294 14 7MB
English Pages 185 Year 1978
Lecture Notes in Mathematics Edited by A Dold and B. Eckmann
640 Johan L. Dupont
Curvature and Characteristic Classes
Springer-Verlag Berlin Heidelberg New York 1978
Author Johan L. Dupont Matematisk Institut Ny Munkegade DK-BOOO Aarhus C/Denmark
AMS Subject Classifications (1970): 53C05, 55F40, 57D20, 58AlO, 55J10 ISBN 3-540-08663-3 Springer-Verlag Berlin Heidelberg New York ISBN 0-387-08663-3 Springer-Verlag New York Heidelberg Berlin This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically those of translation, reprinting, re-use of illustrations, broadcasting, reproduction by photocopying machine or similar means, and storage in data banks. Under § 64 of the German Copyright Law where copies are made for other than private use, a fee is payable to the publisher, the amount of the fee to be determined by agreement with the publisher.
© by Springer-Verlag Berlin Heidelberg 1978 Printed in Germany Printing and binding: Beltz Offsetdruck, Hemsbach/Bergstr. 2141/3140-643210
INTRODUCTION
These notes are based on a series of lectures given at the Mathematics Institute, University of Aarhus, during the academic year 1976-77. The purpose of the lectures was to give an introduction to the classical Chern-Wei 1 theory of characteristic classes with real coefficients presupposihg only basic knowledge of differentiable manifolds and Lie groups together with elementary homology theory. Chern-Wei 1 theory is the proper generalization to higher dimensions of the classical Gauss-Bonnet theorem which states that for
M
a compact surface of genus
( 1)
where
g
in 3-space
2 (1-g)
K
is the Gaussian curvature. In particular
topological invariant of
M.
JM
K
In higher dimensions where
a compact Riemannian manifold,
K
in (1)
is a M
is
is replaced by a
closed differential form (e.g. the Pfaffian or one of the Pontrjagin forms, see chapter
examples 1 and 3)
associated to
the curvature tensor and the integration is done over a singular chain in
M.
In this way there is defined a singular cohomology
class (e.g. the Euler class or one of the Pontrjagin classes) which turns out to be a differential topological invariant in the sense that it depends only on the tangent bundle of
M
considered as a topological vector bundle. Thus a repeating theme of this theory is to show that certain quantities which
a
priori depend on the local differential
geometry are actually global topological invariants. Fundamental
IV
in this context is of course the de Rham theorem which says that every real cohomology class of a manifold
M
can be re-
presented by integrating a closed form over singular chains and on the other hand if integration of a closed form over singular chains represents the zerococycle then the form is exact. In chapter 1 we give an elementary proof of this theorem (essentially due to A. Weil [34]) which depends on 3 basic tools used several times through the lectures: gration operator of the Poincare lemma, covering,
(i) the inte-
(ii) the nerve of a
(iii) the comparison theorem for double complexes
(I have deliberately avoided all mentioning of spectral sequences) . In chapter 2 we show that the de Rham isomorphism respects products and for the proof we use the opportunity to introduce another basic tool:
(iv) the WhitneyThornSullivan theory of
differential forms on simplicial sets. The resulting simplicial de Rham compleK, as we call it, connects the calculus of differential forms to the combinatorial methods of algebraic topology, and one of the main purposes of these lectures is to demonstrate its applicability in the theory of characteristic classes occuring in differential geometry. Chapter 3
contains an account of the theory of connection
and curvature in a principal Gbundle
(G
a Liegroup) essential-
ly following the exposition of Kobayashi and Nomizu [17>]. The chapter ends with some rather long exercises (nos. 7 and 8) explaining the relation of the general theory to the classical theory of an affine connection in a Riemannian manifold. Eventually, in chapter 4 we get to the ChernWei I construction in the case of a principal Gbundle a connection
e
and curvature
manifold mentioned above
G = O(n)
IT:
E
M with
(in the case of a Riemannian and
E
is the bundle of
v orthonormal tangent frames).
In this situation there is
associated to every G-invariant homogeneous polynomial the Lie algebra
p(n k)
a closed differential form
defining in turn a cohomology class
WE(P)
EH
2k
P
on
on
M
(M,lR).
Before proving that this class is actually a topological invariant of the principal G-bundle we discuss in chapter 5 the general notion of a characteristic class for topological principal G-bundles. By this we mean an assignment of a cohomology class in the base space of every G-bundle such that the assignment behaves naturally with respect to bundle maps. The main theorem (5.5) of the chapter states that the ring of characteristic classes is isomorphic to the cohomology ring of the classifying space
BG.
Therefore, in order to define the characteristic class WE(P)
for
E
any topological G-bundle it suffices to make the
Chern-Weil construction for the universal G-bundle BG.
Now the point is that although
BG
EG
over
is not a manifold it
is the realization of a simplicial manifold,
that is, roughly
speaking, a simplicial set where the set of p-simplices constitute a manifold. Therefore we generalize in chapter 6 the simplicial de Rham complex to simplicial manifolds, and it turns out that the Chern-Weil construction carries over to the universal bundle. In this way we get a universal Chern-Weil homomorphism W:
where
I*(G)
r*(G) ... H*(BG,lR)
denotes the ring of G-invariant polynomials on
the Lie algebra In chapter 7 we specialize the construction to the classical groups obtaining in this way the Chern and Pontrjaging classes
VI
with real coefficients. We also consider the Euler class defined by the Pfaffian polynomial and in an exercise we show the GaussBonnet formula in all even dimensions. Chapter 8 is devoted to the proof of the theorem (8.1) due to H. Cartan that for
G
H*(BG,JR)
is an isomorphism
a compact Lie group. At the same time we prove A.Borel's
theorem that of
w: I*(G)
H*(BG,JR)
H*(BT,JR)
is isomorphic to the invariant part
under the Weyl group
W of a maximal torus
T.
The corresponding result for the ring of invariant polynomials (due to C. Chevalley) depends on some Lie group theory which is rather far from the main topic of these notes, and I have therefore placed the proof in an appendix at the end of the chapter. The final chapter 9 deals with the special properties of characteristic classes for Gbundles with a flat connection or equivalently with constant transition functions. If
G
is
compact it follows from the above mentioned theorem 8.1 that every characteristic class with real coefficients is in the image of the ChernWeil homomorphism and therefore must vanish. In general for
K
G
a maximal compact subgroup we derive a
formula for the characteristic classes involving integration over certain singular simplices of
G/K.
As an application we
prove the theorem of J. Milnor [20] that the Euler number of a flat
Sl(2,JR)bundle on a surface of genus
value less than
h
has numerical
h.
I have tried to make the notes as selfcontained as possible giving otherwise proper references to wellknown textbooks. Since our subject is classical, the literature is quite large, and especially in recent years has grown rapidly, so I have made no attempt to make the bibliography complete.
VII
It should be noted that many of the exercises are used in the main text and also some details in the text are left as an exercise. In the course from which these notes derived the weekly exercise session played an essential role. I am grateful to the active participants in this
course, especially to
Johanne Lund Christiansen, Poul Klausen, Erkki Laitinen and S¢ren Lune Nielsen for their valuable criticism and suggestions. Finally I would like to thank Lissi Daber for a careful typing of the manuscript and prof. Albrecht Dold and the Springer-Verlag for including the notes in this series.
Aarhus, December" 15, 1977.
CONTENTS Chapter
PClge
1.
Differential forms and cohomology
2.
Multiplicativity.
3.
Connections in principal bundles
38
4.
The Chern-Weil homomorphism
61
5.
Topological bundles and classifying spaces
71
6.
Simplicial manifolds. for BG
89
The simplicial de Rham complex
20
The Chern-Weil homomorphism
7.
Characteristic classes for some classical groups
97
8.
The Chern-Weil homomorphism for compact groups
114
9.
Applications to flat bundles
144
References
165
List of symbols
168
Subject index
170
CURVATURE AND CHARACTERISTIC CLASSES
1.
Differential forms and cohomology
First let us recall the basic facts of the calculus of differential forms on a differentiable manifold differential form
w
vector fields
•• 'X
w(X 1 , .•. ,X
k)
w(X 1 , ••. ,X k)p
X1"
of degree
k
associates to
a real valued
k
M.
COO
A k
function
such that it has the "tensor property" depends only on
X1p""'Xk p
For an l-form
and a k-form
(i.e. p E M)
for all
x1 , ... ,X k.
and such that it is multilinear and alternating in
the
COO
is
the product
(k+l)-form defined by
where
runs through all permutations of
0
1, ... ,k+l.
This
product is associative and graded commutative, i.e.
Furthermore there is an exterior differential k-form
w
associates a
(k+1)-form
dw
d
defined by
1 [k+1 i+' (do ) (X1, ... ,X k+ 1) = k+1 J1(-1) Xi(w(X"
+
L
i
M,
cc- (U) ,3)
such that and
rc- (U), 0)
be the corresponding chain or co chain complexes (called "with support in
un) and let C* (M)
->
C* (U )
be the natural maps induced by the inclusion
10
1
c SOO (M). n
SOO (U)
:
n
Then
1
*
and
1
*
are chain
equivalences, in particular they induce isomorphisms H(C*(M)),
H(C*(U))
H(C*(M))
H(C*(U)).
We now define a natural map
by the formula (1.13)
I
I (w) a =
a E Soo{M).
a*w,
n
is clearly a natural transformation of functors, that is, if COO
is a
map,
I where
f*:
A*(N)
then
f* = f#
0
A*(M)
and
0
I,
f#
C*(N)
C*{M)
are the
induced maps. Lemma 1.14.
I
is a chain map, i.e. I
In particular
o
d
0
l(dw)T
I.
induces a map on homology
I
H(C*
1: H(A*(M)) Proof.
0
(M)) •
This simply follows using exercise 2 above:
r n+1
I
oJ
n-e l i=O W
. (_1)1
dT*w
.
{E1)*T*W
{_1)1
i=O
I
r
+1 T* (dw)
I(w)
E An(M),
T
()= Ei
E
o(l(w))T'
T
n+l (M) .
SOO
11
Theorem 1.15.
(de Rham).
an isomorphism for any
COO
I : H*(A*(M»
manifold
H(C*(M)
is
M.
First notice: Lemma 1.16.
Theorem 1.15 is true for
M diffeomorphic to
a star shaped open set in llin . Proof.
It is clearly enough to consider
open set star shaped with respect to consider the homotopy and
g(-,O) = e
g: U
x
e
E U.
U
[0,1]
M = U clli
n
an
As in Lemma 1.2
with
g(-,l)
=
id
given by g(x,s) = sx + (l-s)e.
By (1.11) the inclusion
{e}
U
induces an isomorphism
in singular cohomology, so the statement follows from (1.10) together with Lemma 1.2 and the commutative diagram H (A* (U»
-
1
H(A* (e)
I
H (C* (U»
I
H (C*
II
(e )
II ]R
JR
Lemma 1.17.
For any
there is an open covering
COO
manifold
U = {Ua}aEL'
M of dimension
such that every non-
U n... n U , aO, •.. ,a p E aO ap diffeomorphic to a star shaped open set of ]Rn. empty finite intersection
Proof.
Choose a Riemannian metric on
point has a neighbourhood every point of
U
U
M.
U).
(i.e., for every
In particular,
U
I,
is
Then every
which is normal with respect to q E U,
eXP q
diffeomorphism of a star shaped neighbourhood of onto
n
is a
oE
T
q
(M)
is geodesically convex, that is,
12
for every pair of points segment in U.
M
joining
p,q E U p
and
there is a unique geodesic
q
and this is contained in
[14, Chapter I Lemma
(For a proof see e.g. S. Helgason
= {Ua}aEZ
6.4).
Now choose a covering
sets.
Then any non-empty finite intersection
U
with such open
n ... n u a k
is again geodesically convex and so is a normal neighbourhood of each of its points.
It is therefore
clearly diffeomorphic to a star shaped region in
(via
the exponential map).
In view of the last two lemmas it is obvious that we want to prove Theorem 1.15 by some kind of formal inductive argument using a covering as in Lemma 1.17.
What is needed
are some algebraic facts about double complexes: We consider modules over a fixed ring shall only use
A complex
R =
with a differential
C
-+
n+1
Similarly, a double complex is a
ip,ql
cp,q,
,
z::
is a
nEZ::, x
(actually we module
such that
dd = O.
Z::-graded module
together with two differentials ... CP
d'
C*
R
+1, q ,
satisfying (1. 18)
d'd'
0,
d"d"
d"d' + d'd"
0,
We shall actually assume that complex, that is, Associated to
cp,q = 0
(C*'*,d' ,d")
C*,*
O.
is a 1. quadrant double or
q < O.
is the total complex
(C* ,d)
if either
p < 0
where d
d'
+ d".
13
For fixed to
d'.
q
we can take the homology of
C*,q
with respect
This gives another bi-graded module
Now suppose
1c*,*
as above, and suppose
and
2C*,*
are two double complexes
f
respecting the grading and commuting with clearly
f
d'
and
d".
Then
gives a chain map of the associated total complexes
and hence induces
Also clearly
f*: H(1c*,d) .... H(2c*,d).
.... EP,q f 1 .. 1EP,q 1 2 1
induces
HP(C*,q,d').
=
Lemma 1.19.
Suppose
We now have: f : 1C*,* .... 2C*,*
of 1. quadrant double complexes and suppose is an isomorphism.
f
Then also
is a homomorphism f 1 : 1E1'* .... 2E1'*
f* : H(1C*) .... H(2C*)
is an
isomorphism. Proof. complex
For a double complex
(C*,d)
(C*,*,d' ,d") F*
define the subcomplexes
q
with total
= C*,
C
q E:
7l,
by
F* q
Then clearly ::>
and
d : F* .... F*. q
isomorphic to
q
F* q-l
::>
F* q
::>
Notice that the complex
(C*,q,d').
Therefore for
is f
1
C*,* ....
2
C*,*
a map of double complexes the assumption that f
1
:
EP'q....
1 1
that
EP,q
2 1
f:
is an isomorphism, is equivalent to saying ....
in homology.
q E: 7l,
Now by induction for
induces an isomorphism
r = 1,2, ...
it follows
from the commutative diagram of chain complexes 0
....
0
....
1F*q+r / 1F*q+r+1 H
2F q+r / 2 F*q+r+l
.... 1F*/ .... 1F*/ q 1F*q+r q 1F*q+r+1 H
H
....
2F*/ q 2F*q+r+1
.... 0
....
2 F*/ q 2 F*q+r
.... 0
14
and the five lemma that f:
F*/ F*
F*/ F*
1 q 1 q+r
2 q 2 q+r
q E
induces an isomorphism in homology for all r
=
C*, *
1,2, . . . .
and
However, for a 1. quadrant double complex
we have
o
for
r > n
so the lemma follows. Remark.
Interchanging
similar lemma with
p
and
replaced by
q
in H
cp,q
q (C p ,
we get a
* , d ")
•
Notice that for a 1. quadrant double complex follows from (1.18) that denoted
d"
d" :
it
induces a differential also for each
E O,q 1
C*,*
In particular, since
p.
ker(d'
we have a natural inclusion of chain complexes e
(Ef'* ,d")
(C* ,d)
(called the "edge-homomorphism"). Corollary 1.20.
o
Suppose
for
p >
o.
Then
induces an isomorphism H(Ef'*,d") Proof.
q
H(C*,d).
is a double complex with
Lemma 1.19 for the natural inclusion
d'
= 0
cp,q.
Apply
e
15
Note.
For more information on double complexes see e.g.
G. Bredon [7, appendix) or S. Mac Lane [18, Chapter 11, 3 and 6). We now turn to Proof of Theorem 1.15. of
M as in Lemma 1.17.
complex as follows:
Choose a covering
p,q
n
0
Aq(U
(a O ' ••• ,a )
a
p
{Ua}aEE
0
consider
n ... n U ) a p
where the product is over all ordered (p+l)-tuples such that
=
Associated to this we get a double
Given
=
U
n... n
U
Ua p
aO
differential is given by
P,q AU
-+
* @.
(ao, ... ,a p )
The "vertical"
AP,q+l U
n... n U ) -+ Aq+ l (U n... n Ua is the aO ap aO p exterior differential operator. The "horizontal" differential
where
d: Aq(U
AP,q U
-+
AP+l,q U
is given as follows: For Aq(u
Cl O
(1.21)
w =
(w
n••• o u
(a O ' ••• ,a p)
Clp + l
(ow)
(Cl
)
O,
) E
,q
the component of
is given by p+l
L
· · · , ap + l )
It is easily seen that
00
=
i=O 0
and
double complex. Now notice that there is a natural inclusion
ow
in
16
Lemma 1.22.
For each
q
the sequence -+
•••
is exact. Proof.
A-1 , q
In fact putting
we can construct
U
homomorphisms
such that (1 .23)
id.
To do this just choose a partition of unity
with
and define (s w)
p
(U
O
, · · · , up _ 1 )
aEL
U
(U
O , · · · ,a p _ 1 ,a) ,
w E It is easy to verify that
s
p
is well-defined and that (1.23)
is satisfied. It follows that p > 0 p
O.
Together with Corollary 1.20 this proves Lemma 1.24.
Let
A*
U
be the total complex of
there is a natural chain map eA
: A*(M) ...
which induces an isomorphism in homology.
Then
17
A*
We now want to do the same thing with the singular cochain functor
replaced by
As before we get a double
complex
n
Cq(U
(aO, .•• ,a ) p
a
0
n... n
U ) a p
where the "vertical" differential is given by
c*(u
the coboundary in the complex
(-1)P
times
n... n
U ) and where a p the "horizontal" differential is given by the same formula as (1.21) above.
"o
Again we have a natural map of chain complexes
and we want to prove
C*(M)
Lemma 1.25.
induces an isomorphism in
homology. Suppose for the moment that Lemma 1.25 is true and let us finish the proof of Theorem 1.15 using this. For
U c M
we have a chain map
A*{U)
I
C*(U)
as defined by (1.13) above.
Therefore we clearly get a map
of double complexes I and we have a commutative diagram
U
u
teA
t e C
A* (M)
C* (M)
A* ----+. C
18
By (1.24) and (1.25) the vertical maps induce isomorphisms in homology.
It remains to show that the upper horizontal
map induces an isomorphism in homology.
Now by the remark
following Lemma 1.19 it suffices to see that for each
is an isomorphism.
However this is exactly Lemma 1.16 applied
n ... n u cx p
to each of the sets Proof of Lemma 1.25. with
A*
replaced by
It is not true that Lemma 1.22 holds
C*.
However, if we restrict to cochains
U
with support in the covering let
Cq(U)
o(6
q
C
it is true.
Thus as in (1.12)
denote the q-cochains defined on simplices
a E Soo(U), i.e. for each q)
p
U cx.
a E SOO(U)
there is a
q
u
with
Then there is a natural restriction map
cq(U)
and the sequence
(1 .26)
-of'
is exact.
•
&.
In fact we construct homomoprhisms
Cp-1 u as follows:
For each
(C- 1 , q U
a E Soo(U)
choose
q
cx(o)
E
L
such that
Then an easy calculation shows that
It follows that the chain map e
C
=
e
C
0 1*,
where
e
1*: C*(M)
c
: C*(M) C*(U)
u
C
factors into
is the natural chain
map as in (1.12) and where the edge homomorphism
19
induces an isomorphism in homology by Corollary 1.20 and the exactness of (1.26).
Since
t*
also induces an isomorphism
in homology by (1.12) this ends the proof of Lemma 1.25 and also of Theorem 1.15. Exercise 4.
For a topological space
X
denote the set of continuous singular n-simplices of and let
and
and cochain complexes. the inclusion
S;(M)
X,
be the corresponding chain Show that for a S;oP(M)
COO
manifold
M
induces isomorphisms in homology H(C*(M)).
(Hint:
Use double complexes for a covering as in Lemma 1 .17).
Hence the homology and cohomology based on
COO
singular
simplices agree with the usual singular homology and cohomology. It follows therefore from Theorem 1.15 that the de Rham cohomology groups are topological invariants. Exercise 5.
Show directly the analogue of the homotopy
property (1.11) for the de Rham complex. Note.
The above proof of de Rham's theorem goes back to
A. Weil [34].
It contains the germs of the theory of sheaves.
For an exposition of de Rham's theorem in this context see e.g. F. W. Warner [33, chapter 5].
2.
Multiplicativity.
The simplicial de Rham complex
In Chapter 1 we showed that for a differentiable manifold M
the de Rham cohomology groups
invariants of
makes
A*(M)
M.
Hk(A*(M»
are topological
As mentioned above the wedge-product
an algebra and it is easy to see that (2.1)
induces a multiplication (2.2) In this chapter we shall show that (2.2) is also a topological invariant.
More precisely, let
be the usual cup-product in singular cohomology; then we shall prove Theorem 2.4.
For any differentiable manifold
M
the
diagram Hk(A*(M»
@ Hl(A*(M»
n Hk(C*(M»
(i9
Hk + l
/\
(A* (M) )
n
I
Hk+l (C* (M) )
@ Hl(C*(M»
commutes. For the proof it is convenient to introduce the simplicial de Rham complex which is a purely combinatorial construction closely related to the cochain complex
C*
but on the other
hand has the same formal properties as the de Rham complex
A*.
21
We shall define it for a general simplicial set: Definition 2.5.
A simplicial set
S = {Sq}' q = 0,1,2, •.. ,
S
is a sequence
of sets together with face operators
E:.
Sq
->
S q- l' i
O, ... ,q,
and degeneracy operators
ni
S
->
Sq+1' i
O, .•• ,q,
which satisfy the identities
1.
q
(i) (ii)
E: _ j 1E: i,
i < j,
n j + 1n i ,
i
f
i < j,
j
=
(iii)
Example 1. S
SOO (M) q
q
i
=
i ' id, "
i
n E: _ , j i 1
i > j + 1.
j+1,
j, i
We shall mainly consider the example, where StoP(M) .
or
0, ... ,q,
j,
q
where
E:
i
Here as in Chapter 1,
nq -
1
->
nq
E:i(O) = 0
0
E: i ,
is defined by
(2.6)
Analogously, the degeneracy operators
ni
(0 )
=0
0
ni, .1. = 0 , ... ,q,
h were
are defined by
nq + 1
->
nq
is defined
by (2.7)
We leave i t to the reader to verify the above identities. A map of simplicial sets is clearly a sequence of maps commuting with the face and degeneracy operators. SOO
and
stop
become functors from the category of
Obviously COO
manifolds (respectively topological spaces) to the category of simplicial sets.
22 Definition 2.8.
Let (jJ
A differential k-form
{S } q
S on
be a simplicial set. (jJ='{(jJa}' a EllS p P
is a family
S
of k-forms such that (i)
(jJa
is a k-form on the standard simplex
i *(jJa' (jJE.a =' (E)
Ei
where
w
is the i-th face map as defined by (2.6). co
Let
for
S =' S (M)
is a k-form on
by putting
O, ••• ,p, a ESp' P =' 1,2, ....
a,
Example 2. if
C
a E
for
(jJ A
If
(jJ E Ak(S),
co
manifold.
(jJ =' {(jJa}
we get a k-form
M
(jJa =' a*w
a
M
Then
on
S'" (M) •
P
The set of k-forms on a simplicial set Ak(S).
for
E Al(S)
S
is denoted
we have again the wedge-product
defined by 0,1, ...
(2.9)
Also, we have the exterior differential
d
defined by
a ESp'
(2.10)
It is obvious that
A
commutative and that (2.11)
dd =' 0
d
satisfies
=' d(jJ A (A*(S) ,A,d)
de Rham complex of then clearly we get (2.12)
is again associative and graded
and
d«(jJ A We shall call
P = 0,1,2, ...
S.
If
+ (_1)k(jJ A
E Al(S).
the simplicial de Rham algebra or f: S
f* : A*(5'} (jJfa'
(jJ E Ak(s),
5' A*(S}
is a simplicial map defined by 0,1, .••
23 and thus
A*
is a contravariant functor.
Remark 1. manifold
M
Notice that by Example 2 we have for any
COO
a natural transformation i
(2.13)
: A*(M)
-->
A*(Soo(M»
which is clearly injective, so we can think of simplicial forms on
Soo(M)
as some generalized kind of forms on
M.
We now want to prove a "de Rham theorem" for any simplicial set
8. The chain complex
with real
C*(8)
coefficients is of course the complex where vector space on
8
a
and
k
C k(8) k
a (a)
'i
(-1)
i=O
-->
C k(8)
is given by
Ck- 1 (8)
i E (a), i
is the free
a E Sk'
Dually the cochain complex with real coefficients is C* (8) = Hom(C* (8) ,JR),
so again a k-cochain is a family
c = (c
0 : Ck(8)
a),
a E 8
k,
and
(2.14 )
(DC)a
=
k+1
'i
i=O
Ck + 1 (8)
-->
is given by
.
(-1)
l
c
EiT
,
T
E 8 k + 1•
Again we have a natural map
defined by (2.15)
I(j)a
(j)
k
E A (8), a E 8
k,
and we can now state Theorem 2.16 (H. Whitney).
I : A*(8}
-->
C*(8)
chain map inducing an isomorphism in homology.
is a
In fact there
24
E : C*(S)
is a natural chain map
k A - 1 (S), k
k
homotopies
sk : A (S)
(2.17)
1
0
d
o
(2.18)
1
0
E
id,
I,
0
A*(S) and natural chain
E
0
E
0
0
1 , 2 , ... ,
suc h th a t
doE
- id
sk+1
0
d + d
= 0,1 , ...
k
For the proof we first need some preparations. usual
sk'
0
As
is the standard p-simplex spanned by the
canonical basis coordinates
{eo, •.• ,e
(to, •.. ,t
p)' respect to each vertex e
have operators each
j
h(j)
and we use the barycentric
p}
Now j,
j
is star shaped with and therefore we
0, ... ,p,
for
:
as defined in the proof of Lemma 1.2.
o
w E
for
Also put
The prooL of Lhe following lemma
is left as an exercise (cf. Exercise 3 of Chapter 1): Lemma 2.19. k
k
The operators
P
k-1
0,1,2, ... , satisfy
k > 0
(2.20)
k (ii)
(2.21 )
(iii)
(2.22)
For
I ,j
0
0, ... ,p i
> j
i
< j
p
25
Next some notation: Consider a fixed integer Let
I
(iO, •.. ,i
satisfying I
is
to
be a sequence of integers
k)
< i < ... < i p, The "d.Li.reris Lon" of k 1 O (for 1=0 put 101 -1). Corresponding
we have the inclusion
dimensional face spanned by have a face map
: Sp
and
E.
=
E.
0 ••• 0
J1
{e.
I
onto the k-
, ... ,e. }
and similarly we
(2.23) (for
WI =
k
L
s=O
j1 >
> jl
and
k + 1
s
p.
A ••• A
E O ••• O
is the
0
A ••• A
dt.
0) and the operator
We can now define
E
(for
k + 1
which lowers the degree by
c = (c
(2.24)
T
E{c)
o
L
=
W
III=k I
which is clearly a k-form on
is defined as follows: (2.25)
=
.
h
0
= id) .
=
L
P
put
c
p < k
the sum is of k- 1{S) sk: Ak{S) A
Similarly
For
put
as follows (a
o E S
(if
course interpreted as zero).
0
below) :
a k-cochain and
)
I =
Ak{S)
Ck{S)
:
motivation is given in Exercise
E Ak{S) III !W I
and
[j E sP
put
A
which is clearly a k-1-form on First we show that (2.24) satisfies Definition 2.8 (ii): Let
1 E {O, ... ,p}
contain
j 1
defined by
on dt.
E
Also associated
A
(-1) t. dt.
I
For
j 1
p
there is the "elementary form"
I
I
u
where
J1
:
Explicitly,
Sk.
complementary sequence to to
o.
0 < i
III = k
I
p >
1.
and suppose
Then for some
s
I = (i
.. ,i does not k) O" and we have is < 1 < i + s 1
26
k .'
x:
\'
L
I I' I=k
since it is easy to see that I' = (ib, ... ,i
o
k)
ib < ••• < i
• c
wI'
L
III=k;lU
llI(a)
wI'
llI(a) = llI' (Ela).
Now since
runs over all sequences satisfying
k
p - 1,
the last expression above equals
E(c)
which was to be proved. Similarly (2.25) is shown Ela to satisfy Definition 2.8 (i) using (2.21) above. Now let us prove the identities (2.17):
The first
identity of (2.17) is proved exactly as Lemma 1.14, so let us concentrate on the second one:
a E S
p
c E Ck(S)
For
and
we have
k:
(2.26)
L
tII=k
(k+1):
L
III=k
dt. 10
dt. 1k
A ••• A
On the other hand (2.27)
(k+1) :
L
(Dc) II
III=k+1 (k+1) :
L
III=k+1 For
J
= (jo'· ··,jk)' 0
terms involving
W I
•
() in (2.27). llJ a (i o , · · · , i l,···,ik+ 1) = (jo,···,jk)·
c
llJ (a)
(a)
k+1
( L
1=0
jo < •.• < jk
c
I
1 (-1)
P
cE
lllI
we shall find the
Now
in (2.27) therefore is
(i O,···,i k+ 1) = (jo, ... ,jl-1,i,jl, ..• ,jk). Now (2.28) equals
iff
The coefficient of
(2.28)
where
(a)) •
27
L
(k+1):
[ L
.... (.JO, ••• ,J. ) s< 1 k
1.-.
(_1)s+lt. d t , A..• Adt. A..• Adt. Adt.Adt. A •. Js J0 Js J 1-1 1. J1
•• Adt. +t.dt. A ••• Adt. + Jk 1. JO Jk +
L.\'
(_1)s+1-1 t. dt. A ...A d t. d] A d t.A d t. A •••A t. A .•• At. JS JO J l- 1 1. Jl JS Jk
(k-e l l !
'lE('J " O
1.
.) ' " Jk
[t.dt. A..• xd t; . + 1. J0 Jk
k
L
+
+
L
S=O
-to dt. A ... Adt. Adt.Adt. A .•• Adt.] JS JO J s- 1 1. J s+ 1 Jk
k
L
L
S=O ilE(jO, .•• ,jk)
L
(k+1):[
-to dt. A..• xd t . Adt.Adt. A ••. Adt.] JS JO J s- 1 1. J s+1 Jk
ilE(jo, ... ,jk)
+
t.dt. A •.. xd t . + 1. JO Jk
k
k p dt. A ••• Adt. - L t. dt. A.•. xd t , AlL d t S=O J S JO Jk S=O J S JO J s- 1 i=O
L t.
j
1.
l xd t , A•. Adt.] J s+1 Jk
p (k+1):
L t.dt.
i=O
I
since
A ••. Adt. = JO Jk
J.
o
dt.
i=O
1.
E(oc)a = (k+1): by
(k-r l I Ld t , A.•. xd t . JO Jk
p
and
L
IJ I=k
L t.
i=O
1.
Hence
1.
dt. A... xd t, . • c () Jo Jk].lJ a
= dE(c)
a
(2.26) which proves the second identity of (2.17). To prove the first equation of
c = (c a), a E Sk (2.24)
E(c)o
(2.18)
and we shall show that
is the k-form on
consider a k-cochain
I(E(c))a = ca'
given by
By
28 k
k c
.
I (-1 ) J t
a j=O
.d to A... Adt . A.•. Adt
J
J
k
k.c [t ..• Adt + I (-1) k j=1 a odt 1A I
. 1 J-
k
t.dt.Adt J
J
A
1A
... Adt.A ... Adt J
Therefore
I (E (c)) a by Exercise 1 of Chapter 1. For the proof of the second equation of (2.18) first observe that an iterated application of (2.20) yields the following Lemma 2.29. I
=
(i
Let
, 0 < r ' ••• , i O r)
Suppose
Now let
k
r.
k w E A (f1,P) , k p,
with
and consider
0,
0 < i
O
< ••• < i
< p. r =
p
(otherwise
Then
E Ak(S)
and
a E S .
there is nothing to prove).
p
Assume
By (2.29)
k
k]
29
(2.30)
sk+1 (do) a
-
ill
L I! I
A
(
'
L (-1)J h ( 1, j=O
-:' ,)(\p» a 0 , ••• , 1 j , ••• , 1 1I I
Also
By (2.22) (-1)
k
h(,
1
' ) (.p ) (e.
0 , · · · , l k _1
a
1
k
)
Therefore adding (2.30) and (2.31) we obtain
E(I(\p»
a
-\p -
a
L 0< I I I
II I
'
L (-1)J h ( 1. j=O
0;-
0 , •• , 1 j , ••
,)(\p» a 1I I
,1
However the last two sums in (2.32) cancel by exactly the same calculations as in the proof that (2.26) equals (2.27) above. This proves the second equation of (2.18) and ends the proof of Theorem 2.16. We now return to the proof of Theorem 2.4. in the commutative diagram
Notice that
30 i
A* (M)
C* (M) all maps induce isomorphism in homology. i
: A*(M)
A*(Soo(M))
Also
is obviously multiplicative.
Theorem
2.4 therefore immidiately follows from Theorem 2.33.
For any simplicial set
S
the following
diagram commutes H (A
* (S))
H (A
+I
H(C*(S))
* (S) ) I
H(C*(S))
where the upper horizontal map is induced by the wedge-product of simplicial forms and the lower horizontal map is the cupproduct. Before proving this theorem let us recall the definition of the cup-product in
H(C*(S)).
Consider the functor
C*
from the category of simplicial
sets to the category of chain-complexes and chain maps (as usual we take coefficients equal to
An approximation to
the diagonal is a natural transformation
(in particular a chain map) such that in dimension zero is given by
It follows using acyclic models that there exists some
and
it is unique up to chain homotopy (see e.g. A. Dold [10, Chapter 6,
§
11, Exercise 4].
The cup-product is now simply
31
induced by the composed mapping *
C* (8 )
C* (8 ) ... Hom (C* (8 )
An explicit choice for
C* (8) , lR) ... C* (8) .
is the Alexander-Whitney map
AW
defined by (2.34)
n I u (a) p=o (O, •.. ,p)
AW(a)
With this choice of
u
(p, •.. ,n
a = (a ) E CP(8) a
then
is represented by the cochain
(2.35)
(avb)a=a
u ( 0 , ••• , p)
Proof of Theorem 2.32.
and
(a)·b
T
() u (p, ... , p+q ) a ,
By Theorem 2.16 every simplicial
form is cohomologous to a form in the image of It is therefore enough to show that for the (p+q) -cochain
I (E (a)
of
H(C*(8).
a
and
b
in
n•
q(8); b = (b ) E C
Let
b
E 8
a
the cup-product is explicitly given
as follows: a
) (a),
A
E (b) )
E: C*(8) ... A*(8).
a E CP(8), b E C
q(8)
represents the cup-product
80 let
* : C*(8)
C*(8) ... C*(8)
E(b»,
a E CP ( 8 ) , b E C
be defined by (2.36)
*{a
b)
I (E(a)
A
q(8).
We claim that there is an approximation to the diagonal inducing (2.36). Put E(a)
i
and consider
a E 8
n.
i O < ••• < p
I = (i ' o
s
n,
0
... ' i p)
and
n. If
I
than two integers in common then obviously I
6
(jo' ••. , jq)
J
jo < ••• < jq
has at least one integer in common.
suppose
Then on
n
,
I a ()w I I I =p III a I'
=
where as usual 0
Let us find an explicit formula for (2.36):
n = p + q
a
and
J
Then and wI
I
and
J
has more
J A
satisfy
W
J
= O.
Now
have exactly two integers in common, say
32
Then
s +r
+ ( -1) 2
2t . 2
t . d t . A.. Adt . Jr 1
s2
A•. Adt. Adt. A.. Adt . A•. Adt . JO Jr Jq P 1
and it is easy to see that these two terms are equal with opposite signs so and
wI A
W
= 0
also in this case.
Finally suppose
have exactly one integer in common, say
J
+
'\ s+k L (-1) t. t. dt. k*r Jk
n
Using
J
L dt;\
;\=0
o
is = jr;
I then
A... xd t , A... xd t.. Adt. A.. xd t . A.. xd t . JO Jk Jq A
we get
It follows that
(E(a)AE(b»
a
=p:q:
'\ r+s L a ()b ()'(-1) t. dt. A... xd t , A.. I I I =p ]JI a ]JJ a A
IJI=q
.. Adt. Adt. A•.. Adt. A... Adt . JO Jr Jq where the sum is taken over
I
and
J
such that for some
s
33
and
r
i
sgn(I,J)
jr and no other integers are common. Now let s be the sign (_1)p-s+r. times the sign of the
permutation taking
(O, •.. ,n)
into
(io,···,is,···,ip,is=jr,jo' " " ; r , ... ,jq);
r
sgn(I,J)J
t-,n
(-1)
r+s
t. dt. A... xd t,
then
A... Adt. xd t . A... Adt. A•.• xd t . JO Jr Jq A
is
1/ (n+1): • Hence (2.37)
I(E(a)
p:q:
(p+q+1):
where again
I
and
l:
III=p IJI=q J
A
E(b»(}
sgn(I,J)a
()b ()
() ()
have exactly one integer in common.
Therefore if we define the map
by 4> (o )
(2.38)
then 4>
4>*
l:
(
p+q=n
n+
1)
\" I L . I I I =p IJI=q
given by (2.36) is the dual map.
is an approximation to the diagonal:
() E Sn
@
We want to show that
Clearly
4>
is natural
and 4>(})
() @ ()
It remains to show that it is enough to see that
for
() E SO'
is a chain map.
4> 4>*
However, for this
is a chain map which is easy:
34
q,*(o(a@b))
q,*(oa@b + (-1)P a@ob) I(E(oa) "E(b)) +
(-1)
P l ( E ( a ) "E(ob))
l(dE(a) "E(b) + (-1)PE(a) "dE(ob) l(d(E(a) "E(b)) = ol(E(a) "[Cb)) = oq,*(a@b). This ends the proof. Remark.
Notice that the term in (2.37) corresponding to
I = (O, ... ,p), J = (p, •.. ,p+q) Whitney cup-product (2.35).
gives exactly the Alexander-
Thus (2.37) is an average of the
Alexander-Whitney cup-product over the permutations given by (I,J)
in order to make the product graded commutative on the
cochain level.
In fact the A-W-product is not graded
commutative on the cochain level as since
"
is graded commutative.
q,*
clearly must be
On the other hand the A-W-
product is associative on the cochain level which not.
q,*
is
In order to achieve both properties it seems necessary
to replace the functor
C*
by the chain equivalent functor
A*. Exercise 1. with
0
Consider for
i O < ••• < i k
k < P
and let
P
a sequence c
I = (iO, ••• ,ik) be the
set = {( to' ..• , t ) I some t. >O} = p
(i.e. we subtract a TIl :
- {t.. =t. = ... =t. =O},
p-k-1-dimensional face).
be the projection 1 s a)
Show that on
I
( t . , ... ,t.
).
Let
35
where
is given by (2.23). b)
Show the following properties of
c)
(i)
(IlI)*W
(ii)
(IlJ)*W
r
dt
I
a
E(c)o
on
c
dt
k
IJI = k, J 'f I.
if
Conclude that for
the form
A ••• A
1
wI:
(co)
=
satisfy:
a k-cochain and
For any
I
a ESp'
(iO, ... ,i
=
k)
as above (2.39) d)
a E Sk
Observe that for
the k-form on
is the simplest choice in order to satisfy the first identity of (2.18).
Show that with this choice for
a E Sk
the
condition (2.39) is a necessary requirement for the choice of for
a ESp' P
Exercise 2.
a)
Let
of simplicial sets. (L)
b)
1
=
f
: S
f*
0
E
(iii)
sk
0
f*
=
f*
sk'
0
Two simplicial maps
(i)
(ii)
be a simplicial map
f*
0
(ii)
O, •.. ,q,
S'
Show that
homotopic if for each i
k.
>
q
1,2, ..•
k
fO,f,
S
S'
there are functions
such that Oh O
=
fa,
q+1 hq
{h.J- 1c 1., , e h . = 1. J h j i-1'
=
f1
if
i < j,
if
i > j+1,
i
j+1 hj+1
=
j+1 hj'
are called hi
36
(iii)
Show that c) and b)
C*(S')
Let
f
O,f 1
imply that
d)
: S
i
> j.
are chain homotopic.
be homotopic.
Let
S
S
->
A*(S)
Show that a) are chain homotopic.
be a simplicial set.
ni
A k-form
is called normal if it furthermore satisfies
(iii)
on
if
Find explicit chain homotopies in c) .
on
(2.7).
< j,
fO,f, : A*(S')
Exercise 3.
where
i
C*(S)
->
S'
->
if
=
i
6 P+ 1
6P
->
Let
O, ••• ,p,
0
E S , P p
= 0,1,2, ..
is the i-th degeneracy map defined by
c Ak(S)
be the subset of normal k-forms
S. a)
f : S
->
Show that S'
d
and
preserve
A
is a simplicial map then
normal forms and if
f*
also preserves
normal forms. b) k
Show that the operators
0,1, ... , j
fl '!'h ( ')'
h(i)h(i) = 0,
(ii) Let
k-cochains
satisfy
J h (L) n·* J { n * h ( i _ 1 )' j
(i)
c)
O, ••• ,p,
=
Ck(S)
c
=
v
i = 0, ••• , k-1 . (L)
(ii) (iii)
h(j)
E
->
sk :
->
i
> j
be the set of normal cochains, i.e., c
niT
Show that ->
j
i = O, ... ,p.
such that
I
i
Ak- 1 (S) N
0
VT E Sk-1'
37
and conclude that
I :
is a chain equivalence.
Hence since the inclusion
C*(S)
is a chain
equivalence (see e.g. S. MacLane [18, Chapter 7, the inclusion
A*(S)
Exercise 4.
§
6] also
is a chain equivalence.
(D. Sullivan).
Let
m)
denote the
set of polynomial forms with rational coefficients, i.e.
w E
is the restriction of a k-form in
lli
n+1
of
the form
where
a.
.
are polynomials in
lO·· .lk
to ... t
n
with rational
coefficients. Now let S
S
be a simplicial set.
is called rational if
Ak(S, W) a)
E
A k-form m)
for
0
E S. p
Show that
multiplication
is a rational vector space
Let
C*(S, m)
(i)
(ii)
d
and exterior
A.
rational values.
denote the complex of cochains with
Show that
I E
C*(S,
A*(S,
and conclude that the Theorems 2.16 and 2.33 hold with and
C*(S) c)
Let
denote the set of rational k-forms.
which is closed under the exterior differential
b)
on
=
replaced by
A*(S)
and
Formulate and prove a normal version of question b)
(see Exercise 3). Note.
For a simplicial complex the construction of the
simplicial de Rham complex goes back to H. Whitney [35, Chapter 7].
3.
Connections in principal bundles
The theory of connections originates from the concept of "parallel translation" in a Riemannian manifold.
So for
motivation consider the tangent bundle TM of a differentiable manifold
M; or more generally a real vector bundle
M of dimension
over
p,q E M and a vector
Given points
n.
V
v E V P
one wants a concept of the corresponding "parallel" vector
E V
T(V)
i.e. we require an isomorphism T : V Vq. However, p q, V is a trivial bundle this seems to be an impossible
unless
requirement.
What is possible is something weaker: the concept
of parallel translation along a curve from suppose to
[a,b]
y
y(b)
q
p
to
q,
that is,
M is a differentiable curve from v E V
and let
y(a)
p
be a given vector; then a
p
\
i
"connection" will associate to these data a differentiable family with
v t E Vy(t)' t E l a b l , j
v
a
parallel translate a basis or frame
V.
space
Therefore let
P
over
11
{v ' ••• ,v
for the vector n} M denote the frame bundle 1
F(V)
:
M, i.e. the bundle whose fibre over
of all bases (frames) for associates to any curve a lift of rCa)
It is of course enough to
= v.
=
e
y and
through 11
0
Y
X E Te(F(V))
[a,b]
M and any point
e, that is, a curve
=
y.
is equal to the set
Then a "connection" simply
p
y
defines a tangent vector vector
V.
p
Now let
q
X E T (M) p
such that
11*X
y : [a,b]
tend to and =
y
X.
T
11(e)
(M)
p;
F(V)
then
with
y
defines a tangent So infinitessimally
a "connection" defines a "horizontal" subspace mapping isomorphically onto
e E F(V)y(a)
for every
He
Te(F(V)) e E F(V).
And
that is actually how we are going to define a connection formally below.
Notice that
F(V)
is the principal
Gl(n,JR )-bundle
39
associated to
V.
So first let us recall the fundamental facts
about principal G-bundles for any Lie group
COO
G.
Let
M
be a
manifold. Definition 3.1.
mapping
TI
:
E
M
A principal G-bundle is a differentiable of differentiable manifolds together with a
differentiable right G-action (i)
For every
P E M E
E x G
(Local triviality)
(ii)
neighbourhood
U
TI
P
-1
E
satisfying is an orbit.
(p)
Every point of
and a diffeomorphism
:
M TI
-1
has an open (U)
U x G,
such that (a)
the diagram TI-
1 (U)
U
x
G
U
commutes, (b)
is equivariant, i.e. .g,
=
where
G
acts trivially on
translation on E
e E
TI-
1
(U),
U
g E G,
and by right
G.
is called the total space,
M
the base space and
1 TI is onto TI(p) is the fibre at p. Notice that by (i) P and by (ii) it is an open mapping so TI induces a homeomorphism
E
to
M.
is free (i.e., xg
= x
of the orbit space G
on
E
given by
g
eg
E/G
Also observe that the action of g
=
1)
and the mapping
is a diffeomorphism for every
e E E
p.
G
E.
P
We shall
often refer to a principal G-bundle by just writing its total space
E
40
Example 1. bundle.
Suppose
V
Then the bundle
M
F(V)
is an n-dimensional vector M of n-frames is a principal
Gl (n, lR ) -bundle. Let
E
isomorphism morphism.
M
M
and
F
(j)
E
F
x
an isomorphism mapping
(j)
G
be two principal G-bundles. Then an
M
is a G-equivariant fibre preserving diffeo-
is of course a trivial principal G-bundle and
(j)
E
M
x
is called a trivialization.
G
in (ii) above is called a local trivialization.
Now consider a principal G-bundle covering (j)
a
The
U
= {Ua}aEL
1
: n- (U ) a (j)S
of
U x G. a 0
n
E
:
M
and choose a
together with trivializations
M
Then if
*¢
U n Us a
consider
-1
(j)a
which is easily seen to be of the form (j)S
0
-1
(j)a (p,a)
gSa: U n Us G is a COO function. This system {gsa} a are called the transition functions for E with respect to U
where
and they clearly satisfy the cocycle condition (3.2)
gSa (p) g
=
aa
gya (p),
=
1.
On the other hand given a covering
=
and a system of a transition functions satisfying (3.2) one can construct a U
{U }
corresponding principal G-bundle as follows: the total space is the quotient space of (p,a) E Ua
liU a a x
G
x
G
with the identifications with
(p,gSa(p) ·a) E Us
Vp E Ua n US' a E G.
x
G
41
Again let f
: N
M be a principal G-bundle and let
M be a differentiable map. f*E
f*n
n: E
N
The "pull-back"
is the principal G-bundle with total space
and
1T (e) }
{{q,e) If (q )
f*E f*1T
given by the restriction of the projection
onto the first factor.
The projection onto the second factor
giveg an equivariant map
f : f*E
l
E
covering
f, i.e. the
f*{E) f* 1T
f
N -----'''---.. M
commutes. Exercise 1.
a)
Show that if
transition functions for U = {Pa}aEl: for
f*E b)
then
f.
0
f}
Let
F
N, E
M
(f,f),
f: F
E
where
is the set of
relative to the covering is the set of transition functions
relative to the covering
map is a pair map and
{gaB
E
{gaB}
-1 f-1 U = {f Ua}aEl:.
be principal G-bundles. f
: N
M
A bundle
is a differentiable
is an equivariant differentiable map covering
Show that any bundle map factorizes into an isomorphism : F
f*E
and the canonical bundle map
Exercise 2.
a)
f*{E)
E
Show that a principal G-bundle
as above. 1T : E
M
is trivial iff it has a section, i.e. a differentiable map s : M b) 1T*E
E
such that Let
1T
E
1T
0
sid.
M be a principal G-bundle.
Show that
is trivial. c)
Let
n : E
M be a principal G-bundle and let
H
G
42
be a closed sUbgroup.
Show that
E
E/H
is a principal H-bundle.
(Hint: First construct local sections of the bundle
G
G/H
using the exponential map). Let
Exercise 3. and let
N
: E
M
be a manifold with a left G-action
associated fibre bundle with fibre : EN
M where
EN
under the G-action and where Show that
be a principal G-bundle
= E
x
(e,x) 0g
G
N
N
N.
is the orbit space of
N
The
E x N
= (eg,g-1 x), e E E, x E N, g E G,
is induced by the projection on EN
x
is the mapping
followed by
E
is a manifold and that the fibre bundle is locally
trivial in the sense that every point of
u
G
with a diffeomorphism
:
-1
(U)
M
U
x
N
has a neighbourhood such that the
diagram -1
(U)
U x N
U
commutes.
In particular
Now let
Hand
G
is open and differentiable. be two Lie-groups and let
be a homomorphism of Lie groups. principal H-bundle and
E
Suppose M
p
= Ep'
Vp EM,
=
'a(h),
Then we will say that to
a
E
"relative to
a
F
M
is a
E
satisfying
and "Ix E F, hE H. to
G
relative
is a reduction of
E
to
(when it is clear what a").
F
is an extension of
or, equivalently, that
relative to
: F
G
is a principal G-bundle and
suppose there is a differentiable map (F ) c
a: H
a
F
is we will omit
H
43
Example 2. principal
An n-dimensional vector bundle
Gl (N, lR) - bundle n lR
act on the left on n lR
with fibre
F (V)
M.
FO(V)
is just the vector bundle. Hence there is a
F(V) F(V)
fibre. Then
Gl (n, lR) - bundles
g E G. 'IT
G
:
G
O(n).
In fact
M is the corresponding orthogonal bundle FO(V)
Riemannian metric on
consider
defines a
consist of the orthonormal frames in each
F(V)
versely a reduction of
Exercise 4.
V
M to the orthogonal group
FO(V)
and the inclusion
Gl (n, lR)
and that the associated fibre bundle
and vector bundles. A Riemannian metric on
let
M has the
Notice that
one-to-one correspondance between principal
reduction of
V
F(V)
defines the reduction.
to
O(n)
clearly gives rise to a
F
M be a principal H-bundle and
V.
a) Let
'IT
:
with the left H-action given by
.
h
g
=
a(h)g, h E H,
Show that the associated fibre bundle with fibre
FG
Con-
M is a G-extension of
'IT
:
and show that an
M,
F
G,
extension is unique. b)
Show that a principal G-bundle
reduction to
u = {u} y {a
0
h
H
relative to
a
'IT
:
E
iff there is a covering
and a set of transition functions for
} with SY
{h
M has a
E
of the form
} a set of functions satisfying SY
(3.2)
Before we introduce the notion of a connection in a principal bundle it is convenient to consider differential forms with coefficients in a vector space. So let and
w
V on
M be a
COO
manifold
a finite dimensional vectorspace. A differential form M of degree
function
k
w(X1' ••. 'X k)
with values in : M
V
V
associates a
to every set of
COO
COO
vector
44
fields
Xl, ••. ,X k
on
w
M;
is again multilinear and alter-
nating and has the "tensor property" as before. a basis
{e 1, ... ,e
n}
w = w1 e 1 + ... + wne n forms.
Let
values in
V
then
where
Ak{M,V) V.
for
Again
w
(w , •• • ,w l
If we choose
is of the form is a set of usual k-
n)
denote the set of kforms on A*{M,V)
M with
has an exterior differential
defined by the same formula as in Chapter 1 and chain complex (that is, dd = 0).
A*{M,V)
d
is a
This time, however, the wedge
product is a map
for w
2
V,W
two vectorspaces.
E Al{M,W)
define
where as usual
w
1
In fact for
A W
2
w
l
E Ak+l(M,V 0 W)
E Ak(M,V)
by
runs through all permutations of
a
and
1, ... ,k+l.
Again we have the formula (3.4)
(dw 1) w
Similarly for
F: M
an induced map
w2
k
+ (-1) w1
k
1
N
E A (M,V), w
a
A*(N,V)
F*
A
COO
Now let
d
map of A*{M,V).
and induced maps
E A (M,W). 00
C
manifolds we have
P: V
Also if
P: A*{M,V) F*
W is
A*{M,W)
as above. G
as usual the set of leftinvariant vector fields on
G.
This
can also be identified with the tangent space of
at the unit
element
1 E G.
For
g E G
let
The Lie algebra
is
of
G
be a Lie group.
dw 2 '
1
2
a linear map it clearly induces a map commuting with
A
Ad (g)
:,
1
G
be the adjoint
45
representation, i.e., the differential at x .... gxg
,
Now let
with
E .... M
:
IT
G .... E
the map
vx
of the map
-1
given by
be a principal G-bundle. 9
....
For
induces an injection
x'g
and the quotient space is naturally identified
.... Tx(E)
That is, we have an exact sequence
TlT(x) (M).
v
IT*
2...,
(3.5)
--->
The vectors in the image of
v
T
lT (x) (M)
--->
O.
are called vertical and we want
x
to single out a complement in
Tx(E)
of horizontal vectors,
i.e., we want to split the exact sequence (3.5).
This of course such that
is equivalent to a linear map (3.6)
id
It is therefore natural to define a connection in 1 8 E: A
be a 1-form
E = M
consider the trivial bundle 1-form on
E
8 (x,g)
Z
left translation by
action on
E
= M
To motivate this
8
g.
Now for
G,
x
9 E: G 9
let
L R
9 9
:
-1 :
E
be the
For
by the right action on
R*8 9
Ad (g-1)
0
:
8
G
A 1 (E'j)
0
8,
....
E
Vg E G,
1 .... A (E'1)
is
denote E,
and the trivial
defined by (3.7) we have
Ad (g -1)
G .... G
on the principal G-bundle
M.
Lemma 3.8.
where
x E: E.
x E: M, 9 E: G,
is the projection and
the map given by the action of i.e. for
8.
G .... M and let
x
OlT z ) * '
=(L g-l
M x G .... G
:
simply to
given by
(3.7)
lT
E
such that (3.6) holds for all
However, we want a further condition on
where
x E: E
is induced by
46
Ad(g-l ) Proof.
e
Since
to consider
= pt.
M
from
is induced via That is,
e
G
it is enough
is the 1-form on
G
defined
by
ey Then (R*e) g
y
With this motivation we have Definition 3.9. n
E
M ( L)
A connection in a principal G-bundle
is a 1-form ex
0
V
x
=
1 e E A
satisfying: is the
where
id
differential of the map (El
R*e g
= Ad(g-l)
where on
g
xg.
Vg E G, is given by the action of
g
E. If
vectors, i.e. R *H
e, E
Remark 1.
(E)'
0
g
H
H x
x
x
Tx(E)
is the subspace of horizontal
then (ii) Vx E E,
In fact (ii) clearly implies (ii)
is equivalent to
Vg E G.
I
and since both sides of (ii)
vanish on horizontal vectors (granted (ii) ')
it is enough to
check (ii) on vertical vectors in which case (ii) from (i) and Lemma 3.8.
is obvious
47
Remark 2.
By Lemma 3.8 the product bundle
has a connection given by (3.7).
E
F
M
M
x
G.
Notice
is an isomorphism of G-bundles and if
8
has a connection
G
x
This is called the flat
connection or the Maurer-Cartan connection of that if
M
then
E
defines a connection in
In particular every trivial bundle has a connection induced from the flat connection in the product bundle.
This is also
called the flat connection induced by the given trivialization. The following proposition is obvious. Proposition 3.10.
Any convex combination of connections
is again a connection.
More precisely:
connections in
E
M
M with
LiA
functions on
TI
connection in
:
paracompact manifold
connection.
i
=
1.
A , ••• ,A 1
Then
8
=
8
k
1,
. . . ,8
k
be
be realvalued
Li Ai 8 i
is again a
E.
Corollary 3.11.
Proof.
and let
Let
Any principal G-bundle M
TI
E
M
on a
has a connection.
By Remark 2 above every trivial bundle has a flat In general local trivializations define
flat
a covering of M. a in EIU a for {Ua}aEZ choose a partition of unity {A } and put 8 = LAa 8 a • It a a follows from Proposition 3.10 that 8 is a connection. connections
8
Exercise 5.
a)
Now
Suppose we have a bundle map of principal
G-bundles
I
F ----+ E
1
If
E
has a connection
1
8
then
I*8
defines a connection in
F.
48
b)
If
E
M is a trivial G-bundle then there is a bundle
map
and the flat connection is just the induced connection of the Maurer-Cartan connection in the G-bundle
G
Now consider a principal G-bundle
e.
For
X E Tx(E)
X
E H
x
X
ex
ker
=
E
:'1
im v x' v x
Now suppose
w
with coefficients in some vectors pace is horizontal if
: E
M with connection
a tangent vector we have already introduced
the term vertical for for
pt.
w(X
1,
... ,X
= 0
k)
A* (E,V)
E
and horizontal
Tx(E),
is a k-form
We will say that
V.
w
whenever just one of the
is vertical. I f V is a (left) X1,··· ,X k E Tx(E) representation of G then we will say that w is equivariant -1 if R*w w, Vg E G. In particular i f V is the trivial g = g
vectors
representation an equivariant form is called invariant. that the invariant horizontal forms on
E
with coefficients in
are exactly the forms in the image of In fact suppose define
w
E
choose
x
E
=
Xi' i
w E A*(E)
Ak(M) -1
=
(p)
1, ... ,k
: A*(M)
For
X1 , ... ,X k
p E
E
M and
Tx(E)
Furthermore if
COO
X 1,.",Xk
are extended to
=
Xi'
so
to
COO
w(X 1 , ... ,Xk )
and
P
X1, ... ,X k.
vector fields
X1"."Xk vector fields satisfying
M we can by local triviality of x
T (M)
E
and put
x
a neighbourhood of
X1 , · · · 'X k
such that
This is then independent of the choices of
on
A*(E).
is horizontal and invariant; then we
as follows: and
Notice
is
COO
E
extend
in a neighbourhood of
in
x.
49
Now consider the connection from that
8
is an equivariant 1-form with coefficients in
the adjoint action of the image of
8
A
8
G.
[-, -]
Proposition 3.12. connection
8.
E
Let
: E Q E
M
x
G
with the flat
be the curvature form defined by d8
(the structural equation). Furthermore
(3.15)
Q
dQ
In particular Proof.
drl
Then
Q
is horizontal and equivariant.
satisfies the Bianchi identity [rl ,8] .
vanishes on sets of horizontal vectors.
a) follows from b) since by Exercise 5
induced from the principal G-bundle
= 0
Then we have:
M be a principal G-bundle with connection
(3.14 )
Q
1
d8
and let
c)
2(E" A
:..., ®
Let
denote
Then
(3.13 )
8
a)
with
[8,8] E
Also let
under the map
induced by the bracket
b)
Observe
G
pt
e
is
and therefore
because it is horizontal by b) . b)
It is obvious that
hence both
d8
and
observe that clearly To see that for any (3.16)
rI
X,Y E Tx(E) (d8) (X,Y)
(e,8]
rI
is equivariant since
and
are equivariant (for the second one
1-
Ad (g)
preserves the Lie bracket) .
is horizontal we must show for with
8
X
x E E
and
vertical that
-\(8,8](X,Y)
-\(8
(X),8 (Y)].
In order to show (3.16) it is enough to consider 1)
Y
vertical
50 and 2)
Y
1)
horizontal.
First notice that for any vector
COO
associated
vector field
where
A*
on
E
A E
JJjr
there is an
= Vx(A)
defined by
as usual is induced by
g
xg.
Observe
that the associated 1-parameter group of diffeomorphisms is {R
},tElR,
gt see that for
where A,B
gt
= exp tA, t
Also it is easy to
E lR.
E'1 [A,B]*
(3.17)
[A*,B*].
In fact by local triviality it is enough to prove this for a trivial G-bundle A
= M
E
x
G
in which case
is the left invariant vector field on
Therefore (3.17) bracket in
1.
A* G
A
= 0
where
associated to
A.
is immidiate from the definition of the Lie
Now, to prove (3.16)
for
X
and
Y
vertical it is clearly
enough to prove (d8) (A*,B*) But since
8 (A*)
A,
(d8) (A* ,B*)
8 (A*),8 (B*)], 8 (B*)
B
A,B E
1- .
are constants we conclude
-'-,8 ( [A*,B*]) = -'-,8 ( [A,B]*) -'-,[A,B] = -'-, [8(A*),8 (B*)].
2)
Again extend Also for
Y
X
to a vector field of the form
horizontal extend it to a horizontal
vector field also denoted by vector field Since
Y
Z
and then put
Y
Y
(first extend
Y
Y
=
Z
Y
- v
y
0
to any
00
C
00
C
8 (Z ), Y E E). Y
Y
is horizontal the right hand side of (3.16) vanishes.
So we must show (3.18)
A*,
(d8) (A*,Y)
o
for
Y
a horizontal
vector field.
51
Now since
e (A*)
A
e(y) = 0
is constant and (de)(A*,Y) =
As remarked in 1)
the 1-parameter group associated to
A*
is
Therefore g
lim 2. (Y t_ y ) t .... O t x x
where
Since g
e (Y x t)
1
Ad (g
)
0
e (Y
xg
0,
-1 ) t
we conclude
which proves (3.18) and hence proves b). c)
Differentiating (3.14) we get
o
+
-
[ de, e] =
-
e] +
[ [e ,e] , e]
-
since
[ [e , e] , e]
o
by the Jacobi identity.
This proves the
proposition. Remark.
Let
X, Y
be horizontal vector fields on
E.
Then
by (3.14) (3.19 )
O(X,Y) =
which gives another way of defining Definition 3.20.
A connection
e
in a principal G-bundle
is called flat if the curvature form vanishes, that is,
= O.
52
Theorem 3.21. IT
:
E
M
A connection
in a principal G-bundle
is flat iff around every point of
neighbourhood
U
restriction of
and a trivialization of to
6
connection in
U
Proof.
EIU
M
EIU
there is a such that the
is induced from the flat
G.
x
is obvious by Proposition 3.12 a).
n
Suppose
q:
6
O.
=
For
x E E
let
subspace of horizontal vectors, i.e.
tiable subbundle of
T(E».
x
X E H
x
E
This clearly defines a distribution on
Tx(E)
H
iff
be the
6 (X)
=
O.
(i.e. a differen-
By (3.19) this is an integrable
distribution hence by Frobenius' integrability theorem defines a foliation (see e.g. M. Spivak [29, Chapter 6]) such that is the tangent space to the leaf through Remark
following Definition 3.9 that
x. R
It follows from E
g
Hx
E, g E G,
maps any leaf diffeomorphically onto some (possibly different) leaf of the foliation. Now let leaf IT
through
Hx
x
U IT
F
of :
p E M
Tp(M) p
x.
Since
x E
Tx(F)
V
is a diffeomorphism.
:
IT-
H X
of
x
Elu;
U
x
G
where
:
U x G
be the connection in
6'
connection in
so
and consider the
and since
in
F
such that s : U
V
Elu
In fact the trivialization is given by
1 (U)
subspace in
(p)
hence by exercise 2
s(q)·g, Now let
-1
The inverse
therefore defines a section of is trivial.
IT
is an isomorphism we can find a neighbourhood
and a neighbourhood U
V
and choose
U
x
G.
and
6'
-1
(U)
is defined by
qEU,gEG.
Elu
induced from the flat
Then it is obvious that the horizontal
T (E), Y E V, g E G, y.g 6
IT
is
(R )*(T (V») g Y
=
R *H g Y
defines the same horizontal subspaces
53
and therefore must agree. Corollary 3.22.
Let
IT
:
E
M be a principal G-bundle.
The following are equivalent: 1)
E
has a connection with vanishing curvature.
2)
There is a covering of
a set of transition functions gaS
Ua n Us 3)
Then
E
{gaS}
is constant for all
Let
G be the group G d has a reduction to G d,
Proof, 2)
G
M by open sets for
E
{Ua}aEL
such that
a,S E L.
with the discrete topology.
2) and 3) are equivalent by Exercise 4. be the
Let
=> 1):
trivializations with the constant transition functions Let
and
gaS'
8
be the connection in E.I Va induced from the flat a connection in U x G. Now there is a commutative diagram of a bundle maps
F
• U n Us a
G
IT
1
8
definition
z
be the Maurer-Cartan connection in
0 8
0
G
• G
G
and let
x
G
pt.
By
is left invariant and therefJre
Therefore and 8 agree on EIU n US' a a S we can define a global connection 8 in E which agree with or equivalently
8 a.
on
Clearly
EIU a·
has for all
8
a..
8
has vanishing curvature since
8
a
54
1)
2):
curvature.
Now let
8
be a connection in
By Theorem 3.21 we can cover
and find trivializations
U .... U
M
is induced from the flat connection in
with vanishing
{Ua} 1U such that Slna x G. Now fix a,S E L
x G
a
E
U a
by open sets
and let
Again let
8
U n Us a
be the flat connection in
0
x
Then
G.
(jJ*SO = 8 0 so (jJ permutes the leaves of the horizontal foliation, i.e. , the sets of the form In (U n US) x g, g E G. a
clearly
(jJ(U n Us a and it follows that
particular
1) = (U
x
a
n
US)
Hence the transition function Exercise 6. and let
Let
x
go
go E G,
for some
is constantly equal to
a : H .... G
be a Lie yroup homomorphism
F .... M be a principal H-bundle with connection
Show that if
(jJ
F .... E
is the extension to
a connection
8E
in
such that
E
(jJ*8
E
G
= a*
0
SF'
then 8
F,
is where
a*
is the induced map of Lie algebras. Exercise 7.
Let
M be a manifold and let
be the frame bundle of the tangent bundle, projection.
...1 (n,m) -r
=
The structure group is n Hom(m n ,m)
isomorphism
Since
x : m n .... T (M) p
with coefficients in
JRn
x
Show that
1-form, where
w on
GI(n,m)
-1
x E
1T
F(M) .... M
n
the
with Lie algebra
(p), P E M,
there is a 1-form
w
is an on
F(M)
defined by W
a)
Gl(n,m)
F(M) = F(TM)
x
-1
F(M) acts on
0
11*.
is a horizontal equivariant m n
by the usual action.
55
b)
For
M=lR
n
F (lR n)
connection in
and for
8EA
1(F(M)),#n,lR))
the
defined by the natural trivialization
show that dw
-8
II
w
where the wedge-product denotes the composite map
(Hint:
Notice that
.coordinates a real
n
x
For
F (lR n) = lRn
x
Gl (n , lR)
lRn
x
JRn
2
wi th
n
y= (Y1""'Yn) E lR n-matrix.
and X={x·'}"1 lJ l,J= , ... ,n 1dX and w = x- 1 d y ) . 8 = X-
Then
M a general manifold and
show that the torsion-form (3.23)
8
a connection in
S E A2(F(M) ,lRn) -8
dw
II
F(M)
defined by
w+ S
is equivariant and horizontal. c)
where
With respect to the canonical basis of
1
w , ••• ,w
e
n
are usual 1-forms on
·· ··· ·'n
='
8
1,
... .. ..
on
••••••••. en
Then (3.23) takes the form
F(M).
([)
lRn
we write
Similarly we write
56
(3.23)
d)
Show that every horizontal 1-form
the form on
i Lifi w
Ci =
,
where
on
Ci
00
are real valued
f.
1
C
functions
M is given a Riemannian metric and let
Now suppose
be a connection in the orthogonal frame bundle
wand
S
be defined on
FO(M)
exactly as for
Show that (3.23) still holds and that on (3.24)
i,j
J
Furthermore show that if by (3.23) and (3.24).
S
=
F(M)
write
Ci
j
f)
=
Lifijwi
FO(M)
e
is uniquely determined
(Hint: Show first that if
Ci =
and if we
a)
Let
F(M).
M be a manifold and Let
K
:
V .... M
an n-
F .... M be the associated
= Hom(mn, m n)
Show that for
e
is the Lie algebra of
E A1 (F,,t(n,ml),
e
(3.14) takes the form
(3.25)
M the
Gl(n,m) -bundle, i.e. the bundle of n-frames in
,l(n, m)
is a
Notice that by Exercise 6 this extends
dimensional vector bundle.
Again
J
has a unique torsion free connection (the
to a well-defined connection in
principal
(Ci.)
as in d), then
Levi-Civi ta connection).
Exercise 8.
above.
FO(M)
Conclude that for every Riemannian manifold
framebundle
Let
1, ... , n.
then
0
FO(M).
row of horizontal 1-forms satisfying
F,
is of
F(H)
F(H) . e)
e
0, ... ,n.
i
I
d8
-8
II
e
+
n
where the wedge-product denotes the composite
V.
GI (n, m)
a connection in
57
{here "(n, JR) of maps of
@
JRn
l'
(n, JR) ... "(n, lR) JRn).
into
canonical basis of
is given by composition
Furthermore, with respect to the
e
?(n,JR),
Si
and
are given by matrices
e11 •••••••••• en1
Si 1··········Sin
en1·
Si 1··········Sin
n en
1
1
n
n
of 1- and 2-forms respectively. Show that (3.25) is equivalent to (3. 25)
I
-1:
]
b)
i, j
k
Observe that
COO
sections of
correspondence with equivariant where
Gl (n, JR)
sections of
acts on
V
C
00
V
1, •..
are in 1-1
Similarly show that
into
n lR
The set of
COO
functions of
JRn in the usual way.
is denoted
,n.
F
r{V). C
00
T*M @ V
sections of
are in 1-1
correspondence with equivariant horizontal 1-forms on coefficients in
JRn.
Alternatively
to every vector
t E r{T*M
an element
E V P
(i) (ii)
@
V)
F
with
associates
such that
A E JR,
if
X
is a
p
tx
P
is a
COO
vector field on
COO
section of
V.
M
then the function
58
c)
Let again
6
be a connection in
V(s) E A (F , lli n ) 1
define
that
is considered as a function of
V(s)
s E r(V)
-6-s + V(s)
ds s
For any
by
(3.26)
(here
F.
F
into
lli
n).
Show
is horizontal and equivariant, hence defines
V(s) E r(T*M ® V). d)
s E r (V)
For
and
X
P
V (s) = V(s)x x
as in c) and let
P
P
V
V(s) E r(T*M ® V)
as defined in b) _
E V P
P
is called the covariant derivative of and
let
E T (M)
s
in the direction
V
P
V (s) ilXp
P
ilV
=
sEr(M), (ii)
If
X
is a
function
p
....
V x
a
C
(s )
P
is a
s
y : [a,b]
6 M a
is a unique liftet curve
Y
=
Y,
(s), p
ilElli.
then the
section of
for
real valued function on
As before let
M
x
V.
Vx(s) .
directional derivative of
that for
C
00
X(f);x(S) + fVx(s)
Vx(fs) 00
vector field on
COO
This is denoted (iii)
6.
satisfies:
Vx +y (s)
(i)
IT 0
Xp
is called the covariant differential corresponding to
Show that
e)
This
s E r (V) M
curve and
y : [a,b]
such that the tangents of
X(f)
and
the
f.
be a connection in COO
f
I
F
IT
:
x E IT-
1
F
M.
(y(a))
Show there
= x,
with
yare all horizontal.
Notice that this lift defines an isomorphism (the "parallel translation along
y")
t E
[a,b] .
59
f) C
00
For
X
E T
p
p
(M)
let
y (0) = p,
curve with
y
y' (0)
be parallel translation along
(3.27)
T
'V X (s)
t
M,
P
Let
T
lift of
v y.
of
Flu
.... V Y(t)
s E r (V)
s(y(t»-s(p)
such that
v
P
t .... O
l
i a.
and
v
0
L a.v.
s =
Now write
the components of
V
t
Show that for
{Hint: Observe that in some neighbourhood a section
be a
s > 0
lim - = - - - : t - - - - -
=
p
X
y. -1
....
[-s,d
:
l
.... lR,
of
p
there is
defines a horizontal
y
where
l
U
U
are
(v 1'··· ,v n)
are
i = 1, ... ,n,
C
00
functions) . g) 8.
2(F'7(n,lR» rl E A
Now let
s E r(V),
Show that for any
function of
F
into
(3.28)
X
and
s
Y
-
rl (X, Y)
(3.29) h)
9.
V
TM
=
'V
(s) •
M
rl
defines a
Show that
Let
8
and let
TM,
w
be the 1-form considered
be a connection in
Observe that for
a section of
A
(s)
Now let
in Exercise 7.
8
vector fields on
rl(X,Y) E r(Hom(V,V».
section
interpreted as an equivariant
we have
rl •
d'V(s)
Notice that for
form
lRn,
be the curvature form of
X, Y
F(M)
with torsion
vector fields on
that is, a new vector field
M
'9(X,Y)
e
defines and show
that this is given by (3.30) where
8(X,Y) 'V
-
'Vy(X)
-
[X,Y])
is defined in d) •
(Hint: Notice first that for any vector field
X on
F(M)
60
which is a lift of a vector field
TI*Xx = XTIX , Vx E F(M))
X
the function
on
(that is,
w(X)
the equivariant function corresponding to Note.
M
is X
as in b)
above.
Our treatment of principal bundles and connections
follows closely the exposition by S. Kobayashi and K. Nomizu [17, Chapter I and II].
4.
The Chern-Weil homomorphism
We now come to the main object of these lectures, namely to construct characteristic cohomology classes for principal G-bundles by means of a connection. Let k
V
let
First some notation:
be a finite dimensional real vector space. Sk{v*)
denote the vector space of symmetric
multilinear real valued functions in Equivalently
For
P E Sk{v*)
k
variables on
is a linear map
V.
P : V ®... ® V
lli
which is invariant under the action of the symmetric group V ®... @ V.
acting on
There is a product
o
defined by
P
(4.1)
Q{v 1 , ... ,v + l ) = k
0
1l)
= (k+ a
where S* (V*)
LoP (v 01'
.•. ,v o k)
Q{v O {k + 1 ) ' " .,vo{k+l))
1, ... ,k+l.
runs through all permutations of
II
(SO (V*) = lli);
Sk (V*)
then
S* (V*)
Let
is a graded
algebra. Exercise 1. lli [x ' ••• ,x 1 n]
k
k
Let
{e
1,
... ,e
n}
be a basis for
V
and let
be the set of homogeneous polynomials of degree
in some variables
Show that the mapping
defined by P{v, ... ,v),
v
62
for
P E Sk(v*), s*(v*)
-->
shows that given by
JR[x , ••• ,x 1
P v
is an isomorphism and that is an algebra isomorphism.
n]
is determined by the P(v, ... ,v).
This
function on
V
is called
The inverse of
polarization. Now let
be a Lie group with Lie algebra
G
the adjoint representation induces an action of for every
k: P (Ad (g
-1
) v 1 ' ••• , Ad (g
-1
v 1,···,vk E1' Let
Ik(G)
G
) v k) ,
E
g
G.
be the G-invariant part of
Notice that
the multiplication (4.1) induces a multiplication (4.2)
I*(G)
In view of Exercise
is called the algebra of invariant
polynomials on Now consider a principal G-bundle differentiable manifold E
with curvature form
M,
and suppose
Q E A2 (E'1)'
llk =llA ... I\QE 2k A (E'1@" so Q
P E Ik(G)
gives rise to a 2k-form
is horizontal also
equivariant and
P
2k-form.
p(Qk)
Hence
also denote by
p(Qk)
invariant
8
-->
M
on a
is a connection in
Then for
we have
k
2k(E'1®k) = A p(Qk) E A2k(E) .
is horizontal, and since p(Qk)
Since Q
is
is an invariant horizontal
is the lift of a 2k-form on
M which we
p(Qk).
Theorem 4.3. Let
.@')
E
n
a)
p(Qk) E A2k(M)
2k(A*(M)) WE(P) E H
is a closed form.
be the corresponding cohomology
63
class. b}
Then WE{P}
does not depend on the choice of connection
and in particular does only depend on the isomorphism class of
E. c}
WE: I*{G}
H{A*{M}}
d}
For
M a differentiable map
Remark. phism.
f
: N
The map
WE
is an algebra homomorphism.
is called the Chern-Wei 1 homomor-
Sometimes we shall just denote it by
W when the
bundle in question is clear from the context. WE{P} to
is called the characteristic class of
P E I*{G}
For E
corresponding
P. Proof of Theorem 4.3.
a}
Since
TI* : A*{M}
= 0
injective it is enough to show that Now since
P
is symmetric and
A*(E) in
is
A*{E}.
a 2-form
{4.4} by {3.15}.
On the other hand since
P E
is invariant
we have {4.5}
P {Ad (g t) y 1
gt
=
I
••• I
Ad (g t) y k } = P (Y 1
exptyo'
Differentiating (4.5) at
t
YO'Y1""
o
I
••• I
'Y k E
1'
we get
o or equivalently 0,
Yk)
I
t E JR.
64
From this it follows that
A
o
A ••• A
which
together with (4.4) ends the proof of a). b)
For this we need the following easy lemma (compare
Chapter 1, Exercise 5 or Lemma 1.2); Lemma 4.6.
Let
h : Ak(M
w
be the operator sending
J1
=
h(w)
s=O
a
--+
Ak- 1 (M),
k
0,1 , ... ,
S to
A
a +
(hw
o
ds
=
[0,1])
x
for
w E A0) .
Then
w E A*(M
dh(w) + h(dw) = iiw - iOw,
(4.7)
where
i 0 (p)
= (p, 0),
eO
Now suppose
= (p, 1 ),
L 1 (p)
and
curvature forms
e1
and
principal G-bundle e E A1 (E x [0,1])
are two connections in
--+
M
x
[ 0,1 ]
and let
be the form given by
By Proposition 3.10
e
be the curvature form of
P(D
k)
with
E
Consider the
(x,s)
obvious that
[0,1])
P EM.
respectively.
E x [0,1]
x
is a connection in
e.
Since
o =
and iiD 0 is a closed 2k-form on E i*D
i(;e Q1.
x
[0,1]
=
E E
x
[0,1].
E x [0,1].
eO'
Now for by a)
i*e = 1
Let
e1
D
it is
PErk (G) , above.
There-
fore by (4.7)
and hence in
H
2k(A*(M)).
and
represent the same cohomology class
This shows that
WE(P)
does not depend on the
65
choice of connectiQn.
The second statement is obvious from
this. c)
For
P E Il(G)
and
Q E Ik(G)
it is straight forward
to verify that (4.8)
from which c) d)
If
trivially follows. e
is a connection in
then clearly
I*e
curvature form
E
is a connection in
M with curvature form
f*E
N
with
Therefore since
(4.9)
d) clearly follows. Remark.
Let
be the algebra of complex valued
G-invariant polynomials on bundle
E
with connection
Then for any principal Ge
we get a similar complex Chern-
Weil homomorphism (G)
(4.10)
H(A* (M,lC»
!O
0
and we can clearly find an open nO (to, ... ,t ) such that V c int(tJ ).
of
nO
Define
and let
h
nO
coordinate of Now let
TI G
nO nO+ 1
n > nO
G
be the map which project onto the first
and suppose we have defined an invariant
74
open set
U EG(n-1} and an equivariant map h U ... G. n- 1 n- 1 n- 1 n n 1 p : 6 x G + ... EG(n} be the natural projection and
Let
observe that W
d6 n
h' W.
G
Shrinking
Now consider
since
be the closed subset
where
W"
w"
Un
h
n- 1 1 6 n x Gn +
W'
W'
6
n
Gn + 1
x
(U n 1)
Now let
U·
Clearly
h"
and
h"
: Un ... G. and
and def-ine
W'"
extends
clearly
U = U Un n
extends to a map
h'
W'.
defined on
W"
On the other hand we
W"' c 6 -1
W
n
n 1
x G
+
such that
(U -
n 1)
is an open G-invariant subset.
= W" n
p(U'}
n
Then
defined by
W is G-invariant.
c
Un = Un- 1 U
h
(Un- 1).
is an open G-invariant set and notice that and hence
-1
-1 -
W ... G
a little we can assume c
Let
is an open neighbourhood of
P
P
P
W p
0
can find a G-invariant open subset
since
EG(n-1}.
into
is an ANR the map
: W' ... G
Clearly
maps
ef+1
x
since
p
h
n- 1
0
h" : U' ... G
p : W ... G
and is equivariant.
is an open invariant set in h
n- 1
EG(n}
and
defines an equivariant extension
This construct h = U h n. n
by
Un
and
h
n
inductively, so let
This ends the proof of the proposition.
We can now state the main result of this chapter: Theorem 5.5. c
The map associating to a characteristic class
for principal G-bundles the element
1-1 correspondence.
c(E(G}} E H*(BG}
is a
75
For the proof we shall study
EG
and
BG
from a
"simplicial" point of view: Let
X = {X }, q = 0,1, ... , q
suppose that each
X
be a simplicial set and
is a topological space such that all
q
face and degeneracy operators are continuous.
Then
X
is
called a simplicial space and associated to this is the socalled fat realization, the space
II X" given by
II X II
with the identifications ( E i t ,x )
(5.6)
(t , E i X) ,
tEALJ n-l, x E X n'
n
Remark 1.
0, ...
1..
=
,n,
1,2, ...
It is common furthermore to require 0, ...
(5.7)
n
=
,n,
0,1, ...
The resulting space is called the geometric realization and is I X I.
denoted by
One can show that the natural map
II X II .... I X I
is a homotopy equivalence under suitable conditions. Remark 2. Example 1. consider
X
Notice that both If
X
=
{X}
II' II
and
I· I
are functors.
is a simplicial set then we can
q
as a simplicial space with the discrete topology.
The name "geometric realization" for the space
Ixi
originates
from this case. Example 2.
Let
X
the simplicial space with
be a topological space and let NX
q
=
X
NX
be
and all face and degeneracy
76
operators equal to the identity. II NX II = II N (pt) II
Then
x
INXI
and
where
x X,
with the apropriate identifications. Example 3.
Let
G
be a Lie group (or more generally any
topological group) and consider the following two simplicial spaces
NG
(Here
and
NG(O) In
NG: NG(q)
G
x ••• x
G
(q+l-times),
NG(q)
G
x ••• x
G
(q-times).
consists of one element, namely the empty a-tuple !).
NG
NG(q)
NG(q-l )
NG(q)
and
->
NG(q+l)
are given by
0, .•. ,q.
i
Similarly in
NG
£i (gl""
£i : NG(q)
,gq) =
->
NG(q-l)
is given by
(g 2 ' ... ,gq) ,
i
o
(gl"" ,gi gi+l"" ,gq)'
i
1, •.. ,q-l
(g l' ... ,gq-l) ,
i
q
{
and
n
i
: NG(q)
->
n· (gl,···,g ) .i, q By definition map
y
: NG
->
EG
NG
NG(q+l)
by
= (gl,···,g· 1,1,g., ,... ,g q ),
i
= O, ... ,q.
= II NG [I and if we consider the simplicial given by
77
(5.8) it is easy to see that there is a commutative diagram II NG II
111 y
II
- -......·IINGII such that the bottom horizontal map is a homeomorphism. will therefore identify
BG with
The simplicial spaces
NG
and
with
and
II NG II NG
We II y II.
above are special cases
of the following: Example 4.
Let
C
be a topological category, i.e. a
"small" category such that the set of objects set of morphisms
Ob (C)
and the
are topological spaces and such
that (i)
Ob(C)
The "source" and "target" maps
are
continuous. (ii)
"Composition":
->
is continuous
where
x
consists of the
pairs of composable morphisms (i.e. (f,f') E Associated to nerve of NC(2)
=
C
C where
source (f)
=
target (f'».
there is a simplicial space NC(O) = Ob(C),
NC
NC(1) =
and generally NC(n) c MM(C)
x ••• x
MM(C)
is the subset of compos able strings
(n
times)
called the
78
That is, i = l , ... ,n-l.
NC(n)
Here (f
(f , f , ... ,f = 1 2 n)
E:
i
2
(f {
n
NC(n)
i:
->
1,
•.. ,f
i
f
0
i
(f , •.. ,f -
NC(n+l)
is given by
, ..• ,f ) , n
n 1
1
and
NC(n-l)
->
i+ 1,
... ,f
0
0 < i < n
n),
) ,
i
=
n
is given by 0, ...
Remark 1.
Notice that
N
,n.
is a functor from the category
of topological categories (where the morphisms are continuous functors) to the category of simplicial spaces. Remark 2.
Observe that a topological group is a topological
category with just one object and it follows that
NG
as
defined in Example 3 is exactly the nerve of
G
as defined in
Example 4.
NG
defined in
Furthermore the simplicial space
Example 3 is exactly the nerve of the category
Ob(G) = G
follows:
target (go,g,) = go y
NG
y
G
NG G
and
= G
and
(go,g,)
Let
is a topological category is a pair )
O
Now i t is easy to see that the natural map
induces an isomorphism in homology, hence by assumption II f II :
(II X II (n) , Ii X II (n-1))
->
induces an isomorphism in homology. five-lemma shows that
II f II :
(II X'II (TI) ,ii X'II (n-1)) Now iterated use of the
II X II (n )
II x'II (n), n
-->
induces an isomorphism in homology and therefore
II X II
->
II X' II
=
1,2, ... ,
II f II :
also induces an isomorphism in homology.
By
the Universal coefficient theorem the result now follows, and thus finishes the proof of Proposition 5.15. Corollary 5.17.
Suppose
f
O
,f
1
: X
->
X'
are simplicially
homotopic simplicial maps of simplicial spaces p
there are continuous maps
satisfying i) -
and let
i
-->
=
O, ... ,p,
iii) of Exercise 2b) of Chapter 2).
II fOIl* Proof.
hi : X p
(i.e., for each
=
II f
1
11* : H*(II X'II)
-->
H*(II X II).
In fact consider the induced maps
c p + 1 ,q (X') p
L
i=O
->
cp,q (X) . #
(-1)lh.
1
be defined by
Then
85
Then
as in Exercise 2 of Chapter 2.
s since ff
f::
6" + 8"
0
p+l
are chain maps
and
C* (X')
....
Furthermore S
0
0
p+l
C*(X' ) .... C*(X ). p+l p
C*(X)
It follows that
are chain homotopic and hence
induce the same map in homology. Proof of Theorem 5.5.
n
class and let
U of
covering : n
-1
E .... X
(U
a
X
) .... U
a
First let
c
be a characteristic
be a principal G-bundle.
Choose a
such that there are trivializations
x G
and consider the diagram (5.12) above.
Notice that there is a commutative diagram
---='---...... where
U)
H (C
is the isomorphism of Lemma 1.25, so that
E* U
is
also an isomorphism. Now by naturality of (5.18)
E
U
U(C (E»
c
= f U(c (EG) )
and since
E
by
and equation (5.18).
c(EG)
is an isomorphism
On the other hand let principal G-bundle the class (5.19 )
E
U(c (E»
Co
c(E)
H*(BG) c(E)
by
is uniquely determined
and define for a
86
we must show that Now if of
X
then
c(E)
is well defined:
=
U
{ua}aEL
W
{Ua nuS} (a,S)ELXL'
and
U'
{US}SEL'
are two coverings
is also a covering of
X
and clearly there is a commutative diagram
(5.20)
Also let ljJW
f W : II NXW"
-->
BG
be the realization of
NljJW
where
is given by the transition functions corresponding to the
US.
trivializations
Then clearly there is a commutative
diagram
(5.21 ) IINXull/ From the diagram (5.20) and (5.21) it follows that it is enough to show that for any covering
q; (cO)
E H* (II NX U II)
U
the element
does not depend on the particular choices
of trivializations So let
and
be two sets of trivializations
} and let ljJ,ljJ' : Xu --> G be the a corresponding continuous functors. We want to show that the associated to
associated maps in cohomology.
Aa : Ua
-->
U =
{U
f U' f U : II NX II U
-->
BG
induce the same map
Now the family of continuous maps
G, a E L,
defined by
87
(x, \1
(x,g) E U
(X) 'g) ,
X
Ct
G,
satisfy A
Ct
Hence A :
A
W
=
W·
(x)
1
is just a continuous natural transformation of the functors
and
f Q= f U *
That
therefore follows from Corollary 5.17 and the following general lemma: Lemma 5.20.
Let
C
of topological categories
D
C, D.
be two continuous functors
If
W
A :
continuous natural transformation then
W'
NW,NW'
is
Q
NC
ND
are
simplicially homotopic simplicial maps. Proof.
We shall construct
hi : NC(p)
ND(p+l), i
satisfying i) - iii) of Exercise 2b) in Chapter 2. simplex in A
For
o
...
f
NC
f
1
ND
O, ... ,p,
Now a p-
is a string
i = O, ... ,p,
simplex in
=
hi
P
A
O'
••• ,A
E
MOJL (C) •
associates to this string the
(p+l)-
p
E Ob (C) ,
given by the string
w'
(f.) l
1jJ'(A.)< l
AA. l
W(f ) P
hi : NC(p)
ND(p+l)
is clearly continuous and it is straight-
forward to check the identities i) - iii) of Exercise 2b) in Chapter 2.
This proves the lemma.
88
It follows that
c(E)
defined by (5.19) is well defined
and it is easily checked that condition (5.2). Note.
c(E)
satisfies the naturality
This ends the proof of Theorem 5.5.
The original construction of a classifying space
is due to J. Milnor [20]. the one in G. Segal [24).
Our exposition follows essentially
6.
Simplicial manifolds. In this chapter
coefficients.
H*
The Chern-Wei I homomorphism for again denotes cohomology with real
We now want to define for a Lie group
w : I*{G)
Chern-Weil homomorphism is that
BG
and
is a simplicial manifold.
NG
is not a manifold.
H*{BG);
However,
the
but the trouble
BG
That is,
G
II N{G) II,
X
a
{X }
q
simplicial set is called a simplicial manifold if all COO
BG
X
manifolds and all face and degeneracy operators are
are
q
COO
maps. Example 1.
X = {X is a q} considered as zero dimensional
Again a simplicial set
simplicial manifold with all
X
q
manifolds. Example 2. space
NM
with
Also if NM{q)
=
M is a
00
C
manifold the simplicial
M and all face and degeneracy operators
equal to the identity is again a simplicial manifold. Example 3. and
NG
For
G
a Lie group the simplicial spaces
are also simplicial manifolds and
y : NG
NG
NG
is a
differentiable simplicial map. Example 4.
U
=
{Ua}aEL
manifold.
For
MaC
00
manifold with an open covering
the simplicial space Finally, if
E
NM U
is also a simplicial
M is a differentiable principal
G-bundle with differentiable trivializations
: U a a a then taking the nerves of the diagrams (5.10) and (5.11) we
obtain the corresponding diagrams of simplicial manifolds and differentiable simplicial maps.
x
G
90
Now let us study the cohomological properties of a simplicial manifold, in particular we want a de Rham theorem. Again in this chapter for
M a manifold
C*(M)
denotes the
cochain complex with real coefficients based on
COO
singular
simplices. Now consider a simplicial manifold Chapter 5 we have the double complex
X = {X}. p
cp,q(X)
=
As in
Cq(X ). p
Notice
that by Lemma 1.19 and Exercise 4 of Chapter 1 the natural map
Cq (X) ... top p induces an isomorphism on homology of the total complexes. Here
We also have the double complex the vertical differential differential in 8'
Aq{X
p)'"
A*{X) p
Aq{X + p 1) 8'
d"
is
(-1)P
times the exterior
and the horizontal differential is defined by p+1
L
i=O
. {-1)lS'!'. l
Furthermore we have an integration map
which is clearly a map of double complexes.
By Theorem 1.15 and
Lemma 1.19 we easily obtain Proposition 6.1. Then
Let
X
Ix: AP,q(X) ... cp,q{X)
H(A*(x»
=
{X} p
be a simplicial manifold.
induces a natural isomorphism
H(C*(X»
H* (II XII).
Now there is even another double complex associated to a simplicial manifold which generalizes the simplicial de Rham complex of Chapter 2:
91
Definition 6.2. manifold
X = {X} p
A simplicial n-form is a sequence
on the simplicial
=
of n-forms
such that
on
(id
(6.3)
x
O, ... ,p,
i
Remark.
Notice that
L'l p - 1 x X ,
on
=
P
P
=
0,1,2, ...
defines an n-form on
II L'l P x X and that (6.3) is the natural condition for a form p=O p on II X II in view of the iden tif ica tions (5.6). In the
following the restriction denoted
of
Notice also that for
is also
to X
discrete Definition 6.2
agrees with Definition 2.8. Let
An(X)
denote the set of simplicial n-forms on L'l P
Again the exterior differential on differential
An+ 1 (X)
d: An(X)
x
X
P
X.
defines a
and also we have the exterior
multiplication 1\
satisfying the usual identities. The complex
(A*(X) ,d)
is actually the total complex of
a double complex
(Ak,l(X) ,d' ,d").
Ak,l(X), k+l = n
iff
x
X
p
Here an n-form
lies in
is locally of the form
=L where L'l P
(to, ... ,t p)
and
{x.} J
as usual are the barycentric coordinates in
are local coordinates in
that
II
k+l=n and that
Ak,l (X)
X. p
It is easy to see
92
d where
d'
d' + d"
is the exterior derivative with respect to the
barycentric coordinates and
d"
is
(_l)k
times the exterior
derivative with respect to the x-variables. Now restricting a over
6
k
6k
(k,l)-form to
x
X k
and integrating
yields a map
which is clearly a map of double complexes.
The following
theorem is now a strightforward generalization of Theorem 2.16: 6.4.
(A*,l(X),d')
and
For each
the two chain complexes
1
(A*,l(X) ,0')
are chain equivalent.
In fact
there are natural maps E
and chain homotopies
such that (6.5)
1
(6.6)
d
6 I
0
d' = 0'
0
E
E
0
0
1,
E
0 '
0
d"
=
d"
0
E
id
(6.7)
o.
(6.8)
In particular
1
6:
Ak,l(X)
Ak,l(X)
induces a natural iso-
morphism on the homology of the total complexes (6.9)
H(A*(X))
H(A* (X))
H* (II X II).
93
Also let us state without proof (see J. L. Dupont [11]) the following generalization of Theorem 2.33: Theorem 6.10.
The isomorphism (6.9) is multiplicative
where the product on the left is induced by the
A-product
and where the product on the right is the cup-product. As an application of Theorem 6.4 let us consider a manifold
M with a covering
U =
{U}
a an:
and let
NM U
be the
simplicial manifold associated to the nerve of the category
MU.
Notice that the natural map
is induced by the natural projections
and that these also induce the natural map A*(M) Corollary 6.11. the natural map
A*(NM ) .
A*(NM) For
A*(M)
U
U=
{U}
a
A*(NM )
an open covering of
M
induces an isomorphism in
U
homology. Proof.
In fact the composite
is the map
eA
of Lemma 1.24.
Now let us turn to Chern-Wei I theory for simplicial manifolds.
A
sequence E =
{E
P
simplicial G-bundle n
p
}, M =
: E
P
{M }
P
M
p
n : E
M
is of course a
of differentiable G-bundles where
are simplicial manifolds,
n: E
M
is
94
a simplicial differentiable map and also right multiplication g E G, R : E .... E, g
by 'IT
:
E
A connection in
is simplicial.
.... M is then a 1-form
on
8
(in the sense of
E
1
Definition 6.2 above) with coefficients in restricted to
t-,P
x E
p
for
8
is a connection in the usual sense in
E .... liP x M P P Again we have the curvature form liP
the bundle
such that
x
P E Ik(G)
we get
p(n k ) E A
n
2k(M)
defined by 3.14 and a closed form
representing a class
such that Theorem 4.3 holds. In particular let us consider the simplicial G-bundle y : NG
.... NG.
There is actually a canonical connection in
this bundle constructed as follows: Let G .... pt.
80
be the Maurer-Cartan connection in the bundle
Also let
qi : liP
8
is simply given over
(6.12 )
liP
NG(p)
x
t
(to, ... ,t
By Proposition 3.10,
p)
be the projection
O, ... ,p,
=
8
where as usual liP.
NG(p) .... G
GP+ 1, i
onto the i-th factor in Then
x
and let
8.
1
by
8
P P
are the barycentric coordinates in 81l1 P x NG(p)
is clearly a connection
in the usual sense and it is also obvious from (6.12) that satisfies (6.3). Theorem 6.13.
We now summarize: a) There is a canonical homomorphism W
such that for
8
I*(G) .... H*(BG)
P E Ik(G),
w(P)
is represented in
2k(NG) A
by
95
p{n k )
n
where
is the curvature form of the connection
e
defined by (6.12). b)
P E Ik{G),
Let for
characteristic class.
w{P) (.)
Then if
TI:
be the corresponding
E
M
is an ordinary
differentiable G-bundle we have W{P) (E)
where
a*
WE : I* (G)
c)
W:
d)
Let
I*{G)
H* (M)
I*{G)
H*{BG)
a: H I*{H)
is the usual Chern-Weil homomorphism.
G
is an algebra homomorphism.
be a Lie group homomorphism and let
be the induced map. a*
I*{G)
-1
H* (BG)
Then the diagram
• I* (H)
1-
Ba*
• H* (BH)
commutes. Proof.
a)
is a definition.
b) Choose an open covering trivializations of
E
U
=
{Ua}
of
M
and
so that we have a commutative diagram
of differentiable simplicial bundles:
}n
NE ... '---
F-u---+' 1
NM+'--- NM U -----'>. NG . By the proof of Theorem 5.5 the pull back of
II NM U II in
is given by
H{A*{NM U)
fii
(w
(P) )
w{P) (E)
to
which clearly is represented
by the Chern-Weil image of
P
for the simplicial
96 G-bundle
NE
e
connection in
E
M
NM
V
U
e'
with connection
defined by (6.12).
On the other hand a connection
induces another connection
the pull-back of
WE(P)
in
by the Chern-Weil image of
H(A*(NM P
induced from the
en
in
NE
NM
V
U
and
is clearly represented
U)
using the connection
However, by the argument of Theorem 4.3, b)
en.
the Chern-Wei 1 image
is independent of the choice of connection, which proves that
where
IOU:
Since
IOU
II NM
U
II
M
is the natural map considered above.
induces an isomorphism in cohomology this ends the
proof of b). c) follows again from the simplicial analogue of Theorem 4.3 c) and Theorem 6.10. d)
is straightforward and the proof is left to the reader.
Note.
Notice that by a),
the total complexes defined elements.
A*(NG)
and
w(P)
is also represented in
C*(NG)
The construction of
by canonically w(P)
in
A*(NG)
is
due to H. Shulman [26] generalizing a construction by R. Bott (see [2], [4], and [5]).
The exposition in terms of simplicial
manifolds follows J. L. Dupont [11].
7.
Characteristic classes for some classical groups
We shall now study the properties of the characteristic classes defined in the examples of Chapter 4. Chern classes. For
G
we considered in Chapter 4 Example 4 the
=
complex valued invariant polynomials defined by (4.13).
C k k,
=
0,1, ... ,n
=
(CO
For a differentiable
TI
:
1),
E
we thus define characteristic classes called the Chern classes 0,1, ... ,n,
(7,1)
where
represented by the complex valued 2k-forms is the curvature form of a connection in
TI
:
E
M.
Q
Notice that
since every complex vector bundle has a Hermitian metric, i.e. a reduction to inclusion
U(n),
Ck(E)
2k(M,B) H
actually lies in the image of the
H2k(M,a:)
(c f . Exercise 4 of Chapter 4).
By Theorem 6.13 we can extend the definition of the Chern classes to any topological
and then use Theorem 5.5. since
C k
Gl(n,cr)-bundle by first defining
Again
restricted to
from Theorem 6.13 d)
c
is a real class. In fact k is a real polynomial it follows
to BU(n) is k a real class (represented by a real valued form), and since the natural inclusion
that the restriction of
j : U(n)
C
Gl(n,cr)
it follows that Bj
BU(n)
B Gl(n,a:)
c
is a homotopy equivalence
M
98 induces an isomorphism in cohomology. Proposition 7.2.
Let
a : H
In general we have
G
be a homomorphism of
two Lie groups which induces an isomorphism in homology (coefficients in a P.I.D.
h).
Then also
Ba : BH
BG
induces an isomorphism in homology as well as in cohomology (with coefficients Proof.
h).
By Runneth's formula
Na(p)
: NH(p)
induces an isomorphism in homology for each
NG(p)
p.
The proposition
therefore follows by Lemma 5.16. Before continuing the study of the Chern classes we make a few definitions: Suppose we consider a topological space
E Then the
X
s:
and a
: E e F
sum
with a principal
X
X
X.
F
is most easily de-
scribed in terms of transition functions as follows: First let
be the homomorphism taking a pair of matrices
(A,B)
to the
matrix A Now choose a covering and
F
and
{haS}
and
F
e
B
U =
have trivializations over
F
X
such that both
Ua' a
and let
E
{gaS}
be the corresponding transition functions for
respectively.
e
Then
with transition functions and
of
{gaS
: e
e
E
F
haS}'
are differentiable then also
E
X
is the bundle
Notice that if
e
F
is.
E
E
99
=
Notice that
=
the multiplicative
group of non-zero complex numbers.
are in 1-1
correspondence with l-dimensional complex vector bundles (also called complex line bundles).
An important example is the H* (II NG II) constant.
\1/
r*
is multiplication by a non-zero
This shows the injectivity of
proves (8.28).
II i 11*
Choose an inner product on
is invariant under the adjoint action of G
is compact).
that is, split
if
Ad(exp(t», t E./-
acts on
(8.31)
21TCi (t) 1
where
Ci
i
M'Z,
--> JR,
G/T
that
by the matrix
0
cos 21TCi (t). 1
° i = 1, ... ,m,
cos 21TCi (t) m
-sin 21TCi m(t)
sin 21TCi m(t)
cos 21TCim (t)
are linear forms on
details see e.g. Adams [1, Chapter 4]). bundle of
for
2m}
-sin 21TCi (t) 1
Sl.n 21TCi (t) 1
: "
(This is possible
into an orthogonal direct sum
1
C'
G.
which
Now make a root space decomposition of
and find an orthonormal basis {e , ••• ,e
Ad(exp(t) )
and hence
It remains to prove the existence of
Proof of Lemma 8.30.
since
II i II *:
0
,i
(for
Notice that the tangent
can be identified with the 2m-dimensional vector
127
bundle G
E;
x T ""'
.... G/T
which is clearly an oriented bundle with the orientation given by the basis
{e
Now let 8
K
1, :
1
,e
2rn}.
,j.
be the orthogonal projection and let
be the canonical connection in
NG
given by (6.12).
Then
clearly
defines a connection in the principal T-bundle
NG .... NG/T
let
P E Im(T)
n
T
be the curvature form.
Also consider
and
given by the polynomial function m
P(v, ... ,v) = (_l)m
1I a..(v),
i=l
Then by Chern-Weil theory the 2m-form on
NG/T
and we let
is a closed form
be the corresponding form on
under the identification (8.29), so clearly satisfied.
It remains to prove (ii).
is just the G .... G/T
v E)..
l
image of
with connection
8
T
P
Now
d'l'
=
N(G;G/T)
is m) = P(n E A2m (G/ T) 0
T
in the principal T-bundle
given by g E G,
and
f
nT
=
d8 T.
Unfortunately it is not so easy to calculate
directly. However, as noticed above the extension G/T of the bundle G .... G/T to the group SO{2m) via the adjoint representation on
is just the tangent bundle of
it is easy to see that
G/T
and
is exactly the Pfaffian form.
On the other hand it follows from (8.31) that the bundle is a Whitney sum of
SO(2)-bundles.
Therefore, as remarked after
128 Proposition 8.11
(cf. Exercise 2 of Chapter 7),
I
(8.32)
G/T
P(o*) = .
Now the right hand side of (8.32) we can compute by the formula (7.33) for a vector field of the following form: Choose a regular element ai' i
=
bundle
V
1, ... ,m) i;,: G x
o
E )
and consider the section 4ft ....
T
G/T
s(gT) where again va
given by
(g, (id-K)
0
Ad(g
-1
)v
0
s
at
g E NT
iff NT/T
W
gT E W
maps a neighbourhood of
gT
{T}
g E G
s(gT)
is
in
vanishes at the
+1.
For this we recall the exp
.... G .... G/T
diffeomorphic onto a neighbour-
0 E
G/T,
Since
Now we claim that the
G/T.
c
s
so
well-known fact that the exponential map
hood of
a) ,
is the orthogonal projection.
finite set of points local index of
of the vector
1 .. ;
K
is regular
s
so we get a local trivialization near
by (g exp x,v) ..... v,
x E
near zero, v E "4(.
It is therefore enough to see that the map
s
....
given
by i(x)
=
(id-K) (Ad(exp(-x))v
g),
v
g
Ad (g
is an orientation preserving diffeomorphism near differential x E
s*
at
0
is given by
i*(x)
-1
O.
) v, x E mt , The
= -[x,v g] = ad(v g) (x),
Differentiating (8.31) and taking the determinant now
gives det(ad v ) g
(2,r)
m.!!1..
Ila.(v) i=1 l g
2
>0
129
so the local index of
J
s
'ji0
at
gT
J
G/T
is +1.
PIn;)
=
It follows that
Iwl >
0
G/T
which proves Lemma 8.30 and finishes the proof of Theorem 8.1 for
G
connected.
For
G
a general compact group we get a diagram similar
to (8 .27) : I* (G)
1
(8.33)
Inv G/ G (I*(G O)) 0
---+
H* (BG)
1
Inv G/G (H* (BGO) )
---+
0
where again the upper horizontal map is the isomorphism (8.2) and the right vertical map is an isomorphism since connected.
Again it suffices to show that if
GO
i : GO
is G
is
the inclusion then (8.34 )
Bi*: H*(BG)
H*(BG
is injective.
O)
As before, this is equivalent to showing that II ill*
i
: H*(II NG II)
N(G;G/G o )
NG
H*(IIN(G;G/G O) II)
is injective, whe r e
is defined as follows: NG(p)
and
i
x
G/G
O
is given by the projection
on the first factor. T
This time A*N(G;(G/G
o) ) )
A*(NG)
is simply given by T(lP) where
llP
x
NG(p)
L
gEG/G llP
s*lP, g
O
x
lP E A*(N(G;G/G o)
N(G;G/G
O)
(p)
is given by putting
130
gG
o
E GIGO
on the last coordinate (notice that
a simplicial map but still easily checked that
T
T
is well-defined).
I GIGO I.
multiplication by
IG/GOI
Hence also
of Theorem
0
T
r
II
0
i*
II *
is
is
where
T* : H*(II N(G;G/G T.
T*
is not
g
Again it is
is a chain map and that
multiplication by
is the map induced by
s
O)
II)
H*(II NG II)
-+
This shows (8.34) and ends the proof
8.'.
Corollary 8.35.
(A. Borel [3]).
connected Lie group, and let a maximal torus.
Then
Bi
i
T
Let -+
BG
BT
G
G
be a compact
be the inclusion of
induces an isomorphism
H* (BG) ------>. rnvwH* (BT) . Proof.
Obvious from the diagram (8.27).
Corollary 8.36.
(i) The Chern classes of
are uniquely determined by the properties a) - d)
of Theorem
7.3.
(ii) Furthermore H* (BU(n)) ';;; m. [c"
H* (BGl (n,(£))
is a polynomial ring with the Chern classes
... ,c
c, , ... ,c
n] n
as
generators. Proof. U(n)-bundles.
As noticed in Chapter 7 it is enough to consider Now let
i : Tn
(
o
-+
U(n)
be the natural inclusion
131
and let j
qj : Tn
1, ... ,n.
=
U(1)
be the projection onto the j-th factor,
It is well-known that
Tn
is a maximal torus so
by Corollary 8.35 Bi* is injective. values on
H*(BT n)
H*(BU(n))
That is, the Chern classes are determined by the
U(n)-bundles which are Whitney sums of
Hence by (7.5) they are determined by
c
on
1
U(1)-bundles.
U(1)-bundles.
This, however, is determined by (7.6) as remarked immediately after Proposition 8.11.
This proves (i).
(ii) By Corollary 8.22 Yj
=
n)
= JR[Y1""'Yn]
2(BTn), (Bqj)*C 1 E H j = 1, ... ,n
the first Chern class. on
H*(BT
Tn
Now
and
c
1
where 2(BU(1)) is E H
W is the symmetric group acting
by permuting the factors, i.e. n
Hence
Invw(BT )
W acts on
H*(BT
n)
by
is a polynomial ring
with generators the elementary symmetric polynomials 1, ... , n,
der Waerden [32,
§ 29]).
(see e.g. B.L. van
in
However, by (7.5) 1, ... In,
which proves the corollary. Corollary 8.37. for
(i) The Euler class with real coefficients
SO(2m)-bundles is uniquely determined by the properties
i), ii), and v)
of Exercise 1 e) of Chapter 7.
In particular
formula (7.39) holds. (ii) Furthermore H*(BG1(2m,JR)+)
H*(BSO(2m)) JR [P1'··· ,Pm-1 ,e]
132
is a polynomial ring with generators the first classes
and the Euler class
m-l
Pontrjagin
e.
(iii) Finally H* (BGl (2m,lR))
H* (BO (2m))
is a polynomial ring in the Pontrjagin classes Proof. be the set
The maximal torus in m T
SO(2m)
cos 21Tx
1
o
1
cos 21TX
o
1
sin 21TX
m. , ••• ,x ) E lRm/ ." IJJ
Again let
m
inclusion and let
-sin 21TX
m
cos 21TX
m
SO(2m)
i
SO (2), j
projection on the j-th factor. injectivityof
is well-known to
of matrices of the form -sin 21Tx
(x
Pl, ... ,Pm.
1, ... ,m,
=
m m
be the be the
As before (i) follows from the
i*: H*(BSO(m))
H*(BT
m)
together with the
remark following Proposition 8.11. m) H* (BT
(ii) Again Yj
=
5.17)
2
(Bq.) *e E H (BTm) . J
that the Weyl group
= lR [Yl'··· 'Ym]
where
It is easily seen (cf. Adams Example W acts on
m) H* (BT
by permuting
the Yj'S and changing the sign on an even number of the Yj's. We want to determine the subring
First notice that
A
changing the sign of
has an involution Y say. 1,
T
:
Then clearly
A A
A
given by
= A+
ffi
A_,
133
where
A+
and
A
are the
±1
eigen spaces for
Notice
W'
is the group generated by the permutations of the
T.
that
where
Yj's together with the transformations which changes the sign of any number of the
where
0.
Yj's.
is the j-th elementary symmetric
'=
]
It is now easily seen that
polynomial
Now every element of
A_
is easily
seen to be divisible by the polynomial £
'=
Hence
Now
Here £
'=
£
2
hence
0,
J
'=
(Bi)*e
{Bi)*p"
J
j
1, ... ,m, by (7.20) and (7.26), and
by (7.24).
This proves (ii).
(iii) By Theorem 8.1 and (8.2) H*(BO(2m))
r-J
I*
(O
(2m))
Inv O(2m)/SO{2m) (I*(SO{2m))). Here
0{2m)/SO{2m) ;;:: 'll./2
acts on
I*{SO{2m))
using the adjoint
action of an orientation reversing orthogonal matrix.
This
clearly fixes the Pontrjagin polynomials and changes the sign
134
of the Pfaffian polynomial (see Chapter 4, Example 1 and 3). Hence the invariant part of ring in the variables
I*(SO(2m))
P 1"",Pm-
is the polynomial
and
1
Pf2 = P. m
This proves
the corollary. In a similar way one proves Corollary 8.38. (L)
H*(BG1(2m+1)+)
H*(BSO(2m+l))
lR[Pl, ... ,Pm]
is a polynomial ring in the Pontrjagin classes. (ii) H*(BGI(2m+l))
H*(BO(2m+l))
H* (BSO(2m+l))
lR [p 1 ' ••• , Pm] . Remark.
In all the cases considered above
Inv w(S*(4*)) In fact if
V
is a polynomial ring.
H*(BG) =
This is no coincidence.
is any real vectorspace of dimension
I
and
W
is a finite group generated by reflections in hyperplanes of then
Invw(S*(V*))
is a polynomial ring in
N. Bourbaki [6, Chapitre V,
§
I
V,
generators (cf.
5, theoreme 3]).
APPENDIX We will in this appendix give a proof of the differentiability of the function
pI
8.3 by the formula (8.5).
defined in the proof of Proposition First we recall some rather standard
facts from the theory of Lie groups. In the following suppose
G
simple Lie group without center.
=
1 @lR
a:
is a compact connected semiLet
the complexification of
complex analytic Lie group
Ga:
be the Lie algebra and
1.
Then there is a
(the complexification of
G)
135
and an injection algebra of
1 .. 1
$
Ga
i1 =
j
:
and
1
without center
G .... Gn: j*
ad
(1)
= End
To see this notice that since
a;'
(1)
Ad: G .... Gl
(1)
is
G
is injective and the image c
Int(1) = Gl(1)
with Lie algebra
defined by
ad(1 ) = {ad (v) I v
E1 ,
i
;; Gl('1rr)
group with complex Lie algebra
1rr .... ad(
XE1
ad(v) (x ) = l v x ] ,
Ga; = Int(,
We can then take
ad :
is the Lie
is the natural inclusion
is the connected subgroup c
such that
(1 n:)
ad
c
is an isomorphism and
}.
the complex analytic End j
(1rr) .
:
Here again
G .... G rr
is given
by the composite
In the following we shall identify
G
with the image in
Grr.
We also need the Jordan-decomposition of elements of
v
For a complex vector space
a linear map
1e '
A E End(V)
has a unique Jordan-decomposition S + N,
A
with for
S 5)
semi-simple and
v E
ad (v) E End ( 1n:) Lemma8.A.1. decomposition
v
is nilpotent and Proof.
(i.e.
V
nilpotent (i.e.
N
particular for
SN
1 rr
NS
has a basis of eigenvectors Nk = 0
for some
k
In
0).
we have a Jordan-decomposition of
and we have For
v E
s + n [s,n]
1rr
there is a unique Jordan-
such that
=
ad v
is semi-simple,
ad n
0
We must show that the semi-simple part of
(and hence also the nilpotent part) lies again in
ad v
136
c:
ad(1 rr) = End(1 rr) ·
1rr
Since
1rr
the Lie algebra of derivations of [14, Chapter II,
Proposition 6.4]),
lies in
iff [ox,y] + [x,oy]
O[x,y]
We must show that if
0
ad(1 rr)
is semi-simple
(see e.g. S. Helgason
o E End (
that is,
,
x,y E
1rr .
So let
0
= S
be the Jordan
+ N
decomposition.
Then there is a direct sum decomposition
'1'rr=
such that
A
A eigenvalue
is the eigenspace of
(4jrr)A
S
S
with
that is
A,
o That
1rr)
is a derivation then also the semi-
simple part is a derivation.
E9
is
for some
k
is a derivation simply means that for
O}.
>
A,
E rr,
This, however, easily follows from the identity k
[x,y]
=
k
k L (.) [ (0-1..) k-i x ,
which is proved by induction on Now let let
k rr
=
T
4®JR
G II
connected Lie group. ad(t) metric.
i
i=O
k.
y],
x,y E
1rr'
k=O, 1,2, •.
This proves the lemma.
be a maximal torus with Lie algebra
irr
and let
Every element
4,
be the corresponding is semi-simple since
is skew-adjoint with respect to a G-invariant Therefore every element of
is semi-simple as well
and we have the root space decomposition (see e.g. Helgason [14, Chapter III, § 4])
r
137
where
et E ,
1
are the roots, i.e.
one-dimensional subspaces and
Furthermore let
t
(8.A.2)
+ c
.t and
Then both (8.A.3)
B
Go:
Ell = aE+ 1a:a
Ell etE+ G:et
t-+
1 er
are subalgebras of
[1a:et' 1a:s]
Also let
be a choice of positive roots and let
l' ' t- +
A. a: Ell
=
are
a:et
since
a,S E .
"da:(et+S)'
be the group with Lie
With this
notation we now have
i7o:'
Lemma 8.A.4.
b) Ad(g)v
+
= t+n E.Ir
is nilpotent.
with
v
is semi-simple and
v E l 0:
there is
tEA-a:' n EJ.+
and
g E Ga:
such that
[t,n]
O.
then the semi-simple part of
v
Further-
is conjugate
t. c) G
The inclusion and
d)
If
there exists
Proof.
v,
If
NT .... NTa:
of normalizers of
and
T
Ta:
Ga:' respectively, induces an isomorphism W
of
-f!r
For every element
more, if
in
is a maximal abelian subalgebra of
Furthermore every element of
every element of
to
a)
= NT/T ----=--. NTa:/Ta: .
s E 4a:
and i f for some
w E NTa:
a)
such that
For
[v,4a:J = 0
let
'1]
= 0,
Re v [Tm
A-rr
then
= Ad(g)s.
0e the complex conjugate
CiT, "a:J
then clearly also
both the real and imaginary part [Re v
Ad(w) s
v
Ad(g)s E
g E Go:,
rm v
and
J
=
0
=
0
satisfy
so
138
so by maximali ty of
j
v = Re v + i 1m v =
is a maximal abelian subalgebra.
o.
This shows that
4- II:
The second statement is
already proved and the last clearly follows from (B.A.3). b)
By the Iwasawa decomposition (see e.g. Helgason [14,
Chapter VI, Theorem 6.3]) we have G . e xp f L
(B.A.5) in particular
B n G
T
J) .
expp
and the inclusion
G
induces
a diffeomorphism
so the Euler characteristic of
is different from zero
(cf. Adams [1, proof of Theorem 4.21]).
For
v E
10::
we there-
fore conclude by Lefschetz' fixed point theorem that there is an element
g E
gB E
such that
is fixed under the
oneparameter group of diffeomorphisms
where
exp(rv)xB, r
hr(XB)
g Hence
-1
E,.t..
Ad (g -1) v
E JR,
that is,
exp(rv)g E B,
Vr E JR.
We can therefore suppose
VEt,
and
we write (B.A.6)
v
Now we claim that we can change of
B
so that
x
a
0
v
only for
is a minimal root so that both
by conjugation by elements a(t)
x
a
*0
=
O. but
In fact suppose a(t)
O.
Then
a
139
1
Exp(ad(a(t) x a)) (v) vt+ where
a'
> a
means that
a' - a
this procedure we can find
b E B t +
Ad(b)v
Therefore we put [t,n] = 0;
hence
4.11:
b E B
x ))i(v) a
a
Ya'
is a positive root.
Iterating
such that
L
n =
Notice that conjugation by in
L
a'>a
aE¢+ a(t)=O
L Z E aE¢+ ct Ad(b)v = t + n
L
a (1 [v,x + t) a]
z. ct
and we clearly have is the Jordan decomposition. does not change the component
in the decomposition (8.A.6) which proves the second
statement in b) . c)
Clearly
NT/T
NT
n G
and since
is injective.
left-multiplication by
Now for
gET
the map
T
a regular element,
g
has a fixed point for every element in
B.
Therefore
the composite
is a bijection so it remains to show that
T
a
=
n B.
however, is trivial from the fact that every element of of the form
a . exp(n)
with
a E Til:
and
n
Let
s E
4a
and
Consider the Lie algebra
g E Gil:
with
Ad(g)s
B
is
This ends
the proof of c). d)
This,
t E
ill:'
140
J= and let
0
Ga:
x E
E1a:
I
[v,t]
O}
be the associated connected sUbgroup of
j a:
Then clear ly
{v
J
Ad (g) ). a:
and also
Ga:'
since for
/.1£ 1. (f
Also
Ad (g) j..
and hence
1£
o.
l xvs l
[Ad (g) (x) , t]
are Cartan subalgebras (i. e. a
nilpotent algebra with itself as normalizer).
Hence by the
conjugacy theorem (see e.g. J. P. Serre [25, Chapitre III, Theoreme 2])
there exists a Ad (g)
Hence
d
-1
g E NTa:
and
d E 0
such that
Arr.
Ad(d
-1
Ad(d)t
g)s
t.
This ends the
proof of the lemma. After these preparations we now return to the proof of the
8.3.
Recall that
Lie group
G
m
pI
differentiability of
in the proof of Proposition
is the Lie algebra of a compact connected
with maximal torus
polynomial of degree
T
and
P
is a homogeneous
on the Lie algebra ), of
k
T.
pI
is defined by the formula pI (v)
P (ad (g) v)
We shall show that pI
a:
on
pI
where
lid (g) v E
4--
for some
g E G.
extends to a complex analytic function
11£.
Since
G
where
is compact
}= is the center and [14, Chapter II,
{v E
11
' is a l Proposition
[v,x]
=0
Vx E
7}
semi-simple ideal (see Helgason 6.6]).
Furthermore, if
Z c G
141
is the center of
'1'
then
G
is naturally identified with G' = G/Z.
the Lie algebra of the group representation factors through Ad(g) (z+v)
and
z + Ad(g' )v,
g' = gZ E G'.
where
A=
f
@
Notice that center.
G'
./.
G'
n
Also
i
Z
Clearly the adjoint
and E}
, v E
I"
g E G,
T' = T/Z
is a maximal torus in
G'
./ n
is the Lie algebra of
T' .
where
7'
is a compact semi-simple Lie group without
Therefore we shall restrict to the case where
semi-simple without center.
G
is
The reader will have no difficulties
in extending the arguments to the general case. The homogeneous polynomial
clearly extends to
P : , , / .... lR
a complex homogeneous polynomial
..
:
and obviously
is invariant under the adjoint action of 8.A.4 c) and the invariance of
For
I rr ....
pI
Now define
C :
v
under the action by
w on
A.
as follows:
0:
choose
Ell!
p
by Lemma
such that
g E
Ad(g)v
t + n
as in Lemma 8.A.4 b), and put
Then this is clearly well-defined by the uniqueness of the Jordandecomposition and Lemma 8.A.4 d) . First we show that let
1T
:
.t . .
j. I!
@
+
-t. = "-I! t
.•
....
is continuous:
be the projection in the and notice that if
then we can write Ad(u)v
P'
g
= Ad(b -1 )
=u
Ad(g)v
. b, u E G, b E B
(t+n)
=
Clearly also
t + n',
with
pl.
For this
decomposition
=
t + n
as above
by (8.A.5) and then n' E
..t +.
142
It follows that
=
(S.A.7)
and by the second part of Lemma S.A.4 b) any
u E: G
such that
To show that
Ad (u) v E:
P'
{v
there is a subsequence choose
uk E: G
fr.
is continuous it suffices to show that
a:
whenever a sequence
this equation holds for
k}, {v
such that
k
=
1,2, ... ,
converges to
v,
k .)
(v).
such that
k.}
(v
l
l
we can assume by taking a subsequence that say.
Hence
Ad(Uk)V
To see that
Ad(u)v
k
p'
uk
Now
is compact
G
Since
Ad(uk)v k E:£.
then
converges to
u,
and so
is actually complex analytic it suffices
c
by the Riemann removable singularity theorem (cf. R. C. Gunning and H. Rossi [13, Chapter 1,
§ C, Theorem
3]) to show that it
is complex analytic outside a closed algebraic set
S
*
For this consider the complex analytic mapping F
defined by F (g , t.)
= Ad (g) t,
= p((t).
and notice that analytic near points at
(g,t).
only if ad (t) 1
=
:
t
J(
= Ad(g)t
for which
Now it is easy to see that
F
F
is strictly bigger than
and let
be the set
is
is non-singular
is singular at
is singular in the sense that the kernel of
1
v
l}
then by Lemma 8.A.4 b),
actually semi-simple so by the above near
of
s
S
v
is
is complex analytic is an algebraic subset
For this let
aO(v) + a 1 (v)
x
+ •.. + an(n» ..
n
=
he the characteristic polynomial of
det(ad(v)-)"l), n ad v.
=
dimo:
10:'
Then clearly
o} which is obviously a closed algebraic set and since
/1 II n
S
=
U ker a. t a.E
there exist elements outside the complex analyticity of 8.3.
S. pI '
consider
{e 1 , ••• ,e
p}
0
JRP
and for
+' .
inductively as follows:
a(gl, ... ,gp) p = 1
6P
a(gl)
: 6'
G/K
For
is given by
as the cone on the face spanned by
Then the restriction of
a(g"
that face must be given by
L 0 a(g2, ... ,g ), g, p this map to the cone using the contraction h s'
... ,gp)
to
and we extend Explicitly
(9.8)
It is now straightforward to check (9.6) inductively. The merit of a filling
a
of
G/K
is that it enables us
to construct explicit Eilenberg-MacLane cochains:
Consider the
subcomplex
A*(G/K)
InvG(A*(G/K»
of the de Rham complex
consisting of G-invariant forms (where the G-action is induced by the left G-action on
by
G/K).
Define the map
148
(9.9) 0,1 Proposition 9.10. b)
J
a)
,2, ..
is a chain map.
The induced map on homology
is independent of the choice of filling. Proof.
a)
By Stoke's theorem and (9.6)
J (dw) (gl'··· ,gp+1)
J p+1
t>
J t>P
+
[L
0
gl
(gl'··· ,gp+1) *dw o(g2,···,g +l)]*w +
0
p
'J
PL (_1)l o(gl,···,g,g'+l,···,g +l)*w + i=l t>P l l P
+ (-1)P+1J P
o(gl, ... ,g )*w p
(w) ) (g 1 ' ... , gp+ 1 )
= w.
We give an alternative description of
J*:
Consider the
map of simplicial manifolds
where
and the face operators are given by i
o i
0, < i
p.
< p,
149
is just given
y
the projection onto the first factor.
the proof of Theorem 8.1. bundle with fibre
a
that if L
(g1 .•. gp)
G/K
associated to
is a filling of -1
II y II
w E Aq{G/K)
Now if
family of forms on
LIP
the projections onto
H{Inv G/K
=
dw G
dw,
x
Y G
d
G/K
is the fibre
y
: EG d
BG d.
Notice
then the family
LIP
a{g1,···,gp)
0
defines a section of
Clearly
The realization of
(Cf.
G/K, gl, ... ,gp E G, P = 0,1,2, ... ,
which explains the definition).
is an invariant form then the corresponding NGd{p)
G/K,
x
G/K, p
=
0,1, ..• ,
defines an element
induced by
w E Aq{N{Gd;G/K)).
so we have an induced map on homology
A*{G/K))
H{A*{N{Gd;G/K))).
On the other hand, since
is contractible
induces an isomorphism in de Rham cohomology by Lemma 5.16 and Theorem 6.4.
Hence the composite map
H{Inv
G
A*{G/K))
is canonically defined (i.e. without a choice of filling) we claim that this is just
J*
get an explicit inverse to
y*
where
a
In fact given a filling
and a
we
0,1,2, ... ,
is
given by
gl , ... ,gp E G, P
0,1,2, .•.
150
w E InvGAP{G/K)
Then obviously for
J
/),
p(L
(g1·· .gp)
_1
0
G{gl,···,gp)]*w
= J",pG(gl,···,gp)*w
J{w){gl,···,gp)·
This proves the proposition. Remark. h
In the proof of Lemma 9.7 we replaced the contraction
by the contraction
s
where
hots)
in order to be able to define the ",p.
of
o (s) = 0 00
map
C
for
near zero
s
on all
G(g l' ... ,gp)
On the other hand the inductive construction (9.8) using
the original contraction makes sense on the open simplex and the corresponding change of parameter does not affect the value of the integral (9. 9).
In particular let us describe
ly for the case where
G
§
TI
:
G
G/K
7])
explicit-
s
is semisimple with finite center:
'7
we can choose a Cartan decomposition (14, Chapter 3,
="
h
=
and the map
TI
0
Then
6l/ '
(see Helgason
exp
G/K
is the projection and
G
(where
the exponential
map) is a diffeomorphism {see Helgason (14, Chapter 6, Theorem 1.1]). Therefore we get a contraction defined by (9. 11 )
The curves
h
s
s
-1
=
(x)
hs{x)
Riemannian metric on
x E G/K,
(x l j ,
s E [0, 1 ] •
are geodesics with respect to a G-invariant G/K
and we shall therefore refer to the
corresponding filling defined inductively by (9.8) as the filling by geodesic simplices. We can now describe the composite map (9.4): Theorem 9.12. Bj*: H*{BG,lR) is represented in
For
Il(K)
P E
H*{BGd,lR)
of
H2 l{C*{NG)) d
the image under
W (P)
E H
2l (BK, lR)
H
2l
(BG, lR)
by the Eilenberg-MacLane co chain
151
1
1
where
above and
J
E InvG(A
21
is defined in step I
(G/K»
is given by (9.9).
That is,
(9.13 ) where
a
is a filling of
Proof.
Let
i
G/K.
:
G
be the inclusion and consider the
commutative diagram of simplicial manifolds N(Gd;G/K)
1y
(9.14 )
N(G where
----L
1
Nj
d)
j : NGd(p)
x
NT
NG/K ,
• NG
G/K
NK/K
1
Ni
0, k E SO(2),
and
b)d '
ad _ b2
1,
a,d>O.
It is easy to see that (9.16) then reduces to
1
r
t
0).
Notice that the numerical value
2nL Ar c (and equal to zero for
b
an
b
(1+a\] j - Arc tan,-S-j
satisfies rr
(9.17)
2"
1
4·
(This inequality can also be deduced directly from Theorem 9.12; see Exercise 2 below).
This has the following consequence due
to J. Milnor [22]: Corollary 9.18.
Let
over an oriented surface class (9.19 )
e(E)
: E X h
X be a flat Sl(2,IR)-bundle h of genus h > 1. Then the Euler
satisfies l1 < h.
154
Proof.
We first need some well-known facts about the
topology of surfaces.
X can be constructed as a 4h-polygon h with pairwise identifications of the sides as on the figure
/
/ / /
/ / /
/
.:..
. ...... Here the sides group
f
x
1,
..• ,x
give generators of the fundamental
2h
with the single relation 1•
Furthermore the universal covering is contractible (see reference in Exercise 2 e) below) .
,'Ie can now define a continuous map
f
For
:
Bf
-+
X h
generators
as follows: x 1 ' ... , x 2h
x
E
r
representing
choose a word in the x
and map
into the corresponding curve in the polygon. over the skeletons of groups
= a
for
Bf
i > 1.
with integral coefficients of C*Nf.
x
x c /',1
x
r
Now extend the map
using the fact that the homotopy Clearly
equivalence by Whitehead's theorem.
of the complex
/', 1
Hence
X h
f
is a homotopy
In particular the homology is isomorphic to the homology
H (C*Nr) :;:; 2Z 2
the generator is represented by the chain
and we claim that
z E C2(Nf)
defined
by z =
-1 -1 -1 -1 (x 1,x 2) + (x 1x 2,x 1) + ... + (xx 2x 1 x 2 ... x2h,x2h-1) +
+ (1,1) -
-1 (x 1,x1 ) + (1,1) -
-1 (x 2,x 2 ) + •.• + (1,1) -
-1 (x2h-1,x2h-1)
155
which is easily checked to be a cycle. is the sum of all the
(4h-2)
In fact
f*z E C*(X
h)
2-simplices in the triangulation
shown in the above figure plus some degenerate simplices. Now any flat a map
Be
:
Sf
S :
SI(2,:rn.)-bundle BSI(2lR)d
where
a : I'
homomorphism (see Exercise 1 below).
=
E
Bf
is induced by
SI(2,:rn.)
is a
It follows that
.
Now it is easy to see from (9.16) that a simplex of the form (x,x
-1
)
contribute zero (since in this case the integrand is
the trace of the product of a skew-symmetric and a symmetric matrix).
Therefore the right hand side consists of
4h-2
terms each of which numerically contribute with less than 1/4. This proves the corollary. Proof qf Theorem 9.15. G = Sp(2n,:rn.) the filling
It is straightforward to check that
is semi-simple so we can apply Theorem 9.12 using by geodesic simplices.
a
First let us reduce the
number of integration variables: In general for
G
semi-simple with maximal compact group
1 = -l J we exp :'1 G/K
and Cartan decomposition 'IT
0
ill
have the diffeomorphism
as in the remark following Proposition 9.11. 1
=
exp
0
-1
:
G/K
G
is an embedding such that the diagram
G/K 2-. G G/K commutes.
Then we have
Lemma 9.20.
For
Therefore
and
K
156
(9.21 ) where
P(gl,g2)
gl g2°
(that is,
P (gl ,g2)
1*P(8 K)
is the geodesic curve in
=
P (gl ,g2) (s)
Proof. in fact
J
=
] (P (nK) ) (g 1 ' g 2)
g 1tp (stp
-1
G/K
(g 2 0)
considered as a form on
P([8
K,8 K])
=
a
since
P
from
to
gl°
) ,
s E [0,1]) .
G
is actually exact;
is K-invariant, hence by (3.14)
and so (9.22) Now by (9.8) the geodesic 2-simplex
o(gl,g2)
6
2
G/K
is given
by (9.23) where
hs(x) = tp(stp
-1
(x)), x E G/K, s E [0,1].
Notice that
vanishes on the tangent fields along any curve of the form s E [0,1],
and since
10 o(gl,g2) 0 £i, i
= 1,2,
exp(sv),
is of this
form we conclude from (9.22) that
I I
6
2
0
1
(o(gl,g2)
(g l , g 2 ) *d ( l * P ( 8 K) )
6
0 £0)*1*P(8 K)
which is just (9.21). Now for
., =ft( 2n, lR)
G
=
Sp ( 2n, lR)
c
Gl (2n, lR) ,
is contained in
fy (2n, lR)
=
{x = C
= -"-(n)
The Lie algebra
=
M(2n, lR)
{x =
of
the Lie algebra as the set of matrices
C, t B K
=
U(n)
=
B}
is the subspace
157
wi th complement in
-f;r'( 2n, lR) : A,
(n)
tB =
B} .
is identified with the vectorspace of Hermitian
n
n
x
complex matrices (as in Example 5 of Chapter 4) by letting
=
x
X=
correspond to
A + iC.
In this notation the first Chern
2
E H (BU(n),lR) is given by the Chern-Wei 1 image of 1 the linear form P E 1 1 (U(n)) given by
class
c
=-
=-
(9.24)
P (X)
Now
is identified with
\l
:
G/K
G/K .... Gl ( 2n, lR)
tr (X)
1 2 'f[ tr (C)
=-
G n P(2n,lR)
1 4 'f[ tr (JX),
X E
.M(n).
via the map
given by g E G
(see G. Mostow [23, p. 20]). embedding
1 : G/K .... G 1 (p)
Also if let along
. p
p = p(s),
Under this identification the
above is given by
p,
pEG n P(2n,lR)
s E [0,1],
denote the derivative, i.e. the tangent vector field p,
Notice that the projection K(X)
For
G n P(2n;lR)
is a curve in
P E 1
1
(U(n))
-
t X) ,
K
:
1J-(2n) ....
X E
-
ff(2n).
given by (9.24) above the form
therefore takes the following form along a curve s E [0, 1 ] ,
in
is given by
Gnp ( 2n, lR) :
1*P(8 p
=
K)
p(s),
158
1 * P (8
K
) (p)
'" -
tr (J ( T -1
8n
t (T -1 i) ) ) , T '" P!;,
i -
so (9.25)
1*P(8 ) (f») '" K
Now suppose
p
1
-4"TI tr(JT
TO exp(SY)T
O'
T '" P
G n P(2n,JR),
is a geodesic in
p(s)
-1. T),
!;,
that is,
s E [0,1], YEt TO E G n P(2n,JR).
Then (9.26)
p
-1 .
-1
TO YT
p
is a constant in " . '" exp(2(s)), 2(s)
O
'" p(O)
-1
p(O)
Q
On the other hand, if we write
E 1 ' s E [0,1],
then
p(s)
(see Helgason [14,
Chapter II, Theorem 1.7]): p
-1 .
l-exp(-ad 2) ad 2
p
(1
+ exp(-ad
(1
where again
T '" P tr(JT
Now since of
-1
2
2'
2
2
(
2 ) 2
2)) (T
Hence by
. 2
-1 . T),
(9.26)
2 -1 t ) '" tr(J(l + exp(-ad 2)) (Q)).
2 E S (2n, JR),
M(2n, JR)
Therefore
'" exp
+ exp (-ad
(Z)
ad Z
is a self adjoint transformation
with respect to the inner product
159
Z -1 - < J, (1 + exp (- ad 2") ) (Q) > Z
«1+exp(-ad2"))
-1
(J),Q>.
Now it is easy to see that tr(JT
-1.
I
T) = -«1 +exp(-Z))
-1
J,Q>
tr (J (1 + exp (- Z) ) -1 Q} tr (J (1 + P-1 ) - 1 p (0) - 1 P(0) ) .
=
Finally let
p
t g1° = g1 g1
to
p(s), s E [0,1],
be the geodesic curve from
t t g1 g2 g2 g1'
g1 g2°
Then
that is,
and we conclude
t
since
g1J
Jg
-1
1.
Theorem 9.15 now clearly follows from Theorem
9.12 together with (9.21) and (9.25). Remark.
It would be interesting to know if the expression
in (9.16) is bounded also for Exercise 1.
Let
X
n > 1.
be a connected locally path-connected
and semi-locally 1-connected topological space so that it has a universal covering space group of a) n :
X
X
and let
Suppose X
G
a : r
is a principal
n :
X
X.
Let
r
be the fundamental
be any Lie group. G
is a homomorphism.
Show that
r-bundle (therefore called a principal
r-covering) and that the associated extension to a principal
G-
160
bundle
TI
b)
:
a.
E
X
a.
Suppose
is a flat G-bundle.
=
r
so that
{1}
=
X
X
Show that every flat G-bundle is trivial. the corresponding c)
Gd-bundle
a. :
r
-->
TI
:
X -->
X
(Hint:
Observe that
is a covering space of
Show that in general every flat
the extension of
is simply connected.
to
G-bundle
Let
G
group.
be a maximal compact subgroup.
Let
K
G
-->
w E InvGA*(G/K),
For
G
be a homomorphism from a discrete
J*w E H*(BGd,Th),
the element
defines a characteristic class for flat Let
TI
TI
Ea.
a.
M
:
-->
M
-->
Exercise 1a) and let bundle with fibre
M
be the corresponding flat Tla.: M xrG/K
G/K.
-->
Show that
of the characteristic class
is just b)
G/K) w
G-bundles.
be a differentiable principal
J*(w)
M
Tla.
in cohomology and that the pull-back
r
induces an isomorphism n*U*(w)(E)) E H*(M xrG/K,Th) a. a.
E H*(M,Th)
a. : r
-->
G
G/K
-->
r\G/K
TI
Again let
TI
= Mr
right action by
xg = g
J* (w) (Ea.) E H* (Mr, Th) whose lift to
the diagonal
G/K
-->
n a. : r \ (G/K x G/K)
-1
x
for
G/K x G/K -->
M . r)
M x G/K
-->
G/K.
x E G/K,
is just
is the covering space provided
r-action on
is represented in G/K
M x G/K
r
is discrete
be the associated
a.
flat G-bundle (first change the left
w
is represented in
is the inclusion of a discrete
of a manifold (this is actually the case and torsion free).
(see
be the associated fibre-
pulled back under the projection
Now suppose
r-covering
G-bundle
by the unique form whose lift to
subgroup such that
form
is
be a Lie group with finitely many
a. : r
A*(M x
X
relative to some homomorphism
G
components and let
and let
on
G.
Exercise 2.
a)
X).
w.
g E fl.
G/K
to a
Show that
A* (M r)
by the unique
(Hint:
Observe that
induces a section of the bundle
161
c) for
Again consider
P E Il(K),
by the form
G, rand
(E ) a where
w(P)
K
for
and show that A 21 (M )
r
is the curvature form of the (Hint:
direct proof by observing that
G of the principal
as in b)
is represented in
connection given in step I.
to
K
Either use b) or give a is the extension
11
a
r \G
K-bundle
r\G/K).
In particular,
dim G/K = 2k, = r
: r
G
1 r
1
J
for all
M
r
and
a
2
: r
G
2
be homomorphisms
are the fundamental groups of two
2
dimensional compact manifolds
M
1
and
M
2
and let
M be the corresponding flat a a 2 2 2 the Hirzebruch proportionality principle:
and
11
E
:
There is a real constant (9.28)
1
for all
P E Ik(K). G
and
Riemannian metrics induced from a left invariant metric on
G/K
Furthermore, if M. 1
=
=
Mr ' i i
vol(M
r1
=
f
r
and
1
1,2, as in b)
)/vol(M
r2)
where
are discrete subgroups of above then Mr.' i
=
c(a 1,a 2) =
1,2,
are given the
1
(which exists since
has an inner product which is invariant
under the adjoint action by e)
2
Now consider
K).
G = PSl(2,lR)
=
Sl(2,lR)/{±1}.
by isometries on the Poincare upper halfplane H
x + iy E iC I y > O}
{z
with Riemannian metric 1 Y
@ dx
+ dy @ dy) .
G
acts
162
The action is given by Z I----->
(az
+ b) /
+ d),
(c z
z E: