimension, or equivalently an open cell.

Classifying Vector Bundles Section 1.2 33

Proposition 1.17. The cells e(σ) are the cells of a CW structure on G

n

(R

k

).

For example G

2

(R

4

) has six cells corresponding to the Schubert symbols (1, 2),

(1, 3), (1, 4), (2, 3), (2, 4), (3, 4), and these cells have dimensions 0, 1, 2, 2, 3, 4

respectively. In general the number of cells of G

n

(R

k

) is

k

n

**, the number of ways of
**

choosing the n distinct numbers σ

i

≤ k.

Proof: Our main task will be to ﬁnd a characteristic map for e(σ). This is a map

from a closed ball of the same dimension as e(σ) into G

n

(R

k

) whose restriction

to the interior of the ball is a homeomorphism onto e(σ). From the echelon forms

described above it is not clear how to do this, so we will use a slightly different sort

of echelon form. We allow the special 1’s to be arbitrary nonzero numbers and we

allow the entries below these 1’s to be nonzero. Then we impose the condition that

the rows be orthonormal and that the last nonzero entry in each row be positive. Let

us call this an orthonormal echelon form. Once again there is a unique orthonormal

echelon form for each n plane since if we let

i

denote the subspace of spanned

by the ﬁrst i rows of the standard echelon form, or in other words

i

= ∩ R

σ

i

,

then there is a unique unit vector in

i

orthogonal to

i−1

and having positive σ

i

th

coordinate.

The i

th

row of the orthonormal echelon form then belongs to the hemisphere H

i

in the unit sphere S

σ

i

−1

⊂ R

σ

i

⊂ R

k

consisting of unit vectors with non-negative σ

i

th

coordinate. In the Stiefel manifold V

n

(R

k

) let E(σ) be the subspace of orthonormal

frames (v

1

, ··· , v

n

) ∈ (S

k−1

)

n

such that v

i

∈ H

i

for each i. We claim that E(σ) is

homeomorphic to a closed ball. To prove this the main step is to show that the pro-

jection π : E(σ)→H

1

, π(v

1

, ··· , v

n

) = v

1

, is a trivial ﬁber bundle. This is equivalent

to ﬁnding a projection p: E(σ)→π

−1

(v

0

) which is a homeomorphism on ﬁbers of

π, where v

0

= (0, ··· , 0, 1) ∈ R

σ

1

⊂ R

k

, since the map π×p: E(σ)→H

1

×π

−1

(v

0

) is

then a continuous bijection of compact Hausdorff spaces, hence a homeomorphism.

The map p: π

−1

(v)→π

−1

(v

0

) is obtained by applying the rotation ρ

v

of R

k

that

takes v to v

0

and ﬁxes the (k − 2) dimensional subspace orthogonal to v and v

0

.

This rotation takes H

i

to itself for i > 1 since it affects only the ﬁrst σ

1

coordinates

of vectors in R

k

. Hence p takes π

−1

(v) onto π

−1

(v

0

).

The ﬁber π

−1

(v

0

) can be identiﬁed with E(σ

¹

) for σ

¹

= (σ

2

−1, ··· , σ

n

−1). By

induction on n this is homeomorphic to a closed ball of dimension (σ

2

−2) +··· +

(σ

n

−n), so E(σ) is a closed ball of dimension (σ

1

−1)+···+(σ

n

−n). The boundary

of this ball consists of points in E(σ) having v

i

in ∂H

i

for at least one i. This too

follows by induction since the rotation ρ

v

takes ∂H

i

to itself for i > 1.

The natural map E(σ)→G

n

sending an orthonormal n tuple to the n plane it

spans takes the interior of the ball E(σ) to e(σ) bijectively. Since G

n

has the quotient

topology from V

n

, the map int E(σ)→e(σ) is a homeomorphism. The boundary of

34 Chapter 1 Vector Bundles

E(σ) maps to cells e(σ

¹

) of G

n

where σ

¹

is obtained from σ by decreasing some

σ

i

’s, so these cells e(σ

¹

) have lower dimension than e(σ).

To see that the maps E(σ)→G

n

(R

k

) for the cells e(σ) are the characteristic

maps for a CW structure on G

n

(R

k

) we can argue as follows. Let X

i

be the union of

the cells e(σ) in G

n

(R

k

) having dimension at most i. Suppose by induction on i that

X

i

is a CWcomplex with these cells. Attaching the (i+1) cells e(σ) of X

i+1

to X

i

via

the maps ∂E(σ)→X

i

produces a CW complex Y and a natural continuous bijection

Y→X

i+1

. Since Y is a ﬁnite CW complex it is compact, and X

i+1

is Hausdorff as a

subspace of G

n

(R

k

), so the map Y→X

i+1

is a homeomorphism and X

i+1

is a CW

complex, ﬁnishing the induction. Thus we have a CW structure on G

n

(R

k

). L¦

Since the inclusions G

n

(R

k

) ⊂ G

n

(R

k+1

) for varying k are inclusions of subcom-

plexes and G

n

(R

∞

) has the weak topology with respect to these subspaces, it follows

that we have also a CW structure on G

n

(R

∞

).

Similar constructions work to give CW structures on complex Grassmann mani-

folds, but here e(σ) will be a cell of dimension (2σ

1

−2)+(2σ

2

−4)+···+(2σ

n

−2n).

The hemisphere H

i

is deﬁned to be the subspace of the unit sphere S

2σ

i

−1

in C

σ

i

con-

sisting of vectors whose σ

i

th

coordinate is real and nonnegative, so H

i

is a ball of

dimension 2σ

i

−2. The transformation ρ

v

∈ SU(k) is uniquely determined by spec-

ifying that it takes v to v

0

and ﬁxes the orthogonal (k − 2) dimensional complex

subspace, since an element of U(2) of determinant 1 is determined by where it sends

one unit vector.

The highest-dimensional cell of G

n

(R

k

) is e(σ) for σ = (k−n+1, k−n+2, ··· , k),

of dimension n(k−n), so this is the dimension of G

n

(R

k

). Near points in these top-

dimensional cells G

n

(R

k

) is a manifold. But G

n

(R

k

) is homogeneous in the sense that

given any two points in G

n

(R

k

) there is a homeomorphism G

n

(R

k

)→G

n

(R

k

) taking

one point to the other, namely, the homeomorphism induced by an invertible linear

map R

k

→R

k

taking one n plane to the other. From this homogeneity it follows that

G

n

(R

k

) is a manifold near all points. Since it is compact, it is a closed manifold.

There is a natural inclusion i : G

n

G

n+1

, i() = R×j() where j : R

∞

→R

∞

is

the embedding j(x

1

, x

2

, ···) = (0, x

1

, x

2

, ···). If σ() = (σ

1

, ··· , σ

n

) then σ(i()) =

(1, σ

1

+1, ··· , σ

n

+1), so i takes cells of G

n

to cells of G

n+1

of the same dimension,

making i(G

n

) a subcomplex of G

n+1

. Identifying G

n

with the subcomplex i(G

n

),

we obtain an increasing sequence of CW complexes G

1

⊂ G

2

⊂ ··· whose union

G

∞

=

n

G

n

is therefore also a CW complex. Similar remarks apply as well in the

complex case.

Appendix: Paracompactness 35

Appendix: Paracompactness

A Hausdorff space X is paracompact if for each open cover ¦U

α

¦ of X there

is a partition of unity ¦ϕ

β

¦ subordinate to the cover. This means that the ϕ

β

’s are

maps X→I such that each ϕ

β

has support (the closure of the set where ϕ

β

≠ 0)

contained in some U

α

, each x ∈ X has a neighborhood in which only ﬁnitely many

ϕ

β

’s are nonzero, and

¸

β

ϕ

β

= 1. An equivalent deﬁnition which is often given is

that X is Hausdorff and every open cover of X has a locally ﬁnite open reﬁnement.

The ﬁrst deﬁnition clearly implies the second by taking the cover ¦ϕ

−1

β

(0, 1]¦. For the

converse, see [Dugundji] or [Lundell-Weingram]. It is the former deﬁnition which is

most useful in algebraic topology, and the fact that the two deﬁnitions are equivalent

is rarely if ever needed. So we shall use the ﬁrst deﬁnition.

A paracompact space X is normal, for let A

1

and A

2

be disjoint closed sets in X,

and let ¦ϕ

β

¦ be a partition of unity subordinate to the cover ¦X−A

1

, X−A

2

¦. Let ϕ

i

be the sum of the ϕ

β

’s which are nonzero at some point of A

i

. Then ϕ

i

(A

i

) = 1, and

ϕ

1

+ϕ

2

≤ 1 since no ϕ

β

can be a summand of both ϕ

1

and ϕ

2

. Hence ϕ

−1

1

(

1

/

2

, 1]

and ϕ

−1

2

(

1

/

2

, 1] are disjoint open sets containing A

1

and A

2

, respectively.

Most of the spaces one meets in algebraic topology are paracompact, including:

(1) compact Hausdorff spaces

(2) unions of increasing sequences X

1

⊂ X

2

⊂ ··· of compact Hausdorff spaces X

i

,

with the weak or direct limit topology (a set is open iff it intersects each X

i

in an

open set)

(3) CW complexes

(4) metric spaces

Note that (2) includes (3) for CW complexes with countably many cells, since such

a CW complex can be expressed as an increasing union of ﬁnite subcomplexes. Using

(1) and (2), it can be shown that many manifolds are paracompact, for example R

n

.

The next three propositions verify that the spaces in (1), (2), and (3) are paracom-

pact.

Proposition 1.18. A compact Hausdorff space X is paracompact.

Proof: Let ¦U

α

¦ be an open cover of X. Since X is normal, each x ∈ X has an open

neighborhood V

x

with closure contained in some U

α

. By Urysohn’s lemma there is a

map ϕ

x

: X→I with ϕ

x

(x) = 1 and ϕ

x

(X −V

x

) = 0. The open cover ¦ϕ

−1

x

(0, 1]¦ of

X contains a ﬁnite subcover, and we relabel the corresponding ϕ

x

’s as ϕ

β

’s. Then

¸

β

ϕ

β

(x) > 0 for all x, and we obtain the desired partition of unity subordinate to

¦U

α

¦ by normalizing each ϕ

β

by dividing it by

¸

β

ϕ

β

. L¦

Proposition 1.19. If X is the direct limit of an increasing sequence X

1

⊂ X

2

⊂ ···

of compact Hausdorff spaces X

i

, then X is paracompact.

36 Chapter 1 Vector Bundles

Proof: A preliminary observation is that X is normal. To show this, it sufﬁces to ﬁnd

a map f : X→I with f(A) = 0 and f(B) = 1 for any two disjoint closed sets A and B.

Such an f can be constructed inductively over the X

i

’s, using normality of the X

i

’s.

For the induction step one has f deﬁned on the closed set X

i

∪(A∩X

i+1

)∪(B∩X

i+1

)

and one extends over X

i+1

by Tietze’s theorem.

To prove that X is paracompact, let an open cover ¦U

α

¦ be given. Since X

i

is

compact Hausdorff, there is a ﬁnite partition of unity ¦ϕ

ij

¦ on X

i

subordinate to

¦U

α

∩X

i

¦. Using normality of X, extend each ϕ

ij

to a map ϕ

ij

: X→I with support

in the same U

α

. Let σ

i

=

¸

j

ϕ

ij

. This sum is 1 on X

i

, so if we normalize each ϕ

ij

by dividing it by max¦

1

/

2

, σ

i

¦, we get new maps ϕ

ij

with σ

i

= 1 in a neighborhood

V

i

of X

i

. Let ψ

ij

= max¦0, ϕ

ij

−

¸

k<i

σ

k

¦. Since 0 ≤ ψ

ij

≤ ϕ

ij

, the collection ¦ψ

ij

¦

is subordinate to ¦U

α

¦. In V

i

all ψ

kj

’s with k > i are zero, so each point of X has a

neighborhood in which only ﬁnitely many ψ

ij

’s are nonzero. For each x ∈ X there

is a ψ

ij

with ψ

ij

(x) > 0, since if ϕ

ij

(x) > 0 and i is minimal with respect to this

condition, then ψ

ij

(x) = ϕ

ij

(x). Thus when we normalize the collection ¦ψ

ij

¦ by

dividing by

¸

i,j

ψ

ij

we obtain a partition of unity on X subordinate to ¦U

α

¦. L¦

Proposition 1.20. Every CW complex is paracompact.

Proof: Given an open cover ¦U

α

¦ of a CW complex X, suppose inductively that we

have a partition of unity ¦ϕ

β

¦ on X

n

subordinate to the cover ¦U

α

∩ X

n

¦. For a

cell e

n+1

γ

with characteristic map Φ

γ

: D

n+1

→X, ¦ϕ

β

Φ

γ

¦ is a partition of unity on

S

n

= ∂D

n+1

. Since S

n

is compact, only ﬁnitely many of these compositions ϕ

β

Φ

γ

can

be nonzero, for ﬁxed γ. We extend these functions ϕ

β

Φ

γ

over D

n+1

by the formula

ρ

ε

(r)ϕ

β

Φ

γ

(x) using ‘spherical coordinates’ (r, x) ∈ I ×S

n

on D

n+1

, where ρ

ε

: I→I

is 0 on [0, 1−ε] and 1 on [1−

ε

/

2

, 1]. If ε = ε

γ

is chosen small enough, these extended

functions ρ

ε

ϕ

β

Φ

γ

will be subordinate to the cover ¦Φ

−1

γ

(U

α

)¦. Let ¦ψ

γj

¦ be a ﬁnite

partition of unity on D

n+1

subordinate to ¦Φ

−1

γ

(U

α

)¦. Then ¦ρ

ε

ϕ

β

Φ

γ

, (1−ρ

ε

)ψ

γj

¦ is

a partition of unity on D

n+1

subordinate to ¦Φ

−1

γ

(U

α

)¦. This partition of unity extends

the partition of unity ¦ϕ

β

Φ

γ

¦ on S

n

and induces an extension of ¦ϕ

β

¦ to a partition

of unity deﬁned on X

n

∪e

n+1

γ

and subordinate to ¦U

α

¦. Doing this for all (n+1) cells

e

n+1

γ

gives a partition of unity on X

n+1

. The local ﬁniteness condition continues to

hold since near a point in X

n

only the extensions of the ϕ

β

’s in the original partition

of unity on X

n

are nonzero, while in a cell e

n+1

γ

the only other functions that can be

nonzero are the ones coming from ψ

γj

’s. After we make such extensions for all n,

we obtain a partition of unity deﬁned on all of X and subordinate to ¦U

α

¦. L¦

Here is a technical fact about paracompact spaces that is occasionally useful:

Appendix: Paracompactness 37

Lemma 1.21. Given an open cover ¦U

α

¦ of the paracompact space X, there is a

countable open cover ¦V

k

¦ such that each V

k

is a disjoint union of open sets each

contained in some U

α

, and there is a partition of unity ¦ϕ

k

¦ with ϕ

k

supported in

V

k

.

Proof: Let ¦ϕ

β

¦ be a partition of unity subordinate to ¦U

α

¦. For each ﬁnite set S of

functions ϕ

β

let V

S

be the subset of X where all the ϕ

β

’s in S are strictly greater

than all the ϕ

β

’s not in S . Since only ﬁnitely many ϕ

β

’s are nonzero near any x ∈ X,

V

S

is deﬁned by ﬁnitely many inequalities among ϕ

β

’s near x, so V

S

is open. Also,

V

S

is contained in some U

α

, namely, any U

α

containing the support of any ϕ

β

∈ S ,

since ϕ

β

∈ S implies ϕ

β

> 0 on V

S

. Let V

k

be the union of all the open sets V

S

such

that S has k elements. This is clearly a disjoint union. The collection ¦V

k

¦ is a cover

of X since if x ∈ X then x ∈ V

S

where S = ¦ ϕ

β

¦ ϕ

β

(x) > 0¦.

For the second statement, let ¦ϕ

γ

¦ be a partition of unity subordinate to the

cover ¦V

k

¦, and let ϕ

k

be the sum of those ϕ

γ

’s supported in V

k

but not in V

j

for

j < k. L¦

Exercises

1. Show that the projection V

n

(R

k

)→G

n

(R

k

) is a ﬁber bundle with ﬁber O(n) by

showing that it is the orthonormal n frame bundle associated to the vector bundle

E

n

(R

k

)→G

n

(R

k

).

The idea of K–theory is to make the direct sumoperation on real or complex vector

bundles over a ﬁxed base space X into the addition operation in a group. There are

two slightly different ways of doing this, producing, in the case of complex vector

bundles, groups K(X) and

`

K(X) with K(X) ≈

`

K(X)⊕Z, and for real vector bundles,

groups KO(X) and

¯

KO(X) with KO(X) ≈

¯

KO(X)⊕Z. Complex K–theory turns out

to be somewhat simpler than real K–theory, so we concentrate on this case in the

present chapter.

Computing

`

K(X) even for simple spaces X requires some work. The case X = S

n

is the Bott Periodicity Theorem, which gives isomorphisms

`

K(S

n

) ≈

`

K(S

n+2

) for all

n, and more generally

`

K(X) ≈

`

K(S

2

X) where S

2

X is the double suspension of X.

This is a deep theorem, so it is not surprising that it has applications of real substance.

We give some of these in §2.3, notably:

(1) The nonexistence of division algebras over R in dimensions other than 1, 2, 4,

and 8, the dimensions of the real and complex numbers, quaternions, and Cayley

octonions.

(2) The nonparallelizability of spheres other than S

1

, S

3

, and S

7

.

The proof of the Bott Periodicity Theorem divides into two parts. The ﬁrst is the hard

technical work, proving an isomorphism K(X×S

2

) ≈ K(X)⊗K(S

2

). This takes about

ten pages, forming the bulk of §2.1. The other half is easier, being more formal in

nature, and this is contained in §2.2 where the cohomological aspects of K(X) are

the main focus.

The Functor K(X) Section 2.1 39

2.1. The Functor K(X)

Since we shall be dealing exclusively with complex vector bundles in this chapter,

let us take ‘vector bundle’ to mean ‘complex vector bundle’ unless otherwise speciﬁed.

Base spaces will always be assumed to be compact Hausdorff, so that the results of

the preceding chapter which have this hypothesis will be available to us.

For the purposes of K–theory it is convenient to take a slightly broader deﬁni-

tion of ‘vector bundle’ which allows the ﬁbers of a vector bundle p: E→X to be vec-

tor spaces of different dimensions. We still assume local trivializations of the form

h: p

−1

(U)→U×C

n

, so the dimensions of ﬁbers must be locally constant over X, but

if X is disconnected the dimensions of ﬁbers need not be globally constant.

Consider vector bundles over a ﬁxed base space X. The trivial n dimensional

vector bundle we write as ε

n

→X. Deﬁne two vector bundles E

1

and E

2

over X to be

stably isomorphic, written E

1

≈

s

E

2

, if E

1

⊕ε

n

≈ E

2

⊕ε

n

for some n. In a similar vein

we set E

1

∼ E

2

if E

1

⊕ε

m

≈ E

2

⊕ε

n

for some m and n. It is easy to see that both ≈

s

and ∼ are equivalence relations. On equivalence classes of either sort the operation

of direct sum is well-deﬁned, commutative, and associative. A zero element is the

class of ε

0

.

Proposition 2.1. If X is compact Hausdorff, then the set of ∼ equivalence classes of

vector bundles over X forms an abelian group with respect to ⊕.

This group is called

`

K(X).

Proof: Only the existence of inverses needs to be shown, which we do by showing

that for each vector bundle π : E→X there is a bundle E

¹

→X such that E⊕E

¹

≈ ε

m

for some m. If all the ﬁbers of E have the same dimension, this is Proposition 1.4.

In the general case let X

i

= ¦ x ∈ X¦

¦ dimπ

−1

(x) = i ¦. These X

i

’s are disjoint open

sets in X, hence are ﬁnite in number by compactness. By adding to E a bundle which

over each X

i

is a trivial bundle of suitable dimension we can produce a bundle whose

ﬁbers all have the same dimension. L¦

For the direct sumoperation on ≈

s

equivalence classes, only the zero element, the

class of ε

0

, can have an inverse since E⊕E

¹

≈

s

ε

0

implies E⊕E

¹

⊕ε

n

≈ ε

n

for some

n, which can only happen if E and E

¹

are 0 dimensional. However, even though

inverses do not exist, we do have the cancellation property that E

1

⊕E

2

≈

s

E

1

⊕E

3

implies E

2

≈

s

E

3

over a compact base space X, since we can add to both sides of

E

1

⊕E

2

≈

s

E

1

⊕E

3

a bundle E

¹

1

such that E

1

⊕E

¹

1

≈ ε

n

for some n.

Just as the positive rational numbers are constructed from the positive integers

by forming quotients a/b with the equivalence relation a/b = c/d iff ad = bc, so we

can formfor compact X an abelian group K(X) consisting of formal differences E−E

¹

of vector bundles E and E

¹

over X, with the equivalence relation E

1

− E

¹

1

= E

2

− E

¹

2

iff E

1

⊕E

¹

2

≈

s

E

2

⊕E

¹

1

. Verifying transitivity of this relation involves the cancellation

40 Chapter 2 Complex K–Theory

property, which is why compactness of X is needed. With the obvious addition rule

(E

1

−E

¹

1

)+(E

2

−E

¹

2

) = (E

1

⊕E

2

)−(E

¹

1

⊕E

¹

2

), K(X) is then a group. The zero element is

the equivalence class of E −E for any E, and the inverse of E −E

¹

is E

¹

−E. Note that

every element of K(X) can be represented as a difference E−ε

n

since if we start with

E −E

¹

we can add to both E and E

¹

a bundle E

¹¹

such that E

¹

⊕E

¹¹

≈ ε

n

for some n.

There is a natural homomorphism K(X)→

`

K(X) sending E − ε

n

to the ∼ class

of E. This is well-deﬁned since if E −ε

n

= E

¹

−ε

m

in K(X), then E⊕ε

m

≈

s

E

¹

⊕ε

n

,

hence E ∼ E

¹

. The map K(X)→

`

K(X) is obviously surjective, and its kernel consists of

elements E−ε

n

with E ∼ ε

0

, hence E ≈

s

ε

m

for some m, so the kernel consists of the

elements of the form ε

m

−ε

n

. This subgroup ¦ε

m

−ε

n

¦ of K(X) is isomorphic to Z.

In fact, restriction of vector bundles to a basepoint x

0

∈ X deﬁnes a homomorphism

K(X)→K(x

0

) ≈ Z which restricts to an isomorphism on the subgroup ¦ε

m

− ε

n

¦.

Thus we have a splitting K(X) ≈

`

K(X)⊕Z, depending on the choice of x

0

. The group

`

K(X) is sometimes called reduced, to distinguish it from K(X).

Ring Structure

Besides the additive structure in K(X) there is also a natural multiplication com-

ing from tensor product of vector bundles. For elements of K(X) represented by

vector bundles E

1

and E

2

their product in K(X) will be represented by the bundle

E

1

⊗E

2

, so for arbitrary elements of K(X) represented by differences of vector bun-

dles, their product in K(X) is deﬁned by the formula

(E

1

−E

¹

1

)(E

2

−E

¹

2

) = E

1

⊗E

2

−E

1

⊗E

¹

2

−E

¹

1

⊗E

2

+E

¹

1

⊗E

¹

2

It is routine to verify that this is well-deﬁned and makes K(X) into a commutative ring

with identity ε

1

, the trivial line bundle, using the basic properties of tensor product

of vector bundles described in §1.1. We can simplify notation by writing the element

ε

n

∈ K(X) just as n. This is consistent with familiar arithmetic rules. For example,

the product nE is the sum of n copies of E.

If we choose a basepoint x

0

∈ X, then the map K(X)→K(x

0

) obtained by re-

stricting vector bundles to their ﬁbers over x

0

is a ring homomorphism. Its kernel,

which can be identiﬁed with

`

K(X) as we have seen, is an ideal, hence also a ring in its

own right, though not necessarily a ring with identity.

The rings K(X) and

`

K(X) can be regarded as functors of X. A map f : X→Y in-

duces a map f

∗

: K(Y)→K(X), sending E−E

¹

to f

∗

(E)−f

∗

(E

¹

). This is a ring homo-

morphism since f

∗

(E

1

⊕E

2

) ≈ f

∗

(E

1

)⊕f

∗

(E

2

) and f

∗

(E

1

⊗E

2

) ≈ f

∗

(E

1

)⊗f

∗

(E

2

).

The functor properties (fg)

∗

= g

∗

f

∗

and 11

∗

= 11 as well as the fact that f = g

implies f

∗

= g

∗

all follow from the corresponding properties for pullbacks of vector

bundles. Similarly, we have induced maps f

∗

:

`

K(Y)→

`

K(X) with the same properties,

except that for f

∗

to be a ring homomorphism we must be in the category of base-

The Functor K(X) Section 2.1 41

pointed spaces and basepoint-preserving maps since our deﬁnition of multiplication

for

`

K required basepoints.

An external product µ : K(X)⊗K(Y)→K(X×Y) can be deﬁned by µ(a⊗b) =

p

∗

1

(a)p

∗

2

(b) where p

1

and p

2

are the projections of X×Y onto X and Y . The tensor

product of rings is a ring, with multiplication deﬁned by (a⊗b)(c ⊗d) = ac ⊗bd, and

µ is a ring homomorphism since µ((a⊗b)(c ⊗d)) = µ(ac ⊗bd) = p

∗

1

(ac)p

∗

2

(bd) =

p

∗

1

(a)p

∗

1

(c)p

∗

2

(b)p

∗

2

(d) = p

∗

1

(a)p

∗

2

(b)p

∗

1

(c)p

∗

2

(d) = µ(a⊗b)µ(c ⊗d).

Taking Y to be S

2

we have an external product

µ : K(X)⊗K(S

2

)→K(X×S

2

)

Most of the remainder of §2.1 will be devoted to showing that this map is an isomor-

phism. This is the essential core of the proof of Bott Periodicity.

The external product in ordinary cohomology is called ‘cross product’ and written

a×b, but to use this symbol for the K–theory external product might lead to confusion

with Cartesian product of vector bundles, which is quite different fromtensor product.

Instead we will sometimes use the notation a ∗b as shorthand for µ(a⊗b).

The Fundamental Product Theorem

The key result allowing the calculation of K(X) in nontrivial cases is a certain for-

mula that computes K(X×S

2

) in terms of K(X). For example, when X is a point this

will yield a calculation of K(S

2

). In the next section we will deduce other calculations,

in particular K(S

n

) for all n.

Let H be the canonical line bundle over S

2

= CP

1

. We showed in Example 1.13

that (H⊗H)⊕1 ≈ H⊕H. In K(S

2

) this is the formula H

2

+1 = 2H, so H

2

= 2H−1.

We can also write this as (H − 1)

2

= 0, so we have a natural ring homomorphism

Z[H]/(H − 1)

2

→K(S

2

) whose domain is the quotient ring of the polynomial ring

Z[H] by the ideal generated by (H−1)

2

. In particular, note that an additive basis for

Z[H]/(H −1)

2

is ¦1, H¦.

We deﬁne a homomorphism µ as the composition

µ : K(X)⊗Z[H]/(H −1)

2

→K(X)⊗K(S

2

) →K(X×S

2

)

where the second map is the external product.

Theorem 2.2. The homomorphism µ : K(X)⊗Z[H]/(H −1)

2

→K(X×S

2

) is an iso-

morphism of rings for all compact Hausdorff spaces X.

Taking X to be a point we obtain:

Corollary 2.3. The map Z[H]/(H −1)

2

→K(S

2

) is an isomorphism of rings.

Thus if we regard

`

K(S

2

) as the kernel of K(S

2

)→K(x

0

), then it is generated as

an abelian group by H −1. Since we have the relation (H −1)

2

= 0, this means that

42 Chapter 2 Complex K–Theory

the multiplication in

`

K(S

2

) is completely trivial: The product of any two elements

is zero. Readers familiar with cup product in ordinary cohomology will recognize

that the situation is exactly the same as in H

∗

(S

2

; Z) and

`

H

∗

(S

2

; Z), with H − 1

behaving exactly like the generator of H

2

(S

2

; Z). In the case of ordinary cohomology

the cup product of a generator of H

2

(S

2

; Z) with itself is automatically zero since

H

4

(S

2

; Z) = 0, whereas with K–theory a calculation is required.

The present section will be devoted entirely to the proof of the preceding theorem.

Nothing in the proof will be used elsewhere in the book except in the proof of Bott

periodicity for real K–theory in §2.4, so the reader who wishes to defer a careful

reading of the proof may skip ahead to §2.2 without any loss of continuity.

The main work in proving the theorem will be to prove the surjectivity of µ.

Injectivity will then be proved by a closer examination of the surjectivity argument.

Clutching Functions

From the classiﬁcation of vector bundles over spheres in §1.2 we know that vec-

tor bundles over S

2

correspond exactly to homotopy classes of maps S

1

→GL

n

(C),

which we called clutching functions. To prove the theoremwe will generalize this con-

struction, creating vector bundles over X×S

2

by gluing together two vector bundles

over X×D

2

by means of a generalized clutching function.

We begin by describing this more general clutching construction. Given a vector

bundle p: E→X, let f : E×S

1

→E×S

1

be an automorphism of the product vector

bundle p×11: E×S

1

→X×S

1

. Thus for each x ∈ X and z ∈ S

1

, f speciﬁes an

isomorphism f(x, z) : p

−1

(x)→p

−1

(x). From E and f we construct a vector bundle

over X×S

2

by taking two copies of E×D

2

and identifying the subspaces E×S

1

via

f . We write this bundle as [E, f], and call f a clutching function for [E, f]. If

f

t

: E×S

1

→E×S

1

is a homotopy of clutching functions, then [E, f

0

] ≈ [E, f

1

] since

from the homotopy f

t

we can construct a vector bundle over X×S

2

×I restricting

to [E, f

0

] and [E, f

1

] over X×S

2

×¦0¦ and X×S

2

×¦1¦. From the deﬁnitions it is

evident that [E

1

, f

1

]⊕[E

2

, f

2

] ≈ [E

1

⊕E

2

, f

1

⊕f

2

].

Here are some examples of bundles built using clutching functions:

1. [E, 11] is the external product E ∗ 1 = µ(E⊗1), or equivalently the pullback of E

via the projection X×S

2

→X.

2. Taking X to be a point, then we showed in Example 1.10 that [1, z] ≈ H where ‘ 1’

is the trivial line bundle over X, ‘ z’ means scalar multiplication by z ∈ S

1

⊂ C, and

H is the canonical line bundle over S

2

= CP

1

. More generally we have [1, z

n

] ≈ H

n

,

the n fold tensor product of H with itself. The formula [1, z

n

] ≈ H

n

holds also for

negative n if we deﬁne H

−1

= [1, z

−1

], which is justiﬁed by the fact that H⊗H

−1

≈ 1.

3. [E, z

n

] ≈ E ∗H

n

= µ(E⊗H

n

) for n ∈ Z.

4. Generalizing this, [E, z

n

f] ≈ [E, f]⊗

H

n

where

H

n

denotes the pullback of H

n

via the projection X×S

2

→S

2

.

The Functor K(X) Section 2.1 43

Every vector bundle E

¹

→X×S

2

is isomorphic to [E, f] for some E and f . To

see this, let the unit circle S

1

⊂ C ∪ ¦∞¦ = S

2

decompose S

2

into the two disks D

0

and D

∞

, and let E

α

for α = 0, ∞ be the restriction of E

¹

over X×D

α

, with E the

restriction of E

¹

over X×¦1¦. The projection X×D

α

→X×¦1¦ is homotopic to the

identity map of X×D

α

, so the bundle E

α

is isomorphic to the pullback of E by the

projection, and this pullback is E×D

α

, so we have an isomorphism h

α

: E

α

→E×D

α

.

Then f = h

0

h

−1

∞

is a clutching function for E

¹

.

We may assume a clutching function f is normalized to be the identity over

X×¦1¦ since we may normalize any isomorphism h

α

: E

α

→E×D

α

by composing it

over each X×¦z¦ with the inverse of its restriction over X×¦1¦. Any two choices of

normalized h

α

are homotopic through normalized h

α

’s since they differ by a map g

α

from D

α

to the automorphisms of E, with g

α

(1) = 11, and such a g

α

is homotopic

to the constant map 11 by composing it with a deformation retraction of D

α

to 1.

Thus any two choices f

0

and f

1

of normalized clutching functions are joined by a

homotopy of normalized clutching functions f

t

.

The strategy of the proof will be to reduce from arbitrary clutching functions

to successively simpler clutching functions. The ﬁrst step is to reduce to Laurent

polynomial clutching functions, which have the form (x, z) =

¸

¦i¦≤n

a

i

(x)z

i

where

a

i

: E→E restricts to a linear transformation a

i

(x) in each ﬁber p

−1

(x). We call

such an a

i

an endomorphism of E since the linear transformations a

i

(x) need not

be invertible, though their linear combination

¸

i

a

i

(x)z

i

is since clutching functions

are automorphisms.

Proposition 2.4. Every vector bundle [E, f] is isomorphic to [E, ] for some Laurent

polynomial clutching function . Laurent polynomial clutching functions

0

and

1

which are homotopic through clutching functions are homotopic by a Laurent

polynomial clutching function homotopy

t

(x, z) =

¸

i

a

i

(x, t)z

i

.

Before beginning the proof we need a lemma. For a compact space X we wish

to approximate a continuous function f : X×S

1

→C by Laurent polynomial functions

¸

¦n¦≤N

a

n

(x)z

n

=

¸

¦n¦≤N

a

n

(x)e

inθ

, where each a

n

is a continuous function X→C.

Motivated by Fourier series, we set

a

n

(x) =

1

2π

S

1

f(x, θ)e

−inθ

dθ

For positive real r let u(x, r, θ) =

¸

n∈Z

a

n

(x)r

¦n¦

e

inθ

. For ﬁxed r < 1, this series

converges absolutely and uniformly as (x, θ) ranges over X×S

1

, by comparison with

the geometric series

¸

n

r

n

, since compactness of X×S

1

implies that ¦f(x, θ)¦ is

bounded and hence also ¦a

n

(x)¦. If we can show that u(x, r, θ) approaches f(x, θ)

uniformly in x and θ as r goes to 1, then sums of ﬁnitely many terms in the series

for u(x, r, θ) with r near 1 will give the desired approximations to f by Laurent

polynomial functions.

44 Chapter 2 Complex K–Theory

Lemma 2.5. As r→1, u(x, r, θ)→f(x, θ) uniformly in x and θ.

Proof: For r < 1 we have

u(x, r, θ) =

∞

¸

n=−∞

1

2π

S

1

r

¦n¦

e

in(θ−t)

f(x, t) dt

=

S

1

1

2π

∞

¸

n=−∞

r

¦n¦

e

in(θ−t)

f(x, t) dt

where the order of summation and integration can be interchanged since the series

in the latter formula converges uniformly, by comparison with the geometric series

¸

n

r

n

. Deﬁne the Poisson kernel

P(r, ϕ) =

1

2π

∞

¸

n=−∞

r

¦n¦

e

inϕ

for 0 ≤ r < 1 and ϕ ∈ R

Then u(x, r, θ) =

S

1

P(r, θ −t)f(x, t) dt . By summing the two geometric series for

positive and negative n in the formula for P(r, ϕ), one computes that

P(r, ϕ) =

1

2π

·

1 −r

2

1 −2r cos ϕ+r

2

Three basic facts about P(r, ϕ) which we shall need are:

(a) As a function of ϕ, P(r, ϕ) is even, of period 2π, and monotone decreasing

on [0, π], since the same is true of cos ϕ which appears in the denominator of

P(r, ϕ) with a minus sign. In particular we have P(r, ϕ) ≥ P(r, π) > 0 for all

r < 1.

(b)

S

1

P(r, ϕ) dϕ = 1 for each r < 1, as one sees by integrating the series for

P(r, ϕ) term by term.

(c) For ﬁxed ϕ ∈ (0, π), P(r, ϕ)→0 as r→1 since the numerator of P(r, ϕ) ap-

proaches 0 and the denominator approaches 2 −2cos ϕ ≠ 0.

Nowto showuniformconvergence of u(x, r, θ) to f(x, θ) we ﬁrst observe that, using

(b), we have

¸

¸

u(x, r, θ) −f(x, θ)

¸

¸

=

¸

¸

¸

S

1

P(r, θ −t)f(x, t) dt −

S

1

P(r, θ −t)f(x, θ) dt

¸

¸

¸

≤

S

1

P(r, θ −t)

¸

¸

f(x, t) −f(x, θ)

¸

¸

dt

Given ε > 0, there exists a δ > 0 such that ¦f(x, t) −f(x, θ)¦ < ε for ¦t −θ¦ < δ and

all x, since f is uniformly continuous on the compact space X×S

1

. Let I

δ

denote

the integral

**P(r, θ −t) ¦f(x, t) −f(x, θ)¦ dt over the interval ¦t −θ¦ ≤ δ and let I
**

¹

δ

denote this integral over the rest of S

1

. Then we have

I

δ

≤

¦t−θ¦≤δ

P(r, θ −t) ε dt ≤ ε

S

1

P(r, θ −t) dt = ε

By (a) the maximum value of P(r, θ −t) on ¦t −θ¦ ≥ δ is P(r, δ). So

I

¹

δ

≤ P(r, δ)

S

1

¦f(x, t) −f(x, θ)¦ dt

The Functor K(X) Section 2.1 45

The integral here has a uniformbound for all x and θ since f is bounded. Thus by (c)

we can make I

¹

δ

≤ ε by taking r close enough to 1. Therefore ¦u(x, r, θ)−f(x, θ)¦ ≤

I

δ

+I

¹

δ

≤ 2ε. L¦

Proof of Proposition 2.4: Choosing a Hermitian inner product on E, the endomor-

phisms of E×S

1

form a vector space End(E×S

1

) with a norm ¦α¦ = sup

¦v¦=1

¦α(v)¦.

The triangle inequality holds for this norm, so balls in End(E×S

1

) are convex. The

subspace Aut(E×S

1

) of automorphisms is open in the topology deﬁned by this norm

since it is the preimage of (0, ∞) under the continuous map End(E×S

1

)→[0, ∞),

α

inf

(x,z)∈X×S

1 ¦ det(α(x, z))¦. Thus to prove the ﬁrst statement of the lemma it

will sufﬁce to show that Laurent polynomials are dense in End(E×S

1

), since a suf-

ﬁciently close Laurent polynomial approximation to f will then be homotopic to

f via the linear homotopy t + (1 − t)f through clutching functions. The second

statement follows similarly by approximating a homotopy from

0

to

1

, viewed as

an automorphism of E×S

1

×I , by a Laurent polynomial homotopy

¹

t

, then combin-

ing this with linear homotopies from

0

to

¹

0

and

1

to

¹

1

to obtain a homotopy

t

from

0

to

1

.

To show that every f ∈ End(E×S

1

) can be approximated by Laurent polynomial

endomorphisms, ﬁrst choose open sets U

i

covering X together with isomorphisms

h

i

: p

−1

(U

i

)→U

i

×C

n

i

. We may assume h

i

takes the chosen inner product in p

−1

(U

i

)

to the standard inner product in C

n

i

, by applying the Gram-Schmidt process to h

−1

i

of the standard basis vectors. Let ¦ϕ

i

¦ be a partition of unity subordinate to ¦U

i

¦

and let X

i

be the support of ϕ

i

, a compact set in U

i

. Via h

i

, the linear maps f(x, z)

for x ∈ X

i

can be viewed as matrices. The entries of these matrices deﬁne functions

X

i

×S

1

→C. By the lemma we can ﬁnd Laurent polynomial matrices

i

(x, z) whose

entries uniformly approximate those of f(x, z) for x ∈ X

i

. It follows easily that

i

approximates f in the ¦ · ¦ norm. From the Laurent polynomial approximations

i

over X

i

we form the convex linear combination =

¸

i

ϕ

i

i

, a Laurent polynomial

approximating f over all of X×S

1

. L¦

A Laurent polynomial clutching function can be written = z

−m

q for a polyno-

mial clutching function q, and then we have [E, ] ≈ [E, q]⊗

H

−m

. The next step is

to reduce polynomial clutching functions to linear clutching functions.

Proposition 2.6. If q is a polynomial clutching function of degree at most n, then

[E, q]⊕[nE, 11] ≈ [(n+1)E, L

n

q] for a linear clutching function L

n

q.

46 Chapter 2 Complex K–Theory

Proof: Let q(x, z) = a

n

(x)z

n

+··· +a

0

(x). Each of the matrices

A =

¸

¸

¸

¸

¸

¸

¸

¸

¸

1 −z 0 ··· 0 0

0 1 −z ··· 0 0

0 0 1 ··· 0 0

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

0 0 0 ··· 1 −z

a

n

a

n−1

a

n−2

··· a

1

a

0

B =

¸

¸

¸

¸

¸

¸

¸

¸

¸

1 0 0 ··· 0 0

0 1 0 ··· 0 0

0 0 1 ··· 0 0

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

0 0 0 ··· 1 0

0 0 0 ··· 0 q

**deﬁnes an endomorphism of (n+1)E by interpreting the (i, j) entry of the matrix as
**

a linear map from the j

th

summand of (n+1)E to the i

th

summand, with the entries

1 denoting the identity E→E and z denoting z times the identity, for z ∈ S

1

. We can

pass from the matrix A to the matrix B by a sequence of elementary row and column

operations in the following way. In A, add z times the ﬁrst column to the second

column, then z times the second column to the third, and so on. This produces 0’s

above the diagonal and the polynomial q in the lower right corner. Then for each

i ≤ n, subtract the appropriate multiple of the i

th

row from the last row to make all

the entries in the last row 0 except for the ﬁnal q. These row and column operations

are not quite elementary row and column operations in the traditional sense since

the entries of the matrices are not numbers but linear maps. However, by restricting

to a ﬁber of E and choosing a basis in this ﬁber, each entry in A becomes a matrix

itself, and then each of the preceding row and column operations can be realized by

a sequence of traditional row and column operations on the expanded matrices.

The matrix B is a clutching function for [nE, 11]⊕[E, q], hence in each ﬁber the

expanded version of B has nonzero determinant. Since elementary row and column

operations preserve determinant, the expanded version of A is also invertible in each

ﬁber. This means that A is an automorphism of (n + 1)E for each z ∈ S

1

, and

therefore determines a clutching function which we denote by L

n

q. Since L

n

q has

the form A(x)z+B(x), it is a linear clutching function. The matrices A and B deﬁne

homotopic clutching functions since the elementary row and column operations can

be achieved by continuous one-parameter families of such operations. For example

the ﬁrst operation can be achieved by adding tz times the ﬁrst column to the second,

with t ranging from 0 to 1. Since homotopic clutching functions produce isomorphic

bundles, we obtain an isomorphism [E, q]⊕[nE, 11] ≈ [(n+1)E, L

n

q]. L¦

Linear Clutching Functions

For linear clutching functions a(x)z +b(x) we have the following key fact:

Proposition 2.7. Given a bundle [E, a(x)z +b(x)], there is a splitting E ≈ E

+

⊕E

−

with [E, a(x)z +b(x)] ≈ [E

+

, 11]⊕[E

−

, z].

Proof: The ﬁrst step is to reduce to the case that a(x) is the identity for all x.

Consider the expression:

(∗) (1 +tz)

¸

a(x)

z +t

1 +tz

+b(x)

¸

=

¸

a(x) +tb(x)

¸

z +ta(x) +b(x)

The Functor K(X) Section 2.1 47

When t = 0 this equals a(x)z + b(x). For 0 ≤ t < 1, (∗) deﬁnes an invertible

linear transformation since the left-hand side is obtained from a(x)z +b(x) by ﬁrst

applying the substitution z

(z + t)/(1 + tz) which takes S

1

to itself (because if

¦z¦ = 1 then ¦(z+t)/(1+tz)¦ = ¦z(z+t)/(1+tz)¦ = ¦(1+tz)/(1+tz)¦ = ¦w/w¦ = 1),

and then multiplying by the nonzero scalar 1+tz. Therefore (∗) deﬁnes a homotopy

of clutching functions as t goes from 0 to t

0

< 1. In the right-hand side of (∗) the

term a(x) + tb(x) is invertible for t = 1 since it is the restriction of a(x)z + b(x)

to z = 1. Therefore a(x) + tb(x) is invertible for t = t

0

near 1, as the continuous

function t

inf

x∈X

¸

¸

det[a(x) +tb(x)]

¸

¸

is nonzero for t = 1, hence also for t near

1. Now we use the simple fact that [E, fg] ≈ [E, f] for any isomorphism g : E→E.

This allows us to replace the clutching function on the right-hand side of (∗) by the

clutching function z + [t

0

a(x) + b(x)][a(x) + t

0

b(x)]

−1

, reducing to the case of

clutching functions of the form z +b(x).

Since z +b(x) is invertible for all x, b(x) has no eigenvalues on the unit circle

S

1

.

Lemma 2.8. Let b: E→E be an endomorphism having no eigenvalues on the unit

circle S

1

. Then there are unique subbundles E

+

and E

−

of E such that:

(a) E = E

+

⊕E

−

.

(b) b(E

:

) ⊂ E

:

.

(c) The eigenvalues of b¦

¦E

+

all lie outside S

1

and the eigenvalues of b¦

¦E

−

all lie

inside S

1

.

Proof: Consider ﬁrst the algebraic situation of a linear transformation T : V→V with

characteristic polynomial q(t). Assuming q(t) has no roots on S

1

, we may factor q(t)

in C[t] as q

+

(t)q

−

(t) where q

+

(t) has all its roots outside S

1

and q

−

(t) has all its

roots inside S

1

. Let V

:

be the kernel of q

:

(T) : V→V . Since q

+

and q

−

are relatively

prime in C[t], there are polynomials r and s with rq

+

+sq

−

= 1. From q

+

(T)q

−

(T) =

q(T) = 0, we have Imq

−

(T) ⊂ Ker q

+

(T), and the opposite inclusion follows from

r(T)q

+

(T) +q

−

(T)s(T) = 11. Thus Ker q

+

(T) = Imq

−

(T), and similarly Ker q

−

(T) =

Imq

+

(T). From q

+

(T)r(T) + q

−

(T)s(T) = 11 we see that Imq

+

(T) + Imq

−

(T) = V ,

and from r(T)q

+

(T) + s(T)q

−

(T) = 11 we deduce that Ker q

+

(T) ∩ Ker q

−

(T) = 0.

Hence V = V

+

⊕V

−

. We have T(V

:

) ⊂ V

:

since q

:

(T)(v) = 0 implies q

:

(T)(T(v)) =

T(q

:

(T)(v)) = 0. All eigenvalues of T ¦

¦V

:

are roots of q

:

since q

:

(T) = 0 on V

:

.

Thus conditions (a)–(c) hold for V

+

and V

−

.

To see the uniqueness of V

+

and V

−

satisfying (a)–(c), let q

¹

:

be the characteristic

polynomial of T ¦

¦V

:

, so q = q

¹

+

q

¹

−

. All the linear factors of q

¹

:

must be factors of

q

:

by condition (c), so the factorizations q = q

¹

+

q

¹

−

and q = q

+

q

−

must coincide up

to scalar factors. Since q

¹

:

(T) is identically zero on V

:

, so must be q

:

(T), hence

V

:

⊂ Ker q

:

(T). Since V = V

+

⊕V

−

and V = Ker q

+

(T)⊕ Ker q

−

(T), we must have

V

:

= Ker q

:

(T). This establishes the uniqueness of V

:

.

48 Chapter 2 Complex K–Theory

As T varies continuously through linear transformations without eigenvalues on

S

1

, its characteristic polynomial q(t) varies continuously through polynomials with-

out roots in S

1

. In this situation we assert that the factors q

:

of q vary continuously

with q, assuming that q, q

+

, and q

−

are normalized to be monic polynomials. To

see this we shall use the fact that for any circle C in C disjoint from the roots of q,

the number of roots of q inside C, counted with multiplicity, equals the degree of

the map γ : C→S

1

, γ(z) = q(z)/¦q(z)¦. To prove this fact it sufﬁces to consider the

case of a small circle C about a root z = a of multiplicity m, so q(t) = p(t)(t −a)

m

with p(a) ≠ 0. The homotopy

γ

s

(z) =

p(sa +(1 −s)z)(z −a)

m

¦p(sa +(1 −s)z)(z −a)

m

¦

gives a reduction to the case (t −a)

m

, where it is clear that the degree is m.

Thus for a small circle C about a root z = a of q of multiplicity m, small per-

turbations of q produce polynomials q

¹

which also have m roots a

1

, ··· , a

m

inside

C, so the factor (z −a)

m

of q becomes a factor (z −a

1

) ··· (z −a

m

) of the nearby

q

¹

. Since the a

i

’s are near a, these factors of q and q

¹

are close, and so q

¹

:

is close

to q

:

.

Next we observe that as T varies continuously through transformations without

eigenvalues in S

1

, the splitting V = V

+

⊕V

−

also varies continuously. To see this,

recall that V

+

= Imq

−

(T) and V

−

= Imq

+

(T). Choose a basis v

1

, ··· , v

n

for V such

that q

−

(T)(v

1

), ··· , q

−

(T)(v

k

) is a basis for V

+

and q

+

(T)(v

k+1

), ··· , q

+

(T)(v

n

) is

a basis for V

−

. For nearby T these vectors vary continuously, hence remain indepen-

dent. Thus the splitting V = Imq

−

(T)⊕ Imq

+

(T) continues to hold for nearby T ,

and so the splitting V = V

+

⊕V

−

varies continuously with T .

It follows that the union E

:

of the subspaces V

:

in all the ﬁbers V of E is a

subbundle, and so the proof of the lemma is complete. L¦

To ﬁnish the proof of Proposition 2.7, note that the lemma gives a splitting [E, z+

b(x)] ≈ [E

+

, z + b

+

(x)]⊕[E

−

, z + b

−

(x)] where b

+

and b

−

are the restrictions of

b. Since b

+

(x) has all its eigenvalues outside S

1

, the formula tz + b

+

(x) for 0 ≤

t ≤ 1 deﬁnes a homotopy of clutching functions from z + b

+

(x) to b

+

(x). Hence

[E

+

, z +b

+

(x)] ≈ [E

+

, b

+

(x)] ≈ [E

+

, 11]. Similarly, z +tb

−

(x) deﬁnes a homotopy of

clutching functions from z +b

−

(x) to z, so [E

−

, z +b

−

(x)] ≈ [E

−

, z]. L¦

For future reference we note that the splitting [E, az + b] ≈ [E

+

, 11]⊕[E

−

, z]

constructed in the proof of Proposition 2.7 preserves direct sums, in the sense that

the splitting for a sum [E

1

⊕E

2

, (a

1

z+b

1

)⊕(a

2

z+b

2

)] has (E

1

⊕E

2

)

:

= (E

1

)

:

⊕(E

2

)

:

.

This is because the ﬁrst step of reducing to the case a = 11 clearly respects sums, and

the uniqueness of the : splitting in Lemma 2.8 guarantees that it preserves sums.

The Functor K(X) Section 2.1 49

Conclusion of the Proof

Now we address the question of showing that the homomorphism

µ : K(X)⊗Z[H]/(H −1)

2

→K(X×S

2

)

is an isomorphism. The preceding propositions imply that in K(X×S

2

) we have

[E, f] = [E, z

−m

q]

= [E, q]⊗

H

−m

= [(n+1)E, L

n

q]⊗

H

−m

−[nE, 11]⊗

H

−m

= [((n+1)E)

+

, 11]⊗

H

−m

+[((n+1)E)

−

, z]⊗

H

−m

−[nE, 11]⊗

H

−m

= ((n+1)E)

+

∗H

−m

+((n+1)E)

−

∗H

1−m

−nE ∗H

−m

This last expression is in the image of µ. Since every vector bundle over X×S

2

has

the form [E, f], it follows that µ is surjective.

To show µ is injective we shall construct ν : K(X×S

2

)→K(X)⊗Z[H]/(H − 1)

2

such that νµ = 11. The idea will be to deﬁne ν([E, f]) as some linear combination of

terms E⊗H

k

and ((n+1)E)

:

⊗H

k

that is independent of all choices.

To investigate the dependence of the terms in the formula for [E, f] displayed

above on m and n we ﬁrst derive the following two formulas, where deg q ≤ n:

(1) [(n+2)E, L

n+1

q] ≈ [(n+1)E, L

n

q]⊕[E, 11]

(2) [(n+2)E, L

n+1

(zq)] ≈ [(n+1)E, L

n

q]⊕[E, z]

The matrix representations of L

n+1

q and L

n+1

(zq) are:

¸

¸

¸

¸

¸

¸

¸

1 −z 0 ··· 0

0 1 −z ··· 0

.

.

.

.

.

.

.

.

.

.

.

.

0 0 0 1 −z

0 a

n

a

n−1

··· a

0

and

¸

¸

¸

¸

¸

¸

¸

1 −z 0 ··· 0 0

0 1 −z ··· 0 0

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

0 0 0 ··· 1 −z

a

n

a

n−1

a

n−2

··· a

0

0

**In the ﬁrst matrix we can add z times the ﬁrst column to the second column to
**

eliminate the −z in the ﬁrst row, and then the ﬁrst rowand column give the summand

[E, 11] while the rest of the matrix gives [(n + 1)E, L

n

q]. This proves (1). Similarly,

in the second matrix we add z

−1

times the last column to the next-to-last column to

make the −z in the last column have all zeros in its row and column, which gives the

splitting in (2) since [E, −z] ≈ [E, z], the clutching function −z being the composition

of the clutching function z with the automorphism −1 of E.

In view of the appearance of the correction terms [E, 11] and [E, z] in (1) and (2),

it will be useful to know the ‘ :’ splittings for these two bundles:

(3) For [E, 11] the summand E

−

is 0 and E

+

= E.

(4) For [E, z] the summand E

+

is 0 and E

−

= E.

Statement (4) is obvious from the deﬁnitions since the clutching function z is already

in the form z + b(x) with b(x) = 0, so 0 is the only eigenvalue of b(x) and hence

50 Chapter 2 Complex K–Theory

E

+

= 0. To obtain (3) we ﬁrst apply the procedure at the beginning of the proof of

Proposition 2.7 which replaces a clutching function a(x)z + b(x) by the clutching

function z + [t

0

a(x) + b(x)][a(x) + t

0

b(x)]

−1

with 0 < t

0

< 1. Specializing to the

case a(x)z +b(x) = 11 this yields z +t

−1

0

11. Since t

−1

0

11 has only the one eigenvalue

t

−1

0

> 1, we have E

−

= 0.

Formulas (1) and (3) give ((n + 2)E)

−

≈ ((n + 1)E)

−

, using the fact that the

: splitting preserves direct sums. So the ‘minus’ summand is independent of n.

Suppose we deﬁne

ν([E, z

−m

q]) = ((n+1)E)

−

⊗(H −1) +E⊗H

−m

∈ K(X)⊗Z[H]/(H −1)

2

for n ≥ deg q. We claim that this is well-deﬁned. We have just noted that ‘minus’

summands are independent of n, so ν([E, z

−m

q]) does not depend on n. To see

that it is independent of m we must see that it is unchanged when z

−m

q is replaced

by z

−m−1

(zq). By (2) and (4) we have the ﬁrst of the following equalities:

ν([E, z

−m−1

(zq)]) = ((n+1)E)

−

⊗(H −1) +E⊗(H −1) +E⊗H

−m−1

= ((n+1)E)

−

⊗(H −1) +E⊗(H

−m

−H

−m−1

) +E⊗H

−m−1

= ((n+1)E)

−

⊗(H −1) +E⊗H

−m

= ν([E, z

−m

q])

The second equality use the relation (H−1)

2

= 0 which implies H(H−1) = H−1 and

hence H −1 = H

−m

−H

−m−1

for all m. The third and fourth equalities are evident.

Another choice which might perhaps affect the value of ν([E, z

−m

q]) is the con-

stant t

0

< 1 in the proof of Proposition 2.7. This could be any number sufﬁciently

close to 1, so varying t

0

gives a homotopy of the endomorphism b in Lemma 2.8. This

has no effect on the : splitting since we can apply Lemma 2.8 to the endomorphism

of E×I given by the homotopy. Hence the choice of t

0

does not affect ν([E, z

−m

q]).

It remains to see that ν([E, z

−m

q]) depends only on the bundle [E, z

−m

q], not on

the clutching function z

−m

q for this bundle. We showed that every bundle over X×S

2

has the form [E, f] for a normalized clutching function f which was unique up to

homotopy, and in Proposition 2.5 we showed that Laurent polynomial approximations

to homotopic f ’s are Laurent-polynomial-homotopic. If we apply Propositions 2.6

and 2.7 over X×I with a Laurent polynomial homotopy as clutching function, we

conclude that the two bundles ((n+1)E)

−

over X×¦0¦ and X×¦1¦ are isomorphic.

This ﬁnishes the veriﬁcation that ν([E, z

−m

q]) is well-deﬁned.

It is easy to check through the deﬁnitions to see that ν takes sums to sums

since L

n

(q

1

⊕ q

2

) = L

n

q

1

⊕ L

n

q

2

and the : splitting in Proposition 2.7 preserves

sums, as we have already observed. It follows that ν extends to a homomorphism

K(X×S

2

)→K(X)⊗Z[H]/(H −1)

2

.

The last thing to verify is that νµ = 11. The group Z[H]/(H−1)

2

is generated by

1 and H, so in view of the relation H + H

−1

= 2, which follows from (H − 1)

2

= 0,

Bott Periodicity Section 2.2 51

we see that K(S

2

) is also generated by 1 and H

−1

. Thus it sufﬁces to show νµ = 11

on elements of the form E⊗H

−m

for m≥ 0. We have νµ(E⊗H

−m

) = ν([E, z

−m

]) =

E

−

⊗(H−1) +E⊗H

−m

= E⊗H

−m

since E

−

= 0, the polynomial q being 11 so that (3)

applies.

This completes the proof of Theorem 2.2. L¦

2.2 Bott Periodicity

In this section we develop a fewbasic tools that make it possible to compute K(X)

for more complicated spaces X, and in particular for all spheres, yielding the Bott

Periodicity Theorem. These tools are formally very similar to the basic machinery of

algebraic topology involving cohomology, and in fact they make K–theory into what

was classically called a generalized cohomology theory, but now is called simply a

cohomology theory.

Exact Sequences

We will be going back and forth between the two versions of K–theory, K(X) and

`

K(X). Sometimes one is more convenient for stating a result, sometimes the other.

We begin by examining a key property of the reduced groups

`

K(X).

Proposition 2.9. If X is compact Hausdorff and A ⊂ X is a closed subspace, then the

inclusion and quotient maps A

i

→X

q

→X/A induce homomorphisms

`

K(X/A)

q

∗

→

`

K(X)

i

∗

→

`

K(A) for which the kernel of i

∗

equals the image of q

∗

.

Since we assume A is a closed subspace of a compact Hausdorff space, it is also

compact Hausdorff. The quotient space X/A is compact Hausdorff as well, with

the Hausdorff property following from the fact that compact Hausdorff spaces are

normal, hence a point x ∈ X −A and the subspace A have disjoint neighborhoods in

X, projecting to disjoint neighborhoods of x and the point A/A in X/A.

There is some standard terminology to described the conclusion of the propo-

sition. A sequence of homomorphisms G

1

→G

2

→··· →G

n

is said to be exact if at

each intermediate group G

i

the kernel of the outgoing map equals the image of in-

coming map. Thus the proposition states that

`

K(X/A)

q

∗

→

`

K(X)

i

∗

→

`

K(A) is exact.

We will be deriving some longer exact sequences below.

Proof: The inclusion Imq

∗

⊂ Ker i

∗

is equivalent to i

∗

q

∗

= 0. Since qi is equal to

the composition A→A/A

X/A and

`

K(A/A) = 0, it follows that i

∗

q

∗

= 0.

For the opposite inclusion Ker i

∗

⊂ Imq

∗

, suppose the restriction over A of a

vector bundle p: E→X is stably trivial. Adding a trivial bundle to E, we may assume

that E itself is trivial over A. Choosing a trivialization h: p

−1

(A)→A×C

n

, let E/h be

52 Chapter 2 Complex K–Theory

the quotient space of E under the identiﬁcations h

−1

(x, v) ∼ h

−1

(y, v) for x, y ∈ A.

There is then an induced projection E/h→X/A. To see that this is a vector bundle

we need to ﬁnd a local trivialization over a neighborhood of the point A/A.

We claim that since E is trivial over A, it is trivial over some neighborhood of A.

In many cases this holds because there is a neighborhood which deformation retracts

onto A, so the restriction of E over this neighborhood is trivial since it is isomorphic

to the pullback of p

−1

(A) via the retraction. In the absence of such a deformation

retraction one can make the following more complicated argument. A trivialization

of E over A determines sections s

i

: A→E which form a basis in each ﬁber over A.

Choose a cover of A by open sets U

j

in X over each of which E is trivial. Via a local

trivialization, each section s

i

can be regarded as a map from A∩U

j

to a single ﬁber,

so by the Tietze extension theorem we obtain a section s

ij

: U

j

→E extending s

i

. If

¦ϕ

j

, ϕ¦ is a partition of unity subordinate to the cover ¦U

j

, X − A¦ of X, the sum

¸

j

ϕ

j

s

ij

gives an extension of s

i

to a section deﬁned on all of X. Since these sections

form a basis in each ﬁber over A, they must form a basis in all nearby ﬁbers. Namely,

over U

j

the extended s

i

’s can be viewed as a square-matrix-valued function having

nonzero determinant at each point of A, hence at nearby points as well.

Thus we have a trivialization h of E over a neighborhood U of A. This induces

a trivialization of E/h over U/A, so E/h is a vector bundle. It remains only to verify

that E ≈ q

∗

(E/h). In the commutative diagram at the right the

−

−

→

−

−

→

−−−−−→

E

−−−−−→

X A

q

p

X/

h E/

quotient map E→E/h is an isomorphism on ﬁbers, so this map

and p give an isomorphism E ≈ q

∗

(E/h). L¦

There is an easy way to extend the exact sequence

`

K(X/A)→

`

K(X)→

`

K(A) to the

left, using the following diagram, where C and S denote cone and suspension:

A CA X X

SA SX A

∪ CA CX X ∪ ∪ ( ) CA CX C X ∪ ∪ ∪ ( ( ) CA X ∪ ( ) )

X/

−−→ −−→ −−→

In the ﬁrst row, each space is obtained fromits predecessor by attaching a cone on the

subspace two steps back in the sequence. The vertical maps are the quotient maps

obtained by collapsing the most recently attached cone to a point. In many cases the

quotient map collapsing a contractible subspace to a point is a homotopy equivalence,

hence induces an isomorphism on

`

K. This conclusion holds generally, in fact:

Lemma 2.10. If A is contractible, the quotient map q : X→X/A induces a bijection

q

∗

: Vect

n

(X/A)→Vect

n

(X) for all n.

Proof: A vector bundle E→X must be trivial over A since A is contractible. A

trivialization h gives a vector bundle E/h→X/A as in the proof of the previous

proposition. We assert that the isomorphism class of E/h does not depend on h.

This can be seen as follows. Given two trivializations h

0

and h

1

, by writing h

1

=

Bott Periodicity Section 2.2 53

(h

1

h

−1

0

)h

0

we see that h

0

and h

1

differ by an element of g

x

∈ GL

n

(C) over each

point x ∈ A. The resulting map g : A→GL

n

(C) is homotopic to a constant map

x

α ∈ GL

n

(C) since A is contractible. Writing now h

1

= (h

1

h

−1

0

α

−1

)(αh

0

), we

see that by composing h

0

with α in each ﬁber, which does not change E/h

0

, we may

assume that α is the identity. Then the homotopy from g to the identity gives a

homotopy H from h

0

to h

1

. In the same way that we constructed E/h we construct

a vector bundle (E×I)/H→(X/A)×I restricting to E/h

0

over one end and to E/h

1

over the other end, hence E/h

0

≈ E/h

1

.

Thus we have a well-deﬁned map Vect

n

(X)→Vect

n

(X/A), E

E/h. This is an

inverse to q

∗

since q

∗

(E/h) ≈ E as we noted in the preceding proposition, and for a

bundle E→X/A we have q

∗

(E)/h ≈ E for the evident trivialization h of q

∗

(E) over

A L¦

From this lemma and the preceding proposition it follows that we have a long

exact sequence of

`

K groups

··· →

`

K(SX)→

`

K(SA)→

`

K(X/A)→

`

K(X)→

`

K(A)

For example, if X is the wedge sum A ∨ B then X/A = B and the sequence breaks

up into split short exact sequences, which implies that the map

`

K(X)→

`

K(A)⊕

`

K(B)

obtained by restriction to A and B is an isomorphism.

Deducing Periodicity from the Product Theorem

We can use the exact sequence displayed above to obtain a reduced version of

the external product, a ring homomorphism

`

K(X)⊗

`

K(Y)→

`

K(X ∧Y) where X ∧Y =

X×Y/X ∨Y and X ∨Y = X×¦y

0

¦ ∪¦x

0

¦×Y ⊂ X×Y for chosen basepoints x

0

∈ X

and y

0

∈ Y . The space X ∧Y is called the smash product of X and Y . To deﬁne the

reduced product, consider the long exact sequence for the pair (X×Y, X ∨Y):

−−→

Y X S K( ( ))

∼

≈ ≈

⊕

∨ ∨ ×

−−→

Y X S K( ( ))

∼

X S K( )

∼

Y S K( )

∼

⊕ X K( )

∼

Y K( )

∼

−−→

Y X K( )

∼

−−→

Y X K ( )

∼

× Y X K( )

∼

∧

The second of the two vertical isomorphisms here was noted earlier. The ﬁrst vertical

isomorphism arises in similar fashion by using a reduced version of the suspension

operator which associates to a space Z with basepoint z

0

the quotient space ΣZ

of SZ obtained by collapsing the segment ¦z

0

¦×I to a point. The quotient map

SZ→ΣZ induces an isomorphism

`

K(SZ) ≈

`

K(ΣZ) by the preceding lemma. For

reduced suspension we have Σ(X ∨Y) = ΣX ∨ΣY , which gives the ﬁrst isomorphism

in the diagram. The last horizontal map in the sequence is a split surjection, with

splitting

`

K(X)⊕

`

K(Y)→

`

K(X×Y), (a, b)

p

∗

1

(a) +p

∗

2

(b) where p

1

and p

2

are the

projections of X×Y onto X and Y . Similarly, the ﬁrst map splits via (Sp

1

)

∗

+(Sp

2

)

∗

.

So we get a splitting

`

K(X×Y) ≈

`

K(X ∧Y)⊕

`

K(X)⊕

`

K(Y).

54 Chapter 2 Complex K–Theory

For a ∈

`

K(X) = Ker(K(X)→K(x

0

)) and b ∈

`

K(Y) = Ker(K(Y)→K(y

0

)) the

external product a ∗ b = p

∗

1

(a)p

∗

2

(b) ∈ K(X×Y) has p

∗

1

(a) restricting to zero in

K(Y) and p

∗

2

(b) restricting to zero in K(X), so p

∗

1

(a)p

∗

2

(b) restricts to zero in both

K(X) and K(Y), hence in K(X ∨ Y). In particular, a ∗ b lies in

`

K(X×Y), and from

the short exact sequence above, a ∗ b pulls back to a unique element of

`

K(X ∧ Y).

This deﬁnes the reduced external product

`

K(X)⊗

`

K(Y)→

`

K(X ∧ Y). It is essentially

a restriction of the unreduced external product, as shown in the following diagram,

≈ ⊕ ⊕ ⊕ Y X K ( ( ) K( ) ) Y X K( )

∼

K( )

∼

Y X K( )

∼

K( )

∼

−−→ −−→

=

=

=

=

=

=

⊗ ⊗ Z

≈ ⊕ ⊕ ⊕ Y X K( ) Y X K(

∼

) Y X K( )

∼

K( )

∼

Z ∧ ×

so the reduced external product is also a ring homomorphism, and we shall use the

same notation a ∗ b for both reduced and unreduced external product, leaving the

reader to determine from context which is meant.

Since S

n

∧X is the n fold iterated reduced suspension Σ

n

X, which is a quotient

of the ordinary n fold suspension S

n

X obtained by collapsing an n disk in S

n

X to a

point, the quotient map S

n

X→S

n

∧X induces an isomorphism on

`

K by Lemma 2.10.

Then the reduced external product gives rise to a homomorphism

β:

`

K(X)→

`

K(S

2

X), β(a) = (H −1) ∗a

where H is the canonical line bundle over S

2

= CP

1

. Here at last is the Bott Periodicity

Theorem:

Theorem 2.11. The homomorphism β:

`

K(X)→

`

K(S

2

X), β(a) = (H −1) ∗a, is an

isomorphism for all compact Hausdorff spaces X.

Proof: The map β is the composition

`

K(X) →

`

K(S

2

)⊗

`

K(X) →

`

K(S

2

X)

where the ﬁrst map a

(H −1) ⊗a is an isomorphism since

`

K(S

2

) is inﬁnite cyclic

generated by H − 1, and the second map is the reduced external product, so the

fact that this is an isomorphism is equivalent to the Product Theorem proved in the

previous section, by the diagram displayed above. L¦

Corollary 2.12.

`

K(S

2n+1

) = 0 and

`

K(S

2n

) ≈ Z, generated by the n fold reduced

external product (H −1) ∗··· ∗(H −1).

Bott Periodicity Section 2.2 55

Extending to a Cohomology Theory

As we sawearlier, a pair (X, A) of compact Hausdorff spaces gives rise to an exact

sequence of

`

K groups, the ﬁrst row in the following diagram:

≈

X S K ( ( ) ) K ( )

∼ ∼

A X K( )

∼

K( )

∼

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

=

−→ −→ −→ −→ −→ −→ −→ X/

2

A A S S K( )

∼

2

X S K( ) K( )

∼ ∼

−−→

≈

−−→

X/ A A S K( )

∼

X K ( ) K ( )

∼ ∼

A X K ( )

∼

K ( )

∼

−→ − → −→ −→ −→ − → − → X

2

K (

∼

2 1

(

1

A A ) X K )

∼

(

1

K

∼

A) , K ( )

∼

X

0 0 0

A X K ( )

∼

K ( )

∼

−→

0 0

A ,

β β

If we set

`

K

−n

(X) =

`

K(S

n

X) and

`

K

−n

(X, A) =

`

K(S

n

(X/A)), this sequence can be

written as in the second row. Negative indices are chosen here so that the ‘coboundary’

maps in this sequence increase dimension, as in ordinary cohomology. The lower

left corner of the diagram containing the Bott periodicity isomorphisms β commutes

since external tensor product with H−1 commutes with maps between spaces. So the

long exact sequence in the second row can be rolled up into a six-term periodic exact

sequence. It is reasonable to extend the deﬁnition of

`

K

n

to positive n by setting

`

K

2i

(X) =

`

K(X) and

`

K

2i+1

(X) =

`

K(SX). Then the six-term exact sequence can be

written

A X K ( )

∼

K ( )

∼

K ( )

∼

X

0 0 0

A ,

A X K ( )

∼

K ( )

∼

K ( )

∼

X

1 1 1

A ,

−−−−→

−

−

→

−

−

→

− − − − → − − − − →

−−−−→

A product

`

K

i

(X)⊗

`

K

j

(Y)→

`

K

i+j

(X ∧ Y) is obtained from the external product

`

K(X)⊗

`

K(Y)→

`

K(X∧Y) by replacing X and Y by S

i

X and S

j

Y . If we deﬁne

`

K

∗

(X) =

`

K

0

(X)⊕

`

K

1

(X), then this gives a product

`

K

∗

(X)⊗

`

K

∗

(Y)→

`

K

∗

(X ∧ Y). The relative

formof this is a product

`

K

∗

(X, A)⊗

`

K

∗

(Y, B)→

`

K

∗

(X×Y, X×B∪A×Y), coming from

the products

`

K(Σ

i

(X/A))⊗

`

K(Σ

j

(Y/B)) →

`

K(Σ

i+j

(X/A ∧ Y/B)) using the natural

identiﬁcation (X×Y)/(X×B ∪A×Y) = X/A∧Y/B.

If we compose the external product

`

K

∗

(X)⊗

`

K

∗

(X)→

`

K

∗

(X ∧ X) with the map

`

K

∗

(X ∧ X)→

`

K

∗

(X) induced by the diagonal map X→X ∧ X, x

(x, x), then we

obtain a multiplication on

`

K

∗

(X) making it into a ring, and it is not hard to check that

this extends the previously deﬁned ring structure on

`

K

0

(X). The general relative form

of this product on

`

K

∗

(X) is a product

`

K

∗

(X, A)⊗

`

K

∗

(X, B)→

`

K

∗

(X, A∪B) which is

induced by the relativized diagonal map X/(A∪B)→X/A∧X/B.

Example 2.13. Suppose that X = A ∪ B where A and B are compact contractible

subspaces of X containing the basepoint. Then the product

`

K

∗

(X)⊗

`

K

∗

(X)→

`

K

∗

(X)

is identically zero since it is equivalent to the composition

`

K

∗

(X, A)⊗

`

K

∗

(X, B)→

`

K

∗

(X, A∪B)→

`

K

∗

(X)

and

`

K

∗

(X, A∪B) = 0 since X = A∪B. For example if X is a suspension we can take A

and B to be its two cones, with a basepoint in their intersection. As a particular case we

56 Chapter 2 Complex K–Theory

see that the product in

`

K

∗

(S

n

) ≈ Z is trivial for n > 0. For n = 0 the multiplication

in

`

K

∗

(S

0

) ≈ Z is just the usual multiplication of integers since R

m

⊗R

n

≈ R

mn

. This

illustrates the necessity of the condition that A and B both contain the basepoint of

X, since without this condition we could take A and B to be the two points of S

0

.

More generally, if X is the union of compact contractible subspaces A

1

, ··· , A

n

containing the basepoint then the n fold product

`

K

∗

(X, A

1

)⊗ ··· ⊗

`

K

∗

(X, A

n

)→

`

K

∗

(X, A

1

∪··· ∪A

n

)

is trivial, so all n fold products in

`

K

∗

(X) are trivial. In particular all elements of

`

K

∗

(X) are nilpotent since their n

th

power is zero. This applies to all compact mani-

folds for example since they are covered by ﬁnitely many closed balls, and the condi-

tion that each A

i

contain the basepoint can be achieved by adjoining to each ball an

arc to a ﬁxed basepoint. In a similar fashion one can see that this observation applies

to all ﬁnite cell complexes, by induction on the number of cells.

Whereas multiplication in

`

K(X) is commutative, in

`

K

∗

(X) this is only true up to

sign:

Proposition 2.14. αβ = (−1)

ij

βα for α ∈

`

K

i

(X) and β ∈

`

K

j

(X).

Proof: The product is the composition

`

K(S

i

∧X)⊗

`

K(S

j

∧X) →

`

K(S

i

∧S

j

∧X ∧X) →

`

K(S

i

∧S

j

∧X)

where the ﬁrst map is external product and the second is induced by the diagonal

map on the X factors. Replacing the product αβ by the product βα amounts to

switching the two factors in the ﬁrst term

`

K(S

i

∧X)⊗

`

K(S

j

∧X), and this corresponds

to switching the S

i

and S

j

factors in the third term

`

K(S

i

∧S

j

∧X). Viewing S

i

∧S

j

as the smash product of i + j copies of S

1

, then switching S

i

and S

j

in S

i

∧ S

j

is

a product of ij transpositions of adjacent factors. Transposing the two factors of

S

1

∧S

1

is equivalent to reﬂection of S

2

across an equator. Thus it sufﬁces to see that

switching the two ends of a suspension SY induces multiplication by −1 in

`

K(SY). If

we view

`

K(SY) as ¹Y, U), then switching ends of SY corresponds to the map U→U

sending a matrix to its inverse. We noted in the proof of Proposition 2.2 that the group

operation in K(SY) is the same as the operation induced by the product in U, so the

result follows. L¦

Proposition 2.15. The exact sequence at the right is an

exact sequence of

`

K

∗

(X) modules, with the maps homo-

morphisms of

`

K

∗

(X) modules.

A

X K ( )

∼

K ( )

∼

K ( )

∼

X A , −−−−→

−

−

→

−

−

→

∗ ∗

∗

The

`

K

∗

(X) module structure on

`

K

∗

(A) is deﬁned by ξ·α = i

∗

(ξ)α where i is the

inclusion A

**X and the product on the right side of the equation is multiplication
**

in the ring

`

K

∗

(A). To deﬁne the module structure on

`

K

∗

(X, A), observe that the

Bott Periodicity Section 2.2 57

diagonal map X→X ∧ X induces a well-deﬁned quotient map X/A→X ∧ X/A, and

this leads to a product

`

K

∗

(X)⊗

`

K

∗

(X, A)→

`

K

∗

(X, A).

Proof: To see that the maps in the exact sequence are module homomorphisms we

look at the diagram

A S S K ( ( ) ) K ( )

∼ ∼

X/

j j

X S K( )

∼

j

A S K( )

∼

j

A S −−−−−→ −−−−−→ −−−−−→

A S S K ( ( ) ) K ( )

∼ ∼

X/

j

S X

i j

X S K( )

∼

j

A S K( )

∼

j

A S −−−→ −−−→ −−−→

A S S K ( ( ) ) K ( )

∼ ∼

X/

j i

X S K( )

∼

j

A S K( )

∼

j

A S −−−−−→ −−−−−→ −−−−−→

+ j i + i + i +

∧ S X

i

∧ S X

i

∧ S X

i

∧

−

−

→

−

−

→

−

−

→

−

−

→

−

−

→

−

−

→

−

−

→

−

−

→

where the vertical maps between the ﬁrst two rows are external product with a ﬁxed

element of

`

K(S

i

X) and the vertical maps between the second and third rows are

induced by diagonal maps. What we must showis that the diagramcommutes. For the

upper two rows this follows from naturality of external product since the horizontal

maps are induced by maps between spaces. The lower two rows are induced from

suspensions of maps between spaces,

A S X∧ X∧X/A X∧X X∧A − − − → − − − → − − − →

A S X/A X A − − − − → − − − − → − − − − →

−

−

→

−

−

→

−

−

→

−

−

→

so it sufﬁces to show this diagram commutes up to homotopy. This is obvious for the

middle and right squares. The left square can be rewritten

A S X∧ X∧ X − − − →

A S X CA − − − − →

−

−

→

−

−

→

∪

CA ∪ ( )

where the horizontal maps collapse the copy of X in X∪CA to a point, the left vertical

map sends (a, s) ∈ SA to (a, a, s) ∈ X ∧SA, and the right vertical map sends x ∈ X

to (x, x) ∈ X ∪ CA and (a, s) ∈ CA to (a, a, s) ∈ X ∧ CA. Commutativity is then

obvious. L¦

It is often convenient to have an unreduced version of the groups

`

K

n

(X), and this

can easily be done by the simple device of deﬁning K

n

(X) to be

`

K

n

(X

+

) where X

+

is

X with a disjoint basepoint labeled ‘+’ adjoined. For n = 0 this is consistent with the

relation between K and

`

K since K

0

(X) =

`

K

0

(X

+

) =

`

K(X

+

) = Ker(K(X

+

)→K(+)) =

K(X). For n = 1 this deﬁnition yields K

1

(X) =

`

K

1

(X) since S(X

+

) = SX ∨ S

1

and

`

K(SX∨S

1

) ≈

`

K(SX)⊕

`

K(S

1

) ≈

`

K(SX) since

`

K(S

1

) = 0. For a pair (X, A) with A ≠ ∅

one deﬁnes K

n

(X, A) =

`

K

n

(X, A), and then the six-term long exact sequence is valid

also for unreduced groups. When A = ∅ this remains valid if we interpret X/∅ as

X

+

.

Since X

+

∧Y

+

= (X×Y)

+

, the external product

`

K

∗

(X)⊗

`

K

∗

(Y)→

`

K

∗

(X∧Y) gives

a product K

∗

(X)⊗K

∗

(Y)→K

∗

(X×Y). Taking X = Y and composing with the map

58 Chapter 2 Complex K–Theory

K

∗

(X×X)→K

∗

(X) induced by the diagonal map X→X×X, x

(x, x), we get a

product K

∗

(X)⊗K

∗

(X)→K

∗

(X) which makes K

∗

(X) into a ring.

There is a relative product K

i

(X, A)⊗K

j

(Y, B)→K

i+j

(X×Y, X×B ∪ A×Y) de-

ﬁned as the external product

`

K(Σ

i

(X/A))⊗

`

K(Σ

j

(Y/B)) →

`

K(Σ

i+j

(X/A∧Y/B)), us-

ing the natural identiﬁcation (X×Y)/(X×B ∪A×Y) = X/A∧Y/B. This works when

A = ∅ since we interpret X/∅ as X

+

, and similarly if Y = ∅. Via the diagonal map

we obtain also a product K

i

(X, A)⊗K

j

(X, B)→K

i+j

(X, A∪B).

With these deﬁnitions the preceding two propositions are valid also for unreduced

K–groups.

Elementary Applications

With the calculation

`

K

∗

(S

n

) ≈ Z completed, it would be possible to derive many

of the same applications that follow from the corresponding calculation for ordinary

homology or cohomology, as in [AT]. For example:

— There is no retraction of D

n

onto its boundary S

n−1

, since this would mean that

the identity map of

`

K

∗

(S

n−1

) factored as

`

K

∗

(S

n−1

)→

`

K

∗

(D

n

)→

`

K

∗

(S

n−1

), but

the middle group is trivial.

— The Brouwer ﬁxed point theorem, that for every map f : D

n

→D

n

there is a point

x ∈ D

n

with f(x) = x. For if not then it is easy to construct a retraction of D

n

onto S

n−1

.

— The notion of degree for maps f : S

n

→S

n

, namely the integer d(f) such that the

induced homomorphism f

∗

:

`

K

∗

(S

n

)→

`

K

∗

(S

n

) is multiplication by d(f). Rea-

soning as in Proposition 2.2, one sees that d is a homomorphism π

n

(S

n

)→Z.

In particular a reﬂection has degree −1 and hence the antipodal map of S

n

,

which is the composition of n+1 reﬂections, has degree (−1)

n+1

since d(fg) =

d(f)d(g). Consequences of this include the fact that an even-dimensional sphere

has no nonvanishing vector ﬁelds.

However there are some things homology can do that would be harder using K–theory

since

`

K

∗

(S

n

) only distinguishes even-dimensional spheres from odd-dimensional

spheres. Also, since we have so far only deﬁned K–theory for compact spaces, it

would take more work to derive some of the other classical applications of homology

such as Brouwer’s theorems on invariance of dimension and invariance of domain, or

the Jordan curve theorem and its higher-dimensional analogs.

Division Algebras Section 2.3 59

2.3. Division Algebras and Parallelizable Spheres

With the hard work of proving Bott Periodicity now behind us, the goal of this

section is to prove Adams’ theoremon the Hopf invariant, with its famous applications

including the nonexistence of division algebras beyond the Cayley octonions:

Theorem 2.16. The following statements are true only for n = 1, 2, 4, and 8:

(a) R

n

is a division algebra.

(b) S

n−1

is parallelizable, i.e., there exist n−1 tangent vector ﬁelds to S

n−1

which

are linearly independent at each point, or in other words, the tangent bundle to S

n−1

is trivial.

A division algebra structure on R

n

is a multiplication map R

n

×R

n

→R

n

such

that the maps x

ax and x

**xa are linear for each a ∈ R
**

n

and invertible if a ≠ 0.

Since we are dealing with linear maps R

n

→R

n

, invertibility is equivalent to having

trivial kernel, which translates into the statement that the multiplication has no zero

divisors. An identity element is not assumed, but the multiplication can be modiﬁed

to produce an identity in the following way. Choose a unit vector e ∈ R

n

. After

composing the multiplication with an invertible linear map R

n

→R

n

taking e

2

to e

we may assume that e

2

= e. Let α be the map x

xe and β the map x

**ex. The new
**

product (x, y)

α

−1

(x)β

−1

(y) then sends (x, e) to α

−1

(x)β

−1

(e) = α

−1

(x)e = x,

and similarly it sends (e, y) to y. Since the maps x

ax and x

**xa are surjective
**

for each a ≠ 0, the equations ax = e and xa = e are solvable, so nonzero elements

of the division algebra have multiplicative inverses on the left and right.

H-Spaces

The ﬁrst step in the proof of the theorem is to reduce it to showing when the

sphere S

n−1

is an H–space.

To say that S

n−1

is an H–space means there is a continuous multiplication map

S

n−1

×S

n−1

→S

n−1

having a two-sided identity element e ∈ S

n−1

. This is weaker than

being a topological group since associativity and inverses are not assumed. For exam-

ple, S

1

, S

3

, and S

7

are H–spaces by restricting the multiplication of complex numbers,

quaternions, and Cayley octonions to the respective unit spheres, but only S

1

and S

3

are topological groups since the multiplication of octonions is nonassociative.

Lemma 2.17. If R

n

is a division algebra, or if S

n−1

is parallelizable, then S

n−1

is

an H–space.

Proof: Having a division algebra structure on R

n

with two-sided identity, an H–space

structure on S

n−1

is given by (x, y)

**xy/¦xy¦, which is well-deﬁned since a division
**

algebra has no zero divisors.

Now suppose that S

n−1

is parallelizable, with tangent vector ﬁelds v

1

, ··· , v

n−1

which are linearly independent at each point of S

n−1

. By the Gram-Schmidt process we

60 Chapter 2 Complex K–Theory

may make the vectors x, v

1

(x), ··· , v

n−1

(x) orthonormal for all x ∈ S

n−1

. We may

assume also that at the ﬁrst standard basis vector e

1

, the vectors v

1

(e

1

), ··· , v

n−1

(e

1

)

are the standard basis vectors e

2

, ··· , e

n

, by changing the sign of v

n−1

if necessary to

get orientations right, then deforming the vector ﬁelds near e

1

. Let α

x

∈ SO(n) send

the standard basis to x, v

1

(x), ··· , v

n−1

(x). Then the map (x, y)

α

x

(y) deﬁnes

an H–space structure on S

n−1

with identity element the vector e

1

since α

e

1

is the

identity map and α

x

(e

1

) = x for all x. L¦

Before proceeding further let us list a few easy consequences of Bott periodicity

which will be needed.

(1) We have already seen that

`

K(S

n

) is Z for n even and 0 for n odd. This comes

from repeated application of the periodicity isomorphism

`

K(X) ≈

`

K(S

2

X), α

**α∗(H−1), the external product with the generator H−1 of
**

`

K(S

2

), where H is

the canonical line bundle over S

2

= CP

1

. In particular we see that a generator of

`

K(S

2k

) is the k fold external product (H −1) ∗··· ∗(H −1). We note also that

the multiplication in

`

K(S

2k

) is trivial since this ring is the k fold tensor product

of the ring

`

K(S

2

), which has trivial multiplication by Corollary 2.3.

(2) The external product

`

K(S

2k

)⊗

`

K(X)→

`

K(S

2k

∧ X) is an isomorphism since it is

an iterate of the periodicity isomorphism.

(3) The external product K(S

2k

)⊗K(X)→K(S

2k

×X) is an isomorphism. This fol-

lows from(2) by the same reasoning which showed the equivalence of the reduced

and unreduced forms of Bott periodicity. Since external product is a ring homo-

morphism, the isomorphism

`

K(S

2k

∧X) ≈

`

K(S

2k

)⊗

`

K(X) is a ring isomorphism.

For example, since K(S

2k

) can be described as the quotient ring Z[α]/(α

2

), we

can deduce that K(S

2k

×S

2

) is Z[α, β]/(α

2

, β

2

) where α and β are the pullbacks

of generators of

`

K(S

2k

) and

`

K(S

2

) under the projections of S

2k

×S

2

onto its

two factors. An additive basis for K(S

2k

×S

2

) is thus ¦1, α, β, αβ¦.

We can apply the last calculation to show that S

2k

is not an H–space if k > 0.

Suppose µ : S

2k

×S

2k

→S

2k

is an H–space multiplication. The induced homomor-

phism of K–rings then has the form µ

∗

: Z[γ]/(γ

2

)→Z[α, β]/(α

2

, β

2

). We claim that

µ

∗

(γ) = α+β+mαβ for some integer m. The composition S

2k

i

→S

2k

×S

2k

µ

→S

2k

is the identity, where i is the inclusion onto either of the subspaces S

2k

×¦e¦ or

¦e¦×S

2k

, with e the identity element of the H–space structure. The map i

∗

for i the

inclusion onto the ﬁrst factor sends α to γ and β to 0, so the coefﬁcient of α in µ

∗

(γ)

must be 1. Similarly the coefﬁcient of β must be 1, proving the claim. However, this

leads to a contradiction since it implies that µ

∗

(γ

2

) = (α + β + mαβ)

2

= 2αβ ≠ 0,

which is impossible since γ

2

= 0.

There remains the much more difﬁcult problem of showing that S

n−1

is not an

H–space when n is even and different from 2, 4, and 8. The ﬁrst step is a simple

construction which associates to a map g : S

n−1

×S

n−1

→S

n−1

a map g : S

2n−1

→S

n

.

Division Algebras Section 2.3 61

To deﬁne this, we regard S

2n−1

as ∂(D

n

×D

n

) = ∂D

n

×D

n

∪ D

n

×∂D

n

, and S

n

we

take as the union of two disks D

n

+

and D

n

−

with their boundaries identiﬁed. Then

g is deﬁned on ∂D

n

×D

n

by g(x, y) = ¦y¦g(x, y/¦y¦) ∈ D

n

+

and on D

n

×∂D

n

by

g(x, y) = ¦x¦g(x/¦x¦, y) ∈ D

n

−

. Note that g is well-deﬁned and continuous, even

when ¦x¦ or ¦y¦ is zero, and g agrees with g on S

n−1

×S

n−1

.

Now we specialize to the case that n is even, or in other words, we replace n

by 2n. For a map f : S

4n−1

→S

2n

, let C

f

be S

2n

with a cell e

4n

attached by f . The

quotient C

f

/S

2n

is then S

4n

, and since

`

K

1

(S

4n

) =

`

K

1

(S

2n

) = 0, the exact sequence

of the pair (C

f

, S

2n

) becomes a short exact sequence

0 →

`

K(S

4n

) →

`

K(C

f

) →

`

K(S

2n

) →0

Let α ∈

`

K(C

f

) be the image of the generator (H −1) ∗··· ∗(H −1) of

`

K(S

4n

) and

let β ∈

`

K(C

f

) map to the generator (H −1) ∗··· ∗(H −1) of

`

K(S

2n

). The element

β

2

maps to 0 in

`

K(S

2n

) since the square of any element of

`

K(S

2n

) is zero. So by

exactness we have β

2

= hα for some integer h. The mod 2 value of h depends only

on f , not on the choice of β, since β is unique up to adding an integer multiple of

α, and (β + mα)

2

= β

2

+ 2mαβ since α

2

= 0. The value of h mod 2 is called the

mod 2 Hopf invariant of f . In fact αβ = 0 so h is well-deﬁned in Z not just Z

2

, as

we will see in §4.1, but for our present purposes the mod 2 value of h sufﬁces.

Lemma 2.18. If g : S

2n−1

×S

2n−1

→S

2n−1

is an H–space multiplication, then the as-

sociated map g : S

4n−1

→S

2n

has Hopf invariant :1.

Proof: Let e ∈ S

2n−1

be the identity element for the H–space multiplication, and let

f = g. In view of the deﬁnition of f it is natural to view the characteristic map Φ of

the 4n cell of C

f

as a map (D

2n

×D

2n

, ∂(D

2n

×D

2n

))→(C

f

, S

2n

). In the following

commutative diagramthe horizontal maps are the product maps. The diagonal map is

external product, equivalent to the external product

`

K(S

2n

)⊗

`

K(S

2n

)→

`

K(S

4n

), which

is an isomorphism since it is an iterate of the Bott periodicity isomorphism.

D e K

2n

( { ) } } ,

∼

× D e

2n

{ × ∂ e K D

2n

({ ) } } ,

∼

D e

2n

{ × ×∂

D K

2n

D

2n

D

2n

( ) ,

∼

× D

2n

× ∂ D K

2n

D

2n

D

2n

( ) ,

∼

× D

2n

×∂ ∂ D K

2n

D

2n

D

2n

( ( )) ,

∼

× D

2n

× −−−−−→

K D C

2n

( ) ,

∼

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→

f

K D C

2n

( ) ,

∼

f

K S C

2n

( ) ,

∼

f +

K C ( )

∼

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→

f

K C ( )

∼

f

K C ( )

∼

f

−

−

→

−

−

→

−

−

→

≈

≈

≈

−

−

→

−

−

→

≈

⊗

⊗

⊗

⊗

⊗

Φ

∗

Φ

∗

Φ

∗

By the deﬁnition of an H–space and the deﬁnition of f , the map Φ restricts to a

homeomorphism from D

2n

×¦e¦ onto D

2n

+

and from ¦e¦×D

2n

onto D

2n

−

. It follows

that the element β⊗β in the upper left group maps to a generator of the group in the

bottom row of the diagram, since β restricts to a generator of

`

K(S

2n

) by deﬁnition.

62 Chapter 2 Complex K–Theory

Therefore by commutativity of the diagram, the product map in the top row sends

β⊗β to :α since α was deﬁned to be the image of a generator of

`

K(C

f

, S

2n

). Thus

we have β

2

= :α, which says that the Hopf invariant of f is :1. L¦

In view of this lemma, Theorem 2.16 becomes a consequence of the following

theorem of Adams:

Theorem 2.19. If f : S

4n−1

→S

2n

is a map whose mod 2 Hopf invariant is 1, then

n = 1, 2, or 4.

The proof of this will occupy the rest of this section.

Adams Operations

The Hopf invariant is deﬁned in terms of the ring structure in K–theory, but in

order to prove Adams’ theorem, more structure is needed, namely certain ring homo-

morphisms ψ

k

: K(X)→K(X). Here are their basic properties:

Theorem 2.20. There exist ring homomorphisms ψ

k

: K(X)→K(X), deﬁned for all

compact Hausdorff spaces X and all integers k ≥ 0, and satisfying:

(1) ψ

k

f

∗

= f

∗

ψ

k

for all maps f : X→Y . (Naturality)

(2) ψ

k

(L) = L

k

if L is a line bundle.

(3) ψ

k

◦ ψ

= ψ

k

.

(4) ψ

p

(α) ≡ α

p

mod p for p prime.

This last statement means that ψ

p

(α) −α

p

= pβ for some β ∈ K(X).

In the special case of a vector bundle E which is a sum of line bundles L

i

, prop-

erties (2) and (3) give the formula ψ

k

(L

1

⊕ ··· ⊕L

n

) = L

k

1

+ ··· + L

k

n

. We would like

a general deﬁnition of ψ

k

(E) which specializes to this formula when E is a sum of

line bundles. The idea is to use the exterior powers λ

k

(E). From the corresponding

properties for vector spaces we have:

(i) λ

k

(E

1

⊕E

2

) ≈

¸

i

¸

λ

i

(E

1

)⊗λ

k−i

(E

2

)

.

(ii) λ

0

(E) = 1, the trivial line bundle.

(iii) λ

1

(E) = E.

(iv) λ

k

(E) = 0 for k greater than the maximum dimension of the ﬁbers of E.

Recall that we want ψ

k

(E) to be L

k

1

+···+L

k

n

when E = L

1

⊕ ··· ⊕L

n

for line bundles

L

i

. We will show in this case that there is a polynomial s

k

with integer coefﬁcients

such that L

k

1

+ ··· + L

k

n

= s

k

(λ

1

(E), ··· , λ

k

(E)). This will lead us to deﬁne ψ

k

(E) =

s

k

(λ

1

(E), ··· , λ

k

(E)) for an arbitrary vector bundle E.

To see what the polynomial s

k

should be, we ﬁrst use the exterior powers λ

i

(E)

to deﬁne a polynomial λ

t

(E) =

¸

i

λ

i

(E)t

i

∈ K(X)[t]. This is a ﬁnite sum by property

(iv), and property (i) says that λ

t

(E

1

⊕E

2

) = λ

t

(E

1

)λ

t

(E

2

). When E = L

1

⊕ ··· ⊕L

n

this implies that λ

t

(E) =

¸

i

λ

t

(L

i

), which equals

¸

i

(1 + L

i

t) by (ii), (iii), and (iv).

Division Algebras Section 2.3 63

The coefﬁcient λ

j

(E) of t

j

in λ

t

(E) =

¸

i

(1 + L

i

t) is the j

th

elementary symmetric

function σ

j

of the L

i

’s, the sum of all products of j distinct L

i

’s. Thus we have

(∗) λ

j

(E) = σ

j

(L

1

, ··· , L

n

) if E = L

1

⊕ ··· ⊕L

n

To make the discussion completely algebraic, let us introduce the variable t

i

for

L

i

. Thus (1+t

1

) ··· (1+t

n

) = 1+σ

1

+··· +σ

n

, where σ

j

is the j

th

elementary sym-

metric polynomial in the t

i

’s. The fundamental theorem on symmetric polynomials,

proved for example in [Lang, p. 134] or [van der Waerden, §26], asserts that every

degree k symmetric polynomial in t

1

, ··· , t

n

can be expressed as a unique polyno-

mial in σ

1

, ··· , σ

k

. In particular, t

k

1

+··· +t

k

n

is a polynomial s

k

(σ

1

, ··· , σ

k

), called a

Newton polynomial. This polynomial s

k

is independent of n since we can pass from

n to n−1 by setting t

n

= 0. A recursive formula for s

k

is

s

k

= σ

1

s

k−1

−σ

2

s

k−2

+··· +(−1)

k−2

σ

k−1

s

1

+(−1)

k−1

kσ

k

To derive this we may take n = k, and then if we substitute x = −t

i

in the identity

(x +t

1

) ··· (x +t

k

) = x

k

+σ

1

x

k−1

+··· +σ

k

we get t

k

i

= σ

1

t

k−1

i

−··· +(−1)

k−1

σ

k

.

Summing over i then gives the recursion relation. The recursion relation easily yields

for example

s

1

= σ

1

s

2

= σ

2

1

−2σ

2

s

3

= σ

3

1

−3σ

1

σ

2

+3σ

3

s

4

= σ

4

1

−4σ

2

1

σ

2

+4σ

1

σ

3

+2σ

2

2

−4σ

4

Summarizing, if we deﬁne ψ

k

(E) = s

k

(λ

1

(E), ··· , λ

k

(E)), then in the case that E

is a sum of line bundles L

1

⊕ ··· ⊕L

n

we have

ψ

k

(E) = s

k

(λ

1

(E), ··· , λ

k

(E))

= s

k

(σ

1

(L

1

, ··· , L

n

), ··· , σ

k

(L

1

, ··· , L

n

)) by (∗)

= L

k

1

+··· +L

k

n

Verifying that the deﬁnition ψ

k

(E) = s

k

(λ

1

(E), ··· , λ

k

(E)) gives operations on

K(X) satisfying the properties listed in the theorem will be rather easy if we make

use of the following general result:

The Splitting Principle. Given a vector bundle E→X with X compact Hausdorff,

there is a compact Hausdorff space F(E) and a map p: F(E)→X such that the in-

duced map p

∗

: K

∗

(X)→K

∗

(F(E)) is injective and p

∗

(E) splits as a sum of line

bundles.

This will be proved later in this section, but for the moment let us assume it and

proceed with the proof of Theorem 2.20 and Adams’ theorem.

Proof of Theorem 2.20: Property (1) for vector bundles, f

∗

(ψ

k

(E)) = ψ

k

(f

∗

(E)),

follows immediately from the relation f

∗

(λ

i

(E)) = λ

i

(f

∗

(E)). Additivity of ψ

k

for

64 Chapter 2 Complex K–Theory

vector bundles, ψ

k

(E

1

⊕E

2

) = ψ

k

(E

1

) + ψ

k

(E

2

), follows from the splitting principle

since we can ﬁrst pull back to split E

1

then do a further pullback to split E

2

, and the

formula ψ

k

(L

1

⊕ ··· ⊕L

n

) = L

k

1

+ ··· + L

k

n

preserves sums. Since ψ

k

is additive on

vector bundles, it induces an additive operation on K(X) deﬁned by ψ

k

(E

1

− E

2

) =

ψ

k

(E

1

) −ψ

k

(E

2

).

For this extended ψ

k

the properties (1) and (2) are clear. Multiplicativity is also

easy from the splitting principle: If E is the sum of line bundles L

i

and E

¹

is the sum

of line bundles L

¹

j

, then E⊗E

¹

is the sum of the line bundles L

i

⊗L

¹

j

, so ψ

k

(E⊗E

¹

) =

¸

i,j

ψ

k

(L

i

⊗L

¹

j

) =

¸

i,j

(L

i

⊗L

¹

j

)

k

=

¸

i,j

L

k

i

⊗L

¹

j

k

=

¸

i

L

k

i

¸

j

L

¹

j

k

= ψ

k

(E)ψ

k

(E

¹

). Thus

ψ

k

is multiplicative for vector bundles, and it follows formally that it is multiplicative

on elements of K(X). For property (3) the splitting principle and additivity reduce

us to the case of line bundles, where ψ

k

(ψ

(L)) = L

k

= ψ

k

(L). Likewise for (4), if

E = L

1

+··· +L

n

, then ψ

p

(E) = L

p

1

+··· +L

p

n

≡ (L

1

+··· +L

n

)

p

= E

p

mod p. L¦

By the naturality property (1), ψ

k

restricts to an operation ψ

k

:

`

K(X)→

`

K(X) since

`

K(X) is the kernel of the homomorphism K(X)→K(x

0

) for x

0

∈ X. For the external

product

`

K(X)⊗

`

K(Y)→

`

K(X ∧Y), we have the formula ψ

k

(α∗β) = ψ

k

(α) ∗ψ

k

(β)

since if one looks back at the deﬁnition of α ∗ β, one sees this was deﬁned as

p

∗

1

(α)p

∗

2

(β), hence

ψ

k

(α∗β) = ψ

k

(p

∗

1

(α)p

∗

2

(β))

= ψ

k

(p

∗

1

(α))ψ

k

(p

∗

2

(β))

= p

∗

1

(ψ

k

(α))p

∗

2

(ψ

k

(β))

= ψ

k

(α) ∗ψ

k

(β).

This will allow us to compute ψ

k

on

`

K(S

2n

) ≈ Z. In this case ψ

k

must be

multiplication by some integer since it is an additive homomorphism of Z.

Proposition 2.21. ψ

k

:

`

K(S

2n

)→

`

K(S

2n

) is multiplication by k

n

.

Proof: Consider ﬁrst the case n = 1. Since ψ

k

is additive, it will sufﬁce to show

ψ

k

(α) = kα for α a generator of

`

K(S

2

). We can take α = H −1 for H the canonical

line bundle over S

2

= CP

1

. Then

ψ

k

(α) = ψ

k

(H −1) = H

k

−1 by property (2)

= (1 +α)

k

−1

= 1 +kα−1 since α

i

= (H −1)

i

= 0 for i ≥ 2

= kα

When n > 1 we use the external product

`

K(S

2

)⊗

`

K(S

2n−2

)→

`

K(S

2n

), which is

an isomorphism, and argue by induction. Assuming the desired formula holds in

`

K(S

2n−2

), we have ψ

k

(α ∗ β) = ψ

k

(α) ∗ ψ

k

(β) = kα ∗ k

n−1

β = k

n

(α ∗ β), and we

are done. L¦

Division Algebras Section 2.3 65

Now we can use the operations ψ

2

and ψ

3

and the relation ψ

2

ψ

3

= ψ

6

= ψ

3

ψ

2

to prove Adams’ theorem.

Proof of Theorem 2.19: The deﬁnition of the Hopf invariant of a map f : S

4n−1

→S

2n

involved elements α, β ∈

`

K(C

f

). By Proposition 2.21, ψ

k

(α) = k

2n

α since α is the

image of a generator of

`

K(S

4n

). Similarly, ψ

k

(β) = k

n

β + µ

k

α for some µ

k

∈ Z.

Therefore

ψ

k

ψ

(β) = ψ

k

(

n

β +µ

α) = k

n

n

β +(k

2n

µ

+

n

µ

k

)α

Since ψ

k

ψ

= ψ

k

= ψ

ψ

k

, the coefﬁcient k

2n

µ

+

n

µ

k

of α is unchanged when k

and are switched. This gives the relation

k

2n

µ

+

n

µ

k

=

2n

µ

k

+k

n

µ

, or (k

2n

−k

n

)µ

= (

2n

−

n

)µ

k

By property (6) of ψ

2

, we have ψ

2

(β) ≡ β

2

mod 2. Since β

2

= hα with h the Hopf

invariant of f , the formula ψ

2

(β) = 2

n

β +µ

2

α implies that µ

2

≡ h mod 2, so µ

2

is

odd if we assume h = :1. By the preceding displayed formula we have (2

2n

−2

n

)µ

3

=

(3

2n

−3

n

)µ

2

, or 2

n

(2

n

−1)µ

3

= 3

n

(3

n

−1)µ

2

, so 2

n

divides 3

n

(3

n

−1)µ

2

. Since 3

n

and µ

2

are odd, 2

n

must then divide 3

n

−1. The proof is completed by the following

elementary number theory fact. L¦

Lemma 2.22. If 2

n

divides 3

n

−1 then n = 1, 2, or 4.

Proof: Write n = 2

**m with m odd. We will showthat the highest power of 2 dividing
**

3

n

−1 is 2 for = 0 and 2

+2

for > 0. This implies the lemma since if 2

n

divides

3

n

−1, then by this fact, n ≤ +2, hence 2

≤ 2

m= n ≤ +2, which implies ≤ 2

and n ≤ 4. The cases n = 1, 2, 3, 4 can then be checked individually.

We ﬁnd the highest power of 2 dividing 3

n

− 1 by induction on . For = 0

we have 3

n

−1 = 3

m

−1 ≡ 2 mod 4 since 3 ≡ −1 mod 4 and m is odd. In the next

case = 1 we have 3

n

− 1 = 3

2m

− 1 = (3

m

− 1)(3

m

+ 1). The highest power of 2

dividing the ﬁrst factor is 2 as we just showed, and the highest power of 2 dividing

the second factor is 4 since 3

m

+ 1 ≡ 4 mod 8 because 3

2

≡ 1 mod 8 and m is

odd. So the highest power of 2 dividing the product (3

m

−1)(3

m

+1) is 8. For the

inductive step of passing from to + 1 with ≥ 1, or in other words from n to

2n with n even, write 3

2n

− 1 = (3

n

− 1)(3

n

+ 1). Then 3

n

+ 1 ≡ 2 mod 4 since n

is even, so the highest power of 2 dividing 3

n

+1 is 2. Thus the highest power of 2

dividing 3

2n

−1 is twice the highest power of 2 dividing 3

n

−1. L¦

The Splitting Principle

The splitting principle will be a fairly easy consequence of a general result about

the K–theory of ﬁber bundles called the Leray-Hirsch theorem, together with a calcu-

lation of the ring structure of K

∗

(CP

n

). The following proposition will allow us to

compute at least the additive structure of K

∗

(CP

n

).

66 Chapter 2 Complex K–Theory

Proposition 2.23. If X is a ﬁnite cell complex with n cells, then K

∗

(X) is a ﬁnitely

generated group with at most n generators. If all the cells of X have even dimension

then K

1

(X) = 0 and K

0

(X) is free abelian with one basis element for each cell.

The phrase ‘ﬁnite cell complex’ would normally mean ‘ﬁnite CW complex’ but we

can take it to be something slightly more general: a space built froma ﬁnite discrete set

by attaching a ﬁnite number of cells in succession, with no conditions on the dimen-

sions of these cells, so cells are not required to attach only to cells of lower dimension.

Finite cell complexes are always homotopy equivalent to ﬁnite CW complexes (by de-

forming each successive attaching map to be cellular) so the only advantages of ﬁnite

cell complexes are technical. In particular, it is easy to see that a space is a ﬁnite cell

complex if it is a ﬁber bundle over a ﬁnite cell complex with ﬁbers that are also ﬁnite

cell complexes. This is shown in Proposition 2.28 in a brief appendix to this section.

It implies that the splitting principle can be applied staying within the realm of ﬁnite

cell complexes.

Proof: We show this by induction on the number of cells. The complex X is obtained

from a subcomplex A by attaching a k cell, for some k. For the pair (X, A) we

have an exact sequence

`

K

∗

(X/A) →

`

K

∗

(X) →

`

K

∗

(A). Since X/A = S

k

, we have

`

K

∗

(X/A) ≈ Z, and exactness implies that

`

K

∗

(X) requires at most one more generator

than

`

K

∗

(A).

The ﬁrst term of the exact sequence K

1

(X/A)→K

1

(X)→K

1

(A) is zero if all

cells of X are of even dimension, so induction on the number of cells implies that

K

1

(X) = 0. Then there is a short exact sequence 0→

`

K

0

(X/A)→

`

K

0

(X)→

`

K

0

(A)→0

with

`

K

0

(X/A) ≈ Z. By induction

`

K(A) is free, so this sequence splits, hence K

0

(X) ≈

Z⊕K

0

(A) and the ﬁnal statement of the proposition follows. L¦

This proposition applies in particular to CP

n

, which has a cell structure with one

cell in each dimension 0, 2, 4, ··· , 2n, so K

1

(CP

n

) = 0 and K

0

(CP

n

) ≈ Z

n+1

. The ring

structure is as simple as one could hope for:

Proposition 2.24. K(CP

n

) is the quotient ring Z[L]/(L−1)

n+1

where L is the canon-

ical line bundle over CP

n

.

Thus by the change of variable x = L−1 we see that K(CP

n

) is the truncated poly-

nomial ring Z[x]/(x

n+1

), with additive basis 1, x, ··· , x

n

. It follows that 1, L, ··· , L

n

is also an additive basis.

Proof: The exact sequence for the pair (CP

n

, CP

n−1

) gives a short exact sequence

0 →K(CP

n

, CP

n−1

) →K(CP

n

)

ρ

→K(CP

n−1

) →0

We shall prove:

(a

n

) (L −1)

n

generates the kernel of the restriction map ρ.

Division Algebras Section 2.3 67

Hence if we assume inductively that K(CP

n−1

) = Z[L]/(L − 1)

n

, then (a

n

) and the

preceding exact sequence imply that ¦1, L − 1, ··· , (L − 1)

n

¦ is an additive basis for

K(CP

n

). Since (L − 1)

n+1

= 0 in K(CP

n

) by (a

n+1

), it follows that K(CP

n

) is the

quotient ring Z[L]/(L −1)

n+1

, completing the induction.

A reason one might expect (a

n

) to be true is that the kernel of ρ can be identi-

ﬁed with K(CP

n

, CP

n−1

) =

`

K(S

2n

) by the short exact sequence, and Bott periodicity

implies that the n fold reduced external product of the generator L − 1 of

`

K(S

2

)

with itself generates

`

K(S

2n

). To make this rough argument into a proof we will have

to relate the external product

`

K(S

2

)⊗ ··· ⊗

`

K(S

2

)→

`

K(S

2n

) to the ‘internal’ product

K(CP

n

)⊗ ··· ⊗K(CP

n

)→K(CP

n

).

The space CP

n

is the quotient of the unit sphere S

2n+1

in C

n+1

under multipli-

cation by scalars in S

1

⊂ C. Instead of S

2n+1

we could equally well take the boundary

of the product D

2

0

× ··· ×D

2

n

where D

2

i

is the unit disk in the i

th

coordinate of C

n+1

,

and we start the count with i = 0 for convenience. Then we have

∂(D

2

0

× ··· ×D

2

n

) =

i

(D

2

0

× ··· ×∂D

2

i

× ··· ×D

2

n

)

The action of S

1

by scalar multiplication respects this decomposition. The orbit space

of D

2

0

× ··· ×∂D

2

i

× ··· ×D

2

n

under the action is a subspace C

i

⊂ CP

n

homeomorphic

to the product D

2

0

× ··· ×D

2

n

with the factor D

2

i

deleted. Thus we have a decomposi-

tion CP

n

=

i

C

i

with each C

i

homeomorphic to D

2n

and with C

i

∩ C

j

= ∂C

i

∩ ∂C

j

for i ≠ j .

Consider now C

0

= D

2

1

× ··· ×D

2

n

. Its boundary is decomposed into the pieces

∂

i

C

0

= D

2

1

× ··· ×∂D

2

i

× ··· ×D

2

n

. The inclusions (D

2

i

, ∂D

2

i

) ⊂ (C

0

, ∂

i

C

0

) ⊂ (CP

n

, C

i

)

give rise to a commutative diagram

D K

2

1

D

2

1

( ) , ∂ D K

2

n

D

2

n

( ) , ∂

−−−−−−−−−−→

−−−−−→ −−−−−→

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→

−−−−−−−−−−−−−−−−−→

− − − − − − − − − − − − − − − − − →

−

−

→

−

−

→

−

−

→

−

−

→

−

−

→

≈

≈

. . .

C K

0 0 0 1

C

0

( ) , ∂ C K

n

C ( ) , ∂

0 0

C K C ( ) , ∂

. . .

K

1

C P ( ) ,

n

n

. . . . . .

≈

≈

≈

C K C P ( ) K

1

C P ( ,

n

n

C C ) ) ,

n

C

K P ( )

n

. . .

C K P ( )

n

C K P ( )

n

C

∪ ∪ K P ( ,

n

C P C

n 1

⊗ ⊗

⊗ ⊗

⊗ ⊗

⊗ ⊗

where the maps from the ﬁrst column to the second are the n fold products. The

upper map in the middle column is an isomorphism because the inclusion C

0

CP

n

induces a homeomorphism C

0

/∂C

0

≈ CP

n

/(C

1

∪···∪C

n

). The CP

n−1

at the right side

of the diagram sits in CP

n

in the last n coordinates of C

n+1

, so is disjoint from C

0

,

hence the quotient map CP

n

/CP

n−1

→CP

n

/(C

1

∪···∪C

n

) is a homotopy equivalence.

The element x

i

∈ K(CP

n

, C

i

) mapping downward to L−1 ∈ K(CP

n

) maps upward

to a generator of K(C

0

, ∂

i

C

0

) ≈ K(D

2

i

, ∂D

2

i

). By commutativity of the diagram, the

product x

1

··· x

n

then generates K(CP

n

, C

1

∪ ··· ∪ C

n

). This means that (L − 1)

n

68 Chapter 2 Complex K–Theory

generates the image of the map K(CP

n

, CP

n−1

)→K(CP

n

), which equals the kernel of

ρ, proving (a

n

). L¦

Since CP

n

is the union of the n+1 balls C

i

, Example 2.13 shows that all products

of n + 1 elements of

`

K(CP

n

) must be zero, in particular (L − 1)

n+1

= 0. But as we

have just seen, (L −1)

n

is nonzero, so the result in Example 2.13 is best possible in

terms of the degree of nilpotency.

Now we come to the Leray-Hirsch theorem for K–theory, which will be the major

theoretical ingredient in the proof of the splitting principle:

Theorem2.25. Let p: E→B be a ﬁber bundle with E and B compact Hausdorff and

with ﬁber F such that K

∗

(F) is free. Suppose that there exist classes c

1

, ··· , c

k

∈

K

∗

(E) that restrict to a basis for K

∗

(F) in each ﬁber F . If either

(a) B is a ﬁnite cell complex, or

(b) F is a ﬁnite cell complex having all cells of even dimension,

then K

∗

(E), as a module over K

∗

(B), is free with basis ¦c

1

, ··· , c

k

¦.

Here the K

∗

(B) module structure on K

∗

(E) is deﬁned by β · γ = p

∗

(β)γ for

β ∈ K

∗

(B) and γ ∈ K

∗

(E). Another way to state the conclusion of the theorem is

to say that the map Φ: K

∗

(B)⊗K

∗

(F)→K

∗

(E), Φ(

¸

i

b

i

⊗i

∗

(c

i

)) =

¸

i

p

∗

(b

i

)c

i

for i

the inclusion F

E, is an isomorphism.

In the case of the product bundle E = F×B the classes c

i

can be chosen to be the

pullbacks under the projection E→F of a basis for K

∗

(F). The theorem then asserts

that the external product K

∗

(F)⊗K

∗

(B)→K

∗

(F×B) is an isomorphism.

For most of our applications of the theorem either case (a) or case (b) will sufﬁce.

The proof of (a) is somewhat simpler than (b), and we include (b) mainly to obtain the

splitting principle for vector bundles over arbitrary compact Hausdorff base spaces.

Proof: For a subspace B

¹

⊂ B let E

¹

= p

−1

(B

¹

). Then we have a diagram

−

−

→

−

−

→

−

−

→

−−−−−→ −−−−−→

∗

∗

⊗

Φ Φ Φ

B B K ( )

( )

,

∗

F K ( )

−−−−−→

∗

⊗ B K ( )

∗

F K ( )

−−−−−→

∗

⊗ B K ( )

∗

F K ( )

−−−−−−−−−−−−−−−→ −−−−−−−−−−−−→

∗

E E K ( ) , −−−−−−−−−−−−−−−−−→

∗

E K ( )

−−−−−−−−−−−→

∗

E K ( )

where the left-hand Φ is deﬁned by the same formula Φ(

¸

i

b

i

⊗i

∗

(c

i

)) =

¸

i

p

∗

(b

i

)c

i

,

but with p

∗

(b

i

)c

i

referring nowto the relative product K

∗

(E, E

¹

)×K

∗

(E)→K

∗

(E, E

¹

).

The right-hand Φ is deﬁned using the restrictions of the c

i

’s to the subspace E

¹

. To

see that the diagram (∗) commutes, we can interpolate between its two rows the row

→K

∗

(E, E

¹

)⊗K

∗

(F) →K

∗

(E)⊗K

∗

(F) →K

∗

(E

¹

)⊗K

∗

(F) →

by factoring Φ as the composition

¸

i

b

i

⊗i

∗

(c

i

)

¸

i

p

∗

(b

i

) ⊗i

∗

(c

i

)

¸

i

p

∗

(b

i

)c

i

.

The upper squares of the enlarged diagram then commute trivially, and the lower

squares commute by Proposition 2.15. The lower row of the diagram is of course

Division Algebras Section 2.3 69

exact. The upper row is also exact since we assume K

∗

(F) is free, and tensoring

an exact sequence with a free abelian group preserves exactness, the result of the

tensoring operation being simply to replace the given exact sequence by the direct

sum of a number of copies of itself.

The proof in case (a) will be by a double induction, ﬁrst on the dimension of B,

then within a given dimension, on the number of cells. The induction starts with the

trivial case that B is zero-dimensional, hence a ﬁnite discrete set. For the induction

step, suppose B is obtained from a subcomplex B

¹

by attaching a cell e

n

, and let

E

¹

= p

−1

(B

¹

) as above. By induction on the number of cells of B we may assume the

right-hand Φ in (∗) is an isomorphism. If the left-hand Φ is also an isomorphism,

then the ﬁve-lemma will imply that the middle Φ is an isomorphism, ﬁnishing the

induction step.

Let ϕ: (D

n

, S

n−1

)→(B, B

¹

) be a characteristic map for the attached n cell. Since

D

n

is contractible, the pullback bundle ϕ

∗

(E) is a product, and so we have a com-

mutative diagram

−

−

→

−

−

→

−−−−−→

∗

⊗

Φ Φ

Φ

B B K ( ) , ,

∗

F K ( ) ⊗

∗

F K ( )

∗

D S K ( )

−−−−−−−−−−−−→

∗

E E K ( ) , ,

∗ ∗

E K ( ( )

−−−−

−

−

−

−

−

−

−

−−→

∗

K (

≈

≈

≈

ϕ

∗

E ( )) ϕ × ×

n n 1

, F F D S )

n n 1

The two horizontal maps are isomorphisms since ϕ restricts to a homeomorphism

on the interior of D

n

, hence induces homeomorphisms B/B

¹

≈ D

n

/S

n−1

and E/E

¹

≈

ϕ

∗

(E)/ϕ

∗

(E

¹

). Thus the diagram reduces the proof to showing that the right-hand

Φ, involving the product bundle D

n

×F→D

n

, is an isomorphism.

Consider the diagram (∗) with (B, B

¹

) replaced by (D

n

, S

n−1

). We may assume

the right-hand Φ in (∗) is an isomorphism since S

n−1

has smaller dimension than

the original cell complex B. The middle Φ is an isomorphism by the case of zero-

dimensional B since D

n

deformation retracts to a point. Therefore by the ﬁve-lemma

the left-hand Φ in (∗) is an isomorphism for (B, B

¹

) = (D

n

, S

n−1

). This ﬁnishes the

proof in case (a).

In case (b) let us ﬁrst prove the result for a product bundle E = F×B. In this case

Ψ is just the external product, so we are free to interchange the roles of F and B.

Thus we may use the diagram (∗) with F an arbitrary compact Hausdorff space and

B a ﬁnite cell complex having all cells of even dimension, obtained by attaching a cell

e

n

to a subcomplex B

¹

. The upper row of (∗) is then an exact sequence since it is

obtained from the split short exact sequence 0→K

∗

(B, B

¹

)→K

∗

(B)→K

∗

(B

¹

)→0 by

tensoring with the ﬁxed group K

∗

(F). If we can show that the left-hand Φ in (∗) is

an isomorphism, then by induction on the number of cells of B we may assume the

right-hand Φ is an isomorphism, so the ﬁve-lemma will imply that the middle Φ is

also an isomorphism.

To show the left-hand Φ is an isomorphism, note ﬁrst that B/B

¹

= S

n

so we may

70 Chapter 2 Complex K–Theory

as well take the pair (B, B

¹

) to be (D

n

, S

n−1

). Then the middle Φ in (∗) is obviously

an isomorphism, so the left-hand Φ will be an isomorphism iff the right-hand Φ is

an isomorphism. When the sphere S

n−1

is even-dimensional we have already shown

that Φ is an isomorphism in the remarks following the proof of Lemma 2.17, and

the same argument applies also when the sphere is odd-dimensional, since K

1

of an

odd-dimensional sphere is K

0

of an even-dimensional sphere.

Now we turn to case (b) for nonproducts. The proof will once again be inductive,

but this time we need a more subtle inductive statement since B is just a compact

Hausdorff space, not a cell complex. Consider the following condition on a compact

subspace U ⊂ B:

For all compact V ⊂ U the map Φ: K

∗

(V)⊗K

∗

(F)→K

∗

(p

−1

(V)) is an isomor-

phism.

If this is satisﬁed, let us call U good. By the special case already proved, each point of

B has a compact neighborhood U that is good. Since B is compact, a ﬁnite number

of these neighborhoods cover B, so by induction it will be enough to show that if U

1

and U

2

are good, then so is U

1

∪U

2

.

A compact V ⊂ U

1

∪U

2

is the union of V

1

= V ∩U

1

and V

2

= V ∩U

2

. Consider

the diagram like (∗) for the pair (V, V

2

). Since K

∗

(F) is free, the upper row of this

diagram is exact. Assuming U

2

is good, the map Φ is an isomorphism for V

2

, so Φ

will be an isomorphism for V if it is an isomorphism for (V, V

2

). The quotient V/V

2

is homeomorphic to V

1

/(V

1

∩V

2

) so Φ will be an isomorphism for (V, V

2

) if it is an

isomorphism for (V

1

, V

1

∩ V

2

). Now look at the diagram like (∗) for (V

1

, V

1

∩ V

2

).

Assuming U

1

is good, the maps Φ are isomorphisms for V

1

and V

1

∩V

2

. Hence Φ is

an isomorphism for (V

1

, V

1

∩V

2

), and the induction step is ﬁnished. L¦

Example 2.26. Let E→X be a vector bundle with ﬁbers C

n

and compact base X.

Then we have an associated projective bundle p: P(E)→X with ﬁbers CP

n−1

, where

P(E) is the space of lines in E, that is, one-dimensional linear subspaces of ﬁbers of

E. Over P(E) there is the canonical line bundle L→P(E) consisting of the vectors in

the lines of P(E). In each ﬁber CP

n−1

of P(E) the classes 1, L, ··· , L

n−1

in K

∗

(P(E))

restrict to a basis for K

∗

(CP

n−1

) by Proposition 2.24. From the Leray-Hirsch theorem

we deduce that K

∗

(P(E)) is a free K

∗

(X) module with basis 1, L, ··· , L

n−1

.

Proof of the Splitting Principle: In the preceding example, the fact that 1 is among the

basis elements implies that p

∗

: K

∗

(X)→K

∗

(P(E)) is injective. The pullback bundle

p

∗

(E)→P(E) contains the line bundle L as a subbundle, hence splits as L⊕E

¹

for

E

¹

→P(E) the subbundle of p

∗

(E) orthogonal to L with respect to some choice of

inner product. Now repeat the process by forming P(E

¹

), splitting off another line

bundle from the pullback of E

¹

over P(E

¹

). Note that P(E

¹

) is the space of pairs

of orthogonal lines in ﬁbers of E. After a ﬁnite number of repetitions we obtain

the ﬂag bundle F(E)→X described at the end of §1.1, whose points are n tuples of

Division Algebras Section 2.3 71

orthogonal lines in ﬁbers of E, where n is the dimension of E. (If the ﬁbers of E

have different dimensions over different components of X, we do the construction

for each component separately.) The pullback of E over F(E) splits as a sum of line

bundles, and the map F(E)→X induces an injection on K

∗

since it is a composition

of maps with this property. L¦

In the preceding Example 2.26 we saw that K

∗

(P(E)) is free as a K

∗

(X) module,

with basis 1, L, ··· , L

n−1

. In order to describe the multiplication in K

∗

(P(E)) one

therefore needs only a relation expressing L

n

in terms of lower powers of L. Such

a relation can be found as follows. The pullback of E over P(E) splits as L⊕E

¹

for

some bundle E

¹

of dimension n−1, and the desired relation will be λ

n

(E

¹

) = 0. To

compute λ

n

(E

¹

) = 0 we use the formula λ

t

(E) = λ

t

(L)λ

t

(E

¹

) in K

∗

(P(E))[t], where

to simplify notation we let ‘ E’ also denote the pullback of E over P(E). The equation

λ

t

(E) = λ

t

(L)λ

t

(E

¹

) can be rewritten as λ

t

(E

¹

) = λ

t

(E)λ

t

(L)

−1

where λ

t

(L)

−1

=

¸

i

(−1)

i

L

i

t

i

since λ

t

(L) = 1 + Lt . Equating coefﬁcients of t

n

in the two sides of

λ

t

(E

¹

) = λ

t

(E)λ

t

(L)

−1

, we get λ

n

(E

¹

) =

¸

i

(−1)

n−i

λ

i

(E)L

n−i

. The relation λ

n

(E

¹

) = 0

can be written as

¸

i

(−1)

i

λ

i

(E)L

n−i

= 0, with the coefﬁcient of L

n

equal to 1, as

desired. The result can be stated in the following form:

Proposition 2.27. For an n dimensional vector bundle E→X the ring K(P(E)) is

isomorphic to the quotient ring K

∗

(X)[L]/

¸¸

i

(−1)

i

λ

i

(E)L

n−i

. L¦

For example when X is a point we have P(E) = CP

n−1

and λ

i

(E) = C

k

for k =

n

i

,

so the polynomial

¸

i

(−1)

i

λ

i

(E)L

n−i

becomes (L−1)

n

and we see that the proposition

generalizes the isomorphism K

∗

(CP

n−1

) ≈ Z[L]/(L −1)

n

).

Appendix: Finite Cell Complexes

As we mentioned in the remarks following Proposition 2.23 it is convenient for

purposes of the splitting principle to work with spaces slightly more general than ﬁnite

CW complexes. By a ﬁnite cell complex we mean a space which has a ﬁnite ﬁltration

X

0

⊂ X

1

⊂ ··· ⊂ X

k

= X where X

0

is a ﬁnite discrete set and X

i+1

is obtained from

X

i

by attaching a cell e

n

i

via a map ϕ

i

: S

n

i

−1

→X

i

. Thus X

i+1

is the quotient space

of the disjoint union of X

i

and a disk D

n

i

under the identiﬁcations x ∼ ϕ

i

(x) for

x ∈ ∂D

n

i

= S

n

i

−1

.

Proposition 2.28. If p: E→B is a ﬁber bundle whose ﬁber F and base B are both

ﬁnite cell complexes, then E is also a ﬁnite cell complex, whose cells are products of

cells in B with cells in F .

Proof: Suppose B is obtained from a subcomplex B

¹

by attaching a cell e

n

. By induc-

tion on the number of cells of B we may assume that p

−1

(B

¹

) is a ﬁnite cell complex.

If Φ: D

n

→B is a characteristic map for e

n

then the pullback bundle Φ

∗

(E)→D

n

is

a product since D

n

is contractible. Since F is a ﬁnite cell complex, this means that

72 Chapter 2 Complex K–Theory

we may obtain Φ

∗

(E) from its restriction over S

n−1

by attaching cells. Hence we may

obtain E from p

−1

(B

¹

) by attaching cells. L¦

Characteristic classes are cohomology classes associated to vector bundles. They

measure in some way how a vector bundle is twisted, or nontrivial. There are four

main kinds:

1. Stiefel-Whitney classes w

i

(E) ∈ H

i

(B; Z

2

) for a real vector bundle E→B.

2. Chern classes c

i

(E) ∈ H

2i

(B; Z) for a complex vector bundle E→B.

3. Pontryagin classes p

i

(E) ∈ H

4i

(B; Z) for a real vector bundle E→B.

4. The Euler class e(E) ∈ H

n

(B; Z) for an oriented n dimensional real vector bundle

E→B.

The Stiefel-Whitney and Chern classes are formally quite similar. Pontryagin classes

can be regarded as a reﬁnement of Stiefel-Whitney classes when one takes Z rather

than Z

2

coefﬁcients, and the Euler class is a further reﬁnement in the orientable case.

For an n dimensional vector bundle E→B to be trivial is equivalent to its clas-

sifying map f : B→G

n

being nullhomotopic, but as with most things in homotopy

theory it can be quite difﬁcult to determine whether this is the case or not. Much

more accessible is the weaker question of whether f induces a nontrivial map on ho-

mology or cohomology, and this is precisely what characteristic classes measure. The

Stiefel-Whitney classes w

i

, like the other characteristic classes, satisfy the important

naturality property that w

i

(f

∗

(E)) = f

∗

(w

i

(E)), so in particular if f is a classifying

map B→G

n

for E we have E = f

∗

(E

n

) and w

i

(E) = w

i

(f

∗

(E

n

)) = f

∗

(w

i

(E

n

)).

Thus if f induces the trivial map on Z

2

cohomology, as the Stiefel-Whitney classes

of E are trivial. The converse statement is also true because the classes w

i

(E

n

) gen-

erate the cohomology ring H

∗

(G

n

; Z

2

). In fact H

∗

(G

n

; Z

2

) is exactly the polynomial

ring Z

2

[w

1

(E

n

), ··· , w

n

(E

n

)]. This is one reason for considering Z

2

rather than Z

coefﬁcients, since H

∗

(G

n

; Z) has a more complicated ring structure, and the Pontrya-

gin classes p

i

(E

n

) are needed in order to describe it completely. Even though this

description is not so nice, it is nevertheless true that f induces the trivial map on co-

74 Chapter 3 Characteristic Classes

homology with arbitrary coefﬁcients iff all the Stiefel-Whitney and Pontryagin classes

of E are zero. If E is orientable we can take its classifying map to be a map to

`

G

n

,

and the corresponding statement here is that this map induces the trivial map on

cohomology with arbitrary coefﬁcients iff the Euler class of E as well as all its Stiefel-

Whitney and Pontryagin classes are trivial. For complex vector bundles the analogous

statements are also true but somewhat simpler since one only needs Chern classes,

as H

∗

(G

n

(C); Z) is the polynomial ring Z[c

1

, ··· , c

n

] on the Chern classes c

i

of the

universal bundle over G

n

(C).

The vanishing of all the characteristic classes of a vector bundle is a necessary

condition for it to be trivial, but it is not always sufﬁcient, as there exist nontrivial

vector bundles whose characteristic classes are all zero. Perhaps the simplest example

is the tangent bundle of the sphere S

5

. One reason why characteristic classes are not

sufﬁcient to determine when a vector bundle is trivial is that, except for the Euler

class, they are stable invariants, meaning that taking the direct sum of a given bundle

with a trivial bundle does not change the characteristic classes. Thus for a vector

bundle which is stably trivial, such as the tangent bundle of a sphere, only the Euler

class has a chance of detecting nontriviality of the bundle. A key feature of the Euler

class is that it is zero for vector bundles that split as the sum of a trivial bundle with

some other bundle. Tangent bundles to odd-dimensional spheres have this property

since they have nonzero sections, splitting off a trivial line bundle. Thus these tangent

bundles have no nonzero characteristic classes, but only S

1

, S

3

, and S

7

have trivial

tangent bundles, as we showed in the previous chapter.

What Exactly Do Characteristic Classes Measure?

Let us see how one might attack the question of what makes a vector bundle non-

trivial. The ﬁrst obstacle to triviality is orientability. For a vector bundle E→B with

B path-connected, orientability is detected by the homomorphism π

1

(B)→Z

2

that

assigns 0 or 1 to each loop according to whether orientations of ﬁbers are preserved

or reversed as one goes around the loop. Since Z

2

is abelian, this homomorphism

factors through the abelianization H

1

(B) of π

1

(B), and homomorphisms H

1

(B)→Z

2

are identiﬁable with elements of H

1

(B; Z

2

). Thus we have an element of H

1

(B; Z

2

)

associated to E which is zero exactly when E is orientable. This is exactly the ﬁrst

Stiefel-Whitney class w

1

(E).

Let us assume now that B is a CW complex. Then E is orientable iff its restriction

to the 1 skeleton B

1

of B is orientable, since all loops in B can be deformed to lie

in B

1

. Furthermore, a vector bundle over a 1 dimensional CW complex is trivial iff

it is orientable, so we can also say that w

1

(E) measures whether E is trivial over the

1 skeleton B

1

.

Having an understanding of when E is trivial over the 1 skeleton we can ask about

triviality over each successive higher skeleton. Assuming that E is trivial over B

1

, then

if the ﬁbers have dimension n we can choose n orthonormal sections over B

1

, and

Characteristic Classes 75

so we ask whether these sections extend to orthonormal sections over each 2 cell. If

we pull E back to a vector bundle over the disk D

2

via a characteristic map D

2

→B

for a 2 cell, the sections over B

1

pull back to sections of the pullback over ∂D

2

. The

pullback bundle is trivial since D

2

is contractible, so by choosing a trivialization the

sections over ∂D

2

determine a map ∂D

2

→O(n). If we choose the trivialization to

give the same orientation as the trivialization of E over B

1

determined by the sections,

we can take this map ∂D

2

→O(n) to have image in the identity component SO(n) of

O(n). Thus we have an element of π

1

(SO(n)) for each 2 cell, and it is not hard to

see that the sections over B

1

extend to orthonormal sections over B

2

iff this element

of π

1

(SO(n)) is zero for each 2 cell. The group π

1

(SO(n)) is obviously 0 for n = 1,

and it is Z for n = 2 since SO(2) is homeomorphic to S

1

. For n ≥ 3 it is not hard to

compute that π

1

(SO(n)) is Z

2

, with the obvious inclusion SO(2)

SO(n) inducing

a surjection on π

1

, so a generator of π

1

(SO(n)) for n ≥ 2 is represented by the loop

given by rotating a plane through 360 degrees, ﬁxing the orthogonal complement of

this plane. We can regard the function which assigns to each 2 cell the element of

π

1

(SO(n)) which is the obstruction to extending the sections over the 2 cell as a

2 dimensional cellular cochain with coefﬁcients in π

1

(SO(n)). It turns out to be a

cellular cocycle, and so deﬁnes an element of H

2

(B; π

1

(SO(n))). This element, call

it ω

2

(E), does not depend on the original choice of the sections over B

1

, and by

rechoosing these sections if necessary one can see that E is trivial over B

2

iff ω

2

(E)

is zero in H

2

(B; π

1

(SO(n))). When n = 1 we have ω

2

(E) = 0 automatically, so E is

trivial over B

2

in this case. When n = 2 the class ω

2

(E) lies in H

2

(B; Z) and equals

the Euler class e(E). For n ≥ 3 we have ω

2

(E) lying in H

2

(B; Z

2

) and it equals the

Stiefel-Whitney class w

2

(E). (When n = 2, w

2

(E) is the image of e(E) under the

homomorphism H

2

(B; Z)→H

2

(B; Z

2

) reducing coefﬁcients mod 2.)

If E is trivial over B

2

we can try to extend n orthonormal sections from B

2

to

B

3

in the same way, cell by cell. For each cell the obstruction to extending lies in

π

2

(SO(n)). This group happens to be zero for all n, so the sections automatically

extend over B

3

. At the next step we seek to extend over 4 cells, where the obstructions

lie in π

3

(SO(n)). These groups can also be computed explicitly with only a little more

effort than for π

1

and π

2

, and they are 0 for n ≤ 2, Z for n = 3, Z⊕Z for n = 4,

and Z again for n ≥ 5. By reasoning analogous to that for 2 cells we then obtain an

obstruction ω

4

(E) ∈ H

4

(B; π

3

(SO(n))). In the cases when the coefﬁcient group is

Z this turns out to be the Pontryagin class p

1

(E). In the special case n = 4 when

π

3

(SO(4)) is Z⊕Z the class ω

4

(E) can be viewed as a pair of elements of H

4

(B; Z),

and these are p

1

(E) and e(E).

Unfortunately things become much more complicated as one proceeds to higher

skeleta B

k

. The ﬁrst problem is that the groups π

k−1

(SO(n)) become increasingly

difﬁcult to compute. They are ﬁnitely generated abelian groups whose nontorsion

is known and fairly easy to describe, but the torsion is quite a mess. One can sim-

76 Chapter 3 Characteristic Classes

plify by restricting to the stable range n > k where Bott periodicity gives the val-

ues of π

k−1

(SO(n)). For k = 4i we have π

4i−1

(SO(n)) = Z, and the Pontrya-

gin class p

i

(E) ∈ H

4i

(B; Z) detects the nontriviality of bundles with obstructions

ω

4i

(E) ∈ H

4i

(B; π

4i−1

(SO(n))) for n > 4i. The other nontrivial stable groups are

π

k−1

(SO(n)) = Z

2

for k = 8i+1, 8i+2, but the obstructions ω

k

(E) in these cases are

not detected by Stiefel-Whitney or Pontryagin classes except when i = 0. Pontryagin

classes also detect the nontorsion in π

k−1

(SO(n)) outside the stable range.

Another drawback to this approach is that it only associates a single cohomol-

ogy class ω

k

(E) to each E, the ﬁrst obstruction to triviality. For example if E is

nonorientable, we get no further characteristic classes associated to E.

A more subtle way of measuring triviality of an n dimensional vector bundle

E→B is to look for obstructions to the existence of fewer than n orthonormal sec-

tions. Suppose for example that we ask whether there is a single section of unit

vectors. As before we proceed by induction over skeleta of B. Assuming we have a

section over B

k−1

, consider the problem of extending it over a k cell. Pulling E back

to the trivial bundle over D

k

via a characteristic map for the cell, we have a section

over ∂D

k

, or equivalently a map ∂D

k

→S

n−1

, and the section over B

k−1

extends over

the k cell iff this map is nullhomotic. Thus if k < n such an extension is always

possible, so the ﬁrst possibility of a nontrivial obstruction occurs when k = n, where

π

k−1

(S

n−1

) = Z. Thus it appears that we get an obstruction ω(E) ∈ H

n

(B; Z) but

there is a technical point that in order to identify the groups π

n−1

(S

n−1

) with Z in a

consistent way it is necessary to assume E is orientable. If we choose an orientation

there is then a well-deﬁned ﬁrst obstruction to a section of unit vectors, and this is

exactly the Euler class e(E). If one only considers the mod 2 value of the element of

π

n−1

(S

n−1

) associated to each n cell then it is not necessary to assume E is orientable

and one obtains a well-deﬁned obstruction in H

n

(B; Z

2

) which is the Stiefel-Whitney

class w

n

(E).

More generally suppose we look for k orthonormal sections. Then the unit sphere

S

n−1

in the ﬁbers of E is replaced by the Stiefel manifold V

k

(R

n

) of orthonormal

k tuples of vectors in R

n

. As we will see, the ﬁrst nonvanishing homotopy group

of V

k

(R

n

) is π

n−k

. This means that there are no obtructions to ﬁnding k orthonor-

mal sections over the (n − k) skeleton of B. Thus we can ﬁnd n sections over the

0 skeleton (this is obvious), n−1 sections over the 1 skeleton, n−2 sections over the

2 skeleton, and so on. The ﬁrst obstruction to ﬁnding k orthonormal sections would

lie in H

n−k+1

(B; π

n−k

(V

k

(R

n

))). This coefﬁcient group is either Z or Z

2

, depending

on what n and k are. In case it is Z we again have the technical problemof not having

a canonical identiﬁcation with Z, so to avoid this problem we reduce the obstruction

mod 2. Then we get a well-deﬁned obstruction in H

n−k+1

(B; Z

2

) to the existence of

k orthonormal sections. This is the Stiefel-Whitney class w

n−k+1

(E).

Complex vector bundles can be treated in a similar fashion. For the complex

Stiefel-Whitney and Chern Classes Section 3.1 77

Stiefel manifold V

k

(C

n

) the ﬁrst nonvanishing homotopy group is π

2n−2k+1

, which is

Z in all cases, and there is a canonical identiﬁcation with Z, so we obtain an obstruction

in H

2n−2k+2

(B; Z) to the existence of k orthonormal sections in a complex vector

bundle E of dimension n. This is the Chern class c

n−k+1

(E), up to a sign at least.

The easiest way to deﬁne Pontryagin classes is in terms of Chern classes rather

than as obstructions, so this is what we will do later in the chapter after Stiefel-Whitney

and Chern classes have been introduced.

3.1. Stiefel-Whitney and Chern Classes

Stiefel-Whitney classes are deﬁned for real vector bundles, Chern classes for com-

plex vector bundles. The two cases are quite similar, but for concreteness we will em-

phasize the real case, with occasional comments on the minor modiﬁcations needed

to treat the complex case.

The deﬁnition of Stiefel-Whitney and Chern classes outlined in the introduction

above, while conceptually appealing, is somewhat awkward to work with when one

tries to prove certain general facts or calculate examples beyond the simplest cases.

For this reason we will start with a different deﬁnition that avoids these drawbacks

and is quite simple and elegant in its own way, although it relies on a little more

algebraic topology, and then afterwards we will show that the two deﬁnitions give the

same result.

A technical point before we begin: We will assume without further mention that

all base spaces of vector bundles are paracompact, so that the fundamental results

of Chapter 1 apply. For the study of characteristic classes this is not an essential

restriction since one can always pass to pullbacks over a CW approximation to a given

base space, and CW complexes are paracompact.

Axioms and Construction

Here is the basic result about Stiefel-Whitney classes giving their most essential

properties, which can be regarded as axioms:

Theorem 3.1. There is a unique sequence of functions w

1

, w

2

, ··· assigning to each

real vector bundle E→B a class w

i

(E) ∈ H

i

(B; Z

2

), depending only on the isomor-

phism type of E, such that

(a) w

i

(f

∗

(E)) = f

∗

(w

i

(E)) for a pullback f

∗

(E).

(b) w(E

1

⊕E

2

) = w(E

1

) w(E

2

) where w = 1 +w

1

+w

2

+··· ∈ H

∗

(B; Z

2

).

(c) w

i

(E) = 0 if i > dimE.

(d) For the canonical line bundle E→RP

∞

, w

1

(E) is a generator of H

1

(RP

∞

; Z

2

).

The sum w(E) = 1+w

1

(E)+w

2

(E)+··· is the total Stiefel-Whitney class. Note that

(c) implies that the sum 1+w

1

(E)+w

2

(E)+··· has only ﬁnitely many nonzero terms,

78 Chapter 3 Characteristic Classes

so this sumdoes indeed lie in H

∗

(B; Z

2

), the direct sumof the groups H

i

(B; Z

2

). From

the formal identity

(1 +w

1

+w

2

+···)(1 +w

¹

1

+w

¹

2

+···) = 1 +(w

1

+w

¹

1

) +(w

2

+w

1

w

¹

1

+w

¹

2

) +···

it follows that the formula w(E

1

⊕E

2

) = w(E

1

) w(E

2

) is just a compact way of

writing the relations w

n

(E

1

⊕E

2

) =

¸

i+j=n

w

i

(E

1

) w

j

(E

2

), where w

0

= 1. This

relation is sometimes called the Whitney sum formula.

Property (d) can be viewed as a nontriviality condition. If this were dropped we

could set w

i

(E) = 0 for all E and all i > 0, and the ﬁrst three conditions would be

satisﬁed.

For complex vector bundles there are analogous Chern classes:

Theorem 3.2. There is a unique sequence of functions c

1

, c

2

, ··· assigning to each

complex vector bundle E→B a class c

i

(E) ∈ H

2i

(B; Z), depending only on the iso-

morphism type of E, such that

(a) c

i

(f

∗

(E)) = f

∗

(c

i

(E)) for a pullback f

∗

(E).

(b) c(E

1

⊕E

2

) = c(E

1

) c(E

2

) for c = 1 +c

1

+c

2

+··· ∈ H

∗

(B; Z).

(c) c

i

(E) = 0 if i > dimE.

(d) For the canonical line bundle E→CP

∞

, c

1

(E) is a generator of H

2

(CP

∞

; Z) spec-

iﬁed in advance.

As in the real case, the formula in (b) for the total Chern classes can be rewritten

in the form c

n

(E

1

⊕E

2

) =

¸

i+j=n

c

i

(E

1

) c

j

(E

2

), where c

0

= 1. The condition (d)

is now a normalization condition as well as a nontriviality condition, since replacing

each function c

i

by k

i

c

i

for a ﬁxed integer k gives new functions satisfying (a)-(c).

The proof of both theorems will be based on the Leray-Hirsch theorem describing

the cohomology of certain ﬁber bundles. If p: E→B is a ﬁber bundle then we can

use cup products to make H

∗

(E; R) into a module over the ring H

∗

(B; R), for R

any commutative ring with identity, by setting αβ = p

∗

(α) β for α ∈ H

∗

(B) and

β ∈ H

∗

(E). Then the Leray-Hirsch theorem asserts that this module H

∗

(E; R) is a

free H

∗

(B; R) module provided that for each ﬁber F the inclusion F

E induces a

surjection on H

∗

(−−; R) and H

n

(F; R) is a free R module of ﬁnite rank for each n.

Moreover, a basis for the H

∗

(B; R) module H

∗

(E; R) can be obtained by choosing any

set of elements in H

∗

(E; R) that map to a basis for H

∗

(F; R) under the map induced

by the inclusion. A proof of this theorem using a fairly straightforward induction

argument can be found in §4.D of [AT].

Proof of 3.1 and 3.2: Associated to a vector bundle π : E→B with ﬁber R

n

is the

projective bundle P(π) : P(E)→B, where P(E) is the space of all lines through the

origin in all the ﬁbers of E, and P(π) is the natural projection sending each line in

π

−1

(b) to b ∈ B. We topologize P(E) as a quotient of the complement of the zero

section of E, the quotient obtained by factoring out scalar multiplication in each ﬁber.

Stiefel-Whitney and Chern Classes Section 3.1 79

Over a neighborhood U in B where E is a product U×R

n

, this quotient is U×RP

n−1

,

so P(E) is a ﬁber bundle over B with ﬁber RP

n−1

.

We would like to apply the Leray-Hirsch theorem for cohomology with Z

2

co-

efﬁcients to this bundle P(E)→B. To do this we need classes x

i

∈ H

i

(P(E); Z

2

)

restricting to generators of H

i

(RP

n−1

; Z

2

) in each ﬁber RP

n−1

for i = 0, ··· , n − 1.

Recall from the proof of Theorem 1.16 that there is a map g : E→R

∞

that is a linear

injection on each ﬁber. Projectivizing the map g by deleting zero vectors and then

factoring out scalar multiplication produces a map P(g) : P(E)→RP

∞

. Let α be a gen-

erator of H

1

(RP

∞

; Z

2

) and let x = P(g)

∗

(α) ∈ H

1

(P(E); Z

2

). Then the powers x

i

for

i = 0, ··· , n − 1 are the desired classes x

i

since a linear injection R

n

→R

∞

induces

an embedding RP

n−1

RP

∞

for which α pulls back to a generator of H

1

(RP

n−1

; Z

2

),

hence α

i

pulls back to a generator of H

i

(RP

n−1

; Z

2

). Note that any two linear injec-

tions R

n

→R

∞

are homotopic through linear injections, so the induced embeddings

RP

n−1

RP

∞

of different ﬁbers of P(E) are all homotopic. We showed in the proof of

Theorem 1.16 that any two choices of g are homotopic through maps that are linear

injections on ﬁbers, so the classes x

i

are independent of the choice of g.

The Leray-Hirsch theoremthen says that H

∗

(P(E); Z

2

) is a free H

∗

(B; Z

2

) module

with basis 1, x, ··· , x

n−1

. Consequently, x

n

can be expressed uniquely as a linear

combination of these basis elements with coefﬁcients in H

∗

(B; Z

2

). Thus there is a

unique relation of the form

x

n

+w

1

(E)x

n−1

+··· +w

n

(E) · 1 = 0

for certain classes w

i

(E) ∈ H

i

(B; Z

2

). Here w

i

(E)x

i

means P(π)

∗

(w

i

(E)) x

i

, by

the deﬁnition of the H

∗

(B; Z

2

) module structure on H

∗

(P(E); Z

2

). For completeness

we deﬁne w

i

(E) = 0 for i > n and w

0

(E) = 1.

To prove property (a), consider a pullback f

∗

(E) = E

¹

, ﬁtting

into the diagram at the right. If g : E→R

∞

is a linear injection on

−

−

→

−

−

→

−−−−−→

E E

−−−−−→

B B

f

f

∼

π π

ﬁbers then so is g

`

f , and it follows that P(

`

f)

∗

takes the canonical

class x = x(E) for P(E) to the canonical class x(E

¹

) for P(E

¹

). Then

P(

`

f)

∗

¸

i

P(π)

∗

¸

w

i

(E)

x(E)

n−i

=

¸

i

P(

`

f)

∗

P(π)

∗

¸

w

i

(E)

P(

`

f)

∗

¸

x(E)

n−i

=

¸

i

P(π

¹

)

∗

f

∗

¸

w

i

(E)

x(E

¹

)

n−i

so the relation x(E)

n

+ w

1

(E)x(E)

n−1

+ ··· + w

n

(E) · 1 = 0 deﬁning w

i

(E) pulls

back to the relation x(E

¹

)

n

+f

∗

(w

1

(E))x(E

¹

)

n−1

+··· +f

∗

(w

n

(E)) · 1 = 0 deﬁning

w

i

(E

¹

). By the uniqueness of this relation, w

i

(E

¹

) = f

∗

(w

i

(E)).

Proceeding to property (b), the inclusions of E

1

and E

2

into E

1

⊕E

2

give in-

clusions of P(E

1

) and P(E

2

) into P(E

1

⊕E

2

) with P(E

1

) ∩ P(E

2

) = ∅. Let U

1

=

P(E

1

⊕E

2

) − P(E

1

) and U

2

= P(E

1

⊕E

2

) − P(E

2

). These are open sets in P(E

1

⊕E

2

)

that deformation retract onto P(E

2

) and P(E

1

), respectively. A map g : E

1

⊕E

2

→R

∞

80 Chapter 3 Characteristic Classes

which is a linear injection on ﬁbers restricts to such a map on E

1

and E

2

, so the

canonical class x ∈ H

1

(P(E

1

⊕E

2

); Z

2

) for E

1

⊕E

2

restricts to the canonical classes

for E

1

and E

2

. If E

1

and E

2

have dimensions m and n, consider the classes ω

1

=

¸

j

w

j

(E

1

)x

m−j

and ω

2

=

¸

j

w

j

(E

2

)x

n−j

in H

∗

(P(E

1

⊕E

2

); Z

2

), with cup product

ω

1

ω

2

=

¸

j

¸¸

r+s=j

w

r

(E

1

)w

s

(E

2

)

¸

x

m+n−j

. By the deﬁnition of the classes w

j

(E

1

),

the class ω

1

restricts to zero in H

m

(P(E

1

); Z

2

), hence ω

1

pulls back to a class in

the relative group H

m

(P(E

1

⊕E

2

), P(E

1

); Z

2

) ≈ H

m

(P(E

1

⊕E

2

), U

2

; Z

2

), and similarly

for ω

2

. The following commutative diagram, with Z

2

coefﬁcients understood, then

shows that ω

1

ω

2

= 0:

−

−

−

−

−

→

−

−

−

−

−

→

−−−−−→

U E P H

n

1

E

2 1

( ( ) ), ⊕ U E P H

m n

1

E

2 1

U 0

2

( ( ) ), ⊕ U E P H

m

1

E

2 2

( ( ) ), ⊕ × =

+

∪

−−−−−−−−−−−−−−−→

E P H

n

1

E

2

( ( ) ) ⊕ E P H

m n

1

E

2

( ( ) ) ⊕ E P H

m

1

E

2

( ( ) ) ⊕ ×

+

Thus ω

1

ω

2

=

¸

j

¸¸

r+s=j

w

r

(E

1

)w

s

(E

2

)

¸

x

m+n−j

= 0 is the deﬁning relation for the

Stiefel-Whitney classes of E

1

⊕E

2

, and so w

j

(E

1

⊕E

2

) =

¸

r+s=j

w

r

(E

1

)w

s

(E

2

).

Property (c) holds by deﬁnition. For (d), recall that the canonical line bundle is

E = ¦ (, v) ∈ RP

∞

×R

∞

¦ v ∈ ¦. The map P(π) in this case is the identity. The

map g : E→R

∞

which is a linear injection on ﬁbers can be taken to be g(, v) = v.

So P(g) is also the identity, hence x(E) is a generator of H

1

(RP

∞

; Z

2

). The deﬁning

relation x(E) +w

1

(E) · 1 = 0 then says that w

1

(E) is a generator of H

1

(RP

∞

; Z

2

).

The proof of uniqueness of the classes w

i

will use a general property of vector

bundles called the splitting principle:

Proposition 3.3. For each vector bundle π : E→B there is a space F(E) and a map

p: F(E)→B such that the pullback p

∗

(E)→F(E) splits as a direct sum of line bun-

dles, and p

∗

: H

∗

(B; Z

2

)→H

∗

(F(E); Z

2

) is injective.

Proof: Consider the pullback P(π)

∗

(E) of E via the map P(π) : P(E)→B. This pull-

back contains a natural one-dimensional subbundle L = ¦ (, v) ∈ P(E)×E ¦ v ∈ ¦.

An inner product on E pulls back to an inner product on the pullback bundle, so we

have a splitting of the pullback as a sum L⊕L

⊥

with the orthogonal bundle L

⊥

hav-

ing dimension one less than E. As we have seen, the Leray-Hirsch theorem applies

to P(E)→B, so H

∗

(P(E); Z

2

) is the free H

∗

(B; Z

2

) module with basis 1, x, ··· , x

n−1

and in particular the induced map H

∗

(B; Z

2

)→H

∗

(P(E); Z

2

) is injective since one of

the basis elements is 1.

This construction can be repeated with L

⊥

→P(E) in place of E→B. After ﬁnitely

many repetitions we obtain the desired result. L¦

Looking at this construction a little more closely, L

⊥

consists of pairs (, v) ∈

P(E)×E with v ⊥ . At the next stage we form P(L

⊥

), whose points are pairs (,

¹

)

where and

¹

are orthogonal lines in E. Continuing in this way, we see that the

Stiefel-Whitney and Chern Classes Section 3.1 81

ﬁnal base space F(E) is the space of all orthogonal splittings

1

⊕··· ⊕

n

of ﬁbers

of E as sums of lines, and the vector bundle over F(E) consists of all n tuples of

vectors in these lines. Alternatively, F(E) can be described as the space of all chains

V

1

⊂ ··· ⊂ V

n

of linear subspaces of ﬁbers of E with dimV

i

= i. Such chains are

called ﬂags, and F(E)→B is the ﬂag bundle associated to E. Note that the description

of points of F(E) as ﬂags does not depend on a choice of inner product in E.

Now we can ﬁnish the proof of Theorem 3.1. Property (d) determines w

1

(E)

for the canonical line bundle E→RP

∞

. Property (c) then determines all the w

i

’s for

this bundle. Since the canonical line bundle is the universal line bundle, property (a)

therefore determines the classes w

i

for all line bundles. Property (b) extends this

to sums of line bundles, and ﬁnally the splitting principle implies that the w

i

’s are

determined for all bundles.

For complex vector bundles we can use the same proof, but with Z coefﬁcients

since H

∗

(CP

∞

; Z) ≈ Z[α], with α now two-dimensional. The deﬁning relation for the

c

i

(E)’s is modiﬁed to be

x

n

−c

1

(E)x

n−1

+··· +(−1)

n

c

n

(E) · 1 = 0

with alternating signs. This is equivalent to changing the sign of α, so it does not

affect the proofs of properties (a)–(c), but it has the advantage that the canonical line

bundle E→CP

∞

has c

1

(E) = α rather than −α, since the deﬁning relation in this

case is x(E) −c

1

(E) · 1 = 0 and x(E) = α. L¦

Note that in property (d) for Stiefel-Whitney classes we could just as well use the

canonical line bundle over RP

1

instead of RP

∞

since the inclusion RP

1

RP

∞

induces

an isomorphism H

1

(RP

∞

; Z

2

) ≈ H

1

(RP

1

; Z

2

). The analogous remark for Chern classes

is valid as well.

Example 3.4. Property (a), the naturality of Stiefel-Whitney classes, implies that a

product bundle E = B×R

n

has w

i

(E) = 0 for i > 0 since a product is the pullback

of a bundle over a point, which must have w

i

= 0 for i > 0 since a point has trivial

cohomology in positive dimensions.

Example 3.5: Stability. Property (b) implies that taking the direct sum of a bundle

with a product bundle does not change its Stiefel-Whitney classes. In this sense Stiefel-

Whitney classes are stable. For example, the tangent bundle TS

n

to S

n

is stably

trivial since its direct sum with the normal bundle to S

n

in R

n+1

, which is a trivial

line bundle, produces a trivial bundle. Hence the Stiefel-Whitney classes w

i

(TS

n

) are

zero for i > 0.

From the identity

(1 +w

1

+w

2

+···)(1 +w

¹

1

+w

¹

2

+···) = 1 +(w

1

+w

¹

1

) +(w

2

+w

1

w

¹

1

+w

¹

2

) +···

82 Chapter 3 Characteristic Classes

we see that w(E

1

) and w(E

1

⊕E

2

) determine w(E

2

) since the equations

w

1

+w

¹

1

= a

1

w

2

+w

1

w

¹

1

+w

¹

2

= a

2

···

¸

i

w

n−i

w

¹

i

= a

n

can be solved successively for the w

¹

i

’s in terms of the w

i

’s and a

i

’s. In particular, if

E

1

⊕E

2

is the trivial bundle, then we have the case that a

i

= 0 for i > 0 and so w(E

1

)

determines w(E

2

) uniquely by explicit formulas that can be worked out. For example,

w

¹

1

= −w

1

and w

¹

2

= −w

1

w

¹

1

−w

2

= w

2

1

−w

2

. Of course for Z

2

coefﬁcients the signs

do not matter, but the same reasoning applies to Chern classes, with Z coefﬁcients.

Example 3.6. Let us illustrate this principle by showing that there is no bundle

E→RP

∞

whose sum with the canonical line bundle E

1

(R

∞

) is trivial. For we have

w(E

1

(R

∞

)) = 1 + ω where ω is a generator of H

1

(RP

∞

; Z

2

), and hence w(E) must

be (1+ω)

−1

= 1+ω+ω

2

+··· since we are using Z

2

coefﬁcients. Thus w

i

(E) = ω

i

,

which is nonzero in H

∗

(RP

∞

; Z

2

) for all i. However, this contradicts the fact that

w

i

(E) = 0 for i > dimE.

This shows the necessity of the compactness assumption in Proposition 1.4. To

further delineate the question, note that Proposition 1.4 says that the restriction

E

1

(R

n+1

) of the canonical line bundle to the subspace RP

n

⊂ RP

∞

does have an

‘inverse’ bundle. In fact, the bundle E

⊥

1

(R

n+1

) consisting of pairs (, v) where

is a line through the origin in R

n+1

and v is a vector orthogonal to is such an

inverse. But for any bundle E→RP

n

whose sum with E

1

(R

n+1

) is trivial we must

have w(E) = 1 + ω + ··· + ω

n

, and since w

n

(E) = ω

n

≠ 0, E must be at least

n dimensional. So we see there is no chance of choosing such bundles E for varying

n so that they ﬁt together to form a single bundle over RP

∞

.

Example 3.7. Let us describe an n dimensional vector bundle E→B with w

i

(E)

nonzero for each i ≤ n. This will be the n fold Cartesian product (E

1

)

n

→(G

1

)

n

of

the canonical line bundle over G

1

= RP

∞

with itself. This vector bundle is the direct

sum π

∗

1

(E

1

)⊕ ··· ⊕π

∗

n

(E

1

) where π

i

: (G

1

)

n

→G

1

is projection onto the i

th

factor, so

w((E

1

)

n

) =

¸

i

(1 +α

i

) ∈ Z

2

[α

1

, ··· , α

n

] ≈ H

∗

((RP

∞

)

n

; Z

2

). Hence w

i

((E

1

)

n

) is the

i

th

elementary symmetric polynomial σ

i

in the α

j

’s, the sum of all the products of i

different α

j

’s. For example, if n = 3 then σ

1

= α

1

+α

2

+α

3

, σ

2

= α

1

α

2

+α

1

α

3

+α

2

α

3

,

and σ

3

= α

1

α

2

α

3

. Since each σ

i

with i ≤ n is nonzero in Z

2

[α

1

, ··· , α

n

], we have

an n dimensional bundle whose ﬁrst n Stiefel-Whitney classes are all nonzero.

The same reasoning applies in the complex case to showthat the n fold Cartesian

product of the canonical line bundle over CP

∞

has its ﬁrst n Chern classes nonzero.

In this example we see that the w

i

’s and c

i

’s can be identiﬁed with elementary

symmetric functions, and in fact this can be done in general using the splitting princi-

ple. Given an n dimensional vector bundle E→B we know that the pullback to F(E)

Stiefel-Whitney and Chern Classes Section 3.1 83

splits as a sum L

1

⊕ ··· ⊕L

n

→F(E). Letting α

i

= w

1

(L

i

), we see that w(E) pulls

back to w(L

1

⊕ ··· ⊕L

n

) = (1 +α

1

) ··· (1 +α

n

) = 1 +σ

1

+··· +σ

n

, so w

i

(E) pulls

back to σ

i

. Thus we have embedded H

∗

(B; Z

2

) in a larger ring H

∗

(F(E); Z

2

) such that

w

i

(E) becomes the i

th

elementary symmetric polynomial in the elements α

1

, ··· , α

n

of H

∗

(F(E); Z

2

).

Besides the evident formal similarity between Stiefel-Whitney and Chern classes

there is also a direct relation:

Proposition 3.8. Regarding an n dimensional complex vector bundle E→B as a

2n dimensional real vector bundle, then w

2i+1

(E) = 0 and w

2i

(E) is the image of

c

i

(E) under the coefﬁcient homomorphism H

2i

(B; Z)→H

2i

(B; Z

2

).

For example, since the canonical complex line bundle over CP

∞

has c

1

a generator

of H

2

(CP

∞

; Z), the same is true for its restriction over S

2

= CP

1

, so by the proposition

this 2 dimensional real vector bundle E→S

2

has w

2

(E) ≠ 0.

Proof: The bundle E has two projectivizations RP(E) and CP(E), consisting of all the

real and all the complex lines in ﬁbers of E, respectively. There is a natural projection

p: RP(E)→CP(E) sending each real line to the complex line containing it, since a real

line is all the real scalar multiples of any nonzero vector in it and a complex line is all

the complex scalar multiples. This projection p ﬁts into a commutative diagram

−

−

−

−

−

→

−

−

−

−

−

→

−

−

−

−

−

→

−−−−−−−−−−→ −−−−−−−−−→

E P P

2

( ) R R

g

p

P( ) R

P R

n 1 ∞

−−−−−−−−−−→ −−−−−−−−−→

E P P ( ) C C

g P( ) C

P C

n 1 ∞

where the left column is the restriction of p to a ﬁber of E and the maps RP(g)

and CP(g) are obtained by projectivizing, over R and C, a map g : E→C

∞

which

is a C linear injection on ﬁbers. It is easy to see that all three vertical maps in

this diagram are ﬁber bundles with ﬁber RP

1

, the real lines in a complex line. The

Leray-Hirsch theorem applies to the bundle RP

∞

→CP

∞

, with Z

2

coefﬁcients, so if

β is the standard generator of H

2

(CP

∞

; Z), the Z

2

reduction β ∈ H

2

(CP

∞

; Z

2

) pulls

back to a generator of H

2

(RP

∞

; Z

2

), namely the square α

2

of the generator α ∈

H

1

(RP

∞

; Z

2

). Hence the Z

2

reduction x

C

(E) = CP(g)

∗

(β) ∈ H

2

(CP(E); Z

2

) of the

basic class x

C

(E) = CP(g)

∗

(β) pulls back to the square of the basic class x

R

(E) =

RP(g)

∗

(α) ∈ H

1

(RP(E); Z

2

). Consequently the Z

2

reduction of the deﬁning relation

for the Chern classes of E, which is x

C

(E)

n

+ c

1

(E)x

C

(E)

n−1

+ ··· + c

n

(E) · 1 = 0,

pulls back to the relation x

R

(E)

2n

+c

1

(E)x

R

(E)

2n−2

+··· +c

n

(E)· 1 = 0, which is the

deﬁning relation for the Stiefel-Whitney classes of E. This means that w

2i+1

(E) = 0

and w

2i

(E) = c

i

(E). L¦

84 Chapter 3 Characteristic Classes

Cohomology of Grassmannians

From Example 3.7 and naturality it follows that the universal bundle E

n

→G

n

must also have all its Stiefel-Whitney classes w

1

(E

n

), ··· , w

n

(E

n

) nonzero. In fact a

much stronger statement is true. Let f : (RP

∞

)

n

→G

n

be the classifying map for the

n fold Cartesion product (E

1

)

n

of the canonical line bundle E

1

, and for notational

simplicity let w

i

= w

i

(E

n

). Then the composition

Z

2

[w

1

, ··· , w

n

] →H

∗

(G

n

; Z

2

)

f

∗

→H

∗

¸

(RP

∞

)

n

; Z

2

≈ Z

2

[α

1

, ··· , α

n

]

sends w

i

to σ

i

, the i

th

elementary symmetric polynomial. It is a classical algebraic re-

sult that the polynomials σ

i

are algebraically independent in Z

2

[α

1

, ··· , α

n

]. Proofs

of this can be found in [van der Waerden, §26] or [Lang, p. 134] for example. Thus

the composition Z

2

[w

1

, ··· , w

n

]→Z

2

[α

1

, ··· , α

n

] is injective, hence also the map

Z

2

[w

1

, ··· , w

n

]→H

∗

(G

n

; Z

2

). In other words, the classes w

i

(E

n

) generate a poly-

nomial subalgebra Z

2

[w

1

, ··· , w

n

] ⊂ H

∗

(G

n

; Z

2

). This subalgebra is in fact equal to

H

∗

(G

n

; Z

2

), and the corresponding statement for Chern classes holds as well:

Theorem 3.9. H

∗

(G

n

; Z

2

) is the polynomial ring Z

2

[w

1

, ··· , w

n

] on the Stiefel-

Whitney classes w

i

= w

i

(E

n

) of the universal bundle E

n

→G

n

. Similarly, in the

complex case H

∗

(G

n

(C

∞

); Z) ≈ Z[c

1

, ··· , c

n

] where c

i

= c

i

(E

n

(C

∞

)) for the univer-

sal bundle E

n

(C

∞

)→G

n

(C

∞

).

The proof we give here for this basic result will be a fairly quick application of the

CW structure on G

n

constructed at the end of §1.2. A different proof will be given

in §3.2 where we also compute the cohomology of G

n

with Z coefﬁcients, which is

somewhat more subtle.

Proof: Consider a map f : (RP

∞

)

n

→G

n

which pulls E

n

back to the bundle (E

1

)

n

considered above. We have noted that the image of f

∗

contains the symmetric poly-

nomials in Z

2

[α

1

, ··· , α

n

] ≈ H

∗

((RP

∞

)

n

; Z

2

). The opposite inclusion holds as well,

since if π : (RP

∞

)

n

→(RP

∞

)

n

is an arbitrary permutation of the factors, then π pulls

(E

1

)

n

back to itself, so fπ = f , which means that f

∗

= π

∗

f

∗

, so the image of f

∗

is

invariant under π

∗

: H

∗

((RP

∞

)

n

; Z

2

)→H

∗

((RP

∞

)

n

; Z

2

), but the latter map is just the

same permutation of the variables α

i

.

To ﬁnish the proof in the real case it remains to see that f

∗

is injective. It sufﬁces

to ﬁnd a CW structure on G

n

in which the r cells are in one-to-one correspondence

with monomials w

r

1

1

··· w

r

n

n

of dimension r = r

1

+2r

2

+··· +nr

n

, since the number

of r cells in a CW complex X is an upper bound on the dimension of H

r

(X; Z

2

) as a

Z

2

vector space, and a surjective linear map between ﬁnite-dimensional vector spaces

is injective if the dimension of the domain is not greater than the dimension of the

range.

Monomials w

r

1

1

··· w

r

n

n

of dimension r correspond to n tuples (r

1

, ··· , r

n

) with

r = r

1

+2r

2

+··· +nr

n

. Such n tuples in turn correspond to partitions of r into at

Stiefel-Whitney and Chern Classes Section 3.1 85

most n integers, via the correspondence

(r

1

, ··· , r

n

) ←→ r

n

≤ r

n

+r

n−1

≤ ··· ≤ r

n

+r

n−1

+··· +r

1

.

Such a partition becomes the sequence σ

1

−1 ≤ σ

2

−2 ≤ ··· ≤ σ

n

−n, corresponding

to the strictly increasing sequence 0 < σ

1

< σ

2

< ··· < σ

n

. For example, when n = 3

we have:

(r

1

, r

2

, r

3

) (σ

1

−1, σ

2

−2, σ

3

−3) (σ

1

, σ

2

, σ

3

) dimension

1 0 0 0 0 0 0 1 2 3 0

w

1

1 0 0 0 0 1 1 2 4 1

w

2

0 1 0 0 1 1 1 3 4 2

w

2

1

2 0 0 0 0 2 1 2 5 2

w

3

0 0 1 1 1 1 2 3 4 3

w

1

w

2

1 1 0 0 1 2 1 3 5 3

w

3

1

3 0 0 0 0 3 1 2 6 3

The cell structure on G

n

constructed in §1.2 has one cell of dimension (σ

1

− 1) +

(σ

2

− 2) + ··· + (σ

n

− n) for each increasing sequence 0 < σ

1

< σ

2

< ··· < σ

n

. So

we are done in the real case.

The complex case is entirely similar, keeping in mind that c

i

has dimension 2i

rather than i. The CWstructure on G

n

(C

∞

) described in §1.2 also has these extra fac-

tors of 2 in the dimensions of its cells. In particular, the cells are all even-dimensional,

so the cellular boundary maps for G

n

(C

∞

) are all trivial and the cohomology with Z

coefﬁcients consists of a Z summand for each cell. Injectivity of f

∗

then follows

from the algebraic fact that a surjective homomorphism between free abelian groups

of ﬁnite rank is injective if the rank of the domain is not greater than the rank of the

range. L¦

One might guess that the monomial w

r

1

1

··· w

r

n

n

corresponding to a given cell of

G

n

in the way described above was the cohomology class dual to this cell, represented

by the cellular cochain assigning the value 1 to the cell and 0 to all the other cells.

This is true for the classes w

i

themselves, but unfortunately it is not true in general.

For example the monomial w

i

1

corresponds to the cell whose associated partition is

the trivial partition i = i, but the cohomology class dual to this cell is w

¹

i

where

1+w

¹

1

+w

¹

2

+··· is the multiplicative inverse of 1+w

1

+w

2

+···. If one replaces the

basis of monomials by the more geometric basis of cohomology classes dual to cells,

the formulas for multiplying these dual classes become rather complicated. In the

parallel situation of Chern classes this question has very classical roots in algebraic

geometry, and the rules for multiplying cohomology classes dual to cells are part of

the so-called Schubert calculus. Accessible expositions of this subject from a modern

viewpoint can be found in [Fulton] and [Hiller].

Applications of w

1

and c

1

We saw in §1.1 that the set Vect

1

(X) of isomorphism classes of line bundles

over X forms a group with respect to tensor product. We know also that Vect

1

(X) =

86 Chapter 3 Characteristic Classes

[X, G

1

(R

∞

)], and G

1

(R

∞

) is just RP

∞

, an Eilenberg-MacLane space K(Z

2

, 1). It is a

basic fact in algebraic topology that [X, K(G, n)] ≈ H

n

(X; G) when X has the homo-

topy type of a CW complex; see Theorem 4.56 of [AT], for example. Thus one might

ask whether the groups Vect

1

(X) and H

1

(X; Z

2

) are isomorphic. For complex line

bundles we have G

1

(C

∞

) = CP

∞

, and this is a K(Z, 2), so the corresponding question

is whether Vect

1

C

(X) is isomorphic to H

2

(X; Z).

Proposition 3.10. The function w

1

: Vect

1

(X)→H

1

(X; Z

2

) is a homomorphism, and

is an isomorphism if X has the homotopy type of a CW complex. The same is also

true for c

1

: Vect

1

C

(X)→H

2

(X; Z).

Proof: The argument is the same in both the real and complex cases, so for def-

initeness let us describe the complex case. To show that c

1

: Vect

1

C

(X)→H

2

(X) is

a homomorphism, we ﬁrst prove that c

1

(L

1

⊗L

2

) = c

1

(L

1

) + c

1

(L

2

) for the bundle

L

1

⊗L

2

→G

1

×G

1

where L

1

and L

2

are the pullbacks of the canonical line bundle

L→G

1

= CP

∞

under the projections p

1

, p

2

: G

1

×G

1

→G

1

onto the two factors. Since

c

1

(L) is the generator α of H

2

(CP

∞

), we know that H

∗

(G

1

×G

1

) ≈ Z[α

1

, α

2

] where

α

i

= p

∗

i

(α) = c

1

(L

i

). The inclusion G

1

∨G

1

⊂ G

1

×G

1

induces an isomorphismon H

2

,

so to compute c

1

(L

1

⊗L

2

) it sufﬁces to restrict to G

1

∨G

1

. Over the ﬁrst G

1

the bundle

L

2

is the trivial line bundle, so the restriction of L

1

⊗L

2

over this G

1

is L

1

⊗1 ≈ L

1

.

Similarly, L

1

⊗L

2

restricts to L

2

over the second G

1

. So c

1

(L

1

⊗L

2

) restricted to

G

1

∨G

1

is α

1

+α

2

restricted to G

1

∨G

1

. Hence c

1

(L

1

⊗L

2

) = α

1

+α

2

= c

1

(L

1

)+c

1

(L

2

).

The general case of the formula c

1

(E

1

⊗E

2

) = c

1

(E

1

) +c

1

(E

2

) for line bundles E

1

and E

2

now follows by naturality: We have E

1

≈ f

∗

1

(L) and E

2

≈ f

∗

2

(L) for maps

f

1

, f

2

: X→G

1

. For the map F = (f

1

, f

2

) : X→G

1

×G

1

we have F

∗

(L

i

) = f

∗

i

(L) ≈ E

i

,

so

c

1

(E

1

⊗E

2

) = c

1

(F

∗

(L

1

)⊗F

∗

(L

2

)) = c

1

(F

∗

(L

1

⊗L

2

)) = F

∗

(c

1

(L

1

⊗L

2

))

= F

∗

(c

1

(L

1

) +c

1

(L

2

)) = F

∗

(c

1

(L

1

)) +F

∗

(c

1

(L

2

))

= c

1

(F

∗

(L

1

)) +c

1

(F

∗

(L

2

)) = c

1

(E

1

) +c

1

(E

2

).

As noted above, if X is a CW complex, there is a bijection [X, CP

∞

] ≈ H

2

(X; Z),

and the more precise statement is that this bijection is given by the map [f]

f

∗

(u)

for some class u ∈ H

2

(CP

∞

; Z). The class u must be a generator, otherwise the map

would not always be surjective. Which of the two generators we choose for u is

not important, so we may take it to be the class α. The map [f]

f

∗

(α) factors

as the composition [X, CP

∞

]→Vect

1

C

(X)→H

2

(X; Z), [f]

f

∗

(L)

c

1

(f

∗

(L)) =

f

∗

(c

1

(L)) = f

∗

(α). The ﬁrst map in this composition is a bijection, so since the

composition is a bijection, the second map c

1

must be a bijection also. L¦

The ﬁrst Stiefel-Whitney class w

1

is closely related to orientability:

Stiefel-Whitney and Chern Classes Section 3.1 87

Proposition 3.11. A vector bundle E→X is orientable iff w

1

(E) = 0, assuming that

X is homotopy equivalent to a CW complex.

Thus w

1

can be viewed as the obstruction to orientability of vector bundles. An

interpretation of the other classes w

i

as obstructions will be given in the Appendix

to this chapter.

Proof: Without loss we may assume X is a CW complex. By restricting to path-

components we may further assume X is connected. There are natural isomorphisms

(∗) H

1

(X; Z

2

)

≈

→Hom(H

1

(X), Z

2

)

≈

→Hom(π

1

(X), Z

2

)

fromthe universal coefﬁcient theoremand the fact that H

1

(X) is the abelianization of

π

1

(X). When X = G

n

we have π

1

(G

n

) ≈ Z

2

, and w

1

(E

n

) ∈ H

1

(G

n

; Z

2

) corresponds

via (∗) to this isomorphism π

1

(G

n

) ≈ Z

2

since w

1

(E

n

) is the unique nontrivial el-

ement of H

1

(G

n

; Z

2

). By naturality of (∗) it follows that for any map f : X→G

n

,

f

∗

(w

1

(E

n

)) corresponds under (∗) to the homomorphism f

∗

: π

1

(X)→π

1

(G

n

) ≈

Z

2

. Thus if we choose f so that f

∗

(E

n

) is a given vector bundle E, we have w

1

(E)

corresponding under (∗) to the induced map f

∗

: π

1

(X)→π

1

(G

n

) ≈ Z

2

. Hence

w

1

(E) = 0 iff this f

∗

is trivial, which is exactly the condition for lifting f to the

universal cover

`

G

n

, i.e., orientability of E. L¦

88 Chapter 3 Characteristic Classes

3.2. Euler and Pontryagin Classes

A characteristic class can be deﬁned to be a function associating to each vector

bundle E→B of dimension n a class x(E) ∈ H

k

(B; G), for some ﬁxed n and k, such

that the naturality property x(f

∗

(E)) = f

∗

(x(E)) is satisﬁed. In particular, for the

universal bundle E

n

→G

n

there is the class x = x(E

n

) ∈ H

k

(G

n

; G). Conversely, any

element x ∈ H

k

(G

n

; G) deﬁnes a characteristic class by the rule x(E) = f

∗

(x) where

E ≈ f

∗

(E

n

) for f : B→G

n

. Since f is unique up to homotopy, x(E) is well-deﬁned,

and it is clear that the naturality property is satisﬁed. Thus characteristic classes

correspond bijectively with cohomology classes of G

n

.

With Z

2

coefﬁcients all characteristic classes are simply polynomials in the Stiefel-

Whitney classes since we showed in Theorem 3.9 that H

∗

(G

n

; Z

2

) is the polyno-

mial ring Z

2

[w

1

, ··· , w

n

]. Similarly for complex vector bundles all characteristic

classes with Z coefﬁcients are polynomials in the Chern classes since H

∗

(G

n

(C); Z) ≈

Z[c

1

, ··· , c

n

]. Our goal in this section is to describe the more reﬁned characteristic

classes for real vector bundles that arise when we take cohomology with integer coef-

ﬁcients rather than Z

2

coefﬁcients.

The main tool we will use will be the Gysin exact sequence associated to an

n dimensional real vector bundle p: E→B. This is an easy consequence of the Thom

isomorphism Φ: H

i

(B)→H

i+n

(D(E), S(E)) deﬁned by Φ(b) = p

∗

(b) c for a Thom

class c ∈ H

n

(D(E), S(E)), a class whose restriction to each ﬁber is a generator of

H

n

(D

n

, S

n−1

). The map Φ is an isomorphismwhenever a Thomclass exists, as shown

in Corollary 4D.9 of [AT]. Thom classes with Z coefﬁcients exist for all orientable real

vector bundles, and with Z

2

coefﬁcients they exist for all vector bundles. This is

shown in Theorem 4D.10 of [AT].

Once one has the Thom isomorphism, this gives the Gysin sequence as the lower

row of the following commutative diagram, whose upper row is the exact sequence

for the pair (D(E), S(E)):

D S E H ( ( ) ) E ( ) E ( ) S E ( ) −−→ , D S E ( ) E ( ) , H ( )

−− − − − −→ −−−−−→ −−−−→

. . . . . .

−

−

→

−

−

→ Φ

−

−

→ Φ

i i

H ( )

−−−−−→

i

S E ( ) H ( )

i

H ( )

i 1

D

j

+

+

B H ( )

−−−→ H ( )

−−−−−−−→ −−−−−→ −−−−→

. . . . . .

i n i

−−−−−→ B

p e

B H ( )

i n 1

∗

∗

p

∗

=

=

≈ ≈ ≈

The vertical map p

∗

is an isomorphism since p is a homotopy equivalence from

D(E) to B. The Euler class e ∈ H

n

(B) is deﬁned to be (p

∗

)

−1

j

∗

(c), or in other words

the restriction of the Thom class to the zero section of E. The square containing

the map e commutes since for b ∈ H

i−n

(B) we have j

∗

Φ(b) = j

∗

(p

∗

(b) c) =

p

∗

(b) j

∗

(c), which equals p

∗

(b e) = p

∗

(b) p

∗

(e) since p

∗

(e) = j

∗

(c). The

Euler class can also be deﬁned as the class corresponding to c c under the Thom

isomorphism, since Φ(e) = p

∗

(e) c = j

∗

(c) c = c c.

As a warm-up application of the Gysin sequence let us use it to give a different

proof of Theorem 3.9 computing H

∗

(G

n

; Z

2

) and H

∗

(G

n

(C); Z). Consider ﬁrst the

Euler and Pontryagin Classes Section 3.2 89

real case. The proof will be by induction on n using the Gysin sequence for the univer-

sal bundle E

n

π

→G

n

. The sphere bundle S(E

n

) is the space of pairs (v, ) where

is an n dimensional linear subspace of R

∞

and v is a unit vector in . There is a nat-

ural map p: S(E

n

)→G

n−1

sending (v, ) to the (n−1) dimensional linear subspace

v

⊥

⊂ orthogonal to v. It is an exercise to check that p is a ﬁber bundle. Its ﬁber is

S

∞

, all the unit vectors in R

∞

orthogonal to a given (n − 1) dimensional subspace.

Since S

∞

is contractible, p induces an isomorphism on all homotopy groups, hence

also on all cohomology groups. Using this isomorphism p

∗

the Gysin sequence, with

Z

2

coefﬁcients, has the form

··· →H

i

(G

n

)

e

→H

i+n

(G

n

)

η

→H

i+n

(G

n−1

) →H

i+1

(G

n

) →···

where e ∈ H

n

(G

n

; Z

2

) is the Z

2

Euler class.

We show ﬁrst that η(w

j

(E

n

)) = w

j

(E

n−1

). By deﬁnition the map η is the com-

position H

∗

(G

n

)→H

∗

(S(E

n

))

≈

← H

∗

(G

n−1

) induced from G

n−1

p

← S(E

n

)

π

→G

n

.

The pullback π

∗

(E

n

) consists of triples (v, w, ) where ∈ G

n

and v, w ∈ with

v a unit vector. This pullback splits naturally as a sum L⊕p

∗

(E

n−1

) where L is the

subbundle of triples (v, tv, ), t ∈ R, and p

∗

(E

n−1

) consists of the triples (v, w, )

with w ∈ v

⊥

. The line bundle L is trivial, having the section (v, v, ). Thus the coho-

mology homomorphism π

∗

takes w

j

(E

n

)) to w

j

(L⊕p

∗

(E

n−1

)) = w

j

(p

∗

(E

n−1

)) =

p

∗

(w

j

(E

n−1

)), so η(w

j

(E

n

)) = w

j

(E

n−1

).

By induction on n, H

∗

(G

n−1

) is the polynomial ring on the classes w

j

(E

n−1

),

j < n. The induction can start with the case n = 1, where G

1

= RP

∞

and H

∗

(RP

∞

) ≈

Z

2

[w

1

] since w

1

(E

1

) is a generator of H

1

(RP

∞

; Z

2

). Or we could start with the trivial

case n = 0. Since η(w

j

(E

n

)) = w

j

(E

n−1

), the maps η are surjective and the Gysin

sequence splits into short exact sequences

0 →H

i

(G

n

)

e

→H

i+n

(G

n

)

η

→H

i+n

(G

n−1

) →0

The image of e : H

0

(G

n

)→H

n

(G

n

) is a Z

2

generated by e. By exactness, this Z

2

is the kernel of η: H

n

(G

n

)→H

n

(G

n−1

). The class w

n

(E

n

) lies in this kernel since

w

n

(E

n−1

) = 0. Moreover, w

n

(E

n

) ≠ 0, since if w

n

(E

n

) = 0 then w

n

is zero for all

n dimensional vector bundles, but the bundle E→RP

∞

which is the direct sum of n

copies of the canonical line bundle has total Stiefel-Whitney class w(E) = (1 + α)

n

,

where α generates H

1

(RP

∞

), hence w

n

(E) = α

n

≠ 0. Thus e and w

n

(E

n

) generate

the same Z

2

, so e = w

n

(E

n

).

Now we argue that each element ξ ∈ H

k

(G

n

) can be expressed as a unique poly-

nomial in the classes w

i

= w

i

(E

n

), by induction on k. First, η(ξ) is a unique polyno-

mial f in the w

i

(E

n−1

)’s by the basic induction on n. Then ξ−f(w

1

, ··· , w

n−1

) is in

Ker η = Im(w

n

), hence has the form ζw

n

for ζ ∈ H

k−n

(G

n

) which is unique since

w

n

is injective. By induction on k, ζ is a unique polynomial g in the w

i

’s. Thus

we have ξ expressed uniquely as a polynomial f(w

1

, ··· , w

n−1

) +w

n

g(w

1

, ··· , w

n

).

90 Chapter 3 Characteristic Classes

Since every polynomial in w

1

, ··· , w

n

has a unique expression in this form, the theo-

rem follows in the real case.

Virtually the same argument works in the complex case. We noted earlier that

complex vector bundles always have a Gysin sequence with Z coefﬁcients. The only

elaboration needed to extend the preceding proof to the complex case is at the step

where we showed the Z

2

Euler class is w

n

. The argument from the real case shows

that c

n

is a multiple me for some m ∈ Z, e being now the Z Euler class. Then

for the bundle E→CP

∞

which is the direct sum of n copies of the canonical line

bundle, classiﬁed by f : CP

∞

→G

n

(C

∞

), we have α

n

= c

n

(E) = f

∗

(c

n

) = mf

∗

(e) in

H

2n

(CP

∞

; Z) ≈ Z, with α

n

a generator, hence m = :1 and e = :c

n

. The rest of the

proof goes through without change.

We can also compute H

∗

(

`

G

n

; Z

2

) where

`

G

n

is the oriented Grassmannian. To

state the result, let π :

`

G

n

→G

n

be the covering projection, so

`

E

n

= π

∗

(E

n

), and let

w

i

= w

i

(

`

E

n

) = π

∗

(w

i

) ∈ H

i

(

`

G

n

; Z

2

), where w

i

= w

i

(E

n

).

Proposition 3.12. π

∗

: H

∗

(G

n

; Z

2

)→H

∗

(

`

G

n

; Z

2

) is surjective with kernel the ideal

generated by w

1

, hence H

∗

(

`

G

n

; Z

2

) ≈ Z

2

[ w

2

, ··· , w

n

].

This is just the answer one would hope for. Since

`

G

n

is simply-connected, w

1

has

to be zero, so the isomorphism H

∗

(

`

G

n

; Z

2

) ≈ Z

2

[ w

2

, ··· , w

n

] is the simplest thing

that could happen.

Proof: The 2 sheeted covering π :

`

G

n

→G

n

can be regarded as the unit sphere bundle

of a 1 dimensional vector bundle, so we have a Gysin sequence beginning

0 →H

0

(G

n

; Z

2

) →H

0

(

`

G

n

; Z

2

) →H

0

(G

n

; Z

2

)

x

→H

1

(G

n

; Z

2

)

where x ∈ H

1

(G

n

; Z

2

) is the Z

2

Euler class. Since

`

G

n

is connected, H

0

(

`

G

n

; Z

2

) ≈

Z

2

and so the map x is injective, hence x = w

1

, the only nonzero element of

H

1

(G

n

; Z

2

). Since H

∗

(G

n

; Z

2

) ≈ Z

2

[w

1

, ··· , w

n

], the map w

1

is injective in all

dimensions, so the Gysin sequence breaks up into short exact sequences

0 →H

i

(G

n

; Z

2

)

w

1

→H

i

(G

n

; Z

2

)

π

∗

→H

i

(

`

G

n

; Z

2

) →0

from which the conclusion is immediate. L¦

The goal for the rest of this section is to determine H

∗

(G

n

; Z) and H

∗

(

`

G

n

; Z),

or in other words, to ﬁnd all characteristic classes for real vector bundles with Z

coefﬁcients, rather than the Z

2

coefﬁcients used for Stiefel-Whitney classes. It turns

out that H

∗

(G

n

; Z), modulo elements of order 2 which are just certain polynomials

in Stiefel-Whitney classes, is a polynomial ring Z[p

1

, p

2

, ···] on certain classes p

i

of

dimension 4i, called Pontryagin classes. There is a similar statement for H

∗

(

`

G

n

; Z),

but with one of the Pontryagin classes replaced by an Euler class when n is even.

Euler and Pontryagin Classes Section 3.2 91

The Euler Class

Recall that the Euler class e(E) ∈ H

n

(B; Z) of an orientable n dimensional vector

bundle E→B is the restriction of a Thom class c ∈ H

n

(D(E), S(E); Z) to the zero

section, that is, the image of c under the composition

H

n

(D(E), S(E); Z)→H

n

(D(E); Z)→H

n

(B; Z)

where the ﬁrst map is the usual passage fromrelative to absolute cohomology and the

second map is induced by the inclusion B

**D(E) as the zero section. By its deﬁnition,
**

e(E) depends on the choice of c. However, the assertion (*) in the construction of a

Thom class in Theorem 4D.10 of [AT] implies that c is determined by its restriction to

each ﬁber, and the restriction of c to each ﬁber is in turn determined by an orientation

of the bundle, so in fact e(E) depends only on the choice of an orientation of E.

Choosing the opposite orientation changes the sign of c. There are exactly two choices

of orientation for each path-component of B.

Here are the basic properties of Euler classes e(E) ∈ H

n

(B; Z) associated to ori-

ented n dimensional vector bundles E→B:

Proposition 3.13.

(a) An orientation of a vector bundle E→B induces an orientation of a pullback

bundle f

∗

(E) such that e(f

∗

(E)) = f

∗

(e(E)).

(b) Orientations of vector bundles E

1

→B and E

2

→B determine an orientation of

the sum E

1

⊕E

2

such that e(E

1

⊕E

2

) = e(E

1

) e(E

2

).

(c) For an orientable n dimensional real vector bundle E, the coefﬁcient homomor-

phism H

n

(B; Z)→H

n

(B; Z

2

) carries e(E) to w

n

(E). For an n dimensional com-

plex vector bundle E there is the relation e(E) = c

n

(E) ∈ H

2n

(B; Z), for a suitable

choice of orientation of E.

(d) e(E) = −e(E) if the ﬁbers of E have odd dimension.

(e) e(E) = 0 if E has a nowhere-zero section.

Proof: (a) For an n dimensional vector bundle E, let E

¹

⊂ E be the complement of

the zero section. A Thom class for E can be viewed as an element of H

n

(E, E

¹

; Z)

which restricts to a generator of H

n

(R

n

, R

n

−¦0¦; Z) in each ﬁber R

n

. For a pullback

f

∗

(E), we have a map

`

f : f

∗

(E)→E which is a linear isomorphism in each ﬁber, so

`

f

∗

(c(E)) restricts to a generator of H

n

(R

n

, R

n

− ¦0¦; Z) in each ﬁber R

n

of f

∗

(E).

Thus

`

f

∗

(c(E)) = c(f

∗

(E)). Passing from relative to absolute cohomology classes

and then restricting to zero sections, we get e(f

∗

(E)) = f

∗

(e(E)).

(b) There is a natural projection p

1

: E

1

⊕E

2

→E

1

which is linear in each ﬁber, and

likewise we have p

2

: E

1

⊕E

2

→E

2

. If E

1

is m dimensional we can view a Thom class

c(E

1

) as lying in H

m

(E

1

, E

¹

1

) where E

¹

1

is the complement of the zero section in E

1

.

Similarly we have a Thom class c(E

2

) ∈ H

n

(E

2

, E

¹

2

) if E

2

has dimension n. Then

the product p

∗

1

(c(E

1

)) p

∗

2

(c(E

2

)) is a Thom class for E

1

⊕E

2

since in each ﬁber

92 Chapter 3 Characteristic Classes

R

m

×R

n

= R

m+n

we have the cup product

H

m

(R

m+n

, R

m+n

−R

n

)×H

n

(R

m+n

, R

m+n

−R

m

)z→H

m+n

(R

m+n

, R

m+n

−¦0¦)

which takes generator cross generator to generator by the calculations in Example 3.11

of [AT]. Passing from relative to absolute cohomology and restricting to the zero sec-

tion, we get the relation e(E

1

⊕E

2

) = e(E

1

) e(E

2

).

(c) We showed this for the universal bundle in the calculation of the cohomology of

Grassmannians a couple pages back, so by the naturality property in (a) it holds for

all bundles.

(d) When we deﬁned the Euler class we observed that it could also be described as the

element of H

n

(B; Z) corresponding to c c ∈ H

2n

(D(E), S(E), Z) under the Thom

isomorphism. If n is odd, the basic commutativity relation for cup products gives

c c = −c c, so e(E) = −e(E).

(e) A nowhere-zero section of E gives rise to a section s : B→S(E) by normalizing

vectors to have unit length. Then in the exact sequence

H

n

(D(E), S(E); Z)

j

∗

→H

n

(D(E); Z)

i

∗

→H

n

(S(E); Z)

the map i

∗

is injective since the composition D(E) →B

s

→S(E)

i

→D(E) is homo-

topic to the identity. Since i

∗

is injective, the map j

∗

is zero by exactness, and hence

e(E) = 0 from the deﬁnition of the Euler class. L¦

Consider the tangent bundle TS

n

to S

n

. This bundle is orientable since its base

S

n

is simply-connected if n > 1, while if n = 1, TS

1

is just the product S

1

×R.

When n is odd, e(TS

n

) = 0 either by part (d) of the proposition since H

∗

(S

n

; Z) has

no elements of order two, or by part (e) since there is a nonzero tangent vector ﬁeld

to S

n

when n is odd, namely s(x

1

, ··· , x

n+1

) = (−x

2

, x

1

, ··· , −x

n+1

, x

n

). However,

when n is even e(TS

n

) is nonzero:

Proposition 3.14. For even n, e(TS

n

) is twice a generator of H

n

(S

n

; Z).

Proof: Let E

¹

⊂ E = TS

n

be the complement of the zero section. Under the Thom

isomorphism the Euler class e(TS

n

) corresponds to the square of a Thom class

c ∈ H

n

(E, E

¹

), so it sufﬁces to show that c

2

is twice a generator of H

2n

(E, E

¹

). Let

A ⊂ S

n

×S

n

consist of the antipodal pairs (x, −x). De-

ﬁne a homeomorphism f : S

n

×S

n

− A→E sending a pair

(x, y) ∈ S

n

×S

n

− A to the vector from x to the point of

intersection of the line through −x and y with the tangent

plane at x. The diagonal D = ¦(x, x)¦ corresponds under

x

y

x

f to the zero section of E. Excision then gives the ﬁrst of

the following isomorphisms:

H

∗

(E, E

¹

) ≈ H

∗

(S

n

×S

n

, S

n

×S

n

−D) ≈ H

∗

(S

n

×S

n

, A) ≈ H

∗

(S

n

×S

n

, D),

Euler and Pontryagin Classes Section 3.2 93

The second isomorphism holds since S

n

×S

n

− D deformation retracts onto A by

sliding a point y ≠ :x along the great circle through x and y to −x, and the third

comes from the homeomorphism (x, y)

(x, −y) of S

n

×S

n

interchanging D and

A. From the long exact sequence of the pair (S

n

×S

n

, D) we extract a short exact

sequence

0→H

n

(S

n

×S

n

, D)→H

n

(S

n

×S

n

)→H

n

(D)→0

The middle group H

n

(S

n

×S

n

) has generators α, β which are pullbacks of a gener-

ator of H

n

(S

n

) under the two projections S

n

×S

n

→S

n

. Both α and β restrict to

the same generator of H

n

(D) since the two projections S

n

×S

n

→S

n

restrict to the

same homeomorphism D ≈ S

n

, so α−β generates H

n

(S

n

×S

n

, D), the kernel of the

restriction map H

n

(S

n

×S

n

)→H

n

(D). Thus α − β corresponds to the Thom class

and (α−β)

2

= −αβ−βα, which equals −2αβ if n is even. This is twice a generator

of H

2n

(S

n

×S

n

, D) ≈ H

2n

(S

n

×S

n

). L¦

It is a fairly elementary theorem in differential topology that the Euler class of

the unit tangent bundle of a closed, connected, orientable smooth manifold M

n

is

¦χ(M)¦ times a generator of H

n

(M), where χ(M) is the Euler characteristic of M;

see for example [Milnor-Stasheff]. This agrees with what we have just seen in the case

M = S

n

, and is the reason for the name ‘Euler class.’

One might ask which elements of H

n

(S

n

) can occur as Euler classes of vector

bundles E→S

n

in the nontrivial case that n is even. If we form the pullback of the

tangent bundle TS

n

by a map S

n

→S

n

of degree d, we realize 2d times a generator,

by part (a) of the preceding proposition, so all even multiples of a generator of H

n

(S

n

)

are realizable. To investigate odd multiples, consider the Thom space T(E). This

has integral cohomology consisting of Z’s in dimensions 0, n, and 2n by the Thom

isomorphism, which also says that the Thom class c is a generator of H

n

(T(E)). We

know that the Euler class corresponds under the Thom isomorphism to c c, so e(E)

is k times a generator of H

n

(S

n

) iff c c is k times a generator of H

2n

(T(E)). This

is precisely the context of the Hopf invariant, and the solution of the Hopf invariant

one problem in Chapter 2 shows that c c can be an odd multiple of a generator

only if n = 2, 4, or 8. In these three cases there is a bundle E→S

n

for which c c

is a generator of H

2n

(T(E)), namely the vector bundle whose unit sphere bundle is

the complex, quaternionic, or octonionic Hopf bundle, and whose Thom space, the

mapping cone of the sphere bundle, is the associated projective plane CP

2

, HP

2

, or

OP

2

. Since we can realize a generator of H

n

(S

n

) as an Euler class in these three cases,

we can realize any multiple of a generator by taking pullbacks as before.

94 Chapter 3 Characteristic Classes

Pontryagin Classes

The easiest deﬁnition of the Pontryagin classes p

i

(E) ∈ H

4i

(B; Z) associated to

a real vector bundle E→B is in terms of Chern classes. For a real vector bundle

E→B, its complexiﬁcation is the complex vector bundle E

C

→B obtained from the

real vector bundle E⊕E by deﬁning scalar multiplication by the complex number i

in each ﬁber R

n

⊕R

n

via the familiar rule i(x, y) = (−y, x). Thus each ﬁber R

n

of E becomes a ﬁber C

n

of E

C

. The Pontryagin class p

i

(E) is then deﬁned to be

(−1)

i

c

2i

(E

C

) ∈ H

4i

(B; Z). The sign (−1)

i

is introduced in order to avoid a sign in the

formula in (b) of the next proposition. The reason for restricting attention to the even

Chern classes c

2i

(E

C

) is that the odd classes c

2i+1

(E

C

) turn out to be expressible in

terms of Stiefel-Whitney classes, and hence give nothing new. The exercises at the

end of the section give an explicit formula.

Here is how Pontryagin classes are related to Stiefel-Whitney and Euler classes:

Proposition 3.15.

(a) For a real vector bundle E→B, p

i

(E) maps to w

2i

(E)

2

under the coefﬁcient

homomorphism H

4i

(B; Z)→H

4i

(B; Z

2

).

(b) For an orientable real 2n dimensional vector bundle with Euler class e(E) ∈

H

2n

(B; Z), p

n

(E) = e(E)

2

.

Note that statement (b) is independent of the choice of orientation of E since the

Euler class is squared.

Proof: (a) By Proposition 3.4, c

2i

(E

C

) reduces mod 2 to w

4i

(E⊕E), which equals

w

2i

(E)

2

since w(E⊕E) = w(E)

2

and squaring is an additive homomorphism mod 2.

(b) First we need to determine the relationship between the two orientations of E

C

≈

E⊕E, one coming from the canonical orientation of the complex bundle E

C

, the

other coming from the orientation of E⊕E determined by an orientation of E. If

v

1

, ··· , v

2n

is a basis for a ﬁber of E agreeing with the given orientation, then E

C

is oriented by the ordered basis v

1

, iv

1

, ··· , v

2n

, iv

2n

, while E⊕E is oriented by

v

1

, ··· , v

2n

, iv

1

, ··· , iv

2n

. To make these two orderings agree requires (2n − 1) +

(2n − 2) + ··· + 1 = 2n(2n − 1)/2 = n(2n − 1) transpositions, so the two orienta-

tions differ by a sign (−1)

n(2n−1)

= (−1)

n

. Thus we have p

n

(E) = (−1)

n

c

2n

(E

C

) =

(−1)

n

e(E

C

) = e(E⊕E) = e(E)

2

. L¦

Pontryagin classes can be used to describe the cohomology of G

n

and

`

G

n

with Z

coefﬁcients. First let us remark that since G

n

has a CW structure with ﬁnitely many

cells in each dimension, so does

`

G

n

, hence the homology and cohomology groups of

G

n

and

`

G

n

are ﬁnitely generated. For the universal bundles E

n

→G

n

and

`

E

n

→

`

G

n

let p

i

= p

i

(E

n

), ` p

i

= p

i

(

`

E

n

), and e = e(

`

E

n

), the Euler class being deﬁned via the

canonical orientation of

`

E

n

.

Euler and Pontryagin Classes Section 3.2 95

Theorem 3.16.

(a) All torsion in H

∗

(G

n

; Z) consists of elements of order 2, and H

∗

(G

n

; Z)/torsion

is the polynomial ring Z[p

1

, ··· , p

k

] for n = 2k or 2k +1.

(b) All torsion in H

∗

(

`

G

n

; Z) consists of elements of order 2, and H

∗

(

`

G

n

; Z)/torsion

is Z[ ` p

1

, ··· , ` p

k

] for n = 2k+1 and Z[ ` p

1

, ··· , ` p

k−1

, e] for n = 2k, with e

2

= ` p

k

in the latter case.

The torsion subgroup of H

∗

(G

n

; Z) therefore maps injectively to H

∗

(G

n

; Z

2

),

with image the image of the Bockstein β: H

∗

(G

n

; Z

2

)→H

∗

(G

n

; Z

2

), which we shall

compute in the course of proving the theorem; for the deﬁnition and basic properties

of Bockstein homomorphisms see §3.E of [AT]. The same remarks apply to H

∗

(

`

G

n

; Z).

The theorem implies that Stiefel-Whitney and Pontryagin classes determine all char-

acteristic classes for unoriented real vector bundles, while for oriented bundles the

only additional class needed is the Euler class.

Proof: We shall work on (b) ﬁrst since for orientable bundles there is a Gysin sequence

with Z coefﬁcients. As a ﬁrst step we compute H

∗

(

`

G

n

; R) where R = Z[

1

/

2

] ⊂ Q, the

rational numbers with denominator a power of 2. Since we are dealing with ﬁnitely

generated integer homology groups, changing from Z coefﬁcients to R coefﬁcients

eliminates any 2 torsion in the homology, that is, elements of order a power of 2, and

Z summands of homology become R summands. The assertion to be proved is that

H

∗

(

`

G

n

; R) is R[ ` p

1

, ··· , ` p

k

] for n = 2k + 1 and R[ ` p

1

, ··· , ` p

k−1

, e] for n = 2k. This

implies that H

∗

(

`

G

n

; Z) has no odd-order torsion and that H

∗

(

`

G

n

; Z)/torsion is as

stated in the theorem. Then it will remain only to show that all 2 torsion in H

∗

(

`

G

n

; Z)

consists of elements of order 2.

As in the calculation of H

∗

(G

n

; Z

2

) via the Gysin sequence, consider the sphere

bundle S

n−1

→S(

`

E

n

)

π

→

`

G

n

, where S(

`

E

n

) is the space of pairs (v, ) where is

an oriented n dimensional linear subspace of R

∞

and v is a unit vector in . The

orthogonal complement v

⊥

⊂ of v is then naturally oriented, so we get a projection

p: S(

`

E

n

)→

`

G

n−1

. The Gysin sequence with coefﬁcients in R has the form

··· →H

i

(

`

G

n

)

e

→H

i+n

(

`

G

n

)

η

→H

i+n

(

`

G

n−1

) →H

i+1

(

`

G

n

) →···

where η takes ` p

i

(

`

E

n

) to ` p

i

(

`

E

n−1

).

If n = 2k, then by induction H

∗

(

`

G

n−1

) ≈ R[ ` p

1

, ··· , ` p

k−1

], so η is surjective and

the sequence splits into short exact sequences. The proof in this case then follows

the H

∗

(G

n

; Z

2

) model.

If n = 2k +1, then e is zero in H

n

(

`

G

n

; R) since with Z coefﬁcients it has order

2. The Gysin sequence now splits into short exact sequences

0 →H

i+n

(

`

G

n

)

η

→H

i+n

(

`

G

n−1

) →H

i+1

(

`

G

n

) →0

Thus η injects H

∗

(

`

G

n

) as a subring of H

∗

(

`

G

n−1

) ≈ R[ ` p

1

, ··· , ` p

k−1

, e], where e now

means e(

`

E

n−1

). The subring Imη contains R[ ` p

1

, ··· , ` p

k

] and is torsionfree, so we

96 Chapter 3 Characteristic Classes

can show it equals R[ ` p

1

, ··· , ` p

k

] by comparing ranks of these R modules in each

dimension. Let r

j

be the rank of R[ ` p

1

, ··· , ` p

k

] in dimension j and r

¹

j

the rank

of H

j

(

`

G

n

). Since R[ ` p

1

, ··· , ` p

k−1

, e] is a free module over R[ ` p

1

, ··· , ` p

k

] with basis

¦1, e¦, the rank of H

∗

(

`

G

n−1

) ≈ R[ ` p

1

, ··· , ` p

k−1

, e] in dimension j is r

j

+ r

j−2k

, the

class e = e(

`

E

n−1

) having dimension 2k. On the other hand, the exact sequence above

says this rank also equals r

¹

j

+r

¹

j−2k

. Since r

¹

m

≥ r

m

for each m, we get r

¹

j

= r

j

, and

so H

∗

(

`

G

n

) = R[ ` p

1

, ··· , ` p

k

], completing the induction step. The induction can start

with the case n = 1, with

`

G

1

≈ S

∞

.

Before studying the remaining 2 torsion question let us extend what we have just

done to H

∗

(G

n

; Z), to show that for R = Z[

1

/

2

], H

∗

(G

n

; R) is R[p

1

, ··· , p

k

], where

n = 2k or 2k + 1. For the 2 sheeted covering π :

`

G

n

→G

n

consider the transfer ho-

momorphism π

∗

: H

∗

(

`

G

n

; R)→H

∗

(G

n

; R) deﬁned in §3.G of [AT]. The main feature

of π

∗

is that the composition π

∗

π

∗

: H

∗

(G

n

; R)→H

∗

(

`

G

n

; R)→H

∗

(G

n

; R) is multi-

plication by 2, the number of sheets in the covering space. This is an isomorphism

for R = Z[

1

/

2

], so π

∗

is injective. The image of π

∗

contains R[ ` p

1

, ··· , ` p

k

] since

π

∗

(p

i

) = ` p

i

. So when n is odd, π

∗

is an isomorphism and we are done. When n

is even, observe that the image of π

∗

is invariant under the map τ

∗

induced by the

deck transformation τ :

`

G

n

→

`

G

n

interchanging sheets of the covering, since πτ = π

implies τ

∗

π

∗

= π

∗

. The map τ reverses orientation in each ﬁber of

`

E

n

→

`

G

n

, so τ

∗

takes e to −e. The subring of H

∗

(

`

G

n

; R) ≈ R[ ` p

1

, ··· , ` p

k−1

, e] invariant under τ

∗

is

then exactly R[ ` p

1

, ··· , ` p

[n/2]

], ﬁnishing the proof that H

∗

(G

n

; R) = R[p

1

, ··· , p

k

].

To show that all 2 torsion in H

∗

(G

n

; Z) and H

∗

(

`

G

n

; Z) has order 2 we use the

Bockstein homomorphism β associated to the short exact sequence of coefﬁcient

groups 0→Z

2

→Z

4

→Z

2

→0. The goal is to show that Ker β/ Imβ consists exactly

of the mod 2 reductions of nontorsion classes in H

∗

(G

n

; Z) and H

∗

(

`

G

n

; Z), that is,

polynomials in the classes w

2

2i

in the case of G

n

and

`

G

2k+1

, and for

`

G

2k

, polynomials

in the w

2

2i

’s for i < k together with w

2k

, the mod 2 reduction of the Euler class. By

general properties of Bockstein homomorphisms proved in §3.E of [AT] this will ﬁnish

the proof.

Lemma 3.17. βw

2i+1

= w

1

w

2i+1

and βw

2i

= w

2i+1

+w

1

w

2i

.

Proof: By naturality it sufﬁces to prove this for the universal bundle E

n

→G

n

with

w

i

= w

i

(E

n

). As observed in §3.1, we can view w

k

as the k

th

elementary symmetric

polynomial σ

k

in the polynomial algebra Z

2

[α

1

, ··· , α

n

] ≈ H

∗

((RP

∞

)

n

; Z

2

). Thus

to compute βw

k

we can compute βσ

k

. Using the derivation property β(x y) =

βx y + x βy and the fact that βα

i

= α

2

i

, we see that βσ

k

is the sum of all

products α

i1

··· α

2

ij

··· α

ik

for i

1

< ··· < i

k

and j = 1, ··· , k. On the other hand,

multiplying σ

1

σ

k

out, one obtains βσ

k

+(k +1)σ

k+1

. L¦

Now for the calculation of Ker β/ Imβ. First consider the case of G

2k+1

. The ring

Z

2

[w

1

, ··· , w

2k+1

] is also the polynomial ring Z

2

[w

1

, w

2

, βw

2

, ··· , w

2k

, βw

2k

] since

Exercises 97

the substitution w

1

w

1

, w

2i

w

2i

, w

2i+1

w

2i+1

+ w

1

w

2i

= βw

2i

for i > 0 is

invertible, being its own inverse in fact. Thus Z

2

[w

1

, ··· , w

2k+1

] splits as the tensor

product of the polynomial rings Z

2

[w

1

] and Z

2

[w

2i

, βw

2i

], each of which is invariant

under β. Moreover, viewing Z

2

[w

1

, ··· , w

2k+1

] as a chain complex with boundary

map β, this tensor product is a tensor product of chain complexes. According to

the algebraic K¨ unneth theorem, the homology of Z

2

[w

1

, ··· , w

2k+1

] with respect to

the boundary map β is therefore the tensor product of the homologies of the chain

complexes Z

2

[w

1

] and Z

2

[w

2i

, βw

2i

].

For Z

2

[w

1

] we have β(w

1

) = w

+1

1

, so Ker β is generated by the even powers

of w

1

, all of which are also in Imβ, and hence the β homology of Z

2

[w

1

] is trivial in

positive dimensions; we might remark that this had to be true since the calculation is

the same as for RP

∞

. For Z

2

[w

2i

, βw

2i

] we have β(w

2i

(βw

2i

)

m

) = w

−1

2i

(βw

2i

)

m+1

,

so Ker β is generated by the monomials w

2i

(βw

2i

)

m

with even, and such monomials

with m > 0 are in Imβ. Hence Ker β/ Imβ = Z

2

[w

2

2i

].

For n = 2k, Z

2

[w

1

, ··· , w

2k

] is the tensor product of the Z

2

[w

2i

, βw

2i

]’s for

i < k and Z

2

[w

1

, w

2k

], with β(w

2k

) = w

1

w

2k

. We then have the formula β(w

1

w

m

2k

) =

w

+1

1

w

m

2k

+mw

+1

1

w

m

2k

= ( +m)w

+1

1

w

m

2k

. For w

1

w

m

2k

to be in Ker β we must have

+ m even, and to be in Imβ we must have in addition > 0. So Ker β/ Imβ =

Z

2

[w

2

2k

].

Thus the homology of Z

2

[w

1

, ··· , w

n

] with respect to β is the polynomial ring in

the classes w

2

2i

, the mod 2 reductions of the Pontryagin classes. By general properties

of Bocksteins this ﬁnishes the proof of part (a) of the theorem.

The case of

`

G

n

is simpler since w

1

= 0, hence βw

2i

= w

2i+1

and βw

2i+1

= 0.

Then we can break Z

2

[w

2

, ··· , w

n

] up as the tensor product of the chain complexes

Z

2

[w

2i

, w

2i+1

], plus Z

2

[w

2k

] when n = 2k. The calculations are quite similar to those

we have just done, so further details will be left as an exercise. L¦

Exercises

1. Show that every class in H

2k

(CP

∞

) can be realized as the Euler class of some vector

bundle over CP

∞

that is a sum of complex line bundles.

2. Show that c

2i+1

(E

C

) = β(w

2i

(E)w

2i+1

(E)).

3. For an oriented (2k +1) dimensional vector bundle E show that e(E) = βw

2k

(E).

Homotopy groups of spheres are notoriously difﬁcult to compute, but some par-

tial information can be gleaned from certain naturally deﬁned homomorphisms

J : π

i

(O(n))→π

n+i

(S

n

)

The goal of this chapter is to determine these J homomorphisms in the stable dimen-

sion range n >> i where both domain and range are independent of n, according to

Proposition 1.14 for O(n) and the Freudenthal suspension theorem [AT] for S

n

. The

real form of Bott periodicity proved in Chapter 2 implies that the domain of the stable

J homomorphism π

i

(O)→π

s

i

is nonzero only for i = 4n −1 when π

i

(O) is Z and

for i = 8n and 8n+1 when π

i

(O) is Z

2

. In the latter two cases we will show that J

is injective. When i = 4n−1 the image of J is a ﬁnite cyclic group of some order a

n

since π

s

i

is a ﬁnite group for i > 0 by a theorem of Serre proved in [SSAT].

The values of a

n

have been computed in terms of Bernoulli numbers. Here is a

table for small values of n:

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−

−

−

−

−

−

−

−

−

−

−

−

−

24 240 264 65520 24 13200 16320 28728 480 504 552

1 2 3 4 5 6 7 8 9 10 11 n

a

n

In spite of appearances, there is great regularity in this sequence, but this becomes

clear only when one looks at the prime factorization of a

n

. Here are the rules for

computing a

n

:

1. The highest power of 2 dividing a

n

is 2

+3

where 2

**is the highest power of 2
**

dividing n.

2. An odd prime p divides a

n

iff n is a multiple of (p − 1)/2, and in this case

the highest power of p dividing a

n

is p

+1

where p

**is the highest power of p
**

dividing n.

The ﬁrst three cases p = 2, 3, 5 are shown in the following diagram, where a vertical

chain of k connected dots above the number 4n−1 means that the highest power of

p dividing a

n

is p

k

.

Lower Bounds on Im J Section 4.1 99

−−−−−

3 7 11 15 19 23 27 31 35 39 43 47 51 55 59 63 67 71 75 79 83 87 91 95 99 91 107 103

3 7 11 15 19 23 27 31 35 39 43 47 51 55 59 63 67 71 75 79 83 87 91 95 99 91 107 103

p 2 : =

−−−−−

p 3 : =

−−−−−

p 5 : =

39 79 119 159 199 239 279 319 359 399 439 479 519 559 599 639 679 719 759 799 839 879 919 959 999

4.1 Lower Bounds for Im J

After starting this section with the deﬁnition of the J–homomorphism, we will

use a homomorphism K(X)→H

∗

(X; Q) known as the Chern character to show that

a

n

/2 is a lower bound on the order of the image of J in dimension 4n − 1. Then

using real K–theory this bound will be improved to a

n

, and we will take care of the

cases in which the domaim of the J–homomorphism is Z

2

.

The simplest deﬁnition of the J homomorphism goes as follows. An element

[f] ∈ π

i

(O(n)) is represented by a family of isometries f

x

∈ O(n), x ∈ S

i

, with f

x

the identity when x is the basepoint of S

i

. Writing S

n+i

as ∂(D

i+1

×D

n

) = S

i

×D

n

∪

D

i+1

×S

n−1

and S

n

as D

n

/∂D

n

, let Jf(x, y) = f

x

(y) for (x, y) ∈ S

i

×D

n

and let

Jf(D

i+1

×S

n−1

) = ∂D

n

, the basepoint of D

n

/∂D

n

. Clearly f = g implies Jf = Jg,

so we have a map J : π

i

(O(n))→π

n+i

(S

n

). We will tacitly exclude the trivial case

i = 0.

Proposition 4.1. J is a homomorphism.

Proof: We can view Jf as a map I

n+i

→S

n

= D

n

/∂D

n

which on S

i

×D

n

⊂ I

n+1

is

given by (x, v)

f

x

(v) and which sends the complement of S

i

×D

n

to the basepoint

∂D

n

. Taking a similar view of Jg, the sum Jf +Jg is obtained by juxtaposing these

two maps on either side of a hyperplane. We may assume f

x

is the identity for x in

the right half of S

i

and g

x

is the identity for x in the left half of S

i

. Then we obtain

a homotopy from Jf + Jg to J(f + g) by moving the two S

i

×D

n

’s together until

they coincide, as shown in the ﬁgure below. L¦

−−→ −−→

We know that π

i

(O(n)) and π

n+i

(S

n

) are independent of n for n > i + 1, so

we would expect the J–homomorphism deﬁned above

100 Chapter 4 The J–Homomorphism

to induce a stable J–homomorphism J : π

i

(O)→π

s

i

, via

−−−−−→

−−−−→

−−→

O n ( ) )

S

S

( )

J

−−−−−→

n

J

π

i

π

( O ( ) ) π

i

(n 1 +

γ

n i +

n 1 +

S ( ) π

n i 1 + +

commutativity of the diagramat the right. We leave it as

an exercise for the reader to verify that this is the case.

Composing the stable J–homomorphismwith the map π

i

(U)→π

i

(O) induced by

the natural inclusions U(n) ⊂ O(2n) which give an inclusion U ⊂ O, we get the stable

complex J–homomorphism J

C

: π

i

(U)→π

s

i

. Our goal is to deﬁne via K–theory a homo-

morphism e : π

s

i

→Q/Z for i odd and compute the composition eJ

C

: π

i

(U)→Q/Z.

This will give a lower bound for the order of the image of the real J–homomorphism

π

i

(O)→π

s

i

when i = 4n−1.

The Chern Character

The total Chern class c = 1 + c

1

+ c

2

+ ··· takes direct sums to cup products,

and the idea of the Chern character is to form an algebraic combination of Chern

classes which takes direct sums to sums and tensor products to cup products, thus

giving a natural ring homomorphism from K–theory to cohomology. In order to make

this work one must use cohomology with rational coefﬁcients, however. The situation

might have been simpler if it had been possible to use integer coefﬁcients instead, but

on the other hand, the fact that one has rational coefﬁcients instead of integers makes

it possible to deﬁne a homomorphism e : π

2m−1

(S

2n

)→Q/Z which gives some very

interesting information about the difﬁcult subject of homotopy groups of spheres.

In order to deﬁne the Chern character it sufﬁces, via the splitting principle, to do

the case of line bundles. The idea is to deﬁne the Chern character ch(L) for a line

bundle L→X to be ch(L) = e

c

1

(L)

= 1 +c

1

(L) +c

1

(L)

2

/2! +··· ∈ H

∗

(X; Q), so that

ch(L

1

⊗L

2

) = e

c

1

(L

1

⊗L

2

)

= e

c

1

(L

1

)+c

1

(L

2

)

= e

c

1

(L

1

)

e

c

1

(L

2

)

= ch(L

1

)ch(L

2

). If the sum

1+c

1

(L)+c

1

(L)

2

/2! +··· has inﬁnitely many nonzero terms, it will lie not in the direct

sum H

∗

(X; Q) of the groups H

n

(X; Q) but rather in the direct product. However, in

the examples we shall be considering, H

n

(X; Q) will be zero for sufﬁciently large n,

so this distinction will not matter.

For a direct sum of line bundles E ≈ L

1

⊕ ··· ⊕L

n

we would then want to have

ch(E) =

¸

i

ch(L

i

) =

¸

i

e

t

i

= n+(t

1

+··· +t

n

) +··· +(t

k

1

+··· +t

k

n

)/k! +···

where t

i

= c

1

(L

i

). The total Chern class c(E) is then (1 +t

1

) ··· (1 +t

n

) = 1 +σ

1

+

···+σ

n

, where σ

j

= c

j

(E) is the j

th

elementary symmetric polynomial in the t

i

’s, the

sum of all products of j distinct t

i

’s. As we saw in §2.3, the Newton polynomials s

k

satisfy t

k

1

+···+t

k

n

= s

k

(σ

1

, ··· , σ

k

). Since σ

j

= c

j

(E), this means that the preceding

displayed formula can be rewritten

ch(E) = dimE +

¸

k>0

s

k

(c

1

(E), ··· , c

k

(E))/k!

The right side of this equation is deﬁned for arbitrary vector bundles E, so we take

this as our general deﬁnition of ch(E).

Lower Bounds on Im J Section 4.1 101

Proposition 4.2. ch(E

1

⊕E

2

) = ch(E

1

) +ch(E

2

) and ch(E

1

⊗E

2

) = ch(E

1

)ch(E

2

).

Proof: The proof of the splitting principle for ordinary cohomology in Proposition

2.3 works with any coefﬁcients in the case of complex vector bundles, in particular

for Q coefﬁcients. By this splitting principle we can pull E

1

back to a sum of line

bundles over a space F(E

1

). By another application of the splitting principle to the

pullback of E

2

over F(E

1

), we have a map F(E

1

, E

2

)→X pulling both E

1

and E

2

back

to sums of line bundles, with the induced map H

∗

(X; Q)→H

∗

(F(E

1

, E

2

); Q) injec-

tive. So to prove the proposition it sufﬁces to verify the two formulas when E

1

and E

2

are sums of line bundles, say E

i

= ⊕

j

L

ij

for i = 1, 2. The sum formula holds since

ch(E

1

⊕E

2

) = ch(⊕

i,j

L

ij

) =

¸

i,j

e

c

1

(L

ij

)

= ch(E

1

) +ch(E

2

), by the discussion preced-

ing the deﬁnition of ch. For the product formula, ch(E

1

⊗E

2

) = ch

¸

⊕

j,k

(L

1j

⊗L

2k

)

=

¸

j,k

ch(L

1j

⊗L

2k

) =

¸

j,k

ch(L

1j

)ch(L

2k

) = ch(E

1

)ch(E

2

). L¦

In view of this proposition, the Chern character automatically extends to a ring

homomorphism ch: K(X)→H

∗

(X; Q). By naturality there is also a reduced form

ch:

`

K(X)→

`

H

∗

(X; Q) since these reduced rings are the kernels of restriction to a

point.

As a ﬁrst calculation of the Chern character, we have:

Proposition 4.3. ch:

`

K(S

2n

)→H

2n

(S

2n

; Q) is injective with image equal to the sub-

group H

2n

(S

2n

; Z) ⊂ H

2n

(S

2n

; Q).

Proof: Since ch(x⊗(H − 1)) = ch(x) ch(H − 1) we have the commutative dia-

gram shown at the right, where the upper map is external

−

−

−

−

−

→

−−−−−→

−−→

X K( )

X H ( ) ;

≈

≈

Q

∗

ch

−

−

−

−

−

→

S X K

2

S X

2

( )

H ( ) ;

Q

∗

ch

tensor product with H − 1, which is an isomorphism by

Bott periodicity, and the lower map is cross product with

ch(H − 1) = ch(H) − ch(1) = 1 + c

1

(H) − 1 = c

1

(H), a

generator of H

2

(S

2

; Z). From Theorem 3.16 of [AT] the lower map is an isomorphism

and restricts to an isomorphism of the Z coefﬁcient subgroups. Taking X = S

2n

, the

result now follows by induction on n, starting with the trivial case n = 0. L¦

An interesting by-product of this is:

Corollary 4.4. A class in H

2n

(S

2n

; Z) occurs as a Chern class c

n

(E) iff it is divisible

by (n−1)! .

Proof: For vector bundles E→S

2n

we have c

1

(E) = ··· = c

n−1

(E) = 0, so ch(E) =

dimE+s

n

(c

1

, ··· , c

n

)/n! = dimE:nc

n

(E)/n! by the recursion relation for s

n

derived

in §2.3, namely, s

n

= σ

1

s

n−1

−σ

2

s

n−2

+··· +(−1)

n−2

σ

n−1

s

1

+(−1)

n−1

nσ

n

. L¦

Even when H

∗

(X; Z) is torsionfree, so that H

∗

(X; Z) is a subring of H

∗

(X; Q),

it is not always true that the image of ch is contained in H

∗

(X; Z). For example, if

102 Chapter 4 The J–Homomorphism

L ∈ K(CP

n

) is the canonical line bundle, then ch(L) = 1+c+c

2

/2+···+c

n

/n! where

c = c

1

(L) generates H

2

(CP

n

; Z), hence c

k

generates H

2k

(CP

n

; Z) for k ≤ n.

The Chern character can be used to show that for ﬁnite cell complexes X, the

only possible differences between the groups K

∗

(X) and H

∗

(X; Z) lie in their tor-

sion subgroups. Since these are ﬁnitely generated abelian groups, this will follow if

we can show that K

∗

(X)⊗Q and H

∗

(X; Q) are isomorphic. Thus far we have de-

ﬁned the Chern character K

0

(X)→H

even

(X; Q), and

it is easy to extend this formally to odd dimensions by

the commutative diagram at the right.

−−−−−→

−−−→

SX K ( ) SX H ( ) ;

Q

even

ch

X H ( ) ; X K ( ) Q

odd

ch

1

0

=

≈

Proposition 4.5. The map K

∗

(X)⊗Q→H

∗

(X; Q) induced by the Chern character

is an isomorphism for all ﬁnite cell complexes X.

Proof: We proceed by induction on the number of cells of X. The result is triv-

ially true when there is a single cell, a 0 cell, and it is also true when there are two

cells, so that X is a sphere, by the preceding proposition. For the induction step,

let X be obtained from a subcomplex A by attaching a cell. Consider the ﬁve-term

sequence X/A→SA→SX→SX/SA→S

2

A. Applying the rationalized Chern charac-

ter K

∗

(−)⊗Q→H

∗

(−; Q) then gives a commutative diagram of ﬁve-term exact se-

quences since tensoring with Q preserves exactness. The spaces X/A and SX/SA

are spheres. Both SA and S

2

A are homotopy equivalent to cell complexes with the

same number of cells as A, by collapsing the suspension or double suspension of

a 0 cell. Thus by induction four of the ﬁve maps between the two exact sequences

are isomorphisms, all except the map K

∗

(SX)⊗Q→H

∗

(SX; Q), so by the ﬁve-lemma

this map is an isomorphism as well. Finally, to obtain the result for X itself we may

replace X by S

2

X since the Chern character commutes with double suspension, as

we have seen, and a double suspension is in particular a single suspension, with the

same number of cells, up to homotopy equivalence. L¦

The e Invariant

Now let us deﬁne the main object we will be studying in this section, the homo-

morphism e : π

2m−1

(S

2n

)→Q/Z. For a map f : S

2m−1

→S

2n

we have the mapping

cone C

f

obtained by attaching a cell e

2m

to S

2n

by f . The quotient C

f

/S

2n

is S

2m

so we have a commutative diagram of short exact sequences

−

−

−

−

−

→

−−−−−→ −−−−−→ −−−−−→ −−−−−→

−−→ −−→ −−→ −−→

C K( )

C H 0 0 ( ) ;

Q

∗

ch

f

2n

2n

f

−−−−−→

S K 0 0

S

( )

H ( ) ;

Q

∗

ch

2m

2m

−−−−−→

S K

S

( )

H ( ) ;

Q

∗

ch

There are elements α, β ∈

`

K(C

f

) mapping from and to the standard generators

(H −1) ∗··· ∗(H −1) of

`

K(S

2m

) and

`

K(S

2n

), respectively. In a similar way there

Lower Bounds on Im J Section 4.1 103

are elements a, b ∈

`

H

∗

(C

f

; Q) mapping from and to generators of H

2m

(S

2m

; Z) and

H

2n

(S

2n

; Z). After perhaps replacing a and b by their negatives we may assume that

ch(α) = a and ch(β) = b +ra for some r ∈ Q, using Proposition 4.3. The elements

β and b are not uniquely determined but can be varied by adding any integer mul-

tiples of α and a. The effect of such a variation on the formula ch(β) = b + ra is

to change r by an integer, so r is well-deﬁned in the additive group Q/Z, and we

deﬁne e(f) to be this element r ∈ Q/Z. Since f = g implies C

f

= C

g

, we have a

well-deﬁned map e : π

2n−1

(S

2m

)→Q/Z.

Proposition 4.6. e is a homomorphism.

Proof: Let C

f,g

be obtained from S

2n

by attaching two 2m cells by f and g, so

C

f,g

contains both C

f

and C

g

. There is a quotient map q : C

f+g

→C

f,g

collapsing

a sphere S

2m−1

that separates the 2m cell of C

f,g

into a pair of 2m cells. In the

upper row of the commutative diagram at the right we

−

−

−

−

−

→

−−−−−→

−−→

C K( )

C H ( ) ;

Q

∗

∗

ch

q

∗

q

f, f g g

f,

−

−

−

−

−

→

C K

C

( )

H ( ) ;

Q

∗

ch

+

f g g +

have generators α

f

and α

g

mapping to α

f+g

and β

f,g

mapping to β

f+g

, and similarly in the second row with

generators a

f

, a

g

, a

f+g

, b

f,g

, and b

f+g

. By restriction

to the subspaces C

f

and C

g

of C

f,g

we obtain ch(β

f,g

) = b

f,g

+ r

f

a

f

+ r

g

b

g

, so

ch(β

f+g

) = b

f+g

+(r

f

+r

g

)a

f+g

. L¦

There is a commutative diagram involving the double suspension:

−−→

S

S

e e

( )

m

π

+

S ( ) π

2

2

2

m 2 2

Q

n 1 2n 1 +

/

−

−

−

−

−

→

−

−

−

−

−

→

Z

Commutativity follows from the fact that C

S

2

f

= S

2

C

f

and ch commutes with the

double suspension, as we sawin the proof of Proposition 4.3. Fromthe commutativity

of the diagram there is induced a stable e invariant e : π

s

2k−1

→Q/Z for each k.

Theorem4.7. If the map f : S

2k−1

→U(n) represents a generator of π

2k−1

(U), then

e(J

C

f) = :β

k

/k where β

k

is deﬁned via the power series

x/(e

x

−1) =

¸

i

β

i

x

i

/i!

Hence the image of J in π

s

2k−1

has order divisible by the denominator of β

k

/k.

The numbers β

k

are known in number theory as Bernoulli numbers. After proving

the theorem we will show how to compute the denominator of β

k

/k.

Thom Spaces

For a vector bundle E→B with unit disk bundle D(E) and unit sphere bundle

S(E) the quotient space D(E)/S(E) is called the Thomspace T(E). When B is compact

this can also be regarded as the one-point compactiﬁcation of E. Thom spaces arise

in the present context through the following:

104 Chapter 4 The J–Homomorphism

Lemma 4.8. C

Jf

is the Thomspace of the bundle E

f

→S

2k

determined by the clutch-

ing function f : S

2k−1

→U(n).

Proof: By deﬁnition, E

f

is the union of two copies of D

2k

×C

n

with the subspaces

∂D

2k

×C

n

identiﬁed via (x, v) ∼ (x, f

x

(v)). Collapsing the second copy of D

2k

×C

n

to C

n

via projection produces the same vector bundle E

f

, so E

f

can also be obtained

from D

2k

×C

n

IC

n

by the identiﬁcation (x, v) ∼ f

x

(v) for x ∈ ∂D

2k

. Restricting to

the unit disk bundle D(E

f

), we have D(E

f

) expressed as a quotient of D

2k

×D

2n

I

D

2n

0

by the same identiﬁcation relation, where the subscript 0 labels this particular

disk ﬁber of D(E

f

). In the quotient T(E

f

) = D(E

f

)/S(E

f

) we then have the sphere

S

2n

= D

2n

0

/∂D

2n

0

, and T(E

f

) is obtained from this S

2n

by attaching a cell e

2k+2n

with characteristic map the quotient map D

2k

×D

2n

→D(E

f

)→T(E

f

). The attaching

map of this cell is precisely Jf , since on ∂D

2k

×D

2n

it is given by (x, v)

f

x

(v) ∈

D

2n

/∂D

2n

and all of D

2k

×∂D

2n

maps to the point ∂D

2n

/∂D

2n

. L¦

In order to compute eJ

C

(f) we need to compute ch(β) where β ∈

`

K(C

Jf

) =

`

K(T(E

f

)) restricts to a generator of

`

K(S

2n

). The S

2n

here is D

2n

0

/∂D

2n

0

for a ﬁber

D

2n

0

of D(E

f

). For a general complex vector bundle E→X a class in

`

K(T(E)) that

restricts to a generator of

`

K(S

n

) for each sphere S

n

coming froma ﬁber of E is called

a Thom class for the bundle.

Let us show how to ﬁnd a Thom class U ∈

`

K(T(E)) for a complex vector bundle

E→X with X compact Hausdorff. We may view T(E) as the quotient P(E⊕1)/P(E)

since in each ﬁber C

n

of E we obtain P(C

n

⊕C) = CP

n

from P(C

n

) = CP

n−1

by attach-

ing the 2n cell C

n

×¦1¦, so the quotient P(C

n

⊕C)/P(C

n

) is S

2n

, which is the part of

T(E) coming fromthis ﬁber C

n

. FromExample 2.24 we knowthat K

∗

(P(E⊕1)) is the

free K

∗

(X) module with basis 1, L, ··· , L

n

, where L is the canonical line bundle over

P(E⊕1). Restricting to P(E) ⊂ P(E⊕1), K

∗

(P(E)) is the free K

∗

(X) module with

basis the restrictions of 1, L, ··· , L

n−1

to P(E). So we have a short exact sequence

0 →

`

K

∗

(T(E)) →K

∗

(P(E⊕1))

ρ

→K

∗

(P(E)) →0

and Ker ρ must be generated as a K

∗

(X) module by some polynomial of the form

L

n

+ a

n−1

L

n−1

+ ··· + a

0

1 with coefﬁcients a

i

∈ K

∗

(X), namely the polynomial

¸

i

(−1)

i

λ

i

(E)L

n−i

in Proposition 2.25, regarded now as an element of K(P(E⊕1)).

The class U ∈

`

K(T(E)) mapping to

¸

i

(−1)

i

λ

i

(E)L

n−i

is the desired Thom class

since when we restrict over a point of X the preceding considerations still apply, so

the kernel of K(CP

n

)→K(CP

n−1

) is generated by the restriction of

¸

i

(−1)

i

λ

i

(E)L

n−i

to a ﬁber.

Observe that

`

K

∗

(T(E)) is a free K

∗

(X) module with the single basis element U

since K

∗

(P(E⊕1)) is a free K

∗

(X) module with basis 1, L, ··· , L

n−1

, U. In particular

we have an isomorphism

`

K

∗

(T(E)) ≈ K

∗

(X), known as the Thom isomorphism.

Similar constructions can be made for ordinary cohomology. The deﬁning relation

for H

∗

(P(E)) as H

∗

(X) module has the form

¸

i

(−1)

i

c

i

(E)x

n−i

= 0 where x =

Lower Bounds on Im J Section 4.1 105

x(E) ∈ H

2

(P(E)) restricts to a generator of H

2

(CP

n−1

) in each ﬁber. Viewed as an

element of H

∗

(P(E⊕1)), the element

¸

i

(−1)

i

c

i

(E)x

n−i

, with x = x(E⊕1) now,

generates the kernel of the map to H

∗

(P(E)) since the coefﬁcient of x

n

is 1. So

¸

i

(−1)

i

c

i

(E)x

n−i

∈ H

∗

(P(E⊕1)) is the image of a Thom class u ∈ H

2n

(T(E)). For

future reference we note two facts:

(1) x = c

1

(L) ∈ H

∗

(P(E⊕1)), since the deﬁning relation for c

1

(L) is x(L)−c

1

(L) = 0

and P(L) = P(E⊕1), the bundle L→E⊕1 being a line bundle, so x(E⊕1) =

x(L).

(2) If we identify u with

¸

i

(−1)

i

c

i

(E)x

n−i

∈ H

∗

(P(E⊕1)), then xu = 0 since the

deﬁning relation for H

∗

(P(E⊕1)) is

¸

i

(−1)

i

c

i

(E⊕1)x

n+1−i

= 0 and c

i

(E⊕1) =

c

i

(E).

For convenience we shall also identify U with

¸

i

(−1)

i

λ

i

(E)L

n−i

∈ K(P(E⊕1)).

We are omitting notation for pullbacks, so in particular we are viewing E as already

pulled back over P(E⊕1). By the splitting principle we can pull this bundle E back

further to a sum

¸

i

L

i

of line bundles over a space F(E) and work in the cohomology

and K–theory of F(E). The Thom class u =

¸

i

(−1)

i

c

i

(E)x

n−i

then factors as a

product

¸

i

(x −x

i

) where x

i

= c

1

(L

i

), since c

i

(E) is the i

th

elementary symmetric

function σ

i

of x

1

, ··· , x

n

. Similarly, for the K–theory Thom class U we have U =

¸

i

(−1)

i

λ

i

(E)L

n−i

= L

n

λ

t

(E) = L

n

¸

i

λ

t

(L

i

) = L

n

¸

i

(1 + L

i

t) for t = −L

−1

, so U =

¸

i

(L −L

i

). Therefore we have

ch(U) =

¸

i

ch(L −L

i

) =

¸

i

(e

x

−e

x

i

) = u

¸

i

[(e

x

i

−e

x

)/(x

i

−x)]

This last expression can be simpliﬁed to u

¸

i

[(e

x

i

− 1)/x

i

] since after writing it as

u

¸

i

e

x

i

¸

i

[(1−e

x−x

i

)/(x

i

−x)] and expanding the last product out as a multivariable

power series in x and the x

i

’s we see that because of the factor u in front and the

relation xu = 0 noted earlier in (2) all the terms containing x can be deleted, or what

amounts to the same thing, we can set x = 0.

Since the Thom isomorphism Φ: H

∗

(X)→H

∗

(D(E), S(E)) ≈

`

H

∗

(T(E)) is given

by cup product with the Thom class u, the result of the preceding calculation can be

written as Φ

−1

ch(U) =

¸

i

[(e

x

i

−1)/x

i

]. When dealing with products such as this it

is often convenient to take logarithms. There is a power series for log[(e

y

− 1)/y]

of the form

¸

j

α

j

y

j

/j! since the function (e

y

−1)/y has a nonzero value at y = 0.

Then we have

log Φ

−1

ch(U) = log

¸

i

[(e

x

i

−1)/x

i

] =

¸

i

log[(e

x

i

−1)/x

i

] =

¸

i,j

α

j

x

j

i

/j!

=

¸

j

α

j

ch

j

(E)

where ch

j

(E) is the component of ch(E) in dimension 2j . Thus we have the general

formula log Φ

−1

ch(U) =

¸

j

α

j

ch

j

(E) which no longer involves the splitting of the

bundle E→X into the line bundles L

i

, so by the splitting principle this formula is

valid back in the cohomology of X.

106 Chapter 4 The J–Homomorphism

Proof of 4.7: Let us specialize the preceding to a bundle E

f

→S

2k

with clutching

function f : S

2k−1

→U(n) where the earlier dimension m is replaced now by k. As

described earlier, the class β ∈

`

K(C

Jf

) =

`

K(T(E

f

)) is the Thom class U, up to a sign

which we can make +1 by rechoosing β if necessary. Since ch(U) = ch(β) = b+ra,

we have Φ

−1

ch(U) = 1 + rh where h is a generator of H

2k

(S

2k

). It follows that

log Φ

−1

ch(U) = rh since log(1 +z) = z −z

2

/2 +··· and h

2

= 0. On the other hand,

the general formula log Φ

−1

ch(U) =

¸

j

α

j

ch

j

(E) specializes to log Φ

−1

ch(U) =

α

k

ch

k

(E

f

) in the present case since

`

H

2j

(S

2k

; Q) = 0 for j ≠ k. If f represents a

suitable choice of generator of π

2k−1

(U(n)) then ch

k

(E

f

) = h by Proposition 4.3.

Comparing the two calculations of log Φ

−1

ch(U), we obtain r = α

k

. Since e(J

C

f)

was deﬁned to be r , we conclude that e(J

C

f) = α

k

for f representing a generator of

π

2k−1

(U(n)).

To relate α

k

to Bernoulli numbers β

k

we differentiate both sides of the equation

¸

k

α

k

x

k

/k! = log[(e

x

−1)/x] = log(e

x

−1) −log x, obtaining

¸

k≥1

α

k

x

k−1

/(k −1)! = e

x

/(e

x

−1) −x

−1

= 1 +(e

x

−1)

−1

−x

−1

= 1 −x

−1

+

¸

k≥0

β

k

x

k−1

/k!

= 1 +

¸

k≥1

β

k

x

k−1

/k!

where the last equality uses the fact that β

0

= 1, which comes from the formula

x/(e

x

−1) =

¸

i

β

i

x

i

/i! . Thus we obtain α

k

= β

k

/k for k > 1 and 1 +β

1

= α

1

. It is

not hard to compute that β

1

= −1/2, so α

1

= 1/2 and α

k

= −β

k

/k when k = 1. L¦

Bernoulli Denominators

The numbers β

k

are zero for odd k > 1 since the function x/(e

x

−1)−1+x/2 =

¸

i≥2

β

i

x

i

/i! is even, as a routine calculation shows. Determining the denominator of

β

k

/k for even k is our next goal since this tells us the order of the image of eJ

C

in

these cases.

Theorem 4.9. For even k > 0 the denominator of β

k

/k is the product of the prime

powers p

+1

such that p −1 divides k and p

**is the highest power of p dividing k.
**

More precisely:

(1) The denominator of β

k

is the product of all the distinct primes p such that p−1

divides k.

(2) A prime divides the denominator of β

k

/k iff it divides the denominator of β

k

.

The ﬁrst step in proving the theoremis to relate Bernoulli numbers to the numbers

S

k

(n) = 1

k

+2

k

+··· +(n−1)

k

.

Proposition 4.11. S

k

(n) =

¸

k

i=0

k

i

β

k−i

n

i+1

/(i +1).

Proof: The function f(t) = 1+e

t

+e

2t

+··· +e

(n−1)t

has the power series expansion

¸

n−1

=0

¸

∞

k=0

k

t

k

/k! =

¸

∞

k=0

S

k

(n)t

k

/k!

Lower Bounds on Im J Section 4.1 107

On the other hand, f(t) can be expressed as the product of (e

nt

−1)/t and t/(e

t

−1),

with power series

¸

∞

i=1

n

i

t

i−1

/i!

¸

∞

j=0

β

j

t

j

/j! =

¸

∞

i=0

n

i+1

t

i

/(i +1)!

¸

∞

j=0

β

j

t

j

/j!

Equating the coefﬁcients of t

k

we get

S

k

(n)/k! =

¸

k

i=0

n

i+1

β

k−i

/(i +1)!(k −i)!

Multiplying both sides of this equation by k! gives the result. L¦

Proof of 4.9: We will be interested in the formula for S

k

(n) when n is a prime p:

(∗) S

k

(p) = β

k

p +

k

1

β

k−1

p

2

/2 +··· +β

0

p

k+1

/(k +1)

Let Z

(p)

⊂ Q be the ring of p integers, that is, rational numbers whose denominators

are relatively prime to p. We will ﬁrst apply (∗) to prove that pβ

k

is a p integer

for all primes p. This is equivalent to saying that the denominator of β

k

contains no

square factors. By induction on k, we may assume pβ

k−i

is a p integer for i > 0.

Also, p

i

/(i + 1) is a p integer since p

i

≥ i + 1 by induction on i. So the product

β

k−i

p

i+1

/(i + 1) is a p integer for i > 0. Thus every term except β

k

p in (∗) is a

p integer, and hence β

k

p is a p integer as well.

Next we show that for even k, pβ

k

≡ S

k

(p) mod p in Z

(p)

, that is, the difference

pβ

k

−S

k

(p) is p times a p integer. This will also follow from (∗) once we see that

each term after β

k

p is p times a p integer. For i > 1, p

i−1

/(i +1) is a p integer by

induction on i as in the preceding paragraph. Since we know β

k−i

p is a p integer, it

follows that each term in (∗) containing a β

k−i

with i > 1 is p times a p integer. As

for the term containing β

k−1

, this is zero if k is even and greater than 2. For k = 2,

this term is 2(−1/2)p

2

/2 = −p

2

/2, which is p times a p integer.

Now we assert that S

k

(p) ≡ −1 mod p if p−1 divides k, while S

k

(p) ≡ 0 mod p

in the opposite case. In the ﬁrst case we have

S

k

(p) = 1

k

+··· +(p −1)

k

≡ 1 +··· +1 = p −1 ≡ −1 mod p

since the multiplicative group Z

∗

p

= Z

p

−¦0¦ has order p−1 and p−1 divides k. For

the second case we use the elementary fact that Z

∗

p

is a cyclic group. (If it were not

cyclic, there would exist an exponent n < p −1 such that the equation x

n

−1 would

have p−1 roots in Z

p

, but a polynomial with coefﬁcients in a ﬁeld cannot have more

roots than its degree.) Let g be a generator of Z

∗

p

, so ¦1, g

1

, g

2

, ··· , g

p−2

¦ = Z

∗

p

. Then

S

k

(p) = 1

k

+··· +(p −1)

k

= 1

k

+g

k

+g

2k

+··· +g

(p−2)k

and hence (g

k

−1)S

k

(p) = g

(p−1)k

−1 = 0 since g

p−1

= 1. If p −1 does not divide

k then g

k

≠ 1, so we must have S

k

(p) ≡ 0 mod p.

Statement (1) of the theorem now follows since if p − 1 does not divide k then

pβ

k

≡ S

k

(p) ≡ 0 mod p so β

k

is p integral, while if p −1 does divide k then pβ

k

≡

S

k

(p) ≡ −1 mod p so β

k

is not p integral and p divides the denominator of β

k

.

To prove statement (2) of the theorem we will use the following fact:

108 Chapter 4 The J–Homomorphism

Lemma 4.12. For all n ∈ Z, n

k

(n

k

−1)β

k

/k is an integer.

Proof: Recall the function f(t) = (e

nt

− 1)/(e

t

− 1) considered earlier. This has

logarithmic derivative

f

¹

(t)/f(t) = (log f(t))

¹

= [log(e

nt

−1) −log(e

t

−1)]

¹

= ne

nt

/(e

nt

−1) −e

t

/(e

t

−1)

We have

e

x

/(e

x

−1) = 1/(1 −e

−x

) = x

−1

[−x/(e

−x

−1)] =

¸

∞

k=0

(−1)

k

β

k

x

k−1

/k!

So

f

¹

(t)/f(t) =

¸

∞

k=1

(−1)

k

(n

k

−1)β

k

t

k−1

/k!

where the summation starts with k = 1 since the k = 0 term is zero. The (k − 1)

st

derivative of this power series at 0 is :(n

k

−1)β

k

/k. On the other hand, the (k−1)

st

derivative of f

¹

(t)(f(t))

−1

is (f(t))

−k

times a polynomial in f(t) and its derivatives,

with integer coefﬁcients, as one can readily see by induction on k. From the formula

f(t) =

¸

k≥0

S

k

(n)t

k

/k! derived earlier, we have f

(i)

(0) = S

i

(n), an integer. So the

(k − 1)

st

derivative of f

¹

(t)/f(t) at 0 has the form m/f(0)

k

= m/n

k

for some

m∈ Z. Thus (n

k

−1)β

k

/k = :m/n

k

and so n

k

(n

k

−1)β

k

/k is an integer. L¦

Statement (2) of the theorem can now be proved. If p divides the denominator

of β

k

then obviously p divides the denominator of β

k

/k. Conversely, if p does not

divide the denominator of β

k

, then by statement (1), p −1 does not divide k. Let g

be a generator of Z

∗

p

as before, so g

k

is not congruent to 1 mod p. Then p does not

divide g

k

(g

k

−1), hence β

k

/k is the integer g

k

(g

k

−1)β

k

/k divided by the number

g

k

(g

k

− 1) which is relatively prime to p, so p does not divide the denominator of

β

k

/k.

The ﬁrst statement of the theorem follows immediately from (1) and (2). L¦

There is an alternative deﬁnition of e purely in terms of K–theory and the Adams

operations ψ

k

. By the argument in the proof of Theorem 2.17 there are formulas

ψ

k

(α) = k

m

α and ψ

k

(β) = k

n

β + µ

k

α for some µ

k

∈ Z satisfying µ

k

/(k

m

− k

n

) =

µ

/(

m

−

n

). The rational number µ

k

/(k

m

− k

n

) is therefore independent of k. It

is easy to check that replacing β by β + pα for p ∈ Z adds p to µ

k

/(k

m

− k

n

), so

µ

k

/(k

m

−k

n

) is well-deﬁned in Q/Z.

Proposition 4.13. e(f) = µ

k

/(k

m

−k

n

) in Q/Z.

Proof: This follows by computing chψ

k

(β) in two ways. First, from the formula

for ψ

k

(β) we have chψ

k

(β) = k

n

ch(β) + µ

k

ch(α) = k

n

b + (k

n

r + µ

k

)a. On the

other hand, there is a general formula ch

q

ψ

k

(ξ) = k

q

ch

q

(ξ) where ch

q

denotes the

component of ch in H

2q

. To prove this formula it sufﬁces by the splitting principle

and additivity to take ξ to be a line bundle, so ψ

k

(ξ) = ξ

k

, hence

ch

q

ψ

k

(ξ) = ch

q

(ξ

k

) = [c

1

(ξ

k

)]

q

/q! = [kc

1

(ξ)]

q

/q! = k

q

c

1

(ξ)

q

/q! = k

q

ch

q

(ξ)

Lower Bounds on Im J Section 4.1 109

In the case at hand this says ch

m

ψ

k

(β) = k

m

ch

m

(β) = k

m

ra. Comparing this with

the coefﬁcient of a in the ﬁrst formula for chψ

k

(β) gives µ

k

= r(k

m

−k

n

). L¦

Books

M. F. Atiyah, K–Theory, W. A. Benjamin, 1967.

R. Bott, Lectures on K(X), W. A. Benjamin, 1969.

J. Dieudonn´e, A History of Algebraic and Differential Topology 1900-1960, Birkh¨auser, 1989.

W. Fulton, Young Tableaux, Cambridge Univ. Press, 1997.

H. Hiller, Geometry of Coxeter Groups, Pitman, 1982.

D. Husemoller, Fibre Bundles, McGraw-Hill, 1966 (later editions by Springer-Verlag).

M. Karoubi, K–Theory: An Introduction, Springer-Verlag, 1978.

S. Lang, Algebra, Addison-Wesley, 1965.

J. W. Milnor and J. D. Stasheff, Characteristic Classes, Princeton Univ. Press, 1974.

N. Steenrod, Topology of Fiber Bundles, Princeton Univ. Press, 1951.

R. Stong, Notes on Cobordism Theory, Princeton Univ. Press, 1968.

B. L. van der Waerden, Modern Algebra, Ungar, 1949.

Papers

J. F. Adams, On the non-existence of elements of Hopf invariant one, Ann. of Math. 72 (1960),

20–104.

J. F. Adams, Vector ﬁelds on spheres, Ann. of Math. 75 (1962), 603–632.

J. F. Adams, On the groups J(X) IV, Topology 5 (1966), 21–71.

J. F. Adams and M. F. Atiyah, K–theory and the Hopf invariant, Quart. J. Math. 17 (1966),

31–38.

M. Atiyah and F. Hirzebruch, Vector Bundles and Homogeneous Spaces, Proc. Sym. Pure

Math.. III (1961), 7–38.

M. Atiyah, Vector bundles and the K¨ unneth formula, Topology 1 (1962), 245–248.

M. Atiyah, Power Operations in K–Theory, Quart. J. Math. Oxford. 17 (1966), 165–193.

M. Atiyah, K–Theory and reality, Quart. J. Math. Oxford. 17 (1966), 367–386.

M. Atiyah and R. Bott, On the periodicity theorem for complex vector bundles, Acta Math.

112 (1964), 229–247.

M. Atiyah, R. Bott and A. Shapiro, Clifford Modules, Topology 3 (1964), 3–38.

R. Bott and J. Milnor, On the parallelizability of spheres, Bull. A.M.S. 64 (1958), 87–89.

R. Bott, The stable homotopy of the classical groups, Ann. of Math. 70 (1959), 313–337.

L. Hodgkin, On the K–Theory of Lie groups, Topology. 6 (1967), 1–36.

M. Mahowald, The order of the image of the J–homomorphism, Bull. A.M.S. 76 (1970), 1310–

1313.

Bibliography 111

J. Milnor, Some consequences of a theorem of Bott, Ann. of Math. 68 (1958), 444–449.

D. Quillen, The Adams conjecture, Topology 10 (1971), 67–80.

D. Sullivan, Genetics of homotopy theory and the Adams conjecture, Ann. of Math. 100 (1974),

1–79.

L. M. Woodward, Vector ﬁelds on spheres and a generalization, Quart. J. Math. Oxford. 24

(1973), 357–366.

Table of Contents

Introduction

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 . . . . . . . . . . . . . . . . . . . . 4 . . . . . . . . . . . . 6

Chapter 1. Vector Bundles

**1.1. Basic Deﬁnitions and Constructions
**

Associated Fiber Bundles 15.

Sections 7. Direct Sums 9. Inner Products 11. Tensor Products 13.

1.2. Classifying Vector Bundles

. . . . . . . . . . . . . . . . .

18

Pullback Bundles 18. Clutching Functions 22. The Universal Bundle 27. Cell Structures on Grassmannians 31. Appendix: Paracompactness 35.

Chapter 2. K–Theory

2.1. The Functor K(X) 2.2. Bott Periodicity

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

38 39

Ring Structure 40. The Fundamental Product Theorem 41. . . . . . . . . . . . . . . . . . . . . . . . . 51

Exact Sequences 51. Deducing Periodicity from the Product Theorem 53. Extending to a Cohomology Theory 55. Elementary Applications 58.

**2.3. Division Algebras and Parallelizable Spheres 2.4. Bott Periodicity in the Real Case 2.5. Vector Fields on Spheres
**

[not yet written]

. . . . . .

59

H–Spaces 59. Adams Operations 62. The Splitting Principle 65.

[not yet written]

Chapter 3. Characteristic Classes

. . . . . . . . . . . . . . . . . . . . . . . . . .

73 77

3.1. Stiefel-Whitney and Chern Classes 3.2. Euler and Pontryagin Classes

Axioms and Constructions 77. Cohomology of Grassmannians 84. . . . . . . . . . . . . . . . . 88

The Euler Class 91. Pontryagin Classes 94.

**Chapter 4. The J–Homomorphism
**

4.1. Lower Bounds on Im J

Bernoulli Denominators 106.

. . . . . . . . . . . . . .

98 99

. . . . . . . . . . . . . . . . . . . .

The Chern Character 100. The e Invariant 102. Thom Spaces 103.

4.2. Upper Bounds on Im J

[not yet written]

Bibliography

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

with an exposition that should be accessible to bright undergraduates familiar with standard material in linear algebra. much simpler than the original proofs using ordinary homology and cohomology. abstract algebra. Some of the best-known applications of algebraic topology in the twentieth century. have relatively elementary proofs using K–theory. which gives the ﬁrst level of depth in the stable homotopy groups of spheres. based on the Periodicity Theorem of Bott proved just a few years earlier. A concrete goal of the later chapters is to tell the full story on the stable J–homomorphism. such as the theorem of Bott and Milnor that there are no division algebras after the Cayley octonions. The ﬁrst portion of this book takes these theorems as its goals. Along the way various other topics related to vector bundles that are of interest independent of K–theory are also developed. approximately the contents of a standard graduate course in algebraic topology. and Pontryagin. and it is also more powerful for certain purposes.Preface Topological K–theory. . the ﬁrst generalized cohomology theory to be studied thoroughly. Later chapters of the book assume more. was introduced around 1960 by Atiyah and Hirzebruch. such as the characteristic classes associated to the names Stiefel and Whitney. In some respects K–theory is more elementary than classical homology and cohomology. and point-set topology. or Adams’ theorem determining the maximum number of linearly independent tangent vector ﬁelds on a sphere of arbitrary dimension. Chern.

and the values of KO(S n ) are given by the following table: n mod 8 KO(S ) n 1 2 3 4 Z 5 0 6 0 7 0 8 Z Z2 Z2 0 For example. As it happens. or one can look for a cruder classiﬁcation. perhaps different n ’s for the two given vector bundles. but the classiﬁcation of all the different vector o bundles over a given base space with ﬁber of a given dimension is quite difﬁcult in general. called the base space of the vector bundle.Introduction Everyone is familiar with the M¨bius band. Then it turns out that the set of stable isomorphism classes of vector bundles over X forms an abelian group under the direct sum operation. This is called Bott Periodicity. KO(S 1 ) is Z2 . as contrasted with an annulus which is the actual product of a circle and a line. Using this. using a weaker equivalence relation than the natural notion of isomorphism for vector bundles. Vector bundles are the natural generalization of the M¨bius band and annulus. and a generator for this group is the M¨bius bundle. there are two directions one can take to make some partial progress on the problem. so let us discuss this ﬁrst. The traditional notation for this group is KO(X) . called the ﬁber of the vector bundle. which in each ﬁber reduces just to direct sum of vector spaces. This has order two since the direct sum of two copies of the o . There is a natural direct sum operation for vector bundles over a ﬁxed base space X . a cyclic group of order two. For example. In the absence of a full classiﬁcation of all the different vector bundles over a given base space. The only two vector bundles with base space a circle and one-dimensional ﬁber are the M¨bius band and the annulus. and the study of vector bundles could be called Linear Algebraic Topology. the twisted product of a circle and a o line. with o the circle replaced by an arbitrary topological space. In the case of spheres the groups KO(S n ) have the quite unexpected property of being periodic in n . and the line replaced by a vector space of arbitrary ﬁnite dimension. when the base space is a high-dimensional sphere and the dimension of the ﬁber is at least three. Vector bundles thus combine topology with linear algebra. then the classiﬁcation is of the same order of difﬁculty as the fundamental but still largely unsolved problem of computing the homotopy groups of spheres. One can either look for invariants to distinguish at least some of the different vector bundles. the latter approach is more elementary in terms of prerequisites. at least if X is compact Hausdorff. one can obtain a weaker notion of isomorphism of vector bundles by deﬁning two vector bundles over the same base space X to be stably isomorphic if they become isomorphic after direct sum with product vector bundles X × Rn for some n .

n . Using a more general form of Bott periodicity. which in turn give product operations in both real and complex K–theory similar to cup product in ordinary cohomology. S 3 . so the period is two rather than eight. The ﬁrst such invariant is orientability. Clifford algebras also provide a nice explanation for the mysterious sequence of groups appearing the in real form of Bott periodicity. in some ways more powerful even than ordinary cohomology. Now let us return to the original classiﬁcation problem for vector bundles over a given base space and the question of ﬁnding invariants to distinguish different vector bundles. and S 7 . the M¨bius bundle is not orientable since o as one goes all the way around the base circle. and the harder part is to show there can be no more than this construction provides. which is an orientable vector bundle. however. is constructed in the same way as KO(X) but using vector bundles whose ﬁbers are vector spaces over C rather than R . of the theorem that there are no ﬁnite dimensional division algebras over R in dimensions other than 1 . and 8 . once the basic machinery of complex K–theory has been set up. There is more algebraic structure here than just the additive group structure. The groups K(X) and KO(X) for varying X share certain formal properties with the cohomology groups studied in classical algebraic topology. it is in fact possible to extend the groups K(X) and KO(X) to a full cohomology theory. The complex version of KO(X) . respectively. called K(X) . and octonions. 2 . and the Cayley octonions. 4 . The complex form of Bott Periodicity asserts simply that K(S n ) is Z for n even and 0 for n odd. Things become simpler if one passes from the real vector spaces to complex vector spaces. With all this extra structure.2 Introduction M¨bius bundle is the product S 1 × R2 . the orientation of the ﬁber lines is reversed. the question of whether all the ﬁbers can be coherently oriented. Another classical problem that can be solved more easily using K–theory than ordinary cohomology is to ﬁnd the maximum number of linearly independent tangent vector ﬁelds on the sphere S n . For example. There is an algebraic construction of the requisite number of vector ﬁelds using Clifford algebras. K–theory becomes a powerful tool. The prime example of this is the very simple proof. In this case complex K–theory is not enough. families of abelian groups K n (X) and KO (X) for n ∈ Z that are periodic in n of period two and eight. exterior powers of vector spaces give natural operations within K–theory. as one can see by embedding two M¨bius bands o o in a solid torus so that they intersect orthogonally along the common core circle of both bands. This does not happen for the annulus. and the added subtlety of real K–theory is needed. Tensor products of vector spaces give rise to tensor products of vector bundles. the dimensions of the classical examples of the real and complex numbers. which is also the core circle of the solid torus. Furthermore. the quaternions. quaternions. the unit spheres in the complex numbers. The same proof shows also that the only spheres whose tangent bundles are product bundles are S 1 .

but with Z coefﬁcients there are others. they have still proved themselves to be quite useful objects. The next Stiefel-Whitney class measures a more reﬁned sort of orientability called a spin structure. . It turns out that Stiefel-Whitney classes generate all characteristic classes for ordinary cohomology with Z2 coefﬁcients. called Stiefel-Whitney classes. called Pontryagin and Euler classes. Although characteristic classes do not come close to distinguishing all the different vector bundles over a given base space.Introduction 3 Orientability is measured by the ﬁrst of a sequence of cohomology classes associated to a vector bundle. the latter being related to the Euler characteristic. except in a few low dimensional cases. Cohomological invariants of vector bundles such as these with nice general properties are known as characteristic classes. and the higher Stiefel-Whitney classes measure whether the vector bundle looks more and more like a product vector bundle over succesively higher dimensional subspaces of the base space.

Vectors vx ∈ Px are thought of as arrows with their tail at x . One can think of T S 2 as a continuous family of vector spaces parametrized by points of S 2 . Is that what T S 2 really is? More each plane Px to the plane {x}× R2 by a vector space isomorphism. then for each ﬁxed nonzero vector v ∈ R2 the family of vectors vx = h−1 (x. It is a classical theorem in algebraic topology that no such vector ﬁeld exists. The simplest continuous family of 2 dimensional vector spaces parametrized by points of S 2 is of course the product S 2 × R2 . as every nonzero vector in R3 occurs as τ(vx ) for inﬁnitely many x . v) with v orthogonal to x . and secondly it is the disjoint union of all the vector spaces Px for x ∈ S 2 . If we regard a vector vx in Px as a vector in R3 . in fact for all x in a great circle hand. At each point x ∈ S 2 there is a tangent plane Px . so τ −1 (0) is a whole sphere. Thus T S 2 is ﬁrst of all a topological space. namely the subspace consisting of pairs (x. If we had such an h . vx in S 2 . (See §2. τ(vx )) . The association vx function τ : T S 2 vx x τ ( vx ) Px over S 2 . and can be used to →R τ(vx ) deﬁnes a 3 where T S 2 is the set of all tangent vectors vx as x ranges topologize T S 2 as a subspace of S 2 × R3 .To motivate the deﬁnition of a vector bundle let us consider tangent vectors to the unit 2 sphere S 2 in R3 . Moreover τ(0x ) = 0 for all x ∈ S 2 . This function τ is surjective but certainly not injective. On the other (x. is injective. the function T S 2 →S 2 × R3 .2 for a proof using precisely we can ask whether there is a homeomorphism h : T S 2 →S 2 × R2 that takes . then the standard convention in linear algebra would be to identify vx with all its parallel translates. This is a 2 dimensional vector space with the point x as its zero vector 0x . and in particular with the unique translate τ(vx ) having its tail at the origin in R3 . v) would be a continuous ﬁeld of nonzero tangent vectors to S 2 .

Regarding R4 as the quaternions.Basic Deﬁnitions and Constructions Section 1. This is true in fact for any y on the same side of P as x . In this case there is a continuous ﬁeld vx of nonzero tangent vectors to S 1 . t2 . an equivalence is the homeomorphism S 3 × R3 →T S 3 sending (x. having the virtue of easily genmorphism p −1 (U)→U × P that restricts to a linear isomorphism p −1 (y)→{y}× P for each y ∈ U . Dropping down a dimension. A similar construction using Cayley octonions shows that T S 7 is equivalent to S 7 × R7 . It is a rather deep theorem. Thus for y in a suitable neighborhood U of x in S 2 the map (y. Furthermore this homeomorphism has the key property of restricting to a linear isomorphism from Py onto P for each y ∈ U . A convenient way of rephrasing this situation. t3 )) to the translation of the vector t1 ix + t2 jx + t3 kx having its tail at x .1 5 techniques from this book. Although T S n is not usually equivalent to the product S n × Rn . obtained by regarding points x ∈ S 1 as unit complex numbers and letting vx be the translation of the vector ix that has its tail at x . is to let p : T S 2 →S 2 be the map (x. Moving up to S 3 . For a point x ∈ S 2 let P be the translate of the tangent plane Px that passes through the origin. there is a sense in which this is true locally. proved in §2. and is not just a disguised form of the product S 2 × R2 . t) to tvx . eralizing. with {x}× R a tangent vector vy to the orthogonal projection of τ(vy ) onto P is a linear iso(y. one could consider in similar fashion the space T S 1 of tangent vectors to the unit circle S 1 in R2 . This leads to a going to the tangent line at x by a linear isomorphism. vx ) x . Thus T S 1 really is equivalent to the product S 1 × R1 . the space T S 3 of tangent vectors is again equivalent to the product S 3 × R3 . πy (vy )) is a consisting of tangent vectors at points of U and with range the product U × P . and then we have a homeo- .) So T S 2 is genuinely twisted.3. S 3 . Take the case of the 2 sphere for example. vy ) homeomorphism with domain the subspace of T S 2 vx x homeomorphism S 1 × R→T S 1 taking (x. that S 1 . and S 7 are the only spheres whose tangent bundle is equivalent to a product. (t1 . the unit sphere in R4 . For points y ∈ S 2 that are sufﬁciently close to x the map πy : Py →P sending morphism.

on S n . then the projection I × R→I induces a map p : E →S 1 which is a 1 dimensional vector bundle. v) = (y. (4) The normal bundle to S n in Rn+1 . The map p : E →S n where E = { (x. choose any point x ∈ S n and (2) If we let E be the quotient space of I × R under the identiﬁcations (0. Since E is homeomorphic to a M¨bius band with its boundary o circle deleted. v) ∈ S n × Rn+1 | x ⊥ v } and we think of v as a tangent vector to sends (x. Here are some examples of vector bundles: (1) The product or trivial bundle E = B × Rn with p the projection onto the ﬁrst factor. v) = x . Then hx is a local trivialization since πx restricts to an isomorphism of pairs (x. local trivializations hx : p −1 (Ux )→Ux × R Ux . such that the following local triviality condition is satisﬁed: There is a cover of B by open sets Uα for each of which there exists a homeomorphism hα : p −1 (Uα )→Uα × Rn taking p −1 (b) to {b}× Rn by a vecthe vector bundle. πx (v)) where πx is orthogonal projection onto the hyperplane p −1 (y) onto p −1 (x) for each y ∈ Ub . v) ∈ S n × Rn+1 such that v is perpendicular to the tangent plane to S n at p(x. In the next chapter complex vector bundles will play the larger role. or in other words. or line bundle. however. we call this bundle the M¨bius bundle. p −1 (x) . we can . letting the rest of the data be implicit. As in the previous example. a line bundle p : E →S n with E consisting of x . and the vector spaces p −1 (b) are the ﬁbers. An n dimensional vector bundle is a map p : E →B together with a real vector tor space isomorphism for each b ∈ Uα .1 Basic Deﬁnitions and Constructions Throughout the book we use the word map to mean a continuous function. Such an hα is called a local trivialization of let Ux ⊂ S n be the open hemisphere containing x and bounded by the hyperplane through the origin orthogonal to x . Deﬁne hx : p −1 (Ux )→Ux × p −1 (x) ≈ Ux × Rn by hx (y.6 Chapter 1 Vector Bundles 1. t) ∼ (1. The map p : E →S n is again given by can be obtained by orthogonal projection of the ﬁbers p −1 (y) onto p −1 (x) for y ∈ (5) Real projective n space RPn is the space of lines in Rn+1 through the origin. Since each such line intersects the unit sphere S n in a pair of antipodal points. To construct local trivializations. Often one abbreviates terminology by just calling the vector bundle E . v) to x . o (3) The tangent bundle of the unit sphere S n in Rn+1 . We will focus on real vector bundles in this chapter. −t) . We could equally well take C in place of R as the scalar ﬁeld. v = tx for some t ∈ R . obtaining the notion of a complex vector bundle. E is the total space. Usually the complex case is entirely analogous. The space B is called the base space. a vector bundle p : E →S n S n by translating it so that its tail is at the head of x . space structure on p −1 (b) for each b ∈ B .

We use the notation E1 ≈ E2 to indicate that E1 and E2 are isomorphic. the normal bundle of S n in Rn+1 is isomorphic to the product bundle S n × R by the map (x. x) ∼ (π . π ]× R with the two ends {0}× R and {π }× R identiﬁed. the space E ⊥ = { ( .1 7 of RPn × Rn+1 consisting of pairs ( . for eiθ ∈ S 1 and t ∈ R . iteiθ ) As a further example. The tangent bundle to S 1 is also isomorphic (eiθ . for which a set in RP∞ is open iff it intersects each RPn in an open set. to the trivial bundle S 1 × R . v) ∈ RPn × Rn+1 | v ⊥ } . an (n − k) dimensional vector bundle consisting of pairs ( . The topology on this space will be deﬁned precisely in §1. along with a canonical k dimensional vector bundle over it consisting of pairs ( . v) = ⊥ . The topology we use on RP∞ is the weak or direct limit topology. is a vector bundle with ﬁbers the orthogonal subspaces . of dimension n . tx) (x. Thus an isomorphism preserves −1 base space B is a homeomorphism h : E1 →E2 taking each ﬁber p1 (b) to the cor- all the structure of a vector bundle. There is also an inﬁnite-dimensional projective space RP∞ which is the union also regard RPn as the quotient space of S n in which antipodal pairs of points are of the ﬁnite-dimensional projective spaces RPn under the inclusions RP n ⊂ RP n+1 coming from the natural inclusions Rn+1 ⊂ Rn+2 . p( . so the vectors in these lines sweep out a rectangle [0. Namely. −x) since rotating a vector through an angle of π produces its negative. and the union of all these is a canonical line bundle over RP∞ . v) with v ∈ identiﬁed. and p( . v) with v orthogonal to An isomorphism between vector bundles p1 : E1 →B and p2 : E2 →B over the same −1 responding ﬁber p2 (b) by a linear isomorphism. the M¨bius bundle in (2) above is isomorphic to the canono ical line bundle over RP1 ≈ S 1 . t) . via (eiθ . v) where is a point in Gk (Rn ) and v is a vector in . again with the direct limit topology. (6) The canonical line bundle over RPn has an orthogonal complement. t) . the space of all k dimenional planes through the origin in Rn . The projection p : E ⊥ →RPn . Local trivializations work just as in the ﬁnite-dimensional case. The canonical line bundle p : E →RPn has as its total space E the subspace . v) = . .2. RP1 is swept out by a line rotating through an angle of π . The condition s(b) ∈ p −1 (b) can also be written as A section of a vector bundle p : E →B is a map s : B →E assigning to each b ∈ B a . Again local trivializations can be deﬁned by orthogonal projection. The inclusions RP n ⊂ RPn+1 induce corresponding inclusions of canonical line bundles. so isomorphic bundles are often regarded as the same. A natural generalization of RPn is the so-called Grassmann manifold Gk (Rn ) . The identiﬁcation is (0.Basic Deﬁnitions and Constructions Section 1. Local trivializations can be obtained once more by orthogonal projection. This too has an orthogonal complement. Sections vector s(b) in the ﬁber p −1 (b) . For example.

Composing with local trivializations reduces Here gx is an element of the group GLn (R) of invertible linear transformations of −1 n× n matrix. In one direction this is evident since the trivial bundle certainly has such sections and an isomorphism of vector bundles takes linearly independent sections to linearly independent sections. For example. As we shall show in §2. ··· . the zero 1 section whose value is the zero vector in each ﬁber. gx (v)) . This means that if gx is regarded as an depends continuously on x since its entries can be expressed algebraically in terms −1 of the entries of gx . Every vector bundle has a canonical section. the identity map of B . since the trivial bundle obviously has a nonvanishing section. its n2 entries depend continuously on x . so we may restrict to an open set to the case of an isomorphism h : U × Rn →U × Rn of the form h(x. What must be checked is that h−1 is continuous. Lemma 1. Not all vector bundles have such a section. ··· . v) = (x. so the complements of the zero sections in E1 and E2 must be homeomorphic. the map h : B × Rn →E given by h(b. v) = (x. A continuous map h : E1 →E2 between vector bundles over the same Proof: The hypothesis implies that h is one-to-one and onto. gx (v)) is continuous. namely. and an isomorphism between vector bundles takes nonvanishing sections to nonvanishing sections. This is a local question. Here a section is just a tangent vector ﬁeld to S n . Conversely. sn (b) are linearly inde- is a linear isomorphism in each ﬁber. S n has a nonvanishing vector ﬁeld iff n is odd. sn such that the vectors s1 (b). We often identity the zero section with its image. t1 . we can see that the M¨bius bundle is not isomorphic to the product bundle S 1 × R since the complement o of the zero section is connected for the M¨bius bundle but not for the product bundle.2. tn ) = i ti si (b) it has n sections s1 . Consider for example the tangent bundle to S n . ··· . From this it follows that the tangent bundle of S n is not isomorphic to the trivial bundle if n is even and nonzero. .8 Chapter 1 Vector Bundles ps = 1 . Therefore h−1 (x. The inverse matrix gx also Rn . Hence h is an isomorphism by the following useful technical result: −1 base space B is an isomorphism if it takes each ﬁber p1 (b) to the corresponding −1 ﬁber p2 (b) by a linear isomorphism. an n dimensional bundle p : E →B is isomorphic to the trivial bundle iff pendent in each ﬁber p −1 (b) . plement of the zero section since any vector bundle isomorphism h : E1 →E2 must One can sometimes distinguish nonisomorphic bundles by looking at the com- take the zero section of E1 onto the zero section of E2 . if one has n linearly independent sections si . gx is 1/(det gx ) times the classical adjoint matrix of −1 gx .1. In fact. o At the other extreme from the zero section would be a section whose values are all nonzero. and gx depends continuously on x . U ⊂ B over which E1 and E2 are trivial. and is continuous since its composition with a local trivialization p −1 (U)→U × Rn is continuous. a subspace of E which projects homeomorphically onto B by p .

x2 . so the tangent bundle to S 7 is also trivial. Thus multiplication by a quaternion in the unit sphere S 3 is an isometry of H . x3 . x2 . v2 ) ∈ E1 × E2 | p1 (v1 ) = p2 (v2 ) } There is then a projection E1 ⊕ E2 →B sending (v1 . iz.1 9 As an example. This leads to the construction of seven orthogonal tangent vector ﬁelds on the unit sphere S 7 . v2 ) to the point p1 (v1 ) = p2 (v2 ) . S 3 . kz . kj = −i . ij = k . a division algebra structure on R8 . x1 ) for (x1 . −x3 . multiplication of octonions is deﬁned by (z1 . x2 ) (−x2 . −x2 ) (−x4 . x1 . x3 ) (−x3 . x4 ) (x1 . the only spheres with trivial tangent bundle are S 1 . x4 ) . and ik = −j . Thinking of R8 as H× H . x4 ) (x1 . The underlying reason why this works is that quaternion multiplication satisﬁes |zw| = |z||w| . w2 ) = (z1 w1 − w 2 z2 . where | · | is the usual norm of vectors in R4 . if we set iz since iz = −x2 + ix1 . In terms of complex numbers. then the unit sphere is S 3 and we can deﬁne three sections of its tangent bundle by the formulas z z z iz jz kz or or or (x1 . x2 . i. z = x1 + ix2 then this section is z There is an analogous construction using quaternions instead of complex numbers. the tangent bundle to S 1 is trivial because it has the section (x1 . x3 . x2 ) ∈ S 1 . As we shall show in §2. The quaternions 1. −x4 . by Lemma 1. so when we multiply them by an arbitrary unit quaternion z ∈ S 3 we get a new orthonormal basis z. x3 . and S 7 . x2 . x1 ) It is easy to check that the three vectors in the last column are orthogonal to each other and to (x1 . Another way of characterizing the trivial bundle E ≈ B × Rn is to say that there is a such a projection together with the bundle projection E →B gives an isomorphism Direct Sums Given two vector bundles p1 : E1 →B and p2 : E2 →B over the same base space B . x4 . x4 ) (−x2 .1. x3 . jk = i .Basic Deﬁnitions and Constructions Section 1. x2 . x3 . Quaternions have the form z = x1 + ix2 + jx3 + kx4 . ki = j . z2 )(w1 . x2 . The same constructions work for the Cayley octonions. ji = −k . since E ≈ B × Rn . This leads us to deﬁne the direct sum of E1 and E2 as the space E1 ⊕ E2 = { (v1 . and hence the tangent bundle to S 3 is trivial. we would like to create a third vector bundle over B whose ﬁber over each point of B is the direct sum of the ﬁbers of E1 and E2 over this point.3. If we identify H with R4 via the coordinates (x1 . and form a division algebra H via the multiplication rules i2 = j 2 = k2 = −1 . so we have three linearly independent nonvanishing tangent vector ﬁelds on S 3 . z2 w 1 + w2 z1 ) and satisﬁes the key property |zw| = |z||w| . The ﬁbers of this projection are the direct sums of the ﬁbers of E1 and E2 . continuous projection map E →Rn which is a linear isomorphism on each ﬁber. jz. x1 . x4 ) . as we . k form the standard orthonormal basis for R4 . j.

deﬁned in example (6) above. so it follows that the direct sum of the M¨bius o bundle with itself is the trivial bundle. the bundle E ⊥ is isomorphic to E itself by the map that rotates each vector in the plane by 90 degrees. the restriction of the product E1 × E2 over the diagonal B = {(b. Applied to T S n this identiﬁcation yields T RPn . b) ∈ B × B} is exactly E1 ⊕ E2 . Specializing to the case n = 1 . As another example. This is saying that a tangent vector to RPn is equivalent to a pair of antipodal tangent vectors to S n . is isomorphic to the trivial bundle RPn × Rn+1 via the map ( . w) ( . v. a digression we shall skip here. The direct sum of two trivial bundles is again a trivial bundle. o Example: n The tangent bundle of real projective space. with ﬁbers the products p1 (b1 )× p2 (b2 ) . We call this the restriction of E over A . v. So the tangent bundle to S n is stably trivial: it becomes trivial after taking the direct sum with a trivial bundle. For example. Then if E1 and E2 both have the same base space B . and the map (x. giving a decomposition of the product S 1 × R2 as o the direct sum of two M¨bius bundles. embed the M¨bius o bundle in the product bundle S 1 × R2 by taking the line in the ﬁber {θ}× R2 that makes an angle of θ/2 with the x axis. We noted earlier that E is isomorphic o to the M¨bius bundle over S 1 = RP1 . the direct sum of the tangent and normal bundles to S n in Rn+1 is the trivial bundle S n × Rn+1 since elements of the direct sum are triples (x. To see this geometrically. (b) Given vector bundles p1 : E1 →B1 and p2 : E2 →B2 . clearly. the tangent bundle to RPn . tx) S ×R n n+1 (x. although a formal proof would require a signiﬁcant digression on what precisely tangent vectors to a smooth manifold are. tx) ∈ S n × Rn+1 × Rn+1 with x ⊥ v . v. then hα × hβ is a local trivialization for E1 × E2 . but the direct sum of nontrivial bundles can also be trivial. For a relatively painless veriﬁcation of the local triviality condition we make two preliminary observations: (a) Given a vector bundle p : E →B and a subspace A ⊂ B . suppose we factor out by the identiﬁcations (x. then p : p −1 (A)→A is clearly a vector bundle. Starting with the isomor- phism S × Rn+1 ≈ T S n ⊕ NS n . it does split in an interesting way as a direct sum of nontrivial line bundles. v) ∼ (−x. What we will show is that even though the direct sum of T RPn with a trivial line bundle may not be trivial as it is for a sphere.10 Chapter 1 Vector Bundles wanted. v + w) for v ∈ and w ⊥ . −v) on both sides of this isomorphism. where NS n is the normal bundle of S n in Rn+1 . and then the orthogonal lines in the ﬁbers form a second copy of the M¨bius bundle. v + tx) gives an isomorphism of the direct sum bundle with . . the direct sum E ⊕ E ⊥ of the canonical line bundle E →RPn with its orthogonal complement. then p1 × p2 : E1 × E2 →B1 × B2 −1 −1 is also a vector bundle. A moment’s reﬂection shows this to be entirely reasonable. For if we have −1 −1 local trivializations hα : p1 (Uα )→Uα × Rn and hβ : p2 (Uβ )→Uβ × Rm for E1 and E2 .

2 . Similarly using octonions we can see that the direct sum of eight copies of the canonical line bundle over RP7 is trivial. −x) is well-deﬁned in the quotient. and with β pact if it is Hausdorff and every open cover has a subordinate partition of unity. tx) ∼ (−x. −v) can be written as (x. a positive deﬁnite symmetric bilinear form. Now let us consider the identiﬁcation (x. 1 . T RP3 is trivial since the three linearly independent tangent vector ﬁelds on S 3 deﬁned earlier in terms of quaternions pass down to linearly independent tangent vector ﬁelds on the quotient RP3 . Inner Products An inner product on a vector bundle p : E →B is a map . t(−x)) .Basic Deﬁnitions and Constructions Section 1. tx) ∼ (−x. or 4 modulo 8 . The claim is that E is just the canonical line bundle over RPn . let us identify S n × R with NS n by the isomorphism (x. Proposition 1. tx) in NS n . Thus we have the identiﬁcation (x. only ﬁnitely many of the ϕβ being nonzero near each point of X . −t) in S n × R . : E ⊕ E →R which restricts in each ﬁber to an inner product. a ϕβ = 1 .1 11 In the normal bundle NS n the identiﬁcation (x. Thus we have shown that the tangent bundle T RPn is stably isomorphic to the direct sum of n + 1 copies of the canonical line bundle over RPn . −v) in S n × Rn+1 .2. 8 . v) ∼ (−x. In §2. For n = 3. This identiﬁcation respects the coordinate factors of Rn+1 . The quotient is the canonical line bundle over RPn since the identiﬁcations x ∼ −x in the ﬁrst coordinate give lines through the origin in Rn+1 . 7 this gives 2ϕ(n) = 4. the same numbers we obtained from the triviality of T RP3 and T RP7 . tx) . so the quotient is the direct sum of n + 1 copies of the line bundle E over RPn obtained by making the identiﬁcations (x.5 we will determine when the sum of k copies of the canonical line bundle over RPn is at least stably trivial. hence ((−x. This identiﬁcation yields the product bundle RPn × R since the section x (−x. and in the second coordinate there are no identiﬁcations so we have well-deﬁned vectors tx in these lines. To see this. The deﬁnition of paracompactness we are using is that a space X is paracomcollection of maps ϕβ : X →[0. If there were a 16 dimensional division algebra after the octonions then one might expect the sum of 16 copies of the canonical line bundle over RP15 to be trivial. t) ∼ (−x. using Urysohn’s . Hence the direct sum of four copies of the canonical line bundle over RP3 is trivial. t) (−x. 1] each supported in some set of the open cover. −t) (x. However this is not the case since 2ϕ(15) = 27 = 128 . (−t)(−x)) = (−x. The answer turns out to be rather subtle: This happens exactly when k is a multiple of 2ϕ(n) where ϕ(n) is the number of integers i in the range 0 < i ≤ n with i congruent to 0 . Constructing such functions is easy when X is compact Hausdorff. tx) . When n = 3 . v) ∼ (−x. An inner product exists for a vector bundle p : E →B if B is compact Hausdorff or more generally paracompact. for example.

sm . w) To see that ⊥ E0 satisﬁes the local triviality condition for a vector bundle. ··· . Proposition 1. where we also show that certain classes of noncompact spaces are paracompact. In the case of complex vector bundles one can construct Hermitian inner products in the same way. sn are independent at b0 . isomorphic to E via the map (v. izations hα : p −1 (Uα )→Uα × Rn . using Lemma 1. Proof: ⊥ With respect to a chosen inner product on E . note ﬁrst that we may assume E is the product B × Rn since the question is local in B . ⊥ | ⊥ This h carries E0 to U × Rm and E0 to U × Rn−m .3. si (b)) of E local sections of E0 to a set of n independent local sections b by choosing sm+1 . ··· . Since E0 is a vector bundle. Note that when the subbundle E0 is equal to E itself. We will show now that the corresponding result holds for vector bundles over a paracompact base. si (b)) near each point b0 ∈ B . We claim v + w . sm+1 . using the given inner product. Apply the Gram-Schmidt orthogonalization process to s1 . to pull back the standard inner product in Rn to an inner product ·. A vector subbundle of a vector of E in a vector subspace. of dimension m say. then taking the same vectors for all nearby ﬁbers. If this is so. then setting v.1. sm . ··· . it has m independent local sections b (b. and the ﬁrst m of them are h : p −1 (U)→U × Rn with h(b. This is done in the appendix to this chapter. sn ﬁrst in the ﬁber p −1 (b0 ) . sn in each ﬁber. · α Proof: An inner product for p : E →B can be constructed by ﬁrst using local trivialon p −1 (Uα ) . then E0 ⊕ E0 is which in each ﬁber consists of all vectors orthogonal to vectors in E0 . then there is a vector subbundle E0 ⊂ E such that E0 ⊕ E0 ≈ E . The explicit formulas for the Gram-Schmidt process show the si ’s are continuous. they will remain independent for nearby b by continuity of the determinant function. w α(β) where {ϕβ } is a partition of unity with the support of ϕβ contained in Uα(β) . let E0 be the subspace of E ⊥ ⊥ that the natural projection E0 →B is a vector bundle. The sections si allow us to deﬁne a local trivialization ⊥ E0 . In linear algebra one can show that a vector subspace is always a direct summand by taking its orthogonal complement. ··· . since if s1 . the last part of the proof shows that for any vector bundle with an inner product it is always possible to choose . Most spaces that arise naturally in algebraic topology are paracompact. such that the restriction p : E0 →B is a vector bundle. ··· . bundle p : E →B has the natural deﬁnition: a subspace E0 ⊂ E intersecting each ﬁber If E →B is a vector bundle over a paracompact base B and E0 ⊂ E ⊥ ⊥ is a vector subbundle. We may enlarge this set of m independent (b. so h | E0 is a local trivialization of a basis for E0 in each ﬁber. si (b)) equal to the i th standard basis vector of Rn . w = β ϕβ p(v) v. to obtain new sections si .12 Chapter 1 Vector Bundles Lemma. sm+1 .

Then for K–theory is tensor product. Proof: To motivate the construction. then g is a linear injection on each ﬁber. Here is a general result result along these lines: Proposition 1. 1] form an open cover of B . ﬁrst this gives an embedding of E in B × RN as a direct summand. let x ∈ B . so the local trivializations are by isometries.Basic Deﬁnitions and Constructions Section 1. 1] that is 0 outside Ux and nonzero at x . Tensor Products In addition to direct sum. a number of other algebraic constructions with vector spaces can be extended to vector bundles. as a set.1 13 local trivializations that carry the inner product to the standard inner product. one or both of which may be nontrivial. The image of f is a subbundle of the product B × RN since projection of RN onto The map g is the second coordinate of a map f : E →B × RN with ﬁrst coordinate −1 the i th Rn factor gives the second coordinate of a local trivialization over ϕi (0. Letting subcover. 1] . By Urysohn’s Lemma there is a map ϕx : B →[0. We have seen several cases where the sum of two bundles. that E is a subbundle of a trivial bundle B × RN . the various gi ’s the coordinates of a map g : E →RN with RN a product of copies of p . as we shall see in Example 3.6. suppose ﬁrst that the result holds and hence this product with the projection of the product onto RN yields a map E →RN that is a linear injection on each ﬁber. One which is particularly important −1 −1 E1 ⊗ E2 . The . For vector bundles p1 : E1 →B and p2 : E2 →B . Thus we have E isomorphic to a subbundle of B × RN so by preceding proposition there is a complementary subbundle E with E ⊕ E isomorphic to B × RN . By compactness this has a ﬁnite gi : E →Rn by gi (v) = ϕi (p(v))[πi hi (v)] where p is the projection E →B and πi hi −1 πi to Rn . Let the corresponding Ux ’s and ϕx ’s be relabeled Ui and ϕi . is the trivial bundle. 1] . Our strategy will be to reverse the logic here. then showing that Each point x ∈ B has a neighborhood Ux over which E is trivial. The topology on this set is deﬁned in the following way. be the disjoint union of the vector spaces p1 (x) ⊗ p2 (x) for −1 hi : pi (U)→U × Rni for each open set U ⊂ B over which E1 and E2 are trivial. the sets ϕx (0. Choose isomorphisms −1 −1 a topology TU on the set p1 (U ) ⊗ p2 (U ) is deﬁned by letting the ﬁberwise tensor −1 −1 product map h1 ⊗ h2 : p1 (U) ⊗ p2 (U )→U × (Rn1 ⊗ Rn2 ) be a homeomorphism. so if we make is the composition of a local trivialization hi : p −1 (Ui )→Ui × Rn with the projection Rn . Then gi is a linear injection on each ﬁber over ϕi (0. This can fail when B is noncompact.4. Composing the inclusion of E into constructing a map E →RN that is a linear injection on each ﬁber. An example is the canonical line bundle over RP∞ . For each vector bundle E →B with B compact Hausdorff there exists a vector bundle E →B such that E ⊕ E is the trivial bundle. Deﬁne −1 x vary.

hence are identically 1 . since the entries of the matrices g1 (x) ⊗ g2 (x) are the products of the entries of g1 (x) and g2 (x) . Tensor product again makes the complex analog Vect1 (B) of Vect1 (B) into an abelian C classes of one-dimensional vector bundles over B is an abelian group with respect replacing its gluing matrices gβα (x) ∈ GL1 (R) with their inverses. then the set Vect1 (B) of isomorphism to the tensor product operation. The inverse of a line bundle E →B is obtained by dition is preserved since 1× 1 matrices commute. As we shall see in §3. When we replace U by an open sub−1 −1 set V .14 Chapter 1 Vector Bundles topology TU is independent of the choice of the hi ’s since any other choices are obtained by composing with isomorphisms of U × Rni of the form (x. the topology on p1 (V ) ⊗ p2 (V ) induced by TU is the same as the topology analogous isomorphisms of U × (Rn1 ⊗ Rn2 ) whose second coordinates g1 ⊗ g2 are TV since local trivializations over U restrict to local trivializations over V . suppose we have two vector bundles E1 →B and E2 →B . then we can reconstruct E as the quotient space of the disjoint union with hβ h−1 (x. the cohomology group H 1 (B. Hence we get a well-deﬁned topology on E1 ⊗ E2 making it a vector bundle over B . being isometries of R . It is also distributive with respect to direct sum. These tensor product constructions work equally well for complex vector bundles. v) for continuous maps gi : U →GLni (R) . Any collection of ‘gluing functions’ gβα satisfying this condition can be used to construct a vector bundle E →B . The gluing functions for E ⊗ E are the squares of these gβα ’s. We can choose an open cover {Uα } with both E1 and E2 trivial over each Uα . gi (x)(v)) continuous maps U →GLn1 n2 (R) . the trivial line bundle. v) ∈ Uα × Rn be viewed as maps gβα : Uα ∩ Uβ →GLn (R) . It is routine to verify that the tensor product operation for vector bundles over a ﬁxed base space is commutative. In the case of tensor products. If we are given a vector bundle p : E →B and an open cover {Uα } of B with lo) obtained by identifying (x. Z2 ) when B is homotopy equivalent to a CW complex. Then the gluing 1 2 functions for the bundle E1 ⊗ E2 are the tensor product functions gβα ⊗ gβα assigning 1 2 to each x ∈ Uα ∩ Uβ the tensor product of the two matrices gβα (x) and gβα (x) . v) ∈ Uβ × Rn whenever x ∈ Uα ∩ Uβ . These satisfy the ‘cocycle condition’ gγβ gβα = gγα on Uα ∩ Uβ ∩ Uγ . If we give E an inner product. The cocycle con- . taking vectors in ﬁbers of E to vectors in R1 of the same length. There is another way to look at this construction that takes as its point of departure a general method for constructing vector bundles we have not mentioned prevical trivializations hα : p −1 (Uα )→Uα × Rn . i and so obtain gluing functions gβα : Uα ∩ Uβ →GLni (R) for each Ei .1. and has an identity element. we may rescale local trivializations hα to be isometries. so E ⊗ E is the trivial line bundle. Then all the values of the gluing functions gβα are ±1 . The functions hβ h−1 can α α α (Uα × R n ously. the group Vect1 (B) is isomorphic to a standard object in algebraic topology. associative. Thus each element of Vect1 (B) is its own inverse. hence h1 ⊗ h2 changes by composing with (x. If we restrict attention to line bundles.

so we cannot expect E ⊗ E to be trivial. The effect on the gluing maps gβα is to replace them by their complex conjugates as well. If V has p : E →B the procedure follows closely what we did for tensor product.1 15 group. then we have the more . but the vector space structure in the ﬁbers is modiﬁed by redeﬁning scalar multiplication by the rule λ(v) = λv where the right side of this equation means scalar multiplication in E and the left side means scalar multiplication in E . Specializing to line bundles. The hα : p −1 (Uα )→Uα × F taking p −1 (b) to {b}× F for each b ∈ Uα . This is a map p : E →B such that there is a cover of B by open sets Uα for each of which there exists a homeomorphism now several different ways of constructing ﬁber bundles from vector bundles. another construction useful in K–theory is the exterior power λk (E) of a vector bundle E . This implies that local trivializations for E are obtained from local trivializations for E by composing with the coordinatewise conjugation map Cn →Cn in each ﬁber. +1 for an even permutation and −1 for an odd permutation. In §3. then we deﬁne a topology on this set via local trivializations. the notion of the conjugate bundle E →B . not necessarily ±1 . E We may as well mention here another general construction for complex vector is the same as E . Thus conjugate bundles provide inverses in Vect1 (B) . Having an inner product on a vector bundle E . ϕ and ψ depends continuously on ϕ and ψ . the values are complex numbers of norm 1 . lengths of vectors are deﬁned.1 we will show that the group Vect1 (B) is isomorphic to H 2 (B. Z) C when B is homotopy equivalent to a CW complex. bundles E →B . Recall from linear algebra that the exterior power λk (V ) of a vector space V is the quotient of the k fold tensor product V ⊗ ··· ⊗ V by the subspace generated by vectors of the form v1 ⊗ ··· ⊗ vk − sgn(σ )vσ (1) ⊗ ··· ⊗ vσ (k) where σ is a permutation of the subscripts and sgn(σ ) = ±1 is its sign. We will describe spaces and replace the model ﬁber Rn by an arbitrary space F . but after rescaling the gluing functions gβα for a complex line bundle E . The key fact about tensor product which we needed before was that the tensor product ϕ ⊗ ψ of linear transformations is that a linear map ϕ : Rn →Rn induces a linear map λk (ϕ) : λk (Rn )→λk (Rn ) which depends continuously on ϕ .Basic Deﬁnitions and Constructions Section 1. C Besides tensor product of vector bundles. and we can consider the subspace S(E) consisting of the unit spheres in all the ﬁbers. This holds since λk (ϕ) is a quotient map of the k fold tensor product of ϕ with itself. Now to deﬁne λk (E) for a vector bundle Associated Fiber Bundles If we modify the deﬁnition of a vector bundle by dropping all references to vector general notion of a ﬁber bundle with ﬁber F . For exterior powers the analogous fact dimension n then λk (V ) has dimension n k . we then have E ⊗ E isomorphic to the trivial line bundle since its gluing maps have values zz = 1 for z a unit complex number. As a topological space. We ﬁrst form the disjoint union of the exterior powers λk (p −1 (x)) of all the ﬁbers p −1 (x) .

Since each unit vector uniquely determines the line containing it. so these bundle D(E)→B with ﬁbers the disks of vectors of length less than or equal to 1 . as the quotient of the complement of the zero section in E obtained by identifying each nonzero vector with all positive scalar multiples of itself. this quotient is U × RPn−1 . S n . where P (E) is the space of all lines through the origin in all the ﬁbers of E . It follows that D(E) can also be deﬁned without invoking a metric. i. with respect to the projection P (E)→B which sends each line in the ﬁber of E over a point b ∈ B to b . one could form the Stiefel bundle Vk (E)→B . The canonical line bundle E →RPn has as its unit sphere bundle S(E) the space local trivializations restrict to local trivializations for S(E) . where points of Vk (E) are k tuples of orthogonal unit vectors in ﬁbers of E .. S(E) is the same as the space of unit vectors in Rn+1 . so P (E) is a ﬁber bundle over B with ﬁber RPn−1 . The ﬁber of F (E) is thus the ﬂag manifold F (Rn ) consisting of n tuples of orthogonal lines through the origin in Rn . We topologize P (E) as the quotient of the sphere bundle S(E) obtained by factoring out scalar multiplication in each ﬁber. It is possible to describe S(E) without reference to an inner product. The ﬁber of Gk (E) is the Grassmann manifold Gk (Rn ) of k planes through the origin in Rn . The ﬁber of Vk (E) is the Stiefel manifold Vk (Rn ) of orthonormal k frames in Rn . This is the quotient space of Vk (E) obtained by identifying two k frames in a ﬁber if they span the same subspace of the ﬁber. This is the quotient space of S(E)× [0. (2) For an n dimensional vector bundle E →B . Similarly there is a disk . Local triviality follows as in the preceding example. there is the Grassmann bundle Gk (E)→B of k dimensional linear subspaces of ﬁbers of E . so Vk (E) is a subspace of the product of k copies of S(E) . Over a neighborhood U in B where E is a product U × Rn . (3) As a reﬁnement of the last example. (4) Generalizing P (E) . of unit vectors in lines through the origin in Rn+1 . More generally.e. the associated ﬂag bundle F (E)→B has total space F (E) the subspace of the n fold product of P (E) with itself consisting of n tuples of orthogonal lines in ﬁbers of E . Here are some more examples. 1] obtained by identifying two points in S(E)× {0} if they have the same image in B .16 Chapter 1 Vector Bundles natural projection S(E)→B is a ﬁber bundle with sphere ﬁbers since we have observed that local trivializations for E can be chosen to be isometries in each ﬁber. namely as the mapping cylinder of the projection S(E)→B . for any k ≤ n one could take k tuples of orthogonal lines in ﬁbers of E and get a bundle Fk (E)→B . (1) Associated to a vector bundle E →B is the projective bundle P (E)→B .

For a vector bundle E →X with a subbundle E ⊂ E . Show that a vector bundle E →X has k independent sections iff it has a trivial k dimensional subbundle.Basic Deﬁnitions and Constructions Section 1. construct a quotient vector bundle E/E →X . 2. .1 17 Exercises 1. up to isomorphism. Show that the orthogonal complement of a subbundle is independent of the choice of inner product. 3.

18

Chapter 1

Vector Bundles

**1.2 Classifying Vector Bundles
**

As was stated in the Introduction, it is usually a difﬁcult problem to classify all the different vector bundles over a given base space. The goal of this section will be to rephrase the problem in terms of a standard concept of algebraic topology, the idea of homotopy classes of maps. This will allow the problem to be solved directly in a few very simple cases. Using machinery of algebraic topology, other more difﬁcult cases can be handled as well, as is explained in §??. The reformulation in terms of homotopy also offers some conceptual enlightment about the structure of vector bundles. For the reader who is unfamiliar with the notion of homotopy we give the basic deﬁnitions in the Glossary [not yet written], and more details can be found in the introductory chapter of the author’s book Algebraic Topology.

Pullback Bundles

We will denote the set of isomorphism classes of n dimensional real vector bundles over B by Vectn (B) . The complex analogue, when we need it, will be denoted f ∗ : Vect(B)→Vect(A) , in the reverse direction. by Vectn (B) . Our ﬁrst task is to show how a map f : A→B gives rise to a function C Given a map f : A→B and a vector bundle p : E →B , then there

Proposition 1.5.

exists a vector bundle p : E

→A

with a map f : E

→E

taking the ﬁber of E over

each point a ∈ A isomorphically onto the ﬁber of E over f (a) , and such a vector bundle E is unique up to isomorphism. From the uniqueness statement it follows that the isomorphism type of E depends only on the isomorphism type of E since we can compose the map f with an isomorphism of E with another vector bundle over B . Thus we have a function f ∗ : Vect(B)→Vect(A) taking the isomorphism class of E to the isomorphism class of E . Often the vector bundle E is written as f ∗ (E) and called the bundle induced by f , or the pullback of E by f .

Proof:

First we construct an explicit pullback by setting E = { (a, v) ∈ A× E | f (a) =

p(v) } . This subspace of A× E ﬁts into the diagram at the right where p (a, v) = a and f (a, v) = v . Notice that the two compositions f p and pf from E to B are equal since they both since a pair (a, v) to f (a) . The formula f p = pf looks a bit like a commutativity relation, which may explain why the word commutative is used to describe a diagram to another are equal. If we let Γf denote the graph of f , all points (a, f (a)) in A× B , then p factors 1× these two maps is the restriction of the vector bundle 1 p : A× E →A× B over the as the composition E

E

−− E −− −→

f f

∼

− − →

p

diagram like this one in which any two compositions of maps from one point in the

− − →

p

A −− B −− −→

→Γf →A ,

(a, v)

(a, p(v)) = (a, f (a))

a . The ﬁrst of

Classifying Vector Bundles

Section 1.2

19

graph Γf , so it is a vector bundle, and the second map is a homeomorphism, so their composition p : E

→A

is a vector bundle. The map f obviously takes the ﬁber E

over a isomorphically onto the ﬁber of E over f (a) . For the uniqueness statement, we can construct an isomorphism from an arbitrary E satisfying the conditions in the proposition to the particular one just constructed by sending v ∈ E to the pair (p (v ), f (v )) . This map obviously takes each ﬁber of E to the corresponding ﬁber of f ∗ (E) by a vector space isomorphism, so by Lemma 1.1 it is an isomorphism of vector bundles. One can be more explicit about local trivializations in the pullback bundle f ∗ (E) constructed above. If E is trivial over a subspace U ⊂ B then f ∗ (E) is trivial over f −1 (U) since linearly independent sections si of E over U give rise to independent sections a (a, si (f (a))) of f ∗ (E) over f −1 (U ) . In particular, the pullback of a trivial bundle is a trivial bundle. This can also be seen directly from the deﬁnition, which in the case E = B × Rn just says that f ∗ (E) consists of the triples (a, b, v) in A× B × Rn with b = f (a) , so b is redundant and we have just the product A× Rn . Now let us give some examples of pullbacks. (1) The restriction of a vector bundle p : E →B over a subspace A ⊂ B can be viewed as a pullback with respect to the inclusion map A is certainly an isomorphism on each ﬁber. (2) Another very special case is when the map f is a constant map, having image a single point b ∈ B . Then f ∗ (E) is just the product A× p −1 (b) , a trivial bundle. tient map S n →RPn . Indeed, T RPn was deﬁned as a quotient space of T S n and the quotient map takes ﬁbers isomorphically to ﬁbers. (4) An interesting example which is small enough to be visualized completely is the pullback of the M¨bius bundle E →S 1 by the two-to-one map f : S 1 →S 1 given by o (3) The tangent bundle T S n is the pullback of the tangent bundle T RPn via the quoB since the inclusion p −1 (A) E

o f (z) = z2 in complex notation. One can regard the M¨bius bundle as the quotient of S 1 × R under the identiﬁcations (z, t) ∼ (−z, −t) , and the quotient map for this o identiﬁcation is the map f exhibiting the annulus S 1 × R as the pullback of the M¨bius bundle. More concretely, the pullback bundle can be thought of as a coat of paint applied to ‘both sides’ of the M¨bius bundle, and the map f sends each molecule o of paint to the point of the M¨bius band to which it adheres. Since E has one halfo twist, the pullback has two half-twists, hence is the trivial bundle. More generally, if o En is the pullback of the M¨bius bundle by the map z zn , then En is the trivial bundle for n even and the M¨bius bundle for n odd. This can be seen by viewing o the M¨bius bundle as the quotient of a strip [0, 1]× R obtained by identifying the o two edges {0}× R and {1}× R by a reﬂection of R , and then the bundle En can be

constructed from n such strips by identifying the right edge of the i th strip to the

20

Chapter 1

Vector Bundles

left edge of the i + 1 st strip by a reﬂection, the number i being taken modulo n so that the last strip is glued back to the ﬁrst. (5) At the end of the previous section we deﬁned the ﬂag bundle F (E) associated to an n dimensional vector bundle E →B to be the space of orthogonal direct sum decompositions of ﬁbers of E into lines. The vectors in the i th line form a line bundle spect to the projection F (E)→B since a point in the pullback consists of an n tuple of lines

1

**Li →F (E) , and the direct sum L1 ⊕ ··· ⊕ Ln is nothing but the pullback of E with re⊥ ··· ⊥
**

n

in a ﬁber of E

L 1 ⊕ ... ⊕ L n −− E −→ − − F ( E ) −−− B −−→

− − →

− − →

together with a vector v in this ﬁber, and v can be expressed uniquely as a sum v = v1 + ··· + vn with vi ∈

i . This construction for pulling

an arbitrary bundle back to a sum of line bundles is a key ingredient in the so-called ‘splitting principle’ which is important in §2.3 and §3.1. Here are some elementary properties of pullbacks: (i) (f g)∗ (E) ≈ g ∗ (f ∗ (E)) . (ii) 1 ∗ (E) ≈ E . 1 (iii) f ∗ (E1 ⊕ E2 ) ≈ f ∗ (E1 ) ⊕ f ∗ (E2 ) . (iv) f ∗ (E1 ⊗ E2 ) ≈ f ∗ (E1 ) ⊗ f ∗ (E2 ) . The proofs are easy applications of the preceding proposition. In each case one just checks that the bundle on the right satisﬁes the characteristic property of a pullback. isomorphism on each ﬁber, so f ∗ (E1 ) ⊗ f ∗ (E2 ) satisﬁes the condition for being the pullback f ∗ (E1 ⊗ E2 ) . Now we come to the main technical result about pullbacks: For example in (iv) there is a natural map from f ∗ (E1 ) ⊗ f ∗ (E2 ) to E1 ⊗ E2 that is an

∗ ∗ then the induced bundles f0 (E) and f1 (E) are isomorphic if A is compact Hausdorff

Theorem 1.6.

Given a vector bundle p : E →B and homotopic maps f0 , f1 : A→B ,

or more generally paracompact.

Proof:

∗ ∗ A× {0} and A× {1} are then f0 (E) and f1 (E) . So the theorem will be an immediate

Let F : A× I →B be a homotopy from f0 to f1 . The restrictions of F ∗ (E) over

consequence of the following result:

Proposition

Proof:

1.7. The restrictions of a vector bundle E →X × I over X × {0} and

X × {1} are isomorphic if X is paracompact. We need two preliminary facts: (1) A vector bundle p : E →X × [a, b] is trivial if its restrictions over X × [a, c] and X × [c, b] are both trivial for some c ∈ (a, b) . To see this, let these restrictions and h2 : E2 →X × [c, b]× Rn be isomorphisms. These isomorphisms may not agree on be E1 = p −1 (X × [a, c]) and E2 = p −1 (X × [c, b]) , and let h1 : E1 →X × [a, c]× Rn

p −1 (X × {c}) , but they can be made to agree by replacing h2 by its composition with

b]× Rn →X × [c. This is because for each x ∈ X we can ﬁnd onto the corresponding ﬁber of Ei−1 .2 21 given by h1 h−1 : X × {c}× Rn →X × {c}× Rn .1 . This means that E is trivial over each Vi × I . The inﬁnite composition h = the restriction over X × {0} since near each point x ∈ X only ﬁnitely many ϕi ’s are nonzero. we can choose an open cover {Uα } of X so that E is trivial over each Uα × I .6 holds for ﬁber bundles as well as vector bundles. Also as before we construct hi : Ei →Ei−1 using the fact that E is trivial over Vi × I .8. (2) For a vector bundle p : E →X × I .i × [ti−1 . h1 h2 ··· is then a well-deﬁned isomorphism from the restriction of E over X × {1} to bijection f : Vectn (B)→Vectn (A) . every vector bundle over a con∗ Corollary 1. Let us ﬁrst deal with the simpler case that X is compact Hausdorff. 1] such that the bundle is trivial over Ux. with the same proof. Relabel these as Ui for i = 1. For i ≥ 0 . Proof: ∗ If g is a homotopy inverse of f then we have f ∗ g ∗ = 1 ∗ = 1 and g ∗ f ∗ = 1 1 1 =1. Now we prove the proposition. ψi (x)) .4 there is a corresponding partition of unity by functions ϕi with the support of ϕi contained in Ui . let the subspace of X × I consisting of points of the form (x. 1 1 We might remark that Theorem 1. Then by (1) the bundle is trivial over Uα × I where Uα = Ux. As shown in Proposition 1. be the restriction of the bundle E over Xi . 1] . 2. and let pi : Ei →Xi ural projection homeomorphism Xi →Xi−1 lifts to a homeomorphism hi : Ei →Ei−1 ψi = ϕ1 + ··· + ϕi . the nat−1 which is the identity outside pi (Ui ) and which takes each ﬁber of Ei isomorphically the isomorphism X × [c. . they 2 deﬁne a trivialization of E . ti ] . By (2). so in particular ψ0 = 0 and ψm = 1 . The composition h = h1 h2 ··· hm is then an isomorphism from the restriction of E over X × {1} to the restriction over X × {0} . such that each Vi is a disjoint union of open sets each contained in some Uα . there exists an open cover {Uα } of X so that each open neighborhoods Ux.k in X and a partition 0 = t0 < t1 < ··· < tk = 1 of [0. ··· . m .1 ∩ ··· ∩ Ux. so there is a neighborhood of x in which all but ﬁnitely many hi ’s are the identity.21 asserts that there is a countable cover {Vi }i≥1 of X and a partition of unity {ϕi } with ϕi supported in Vi . As before we let ψi = ϕ1 + ··· + ϕi and let pi : Ei →Xi be the restriction of E over the graph of ψi . Lemma 1. Once h1 and h2 agree on E1 ∩ E2 . b]× Rn which on each slice X × {x}× Rn is restriction p −1 (Uα × I)→Uα × I is trivial. In the general case that X is only paracompact. Then a ﬁnite number of Uα ’s cover X . using compactness of [0. ··· . Ux. Let Xi be the graph of ψi .k . In particular. A homotopy equivalence f : A→B of paracompact spaces induces a tractible paracompact base is trivial.Classifying Vector Bundles Section 1. Since E is trivial over Ui × I .

so from a map f : S k−1 →GLn (C) one obtains Example 1. described earlier in the discussion of tensor products. z1 ] . the angle of rotation increases from 0 to 4π . w− ) . 2 2 we obtain vector ﬁelds w+ and w− on D+ and D− .10. The map f is called a clutching function for Ef . Thus f rotates (v+ . starting from a point where the two trivializations agree. f (x)(v)) ∈ ∂D+ × Rn . Given a map f : S k−1 →GLn (R) . By reﬂecting across the plane of the equator we obtain a corresponding vector ﬁeld v− 2 on D− . Consider the vector ﬁeld v+ on D+ obtained by choosing a tangent vector at the north pole and transporting this along each meridian circle in such a way as to maintain a constant angle with the meridian. The space CP1 is the quotient of C2 − {0} under the equivalence relation (z0 . There is then a natural projection Ef →S k and this is an n dimensional vector bundle. w+ ) to (v− . as one can most easily see by taking an equivalent deﬁnition in which the two hemispheres of S k are enlarged slightly to open balls and the identiﬁcation occurs over their intersection. z1 ) ∼ λ(z0 . v) ∈ ∂D− × Rn with (x. We can also write points of CP1 as ratios z = z0 /z1 ∈ C ∪ {∞} = S 2 . with the map f used in each slice S k−1 × {t} . f (x)(v)) ∈ Let us describe a way to construct vector bundles E →S k with base space a sphere. S k−1 . 1 2 Let us ﬁnd a clutching function for the canonical complex line bundle over CP = S . The vector ﬁelds v± and w± give trivializations of T S 2 over 2 2 D± . The same construction a complex vector bundle Ef →S k .9. z1 ) for λ ∈ C − {0} . Example 1. let Ef be the quotient of the disjoint union k ∂D+ × Rn . Denote the equivalence class of (z0 . v) ∈ ∂D− × Rn with (x. From this viewpoint the construction of Ef is a special case of the general construction of vector bundles by gluing together products. w− vectors in the v+ . by analogy with the mechanical clutch which engages and disengages gears in machinery. then f (θ) is rotation by 2θ . with D+ ∩D− = k D+ × Rn k k D− × Rn obtained by identifying (x. z1 ) by [z0 . w+ coordinate system. Let us see how the tangent bundle T S 2 to S 2 can be described in 2 these terms. a product S k−1 × (−ε. We can then recon2 2 struct T S 2 as a quotient of the disjoint union of D+ × R2 and D− × R2 by identifying along S 1 × R2 using the clutching function f which reads off the coordinates of the v− . As we go around the equator S 1 in the counterclockwise direction when viewed from above.22 Chapter 1 Vector Bundles Clutching Functions k k k k Write S k as the union of its upper and lower hemispheres D+ and D− . since the deﬁnition of a clutching k k function was that it identiﬁes (x. If we parametrize S 1 by the angle θ from the starting point. ε) . allowing us to identify these two halves of T S 2 with D± × R2 . Rotating each vector v+ and v− by 90 degrees counterclockwise when viewed from outside the sphere. works equally well with C in place of R . If .

From this we conclude that h+ and h− are unique up to homotopy. z1 /z0 ] = [1. 2 2 function f : S 1 →GL1 (C) deﬁned by f (z) = (z) . Y ] the set of homotopy classes of maps X →Y . and points in the disk D∞ 2 Over D0 a section of the canonical line bundle is given by [z. The set of diagonal matrices in GLn (C) is path-connected since it is homeomorphic to the product of n copies of C − {0} . then the same sort of clutching construction can be used to produce a vector bundle EF →S k × I that restricts to Ef over S k × {0} k−1 A basic property of the construction of bundles Ef →S k via clutching functions → Vectn (S k ) . For if we have a homotopy F : S k−1 × I →GLn (R) from f to g . k k k k p : E →S k . These sections determine trivializations of the canonical line bundle over these two disks. 1] = [z. 1] 2 D∞ a section is [1. Proof: The map Φ : [S k−1 . Since D± is contractible. z−1 ) . (z. Proposition 1. and over their common boundary S 1 we 2 2 pass from the D∞ trivialization to the D0 trivialization by multiplying by z . then points in the disk D0 inside the unit circle S 1 ⊂ C can be expressed 2 uniquely in the form [z0 /z1 . Choose trivializations h± : E± →D± × Cn . its restrictions E+ and E− over D+ and D− are trivial since D+ and D− We construct an inverse Ψ to Φ . then the association Ef gives a well-deﬁned map Φ : [S k−1 . Hence Ef and Eg are isomorphic by Proposition 1. such a map is homotopic to a constant map k by composing with a contraction of D± . Now we need the fact that GLn (C) is path- see that Ψ (E) is well-deﬁned. By such operations any matrix in GLn (C) can be diagonalized. which means that − inverses of each other. GLn (C)] . Thus if we denote by [X. The elementary row operation of modifying a matrix in GLn (C) by adding a scalar multiple of one row to another row can be realized by a path in GLn (C) . whose homotopy class is by deﬁnition Ψ (E) ∈ [S k−1 . composition h+ h−1 : S k−1 →GLn (C) is also unique up to homotopy. To k are contractible. Then h+ h−1 deﬁnes a map − k k D± →GLn (C) .11. GLn (C)] . note ﬁrst that any two choices of h± differ by a map connected.2 23 2 we do this. 1] with |z| ≤ 1 . There is a similarly-deﬁned map Φ for complex vector bundles. This turns out to have slightly better behavior than in the real case. 1) and over (1. Hence the Ψ is a well-deﬁned map Vectn (S k )→[S k−1 .7. GLn (C)]→Vectn (S k ) which sends a clutching C function f to the vector bundle Ef is a bijection. Thus if 2 2 2 2 we take D∞ as D+ and D0 as D− then the canonical line bundle has the clutching 2 than D∞ then the clutching function would have been f (z) = (z−1 ) . It is clear that Ψ and Φ are C . z−1 ] outside S 1 can be written uniquely in the form [1. GLn (R)] f →GLn (R) is that Ef ≈ Eg if f and g are homotopic. If we had taken D+ to be D0 rather f :S and Eg over S k × {1} . Given an n dimensional vector bundle S k−1 →GLn (C) . so we will prove the following basic result about the complex form of Φ before dealing with the real case.Classifying Vector Bundles Section 1. just by inserting a factor of t in front of the scalar multiple and letting t go from 0 to 1 . z−1 ] with |z−1 | ≤ 1 .

Thus both cosets are path-connected and so GLn (R) has αβ is a homeomorphism of GLn (R) with connected. Then by a path of diagonal matrices we can make all the diagonal entries +1 or −1 . as we will now show. Just as in the complex case we can construct a path from an arbitrary matrix in GLn (R) to a diagonal matrix. Then the matrix product 1)α 1 1 (f ⊕ 1 t (1 ⊕ g)αt gives a homotopy from f ⊕ g to f g ⊕ 1 . which is the case as we saw in the proof of the proposition. two ordered bases being equivalent if the invertible matrix taking the ﬁrst basis to the second has positive determinant. This can be seen by looking at the clutching functions for these two bundles. The closest analogy with the complex case is obtained by considering oriented real vector bundles.24 Chapter 1 Vector Bundles Example 1. Example 1. GLn (C)] has a single element. or in other words that GLn (C) is path-connected. The preceding analysis does not quite work for real vector bundles since GLn (R) is not path-connected. Recall from linear algebra that an orientation of a real vector space is an equivalence class of ordered bases.12. The two cosets n n are homeomorphic since the map β inverse the map β two path-components. tion (H ⊗ H) ⊕ 1 ≈ H ⊕ H where 1 is the trivial one-dimensional bundle. α−1 β .13. Since GL2n (C) is path-connected. which is the clutching function for Ef g ⊕ n . Two −1 ’s represent a 180 degree that the subgroup GL+ (R) consisting of matrices of positive determinant is pathn cosets GL+ (R) and αGL+ (R) for α a ﬁxed matrix of determinant −1 . the other two blocks being zero. f g(x) = The bundle Ef ⊕ Eg has clutching function the map f ⊕g : S k−1 →GL2n (C) having the matrices f (x) in the upper left n× n block and the matrices g(x) in the lower right n× n block. This shows . In fact GLn (R) has exactly two path-components. More generally. and GLn (R) is the disjoint union of the two rotation of a plane so they can be replaced by +1 ’s via a path in GLn (R) . let us show the formula Ef g ⊕ n ≈ Ef ⊕ Eg for n dimensional vec- f g is the clutching function obtained by pointwise matrix multiplication. which are the maps S 1 →GL2 (C) given by Let us show that the canonical line bundle H →CP1 satisﬁes the rela- z z2 0 0 1 and z z 0 0 z tor bundles Ef and Eg over S k with clutching functions f . where f (x)g(x) . Every complex vector bundle over S 1 is trivial. g : S k−1 →GLn (C) . there is a path αt ∈ GL2n (C) from the identity matrix to the matrix of the transformation which interchanges the two factors of Cn × Cn . This subgroup has index 2 . since by the proposi- tion this is equivalent to saying that [S 0 . We can see that there are at least two path-components since the determinant function is a continuous surjection GLn (R)→R − {0} to a space with two path-components.

This is because an oriented line bundle o over a paracompact base is always trivial since it has a canonical section formed by the unit vectors having positive orientation. the 0 n dimensional vector bundles over S k with an orientation speciﬁed in the ﬁber over one point x0 ∈ S k−1 . This implies that the natural map n n Vectn (S k )→Vectn (S k ) is a bijection. We conclude that Vectn (S 1 ) has exactly two elements for each trivial. The nontrivial bundle is nonorientable since an orientable bundle over S 1 is When k > 1 the sphere S k−1 is path-connected. Not all vector bundles can be given an orientation. These bundles all have automorphisms reversing orientations of ﬁbers. determined by an orientation in one ﬁber. GL+ (R)] n GL+ (R) is path-connected. It follows that the map Vectn (S k )→Vectn (S k ) is surjective and at most two-to-one. represented by maps taking x0 to the identity and the other point n of S 0 to either the identity or a reﬂection. any two clutching functions S 0 →GL+ (R) being homotopic. and since + Proposition 1. + . so we obtain a bijection of Vectn (S k ) with the homotopy classes 0 The map Vectn (S k )→Vectn (S k ) that forgets the orientation over x0 is a sur0 jection that is two-to-one except on vector bundles that have an automorphism (an isomorphism from the bundle to itself) reversing the orientation of the ﬁber over x0 . when n = 1 .2 25 An orientation of a real vector bundle p : E →B is a function assigning an orientation to each ﬁber in such a way that near each point of B there is a local trivialization h : p −1 (U)→U × Rn carrying the orientations of ﬁbers in p −1 (U ) to the standard oribe to say that the orientations of ﬁbers of E can be deﬁned by ordered n tuples of independent local sections. so maps S k−1 →GLn (R) takWhen k = 1 there are just two homotopy classes of maps S 0 →GLn (R) taking ing x0 to GL+ (R) take all of S k−1 to GL+ (R) . Another way of stating this condition would → Vectn (S k ) . the argument from the complex case shows: n entation of Rn in the ﬁbers of U × Rn . where isomorphisms are required to preserve orientations. GL+ (R)] n → Vectn (S k ) is a bijection.Classifying Vector Bundles Section 1. We can choose the trivializations h± over D± to carry this orientation to a standard orientation of Rn . or the direct sum of the M¨bius bundle o o with a trivial bundle when n > 1 . the M¨bius bundle. with the equivalence relation of isomorphism preserving the k orientation of the ﬁber over x0 . The clutching construction deﬁnes a map Φ : [S k−1 . Thus every vector bundle over S k is orientable + 0 and has exactly two different orientations.14. x0 to GL+ (R) . n n ≥ 1 . where it is one-to-one. Let Vectn (B) denote the set of isomorphism classes of oriented n dimensional + real vector bundles over B . + To analyze Vectn (S k ) let us introduce an ad hoc hybrid object Vectn (S k ) . and then h± is unique up to of maps S k−1 →GLn (R) taking x0 to GL+ (R) . n homotopy as before. The corresponding bundles are the trivial bundle and. For example the M¨bius bundle is not orientable. The map Φ : [S k−1 .

as we saw in Example 1. function for an oriented 2 dimensional vector bundle Em →S 2 . and for the second operation one can just insert a scalar factor of t in front of the linear combination of vectors being subtracted and let t go from 0 to 1 . SO(2) is just a circle. and twoto-one otherwise.26 Chapter 1 Vector Bundles It is one-to-one on bundles having an orientation-reversing automorphism. It is a fact proved in every zm . Both operations are realizable by a homotopy.9 as T S 2 . Composing maps S k−1 →SO(n) with the inclusion SO(n) function [S k−1 a map into SO(n) . z rotations. and E1 is the real vector bundle underlying the canonical complex introductory algebraic topology course that an arbitrary map f : S 1 →S 1 is homotopic to one of the maps z of f . We encountered E2 in line bundle over CP1 = S 2 . the orthogonal matrices of determinant +1 . SO(n)]→[S k−1 . since orientation-preserving isometries of R2 are we have for each integer m ∈ Z the map fm : S 1 →S 1 . This is the path-component of O(n) containing the identity matrix since deformation retractions preserve path-components. Thus via the bundles Em we have a bijection Vect2 (S 2 ) ≈ Z . which is the clutching retraction of GL+ (R) onto SO(n) gives a homotopy of any map S k−1 →GL+ (R) to n n . Hence the sequence of operations provides a deformation retraction of GLn (R) onto O(n) . the n× n matrices over C whose columns form an orthonormal basis using the standard Hermitian inner product. The Gram-Schmidt orthogonalization process provides a deformation retraction of GLn (R) onto the subspace O(n) .10. The other component consists of orthogonal matrices of determinant −1 . Taking k = 2 as well. so elements of Vect2 (S 2 ) correspond bijectively . that is. In the complex case the same argument applied to GLn (C) gives a deformation retraction onto the unitary group U (n) . Restricting to the component GL+ (R) . we have a deformation retraction of this component onto n the special orthogonal group SO(n) . called the degree orientation of Em changes it to E−m . n GL+ (R) gives a n This is a bijection since the deformation Example 1. The explicit algebraic formulas show that both operations depend continuously on the initial basis and have no effect on an orthonormal basis. matrices whose columns form an orthonormal basis. and if two such maps to SO(n) are homotopic in GL+ (R) then n they are homotopic in SO(n) by composing each stage of the homotopy with the retraction GL+ (R)→SO(n) produced by the deformation retraction. For n example. the n× n matrices A with AAT = I . determined by the angle of rotation. n The advantage of SO(n) over GL+ (R) is that it is considerably smaller. This is obvious for rescaling. Inside GLn (R) is the orthogonal group O(n) . Reversing the + zm for some m uniquely determined by f . the subgroup consisting of isometries. or in matrix terms. Each step of the process is either normalizing the i th vector of a basis to have length 1 or subtracting a linear combination of the preceding vectors from the i th vector in order to make it orthogonal to them. a point in S 1 . GL+ (R)] .

Namely. SO(n)] has a natural group structure coming from f (x)g(x) . the path-connected. and zm zn = zm+n . SO(n)] obtained by ignoring basepoint data. the product of two maps f . suppose we are given a map f : S i →SO(n) . so all →S 1 are homotopic if k > 2 . The generalization to n > 1 will consist in . This is not too hard to show using a little basic homotopy classes of maps S 3 →SO(n) and S 3 →U (n) there are. U (n)] are isomorphic to the homotopy groups πi SO(n) and πi U(n) .we obtain a similar bijection Vect2 (S 2 ) ≈ Z .) Hence every 2 dimensional vector bundle over S k is trivial when k > 2 .2 27 with integers m ≥ 0 . as we show in §??. and suppose ft is a homotopy from f0 to f1 that may not take basepoint to basebasepoint to basepoint for all t . point at all times. When n = 1 this bundle will be just the canonical line bundle over RP∞ . Then [ft (s0 )]−1 ft is a new homotopy from f0 to f1 that does take choose as the basepoint of SO(n) . SO(n)] The Universal Bundle We will show that there is a special n dimensional vector bundle En →Gn with the property that all n dimensional bundles over paracompact base spaces are obtainable as pullbacks of this single bundle. vector bundles over S 3 are trivial. where homotopies are also required to take s0 to x0 at all times. g : X →SO(n) is the map x z [S 1 . C Another fact proved near the beginning of algebraic topology is that any two maps S k−1 and R is contractible. Since SO(n) is path-connected. Note to the reader: The rest of this chapter will not be needed until Chapter 3. This basepoint data is needed in order to deﬁne a group structure in πi X when i > 0 . SO(n)] and [S i . there is an evident map πi SO(n)→[S i . (When i = 0 there is no natural group structure in general. special cases of the homotopy groups πi X deﬁned for all spaces X and central to algebraic topology. the product of zm and z zm zn . where we also compute how many For any space X the set [X. It is a fact that any two maps S 2 →SO(n) or S 2 →U (n) are homotopic.Classifying Vector Bundles Section 1. as is any complex line bundle over S k for k > 2 . and then πi X is the set of homotopy classes of maps S i →X taking a suppose that f0 and f1 are two maps S i →SO(n) taking s0 to the identity matrix. SO(2)] ≈ Z is a group isomorphism since for maps S 1 →S 1 . U (n)] is a group. The groups [S . via the bundles Em . To see that this is surjective.) However. (Such a map lifts to the covering space R of S 1 . which we SO(n) from the identity matrix to f (s0 ) . machinery of algebraic topology. For injectivity of the map πi SO(n)→[S i . Taking X = SO(n) . For example the bijection zn is z i the group structure in SO(n) . Similarly [X. Since U (1) is also homeomorphic to S 1 . The same argument also applies for U (n) . To deﬁne πi X requires choosing a basechosen basepoint s0 ∈ S i to x0 . and then the matrix product α−1 f deﬁnes t group structure on πi X turns out to be independent of the choice of x0 when X is point x0 ∈ X . deﬁned earlier. there is a path αt in a homotopy from f to a map S i →SO(n) taking s0 to the identity matrix.

It is a closed subspace since orthogonality of two vectors can of spheres is compact.28 Chapter 1 Vector Bundles replacing RP∞ . π (v)) is a local trivialization of En (Rk ) . ··· . Since the value of this determinant is obviously a continuous function of v1 . For U will show that U is open in Gn (Rk ) and that the map h : p −1 (U )→U × ≈ U × Rn to be open is equivalent to its preimage in Vn (Rk ) being open. namely. As a set this is the collection of all n dimensional vector subspaces of Rk . π (vn ) are independent. each basis vector will lie in some Rk . by the space of n dimensional vector subspaces of R∞ . is a vector bundle. Proof: The projection p : En (Rk )→Gn (Rk ) . Avn have nonzero determinant. v) = ( . so a set in Gn (R∞ ) is open iff it intersects each Gn (Rk ) in an open set. We give Gn (R∞ ) the weak or direct En (Rk ) . ··· . n dimensional planes in Rk passing through the origin. There are canonical n dimensional vector bundles over Gn (Rk ) and Gn (R∞ ) . let π : Rk → be orthogonal projec| tion and let U = { ∈ Gn (Rk ) | π ( ) has dimension n } . the subspace of orthogonal n tuples. ∈ U . First we deﬁne the Grassmann manifold Gn (Rk ) for nonnegative integers n ≤ k . ··· . v) = . hence there will be an Rk containing all the limit topology. vn such that π (v1 ). This with respect to the standard basis in . ··· . vn yielding a nonzero determinant form an open set in Vn (Rk ) . again with Lemma 1. that is. Later in this section we will construct a ﬁnite CW complex structure on Gn (Rk ) and in the process show that it is Hausdorff and a manifold of dimension n(k − n) . First suppose k is ﬁnite. we let Gn (R∞ ) = ∞ The inclusions Rk ⊂ Rk+1 ⊂ ··· give inclusions Gn (Rk ) ⊂ Gn (Rk+1 ) ⊂ ··· . both for ﬁnite and inﬁnite k . In particular. ··· . give inclusions En (Rk ) ⊂ En (Rk+1 ) ⊂ ··· and we set En (R∞ ) = the weak or direct limit topology. The condition on v1 . . the space of 1 dimensional vector subspaces of R∞ . it follows that the frames v1 .. This is a subspace of the product of n copies of the unit sphere S k−1 . vn is preimage consists of orthonormal frames v1 . in other words. ∞ This is the set of all n dimensional vector subspaces of the vector space R since if we choose a basis for a ﬁnite-dimensional subspace of R . Deﬁne En (Rk ) = { ( . To deﬁne a topology on Gn (Rk ) we ﬁrst deﬁne the Stiefel manifold Vn (Rk ) to be the space of orthonormal n frames in Rk . Hence Vn (Rk ) is compact since the product n frame to the subspace it spans. The inclusions Rk ⊂ Rk+1 ⊂ ··· k basis vectors and therefore the whole subspace. Let A be the matrix of π the domain R k and any ﬁxed basis in the range then that the n× n matrix with columns Av1 . ··· . and Gn (Rk ) is topologized by giving it the quotient topology with respect to this surjection. There is a natural surjection Vn (Rk )→Gn (Rk ) sending an be expressed by an algebraic equation.15. We deﬁned by h( . For ∈ Gn (Rk ) . p( . vn . and k k Gn (R ) . So Gn (Rk ) is compact as well. v) ∈ Gn (Rk )× Rk | v ∈ } . n tuples of orthonormal vectors in Rk .

··· . bundle p : E →X . [f ] f ∗ (En ) . Gn is called the classifying space for n dimensional vector bundles and En →Gn is called the universal bundle. and to simplify notation we will write Gn for Gn (R∞ ) and En for En (R∞ ) . suppose ﬁrst that we have a map . depends continuously on — B sends the standard basis to v1 .Classifying Vector Bundles Section 1. v) ∈ S n × Rn+1 | x ⊥ v } . We . vn . we use the notation [X. v) = ( . vn for since Gn (R ) has the quotient topology continuous. — A sends the standard basis to π (v1 ). Via the inclusion Rn+1 R∞ we can view this { (x. Namely. Gn ]→Vectn (X) . vector bundles over a ﬁxed base space are classiﬁed by homotopy classes of maps into Gn . vn . consider the tangent bundle to S n . and E is exactly the pullback f ∗ (En ) . vn . L−1 (w)) and L−1 depends continuously on . so we have as a map f : S n →Gn (R∞ ) = Gn . Similarly. When k = ∞ one takes U to be the union of the U ’s for increasing k . continuity being automatic since we use the weak topology. ··· . But since L follows that L depends only on depends continuously on = AB −1 depends continuously . we see that h is Theorem 1.16: The key observation is the following: For an n dimensional vector is a linear injection on each ﬁber. it follows that the product L on v1 . ··· . vn+1 . matrix inversion being continuous. an isomorphism E ≈ f ∗ (En ) is equivalent to a map g : E →R∞ that f : X →Gn and an isomorphism E ≈ f ∗ (En ) . w) = ( . Then we have a commutative diagram Proof of 1. ··· . As earlier in this section. Y ] for the set of homotopy classes of maps f : X →Y . so h−1 is continuous. To see this. vn an orthonormal basis for . the map [X. ··· . ··· . Since matrix multiplication and matrix inversion are continuous operations (think of the ‘classical adjoint’ formula for the inverse of a matrix). Each ﬁber p −1 (x) is a point in Gn (Rn+1 ) .2 29 It is clear that h is a bijection which is a linear isomorphism on each ﬁber. not on the basis v1 . This is the vector bundle p : E →S n where E = a map S n →Gn (Rn+1 ) . vk . π (vn ). ··· .16. from Vn (Rk ) . We linear map L : Rk →Rk restricting to π on can write L as a product AB −1 where: and vn+1 . As an example of how a vector bundle could be isomorphic to a pullback f ∗ (En ) . π (v)) = ( . This ﬁnishes the proof for ﬁnite k . it k . Since we have h( . For ∈ U there is a unique invertible ⊥ and the identity on = Ker π . h−1 ( . Both A and B depend continuously on v1 . vk with v1 . x p −1 (x) . The local trivializations h constructed above for ﬁnite k then ﬁt together to give a local trivialization over this U . ··· . vk a ﬁxed basis for ⊥ need to check that h and h−1 are continuous. Because of this. vn+1 . For paracompact X . ··· . L (v)) . we claim that L . is a bijection. We will mainly be interested in the case k = ∞ now. regarded as a k× k matrix. Thus.

The composition across the top row is a map g : E →R∞ that is given a map g : E →R∞ that is a linear injection on each ﬁber. suppose p : E →X is an . so this theorem is of limited usefulness in enumerating all the different vector bundles over a given base space. let gt = (1 − t)g0 + tg1 . letting f (x) be the n plane g(p −1 (x)) . Near open cover {Ui } of X such that E is trivial over each Ui . then f0 and f1 will The ﬁrst step in constructing a homotopy gt is to compose g0 with the homotopy Lt : R∞ →R∞ deﬁned by Lt (x1 . Gn ] a linear injection on each ﬁber. bundles over a paracompact base have inner products. x2 . then these give maps g0 . For example. through maps gt that are linear injections on ﬁbers. Among other things. x2 . Its importance is due more to its theoretical implications. and this inner product on En induces an inner product on every − − → X − − Gn −− −→ f → Vectn (X) . To show surjectivity of the map [X. as in the ﬁrst paragraph of the proof. since both f and π have this property. if we have isomorphisms E ≈ f0 (En ) and E ≈ f1 (En ) for two maps f0 . it is easy to deduce that vector an obvious inner product obtained by restricting the standard inner product in R∞ pullback f ∗ (En ) . so Lt is injective. ···) . If this is so. Let {Uα } be an open cover of X such that E is trivial over each Uα . v ϕi (p(v))gi (v) . and at least one ϕi is nonzero. and there is a partition each point of X only ﬁnitely many ϕi ’s are nonzero. 0. since the bundle En →Gn has to each n plane. Similarly we can homotope g1 into the even-numbered coordinates. deﬁne f : X →Gn by as above. Still calling the new g ’s g0 and g1 . We claim g0 and g1 are homotopic be homotopic via ft (x) = gt (p −1 (x)) . so these extended (ϕi p)gi ’s are the coordinates of a map g : E →(Rn )∞ = R∞ that is ∗ ∗ For injectivity. By Lemma 1. a linear injection on each ﬁber. This is linear and injective on ﬁbers for each t since g0 and g1 are linear and injective on ﬁbers. Let gi : p −1 (Ui )→Rn be the composition of a trivialization p −1 (Ui )→Ui × Rn with projection onto Rn . extends to a map E →Rn that is zero outside p −1 (Ui ) . Composing the homotopy Lt with g0 moves the image of g0 into the odd-numbered coordinates. Gn ] is usually beyond the reach of what is possible technically.21 in the Appendix to this chapter there is a countable of unity {ϕi } with ϕi supported in Ui . v) = v .30 Chapter 1 Vector Bundles ∼ f ∗ π E ≈ f ( En ) −− En −− R −→ −→ ∞ − − − − −− −− −→ p − − → where π ( . This clearly yields a commutative diagram n dimensional vector bundle. The map (ϕi p)gi . ···) = (1 − t)(x1 . 0. g1 : E →R∞ that are linear injections on ﬁbers. it can reduce the proof of a general statement to the special case of the universal bundle. x2 . ···) + t(x1 . Conversely. An explicit calculation of [X. f1 : X →Gn . For each t this is a linear map whose kernel is easily computed to be 0 .

it and . and then fv ( ) = |v| > fv ( ) . + have a single element when X is a point. v) ∈ Gn (R∞ )× R∞ with v ∈ Gn (R )→Gn (R∞ ) . using for example the local trivializations constructed in Lemma 1. is simply-connected. since we will need this fact later sufﬁces to ﬁnd a continuous map f : Gn (Rk )→R taking different values on k when we construct the CW structure. Given two n planes and in Gn (Rk ) .16 for oriented vector bundles. In other words. For a vector v ∈ R let fv ( ) be the length of the orthogonal projection of v onto This is a continuous function of for v1 . Gn (R∞ )] ∞ Forming the union over increasing k we then have the space Gn (R∞ ) . In particular. En (R∞ ) is the pullback of En (R∞ ) via the natural projection show that Vectn (X) ≈ [X. Small modiﬁcations in the proof that Vectn (X) ≈ [X. because of the triviality of Vectn (S 1 ) ≈ [S 1 . Gn (C∞ )] . Recall that any n× k matrix A can be put into an echelon form by a ﬁnite sequence of elementary row operations . vn .Classifying Vector Bundles Section 1. Let Gn (Rk ) be the space of oriented n planes in Rk . In fact Gn (R∞ ) is the universal cover of Gn (R∞ ) since it also observe that a vector bundle E ≈ f ∗ (En (R∞ ) is orientable iff its classifying map spond bijectively with lifts f .2 31 The preceding constructions and results hold equally well for vector bundles over C . and so on.15. it would be good to have a better grasp on their topology. Similar statements hold in the complex case as well. Now for − . The natural projection Gn (R∞ )→Gn (R∞ ) obtained by ignoring orientations is two-to-one. showing that Vectn (X) ≈ [X. . vn hence in an n plane ≠ then fv ( ) = (v · v1 )2 + ··· + (v · vn )2 k 1/2 since if we choose an orthonormal basis v1 . with Gn (Ck ) the space of n dimensional C linear subspaces of Ck . C There is also a version of Theorem 1. ··· . the proof of Theorem 1. ··· .16 translates directly to complex vector bundles. oriented bundle En (R∞ ) over Gn (R∞ ) consists of pairs ( . choose v ∈ There is a nice description of the cells in the CW structure on Gn (Rk ) in terms of the familiar concept of echelon form for matrices. Here we show that Gn (R∞ ) has the structure of a CW complex with each Gn (Rk ) a ﬁnite subcomplex. and readers familiar with the notion of a covering space will have no trouble in recognizing that this two-to-one Both Gn (R∞ ) and Gn (R∞ ) are path-connected since Vectn (X) and Vectn (X) + projection map is a covering space. We will also see that Gn (Rk ) is a closed manifold of dimension n(k − n) . the quotient space of Vn (Rk ) obtained by identifying two n frames when they determine the same oriented subspace of Rk . For a start let us show that Gn (Rk ) is Hausdorff. One can + f : X →Gn (R∞ ) lifts to a map f : X →Gn (R∞ ) . Gn (R∞ )] . and in fact orientations of E corre- Cell Structures on Grassmannians Since Grassmann manifolds play a fundamental role in vector bundle theory. which is certainly continuous in since Gn (R ) has the quotient topology from Vn (Rk ) . The universal . Gn (R∞ )] . with Gn (Ck ) a closed manifold of dimension 2n(k − n) .

32

Chapter 1

Vector Bundles

consisting either adding a multiple of one row to another, multiplying a row by a nonzero scalar, or permuting two rows. In the standard version of echelon form one strives to make zeros in the lower left corner of the matrix, but for our purposes it will be more convenient to make zeros in the upper right corner instead. Thus a matrix in echelon form will look like the following, with the asterisks denoting entries that are arbitrary numbers:

∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ 1 0 0 0 0 ∗ ∗ ∗ 0 1 0 0 0 0 1 0 0 0 0 ∗ 0 0 0 ∗ 0 0 0 1 0 0 0 0

We are assuming that our given n× k matrix A has rank n and that n ≤ k . The shape of the echelon form is speciﬁed by which columns contain the special entries 1 , say in the columns numbered σ1 < ··· < σn . The n tuple (σ1 , ··· , σn ) is called the Schubert symbol σ (A) . To see that this depends only on the givne matrix A and not any particular reduction of A to echelon form, consider the n plane

A

in Rk

spanned by the rows of A . This is the same as the n plane spanned by the rows of an echelon form of A since elementary row operations do not change the subspace spanned by the rows. Let pi : Rk →Rn−i be projection onto the last n − i coordinates. As i goes from 0 to k the dimension of the subspace pi ( as is apparent from the echelon form since pi (

A) A)

of Rk decreases from

n to 0 in n steps and the decreases occur precisely when i hits one of the values σj , has basis the rows of the matrix obtained by deleting the ﬁrst i columns of the echelon form. Thus the Schubert symbol only depends on the n plane spanned by the rows of A . In fact the echelon form itself depends only on the n plane spanned by the rows. To see this, consider projection onto the n coordinates of Rk given by the numbers σi . This projection is surjective when restricted to the n plane, hence is also injective, and the n rows of the echelon form are the vectors in the n plane that project to the standard basis vectors. As noted above, the numbers σj in the Schubert symbol of an n plane are the numbers i for which the dimension of the image pi ( ) decreases as i goes from 0 to n . Equivalently, these are the i ’s for which the dimension of the kernel of the | restriction pi | increases. This kernel is just ∩ Ri where Ri is the kernel of pi , the subspace of Rk spanned by the ﬁrst i standard basis vectors. This gives an alternative deﬁnition of the Schubert symbol σ ( ) . Given a Schubert symbol σ one can consider the set e(σ ) of all n planes in Rk having σ as their Schubert symbol. In terms of echelon forms, the various n planes in e(σ ) are parametrized by the arbitrary entries in the echelon form. There are σi −i of these entries in the i th row, for a total of (ß1 − 1) + ··· (σn − n) entries. Thus e(σ ) is homeomorphic to a euclidean space of this dimension, or equivalently an open cell.

Classifying Vector Bundles

Section 1.2

33

Proposition 1.17.

The cells e(σ ) are the cells of a CW structure on Gn (Rk ) .

For example G2 (R4 ) has six cells corresponding to the Schubert symbols (1, 2) , (1, 3) , (1, 4) , (2, 3) , (2, 4) , (3, 4) , and these cells have dimensions 0 , 1 , 2 , 2 , 3 , 4 respectively. In general the number of cells of Gn (Rk ) is choosing the n distinct numbers σi ≤ k .

k n

, the number of ways of

Proof:

Our main task will be to ﬁnd a characteristic map for e(σ ) . This is a map

from a closed ball of the same dimension as e(σ ) into Gn (Rk ) whose restriction to the interior of the ball is a homeomorphism onto e(σ ) . From the echelon forms described above it is not clear how to do this, so we will use a slightly different sort of echelon form. We allow the special 1 ’s to be arbitrary nonzero numbers and we allow the entries below these 1 ’s to be nonzero. Then we impose the condition that the rows be orthonormal and that the last nonzero entry in each row be positive. Let us call this an orthonormal echelon form. Once again there is a unique orthonormal echelon form for each n plane since if we let

i

**denote the subspace of
**

i

spanned = ∩ Rσi ,

by the ﬁrst i rows of the standard echelon form, or in other words then there is a unique unit vector in coordinate.

i

orthogonal to

i−1

and having positive σi th

in the unit sphere S σi −1 ⊂ Rσi ⊂ Rk consisting of unit vectors with non-negative σi th coordinate. In the Stiefel manifold Vn (Rk ) let E(σ ) be the subspace of orthonormal frames (v1 , ··· , vn ) ∈ (S k−1 )n such that vi ∈ Hi for each i . We claim that E(σ ) is homeomorphic to a closed ball. To prove this the main step is to show that the projection π : E(σ )→H1 , π (v1 , ··· , vn ) = v1 , is a trivial ﬁber bundle. This is equivalent π , where v0 = (0, ··· , 0, 1) ∈ Rσ1 ⊂ Rk , since the map π × p : E(σ )→H1 × π −1 (v0 ) is then a continuous bijection of compact Hausdorff spaces, hence a homeomorphism.

The i th row of the orthonormal echelon form then belongs to the hemisphere Hi

to ﬁnding a projection p : E(σ )→π −1 (v0 ) which is a homeomorphism on ﬁbers of

The map p : π −1 (v)→π −1 (v0 ) is obtained by applying the rotation ρv of Rk that takes v to v0 and ﬁxes the (k − 2) dimensional subspace orthogonal to v and v0 . This rotation takes Hi to itself for i > 1 since it affects only the ﬁrst σ1 coordinates of vectors in Rk . Hence p takes π −1 (v) onto π −1 (v0 ) . The ﬁber π −1 (v0 ) can be identiﬁed with E(σ ) for σ = (σ2 − 1, ··· , σn − 1) . By

induction on n this is homeomorphic to a closed ball of dimension (σ2 − 2) + ··· + (σn −n) , so E(σ ) is a closed ball of dimension (σ1 −1)+···+(σn −n) . The boundary of this ball consists of points in E(σ ) having vi in ∂Hi for at least one i . This too follows by induction since the rotation ρv takes ∂Hi to itself for i > 1 . The natural map E(σ )→Gn sending an orthonormal n tuple to the n plane it spans takes the interior of the ball E(σ ) to e(σ ) bijectively. Since Gn has the quotient topology from Vn , the map int E(σ )→e(σ ) is a homeomorphism. The boundary of

34

Chapter 1

Vector Bundles

E(σ ) maps to cells e(σ ) of Gn where σ is obtained from σ by decreasing some σi ’s, so these cells e(σ ) have lower dimension than e(σ ) . To see that the maps E(σ )→Gn (Rk ) for the cells e(σ ) are the characteristic

maps for a CW structure on Gn (Rk ) we can argue as follows. Let X i be the union of the cells e(σ ) in Gn (Rk ) having dimension at most i . Suppose by induction on i that the maps ∂E(σ )→X i produces a CW complex Y and a natural continuous bijection subspace of Gn (Rk ) , so the map Y →X i+1 is a homeomorphism and X i+1 is a CW X i is a CW complex with these cells. Attaching the (i+1) cells e(σ ) of X i+1 to X i via

Y →X i+1 . Since Y is a ﬁnite CW complex it is compact, and X i+1 is Hausdorff as a

complex, ﬁnishing the induction. Thus we have a CW structure on Gn (Rk ) .

plexes and Gn (R∞ ) has the weak topology with respect to these subspaces, it follows that we have also a CW structure on Gn (R∞ ) . Similar constructions work to give CW structures on complex Grassmann maniThe hemisphere Hi is deﬁned to be the subspace of the unit sphere S 2σi −1 in Cσi consisting of vectors whose σi th coordinate is real and nonnegative, so Hi is a ball of dimension 2σi − 2 . The transformation ρv ∈ SU (k) is uniquely determined by specifying that it takes v to v0 and ﬁxes the orthogonal (k − 2) dimensional complex subspace, since an element of U (2) of determinant 1 is determined by where it sends one unit vector. The highest-dimensional cell of Gn (Rk ) is e(σ ) for σ = (k−n+1, k−n+2, ··· , k) , of dimension n(k − n) , so this is the dimension of Gn (Rk ) . Near points in these topdimensional cells Gn (Rk ) is a manifold. But Gn (Rk ) is homogeneous in the sense that given any two points in Gn (Rk ) there is a homeomorphism Gn (Rk )→Gn (Rk ) taking one point to the other, namely, the homeomorphism induced by an invertible linear Gn (Rk ) is a manifold near all points. Since it is compact, it is a closed manifold. There is a natural inclusion i : Gn folds, but here e(σ ) will be a cell of dimension (2σ1 −2)+(2σ2 −4)+···+(2σn −2n) .

Since the inclusions Gn (Rk ) ⊂ Gn (Rk+1 ) for varying k are inclusions of subcom-

map Rk →Rk taking one n plane to the other. From this homogeneity it follows that Gn+1 , i( ) = R× j( ) where j : R∞ →R∞ is

the embedding j(x1 , x2 , ···) = (0, x1 , x2 , ···) . If σ ( ) = (σ1 , ··· , σn ) then σ (i( )) = (1, σ1 + 1, ··· , σn + 1) , so i takes cells of Gn to cells of Gn+1 of the same dimension, making i(Gn ) a subcomplex of Gn+1 . Identifying Gn with the subcomplex i(Gn ) , we obtain an increasing sequence of CW complexes G1 ⊂ G2 ⊂ ··· whose union G∞ =

n

Gn is therefore also a CW complex. Similar remarks apply as well in the

complex case.

For the converse. and β is a partition of unity {ϕβ } subordinate to the cover. and −1 and ϕ2 (1/2 . for example Rn . and we obtain the desired partition of unity subordinate to β {Uα } by normalizing each ϕβ by dividing it by ϕβ .18. It is the former deﬁnition which is most useful in algebraic topology. 1] are disjoint open sets containing A1 and A2 . This means that the ϕβ ’s are ϕβ = 1 . each x ∈ X has an open −1 map ϕx : X →I with ϕx (x) = 1 and ϕx (X − Vx ) = 0 . Most of the spaces one meets in algebraic topology are paracompact. and we relabel the corresponding ϕx ’s as ϕβ ’s. Using (1) and (2). Proposition 1. A paracompact space X is normal. 1]} . 1] be the sum of the ϕβ ’s which are nonzero at some point of Ai . Let {Uα } be an open cover of X . Then ϕi (Ai ) = 1 . Since X is normal. and (3) are paracompact.Appendix: Paracompactness 35 Appendix: Paracompactness A Hausdorff space X is paracompact if for each open cover {Uα } of X there maps X →I such that each ϕβ has support (the closure of the set where ϕβ ≠ 0 ) contained in some Uα . So we shall use the ﬁrst deﬁnition. Hence ϕ1 (1/2 . for let A1 and A2 be disjoint closed sets in X . Proof: A compact Hausdorff space X is paracompact. respectively. (2). with the weak or direct limit topology (a set is open iff it intersects each Xi in an open set) (3) CW complexes (4) metric spaces Note that (2) includes (3) for CW complexes with countably many cells. including: (1) compact Hausdorff spaces (2) unions of increasing sequences X1 ⊂ X2 ⊂ ··· of compact Hausdorff spaces Xi . then X is paracompact. since such a CW complex can be expressed as an increasing union of ﬁnite subcomplexes. each x ∈ X has a neighborhood in which only ﬁnitely many ϕβ ’s are nonzero. −1 The ﬁrst deﬁnition clearly implies the second by taking the cover {ϕβ (0. see [Dugundji] or [Lundell-Weingram]. An equivalent deﬁnition which is often given is that X is Hausdorff and every open cover of X has a locally ﬁnite open reﬁnement. Then β ϕβ (x) > 0 for all x . The next three propositions verify that the spaces in (1). If X is the direct limit of an increasing sequence X1 ⊂ X2 ⊂ ··· of compact Hausdorff spaces Xi . and let {ϕβ } be a partition of unity subordinate to the cover {X − A1 . Proposition 1.19. By Urysohn’s lemma there is a X contains a ﬁnite subcover. . X − A2 } . 1]} of neighborhood Vx with closure contained in some Uα . Let ϕi −1 ϕ1 + ϕ2 ≤ 1 since no ϕβ can be a summand of both ϕ1 and ϕ2 . The open cover {ϕx (0. and the fact that the two deﬁnitions are equivalent is rarely if ever needed. it can be shown that many manifolds are paracompact.

it sufﬁces to ﬁnd compact Hausdorff. {ϕβ Φγ } is a partition of unity on cell eγ have a partition of unity {ϕβ } on X n subordinate to the cover {Uα ∩ X n } . so each point of X has a neighborhood in which only ﬁnitely many ψij ’s are nonzero.20. This partition of unity extends the partition of unity {ϕβ Φγ } on S n and induces an extension of {ϕβ } to a partition n+1 and subordinate to {Uα } . Given an open cover {Uα } of a CW complex X . x) ∈ I × S n on D n+1 . 1−ε] and 1 on [1− ε/2 . Since S n is compact. We extend these functions ϕβ Φγ over D n+1 by the formula −1 a partition of unity on D n+1 subordinate to {Φγ (Uα )} . To show this. Such an f can be constructed inductively over the Xi ’s. (1 − ρε )ψγj } is −1 functions ρε ϕβ Φγ will be subordinate to the cover {Φγ (Uα )} . 1]. Thus when we normalize the collection {ψij } by dividing by i. extend each ϕij to a map ϕij : X →I with support in the same Uα . there is a ﬁnite partition of unity {ϕij } on Xi subordinate to Xi . where ρε : I →I is 0 on [0. for ﬁxed γ . Since Xi is {Uα ∩ Xi } . Let {ψγj } be a ﬁnite be nonzero. For a S n = ∂D n+1 . we get new maps ϕ max{ /2 i ij Proof: A preliminary observation is that X is normal. For each x ∈ X there is a ψij with ψij (x) > 0 . Doing this for all (n+1) cells of unity deﬁned on X n ∪eγ n+1 gives a partition of unity on X n+1 .j ψij we obtain a partition of unity on X subordinate to {Uα } . only ﬁnitely many of these compositions ϕβ Φγ can ρε (r )ϕβ Φγ (x) using ‘spherical coordinates’ (r . After we make such extensions for all n . Here is a technical fact about paracompact spaces that is occasionally useful: . Since 0 ≤ ψij ≤ ϕij . Using normality of X . ϕij − k<i σk } . Let ψij = max{0.36 Chapter 1 Vector Bundles a map f : X →I with f (A) = 0 and f (B) = 1 for any two disjoint closed sets A and B . In Vi all ψkj ’s with k > i are zero. σ }. since if ϕij (x) > 0 and i is minimal with respect to this condition. This sum is 1 on 1 . while in a cell eγ nonzero are the ones coming from ψγj ’s. suppose inductively that we n+1 with characteristic map Φγ : D n+1 →X . To prove that X is paracompact. If ε = εγ is chosen small enough. Let σi = by dividing it by j ϕij . Then {ρε ϕβ Φγ . then ψij (x) = ϕij (x) . we obtain a partition of unity deﬁned on all of X and subordinate to {Uα } . Proof: Every CW complex is paracompact. these extended −1 partition of unity on D n+1 subordinate to {Φγ (Uα )} . the collection {ψij } is subordinate to {Uα } . Proposition 1. The local ﬁniteness condition continues to eγ hold since near a point in X n only the extensions of the ϕβ ’s in the original partition n+1 the only other functions that can be of unity on X n are nonzero. For the induction step one has f deﬁned on the closed set Xi ∪ (A ∩ Xi+1 ) ∪ (B ∩ Xi+1 ) and one extends over Xi+1 by Tietze’s theorem. so if we normalize each ϕij with σi = 1 in a neighborhood Vi of Xi . using normality of the Xi ’s. let an open cover {Uα } be given.

and there is a partition of unity {ϕk } with ϕk supported in Proof: Let {ϕβ } be a partition of unity subordinate to {Uα } . This is clearly a disjoint union. let {ϕγ } be a partition of unity subordinate to the cover {Vk } .21. For each ﬁnite set S of functions ϕβ let VS be the subset of X where all the ϕβ ’s in S are strictly greater than all the ϕβ ’s not in S . Vk . so VS is open. Exercises 1. showing that it is the orthonormal n frame bundle associated to the vector bundle . VS is contained in some Uα . Since only ﬁnitely many ϕβ ’s are nonzero near any x ∈ X . Also. Let Vk be the union of all the open sets VS such that S has k elements. Given an open cover {Uα } of the paracompact space X . Show that the projection Vn (R k )→Gn (Rk ) is a ﬁber bundle with ﬁber O(n) by En (Rk )→Gn (Rk ) . since ϕβ ∈ S implies ϕβ > 0 on VS . namely. VS is deﬁned by ﬁnitely many inequalities among ϕβ ’s near x . For the second statement. any Uα containing the support of any ϕβ ∈ S . and let ϕk be the sum of those ϕγ ’s supported in Vk but not in Vj for j < k. The collection {Vk } is a cover of X since if x ∈ X then x ∈ VS where S = { ϕβ | ϕβ (x) > 0 } . there is a countable open cover {Vk } such that each Vk is a disjoint union of open sets each contained in some Uα .Appendix: Paracompactness 37 Lemma 1.

and this is contained in §2. . which gives isomorphisms K(S n ) ≈ K(S n+2 ) for all n . 2 . groups K(X) and K(X) with K(X) ≈ K(X) ⊕ Z . S 3 . and for real vector bundles.2 where the cohomological aspects of K(X) are the main focus. the dimensions of the real and complex numbers. and more generally K(X) ≈ K(S 2 X) where S 2 X is the double suspension of X . so we concentrate on this case in the present chapter. quaternions. We give some of these in §2. Complex K–theory turns out to be somewhat simpler than real K–theory. proving an isomorphism K(X × S 2 ) ≈ K(X) ⊗ K(S 2 ) . The proof of the Bott Periodicity Theorem divides into two parts. forming the bulk of §2. producing.The idea of K–theory is to make the direct sum operation on real or complex vector bundles over a ﬁxed base space X into the addition operation in a group. and Cayley octonions. The other half is easier. 4 . groups KO(X) and KO(X) with KO(X) ≈ KO(X) ⊕ Z .1. and 8 . (2) The nonparallelizability of spheres other than S 1 . notably: (1) The nonexistence of division algebras over R in dimensions other than 1 . This is a deep theorem. The ﬁrst is the hard technical work. in the case of complex vector bundles. being more formal in nature. The case X = S n is the Bott Periodicity Theorem. so it is not surprising that it has applications of real substance. Computing K(X) even for simple spaces X requires some work.3. There are two slightly different ways of doing this. and S 7 . This takes about ten pages.

so we can form for compact X an abelian group K(X) consisting of formal differences E−E of vector bundles E and E over X . hence are ﬁnite in number by compactness. The Functor K(X) Since we shall be dealing exclusively with complex vector bundles in this chapter. so the dimensions of ﬁbers must be locally constant over X . then the set of ∼ equivalence classes of vector bundles over X forms an abelian group with respect to ⊕ . commutative. Just as the positive rational numbers are constructed from the positive integers by forming quotients a/b with the equivalence relation a/b = c/d iff ad = bc .The Functor K(X) Section 2. If X is compact Hausdorff. we do have the cancellation property that E1 ⊕ E2 ≈s E1 ⊕ E3 implies E2 ≈s E3 over a compact base space X . In the general case let Xi = { x ∈ X | dim π −1 (x) = i } . tion of ‘vector bundle’ which allows the ﬁbers of a vector bundle p : E →X to be vecFor the purposes of K–theory it is convenient to take a slightly broader deﬁni- h : p −1 (U)→U × Cn . Consider vector bundles over a ﬁxed base space X . Base spaces will always be assumed to be compact Hausdorff. We still assume local trivializations of the form stably isomorphic. Deﬁne two vector bundles E1 and E2 over X to be we set E1 ∼ E2 if E1 ⊕ εm ≈ E2 ⊕ εn for some m and n . this is Proposition 1. with the equivalence relation E1 − E1 = E2 − E2 iff E1 ⊕ E2 ≈s E2 ⊕ E1 . which can only happen if E and E are 0 dimensional. the class of ε0 .1. However. only the zero element. which we do by showing →X such that E ⊕ E ≈ εm for some m . let us take ‘vector bundle’ to mean ‘complex vector bundle’ unless otherwise speciﬁed. If all the ﬁbers of E have the same dimension. since we can add to both sides of E1 ⊕ E2 ≈s E1 ⊕ E3 a bundle E1 such that E1 ⊕ E1 ≈ εn for some n . It is easy to see that both ≈s and ∼ are equivalence relations. This group is called K(X) . can have an inverse since E ⊕ E ≈s ε0 implies E ⊕ E ⊕ εn ≈ εn for some n . if E1 ⊕ εn ≈ E2 ⊕ εn for some n .1 39 2. In a similar vein Proposition 2. vector bundle we write as εn →X . These Xi ’s are disjoint open | sets in X . On equivalence classes of either sort the operation of direct sum is well-deﬁned. A zero element is the class of ε0 . even though inverses do not exist. that for each vector bundle π : E →X there is a bundle E Proof: Only the existence of inverses needs to be shown. and associative. so that the results of the preceding chapter which have this hypothesis will be available to us. By adding to E a bundle which over each Xi is a trivial bundle of suitable dimension we can produce a bundle whose ﬁbers all have the same dimension. written E1 ≈s E2 . but if X is disconnected the dimensions of ﬁbers need not be globally constant. Verifying transitivity of this relation involves the cancellation . For the direct sum operation on ≈s equivalence classes.4.1. The trivial n dimensional tor spaces of different dimensions.

we have induced maps f ∗ : K(Y )→K(X) with the same properties. then the map K(X)→K(x0 ) obtained by restricting vector bundles to their ﬁbers over x0 is a ring homomorphism. and the inverse of E − E is E − E . Similarly. This is a ring homog The rings K(X) and K(X) can be regarded as functors of X . the trivial line bundle. restriction of vector bundles to a basepoint x0 ∈ X deﬁnes a homomorphism Ring Structure Besides the additive structure in K(X) there is also a natural multiplication coming from tensor product of vector bundles. If we choose a basepoint x0 ∈ X . though not necessarily a ring with identity. For example. duces a map f ∗ : K(Y )→K(X) . depending on the choice of x0 . This is consistent with familiar arithmetic rules. so for arbitrary elements of K(X) represented by differences of vector bundles. their product in K(X) is deﬁned by the formula (E1 − E1 )(E2 − E2 ) = E1 ⊗ E2 − E1 ⊗ E2 − E1 ⊗ E2 + E1 ⊗ E2 It is routine to verify that this is well-deﬁned and makes K(X) into a commutative ring with identity ε1 . of E . is an ideal. using the basic properties of tensor product of vector bundles described in §1. hence E ≈s εm for some m .40 Chapter 2 Complex K–Theory property. This subgroup {εm − εn } of K(X) is isomorphic to Z . which is why compactness of X is needed. so the kernel consists of the hence E ∼ E . This is well-deﬁned since if E − εn = E − εm in K(X) . =g ∗ all follow from the corresponding properties for pullbacks of vector . K(X) is then a group. With the obvious addition rule (E1 − E1 ) + (E2 − E2 ) = (E1 ⊕ E2 ) − (E1 ⊕ E2 ) . The zero element is the equivalence class of E − E for any E . and its kernel consists of elements of the form εm − εn . We can simplify notation by writing the element εn ∈ K(X) just as n .1. elements E −εn with E ∼ ε0 . hence also a ring in its own right. then E ⊕ εm ≈s E ⊕ εn . the product nE is the sum of n copies of E . Note that every element of K(X) can be represented as a difference E − εn since if we start with E − E we can add to both E and E a bundle E There is a natural homomorphism K(X)→K(X) sending E − εn to the ∼ class such that E ⊕ E ≈ εn for some n . A map f : X →Y in- morphism since f ∗ (E1 ⊕ E2 ) ≈ f ∗ (E1 ) ⊕ f ∗ (E2 ) and f ∗ (E1 ⊗ E2 ) ≈ f ∗ (E1 ) ⊗ f ∗ (E2 ) . K(X)→K(x0 ) ≈ Z which restricts to an isomorphism on the subgroup {εm − εn } . Its kernel. For elements of K(X) represented by vector bundles E1 and E2 their product in K(X) will be represented by the bundle E1 ⊗ E2 . which can be identiﬁed with K(X) as we have seen. In fact. Thus we have a splitting K(X) ≈ K(X) ⊕ Z . 1 1 The functor properties (f g)∗ = g ∗ f ∗ and 1 ∗ = 1 as well as the fact that f implies f ∗ except that for f ∗ to be a ring homomorphism we must be in the category of base- bundles. The map K(X)→K(X) is obviously surjective. to distinguish it from K(X) . The group K(X) is sometimes called reduced. sending E −E to f ∗ (E)−f ∗ (E ) .

2. with multiplication deﬁned by (a ⊗ b)(c ⊗ d) = ac ⊗ bd . Taking Y to be S 2 we have an external product µ : K(X) ⊗ K(S 2 )→K(X × S 2 ) Most of the remainder of §2. The map Z[H]/(H − 1)2 →K(S 2 ) is an isomorphism of rings. For example. which is quite different from tensor product. The tensor product of rings is a ring.The Functor K(X) Section 2.1 41 pointed spaces and basepoint-preserving maps since our deﬁnition of multiplication for K required basepoints. note that an additive basis for We deﬁne a homomorphism µ as the composition µ : K(X) ⊗ Z[H]/(H − 1)2 Z[H]/(H − 1)2 is {1. Instead we will sometimes use the notation a ∗ b as shorthand for µ(a ⊗ b) . This is the essential core of the proof of Bott Periodicity. The homomorphism µ : K(X) ⊗ Z[H]/(H − 1)2 →K(X × S 2 ) is an iso- morphism of rings for all compact Hausdorff spaces X . so we have a natural ring homomorphism Z[H]/(H − 1)2 →K(S 2 ) whose domain is the quotient ring of the polynomial ring Z[H] by the ideal generated by (H − 1)2 . when X is a point this will yield a calculation of K(S 2 ) . We showed in Example 1. In K(S 2 ) this is the formula H 2 + 1 = 2H . and ∗ ∗ µ is a ring homomorphism since µ((a ⊗ b)(c ⊗ d)) = µ(ac ⊗ bd) = p1 (ac)p2 (bd) = ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ p1 (a)p1 (c)p2 (b)p2 (d) = p1 (a)p2 (b)p1 (c)p2 (d) = µ(a ⊗ b)µ(c ⊗ d) . in particular K(S n ) for all n . Let H be the canonical line bundle over S 2 = CP1 . The external product in ordinary cohomology is called ‘cross product’ and written a× b . Theorem 2. In particular.13 that (H ⊗ H) ⊕ 1 ≈ H ⊕ H . The Fundamental Product Theorem The key result allowing the calculation of K(X) in nontrivial cases is a certain formula that computes K(X × S 2 ) in terms of K(X) . We can also write this as (H − 1)2 = 0 .3. ∗ ∗ p1 (a)p2 (b) An external product µ : K(X) ⊗ K(Y )→K(X × Y ) can be deﬁned by µ(a ⊗ b) = where p1 and p2 are the projections of X × Y onto X and Y . → K(X) ⊗ K(S 2 ) → K(X × S 2 ) where the second map is the external product. In the next section we will deduce other calculations. Taking X to be a point we obtain: Corollary 2. then it is generated as . an abelian group by H − 1 . H} . Since we have the relation (H − 1)2 = 0 .1 will be devoted to showing that this map is an isomorphism. but to use this symbol for the K–theory external product might lead to confusion with Cartesian product of vector bundles. so H 2 = 2H − 1 . this means that Thus if we regard K(S 2 ) as the kernel of K(S 2 )→K(x0 ) .

42 Chapter 2 Complex K–Theory the multiplication in K(S 2 ) is completely trivial: The product of any two elements is zero. Z) . z−1 ] . . Z) and H ∗ (S 2 . f1 ] over X × S 2 × {0} and X × S 2 × {1} . Generalizing this. ‘ z ’ means scalar multiplication by z ∈ S 1 ⊂ C . zn ] ≈ H n . Taking X to be a point. then we showed in Example 1. and call f a clutching function for [E. In the case of ordinary cohomology the cup product of a generator of H 2 (S 2 . From E and f we construct a vector bundle 1 bundle p× 1 : E × S 1 →X × S 1 . creating vector bundles over X × S 2 by gluing together two vector bundles over X × D 2 by means of a generalized clutching function. The main work in proving the theorem will be to prove the surjectivity of µ . f0 ] and [E. then [E. so the reader who wishes to defer a careful reading of the proof may skip ahead to §2. zn f ] ≈ [E. which is justiﬁed by the fact that H ⊗ H −1 ≈ 1 . Nothing in the proof will be used elsewhere in the book except in the proof of Bott periodicity for real K–theory in §2. f1 ] since f . let f : E × S 1 →E × S 1 be an automorphism of the product vector We begin by describing this more general clutching construction. with H − 1 behaving exactly like the generator of H 2 (S 2 . [E. The formula [1. f2 ] ≈ [E1 ⊕ E2 . the n fold tensor product of H with itself. More generally we have [1. z) : p −1 (x)→p −1 (x) . 4. [E. f speciﬁes an from the homotopy ft we can construct a vector bundle over X × S 2 × I restricting evident that [E1 .10 that [1. If isomorphism f (x. zn ] ≈ E ∗ H n = µ(E ⊗ H n ) for n ∈ Z .2 we know that vec- which we called clutching functions. 1 is the external product E ∗ 1 = µ(E ⊗ 1) . [E. f ] ⊗ H n where H n denotes the pullback of H n via the projection X × S 2 →S 2 .2 without any loss of continuity. The present section will be devoted entirely to the proof of the preceding theorem. f ] . Z) = 0 . f0 ] ≈ [E. f ] . We write this bundle as [E. Injectivity will then be proved by a closer examination of the surjectivity argument.4. To prove the theorem we will generalize this construction. to [E. and H is the canonical line bundle over S 2 = CP1 . Given a vector over X × S 2 by taking two copies of E × D 2 and identifying the subspaces E × S 1 via ft : E × S 1 →E × S 1 is a homotopy of clutching functions. Thus for each x ∈ X and z ∈ S 1 . bundle p : E →X . From the classiﬁcation of vector bundles over spheres in §1. f1 ⊕ f2 ] . or equivalently the pullback of E 1] via the projection X × S 2 →X . 3. From the deﬁnitions it is 2. zn ] ≈ H n holds also for negative n if we deﬁne H −1 = [1. f1 ] ⊕ [E2 . Z) with itself is automatically zero since H 4 (S 2 . Clutching Functions tor bundles over S 2 correspond exactly to homotopy classes of maps S 1 →GLn (C) . whereas with K–theory a calculation is required. z] ≈ H where ‘ 1 ’ is the trivial line bundle over X . Here are some examples of bundles built using clutching functions: 1. Readers familiar with cup product in ordinary cohomology will recognize that the situation is exactly the same as in H ∗ (S 2 . Z) .

i polynomial clutching functions. Laurent polynomial clutching functions t (x. θ) = the geometric series n∈Z an (x)r |n| einθ .The Functor K(X) Every vector bundle E Section 2. by comparison with n r n . and this pullback is E × Dα . Any two choices of normalized hα are homotopic through normalized hα ’s since they differ by a map gα from Dα to the automorphisms of E . θ) ranges over X × S 1 . t)zi . and let Eα for α = 0. For a compact space X we wish an (x)zn = |n|≤N an (x)einθ .1 43 see this. The ﬁrst step is to reduce to Laurent ai : E →E restricts to a linear transformation ai (x) in each ﬁber p be invertible. We may assume a clutching function f is normalized to be the identity over and D∞ . then sums of ﬁnitely many terms in the series for u(x. The strategy of the proof will be to reduce from arbitrary clutching functions to successively simpler clutching functions. The projection X × Dα →X × {1} is homotopic to the projection. θ) approaches f (x. we set S1 For positive real r let u(x. . where each an is a continuous function X →C . z) 0 polynomial clutching function and which are homotopic through clutching functions are homotopic by a Laurent = i polynomial clutching function homotopy ai (x. ∞ X × {1} since we may normalize any isomorphism hα : Eα →E × Dα by composing it over each X × {z} with the inverse of its restriction over X × {1} . and such a gα is homotopic 1 1 to the constant map 1 by composing it with a deformation retraction of Dα to 1 . r . ∞ be the restriction of E over X × Dα . To identity map of X × Dα . r .4. Thus any two choices f0 and f1 of normalized clutching functions are joined by a homotopy of normalized clutching functions ft . θ)e−inθ dθ Motivated by Fourier series. ] for some Laurent . r . so the bundle Eα is isomorphic to the pullback of E by the Then f = h0 h−1 is a clutching function for E . For ﬁxed r < 1 . to approximate a continuous function f : X × S 1 →C by Laurent polynomial functions |n|≤N Before beginning the proof we need a lemma. with gα (1) = 1 . f ] is isomorphic to [E. which have the form (x. an (x) = 1 2π f (x. with E the →X × S 2 1 is isomorphic to [E. so we have an isomorphism hα : Eα →E × Dα . though their linear combination are automorphisms. We call such an ai an endomorphism of E since the linear transformations ai (x) need not ai (x)zi is since clutching functions Proposition 2. let the unit circle S ⊂ C ∪ {∞} = S 2 decompose S 2 into the two disks D0 restriction of E over X × {1} . since compactness of X × S 1 implies that |f (x. z) = |i|≤n ai (x)z −1 i where (x) . If we can show that u(x. f ] for some E and f . 1 Every vector bundle [E. θ) uniformly in x and θ as r goes to 1 . this series converges absolutely and uniformly as (x. θ)| is bounded and hence also |an (x)| . θ) with r near 1 will give the desired approximations to f by Laurent polynomial functions.

ϕ) = 1 2π ∞ n=−∞ r |n| einϕ for 0 ≤ r < 1 and ϕ ∈ R Then u(x. P (r . θ) = S1 P (r . θ) dt S1 P (r . θ) dt Given ε > 0 . there exists a δ > 0 such that |f (x. P (r . δ) . t) − f (x. Now to show uniform convergence of u(x. ϕ) . t) dt − S1 P (r . θ − t) on |t − θ| ≥ δ is P (r . Then we have Iδ ≤ |t−θ|≤δ P (r . θ) to f (x. θ − t) ε dt ≤ ε S1 P (r . of period 2π . θ) uniformly in x and θ . δ) S1 |f (x. t) − f (x. r .44 Chapter 2 Complex K–Theory Lemma 2. ϕ) approaches 0 and the denominator approaches 2 − 2 cos ϕ ≠ 0 . ϕ) ≥ P (r .5. r . t) dt 1 2π where the order of summation and integration can be interchanged since the series in the latter formula converges uniformly. by comparison with the geometric series n r n . Deﬁne the Poisson kernel P (r . ϕ) = Three basic facts about P (r . θ)| < ε for |t − θ| < δ and all x . π ] . θ) − f (x. By summing the two geometric series for 1 1 − r2 · 2π 1 − 2r cos ϕ + r 2 positive and negative n in the formula for P (r . θ)→f (x. t) dt . θ)| dt over the interval |t − θ| ≤ δ and let Iδ denote this integral over the rest of S 1 . θ) we ﬁrst observe that. we have u(x. t) dt r |n| ein(θ−t) f (x. θ) = = 1 2π n=−∞ S1 S1 ∞ n=−∞ r |n| ein(θ−t) f (x. ϕ) is even. Let Iδ denote the integral P (r . ϕ) dϕ = 1 for each r < 1 . u(x. P (r . ∞ For r < 1 we have u(x. θ)| dt . π ) > 0 for all r < 1. ϕ) which we shall need are: (a) As a function of ϕ . In particular we have P (r . θ − t) dt = ε By (a) the maximum value of P (r . θ − t)f (x. θ − t) f (x. and monotone decreasing on [0. r . θ − t)f (x. t) − f (x. ϕ)→0 as r →1 since the numerator of P (r . since the same is true of cos ϕ which appears in the denominator of P (r . Proof: As r →1 . one computes that P (r . π ) . as one sees by integrating the series for (c) For ﬁxed ϕ ∈ (0. (b) S1 P (r . using (b). θ − t)f (x. r . θ) = ≤ S1 P (r . r . ϕ) with a minus sign. t) − f (x. since f is uniformly continuous on the compact space X × S 1 . So Iδ ≤ P (r . θ − t) |f (x. ϕ) term by term.

q] ⊗ H −m . Ln q] for a linear clutching function Ln q . From the Laurent polynomial approximations = i ϕi over Xi we form the convex linear combination approximating f over all of X × S 1 . so balls in End(E × S 1 ) are convex. Via hi . Thus by (c) we can make Iδ ≤ ε by taking r close enough to 1 .The Functor K(X) Section 2. To show that every f ∈ End(E × S 1 ) can be approximated by Laurent polynomial hi : p −1 (Ui )→Ui × Cni . since a sufﬁciently close Laurent polynomial approximation f via the linear homotopy t + (1 − t)f through clutching functions. θ) − f (x. By the lemma we can ﬁnd Laurent polynomial matrices approximates f in the · i (x. The subspace Aut(E × S 1 ) of automorphisms is open in the topology deﬁned by this norm since it is the preimage of (0. θ)| ≤ Iδ + Iδ ≤ 2ε . 1 ≈ [(n + 1)E. then 1] [E. ﬁrst choose open sets Ui covering X together with isomorphisms to the standard inner product in Cni .6.1 45 The integral here has a uniform bound for all x and θ since f is bounded. Therefore |u(x. The triangle inequality holds for this norm. by applying the Gram-Schmidt process to h−1 i whose i i entries uniformly approximate those of f (x. It follows easily that norm. the linear maps f (x. α will sufﬁce to show that Laurent polynomials are dense in End(E × S 1 ) . q] ⊕ [nE. inf (x. Thus to prove the ﬁrst statement of the lemma it to f will then be homotopic to to t. ∞) . the endomor- phisms of E × S 1 form a vector space End(E × S 1 ) with a norm α = sup|v|=1 |α(v)| . Proposition 2. The entries of these matrices deﬁne functions Xi × S 1 →C . a compact set in Ui . i . z))| . r . ∞) under the continuous map End(E × S 1 )→[0. Proof of Proposition 2. We may assume hi takes the chosen inner product in p −1 (Ui ) of the standard basis vectors. z) for x ∈ Xi can be viewed as matrices. 1.z)∈X × S 1 | det(α(x. . ] ≈ [E.4: Choosing a Hermitian inner product on E . The next step is to reduce polynomial clutching functions to linear clutching functions. z) endomorphisms. a Laurent polynomial A Laurent polynomial clutching function can be written = z−m q for a polyno- mial clutching function q . by a Laurent polynomial homotopy ing this with linear homotopies from from 0 0 then combint to 0 and 1 to 1 to obtain a homotopy to 1. The second statement follows similarly by approximating a homotopy from 1 0 viewed as an automorphism of E × S × I . z) for x ∈ Xi . and then we have [E. If q is a polynomial clutching function of degree at most n . Let {ϕi } be a partition of unity subordinate to {Ui } and let Xi be the support of ϕi .

. . deﬁnes an endomorphism of (n + 1)E by interpreting the (i. Since homotopic clutching functions produce isomorphic 1] bundles. . there is a splitting E ≈ E+ ⊕ E− with [E. . . . each entry in A becomes a matrix itself. the expanded version of A is also invertible in each ﬁber. These row and column operations are not quite elementary row and column operations in the traditional sense since the entries of the matrices are not numbers but linear maps. for z ∈ S 1 . Since elementary row and column operations preserve determinant. ··· 1 0 ··· 0 q ··· 0 ··· 0 ··· 0 . . a(x)z + b(x)] . Proof: (∗) Given a bundle [E. .7. A= . . . a(x)z + b(x)] ≈ [E+ . . we obtain an isomorphism [E. with the entries Linear Clutching Functions For linear clutching functions a(x)z + b(x) we have the following key fact: Proposition 2. This means that A is an automorphism of (n + 1)E for each z ∈ S 1 . Since Ln q has the form A(x)z + B(x) . . However. For example the ﬁrst operation can be achieved by adding tz times the ﬁrst column to the second. q] ⊕ [nE. . j) entry of the matrix as 1 denoting the identity E →E and z denoting z times the identity. 0 0 0 0 0 0 ··· 1 −z 0 0 0 an an−1 an−2 ··· a1 a0 0 0 0 . Ln q] . The matrices A and B deﬁne homotopic clutching functions since the elementary row and column operations can be achieved by continuous one-parameter families of such operations. (1 + tz) a(x) z+t + b(x) = a(x) + tb(x) z + ta(x) + b(x) 1 + tz Consider the expression: .46 Chapter 2 Complex K–Theory Proof: Let q(x. . hence in each ﬁber the expanded version of B has nonzero determinant. . This produces 0 ’s above the diagonal and the polynomial q in the lower right corner. z] . it is a linear clutching function. 1 ⊕ [E− . We can pass from the matrix A to the matrix B by a sequence of elementary row and column operations in the following way. a linear map from the j th summand of (n + 1)E to the i th summand. then z times the second column to the third. . and then each of the preceding row and column operations can be realized by a sequence of traditional row and column operations on the expanded matrices. B=. 1 ≈ [(n + 1)E. by restricting to a ﬁber of E and choosing a basis in this ﬁber. . 1 ⊕ [E. . z) = an (x)zn + ··· + a0 (x) . and therefore determines a clutching function which we denote by Ln q . 1] The matrix B is a clutching function for [nE. . 1] The ﬁrst step is to reduce to the case that a(x) is the identity for all x . Each of the matrices 1 0 0 1 −z 0 ··· 0 0 0 1 0 0 1 −z ··· 0 0 0 0 1 0 0 1 ··· 0 0 . . . . . . Then for each i ≤ n . add z times the ﬁrst column to the second column. In A . with t ranging from 0 to 1 . . and so on. subtract the appropriate multiple of the i th row from the last row to make all the entries in the last row 0 except for the ﬁnal q . q] .

Since z + b(x) is invertible for all x . in C[t] as q+ (t)q− (t) where q+ (t) has all its roots outside S 1 and q− (t) has all its . Therefore (∗) deﬁnes a homotopy of clutching functions as t goes from 0 to t0 < 1 . and similarly Ker q− (T ) = 1 Im q+ (T ) . Let b : E →E be an endomorphism having no eigenvalues on the unit (c) The eigenvalues of b | E+ all lie outside S 1 and the eigenvalues of b | E− all lie | | inside S 1 . Let V± be the kernel of q± (T ) : V →V . let q± be the characteristic | polynomial of T | V± . Assuming q(t) has no roots on S 1 . We have T (V± ) ⊂ V± since q± (T )(v) = 0 implies q± (T )(T (v)) = T (q± (T )(v)) = 0 . Thus Ker q+ (T ) = Im q− (T ) . and then multiplying by the nonzero scalar 1 + tz . 1 and from r (T )q+ (T ) + s(T )q− (T ) = 1 we deduce that Ker q+ (T ) ∩ Ker q− (T ) = 0 . so the factorizations q = q+ q− and q = q+ q− must coincide up to scalar factors. All eigenvalues of T | V± are roots of q± since q± (T ) = 0 on V± . so must be q± (T ) . reducing to the case of clutching functions of the form z + b(x) . All the linear factors of q± must be factors of q± by condition (c). Since q± (T ) is identically zero on V± . From q+ (T )r (T ) + q− (T )s(T ) = 1 we see that Im q+ (T ) + Im q− (T ) = V .The Functor K(X) Section 2. f g] ≈ [E. b(x) has no eigenvalues on the unit circle S . Then there are unique subbundles E+ and E− of E such that : (a) E = E+ ⊕ E− . we may factor q(t) roots inside S 1 . To see the uniqueness of V+ and V− satisfying (a)–(c). 1 Lemma 2. Since V = V+ ⊕ V− and V = Ker q+ (T ) ⊕ Ker q− (T ) . This establishes the uniqueness of V± . Now we use the simple fact that [E. Therefore a(x) + tb(x) is invertible for t = t0 near 1 . circle S 1 . Consider ﬁrst the algebraic situation of a linear transformation T : V →V with Proof: characteristic polynomial q(t) . Hence V = V+ ⊕ V− . as the continuous function t 1 . inf x∈X det[a(x) + tb(x)] is nonzero for t = 1 .1 47 When t = 0 this equals a(x)z + b(x) . and the opposite inclusion follows from 1 r (T )q+ (T ) + q− (T )s(T ) = 1 . f ] for any isomorphism g : E →E . hence V± ⊂ Ker q± (T ) . hence also for t near This allows us to replace the clutching function on the right-hand side of (∗) by the clutching function z + [t0 a(x) + b(x)][a(x) + t0 b(x)]−1 . (∗) deﬁnes an invertible linear transformation since the left-hand side is obtained from a(x)z + b(x) by ﬁrst applying the substitution z (z + t)/(1 + tz) which takes S 1 to itself (because if |z| = 1 then |(z +t)/(1+tz)| = |z(z +t)/(1+tz)| = |(1+tz)/(1+tz)| = |w/w| = 1 ). we have Im q− (T ) ⊂ Ker q+ (T ) . Since q+ and q− are relatively prime in C[t] . For 0 ≤ t < 1 . we must have V± = Ker q± (T ) . so q = q+ q− . In the right-hand side of (∗) the term a(x) + tb(x) is invertible for t = 1 since it is the restriction of a(x)z + b(x) to z = 1 . | Thus conditions (a)–(c) hold for V+ and V− .8. there are polynomials r and s with r q+ +sq− = 1 . From q+ (T )q− (T ) = q(T ) = 0 . (b) b(E± ) ⊂ E± .

48 Chapter 2 Complex K–Theory As T varies continuously through linear transformations without eigenvalues on S . Hence 1] [E+ . 1 This is because the ﬁrst step of reducing to the case a = 1 clearly respects sums. am inside C . assuming that q . It follows that the union E± of the subspaces V± in all the ﬁbers V of E is a subbundle. q− (T )(vk ) is a basis for V+ and q+ (T )(vk+1 ). Since the ai ’s are near a . Since b+ (x) has all its eigenvalues outside S 1 . note that the lemma gives a splitting [E.7. In this situation we assert that the factors q± of q vary continuously with q . b+ (x)] ≈ [E+ . the splitting V = V+ ⊕ V− also varies continuously. az + b] ≈ [E+ . z] . in the sense that the splitting for a sum [E1 ⊕ E2 . small perturbations of q produce polynomials q which also have m roots a1 . 1] For future reference we note that the splitting [E. q+ (T )(vn ) is a basis for V− . recall that V+ = Im q− (T ) and V− = Im q+ (T ) . the formula tz + b+ (x) for 0 ≤ t ≤ 1 deﬁnes a homotopy of clutching functions from z + b+ (x) to b+ (x) . so the factor (z − a)m of q becomes a factor (z − a1 ) ··· (z − am ) of the nearby q . counted with multiplicity. 1 ⊕ [E− . γ(z) = q(z)/|q(z)| . so q(t) = p(t)(t − a)m with p(a) ≠ 0 . The homotopy γs (z) = p(sa + (1 − s)z)(z − a)m |p(sa + (1 − s)z)(z − a)m | the map γ : C →S 1 . the number of roots of q inside C . equals the degree of case of a small circle C about a root z = a of multiplicity m . For nearby T these vectors vary continuously. z + b− (x)] ≈ [E− . z + tb− (x) deﬁnes a homotopy of clutching functions from z + b− (x) to z . Thus the splitting V = Im q− (T ) ⊕ Im q+ (T ) continues to hold for nearby T . . z + b− (x)] where b+ and b− are the restrictions of b . Choose a basis v1 . Similarly. where it is clear that the degree is m . To ﬁnish the proof of Proposition 2. these factors of q and q are close. (a1 z+b1 )⊕(a2 z+b2 )] has (E1 ⊕ E2 )± = (E1 )± ⊕ (E2 )± . ··· . and so the proof of the lemma is complete. vn for V such that q− (T )(v1 ). so [E− . q+ . z] constructed in the proof of Proposition 2. Thus for a small circle C about a root z = a of q of multiplicity m . and so the splitting V = V+ ⊕ V− varies continuously with T . 1 . z + b+ (x)] ⊕ [E− . z + b+ (x)] ≈ [E+ .7 preserves direct sums. and the uniqueness of the ± splitting in Lemma 2. ··· . ··· . hence remain independent.8 guarantees that it preserves sums. its characteristic polynomial q(t) varies continuously through polynomials without roots in S 1 . z + b(x)] ≈ [E+ . To see this. To see this we shall use the fact that for any circle C in C disjoint from the roots of q . and q− are normalized to be monic polynomials. ··· . and so q± is close to q± . Next we observe that as T varies continuously through transformations without eigenvalues in S 1 . To prove this fact it sufﬁces to consider the 1 gives a reduction to the case (t − a)m .

. The idea will be to deﬁne ν([E. . . Similarly. it will be useful to know the ‘ ± ’ splittings for these two bundles: 1] (3) For [E. . In the ﬁrst matrix we can add z times the ﬁrst column to the second column to eliminate the −z in the ﬁrst row. (4) For [E. . 1 H −m + [((n + 1)E)− . . f ] = [E. Ln+1 (zq)] ≈ [(n + 1)E. ··· 1 ··· a0 −z 0 0 0 . Ln q] ⊕ [E. z] in (1) and (2). the clutching function −z being the composition of the clutching function z with the automorphism −1 of E . . . Since every vector bundle over X × S 2 has the form [E. f ] .1 49 Now we address the question of showing that the homomorphism µ : K(X) ⊗ Z[H]/(H − 1)2 →K(X × S 2 ) is an isomorphism. . −z] ≈ [E. . Ln q] ⊕ [E. . . 0 0 0 0 1 −z 0 0 an an−1 ··· a0 an an−1 ··· ··· 0 −z . where deg q ≤ n : 1] (1) [(n + 2)E. f ] displayed above on m and n we ﬁrst derive the following two formulas. . The preceding propositions imply that in K(X × S 2 ) we have [E. in the second matrix we add z−1 times the last column to the next-to-last column to make the −z in the last column have all zeros in its row and column. which gives the splitting in (2) since [E. Ln q]⊗H −m − [nE. Ln+1 q] ≈ [(n + 1)E. Statement (4) is obvious from the deﬁnitions since the clutching function z is already in the form z + b(x) with b(x) = 0 . . z] the summand E+ is 0 and E− = E . f ]) as some linear combination of 1 terms E ⊗ H k and ((n + 1)E)± ⊗ H k that is independent of all choices. 1 while the rest of the matrix gives [(n + 1)E. 1 (2) [(n + 2)E. 1] In view of the appearance of the correction terms [E. q]⊗H −m = [(n + 1)E. 0 an−2 0 0 . . z] . . 1 H −m 1]⊗ 1]⊗ = ((n + 1)E)+ ∗ H −m + ((n + 1)E)− ∗ H 1−m − nE ∗ H −m This last expression is in the image of µ . To show µ is injective we shall construct ν : K(X × S 2 )→K(X) ⊗ Z[H]/(H − 1)2 such that νµ = 1 . z−m q] = [E. To investigate the dependence of the terms in the formula for [E. This proves (1).The Functor K(X) Conclusion of the Proof Section 2. . . 1 H −m 1]⊗ = [((n + 1)E)+ . . z] The matrix representations of Ln+1 q and Ln+1 (zq) are: 1 −z 0 ··· 0 1 −z −z ··· 0 1 0 1 0 . . Ln q] . . it follows that µ is surjective. and . and then the ﬁrst row and column give the summand 1] [E. so 0 is the only eigenvalue of b(x) and hence . z]⊗H −m − [nE. 1 the summand E− is 0 and E+ = E . 1 and [E. .

z−m−1 (zq)]) = ((n + 1)E)− ⊗(H − 1) + E⊗(H − 1) + E⊗H −m−1 = ((n + 1)E)− ⊗(H − 1) + E⊗(H −m − H −m−1 ) + E⊗H −m−1 = ((n + 1)E)− ⊗(H − 1) + E⊗H −m = ν([E.7 preserves K(X × S 2 )→K(X) ⊗ Z[H]/(H − 1)2 . Another choice which might perhaps affect the value of ν([E.8. as we have already observed. We claim that this is well-deﬁned. not on 1 and H . z−m q] .6 and 2. The group Z[H]/(H − 1)2 is generated by It remains to see that ν([E. . So the ‘minus’ summand is independent of n . To see that it is independent of m we must see that it is unchanged when z−m q is replaced by z−m−1 (zq) . Since t0 1 has only the one eigenvalue function z + [t0 a(x) + b(x)][a(x) + t0 b(x)]−1 with 0 < t0 < 1 .7. z−m q]) . z−m q]) = ((n + 1)E)− ⊗ (H − 1) + E ⊗ H −m ∈ K(X) ⊗ Z[H]/(H − 1)2 for n ≥ deg q . z−m q]) The second equality use the relation (H −1)2 = 0 which implies H(H −1) = H −1 and hence H − 1 = H −m − H −m−1 for all m .50 Chapter 2 Complex K–Theory E+ = 0 . The third and fourth equalities are evident. Specializing to the −1 t0 > 1 . z−m q]) does not depend on n . and in Proposition 2. we This ﬁnishes the veriﬁcation that ν([E. z−m q]) is well-deﬁned. We showed that every bundle over X × S 2 has the form [E.8 to the endomorphism of E × I given by the homotopy. z−m q]) depends only on the bundle [E. Suppose we deﬁne ν([E. By (2) and (4) we have the ﬁrst of the following equalities: ν([E. sums. Hence the choice of t0 does not affect ν([E.5 we showed that Laurent polynomial approximations to homotopic f ’s are Laurent-polynomial-homotopic. If we apply Propositions 2. we have E− = 0 .7 which replaces a clutching function a(x)z + b(x) by the clutching −1 −1 1 1 1 case a(x)z + b(x) = 1 this yields z + t0 1 . conclude that the two bundles ((n + 1)E)− over X × {0} and X × {1} are isomorphic. the clutching function z−m q for this bundle. so varying t0 gives a homotopy of the endomorphism b in Lemma 2. This could be any number sufﬁciently close to 1 . This has no effect on the ± splitting since we can apply Lemma 2. f ] for a normalized clutching function f which was unique up to homotopy. which follows from (H − 1)2 = 0 .7 over X × I with a Laurent polynomial homotopy as clutching function. It is easy to check through the deﬁnitions to see that ν takes sums to sums since Ln (q1 ⊕ q2 ) = Ln q1 ⊕ Ln q2 and the ± splitting in Proposition 2. using the fact that the ± splitting preserves direct sums. so ν([E. To obtain (3) we ﬁrst apply the procedure at the beginning of the proof of Proposition 2. Formulas (1) and (3) give ((n + 2)E)− ≈ ((n + 1)E)− . We have just noted that ‘minus’ summands are independent of n . It follows that ν extends to a homomorphism 1 The last thing to verify is that νµ = 1 . z−m q]) is the constant t0 < 1 in the proof of Proposition 2. so in view of the relation H + H −1 = 2 .

Since we assume A is a closed subspace of a compact Hausdorff space. We begin by examining a key property of the reduced groups K(X) . A sequence of homomorphisms G1 →G2 → ··· →Gn is said to be exact if at coming map. hence a point x ∈ X − A and the subspace A have disjoint neighborhoods in X . The quotient space X/A is compact Hausdorff as well. let E/h be For the opposite inclusion Ker i∗ ⊂ Im q∗ . suppose the restriction over A of a . We have νµ(E ⊗ H −m ) = ν([E. and in particular for all spheres. it follows that i∗ q∗ = 0 . Thus it sufﬁces to show νµ = 1 1 on elements of the form E ⊗ H −m for m ≥ 0 . vector bundle p : E →X is stably trivial. yielding the Bott Periodicity Theorem.2.2 Bott Periodicity In this section we develop a few basic tools that make it possible to compute K(X) for more complicated spaces X . with the Hausdorff property following from the fact that compact Hausdorff spaces are normal. Proof: The inclusion Im q∗ ⊂ Ker i∗ is equivalent to i∗ q∗ = 0 . but now is called simply a cohomology theory. z−m ]) = 1 E− ⊗ (H − 1) + E ⊗ H −m = E ⊗ H −m since E− = 0 . and in fact they make K–theory into what was classically called a generalized cohomology theory.2 51 we see that K(S 2 ) is also generated by 1 and H −1 . Exact Sequences We will be going back and forth between the two versions of K–theory. These tools are formally very similar to the basic machinery of algebraic topology involving cohomology. Choosing a trivialization h : p −1 (A)→A× Cn . 2. it is also compact Hausdorff.Bott Periodicity Section 2.9. Since qi is equal to X/A and K(A/A) = 0 . then the q q∗ inclusion and quotient maps A i → K(A) for which the kernel of i∗ i → X → X/A induce homomorphisms K(X/A) → equals the image of q∗ . K(X) ∗ If X is compact Hausdorff and A ⊂ X is a closed subspace. sition. K(X) and K(X) . Adding a trivial bundle to E . we may assume that E itself is trivial over A . sometimes the other. the polynomial q being 1 so that (3) applies. Proposition 2. projecting to disjoint neighborhoods of x and the point A/A in X/A . Sometimes one is more convenient for stating a result. Thus the proposition states that K(X/A) We will be deriving some longer exact sequences below. This completes the proof of Theorem 2. the composition A→A/A q∗ i → K(X) → K(A) ∗ There is some standard terminology to described the conclusion of the propo- each intermediate group Gi the kernel of the outgoing map equals the image of inis exact.

so E/h is a vector bundle.10. v) for x. A trivialization of E over A determines sections si : A→E which form a basis in each ﬁber over A . each space is obtained from its predecessor by attaching a cone on the subspace two steps back in the sequence. Thus we have a trivialization h of E over a neighborhood U of A . over Uj the extended si ’s can be viewed as a square-matrix-valued function having nonzero determinant at each point of A . We assert that the isomorphism class of E/h does not depend on h . In the absence of such a deformation retraction one can make the following more complicated argument. the sum j ϕj sij gives an extension of si to a section deﬁned on all of X . A − − → − − → SA − − → SX . the quotient space of E under the identiﬁcations h−1 (x. Given two trivializations h0 and h1 . X −− X/A −− −→ q There is an easy way to extend the exact sequence K(X/A)→K(X)→K(A) to the left. Via a local so by the Tietze extension theorem we obtain a section sij : Uj →E extending si . X − A} of X . the quotient map q : X →X/A induces a bijection trivialization h gives a vector bundle E/h→X/A as in the proof of the previous proposition. The vertical maps are the quotient maps obtained by collapsing the most recently attached cone to a point. hence at nearby points as well. If {ϕj . This conclusion holds generally. trivialization. If A is contractible. Choose a cover of A by open sets Uj in X over each of which E is trivial. they must form a basis in all nearby ﬁbers. y ∈ A . it is trivial over some neighborhood of A . v) ∼ h−1 (y. ϕ} is a partition of unity subordinate to the cover {Uj . so the restriction of E over this neighborhood is trivial since it is isomorphic to the pullback of p −1 (A) via the retraction. In many cases the quotient map collapsing a contractible subspace to a point is a homotopy equivalence. This induces a trivialization of E/h over U/A . so this map that E ≈ q∗ (E/h) . It remains only to verify quotient map E →E/h is an isomorphism on ﬁbers. by writing h1 = Proof: A vector bundle E →X must be trivial over A since A is contractible. We claim that since E is trivial over A . In many cases this holds because there is a neighborhood which deformation retracts onto A .52 Chapter 2 Complex K–Theory There is then an induced projection E/h→X/A . Since these sections form a basis in each ﬁber over A . where C and S denote cone and suspension: A X X ∪ CA X/A ( X ∪ CA ) ∪ CX (( X ∪ CA ) ∪ CX ) ∪ C ( X ∪ CA ) In the ﬁrst row. Namely. In the commutative diagram at the right the p E −− E/ h −− −→ − − → − − → and p give an isomorphism E ≈ q∗ (E/h) . in fact: q∗ : Vectn (X/A)→Vectn (X) for all n . using the following diagram. To see that this is a vector bundle we need to ﬁnd a local trivialization over a neighborhood of the point A/A . Lemma 2. hence induces an isomorphism on K . This can be seen as follows. each section si can be regarded as a map from A ∩ Uj to a single ﬁber.

splitting K(X) ⊕ K(Y )→K(X × Y ) . which does not change E/h0 . hence E/h0 ≈ E/h1 . if X is the wedge sum A ∨ B then X/A = B and the sequence breaks up into split short exact sequences. we 0 see that by composing h0 with α in each ﬁber. The space X ∧ Y is called the smash product of X and Y . Then the homotopy from g to the identity gives a homotopy H from h0 to h1 . To deﬁne the reduced product. So we get a splitting K(X × Y ) ≈ K(X ∧ Y ) ⊕ K(X) ⊕ K(Y ) . the ﬁrst map splits via (Sp1 )∗ +(Sp2 )∗ . ∗ ∗ (h1 h−1 )h0 we see that h0 and h1 differ by an element of gx ∈ GLn (C) over each 0 Thus we have a well-deﬁned map Vectn (X)→Vectn (X/A) . In the same way that we constructed E/h we construct a vector bundle (E × I)/H →(X/A)× I restricting to E/h0 over one end and to E/h1 over the other end. E E/h . For reduced suspension we have Σ(X ∨ Y ) = ΣX ∨ ΣY . which implies that the map K(X)→K(A) ⊕ K(B) obtained by restriction to A and B is an isomorphism. X ∨ Y ) : We can use the exact sequence displayed above to obtain a reduced version of ∼ ∼ ∼ ∼ ∼ K ( S (X × Y )) − K ( S (X ∨ Y )) − K (X ∧Y ) − K (X × Y ) − K (X ∨ Y ) − → − → − → − → ∼ ∼ K ( SX ) ⊕ K ( S Y ) ≈ ∼ ∼ K (X ) ⊕ K (Y ) ≈ The second of the two vertical isomorphisms here was noted earlier. a ring homomorphism K(X) ⊗ K(Y )→K(X ∧ Y ) where X ∧ Y = X × Y /X ∨ Y and X ∨ Y = X × {y0 } ∪ {x0 }× Y ⊂ X × Y for chosen basepoints x0 ∈ X and y0 ∈ Y . consider the long exact sequence for the pair (X × Y . inverse to q A From this lemma and the preceding proposition it follows that we have a long exact sequence of K groups ··· →K(SX)→K(SA)→K(X/A)→K(X)→K(A) For example.2 53 point x ∈ A . Writing now h1 = (h1 h−1 α−1 )(αh0 ) . The quotient map .Bott Periodicity Section 2. and for a Deducing Periodicity from the Product Theorem the external product. we may assume that α is the identity. Similarly. The ﬁrst vertical isomorphism arises in similar fashion by using a reduced version of the suspension operator which associates to a space Z with basepoint z0 the quotient space ΣZ SZ →ΣZ induces an isomorphism K(SZ) ≈ K(ΣZ) by the preceding lemma. b) ∗ ∗ p1 (a) + p2 (b) where p1 and p2 are the of SZ obtained by collapsing the segment {z0 }× I to a point. The resulting map g : A→GLn (C) is homotopic to a constant map x α ∈ GLn (C) since A is contractible. (a. This is an bundle E →X/A we have q∗ (E)/h ≈ E for the evident trivialization h of q∗ (E) over since q (E/h) ≈ E as we noted in the preceding proposition. The last horizontal map in the sequence is a split surjection. which gives the ﬁrst isomorphism in the diagram. with projections of X × Y onto X and Y .

and from the short exact sequence above.11. the quotient map S n X →S n ∧ X induces an isomorphism on K by Lemma 2. so p1 (a)p2 (b) restricts to zero in both For a ∈ K(X) = Ker(K(X)→K(x0 )) and b ∈ K(Y ) = Ker(K(Y )→K(y0 )) the K(X) and K(Y ) . is an isomorphism for all compact Hausdorff spaces X .10. and we shall use the same notation a ∗ b for both reduced and unreduced external product. a ∗ b lies in K(X × Y ) . and the second map is the reduced external product. where H is the canonical line bundle over S 2 = CP1 . by the diagram displayed above. as shown in the following diagram. generated by the n fold reduced external product (H − 1) ∗ ··· ∗ (H − 1) . β(a) = (H − 1) ∗ a point. Proof: The homomorphism β : K(X)→K(S 2 X) . which is a quotient of the ordinary n fold suspension S n X obtained by collapsing an n disk in S n X to a Then the reduced external product gives rise to a homomorphism β : K(X)→K(S 2 X). This deﬁnes the reduced external product K(X) ⊗ K(Y )→K(X ∧ Y ) . so the fact that this is an isomorphism is equivalent to the Product Theorem proved in the previous section. In particular. It is essentially a restriction of the unreduced external product. leaving the reader to determine from context which is meant. β(a) = (H − 1) ∗ a .12. ∼ ∼ ∼ ∼ K ( X ) ⊗ K ( Y ) ≈ ( K ( X ) ⊗ K (Y ) ) ⊕ K ( X ) ⊕ K (Y ) ⊕ Z K (X × Y ) so the reduced external product is also a ring homomorphism. Here at last is the Bott Periodicity Theorem: Theorem 2. − − → ≈ ∼ K (X ∧ Y ) → K(S 2 ) ⊗ K(X) → K(S 2 X) (H − 1) ⊗ a is an isomorphism since K(S 2 ) is inﬁnite cyclic − → ∼ ∼ ⊕ K ( X ) ⊕ K (Y ) ⊕ Z == == == Corollary 2. The map β is the composition K(X) where the ﬁrst map a generated by H − 1 .54 Chapter 2 Complex K–Theory ∗ ∗ ∗ external product a ∗ b = p1 (a)p2 (b) ∈ K(X × Y ) has p1 (a) restricting to zero in ∗ ∗ ∗ K(Y ) and p2 (b) restricting to zero in K(X) . . a ∗ b pulls back to a unique element of K(X ∧ Y ) . Since S n ∧ X is the n fold iterated reduced suspension Σn X . K(S 2n+1 ) = 0 and K(S 2n ) ≈ Z . hence in K(X ∨ Y ) .

B)→K ∗ (X. X × B ∪A× Y ) . Then the product K ∗ (X) ⊗ K ∗ (X)→K ∗ (X) is identically zero since it is equivalent to the composition K ∗ (X. For example if X is a suspension we can take A and B to be its two cones. A ∪ B) which is this extends the previously deﬁned ring structure on K 0 (X) . this sequence can be written as in the second row. subspaces of X containing the basepoint.2 55 Extending to a Cohomology Theory As we saw earlier. x) . The lower left corner of the diagram containing the Bott periodicity isomorphisms β commutes since external tensor product with H −1 commutes with maps between spaces. x induced by the relativized diagonal map X/(A ∪ B)→X/A ∧ X/B . the ﬁrst row in the following diagram: ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ K ( S 2X ) − K ( S 2A ) − K ( S ( X/A )) − K ( SX ) − K ( SA ) − K ( X/A ) − K ( X ) − K ( A) → → → → → → → == ∼ ∼ ∼ ∼ ∼ ∼ ∼ ∼ K 2 ( X ) − K 2 ( A ) − K 1( X. Suppose that X = A ∪ B where A and B are compact contractible == ∼ ∼ ∼ K 0( X. A) ⊗ K ∗ (X. coming from the products K(Σi (X/A)) ⊗ K(Σj (Y /B)) K 0 (X) ⊕ K 1 (X) . A) ⊗ K ∗ (X. If we deﬁne K ∗ (X) = identiﬁcation (X × Y )/(X × B ∪ A× Y ) = X/A ∧ Y /B .13. It is reasonable to extend the deﬁnition of K n to positive n by setting K 2i (X) = K(X) and K 2i+1 (X) = K(SX) . A ) −− K 0( X ) −− K 0(A ) −→ − −→ − ∼ ∼ ∼ K 1(A ) →− K 1( X ) →− K 1( X. as in ordinary cohomology. The relative K(X) ⊗ K(Y )→K(X ∧Y ) by replacing X and Y by S i X and S j Y . obtain a multiplication on K ∗ (X) making it into a ring. A ∪ B)→K ∗ (X) Example and K ∗ (X. A) of compact Hausdorff spaces gives rise to an exact sequence of K groups. with a basepoint in their intersection. then this gives a product K ∗ (X) ⊗ K ∗ (Y )→K ∗ (X ∧ Y ) . As a particular case we == A product K i (X) ⊗ K j (Y )→K i+j (X ∧ Y ) is obtained from the external product If we compose the external product K ∗ (X) ⊗ K ∗ (X)→K ∗ (X ∧ X) with the map (x. Negative indices are chosen here so that the ‘coboundary’ maps in this sequence increase dimension. Then the six-term exact sequence can be written − → − − → form of this is a product K ∗ (X. A ) − K 0( X )→ K 0( A) → → → − → → − β ≈ β ≈ ∼ ∼ K 0 ( X ) − K 0( A) → − → If we set K −n (X) = K(S n X) and K −n (X. then we 2. a pair (X. A∪B) = 0 since X = A∪B . A) = K(S n (X/A)) . So the long exact sequence in the second row can be rolled up into a six-term periodic exact sequence. A) ⊗ K ∗ (Y . The general relative form K ∗ (X ∧ X)→K ∗ (X) induced by the diagonal map X →X ∧ X . A ) −− − −− − == == → K(Σi+j (X/A ∧ Y /B)) == == == − − → using the natural . A ) − K 1( X ) →K 1( A) − K 0( X. B)→K ∗ (X × Y . and it is not hard to check that of this product on K ∗ (X) is a product K ∗ (X. B)→K ∗ (X.Bott Periodicity Section 2.

We noted in the proof of Proposition 2. exact sequence of K (X) modules. and the condition that each Ai contain the basepoint can be achieved by adjoining to each ball an arc to a ﬁxed basepoint. by induction on the number of cells. This illustrates the necessity of the condition that A and B both contain the basepoint of X . since without this condition we could take A and B to be the two points of S 0 . then switching S i and S j in S i ∧ S j is a product of ij transpositions of adjacent factors. A ) −− K ( X ) −→ ∗ − . Replacing the product αβ by the product βα amounts to switching the two factors in the ﬁrst term K(S i ∧X) ⊗ K(S j ∧X) . A1 ) ⊗ ··· ⊗ K ∗ (X. observe that the − − → ∗ ∼∗ K (A ) → − − Proposition 2. and this corresponds to switching the S i and S j factors in the third term K(S i ∧ S j ∧ X) . If sending a matrix to its inverse. if X is the union of compact contractible subspaces A1 . The K ∗ (X) module structure on K ∗ (A) is deﬁned by ξ·α = i∗ (ξ)α where i is the inclusion A X and the product on the right side of the equation is multiplication in the ring K ∗ (A) . Thus it sufﬁces to see that we view K(SY ) as Y . In a similar fashion one can see that this observation applies to all ﬁnite cell complexes.14.15. A) . in K ∗ (X) this is only true up to sign: Proposition 2. Proof: αβ = (−1)ij βα for α ∈ K i (X) and β ∈ K j (X) . Transposing the two factors of S 1 ∧ S 1 is equivalent to reﬂection of S 2 across an equator. so all n fold products in K ∗ (X) are trivial. U .2 that the group operation in K(SY ) is the same as the operation induced by the product in U .56 Chapter 2 Complex K–Theory see that the product in K ∗ (S n ) ≈ Z is trivial for n > 0 . For n = 0 the multiplication in K ∗ (S 0 ) ≈ Z is just the usual multiplication of integers since Rm ⊗ Rn ≈ Rmn . An containing the basepoint then the n fold product K ∗ (X. ··· . A1 ∪ ··· ∪ An ) is trivial. An )→K ∗ (X. then switching ends of SY corresponds to the map U →U switching the two ends of a suspension SY induces multiplication by −1 in K(SY ) . This applies to all compact manifolds for example since they are covered by ﬁnitely many closed balls. The exact sequence at the right is an ∼∗ ∼ K ( X. Viewing S i ∧ S j as the smash product of i + j copies of S 1 . In particular all elements of K ∗ (X) are nilpotent since their n th power is zero. so the result follows. To deﬁne the module structure on K ∗ (X. Whereas multiplication in K(X) is commutative. The product is the composition K(S i ∧ X) ⊗ K(S j ∧ X) → K(S i ∧ S j ∧ X ∧ X) → K(S i ∧ S j ∧ X) where the ﬁrst map is external product and the second is induced by the diagonal map on the X factors. More generally. with the maps homomorphisms of K ∗ (X) modules.

It is often convenient to have an unreduced version of the groups K n (X) . and then the six-term long exact sequence is valid also for unreduced groups. s) ∈ CA to (a.Bott Periodicity Section 2. the external product K ∗ (X) ⊗ K ∗ (Y )→K ∗ (X ∧Y ) gives a product K ∗ (X) ⊗ K ∗ (Y )→K ∗ (X × Y ) . A)→K ∗ (X. Commutativity is then obvious. s) ∈ SA to (a. A) with A ≠ ∅ one deﬁnes K n (X. − − → − − → − − → SA → − − X/A → − − X → − − A −− −− −− so it sufﬁces to show this diagram commutes up to homotopy. x) ∈ X ∪ CA and (a. s) ∈ X ∧ SA . For n = 1 this deﬁnition yields K 1 (X) = K 1 (X) since S(X+ ) relation between K and K since K 0 (X) = K 0 (X+ ) = K(X+ ) = Ker(K(X+ )→K(+)) = SX ∨ S 1 and K(SX ∨ S 1 ) ≈ K(SX) ⊕ K(S 1 ) ≈ K(SX) since K(S 1 ) = 0 . a. This is obvious for the middle and right squares. The lower two rows are induced from suspensions of maps between spaces. For a pair (X. For n = 0 this is consistent with the K(X) . and this can easily be done by the simple device of deﬁning K n (X) to be K n (X+ ) where X+ is X with a disjoint basepoint labeled ‘+’ adjoined. The left square can be rewritten − − → SA → − − X ∪ CA −− where the horizontal maps collapse the copy of X in X ∪CA to a point. Since X+ ∧Y+ = (X × Y )+ . diagonal map X →X ∧ X induces a well-deﬁned quotient map X/A→X ∧ X/A . Taking X = Y and composing with the map − − → X ∧ SA →− X ∧ ( X ∪ CA) − − − − → − − → − − → X ∧ SA →− X ∧ X/A →− X ∧ X →− X ∧ A − − − − − − − − → − − → − − → − − → − − → − − → . and Proof: To see that the maps in the exact sequence are module homomorphisms we look at the diagram ∼ ∼ ∼ ∼ K ( S jSA ) − − − K ( S j ( X/A )) − − − K ( S jX ) − − − K ( S jA ) −−→ −−→ −−→ ∼ ∼ ∼ ∼ K ( S iX ∧ S jSA ) −→ K ( S iX ∧ S j ( X/A )) −→ K ( S iX ∧ S jX ) −→ K ( S iX ∧ S jA ) − − − − − − ∼ ∼ ∼ ∼ K ( S i + jSA ) −−− K ( S i + j ( X/A )) − − − K ( S i + jX ) − − − K ( S i + jA ) −−→ −−→ −−→ where the vertical maps between the ﬁrst two rows are external product with a ﬁxed element of K(S i X) and the vertical maps between the second and third rows are induced by diagonal maps. When A = ∅ this remains valid if we interpret X/∅ as X+ . A) . a.2 57 this leads to a product K ∗ (X) ⊗ K ∗ (X. A) = K n (X. What we must show is that the diagram commutes. A) . the left vertical map sends (a. For the upper two rows this follows from naturality of external product since the horizontal maps are induced by maps between spaces. s) ∈ X ∧ CA . and the right vertical map sends x ∈ X to (x.

us- ing the natural identiﬁcation (X × Y )/(X × B ∪ A× Y ) = X/A ∧ Y /B . that for every map f : D n →D n there is a point x ∈ D n with f (x) = x . as in [AT]. For example: — There is no retraction of D n onto its boundary S n−1 . B)→K i+j (X × Y . A) ⊗ K j (X. it would take more work to derive some of the other classical applications of homology such as Brouwer’s theorems on invariance of dimension and invariance of domain. Via the diagonal map With these deﬁnitions the preceding two propositions are valid also for unreduced Elementary Applications With the calculation K ∗ (S n ) ≈ Z completed.58 Chapter 2 Complex K–Theory (x. A ∪ B) . since we have so far only deﬁned K–theory for compact spaces. X × B ∪ A× Y ) de- → K(Σi+j (X/A ∧ Y /B)) . However there are some things homology can do that would be harder using K–theory since K ∗ (S n ) only distinguishes even-dimensional spheres from odd-dimensional spheres. — The Brouwer ﬁxed point theorem. Also. x ﬁned as the external product K(Σi (X/A)) ⊗ K(Σj (Y /B)) There is a relative product K i (X. A = ∅ since we interpret X/∅ as X+ . — The notion of degree for maps f : S n →S n . . we get a product K (X) ⊗ K ∗ (X)→K ∗ (X) which makes K ∗ (X) into a ring. and similarly if Y = ∅ . This works when we obtain also a product K i (X. B)→K i+j (X. Rea- In particular a reﬂection has degree −1 and hence the antipodal map of S n . Consequences of this include the fact that an even-dimensional sphere has no nonvanishing vector ﬁelds. x) .2. it would be possible to derive many of the same applications that follow from the corresponding calculation for ordinary homology or cohomology. since this would mean that the identity map of K ∗ (S n−1 ) factored as K ∗ (S n−1 )→K ∗ (D n )→K ∗ (S n−1 ) . K–groups. ∗ K ∗ (X × X)→K ∗ (X) induced by the diagonal map X →X × X . but the middle group is trivial. or the Jordan curve theorem and its higher-dimensional analogs. induced homomorphism f ∗ : K ∗ (S n )→K ∗ (S n ) is multiplication by d(f ) . A) ⊗ K j (Y . one sees that d is a homomorphism πn (S n )→Z . has degree (−1)n+1 since d(f g) = d(f )d(g) . which is the composition of n + 1 reﬂections. For if not then it is easy to construct a retraction of D n onto S n−1 . namely the integer d(f ) such that the soning as in Proposition 2.

An identity element is not assumed. S 1 . Division Algebras and Parallelizable Spheres With the hard work of proving Bott Periodicity now behind us. i. H-Spaces The ﬁrst step in the proof of the theorem is to reduce it to showing when the sphere S n−1 is an H–space.3 59 2. y) algebra has no zero divisors. an H–space xy/|xy| . and Cayley octonions to the respective unit spheres. After xe and β the map x −1 ex .16. and 8 : is a division algebra. For example. y) (x)β −1 (y) then sends (x. an H–space. with tangent vector ﬁelds v1 . (b) S n−1 is parallelizable. quaternions. Since the maps x ax and x for each a ≠ 0 . e) to α (x)β −1 (e) = α−1 (x)e = x . the equations ax = e and xa = e are solvable.Division Algebras Section 2. vn−1 which are linearly independent at each point of S n−1 . (a) R n The following statements are true only for n = 1 . the goal of this section is to prove Adams’ theorem on the Hopf invariant. S n−1 × S n−1 →S n−1 having a two-sided identity element e ∈ S n−1 . the tangent bundle to S n−1 is trivial. with its famous applications including the nonexistence of division algebras beyond the Cayley octonions: Theorem 2. Since we are dealing with linear maps Rn →Rn .3. By the Gram-Schmidt process we . A division algebra structure on Rn is a multiplication map Rn × Rn →Rn such that the maps x ax and x xa are linear for each a ∈ Rn and invertible if a ≠ 0 . Let α be the map x α −1 to produce an identity in the following way. there exist n − 1 tangent vector ﬁelds to S n−1 which are linearly independent at each point. To say that S n−1 is an H–space means there is a continuous multiplication map Lemma 2. S 3 . or in other words. invertibility is equivalent to having trivial kernel. xa are surjective and similarly it sends (e. but the multiplication can be modiﬁed composing the multiplication with an invertible linear map Rn →Rn taking e2 to e we may assume that e2 = e . or if S n−1 is parallelizable. ··· . and S 7 are H–spaces by restricting the multiplication of complex numbers. If Rn is a division algebra.e. which is well-deﬁned since a division structure on S n−1 is given by (x. 2 . The new product (x. so nonzero elements of the division algebra have multiplicative inverses on the left and right. which translates into the statement that the multiplication has no zero divisors. y) to y . 4 .. This is weaker than being a topological group since associativity and inverses are not assumed. Now suppose that S n−1 is parallelizable. but only S 1 and S 3 are topological groups since the multiplication of octonions is nonassociative.17. then S n−1 is Proof: Having a division algebra structure on Rn with two-sided identity. Choose a unit vector e ∈ Rn .

3. β]/(α2 . y) an H–space structure on S n−1 αx (y) deﬁnes with identity element the vector e1 since αe1 is the identity map and αx (e1 ) = x for all x . vn−1 (x) orthonormal for all x ∈ S n−1 . There remains the much more difﬁcult problem of showing that S n−1 is not an H–space when n is even and different from 2 . We claim that is the identity. proving the claim. The induced homomori → S 2k × S 2k → S 2k µ µ ∗ (γ) = α + β + mαβ for some integer m . For example. where H is the canonical line bundle over S 2 = CP1 . Similarly the coefﬁcient of β must be 1 . Since external product is a ring homomorphism. ··· . α α ∗ (H − 1) . ··· . (1) We have already seen that K(S n ) is Z for n even and 0 for n odd. The ﬁrst step is a simple construction which associates to a map g : S n−1 × S n−1 →S n−1 a map g : S 2n−1 →S n . β2 ) where α and β are the pullbacks of generators of K(S 2k ) and K(S 2 ) under the projections of S 2k × S 2 onto its two factors. en . by changing the sign of vn−1 if necessary to get orientations right. . The composition S 2k phism of K–rings then has the form µ : Z[γ]/(γ )→Z[α. the isomorphism K(S 2k ∧ X) ≈ K(S 2k ) ⊗ K(X) is a ring isomorphism. Let αx ∈ SO(n) send the standard basis to x. we can deduce that K(S 2k × S 2 ) is Z[α. which is impossible since γ 2 = 0 . v1 (x). this leads to a contradiction since it implies that µ ∗ (γ 2 ) = (α + β + mαβ)2 = 2αβ ≠ 0 . ∗ 2 of the ring K(S 2 ) . the external product with the generator H − 1 of K(S 2 ) . vn−1 (x) . We note also that the multiplication in K(S 2k ) is trivial since this ring is the k fold tensor product (2) The external product K(S 2k ) ⊗ K(X)→K(S 2k ∧ X) is an isomorphism since it is (3) The external product K(S 2k ) ⊗ K(X)→K(S 2k × X) is an isomorphism. β2 ) . an iterate of the periodicity isomorphism. ··· . then deforming the vector ﬁelds near e1 . However. β]/(α2 . the vectors v1 (e1 ). v1 (x). since K(S 2k ) can be described as the quotient ring Z[α]/(α2 ) . This follows from (2) by the same reasoning which showed the equivalence of the reduced and unreduced forms of Bott periodicity. so the coefﬁcient of α in µ ∗ (γ) must be 1 . The map i∗ for i the inclusion onto the ﬁrst factor sends α to γ and β to 0 . An additive basis for K(S 2k × S 2 ) is thus {1. vn−1 (e1 ) are the standard basis vectors e2 . ··· . We can apply the last calculation to show that S 2k is not an H–space if k > 0 .60 Chapter 2 Complex K–Theory may make the vectors x. with e the identity element of the H–space structure. where i is the inclusion onto either of the subspaces S 2k × {e} or {e}× S 2k . Then the map (x. and 8 . which has trivial multiplication by Corollary 2. α. Before proceeding further let us list a few easy consequences of Bott periodicity which will be needed. Suppose µ : S 2k × S 2k →S 2k is an H–space multiplication. In particular we see that a generator of K(S 2k ) is the k fold external product (H − 1) ∗ ··· ∗ (H − 1) . We may assume also that at the ﬁrst standard basis vector e1 . This comes from repeated application of the periodicity isomorphism K(X) ≈ K(S 2 X) . αβ} . 4 . β.

and g agrees with g on S n−1 × S n−1 . The value of h mod 2 is called the mod 2 Hopf invariant of f . not on the choice of β .Division Algebras Section 2. For a map f : S 4n−1 →S 2n . y) = |y|g(x. −−−− −−−− −−−−− −−−−−− −−−−−− →−−−−− → − − → − − − − → − − → − − → . sociated map g : S 4n−1 →S 2n has Hopf invariant ±1 . S ) K −−−−−−−−−−− ∼ −−−−−−−−−−→ ∗ ∗ ∗ Φ ⊗Φ Φ ≈ ∼ 2n ∼ 2n 2n 2n 2n 2n 2n 2n K ( D × D . and (β + mα)2 = β2 + 2mαβ since α2 = 0 . In view of the deﬁnition of f it is natural to view the characteristic map Φ of the 4n cell of Cf as a map (D 2n × D 2n . which is an isomorphism since it is an iterate of the Bott periodicity isomorphism. by 2n . ∂(D 2n × D 2n ))→(Cf .1. ∂D × {e } ) ⊗ K ({e } × D . ∼ ∼ K ( Cf ) ⊗ K ( Cf ) −−−−−−−−−−−−−−→ K ( Cf ) −−−−−−−−−−−−−− ∼ −−−−−−−−−−−−−− ≈ ∼ ∼ 2n 2n 2n ( Cf . Note that g is well-deﬁned and continuous. {e } × ∂D ) By the deﬁnition of an H–space and the deﬁnition of f . S 2n ) becomes a short exact sequence 0 Now we specialize to the case that n is even. In the following commutative diagram the horizontal maps are the product maps. D + ) −−−−−−−−−−− K ( Cf . ∂D × D ) ⊗ K ( D × D . since β is unique up to adding an integer multiple of α . D × ∂D ) −− K (D × D . even when |x| or |y| is zero. In fact αβ = 0 so h is well-deﬁned in Z not just Z2 .18. Then n g is deﬁned on ∂D n × D n by g(x. then the as- Proof: Let e ∈ S 2n−1 be the identity element for the H–space multiplication. D ) ⊗ K ( Cf . the exact sequence → K(S 4n ) → K(Cf ) → K(S 2n ) → 0 Let α ∈ K(Cf ) be the image of the generator (H − 1) ∗ ··· ∗ (H − 1) of K(S 4n ) and let β ∈ K(Cf ) map to the generator (H − 1) ∗ ··· ∗ (H − 1) of K(S 2n ) . y/|y|) ∈ D+ and on D n × ∂D n by n g(x.3 61 To deﬁne this. as we will see in §4. ∂ (D × D ) ) −→ ∼ 2n 2n 2n 2n − − ≈ ≈ ∼ 2n ∼ 2n 2n 2n K ( D × {e }. but for our present purposes the mod 2 value of h sufﬁces. let Cf be S 2n with a cell e4n attached by f . and since K 1 (S 4n ) = K 1 (S 2n ) = 0 . y) ∈ D− . If g : S 2n−1 × S 2n−1 →S 2n−1 is an H–space multiplication. The diagonal map is external product. The element β2 maps to 0 in K(S 2n ) since the square of any element of K(S 2n ) is zero. S 2n ) . the map Φ restricts to a 2n 2n homeomorphism from D 2n × {e} onto D+ and from {e}× D 2n onto D− . The mod 2 value of h depends only on f . Lemma 2. The of the pair (Cf . since β restricts to a generator of K(S 2n ) by deﬁnition. equivalent to the external product K(S 2n ) ⊗ K(S 2n )→K(S 4n ) . or in other words. we regard S 2n−1 as ∂(D n × D n ) = ∂D n × D n ∪ D n × ∂D n . y) = |x|g(x/|x|. So by exactness we have β2 = hα for some integer h . and S n we n n take as the union of two disks D+ and D− with their boundaries identiﬁed. we replace n quotient Cf /S 2n is then S 4n . and let f = g . It follows that the element β ⊗ β in the upper left group maps to a generator of the group in the bottom row of the diagram.

··· . Theorem 2. ··· . + Li t) by (ii). In view of this lemma. we ﬁrst use the exterior powers λi (E) to deﬁne a polynomial λt (E) = this implies that λt (E) = i λi (E)t i ∈ K(X)[t] .16 becomes a consequence of the following theorem of Adams: If f : S 4n−1 →S 2n is a map whose mod 2 Hopf invariant is 1 . From the corresponding properties for vector spaces we have: (i) λk (E1 ⊕ E2 ) ≈ 0 i λi (E1 ) ⊗ λk−i (E2 ) .19. but in order to prove Adams’ theorem. more structure is needed. The idea is to use the exterior powers λk (E) . the product map in the top row sends β ⊗ β to ±α since α was deﬁned to be the image of a generator of K(Cf .62 Chapter 2 Complex K–Theory Therefore by commutativity of the diagram. and property (i) says that λt (E1 ⊕ E2 ) = λt (E1 )λt (E2 ) . λk (E)) . The proof of this will occupy the rest of this section. or 4 . and (iv). λk (E)) for an arbitrary vector bundle E . deﬁned for all compact Hausdorff spaces X and all integers k ≥ 0 . This last statement means that ψp (α) − αp = pβ for some β ∈ K(X) . To see what the polynomial sk should be. (4) ψp (α) ≡ αp mod p for p prime. 2 . This is a ﬁnite sum by property which equals i (1 (iv). n = 1 . Thus we have β2 = ±α . (iv) λk (E) = 0 for k greater than the maximum dimension of the ﬁbers of E . the trivial line bundle.20. This will lead us to deﬁne ψk (E) = 1 n sk (λ1 (E). We will show in this case that there is a polynomial sk with integer coefﬁcients such that Lk + ··· + Lk = sk (λ1 (E). (iii) λ1 (E) = E . Adams Operations The Hopf invariant is deﬁned in terms of the ring structure in K–theory. and satisfying: (1) ψk f ∗ = f ∗ ψk for all maps f : X →Y . S 2n ) . There exist ring homomorphisms ψk : K(X)→K(X) . . namely certain ring homomorphisms ψk : K(X)→K(X) . We would like 1 n a general deﬁnition of ψk (E) which specializes to this formula when E is a sum of line bundles. Recall that we want ψk (E) to be Lk + ··· + Lk when E = L1 ⊕ ··· ⊕ Ln for line bundles 1 n Li . When E = L1 ⊕ ··· ⊕ Ln i λt (Li ) . (iii). which says that the Hopf invariant of f is ±1 . (Naturality) (2) ψk (L) = Lk if L is a line bundle. (ii) λ (E) = 1 . (3) ψk ◦ ψ = ψk . properties (2) and (3) give the formula ψk (L1 ⊕ ··· ⊕ Ln ) = Lk + ··· + Lk . then Theorem 2. Here are their basic properties: Theorem 2. In the special case of a vector bundle E which is a sum of line bundles Li .

··· . tn can be expressed as a unique polynok k mial in σ1 .3 63 + Li t) is the j th elementary symmetric function σj of the Li ’s. The recursion relation easily yields for example s1 = σ1 2 s2 = σ1 − 2σ2 3 s3 = σ1 − 3σ1 σ2 + 3σ3 4 2 2 s4 = σ1 − 4σ1 σ2 + 4σ1 σ3 + 2σ2 − 4σ4 Summarizing. t1 + ··· + tn is a polynomial sk (σ1 . λk (E)) . λk (E)) = sk (σ1 (L1 . then in the case that E is a sum of line bundles L1 ⊕ ··· ⊕ Ln we have ψk (E) = sk (λ1 (E). σk ) . and then if we substitute x = −ti in the identity k k−1 (x + t1 ) ··· (x + tk ) = x k + σ1 x k−1 + ··· + σk we get ti = σ1 ti − ··· + (−1)k−1 σk . ··· . ··· .20: Property (1) for vector bundles. The fundamental theorem on symmetric polynomials. ··· . the sum of all products of j distinct Li ’s. let us introduce the variable ti for Li . ··· .20 and Adams’ theorem. ··· . 134] or [van der Waerden. Proof follows immediately from the relation f ∗ (λi (E)) = λi (f ∗ (E)) . ··· . ··· . proved for example in [Lang. called a Newton polynomial. A recursive formula for sk is sk = σ1 sk−1 − σ2 sk−2 + ··· + (−1)k−2 σk−1 s1 + (−1)k−1 kσk To derive this we may take n = k . Thus (1 + t1 ) ··· (1 + tn ) = 1 + σ1 + ··· + σn . asserts that every degree k symmetric polynomial in t1 . This polynomial sk is independent of n since we can pass from n to n − 1 by setting tn = 0 . if we deﬁne ψk (E) = sk (λ1 (E). Ln )) = Lk 1 + ··· + Lk n by (∗) Verifying that the deﬁnition ψk (E) = sk (λ1 (E). of Theorem 2. Thus we have (∗) λj (E) = σj (L1 . Ln ). In particular. there is a compact Hausdorff space F (E) and a map p : F (E)→X such that the in- The Splitting Principle. Given a vector bundle E →X with X compact Hausdorff. Summing over i then gives the recursion relation. where σj is the j th elementary symmetric polynomial in the ti ’s. §26]. p. f ∗ (ψk (E)) = ψk (f ∗ (E)) . σk (L1 .Division Algebras The coefﬁcient λj (E) of t j in λt (E) = i (1 Section 2. Ln ) if E = L1 ⊕ ··· ⊕ Ln To make the discussion completely algebraic. but for the moment let us assume it and proceed with the proof of Theorem 2. This will be proved later in this section. ··· . ··· . Additivity of ψk for . λk (E)) gives operations on K(X) satisfying the properties listed in the theorem will be rather easy if we make use of the following general result: duced map p ∗ : K ∗ (X)→K ∗ (F (E)) is injective and p ∗ (E) splits as a sum of line bundles. σk .

one sees this was deﬁned as ∗ ∗ ψk (α ∗ β) = ψk (p1 (α)p2 (β)) ∗ ∗ = ψk (p1 (α))ψk (p2 (β)) ∗ ∗ = p1 (ψk (α))p2 (ψk (β)) = ψk (α) ∗ ψk (β). Multiplicativity is also easy from the splitting principle: If E is the sum of line bundles Li and E is the sum of line bundles Lj . In this case ψk must be multiplication by some integer since it is an additive homomorphism of Z . For property (3) the splitting principle and additivity reduce us to the case of line bundles. For the external since if one looks back at the deﬁnition of α ∗ β . ψk (E1 ⊕ E2 ) = ψk (E1 ) + ψk (E2 ) . Proposition 2. and argue by induction. n By the naturality property (1). and it follows formally that it is multiplicative on elements of K(X) .64 Chapter 2 Complex K–Theory vector bundles. Assuming the desired formula holds in K(S 2n−2 ) . Likewise for (4). Thus k product K(X) ⊗ K(Y )→K(X ∧ Y ) . we have ψk (α ∗ β) = ψk (α) ∗ ψk (β) = kα ∗ kn−1 β = kn (α ∗ β) . and we are done. Since ψk is additive on 1 n vector bundles. Then ψk (α) = ψk (H − 1) = H k − 1 k by property (2) since αi = (H − 1)i = 0 for i ≥ 2 = (1 + α) − 1 = 1 + kα − 1 = kα When n > 1 we use the external product K(S 2 ) ⊗ K(S 2n−2 )→K(S 2n ) . it will sufﬁce to show ψ (α) = kα for α a generator of K(S 2 ) . Proof: k ψk : K(S 2n )→K(S 2n ) is multiplication by kn . ψk restricts to an operation ψk : K(X)→K(X) since p i. then E ⊗ E is the sum of the line bundles Li ⊗ Lj . Since ψk is additive.j k ψk (Li ⊗ Lj ) = i. then ψp (E) = L1 + ··· + Lp ≡ (L1 + ··· + Ln )p = E p mod p . and the formula ψk (L1 ⊕ ··· ⊕ Ln ) = Lk + ··· + Lk preserves sums. it induces an additive operation on K(X) deﬁned by ψk (E1 − E2 ) = ψk (E1 ) − ψk (E2 ) . so ψk (E ⊗ E ) = ψ is multiplicative for vector bundles. follows from the splitting principle since we can ﬁrst pull back to split E1 then do a further pullback to split E2 . hence K(X) is the kernel of the homomorphism K(X)→K(x0 ) for x0 ∈ X .j Lk ⊗ Lj = i k i Lk i j Lj = ψk (E)ψk (E ) . Consider ﬁrst the case n = 1 . if E = L1 + ··· + Ln . For this extended ψk the properties (1) and (2) are clear. we have the formula ψk (α ∗ β) = ψk (α) ∗ ψk (β) ∗ ∗ p1 (α)p2 (β) . where ψk (ψ (L)) = Lk = ψk (L) . We can take α = H − 1 for H the canonical line bundle over S 2 = CP1 . which is an isomorphism.21.j (Li ⊗ Lj ) k = i. This will allow us to compute ψk on K(S 2n ) ≈ Z . .

By Proposition 2. We will show that the highest power of 2 dividing +2 Proof: Write n = 2 = 0 and 2 for > 0 . 4 can then be checked individually. Since β2 = hα with h the Hopf invariant of f . write 32n − 1 = (3n − 1)(3n + 1) . together with a calculation of the ring structure of K ∗ (CPn ) . or (k2n − kn )µ = ( − n )µk By property (6) of ψ2 . 2. Therefore ψk ψ (β) = ψk ( n The deﬁnition of the Hopf invariant of a map f : S 4n−1 →S 2n β + µ α) = kn n β + (k2n µ + n n µk )α Since ψk ψ = ψk = ψ ψk . The Splitting Principle The splitting principle will be a fairly easy consequence of a general result about the K–theory of ﬁber bundles called the Leray-Hirsch theorem. 2 . or 2n (2n − 1)µ3 = 3n (3n − 1)µ2 . the formula ψ2 (β) = 2n β + µ2 α implies that µ2 ≡ h mod 2 . or 4 . The proof is completed by the following elementary number theory fact.3 65 Now we can use the operations ψ2 and ψ3 and the relation ψ2 ψ3 = ψ6 = ψ3 ψ2 to prove Adams’ theorem. The highest power of 2 dividing the ﬁrst factor is 2 as we just showed. Then 3n + 1 ≡ 2 mod 4 since n is even. So the highest power of 2 dividing the product (3m − 1)(3m + 1) is 8 . By the preceding displayed formula we have (22n −2n )µ3 = (32n − 3n )µ2 . 3 − 1 is 2 for n If 2n divides 3n − 1 then n = 1. the coefﬁcient k2n µ + and are switched. Proof of Theorem 2.Division Algebras Section 2. We ﬁnd the highest power of 2 dividing 3n − 1 by induction on we have 3 − 1 = 3 case n m − 1 ≡ 2 mod 4 since 3 ≡ −1 mod 4 and m is odd. ψk (β) = kn β + µk α for some µk ∈ Z .21. which implies . 3. Since 3n and µ2 are odd. This implies the lemma since if 2n divides + 2 . ψk (α) = k2n α since α is the image of a generator of K(S 4n ) . The following proposition will allow us to compute at least the additive structure of K ∗ (CPn ) . For ≤2 = 0 3n − 1 . so the highest power of 2 dividing 3n + 1 is 2 . This gives the relation k2n µ + n 2n µk of α is unchanged when k 2n µk = µk + kn µ . Lemma 2. The cases n = 1. For the inductive step of passing from to + 1 with ≥ 1 . and the highest power of 2 dividing the second factor is 4 since 3m + 1 ≡ 4 mod 8 because 32 ≡ 1 mod 8 and m is odd. m with m odd. then by this fact. Similarly. we have ψ2 (β) ≡ β2 mod 2 . β ∈ K(Cf ) .22. 2n must then divide 3n − 1 .19: involved elements α. . or in other words from n to 2n with n even. Thus the highest power of 2 dividing 32n − 1 is twice the highest power of 2 dividing 3n − 1 . In the next = 1 we have 3n − 1 = 32m − 1 = (3m − 1)(3m + 1) . hence 2 ≤ 2 m = n ≤ and n ≤ 4 . so 2n divides 3n (3n − 1)µ2 . so µ2 is odd if we assume h = ±1 . n ≤ + 2 .

for some k . This is shown in Proposition 2. 2n . L. ··· . generated group with at most n generators. Thus by the change of variable x = L−1 we see that K(CPn ) is the truncated polynomial ring Z[x]/(x n+1 ) . The ﬁrst term of the exact sequence K 1 (X/A)→K 1 (X)→K 1 (A) is zero if all cells of X are of even dimension. Then there is a short exact sequence 0→K 0 (X/A)→K 0 (X)→K 0 (A)→0 Z ⊕ K 0 (A) and the ﬁnal statement of the proposition follows. so K 1 (CPn ) = 0 and K 0 (CPn ) ≈ Zn+1 . A) we have an exact sequence K ∗ (X/A) ∗ → K ∗ (X) → K ∗ (A) . Proof: We show this by induction on the number of cells. with no conditions on the dimensions of these cells. so cells are not required to attach only to cells of lower dimension. and exactness implies that K (X) requires at most one more generator than K ∗ (A) . It implies that the splitting principle can be applied staying within the realm of ﬁnite cell complexes. This proposition applies in particular to CPn . we have K (X/A) ≈ Z . 2. For the pair (X. By induction K(A) is free. ··· . ··· . so this sequence splits. it is easy to see that a space is a ﬁnite cell complex if it is a ﬁber bundle over a ﬁnite cell complex with ﬁbers that are also ﬁnite cell complexes. Finite cell complexes are always homotopy equivalent to ﬁnite CW complexes (by deforming each successive attaching map to be cellular) so the only advantages of ﬁnite cell complexes are technical. then K ∗ (X) is a ﬁnitely Proposition 2. It follows that 1. In particular. so induction on the number of cells implies that K 1 (X) = 0 . ∗ Since X/A = S k . . Proof: The exact sequence for the pair (CPn . x. CPn−1 ) → K(CPn ) → K(CPn−1 ) → 0 ρ We shall prove: (an ) (L − 1)n generates the kernel of the restriction map ρ . with additive basis 1. K(CPn ) is the quotient ring Z[L]/(L−1)n+1 where L is the canon- ical line bundle over CPn . 4. CPn−1 ) gives a short exact sequence 0 → K(CPn . The phrase ‘ﬁnite cell complex’ would normally mean ‘ﬁnite CW complex’ but we can take it to be something slightly more general: a space built from a ﬁnite discrete set by attaching a ﬁnite number of cells in succession. The complex X is obtained from a subcomplex A by attaching a k cell. Ln is also an additive basis. which has a cell structure with one cell in each dimension 0.23.28 in a brief appendix to this section. The ring structure is as simple as one could hope for: with K 0 (X/A) ≈ Z .24. x n . hence K 0 (X) ≈ Proposition 2. If all the cells of X have even dimension then K 1 (X) = 0 and K 0 (X) is free abelian with one basis element for each cell.66 Chapter 2 Complex K–Theory If X is a ﬁnite cell complex with n cells.

Thus we have a decomposi- tion CPn = for i ≠ j . . . CPn−1 ) = K(S 2n ) by the short exact sequence. ··· . . (L − 1)n } is an additive basis for K(CPn ) .. . ∂i C0 ) ⊂ (CPn . This means that (L − 1)n −−→ −−− −−− −−− −−− −−− n K(C P . C 1 ∪ . Since (L − 1)n+1 = 0 in K(CPn ) by (an+1 ) . it follows that K(CP n ) is the quotient ring Z[L]/(L − 1)n+1 . Instead of S 2n+1 we could equally well take the boundary 2 2 2 of the product D0 × ··· × Dn where Di is the unit disk in the i th coordinate of Cn+1 . ∂D 1 ) ⊗ 2 2 − → K ( C 0 .. completing the induction. By commutativity of the diagram. Cn ) −− K ( C P n. The orbit space 2 2 2 of D0 × ··· × ∂Di × ··· × Dn under the action is a subspace Ci ⊂ CPn homeomorphic 2 2 2 to the product D0 × ··· × Dn with the factor Di deleted. the product x1 ··· xn then generates K(CPn . ∂Di ) . The CPn−1 at the right side of the diagram sits in CPn in the last n coordinates of Cn+1 . ∂Dn ) . i Ci with each Ci homeomorphic to D 2n and with Ci ∩ Cj = ∂Ci ∩ ∂Cj 2 2 Consider now C0 = D1 × ··· × Dn . so is disjoint from C0 . The element xi ∈ K(CPn . ⊗ K (Dn .Division Algebras Section 2. and we start the count with i = 0 for convenience. A reason one might expect (an ) to be true is that the kernel of ρ can be identiﬁed with K(CPn . then (an ) and the preceding exact sequence imply that {1. ∂C0 ) ≈ ≈ ⊗ K ( C P n. . The inclusions (Di . L − 1. and Bott periodicity implies that the n fold reduced external product of the generator L − 1 of K(S 2 ) with itself generates K(S 2n ) . − → −−−−− ≈ −−−−− −−−−− −−−−− −−−−− −−−−− ⊗ . ⊗ K ( C P n ) −−−−−− K ( C P n ) −−−−−− −−−−−→ where the maps from the ﬁrst column to the second are the n fold products. C P n 1 ) −− −→ −− −→ ≈ 0 − − → − − → . Its boundary is decomposed into the pieces 2 2 2 2 2 ∂i C0 = D1 × ··· × ∂Di × ··· × Dn . . . C1 ) ⊗ . to relate the external product K(S 2 ) ⊗ ··· ⊗ K(S 2 )→K(S 2n ) to the ‘internal’ product The space CPn is the quotient of the unit sphere S 2n+1 in Cn+1 under multiplication by scalars in S 1 ⊂ C . Then we have 2 2 ∂(D0 × ··· × Dn ) = 2 i (D0 × 2 2 ··· × ∂Di × ··· × Dn ) The action of S 1 by scalar multiplication respects this decomposition. C1 ∪ ··· ∪ Cn ) . . To make this rough argument into a proof we will have K(CPn ) ⊗ ··· ⊗ K(CPn )→K(CPn ) . ∂1C 0 ) n K(C P ) ⊗ . ⊗ K ( C .3 67 Hence if we assume inductively that K(CPn−1 ) = Z[L]/(L − 1)n . Ci ) give rise to a commutative diagram K ( D 1 . hence the quotient map CPn /CPn−1 →CPn /(C1 ∪ ··· ∪ Cn ) is a homotopy equivalence. The upper map in the middle column is an isomorphism because the inclusion C0 CPn induces a homeomorphism C0 /∂C0 ≈ CPn /(C1 ∪···∪Cn ) . ∪ Cn ) −− K ( C P n. ∂ C ) − −− K ( C − −− − −− − → → − → ≈ 0 n 0 2 2 . Ci ) mapping downward to L−1 ∈ K(CPn ) maps upward 2 2 to a generator of K(C0 . ∂i C0 ) ≈ K(Di . ∂Di ) ⊂ (C0 .

i bi ⊗ i∗ (ci )) = i p ∗ (bi )ci for i the inclusion F pullbacks under the projection E →F of a basis for K ∗ (F ) . Example 2. But as we have just seen. is an isomorphism. in particular (L − 1)n+1 = 0 . which will be the major theoretical ingredient in the proof of the splitting principle: K ∗ (E) that restrict to a basis for K ∗ (F ) in each ﬁber F . ··· . For most of our applications of the theorem either case (a) or case (b) will sufﬁce. which equals the kernel of ρ . For a subspace B ⊂ B let E = p −1 (B ) . as a module over K ∗ (B) . ∗ ∗ where the left-hand Φ is deﬁned by the same formula Φ( i bi ⊗ i∗ (ci )) = p ∗ (bi )ci . To see that the diagram (∗) commutes. If either (a) B is a ﬁnite cell complex. ··· . proving (an ) .68 Chapter 2 Complex K–Theory generates the image of the map K(CPn . we can interpolate between its two rows the row → K ∗ (E. or (b) F is a ﬁnite cell complex having all cells of even dimension. The upper squares of the enlarged diagram then commute trivially.13 shows that all products of n + 1 elements of K(CPn ) must be zero. B ) ⊗ K ∗( F ) −− K ∗( B ) ⊗ K ∗( F ) −− K ∗( B ) ⊗ K ∗( F ) − −→ − − −→ − − −→ − − − − − − → − − → − − → − − → Φ Φ Φ −−−− K −−−− −−−→ − ∗ ∗ ( E . Let p : E →B be a ﬁber bundle with E and B compact Hausdorff and then K ∗ (E) . E ) ⊗ K ∗ (F ) → K ∗ (E) ⊗ K ∗ (F ) → K ∗ (E ) ⊗ K ∗ (F ) → by factoring Φ as the composition i bi ⊗ i∗ (ci ) i p ∗ (bi ) ⊗ i∗ (ci ) i p ∗ (bi )ci . (L − 1)n is nonzero. with ﬁber F such that K ∗ (F ) is free. and we include (b) mainly to obtain the splitting principle for vector bundles over arbitrary compact Hausdorff base spaces. Then we have a diagram In the case of the product bundle E = F × B the classes ci can be chosen to be the Proof: (∗) −− K ∗( B. Here the K ∗ (B) module structure on K ∗ (E) is deﬁned by β · γ = p ∗ (β)γ for β ∈ K ∗ (B) and γ ∈ K ∗ (E) . The proof of (a) is somewhat simpler than (b). E ) .25. Φ( E . E ) −−−−−→ K ( E ) − − − − − K ( E ) −−−−− −−−−− −−−−→ −−−− −−−− ∗ ∗ −−− − −− − −→ − − − i but with p (bi )ci referring now to the relative product K (E. E )× K (E)→K ∗ (E.15. ck ∈ Theorem 2. The right-hand Φ is deﬁned using the restrictions of the ci ’s to the subspace E . ck } . Suppose that there exist classes c1 . The lower row of the diagram is of course . CPn−1 )→K(CPn ) .13 is best possible in terms of the degree of nilpotency. Since CPn is the union of the n+1 balls Ci . Another way to state the conclusion of the theorem is to say that the map Φ : K ∗ (B) ⊗ K ∗ (F )→K ∗ (E) . is free with basis {c1 . Now we come to the Leray-Hirsch theorem for K–theory. The theorem then asserts that the external product K ∗ (F ) ⊗ K ∗ (B)→K ∗ (F × B) is an isomorphism. so the result in Example 2. and the lower squares commute by Proposition 2.

hence induces homeomorphisms B/B ≈ D n /S n−1 and E/E ≈ ϕ∗ (E)/ϕ∗ (E ) . To show the left-hand Φ is an isomorphism. ϕ ( E ) ) ≈ K ( D n × F . Since mutative diagram Φ Φ ∗ ∗ ∗ K ( E . D n is contractible. B ) = (D n . and let E = p −1 (B ) as above. Consider the diagram (∗) with (B. so we are free to interchange the roles of F and B . S n−1 )→(B. then within a given dimension. The upper row of (∗) is then an exact sequence since it is tensoring with the ﬁxed group K ∗ (F ) . and tensoring an exact sequence with a free abelian group preserves exactness. the result of the tensoring operation being simply to replace the given exact sequence by the direct sum of a number of copies of itself. note ﬁrst that B/B = S n so we may obtained from the split short exact sequence 0→K ∗ (B. The induction starts with the trivial case that B is zero-dimensional. involving the product bundle D n × F →D n . is an isomorphism. and so we have a com∗ ∗ ∗ ≈ K ( B . In this case Ψ is just the external product. If the left-hand Φ is also an isomorphism. B )→K ∗ (B)→K ∗ (B )→0 by −→ − −→ − .3 69 exact. The proof in case (a) will be by a double induction. ﬁrst on the dimension of B . obtained by attaching a cell en to a subcomplex B . then the ﬁve-lemma will imply that the middle Φ is an isomorphism. The upper row is also exact since we assume K ∗ (F ) is free. on the number of cells. If we can show that the left-hand Φ in (∗) is an isomorphism. B ) be a characteristic map for the attached n cell. Thus we may use the diagram (∗) with F an arbitrary compact Hausdorff space and B a ﬁnite cell complex having all cells of even dimension. S n−1 ) . This ﬁnishes the proof in case (a). In case (b) let us ﬁrst prove the result for a product bundle E = F × B . B ) ⊗ K ( F ) −→ K ( D n. so the ﬁve-lemma will imply that the middle Φ is also an isomorphism. then by induction on the number of cells of B we may assume the right-hand Φ is an isomorphism. S n 1 × F ) −−− −−− ∗ ∗ −≈ − − −− Φ −− −− −− −− −− −→ The two horizontal maps are isomorphisms since ϕ restricts to a homeomorphism on the interior of D n . suppose B is obtained from a subcomplex B by attaching a cell en . We may assume the right-hand Φ in (∗) is an isomorphism since S n−1 has smaller dimension than the original cell complex B . B ) replaced by (D n . ﬁnishing the induction step. the pullback bundle ϕ∗ (E) is a product. By induction on the number of cells of B we may assume the right-hand Φ in (∗) is an isomorphism. For the induction step. S n 1 ) ⊗ K ( F ) − ∗ − − − Let ϕ : (D n . E ) −−−→ K (ϕ ( E ) . Therefore by the ﬁve-lemma the left-hand Φ in (∗) is an isomorphism for (B. Thus the diagram reduces the proof to showing that the right-hand Φ . The middle Φ is an isomorphism by the case of zerodimensional B since D n deformation retracts to a point.Division Algebras Section 2. S n−1 ) . hence a ﬁnite discrete set.

Now repeat the process by forming P (E ) . a ﬁnite number of these neighborhoods cover B . that is. hence splits as L ⊕ E for E basis elements implies that p ∗ : K ∗ (X)→K ∗ (P (E)) is injective. V2 ) . S n−1 ) . ··· . V2 ) if it is an isomorphism for (V1 . but this time we need a more subtle inductive statement since B is just a compact Hausdorff space. then so is U1 ∪ U2 . so Φ will be an isomorphism for V if it is an isomorphism for (V . Consider Then we have an associated projective bundle p : P (E)→X with ﬁbers CPn−1 . Note that P (E ) is the space of pairs the ﬂag bundle F (E)→X described at the end of §1. the maps Φ are isomorphisms for V1 and V1 ∩ V2 . since K 1 of an odd-dimensional sphere is K 0 of an even-dimensional sphere. In each ﬁber CPn−1 of P (E) the classes 1.24. A compact V ⊂ U1 ∪ U2 is the union of V1 = V ∩ U1 and V2 = V ∩ U2 . Ln−1 . V2 ) . splitting off another line bundle from the pullback of E over P (E ) . V1 ∩ V2 ) . Hence Φ is an isomorphism for (V1 . Then the middle Φ in (∗) is obviously an isomorphism. Consider the following condition on a compact subspace U ⊂ B : For all compact V ⊂ U the map Φ : K ∗ (V ) ⊗ K ∗ (F )→K ∗ (p −1 (V )) is an isomorphism. not a cell complex. each point of B has a compact neighborhood U that is good. When the sphere S n−1 is even-dimensional we have already shown that Φ is an isomorphism in the remarks following the proof of Lemma 2. V1 ∩ V2 ) . By the special case already proved. the fact that 1 is among the p ∗ (E)→P (E) contains the line bundle L as a subbundle. Since B is compact. the upper row of this diagram is exact.17. If this is satisﬁed. ··· .70 Chapter 2 Complex K–Theory as well take the pair (B. restrict to a basis for K ∗ (CPn−1 ) by Proposition 2. L. From the Leray-Hirsch theorem Proof of the Splitting Principle: In the preceding example.1. the diagram like (∗) for the pair (V . B ) to be (D n . P (E) is the space of lines in E . Let E →X be a vector bundle with ﬁbers Cn and compact base X . After a ﬁnite number of repetitions we obtain . Ln−1 in K ∗ (P (E)) we deduce that K ∗ (P (E)) is a free K ∗ (X) module with basis 1. so the left-hand Φ will be an isomorphism iff the right-hand Φ is an isomorphism. let us call U good. where Example 2. L. Now we turn to case (b) for nonproducts. Over P (E) there is the canonical line bundle L→P (E) consisting of the vectors in the lines of P (E) . and the induction step is ﬁnished. and the same argument applies also when the sphere is odd-dimensional. Assuming U2 is good.26. Now look at the diagram like (∗) for (V1 . so by induction it will be enough to show that if U1 and U2 are good. the map Φ is an isomorphism for V2 . The pullback bundle →P (E) the subbundle of p ∗ (E) orthogonal to L with respect to some choice of inner product. one-dimensional linear subspaces of ﬁbers of E . The proof will once again be inductive. The quotient V /V2 is homeomorphic to V1 /(V1 ∩ V2 ) so Φ will be an isomorphism for (V . Assuming U1 is good. V1 ∩ V2 ) . whose points are n tuples of of orthogonal lines in ﬁbers of E . Since K ∗ (F ) is free.

as desired. we do the construction for each component separately. Thus Xi+1 is the quotient space of the disjoint union of Xi and a disk D ni under the identiﬁcations x ∼ ϕi (x) for x ∈ ∂D ni = S ni −1 . (If the ﬁbers of E have different dimensions over different components of X . where to simplify notation we let ‘ E ’ also denote the pullback of E over P (E) . i (−1) The relation λn (E ) = 0 λ (E)L n−i = 0 .28.Division Algebras Section 2.) The pullback of E over F (E) splits as a sum of line bundles. By induc- If Φ : D n →B is a characteristic map for en then the pullback bundle Φ∗ (E)→D n is tion on the number of cells of B we may assume that p −1 (B ) is a ﬁnite cell complex. L. X0 ⊂ X1 ⊂ ··· ⊂ Xk = X where X0 is a ﬁnite discrete set and Xi+1 is obtained from Proposition 2. with basis 1. this means that . The pullback of E over P (E) splits as L ⊕ E for some bundle E of dimension n − 1 . becomes (L−1) and we see that the proposition Appendix: Finite Cell Complexes As we mentioned in the remarks following Proposition 2. Ln−1 .3 71 orthogonal lines in ﬁbers of E . and the desired relation will be λn (E ) = 0 . where n is the dimension of E .23 it is convenient for purposes of the splitting principle to work with spaces slightly more general than ﬁnite CW complexes. If p : E →B is a ﬁber bundle whose ﬁber F and base B are both ﬁnite cell complexes. The equation λt (E) = λt (L)λt (E ) can be rewritten as λt (E ) = λt (E)λt (L)−1 where λt (L)−1 = λt (E ) = λt (E)λt (L)−1 .26 we saw that K ∗ (P (E)) is free as a K ∗ (X) module. with the coefﬁcient of Ln equal to 1 . Equating coefﬁcients of t n in the two sides of n−i i λ (E)Ln−i . Proof: Suppose B is obtained from a subcomplex B by attaching a cell en . For an n dimensional vector bundle E →X the ring K(P (E)) is i i n−i i (−1) λ (E)L . To compute λn (E ) = 0 we use the formula λt (E) = λt (L)λt (E ) in K ∗ (P (E))[t] . The result can be stated in the following form: isomorphic to the quotient ring K ∗ (X)[L]/ Proposition 2. then E is also a ﬁnite cell complex. Since F is a ﬁnite cell complex. i (−1) λ (E)L i i n−i n . In order to describe the multiplication in K ∗ (P (E)) one therefore needs only a relation expressing Ln in terms of lower powers of L . By a ﬁnite cell complex we mean a space which has a ﬁnite ﬁltration Xi by attaching a cell eni via a map ϕi : S ni −1 →Xi . Such a relation can be found as follows. a product since D n is contractible. whose cells are products of cells in B with cells in F .27. we get λn (E ) = can be written as i (−1) i i i i i i (−1) L t since λt (L) = 1 + Lt . In the preceding Example 2. n i For example when X is a point we have P (E) = CPn−1 and λi (E) = Ck for k = so the polynomial generalizes the isomorphism K ∗ (CPn−1 ) ≈ Z[L]/(L − 1)n ) . and the map F (E)→X induces an injection on K ∗ since it is a composition of maps with this property. ··· .

Hence we may obtain E from p −1 (B ) by attaching cells.72 Chapter 2 Complex K–Theory we may obtain Φ∗ (E) from its restriction over S n−1 by attaching cells. .

The converse statement is also true because the classes wi (En ) gen- coefﬁcients. Thus if f induces the trivial map on Z2 cohomology. sifying map f : B →Gn being nullhomotopic. Pontryagin classes can be regarded as a reﬁnement of Stiefel-Whitney classes when one takes Z rather than Z2 coefﬁcients. as the Stiefel-Whitney classes erate the cohomology ring H ∗ (Gn . The Euler class e(E) ∈ H n (B. ··· . wn (En )] . and the Pontryagin classes pi (En ) are needed in order to describe it completely. since H ∗ (Gn . There are four main kinds: 2. 1. Z2 ) . The naturality property that wi (f ∗ (E)) = f ∗ (wi (E)) . In fact H ∗ (Gn . Even though this description is not so nice. it is nevertheless true that f induces the trivial map on co- . Z) for a complex vector bundle E →B . Z) for a real vector bundle E →B .Characteristic classes are cohomology classes associated to vector bundles. The Stiefel-Whitney and Chern classes are formally quite similar. For an n dimensional vector bundle E →B to be trivial is equivalent to its clas- 3. Stiefel-Whitney classes wi (E) ∈ H i (B. or nontrivial. like the other characteristic classes. so in particular if f is a classifying Stiefel-Whitney classes wi . Much more accessible is the weaker question of whether f induces a nontrivial map on homology or cohomology. Z2 ) is exactly the polynomial ring Z2 [w1 (En ). They measure in some way how a vector bundle is twisted. Z2 ) for a real vector bundle E →B . satisfy the important map B →Gn for E we have E = f ∗ (En ) and wi (E) = wi (f ∗ (En )) = f ∗ (wi (En )) . This is one reason for considering Z2 rather than Z of E are trivial. Z) has a more complicated ring structure. but as with most things in homotopy theory it can be quite difﬁcult to determine whether this is the case or not. Chern classes ci (E) ∈ H 2i (B. 4. Pontryagin classes pi (E) ∈ H 4i (B. and this is precisely what characteristic classes measure. and the Euler class is a further reﬁnement in the orientable case. Z) for an oriented n dimensional real vector bundle E →B .

but only S 1 . and . meaning that taking the direct sum of a given bundle with a trivial bundle does not change the characteristic classes. Since Z2 is abelian. except for the Euler class. and the corresponding statement here is that this map induces the trivial map on cohomology with arbitrary coefﬁcients iff the Euler class of E as well as all its StiefelWhitney and Pontryagin classes are trivial. This is exactly the ﬁrst Stiefel-Whitney class w1 (E) . such as the tangent bundle of a sphere. What Exactly Do Characteristic Classes Measure? trivial. The ﬁrst obstacle to triviality is orientability. cn ] on the Chern classes ci of the universal bundle over Gn (C) . and homomorphisms H1 (B)→Z2 are identiﬁable with elements of H 1 (B. ··· . orientability is detected by the homomorphism π1 (B)→Z2 that assigns 0 or 1 to each loop according to whether orientations of ﬁbers are preserved or reversed as one goes around the loop. For complex vector bundles the analogous statements are also true but somewhat simpler since one only needs Chern classes. Z2 ) . Thus we have an element of H 1 (B. If E is orientable we can take its classifying map to be a map to Gn . A key feature of the Euler class is that it is zero for vector bundles that split as the sum of a trivial bundle with some other bundle. Having an understanding of when E is trivial over the 1 skeleton we can ask about triviality over each successive higher skeleton. and S 7 have trivial tangent bundles. Thus for a vector bundle which is stably trivial. then if the ﬁbers have dimension n we can choose n orthonormal sections over B 1 . as we showed in the previous chapter. they are stable invariants. only the Euler class has a chance of detecting nontriviality of the bundle. as H ∗ (Gn (C). splitting off a trivial line bundle. Then E is orientable iff its restriction to the 1 skeleton B 1 of B is orientable. Perhaps the simplest example is the tangent bundle of the sphere S 5 . but it is not always sufﬁcient. S 3 . since all loops in B can be deformed to lie in B 1 .74 Chapter 3 Characteristic Classes homology with arbitrary coefﬁcients iff all the Stiefel-Whitney and Pontryagin classes of E are zero. Z) is the polynomial ring Z[c1 . Z2 ) associated to E which is zero exactly when E is orientable. One reason why characteristic classes are not sufﬁcient to determine when a vector bundle is trivial is that. a vector bundle over a 1 dimensional CW complex is trivial iff it is orientable. Let us assume now that B is a CW complex. For a vector bundle E →B with Let us see how one might attack the question of what makes a vector bundle non- B path-connected. so we can also say that w1 (E) measures whether E is trivial over the 1 skeleton B 1 . Assuming that E is trivial over B 1 . Furthermore. Thus these tangent bundles have no nonzero characteristic classes. this homomorphism factors through the abelianization H1 (B) of π1 (B) . The vanishing of all the characteristic classes of a vector bundle is a necessary condition for it to be trivial. Tangent bundles to odd-dimensional spheres have this property since they have nonzero sections. as there exist nontrivial vector bundles whose characteristic classes are all zero.

the sections over B 1 pull back to sections of the pullback over ∂D 2 . where the obstructions lie in π3 (SO(n)) . It turns out to be a cellular cocycle. so the sections automatically extend over B 3 . π1 (SO(n))) . and so deﬁnes an element of H 2 (B.) B 3 Stiefel-Whitney class w2 (E) . does not depend on the original choice of the sections over B 1 . Thus we have an element of π1 (SO(n)) for each 2 cell.Characteristic Classes 75 we pull E back to a vector bundle over the disk D 2 via a characteristic map D 2 →B so we ask whether these sections extend to orthonormal sections over each 2 cell. Z for n = 3 . Z) . and it is not hard to see that the sections over B 1 extend to orthonormal sections over B 2 iff this element of π1 (SO(n)) is zero for each 2 cell. When n = 2 the class ω2 (E) lies in H 2 (B. and Z again for n ≥ 5 . w2 (E) is the image of e(E) under the If E is trivial over B 2 we can try to extend n orthonormal sections from B 2 to in the same way. with the obvious inclusion SO(2) SO(n) inducing we can take this map ∂D 2 →O(n) to have image in the identity component SO(n) of a surjection on π1 . The ﬁrst problem is that the groups πk−1 (SO(n)) become increasingly difﬁcult to compute. This element. These groups can also be computed explicitly with only a little more effort than for π1 and π2 . One can sim- . If we choose the trivialization to pullback bundle is trivial since D 2 is contractible. This group happens to be zero for all n . When n = 1 we have ω2 (E) = 0 automatically. Z ⊕ Z for n = 4 . For each cell the obstruction to extending lies in π2 (SO(n)) . cell by cell. and these are p1 (E) and e(E) . Unfortunately things become much more complicated as one proceeds to higher skeleta B k . At the next step we seek to extend over 4 cells. O(n) . Z)→H 2 (B. Z) and equals the Euler class e(E) . Z2 ) reducing coefﬁcients mod 2 . π3 (SO(n))) . ﬁxing the orthogonal complement of this plane. and by rechoosing these sections if necessary one can see that E is trivial over B 2 iff ω2 (E) is zero in H 2 (B. If for a 2 cell. In the special case n = 4 when π3 (SO(4)) is Z ⊕ Z the class ω4 (E) can be viewed as a pair of elements of H 4 (B. (When n = 2 . We can regard the function which assigns to each 2 cell the element of π1 (SO(n)) which is the obstruction to extending the sections over the 2 cell as a 2 dimensional cellular cochain with coefﬁcients in π1 (SO(n)) . By reasoning analogous to that for 2 cells we then obtain an obstruction ω4 (E) ∈ H 4 (B. and they are 0 for n ≤ 2 . They are ﬁnitely generated abelian groups whose nontorsion is known and fairly easy to describe. but the torsion is quite a mess. The sections over ∂D 2 determine a map ∂D 2 →O(n) . For n ≥ 3 it is not hard to compute that π1 (SO(n)) is Z2 . For n ≥ 3 we have ω2 (E) lying in H 2 (B. The group π1 (SO(n)) is obviously 0 for n = 1 . and it is Z for n = 2 since SO(2) is homeomorphic to S 1 . Z2 ) and it equals the homomorphism H 2 (B. so by choosing a trivialization the give the same orientation as the trivialization of E over B 1 determined by the sections. call it ω2 (E) . so E is trivial over B 2 in this case. so a generator of π1 (SO(n)) for n ≥ 2 is represented by the loop given by rotating a plane through 360 degrees. π1 (SO(n))) . In the cases when the coefﬁcient group is Z this turns out to be the Pontryagin class p1 (E) .

This is the Stiefel-Whitney class wn−k+1 (E) . πn−k (Vk (Rn ))) . the ﬁrst obstruction to triviality. This coefﬁcient group is either Z or Z2 . so to avoid this problem we reduce the obstruction mod 2 . so the ﬁrst possibility of a nontrivial obstruction occurs when k = n . For example if E is nonorientable. and this is exactly the Euler class e(E) . The ﬁrst obstruction to ﬁnding k orthonormal sections would lie in H n−k+1 (B. Another drawback to this approach is that it only associates a single cohomology class ωk (E) to each E . More generally suppose we look for k orthonormal sections. For k = 4i we have π4i−1 (SO(n)) = Z . Z2 ) to the existence of k orthonormal sections. As we will see. n−2 sections over the 2 skeleton. Thus we can ﬁnd n sections over the 0 skeleton (this is obvious). Pontryagin classes also detect the nontorsion in πk−1 (SO(n)) outside the stable range. where πk−1 (S n−1 ) = Z . Assuming we have a section over B k−1 . or equivalently a map ∂D k →S n−1 . As before we proceed by induction over skeleta of B . In case it is Z we again have the technical problem of not having a canonical identiﬁcation with Z . π4i−1 (SO(n))) for n > 4i . If we choose an orientation there is then a well-deﬁned ﬁrst obstruction to a section of unit vectors. and so on. Complex vector bundles can be treated in a similar fashion. we have a section over ∂D k . Z2 ) which is the Stiefel-Whitney class wn (E) . E →B is to look for obstructions to the existence of fewer than n orthonormal sections. Suppose for example that we ask whether there is a single section of unit vectors. depending on what n and k are. This means that there are no obtructions to ﬁnding k orthonormal sections over the (n − k) skeleton of B . For the complex . and the Pontryagin class pi (E) ∈ H 4i (B. but the obstructions ωk (E) in these cases are not detected by Stiefel-Whitney or Pontryagin classes except when i = 0 . Thus if k < n such an extension is always possible. n−1 sections over the 1 skeleton.76 Chapter 3 Characteristic Classes plify by restricting to the stable range n > k where Bott periodicity gives the values of πk−1 (SO(n)) . If one only considers the mod 2 value of the element of πn−1 (S n−1 ) associated to each n cell then it is not necessary to assume E is orientable and one obtains a well-deﬁned obstruction in H n (B. Pulling E back to the trivial bundle over D k via a characteristic map for the cell. Z) but there is a technical point that in order to identify the groups πn−1 (S n−1 ) with Z in a consistent way it is necessary to assume E is orientable. The other nontrivial stable groups are πk−1 (SO(n)) = Z2 for k = 8i+1. and the section over B k−1 extends over the k cell iff this map is nullhomotic. Z) detects the nontriviality of bundles with obstructions ω4i (E) ∈ H 4i (B. the ﬁrst nonvanishing homotopy group of Vk (Rn ) is πn−k . consider the problem of extending it over a k cell. 8i+2 . Then we get a well-deﬁned obstruction in H n−k+1 (B. Then the unit sphere S n−1 A more subtle way of measuring triviality of an n dimensional vector bundle in the ﬁbers of E is replaced by the Stiefel manifold Vk (Rn ) of orthonormal k tuples of vectors in Rn . Thus it appears that we get an obstruction ω(E) ∈ H n (B. we get no further characteristic classes associated to E .

which is Z in all cases. There is a unique sequence of functions w1 . Theorem 3. and CW complexes are paracompact. Z2 ) . so that the fundamental results of Chapter 1 apply.Stiefel-Whitney and Chern Classes Section 3. Note that (c) implies that the sum 1 + w1 (E) + w2 (E) + ··· has only ﬁnitely many nonzero terms.1 77 Stiefel manifold Vk (Cn ) the ﬁrst nonvanishing homotopy group is π2n−2k+1 . For the study of characteristic classes this is not an essential restriction since one can always pass to pullbacks over a CW approximation to a given base space. The two cases are quite similar. Chern classes for complex vector bundles. and then afterwards we will show that the two deﬁnitions give the same result. A technical point before we begin: We will assume without further mention that all base spaces of vector bundles are paracompact. The easiest way to deﬁne Pontryagin classes is in terms of Chern classes rather than as obstructions. Z2 ) . but for concreteness we will emphasize the real case. Axioms and Construction Here is the basic result about Stiefel-Whitney classes giving their most essential properties. This is the Chern class cn−k+1 (E) . w2 . is somewhat awkward to work with when one tries to prove certain general facts or calculate examples beyond the simplest cases. depending only on the isomorphism type of E . so this is what we will do later in the chapter after Stiefel-Whitney and Chern classes have been introduced. w(E2 ) where w = 1 + w1 + w2 + ··· ∈ H ∗ (B. while conceptually appealing. such that (a) wi (f ∗ (E)) = f ∗ (wi (E)) for a pullback f ∗ (E) .1.1. Stiefel-Whitney and Chern Classes Stiefel-Whitney classes are deﬁned for real vector bundles. (b) w(E1 ⊕ E2 ) = w(E1 ) (c) wi (E) = 0 if i > dim E . 3. up to a sign at least. although it relies on a little more algebraic topology. ··· assigning to each (d) For the canonical line bundle E →RP∞ . Z2 ) . The sum w(E) = 1+w1 (E)+w2 (E)+··· is the total Stiefel-Whitney class. with occasional comments on the minor modiﬁcations needed to treat the complex case. The deﬁnition of Stiefel-Whitney and Chern classes outlined in the introduction above. Z) to the existence of k orthonormal sections in a complex vector bundle E of dimension n . and there is a canonical identiﬁcation with Z . w1 (E) is a generator of H 1 (RP∞ . which can be regarded as axioms: real vector bundle E →B a class wi (E) ∈ H i (B. . For this reason we will start with a different deﬁnition that avoids these drawbacks and is quite simple and elegant in its own way. so we obtain an obstruction in H 2n−2k+2 (B.

where c0 = 1 . Z2 ) . This relation is sometimes called the Whitney sum formula. R) module H ∗ (E. such that (a) ci (f ∗ (E)) = f ∗ (ci (E)) for a pullback f ∗ (E) . and P (π ) is the natural projection sending each line in π −1 (b) to b ∈ B . c2 . ··· assigning to each c(E2 ) for c = 1 + c1 + c2 + ··· ∈ H ∗ (B. for R any commutative ring with identity. where P (E) is the space of all lines through the origin in all the ﬁbers of E . Z2 ) . (d) For the canonical line bundle E →CP∞ . and the ﬁrst three conditions would be satisﬁed. c1 (E) is a generator of H 2 (CP∞ . depending only on the isomorphism type of E . R) is a free H ∗ (B. Property (d) can be viewed as a nontriviality condition. (b) c(E1 ⊕ E2 ) = c(E1 ) (c) ci (E) = 0 if i > dim E . iﬁed in advance. a basis for the H ∗ (B. Z) spec- ci (E1 ) cj (E2 ) . the direct sum of the groups H i (B. A proof of this theorem using a fairly straightforward induction argument can be found in §4. As in the real case. From the formal identity (1 + w1 + w2 + ···)(1 + w1 + w2 + ···) = 1 + (w1 + w1 ) + (w2 + w1 w1 + w2 ) + ··· it follows that the formula w(E1 ⊕ E2 ) = w(E1 ) writing the relations wn (E1 ⊕ E2 ) = i+j=n wi (E1 ) w(E2 ) is just a compact way of wj (E2 ) . the quotient obtained by factoring out scalar multiplication in each ﬁber.2. of 3.2: Associated to a vector bundle π : E →B with ﬁber Rn is the Proof . the cohomology of certain ﬁber bundles. by setting αβ = p ∗ (α) β for α ∈ H ∗ (B) and E induces a The proof of both theorems will be based on the Leray-Hirsch theorem describing β ∈ H ∗ (E) . R) into a module over the ring H ∗ (B. Z) .D of [AT]. where w0 = 1 . Z) .1 and 3. If this were dropped we could set wi (E) = 0 for all E and all i > 0 . R) module provided that for each ﬁber F the inclusion F surjection on H ∗ (−−.78 Chapter 3 Characteristic Classes so this sum does indeed lie in H ∗ (B. R) is a free R module of ﬁnite rank for each n . There is a unique sequence of functions c1 . If p : E →B is a ﬁber bundle then we can use cup products to make H ∗ (E. For complex vector bundles there are analogous Chern classes: complex vector bundle E →B a class ci (E) ∈ H 2i (B. R) that map to a basis for H ∗ (F . projective bundle P (π ) : P (E)→B . We topologize P (E) as a quotient of the complement of the zero section of E . since replacing each function ci by ki ci for a ﬁxed integer k gives new functions satisfying (a)-(c). The condition (d) is now a normalization condition as well as a nontriviality condition. R) . R) can be obtained by choosing any set of elements in H ∗ (E. Then the Leray-Hirsch theorem asserts that this module H ∗ (E. R) under the map induced by the inclusion. Moreover. the formula in (b) for the total Chern classes can be rewritten in the form cn (E1 ⊕ E2 ) = i+j=n Theorem 3. R) and H n (F .

so P (E) is a ﬁber bundle over B with ﬁber RPn−1 . ··· . The Leray-Hirsch theorem then says that H ∗ (P (E).1 79 Over a neighborhood U in B where E is a product U × Rn . Z2 ) . and it follows that P (f )∗ takes the canonical P (f )∗ i class x = x(E) for P (E) to the canonical class x(E ) for P (E ) . Z2 ) module structure on H ∗ (P (E). To prove property (a). ··· . To do this we need classes xi ∈ H i (P (E). ﬁtting for certain classes wi (E) ∈ H i (B. Here wi (E)x i means P (π )∗ (wi (E)) x i . by RP∞ of different ﬁbers of P (E) are all homotopic. an embedding RPn−1 i = 0. By the uniqueness of this relation. this quotient is U × RPn−1 . Z2 ) . A map g : E1 ⊕ E2 →R∞ P (E1 ⊕ E2 ) − P (E1 ) and U2 = P (E1 ⊕ E2 ) − P (E2 ) . Z2 ) module with basis 1. x n−1 . consider a pullback f ∗ (E) = E . so the induced embeddings RPn−1 Theorem 1. Z2 ) is a free H ∗ (B. ··· . Z2 ) We would like to apply the Leray-Hirsch theorem for cohomology with Z2 co- Recall from the proof of Theorem 1. erator of H 1 (RP∞ .16 that there is a map g : E →R∞ that is a linear injection on each ﬁber. Proceeding to property (b). Projectivizing the map g by deleting zero vectors and then factoring out scalar multiplication produces a map P (g) : P (E)→RP∞ . Then P (π )∗ wi (E) x(E)n−i = i B −−− B −−→ P (f )∗ P (π )∗ wi (E) P (π )∗ f ∗ wi (E) i P (f )∗ x(E)n−i x(E )n−i = so the relation x(E) + w1 (E)x(E) n n−1 back to the relation x(E )n + f ∗ (w1 (E))x(E )n−1 + ··· + f ∗ (wn (E)) · 1 = 0 deﬁning wi (E ) . Let U1 = that deformation retract onto P (E2 ) and P (E1 ) . Z2 ) . so the classes x i are independent of the choice of g . Let α be a gen- restricting to generators of H i (RPn−1 . Z2 ) . If g : E →R∞ is a linear injection on E −−− E −−→ f ∼ f − − → π − − → π ﬁbers then so is g f . Consequently. x. Then the powers x i for hence αi pulls back to a generator of H i (RPn−1 . Z2 ) .16 that any two choices of g are homotopic through maps that are linear injections on ﬁbers. the inclusions of E1 and E2 into E1 ⊕ E2 give inclusions of P (E1 ) and P (E2 ) into P (E1 ⊕ E2 ) with P (E1 ) ∩ P (E2 ) = ∅ . wi (E ) = f ∗ (wi (E)) . n − 1 are the desired classes xi since a linear injection Rn →R∞ induces RP∞ for which α pulls back to a generator of H 1 (RPn−1 . Z2 ) and let x = P (g)∗ (α) ∈ H 1 (P (E). efﬁcients to this bundle P (E)→B . Z2 ) . Z2 ) in each ﬁber RPn−1 for i = 0. n − 1 . respectively. For completeness we deﬁne wi (E) = 0 for i > n and w0 (E) = 1 . Thus there is a unique relation of the form x n + w1 (E)x n−1 + ··· + wn (E) · 1 = 0 the deﬁnition of the H ∗ (B. x n can be expressed uniquely as a linear combination of these basis elements with coefﬁcients in H ∗ (B. We showed in the proof of into the diagram at the right. These are open sets in P (E1 ⊕ E2 ) + ··· + wn (E) · 1 = 0 deﬁning wi (E) pulls . Note that any two linear injections Rn →R∞ are homotopic through linear injections.Stiefel-Whitney and Chern Classes Section 3.

v) ∈ RP∞ × R∞ | v ∈ Property (c) holds by deﬁnition. hence ω1 pulls back to a class in the relative group H m (P (E1 ⊕ E2 ). Looking at this construction a little more closely. U 1 ∪ U 2 ) = 0 −− −→ H m ( P ( E 1 ⊕ E 2 ) ) × H n( P ( E 1 ⊕ E 2 ) ) Thus ω1 ω2 = j r +s=j Stiefel-Whitney classes of E1 ⊕ E2 . Z2 ) . The map P (π ) in this case is the identity. consider the classes ω1 = j wj (E1 )x m−j and ω2 = j r +s=j j wj (E2 )x n−j in H ∗ (P (E1 ⊕ E2 ). This construction can be repeated with L⊥ →P (E) in place of E →B . This pull}.80 Chapter 3 Characteristic Classes which is a linear injection on ﬁbers restricts to such a map on E1 and E2 . At the next stage we form P (L⊥ ) . back contains a natural one-dimensional subbundle L = { ( . U 1 ) −− H m + n ( P ( E1 ⊕ E2 ) . and so wj (E1 ⊕ E2 ) = E = { ( . Z2 ) module with basis 1. So P (g) is also the identity. so H ∗ (P (E). and p ∗ : H ∗ (B. Z2 ) ≈ H m (P (E1 ⊕ E2 ). Z2 )→H ∗ (F (E). bundles called the splitting principle: The proof of uniqueness of the classes wi will use a general property of vector dles. ··· . hence x(E) is a generator of H 1 (RP∞ . recall that the canonical line bundle is map g : E →R∞ which is a linear injection on ﬁbers can be taken to be g( . Z2 ) . U2 . As we have seen. For each vector bundle π : E →B there is a space F (E) and a map Proof: Consider the pullback P (π )∗ (E) of E via the map P (π ) : P (E)→B . P (E1 ). where and ) are orthogonal lines in E . U 2 ) × H n( P ( E1 ⊕ E2 ) . For (d). with Z2 coefﬁcients understood. so we have a splitting of the pullback as a sum L ⊕ L⊥ with the orthogonal bundle L⊥ having dimension one less than E . so the canonical class x ∈ H 1 (P (E1 ⊕ E2 ). Z2 ) is injective since one of − − − − − → −−−−−→ H m + n ( P ( E1⊕ E2) ) −−−−− −−−−− r +s=j wr (E1 )ws (E2 ) x m+n−j = 0 is the deﬁning relation for the wr (E1 )ws (E2 ) . v) = v . and similarly for ω2 . with cup product ω1 ω2 = wr (E1 )ws (E2 ) x m+n−j . whose points are pairs ( .3. The relation x(E) + w1 (E) · 1 = 0 then says that w1 (E) is a generator of H 1 (RP∞ . then shows that ω1 ω2 = 0 : H m( P ( E1 ⊕ E2 ) . v) ∈ P (E)× E with v ⊥ . p : F (E)→B such that the pullback p ∗ (E)→F (E) splits as a direct sum of line bun- Proposition 3. Z2 ) . x. Z2 ) is the free H ∗ (B. After ﬁnitely many repetitions we obtain the desired result. x n−1 the basis elements is 1 . the Leray-Hirsch theorem applies to P (E)→B . Z2 ) is injective. Z2 ) for E1 ⊕ E2 restricts to the canonical classes for E1 and E2 . − − − − − → . By the deﬁnition of the classes wj (E1 ) . Z2 ) . If E1 and E2 have dimensions m and n . the class ω1 restricts to zero in H m (P (E1 ). we see that the and in particular the induced map H ∗ (B. Z2 ) . The deﬁning } . v) ∈ P (E)× E | v ∈ An inner product on E pulls back to an inner product on the pullback bundle. L⊥ consists of pairs ( . Z2 )→H ∗ (P (E). Continuing in this way. The following commutative diagram.

Example 3. Z2 ) ≈ H (RP . with α now two-dimensional. Hence the Stiefel-Whitney classes wi (T S n ) are zero for i > 0 . Example 3. Property (b) implies that taking the direct sum of a bundle with a product bundle does not change its Stiefel-Whitney classes.5: Stability. Note that in property (d) for Stiefel-Whitney classes we could just as well use the canonical line bundle over RP1 instead of RP∞ since the inclusion RP1 1 ∞ 1 1 called ﬂags. Z2 ) . The deﬁning relation for the ci (E) ’s is modiﬁed to be x n − c1 (E)x n−1 + ··· + (−1)n cn (E) · 1 = 0 with alternating signs. Since the canonical line bundle is the universal line bundle. and F (E)→B is the ﬂag bundle associated to E . From the identity (1 + w1 + w2 + ···)(1 + w1 + w2 + ···) = 1 + (w1 + w1 ) + (w2 + w1 w1 + w2 ) + ··· . F (E) can be described as the space of all chains V1 ⊂ ··· ⊂ Vn of linear subspaces of ﬁbers of E with dim Vi = i . Z) ≈ Z[α] . For complex vector bundles we can use the same proof. Such chains are of points of F (E) as ﬂags does not depend on a choice of inner product in E . Note that the description Now we can ﬁnish the proof of Theorem 3. which must have wi = 0 for i > 0 since a point has trivial cohomology in positive dimensions.Stiefel-Whitney and Chern Classes ﬁnal base space F (E) is the space of all orthogonal splittings Section 3. since the deﬁning relation in this case is x(E) − c1 (E) · 1 = 0 and x(E) = α . the naturality of Stiefel-Whitney classes. Property (b) extends this to sums of line bundles. but it has the advantage that the canonical line bundle E →CP∞ has c1 (E) = α rather than −α . and the vector bundle over F (E) consists of all n tuples of vectors in these lines.1. The analogous remark for Chern classes is valid as well. Property (c) then determines all the wi ’s for this bundle.4. property (a) therefore determines the classes wi for all line bundles. for the canonical line bundle E →RP∞ . Property (d) determines w1 (E) RP∞ induces an isomorphism H (RP . but with Z coefﬁcients since H ∗ (CP∞ . Alternatively. so it does not affect the proofs of properties (a)–(c). produces a trivial bundle. which is a trivial line bundle. For example. Property (a). In this sense StiefelWhitney classes are stable. implies that a product bundle E = B × Rn has wi (E) = 0 for i > 0 since a product is the pullback of a bundle over a point.1 ⊕ ··· ⊕ 81 1 n of ﬁbers of E as sums of lines. and ﬁnally the splitting principle implies that the wi ’s are determined for all bundles. This is equivalent to changing the sign of α . the tangent bundle T S n to S n is stably trivial since its direct sum with the normal bundle to S n in Rn+1 .

v) where E1 (Rn+1 ) of the canonical line bundle to the subspace RPn ⊂ RP∞ does have an is such an is a line through the origin in Rn+1 and v is a vector orthogonal to inverse. Let us describe an n dimensional vector bundle E →B with wi (E) the canonical line bundle over G1 = RP∞ with itself. but the same reasoning applies to Chern classes. Hence wi ((E1 )n ) is the w((E1 )n ) = i elementary symmetric polynomial σi in the αj ’s. αn ] ≈ H ∗ ((RP∞ )n . if E1 ⊕ E2 is the trivial bundle. Z2 ) . Z2 ) . Given an n dimensional vector bundle E →B we know that the pullback to F (E) . nonzero for each i ≤ n . Of course for Z2 coefﬁcients the signs do not matter. ··· . note that Proposition 1. 2 w1 = −w1 and w2 = −w1 w1 − w2 = w1 − w2 .7. wi (E) = 0 for i > dim E . This vector bundle is the direct + αi ) ∈ Z2 [α1 . αn ] . if n = 3 then σ1 = α1 +α2 +α3 . so i (1 th Example 3. then we have the case that ai = 0 for i > 0 and so w(E1 ) determines w(E2 ) uniquely by explicit formulas that can be worked out. In this example we see that the wi ’s and ci ’s can be identiﬁed with elementary symmetric functions. the bundle E1 (Rn+1 ) consisting of pairs ( . Example w(E1 (R∞ )) = 1 + ω where ω is a generator of H 1 (RP∞ . For we have 3.4. So we see there is no chance of choosing such bundles E for varying n so that they ﬁt together to form a single bundle over RP∞ . we have an n dimensional bundle whose ﬁrst n Stiefel-Whitney classes are all nonzero. For example. and σ3 = α1 α2 α3 .6. This will be the n fold Cartesian product (E1 )n →(G1 )n of ∗ ∗ sum π1 (E1 ) ⊕ ··· ⊕ πn (E1 ) where πi : (G1 )n →G1 is projection onto the i th factor. and hence w(E) must be (1 + ω)−1 = 1 + ω + ω2 + ··· since we are using Z2 coefﬁcients. the sum of all the products of i different αj ’s.4 says that the restriction ⊥ ‘inverse’ bundle. For example. E →RP∞ whose sum with the canonical line bundle E1 (R∞ ) is trivial. To further delineate the question. Since each σi with i ≤ n is nonzero in Z2 [α1 . Thus wi (E) = ωi .82 Chapter 3 Characteristic Classes we see that w(E1 ) and w(E1 ⊕ E2 ) determine w(E2 ) since the equations w1 + w1 = a1 w2 + w1 w1 + w2 = a2 ··· i wn−i wi = an can be solved successively for the wi ’s in terms of the wi ’s and ai ’s. this contradicts the fact that This shows the necessity of the compactness assumption in Proposition 1. However. and in fact this can be done in general using the splitting principle. In particular. and since wn (E) = ωn ≠ 0 . Z2 ) for all i . ··· . The same reasoning applies in the complex case to show that the n fold Cartesian product of the canonical line bundle over CP∞ has its ﬁrst n Chern classes nonzero. In fact. σ2 = α1 α2 +α1 α3 +α2 α3 . E must be at least n dimensional. with Z coefﬁcients. Let us illustrate this principle by showing that there is no bundle which is nonzero in H ∗ (RP∞ . But for any bundle E →RP n whose sum with E1 (R n+1 ) is trivial we must have w(E) = 1 + ω + ··· + ωn .

consisting of all the real and all the complex lines in ﬁbers of E . then w2i+1 (E) = 0 and w2i (E) is the image of For example. 2n dimensional real vector bundle. which is x C (E)n + c 1 (E)x C (E)n−1 + ··· + c n (E) · 1 = 0 . Z2 ) . Z2 ) . Z2 ) . over R and C . Z) . back to w(L1 ⊕ ··· ⊕ Ln ) = (1 + α1 ) ··· (1 + αn ) = 1 + σ1 + ··· + σn . Z2 ) . so if β is the standard generator of H 2 (CP∞ . the real lines in a complex line. respectively. This means that w2i+1 (E) = 0 and w2i (E) = c i (E) . This projection p ﬁts into a commutative diagram line is all the real scalar multiples of any nonzero vector in it and a complex line is all RP 2 n CP n 1 −−−→ RP ( E ) −−−− RP ∞ −−− −−− −−−→ −−− − − − − − → − − − → p CP ( g ) RP ( g ) where the left column is the restriction of p to a ﬁber of E and the maps RP(g) and CP(g) are obtained by projectivizing. so wi (E) pulls splits as a sum L1 ⊕ ··· ⊕ Ln →F (E) . the Z2 reduction β ∈ H 2 (CP∞ . since the canonical complex line bundle over CP∞ has c1 a generator of H 2 (CP∞ . so by the proposition this 2 dimensional real vector bundle E →S 2 has w2 (E) ≠ 0 . a map g : E →C∞ which is a C linear injection on ﬁbers. There is a natural projection p : RP(E)→CP(E) sending each real line to the complex line containing it. Z2 ) such that wi (E) becomes the i th elementary symmetric polynomial in the elements α1 . Hence the Z2 reduction x C (E) = CP(g)∗ (β) ∈ H 2 (CP(E).8. Z) . Letting αi = w1 (Li ) .1 83 back to σi . Z2 ) of the this diagram are ﬁber bundles with ﬁber RP1 . The back to a generator of H 2 (RP∞ . Z)→H 2i (B.Stiefel-Whitney and Chern Classes Section 3. with Z2 coefﬁcients. Z2 ) in a larger ring H ∗ (F (E). ··· . It is easy to see that all three vertical maps in Leray-Hirsch theorem applies to the bundle RP∞ →CP∞ . Proof: The bundle E has two projectivizations RP(E) and CP(E) . Regarding an n dimensional complex vector bundle E →B as a ci (E) under the coefﬁcient homomorphism H 2i (B. − − − → 1 −−−→ C P ( E ) −−−− C P ∞ −−− −−− −−−→ −−− . Consequently the Z2 reduction of the deﬁning relation for the Chern classes of E . the same is true for its restriction over S 2 = CP1 . αn of H ∗ (F (E). we see that w(E) pulls Besides the evident formal similarity between Stiefel-Whitney and Chern classes there is also a direct relation: Proposition 3. Z2 ) pulls H 1 (RP∞ . since a real the complex scalar multiples. pulls back to the relation xR (E)2n + c 1 (E)xR (E)2n−2 + ··· + c n (E) · 1 = 0 . Z2 ) . which is the deﬁning relation for the Stiefel-Whitney classes of E . namely the square α2 of the generator α ∈ basic class xC (E) = CP(g)∗ (β) pulls back to the square of the basic class xR (E) = RP(g)∗ (α) ∈ H 1 (RP(E). Thus we have embedded H ∗ (B.

A different proof will be given in §3. Proofs of this can be found in [van der Waerden. then π pulls (E1 )n back to itself. Thus the composition Z2 [w1 . Z2 ) as a Z2 vector space. ··· . which means that f ∗ = π ∗ f ∗ . ··· . Similarly. It sufﬁces r to ﬁnd a CW structure on Gn in which the r cells are in one-to-one correspondence r with monomials w11 ··· wnn of dimension r = r1 + 2r2 + ··· + nrn . Z2 ≈ Z2 [α1 . Z2 ) . αn ] . since if π : (RP∞ )n →(RP∞ )n is an arbitrary permutation of the factors. It is a classical algebraic result that the polynomials σi are algebraically independent in Z2 [α1 . the i th elementary symmetric polynomial. ··· . the classes wi (En ) generate a poly- Whitney classes wi = wi (En ) of the universal bundle En →Gn . αn ] ≈ H ∗ ((RP∞ )n .2. Z2 ) . and the corresponding statement for Chern classes holds as well: 3. Then the composition Z2 [w1 . r Monomials w11 ··· wnn of dimension r correspond to n tuples (r1 . Z2 )→H ∗ ((RP∞ )n . Let f : (RP∞ )n →Gn be the classifying map for the must also have all its Stiefel-Whitney classes w1 (En ). Such n tuples in turn correspond to partitions of r into at . Z2 ) . ··· .9. p. This subalgebra is in fact equal to H ∗ (Gn . ··· . H ∗ (Gn . ··· . Proof: considered above. in the sal bundle En (C∞ )→Gn (C∞ ) . ··· . wn ]→H ∗ (Gn . hence also the map nomial subalgebra Z2 [w1 . wn (En ) nonzero. complex case H ∗ (Gn (C∞ ). cn ] where ci = ci (En (C∞ )) for the univer- Theorem The proof we give here for this basic result will be a fairly quick application of the CW structure on Gn constructed at the end of §1. ··· . ··· . since the number of r cells in a CW complex X is an upper bound on the dimension of H r (X. ··· . αn ] sends wi to σi .7 and naturality it follows that the universal bundle En →Gn much stronger statement is true. We have noted that the image of f ∗ contains the symmetric polynomials in Z2 [α1 . Z2 ) . In other words. Z2 ) is the polynomial ring Z2 [w1 . Z2 ) .84 Chapter 3 Characteristic Classes Cohomology of Grassmannians From Example 3. Z) ≈ Z[c1 . wn ] ⊂ H ∗ (Gn . αn ] is injective. 134] for example. To ﬁnish the proof in the real case it remains to see that f ∗ is injective. and for notational simplicity let wi = wi (En ) . wn ] on the StiefelZ2 [w1 . which is somewhat more subtle. and a surjective linear map between ﬁnite-dimensional vector spaces is injective if the dimension of the domain is not greater than the dimension of the range. wn ]→Z2 [α1 . wn ] → H ∗ (Gn . so f π f . §26] or [Lang. ··· . so the image of f ∗ is Consider a map f : (RP∞ )n →Gn which pulls En back to the bundle (E1 )n invariant under π ∗ : H ∗ ((RP∞ )n . rn ) with r r = r1 + 2r2 + ··· + nrn .2 where we also compute the cohomology of Gn with Z coefﬁcients. ··· . In fact a n fold Cartesion product (E1 )n of the canonical line bundle E1 . Z2 ) → H ∗ f∗ (RP∞ )n . but the latter map is just the same permutation of the variables αi . The opposite inclusion holds as well.

i For example the monomial w1 corresponds to the cell whose associated partition is the trivial partition i = i . σ3 − 3) 0 0 0 0 1 0 0 0 0 1 0 1 1 0 0 1 1 2 1 2 3 (σ1 . the cells are all even-dimensional. We know also that Vect1 (X) = . ··· . In particular. rn ) Section 3. σ3 ) 1 1 1 1 2 1 1 2 2 3 2 3 3 2 3 4 4 5 4 5 6 dimension 0 1 2 2 3 3 3 The cell structure on Gn constructed in §1. In the parallel situation of Chern classes this question has very classical roots in algebraic geometry. Accessible expositions of this subject from a modern viewpoint can be found in [Fulton] and [Hiller]. r2 . represented by the cellular cochain assigning the value 1 to the cell and 0 to all the other cells. the formulas for multiplying these dual classes become rather complicated.2 has one cell of dimension (σ1 − 1) + (σ2 − 2) + ··· + (σn − n) for each increasing sequence 0 < σ1 < σ2 < ··· < σn .1 85 ← → rn ≤ rn + rn−1 ≤ ··· ≤ rn + rn−1 + ··· + r1 . corresponding to the strictly increasing sequence 0 < σ1 < σ2 < ··· < σn . Applications of w1 and c1 We saw in §1. If one replaces the basis of monomials by the more geometric basis of cohomology classes dual to cells. So we are done in the real case. Injectivity of f ∗ then follows from the algebraic fact that a surjective homomorphism between free abelian groups of ﬁnite rank is injective if the rank of the domain is not greater than the rank of the range.Stiefel-Whitney and Chern Classes most n integers. via the correspondence (r1 . For example. so the cellular boundary maps for Gn (C∞ ) are all trivial and the cohomology with Z coefﬁcients consists of a Z summand for each cell. r One might guess that the monomial w11 ··· wnn corresponding to a given cell of r The complex case is entirely similar. The CW structure on Gn (C∞ ) described in §1. and the rules for multiplying cohomology classes dual to cells are part of the so-called Schubert calculus. This is true for the classes wi themselves.1 that the set Vect1 (X) of isomorphism classes of line bundles over X forms a group with respect to tensor product. σ2 − 2. but unfortunately it is not true in general.2 also has these extra factors of 2 in the dimensions of its cells. r3 ) 1 w1 w2 2 w1 w3 w1 w2 3 w1 0 1 0 2 0 1 3 0 0 1 0 0 1 0 0 0 0 0 1 0 0 (σ1 − 1. rather than i . keeping in mind that ci has dimension 2i Gn in the way described above was the cohomology class dual to this cell. when n = 3 we have: (r1 . σ2 . Such a partition becomes the sequence σ1 − 1 ≤ σ2 − 2 ≤ ··· ≤ σn − n . but the cohomology class dual to this cell is wi where 1 + w1 + w2 + ··· is the multiplicative inverse of 1 + w1 + w2 + ··· .

c1 (L) is the generator α of H 2 (CP∞ ) . Hence c1 (L1 ⊗ L2 ) = α1 +α2 = c1 (L1 )+c1 (L2 ) . Since initeness let us describe the complex case. Z) . C Proposition 3. The inclusion G1 ∨G1 ⊂ G1 × G1 induces an isomorphism on H 2 .10. Which of the two generators we choose for u is as the composition [X. As noted above. and this is a K(Z. so since the composition is a bijection. an Eilenberg-MacLane space K(Z2 . Z) . L1 ⊗ L2 restricts to L2 over the second G1 .86 Chapter 3 Characteristic Classes [X. and true for c1 : Vect1 (X)→H 2 (X. α2 ] where so to compute c1 (L1 ⊗ L2 ) it sufﬁces to restrict to G1 ∨G1 . the second map c1 must be a bijection also. For complex line bundles we have G1 (C∞ ) = CP∞ . so the restriction of L1 ⊗ L2 over this G1 is L1 ⊗ 1 ≈ L1 . if X is a CW complex. and the more precise statement is that this bijection is given by the map [f ] f ∗ (u) for some class u ∈ H 2 (CP∞ . we ﬁrst prove that c1 (L1 ⊗ L2 ) = c1 (L1 ) + c1 (L2 ) for the bundle L→G1 = CP∞ under the projections p1 . K(G. see Theorem 4. The function w1 : Vect1 (X)→H 1 (X. 2) . Z) . The class u must be a generator. C is an isomorphism if X has the homotopy type of a CW complex. for example. CP∞ ] ≈ H 2 (X. To show that c1 : Vect1 (X)→H 2 (X) is C Proof: The argument is the same in both the real and complex cases. n)] ≈ H n (X. CP∞ ]→Vect1 (X)→H 2 (X. G1 (R∞ )] . So c1 (L1 ⊗ L2 ) restricted to G1 ∨G1 is α1 +α2 restricted to G1 ∨G1 . f2 : X →G1 . The ﬁrst Stiefel-Whitney class w1 is closely related to orientability: . Similarly. so c1 (E1 ⊗E2 ) = c1 (F ∗ (L1 )⊗F ∗ (L2 )) = c1 (F ∗ (L1 ⊗L2 )) = F ∗ (c1 (L1 ⊗L2 )) = F ∗ (c1 (L1 ) + c1 (L2 )) = F ∗ (c1 (L1 )) + F ∗ (c1 (L2 )) = c1 (F ∗ (L1 )) + c1 (F ∗ (L2 )) = c1 (E1 ) + c1 (E2 ). [f ] C not important. f2 ) : X →G1 × G1 we have F ∗ (Li ) = fi∗ (L) ≈ Ei . so for def- L1 ⊗ L2 →G1 × G1 where L1 and L2 are the pullbacks of the canonical line bundle ∗ αi = pi (α) = c1 (Li ) . Z2 ) is a homomorphism. 1) . and G1 (R∞ ) is just RP∞ . Thus one might ask whether the groups Vect1 (X) and H 1 (X. The ﬁrst map in this composition is a bijection. there is a bijection [X. ∗ ∗ and E2 now follows by naturality: We have E1 ≈ f1 (L) and E2 ≈ f2 (L) for maps The general case of the formula c1 (E1 ⊗ E2 ) = c1 (E1 ) + c1 (E2 ) for line bundles E1 f1 . we know that H ∗ (G1 × G1 ) ≈ Z[α1 . For the map F = (f1 . It is a basic fact in algebraic topology that [X. so we may take it to be the class α .56 of [AT]. so the corresponding question is whether Vect1 (X) is isomorphic to H 2 (X. The map [f ] f ∗ (L) f ∗ (α) factors c1 (f ∗ (L)) = f ∗ (c1 (L)) = f ∗ (α) . otherwise the map would not always be surjective. Z2 ) are isomorphic. Z) . Over the ﬁrst G1 the bundle L2 is the trivial line bundle. p2 : G1 × G1 →G1 onto the two factors. G) when X has the homotopy type of a CW complex. Z) . The same is also a homomorphism.

orientability of E . assuming that X is homotopy equivalent to a CW complex.e. w1 (E) = 0 iff this f∗ is trivial. An interpretation of the other classes wi as obstructions will be given in the Appendix to this chapter. Z2 ) corresponds via (∗) to this isomorphism π1 (Gn ) ≈ Z2 since w1 (En ) is the unique nontrivial elf ∗ (w1 (En )) corresponds under (∗) to the homomorphism f∗ : π1 (X)→π1 (Gn ) ≈ corresponding under (∗) to the induced map f∗ : π1 (X)→π1 (Gn ) ≈ Z2 . i. we have w1 (E) ement of H 1 (Gn . and w1 (En ) ∈ H 1 (Gn . Hence universal cover Gn . Thus if we choose f so that f ∗ (En ) is a given vector bundle E .Stiefel-Whitney and Chern Classes Section 3.. By naturality of (∗) it follows that for any map f : X →Gn . Thus w1 can be viewed as the obstruction to orientability of vector bundles. which is exactly the condition for lifting f to the . Z2 ) ≈ Hom(π1 (X).1 87 Proposition 3. A vector bundle E →X is orientable iff w1 (E) = 0 . Z2 ) . By restricting to path≈ components we may further assume X is connected. When X = Gn we have π1 (Gn ) ≈ Z2 . Z2 . There are natural isomorphisms H 1 (X. Z2 ) → Hom(H1 (X).11. Proof: (∗) Without loss we may assume X is a CW complex. Z2 ) → from the universal coefﬁcient theorem and the fact that H1 (X) is the abelianization of π1 (X) .

and it is clear that the naturality property is satisﬁed.. Whitney classes since we showed in Theorem 3. . Consider ﬁrst the . S(E)) . ∗ ∗ ∗ Euler class can also be deﬁned as the class corresponding to c isomorphism. G) . −− The vertical map p ∗ is an isomorphism since p is a homotopy equivalence from D(E) to B . The square containing the map p (b) ∗ e commutes since for b ∈ H i−n (B) we have j ∗ Φ(b) = j ∗ (p ∗ (b) j (c) .88 Chapter 3 Characteristic Classes 3. Z) . for the mial ring Z2 [w1 . wn ] . − H i( D ( E ). The map Φ is an isomorphism whenever a Thom class exists. Our goal in this section is to describe the more reﬁned characteristic classes for real vector bundles that arise when we take cohomology with integer coefﬁcients rather than Z2 coefﬁcients. The c under the Thom c. Thus characteristic classes correspond bijectively with cohomology classes of Gn . − → − − − → − − → − − → − − − → ∗ → − − i n → − − → − − == ≈ Φ ≈ p∗ i . S ( E ) ) − . Z2 ) and H ∗ (Gn (C). Z2 ) is the polynoWith Z2 coefﬁcients all characteristic classes are simply polynomials in the Stiefel- element x ∈ H k (Gn . Thom classes with Z coefﬁcients exist for all orientable real vector bundles. this gives the Gysin sequence as the lower row of the following commutative diagram. S(E)) deﬁned by Φ(b) = p ∗ (b) class c ∈ H (D(E).2. . S ( E ) ) − H i ( D ( E )) − H i(S ( E )) − H i +1( D ( E ). Once one has the Thom isomorphism.9 that H ∗ (Gn . or in other words the restriction of the Thom class to the zero section of E . ··· . Similarly for complex vector bundles all characteristic classes with Z coefﬁcients are polynomials in the Chern classes since H ∗ (Gn (C). Since f is unique up to homotopy.. cn ] . as shown in Corollary 4D. S n−1 ) . Euler and Pontryagin Classes bundle E →B of dimension n a class x(E) ∈ H k (B. ··· . and with Z2 coefﬁcients they exist for all vector bundles. − − H −→ ( B ) −− − H ( B ) −− H (S ( E )) − − H −− → − − −− −→ −− −→ i e p∗ ≈ Φ i n +1 ( B ) −−→ .9 of [AT]. G) deﬁnes a characteristic class by the rule x(E) = f ∗ (x) where universal bundle En →Gn there is the class x = x(En ) ∈ H k (Gn . In particular. . . a class whose restriction to each ﬁber is a generator of H n (D n . The Euler class e ∈ H n (B) is deﬁned to be (p ∗ )−1 j ∗ (c) . which equals p (b ∗ ∗ ∗ c) = e) = p (b) c = j (c) ∗ ∗ p (e) since p (e) = j (c) . x(E) is well-deﬁned. G) . such A characteristic class can be deﬁned to be a function associating to each vector E ≈ f ∗ (En ) for f : B →Gn . This is an easy consequence of the Thom c for a Thom n The main tool we will use will be the Gysin exact sequence associated to an isomorphism Φ : H i (B)→H i+n (D(E). Z) ≈ Z[c1 . This is shown in Theorem 4D. whose upper row is the exact sequence for the pair (D(E).10 of [AT].9 computing H ∗ (Gn . . . any that the naturality property x(f ∗ (E)) = f ∗ (x(E)) is satisﬁed. S(E)) : j . since Φ(e) = p (e) c=c As a warm-up application of the Gysin sequence let us use it to give a different proof of Theorem 3. Conversely. n dimensional real vector bundle p : E →B . for some ﬁxed n and k .

where G1 = RP∞ and H ∗ (RP∞ ) ≈ case n = 0 . hence wn (E) = αn ≠ 0 . so η(wj (En )) = wj (En−1 ) . where α generates H 1 (RP∞ ) . with Z2 coefﬁcients. By deﬁnition the map η is the comπ → Gn . Thus e and wn (En ) generate wn (En−1 ) = 0 . Now we argue that each element ξ ∈ H k (Gn ) can be expressed as a unique polynomial in the classes wi = wi (En ) . and p ∗ (En−1 ) consists of the triples (v. ) mology homomorphism π ∗ takes wj (En )) to wj (L ⊕ p ∗ (En−1 )) = wj (p ∗ (En−1 )) = By induction on n . ) . v. ) . Thus the coho- p ∗ (wj (En−1 )) . η(ξ) is a unique polynomial f in the wi (En−1 ) ’s by the basic induction on n . Its ﬁber is S . by induction on k . w. wn−1 ) + wn g(w1 . ) where . v a unit vector. ζ is a unique polynomial g in the wi ’s. position H ∗ (Gn )→H ∗ (S(En )) ← H ∗ (Gn−1 ) induced from Gn−1 ← S(En ) ≈ p We show ﬁrst that η(wj (En )) = wj (En−1 ) . has the form ··· → H i (Gn ) → H i+n (Gn ) → H i+n (Gn−1 ) → H i+1 (Gn ) → ··· η e where e ∈ H n (Gn . The sphere bundle S(En ) is the space of pairs (v.2 89 real case. w. this Z2 n dimensional vector bundles. ) where subbundle of triples (v. Since η(wj (En )) = wj (En−1 ) . p induces an isomorphism on all homotopy groups. The induction can start with the case n = 1 . The pullback π ∗ (En ) consists of triples (v. Z2 ) is the Z2 Euler class. Using this isomorphism p ∗ the Gysin sequence. By exactness. the same Z2 . wn ) . wn−1 ) is in Ker η = Im( wn ) . but the bundle E →RP∞ which is the direct sum of n copies of the canonical line bundle has total Stiefel-Whitney class w(E) = (1 + α)n . all the unit vectors in R∞ orthogonal to a given (n − 1) dimensional subspace. ) to the (n − 1) dimensional linear subspace v⊥ ⊂ ∞ orthogonal to v . so e = wn (En ) . It is an exercise to check that p is a ﬁber bundle. j < n . By induction on k . ··· . This pullback splits naturally as a sum L ⊕ p ∗ (En−1 ) where L is the ∈ Gn and v. tv. hence also on all cohomology groups. Or we could start with the trivial → H i (Gn ) → H i+n (Gn ) → H i+n (Gn−1 ) → 0 η e is the kernel of η : H n (Gn )→H n (Gn−1 ) . Moreover. The proof will be by induction on n using the Gysin sequence for the univeris an n dimensional linear subspace of R∞ and v is a unit vector in sal bundle En π → Gn . The class wn (En ) lies in this kernel since e : H 0 (Gn )→H n (Gn ) is a Z2 generated by e . First. ··· . H ∗ (Gn−1 ) is the polynomial ring on the classes wj (En−1 ) . Since S ∞ is contractible. wn (En ) ≠ 0 .Euler and Pontryagin Classes Section 3. Then ξ − f (w1 . There is a nat- ural map p : S(En )→Gn−1 sending (v. t ∈ R . having the section (v. Z2 ) . hence has the form ζ wn for ζ ∈ H k−n (Gn ) which is unique since wn is injective. Thus we have ξ expressed uniquely as a polynomial f (w1 . ··· . w ∈ with with w ∈ v ⊥ . The line bundle L is trivial. since if wn (En ) = 0 then wn is zero for all . the maps η are surjective and the Gysin sequence splits into short exact sequences 0 The image of Z2 [w1 ] since w1 (E1 ) is a generator of H 1 (RP∞ .

the only nonzero element of w1 is injective in all x where x ∈ H 1 (Gn . is a polynomial ring Z[p1 . ··· . Z) . The argument from the real case shows that cn is a multiple me for some m ∈ Z . state the result. wn ] . hence x = w1 . Z2 )→H ∗ (Gn . Z) ≈ Z . the theorem follows in the real case. H 0 (Gn . or in other words. so En = π ∗ (En ) . Since Gn is connected. Z2 ) → H 1 (Gn . so we have a Gysin sequence beginning 0 → H 0 (Gn . Z2 ) → H 0 (Gn . hence m = ±1 and e = ±cn . Z2 ) π→ H i (Gn . Z2 ) is surjective with kernel the ideal We can also compute H ∗ (Gn . Since H (Gn . to be zero. Z2 ) ≈ Z2 [w1 . We noted earlier that complex vector bundles always have a Gysin sequence with Z coefﬁcients. Z2 ) where Gn is the oriented Grassmannian. To generated by w1 . the map w1 ∗ ∗ dimensions. classiﬁed by f : CP∞ →Gn (C∞ ) . Since Gn is simply-connected. This is just the answer one would hope for. ··· . ··· . Z2 ) ≈ Z2 [w2 . The only elaboration needed to extend the preceding proof to the complex case is at the step where we showed the Z2 Euler class is wn . p2 . Z2 ) → 0 from which the conclusion is immediate. wn has a unique expression in this form. but with one of the Pontryagin classes replaced by an Euler class when n is even. Z2 ) ≈ Z2 and so the map 1 H (Gn .90 Chapter 3 Characteristic Classes Since every polynomial in w1 . wi = wi (En ) = π ∗ (wi ) ∈ H i (Gn . Z2 ) ≈ Z2 [w2 . Z2 ) → H i (Gn . wn ] . The rest of the proof goes through without change. Z) and H ∗ (Gn . Z) . modulo elements of order 2 which are just certain polynomials in Stiefel-Whitney classes. ··· . we have αn = cn (E) = f ∗ (cn ) = mf ∗ (e) in for the bundle E →CP∞ which is the direct sum of n copies of the canonical line H 2n (CP∞ . ···] on certain classes pi of dimension 4i . . Proposition 3. Virtually the same argument works in the complex case.12. It turns out that H ∗ (Gn . so the Gysin sequence breaks up into short exact sequences 0 → H i (Gn . Z) . Z2 ) → H 0 (Gn . where wi = wi (En ) . wn ] is the simplest thing that could happen. and let π ∗ : H ∗ (Gn . called Pontryagin classes. to ﬁnd all characteristic classes for real vector bundles with Z coefﬁcients. Z2 ) x is injective. e being now the Z Euler class. with αn a generator. Z2 ) . Z2 ) . let π : Gn →Gn be the covering projection. There is a similar statement for H ∗ (Gn . hence H ∗ (Gn . The goal for the rest of this section is to determine H ∗ (Gn . rather than the Z2 coefﬁcients used for Stiefel-Whitney classes. so the isomorphism H ∗ (Gn . Then bundle. Z2 ) is the Z2 Euler class. w1 has Proof: The 2 sheeted covering π : Gn →Gn can be regarded as the unit sphere bundle of a 1 dimensional vector bundle.

S(E). (a) An orientation of a vector bundle E →B induces an orientation of a pullback bundle f ∗ (E) such that e(f ∗ (E)) = f ∗ (e(E)) . the image of c under the composition H n (D(E). Then ∗ p2 (c(E2 )) is a Thom class for E1 ⊕ E2 since in each ﬁber . Z) where the ﬁrst map is the usual passage from relative to absolute cohomology and the second map is induced by the inclusion B D(E) as the zero section. (b) Orientations of vector bundles E1 →B and E2 →B determine an orientation of the sum E1 ⊕ E2 such that e(E1 ⊕ E2 ) = e(E1 ) e(E2 ) . the coefﬁcient homomorplex vector bundle E there is the relation e(E) = cn (E) ∈ H 2n (B. f ∗ (E) . Rn − {0}. E .2 91 The Euler Class bundle E →B is the restriction of a Thom class c ∈ H n (D(E). Z2 ) carries e(E) to wn (E) . that is. we get e(f ∗ (E)) = f ∗ (e(E)) . and the restriction of c to each ﬁber is in turn determined by an orientation of the bundle. Z) in each ﬁber Rn of f ∗ (E) . Z) associated to oriRecall that the Euler class e(E) ∈ H n (B. Z)→H n (B. However. If E1 is m dimensional we can view a Thom class c(E1 ) as lying in H m (E1 . For a pullback f ∗ (c(E)) restricts to a generator of H n (Rn . Z) of an orientable n dimensional vector Proposition 3. Z) in each ﬁber Rn . we have a map f : f ∗ (E)→E which is a linear isomorphism in each ﬁber. For an n dimensional com- Proof: (a) For an n dimensional vector bundle E . S(E). and likewise we have p2 : E1 ⊕ E2 →E2 . for a suitable choice of orientation of E . Passing from relative to absolute cohomology classes (b) There is a natural projection p1 : E1 ⊕ E2 →E1 which is linear in each ﬁber. ∗ the product p1 (c(E1 )) Similarly we have a Thom class c(E2 ) ∈ H n (E2 . Rn − {0}.10 of [AT] implies that c is determined by its restriction to each ﬁber. (e) e(E) = 0 if E has a nowhere-zero section. A Thom class for E can be viewed as an element of H n (E. (d) e(E) = −e(E) if the ﬁbers of E have odd dimension. By its deﬁnition. There are exactly two choices of orientation for each path-component of B . ented n dimensional vector bundles E →B : Here are the basic properties of Euler classes e(E) ∈ H n (B. (c) For an orientable n dimensional real vector bundle E . Z) . E1 ) where E1 is the complement of the zero section in E1 . phism H n (B.Euler and Pontryagin Classes Section 3. so Thus f ∗ (c(E)) = c(f ∗ (E)) . Z)→H n (D(E). e(E) depends on the choice of c . Choosing the opposite orientation changes the sign of c . the assertion (*) in the construction of a Thom class in Theorem 4D. so in fact e(E) depends only on the choice of an orientation of E . let E ⊂ E be the complement of the zero section. Z)→H n (B. E2 ) if E2 has dimension n . Z) to the zero section. Z) which restricts to a generator of H n (Rn .13. and then restricting to zero sections.

11 of [AT]. so it sufﬁces to show that c 2 is twice a generator of H 2n (E.92 Chapter 3 Characteristic Classes Rm × Rn = Rm+n we have the cup product H m (Rm+n . Z) s i → B → S(E) → D(E) is homo∗ vectors to have unit length. so e(E) = −e(E) . the map j ∗ is zero by exactness. e(T S n ) = 0 either by part (d) of the proposition since H ∗ (S n . xn ) . −xn+1 . Z) under the Thom isomorphism.14. Proof: For even n . −x) . e(T S n ) is twice a generator of H n (S n . y) ∈ S n × S n − A to the vector from x to the point of plane at x . E ) . Let A ⊂ S n × S n consist of the antipodal pairs (x. so by the naturality property in (a) it holds for all bundles. c ∈ H 2n (D(E). namely s(x1 . Let E ⊂ E = T S n be the complement of the zero section. However. S n × S n − D) ≈ H ∗ (S n × S n . (d) When we deﬁned the Euler class we observed that it could also be described as the element of H n (B. S(E). we get the relation e(E1 ⊕ E2 ) = e(E1 ) e(E2 ) . When n is odd. E ) ≈ H ∗ (S n × S n . D). If n is odd. (c) We showed this for the universal bundle in the calculation of the cohomology of Grassmannians a couple pages back. T S 1 is just the product S 1 × R . and hence Consider the tangent bundle T S n to S n . Since i∗ is injective. Excision then gives the ﬁrst of the following isomorphisms: H ∗ (E. x)} corresponds under f to the zero section of E . De(x. Z) corresponding to c c c = −c c . while if n = 1 . Z) . topic to the identity. The diagonal D = {(x. A) ≈ H ∗ (S n × S n . ··· . Z) j∗ i → H n (D(E). or by part (e) since there is a nonzero tangent vector ﬁeld to S n when n is odd. E ) . Under the Thom isomorphism the Euler class e(T S n ) corresponds to the square of a Thom class c ∈ H n (E. Rm+n − {0}) which takes generator cross generator to generator by the calculations in Example 3. Rm+n − Rm )z→H m+n (Rm+n . Rm+n − Rn )× H n (Rm+n . This bundle is orientable since its base S n is simply-connected if n > 1 . Then in the exact sequence the map i∗ is injective since the composition D(E) e(E) = 0 from the deﬁnition of the Euler class. ﬁne a homeomorphism f : S n × S n − A→E sending a pair x intersection of the line through −x and y with the tangent y x . Passing from relative to absolute cohomology and restricting to the zero section. S(E). ··· . xn+1 ) = (−x2 . x1 . the basic commutativity relation for cup products gives (e) A nowhere-zero section of E gives rise to a section s : B →S(E) by normalizing H n (D(E). when n is even e(T S n ) is nonzero: Proposition 3. Z) has no elements of order two. Z) → H n (S(E).

are realizable. which equals −2αβ if n is even. connected. n . where χ (M) is the Euler characteristic of M . Thus α − β corresponds to the Thom class of H 2n (S n × S n . see for example [Milnor-Stasheff]. and 2n by the Thom isomorphism. D) ≈ H 2n (S n × S n ) . or octonionic Hopf bundle. Both α and β restrict to same homeomorphism D ≈ S n . HP2 . or OP2 . so all even multiples of a generator of H n (S n ) has integral cohomology consisting of Z ’s in dimensions 0 . 4 . This agrees with what we have just seen in the case M = S n . is the associated projective plane CP2 . quaternionic. and the solution of the Hopf invariant only if n = 2 . β which are pullbacks of a generthe same generator of H n (D) since the two projections S n × S n →S n restrict to the ator of H n (S n ) under the two projections S n × S n →S n . so e(E) c is k times a generator of H 2n (T (E)) . we can realize any multiple of a generator by taking pullbacks as before. y) sequence (x. consider the Thom space T (E) . D) . −y) of S n × S n interchanging D and A . To investigate odd multiples. and whose Thom space. From the long exact sequence of the pair (S n × S n . D)→H n (S n × S n )→H n (D)→0 The middle group H n (S n × S n ) has generators α . This is twice a generator the unit tangent bundle of a closed. . D) we extract a short exact 0→H n (S n × S n .Euler and Pontryagin Classes Section 3. This is precisely the context of the Hopf invariant.2 93 The second isomorphism holds since S n × S n − D deformation retracts onto A by sliding a point y ≠ ±x along the great circle through x and y to −x . and the third comes from the homeomorphism (x. which also says that the Thom class c is a generator of H n (T (E)) . orientable smooth manifold M n is | χ (M)| times a generator of H n (M) . This c . we realize 2d times a generator. namely the vector bundle whose unit sphere bundle is the complex. and is the reason for the name ‘Euler class. We know that the Euler class corresponds under the Thom isomorphism to c is k times a generator of H (S ) iff c one problem in Chapter 2 shows that c n n One might ask which elements of H n (S n ) can occur as Euler classes of vector tangent bundle T S n by a map S n →S n of degree d .’ bundles E →S n in the nontrivial case that n is even. or 8 . In these three cases there is a bundle E →S n for which c c can be an odd multiple of a generator c is a generator of H 2n (T (E)) . the mapping cone of the sphere bundle. It is a fairly elementary theorem in differential topology that the Euler class of and (α − β)2 = −αβ − βα . Since we can realize a generator of H n (S n ) as an Euler class in these three cases. the kernel of the restriction map H n (S n × S n )→H n (D) . so α − β generates H n (S n × S n . If we form the pullback of the by part (a) of the preceding proposition.

so does Gn . v2n . First let us remark that since Gn has a CW structure with ﬁnitely many Gn and Gn are ﬁnitely generated. and hence give nothing new. tions differ by a sign (−1)n(2n−1) = (−1)n . v2n . c2i (E C ) reduces mod 2 to w4i (E ⊕ E) . The reason for restricting attention to the even Chern classes c2i (E C ) is that the odd classes c2i+1 (E C ) turn out to be expressible in terms of Stiefel-Whitney classes. ··· . Here is how Pontryagin classes are related to Stiefel-Whitney and Euler classes: of E becomes a ﬁber Cn of E C . pn (E) = e(E)2 . The sign (−1)i is introduced in order to avoid a sign in the formula in (b) of the next proposition. (b) First we need to determine the relationship between the two orientations of E C ≈ E ⊕ E . Z2 ) . The exercises at the end of the section give an explicit formula. the other coming from the orientation of E ⊕ E determined by an orientation of E . which equals w2i (E)2 since w(E ⊕ E) = w(E)2 and squaring is an additive homomorphism mod 2 . its complexiﬁcation is the complex vector bundle E C →B obtained from the The easiest deﬁnition of the Pontryagin classes pi (E) ∈ H 4i (B. Note that statement (b) is independent of the choice of orientation of E since the (b) For an orientable real 2n dimensional vector bundle with Euler class e(E) ∈ Euler class is squared.4. For the universal bundles En →Gn and En →Gn canonical orientation of En . Proof: (a) By Proposition 3. Thus we have pn (E) = (−1)n c2n (E C ) = Pontryagin classes can be used to describe the cohomology of Gn and Gn with Z coefﬁcients. ··· . H 2n (B. If v1 . pi (E) maps to w2i (E)2 under the coefﬁcient homomorphism H 4i (B. (a) For a real vector bundle E →B . Z)→H 4i (B. the Euler class being deﬁned via the . ··· . Z) . iv2n . x) . so the two orienta(−1)n e(E C ) = e(E ⊕ E) = e(E)2 . Z) associated to real vector bundle E ⊕ E by deﬁning scalar multiplication by the complex number i in each ﬁber Rn ⊕ Rn via the familiar rule i(x. iv2n . cells in each dimension. For a real vector bundle E →B . v2n is a basis for a ﬁber of E agreeing with the given orientation. iv1 . The Pontryagin class pi (E) is then deﬁned to be Proposition 3. iv1 . Thus each ﬁber Rn (−1)i c2i (E C ) ∈ H 4i (B. and e = e(En ) .15. while E ⊕ E is oriented by v1 . To make these two orderings agree requires (2n − 1) + (2n − 2) + ··· + 1 = 2n(2n − 1)/2 = n(2n − 1) transpositions. one coming from the canonical orientation of the complex bundle E C . pi = pi (En ) . ··· . y) = (−y. then E C is oriented by the ordered basis v1 . hence the homology and cohomology groups of let pi = pi (En ) . Z) .94 Chapter 3 Characteristic Classes Pontryagin Classes a real vector bundle E →B is in terms of Chern classes.

Proof: We shall work on (b) ﬁrst since for orientable bundles there is a Gysin sequence with Z coefﬁcients. As in the calculation of H ∗ (Gn . ··· . ··· . e] for n = 2k .2 95 Theorem 3. Z)/torsion is the polynomial ring Z[p1 . and H ∗ (Gn . pk−1 ] . ··· . which we shall compute in the course of proving the theorem. (a) All torsion in H ∗ (Gn . pk−1 . The assertion to be proved is that H ∗ (Gn . pk ] and is torsionfree. Z2 ) model. Z) therefore maps injectively to H ∗ (Gn . ··· . Since we are dealing with ﬁnitely generated integer homology groups. Z2 ) via the Gysin sequence.E of [AT]. Z) consists of elements of order 2 . The Gysin sequence now splits into short exact sequences 0 → H i+n (Gn ) → H i+n (Gn−1 ) → H i+1 (Gn ) → 0 η Thus η injects H ∗ (Gn ) as a subring of H ∗ (Gn−1 ) ≈ R[p1 . Z) . The theorem implies that Stiefel-Whitney and Pontryagin classes determine all characteristic classes for unoriented real vector bundles. so we . Z) consists of elements of order 2 . Z2 )→H ∗ (Gn . with e2 = pk in the latter case. e] . pk−1 . ··· . so we get a projection → H i (Gn ) → H i+n (Gn ) → H i+n (Gn−1 ) → H i+1 (Gn ) → ··· where η takes pi (En ) to pi (En−1 ) . pk ] for n = 2k or 2k + 1 . elements of order a power of 2 . changing from Z coefﬁcients to R coefﬁcients eliminates any 2 torsion in the homology. The of v is then naturally oriented. The same remarks apply to H ∗ (Gn . then by induction H ∗ (Gn−1 ) ≈ R[p1 . ··· . As a ﬁrst step we compute H ∗ (Gn . pk ] for n = 2k + 1 and R[p1 . where implies that H ∗ (Gn . e] for n = 2k . Then it will remain only to show that all 2 torsion in H ∗ (Gn . Z2 ) . that is.Euler and Pontryagin Classes Section 3. The proof in this case then follows the H ∗ (Gn . then e is zero in H n (Gn . R) is R[p1 . with image the image of the Bockstein β : H ∗ (Gn . (b) All torsion in H ∗ (Gn . where e now means e(En−1 ) . pk−1 . If n = 2k . This stated in the theorem. the rational numbers with denominator a power of 2 . ··· . Z) consists of elements of order 2 . R) since with Z coefﬁcients it has order 2 . R) where R = Z[1/2 ] ⊂ Q. ) where is . for the deﬁnition and basic properties of Bockstein homomorphisms see §3. and Z summands of homology become R summands. The subring Im η contains R[p1 . Z2 ) . Z) has no odd-order torsion and that H ∗ (Gn . Z)/torsion is Z[p1 . while for oriented bundles the only additional class needed is the Euler class. so η is surjective and the sequence splits into short exact sequences. The Gysin sequence with coefﬁcients in R has the form ··· η π → S(En ) → Gn . Z)/torsion is as S(En ) is the space of pairs (v.16. pk ] for n = 2k + 1 and Z[p1 . ··· . If n = 2k + 1 . and H ∗ (Gn . consider the sphere bundle S n−1 an oriented n dimensional linear subspace of R∞ and v is a unit vector in orthogonal complement v ⊥ ⊂ e p : S(En )→Gn−1 . The torsion subgroup of H ∗ (Gn .

R)→H ∗ (Gn . ··· . observe that the image of π ∗ is invariant under the map τ ∗ induced by the implies τ ∗ π ∗ = π ∗ . For the 2 sheeted covering π : Gn →Gn consider the transfer hoof π∗ is that the composition π∗ π ∗ : H ∗ (Gn . The induction can start with the case n = 1 . p[n/2] ] . ··· . ﬁnishing the proof that H ∗ (Gn . w2k+1 ] is also the polynomial ring Z2 [w1 . we get rj = rj . with G1 ≈ S ∞ . Z) . R) = R[p1 . takes e to −e . pk−1 . H ∗ (Gn . Before studying the remaining 2 torsion question let us extend what we have just n = 2k or 2k + 1 . ··· . i to compute βwk we can compute βσk . e} . pk ] . that is. As observed in §3.17. Z) has order 2 we use the Bockstein homomorphism β associated to the short exact sequence of coefﬁcient groups 0→Z2 →Z4 →Z2 →0 . The subring of H ∗ (Gn . ··· . Z) and H ∗ (Gn . R) deﬁned in §3. Proof: βw2i+1 = w1 w2i+1 and βw2i = w2i+1 + w1 w2i . R)→H ∗ (Gn . βw2 . since π τ = π is even. the mod 2 reduction of the Euler class. ··· . pk ] in dimension j and rj the rank of H j (Gn ) . Using the derivation property β(x βx we see that βσk is the sum of all products αi1 ··· α2 ··· αik for i1 < ··· < ik and j = 1. w2k . pk ] . ··· . Now for the calculation of Ker β/ Im β . First consider the case of G2k+1 . ··· . pk ] with basis {1. Z) and H ∗ (Gn . the number of sheets in the covering space. ··· . Lemma 3. we can view wk as the k th elementary symmetric polynomial σk in the polynomial algebra Z2 [α1 . This is an isomorphism π ∗ (pi ) = pi . The main feature plication by 2 . Since R[p1 . By naturality it sufﬁces to prove this for the universal bundle En →Gn with wi = wi (En ) . pk−1 . αn ] ≈ H ∗ ((RP∞ )n . ··· . R)→H ∗ (Gn . pk ] by comparing ranks of these R modules in each dimension. pk ] since then exactly R[p1 .1. the class e = e(En−1 ) having dimension 2k . By general properties of Bockstein homomorphisms proved in §3. and momorphism π∗ : H ∗ (Gn . The map τ reverses orientation in each ﬁber of En →Gn . ··· . ··· . So when n is odd. ··· . the rank of H ∗ (Gn−1 ) ≈ R[p1 . so τ ∗ for R = Z[1/2 ]. e] in dimension j is rj + rj−2k . ··· . pk−1 . to show that for R = Z[1/2 ].G of [AT]. w2 . Z2 ) . The ring Z2 [w1 . Thus y) = y +x βy and the fact that βαi = α2 .E of [AT] this will ﬁnish the proof. On the other hand. so π ∗ is injective.96 Chapter 3 Characteristic Classes can show it equals R[p1 . and for G2k . completing the induction step. Since rm ≥ rm for each m . ij multiplying σ1 σk out. Let rj be the rank of R[p1 . ··· . polynomials 2 in the w2i ’s for i < k together with w2k . k . R) is R[p1 . e] is a free module over R[p1 . the exact sequence above so H ∗ (Gn ) = R[p1 . When n deck transformation τ : Gn →Gn interchanging sheets of the covering. π ∗ is an isomorphism and we are done. R) is multidone to H ∗ (Gn . ··· . On the other hand. where says this rank also equals rj + rj−2k . R) ≈ R[p1 . βw2k ] since . pk ] . 2 polynomials in the classes w2i in the case of Gn and G2k+1 . one obtains βσk + (k + 1)σk+1 . e] invariant under τ ∗ is To show that all 2 torsion in H ∗ (Gn . The goal is to show that Ker β/ Im β consists exactly of the mod 2 reductions of nontorsion classes in H ∗ (Gn . The image of π ∗ contains R[p1 . Z) .

Hence Ker β/ Im β = Z2 [w2i ] . all of which are also in Im β . We then have the formula β(w1 w2k ) = m m m m w1 +1 w2k + mw1 +1 w2k = ( + m)w1 +1 w2k . For Z2 [w2i . Show that c2i+1 (E C ) = β(w2i (E)w2i+1 (E)) . and hence the β homology of Z2 [w1 ] is trivial in positive dimensions. w2k ] . βw2i ] ’s for m i < k and Z2 [w1 . so further details will be left as an exercise. > 0 . ··· .Exercises the substitution w1 w1 . plus Z2 [w2k ] when n = 2k . Exercises 1. . For w1 w2k to be in Ker β we must have + m even. βw2i ] . even. viewing Z2 [w1 . wn ] with respect to β is the polynomial ring in 2 the classes w2i . w2k+1 ] splits as the tensor product of the polynomial rings Z2 [w1 ] and Z2 [w2i . and to be in Im β we must have in addition 2 Z2 [w2k ] . 2. hence βw2i = w2i+1 and βw2i+1 = 0 . wn ] up as the tensor product of the chain complexes Z2 [w2i . w2i w2i . Show that every class in H 2k (CP∞ ) can be realized as the Euler class of some vector bundle over CP∞ that is a sum of complex line bundles. βw2i ] we have β(w2i (βw2i )m ) = w2i (βw2i )m+1 . w2k+1 ] with respect to u the boundary map β is therefore the tensor product of the homologies of the chain complexes Z2 [w1 ] and Z2 [w2i . ··· . with β(w2k ) = w1 w2k . Z2 [w1 . the mod 2 reductions of the Pontryagin classes. For an oriented (2k + 1) dimensional vector bundle E show that e(E) = βw2k (E) . The calculations are quite similar to those we have just done. ··· . we might remark that this had to be true since the calculation is −1 the same as for RP∞ . being its own inverse in fact. Moreover. ··· . Thus Z2 [w1 . so Ker β is generated by the monomials w2i (βw2i )m with 2 with m > 0 are in Im β . so Ker β is generated by the even powers of w1 . According to the algebraic K¨nneth theorem. w2i+1 97 w2i+1 + w1 w2i = βw2i for i > 0 is invertible. So Ker β/ Im β = Thus the homology of Z2 [w1 . 3. this tensor product is a tensor product of chain complexes. The case of Gn is simpler since w1 = 0 . the homology of Z2 [w1 . w2i+1 ] . ··· . ··· . each of which is invariant under β . For Z2 [w1 ] we have β(w1 ) = w1 +1 . Then we can break Z2 [w2 . w2k ] is the tensor product of the Z2 [w2i . and such monomials For n = 2k . βw2i ] . w2k+1 ] as a chain complex with boundary map β . By general properties of Bocksteins this ﬁnishes the proof of part (a) of the theorem.

Here is a table for small values of n : −−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− −−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− an 24 240 504 480 264 65520 24 16320 28728 13200 552 n −−− −−− −−− −−− 1 2 3 4 5 6 7 8 9 10 11 In spite of appearances. where a vertical chain of k connected dots above the number 4n − 1 means that the highest power of p dividing an is p k . An odd prime p divides an iff n is a multiple of (p − 1)/2 .Homotopy groups of spheres are notoriously difﬁcult to compute. 3. according to Proposition 1. The values of an have been computed in terms of Bernoulli numbers.14 for O(n) and the Freudenthal suspension theorem [AT] for S n . but this becomes clear only when one looks at the prime factorization of an . The real form of Bott periodicity proved in Chapter 2 implies that the domain of the stable s J homomorphism πi (O)→πi is nonzero only for i = 4n − 1 when πi (O) is Z and for i = 8n and 8n + 1 when πi (O) is Z2 . and in this case the highest power of p dividing an is p dividing n . Here are the rules for computing an : 1. The highest power of 2 dividing an is 2 dividing n . When i = 4n − 1 the image of J is a ﬁnite cyclic group of some order an s since πi is a ﬁnite group for i > 0 by a theorem of Serre proved in [SSAT]. The ﬁrst three cases p = 2. but some partial information can be gleaned from certain naturally deﬁned homomorphisms J : πi (O(n))→πn+i (S n ) The goal of this chapter is to determine these J homomorphisms in the stable dimension range n >> i where both domain and range are independent of n . +1 +3 where 2 is the highest power of 2 where p is the highest power of p . 2. In the latter two cases we will show that J is injective. 5 are shown in the following diagram. there is great regularity in this sequence.

as shown in the ﬁgure below.1 99 −−− −− 3 7 11 15 19 23 27 31 35 39 43 47 51 55 59 63 67 71 75 79 83 87 91 95 99 103 107 p = 2: −−− −− 3 7 11 15 19 23 27 31 35 39 43 47 51 55 59 63 67 71 75 79 83 87 91 95 99 103 107 p = 3: −−− −− 39 79 119 159 199 239 279 319 359 399 439 479 519 559 599 639 679 719 759 799 839 879 919 959 999 p = 5: 4. v) ∂D . Taking a similar view of Jg . we will Jf (D i+1 × S n−1 ) = ∂D n . with fx the identity when x is the basepoint of S i . y) = fx (y) for (x. let Jf (x. Then we obtain a homotopy from Jf + Jg to J(f + g) by moving the two S i × D n ’s together until they coincide. so . Proof: n J is a homomorphism. and we will take care of the cases in which the domaim of the J–homomorphism is Z2 . g implies Jf Jg . The simplest deﬁnition of the J homomorphism goes as follows. We will tacitly exclude the trivial case n After starting this section with the deﬁnition of the J–homomorphism.Lower Bounds on Im J Section 4. Clearly f i = 0.1.1 Lower Bounds for Im J use a homomorphism K(X)→H ∗ (X. the sum Jf + Jg is obtained by juxtaposing these two maps on either side of a hyperplane. the basepoint of D n /∂D n . We may assume fx is the identity for x in the right half of S i and gx is the identity for x in the left half of S i . Proposition 4. An element [f ] ∈ πi (O(n)) is represented by a family of isometries fx ∈ O(n) . Q) known as the Chern character to show that an /2 is a lower bound on the order of the image of J in dimension 4n − 1 . y) ∈ S i × D n and let so we have a map J : πi (O(n))→πn+i (S ) . Writing S n+i as ∂(D i+1 × D n ) = S i × D n ∪ D i+1 × S n−1 and S n as D n /∂D n . − − → we would expect the J–homomorphism deﬁned above − − → We know that πi (O(n)) and πn+i (S n ) are independent of n for n > i + 1 . Then using real K–theory this bound will be improved to an . We can view Jf as a map I n+i →S n = D n /∂D n which on S i × D n ⊂ I n+1 is fx (v) and which sends the complement of S i × D n to the basepoint given by (x. x ∈ S i .

··· . Q) will be zero for sufﬁciently large n . via the splitting principle. This will give a lower bound for the order of the image of the real J–homomorphism s πi (O)→πi when i = 4n − 1 . to do bundle L→X to be ch(L) = ec1 (L) = 1 + c1 (L) + c1 (L)2 /2! + ··· ∈ H ∗ (X. it will lie not in the direct sum H ∗ (X. Q) but rather in the direct product. via commutativity of the diagram at the right. the Newton polynomials sk k k satisfy t1 + ··· + tn = sk (σ1 . Q) of the groups H n (X. this means that the preceding displayed formula can be rewritten ch(E) = dim E + k>0 sk (c1 (E).100 Chapter 4 The J–Homomorphism s to induce a stable J–homomorphism J : πi (O)→πi . Our goal is to deﬁne via K–theory a homo- the natural inclusions U(n) ⊂ O(2n) which give an inclusion U ⊂ O . In order to make this work one must use cohomology with rational coefﬁcients. so we take this as our general deﬁnition of ch(E) . The situation might have been simpler if it had been possible to use integer coefﬁcients instead. H n (X. but on the other hand. As we saw in §2. in the case of line bundles. For a direct sum of line bundles E ≈ L1 ⊕ ··· ⊕ Ln we would then want to have ch(E) = i ch(Li ) = i k k eti = n + (t1 + ··· + tn ) + ··· + (t1 + ··· + tn )/k! + ··· where ti = c1 (Li ) . If the sum 1+c1 (L)+c1 (L)2 /2!+··· has inﬁnitely many nonzero terms. − − − − − → − − → The Chern Character The total Chern class c = 1 + c1 + c2 + ··· takes direct sums to cup products. πi (O (n)) −→ πi (O (n +1)) − − − J J γ S n +1 n π n +i ( S ) − π n +i +1( S ) − → Composing the stable J–homomorphism with the map πi (U )→πi (O) induced by s complex J–homomorphism JC : πi (U )→πi . σk ) . . Since σj = cj (E) . however. the sum of all products of j distinct ti ’s. so this distinction will not matter. The total Chern class c(E) is then (1 + t1 ) ··· (1 + tn ) = 1 + σ1 + ···+σn . we get the stable s morphism e : πi →Q/Z for i odd and compute the composition eJC : πi (U )→Q/Z . However. where σj = cj (E) is the j th elementary symmetric polynomial in the ti ’s. the fact that one has rational coefﬁcients instead of integers makes it possible to deﬁne a homomorphism e : π2m−1 (S 2n )→Q/Z which gives some very interesting information about the difﬁcult subject of homotopy groups of spheres. so that ⊗ ch(L1 ⊗ L2 ) = ec1 (L1 L2 ) = ec1 (L1 )+c1 (L2 ) = ec1 (L1 ) ec1 (L2 ) = ch(L1 )ch(L2 ) . and the idea of the Chern character is to form an algebraic combination of Chern classes which takes direct sums to sums and tensor products to cup products. ··· . ck (E))/k! The right side of this equation is deﬁned for arbitrary vector bundles E .3. thus giving a natural ring homomorphism from K–theory to cohomology. Q) . In order to deﬁne the Chern character it sufﬁces. We leave it as an exercise for the reader to verify that this is the case. The idea is to deﬁne the Chern character ch(L) for a line the examples we shall be considering.

a K( X ) − − K( S 2X ) −≈ → − − H ( X.4.Lower Bounds on Im J Section 4. Proof: group H 2n (S 2n . For the product formula. By another application of the splitting principle to the pullback of E2 over F (E1 ) . Even when H ∗ (X. starting with the trivial case n = 0 .16 of [AT] the lower map is an isomorphism and restricts to an isomorphism of the Z coefﬁcient subgroups. E2 )→X pulling both E1 and E2 back tive. Q ) − → ∗ generator of H 2 (S 2 . ch : K(S 2n )→H 2n (S 2n . As a ﬁrst calculation of the Chern character. E2 ). Q)→H ∗ (F (E1 . the result now follows by induction on n . From Theorem 3. Z) is torsionfree.k i. Q ) − H (S X.1 101 Proposition 4. ··· . Q) . so ch(E) = dim E +sn (c1 . Z) occurs as a Chern class cn (E) iff it is divisible Proof: For vector bundles E →S 2n we have c1 (E) = ··· = cn−1 (E) = 0 . So to prove the proposition it sufﬁces to verify the two formulas when E1 and E2 are sums of line bundles. By this splitting principle we can pull E1 back to a sum of line bundles over a space F (E1 ) . Q) is injective with image equal to the subch(H − 1) we have the commutative dia- Since ch(x ⊗ (H − 1)) = ch(x) gram shown at the right. so that H ∗ (X. by the discussion preced- ing the deﬁnition of ch .k (L1j ⊗ L2k ) = ch(L1j ⊗ L2k ) = j. An interesting by-product of this is: − − − − − → − − − − − → ch ch ≈ ∗ 2 Corollary 4. For example. Z) .j to sums of line bundles.2. Proof: ch(E1 ⊕ E2 ) = ch(E1 ) + ch(E2 ) and ch(E1 ⊗ E2 ) = ch(E1 )ch(E2 ) .3.k ch(L1j )ch(L2k ) = ch(E1 )ch(E2 ) . Q) injec- ec1 (Lij ) = ch(E1 ) + ch(E2 ) . and the lower map is cross product with ch(H − 1) = ch(H) − ch(1) = 1 + c1 (H) − 1 = c1 (H) . where the upper map is external tensor product with H − 1 . with the induced map H ∗ (X. 2 . Q) . say Ei = ⊕ j Lij for i = 1.j Lij ) = j. By naturality there is also a reduced form point. Z) ⊂ H 2n (S 2n . which is an isomorphism by Bott periodicity.3 works with any coefﬁcients in the case of complex vector bundles. The proof of the splitting principle for ordinary cohomology in Proposition 2. if .3. namely. sn = σ1 sn−1 − σ2 sn−2 + ··· + (−1)n−2 σn−1 s1 + (−1)n−1 nσn . by (n − 1)! . homomorphism ch : K(X)→H ∗ (X. Z) . cn )/n! = dim E ±ncn (E)/n! by the recursion relation for sn derived in §2. the Chern character automatically extends to a ring ch : K(X)→H ∗ (X. The sum formula holds since ch(E1 ⊕ E2 ) = ch( ⊕ i. A class in H 2n (S 2n . Q) . Taking X = S 2n . it is not always true that the image of ch is contained in H ∗ (X. in particular for Q coefﬁcients. we have a map F (E1 . Q) since these reduced rings are the kernels of restriction to a Proposition 4. we have: In view of this proposition. Z) is a subring of H ∗ (X. ch(E1 ⊗ E2 ) = ch ⊕ j.

ch K 0 ( SX ) −− H −→ − − even ( SX. the homo- cone Cf obtained by attaching a cell e2m to S 2n by f . Since these are ﬁnitely generated abelian groups. Q) . up to homotopy equivalence. with the same number of cells. Proof: The map K ∗ (X) ⊗ Q→H ∗ (X. The Chern character can be used to show that for ﬁnite cell complexes X . The quotient Cf /S 2n is S 2m so we have a commutative diagram of short exact sequences 0 −− K(S −− −→ 0 − H (S − → ∗ 2m ) −−− K(Cf ) − − K(S 2n ) −− 0 −−→ −− −→ −− −→ ch ∗ ch ∗ ch 2m − → − → . and it is easy to extend this formally to odd dimensions by the commutative diagram at the right. a 0 cell. We proceed by induction on the number of cells of X . Z) . For the induction step. In a similar way there ≈ = − − − − − → − − − − − → − − − − − → . Q ) −→ odd Proposition 4. The result is triv- ially true when there is a single cell. by the preceding proposition. Q) induced by the Chern character is an isomorphism for all ﬁnite cell complexes X . to obtain the result for X itself we may replace X by S 2 X since the Chern character commutes with double suspension. all except the map K ∗ (SX) ⊗ Q→H ∗ (SX. Q ) ch 1 K ( X ) − − H ( X. so that X is a sphere.Q ) − 0 There are elements α. and it is also true when there are two cells. Both SA and S 2 A are homotopy equivalent to cell complexes with the same number of cells as A . The spaces X/A and SX/SA are spheres. this will follow if we can show that K ∗ (X) ⊗ Q and H ∗ (X. Q ) − H (Cf . Thus far we have deﬁned the Chern character K 0 (X)→H even (X. respectively. Finally.102 Chapter 4 The J–Homomorphism L ∈ K(CPn ) is the canonical line bundle. Consider the ﬁve-term sequence X/A→SA→SX →SX/SA→S 2 A . Q ) − H (S 2n − → . Q) then gives a commutative diagram of ﬁve-term exact sequences since tensoring with Q preserves exactness. Applying the rationalized Chern charac- ter K ∗ (−) ⊗ Q→H ∗ (−. hence c k generates H 2k (CPn . then ch(L) = 1+c +c 2 /2+···+c n /n! where c = c1 (L) generates H 2 (CPn . and a double suspension is in particular a single suspension. by collapsing the suspension or double suspension of a 0 cell.5. Q) are isomorphic. Q) . β ∈ K(Cf ) mapping from and to the standard generators (H − 1) ∗ ··· ∗ (H − 1) of K(S 2m ) and K(S 2n ) . so by the ﬁve-lemma this map is an isomorphism as well. as we have seen. Z) for k ≤ n . let X be obtained from a subcomplex A by attaching a cell. Z) lie in their torsion subgroups. For a map f : S 2m−1 →S 2n we have the mapping Now let us deﬁne the main object we will be studying in this section. the only possible differences between the groups K ∗ (X) and H ∗ (X. The e Invariant morphism e : π2m−1 (S 2n )→Q/Z . Thus by induction four of the ﬁve maps between the two exact sequences are isomorphisms.

then x/(ex − 1) = βi x i /i! e(JC f ) = ±βk /k where βk is deﬁned via the power series i s Hence the image of J in π2k−1 has order divisible by the denominator of βk /k . so r is well-deﬁned in the additive group Q/Z . By restriction ch(βf +g ) = bf +g + (rf + rg )af +g . deﬁne e(f ) to be this element r ∈ Q/Z . Thom spaces arise in the present context through the following: .1 103 are elements a. After proving the theorem we will show how to compute the denominator of βk /k . Q ) − H (Cf + g .g + rf af + rg bg .g into a pair of 2m cells.3. When B is compact this can also be regarded as the one-point compactiﬁcation of E .g we obtain ch(βf . ag . so − − − − − → − − − − − → ch q∗ ch −− −− −→ e Commutativity follows from the fact that CS 2 f = S 2 Cf and ch commutes with the double suspension. Z) and H 2n (S 2n . and we well-deﬁned map e : π2n−1 (S 2m )→Q/Z . and bf +g . g .g collapsing a sphere S 2m−1 that separates the 2m cell of Cf . There is a commutative diagram involving the double suspension: S 2m 2m + 2 ) π2 n 1 ( S ) − → π 2n +1 ( S − 2 Let Cf .Lower Bounds on Im J Section 4. The numbers βk are known in number theory as Bernoulli numbers. Proof: e is a homomorphism. Z) . Cf . There is a quotient map q : Cf +g →Cf . using Proposition 4. as we saw in the proof of Proposition 4. The elements β and b are not uniquely determined but can be varied by adding any integer multiples of α and a .g contains both Cf and Cg .3. af +g . The effect of such a variation on the formula ch(β) = b + r a is to change r by an integer. Q) mapping from and to generators of H 2m (S 2m .g . we have a Proposition 4. After perhaps replacing a and b by their negatives we may assume that ch(α) = a and ch(β) = b + r a for some r ∈ Q . bf . In the have generators αf and αg mapping to αf +g and βf . If the map f : S 2k−1 →U (n) represents a generator of π2k−1 (U ) . From the commutativity s of the diagram there is induced a stable e invariant e : π2k−1 →Q/Z for each k .7. g) − − K(Cf + g ) −− −→ H (Cf. Q ) − → ∗ ∗ q∗ to the subspaces Cf and Cg of Cf . so upper row of the commutative diagram at the right we K(Cf.g be obtained from S 2n by attaching two 2m cells by f and g . Q/ Z − − −e→ −− Theorem 4. Since f g implies Cf Cg .g ) = bf . and similarly in the second row with generators af .g mapping to βf +g . Thom Spaces For a vector bundle E →B with unit disk bundle D(E) and unit sphere bundle S(E) the quotient space D(E)/S(E) is called the Thom space T (E) . b ∈ H ∗ (Cf .6.

since on ∂D 2k × D 2n it is given by (x. K ∗ (P (E)) is the free K ∗ (X) module with basis the restrictions of 1.25. v) D 2n with characteristic map the quotient map D 2k × D 2n →D(Ef )→T (Ef ) . In order to compute eJC (f ) we need to compute ch(β) where β ∈ K(CJf ) = 2n D0 of D(Ef ) . The S 2n here is D0 /∂D0 for a ﬁber restricts to a generator of K(S n ) for each sphere S n coming from a ﬁber of E is called a Thom class for the bundle. known as the Thom isomorphism. Restricting to P (E) ⊂ P (E ⊕ 1). The deﬁning relation for H ∗ (P (E)) as H ∗ (X) module has the form i n−i i (−1) ci (E)x = 0 where x = . Restricting to the unit disk bundle D(Ef ) . In the quotient T (Ef ) = D(Ef )/S(Ef ) we then have the sphere 2n 2n S 2n = D0 /∂D0 . In particular we have an isomorphism K ∗ (T (E)) ≈ K ∗ (X) . Ef is the union of two copies of D 2k × Cn with the subspaces ∂D 2k × Cn identiﬁed via (x. regarded now as an element of K(P (E ⊕ 1)) . Ln−1 to P (E) . fx (v)) . Similar constructions can be made for ordinary cohomology. CJf is the Thom space of the bundle Ef →S 2k determined by the clutch- Proof: By deﬁnition. so Observe that K ∗ (T (E)) is a free K ∗ (X) module with the single basis element U since K ∗ (P (E ⊕ 1)) is a free K ∗ (X) module with basis 1. U . For a general complex vector bundle E →X a class in K(T (E)) that 2n 2n K(T (Ef )) restricts to a generator of K(S 2n ) .104 Chapter 4 The J–Homomorphism ing function f : S 2k−1 →U(n) . Ln−1 . E →X with X compact Hausdorff. ··· .24 we know that K ∗ (P (E ⊕ 1)) is the Let us show how to ﬁnd a Thom class U ∈ K(T (E)) for a complex vector bundle free K ∗ (X) module with basis 1. and T (Ef ) is obtained from this S 2n by attaching a cell e2k+2n map of this cell is precisely Jf . namely the polynomial i i n−i i (−1) λ (E)L in Proposition 2. ··· . L. ··· . Collapsing the second copy of D 2k × Cn to Cn via projection produces the same vector bundle Ef . From Example 2. where the subscript 0 labels this particular disk ﬁber of D(Ef ) . We may view T (E) as the quotient P (E ⊕ 1)/P (E) since in each ﬁber Cn of E we obtain P (Cn ⊕ C) = CPn from P (Cn ) = CPn−1 by attaching the 2n cell Cn × {1} . L.8. so Ef can also be obtained from D 2k × Cn Cn by the identiﬁcation (x. v) ∼ (x. The attaching fx (v) ∈ /∂D 2n and all of D 2k × ∂D 2n maps to the point ∂D 2n /∂D 2n . where L is the canonical line bundle over P (E ⊕ 1) . so the quotient P (Cn ⊕ C)/P (Cn ) is S 2n . we have D(Ef ) expressed as a quotient of D 2k × D 2n 2n D0 by the same identiﬁcation relation. v) ∼ fx (v) for x ∈ ∂D 2k . L. Lemma 4. which is the part of T (E) coming from this ﬁber Cn . i i n−i i (−1) λ (E)L The class U ∈ K(T (E)) mapping to is the desired Thom class i i n−i i (−1) λ (E)L the kernel of K(CPn )→K(CPn−1 ) is generated by the restriction of to a ﬁber. So we have a short exact sequence 0 → K ∗ (T (E)) → K ∗ (P (E ⊕ 1)) → K ∗ (P (E)) → 0 ρ and Ker ρ must be generated as a K ∗ (X) module by some polynomial of the form Ln + an−1 Ln−1 + ··· + a0 1 with coefﬁcients ai ∈ K ∗ (X) . Ln . since when we restrict over a point of X the preceding considerations still apply.

For future reference we note two facts: (1) x = c1 (L) ∈ H ∗ (P (E ⊕ 1)) . or what amounts to the same thing. i n+1−i i (−1) ci (E ⊕ 1)x For convenience we shall also identify U with i i n−i i (−1) λ (E)L We are omitting notation for pullbacks. since the deﬁning relation for c1 (L) is x(L)−c1 (L) = 0 and P (L) = P (E ⊕ 1) . The Thom class u = then factors as a i (x − xi ) where xi = c1 (Li ) . By the splitting principle we can pull this bundle E back further to a sum product i Li of line bundles over a space F (E) and work in the cohomology i n−i i (−1) ci (E)x th and K–theory of F (E) . so in particular we are viewing E as already pulled back over P (E ⊕ 1) . i (−1) ci (E)x with x = x(E ⊕ 1) now. ··· . There is a power series for log[(ey − 1)/y] j αj y j /j! since the function (ey − 1)/y has a nonzero value at y = 0 . x(L) . for the K–theory Thom class U we have U = i i n−i i (−1) λ (E)L i (L = Ln λt (E) = Ln − Li ) = i λt (Li ) = Ln i (1 + Li t) for t = −L−1 . the result of the preceding calculation can be written as Φ−1 ch(U) = of the form Then we have log Φ−1 ch(U) = log = j i [(e xi − 1)/xi ] . (2) If we identify u with ∗ ∈ H ∗ (P (E ⊕ 1)) . deﬁning relation for H (P (E ⊕ 1)) is ci (E) . S(E)) ≈ H ∗ (T (E)) is given i [(e xi by cup product with the Thom class u . So ∈ H ∗ (P (E ⊕ 1)) is the image of a Thom class u ∈ H 2n (T (E)) . − 1)/xi ] = log[(exi − 1)/xi ] = αj xi /j! j i i. xn . so U = xi − Li ) . so by the splitting principle this formula is j αj chj (E) which no longer involves the splitting of the valid back in the cohomology of X .Lower Bounds on Im J Section 4. Thus we have the general formula log Φ−1 ch(U) = bundle E →X into the line bundles Li .j αj chj (E) where chj (E) is the component of ch(E) in dimension 2j . Similarly. then xu = 0 since the = 0 and ci (E ⊕ 1) = ∈ K(P (E ⊕ 1)) . since ci (E) is the i elementary symmetric function σi of x1 .1 105 x(E) ∈ H 2 (P (E)) restricts to a generator of H 2 (CPn−1 ) in each ﬁber. Therefore we have ch(U) = i ch(L i (e x − e xi ) = u i [(e xi i [(e − ex )/(xi − x)] This last expression can be simpliﬁed to u u ie xi i [(1 − e x−xi − 1)/xi ] since after writing it as )/(xi − x)] and expanding the last product out as a multivariable power series in x and the xi ’s we see that because of the factor u in front and the relation xu = 0 noted earlier in (2) all the terms containing x can be deleted. . so x(E ⊕ 1) = i n−i i (−1) ci (E)x i n−i . Since the Thom isomorphism Φ : H ∗ (X)→H ∗ (D(E). we can set x = 0 . the bundle L→E ⊕ 1 being a line bundle. Viewed as an element of H ∗ (P (E ⊕ 1)) . When dealing with products such as this it is often convenient to take logarithms. the element i n−i i (−1) ci (E)x generates the kernel of the map to H ∗ (P (E)) since the coefﬁcient of x n is 1 .

7: Let us specialize the preceding to a bundle Ef →S 2k with clutching αk ch (Ef ) in the present case since H (S 2k . (2) A prime divides the denominator of βk /k iff it divides the denominator of βk .106 Chapter 4 The J–Homomorphism function f : S 2k−1 →U(n) where the earlier dimension m is replaced now by k . Theorem 4. Determining the denominator of βk /k for even k is our next goal since this tells us the order of the image of eJC in these cases.11. Since ch(U ) = ch(β) = b + r a . powers p +1 For even k > 0 the denominator of βk /k is the product of the prime such that p − 1 divides k and p is the highest power of p dividing k . as a routine calculation shows. On the other hand. To relate αk to Bernoulli numbers βk we differentiate both sides of the equation k x x k αk x /k! = log[(e − 1)/x] = log(e − 1) − log x . we obtain r = αk . Thus we obtain αk = βk /k for k > 1 and 1 + β1 = α1 . Bernoulli Denominators The numbers βk are zero for odd k > 1 since the function x/(ex − 1) − 1 + x/2 = i≥2 βi x i /i! is even.3. It follows that log Φ−1 ch(U) = r h since log(1 + z) = z − z2 /2 + ··· and h2 = 0 . we conclude that e(JC f ) = αk for f representing a generator of π2k−1 (U(n)) . which comes from the formula x/(ex − 1) = i βi x i /i! . If f represents a suitable choice of generator of π2k−1 (U (n)) then chk (Ef ) = h by Proposition 4. It is not hard to compute that β1 = −1/2 . More precisely : (1) The denominator of βk is the product of all the distinct primes p such that p − 1 divides k . obtaining k≥1 j 2j αj chj (E) specializes to log Φ−1 ch(U ) = αk x k−1 /(k − 1)! = ex /(ex − 1) − x −1 = 1 + (ex − 1)−1 − x −1 = 1 − x −1 + =1+ k≥1 βk x β x k≥0 k k−1 /k! k−1 /k! where the last equality uses the fact that β0 = 1 .9. Since e(JC f ) was deﬁned to be r . The ﬁrst step in proving the theorem is to relate Bernoulli numbers to the numbers Sk (n) = 1k + 2k + ··· + (n − 1)k . so α1 = 1/2 and αk = −βk /k when k = 1 . up to a sign which we can make +1 by rechoosing β if necessary. we have Φ−1 ch(U) = 1 + r h where h is a generator of H 2k (S 2k ) . the general formula log Φ−1 ch(U ) = k Proof of 4. Proposition 4. As described earlier. the class β ∈ K(CJf ) = K(T (Ef )) is the Thom class U . Proof: Sk (n) = n−1 =0 k i=0 k i βk−i ni+1 /(i + 1) . Q) = 0 for j ≠ k . ∞ k=0 The function f (t) = 1 + et + e2t + ··· + e(n−1)t has the power series expansion ∞ k=0 k k t /k! = Sk (n)t k /k! . Comparing the two calculations of log Φ−1 ch(U ) .

This will also follow from (∗) once we see that each term after βk p is p times a p integer. For k = 2 . p i−1 /(i + 1) is a p integer by induction on i as in the preceding paragraph. there would exist an exponent n < p − 1 such that the equation x n − 1 would have p − 1 roots in Zp . Proof of 4. We will ﬁrst apply (∗) to prove that pβk is a p integer for all primes p . So the product βk−i p i+1 /(i + 1) is a p integer for i > 0 . we may assume pβk−i is a p integer for i > 0 . p i /(i + 1) is a p integer since p i ≥ i + 1 by induction on i . which is p times a p integer. this term is 2(−1/2)p 2 /2 = −p 2 /2 . it follows that each term in (∗) containing a βk−i with i > 1 is p times a p integer. with power series ∞ i=1 ni t i−1 /i! ∞ j=0 βj t j /j! = k i=0 ∞ i=0 ni+1 t i /(i + 1)! ∞ j=0 βj t j /j! Equating the coefﬁcients of t k we get Sk (n)/k! = ni+1 βk−i /(i + 1)!(k − i)! Multiplying both sides of this equation by k! gives the result. pβk ≡ Sk (p) mod p in Z(p) . rational numbers whose denominators are relatively prime to p . so {1. Statement (1) of the theorem now follows since if p − 1 does not divide k then pβk ≡ Sk (p) ≡ 0 mod p so βk is p integral. so we must have Sk (p) ≡ 0 mod p . while Sk (p) ≡ 0 mod p in the opposite case. By induction on k . For p cyclic. (If it were not p since the multiplicative group Z∗ = Zp − {0} has order p − 1 and p − 1 divides k . ··· . Now we assert that Sk (p) ≡ −1 mod p if p − 1 divides k .) Let g be a generator of Z∗ . but a polynomial with coefﬁcients in a ﬁeld cannot have more roots than its degree. while if p − 1 does divide k then pβk ≡ Sk (p) ≡ −1 mod p so βk is not p integral and p divides the denominator of βk . As for the term containing βk−1 . that is.9: (∗) We will be interested in the formula for Sk (n) when n is a prime p : Sk (p) = βk p + k 1 βk−1 p 2 /2 + ··· + β0 p k+1 /(k + 1) Let Z(p) ⊂ Q be the ring of p integers. that is. In the ﬁrst case we have Sk (p) = 1k + ··· + (p − 1)k ≡ 1 + ··· + 1 = p − 1 ≡ −1 mod p the second case we use the elementary fact that Z∗ is a cyclic group. Since we know βk−i p is a p integer. this is zero if k is even and greater than 2 . This is equivalent to saying that the denominator of βk contains no square factors. Then p p Sk (p) = 1k + ··· + (p − 1)k = 1k + g k + g 2k + ··· + g (p−2)k and hence (g k − 1)Sk (p) = g (p−1)k − 1 = 0 since g p−1 = 1 .1 107 On the other hand. g p−2 } = Z∗ . Also. the difference pβk − Sk (p) is p times a p integer. To prove statement (2) of the theorem we will use the following fact: . Next we show that for even k . f (t) can be expressed as the product of (ent −1)/t and t/(et −1) .Lower Bounds on Im J Section 4. If p − 1 does not divide k then g k ≠ 1 . For i > 1 . Thus every term except βk p in (∗) is a p integer. and hence βk p is a p integer as well. g 1 . g 2 .

so g k is not congruent to 1 mod p . the (k − 1) st derivative of f (t)(f (t))−1 is (f (t))−k times a polynomial in f (t) and its derivatives. an integer. from the formula for ψk (β) we have ch ψk (β) = kn ch(β) + µk ch(α) = kn b + (kn r + µk )a . If p divides the denominator of βk then obviously p divides the denominator of βk /k . so ψk (ξ) = ξ k . Then p does not p divide g k (g k − 1) . Recall the function f (t) = (ent − 1)/(et − 1) considered earlier.13. Let g be a generator of Z∗ as before. Thus (nk − 1)βk /k = ±m/nk and so nk (nk − 1)βk /k is an integer. It is easy to check that replacing β by β + pα for p ∈ Z adds p to µk /(km − kn ) . if p does not divide the denominator of βk . There is an alternative deﬁnition of e purely in terms of K–theory and the Adams operations ψk . On the other hand. then by statement (1). By the argument in the proof of Theorem 2. The ﬁrst statement of the theorem follows immediately from (1) and (2). First. Conversely.108 Chapter 4 The J–Homomorphism Lemma 4. To prove this formula it sufﬁces by the splitting principle and additivity to take ξ to be a line bundle. On the other hand. From the formula f (t) = k≥0 Sk (n)t k /k! derived earlier. Proposition 4. This follows by computing ch ψk (β) in two ways. Proof: e(f ) = µk /(km − kn ) in Q/Z . with integer coefﬁcients. This has logarithmic derivative f (t)/f (t) = (log f (t)) = [log(ent − 1) − log(et − 1)] = nent /(ent − 1) − et /(et − 1) We have ex /(ex − 1) = 1/(1 − e−x ) = x −1 [−x/(e−x − 1)] = So f (t)/f (t) = ∞ k=1 ∞ k=0 (−1)k βk x k−1 /k! (−1)k (nk − 1)βk t k−1 /k! where the summation starts with k = 1 since the k = 0 term is zero. as one can readily see by induction on k . The rational number µk /(km − kn ) is therefore independent of k . p − 1 does not divide k . Statement (2) of the theorem can now be proved. nk (nk − 1)βk /k is an integer. Proof: For all n ∈ Z .17 there are formulas ψk (α) = km α and ψk (β) = kn β + µk α for some µk ∈ Z satisfying µk /(km − kn ) = µ /( m − n ) . there is a general formula chq ψk (ξ) = kq chq (ξ) where chq denotes the component of ch in H 2q . so p does not divide the denominator of βk /k .12. hence chq ψk (ξ) = chq (ξ k ) = [c1 (ξ k )]q /q! = [kc1 (ξ)]q /q! = kq c1 (ξ)q /q! = kq chq (ξ) . so µk /(km − kn ) is well-deﬁned in Q/Z . hence βk /k is the integer g k (g k − 1)βk /k divided by the number g k (g k − 1) which is relatively prime to p . The (k − 1) st derivative of this power series at 0 is ±(nk − 1)βk /k . So the (k − 1) st derivative of f (t)/f (t) at 0 has the form m/f (0)k = m/nk for some m ∈ Z . we have f (i) (0) = Si (n) .

.1 109 In the case at hand this says chm ψk (β) = km chm (β) = km r a . Comparing this with the coefﬁcient of a in the ﬁrst formula for ch ψk (β) gives µk = r (km − kn ) .Lower Bounds on Im J Section 4.

Cambridge Univ. Modern Algebra. F. A. J. Addison-Wesley. J. 1966 (later editions by Springer-Verlag). Atiyah. On the K–Theory of Lie groups. K–Theory and reality. Atiyah and R. W. Acta Math. K–Theory: An Introduction.M. F. Atiyah and F. F. 31–38. Atiyah. Papers J. e a W. 7–38. Math. Math. A History of Algebraic and Differential Topology 1900-1960. L. N. Atiyah. L. M. 17 (1966). M. Adams. The order of the image of the J–homomorphism. Princeton Univ. Geometry of Coxeter Groups. M. Adams. Pure Math. 1974. D. Topology. Steenrod. Quart. 3–38. Benjamin. of Math. Fibre Bundles. A. Vector Bundles and Homogeneous Spaces. Stasheff. Hodgkin. Press. J. u M. Fulton. Bott. Sym. 1–36. Quart. Topology 1 (1962). 367–386.S. Shapiro. 313–337. F. F. R. Milnor and J. Bott and J. 1969. R. Princeton Univ. 1949. Adams. Bott. 1968. Hiller. 603–632. On the groups J(X) IV. Hirzebruch. K–Theory. 70 (1959). 1997. Pitman. B. 1965. 1978. K–theory and the Hopf invariant. Bull. M. Power Operations in K–Theory.M. Press. McGraw-Hill. J. 75 (1962). 1967. Press. Clifford Modules. Quart. 1310– 1313. of Math. Ungar. Ann. 64 (1958). A. 20–104. Proc. A. . Vector ﬁelds on spheres. S. R. Bott and A.Books M. Atiyah. The stable homotopy of the classical groups. 1982. Milnor. Math. Mahowald. 1951.S. III (1961). of Math. Characteristic Classes. Birkh¨user. Press. Ann. 72 (1960). Adams and M. Topology of Fiber Bundles. Princeton Univ. R. Oxford. 76 (1970). 21–71. 229–247. R. 17 (1966). F. Lang. On the non-existence of elements of Hopf invariant one. Atiyah. Topology 3 (1964). J. 165–193. M. 245–248. Benjamin. M. 112 (1964). 6 (1967). Algebra. Young Tableaux.. Topology 5 (1966). Husemoller. 17 (1966). Notes on Cobordism Theory. On the periodicity theorem for complex vector bundles. 87–89. W. J. Vector bundles and the K¨nneth formula. W. M. Bull. D. Springer-Verlag. 1989. Ann. Stong. Dieudonn´. On the parallelizability of spheres. Bott. J. van der Waerden. H. J. Oxford. Karoubi. Lectures on K(X). Atiyah.

J. Oxford. Genetics of homotopy theory and the Adams conjecture. of Math. of Math. Some consequences of a theorem of Bott. Quillen. Sullivan. Topology 10 (1971). Quart. Vector ﬁelds on spheres and a generalization. 1–79. 357–366. 24 (1973). Woodward. Math. 68 (1958). L. 111 D. Ann. 444–449. 67–80.Bibliography J. 100 (1974). The Adams conjecture. Ann. Milnor. . M. D.

- Vector Bundles and K Theory - Hatcher
- Vector Bundles and K Theory - Allen Hatcher
- Hatcher-Vector Bundles and K-Theory
- Art-Fibre Bundle, Jet Manifolds and Lagrangian Theory
- Gauss Bonet Theorem
- Stage 2014
- VB
- Vector Bundles and K-Theory
- vector bundles
- 955Ass1_2010
- Basics on Hermitian Symmetric Spaces
- Differential Geometry
- Moment
- Vector Bundles UCL
- Dupont Fibre Bundles
- Lie Notes
- Damien Calaque and Carlo A. Rossi- Lectures on Duflo isomorphisms in Lie algebra and complex geometry
- Riemannian Geometry
- lie groups of type G_2
- geomim
- The Topology of Fiber Bundles by Cohen
- Problem Sheet
- Linear Span
- 1-s2.0-S0196885814000220-main
- seiberg witten rev2edition
- Lie Reps
- Vector Projection
- la01-rips
- 554midterm2 Solutions
- Hatcher - Vector Bundles and K-Theory

Are you sure?

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

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd