Lectures on the

Geometry of
Manifolds
Third Edition
Other World Scientific Titles by the Author

Introduction to Real Analysis

ISBN: 978-981-121-038-9
ISBN: 978-981-121-075-4 (pack)
 Lectures on the

Geometry of
Manifolds
Third Edition

Liviu I Nicolaescu
University of Notre Dame, USA

World Scientific
NEW JERSEY . LONDON . SINGAPORE . BEIJING . SHANGHAI . HONG KONG . TAIPEI . CHENNAI . TOKYO
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA ojice: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

Library of Congress Cataloging-in-Publication Data

Names: Nicolaescu, Liviu I., author.
Title: Lectures on the geometry of manifolds / Liviu I. Nicolaescu, University of Notre Dame, USA.
Description: Third edition. | New Jersey : World Scientific Publishing Co., [2021] |
Includes bibliographical references and index.
Identifiers: LCCN 2020037906 | ISBN 9789811214813 (hardcover) | ISBN 9789811215957 (paperback) |
ISBN 9789811214820 (ebook) I ISBN 9789811214837 ( b o o k other)
Subjects: LCSH: Geometry, Differential. | Manifolds (Mathematics)
Classification: LCC QA649 .N53 2021 | DDC 516.3/62--dc23
LC record available at https://lccn.loc.gov/2020037906

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.

Copyright © 2021 by World Scientific Publishing Co. Pte. Ltd.

All rights reserved. This book, or parts thereof may not be reproduced in any form or by any means, electronic or
mechanical, including photocopying, recording or any information storage and retrieval system now known or to
be invented, without written permission from the publisher

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center,
Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from
the publisher.

For any available supplementary material, please visit

https://www.worldscientiflc.com/worldscibooks/ l0.1142/ l l 680#t=suppl

Desk Editor: Liu Yumeng

Printed in Singapore
To the magical summer nights of my hometown
This page intentionally left blank
 Preface

Shape is a fascinating and intriguing subject which has stimulated the imagination of many
people. It suffices to look around to become curious. Euclid did just that and came up
with the first pure creation. Relying on the common experience, he created an abstract
world that had a life of its own. As the human knowledge progressed so did the ability
of formulating and answering penetrating questions. In particular, mathematicians started
wondering whether Euclid's "obvious" absolute postulates were indeed obvious and/or ab-
solute. Scientists realized that Shape and Space are two closely related concepts and asked
whether they really look the way our senses tell us. As Felix Klein pointed out in his Er-
langen Program, there are many ways of looking at Shape and Space so that various points
of view may produce different images. In particular, the most basic issue of "measuring
the Shape" cannot have a clear cut answer. This is a book about Shape, Space and some
particular ways of studying them.
Since its inception, the differential and integral calculus proved to be a very versatile
tool in dealing with previously untouchable problems. It did not take long until it found
uses in geometry in the hands of the Great Masters. This is the path we want to follow in
the present book.
In the early days of geometry nobody wonted about the natural context in which the
methods of calculus "feel at home". There was no need to address this aspect since for the
particular problems studied this was a non-issue. As mathematics progressed as a whole
the "natural context" mentioned above crystallized in the minds of mathematicians and it
was a notion so important that it had to be given a name. The geometric objects which can
be studied using the methods of calculus were called smooth manifolds. Special cases of
manifolds are the curves and the surfaces and these were quite well understood. B. Riemann
was the first to note that the low dimensional ideas of his time were particular aspects of a
higher dimensional world.
The first chapter of this book introduces the reader to the concept of smooth mani-
fold through abstract definitions and, more importantly, through many we believe relevant
examples. In particular, we introduce at this early stage the notion of Lie group. The
main geometric and algebraic properties of these objects will be gradually described as we
progress with our study of the geometry of manifolds. Besides their obvious usefulness in
geometry, the Lie groups are academically very friendly. They provide a marvelous testing

vii
viii Lectures on the Geometry of Manzfolds

ground for abstract results. We have consistently taken advantage of this feature through-
out this book. As a bonus, by the end of these lectures the reader will feel comfortable
manipulating basic Lie theoretic concepts.
To apply the techniques of calculus we need things to derivate and integrate. These
"things" are introduced in Chapter 2. The reason why smooth manifolds have many dif-
ferentiable objects attached to them is that they can be locally very well approximated
by linear spaces called tangent spaces. Locally, everything looks like traditional calculus.
Each point has a tangent space attached to it so that we obtain a "bunch of tangent spaces"
called the tangent bundle. We found it appropriate to introduce at this early point the notion
of vector bundle. It helps in structuring both the language and the thinking.
Once we have "things to derivate and integrate" we need to know how to explicitly
perform these operations. We devote the Chapter 3 to this purpose. This is perhaps one of
the most unattractive aspects of differential geometry but is crucial for all further develop-
ments. To spice up the presentation, we have included many examples which will found
applications in later chapters. In particular, we have included a whole section devoted
to the representation theory of compact Lie groups essentially describing the equivalence
between representations and their characters.
The study of Shape begins in earnest in Chapter 4 which deals with Riemann manifolds.
We approach these objects gradually. The first section introduces the reader to the notion of
geodesics which are defined using the Levi-Civita connection. Locally, the geodesics play
the same role as the straight lines in an Euclidian space but globally new phenomena arise.
We illustrate these aspects with many concrete examples. In the final part of this section
we show how the Euclidian vector calculus generalizes to Riemann manifolds.
The second section of this chapter initiates the local study of Riemann manifolds. Up
to first order these manifolds look like Euclidian spaces. The novelty arises when we study
"second order approximations" of these spaces. The Riemann tensor provides the complete
measure of how far is a Riemann manifold from being flat. This is a very involved object
and, to enhance its understanding, we compute it in several instances: on surfaces (which
can be easily visualized) and on Lie groups (which can be easily formalized). We have also
included Caftan's moving frame technique which is extremely useful in concrete computa-
tions. As an application of this technique we prove the celebrated Theorem Egregium of
Gauss. This section concludes with the first global result of the book, namely the Gauss-
Bonnet theorem. We present a proof inspired from [38] relying on the fact that all Riemann
surfaces are Einstein manifolds. The Gauss-Bonnet theorem will be a recurring theme in
this book and we will provide several other proofs and generalizations.
One of the most fascinating aspects of Riemann geometry is the intimate correlation
"local-global". The Riemann tensor is a local object with global effects. There are cur-
rently many techniques of capturing this correlation. We have already described one in the
proof of Gauss-Bonnet theorem. In Chapter 5 we describe another such technique which
relies on the study of the global behavior of geodesics. We felt we had the moral obliga-
tion to present the natural setting of this technique and we briefly introduce the reader to
the wonderful world of the calculus of variations. The ideas of the calculus of variations
 Preface ix

produce remarkable results when applied to Riemann manifolds. For example, we explain
in rigorous terms why "very curved manifolds" cannot be "too long".
In Chapter 6 we leave for a while the "differentiable realm" and we briefly discuss the
fundamental group and covering spaces. These notions shed a new light on the results
of Chapter 5. As a simple application we prove Weyl's theorem that the semisimple Lie
groups with definite Killing form are compact and have finite fundamental group.
Chapter 7 is the topological core of the book. We discuss in detail the cohomology of
smooth manifolds relying entirely on the methods of calculus. In writing this chapter we
could not, and would not escape the influence of the beautiful monograph [22], and this
explains the frequent overlaps. In the first section we introduce the DeRham cohomology
and the Mayer-Vietoris technique. Section 2 is devoted to the Poincaré duality, a feature
which sets the manifolds apart from many other types of topological spaces. The third sec-
tion offers a glimpse at homology theory. We introduce the notion of (smooth) cycle and
then present some applications: intersection theory, degree theory, Thom isomorphism and
we prove a higher dimensional version of the Gauss-Bonnet theorem at the cohomologi-
cal level. The fourth section analyzes the role of symmetry in restricting the topological
type of a manifold. We prove Elie Cartan's old result that the cosmology of a symmetric
space is given by the linear space of its bi-invariant forms. We use this technique to com-
pute the lower degree cohomology of compact semisimple Lie groups. We conclude this
section by computing the cosmology of complex Grassmannians relying on Weyl's inte-
gration formula and Schur polynomials. The chapter ends with a fifth section containing a
concentrated description of tech cohomology.
Chapter 8 is a natural extension of the previous one. We describe the Chern-Weil con-
struction for arbitrary principal bundles and then we concretely describe the most important
examples: Chern classes, Pontryagin classes and the Euler class. In the process, we com-
pute the ring of invariant polynomials of many classical groups. Usually, the connections in
principal bundles are defined in a global manner, as horizontal distributions. This approach
is geometrically very intuitive but, at a first contact, it may look a bit unfriendly in concrete
computations. We chose a local approach build on the reader's experience with connec-
tions on vector bundles which we hope will attenuate the formalism shock. In proving the
various identities involving characteristic classes we adopt an invariant theoretic point of
view. The chapter concludes with the general Gauss-Bonnet-Chern theorem. Our proof is
a variation of Chern's proof.
Chapter 9 is the analytical core of the book.' Many objects in differential geometry
are defined by differential equations and, among these, the elliptic ones play an important
role. This chapter represents a minimal introduction to this subject. After presenting some
basic notions concerning arbitrary partial differential operators we introduce the Sobolev
spaces and describe their main functional analytic features. We then go straight to the core
of elliptic theory. We provide an almost complete proof of the elliptic a priori estimates
(we left out only the proof of the Calderon-Zygmund inequality). The regularity results are
then deduced from the a priori estimates via a simple approximation technique. As a first

lin the current edition, this is Chapter 10.

X Lectures on the Geometry of Manzfolds

application of these results we consider a Kazhdan-Warner type equation which recently

found applications in solving the Seiberg-Witten equations on a Kohler manifold. We adopt
a variational approach. The uniformization theorem for compact Riemann surfaces is then
a nice bonus. This may not be the most direct proof but it has an academic advantage. It
builds a circle of ideas with a wide range of applications. The last section of this chapter
is devoted to Fredholm theory. We prove that the elliptic operators on compact manifolds
are Fredholm and establish the homotopy invariance of the index. These are very gen-
eral Hodge type theorems. The classical one follows immediately from these results. We
conclude with a few facts about the spectral properties of elliptic operators.
The last chapter is entirely devoted to a very important class of elliptic operators namely
the Dirac operators. The important role played by these operators was singled out in the
works of Atiyah and Singer and, since then, they continue to be involved in the most dra-
matic advances of modern geometry. We begin by first describing a general notion of Dirac
operators and their natural geometric environment, much like in [14]. We then isolate a
special subclass we called geometric Dirac operators. Associated to each such operator is
a very concrete WeitzenbOck formula which can be viewed as a bridge between geometry
and analysis, and which is often the source of many interesting applications. The abstract
considerations are backed by a full section describing many important concrete examples.
In writing this book we had in mind the beginning graduate student who wants to spe-
cialize in global geometric analysis in general and gauge theory in particular. The second
half of the book is an extended version of a graduate course in differential geometry we
taught at the University of Michigan during the winter semester of 1996.
The minimal background needed to successfully go through this book is a good knowl-
edge of vector calculus and real analysis, some basic elements of point set topology and
linear algebra. A familiarity with some basic facts about the differential geometry of curves
of surfaces would ease the understanding of the general theory, but this is not a must. Some
parts of the chapter on elliptic equations may require a more advanced background in func-
tional analysis.
The theory is complemented by a large list of exercises. Quite a few of them contain
technical results we did not prove so we would not obscure the main arguments. There
are however many non-technical results which contain additional information about the
subjects discussed in a particular section. We left hints whenever we believed the solution
is not straightforward.

Personal note It has been a great personal experience writing this book, and I sincerely
hope I could convey some of the magic of the subject. Having access to the remarkable
science library of the University of Michigan and its computer facilities certainly made my
job a lot easier and improved the quality of the final product.
I learned differential equations from Professor Viorel Barbu, a very generous and en-
thusiastic person who guided my first steps in this field of research. He stimulated my
curiosity by his remarkable ability of unveiling the hidden beauty of this highly technical
subject. My thesis advisor, Professor Tom Parker, introduced me to more than the funda-
mentals of modern geometry. He played a key role in shaping the manner in which I regard
 Preface xi

mathematics. In particular, he convinced me that behind each formalism there must be a

picture, and uncovering it, is a very important part of the creation process. Although I did
not directly acknowledge it, their influence is present throughout this book. I only hope the
filter of my mind captured the full richness of the ideas they so generously shared with me.
My friends Louis Funar and Gheorghe Ionesez read parts of the manuscript. I am
grateful to them for their effort, their suggestions and for their friendship. I want to thank
Arthur Greenspoon for his advice, enthusiasm and relentless curiosity which boosted my
spirits when I most needed it. Also, I appreciate very much the input I received from the
graduate students of my "Special topics in differential geometry" course at the University
of Michigan which had a beneficial impact on the style and content of this book.
At last, but not the least, I want to thank my family who supported me from the begin-
ning to the completion of this project.

Ann Arbor, 1996.

Preface to the second edition

Rarely in life is a man given the chance to revisit his "youthful indiscretions". With this
second edition I have been given this opportunity, and I have tried to make the best of it.
The first edition was generously sprinkled with many typos, which I can only attribute
to the impatience of youth. In spite of this problem, I have received very good feedback
from a very indulgent and helpful audience, from all over the world.
In preparing the new edition, I have been engaged on a massive typo hunting, supported
by the wisdom of time, and the useful comments that I have received over the years from
many readers. I can only say that the number of typos is substantially reduced. However,
experience tells me that Murphy's Law is still at work, and there are still typos out there
which will become obvious only in the printed version.
The passage of time has only strengthened my conviction that, in the words of Isaac
Newton, "in learning the sciences examples are of more use than precepts". The new
edition continues to be guided by this principle. I have not changed the old examples, but
I have polished many of my old arguments, and I have added quite a large number of new
examples and exercises.
The only major addition to the contents is a new chapter (Chapter 9) on classical integral
geometry. This is a subject that captured my imagination over the last few years, and since
the first edition developed all the tools needed to understand some of the juiciest results
in this area of geometry, I could not pass the chance to share with a curious reader my
excitement about this line of thought.
One novel feature in our presentation of the classical results of integral geometry is the
use of tame geometry. This is a recent extension of the better know area of real algebraic
geometry which allowed us to avoid many heavy analytical arguments, and present the
geometric ideas in as clear a light as possible.
Notre Dame, 2007.
2He passed away while I was preparing the second edition. He was the ultimate poet of mathematics.
xii Lectures on the Geometry of Manifolds

Preface to the third edition

I started writing the first edition 25 years ago as a fresh PhD and I was warned by
many "adults" that this was a wrong career move. I embarked in this project with all the
energy, enthusiasm, inexperience and confidence of young age. It was meant to be an
honest presentation of the basic elements of differential geometry used in global analysis.
By "honest" presentation I understood that I should include clear and detailed explanations
for many of the folklore results, examples and points of view that are harder to trace in the
literature and had helped in my research.
It was an exciting experience writing the first edition and I was rewarded for my effort it
in many ways. There was an immediate reward for the time spent immersed in the minutia
of many examples and proofs. The detailed understanding I achieved allowed me to push
my own research to new directions and in such a depth that I did not believe I was capable
of at that age.
There was a long term reward since it turns out that the useful facts and examples and
computations that were harder to trace are still useful and harder to trace, but now I can
always turn to this book for details. Personal bias aside, I keep a copy of my book on my
desk since I frequently need to look up something in it.
There is a personal reward, as an author. I have been receiving input and acknowledg-
ments from many readers from all the corners of the world. I have implemented all the
corrections and suggestions I have received. A book lives through its readers and appar-
ently the present one is still alive.
So what is new in the third edition? There are some obvious additions reflecting my
current research interests. There is the new Chapter 11 on spectral geometry leading to the
original results in Subsection 11.4.5. This is the first place where they appear in published
form.
Describing new results in a monograph rather than in a research journal has allowed me
to go to a level of detail and provide perspective that is not possible in a journal. In particu-
lar, I have added two new Subsections 4.2.5 and 5.2.3 on Riemannian geometry containing
facts that, surprisingly, are not familiar to many geometers. These play an important role
in Subsection 11.4.5.
There are less conspicuous changes. I have been using various parts of the book for
graduate courses I taught over the years. The new edition contains the useful feedback
I have received from my students. I have enhanced and cleaned many proofs and I have
added new examples. Of course I have corrected the typos pointed out to me by readers. It
is likely I have introduced new ones.
Notre Dame, 2019.
 Contents

Preface vii

1. Manifolds 1
1.1 Preliminaries 1
1.1.1 Space and Coordinatization 1
1.1.2 The implicit function theorem 3
1.2 Smooth manifolds . 6
1.2.1 Basic definitions . . . . . . . . 6
1.2.2 Partitions of unity 9
1.2.3 Examples . 10
1 2 4 How many manifolds are there? 20

2 • Natural Constructions on Manifolds 23

2.1 The tangent bundle 23
2.1.1 Tangent spaces. • . . . . o f o f 23
2.1.2 The tangent bundle . . . . . . . 27
2.1.3 Transversality 29
2.1.4 Vector bundles 33
2.1.5 Some examples of vector bundles 37
2.2 A linear algebra interlude 41
2.2.1 Tensor products 41
2.2.2 Symmetric and skew-symmetric tensors 46
2.2.3 The "super" slang 53
2.2.4 Duality................ 56
2.2.5 Some complex linear algebra . . . . . 64
2.3 Tensor fields 67
2.3.1 Operations with vector bundles 67
2.3.2 Tensor fields 69
2.3.3 Fiber bundles . . . . . . . . . . 73

xiii
xiv Lectures on the Geometry of Manifolds

3. Calculus on Manifolds 79
3.1 The Lie d e r i v a t i v e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.1.1 Flows on manifolds 79
3.1.2 The Lie d e r i v a t i v e . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.1.3 Examples . 86
3.2 Derivations of Q' (M) 88
3.2.1 The exterior derivative . . . . . . . . . . . . . . . . . . . . . . . 88
3.2.2 E x a m p l e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.3 Connections on vector bundles . . . . . . . . . . . . . . . . . . . . . . . 94
3.3.1 Covariant derivatives . . . . . . . . . . . . . . . . . . . . . . . . 94
3.3.2 Parallel t r a n s p o r t . . . . . . . . . . . . . . . . . . . 100
3.3.3 The curvature of a connection . . . . . . . . . . . . . 101
3.3.4 Holonomy....................... 105
3.3.5 The Bianchi identities . . . . . . . . . . . . . . . . . 108
3.3.6 Connections on tangent bundles . . . . . . . . . . . . 109
3.4 Integration on manifolds . . . . . . . . . . . . . . . . . . . . 111
3.4.1 Integration of 1-densities . . . . . . . . . . . . . . . 111
3.4.2 Orientability and integration of differential forms . . . 115
3.4.3 Stokes'formula 123
3.4.4 Representations and characters of compact Lie groups . . . . . . 127
3.4.5 Fiberedca1culus..........................133

4. Riemannian Geometry 139

4.1 Metric properties 139
4.1.1 Definitions and examples . . . . . . . . . . 139
4.1.2 The Levi-Civita connection 143
4. 1.3 The exponential map and normal coordinates 147
4. l .4 The length minimizing property of geodesics 150
4.1.5 Calculus on Riemann manifolds . . . . . . 155
4.2 The Riemann curvature 165
4.2.1 Definitions and properties . . . . . . . . . . . . 165
4.2.2 E x a m p l e s . . . . . . . . . . . . . . . . . . . . . 169
4.2.3 Ca1"tan's moving frame method . . . . . . . . . 172
4.2.4 The geometry of submanifolds . . . . . . . . . 175
4.2.5 Correlators and their geometry 182
4.2.6 The Gauss-Bonnet theorem for oriented surfaces 192

5. Elements of the Calculus of Variations 201

5.1 The least action principle 201
5.1.1 The 1-dimensional Euler-Lagrange equations 201
5.1.2 Noether's conservation principle 207
5.2 The variational theory of geodesics . . . 210
 Contents xv

5.2.1 Variational formulae 211

5.2.2 Jacobi fields 214
5.2.3 The Hamilton-Jacobi equations 220

6. The Fundamental Group and Covering Spaces 227

6.1 The fundamental group . . ..228
6.1.1 Basic n o t i o n s . . . . . . . . . . . . . . ..228
6.1.2 Of categories and functors . . . ... o n .232
6.2 Covering Spaces . . . . . . . . . . . . . .. .. .. 233
6.2.1 Definitions and examples . . . . . . . ..233
6.2.2 Unique lifting property . . . . . . . . ..235
6.2.3 Homotopy lifting property . . . t o o . . .236
6.2.4 On the existence of lifts .. . . . . 237
6.2.5 The universal cover and the fundamental group ..239

7. Cohomology 241
7.1 DeRham c o h o m o l o g y . . . . . . . . . . . . . . . . . . . . . . . . . . 241
7.1.1 Speculations around the Poincaré lemma . . . . . . . 241
7.1.2 (tech vs. DeRham . . . . . . . . . . . 245
7.1.3 Very little homological algebra . . . . . . . . . . . . . . . 247
7. 1.4 Functorial properties of the DeRham cohomology . . 254
7.1.5 Some simple examples . . . . . . . . . . . . . . . . . . . . . . . 257
7.1.6 The Mayer-Vietoris principle . . 259
7.1.7 The Kenneth formula . . . . . . . . . .. . 262
7.2 The Poincaré duality . . . . . . . . . . . . . . . . 265
7.2.1 Cohomology with compact supports • . . 265
7.2.2 The Poincaré duality . . . . . . • 268
7.3 Intersection theory . . . . . . . . . . . . . . . • • 272
7.3.1 Cycles and their duals . . . . . • 272
7.3.2 Intersection theory . . . . . . . . . . . • • 277
7.3.3 The topological degree . . . . . . . . . • • 282
7.3.4 The Thom isomorphism theorem 284
7.3.5 Gauss-Bonnet revisited . . . . 287
7.4 Symmetry and topology . . 291
7.4.1 Symmetric spaces . . . . . . . . . . . 291
7.4.2 Symmetry and cosmology . . . . . . . . . . . . . . 294
7.4.3 The cosmology of compact Lie groups 298
7.4.4 Invariant forms on Grassmannians and Weyl's integral formula 299
7.4.5 The Poincaré polynomial of a complex Grassmannian 306
7.5 éechcohomology................ 312
7.5.1 Sheaves and presheaves . . . . 313
7.5.2 tech cosmology . . . . . . . 317
xvi Lectures on the Geometry of Manzfolds

8. Characteristic Classes 329

8.1 Chern-Weil theory 329
8.1.1 Connections in principal G-bundles 329
8.1.2 G-vector bundles . 335
8.1.3 Invariant polynomials 336
8. 1.4 The Chern-Weil Theory 339
8.2 Important examples 343
8.2.1 The invariants of the torus T" 343
8.2.2 Chern classes . 343
8.2.3 Pontryagin classes 346
8.2.4 The Euler class . 348
8.2.5 Universal classes . 351
8.3 Computing characteristic classes 357
8.3.1 Reductions 358
8.3.2 The Gauss-Bonnet-Chern theorem 363

9. Classical Integral Geometry 373

9.1 The integral geometry of real Grassmannians . ..n o n o o.373
f n o

9.1.1 C o - a r e a f o r m u l a e . . . . . . . . . . . . . . . . . . . . . . . . . . 373
9. 1.2 Invariant measures on linear Grassmannians . . . . . o f . . 386 O .

9. 1.3 Affine Grassmannians . 395

9.2 G a u s s - B o n n e t a g a i n ? ! ' ? . . . . . . . . . . . . . . . . . . . . . . . . . . . 398
9.2.1 The shape operator and the second fundamental form . . . . . . 398
9.2.2 The Gauss-Bonnet theorem for hypersurfaces of an Euclidean
space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
9.2.3 Gauss-Bonnet theorem for domains in an Euclidean space . 405
9.3 Curvature m e a s u r e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
9.3.1 Tame geometry . . . . . . . . . 409
9.3.2 Invariants of the orthogonal group . . . . . . . . . . . . . . . . 414
9.3.3 The tube formula and curvature measures . . . . . . . . . . . . 418
9.3.4 Tube formula => Gauss-Bonnet formula for arbitrary
submanifolds of an Euclidean space . . . . . . . . . . . . . . . 429
9.3.5 Curvature measures of domains in an Euclidean space 431
9.3.6 Crofton formulae for domains of an Euclidean space . 433
9.3.7 Crofton formulae for submanifolds of an Euclidean space 443

10. Elliptic Equations on Manifolds 451

10.1 Partial differential operators: algebraic aspects 451
10.1.1 Basic notions . 451
10.1.2 Examples . 457
10.1.3 Formal adjoints 459
10.2 Functional framework . 464
 Contents xvii

10.2.1 Sobolev spaces in RN . . . . . . . . . . . . . 464

10.2.2 Embedding theorems: integrability properties. . . . 471
10.2.3 Embedding theorems: differentiability properties . 476
10.2.4 Functional spaces on manifolds . . . . . . 480
10.3 Elliptic partial differential operators: analytic aspects 484
10.3.1 Elliptic estimates in RN 485
10.3.2 Elliptic regularity 489
10.3.3 An application: prescribing the curvature of surfaces 494
10.4 Elliptic operators on compact manifolds 504
10.4.1 Fredholm theory . 504
10.4.2 Spectral t h e o r y . . . . . . . . . . . . . . . . . . . . 513
10.4.3 Hodge theory . 518

11. Spectral Geometry 523

11.1 Generalized functions and currents . . . . . . . . . . . . .. . 523
11.1.1 Generalized functions and operations withy them . .. o f 0 . . . 523
11.1.2 C u r r e n t s . . . . . . . . . . . . . . . . . . . . . 528
11.1.3 Temperate distributions and the Fourier transform . . 529
11.1.4 Linear differential equations with distributional data 531
11.2 Important families of generalized functions 536
11.2.1 Some classical generalized functions on the real axis . 536
11.2.2 Homogeneous generalized functions . . . . . . . . . 542
11.3 The wave equation 548
11.3.1 Fundamental solutions of the wave 548
11.3.2 The wave family 550
11.3.3 Local parametrices for the wave equation with variable
coefficients....................... 567
I I .4 Spectral geometry 574
l 1.4.1 The spectral function of the Laplacian on a compact manifold 574
11.4.2 Short time asymptotic for the wave kernel 580
11.4.3 Spectral function asymptotics . . . . . . . . . . . . . . . . . 588
11.4.4 Spectral estimates of smoothing operators 593
11.4.5 Spectral perestroika . . . . . . . . . . . . . . . . . . . . . . 598

12. Dirac Operators 611

12.1 The structure of Dirac operators . . . . 611
12.1.1 Basic definitions and examples 611
12.1.2 Clifford algebras 614
12.1.3 Clifford modules: the even case 617
12.1.4 Clifford modules: the odd case 622
12.1.5 A look ahead . 623
12.1.6 The spin group . 625
xviii Lectures on the Geometry of Manifolds

12.1.7 The complex spin group 633

12.1.8 Low dimensional examples . . 635
12.1.9 Dirac bundles 640
12.2 Fundamental examples 644
12.2.1 The Hodge-DeRham operator . 644
12.2.2 The Hodge-Dolbeault operator 648
12.2.3 The spin Dirac operator . . . . 654
12.2.4 The spine Dirac operator 660

Bibliog raphy 669

Index 675
 Chapter 1

Manifolds

1.1 Preliminaries

1.1.1 Space and Coordinatization

Mathematics is a natural science with a special modus operandi. It replaces concrete natural
objects with mental abstractions which serve as intermediaries. One studies the properties
of these abstractions in the hope they reflect facts of life. So far, this approach proved to be
very productive.
The most visible natural object is the Space, the place where all things happen. The first
and most important mathematical abstraction is the notion of number. Loosely speaking,
the aim of this book is to illustrate how these two concepts, Space and Number, fit together.
It is safe to say that geometry as a rigorous science is a creation of ancient Greeks.
Euclid proposed a method of research that was later adopted by the entire mathematics.
We refer of course to the axiomatic method. He viewed the Space as a collection of points,
and he distinguished some basic objects in the space such as lines, planes etc. He then
postulated certain (natural) relations between them. All the other properties were derived
from these simple axioms.
Euclid's work is a masterpiece of mathematics, and it has produced many interesting
results, but it has its own limitations. For example, the most complicated shapes one could
reasonably study using this method are the conics and/or quadrics, and the Greeks certainly
did this. A major breakthrough in geometry was the discovery of coordinates by René
Descartes in the 17th century. Numbers were put to work in the study of Space.
Descartes' idea of producing what is now commonly referred to as Cartesian coordi-
nates is familiar to any undergraduate. These coordinates are obtained using a very special
method (in this case using three concurrent, pairwise perpendicular lines, each one en-
dowed with an orientation and a unit length standard. What is important here is that they
produced a one-to-one mapping

Euclidian Space -> R3 7 P I

> (;v(P)?y(Pl»2(P))-
We call such a process coordinatization. The corresponding map is called (in this case)
Cartesian system of coordinates. A line or a plane becomes via coordinatization an alge-
braic object, more precisely, an equation.

1
2 Lectures on the Geometry of Manzfolds

In general, any coordinatization replaces geometry by algebra and we get a two-way

correspondence

Study of Space < > Study of Equations.

The shift from geometry to numbers is beneficial to geometry as long as one has efficient
tools do deal with numbers and equations. Fortunately, about the same time with the in-
troduction of coordinates, Isaac Newton created the differential and integral calculus and
opened new horizons in the study of equations.
The Cartesian system of coordinates is by no means the unique, or the most useful coor-
dinatization. Concrete problems dictate other choices. For example, the polar coordinates
represent another coordinatization of (a piece of the plane) (see Figure 1.1).

Fig. 1.1 Polar coordinates.

P >(1~(p),9(p))@(0,oo) X ( or, fl-

This choice is related to the Cartesian choice by the well known formulae

at rosH y=rsin9. (1.1.1)

A remarkable feature of (l.1.l) is that ;L°(Pl and y(P) depend smoothly upon r(Pl and
@(p).
As science progressed, so did the notion of Space. One can think of Space as a config-
uration set, i.e., the collection of all possible states of a certain phenomenon. For example,
we know from the principles of Newtonian mechanics that the motion of a particle in the
ambient space can be completely described if we know the position and the velocity of the
particle at a given moment. The space associated with this problem consists of all pairs
(position, velocity) a particle can possibly have. We can coordinatize this space using six
 Manifolds 3

functions: three of them will describe the position, and the other three of them will de-
scribe the velocity. We say the configuration space is 6-dimensional. We cannot visualize
this space, but it helps to think of it as an Euclidian space, only "roomier".
There are many ways to coordinatize the configuration space of a motion of a particle,
and for each choice of coordinates we get a different description of the motion. Clearly, all
these descriptions must "agree" in some sense, since they all reflect the same phenomenon.
In other words, these descriptions should be independent of coordinates. Differential ge-
ometry studies the objects which are independent of coordinates.
The coordinatization process had been used by people centuries before mathematicians
accepted it as a method. For example, sailors used it to travel from one point to another
on Earth. Each point has a latitude and a longitude that completely determines its position
on Earth. This coordinatization is not a global one. There exist four domains delimited
by the Equator and the Greenwich meridian, and each of them is then naturally coordina-
tized. Note that the points on the Equator or the Greenwich meridian admit two different
coordinatizations which are smoothly related.
The manifolds are precisely those spaces which can be piecewise coordinatized, with
smooth correspondence on overlaps, and the intention of this book is to introduce the reader
to the problems and the methods which arise in the study of manifolds. The next section
is a technical interlude. We will review the implicit function theorem which will be one of
the basic tools for detecting manifolds.

1.1.2 The implicit function theorem

We gather here, with only sketchy proofs, a collection of classical analytical facts. For
more details one can consult [34].
Let X and Y be two Banach spaces and denote by L ( X , Y) the space of bounded linear
operators X -> Y. For example, if X = R", Y = RT", then L ( X , Y) can be identified
with the space of m X n matrices with real entries. For any set S we will denote by k g the
identity map S -> S.
Definition 1.1.1. Let F U C X -> Y be a continuous function (U is an open subset of
X). The map F is said to be (Fréchet) differentiable at u 6 U if there exists T e L ( X , Y )
such that
lIF<~@+ h) - F(u@) - T h e y I <»(llhllx) as h -> 0,
i.e.,
l
IIHIIx num0
lim h) ' F(Uol - Telly = 0.
h,->0

Loosely speaking, a continuous function is differentiable at a point if, near that point, it
admits a "best approximation" by a linear map.
When F is differentiable at U0 6 U, the operator T in the above definition is uniquely
determined by
d . 1
Th i - F(u0 -|- t h ) nm - (F(u0 -|- th) - F(u0l)
dt t->0 t
4 Lectures on the Geometry of Manzfolds

We will use the notation T = DU0 F and we will call T the Fréchet derivative of F at U0 .
Assume that the map F : U -> Y is differentiable at each point u 6 U. Then F is said
to be of class C1, i f the map u +-> D'LLF 6 L ( X , Y) is continuous. F is said to be of class
C2 if u +-> D'LLF is of class C1. One can define inductively Ck and coo (or smooth) maps.

Example 1.1.2. Consider F : U C R" -> Run. Using Cartesian coordinates

QUO .CE
1
7 . .. ,$ n
in R" and
1 m
'LL 'LL 7 ° .
. ,U

in Run, we can think of F as a collection of m functions on U

1 1 l
'LL 'u,(;13,...,a3"), ,um U'771(;U17°°'7a:n)'

The map F is differentiable at a point p = (p1, . . . ,1>") 6 U if and only if the functions u'll
are differentiable at p in the usual sense of calculus. The Fréchet derivative of F at p is the
linear operator DpF : R" -> RM given by the Jacobian matrix
31/,1
(al 5':1:2 (al (p)
3(7u,1 um
7 7 31/,2 nu2 51/,2
D,,F in 3:151 (p) 31/32 (al 3515" (p)
3(a21 7 7

(p)
sum
31/32 (p)
sum
3113" (p) _
The map F is smooth if and only if the functions ) are smooth.

Exercise 1.1.3. (a) Let 'LL C LGR" ,WI denote the set of invertible n X n matrices. Show
that U is an open subset of L(]R"', Rn).
(b) Let F : 'LL -> U be defined as A -> A-1_ Show that DAF(H) = -A-1HA-1 for any
n X n matrix H .
(c) Show that the Fréchet derivative of the map det : L(]Rn*,R") -> R, A »-> det A, at
A = Ilia" 6 L(Rn*, Rn) is given by tr H, i.e.,
d
L 0
det(J1Rn + t H ) = t H , oH € L(]R",]R").

Theorem 1.1.4 (Inverse function theorem). Let X , Y be two Banach spaces, U C X

open and F : U C X -> Y a smooth map. If at a point U 0 6 U the derivative DU0 F E
L(X,Y) is invertible, then there exits an open neighborhood U1 of U 0 in U such that
F(U1) is an open neighborhood of Uo = F(u0) in Y and F : U1 -> F(U1) is bijective,
with smooth inverse.

The spirit of the theorem is very clear: the convertibility of the derivative DU0 F "propa-
gates" locally to F because DU0 F is a very good local approximation for F.
 Manifolds 5

More formally, if we set T DuoF, then

F(u0 + h) = F(u0) + Th + ¢~(h),
where r(h) = o(llhll) as h -> 0. The theorem states that, for every 'u s u f i ciently close to
U0, the equation F('u.) = 'U has a unique solution 'LL = up -|- h, with h very small. To prove
the theorem one has to show that, for llv - 'U0 fly sufficiently small, the equation below
t)0+Th+v°(h) ='u

has a unique solution. We can rewrite the above equation as Th = 'U - U0 - r(h) or,
equivalently, as h = T"1('v - U0 - r(h)). This last equation is a fixed point problem that
can be approached successfully via the Banach fixed point theorem.

Theorem 1.1.5 (Implicit function theorem). Let X, Y , Z be Banach spaces, W C X ,

"//c Y open sets and
F :W X "If -> Z
a smooth map. Let (330, ygl 6 W X "//, and Z0 := F(:z;0, y0l. Set
$2 : "// -> Z, F2(Y) = FllimYl-
Assume that D90 F2 G L(Y, Z ) is invertible. Then there exist open neighborhoods U C W
of £170in X , V C 7/ of Yo in Y , and o smooth map G : U -> V such that the set S of
solution (as, y ) of the equation F(:c, y ) = Z0 which lie inside U X V can be identified with
the graph of G, i.e.,

{(a:,y)€UxV; F(w,y)=20}={(w,G(w)l G U x V ; a2€U}.

In pre-Bourbaki times, the classics regarded the coordinate y as a function of :u defined
implicitly by the equality F(a2, yl = Z0»

Proof. Consider the map

H:X X Y->X X Z, § = ( a : , y ) »-> (:c,F(;z3,y)).
The map H is a smooth map, and at £0 = (1U073/0) its derivative D€01-I : X X Y -> X X Z
has the block decomposition

D H=
so -D€oF1 D€0 F2
Above, DF1 (respectively DF2) denotes the derivative of ac »-> F(:L°, 3/ol (respectively the
derivative of y »-> F(x0, y)). The linear operator Do0 H is invertible, and its inverse has
the block decomposition
Hx 0
(D50H)
(D1oe)-1 o (D§0F1) (D 0F2)*
Thus, by the inverse function theorem, the equation (as, F(;u, y)) = (in, zo) has a unique
solution (M) = H-1(x, Zeal in a neighborhood of (5130, yo). It obviously satisfies 5: = as
and F(i;,g) = 20. Hence, the set {(x,y); F(ac,y) = z0} is locally the graph of ac +->
H-1 (ac, 20).
6 Lectures on the Geometry of Manzfolds

1.2 Smooth manifolds

1.2.1 Basic definitions

We now introduce the object which will be the main focus of this book, namely, the concept
of (smooth) manifold. It formalizes the general principles outlined in Subsection 1.1.1.

Definition 1.2.1. A smooth manifold of dimension m is a locally compact, second count-

able' Hausdorff space M together with the following collection of data (henceforth called
atlas or smooth structure) consisting of the following.

(a) An open cover {U'i}i@I of M,

(b) A collection of homeomorphisms

\Pi u -> \p(u-> C Run; i 6 I}

(called charts or local coordinates) such that, 'Pt (Up) is open in RM and, if UZ-O(/'j ¢ (Z),
then the transition map

\IJjo 'Pi 1 \P¢(U¢ M Url C RM \Ifjlu¢ O Url C RM

is smooth. (We say that the various Cha s are smoothly compatible, see Figure 1.2).

J
Ui J

'Vi W.
J

/
"Ii 'Vi-1
Run

Fig. 1.2 Transition maps.

l A second countable space is a topological space that admits a countable basis of open sets.
 Manifolds 7

Remark 1.2.2. (a) Each chart 'Pi : Uz- -> RM can be viewed as a collection of m functions
( X 1 7 ° ° ° 7 a 3 M 7 on Up,

l 1-1
1,2
(p)
(p) 1
We (p)

(p)
at771

Similarly, we can view another chart 'PJ as another collection of functions (y1, - - JW)-
•

The transition map 'Pa o 'Pi l can then be interpreted as a collection of maps
(x1,...,;u"") +-> l3/1(a31,...,a:""),...,yM(ac1,...,a:M)l.
(b) Since a manifold is a second countable space we can always work with atlases that are
at most countable.

The first and the most important example of manifold is R" itself. The natural smooth
structure consists of an atlas with a single chart, Ilia" : R" -> R". To construct more
examples we will use the implicit function theorem.

Definition 1.2.3. (a) Let M, N be two smooth manifolds of dimensions m and respectively
n. A continuous map f : M -> N is said to be smooth if, for any local charts QS on M
w w
and on N, the composition o f o (V1 (whenever this makes sense) is a smooth map
Run -> Rn_

(b) A smooth map f : M -> N is called a dyffeomorphism if it is invertible and its inverse
is also a smooth map.

Example 1.2.4. The map t »-> et is a diffeomorphism ( -oo, oo) -> (0,oo). The map
t +-> t3 is a homeomorphism R -> R, but it is not a diffeomorphism!

If M is a smooth m-dimensional manifold, we will denote by cOO (M) the linear space
of all smooth functions f : M -> R. Let us point out a simple procedure that we will use
frequently in the sequel. Suppose that f : M -> R is a smooth function. If (U, \If) is a
local chart on M so \P(Ul is an open subset of Run, then, by definition, the composition
f o \I/-1 : \P(Ul -> R is a smooth function on the open set \If(U) C R". If we denote
w
by x1, . . . ,513771 the canonical Euclidean coordinates on Run, then o \I/-1 is a function
depending on the m variables .CU1 . . m and we will use the notation f lx, . . . ,LU m
7 '

when referring to this function.

Remark 1.2.5. Let U be an open subset of the smooth manifold M (dim M m ) and
\If:U-HRM
a smooth, one-to-one map with open image and smooth inverse. Then \If defines local
coordinates over U compatible with the existing atlas of M. Thus (U, \I/) can be added to
the original atlas and the new smooth structure is diffeomorphic with the initial one. Using
Zermelo's Axiom we can produce a maximal atlas (no more compatible local chart can be
added to it).
8 Lectures on the Geometry of Manzfolds

Our next result is a general recipe for producing manifolds. Historically, this is how
manifolds entered mathematics.

Proposition 1.2.6. Let M be a smooth manzfold of dimension m and f l , 7 fk 6

cOO ( M ) . Define

Z Z»(f1, 7 fol {ppM; MP) = =f1=(p) =0}-

Assume that the functions fl, . . . , fk are functionally independent along Z, i.e., for each
p 6 Z, there exist local coordinates (331, . . . I dejined in a neighborhood of p in M
such that so (p) = 0, i = 1, . . . ,m, and the matrix
6f1 $f1 3f1
3x1 3x2 hmm
. a

95;
p ask 6 f k ask
3x1 3x2 hmm :n 1 :1;'"=0

has rank k. Then Z has a natural structure of smooth manifold of dimension m - k.

Proof. Step 1: Constructing the charts. Let Po E Z., and denote by ( X 1 7 ' ° ° 7 X M 7 local
coordinates near Po such that x'll (pg) = 0. One of the A: X k minors of the matrix
fifl
hmm
..
3:5 3fk
p
31/31 31/32 hmm 11:1 :nm 0
is nonzero. Assume this minor is determined by the last k columns (and all the k lines).
We can think of the functions fl, . . . , fk as defined on an open subset U of Run. Split
RM as R m - k X Rk, and set
al 1
(:z3,...,a:M -k
),
ml/
.. ("-'°+1, . . . , mM).
We are now in the setting of the implicit function theorem with
x=Rm-'" 7 y=1Re.'" 7 z=R'* 7

and F X X Y - > Z given by

f1l£ul
an-> 6 R/*

fk(x))
OF
In this case, DF2 = ( 5933" ) : Re' -> Rk is invertible since its determinant corresponds to
our nonzero minor. Thus, in a product neighborhood UP0 = UPo X U" Po
of Po, the set Z, is
the graph of some function
g _
. UPo) C RM- k >U/I
Po
CRk*,

i.e.,
Z O UPo (;u/,g(xI) ) 6 Rm-k X Re. co' 67 UPo) 7 at / small .
 Manifolds 9

We now define MY I Z FW UP() -> ]RM"* by

(22', 9(11') ) +¢P0> 513I E R""-k.

The map 1% is a local chart of Z near Po.

Step 2. The transition maps for the charts constructed above are smooth. The details are
left to the reader.

Exercise 1.2.7. Complete Step 2 in the proof of Proposition 1.2.6.

Definition 1.2.8. Let M be a m-dimensional manifold. A codimension k submamfold

of M is a subset S C M locally defined as the common zero locus of k functionally
independent functions fl, . . . , fk 6 COO(M). More precisely, this means that, for any
Po 6 S, there exists an open neighborhood U of Po e M and k functionally independent
smooth functions

f17...7fk¢2U- R
such that p 6 U m S if and only if fl (p) fA¢(Pl=0-

Proposition 1.2.6 shows that any submanifold N C M has a natural smooth structure
so it becomes a manifold per se. Moreover, the inclusion map z' : N <-> M is smooth.

Exercise 1.2.9. Suppose that M is a smooth m-dimensional manifold. Prive that S C M is

a codimension k-submanifold of M if and only if, for any Po 6 M, there exists a coordinate
chart (U, \I/) with local coordinates (x1, . . . ,arM) such that Po 6 U and

urns { p a U ; 1v1(p) 1vI"(p) 0 }~

1.2.2 Partitions of unity

This is a very brief technical subsection describing a trick we will extensively use in this
book. Recall that manifolds are locally compact, second countable topological spaces. As
such, they are paracompact so they admit continuous partitions of unity, see [30, §3.7]. A
much more precise result is in fact true.

Definition 1.2.10. Let M be a smooth manifold and (UJQQA an open cover of M. A

(smooth) partition of unity subordinated to this cover is a family (fB)Be8 C COOIMI
satisfying the following conditions.

(i) 0 S f B S 1.
(ii) EIQS : 8 -> A such that supp fa C U¢(/3)-
(iii) The family (supp f ii) is locally finite, i.e., any point as 6 M admits an open neighbor-
hood intersecting only finitely many of the supports supp fa .
(iv) EB f g l i ) = 1 for allay 6 M .
10 Lectures on the Geometry of Manifolds

We include here for the reader's convenience the basic existence result concerning par-
titions of unity. For a proof we refer to [128].

Proposition 1.2.11. (a) For any open cover U = ( U ) o € A of a smooth manifold M there
exists at least one smooth partition of unity IJBIBQ8 subordinated to U such that supp f B
is compact for any B.
(b) If we do not require compact supports, then we can _lind a partition of unity in which
'B = A and QUO = HA.

Exercise 1.2.12. Let M be a smooth manifold and S C M a closed submanifold, i.e., S is

a closed subset of M. Prove that the restriction map

7:Croo(M)->CYOO(S) f *->fls

is surjective. Deduce that for any finite set X C M and any function g : F -> R there exists
a smooth, compactly supported function f : M -> R such that f(:z:) = glatl, Van 6 X .

1.2.3 Examples
Manifolds are everywhere, and in fact, to many physical phenomena which can be modelled
mathematically one can naturally associate a manifold. On the other hand, many problems
in mathematics find their most natural presentation using the language of manifolds. To
give the reader an idea of the scope and extent of modern geometry, we present here a she
list of examples of manifolds. This list will be enlarged as we enter deeper into the study
of manifolds.

Example 1.2.13. (The n-dimensional sphere) This is the codimension l submanifold of

Rn+l given by the equation
n

I I
:u 2
(W T'
2
, at
0
.CU 7 7 so'n) 6 R'n+l
'i 0

One checks that, along the sphere, the differential of laval2 is nowhere zero, so by Proposi-
tion 1.2.6, S" is indeed a smooth manifold. In this case one can explicitly construct an atlas
(consisting of two charts) which is useful in many applications. The construction relies on
stereographic projections.
Let N and S denote the North and resp. South pole of S" (N = (0, . . . ,0, l) 6 lR"+1,
S = (0, . . . ,0, -1) 6 R"+1). Consider the open sets Un = S"\{N} and Us = S"\{S}.
They form an open cover of S". The stereographic projection from the North pole is the
map uN : Un -> R" such that, for any P 6 Un, the point oN (P) is the intersection of the
line NP with the hyperplane {;u" = 0} 2 R" .
The stereographic projection from the South pole is defined similarly.

.. For P Q U n we denote by lY1lPl)

For Q GUs we denote by (zwQ),
7 y"(P) ) the coordinates of uN (P).
z"(Q) ) the coordinates of us (Q).
7
 Manifolds 11

A simple argument shows the map

(y1(p),...,y"(p)) »-> (21(P)7...72"(Pl)7 P G Un VW Us,

is smooth (see the exercise below). Hence {(UN, oN), (US, 0g)} defines a smooth struc-
tune on S".

Exercise 1.2.14. Show that the functions y Zi constructed in the above example satisfy
ni
Zi
y vi = 1, ,n.
lE]=1lY]l2
Example 1.2.15. (The n-dimensional torus) This is the codimension n submanifold of the
vector space R211 with Cartesian coordinates (al , y1, , can, yn), defined by the equalities
2 2 2
£I31+?/1 Mn 2
-l-yn = 1.
Note that T1 is diffeomorphic with the 1-dimensional sphere Sl (unit circle). As a set T"
is a direct product of n circles (see Figure 1.3)
T"={=v21+yI=1} X {:u2 2
Yn 1} S1 X X 51.

I I

Fig. 1.3 The 2-dimensional torus.

The above example suggests the following general construction.

Example 1.2.16. Let M and N be smooth manifolds of dimension m and respectively

n. Then their topological direct product has a natural structure of smooth manifold of
dimension m - i - n .

Example 1.2.17. (The connected sum of two manifolds) Let M1 and M2 be two manifolds
of the same dimension m. Pick Pi 6 Mt (i = 1, 2), choose small open neighborhoods Up
of Pt in M and then local charts QM identifying each of these neighborhoods with 8220),
the ball of radius 2 in RM.
Let Vi C Up correspond (via ) to the annulus {1/2 < lac°| < 2} C Run. Consider
at
go: {1/2<lxl <2}->{l/2<l33l <2}, ( )
I II2
The action of <l> is clear: it switches the two boundary components of {1/2 < laval < 2},
and reverses the orientation of the radial directions.
12 Lectures on the Geometry of Manifolds

Now "glue" V1 to VI using the "prescription" given by $2 1 o QUO o zp1 : V1 -> VI. In
this way we obtain a new topological space with a natural smooth structure induced by
the smooth structures on Mi. Up to a diffeomeorphism, the new manifold thus obtained is
independent of the choices of local coordinates ([24]), and it is called the connected sum of
M1 and M2 and is denoted by M1#M2 (see Figure 1.4).

V V2
1
l
\
I
I
I

WI
'T
l l
l l
I
I1 I

Fig. 1.4 Connected sum of tori.

Example 1.2.18. (The real projective space ]RIP'") As a topological space RIP" is the quo-
tient of Rn-I-1 modulo the equivalence relation

as y ~:de6> EA E R* : at Ay.

The equivalence class of as = (xO, . . . , in 6 R \ {0} is usually denoted by

ac0,...,;z:" . Alternatively, RIP" is the set of all lines (directions) in lR"+1. Tradition-
ally, one attaches a point to each direction in ]R"+1, the so called "point at infinity" along
that direction, so that RIP" can be thought as the collection of all points at infinity along all
the directions in lRTL+1.
The space RIP" has a natural structure of smooth manifold. To describe it consider the
sets

Up: [ a c 0 , . . . , a 3 " ] € R P " ; m k ¢ 0}, If 0, ,n.

Now define
0 wk: k
] » - > (az / 7 . ° 7 ; U k
-1 /
'l,Dk :Up->Rn [110 as
, ...,x"
 Manifolds 13

The maps wk define local coordinates on the projective space. The transition map on the
overlap region Up O Um = {[;z:0 . . . , asn] , ;z:k;u"" 96 0} can be easily described. Set
7

¢k([£U07 . . . ,we
_ (€1,..-,€nl7 ¢m(llv0 7 ° ° ° 7 ; U n D = (m,---n77)-
The equality

[¢v°». • ' ,iv"] I [§1, ' • • 7g/<7-17 l a , 7 in] [7117 7 77771-17 17 U m ) 7 Un1
immediately implies (assume Is < m)

61 = u1/Dk, 7 Uk-1/Uk, §k+1 Up

fk = U k + 1 / U k , 7 17m-1/nk, €m-1 I 1/*71<: (1.2.1)
am I 77'rn77ka 7 Un/Wk
This shows the map Wk, o ism-1 is smooth and proves that RIP" is a smooth manifold. Note
that when n = 1, RIP1 is diffeomorphic with Sl. One way to see this is to observe that
the projective space can be alternatively described as the quotient space of S" modulo the
equivalence relation which identifies antipodal points.

Example 1.2.19. (The complex projective space ClP'") The definition is formally identical
to that of RIP". CIP" is the quotient space of C"--1 \ {0} modulo the equivalence relation
def
at f\J
y '\
E1/\€lC*: at Ay.

The open sets Up are defined similarly and so are the local charts Wk : Up -> CC". They
satisfy transition rules similar to (1.2.l) so that CIP" is a smooth manifold of dimension 271.

Exercise 1.2.20. Prove that (CIP1 is diffeomorphic to S2 .

In the above example we encountered a special (and very pleasant) situation: the gluing
maps not only are smooth, they are also holomorphic as maps W o ism - 1 : U -> V where
U and V are open sets in lcn This type of gluing induces a "rigidity" in the underlying
manifold and it is worth distinguishing this situation.

Definition 1.2.21 (Complex manifolds). A complex manifold is a smooth, 2n-

dimensional manifold M which admits an atlas UmM : Up -> @"} such that all transi-
tion maps are holomorphic.

The complex projective space is a complex manifold. Our next example naturally general-
izes the projective spaces described above.

Example 1.2.22. (The real and complex Grassmannians Gr 1<(R"l» Gr k(<Q").)

Suppose V is a real vector space of dimension n. For every 0 < lc < n we denote by
Gr k(Vl the set of Is-dimensional vector subspaces of V. We will say that Grk(V) is the
linear Grassmannian of It-planes in E. When V Rn we will write Grk (R) instead of
Gr a l l n ) .
14 Lectures on the Geometry of Manifolds

We would like to give several equivalent descriptions of the natural structure of smooth
manifold on Grk (V). To do this it is very convenient to fix an Euclidean metric on V.
Any k-dimensional subspace L C V is uniquely determined by the orthogonal projec-
tion onto L which we will denote by PL. Thus we can identify Gr MV) with the set of
rank k projectors
ProjkM {p= V - > V ; P* 1}.

Q_*
.. P2 rankP

II
II
7

Let use observe that the rank of an orthogonal projector is determined by the equality
rank P tr P.
We have a natural map
P Gr k(V ) -> Projk(vl, L I-) PL
with inverse P *-> Range (P).
The set Proj k l ) is a subset of the vector space of symmetric endomorphisms
End+(v) == {A E EndlVl, A* = A }.
The space End+ (V) is equipped with a natural inner product

( B ) := 5l tr(AB), vA, B E End-*(V). (1.2.2)

We denote by II II the norm on End+ (V) induced by this inner product,

l
11A112 = 5 ttlA2l.
Note that the subset Proj J V ) C End+ (V) can alternatively be described by the equalities
P2 = p, t r P = l .
This proves that Projm / ) is a closed subset of End+ (V). From the equality
1 k
lIPII 2 - t I P 2 = § t r P = §» 'v'P G Projm / )
2
we deduce that Projm / ) is also bounded subset of End+ (V). The bijection
P : Gr k l ) -> Projk(V), L »-> PL
induces a topology on Grk (V), and with this topology Grk (V) is a compact metric space.
We want to show that Grk,(V) has a natural structure of smooth manifold compatible
with this topology. To see this, we define for every L C Gr k(V) the set
Gr : = { U € G r MV); U M L J ' = 0 } .

Lemma 1.2.23. (a) Let L G Gr /K7(V)- Then

U F) L; = 0 ~< >» II - PL -|- PU : V -> V is an isomorphism. (1.2.3)

(b) The set Grk(V, L ) is an open subset ofGrk(V).

 Manifolds 15

Proof. (a) Note first that a dimension count implies that

UmL_L i 0 `\I \, P
U+Li i V `\I \, P
U re L 0.
Let us show that U O L_L = 0 implies that ]1 - PL -|- PL is an isomorphism. It suffices to
show that
Ker(]l - PL -|- Pu) = 0.

Suppose 'u E ker(]1 - PL -|- PU). Then

0=pL(11-pL 'l'PU)'U=PLPU'U=0--)>PU'U Q Urwke1~pL =Ural = 0 .

Hence PU?) = 0, so thats E U I . From the equality (11 - pa - Pu)V 0 we also deduce
(II - PL)V = 0 so thats E L. Hence

uQU1'-LM L 0.

Conversely, we will show that if II - PL + PU = PL_L + PU is onto, then U -|- LJ- = V.

Indeed, let 'u E V. Then there exists :IJ E V such that

FU=PLX+Pua;eLl+U.
(b) We have to show that, for every K 6 Grk(V, L), there exists 6 > 0 such that any U
satisfying

lIP - PK I < e
intersects LJ- trivially. Since K e Gr k(V, L) we deduce from (a) that the map It - PL
PK : V -> V is an isomorphism. Note that

I (11 PL PK) - (11 - PL - PUlll = lInK - PU||-

The space of isomorphisms of V is an open subset of End(Vl. Hence there exists 6 > 0
such that, for any subspace U satisfying I PU - PK I < 5, the endomorphism (it - PL - PU )
is an isomorphism. We now conclude using pa (a).

Since L 6 Gr k l , L l , v L @ Gr h e ) , the collection

Gr k(V§Ll; L Q Gr I<(V)

is an open cover of G1°k1J(V). Note that for every L E Gr k l v ) we have a natural map
F :Hom(L, Ll) -> Gr k(V, L), (1 .2.4)

that associates to each linear map S : L -> L-L its graph (see Figure 1.5)
Fs = {;u-i-S566 L + Ll V; $ 6 L}.

The map (1.2.4) is obviously injective. We claim that it is in fact a bijection.

Indeed, if U E Gr k:(V, L), then the restriction of PL to U is injective since
Unke1~pL = U n L l = 0 .
16 Lectures on the Geometry of Manifolds

LJ_

v
vL
L 1 rS
\
\
Sx "

I
I
I
I
v x L
L

Fig. 1.5 Subspaces as graphs o f linear operators.

Thus, the linear map PLlu : U -> L, U 9 Up »-> PLU is a linear isomorphism because
dim U dimL
i k. Denote by HU its inverse, HU : L -> U. It is easy to see that U is
Z

the graph of the linear map

5 : L -> L l , Sac = PLiHUX.

We will show that the bijection (1.2.4) is a homeomorphism. We first prove that it is
continuous by providing an explicit description of the orthogonal projection PpS
Observe first that the orthogonal complement of FS is the graph of -S* : L-L -> L.
More precisely,
1"SJ.=1*_S*={y-s*yeL_L+L=v; y Q L - L } .
Let ' U = P L ' U + P L J . ' U = ' U L - l - U L + E V (see Figure 1.5). Then
'U
P1*S'U=LU+SCC, :UE L \
(;z:+Sa9l€II;9§

/
\
\.

I §1xEL, Z / € L J - such that

at + S*y UL
Sac y 'UL_L

Consider the operator S : L ® LJ- -> L ® L_L which has the block decomposition

Then the above linear system can be rewritten as

Now observe that

8.
H
:u

y
'UL

'UL-L ]
S2 ]lL+S*S 0
0 HL +SS*
Hence S is invertible, and
S-12 (]lL-I-S*S) 0
S
0 l:l]_L_l_ -l-SS*)'1
 Manifolds 17

l]l L -I-S*S)'1 (HL + S*S)'1S*

u +ss*)-1s -(EL -|- SS*l-1
We deduce
QUO l]lL + S*S)-1'UL -|- (HL -|- S*Sl-1S*'ULJ_
and,
QUO

Hence PpS has the block decomposition

Pos?)
[ ] S33

IlL
Pos
S ] [lm + s*s)-1 (aL + s * 5 > - l s *

l]l L -I-S*Sl-1 (aL -|- S*S)'1S*

(l.2.5)
so + 5*s) S(J1L +S *s) -15 *
This proves that the map
Hom(L, Ll) 9 S »-> Pos E Grklv, L)
is continuous. Note that if U 6 Grk(V, L), then with respect to the decomposition V
L + L-L the projector PU has the block form
PLPuIL PLPUIL
PU
A
C
B
D
PLJ.PUIL PLLPUILJ.
where for every subspace K <-> V we denoted by IK : K -> V the canonical inclusion,
] 7

then U = Fs. If U = Fs, then (l.2.5) implies

PLPuIL A(U) = (HL +5* 5>- 1 0(U) = S(11L + s*5> -1 7
7 PLPUILJ.
so S = C A 1 Since A and C depend continuously on PU we deduce shows that the
inverse of the graph map
H o m ( L , L l 9 S »-> FS 6 Gr 1<(Vl
is also continuous. Moreover, the above formulae show that if U E Gr k(V, Low
Gr k (V, Ll), then we can represent U in two ways,
U = FSO = FS1, Si G H0>mlL1,Ll), i = 0, 1,
and the correspondence So -> $1 is smooth. This shows that Grk(V) has a natural struc-
ture of smooth manifold of dimension
dim Grklvl = dim Hom(L, Ll) k ( n - k).
Gr k(@") is defined as the space of complex k-dimensional subspaces of Q". It can be
structured as above as a smooth manifold of dimension 2k(n - k). Note that Gp1 (RTLI =
~
~
RP7n-1, and Gp1 (ml = <clpn-1. The Grassmannians have important applications in many
classification problems.
18 Lectures on the Geometry of Manifolds

Exercise 1.2.24. Show that Gr llCTL) is a complex manifold of complex dimension k(n
1<)).

Example 1.2.25. (Lie groups). A Lie group is a smooth manifold G together with a group
structure on it such that the map

G X G->G (9,I2)>9-1V1

is smooth. These structures provide an excellent way to formalize the notion of symmetry.
(a) (Rn, -I-) is a commutative Lie group.
(b) The unit circle S1 can be alternatively described as the set of complex numbers of
norm one and the complex multiplication defines a Lie group structure on it. This is a
commutative group. More generally, the torus T" is a Lie group as a direct product of n
circles.2
(c) The general linear group GL(n, K) defined as the group of invertible n X n matrices
with entries in the field K = R, (C is a Lie group. Indeed, GL(n, K) is an open subset (see
Exercise l.1.3) in the linear space of n X n matrices with entries in K. It has dimension
den2, where do is the dimension of K as a linear space over R. We will often use the
alternate notation GL(K'") when referring to GL(n, K).
(d) The orthogonal group O(n) is the group of real n X n matrices satisfying

T.Tt Jl.

To describe its smooth structure we will use the Cayley transform trick as in [113] (see also
the classical [133]). Set

m,,(1R()# {T E MAR); det(l +Tl ¢ 0 }.

o f

The matrices in m,,(1R()# are called non exceptional. Clearly II E O(n)# = O(n) VI
Mn (lR)# so that O(n)# is a nonempty open subset of O(nl. The Cayley transform is the
map # : My(Rl# -> m(02a) defined by
A »-> A# <11 A)(]1 + A )
The Cayley transform has some very nice properties.
(i) A# E m,,(nl# for every A E Mn(R)#.
(ii) # is involutive, i.e., ( A # l # = A for any A E m,,(n)#.
(iii) For every T 6 O(n)# the matrix T# is skew-symmetric, and conversely, if A E
m,,(n)# is skew-symmetric then A# E O(n).
Thus the Cayley transform is a homeomorphism from O(n)# to the space of non-
exceptional, skew-symmetric, matrices. The latter space is an open subset in the linear
space of real n X n skew-symmetric matrices, Q(n).
Any T 6 O(n) defines a self-homeomorphism of O(n) by left translation in the group
LT : O(n) -> O(n) S »-> LT(S) = T S. -
20ne can show that any connected commutative Lie group has the from Tn X R "
 Manifolds 19

We obtain an open cover of O(n):

O(n) = U T - O(n)#.
T€O(n)

Define QT : T - O ( n ) # -) Q(n) by S »-> (T-1 - S)#. One can show that the collection

IT . o<n>#,@T ) T€0(n)

defines a smooth structure on O(n). In particular, we deduce

dim O(n) n(n - 1)/2.

Inside O(n) lies a normal subgroup (the special orthogonal group)

SO(n) = {T 6 O(n) det T = 1}.

The group SO(n) is a Lie group as well and dim SO(n) = dim O(n).
(e) The unitary group U (in) is defined as

U(n) = {T 6 GL (n,(Cl; T - T * = ] l } ,

where T* denotes the conjugate transpose (adjoint) of T. To prove that U (n) is a manifold
one uses again the Cayley transform trick. This time, we coordinatize the group using the
space u(n) of skew-adjoint (skew-Hermitian) n X n complex matrices (A = -A*). Thus
U (n) is a smooth manifold of dimension

dim U(n) = dim_u(n) n2

Inside U(n) sits the normal subgroup SU(n), the kernel of the group homomorphism det :
U(n) -> S1. SU(nl is also called the special unitary group. This a smooth manifold of
dimension 112 - 1. In fact the Cayley transform trick allows one to coordinatize SU(n)
using the space

W) {A E _u(nl , teA 0}.

Exercise 1.2.26. (a) Prove the properties (i)-(iii) of the Cayley transform, and then show
-
that (T 0(n)#, \ P T ) T € 0 ( n ) defines a smooth structure on O(nl.
(b) Prove that U(n) and SU(n) are manifolds.
(c) Show that O(n), SO(n), U(n), SU(n) are compact spaces.
(d) Prove that SU(2) is diffeomorphic with S3. (Hint: think of S3 as the group of unit
quaternions.)

Exercise 1.2.27. Let SL(n; K) denote the group of n X n matrices of determinant 1 with
entries in the field K = K, C Use the implicit function theorem to show that SL(n; K) is
a smooth manifold of dimension dK(77,2 - 1), where did = dim K.
20 Lectures on the Geometry of Manifolds

Exercise 1.2.28. (Quillen). Suppose Vo, V1 are two real, finite dimensional Euclidean
space, and T : V0 -> V1 is a linear map. We denote by T* is adjoint, T* : V1 -> Vo,
and by FT the graph of T,
FT = { ( U 0 7 U1l E V0 ® V13 'U1 I T'U0)}.

We form the skew-symmetric operator

0 T*
[ ] [ N ]
00 U0
XzV0QEV1->V0@BV1, X
U1 T 0 U1

We denote by CT the Cayley transform of X ,

C'T=(]1-Xl(]l-i-X) 7

and by R0:V0€BV1->V0€BV1 the reflection

Ro

Show that RT = CTRL is an orthogonal involution, i.e.,

R§~=1, R T = R T ,
and ker(]1 - RT) FT. In other words, RT is the orthogonal reflection in the subspace
FT,
RT = 2PPT 11,
where PPT denotes the orthogonal projection onto FT.

Exercise 1.2.29. Suppose G is a Lie group, and H is an abstract subgroup of G. Prove that
the closure of H is also a subgroup of G.

Exercise 1.2.30. (a) Let G be a connected Lie group and denote by U a neighborhood of
1 6 G. If H is the subgroup algebraically generated by U show that H is dense in G.
(b) Let G be a compact Lie group and g 6 G. Show that l 6 G lies in the closure of
{9" ; n G Z \ {0}}-
Remark 1.2.31. If G is a Lie group, and H is a closed subgroup of G, then H is in fact a
smooth submanifold of G, and with respect to this smooth structure H is a Lie group. For a
proof we refer to [61, 128]. In view of Exercise 1.2.29, this fact allows us to produce many
examples of Lie groups.

1.2.4 How many manifolds are there?

The list of examples in the previous subsection can go on for ever, so one may ask whether
there is any coherent way to organize the collection of all possible manifolds. This is too
general a question to expect a clear cut answer. We have to be more specific. For example,
we can ask
 Manifolds 21

Question 1: Which are the compact, connected manifolds of a given dimension d?

For d = 1 the answer is very simple: the only compact connected 1-dimensional mani-
fold is the circle S1. (Can you prove this?)
We can raise the stakes and try the same problem for d = 2. Already the situation
is more elaborate. We know at least two surfaces: the sphere S2 and the torus T2. They
clearly look different but we have not yet proved rigorously that they are indeed not diffeo-
morphic. This is not the end of the story. We can connect sum two tori, three tori or any
number g of tori. We obtain doughnut-shaped surface as in Figure 1.6

Fig. 1.6 Connected sum o f 3 tori.

Again we face the same question: do we get non-diffeomorphic surfaces for different
choices of g? Figure 1.6 suggests that this may be the case but this is no rigorous argument.
We know another example of compact surface, the projective plane RIP2, and we nat-
urally ask whether it looks like one of the surfaces constructed above. Unfortunately, we
cannot visualize the real projective plane (one can prove rigorously it does not have enough
room to exist inside our 3-dimensional Universe). We have to decide this question using a
little more than the raw geometric intuition provided by a picture. To kill the suspense, we
mention that RIP2 does not belong to the family of donuts. One reason is that, for example,
a torus has two faces: an inside face and an outside face (think of a car rubber tube). RP2
has a weird behavior: it has "no inside" and "no outside". It has only one side! One says
the torus is orientable while the projective plane is not.
We can now connect sum any numbers of lRlPD2's to any donut an thus obtain more and
more surfaces, which we cannot visualize and we have yet no idea if they are pairwise
distinct. A classical result in topology says that all compact surfaces can be obtained in
this way (see [90]), but in the above list some manifolds are diffeomorphic, and we have
to describe which. In dimension 3 things are not yet settled3 and, to make things look
hopeless, in dimension > _ 4 Question 1 is algorithmically undecidable .
We can reconsider our goals, and look for all the manifolds with a given property X . In
many instances one can give fairly accurate answers. Property X may refer to more than
the (differential) topology of a manifold. Real life situations suggest the study of manifolds
with additional structure. The following problem may give the reader a taste of the types
of problems we will be concerned with in this book.
3 Things are still not settled in 2007, but there has been considerable progress due to G. Perelman's proof of the
Poincaré's and Thursron's conjectures.
22 Lectures on the Geometry of Manifolds

Question 2: Can we wrap a planar piece of canvas around a metal sphere in a one-to-one
fashion? (The canvas is flexible but not elastic).
A simple do-it-yourself experiment is enough to convince anyone that this is not pos-
sible. Naturally, one asks for a rigorous explanation of what goes wrong. The best ex-
planation of this phenomenon is contained in the celebrated Theorem Egregium (Golden
Theorem) of Gauss. Canvas surfaces have additional structure (they are made of a special
material), and for such objects there is a rigorous way to measure "how curved" are they.
One then realizes that the problem in Question 2 is impossible, since a (canvas) sphere is
curved in a different way than a plane canvas.
There are many other structures Nature forced us into studying them, but they may not
be so easily described in elementary terms.

A word to the reader. The next two chapters are probably the most arid in geometry but,
keep in mind that, behind each construction lies a natural motivation and, even if we do
not always have the time to show it to the reader, it is there, and it may take a while to
reveal itself. Most of the constructions the reader will have to "endure" in the next two
chapters constitute not just some difficult to "swallow" formalism, but the basic language
of geometry. It might comfort the reader during this less than glamorous journey to carry
in the back of his mind Hermann Weyl's elegantly phrased advise.
"It is certainly regrettable that we have to enter into the purely formal aspect in such detail
and to give it so much space but, nevertheless, it cannot be avoided. Just as anyone who
wishes to give expressions to his thoughts with ease must spend laborious hours learning
language and writing, so here too the only way we can lessen the burden of formulae is to
master the technique of tensor analysis to such a degree that we can turn to real problems
that concern us without feeling any encumbrance, our object being to get an insight into
the nature of space [...]. Whoever sets out in quest of these goals must possess a perfect
mathematical equipment from the outset."
H. Weyl: Space, Time, Mattel:
 Chapter 2

Natural Constructions on Manifolds

The goal of this chapter is to introduce the basic terminology used in differential geometry.
The key concept is that of tangent space at a point which is a first order approximation of
the manifold near that point. We will be able to transport many notions in linear analysis
to manifolds via the tangent space.

2.1 The tangent bundle

2.1.1 Tangent spaces

We begin with a simple example which will serve as a motivation for the abstract defini-
dons
Example 2.1.1. Consider the sphere 2
S2) 2 + Z 2 l in RB.
( I +
We want to find the plane passing through the North pole N (0, 0, 1) that is "closest" to the
sphere. The classics would refer to such a plane as an oscillator plane.
The natural candidate for this oscillator plane would be a plane given by a linear equa-
tion that best approximates the defining equation 3132 +3/2 -l- Z2 = 1 in a neighborhood of the
North pole. The linear approximation of 232 -1- y2 -l- Z2 near N seems like the best candidate.
We have
1v2+y2+22 l 2(z 1) + O(2),
where 0(2) denotes a quadratic error. Hence, the oscillator plane is Z = 1. Geometrically,
it is the horizontal amine plane through the North pole. The linear subspace {z = 0} C $3
is called the tangent space to S2 at N.
The above construction has one deficiency: it is not intrinsic, i.e., it relies on objects
"outside" the manifold S2. There is one natural way to fix this problem. Look at a smooth
path 'y(t) on S2 passing through N at t = 0. Hence, t »-> 'y(t) 6 R3, and
Iv(t)l 2 = 1. (2.1.1)
If we differentiate (2.1.1) at t = 0 we get 07(0), 'y(0)) = 0, i.e., (0) J_ 'y(0), so that (0)
lies in the linear subspace Z = 0. We deduce that the tangent space consists of the tangents
to the curves on S2 passing through N.

23
24 Lectures on the Geometry of Manifolds

This is apparently no major conceptual gain since we still regard the tangent space
as a subspace of R3, and this is still an extrinsic description. However, if we use the
stereographic projection from the South pole we get local coordinates (u, Fu) near N, and
any curve 'y(t) as above can be viewed as a curve t »-> (u(t), v(t)) in the (u, UI plane. If
(WI is another curve through N given in local coordinates by t »-> (_
'u,(t),y(t)), then

(U°(0l»'ul(0) )
`\I \, P
(0l (/5(0)
The right-hand side of the above equality defines an equivalence relation on the set of
smooth curves passing trough (0, 0). Thus, there is a bijective correspondence between the
tangents to the curves through N, and the equivalence classes of "~". This equivalence
relation is now intrinsic modulo one problem: "~" may depend on the choice of the local
coordinates. Fortunately, as we are going to see, this is a non-issue.

Definition 2.1.2. Let M771 be a smooth manifold and Po a point in M. Two smooth paths
a, B : (-e, 5 -> M such that a(0) = B(0) = Po are said to have a ]?rst order contact at
Po if there exist local coordinates (x) = (xl, . . . ,:z:m*) near Po such that

w'o(0) i £i35(0l,
where

a(t) = ( m ) (al (t), • • • ,x;"(t) ) 7

and

it) = ( t ) = (we), . . »w2"(t))-•

We write this O d ~ 1

Lemma 2.1.3. ~1 is an equivalence relation.

Sketch of proof. The binary relation -1 is obviously reflexive and symmetric, so we only
have to check the transitivity. Let a -1 B and B -1 fy. Thus there exist local coordinates
(w)=(x% ,xm) and (y) = Ly1, . 7 ym) near Po such that
(mum) = (am) and (y°@(0) ) IzM0II-
The transitivity follows from the equality
<93/i 33/i
9(0) Z yaw) 3307
in) 3307
°j
11» (0) i y°j (0)-
j J

Definition 2.1.4. A tangent vector to M at p is a first-order-contact equivalence class of

curves through p. The equivalence class of a curve a(t) such that a(0) = p will be tem-
porarily denoted by [a(0)]. The set of these equivalence classes is denoted by TpM, and is
called the tangent space to M at p.

Lemma 2.1.5. TpM has a natural structure of vector space.

 Natural Constructions on Manifolds 25

Proof. Choose local coordinates (x1, . . . , a s ) near p such that xi (p) = 0, W, and let a
and 5 be two smooth curves through p. In the above local coordinates the curves a, B
become (w;(t)) (w;<t))~ Construct a new curve 'Y through p whose as coordinates are
given by

( ) = (wav) +w;(t))-
We want to emphasize that the above curve depends on a, 3 and the choice of local coor-
dinates (x1, . . . , a s ) . We will indicate this using the notation a +x B when referring to by.
Set

[Ci(0)] -|- [B(0)1 ii Mon.

The zero vector in TpM is described by a curve a such that go; (0) 0, W. For this
operation to be well defined one has to check several things.

(a) The equivalence class ['y°(0)] is independent of coordinates. In other words, if

x 1 , . . . , x M and (y1, . . . , am) are local coordinates near p such that

t'(p)==y"(p)= 0, am,
then

(b) If a curve represents the zero vector in some local coordinates (x'i) at p, then it repre-
sents the zero vector in any other choice of local coordinates at p.
(c) If [o21(0)] = [o22(0)] and [B1(0)] = [B2(0)], then
[021(0)] + [B1(0)] [022(0)] + [82(0)]-
We let the reader supply the routine details.

Exercise 2.1.6. Finish the proof of the Emma 2. l .5.

From this point on we will omit the brackets [ - ] in the notation of a tangent vector.
Thus, [ci(0)1 will be written simply as 0;(0).
As one expects, all the above notions admit a nice description using local coordinates.
Let (1131, . . . ,as"") be coordinates near p E M such that xi (al = 0, Vi. Consider the curves
1 7 . . . 7 MY
where 6; denotes Kronecker's delta symbol. For example
€1ltl=lt,0,0,...,0< €2ltl=l0,t,0,...,0l,
We set
@xk(p) Z= éklol I . .7Mk ) . (2.1.2)

Often, when the point p is clear from the context, we will omit it in the above notation.
" N o t e that the vectors a (p) depend on the choice of local coordinates (5131 7 7 :um .

Lemma 2.1.7. The collection (3wk ( p ) llgkgtm is a basis ofT pM.

26 Lectures on the Geometry of Mamfolds

Proof. Given a smooth path

a(t) Z ( l (t), 7 xi*(t))» 0z(Ul=p,
we have
do) = (f1 (0), 7 ;'(0)) = f1 (0)(1,0, ,0)+ +12gl(0)(0,...,0,1)
TTI
(2.l.3)
531 (0)31 + -|- 52T(0)<9'm i:;(0l<9Q3
i 1

Exercise 2.1.8. Suppose that M C R" is an m-dimensional smooth submanifold of R" .

Let Po E M and suppose that a, 5 : (-et, 5) -> M are two smooth paths such that a(0) =
5(0) P0- Prove that the following statements are equivalent.
(1) The paths a have first order contact at Po .
(2)
day) = d,B(O)
as vectors in Rn .
dt dt

Deduce from the above that Thor can be identified with the vector subspace of R"
spanned by the vectors U e R" with the property that there exists a smooth path a :
(6, 5) -> M C R" such that
da(0)
04(0) Po, 17.
dt
Exercise 2.1.9. Let F : RN -> Rkbe a smooth map. Assume that
(a) M = F-1(0) # V);
(b) rank D11:F = k, for all zz: 6 M.
Then M is a smooth manifold of dimension N - k and
Lm r her Dx3 F, Vcc E M.
Example 2.1.10. We want to describe T11 G, where G is one of the Lie groups discussed in
Section 1.2.2.
(a) G = O(n). Let (-6, 6) 9 .s »-> T(.s) be a smooth path of orthogonal matrices such that
T(0l = ]1. Then Tt(s) T(sl = JL Differentiating this equality at .s = 0 we get
Tt(0) -|- T(0) = 0.
The matrix T ( 0 ) defines a vector in T1O(nl, so the above equality states that this tangent
space lies inside the space of skew-symmetric matrices, i.e., To O(n) C Q(n). On the other
hand, we proved in Section 1.2.2 that dim G = dim Q(n) so that
T]1O(nl = Q(n).
(b) G = SL(n;]R). Let ( - e , 5 9 s »-> T(s) be a smooth path in SL (n;]R.) such that
T(0l = 1. Then det T ( s ) = 1 and differentiating this equality at S = 0 we get (see
Exercise 1.1.3)
tr T(0) = 0.
Thus, the tangent space at ]1 lies inside the space of traceless matrices, i.e., To SL(n; R) C
QW; R). Since (according to Exercise 1.2.27) dim SL(n; R) = d i m ( n ; R) we deduce
T]1SL(n;R) = M; R).
Exercise 2.1.11. Show that T]1U(n) = g(n) and To SU(n) =
 Natural Constructions on Manycolds 27

2.1.2 The tangent bundle

In the previous subsection we have naturally associated to an arbitrary point p on a manifold
M a vector space TpM . It is the goal of the present subsection to coherently organize the
family of tangent spaces lTpM)p€M° In particular, we want to give a rigorous meaning to
the intuitive fact that TpM depends smoothly upon p.
We will organize the disjoint union of all tangent spaces as a smooth manifold TM .
There is a natural surjection
'ii:TM= U ToM->M,
ppM
7r(v)=p< >v€TpM.

Any local coordinate system .cc = ( x ) defined over an open set U C M produces a natural
basis l88,(p)l of TpM, for any p 6 U. Thus, an element U 6 TU LIpaU TpM is T

completely determined if we know

.. which tangent space does it belong to, i.e., we know the point p
the coordinates of 'U in the basis (6,,,, (p) ),
7r(0) E M ,

§jx(v)6(p).
D
II

We thus have a bijection

m,,,.TU->U°"><1R(TT C R " X R " 7 'U * >(:v(7r(v)), X1(v), . . . 7 XM('v))7
where U" is the image of U in RM via the coordinates ( i ) . We can now use the map mm
to transfer the topology on Run X RM to TU. Again, we have to make sure this topology is
independent of local coordinates.
To see this, pick a different coordinate system y = ( y ) on U. The coordinate indepen-
dence referred to above is equivalent to the statement that the transition map

\Py O m U"'x]R€.'" >TU >UYx]R'"'

is a homeomorphism.
Let A := (T,X) 6 U" X Run. Then .3 1 (A) up, <5v(0))» where Hp) iv, and
a(t) C U is a path through p given in the :zz coordinates as
a(t) :1z-I-tX.

Denote by F U"->UY the transition map :zz »-> y. Then

\I'y°\1'm1(A)=(y(T);Y1, »Y'") 7

where <5((0) (QP (0)) =2yw,,(p), and ( y ( t ) ) i s the description of the path a(t) in
Z

the coordinates by . Applying the chain rule we deduce

33,15 33/5 .
YJ y°j (0)
t
(0) Xz (2.l.4)
33% 350'L

This proves that \1/y O Q / 1 is actually smooth.

28 Lectures on the Geometry of Manifolds

The natural topology of TM is obtained by patching together the topologies of T U ,

where (U ( f ) W is a countable atlas of M. A set D C TM is open if its intersection with
any TU is open in T U . The above argument shows that TM is a smooth manifold with
(TU", \I/'Y) a defining atlas. Moreover, the natural projection 'II' : TM -> M is a smooth
map.

Definition 2.1.12. The smooth manifold TM described above is called the tangent bundle
ofM.

Proposition 2.1.13. A smooth map f : M -> N induces a smooth map D f : T M -> T N

such that
(al D f l T p m l C Tf(p)N) V p 6 M
(b) The restriction to each tangent space DpF : TpM -> Tf(p)N i5 linear: The map D f is
called the differential of f , and one often uses the alternate notation f* = Df.
Proof. Recall that TpM is the space of tangent vectors to curves through p. If a ( t ) is such
a curve (a(0) = p), then B(t) = f(a(t)) is a smooth curve in N through q = f(p), and
we define

Df(d(0)) Q; Be).
One checks easily that if a l ~1 Ag, then f(a1) ~1 f(a2), so that Df is well defined.
To prove that the map Df : TpM -> TqN is linear it suffices to verify this in any partic-
ular local coordinates ( f f l , . . . ,;I:M) near p, and (g1, . . . ,y"') near q, such that :1ci(p) = 0,
by (q) = 0, W, j, since any two choices differ (infinitesimally) by a linear substitution.
Hence, we can regard f as a collection of maps
(511»»~»»f1>M) »-> (y1(:1:1, ,a:'"),...,y"(:c1, . . . ,a:'")).

A basis in TpM is given by { 3 . while a basis of TqN is given by { by,

If a, 5 : (-6, 5) -> M are two smooth paths such that a(0) = 5(0) = p, then in local
coordinates they have the description

a(t) 1 (5/31 (al, »H1Q"(t))» B(t)=(H1E(t)» 7 HvZ"(t))»

do) Z (131 (go, 7 ia""(0))» 5(0)=(i:}2(0)» 7 =i:Z'(0) l.

Then
F(@(¢)) = (y1(t), ,y;:(t)) I (1/1(=@;()), 7 y"(:v2,(t) )),

F(@()) = (1/é»(t)» »yE(t)) = (3/1(=@;()), , y " ( ; ( t ) ) ),

Flag) -|- aw) Z (y1(w;lt) -|- H1200 ), ,y"(£uZ,(t)+w2(t)))-

DF( co) )
d
llllllllll

dt F(@(t))
( 2- 1- 3)
'rL

j=1
..
3/3(0)3yi
iii
j= 1 i=1
691
33% I
=i;;(0) E9yj 7
 Natural Constructions on Manifolds 29

DF(B(0) )
d
llllllllll

dt F(@(t))<"'3>
j 1
y°g(0)@,, iii
.

§=1
1:1
8?/1
3581
(0) 3y3 7

DF(@(0) -|- /3(0) ) d

dt F( @ (t) + 5(0) iii
j=1
<93/i

i=1 3113'
(5;;(0)+:i:g(0)) E9yj

DF(@(0)) + DF(/3'(0)).
This shows that Df : TpM -> TqN is the linear operator given in these bases by the matrix
/ . \ . . . .
I..

A J l . In paltlcular, this
•
implies that Df is also smooth.
1SJSH, 1§i§m

2.1.3 Transversality
In this subsection we want to explain rigorously a phenomenon with which the reader may
already be intuitively acquainted. We describe it first in a special case.
Suppose M is a submanifold of dimension 2 in R3. Then, a simple thought experiment
suggests that most horizontal planes will not be tangent to M. Equivalently, if we denote
by f the restriction to M of the function (as, y, z) +-> z, then for most real numbers h the
level set f-1(h) = M n {z = h} does not contain a point where the differential of f is
zero, so that most level sets f-1 (in) are smooth submanifolds of M of codimension l, i.e.,
smooth curves on M.
We can ask a more general question. Given two smooth manifolds X, Y, a smooth map
f : X -> Y, is it true that for "most" y E Y the level set f - 1 (y) is a smooth submanifold of
X of codimension dim Y? This question has a positive answer, known as Sard's theorem.

Definition 2.1.14. Suppose that Y is a smooth, connected manifold of dimension m.

(a) We say that a subset S C Y is negligible if, for any coordinate chart of Y, \I/ : U ->
Run, the set \I1(S VW U) C RM has Lebesgue measure zero in RM .
(b) Suppose F : X -> Y is a smooth map, where X is a smooth manifold. A point
£13 6 X is called a critical point of F, if the differential DasF : TasX -> Tp(:,3)Y is not
surjective.

We denote by CTF the set of critical points of F, and by As C Y its image via F. We will
refer to As as the discriminant set of F. The points in As are called the critical values
of F.

Exercise 2.1.15. Define

Z (at, as b,c)€]R4; a:132-i-bzc-I-c=0; a # 0 }.

II

(a) Prove that Z is a smooth submanifold of R4.

(b) Define 'ii : Z -> R3 by We a, b, c) I "> (a, b, c). Describe explicitly the discriminant set
of Tr.
30 Lectures on the Geometry of Manifolds

Definition 2.1.16. Suppose U, V are finite dimensional real Euclidean vector spaces, (9 C
U is an open subset, and F : (9 -> V is a smooth map. Then a point u 6 (9 is a critical
point of F if and only if
rank (D'UF : U -> V) < m i n ( dim U, d i v ) .
Exercise 2.1.17. Show that if F : M -> N is a smooth map and dim N < _ dim M, hen for
every q 6 N \ As the fiber F-1(q) is either empty, or a submanifold of M of codimension
dim N.

Theorem 2.1.18 (Sard). Suppose U, V are ]?nite dimensional real Euclidean vector
spaces, (9 C U is an open subset, and F : C) -> V is a smooth map. Assume
dim V < dim U . Then the discriminant set As is negligible.

Proof. We follow the elegant approach of J. Milnor [97] and L. Pontryagin [111]. Set
n = dim U, and m = dim V. We will argue inductively on the dimension n.
For every positive integer k we denote by Cfrkp C CTF the set of points 'LL G (9 such
that all the partial derivatives of F up to order A: vanish at u. We obtain a decreasing
filtration of closed sets
CTF nor; nor; D .
The case n = 0 is trivially true so we may assume n > 0, and the statement is true for any
n' < n, and any m < n'. The inductive step is divided into three intermediary steps.
Step 1. The set F ( C T p \ C r ) is negligible.
Step 2. The set F(CT'F \ CT"'F+1) is negligible for all k > l.
Step 3. The set F(CT'F) is negligible for some s u f i ciently large ac.
Step 1. Set CTF := CTF u p . We will show that there exists a countable open cover
{(97}j>1 of CTF such that F ( 0 j VI CTFI is negligible for all j > 1. Since CT'F is
separable, it suffices to prove that every point u 6 CTF admits an open neighborhood N
such that F ( N VI CTF ) is negligible.
Suppose up 6 C T F . Assume first that there exist local coordinates (5/31, . . .,;1:") de-
fined in a neighborhood N of up, and local coordinates Ly1 , . . . , ym*) near 100 = F(u0) such
that,
@i('U,0) =0, W 1, Na W = 1 MY
3/(v0) = 0 , 7

and the restriction of F to N is described by functions ye yi(:1:1,...,a:M) such that

al x1.
For every t E R we set
3\ft:= {(:1:1,...,:1:")€3\f, : c 1 = t } 7

and we define
m)
Gt I Nt -> R m - 1 7 ( . . . a 2 2
(3/ (t : c , . . . , §UTL)a°°°7YMlt7X27°°°aIn) I.
Observe that
l->
F(NN C T F I = U { t } X Gt(C'rGt ).
t
 Natural Constructions on Manifolds 31

The inductive assumption implies that the sets Gt (Cfrgt) have trivial (m - l)-dimensional
Lebesgue measure. Using Cavalieri's principle or Fubini's theorem we deduce that F(N O
CTF) has trivial m-dimensional Lebesgue measure.
To conclude Step l is suffices to prove that the above simplifying assumption con-
cerning the existence of nice coordinates is always fulfilled. To see this, choose local
coordinates (so, . . . , s") near U0 and coordinates l y , . . . ,yM) near U0 such that
s£('u,0)=0, 'v'i=1,...nI y7(v0)=0, 'v'j=1,...m,

The map F is then locally described by a collection of functions yj(s17 7 sn), j

l , . . . , n . Since u 6 C T F/ , we can assume, after an eventual re-labelling of coordinates,

that 31/1381
Now define
a: l y 1 (s 1 , . . . , s n ), a:¢ -_ s~ ' , ' v-' z_ - 2 , . . . , n .

The implicit function theorem shows that the collection of functions (1131 ,...,.cc") defines
a coordinate system in a neighborhood of U0. We regard by as functions of so. From the
definition we deduce y1 = acl .
Step 2. Set
k
( )
CTF
• k
:= CHI; Crl+1. Since U0 6 C r i ) , we can find local coordinates
.
(s1, . . .,s") near U0 and coordinates (y1, . . . , ym) near "U0 such that
smog Z 0, Vi I l,...n, ye I v ) = 0, 'v'j Z l,...m,

31y1
(Uu) 0, v j = 1 , . . . , k ,
6(81)j
and
3/c-I-lyl

3(81)k+1 (Uo) ¢ 0.
Define
alkyl
2(8) 6(51)k 7

and set 1:i := s', V11 2, . . . , n.

Then the collection ( x ) defines smooth local coordinates on an open neighborhood N
o f U Q , and Cfrkp HN is contained in the hyperplane {a:1 = 0}. Define

G : NVW{:I:1 = 0 } -> V ; G(:z:2,...,:1:M) = F(0,:1:2,.. 7 so" .

Then
co;my = Org, F(C1°kF mol = G(c¢~'g),

and the induction assumption implies that G(C'r'é) is negligible. By covering CTF( k ) with
a countably many open neighborhood {Ne}e>1 such that F(C'r'§" Vulg) is negligible we
conclude that F(C'rF( k ) ) is negligible.
Step 3. Suppose k > Q . We will prove that F(C'rk'F) is negligible. More precisely, we
will show that, for every compact subset S C (9, the set F ( S o C7r°k'F) is negligible.
32 Lectures on the Geometry of Manifolds

From the Taylor expansion around points in C'rak'F OS we deduce that there exist num-
bers To 6 (0, l ) and A0 > 0, depending only on S, such that, if C is a cube with edge
1° < to which intersects CW; OS, then

diamF(C) < A0rk 7

where for every set A C V we define

diam(A) := sup{ la1 G12l5 611102 pA}.
In particular, if um denotes the m-dimensional Lebesgue measure on V, and ,un denotes
the n-dimensional Lebesgue measure on U, we deduce that there exists a constant A1 > 0
such that

urn(F(cll < 17mk I )'1l*nlC)Mk/'N'

Cover C'r_kF OS by finitely many cubes {Ce}1gegL, of edges < kg, such that their interiors
are disjoint. For every positive integer P, subdivide each of the cubes Ce into P" sub-cubes
C of equal sizes. For every sub-cube $6 which intersects CH; we have
A
,UmlFlCll S I\1,Un(C?)Mk/N
We deduce that

mm( F(@ re 0 1 ) < um( Fw; re Grim) S p"-mMv).

O'

n
If we let P -> oo in the above inequality, we deduce that when k > m
we have

ralF(Ce CT) ) 0, W = 1 , . . . , L .

Theorem 2.1.18 admits the following immediate generalization.

Corollary 2.1.19 (Sard). Suppose F : X -> Y $5 a smooth mop between two smooth
connected, separable, manifolds. Then its discriminant set is negligible.

Definition 2.1.20. A smooth map f : M -> N is called immersion (resp. submersion) if

for every p 6 M the differential D i f : TpM -> Tf(p)N is injective (resp. surjective). A
smooth map f : M -> N is called an embedding if it is an injective immersion.

Suppose F : M -> N is a smooth map, and dim M > dim N. Then F is a submersion
if and only if the discriminant set As is empty.

Exercise 2.1.21. Suppose F : M -> N is a smooth map, and S C N is a smooth subman-

ifold of N. We say that F is transversal to S if for every .in 6 F-1(S) we have
-
`
TF(£,3)S -|- DmFlTxM).
TF(:n)N

Prove that if F is transversal to S, then F' 1 (S) is a submanifold of M whose codimension

is equal to the codimension of S in N.
 Natural Constructions on Manifolds 33

Exercise 2.1.22. Suppose A, x, Y are smooth, connected manifolds, and F : A X X->Y

is a smooth map
A x X 9 (A,:1:) »-> FIZUI 6 Y.
Suppose S is a submanifold of Y such that F is transversal to S. Define
ZS Z = F-1lgl C A X X.
Prove that if A0 6 A is a regular value of the natural map
ZsC A xX->A,
then the map F o : X -> Y is transversal to S.

2.1.4 Vector bundles

The tangent bundle TM of a manifold M has some special features which makes it a
very particular type of manifold. We list now the special ingredients which enter into this
special structure of TM since they will occur in many instances. Set for brevity E := TM,
and F := RM (m = dim M). We denote by Aut(F) the Lie group GL(n, R) of linear
automorphisms of F. Then

(a) E is a smooth manifold, and there exists a surjective submersion 7r E -> M. For
every U C M we set E IN:= 7r"1(U).
(b) From (2.l.4) we deduce that there exists a trivializing cover, i.e., an open cover U of
M , and for every U G 'LL a diffeomorphism
\ItuZElu->UXF, U »-> (p = v), @g(@)).
(bl) (Dp is a diffeomorphism Ep -> F for any p € U.
(b2) If U, V G 'LL are two trivializing neighborhoods with non empty overlap U n V
then, for any p Q U Fl V, the map <I>VU(p) <1>pV o (<I>8) : F -> F is a linear
isomorphism, and moreover, the map
p I-> @VU(P) G Aut(F)
is smooth.

In our special case, the map QvUIP) is explicitly defined by the matrix (2. 1.4)
I

App) i (p)
1<'£,j<m

In the above formula, the functions l ) are local coordinates on U, and the functions (ye l
are local coordinates on V.
The properties (a) and (b) make no mention of the special relationship between E =
TM and M. There are many other quadruples (E, 7r, M, F) with these properties and they
deserve a special name.

Definition 2.1.23. A vector bundle is a quadruple (E,1r,M,F) such that

34 Lectures on the Geometry of Manifolds

.. E, M are smooth manifolds,

.. 7T : E -> M is a surjective submersion,

F is a vector space over the field K = R, (I, and
the conditions (a) and (b) above are satisfied.

The manifold E is called the total space, and M is called the base space. The vector
space F is called the standard]iber, and its dimension (over the field of scalars K) is called
the rank of the bundle. A line bundle is a vector bundle of rank one.

Roughly speaking, a vector bundle is a smooth family of vector spaces. Note that the
properties (bl) and (b2) imply that the fibers 7r"1(p) of a vector bundle have a natural
structure of linear space. In particular, one can add elements in the same fiber. Moreover,
the addition and scalar multiplication operations on 7r"1(p) depend smoothly on p. The
smoothness of the addition operation this means that the addition is a smooth map

+ : E X m E = { ( U , V ) e E X E ; 1r('u,)=1r(v)}->E.

The smoothness of the scalar multiplication means that it is smooth map

RxE->E.
There is an equivalent way of defining vector bundles. To describe it, let us introduce a
notation. For any vector space F over the field K = R, (C we denote by GLK(F), (or
simply GL(F) if there is no ambiguity concerning the field of scalars K) the Lie group of
linear automorphisms F -> F.
~
According to Definition 2.1.23, we can find an open cover (Ua) of M such that each of
the restrictions Ea = E INa is isomorphic to a product \110, : Ea = F X UQ- Moreover, on
the overlaps UQ VW UP, the transition maps .905 = \1/o\1/0 1 can be viewed as smooth maps
.QoB : Uo, VI UB -> GL(F).

They satisfy the cocycle condition

(a) gas I HF
(b) QQBQBWQwQ up over Ua O UB H U p
Conversely, given an open cover (Ua) of M, and a collection of smooth maps
£106 : Uo, O UB -> GL(F)
satisfying the cocycle condition, we can reconstruct a vector bundle by gluing the product
bundles Ea = F X UCI on the overlaps Uo, n UB according to the gluing rules
the point (v,.5c) 6 Ea is identified with the point (9@(:@)w) 6 EB Va: 6 Uo, re UB-

The details are can*ied out in the exercise below.

We will say that the map 96a is the transition from the a-trivialization to the B-
trivialization, and we will refer to the collection of maps (950) satisfying the cocycle
 Natural Constructions on Manifolds 35

condition as a gluing cocycle. We will refer to the cover (Ua) as above as a trivializing
cover.

Exercise 2.1.24. Consider a smooth manifold M, a vector space V, an open cover (UQ),
and smooth maps
Q0B : Uo, re UB -> GLIV)
satisfying the cocycle condition. Set
X:= UV x U O x { a } .
Of

We apologize X as the disjoint union of the topological spaces U a X V , and we define a

relation ~ C X X X by
def a
V X U a x { a } 9 ('u,,:1:,a) i v (v, ' a B) Q V X UB X {/5'} -: zu, 'U =gg0,l£l3)'LL.

(a) Show that is an equivalence relation, and E = X/ equipped with the quotient
I'\J

topology has a natural stnicture of smooth manifold.

(b) Show that the projection 77' : X -> M , (7u,,a:,a) »-> as descends to a submersion
E -> M.
(c) Prove that (E, 7r7 M, V) is naturally a smooth vector bundle.

Definition 2.1.25. A description of a vector bundle in terms of a trivializing cover, and a

gluing cocycle is called a gluing cocycle description of that vector bundle.

Exercise 2.1.26. Find a gluing cocycle description of the tangent bundle of the 2-sphere.

In the sequel, we will prefer to think of vector bundles in terms of gluing cocycles.
Definition 2.1.27. (a) A section in a vector bundle E l> M defined over the open subset
U C M is a smooth map S : U -> E such that
s(p) E Ep '/r"1(p), Vp E U 4--> f r o s = ]1U.
The space of smooth sections of E over U will be denoted by I`(U, E) or COO(U, E) . Note
that F(U, E) is naturally a vector space. We will use the simpler notation C°°(E) when
referring to the space of sections of E over M.
(b) A section of the tangent bundle of a smooth manifold is called a vector field on that
manifold. The space of vector fields over on open subset U of a smooth manifold is denoted
by Vect (U).

Proposition 2.1.28. Suppose E -> M is a smooth vector bundle with standard fiber F,
defined by an open cover lUa)a6AJ and gluing cocycle
960 : Uog -> GL(F).
Then there exists a natural bueetion between the vector space of smooth sections of E, and
the set of families of smooth maps { sa : Uo, -> F; o f 6 A }, satisfying the following
gluing condition on the overlaps
sO(:1:) = go5(:1:)s5(:1:), Va: 6 Uo, UB-
36 Lectures on the Geometry of Manifolds

Exercise 2.1.29. Prove the above proposition.

Definition 2.1.30. (a) Let Ei 1 Mt be two smooth vector bundles. A vector bundle map
consists of a pair of smooth maps f : M1 -> M2 and F : E1 -> E2 satisfying the
following properties.

. The map F covers f , i.e., F(E,} )

diagram below is commutative
Equivalently, this means that the

E1 F ET
>

W1 '/T2

f
M1 M2
.
)

The induced map F .• E p1 - > E f l2p l is linear.

The composition of bundle maps is defined in the obvious manner and so is the identity
morphism so that one can define the notion of bundle isomorphism in the standard way.
(b) If E and F are two vector bundles over the same manifold, then we denote by
Hom(F, F) the space of bundle maps E -> F which cover the identity ]lM. Such bundle
maps are called bundle morphisms.

For example, the differential Df of a smooth map f M - > N is a bundle map

Df : TM -> TN covering f .
Definition 2.1.31. Let E l> M be a smooth vector bundle. A bundle endomorphism of
E is a bundle morphism F : E -> E. An automorphism (or gauge transformation) is an
invertible endomorphism.

Example 2.1.32. Consider the trivial vector bundle M -> M over the smooth manifold
M. A section of this vector bundle is a smooth map u : M -> R". We can think of u as a
smooth family of vectors (141:) G RnlrrélVf-
An endomorphism of this vector bundle is a smooth map A : M -> EI1dRlRnl. We
can think of A as a smooth family of n X n matrices
GW) 0é(:11) al (.l
021(H3) a ) a2 (.1)
Ax 7 a;(a3) 6 C°°(M).

61/i'(H13 @§(f1>) 0,26/1)

The map A is a gauge transformation if and only if det A ; é 0 , V : 1 : € M .

Exercise 2.1.33. Suppose El, E2 -> M are two smooth vector bundles over the smooth
manifold with standard fibers F1, and respectively F2. Assume that both bundles are de-
fined by a common trivializing cover (U0,)0,6A and gluing cocycles
.QBQ : Uog -> GL(F1)7 h e : Uog -> GL(F2)-
 Natural Constructions on Manifolds 37

Prove that there exists a bijection between the vector space of bundle morphisms
Hom(E, F), and the set of families of smooth maps
{TO, : Uo -> Hom(F1,F2); a 6 A },
satisfying the gluing conditions
Tglilt) = h 0 0 ( ) T O ( ) ( x ) 9 0 1, V113 6 Ugh.

Exercise 2.1.34. Let V be a vector space, M a smooth manifold, um

an open cover of
M , and gag, hog : Uo, O UB -> GL(V) two collections of smooth maps satisfying the
cocycle conditions. Prove the two collections define isomorphic vector bundles if and only
they are cohomologous, i.e., there exist smooth maps (pa : Uo, -> GL(V) such that
-l
h ¢a9QB¢5

2.1.5 Some examples of vector bundles

In this section we would like to present some important examples of vector bundles and
then formulate some questions concerning the global structure of a bundle.

Example 2.1.35. (The tautological line bundle over RIP" and (CIP'"). First, let us recall
that a rank one vector bundle is usually called a line bundle. We consider only the complex
case. The total space o f the tautological or universal line bundle over CIP" is the space
un = UC = { (z,L) G (c"+1 x @1p"§ Z belongs to the line L C (STL-l-1

Let 77` : UC -> CIP" denote the projection onto the second component. Note that for every
L Q CIP", the fiber through 7r-1 (L) = 'LLC L coincides with the one-dimensional subspace
in c"+1 defined by L.

Example 2.1.36. (The tautological vector bundle over a Grassmannian). We consider

here for brevity only complex Grassmannian Gr klCCnl . The real case is completely similar.
The total space of this bundle is
Uk,n = up" = (z, L) E (Zn X Gr I¢(@N) ; Z belongs to the subspace L C Q" }.
If or denotes the natural projection 'fr : U p T L -> Gr I<(<€"), then for each L 6 Gr llCn) the
fiber over L coincides with the subspace in C" defined by L. Note that 'L£<€_1 = u (C.

(C C
Exercise 2.1.37. Prove that 'Un and Uk,n are indeed smooth vector bundles. Describe a
gluing cocycle for UC.

Example 2.1.38. A complex line bundle over a smooth manifold M is described by an

open cover l e a ) a € A and smooth maps
7

.9B0 : UQ M UB -> GL(l,(C) 2 IC* 7

38 Lectures on the Geometry of Manifolds

satisfying the cocycle condition

QMHI) = 9vB(33) • a W ) , W: 6 Uo, re UB re Up

Consider for example the manifold M = S2 C R3. Denote as usual by N and S the North
and respectively South pole. We have an open cover

s2:u0LJu>O, u0=s2\{5}, U1 52\{N}.

In this case, we have only a single nontrivial overlap, Un O Us- Identify U0 with the
complex line (C, so that the North pole becomes the origin Z = 0.
For every n 6 Z we obtain a complex line bundle Ln->S2 defined by the open cover
9

{U0, U1} and gluing cocycle

910(Z) Z
n
7 'v'z€(C* UP \ w
A smooth section of this line bundle is described by a pair of smooth functions

u0:U0-HC, u1:U1-MC,

which along the overlap U0 n U1 satisfy the equality u1(z) Z '"'u,0(z). For example, if
n > 0, the pair of functions

u0(z) = z"', u1(p) = l, Vp E U1

defines a smooth section of Ln.

Exercise 2.1.39. We know that (CIP1 is diffeomorphic to S2_ Prove that the universal
line bundle Un -> (III1 is isomorphic with the line bundle L-1 constructed in the above
example.

Exercise 2.1.40. Consider the incidence set

j {(flaL) E (<€"+1\{0}) X (CIP'"; Z E L}.

Prove that the closure of 3 in C"+1 X CIP" is a smooth manifold diffeomorphic to the total
space of the universal line bundle Un -> CIP". This manifold is called the complex blowup
of lcn-1-1 at the origin.

The family of vector bundles is very large. The following construction provides a very
powerful method of producing vector bundles.

Definition 2.1.41. Let f : X -> M be a smooth map, and E a vector bundle over M
defined by an open cover (UQ) and gluing cocycle (gag). The pullback of E by f is the
vector bundle f*E over X defined by the open cover f-1(UO,), and the gluing cocycle
(QQ5 o f)-
 Natural Constructions on Manycolds 39

One can check easily that the isomorphism class of the pullback of a vector bundle E is
independent of the choice of gluing cocycle describing E. The pullback operation defines
a linear map between the space of sections of E and the space of sections of f*E.
More precisely, if s 6 I`(E) is defined by the open cover (UQ), and the collection
of smooth maps (so, then the pullback f*s is defined by the open cover f-1(U0,), and
7

the smooth maps (sOf o f ). Again, there is no difficulty to check the above definition is
independent of the various choices.

Exercise 2.1.42. For every positive integer lc consider the map

Pa : QIP1 -> QIP1, P1<([20»21]) = [20'"»21'"]~

Show that paLn f\J

Lkna where Ln is the complex line bundle Ln (CIP1 defined in
Example 2.1.38.

Exercise 2.1.43. Let E -> X be a rank k (complex) smooth vector bundle over the mani-
fold X . Assume E is ample, i.e., there exists a finite family 81, . . . , S N of smooth sections
of E such that, for any a: E X, the collection {s1(:1:), . . . , sN(:1:)} spans Et. For each
CU € X we set

S 6 (CN ; 'u'si(a:) 0
..II

'U

Note that dim So, = N - k. We have a map F : X -> Gr I<(<€N) defined by aw->SJ-.
(a) Prove that F is smooth.
(b) Prove that E* is isomorphic with the pullback F*"llk n -

Exercise 2.1.44. Show that any vector bundle over a smooth compact manifold is ample.
Thus any vector bundle over a compact manifold is a pullback of some tautological bundle!

The notion of vector bundle is trickier than it may look. Its definition may suggest that
a vector bundle looks like a direct product manifold X vector space since this happens at
least locally. We will denote by M the bundle K" X M -> M.
'ii
Definition 2.1.4S. A rank T' vector bundle E -> M (over the field K = R, Q) is called
trivial or trivializable if there exists a bundle isomorphism E & Km. A bundle isomor-
phism E -> KM is called a trivialization of E, while an isomorphism K' -> E is called a
framing of E.
A pair (trivial vector bundle, trivialization) is called a trivialized, or framed bundle.

Remark 2.1.46. Let us explain why we refer to a bundle isomorphism up : M -> E as a

framing.
Denote by (el, . . . , 61" the canonical basis of KT'_ We can also regard the vectors et
as constant maps M -> KT', i.e., as (special) sections of Km. The isomorphism go deter-
mines sections f = (P(€¢) of E with the property that for every :1; 6 M the collection
(fl (513), . . . , f,~(:13) ) is a frame of the fiber Et.
40 Lectures on the Geometry of Manifolds

This observation shows that we can regard any framing of a bundle E -> M of rank T'
as a collection of 7° sections up, . . . , u which are pointwise linearly independent.

One can naively ask the following question. Is every vector bundle trivial? We can even
limit our search to tangent bundles. Thus we ask the following question.
15 it true that for every smooth manzfold M the tangent bundle T M i5 trivial (as a
vector bundle)?
Let us look at some positive examples.

Example 2.1.47. TS1 'E KS.. Let 9 denote the angular coordinate on the circle. Then 36
is a globally defined, nowhere vanishing vector field on S1. We thus get a map
£51 -> TS1 7 (s, 9) 1-> (S3979) E $951
which is easily seen to be a bundle isomorphism.
Let us carefully analyze this example. Think of S1 as a Lie group (the group of complex
numbers of norm l). The tangent space at Z = l, i.e., 9 = 0, coincides with the subspace
Re Z = 0, and - 69 11 is the unit vertical vector j.
8

Denote by Re the counterclockwise rotation by an angle 9. Clearly RE is a diffeomor-

phism, and for each 9 we have a linear isomorphism
De l0=0 RE : T151 -> T9s1.
Moreover,

69 = DO l9=0 Red
The existence of the trivializing vector field 36 is due to to our ability to "move freely and
coherently" inside S1. One has a similar freedom inside a Lie group as we are going to see
in the next example.

Example 2.1.48. For any Lie group G the tangent bundle TG is trivial.
To see this let n = dim G, and consider e l , . . . , en a basis of the tangent space at the
origin, To G. We denote by Rg the right translation (by g) in the group defined by
R g : :co-> : I : - g , VJJGG.
Rg is a diffeomorphism with inverse R9-1 so that the differential DRg defines a linear
isomorphism DRg : TIG -> TgG. Set
Eilis ) = DRgl€¢) 6 TAG, i = 1, ,n.
-
Since the multiplication G X G -> G, (g, h) »-> g h is a smooth map we deduce that the
vectors Et (g) define smooth vector fields over G. Moreover, for every g Q G, the collection
{E1 (g), . . . , En (g)} is a basis of TgG so we can define without ambiguity a map
<1> =@G -> TG, I
(9;X1 ,. . . ,XTL) »-> 9; XE¢l9) ).
One checks immediately that <I> is a vector bundle isomorphism and this proves the claim.
In particular TS3 is trivial since the sphere S3 is a Lie group (unit quaternions). (Using the
Cayley numbers one can show that TS7 is also trivial, see [112] for details).
 Natural Constructions on Manifolds 41

We see that the tangent bundle TM of a manifold M is trivial if and only if there exist
vector fields Xl, . . . ,Xm (m = d i M ) such that for each p 6 M, X1(p), . . . , X m ( p )
span T p M . This suggests the following more refined question.
Problem Given a manifold M , compute v ( M ) , the maximum number ofpointwise linearly
independent vector fields over M . Obviously 0 < v ( M ) < dim M and T M i5 trivial if
and only if v ( M ) = dim M . A special instance of this problem is the celebrated vector
field problem: compute vlgnl for any n > l.
We have seen that 'u(S") = n for n = 1, 3 and 7. Amazingly, these are the only cases
when the above equality holds. This is a highly nontrivial result, first proved by J.F. Adams
in [3] using very sophisticated algebraic tools. This fact is related to many other natural
questions in algebra. For a nice presentation we refer to [91].
The methods we will develop in this book will not suffice to compute v(S'") for any n,
but we will be able to solve "half" of this problem. More precisely we will show that
v(S") = 0 if and only if n is an even number.
In particular, this shows that TS211 is not trivial. In odd dimensions the situation is far more
elaborate (a complete answer can be found in [3]).

Exercise 2.1.49. vls2k 1) > l for any k l.

The quantity v ( M ) can be viewed as a measure of nontriviality of a tangent bundle.

Unfortunately, its computation is highly nontrivial. In the second part of this book we will
describe more efficient ways of measuring the extent of nontriviality of a vector bundle.

2.2 A linear algebra interlude

We collect in this section some classical notions of linear algebra. Most of them might be
familiar to the reader, but we will present them in a form suitable for applications in differ-
ential geometry. This is perhaps the least glamorous part of geometry, and unfortunately
cannot be avoided.
"Convention. All the vector spaces in this section will tacitly be assumed finite dimen-
signal, unless otherwise stated.

2.2.1 Tensor produets

Let E, F be two vector spaces over the field K (K lm). Consider the (infinite) direct
sum
'J'(E, F) K.
(e,f)€EXF
Equivalently, the vector space 'J'(E, F) can be identified with the space of functions C :
E X F -> K with finite support. The space 'J'(E, F) has a natural basis consisting of
"Dirac functions"
l if (:13,y) (6,f)
Z
5ef z E X F - > ] K , (.ac,y)»->
0 if (Hay) vi (6,/')
42 Lectures on the Geometry of Mamfolds

In particular, we have an injection 1

E X F Q m , h
Inside 'J'(E, F) sits the linear subspace R(E, F) spanned by

*6e,f - 5»e,f» A5e,f - 5e,Af» (5e+e',f - 5e,f - 5e',f» 5e,f+f' - 5e,f - 5e,f' 7

where 6,6I 6 E , f , f ' 6 F,andA 6 K. Now define

E ®11< F : = or(E,F)/R(E,F),

and denote by 71' the canonical projection or : 'J'(E, F) -> E (X) F. Set

f 7f((5e,f)~
a

€ I 1

We get a natural map

E F->E (X) F,eX f »->e®f.

Obviously L is bilinear. The vector space E ®1K F is called thetensor product of E and F
over K. Often, when the field of scalars is clear from the context, we will use the simpler
notation E (X) F. The tensor product has the following universality property.

Proposition 2.2.1. For any bilinear map QUO : E X F -> G there exists a unique linear map
CD : E ( X ) F -> G such that the diagram below is commutative.
I,
E X F > E (X) F
:l<I>
d> I
G

The proof of this result is left to the reader as an exercise. Note that if (80 is a basis of
E, and (fy) is a basis of F, then (6z ® fy) is a basis of E ® F, and therefore

dirge E ®K F = ldiII1K El - ldimg F) .

~
Exercise 2.2.2. Using the universality property of the tensor product prove that there exists
a natural isomorphism E (X) F = F (X) E uniquely defined by € f »-> f (X) e.

The above construction can be iterated. Given three vector spaces $1 , Et, E3 over the
same field of scalars K we can construct two triple tensor products:

(El (X) EQ) (X) E3 and E1 (X) (Ez (X) E3).

Exercise 2.2.3. Prove there exists a natural isomorphism of K-vector spaces

~
(El (X) EQ) (X) E3 = E1 ® (E2 ® E3).
lA word of cuufiwl; 6 is not linear!
 Natural Constructions on Manifolds 43

The above exercise implies that there exists a unique (up to isomorphism) triple tensor
product which we denote by E1 (X) E2 (X) E3. Clearly, we can now define multiple tensor
products: E1 (X) ---
(X) En .

Definition 2.2.4. (a) For any two vector spaces U, V over the Held K we denote by
Hom(U, V), or HOMK(U, V) the space of IN-linear maps U -> V.
(b) The dual of a K-linear space E is the linear space E* deaned as the space HOMK(E, K)
of K-linear maps E -> K. For any e* 6 E* and e 6 E we set
(e*,e) := e*(e).

The above constructions are factorial. More precisely, we have the following result.

Proposition 2.2.5. Suppose Ei, Ft, Gt, i = 1,2 are K-vector spaces. Let To 6
Hom (Ei, Fi), Si 6 Hom(Fi, Gi), i = 1, 2, be two linear operators. Then they naturally
induce a linear operator
T = T 1 ® T 2 I E 1 ® E 2 - F 1 ® F 2 » S1®S2-F1®F2
uniquely deaned by

T1 ® T2(€1 ® 821 = lT181l ® lT2€2l, \V/€¢ G E

and satisfying

(51 ® SQ) O (Tl (X) TQ) = (51 O To) (X) (52 O TQ).
(b) Any linear operator A : E -> F induces a linear operator AT : F* -> E* uniquely
defined by
(A*f*,@) = (f*,Ae)> ve E E , f* E F*.
The operator At is called the transpose or adjoint of A. Moreover
(A o BV = BT o AT, VA E Hom(F,G), B E Hom(E,F).

Exercise 2.2.6. Prove the above proposition.

Remark 2.2.7. Any basis (€zl1<1;$n

_ of the n-dimensional IN-vector space determines a
basis (€@l1sz$n of the dual vector space V* uniquely deaned by the conditions

1
(e) by)
0
We say that the basis (evil is dual to the basis (eil- The quantity (65 ) is called the Kronecker
symbol.
A vector 'U 6 V admits a decomposition
n
ii
g
II

'V€i7
i t
44 Lectures on the Geometry of Mamfolds

while a vector v* QV* admits a decomposition

TL

u* fu- ez
1l=1
Moreover,
TL

('v*/u) U-U
ii 1

Classically, a vector 'U in V is represented by a one-column matrix

,U1

Q
II
7

,Un

while a vector u* is represented by a one-row matrix

IU*
[ful U;;].
Then
'01

('v*/u) I [U1 7

,Un

where the - denotes the multiplication of matrices.

Using the functoriality of the tensor product and of the equalization construction one proves
easily the following result.

Proposition 2.2.8. (a) There exists a natural isomorphism

E* (X) F* E (E (X) F)* 7
uniquely defined by
E*® F* 9€*®at* I
I
7

where

(Le*®f*vX®Y)= (€*»fv)(f*,y), 've;I:€F, 3/QF.

In particular, this shows E* (X) F* can be naturally identified with the space of bilinear
maps E X F -> K.
(b) The adjunction morphism E* (X) F -> Hom ( E , F), given by
F* (X) F 9e*®f I >Te*®]- €Hom(E,F),
where

To*®f($) :z (€*»x1 )f> Vat E E,

2
is an isomorphism.

2 The finite dimensionality of E is absolutely necessary. This adjunction formula is known in Bourbaki circles
as "for rule d'adjonction char 61 Carfare".
 Natural Constructions on Manifolds 45

Exercise 2.2.9. Prove the above proposition.

Let V be a vector space. For r, S > 0 set

my) == V ®'I"
(X)
(V*)®s
7

where by definition V ®0 lV*l®0 = K. An element of in; is called tensor of lype (is).

Example 2.2.10. According to Proposition 2.2.8 a tensor of type ( L l ) can be identified

with a linear endomorphism of V, i.e.,
<r}(v) 2- End (v),
while a tensor of type (0, of) can be identified with a k-linear map
Vx -- x V->]K.
k

A tensor of type (v, 0) is called contra variant, while a tensor of type (0, s) is called covari-
ant. The tensor algebra of V is deaned to be

<r(v) my).

We use the term algebra since the tensor product induces bilinear maps
I /

®:'J';x'J';, ->q'§++§,.

The elements of 'J'(V) are called tensors.

Exercise 2.2.11. Show that ('T(V)» 7 (X) I is an associative algebra.

Example 2.2.12. It is often useful to represent tensors using coordinates. To achieve this
pick a basis (80 of V, and let (ei) denote the dual basis in V* uniquely defined by
l if'i=j
(7 €j)
0 ifigéj

We then obtain a basis of <r;(v)

{ e¢1®~~-®e¢,,®e31® ®€ i5/ 1 <i0 ~j@ < dim V}-

Any element T € €T;°(V) has a decomposition
T¢1...¢,.
T .]1..._]5
6111 (X) ®e, @ € j 1 (X) 7 ®ejs 7

where we use Einstein convention to sum over indices which appear twice, once as a su-
perscript, and the second time as a subscript.
Using the adjunction morphism in Proposition 2.2.8, we can identify the space or (V)
with the space End(V) a linear isomorphisms. Using the bases (QI and (co), and Einstein's
convention, the adjunction identification can be described as the correspondence which
associates to the tensor A = are (X) of Q :ri(V), the linear operator LA : V -> V which
maps the vector 'U = up et to the vector LAU = ajfuf et.
46 Lectures on the Geometry of Mamfolds

On the tensor algebra there is a natural contraction (or trace) operation

-1
tr : T; -> _1

uniquely defined by
1 1 ®""U»,~®U2®°_® S
tr 'U1 ®-~-®v~® 'LL ®---®uS` (Fu, 7 U1),U2 'LL 7

'v'"7€,u7€V*
In the coordinates determined by a basis 060 of V, the contraction can be described as
¢2...¢,.
(tr T ) . ] 2 . . . _ ] 5 (Tub2 ) 7

where again we use Einstein's convention. In particular, we see that the contraction coin-
cides with the usual trace on by; (V) 2 End (V).

2.2.2 Symmetric and skew-symmetric tensors

Let V be a vector space over K = R, (C. We set W W ) := :kg (V), and we denote by S the
group of permutations of T' objects. When T' = 0 we set 80 := {1}.
Every permutation O' 6 S determines a linear map 'J'""(V) -> 'J"l(V), uniquely deter-
mined by the correspondences

01 (X) ®v1~»->00(1)®---®00(¢), 'v'v1,...,'v',°€V

We denote this action of O' 6 S on an arbitrary element t 6 U""(V) by it.

In this subsection we will describe two subspaces invariant under this action. These
are special instances of the so called Schur functors. (We refer to [46] for more general
constructions.) Define
1
S T' 'J"l(V) -> 'T"(V), S 07t,
t~(t): al
u'S

and

L if dim V
A 'T"lvl -> 'T"(V)7 A t) EGGS, 6(c7)(7t
s 0

T'
0 if dim V
Above, we denoted by 6(0) the signature of the permutation U. Note that

A0 SO HK.

The following results are immediate. Their proofs are left to the reader as exercises.

Lemma 2.2.13. The Op€l'C1tOl"S A 7° and S are projectors of 'T"° i.e.,

SO S fr 7 A 2or A or.

Moreover,

0S,~(t) = S,~(ot) S ,~(t), 0A,~(t) A 1-(0¢) = 6(0) A ,~(t), Vt Q gmv).

 Natural Constructions on Manifolds 47

Definition 2.2.14. A tensor T 6 'T"(V) is called symmetric (respectively skew-symmetric)

if
S MT) =T (respectively A ,,(T) T).

The nonnegative integer 1° is called the degree of the (skew -)symmetric tensor.
The space of symmetric tensors (respectively skew-symmetric ones) of degree r will be
denoted by S V (and respectively A"°V).

Set

S'V .
o
Ql§s'°v, andA'V @A s
'r'>0 'r">0

Definition 2.2.15. The exterior product is the bilinear map

A I A V x ASV -> A'°+sv,
deaned by
(r + s)!
w"A17S A v~+sl°u (X) to), Vi G AWE Us G A.SV.
res!
Proposition 2.2.16. The exterior product has the following properties.
(a) (Associativity)

(04A/3)A7=@/\(/5'/W)\ \1a, l5'v € A°v.

In particular;

U1 A - - - /\ up = k!Ak(v1 ® (X) up) = 6(07)U0(1) (X) ®UU(k)7 \V'Ui G V.

o'€Sk

(b) (Super-commutativily)

wAsS I-l)'r'g17S /\ w , Vi Q AV, wS E ASV.

Proof. We first define a new product "Al" by

CU/\17]S A 1°-i-slW (X) UI)

which will turn out to be associative and will force A to be associative as well.
To prove the associativity of /\1 consider the quotient algebra Q* = T* IN*, where 'T*
is the associative algebra (@ ,~>0T'°(V), -|-, ®), and al* is the bilateral ideal generated by
the set of squares { v ® v / U 6 V } . Denote the (obviously associative) multiplication in Q*
by U. The natural projection 'fl' : 'T* -> Q* induces a linear map 7'l` : A' V -> Q*. We will
complete the proof of the proposition in two steps.
Step 1. We will prove that the map Tl' : A' V -> Q* is a linear isomorphism, and moreover
7r(aA1B) = tr(a) U B). (2.2.1)

In particular, /\1 is an associative product.

48 Lectures on the Geometry of Mamfolds

The crucial observation is

7r(T) = or( A ,-(T)), Vt 6 frlvl. (2.2.2)

It suffices to check (2.2.2) on monomials T ; € 1 ® . . . (X) e , et 6 V. Since

@+v2e,wvev
u(mod U*). Hence, for any 0 QS

7r(€1 (X) - -- (X) e) = e(o)tr(eu(1l (X) (X) €o('r)l° (2.2.3)

When we sum over O' 6 S in (2.2.3) we obtain (2.2.2).

To prove the injectivity of 71' note First that A* w*) 0. If tr(w) 0 for some Ld 6
As V, then w 6 kerw = U* re A' V so that

w A *(w) 0.

The surjectivity of 7/ follows immediately from (2.2.2). Indeed, any 7r(T) can be alterna-
tively described as tr(w) for some w 6 A' V. It suffices to take w = A* (T).
To prove (2.2.1) it suffices to consider only the special cases when a and B are mono-
mia1s:

a A 7 ° l € 1 ® ' ° ' ® € r l 7 /3 As(f1 ( X ) (X) fs)-

We have

(X) (X) mfr) ( X ) A 5(f1 (X) (X) full

®e,~)® A 5l.t1® (X) fs) )
®e,~))U of( A 5(f1 (X) (X) f5)) = we) LJ w).
Thus /\1 is associative.
Step 2. The product A is associative. Consider a 6 AT'V, B6 ASV and v 6 Ate. We
have
f( + )! (r+sl! iT+s+tl!
(aAB)A'y QA1/5 As Hs! (7° -|- s)!t! (@A15)
A17

(r -|- S + t)! (T + S -|- t)!

r!s!t! (@A15 lA17 r!8!t! &A1(5Aw) = a /\ (0 /\ vl-
The associativity of A is proved. The computation above shows that

e1A---Aek=k!Ak(e1®-~-®ek).

(b) The supercommutativity of A follows from the supercommutativity of /\1 (or U). To
prove the latter one uses (2.2.2). he details are left to the reader.

Exercise 2.2.17. Finish the proof of part (b) in the above proposition.
 Natural Constructions on Manifolds 49

The space A' V is called the exterior algebra of V. The operation A is called the
exterior product. The exterior algebra is a Z-graded algebra, i.e.,

( A V /\ ASK pA*l+SV, 'v'1°,s.

Note that AT'V 0 for 1°> dim V (pigeonhole principle).

Definition 2.2.18. Let V be an n-dimensional K-vector space. The one dimensional vector
space A'"V is called the determinant line of V, and it is denoted by det V.

There exists a natural injection LV :V<->A' V, 1,V(0) U, such that

LV('u) /\ Le (Fu) 0, Vu € V
This map enters crucially into the formulation of the following universality property.

Proposition 2.2.19. Let V be a vector space over K. For any K-algebra A, and any linear
map QUO : V -> A such that Tb(;z:)2 = 0, there exists an unique morphism of K-algebras
® : A' V -> A such that the diagram below is commutative

V ` r
LV
> A. V
I
I
I <1> 7
I
I

A
i.e., ® O LV ¢,

Exercise 2.2.20. Prove Proposition 2.2.19.

The space of symmetric tensors S' V can be similarly given a structure of associative
algebra with respect to the product

a-B S ~+5(a®B), v a e s ' ° v , @ @ s 8v

s
•

The symmetric product "-" is also commutative.

Exercise 2.2.21. Formulate and prove the analogue of Proposition 2.2.19 for the algebra
s' V.

It is often convenient to represent (skew-)symmetric tensors in coordinates. If

617 , en is a basis of the vector space V then, for any l < 7° < n, the family
6111 /\ /\€,/ 1 Gil < <1l1~<n
is a basis for AT'V so that any degree 7° skew-symmetric tensor CU can be uniquely repre-
sented as
.1 . . . . . .
cu- cu' "'"61 A - ~ A € 2 ,
1Si1<---<i»,-S'f1
50 Lectures on the Geometry of Mamfolds

Symmetric tensors can be represented in a similar way. To any multi-index I

(in, . . . , in) 6 Z20 we can associate the symmetric monomial
eI a
I 1
e11 in
en
and the resulting collection (of) is a basis of S'V.
The A' and the S' constructions are factorial in the following sense.

Proposition 2.2.22. Any linear map L : V -> W induces a natural morphisms of algebras
A' L : A' V -> A' W , S ' L : S ' V -> s ' w
uniquely dejined by their actions on monomials
A. Ll'U1 /\ -- A U ) = (Lil) /\ - - /\ ILUI,
and
S. Ll'U1, . . . ,u,~) = (Lil) - - - ( L ) .
B
Moreover; If U A) V >W are Mo linear maps, then
ATI(BAl = (ATB)(A'°A)7 S"(BA) = (S"B)(STA).
Exercise 2.2.23. Prove the above proposition.

In particular, if n = dirn]K V, then any linear endomorphism L : V -> V defines an

endomorphism
A"L : derV = A"V -> derV.
Since the vector space det V is 1-dimensional, the endomorphism A"L can be identified
with a scalar det L, the determinant of the endomorphism L.

Definition 2.2.24. Suppose V is a finite dimensional vector space and A : V -> V is an

endomorphism of V. For every positive integer T' we denote by 07,~(A) the trace of the
induced endomorphism
AT'A A*V -> A V ,
and by ,MAl the trace of the endomorphism AT` : V -> V. We define
A) = 1, A) = d i v .

Exercise 2.2.25. Suppose V is a complex n-dimensional vector space, and A is an endo-

morphism of V .
(a) Prove that if A is diagonalizable and its eigenvalues are CL17 7 A N a then

Ur(/4) Z (Lil ° ' ° & ¢ _ , ¢,.(A) ICL1 -I- + Ag.

1$1;1<---<i,.<n

(b) Prove that for every sufficiently small Q Q we have the equalities
Z

d
det(]lv -|- zA) Ur ( A w , - -dol o g det ( ]1 pA) ¢,,(A)z"-1
j>0 'r>1
 Natural Constructions on Manifolds 51

The functors A' and S' have an exponential like behavior, i.e., for any finite dimen-
sional vectors spaces V , W, there exists natural isomorphisms of vector spaces
A'V (X) A'W§A°(V ® (2.2.4a)

S'V (X) S ' W § S ' (v ® w). (2.2.4b)

To define the isomorphism in (2.2.4a) consider the bilinear map

<15 : A ' V X A ' W - > A ' ( V ® w),

uniquely determined by
g25('u1A-~-A0,~,w1A-~-/\wS)=v1/\---Av,~Aw1A---/\w5.

The universality property of the tensor product implies the existence of a linear map
<I> A ' V ® A ' W - > A ' IVQBW),

such that Q o L = go, where L is the inclusion of A' V X A'WinA'V (X) A' W. This map
is clearly surjective and it is an isomorphism since
dim A' V X A. w = dim A' (V GB W) dim V+dim W

This proves (2.2.4a).

To prove (2.2.4b) observe as above that there exists a linear map

(b S' V ® S' W -> S°(V ® W)

uniquely determined by the equality
'Up, 'U11 U 1 ° ° ° U j ' 'WE •
(150)1 Wk. c » u

To prove that it is an isomorphism, fix basis (fifzllgigm of V and basis (fjhgjgn of W.

Note that the collection
{ eI (X) fJ; I 6 Z ; * 0 , J e Z ' 2 0 }
is a basis of S' V (X) S' W that is mapped bijectively by QUO to the collection
{ 6] - fJ; I e 220, J 6 ago }
which is a basis of S' (V ® W).
We want to mention a few general facts about Z-graded vector spaces, i.e., vector spaces
equipped with a direct sum decomposition
V
n Z
vn _
(We will always assume that each Vn is finite dimensional.) The vectors in Vn are said to
be homogeneous, of degree n. For example, the ring of polynomials lK[x] is a K-graded
vector space. The spaces A' V and S' V are Z-graded vector spaces.
The direct sum of two Z-graded vector spaces V and W is a Z-graded vector space
with
(V ® weTL := Vn ® Wn-
52 Lectures on the Geometry of Mamfolds

The tensor product of two Z-graded vector spaces V and W is a Z-graded vector space
with
(V WM Vf Ws.

s o
(X) (X)

T'-l-8277,

To any Z-graded vector space V one can naturally associate a formal series Pvltl E
Z[[t, V111 by

n Z
The series Pv (t) is called the Poincaré series o f V.
Example 2.2.26. The Poincaré series of ]K[ac] is
_l.t** 1 _ l
PKlx1(t)= 1 + t t2
- 1 - 15'
Exercise 2.2.27. Let V and W be two Z-graded vector spaces. Prove the following state-
ments are true (whenever they make sense).
(21) Pveswltl = PvU) + Pw(15l-
-
(b) Pv®w(15) = Pv(t) pW(t).
(C) dimV = Pall.

Definition 2.2.28. Let V be a Z-graded vector space. The Euler characteristic of V, de-
noted by X(V), is defined by
x(V) == Pv(-1l n Z
(-l)"'dimvm

whenever the sum on the right-hand side makes sense.

Remark 2.2.29. If we try to compute xllK[a2]) using the first formula in Definition 2.2.28
we get xumwl) = l I2, while the second formula makes no sense (divergent series).

Proposition 2.2.30. Let V be a K-vector space of dimension n. Then

PA° alt) = (l -|- U",
and
n d n-1
l
Ps' v(t)

In particular, dim A' V = 2" and X(A' V)

I
n
0.
1>! C-) dt

Proof. From (2.2.4a) and (2.2.4b) we deduce using Exercise 2.2.27 that for any vector
spaces V and W we have
PA- (ve9w) (to = PA-V(t) • PA- w('5l and Ps- (vebw) (to = P5»V(t) ' P$»W(t).
In particular, if V has dimension n, then V 2 K" so that
PA~ vltl (PA~ K(t))" and Ps' vltl (P5'°
Z . Z

The proposition follows using the equalities

.
PA H<(15) 1 -|- 15, and Ps' K05) = PK[x] (t)
1
1-t
.
 Natural Constructions on Manifolds 53

2.2.3 The "super" slang

The aim of this very brief section is to introduce the reader to the "super" terminology. We
owe the "super" slang to the physicists. In the quantum world many objects have a special
feature not present in the Newtonian world. They have parity (or chirality), and objects
with different chiralities had to be treated differently.
The "super" terminology provides an algebraic formalism which allows one to deal
with the different parities on an equal basis. From a strictly syntactic point of view, the
"super" slang adds the attribute super to most of the commonly used algebraic objects. In
this book, the prefix "s-" will abbreviate the word "super"

Definition 2.2.31. (a) A s-space is a Z2-graded vector space, i.e., a vector space V
equipped with a direct sum decomposition V = V0 ® VI .
(b) A s-algebra over K is a Z2-graded K-algebra, i.e., a IN-algebra A together with a direct
sum decomposition A = A0 ® Al such that A i . As C Ai-*7 (mod 2). The elements in A'i
are called homogeneous of degree i. For any a 6 A we denote its degree (mod 2) by lal.
The elements in A0 are said to be even while the elements in Al are said to be odd.
(c) The supercommutator in a s-algebra A = A0 ® A1 is the bilinear map

..
[,]S:A X A->A,

defined on homogeneous elements wi E A , Do E As by

lwnvjls == w t v j - (-1)"17i(»Jj

An s-algebra is called .9-commutative, if the suppercommutator is trivial, [o 7 .] S E 0.

Example 2.2.32. Let E = E0 ® E1 be a 5-space. Any linear endomorphism T 6 End ( E )

has a block decomposition
II

where Tri Q End (18 Et ). We can use this block decomposition to describe a structure of
s-algebra on End (E). The even endomorphisms have the form

while the odd endomorphisms have the form

54 Lectures on the Geometry of Mamfolds

Example 2.2.33. Let V be a finite dimensional space. The exterior algebra A'V is natu-
rally a s-algebra. The even elements are gathered in
Aeven V AT'V ,
T' even
while the odd elements are gathered in
Aoddv = As.
T' odd
The .9-algebra A'V is .9-commutative.

Definition 2.2.34. Let A = A0 ® A1 be a s-algebra. An s-derivation on A is a linear

operator on D 6 End (A) such that, for any an 6 A,
[D, Lx]End(Al Z I/Da: a (2.2.5)
where [ , SEnd(.A) denotes the supercommutator in End (A) (with the s-structure deaned
in Example 2.2.32), while for any Z 6 A we denoted by LZ the left multiplication operator
a »-> Z a.
a

An .9-derivation is called even (respectively odd), if it is even (respectively odd) as an

element of the .9-algebra End (A).

Remark 2.2.35. The relation (2.2.5) is a super version of the usual Leibniz formula. In-
deed, assuming D is homogeneous (as an element of the s-algebra End (All then equality
(2.2.5) becomes
D ) = (Dlv)y + (-1)l""Dlw(Dy),
for any homogeneous elements Xv by 6 A.

Example 2.2.36. Let V be a vector space. Any u* 6 V* defines an odd 5-derivation of

A'V denoted by in* uniquely determined by its action on monomials.
T'

in* l'U0 /\ U1 /\ AU (-1Y<u*7 'Ui>'U0 /\ U1 /\ A A A AU.

i 0
As usual, a hat indicates a missing entry. The derivation in* is called the interior derivation
by u* or the contraction by u*. Often, one uses the alternate notation u*J to denote this
derivation.

Exercise 2.2.37. Prove the statement in the above example.

Definition 2.2.38. Let A = (A0 ® A1, +A , ]) be an s-algebra over K, not necessarily

associative. For any ac 6 A we denote by Ras the right multiplication operator, a »-> [0, ac].
A is called an s-Lie algebra if it is s-anticommutative, i.e.,
[as, y] -|- (-l)|""||Y|[y, at] = 0, for all homogeneous elements x, y e A,
and Van 6 A, Ras is a s-derivation.
When A is purely even, i.e., A1 = {0}, then A is called simply a Lie algebra. The
multiplication in a (s-) Lie algebra is called the (s-)bracket.
 Natural Constructions on Manifolds 55

The above definition is highly condensed. In down-to-earth terms, the fact that Ras is a
s-derivation for all as 6 A is equivalent with the super Jacobi identity

Hy, 2], x] Hy, 111, z] + l-1ll33ll1/l[y [z, SDH 7 (2.2.6)

for all homogeneous elements cc, y, Z Q A. When A is a purely even K-algebra, then A is
a Lie algebra over K if [ , ] is anticommutative and satisfies (2.2.6), which in this case is
equivalent with the classical Jacobi identity,

Hw,y], z] + He, z],;u] -I- [[w1]»y] 0> Van°,y,z€A. (2.2.7)

Example 2.2.39. Let E be a vector space (purely even). Then A End (El is a Lie
algebra with bracket given by the usual commutator: [a, 6] = ab - be.

Proposition 2.2.40. Let A = A0 ® A1 be a 5-algebra, and denote by DerS (Al the vector
space of s-derivations of A.
(a) For any D 6 DerS(A), its homogeneous components Do, D1 6 End (Al are also
s-derivations.
(b) For any D, D' 6 DerS(A), the s-cornrnutator [D, D']End(Al is again an 5-derivation.
(c) Vi 6 A the bracket B 11: : a »-> [a, buzz]S is a s-derivation called the bracket derivation
determined by at. Moreover

[B"', By]End(A) = B[""i"]S, Vat, y G A.

Exercise 2.2.41. Prove Proposition 2.2.40.

Definition 2.2.42. Let E = E0 ® E1 and F = F0 ® F1 be two 5-spaces. Their 5-tensor

product is the 5-space E (X) F with the Z2-grading,

Fr E¢®Fi, 0, 1.
..II

(E (X) 6
'i-l-jEe (2)

To emphasize the super-nature of the tensor product we will use the symbol sc®9s instead of
the usual "®".

Exercise 2.2.43. Show that there exists a natural isomorphism of .9-spaces

V* A°V f"*J
DerS (A°V),
(X)

Do*®w
uniquely determined by u* X w »-> DU*®"", where is s-derivation deaned by

D'U*®wlul (u*,u)w, 've 6 V

Notice in particular that any s-derivation of A°V is uniquely determined by its action on
A1V. (When Ld = 1, D'U* ®1 coincides with the internal derivation discussed in Exam-
ple 2.2.36.)
56 Lectures on the Geometry of Mamfolds

Let A = A0 ® A1 be an .9-algebra over K R,(c. A supertrace on A is a K-linear

map 7' A -> K such that,
T([22,y]s) 0 Very 6 A.
i

If we denote by 1A»A1s the linear subspace of A spanned by the supercommutators

{[X7Y]S 3 $79 E A }»
then we see that the space of s-traces is isomorphic with the dual of the quotient space
A/[A,A]S.
Proposition 2.2.44. Let E = E0 ® E1 be ajinite dimensional 5-space, and denote by A the
5-algebra of endomorphisms of E. Then there exists a canonical .9-trace trS on A uniquely
dejined by
tI's]1E I dim E0 - dim E l .
In fact, If T 6 A has the block decomposition
$4
II

then
tr . T = tr T00 - tr T11.

Exercise 2.2.45. Prove the above proposition.

2.2.4 Duality
Duality is a subtle and fundamental concept which permeates all branches of mathematics.
This section is devoted to those aspects of the atmosphere called duality which are relevant
to differential geometry. In the sequel, all vector spaces will be tacitly assumed finite
dimensional, and we will use Einstein's convention without mentioning it. K will denote
one of the fields R or C
Definition 2.2.46. A pairing between two K-vector spaces V and W is a bilinear map
B V X W->]K.

Any pairing B V W-> K defines a linear map

.
X

] I 8 . V - > W * , v»->B(u, )kW* 7

called the adjunction morphism associated to the pairing.
Conversely, any linear map L : V -> W* defines a pairing
BL V X W -> K , B(v,w) = (L'u)(w), Vu G V, w € W.
Observe that EBL = L. A pairing B is called a duality if the adjunction map III is an
isomorphisms.
..
Example 2.2.47. The natural pairing ( , ) : V* X V -> K is a duality. One sees that
H( 0 , 0 ) = ] L V * Z V * - > V * This pairing is called the natural duality between a vector space
and its dual.
 Natural Constructions on Manifolds 57

generate quadratic form

natural isomorphism
..
Example 2.2.48. Let V be a finite dimensional real vector space. Any symmetric nonde-
) : V X V -> R defines a (self)duality, and in particular a
7

.. U( 0 , 0 I Z V - > V *
When , ) is positive definite, then the operator L is called metric duality or lowering-
the-indices map. This operator can be nicely described in coordinates as follows. Pick a
basis ( I of V, and set
gig
Let ( e ) denote the dual basis of V* defined by
(et, € ) 'Soj , W, j.
The action of L is then
Lei 9¢j€J

Example 2.2.49. Consider V a real vector space and Ld : V X V -> R a skew-symmetric

bilinear form on V. The form Ld is said to be symplectic if this pairing is a duality. In this
case, the induced operator Hw : V -> V* is called symplectic duality.

Exercise 2.2.50. Suppose that V is a real vector space, and w : V x V - > ] R is a symplectic
duality. Prove the following.
(a) The V has even dimension.
(b) If (80 is a basis of V, and we OJ(e£ , € j l , then d€t(w¢j)1g¢,jgdimv ¢ 0.

The notion of duality is compatible with the factorial constructions introduced so far.

Proposition 2.2.51. Let B : V X We -> R (i = 1, 2) be two pairs of spaces in duality.

Then there exists a natural duality
B = 81 (X) B2 I (V1 (X) V2) X (W1®W2)->]R,
uniquely determined by
; \
HB1®B2 = HB1 (X) HB2 \ I
8(U1 (X) 'U2,'U)1 (X) 'U1QI = B1l'U1,'U)1l ° B 2 ( U 2 , W 2 l -

Exercise 2.2.52. Prove Proposition 2.2.51.

Proposition 2.2.51 implies that given two spaces in duality B : V X W -> K there is a
naturally induced duality
B (Xm : v®'" x W ®"f* ->K.
This defines by restriction a pairing
AfrB : ATV X A"°W -> K
uniquely determined by
AB I\'v»,~,w1 /\ Aw det ( B
..II

l'Ul /\ (U p, W y l l l g g r •

Exercise 2.2.53. Prove the above pairing is a duality.

58 Lectures on the Geometry of Manifolds

In particular, the natural duality ( , .. ):V* V - > K induces a duality

<. .):ATV*
X

7 X AT'V->]R,
and thus defines a natural isomoiphism
AW* 2 (A*v>*
This shows that we can regard the elements of AfrV* as skew-symmetric 1°-linear forms
V" ->K.
A duality B : V X W-> K naturally induces a duality B U V * X W * - > K by
Bt('u*,w*) =l'u',I[Blw*),
where IIB : V -> W* is the adjunction isomorphism induced by the duality B.

..
Now consider a (real) Euclidean vector space V. Denote its inner product by (
The self-duality defined by l 7 ) induces a self-duality
. .>-
7

l. . , ):A"V X ATIV->]R,
determined by
(U1 A - - - A v , ~ , w 1 A - - - A w , - ) .. det((v,w))1s s (2.2.8)
The symmetric 7° X T matrix
(U1,'U1 ('U1»U2l ('U1»Uv~l

(U2»U1l ('U2»U2l IU?) u)

G (v1,...,v,~) ..

( u m ) ('u»,°,v2) l'07f°7'0'f°l

is called the Grammian of the vectors "up, . . . , v . It is positive semidefinite and, if the
vectors '01, . . . , o are linearly independent, then it is also nonsingular. This shows that
the bilinear form in (2.2.8) is symmetric and positive definite. We have thus proved the
following result.

Corollary 2.2.54. An inner product on a real vector space V naturally induces an inner
product on the tensor algebra y ( V ) , and in the exterior algebra A°V.

In a Euclidean vector space V the inner product induces the metric duality L : V - > V * .
This induces an operator L : T; (V) -> T§°+11(V) defined by
L ( v 1 ® . . . ® v , - ® u1 (X)
...® uS) ) ® IILu1 (X) u (X)
; (112 (X) (X) 'u»S)- (2.2.9)
(X) v
1

The operation defined in (2.2.9) is classically referred to as lowering the indices.

The reason for this nomenclature comes from the coordinate description of this opera-
tion. If T 6 'y§(V) is given by
T il-
I
Tm...j;€i1 (X) ® e , ®ei1 (X) (X) eds 7
then
LT ¢.2....¢,. = ..T¢¢2....¢,
( IJJ1---§v~ 91] J 1 . - - a s 7

where gig I
( i i 7 o i l . The inverse of the metric duality L-1 V* -> V induces a linear
operation 7; (V) -> <r;+; (V) called raising the indices.
 Natural Constructions on Manifolds 59

Exercise 2.2.55 (Cartan). Let V be a Euclidean vector space. For any 'U Q V denote by
6v (resp. i'U) the linear endomorphism of A*V defined by e'Uw = 'U A Ld (resp. i'U /Le*

where iv* denotes the interior derivation defined by u* 6 V*-the metric dual of "u, see
Example 2.2.36). Show that for any u, 'U Q V
[e 7'u,]s I e v e -|- iuev I (to, U ) ] l A . V .

Definition 2.2.56. Let V be a real vector space. A volume form on V is a nontrivial linear
form on the determinant line of V , / u : d e t V - > R

Equivalently, a volume form on V is a nontrivial element of det V* (n = dim V).

Since det V is 1-dimensional, a choice of a volume form corresponds to a choice of a basis
of det V.

Definition 2.2.57. (a) An orientation on a vector space V is a continuous, surjective map

or : derV \ {0} -> {::1}.
We denote by O'r(V) the set of orientations of V. Observe that O'r(V) consists of pre-
cisely two elements.
(b) A pair (V, or), where V is a vector space, and or is an orientation on V is called an
oriented vector space.
(c) Suppose or E Or(V). A basis LU offset V is said to be positively oriented if or(w) > 0.
Otherwise, the basis is said to be negatively oriented.

There is an equivalent way of looking at orientations. To describe it, note that any
nontrivial volume form n on V uniquely specifies an orientation or given by
0ru(W) := sign,u(w), Vi E det V \ {0}.
We define an equivalence relation on the space of nontrivial volume forms by declaring
H1 H2 1 ' u1(w)u2((»J) > 0, Vi 6 derV \ {0}.
Then
M1 iv M2 KI ,I
\
07'/-01 I OTm2 .
To €V€I'Y orientation 07' WC ChIll ElSSOCi2lt€ 2111 €qI.liV211€I1C€ class [/J]o»,» of volume forms such
that
1u(w)o'r(w) > 0, Vi 6 det V \ {0}.
Thus, we can identify the set of orientations with the set of equivalence classes of nontrivial
volume forms.
Equivalently, to specify an orientation on V it suffices to specify a basis w of det V.
The associated orientation orb, is uniquely characterized by the condition
orw(w) = 1.
60 Lectures on the Geometry of Manifolds

To any basis {et, ..., en} of V one can associate a basis el A ---
A eTL of det V. Note that a
permutation of the indices 1, . . . , n changes the associated basis of det V by a factor equal
to the signature of the permutation. Thus, to define an orientation on a vector space, it
suffices to specify a total ordering of a given basis of the space.
An ordered basis of an oriented vector space (V, or) is said to be positively oriented if
so is the associated basis of det V.
Definition 2.2.58. Given two orientations 0T1, 0T2 on the vector space V we define
0 r 1 / 0 r 2 Q {d:1}

to be
0m/0r2 := or1(w)or2(w), Vi 6 det V \ {0}.
We will say that 0T1 /org is the relative signature of the pair of orientations o f , 0T2.

g( 7 ..
Assume now that V is a Euclidean space. Denote the Euclidean inner product by
The vector space det V has an induced Euclidean structure, and in particular,
there exist exactly two length-one-vectors in det V. If we fix one of them, call it w, and we
think of it as a basis of det V, then we achieve two things.

. First, it determines a volume form #9 defined by

. #go/\Wl
Second, it determines an orientation on V.
Z A.

Conversely, an orientation or 6 Or(V) uniquely selects a length-one-vector w Wor I

in det V, which determines a volume form ,ug = 1 g". Thus, we have proved the following
result.

Proposition 2.2.59. An orientation 07° on a Euclidean vector space (V, g ) canonically

selects a volume form on V , henceforth denoted by Detg = D e t g .

Exercise 2.2.60. Let (V, g) be an n-dimensional Euclidean vector space, and 07° an orien-
tation on V. Show that, for any basis 111, . . . 'un of V, we have
Detg'°(v1 /\ - - - As" 07° 01 A - - - A v n ) v<detg<@~»)).
If V = R2 with its standard metric, and the orientation given by 61 /\ 62, prove that
IDetg'°(u1 /\ 'U2 lI
is the area of the parallelogram spanned by U1 and UQ.

Definition 2.2.61. Let (V,g, or) be an oriented, Euclidean space and denote by Detg the
associated volume form. The Berezin integral or (be rezinian) is the linear form

A°V ->]R,
g
defined on homogeneous elements by
n
_* / 0
if deg Ld < dim V
'9
. Detg"°w if deg Ld dim V
 Natural Constructions on Manifolds 61

Definition 2.2.62. Let CU E A2v, where (V»9, or) is an oriented, Euclidean space. We
define its pfa173an as
I

0 if dim V is odd

..II
p f (w) Pfg°" ((»Jo
gexpa -Detg"°(wA") if dim V = 271 7

where exp w denotes the exponential in the (nilpotent) algebra A°V,

wk
exp w i

..
II
k>0

For any Euclidean space (V, g) we denote End_ (V) End- (v, 9) the space of skew-
symmetric linear maps V -> V a natural isomorphism
End_ (VI -> Arv*, I; End_(V) 9 A »-> WA 6 Arv,
(2.2.l0)
wA(v1,v2) = g(Av1,v2), 'v'u1,v2 € V.
If 61 7 , e n ) is an orthonormal basis of V then
l 1
WA=
¢<j
QIA6i, € j ) € ¢ /\ € j
52
2,9
glA€11, € j > € i /\ (in
2
i<j
0»ij€z /\ (in 7

where 0l1 are the entries of the N X N matrix describing the endomorphism A in the basis
(Q),
Cbij 9(A€j»€il = 9(€¢»A6jl-
Given an orientation or on (V, 9) we define the pfaffan of A 6 End_ (V) by
p f (A) = P .f9'" (A) := p f (wA)-
Example 2.2.63. Let V = R2 denote the standard Euclidean space oriented by 61 /\ 82,

where 8182 denotes the standard basis. If

r

A 7

then WA = 961 A et so that P f ( A ) = 9.

Exercise 2.2.64. Let A : V -> V be a skew-symmetric endomorphism of an oriented

Euclidean space V. Prove that palA)2 det A.

Exercise 2.2.65. Let (V, g, or) be an oriented Euclidean space of dimension 271. Consider
A I V > V a skewsymmetric endomorphism and a positively oriented orthonormal frame
e 1 7 ° ° ' 7 € 2 ' N , » Prove that

(-l)"
p .kg"°lAl Znn!
€(07lG»0(1)0(2
) ao(2n-1)o(2n)
0652n

( 1)" €(O)a0'l1)0'(2l U»0(2n-1)0-(Qn) 7

ug8§n
where all = glen 7 Adj) is the "i, j)-th entry in the matrix representing A in the basis (€)»
and SQTL denotes the set of permutations O' Q S211 satisfying
$7(2k - 1) < min{07(2k), 07(2k -|- l)}, We.
62 Lectures on the Geometry of Manifolds

Let (V, g, or) be an n-dimensional, oriented, real Euclidean vector space. The metric
duality Lg : V -> V* induces both a metric, and an orientation on V*. In the sequel we
will continue to use the same notation Le to denote the metric duality 7§ (V) -> in; (V*) 2
in;(V),
Definition 2.2.66. Suppose (VZ 9, or) oriented Euclidean space, and T' is a nonnegative
integer, T < dim V. The 7°-th Hodge pairing is the pairing
I-I
ATIV* X A""IV -> R,
defined by
':'(u)TI7,rl'f'l,-T'l
Detg'r'(Lg 1wT' nn-r l, wT' Q A"V*, ,nn-7° Q A"'°°V

Exercise 2.2.67. Prove that the Hodge pairing is a duality.

Definition 2.2.68. The Hodge *-operator is the adjunction isomoiphism

* TV* _> AN-TV* .

induced by the Hodge duality.

The above definition obscures the meaning of the *-operator. We want to spend some
time clarifying its significance.
Let a G AT'V* so that * Q G A""""V*. Denote by ( 7 ) the standard pairing ..
A"-*v* x A"-*v -> R,
..
and by ( , ) the induced metric on A""IV*. Then, by definition, for every 5
the operator * satisfies
6 An-sv*

Detg(Lg 1 a A L g 1 B ) < *0¢, £g 1p> (*a,B), V6€A""IV* (2.2.11)

Let to denote the unit vector in det V defining the orientation. Then (2.2.l 1) can be rewrit-
ten as

la /\ !3,(»J) = (*@,!5'l, WE 6 AH-TV*

Thus
a A /3 = (w, B) Deaf, Va 6 A V * 7 WE Q A"-*v* (2.2.12)

Equality (2.2.l2) uniquely determines the action of *.

Example 2.2.69. Let V be the standard Euclidean space R3 with standard basis €1,€2,€3,
and orientation determined by el A 62 A et. Then
*€1 = 62 /\ 63, *€2 = 63 /\ 617 *€3 = 61 /\ 627

*1=€1A€2A€3, *l€1/\€2/\€3)=l,

* 62/\€3 el? * 63 €1 627 * 61 62 63.

 Natural Constructions on Manifolds 63

The following result is left to the reader as an exercise.

Proposition 2.2.70. Suppose (V, g, o r ) is an oriented, real Euclidean space of dimension

n. Then the associated Hodge *-operator satisjies

*I = <-1>p('~-pM Vi Q APP* 7

Detg(*ll = 1,

and

Of A *B = (We) * 1 , Va e A'*v*, we € An-kv*

Exercise 2.2.71. Let (V»9, 6) be an n-dimensional, oriented, Euclidean space. For every
t > 0 Denote by it the rescaled metric it = t2g. If * is the Hodge operator corresponding
to the metric g and orientation or, and *t is the Hodge operator corresponding to the metric
it and the same orientation, show that

Dettg = t" Detg 7

and

*to = tn-2p *w Vi G ApV*

We conclude this subsection with a brief discussion of densities.

Definition 2.2.72. Let V be a real vector space. For any T' > 0 we define an 1°-density to
be a function f : det V -> R such that

f(A~) = Il.°f(u), Vu Q derV \ {0}, vA ¢ 0.

The linear space of r-densities on V will be denoted by IAIV. When T' 1 we set
IAlv := III,, and we will refer to 1-densities simply as densities.
Example 2.2.73. Any Euclidean metric g on V defines a canonical l-density | Detg | E
no which associated to each w € d e t V its length, Iw I g.

Observe that an orientation or Q O'r(V) defines a natural linear isomorphism

to, : derV* -> IAlv derV* 9 M »-> u E IAIv, (2.2.13)

where

z0,..u,(w) or(w),u(w), Vi 6 det V \ {0}.

In particular, an orientation in a Euclidean vector space canonically identifies IAlv with R
64 Lectures on the Geometry of Manifolds

2.2.5 Some complex linear algebra

It is convenient to have a coordinate description of the abstract objects introduced above.

Let V be an n-dimensional complex vector space and h a Hermitian metric on it. Pick an
unitary basis 61, ..., iiTL of V, i.e., n = d i n g V, and h(ei, co) = i i . For each j, we denote
by f j the vector i e j . Then the collection e l , fl, . . . , en, fn is an R-basis of V. Denote by
e1, f1, . . . , e", fn the dual R-basis in V*. Then
Re h ( i i » € j l 60 R e d ( f , f j ) and Re h l € z » f j l = 0 ,

Re h Le ®€'i
-I-f (X) fz)-
ii

Also

Lull¢,fjl I -lm h(€i,'i€jl = (So 7 Wlezaejlz 'Jlffjl 0 WJ,

which shows that

CU Imh et /\ f'L.

Any complex space V can be also thought of as a real vector space. The multiplication
by ii = v -1 defines a real linear operator which we denote by J. Obviously J satisfies
$2 I -iv.
Conversely, if V is a real vector space then any real operator J V -> V as above
defines a complex structure on V by
( a + b ' £ ) ' v = a v + b J v , VuGV, a + b i € l C .

We will call an operator J as above a complex structure.

Let V be a real vector space with a complex structure J on it. The operator J has no
eigenvectors on V. The natural extension of J to the complexification of V, Vc = V ® (C,
has two eigenvalues ii, and we have a splitting of Vi as a direct sum o f complex vector
spaces (eigenspaces)

Va: = her (J - i) ® (her J -|- i).

Exercise 2.2.74. Prove that we have the following isomorphisms of complex vector spaces
V&"ker(J-fi) V%lkerJ-I-i).

Set

V1,0 :z Ker(J-fi) V V0,1 == Ker(J+i) V.

Thus V(: 2 V.'1,0 ® V0,1 2 V® V. We obtain an isomorphism of Z-graded complex
vector spaces

A°V(C f"*J A.Vl,0 (X) Ai/0,1

 Natural Constructions on Manifolds 65

If we set
Aznqy :I ApV1 ,0 AqV0 ,1
®(F: 7

then the above isomoiphism can be reformulated as

Am; 2 AMS. (2.2.14)
p-l-qzk

Note that the complex structure J on V induces by duality a complex structure J* on v; 7

and we have an isomorphism of complex vector spaces

v; (v, J); 2 ( w ) .
We can define similarly AWIV* as the Aw?-construction applied to the real vector space
v: equipped with the complex structure J*. Note that
A1,0V*
~_ (A1,0V)2»
A0,1V* g (A071V)2»

and, more generally

Ap»<1v* 5 (Ap»'1v);

If h is a Hermitian metric on the complex vector space (v, JI, then we have a natural
isomorphism of complex vector spaces
vi* iv
(V~*» J*l c (V, -J) go V,
SO that

A'°'qV* go A<1»pV.

The Euclidean metric g = Re h, and the associated 2-form w = -lm h are related by
g ( u , v l = w ( u , Jo), w(u,ul = g ( J u , v ) , Vu,v Q V. (2.2.l5)

Moreover, Ld is a (l, l)-form. To see this it suffices to pick a unitary basis (60 of V, and
construct as usual the associated real orthonormal basis {et, fl, - - ,am fn} (f = Jo)» -
Denote by { @ ¢ , f ; = l, . . . , n } the dual orthonormal basis in W, Then J*e'i = _ f
and if we set
et
/ § ( e + » i f " ) , 5.7 (@3 -'if3l>

then

A 0V* = span@{5} AO,1 = span@{€i},

and

w ii &i AS
66 Lectures on the Geometry of Manifolds

Let V be a complex vector space, and e l , . . . , eTL be a basis of V over (C. This is not a real
basis of V since dim]R V = 2 dim@ V. We can however complete this to a real basis. More
precisely, the vectors 61, ie1, . . . , en, 'Len form a real basis of V.

Proposition 2.2.75. Suppose (el, . . . , en) and (fl, . . . , fTL) are two complex bases of V,
and Z = (2)1§j,k is the complex matrix describing the transition from the basis e to the
basis f, i.e.,

fk €3, vi < lc S n.
j

Then

flAfzflA---Af,,Aif,,=ldetzl2elAfzelA-»-AenAi@,,.

Proof. We write
.
z jk _- : z : kj + z y kJ, : zj : k , yj k € R , e - z e ,
A .
fk-zfk.
A n

Then

fk
J
(ii + '5?/i)€3 2J tie; + 2J yiéw
and

f 2 Yk€.7 + 2 ccke3.
j
j uJ

j
A.

l_lln('n-1)/29
Then, if we set in we deduce

f l / \ i f 1 / \ ° ' ° / \ f n / \ i f n = € n f 1 /\ AfnAf1 A Asn

e,,(det Z)e1 A A€nAé1 A A é'n = (det 2)€1 A 'i€1 A A en A Ian,

where 2 is the 277, X 2n real matrix

Re Z IWLZ X
(N
II

7
ImZ ReZ Y
and X, Y denote the n i n real matrices with entries (to) and respectively (yi)-
We want to show that
set? = ldetZI2 7 Hz 6 End@((c"). (2.2.16)

Let

.A {z 6 End@;(lC"); ldetZI2 det2}.

II
..

We will prove that A = End@;(lC").

The set A is nonempty since it contains all the diagonal matrices. Clearly A is a closed
subset of End@((C"), so it suffices to show that A is dense in End@(@").
 Natural Constructions on Manifolds 67

Observe that the correspondence

End©((c") 9 Z -> 2 Q ENdR(1R()
is an endomorphism of R-algebras. Then, for every Z 6 End@(@"), and any complex
linear automorphism T Q AL1138 C(@"), we have

TZT-1 = T22/-1.
Hence
Z E .A -< >» TZT-1 6 A, VT E Aut@(cnt).
In other words, if a complex matrix Z satisfies (2.2.16), then so will any of its conjugates.
In particular, A contains all the diagonalizable n X n complex matrices, and these form a
dense subset of End@((C") (see Exercise 2.2.76).

Exercise 2.2.76. Prove that the set of diagonalizable n x n complex matrices form a dense
subset of the vector space End@(@"'l.

Definition 2.2.77. The canonical orientation of a complex vector space V, d i g V = n,

is the orientation defined by 6 1 A i € 1 A A en A 'Ion Q Aye, where {et, . . . , e,,} is any
complex basis of V .

Suppose h is a Hermitian metric on the complex vector space V. Then g = Re h

defines is real, Euclidean metric on V regarded as a real vector space. The canonical
orientation orc on V, and the metric g define a volume form Det0"°g
6 Apy6~* n 9

dime; V, and a pffafian

1
Pfd pfh'rc : Arv* -> R, AM* 9 n »-> g(v"7D€tg).

Ifs lm h is real 2-form associated with the Hermitian metric h, then

l
Det Dotg'rc = -w"
9 in! 7

and we conclude

pf hw l.

2.3 Tensor Iields

2.3.1 Operations with vector bundles

We now return to geometry, and more specifically, to vector bundles.

Let K denote one of the fields R or (C, and let E -> M be a rank T' K-vector bundle
over the smooth manifold M . According to the definition of a vector bundle, we can find
an open cover (Url of M such that each restriction E I UQ is trivial: E | Up, V X Up, f"*J

where V is an 1°-dimensional vector space over the field K. The bundle E is obtained by
68 Lectures on the Geometry of Manifolds

gluing these trivial pieces on the overlaps Ua re UB using a collection of transition maps
QQB : UO( O UB -> GL(V) satisfying the cocycle condition.
Conversely, a collection of gluing maps as above satisfying the cocycle condition
uniquely defines a vector bundle. In the sequel, we will exclusively think of vector bundles
in terms of gluing cocycles.
Let E, F be two vector bundles over the smooth manifold M with standard fibers VE
and respectively Vp, given by a (common) open cover (Ua), and gluing cocycles

QQB : Uog -> GL(VE ), and respectively hog : UQ5 -> GLWFI-

Then the collections

QQB ® hog : UQ5 -> GL(VE ® Vp), QQB (X) hog : Uog -> GL(VE (X) Vp),

(ITB) :U -> GLWE >, ATM : u -> GL(A'°vE),

where T denotes the transpose of a linear map, satisfy the cocycle condition, and therefore
define vector bundles which we denote by E ® F, E (X) F, E*, and respectively A"IE. In
particular, if T' = ranklE, the bundle AT'E has rank 1. It is called the determinant line
bundle of E, and it is denoted by det E. Given the adjunction isomorphism VE (X) VF E
HOM(VE, Vp), we set

Hom(E, F) := E* (X) F.

We set

End(E) := Hom(E,E).

The reader can check easily that these vector bundles are independent of the choices of
transition maps used to characterize E and F (use Exercise 2.l.34). The bundle E* is
called the dual of the vector bundle E. The direct sum E ® F is also called the Whitney
sum of vector bundles. All the factorial constructions on vector spaces discussed in the
previous section have a vector bundle correspondent. (Observe that a vector space can be
thought of as a vector bundle over a point.)
These above constructions are natural in the following sense. Let E' and F' be vector
bundles over the same smooth manifold M/. Any bundle maps S : E -> E' and T : F ->
F' , both covering the same dweomorphism go : M -> M', induce bundle morphisms

S ® T : E ® F->E' ® T/ 7 S (X) T:E (X) F->E' (X) FI 7

covering (25, a morphism

s* : (E')* -> E* 7

covering (a 1
etc.

Exercise 2.3.1. Prove the assertion above.

 Natural Constructions on Manifolds 69

Example 2.3.2. Let E, F, E' and F' be vector bundles over a smooth manifold M. Con-
sider bundle isomorphisms S : E -> E' and T : F -> F' covering the same diffeomor-
phism of the base, <l> : M -> M. Then (S-1)t : E* -> (EI)* is a bundle isomorphism
covering go, so that we get an induced map (5-1)'r (X) T : E* (X) F -> (E/)* (X) F'. Note that
we have a natural identification

COO( Hom(E, F ) ) = c°°(E* (X) F) E Hom(E, F),

where we recall that Hom(E, F) denotes the space of smooth bundle moiphisms E ->F.

Definition 2.3.3. Let E -> M be a K-vector bundle over M . A metric on E is a section h

OfE* ®1I< E* (F: Eif11< = R) such that, for any m Q M, h(m) defines a metric on Et
(Euclidean if K = R or Hermitian if K G).

2.3.2 Tensor jields

We now specialize the previous considerations to the special situation when E is the tangent
bundle of M , E 5 T M . The cotangent bundle is then

T*M (TM)*.
We define the tensor bundles of M

<T;°(m) : =t;(Tm) (TM)®" (X) (T*M)®s.

Definition 2.3.4. (a) A tensorjield of type (T, s) over the open set U C M is a section of
'J';l(M) over U.
(b) A degree 7" dqferentialform (r-form for brevity) is a section of A"(T*M). The space
of (smooth) 1°-forms over M is denoted by QT(M). We set
QUO
mm).
..

(M)
II

'r>0

(c) A Riemannian metric on a manifold M is a metric on the tangent bundle. More pre-
cisely, it is a symmetric (0, 2)-tensor g, such that for every 11: E M, the bilinear map

go:TasMxTasM->H

defines a Euclidean metric on TacxM.

If we view the tangent bundle as a smooth family of vector spaces, then a tensor field
can be viewed as a smooth selection of a tensor in each of the tangent spaces. In particular,
a Riemann metric defines a smoothly varying procedure of measuring lengths of vectors in
tangent spaces

Example 2.3.5. It is often very useful to have a local description of these objects. If
txa't . . ,a:")6' are local coordinates on an open set U C M, then the vector fields
( 3:17 7 311271 ) trivialize TM IN, i.e., they define a framing of the restriction of T M to
3
70 Lectures on the Geometry of Manifolds

U. We can form a dual framing of T*M In, using the 1-formsdx, i 1, ,n. They
satisfy the duality conditions

(do'L, 3a 3> = 537 'v'i,j.

A basis in 'J'.§(T,,M) is given by

w (X) (X) a, ® do ® (X) dxjs, 1 S 7317 7 i , ~ < n , 1 <j1> 7 j.S S n

Hence, any tensor T 6 7 § ( M ) has a local description

.
T Tel---j; ex11 (X) (X) 33,,,. (X) data (X) (X) dzvjs

In the above equality we have used Einstein's convention. In particular, an 1°-form CU has
the local description

w= Ldi1...i d.CU11 /\ Ad;z:1*, Wil...¢,. Wlaxll 7 7 a 7

1$¢1<---<»z,.5n

while a Riemann metric g has the local description

.
.~d:z:z
® drvj) gM
~.=g( ax." 6 a)
x °
DO
II

92]
z ,j
=g.7?

Remark 2.3.6. (a) A covariant tensor field, i.e., a (0, 5)-tensor field S, naturally defines a
cOO (M)-multilinear map

.S

S •
. Vect (M) -> C'°°(M),
1

(Xl: ,xs) (p sp(x1(p), »X s(p )) e c°°(m>.

Conversely, any such map uniquely defines a (0, 5)-tensor field. In particular, an 1°-form 77
can be identified with a skew-symmetric cOO (M)-multilinear map
T'

17 Vect (M) -> C°°(M).

1

Notice that the wedge product in the exterior algebras induces an associative product in
Qc (M) which we continue to denote by /\.

(b) Let f E C'°°(M). Its Fréchet derivative Df : TM -> TR 2 R X IR is naturally a

1-form. Indeed, we get a smooth cOO (M)-linear map df : Vect (M) -> cOO (M) defined
by

of(x)m .
a
Df (Xlf('rn) E Tf(m)R E R.

In the sequel we will always regard the differential of a smooth function f as a 1-form and
to indicate this we will use the notation df (instead of the usual Df).
 Natural Constructions on Manifolds 71

Any diffeomorphism f : M -> N induces bundle isomorphisms Df : TM -> TN

and lpf-1)'r : T * M -> T*N covering f. Thus, a diffeomorphism f induces a linear map

f* :UCI(Ml->UZI(N)V (2.3.1)
called the push-forward map. In particular, the group of diffeomorphisms of M acts natu-
rally (and linearly) on the space of tensor fields on M .

Example 2.3.7. Suppose f : M -> N is a diffeomolphism, and S is a (0, k)-tensor field

on M, which we regard as a cOO (M)-multilinear map
S : Vect (M) X X Vect (M) -> C°°(M).
k

Then f*s is a (0, is) tensor field on N . Let q 6 N , and set p f-l(q)~ Then, fOrany

..II
Y1 7 . 7 Y1K7 Q TqN, we have

l*f*SlqlYl 7 ,Ye) so (f* $y1, (A)-1yk)

• • • 7

So ((D,»f)-1}/1, - - 9 (Dif)-ln) -
For covariant tensor fields a more general result is true. More precisely, any smooth
map f : M -> N defines a linear map

f* : <rQlnl -w/~(M),
called the pullback by f . Explicitly, if S is such a tensor defined by a C'°°(M)-multilinear
map
S : (Vect ( N W -> o°°<n>,
then f*S is the covariant tensor field defined by

(f*S)p(X1(p)»---»Xs(p)) I= S.f(p)(Dpf(X1)»---»Dpf(Xsll»
'v'XI, . . . ,Xs G Vect ( M l , p G M. Note that when f is a dweomorphism we have

f* = (f* 1lT,
where f* is the push-forward map defined in (2.3.1).
Example 2.3.8. Consider the map
F* : (0,ool X (0,27r) -> R2, (KH) »-> (LU = rcos9,y = rsinH).

The map F defines the usual polar coordinates. It is a diffeomorphism onto the open subset

\ u , 0); ;U> 0}-

For simplicity, we will write 3 ax instead of -a 3
31° ' ex etc. We have
F*d;z: = d(1° cos 9) = cos cdr - rsin9d9, F*dy = d(r sin 9) = sin9d1° -|- 7° cos 9d9,

F*(da: /\ dy) = d(1° cos 9) /\ d(rsin9) = (cos 9dr - rsin9d9) /\ (sin9d1° -|- 1° cos 9d9)

= r(cos2 9 -|- sin29)dr /\ d9 = pdf /\ d9.

72 Lectures on the Geometry of Manifolds

To compute F* <1 and F* 30 we use the chain rule which implies

3:13 by l
F* an <9r
-I-
31° y
cos 933 -|- sin 9<9y
7°
(33339 -|- y3y)
1
(112 + y2)1/2 (33333 + Wi'/>'

The Euclidean metric is described over U by the symmetric (0, 2)-tensor g = d;u2 -|- dg/2
The pullback of g by F is the symmetric (0, 2)-tensor
2 2
F*(dx2 + dye) = (d cos 9>) -|- ld(r sin 9) )
= (cos Hdr - 1°sin9d9)2 -|- (sin9d1° -|- T' cos 9d9)2 = do,.2 -|- M92.
To compute F* dr we need to express T' as a function of x and y, T' I I _|_ y2l 1/2 and
then we have
F*dr = (F"1)*d1° = d(a32 + y2)1/2 = :u(x2 y/)-1/2dzv ylQ32 y2)-1/2dy.

All the operations discussed in the previous section have natural extensions to tensor fields.
There exists a tensor multiplication, a Riemann metric defines a duality L : Vect (M) ->
Q1 (M) etc. In particular, there exists a contraction operator
tr : <1';iI(m) -> <I;(m)
defined by
tr(x0®---®x,.)®(w0®- (X) Ws) =WolXol(X1 (X) . . . X (X)
W1
®°°°®WSl,
\V/Xi G Vect lM),\VWj G In local coordinates the contraction has the form
lT'L0...Z»f»
{nr 30.../)} = }-
Let us observe that a Riemann metric g on a manifold M induces metrics in all the associ-
ated tensor bundles 7 § ( M ) , If we choose local coordinates (so) on an open set U then, as
explained above, the metric g can be described as
g Z
gzj day (X) dccj 7
Z

M
while a tensor field T of type (T, s) can be described as
T T*1~~'»~a ®®da;°J 1 ( X ) ® dgjjs
® aas
- .71~~~js or

If we denote by (gin) the inverse of the matrix l9zjl7 then, for every point p 6 U, the length
of T ( p ) 6 u§(M)p is the number IT(p)lg defined by
ITlPllg = milk. ---91,-k,~93.
'e
. -,go
•
'ses . ...',
T¢:...jT'°'°
r
7

where in the above equalities we have used Einstein's convention.

The exterior product defines an exterior product on the space of smooth differential
forms
/\ : Q' ( M ) X Q' (M)- Q' (M)-
The space Q' ( M ) is then an associative algebra with respect to the operations -|- and A.

Proposition 2.3.9. Let f : M -> N be a smooth map. The pullback by f defines a

morphism of associative algebras f* : Q' ( N ) -> Q' ( M ) .

Exercise 2.3.10. Prove the above proposition.

 Natural Constructions on Manifolds 73

2.3.3 Fiber bundles

We consider useful at this point to bring up the notion of fiber bundle. There are several
reasons to do this.
On one hand, they arise naturally in geometry, and they impose themselves as worth
studying. On the other hand, they provide a very elegant and concise language to describe
many phenomena in geometry.
We have already met examples of fiber bundles when we discussed vector bundles.
These were "smooth families of vector spaces". A fiber bundle wants to be a smooth
family of copies of the same manifold. This is a very loose description, but it offers a first
glimpse at the notion about to be discussed.
The model situation is that of direct product X = F X B, where B and F are smooth
manifolds. It is convenient to regard this as a family of manifolds (Fb)beB. The manifold
B is called the base, F is called the standard (model) fiber, and X is called the total space.
This is an example of trivial fiber bundle.
In general, a fiber bundle is obtained by gluing a bunch of trivial ones according to a
prescribed rule. The gluing may encode a symmetry of the fiber, and we would like to
spend some time explaining what do we mean by symmetry.
Definition 2.3.11. (a) Let M be a smooth manifold, and G a Lie group. We say the group
G acts on M from the left (respectively right), if there exists a smooth map
< I > : G x M - > M , (g,m)»-><I>9m,
such that <I>1 E JIm and
®g(®;,m) = <I>gI,m (respectively <I>g(i>I,m) = <I>I19m) Vg, h 6 G, m 6 M.
In particular, we deduce that Vg 6 G the map <I>g is a diffeomorphism of M . For any
m 6 M the set
G-m={©gm; g a G }
is called the orbit of the action through m.
(b) Let G act on M. The action is called free if Vg 6 G \ {1}, and 'v'm E M Tgm 7; m.
The action is called ejective if, Vg 6 G \ {1}, ®9 7; JIm.

It is useful to think of a Lie group action on a manifold as encoding a symmetry of that

manifold.
Example 2.3.12. Consider the unit 3-dimensional sphere
S2 = {(x,y,z) 6 R3; $1:2 +3/2 + 2 2 = l}.
Then the counterclockwise rotations about the z-axis define a smooth left action of S1 on
S2. More formally, if we use cylindrical coordinates (T, 9, 2)
Hz: = TCOSH, y = r s i n 9 , z=0,
then for every go 6 IR mod 27r 2 S1 we define RS0 : S2 -> S2 by
RSOITQQQZ) = (r, (9 -|- (pl mod Qtr, z).
The resulting map R : S1 X S2 -> S2, (go, p) +-> R(plp) defines a left action of S1 on S2
encoding the rotational symmetry of S2 about the z-axis.
74 Lectures on the Geometry of Manifolds

Example 2.3.13. Let G be a Lie group. A linear representation of G on a vector space V

is a left action of G on V such that each Tg is a linear map. One says V is a G-module. For
example, the tautological linear action of SO(n) on R" defines a linear representation of
SO(n).

Example 2.3.14. Let G be a Lie group. For any g 6 G denote by Lg (resp. Rg) the left
(resp. right) translation by g. In this way we get the tautological left (resp. right) action of
G on itself.

Definition 2.3.15. A smoothjber bundle is an object composed of the following:

(a) a smooth manifold E called the total space,

(b) a smooth manifold F called the standard fiber,
(c) a smooth manifold B called the base,
(d) a surjective submersion 'IT : E -> B called the natural projection,
(e) a collection of local trivializations, i.e., an open cover (I/1a) of the base B, and diffeo-
morphisms \1/0, : F X Ua -> 7r"1(Uo) such that
7T oWe v be : : b, */(f, be 6 F X UQ,
i.e., the diagram below is commutative.
'PQ
F XU0 '7T 1 U/0J

UQ
We can form the transition (gluing) maps \I/0 . F x Uag - > F x Ua ,where l/'ag =
UQ O UP, defined by Wag = upa 1 o W . According to (e), these maps can be written as
1 b a 0 ( f » be = (Toflilblf, be,

where T<.l3 lbl is a diffeomorphism of F depending smoothly upon b G U0g.

We will denote this fiber bundle by (E, or, F, Be.

If G is a Lie group, then the bundle ( E , Tr, F, B) is called a G-fiber bundle if it satisfies
the following additional conditions.

(D There exists an effective left action of the Lie group G on F,

GF 9 l9»;z:)+->g- QUO TQQCGF.
X

The group G is called the symmetry group of the bundle.

(g) There exist smooth maps gag : Uog -> G satisfying the cocycle condition
.gas I
1 E G, .Eva I .975 • QBQ) V@,B,v,
and such that
TQBUPI = T9o¢=(b) -
We will denote a G-fiber bundle by ( E , or, F, B, G).
 Natural Constructions on Manifolds 75

The choice of an open cover (Ua) in the above definition is a source of arbitrariness
since there is no natural prescription on how to perform this choice. We need to describe
when two such choices are equivalent.
Two open covers (Url and (W), together with the collections of local trivializations
ea . F x u -> for-HUa) and w :F x v -> w-1(v;)
are said to be equivalent if, for all Q, 11, there exists a smooth map
Tai . UQ VW * G,
such that, for any m e UmM/Q, and any f Q F , we have

<1>0}@i(f,x) Z (Ta¢(93lf»

A G-bundle structure is defined by an equivalence class of trivializing covers.

As in the case of vector bundles, a collection of gluing data determines a G-fiber bundle.
Indeed, if we are given a cover (U0,)0,€A of the base B, and a collection of transition maps
QQB : Ua n UB -> G satisfying the cocycle condition, then we can get a bundle by gluing
the trivial pieces Uo, X F along the overlaps.
More precisely, if b 6 Uamltg, then the element (f, IQIQFXUa is identified with the
element (g5O,(bl • f , be G F X UB-
7T
Definition 2.3.16. Let E -> B be a G-fiber bundle. A G-automorphism of this bundle is a
diffeomorphism T : E -> E such that or o T = or, i.e., T maps fibers to fibers, and for any
trivializing cover (I/'ol (as in Definition 2.3.15) there exists a smooth map go, : Uo, -> G
such that

Q 1Tq/aU, be (9o (6) f,b ), w>,f.

Definition 2.3.17. (a) A fiber bundle is an object defined by conditions (a)-(d) and (f) in
the above definition. (One can think the structure group is the group of diffeomorphisms of
the standard fiber).
7T
(b) A section of a fiber bundle E -> B is a smooth map S : B -> E such that 7T O S ]DB»
i.e., 5(5) G -1(6>, v i G B.

Example 2.3.18. A rank T' vector bundle (over K = R, (I) is a GL(v", K) -fiber bundle with
standard fiber KT, and where the group GL(r, K) acts on KT' in the natural way.

Example 2.3.19. Let G be a Lie group. A principal G-bundle is a G-fiber bundle with
fiber G, where G acts on itself by left translations. Equivalently, a principal G-bundle over
a smooth manifold M can be described by an open cover 'Ll of M and a G-cocycle, i.e., a
collection of smooth maps
_ Q u v : U M V - G U,Ve'LL,
such that V x 6 UHVV1W(U,V,W€"L()

9uv(33l9vw($l9wu(22l = 1 6 G.
76 Lectures on the Geometry of Manifolds

Exercise 2.3.20 (Alternate definition of a principal bundle). Let P be a fiber bundle

with fiber a Lie group G. Prove the following are equivalent.
(a) P is a principal G-bundle.
(b) There exists a free, right action of G on G,

P X G - > P , (19,9)+->19~9,
such that its orbits coincide with the fibers of the bundle P, and there exists a trivializing
cover
{w@ : G x Ua ->7r-1(Ua) } 7

such that

\I/ 0(h9/'~I,)= *I'0(h>u)'9» V g , h € G , 'u,6 U0.

Exercise 2.3.21 (The frame bundle of a manifold). Let M n be a smooth manifold. De-
note by F(M) the set of frames on M, i.e.,

F ( M ) = {(m,X1, ,Xnl, m 6 M, Xi 6 TmM and span(Xl, ,X,,) = T m M } .

(a) Prove that F(M) can be naturally organized as a smooth manifold such that the natural
projections : F ( M ) -> M, (m, Xi, ,X,,) »-> m is a submersion.
(b) Show F ( M ) is a principal GL(n, R)-bundle. The bundle F ( M ) is called the frame
bundle of the manifold M.
Hint: A matrix T (To 6 GL(n, K) acts on the right on F ( M ) by

~17 ...,x,,) »-> (my lT-1l1Xi9 (T'1)X¢).

Example 2.3.22 (Associated fiber bundles). Let 77 : P -> G be a principal G-bundle.

Consider a trivializing cover lU0la€A9 and denote by gag : Up Fl UB -> G a collection of
gluing maps determined by this cover. Assume G acts (on the left) on a smooth manifold F

X F - > F , (Q, f ) »-> Tlglf-

The collection Tag = 'T(90@) : Uog -> Diffeo (F) satisfies the cocycle condition and can
be used (exactly as we did for vector bundles) to define a G-fiber bundle with fiber F. This
new bundle is independent of the various choices made (cover (UQ) and transition maps
QQBI. (Prove this!) It is called the bundle associated to P via T and is denoted by P XT F.

Exercise 2.3.23. Prove that the tangent bundle of a manifold M n is associated to F(m>
via the natural action of GL(n, R) on R"

Exercise 2.3.24 (The Hopf bundle). If we identify the unit odd dimensional sphere
S271-1 with the submanifold

217 7 Zn) E (Zn l20l2 -I- + IonI 2 w

 Natural Constructions on Manifolds 77

then we detect an S1-action on S2n-1 given by

et - (z1, zn) = (eWzl, eweTL).
The space of orbits of this action is naturally identified with the complex projective space
CP"l1.
(a) Prove that p : S272-1 -> <clpn-1 is a principal S1 bundle called Hopf bundle. (p is the
obvious projection.) Describe one collection of transition maps.
(b) Prove that the tautological line bundle over (CIPn-1 is associated to the Hopf bundle via
the natural action of Sl on (Cl .

Exercise 2.3.25. Let E be a vector bundle over the smooth manifold M. Any metric h on
E (euclidian or Hermitian) defines a submanifold S(E) C E by

S(E) = {v 6 E; IUlh l }.
Prove that S (E) is a libration over M with standard fiber a sphere of dimension rank E - 1.
The bundle S(E) is usually called the sphere bundle o f E.
This page intentionally left blank
 Chapter 3

Calculus on Manifolds

This chapter describes the "kitchen" of differential geometry. We will discuss how one can
operate with the various objects we have introduced so far. In particular, we will introduce
several derivations of the various algebras of tensor fields, and we will also present the
inverse operation of integration.

3.1 The Lie derivative

3.1.1 Flows on manifolds

The notion of flow should be familiar to anyone who has had a course in ordinary differ-
ential equations. In this section we only want to describe some classical analytic facts in a
geometric light. We strongly recommend [6, 12] for more details, and excellent examples.
A neighborhood N of {0} X M in R X M is called balanced if, Vm 6 M, there exists
r 6 (0, oo] such that
(R X {m} ) m N = (-m~) X {m}.
Note that any continuous function f : M -> (0, oo) defines a balanced open
Nf := { ( t , m ) 6 R X M; Itl < f(m) }~
Definition 3.1.1. A loealflow is a smooth map Q : N -> M, (t, m) +-> <1>t(m), where N is
a balanced neighborhood of {0} X M in R X M, such that

(a) q>0(m) = m, Vm 6 M.
(b) q>t(<1>s(ml) 1 @+s(m) for all s , t 6 R m 6 M such that
( s , m l , (s -I- t , m ) , l¢,q>S(mll 6 N.

When N R XM , Q? is called allow.

The conditions (a) and (b) above show that a flow is nothing but a left action of the
additive (Lie) group (R, +) on M .
Example 3.1.2. Let A be a e x n real matrix. It generates a flow <I>iA on R" by
to

(2 )
t
i>Aa3 e t A as -Ak ac.
k=0 k!

79
80 Lectures on the Geometry of Manifolds

Definition 3.1.3. Let Q :N-> M be a local flow on M . The infinitesimal generator of Q

is the vector field X on M defined by
d
X(p) = X<I>(P) llllllllll

dt <1>t(p), V P G M ,

i.e., X (p) is the tangent vector to the smooth path t »-> <I>t(p) at t = 0. This path is called
the flow line through p.

Exercise 3.1.4. Show that X<I> is a smooth vector field.

Example 3.1.5. Consider the flow €¢A on R" generated by a n X n matrix A. Its generator
is the vector field XA on R" defined by
d
XAIU/I dt
e
tA
u Au.
t O

Proposition 3.1.6. Let M be a smooth n-dimensional manzfold. The map

X : {Local flows on M} -> Vect (M), i> »-> X<I>,

is a surjection. Moreover; If @¢ : N -> M (i=],2) are two loealflows such that Xi1
X<I>2, then <I>1 = CD2 on N1 D No.

Proof. Surjeetivily. Let X be a vector field on M . An integral curve for X is a smooth

curve by : (a, b) -> M such that

WI Z X(7(t))-
In local coordinates ( f ) over on open subset U C M this condition can be rewritten as
Wt) =x(:¢1(t),...,x"(z)), vi 1, ANa (3.1. 1)

where W) = (;r31(tl, ...,x"(t)), and X = Xfi 3:1% a The above equality is a system of
ordinary differential equations. Classical existence results (see e.g. [12, 60]) show that, for
any precompact open subset K C U, there exists € > 0 such that, for all at G K, there
exists a unique integral curve for X , 'Ye : ( - 5 , 5) -> M satisfying
'yas(0l = LU. (3.l.2)

Moreover, as a consequence of the smooth dependence upon initial data we deduce that the
map

@K:3\fK=l-(5) X K ->M, (ac,il*->'yx(tl,

is smooth.
Now we can cover M by open, precompact, local coordinate neighborhoods (KQIQQA 7

and as above, we get smooth maps Qa : Na = (-eyea) X Ka -> M solving the initial
value problem (3. 1.1-2). Moreover, by uniqueness, we deduce
QQ - QQ on No VWN5.
 Calculus on Manifolds 81

Define
N: U No
06.A
7

andseti1'> : N-> M,@= paa onU\fo.

The uniqueness of solutions of initial value problems for ordinary differential equations
implies that i1'> satisfies all the conditions in the definition of a local flow. Tautologically,
X is the infinitesimal generator of i1'>. The second part of the proposition follows from the
uniqueness in initial value problems.

The family of local flows on M with the same infinitesimal generator X 6 Vect(M) is
naturally ordered according to their domains,
lfbl I N1 -> M) 4 l@2:N2->Ml
if and only if N1 C N2. This family has a unique maximal element which is called the
local flow generated by X, and it is denoted by i1'>x .

Exercise 3.1.7. Consider the unit sphere

S2 = {(a2,y,z) e R3; £1:2 +y2 +z2 = 1
For every point p 6 S2 we denote by X (p) e TOR3, the orthogonal projection of the vector
k = (0, 0, 1) onto TpS2.
(a) Prove that p »-> X (p) is a smooth vector field on S2, and then describe it in cylindrical
coordinates (z, 0), where
as rcos0, y = T' sin 0 , T'
lx2_l_y2l1/2.

(b) Describe explicitly the flow generated by X .

3.1.2 The Lie derivative

Let X be a vector field on the smooth n-dimensional manifold M and denote by <I> = ®X
the local flow it generates. For simplicity, we assume <I> is actually a flow so its domain is
R x M. The local flow situation is conceptually identical, but notationally more compli-
cated.
For each t 6 R, the map <I>t is a diffeomorphism of M and so it induces a push-forward
map on the space of tensor fields. If S is a tensor field on M we define its Lie derivative
along the direction given by X as
• l
LxS1n hm - ((@i>s)m-sm) V m 6 M. (3.l.3)
t->0 t

Intuitively, LxS measures how fast is the flow ® changing! the tensor S.
If the limit in (3.13) exists, then one sees that LxS is a tensor of the same type as S.
To show that the limit exists, we will provide more explicit descriptions of this operation.
lArnold refers to the Lie derivative Lx as the 'fisherman's derivative". Here is the intuition behind this very
suggestive terminology. We place an observer (fisherman) at a fixed point p E M, and we let him keep track of
the the sizes of the tensor S carried by the flow at the point p. The Lie derivatives measures the rate of change in
these sizes.
82 Lectures on the Geometry of Mamfolds

Lemma 3.1.8. For any X 6 Vect (M) and f 6 C°°(M) we have

Above, ( .. o f :z L x f
(df,x) df(X)- I

) denotes the natural duality between T * M and T M ,

I

..I , : C°°(T*M) X C°°(TM) -> C°°(M),

COO(T*M) X C°°(TM) 9 ( 0 , X ) »-> a ( X ) 6 C°°(M).
In particular Lx is a derivation of cOO ( M l .

Proof. Let Qt = QQ be the local flow generated by X. Assume for simplicity that it is
defined for all t. The map Qt acts on cOO (M) by the pullback of its inverse, i.e.,
<I>* = (<1>-t)*.
Hence, for point p E M we have
. 1 d
Lx f (p) hm - (f(p) - f(<I>-tp)) f(<I>ltp) <df,X>p.
t->0 t
Z
dt t 0

Exercise 3.1.9. Prove that any derivation of the algebra COO(M) is of the form Lx for
some X 6 Vect(M), i.e.,
Der (C'°°(M)) E Veczt (M).

Lemma 3.1.10. Let X, Y 6 Vect (M). Then the Lie derivative of Y along X i5 a new
vectorjield LxY which, viewed as a derivation O f f ° ° ( M ) , coincides with the commutator
of the two derivations of GOO ( M ) defined by X and Y , i.e.,
L x Y f = [x, Yu, o f 6 C>°°(m).
The veetorjield [X, Y ] = LxY is called the Lie bracket o f X and Y . In particular the Lie
bracket induces a Lie algebra structure on Vect (M).

Proof. We will work in local coordinates ( w ) near a point m E M so that

X xa, and Y = Wa .
We first describe the commutator [X, Y1. If f € C°°(M), then
. .of of )
[X, Ylf
. .
(
( XZ3x 71 ) Y .7 33123.
82f
) ( ywxj ) Xi
-<9Yi o f
323'.

. . 92f EX' o f
XzYJ .
exz 3;I:] . -I- Xz .
3;u' (9:6]
.
3a:"<9;v9 ) (
314 811.i
XZYJ

so that the commutator of the two derivations is the derivation defined by the vector field
o . +W 7

X i ask _ ii axk
[X, Y]
31 333]
as. (3.1.4)

Note in particular that [d 3acj] = 0, i.e., the basic vectors 3gg'L commute as derivations.
So far we have not proved the vector field in (3.l.4) is independent of coordinates. We
will achieve this by identifying it with the intrinsically defined vector field L x Y .
 Calculus on Manifolds 83

\i
Set 'y(t) 1 M so that we have a parametrization 'y(tl I (we) with xi_ Then
q>"'m = 7(-t) = 7(0) - 1(0)t + 0(9) = (x0 - t + o(t2)) 7

and
. 35/j
YJ txfi .-l-Ot2. (3.l.5)
Yi(-> 351 ( )
Note that is* t T(0)M -> Tv(-t)M is the linearization of the map
(of) »-> ( - tx + O(t2)) 7

so it has a matrix representation

<I>*t ]1 t 1 331]
. + 0(t2). (3.l.6)
M
In particular, using the geometric series
(11-A) =]l-i-A-i-A2+ 7

where A is a matrix of operator norm strictly less than 1, we deduce that the differential
(Dt = (®* t)-1 : Tv(_t)M -> T"(0)M7

has the matrix form

11 -I-t I 3:z:wj s o
+ 0(t2l. (3.l.7)

Using (3.l.7) in (3.1.5) we deduce

k . ay'< YJ axk
Yk - (@ty-m) t XZ
316z
.
3937
+ 0(t2 l.
This concludes the proof of the lemma.

Lemma 3.1.11. For any dQ§"erential form w E Q1(M), and any vector welds X , Y E
V e c t ( M ) we have
(LxWl(Y) = LX (w(Y)l - (»J(X7Y])7 (3.l.8)
-
where X w ( Y ) denotes the (Lie) derivative of the funetion w ( Y l along the veetorfield X .

Proof. Denote by <I>t the local flow generated by X . We have <I>tw = (<I>-tl*w, i.e., for
any p 6 M, and any Y 6 Vect ( M ) , we have
(<1>%v)p(yp) = °"<I>'tpl Q* top ).
Fix a point p E M, and choose local coordinates ( i l near p. Then
cu = wdx, X Xa Y 7y*'@, . Z
11

Denote by 'y(t) the path t +-> <I>t(p). We set Ldiltl lilYltll' XI Xi(p ), and YS
Yfi(p). Using (3.1.6) we deduce

l@tw)p(Ypl Wil't) YS
84 Lectures on the Geometry of Manifolds

Hence
d exfi
- ( L x ( » ) y = dt
llllllllll

_0 l@tl*'lplYpl @¢(0)Y5 Milo) Yo".

3567
'As
One the other hand, we have

-
X w(Y)
d
w,<t>y%:) C;1il0ly0Z+ lJJZl0)X07
aw
dt t 0 , 3307
M
We deduce that

X8 aw EXfi

J
-
X w(Y) - (LxC,U)Y (Milo)
3561 3337 Y ( )
wp([X»Y]pl-
M
Observe that if S, T are two tensor fields on M such that both LxS and LxT exist,
then using (3.l.3) we deduce that L x ( S ( X ) T) exists, and
Lx(S®T)=LxS®T-I-S (X) LxT. (3.l.9)

Since any tensor field is locally a linear combination of tensor monomials of the form
XI ®~~®X»,~®w1 ®~~®uJ5, Xi 6 V e c t ( M ) , we 6 Q1(M),
we deduce that the Lie derivative LxS exists for every X 6 Vect (M), and any smooth
tensor field S. We can now completely describe the Lie derivative on the algebra of tensor
fields.

Proposition 3.1.12. Let X be a vector ariela' on the smooth manifold M . Then the Lie
derivative Lx is the unique derivation of'T** ( M ) with the following properties.
Ra) L x f = ( d f . x ) = x f , o f 6 c°°lm).
(b) LxY = 1X, Y], 'v'X,Y 6 Vect (M).
(c) Lx commutes with the contraction tr : ( M ) -> 'In(Ml.
Moreover; LX is a natural operation, i.e., for any dQ§'eomorphisrn (15 : M -> N we have
s* O LX = L¢>*X O ¢*, VX 6 Vect (Ml, i.e., QZ5*(LX) = L¢>*X.
Proof. The fact that LX is a derivation follows from (3.1.9). Properties (a) and (b) were
proved above. As for part (c), in its simplest form, when T = Y®w, where Y 6 Vect (M),
and w 6 Q1 ( M ) , the equality
LX trT I tI'LxT
is equivalent to

Lx (wlyll (Lx°/)(Y) + WlLx (YI, (3.1.10)

which is precisely (3.1.8).
Since Lx is a derivation of the algebra of tensor fields, its restriction to COO(M) ®
Vect (M) ® Q1 (M) uniquely determines the action on the entire algebra of tensor fields
which is generated by the above subspace. The reader can check easily that the general
case of property (c) follow from this observation coupled with the product rule (3. 1.9).
 Calculus on Manzfolds 85

The naturality of Lx is another way of phrasing the coordinate independence of this

operation. We leave the reader to fill in the routine details.

Corollary 3.1.13. For any X, Y 6 Vect ( M ) we have

]Lxa Ly] I L]X,Y]7
as derivations of the algebra of tensor fields on M . In particular this says that Vect ( M l
as a space of derivations of :T** (M ) is a Lie subalgebra of the Lie algebra of derivations.

Proof. The commutator [LX, Ly] is a derivation (as a commutator of derivations). By

Lemma 3.1.10, ]LX, Ly] = LlX,Y1 on GOO ( M ) . Also, a simple computation shows that
]Lx,Ly]Z = L]X,Y]Z7 VZ 6 Vect ( M ) ,
so that ]LX, Ly] = LlX,Y1 on Vect (M). Finally, since the contraction commutes with
both Lx and Ly it obviously commutes with L i L y - L y L x . he corollary is proved.

Exercise 3.1.14. Prove that the map

11) : Vect ( M ) ® End ( T M ) -> D@t°(7l'(M))
given by LD(X,S) Lx -|- S is well defined and is a linear isomorphism. Moreover,

[1D(X1»51)»1D(X2»52)] = iD([X1»X2]» [S1»52]l-

The Lie derivative Lx is a derivation of 'TI with the remarkable property
Lx(Q*(Mll c o * ( M ) .
The wedge product makes Q*(M) a s-algebra, and it is natural to ask whether Lx is an
s-derivation with respect to this product.

Proposition 3.1.15. The Lie derivative along a vector field X i5 an even 5-derivation of
Q * ( M > , i.e.,
Q*
Lx((»J /\ UI = (LXUJ) /\ 77 -|- w /\ (Lxv), Vi, 77 6 (M)-

Proof. As in Subsection 2.2.2, denote by A the anti-symmetrization operator A :

iT*ml®k -> m m ) . The statement in the proposition follows immediately from the
straightforward observation that the Lie derivative commutes with this operator (which is a
projector). We leave the reader to fill in the details.

Exercise 3.1.16. Let M be a smooth manifold, and suppose that <I>, \P : R X M -> M are
two smooth flows on M with infinitesimal generators X and respectively Y. We say that
the two flows commute if
s
<1>*'oq/s off t , iv's,t€]R.

Prove that if

{p 6 M; X(p)=0}= { p p M ; Y(p)=0}-
the two flows commute if and only i f [ x , y ] I 0.
86 Lectures on the Geometry of Manifolds

3.1.3 Examples

Example 3.1.17. Let w = Ujida be a 1-form on R". If X X13,,j is a vector field on

Rn, then LxLU = lLxwlkd£L.k is defined by

( L x w ) k = (Lx¢»J)(8mk) = X w ( 3 x k ) - (»J(Lx3xk) = X ' wk +w

Hence
301k . 8X1
LxLU Xi duck .
35137 -|- WJ 3;3/*

In particular, if X aas = E j (SU a , then

n
31,0/6
LuLU Law .d k
333'L QUO
k=1

If X is the radial vector field X 'II £I°i§mi, then

LxLU (X-wk we)d:z:k.

k

Example 3.1.18. Consider a smooth vector field X = F633 -|- Gdy + HEL on R3. We want
to compute Lx do, where do is the volume form on R3, do = do: A dy A dz. Since Lx is
an even s-derivation of Q* ( M ) , we deduce
Lx(d:U /\ dy A d z ) = ( L x d x l A dy /\ dz -|- da: A ( L a d y ) A dz -|- da: A dy A ( L o d z ) .
Using the computation in the previous example we get
OF OF OF
:= day 3 do + dz, Lx(dy) = dG, L x ( d z ) dH
31: + y Hz
SO that
oF OH_|_ KG
Lx (do) do = (div X ) d v .
+
3312 <99 32:
In particular, we deduce that if div X = 0, the local flow generated by X preserves the
form do. We will get a better understanding of this statement once we learn integration on
manifolds, later in this chapter.

Example 3.1.19 (The exponential map of a Lie group). Consider a Lie group G. Any
element g 6 G defines two diffeomorphisms of G: the left (Lg), and the right translation
(Rg) on G,
Lglhl = g - by, Rg(/2) = h g, WE 6 G. -
A tensor field T on G is called left (respectively right) invariant if for any g 6 G (Lg)*T =
T (respectively (Rg)*T = T). The set of left invariant vector fields on G is denoted by
LG.
The naturality of the Lie bracket implies

(L9)*[X»Y] I [(L9)*X» (L9)*Y]»

 Calculus on Manifolds 87

so that VX, Y E LG, [X, Y] 6 LG. Hence LG is a Lie subalgebra of Vect (G). It is called
called the Lie algebra of the group G.
Fact 1. dim LG = dim G. Indeed, the left invariance implies that the restriction map
LG -> T1G, X »-> XI is an isomorphism (Exercise). We will often find it convenient to
identify the Lie algebra of G with the tangent space at 1.
Fact 2. Any X Q LG defines a local flow i>tx on G that is is defined for all t Q R In other
words, il'>tx is a flow. (Exercise) Set
exp(tX) : = @ ( 1 ) .
We thus get a map
exp : T1G 2 LG -> G, X »-> exp(Xl
called the exponential map of the group G.
-
Fact 3. il'>tx (g) = g exp (ISX), i.e.,

<I>§< = Rexp(tX)-
Indeed, it suffices to check that
d
llllllllll

dt (9 . exp (TX)) = X9-

t=0
-
We can write (g exp(tX)) = Lg exp(tX) so that
d d
dt t=0
(Le exp(tx)) ; (Le)*
( dt t=0
exp(tx) (Lg)*X Xga (left invariance).

The reason for the notation exp (TX) is that when G = GL(n, K), then the Lie algebra of
G is the Lie algebra gl(n, K) of all n X n matrices with the bracket given by the commutator
of two matrices, and for any X 6 LG we have (Exercise)
l
exp(Xl eX Xi.
k>0
k!

Exercise 3.1.20. Prove the statements left as exercises in the example above.

Exercise 3.1.21. Let G be a matrix Lie group, i.e., a Lie subgroup of some general linear
group GL(N, K). This means the tangent space T1G can be identified with a linear space
of matrices. Let X, Y E T1G, and denote by exp(tX ) and exp(tY) the 1-parameter groups
with they generate, and set
g(s, t) = exp(sX) exp(tY) eXP(-SX) exp(-tY).

(a) Show that

g.s,t = 1 -1- [X,Y]algst -|- O((s2 -|- t2)3/2) as s , t -> 0,

where the bracket [X, Y]al9 (temporarily) denotes the commutator of the two matrices
X and Y.
88 Lectures on the Geometry of Manifolds

(b) Denote (temporarily) by [X, Ylgeovrz the Lie bracket of X and Y viewed as left invari-
ant vector fields on G. Show that at 1 6 G

[X, Ylalg = [X, Ylgeom-

(c) Show that Q(n) C n, R) (defined in Section 1.2.2) is a Lie subalgebra with respect
to the commutator [- , -]. Similarly, show that _u(n), M) C n, (C) are real Lie
subalgebras of M, (Cl, while we,
(C) is even a complex Lie subalgebra of M, (C).
(d) Prove that we have the following isomorphisms of real Lie algebras,

Lo(n) 2 Q(t1)» L U ( n ) 2 Mtl), LsU(n) 2 LW) and £sL(n,<:> 2 QW, (C)-

Remark 3.1.22. In general, in a non-commutative matrix Lie group G, the traditional

equality
exp(tX) exp(tY) = exp(t(X -|- Y))
no longer holds. Instead, one has the Campbell-Hausdorfformula
exp(tXl -exp(tY) = exp ( ¢d1(x, y) + t2d2(x, y) + t3d3(x, y) + - -) 7

where do are homogeneous polynomials of degree k in X, and Y with respect to the multi-
plication between X and Y given by their bracket. The do's are usually known as Dynkin
polynomials. For example,
l
d1 IX, y) X + Y , d2(X,Y) -21X 7Y L
_ l
d3(X, Y) -([X, [X, YH -1- [Y, [Y, XH) etc.
12 '

For more details we refer to [58, 113].

3.2 Derivations of Q' (M)

3.2.1 The exterior derivative

The super-algebra of exterior forms on a smooth manifold M has additional structure, and
in particular, its space of derivations has special features. This section is devoted precisely
to these new features.
The Lie derivative along a vector field X defines an even derivation in Q°(M). The
vector field X also defines, via the contraction map, an odd derivation iX , called the interior
derivation along X , or the contraction by X ,
ix : =tr (X (X) cu), Vi 6 Q"(M).
More precisely, ix is the (r - 1)-form determined by
lix(.4JllX1, . . . 7X,"_1) I Wlx,x1, . . . 7X7~-1)7 VXI, 7XT°-1 6 Vect (M).
 Calculus on Manifolds 89

The fact that iX is an odd .9-derivation is equivalent to

ix(w /\ 17) = (1lXw) /\ 17 -|- (-l)deg"Jw /\ ( i X n ) , V i , 17 6 Q * ( M ) .
Often the contraction by X is denoted by XJ .

Exercise 3.2.1. Prove that the interior derivation along a vector field is a .9-derivation.

Proposition 3.2.2. (61) Hx, iY]S = ixia -|- iyix = 0.

(b) The super-commutator Of Lx and i n as .9-derivations of Q* ( M ) is given by
[Lx,iy]5 = LXiY - iYLX = i[x,y]-
The proof uses the fact that the Lie derivative commutes with the contraction operator,
and it is left to the reader as an exercise.
The above s-derivations by no means exhaust the space of s-derivations of Q' (M)- In
fact we have the following fundamental result.

Proposition 3.2.3. There exists an odd 5-derivation d on the 3-algebra ofdy erentialforms
gr( - I uniquely characterized by the following conditions.

(a) For any smooth function f E QOIMI, df coincides with the deferential o f f .
(b) d2 0.
Z

(c) d is natural, i.e., for any smooth function $5 : N -> M , and for any form w on M , we
have
dg25*w = Cl5*dw('I \. w, dl = 0).
The derivation d is called the exterior derivative.

Proof. Uniqueness. Let U be a local coordinate chart on M n with local coordinates

1
7 ,as"'). Then, over U, any 1°-form w can be described as
z'
CU (.Ui1...i_d.CU 1 /\ /\ doi r
1$11<---<i.,.$n

Since d is an 5-derivation, and d(da;.i) = 0, we deduce that, over U

do = (do 1 . . .'L T' ) /\ (del /\ - - - /\ day)
1Si1<~~~<¢~S'f1
/'111...i,-
do A<dx'1A-~-Adm>. (3.2.1)

Thus, the form do is uniquely determined on any coordinate neighborhood, and this com-
pletes the proof of the uniqueness of d.

Existence. Consider an r-form w. For each coordinate neighborhood U we define do In as

in (3.2.1). To prove that this is a well defined operation we must show that, if U, V are two
coordinate neighborhoods, then
dwlu dwlv 0numv.
90 Lectures on the Geometry of Manifolds

Denote by (x1, , in the local coordinates on U, and by (g1, ,y") the local coordi-
nates along V, so that on the overlap U O V we can describe the y's as functions of the ac's.
Over U we have
'i i
cu = Wi1...id$ 1 I\...Adac° 'r'

1$i1<---<i,~$n

div /\ (day21 /\ /\ d;z:"`),

while over V we have

-
w- * _ _ al /\
cu31...3'r'dy . . . /\ dy j "
l<j1<°--<j,~§n

J _]1..._]7"
do dye ldYj1/\---/\dyj'")-

The components Wi1..."£,- and Wil---jT' are skew-symmetric, i.e., V0 GS

Wi,,(1)...i,,(,.> = €(0lWi1...'£,. 7

and similarly for the aT/s. Since w lu= w Iv over U n V we deduce

3?/i"
Wi1...i,-
3:01 3xi,~ Wall---jr'
Hence
7"
gall ..-iT J1 a2yjk 3yj"` all---jr
ECk + 3£ui1 $y7"
A

311
1
3 1 ex 3xik 351 Wj1~»-jr 351 311%
7

where in the above equality we also sum over the indices j1 , • • 'Q according to Einstein's
convention. We deduce
a . Z . _ .
2 2 wa.z
I I I 7"
dag'/\da;@./\.../\dx@v~
:u
1$¢1<--~<¢,.$n

T'
33/11 §2yj1¢ 33/i"
22 k = l 33321
ex 3xk 3£61, do# /\ do" /\ Adm

T' 63/j1
3y7* 3¢TJi
-I- 22
i k=1 3a:'1
3;u2t
.
1
3x2 'i' j'"d QUO1. /\ d:u z 1 /\ '
I\d1I'. (3.2.2)

Notice that
62 32
7
3ac.'é3;c'k 335333i

while do /\ day = -dan /\ do so that the first term in the right-hand side of (3.2.2)
vanishes. Consequently on U (1 V

30011 . . .i
day /\ do /\ do 3?/" aWj1---j1~ /\ day /\ do day"
35 3351 351 351%
 Calculus on Manifolds 91

ayjl _ 3?/" .
am doi
) ( /\ .
35321
d
ac
'L
1
) /\ /\
( 351" do' )
ldWj1---jrl /\ dyjl /\ A dye" I 3W7i1°°°i'"
3yi
This proves do In= do Iv over U O V. We have thus constructed a well defined linear
map
co : o°<m) -> o°+1(m).
To prove that d is an odd s-derivation it suffices to work in local coordinates and show that
the (super)product rule on monomials.
Let H = fox 1 A ---
A do and CU = gdmjl /\ /\ dxjs . We set for simplicity
dat] :z d /\ a /\ d a y " and dif] == 4331. /\ - - - /\ dzvjs
Then
d(0 /\ cu) d( fgdQJ /\ do.]) = d ( f 9 ) /\ do] /\ d#
( d f - g - 1 - f ~dg)Ad:cIAd:z:']
i f /\ do] /\ d:z:'] + ( - I ) ' ( f /\ d i f ) /\ (do /\ d:z:'1)
do /\ cu + ( - 1 l d e g 0 0 /\ do.
We now prove d2 = 0. We check this on monomials fizzI as above.
d2(fd»#) = d(df /\ i f ) = (d2fl /\ data.
Thus, it suffices to show d2 f = 0 for all smooth functions f. We have
of2
d2f = dat /\ dxj .
6l1/.i3£1Zlj
The desired conclusion follows from the identities
c'9f2 I c'9f2 I

and day /\ data -dace A dcc .

3:c"13:87 3227 do
Finally, let go be a smooth map N -> M and w = EI wfdail be an T-form on M. Here I
runs through all ordered multi-indices l < ii < < i < dim M . We have ---
dl¢*w) (¢zN(¢*w/l /\ ¢*(dw') -|- ¢*(,J' /\ d(¢*d¢¢')) .
I
For functions, the usual chain rule gives dn(gz5*wI) = C15*(dMw1l. In terms of local coor-
dinates ( ) the map (15 looks like a collection of smooth n functions ( i = xi o <15 : WWI
and since the pullback is a morphism with respect to the exterior multiplication we get
¢*(dw') = dqsf = ¢*d$i1 /\ a - /\ ¢*da¥
do" /\ --- /\ dn¢".
In particular, i n ( d ¢ f ) = 0. We put all the above together and we deduce
in(¢*wl = ¢*(dmWI) /\ dqsf = ¢*(dMw') /\ ¢*daI = ¢*(dmWl-
The proposition is proved.

Proposition 3.2.4. The exterior derivative satisfies the following relations.

92 Lectures on the Geometry of Manifolds

(0)
[d, (qS = 2d2 = 0.
lb) (Caftan's homotopy formula)
[d,iX]S = dix +ixd = Lx, VX 6 Vect (M).
(c)
[d, Lx]s = dLx - Lxd = 0, VX E Vect

Above, [ , ]5 denotes the super-commutator in the s-algebra of real endomorphisms of

Qc
(M)
Proof. To prove the homotopy formula set
'D [d,ix]s = Dix + i d .
The operator QD is an even s-derivation of Q*(M). It is a local s-derivation, i.e., if Ld 6
Q*(M) vanishes on some open set U then Day vanishes on that open set as well. The
reader can check easily by direct computation that Be = Law, Vi 6 Q0(M) ® Ql ( M ) .
The homotopy formula is now a consequence of the following technical result left to the
reader as an exercise.

Lemma 3.2.5. Let QD, QD' be two local 5-derivations o f f ' ( M ) which have the same parity,
i.e., they are either both even or both odd. IfQD = QD' on Q0(M) ® Ql ( M ) , then QD = QD'
on Q°(M).

Part (iii) of the proposition is proved in a similar way.

An immediate consequence of the homotopy formula is the following invariant descrip-

tion of the exterior derivative:
T'

ICZQJIIXo, XI7 7 Xl l-1lixi(wlx07

i=0
(3.23)
-I- (-1)i+li(»J([Xi7 Xi]7 X0, ..., X, ..., Xi, ..., X ) .
O<i<.j<'r

Above, the hat indicates that the corresponding entry is missing.

Equality (3.2.3) is a simple consequence of the homotopy formula. We prove it in two
special case T' = l and T' = 2.
The case 7' I 1. Let w be an 1-form and let X , Y 6 Vect ( M ) . We deduce from the
homotopy formula

dw(X, Y) I ('id¢*J)(Yl I (Lx(»J)(Yl (d(»J(Xll(Y)-

On the other hand, since Lx commutes with the contraction operator, we deduce

X(,u(Y) I Lx((»(y)) I (Lx~J)(y) +w([X, y1).

 Calculus on Manifolds 93

Hence
d w ( X , Y) = Xw(Y) - (»J([X7Y]) - (d°J(X))(Y) = Xw(Y) - Yw(X) - w([X7Y])-
This proves (3.2.3) in the case T' = 1.
The case r = 2. Consider a 2-form CU and three vector fields X, Y and Z. We deduce from
the homotopy formula
(d~»)(x, y, z) = lixdwlly, 2) = ( L x - ¢iiX>(,J(y, z ) . (3.2.4)
Since Lx commutes with contractions we deduce
lLXWllY» X ) X((»J(Y»Zll
I w([X,Y],Z) w(Y, [X»Z])- (3.2.5)
We substitute (3.2.5) into (3.2.4) and we get
(do)(X,Y, Z) = X((»J(Y7 Z)) - w([X,Y], Z) - w(Y, [X, Z]) - d(iXw)(Y,X). (3.2.6)
We apply now (3.23) for r = 1 to the 1-form i x . We get
<1(¢x~J)(y/x) = y(ix~»(2)) - z(¢x~(y)) - ( i x ) ( [ y , 21)
= Y w ( X , Z) - Zn(X,Y) - w(X, [Y, Z]). (3.2.7)
If we use (3.2.7) in (3.2.6) we deduce
(dwllx, Y, 2) Xw(Y, z) - Yw(X, z) + Zw(X, iI
-w([X,Y], Z) +w([X, Z],Y) w([Y, Z],X). (3.2.8)
The general case in (3.2.3) can be proved by induction.

Exercise 3.2.6. Prove Lemma 3.2.5.

Exercise 3.2.7. Finish the proof of (3.2.3) in the general case.

3.2.2 Examples

Example 3.2.8. (The exterior derivative in R3).

(a) Let f Q C>°(]R2.3). Then

<9
df = f o x -|- f l y
a -|-
aadz.
fix <99 32:
The differential df looks like the gradient of f .
(b) Let cu 6 Q1(lR3), to = P d f : -|- Qdy -|- Rdz. Then
do= dPAda3-l-dQAdy+dRAdz
<'3Q UP <9R (QQ <9P OR
( ex - by ) day°Ady+

so that do looks very much like a curl.

by - ( 3,2
) d y A d s -I-
( 3,2 )
3:6
d Ad
z E'

(c) Let w = Pdy /\ dz -|- Qdz /\ do: -|- Rdac /\ do 6 Q2(R3). Then
UP IQ AR
do = -|- -I- 3,2 d;z:AdyAdz.
3:6 by
This looks very much like a divergence.
94 Lectures on the Geometry of Manifolds

Example 3.2.9. Let G be a connected Lie group. In Example 3.1.19 we defined the Lie
algebra LG of G as the space of left invariant vector fields on G. Set
Q,eft(G) L left invariant 1°-forms on G.
In particular, LG f\J

Qlleft If we identify L G 2 T1 G, then we get a natural isomor-

phism
~
=
Q;,(G) NLG.
The exterior derivative of a form in Q l e f t can be described only in terms of the algebraic
structure of LG.
Indeed, let Ld Q LG = o}e..¢(G). For X, Y Q LG we have (see (3.2.3))
dw(X, Y) = Xw(Y) - Ye(Xl - w([X, Y]).
Since w, X and Y are left invariant, the scalars w(X) and w(Y) are constants. Thus, the
first two terms in the above equality vanish so that
do(X,Y) = -°J([X7Yl)-
More generally, if w E Q r e f t ' then the same arguments applied to (3.2.3) imply that for all
X07 . . . 7 X G LG we have
dwlx0, Xl, ...7 X,,-) l-1lH-jWl[XiaXj]7X1' C C C 7 X, Xi X,,,). (3.2.9)
0<i<j<T"

3.3 Connections on vector bundles

3.3.1 Covariant derivatives

We learned several methods of differentiating tensor objects on manifolds. However, the

tensor bundles are not the only vector bundles arising in geometry, and very often one is
interested in measuring the "oscillations" of sections of vector bundles.
Let E be a K-vector bundle over the smooth manifold M (K = R, (C). Technically, one
would like to have a procedure of measuring the rate of change of a section u of E along a
direction described by a vector field X. For such an arbitrary E, we encounter a problem
which was not present in the case of tensor bundles. Namely, the local flow generated by
the vector field X on M no longer induces bundle homomorphisms.
For tensor fields, the transport along a flow was a method of comparing objects in dif-
ferent fibers which otherwise are abstract linear spaces with no natural relationship between
them.
To obtain something that looks like a derivation we need to formulate clearly what
properties we should expect from such an operation.

(a) It should measure how fast is a given section changing along a direction given by a
vector field X . Hence it has to be an operator
V : Vect (M) X COO(E) -> COO(E), (X,'u,) »-> VxU,
where we recall (see Definition 2.1.27) that coo (E) denotes the space of smooth sec-
tions of E over M.
 Calculus on Manifolds 95

(b) If we think of the usual directional derivative, we expect that after "rescaling" the
direction X the derivative along X should only rescale by the same factor, i.e.,
'v'f€COO(Ml: V f x U = fVx'LL.
(c) Since V is to be a derivation, it has to satisfy a sort of (Leibniz) product rule. The only
product that exists on an abstract vector bundle is the multiplication of a section with a
smooth function. Hence we require
Vx(fu1 = (Xf)u-l-fVxu, 'v'f Q C°°(M), u € C°°(E).

The conditions (a) and (b) can be rephrased as follows: for any u p C°°(E), the map
Vu : Vect (M) -> COO(E), X »-> V x U ,
is COO (M)-linear so that it defines a bundle morphism (see Definition 2.1.30(b) and Exam-
ple 2.3.2)
~
Vu 6 Hom(TM, E) = CIOC(T*M (X) El.
Summarizing, we can formulate the following definition.

Definition 3.3.1. A covariant derivative (or linear connection) on E is a K-linear map

V : C'°°(E) -> CYOO(T*M ® E ) ,
such that, Vf 6 C ° ° ( M ) , and Vu 6 C°°(E), we have
V(f'u.) = df (X) u fV'u..
Example 3.3.2. Let M E K7" X M be the rank T' trivial vector bundle over M. The space
cfoo (KM) of smooth sections coincides with the space C°°(M, Kl of K"-valued smooth
functions on M. We can define
v0 : C°°(M,]K") -> c°°(m,T*m (X) HC)
v " ( f , ...,f) = (dfl, ...,df~).
One checks easily that V is a connection. his is called the trivial connection.

Remark 3.3.3. Let V0, V1 be two connections on a vector bundle E -> M. Then for any
QUO Goo (M) the map
V Z QV1 -|- (l - a)V0 : C°°(E) -> COO(T* ® E)
is again a connection.

@'Notation. For any vector bundle F over M we set

Qklpl : =c°°(A'<=T*m (X)

We will refer to these sections as deferential Le-forms with coejicients in the vector
bundle F.

Proposition 3.3.4. Let E be a vector bundle. The space A ( E ) of linear connections on E

is an a]§'ine space modeled on Q1 (End (E)).
96 Lectures on the Geometry of Manifolds

Proof. We first prove that A (E) is not empty. To see this, choose an open cover {UO} of
M such that E I UQ is trivial Va. Next, pick a smooth partition of unity (fig) subordinated
to this cover.
Since E IUa is trivial, it admits at least one connection, the trivial one, as in the above
example. Denote such a connection by Va. Now define

a
in
One checks easily that V is a connection so that .A(E) is nonempty. To check that A (E)
is an affine space, consider two connections V0 and V l . Their difference A = V1 - V0 is
an operator

A C°°(E)->C°°(T*M (X) E),

satisfying A(fu) = f A r ) , Vu 6 C°°(E). Thus,

A E c°°(H<>m(E,T*m®E)) iv
C°°(T*M®E*®E)
~= Q1(E*®E) 5 Q1(End(E) ).
Conversely, given V0 G A(E) and A 'E Q1 ( End(E)) one can verify that the operator

vA = v0 + A : c°°(E> -> Q1(El.

is a linear connection. This concludes the proof of the proposition.

The tensorial operations on vector bundles extend naturally to vector bundles with con-
nections. The guiding principle behind this fact is the product formula.

. If E ii = 1, 2) are two bundles with connections V , then E1 (X) E2 has a naturally

induced connection VE1®E2 uniquely determined by the product rule,

VE1®E2 (U1 ® url = (Vlul) ® U2 + U1 (X) V21IQ.

. A connection on a bundle E induced a connection V* on the dual bundle E* deter-

mined by the identity

X('U, u) (Vxv,u) -|- ( v , V X u ) , Vu € C°°(E), U € C°°(E*), X € Vect (M),

where

(.., )=c*°°(E*) X C°°(E)->C°°(M)

. is the pairing induced by the natural duality between the fibers of E* and E.
Any connection VE on a vector bundle E induces a connection v E n d ( E ) on
End (E) 2 E* (X) E defined by

<vE"d<E)T)(u) = VI(Tu) - T ( V E u ) = WE, T]u, (3.3.1)

VT Q End(E), u Q c<>°(E).
 Calculus on Manifolds 97

A metric h on a real vector bundle E -> M is a section of E* (X)

E* A connection V
on E is said to be compatible with a metric h on E if
VE* ®E* h 0.
If V is compatible with a metric h then, for any u, u p COO(E) and any X Q Vect (M) we
have
X - h(u,'u) = h(VxD'U,, u) -|- h(u, VxU). (3.3.2)

It is often useful to have a local description of a covariant derivative. This can be obtained
using Cartan's moving frame method.
Let E -> M be a K-vector bundle of rank T' over the smooth manifold M. Pick a
coordinate neighborhood U such E In is trivial. A moving frame2 is a bundle isomorphism
¢¢U=]I<.°xm->EIU.
Consider the sections ea = C$5(50,), a = 1, ..., r, where 6O are the natural basic sections
of the trivial bundle u , As Hz: moves in U, the collection l€1l$l3), ..., €t~(11)) describes
a basis of the moving fiber E33, whence the terminology moving frame. A section u e
0°°(Elu) can be written as a linear combination
u u ae a ua E C°°(U,]K).
Hence, if V is a covariant derivative in E, we have
Vu = duo" (X) ea + u°'Vea.
Thus, the covariant derivative is completely described by its action on a moving frame.
To get a more concrete description, pick local coordinates (xi) over U. Then V8O E
Q1 (E IU) so that we can write
B W
Vea = r o ® 65, F 5
pa
E c°°(U, KI.

Thus, for any section usea of E In we have

Vu = duo" (X) ea + FouOdxztz (X) 65. (3.3.3)

It is convenient to view
It B
pa
) as an T' X 1°-matrix valued 1-form, and we write this as

do (X) F .
The form F = do (X) F is called the connection I -form associated to the choice of local
gauge. A moving frame allows us to identify sections of E IU with K"-valued functions on
U, and we can rewrite (3.33) as
Vu = du -|- Fu. (33.4)
If the connection V is compatible with a metric h and we choose the local moving frame
(ea) defined to be h-orthonormal, then (33.2) implies that for every vector field CX we
have
fLlVX€o, 651 = -/2(€o 7 Vx€0)-
2A moving frame is what physicists call a choice of local gauge.
98 Lectures on the Geometry of Manifolds

Choosing local coordinates w) we deduce that if V is compatible with a metric h and (60)
is an orthonormal moving frame, then the associated connection 1-form
F div (X) FE

satisfies the additional property that Fr are skew-symmetric matrices.

A natural question arises: how does the connection 1-form changes with the change of
the local gauge?
Let f = (to) be another moving frame of E IN. The correspondence € a * - f 0 defines
an automorphism of E In. Using the local frame e we can identify this correspondence
with a smooth map g : U -> GL(r; K). The map g is called the local gauge transformation
relating e to f .
Let f` denote the connection 1-form corresponding to the new moving frame, i.e.,

Vfo Fgfg.
Consider a section O' of E In. With respect to the local frame Leal the section O' has a
decomposition
0' uaeo,

while with respect to (fa) it has a decomposition

0' '&f@.
The two decompositions are related by
'U g'&. (3.3.5)

Now, we can identify the E-valued 1-form Vo with a K1°-valued 1-form in two ways: either
using the frame e, or using the frame f . In the first case, the derivative V0 is identified
with the K-valued 1-form

du + Fu,
while in the second case it is identified with
do + Fa.
These two identifications are related by the same rule as in (3.3.5):
du -|- Fu g(diL -|- Fa).

Using (3.3.5) in the above equality we get

(dgM -|- gdia -|- Fgvl = gd?) -|- g l l .

Hence
F 9
-1
dg+9
-1
F9-
The above relation is the transition rule relating two local gauge descriptions of the same
connection.
 Calculus on Manifolds 99

1@A word of warning. The identification

{mo'uing f T°ames} 2 {local t1°ivz'alization}

should be treated carefully. These are like an object and its image in a mirror, and there is
a great chance of confusing the right hand with the left hand.

N=>
More concretely, if ta : Ea > K" X UO( (respectively to : EB
Z K" X UB ) is a
trivialization of a bundle E over an open set Ua (respectively UB ), then the transition map
"from Q to B" over Uo n UB is 96a = pa o ta 1. The standard basis in K , denoted by (6,-),
induces two local moving frames on E:
€ayi I to 1ldz) and €[3,¢ = to lldzl.
On the overlap U o U 5 these two frames are related by the local gauge transformation

85,1 = go l e a , i Z gaBea,'1l-

This is precisely the opposite way the two trivializations are identified.

The above arguments can be reversed producing the following global result.

Proposition 3.3.5. Let E -> M be a rank r smooth vector bundle, and (Ua) a trivializing
cover with transition maps gal? : Uo, VW UB -> G L ( r , K ) . Then any collection of matrix
valued 1-forms Fo, E Q1(End U ) satisfying

FB (goédgoel + gogfogae = -(dg go 1 + QBQFQQQ 1 over Ua D UP, (33.6)

uniquely defines a covariant derivative on E.

Exercise 3.3.6. Prove the above proposition.

We can use the local description in Proposition 3.3.5 to define the notion of pullback of
a connection. Suppose we are given the following data.

.. A smooth map f : N -> M.

A rank T' K-vector bundle E -> M defined by the open cover (Ua), and transition

. maps Quo : Ua VW UP -> GL(]K'°).

A connection V on E defined by the l-forms Fa 6 Q1 l E n d l U o , ) l satisfying the
gluing conditions (3.3.6).

Then, these data define a connection f*v on f*E described by the open cover
f-1(Ua), transition maps QBQ o f and l-forms f*II0. This connection is independent
of the various choices and it is called the pullback of V by f

Example 3.3.7 (Complex line bundles). Let L -> M be a complex line bundle over the
smooth manifold M . Let {Ua} be a trivializing cover with transition maps zag : Ua O
U5 -> Q* = GL(1, (I). The bundle of endomorphisms of L, End (L) = L* (X) L is
~
trivial since it can be defined by transition maps (zofil -1 ®ZoB = 1. Thus, the space of
100 Lectures on the Geometry of Manzfolds

connections on L, .A (L) is an affine space modeled by the linear space of complex valued
1-forms. A connection on L is simply a collection of (C-valued 1-forms wa on UO( related
on overlaps by
d
we = ZAg +w0' = dlogzog +w0'
zag

3.3.2 Parallel transport

As we have already pointed out, the main reason we could not construct natural derivations
on the space of sections of a vector bundle was the lack of a canonical procedure of iden-
tifying fibers at different points. We will see in this subsection that such a procedure is all
we need to define covariant derivatives. More precisely, we will show that once a covariant
derivative is chosen, it offers a simple way of identifying different fibers.
Let E -> M be a rank r K-vector bundle and V a covariant derivative on E. For any
smooth path by : [0, 1] -> M we will define a linear isomorphism T'Y : E'y(0) -> E'YU)
called the parallel transport along by. More exactly, we will construct an entire family of
linear isomorphisms

To : Ev(0) -> Et(t)°

One should think of this To as identifying different fibers. In particular, if U 0 6 ECU) then
-
the path t +-> ut - Ttu0 6 E7U) should be thought of as a "constant" path. The rigorous
way of stating this "constancy" is via derivations: a quantity is "constant" if its derivatives
are identically 0. Now, the only way we know how to derivate sections is via V, i.e., ut
should satisfy
d
V2ut 0, where it
dt

The above equation suggests a way of defining To. For any 'U,0 e E7(0)° and any t e [0, 11,
define Ttuo as the value at t of the solution of the initial value problem

V dtd u(tl 0
(3.3.7)
»U,(0) U0

The equation (3.3.7) is a system of linear ordinary differential equations in disguise.

To see this, let us make the simplifying assumption that W) lies entirely in some co-
ordinate neighborhood U with coordinates (LV1, ..., at" , such that E IU is trivial. This is
always happening, at least on every small portion of 7. Denote by l€O()1§a§v° a local mov-
ing frame trivializing E In so that u = uaeo. The connection 1-form corresponding to this
moving frame will be denoted by F E Q1(End ( I I . Equation (33.7) becomes (using
Einstein's convention)

-|- Ft5ug 0
(3.3.8)
u'""(0) up 7
 Calculus on Manifolds 101

where
d
Ft I - F *y J F E QUO ( End(11<T'l) End(KT) .
dt J
More explicitly, if the path 'y(t) is given by the smooth map
t»-> W) = (x1(t)»---»22"(tl) 7

then Ft is the endomorphism given by

Ft€,3 LEAF 'Bea-
The system (3.38) can be rewritten as
-|- pfgfgimg 0
7 (3.3.9)
we) US'
This is obviously a system of linear ordinary differential equations whose solutions
exist for any t. We deduce
TWo . (3.3.l0)
This gives a geometric interpretation for the connection 1-form F: for any vector field X,
the contraction
-ixr = -F(x) e c°°(End(E))
describes the infinitesimal parallel transport along the direction prescribed by the vector
field X, in the non-canonical identification of nearby fibers via a local moving frame.
In more intuitive terms, if fy(t) is an integral curve for X , and To denotes the parallel
transport along 'Y from E»V(0) to E7(t)' then, given a local moving frame for E in a neigh-
borhood of '1/(0), To is identified with a t-dependent matrix which has a Taylor expansion
of the form
To = ]1 - F015 -|- O(t2), t very small, (3.3.11)
with F0 I (1§xF) Io )-
If the connection V is also compatible with a metric h, then (3.3.2) shows the parallel
transport

To : E t ) -> E a )
is an isometric.

3.3.3 The curvature of eonneetion

Consider a rank k smooth K-vector bundle E -> M over the smooth manifold M, and let
V : Q0IE) -> o1(El be a covariant derivative on E.

Proposition 3.3.8. The connection V has a natural extension to an operator

do : oriel -> QT'-I-IIEI
uniquely defined by the requirements,
102 Lectures on the Geometry of Manzfolds

la)

dV IQ0(E) V,

lb) Vi 6 Q'°(m), 77 6 QSIE)

dv(w /\ 77) = do /\ 77 + (-l)"°w /\ den.

Outline of the proof Existence. F o r m 6 S2'°(M), u 6 Q0(E) set

dv(w (X) u) = do (X) 'LL -|- (-l)7°wVu. (3.3.l2)

Using a partition of unity one shows that any 17 6 Q""(E) is a locally finite combination of
monomials as above so the above definition induces an operator Q"°(E) -> QQ"-I-1 (E). We
let the reader check that this extension satisfies conditions (a) and (b) above.
Uniqueness. Any operator with the properties (a) and (b) acts on monomials as in (3.3.l2)
so it has to coincide with the operator described above using a given pa ition of unity.

Example 3.3.9. The trivial bundle M has a natural connection V0 - the trivial connec-
tion. This coincides with the usual differential d : Q0(M) (X) K -> Q1(M) (X) K. The
extension do" is the usual exterior derivative.

There is a major difference between the usual exterior derivative d, and an arbitrary
do_ In the former case we have d2 = 0, which is a consequence of the commutativity
[63 7 31135 ] = 0, where l ) are local coordinates on M. In the second case, the equality
(do)2 = 0 does not hold in general. Still, something very interesting happens.

Lemma 3.3.10. For any smooth function f 6 C°°(M), and any w 6 Q"(E) we have
V _ V
(d )2(f(»J) - f { ( d )2(»J}-

Hence (do)2 is a bundle morphism A'°T*M (X) E -> AT'+2T*m ® E.

Proof. We compute

(dV)2(f(»J) = dvwf A w + f d v w )
I -if A dew if dew f(d")2(» I f(d")2<».

As a map S20(E) -> Q2(El, the operator (do)2 can be identified with a section of
~
HomK (E, A2T*M ®R E ) 2 E* (X) A2T*M ®1R E = A2T*M ®1R EndK (E).

Thus, ldv)2 is an EI1dK (E)-valued 2-form.

Definition 3.3.11. For any connection V on a smooth vector bundle E -> M, the object
(do)2 6 Q2 lEI1d]K (E)) is called the curvature of V, and it is usually denoted by F(vl.
 Calculus on Manifolds 103

Example 3.3.12. Consider the trivial bundle Km. The sections of this bundle are smooth
K"-valued functions on M. The exterior derivative d defines the trivial connection on KM ,
and any other connection differs from d by a M(K)-valued 1-form on M. If A is such a
form, then the curvature of the connection d -|- A is the 2-form F(A) defined by
F(A)u = (d -|- A)2u = (DA -|- A /\ A)u, Vu E 0°°(m, KU.
The A-operation above is defined for any vector bundle E as the bilinear map

Qflmd (E)) X Qk(End (E)) -> Q9+k(End lm),

uniquely determined by
wi no (X) AB, A, B a c ° ° ( E n d ( E ) ) .
((»J'° (X) A)/\(t1S (X) B)

We conclude this subsection with an alternate description of the curvature which hope-
fully will shed some light on its analytical significance.
Let E -> M be a smooth vector bundle on M and V a connection on it. Denote
its curvature by F = F(V) 6 §22(End (E)). For any X , Y 6 Vect (M) the quantity
F(X, Yl is an endomorphism of E. In the remaining part of this section we will give a
different description of this endomorphism.
For any vector field Z, we denote by in AZ"(E) -> Q-1(18l the COO(M) - linear
operator defined by

iz(w ® u) = (izzy) (X) u, Vi E Q"(M), u 6 Q0(E).

For any vector field Z, the covariant derivative VZ : cOO (E) -> cOO (E) extends naturally
as a linear operator Q"l(E) -> AZ"(E), which we continue to denote by Vi, uniquely
defined by the requirements

Vzlw®ul lLzwl®u+w®Vzu.

The operators do, in, Vz AZ°(E) ->

QUO
satisfy the usual super-commutation iden-
titles.

i" -|- (Mz = Vi. (3.3.13)

ix?iy -I- 1yix 0. (3.3.14)

VXiY - iYVX = i[x,y]- (3.3.15)

For any 'LL 6 Q0(E) we compute using (33.13)-(3.3.15)

F ( x , Y)u = iy¢x(dv)2u = ¢Y(zx¢zv)vu

= bY(VX - do¢x)vu = ( z y v x ) v u - liydvlvx'LL

= ( V x i y - z'[Xy])Vu - V;/Vxu = (VxVy - V;/Vx - V[x,y]l"~'-

104 Lectures on the Geometry of Manzfolds

Hence

F(X,Y) I [Vx,Vy] - VlX,Y-- (3.3.16)

If in the above formula we take X a and Y a where ( a ) are local coordinates

9

on M , and we set Vi := Va m i , Vi pa then we deduce

7

FU Fez' Flux, gmj [Vow VJ]- (3.3.l7)

Thus, the endomorphism F,-i measures the extent to which the partial derivatives Vi, Vi
fail to commute. This is in sharp contrast with the classical calculus and an analytically
oriented reader may object to this by saying we were careless when we picked the con-
nection. Maybe an intelligent choice will restore the classical commutativity of partial
derivatives SO we should concentrate from the very beginning to covariant derivatives V
such that F(V) = 0.

Definition 3.3.13. A connection V such that F(V) = 0 is called flat.

Proposition 3.3.14. If V is a connection on a real vector bundle that is compatible with a

metric h, then FM 6 Q2 ( End_ III) where End_ ( E ) denotes the bundle of endomor-
phisms of E that are skew-symmetric with respect to h.

Proof. Let X, Y 6 Vect ( M ) and u, U 6 COO(M). Using (33.2) we deduce

XYh(u, u) = Xh(Vyu, u) + Xh(u, Viv)

= hlVxV§/U, 'UI + h(Vy'LL, VxUI + h(Vx'lL, Vyvl + h(u, VxVyUI.

Hence

[X, Y]h(u, UI = XY/z(u, UI - Y X h ( u v )

(3.3.18)
= h([VX, V5/]'LL, UI + h(u, [Vx 7 V3/]Ul.

On the other hand,

[X, Y]h(u, 'UI = h(V[Xy]'Ll,, 'UI -|- h(u, V[x,y]'U)- (3.3.19)

The desired conclusion is reached after subtracting (33.19) from (3.3.18).

A natural question arises: given an arbitrary vector bundle E -> M do there exist flat
connections on E? If E is trivial then the answer is obviously positive. In general, the
answer is negative, and this has to do with the global structure of the bundle. In the second
half of this book we will discuss in more detail this fact.
 Calculus on Manifolds 105

3.3.4 Holonomy

The reader may ask a very legitimate question: why have we chosen to name curvature,
the deviation from commutativity of a given connection. In this subsection we describe the
geometric meaning of curvature, and maybe this will explain this choice of terminology.
Throughout this subsection we will use Einstein's convention.
Let E -> M be a rank r smooth K-vector bundle, and V a connection on it. Consider
local coordinates (ac1, ac") on an open subset U C M such that E In is trivial. Pick a
moving frame (el, ..., of E over U. The connection 1-form associated to this moving
frame is
F 7 1 < a <71

It is defined by the equalities (Vi I= Viat)

V 26,3 Z FQBQQ. (3.3.20)

Using (3.3.l7) we compute

F i j i = (WW - V§VDw3 = V¢(Fj€0) - vjlr B)

'61`?@. 3%. \ a + F'Y
333'L 3933 e JB F -I mFr ea'

SO that
6Ili 32
.F
3:0 3113?
+ 11; pa pal* (3.3.21)

Though the above equation looks very complicated it will be the clue to understanding the
geometric significance of curvature.

p
.9 _
1
0 I

Fig. 3.1 Parallel transport along a coordinate parallelogram.

Assume for simplicity that the point of coordinates (0, ..., 0) lies in U. Denote by ToS
the parallel transport (using the connection V) from (ac1, ..., ac" l to (x1 -|- s, 332, ..., in
along the curve T »-> (x1 -|- T, x2, ..., mn). Define TO in a similar way using the coordinate
x2 instead of al.
106 Lectures on the Geometry of Manzfolds

Look at the parallelogram Ps,t in the "plane" (x1, 5132) described in Figure 3.1, where

P 0 = l 0 ) ° ° ' v 0 ) ) PI (s,0,...,0), $2 = (s,t,0,...,0), $3 = (0,t,0,...,0).

We identify E In with the trivial bundle E0 X U where E0 is the fiber of E over Po. We
now perform the counterclockwise parallel transport along the boundary of Ps,t. The out-
come is a linear map Jis,t : E0 -> Eo. Set $12 := F(66'1 ex2 a
) l ( 0 . 0 ) . F12 is also an
endomorphism of E0.

Proposition 3.3.15. We have

62
F12
3s3t
75
,t
_

We see that the parallel transport of an element u 6 E0 along a closed path may not
return it to itself. The curvature is an infinitesimal measure of this deviation.

Proof. The parallel transport along 3Pst can be described as

Is,t = T2-tTl-ST;T1.

The parallel transport To E0 -> EP1 can be approximated using (3.3.10)

TO = 11 - 51"1(p0) + 0152 + O(83). (33.22)

The term 01 is an endomorphism of E0 whose exact form is not relevant to our computa-
tions. In the sequel the letter C' (eventually indexed) will denote constants.
+ CQ# + o(i3))T1
$2

(11 iF2(P1)
II

(11 ip2(p1) + CQ# + O(i3)) (11 - SF1lPol + 0152 + O(53))

II

(11 8F1(Pol iF2(P1) + is/l2lPllFllPoll -|- C182 -|- C29 + 0(3)-

II

'

Above and in the sequel O(k) denotes an error < C'(82 + i 2 ) I ¢ / 2 as s,t -> 0. Now use the
approximation
3F2
F Qlpgl -1-
S
F2(P1l
to deduce

m
a
T;T1 11 - 511 - - st
33:1 )} l,,O
- F2F1
(3.3.23)
-|- C132 -|- c29 -|- O(3).
Similarly, we have

T1IST5Ti (11 + 5F1(p2) + c352 + 0(3))T2tT1s.

The F-term in the right-hand-side of the above equality can be approximated as
an a 1
n(p2) = FllPol + S 3561 Pol + t o IPO) + o(2).
 Calculus on Manifolds 107

Using (3.3.23) we get after an elementary computation

f ]1 _ I'
31"1 3II2
T1 ST2tT1S tII2 + st -|- F2F1 lp,
3362 3:131 t»») (3.3.24)
+ c452 + c59 + o(3).
Finally, we have
T2-tT1-.sT;T1 _- (11 + tF2 (193) + 06? + o(3))T1-8T;T1 7

with
as
r2(p3) = t o ) + ¢6§ Pol + (M + o(3).
Using (33.24) we get

T2-*T1-ST;Tl - 11 + st
an _ a + F2F1 - F11 1Po
3312 3551

+ 0852 + 099 + 0(3)

11 - StF12lP0) + $852 + C99 + o(3).
321/4
Clearly 356% -F12lP0) as claimed.

Remark 3.3.16. If we had kept track of the various constants in the above computation we
would have arrived at the conclusion that C8 = C9 = 0, i.e.,
'J'9 t = II - stF12 + O(3). (3.3.25)
Alternatively, the constant C8 is the second order correction in the Taylor expansion of
S»-> 790 = II, Vs, so $8 = 0. The same goes for CQ. Thus we have
dIs,t dTf,¢5
F12
dareaPst ds
Loosely speaking, the last equality states that the curvature is the "amount of holonomy per
unit of area".

The result in the above proposition is usually formulated in terms of holonomy.

Definition 3.3.17. Let E -> M be a vector bundle with a connection V. The holonomy of
V along a closed path 'Y is the parallel transpo along by.

We see that the curvature measures the holonomy along infinitesimal parallelograms.
A connection can be viewed as an analytic way of trivializing a bundle. We can do so
along paths starting at a fixed point, using the parallel transport, but using different paths
ending at the same point we may wind up with trivializations which differ by a twist. The
curvature provides an infinitesimal measure of that twist.

Exercise 3.3.18. Prove that any vector bundle E over the Euclidean space Rn is trivializ-
able.
Hint: Use the parallel transport defined by a connection on the vector bundle E to produce
a bundle isomorphism E -> E0 X R", where E0 is the fiber of E over the origin.
108 Lectures on the Geometry of Manzfolds

3.3.5 The Bianchi identities

Consider a smooth K-vector bundle E -> M equipped with a connection V = VE . We
have seen that the associated exterior derivative do : Q/>(E) -> Qp+1 (E) does not satisfy
the usual (do)2 = 0, and the curvature is to blame for this. The Bianchi identity describes
one remarkable algebraic feature of the curvature.
~
Recall that VE induces a connection in any tensor bundle constructed from E. In
particular, it induces a connection in E* (X) E = End (E) which we denote by V E n d ( E l .
This extends to an "exterior derivative"
DE do
End(E)
: op(End (E)) -> Op+1(End (E)).

Proposition 3.3.19 (Bianchi identity). Let E -> M be a K-vector bundle on M and V E

a connection on E. Then
D E F W E ) = 0.

Roughly speaking, the Bianchi identity states that (dV)3 is 0.

Proof. We will use the identities (3.3.13)-(3.3.15). For any vector fields X , Y , Z we have
ZxDE = VExNd(E) - DE¢x .
Hence,
(DEFIIX, y, z) ZzZyZxDEF = ¢Z¢y(v§"'*lE) - DEix)F
¢Z(vEXnc*(E)iy - i[x,y])F - in(v§nd(E) - D E / ) ¢ x F
land(E)iziy - i[x,z]iy - ii[x,y]lF
-(vi"d(E)¢ZiX - i[y,z]ix - VEz"dlE)¢y@x)F

+ i[y,z]ix + i[2 X ] i ¥ ) F
(ilx,y]i2

-(v§<"d(E)i,/i2 + VEy"dlE)¢Z¢X + vi"dlEl¢x¢y )F

A(x, Y, z) + A(y, Z, x ) + A(z, x, y),

where
A(x, y, zl ii li[x,yliz - vi"dlEliy¢z)F, VX, Y, Z 6 Vect ( m ) .
We compute immediately
'i[xy]izF F<z, [x,y1) = [vi, VSQy1] V[x,y1] .
Also for any u 6 Q0(E) we have
(vi"d(E)¢Y¢ZF)u = vi(Flz, elul - F(z, Y ) v § u = ii, F(z, y)] 'LL

[vi»v€,z u- ii, WyE, VE]1 u.

Hence
A(x,y,z) = ii, [eyE,v§]]
The Bianchi identity now follows from the classical Jacobi identity for commutators.
 Calculus on Manifolds 109

Example 3.3.20. Let K be the trivial line bundle over a smooth manifold M. Any connec-
tion on K has the form V" = d-I-cu, where d is the trivial connection, and Ld is a K-valued
1-form on M. The curvature of this connection is

FM do.

The Bianchi identity is in this case precisely the equality d2w 0.

3.3.6 Connections on tangent bundles

The tangent bundles are very special cases of vector bundles so the general theory of con-
nections and parallel transport is applicable in this situation as well. However, the tangent
bundles have some peculiar features which enrich the structure of a connection.
Recall that, when looking for a local description for a connection on a vector bundle,
we have to first choose local coordinates on the manifolds, and then a local moving frame
for the vector bundle. For an arbitrary vector bundle there is no correlation between these
two choices.
For tangent bundles it happens that, once local coordinates (x'i) are chosen, they au-
tomatically define a moving frame of the tangent bundle, la gel and it is thus very
natural to work with this frame. Hence, let V be a connection on TM. With the above
notations we set

Vial rkak lV?3=V3'i)°

The coefficients PkJ are usually known as the Christoffel symbols of the connection. As
usual we construct the curvature tensor

F(x, y) = We, Vy] - v[XY1 6 (2°°(End(Tm)).

Still, this is not the only tensor naturally associated to V.

Lemma 3.3.21. For X, Y € Vect ( M ) consider

T(X, Y) = VxY - V)/X - [X, Y] 6 Vect (M).

ThenVf€C°°(Ml

T(fX»Y) i T(X,fY) i fT.(x,y),

so that T ( ¢ , o ) is a tensor T 6 Q2 ( T M I , i.e., a 2-form whose coe]§"icients are vector fields
on M. The tensor T is called the torsion of the connection V .

The proof of this lemma is left to the reader as an exercise. In terms of Christoffel
symbols, the torsion has the description

T(<9¢»<9j)
k
; FjkZlak .
Definition 3.3.22. A connection on TM is said to be symmetric i f T = 0.
110 Lectures on the Geometry of Manzfolds

We guess by now the reader is wondering how the mathematicians came up with this
object called torsion. In the remaining of this subsection we will try to sketch the geomet-
rical meaning of torsion.
To find such an interpretation, we have to look at the finer structure of the tangent space
at a point cz: 6 M. It will be convenient to regard TxM as an affine space modeled by R" ,
n = dim M. Thus, we will no longer think of the elements of TxM as vectors, but instead
we will treat them as points. The tangent space Ta,M can be coordinatized using amine
frames. These are pairs (p, e), where p is a point in T,,,M, and e is a basis of the underlying
vector space. A frame allows one to identify TasM with R", where p is thought of as the
origin.
Suppose that A, B are two affine spaces, both modelled by R", and (p, e), (p, f ) are
affine frames of A and respectively B. Denote by (we the coordinates in A induced by
the frame (p, e), and by (y.7 ) the coordinates in B induced by the frame (q, f). An affine
isomorphism T : A -> B can then be described using these coordinates as
TO;->]Ry ate->y: S:r:+v,
where 'U is a vector in R", and S is an invertible n X n real matrix. Thus, an affine map is
described by a "rotation" S, followed by a translation 'U. This vector measures the "drift"
of the origin. We write T = Six.
If now (w) are local coordinates on M, then they define an affine frame Ax at each
az 6 M: (AA, = (0; (80)l- Given a connection V on TM, and a smooth path by : I -> M,
we will construct a family of affine isomorphisms To : T7(0) -> T*yU) called the amine
transport of V along by. In fact, we will determine To by imposing the initial condition
TO = II, and then describing To.
This is equivalent to describing the infinitesimal affine transport at a given point 5120 6
M along a direction given by a vector X = X a , 6 T1120M. The affine frame of T10M is
Ago (0; (60)-
i

If £1% is a point along the integral curve of X , close to 5170 then its coordinates satisfy
x0 HW +O(t2).
This shows the origin S130 of A230 "drifts" by tX -|- O(t2). The frame (am) suffers a parallel
transport measured as usual by 11 - tixI` -|- O(i2). The total affine transport will be
Tt = (1 - ¢¢X1"lL¢x + o(t2).
The holonomy of V along a closed path will be an affine transformation and as such it has
two components: a "rotation" and a translation. As in Proposition 3.3.15 one can show
the torsion measures the translation component of the holonomy along an infinitesimal
parallelogram, i.e., the "amount of drift per unit of area". Since we will not need this fact
we will not include a proof of it.

Exercise 3.3.23. Consider the vector valued 1-form Ld E Q1(Tm) defined by

w(X) = X VX 6 Vect (M).
Show that if V is a linear connection on TM, then dew = TV, where TV denotes the
torsion of V.
 Calculus on Manifolds 111

Exercise 3.3.24. Consider a smooth vector bundle E -> M over the smooth manifold
M . We assume that both E and TM are equipped with connections and moreover the
connection on TM is torsionless. Denote by V the induced connection on A2T*M (X)
End ( E l . Prove that VX, Y, Z 6 Vect (M)
(VxF) (y, z) -|- (vypl (z, x) WzFMX, y) = 0.

3.4 Integration on manifolds

3.4.1 Integration of 1-densities

We spent a lot of time learning to differentiate geometrical objects but, just as in classical
calculus, the story is only half complete without the reverse operation, integration.
Classically, integration requires a background measure, and in this subsection we will
describe the differential geometric analogue of a measure, namely the notion of 1-density
on a manifold.
Let E -> M be a rank k, smooth real vector bundle over a manifold M defined by
an open cover (Ua) and transition maps 9g0 : U05 -> GL(k,]R) satisfying the cocycle
condition. For any r 6 R we can form the real line bundle IAI'"(E) defined by the same
open cover and transition maps
tea ldet goal ldetg&5| : Uog R>0 "-> GL(1,]R).
The fiber at p G M of this bundle consists of 1°-densities on Ep (see Subsection 2.2.4).

Definition 3.4.1. Let M be a smooth manifold and or > 0. The bundle of 1°-densities on M
is

IAIRI =IAI'°(TMl.
When T' l we will use the notation IAIN = IAI}w. We call IAIN the density bundle o f
M.

Denote by C°°(IAIM) the space of smooth sections of IAIN, and by C3°(IAIM) its
subspace consisting of compactly supported densities.
It helps to have local descriptions of densities. To this aim, pick an open cover of M
consisting of coordinate neighborhoods (Url . Denote the local coordinates on Uo by (we).
This choice of a cover produces a trivializing cover of TM with transition maps

330
T
( ) j
3565
1<i,j<n
7

where n is the dimension of M. Set 600 I det Tag I . A 1-density on M is then a collection
of functions nO( 6 cOO (Ua) related by
-1
(Sa5/15.
112 Lectures on the Geometry of Manzfolds

It may help to think that for each point p 6 Ua the basis exa1 , ..., of TpM spans an
infinitesimal parallelepiped and uo (p) is its "volume". A change in coordinates should be
thought of as a change in the measuring units. The gluing rules describe how the numerical
value of the volume changes from one choice of units to another.
The densities on a manifold resemble in many respects the differential forms of max-
imal degree. Denote by det T M = AdiMMTM the determinant line bundle of TM. A
density is a map
lJ, : C°°(det T M ) -> C°°(M),
such that 1u(fe) = l f l/i(e), for all smooth functions f : M -> R and all e 6
C°°(det T M ) . In particular, any smooth map (b M -> N between manifolds of the
same dimension induces a pullback transformation

(l)* : 0°O>(IAln) -> c°°(IAnm),

described by
(gb*,u)(6) = ,up(det (M . e) = ldetq15*| 1u(e), We 6 C°°(detTM).
Example 3.4.2. (a) Consider the special case M = ]R'"'". Denote by e l , ..., en the canonical
basis. This extends to a trivialization of TR" and, in particular, the bundle of densities
comes with a natural trivialization. It has a nowhere vanishing section ldvv,l defined by
ldv,,,l(€1 /\ AAn) = 1.

In this case, any smooth density on R" takes the form ,LL = f ld'un |, where f is some smooth
function on ]R"'. The reader should think of ldv'nl as the standard Lebesgue measure on R" .
If go : R" -> R" is a smooth map, viewed as a collection of n smooth functions
1 1
651 - Q51(CC 7°°°7$N)7 • • • ,sin - 7°°'7$N)7

then,

<9d)¢
(25*(ldvnll = det
( )
3111

(b) Suppose M is a smooth manifold of dimension m. Then any top degree form CU 6
ldvnl.

§2m*(M) defines a density lwl on M which associates to each section e 6 C°°(det TM)
the smooth function
QUO »-> lwas( e(a2) )|.
Observe that lwl = | - wl, so this map §2M(Ml -> @OO(IAlml is not linear.
(c) Suppose g is a Riemann metric on the smooth manifold M. The volume density defined
by g is the density denoted by ldvgl which associates to each e E C°°(det TM) the
pointwise length

ac »-> le(.)lg~
If (Um ( 3 ) ) is an atlas of M , then on each UO( we have top degree forms
d¢¢¢=dw1A»--Adg" 7
 Calculus on Manifolds 113

to which we associate the density ldac°OI. In the coordinates (51:23 ) the metric g can be
described as
.
..doZ
ga,.7 j
dgj0
.
®

m
II
TM
We denote by l9al the determinant of the symmetric matrix I.go lga;ij)1§i,j<_m- Then the
restriction of ldvgl to UO( has the description

ldVg I v lgol lda3ol.

The importance of densities comes from the fact that they are exactly the objects that
can be integrated. More precisely, we have the following abstract result.

Proposition 3.4.3. There exists a natural way to associate to each smooth manQ'ola' M a
linear map

07°(IAlm) -> R
M

uniquely defined by the following conditions.

(a) FM is invariant under do eomorphisms, i.e., for any smooth manifolds M, N of the
same dimension n, any dQ§"eomorphism cl) : M -> N, and any ,LL 6 05° (IAIM), we nave

,a_
'n
(b) /M is a local operation, i.e., for any open set U C M, and any ,u 6 o3°(IAIM) with
supp 1,L C U , we have

,u.
'u
(c) For any p 6 08° (R'nl we have

/Rn
aldun I /Rn
p(:13)d;u,

where in the right-hand-side stands the Lebesgue integral of the compactly supportedfunc-
tion p. FM
is called the integral on M .

Proof. To establish the existence of an integral we associate to each manifold M a collec-

tion of data as follows.

(i) A smooth partition of unity A C 08° (Ml such that Va 6 A the support supp a lies
entirely in some precompact coordinate neighborhood UQ, and such that the cover (UO, )
is locally finite.
(ii) For each Uo we pick a collection of local coordinates ( x ) , and we denote by id;ual
(n = dim M) the density on UO( defined by
.-I

lda3a I
(
II
114 Lectures on the Geometry of Manzfolds

For any ,u E 0°O(|A|), the product as, is a density supported in UQ, and can be written
as

04/1 = ;U#adaoc 7

where ua is some smooth function compactly supported on Uo, The local coordinates
allow us to interpret uo as a function on R". Under this identification Idwf; I corresponds
to the Lebesgue measure ldvnl on R", and uOt is a compactly supported, smooth function.
We set
IUa¢ld53ol.
VULG f]n

Finally, define
A

M
,u
/M
,u
def

2/
QUA UQ
up/,.

The above sum contains only finitely many nonzero terms since supp ,L/, is compact, and
thus it intersects only finitely many of the U' S which form a locally finite cover.
To prove property (a) we will first prove that the integral defined as above is independent
of the various choices: the partition of unity A C 0go>(M), and the local coordinates
ll";laeA-
• Independence of coordinates. Fix the partition of unity A, and consider a new collection
of local coordinates (ya) on each UQ. These determine two densities id:z;;l and respectively
Idyll |. For each ,LL 6 05°(IAIM) we have

co = aden any ldYo¢lv

where #3 #Z 6 05'°(UQ) are related by

The equality

uiildwol my Idyo I
R" Rn
is the classical change in variables formula for the Lebesgue integral.
¢Independence of the partition of unit y. LetA,8 C C§°(M) two partitions of unity on
M. We will show that
A 8

Form the partition of unity

/ / M M

A*8 : {0¢B; ( a , B ) € .A X 8}<05°(m).

Note that supp QB C Uog = Ua O UB- We will prove
A _A*8 8

/ /
M M / M
 Calculus on Manifolds 115

Let ,u E C`6'°(|A|m)- We can view as as a compactly supported function on R". We have

/
Up,
Q#
2/ B UaCR"
BQ#
2/
B UQ#
aBp,. (3.4.1)

Similarly

/U5
Be 2/a UG B
QB#- (3.4.2)

Summing (3.4.1) over a and (3.4.2) over B we get the desired conclusion.
To prove property (a) for a diffeomorphism QUO : M -> N, consider a partition of unity
A C C'§°(N). From the classical change in variables formula we deduce that, for any
coordinate neighborhood Ua containing the support of a 6 A, and any ,u 6 0gO>(IAI N) we
have

(¢*@)¢*u ow-
(b'1(Ua) UQ

The collection (l ( = a s)
QEA
forms a partition of unity on M. Property (a) now
follows by summing over a the above equality, and using the independence of the integral
on partitions of unity.
To prove property (b) on the local character of the integral, pick U C M, and then
choose a partition of unity 8 C 0golu) subordinated to the open cover IVBIBQQ3- For
any partition of unity A C C'§°(M) with associated cover IVaIQEA we can form a new
partition of unity A * 8 of U with associated cover Vol = Vo, n VB- We use this partition
of unity to compute integrals over U. For any density ,u on M supported on U we have

/ 2/
M
,LL
a Vo
co
22/
0GABGB V06
QB# ::
2/
0,B€.A*8 VQB
QB#
/
U
,u.

Property (c) is clear since, for M = Rn, we can assume that all the local coordinates chosen
are Cartesian. The uniqueness of the integral is immediate, and we leave the reader to fill
in the details.

3.4.2 Orientabilily and integration of dzjferentialforms

Under some mild restrictions on the manifold, the calculus with densities can be replaced
with the richer calculus with differential forms. The mild restrictions referred to above
have a global nature. More precisely, we have to require that the background manifold is
oriented.
Roughly speaking, the oriented manifolds are the "2-sided manifolds", i.e., one can
distinguish between an "inside face" and an "outside face" of the manifold. (Think of a
2-sphere in R3 (a soccer ball) which is naturally a "2-faced" surface.)
The 2-sidedness feature is such a frequent occurrence in the real world that for many
years it was taken for granted. This explains the "big surprise" produced by the famous
counter-example due to MObius in the first half of the 19th century. He produced a 1-sided
surface nowadays known as the MObius band using paper and glue. More precisely, he
116 Lectures on the Geometry of Manzfolds

Il

l l

` \
\
\
1
I
,r
/
/
/
/

Fig. 3.2 The Mobius band.

glued the opposite sides of a paper rectangle attaching arrow to arrow as in Figure 3.2. The
2-sidedness can be formulated rigorously as follows.
Definition 3.4.4. A smooth manifold M is said to be orientable if the determinant line
bundle det TM (or equivalently det T * M ) is trivializable.

We see that det T*M is trivializable if and only if it admits a nowhere vanishing sec-
tion. Such a section is called a volume form on M. We say that two volume forms W i and
W2 are equivalent if there exists f 6 C°°(M) such that
C02 étwl .
This is indeed an equivalence relation, and an equivalence class of volume forms will be
called an orientation of the manifold. We denote by O ' r ( M ) the set of orientations on
the smooth manifold M. A pair (orientable manifold, orientation) is called an oriented
manifold.
Let us observe that if M is orientable, and thus Or(Ml 73 (D, then for every point
p 6 M we have a natural map
Or(M) -> O'r(TpM), O'r(M) 9 or »-> orp 6 O'r°(TpM),
defined as follows. If the orientation or on M is defined by a volume form co, then wp 6
det Tp M is a nontrivial volume form on T p M , which canonically defines an orientation
o r , on T p M . It is clear that if W i and W2 are equivalent volume form then 0'rw1,p =
orw2 ,p-
This map is clearly a surjection because or_,vp = - o p , for any volume form CU.
Proposition 3.4.5. If M is a connected, orientable smooth manifold M , then for every
p 6 M the map
O'r(M) 9 or »-> o f p E Or(TpM)
is a bisection.
 Calculus on Manifolds 117

Proof. Suppose or and or' are two orientations on M such that O'T'p = o p . The function
M 9 q *-> 6(q) = o r / o r q 6 {d:1}
is continuous, and thus constant. In particular, 6(q) = 6(p) = 1, 'v'q 6 M.
If or is given by the volume form w and or' is given by the volume form w' , then there
exists a nowhere vanishing smooth function p : M -> R such that w' = pa. We deduce
sign »0(q) = €(q), Vq 6 M .
This shows that the two forms w I and w are equivalent and thus or = or' .
The last proposition shows that on a connected, orientable manifold, a choice of an
orientation of one of its tangent spaces uniquely determines an orientation of the manifold.
A natural question arises.
How can one decide whether a given manyL'old is orientable or not.
We see this is just a special instance of the more general question we addressed in Chap-
ter 2: how can one decide whether a given vector bundle is trivial or not. The orientability
question can be given a very satisfactory answer using topological techniques. However, it
is often convenient to decide the orientability issue using ad-hoc arguments. In the remain-
ing part of this section we will describe several simple ways to detect orientability.

Example 3.4.6. If the tangent bundle of a manifold M is trivial, then clearly TM is ori-
entable. In particular, all Lie groups are orientable.

Example 3.4.7. Suppose M is a manifold such that the Whitney sum 4 ® TM is trivial.
Then M is orientable. Indeed, we have
det(Bk ® TM) det Bk (X) det TM.
Both det Bk and det(Rk ® TM) are trivial. We deduce det TM is trivial since
~
det TM = det(Rk ® TM) (det Rel * (X)

This trick works for example when M 2" S". Indeed, let 1/ denote the normal line bundle.
The fiber of 1/ at a point p 6 S" is the 1-dimensional space spanned by the position vector
of p as a point in R", (see Figure 3.3). This is clearly a trivial line bundle since it has a
tautological nowhere vanishing section p »-> p 6 Vp- The line bundle 1/ has a remarkable
feature:
U ® TS" = IRT),-l-1
Hence all spheres are orientable.

Important convention. The canonical orientation on R" is the orientation defined by

the volume form day1 /\ ---/\ dat", where x1, ..., x" are the canonical Cartesian coordinates.
The unit sphere S" C lR"+1 is orientable. In the sequel we will exclusively deal with
its canonical orientation. To describe this orientation it suffices to describe a positively
oriented basis of det TpM for some p 6 S". To this aim we will use the relation
IRR-l-1 up ® TpS"
i v
118 Lectures on the Geometry of Manzfolds

I
I
| /
/
\
\ I /
\ | /
\ | /
\
\ | /
X.- - _ _ / _
*-
\ I / "\u.

< _ " L _ _ -
°-~.

, ' \
- - - _ ; . _'°°~.,
>
\
I \
I \

I `\
I \
L
I " .
I
I
I

.
l/

Fig. 3.3 The normal line bundle to the round sphere.

An element cu E det TpS" defines the canonical orientation if p-» A w 6 det Rn-i-1 defines
the canonical orientation of R"+*. Above, by p-» we denoted the position vector of p as a
point inside the Euclidean space lR"+1. We can think of 15' as the "outer" normal to the
round sphere. We call this orientation outer normaljirst. When n = l it coincides with the
counterclockwise orientation of the unit circle S1 .

Lemma 3.4.8. A smooth manifold M is orientable zfand only of there exists an open cover
lUa)0 E A ) and local coordinates (5131 , ..., kg) on Uo, such that
/
a;
det l 35135
j
> 0 OHU0 VWUQ. (3.4.3)
\ I

Proof. 1. We assume that there exists an open cover with the properties in the lemma, and
we will prove that det T*M is trivial by proving that there exists a volume form.
Consider a partition of unity 8 C c3° (Ml subordinated to the cover (UoJQQA i.e., 9

there exists a map SO : 8 -> A such that

suppl C U(/)(B) VI E 8.
Define
w
II
a s

Bongo(,B))
B
where for all a Q A we define wa := do1 A A dog. The form w is nowhere vanishing ---
since condition (3.43) implies that on an overlap U O re Uom the forms was, ..., com ---
differ by a positive multiplicative factor.
2. Conversely, let CU be a volume form on M and consider an atlas (Uo (x13))~ Then
w INa= ,L/'fadac1 /\ - - - /\ dog,
 Calculus on Manifolds 119

where the smooth functions ,Uo are nowhere vanishing, and on the overlaps they satisfy the
gluing condition

,UB
A 015 det I I I I I I I I I I I
.
la

A permutation up of the variables 51:1 , ..., x will change day1 A ---

A dog by a factor €((/2)
so we can always arrange these variables in such an order so that /Ja > 0. This will insure
the positivity condition

A as > 0.

The lemma is proved.

We can rephrase the result in the above lemma in a more conceptual way using the
notion of orientation bundle. Suppose E -> M is a real vector bundle of rank r on the
smooth manifold M described by the open cover l U a l a € A and gluing cocycle 7

gg0 : U05 -> GL(r, R).

The orientation bundle associated to E is the real line bundle ® (E) -> M described by
the open cover lUa)0EA9 and gluing cocycle

650 := sign det 950 : U0g -> R* = GL(l,]R).

We define orientation bundle ( 9 M of a smooth manifold M as the orientation bundle asso-

ciated to the tangent bundle of M, ® M @(TM).
The statement in Lemma 3.4.8 can now be rephrased as follows.

Corollary 3.4.9. A smooth manifold M is orientable If and only If the orientation bundle
@M is trivializable.

From Lemma 3.4.8 we deduce immediately the following consequence.

Proposition 3.4.10. The connected sum of Iwo orientable manifolds is an orientable man-
fold.

Exercise 3.4.11. Prove the above result.

Using Lemma 3.4.8 and Proposition 2.2.75 we deduce the following result.

Proposition 3.4.12. Any complex manifold is orientable. In particular, the complex Grass-
mannians Grk ( m l are orientable.

Exercise 3.4.13. Supply the details of the proof of Proposition 3.4. 12.
120 Lectures on the Geometry of Manzfolds

The reader can check immediately that the product of two orientable manifolds is again
an orientable manifold. Using connected sums and products we can now produce many
examples of manifolds. In particular, the connected sums of g tori is an orientable manifold.
By now the reader may ask where does orientability interact with integration. The
answer lies in Subsection 2.2.4 where we showed that an orientation or on a vector space
V induces a canonical, linear isomorphism 2O1" : det V* -> IAIV, see (2.2.l3).
Similarly, an orientation or on a smooth manifold M defines an isomorphism
ZOO' C°°(detT*m) -> C°°(IAlml.
For any compactly supported differential form Ld on M of maximal degree we define its
integral by

z0,,.w.
'm 'm
We want to emphasize that this definition depends on the choice of orientation.
We ought to pause and explain in more detail the isomorphism 2O7' : cOO (det T * M ) ->
cOO (IAIN). Since M is oriented we can choose a coordinate atlas ( u , I ) such that
I' Q ii `l
det > 0, n dim M , (3.4.4)
1§,'i,j§M
and on each coordinate patch UQ the orientation is given by the top degree form doa
dol A---Adccg.
A top degree differential form LU is described by a collection of forms
We padsa Po E CYOO(U0),
and due to the condition (3.4.4) the collection of densities
Ila I Padua' E C°°(U0,, I III lm)
satisfy ,ua = ,up on the overlap Uog. Thus they glue together to a density on M, which is
precisely ZOT'LU.

Example 3.4.14. Consider the 2-form on R3, w = :cdy /\ do, and let S2 denote the unit
sphere. We want to compute j S2 ( ; I $2 , where S2 has the canonical orientation. To com-
pure this integral we will use spherical coordinates (T, go, H). These are defined by (see
Figure 3.4.)
ac r sin go COS H
y 7" sin up sin 0
Z TCOS up

At the point p (1,0,0) we have

c'91,"=3,,=15', 3023y(p:-(9Z,
SO that the standard orientation on S2 is given by d o /\ do. On S2 we have T' 1 and
dr E 0 so that

Andy /\ do 152 = sin go COS 9 (cos 9 sin cpd -|- sin 9 COS godc,o)) /\ (- sin (p)dgo
= sin3 go COS2 9d(p /\ d9.
 Calculus on Manifolds 121

II
z

up P
r I
I
I

L
I

O .<
\
\
I
I /
/

\ I /
\ | /
\ /
9 \
\ \
I
| /
/

\ | /
\ /
\ I /
v

Fig. 3.4 Spherical coordinates.

The standard orientation associates to this form de density sin3 (ac:os2 0ldgod0l, and we
deduce
11' 2tr
sin3 up cos2 0ld(pd0| sing cpdcp l
COS2 0d6
lm] x [0,27/] 0 0

_ 47r
volume of the unit ball 83 C R3.
' 3
As we will see in the next subsection the above equality is no accident.

Example 3.4.15 (Invariant integration on compact Lie groups). Let G be a compact,

connected Lie group. Fix once and for all an orientation on the Lie algebra LG. Consider
a positively oriented volume element w 6 det LG. We can extend w by left translations
to a left-invariant volume form on G which we continue to denote by w. This defines an
orientation, and in particular, by integration, we get a positive scalar
U
II

CU.
G

Set dVG = in
e
so that

we 1. (3.4.5)
G

The differential form dVG is the unique left-invariant n-form (n = dim G) on G satisfying
(3.4.5) (assuming a fixed orientation on G). We claim dVG is also right invariant.
To prove this, consider the modular function G 9 h »-> A(h) 6 R defined by

Rfldvg ) A(h)dVG.
122 Lectures on the Geometry of Manzfolds

The quantity A(h) is independent of h because RhdVG is a left invariant form, so it has to
be a scalar multiple of dVG. Since (R;»,1/'L2)* = (R/12 Rhll* = Rh1Rh2 we deduce

A(h1h2) = A(/21lA(h2) Wu, 122 6 G.

Hence h +-> Ah is a smooth morphism

G -> (M {0},~).

Since G is connected A(G) C R+, and since G is compact, the set A(G) is bounded.
If there exists QUO Q G such that A(x) go 1, then either A(x) > 1, or A(a:-1) > 1, and
in particular, we would deduce the set lAl5'3nllne Z is unbounded. Thus A E 1 which
establishes the right invariance of dVG.
The invariant measure dVG provides a very simple way of producing invariant objects
on G. More precisely, if T is tensor field on G, then for each so 6 G define

To ((Lg)*T)mdVG(9)-
G

Then as »-> T, defines a smooth tensor field on G. We claim that T is left invariant. Indeed,
for any h G G we have

(Lhl*T
/
G(Lh)*((L9)*TldVG(9) G((Lhg)*T)dVG(9l

-TZ
u hg

/
G
(Lu)*TLh_1dVG(fu,) (Lh_1dVG Z

If we average once more on the right we get a tensor

GEQH-> ( (Rg)i' )¢zvG,

G

which is both left and right invariant.

Exercise 3.4.16. Let G be a Lie group. For any X G LG denote by ad(X) the linear map
LG->»CG defined by

LG 9 Y»-> [x,y] QLG.

(a) If w denotes a left invariant volume form prove that V/XGLG

LxW = - t r ad(X)w.

(b) Prove that if G is a compact Lie group, then tr a d ( X ) = 0, for any X Q LG.
 Calculus on Manifolds 123

3.4.3 Stokes' formula

The Stokes' formula is the higher dimensional version of the fundamental theorem of cal-
culus (Leibniz-Newton formula)
b
if f(1») f(&),

where f : [0, b] -> R is a smooth function and df = f ' (t)dt. In fact, the higher dimensional
formula will follow from the simplest l-dimensional situation.
We will spend most of the time finding the correct formulation of the general version,
and this requires the concept of manifold with boundary. The standard example is the lower
half-space
n
(;u1,...,a:") Q R" ;ac
1 _ 0}.
<

Definition 3.4.17. A smooth manifold with boundary of dimension n is a topological space

with the following properties.

(a) There exists a smooth n-dimensional manifold m- that contains M as a closed subset.
(b) The interior of M, denoted by M0, is non empty.
(c) For each point p 6 AM := M \ M0, there exist smooth local coordinates (x1, ..., ;z:" )
defined on an open neighborhood N of p in such that M
(01) M O O N {q€N; :121(q) < 0, }.
(02) <9MnN {:v1=0}-

The set $M is called the boundary of M. A manifold with boundary M is called

orientable if its interior M0 is orientable.

Example 3.4.18. (a) A closed interval I = [a, b] is a smooth 1-dimensional manifold with
boundary <91 = {a, b}. We can take m- = R
(b) The closed unit ball B3 C R3 is an orientable manifold with boundary 91-33 = SQ.
We can take m- = R3.
(c) Suppose X is a smooth manifold, and f : X -> R is a smooth function such that
0 6 R is a regular value of f , i.e.,

f(33) 0 \.
J

df(w) ¢ 0.

Define M := {a: 6 X ; f (az) < _ 0 A simple application of the implicit function theorem
shows that the pair (X, M) defines a manifold with boundary. Note that examples (a) and
(b) are special cases of this construction. In the case (a) we take X = R and f (as) =
( x - a)(;u - b), while in the case (b) we take X = R3 and f(x, y, Z = (x2 -|- y2 -|- z2) - 1.
124 Lectures on the Geometry of Manzfolds

Definition 3.4.19. Two manifolds with boundary M1 C Ml, and M2 C M2 are said to be
dweomorphic if, for every i = 1, 2 there exists an open neighborhood Up of M in m and
a diffeomorphism F : U1 -> U2 such that F(M1) = M2.

Exercise 3.4.20. Prove that any manifold with boundary is diffeomorphic to a manifold
with boundary constructed via the process described in xample 3.4. 18(c).

Proposition 3.4.21. Let M be a smooth manu'old with boundary. Then its boundary AM is
also a smooth manifold of dimension dim AM = dim M - 1. Moreover, f M is orientable,
then so is its boundary.

The proof is left to the reader as an exercise.

'important convention. Let M be an orientable manifold with boundary. There is a
(non-canonical) way to associate to an orientation on M0 an orientation on the boundary
311/I. This will be the only way in which we will orient boundaries throughout this book. If
we do not pay attention to this convention then our results may be off by a sign.
We now proceed to described this induced orientation on AM. For any p e AM
choose local coordinates all, ...,:v" as in Definition 3.4.17. Then the induced orientation
of Tp3M is defined by
Adm? /\ - - - A d m " 6 detTp3M, 6 it,
where 6 is chosen so that for ac1 < 0, i.e., inside M, the form
ed:c1 Ada22A---Adm"
is positively oriented. The differential do1 is usually called an outer conformal since ac1
increases as we go towards the exterior of M. -day1 is then the inner eonormal for anal-
ogous reasons. The rule by which we get the induced orientation on the boundary can be
rephrased as
{outer conorrnal} A {orientation on boundary} = {orientation in the interior}.
We may call this rule "outer normal first" for obvious reasons.

Example 3.4.22. The canonical orientation on Sn C ]R"+1 coincides with the induced
orientation of Sn-I-1 as the boundary of the unit ball B"+1.

Exercise 3.4.23. Consider the hyperplane Ht C R" defined by the equation { xi = 0}.
Prove that the induced orientation of H ii as the boundary of the half-space H n,z' _ 0}
+ = {%>
is given by the (n - 1)-form (-1)ida31 /\- - -AJ A- - - dx" where, as usual, the hat indicates
a missing term.

Theorem 3.4.24 (Stokes formula). Let M be an oriented n-dimensional manifold with

boundary AM and w 6 Q'I1-1 ( M ) a compactly supported form. Then
r r
w.
_'am
In the above formula d denotes the exterior derivative, and AM has the induced orientation.
 Calculus on Manifolds 125

Proof. Via partitions of unity the verification is reduced to the following two situations.
Case 1. The n - 1)-form w is compactly supported in Rn. We have to show

do 0.
Rn

It suffices to consider only the special case

w f lQ;)da¢2 /\ -- a /\ do" 7

where f (az) is a compactly supported smooth function. The general case is a linear combi-
nation of these special situations. We compute

since
fR'n fun
of
3331
daz1 /\ /\ do"
Rn 1 It 511 )
R
dayl day2 /\ /\ do" 0,

of dat
3:13
1 f(00»132» ...,:z:") - f ( -on,;z:2 , ...,:z3"),
R

and f has compact support.

Case 2. The n 1)-form w is compactly supported in Hn . Let

f'L(:c)d:c1 A - - - A c a ' A - - - A d m " .

3
II

Then

1-1-1
do ( 1 of d:c1 A - - - A d m " .
) 33%

One verifies as in Case 1 that

For Z 1 we have
L 31". d 1 A
332'L 513
A day" 0 for i go 1.

n
@f
33121 day 1 /\ /\ das"
n 1
(
0
god ) d:r32 A /\ dat"
+ OO

lf(0,a32,...,ac") - f ( -on,;z:2,..., ;z:"')l d:I:2 / \ - - - A d m "

Rn 1

f(0,x2, ;u")d;u2 /\ /\ do" w.

Rn-1 3 n

The last equality follows from the fact that the induced orientation on a TL
is given by
day2 A ---
A do". This concludes the proof of the Stokes formula.

Remark 3.4.25. Stokes formula illustrates an interesting global phenomenon. It shows

that the integral FM
do is independent of the behavior of w inside M. It only depends on
the behavior of CU on the boundary.
126 Lectures on the Geometry of Manzfolds

Example 3.4.26.
ctr
acdy /\ do = do dy do vol (83) 3°
fs2 f83

Remark 3.4.27. The above considerations extend easily to more singular situations. For
example, when M is the cube [0, 1]'" its topological boundary is no longer a smooth mani-
fold. However, its singularities are inessential as far as integration is concerned. The Stokes
formula continues to hold
w Vi E Q"-1<I">.
'/71
'of
The boundary is smooth outside a set of measure zero and is given the induced orientation:
"outer (co)normal first". The above equality can be used to give an explanation for the
terminology "exterior derivative" we use to call d. Indeed if CU Q Qn' 1 (Rn) and 1h = [0, h]
then we deduce
do |a:=0= lim h-n LU. (3.4.6)

When n =l this is the usual definition of the derivative.

Example 3.4.28. We now have sufficient technical background to describe an example of

vector bundle which admits no flat connections, thus answering the question raised at the
end of Section 3.3.3.
Consider the complex line bundle L n - > S 2 constructed in Example 2.1.38. Recall that
Ln is described by the open cover
S2=U0LJU1,U0 S2 \ {South pole}, U1 S2 \ {North pole},
and gluing cocycle
Z n
910 : U0 re U1 - (C*, 910(Zl I Z 901(Zl -17

where we identified the overlap U0 Fl U1 with the punctured complex line (c*
A connection on Ln is a collection of two complex valued forms We 6 Q1 (U1l (X) (C,
C01 6 Q1 (U1) (X) (I, satisfying a gluing relation on the overlap (see Example 3.3.7)

dg10 do
W1 = - -|- LU() = n Wo-
910 Z

If the connection is flat, then

dog = 0 on U0 and d l = 0 on U1-
Let E+ be the Equator equipped with the induced orientation as the boundary of the north-
ern hemisphere, and E' the equator with the opposite orientation, as the boundary of the
southern hemisphere. The orientation of E+ coincides with the orientation given by the
form dH, where Z = exp('i9l.
We deduce from the Stokes formula (which works for complex valued forms as well)
that

E+
We 0
/E+
W1
/ E
$01 0.
 Calculus on Manifolds 127

On the other hand, over the Equator, we have

do
L01 - We n n'id9,
Z

from which we deduce

.
0 We - w1 = no do = 2n7ri !!!
E+ E+
Thus there exist no flat connections on the line bundle Ln, n go 0, and at fault is the gluing
cocycle defining L. In a future chapter we will quantify the measure in which the gluing
data obstruct the existence of flat connections.

3.4.4 Representations and characters of compact Lie groups

The invariant integration on compact Lie groups is a very powerful tool with many uses.
Undoubtedly, one of the most spectacular application is Hermann Weyl's computation of
the characters of representations of compact semi-simple Lie groups. The invariant inte-
gration occupies a central place in his solution to this problem.
We devote this subsection to the description of the most elementary aspects of the rep-
resentation theory of compact Lie groups.
Let G be a Lie group. Recall that a (linear) representation of G is a left action on a
(finite dimensional) vector space V
G X V->V (g,v)»->T(g)v€V,

such that the map T(g) is linear for any g. One also says that V has a structure of G-module.
If V is a real (respectively complex) vector space, then it is said to be a real (respectively
complex) G-module.

Example 3.4.29. Let V C". Then G GL(n, (C) acts linearly on V in the tautological
manner. Moreover V*, v ® k , A""V and SW are complex G-modules.

Example 3.4.30. Suppose G is a Lie group with Lie algebra LG - T1G. For every g QG,
the conjugation
Cg:G->G, h+-> Cg(h )=g/ zg 7

sends the identity 1 Q G to itself. We denote by Add : LG -> L the differential of

h +-> Cg(/1) at h 1. The operator Adg is a linear isomorphism of LG, and the resulting
map
G I-> Alltllgl, g +-> Add ,
is called adjoint representation of G.
For example, if G = SO(n), then LG = (n), and the adjoint representation Ad
SO(n) -> Aut(&(77»)l, can be given the more explicit description
Ad
QM) 9 X I ( ) gxg Q @(n), Vg 6 SO(n).
128 Lectures on the Geometry of Manzfolds

Exercise 3.4.31. Let G be a Lie group, with Lie algebra LG. Prove that for any X, Y QLG
we have
d
dt l=@ Adexp(x)(y)
llllllllll

Z ad(X)Y = [X, ye.

Definition 3.4.32. A morphism of G-modules VI and V2 is a linear map L : VI -> VI such

that, for any g Q G, the diagram below is commutative, i.e., T2(g)L = LTD (g).

VI L > VI
T1 (g) T2-lgl

VI L > V2
The space of morphisms of G-modules is denoted by HOMG(V1, VI). The collection of
isomorphisms classes of complex G-modules is denoted by G - Mod.

If V is a G-module, then an invariant subspace (or submodule) is a subspace U C V

such that T(g) (U) C U, Vg 6 G. A G-module is said to be irreducible if it has no invariant
subspaces other than {0} and V itself.

Proposition 3.4.33. The direct sum " ® ", and the tensor product " ® " delne a structure of
semi-ring with l on G - Mod. 0 is represented by the null representation { 0 } , while l is
represented by the trivial module G -> Aut (QI, g +-> 1.

The proof of this proposition is left to the reader.

Example 3.4.34. Let T : G -> Aut lu'il (i = 1, 2) be two complex G-modules. Then U1
is a G-module given by (g, 'u,*) »-> T1 (g-1)l u*. Hence Hom (UI, UQ) is also a G-module.
Explicitly, the action of g G G is given by

(g, L)
I
I > T2l9lLT1lg-1l VL€Hom(U1 U2).
We see that HOMG(U1, UQ) can be identified with the linear subspace in Hom IU17 UQ)
consisting of the linear maps U1 -> U2 unchanged by the above action of G.

Proposition 3.4.35 (Weyl's unitary trick). Let G be a compact Lie group, and V a com-
plex G-module. Then there exists a Hermitian metric h on V which is G-invariant, i.e.,
h(gv1,gv2) = h(u1,v2), V u 1 , 0 2 G V.

Proof. Let h be an arbitraxy Hermitian metric on V. Define its G-average by

In, u)
..

h(gu, 9v)dVG(9),
II

where dVG (g) denotes the normalized bi-invariant measure on G. One can now check
easily that h is G-invariant.
 Calculus on Manifolds 129

In the sequel, G will always denote a compact Lie group.

Proposition 3.4.36. Let V be a complex G-module and h a G-invariant Hermitian metric.

If U is an invariant subspace of V then so is UJ_, where " J _ " denotes the orthogonal
complement with respect to h.

Proof. Since h is G-invariant it follows that, Vg Q G, the operator T(g) is unitary,

T(g)T*(9) = iv. Hence, T*(9) = T'1(g) = T(9'1), Vg € G-
I f x Q U , thenforallu E U, andVg Q G
h(T(g);l3, u) i h((x, T*(g)ul h(a:, T(g-1)'u,) = 0.
Thus T(g);13 € UJ- so that UJ- is G-invariant.
9

Corollary 3.4.37. Every G-module V can be decomposed as a direct sum of irreducible

ones.

If we denoted by Ir'r°(G) the collection of isomorphism classes of irreducible G-

modules, then we can rephrase the above corollary by saying that Irr(G) generates the
semigroup (G - Mod, e).
To gain a little more insight we need to use the following remarkable trick due to Isaac
Schur.

Lemma 3.4.38 (Schur lemma). Let VI, V2 be two irreducible complex G-modules. Then
l if VI = V2
dimqj HOMG(V1, VI)
0 if VI go V2 •

Proof. Let L 6 HOMG(V1, V2). Then her L C VI is an invariant subspace of VI. Similarly,
Range (L) C V2 is an invariant subspace of V2. Thus, either her L = 0 or her L = V1.
The first situation forces RangeL ;é 0 and, since $2 is irreducible, we conclude
RangeL = V2- Hence, L has to be in isomorphism of G-modules. We deduce that, if
~
VI and V2 are not isomorphic as G-modules, then HOMG(V1, V2) = {0}.
Assume now that VI = V2 and S : V1 -> V2 is an isomorphism of G-modules. Accord-
ing to the previous discussion, any other nontrivial G-morphism L : VI -> V2 has to be an
isomorphism. Consider the automorphism T = S-1L : $1 -> V1. Since $1 is a complex
vector space T admits at least one (non-zero) eigenvalue A.
The map A]1V1 - T is an endomorphism of G-modules, and her (Ally, - T) go 0.
Invoking again the above discussion we deduce T E A1V1, i.e., L =
AS. This shows
dirnHornG(V1, V2) = 1.

Schur's lemma is powerful enough to completely characterize S1 Mod, the repre-

sentations of S1.

Example 3.4.39 (The irreducible (complex) representations of S1). Let V be a com-

plex irreducible S1-module
S1 X V9(eW,u)l-->T0v€V,
130 Lectures on the Geometry of Manzfolds

-
where TO1 TO2 = T@1+92 mod 2'rr. In particular, this implies that each T9 is an S1-
automorphism since it obviously commutes with the action of this group. Hence T9 =
A(0)]1V which shows that d i V = 1 since any 1-dimensional subspace of V is S1-
invariant. We have thus obtained a smooth map
A:S1->(C* 7

such that
Aw" girl = >(e")A(@@>,
Hence A : Sl -> KI* is a group morphism. As in the discussion of the modular function we
deduce that l,\l E 1. Thus, A looks like an exponential, i.e., there exists a Q R such that
(vergCy./)
A(@i*'> = exp('iaH), we 6 R.
Moreover, exp(27r'ia) = 1, so that a Q Z.
Conversely, for any integer n 6 Z we have a representation
'0 . 0
Sl &l Aut (el (6'a ,zl +-> ezn z.
The exponentials exp('£n0) are called the characters of the representations ,On-

Exercise 3.4.40. Describe the irreducible representations of T"-the n-dimensional torus.

Definition 3.4.41. (a) Let V be a complex G-module, g +-> T(g) 6 Aut ( V ) . The charac-
ter of V is the smooth function
Xv : G -> C, xv(9) trT(g). ==
(b) A elassfunction is a continuous function f : G -> (C such that
_1 _
f(h9h ) - f(9) V g , h € G .
(The character of a representation is an example of class function.)

Theorem 3.4.42. Let G be a compact Lie group, U1, U2 complex G-modules and XU their
characters. Then the following hold.
(0) XU1@U2 = XU1 -|- XU2' XU1®U2 = XU1 • XU1-
(19) XU,ll) = dim Ui-
(c) XU.* = Y M -the complex conjugate Of XU.
(d)

XU (g)dvG (go : dim UG 7

G
where UG denotes the space of G-invariant elements of Up,
HG {as E u , Hz: = T¢l9)93 we E G}.
(6)

XU1 (9) 'Yi/5 (9)dVG(9) = dim HOMG(U2, U1)-

G
 Calculus on Manifolds 131

Proof. The parts (a) and (b) are left to the reader. To prove (c), fix an invariant Hermitian
metric on U = Ui- Thus, each T(g) is a unitary operator on U. The action of G on U* is
given by T(9-1)*. Since T(g) is unitary, we have T(g-1)* = T(g). This proves (c).
Proof of (d). Consider

P : U -> U, Pa T(g)u dvG(g).

G
Note that PT(h) = T(h)P, WE E G, i.e., P E HOMG(U, U). We now compute

T(h)Pu
G
T(h9)u dVG(9)
G
TWO Rh-1dVG('y),

T('y)'u, dvG('y) = Pa.

G
Thus, each Pa is G-invariant. Conversely, if as 6 U is G-invariant, then

Pa:
/
G
T(9):vd9
/ G
:IJ dVG (9) Xv

i.e., UG = Range P. Note also that P is a projector, i.e., P2 = P. Indeed,

Gaga 'U»dvG(9 ) -- P
u
P2u T
l9)p G
P
G
Hence P is a projection onto UG, and in particular

d i l l ; UG = t r P trT(g) dvG(g) Xu(g) dvG(g).

G G
Proof of (6).

G
XU1 '36-U2dVGl9l
G
XU1 -xu;dvG(9l G
XU1®U; dvG(9l
G
XHom(U2,U1)

ding (Hom (up, u1))G dim() HOIHG(U2, U1),

since Home coincides with the space of G-invariant morphisms.

Corollary 3.4.43. Let U , V be irreducible G-modules. Then

l ,U = V
~
XU» XV) XU °Yvd9 = 61/v 0 ,U %V
G

Proof. Follows from Theorem 3.4.42 using Schur's lemma.

Corollary 3.4.44. Let U , V be two G-modules. Then U 'E V if and only z f x u = XV.

Proof. Decompose U and V as direct sums of irreducible G-modules

V e
€91(Nj)-

Hence XU = Z:MZx\ and Xv = 2 nix . The equivalence "representation" ~: >»

"characters" stated by this corollary now follows immediately from Schur's lemma and the
previous corollary.
132 Lectures on the Geometry of Manzfolds

Thus, the problem of describing the representations of a compact Lie group boils down
to describing the characters of its irreducible representations. This problem was completely
solved by Hermann Weyl, but its solution requires a lot more work that goes beyond the
scope of this book. We will spend the remaining part of this subsection analyzing the
equality (d) in Theorem 3.4.42.
Describing the invariants of a group action was a very fashionable problem in the sec-
ond half of the nineteenth century. Formula (d) mentioned above is a truly remarkable
result. It allows (in principle) to compute the maximum number of linearly independent
invariant elements.
Let V be a complex G-module and denote by Xv its character. The complex exterior al-
gebra ATV* is a complex G-module, as the space of complex multi-linear skew-symmetric
maps
V X X V->(C.

Denote by bglvl the complex dimension of the space of G-invariant elements in A/<=v*.
One has the equality

be) XA'°V*dVGl9)-
G
These facts can be presented coherently by considering the Z-graded vector space
J; M ¢= @ Akinv V*
k

Its Poincaré polynomial is

P3;(v)ltl t'"6Z(V) 15kXA1¢v* dVG

G
To obtain a more concentrated formulation of the above equality we need to recall some
elementary facts of linear algebra.
For each endomorphism A of V denote by (7klAl the trace of the endomorphism
MA : AW -> AW.
Equivalently, (see Exercise 2.2.25) the number ok1.I(A) is the coefficient of tk in the charac-
teristic polynomial
cHIlI = d€tl]1v -|- tA).

Explicitly, 0'klAl is given by the sum

O'k,lAl det I aiai5 ) (n = dim V).

1<i1~ ° ° i k < N

If g 6 G acts on V b y T ( g ) , then g acts on Ate* b y A"'T(9-1)* = A1"T(g). (We

implicitly assumed that each T (g) is unitary with respect to some G-invariant metric on
V.) Hence

XAW* = U1<:lT(9)l~ (3.4.7)

 Calculus on Manifolds 133

We conclude that

P3;(v)ltl IN; ¢'<=0,(T(9> )dvG(g)

G G
dell ]iv + t T ( g ) ) dvG(g). (3.4.8)

Consider now the following situation. Let V be a complex G-module. Denote by A,,1V the
space of R-multi-linear, skew-symmetric maps
V X X V->]R.

The vector space A,,1V* is a real G-module. We complexity it, so that A; V ®(C is the space
of R-multi-linear, skew-symmetric maps

V x - ~ - x V ->(C,
and as such, it is a complex G-module. The real dimension of the subspace gklvl of G-
invariant elements in Akv* will be denoted by bk(V), so that the Poincaré polynomial of
J;(v) = €BIlg3k is

Pf1:(v) (t)
Ikb7°klyl.
k

On the other hand, bk (V) is equal to the complex dimension of Ak V* (X) C. Using the results
of Subsection 2.2.5 we deduce
A;v (X) (C
r a
Age* (Xm: A;V*
k
laB+y=kA;V* (X) AW*I • (3.4.9)

Each of the above summands is a G-invariant subspace. Using (3.4.7) and (3.4.9) we
deduce

Pa;(v) (t) 07fz ( T ( g ) )(ti(T(9) )t¢+j avG(gl

k Gi+j k

det( iv -|- tT(g)l dell iv + t T ( g ) l dvG(g) (3.4.10)

'detl NV + tT (g ))1 2 dog).

G

We will have the chance to use this result in computing topological invariants of mani-
folds with a "high degree of symmetry" like, e.g., the complex Grassmannians.

3.4.5 Fibered ealeulus

In the previous section we have described the calculus associated to objects defined on a
single manifold. The aim of this subsection is to discuss what happens when we deal with
an entire family of objects parameterized by some smooth manifold. We will discuss only
the fibered version of integration. The exterior derivative also has a fibered version but its
true meaning can only be grasped by referring to Leray's spectral sequence of a libration
and so we will not deal with it. The interested reader can learn more about this operation
from [55], Chapter 3, Sec. 5.
134 Lectures on the Geometry of Manzfolds

Assume now that, instead of a single manifold F, we have an entire (smooth) family of
them (Fe)be B . In more rigorous terms this means that we are given a smooth fiber bundle
p : E -> B with standard fiber F.
On the total space E we will always work with split coordinates ( a , by ), where w)
are local coordinates on the standard fiber F, and (yj ) are local coordinates on the base B
(the parameter space).
The model situation is the bundle
p
E=R'<xRm£>1Re.m Z B, (x,y) I
>y.
We will first define a fibered version of integration. This requires a fibered version of
orientability.

Definition 3.4.45. Let p : E -> B be a smooth bundle with standard fiber F. The bundle
is said to be orientable if the following hold.

(a) The manifold F is orientable,

(b) There exists an open cover (Ua), and trivializations P'1(Ual > F x UQ, such that
the gluing maps
W 5 o \ P a 1 : F x Uag F X U05 IU0B = UQ VW Url
are fiberwise orientation preserving, i.e., for each y G Ugh, the diffeomorphism
F9f F
preserves any orientation on F .

Exercise 3.4.46. If the base B of an orientable bundle p : E -> B is orientable, then so is

the total space E (as an abstract smooth manifold).

Important convention. Let p : E -> B be an orientable bundle with oriented basis

B. The natural orientation of the total space E is defined as follows.
If E F X B then the orientation of the tangent space T(f,b)E is given by QF X
I

we, where we 6 det TfF (respectively we 6 det TbB) defines the orientation of TfF
(respectively TbB).
The general case reduces to this one since any bundle is locally a product, and the
gluing maps are fiberwise orientation preserving. This convention can be briefly described
as
orientation total space = orientation fiber /\ orientation base.
The natural orientation can thus be called the fiber-first orientation. In the sequel all ori-
entable bundles will be given the fiber-first orientation.

Let p : E -> B be an orientable fiber bundle with standard fiber F.

Proposition 3.4.47. There exists Cl linear operator

p* : Q;ptlE) -> Qgpt (B) , T' = dim F,

E/B
 Calculus on Manifolds 135

uniquely defined by its action on forms supported on domains D of split coordinates

~
D = RR x Run 2> Run , (33; y) »-> y-
I f s = fda! /\ dy'], f 6 CCpt(]R"+M), then

0 7 III ¢ T'
E/B (IR f d f I l dyJ 7 III = T' •

The operator I5/8 is called the integration-along-fibers operator

The proof goes exactly as in the non-parametric case, i.e., when B is a point. One shows
using partitions of unity that these local definitions can be patched together to produce a
well defined map

QZpt -> Q;ptlBl°

E/B
The details are left to the reader.

Proposition 3.4.48. Let p : E -> B be an orientable bundle with an T-dimensional stan-

dardjiber F. Then for any LU 6 Qe*pt ( E ) and 17 6 welpt(B) such that deg w+deg n = dim E
we have
dEuJ = l-1l7dB w.
E/B
If B is oriented and CU 17 at as above then

f
E
CU I\p*('7)
I(/ )B E/B
w Av- (Fubini)

The last equality implies immediately the projection formula

p*(w /\ PW = P ) /\ 77- (3.4.11)

Proof. It suffices to consider only the model case

p:E=Rx]RM->]RM=B, (atgy) 3, y,
and w fdail / \ b Y ] . Then
a . @f dat] /\ dye /\ dye.
dew f ax' Adatl A d 3 / J + (-1>III
3:6z
ii j byJ

/E/B
dEad
(f Rt Hz
@f
35
div /\ do] -|- (-1)|I|
(L
The above integrals are defined to be zero if the corresponding forms do not have degree
@f . d I
é9y3 QUO
/\ dye /\ dyJ.

T.
Stokes' formula shows that the first integral is always zero. Hence

E/B
dEw I (-1) I/I 3
<9?/i
(
The second equality is left to the reader as an exercise.
j
do]
) A dye A dy'I = (-1>*d8
E/B
w.
136 Lectures on the Geometry of Manzfolds

Exercise 3.4.49 (Gelfand-Leray). Suppose that p : E -> B is an oriented libration, wE

is a volume form on E, and we is a volume form on B.
(a) Prove that, for every b 6 B, there exists a unique volume form W E / B on Et p-1lbl
with the property that, for every as 6 Et, we have
wE(a:) = wg/B(a3) /\ (p*w8)(a3) 6 AdiMET;E.
This form is called the Gelfand-Leray residue of wE rel p.
(b) Prove that for every compactly supported smooth function f : E -> R we have

LP*lf°"Ellb Cl Et
f¢<JE/8
) WB(b),Vb E B,

p
/E
freE
I(/
B Eb
f¢*JE/B
) w8-

(c) Consider the libration R2 -> R (33, y) »->t ax -I-by,a2+b2 ¢ 0. Compute the
Gelfand-Leray residue d x dt
Ady
along the fiber p(a:, y) 0.

Definition 3.4.50. A 3-bundle is a collection ( E , 3E,p, B ) consisting of the following.

(a) A smooth manifold E with boundary BE.

(ii) A smooth map p : E -> B such that the restrictions p I t E -> B and p : BE -> B
are smooth bundles.

The standard fiber of p : I t E -> B is called the interiorjiber.

One can think of a 3-bundle as a smooth family of manifolds with boundary.

Example 3.4.51. The projection

p:[0,l]xM->M (t;m)»->m,

defines a <9-bundle. The interior fiber is the open interval (0, 1). The fiber of p : c9(I X

M) -> M is the disjoint union of two points.

Standard Models. A <9-bundle is obtained by gluing two types of local models.

.. Interior models.
Boundary models.
R""xR1"->]R""
H + x]R2.'"->]RM, where
T'

as?"
H+T'
:={(;I31,.-~
7 oR; x1 > 0}.
Remark 3.4.S2. Let p : (E, BE) -> B be a <9-bundle. If p : I t E -> B is orientable and
the basis B is oriented as well, then on BE one can define two orientations.

(i) The fiber-first orientation as the total space of an oriented bundle 3E->B.
(ii) The induced orientation as the boundary of E.

These two orientations coincide.

Exercise 3.4.S3. Prove that the above orientations on BE coincide.

 Calculus on Manifolds 137

Theorem 3.4.S4. Let p : ( E , B E ) -> B be an orientable 3-bundle with an 1°-dimensional

interiorjiben Then for any w 6 Q;pt(E) vm have

dEad - (-1)*d8 w (Homotopy formula).

I3E/B IE/8 E/B

The last equality can be formulated as

/AE/B /E/B
dE - (-l)"d8
E/B

This is "the mother of all homotopy formulae". It will play a crucial part in Chapter 7 when
we embark on the study of DeRham cosmology.

Exercise 3.4.55. Prove the above theorem.

This page intentionally left blank
 Chapter 4

Riemannian Geometry

Now we can finally put to work the abstract notions discussed in the previous chapters.
Loosely speaking, the Riemannian geometry studies the properties of surfaces (manifolds)
"made of canvas". These are manifolds with an extra structure arising naturally in many
instances. On such manifolds one can speak of the length of a curve, and the angle between
two smooth curves. In particular, we will study the problem formulated in Chapter 1: why a
plane (flat) canvas disk cannot be wrapped in an one-to-one fashion around the unit sphere
in R3. Answering this requires the notion of Riemann curvature which will be the central
theme of this chapter.

4.1 Metric properties

4.1.1 Definitions and examples

To motivate our definition we will first try to formulate rigorously what do we mean by a
"canvas surface".
A "canvas surface" can be deformed in many ways but with some limitations: it cannot
be stretched as a rubber surface because the fibers of the canvas are flexible but not elastic.
Alternatively, this means that the only operations we can perform are those which do not
change the lengths of curves on the surface. Thus, one can think of "canvas surfaces" as
those surfaces on which any "reasonable" curve has a well defined length.
Adapting a more constructive point of view, one can say that such surfaces are endowed
with a clear procedure of measuring lengths of piecewise smooth curves.
Classical vector analysis describes one method of measuring lengths of smooth paths
in R3. If by [0, 11 -> R3 is such a paths, then its length is given by
1
length (w) l(t)udt,
0

where l'Y(t)l is the Euclidean length of the tangent vector WI,

We want to do the same thing on an abstract manifold, and we are clearly faced with
one problem: how do we make sense of the length lf}/(t)l? Obviously, this problem can be
solved if we assume that there is a procedure of measuring lengths of tangent vectors at any
point on our manifold. The simplest way to do achieve this is to assume that each tangent

139
140 Lectures on the Geometry of Manzfolds

space is endowed with an inner product (which can vary from point to point in a smooth
way).

Definition 4.1.1. (a) A Riemann manifold is a pair (M, g) consisting of a smooth manifold
M and a metric g on the tangent bundle, i.e., a smooth, symmetric positive definite (0, 2)-
tensor field on M. The tensor g is called a Riemann metric on M.
(b) Two Riemann manifolds (Mi, oil (i = 1, 2) are said to be isometric if there exists a
diffeomorphism QUO : M1 -> M2 such that (l5*g2 = 91-

If ( M , g ) is a Riemann manifold then, for any ;136 M , the restriction

g,,:TxMxTxM->H

.. ..
is an inner product on the tangent space T x M . We will frequently use the alternative
notation ( I's = gx( , . The length of a tangent vector 'U 6 TxM is defined as usual,
7

Ivlx gLC:l'U,'U)1/2

If by [a, b] -> M is a piecewise smooth path, then we define its length by

1('y) b l'ltllv(t)dt°
a

If we choose local coordinates (al , . . . ,a:") on M, then we get a local description of g as

g
rd:d x i
id l
(a azj 78
gi 9 $1

I
Proposition 4.1.2. Let M be a smooth manifold, and denote by RM the set of Riemann
metrics on M . Then RM is a non-empty convex cone in the linear space of symmetric
(0, 2)-tensors.

Proof. The only thing that is not obvious is that (Qm is non-empty. We will use again par-
titions of unity. Cover M by coordinate neighborhoods ( U 6 ) o € A . Let (kg) be a collection
of local coordinates on UQ. Using these local coordinates we can construct by hand the
metric ga on Ua by
go ; (d;u1)2-I-°~ + (dw:;)2.

Now, pick a partition of unity 8 C 05° (M) subordinated to the cover qUo)oeA, i.e., there
exits a map go : 8 -> A such that VB 6 8 suppl C U¢(g)~ Then define

g I39</w0) .
Bea;
The reader can check easily that g is well defined, and it is indeed a Riemann metric on M .

Example 4.1.3 (The Euclidean space). The space R" has a natural Riemann metric
90 = (d231)2 -I- + (dw")2.
The geometry of (RTL, g0l is the classical uclidean geometry.
 Riemannian Geometry 141

Example 4.1.4 (Induced metrics on submanifolds). Let ( M , g ) be a Riemann manifold

and S C M a submanifold. If 'L : S -> M denotes the natural inclusion then we obtain by
pull back a metric on S
gs='L*g=gls
For example, any invertible symmetric n o n matrix defines a quadratic hypersurface in R"
by
1 6 R" (A;u,;u) = ii
..
QUO 7

where ( 7 ) denotes the Euclidean inner product on Rn. <%"A has a natural metric induced
by the Euclidean metric on R". For example, when A = In, then <%?,, is the unit sphere in
R", and the induced metric is called the round metric of S""1.

/ -
l
l
l
l
l
l
l
_ - - | » _
I
I
I
I
I
I
I
I
` L /

Fig. 4.1 The unit sphere and an ellipsoid look "dilTerent".

......................... . . .

>

..............................

Fig. 4.2 A plane sheet and a half cylinder are "not SO different

Remark 4.1.5. On any manifold there exist many Riemann metrics, and there is no natural
way of selecting one of them. One can visualize a Riemann structure as defining a "shape"
142 Lectures on the Geometry of Manzfolds

of the manifold. For example, the unit sphere $1:2 -|- y2 -|- Z2 = 1 is diffeomorphic to
y2
the ellipsoid % -|- -
Z2

22
-|- -
32
= 1, but they look "different" (see Figure 4.1). However,
appearances may be deceiving. In Figure 4.2 it is illustrated the deformation of a sheet of
paper to a half cylinder. They look different, but the metric structures are the same since
we have not changed the lengths of curves on our sheet. The conclusion to be drawn from
these two examples is that we have to be very careful when we use the attribute "different".

Example 4.1.6 (The hyperbolic plane). The Poincaré model of the hyperbolic plane is
the Riemann manifold (D, 9) where D is the unit open disk in the plane R2 and the metric
g is given by
1
g
1 $2 _ y2 ( da:2 dy2).

Exercise 4.1.7. Let 11-f denote the upper half-plane

17-(={('u,,v)€R2,v>0},

endowed with the metric

hi
1
7u? + dU2)
4'u2 (d
Show that the Cayley transform

-I-iy»->w
.z + i
u-I-w
.
Z .CU zZ •
_ z

establishes an isometric (D , 8 ) % ( m , h ) .

Example 4.1.8 (Left invariant metrics on Lie groups). Consider a Lie group G, and de-
-
note by LG its Lie algebra. Then any inner product (w ) on LG induces a Riemann metric
h = (w °>9 on G defined by

Ilglx»yl I (X»Y)9 <(/29-1)*X» (L9-1)*Y), Vg 6 G, X , Y Q TgG,

where (Lg- 1)* . TgG -> T1G is the differential at g k G of the left translation map L9 1.

One checks easily that the correspondence

G99'-<')'>9
is a smooth tensor field, and it is left invariant, i.e.,

L g h h 'v'g€G.

If G is also compact, we can use the averaging technique of Subsection 3.4.2 to produce
metrics which are both left and right invariant.
 Riemannian Geometry 143

4.1.2 The Levi-Civita eonneetion

To continue our study of Riemann manifolds we will try to follow a close parallel with
classical Euclidean geometry. The first question one may ask is whether there is a notion
of "straight line" on a Riemann manifold.
In the Euclidean space R3 there are at least two ways to define a line segment.

(i) A line segment is the shortest path connecting two given points.
(ii) A line segment is a smooth path 'Y : [0, 1] -> R3 satisfying
w) 0. (4.l.1)

Since we have not said anything about calculus of variations which deals precisely with
problems of type (i), we will use the second interpretation as our starting point. We will
soon see however that both points of view yield the same conclusion.
Let us first reformulate (4.1.1). As we know, the tangent bundle of R3 is equipped with
a natural trivialization, and as such, it has a natural trivial connection V0 defined by
v931 = 0 iv"£,j, where a a 7 Vi Va 9

i.e., all the Christoffel symbols vanish. Moreover, if 90 denotes the Euclidean metric, then
V05 k 90(V06j, 60 - 90(6j, V06 0,
i.e., the connection is compatible with the metric. Condition (4.l.l) can be rephrased as

V(¢)(tl 0, (4.l.2)
so that the problem of defining "lines" in a Riemann manifold reduces to choosing a "nat-
ural" connection on the tangent bundle.
Of course, we would like this connection to be compatible with the metric, but even
so, there are infinitely many connections to choose from. The following fundamental result
will solve this dilemma.

Proposition 4.1.9. Consider a Riemann manifold ( M , g). Then there exists a unique sym-
metric connection V on TM compatible with the metric g, i.e.,
T<v>=0, V g = 0 .
The connection V is usually called the Levi-Civita connection associated to the metric g.

Proof. Uniqueness. We will achieve this by producing an explicit description of a connec-

tion with the above two properties.
Let V be such a connection, i.e.,
V g = 0 and V x Y - V ; / X : [X,Y], VX,Y 6 V e c t ( M ) .
For any X, Y, Z 6 Vect (M) we have
Z g ( X , Y) = 9(VzX, Y) -|- g(X, VzYI
since Vg = 0. Using the symmetry of the connection we compute
Zg(X,Y) - Y g ( Z , X ) +Xg(Y,Z) = 9(VzX»Y) - 9 ( V y Z , X l *l'glVxY,Zl
144 Lectures on the Geometry of Manzfolds

+ g ( X , VzY ) - 9(Z, VyXI + 9(Y» VxZ)

9([Z» YL X) + 9([X» YL Z) + g([Z» XL Y) -|- 2g(VxZ, Y)~
We conclude that
QIVxZ, Y) §{X9(Y» Zl - YQ (Z ,X ) -|- Zg(X,Y)

-9([x,y],z) + g([Y» ZLX) - 9([ z,x ],yl }. (4.l.3)

The above equality establishes the uniqueness of V.
Using local coordinates (x1 . . . ,ac") on M we deduce from (4.l.3), with X <2
a , Y = 3 k = 3 , , k , Z = 3 =j 3x3 la that
1
Q( i3j,31<=l glee (3i9j /<: - Mk 9 i j + <9j9i1<=)
Above, the scalars F e23 denote the Christoffel symbols of V in these coordinates, i.e.,

If ( g ) denotes the inverse of lgwl we deduce

1
F f = 59kg l3i9jA: - 3I<91i + 3j9iI<=) (4.l.4)
Existence. It boils down to showing that (4.1.3) indeed defines a connection with the re-
quired properties. The routine but rather tedious details are left to the reader.

We can now define the notion of "straight line" on a Riemann manifold.

Definition 4.1.10. A geodesic on a Riemann manifold (M, g) is a smooth path
b y : (a,b) ->M,
satisfying
v (4.l.5)
where V is the Levi-Civita connection.

Using local coordinates (501, ..., so'n) with respect to which the Christoffel symbols are
(Pk ), and the path by is described by 'y(tl = (ac1 (t), ..., a:'*(t)), we can rewrite the geodesic
equation as a second order, nonlinear system of ordinary differential equations. Set
d
_W) g.
Then,
v_ ( ¢ )
d
dt
= a+ p a = Zii + ¢z¢3vi@,
= iekak -|- r x w a k (FZ = paj i l )
so that the geodesic equation is equivalent to
k
at -|-
.¢ ..
Pkml' at" 0 Wk l,...,n. i i (4. 1.6)
Since the coefficients Fk*
'in = Fk
if (13) depend smoothly upon co, we can use the classical
Banach-Picard theorem on existence in initial value problems (see e.g. [6]). We deduce the
following local existence result.
Proposition 4.1.11. Let ( M , g) be a Riemann manifold. For any compact subset K C T M
there exists 6 > 0 such that for any (as, X ) 6 K there exists a unique geodesic by 'Yx,X
(-5, s) -> M such that 7(0) = co, fi/(0) = X.
 Riemannian Geometry 145

One can think of a geodesic as defining a path in the tangent bundle t »-> ('y(tl, 'Y(t)).
The above proposition shows that the geodesics define a local flow Q on T M by

<1>"(w,x) = ( v ( t l / 7 ( t ) ) 'Y Z Yx,X-

Definition 4.1.12. The local flow defined above is called the geodesic flow of the Riemann
manifold (M, g). When the geodesic glow is a global flow, i.e., any 'Y1I:,X is defined at each
moment of time t for any (ac, X) 6 TM, then the Riemann manifold is called geodesically
complete.

The geodesic flow has some remarkable properties.

Proposition 4.1.13 (Conservation of energy). If the path 'y(t) is a geodesic, then the
length of W) is independent of time.

Proof. We have
d 2 d . •
9(v(t)/7(t% = 2 9 ( V ( ¢ ) , 1(0)
dt l ( t ) l 0.
Thus, if we consider the sphere bundles

5,.(m> = {x 6 T M ; IXI = w,
we deduce that s,,.(m> are invariant subsets of the geodesic flow.

Exercise 4.1.14. Describe the infinitesimal generator of the geodesic flow.

Example 4.1.1S. Let G be a connected Lie group, and let LG be its Lie algebra. Any
X 6 LG defines an endomorphism ad(X) of LG by
ad(X)Y [X, Y].
The Jacobi identity implies that

ad([X, YE) = law), ado/)1,

where the bracket in the right hand side is the usual commutator of two endomorphisms.
Assume that there exists an inner product ( ) on LG such that, for any X 6 LG, the
operator ad(X) is skew-adjoint, i.e.,

<[X,Y],Z) = - ( y J m l ) . (4.1.7)

We can now extend this inner product to a left invariant metric h on G. We want to describe
its geodesics.
First, we have to determine the associated Levi-Civita connection. Using (4.13) we
get

h(vxz, y) §{xh(y, zl - y(z,x) + Z h ( X , y)

-h([X,Y], 2) + h([Y, Z],X) h([Z7X]7Y)}.
146 Lectures on the Geometry of Manzfolds

If we take X, Y, Z 6 LG, i.e., these vector fields are left invariant, then h(Y, Z) const.,
h(Z,X) = const., h(X,Y) = const. so that the first three terms in the above formula
vanish. We obtain the following equality (at l G G )
1
(VX z, Y) -<[x,y1, Z) + <[y,.z],X> <[z,x],y>}.
Using the skew-symmetry of ad(X) and ad(Z) we deduce
1
<[X,z],y>,
SO that, at 1 k G , we have
1
VxZ §[X»2] \V/X,Z6£JG. (4.l.8)

This formula correctly defines a connection since any X 6 Vect (G) can be written as a
linear combination

0£iXZ° 6 COO(G) Xi G LG.

II

Old

If 'y(t) is a geodesic, we can write W ) = 2 'Y/iXi, so that

1
0 V(¢)('5l fm + 2 ..
'Ywj [X XJ]-
z,.7

Since [X¢»Xi] = -[Xi»Xi]» we deduce i = 0, i.e.,

W) .2'l'»;lOlX»; = X .
This means that by is an integral curve of the left invariant vector field X so that the
geodesics through the origin with initial direction X 6 T1G are

'Yx(tl = exp(tx).

Exercise 4.1.16. Let G be a Lie group and h a bi-invariant metric on G. Prove that its
restriction to LG satisfies (4.1.7). In particular, on any compact Lie groups there exist
metrics satisfying (4. 1.7).

Definition 4.1.17. Let L be a finite dimensional real Lie algebra. The Killing pairing or
form is the bilinear map
H, :L X L -> R, IQ(x,yl := -tr (ad(x) ad(y)) x , y 6 L.

The Lie algebra ,G is said to be semisimple if the Killing pairing is a duality. A Lie group
G is called semisimple if its Lie algebra is semisimple.

Exercise 4.1.18. Prove that SO(n) and SU(n) and SL(n, RI are semisimple Lie groups,
but U(n) is not.
 Riemannian Geometry 147

Exercise 4.1.19. Let G be a compact Lie group. Prove that the Killing form is positive
semi-definite1 and satisfies (4.1.7).
Hint: Use Exercise 4.1.16.

Exercise 4.1.20. Show that the parallel transport of X along exp(tY) is

(L@xp(%y)>* l R e x p ( § Y ) l * X °

Example 4.1.21 (Geodesics on flat tori, and on SU(2)). The n-dimensional torus T n r\/

S1 X X S1 is an Abelian, compact Lie group. If (91 7 • . • 7 @") are the natural angular

coordinates on T", then the flat metric is defined by

(do1)2 + + (do")2_
m
II

The metric g on T" is bi-invariant, and obviously, its restriction to the origin satisfies the
skew-symmetry condition (4.l.7) since the bracket is 0. The geodesics through 1 will be
the exponentials

. . . a ( ¢ ) t+->(e°"1t 7 .. . e t )
7 end.
If the numbers as are linearly dependent over Q, then obviously ' y 0 1 1 . . . O n (t) is a closed
curve. On the contrary, when the a's are linearly independent over Q then a classical result
of Kronecker (see e.g. [59]) states that the image of ' y o i l . . . o n is dense in Tn!!! (see also
Section 7.4 to come).
The special unitary group SU(2) can also be identified with the group of unit quater-
nions

{a+bfi+cj+dk; a2+b2+02+d2 1},

so that SU(2) is diffeomorphic with the unit sphere S3 C R4. The round metric on S3
is bi-invariant with respect to left and right (unit) quaternionic multiplication (verify this),
and its restriction to (1 0 0, 0) satisfies (4.l.7). The geodesics of this metric are the 1-
7 7

parameter subgroups of S3, and we let the reader verify that these are in fact the great
circles of S3, i.e., the circles of maximal diameter on S3. Thus, all the geodesics on S3 are
closed.

4.1.3 The exponential map and normal coordinates

We have already seen that there are many differences between the classical Euclidean ge-
ometry and the the general Riemannian geometry in the large. In particular, we have seen
examples in which one of the basic axioms of Euclidean geometry no longer holds: two
distinct geodesic (read lines) may intersect in more than one point. The global topology of
the manifold is responsible for this "failure".
lThe converse of the above exercise is also true, i.e., any semisimple Lie group with positive definite Killing
form is compact. This is known as Weyl's theorem. Its proof, which will be given later in the book, requires
substantially more work.
148 Lectures on the Geometry of Manzfolds

Locally however, things are not "as bad". Local Riemannian geometry is similar in
many respects with the Euclidean geometry. For example, locally, all of the classical inci-
dence axioms hold.
In this section we will define using the metric some special collections of local coordi-
nates in which things are very close to being Euclidean.
Let (M, g) be a Riemann manifold and U an open coordinate neighborhood with coor-
dinates (ac1, ..., mTL). We will try to find a local change in coordinates (al +-> (ye) in which
the expression of the metric is as close as possible to the Euclidean metric 90 = (S d g d y j .
Let q 6 U be the point with coordinates (0,...,0) . Via a linear change in coordinates
we may as well assume that

9¢i(<1) = 5¢j~
We can formulate this by saying that lgtjl is Euclidean up to order zero.
We would like to "spread" the above equality to an entire neighborhood of q. To achieve
this we try to find local coordinates (ye ) near q such that in these new coordinates the metric
is Euclidean up to order one, i.e.,
391g 360
gin (Q) (So 0. vi,j,I@.
5'y": (Q) 5'y"" (Q)
We now describe a geometric way of producing such coordinates using the geodesic flow.
Denote as usual the geodesic from q with initial direction X 6 TqM by 'Yq,X (15). Note
the following simple fact.
iv's > 0 'ye 5x(tl = 'Yq,xl5tl-

Hence, there exists a small neighborhood V o f 0 6 Tqll/I such that, for any X 6 V, the
geodesic 'Yq,X (to is defined for all ltl < 1. We define the exponential map at q by
expq : V C TqM -> M, X »-> 'Yq,X(1).
The tangent space TqM is a Euclidean space, and we can define Dq(1°) C T q M , the open
"disk" of radius T' centered at the origin. We have the following result.

Proposition 4.1.22. Let ( M , go and q 6 M as above. Then there exists T' > 0 such that the
exponential map
expq : D q(al -> M
i5 a dyeomorphism onto. The supreme of all radii T' with this property i5 denoted by
DmM)-

Definition 4.1.23. The positive real number pm(q) is called the injectivily radius o f M at
q. The infimum

Pm i18f Pm (Q)
is called the injectivily radius o f M .
 Riemannian Geometry 149

The proof of Proposition 4.1.22 relies on the following key fact.

Lemma 4.1.24. The Fréchet deferential at 0 6 TqM of the exponential map

DO expo . TOM Texpq(0)M I TqM
i5 the identity TqM -> TqM.

Proof. Consider X 6 TqM. It defines a line t »-> TX in TqM which is mapped via the
exponential map to the geodesic 'Yq,X (t). By definition
(Do expq)X = ' q,Xl0) X.
Proposition 4.1.22 follows immediately from the above lemma using the inverse func-
tion theorem.

Now choose an orthonormal frame (el, ..., en) ofTqM, and denote by (as1, ..., asn) the
resulting Cartesian coordinates in TqM. For 0 < 7° < pM(q), any poi f t p 6 expq(D q(7"))
can be uniquely written as

P : : expo(we),
SO that the collection (ml, ..., mol provides a coordinatization of the open set

expq(D q(¢~)) C M. The coordinates thus obtained are called normal coordinates at q,
the open set expq (D ql'r°)l is called a normal neighborhood, and will be denoted by B AQ)
for reasons that will become apparent a little later.

Proposition 4.1.25. Let W ) be normal coordinates at q 6 M, and denote by g'in the

expression of the metric tensor in these coordinates. Then we have
<99 in
g fzjlql and 0 W, j, k .
@a:'" (Q)
($1j

Thus, the normal coordinates provide a ]inst order contact between g and the Euclidean
metric.
a form an orthonormal basis of TqM and this
Proof. By construction, the vectors ii = 3:12 i
proves the first equality. To prove the second equality we need the following auxiliary
result.

Lemma 4.1.26. In normal coordinates ( M i l (at q) the Christoffel symbols F vanish at q.

Proof. For any (m1, ...,m") 6 R" the curve t »-> la:"(t))1<<n, w i l t ) = m t , W is the
geodesic t »-> expq (E mt633) so that
F ]k(a:(t))m3mk = 0.
In particular,
F k(0lm3mk 0 Vmj 6 R"
from which we deduce the lemma.
150 Lectures on the Geometry of Manzfolds

The result in the above lemma can be formulated as

g V 0, b y k
3:23
.
amz 7
gmk i

so that,
a
V a . 33% 0 at q, W J.
•
(4.l.9)
3:13-7

691g
Using Vg = 0 we deduce 3mk (q)=(;1g z`]llCI=0°
The reader may ask whether we can go one step further, and find local coordinates
which produce a second order contact with the Euclidean metric. At this step we are in for
a big surprise. This thing is in general not possible and, in fact, there is a geometric way
of measuring the "second order distance" between an arbitrary metric and the Euclidean
metric. This is where the curvature of the Levi-Civita connection comes in, and we will
devote an entire section to this subject.

4.1.4 The length minimizing properly of geodesies

We defined geodesics via a second order equation imitating the second order equation defin-
ing lines in an Euclidean space. As we have already mentioned, this is not the unique way
of extending the notion of Euclidean straight line to arbitrary Riemann manifolds.
One may try to look for curves of minimal length joining two given points. We will
prove that the geodesics defined as in the previous subsection do just that, at least locally.
We begin with a technical result which is interesting in its own. Let (M, g) be a Riemann
manifold.

Lemma 4.1.27. For each q 6 M there exists 0 < T' < pM(q), and 6 > 0 such that,
Wm E B ,~(q), we have e < Pm ( m ) and B € ( m ) D B rlql. In particular; any two points of
B ,.(q) can be joined by a unique geodesic of length < e.

We must warn the reader that the above result does not guarantee that the postulated
connecting geodesic lies entirely in B ,~(q). This is a different ball game.

Proof. Using the smooth dependence upon initial data in ordinary differential equations
we deduce that there exists an open neighborhood V of (q, Of 6 TM such that exp X is
well defined for all (m, X ) 6 V. We get a smooth map
F:V->M X M (m,X)»->(m,expmXl.
We compute the differential of F at (q, 0). First, using normal coordinates (wi) near q we
get coordinates W; Xi ) near (q, 0) 6 TM. The partial derivatives of F at (q, Of are
/ Q \
3 _|_ <9 / pa \
_ a
D(q,0)F 1 \
Z
ax in D(<1,0)F 1 - ax
Thus, the matrix defining D(q,0)F has the form

and in particular, it is nonsingular.

 Riemannian Geometry 151

It follows from the implicit function theorem that F maps some neighborhood V of
(q, 0) 6 TM diffeomorphically onto some neighborhood U of (q, q) 6 M X M. We can
choose V to have the form { ( m , X) ; IX lm < 6, m y B 6 (q)} for some sufficiently small
6 and 6. Choose 0 < 1° < min(€, pM(q)) such that

M17 M2 E B 7(ql \.
1
( M M 2 ) E U.

In particular, we deduce that, for any m 6 B1(q), the map exp D (m) C TmM -> M
is a diffeomorphism onto its image, and

B ~=:lml=exPml D @(m))D B rltll-

Clearly, for any m 6 M, the curve t +-> exp,,n(tX) is a geodesic of length < 5 joining m
to expm(X). It is the unique geodesic with this property since F : V -> U is injective.

We can now formulate the main result of this subsection.

Theorem 4.1.28. Let q, T' and 5 as in the previous lemma, and consider the unique geodesic
by : [0, 1] -> M of length < e joining two points of B ,~(q). I f s : [0, 1] -> M is piecewise
smooth path with the same endpoints as by then

/ Wtlldt /'
0
<
0
law°(t)ldt

with equality If and only If w([0, 1]) = *y([0, 1]). Thus, Fy is the shortest path joining its
endpoints.

The proof relies on two lemmata. Let m G M be an arbitrary point, and assume 0 < R <
Pm(7"l~

Lemma 4.1.29 (Gauss). In B R(m) C M , the geodesics through m are orthogonal to the
hype rsurjfaces

26 = €XPq(56(0)) {@XP177,(Xl; l x l = 5 } , 0 < 6 < R.

Proof. Let t »-> X ( t ) , 0 < t < l denote a smooth curve in T,1,M such that txt¢)lm = 1,
i.e., X ( t ) is a curve on the unit sphere $1 (0) C T71,M. We assume that the map t »->
X (15) 6 S1 (0) is an embedding, i.e., it is injective, and its differential is nowhere zero. We
have to prove that the curve t »-> exp,m(5X (t)) are orthogonal to the radial geodesics

S +-> exam(sx(t))» 0 < s < R .

Consider the smooth map

f : [0,R] X [0, 1] -> M, f(s,t) = exam(sx(¢)l (s,t) 6 (0,R) X (0, 1).

If we use normal coordinates on BR(M) we can express f as the embedding

(0, R) X (0, 1) -> TmM, (s,t) »-> sX(t).

152 Lectures on the Geometry of Manzfolds

Set

* E Tf(s»t)M'

Define 315 similarly. The objects 35, and <9¢ are sections of the restriction of TM to the
image of f. Using the normal coordinates on BR(M) we can think of 3S as a vector field
on a region in T T T I M , and as such we have the equality 3S = X (t). We have to show

(35> at) i 0 'v'(s, tlv

where ( ) denotes the inner product defined by g.
Using the metric compatibility of the Levi-Civita connection along the image of f we
compute

aS(6S7 at) I (V3S35,3t) -|- <387 VaS3¢)-

Since the curves s »-> f ( s , t = const.) are geodesics, we deduce
VIS 3S = 0.
On the other hand, since is, a ] = 0, we deduce (using the symmetry of the Levi-Civita
connection)
l
(Asa V65 at) I (as) VI 95) 56 l@8l2 I 0»
since $3.S1 = IX(t)l = 1. We conclude that the quantity (6.S76t) is independent of 8. For
S = 0 we have f(0, t) = expm(0l so that 6 t l s = 0 = 0, and therefore

(55, fit) Z 0 Ws, t),

as needed.

Now consider any continuous, piecewise smooth curve

[a, b] -> B R(ml \ {M}~
Each w ( t l can be uniquely expressed in the form

w(t) i expmlpltlx(tll IX(t)l 1 0 < up<¢)l < R.

Lemma 4.1.30. The length of the curve w(tl i5 > lp(bl - p(all. The equality holds If
and only gr X (t) = const and ,O(tl > 0. In other words, the shortest path joining l o
concentrical shells 26 is a radial geodesic.
Pal( p
Proof. Let f(0, t) eX x (t))» so that w) f(»0(t)» t ) . Then
of <9f
u= o p+
f) at '

Since the vectors 5; and Qi

fit
are orthogonal, and since

IX(t)l 1
 Riemannian Geometry 153

we get

The equality holds if and only if

l(»J'l

go
2

;
IPI2+
uu
of
(915

0, i.e., X
2 >
II2
0. Thus
b b

L lwldt >
L lbldt > lp(b) p(Gll.
Equality holds if and only if p(t) is monotone, and X (t) is constant. This completes the
proof of the lemma.

The proof of Theorem 4.1.28 is now immediate. Let mg, Ml E B rlql» and consider a
geodesic 'Y : [0, 1] -> M of length < 6 such that W) = m , i = 0, 1. We can write
'y(t) = expm0(tX) X 6 D @(Mol.

Set R = IX |. Consider any other piecewise smooth path w : [a, b] -> M joining mg to
m l . For any 6 > 0 this path must contain a portion joining the shell E6(m0) to the shell
ER(m0) and lying between them. By the previous lemma the length of this segment will
be > R - 6. Letting 6 -> 0 we deduce

l((»J) >
_R =:
l('y).
If w([a, b]) does not coincide with 'yl[0, 11) then we obtain a strict inequality.

Any Riemann manifold has a natural structure of metric space. More precisely, we set

d(p, Q) = inf{z ((»J); w : [0, 1] -> M piecewise smooth path joining p to q}.
A piecewise smooth path LU connecting two points p, q such that l(w) = dup, Q) is said to
be minimal. From Theorem 4. 1.28 we deduce immediately the following consequence.

Corollary 4 .1.31. The image of any minimal path coincides with the image of a geodesic.
In other words, any minimal path can be reparametrized such that it satisfies the geodesic
equation.

Exercise 4.1.32. Prove the above corollary.

Theorem 4.1.28 also implies that any two nearby points can be joined by a unique
minimal geodesic. In particular we have the following consequence.

Corollary 4.1.33. Let q 6 M . Then for all T' > 0 sujiciently small

expo(D »~(0))(= B ~(q)) = {p 6 M d(p,q) < r }- (4.1.10)

Corollary 4.1.34. For any q 6 M we have the equality

pm(q) = sup{ T; 7° satisfies (41.10) }.

154 Lectures on the Geometry of Manzfolds

Proof. The same argument used in the proof of Theorem 4.1.28 shows that for any 0 <
< pm(q) the radial geodesics expo(tX) are minimal.
1°

Definition 4.1.35. A subset U C M is said to be convex if any two points in U can be

joined by a unique minimal geodesic which lies entirely inside U .

Proposition 4.1.36. For any q 6 M there exists 0 < R < LM(q) such that for any 7° < R
the ball B ,~(q) is convex.

Proof. Choose ax 0 < 5 < pM(q), and 0 < R < 6 such that any two points mg, ml
in BR(q) can be joined by a unique minimal geodesic [0, 11 9 t +-> 'Ymo,m1 (t) of length
< 5, not necessarily contained in B Rlfll . We will prove that Vm0, ml 6 B RIG) the map
t +-> d(q, fym0 m1 (t)) is convex and thus it achieves its maxima at the endpoints t = 0, 1.
Note that
d(q,v(t)) < R+6 < 0m(q)-
The geodesic 'Ym0,m1 (t) can be uniquely expressed as
tYmon. (t) Z expq(1'(t)x(t)l XG) 6 TqM with t) dlq/Ym0,m1 (ill- i

It suffices to show (r2) > 0 for t 6 [0, 1] if d(q, mo) and d(q, m l ) are sufficiently small.
At this moment it is convenient to use normal coordinates ( w ) near q. The geodesic
f y m m l takes the form (we), and we have
7'
2 (w1)2+~~~+(w")2.

We compute easily
do .
+~--+a:")+la¢l 2
n no

a31 (4.l.11)
. d¢2("2) =2r2(
where :i:(t) = Z £i3z€z' 6 TqM. The path by satisfies the equation
+ p 9 k ( m > ¢ 2 ¢ k = 0.
Since l`ik(0) - 0 (normal coordinates), we deduce that there exists a constant C > 0
'

(depending only on the magnitude of the second derivatives of the metric at q) such that
up;(w)l <
_ Cl;1;'l.
Using the geodesic equation we obtain
> _civil&fl 2
We substitute the above inequality in (4.l.l 1) to get
d2 2
1 QuffU-wwW
> (4.l.12)
co# (7°
If we chose from the very beginning
R -I- 5 S (nC)-1/3 7

_ R -|- 5, the right-hand-side of (4.1.l2) is

then, because along the geodesic we have 1221 <
nonnegative. This establishes the convexity of t +-> ,r,2 (t) and concludes the proof of the
proposition.

In the last result of this subsection we return to the concept of geodesic completeness.
We will see that this can be described in terms of the metric space structure alone.
 Riemannian Geometry 155

Theorem 4.1.37 (Hopf-Rinow). Let M be a Riemann manifold and q 6 M . The following

assertions are equivalent:
(a) expq is defined on all ofTqM.
(b) The closed and bounded (with respect to the metric structure) sets of M are compact.
(e) M is complete as a metric space.
(d) M is geodesically complete.
(e) Tnere exists a sequence of eompact sets Kn C M, Kn C Kn-i-1 and Un K T L = M such
that impn Q Kn then d(q,pn) -> oo.
Moreovel; on a (geodesically) complete manifold any halo points can be joined by a
minimal geodesic.

Exercise 4.1.38. Prove the Hopf- inow theorem.

Remark 4.1.39. (a) Every smooth manifold admits complete Riemann metrics. We refer
to [108] for a proof and more refined statements.
(b) On a complete manifold there could exist points (sufficiently far apart) which can be
joined by more than one minimal geodesic. Think for example of a manifold where there
exist closed geodesic, e.g., the tori T".

4.1.5 Calculus on Riemann manifolds

The classical vector analysis extends nicely to Riemann manifolds. We devote this subsec-
tion to describing this more general "vector analysis".
Let ( M , g) be an oriented Riemann manifold. We now have two structures at our
disposal: a Riemann metric, and an orientation, and we will use both of them to construct
a plethora of operations on tensors.
First, using the metric we can construct by duality the lowering-the-indices isomor-
phism L : Vect (Ml -> Q1(M), see Example 2.2.48.
Example 4.1.40. Let M R3 with the Euclidean metric. A vector field V on M has the
form
v=p-+
a <9
-+R-
<9
fix Q 331 <9z
Then
W LV = Pdf? -|- Qdy -|- Rdz.
If we think of V as a field of forces in the space, then W is the infinitesimal work of V .

On a Riemann manifold there is an equivalent way of describing the exterior derivative.

Proposition 4.1.41. Let
o°°(T*m (X) A'°T*m) -> o°°(Ak-I-1T*m)
denote the exterior multiplication operator
e(a (X) el a A B, Va 6 M m ) , /3 6 Qk(ml_
156 Lectures on the Geometry of Manzfolds

Then the exterior derivative d is related to the Levi-Civita on A ' T * m connection via the
equality d = 5 o V.

Proof. We will use a strategy useful in many other situations. Our discussion about normal
coordinates will payoff. Denote temporarily by D the operator 5 o V.
The equality d = D is a local statement, and it suffices to prove it in any coordinate
neighborhood. Choose (as) normal coordinates at an arbitrary point p 6 M, and set
<9i := 3:13a ii . Note that

do? /\ v , v
m
II - pa

Let w 6 Q1<=(M). Nearp it can be written as

fbI,
3

CU]
II

where as usual, for any ordered multi-index I: (1 < 'it < < in < n , we set
day == MY A ~ - A ¢ W k .
In normal coordinates at p we have (Vi<9i) lp= 0 from which we get the equalities

(vim) In (Mk) = -(d2¢i(V¢31<:ll In= 0.

Thus, at p,

Do
; do? /\ vdmfl

co A <9i(,JI = do.
2 do? A ( a j d w f
I
+ w/v¢(dwI))
I

Since the point p was chosen arbitrarily this completes the proof of Proposition 4. l .41 .

Exercise 4.1.42. Show that for any k-form w on the Riemann manifold (M, go the exterior
derivative do can be expressed by
k
dC4JlX0, Xl, 7 Xi) (-1Y(Vx,(»J)(X0» 7XZ7 7 Xi),
i 0

for all Xi, 7 Xi 6 Vect ( M ) . (V denotes the Levi-Civita connection.)

The Riemann metric defines a metric in any tensor bundle 'I§(M) which we continue
to denote by g. Thus, given two tensor fields To, T2 of the same type (T, 3) we can form
their pointwise scalar product

M 9 p »-> 9(T, s)1, = 9p(T1(p),T2(p))-

In particular, any such tensor has a pointwise norm

M 9 P »-> ITg,p = (T,Tl;/2.
 Riemannian Geometry 157

Using the orientation we can construct (using the results in Subsection 2.2.4) a natural
volume form on M which we denote by d l , , and we call it the metric volume. This is the
positively oriented volume form of pointwise norm E 1.
If (5131, ..., son) are local coordinates such that do1 A - -
A do" is positively oriented,
then
dog v lgld;u1 A~-~/\d;I3" 7

where lgl detlgtjl. In particular, we can integrate (compactly supported) functions on

M by

/ /
of OO
. do fdvg o f 6 CO ( m ) .
(M,g) M
We have the following not so suiprising result.

Proposition 4.1.43. vxdvg 0, VX 6 Vect (M).

Proof. We have to show that for any p 6 M

(vxdvg)(e1, ...,@,,) = 0, (4.l.13)
a
where e l , ..., ep is a basis of T p M . Choose normal coordinates (wall near p. Set 6% = 3:9 9

gfzj = g(3i, 61), and et = 6% lp. Since the expression in (4.1.l3) is linear in X, we may as
well assume X = II for some k = 1, ..., n. We compute
(VxdVg)(e1, en) = X(dVg(31, a l l lp
(4.1.14)
dfvg(e1, ..., (VX<9v)lp, ...<9nl.

We consider each term separately. Note first that dVg(c91, ..., and = ldetlgzjll1/2' I that

X(d€t(9z]))1/2 lp= <9k(d€t(9U))1/2 In

is a linear combination of products in which each product has a factor of the form dkgl] lp.
Such a factor is zero since we are working in normal coordinates. Thus, the first term in
(4. l .14) is zero. The other terms are zero as well since in normal coordinates at p we have
the equality

Vx<l = V0k< = 0.

Proposition 11.3.3 is proved.

Once we have an orientation, we also have the Hodge *-operator

* W M ) -> Q"-k(m),
uniquely determined by
a /\ *B = (Q, gldvg, Va 5 E Q1<(M). (4.1.15)

In particular, *l avg.
158 Lectures on the Geometry of Manzfolds

Example 4.1.44. To any vector field F pa + QQ, + RaZ on R3 we associated its

infinitesimal work
WF = L ( F ) = P d f -|- Qdy -|- Rdz.
The infinitesimal energy flux of F is the 2-form
®p = *WF = PdyAdz+QdzAdx-l-RdacAdy.
The exterior derivative of WF is the infinitesimal flux of the vector field curl F
dwp (3y12 - 3zQldy /\ dz -|- (3ZP - 3xR)dz /\ do -|- (63 Q 3yP)dac /\ dy
®curlF I *Wcurl F ~

The divergence of F is the scalar defined as

d'ivF _ *d* WF - *d®p
-

* {(<9,,,P -l- 33/62 -|- 3ZR)d;z: /\ dy /\ do} <8;,p+<8},Q+<8'ZR

If f is a function on R3, then we compute easily
*d*df= a2f auf a2f Af.
Definition 4.1.45. (a) For any smooth function f on the Riemann manifold ( M , g l we
denote by grad f , or gradg f , the vector field g-dual to the 1-form df. In other words
g ( g ' r a d f , X ) = d f ( x ) = X - f VX 6 Vect ( M ) .
(b) If ( M , g ) is an oriented Riemann manifold, and X 6 Vect ( M ) , we denote by div X ,
or dfifvg X, the smooth function defined by the equality
LxdVg (div)dVg.
Exercise 4.1.46. Consider the unit sphere
S2 = { (a:,y,z) GR3; £132-'ry2-I-z2 =1},
and denote by g the Riemann metric on S2 induced by the Euclidean metric 90 do:2 +
do2 + dz2 on R3.
(a) Express g in the spherical coordinates (v, 0, (al defined as in Example 3.4. 14.
(b) Denote by h the restriction to S2 of the function h(;z:, y, Z = z. Express gradg f in
spherical coordinates.

Proposition 4.1.47. Let X be a v e t o field on the oriented Riemann manifold ( M , g), and
denote by a the 1-form dual to X . Then
(a) d i v X = tr ( V X ) , where we view V X as an element of coo (End ATM)) via the
identQ'ication5
~
=
VX 6 sz1(Tm) 'E COO(T*M (X) T M ) C°°(End(TM)).
(b) d i v X = *d * a.
(c) If (ac1, ..., ;u") are local coordinates such that do1 /\ /\ do" is positively oriented,
then
1
divX = ail x),
v 191
where X Xa,
 Riemannian Geometry 159

..
The proof will rely on the following technical result which is interesting in its own. For
simplicity, we will denote the inner products by l , ), instead of the more precise g(¢, l. .
Lemma 4.1.48. Denote by 6 the operator
6 = *d* : Q1<(M) -> of-Hm).
Let a be a (k 1)-form and B a k-form such that at least one of them is compactly sup-
ported. Then

(do(,/5')dVg -1 II/n,k
( a , 5B)dVg 7
M M
where I/n,k = no -|- n -|- 1.

Proof. We have

M
(do, mdvg
/ M
do I\*@
/M
d(0 A*5)+(-1)k*

The first integral in the right-hand-side vanishes by the Stokes formula since a A * B has
M
Q/\d*@.

compact support. Since

d* B 6 Qn-k+1lml and *2 = (-1)(n-I¢+1)(k-1) on Qn-k+llml
we deduce

(do, Mdvg = l--1lk+(n-k-I-1)(n-k) a /\ *DB.

M M
This establishes the assertion in the lemma since
(n-k+1)(k-1) k; 1/n,k (mod2).

Definition 4.1.49. Define d* : Qk(M) -> Qk-1(M) by

d* := (-1)""""c5 = (-1)"~»= *d*
Proof of the proposition. Set $2 := dvg, and let (XI» ..., XTL) be a local moving frame
of TM in a neighborhood of some point. Then
(LX§2)(X1, ...,X,,) = X(§2(X1, ».07XTL)) ZMX1, ..., [X,Xi], ..,x>. (4.l.16)
al

Since VS) = 0 we get

X (m, -»XTL)) Z §2(x1, v e x , ...,Xn).

Using the above equality in (4.1.16) we deduce from VxY - [X, Y] = V5/X that
(LX§2)(X1,...,XM Zn (Xl, ~.~,VxX7 ...,Xn). (4.l.17)

Over the neighborhood where the local moving frame is defined, we can find smooth func-
tions f ? , such that
v e x = fixe tr (ex) f
160 Lectures on the Geometry of Manzfolds

Part (a) of the proposition follows after we substitute the above equality in (4.l.17).
Proof of ( b ). For any f 6 cO>(ml we have
Lx(f(»J) = ( x f m + f(dfzvx)Q.
On the other hand,
Lx(fQ) = (mi -|- ¢zix)(fQ> = d ¢ x ( f m .
Hence

{(Xf) -|- f ( d i v X)} We dlixfM.

Since the form f Q is compactly supported we deduce from Stokes formula

dlixfM 0.
M

We have thus proved that for any compactly supported function f we have the equality

f (div X )dvg (Xf)dv9 i f (X)dV9

M M M

(grad f , x)dvg ( i f , a)dVg .

M M
Using Lemma 4.1.48 we deduce

f (d'£'v X)dvg féadvg *of 6 05'°(m).

M M

This completes the proof of (b).

Proof of (c). We use the equality

L x ( v I9ldcv1 /\ A darn) = d i v ( x ) l / m d x l A a • u A den).

The desired formula follows derivating in the left-hand-side. One uses the fact that LX is
an even s-derivation and the equalities

Lxd:c = 3 Xfid£l:Z (no summations,

proved in Subsection 3.1.3.

Exercise 4.1.50. Let ( M , go be a Riemann manifold and X 6 Vect ( M ) . Show that the
following conditions on X are equivalent.
(a) LxQ I 0.
(b) g ( v y x , Z ) -|- g(Y, VzX) = 0 for all Y, Z 6 Vect ( M ) .
(A vector field X satisfying the above equivalent conditions is called a Killing vector field.)

Exercise 4.1.51. Consider a Killing vector field X on the oriented Riemann manifold
(M, g), and denote by 17 the 1-form dual to X. Show that 517 = 0, i.e., d i v ( X ) = 0.
 Riemannian Geometry 161

Definition 4.1.S2. Let (M, g) be an oriented Riemann manifold (possibly with boundary).
For any k-forms Q, 3 define

(05) I <05M M
(a,@)dVg
M
a /\ *

whenever the integrals in the right-hand-side are finite.

Let (M, g) be an oriented Riemann manifold with boundary <9M. By definition, M is a

closed subset of a boundary-less manifold M of the same dimension. Along AM we have
a vector bundle decomposition

( T M ) l@m= Team) ® n
where n = (Tam)J_ is the orthogonal complement of T<9M in (TM)16m. Since both M
and AM are oriented manifolds it follows that I/ is a trivial line bundle. Indeed, over the
boundary

det T M = det(T3M) (X) n

so that
iv
n detTM (X) det(T3M)*.

In particular, n admits nowhere vanishing sections, and each such section defines an orien-
tation in the fibers of n.
An outer normal is a nowhere vanishing section O' of n such that, for each :Hz 6 AM,
and any positively oriented w11: 6 det Txc'3M, the product ox A wx is a positively oriented
element of det TazM. Since TL can*ies a fiber metric, we can select a unique outer normal of
pointwise length E 1. This will be called the unit outer normal, and will be denoted by 17.
Using partitions of unity we can extend 17 to a vector field defined on M .

Proposition 4.1.53 (Integration by parts). Let ( M , g) be a compact, oriented Riemann

manifold with boundary, a E §Zk-1(M) and 5 E §2k(Ml. Then

M
(do, B)dVg
/AM'
(Q /\ *Be liM -|-
~m
(Q, d*BldVg

/am
a lam I\5(i175) lam -I-
M
(Q, d*5)dVg

where »2 denotes the Hodge *-operator on AM with the induced metric g and orientation.

Using the ( ) notation of Definition 4.1.52 we can rephrase the above equality as

(do£,5)m = (O¢»i175)am -I- (0(»d*5)m-

162 Lectures on the Geometry of Manzfolds

Proof. As in the proof of Lemma 4.1.48 we have

(do, plug = do /\ *B = d(a /\ *B) + (-l)ka /\ d * B.

The first part of the proposition follows from Stokes formula arguing precisely as in
Lemma 4.1.48. To prove the second part we have to check that

(Q /\ 3) lam= a lam A2(i75) lim

This is a local (even a pointwise) assertion so we may as well assume
1
M H n+ {(x1,...,x") 6 R" ;:L° 2 0}7
and that the metric is the Euclidean metric. Note that 17 = -91. Let I be an ordered (k - l)-
index, and J be an ordered k-index. Denote by JC the ordered (n-k)-index complementary
to J so that (as sets) J U JC = {l, ..., n}. By linearity, it suffices to consider only the cases
a d;L.I,B day] . We have
* d:z:'1 = eJd:c']c (et it ) (4.l.18)

and
0 1
z 7dil? J _d:z:'] 7 7
7 1

where J' = J {1}. Note that, if 1 Q J, then 1 € JC so that

(Q /\ */3) lam= 0 a lam A§(i75l lam»

and therefor, the only nontrivial situation left to be discussed is 1 E J . On the boundary
we have the equality

>T<(il7d:z:J) = - § ( d x J ) = -eJdxJC (e.1 = :l:1). (4.1.19)

We have to compare the two signs 6.1 and ] in (4.1.l8) and (4.1.19). The sign 6.1 is the
signature of the permutation JO'J° of {1, ..., n} obtained by writing the two increasing
multi-indices one after the other, first J and then JC. Similarly, since the boundary AM has
the orientation -d;u2 /\ ---A das", we deduce that the sign 6 J i s (-1) X (the signature of
the permutation J'O'JC of {2, ..., n}). Obviously

sign (JOJe) = sign (J'LJ'J°),

so that 6.1 of The proposition now follows from (4.1.18) and (4. 1. 19).

Corollary 4.1.54 (Gauss). Let ( M , g ) be a compact, oriented Riemann manifold with

boundary, and X a vector field on M . Then

div(X)dVg ( X , 17)duga 7
M am
where go = 9 lam.
 Riemannian Geometry 163

Proof. Denote by a the 1-form dual to X . We have

M
div(X)dVg
/M
l/\*d*oddvg
M
(1, *d * aldVg
M
(1, d*a)dVg

/am
alljldvge
am
(X, 17)duga .

Remark 4.1.55. The compactness assumption on M can be replaced with an integrability

condition on the forms c so that the previous results hold for noncompact manifolds as
well provided all the integrals are finite.

Definition 4.1.56. Let (M, g) be an oriented Riemann manifold. The geometric Laplacian
is the linear operator AM : COO(M) -> C°°(M) defined by

Am=d*df=- * d * df = - div(g1°ad fl-

A smooth function f on M satisfying the equation AM f = 0 is called harmonic.

Using Proposition 4.1.47 we deduce that in local coordinates (x1 7 13" , the geome-
ten's Laplacian takes the form
in 3
Am 7

where (QU ) denotes as usual the matrix inverse to (gm ) - Note that when g is the Euclidean
metric, then the geometers' Laplacian is
AS -(as + . . . + a2),
which differs from the physicists' Laplacian by a sign.

Corollary 4.1.57 (Green). Let ( M , g ) as in Proposition 4.].53, and f , g G C°°(M). Then

(f, AM9)M = (if» dg)m 7

am
and
@f
(f,AMg)M - (AMf»g)M 617' 9 am f, Q
317 am
Proof. The first equality follows immediately from the integration by parts formula
(Proposition 4.l.53), with a = f , and 5 = dg. The second identity is now obvious.

Exercise 4.1.58. (a) Prove that the only harmonic functions on a compact oriented Rie-
mann manifold M are the constant ones.
(b) I f u , f Q COO(M) are such that AmU, == f show that FM
f = 0.
164 Lectures on the Geometry of Manzfolds

Exercise 4.1.59. Denote by (u1, . . . ,u") the coordinates on the round sphere S" *-> RTL+l
obtained via the stereographic projection from the south pole.
(a) Show that the round metric 90 on S" is given in these coordinates by
4
90 1
-i-v2
{ (do -I- -|- (du")2 },

where 7,2 = (u1)2 -|- ---

+ ('u,")2.
(b) Show that the n-dimensional "area" of S" is

where F is Euler's Gamma function

1*(so zs-1€-tdz.
0

Hint: Use the "doubling formula"

7r1/2II(28) = 225-1II(8)II(s + 1/2),

and the classical Beta integrals (see [54], or [134], Chapter XII)
OO ,rn-1
d (F (H/2))2
0 ll+T2)n T' 2II(n)

Exercise 4.1.60. Consider the Killing form on U) (the Lie algebra of SU(2)) defined by

(x, Y) - t r X Y. -
(a) Show that the Killing form defines a bi-invariant metric on SU(2), and then compute the
volume of the group with respect to this metric. The group SU(2) is given the orientation
defined by 61 A 62 /\ et 6 A 3 ( 2 l 9 where €1 E Q ) are the Pauli matrices

i 0 1 0 i
€1 62 63
0 l 0 i 0

(b) Show that the trilinear form on Q) defined by

B(x,y,z) i <[X,y],z),

is skew-symmetric. In particular, B Q A3w(2)*.

(c) B has a unique extension as a left-invariant 3-form on SU (2) known as the Cartanform
on SU (2)) which we continue to denote by B. Compute I$m) B.
~
Hint: Use the natural diffeomorphism SU (2) = S3, and the computations in the previous
exercise.
 Riemannian Geometry 165

4.2 The Riemann curvature

Roughly speaking, a Riemann metric on a manifold has the effect of "giving a shape" to
the manifold. Thus, a very short (in diameter) manifold is different from a very long one,
and a large (in volume) manifold is different from a small one. However, there is a lot more
information encoded in the Riemann manifold than just its size. To recover it, we need to
look deeper in the structure, and go beyond the first order approximations we have used so
far.
The Riemann curvature tensor achieves just that. It is an object which is very rich in
information about the "shape" of a manifold, and loosely speaking, provides a second order
approximation to the geometry of the manifold. As Riemann himself observed, we do not
need to go beyond this order of approximation to recover all the information.
In this section we introduce the reader to the Riemann curvature tensor and its asso-
ciates. We will describe some special examples, and we will conclude with the Gauss-
Bonnet theorem which shows that the local object which is the Riemann curvature has
global effects.
l@Unless otherwise indicated, we will use Einstein 's summation convention.

4.2.1 Definitions and properties

Let ( M , g ) be a Riemann manifold, and denote by V the Levi-Civita connection.

Definition 4.2.1. The Riemann curvature is the tensor R R(g), defined as

R(9) = FW ),
where F(V) is the curvature of the Levi-Civita connection.

The Riemann curvature is a tensor R G §22(End ( T M ) ) explicitly defined by

R ( X , Y ) Z = [Vx,Vy]Z - V7Xy]Z.
In local coordinates al 7 . . , mn) we have the description
Refzjk3£ R(81, <gkl3..
In terms of the Christoffel symbols we have

Reink = 31F air +p" Jre Z

.
'rn
F771kl-"ij

Lowering the indices we get a new tensor

Rijké 9i'rnR<:£ g( R(é9k, @g)@j, at ) 9l3¢, R(a,,, 3037 ).

Theorem 4.2.2 (The symmetries of the curvature tensor). The Riemann curvature ten-
or R satisfies the following identities ( X , Y, Z, U, V Q Vect I I I ) .

(a) 9(R(X, Y) U, v ) -glRlly, x), U, V)-

(b) 9(R(X, Y) U, v) -g(R(X» YlVv U)-
166 Lectures on the Geometry of Manzfolds

(c) (The Is Bianchi identity)

R ( X , Y)Z -|- R(Z,X)Y -|- R(Y, Z)X = 0.
(d)

g(R(X,Y)U, V) i g(R(U, V)X,Y).

(e) (The 2nd Bianchi identity)

(vXR)(y, Z) -|- (VyRllZ» X) -|- lVzR)(X» Y) i 0.

In local coordinates the above identities have the form

Rig/e = -Rjfuqe = -Rijek, (4.2.la)

Rijké Rkévij 7 (4.2.1b)

Rake -I- Reek + Rkej 0, (4.2.lc)

(MV be + (WW in + (vkR)"' et :: 0. (4.2.1d)

Proof. (a) It follows immediately from the definition of R as an End(TM)-valued skew-

symmetric bilinear map (X, Y) »-> R(X, Y).
(b) This is a special case of Proposition 3.3. 14 about the curvature of connections compat-
ible with a metric.
(c) This is a local statement so we suffices to prove it holds on each coordinate patch. The
identity is also C'°°(M)-multilinear so it suffices to prove it only in the special case when
the vector fields X, Y, Z pairwise commute. The 1st Bianchi identity is then equivalent to
VxVyZ - VyVxZ + VzVxY - VxVzY + VyVzX - VzVyX = 0.
The identity now follows from the symmetry of the connection: V x Y = V}/X etc.
(d) We will use the following algebraic lemma ([80], Chapter 5).

Lemma 4.2.3. Let R : E X E X E X E -> R be a quadrilinear map on a real vector space

E. Define
S(X1,X2,X3,X4) R(X1,X2,X3,X4) + RlX27 X37 Xl7 X4) -|- R(X3,X1,X2, X4)-
If R satisfies the symmetry conditions
R(X1,X2,X3,X4) = -R(X2,X1,X3,X4)

R(X1, X2,X3,X4) Z -R(X1,X2,X4, X3),

then

R(X1>X2,X3,X4) - RlX37 X49 Xl7 X2)

l
5 {S(X1,X2,X3,X4) - S(X2,X3,X4» Xl)
-5(X3,X4»X1,X2) -|- S(X47X37X17
 Riemannian Geometry 167

The proof of the lemma is a straightforward (but tedious) computation which is left to
the reader. The Riemann curvature R = glx3, R(X1, x2>x4) satisfies the symmetries
required in the lemma and moreover, the 1st Bianchi identity shows that the associated
form S is identically zero. This concludes the proof of (d).
(e) This is the Bianchi identity we established for any linear connection (see Exer-
cise 3.3.24).

Exercise 4.2.4. Denote by (in of n-dimensional curvature tensors, i.e., tensors (Rijkll E
( ~ ) ® 4 satisfying the conditions,

Rilke = Rkeij = -Rjike, Rilke + Riejk + Riker = 0, 'v'i,j, k,€.

Prove that

kg) + n 1 n n + n
dim (in
=( 2 'l() 2 (2>((2l 1 )
(Hint: Consult [16], Chapter 1, Section G.)

Remark 4.2.5. The Riemann tensor R of a metric g is naturally a section o f A2T*M (X)

End_(TM). The tensor RE obtained from R by lowering the indices

Rt(X,Y,UV) 9(x, RIU, V)Y )

is a section of A2T*M (X) T*M. Using the isomorphism End_ T*M -> A2T*M defined
in (2.2.10) we obtain another tensor WR 6 COO(A2T*M (X) A2T*m). The tensors RE and
WR are related by the equality

RE ZLUR. (4.2.2)

Using the metric g we can identify A2T*M (X) A2T*M with the bundle End (A2T*m) .
The identity (d) in Theorem 4.2.2 states that the tensor R+ is symmetric when regarded as
an endomorphism of the bundle A2T*m.

The Riemann curvature tensor is the source of many important invariants associated to
a Riemann manifold. We begin by presenting the simplest ones.

Definition 4.2.6. Let ( M , g) be a Riemann manifold with curvature tensor R. Any two
vector fields X, Y on M define an endomorphism of T M by

U »-> R(U,X)Y

The Ricci curvature is the trace of this endomorphism, i.e.,

Ric (X, Y) = tr (U »-> R(U,X)Y).

We view it as a (0,2)-tensor (X, Y) »-> Ric (X, Y) E C°°(M).

168 Lectures on the Geometry of Manzfolds

If . . . , mn) are local coordinates on M and the curvature R has the local expression
(5131,

R ( R e ) then the Ricci curvature has the local description

Ric lRicz'jl ZR .
e
The symmetries of the Riemann curvature imply that Ric is a symmetric (0,2)-tensor (as
the metric).

Definition 4.2.7. The scalar curvature S of a Riemann manifold is the trace of the Ricci
tensor. In local coordinates, the scalar curvature is described by
S = Q" Rich = QURQ, , (4.2.3)
where ( g j ) is the inverse matrix of (90j)

Let ( M g ) be a Riemann manifold and p 6 M . For any linearly independent X, Y G

TpM set
(R(x,y)y,x)
K1,(x, Y )
IX A YI
7

where IX AY I denotes the Gramm determinant

(x,x) (Xy)
7
Ly,xl lay)
which is non-zero since X and Y are linearly independent. (IX A YI1/2 measures the area
of the parallelogram in TpM spanned by X and Y.)
If we choose local coordinates ( f ) near p than we set
Rijij
KfzjlPl = Kp(3x¢,6xj) I .. .. 2
927933 gig
Remark 4.2.8. (a) The sectional curvature determines the Riemann tensor. For a proof we
refer to Thm. 3.8 in [48].
(b) Given a metric g on a smooth manifold M, and a constant A > 0, we obtained a new,
rescaled metric PA = I\2g. A simple computation shows that the Christoffel symbols and
Riemann tensor of 9> are equal with the Christoffel symbols and the Riemann tensor of the
metric g. In particular, this implies
R1cg = Rlcg
However, the sectional curvatures are sensitive to metric rescaling.
For example, if g is the canonical metric on the 2-sphere of radius l in R3, then QA is
the induced metric on the 2-sphere of radius A in R3. Intuitively, the larger the constant A,
the less curved is the corresponding sphere.
In general, for any two linearly independent vectors X, Y Q Vect ( M ) we have
K9AlX7 y) = A-2Kg(x, y).
In particular, the scalar curvature changes by the same factor upon rescaling the metric.
If we thing of the metric as a quantity measured in meter2 , then the sectional curvatures
are measured in m€[€t-2.
 Riemannian Geometry 169

Exercise 4.2.9. Let x,y,z, W 6 TpM such that span(X,Y) span(Z,W) is a 2-

II
dimensional subspace of TpM prove that Kp(X, Y ) K,,(z, W)-

According to the above exercise the quantity K p ( X , Y) depends only upon the 2-plane
in TpM generated by X and Y. Thus Kp is in fact a function on Gp2 (p) the Grassmannian
of 2-dimensional subspaces of T p M .

Definition 4.2.10. The function Kp Gr 2(Pl -> R defined above is called the sectional
curvature o f M at p.

Exercise 4.2.11. Prove that

Gr 2(Ml = disjoint union of GF2 (p) p6 M
can be organized as a smooth fiber bundle over M with standard fiber Gr 2lRnl» n
dim M, such that, if M is a Riemann manifold, G r 2 ( M ) 9 (p; or) »-> Kp(7r' is a smooth
map.

4.2.2 Examples

..
Example 4.2.12. Consider again the situation discussed in Example 4.1.15. Thus, G is a
Lie group, and ( 7 ) is a metric on the Lie algebra LG satisfying
(ad(X)Y, z) = -<y, ad(X)Z).
..
In other words, ( , ) is the restriction of a bi-invariant metric in on G. We have shown that
the Levi-Civita connection of this metric is
_ l
VxY -[X,Y], V/X,Y€»CG.
- 2
We can now easily compute the curvature
l
R(X, Y ) Z Z {[X» [Y, ZH - [Y, [X, ZH} 2 HX, YL 2]
1 1 1 1
(Jacobi identity) = -Hx
4 1, Z] -|- ZW) [X, ZH - 4 [y, [x, nu 2[[X»Y]»Z]
1
4[[X7Y]7 Z]~
We deduce
l
( R ( x , Y)Z, W) 4(ux,y],2],w) = 4(a»d(Zl[X»Y]»W)
l
= -([X» YL ad(Z)W) 4 ([X, YL [Z, W])~
Now let 7r 6 G I I 2 log G) be a 2-dimensional subspace of TAG, for some g Q G. If (X, Y)
is an orthonormal basis of or, viewed as left invariant vector fields on G, then the sectional
curvature along or is

KQIW) 4([X»Y]»[X»Y]) > 0.

170 Lectures on the Geometry of Manzfolds

Denote the Killing form by I<,(X, Y) = -tr lad(X) ad(Y) ), To compute the Ricci cur-
vature we pick an orthonormal basis E1 , ,En o f L G . ForanyX = X E , Y = ye ET 6
LG we have

Ric ( x , y) l tr (Z»-> [[X»Z]»Y])

4
l 1
12 4 (adlyl[x, E,~],
l l
12 ([X» E ) [Y, 4
(adlXlEz', adlYlEj)

_l Z l tr
l
zI{lX7Yl_
4 (ad(y)ad(x) )

In particular, on a compact semisimple Lie group the Ricci curvature is a symmetric posi-
tive definite (0, 2)-tensor, and more precisely, it is a scalar multiple of the Killing metric.
We can now easily compute the scalar curvature. Using the same notations as above we
get
1
S (E ET 4
lislE»,;,

In particular, if G is a compact semisimple group and the metric is given by the Killing
form then the scalar curvature is
1
s(/~z) = 4 dim G.

Remark 4.2.13. Many problems in topology lead to a slightly more general situation than
the one discussed in the above example namely to metrics on Lie groups which are only
left invariant. Although the results are not as "crisp" as in the bi-invariant case many nice
things do happen. For details we refer to [96].

Example 4.2.14. Let M be a 2-dimensional Riemann manifold (surface), and consider

local coordinates on M, (ac1, 5132). Due to the symmetries of R,

Rijkl -Rijlk Rklij 7

we deduce that the only nontrivial component of the Riemann tensor is R R1212- The
sectional curvature is simply a function on M
l l
K -R1212
191
i 55(9), where 191 detlgijl-

In this case, the scalar K is known as the total curvature or the Gauss curvature of the
surface.
 Riemannian Geometry 171

In particular, if M is oriented, and the form do1 A day2 defines the orientation, we can
construct a 2-form
l l l
5(9) Kd'Ug g R1212d5U1 /\ d;132.
27rV 191
The 2-form e(g) is called the Euler form associated to the metric g. We want to emphasize
that this form is defined only when M is oriented.
We can rewrite this using the pfaffian construction of Subsection 2.2.4. The curvature
R is a 2-form with coefficients in the bundle of skew-symmetric endomorphisms of TM
SO we can write

R A ® dog, A
1 [ 0 £1212
0
vlgl R2112

Assume for simplicity that (xl , x2) are normal coordinates at a point q E M. Thus at q,
lgl = l since 31, <92 is an orthonormal basis of TqM. Hence, at q, dvg = del /\ d:z:2, and
l l
5(9) 9lR(317 62 )@2, glldilil /\ d:z:2 = -R1212d£E1
27r
/\ dx2.

Hence we can write

1 1
6(9) - P g l'Aldu9 - P 9(-RX
27r f 27r f
The Euler form has a very nice interpretation in terms of holonomy. Assume as before
that (ac1, 5122) are normal coordinates at q, and consider the square St = [0, 1/-t] X [0, \/2]
in the (ac1, 5122) plane. Denote the (counterclockwise) holonomy along 3st by 'Jet. This is
an orthogonal transformation of TqM, and with respect to the orthogonal basis (61, (92) of
T q M , it has a matrix description as

'L
[cos 9(t) - sin 9(t)
sin 9(t) cos 9(t)

The result in Subsection 3.3.4 can be rephrased as

sin§(t) = -i9<R<a1, a2>a2, al) + 0052),

so that

R1212 =
Hence R1212 is simply the infinitesimal angle measuring the infinitesimal rotation suffered
by <91 along St. We can think of the Euler form as a "density" of holonomy since it measures
the holonomy per elementary parallelogram.
172 Lectures on the Geometry of Manzfolds

4.2.3 Carton's moving frame method

This method was introduced by Elie Carman at the beginning of the 20th century. Cartan's
insight was that the local properties of a manifold equipped with a geometric structure can
be very well understood if one knows how the frames of the tangent bundle (compatible
with the geometric structure) vary from one point of the manifold to another. We will
begin our discussion with the model case of R". Throughout this subsection we will use
Einstein's convention.

Example 4.2.15. Consider an orthonormal moving frame on R", Xa xgp, Of

6'
l , ..., n, where (x1, . . . ,acn) are the usual Cartesian coordinates, and <2 := ex% Denote by
Baal the dual coframe, i.e., the moving frame of T*Rn* defined by
9°"(XB) i 6@~
The l-forms 90" measure the infinitesimal displacement of a point P with respect to the
frame (Xa). Note that the TM-valued l-form 9 = HOX0, is the differential of the identity
map 11 : R" -> R" expressed using the given moving frame.
Introduce the l-forms we"
defined by
dx5 = wgXo, (4.2.4)
where we set

xo,
..
II

3:27
We can form the matrix valued 1-form w = (we ) which measures the infinitesimal rotation
suffered by the moving frame (Xa) following the infinitesimal displacement az »-> cl: + do.
In particular, w = (we) is a skew-symmetric matrix since
0 d<Xo»X@> = (ca X0-XB> + ( X a , w 'X@)-
u

Since 9 = dll we deduce

0:=: d2 it ==do (we ® pa - 95 ® dx B (CW - 9" A we) ® xa,
and we can rewrite this as
doa* = H5 /\ "B 7 or do = wAs. (4.2.5)
Above, in the last equality we interpret w as a n X n matrix whose entries are 1-forms, and
9 as a column matrix, or a n X 1 matrix whose entries are 1-forms.
Using the equality d2 XI = 0 in (4.2.4) we deduce
dog = -W'Y /\ we, or equivalently, do wAs. (4.2.6)
The equations (4.2.5)-(4.2.6) are called the structural equations of the Euclidean space.
The significance of these structural equations will become evident in a little while.

We now try to perform the same computations on an arbitrary Riemann manifold

(M, g), d i M n. We choose a local orthonormal moving frame (Xa)1<0<n9 and
we construct similarly its dual coframe (9°"l1$o$n~ Unfortunately, there is no natural way
to define dxa to produce the forms we
entering the structural equations. We will find them
using a different (dual) search strategy.
 Riemannian Geometry 173

Proposition 4.2.16 (E. Cartan). There exists a collection of I-forms ((»J§")1$a,B$n

uniquely defined by the requirements

(a) we = -we, 'v'a,B.

(b) d90' 9B Aug, Va.

Proof. Since the collection of two forms (9'"" /\ 9B)1<0<,3<n

_ defines a local frame of
A2T*R'", there exist functions go
uniquely determined by the conditions

d90' - 2957
l a 9B /\ Q'Y 957 I -9vB'
Uniqueness. Suppose that there exist forms we satisfy the conditions (a) & (b) above. Then
there exist functions f such that

7.
I79
Then the condition (a) is equivalent to
(al) fg., = B
.fa'y'
while (b) gives
(bl) ff - f'yB I 92%-
The above two relations uniquely determine the f's in terms of the g's via a cyclic
permutation of the indices 0¢,/3,v
1 'Y
f§"~ - 2 Iggy + gfl/B0 - 905)° (4.2.7)

Existence. Define = fawe

Be 07, where the f 's are given by (4.2.7). We let the reader check
that the forms we
satisfy both (a) and (b).

The reader may now ask why go through all this trouble. What have we gained by
constructing the forms w, and after all, what is their significance?
To answer these questions, consider the Levi-Civita connection V. Define of by

VX (QBXQ.
Hence
VxXB I CTJg(X)Xo,
Since V is compatible with the Riemann metric, we deduce in standard manner that (mg
_ B

The differential of gO( can be computed in terms of the Levi-Civita connection (see
Subsection 4.1.5), and we have

d9°'(X07X'yl XB9Q(Xw7 - Xv9°"(X5) - 9 " ( v X x v > + @alVx,yXgl

(use 9°'(X@) 65 - const) = -9 a (°Jv(XB)X6)+9
6 a 6
((»J@(XvlX6l
TDglx'yl - 7 (XI) = (HB /\@gllXB7X'Yl-
174 Lectures on the Geometry of Manzfolds

Thus the ¢?J's satisfy both conditions (a) and (b) of Proposition 4.2. 16 so that we must have
cog = we.
In other words, the matrix valued l-form (wall is the l-form associated to the Levi-Civita
connection in this local moving frame. In particular, using the computation in Exam-
ple 3.3.12 we deduce that the 2-form
Q (do-I-wAw)
is the Riemannian curvature of g. The Carton structural equations of a Riemann manifold
take the form
d9=-wA9, do+wAw=Q. (4.2.8)
Comparing these with the Euclidean structural equations we deduce another interpretation
of the Riemann curvature: it measures "the distance" between the given Riemann metric
and the Euclidean one". We refer to [123] for more details on this aspect of the Riemann
tensor.
The technique of orthonormal frames is extremely versatile in concrete computations.

Example 4.2.17. We will use the moving frame method to compute the curvature of the
hyperbolic plane, i.e., the upper half space
H+ I
{(f1%y); y > 0}
endowed with the metric g = y'2(d;/:2 -|- dy2l.
The pair (We, y<'9y ) is an orthonormal moving frame, and (Hx i d , by
y
idol is its
dual coframe. We compute easily
1
do"' d -do:
y
l
_do 1
( ) ( )
c>

doy d --do: A 9rr;1

II

II

y y
Thus the connection 1-form in this local moving frame is
0
J] day.
3
II

1 3/
y

Note that w /\ w 0. Using the structural equations (4.28) we deduce that the Riemann
curvature is
1
0 y2 0 1
so do 1
0
dyAda3
1 0
9x3 A 93/.
y2
llllllllll

The Gauss curvature is

l y4 / 1 \
K glglax) <8y)<8y» <8'm)
Exercise 4.2.18. Suppose (M, g) is a Riemann manifold, and 'LL 6 C°°(M). Define a new
metric gu := 62/g. Using the moving frames method, describe the scalar curvature of g'LL
in terms of u and the scalar curvature of g.
 Riemannian Geometry 175

4.2.4 The geometry of submanfolds

We now want to apply Cartan's method of moving frames to discuss the the local geometry
of submanifolds of a Riemann manifold.
Let (M, g) be a Riemann manifold of dimension m, and S a k-dimensional submanifold
in M. The restriction of g to S induces a Riemann metric is on S. We want to analyze the
relationship between the Riemann geometry of M (assumed to be known) and the geometry
of S with the induced metric.
Denote by VM (respectively VS) the Levi-Civita connection of (M, g) (respectively
of (S, go). The metric g produces an orthogonal splitting of vector bundles

(TM)l5~= T S ® NsS.
The Ns is called the normal bundle of S --> M, and its is the orthogonal complement of
T S in (TM)15. Thus, a section of ( T M ) 15, that is a vector field X of M along S, splits
into two components: a tangential component XT, and a normal component, X".
Now choose a local orthonormal moving frame (Xl, ..., Xi; X;,,+1, ..., Xm) such that
the first k vectors lX17 .C.7Xk) are tangent to S. Denote the dual coframe by (0°')1sa$tn.
Note that

9"'l$=0 f o r a > k .

Denote by (#§)» (1 < a, B g m) the connection l-forms associated to VM by this frame,

and let (to, (l < a, B g k) be the connection 1-forms of VS corresponding to the frame
(Xl, . . . ,Xi1.1). We will analyze the structural equations of M restricted to S <-> M .

d9°=9@A/1g 1 _<a,/3 <m. (4.2.9)

We distinguish two situations.

A. l < a < k. Since 95 l$= 0 for B > k the equality (4.2.9) yields
k
dH' H IW0, u 5 : - # B l S@,BSk-
/3=1

The uniqueness part of Proposition 4.2.16 implies that along S

08=/3 1 SO <k.
This can be equivalently rephrased as

vey = (VXMYV ex, Y E Vect (s). (4.2.l0)

B. k < a _< m. We deduce

k
5 A/1g.
B 1

At this point we want to use the following elementary result.

176 Lectures on the Geometry of Manzfolds

Lemma 4.2.19 (Caftan). Let V be a d-dimensional real vector space and consider p lin-
early independent elements w1, ,wp 6 A1V, p < d. If91, ,up 6 A1V are such
that
p

26 I\(,4Ji
'Z 1
0,

then there exist scalars All, l < i , j < p such that A = Ali and
p

$i Aijwj .
j=1

Proof. Extend the collection lwi)1$i<p to El basis (6,k)1<k<d of A1V. Then

d

91 Aikwk
k=1

and
p d

29
c>

/\Low
II

I\(.4J¢ Aikwk
'£=1 k = 1
p

1<k<'i<p
(A - Akfzlwk A w + EZ
i 1 k>p
AM/Jk I\(.4J¢.

This proves that A = 0 for k > p and A = As for l i, k _ p.

Using Caftan lemma we can find smooth functions f8,v, A > k , 1 _< B, by < k satisfying
A _ A
and #BA -
_ A
far - f f -
Now define
N f8v 9@ ®97 ®XA»
We can view N as a symmetric bilinear map
Vect (S) X Vect (S) -> C'°°(N$).
If U, V 6 Vect (S)

U= u@x@ = 9B(u)x@ 1 S B S k,
and

V =WX = m(V)x'y l _< v g k ,

then
k

n(U,v> E
A>k
;B,v=1 f8@@(u)@9(v)
) x*=
2(2
A>k B
~2(v)u5
) X .
 Riemannian Geometry 177

The last term is precisely the normal component of vI U. We have thus proved the fol-
lowing equality.

(v/')" = NIU, v) NU/1 U) (veyI V) V

(4.2.l 1)

The map N is called the Zndfundamentalform2 of S -> M.

There is an alternative way of looking at N. Choose

U , V € V e c t ( S ) , N € COO(Nsl.

If we write g(¢ 7 ¢ )=( .. )»

7
then

(~(u,v>,n) Z ( (v%v)" , n)=(v£¥'vcn)

=U-(V,N) -(v,vgyfn)=-(v,lvgyIn)).
We have thus established

(v, (vgyIn)T) = (n(U,v),n) = (n(v,U),n) = -(U, (wvv)- (42.12)

The second fundamental form can be used to determine a relationship between the curva-
ture of M and that of S. More precisely we have the following celebrated result.

Theorem Egregium (Gauss). Let RM (resp. RS) denote the Riemann curvature
of (M, g) (resp. (S, g Is)» Then for any X, Y, Z, T E Vect (S) we have

(RM
Ix
'Y)z
aT) =<8S
-I-
Y IZNT
>
II>x2) (
)
T N
(x l ' (Ya e
(4.2.13)
( ' ) - ( N7

D
Proof. Note that YT
VXMY = viiI + n(x, y).
We have

RM(X, Y)X = WxM, vym]z - V y Z

= VxMW?Z + MY, 2 ) ) - W" ( viz + MX, z) ) - Vi 12 - NIX, ii, z).

Take the inner product with T of both sides above. Since N(¢, .) is N5-valued, we deduce
using (4.2.10)-(4.2.l2)

(RM(x, y)z, T I (VXMV§,Z,T) + (VXMN(Y, z ) , T )

-(v¥v§K'z T ) - (VYMMX, z>,T) - (Vi)<»Y]Z7T)
ii, VW, T (My, z),n(x,T) ) + (Mx, z>,n(y, T) ) - ( V < , y ] Z 7 T ) ~
This is precisely the equality (4.2.13).
2 The first fundamental form is the induced metric.
178 Lectures on the Geometry of Manzfolds

The above result is especially interesting when S is a transversally oriented hypersur-

face, i.e., S is a a codimension 1 submanifold such that the normal bundle Ns is trivial.3
Pick an orthonormal frame n of Ns, i.e., a length 1 section of Ns, and choose an orthonor-
mal moving frame (Xl, ..., XT7,_1) of TS.
Then (Xl, ..., X m _ 1 , n) is an orthonormal moving frame of ( T M ) 15, and the second
fundamental form is completely described by
Nn(X,Y) (N(X,Y),n).

..
II
N," is a bona-fide symmetric bilinear form, and moreover, according to (4.2. 12) we have
n,,(x,y)=-(vXMn,y) (VyMN,X), (4.2.14)
In this case, Gauss formula becomes

(RS(x, y>z, T ) (RM(x, y)z, T ) N,,,(X, Z)

only, Z)
n,,,<x, T)
only, T)
n •

Let us further specialize, and assume M Run, so S is a hypersurface in Run. Then

(RS(x, y>z, T )
nn IX, T) n,,(x, z)
(4.2.l5)
nn Ly, T) Nn(Y, 2)
In particular, the sectional curvature along the plane spanned by X, Y is
(R$(X,Y)Y,X) = N,-,(X,X) - No(Y,Y) - INn(X7Y)l 2
Remark 4.2.20. (a) The equality (4.2.l5) is a truly remarkable result. On the right-hand-
side we have an extrinsic term (it depends on the "space surrounding S"), while in the
left-hand-side we have a purely intrinsic term (which is defined entirely in terms of the
internal geometry of S). Historically, the extrinsic term was discovered first (by Gauss),
and very much to Gauss surprise ('?!'?) one does not need to look outside S to compute
it. This marked the beginning of a new era in geometry. It changed dramatically the way
people looked at manifolds and thus it fully deserves the name of The Golden (egregium)
Theorem of Geometry.
We can now explain rigorously why we cannot wrap a plane canvas around the sphere.
Notice that, when we deform a plane canvas, the only thing that changes is the extrinsic
geometry, while the intrinsic geometry is not changed since the lengths of the"fibers" stays
the same. Thus, any intrinsic quantity is invariant under "bending". In particular, no matter
how we deform the plane canvas we will always get a surface with Gauss curvature 0 which
cannot be wrapped on a surface of constant positive curvature! Gauss himself called the
total curvature a"bending invariant".

Example 4.2.21. Suppose that S is a co-oriented hypersurface of Run. The choice of CO-
orienting unit normal vector field n can be viewed a smooth map n : S -> STT1-1 where
S771-1 denotes the unit sphere in Run. Observe that, for any p E S, we have an equality of

subspaces in RM
TpS I Trap)
Sm 1

3 Locally, all hypersurfaces are transversally oriented since Ns is locally trivial by definition.
 Riemannian Geometry 179

The second fundamental form N of S is symmetric bilinear form on TpS. Using the metric
on TpS we can identify it with a symmetric operator

N : T,,s -> Tps -_ To) S771 1

7 (nx,Y ) N(X, Y).

Using the equality (4.2.14) we deduce

{NX,Y) I No(X,Y) = -(VIM No X ) = - ( V Na Y

This shows that - N is the differential of the map

§ : S - > S M - 1 7 S3»p»->n(p)-E
Sm 1

The map 9 is called the Gauss map of the hypersurface S C lR71. The differential of the
Gauss map is known as the shape operator of the hypersurface. We will have more to say
about the Gauss map in Section 9.2.1.
Suppose now that S is the sphere of radius 7° in RM centered at the origin, and n is the
our unit normal vector field along S. In this case the Gauss map is
l
S9;z:»->-:z3€SM 1
T'

Thus
l
Do 9 -].TxS.
T'

1
We deduce from (4.2.15) that the sectional curvature of S along any 2-plane is F'

Example 4.2.22. (Quadrics in R3). Let A : R3 -> R3 be a self-adjoint, invertible linear

operator with at least one positive eigenvalue. This implies the quadric

Q A = { u € R 3 ; (Au,u)=1},

is nonempty and smooth (use the implicit function theorem to check this). Let 'LLO 6 QA.
Then

TuoQA={$€R3; ( A ) 0} I./4'LL()l-L»

QA is a transversally oriented hypersurface in R3 since the map QA 9 u »-> Au defines a

1
nowhere vanishing section of the normal bundle. Set n = IAuI Au.
Consider an orthonormal frame (so, e l , 62 of R3 such that 60 = n(u0). Denote the
Cartesian coordinates in R3 with respect to this frame by (x0 ,ac1,a:2), and set <9i := 3as%.
Extend (el, et) to a local moving frame of TEA near U0 .
The second fundamental form of QA at U 0 is

Nnlai7 ajl = ( 3i"I,7 as MY

I
180 Lectures on the Geometry of Manzfolds

We compute
Au 1
<9i n <2
( IAuI) Au
)
- 1 / 2 ) Au 'Au'

( QAm, Au ) 1
A
IAul3/2 'LL IAuI
Hence
|
n, "j ) U0
1
Aa
lAural (
Nnl3i)3jllu0

1
Adj ) |u o : 1
et? A6 3. ) .
lAural IA'LLol (4.2.l6)

We can now compute the Gaussian curvature at U0 .

l ( A € 1 , 61 ) ( A 6 1 , 62 )
Kuo
IAuol2 (/4627 81 ) ( A 6 2 , 62 )
In particular, when A 7°2/ so that QA is the round sphere of radius T' we deduce
1
K Vlul T.

Thus, the round sphere has constant positive curvature.

Example 4.2.23. (Gauss). Let E be a transversally oriented, compact surface in R3, e.g.,
a connected sum of a finite number of tori. Note that the Whitney sum $2 ® T2 is the
trivial bundle R32. We orient No: such that
orientation No: /\ orientation TO = orientation R3.
Let n be the unit section of No: defining the above orientation. The Gauss map
9:2->S2={u€]R3; l'u,l=1}, 2933+->n(;ulES2 7

depends on how Z) is embedded in R3 so it is an extrinsic object. Denote by Nn the second

fundamental form of 2 --> R3, and let (x1, as2 l be normal coordinates at q 6 Z] such that
orientation TqX] - 31 /\ 32.
'

Consider the Euler form so on Z] with the metric induced by the Euclidean metric in R3
Then, taking into account our orientation conventions, we have
[N,,(31, 31) Nnlala 32)
27T€2(31, 32) = R1212 = det (4.2.17)
:Nnla27 31) :Nnla27 32)
Now notice that

37, nn Z
-N,,(<92,31)31 - N,,,(3¢,<92)32.
We can think of n, <91 lq and <92 lq as defining a (positively oriented) frame of R3. The last
equality can be rephrased by saying that the derivative of the Gauss map
9* . Tqz -> To(q)$2
 Riemannian Geometry 181

acts according to
<1 P-> -Nn(6i7 <91)<91 - Nnlai; c92)<92.
Hence
-Nealy 31) -Nnlgl <92)
§*d'u$2 = (det 9*)dvz = det
[-W, 31)
' d
-n,,(@2,@2) UP
d Nnlal 7 31)
et Nn(32,31)
Nu law 32)
Nu (55, 82)
] dUE 2W€2.

Hence
62 = 92d0S2 . (4.2.l8)

This is one form of the celebrated Gauss-Bonnet theorem. We will have more to say about
it in Subsection 4.2.6
Note that the last equality offers yet another interpretation of the Gauss curvature. From
this point of view the curvature is a "distortion factor". The Gauss map "stretches" an
infinitesimal parallelogram to some infinitesimal region on the unit sphere. The Gauss
curvature describes by what factor the area of this parallelogram was changed. In Chapter 9
we will investigate in greater detail the Gauss map o f arbitrary submanifolds of a Euclidean
space.

Example 4.2.24. Suppose that S is a k-dimensional submanifold of RM equipped with the

induced metric. Fix a point Po 6 S and (u1, . . . ,u":) are normal coordinates on S at Po.
The point on S with coordinates (u1, . . . ,1/") can be identified with a point az = a:(u) E
RT". Classically a:('u,) is called the position vector of the corresponding point in S.
The tangent spaces TpS can be identified with a subspaces of Run. As such, the frame
moving frame l6uil1<_i<k along S can be identified with the collection <M/1)1$z$k of
vectors in RM.
Since the coordinates u are normal at Po we deduce that
Vg a, = 0 at Po .
//
In other words, at Po» the vectors : z a:Z'uJ are normal to TPos. Hence
Jan'
l/
NPo 7 3149 ) .
Using (4.2.13) we deduce

R¢jke(pol = (3k»Rp0(3i»3j)3€) = ( $ $ j I e ) ' <384»38iIk>-

If we define
K : S XS->]R, K ( p , q ) :
( p, q ) 7
then in a neighborhood of (Po, Pol we can view K as a function of two pairs of variables
u = (up, . . .,u"') and 'U = (pa . , up) and we can rewrite the last equality as

Rijké
34. . K(u,0)
'L1,1"l.l,k'1).7'I.)€
I
as u g h ] u k K l u , U)
u 'U O
(4.2.19) I u v 0°
182 Lectures on the Geometry of Manzfolds

4.2.5 Correlators and their geometry

The concept of correlate has its origin in probability (see Remark 4.2.30) but we will not
dwell too much on this aspect. This concept sheds a new light on the results about the
geometry of submanifolds discussed in the previous subsection but, unfortunately, few ge-
ometers are familiar with it. This subsection aims to introduce the audience to this concept
and some of its applications. We follow the presentation in [104] and [105].
First some terminology. Given smooth bundles E -> Mi, z' = 0, 1, over the smooth
manifolds Mo, M1 we denote by E0 ET E1 and E0 IZ E1 the bundles

tfolEol e 7T1(E1)» t0(E0l (X) W1lE1)->M0 X Ml,

where 711- : M0 X M1 -> Mi denotes the canonical projection.
For f Q C°°(M0 X Ml) we denote by a y f the section of T*M0 [Z T*M1 defined
as follows. For Po 6 M0 and Pl Q U1 choose local coordinates ac near P0 and y near PI.
Then

@2 ,yfl U0 XU1
a2 , fd (X) dye .
M
While the above description is local, we leave the reader to convince him/herself that the
resulting object is a well defined global section of T*M0 X T*M1.

Definition 4.2.25. A correlate on the smooth manifold M is a smooth function

K :M X M -> R
satisfying the following conditions.

(i) The function K is symmetric, i.e., K ( p , q ) = K ( q , p ) , V ( p , q ) € M X M .

(ii) The symmetric bilinear form l a 2 , y K l p , q i1>=q E T p M ® T p M is positive definite for
any p 6 M .

Definition 4.2.26. Let M be smooth manifold and V C cOO (M) a vector space of smooth
functions on M. We say that V is ample if, for any p E M the linear map
V 91)l->dv(p) QTpM
is onto.

Example 4.2.27. Suppose that we are given smooth functions

w/,...,wn.m->R
such that span{ \I/1,...,\I/N}is an ample subspace o f C ° ° ( M ) . DefineK : M X M -> R
N

K( p,q ) ZW (p)\I/¢(q)»
'i 1
ii VP, q E M.
 Riemannian Geometry 183

Clearly, condition (i) is satisfied. Next, observe that

N
2
6(B,;l/KPaq
'i 1

and, for any p E M and any X Q T p M , we have

N
329Kpp(X, X ) d (X)2 2 0.
ii 1

Since the map

span{\P1,...,\I/N} 9 \I/ »-> d\I1(p) E TpM

is onto, for any X Q TpM \ {0} there exists i Q {1, . . . , N} such that dpI/i(X) 96 0. This
proves that a2 y K p h , is positive definite.
As a special case of this construction suppose that M is embedded in RN . Denote by
e 1, . . . , e N the canonical basis of Hom(]RN , R). We denote by \I/¢ the restriction of e'i to
M . In this case
N N
KM q) a m » ) w (q)
ii 1
2 e'(p)e'( q )
'i l
(19, q IN 7 (4.2.20)

where (-, - )R- denotes the canonical inner product in RN . The ampleness follows from
the fact that the natural map Tp R" -> TpM is onto, Vp G M .
We will refer to the correlate constructed in this fashion as the correlate associated
to an embedding. More generally if F : M -> R" is an immersion, then the function
K;MxM->R, K(p,q ) lF(p),F( q ) )
is also a correlator.

Suppose that M is a smooth m-dimensional manifold and K is a correlate on M. For

simplicity we denote by E the tangent bundle TM. We set

C == a2 yK 6 c°°(E* m E*).
From the definition of a correlate we see that C defines a metric (-, -In on E. Note that
when K is the correlate determined by an embedding, the metric defined by C is none
other than the induced metric.
For any az, y Q M we can identify C , 6 Ern (X) Ey with an element of
Saw G Hom(Ey, E;).

Since the bilinear form C m a : : E m XEQ,->R is nondegenerate, the induced map

Sm,m : EQ, -> E;
is an isomorphism. We set
T(a:, y) ¢= s,,,} s , E H0>(Ey,E,,,,).
184 Lectures on the Geometry of Manzfolds

We will refer to T(a:, y) as the tunneling map associated to the correlate K. Note that
T:1:,m $1E,,.
_ _ in a neighborhood 6 of P0 in
Fix a point P0 E M and local coordinates (£!3i)1<i<m
M. Suppose that §(x) = l e l ) ) 1 s a m is a local frame of El@). We regard it as an
isomorphism of bundles @(§) : RE -> El@). The tunnelings define a smooth map
-1
To : @ X 6 -> ENdlRm)) Ti(a 7 y) I <I>m(§) T=1=,yq>yl§)~

If we denote by ( e ) the dual moving frame of E*l@, then Cm,U is described by the
matrix S£(;I3, y) ( S06/3(a;7 )1 <0£7,@<'f7l

Co:,y 8013 (at, y)6°" (X) 65 lYla

GLB

0O,5l£U,Y) = <8le(m)<9e@(y)K(:u7

The fact that Cx,:1: is a nondegenerate symmetric bilinear form implies that the matrix
Se(:z:) So(:z:, ac) is invertible. Let

s(=v,=)'1 = ( _gag )1<0,,5<M

Then TglXv few) is represented by the matrix with entries

TE" (as, y) sOWY IZ)SAQ (as, y).

'y,A

If we choose the local frame Q to be orthonormal with respect to the metric ( 9 )c, then

8o,0(€67 f1¢2 = 60,5.

Hence, in this case T£(a2, y) is represented by the matrix 30¢,3lX7Yl 1<a,B<m

'
l l
If we denote by T w y 6 Hom(Ey, E:zz) the adjoint of T a , with respect to the metric
( v ac, then the symmetry of K implies that

So¢,3lXvYl = S,3oll3/valv

i.e.,

Typ;
_ Tfray'
*

Lemma 4.2.28. Let Q be a frame of E | @ orthonormal with respect to the co rrelator metric
( » l c - Then, for any i = 1, . . . ,m, the operator

3ac T§l1°7?!ll1:=y : y -> y 7

is skew-symmetric.
 Riemannian Geometry 185

Proof. We identify 6 X 6 with an open neighborhood of (0, 0) Q RM x RM with co-

ordinates ( ; v e i l Introduce new coordinates Z'i := x'i - y , So := ask -|- yj, so that
a = (9 -|- 3S1. We view the map T(§) as depending on the variables z, S. Note that
T£(0, 3) = 11, TO(-z,s) To(z,s)* 7 Vi, s.

We deduce that
aSI<0, S) = aSI(o, s)* = 0,

é9x1T£(0, So = 3Z1T£(0, s) + a T§(0, s) = 32 T£(0, 3),

*
(3x;T§(0, S) ) = é9x1T§(0, so* = -aZT;(0, s) + aSmlo, S) = -3x1T§(0, so.

Given a coordinate neighborhood with coordinates ( x ) and a local frame Q defining and
isomorphism vector bundles Q>(§) : Kg -> Elga as above, we define the endomorphisms

Fi(§) : KG -> R23 'i = 1 , . . . , m = d i m M ,

(4.2.21)
Filély Z - 3x.Tgl377Y)|:1r=y-

We obtain a 1-form with matrix coefficients I`(g) E Fi léldy The operator

V; =d+ I`(g) (4.2.22)

is then a connection on go, When Q is orthonormal this connection is also compatible

with the natural metric on this trivial bundle. The isomorphism <I>(§) induces a connection
@(e)*V§ on El@.
Suppose that <I>(iMBE -> Elf is another isomorphism determined by the frame _f of
Ea' related to Q via a transition map
g : 6 -> GLm(R), g.

Then

Ti(2% y) = 9-1(;3)T§(22» y)9(y)-

We denote by do the differential with respect to the x variable. We deduce

o (fL y -doT( f ) x , y l x = y
-( dx9l1(;vl )x=y • T§lYv y) ~9(y) 9-1 (33) ( dmT§(a3, y)) Ix=yg(:u)

9'1(y)d9(y)9'1(y) -9(y) + 9`1(y)F(§)y9(yl

9(y)'1d9(yl + 9'1(y)F(§)y9(y)-
Thus

F(§ 9) = stldg + 9-1F(§)9-

186 Lectures on the Geometry of Manzfolds

Proposition 3.3.5 shows that the collection of locally defined 1-forms Hg), Q local frame of
E, defines a connection VC on E. This connection is compatible with the metric 7 -)c
because the local connections VI are such when Q is an orthonormal frame.

Proposition 4.2.29. Any correlate K on M induces a canonical metric ( - , - )c on T M .

The associated Levi-Civita connection coincides with the connection V C described in a
local frame Q by

vC d+ 7

where the m y m-matrix F Q ) is given by (4.2.2]).

Proof. It suffices to prove that the connection VC is symmetric, i.e., its torsion is 0. Once
we choose local coordinates ( x ) on 6 we have a natural frame of El @

t o ; 'a (mot

With these choices Cx'y is represented by the matrix S§(a:, zu) with entries

8a@(x, y) = c92'*,:u5K(a2, y).

We denote by x) the inverse of (305(3?= wll- Then F1(X) is an m x m matrix

`
a
F UI Fmm) )1$0,BS"1°

File,£ul = -c9:/52 l5e(x,x)-15£(,;,y)ll

-3:IJ; S2(a:, $2)-1 Se(;z3, 1/3) S§(;z3, 5C>-16x1S£(£U7 311

S§(:1;°,;r3l'1(9Q3 S£(;r3,a;°l - S£(;u,;ul'13x. Sgl5U)Yly=x

SglXv a;')'1 ex S£(a2, yl -|-

1 SglXa y) ) s , l-1a,se(x, y>{
Sglx, :z;'l-1<8y Sola;', y) |
yas'

The last equality shows that

F s"~()@3yyK(w,y)u=y» (4.2.23)
'Y

Note that this implies that

kg F

which shows that VC is a torsion free connection.

 Riemannian Geometry 187

Let us compute curvature RC' of VC at a point Po. Fix a small coordinate neighborhood
6" with coordinates (w) such that gelp of = 0, W = 1, . . . , m . Using a local frame Q we
can write
Re RMS, M /\ dxj E End(1Re.m) (X) 92(6
1<i<j<m

where

Rig (Q, 11) I ARm (Q, l 33) 1<a,,3<'rn7 Z

a
Rgfj ea
Rijég,

is given by

Run (Q, y) = C9xJli(§, 33) - am FAQ, xi) + [F¢(§, ¢vl» Pa (Q, I) ] -

Using the local frame Q we represent S§(x, y) as an m X m-matrix

SO(w,y) = (sa@(11»y) )1<_a,Q<_m' sa5(=v,y) = O / ( Q o ( w ) » Qe(y) )~

Then T£(:z:, y) = So(:c, :z:)-1S£(x, y), and

new) = s(w,x)-l@ysQ(w,y)4y=,

3,:2 I`i (Q, CC) = a l s(a¢, up)-lay, s ( x , al ynez )

= -556, )-1 la,s(x, ) IS, )-lays lx, al(,,=y
+5(@¢, LUl-1@x» la ii Se(:z:, y)l=y )
= -5<¢¢,>~1(a,5e(@¢,y> + ay,5(@¢,y>14 x=y s§(,¢ >-1a y,5e( x,/>l
+s(¢¢, a - 1 a 2 z y , s ( x , al ' x = y + SQ(a:, x)-1a§,ny, sum, y) Ix=y°
Suppose now that ea 5 0 and Q is orthonormal at P0 where at y = 0. More precisely,
assume that

C0,0l3xo 7 <9 to) = 60Ba L > 0.

Then
sl0,0l-1 = 11,

and

<9,,j F
la $e (so, y) lx=y=03y3 $e (so, y)l ac y 0 -I- 3x5 So(:u, y) l$=y=03Y' Se (as, y) | x y 0

+(a2,y, $el5U» y)l as y O

a2
S£(:c, y) I as y 0 7

and

[Fi (Q, Of? Fi (Q, 0)] Z iii So(:c, y) l;,8=y=06Y-7 S6 (so, y) I as y O

39/ Se (33, y) lx= y= 06 Y
188 Lectures on the Geometry of Manzfolds

If, additionally, the frame ( ) is synchronous at Po» i.e.,

kg (Po) = 0, w, Q ,
then
c9yj Se (33, y) I ac y 0 FQ, Of I 0,
SO that
Raj (Q, 0) I au, Se(ac°, y) I x y 0
aw
2
Se(ac°, y)l as y 0

and
(4.2.24)

This should be compared with (4.2. 19).

Remark 4.2.30. We cannot end this section without giving the reader an idea about the
probabilistic flavor of the concept of correlate. To keep things simple suppose that
K :M X M -> R
is the correlate determined by an embedding M <-> RN .
Denote by e1, . . . , eN the canonical basis of the dual Hom(RN , R). Denote by 'Pa
the restriction to M of the linear function ea : RN -> R. Now choose N independent
random variables Ai, . . . , An with mean 0 and variance 14 and form the random linear
combination
N
\11 ZAQ/Q.
a 1
The random object \I/ is known as a random function or random field on M . The value of
\I/ at a fixed point p is the random variable

WP) I 2 Aowo(p
a

Its mean or expectation is

Z \P1/:(P)E[A0 1
<5

IE[\P(p)]
II

The correlation or covariance kernel of the random function \If is the function
K :M M -> R, K ( p, q ) = IE[*P(pl\P(q)].
X

In other words, K ( p , q ) computes the covariance of the random variables \If(pl and \I/(q)
Note that

K(p,q IE ( 2 A ~ v ( p ) ) ICA B Q/ B( q I
>}
a B
h i m pwe( p) M A Y B 1 §:\1'o(q) Ql l <1-
a
04,5 04,5 Sag

This implies that ]E[AiAj] = 6»»;j, 1 S 'LJ S N, where IE[X ] denotes the expectation of the random
variable X .
 Riemannian Geometry 189

In other words, K (p, q) coincides with what we dubbed the correlate determined by the
embedding M + RN .
Given two vector fields X, Y Q Vect (M) we obtain two new random functions X \I/ -
-
and Y \I/. The metric g defined by the correlate K has the probabilistic description

QpIXp) Ye) = IH (X\P(p)) ' (Y\P(p))1-

To see this it suffices to work in local coordinates ( l near p such that x (p) 0 and
assume X = (Le, Y = 83 . Then
1P3[ (X\P(p)) a (Y\P(p)) ]
IE
K Z A a a x l f ( 0 ) Z Aga
Of

3x,1I1ol0)3mjq,l0lE[AoAB ]
)( Of
\I/ B(0)
M
ii
@pa(0)a 1,,(0) = <8l2»ykK(x,y){ at y 0'
a

The curvature formula (4.2.24) also has a probabilistic interpretation, but it is a bit
more elaborate. To describe it we need to introduce an appropriate language. Given an
m-dimensional vector space V and integers 0 < p, q < m we set
As»qv* == APP* (X) Arv*
There exists a natural bilinear product
@ : Apmq0v* X AP1,<11V* _) APO-I-P1»qO+q1V*

uniquely determined by the requirement

lc*f0 (X) 5 0 l ® ( C * f 1 (X) 51) ICYO/\O£1) (X) l50/\51l-

For example AUV* is the space of bilinear maps V X V->]R.If€17 78m is a basis of
V, e1, . . . , et is the dual basis of V* and
B Z 5 3 8 1 (X) ey
"As
Then

B ®B
in
bkdmf' (X) dick
) (Et
@
j,e
jgd33j ® do"

b k b j d m A dxj ® dick A do" Q A2,2V*.

iJJ<:,£

The Hessian of a function f on an arbitrary manifold is intrinsically well defined only at the
critical points of f . However, on a Riemann manifold (M, g) with Levi-Civita connection
V, we can define the Hessian of the function f to be the map
Hf : V e c t ( M ) X V e c t ( M ) - > COO(M), Hf(X7Yl ¢= XYf - (vxy)f.
190 Lectures on the Geometry of Manzfolds

The symmetry of the Levi-Civita connection implies that Hf is indeed a symmetric tensor.
If ( ) are normal coordinates at a point p, then in these coordinates, the Hessian of f at p
is given by the familiar formula
2
H? M
axixj f ( p ldl (X) dxj Q A1,1TpM.

Note that

Hjp@Hfp <9%,f(0)@% f (0lda3 /\ dxj (X) duek /\ d o . (4,225)

M,/<>,@

If we specialize to the case when g is the metric associated to the correlate

K(p,q) = Z\Pa(p) Q/ Q(<1)

o

we deduce from (42.25) that

2 HE
a
@ Ha; i 2
i,j,k,€
a4,xky,y K(x, y)'
x=y=0
do A dace (X) do," A do"

i<j,k,£
l34'rrkny_7ygKlI, yl' x y 0
4
3 x J x k y 1 y e Kl33, y)l Qjzyz
0 I do /\ dxj (X) dok A doe

(4.2.24) Rkejd A data ® dark A dx£<4'2'1">2 R k e j d i A dczzj (X) dank A do"

i<j,k,£ i<j,k<€
(4.2.1b) - -
= 2 R¢jk1 d.cUz /\ d.ClZ'°7 ® dick A do" = -2 oR,dx'" A M,
i<j,k<£ k<£

Z(*JR

where Run = R(3xk, 3,,,@) 6 End_(TM) and A »-> WA is the isomorphisms

End_(TM) -> AQTpM, AIX, y ) = 9,,(Ax, y )
described in (2.2.10).
Now observe that the Hessian at p of the random function \I/ is the random bilinear
form

H313 EAQHQQ
a

A simple computation shows that

I B ; [ H § @ H § ] = Z H § ® Hg
a

and we deduce the following equality

IE Hg ® Hgt ZLUR. (4.2.26)

This is a special case of a much more general result of R. Adler and J. Taylor [2, Lemma
12.2.l].
 Riemannian Geometry 191

Example 4.2.31 (Reproducing kernels). We have seen that any immersion of a smooth
manifold into a Euclidean space produces a correlate on that manifold. It turns out that
all correlators satisfying a stronger positivity conditions are obtained in this fashion. This
follows using the concept of reproducing kernel Hilbert space that plays an important role
in varied areas of mathematics [110].
Consider a smooth submanifold M C RN . It is equipped with a natural correlate
(4.2.20)
K ¢ M x M - > R , K ( p , q ) (p, q >- (4.2.27)
For any positive integer n and any pa, . . . , p in 6 M define the symmetric n X n matrix

- KlP1)P1l KlP1)P2l KIP1 7 in) -

KlP2)P1l KlP2)P2l K(p2,pn)

CKIP1) 7 pn

_K(pn,pi) K(p,,,p2l K(p,,,pn _

Note that if K is defined by (4.2.27) then, for any n 6 N and any p l , . . . ,pn 6 M the
symmetric matrix C'K(pl, . . . , pn) is positive semidefinite.
A correlate K on a smooth manifold M is called a reproducing kernel if for any
pl, . . . , pn E M the symmetric matrix CK (pl, . . . , p n is a positive semidefinite. The
reason for this terminology stems from the following remarkable reproducing property.
For each q 6 M define
Kg 6 c°°<m>, sKq(p) = K(p, q >-
Denote by W C cOO (M) the linear span of the family
1 Kq;
q EM
Then the following hold (see [ l 10, Chap. 2]).

(i) If
u = UP Kg 6 W , 'u u Kg 71 Q W,
iiI 1l€I
then the quantity
(u, v)K 'U/7;UQ; K( q in q, )
o n

Mel
is independent of the decomposition of u and 'U as linear combination of functions Kq .
Moreover, it defines an inner product on W. Denote by 17-C the completion of W with
respect to this inner product.
(ii) For any p E M, the linear functional
e t p : W -> R, evp(w) = w(p),
is continuous with respect to the norm on W induced by the inner product ( 7 )K
and thus defines an element etp in the dual 17-C* of H.
192 Lectures on the Geometry of Manzfolds

(iii) The resulting map

M Y p »-> e t p GH*
is continuous. Moreover

K(p, q €'Up , e t q g.C*

Since K is smooth the above map is smooth. Moreover, the condition (ii) in Defini-
tion 4.2.26 of a correlate shows that this map is also an immersion.

4.2.6 The Gauss-Bonnet theorem for oriented surfaces

We conclude this chapter with one of the most beautiful results in geometry. Its meaning
reaches deep inside the structure of a manifold and can be viewed as the origin of many
fertile ideas.
Recall one of the questions we formulated at the beginning of our study: explain unam-
biguously why a sphere is "different" from a torus. This may sound like forcing our way
in through an open door since everybody can "see" they are different. Unfortunately this is
not a conclusive explanation since we can see only 3-dimensional things and possibly there
are many ways to deform a surface outside our tight 3D Universe.
The elements of Riemann geometry we discussed so far will allow us to produce an
invariant powerful enough to distinguish a sphere from a torus. But it will do more than
that.

Theorem 4.2.32 (Gauss-Bonnet Theorem. Preliminary version). Let S be a compact

oriented surface without boundary. If 90 and 91 are two Riemann metrics on S and 691 ( S )
( i = 0 , 1 ) are the corresponding Euler forms then

/S
690

Hence the quantity j S kg(sl is independent

(So
/of
S
691

the Riemann metric g SO that it really de-

ponds only on the topology of S!!!

The idea behind the proof is very natural. Denote by it the metric it 90+t(91 -90 )-
We will show
d
0 W e [0,1].
dt S ggi
It is convenient to consider a more general problem.

Definition 4.2.33. Let M be a compact oriented manifold. For any Riemann metric g on
E define

8m(M =9l s(9)dvg 7

M
where s(g) denotes the scalar curvature of (M, g). The functional g »-> €(g) is called the
Hilbert-Einstein functional.
 Riemannian Geometry 193

We have the following remarkable result.

Lemma 4.2.34. Let M be a compact oriented manifold without boundary and gt = (gii )
be a I -parameter family of Riemann metrics on M depending smoothly upon t 6 R. Then
d l t t -t
-<'3(9t) RiCgt §5(9 )9 ,Q dVgt , Vt.
dt M t

In the above formula (-, -)t denotes the inner product induced by gt on the space of sym-
metric (0,2)-ten5ors while the dot denotes the t-derivative.

Definition 4.2.35. A Riemann manifold (M, g) of dimension n is said to be Einstein if the

metric g satisfies Einstein '5 equation

Ricg = S ( g ,

where s(:z:) denotes the scalar curvature.

Example 4.2.36. Observe that if the Riemann metric g satisfies the condition

Ricg(x) = A(x)g(x) (4.2.28)

for some smooth function A Q C°°(M), then by taking the traces of both sides of the above
equality we deduce

/\(a:l 5(9)(;v) , n dim M .

..
II

n
Thus, the Riemann manifold is Einstein if and only if it satisfies (4.2.28).
Using the computations in Example 4.2.12 we deduce that a certain constant multiple
of the Killing metric on a compact semisimple Lie group is an Einstein metric.
We refer to [16] for an in depth study of the Einstein manifolds.

The (0, 2) -tensor

1
gin 2= Rid($) - §5(w)91,3(w)
is called the Einstein tensor of ( M , g).

Exercise 4.2.37. Consider a 3-dimensional Riemann manifold ( M ) - Show that

S
Rijké 5ik9i£ fiwgjk -I- €jegik 8jkg1 e + §l9@e931<: 9¢I<:9j£)~

In particular, this shows that on a Riemann 3-manifold the full Riemann tensor is com-
pletely determined by the Einstein tensor.

Exercise 4.2.38 (Schouten-Struik [117]). Prove that the scalar curvature of an Einstein
manifold of dimension > 3 is constant.
Hint: Use the 2nd Bianchi identity.
194 Lectures on the Geometry of Manzfolds

Notice that when (S, 9) is a compact oriented Riemann surface two very nice things
happen.
(i) (S, g) is Einstein. (Recall that only R1212 is nontrivial.)
(ii) €(gl = 2/$ et.
Theorem 4.2.32 is thus an immediate consequence of Lemma 4.2.34.
Proof of the lemma. We will produce a very explicit description of the integrand
d d t d
lSlgtldVgt )
llllllllll

dt ( dt )) dVgt -|-
(
Slgtl - d vof
dt ) (4.2.29)

of to (g t . We will adopt a "roll up your sleeves, and just do it" strategy reminiscent to the
good old days of the tensor calculus frenzy. By this we mean that we will work in a nicely
chosen collection of local coordinates, and we will keep track of the zillion indices we will
encounter. As we will see, the computations are not as hopeless as they may seem to be.
We will study the integrand (4.2.29) at t = 0. The general case is entirely analogous.
For typographical reasons we will be forced to introduce new notations. Thus, Q will denote
(gt) for t = 0, while gt will be denoted simply by g. A hat over a quantity means we think
of that quantity at t = 0, while a dot means differentiation with respect to t at t = 0.
Let q be an arbitrary point on S, and denote by (131, . . . , ;z:") a collection of §;-normal
coordinates at q. Denote by V the Levi-Civita connection of g and let F denote its
Christoffel symbols in the coordinates ( x ) .
Many nice things happen at q, and we list a few of them which will be used later.
in
§ij g viz» 3I<§1ij = 0. (4.2.30)

as 0, F 0. (4.2.31)

If a = adar is a 1-form then, at q,

(4.2.32)
z'

In particular, for any smooth function u we have

(4.2.33)
al

Set

h = (Hal :z (Q) = (Qu)-

The tensor h is a symmetric (0,2) -tensor. Its Q-trace is the scalar
trgh = § " h j = tra-1(h),
where L is the lowering the indices operator defined by Q. In particular, at q
tag h his • (4.2.34)

The curvature of g is given by

e
Rmy = -R eink = akron - @,1"¢, + Fe 1QFij - F" jai/q:~
 Riemannian Geometry 195

The Ricci tensor is

LU
k
iikj ark - a r k + pa Kr - Pa' 2rggg.
Finally, the scalar curvature is
S = trg R = gjRj (@`k1"Z ajrkk, + pk kegg - pk

.Jrggl
9"
Differentiating S at t = 0, and then evaluating at q we obtain

.é = Q" la in - I
a,1*/<=i k + 6" la in - 3]Fik
.°1<= I
= WR + la ii '
I.
32.°1<=
Pik
(4.2.35)

'k _ _
The term QU can be computed by derivating the equality gz, gawk - (So at = 0. We get

9""§,k + Qwhjk = 0,
SO that

Q" -he. (4.2.36)

To evaluate the derivatives 1'8 we use the known formulae

m 1 km
l3¢9jk - 61<=9¢j + 31%) 7

which, upon differentiation at t = 0, yield

1 +l
• m
= 5 (3i9jI< - 3A:§1ij + fijfltzkl
A

29 k m l f h j k - <9khii + c9jhk)
(4.2.37)
l _
(Wim - gm./ij + é i h m )
2 '

We substitute (4.2.36)-(4.2.37) in (4.2.35), and we get, at q

1 1
.é hijéij + 5 Z
( a a h k - aW + 31¢3¢/Mkl 2
(afhkk - as h e )
"BJ i,k ¢,1<¢

+ 3¢6khik
II
I

hijéij 6¢2hkk
"BJ ¢,1<¢ i,k

= -(Ric, -|- Am,§tI`g -|- 6¢6khik . (4.2.38)

i,k

To get a coordinate free description of the last term note that, at q,

(Vhf)<a, am) = mm.

The total covariant derivative Vh is a (0, 3)-tensor. Using the Q-trace we can construct a
(0, l)-tensor, i.e., a 1-form
ttgwh) = tr(L§ h),
where LQ 1 is the raising the indices operator defined by Q. In the local coordinates w) we
have
tf§l/z) QU ( vihljkdxk
A
196 Lectures on the Geometry of Manzfolds

Using (4.2.30), and (4.2.32) we deduce that the last term in (4.2.38) can be rewritten (at q)
as
(Sure ( h ) = 5t1l§(W).

We have thus established that

.é = -(RiC, -|- AA/[§tI`§ -|- 5t1`§(W1l. (4.2.39)

The second term of the integrand (4.2.29) is a lot easier to compute.

asg iv l9ld5 U1/\ ~-~/\ d;z:" 7

SO that
. l d 1
dV i§l9l -1/2 -l9ldw A~--Adm".
A

9 ¢it
At q the metric is Euclidean, §ij = 5¢j, and
d
-191
dt = EQ = l§l - tll§(Q) = (§,Q)§ |§|-

Hence
1 A A • . avg + (A tag g + 5t1l§(W1) ) avg .
é(9) §s(glg-R1cg ,g
M g M

Green's formula shows the last two terms vanish and the proof of the Lemma is concluded.

Definition 4.2.39. Let S be a compact, oriented surface without boundary. We define its
Euler characteristic as the number
l
x(S) 27r @(9), lllllllllll

where g is an arbitrary Riemann metric on S. The number

1
_
9(5) = 2 ( 2 _ XlS))
is called the genus of the surface.

Remark 4.2.40. According to the theorem we have just proved, the Euler characteristic is
independent of the metric used to define it. Hence, the Euler characteristic is a topological
invariant of the surface. The reason for this terminology will become apparent when we
discuss DeRham cohomology, a Z-graded vector space naturally associated to a surface
whose Euler characteristic coincides with the number defined above. So far we have no
idea whether X(S) is even an integer.

Proposition 4.2.41.
x(S2l = 2 and X(T2) = 0.
 Riemannian Geometry 197

Proof. To compute x(S2l we use the round metric 90 for which K = l so that
l l
x(S2l 27r S2 do90 area90 (S2 ) 2.

To compute the Euler characteristic of the torus we think of it as an Abelian Lie group with
a bi-invariant metric. Since the Lie bracket is trivial we deduce from the computations in
Subsection 4.2.2 that its curvature is zero. his concludes the proof of the proposition.

Proposition 4.2.42. If $¢ (i = I, 2) are M o compact oriented surfaces without boundary

then

X(51#$2l = X($1) -|- X($2) - 2.

Thus upon iteration we get
k

XlSl# #go) xIS - 2(k - II)

i t

for any compact oriented sultaces 51, . ..,Mk. In terms of genera, the last equality can be
rephrased as
k

QIS1# . . #Mk)
• 9lSil-
'in l

In the proof of this proposition we will use special metrics on connected sums of sur-
faces which require a preliminary analytical discussion.
Consider f : (-4, 4) -> (0, oo) a smooth, even function such that

(i) f(a:) = 1 for la:l < 2. _

(ii) f(22) = \ 1 (at + 3)2 for at E [-4, -3.5].
(iii) f(22l = \ 1 (at - 312 for at 6 [3.5, 41.
(iv) f is non-decreasing on [-4, 0].

One such function is graphed in Figure 4.3

Denote by Sf the surface inside R3 obtained by rotating the graph of f about the :zz-axis.
Because of properties (i)-(iv), Sf is a smooth surface diffeomorphic5 to S2. We denote by
g the metric on Sf induced by the Euclidean metric in R3. Since Sf is diffeomorphic to a
sphere

K9dvg _ QWX ( S 4 •
Sf _ ) Z
'ii
Set

Sf:|: e

.- SfM{:|:£U>0}, Sf:al e

. - SfM{:|:$ > 1}.

5 One such diffeomorphism can be explicitly constructed projecting along radii starting at the origin.
198 Lectures on the Geometry of Manzfolds

/l

\
l >

II

S
f
/
/ /
/
/ /

l I /
/
/
I I
I
I
+ :>

l l I
l
\ \
\ \
\
\
\
/
\ \
\ \
n

S- +
S
f f

Fig. 4.3 Generating a hot-dog-shaped surface.

Since f is even we deduce

1
Kgdvg Kgdvg = 2¢r. (4.2.40)
:|: 2. ff
Sf

On the other hand, on the neck C = {l22l < 2} the metric g is locally Euclidean g
d:z32 -|- d92, so that over this region Kg 0. Hence

Kgdfug 0. (4.2.41)
C'

Proof of the Proposition 4.2.42. Let Di C Si (z' = l , 2) be a local coordinate neighbor-

hood diffeomorphic with a disk in the plane. Pick a metric ii on $i such that (D1, go) is
isometric with 5? and (D27 92) is isometric to Sf . The connected sum S1#S2 is obtained
by chopping off the regions s;
from D1 and Sf 1 from D2 and (isometrically) identifying
the remaining cylinders 5? n {l;13l < l} = C and call O the overlap created by gluing (see
 Riemannian Geometry 199

S
1
D 1'

l
$2

l
or /
2

( l
0
l

Fig. 4.4 Special metric on a connected sum.

Figure 4.4). Denote the metric thus obtained on S1#S2 by Q. We can now compute

X($1#$2l : 'or
1 /S1#S2
Kggdvg

1 f 1 1
K91 dV91 + 271' S2\D2 K 92 dV92 + Kg dvg
- Qtr S1\D1
I
Qtr

(4.2.41) l 1
llllll--I

K91 dvgl 27r D. Kgdvg

lllllllllllll

2tr J1

+ 1 1 (4.2.40)
K92 CWQ2 27r D2 Kgdvg - xIS1) -|- XISQ) - 2.
lllllllll.ll.l

27r S2
This completes the proof of the proposition.

Corollary 4.2.43 (Gauss-Bonnet). Let 29 denote the connected sum of g-tori. (By defini-
son 20 = s2.) Then
6

Mig) = 2 - 29 and 9(29)

II

In particular; a sphere is not dijeomorphic to a torus.

Remark 4.2.44. It is a classical result that the only compact oriented surfaces are the con-
nected sums of g-tori (see [90]), so that the genus of a compact oriented surface is a com-
plete topological invariant.
This page intentionally left blank
 Chapter 5

Elements of the Calculus of Variations

This is a very exciting subject lieing at the frontier between mathematics and physics. The
limited space we will devote to this subject will hardly do it justice, and we will barely
touch its physical significance. We recommend to anyone looking for an intellectual feast
the Chapter 16 in vol. 2 of "The Feynmann Lectures on Physics" [47], which in our opinion
is the most eloquent argument for the raison d'etre of the calculus of variations.

5.1 The least action principle

5.1.1 The 1-dimensional Euler-Lagrange equations

From a very "dry" point of view, the fundamental problem of the calculus of variations can
be easily formulated as follows.
Consider a smooth manifold M, and let L : R X TM -> R by a smooth function
called the lagrangian. Fix two points Po, PI E M . The action of a piecewise smooth path
'Y : [0, 1] -> M connecting these points is the real number S(*y) = $L (7) defined by
1
SLl')'l
.

SW) L(¢»v'(tl»v(t))dt-
0

In the calculus of variations one is interested in those paths as above with minimal action.

Example 5.1.1. Given a smooth function U : R 3 -> R called the potential, we can form
the lagrangian

L(q°, Q) : R3 x]R3%T]R3->]R,

given by

L= Q- U
.
= klnetlc•
energy
.
- potential energy =
1
5mlql•2
U(q)-

The scalar m is called the mass. The action of a path (traj ectory) *y : [0, l] -> R3 is a
quantity called the Newtonian action. Note that, as a physical quantity, the Newtonian
action is measured in the same units as the energy.

201
202 Lectures on the Geometry of Manzfolds

Example 5.1.2. To any Riemann manifold ( M , g) one can naturally associate two la-
grangians Ll, L2 : TM -> R defined by
L1('u,q) = go(u,ul1/2 (u 6 TqM),

and
l
L2 (u, Q) 59q(07'Ul~

We see that the action defined by L1 coincides with the length of a path. The action defined
by L2 is called the energy of a path.

Before we present the main result of this subsection we need to introduce a bit of notation.
Tangent bundles are very peculiar manifolds. Any collection (q1, . . . , q"l of local co-
ordinates on a smooth manifold M automatically induces local coordinates on TM. Any
point in TM can be described by a pair (u, q), where q 6 M , 'U 6 TqM. Furthermore, 'U
has a decomposition

'U 11Z3i, where 3%

a
3q
'Ui
We set q'fi so that

The collection (q°1 . . . , q", q1, . . . , q") defines local coordinates on TM. These are said to
be holonomic local coordinates on TM. This will be the only type of local coordinates we
will ever use.

Theorem 5.1.3 (The least action principle). Let L : R X T M -> R be a lagrangian, and
pg, Pl 6 M zwo fixed points. Suppose 'Y [0, l] -> M is a smooth path such that the
following hold.

(i) W ) = Pi, 'i = 0, 1.

( i i ) SLY) < SL (if), for any smooth path if : [0, 1] -> rejoining Po to P1-

Then the path by satisfies the Euler-Lagrange equations

d a Lu, , a
dt 37 'YI _L
<97 (t
7
Wm)-
More precisely, ,fly , q'Z) are holonornic local coordinates on T M such that 'y(t) = (q (to),
and by = (as (i)), then by is a solution of the system of nonlinear ordinary differential
equators
d aL aL
dim M .
dt 9'k ( t , dj, q') (t,<ij,ql» k = 1, ,n
3qk¢

Definition 5.1.4. A path 'Y : [0,l] -> M satisfying the Euler-Lagrange equations with
respect to some lagrangian L is said to be an extremal o f L.
 Elements of the Calculus of Variations 203

To get a better feeling of these equations consider the special case discussed in Exam-
ple 5.1.1
l • 2
L 2 m1q1 U(q).

Then
a
_L
aL
Z
ma, lllllllllll

-VU(q),
<8lq <9q
and the Euler-Lagrange equations become

We VU(q). (5.l.l)

These are precisely Newton's equation of the motion of a particle of mass m in the force
field -VU generated by the potential U.
In the proof of the least action principle we will use the notion of variation of path.

Definition 5.1.5. Let 'Y [0, l] -> M be a smooth path. A variation of by is a smooth map
a = aS(t) : (-6,6) X [0,l] -> M,
such that a0(t) = 'y(t). If moreover, as(i) = Pi Vs, i 0, 1, then we say that a is a
variation rel endpoints.

Proof of Theorem 5.1.3. Let 048 be a variation of by rel endpoints. Then

SL(O¢ol S SL(O¢5) Vs,

SO that
d
llllllllll

ds
SLIOJsI = 0.
s=0

Assume for simplicity that the image of Fy is entirely contained in some open coordinate
neighborhood U with coordinates lq1, . . . q'nl. Then, for very small lsl, we can write
7

do.S
Odsltl i (q 5,¢5) and (q"(s,¢)).
dt
Following the tradition, we set

(So
pa fig
<% (So :=
a | do8 éqf a
..
II

38 8=0 35 35 s=0 dt 35 3q'j

The quantity (in is a vector field along *y called infinitesimal variation (see Figure 5. 1 ). In
fact, the pair (pa; pa) 6 T ( T M ) is a vector field along t »-> ('y(t), I(t)) 6 TM. Note that
(SOn = co, and at endpoints (So = 0.
Exercise 5.1.6. Prove that if t »-> X (t) 6 T ' y ( t ) M is a smooth vector field along by, such
that X (15) = 0 for t 0, l then there exists at least one variation rel endpoints a such that
(So = X.
Hint: Use the exponential map of some Riemann metric on M .
204 Lectures on the Geometry of Manzfolds

Y
/
pa

Fig. 5.1 Deforming a path rel endpoints.

We compute (at s = 0)
1 -1
d d aL l aL
0 = -SL((I8) = -
ds ds 0
L(t, 0457
• 045) =
3q'l
.da'Ldt + ..(sadt.
0 3119
u

(5.l.2)

Integrating by parts in the second term in the right-hand-side we deduce

1 aL. - -d aL
0 /{
fig'l dt off ( l}
(SQidt_ (5. 1.3)

The last equality holds for any variation a. From Exercise 5.1.6 we deduce that it holds
for any vector field 6afidi along *y. At this point we use the following classical result of
analysis.
If f (to is a continuous function on [ 0 , ] ] such that
1
f (t)glt)dt = 0 Vg 6 C0 (0,1),
0
then f is identically zero.
Using this result in (5.1.3) we deduce the desired conclusion.

Remark 5.1.7. (a) In the proof of the least action principle we used a simplifying assump-
tion, namely that the image of by lies in a coordinate neighborhood. This is true locally,
and for the above arguments to work it suffices to choose only a special type of variations,
localized on small intervals of [0, 11
In terms of infinitesimal variations this means we
need to look only at vector fields along by supported in local coordinate neighborhoods. We
leave the reader fill in the details.
(b) The Euler-Lagrange equations were described using holonomic local coordinates.
The minimizers of the action, if any, are objects independent of any choice of local coor-
dinates, so that the Euler-Lagrange equations have to be independent of such choices. We
check this directly.
 Elements of the Calculus of Variations 205

If (of) is another collection of local coordinates on M and ( i j 7 Ii) are the coordinates
induced on TM, then we have the transition rules
azi n 31-k
=u'(q1 7 ,q"),
gawk q 7

SO that
a 3337 9 _I_ 32567 ~k 3
3q go'Z 3367 3q'*3q'q éij
a a a 33 3
3qj <9qj a 3q iii •

Then
aL _ Ami aL + 3233.7 ~k aL
. <1
3q'il - 3q'll 3331 é1q":3q'L 3.5537

d d 3561 aL aL _|_ 331J d

3251:i q.k
at ('")
W at ( §q" <9;i:i aqkaq
3937 3q dt 3;ti
We now see that the Euler-Lagrange equations in the q-variables imply the Euler-Lagrange
('"l
in the Hz:-variable, i.e., these equations are independent of eoordinates.
The esthetically conscious reader may object to the way we chose to present the Euler-
Lagrange equations. These are intrinsic equations we formulated in a coordinate dependent
fashion. Is there any way of writing these equation so that the intrinsic nature is visible "on
the nose"'?
If the lagrangian L satisfies certain nondegeneracy conditions there are two ways of
achieving this goal. One method is to consider a natural nonlinear connection VL on TM
as in [100] . The Euler-Lagrange equations for an extremal fy(t) can then be rewritten as a
"geodesics equation"
V 7.
L .
The example below will illustrate this approach on a very special case when L is the la-
grangian L2 defined in Example 5. 1.2, in which the extremals will turn out to be precisely
the geodesics on a Riemann manifold.
Another far reaching method of globalizing the formulation of the Euler-Lagrange
equation is through the Legendre transform, which again requires a nondegeneracy con-
dition on the lagrangian. Via the Legendre transform the Euler-Lagrange equations be-
come a system o f first order equations on the cotangent bundle T*M known as Hamilton
equations.
These equations have the advantage that can be formulated on manifolds more general
than the cotangent bundles, namely on symplectic manifolds. These are manifolds carrying
a closed 2-form whose restriction to each tangent space defines a symplectic duality (see
Subsection 2.2.4.)
Much like the geodesics equations on a Riemann manifold, the Hamilton equations
carry a lot of information about the structure of symplectic manifolds, and they are cur-
rently the focus of very intense research. For more details and examples we refer to the
monographs [7, 38].
206 Lectures on the Geometry of Manzfolds

Example 5.1.8. Let (M, g) be a Riemann manifold. We will compute the Euler-Lagrange
equations for the lagrangians Ll, L2 in Example 5.1.2.
. l ~i .-
L2(q, Q) = 5912(q)q q.7,
SO that
aL2 3L2 l agij

aiff gjkq.7
3q/* 2 3qk*
qqj
The Euler-Lagrange equations are
99 . l agij
qjgjk + 3%
q q°j
2 3qk*
qqj (5.l.4)

Since g k TTLgarn
'rn
(Si, we get

(5.l.5)

When we derivate with respect to t the equality g k q gjkqj we deduce

39%
gum g g j k
3q'
qq 1
29
km /
+ 39
3q'l
kW
n 3117
We substitute this equality in (5.l.5), and we get

(5.l.6)

m
The coefficient of q°¢q'J in (5.1.6) is none other than the Christoffel symbol SO this
equation is precisely the geodesic equation.

/2
Example 5.1.9. Consider now the lagrangian L1 (re, Q) = (giiq°'iq°j)1 Note that the action
P1
3L(q(15)) MQ, qldt,
Po
is independent of the parametrization t »-> q(t) since it computes the length of the path
t »-> q(t). Thus, when we express the Euler-Lagrange equations for a minimizer 70 of
this action, we may as well assume it is parameterized by arclength, i.e., to I = 1. The
Euler-Lagrange equations for L1 are

d
39i' - '
3113 q'q.7
-.
g k j q'j

gfzj q 'in 2V5lij q iv

Along the extremal we have 9ziqq.7 l (arclength parametrization) so that the previous
equations can be rewritten as
d 1 agij
lg1<1'in ) 2 3qk¢
qqj
We recognize here the equation (5. 1.4) which, as we have seen, is the geodesic equation in
disguise. This fact almost explains why the geodesics are the shortest paths between two
nearby points.
 Elements of the Calculus of Variations 207

5.1.2 Noether's conservation principle

his subsection is intended to offer the reader a glimpse at a fascinating subject touching
both physics and geometry. We first need to introduce a bit of traditional terminology
commonly used by physicists.
Consider a smooth manifold M. The tangent bundle TM is usually referred to as the
space of states or the lagrangian phase space. A point in TM is said to be a state. A
lagrangian L : R X TM -> R associates to each state several meaningful quantities.

.. The generalized momenta: Pi aL

or
l l l l l l l l l - - _

. The energy: H = pin - L.

aL
The generalizedforee: F = Gq'L l l l l l l l l l - - _

This terminology can be justified by looking at the lagrangian of a classical particle in

a potential force field, F = -VU,
1
L = -m
2 Ill 2
' U(q).

The momenta associated to this lagrangian are the usual kinetic momenta of the Newtonian
mechanics

PI my 7

while H is simply the total energy

1 •
H hmM+UM%
It will be convenient to think of an extremal for an arbitrary lagrangian L(t, ii, q) as de-
scribing the motion of a particle under the influence of the generalized force.

Proposition 5.1.10 (Conservation of energy). Let 'y(tl be an extremal of time indepen-

dent lagrangian L = L(q, ql. Then the energy is conserved along 7, i.e.,
d
WI
o'

-HW,
II

dt
Proof. By direct computation we get
d d of; <9L ~¢ of;
dt dt l p f i £ avi (1 <9q q avi (1
d of; AL
{ ( go )
llllllllll

dt 3% q } 0 (by uler- agrange).

At the beginning of the 20th century (1918), Emmy Noether discovered that many of the
conservation laws of the classical mechanics had a geometric origin: they were, most of
them, reflecting a symmetry of the lagrangian! !!
This became a driving principle in the search for conservation laws, and in fact, con-
servation became almost synonymous with symmetry. It eased the leap from classical to
quantum mechanics, and one can say it is a very important building block of quantum
208 Lectures on the Geometry of Manzfolds

physics in general. In the few instances of conservation laws where the symmetry was not
apparent the conservation was always "blamed" on a "hidden symmetry". What is then this
Noether principle 7
To answer this question we need to make some simple observations.
Let X be a vector field on a smooth manifold M defining a global flow <I>S. This flow
induces a flow \1/S on the tangent bundle TM defined by
\I'S(u,;u) = l<I>S*(vl, <I>S(a3)

One can think of \1/S as defining an action of the additive group R on TM. Alternatively,
physicists say that X is an in]initesimal symmetry of the given mechanical system described
by the lagrangian L.

Example 5.1.11. Let M be the unit round sphere S2 C R3. The rotations about the z-axis
a
define a 1-parameter group of isometrics of S2 generated by -
be, where H is the longitude
on S2 .

Definition 5.1.12. Let L be a lagrangian on TM, and X a vector field on M. The la-
grangian L is said to be X-invariant if
Lo\I/S=L, Vs.

Denote by DC Q Vect ( T M ) the infinitesimal generator of \PS, and by Lx the Lie deriva-
tive on TM along X. We see that L is X-invariant if and only if
Lx1; 0.
We describe this derivative using the local coordinates (quo, q'i). Set

(do (s), <11(8) ) == \If sw, q)-

Then

To compute d
|S 8
d
lllllllllll

ds is 0
q ( s ) = Xk5k.

ds 0 61(5) $q7 we use the definition of the Lie derivative on M

-kaxe a
d .~ 8
dsq 3qj - Lx
since <96j/W
la
0 on TM. Hence
8
3qi ) (
X i éq
3qk q $q
8 _.3X3
q 3% (9q7
7

. a. 3xi a
...
fx = x'
3q'L + <1~k k
3q 3q3
Corollary 5.1.13. The lagrangian L is X-invariant if and only If
Xi AL. + q 8xi aL
~k
0. (5.1.7)
3q'L 3q/* 3qj
 Elements of the Calculus of Variations 209

Theorem 5.1.14 (E. Noether). If the lagrangian L is X -invariant, then the quantity
X i aL Z

Pa XPfz
64
is conserved along the extremal of L.

Proof. Consider an extremal v = v(q'(t)) of L. We compute

d d f C EXi ~k aL
( \ T \
. d
dt dt 1Xz('V(t)) 6 3q'*q 3q°i
+ X'
dt
llllllllll

(W)..
3q"
Euler-Lagrange EX z
q.k AL 3L(5.1.7)
+X 0.
- 3q/* 3 3q'll -
The classical conservation-of-momentum law is a special consequence of Noether' S theo-
rem.
Corollary 5.1.15. Considers lagrangian L = L(t, q, q, I on R". If = 0, i.e., the i-tn
component of the force is zero, then = 0 along any extremal, i.e., the i-th component
of the momentum i5 conserved.

Proof. Take X = - in Noether's conservation law.

The conservation of momentum has an interesting application in the study of geodesics.

Example 5.1.16 (Geodesics on surfaces of revolution). Consider a surface of revolution

S in R3 obtained by rotating about the z-axis the curve y = f (z) situated in the yz plane.
If we use cylindrical coordinates (T, H, z) we can describe S as T' = f(z).
II Z

Y
I- >

/
/'
/
X

Fig. 5.2 A surface o f revolution.

In these coordinates, the Euclidean metric in R3 has the form

ds = d7°2 + dz2 + M62
210 Lectures on the Geometry of Manzfolds

We can choose Z: 9) as local coordinates on S, then the induced metric has the form
is = {1 + (f'(Z))2}dZ2 + f2(z)d92 A(z)dz2 -|- 12402, T = f(z>.
The lagrangian defining the geodesics on S is
L 1
5 [A
Z
-|- 7292

We see that L is independent of 9: Q39 = 0, so that the generalized momentum

aL
7,29
69
is conserved along the geodesics.

This fact can be given a nice geometric interpretation. Consider a geodesic 'y(t)
(z(t), 0(t)) on our surface of revolution S. The angle <l> between the geodesic and the
parallels Z = const can be determined from the equality

(36) v29
COS QUO
we . l<991 to 7

i.e., rcoscl) = 1.29l'YI-1. The conservation of energy implies that we 2

2L His
constant along the geodesics. We deduce the following classical result.

Theorem 5.1.17 (Clairaut). On a a surface of revolution T' = f (z) the quantity T' COS ¢ is
constant along any geodesic, where QUO E ( - 7 r , Tr) is the oriented angle the geodesic makes
with the parallels Z = const.

Exercise S.1.18. Describe the geodesics on the round sphere S2 and on the cylinder {a:2 +
9

3/2
1} C R3.

5.2 The variational theory of geodesics

We have seen that the paths of minimal length between two points on a Riemann manifold
are necessarily geodesics.
However, given a geodesic joining two points quo, C11 it may happen that it is not a min-
imal path. This should be compared with the situation in calculus, when a critical point of
a function f may not be a minimum or a maximum.
To decide this issue one has to look at the second derivative. This is precisely what we
intend to do in the case of geodesics. This situation is a bit more complicated since the
action functional
1
S l'7l2dt
5
is not defined on a finite dimensional manifold. It is a function defined on the "space of all
paths" joining the two given points. With some extra effort this space can be organized as
an infinite dimensional manifold. We will not attempt to formalize these prescriptions, but
rather follow the ad-hoc, intuitive approach of [94].
 Elements of the Calculus of Variations 211

5.2.1 Variational formulae

Let M be a connected Riemann manifold, and consider p, q 6 M. Denote by Q p q =

Qpq(M) the space of all continuous, piecewise smooth paths *y : [0, 1] -> M connecting p
to q.
An infinitesimal variation of a path by E Q p q is a continuous, piecewise smooth vector
field V along by such that V(0) = 0 and V(1) = 0 and
lim vlf i h)
h\,0

exists for every t E [0, 1], they are vectors V(t)i E T"(t)M, and v(t)+ = v(t)- for all
but finitely many t-s. The space of infinitesimal variations of 'y is an infinite dimensional
linear space denoted by T'Y - T'YQp,q~

Definition 5.2.1. Let by E Qp,q° A variation of by is a continuous map

Q = aS(t) : (-6,6) X [0, 1] -> M
such that

Vs E (-5, 6), as 6 Qp,q.

(ii) There exists a partition 0 = to < ii < tk_1 < to - l of [0,1] such that the
restriction of a to each (-e, 8) X (ti_1, t ) is a smooth map.

Every variation a of by defines an infinitesimal variation

3
pa
..
II

35 S 0

Exercise 5.2.2. Given V E T'7 construct a variation a such that pa V.

Consider now the energy functional

1
1
E : Qpq -> R, E('y) 2 0 Iw( ) | 2it.
Fix 7 6 Qp# and let a be a variation of by. The velocity W) has a finite number of
discontinuities, so that the quantity
A ° = . +h
lim (W
W h_>0+ ) 'W to)
is nonzero only for finitely many t's.

Theorem 5.2.3 (The first variation formula).

1
d
E*(5a) lllllllllll

EMs) <(60¢)(t)% Aw) (co,V_*}/)dt,

d (5.2.1)
ds s O t 0
dt

where V denotes the Levi-Civita connection. (Note that the right-hand-side depends on a
only through (So SO it is really a linear function on T . )
212 Lectures on the Geometry of Manu'olds

30
Proof. Set $1S 6't
S
We differentiate under the integral sign using the equality

a <5¢SI
2
2<v
33 38

and we get
~1
d
E(0¢5) (v a 619, dS) l s = 0 dt.
ds J
38

a and
Since the vector fields -
38
%
commute we have V a - 30
as it
= V a pa
f i t 35
L€[0=t0<t2<° -
< to = 1 be a partition of [0,1] as in Definition 5.2.1. Since
as by for S = 0 we conclude

k to

We use the equality

a
315 <5a,'> IV 57 6a/9)-I-(5a, V g
8 )

to integrate by parts, and we obtain

k to
E* (5o~) (co, V ' § / ) d t .

This is precisely equality (5.2.1).

Definition 5.2.4. A path 'Y 6 Qzmq is called critical if

E*(V) : 0, W 6 To'

Corollary 5.2.5. A path 'Y 6 Qpq is critical zf and only If it is a geodesic.

Exercise 5.2.6. Prove the above corollary.

Remark 5.2.7. Note that, a priori, a critical path may have a discontinuous first derivative.
The above corollary shows that this is not the case: the criticality also implies smoothness.
This is a manifestation of a more general analytical phenomenon called elliptic regularity.
We will have more to say about it in Chapter I I .

The map E* : T'y -> R, (So »-> E* (do) is called the first derivative of E at by E Q p q . We
want to define a second derivative of E in order to address the issue raised at the beginning
of this section. We will imitate the finite dimensional case which we now briefly analyze.
Let f : X -> IRE. be a smooth function on the finite dimensional smooth manifold X . If
kg is a critical point of f , i.e., df (xo) = 0, then we can define the Hessian at kg

f** : T12,0 x X TS2,0 X -> IRE.

 Elements of the Calculus of Variations 213

as follows. Given VI, VI Q T 0x, consider a smooth map ls1,s2l P-> O£lS1,S2) E X such
that
pa
a(0, 0)
5130
and
387;
(0, 0) z 1,2. (5.2.2)

Now set
92f( 0¢(31» S2) )
f**lV1, VI) 3.91382
(0,0)

Note that since xo is a critical point of f , the Hessian f** (V1> VI) is independent of the
function Q satisfying (5.2.2).
We now return to our energy functional E : Q p q -> R Let 'Y Q Qzmq be a critical path.
Consider a 2-parameter variation of 'Y
a := 0581,82 : (-e,e) X (-e,e) X [0, 1] -> M, (s1,s2,t) »-> a81,52ltl°

Set U := (-e,e) X (-s,e) C R2 and 'y := 0Y0,0. The map Q is continuous, and has
second order derivatives everywhere except maybe on finitely many "coordinate" planes
30
Si = const, or t = const. Set 511a := l<0,0)»i = 1, 2. Note that (ii 0£ 6 To.
3811

Exercise 5.2.8. Given VI, V2 6 T'Y construct a 2-parameter variation a such that VE (5i04.

We can now define the Hessian o f E at 'Y by

E**l(510£, ($201) 32E(@s1,s2 )

3.91382
(0,0)

Theorem 5.2.9 (The second variation formula).

E** l(S1CY, (5201) ((52CY, At(S1C¥)

1
(S2CY7 V2-(S1CY
a R ( 5 1 @ ) ) it, (5.23)
t
0 at
where R denotes the Riemann curvature. In particular, E** is a bilinear functional on Tv-

Proof. According to the first variation formula we have

BE
682

Hence
62E
881382
t
1
t ( (5204, a
351
( ))
(V 3 (5201, V _'Y)dt
a dt. (5.2.4)
381 $t
0

Since 'Y is a geodesic, we have

Am I 0 and va it
= 0.
214 Lectures on the Geometry of Manzfolds

Using the commutativity of -a with 6'

we deduce
fit 681

pa pa
v 351
a
( 316
Finally, the definition of the curvature implies
) At
(v 351
a
316 ) At
(v Q (5101
it
.

v 3
381
V Q
it
i v Q
it
V a
381
+ Rl61Q,F/l_
Putting all the above together we deduce immediately the equality (5.2.3).

Corollary 5.2.10.
E**(V1,V2) I E**lV2,V1)» VV1,V2 E T7-

5.2.2 Jacobi fields

In this subsection we will put to work the elements of calculus of variations presented so
far. Let (M, g) be a Riemann manifold and p, q E M.
Definition 5.2.11. Let 'y E Qw; be a geodesic. A geodesic variation of by is a smooth map
aS(t) : (-5,8) X [0, 1] -> M such that, to = by, and t »-> as(t) is a geodesic for all 8. We
set as usual pa = Z?l1.s=0.
llllllllll

Proposition 5.2.12. Let 'Y E Qw] be a geodesic and (045) a geodesic variation offs. Then
the infinitesimal variation 504 satisfies the Jacobi equation

vim = no, 6oi)'7° (it

Proof.
V353 Vt (v I v (pa
316

VG l ) RW, cal
pa
315 RW, cal 315
30

Definition S.2.13. A smooth vector field J along a geodesic 'Y is called a Jacobi field if it
satisfies the Jacobi equation
v31 = Run Jo
Exercise S.2.14. Show that if J is a Jacobi field along a geodesic 'Y» then there exists a
geodesic variation as of by such that J = do.

Exercise 5.2.15. Let 'Y Q Q12,q» and J a vector field along fy.
(a) Prove that J is a Jacobi field if and only if
E**(J, v ) = 0, W Q T,.
(b) Show that a vector field J along *y which vanishes at endpoints, is a Jacobi field if any
only if E** (J, W) = 0, for all vector fields W along fy.
 Elements of the Calculus of Variations 215

Exercise 5.2.16. Let *y E Q p q be a geodesic. Define gp to be the space of Jacobi fields V

along 'Y such that V ( p ) = 0. Show that dim gp = dim M, and moreover, the evaluation
map
evq : 3/, -> TpM V »-> V t v ( p )
is a linear isomorphism.

Fig. 5.3 The poles are conjugate along meridians.

Definition S.2.17. Let 'y(t) be a geodesic. Two points 'y(t1) and 'Yltzl on *y are said to be
conjugate along 'Y if there exists a nontrivial Jacobi field J along by such that ](t1;l = 0,
i = 1, 2.

Example 5.2.18. Consider by : [0,2¢r] -> S2 a meridian on the round sphere connection
the poles. One can verify easily (using Clairaut's theorem) that by is a geodesic. The
counterclockwise rotation by an angle H about the z-axis will produce a new meridian,
hence a new geodesic 79 • Thus (79) is a geodesic variation of 'Y with fixed endpoints. 57
is a Jacobi field vanishing at the poles. We conclude that the poles are conjugate along any
meridian (see Figure 5.3).

Definition 5.2.19. A geodesic 'Y 6 S2pqis said to be nondegenerate if q is not conjugated

to p along fy.

The following result (partially) explains the geometric significance of conjugate points.
Theorem 5.2.20. Let by 6 Q p q be a nondegenerate, minimal geodesic. Then p is conjugate
with no point on *y other than i t e m In particular, a geodesic segment containing conjugate
points cannot be minimal !
216 Lectures on the Geometry of Manzfolds

Proof. We argue by contradiction. Let PI 'Yltll be a point on 'Y conjugate with p.

Denote by Jt a Jacobi field along by l[0,t1] such that Jo - 0 and Jt1 0. Define V Q T , b Y

Jf, t€[0,t1]
Vt
0, t > to.

We will prove that W is a Jacobi field along by which contradicts the nondegeneracy of by.
Step 1.
E**(U, U) 2 0 VU Q To. (5.2.5)

Indeed, let 048 denote a variation of by such that do = U. One computes easily that
d2
d82 E(0¢S2l 2E**(U, U)~

Since 'y is minimal for any small S we have length (QS2) > length (Ag) so that
2
l 1 l
E(0152) > 2
(/ 0 16482 lot
) = length (0452 > length (0¢0)2
1
= -length (by) 2 E(0¢0)-

Hence
d2
EIOA8QI > 0.
ds2 S 0
This proves (5.2.5).
Step 2. E**(v, VI = 0. This follows immediately from the second variation formula and
the fact that the nontrivial portion of V is a Jacobi field.
Step 3.
E**(U, V) 0, w6i/1,.
From (5.2.5) and Step 2 we deduce
0 i E**(V,V) < E**(V+ 7'U,V-|- TU) = fU(7') VT.
Thus, 7' 0 is a global minimum of fU(7'l so that

f{,(0) 0.
Step 3 follows from the above equality using the bilinearity and the symmetry of E**. The
final conclusion (that V is a Jacobi field) follows from xercise 5.2.15.

Exercise 5.2.21. Let 'Y : R -> M be a geodesic. Prove that the set
{ t E IR ; 'y(t) is conjugate to 7(0) }
is discrete.
 Elements of the Calculus of Variations 217

Definition 5.2.22. Let 'Y Q Qzmq be a geodesic. We define its index, denoted by ind (by), as
the cardinality of the set

Co = {t E (0, 1) ; is conjugate to 7(0) }

which by Exercise 5.2.21 is finite.

Theorem 5.2.20 can be reformulated as follows: the index of a nondegenerate minimal

geodesic is zero.
The index of a geodesic obviously depends on the curvature of the manifold. Often,
this dependence is very powerful.

Theorem 5.2.23. Let M be a Riemann manifold with non-positive sectional curvature, i.e.,

(R(X,Y)Y,X) < 0 VX,Y 6 T,,,M Van' 6 M . (5.2.6)

Then for any p, q 6 M and any geodesic by 6 Qp,q» ind (7) = 0.

Proof. It suffices to show that for any geodesic 'y : [0,1] -> M the point 7(1) is not
conjugated to 'l/(0).
Let Jt be a Jacobi field along by vanishing at the endpoints. Thus

WJ Z RW, Jo,
so that
1 1 1
(vim .])dt (RW, J o , .])dt (-R(.L w, .])dt.
0 0 0

We integrate by parts the left-hand-side of the above equality, and we deduce

1 1
1
(w,J)l 0 IVtJI2dt (Re, i)% J)dt.
0 0

Since J(7') = 0 for T = 0, 1, we deduce using (5.2.6)

1
IVtJI2dt < 0.
0

This implies V t J 0 which coupled with the condition J(0) = 0 implies J 0. The
proof is complete.

The notion of conjugacy is intimately related to the behavior of the exponential map.

Theorem 5.2.24. Let ( M , g) be a connected, complete, Riemann manifold and 6/0 6 M . A

point q 6 M is conjugated to quo along some geodesic If and only U" it is a critical value of
the exponential map

expC10 . T<10 M -> M.

218 Lectures on the Geometry of Manzfolds

Proof. Let q = expGo "u, 'U 6 T<10 M. Assume first that q is a critical value for e x p o , and 'U
is a critical point. Then D'U exp<10 (X) = 0, for some X E T'U(TC10M). Let 11(5) be a path
in T<10 M such that 11(0) = 'u and 13(0) = X. The map (5, t) »-> expqo (tv(s)) is a geodesic
variation of the radial geodesic 'Ye t »-> exp<10 (to). Hence, the vector field

W=
a
-
38 ¢s=O
e x p o (t'u(s))
is a Jacobi field along 'Yu - Obviously W(0) = 0, and moreover
W(1) =
a expC10(u(s)) = D'U expC10 (X) = 0.
On the other hand this is a nontrivial field since
a
VtW :- VS ' s = 0
expq0(w(s))
lllllllllll

VSv(s) l5=0;Ȏ 0.
(916
This proves quo and q are conjugated along 7'U.
Conversely, assume 'U is not a critical point for expo. For any X 6 T'U (TC10 M ) denote
by Jx the Jacobi field along 'Y'U such that
JX(q0) = 0. (5.2.7)
The existence of such a Jacobi field follows from Exercise 5.2.16. As in that exercise,
denote by 3qo the space of Jacobi fields J along 7'U such that J(q0) = 0. The map
T»v(Tq-,Ml -) 390 X I-> Jx
is a linear isomorphism. Thus, a Jacobi field along v'U vanishing at both quo and q must have
the form Jx, where X E T'U(T<10M) satisfies D'U exp<10 (X) = 0. Since 'u is not a critical
point, this means X = 0 so that Jx E 0.

Corollary 5.2.25. On a complete Riemann manifold M with non-positive sectional curva-

ture the exponential map expq has no critical values for any q E M .

We will see in the next chapter that this corollary has a lot to say about the topology
of M.
Consider now the following experiment. Stretch the round two-dimensional sphere of
radius l until it becomes "very long". A possible shape one can obtain may look like in
Figure 5.4. The long tube is very similar to a piece of cylinder so that the total (= scalar)
curvature is very close to zero, in other words is very small. The lesson to learn from this
intuitive experiment is that the price we have to pay for lengthening the sphere is decreasing
the curvature. Equivalently, a highly curved surface cannot have a large diameter. Our next
result offers a more quantitative description of this phenomenon.

Theorem 5.2.26 (Myers). Let M be an n-dimensional complete Riemann manifold. If for

a l l X E Vect (M)
Ric(X,X) > ( N ,r'2 Umm 2
7

then every geodesic of length > 'ITT' has conjugate points and thus is not minimal. Hence
diam (M) = sup{dist (p, q) ; p, q E M } < or,
and in particular, Hopf-Rinow theorem implies that M must be compact.
 Elements of the Calculus of Variations 219

Fig. 5.4 Lengthening a sphere.

Proof. Fix a minimal geodesic *y : [0, 6] -> M of length Z, and let €¢(0) be an orthonormal
basis of vector fields along by such that e,1(t) = W) Set Qo = '7(0)» and QUO = by(€). Since
*y is minimal we deduce

E**(v, v) > _ 0 W e T.
Set Wz- = sin(1rt/€)ei. Then
~£

f~Z
sin2(1rt/6) (TR/£2 - (R(€z'» w, @)) it.
We sum overs = 1, ,n 1, and we obtain
n-1 e
sin2 Trt/K ((n - lltr2/H2 - Ric Idyll dt > 0.
11 1 0

If 6> try, then

(n - 1)tr2/62 - Ric (M) < 0,
so that,
n 1

i 1
Hence, at least for some W , we have < 0. This contradicts the minimality
of by. The proof is complete.

We already know that the Killing form of a compact, semisimple Lie group is positive
definite, see Exercise 4.1.19. The next result shows that the converse is also true.
Corollary 5.2.27. A semisimple Lie group G with positive definite Killing pairing is com-
pact.
220 Lectures on the Geometry of Manzfolds

Proof. The Killing form defines in this case a bi-invariant Riemann metric on G. Its
geodesics through the origin 1 6 G are the 1-parameter subgroups exp(tX ) which are
defined for all t 6 R Hence, by Hopf-Rinow theorem G has to be complete.
On the other hand, we have computed the Ricci curvature of the Killing metric, and we
found
1
Ric (X, Y) X / ( x , y ) VX,Y Q LG.

The corollary now follows from Myers' theorem.

Exercise 5.2.28. Let M be a Riemann manifold and q E M . For the unitary vectors
X , Y e TqM consider the family of geodesics
'yS(t) = expq t(x + bY).
2
Denote by Wt 6%8 the associated Jacobi field along Volt)- Form f(t) Prove
the following.
(a) Wt = Dux €Xpq (Y) = Frechet derivative of 'U »-> expq(v).
1
(b) f(t) I t2 5 <R(y, x)x, y>q¢4 + o(¢5).
'

(c) Denote by mi a collection of normal coordinates at q. Prove that

l i .
gkgl$l = 6k£ - §Rkzge$ $3 -|- O(3).

det .Q 1 -§RZ]a:'a:7 -|- O(3).

(d) Let

D AQ) = {x € TqM ; Ia2l< r}.

Prove that if the Ricci curvature is negative definite at q then

VOID(D t~(qll D Aqll

<_volg(expq(

for all 1° sufficiently small. Above, v010 denotes the Euclidean volume in TqM while v01g
denotes the volume on the Riemann manifold M.

Remark 5.2.29. The interdependence "curvature-topology" on a Riemann manifold has

deep reaching ramifications which stimulate the curiosity of many researchers. We refer
to [35] or [94] and the extensive references therein for a presentation of some of the most
attractive results in this direction.

5.2.3 The Hamilton-Jacobi equations

Suppose that (M, g) is a smooth, m-dimensional manifold, Po E M, U is an open, geodesi-
cally convex neighborhood of Po and (x1, . . . , 1.m are normal coordinates on U centered
at Po. We have a smooth function
1 7 : U x U - > [ 0 , o o ) , 7 7 ( p,q ) S( p , q I2 7 S(p, q distg(p, q )-
 Elements of the Calculus of Variations 221

We want to investigate the partial derivatives of 17 at (p Of p 0)- Using the above normal
coordinates we regard T/ as a function 17 = n(ac, y) defined in an open neighborhood of
(0, 0) E RM X Run. The next result is classical and it is the main tool that will help us find
the Taylor expansion of n(ac, y) at (0, 0).

Theorem 5.2.30. Fix y 6 U . The function U 9 al-> 17(ac°, y) satisfies the Hamilton-Jacobi
equation.
317(a3, y) é117(;v, y)
gklxl <9;z:'* 331e
4'7(23,y), Way- (5.2.8)

Proof. Our approach is inspired from [114, pp. 167-170]. Denote by ' Y y Z 'Yayltl the
unique geodesic that connects Hz: to y parametrized such that
1
S(a:,y) = distg(x,y) |%,y(tlldt-
0

Since t »-> l%,y(tll is constant along a geodesic we have

1
f7(:/1» y) = S(:z;, y)2 l%,y(t)l2d¢ L(vm,y, rryldta
0 0
where

L(q , ii) 9~i(q)<i"<i1

M
We now let the point :u vary along an arbitrary smooth path co: 55(5), lsl < l in U. Set
130 x(0) E U. We write

Us (t) = Y:1:(s),9ltl, 5u(t)

d
lllllllllll

ds
IS 0'u»5(t)-

We have (using Einstein's conventions)

~1
877
nu griz 607 to)
d
ds 15 0
S(:13(sl, y)
d
ds
L(u1.9,u°S9)d8

(5.1.2) I'1 A I . ~1 of; .

i
5u'dt -I- .5u'Jdt
'0 '0 <9q3
(integrate by parts)
1 AL d of;
/(
0 <9qi1 dt go"Z
=0

Since the path 33(5) was arbitrary we deduce

as aL (Had)
. , v. •

35 W z
We set
1 AL 1 1.
Pa 29 i 2g7k 7
222 Lectures on the Geometry of Manzfolds

and we introduce the Hamiltonian

l l -.
H(x1»12) glhPj 2 9 gikw Pa = §5 xkp]

l
_ j _= l = L(a3 l
2 PJ 29" 7 .
The quantity L(aC, H?) is constant along the geodesic 'Y2:,y and, at the endpoint x, we have
_
t1(x1 y) = L(;z:,a:) = H(x112) =
1
XH(x', 3xj 17(a:, y) )-
.

Expanding we get
_ l
_ ZN

Besides their intrinsic interest, the Hamilton-Jacobi equations can be used to produce an
explicit description of the Taylor expansion for the square of the geodesic distance n(ac, y).
This Taylor expansion will play an important part in Subsection 11.4.5.
With this goal in mind we introduce some notation. If f : f(t1, . . . ,TN) is a smooth
function defined in a neighborhood of 0 6 RN and A: is a nonnegative integer, then we
denote by [f] k the degree k-homogeneous part in the Taylor expansion of f at 0, i.e.,
1
[f]k=k, lllllllll

@?f4 ¢=0ta 6 IRM 7 ,tN].

' lot=I¢

In the coordinates ( w the metric g has the form (using Einstein's summation convention
throughout)

g god$d$j 7

where gig satisfy the estimates (see Exercise 5.2.28)

1 . .
go! = (5k2 - §RZ/¢3£(0l.Q?z$l73 + (5.2.9)

For any e x n matrix A that is small enough we have

(1-A) l -I-A-I-A2-I-~'~.

We deduce that
1 . .
gk2 = (Sky -|- §R»L1¢]£l0l3l3?£U] -|- (5.2.10)

The function 17 satisfies a Hamilton-Jacobi equation (5.2.8) and the obvious symmetry con-
ditions
771
2
HGH, y) = VW, 33), v(0» 23) VW, 0) = 1221 (w, 0. (5.2.11)
o n

= T)(ac,a2l
i 1

The degree 2 part of the Taylor expansion of T) satisfies similar symmetries. This implies
TTY
. " 2
2
[7712 lac° al EQ, y') (5.2.12)
ii 1
 Elements of the Calculus of Variations 223

The symmetries (5.2.11) suggest the introduction of new coordinates (u, u) on U X U,

ni ni .
Up $ ?J 7 UP ask-I-yj.

Then
xi l 1
2
(Up -|- U i l 7 yJ
2
l'Uj - up), ex% = 3 +3
The equality (5.2.10) can be rewritten as
1
gke(Q3) (SM + § Ri1<:je(U¢ -|- Ui)(U8 -|- Ujl -|- O(3). (5.2.l3)
z,_7

The symmetry relations (5.2.l 1) become

2
n(u,v)=n( U7 UI, 17(u»u)=luI 7 (5.2.l4)

while (5.2.l2) changes to

[7711 = 0, ['7]2 = III 2 (5.2.l5)

The equality (5.2.8) can be rewritten

kl
g (H1)('7'k + v ' ) ( v Q +t7') =4v- (5.2.16)
k,l
zAp zAg

Note that

[Ago = [Ado = [9kg]1 = 0, (5.2.17)

while (5.2.15) implies that

[Ak]1 2'LLk .
We deduce

4[77]3 [91¢i]0( [Ash [Ada -|- [A1<]2[A@]1 ) 2[Ak]2[Ak]1 4

1¢,e k k

We can rewrite this last equality as a differential equation for [tv]3 namely

['7]3 Uklguk + 8'Uk)[77]3'

k

We set P := [17]3 so that P is a homogeneous polynomial of degree 3 in the variables u, 'U.

Moreover, according to (5.2.14), the polynomial P is even in u and P(u, Fu) = 0. Thus P
has the form
P CZ~(u)u,~ + P0(0),
sP2

where Co (u) is a homogeneous polynomial of degree 2 in the variables u, and Po (u) is

homogeneous of degree 3 in the variables 'U.
224 Lectures on the Geometry of Manzfolds

We have

U/<78'Uk1.r P2 C'k(u)uk, QUO U1¢3vk P07 U1<:3uk P0 = 0.

k k k k

IQ3

The classical Euler equations imply

Ukgukp2 I 2P2.
k

We deduce
P 2P2 + QUO + QUO 7
where the polynomials QUO and Q1 are odd in the variable u. Since P is even in the variable
we deduce
'LL

QUO -|- Q1 Z 0,
so that P2 -|- P0 = P = 2P2. Hence P2 = P0 = 0 and thus

U]3 :: (). (5.2.18)

In particular

[Ak]2 = 0, Vi. (5.2.19)

Going back to (5.2.l6) and using (5.2.l7) and (5.2.19) we deduce

4['7]4 [9ke]2[Ak]1[A€]1 + E[9ke]0l[AI<]1[Ads + [Ak]3[A@11)

I<,e k,e
(5.2.20)
4 Z[gk£]2ukug + 2 Z 'u,k[Ak]3.
k,€ k

We set P := [17]4. The polynomial P is homogeneous of degree 4 in the variables u, u, and

it is even in the variable u. We can write P = P0 -|- P2 -|- P4, where
\

I
Czj1¢z'u»zuj'uk'u2, P2 ZQ@jlu)'uz'uj 7

k as
and P0 is homogeneous of degree 4 in the variables u, Qzj (UI is a homogeneous quadratic
polynomial in the variables u. We have

U1¢[A1¢]3 Uk(6uk -|- 8,,k)P.

k k

We have
uk3UkP2,, = 21/P2l/, U = 0 , l , 2, uk3UkP4 = 0,
k k

uk<9@kP2 ukQ¢j(5k¢vj + 5kiv¢) lQkau1<=uy + Q j k v j u k I.

k i n lm
 Elements of the Calculus of Variations 225

Using these equalities in (5.2.20) we deduce

mp4 + 4p2 + up0 = 4 [gklzukue + 4P4 + 2P2 + 7U//<78'I)k: P0
k,£ k

+ (Q *|' Q k j )'Ll,k'l}3 .
kJ
This implies P0 = 0 so that P P4 + PQ, and we can then rewrite the above equality as
PA 2 Z[9k]2uku@
I<:,e
+ 2; (QUO/
kJ
-I-Q1¢j
) 'LLk'Uj (5.2.21)

Note that the equality 1°(u,'u,) = lul2 implies P ( u , u) = 0 so that

P4(u) = P4('u,,u1 = -P2(u,u).

Therefore it suffices to determine P2~ This can be achieved using the equality (5.2.l3) in
(5.2.2l). We have
1
2 [gk£]2Uk'LL£ = Rik je -l- 'Ui)('Uj -l- U j ) u k u g
6
k,£ i,j,k,£
1
6
i,j
ii k,£
Rzkgeukuel 'U»;'Uj -l- Sj (U)'Uj,
J

@ii('u~)
where Sf (u) denotes a homogeneous polynomial of degree 3 in u. The equality (5.2.21)
can now be rewritten as
1
Z Qw(UlU¢Uj = 6Z y('0lu1Ua + .2 51(u)Uj (Q -'r Q k j >'LLk'U]
M M J
From this we read easily
l A

QMu) = gQwlul ikjeukue •

k,€

This determines P2 .
1
P2(u» u) = 6 Z QMUIWJJ (5.2.22)
'in
As we have indicated above P2 determines P4 .
l
P4('u,) = -P2 (u, Fu) R¢1<jeU¢UjU1<>Ue~ (5.2.23)
6
i,j,k,€

The skew symmetries of the Riemann tensor imply that P4 = 0 so that

l
['7]4(U» u) 6Z y(uluwy» QzJIU) = ZR@kgeukue- (5.2.24)
'As k,e

Example 5.2.31. Suppose that M is a surface, i.e., m = 2. Set

K i $1212 i R2121 i -R1221~
226 Lectures on the Geometry of Manzfolds

Note that K is the Gaussian curvature of the surface. Then

. 2 .
2
Q11 = Rlkleukue = Kun, Q22 = Rzkzeukue Ku1,
k,e k,£

Q12 = Rlkzeukue = -K'LL1'LL2 = Q21-

k,£

Hence

P2(u,v) =
K (u2'ul
2 2 2 2
-|- u102 - 2'Ll,1'LI2'U1'U2l = - ( U W 2 - UQU1)
K 2
6
In higher dimensions, if we restrict (5.2.24) to the surface

E x i M; ac°k=;u£=0, ' v ' k , € § Z { i , j }

II
O
.

we deduce
A

.. 2 .. 2
I n j - Kwua) Q J J , - KzJ 7

A K. . 2
Qzj'2 J = 'LJ
6 ( 'U/jV j 'U,j'Ui
 Chapter 6

The Fundamental Group and Covering Spaces

In the previous chapters we almost exclusively studied local properties of manifolds. This
study is interesting only if some additional structure is present since otherwise all manifolds
are locally alike.
We noticed an interesting phenomenon: the global "shape" (topology) of a manifold
restricts the types of structures that can exist on a manifold. For example, the Gauss-
Bonnet theorem implies that on a connected sum of two tori there cannot exist metrics with
curvature everywhere positive because the integral of the curvature is a negative universal
constant.

L
no
\

Fig. 6.1 Looking for unshrinkable loops.

We used the Gauss-Bonnet theorem in the opposite direction, and we deduced the
intuitively obvious fact that a sphere is not diffeomorphic to a torus because they have
distinct genera. The Gauss-Bonnet theorem involves a heavy analytical machinery which
may obscure the intuition. Notice that S2 has a remarkable property which distinguishes it
from T2: on the sphere any closed curve can be shrunk to a point while on the torus there
exist at least two "independent" unshrinkable curves (see Figure 6.1). In particular, this
means the sphere is not diffeomorphic to a torus.
This chapter will set the above observations on a rigorous foundation.

227
228 Lectures on the Geometry of Manzfolds

6.1 The fundamental group

6.1.1 Basic notions

In the sequel all topological spaces will be locally path connected spaces.

Definition 6.1.1. (a) Let X and Y be two topological spaces. Two continuous maps fo,f1 :
X -> Y are said to be homotopic if there exists a continuous map
F : [0, l] X X -> Y (t,;u) +-> Ftlzul,
such that, E E f form = 0, l. We write this as to al .
(b) Two topological spaces X, Y are said to be homotopy equivalent if there exist maps
f : X - > Y a n d g : Y - > X s u c h t h a t f o g ~ fly a n d g o f ~ ] l x . W e w r i t e t h i s X ~ Y .
(c) A topological space is said to be contractible if it is homotopy equivalent to a point.

Example 6.1.2. The unit disk in the plane is contractible. The annulus {l <lzl< 2} is
homotopy equivalent to the unit circle.

Definition 6.1.3. (a) Let X be a topological space and 51:0 6 X . A loop based at £130 is a
continuous map
'Y [0, l] -> X, such that 7(0) 7(1) 5120-

The space of loops in X based at 1:0 is denoted by Q(X, 230).

(b) Two loops '70, 'Y1 : I -> X based at 51:0 are said to be homotopie rel 130 if there exists a
continuous map
[0, l] -> X, (t, s) »-> I`%(8),
such that
Fi l 8 ) = ' Y f z ( 3 l ' é = 0 , l ,
and
(5 »-> Ft(.s)) 6 Q(X,£60) Vt 6 [0, 11.
We write this as '70 r\./$0 'Yl-

Note that a loop is more than a closed curve, it is a closed curve + a description of a
motion of a point around the closed curve.
Example 6.1.4. The two loops m m I ->@,v/(=(t) exp(2k1rt), k l , 2 are different
though they have the same image.

Definition 6.1.5. (a) Let '711 'Y2 be two loops based at 130 E X . The product o f 'Y1 and 'Y2 is
the loop
'Y1(28l» 0 S sl/2
'Y1 * 'Y2(3) ')/2(2S - II, l /2 < 3 < l
The inverse of a based loop by is the based loop 'Y defined by
'Y (so = v(1 S •

(c) The identity loop is the constant loop eH30 (5) $0.
 The Fundamental Group and Covering Spaces 229

Intuitively, the product of two loops 'Yi and 'Y2 is the loop obtained by first following 'Yi
(twice faster), and then 'Y2 (twice faster) .
The following result is left to the reader as an exercise.

Lemma 6.1.6. Let 050 "mo 011, 50-x051 and 70-x071 be three pairs of homotopic based
loops. Then
(0) * @0"'m0011 * 51.
O10
(b) Ag* 00 -IB()ex.

(c) to * e000-12000.

(d) (010 * 50) * 'Y02x00£0 * (Bo * 'Yol-

Hence, the product operation descends to an operation "~" on Q(X, x 0 ) / - a c 0 , the set of
homotopy classes of based loops. Moreover the induced operation is associative, it has a
unit and each element has an inverse. Hence (Q(X, 5 1 3 0 ) / - x 0 , -) is a group.

Definition 6.1.7. The group (Q(X,;u0)/~_,,30, .l is called the fundamental group (or the
Poincaré group) of the topological space X, and it is denoted by W1 (X, 5150). The image of
a based loop by in tf1 (X, xo) is denoted by iv] -

The elements of al (X, xo) are the "unshrinkable loops" discussed at the beginning of
this chapter.
The fundamental group w1(X,x0) "sees" only the connected component of X
which contains xo. To get more information about X one should study all the groups
{f/T1 (X, 3?)} a:€X -

Proposition 6.1.8. Let X and Y be two topological spaces. Fix two points, £130 6 X and
Yo 6 Y . Then any continuous map f : X -> Y such that f (1130) = Yo induces o morphism
of groups

f* : tfllX,€col -> W1(Y»3/ol,

satisfying the following functorialily properties.
(al (lx)* ]]'1r1lX,a:0)~
(b) If

(X, 5130) L (Y, 3/ol i> (Z, ZUNI

are continuous maps, such that f (;130l = 3/0 and g(y0) = Z0, then (9 o f)* I Q* o f*~
(C) Let f0,f1 : (X, 320) -> (Y, y0) be two base-point-preserving continuous maps. Assume
f0 is homotopic to al rel 5130, i.e., there exists a continuous map F : I X X -> Y , (t,x) +->
Fe(ac) such that, Fig(£cl E f l y ) for t = 0, l and Fe(to) E Yo. Then (f0)* = (f1l*-

Proof. Let 'Y E QlX,$0l_ Then f('yl E so, y0)» and one can check immediately that
y0
v;ftov' => f('y) I\.!
f('y"l.
Hence the correspondence
QIXMUo) 9 'Y * f('y) 6 Q(Y»Y0)
230 Lectures on the Geometry of Manzfolds

descends to a map f : tf1 lx, 5130) -> 7I'1 LY, y0l. This is clearly a group morphism. The
statements (a) and (b) are now obvious. We prove (c).
Let to, fl : (X, k g ) -> (Y, y0) be two continuous maps, and Ft a homotopy rel 5130
connecting them. For any by 6 Q(X, xo) we have

50 fo('Y) f1('Y) 51.

N-:yo
The above homotopy is realized by 8t I Ftl'}').

A priori, the fundamental group of a topological space X may change as the base point
varies and it almost certainly does if X has several connected components. However, if
X is connected, and thus path connected since it is locally so, all the fundamental groups
'/T1 (X, ac), cz: 6 X are isomorphic.

X
0< a<
r <> > X
Y 1
> a >

Fig. 6.2 Connecting base points.

Proposition 6.1.9. Let X be a connected topological space. Any continuous path a

[0, l ] -> X joining :UO to £111 induces on isomorphism

go* :7r1(X,:z:0l -> 771lX,£U1l,

defined by, CY* (M ) : a * by * a],° see figure 6.2.

Exercise 6.1.10. Prove Proposition 6.1.9.

Thus, the fundamental group of a connected space X is independent of the base point
modulo some isomorphism. We will write Try (X, pt) to underscore this weak dependence
on the base point.

Corollary 6.1.11. Two homotopically equivalent connected spaces have isomorphic fun-
damental groups.

Example 6.1.12. (a) 7l'1lRTL,ptl Tf1(pt,pt) = { l } -

(b) 'rr1(annulus) 7T1lS1l.

Definition 6.1.13. A connected space X such that '/T1 (X, pt) {l} is said to be simply
connected.
 The Fundamental Group and Covering Spaces 231

Exercise 6.1.14. Prove that the spheres of dimension _ 2 are simply connected.
Exercise 6.1.15. Let G be a connected Lie group. Define a new operation "*" on Q(G, l)
by
(Q * BMS) 0(s) . ms),
-
where denotes the group multiplication.
(a) Prove that a * B ""1 a * 5
(b) Prove that '/T1 (G, l ) is Abelian.

Exercise 6.1.16. Let E -> X be a rank T' complex vector bundle over the smooth manifold
X, and let V be a flat connection on E, i.e., F(V) = 0. Pick ac() 6 X, and identify the fiber
Exp with CC". For any continuous, piecewise smooth by 6 Q(X, xo) denote by T'Y = T l v l
the parallel transport along by, so that T'Y 6 GL (U).
(a) Prove that 0 - 3 3 o B => Ta = TB.
= To o Ta.
T,B*a

(b)
Thus, any flat connection induces a group morphism
T : 7r1(X,:z:0) -> GL((C") *y »->T'7- 1 .
This morphism (representation) is called the monodromy of the connection.

Example 6.1.17. We want to compute the fundamental group of the complex projective
space CIP". More precisely, we want to show it is simply connected. We will establish this
by induction.
~
For n = l, (CIP1 = S2, and by Exercise 6.1.14, the sphere S2 is simply connected. We
next assume (c1p'" is simply connected for k < n and prove the same is true for n.
Notice first that the natural embedding (jk-l-1 <-> C"-*1 induces an embedding (c1p'" <->
CIP". More precisely, in terms of homogeneous coordinates this embedding is given by
[ z 0 , . . . : Zn]»-> [z0,...,z;,,0,...,0] € UP".
Choose as base point pt = [1,0, . . . ,0] 6 CIP", and let by 6 Q((ClP",pt). We may assume
by avoids the point P = [0, . . . , 0, 1] since we can homotop it out of any neighborhood of
P.
We now use a classical construction of projective geometry. We project by from P to
the hyperplane ii" = Gp"-1 C--> CIP". More precisely, if C = [z0, . . . , zn] 6 UP", we
denote by ¢r(§) the intersection of the line PQ with the hyperplane J . In homogeneous
coordinates
: : iz0I1 - "ala • • • Z n - 1 ( 1 - ZnlvOl (= 2107
7 7 2 N - 1 7 0] when Zn ¢ 1).
Clearly or is continuous. For t 6 [0, l] define
7Tt lcl [z0(l - HzTL), . . . ,zn_1(l - toTL), (1 - t)zn].
Geometrically, '/To flows the point Q along the line PC until it reaches the hyperplane 32
Note that 'ii,¢(§) = C, Vt, and VC 6 %. Clearly, '/To is a homotopy rel pt connecting
*y = two (by) to a loop 'Y1 in Qi" 2 (CIPn-1 based at pt. Our induction hypothesis shows that
'Y1 can be shrunk to pt inside Qi" . his proves that CIP" is simply connected.
232 Lectures on the Geometry of Manzfolds

6.1.2 Of categories and functors

The considerations in the previous subsection can be very elegantly presented using the
language of categories and functors. This brief subsection is a minimal introduction to this
language. The interested reader can learn more about it from the monograph [74, 89].
A category is a triplet € = ( Ob (€), H o m ( € ) , o) satisfying the following conditions.

(i) Ob (6) is a set whose elements are called the objects of the category.
(ii) Hom((3) is a family of sets Hom (X, Y), one for each pair of objects X and Y. The
elements of Hom (X, Y) are called the morphisms (or arrows) from X to Y.
(iii) o is a collection of maps
O Hom(X, Yl X Hom(Y, Z) -> Hom(X, Z), ( f , g) +-> g O f,
which satisfies the following conditions.

(Cl) For any object X, there exists a unique element Hx wHom(X,X) such that,
fo]1x =f go]lx = g Vf 6 Hom(X,Y), Vg 6 Hom(Z,X).
(C2) 'v'f G Hom(X,Y), g E Hom(Y, Z), h 6 Hom(Z, W)
h<>(9of)=(ho9)of-

Example 6.1.18. • Top is the category of topological spaces. The objects are topological
spaces and the morphisms are the continuous maps. Here we have to be careful about one
foundational issue. Namely, the collection of all topological spaces is not a set. To avoid
this problem we need to restrict to topological spaces whose subjacent sets belong to a

.
certain Universe. For more about this foundational issue we refer to [74].
(Top, *) is the category of marked topological spaces. The objects are pairs (X, *), where
X is a topological space, and * is a distinguished point of X . The morphisms
(X, *I L (Y, o)
are the continuous maps f : X -> Y such that f(*) = <>.
. ]pVect is the category of vector spaces over the field IF. The morphisms are the IF-linear

.
maps.
Gr is the category of groups, while Ab denotes the category of Abelian groups. The

.
morphisms are the obvious ones.
RMOd denotes the category of left R-modules, where R is some ring.

Definition 6.1.19. Let 81 and (32 be two categories. A covariant (respectively contravari-
ant) functor is a map
St: Ob (81) X Ho1n(€1) -> Ob 1621 X Ho1n(€2),

( X , f ) »-> (5'(X)»5'(f))»
_gr ( f )
such that, if X i> Y, then 5"(X) 5"(fQ 5"(Y) (respectively 5'(X) 5'(Y)), and
(i) 5"(]1x) = ]1;~"<x),
(ii) 9(9) o 5'(f) = 5'(9 o f ) (respectively 5'(f) o 9(9) = 5'(9 O f))-
 The Fundamental Group and Covering Spaces 233

Example 6.1.20. Let V be a real vector space. Then the operation of right tensoring with
V is a covariant functor
L ®
®V R Veet * R Veet 7 U *V*-> U (X) v,
(UP -> UQ) (U1 (X) V L ->V
*V*-)1 U2 ®V).

On the other hand, the operation of taking the dual defines a contravariant functor,
* L *Lt *
R Vect * R Vect V -> v* 7 (u -> v) 'v-> (V - > U ) .
7

The fundamental group construction of the previous is a covariant functor

'/T1 (Top, *I -> Gr.
In Chapter 7 we will introduce other functors very important in geometry. For more infor-
mation about categories and functors we refer to [74, 89].

6.2 Covering Spaces

6.2.1 Definitions and examples

As in the previous section we will assume that all topological spaces are locally path con-
nected.

Definition 6.2.1. (a) A continuous map '7l' : X -> Y is said to be a covering map if, for
any y 6 Y, there exists an open neighborhood U of y in Y, such that 7r-1 (U) is a disjoint
union of open sets $2 C X each of which is mapped homeomorphically onto U by 7r. Such
a neighborhood U is said to be an evenly covered neighborhood. The sets ld are called the
sheets of 'fr over U.
(b) Let Y be a topological space. A covering space of Y is a topological space X, together
with a covering map '7T : X -> Y.
(c) If 7T : X -> Y is a covering map, then for any y 6 Y the set 7r"1 (kg) is called the fiber
over y.

Example 6.2.2. Let D be a discrete set. Then, for any topological space X, the product
X X D is a covering of X with covering projection or(;u, d) = at. This type of covering
space is said to be trivial.

Exercise 6.2.3. Show that a libration with standard fiber a discrete space is a covering.

Example 6.2.4. The exponential map exp : R -> S1, t »-> exp(2trit) is a covering map.
However, its restriction to (0, OO) is no longer a a covering map. (Prove this!)

Exercise 6.2.5. Let (M, g) and (M, go be two Riemann manifolds of the same dimension
such that (M, go is complete. Let Q25 : M -> M be a surjective local isometric, i.e., <I> is
smooth and
l'UI9 I
ID(25('v)l§ Vu E TM.
Prove that ¢ is a covering map.
234 Lectures on the Geometry of Manzfolds

The above exercise has a particularly nice consequence.

Theorem 6.2.6 (Cartan-Hadamard). Let (M, g) be a complete Riemann manifold with

non-positive sectional curvature. Then for every point q E M , the exponential map
expq . TqM -> M
is a covering map.

Proof. The pull-back h = €Xpq (9) is a symmetric non-negative definite (0, 2)-tensor field
on TqM. It is in fact positive definite since the map expq has no critical points due to the
non-positivity of the sectional curvature.
The lines t »-> to through the origin of TqM are geodesics of h and they are defined
for all t 6 R. By Hopf-Rinow theorem we deduce that (TqM, h) is complete. The theorem
now follows from Exercise 6.2.5.

Exercise 6.2.7. Let c? and G be two Lie groups of the same dimension and Q25 : 68 -> G
a smooth, surjective group morphism. Prove that go) is a covering map. In particular, this
explains why exp : R -> S1 is a covering map.

Exercise 6.2.8. Identify S3 C R4 with the group of unit quaternions

S 3 = { q 6 H ; lql=1}~
The linear space R3 can be identified with the space of purely imaginary quaternions
R3 = Im]H[ = {a:'£ + y j + z k } .
(a) Prove that q:z:q-1 6 lm H, */q 6 S3.
(b) Prove that for any q 6 S3 the linear map
Tq . lm H -> Imll-Il az »-> q;uql1
is an isometric so that Tq 6 SO(3). Moreover, the map
S3 9 q»->Tq 6 SO(3)
is a group morphism.
(c) Prove the above group morphism is a covering map.

Example 6.2.9. Let M be a smooth manifold. A Riemann metric on M induces a metric

on the determinant line bundle det TM. The sphere bundle of det TM (with respect to
this metric) is a 2 : l covering space of M called the orientation cover o f M .

Definition 6.2.10. Let X1 3 Y and X2 3 Y be two covering spaces of Y . A morphism

of covering spaces is a continuous map F : XI -> X2 such that 7r2 o F = '/VI, i.e., the
diagram below is commutative.
x1 F ) XI

X YA
If F is also a homeomorphism we say F is an isomorphism of covering spaces.
 The Fundamental Group and Covering Spaces 235

7r
Finally, if X -> Y is a covering space then its automorphisms are called deck transfor-
mations. The deck transformations form a group denoted by Deck (X, w).

Exercise 6.2.11. Show that Deck l]Rg_'§l'S1l

~=z.
Exercise 6.2.12. (a) Prove that the natural projection S" -> RIP" is a covering map.
(b) Denote by u the tautological (real) line bundle over RP". Using a metric on this line
bundle form the associated sphere bundle S Q L ) -> RIP". Prove that this fiber bundle
defines a covering space isomorphic with the one described in part (a).

6.2.2 Unique lzfting properly

'lr
Definition 6.2.13. Let X -> Y be a covering space and F : Z - > Y a continuous map. A
ly? of f is a continuous map F : Z -> X such that 7T o F =f,i.e., the diagram below is
commutative.

X
/VF
F'//
'ii
/

Z
.
/
f
>Y

Proposition 6.2.14 (Unique Path Lifting). Let X 1> Y be a covering map, by : [0, l] ->
Y a path in Y and :UQ a point in thejiber over Yo = 7/(0), £130 G tr"1 (yo). Then there exists
at most one lift offs, F : [0, 1] -> Y such that r(0) = 500.

Proof. We argue by contradiction. Assume there exist two such lifts, Fl, F2 : [0, 1] -> X .
Set

S t € [0, 1]; 1"1(t) F2(t) }.

II

The set S is nonempty since 0 6 S. Obviously S is closed so it suffices to prove that it is

also open. We will prove that for every S C S, there exists 5 > 0 such that 18 - 6, S -|- 51
[0, 11 C S. For simplicity, we consider the case S = 0. The general situation is entirely
similar.
We will prove that there exists To > 0 such that [0, kg] C S. Pick a small open neigh-
borhood U of kg such that 7T restricts to a homeomorphism onto rr(U). There exists To > 0
l
such that vi( [0, r01 C U,1l = l , 2. Since 'rroII1 = VTOF2, we deduce F1 1107T'01- F2 l[0:T01°
The proposition is proved.

Theorem 6.2.15. Let X l> Y be a covering space, and f : Z -> Y be a continuous map,
where Z is a connected space. Fix Z0 6 Z , and 5130 6 7r"1(y0), where Yo = f(z0). Then
there exists at most one lzft F : Z -> X of f such that F (20) = kg.
236 Lectures on the Geometry of Manu'olds

Proof. For each Z 6 Z let aZ be a continuous path connecting Zo to Z. If Fl, F2 are two
lifts of f such that F1 (to) = F2(z0) = xo then, for any Z 6 Z, the paths F1 = F(aZ), and
F2 = F2 (az) are two lifts of by = f (aZ) starting at the same point. From Proposition 6.2.14
we deduce that F1 E FQ, i.e., F1 (z) = F2(z), for any z 6 Z.

6.2.3 Homotopy lu'ting properly

Theorem 6.2.16 ( Homotop y lifting propert y). Let X l> Y be a covering space, f :
Z -> Y be a continuous map, and F : Z -> X be a ly? o f f . If
h : [0,l] X Z -> Y (t,z) »-> ht(z)
is a homotopy of f ( ho (z) f (z) ), then there exists a unique [Ut of h
H : [0, l] x Z -> X (t , z) *-> He(zl»
such that H0(z)

Proof. For each Z 6 Z we can find an open neighborhood UZ of Z 6 Z, and a partition

0 = to < to < . . . < tn = l, depending on z, such that h a p s [ti_1, to] XUZ into an evenly
covered neighborhood of h i i _ 1 (z). Following this partition, we can now successively lift
h IIXUz to a continuous map H = Hz : I X Uz -> X such that H0(§) = F(§), VC 6 Uz.
By unique lifting property, the liftings on I X Uz, and I X Uz, agree on I X (UZ, n UZ2),
for any 217 212 6 Z, and hence we can glue all these local lifts together to obtain the desired
lift H on I X Z .

Corollary 6.2.17 (Path lifting property). Suppose that X l> Y is a covering map, y0 6
Y , and '7 : [0, l] -> Y is a continuous path starting at to. Then, for every £130 G 7r"1(y0l
there exists a unique lzft F : [0, l] -> X offs starting at 1/30.

Proof. Use the previous theorem with f : {pt} -> Y , f (pi = 1/(Ol, and htlPtl 'Yltl~
Corollary 6.2.18. Let X l> Y be a covering space, and Yo E Y. Iffy0, 71 € Q(Y, 3/ol are
homotopic rel 3/0, then any lifts Fo, F1 which start at the same point also end at the same
point, i.e., poll) = I`1(1l.

Proof. Lift the homotopy connecting 'Yo to 'Yi to a homotopy in X . By unique lifting
property this lift connects F0 to Fl. We thus get a continuous path p¢(1) inside the fiber
,/T-1 (yo) which connects I`0(1) to F1(1l~ Since the fibers are discrete, this path must be
constant.
'IT
Let X -> Y be a covering space, and Yo 6 Y. Then, for every ac 6 7T-1(?/0)» and any
*y 6 QW, yo), denote by Fx the unique lift of by starting at x. Set

ac-*y := Fas (1).

By Corollary 6.2.18, if U.) 6 Q(Y, yo) is homotopic to by rel Yo, then
ac-fy ac-oJ.
 The Fundamental Group and Covering Spaces 237

Hence, zz:-7 depends only upon the equivalence class [by] E Try (Y, y0l. Clearly
at . (['y] . M) i (co . ['y]l . [~J],
and
£8 • bY() My

so that the correspondence

-1 -1
7r (Yu) X 7T1(Y»3/ol 9 ( J o ) +-> 2 3 ° ' y Q 3 3 - 7 Q 7r (to)
defines a right action of 7r1(Y, y0) on the fiber 7r"1(y0). This action is called the mon-
odromy of the covering. The map as »-> Hz: by is called the monodromy along by. Note
that when Y is simply connected, the monodromy is trivial. The map 'IT induces a group
morphism

7r* W1(X7$0) -)tFlly,y0l 5130 € t l ' - 1 l Y 0 l .

Proposition 6.2.19. or* is injective.

Proof. Indeed, let by 6 Q(X, 1110) such that WW) is trivial in 7r1(Y, to). The homotopy
connecting 7r('y) to ey0 lifts to a homotopy connecting by to the unique lift of 2Y0 at $150,
which is efro .

6.2.4 On the existence of l#ts

Theorem 6.2.20. LetX l> Y be a covering space, :cg E X, Yo = tr(a:0l 6 Y , f : Z -> Y
a continuous map and 2 0 6 Z such that f (z0) = Yo~ Assume the spaces Y and Z are
connected (and thus path connected). f admits a lzft F : Z -> X such that F(z0) = £130 If
and only If
f* (7t1(27 Zgll C 7T* (7T1 lx,cc0ll (6.2. 1)

Proof. Necessity. If F is such a lift then, using the functoriality of the fundamental group
construction, we deduce f* = Tr* o F*. This implies the inclusion (6.2.l).
Sufficiency. For any Z 6 Z, choose a path 'Ye from Zo to Z. Then oz f(vzl is a
path from Yo to y = f(zl. Denote by A z the unique lift of aZ starting at 1130, and set
F(z) = AZ(1). We claim that F is a well defined map.
Indeed, let cuZ be another path in Z connecting Z0 to Z. Set AZ := f(wz), and denote by
AZ its unique lift in X starting at 330. We have to show that Az(1) = AZ (1). Construct the
loop based at Z0

Be We * 'Ye
Then f(5z) is a loop in Y based at Yo. From (6.2. 1) we deduce that the lift B Z of f(5z) at
E X is a closed path, i.e., the monodromy along f(@z) is trivial. We now have
330

AZ(1) I BZ(1/2) I AGO) I A2(1).

238 Lectures on the Geometry of Manzfolds

This proves that F is a well defined map. We still have to show that this map is also
continuous.
Pick Z 6 Z. Since f is continuous, for every arbitrarily small, evenly covered neigh-
borhood U of f (z) 6 Y there exists a path connected neighborhood V of Z E Z such that
f ( V ) C U. For any C E V pick a path 0 = UQ in V connecting z to Q. Let w denote
the path w = 'Ye * UQ (go from Z0 to Z along Vi, and then from Z to C along up). Then
F(§) = Q(1), where Q is the unique lift of f(w) starting at 5130. Since (f(C) E U, we
deduce that 9(1) belongs to the local sheet E, containing F(z), which homeomorphically
covers U. We have thus proved Z E V C F-1(E). The proof is complete since the local
sheets E form a basis of neighborhoods of F(z).
'ii
Definition 6.2.21. Let Y be a connected space. A covering space X -> Y is said to be
universal if X is simply connected.

Corollary 6.2.22. Let XI Y (i=0,]) be two covering spaces of Y . Fix Xi E Xi such

that p0(:v0) = p(:u1) = Yo E Y . If X0 is universal, then there exists a unique covering
morphism F : X0 -> XI such that F(:z:0) = al.

Proof. A bundle morphism F : X0 -> XI can be viewed as a lift of the map Po : X0 Y

to the total space of the covering space defined by lU1» The corollary follows immediately
from Theorem 6.2.20 and the unique lifting property.

Corollary 6.2.23. Every space admits at most one universal covering space (up to isomor-
phis).

Theorem 6.2.24. Let Y be a connected, locally path connected space such that each of its
points admits a simply connected neighborhood. Then Y admits an (essentially unique)
universal covering space.

Sketch of proof. Assume for simplicity that Y is a metric space. Fix Yo 6 Y. Let 90
denote the collection of continuous paths in Y starting at 3/0» It can be topologized using
the metric of uniform convergence in Y. Two paths in TYoare said to be homotopic rel
endpoints if they we can deform one to the other while keeping the endpoints fixed. This
defines an equivalence relation on Ty0. We denote the space of equivalence classes by Y,
and we endow it with the quotient topology. Define p : Y -> Y by

p(['y]) = 7(1) W Q Ty/0°

Then l},pl is a universal covering space of Y.

Exercise 6.2.25. Finish the proof of the above theorem.

Example 6.2.26. The map 0 2 3 5 1 is the universal cover of S1. More generally, exp :
R"->T"
(to, . . . Jn) +-> (exp(27rit11, . . . ,exp(2¢ritn1)), ii \ -1,
 The Fundamental Group and Covering Spaces 239

is the universal cover of T". The natural projection p STL -> RIP" is the universal cover
of RIP".

Example 6.2.27. Let (M, g) be a complete Riemann manifold with non-positive sectional
curvature. By Cartan-Hadamard theorem, the exponential map expq : TqM -> M is a
covering map. Thus, the universal cover of such a manifold is a linear space of the same
dimension. In particular, the universal covering space is contractible! ! !
We now have another explanation why exp : R" -> T" is a universal covering space
of the torus: the sectional curvature of the (flat) torus is zero.

Exercise 6.2.28. Let ( M , g l be a complete Riemann manifold and p : M -> M its univer-
sal covering space.
(a) Prove that M has a natural structure of smooth manifold such that p is a local diffeo-
morphism.
(b) Prove that the pullback p*g defines a complete Riemann metric on M locally isometric
with g.

Example 6.2.29. Let ( M g ) be a complete Riemann manifold such that

Ric: ( X , X ) > const.IXI2go (6.2.2)
where const denotes a strictly positive constant. By Myers theorem M is compact. Us-
ing the previous exercise we deduce that the universal cover M is a complete Riemann
manifold locally isometric with (M, g). Hence the inequality (6.2.2) continues to hold on
the covering M. Myers theorem implies again that the universal cover M is compact!! In
particular, the universal cover of a semisimple, compact je group is compact!!!

6.2.5 The universal cover and the fundamental group

Theorem 6.2.30. Let X' 3> X be the universal cover of space X . Then
7TH lX, oil 2 Deck (X -> x).
Proof. Fix £0 6 X and set £130 = Plfol- There exists a bijection
Ev :Deck (X) -> p - o > ,
given by the evaluation
Ev (Fl : F(€0).
For any E Tr' 1 (xo), let 'kg be a path connecting go to g. Any two such paths are homotopic
rel endpoints since X is simply connected (check this). Their projections on the base X
determine identical elements in 711 ( X , 330). We thus have a natural map
\I/ Deck(X) -> tf1lX»1°ol F *->p(vF(€0))-
The map \11 is clearly a group morphism. (Think monodromy!) The injectivity and the
surjectivity of \If are consequences of the lifting prope ies of the universal cover.

Corollary 6.2.31. If the space X has a compact universal cover, then w1 ( X , pt) isjinite.
240 Lectures on the Geometry of Manzfolds

Proof. Indeed, the fibers of the universal cover have to be both discrete and compact.
Hence they must be finite. The map Ev in the above proof is a bijection onto Deck (X).

Corollary 6.2.32 (H. Weyl). The fundamental group of a compact semisimple group is
finite.

Proof. Indeed, we deduce from Example 6.2.29 that the universal cover of such a group is
compact.

Example 6.2.33. From Example 6.2.11 we deduce that tel IS1) r\/

(Z, >-
~=
Exercise 6.2.34. (a) Prove that 711 (]R]P'" ,pt) Z2,'v'n > 2.
(b) Prove that W1 (Tn) r\/ Zn

Exercise 6.2.35. Show that the natural inclusion U(n - 1) <-> U(n) induces an isomor-
phis between the fundamental groups. Conclude that the map
det : U(n) -> S1
induces an isomorphism
~ 7T1(S1, 1) ~= Z.
7r1(U(n), 1) =
 Chapter 7

Cohomology

7.1 DeRham cosmology

7.1.1 Speculations around the Poincaré lemma

To start off, consider the following partial differential equation in the plane. Given two
smooth functions P and Q, find a smooth function u such that
311,
llllllllllll-

p, Q Q. (7.l.1)
3:6 <99
As is, the formulation is still ambiguous since we have not specified the domains of the
functions u, P and Q. As it will turn out, this aspect has an incredible relevance in geome-
try.
Equation (7.l.l) has another interesting feature: it is overdetermined, i.e., it imposes
too many conditions on too few unknowns. It is therefore quite natural to impose some ad-
ditional restrictions on the data P, Q just like the zero determinant condition when solving
overdetermined linear systems.
To see what restrictions one should add it is convenient to introduce the 1-form a
Pdat + Qdy. The equality (7.l.1) can be rewritten as

du a. (7.1.2)

If (7.1.2) has at least one solution u, then 0 du do, so that a necessary condition for
existence is

do 0, (7.13)

i.e.,
UP _ 662
fly - 3:13

A form satisfying (7.1.3) is said to be closed. Thus, if the equation du = a has a solution
then a is necessarily closed. Is the converse also true?
Let us introduce a bit more terminology. A form a such that the equation (7. 1.2) has a
solution is said to be exact. The motivation for this terminology comes from the fact that

241
242 Lectures on the Geometry of Manzfolds

sometimes the differential form du is called the exact differential of u. We thus have an
inclusion of vector spaces
{exact forms} C {closed forms}.
Is it true that the opposite inclusion also holds?
Amazingly, the answer to this question depends on the domain on which we study
(7.1.2). The Poincaré lemma comes to raise our hopes. It says that this is always true, at
least locally.

Lemma 7.1.1 (Poincaré lemma). Let C be an open convex set in R" and a 6 Qk (C).
Then the equation
du = a (7.l.4)

has a solution u p Q k - 1 (Cl if and only if a is closed, do = 0.

Proof. The necessity is clear. We prove the sufficiency. We may as well assume that
0 E C. Consider the radial vector field on C
F £17i3$i 7

and denote by (Dt the flow it generates. More explicitly, <I>t is the linear flow
q>t(:v) = e t , at E R"

The flow lines of <I>t are half-lines, and since C is convex, for every a 2 € C , and any t < 0
we have q>t(x3 E C.
We begin with an a priori study of (7.1.4). Let u satisfy du a. Using the homotopy
formula,
L; d'£,=' + i»,='d,
we get

dL;/u = d(di;~' -|- i,=°dlu = di;-'a d(L,=anu - i,=*&) = 0.

This suggests looking for solutions up of the equation
Lw = , (p E sz'°-1(c). (7.1.5)
If go is a solution of this equation, then
L,=*dgo = dL,=*§O = Di,=*& I L,=*(I - iR*d& I L,=*&.
Hence the form SO also satisfies
L,,='(dgo - Q) = 0.
Set w := dgo -01 : E l l d m l Using the computations in Subsection 3.1.3 we deduce
L i m = do 7

SO that
L lL,=WI)d$l3I = 0.
I
 Cohomology 243

We deduce that L,=-cuI = 0, and consequently, the coefficients W I are constants along the
flow lines of ®t which all converge at 0. Thus,
Ld] C] const.
Each monomial cfdacI is exact, i.e., there exist T1I E QL6-1(Cl such that dn; efdaxl. For
example, whenI = 1 < 2 < < Is
day1 Add2 /\ -- /\ dank = d(xldx2 /\ - Adel.
•

Thus, the equality LfrW = 0 implies w is exact. Hence there exists 17 E Qk-1lcl such that
d((0 - H) &7

i.e., the differential form 'LL := SO 17 solves (7.1.4). Conclusion: any solution of (7.l.5)
produces a solution of (7.1.4).
We now proceed to solve (7.1.5), and to this aim, we use the flow ®%. Define
0
u l@t)* l'iv=*Oz)dt. (7.l.6)

Here the convexity assumption on C' enters essentially since it implies that
<I>t*(C)cC who,
so that if the above integral is convergent, then u is a form on C. If we write

l<I>t)* l'iv='(}£l i WW)dwI 7

III=k-1

then
0
u(a:) u1(w)dt)
t do (7.1.7)
ECU
I -OO

We have to check two things.

A. The integral in (7.1.7) is well defined. To see this, we first write
a= ajdai J 7
III=I<=
and then we set

A(22) max la.](7'x)l.

J£<T<1
Then
t.
(<1>v)*(w) 62? a.](et;u)d;u'] 7

SO that

l77§(g3)l < C€tl$UIAl3Il Vt < 0.

This proves the integral in (7, 1.7) converges.
244 Lectures on the Geometry of Manzfolds

B. The differential form 'LL defined by (7.1.6) is a solution of (7.1.5). Indeed,

1
L,,="u = lim ((<1>S)*u - u
s ( ) S
0 0
1
= lim -
s ( ) S ( [OO -[of (<I>t)*(i,=»a)dt )
(<1>Sl* (<I>t)*(i,,='a)dt
0
lim
s ( ) (/~ I -[of (<I>t)*(1l,=»0)dt )
S
_OO
(Qt * ii,=°Ozldt

lim
8->O

The Poincaré lemma is proved.

/ 0
(<1>')*(=~a)dt Z (®0)*(if0) 'i,=*&.

The local solvability does not in any way implies global solvability. Something happens
when one tries to go from local to global.

Example 7.1.2. Consider the form do on R2 \ {0} where or, 9) denote the polar coordinates
in the punctured plane. To write it in Cartesian coordinates (so, y) we use the equality
tan 9 y
cz:
so that
(1 + tank me = y
-do:
$2
+
d
:u
y and
lllllllllll
l + dO
-ydfzz + ardy
1:2 7

i.e.,
-ydx + :Ody
do a.
i
£u2 + y2
Obviously, do = d29 = 0 on R2 \ {0} so that a is closed on the punctured plane. Can we
find a smooth function u on R2 \ {0} such that du = a?
We know that we can always do this locally. However, we cannot achieve this globally.
Indeed, if this was possible, then

du a do = 2tr.
Sl Sl Sl
On the other hand, using polar coordinates u = u(7°' H) we get
au 27r NU
du -do 0 d9=u(1,2trl -u(1,0) = 0 .
S1 $1 <99
Hence on R2 \ {0}
{exact forms} 7é {closed forms}.
We see what a dramatic difference a point can make: R2 \ {point} is structurally very
different from R2.

The artifice in the previous example simply increases the mystery. It is still not clear
what makes it impossible to patch-up local solutions. The next subsection describes two
ways to deal with this issue.
 Cohomology 245

7.1.2 beef vs. DeRham

Let us try to analyze what prevents the "spreading" of local solvability of (7.1.4) to global
solvability. We will stay in the low degree range.

The tech approach. Consider a closed 1-form Ld on a smooth manifold. To solve the
equation du = w, 'LL E coo (M) we first cover M by open sets (Ua) which are geodesically
convex with respect to some fixed Riemann metric.
Poincaré lemma shows that we can solve du = Ld on each open set Ua so that we can
find a smooth function fa E C'°°(UO,) such that dfa = w. We get a global solution if and
only if

f<w=fo-f 6 0 oneach Uo,5 =UOfMU5;£(l).

For fixed a, the solutions of the equation du = Ld on Uo, differ only by additive constants,
i.e., closed 0-forms.
The quantities fem satisfy d o g = 0 on the (connected) overlaps U ; so they are con-
stants. Clearly they satisfy the conditions

.taB -'r fBv -1- f»yo = 0 on every U0 Be' Uaf'WU5VWU§£@. (7.l.8)

On each UO5we have, as we have seen, several choices of solutions. Altering a choice is
tantamount to adding a constant fa -> to, + ca. The quantities fan change according to

fog -> fag + Co - CO.

Thus, the global solvability issue leads to the following situation.

Pick a collection of local solutions fa. The equation du = Ll.) is globally solvable if we
can alter each fa by a constant ca such that

foe ICE ca) Va,B such that Uog ¢ (D. (7.1.9)

We can start the alteration at some open set Up, and work our way up from one such open
set to its neighbors, always trying to implement (7.l.9). It may happen that in the process
we might have to return to an open set whose solution was already altered. Now we are in
trouble. (Try this on S1 , and w = dH.) After several attempts one can point the finger to the
culprit: the global topology of the manifold may force us to always return to some already
altered local solution.
Notice that we replaced the partial differential equation du = CU with a system of linear
equations (7.l.9), where the constants fog are subject to the constraints (7.l.8). This is no
computational progress since the complexity of the combinatorics of this system makes it
impossible solve in most cases.
The above considerations extend to higher degree, and one can imagine that the com-
plexity increases considerably. This is however the approach tech adopted in order to
study the topology of manifolds, and although it may seem computationally hopeless, its
theoretical insights are invaluable.
246 Lectures on the Geometry of Manzfolds

The DeRham approach. This time we postpone asking why the global solvability is not
always possible. Instead, for each smooth manifold M one considers the Z-graded vector
spaces

B°(M) 8'"(M)» ( 8 0 ( M ) w) z°(m) .

o

kW),
k>o k>0

where

Bk(M) = { do E Qk:(M); w E Qk-1(M) }= exact k - forms,

and

Zk(M= {v€9"'(M); do= 0 } closed k - forms.

Clearly Bk C Zk*. We form the quotients,
M) l /Bkl
H1¢( - zklm M ).
Intuitively, this space consists of those closed k-forms w for which the equation du = LU
has no global solution u E Qk-l ( M ) . Thus, if we can somehow describe these spaces, we
may get an idea "who" is responsible for the global nonsolvability.

Definition 7.1.3. For any smooth manifold M the vector space Hk (M) is called the k-th
DeRham cosmology group of M .

Clearly HI<=(M) = 0 for /Q > dim m.

Example 7.1.4. The Poincaré lemma shows that Hk UR") 0 for k > 0. The discussion
in Example 7.1.2 shows that H1(R2 \ {0}) ¢ 0.

Proposition 7.1.5. For any smooth manifold M

dim H0(Ml = number of connected components of M .

Proof. Indeed

H"( M) = 20( M) = { f Q C M ) , i f = 0}.

Thus, H0(Ml coincides with the linear space of locally constant functions. These are
constant on the connected components of M.

We see that H 0 ( M ) , the simplest of the DeRham groups, already contains an important
topological information. Obviously the groups Hk are diffeomorphism invariants, and its
is reasonable to suspect that the higher cohomology groups may contain more topological
information.
Thus, to any smooth manifold M we can now associate the graded vector space
H.(m) ¢= 69 H'<=(M).
k>o
 Cohomology 247

A priori, the spaces Hk (Ml may be infinite dimensional. The Poincaré polynomial of M,
denoted by Pm (t), is defined by

Pm (t) H* dim H"'(m),

k>o

every time the right-hand-side expression makes sense. The number dim Hk (M) is usually
denoted by bk(M), and it is called the k-zh Betti number of M. Hence

Pm (t) be(m)tk*.
k

The alternating sum

x(M) l-1lkbklMlv
k

is called the Euler characteristic o f M .

Exercise 7.1.6. Show that Psi (t) l t.

We will spend the remaining of this chapter trying to understand what is that these
groups do and which, if any, is the connection between the two approaches outlined above.

7.1.3 Very little homological algebra

At this point it is important to isolate the common algebraic skeleton on which both the
DeRham and the (lech approaches are built. This requires a little terminology from homo-
logical algebra.
" i n the sequel all rings will be assumed commutative with 1.

Definition 7.1.7. (a) Let R be a ring, and let

C Cn Ly Dn
HGT n Z

be two Z-graded left R-modules. A degree k-morphism <15 : C' -> D' is an R-module
morphism such that

<p(c"> C Dn+k 'v'n E z.

(b) Let
C. Cn
n Z

be a Z-graded R-module. A boundary (respectively coboundary) operator is a degree -1

(respectively a degree 1) endomorphism d : C' -> C' such that d2 = 0.
A chain (respectively c o c h i n ) complex over R is a pair (C°, d), where C' is a Z-
graded R-module, and d is a boundary (respectively a coboundary) operator.
248 Lectures on the Geometry of Manzfolds

In this book we will be interested mainly in cochain complexes so in the remaining part
of this subsection we will stick to this situation. In this case cochain complexes are usually
described as (C' = €Bn€ZC", d). Moreover, we will consider only the case C" = 0 for
n<0.
Traditionally, a cochain complex is represented as a long sequence of R-modules and
morphisms of R-modules

(C'°»d) ->c"-1 on > c" 1

"">c7n-I-1 -> 7

such that range ldn_1l C her ldnl, i.e., dndn-1 = 0.

Definition 7.1.8. Let

dn- dn
->Cyn 1 HJ" >Cv'rL-I-1->-~-
be a cochain complex of R-modules. Set
Z"(C') := her dn B"(C) :: range (d,,_1).
The elements of Z" (C) are called cocycles, and the elements of B" (C) are called cobound-
aries.
Two cocycles c,c' E Zn*(C') are said to be eohomologous if C - c' E Bn*(C). The
quotient module
H"(Cl Z"(C)/B"(Cl
is called the n-th cosmology group (module) of C. It can be identified with the set of
equivalence classes of cohomologous cocycles. A cochain complex complex C' is said to
be acyclic i f H"(C') = 0 for all n > 0.

For a cochain complex (C', d) one usually writes

H'((1) = H . ( a d ) = @ H"((>).
n>0

Example 7.1.9 (The DeRham complex). Let M be an m-dimensional smooth manifold.

Then the sequence
0 ->&0(M >i>Q1
d
-> i>szmlm) -> 0
(where d is the exterior derivative) is a cochain complex of real vector spaces called the
DeRham complex. Its cosmology groups are the DeRham cosmology groups of the
manifold.

Example 7.1.10. Let (g, [., -]) be a real Lie algebra. Define
d : mg* -> Ak-I-19*,
by
(d(»Jl(X07 X I , 7 Xi) <-1>¢+9(»v({X@» xi], XG) ,X 7 in 7 7 Xi),
0<i<j<k
where, as usual, the hat"indicates a missing argument.
According to the computations in Example 3.2.9 the operator d is a coboundary op-
erator, so that (A°g*, d) is a cochain complex. Its cosmology is called the Lie algebra
cosmology, and it is denoted by H' (g).
 Cohomology 249

Exercise 7.1.11. (a) Let g be a real Lie algebra. Show that

H1(9) 2 (9/[9u9 D* 7
where [g,g] = span{ [X,Y] ; X , Y E g }.
(b) Compute H1 ( M , R)), where ii, R) denotes the Lie algebra of n X n real matrices
with the bracket given by the commutator.
(c) (Whitehead) Let g be a semisimple Lie algebra, i.e., its Killing pairing is nondegenerate.
Prove that H1 (g) = {0}- (Hint: Prove that [g, g]J_ = 0, where J_ denotes the orthogonal
complement with respect to the Killing pairing.)

Proposition 7.1.12. Let

dn- dn
(C'°,d): ->c"-1 § C" > Crn,-i-1 -> ,
be a cochin complex of R-modules. Assume moreover that C' i5 also a Z-graded R-
algebra, i.e., there exists an associative multiplication such that
C" - cm C C'M+"' 'v'm,n.
If d is a quasi-derivation, i.e.,

d(:z: y) = ::(d:u) • y :|: :IJ (by) Way Q c,

then H°(C') inherits a structure of Z-graded R-algebra.

A cochain complex as in the above proposition is called a deferential graded algebra

or DGA .

-
Proof. It suffices to show Z°(C') Z°(C') C Z°(C'), and B°(C') - B°(C') C B°(C').
If do: = dy = 0, then d(a:y) = ::(do:)y dz :z3(dy) = 0. If a: = do' and y = dy' then,
since d2 = 0, we deduce my = :=(d;z3Idy/).

Corollary 7.1.13. The DeRhom cosmology of smooth manifold has an R-algebra struc-
ture induced by the exterior multiplication o d y erentiolforms.

Definition 7.1.14. Let ( A ° , d) and ( B ° , 6) be two cochin complexes of R-modules.

(a) A cochin morphism, or morphism of cochin complexes is a degree 0 morphism <15
A' -> B' such that QUO o d = 6 o (IS, i.e., the diagram below is commutative for any n.
An dn ) An+1

(in ¢)n+l

Be 5"
> Bn-l-1

(b) Two cochain morphisms aw : A' -> B' are said to be cochin homotopic, and we
write this <2) up, if there exists a degree - 1 morphism X : {ATL -> Bn-1} such that
(Ma) MG) i in o x(al :|: X O
d(al.
250 Lectures on the Geometry of Manzfolds

(c) Two cochain complexes (A', d), and (B°, 6) are said to be homotopic, if there exist
cochain morphism

QUO A -> B and Tb: B ->A,

such that w O
<2) i v
lA, and <Z50¢';]18-

Example 7.1.15. The commutation rules in Subsection 3.2.1, namely [LX, d] = 0, and
[iX, d]S = Lx, show that for each vector field X on a smooth manifold M, the Lie deriva-
tive along X, Lx : Qs ( M ) -> Q ° ( M ) is a cochain morphism homotopic with the trivial
map (E 0). The interior derivative ix is the cochain homotopy achieving this.

Proposition 7.1.16. (a) Any cochin morphism <l> (A°, d) -> (B°, 5) induces a degree
zero morphism in cosmology

d H°(A) -> H°(13).

(b) If the enchain maps QS, Tb : A -> B are c o c h i n homotopie, then they induce identical
morphisms in chorology, QS* = Tb*.
(c) (]lA)* = ]1H°(A>» and If (AO, d0) <l> > (A1,dl) ¢> (AQ, d2) are c o c h i n morphisms,
then (in o ¢)* : up* o gzb*.

Proof. (H) It boils down to checking the inclusions

¢(Z"(A)) C zn(13) and ¢(B"(A)) C B"(8).

These follow immediately from the definition of a cochin map.

(b) We have to show that gb(cocycle) - 1b(cocycle) = coboundary. Let do 0. Then

qb(a) - , a l = d:6(X(a) :|: x(dal = 5(:l:X(al) = coboundary in B.

(c) Obvious.

Corollary 7.1.17. If Iwo coe fain complexes (A°, d) and ( B ' , 5) are cochin homotopic,
then their cosmology modules are isomorphic.

Proposition 7.1.18. Let

0 -> (A°,dA> > (82,B) w > (CMC') -> 0

be a short exact sequence of cochin complexes of R-modules. This means that we have a
commutative diagram
 Cohomology 251

DA dB do

An+1 ¢n+ 1 ) Bn-i-1 ¢n+

0 > 1) C171-l-1 > 0

DA dB do

ATL
¢n Be Ion
Cm
0 0 (7.l.l0)

DA dB do

> An-1 Bn-l

Ion 1 ) Cn 1
0 71bn-1) > 0

DA dB do

in which the rows are exact. Then there exists a long exact sequence

a
- - - - > H"-1<0) 631 H"(.4> s H"(1-3) is H"(c) is H"+1(A> -> (7.l.11)

We will not include a proof of this proposition. We believe this is one proof in homo-
logical algebra that the reader should try to produce on his/her own. We will just indicate
the construction of the connecting maps 3%. This construction, and in fact the entire proof,
relies on a simple technique called diagram chasing.
Start with at 6 H"(C). The cosmology class at can be represented by some cocycle
c 6 Z"(C'). Since 1% is surjective there exists b E B" such that C b). From
the commutativity of the diagram (7.l.l0) we deduce 0 = dC¢,,(b) w-1-1d8 (be, i.e.,
dB (be E her W,,+1 = range Cl5,;+1. In other words, there exists CL E An+1 such that
<25,+1 (a) = d B . We claim that a is a cocycle. Indeed,
A
¢N+2dn+1& dB
N I
Since g25n+2 is injective we deduce dA+1a = 0, i.e., a is a cocycle.
If we trace back the path which lead us from C E Z" (C) to a E
-
dB -*-1d8b 0

Zn-l-1 (A), we can write

a ( l o dB o 1 loc) ( l o d8b.

This description is not entirely precise since a depends on various choices. We let the reader
check that the correspondence Zn*(C) 9 C +-> a E Zn-l-1 (A) above induces a well defined
252 Lectures on the Geometry of Manzfolds

map in cosmology, 3n : H"(C) -> Hn+1(A) and moreover, the sequence (7.1.11) is
exact.

Exercise 7.1.19.1 Suppose R is a commutative ring with l. For any cochain complex
(K', do) of R-modules, and any integer n we denote by K[n]' the complex defined by
K[n]"" = K"+M, dK[n] = (-1)"dK. We associate to any cochain map f : K' -> L. two
new cochain complexes:
(a) The cone ( c ( f ) ° , d c ( f ) ) where

H 1
ki-i-1 Ki-I-l
-do 0
c(fl' Z K[l]° ® L. 7 dc(f) Zi
] f dL
i

(b) The cylinder ( c l y u f ) , d c y u f )

Ki Ki

1 H1
do -:l]_K»i+1 0
iv Ko -do ki-I-1
0 0 19+1
C'yl(f)° ® 0(f)° 7 dc'yl(f)
i go dL
f 0
We have canonical inclusions t
L. -> C y l ( f ) , : K' -> C'yl(f)', a canonical projec-
sons B : C'yl(f)° -> L', 6 = (5(f):0(f)' -> K[l]°,and7r Cyi(f) -> 0W(f).

(i) Prove that a, Bf-, 6(f) are cochain maps, B o a = 11 L and a o B is cochain homotopic to
11cyI(f)-
(ii) Show that we have the following commutative diagram of cochain complexes, where
the rows are exact.
5(f) >
> L.
7T
0 > C(f)' K[1]' > 0

a ]lC(f)

0 > K. 'ii
> 0
f > Cal(fl C(f)

11 K B

K. f L.

(iii) Show that the connecting morphism in the long exact sequence corresponding to the
short exact sequence

0-> KO f > C a l ( f l ">c*(f)->0

coincides with the morphism induced in cosmology by 6(f) : C ( f ) -> K111..
(iv) Prove that f induces an isomorphism in cosmology if and only if the cone of f is
acyclic.
lThis exercise describes additional features of the long exact sequence in cohomology. They are particularly
useful in the study of derived categories.
 Cohomology 253

Exercise 7.1.20 (Abstract Morse inequalities). Let C' = QQH20 C" be a cochain com-
plex of vector spaces over the field IF. Assume each of the vector spaces C" is finite
dimensional. Form the Poincaré series
7

and
PH.(c) (t) TN dimly Hnlgl.
n>0
Prove that there exists a formal series R(t) E ZW] with non-negative coefficients such
that
pc- (t) = PH-(c)l'5) + (1 + t)R(t).
In particular, whenever it makes sense, the graded spaces C* and H* have identical Euler
characteristics
x(C") Pa(-ll = PH- (ml - l ) = x ( H ' ( 0 l ) -
Exercise 7.1.21 (Additivity of Euler characteristic). Let
0->A'->B'->C'->0
be a short exact sequence of cochain complexes of vector spaces over the field IF. Prove
that if at least two of the cosmology modules H'(A), H°(B) and H'(C') have finite
dimension over IF, then the same is true about the third one, and moreover
x(H°(B)) = x( H' (All x(H°((1)).

Exercise 7.1.22 (Finite dimensional Hodge theory). Let

/ \
iv: 69 v", d..
,
..

d)
II

/ 7120

be a cochin complex of real vector spaces such that dim V" < oo, for all n. Assume that
each V" is equipped with a Euclidean metric, and denote by dTL : Vn-I-1 -> V" the adjoint
of dn. We can now form the Laplacians
An . V" -> V", An
(a) Prove that &B,,,20An = (d -|- d*)2.
(b) Prove that A'no = 0 if and only if dna = 0 and 0. In particular, her A'n C
Z"(V'l~
(c) Let C 6 Zn*(V). Prove that there exists a unique E 6 Z"(V) cohomologous to C such
that
III = min{ la/l; C - c' 6 Bn (C)},
-
where | | denotes the Euclidean norm in V" .
(d) Prove that E determined in part (c) satisfies An? = 0. Deduce from all the above that
the natural map
her ATL -> H"lv°l
is a linear isomorphism.
254 Lectures on the Geometry of Manzfolds

Exercise 7.1.23. Let V be a finite dimensional real vector space, and UoGV. Define
»
do, = d g ) : Arv -> Ak:+1V o r > u 0 A w .
(a) Prove that

Aka do Ak-I-1V
do

is a cochain complex. (It is known as the Koszul complex.)

(b) Use the Cartan identity in Exercise 2.2.55 and the finite dimensional Hodge theory
described in previous exercise to prove that the Koszul complex is acyclic if U0 go 0, i.e.,
H'"(A'v, d(00l) = 0, We _ 0.
7.1.4 Funetorialproperties of the DeRham eohomology

Let M and N be two smooth manifolds. For any smooth map QS : M -> N the pullback
(l>* Q' Q' (N) -> (M)
is a cochain morphism, i.e., d)*dn = drab*, where do and respectively dn denote the
exterior derivative on M, and respectively N . Thus, (l)* induces a morphism in cohomology
which we continue to denote by d)*
H.
co*(N) -> H'(M).
In fact, we have a more precise statement.

Proposition 7.1.24. The DeRhom cohomology construction i5 a controvoriontfunctorfrom

the category of smooth momfolds and smooth maps to the category of Z-graded vector
spaces with degree zero morphisms.

Note that the pullback is an algebra morphism QS* Q. ( N ) -> Q' I M ) ,

¢*(@A@) Z (¢*00 A (¢*e), 'v'a,B 6 Q' IN,
and the exterior differentiation is a quasi-derivation, so that the map it induces in cohomol-
ogy will also be a ring morphism.

Definition 7.1.25. (a) Two smooth maps Q50, Q51 : M -> N are said to be (smoothly)
homotopic, and we write this Q50 8h Q51, if there exists a smooth map
N

@:I X M -> N (t,m) »-> ®t(ml,

such that is = QUO-, form = 0, 1.
(b) A smooth map (15 : M -> N is said to be a (smooth) homotopy equivalence if there
exists a smooth map uP : N -> M such that <15 o uP Hn, and o QUO
FiSh He. w FiSh

(c) Two smooth manifolds M and N are said to be homotopy equivalent if there exists a
homotopy equivalence QUO : M -> N .

Proposition 7.1.26. Let <250, <251 : M -> N be two homotopic smooth maps. Then they
induce identical maps in cohomology
(25Z3=(z5i' H' (N)-> H' (Ml-
 Cohomology 255

Proof. According to the general results in homological algebra, it suffices to show that the
pullbacks
<¢50»<251 : Q'(N) -> Q'(M)
are cochain homotopic. Thus, we have to produce a map
X : Q ' ( N ) -> Q*°-1(Ml,
such that
w) - Colwl ix(d(»J) i dew, Vi 6 o'(m>.
I

At this point, our discussion on the fibered calculus of Subsection 3.4.5 will pay off.
The projection <I> : I X M -> M defines an oriented 8-bundle with standard fiber I .
For any w 6 Q ' ( N ) we have the equality

Cbllw) - (l50(Wl = 'I>*((»Jl llxM -'1>*((»J) l0xM= '1>*((»Jl-

(8IxM)/M
We now use the homotopy formula in Theorem 3.4.54 of Subsection 3.4.5, and we deduce

(alxM)/M
@* lull
/(IXM)/M
dIxm@*lWl - aM
(IXM)/M
@*oWl

Thus
/(IXM)/M
@*ldNL4J) - dM
(IXM)/M
@*oWl.

x(wl <I>*(,,)
(IXM)/M
is the sought for cochain homotopy.

Corollary 7.1.27. Two homotopy equivalent spaces have isomorphic cosmology rings.

Consider a smooth manifold M, and U, V two open subsets of M such that M = UUV.
Denote by nU (respectively W) the inclusions U <-> M (resp. V <-> M). These induce the
restriction maps
. Q .(M )_> w»_> w
nU
•
QUO
(U)
'
IU'
and
Q'(M) ->
QUO
iv), w e - > w l v
We get a cochin morphism
QUO
(M) -> Q'lU) ® QUO
(V), w +-> (zuw,zvw).
There exists another cochin morphism
6 : we) ® m y ) -> Q' (Um Vl» (we) +-> -¢*JIUnv +'7IUrwv

Lemma 7.1.28. The short Mayer-Vietoris sequence

0 -> Q°(m) '">Q'(U) ® Q'(v> 6 >Q*(UVWV)-> 0
is exact.
256 Lectures on the Geometry of Manzfolds

Proof. Obviously 7° is injective. The proof of the equality Range 7° = her 6 can be safely
left to the reader. The surjectivity of 6 requires a little more effort.
The collection {U, V} is an open cover of M, so we can find a partition of unity
{(,oU, (,CV} C COO(M) subordinated to this cover, i.e.,
supp(/)u C U, supp(/)v C ii 0 S ( p u , (PV S l, (PU +l0v = 1.
Note that for any w 6 Q*(U n V) we have
supp (Pvw C supp (PV C v,
and thus, upon extending (PVLU by 0 outside V, we can view it as a form on U. Similarly,
(PULU 6 Q*(V). Note that

6(-(0vw, (www) = ( i v + (,f)u)°J w.

This establishes the surjectivity of 6.

Using the abstract results in homological algebra we deduce from the above lemma the
following fundamental result.

Theorem 7.1.29 (Mayer-Vietoris). Let M = U U V be an open cover of the smooth

manifold M. Then there exists a long exact sequence
--- -> H'<=lml -> H"'(U) ® H W ) -> Hk=(U n vI -> Hk+1(m) ->
*r' 6 3
,
called the long Mayer-Vietoris sequence.

The connecting morphisms 3 can be explicitly described using the prescriptions fol-
lowing Proposition 7.1.18 in the previous subsection. Let us recall that construction.
Start with CU 6 Q"(U D V) such that do = 0. Writing as before
cu = (Pvw + (PUT,
we deduce
d(govw) = d(-(pgw) on U D V.
Thus, we can find 17 6 Qk+1(M) such that
tllu= dWvw) '7lv= d(-¢UWl-
Then
k w : T).
The reader can prove directly that the above definition is independent of the various choices.
The Mayer-Vietoris sequence has the following factorial property.
Proposition 7.1.30. Let QUO : M -> N be a smooth map and {U, V } an open cover of N .
Then U/ = Q5-1(U)7 V' = cl5-1(V) form an open cover o f M and moreover the diagram
below is commutative.
Hk(nl > Hk(U) ® H'°(v)
.°
6
>HUNV)
6' He+llnl
d>* d>* ¢* ¢>*

,my 6/ 3/
H'<=(M) > HW) ® HW) > H'"<U' re v') > Hk+1lml
 Cohomology 257

Exercise 7.1.31. Prove the above proposition.

7.1.5 Some simple examples

The Mayer-Vietoris theorem established in the previous subsection is a very powerful tool
for computing the cohomology of manifolds. In principle, it allows one to recover the
cohomology of a manifold decomposed into simpler parts, knowing the cohomologies of
its constituents. In this subsection we will illustrate this principle on some simple examples.

Example 7.1.32 (The cosmology of spheres). The cosmology of S1 can be easily

computed using the definition of DeRham cosmology. We have H0(S1) = R since S1 is
connected. A 1-form 17 6 Q1 (51) is automatically closed, and it is exact if and only if

T) 0.
S1

Indeed, if 17 dF, where F : R -> R is a smooth 21r-periodic function, then

77=F(27r)-F(0)=0.
S1

Conversely if 17 f (0)d0, where f : R -> R is a smooth 27r-periodic function and

27r
f (0)d0 0,
0

then the function

F(t) f (s)ds
0

is smooth, 21r-periodic, and dF 17.

Thus, the map

:Q1<s1>->R, 17+-> 17,

S1 $1

induces an isomorphism H1 (51) ->R, and we deduce

pS1(t)= 1 +

To compute the cosmology of higher dimensional spheres we use the Mayer-Vietoris

theorem.
The (n -|- 1)-dimensional sphere Sn-l-1 can be covered by two open sets
USOuth = Sn-I-1 \ {North pole} and North = Sn-I-1 \ {South pole}.
Each is diffeomorphic to lRTL+1. Note that the overlap N o r t h n USOuth is homotopically
equivalent with the equator S". The Poincaré lemma implies that
Hk+1lUnorthl ® Hk+1(Usouthl
f\J
0
for k > 0. The Mayer-Vietoris sequence gives
HklUnorthl ® Hklljsouth) Hklljno'rth VW Usouthl Hk+1lsn+1l 0.
258 Lectures on the Geometry of Manzfolds

For k > 0 the group on the left is also trivial, SO that we have the isomorphisms
H'"(s") 2 H'"(UnO,t,, re USOuthl 2 H'<:+1(s"+1) If > 0.
Denote by Pn (t) the Poincare polynomial of S" and set Qn (t) = Pn (t) - Pn (0) Pm (t)
l. We can rewrite the above equality as
Qn+1(t) t o M ) n > 0.
Since Q1(t) = t we deduce Qn1(t) = tn, i.e.,
TN
Ps" (t) l

Example 7.1.33. Let {U, V} be an open cover of the smooth manifold M. We assume that
all the Betti numbers of U, V and U D V are finite. Using the Mayer-Vietoris short exact
sequence, and the Exercise 7.1.21 in Subsection 7. 1.3, we deduce that all the Betti numbers
of M are finite, and moreover
x(ml = x(U) -|- x(V) - x ( u O Vl- (7, 1.12)
This resembles very much the classical inclusion-exclusion principle in combinatorics. We
will use this simple observation to prove that the Betti numbers of a connected sum of g
tori is finite, and then compute its Euler characteristic.
Let E be a surface with finite Betti numbers. From the decomposition
2 (2 \ disk) LJ disk,
we deduce (using again Exercise 7.1.21)
X(2) = X(2 \ disk) + x(disk) - X ( (2 \ disk) O (disk) ) .
Since (E \ disk) al disk is homotopic to a circle, and x(disk) = l, we deduce
X(z) = x(2 \ disk) + 1 - X(s1) = x(2 \ disk) + 1.
If now 21 and 22 are two surfaces with finite Betti numbers, then
21#E2 = (21 \disk) U (22 \disk),
where the two holed surfaces intersect over an entire annulus, which is homotopically a
circle. Thus
xlE1#E2l Xl21 \ disks -|- Xl22 \ disks xIS1)
X(21l + Xl22l - 2.
This equality is identical with the one proved in Proposition 4.2.42 of Subsection 4.2.5.
We can decompose a torus as a union of two cylinders. The intersection of these cylin-
ders is the disjoint union of two annuli so homotopically, this overlap is a disjoint union of
two circles. In particular, the Euler characteristic of the intersection is zero. Hence
x(torus) = 2x(cylinder) = 2x(circle) = 0.
We conclude as in Proposition 4.2.42 that
x(connec:ted sum of g tori) = 2 - 2g.
This is a pleasant, surprising connection with the Gauss-Bonnet theorem. And the story is
not over.
 Cohomology 259

7.1.6 The Mayer-Vetoris principle

We describe in this subsection a "patching" technique which is extremely versatile in estab-
lishing general homological results about arbitrary manifolds building up from elementary
ones.

Definition 7.1.34. A smooth manifold M is said to be ofjinite type if it can be covered by

finitely many open sets $1, . . . , Um such that any nonempty intersection U O ---
D Upk
(k > l ) is diffeomorphic to Rdim M . Such a cover is said to be a good cover.

Example 7.1.35. (a) All compact manifolds are of finite type. To see this, it suffices to
cover such a manifold by finitely many open sets which are geodesically convex with re-
spect to some Riemann metric.
(b) If M is a finite type manifold, and U C M is a closed subset homeomorphic with
the closed unit ball in ] R d i m m , then M \ U is a finite type non-compact manifold. (It
suffices to see that R" closed ball is of finite type.)
(c) The connected sums, and the direct products of finite type manifolds are finite type
manifolds.

Proposition 7.1.36. Let p : E -> B be a smooth vector bundle. If the base B is ofjnite
type, then so is the total space E.

In the proof of this proposition we will use the following fundamental result.

Lemma 7.1.37. Let p : E -> B be a smooth vector bundle such that B is dyeomorphic to
Rn. Then p : E -> B is o trivializable bundle.

Proof of Proposition 7.1.36. Denote by F the standard fiber of E. The fiber F is a

vector space. Let (Uz~)19$1/ be a good cover of B. For each ordered multi-index I :=
{it < -- - ---
< ik} denote by UI the multiple overlap U m (W Uik. Using the previous
lemma we deduce that each E, E IU1 is a product F X Up, and thus it is diffeomorphic
with some vector space. Hence (ED is a good cover.

Exercise 7.1.38. Prove Lemma 7.1.37.

Hint: Assume that E is a vector bundle over the unit open ball B C R". Fix a connection
V on E, and then use the V-parallel transport along the half-lines L , [0,oo) 9 t +-> to E
R", ac 6 R" \ {0}.

@»We denote by M n the category _finite type smooth manifolds of dimension n. The
morphisms of this category are the smooth embeddings, i.e., the one-to-one immersions
M1 <--> MQ, Mi E Mn.

Definition 7.1.39. Let R be a commutative ring with l. A contravariant Mayer-Wetoris

functor (or M V-functor for brevity) is a contravariant functor from the category Mn, to
260 Lectures on the Geometry of Manzfolds

the category of Z-graded R-modules

3' ®n€Z3 N * GradRl\/Iod 7 M»-> 9'"(M),
n

with the following property. If {U, V } is a MV-cover of M 6 Mn, i.e., U, V, U re V 6

Mn, then there exist morphisms of R-modules
an : 9*'"(U re v) -> 3rrz-I-1(m),

such that the sequence below is exact

r*
a
- > 3 r l M ) -> 9""(U) ® m y ) i> 5t""(Urwv) s 3rH+1lml -> 7

,r,*
where is defined by
T*
5"('¢u) ® 5"('¢vl»
and 6 is defined by
5(a2 ® y) = 5"(2U0vl(3/I - 5"('¢Ur1vl(tl-
(The maps z, denote natural embeddings.) Moreover, if N 6 Mn is an open submanifold
of N , and {U, V } is an M V - c o v e r o f M such that {Uri N , V O N } is an MV-cover of N,
then the diagram below is commutative.
an gw+1lml
3'"lumvl )

3'"(umvmnl an > 9I'17~+1 TNT

The vertical arrows are the morphisms 5"(z,) induced by inclusions.

The covariant MV-functors are defined in the dual way, by reversing the orientation of
all the arrows in the above definition.

Definition 7.1.40. Let S", 9 be two contravariant MV-functors,

f } ' , § : M n - > Grad _R1\/Iod.
A Col'l'€spol'ld€l'lc€2 between these functors is a collection of R-module morphisms
(25m ® (We =®
TLGZ
> 9"(m),

one morphism for each M Q Mn, such that, for any embedding M1 <1 MQ, the diagram
below is commutative
( )
9'"(M2) 5°"<f> > 3""(M1)

¢>m2 ( l 7

9((f))
9" (MQ) > 9" (Ml )
2 Our notion of correspondence corresponds to the categorical notion of morphism of functors.
 Cohomology 261

and, forany M e n T L and any MV-cover {U, V} of M, the diagram below is commuta-
five.
a 3r%+1lMl
g'"lumvl >

¢Urwv d>m

3 ) in-I -lljw l
QnIUVWV)

The correspondence is said to be a natural equivalence if all the morphisms <l>m are iso-
morphisms.

Theorem 7.1.41 (Mayer-Vietoris principle). Let S", 9 be two (contravariant) Mayer-

Wetoris functors on Mn and (15 : 5' -> 9 a correspondence. If
(/)§ : @t'"(R") -> 9k(R")
i5 an isomorphism for any k 6 Z, then cl) i5 a natural equivalence.

Proof. The family of finite type manifolds Mn has a natural filtration

Ml CM2 c - - - c M Q c - - - ,
where M; is the collection of all smooth manifolds which admit a good cover consisting
of at most 7" open sets. We will prove the theorem using an induction over r.
The theorem is clearly true for r = l by hypothesis. Assume ¢'; is an isomorphism
for all M 6 m;-1. Let M 6 M; and consider a good cover {U1, . . . , U,.} of M. Then
{ U = U1 u---LJU,-_1,U,,-}
is an MV-cover of M. We thus get a commutative diagram

5"'(u) ® 2""(u~) > £¥n*(Urw U,.)

8
> 9:'l1-I-lljwl > N-I-IIUI @ 912-I-I(U)

a
9"(U) ® 9"(uI) \
9"(U n U,.) 97.-I-1(jwl \ 5""+(u) ® 9%-I-IIUI
The vertical arrows are defined by the correspondence go. Note the inductive assumption
implies that in the above infinite sequence only the morphisms (ISm may not be isomor-
phisms. At this point we invoke the following technical result.

Lemma 7.1.42 (The five lemma). Consider the following commutative diagram of R-
modules.
AS > AS > AS A1 A2
f 2

B_2
1 f 1

> B_ 1
1 >
to
B0
1 fl

B1
f2

B2
If f is an isomorphismfor any i ¢ 0, then SO is f0.
262 Lectures on the Geometry of Manzfolds

Exercise 7.1.43. Prove the five lemma.

The five lemma applied to our situation shows that the morphisms ( z ) m must be isomor-
phisms.

Remark 7.1.44. (a) The Mayer-Vietoris principle is true for covariant MV-functors as
well. The proof is obtained by reversing the orientation of the horizontal arrows in the
above proof.
(b) The Mayer-Vietoris principle can be refined a little bit. Assume that 5* and 9 are
functors from Mn to the category of Z-graded R-algebras, and go : 3' -> 9 is a correspon-
dence compatible with the multiplicative structures, i.e., each of the R-module morphisms
<l>m are in fact morphisms of R-algebras. Then, if (be are isomorphisms of Z-graded
R-algebras, then so are the QSm 's, for any M 6 Mn .
(c) Assume R is a field. The proof of the Mayer-Vietoris principle shows that if 5* is a
MV-functor and diIIlR 5** (Rn) < oo then dim 5**(M) < oo for all M 6 Mn.
(d) The Mayer-Vietoris principle is a baby case of the very general technique in alge-
braic topology called the acyclic models principle.

Corollary 7.1.45. Any finite type manifold has #nite etti numbers.

7.1.7 The Kenneth formula

We learned in principle how to compute the cosmology of a "union of manifolds". We
will now use the Mayer-Vietoris principle to compute the cosmology of products of man-
ifolds.

Theorem 7.1.46 (Kenneth formula). Let M 6 M m and N E Mn. Then there exists a
natural isomorphism of graded R-algebras

H. H.
(M X N)
f

( M ) ® H* ( n ) 6969
7120 p-I-q=n
H'"(Ml (X) HQIN)

In particular, we deduce
ProN(¢) = PMT . PW).

Proof. We construct two functors

529 Mm -> Grad R All,

5':M+->
'r">0
3""'l(Ml 6969
'r>0 p-l-q=r
H'"(M) (X) H"(N) 7

and
§:M+-> 9"(M) H" ( M x N ) ,
'r'>0 'r'>0
 Cohomology 263

where

5I'(f)
f* IHp(M2) ®]1H<1(n) 7 W' I M1 --> 27
p+q or

and

9(f) (f X ]iN)* IH'*(m2 xn)V W' : M1 "-> M2-

'r">0

We let the reader check the following elementary fact.

Exercise 7.1.47. 5* and 9 are contravariant MV-functors.

For M 6 Mm, define ¢M : st(m) -> 9(n) by

H.
( b e ( w ® t I ) = ( » J X 17 WmWAWV (USG H'(m), T) 6 (MU,
where arm (respectively W N ) are the canonical projections M X N -> M (respectively
M X N -> N). The operation
X : H°(M) (X) H ' ( N ) -> H' (M X N) ((»J®t7)*->¢~vx17,
is called the cross product. The Kenneth formula is a consequence of the following lemma.

Lemma 7.1.48. (a) go is a correspondence of MV-functors.

(19) gz5Rm is an isomorphism.

Proof. The only nontrivial thing to prove is that for any MV-cover {U, V} of M @Mm
the diagram below is commutative.
3
®p+q=THPlU n V ) ® H<*(n) > ®p+q=rHP+1lMl (X) Hq(N)

¢umv do
3/
HT'<<u X n)rw(v X N)) > H'-+1(m X
N)

We briefly recall the construction of the connecting morphisms <9 and <'3I.
One considers a partition of unity {cpu, gov} subordinated to the cover {U, V}. Then,
the functions ¢U = 7VM 90U and iv = 7rm(pv form a partition of unity subordinated to the
cover
{u X n,v X N
ofM X N.Ifw (X) N e H'(U re) (X) H. (no,

w®M=@ (X) W
where

(8 IU= -d((Pv°Jl 69 = d((0U°J)-

On the other hand, ¢ u m v ( w ® u l = w x u 7 and 3I(wx17)=cD X T). This proves (a)-
264 Lectures on the Geometry of Manzfolds

To establish (b), note that the inclusion

] : N - - > ] R " " x N , 513+->(0,x)

is a homotopy equivalence, with iN a homotopy inverse. Hence, by the homotopy invari-

ance of the DeRham cosmology we deduce

Using the Poincaré lemma and the above isomorphism we can identify the morphism qz5Rm

Example 7.1.49. Consider the n-dimensional torus, T n _ By writing it as a direct product

of n circles we deduce from Kenneth formula that

PT. (t) {P51(to}" (1 + TY".

Thus
n
belTn)
()k
7 dim H*(T"") = 2" 7

and X(T") = 0.
One can easily describe a basis of H*(T"). Choose angular coordinates (01, . . . , gel
on T". For each ordered multi-index I = (l S 'ii < ---
< 'in < n) we have a closed, non-
exact form deI. These monomials are linearly independent (over R) and there are 2" of
them. Thus, they form a basis of H* (T'n). In fact, one can read the multiplicative structure
using this basis. We have an isomorphism of R-algebras
H. A'lR"'.
(1/"") L

Exercise 7.1.50. Let M G M , and N 6 Mn- Show that for any We G H* (ML to E
H*(nl7 i 7 j = 0 , 1, the following equality holds.
deg
(We X Dol/\(W1 X '71)= l-1ldeg770
0J1
(We Cell) X 070 771)-

Exercise 7.1.51 (Leray-Hirsch). Let p E -> M be smooth bundle with standard fiber
F. We assume the following:

(a) Both M and F are of finite type.

(b) There exist cosmology classes 61, . . . 7 e 6 H. ( E ) such that their restrictions to any
fiber generate the cosmology algebra of that fiber.

The projection p induces a H' (M)-module structure on H' (E) by

w T):p*wA17, Vwé
H. (M), T16 H. (E).

Show that H' (E) is a free H' (M)-module with generators 617 ..,e,~.
 Cohomology 265

7.2 The Poincaré duality

7.2.1 Co ho mo logy with eompaet supports

Let M be a smooth n-dimensional manifold. Denote by Qkptlml the space of smooth

compactly supported k-forms. Then
d d
0 -> Q 0 p t ( M l -> -> Qgpt(m) -> 0
is a cochain complex. Its cohomology is denoted by H$pt(m), and it is called the DeR-
ham cosmology with compact supports. Note that when M is compact this cosmology
coincides with the usual DeRham cohomology.
Although it looks very similar to the usual DeRham cohomology, there are many im-
portant differences. The most visible one is that if QUO : M -> N is a smooth map, and
w 6 Q;pt(N), then the pull-back qz5*w may not have compact support, so this new con-
struction is no longer a contravariant functor from the category of smooth manifolds and
smooth maps, to the category of graded vector spaces.
On the other hand, if dim M = dim N, and go is an embedding, we can identify M with
an open subset of N, and then any 17 6 §2e'pt(M) can be extended by 0 outside M C N.
This extension by zero defines a push-forward map
co* I Q;ptlM) -> Q;ptlN)-
One can verify easily that ¢* is a cochain map so that it induces a morphism
QS* : H;ptlm) -> Hc.ptlNl°
In terms of our category Mn we see that Hgpt is a covariantfunetor from the category Mn,
to the category of graded real vector spaces. As we will see, it is a rather nice functor.

Theorem 7.2.1. Hcpt

. is a covariant MV-functon and moreover
07 k<n
Hkpt (Rn ) R k n7

The last assertion of this theorem is usually called the Poincaré lemma for compact
supports.
We first prove the Poincaré lemma for compact supports. The crucial step is the follow-
ing technical result that we borrowed from [14].

Lemma 7.2.2. Let E 1 B be a rank T' real vector bundle which i5 orientable in the sense
described in Subsection 3.4. 5. Denote by p* the integration-along-fibers map
P* I Qbpt (El -> Q;ptlBl'
Then there exists a smooth bilinear map
in : Qpt(E) X Q"pt(E)-> 7,-I-j-T"-1(E)
Qept

such that,
p*p*o¢ A /3 - a A pa I (-1)"d(m(@~, in) - m(da, f ) + (-1)deg"m(@, do.
266 Lectures on the Geometry of Manzfolds

Proof. Consider the 3-bundle

7r :8 I I X (E ® E) -> E, or : (t;v0,v1) »-> (t;u0 -I-t('u1 - v0)).

Note that
38=({0}x(E®E)) LI ({l} X (ElBE)).
Define 'IrTt E ® E -> E a s the composition
E®E%{t}XE®E<--> I X (E®E)l>E.
Observe that

3% = tflaa= (-w0) U Tr1.

For (Q, B) 6 Qgpt X Q;ptlEl define a ® B 6 Q;pt(E®E) by

QQB Q; (#")*@ A (We.

(Verify that the support of a GM? is indeed compact.) For a G QptlEl' and /3 6 QJptlEl
we have the equalities

p*p*(1 /\ B= 7r1(a G) B) G Q;;v(E), a I\p*p*!3 = tr0(a G) B) 6 Q;;v(E).

Hence

D(o¢,B)=1>*p*o¢/\!5'- a Ap*p*@=w1(<1®/3)-w"(@®6) Eo®¢2.

/
We now use the fibered Stokes formula to get

D(@,e) Ci8']"'*(O®5) + (-1)ldE I*(a®/3),

a/E E/E
where 'T is the natural projection 8 =I X (E ® E) -> E ® E. The lemma holds with

in low, B) 'J~*(0¢®5).
5/E

Proof of the Poincaré lemma for compact supports. Consider 6 6 05° (Rn) such that

0<6<1 7 5(a2)da: l.
Rn

Define the, closed, compactly supported n-form

7' := 5(;uld;u1 /\ - - - /\ do".
We want to use Lemma 7.2.2 in which E is the rank n bundle over a point, i.e., E
p
{pt} o R " - > {pt}. The integration along fibers is simply the integration map.
" R CU »
-> 0 7 degw<n
P* . " o R ) - >
O
QU
7 P*w
Rn w = degw=n.
If now Ld is a closed, compactly supported form on R", we have
w (12*p*T) /\ w.
 Cohomology 267

Using Lemma 7.2.2 we deduce

w 7' /\ P*P*w = (-l)"dm(T,w).

Thus any closed, compactly supported form Ld on R" is cohomologous to T A p*p*w. The
latter is always zero if deg < n. When deg = n we deduce that w is cohomologous to
(f w)7'. This completes the proof of the Poincaré lemma.

To finish the proof of Theorem 7.2.1 we must construct a Mayer-Vietoris sequence.

Let M be a smooth manifold decomposed as an union of two open sets M = U U V, The
sequence of inclusions

U O V <-> U,V <-> M ,

induces a short sequence

0 Q;ptlu VW V) 1> Q;ptlul ® Qgptlv) 1> Q;ptlml 0:

where

i(w)=(cD,cD), j(w,17) f? co.

A

The hat denotes the extension by zero outside the support. This sequence is called the
Mayer-Vietoris short sequence for compact supports.

Lemma 7.2.3. The above Mayer-Vietoris sequence is exact.

Proof. The morphism i is obviously injective. Clearly, Range (in ker(j). We have to
prove that j is surjective.
Let (toy, gov) a partition of unity subordinated to the cover {U, V}. Then, for any
T] 6 Q§pt(M), we have

SOU77 G Q;ptlul and (PVD G Qgptl/I'

In particular, 77 = j(-(pun, (pa), which shows that .j is surjective.

We get a long exact sequence called the long Mayer-Vietoris sequence for compact
supports.

-> H';t(U nv) -> H ' ; ( U ) ® H'¢,(v> -> H k p t l m l i> H'§;;1 ->
The connecting homomorphism can be explicitly described as follows. If w E Qkpt(M) is
a closed form then

dlS0 U(»J ) = ii(-(pvw) on U re VL

We set do := d((pUw). The reader can check immediately that the cosmology class of do
is independent of all the choices made.
268 Lectures on the Geometry of Manzfolds

If QUO : N <-> M is a morphism of Mn then for any MV-cover of {U, V} of M

{¢-1(U),¢-1(V)} is an MV-cover of N. Moreover, we almost tautologically get a com-
mutative diagram

Hkptlnl 5 > H'3;1((l>(umv))

¢*
l ,l
eptlm 6
d>*
l re v)
HC H';;;1(U

This proves Hgpt is a covariant Mayer-Vietoris sequence.

Remark 7.2.4. To be perfectly honest (from a categorical point of view) we should have
considered the chain complex

@Q eptk 7

k<o go

and correspondingly the associated homology

AV

. kept'

This makes the sequence

0 4-Q"* <- +-Q-1<-Q0<-0

a chain complex, and its homology E, is a bona-fide covariant Mayer-Vietoris functor

since the connecting morphism 6 goes in the right direction E, -> PI.-1. However, the
simplicity of the original notation is worth the small formal ambiguity, so we stick to our
upper indices.

From the proof of the Mayer-Vietoris principle we deduce the following.

Corollary 7.2.5. For any M G Mn, and any k n we have dim Hkpt(m) < oo.

7.2.2 The Poincaré duality

Definition 7.2.6. Denote by M+ the category of n-dimensional, finite type, oriented man-
ifolds. The morphisms are the embeddings of such manifold. The MV functors on M+
are defined exactly as for M'n .

Given M U M , there is a natural pairing

..
, )=Qk(m)xQ;;,J(m)4R,
defined by

(wpM w /\ 77-
M
 Cohomology 269

This pairing is called the Kronecker pairing. We can extend this pairing to any ((»J»v) G
Q' X QUOpt as
0, deg -I-degn go n
(¢LJJ1) deg -I-degn : n
few As,
The Kronecker pairing induces maps
'D I
'Dk : W M ) -> lQgptk(m))*, <qp(wm>
..
I (W)77)f<.

Above, ( ) denotes the natural pairing between a vector space V and its dual V*,
..
7

( 7 ):V* X V -> R.
If CUis closed, the restriction of QD(w) to the space Zenptk of closed, compactly supported
n - la)-forms vanishes on the subspace Bcnptk = doTgp/H ( m ) . Indeed, if 77 = d17',
,all then
6 Qgptk-llMl
S k
(iD(w)» '7) cu /\ dn' i i dwA17'=0.
M M

Thus, if LU is closed, the linear functional QD(w) defines an element of (H"ptk(M))*- If

moreover w is exact, a computation as above shows that D(w) 0 6 (H",J(m))*.
Hence QD descends to a map in cosmology

QD : -> ( H / " ( M l ) *
Equivalently, this means that the Kronecker pairing descends to a pairing in cosmology,

( ..M
7
H"'(m) X Hept '"(m) -> R.
Theorem 7.2.7 (Poincaré duality). The Kronecker pairing in cosmology is a dualilyfor
all M E M+.

Proof. The functor M+ -> Graded Vector Spaces defined by

M-> Q8'P1'"<mI lHenptklMl )*
k k

is a contravariant MV-functor. (The exactness of the Mayer-Vietoris sequence is pre-

served by transposition. This is where the fact that all the cosmology groups are finite
dimensional vector spaces plays a very important role.)
For purely formal reasons which will become apparent in a little while, we re-define
the connecting morphism

FMU re v) i Hk-I-1(U U v),

_ fra-
to be (-1)/<=5+, where Hcpt k 1(Uov)
5 'ra-
-> Hept k(Uov) denotes the connecting morphism
in the DeRham cohomology with compact supports, and 6+ denotes its transpose.
The Poincaré lemma for compact supports can be rephrased
R, k 0
Pfklnaqnl I

0, k>0 '
270 Lectures on the Geometry of Manzfolds

The Kronecker pairing induces linear maps

DM : H'"(m) -> Hk(ml.

Lemma 7.2.8. €9k Dk i5 a correspondence of MV functors.

Proof. We have to check two facts.

Fact A. Let mi N be a morphism in M+. Then the diagram below is commutative.
¢>*
HWy) > H/"(M)
I I
Dn
g,*
PIN) PPM)

Fact B. If {U, V } is an MV-cover of M 6 M+, then the diagram bellow is commutative

a He+llml
Hkwrw)
®Urwv
l (-1)'°6*
Pm

Hk+1lml
l
HUOV) >

..
Proof of Fact A. Let w 6 H"'(N). Denoting by ( » ) the natural duality between a vector
space and its dual we deduce that for any u € HZL/°(M) we have

<¢* o iln(w)»'7) = (l¢*l iDn(W,0?7) = (Dn((»J), (M )

P p

1 /\ do? w I M I\'l7 ¢*(»J /\ 77 = (iD(¢*(»J)v u)-

'n 'm<->n M

Hence €2*§*0iDn =iDmOQ5*-

Proof of Fact B. Let (PUNPV be a partition of unity subordinated to the MV-cover {U, V}
of M G M+_ Consider a closed k-form w e Qk(U VW V). Then the connecting morphism
in usual DeRham cosmology acts as
d(-goyw) on U
<9w
d((puw) on V
n-k-1
Choose 17 G Qept such that dr) = 0. We have

<®m3W, u)
/ M
3wA'7
/
U
3wA17+
/
V
3wA17
/
Unv
3wA17

/ U
d((/)v(»J) /\ U +
/ V
d(/)u(»J) /\ 77 -|-
/
Unv
CZISUvUJ) /\ Tl-

Note that the first two integrals vanish. Indeed, over U we have the equality

(d((/)v(»J)) A U = d ( ¢ v W A V ) 7
 Cohomology 271

and the vanishing now follows from Stokes formula. The second term is dealt with in a
similar fashion. As for the last term, we have

UNV
d(wv(»J) /\ 77
UNV
do /\ w /\ 77 = (-1)deg'"
Unv
cu /\ (do /\ al
= (-u w /\ 677 = (-1>'"<;1>U5V,J, 5'7> = <<-1>'5*a>UOV, 17)»
Unv
This concludes the proof of Fact B. The Poincaré duality now follows from the Mayer-
Vietoris principle.

Remark 7.2.9. Using the Poincaré duality we can associate to any smooth map f : M ->
N between compact oriented manifolds of dimensions m and respectively n a natural push-
forward or Gysin map
f * : H . (M)->H°+</(N)V q : = d i m N - d i m M = n - m )
defined by the composition
.
H (M)
D
f (He
-.
(M))
* (f*)T
-> ( H e
-.
(N))
* ID-1
H
. -I-n
-
mol),
where (f*lt denotes the transpose of the pullback morphism.

Corollary 7.2.10. IfM G M+ then

. iv 'n-k .
Proof. Since I-I"'pt(M) is finite dimensional, the transpose
1>tm : (HZ,/"(M))** -> (Hk<=(M))*
is an isomorphism. On the other hand, for any finite dimensional vector space there exists
a natural isomorphism
v**-v.
Corollary 7.2.11. Let M be a compact, connected, oriented, n-dimensional manifold.
Then the pairing
H'"(m) X H"-'"(m) -> R (wt7l »-> w /\ 77
M
i5 a duality. In particular; 6k(Ml = b,,_k(Ml, We.

If M is connected H0(M) H " ( m ) R so that H" (M) is generated by any volume

f\_./

form defining the orientation.

The symmetry of Betti numbers can be translated in the language of Poincaré polyno-
mials as
PMltl-
tn Pm
(1)t (7.2.1)

Example 7.2.12. Let Eg denote the connected sum of g tori. We have shown that
Xlzgl =b0-b1-l-b2=2-Qg.
Since 2g is connected, the Poincaré duality implies 62 = 60 = 1. Hence, bi = 2g, i.e.,
pag (t) = 1 + Qgt + t2.
272 Lectures on the Geometry of Manzfolds

Consider now a compact oriented smooth manifold such that dim M Qk. The Kro-
necker pairing induces a non-degenerate bilinear form

J=H'<=(m) X H'"(m)-mR( U((»J,v) wAs.

M
The bilinear form I] is called the cosmological intersection form of M.
When k is even (so that n is divisible by 4) U is a symmetric form. Its signature is called
the signature of M, and it is denoted by o ( M ) .
When lc is odd, :J is skew-symmetric, i.e., it is a symplectic form. In particular,
H 2 k + l (M) must be an even dimensional space.

Corollary 7.2.13. For any compact manifold M 6 M4+k+2 the middle Betti number
b2k+1 ( M ) i5 even.

Exercise 7.2.14. (a) Let P Q ZH] be an odd degree polynomial with non-negative integer
coefficients such that P(0) = 1. Show that if P satisfies the symmetry condition (7.2.1)
there exists a compact, connected, oriented manifold M such that Pm (t) = P(t).
(b) Let P 6 Z[t] be a polynomial of degree 2k with non-negative integer coefficients.
Assume P(0) = l and P satisfies (7.2.1). If the coefficient of tk is even then there exists a
compact connected manifold M 6 myK: such that Pm(t) = P(t).
Hint: Describe the Poincaré polynomial of a connect sum in terms of the polynomials of
its constituents. Combine this fact with the Kenneth formula.

Remark 7.2.15. The result in the above exercise is sharp. Using his intersection theorem
F. Hirzebruch showed that there exist no smooth manifolds M of dimension 12 or 20 with
Poincaré polynomials 1 + 156 + 1512 and respectively 1 + t o -|- t20. Note that in each of these
cases both middle Betti numbers are odd. For details we refer to J. P. Serre, "Travaux de
Hirzebruch sur la topologic des variété5", Seminaire ourbaki 1953/54, nO 88.

7.3 Intersection theory

7.3.1 Cycles and their duals

Suppose that M is a smooth manifold.

Definition 7.3.1. A k-dimensional cycle in M is a pair (S, a , where S is a compact,

oriented k-dimensional manifold without boundary, and QUO : S -> M is a smooth map.
We denote by € k ( M ) the set of k-dimensional cycles in M.

Definition 7.3.2. (a) Two cycles (SQ, (150), (So, $51) E € k ( M ) are said to be concordant,
and we write this (SO, QSo) -C (So, $51), if there exists a compact, oriented manifold with
boundary 2, and a smooth map Q : Z] -> M such that the following hold.
(al) 32 = (-5ol l_l $1 where -So denotes the oriented manifold SO equipped with the
opposite orientation, and "LI" denotes the disjoint union.
(a2) <I>l Si (15i, i = 0, 1.
 Cohomology 273

(b) A cycle (S, go) Q (`3k(Ml is called trivial if there exists a (k -|- 1)-dimensional,
oriented manifold Z] with (oriented) boundary S, and a smooth map <I> : Z] -> M such that
@law = QS. We denote by 'I/c(M) the set of trivial cycles.
(c) A cycle (S, go) Q (`3k(Ml is said to be degenerate if it is concordant to a constant
cycle, i.e., a cycle (S', q25/) such that QS' is map constant on the components of S'. We
denote by Do (M) the set of degenerate cycles.

Exercise 7.3.3. Let ls0a¢0l Ne (S17¢1l° Prove that (-50 LIS1» (250 UQ51) is a trivial cycle.

S
1

Fig. 7.1 A cobordism in R3

The cobordism relation on Go1I (M) is an equivalence relation,3 and we denote by Zh, (M)
the set of equivalence classes. For any cycle (S, QS) 6 Mk (Ml we denote by [S, QUO] its image
in Z,k(M). Since M is connected, all the trivial cycles are concordant, and they define an
element in Zk1I(M) which we denote by [0].
Given 1S0, gz501, 1S1, Q511 6 Z,k(M) we define

[So <2501 + [So $51] [SQ U S1,(l50 U

O

(251],
.

where "Ll" denotes the disjoint union.

Proposition 7.3.4. Suppose M is a smooth manifold.

(a) Let [S Q , [S'-, QS'-1 Q €k(M), i = 0, 1. 1f[S17 Of] -6 [S'~, q25'~], fori = 0, 1, then

[So U $1, (150 U Q51] No [56 l-l $3 (156 U (1531

so that the above map + : Zk(M) X Zk(M) -> Zn ( M ) i5 well defined.
oWe urge the reader to supply a proof of this fact.
274 Lectures on the Geometry of Manzfolds

(b) The binary operation -|- induces a structure ofAbelian group on Zn ( M ) . The trivial
element i5 represented by the trivial cycles. Moreover;

-m = [-5,¢] 6 mm).
Exercise 7.3.5. Prove the above proposition.

We denote by ?€k(M) the quotient of Z k ( M ) modulo the subgroup generated by the

degenerate cycles. Let us point out that any trivial cycle is degenerate, but the converse is
not necessarily true.
Suppose M Q M+. Any k-cycle (S, (15) defines a linear map H k ( M ) -> R given by

H'"<m> 9 w +-> q25*w.

.J ?

Stokes formula shows that this map is well defined, i.e., it is independent of the closed form
representing a cosmology class.
Indeed, if LU is exact, i.e., w = du/, then
¢*dw/ dl)*wI 0.
w w
> >

In other words, each cycle defines an element in ( H k ( M ) )*. Via the Poincaré duality we
identify the vector space (H1<(M) )* with Henpt k ( M ) . Thus, there exists 55 6 Henpt k (M)
such that

CU /\ 65 = g2S*w Vi E H'"(m).
r w
>

The compactly supported cosmology class is is called the Poincaré dual of (S, QS).
There exist many closed forms 17 6 Q'nptk ( M ) representing is» When there is no risk
of confusion, we continue denote any such representative by is .

Example 7.3.6. Let M R", and S is a point, S {pt} C R". pt is canonically a

/l

- _ >

Fig. 7.2 The dual o f a point is Dirac's distribution.

 Cohomology 275

0-cycle. Its Poincaré dual is a compactly supported n-form w such that for any constant A
(i.e., closed 0-form)

A A,

i.e.,

CU 1.
Rn

Thus opt can be represented by any compactly supported n-form with integral 1. In partic-
ular, we can choose representatives with arbitrarily small supports. Their "profiles" look
like in Figure 7.2. "At limit" they approach Dirac's delta distribution.

Example 7.3.7. Consider an n-dimensional, compact, connected, oriented manifold M.

We denote by [M] the cycle (M, HM). Then 6[M] = l 6 H 0 ( M ) .

@)For any do erentialform w, we set (for typographical reasons), lwl := deg w.

Example 7.3.8. Consider the manifolds M 6 M+ and N 6 M+_ (The manifolds M and
N need not be compact.) To any pair of cycles (S, co) 6 @p(M), and (T, QD) G €q(N) we
can associate the cycle (S X T , go X up) 6 @p+q(M X N). We denote by 7rm (respectively
7rN) the natural projection M X N -> M (respectively M X N -> N). We want to prove
the equality

(SSxT I l-1l(M -P)q6 S X IT 7 (7.3.1)

where
QUO
cu X 17 1rmw /\ 1rN17, 'v'(w,17) Q Q'(M) X (N)-
Pick (w, 17) E Q' (Ml X Q' ( N ) such that, deg -|- degn = (m+n) - (p+ql~ Then,
using Exercise 7. 1.50

(w X t/)A(6s X 5Tl=l-1l(m-p)lt1l (WAS) X (nAsT).

MXN MXN

The above integral should be understood in the generalized sense of Kronecker pairing.
The only time when this pairing does not vanish is when lwl = p and 1171 = q. In this case
the last term equals

A 65 A IT q25*w
(_1lq(tn-p)
(f S
CU
)(/ T
T)
) I l_1)q("®-p)
(f S )(/ )T
zp*T)

= (-1)q(tf.-p)
/SXT
() X w* (cu A H)-
This establishes the equality (7.3. 1).
276 Lectures on the Geometry of Manzfolds

Example 7.3.9. Consider a compact manifold M 6 M+. Fix a basis (Wil of H' (M) such
that each We is homogeneous of degree = do. and denote by wi the basis of H ' ( M )
dual to (We) with respect to the Kronecker pairing, i.e.,

( w ' i w j ) , . = I-1)l°JI°lw3l(w3 ii) 55.

In M X M there exists a remarkable cycle, the diagonal

A = A m : M - > M X M , 33+->(ac,ac).

We claim that the Poincaré dual of this cycle is

(SA = $M ii i
( 1 ) I IU.) XC4J»,;. (7.3.2)

Indeed, for any homogeneous forms a,B 6 Q' (M) such that 101 +I/3l n, we have

(Q X BIT $m Z( 1) Iwil (Q X B) /\ (w X ii)

MXM MXM

(-1)l<»J'l(-1)IBI'l(»'l (we) x ( B )
MXM

(_1llw'll_1llBI'lw'l
(f M
QA
)(/ M
5/\UJ1
)
The z'-th summand is nontrivial only when III Iii I, and III Z lwil° Using the equality
l,yl -|- |,y|2 E 0(mod 2) we deduce

A/IXM
(a X mA am El/ M
QA
)(/ M
5/\0Ji
) <@»w%(0»w¢)-

From the equalities

a w(w',0),. and B <5»°J3 »)»<,w1

ii j

we conclude

/
M
A*(0 X/3)
/ / (Z
M
QAB
M i
We(wz,Q)HlA (2
j
(B=Wj)HWjl

/2 M
"M
(w¢»@)(6»wi)»¢wi Awe Z
M
(ii»&)(B,wi)~(wi»wi)~

(_1)lwil(n-lwil)(& wi)l_1)lwl(n-ImI) (5 0)11):.c

(Q, 'Z>H<>Wi>f.°
Equality (7.3.2) is proved.
 Cohomology 277

Proposition 7.3.10. Let M 6 M+ be a manzfold, and suppose that (S¢,(z5¢) 6 €klml

( i = 0, 1) are Mo k-cycles in M.
(6l)Ifl$0, (Z5ol ~¢ ($1» Q51l» then 650 = 651 in Henpt k(M)°
(b) If(S0, gb0) is trivial, then 5$0 = 0 in Hcnpt '<(M).
(C) (5$OU$1 = 550 -|- 551 in HCnpt k(M)°
ld) (LSO = -550 in H§ptk(m).

Proof. (a) Consider a compact manifold Z] with boundary <92 = -SO l_l $1 and a smooth
map i1'> : 2 -> M such that i1'> law= Q50 l_l (al. For any closed k-form to Q m m ) we have

/ Q*
/ d@*w Stokes
/ / /
o

(do) w Qblw (2500)-

II

2 2 as $1 so
Part (b) is left to the reader. Part (c) is obvious. To prove (d) consider 2 = [0, 1] X SO and
QUO : [0, 1] X SO -> M , <I>(t,;u) = q250(a3), 'v'(t,a3l 6 2.
Note that 32 = (-S0) LI So so that
<L$0 (550 = 6-$0u$0 = 502 = 0.
The above proposition shows that the correspondence
(MM) 9 (5, (/)) »-> 65 e H ; ; k ( m )
descends to a map
5 : oH,<,(ml -> H"pk(ml.
This is usually called the homological Poincaré duality. We are not claiming that 6 is an
isomorphism.

7.3.2 Intersection theory

Consider M € 3V[+, and a k-dimensional compact oriented submanifold S of M. We
denote by 'Z : S <-> M inclusion map so that (S, z) is a k-cycle.
Definition 7.3.11. A smooth map <15 T -> M from an (n - k)-dimensional, oriented
manifold T is said to be transversal to S, and we write this S Rh QS, if the following hold.

(a) g25-1 (S) is a finite subset of T,

(b) for every an 6 (15-1 (S) we have
go* (T,,,T) -|- T¢>(x)S ; T¢(m)M (direct sums.

If S Rh ®, then for each ac' E q25-1 (So we define the local intersection number at at to be
(or = orientation)
i,,(S, T) 1, OT' iT¢<m)S) A or((l5*TasT) t°(T¢» (x)M)
1, o'°(T¢(x)S) A O ( ¢ * T T ) -OTlT¢(xlM)
Finally, we define the intersection number of S with T to be
-
S T := i,,(S, To.
w€d>-1(S)
278 Lectures on the Geometry of Manzfolds

1/ 1

Fig. 7.3 The intersection number o f the two cycles on T 2 is 1.

Our next result offers a different description of the intersection number indicating how
one can drop the transversality assumption from the original definition.

Proposition 7.3.12. Let M E 3vI+. Consider a compact, oriented, k-dimensional sub-

manifold S <-> M, and (T, go) E €,,_k (]l/I) a ( n - it)-dimensional cycle intersecting S
transversally, i.e., S Rh QS. Then

S-T 65 /\ (5:r, (7.3.3)

where 6. denotes the Poincaré dual of o.

The proof of the proposition relies on a couple of technical lemmata of independent

interest.

Lemma 7.3.13 (Localization lemma). Let M 6 M+ and (S, (be 6 € k ( M ) . Then, for any
open neighborhood N of Q5(S ) in M there exists 65n 6 Q;*'"(m) such that
(a) 65n represents the Poincaré dual 65 Q HCnpt '"(M),-
(b) supp $5n C N.

Proof. Fix a Riemann metric on M. Each point p 6 Cl5(S) has a geodesically convex open
neighborhood entirely contained in N. Cover g25(S) by finitely many such neighborhoods,
and denote by N their union. Then N Q M+, and (S, (15) E ( N ) .
Denote by (SsN the Poincaré dual of S in N. It can be represented by a closed form in
Qgptk (N) which we continue to denote by (SsN . If we pick a closed form w E Qk ( M ) , then
w | N is also closed, and

WA5sN W A 5sN Cl5*w.

r r T
>
 Cohomology 279

Hence, (SsN represents the Poincaré dual of S in Hype '"(m), and moreover, supp (SsN C N.

Definition 7.3.14. Let M E M+, and S --> M a compact, k-dimensional, oriented sub-
manifold of M. A local transversal at p E S is an embedding
Q) : B C R"-'" -> m, B = open ball centered at 0 e t k7

such that S Hn Q) and (l5-1($) I {0}~

Lemma 7.3.15. Let M Q M"L, S <-> M a compact, k-dimensional, oriented submode zfold
of M and ( B , (15) a local transversal at p 6 S. Then for any sujiciently "thin " closed
neighborhood N of S C M we have

s~(8,(l)) B¢*6SN

Proof. Using the transversality S Rh co), the implicit function theorem, and eventually re-
stricting go to a smaller ball, we deduce that, for some sufficiently "thin" neighborhood N of
S, there exist local coordinates (5151, . . . ,x") defined on some neighborhood U of p 6 M
diffeomorphic with the cube
{laval < 1, of;}
such that the following hold.
(i)SVlU={:z:k:+1 =--~= n
= 0 } , p = ( 0 , . . . 7 0).
(ii) The orientation of S n U is defined by dzc1 A - - - /\ dark.
'n-k
(iii) The map QUO : B ->M is expressed in these coordinates as
C R(y1»---,y""°)

cc
1
= 0, ..., co:k =0,$ck+1 = y1 7 7
in in k

( i v ) N O U = {l:uil < 1 / 2 ; j = 1 , . . . , n } .
Let e :l:1 such that edit1 /\ . . . /\ d s " defines the orientation of TM. In other words,
e s~ (B, ¢).
For each f o l 7 . , xi) 6 S al U denote by Ps the (n - k)-"plane"
Pe +1 iN
7

{(§§£z:k . . . 7l:1:il<1 j > k } .

7

(n - l- 1 /\
We orient each Pe using the k)°foI'm d$k A do", and set

55N .
..II

U(6)
Pe
Equivalently,

u(gl (z5§5sN 7
B

where (25€ : B -> M is defined by

1 1
My , - - - , y " " " ) = (€,y ,---,y""")-
280 Lectures on the Geometry of Manzfolds

To any function up go(§l Q COO(S re U) such that

suppgoc no _< 1 / 2 , ¢_</},

we associate the k-form

WW lode A • A dank = (pdf Q Q'*pt(srw u).

Extend the functions a 1c , . . . , wk € C°°(Umn) to smooth compactly supported functions
i 6 03°>(M) -> 10, 11.
The form WW is then the restriction to U O S of the closed compactly supported form

l;'(/) 90(§:1,...,iz/")d§:1 /\ --~Adi~":.

We have

/ M
l;'(/) /\ 6?
The integral over U can be evaluated using the Fubini theorem. Write
/ U
Wav /\ (SsN
=/ S
WW
/ Rk
(p(§)d§- (7.3.4)

5SN= fd;d"+1 A ~ » ~ A d x " + Q,

where Q is an (.7, - k)-form not containing the monomial dxk+1 /\ /\ do". Then
A A

oJA5 f (,od:z:1 A - - - A d m "

r
JU
= 6
/ U
fc,old;z31 /\ -- /\ d:z:"l (lda21 /\ /\ d;/:"l = Lebesgue density)

Fubini
= e
I ( / SOU
s)
Pe
fly@:k-I-1 A /\ d:z:"l Idol

E
I ( / )
S
(,f2(€l
Pi
65" Idol E
/
S(/)(€)u(€l|d€|-

Comparing with (7.3.4), and taking into account that SO was chosen arbitrarily, we deduce
that

B¢5*6sN=v(0) E.

The local transversal lemma is proved.

Proof of Proposition 7.3.12. Let

<!>-1(5) = {p1,...,pm}.
The transversality assumption implies that each PI has an open neighborhood Be' diffeo-
morphic to an open ball such that (bi (la I B is a local transversal at Yi = q25(pi). Moreover,
I

we can choose the neighborhoods Be' to be mutually disjoint. Then

S~T
 Cohomology 281

The compact set K := <25(T \ UBi) does not intersect S so that we can find a "thin", closed
neighborhood" N of S <-> M such that K O N = 0). Then, QS* (SsN is compactly supported
in the union of the BE's and

M
68N /\ (ST
From the local transversal lemma we get
T
¢* (SSN
2/ Bi
(15I6sN-

8 ¢*6§ = S ( l m ) 21% (sJ)-

Equality (7.3.3) has a remarkable feature. Its right-hand-side is an integer which is
defined only for cycles S, T such that S is embedded and S to T.
The left-hand-side makes sense for any cycles of complementary dimensions, but a
priori it may not be an integer. In any event, we have a remarkable consequence.
Corollary 7.3.16. Let ( s ) 6 ( M M ) and ((T¢,¢¢) Q (2,,-k(m) where M G m+,
0, 1. 1f

(a) SO N o SO) To N o T19

(b) the cycles S are embedded, and
(c) Si Hn To,

then
S 0 ° T 0 = $ 1 -$1.

Definition 7.3.17. The homological intersection pairing is the Z-bilinear map

j = j m : j l f k l M ) X j{n_kl]ll)->]R,
(M E JV[+) defined by

3(S, T) is (iT.
M

We have proved that in some special instances 3(S, T ) Q Z. We want to prove that,
when M is compact, this is always the case.
Theorem 7.3.18. Let M 6 M+ be compact manifold. Then, for any (S, To 6 1TCk(Ml X

17'(n_k III, the intersection number U(S, To is an integer:

The theorem will follow from two lemmata. The first one will show that it suffices to
consider only the situation when one of the two cycles is embedded. The second one will
show that, if one of the cycles is embedded, then the second cycle can be deformed so that
it intersects the former transversally. (This is called a general position result.)

Lemma 7.3.19 (Reduction-to-diagonal trick). Let M, S and T as in Theorem 7.8'.]8.

Then

:](S,T) I (-1)"-k 5SXT /\ (YA,

MXM
where A is the diagonal cycle A : M -> M X M, as +-> (co, co). (It is here where the
compactness of M i5 essential, since otherwise A would not be a cycle. )
282 Lectures on the Geometry of Manzfolds

Proof. We will use the equality (7.3.l)

(SSxT I (-1)"- '"6s X 6:1

Then
( 1)" k
/MXM
(SSxTA(IA
MXM
ICes X IT) /\ PA

/M
A * ((is X IT)
M
55 (iT.

Lemma 7.3.20 (Moving lemma). Let S E @1K7(M)) and T 6 € n _ k ( M ) be two cycles in

M 6 M-I-. If S i5 embedded, then T is concordant to a cycle T such that S Hn T.

The proof of this result relies on Sard's theorem. For details we refer to [62], Chapter 3.
Proof of Theorem 7.3.18. Let l5,Tl E €klml X (211-k(M)~ Then,
:1<s,T> = (-1>"-'ms x T, Al.
Since A is embedded we may assume by the moving lemma that (S X T) Hn A, so that
U(S X T , A) Q Z.

7.3.3 The topological degree

Consider two compact, connected, oriented smooth manifolds M, N having the same di-
mension n. Any smooth map F : M -> N canonically defines an n-dimensional cycle in
M XN
FF : M - > M x N :Ur->(x,F(a:)).
FF is called the graph off.
Any point y E N defines an n-dimensional cycle M X {y}. Since N is connected all
these cycles are concordant so that the integer FF (M X {y}l is independent of y.

Definition 7.3.21. The topological degree of the map F is defined by

degF := (M x {y}) - FF.

Note that the intersections of FF with M X {y} correspond to the solutions of the
equation Thus the topological degree counts these solutions (with sign).

Proposition 7.3.22. Let F : M -> N as above. Then for any n-form w E AZ"(N)
n

F*w = deaF w.
r Jn
Remark 7.3.23. The map F induces a morphism

R r\/
H**(n) 1 H " ( M ) f"*J
R

which can be identified with a real number. The above proposition guarantees that this
number is an integer.
 Cohomology 283

Proof of the proposition. Note that if w E Q"(N) is exact, then

P P

F*w = 0.
'm
Thus, to prove the proposition it suffices to check it for any particular form which generates
Hn(N). Our candidate will be the Poincaré dual dy of a point y 6 N. We have

'So 1,
N
while equality (73.1) gives
5MX{9} = (Ym X (So - 1 X 53/~

We can then compute the degree of F using Theorem 7.3.18

deaF = deaF
/
N
(Sy
/MXN
(1 X (sys $pF
M
FF(l X (sys
M
F*6y.

Corollary 7.3.24 (Gauss-Bonnet). Consider a connected sum of g-tori E 2g embed-

ded in R3 and let 92 : E -> S2 be its Gauss map. Then
deg 92 = X(2) = 2 - 2g.

This corollary follows immediately from the considerations at the end of Subsec-
tion 4.2.4.

Exercise 7.3.25. Consider the compact, connected manifolds MI,M1,N 6 M+ and the
smooth maps F : MZ- -> N, i = 0, 1. Show that if F0 is concordant to F1 then deg F0 =
deg Fl. In particular, homotopic maps have the same degree.

Exercise 7.3.26. Let A be a nonsingular n X n real matrix. It defines a smooth map

Ax
FA : SN-1 -> SN-1 7 at »->
III
Prove that deg FA = sign det A.
-
Hint: Use the polar decomposition A = P O (where P is a positive symmetric matrix
and O is an orthogonal one) to deform A inside GL(n, RI to a diagonal matrix.

F
Exercise 7.3.27. Let M -> N be a smooth map (M, N are smooth, compact oriented of
dimension n). Assume y E N is a regular value of F, i.e., for all :u 6 F-1(y) the derivative
D F : T,,m -> TyN
is invertible. For QUO Q F-1(y) define
1, DasF preserves orientations
deg(F, so)
1, otherwise
Prove that
deg F = deg(F, x).
F(fv)=y
284 Lectures on the Geometry of Manzfolds

Exercise 7.3.28. Let M denote a compact oriented manifold, and consider a smooth map
F : M -> M. Regard H' (M) as a superspace with the obvious Z2 -grading
)
H. AS' H even IMl €l9 H 0 d d lm
( ) I

and define the Lefschetz number A(F) of F as the supertrace of the pull back
F* H' (M)-> H' (M).
Prove that A(F) = A FF, and deduce from this the Lefschetz fixed point theorem: A(F) ¢
0 => F has a fixed point.

Remark 7.3.29. For an elementary approach to degree theory, based only on Sard's theo-
rem, we refer to the beautiful book of Milnor [97].

7.3.4 The Thom isomorphism theorem

Let p : E -> B be an orientable fiber bundle (in the sense of Definition 3.4.45) with
standard fiber F and compact, oriented basis B. Let dim B = m and dim F = r. The total
space E is a compact orientable manifold which we equip with the fiber-first orientation.
In this subsection we will extensively use the techniques of fibered calculus described
in Subsection 3.4.5. The integration along fibers

= p* : Q;p(E) -> 9'-~'°(8)

E/B
satisfies
P* dE (-1)dBP* 7
so that it induces a map in cosmology
P* : H;,,(E) -> H'-w8).
This induced map in cohomology is sometimes called the Gysin map.

Remark 7.3.30. If the standard fiber F is compact, then the total space E is also compact.
Using Proposition 3.4.48 we deduce that for any w 6 Qs (El, and any 17 6 Q' (B) such that
deg -|- degn = dim E,
we have
<P*(W)v 77)f'é I (¢*t»P*l'7))/<~
If we denote by Com : Q' (M) _> QdimM-olml * the Poincare duality isomorphism on a
compact oriented smooth manifold M, then we can rewrite the above equality as

(iD81>*((»J)n7) I (DEW,P*'7) I (lP*)TDEW,77)

\.
1

D8p*lWl I l]9*lTDEW.

Hence
P* I ®B1lP*l>E»
so that, in this case, the Gysin map coincides with the pushforward (Gysin) map defined in
Remark 7.2.9.
 Cohomology 285

Exercise 7.3.31. Consider a smooth map f : M -> N between compact, oriented man-
ifolds M, N of dimensions m and respectively n. Denote by if the embedding of M in
M XN as the graph of f

M3a:»->(a;,f(a:)) € M X N.

The natural projection M X N -> N allows us to regard M X N as a trivial fiber bundle

over N.
Show that the push-forward map f* : H ' ( M ) -> H'+"""(N) defined in Re-
mark 7.2.9 can be equivalently defined by

f* =7r* o 7

where 7r* : H ' ( M X N)-> H"""(N) denotes the integration along fibers while

if)* H ' ( m ) -> H.+"(m X al,

is the pushforward morphism defined by the embedding i f .

Let us return to the fiber bundle p : E -> B. Any smooth section o : B -> E defines
an embedded cycle in E of dimension m = dim B. Denote by 5O' its Poincaré dual in
Hgpt
Using the properties of the integration along fibers we deduce that, for any w E Qm(B),
we have

E
5O' /\ p*w
I(/ B E/B )
5O' w.

On the other hand, by Poincaré duality we get

E
5O' Ap*w = (-1y"fn',
/ E
p*w /\ 6O'

(-am
/ B
O,*p*w=(-ll1°m
B
(po)*~»:(-1)*m
B
w.

Hence

p*50 6O' = (-l)""' E Q0(B).

E/B

Proposition 7.3.32. Let p : E -> B a bundle as above. If it admits at least one section,
then the Gysin map

P* Hgpt(El -> H'-*(B)

i5 surjective.
286 Lectures on the Geometry of Manzfolds

Proof. Denote by To the map

To : H' (B) -> '+'°(E)

Hept w +-> ( - 1 ) m 5 , /\p*w = P*w /\ 50.

Then To is a right inverse for p*. Indeed

w (-1)'°""p*6o w (-l)"Mp* (60 p*w) = p* (fO'wl.
The map p* is not injective in general. For example, if (S, go) is a k-cycle in F, then it
defines a cycle in any fiber 'ii-1 (6), and consequently in E. Denote by 68 its Poincaré dual
in Hip-klgl. Then for any w E §2M"*(B) we have

B
lp*5sl /\ cu
E
is /\ P*w d:
/
D
(1)*1>*(»J
/
since p o (b is constant. Hence p* 55 0, and we conclude that that if F carries nontrivial
S
(p o ¢)*(»J = 0,

cycles her p* may not be trivial.

The simplest example of standard fiber with only trivial cycles is a vector space.

Definition 7.3.33. Let p : E -> B be an orientable vector bundle over the compact oriented
manifold B (dirndl = m, rank (E) = T). The Thom class of E, denoted by TE is the
Poincaré dual of the cycle defined by the zero section go : B -> E, b +-> 0 6 Et. Note that
TE 6 HeptlEl°
Theorem 7.3.34 (Thom isomorphism). Let p : E -> B as in the above definition. Then
the map
7' H°(B) -> °+T(El
Hcpt w »-> TE Ap*w
is an isomorphism called the Thom isomorphism. Its inverse is the Gysin map

(-1)"'"p* : H;,..(E) -> H'-t(1-31.

Proof. We have already established that in = (-1)'~"», To prove the reverse equality,
To* = (-1)~"., we will use Lemma 7.2.2 of Subsection 7.2.1. For B 6 Q;pt(E) we have

(P*P*7'El /\ 5 TE /\ (p*p*B) Z (-ll'°d(\"(TE» in),

where in ( E , IN) G Q;ptlEl° Since P*P*TE ; (-1)rm, we deduce
(-l)T7716 = (TE As*(p*@) -|- exact form => (-1)"r°mB = TE OMB) l HZpt(E)~
Exercise 7.3.35. Show that TE : <*1, where C» » H ' ( M ) -> .+dimMlEt
Hcpt is the push-
forward map defined by a section Q : M -> E.

Remark 7.3.36. Suppose M 6 M+F, and S --> M is a compact, oriented, k-dimensional

manifold. Fix a Riemann metric g on M, and denote by Ns the normal bundle of the
embedding S --> M, i.e., the orthogonal complement of TN in (TM)IS.
The exponential map defined by the metric g defines a smooth map
exp:Ng->M,
 Cohomology 287

which induces a diffeomorphism from an open neighborhood 6 of S in Ns, to an open

(tubular) neighborhood N of S in M. Fix a closed form 65n E on-1<=(Ml with compact
support contained in N, and representing the Poincaré dual of (S, z'), where z' : S --> M
denotes the canonical inclusion.
Then exp*58N is the Poincaré dual of the cycle (S, Q) in Ns, where Q : S -> Ns denotes
the zero section. This shows that exp* so
is a compactly supported form representing the
Thom class of the normal bundle Ns .
Using the identification between O and N we obtain a natural submersive projection
'or : N -> S corresponding to the natural projection Ns -> S. In more intuitive terms, 71'
associates to each ac E N the unique point in S which closest to as. One can prove that the
Gysin map
H.
i* (S) -> H ; , ; ; ( " ( m ) ,
is given by,
H. (S) 9 cu |. > exp* (SsN /\ tr*w 6 H l " " " ) ( M ) . (7.3.5)

Exercise 7.3.37. Prove the equality (7.3.5).

7.3.5 Gauss-Bonnet revisited

We now examine a very special type of vector bundle: the tangent bundle of a compact,
oriented, smooth manifold M. Note first the following fact.

Exercise 7.3.38. Prove that M is orientable if and only if TM is orientable as a bundle.

Definition 7.3.39. Let E -> M be a real orientable vector bundle over the compact, ori-
ented, n-dimensional, smooth manifold M. Denote by TE 6 H"cpt the Thom class of
E. The Euler class of E is defined by
e(El CoTE E H"(M),
where QUO : M -> E denotes the zero section. e ( T M ) is called the Euler class of M, and it
is denoted by e ( M ) .

Note that the sections of TM are precisely the vector fields on M. Moreover, any such
section (7 : M -> TM tautologically defines an n-dimensional cycle in TM, and in fact,
any two such cycles are homotopic: try a homotopy, affine along the fibers of TM.
Any two sections 0 0 , 071 : M -> TM determine cycles of complementary dimension,
and thus the intersection number Oo • 0 1 is a well defined integer, independent of the two
sections. It is a number reflecting the topological structure of the manifold.

Proposition 7.3.40. Let 0 0 , 0 1 : M -> T M be two sections ofTer. Then

e(M) Oo • 01.
M
288 Lectures on the Geometry of Manzfolds

In particular If dim M i5 odd then

e(M) 0.
M

Proof. The section 00, 01 are concordant, and their Poincaré dual in Hcnpt (TM) is the Thom
class TM. Hence

0'0'0'1 (See /\ (So TM /\ TM

.ITM JTM

If dim M is odd then

TM
TM /\ (SCO
/M
CoTM
/ M

e(M) O'0°(71=-0'1°O'0 e(M).

M M

Theorem 7.3.41. Let M be a compact oriented n-dimensional manifold, and denote by

e ( M ) E H",(M) its Euler class. Then the integral of e ( M ) over M i5 equal to the Euler
enaracteristie of M,
n

e(M) = x(M) (--)'"bk(m).

M k=0

In the proof we will use an equivalent description of X(M).

Lemma 7.3.42. Denote by A the diagonal cycle A QQ-['nljlj X M).Thenx(M)=A~A.

Proof. Denote by AM E H;'pt(M X M) the Poincaré dual of A. Consider a basis (we) of

H. (M) consisting of homogeneous elements. We denote by (w"L) the dual basis, i.e.,

<W'l7 W3> "Z

61° vi, j.
According to (7.3.2), we have

AM (-1ll(»J'lw'L X Wi-

Similarly, if we start first with the basis In'i), then its dual basis is

('1)IWIIW'IW1»
so, taking into account that l¢~'il + Iwjl ' l°Jjl E "»l¢~8jl (mod 2% we also have
am = (-1)"~l°Jjl¢,,i X v i

Using Exercise 7.1.50 we deduce

A -A
MXM
AM /\ am
/ (
MXM ¢
i i 'i
( I I I 'cu x We
)( /\
j
( 1l n l w y l w j X v i
 Cohomology 289

l_1)lw°l+"°l°J.7ll_1)lw=l°l°Jylwi /\UJj X We x v i
MXM M

(-1llw'l+n~lw3l l-1ll°Jil°l°Jvl ( w , We)I<, ( W i n Wj)f<,-

M
In the last expression we now use the duality equations

(wi,w3) leVi l-IWI 51)

j
( l )
and the congruences
0 mod 2, lwl +nl~JI + W + lW¢l - we E we mod 2,
to conclude that
A.A l-1ll(»J"l = x(M)-
w '11

Proof of Theorem 7.3.41. The tangent bundle of M X M restricts to the diagonal A

as a rank 2n vector bundle. If we choose a Riemann metric on M X M then we get an
orthogonal splitting

T(M X M) IA= NA ® TA.

The diagonal map M -> M X M identifies M with A so that TA iv
T M . We now have
the following remarkable result.

Lemma 7.3.43. NA TM.

Proof Use the isomorphisms

T(M X MalA TA€BNA§TM€BNA

and

T(M X M)IA=TM€ETM.
From this lemma we immediately deduce the equality of Thom classes
TNA TM. (7.3.6)

At this point we want to invoke a technical result whose proof is left to the reader as an
exercise in Riemann geometry.

Lemma 7.3.44. Denote by exp the exponential map of a Riemann metric g on M X M .

Regard A as a submanzfold in N A via the embedding given by the zero section. Then there
exists an open neighborhood 'Ll of A C N A C T ( M X M ) such that
explu:U->M X M
i5 an embedding.
290 Lectures on the Geometry of Manzfolds

Let U be a neighborhood of A C NA as in the above lemma, and set N exp("L().

Denote by so the Poincare dual of A in "Ll, so
6 He*pt(U), and by 52
the Poincare dual of
A in N , (SAN E HcnptlNl° Then

62/mX 6AN*A6AN AM /\ am x(M)-

u N MXM

The cohomology class so is the Thom class of the bundle NA -> A, which in view of
(7.3.6) means that = so TNA = TM. We get

Hence
u
6XA6X / (so / CSTNA /
A A M
CSTm /
M
e(M).

M
e(M) xIMI-
If M is a connected sum of g tori then we can rephrase the Gauss-Bonnet theorem as
follows.

Corollary 7.3.45. For any Riemann metric h OH Cl C0HH6C1601 Sum of g-tori Fig WE have
1
-e(h) = shd'l)h elxg) in H*(29)~
The remarkable feature of the Gauss-Bonnet theorem is that, once we choose a metric, we
can explicitly describe a representative of the Euler class in terms of the Riemann curvature.
The same is true for any compact, oriented, even-dimensional Riemann manifold. In
this generality, the result is known as Gauss-Bonnet-Chern, and we will have more to say
about it in the next two chapters.
We now have a new interpretation of the Euler characteristic of a compact oriented
manifold M.
Given a smooth vector field X on M, its "graph" in T M ,
FX = {(a2,X(x)) E TxM , :UE M},
is an n-dimensional submanifold of TM. The Euler characteristic is then the intersection
number
x(]M) F x • M ,
where we regard M as a submanifold in TM via the embedding given by the zero section.
In other words, the Euler characteristic counts (with sign) the zeroes of the vector fields on
M. For example if X(Ml 7é 0 this means that any vector field on M must have a zero! We
have thus proved the following result.

Corollary 7.3.46. I f x ( M ) 7é 0 then the tangent bundle T M is nontrivial.

The equality X(S2"l = 2 is particularly relevant in the vector field problem discussed
in Subsection 2.1.4. Using the notations of that subsection we can write
v(S2") 0.

We have thus solved "half" the vector field problem.

 Cohomology 291

Exercise 7.3.47. Let X be a vector field over the compact oriented manifold M. A point
as E M is said to be a non-degenerate zero of X if X (0) = 0 and

det

for some local coordinates w ) near kg such that the orientation of TU0M is given by
day1 A ---
A do". Prove that the local intersection number of FX with M at xo is given by

3xi
ix() (Fx» Ml sign det
( 3337 M ac 390

This is sometimes called the local index of X at to, and it is denoted by Il;(X7;L0l.

From the above exercise we deduce the following celebrated result.

Corollary 7.3.48 (Poincaré-Hopf). If X i5 a vectorjield along a compact, oriented man-

#old M , with only non-degenerated zeros 5131, . . . ,ask, then

x(M) 'i(X7£Cj).
j

Exercise 7.3.49. Let X be a vector field on Rn and having a non-degenerate zero at the
origin.
(a) Prove that for all r > 0 sufficiently small X has no zeros on S = {l:z:l = r}.
(b) Consider F : S -> S n - 1 defined by

l
X(a2).
F'""> 1 lx(w)l
Prove that i(X, 0) = deg F for all T > 0 sufficiently small.
Hint: Deform X to a linear vector field.

7.4 Symmetry and topology

The symmetry properties of a manifold have a great impact on its global (topological) struc-
ture. We devote this section to a more in depth investigation of the correlation symmetry-
topology.

7.4.1 Symmetric spaces

Definition 7.4.1. A homogeneous space is a smooth manifold M acted transitively by a

Lie group G called the symmetry group.
292 Lectures on the Geometry of Manzfolds

Recall that a smooth left action of a Lie group G, on a smooth manifold M

G X M - > M (g,m)*->9~m,
is called transitive if, for any m 6 M , the map
\Ifm:G9g+->g-mEM
is surjective. For any point ac of a homogeneous space M we define the isotropy group at QUO
by
U = {g Q G 9 11}-

Lemma 7.4.2. Let M be a homogeneous space with symmetry group G and ac, y 6 M.
Then
(a) Ux i5 a closed subgroup off,°
(b) Ux 2 39.

-
Proof. (a) is immediate. To prove (b), choose g E G such that y = g at. Then note that
Uy = g:]xg-l.

Remark 7.4.3. It is worth mentioning some fundamental results in the theory of Lie groups
which will shed a new light on the considerations of this section. For proofs we refer
to [61, 128].
Fact 1. Any closed subgroup of a Lie group is also a Lie group (see Remark 1.2.31). In
particular, the isotropy groups U33 of a homogeneous space are all Lie groups. They are
smooth submanifolds of the symmetry group.
Fact 2. Let G be a Lie group, and H a closed subgroup. Then the space of left cosets,
G / H : = { 9 ° H ; gaG},
can be given a smooth structure such that the map
G X (G/H) G/H lg1ȣ12Hl ** (9192) ' H
is smooth. The manifold G/H becomes a homogeneous space with symmetry group G.
All the isotropy groups subgroups of G conjugate to H.
Fact 3. If M is a homogeneous space with symmetry group G, and at 6 M, then M is
equivariantly diffeomorphic to G/jx, i.e., there exists a diffeomorphism
(l5:M->G/jx,
such that My y) = 9 - (My)

We will be mainly interested in a very special class of homogeneous spaces.

Definition 7.4.4. A symmetric space is a collection of data (M, h, G, (7, i) satisfying the
following conditions.
(a) (M, h) is a Riemann manifold.
(b) G is a connected Lie group acting isometrically and transitively on M
GxM9(g,m)»->g~m€M.
(c) 0 : M X M -> M is a smooth map (m1,m2) +-> 0m1 (mg) such that the following
hold.
 Cohomology 293

(cl) We 6 M of : M -> M is an isometric, and 0m(m) m.

(CZ) (to 0 0 m = He.
(63) D0mlTmM= -llTmm-
(04) ugh = gvmg-1.

(d)i:M X G->G,(m,g) +-> mg is a smooth map such that the following hold.

(dl) 'v'm E M, it : G -> G is a homomorphism of G.

(do) iTTI o iTTI = 1G~
(d3) igm = gi771g-1, am,
9) E M x G.

(e) 0771g0'TTI1(x) = 0mg0771(x) = im(g) ~a3, V m , x 6 M , g 6 G.

Remark 7.4.5. This may not be the most elegant definition of a symmetric space, and
certainly it is not the minimal one. As a matter of fact, a Riemann manifold (M, g) is a
symmetric space if and only if there exists a smooth map O' : M X M -> M satisfying
the conditions (cl), (c2) and (c3). We refer to [61,70] for an extensive presentation of this
subject, including a proof of the equivalence of the two descriptions. Our definition has one
academic advantage: it lists all the properties we need to establish the topological results
of this section.

The next exercise offers the reader a feeling of what symmetric spaces are all about. In
particular, it describes the geometric significance of the family of involutions of .
Exercise 7.4.6. Let ( M , hl be a symmetric space. Denote by V the Levi-Civita connec-
son, and by R the Riemann curvature tensor.
(a) Prove that VR = 0.
(b) Fix m E M , and let 'y(t) be a geodesic of M such that 7(0) m. Show that
(tmv(t) i by(-tl.
Example 7.4.7. Perhaps the most popular example of symmetric space is the round sphere
S" C lR7*+1. The symmetry group is SO(n -|- l), the group of orientation preserving
"rotations" of lR3."'+1. For each m 6 Sn-1-1 we denote by of the orthogonal reflection
through the 1-dimensional space determined by the radius Om. We then set i71,.(T) =
<7mT(7m1, VT E SO(n -|- 1). We let the reader check that O' and i satisfy all the required
axioms.

Example 7.4.8. Let G be a connected Lie group and m a bi-invariant Riemann metric on
G. The direct product G X G acts on G by
-1
h glhg2
191>92) '

This action is clearly transitive and since m is bi-invariant its is dsc isometric. Define
0:G X G - > G 0gh=gh-1g-1
and
i :G X (G X G)->G X G ig(g1,g2)=(9919 ,9g2g-1)-
294 Lectures on the Geometry of Manzfolds

We leave the reader to check that these data do indeed define a symmetric space structure
on (G, in). The symmetry group is G X G.

Example 7.4.9. Consider the complex Grassmannian M = Gr kl@"l . Recall that in Ex-
ample 1.2.22 we described Gr I<:(<€") as a submanifold of End+(v) - the linear space
of self-adjoint n X n complex matrices, via the map which associates to each complex
subspace S, the orthogonal projection Ps : C" -> CC" onto S.
l
The linear space End+"(C") has a natural metric g0(A, B) = -Re
2 tr (AB*) that re-
stricts to a Riemann metric g on M. The unitary group U(n) acts on End+ (ml by conju-
gation,

U(n) X End+((c">;<T,Al I
I >T*A TAT* •

Note that U(n) * M M , and 90 is U(n)-invariant. Thus U(n) acts transitively, and
isometrically on M.
For each subspace S 6 M define

RS Z= P5-PS =2P$-1.

The operator Rs is the orthogonal reflection through S-L_ Note that Rs E U(n), and
Re = 2. The map A »-> Rs *A is an involution of End+ ((Cn*). It descends to an involution
of M. We thus get an entire family of involutions

O' :M X M -> M, lP51,Ps2l *-> R$1 *PA2

Define

i :M X U(n) -> U(nl, isT = RsTRs.

We leave the reader to check that the above collection of data defines a symmetric space
structure on Gr ICl<€"l-

Exercise 7.4.10. Fill in the details left out in the above example.

7.4.2 Symmetry and eohomology

Definition 7.4.11. Let M be a homogeneous space with symmetry group G. A differential

form LUG Q *(M) is said to be (left) invariant if Zgw = CU Vg E G, where we denoted by

g
Q' (M)- Q' (M)
the pullback defined by the left action by g : m + - > g - m .

Proposition 7.4.12. Let M be a compact homogeneous space with compact, connected

symmetry group G. Then any chorology class of M can be represented by a (not neces-
sarily unique) invariant form.
 Cohomology 295

Proof. Denote by dvG(g) the normalized bi-invariant volume form on G. For any form
w E Q* (M) we define its G-average by

£gw dVG (g).

G
The form w is an invariant form on M. The proposition is a consequence of the following
result.

Lemma 7.4.13. Ifs i5 a closed form on M then G i5 closed and cohomologous to w.

Proof. The form w is obviously closed so we only need to prove it is cohomologous to w.

Consider a bi-invariant Riemann metric m on G. Since G is connected, the exponential
map
exp : LG -> G X +-> exp(tX)
is surjective. Choose 7 ° > 0 sufficiently small so that the map
exp : D, = {txt,,, 1° 7 XI LG} -> G
is an embedding. Set B = exp D . We can select finitely many go, . . , g r o G such that
m
G= U Be, Be := gjB'r-
j 1

Now pick a partition of unity (as) C C10°(G) subordinated to the cover (Bi), i.e.,

0g as g l, suppaj C Be, Old = 1.

j

Set

Since the volume of G is normalized to 1, and in as l, we deduce that in "J = l. For

any j = 1, , m define TO : so*(m) -> so*<m> by

To Ld as(g)€gw dvGl9l-
•
O

Note that

T joe) and dTjLd I Tjdw.

j

Each To is thus a cochain morphism. It induces a morphism in cosmology which we

continue to denote by Tj- The proof of the lemma will be completed in several steps.
Step 1. ]lonH*(M), for all g E G . Let X QLG such that g =expX. Define
f:I X M -> M ft(m) := exp(tX) m = exp(tX)m.
The map f is a homotopy connecting Hm with This concludes Step 1.
296 Lectures on the Geometry of Manzfolds

Step 2.

To
For t E [0, 1], consider Q5j,t : Be -> G defined as the composition
exp-1 -
t -) exp)

KJ
By > B, > Do" Dtr Be, G.

Define T Q*(M)->§2*(M) by

TjtLd
G
Old (g)€¢ ,¢(t)w dVG (g) = T]q25.7 tw. (7.4.1)

We claim that T80 is cochin homotopic to T .

To verify this claim sett := 6S, -oo < S < 0 and
gS = exp(eS exp-1(g)) Vg 6 B .
Then

w a(g]g)€gs€gw dvG(g).
Br Br

For each g E B the map (5, m) »-> \P.S(m) is ml defines a local flow on M. We denote
by Xg its infinitesimal generator. Then
d
ds (Usual / Br
011 (9jg)Lxg KgS £9 w dVG (9)

/ B,
as(gjg)(dmixg + ixgdm)(€9s€9 w) dvG(g).

Consequently,

/ (f as(g]gl(dixg + ixgd)(€gs£g w) dvG(g) ds.

II
I

-oo B,

(An argument entirely similar to the one we used in the proof of the Poincaré lemma shows
that the above improper integral is pointwise convergent.) From the above formula we
immediately read a cochain homotopy X : §Z'(M) -> §2"1(Ml connecting U-oo to Uo.
More precisely
0

{XlWl}la¢6MI=
/ (f
-oo B
@j(9j 9l 1§xg€9s*9,°/
}L dvG(9l ds.

Now notice that

To10(.4J
(/ 81
Oi (QICWG (9)
.
afgw,
*

while To,16,0 Tow. Taldng into account Step 1, we deduce T as Step 2 is

completed.
 Cohomology 297

The lemma and hence the proposition follow from the equality

CLj]1H*(m) To G average.
j j

The proposition we have just proved has a greater impact when M is a symmetric space.

Proposition 7.4.14. Let ( M , h) be an, oriented symmetric space with symmetry group G.
Then the following are true.
(a) Every invariant form on M is closed.
(b) If moreover M is compact, then the only invariant form cohomologous to zero is the
trivial one.

Proof. (a) Consider an invariant ac-form w. Fix Mg Q M and set (D um0w. We claim (D
is invariant. Indeed, Vg e G
* A

Zgrl: : : Z9afNow = (0m0g)*w = (gig0mog)*w = "t/°(9)'*' um w w.

Since Dam) ITeM= -]ITeM, we deduce that, at mo E M, we have (IJ = (-1)k:w.

Both w and (If are G-invariant, and we deduce that the above equality holds at any point
-
m = 9 mo. Invoking the transitivity of the G-action we conclude that

al: (-1)'"w on M .
In paiticular, do; = (-l)k*dw on M.
The (k + l)-forms do; = omodw 85, and do are both invariant and, arguing as
above, we deduce

do CZ (-llk+1dw'
The last two inequalities imply do = 0.
(b) Let w be an invariant form cohomologous to zero, i.e., cu = do. Denote by * the Hodge
*-operator corresponding to the invariant metric h. Since G acts by isometrics, the form
17 = *w is also invariant, so that d17 = 0. We can now integrate (M is compact), and use
Stokes theorem to get

wAs daA=d: aAd17=0.

M M M
This forces w E 0.

From Proposition 7.4.12 and the above theorem we deduce the following celebrated
result of Elie Caftan ([25])

Corollary 7.4.15 (E. Cartan). Let ( M , h ) be a compact, oriented symmetric space with
compact, connected symmetry group G. Then the chorology algebra H * ( M ) of M is
isomorphic with the graded algebra Qinv ( M ) of invariant forms on M.

In the coming subsections we will apply this result to the symmetric spaces discussed
in the previous subsection: the Lie groups and the complex Grassmannians.
298 Lectures on the Geometry of Manzfolds

7.4.3 The eohomology of eompaet Lie groups

Consider a compact, connected Lie group G, and denote by LG its Lie algebra. According
to Proposition 7.4. 12, in computing its cosmology, it suffices to restrict our considerations
to the subcomplex consisting of left invariant forms. This can be identified with the exterior
algebra A°LG. We deduce the following result.

Corollary 7.4.16. H°(G) 5 H' (LG) -2 Air?/CG) where AULZG denotes the algebra of
bi-invariantforms on G, while H' (LG) denotes the Lie algebra conomology introduced in
Example 7.1 .I0.

Using the Exercise 7.1.11 we deduce the following consequence.

6
Corollary 7.4.17. If G is a compact semisimple Lie group then H l (G)

II
Proposition 7.4.18. Let G be a compact semisimple Lie group. Then H2(G) = 0.

Proof. A closed bi-invariant 2-form w on G is uniquely defined by its restriction to LG,

and satisfies the following conditions.
do = 0 ' \I I\ i
W([X0» X1]»X2) - W([X0»X2]» Xl) + W([X1» X2]» Xo) = 0,
and (right-invariance)

(Lx0W)(X1»X2) = 0 VX0 G LG ~: \ °J([X0»X1]»X2) - W([X0»X2]»X1) = 0.

Thus

W([X0»X1]»X2) 0 VX0,X1,X2 E LG.

On the other hand, since H1 (LG) = 0 we deduce (see Exercise 7.l.11) LG [£G,£G],
so that the last equality can be rephrased as
w(X,Y) 0 VX,Y E i5G~

Definition 7.4.19. A Lie algebra is called simple if it has no nontrivial ideals. A Lie group
is called simple if its Lie algebra is simple.

Exercise 7.4.20. Prove that SU(n) and SO(m) are simple.

Proposition 7.4.21. Let G be a compact, simple Lie group. Then H3(G) r\/
R. Moreover,
H3 (G) is generated by the Caftan form
a(X,Y,Z) 1 I=z([X,Y],Z),
where ii denotes the Killing pairing.

The proof of the proposition is contained in the following sequence of exercises.

Exercise 7.4.22. Prove that a simple je algebra is necessarily semi-simple.

 Cohomology 299

Exercise 7.4.23. Let w be a closed, bi-invariant 3-form on a Lie group G. Then

w(X,Y, [z,T]) = w([X,Y], z,T) VX,Y, z,T 6 LG.

Exercise 7.4.24. Let U.) be a closed, bi-invariant 3-form on a compact, semisimple Lie
group.
(a) Prove that for any X E LG there exists a unique left-invariant form 17X E Q1 (G) such
that

lixw)lY»Zl = *7xl[Y»Z])-
Moreover, the correspondence X »-> 17X is linear. Hint: Use Hl(Gl = H2(G) = 0.
(b) Denote by A the linear operator LG -> LG defined by
f<z(AX,Y) = nX(Y).
Prove that A is self-adjoint with respect to the Killing metric.
(c) Prove that the eigenspaces of A are ideals of LG. Use this to prove Proposition 7.4.21.

Exercise 7.4.25. Compute

a and as
SU(2) SO(3)

where a denotes the Cartan form. (These groups are oriented by their Cartan forms.)
Hint: Use the computation in the Exercise 4.1.60 and the double cover SU(2) -> SO(3)
described in the Subsection 6.2.1. Pay very much attention to the various constants.

7.4.4 Invariant forms on Grassmannians and Weyl's integral formula

We will use the results of Subsection 7.4.2 to compute the Poincaré polynomial of the
complex Grassmannian GrkllCn). Set H = n - k.
As we have seen in the previous subsection, the Grassmannian Grk(C"') is a sym-
metric space with symmetry group U(n). It is a complex manifold so that it is orientable
(cf. Exercise 3.4.13). Alternatively, the orientability of GrkllCn) is a consequence of the
following fact.

Exercise 7.4.26. If M is a homogeneous space with connected isotropy groups, then M is

orientable.

We have to describe the U(n)-invariant forms on GrkllCnl. These forms are com-
pletely determined by their values at a particular point in the Grassmannian. We choose this
point to correspond to the subspace S0 determined by the canonical inclusion (Ck --> Q" .
The isotropy of SO is the group H = U(k) X U(€). The group H acts linearly on the
tangent space Vo = TSO Grk lm). If w is an U (n)-invariant form, then its restriction to V0
is an H-invariant skew-symmetric, multilinear map
Vo X X V0->R.
300 Lectures on the Geometry of Manzfolds

Conversely, any H-invariant element of A' Vo* extends via the transitive action of U(n)
to an invariant form on Grk((C"). Denote by Aivw the space of H-invariant elements of
A' V0*. We have thus established the following result.

Proposition 7.4.27. There exists an isomer rphism of graded R-algebras:

H.

We want to determine the Poincaré polynomial of the complexified Z-graded space,

A'nv (X) (I

Pkg(t) to dimqj Ajnv (X) (C PGrk l(C'n,) (t) .

j
Denote the action of H on V0 by
H 9 h »-> Th 6 Aut (Vo).
Using the equality (3.4.10) of Subsection 3.4.4 we deduce

pklfl | detmvo + tT,,)l2dvH(h), (7.4.2)

H
where dVH denotes the normalized bi-invariant volume form on H.
At this point, the above formula may look hopelessly complicated. Fortunately, it can
be dramatically simplified using a truly remarkable idea of H. Weyl.
Note first that the function
H 9 h »-> (/)(h) = ldetl0v0 + tT,,)l2
is a class function, i.e., up(ghg-1) = (,o(h), Vg, h E H.
Inside H sits the maximal torus 'IT = Tk X Te, where
To { drag (61917 . . ,eW'°) E U ( k ) } andbT
•
= {dig l€"1¢1 , . . . , e 1305e ) G vi.
Each h E U ( k ) X U(€) is conjugate to diagonal unitary matrix, i.e., there exists g E H
such that ghg'l 6 'Iii
We can rephrase this fact in terms of the conjugation action of H on itself
C :HH -> H (9,h)*>C9(I1)
X gag
The class functions are constant along the orbits of this conjugation action, and each such
orbit intersects the maximal torus T. In other words, a class function is completely de-
termined by its restriction to the maximal torus. Hence, it is reasonable to expect that we
ought to be able to describe the integral in (7.4.2) as an integral over 'll`. This is achieved in
a very explicit manner by the next result.
Define As, : Tk -> (C by
(610'L _ e ) ,
As (91, . ,@'°) H
l<i<j<n
i

On the unitary group U(k) we fix the bi-invariant metric m = Mk such that at T1U (k) we
have
m(x, Y) == Retr(XY*).
 Cohomology 301

If we think of X, Y as k X ac matrices, then

m(X, Y) Re E fzjy-ij •
M
We denote by dvk the volume form induced by this metric, and by Vi the total volume

Vi d'UI¢.

II
O
.
U(k)

Theorem 7.4.28 (Weyl's integration formula). Consider a class function SO on the group
G U(k1) X X U(k S), denote by dg the volume form

dg = dvk1 /\ /\ d'UI¢s 7

and by V the volume ofG, V L Vi1 . vis . Then

S
l l l
V
llllllllll

G co( g )dg Z S
Hj 1 ki!vo1<'r) or so(0 la
7 0 s)
H Ink,(0M2¢£01
j 1
/\ A dos,

S
l
H j= 1(2'")'° kg I 'll` (0(01, 7 0 so H IA/<,(01)l2d91 /\ A dos.
j 1

Above, for every j = l , . . . , s, we denoted by 0 j the angular coordinates on Irv ,

0 j .- 1 . . . ,Mk'), o

while dog denotes the bi-invariant volume on T ,

The remainder of this subsection is devoted to the proof of this proposition. The reader
may skip this part at the first lecture, and go directly to Subsection 7.4.5 where this for-
mula is used to produce an explicit description of the Poincaré polynomial of a complex
Grassmannian.
Proof of Weyl's integration formula. To keep the main ideas as transparent as possible,
we will consider only the case S = 1. The general situation is entirely similar. Thus,
G = U(k) and T = Tk. In this case, the volume form dg is the volume form associated to
metric my .
We denote by Ad : G -> AUt(LG) the adjoint representation of G, see Example 3.4.30.
In this concrete case, Ad can be given the explicit description

Adg(x) QXQ 17 VgEU(k), X E _u(/<=l-

Denote the angular coordinates 0 on 'll` by 0 (91,... ek). The restriction of the metric
7

m to ']l` is described in these coordinates by

in TI (do1)2 + + (d9'°)2-
302 Lectures on the Geometry of Manzfolds

To any class function ( p : G - > (C we associate the complex valued differential form
(/)(9)d9.
WW
Consider the homogeneous space G/T, and the smooth map
Q z T X G / T > G ( t , g T ) = gag 1
Note that if g1T = g2'll`, then glug1 1 = g1 to2 , so the map q is well defined.
We have a m-orthogonal splitting of the Lie algebra LG
LG Lay 6BLé.
The tangent space to G/T at l T E G/T can be identified with £ . For this reason we
will write LE/T instead of L .
Fix ac E G/T. Any g E G defines a linear map Lg : T,,G/T -> T G / T . Moreover,
if go: = has = y, then Lg and Lh differ by an element in the stabilizer of as E G/T. This
stabilizer is isomorphic to T, and in particular it is connected.
Hence, if w E det TaG/T, then L_gw E det TyG/T and Law E det TyG/T define the
same orientation of TaG/T. In other words, an orientation in one of the tangent spaces of
G/T "spreads" via the action of G to an orientation of the entire manifold. Thus, we can
orient G/T by fixing an orientation on LE/T.
We fix an orientation on LG, and we orient LT using the form do do1 /\ A dok.
The orientation on £G/ir will be determined by the condition
or(LG) L r(L'y) /\ o7r°(LG/T).
The proof of Weyl's integration formula will be carried out in three steps.
Step 1.
q*WS0 Vi.
'G /TXG/T
Step 2. We prove that there exists a positive constant Co such that for any class function
SO on G we have
k! q*l-Ucp Co (/)( 0 )oAk(0 )l2d0.
I'IFX G/'II` 'IF
Step 3. We prove that
Vi VIc
Co
vol (T) (wk
Step 1. We use the equality

q*w(p = degq we,

'II'x G/'II` G
so it suffices to compute the degree of q.
Denote by N('ll`) the normalizer of 'IT in G, i.e.,
NoT)I{gQG;gTg CT 7

and then form the Weyl group

W N('Ill)/'ll`.

Lemma 7.4.29. The Weyl group W i5 isomorphic to the group So of permutations of ac

symbols.
 Cohomology 303

Proof. This is a pompous rephrasing of the classical statement in linear algebra that two
unitary matrices are similar if and only if they have the same spectrum, multiplicities in-
cluded. The adjoint action of N (T) on T: diagonal unitary matrices simply permutes the
entries of a diagonal unitary matrix. This action descends to an action on the quotient W
so that W C So.
Conversely, any permutation of the entries of a diagonal matrix can be achieved by a
conjugation. Geometrically, this corresponds to a reordering of an orthonormal basis.
1
Lemma 7.4.30. Let a1, . . . , ak) Q Rk such that 1, - a
2'rr 7 . . , are linearly independent
over I . Set T := (exp(ial), . . . ,€xp('i0/")) E To. Then the sequence lTd)nez is dense in
T k . (The element T is said to be a generator of Tk.)

For the sake of clarity, we defer the proof of this lemma to the end of this subsection.

Lemma 7.4.31. Let T Q T o C G be a generator 0f"r f f . Then q - l l T ) C T X G/'ll` consists

OflWI = kl points.
Proof.

q(5,9T) 7' I
r\
\
9891 T 4 '»gag-1 SET.
In particular, g7rng-1 = s" E T, 'v'n € Z. Since ('r") is dense in T, we deduce

gTg-1 C TgeNWY
Hence

Q-1(T) = { (9'1'tQ,QT) € T x G/T ; Q G NUT) }»

and thus q-1 (T) has the same cardinality as the Weyl group W.

The metric m on LE/T extends to a G-invariant metric on G/']l`. It defines a left-

invariant volume form du on G/']l`. Observe that the volume form on T induced by the
metric m is precisely
dO 91 A»--AdH'*'.

Lemma 7.4.32. q*dg = IAk(9)l2d9 /\ du~ In particular, any generator T of irk C G is a

regular value ofq since Ak(O') 7£ 0, VI 6 q-1 (T).

Proof. Fix a point CU0 = (t0,g0']I`) E T X G/'ll`. We can identify TxolHlp X G/T) with
LT QE LE/T using the isometric action of 'll` X G on T X G/'ll`.
Fix X 6 LT, Y E LE/T. For every real number S consider
hS = Cllto exp(sX), go exp(sY)'Jlll = 90 exp(sY)t0 exp(sX) exp(-sY)g0 1 6 G.
We want to describe
d
h0 1h8 E T1G = LG.
ds 8=0
304 Lectures on the Geometry of Manzfolds

Using the Taylor expansions

exp(sX) = l + sX + O(s2), and exp(sY) l -|- sY + O(s2),

we deduce

h0 1h8 got() Ill + 5y)i0(1 + sX)(1 - $y)90 1 + O(s2)

l -|- .9 (90i0 1Yi0.90 1 -|- 90X90 1 " QoYQ() 1l + 0(32)-
Hence, the differential
Do Tx0('ll` X G/'ll`) -2 LT GO 5G/wr -> LT GO 5G/'lr -2 £G»

can be rewritten as
DH20q(x ® Y) = Ad90(Adt0 1 - ]1)Y + Ad90x,
or, in block form,
0 0
Dmoq Ad
go 0 Adt-1 l£c/wr
o

The linear operator Adg is an m-orthogonal endomorphism of LG, so that det Adg - i t .
On the other hand, since G = U(k) is connected, det Adg = det Ad1 = 1. Hence,
det D20q = detlAdt-1 G).
Now observe that LE/T = LT is equal to
{ X € u ( / ¢ l ; Xii = 0 v j = 1 , . . . , k } C _ H l k ) = £ G -
Given t = d i g (exp(f£911, . . . ,exp(i9/")) G To C U ( k ) , we can explicitly compute the
eigenvalues of Adt-1 acting on LE/T. More precisely, they are

{€XP('i(9z-9j)) ; 1 < wt' <k }~

Consequently

det D 0q = d€tlAdt-1 1 ) = Ink(t)l 2

Lemma 7.4.32 shows that q is an orientation preserving map. Using Lemma 7.4.31 and
Exercise 7.3.27 we deduce degq = IWI = kl. Step 1 is completed.
Step 2 follows immediately from Lemma 7.4.32. More precisely, we deduce

/ G
WW
1
k! U G/'II'

:Co
do" )/ Tk
(,f>(9)Ak(@>d@.

To complete Step 3, that is, to find the constant Co, we apply the above equality in the
special case (p = l so that W(P = dvk. We deduce
Co
Vi As( 0 )d0 7
Ki To
 Cohomology 305

so that
vick!
Co
fTo Aklgldg
Thus, we have to show that

As( 0 )d0 k!vo1(T) = (2'rr)kk!.

To

To compute the last integral we observe that Ak9.r(9) can be expressed as a Vandermonde
determinant
l l l
ew1 ew2 ewk
€2i91 €2i02 €2»z9'"
A/A9)

€m-1)61 €m-1)92 €i(/Q-1)9'<

This shows that we can write As as a trigonometric polynomial

As( 0 ) = €lo'l€ul0 ),
0€5I;

where, for any permutation o of {l, 2, 7 k} we denoted by 6(0) its signature, and by
eO'(0 ) the trigonometric monomial

to( 0 )=€O'l917 ,0'")

Hkj= €io(i)9j
Hkj= et?
The monomials 60 are orthogonal with respect to the L2-metric on To, and we deduce
k

To
IA,(0>l2d9=
2/
u€Sk, 'IT
let( 0 )l2d0 (2 Ye/<!.

This completes the proof of Weyl's integration formula.

Proof of Lemma 7.4.30. We follow Weyl's original approach ([129, 130]) in a modern
presentation.
Let X = C(']l`, Q) denote the Banach space of continuous complex valued functions on
'll`. We will prove that
n
1
lim n + fw) felt, WAX. (7.43)
n->oo 1 T
j 0

If U p 'll` is an open subset and f is a continuous, non-negative function supported in U

( f ¢ 0) then, for very large n,
n
l
n+ l fw) re
r a fdf e 0.
j 0 T

This means that f(T7) go 0, i.e., TO Q U for some j.

306 Lectures on the Geometry of Manzfolds

To prove the equality (7.4.3) consider the continuous linear functionals Ln, L : X - > l C
n
1
Ln( f) n+ 1 Fofo), L(f) Tfdt.

We have to prove that

lim Ln( f) L(f) f i X . (7.4.4)
n->oo

It suffices to establish (7.4.4) for any f Q S, where S is a subset of X spanning a dense

subspace. We let S be the subset consisting of the trigonometric monomials
@(@1, . . . , et) = exp(i<191) • €XP(iC/<9k)7 Q : ( g , . . . , Cal Q Mk.
The Weierstrass approximation theorem guarantees that this S spans a dense subspace. We
compute easily
n
l l €Claln+1 1
Le(€<) n+ l €i<(0¢) n + l
j 0 €<(c¥) - l
1 1
Since l , -27r
ala -as
727r are linearly independent over Q we deduce that €€lOz) ,é l for all
§ € Z ' " Hence

lim Lnlec) 0 €§dt = Ll€gl.

n->oo T

Lemma 7.4.30 is proved.

7.4.5 The Poincaré polynomial of complex Grassmannian

After this rather long detour we can continue our search for the Poincaré polynomial of
Gr k(<€")-
Let $0 denote the canonical subspace Ck <-> C". The tangent space of Grk((C") at $0
can be identified with the linear space 8 of complex linear maps (Co -> ICE Z = n - k. The 7

isotropy group at S0 is H = U(k) X U(€).

Exercise 7.4.33. Prove that the isotropy group H acts on 8 = { L : (ck -> (ce} by
(T,S) • L Z SLT* V L 6 8 , T 6 U(/<), S 6 U(€).

Consider the maximal torus Tk X TZ C H formed by the diagonal unitary matrices.

We will denote the elements of T k by
5 : (61, . . . ,(=:;), pa 6
27rr'i'r0'
7

and the elements of Te by

e: el? 7 be), €j €XP(27ri9i l7
The normalized measure on Tk is then
df del /\ /\ df," 7
 Cohomology 307

and the normalized measure on Te is

dO 1 do1 /\ /\ dee.

The element (Q, Q) 6 To X TZ viewed as a linear operator on 8 has eigenvalues

{€0€j l Of < ac 1 < J< Q,
Using the Weyl integration formula we deduce that the Poincaré polynomial of Gr A¢(@")
is
l
Pkfltl k!€ H
II + t § 0 € j l 2 I A k l § l T ) ) l 2 I A g ( § ( 0 ))l2dT /\ dO
Z
! T k x T 3 GJ

l
Ki!! TlCXT3
H l80 +
GJ
t € j l 2 I A k l § l T ) ) l 2 I A g ( Q l 0 ))l2dT /\ dH.

We definitely need to analyze the integrand in the above formula. Set

H
..II

f k l f l (pa -l- t@j)A1<=(§)Ae(Ql»

a,j

so that
l
Pkg(t) I al ad d0. (7.4.5)
km! T"=xTf e ( ) km() T A
We will study in great detail the formal expression

J/,e(t; w;y) H
GJ
lata + Wy)-
The Weyl group W = So X 86 acts on the variables (is y) by separately permuting the
:Hz-components and the y-components. If (0, up) E W then

J/¢,e(ts UCC); (/)(y)) ]/¢,e(t§ 113; y ) .

Thus we can write Jo as a sum

Jol t) 15dQdl5ulRdl3/)
d>0

where Qdlél?) and Rd(y) are symmetric polynomials in Q: and respectively y.

To understand the nature of these polynomials we need to introduce a very useful class
of symmetric polynomials, namely the Schur polynomials. This will require a short trip in
the beautiful subject of symmetric polynomials. An extensive presentation of this topic is
contained in the monograph [88].
A partition is a compactly supported, decreasing function
A : { 1 , 2 , ...} -> {0,1,2,...}.

We will describe a partition by an ordered finite collection ()\17>\27 ,An), where A1 >
> An > An-I-1 = 0. The length of a partition A is the number
L(A) := max{n ; An ¢ 0}.
308 Lectures on the Geometry of Manzfolds

6 5
5 4
5 4
3 3
1 3
1

Fig. 7.4 The conjugate o f (6,5,5,3,1) is (5,4,4,3,3,1).

The weight of a partition A is the number

II An.
'rz>1

Traditionally, one can visualize a partition using Young diagrams. A Young diagram is an
array of boxes arranged in left justified rows (see Figure 7.4). Given a partition (Al 2
- - > An), its Young diagram will have A1 boxes on the first row, A2 boxes on the second
row etc.
Any partition A has a conjugate X defined by

Rn #{j20;Aj2N}.
The Young diagram of §\ is the transpose of the Young diagram of A (see Figure 7.4).
A strict partition is a partition which is strictly decreasing on its support. Denote by
9% the set of partitions of length g n, and by yn the set o f strict partitions A of length
n - l < L(,\) g n. Clearly Tn C Tn- Denote by 5 = 6" E Tn the partition (n - l , n -
2,...,1,0,...).

Remark 7.4.34. The correspondence ?n 9 A H A + 6 n E yn is a bijection.

To any A 6 Tn we can associate a skew-symmetric polynomial

aA(S U1» . . . , a ; n ) d€tl.CU.7A ) e(0);Ul
U6SH

Note that asn is the Vandermonde determinant

a'5lxla 7 son) = det(x- -1-fi)

H
i<j
5131 to An (at).

For each A 6 Tn we have A -|- 6 E Tn, so that

0,A+6 is well defined and nontrivial.
Note that 0»A+6 vanishes when = $31, SO that the polynomial 0»>+6l5U1 is divisible by
S13i
each of the differences SQ - zvj ) and consequently, it is divisible by as. Hence

G,>\-I-5(£I3)
SAILU)
a6(;u)
is a well defined polynomial. It is a symmetric polynomial since each of the quantities
ClA+6 and 06 is skew-symmetric in its arguments. The polynomial $A (x) is called the
 Cohomology 309

Schur polynomial corresponding to the partition A. Note that each Schur polynomial SA is
homogeneous of degree III. We have the following remarkable result.

Lemma 7.4.35.
Jkg(t) tIIS(22)SX(?J)»
A€Tk,e

where

T o :={A A1 ac L(A) < 2

For each A E j)kg we denoted by X the complementary partition
X= ( x ¢ - A x , / < - A @ _ 1 , . . . , / < ; - A 1 ) .

___
6 7
4 6
>
4 5
2 3
1 3
0 I I II I 1
Fig. 7.5 The complementary of(6, 4, 4, 2, 1, 0) G 3'7,6 is (7, 6, 5, 3, 3, 1) E 9>6,7~

Geometrically, the partitions in T o are precisely those partitions whose Young dia-
grams fit inside a E X k rectangle. If A is such a partition then the Young diagram of the
complementary of X is (up to a 180° rotation) the complementary of the diagram of A in
the E X k rectangle (see Figure 7.5).
For a proof of Lemma 7.4.35 we refer to [88], Section 1.4, Example 5. The true essence
of the Schur polynomials is however representation theoretic, and a reader with a little more
representation theoretic background may want to consult the classical reference [87], Chap-
ter VI, Section 6.4, Theorem V, for a very exciting presentation of the Schur polynomials,
and the various identities they satisfy, including the one in Lemma 7.4.35.
Using (7.4.5), Lemma 7.4.35, and the definition of the Schur polynomials, we can
describe the Poincaré polynomial of Grk (@") as
2
l
p t ) I I tIAla$+(5k(§laX+5£(§) CUT /\ do. (7.4.6)
. l T k x T / 3 lApP/
e

The integrand in (7.4.6) is a linear combination of trigonometric monomials a11 . 5`k

e11 ---e , where the T'-S and S-S are nonnegative integers.
310 Lectures on the Geometry of Manzfolds

Note that if A, ,LL Q 9"k,e are distinct paltitions, then the terms "$\+al§) and 61/1+6(§)
have no monomials in common. Hence

To
0§+6(§)0N+6(§) df = 0.
Similarly,

TO
&x+5(Q)<»n+5(§)d0 = 0, if A ¢ ll-
On the other hand, a simple computation shows that

To
la§\-I-6(§)l2dT = Ki,
and

laX+6(§)l2d0 = € ! .
To

In other words, the terms

1/2
l

\ /

form an orthonormal system in the space of trigonometric (Fourier) polynomials endowed

with the L2 inner product. We deduce immediately from (7.4.6) that
_ 2X (7.4.7)
Pkg (t) - 1: I I
A€3 'k,e

The map
9>,, 9 A »-> X E 9'e,k 7
is a bijection so that
Pa:£(t) t2IAI p&klt). (7.4.8)
A€Te,k=

Computing the Betti numbers, i.e., the number of partitions in IP with a given weight is a
very complicated combinatorial problem, and currently there are no exact general formulae.
We will achieve the next best thing, and rewrite the Poincare polynomial as a "fake" rational
function.
Denote by I)I glw) the number of partitions A E 9P6 k with weight I,\l w . Hence
KB
Pkg(t) bk,glW)t2w

Alternatively, 6k,e (w) is the number of Young diagrams of weight w which fit inside a k x i
rectangle.

Lemma 7.4.36.
bk+1,e+1l1Ul = bk,e+1l'Lul -|- 61¢+1,el1U Z 1).
 Cohomology 311

k 1
R l+1 <
Is; >-

I
k+1
1 \ ¢l"""""

Fig. 7.6 The rectangles Re+1 and Rk-I-1 are "framed" inside R53§

Proof. Look at the (k + l) X 6 rectangle Rk-I-l inside the (k + l) X (Z + l)-rectangle

(see Figure 7.6). Then

61<:+1,e+1(w) # { diagrams of weight w which fit inside R513 }

#{ diagrams which Ht inside Rk-i-1 }

+ #{ diagrams which do not fit inside R/C-i-1

On the other hand,

bk+1 ,£ = #{ diagrams which fit inside Rk-I-1

If a diagram does not fit inside Rk-l-1 this means that its first line consists of Z + l boxes.
When we drop this line, we get a diagram of weight w - K - l which fits inside the k X
(E + 1) rectangle Re+1 of Figure 7.6. Thus, the second contribution to 6I¢+1,e+1lW) is
61¢,£+1l'W Z 1).

The result in the above lemma can be reformulated as

Pk+1,e+1(15) = Pk+1,€l15l + t2(€-t-1)Pk£+1(t).

Because the roles of k and Z are symmetric (of. (7.4.8)) we also have

Pk+1,1i+1(t) = Pk,e+1(15) -|- t2("°+1)Pk+1 g(t).

These two equality together yield
t2(k7+1l)
p%,+1(1 - t2(€-i-1) = PA¢+1,ell -

Let m = k + Z + l and set Qdnltl - Pd,m-d(?5)

rephrased as
-
- PG.(m,<:) (t)- The last equality can be

1 - tm-k
Qk+1,mlt) = C2/<7,7n l _ t2(k-I-1) 7
312 Lectures on the Geometry of Manzfolds

so that
1 _ t2lM-/C) 1 _ t2l'r7l-k-l-ll 1 _ t2l'r7,-1l
Qk+1,m(t) 1 _ t2(k-i-1) 1 _ tack
a
1 - t4 '

Now we can check easily that blyrn-1 (w) = 1, i.e.,

1 tom
Q1,fm(t) = 1 + £2
_|_ t2(m-1)
-|- t4 -I-
1 t2
Hence
<1 _ 152(m-1<+1)) . . . <1 _ 52nd)
PGk(m, (c) (t) Q1<,m(t) <1-¥y.}-
(1-¥)~1-Qm)
(1 - t2) • • (1 - z2/)(1 - ¢ 2 ) - - (1 - t2(m-k)l

Remark 7.4.37. (a) The invariant theoretic approach in computing the cosmology of
Gr 1<(<€") was used successfully for the first time by C. Ehresmann [41]. His method
was then extended to arbitrary compact, oriented symmetric spaces by H. Iwamoto [69].
However, we followed a different avenue which did not require Caltan's maximal weight
theory.
(b) We borrowed the idea of using the Weyl's integration formula from Weyl's classical
monograph [133] In turn, Weyl attributes this line of attack to R. Brauer. Our strategy is
however quite different from Weyl's. Weyl uses an equality similar to (7.4.5) to produce
an upper estimate for the Betti numbers (of U(n) in his case) and then produces by hand
sufficiently many invariant forms. The upper estimate is then used to established that these
are the only ones. We refer also to [123] for an explicit description of the invariant forms
on Grassmannians.

Exercise 7.4.38. Show that the cosmology algebra of CIP" is isomorphic to the truncated
ring of polynomials

R-[t]/(w"+1)»
where .CU is a formal variable of degree 2 while (;v"+1) denotes the ideal generated by mn-}-1
Hint: Describe A'vlnv QP" explicitly.

7.5 tech cosmology

In this last section we return to the problem formulated in the beginning of this chapter:
what is the relationship between the Cech and the DeRham approach. We will see that
these are essentially two equivalent facets of the same phenomenon. Understanding this
equivalence requires the introduction of a new and very versatile concept namely, the con-
cept of sheaf. This is done in the first part of the section. The second part is a fast paced
introduction to Cech cohomology. A concise yet very clear presentation of these topics can
be found in [63]. For a very detailed presentation of the classical aspects of subject we
refer to [52]. For a modern presentation, from the point of view of derived categories we
refer to the monographs [68, 73].
 Cohomology 313

7.5.1 Sheaves and presheaves

Consider a topological space X . The topology 'To on X , i.e., the collection of open subsets,
can be organized as a category. The morphisms are the inclusions U --> V . A presheaf of
Abelian groups on X is a contravariant functor S : 'To -> Ab.
In other words, S associates to each open set an Abelian group S(U), and to each
inclusion U -> V, a group morphism ToU : sly) -> S(U) such that, if U *-> V -> W,
then g= TvU o , If s 6 S(V) then, for any U --> V, we set

S lu= t~v"(s) 6 so).

I f f 6 S(U) then we define domf := U.
The presheaves of rings, modules, vector spaces are defined in an obvious fashion.

Example 7.5.1. Let X be a topological space. For each open set U C X we denote by
C'(U) the space of continuous functions U -> R The assignment U »-> C'(U) defines a
presheaf of R-algebras on X. The maps ToU are determined by the restrictions IN: C'(V) ->
co).
If X is a smooth manifold we get another presheaf U +-> COO(U). More generally, the
differential forms of degree k can be organized in a presheaf Qk (s). If E is a smooth vector
bundle, then the E-valued differential forms of degree k can be organized as a presheaf of
vector spaces
U »-> Q':=E(U) = Q'°(EIU).
If G is an Abelian group equipped with the discrete topology, then the G-valued continuous
functions C'(U, G) determine a presheaf called the constant G-presheaf which is denoted
by Qx -
Definition 7.5.2. A presheaf S on a topological space X is said to be a sheaf if the follow-
ing hold.

(a) If (UQ) is an open cover of the open set U, and f , 9 € 5 ( U ) satisfy fIUO,=9IU¢,»VCY,
then f = 9-
(b) If (UQ) is an open cover of the open set U , and to 6 5(Ua) satisfy
flue=fnu $U5 VUo, (W UB ,Q (D
then there exists f 6 S(U) such that, f Ina= fo,'v'a.

Example 7.5.3. All the presheaves discussed in Example 7.5.1 are sheaves.

Example 7.5.4. Consider the presheaf S over R defined by

S(U) := continuous, bounded functions f : U -> R.
We let the reader verify this is not a sheaf since the condition (b) is violated. The reason
behind this violation is that in the definition of this presheaf we included a global condition
namely the boundedness assumption.
314 Lectures on the Geometry of Manzfolds

Definition 7.5.5. Let X be a topological space, and R a commutative ring with 1. We

equip R with the discrete topology
(a) A space of germs over X ( " s p a c e étalé " in the French literature) is a topological space
8, together with a continuous map 'IT : 8 -> X satisfying the following conditions.

(al) The map 'IT is a local homeomoiphism that is, each point e 6 8 has a neighborhood
U such that 'IT In is a homeomorphism onto the open subset 7r(U) C X. For every
ac 6 X, the set fix := 7r"1(a3), is called the stalk at ac.
(a2) There exist continuous maps
R x 8->€,
and
+:{('u,,v)€€ X 8; tr(u)=7r(u) ->€,

such that
V T G R , 'v/:I:6X, V'u,,v€€,,, we have 1°-u6€a13, u + ' v € € : 1 3 ,

and with respect to the operations +, ~, the stalk is an R-module, Van ex

(b) A section of a space of germs 7r : 8 -> X over a subset Y C X is a continuous function
S : Y -> 8 such that s(y) 6 So, /y 6 Y. The spaces of sections defined over Y will be
denoted by 8(Y).

Example 7.5.6 (The space of germs associated to a presheaf). Let S be a presheaf of

Abelian groups over a topological space X. For each :u 6 X define an equivalence re-
lotion Nm on

U so),
U 81:
U runs through the open neighborhoods of X ,

by
f Nx g ~: '» EI open U 9 :u such that flu= g l u .
The equivalence class of f 6 Una: S(U) is denoted by [f]x, and it is called the germ of f
at as. Set
So { [ f ] x ; d 0 > f 9:6 }» and 3 := U So.
xiX

There exists a natural projection 7T : 3 -> X which maps [fix to ac. The "fibers" of this
map are 1r"1(a:) = Sac - t h e germs a t e 6 X. Any f 6 S(U) defines a subset

Up={[f]u;u€U co.
We can define a topology in 3 by indicating a basis of neighborhoods. A basis of open
neighborhoods of [f]as 6 S is given by the collection

{UQ ; U 9 My g E S(U) [9]w i lm }~

 Cohomology 315

We let the reader check that this collection of sets satisfies the axioms of a basis of neigh-
borhoods as discussed e.g. in [77].
With this topology, each f 6 S(U) defines a continuous section of 'IT over U

[ f ] : U9'u,»->[f]»,,, €Su.

Note that each fiber S ac has a well defined structure of Abelian group

[fly -|- [g],, = [(f + g ) lu]x U 9 an is open and U C dom O doing.

(Check that this addition is independent of the various choices.) Since 'IT : f ( U ) -> U is a
homeomorphism, it follows that 'IT : S -> X is a space of germs. It is called the space of
germs associated to the presheaf S.

If the space of germs associated to a sheaf S is a covering space, we say that S is a sheaf
of locally constant functions (valued in some discrete Abelian group). When the covering
is trivial, i.e., it is isomorphic to a product X X { discrete set 1, then the sheaf is really the
constant sheaf associated to a discrete Abelian group.

Example 7.5.7 (The sheaf associated to a space of germs). Consider a space of germs
a l> X over the topological space X. For each open subset U C X, we denote by
_ space of continuous sections U -> 8. The correspondence U »-> 8(U) clearly a
a (U) the
sheaf. 8 is called the sheaf associated to the space of germs.

Proposition 7.5.8. (a) Let 8 l> X be a space of germs. Then 3 = 8. _

(b) A presheafS over the topological space X is a sheaf zuzana' only 1f3 S.

Exercise 7.5.9. Prove the above proposition.

Definition 7.5.10. If S is a presheaf over the topological space X , then the sheaf 3 is called
the sheaf associated to S , or the sheafification of S.

Definition 7.S.11. (a) Let A morphism between the (pre)sheaves of Abelian groups (mod-
ules etc.) S and S over the topological space X is a collection of morphisms of Abelian
groups (modules etc.) he : S(U) -> SIU), one for each open set U C X, such that, when
V C U, we have he o TuV = ,=Up o he. Above, TuV denotes the restriction morphisms of S,
while the ,:UV denotes the restriction morphisms of 3. A morphism h is said to be injective
if each he is injective.
(b) Let S be a presheaf over the topological space X. A sub-presheaf of S is a pair (T, 2),
where 'J' is a presheaf over X , and 'L : 'J' -> S is an injective morphism. The morphism 'L is
called the canonical inclusion of the sub-presheaf.
316 Lectures on the Geometry of Manzfolds

Let h : S -> 'J' be a morphism of presheaves. The correspondence

U »-> kerhu C 5(U>,

defines a presheaf called the kernel of the morphism h. It is a sub-presheaf of S.

Proposition 7.5.12. Let h : S -> 'J' be a morphism of presheaves. If both S and 'J' are
sheaves, then so is the kernel of h.

The proof of this proposition is left to the reader as an exercise.

Definition 7.5.13. (a) Let S 1 X (i = 0, 1) be two spaces of germs over the same
topological space X. A morphism of spaces of germs is a continuous map h : 80 -> 81
satisfying the following conditions.

(al) tf1 o h = 7r0, i.e., h(tr0 1(x)) C Tr1 1(a:), Va: 6 X.

(a2) For any so 6 X, the induced map ha, : 7r0 1(:z:) -> w1 1(:I:) is a morphism of Abelian
groups (modules etc.).

The morphism h is called injective if each hrr: is injective.

(b) Let 8 l> X be a space of germs. A subspace of germs is a pair (313), where 3' is a
space of germs over X and j : 3' -> 8 is an injective morphism.

Proposition 7.5.14. (a) Let h : 80 -> 81 be a morphism between two spaces of germs over
X . Then Algol 3 X is a space of germs over X called the image of h. It is denoted by
lm h, and it is a subspace of 81.

Exercise 7.5.15. Prove the above proposition.

Lemma 7.5.16. Consider Iwo sheaves S and 'J', and let h : S -> 'T be a rnorpnism. Then h
induces a morphism between the associated spaces of germs h : S -> 'J'.

The definition of /E should be obvious. If f 6 So, and x 6 U, then he) = [h(fl]m,

where h ( f ) is now an element of 'J'(U). We let the reader check that l is independent of
the various choices, and that it is a continuous map S -> 'ii with respect to the topologies
described in Example 7.5.6.
The sheaf associated to the space of germs lm /E is a subsheaf of 'T called the image of
h, and denoted by lm h.

Exercise 7.5.17. Consider a morphism of sheaves over X, h : S -> 'J'. Let U C X be

an open set. Show that a section g 6 'J'(U) belongs to ( l m h)(U) if an only if, for every
as 6 X, there exists an open neighborhood Vx C U, such that g I v h ( f ) , for some
f 6 S(V).
 Cohomology 317

" D u e to Proposition 7.5.8(a), in the sequel we will make no distinction between sheaves
and spaces of germs.

Definition 7.5.18. (a) A sequence of sheaves and morphisms of sheaves,

-> sn 3
in
Sn-I-1 41 1
Sn-l-2 -> 7

is said to be exact if lm hn - her hn+1, 'v'n.

(b) Consider a sheaf S over the space X. A resolution of S is a long exact sequence

0 +5+80@>51@>~-~->s"s8H+1->

Exercise 7.5.19. Consider a short exact sequence of sheaves

0 - > S _ 1 ->S0->S1 ->0.
For each open set U define S(U) = S0(U)/S_1(U).
(a) Prove that U +-> S(U) is a presheaf.
~
(b) Prove that 81 = S = the sheaf associated to the presheaf S.

Example 7.5.20 (The DeRham resolution). Let M be a smooth n-dimensional manifold.

Using the Poincaré lemma and the Exercise 7.5.17 we deduce immediately that the se-
quence
d d d
0 <->RM<->Q0M->Q}W-> ->Q"M-> 0
is a resolution of the constant sheaf Be, mM denotes the sheaf of k-forms on M while d
denotes the exterior differentiation.

7.5.2 beef eohomology

Let U »-> S(U) be a presheaf of Abelian groups over a topological space X. Consider
an open cover u = ( U S A o f X . A simplex associated to 'LL is a nonempty subset
A = { O d 0 , . . . , C ¥ q C A , such that
q

UA UOH ¢ (D.
II

We define d m A to be one less the cardinality of A, dim A III 1. The set of all
q-dimensional simplices is denoted by Nq(U). Their union,

U NG (U)
q

is denoted by N(U), and it is called the nerve of the cover.

For every nonnegative integer q we set Aq := {0, . . . , q}, and we define an ordered
q-simplex to be a map O' : Aq -> A with the property that o(Aq) is a simplex of N(U),
possibly of dimension < q. We will denote by 31
(U) the set of ordered q-simplices, and
318 Lectures on the Geometry of Manzfolds

we will use the symbol Q07 • • • 705q to denote an ordered q-simplex 0 : Aq -> .A such that
0(k) Oak.
Define
Cq(S, U) H
ueJ<rq(u)
So, 80- S(Uo(Aql).

The elements of QQ (S, U) are called Cech q-cochains subordinated to the cover U. In other
words, a q-cochain C associates to each ordered q-simplex O' an element 0) 6 So.. ac,
We have face operators
of = of : Mu) -> J<rq_1('u), J= 0, 7 q,
where for a ordered q-simplex O' = (a0,...,aq),we set
$i0: 3gO'= lQ0)... ,Of 7 QQ) 6 U(<1-1) 7

and where a hat "Asa indicates a missing entry.

Exercise 7.5.21. Prove that aq_aq

i j . . j-1 i
- a q _ l a q , for all j > i .

We can now define an operator

5 : cQ-1<8,u> -> C'q(S,U),
which assigns to each (q 1)-cochain c, a q-cochain 50 whose value on an ordered q-
simplex O' is given by
q
(50, 0) (-1>j<0, 8j(7) In,
..II

j 0

Using Exercise 7.5.21 above we deduce immediately the following result.

Lemma 7.5.22. 62 0 so that

6 6 6
0 --> c0(8,u) i> c1(8,u> -> -> Cq(S, U) ->
is a cochin complex.

The cosmology of this cochain complex is called the tech cosmology of the cover
U with coefficients in the pre-sheaf S.

Example 7.5.23. Let U and S as above. A 0-cochain is a correspondence which associates

to each open set Uo, 6 U an element co 6 S(Uo). It is a cocycle if, for any pair (Q, B) such
that Ua n B as 0), we have

CB Co 0.
Observe that if S is a sheaf, then the collection (ea) determines a unique section of S.
A 1-cochain associates to each pair (Q, B) such that Uog go (D an element
Cog E 5l(/Vo5l.
 Cohomology 319

This correspondence is a cocycle if, for any ordered 2-simplex ( 0 0 / ) we have

i 0.
For example, if X is a smooth manifold, and U is a good cover, then we can associate to
each closed 1-form w 6 Q1 (M) a tech 1-cocycle valued in BX as follows.
First, select for each Ua a solution fa 6 C°°(Ua) of
dfa w.
Since d(fO - few E 0 on Uo5, we deduce there exist constants Cog such that fa - f B = cog .
Obviously this is a cocycle, and it is easy to see that its cohomology class is independent
of the initial selection of local solutions to. Moreover, if w is exact, this cocycle is a
coboundary. In other words we have a natural map
Hl(Xl -> H1(n<u),RX).
We will see later this is an isomoiphism.

Definition 7.5.24. Consider two open covers U qUa¢)aEA and V = (V5)B€Q8 of the 9

same topological space X. We say V is finer than U, and we write this 'LL -< V, if there
exists a map Q : 8 -> A, such that
VI C UQU3) WE G B.
The map Q is said to be a rejinement map.

Proposition 7.5.25. Consider two open covers 'LL = (U6)0,€A and V = (V5)568 of the
same topological space X such that U -< V. Fix a sheaf of Abelian groups S. Then the
following are true.
(a) Any refinement map Q induces a coe fain morphism

Q* Cvq(U,Sl -> Cvq(\7,S).

q q

(b) If : A -> 8 is another refinement map, then Q* is coe rain nomotopic to v* . In

T'
particular, any relation U -< V defines a unique morphism in eonomology
eX : H°l'LL,5l -> H. iv,sl.
(C)IfU -< V -< W , then

23" = o y,
Proof. (a) We define Q* : C<1(u) -> Cq(\7) by

so) 9 (@*(@),0) :i (C, @(al) Iv 'v'c E cow) O' E \7(q)),

where, by definition, 9(0) 6 As is the ordered q-simplex (9(B0), . . . , Q(Bq)). The fact that
Q* is a cochain morphism follows immediately from the obvious equality

Qo<9§=<9§o@-
320 Lectures on the Geometry of Manzfolds

(b) We define he : Nq_1(vl -> Mu) by

hj(50, • • 7 q-1) I lQl@0)) • • • 7 @(5jl»7°(@j)» ' • ' (/5'q-1))~
The reader should check that hj(o) is indeed an ordered simplex of U for any ordered
simplex 0 of V. Note that V0' C UhJ(o) W. Now define

X == Xi : cow) -> cQ-1(v)

by
q 1

( X q l C l v ft) (-w(@, h ) IV 'v'c 6 cow) VI 6 V01-1)°

j 0

We will show that

6 o Xqlcl + Xq+1 o 5(0) Z Q* (c) T* (c) 'v'c E C"1(U)-

Let O' (50, ,0q) 6 Nqlvl, and set
@(0) IAO)---)}'q1» r(0) Z (n0.-,uq) €U(q),

so that,
I7/jlo'l = (\0»~ .,/\j,/-Lj,. ..,/in).
Then

(x O 5(c),0) (-1)j(50, he((f)> Iv,

q q+1

(-1)J' Z( l )'"(c» 9é"+1h,(@)) Iv,

j 0

q j

(-1>i (-1l1"(0» (As, : j a w\j,llj, ,llql) Iv,

j 0 k 0

q
-|- l-1)Z_.l_1(C, (Au, w\j»llj, »lik, 7 Nql) Iv
€=j

(-w (-1l/°(0» (As, : p a 7 / \ j 7 / j ? )la»ql) Iv,

j 0 k 0

q
-I- l-1l@-}-1(07 I Q , 7 ) \ j 7 / / j 7 »lik, 7 /Jql) Iv
e=j+1

q .
-|- (-1)"{ ac, ( m • :Aj-lafUja ),Uql) IV
j=0

+(c, ( m . . . 7 * j » I'Lj-!r-17 7 llql) Iv

 Cohomology 321

The last term is a telescopic sum which is equal to

(C, (}\0 v • • -,As Iv -(C, ILLQ, . . 7 l L q l )

• lo= (Q*c,0) (1°*c,0l.
If we change the order of summation in the first two term we recover the term (-5Xe,o).
Part (c) is left to the reader as an exercise.

We now have a collection of graded groups

{ H°(U,S) ; U - open cover of X },
and morphisms

,Qi . H,lib 5) _> u '< V

V) 5%
H. I
7
•

such that,
u W
II and Zu W 'V
0?/LL,
Zu 3 ? / V

whenever U -< V -< W. We can thus define the inductive (direct) limit
H°(x, go : lim H°('L(, S).
u
The group H' (X, S) is called the beef cosmology of the space X with coefficients in the
pre-sheaf S.
Let us briefly recall the definition of the direct limit. One defines an equivalence relation
on the disjoint union

H I-I'(U,S),
'II

by
H. > 2w>'L1,v= @uWf=@9.
(U)9f g 6 I-I'(\7) is

We denote the equivalence class of f by j. Then

iium H' (up r

Note that we have canonical morphisms,

nu : H' (U, so -> H. (x,s).

Example 7.5.26. Let S be a sheaf over the space X. For any open cover 'LL = (Up)» a
0-cycle subordinated to U is a collection of sections f 6 S(Ua) such that, every time
Ua re UB go (Z), we have

JU0 IU0BI f 5IUo0

According to the properties of a sheaf, such a collection defines a unique global section
f 6 S(X). Hence, H0(X,S) = S(X).
~
322 Lectures on the Geometry of Manzfolds

Proposition 7.5.27. Any morphism of pre-sheaves h : S0 -> 51 over X induces a mor-

phis in cosmology
H. H' (X, 81).
h* (X>50) -)

Sketch of proof. Let U be an open cover of X . Define

h : cq<u,50> -> c'1<u,s1),

by
I/z*c,0) = ho(lc,o)) Vc 6 Cvq(U,S0) 07 6 U(q).

The reader can check easily that in* is a cochain map so it induces a map in cosmology
Hui H• (to) -> H°(U, 81)
which commutes with the refinements eX. The proposition follows by passing to direct
limits.

Theorem 7.5.28. Let

p
0 - > 8 _ 1 l>S0->S1 ->0,
be an exact sequence of sheaves over a paracompact space X . Then there exists a natural
long exact sequence
J* 6*
->H<1(x,8_1) -> H<1(x,80) is H<1(x,81) -> H'1+1(x,8_1> -> .

Sketch of proof. For each open set U C X define S(U) := S0(U)/S_1(U). Then the
correspondence U »-> S(U) defines a pre-sheaf on X. Its associated sheaf is isomorphic
with 81 (see Exercise 7.5. 19). Thus, for each open cover U we have a short exact sequence
0 -> cq(u,s_1l l> c'1<u,50> l> co(u,s) -> 0.
We obtain a long exact sequence in cosmology
-> H<1<u,8_1> -> Hq(u,50) -> H'1('u,5> -> H+11(u,x> ->
Passing to direct limits we get a long exact sequence
-> Hq(X,S_1) -> Hq(X,S0) -> Hq(X,S) -> Hq-l-1(x,5_1) ->
To conclude the proof of the proposition we invoke the following technical result. Its proof
can be found in [ l 19].

Lemma 7.5.29. If two pre-sheaves S, S' over a paracompact topological space X have
isomorphic associated sheaves then
H.
(x,5) 'E H ' ( x , s'
Definition 7.5.30. A sheaf S is said to be fine if, for any locally finite open cover U
(UQMQA there exist morphisms ha : S -> S with the following properties.
 Cohomology 323

(a) For any $ 6 .A there exists a closed set CO( C UQ, and ha(sas) = 0 for ac Q Co, where
So denotes the stalk of S at cz: 6 X .
(b) Za ho is. This sum is well defined since the cover U is locally finite.

Example 7.5.31. Let X be a smooth manifold. Using partitions of unity we deduce that
the sheaf of of smooth k-forms is fine. More generally, if E is a smooth vector bundle
over X then the space Q; of E-valued k-forms is fine.

Proposition 7.5.32. Let S be a fine sheaf over a paracompact space X . Then HQ (X, S) r\/

Oforq > 1.

Proof. Because X is paracompact, any open cover admits a locally finite refinement. Thus,
it suffices to show that, for each locally finite open cover U = l U a l a € A » the cohomology
groups HQ (U, S) are trivial for q > 1. We will achieve this by showing that the identity
map C'q(U, S) -> C"I(U, S) is cochain homotopic with the trivial map. We thus need to
produce a map
Xi : 0<1(u,s> -> cQ-1(u,8l,
such that
XC]-I-15q -|- 5q-1Xq = JL (7.5.l)
Consider the morphisms ho, : S -> S associated to the cover U postulated by the definition
of a fine sheaf. For every a 6 A, O' 6 Nqwl, and every f 6 C'q(U,S), we construct
Ito, ( f ) , of) 6 sl(/'O'l as follows. Consider the open cover of Uo-
{ V = U , m Uo, w ==
uO' \ Co, }, (suppla C ca).
Note that haf lvfww= 0 and, according to the axioms of a sheaf, the section ha(f IV) can
be extended by zero to a section (t0,(f), 0) 6 s(UO'). Now, for every f 6 C'q(U, S), define
xqf 6 C'q-1(U,S) by
..II

(Xqlfl) 0) <ia(f), 0)-

a
The above sum is well defined since the cover 'LL is locally finite. We let the reader check
that Xi satisfies (7.5.l).

Definition 7.5.33. Let S be a sheaf over a space X . Amine resolution is a resolution

d
0 ->8+50i>s1-> 7

such that each of the sheaves Si is fine.

Theorem 7.5.34 (Abstract DeRham theorem). Let

dl
0-->S->S0>S1->-~
be a jine resolution of the sheaf S over the paracompact space X . Then
0 -> s0<x> $1 51(X)
d

is a coe fain complex, and there exists a natural isomorphism

~
=
He(X,S) Hq(5q(X)l~
324 Lectures on the Geometry of Manzfolds

_ 1, denote
Proof. The first statement in the theorem can be safely left to the reader. For q >
by Zn the kernel of the sheaf morphism do. We set for uniformity Z0 := S. We get a short
exact sequence of sheaves

0 ->Z"->Sq->Z"+1-> 0 q 2 0 . (7.5.2)

We use the associated long exact sequence in which Hk (X, sql 0, for k > 1, since Sq is
a fine sheaf. This yields the isomorphisms
~
H k - 1 ( x , z<1+1> = H'"(x, Zn) If 2 2.
We deduce inductively that
~=
H71(x, 20) Hl(x, am-ll m > 1 . (7.5.3)

Using again the long sequence associated to (7.5.2), we get an exact sequence

H0(x, zm-1) dL H0(x, am) -> H1(x, am-ll -> 0.

We apply the computation in Example 7.5.26, and we get
H1(x, am-ll n
o
z,"*(xl/dm-1 liM-IIXII

This is precisely the content of the theorem.

Corollary 7.5.35. Let M be a smooth manifold. Then

iv H.
H. (M, Run ) (M)-
Proof. The manifold M is paracompact. We conclude using the fine resolution

0->RM<->Q0Mi>Q}W->

Exercise 75.36. Describe explicitly the isomorphisms

~=
HW) H1(m,Bm) and H2(M) 2 H2(m,Bm).
Remark 7.5.37. The above corollary has a surprising implication. Since the Cech coho-
mology is obviously a topological invariant, so must be the DeRham cosmology which is
defined in terms of a smooth structure. Hence if two smooth manifolds are homeomorphic
they must have isomorphic DeRham groups, even if the manifolds may not be diffeomor-
phic.
Such exotic situations do exist. In a celebrated paper [93], John Milnor has constructed
a family of nondiffeomorphic manifolds all homeomorphic to the sphere S7_ More recently,
the work of Simon Donaldson in gauge theory was used by Michael Freedman to construct
a smooth manifold homeomorphic to R4 but not diffeomorphic with R4 equipped with the
natural smooth structure. (This is possible only for 4-dimensional vector spaces!) These
three mathematicians, J. Milnor, S. Donaldson and M. Freedman were awarded Fields
medals for their contributions.
 Cohomology 325

Theorem 7.5.38 (Leray). Let M be a smooth manifold and U IUoQQA a good cover
ofM, i.e.,
dim M
O'

Then
f\J H.
Ho (U, Be ) (m).
Proof. Let Zn denote the sheaf of closed k-forms on M. Using the Poincaré lemma we
deduce
2'"(u;) = dak'-1lUO').
We thus have a short exact sequence
0 -> cqlu, zkl -> Cow, Qk-1) c'1(u, z'*) -> 0.
Using the associated long exact sequence and the fact that Qk-l is a fine sheaf, we deduce
as in the proof of the abstract DeRham theorem that
~=
H"(U,Bml HQ-1(u,z,1>
~= ~ H. (ml.
iv
H1(u, z,<1-1) H0(u, zq>/d*H0(u, QQ-1) = zq(m)/dszq-1(m) iv

Remark 7.5.39. The above result is a special case of a theorem of Leray: if S is a sheaf on
a paracompact space X, and 'LL is an open cover such that
_l A
Hqlz/'A,s) = 0, vi > 6 elul,
then I-I'(U, S) = H' (x,8>. For a proof we refer to [52].
When QM is a constant sheaf, where G is an arbitrary Abelian group, we have a
Poincaré lemma (see [42], Chapter IX, Thm. 5.1.),
I-Iq(]R"',Q'_)=0 q > 1 ,
Hence, for any good cover U
H.
H.(m,Qm)
Remark 7.5.40. A combinatorial simplicial complex, or a simplicial scheme, is a collection
UC of nonempty, finite subsets of a set V with the property that any nonempty subset B of a
subset A 6 fK must also belong to the collection UC, i.e.,
A ex, B CA, B ;é(D-» B ex.
The set V is called the vertex set of UC, while the subsets in U( are called the (open) faces)
of the simplicial scheme. The nerve of an open cover is an example of simplicial scheme.
To any simplicial scheme UC, with vertex set V, we can associate a topological space
l5<l in the following way.
Denote by RV the vector space of all maps f : V -> R with the property that f (Fu) 0,
for all but finitely many 'u's. In other words,
RV R.
'UGV
326 Lectures on the Geometry of Manzfolds

Note that RV has a canonical basis given by the Dirac functions

1 u 'U
61, : V -> R, 5U,u)
0 u ¢ 'U.

We apologize RV by declaring a subset C C RV closed if the intersection of C with any

finite dimensional subspace of RV is a closed subset with respect to the Euclidean topology
of that finite dimensional subspace.
For any face F C UC we denote by As C RV the convex hull of the set {6'U§ '06 F }.
Note that As is a simplex of dimension IFI - 1. Now set

l5<l U AF.
FGUC

We will say that l9<l is the geometric realization of the simplicial scheme UC.
A result of Borsuk-Weil (see [$9]) states that if U = (UoJQQA is a good cover of a com-
pact manifold M, then the geometric realization of the nerve elul is homotopy equivalent
to M. Thus, the nerve of a good cover contains all the homotopy information about the
manifold. Since the DeRham cosmology is homotopy invariant, Leray's theorem comes
as no suprise. What is remarkable is the explicit way in which one can extract the cohomo-
logical information from the combinatorics of the nerve.
While at this point, we should remark that any compact manifold admits finite good
covers. This shows that the homotopy type is determined by a finite, albeit very large, set
of data.

Example 7.5.41. Let M be a smooth manifold and 'LL = (UoJQQA a good cover of M. A
1-cocycle of RM is a collection of real numbers fog - one for each pair (Q, B) 6 \A2 such
that U05 75 (Z) satisfying

fem + far + fro 0,

whenever UQBW ¢ (Z). The collection is a coboundary if there exist the constants fa such
that fan = f B - fa- This is precisely the situation encountered in Subsection 7.1.2. The
abstract DeRham theorem explains why the Cech approach is equivalent with the DeRham
approach.

Remark 7.5.42. Often, in concrete applications it is convenient to work with skew-

symmetric Cech cochains. A cochain C 6 Cq(S, up is skew-symmetric if for any ordered
q-simplex O' = (to, . . . , aq), and for any permutation up of Aq we have

(c, (0407 7 Qql) i €((/>)(c» l&(,O(0l: 7 0£(f)(<I)l)°

One can then define a "skew-symmetric" Cech cohomology following the same strategy.
The resulting cosmology coincides with the cohomology described in this subsection. For
a proof of this fact we refer to [119].
 Cohomology 327

Exercise 7.5.43. A sheaf of Abelian groups S on a paracompact space X is called soft if

for every closed subset C C X the restriction map S(X) -> S(C) is surjective. The sheaf
S is called flabby if, for any open subset U C X, the restriction map S(X) -> S(U) is
surjective.
(a) Prove that any flabby sheaf is also soft.
(b) Suppose 0 -> 80 -> 51 -> 82 -> 0 is a short exact sequence of sheaves on X. Show
that if 50 is a flabby sheaf, then the sequence of Abelian groups
0 ->S0(X) ->S1(X) ->S2(X) -> 0
is exact.

Exercise 7.5.44. Suppose X is a topological space and S is a sheaf of Abelian groups on X .

(a) Prove that S admits a flabby resolution, i.e., there exists a sequence of flabby sheaves
over X, So, k > 0, and morphisms of sheaves fk : 5 k - 1 -> So, k . 0, 1, . . . , 5-1 = S,
such that the sequence below is exact

0->S I0>80 h > 8 1 >

(b) Prove that if

0 - > S I0>80 f1>81 >

is a flabby resolution of S, then the cosmology groups Hk:(X, S) are isomorphic to the
cosmology groups of the cochain complex

S@(Xl to>sllxl >

This page intentionally left blank
 Chapter 8

Characteristic Classes

We now have sufficient background to approach a problem formulated in Chapter 2: find a

way to measure the "extent of nontriviality" of a given vector bundle. This is essentially a
topological issue but, as we will see, in the context of smooth manifolds there are powerful
differential geometric methods which will solve a large part of this problem. Ultimately,
only topological techniques yield the best results.

8.1 Chem-Weil theory

8.1.1 Connections in principal G-bundles

All the Lie groups we will consider will be assumed to be matrix Lie groups, i.e., Lie
subgroups of a general linear group GL(n, K) = GL(K"').
This restriction is neither severe, nor necessary. It is not severe since, according to a
nontrivial result (Peter-Weyl theorem), any compact Lie group is isomorphic with a matrix
Lie group, and these groups are sufficient for most applications in geometry. It is not
necessary since all the results of this subsection are true for any Lie group. We stick with
this assumption since most proofs are easier to "swallow" in this context.
The Lie algebra g of a matrix Lie group G is a Lie algebra of matrices in which the
bracket is the usual commutator.
Let M be a smooth manifold. Recall that a principal G-bundle P over M can be defined
by an open cover (Url of M and a gluing cocycle
gag : U0g -> G.
The Lie group G operates on its Lie algebra g via the adjoint action
Ad : G -> GL(g), g »-> Ad(g) E GL(g),
where
1
Ad(glx :: gxg 7 VX E g, g E G.
We denote by Ad(P) the vector bundle with standard fiber g associated to P via the adjoint
representation. In other words, Ad(P) is the vector bundle defined by the open cover (Ua),
and gluing cocycle
Adlgog) : UQ5 -> GLIEI.

329
330 Lectures on the Geometry of Manzfolds

The bracket operation in the fibers of Ad(P) induces a bilinear map

1-, .1 : Q'*(Ad(p) ) X Q'(Ad(p>) -> Qk+"(Ad(p) ) 7

defined by
k
> V @ (X) Y]
w (X) (wk A s ) (X) [X,y], (8.l.1)
"( Y
for all wk Q Qklml, 77 2Q Q and X
7 Q Q0( Ad(p)).
7

Q*
Exercise 8.1.1. Prove that for any inv dlpl) the following hold.
lA l[77
-Q I.l?7
[(»J,,,] = (-llloJ (8.l.2)
"*J],
[ |¢|[
7]7] - )|w|
.
[[W77 - [w, (I5],77] -|- l - l
We [WI 1. (8.l.3)

In other words, Q'I I Ad(P) ), [ , ] ) is a super je algebra.

Using Proposition 3.3.5 as inspiration we introduce the following fundamental concept.

Definition 8.1.2. (a) A connection on the principal bundle P defined by the open cover
"LL l e a ) a € A 9 and the gluing cocycle .Quo : Uog -> G is a collection
2

AA 6 Q1(U> (X) Q,

satisfying the transition rules

Aglilil g 5(w)d g@(w ) + g ; ( x ) A ( w ) g @ ( x )

- ( dg@a(:u) log 1 (so) + g@a(2¢)A0(w)g5 1 (Lil, Vcc E Uog.
We will denote by A(U, 9-») the set of connections defined by the open cover U and the
gluing cocycle g., U.. -> G.
(b) The curvature of a connection A 6 All, 9--) is defined as the collection Fa E
Q2(Ua) (X) g where
l
Fa ¢ZAo, -2[A O A o ,]

Remark 8.1.3. Given an open cover (Ua) of M, then two gluing cocycles
g,B0 a h e : U0g G,
define isomorphic principal bundles if and only if there exists smooth maps
Ta:Uo->G,

such that

h@o(¢ul Z T@(t)9@0(@@)T0(1'l'1> ,'v'0,B, Van E Uo0.

If this happens, we say that the cocycles g., and h., are cohomologous.
Suppose (gag), (hog) are two such cohomologous cocycles. Then the maps To induce
a well defined correspondence
'T : A('L(, g .. ->A('L(,h ..
 Characteristic Classes 331

given by

A("L(, 9--) EIAaI I

u'
> ( Bo == - ( d T ) T 1 + TAOlTa 1) E Am, h.. .
The correspondence T is a bisection.
Suppose 'Ll = (U0,)o€A and V = (Vz l i € I are open covers, and QBQ : Uog -> G is a
gluing cocycle. Suppose that the open cover l v i l i € I is finer that the cover q U a ) a € A 9 i.e.,
there exists a map (p : I -> A such that
C U(f)(i)) W E I.
We obtain a new gluing cocycle Q; : Vi -> G given by
ac'

This new cocycle defines a principal bundle isomorphic to the principal bundle defined by
the cocycle (gg0). If (Aa) is a connection defined by the open cover (UOf) and the cocycle
lg5al» then it induces a connection
Ag' Ago(i) Iv
defined by the open cover (W), and the cocycle gin. We say that the connection Aw is the
V-refinement of the connection A. The correspondence

'Tau :A(U»g--) -> A(\7,gCD

so 7
A+->ALf'
is also a bijection.
Finally, given two pairs (open cover, gluing cocycle), (U, g,.) and (V, h,,), both de-
scribing the principal bundle P -> M, there exists an open cover W, finer that both U and
W, and cohomologous gluing cocycles in, B.. : W., ->G refining g.. and respectively
h.,, such that the induced maps
Aw, 9--) >A(W, §..) >A(W,l .. < Aw, h..
are bijections.
For simplicity, we will denote by A(P), any of these isomorphic spaces A(U, on).
" A l l the constructions that we will perform in the remainder of this section are com-
patible with the above isomorphisms but, in order to keep the presentation as transparent
as possible, in our proofs we will not keep track of these isomorphisms. We are convinced
that the reader can easily supply the obvious, and repetitions missing details.

Proposition 8.1.4. Suppose P -> M $5 a principal G-bundle described by an open cover

'Ll = l U a l a 6 A 9 and gluing cocycle go, : UQ5 -> G.
(a) The set A(P) of connections on P i5 an a]§"ine space modeled by Q1( Ad(P) l.
(b) For any connection A = ( A ) Q A(U, g,,l, the collection (Fa) dejines a global
Ad(P)-valued 2-form. We will denote it by FA or F(A), and we will refer to it as the
curvature of the connection A.
(c) (The Bianchi identity.)
dFa -|- [ A O F ] = 0, Va. (8.1.4)
332 Lectures on the Geometry of Manzfolds

Proof. (a) If IAQIQQA, (Bo»)¢»€A E A(P), then their difference Ca Ao Bo satisfies

the gluing rules

CB QBQCQQQ 17

so that it defines an element of Q1(Ad(p)). Conversely, if (Aa) E A(P), and w E

Q1 (Ad(p) ), then CU is described by a collection of g-valued 1-forms to 6 Q1(ual (X) g,
satisfying the gluing rules

°JBIUo,B I gpowoluaggg 1.
The collection A' := Aa -|- wa is then a connection on P. This proves that if A(P) is
nonempty, then it is an affine space modeled by O1( Ad(P) l.
To prove that .A(P) go U) we consider a partition of unity subordinated to the cover
UQ- More precisely, we consider a family of nonnegative smooth functions uO! : M -> R
a E A, such that supp ua C U Va, and

Up 1.
a

For every a E A we set

-1 1
..II

Bo u g d g w 6 Q (UQ) ® 9-
'7'

Since g = ga we deduce g 1dg = - ( d g a ) g 0 , so that

-1
Ba Uvldgavlga»'y '
'Y

Then, on the overlap Ugh, we have the equality

- _1
BB 95Q8095 1 i 2 U"lyldgB"ylg,B» + 2 'U"'7g,50 (d907)90795 1°
'Y 'Y

The cocycle condition implies that

d907 = (dgaalgm -|- glhldgavl)

so that

( dg5'ylg5-'y1 -- ldgWgw = l d g B a l g o @ -|- g5aldga)gB-

Hence

Uvld9Bv)9g Z -ld9B01l9a¢B UWQBQ (¢19ow)9w0 -

'Y 7

Using the cocycle condition again, we deduce 9O 90 1 = 9'yB» so that

_1 _
U'ygBald9a'y)9o¢95 1 U'ygBa ld9awl9vB -
'y 'y

Hence, on the overlap U's we have

>9_1 + Q _1 -1
BB ldgga QBQ
QQ QQBCZQQB -|- QQBBQQQ5'
go
 Characteristic Classes 333

This shows that the collection (Bo) defines a connection on P.

(b) We need to check that the forms Fo satisfy the gluing rules
_
FB - 9
-1
For
7

where g QoB = 9 g 1 - We have

1
FB = CMB -|- 51AB» AB]
1
= d(9-1d9 + Q'1AQ9) + §[9-ldg + g-1AQ9, 9-1d9 + 9-1Aog1-
Set w g-1dg. Using (8.1.2) we get
1
FB = do + §[W7 W]
-1 -1 1 -1 -1
+ d(g Aug) + [mg Aw] + 519 A y Aw]- (8.l.5)

We will check two things.

A. The Maurer-Cartan structural equations.
1
dw-|- §[w,w] = 0.

B.

d(gllAag) + [w, 9'1Ao9l = g`1(dAo)g-

Proof of A. Let us first introduce a new operation. Let M, K) denote the associative
algebra of K-valued n X n matrices. There exists a natural operation
/\ : Q'"(u> ® m, K) x M U ) ® m, K) -> Qk+'lUo,) ® Q , K),
uniquely defined by
(wk®A) A (ne®B) =( J A ) ®(A-Bl, (81.6)
where wk' 6 Q":(U0,), 17e E Qt((/o) and A, B 6 (n,Kl (see also Example 3.3.l2). The
space M, K) is naturally a Lie algebra with respect to the commutator of two matrices.
This structure induces a bracket

l7.. 1.§M(u al®

K
QIIlf '
)X Q ) SEIne
(X)
k+
2 ,
Q)
(X) (n,Kl In
defined as in (8.1.1). A very simple computation yields the following identity.
l
w / \ 77 = 5lW»*7l We E QUO) (X) QW Kl. (8.1.7)

The Lie group lies inside GL(n, K), so that its Lie algebra g lies inside , K). We can
think of the map QoB as a matrix valued map, so that we have

do = d(glldg) = (d9'1) /\ do
= -(9-1 -do g-1)d9 = -(9-1d9) /\ (g-1d9)
w A w(8.1.7) 1
- §[W»W]-
334 Lectures on the Geometry of Manzfolds

Proof of B. We compute
d(g'lAo9) = (d9llAo) ~g + 9-1ldAa¢l9 + g'1Aod9
= -9-1 - do - 9-1 /\ As ' Q + g`1(dAo)g + (g'1Ao9) /\ g'ldg
= - w A g - y + g - M g A w + 9-1ldAa¢l9
(8.1.7) 1 1
= - §[W»9 Mag] + §[W»91Ao9] + 9 1(dAo)9

(81.2) - [w',9'1Ao9] -|- g"(dAo)9-

Part (b) of the proposition now follows from A, B and (8.1.5).
(c) First, we let the reader check the following identity

d[(»Jn7] = [dim] + (-1) '"'[w» dt1]» (8.l.8)

where w,17 E Q*(UO) (X) g. Using the above equality we get
1 lg.1.2l[dAa As]
d(Fa) i §{[dA0,AO,] [A<»dA0]}
1
Z [FoHAQI - 5 uA,Ao1,A1 (8 .1 .3 ) IFA1.
The proposition is proved.

Exercise 8.1.5. Let to, E Q w v ) (X) g satisfy the gluing rules

-1
We = QBQWQQQQ on Uo0.
In other words, the collection to, defines a global k-form w E Q"'(Ad(P))- Prove that the
collection
dw0 + [Am Was
defines a global Ad(P)-valued (k + l)-form on M which we denote by lAw. Thus, the
Bianchi identity can be rewritten as dAF(A) = 0, for any A E A(P) .

Remark 8.1.6. Suppose P -> M is a principal G-bundle given by the open cover U =
(Ua) and gluing cocycle QBQ : Uog -> G. A gauge transformation o f P is by definition, a
collection of smooth maps
Ta:Uo->G

satisfying the gluing rules

T5(ac) = 90,()T(g3)95(x)-1, Vi E USB.
Observe that a gauge transformation is a section of the G-fiber bundle
G <-> C(P) > M,
(see Definition 2.3.15) with standard fiber G, symmetry group G, where the symmetry
group G acts on itself by conjugation
G X G 9 (9,/2) I " > c g ( h ) :z 9119. Q G.
 Characteristic Classes 335

The set of gauge transformations forms a group with respect to the operation

(SO) ' (Ta) :z ISQTQI-

We denote by §(U, g ..this group. One can verify that if P is described the (open cover,
gluing cocycle)-pair (V, h ».) , then the groups §("Ll, g..I and 907% h ».) are isomoiphic.
We denote by glpl the isomorphism class of all these groups.
The group §('L[, g ..acts on A('Ll, g.. according to

9(p) X A(p) 9 (T, A) »-> TAT-1 ¢ = ( - ldT)Ta 1 -|- T A a T 1 ) Q A(Pl.

The group glpl also acts on the vector spaces Q°(Ad(p)) and, for any A 6 A(P),
T E glpl we have
FTAT-1 = TFAT-1
We say that two connections A0, A1 E A(P ) are gauge equivalent if there exists T 6 9(P)
such that A1 = TA0T-l.

8.1.2 G-vector bundles

Definition 8.1.7. Let G be a Lie group, and E -> M a vector bundle with standard fiber a
vector space V . A G-structure on E is defined by the following collection of data.

(a) A representation p : G -> GL(V).

(b) A principal G-bundle P over M such that E is associated to P via p. In other words,
there exists an open cover (Ua) of M, and a gluing cocycle gal? : Uog -> G, such that
the vector bundle E can be defined by the cocycle

p(9o@l : Uow -> G L W ) .

We denote a G-structure by the pair (P, p).
Two G-structures (Pi, p on E, i = l , 2, are said to be isomorphic, if the representa-
tions Pt are isomorphic, and the principal G-bundles P are isomorphic.

Example 8.1.8. Let E -> M be a rank T' real vector bundle over a smooth manifold M. A
metric on E allows us to talk about orthonormal moving frames. They are easily produced
from arbitrary ones via the Gramm-Schimdt orthonormalization technique. In particular,
two different orthonormal local trivializations are related by a transition map valued in the
orthogonal group OU), so that a metric on a bundle allows one to replace an arbitrary
collection of gluing data by an equivalent (cohomologous) one with transitions in O(rl. In
other words, a metric on a bundle induces an O(7") structure. The representation p is in this
case the natural injection O(r) <-> GL(1°, R).
Conversely, an O(7"l structure on a rank r real vector bundle is tantamount to choosing
a metric on that bundle.
Similarly, a Hermitian metric on a rank k complex vector bundle defines an um)-
structure on that bundle.
336 Lectures on the Geometry of Manzfolds

Let E = (P, p, V) be a G-vector bundle. Assume P is defined by an open cover ( U I ,

and gluing cocycle

QoB : U0g -> G.

If the collection { A a E Q1 ( U ) (X) g} defines a connection on the principal bundle P, then

the collection p*(AO) defines a connection on the vector bundle E. Above, p* : g ->
End(V) denotes the derivative of p at 1 E G. A connection of E obtained in this manner
is said to be compatible with the G-structure. Note that if F(A0,) is the curvature of the
connection on P, then the collection p* (F(AO)) coincides with the curvature F(p* (Ao))
of the connection p* (Aa).
For example, a connection compatible with some metric on a vector bundle is compati-
ble with the orthogonal/unitary structure of that bundle. The curvature of such a connection
is skew-symmetric which shows the infinitesimal holonomy is an infinitesimal orthogo-
nal/unitary transformation of a given fiber.

8.1.3 Invariant polynomials

Let V be a vector space over K R, C. Consider the symmetric power
sklv*l C (v*)®'" 7

which consists of symmetric, multilinear maps

9o:Vx--»xV->K.
Note that any SO E 5I<(V*) is completely determined by

PnpM (p('v, . . . Yu).

This follows immediately from the polarization formula

l gk
§0l'U1,...,'U]'Cl P t -I- + ikok).
/<!a¢1~~@¢, (f)(1U1

If dim V = n then, fixing a basis of V, we can identify sklwl with the space of degree k
homogeneous polynomials in n variables.
Assume now that A is a K-algebra with 1. Starting with up 6 Sk(V*) we can produce
a K-multilinear map

90AI(~/1 (X) V) X (A®V)->A,

S-
II

uniquely determined by

(P(611 (X) ' U 1 7 ° " 7 a k (X) url = (,0('u1,...,vk)a1a2--~ak1J EA.

If moreover the algebra A is commutative, then (PA is uniquely determined by the polyno-
mial

P<p(;v)=(0A(x1---,w) a¢eA (X) v;

 Characteristic Classes 337

Remark 8.1.9. Let us emphasize that when A is not commutative, then the above function
is not symmetric in its variables. For example, if CL10,2 = -a2a1, then

P(U»1X1,C12X22°'°) Pl0»2X2,U»1X1)'°')-

For applications to geometry, A will be the algebra Q ° ( M ) of complex valued differential

forms on a smooth manifold M. When restricted to the commutative subalgebra
Qeven
Q2klml (X) (c,
k>0

we do get a symmetric function.

Example 8.1.10. Let V _

gl(n, (c). For each matrix T E V we denote by ck(T) the
coefficient of Ak in the characteristic polynomial

Co,lTlAk, (i=VI>.
k>0

Then, Co ( T ) is a degree k homogeneous polynomial in the entries of T. For example,

n
1
tr T, cn(T) = - 27r'£ det T.
C1lT) I - 27ri /

Via polarization, ck(T) defines an element of Sk(('rL, C)*).

If A is a commutative (C-algebra with 1, then .A (X) gilt, C) can be identified with the
space W, A) of n X n matrices with entries in A. For each T E (n,A) we have

det
( ]l 6 A[A],

and Co (T) continues to be the coefficient of As' in the above polynomial.

Consider now a matrix Lie group G. The adjoint action of G on its Lie algebra g
induces an action on Sk(g * still denoted by Ad. We denote by MG) the Ad-invariant
elements of Sk (g * . It consists of those (Q G S'"(Q*) such that
-1
S0l9X1g 7 7 9XI<:g-1) I S0(X1»---»XI<=)»

for all X17 7 Xi G g. Set

k>0 k>0

The elements of I°(G) are usually called invariant polynomials. The space I " ( G ) can
be identified (as vector space) with the space of Ad-invariant formal power series with
variables from g*.

Example 8.1.11.
(n,lC) 9 X »-> tr exp(X),
338 Lectures on the Geometry of Manzfolds

defines an element of I" (GL(n, (Cl). To see this, we use the "Taylor expansion"

1
€x1>(X) ZN
go
7

which yields

tr exp(X) Z 1 tIIX . k

/20

For each If, t r X k Q I'(GL(n,(c)) since

tr lgXg-1)k = trgXkg-l = t r X k .
Proposition 8.1.12. Letch 6 M G ) . Thenforany X , X l , . . . , X i E g we have

90([X)Xl]7X27' . . , X i ) -|- o o • +90(X17X2=°°°7[X7XI<:]) = 0. (8.l.9)

Proof. The proposition follows immediately from the equality

d -tx
llllllllll

(p(et X X16 7 7 € t x x k 6 - t x ) = 0.
dt t=0

Let us point out a useful identity. If P E Ik (g), U is an open subset of R", and

F = We ® x 6 MU) (X) 9, A w (X) X E Qd(U) (X) g

then

P(F1,...,Fi- »[A»Ftz]»F@+1~--»F1<:l

(-1)d(d1+~~~d_1)ww1 ' ' ' Wkplxl, ' ' ' ,[X, X , a • a Xi).

In particular, if F1, 7 Fe 1 have even degree, we deduce that for every l, . . . , k we

have

PlF1,. . ° a F i - 1 7 [A7F¢]7F¢+1». . .,Fol £41601 L<.JkP(X1,. . .7[X7X'i]v° . . ,Xkl.

Summing over 11, and using the Ad-invariance of the polynomial P, we deduce

ZHF1, 7 Fi-17 [A7 F¢]7F¢+17 . . .,Fol = 0, (8.1.10)

'Z 1

\V/F1,...,Fk_1 €Q€'Uen(Ul (X) 9» Fe, A E §2°(U) (X) 9-

 Characteristic Classes 339

8.1.4 The Chem-Weil Theory

Let G be a matrix Lie group with Lie algebra g, and P -> M be a principal G-bundle over
the smooth manifold M .
Assume P is defined by an open cover (UQ) and a gluing cocycle
QoB : U0g -> G.
Pick A E A(P) defined by the collection As, E Q1(Ua) (X) g. Its curvature is then defined
by the collection
1
Fa I dAa + [AORTAOf].
2
Given (Is E MG), we can define as in the previous section (with A Qeven (UP)»
V 9)
P<l>(Fal : (z5(Fo 7 7 F<.) E Q U ) .

Because (b is Ad-invariant, and FE = QBQFaQQ la we deduce

P¢(Fo) = PMFB) on U0g a

so that the locally defined forms P¢(Fal patch-up to a global 2k-form on M which we
denote by go(FA).

Theorem 8.1.13 (Chem-Weil). (a) The form q25(FA) is closed, VA 6 A(P).

(b) I f A 0 , Al 6 A ( P ) , then the forms q5(FAo l, and qb(FA1) are cohomologous. In other
words, the closed form qz5(FA) defines a chorology class in H 2 k ( M ) which is independent
of the connection A E A(P).
Proof. We use the Bianchi identity dFa =-[As»F€]. The Leibniz' rule yields
¢1¢(FO,, . . . , Fo) ¢ldFa»Faa---a-Fal+°'°+¢lFa)-°-aFa)dF0l

=-¢l[Aa»Fa]»Fa»--- Fa ) 7 ¢ ( F , . . . , F , [A Far (8.1.9)0

(b) Let As' E A(P) (i = 0, 1) be defined by the collections

A; 6 Q1<z/) ® g.
Set CO( := $1 - AO. For 0 S t < l we define At 6 Q l l y ) (X) g by At := A0 -|- tco.
The collection (At ) defines a connection At E A(P), and t »-> At E A(P) is an
(affine) path connecting A0 to A1. Note that

C (cal A
We denote by Ft = ( F t l the curvature of At. A simple computation yields
to
Ft = F0 -|- mica -|- [A07 CON) _|_ [ C , C ] .
2
_ (8.1.11)

Hence,
F.t = do + [A0 , ca] -|- ¢[c,,, c ] = doa -|- [At ,cO,].
340 Lectures on the Geometry of Manzfolds

Consequently,

<l>(F1) - ¢(F'")
/ F*
0
1
¢(F* , , ,Ftl-p + ¢(F' 7 7 F t , F t ) dt

{¢(¢zc;, , Ft F'*)+ (25(Ft 7 Ft , cO,>} it

/
7 -|- 7
0
t
.
/
-I-
0
'0a],Ft
Because the algebra Qeven (Ua) is commutative, we deduce
--,F)
-I- + (/5(F 7 7 Ft, [At ,ca])} do.

(25lW0(1)> ' • • aW0(k)l I

Cl5((»J17 . . . ,wk),
for all O' aSk and any w1, . . . , w k E Q€'Ue"(Ua) (X) g. Hence
1
(/)(F1)-¢(F°)=1< 0 <l)(Ft*, , F t , too, + [A cO,])dt.
We claim that
cl)(Ft,...,Ft,dC'o+[At,C'a]) = d ¢ ( F * , ,Ft,ca).
Using the Bianchi identity we get
d¢(F¢,.CC,F*,cO,>
(!)(Ft.»°°° , F t , d c O , ) + ¢ ( d F t , ~ - 7 Ft,C'al+ + ¢(Ft, ,dFt,C'a)
(I) Ft,...,Ft,dC'o
( t t )t t t t t t
-¢lCaa[AQ7Fo]7Fa)'°')Fa)-°°°-¢lCa7Fa)°'°)Fo7[AQ7Fa]l
¢(Ft,...,Ft,dcO,+[At,cO,])-¢(Ft,...,Ft,[A,ca]>
-(/)(lA )Ft])Ft9 . . . , F t , C ' 0 ) - - - - - ( 2 5 ( F t , . . . , F t , [ A t , F t ] , C ' O , )
(8.1.10)
i ¢(Ft»..- ,Ft,dC'O,+[At,Co]) = ¢ ( d C a + [ A t , C a ] , F t , . . . , F t ) .
Hence
1

¢(F1) - <l)(F0) d 1<¢(At,Ft, ,Ftldt. (8.l.12)

0
We set
1
T<,,(A1 , .AU ) 1¢(A* , F , ,Ftldt.
.•

0
Since $5 = Q B 0 C O Q Q 19 and FB = QBQFaQQ 1 on Uog, we conclude from the Ad-
invariance of QUO that the collection T l A 1 , A0 ) defines a global (Zn - 1)-form on M which
we denote by T,,(A1, A0), and we name it the Q5-transgression from A0 to A1. We have
thus established the transgression formula
€25(FA1) - <l)(FA0>) = dT,,(A1,A0). (8.l.13)
The Chem-Weil theorem is proved.

Remark 8.1.14. Observe that for every Ad-invariant polynomial QUO 6 Ik (g), any principal
G-bundle P, any connection A 6 A(P), and any gauge transformation T E §(P) we have
(PIFAI = (Zeal FTAT-1 >-
We say that the Chem-Weil construction is gauge invariant.
 Characteristic Classes 341

Example 8.1.15. Consider a matrix Lie group G with Lie algebra g, and denote by P0 the
trivial principal G-bundle over G, P0 = G X G. Denote by w the tautological 1-form
w = g-1dg E Q1 (G) (X) g. Note that for every left invariant vector field X E g we have

w(X) X.
Denote by d the trivial connection on PQ. Clearly d is a flat connection. Moreover, the
Maurer-Cartan equation implies that' d -|- w is also a flat connection. Thus, for any <25 E
Ik(g)
(25lFdl i ¢)(F(d+wl i 0.
The transgression formula implies that the form
1
T¢,=T¢ld'l'ZD,d)=]€ 0 (z5(w» Fd-ptw 7 7 Fd-l-twldt E Q2k-1lG)

is closed.
A simple computation using the Maurer-Cartan equations shows that
k
- d)(w» [w, we
't¢ Mk 1
(/ ltd
0
k
1)"' 1do
22k-1Ia(I@ - 1>!
7 lwwwl)
l 1)'" 1
2/<5-1 (2k)! ¢(w, [w, w], »[w»w])
Mk
l-1)k-1 (My, [777/']7 7 lwwwll-
(25)
We thus have a natural map T : I' (G) -> Hodd(G) called transgression. The elements
in the range of 'r are called transgressive. When G is compact and connected, then a non-
trivial result due to the combined efforts of H. Hopf, C. Chevalley, H. Cartan, A. Weil and
L. Koszul states that the cosmology of G is generated as an R-algebra by the transgressive
elements. We refer to [26] for a beautiful survey of this subject.

Exercise 8.1.16. Let G = SU(2). The Killing form H; is a degree 2 Ad-invariant polyno-
mial on Q). Describe T 6 QSIGI and then compute

T .
G

Compare this result with the similar computations in Subsection 7.4.3.

Let us now analyze the essentials of the Chem-Weil construction.

Input: (a) A principal G-bundle P over a smooth manifold M, defined by an open cover
(Url ,and gluing cocycle QQB : Uog -> G.
(b) A connection A E A(P ) defined by the collection

As, Q Q 1 (Ua ) (X) Q,

lThe connections d and d -|- w are in fact gauge equivalent.
342 Lectures on the Geometry of Manzfolds

satisfying the transition rules

AB = g a d g o g -|- 9o§A09o@ on UQ5.
(CW 6 I'"(G)-
Output: A closed form q5(F(A)) 6 Q2'*(M), whose cohomology class is independent of
the connection A. We denote this cohomology class by gz5(P).
Thus, the principal bundle P defines a map, called the Chem-Weil correspondence
Cap : raga -) H• (M) ¢*->(z5(P)-
One can check easily the map cwp is a morphism of R-algebras.

Definition 8.1.17. Let M and N be two smooth manifolds, and F : M -> N be a smooth
map. If P is a principal G-bundle over N defined by an open cover (UOf), and gluing
cocycle .9o15 : Uog -> G, then the pullback of P by F is the principal bundle F* (P) over
M defined by the open cover F-1(Ua), and gluing cocycle
- F a
F 1(Up@) -> 1/0,5 f G.
The pullback of a connection on P is defined similarly. The following result should be
obvious.

Proposition 8.1.18. (a) If P is a trivial G-bundle over the smooth manifold M , then
¢ ( p ) = 0 6 H ° ( M ) , for any go G I . (G).
(b) Let M 5 N be a smooth map between the smooth manifolds M and N . Then, for
every principal G-bundle over N , and any Q25 6 I' (G) we have

¢(F*(Pll Z F*(d)(P))-
Equivalently, this means the diagram below is commutative.

I°(G) cwi H°(N)

CUJFO

H.
If"
(M)
Denote by TG the collection of smooth principal G-bundles (over smooth manifolds).
For each P Q TG we denote by up the base of P. Finally, we denote by H' a contravari-
ant functor from the category of smooth manifolds (and smooth maps) to the category of
Abelian groups.

Definition 8.1.19. An 3"-valued G-characteristic class is a correspondence

TG 9 P P-> e(P) 6 3"(8p),
such that the following hold.

(a) e(P) = 0 if P is trivial.

(b) 9'(F)(e(P)) = c(F*(P)), for any smooth map F : M -> N, and any principal G-
bundle P -> N.
 Characteristic Classes 343

Hence, the Chern-Weil construction is just a method of producing G-characteristic

classes valued in the DeRham cosmology.

Remark 8.1.20. (a) We see that each characteristic class provides a way of measuring the
nontriviality of a principal G-bundle.
(b) A very legitimate question arises. Do there exist characteristic classes (in the DeRham
cosmology) not obtainable via the Chern-Weil construction?
The answer is negative, but the proof requires an elaborate topological technology
which is beyond the reach of this course. The interested reader can find the details in
the monograph [98] which is the ultimate reference on the subject of characteristic classes.
(b) There exist characteristic classes valued in contravariant functors other then the
DeRham cohomology. E.g., for each Abelian group A, the tech cosmology with coef-
ficients in the constant sheaf _A defines a contravariant functor H°(-, A), and using topo-
logical techniques, one can produce H' (-, A)-valued characteristic classes. For details we
refer to [98], or the classical [124].

8.2 Important examples

We devote this section to the description of some of the most important examples of char-
acteristic classes. In the process we will describe the invariants of some commonly encoun-
tered Lie groups.

8.2.1 The invariants of the torus T"

--
The n-dimensional torus T" = U(l) X - X U(l) is an Abelian Lie group, so that the
adjoint action on its Lie algebra t" is trivial. Hence
I.
(T*) = 5°((i")*)_
In practice one uses a more explicit description obtained as follows. Pick angular coordi-
nates 0 < H'L < 2tr, l < i < n, and set
1 .
to do] •
2tr'i
The acts form a basis of (tn * and now we can identify
/
7

I . lTN)£R[$1, fn]-
7

8.2.2 Chem classes

Let E be a rank T' complex vector bundle over the smooth manifold M. We have seen
that a Hermitian metric on E induces an U (T)-structure (P, p), where p is the tautological
representation
p : U(1°) <--> GL(1°,lC).

Exercise 8.2.1. Prove that different Hermitian metrics on E define isomorphic U(7°)-
structures.
344 Lectures on the Geometry of Manzfolds

Thus, we can identify such a bundle with the tautological principal U (T)-bundle of uni-

..
tary frames. A connection on this U (T)-bundle is then equivalent with a linear connection
V on E compatible with a Hermitian metric ( , ) , i.e.,
Vi (Au, u) A{(VxU,U) -|- l'u,, V X v } } ,
VA 6 (C, 'u,,v 6 COO(E), X 6 Vect ( M ) .
The characteristic classes of E are by definition the characteristic classes of the tauto-
logical principal U(1°)-bundle. To describe these characteristic classes we need to elucidate
the structure of the ring of invariants I°(U(r)).
The ring I'(U(r)) consists of symmetric, 1°-linear maps

<2) : QU) X • a • X u(t") -> R,

invariant with respect to the adjoint action
u(t") 9 X »-> TXT-1 E QU), T E up.
It is convenient to identify such a map with its polynomial form
P¢(X) = d ) ( X , . . . , X ) .
The Lie algebra _u(1°) consists of T' X T' complex skew-hermitian matrices. Classical results
of linear algebra show that, for any X 6 _u(r), there exists T 6 U(r), such that TXT-1 is
diagonal
TXT-1 idiag(/\1, . . . , /\v*).
The set of diagonal matrices in Q ) is called the Carton algebra of_u(r), and we will denote
it by €2(1,°). It is a (maximal) Abelian Lie subalgebra of QU). Consider the stabilizer
{ T 6 U(1~) ; TXT-1 = x, VX e @(.~)}»
..II

SU(7`)

and the normalizer

NU(r) := {T E UITI; T€!(1°)T-1 C @Q(1°)}°

The stabilizer SU(1°) is a normal subgroup of NU(,.°, so we can form the quotient

Wu(¢~) 2 = Nu(1~)/51/('~l

called the Weyl group o f U ( r l . As in Subsection 7.4.4, we see that the Weyl group is iso-
morphic with the symmetric group S because two diagonal skew-Hermitian matrices are
unitarily equivalent if and only if they have the same eigenvalues, including multiplicities.
We see that P <l> is Ad-invariant if and only if its restriction to the Cartan algebra is
invariant under the action of the Weyl group.
The Carman algebra is the Lie algebra of the (maximal) torus T" consisting of diagonal
unitary matrices. As in the previous subsection we introduce the variables
l
dO j .
..
II
I

17
2'/Ti
The restriction of P<l> to €_u(#°) is a polynomial in the variables all, . . . , a s . The Weyl group
S permutes these variables, so that P¢ is Ad-invariant if and only if PIbléli1, . . . , SET is a
 Characteristic Classes 345

symmetric polynomial in its variables. According to the fundamental theorem of symmetric

polynomials, the ring of these polynomials is generated (as an R-algebra) by the elementary
ones

Cr £111 337r

Thus
I.
(UW) R[C1, CQ, c 1.
In terms of matrices X 6 QU) we have

Co (Xltk det ]l- ,X 6 I . (U(7°))[¢].

2'rrz
k
The above polynomial is known as the universal rank T' Chem polynomial, and its c o e f f i -
cients are called the universal, rank 7" Chem classes.
Returning to our rank T' vector bundle E, we obtain the Chem classes of E
bE) == co(F(Vll G H2"=(m),
and the Chem polynomial of E

Ct (MIM-
Above, V denotes a connection compatible with a Hermitian metric (
F(V) denotes its curvature.
..
7 ) on E, while

Remark 8.2.2. The Chern classes produced via the Chern-Weil method capture only a part
of what topologists usually refer to characteristic classes of complex bundles. To give the
reader a feeling of what the Chern-Weil construction is unable to capture we will sketch a
different definition of the first Chern class of a complex line bundle. The following facts
are essentially due to Kodaira and Spencer [81], see also [55] for a nice presentation.
Let L -> M be a smooth complex Hermitian line bundle over the smooth manifold M.
~
Upon choosing a good open cover (Ua) of M we can describe L by a collection of smooth
maps zag : Uog -> U(1) = S1 satisfying the cocycle condition
ZQBZBWZWQ II W # - (8.2. 1)
If we denote by COO(-, S1) the sheaf of multiplicative groups of smooth S1-valued func-
tions, we see that the family of complex line bundles on M can be identified with the Czech
group H1 (M, COO(-, g11l. This group is called the smooth Picard group of M. The group
multiplication is precisely the tensor product of two line bundles. We will denote it by
PicOO ( M ) .
If we write zag = exp(2'rri9o@) (93O = -Hog 6 COO(Uo5,]R)) we deduce from (8.2. 1)
that VUQB7 ¢ (Z)
Hog -|- HM -|- H = "QM E Z.
346 Lectures on the Geometry of Manzfolds

It is not difficult to see that, VU ¢ QL we have

"B76 - "OMS -|- "0M36 - "QM i 0.

In other words, the collection "QM defines a Czech 2-cocycle of the constant sheaf Z.
On a more formal level, we can capture the above cocycle starting from the exact se-
quence of sheaves
exp(2'rr"i~)
0 -> Z--> C°°(-JR) > COO<-,SU -> 0.
The middle sheaf is a fine sheaf so its cosmology vanishes in positive dimensions. The
long exact sequence in cohomology then gives

0 -> Pic" (m) i> H2(m, z ) -> 0.

The cocycle ( "QM ) represents precisely the class 6(L).
The class 6(L), L 6 Pic°° (M) is called the topological first Chern class and is denoted
by etlop (L). This terminology is motivated by the following result of Kodaira and Spencer,
[81 ] :
The image of Diop ( L ) in the DeRham chorology via the natural morphism
H*(M,Z) -> H*(M,lR) =
~ HERIMI
coincides with the ]inst Chem class obtained via the Chem-Weil procedure.
The Chern-Weil construction misses precisely the torsion elements in H2(M, Z). For
example, if a line bundle admits a flat connection then its first Chern class is trivial. This
may not be the case with the topological one, because line bundle may not be topologically
trivial.

8.2.3 Ponhyagin classes

Let E be a rank T' real vector bundle over the smooth manifold M. An Euclidean metric on
E induces an O(r) structure (P, p). The representation p is the tautological one
p : O(1°) <--> GL(7°,R).

Exercise 8.2.3. Prove that two metrics on E induce isomorphic OU) -structures.

Hence, exactly as in the complex case, we can naturally identify the rank 1°-real vector
bundles equipped with metric with principal O(r)-bundles. A connection on the principal
bundle can be viewed as a metric compatible connection in the associated vector bundle.
To describe the various characteristic classes we need to understand the ring of invariants
I.
(O(1"ll-
As usual, we will identify the elements of In(O(r)) with the degree k, Ad-invariant
polynomials on the Lie algebra Q(1°) consisting of skew-symmetric T' X T' real matrices. Fix
P 6 Ik:(O(1°)). Set m = [T/2], and denote by J the 2 X 2 matrix

J
..
II
 Characteristic Classes 347

Consider the Carton algebra

{A1J® ® AmJ 6 QU); *j 6 ]R}, 7° 2m
€Q(r) {A1J® ® AmJ ® 0 6 QU); *j 6 R}, 7° 2m -|- l
The Cartan algebra 6900 is the Lie algebra of the (maximal) torus
RE1 ® ---®R@m 6 O(1°), 7° 2m
® Ram ® HR E OITI,
7
RE1 ® 7° 2m -I- l

where for each H E [0, 27r] we denoted by RE the 2 X 2 rotation

- cost _ ' Q
RE II
;;M
..
sin H

As in Subsection 8.2.1 we introduce the variables

1
"la »-doj .
27r
Using standard results concerning the normal Jordan form of a skew-symmetric matrix,
we deduce that, for every X 6 Q(1°), there exists T 6 O(1") such that TXT-1 G €9(r)°
Consequently, any Ad-invariant polynomial on Q(1") is uniquely defined by its restriction to
the Caftan algebra.
Following the approach in the complex case, we consider

SO(1~) { T e of); TXT-1 = x, VX e @Q(»~) 7

Nom {T G Olrl; T € 9 ( t ~ ) T - 1 C @9(»,~)}-

The stabilizer SO('r') is a normal subgroup in N(O(r)), so we can form the Weyl group
..
WO(T) NO(r)/SO(r)°

Exercise 8.2.4. Prove that We(Tl is the subgroup of GL(m, R) generated by the involu-
sons
(7 j (X17'°'7Xi7"°7Xj7"'7 to) »-> (-17 7 X j 7 ° ' ° 7 X i 7 ' ° ' 7 aim

go: (Xl 7 ,.ZUlu,..., am) | (-17 ,-$j,..., as .

The restriction of P 6 Ik (O(1°)) to €9(',°) is a degree k homogeneous polynomial in the
variables 5131, . . . , :am invariant under the action of the Weyl group. Using the above exercise
we deduce that P must be a symmetric polynomial P = P(x1, . . . m), separately even
in each variable. Invoking once again the fundamental theorem of symmetric polynomials
we conclude that P must be a polynomial in the elementary symmetric ones
PI = Z j"2
2 2
P2 Z 2 i<j zzizvj

2 2
Pm $1 Mm
348 Lectures on the Geometry of Manzfolds

Hence
I°(0(t~)) RlPl 7 7 Plr/2l]°
In terms of X 6 Q(r) we have

X I.
P¢lXl = pa (X )t2i det ]1
27r E (0(t"))[t]-
j

The above polynomial is called the universal rank T' Pontryagin polynomial, while its co-
efficients pa (X ) are called the universal rank 7° Pontryagin classes.
The Pontryagin classes PI (E), . . . , pm (E) of our real vector bundle E are then defined
by the equality

PE) p j ( E l t 2 j = det
( ]1
W) E H°(m)m,

where V denotes a connection compatible with some (real) metric on E, while F(V) de-
notes its curvature. Note that pj(E) 6 H42' ( M ) . The polynomial pt(E) 6 H'(M)[t] is
called the Pontryagin polynomial o f E .

8.2.4 The Euler class

Let E be a rank r, real oriented vector bundle. A metric on E induces an O(r)-structure,
but the existence of an orientation implies the existence of a finer structure, namely an
SO(r)-symmetry.
The groups O(r) and SO(r) share the same Lie algebra U) = Q(r). The inclusion
Z : SO ( ) <-> OITI,
induces a morphism of R-algebras
Z* I.
(0(7')) -> I"($O('f"ll-
Because U) = Q(r) one deduces immediately that Z* is injective.
Lemma 8.2.5. When T' is odd then Z* I . (O(1°ll -> I°(50(¢~)) is an isomorphism.

Exercise 8.2.6. Prove the above lemma.

The situation is different when 7° is even, T' Qm. To describe the ring of invariants
I . (50(2ml), we need to study in greater detail the Cartan algebra

€Q(2m)
= { AIJ ® ® AmJ 6 Q<2m> }
and the corresponding Weyl group action. The Weyl group Wso(2m)» defined as usual as
the quotient

Wso(2m) = Nso(2m)l5so(2m) 7
is isomorphic to the subgroup of GL((3Q(2,,,,)) generated by the involutions
Oiij (Al,..,A¢,...,Ai,...,ATTIl»->(A1,...,Aj,...,A,...,ATTI),
 Characteristic Classes 349

and
5 : ( u ... 7 All P-> l€1A17 . . swAm),
7

where 61, . . . ,am = :l:l, and 6 1 ---

am = l . (Check thi5!)
Set as usual Xi := -Ai/27r. The Pontryagin O(2m)-invariants
)2
Pj(SU1, 7 :am .CU Maj 7

1<'i1<---<ijS'f7l

continue to be Wsol27Tll invariants. There is however a new invariant,

Alélil, 711'rn, to
j

In terms of
X = ~A1] as ®A,] E @so(2m,),

we can write
m

AIX) p f (Xl,

where Pf denotes the pfaffian of the skewsymmetric matrix X viewed as a linear map
R2m -> R2M, when R2m is endowed with the canonical orientation. Note that pm = A2.

Proposition 8.2.7.
I°(SO(2m)) N
: ]R[Zl, Z2, . . . 7 Zm; Y]/(}/2 - gm) Izzy = Pa, Y A)
where Ly2 - Zm) denotes the ideal generated by the polynomial Y 2 - gm.

Proof. We follow the approach used by H. Weyl in describing the invariants of the alternate
group ([133], Sec. IL2). The isomorphism will be established in two steps.
Step 1. The space I ° ( S O ( 2 m ) ) is generated as an R-algebra by the polynomials
Plum' 7P71l7A°

Step 2. The kernel of the morphism

R[z1, . . . , Z m , ; Y ] i I°(SO(2m))

defined by ZN »-> pa, Y »-> A is the ideal (YQ - Zm).

Proof of Step 1. Note that W S o ( 2 m ) has index 2 as a subgroup in W 0 ( 2 m l . Thus W s 0 ( 2 m l
is a normal subgroup, and

9
The group 9 = {]1, e} acts on I°(SO(2m)) by
l€F)l£U1,£C2, ...736i) I Fl-£C1,£U2, 7 am ... -F(-17§627 7 mm 7

and moreover,
I°lo(2mll = Ker(]l - 2).
350 Lectures on the Geometry of Manzfolds

For each F 6 I ° ( S O ( 2 m ) ) we define F+ := (II -|- e)F, and we observe that F+ 6

ker(l - 2). Hence
F+ P(p1, . . . , p T T I ) .

On the other hand, the polynomial F' (11- e)F is separately odd in each of its variables.
Indeed,

F < -X17X27°°°7 act) = eF'(;u1,...,ac2,71)

e(]l - e)F(;v1,...,;z:m) L -(11 - elF(a:1,...,xm) = -F'(;u1,...,xm).

F_
Hence, F' vanishes when any of its variables vanishes so that is divisible by their
product A = 231 as, ---
F- i A . G.
Since oF' = -F', and PA - A we deduce eG = G, i.e., G 6 I°(O(2m)). Conse-
quently, G can be written as

G Z Q(P1». • • 7 P 7 7 l ) 7
so that

F-=A°QlP1a°°°aPml-
Step 1 follows from
1 1
F = §(F" + F' I §lPlP1, ,pm)+A~C2(p1, ,pm) )-
Proof of Step 2. From the equality
detX = p f (X)2 ex G @(2m)7
we deduce

ICQ - z771 l C Kemp,

so that we only need to establish the opposite inclusion.
Let P = P(Z1, Z2, . . . , Zm;Y) 6 kerzp. Consider P as a polynomial in Y with
coefficients in R[Z1, . . . , Zm]. Divide P by the quadratic polynomial (in Y) Y2 - gm.
The remainder is linear
R = A ( Z 1 , . . . , z'rn)y + Blzl, . 7 zTTI>.

Since Y2 Zm, P 6 her Q , we deduce R 6 her Tb. Thus

A(p1, . . . , p a l A -|- B(p1, . .
Applying the morphism e we get

-A(p1,---,pm)A+B(p1 mm) 0.
Hence A E B E 0 so that P is divisible by YE - Z
 Characteristic Classes 351

Let E be a rank Zm, real, oriented vector bundle over the smooth manifold M. As in
the previous subsection we deduce that we can use a metric to naturally identify E with a
principal SO(2m)-bundle and in fact, this principal bundle is independent of the metric.
Finally, choose a connection V compatible with some metric on E.

Definition 8.2.8. (a) The universal Euler class is defined by

l
e e(X) ( w p f ( - x ) 6 1m<so(2m>>.

(b) The Euler class of E, is the cosmology class e(E) E H2""(M) represented by the
Euler form
l
et) (21T)*" p .f ( F(V )) 6 Q W M ) .
(According to the Chem-Weil theorem this cosmology class is independent of the metric
and the connection.)

Example 8.2.9. Let (E, g) be a compact, oriented, Riemann surface and denote by VI the
Levi-Civita connection. The the Euler form
1
6(9) =9l9ld'ug
coincides with the Euler form e(V9) obtained via the Chern-Weil construction.

Remark 8.2.10. Let E be a rank 2m, real, oriented vector bundle over the smooth, com-
pact, oriented manifold M. We now have two apparently conflicting notions of Euler
classes.
A topological Euler class et0plEl 6 H2M(M) defined as the pullback of the Thom class
via an arbitrary section of E.
A geometric Euler class geom (E) 6 H2m (Ml defined via the Chern-Weil construction.
The most general version of the Gauss-Bonnet theorem, which will be established later
in this chapter, will show that these two notions coincide !

8.2.5 Universal classes

In each of the situations discussed so far we defined characteristic classes for vector bundles
with a given rank. In this subsection we show how one can coherently present these facts
all at once, irrespective of rank. The algebraic machinery which will achieve this end is
called inverse limit. We begin by first describing a special example of inverse limit.
A projective sequence of rings is a a sequence of rings {R2'L}fn120, together with a se-
(in
quence of ring morphisms Rn ( Rn-|-1- The inverse limit of a projective system (RTL , (in )
is the subring
lim Rn C
<- H Rn
n>0

consisting of the sequences (33n+1l 3719 V1220.

352 Lectures on the Geometry of Manzfolds

Example 8.2.11. Let Rn HX1, . . . , XA] be the ring of formal power series in n
variables with coefficients in the commutative ring with unit 32. (Ro = R.) Denote by
• Qn+1 -> Rn the natural morphism defined by setting Xn-I-1 = 0. The inverse limit of
,L
.. _

this projective system is denoted by RHX1, X2, . . .]].

Given a sequence FTL 6 9€[[X]] (n > l) such that Fn (0) = l, we can form the sequence
of products
ll,Fllxll,FllxllF2lx2l, . . .,F1lX1) ---Fn(XTL), . . . 7

which defines an element in 92HX1, X2, . . .H denoted by FI (X1lF2(X2) When FI

--
F2 = -Fn = ---
= F the corresponding elements is denoted by I F O O

Exercise 8.2.12. (a) Let 3€[[x]]7 denote the set of formal power series F E Hmm such that
F(0) = l. Prove that (92[[;z:]]7, -) is an abelian group.
(b) Prove that VF, G 6 32[[x]]7

(F G)°° = (F)°° (G)°°.

a

Similarly, given Gn 6 52[[m]], (n > 1) such that Fn(0) = 0, we can from the sequence
of sums
(07G1(X1l7G1(X1)+G2(X2)»... ,G1(X1) + ---€BG'n(Xnl,... 7

which defines an element in 32[[X1, X2, . . .]] denoted by G1(X1) + G2(X2) -I- When
-- - -
G1 = G2 = - = Fn = - = F we denote the corresponding element Igloo.

In dealing with characteristic classes of vector bundles one naturally encounters the
increasing sequences
u(1) <-> z/(2) + (8.2.2)

(in the complex case) and (in the real case)

0(1) --> 0(2) + (8.2.3)

We will discuss these two situations separately.

The complex case. The sequence in (8.2.2) induces a projective sequence of rings
R <- I ° ' ( U ( l ) ) <- I°°(u(2)) <- (8.2.4)

We know that I°°(U(n)) = So*[[;z:j]]= the ring of symmetric formal power series in n
variables with coefficients in R. Set

5°OM : 1imS".
<-

Given a rank n vector bundle E over a smooth manifold M, and QUO 6 I.. (U(n)), the
characteristic class cl5(E) is well defined since dim M-I-1 = 0, for any u 6 Q*(M). Thus
 Characteristic Classes 353

we can work with the ring I°°(U(n)) rather than I ' ( U ( n ) ) as we have done so far. An
element
¢=(¢1»¢2»---I E f°°(u(ooll lim I.. (U("ll i 5°°[[H3i]]
<-
is called universal characteristic class.
If E is a complex vector bundle, we set ¢(E)== = Cl5,.(E), where T' = rank E. More
precisely, to define (la (E) we need to pick a connection V compatible with some Hermitian
metric on E, and then set
(ME) i (z5»~(F(V))-
Example 8.2.13. We denote by cg) ln 2 lf the elementary symmetric polynomial in n
variables
@;;> ml . . . iiiik
1§i1<°°°<ik§'r7,

Then the sequence

(k) (k-I-ll
(0, 707Ck 7 Co»I 7

defines an element in 8°O [[:z:j]] denoted by et, which we call the universal k-rh Chem class.
Formally, we can write
Co $i1$i2 1/3ik
1<_¢1<---<¢k<<)o

We can present the above arguments in a more concise form as follows. Consider the
function F(a:) = (l + to) 6 in[[x]], where R = la[[t]] is the ring of formal power series in
the variable t. Then (F)OO defines an element in LRHX1, X27 . . .]]. One sees immediately
that in fact, (F)OO 6 ~SOO[[a2j]] [[t]] and moreover
(Fl°°($17 U327 . . .> = <1 + w;1><1 + UZQI 1-I-c1t-I-c2t2 +
We see that
Ct H
j>1

is the universal Chem polynomial.

We can perform a similar operation with any F E RHHIH such that Flo) = 1. We get a
semigroup morphism
(R[[¢v]]7» ») 9 F »-> (F)°° E l5°O[[53j]], .).
(See Exercise 8.2. 12.) One very important example is
OO
QUO l l k-1 Be 2k 6
F(1vl l e a c
=l - a 3 - I - 1 9 0 2 + - - - = 1 + QUO -|-
-I- 2 12 2
(-1
) <2I<>!" RHHIH-
k 1

The coefficients Be are known as the Bernoulli numbers. The product

331 332
<F)°°
( l e'Xl
)( l e'X2
) 7
354 Lectures on the Geometry of Manzfolds

defines an element in soo [[a:j]] called the universal Todd class. We denote it by Td. Using
the fundamental theorem of symmetric polynomials we can write
Td l -|- Td 1 + Td 2 + 7

where Tdn E sun is an universal symmetric, homogeneous "polynomial" of degree n,

hence ex ressible as a combination of the elemental s mmetric " of no rials" 01 7 02 7 . . ..
By an universal"polynomial" we understand element in the inverse limit
R[$1,$2, .. limR[.cl31, $82,
<-

An universal "polynomial" P is said to be homogeneous of degree d, if it can be represented

as a sequence
P i (p1,p2, . . .),

where Pm is a homogeneous polynomial of degree d in m variables, and

Pm+1(l"l, 11327 ,:z3m,0) = Pm(:c1,...:cm).

For example,
1 1 1
12 (021 + C2l,
Td1 -017 Td 2 Td 3 0102 etc.

Analogously, any function G G RHo]] such that G(0) = 0 defines an element

(GIro = GCu1) + G(1U2l + u ° ° 6 8°°[[1Uj]]-

We have two examples in mind. First, consider G(a:) = wk. We get the symmetric function

so = (a:k)OO = (G)OO =x§+x§+--- ,

called the universal k-th power sum. We will denote these by So.
Next, consider
mm

1n>1
ml

We have

so 6 5(><>[[g3i]].

Given a complex vector bundle E we define

oh ( E ) = rank E -I- (ex - 1)oo(E).

The cohomology class oh ( E ) is called the Chem character of the bundle E. If V is a

connection on E compatible with some Hermitian metric, then we can express the Chem
character of E as

ch (El I tr
( €F (V )
) rank ( E ) +§ l t r (F(v)Ak<),
k!
k=1
 Characteristic Classes 355

where /\ is the bilinear map

s2(End (E)) X Qi (End (El) -> Q i + j (End (E))
defined in Example 3.3.12.

Proposition 8.2.14. Consider two complex vector bundles E l , E2 over the some manifold
M . Then
oh (El ® E2) oh (El) + ch (E2l»
and

C1IIE1 (X) E2) I Ch(E1 I /\ oh (E2) E H ° ( m ) .

Proof. Consider a connection Vi on Er compatible with some Hermitian metric h i , i =

l , 2. Then V1 ® V2 is a connection on E1 ® E2 compatible with the metric hi GB h2,and
moreover
F(vl ® V 2 ) = F ( V 1 ) ® F(v2).
Hence
exp(F(Vl ® v2))=exp(F(v1ll ® exp(F(v2)),
from which we deduce the first equality.
As for the second equality, consider the connection V on E1 ®E2 uniquely defined by
the product rule
VIs1 (X) 82) (V181) (X) 82 + 81 (X) (V282), So 6 C°°(E),

where the operation

(X) Qk(E1l X 9£(E2) -> Qk-I-€(E1 (X) EQ)
is defined by

(w ® 81) (X) ( 8 2 ) = (wk A of) ® (51 (X) S2 (8.2.5)

6 C°°(E), Vi," 6 m m ) , and V176 E Q Q M ) .
'v'81
We compute the curvature of V using the equality F(V) = (dV)2 If $i 6 (jOO(E),
then
FIVNs1 (X) 32) = dV{<v151) ® S2 -|- 81 (X) ($282)}

{(FlV1ls1l (X) 82 (V181) (X) (V282) + lvlsll (X) (V282) + S1 ® (F(v2))}

= F(v1) (X) HE2 + 11E (X) FW2).

The second equality in the proposition is a consequence the following technical lemma.

Lemma 8.2.15. Let A (respectively B ) be a skew-adjoint, n X n (respectively m X m)

complex matrix. Then
tr (exp(A (X) ]1@m -|- Ile" (X) Bll = tr (exp(A)) - tr (exp(Bl).
356 Lectures on the Geometry of Manzfolds

Proof. Pick an orthonormal basis (et) of C" and an orthonormal. basis (fy) of Cm) such
that, with respect to these bases A = diag(A1, . . . ,An) and B = dlagl/11, - - - ,,u,m)), Then,
with respect to the basis lai (X) fol of CC" (X) (CM, we have
(A (X) Mm -|- ]llcn (X) Be = drag( As -|- M

Hence
exp(A (X) ]1cm -|- M" (X) B ) = diag(eX"€/*3),
so that,
tr (A (X) 11cm + loc" (X) B) €Ai€Hj tr (QA)-tr (QB).
The proposition is proved.

Exercise 8.2.16 (Newton's formulae). Consider the symmetric polynomials

Co £/1 1/3ik 6 R[$1,. . . ,£1Z'n],

<_~~-'ik

1$i1 S71

and (120)

G R[$1,. . . » i n ] .
T'
S 1
j

Set
n

f(t) H (1 - tif)-
j 1

(a) Show that

f'(tl 5,~t'° - 1
f(t) 1°

(b) Prove the Newton formulae

T'

j 1

(c) Deduce from the above formulae the following identities between universal symmetric
polynomials.
2 3
S1 C17 S2 01 2027 S3 01 30102 303 .

The real case. The sequence (8.23) induces a projective system

I°°(0(1)) <- I°°(0(2)) 4- .
We have proved that I°° (O(n)) = S L"/2J [[xi]] = the ring of even, symmetric power series
in In/Zl variables. The inverse limit of this system is
I ° o (O(oo) = lim I°°(0(7~l° s°°uw§H lim 5""[[xV]]-
 Characteristic Classes 357

As in the complex case, any element of this ring is called a universal characteristic class. In
fact, for any real vector bundle E and any (15 6 soo [[a:]] there is a well defined characteristic
class qz5(E) which can be expressed exactly as in the complex case, using metric compatible
connections. If F 6 ]R[[ac]]7 then (F(232ll°° defines an element of sOO [[a:]].
In topology, the most commonly encountered situations are the following.
A.
22k-1 _ l k

22k-1(2k)!B"' •

The universal characteristic class (F(23ll°° is denoted by Z, and it is called the -genus.
We can write
<<:

l+A71 + A72 -I-

II

where As, are universal, symmetric, even, homogeneous "polynomials", and as such they
can be described using universal Pontryagin classes
2 2
Pm "in Xjm
1<j1<~~jm

The first couple of terms are

B. Consider
22k
F(a5) \/5 1 + 1-at + -1 2 + = l + Z(-1)'"'1 Bkxk.
tank \/5 3 45" A21 (2k)!

The universal class (F(:v2))°° is denoted by L, and it is called the L-genus. As before, we
can write
L 1+/1+ L2 -|- 7

where the Le's are universal, symmetric, even, homogeneous "polynomials". They can be
expressed in terms of the universal Pontryagin classes. The first few terms are
1 1
L1 §P17 L 2 45( 7P2
lllllllllllII
I921) etc.

8.3 Computing characteristic classes

The theory of characteristic classes is as useful as one's ability to compute them. In this
section we will describe some methods of doing this.
Most concrete applications require the aplication of a combination of techniques from
topology, differential and algebraic geometry and Lie group theory that go beyond the
scope of this book. We will discuss in some detail a few invariant theoretic methods and
we will present one topological result more precisely the Gauss-Bonnet-Chern theorem.
358 Lectures on the Geometry of Manzfolds

8.3.1 Reductions

In applications, the symmetries of a vector bundle are implicitly described through topo-
logical or geometric properties.
For example, if a rank r complex vector bundle E splits as a Whitney sum E = $1 GBEQ
with rank E, H, then E, which has a natural U(r)-symmetry, can be given a finer
structure of U(1°1) X U(r2) vector bundle.
More generally, assume that a given rank 7° complex vector bundle E admits a G-
structure (P, p), where G is a Lie group, P is a principal G bundle, and p : G -> U(r)
is a representation of G. Then we can perform two types of Chern-Weil constructions:
using the U (T) structure, and using the G structure. In particular, we obtain two collections
of characteristic classes associated to E. One natural question is whether there is any
relationship between them.
In terms of the Whitney splitting E = $1 ® E2 above, the problem takes a more
concrete form: compute the Chern classes of E in terms of the Chern classes of E1 and E2.
Our next definition formalizes the above situations.

Definition 8.3.1. Let (p : H -> G be a smooth morphism of (matrix) Lie groups.

(a) If P is a principal H-bundle over the smooth manifold M defined by the open cover
(Ua), and gluing cocycle
hog : Uog -) H,
then the principal G-bundle defined by the gluing cocycle

90¢a=(,o<>ho5:U(,0->G

is said to be the up-associate of P, and it is denoted by go(pl.

(b) A principal G-bundle Q over M is said to be go-reducible, if there exists a principal
H-bundle P -> M such that Q = $o(P).

The morphism up H -> G in the above definition induces a morphism of R-algebras

p*
1° (G) -> 1° (H)-
The elements of her go* C I°(G) are called universal identities.
The following result is immediate.

Proposition 8.3.2. Let P be a principal G-bundle which can be reduced to a principal

H-bundle Q. Then for every 17 6 her go* we have
17(P) = 0 in H *<m>.

Proof. Denote by LG (respectively L H) the Lie algebra of G (respectively H), and by 90*
the differential of up at l 6 H ,
J H £ / G .
 Characteristic Classes 359

Pick a connection ( A l on Q, and denote by (Fa) its curvature. Then the collection
go* (Aa) defines a connection on P with curvature up* (Fa). Now

v((p*(Fal)= ((/)*t1)(Fo) 0.
The above result should be seen as a guiding principle in proving identities between
characteristic classes, rather than a rigid result. What is important about this result is the
simple argument used to prove it.
We conclude this subsection with some simple, but very important applications of the
above principle.

Example 8.3.3. Let E and F be two complex vector bundles over the same smooth mani-
fold M of ranks r and respectively S. Then the Chem polynomials of E, F and E ® F are
related by the identity

CtlE ® Fl = CE) • CtlFl» (8.3.1)

where the "-" denotes the A-multiplication in Heven ( M ) . Equivalently, this means

et(E€BF)= C¢(El'CJ(Fl-
i+j=k

To check this, pick a Hermitian metric g on E and a Hermitian metric h on F. g ® h is

a Hermitian metric on E ® F. Hence, E ® F has an U (r -|- 5) structure reducible to an
U(r) X U(s) structure.
The Lie algebra of U(r) X U(s) is the direct sum u(r) Q; _u(s). Any element X in this
algebra has a block decomposition XE

XT' GX87
II

where X (respectively XS) is an T' X T' (respectively S X s) complex, skew-hermitian matrix.

Let z denote the natural inclusion QU) QBQISI <-> _u(1°+s) and denote by 6 I* (I/(V1) [t] ii"
the Chem polynomial.
We have
* t
( )
1°-I-8
z (c ) )(XT- ® X8) I det ]11°+s . X ® XI
2tr'I
t
( w( (s)
det 11 T .X 118 .Xs cy) Ct (8).
2Trz 2tr'I

The equality (83.1) now follows using the argument in the proof of Proposition 8.3.2.

Remark 8.3.4. Consider the Grassmannian Grk((C") of complex k-dimensional sub-

spaces in (Cn. The universal complex vector bundle U p n -> Gr k(<C" ) is a subbundle
_
of the trivial bundle Q" -> Grk((CTL). The trivial bundle Cn is equipped with a canonical
Hermitian metric. We denote by Q k , n the orthogonal complement of U p n in Q" so that we
have an isomorphism
Qn Uknggkn-
360 Lectures on the Geometry of Manzfolds

From the above example we deduce that

)
Ct lUk," )@( Qw I
C t (Q" ) l.
Denote by Up the j-th Chem class of U p T L and by 'Up the 6-th Chem class of Qkyn- Then
k n-k

Ct I Uk,n ) ujtj, Cal Q/c ," ) ugh.

j 0 e=o

I
One can then prove that the cosmology ring H' Gr 1<(@ "l» lm) is generated by the
classes 'UȢ,Uj, which are subject to the single relation above. More formally

H' ( Gr I<(<€")»Rl R["~Li»Uj5 0<i <k, 0 < j <

deg 'Up = Zi, deg Up = 2 . The proof requires a more sophisticated topological machinery.
(l. )la) )
i
'Up

j
Up l 7

For details we refer to [99], Chapter v3, Section 6.

Exercise 8.3.5. Let E and F be two complex vector bundles over the same manifold M.
Show that
Td ( E ® F ) Td ( E ) • Td ( F ) .
Exercise 8.3.6. Let E and F be two real vector bundles over the same manifold M . Prove
that
Pt(E ® F) = PtlEl °P¢lFl» (8.3.2)

where Pt denotes the Pontryagin polynomials.

Exercise 8.3.7. Let E and F be two real vector bundles over the same smooth manifold
M. Show that
L (EQBFI L(E) - L ( F )

21(E ® F) AS(E) -21(Fl-

Example 8.3.8. The natural inclusion R" '--> (C" induces an embeddings : O(n) --> U(n).
(An orthogonal map T : R" -> R" extends by complexification to an unitary map To :
(EnL ->@'*.) This is mirrored at the Lie algebra level by an inclusion

201) --> _u(")-

We obtain a morphism Z* I. I.
(I/(n)) -> (0(v~,))), and we claim that
i* (C21<=+1) 0, and z*(c2k11) = ( )kpa.

Indeed, for X 6 Q(n) we have

2k-I-1

Z* (02/<+1 ) ( X l M (X) • ' • )'12k+1 (Xl,

271i /
1<i1<~~~<'i2k-I-1<7V1
 Characteristic Classes 361

where *J (X) are the eigenvalues of X over (I. Since X is, in effect, a real skew-symmetric
matrix we have
MY) = Aj(X ) = - i ( X ) -
Consequently,
2k+1

7*(C2k+1)(X) = 2*lC21<:+1 (YD 27ri/

As; (X) ' ' ' )"l12k-l-1 (X)
1<i1<---<i2k+1<N
2k-I-1
l_ll2k-I-l *
' )"i2k-I-1 (Xl Z (C2k+1l(X)-
27r'£/
1<i1<---<i2/¢+1<"
The equality 2* (0210 = (-1lk:Pk is proved similarly.

From the above example we deduce immediately the following consequence.

Proposition 8.3.9. If E -> M is a real vector bundle and E (X) (C is its complexification
then
pk(E) =(-l)" '@k(E (X) (c), I<;=1, 2, (8.3.3)
In a more concentrated form, this means
Pa(El = P - t ( E l = CME ® ©-
Exercise 8.3.10. Let E -> M be a complex vector bundle of rank r.
(a) Show that CklE*l = (-l)"'0I<(E)9 i.e.,
CtlE*) C_t(E)-
(b) One can also regard E as a real, oriented vector bundle ER. Prove that
PMER) .E(-1)k¢2k<p,(ER)
k
I CtlEl . C-tlEl,
and
c,"(El = €(E1R)-

Exercise 8.3.11. (a) The natural morphisms I°°(U(1°)) -> I"(O(r)) described above
induce a morphism
<I>OO : I"°(U(OOll -> 1'°'(O(oo)).
As we already know, for any F 6 R[[a2]]7, the product (FIOO is an element of I" (U(0O)),
and if moreover F is even, then (F)OO can be regarded as an element of I** (O(oo)). Show
that for every F 6 R[[;z:]]7
@of((F)°°)= (F.F-)<)<) E I"(O(oo))»
where F'(x) = F(-ac).
(b) Let E be a real vector bundle. Deduce from part (a) that
Td ( E ® © ) 21III2.
362 Lectures on the Geometry of Manzfolds

Exercise 8.3.12. For every square matrix X we set2

OO
1/2 Xi T' T' 7° 1) T' k+l)
(11+x>1/2
k 0
(-1l'<=
( ) k
7
()k ki 7

and d€t1/2(]1 -|- x) := det(]1 -|- x)1/2

Suppose that E -> M is a rank T' real vector bundle, equipped with a metric and a
compatible connection V. The connection V induces a Hermitian connection VC on the
complexification Ee := E ®1R (C. Prove that

F(VC) F(V e
L Iv) = detl/2 7 AS(v) = detl/2 e
tank(F(V°) sinhlFlv

Example 8.3.13. Consider the inclusion

z : solo) X so(2€>> <-> so(2If + $e).
This induces a ring morphism
I°(5o(2A: + be) -> I°(so(21l X so(2e>>.
Note that
I'(SO(2k) X
~
SO(2€)) = I'(SO(2k)) ®1R I'(SO(2£).
Denote by e(") the Euler class in I°(SO(2u)/. We want to prove that
Z*(€(k-1-Q> = 6(k ) (X)
6(6)

LetX = X1<5:€9X£ 6 $(21@)®(2€). Modulo aconjugationby ( S , T ) 6 SO(2k)xSO(2€)

we may assume that
Xi = A1] ® ® As] and Xe = # 1 ] € 9 ® MJ,
where J denotes the 2 X 2 matrix
rw
II

We have
k-I-E
( (1t+
')>(X
/\1'°°/v<:°/l1_#2 I e(1©>(Xk< . €(£)(X£)
>

€(k) (X) €(£)(Xk ® Xi).

The above example has an interesting consequence.

Proposition 8.3.14. Let E and F be two real, oriented vector bundle of even ranks over
the same manifold M . Then
e ( E ® F ) = e ( E ) e(F), -
-
where denotes the A-multiplication in Heven ( M ) .
oWe are not won'ied about convergence issues because the matrices for which we intend to apply the formula
have nilpotent entries.
 Characteristic Classes 363

Example 8.3.15. Let E be a rank 2h real, oriented vector bundle over the smooth manifold
M. We claim that Q' E admits a nowhere vanishing section 5 then e ( E ) = 0.
To see this, fix an Euclidean metric on E so that E is now endowed with an SO(2h)-
structure. Denote by L the real line subbundle of E generated by the section g. Clearly, L
is a trivial line bundle, and E splits as an orthogonal sum

E L ® Ll.
The orientation on E, and the orientation on L defined by f induce an orientation on L_L 7

so that LJ- has an SO(2k - 1)-structure.

In other words, the SO(2k) structure of E can be reduced to an SO(l) X SO(2k-1) r\/

SO(2k - 1)-structure. Denote by 2* the inclusion induced morphism

I*(50<2/<)> -> I*<so(21@ - 1)).

c5
Since 'L*l€lkll 0, we deduce from the Proposition 8.3.2 that e(E)

II
The result proved in the above example can be reformulated more suggestively as fol-
lows.

Corollary 8.3.16. Let E be a real oriented vector bundle of even rank over the smooth
manifold M. If e(E) ;é 0, then any section of E must vanish somewhere on M !

8.3.2 The Gauss-Bonnet-Chern theorem

If E -> M is a real oriented vector bundle over a smooth, compact, oriented manifold M

.
then there are two apparently conflicting notions of Euler class naturally associated to E.
The topological Euler class

etoplEl I C07'E7

where TE is the Thom class of E and go : M -> E is the zero section.

.The geometric Euler class
1
(g71ry pf (-F(V)) if rank (E) is even
eger (El if rank (El is odd
7
0

where V is a connection on E compatible with some metric and 21° = rank (E). The next
result, which generalizes the Gauss-Bonnet theorem, will show that these two notions of
Euler class coincide.
'ii
Theorem 8.3.17 (Gauss-Bonnet-Chern). Let E -> M be a real, oriented vector bundle
over the compact oriented manifold M . Then

et0plEl egeomlE)~
364 Lectures on the Geometry of Manzfolds

Proof. We will distinguish two cases.

A. rank (E) is odd. Consider the automorphism of E
i : E - > E ur->-'u,'v"u€E.
Since the fibers of E are odd dimensional, we deduce that i reverses the orientation in the
fibers. In particular, this implies
or*i*TE = -tl'*TE = tr* (-TEL

where or* denotes the integration along fibers. Since 7T* is an isomoiphism (Thom isomor-
phis theorem), we deduce
i*TE TE .
Hence
etoplE) I -C0i*7-E. (8.3.4)

On the other hand, notice that

Cui* C0-
Indeed,

C0 i* = (i§0)* Z (-Co)* = (§)* (Co = -Col-

The equality stop = e g e r now follows from (8.3.4).
B. rank (E) = Qk. We will use a variation of the original argument of Chern [29]. Let V
denote a connection on E compatible with a metric g. The strategy of proof is very simple.
We will explicitly construct a closed form w 6 Q2';¢(E> such that

(i) Tr*w 1G QOIMI.

(ii) ggw €(Vl Z (2%) P f (-F(v)).
The Thom isomorphism theorem coupled with (i) implies that w represents the Thom class
in H2p"'t(E). The condition (ii) simply states the sought for equality stop = e g e r .
Denote by S(E) the unit sphere bundle of E,

S(E) Z {'u, 6 E , IUI9 1}.

Then S ( E ) is a compact manifold, and
dimS(E) = d i m M + 2 k - 1 .
Denote by fro the natural projection S(E) -> M, and by fro (E) -> S (E) the pullback of E
to S ( E ) via the map two. The vector bundle w0 (E) has an SO(2k)-structure, and moreover,
it admits a tautological, nowhere vanishing section
T SlEx)9€*->€€Ex (7f0(E)x)e (co E M ) -
Thus, according to Example 8.3.15 we must have

egeom (770 (E) ) 0 E H2'"(s(E)),

 Characteristic Classes 365

where geom (fro E) denotes the differential form

l
geom (fro V) (27f)k P f (-F(710V))-

Hence there must exist w 6 Q21<-1(5(E)) such that

do = egeom(7r0E).

The decisive step in the proof of Gauss-Bonnet-Chern theorem is contained in the follow-
ing lemma.

Lemma 8.3.18. There exists \11 \11 (V) 6 Q2""1($(Ell such that

d\If(V) I €9€ovf(7T0(E)) (8.3.5)

and

m y ) = -1 6 Q"(M)- (8.3.6)
S(E)/M

The form \P(V) is sometimes referred to as the global angular form of the pair (E, V).
For the clarity of the exposition we will conclude the proof of the Gauss-Bonnet-Chern
theorem assuming Lemma 8.3. 18 which will be proved later on.
Denote by T' : E -> R+ the norm function
E 9e»-> lel9.
If we set E0 E \ { zero section }, then we can identify

E0'§(0,oo) X S(E) e»-> I6 I 7

Consider the smooth cutoff function

P = 0(1°) : [0»oo> -> R,

such that W) -1 for v i [0, 1/4], and p(rl = 0 form > 3/4. Finally, define
w = (»»(v) = -p'(7")dr A W) - p(~)#(e(v))
The differential form w is well defined since p' (T) E 0 near the zero section. Obviously w
has compact support on E, and satisfies the condition (ii) since

Cow Z -,0(0lCo7T*€(V) = 6(V)-

From the equality

p(t~we(v) = 0,
E/M

we deduce

IE/m IE/m
p'(1°)d1° /\ W)
I°° 0
0'(1°ld7"
/
S(E)/M
\11 (V)
366 Lectures on the Geometry of Manzfolds

(8.3.6)
==-(p(l)-~p(0l) \I/(V) l.
S(E)/M

To complete the program outlined at the beginning of the proof we need to show that Ld is
closed.

do I p'(r)d7° A d o ) - 0'(t") A or*elVl

( ¢ . ) *
8 3 5 p'(f~)dti A mew) Tt*6(V)}-

The above form is identically zero since ¢r0e(V) = ¢r*e(V) on the support of p' . Thus w
is closed and the theorem is proved.

Proof of Lemma 8.3.18. We denote by V the pullback of V to fro E. The tautological

section T s(E) -> tr0E can be used to produce an orthogonal splitting
%E L@Ll 7

where L is the real line bundle spanned by T, while L_L is its orthogonal complement in
7r0 E with respect to the pullback metric g. Denote by

P 7r0 E -> 7r0 E

the orthogonal projection onto LJ-_ Using P, we can produce a new metric compatible
connection V on 7r0 E by
= (trivial connection on L ) ® POP.
We have an equality of differential forms
1
1r0*e(V) :: V Z
p f (-Fw.

Since the curvature of splits as a direct sum

FW) I 0 Q FW),
where F'(Vl denotes the curvature of IL we deduce 7

P f I F I I = 0.
We denote by Vt the connection V -|- t(V - e) so that V0 = V, and V1 V If Ft is
the curvature of Vt, we deduce from the transgression formula (S.1.12) that

7r0 e(V) = e(V) - e(V) = d

111111»

Pf(V v, Ft 7 ,Ftldt

We claim that the form

k 1

W) k P f (V FR ,F")dt
0

satisfies all the conditions in Lemma 83.18.

 Characteristic Classes 367

By construction,
div) = 7V0@(V)7
so all that we need to prove is

m y ) = -1 6 Q°(M)~
S(E)/M

It suffices to show that for each fiber Eas of E we have

\I/(V) l.
Et

Along this fiber or0 E is naturally isomorphic with a trivial bundle

7I*o
EE u _ <E" xEX
_>
E ).

Moreover, the connection restricts as the trivial connection.

ac By choosing an orthonormal
basis of E33 we can identify 7r0 E | E with the trivial bundle RQ," over lR2":. The unit sphere
S(E,,,) is identified with the unit sphere S2k-1 C R 2 k _ The splitting L ® LJ- over S(E)
restricts over S(Er/,) as the splitting
R2k
I/ Q; TS2/<3-1 7

where u denotes the normal bundle of S2k-1 <-> lR2k. The connection V is then the direct
sum between the trivial connection on u, and the Levi-Civita connection on TS2k-1.
Fix a point p 6 S2""1, and denote by (x1, . . . ,:z:2""1) a collection of normal co-
ordinates near p, such that the basis (3,,,'i In) is positively oriented. Set <% a for
' £ = l , . . . , 2 k - l . Denote the unit outer normal vector field by 30. For a = 0 , 1 , . . . , 2 k - l
set fa = 30, 10. The vectors f a form a positively oriented orthonormal basis of R .
We will use Latin letters to denote indices running from l to 2k - l, and Greek letters
indices running from 0 to 2k - l.

Vi as (¢<%)" + (V<90,)T 7

where the superscript I/ indicates the normal component, while the superscript T indicates
the tangential component. Since at p

we deduce

V;5% (V¢3j)" at p.
Hence
as l z g j a <90)<90 l(9j, @30)30.
Recalling that is the trivial connection in R2k , we deduce

Via() p <90 In f i a Ip
368 Lectures on the Geometry of Manzfolds

Consequently,
iii -(5ji@07 at p.

If we denote by 92' the local frame of T*S21*'1 dual to (am), then we can rephrase the above
equality as
w5 I -(al _l_..._l_92k-1l (X) 30.

On the other hand, 'L 90 - 31, i.e.,

W0 = 91 al -|- - - -|- 9 - 1
(X) (X) 32k-1-

Since (x1, ---

,ac2k-1) are normal coordinates with respect to the Levi-Civita connection
V, we deduce that VIo = 0, Va so that
0 91 92k 1
91 0 0
92 0 0
A (v v)
92k 1
0 0

Denote by F0 the curvature of V0 = at p. Then F0 = 0 ® R, where R denotes the

Riemann curvature of at p.
The computations in Example 4.2.22 show that the second fundamental form of the
embedding
S2k-1 <_> R2k
7

coincides with the induced Riemann metric (which is the first fundamental form). Using
Teorema Egregium we get

(R(31, c'3i)<9k, 63)

In matrix format we have
F0 = 0 ® QuI)
where Q = i /\ 93 . The curvature F* at p of V* = -|- tA can be computed using the
equation (8.1.11) of Subsection 8.1.4, and we get
Ft = F 0 + t 2 A A A = 0 ® (1 -t2lF0
We can now proceed to evaluate \P(Vl.
k 1

WV) In k p f (A, (1 - z2)F0, - -- , ( 1 - t2)F0)dt

0
k 1
p f (A, FO, FU, . . . , FO).
k
(/ 0
(1 - t2)k-1dt (8.3.7)

We need to evaluate the pfaffian in the right-hand-side of the above formula. Set for sim-
plicity F ¢ = F0 .
 Characteristic Classes 369

Using the polarization formula and Exercise 2.2.65 in Subsection 2.2.4, we get

(-1)'<
P f (A,F,F, ,F) 2I*k! € l U l A 0 o 0 1 F0203 `F02k-202k-1°

065%

Ford 0,1, define

{0 E 82/ 7 Of = 0}.
We deduce

2'"/<! p f ( A , F , ,F) = (-1)/<= €(0)(-9U1) /\ 9°2 /\ 9°'3 /\ /\ 902k-2 /\ 90'2k-1

uQSO

-|- (-1)'" €<u>@"° /\ 902 /\ 903 /\ /\ 9U2k-2 /\ 90'2 k-1

0€S1

For each O' 6 5 0 we get a permutation

go: (O1,072,°°° ,O2A¢-ll 6 82k-17

such that 6(0) = €(qz5). Similarly, for 0' 6 Sl, we get a permutation

QUO I (U07 U27 • • a 1072k-ll Q 82k-17

such that @(0>==@(¢)_ Hence

2kk1 P f ( A , F, ,F) = 2(-1)'°+1 €(<1>)9¢1 /\ 9¢2k-1

<D682I¢-1

2(-1)'"+1(21<: - 1)!91 /\ /\ 92k-1 = 2(- 1 l k + 1 d v s 2 k - 1 7

S2k-1_
where dvs2k-1 denotes the Riemannian volume form on the unit sphere Using the
last equality in (8.3.7) we get

\P(V) In lk-|-1l2k- 1 l!dVs2k-.

(2k)!
l47I'lkk!
<1 t2lk'1dt dVs2k-1
0

Using the Exercise 4.1.59 we get that

and consequently
(2k)!
52k 1
W) W2k 1
(47r)k*k!
<1 t2)k-1dt
0

t2)k-1dt .
370 Lectures on the Geometry of Manzfolds

The above integral can be evaluated inductively using the substitution t = COS (p. We have
1 »'rr/2

In
It 0
(1 - z2)k-1dt (cos ¢)2k-1d¢

(cos up)
2k- 2 - 'rr 2 _|_
31119010/ lgk - QI 'rrr/2 (cos §0)2k-3(3in (p)2d(p
0

(2I< - 1>I,,_ - lzk - 2)1,,

so that
2k 3
In 21k_1.
2k
One now sees immediately that

q/(V) = - l .
S2k-1

Lemma 8.3.18 is proved.

Corollary 8.3.19 (Chem). Let (M,g) be a compact, oriented Riemann Manifold of di-
mension 271. If R denotes the Riemann curvature then
1
x(M) P f -R .
(2w)" M ( )
The next exercises provide another description of the Euler class of a real oriented
vector bundle E -> M over the compact oriented manifold M in terms of the homological
Poincaré duality. Set T' = rank E, and let TE be a compactly supported form representing
the Thom class.

Exercise 8.3.20. Let <I> : N -> M be a smooth map, where N is compact and oriented. We
denote by <I># the bundle map <I>*E -> E induced by the pullback operation. Show that
<I>*TE := (<I>#)*TE 6 Q""(®*E) is compactly supported, and represents the Thom class of
<1>*E.

In the next exercise we will also assume M is endowed with a Riemann structure.

Exercise 8.3.21. Consider a nondegenerate smooth section S of E, i.e.,

(i) the set Z, = s-1(0) is a codimension 7° smooth submanifold of M, and

(ii) there exists a connection V0 on E such that
ne
Z
0 &1O
VOl s #
for any vector field X normal to Z (along Z).
 Characteristic Classes 371

(a) (Adjunction formula.) Let V be an arbitrary connection on E, and denote by No the

normal bundle of the embedding Z, <-> M. Show that the adjunction map
Up : NZ -> E | Z )
defined by
XI->VxS x€c °°(n 2),
is a bundle isomorphism. Conclude that Z is endowed with a natural orientation.
(b) Show that there exists an open neighborhood N of Z -> NZ -> TM such that
exp : N -> exp(N)
is a diffeomorphism. (Compare with Lemma 7.3.44.) Deduce that the Poincaré dual of Z
in M can be identified (via the above diffeomorphism) with the Thom class of No .
(c) Prove that the Euler class of E coincides with the Poincaré dual of Z .

The part (c) of the above exercise generalizes the Poincaré-Hopf theorem (see Corol-
lary 7.3.48). In that case, the section was a nondegenerate vector field, and its zero set was
a finite collection of points. The local index of each zero measures the difference between
two orientations of the normal bundle of this finite collection of points: one is the orienta-
tion obtained if one tautologically identifies this normal bundle with the restriction of TM
to this finite collection of points while, and the other one is obtained via the adjunction
map.
In many instances one can explicitly describe a nondegenerate section and its zero set,
and thus one gets a description of the Euler class which is satisfactory for most topological
applications.

Example 8.3.22. Let U denote the tautological line bundle over the complex projective
space CIP". According to Exercise 8.3.10
C1('ul 1 et),
when we view U as a rank 2 oriented, real vector bundle. Denote by [H] the (Zn 2)-cycle
defined by the natural inclusion
QPU-1
'--)
(CHM 7 [Zo 7 . . . , Z , n - 1 ] l [ z 0 . .. ,z n -1 ,0 ]€ @
IP n

We claim that the (homological) Poincaré dual of 01 (Tn) is -1H]. We will achieve this by
showing that C1 lm) = -C1 (II) is the Poincaré dual of [H1.
Let P be a degree l homogeneous polynomial P 6 lC[z0,...,zn] . For each complex
line L -> (En-I-1 the polynomial P defines a complex linear map L -> (C, and hence an
9

element of L*, which we denote by P | L . We thus have a well defined map

CIP" 9 L »-> PILe L* = U* IL,
and the reader can check easily that this is a smooth section of II*, which we denote by [P] .
Consider the special case P0 = Zn. The zero set of the section lP0l is precisely the image
of [H1. We let the reader keep track of all the orientation conventions in Exercise 83.21,
and conclude that C1 (ml is indeed the Poincaré dual of [H] .
372 Lectures on the Geometry of Manzfolds

~
Exercise 8.3.23. Let Us = S2 \ {north pole}, and Un = S2 \ {south pole} = (C. The
overlap Un O Us is diffeomorphic with the punctured plane CC* = (I \ {0}. For each map
~
g : C* -> U(1) = S1 denote by Lg the line bundle over S2 defined by the gluing map

QNS : Un re Us U(1), 9NS(Zl = 9(2)-

Show that

01( g)
L =d t ,
$2
e 9
where deg g denotes the degree of the smooth map g |S1 be* -> S1.
 Chapter 9

Classical Integral Geometry

Ultimately, mathematics is about solving problems, and we can trace the roots of most
remarkable achievements in mathematics to attempts of solving concrete problems which,
for various reasons, were deemed very interesting by the mathematical community.
In this chapter, we will present some very beautiful applications of the techniques de-
veloped so far. In a sense, we are going against the natural course of things, since the
problems we will solve in this chapter were some of the catalysts for the discoveries of
many of the geometric results discussed in the previous chapters.
Integral geometry, also known as geometric probability, is a mixed breed subject. The
question it addresses have purely geometric formulations, but the solutions borrow ideas
from many mathematical areas, such as representation theory and probability. This chapter
is intended to wet the reader's appetite for unusual questions, and make him/her appreciate
the power of the technology developed in the previous chapters.

9.1 The integral geometry of real Grassmannians

9.1.1 Co-area formulae

As Gelfand and his school pointed out, the main trick of classical integral geometry is a
very elementary one, namely the change of order of summation. Let us explain the bare
bones version of this trick, unencumbered by various technical assumptions.
Consider a "roof"
X
B

A B
a B
where X -> A and X -> B are maps between finite sets. One should think of a and 5 as
defining two different fibrations with the same total space X .
Suppose we are given a function f : X -> R. We define its "average" over X as the
"integral"

(f)x f(33)-
a:6X

373
374 Lectures on the Geometry of Manu'olds

We can compute the average ( f }X in two different ways, either summing first over the
fibers of a, or summing first over the fibers of 5, i.e.,

2 2
a A (0(:1:)=a
f(33l
) (f)x
boB
2 2
( 5(2vl=b
f(f/3)

We can reformulate the above equality in a more conceptual way by introducing the func-
) (9.l.1)

tions
a * l f l : A - M C , 0,*(f)(0)= f(f/3)»
a(:13)=a

and
I5'*(f) B ->(C, B*(f)(b)= f(<13l-
5(r)=b
The equality (9.1.1) implies that
(O'*lfl)A =(@*(f))B.
All the integral geometric results that we will prove in this chapter will be of this form. The
reason such formul2e have captured the imagination of geometers comes from the fact that
the two sides (a* ( f ) ) A and (B* (al) B have rather unrelated interpretations.
The geometric situation we will investigate is a bit more complicated than the above
baby case, because in the geometric case the sets X, A, B are smooth manifolds, the maps
a and 5 define smooth fibrations, and the function f is not really a function, but an object
which can be integrated over X , i.e., a density. It turns out that the push forward operations
do make sense for densities resulting in formulas of the type
( H x = (a*lfl)A°
These formulae generalize the classical Fubini theorem and they are often referred to as
coareaformulce for reasons which will become apparent a bit later. To produce such results
we need to gain a deeper understanding of the notion of density introduced in Subsec-
tion 3.4.1.
Suppose V is a finite dimensional real vector space. We denote by det V the top exterior
power of V. Given a real number S we define an 5-density on V to be a map A : det V -> R
such that
A(tQ) = l¢ISAlQ), W 6 R*, Q Q det Vi
We denote by IAIS(V) the one dimensional space of s-densities. Note that we have a
canonical identification IAIo(E) = R We will refer to 1-densities simply as densities, and
we denote the corresponding space by IAI ( V ) .
We say that an s-density A : det V -> R is positive if
A(det V \ 0) C (0, oo).
We denote by III+(V) the cone of positive densities.
Note that any basis 'U17 . . . , U " of V defines linear isomorphisms
)
IAlsV- )\»_>A(v1
> IRE., A° Asn .
 Classical Integral Geometry 375

In particular, we have a canonical identification

IAI S(R") 2 R, A +-> A(€1 A • • A en),

where (el, . . . , en) is the canonical basis of R" .

If V0 and VI are vector spaces of the same dimension n, and g : V0 -> VI is a linear
isomorphism then we get a linear map

g* : IAls(V1) -> Ills(%l» IAls(V1) 9 A +-> g*A,

where

( Q ) (/WMI I Al/\z'lg'l)¢l )» \V/U17 7 vn€%~

If V0 = VI = V so that g 6 Aut(V), then

9*>\ ldetglSA.

For every g, h 6 Aut(V) we gave (oh)* h*g*9 and thus we have a left action of Aut(V)
on IAIS(V)
Aut(V) X IAIS(V) -> IAIS(V)7

Aut(V) X IAls(V) 9 (9, A) »-> QM = (9`1)*% = ldetgl'sA.

We have bilinear maps

IAI5(V) (X) III¢(V) * |A|s-I-tMv (Ml) *- A' ,U-

To any short exact sequence of vector spaces

0 U i v 15w a d l l l. u Ma d i V = Ma dimW=p
we can associate maps

lAI+(u) X IAls(V) -> IAIS(w),

I=IAls(V) X III+(w) -> Ills(Ul»

and
X IAls(U) X IAls(W) -> IAls(V)
as follows.

. Lets, 6 III+(U), A 6 IAIS(V)9 and suppose

lifts 'Up 6 V o f we, such that Blvj)
lwj)1SjSp
We, and a basis
is a basis of W. Now choose
of U such that
l'UȢ)1gi$1m,

a(u1), . . . ,a(um71), U1 7 • • ¢ 7 'Up 7

is a basis of V. We set

(MW ( Ajwj)
*( (/\¢O¢(U¢l ) A ( / \ j ' U j l
m( Am )
It is easily seen that the above definition is independent of the choices of v's and 'u,'s.
376 Lectures on the Geometry of Manzfolds

. Let A 6 IAIS(V) and u 6 LAI+(W). Given basis l'u»i)1gi$vm of U, extend the linearly
independent set ( ) C V to basis
{a(u1), . . . , a(um),01, :Van

of V and now define

*( (/\@O¢(U¢)) /\ (/\j'Uj) ) .
(Aw) ( I\i'LLi ) :z (91.2)
lal I\i5('Vi) )
. Again it is easily verified that the above definition is independent of the various choices.
Let I.L 6 IAIS(U) and u 6 IAIS(W). To define I.L x u : det V -> R it suffices to indicate
its value on a single nonzero vector of the line det V. Fix a basis (Uillgfzgm of U and
a basis l w j l l é j é p of W. Choose lifts (vi) of we to V. Then we set
Ill X v)( (A¢@(U51) /\ lAJwj)l i ,LLI/\¢Uil1/I/\jUjl.
Again one can check that this is independent of the various bases (UU and (We

Exercise 9.1.1. Prove that the above constructions are indeed independent of the various
choices of bases.

Remark 9.1.2. The constructions \, /, and X associated to a short exact sequence of vector
spaces
> V B> w -> 0,
a
0 -> U
do depend on the maps a and B! For example if we replace a by Ut to and 5 by
B7' = TB, t,T > 0, and if
l~* G IAI5(U)> A G I A l5(V)» I/ G IAI3(W),
then
\0¢»5-r A (t/T)S#\o,@ A, A/Omg-r I/ (t/7')SA/01,5 VI i

A x O r t 5 V = (t/7')-sp X05 u.
The next example illustrates this.

Example 9.1.3. Consider the short exact sequence

a 5
0->U=R > v=R2 >W=]R->0
given by
@(s) = (45, 105), l3(w,y) = 5:15 - 23/-
Denote by e the canonical basis of U, by (el, et) the canonical basis of V and by f the
canonical basis of W. We obtain canonical densities ,\U on U, AV on V and AW on W
given by
/\Ul€l I\V(€1 /\ 6 2 ) = w ( f ) 1.
I I

We would like to describe the density >V/5*,\W on V. Set

.ii 0(e) = (4, 10).
I

We choose f 2 6 V such that B(f 2) = f , for example, f2 = (1, 2). Then

A v / 5 * w ( € l = Av(f1 /\ f 2)/>\W(f )
4 1
ldetl H
N
II

10 2
Hence Av/@*Aw = ZAp.
 Classical Integral Geometry 377

Suppose now that E -> M is a real vector bundle of rank n over the smooth manifold
M. Assume that it is given by the open cover (Ua) and gluing cocycle
QBQ : U0g -> Aut(V),
where V is a fixed real vector space of dimension n. Then the bundle of s-densities associ-
ated to E is the real line bundle IAISE given by the open cover (Ua) and gluing cocycle

ldetg50|-3 : U0;-3 -> Aut( IAIS(V)) 2 R*

We denote by cOO (IAISE) the space of smooth sections of IAISE. Such a section is given
by a collection of smooth functions Aa : Ua -> IAIS(V) satisfying the gluing conditions

A;-xlflil 1 (DB 1)*>\o(g3) 1 ldet g5ol-sAol£ul, Va,B, at 6 UQ;-3.

Let us point out that if V = R", then we have a canonical identification IAIS(]R") -> R
and in this case a density can be regarded as a collection of smooth functions AO! : Ua -> R
satisfying the above gluing conditions.
An 5-density A E C'°°(IAIS E) is called positive if for every : B E M we have My) E
III+(E,,,l.
If QUO : N -> M is a smooth map, and E -> M is a smooth real vector bundle, then we
obtain the pullback bundle 1r*E -> N. We have canonical isomorphisms

IAI S E § tr* IAISE,

and a natural pullback map
~
¢* =0*°°(IA15 E) -> C'°°(tr* Ills E) = 0°°(IA1 tr*E).
Given a short exact sequence of vector bundles

0->E0->E1->E2->0

over M , we obtain maps

\¢C(>°(IAI+ EQ) X 6®(IA18 El) -> C7°°(IAI5 ET)»

I:COO>(IA1S El) X 0'>°(IA1+ EQ) -> COOllAI5 Eol,

and

X :COO(l AIS Et) X C°°(IAlsE2) -> C`°°(IAI5 E1)-

Observe that for every positive smooth function f : M -> (0700) we have

(fml\A i (f-1(/\A\, ,v(fy) i (f"1)(/1/M

In the sequel we will almost exclusively use a special case of the above construction, when
E is the tangent bundle of the smooth manifold M. We will denote by IAIS(M) the line
bundle IAIS(TM), and we will refer to its sections as (smooth) s-densities on M. When
s = 1, we will use the simpler notation IAIN to denote IAI1(M).
378 Lectures on the Geometry of Manzfolds

As explained in Subsection 3.4.1, an s-density on M is described by a coordinate atlas

IUQ, l, and smooth functions Ao : Uo, -> R satisfying the conditions
/ ii \

AB Ida I -SAG )
where d = det 1 , n dim M. (9.l.3)
1<'é,j§n

We deduce that the smooth 0-densities on M are precisely the smooth functions.

Example 9.1.4. (a) Suppose w 6 Q " ( M ) is a top degree differential form on M. Then in
a coordinate atlas (Um (1112)) this form is described by a collection of forms
wa = Aadxl A~~»Adwg.

The functions Aa satisfy the gluing conditions AB = do 1A0, and we conclude that the
collection of functions IAalS defines an s-density on M which we will denote by lwlS.
Because of this fact, the s-densities are traditionally described as collections
Aldmols, data == data A-_-Adwg.
(b) Suppose M is an orientable manifold. By fixing an orientation we choose an atlas
(UQ, (2312)) so that all the determinants d g , are positive. If w is a top dimensional form
on M described locally by forms wa = Aa data, then the collection of functions Aa defines
a density on M. Thus, a choice of orientation produces a linear map

§2"(M) -> 0°°(|A|(M))-

As explained in Subsection 3.4.2, this map is a bijection.
(c) Any Riemann metric g on M defines a canonical density on M denoted by ldvg I
and called the volume density. It is locally described by the collection

V lgO£lld:EC!l7
where lgOI denotes the determinant of the symmetric matrix representing the metric g in
the coordinates (HIL)-

The densities on a manifold serve one major purpose: they can be integrated. We
denote by C0(IAI (Ml ) the space of continuous densities with compact support. Consider
the integration map defined in Subsection 3.4.1

C@(IAI(Mll -> R, Idol »-> Idul-

M M

Note that if f is a continuous, compactly supported function on M, and ldIJ,l is a density,

then f ldlu,l is a continuous compactly supported density, and thus there is a well defined
integral

f|d/J»|-
M
We obtain in this fashion a natural pairing

CMM) X C1(IAlm), (f,ldul)+-> mf|diJ»|-

 Classical Integral Geometry 379

Let us observe that if ldpl and we are two positive densities, then there exists a positive
function f such that

Idol = fldfl-
The existence of this function follows from the Radon-Nycodim theorem. For every :1: 6 M
we have

lim
few ldp
f(33l u->{x} fUld¢l 7

where the above limit is taken over open sets shrinking to x. We will use the notation

f : Idol 7
do'
and we will refer to f as the Jacobian of ld,0l relative to ldTI. If <l> : M -> N is a
diffeomoiphism and ldpl = (UQ, Po, Idyll is a density on N, then we define the pullback
of ld,0l by (I) to be the density go* ldpl on M defined by

d)*ldf0l = (¢>-1(uo),pldyl) 7 y; fzzgogb.

The classical change in variables formula now takes the form

/ N
ldpl
/M
(25*ldpl.

Example 9.1.5. Suppose (b : M -> N is a diffeomorphism between two smooth

L 0 6 Qm*(M), and lwl is the associated density. Then
dimensional manifolds,

¢*lw l Z |¢*w|-
Suppose a Lie group G acts smoothly on M. Then for every g 6 G, and any density
ldpl, we get a new density g* ldpl. The density ldpl is called G-invariant if
g*ldpl = Idpl, Vg € G.
Note that a density is G-invariant if and only if the associated Borel measure is G-invariant.
A positive density is invariant if the Jacobian g* ldpl is identically equal to 1,'v'g E G.
Id/>l
Proposition 9.1.6. Suppose ldpl and ld7'l are two G-invariant positive densities. Then the
Jacobian J Idol
Idol is a G-invariant smooth, positive function on G.
Proof. Let :BE M and g a G . Then for every open neighborhood U o f x we have
In add Idol
U
Idol / g(U)
ldpl,
/ U
we 9(U)
Idol 1
\.

ft] do
I I(U)

f g ( U ) Idol
7

and then letting U -> {:z:} we deduce

J(w)=J(9f11), V:1;eM,geG.
Corollary 9.1.7. If G acts smoothly and transitively on the smooth many'old M then, up to
a positive multiplicative constant there exists at most one invariant positive density.
380 Lectures on the Geometry of Manzfolds

Suppose <I> : M -> B is a submersion. The kernels of the differentials of <I> form a
vector subbundle TVM -> TM consisting of the planes tangent to the fibers of ®. We
will refer to it as the vertical tangent bundle. Since <I> is a submersion, we have a short
exact sequence of bundles over M.

0 -> TVM <-> TM <1>*TB -> 0.

Observe that any (positive) density ldul on B defines by pullback a (positive) density
<I>*ld1/l associated to the bundle ®*TB -> M. If A is a density associated to the verti-
cal tangent bundle TVM, then we obtain a density A X <I>* ld1/l on M.
Suppose ldlnl is a density on M such that <I> is proper on the support of ldnl. Set
k := dim B, T' := d i M - dim B. We would like to describe a density ®*ld/nl on B
called the pusnforward of ldll by <I>. Intuitively, ®* id/nl is the unique density on B such
that for any open subset U C B we have

U
®*ld,LLI Idul-
<I>-1(U)

Proposition 9.1.8 (The pushforward of a density). There exists a smooth density <I>* ld,a |
on B uniquely characterized by the following condition. For every density ld1/l on B we
have

'1>*ld#l = V,,ld1/l,
where VV G C°°(B) is given by

v,(b> ld~l/<1>*ld1/I.
..
II

<1>-1(b)

Proof. Fix a positive density ld1/l on B. Along every fiber Mb = <D`1 (b) we have a density
I
Idulb/<I>*ld1/I Q C O O ( III(Mb) corresponding to the short exact sequence
0 -> Tm1, -> (Tm)lm,, Mi (<1>*TB)IMb -> 0.
To understand this density fix at 6 Mb- Then we can find local coordinates (3/i)1$J$I< near
_ _ defined in a neighborhood V of 513 in M such that
b 6 B, and smooth functions (f13')1<@<¢~
the collection of functions Co' , iff) defines local coordinates near :zz on M, and in these
coordinates the map <I> is given by the projection (az, y) »-> y.
In the coordinates y on B we can write

Idl/I i »of(y)ldyl and ldP*l i Pmlx7 ylldx /\ dl.

Then along the fibers y = const we have
Pm (at, y) ld:1:l.
Idulb/®*ldul
08(9)
We set

v,<6> |du|b/@*|du|-
Mb
 Classical Integral Geometry 381

The function V, : B -> R is smooth function. We can associate to Vi the density VVld1/l
which a priori depends on I/.
Observe that if ld19l is another density, then there exists a positive smooth function
B -> R such that
ld/9l = wld1/l.
Then
V
ldulb I<1>* I d"| w- ldulb/<I>*ldyl, V0=w'1
SO that

v>ld/>l = V,,ld1/l.
In other words, the density VVld1/l on B is independent on I/. It depends only on Idul-
Using partitions of unity and the classical Fubini theorem we obtain the coareaformula
for densities.

Theorem 9.1.9 (Coarea formula). Suppose the <I> : M -> B is a submersion, ldvI is a
density on B and Idol is a density on M such that <I> is proper on the support of ldlul.

/M
Idol / B
®*ldpl
I(/ B <1>-1(b)
l/<I>*ld1/I Id1/l~ (9.l.4)

Remark 9.1.10. Very often the submersion <D : M -> B satisfies the following condition.
For every point on the base b e B there exist an open neighborhood U of b in B, a
nowhere vanishing form w G Qk' (U), a nowhere vanishing form Q € Q'<+""(./WU), ( M u :=
<I>-1 (UI), and a form 17 6 Q"(MU) such that

Q 17 /\ tr*w.

Then we can write id,LLI = pIQue, for some p E COO(MU). The form w defines a density lwl
on U, and we conclude that

Idol/f*lwl = pltll, <I>*ldul = flail,

where for every u G U C B we have

f(u) Idol/f*lwl 0IUI/lf*¢»JI plvl-

Mu Mu Mu

In particular, using (9.1.4) we obtain the generalized eoareaformula

MU
Idol
MU
ply A 7r* (»JI
/ U
f(u)l(»JI
lU )
U m
pM |(»J|- (9.l.5)
382 Lectures on the Geometry of Manzfolds

Example 9.1.11 (The classical coarea formula). Suppose ( M , g ) is a Riemann manifold

of dimension m -|- 1, and f : M -> R is a smooth function without critical points. For
simplicity, we also assume that the level sets Mt = f - 1 (15) are compact.
On M we have a volume density ldVgl. We would like to compute the pushforward
density f*ldvgl on R. We seek f*ldvgl of the form

f*ldvgl i v(t)ldtl,
where ldtl is the Euclidean volume density on R and 'U is a smooth function.
For t 6 R the fiber Mt is a compact codimension 1 submanifold of M. We denote by
ldl/2l the volume density on Mt defined by the induced metric it := glut. We denote by
Vf the g-gradient of f , and we set n := IvfI 1
Vf.
Fix to 6 R. For every point p 6 Mt0 we have df (p) go 0, and from the implicit
function theorem we deduce that we can find an open neighborhood U, and smooth function
931, . . . ,mm such that (f, 513l, . . . ,:z:M) are local coordinates on U. Then along U we can
write
ldvgl = pldf/\ dfvl /\ - - A d:1:ml,
where p is a smooth function. The vector field n is a unit normal vector field along Mt0 OU,
so that we have the equality

IUfl1M¢0 PlnJ (dfA da;1 /\ . A d:z:"*) lurwmt0 %

0IVfl° l(d:13 1A~. A d a m ) IUf1M¢0
Now observe that along U we have
p(dfA da:1 A ~ - - A d m m ) »0(f*dt /\ daze A Adair).
Hence

ldV9l/f*ldtl Z pl(d:u1 A . • • /\ do,

so that
1
umm num0 Ivfl udvgu/f*udtu and ldVgl/f*ldtl
IVfl
ldVt0l lufwmt0

Hence
1
f*ldvgl = v ( t ) l d t l , (9.1.6)
v(t)
Mt I f I Adm,
and we obtain in this fashion the classical coarea formula

/M
ldvg I
/U
R Mt
1
I of I www
an.
To see how this works in practice, consider the unit sphere S" C ]R"+1. We denote the
(9.1.7)

coordinates in ]R"+1 by (t, 5131, . . . ,:1;"). We let Pi 6 S" denote the poles given t = :l:1.
We denote by ldvnl the volume density on S", and by 7T : S" -> R the natural projec-
tion given by
(t,;v1,...,:1:") »->t.
 Classical Integral Geometry 383

P 1
9
9 t
1

O t p' I I
>

Fig. 9.1 Slicing a sphere by hyperplanes.

The map 'ii is a submersion on the complement of the poles, M = S" \ {Pi }, and 7r(M) =
(-1, 1). We want to compute or* ldvv,l.
Observe that 7/"1 (t) is the (in - 1)-dimensional sphere of radius (1 - t 2 l 1 / 2 . To find
the gradient V'/r observe that for every p e Sn the tangent vector V7r(p) is the projection
of the vector it on the tangent space TpS", because 3,5 is the gradient with respect to the
Euclidean metric on Rn-l-1 of the linear function 'or : Rn-1-1 -> R tr(t, x ) = t.
We denote by 9 the angle between it and TpS"', set p' = 7r(p), and by t the coordinate
of p' (see Figure 9.1). Then 9 is equal to the angle at p between the radius [0, p] and the
segment [p, pI]. We deduce
cos 9 = length [p,p/] = (1 - t2)1/2
Hence

IVy(p)l = (1 - t2)1/2 7
and consequently,
1
T-1(¢l I Vi | Adm
= <1 >-1/2 we = 0,,_1<1 -t2)
TL
2
2
7

r-wt)

where am denotes the m-dimensional area of the unit m-dimensional sphere am_ The last
formula implies
1 1 \

n 2 TL 2 1 n
O' n 20'n (1-t2) 2
Idol Un 1 (1-s) 2 S-1/2ldSI Un 1B I 7
2 2
7
0 0 \

where B denotes the Beta function

1

L 513 1II - s)q-1ds, p, q > 0. (9.l.8)

o n

Bop, q)

It is known (see [134], Section $2.4l) that

OO up do: F(P)F(ql
Bop, Q) 7 (9.l.9)
0 (1 _|_ £I3lp+q QUO F(P -|- q)
384 Lectures on the Geometry of Manzfolds

where I`(;z;) denotes Euler's Gamma function

F(fv) €":t513-1dt.
0
We deduce
2F()H+1
O'
n - r(ngl) (9.l.10)

The equality (9.1.7) can be further generalized. We start with a simple linear algebra
result.

Lemma 9.1.12. Suppose that U and V are Euclidean spaces of dimensions n + k and
respectively k, n 2 0, and A : U -> V is a surjective linear map. Then there exist
Euclidean coordinates bel, . . . ,;13""-k on U, Euclidean coordinates y l , . . . ,gk on V and
positive numbers ,u1, . . . , ,up such that, in these coordinates the operator A is described by

ye l.LjfDj 7 1§ J S If.
The numbers #21 . . . , p are the eigenvalues of the positive symmetric operator AA*
V -> V so that

P'1°'°llk = \» detAin*

Proof. Let W denote the orthogonal complement of kerA in U. Denote by A0 the re-
striction of A to W so that A0 : W -> V is a linear isomorphism. Note that W coincides
with the range of the adjoint operator A* : V -> U so that
AoA; Z AA*.
We want to find a linear isometric R : V ->W such that the operator
B >= AoR : V ->
V
is symmetric. Note that since R is an isometric we have R-1 R* Moreover we have a
commutative diagram
Ao
w V
R
I I
V \
1
V
iv

Note that AoA* V -> V is positive and symmetric. We define

1/2
R := A0(A0,40) .• V -> W.
Let us show that R is indeed an isometric. Indeed, for any 'U 6 V we have
( R e , R v ) = (A0(A0A0)-1/2/0, A0(A0A0)'1/2u )
Z ((A@A0)-1/2), A@A0(A0A0)-Mv)
= l l A 0 A 0 ) - 1 / 2 U 7 I A 0 A 0 I 1 / 2 ' U l = (m).
 Classical Integral Geometry 385

Clearly AoR = A0A0(A0A0)-1/2 = (A0A0)112 is symmetric. Now choose an orthonor-

mal basis that diagonalizes B. Transport it via R to an orthonormal basis of W. With
respect to these bases of W and V the operator A is described by a diagonal matrix with
entries consisting of the eigenvalues of AoR = (A0A0)1/2.

Suppose now that A : U1 -> U2, dim U1 = n -|- lc, dim U2 = Is is a surjective linear
map between Euclidean spaces. Let lit 6 III(U¢) denote the Euclidean density on U p ,
i = 1, 2. Set U0 := her A and denote by Mo the Euclidean density on U0 .
Using Lemma 9.1.12 we can find Euclidean coordinates 513l, . . . ,:1:n+'° on U1, Eu-
clidean coordinates y1, . . . , yn on U2, and positive numbers up, . . . , ,on such that A is
described the equations,

Yj=lJ, j=l>-.~,H-

Then

/Io = id:v"+l A • - • /\ dx"+'"l, M1 = ld:1:1 A • • - /\ dm"+'"l,

1
# 2 - I d y A»-~Ady"l.

Consider now the short exact sequence of vector spaces

ii A
0 -> U0 <-> U1 > $2 -> 0,

where i is the canonical inclusion. Using (9.l.2) we deduce

l l
M2/A*M1 = ldxn+1 /\ A do:"`*'" | 0. (9.l.11)
I.L1 . • a ,Un JA M
Suppose now that (M, gel, (B, gel are Riemannian manifolds, <I> : M -> B is a submer-
sion and f : M -> R is a continuous function such that <I> is proper on the support of f.
Denote by ldVml the metric volume density on M and by dVB. We define its Jacobian to
be the function

J<I> : M -> R, ]<I>(P) \ det(D<I>(a:))*D<I>(p 7

where D<I>(p) is the differential of <I> at p. From (9.1.11) we deduce

that
1
ldVml/<I>*ldVNI l<>>-1(5) JI, ldV<1>-1(b)l»

where ldVq>-1 lb) | denotes the metric volume density along the fiber ®'1 (6), b 6 B.
From the coarea formula (9. 1.4) we deduce that metric coarea formula

M
fldvml W B
<I>_1(5) JI, <1> b B (9.l.12)
386 Lectures on the Geometry of Manzfolds

9.1.2 Invariant measures on linear Grassmannians

Suppose that V is an Euclidean real vector space of dimension n. We denote by the inner
product on V. For every subspace U C V we denote by PU the orthogonal projection onto
.
U. We would like to investigate a certain natural density on Gr I<:(V)» the Grassmannian
of k dimensional subspaces of V. We are forced into considering densities rather than
forms because very often the Grassmannians Grklvl are not orientable. More precisely,
Gr I<(Vl is orientable if and only if dim V is even.
In Example 1.2.22 we have shown that, via the (orthogonal) projection map U +-> PU ,
we can regard Gr I<(Vl as a submanifold of End+(v>, the vector space of symmetric
endomorphisms of V. The Euclidean metric (l.2.2) on End(V), defined by

( B ) = 51 to<AB*), VA,B 6 End(V),

induces a metric h = hn,k on Grk(V).

Denote by O(V) the group of orthogonal transformations of V. The group O(V) acts
smoothly and transitively on Grklvl

O(V) X Grklvl 9 ( 9 , L ) » - > 9 ( L ) € Gr I<(V)-

Note that

PAL = gPLg-1-
The action of O(V) on End+ (V) by conjugation preserves the inner product on End+ (V),
and we deduce that the action of O(V) on Grk (V) preserves the metric h.
We would like to express this metric in the graph coordinates introduced in Exam-
ple 1.2.22. Consider L G Gr k(V)9 and S Q Hom(L,Ll). Then, for every t € R, we
denote by Ut the graph of the map iS : L -> L-L ,

Ut := Pts € Grk(V,L).
If U0 denotes the tangent to the path t *->Ut at t 0, then
.2 .
he, Uo)
1 d
dt
| t 0
PHS?

If we write Pt I PHS we deduce from (1 .2.5) that

(HL -I-t2S*S)-1 t(11L + t2s*S)-15*
Pt
tS(]1L +t 2S *S )-1 t2s(]1L +t2s*s>-1s*
Hence

pt 0 S*PLJ_ -I- SPL, (9.1.13)

SO that
1
h(U0, Uo) 2(tr(SS*) + t1~(5*5l) = tr(SS*).

We can be even more concrete.

 Classical Integral Geometry 387

Let us choose an orthonormal basis l€-f2l1<i<1<: _ _ of L, and an orthonormal basis

leak<a§n of L-L_ With respect to these bases, the map S : L -> LJ- is described by
a matrix l 8 a ' é l 1 § i § k < 0 § n » and then
tr(SS*) tr(S*S) ISay I 2
i,a

We can think of the collection l3o¢'él as defining local coordinates on the open subset
Gr k(V, L) C Gr k(Vl consisting of k-planes intersecting L-L transversally. Hence
h(U0, UQ) l301il 2 . (9.l.l4)
i,a

In integral geometric computations we will find convenient to relate the above coor-
dinates to the classical language of moving frames. In the sequel we make the following
notational conventions.

.. We will use lower case Latin letters ii, j, to denote indices in the range {1, ..., k}.
We will use lower case Greek letters a m to denote indices in the range

. {k+l,...,n}.
We will use upper case Latin letters A,B,C,
{1 . . . ,n}.
to denote indices in the range

Suppose we have a smooth l-parameter family of orthonormal frames

(lAI = (pA(il)» Itl < l.

This defines a smooth path
to-> Lt = span (ei(t)) 6 Gr /<=(Vl-
We would like to compute A(L0, LO).
Observe that we have a smooth path t »-> it E O(V), defined by

gieAl0l i kAlil.
With respect to the fixed frame (eA(0) ), the orthogonal transformation it is given by a
matrix ( sAB(t) ), where
SAY eA(0) . l£1t€Bl0ll-
Observe that 90 iv. Let Pt denote the projection onto Lt. Then
-1
Pt g¢P09¢ 7

d
so that, if we set X = -
dt lt=09t9 we have
p0 I [X, P0].
With respect to the decomposition V = L0 -|- Lo_L the projector P0 has the block decompo-
sition

Po
388 Lectures on the Geometry of Manzfolds

The operator X is represented by a skew-symmetric matrix with entries

SUAB = S A B = $A . dB»

which has the block form

II
7

I0-,145

where XL6L,L0 denotes a map L0 -> LoJ_ etc. We deduce

lx,p@1

Hence
hlL0,L0> = 12 tr(P0, P0) = tIllXL6L,L0XLO_L,L0) I 'éavl I 2 (9.l.l5)
a,i

We want to interpret this in the language of moving frames.

Suppose M is a smooth m-dimensional manifold and L : M -> G r k ( V ) is a smooth
map. Fix a point Po 6 M and local coordinates ( U l 1 <_i <_m near Po such that ,ufi (pg) = 0.
The map Lt can be described near Po via a moving frame, i.e., an orthonormal frame
(eA) of V, varying smoothly with (nil, such that
L(u) = span ( ei(u) ).

The above computations show that the differential of L at Po is described by the (n - k) X k

matrix of 1-forms on M
DpoL = Wail, 90 ii ea . dB.
More precisely, this means that if X = ( W ) E TpoM> and we let " " denote the differen-
aviation along the flow lines of X, then
Dp0Llxl = ILUmI G TL(0) Gr : V I a Mai ea . ii be
a
. derL'lx)_ (9.l.16)

If M happens to be an open subset of Gr I<(Vl , then we can use the forms Had to describe
the metric h. More precisely, the equalities (9. 1.14) and (9. 1.15) show that
h QM (X) Q a i - (9.1.17)
0,i

In other words, the collection We) is an orthonormal frame of the cotangent bundle
T* Gr kM.
The metric h = hn,k is O(V)-invariant so that the associated Riemannian volume
defines an invariant density. We will denote this invariant density by id'Yn,kl9 where n =
dim V, and we will refer to it as the kinematic density on Gr k M . Since the action of
the group O(V) is transitive, we deduce that any other invariant density is equivalent to
a constant multiple of this metric density. We would like to give a local description of
ld'Yn,/<:|-
 Classical Integral Geometry 389

Set n := dim V. If 6 is a sufficiently small open subset of Grk(V) then we can find
smooth maps
eA=@°->v, A = l , - - - , n ,

with the following properties.

. I
. For every L 6 6" the collection eA(L) )1<A<n is an orthonormal frame of V.
For every L 6 6" the collection Le©(Ll llgzgk is an orthonormal frame of L.
For every l < i g lc, and every k + l < a < l, we have a l-form
e 6 QUO(@), Qui = ea . de = -dea . € .

As explained above, the metric h is described along 6 by the symmetric tensor

h= gym ® H ,
a,i

and the associated volume density is described by

ld'Yn,kl H 90i
..
II

Qui
0,i 0,11

Example 9.1.13. To understand the above construction it is helpful to consider the spe-
cial case of the Grassmannian Gr1(lR2) of lines through the origin in R2. This space is
diffeomorphic to the real projective line RIP1, which in turn, it is diffeomorphic to a circle.
A line L in R2 is uniquely determined by the angle 9 6 [0, Tr] that it forms with the
cc-axis. For such an angle 9, we denote by LE the corresponding line. The line L0 is also
represented by the the orthonormal frame

e1(9) = (cos9,sin 9), e2(9) = (- sin 9,cos 9), L0 = span(e1(9) ),

where the first vector 81 (9) gives the direction of the line. Then 921
ld")/271l = Idol.
= €2 . d81 = d9, and

We would like to compute the volumes of the Grassmannians G r k ( V ) , dim V n

with respect to the kinematic density ld'7n,kl9 i.€., we would like to compute
..
II

Cn,k ld'Y'n,k|-
Grk(vl

Denote by can the volume of the unit ball B" C R", and by U n 1 the 1)-dimensional
"surface area" of the unit sphere .SITL-1, so that
'ITk
n Qk,
F(l/2)'"
k! 7
Un 1 non , and can U1 7
+n/2 ) 22k-*-17Tk]€!
, = 2k + l ,
(2/f + 1)! n
390 Lectures on the Geometry of Manzfolds

where I`(;z:) is the Gamma function. We list below the values of can for small n.
n 0 l 2 3 4
Un l 2 or M
3
-71'2

2
I

I } .

To compute the volume of the Grassmannians we need to use yet another description of the
Grassmannians, as homogeneous spaces.
Note first that the group O(V) acts transitively on Gr k(Vl- Fix Lo Q Gr klvl~
Then the stabilizer of L0 with respect to the action of O(V) on Gr I<(V) is the subgroup
O(L0) X O(L(. and thus we can identify Gr /K(V) with the homogeneous space of left
cosets OW)/O(L0l x O(L5L).
The computation of Cn,k is carried out in three steps.
Step 1. We equip the orthogonal groups O(lR'") with a canonical invariant density ld'Y'n I
called the kinematic density on O(n). Set

Cn Id%nl-
..

O(R'*)

Step 2. We show that

Cn
Cn,k
CoCn-k
Step 3. We show that
1 man
Cn,1 7
2
and then compute Cn inductively using the recurrence relation from Step 2

Cn-i-1 = lC1Cn,1lCna

and the initial condition

C2 = vo1(O(2l) = 261.

Step 1. The group O(V) is a submanifold of End(Vl consisting of endomorphisms S

satisfying SS* = S*S = Ly. We equip End(Vl with the inner product
l
( B ) tr(AB*).
2
This metric induces an invariant metric h on O ( V ) . We would like to give a more concrete
description of this metric.
Denote by End' (V) the subspace of End(V) consisting of skew- symmetric operators.
For any SO 6 O(V) we have a map
expo0 : End- (VI -> O(V), End' (V) I ) So • € X p l x ) .
This defines a diffeomorphism from a neighborhood of 0 in End' (V) to a neighborhood
of So in O(V). Two skew-symmetric endomorphisms X, Y E End- (V) define paths
'YX7'YY : R -> O(V), 'yX(t) = SO exp(tX), 'Yy(tl = SO exp(tY),
 Classical Integral Geometry 391

originating at 50. We set

X == *rx(0) Q Ts0O(V) C End(V), Y == '71/(Ol Q Ts0O(V) C End(V).

-
- SoX, Y - soy, and
Then X -
l l l
h(x,y) = 5 tr ( (s0x><s@Y)*) : 2 -.1~(50xy*s0 ) : 2 TISoSoXY* )
=-1 1;r (xy* ) .
2
If we choose an orthonormal basis (eA) of V so that X and Y are given by the skew
symmetric matrices (QCAB), (3/AB), then we deduce

h<X,Y) = HSABYJAB-
A>B

If we set ]fAr) exp(tX)eA, then we deduce

IIJAB = 8A . j°8(0) = fA(0) . j"8(0).

More generally, if we define

fA : O(V) -> V, f A(Sl Z= S B A ,

then we obtain the angular forms

description
TAB fA . d f 8 . The above metric metric has the

h= TAB ® 9AB-
A>B

The associated volume density is

'd'Yn I TAB
A>B

Step 2. Fix an orthonormal frame ( A l of V such that L0 = span (et ; 1 < i < of). We
can identify V with R", O(V) with O(nl, and L0 with the subspace

Rh ® On-k C R".

An orthogonal n X n matrix T is uniquely determined by the orthonormal frame lT€Al via

the equalities

TAB pA . TE8.

Define

p : O (n)-> Gr klfnl, P ( T l = T ( L 0 ) -

More explicitly, we have

p(T) - span l T € i ) 1 g i $ k -
392 Lectures on the Geometry of Manu'olds

We will prove that we have a principal libration

o<z<> X O(n - k) " \

J
O(n)
p 7

Gr 1<(R")

and that

I ldvnl I CkCn-kldyn,kl~
Once we have this we deduce from the Fubini formula (9. 1.5) that

Cn - CkCn-kCn,k-
Let us prove the above facts.
For every sufficiently small open subset U C Gr kM we can find a smooth section
QUO : U -> O(nl
of the map p : O(n) -> Grk,(R"). The section can be identified with a smooth family of
orthonormal frames l¢A(Ll, L E U ) 1 5 A g n o R " , such that

L = span (¢¢(L); 1 $ i < If).

To such a frame we associate the orthogonal matrix ¢(L) E O(n) which maps the fixed
frame leA) to the frame (¢8 I, It is a given by a matrix with entries
<I5(/UA8 = GA • Q58-
Then we have a smooth map
\P 0(1<) X O(n-k) X U->p'1(U)»
defined as follows.

. Given (s, t , Ll 6 O(k) x O('n, - Ki)

as a (n kl x (n k) matrix (to) .
X U, express 8 as a k X k matrix S (83% and t
. - . - .

Define the orthonormal frame of IRE."

( f A)2=l¢B) * (SJ),
via the equalities
j 1 < i </f,
ii f il3)L) Sf; (9.l.18)
j

L;
fa f (t,L) Z §jt"¢(L) E 7 k-I-1<0<n. (9.l.19)
B
. Now define \11 = \I/(s, t, L) to be the orthogonal transformation of R" which maps the
frame (GA) to the frame (dB), i.e.,
pA Z \I/€A: VA.
 Classical Integral Geometry 393

The map \11 is a homeomorphism with inverse

O(n) 9 T +-> Q/-1(T> = (s,t; L) € O(k)
O(n - k) X X L
defined as follows. We set pA f A(T) ¥ = TeA, l < A < n. Then
L LT span( .f i ) 1 $ ¢ $ k »
Z Z

while the matrices (so) and (to) are obtained via (9. 1.18) and (9. 1.19). More precisely, we
have
So Z ¢¢lLTl - f j) 5E-1" = ¢olLTl - f
Observe that, V30, S1 Q O(k), t0,t1 Q O(n - je), we have
((<l58) * (30,150)l * (Snell = (¢8) * ( S 0 S 1 » t 0 t 1 ) -
This means that \11 is equivariant with respect to the right actions of O(k) X O ( n - je) on
O(k) X O(n - k) X U and O(n). We have a commutative diagram
o(1<> x O(n - fl xU
\I/
> p'l(U)
1r L p
U C GI'klRn)
In particular, this shows that p defines a principal O(k) X O(n - k)-bundle.
Observe now that p* Idvnl is an invariant density on Grk (Rn), and thus there exists a
constant C such that
zhldvnl = 0|d'7n,k|-
This constant is given by the integral of the density ld'Yn|/P*|d'Yn,k| along the fiber
P`1(L0)-
Hence
Cn = Idvnl = p*ldvnl C Idmkl ; CCM-
O(n)
Recall that if we define f A : O(n) -> R" by f A(T)
G1'k(R"l GPklRn)
T € A , 3 I l d QAB fA . dfB» then

ld'Ynl =
A>B
TAB |
We write this as

/\ Had
The form l /\a,z' l
(>(l(/\ He
i>j
/\
Q>,B
15 /\
0,'é

is the pullback of a nowhere vanishing form defined in a neighbor-

)
hood of L0 in Grk (RTL), whose associated density is ld.*yTLkl. We now find ourselves in the
situation described in Remark 9.1 10. We deduce

C
" P 1(Lo)
I D )/\ Qu
i>j
A
a>,5'
H

(f O(k)
ldVkl
)(/ O(n-k)
ldVn-kl
u cucak.
394 Lectures on the Geometry of Manzfolds

Hence
Cn
Cn,k
CkCn-k

Step 3. Fix an orthonormal basis {eA} of V, and denote by S n+ 1

the open hemisphere
S n+ l
{6'€ V; III i 1, 'U . 81 > 0 }.
Note that Gr1(V) 2- RIP7n-1 is the Grassmannian of lines in V. The set of lines that do
not intersect S n+ 1 is a smooth hypersurface of GII1 (V) diffeomorphic to ]R]Pm-2 and thus
has kinematic measure zero. We denote by Gr 1(Vl* the open subset consisting of lines
intersecting S n+ 1 . We thus have a map
w Gr (V) -> S n-|- 1
7 6+-MO S n+ 1

This map is a diffeomorphism, and we have

Cn,1 ld'Yn,1l ld'Yn,ll (¢'1)*|dvn,1|-

Gr1(vl Gr{(v) s+
Now observe that w is in fact an isometric, and thus we deduce
Cn,1
1
= §area(S" 1l = Un 1
__ non
2 .
2
Hence
Cn-I-1 = CnClCn+1,1 = UHCH 7

which implies inductively that

n-1 'n-1

Cn--IO'n_1°°°U'2C2=2
H
k=1
Of H
j=0
U3-

In particular, we deduce the following result.

Proposition 9.1.14. For every 1 < k < n we have

Cn,k
H H UP

H )~( H we)

Following [79], we set

n
l Un 1 Town win!
[to]
2 Wn-1 20.1n-1
7 [up H la]
k 1
in 7

[nli- k
H
n
k l[ /€][
ll[
]!>
n
()k w
Un
(9.1.20)
 Classical Integral Geometry 395

Denote by d]/7L,k the unique invariant density on Gr n such that

(9.1.21)

We have

d]/n,k I ["]
Cr"kk ld'Ynk

Example 9.1.15. Using the computation in Example 9.1.13 we deduce

ld'Y2,1l = Idol, 0S9< 71,

which implies that

L02
02,1
0
Idol 2 2,
W1

as predicted by Proposition 9.1.14. We have

$2
2 2 02,17
W1

so that 1011/2,11 = |d72,1|-

9.1.3 Amine Grassmannians

We denote by Graff k(Vl the set of k-dimensional affine subspaces of V. We would like
to describe a natural structure of smooth manifold on Graffk (V).
We have the tautological vector bundle U = U n , k -> Gr A<(V). It is naturally a sub-
bundle of the trivial vector bundle L/'_ = V X Grk,(V) -> Gr I<(V) whose fiber over
L 6 Grk(V) is the vector subspace L. The trivial vector bundle _V is equipped with a
natural metric, and we denote by uJ_ -> GIlk(V) the orthogonal complement of U in _V.
The fiber of u_L over L G Gr MV) is canonically identified with the orthogonal com-
plement L-L of L in V. The points of uJ- are pairs (6, L), where L G Gr k(V), and Eis a
vector in L-L .
We have a natural map A UL -> Graff kM given by
(E, L) I > 5+ L.
This map is a bijection with inverse
Graff M) 9 S +-> ( S o [s we,
where [S] Q Grk (V) denotes the affine subspace through the origin parallel to S,
[S]=S-S={S1-32, 51,52-E 5 1.
We set
6(5) :z S re [s
and we say that 5(5) is the center of the affine plane S.
396 Lectures on the Geometry of Manzfolds

We equip Graffk (V) with the structure of smooth manifold which makes A a diffeo-
morphism. Thus, we identify Graff k(Vl with a vector subbundle of the trivial bundle
V X Grk(V) described by
ul = {(6,L) Q V X Gr MV); PLY'= 0 }»
where PL denotes the orthogonal projection onto L.
The projection or : u_L -> Gr MV) is a submersion. The fiber of this submersion
over L 6 Gr I<(V) is canonically identified with the vector subspace L-L C V. As such,
it is equipped with a volume density ldVLi |. We obtain in this fashion a density ldVLi |
associated to the vertical subbundle her Dur C Tel .
The base Grk(V) of the submersion $r : Graff 1<3(Vl -> Gr k(V) is equipped with a
density id'Y'n,kl9 and thus we obtain a density Tr* id'Y'n,kl associated to the bundle
7T*T Grk iv) -> Graff /K(V)-
Using the short exact sequence of vector bundles over Graffk (V),
0 -> her Do --> T Graffk(V) > 1r*TGrk(Vl -> 0,
we obtain a density n,k on Graff k(Vl given by
ldnkl = ldVLJ.l X '"*ld'Yn,kl.

Let us provide a local description for this density.

Fix a small open subset (9 C Gr k(Vl9 and denote by 5 its preimage in Graffk (V) via
the projection 7r. Then we can find smooth maps
6A=(9->v, 15->v
with the following properties.

. For every S E (5, (eA(S) ) is an orthonormal frame of V, and

[L] = span(e'l([L])).
. For every S Q 5 we have
S = 17(5) + [5].

We rewrite the last equality as

S I S(F, 8 1 ) .
Observe that the center of this affine plane is the projection of F onto [so

6(5)
a

Following the tradition we introduce the (locally defined) l-forms

Ha := ea . dF', pOi := ea . de.
For fixed L E Gr kM the density on the fiber ULJ_ = LJ- is given by
ldvLI = l/\@l
 Classical Integral Geometry 397

The volume density on Graffk (V) is described along 5 by

The Fubini formula for densities (9.1.4) implies the following result.

Theorem 9.1.16. Suppose f Graff k(Vl -> R i5 a compactly supported ld'"n,kl'

integrable function. Then

GraHlk(V)
f($)ld%,k(S)l
L (/ MV)
LJ. f(p + L)ldVL (1>)l ld.%,k(Lll»

where ldVL | denotes the Euclidean volume density on LJ-_

Denote by Iso(V) the group of affine isometrics of V, i.e., the subgroup of the group of
affine transformations generated by translations, and by the rotations about a fixed point.
Any affine isometric T : V -> V is described by a unique pair (t, S) 6 V X O(V) so that

T ( v ) = S v - I - t , VuGV.

The group Iso(V) acts in an obvious fashion on Graff I<(V)9 and a simple computation
shows that the associated volume density d ' n , , k l is Iso(V) invariant.
If instead of the density ld'Yn,kl on Gr I<(Vl we use the density l a I / n k l , we obtain a
density ld§,k1 on Gra{'Ilk1J(V) which is a constant multiple of ld.,,kl.

ld'z7'nkl cl["]uj A /

ld'7nk |- (9.l.22)

Example 9.1.17. Let us unravel the above definition in the special case Graff 1 (1Ra.2), the
Grassmannians of affine lines in R2. Such a line L is determined by two quantities: the
angle 9 6 [0, Tr) is makes with the :u-axis, and the signed distance ,0 E (-on,ool from
the origin. More precisely, for every p 6 R and 9 e [0, Tr) we denote by LE p the line is
given in Euclidean coordinates by the equation

xsin9-ycos9 =p.

As a manifold, the Grassmannian Graff1(]R2) is diffeomorphic to the interior of the

MObius band. The Fubini formula in Theorem 9.1.16 can now be rewritten
OO

Graffl (R2)
f(L) ld%,1l(L)
OO (/" flL9,plld9l
0
ldpl,

W' E Ceptl Graff 1(R2)

398 Lectures on the Geometry of Manzfolds

9.2 Gauss-Bonnet again?!?

The Grassmannians enter into the study of the geometry of a submanifold via a fundamental
observation going back to Gauss. To understand the shape of a submanifold M -MR" 7

m dim M, it is very productive to understand how a tangent plane TpM varies, as p

varies along M.
Observe that if M is flat, i.e., it is an open subset of an affine subspace, then the tangent
plane TpM does not vary as p moves around M. However, if M is curved, this tangent
plane changes its location in the Grassmannian Grm (W) , and in fact, "the more curved is
M, the faster will the tangent plane move inside this Grassmannian".
In Example 4.2.23 we have seen this heuristic principle at work in the case of a hyper-
surface in R3, where the variation of the unit normal along the surface detects the Riemann
curvature of the surface. Note that the tangent plane is uniquely determined by the unit
normal.
The goal of this section is to analyze in great detail the map
M 9p+->TpM€ Grm (R"),
which, for obvious reasons, is named the Gauss map of the submanifold M .

9.2.1 The shape operator and the secondfundamentalform

Suppose M is a smooth submanifold M of R", and set m dim M. To minimize the

..
II

notational burden we will use the following conventions.

. We will use lower case Latin letters i , j , . . . to denote indices in the range
l < i 7 j , . . . < m = dim M .
. We will use lower case Greek letters g? to denote indices in the range
m<a,B, <n.
. We will use the capital Latin letters A,B,C to denote indices in the range
l <A,B,C'<n.
. We will use Einstein's summation convention.

The Euclidean metric of R" induces a metric g on M. In Subsection 4.2.4 we ex-

plained how to determine the Levi-Civita connection VM of g using Cartan's moving
frame method. Let us recall this construction using notations more appropriate for the
applications in this chapter.
Denote by D the Levi-Civita connection of the Euclidean metric on R". Every vector
field on R" can be regarded as an n-uple of functions
XI

Xn
 Classical Integral Geometry 399

Then

I
'
Q

'U
11111111111111111

11111111111111111
II

II
The restriction to M of the tangent bundle TR" admits an orthogonal decomposition

(TO")lm i TM ® (TM)l,
Correspondingly, a section X of (TlR")l M decomposes into a tangential part XT, and a
normal part X". Fix a a point Po Q M, an open neighborhood U of Po in R", and a
local orthonormal frame (eA) of TR" along U such that the collection (€¢l M) is a local
orthonormal frame of TM along U re M. We denote by (9 A) the dual coframe of (eA),
i.e.,
9A ( € 8 ) = pA . dB = 5AB-

If X is a section of TR"l M then

x* = 9i(x)€1, x" i 9a(X)€&_
We denote by (go the l-forms associated to D by the frame (BAI- They satisfy Caftan's
equations
doA = -(9g /\ HB 7 DEB = @ g e t , G); = -@§.
If we write
A A , A
<25 0 All/Ia QB _ z ®BI]M7

then, as explained Subsection 4.2.4, we deduce that (<I>§) are the l-forms associated to the
Levi-Civita connection VM by the local orthonormal frame (et | M). This implies that
M i
V of = @jo = the tangential component of <I>jeA = D of ,
and thus,
VXMY = ( D X YV, VX,Y Q Vect l m ) .
Consider the Gauss map
9= am : M -> Gr ,,,,(R"), :u »->TasM.
The shape operator of the submanifold M --> R" is, by definition, the differential of the
Gauss map. Intuitively, the shape operator measures how fast the target plane changes its
location as a point in a Grassmannian.
We denote the shape operator by SM , and we would like to relate it to the structural
coefficients (pg .
As explained in Subsection 9.1.2, in the neighborhood U of Po, the "moving plane"
as +-> TasM can be represented by the orthonormal frame (eA) which has the property that
the first m vectors e1(;u), . . . ,eT7,(;z3) span T513M. The differential of the Gauss map at
QUO Q U D M i s a linear map

DO : TxM -> T9(a@) Gr m(R"") Hom(T,,M, ( H e ) .

400 Lectures on the Geometry of Manzfolds

As explained in (9. 1.16), this differential described by the (n- m x m matrix of l-forms

(to . de.7'la,1l : (e O D ey 'la,z

On the other hand, D ey <I>jA€A, SO that ea A D of = <1>§u. Define
<I>g == € J <1>§' Q Q0(m m U),
SO that

<1>§4=<1>gA¢z

We have thus obtained the following result.

Proposition 9.2.1. The shape operator of M , that i5 the deferential of the Gauss map, is
locally described by the matrix of 1-forms (i1'>§")19<_m<0<n. More precisely the operator
H mm ( T m l )
SMI 82) Q 0mlT )
i5 given by
I
SM le j i (D e l f
a
7
»j.

In Subsection 4.2.4 we have defined the second fundamental form of the submanifold
M to be the CYOO (M)-bilinear and symmetric map

Sm:VeCt(M) X vent<m>->c°°((Tm)_L), (X,Y)»->( D x Y ) " .

Note that

SMl€1i» by) = (25"'( D et e j l e a ' j ' Zo 5M(€¢l@J~ (9.2.1)

The equality (4.2. 12) can be rewritten as

ef . (D e r a ) = - ( D e i e j l . ea ea . $ml€1»€Jl» 'v'i,j, Q. (9.2.2)

Theorem Egregium states that the Riemann curvature tensor of M is completely deter-
mined by the second fundamental form. We recall this result below.

Theorem 9.2.2. Suppose M is a submanifold of R". We denote by g the induced metric on

M and by SM the second fundamentalform of the embedding M <--> R". Denote by R the
Riemann curvature of M with the induced metric. Then for any Xl, . . . ,X4 6 Vect ( M l
we have

9(X1»R(X3»X4lX2l = $m(X;»,,X1) SM(X4,X2) - $ m ( X 3 , X 2 ) . . SM(X4,X1),

.
where denotes the inner product in Rn.
 Classical Integral Geometry 401

9.2.2 The Gauss-Bonnet theorem for hypersurfaces ofan Euclidean space

The results in the previous subsection have very surprising consequences. Suppose M is
a compact, orientable 1 hypersurface of ]Rm-i-1 i.e., an orientable submanifold of codimen-
sion l. If we fix an orientation on M, then we obtain a normal vector field
n M -> R'ITz-i-1 7 )L TM, in(x)l = 1, Vi E m.
If we choose an oriented, local orthonormal frame e l , . . . , et of TM, then the ordered set
{n(ac), 81 (x), . . . , e t } is an oriented, local orthonormal frame of l'"+1. In this case we
can identify the second fundamental form with a genuine symmetric bilinear form

S M 6 C O O T*
_> ( M
®2
), 9
Y
5M(X ) = n ( D x y z . .
The Gauss map So : M Gr m(]R*"+1) of the embedding M <-->]R""*1 can be given the
description,

M 9 at »-> (n(ac°))i := the vector subspace orthogonal to

On the other hand, we have an oriented Gauss map

8M:]W->SVM" 7 cc->n(:z:),

and a double cover

S -> Gr mlRm+1), S m 91->lfE) 7

so that the diagram below is commutative

§ >
M Sm
7T
9
GrmlRm+1l

We fix an oriented orthonormal frame (fM, , fol

of Rm+1 and we orient the unit 7

sphere So C ]Rm+1 so that the orientation of Tfz Sm is given by the ordered frame

(fl, . • . fol,
7

In the remainder of this subsection we will assume that m is even, m = 2h.

Denote by dA,,,, E QM(S 771 the "area" form on the unit m-dimensional sphere Sm
Recall that um denotes the "area" of So,
O' 22h7II-hhI!
O' 2h (Qh + l)(.4.)2¢'+1 -->> 2oh (9.23)
(2h)!
Hence
l
(mTTI 1,
Sm O' 'rn
lThe orientability assumption is superfluous. It follows from the Alexander duality theorem with Z/2 coeffi-
cients that a compact hypersurface of a vector space is the boundary of a compact domain, and in particular, it is
orientable.
402 Lectures on the Geometry of Manzfolds

so that
1
llllllllllllll

Um M
§*M mm deg am.
Denote by g the induced metric on M, and by R the curvature of g. We would like to prove
that the integrand §MdAm has the form
rdAm P(Rm)dVm,
where dVm denotes the metric volume form on M and P(Rm) is a universal polynomial
m
of degree -2
in the curvature R of M.
Fix a positively oriented orthonormal frame é' = (el, . . . , et) of TM defined on some
open set U C M, and denote by 6 (0 1, . . . , 0 ml the dual coframe. Observe that
(We 01 A A 0 'rn

We set
SU Sm(€i» o i l . Na Rfzjke o l e , R ( e k , , eelej

-
Theorem 9.2.2 implies that
>'¢A¢ 512
Rijké S¢I<:Sje - S¢eSj1< (9.2.4)
sdk So I
Observe that Rtjke ;£ 0 1, k g é f , and in this case the matrix
>'ik 512

>'1l< Sn
is the 2 X 2 submatrix of S = (5¢j)1<¢ _ ,j<_m lieing at the intersection of the z', j-rows with
the k, 6-columns. We can rephrase the equality (9.2.4) in a more convenient form.
First, we regard the curvature Rtjke at a point 32 E M as a linear map

A2TxM -> A2T;m, R(ek /\ et) Z R i k e 0 'Z A 0 j

¢<j

Next, regard S as a linear map

S : Txll/I -> T * M , Séj - S 01
Then S induces linear maps
ANS : A'"T,,,m -> APT*M,
defined by
Slew1 /\ . ' . /\ Bps = (S€i1) /\ /\ lsépl, vi < 111 < <ip§m.

The equality (9.2.4) can now be rephrased as

R i A25. (9.2.5)

Along U we have the equality

(9mdAM) II= A'"S(61 A -. A ml = (dotS)
01 A /\
0 T7 (det s)dvm .
 Classical Integral Geometry 403

We want to prove that det S can be described in terms of A2S. To see this observe that
(A2h5l(€1 /\ 82 /\ ' ' • 8211-1 / \ € 2 h l (A2slle1 /\ 62> /\ . a /\ (A25)(€2h-1 /\ ezhl
\ ..
h h

Z
i<j
Rij,2s-1,25 91
/\ 0 j
(p€8'
€((p) H
s=1
R(p(2s-1)(p(2s),2s-1,25 dVm,

where S' denotes the set of permutations up of {l, 2, . . . , m = 2h} such that
(Ml) < (/)(2)» - -- 7 (/)(2h - l ) < (We),
and e(go) = i t denotes the signature of a permutation. Observe that
2h 2h-2_ 2 (22")l
2#8' 2
(9.2.6)

We would like to give an alternate description of det S using the concept of pfa]jian.
The Riemann curvature, like the curvature of any metric connection, can be regarded
as a 2-form with coefficients in the bundle of skew-symmetric endomorphisms of TM. It
is locally described by a m X m skew-symmetric matrix
G) = @9 . (GM :
1<'i,j<_1n

whose entries are 2-forms on U,

®in E Rilke 9 k /\ 0e 5l E Rilke 0 k /\ 0

e A2Sle7; /\ e i ) € Q 2 ( U ) .
k<€ 1¢,e

Recall from Subsection 8.2.4 that the pfaffian of ® is the differential form

pf@ (-II" Q2h'(U),

( ) 2hh1 do) ® <p(1)>(2) A A 9 (p(2h-1)(p(2n,) 6
'P€5m

where So denotes the group of permutations of {1, 2, .,m . Observe that

pre) (-1)h /\ G)gpl2h,_1l9P(2h'l 7
( ) Z hl €(90l®¢(1)9p(2) /\
. ¢€§

/I
we can simplify this some more if we introduce the set Sm consisting of permutations
go 6 S' such that
go(l) < 9o(3) < < go(2h l)»
Observe that
/I 8'M (2h)!
#9 =(#8"lh! \.
l 3---(2h- l ) : 7 (2h).
#so
1

' Hz 2hh!
Then
p r e( ) l = ( - l ) " e(lo)@</)(1)(p(2) /\ /\ €)qpl2f'L- 1)</)(2h)
•

4/£843
We have
h
1
p f (- G)) I
H (Cr,(p)ES' XS'
6((t(0) H
j=1
R<0(2j-1)<p(2j)<t(2j-1)0(2j) dVm.
404 Lectures on the Geometry of Manzfolds

On the other hand,

lA2hs)(e1 /\ 62 /\ ' ° ' / \ 6211-1 /\ €2h)
1
€(0)At"5(ep(1> /\ 8(p{2) /\ • '
/\ 8(pl2h-1) /\ €</)(2h)l
#8'm </)€5'
1
Q /\ G)(p(2h_1)90(2h)
#8'm </)€5' do) (/)(1)¢(2) /\

h
1

(0r,(p)€S' XS'
e(c7cp) H
j=1
R(p(2j-l)(P(2j)<t(2j-1)U(2jl dVm

Hence
h!
§;;,dAm (A\h5)(€1 /\ 82 /\ /\ 8 2 h _ 1 /\ B2hl pf G) 7
#8'm
so that
SRI O'2h,
pf ( Q I
K

The last form is the Euler form e(VM ) associated to the Levi-Civita connection VM.
Using the Gauss-Bonnet-Chern theorem we obtain the following result.

Theorem 9.2.3. If m 2 h C R2h+1 is a compact, oriented hypersurface, and g denotes the

induced metric, then
2 -» 1
0' rdAm = e ( V M ) (W p f (- Q 9),
and
-» M _
2 deg am M
et ) - x(m).
Moreover; ifl€17 , €2hl is a local, positively oriented orthonormal frame ofTer, then
h
l
pf
99) h!
(o,(p)€Sh X S '
€((t(el H
j=1
R<p(2i-1)>p(2i)<t(21-1)<t(2i) 'WM (9.2.7a)

do) ® p(1)<p(2) A /\ ®<p(2h-1)(p(2h}v (9.2.7b)

90652Ih

where gogh denotes the set of permutations go of { 1 , . . . , 2h} such that

(/)(2J - 1) < (0(2J), vi S i S h,

S" denotes the set of permutations go Q S' such that
(0(1) < (0(3) < ~~ < (/)(2h - 1),
and

k<£
 Classical Integral Geometry 405

Example 9.2.4. (a) If dim M = 2 then

pf ( g) = R1212dVm (the Gaussian curvature of M) X COM
(b) If dim M = 4, then SZ consists of 3 permutations

17 27 37 47 6 I
1 > @1212®34347
17372747 6 = -1 > -@1312 ® 2434
17472737 e = 1 > @1412 ® 2334
We deduce

pf ( go = 912 /\ 934 ®13 /\ 924 -|- 914 /\ ®23

1 l
i ZlR12z.7 071 /\ 0i)/\(R34ke 0 /\ 0
k
al k
4 lR13fzj Hi /\ 0 i)/\lR241¢e 0 /\ 0')
1 k
+ zlR14@y Hi A 01 ) / \ ( R 2 3 k £ 0 /\ 0 e ).

Each of the three exterior products above in involves only six nontrivial terms, because
9'Z A go = 0, when i = j. Thus the total number of terms in the above expression is 36.
However, there are many repetitions due to the symmetries of the Riemann tensor

Rilke = Rketzj = -Rigel

We have
k e
(Raj Hi /\ 0' ) /\ (R34I<e 0 /\ 0 )
»

+ R1213R3442 + R1214R3423
IR1212 R3434

+ R1223R3414 + R1242R3413 + R1234R34121 dVm, etc.

(c) We can choose the positively oriented local orthonormal frame (el, . . . , ml so that it
diagonalizes SM at a given point co 6 M. Then the eigenvalues of SM at at are called the
principal curvatures at at and are denoted by 14,1 (at), . . . ,/<zm(:z3). Then
P f (-(al = pdvm, p G C°°(M),
where
771

,0(x) = (2h - 1>n H Hi($), Vat 6 M.

k=1

9.2.3 Gauss-Bonnet theorem for domains in an Euclidean space

Suppose D is a relatively compact open subset of an Euclidean space Rm-I-1 with smooth
boundary 81). We denote by TL the outer normal vector field along the boundary. It defines
an oriented Gauss map
§0 : 3D -> Sm
We denote by dAm the area form on the unit sphere SM so that
1
deg §0
0'
saDdAm.
'rn an
406 Lectures on the Geometry of Manzfolds

If m is even then the Gauss-Bonnet theorem for the hypersurface 3D implies

1 1
§',§,dAm 2X( an1 i •
O' 'rn an
Using the Poincaré duality for the oriented manifold with boundary D we deduce X(3D )
2X(D), so that
1
O' 'rn an
We want to prove that the above equality holds even when m is odd. Therefore in the
remainder of this section we assume m is odd.
We first describe the integrand saDdAm. Let SD denote the second fundamental form
of the hypersurface
SD(X,Y) n . (D x v ) = - x . ( D Y al, 'v'X,Y€Vect(<9D).
We deduce
1 1
§;,dA,,,, detl-SD)dV6D»
O' 'rn O' 'rn

where dV3D denotes the volume form on 31).

A smooth vector field on D,
x 115->Rm-I-1
is called admissible if along the boundary it points towards the exterior of D,
X . n > 0 , on as.
For an admissible vector field X we define
1
x' : aD-> s m » X-(p) Ix(p)l X(p), V p E 31).

Let us observe that the map X- is homotopic to the map §0. Indeed, for t E [0, l], define
l
t + UZ.
l(1 t)n + txl ((l )n
Observe that this map is well defined since
l(1-t)n+¢XI 2 = z 2 + ( l - t ) 2 + 2 ¢ ( l - z ) ( n . X) > 0.
Hence
deggp = degX-,

for any admissible vector field X.

Suppose X is a nondegenerate admissible vector field. This means that X has a finite
number of stationary points,
Zx = {I>1, • • - 7PV}7 X(p¢) 0,
and all of them are nondegenerate, i.e., for any p G Z x , the linear map
AX,p : TpRm+1 -> TpRm+1, TpRm+1 9 U <
»-> D uXIIPI
 Classical Integral Geometry 407

is invertible. Define
6X : Zx -> {$1}» €(p) = sign det App.

For any 5 > 0 sufficiently small the closed balls of radius 5 centered at the points in Zx
are disjoint. Set

DO D U BE (p) -
PQZx

The vector field X does not vanish on D69 and we obtain a map
1
X:D€->SM 17 X X.
IX I
Set
l
Q X*dAm.
0' 'rn

Observe that
1
do = x*d(dAm) = 0 on D .
O' m

Stokes' theorem then implies that

/3D lD€
do 0 -:'» deg aD / an
Q=
2/
PGZx 3B@(1>)
Q,

where the spheres 335 (p) are oriented as boundaries of the balls BAP) If we let e -> 0
we deduce
deg aD €x(P)» (9.2.8)
PQZx

for any nondegenerate admissible vector field X.

To give an interpretation of the right-hand side of the above equality, consider the dou-
ble olD. This is the smooth manifold 15 obtained by gluing D along 3D to a copy of itself
equipped with the opposite orientation,

Uan (-D) ~
The manifold without boundary 15 is equipped with an orientation reversing involution
go : D -> D whose fixed point set is AD. In particular, along AD C D we have a cp-
invariant decomposition

Tl5l6D = T619 @ L,
where L is a real line bundle along which the differential of go acts as -II L . The normal
vector field n defines a basis of L. If X is a vector field on D which is equal to n along
31), then we obtain a vector field X on D by setting

X on D
(><
II

SO* ( X ) on -D.
408 Lectures on the Geometry of Manzfolds

If X is nondegenerate, then so is 2, where the nondegeneracy of 2? is defined in terms of

an arbitrary connection on TD. More precisely, if V is a connection on TD, and q Q Zx* ,
then q is nondegenerate if the map

A : Tq15 -> Tq15, 'u +->

is an isomorphism. This map is independent of the connection V, and we denote by @;(<1)

the sign of its determinant. Moreover
ZN =ZxUS 0lZX) )

and, because m i5 odd, the map

€X'~ : ZN -> { i t }
satisfies

€x(P) = €X-((0(pll-
Hence
6X lql(9.2.8)2 deg g, •

62(9) 2
q€Z.X- P€2x

On the other hand, the general Poincaré-Hopf theorem (see Corollary 7.3.48) implies that

eX*(q) = x(D).
QEZx

Using the Mayer-Vietoris theorem we deduce

xi?) = 2x(D) - x(<9D)-

Since AD is odd dimensional and oriented we deduce that X(<'9Dl = 0, and therefore

2x(Dl Z xU) =2 fx(q)=2deg§D.

PGZx

We have thus proved the following result.

Theorem 9.2.5 (Gauss-Bonnet for domains). Suppose D is a relatively compact open

subset of R""+1 with smooth boundary 61). We denote by 9 D the oriented Gauss map

§0 : 3D -> S " , (BD 9 p »-> 'n(pl = unit outer normal,

and by SD the secondfundamentalform of 31),
SD ( X , Y) TL . (D X Y), VX, Y 6 Vect ( u p ) ,
Then
1
d€tl-sDldv3D : : deg §0 x(Dl-
0' TTY an
 Classical Integral Geometry 409

9.3 Curvature measures

In this section we introduce the main characters of classical integral geometry, the so called
curvature measures, and we describe several probabilistic interpretations of them known
under the common name of Crofton formulae.
We will introduce these quantities trough the back door, via the beautiful tube formula
of Weyl, which describes the volume of a tube of small radius r around a compact subman-
ifold of an Euclidean space as a polynomial in r whose coefficients are these geometric
measures. Surprisingly, these coefficients depend only on the Riemann curvature of the
induced metric on the submanifold.
The Crofton Formulae state that these curvature measures can be computed by slicing
the submanifold with affine subspaces of various codimensions, and averaging the Euler
characteristics of such slices. To keep the geometric ideas as transparent as possible, and
the analytical machinery to a minimum, we chose to describe these facts in the slightly
more restrictive context of tame geometry.

9.3.1 Tame geometry

The category of tame spaces and tame maps is sufficiently large to include all the compact
triangulable spaces, yet sufficiently restrictive to rule out pathological situations such as
Cantor sets, Hawaiian rings, or nasty functions such as sin(1/t).
We believe that the subject of tame geometry is one mathematical gem which should be
familiar to a larger geometric audience, but since this is not yet the case, we devote this sec-
tion to a brief introduction to this topic. Unavoidably, we will have to omit many interesting
details and contributions, but we refer to [31 , 36, 37] for more systematic presentations. For
every set X we will denote by IP(X) the collection of all subsets of X .
An R-structure2 is a collection S = {8"}n219 5" C IP(]R"), with the following proper-
ties.

Ei- Sn contains all the real algebraic subsets of R", i.e., the subsets described by finitely
many polynomial equations.
EQ. Sn contains all the closed affine half-spaces of R"
P1. Sn is closed under boolean operations, U, (W and complement.
PQ. If A E S o , and B Q S", then A X B E SM-V".
P3. If A E So, and T : RM -> R" is an affine map, then T(A) E S"

Example 9.3.1 (Semialgebraic sets). Denote by 521g the collection of semialgebraic sub-
sets of R", i.e., the subsets S I R " which are finite unions
S = S1 U USu,
where each 5 is described by finitely many polynomial inequalities. The Tarski-
Seidenberg theorem (see [20]) states that S a l g is a structure.
2This is a highly condensed and special version of the traditional definition of structure. The model theoretic
definition allows for ordered fields, other than R such as extensions of R by "infinitesimals". This can come in
handy even if one is interested only in the field R
410 Lectures on the Geometry of Manzfolds

To appreciate the strength of this theorem we want to discuss one of its many conse-
quences. Consider the real algebraic set
Zn Q R " +1.7 a0-i-a1a3+--a,,_1a:" - 1 = 0 }.
• •

l£I',(I,0,...,(I,n_1l -I-.Clin

The Tarski-Seidenberg theorem implies that the projection of Zn on the subspace with
coordinates ( G i l is a semialgebraic set
in 6 E Rn; Ha: E R, PA l$l3l 0 }7
• •

007 7 an-ll

where
P5(x) = Ag -|- @158 + - &n_1$n
-1
-|- ax"

In other words, the polynomial PA has a real root if and only if the coefficients co' satisfy at
least one system of polynomial inequalities from a /nite, universal, collection of systems
of polynomial inequalities.
For example, when n = 2, the resolvent set 322 is described by the well known inequal-
ity as
- 4a0 2 0.
When n = 3, we can obtain a similar conclusion using Cardano's formulae. For n > 5,
we know that there cannot exist any algebraic formulae describing the roots of a degree n
polynomial, yet we can find algebraic inequalities which can decide if such a polynomial
has at least one real root.

Suppose 8 is a structure. We say that a set is S-definable if it belongs to one of the S"'s.
If A, B are S-definable then a function f : A -> B is called S-de_lQnable if its graph
Ff := {(a,b) E A X B; b = f(a) }
is 8-deiinable. The reason these sets are called definable has to do with mathematical logic.
A formula3 is a property defining a certain set. For example, the two different looking
formulas
2
{;z3€R; a 2 2 0 } , {a;'€]R; 2 3 / E R : LU y }
describe the same set, [0, ool.
Given a collection of formulas, we can obtain new formulas, using the logical opera-
tions A, v, -, and quantifiers EI, V. If we start with a collection of formulas, each describing
an S-definable set, then any formula obtained from them by applying the above logical
transformations will describe a definable set.
To see this, observe that the operators As V, correspond to the boolean operations,
-l

O, U, and taking the complement. The existential quantifier corresponds to taking a projec-
tion. For example, suppose we are given a formula (l5(a, b), (a, b) 6 A X B, A, B definable,
describing a definable set C C A X B. Then the formula
{ s e A ; Elb E B : ¢(a,b)}
describes the image of the subset C C A X B via the canonical projection A X B -> A. If
A C RT", B C R", then the projection A X B -> A is the restriction to A X B of the linear
oWe are deliberately vague on the meaning of formula.
 Classical Integral Geometry 411

projection RM X R" -> RM and P3 implies that the image of C is also definable. Observe
that the universal quantifier can be replaced with the operator -E-\.

Example 9.3.2. (a) The composition of two definable functions A i> B i> C is a definable
function because
Fg0f= 1(a,c)€ A X C;§lb€ B : (a,b) € Ilf, ( b , c ) € I ` g } .

Note that any polynomial with real coefficients is a definable function.

(b) The image and the preimage of a definable set via a definable function is a definable
set.
(c) Observe that 'Salg is contained in any structure S. In particular, the Euclidean norm

I . II]Rn->]R, l(3317 7 son) I

'é 1
2
3%
1/2

is 8-definable.
Observe that any Grassmannian Gr k(R"l is a semialgebraic subset of the Euclidean
space End+ (R) of symmetric operators R" -> R" because it is defined by the system of
algebraic (in)equalities

Gr 1<(R") = {p 6 End+(R"); P2 = p, A'°p ¢ 0, AmP = 0, am > If },

where As P denotes the endomorphism of AND" induced by the linear map P 6 End(R").
In the above description, to the orthogonal projection P of rank k, we associate its range,
which is a vector subspace of dimension k.
Similarly, the affine Grassmannian GraHk(R") is a semialgebraic set because its is
described by
Graff 1<(R") = {(m,p) E R" >< Gr,<,(R")§ Pa: = 0 }.
In the above description, to the pair (ac, P) we associate the affine plane ac -|- Range (Pa .
(d) Any compact, affine simplicial complex K C R" is S-definable because it is a finite
union of affine images of finite intersections of half-spaces.
(e) Suppose A C R" is S-definable. Then its closure el(A) is described by the formula

{ x € R " ; 'v'€>0, § l a € A : at al<€ },

and we deduce that el(Al is also .S-definable. Let us examine the correspondence between
the operations on formulas and operations on sets on this example.
We rewrite this formula as
Ve ( e > 0 ) = > 3 a ( a e A ) A ( : v e R " ) A ( l a : - a l < € )

In the above formula we see one free variable co, and the set described by this formula
consists of those ac for which that formula is a true statement.
The above formula is made of the "atomic" formulae,
(a 6 A), (33 Q RTL), (lac - al < 5), (5 > Of,
412 Lectures on the Geometry of Manzfolds

which all describe definable sets. The logical connector => can be replaced by V-. Finally,
we can replace the universal quantifier to rewrite the formula as a transform of atomic
formulas via the basic logical operations.

E L f ( (6 > 0 ) = > § l a ( a € A ) A ( a 3 € ] R " ) / \ ( l x - a l < € )

Arguing in a similar fashion we deduce that the interior of an S-definable set is an S-

definable set.

Given an R-structure S, and a collection A - (A.,,)v,21, An C U'(]R"), we can form

a new structure 8(A), which is the smallest structure containing 8 and the sets in An. We
say that 8(A) is obtained from 8 by adjoining the collection A.

Definition 9.3.3. An R-structure 8 is called tame, or o-minimal (order minimal) if it satis-

fies the property

T. Any set A E S1 is a _finite union of open intervals (a,b), < a < b < oo, and
singletons {r}.

Example 9.3.4. (a) The collection of real semialgebraic sets S a l g is a tame structure.
(b) (Gabrielov-Hironaka-Hardt) A restricted real analytic function is a function f : R" ->
R with the property that there exists a real analytic function JF defined in an open neighbor-
hood U of the cube Cn := [-1, 1]"' such that

.f(w) :u E co,
f(:v)
0 x€R"\C'n
we denote by S a n the structure obtained from Sals by adjoining the graphs of all the re-
stricted real analytic functions. For example, all the compact, real analytic submanifolds of
R" belong to S a n . The structure 'San is tame.
(c) (Walkie, van den Dries, Macinlyre, Marker) The structure obtained by adjoining to San
the graph of the exponential function R -> R t »-> e*, is a tame structure.
(d) (Khovanski-Speissegger) There exists a tame structure S' with the following properties

(dl) San C 8/.

(do) If U C R " is open, connected and S'-definable, Fl, . . . , Fn . U X R -> R are S'-
definable and C1, and f : U -> R is a C1 function satisfying
8
a f = H n » VQCQR, 7 QQ l,...,n, (9.3.1)

then f is S'-definable.

The smallest structure satisfying the above two properties, is called the pfajian closure4 of
S a n , and we will denote it by San .

4 Our definition of pfaMan closure is more restrictive than the original one in [78, 122], but it suffices for many
geometric applications.
 Classical Integral Geometry 413

Observe that if f (a, b) -> R i s 01, 3 -definable, and kg 6 (a,b) then the antideriva-
five F : ( a , b ) - > ] R

F(23) f (t)dt , 1 (a, b),

000

is also -definable.

The definable sets and functions of a tame structure have rather remarkable tame be-
havior which prohibits many pathologies. It is perhaps instructive to give an example of
function which is not definable in any tame structure. For example, the function ac +-> sin as
is not definable in a tame structure because the intersection of its graph with the horizontal
axis is the countable set we which violates the 0-minimality condition T.
We will list below some of the nice properties of the sets and function definable in a
tame structure S. Their proofs can be found in [3l,36].
. (Piecewise smoothness of tame funetions and sets.) Suppose A is an S-definable set, p is
a positive integer, and f : A -> R is a definable function. Then A can be partitioned into
finitely many 8 definable sets 51, . . . , So, such that each S is a Co-manifold, and each of
the restrictions f | s. is a Co-function. The dimension of A is then defined as max dim S .
..(Dimension of the boundary.) If A is an S-definable set, then dim(el(A) \ A) < dim A.
(Closed graph theorem.) Suppose X is a tame set and f : X -> R" is a tame bounded
function. Then f is continuous if and only if its graph is closed in X X R" .
.(Curve selection.) If A is an 8-definable set, h > 0 an integer, and :Hz 6 el(A) \ A, then

.
there exists an 8-definable Ck-map *y : (0, 1) -> A such that :u = limt_,0 'y(t).
(Triangulability) For every compact definable set A, and any finite collection of defin-
able subsets {517 . . . , S;.,}, there exists a compact simplicial complex K, and a definable
homeomorphism
<I> K -> A

..
such that all the sets <I>-1 (Si) are unions of relative interiors of faces of K.
Any definable set has finitely many connected components, and each of them is definable.
(Definable selection.) Suppose A, A are S-definable. Then a definable family of subsets
of A parameterized by A is a subset
SEA X A.
We set
{ a A ; (a,A>€s},
and we denote by As the projection of S on A. Then there exists a definable function
S : AS -> S such that

s(A) Q S , VA Q As.
. (Dejinability of dimension.) If (SA) AGA is a definable family of definable sets, then the
function
A 9 A »-> dim SA Q R
is definable. In particular, its range must be a _finite subset of Z.
414 Lectures on the Geometry of Manzfolds

. (De/nabilily of Euler characteristic.) Suppose (SA) AeA is a definable family of compact

tame sets. Then the map
A 9 A +-> X(SAl the Euler characteristic of SA 6 Z

.
is definable. In particular, the set { x(5A); A Q A, } C Z is finite.
(Scissor equivalence.) If A and B are two compact definable sets, then there exists a
definable bijection go : A -> B if and only if A and B have the same dimensions and the

.
same Euler characteristics. (The map go need not be continuous.)
(Dejinable triviality of tame maps.) We say that a tame map Q) : X -> S is dejinably
trivial if there exists a definable set F, and a definable homeomorphism T : X -> F X S
such that the diagram below is commutative
T
X > S X F

<I> its

S
If \I/ : X -> Y is a definable map, and p is a positive integer, then there exists a partition of
Y into definable Co-manifolds Y1, . . . , Ye such that each the restrictions
if : \P-1(Y1<l -> ye
is definably trivial.

Definition 9.3.5. A subset A of some Euclidean space Rn is called tame if it is definable

within a tame structure S.

Exercise 9.3.6. Suppose S is a tame structure and up : [0, l] -> R2 is an S-definable map.
(a) Prove that up is differentiable on the complement of a finite subset of [0, l] .
(b) Prove that the set
{ (t, m) G [0, l] X R 2, . go is differentiable at t and m (p'(tl }
is 8-definable.
(c) Prove that the curve

C= {(t(0(ill; t€[0,1]}
has finitely many components, and each of them has finite length.

9.3.2 Invariants of the orthogonal group

In the proof of the tube formula we will need to use H. Weyl's characterization of polyno-
mials invariant under the orthogonal group.
Suppose V is a finite dimensional Euclidean space with metric (-, -) We denote by
( - 7 - ) the canonical pairing
(-,-)¢v* X V-HR, ( 7 u) (U), V u Q V , ATV* = Hom(V, R).
We denote by O(V) the group of orthogonal transformations of the Euclidean space V.
 Classical Integral Geometry 415

Recall that an O(V)-module is a pair (E, 0% where E is a finite dimensional real vector
space, while p is a group morphism
p : O(V) -> Aut(E), g +-> p(g).

A morphism of O(V)-modules (E¢»P¢)» ¢ = 0, 1, is a linear map A E0->E1 such that

for every g Q O(V) the diagram below is commutative

E0 A>E1

90(9)
l
ET >
l
E1
p1(9)

We will denote by Hom0(v) (EO, El) the spaces of morphisms of O(V)-modules.

The vector space V has a tautological structure of O(V)-module given by
T : O(V) -> Aut(V), T(g)v = go, Vg 6 O(V), 'u 6 V
It induces a structure of O(V)-module on V* = Hom(V, R) given by

Pt : UW) -> Aut(V*l, 9 »-> pt(9l»

where

(Mg) A u) (A ,9-1v),v A Q V*,v 6 M

Equivalently, 0+(9) p(g-1)t, where p(h)t denotes the transpose of p(h). We obtain an
action on (V*)®n,

( p ) ® " : olv) -> Auto (v*)®" ), 9 »-> /)1(9)®"-

We denote by (V*)§FV> the subspace consisting of invariant tensors,
®n
w6 (v*)§W> < lp+(9ll CU cu, Vg E O(V).

Observe that (V*)®" can be identified with the vector space of multi-linear maps
TL
w:V V X X V ->R,
n

so that (V*)§FV) can be identified with the subspace of O(V)-invariant multilinear maps
V" -> R
Hermann Weyl has produced in his classic monograph [133] an explicit description
of (V*)V> . We would like to present here, without proof, this beautiful result of Weyl
since it will play an important role in the future. We follow the elegant and more modern
presentation in Appendix I of [8] to which we refer for proofs.
Observe first that the metric duality defines a natural isomorphism of vector spaces
D:V->V* 7 v+->vT 7
defined by
(UT, u ( p a ) , iv"u,,v 6 V
416 Lectures on the Geometry of Manzfolds

This isomorphism induces an isomer rphism of O(V)-modules

D : (VZ al -> (V*»p1l~
We conclude that for ever nonnegative integers r, S we have isomorphisms of G-modules
IV-*l®(v"-I-s) (V*®1") V®8 %" Hom(V®", v ® s > .
(X)

In particular,
( (v*)®""+8" )0(v> 4 lHom(V ®'r 7 v ® s > low) HO1T1o(v)(V®» v ® s > .
I

Let us observe that if we denote by S the group of permutations of {l, . . . ,t°}» then for
every up Q S we obtain a morphism of O(V)-modules
®'r
T¢> 6 HOM0(v)(V 7 v®*>, Tpl'U1 (X) (X) Ur U</)(l) (X) (X)
'U(p(r)

Weyl's First Main Theorem of Invariant Theory states that

H0Mo(v)(V®", v ® s ) ye 0 1 \
I r 5,

and that

®1°
,v®'°) = ]R[S,.] Co E R |
..II

H0mo(v)(V C90i 4p§

I
iP€g7"

We can translate this result in terms of invariant multi-linear forms. Thus

(V*)§?V) ¢ 0 = 21", T' E Z>_0,

and lV*l \'l is spanned by the multilinear forms

: v" -> R, P(/) (of) E 8,.>,
defined by
Pgo tut . . . 7 U r 7
7 V17 7 v) (u1»v<p(1))'~('"»~»v(p(~>)~

The above result has an immediate consequence. Suppose we have a map

f :V X X V ->]R, (v1,...,vn) +-> f(v1,...,v,,),
n

which is a homogeneous polynomial of degree , n . This

form determines a multilinear form
pol, : vol x---xVd'*-HR

obtained by polarization in each variables separately,

1 d
Po1f(/u,1,...,u11, .,am1 dn
gun )
the coefficient of the monomial 511512 • ' °t1d1 ' ° ' t n 1 ' . 'andn in the polynomial
1
LL1 Un

pf(t11,t12,...,t1d1,...,tn1,...,tndn) = f j
iii up, 7 t u. nj
j=1 j=1 /
 Classical Integral Geometry 417

Observe that
f('V1» 7 Vn POlf ('U1, . . . #11, 7'0717 7'071 7

d1 dn

and f is O(V)-invariant if and only if Pole is O(V) invariant.

Note that every function
X X V->R
TL

which is polynomial in each of the variables is a linear combination of functions which is

polynomial and homogeneous in each of the variables. For every l < i < j < n we define
viz I V i X V->R, €Iij('U1>...)'Un) I-' ( ' U p » V j ) .

Theorem 9.3.7 (Weyl ). [ f f : V X X V -> R is a polynomial map then f i5 O(V)-

»

invariant if and only if there exists a polynomial P in the variables iii such that

f('V1» ' • ' in) Z €I@j('U1v . • . 1'Unl1<z<j<n

Example 9.3.8. (a) Consider the space E = v®'°. Observe that a degree n homogeneous
polynomial P on E can by identified with an element in the symmetric tensor product
Symd(E*) C lV* )®2 1¢n .

The polynomial P is called a degree n orthogonal invariant of tensors T G V®'" if it is

invariant as an element of (V*>®'<" For example, Weyl's theorem implies that the only
degree l invariant of a tensor
T ZT zJ Bz (X) 61 Q V ® 2

'M
is the trace
tr(T)
'M ii

The space of degree 2 invariants is spanned by the polynomials

ltt(Tll2, C2(T)
"in "id
Here is briefly how to find a basis of the space of degree k invariant polynomials
P : V ®h -HR.
First of all, such polynomials exist if and only if he is even, he Zm. Fix an orthonormal
-
basis {et, -~ ,ed} of V. A tensor T Q V®h decomposes as
T T 1 -..ih €i1 (X) (X) €
1§'é1,...,ih§d

A degree k polynomial P : V®h -> R is then a polynomial in the dh variables Tfz1...¢h .

418 Lectures on the Geometry of Manzfolds

We will construct a bijection between a basis of invariant polynomials and certain com-
binatorial structures called matchings.
A matching on the set I.k := {l, . . . , he} is an equivalence relation with the prop-
erty that each equivalence class consists of two elements. Thus to produce a matching
we need to choose a subset A C 1-hk of cardinality m = M2 and then a bijection 9

(p : A -> Ihk \ A. Observing that the matching associated to (A, up) is the same as the
matching associated to (Ihk \ A, (p-1), we deduce that the number of matchings is

Denote by 91, the set of functions

1
-Impel
2 II m
(2m)!
2(m!l

llifhk -> {l,27...7 d}

such that i j => /u,(i) 1u,(j). Equivalently, 3",,, can be identified with the set of
functions ,L/, : I h k / ~-> {1, , d}, so there are am such functions.
For any matching define
P- (Tl I T,,(1)...u(h) ' ' • T;1(hk-h+1)...p(hk) -
116Cv"-

The collection

{P~§ iv is a matching of Ihk }

is a basis of the space of degree k invariant polynomials v®/* -> R
For example, the space of degree l invariant polynomials in tensors of order 4 has
dimension 2 2 1 ) = 6. Each polynomial in a basis constructed as above is a sum of d2
terms. We see that things are getting "hairy" pretty fast.

9.3.3 The tube formula and curvature measures

Suppose that M is an m-dimensional submanifold of R". We set

C := codi M n m.

In this section we will assume that M is compact and without boundary, but we will not
assume that it i5 orientable.
For T' > 0 we define the tube of radius T' around M to be the closed set
1ll,.(M) := {:u Q M; dist (;u,Ml < ,~},
and we denote by V(M, al its volume.
Note that a first approximation for V ( M , T) is
V ( M , 7 ) z vol ( M ) - wCT' C 7

where wCre is the volume of a ball of dimension C and radius

We want to prove that there exists a polynomial
Pm(7ll = PJC + - - +1w°" 7

such that, for all sufficiently small r, we have

 Classical Integral Geometry 419

V ( M , r ) = PM(7°l,
-
pC = we vol ( M ) ,
all the coefficients of Pm are described, up to some universal multiplicative constants,
by certain integral, intrinsic geometric invariants of M .

Let N(M) denote the orthogonal complement of T M in (TWaIn, and we will call it
the normal bundle of M --> Rn. We define

n;),,(R") := {('u,p); p Q R" , 'U Q TpR", lvl < r C TR" 7

and we set

UMM) :z NIMI n 1D>,.(R").

The manifold with boundary D ( R " ) is a bundle of n-dimensional disks over R", while
MIMI is a bundle of e-dimensional disks over M.
The exponential map IE:TR"->R" restricts to an exponential map

Observe that because M is compact, there exists To = 1°0(M) > 0 such that, for every
T' G (0, to), the exponential map IEm induces a diffeomorphism

EM : n,.(m) -> Mm).

If we denote by ldvn I the Euclidean volume density on R" we deduce

v(m,¢~) vol ('l',.(Ml ) Adm ]P3MdVn|-

1r,(m> N,~(M)

If 77 N,.(M) -> M denotes the canonical projection, then we deduce from Fubini's
theorem that

V(M,rl 1r*IEMldV,,l. (9.3.2)

M

We want to give a more explicit description of the density Tr* IBM ldV,,l. We will continue
to use the indexing conventions we have used in Subsection 9.2. 1.
Fix a local orthonormal frame leA) of (T]R")l M defined in a neighborhood U C M of
a point Po 6 M such that for all l < i g m vector field ii is tangent to U. We define

D;
t D
tal I ltm-f-1
,t") € Re; we 2
<r
. 0

Note that we have a diffeomorphism

Dre X U >N,°(U) I N~(M)lu, 22) * (tO'ea(a:),a:) E NrlM)a

and thus we can identify D; X U with the open subset 71' -1(U) C N,°(M), and we can use
as Q M and Fe D; as local coordinates on 7r"1(U). Define
1r,,(U) ¢= IEm(N»~(U)) C R" 7
420 Lectures on the Geometry of Manzfolds

and
pA IU) -> R" by éA(;13-I-taeO) = et(ac).
We have thus extended in a special way the local frame (6A) of (TO-")lu to a local frame
of lTR")l1r,.(u) so that
Do pA = 0, 'v'a,A. (9.3.3)

We denote by (0 A) the coframe of IU) dual to pA.

Over D; X U we have a local frame later, , e ) with dual coframe (¢>A) defined by
bi To* 0 dtO! . I
v (Ba Z

Consider the 1-forms Q1(1r,(u)) associated to the Levi-Civita connection D by

(9 A
B 6
the frame (pA) on or(II), and set
A A
@C B ac J Q BE W,

so that
A
D ec 6 B = ®CBeA.

Using (93.3) we deduce

A A A
9 aB 0, 'v'a=> 9 B 9 iB 0 (9.3.4)

Finally set

'Pg = 7
AIM
r*(®B
)
Q
QUO ( n "(U ),
)
AB

7 r* ( ®iABl
ul E
COo (n "(U) ).
The equalities (9.3.4) imply
A A i
'DB Cb'iB¢
On N,-(M)lu we use <a 22) as coordinates and we have

IEM(t°'eo(:u),:z3) = x + t " e o .

We have
EmHA leA A D ii + leA ata .
11

Z
i
8A . (et +t°" D €1 ealcb' +
a
5Adt°" = (SA + t"®;¢¢ + 6AO,di'*

Hence
IBM 0 j = (N + t""<I>ia<1 = (of i°"<1>¢", EM 0 B =d¢@.
a

We find it convenient to set

®ij = ( 2 3 - 1 7 ,®¢i):U-HRC 7
so that

EM 0 ;(/)l (t . q>@j)(z5"
 Classical Integral Geometry 421

Define the m X m symmetric matrix

S : Sl1;$)=l . @¢jl£E)l 1 < i , j < m

Note that the volume density on IRE." is

ldVnl=l9m-I-1/\~~-AH"/\ H1 I\.../\9*"l.

1E;;,dv,, = ldet(]1 - s(€;w) )| ld£'A d¢l = det(]1 s(£ x)) ld£'A d¢l, (9.35)

d£'= dtm+l / \ - - - Adi", ¢z¢)=d¢1 A---Adq>m.

Recalling that I A d 0i ll is the metric volume density ldVml on M, we deduce
1E;,ldv,,l = det(]l - s ) ) ld£] X 7T*ldVMI,

where Idol
denotes the volume density on Re_ For simplicity we write ldVm I instead of
7I-*ldVMI. Now set
1 -»
_5»
p IH, w
p
and denote by Idol the area density on the unit sphere in Re. Then

IEMldVnl =det(]1-pS(w,;z3llpe-1ldpl X Idol X ldVMI. (9.3.6)

Observe that

det(]l - pS(w,a2)l
m

(-1)"p"P~( <I>¢i($) .we,

where PV denotes a homogeneous polynomial of degree V in the m2 variables

'U4j€Re» l S'i,j<m-

We set

F,,(<1>if (23) ) I

(we) .
•

wlldwl.
.

V
So 1

I
Above, F, ( u in is an 0(0)-invariant, homogeneous polynomial of degree V in the vari-
ables UU 6 Re, l < i , j < m. We conclude,
m

7'l'*IE8°'/IldVnl (-II"
c + U 7"°+"P,,l <1> (w) ) udvm(w)l.
(9.3.7)

We would like to determine the invariant polynomials P, ( ii l,

Theorem 9.3.7 on invariants of the orthogonal group O(e) implies that FV must be a
polynomial in the quantities

i i , j , k , e 1= Nile . Ujg.

Because these quantities are homogeneous of degree 2 in the variables UU we deduce

Pa 0 if V is odd. Assume therefore V = 2/2, h Q Z20.
422 Lectures on the Geometry of Manzfolds

For every FORe span ($m+17 7 am-i-el we form the linear operator

U(£l Ulu ij 7 al : Rm -> Rm

given by the m x m matrix (Un . )1S©,JS'"1' We deduce that

(-1)"p(- .F) tr AVUIU 7 al

the sum of all the V X V minors of up?) symmetric with respect to the diagonal.

.
These minors are parametrized by the subsets I C {l, . . . , m } of cardinality #I = V. For
every w Q Re we denote by of( uij w ) the corresponding minor of U(w), and by EI its
average,

F] (Um ) #Mum . w)dw.

0
.

Se-1

Note that HI is an O(c)-invariant polynomial in the variables {'~*@j } i , j € I ~

Let

I {1<'i1<'12<°°'<'i2h<M}C{17lCu7'ITL}7

and denote by 81 the group of permutations of I. For go 6 81 we set

..,2h.

For any C7,9o 6 81 we denote by go 1

€(0, (p) the signature of the permutation O' O 9 and by
QW the invariant polynomial

QI,o,¢ = H
h

j=1
q(P2j-1,(P2j,O2j-1,o2j H
j=1
h

IU'<P25-1u21-1
. Usvzj v 2 j

Lemma 9.3.9. There exists a con5tant§ = ,gm,u,e depending only on m, V and C such that

H-] = $621, QI I= €(U»%0)QI,<1,(f)°

(,o,o€ SI

Proof. We regard UI as a function on the vector space of (2h) X (2h) matrices U with
entries in Re

U ; Iu i j ] i , j € I -
We observe that UI satisfies the following determinant like properties.

... DIDI
DI changes sign if we switch two rows (or columns).
is separately linear in each of the variables U .
is a homogeneous polynomial of degree h in the variables q i , j , k , £ -
 Classical Integral Geometry 423

We deduce that UI is a linear combination of monomials of the form

q/61,/<:2,€1,€2 `' ' qIc211-1k2h,€2h-1,22h 7

where
{It17 . . . k2h} and { 1 , . . . ,€2h}
7

are permutations of I. The skew-symmetry of EI with respect to the permutations of rows

and columns now implies that UI must be a multiple of Q I .

The constant 5 satisfies

I Fflu ij ) RC
f m,z/,e Q, 7 \V/U 6

so it suffices to compute the numerator and denominator of the above fraction for some
special values of U . We can assume I = {1, 2, . . . , 2h} and we choose
1
0
u in GRC

0
Then, if we set
al
to
1
1*

E Re7 w t
II

I to
to

we deduce
t1 0 0

U('u,in?
0% 0
7 Mlluij . V

00 t1
Hence

#JIN "UI Sc-1

|°v1|2"|dw|- (9.3.8)

On the other hand, we have

Q1,0,(0
h

H
j=1
'U'l,02j-10'2j-1
. U9O2j 02j 7

which is nonzero if and only if O' up. We conclude that for this particular choice of uij
we have
QI (2h)!.
424 Lectures on the Geometry of Manzfolds

Hence
1 1 2h
5m,u,e So 1

At this point we invoke the following result whose proof is deferred to the end of this
subsection.

Lemma 9.3.10. For any even, nonnegative integers 2/11, . , 2/2C we have
2F(2h12-I-ll . . . Il2hC2-l-l)

'Sc-1
we 2hC
ldwl FIc-l2hl 7

whereh h1-I--~-I-h6.

We deduce
5 _ 2r(2"2:H)r(1/2lc-1
9 (9.3.9)
m,2h,,c - (2h)!1"(°+;")

and

F (UU) €'rn,2h,e (9.3.10)

#I=u

We denote by 82 the group of permutations of a linearly ordered set with two elements. We
observe that every element

7' T1 7 ,Thl€G 82 X X 82
h

defines a permutation of I by regarding T1 as a permutation of {it, i2}, T2 as a permutation

of {i3, i4} etc. Thus G is naturally a subgroup of 51- The space Sk,/G of left cosets of
this group can be identified with the subset SI C 5] consisting of bijections go : I -> I
satisfying the conditions

901 < 902, S03 < 904, - ' Q 90212-1 < 9021-

We deduce that if 'U 51i (¢v)» then for every u ' S ] we have

€ ( 0 , 0lQI,0,¢ = €((t, (PTlQI,o,¢T

S0681 (p6S'I,T6G
h

(poS'I
€(0, (Q) H
j=1
lq(P2j-1»(P2j,U2j-1,O2j - q(P2j»(P2j-1,02j-1,02.1

9068;
€(0, (Q) H
j=1
R(P2j-1(P2J O ' 2 j - 1 O'2j
 Classical Integral Geometry 425

Using the skew-symmetry Rilke -Rijék we deduce

h

O',(P€5I
€(0, S0)QI,v,(p = 22
uGSI 9068;
€(0, (p) H
j=1
R(P2j-1(P2j02j-1025

uGSJGG
€(0T, (p) H
j=1
R(P2j-1(P2j02j-102j

h
- 2h
o,(p€S
€(0 #P)
HR
j=1
(P2_1-1(P2_102j-102j

ZQIIRI
We conclude that
F,,(¢¢j) = 2h€m,2h,c Q1 (9.3.ll)
#I=2h
Using (93.2) and (93.7) we deduce that
Lm/2J h
vol w 2hT'€+22 2 €m,2h,e
R dV .
VlM,T) llr,,lmll c+
(¢+2h>we+2h M
Q
h( >l ml
h=o
Let us observe that the constant
2h§m,2h,c
(e -|- 2h)we+2h
is independent of the codimension C. It depends only on h. Indeed, we have
2h€m,2h,e _ 2h§m,2h,e _ oh 2F(2h2+1)1l(1/2)°'1
(c -|- 2I1)W0+2h O'0+2h-1 U¢+2h-1 (2h)!I`( ¢+22h )
_ 2"1"(h + 1/2) _ '¥(2h) _ 1
' p(1/2)1+2h(2h)! - 7rIV"(2h)! - (27r)"h! '
We have thus obtained the following celebrated result of Hermann Weyl [132].

Theorem 9.3.11 ('l\1be formula). Suppose M is a closed, compact sub ran fold of R",
dim M = m, C = n - m. Denote by R the Riemann curvature of the induced metric on
M . Tnenfor all r > 0 sufficiently small we have
L1"/2J
V(M, al = vol ('r,~(M)) = We+2hTc+2hum_2hlMl 7

h=0
l
I
/J#f m - 2h( M = (2'/r)hh! M Q h( R )|dVml ,
where Qh is a polynomial of degree h in the curvature. By choosing a local, orthonormal
frame ( e l , . . . , am) o f T M we can express the polynomial QMR) as
Qh(Rl :z QIlR)v
#I=2h
where for every I = {it < 1 2 < < '12h} C {l, . . . , m } we define
h

QI(Rl
0,(,o€S
€(0»(2) H
j=1
R(f)(i2y-1)(f)('i2y)0('i2J-1)0(@2y)°
426 Lectures on the Geometry of Manzfolds

Example 9.3.12. (a) If h 0 then

leTTI(Ml = vol(M).
(b) Assume now that m is even, m 2/1, and oriented. Then c + 2h m+c 77, and
1 1
LLoIMI -Qh(R)dVm .
(21T)" M h!
Comparing the definition of Qh (R) with (9.2.7a) we deduce that the top dimensional form
1 1
-oh R (We 6 Q" M

is precisely the Euler form associated with the orientation of M and the induced metric, SO

that

u0(M,9) €(M,9)- (9.3.l2)

M

(c) Suppose now that M is a hypersurface in ]RM+1. Consider the second fundamental form
S ;
(5¢1)1<_¢,jsm» S11 = (D e i $i) . 6m+1,

where we recall that 6m-l-1 is in fact the oriented unit normal vector field along M. Fix a
point £1:0 6 M and assume that at this point the frame (el, . . . , e t ) diagonalizes the second
fundamental form so that
S I~©fz 6ij .

The eigenvalues 17 . . . 7 H/Tn are the principal curvatures at the point 5130. We denote by
CV(/Q) the elementary symmetric polynomial of degree U in the variables Hi. In this case

c 1, and we have
m
IEmdVRm+1 = det(]l - is)dz A dVm (-1)"t"c,,(rz)dt A dVm

m T' Q L1"/2J V22_l.1

or* EM dV]Rm+1
2(/
V 0 'f`
£Vdi
) CV/lfQld'/A4
1.:0 2h-|- 1
Cghllildvjl/1

so that
Lm/2J L1"/2J 'V'2h_i_1

W1+2h1l1+2hMm-2hlM) = V(M,r) 2 h=o 2h+ cg(/<;)dVM.

h 0 M

We conclude that
2
Mm-2h(M) = C2h(f©ldVm» O'2h I (Zh *|- 1lW1+2h.
O'2h M

If M Sm -> ]Rm-l-1 is the unit sphere then Iii = l and we deduce that
Um
#fn-2hlSm)
2
0th ( ) 2h
(9.3.13)
 Classical Integral Geometry 427

(d) Using the definition of the scalar curvature we deduce that for any m-dimensional
submanifold m771 <-> R" we have
#in-zIM7 g) = const sgldVgl,
M

where S denotes the scalar curvature of the induced metric g, and const is an universal
constant, depending only on m. We see that the map g -> Iim_2(M, g) is precisely the
Hilbert-Einsteinfunctional we discussed in Subsection 4.2.6, Definition 4.2.33.
To find the const we compute lim-QIM) when M = So. Using (93.13) we deduce
20' m m
O' 2 () 2 M
1,
where Sfround denotes the scalar curvature of the round metric on the unit sphere.
Observe that the Grassmannian of oriented codimension one subspaces of lR'"+1 can
be identified with the unit sphere So. Hence, the oriented Gauss map of the unit sphere Sm
is the identity. In particular, the shape operator of Sm <-> Rm-l-1 is the identity operator.
From Theorem Egregium we deduce that all the sectional curvatures of S" are equal to
one. Using the equalities (4.2.3) we deduce

Sround 2
"id
Rijij
M
1 2() m
2

Hence

-
2o m m m
U2 C)
2
Zo m
() 2 COI1Stm

1 1 1
const = o2
llll...l.lllll

411'
> Um-2lM)9) I 471' S9|dV9|-
M
(e) The polynomial QUO still has a "reasonable form"
Q2(Rl= QUO
#I=4
/
Then #5, - 6 and
QI I €(07»S0)Ru102(p1¢2R0304(/)3(04

o,(p€S

R 0 1 0 2 U 1 02 R U 3 0 4 0 3 0 4 -|- €(07, §0)Ru102(P1(P2R0304(P3(P4°

u'S] o;écp€S

The first sum has only three different monomials, each of them appearing twice is them
sum. The second sum has different monomials (corresponding to subsets of cardinality
2 of SQ) and each of them appears twice.

Definition 9.3.13. If (M, g) is a closed, compact, oriented, Riemann manifold, m

dim M, and w is nonnegative integer. If m - w is odd we set

up(ml 0.
428 Lectures on the Geometry of Manu'olds

If m - w is an even, nonnegative integer, m w 211, then we set

1
uw(M,9) l27Tlhh! M Qh ( R )| dVm I .

We will say that P*wlM) g) is the weight w curvature measure of (M, g). We set
1
ld.»Uw I := (2.rl"h! Qh ( R )|dVm I ,
and we will refer to it as the (weight w ) curvature density.

Remark 9.3.14. (a) Let us observe that for any Riemann manifold M, orientable or not,
the quantities Iduwl are indeed well defined, i.e., independent of the choice of local frames
used in their definition. The fastest way to argue this is by invoking Nash embedding
theorem which implies that any compact manifold is can be isometrically embedded in an
Euclidean space. For submanifolds of R", the proof of the tube formula then implies that
these densities are indeed well defined.
We can prove this by more elementary means by observing that, for any finite set I, the
relative signature e(0, LPI of two permutations up, O' : I -> I is defined by choosing a linear
ordering on I , but it is independent of this cnoiee.
(b) The weight of the curvature density has a very intuitive meaning. Namely, we should
think of l w ( M , g) as a quantity measured in meter" .

Proof of Lemma 9.3.10. Consider the integral

2
We have e 6-lt°l2 SO that
C OO C OO
AC e
I(h1, . . . , hc) = H (/
j=1 -oo
6-8252h1 ds
H (/
J 1 0
52
52hj ds

(u 32)

_
C OO C
2hi-I-1\
H (/
j=1 0
e'"'th'j'1/2du
)p HF(
j 1
1
F t; and recalling
•
On the other hand, using spherical C00td1 yes
N3 ' laI» w that h
$1 -|----|- he, we deduce that

lwel2hcdw -p 2

) (/ p2h+
oo C_ 1
I(h1, 7 e dp
0
2
(u p )
OO
1 c-l-2h

2
lwll2h1 -~~ lwel2h"doJ e'"u 2 'ldu

I
SC-1 0
1 e+2h
I$1 I 2h1 . . . ac Zhedw
2
F
2 ) (/ Sc-1
I .
9.3.4 Tube formula
an Euclidean space
- Classical Integral Geometry

Gauss-Bonnet formula for arbitrary submamfolds

Suppose Mm C R" is a closed, compact submanifold of Rn. We do not assume that M is

of
429

orientable. As usual, set C = n - m, and we denote by g the induced metric on M. For

every sufficiently small positive real number 1° we set
M := {;z: E RTL; dist (at, Ml = T'} = 3'lll,.(M).
The closed set M is a compact hypersurface of R", and we denote by gr the induced
metric. Observe that for r and 5 sufficiently small we have
MMfr) = 'r'I"+(M) - 'r,,-(M)
so that,
V(M,.,e) = V(M,v° + 5) - V ( M , 7° - 5),
which implies that
1+2h
W1+2h(°$ Mn-1-2h(Mr» g)
h>_ 0

k k
(T + 6)
c C
c-l-2k +2 -(t~-6) +2 }un-¢-2k(M,9)-
k>0

We deduce
2 e-I-2k
( )
- 1 2k-2h
Iin-1-2hlMragrl I U1 2h c+2k 'ac + Hn-e-2klMag)°
+ w l+2h
k>0

We make a change in variables. We set

p : = n - l -2h, w n-c-2k=m-Zk.

T h e n c + 2 k = ' n , - w , l + 2 h = n - p , e - | - 2 k - l - 2 h = p - w , so that we can rewrite

the above formula as
m
n W Un w
UplMr»grl I 2
w=O ( ) 71-17
r p ' w p w (M, 9)- (9.3.14)

In the above equality it is understood that uw(m) 0 if m w is odd. In particular, we

deduce that If the codimension of M i5 odd then
1 MM Mm Q) : : 2u»p(M>9)> V0 S P < m dim M. (9.3.15)

If in the formula (9.3.14) we assume that the manifold M is a point, then we deduce that
M is the (n - 1)-dimensional sphere of radius r, M = sp-1, and we conclude that

where
l"LPIST
n-1 _ n
)- 2 p
()
can
wn_p
up: 2
n
We p H
up, n - p = 1 mod 2,

is defined by (9.1.20). The last equality agrees with our previous computation
(9.3.16)

[ p]
(9.3.13).
430 Lectures on the Geometry of Manzfolds

If in the formula (93.14) we let p = 0 we deduce

i0lMr)grl I 2/i0(M,gl, V0 < T' << 17 if coding M is odd.

Observe that the tube 'lll,.(M) is naturally oriented, even though the manifold M may not
be orientable. The Gauss-Bonnet theorem for oriented hypersurfaces implies

x(Mv~) I PL0lMr)grl

so that
1
u0(M,9) §XlMv~l) V0 <1°<< 1. (9.3.l7)

Theorem 9.3.15 (Gauss-Bonnet). Suppose M $5 a closed, compact submanzfold of an

Euclidean space Rn Denote by g the induced metric. Then

u0(M,g) = x ( M ) -

Proof. If m = dim M is odd, then both x ( M ) and no (M) are equal to zero and the identity
is trivial. Assume therefore that m is even. If the dimension n of the ambient space is odd,
then the Poincaré duality for the oriented n-dimensional manifold with boundary 'f[',.(M)
implies

x(-mfl I x( <'9'll»»~(M)) Z QX('1',.(M)l = 2x(M)»

and the theorem follows from (9.3. 17).
If n is even, we apply the above argument to the embedding

M <->]R" ->R"+1 7

where we observe that the metric induced by the embedding M > Rn-P1 coincides with
the metric induced by the original embedding M <-> R" .

Remark 9.3.16. (a) We want emphasize again that in the above theorem we did not require
that M be orientable which is the traditional assumption in the Gauss-Bonnet theorem.
(b) By invoking the Nash embedding theorem, we deduce that the tube formula implies the
Gauss-Bonnet formula for any compact iemann manifold, orientable or not.

Let us record for later use the following corollary of the above proof.

Corollary 9.3.17. For every closed compact, smooth submamfold M of an Euclidean

space V such that dim V - dim M i5 odd we have

x ( M ) = 2x(<9T»~(M) ), to < 1° 1.
 Classical Integral Geometry 431

9.3.5 Curvature measures of domains in an Euclidean space

The second fundamental form of a submanifold is in fact a bilinear form with values in
the normal bundle. If the submanifold happens to be the boundary of a domain, then the
normal bundle admits a canonical trivialization, and the second fundamental form will be
a scalar valued form. The next definition formalizes this observation.

Definition 9.3.18. For any relatively compact open subset D of an Euclidean space V, with
smooth boundary 3D, we define the co-oriented second fundamentalform of D to be the
symmetric bilinear map

SD : Vect (up) X Vect lap) -> C°°(c9D),

5D(X»Y)=(D x Y ) . No X,Y€Vect(3D),
where n : 3D -> V denotes the outer unit normal vector field along AD.

Suppose D C Rm-l-1 is an open, relatively compact subset with smooth boundary M :=

BD. We denote by n the unit outer normal vector field along M := BD, and by S = SD
the co-oriented second fundamental form of D. For every symmetric bilinear form B on
an Euclidean space V we define try (B) the j-th elementary symmetric polynomial in the
eigenvalues of B, i.e.,
ZN tIli(B) = det(]1v + zB).
j20

Equivalently,

try B = tr IAA B =AW->A2v)

We define the tube of radius 7' around D to be

MD) 513 E Rm+1; dist D) T' .

We denote by IBM the exponential map

IBM : lTRM+1)lm -> Run-pl 7

(x,pl I ) 1E;mlX,P) =p+x, pE m, X 6 TpRm+1

For 1°> 0 we denote by A C (TR'"+1)IM the closed set

A := { ( M ( p ) » p ) ; P 6 M, t G [0»7°] }-
For sufficiently small 7° the map Em defines a diffeomorphism
EM : A -> It,.(D) \D,
so that

vol (MD) ) = vol (D) + IEMdVRm+. .

A
432 Lectures on the Geometry of Manzfolds

Fix Po E M and a local, positively oriented local orthonormal frame

e17°°°7e'f7l

of TM defined in a neighborhood U of Po in M, such that, for every p 6 U, the collection

(My), € 1 ( P ) » - - - »am(p)l
is a positively oriented, orthonormal frame of R". We obtain a dual coframe 0, 01, . . 7 0n
As in the previous section, the pullback of IEmdVRm+1 to A, has the description
1EmdvRm+. = det(]l - ¢Sm )dt A d 01 A /\ 771
0
771

Z t r j l -SmW dt /\ do1 A A 0 'rn

j 1

We deduce

As
EmdVRm+1
j=0
m
.
JI +1
-1

(f M
trill-smldvm I •

Define
1
mm-MD)
UP (f M
try(-Sm)dVm 7 0 g j g m,

and
,Um+1lD) := vol (D)
so that using the equality oy (j + llwj+1 we deduce the tubeformulafor domains,
m-I-1
+1-k
vol (11",-(D) ) (Um-I-1-k7"M /lk(Dl- (9.3.18)
k=0

Theorem 9.2.5 shows that, just as in the case of submanifolds, we have ,u0IDI x( D).
Definition 9.3.19. Suppose D is a relatively compact domain with smooth boundary of an
Euclidean space V, dim V = n. Then the curvature densities of D are the densities Idle |
on 3D defined by
l
I t f n - j l-SD) ldVaD IN
o n

ld.,LLj
O'

where ldV3DI denotes the volume density on 3D induced by the Euclidean metric on V,
and SD denotes the co-oriented second fundamental form of 31).

We denote by Iwm-I-1
T'
the ball of radius T' in ]Rm-l-1_ Then,

T€lDr
'rn+
II -_ ]D)"r-I-51
7r+
7

SO that

W m + 1 l T -|-
&)M-I-1 w 7r+1-k&""+1-k IJ»k(]D)}'~"+1)~
k>o
 Classical Integral Geometry 433

We conclude

I-'kl]Dn+1)
k
Wm+1-k k ( 'l T'

Suppose X --> Rm-1-1 is a closed, compact smooth submanifold. Then for every suffi-
k
[
m-| -

1
T'
k
(9.3.l9)

ciently small T' > 0, the tube D := 'ill,(X) is a compact domain with smooth boundary
and
MDT') = 'r'p.+(X).
The tube formula for X implies that
w 963/l»fn+1-j(D1~) wk If + (°5)k/lm+1-1<=lX)-
j>0 k>0

We deduce that
l If lg_ .
,U'rn+1-j(D.°) =
wi k2j
wk
()
J
1° 3/1m+1-/<=(X)~

We set n m + l and we make the change in variables

p n J, w n-k.
Then k j = p - w and we obtain the following generalization of the tube formula

HPITTIXI )
1
n-w
II
n w
n_p TO-w,UwlXl

(9.3.20)
7°P-wl~'wlXl.
w

In particular, we have
ii_1}}up(1II~(X)l HpIXI to g 1 g dimX.
7 (9.3.21)

9.3.6 Crofton formula for domains o f f Euclidean space

Suppose D is an open, relatively compact subset of the Euclidean space R" with smooth
boundary M = 31). We denote by g the induced metric on M, by Gr C the Grassmannian
of linear subspaces of R" of codimension c, and by Graff C the affine Grassmannian of
codimension e affine subspaces of R" .
Recall that on Gr C we have a natural metric with volume density ld'YCI and total volume
n 1

Vi
HJ 0 O' j
o n

C
_1 w(
m 1
j -
e
O'

We rescale this volume density as in (9.1.21) to obtain a new volume density ld1/C I with
total volume

Gr C
ld1/C I
H
n
C

As explained in Subsection 9.1.3, these two densities produce two invariant densities la@ I
(9.3.22)

and ld'17CI on Graff e which differ by a multiplicative constant.

434 Lectures on the Geometry of Manzfolds

Theorem 9.3.20 (Crofton formula). Let 1 <p g n - 0, and consider the function

f : Graff" -> R, f ( L ) = l»p(L re D).

If the function f i5 ld17l-integrable, then
p+ C

[ ] p
UP+c (DI
raffC
Up(LfwDlld'17C(Lll.

Proof. For simplicity, we set V ] R " , m = n - 1 = d i m M . We will carry out the proof
in several steps.
Step 1. We will prove that there exists a constant Qw, depending only on m, c, and p
such that

§'n,e,p,Up+c (D)
- GraffC l'p(L n D)ld'17e(L)l.

p+c
Step 2. We will show that the constant g is equal to by explicitly computing both
sides of the above equality in the special case D = ]
Step 1. We will rely on a basic trick in integral geometry. For every S G Graff" we denote
by [S] 6 GrC the parallel translate of S containing the origin, [S] = S - S. We introduce
the incidence relation
C
U={('v,S)€V X Graff v@S}CV X Graff
Observe that we have a diffeomorphism
U->V X Gr e 7 U 9 ( v , S ) |. >('u,[S])€V X Gr c 7
with inverse
V X Gr°(V)3(v,L) |. >(v, 'U + L ) € U .
We obtain a double libration
j

C
V Graff
and we set

:1<m> := £-1(M) = { (v,S) 6 V X Graff e . u € S V W M } .

7

Since dime = d i V + d i m GrC = n-I-c(n - c ) we deduce

dimU(M) = n-|- c(n - c) -codimM = 11+ e(n - c).
Again we have a diagram

U(M)
e

M Graff C
 Classical Integral Geometry 435

The map T' need not be a submersion. Fortunately, Sard's theorem shows that 1° fails to be
a surjection on a rather thin set.
Denote by Graff ° ( m ) the set of codimension C affine planes which intersect M
transversally. The set GraHI°(M) is an open subset of GraffC, and Sard's theorem implies
that its complement has measure zero. We set
U(M)* := 7°"1 l Graff eIM>).
The set U(M)* is an open subset of U(M), and we obtain a doublejbration

U(M)*
(9.3.23)
M Graff °(Ml

The fiber of T' over L 6 Graffe is the slice ML L VI M which is the boundary of the
domain DL :: IL O D) C L. o n

The vertical bundle of the libration T' : 3*(M) -> Graff ° ( m ) is equipped with a
natural density given along a fiber L O M by the curvature density id,uk1Ll | of the domain DL .
We will denote this density by id;i,l. As explained in Subsection 9.1. 1, using the pullback
r* ldel we obtain a density

luAu dl~*pLI X r*ldCI

on U*(M) satisfying,

:1*(m)
lAI / GraHI°(M) (f LOM
lduLI ld.%(L)l
P ) ,lGraH'c /Jpl-L nD)ld2;¢(L)l_

To complete Step 1 in our strategy it suffices to prove that there exists a constant f , depend-
ing only on m and C such that

€*ldAI 1 old,Ucl,
where the curvature density is described in Definition 9.3.13.
Set h := ( m - C The points in U(M) are pairs (x, L), where at 6 M, and L is an
•

affine plane of dimension h -|- l. Suppose (3120, Lo) E 3*(M). Then we can parametrize a
small open neighborhood of (330, Lo) in U* (M) by a family

(16, € 0 ( S l » ellSla ' ' • 7 Ehlsl, € h + 1 l S l » ' 7 em(s)),

where :u runs in a small neighborhood of :UQ 6 M, S runs in a small neighborhood U0 of
lLol in Gr C so that the following hold for every S.
. The collection

{€00§)»€19)» ,oh(Sl, €h+1(S)» »am($)}

.. is an orthonormal frame of R"

The collection {en, 81, . 7oh} is a basis of S.
The collection span {$1, . . .,e;,} is basis ofTH30M re S.
436 Lectures on the Geometry of Manzfolds

. The above condition imply that 80 is a normal vector to the boundary of the domain
D re L C L. We require that 80 is the outer normal.

A neighborhood of (5130, Lo) in U is parametrized by the family

(vi, e0(s), 81lSl, . . . , € h l $ l 7 € h + 1 l $ ) 7 . . . 7 €,,,,(5) ),
where F' runs in a neighborhood of 5130 in the ambient space V.
We denote by SD, the co-oriented second fundamental form of D, by SL the co-
oriented second fundamental form of DL C L, and by ldVLmMI the metric volume density
o n L y . Then, i f w e sets = d i m L - p = m - c - p , w e d e d u c e
1
ld.,LLpLI I k tfkl-sLlldvLqml.
O'

In the sequel we will use the following conventions.

'£,j,k denote indices running in the set {0, . . . , h}.

0¢,B,'y denote indices running in the set {h + 1, . 7 m .
A, B, C' denote indices running in the set {0, 1, . . 7 m •

We denote by (HA) the dual coframe of (cAl» and set

TAB eA . (D dBl.
Then, the volume density of the natural metric on GrC is

ld'Ycl /\$0i
0,i

Then

am I a
D Foea X ld'Ycl AS
a
X ld'Yc I 7

and
L l
lAI X
U' k
d€tl-sLmmlldvLmml X I\@<* X ld'Ycl° (9.3.24)
a

The fiber of K : U(M) -> M over £130 is described by

Go I={lt*,€AlSll E a(m),
We set
G;0 := Gm() VW U * ( M ) .

Gxo ( M ) can be identified with the space of linear subspaces S of codimension C such that
T30M + S = V, i.e., the affine subspace kg -|- S intersects M transversally at kg .
Denote by n a smooth unit normal vector field defined in a neighborhood of xo in M ,
i.e.,
n(a:) J_ TxM , l7*($I3)l = 1.
 Classical Integral Geometry 437

Lemma 9.3.21. Suppose 5130 + S intersects M transversally at 5130. We set e A .. eArs)_

Then at the point k g 6 M we have
l(~ . €0ll . ldvml l91 A - - ~ 9 m l ,
i.e., for any X l , . . . , X m 6 TH30 M we have
.
l(n €0ll ' ldVMI(X17 . . . 7 X T T I l = 191 A - - A @ T T I I ( x 1 , . . . , x T T I ) .

Proof. It suffices to verify this for one basis XI , . , X m o f TG90 M which we can choose to
consists of the orthogonal projections f l , . . 7 f TTI of e l , . . . , et. These projections form
a basis since S intersects T$0 M transversally.
Observe that
.f i z e i a 'v'1 < i < h , f a eOf ea i n . .
Then
ldVMI(.f17°°° , f ml .
d€tl.fA f B)15A,B<_»m~
We observe that
. .
f f
fa . f 5 = 600 -
f 'in fa
Hang, na n . ea,
We deduce
ldVMllf1»
, f m ) 2 = det(]l - A),
• o n

where A denotes the e X c symmetric matrix with entries Rang, h -|- l p a , B < m = h + c .
If we denote by 'ii' the vector
Nh-+1
TO

6 Re,
II

Nh-}-c

which we also regard as a C X l matrix, then we deduce that A = '&"&'l .

This matrix has a C - l dimensional kernel corresponding to vectors orthogonal to ii.
The vector ii itself is an eigenvector of A, and the corresponding eigenvalue A is obtained
from the equality
AA: lal2a=> A = 1612 . 801
2

We conclude that
det(]1 Al= in . 801
2 =>
ldVml( f 17 7 f ml in €0~
On the other hand,
I91 /\ /\
We have again
. .
€ J 171j 7
. et
f
fa : O f 'v'l<1l,j<h<a,

SO that
l91 A---A9""l( f 17 7 f m) in 80
2

The lemma is now proved.

438 Lectures on the Geometry of Manzfolds

Lemma 9.3.22 (Euler-Meusnier). Suppose L 6 G r a f f e intersects M transversally, and

330 6 L H M . [ f n is a unit vector perpendicular to T000 M , then

$L : (n . €0)SDITmorw[L],

that is,
5L(€i»€j) = (n . €0)5D(€i,€jl» vi S M S zh.

Proof. We have
Bal 80 SL(€',1, et of ' I .
Let us now observe that the vector lDez e t ) is parallel with the plane L because the vectors
ii and of lie in this plane. Thus, De et decomposes into two components, one component
parallel to so, and another component, (Dee et)7', tangent to L n M. Hence
h
k
D ei ei SL(e,ej)e0 -|- Sjek.
k=1
Taking the inner product with n we deduce

SD(€i»€jl=( D eiejl . n =5L(@¢»€J)(€0 . H)-

From the above lemma we deduce

tIlkl'SLlr12ol Z in . e0lktIlkl'SDITx0Mn[L]))

for any (kg, L) 6 U*(M). In a neighborhood of (130, Lo) 6 J*(M) we have

1
ldAI(a:,L) = tfkl-sLlldvLmml X I\@°' X ld'Ycl
Of
a

in. e(n'°(tt(-sDITm~1L1))
7n

= 1 PA X ld'Yc I
Of A=1
(use Lemma 9.3.21)
l
U' k
in . €olk:-I1(ttk(-SDITmrw[L]l)ldVml X ld'ycl~
This proves that along the fiber G$0 we have

lAI/ldvml
l
Of
in . 801
k+1
lt1`1<>('SD|Tmomn[L])l|d7c|l[L])-
If we denote by 9(1L1, TU0M) the angle between [L] and the hyperplane T230 M we deduce
l k+1
ld.,\l/ldvMI ICOS 9([L], Tx0-MI ltfkl-sDITfvo M y [ L ] l ) |d%@|([L])-
Of
The map 9 ( k g , L) I > [L] 6 Gr" identifies G000) with an open subset of Gr" whose
G$0)

complement has measure zero. We now have the following result.

Lemma 9.3.23. Suppose that V is an Euclidean space, d i V = m -|- 1, H C V is a

hyperplane through the origin, and B : H X H -> R a symmetric bilinear map. Denote by
 Classical Integral Geometry 439

O ( H ) the subgroup of orthogonal transformations of V which map H to itself and suppose

that
f : Gr C ->]R

i5 an O(H)-invariantfunction. Define
GrH {S 6 Gre, S intersects H trasversally
e
..II
Then, for every 0 <k<m 0, there exists a constant g 'in,e,k» depending only on m,
C and k, such that

f(s) trklBIHms)ld»cl(s) trkTlBl f|d%|(Sl-

n o

i §'fTl,C,]< Z
rC

Proof. Observe that, for fixed f , the map B +-> Ik3J (f, B) is an O(H)-invariant homo-
geneous polynomial of degree k in the entries of B. We can therefore express it as a
polynomial

I k l f , Be pt(t1~1(B), . . . ,tIlklBl )
of t1'1<(Bl + C2f(tr1(B), ,trk;_1(Bl

Let us prove that Qf 0. To do this, we apply the above formula to a symmetric bilinear
form B such that
dim her B > m - k .
Thus, at least m k-| - l of the m eigenvalues of B vanish, so that trk(B) = 0. For such
forms we have

Ik(f,8) = Q(tt1(8), • • ,tIlk_1lB) ) .

On the other hand, for almost all S 6 GrH
C
we have

-
dims FW herB >m C k.
The restriction of B to S O H has m - C eigenvalues, and from the above inequality we
deduce that at least m - C - k of them are trivial. Hence
1 k l f , B l = 0, VB, dimkerB > m - is Qf = 0.
Now choose B to be the bilinear form corresponding to the inner product on H. Then
m
vs 6
trklBl
() k
and try(BIHnsl 7 GrH,

and we conclude that

m _c
k M
Now apply the above lemma in the special case
C
fld'y@l($) = of
m
()
k

1
H BoM, B -SD, f(S) = lcos9(S , H)lk+1
Of
440 Lectures on the Geometry of Manzfolds

to conclude that

we = we(-sD)ldvml = end/e
SO that

g/Lp+clDl
/M
€*ldAI
al*(M)
we
/GraH"°(M)
T* Adm Graff°(M)
#oIL n D)|d'7¢|(L)-

Thus, rescaling ld2;el to i n , we deduce that there exists a constant g depending only on
m and C such that

§/p+c ( D) Z /Graff °(M)

up(LVWD)|d'17c|(L)-

Step 2. To determine the constant g in the above equality we apply it in the special case
D = lD)71+1. Using (93.19) we deduce

§,Up+elDm+1l §Wp+c
l '1
m+
p+c

Now observe that for L 6 GraffC we set r = r(L) = dist (L,0). Then L D Dm-}-1 is
g
(-*)m-}-1

'Um-I-1-c-p P-I-c •

empty if r > 1, and it is a disk of dimension (m + 1 - C = dimL and radius (1 - r2)1/2

if T' < 1. We conclude that
,V.2lp/2 'r<1
L re HWY) = NplMm-H-cl X
p>1.
We set

llwmzv
__ NplMm
' -
+1_
C) - We
_ C _
-
(-U'rn-I-1-c m +l C

P (.0m-}-1-c-p P
Using Theorem 9. 1.16 we deduce

#oIL n ]n>m+1)ld/7CI(L)

(f [LP /*p lDm-I-1 m (az -|- [L])) ldV[L]; l(22l Id1/@l([L])

ld1/Cll[L]l
llln,c,p
L (f x €[L ] l ,
x
I l<1
(1 - | II 2) ld.[L];l(x)
P /2

=IIc,p

(9.3.22) m -|- 1
1"'"ftl,c,pIc,p Id1/¢(5) I = 1'fN,C,PIC,P C
rC

Hence

g
U-Jm-I-1

U-Jfrn-I-1-e-p
m+
P '*'
1
C
U.1m-I-1-e

U-7m-l-1-e-p P
C

w
c,p
+
C
1
 Classical Integral Geometry 441

Using spherical coordinates on Re we deduce

1

/am Re
(1 l¢vl2)'"/2dVRc 0' C 1 c-1(1 -t~2)p/2dt~
0

5 2 ( -sY'/2ds

(9.1.9) 0'0 1 5)
1"(§)1"(1 +
2 FU+§+§)
(9.1.10)
F (1/2) C m
F( go§ Wp+c

We
Hence
m-| -
p+c
1 U-)m+l-e(-Up-l-e [m + 1
§W7Tl+1 We C p

We deduce
U,)m+l(-4.)p-l-e

(.4.)p(.4.)€
(
m -I-
C I m-|- 1
p
c

Wp+c lm-I-ll Wp+c p _1_ e

g wpme
( p i ) lm+1-cl WPW6 ( ) p W p

We now describe a simple situation when the function Graff e E L I - ) L 0 l L M M l is

integrable.

Proposition 9.3.24. If the bounded domain D C R " with smooth boundary is tame, then
the function
Graff C 9 L » - > x ( L r W D ) = / l o w / / N D I
i5 bounded and has compact support. In particular; it i5 integrable and

muD) X(L n D) ld/7l(0l.

raff C

Proof. We know that there exists a tame structure S such that D 6 S". In particular,
M = 3D 6 S". Since the Grassmannian Graff" is semialgebraic, we deduce that GraffC
is also S-definable. Hence, the incidence set
U = ~{((x,Ll 6 M X Graff $ 6 L}
is also S-definable, and so is the map

The fiber of 'IT over L is the intersection L O M. From the definability of Euler characteristic
we deduce that the map
Graff C 9 L + - > X ( L f W M ) € Z
442 Lectures on the Geometry of Manu'olds

is definable. Its range is a definable subset of Z, i.e., a finite set. To see that it has compact
support it suffices to observe that since M is compact, an affine plane which is too far from
the origin cannot intersect M.

Definition 9.3.25. For any compact tame subset S C R", and for every integer 0 < p < 7 1
we define

a,,(s) x(L D) ld;;1(Ll.

II
raff P(R")

The proof of Proposition 9.3.24 shows that the integral in the above definition is well
defined and finite. This proposition also shows that

aplDl : : MpIDI
if D C R" is a tame domain with smooth boundary.
The above definition has a "problem": a priori, the quantity /i ,,(s) depends on the
dimension of the ambient space R" D S. The next exercise asks the reader to prove that
this is not the case.

Exercise 9.3.26. Fix tame structure 5, and suppose S C R" is a compact, S-definable set.
Let N > n and suppose z' : R" -> RN is an amine, isometric embedding.
(a) Prove that

GraH7'(R")X (Lm5)ld-I(L)
1/ GraH7 '(RN)X l U O z(' s l ) l of1/I( U )»

so that [Lp(S) is independent of the dimension of the ambient space.

(b) Prove that for every real number A we have

ii pa/\Sl: IAINa,,($)-
(c) Suppose 51, S2 C R" are two compact S-definable sets. Prove that

ii pIS1 u S2) ii p(51) + l 2lS2l ii pIS n S ) .

(d) Suppose T C R'n is a triangle, i.e., the convex hull of three non collinear points. Com-
pute Ii p(T), Vp > 0.

Proposition 9.3.27. Fix a tame structure 8. Suppose K C R" $5 a compact, S-definable

set, A $5 an arbitrary S-definable set, and A C K X A i5 a closed .S-definable subset of
K X A. For every A 6 A we set
A II {& E K; (a7>\) E A}_
Then the function A 9 A I >fpl)'l p P (AAA 6 R is bounded and Bore] measurable.
A

Proof. Denote by Graff pIK) the set of affine planes of codimension p in IRE." which inter-
sect K. This is a compact, definable subset of Graff". Define
F :A X Graff "(K) -> z, (A, L) »-> X<AA\ re L) 6 z.
 Classical Integral Geometry 443

Since the family {A,\ re L}A,L is definable, we deduce that the function F is definable.
Hence, its range must be ajinite (!?!) subset of Z. In particular, this also shows that F is
measurable and bounded.
Now observe that

fp(>\) FIG, L) ld'17l(L).

Graff p ( K )

The measurability follows from the classical Fubini theorem. The boundedness is obvious.

The above result has the following immediate corollary.

Corollary 9.3.28. Suppose the bounded domain D C R " i5 tame and has smooth bound-
ary. Denote by G r a f f D the set of a]§€ne planes of codimension C which intersect 3D
transversally. Then for any integer; 0 < p S n the function
GraffD 9 L +-> ,up(LVl D)
is bounded, measurable, and has compact support. In particular; it is integrable, and
p+c
[ ]
e ,Up+elDl
r a f fCs
la»plL ODD Id1/(Lll
raffC
nplLnDlld;;ll/;l.
Proof. The set GraffDe is a tame open and dense subset of Graff" SO that its complement
has measure zero. Moreover
#pIL OD) Z imp(/z cpl, VL e Graff Do C

while, by Proposition 9.3.27, the function L »-> np

(LOD) is bounded and measurable. This
function has compact support because the planes too far from the origin cannot intersect D.

9.3.7 Crofton formulwfor submanuolds o f f Euclidean space

In this last subsection we fix a tame structure S, and we assume that M is a closed, compact,
smooth, S-definable submanifold of dimension m of the Euclidean space R". We continue
to denote by Graff C the Grassmannian of affine planes in R" of codimension C. We want
to prove the following result.

Theorem 9.3.29 (Crofton Formula. Part 1). Denote by Graff ° ( M ) the subset of
Graff C consisting of a]§'ine planes which intersect M transversally. Then

MM)
Graff °(M)
,LLoIL (1 M)IJ;1(L)
I -ltraffc IJ»0(L nM)ld;;(L)l,

[Hal is,

ac(M) i MM)-
444 Lectures on the Geometry of Manzfolds

Proof. We can assume that k = codimM < 1. For every ac 6 R" set d(a:) := dist (ac, M).
Fix R > 0 such that for any ac such that d(ac) < R there exists a unique point x 6 M such
that

For every T ' < R consider the tube of radius r, around M,

D := 'lll,.(M),
and set M 8'lll,.(M). For uniformity, we set D0 i M. From (93,15) we deduce
IC(M) = lim #e(D,~).
v ( )

The tube D is a tame domain with smooth boundary, and Theorem 9.3.20 implies

MDN x(L (1 D»~l|du|(L)-

--raffC
Thus it suffices to show that

lim
fr->0
/
Gray
X(L VW D,,,)ld1/I(/;)
lGraH'e xILMI.
For 1° 6 [0,Rl we define
f : Graff c -> R, f,,,lLl = x(L VW DI.

Lemma 9.3.30. There exists C' > 0 such that

lf,.(L)l g C, VL Q Graff c 7 r 6 [0,R).

Proof. This follows from Proposition 93.27.

Let
Graff°(M,.) {LC Graff e .7 L intersects M transversally
Observe that Graff°(Ml is an open subset of Graff C with negligible complement. For
every r > 0 we set

9C,~ {L 6 Graff °(Ml; L 6 Graff CIrs; x(LV\D8) =x(LVWMl, Vs 6 (Owl

Observe that
Graff CIM 1) C Graff ° ( M , 7 ° 0 l » WE >_ T 0 -
To proceed further we need the following technical result, whose proof will presented at
the end of this subsection.

Lemma 9.3.31. The sets X are measurable in Graff" and

U 36, I Graff ° ( m ) .
"r>0
 Classical Integral Geometry 445

Set

GraffC* := { L 6 Graffe; L O D R ¢ vi
The region Graffi (M) is a relatively compact subset of GraffC, and thus it has finite
measure. Define
Xi: rem Graff ii : Graff C*
For 0 < 7° < R we have

MD) f2~(L)ld'17@l(L) f2~(L)ld'17@l(L)

r C*

/or;
flLlld;;cl +
/ H8;
f,,(Llld;;CI = 2
or:
f0(L)ld'7¢l +
H:
I,,(L)ld'7CI.

Hence

clm,,l- 'x Io(L)IJ;CI < If»»(L)lld'7¢l §C'vol(H;',).

;< 9;
We now let 1° -> 0, and since vol ( H ) -> 0 we conclude that

MM) : : lim I~'clDv'l = r->0

1°->0
lim
Xr
f0(L)ld'17¢l
- l}raH'e
This concludes the proof of Theorem 9.3.29.

Proof of Lemma 9.3.31. We will prove that for any given L0 6 Graff °(m) there exists
and p = P(Lol such that
Lo 6 Xp.

The measurability follows from the fact that DC is described using countably many boolean
operations on measurable sets.
Consider the normal bundle
N : (TM -> m.
For x in M we denote b Nx the fiber of N over at.
Let y 6 L0 VW M. We denote by fig the orthogonal complement in L0 o f Ty(L0 re m ) ,
fig = L0 al (Ty(L0 ml)_L.
We think of fig as an affine subspace of R" containing y . Because L0 intersects M
transversally we have
dimery = dim fig = k = codi M.
For every 1° > 0 we set Ny0(r) := fig re D , see Figure 9.2.
The collection (No lye L0rwm forms a vector subbundle N0 -> L0 O M of I T R H I I L o V - M '
The exponential map on TR" restricts to a map

EL 0oM I NO * Lo.
446 Lectures on the Geometry of Manzfolds

I
\
I M
\
/

l
|
I
I
l
I

m
|
r
l
I Not)
LE
/
l
I
l
I
l

•y
|

/
\ l
l
1
I
I
I
I
I
I
I Dr
I
I
p * " - - 1 - - " § q
-' C * *
I
- - *-.
I
¢ " l ~* `
I 5
I 11
1
1 \
l

Fig. 9.2 Slicing the tube Dr around the submanifold M by a plane L0 .

Denote by 6 the pullback to N0 of the distance function :u »-> d(:z;) = dist (az, M),
AL0 = d O IEL0nm : N0 ->]R.
For every y 6 L0 ii M, the restriction to fig of the Hessian of d at y is positive definite.
Hence, there exists p = p(L0) > 0 such that the map
EL0riM : {a: G NO; 5(;z3) g p} - > L 0 O D p
is a diffeomorphism. We deduce that we have a natural projection
L0 VI Dp -> L0 M M,
which is continuous and defines a locally trivial libration with fibers fig (p) .
For every y 6 L0 O M, the fiber Ny°(0) is homeomorphic to a disk of dimension k.
Thus, L0 O Dp is homeomorphic to a tube in L0 around L0 (W M C LQ, so that
xIL0 M pp. I oIL() VW
The downward gradient flow of the restriction to L0 O Dp of the distance function d(a2)
produces diffeomorphisms of manifolds with boundary
L0V1Dp%L0MD,., 'v'r€(0,p).

Hence
xIL0 M D ) xIL0 M pp) xIL0 M ml, VT E
I I

Since the restriction to L0 re Dp of the distance function d(ac) has no critical points other
than the minima y 6 L0 re M, we deduce that L0 is transversal to the level sets
{d(;13) = 1°} = M , VT 6 (0,p].
This proves L0 6 Xp.
 Classical Integral Geometry 447

Corollary 9.3.32. Suppose C C R2 i5 a smooth, closed, compact tame curve. For every
line L 6 Gr 1(]R2) : Gr 1inv) we set
7"Lc(Ll Z = # I L VW cl.
Then the function L »-> n g ( L ) belongs to L°°(Gr1(]R2), ld17l21), has compact support
and
length (C) NclLl\d12,1 I (L)-
Gp1lR2)

Remark 9.3.33. Theorem 9.3.29 offers a strange interpretation to the Hilbert-Einstein

functional. Suppose M is a tame, compact, orientable submanifold of the Euclidean space
R". Let m = dim M, and denote by h the induced metric on M. Then
l
#vi-zIM) HI = -27r
€M(h),
where €m( h) is the Hilbert-Einstein functional
am(h) M s(hldvm(h).

Denote by Graff m-2(M) = Graff ( M ) the set of codimension - 2) affine planes in

Rn which intersect M along a nonempty subset, and by Graff*m 2lml C Graff m-2(M)
the subset consisting of those planes intersecting M transversally. For every L E
Graff;n-2 ( M ) , the intersection LOM is a, possible disconnected, smooth, orientable Rie-
mann surface. By Sard's theorem the complement of Graff;n-2(]l/I) in Graff 1"-2(M)
has zero measure.
For every R > 0 we denote by Graff M-2(R) the set of codimension (m - 2) planes
in R" which intersect the disk IDS of radius R and centered at 0. Note that since M is
compact, there exists R0 > 0 such that
Graff '*"°'2(M) C Graff rz-2(R), VR > Ro.
We set
cIR) ld.;;(L)l, c(Ml Q; ld;;(/;l1,
GraH"m'2(R) GraH'm'2(Ml
Using Crofton's formula in Proposition 9.3.24 and the simple observation that
X(LVW]D)"R)=1, VL Q Graff M-2(Rl»
we deduce that

C(Rl Z #fn-2(D"R) Lim QRm 2

Observe now that cllm ld;7(/;l1 is a probability measure on Graff '"'2(R), and we can
regard the correspondence
Graff m-2(R) 9 L I > x ( L r 1 M ) e Z
as a random variable OR whose expectation is
1
(oR) m-2 M = 1 8 h.
c(R)/LL ( ) 2'rrc(R) m( )
We deduce
amah) = Q 'Wm-2RM-2('lR) = 27r I
lim RM-2<€R)-
Wm-2 R_>oo
I
448 Lectures on the Geometry of Manzfolds

Theorem 9.3.34 (Crofton Formula. Part 2). Denote by Graff °(M) the subset of
Graff C consisting of amine planes which intersect M transversally. Then, for every
0 < p < m-ewehave
19+ C
[ ]p P¢p+clM) I / Graff° ( M )
lplLm M)ld;;I(Ll I /Graff"
a,,(Lim)lw>(L)l.

Proof. We continue to use the same notations we used in the proof of Theorem 9.3.29.
From (93.15) we deduce

lip+elMl : 12% »Up+clDv°l-

The set D is a tame domain with smooth boundary, and Corollary 9.3.28 implies
p-I-c
C ] »U»p+clDv°)
= /
Graff"
lJfpll. m D~)ld'17l(L) = / GraffC
np(LeD,,)l¢{;l(Ll.
Thus it suffices to show that

lim
1*->0
/
Graff"
a,,(L re D,,,)ld;7(L)l = /
Graff C
a,,(L m M)ld'7(L)l.

For r E 10,R) we define

gr Graff C -> R, 9,.(L> = a,,(L m D,.>.

From Proposition 9.3.27 we deduce that
EIC' > 0 : lg»,~(Lll < C, VL E Graff C
7 1° E [0,R). (9.3.25)

Lemma 9.3.3S.
1- 90(L)
gT`lL) :

for all L 6 Graff °(Ml-

Proof. Let L 6 Graff °(Ml and set S := L nD. We have to prove that
lim
?°\,0
apluS""l : ii pls0l'

Denote by Graff "(SO) the set of codimension p affine planes in R" which intersect SO
transversally. The set Graff" (50) is an open subset of Graffp, and its complement has
measure zero. Then,

9'rlLl ii (50 X(U O S»~ll<i'171"(U)l x(U VW $.~)|d'17"(Ul|-

,Graff? Graffp(S0)
(9.3.26)
From Proposition 9.3.27 we deduce that the function
Graff P X [0,R) 9 (u,t~) »->X(UfWS,~) e t
is definable, and thus bounded.
 Classical Integral Geometry 449

Arguing exactly as in the proof of Lemma 9.3.31 we deduce that, for every U E
Graff p (SQ), there exists p E (0, R) such that, for every 7° E (0, p), the set U re S, is
homotopy equivalent to U n So- Hence
1- s l - (U VU Q Graff
X l U m 7" - X VWS0I
r) pls0l.
If we let T' go to zero in (9.3.26), we deduce from the dominated convergence theorem that

limg,.(L) x(umgolld;7p(Ull Ii p(S0)-

'r\,0 Graff?
Using the fact that S0 is a smooth, compact tame manifold we deduce from Theorem 9.3.29
that

ii p(Sol Mpl5ol 90(Ll-

Using (9.3.25) and Lemma 93.35, we deduce from the dominated convergence theorem
that

lim
'r\0
/
Graff
i i i,,(L n Dt~)ld17(L)l
IGraH'c M plLmmlld,7(Lll

l}raH` °(M)
l»LplL M)ld'7(L)l.

Remark 9.3.36. (a) The tameness assumption in the Crofton formulate is not really needed,
and they are true in much more general situations, see [45], Sect. 2.10, 3.2.
We have chosen to work in this more restrictive context for two reasons. First, the
tameness assumption eliminates the need for sophisticated analytical arguments. Second,
we believe that geometers should become more familiar with tame context in which the
geometric intuition is not distracted by analytical caveats.
(b) The subject of integral geometry owes a great deal to the remarkable work of
S.S. Chern who was the first to have reformulated the main problems from a modern point
of view, and push our understanding of this subject to remarkable heights.
It was observed by Chern that the Crofton formulae that we have proved in this section
are only special cases of the so called kinematic formulae which explain how to recover
information about a submanifold of a homogeneous space such as R", by intersecting it
with a large family of simpler shapes. In the Euclidean space, these other simpler shapes
could be affine planes, spheres of a given radius, balls of a given radius.
For example, given a compact, m-dimensional submanifold M C R", and a radius R,
we can ask what is the value of the integral

Rn
X(M m BRILII) )ldH2l»
where BR(X) denotes the ball BRILE) yeaR"; lY-HIIS R }. One of the kinematic
formulae states that
n

Rn
X(M re BRILE) ) Cv7z,n,i#zlMlRn- 7

i 0

where C'rn,n,i are universal constants, independent of M .

450 Lectures on the Geometry of Manzfolds

The above equality is strikingly similar to the tube formula. To see the origin of this
similarity note that BR(;z:) re M 7é 0) ~: 'lllR(M). If R is sufficiently small, the above
formula does indeed specialize to the tube formula, but the kinematic formula extends
beyond the range of applicability of the tube formula.
A detailed presentation of these ideas will send us too far, and instead of pursuing
further this line of thought, we recommend to the reader to open the classical, yet very
much alive monograph of L. SantalO [115], for a more in depth look at this subject.
 Chapter 10

Elliptic Equations on Manifolds

Almost all the objects in differential geometry are defined by expressions involving partial
derivatives. The curvature of a connection is the most eloquent example.
Many concrete geometric problems lead to studying such objects with specific proper-
ties. For example, we inquired whether on a given vector bundle there exist flat connections.
This situation can be dealt with topologically, using the Chern-Weil theory of characteristic
classes.
Very often, topological considerations alone are not sufficient, and one has look into
the microstructure of the problem. This is where analysis comes in, and more specifically,
one is led to the study of partial differential equations. Among them, the elliptic ones play
a crucial role in modern geometry.
This chapter is an introduction to this vast and dynamic subject which has numerous
penetrating applications in geometry and topology.

10.1 Partial differential operators: algebraic aspects

10.1.1 Basie notions

We first need to introduce the concept of partial differential operator (p.d.o. for brevity) on
a smooth manifold M. To understand the forthcoming formalism, it is best that we begin
with the simplest of the situations, M = RN .
Perhaps the best known partial differential operator is the Laplacian,

A C°°(]RN)->C°°(RNl, Au <%2U,

where, as usual, 3; := 3,,,¢. This is a scalar operator in the sense that it acts on scalar valued
functions. Note that our definition of the Laplacian differs from the usual one by a sign.
The Laplacian defined as above is sometimes called the geometers' Laplacian.
Next in line is the exterior derivative
d : Qk(1Re.N) -> Q'<=+1(RN).
This is a vectorial operator in the sense it acts on vector valued functions. A degree k
I
form to on RN can be viewed as a collection of Nk ) smooth functions or equivalently, as a

451
452 Lectures on the Geometry of Manzfolds

smooth function an : RN -> RCE). Thus d can be viewed as an operator

d : COO IN, R(il)) -> IN,1RQ.(k+.)) _
coo
N

It is convenient to think of cOO (RN , RV) as the space of smooth sections of the trivial
bundle R" over RN.
For any smooth K = R, (C-vector bundles E, F over a smooth manifold M we denote
by O p ( E , F) the space of K-linear operators
c°°<E> -> c°°(F>.
The space Op ( E , El is an associative K-algebra.
The spaces of smooth sections cOO (E) and cOO (F) are more than just K-vector spaces.
They are modules over the ring of smooth functions cOO ( M ) . The partial differential oper-
ators are elements of Op which interact in a special way with the above C'°°(M)-module
structures. First define
PDO 0(E,F) := Hom(E,F).
Given T E PDO0, u E C'°°(E), and f e Cvoo(M), we have T(fu) - f ( T u ) 0 or, in
terms of commutators,

IT, flu = T(fu) - f ( T u ) = 0. (10.l.1)

Each f E CV0°(M) defines map

ad(f) U p (E,F) -> 019 (E,F),

by
ad(f)T == T of -f O
T = IT, fl, W 6 Op(E, F).
Above, f denotes the C'°°(M)-module multiplication by f. We can rephrase the equality
(10. 1. 1) as

PDO " L E , F ) = { T e U p ( E J ) , ad(f)T 0 W' E C°°(M)} her ad .

Define

PDO (fol) (E, Fl := her adm+1

{T E Op (E,F); T E kerad(f0)ad(f1) - .ad(fml, Vfi E Cvoo(M)} .
The elements of PDO (fol) are called partial deferential operators of order § m. We set
PDO (E, F) == U PDO (m) ( E , F).
7m>0

Remark 10.1.1. Note that we could have defined PDO lm) inductively as
PDO( m ) I {T Q Op [T, f] Q PDO (m-1) , of Q c°°<m>}.
This point of view is especially useful in induction proofs.
 Elliptic Equations on Manifolds 453

Example 10.1.2. Denote by R the trivial line bundle over RN . The sections of K are
precisely the real functions on RN . We want to analyze PDO (1) = PDO( 1 ) ( m ) .
Let L E PDO ( l ) , u,f E cOO(nN). Then [L,f]u = o(f) u, for some smooth -
function o(f) E cOO (RN). On the other hand, for any f , g E cOO (RN)

0(fg)u = [L, flu = [L, f](gu) -|- -

f([L» 9110 = 0 ( f ) 9 u -|- f0(9) 'u-

Hence 0(fg) = o(f)g -|- fo(g). In other words, the map f »-> o(f) is a derivation of
c°°(RN), and consequently (see Exercise 3.1.9) there exists a smooth vector field X on
RN such that
(t(f) x-f8 of E COOIRN).
Let ,u := L(1) E c°°<RN). Then, for all u E C'°°(]RN), we have

Lu=L(u~1)= [L,u]~1+'u,-L(1)=X~u+/i~u.

Lemma 10.1.3. Any L E PDO("")(E, F) is a local operator, i.e., V u e C'°°(E),

supp Lu C supp u.

Proof. We argue by induction over m. For m = 0 the result is obvious. Let L 6

PDO (m-I-1) and u E C°0(E). For every f E COO(M) we have
L(fu) = [L, flu + f L u .
Since [L, f] G PDO (m) we deduce by induction

supp L(fu) C s u p p u U s u p p f, 'v'f E C ° ° ( M ) .

For any open set (9 such that (9 D supp u we can find f E l on supp u and f E 0 outside
(9 so that few E u). This concludes the proof of the lemma.

The above lemma shows that, in order to analyze the action of a p.d.o., one can work
in local coordinates. Thus, understanding the structure of an arbitrary p.d.o. boils down to
understanding the action of a p.d.o. in PDO(m)(K'°,Kql» where Kp, and Kq are trivial
K-vector bundles over RN . This is done in the exercises at the end of this subsection.

Proposition 10.1.4. Let E , F, G be smooth K-vector bundles over the same manifold M.
1fp E PDO (""*)(F, GI and Q Q PDO ("*>(E, F) then P o Q Q PDO ("*+")lE, G).

Proof. We argue by induction over m -|- n. For m -|- n 0 the result is obvious. In general,
iff E cOO ( M ) , then
[P o o

By induction, the operators in the right-hand-side have orders < m-I-n-1. The proposition
is proved.
454 Lectures on the Geometry of Manzfolds

Corollary 10.1.5. The operator

L IaO(ac°)30' : c°°(RN) -> c°°(1Re.N),
lol<m
a ala 7 04N
N
E Z>0, Ia I Qi 7 a<.=a. ...ING n 7

'in
i5 a p.d.o. of order < m.
Proof. According to the computation in Example 10.1.2, each partial derivative <9i is a 1st
order p.d.o. According to the above proposition, multiple compositions of such operators
are again p.d.o.'s.

Lemma 10.1.6. Let E , F -> M be two smooth vector bundles over the smooth manifold
M. ThenforanyP€ PDO ( E , F ) , a n d a n y f , g € C'°°(M)
ad(fl . (adlai)pl Z ad(g) . (f1d(f)Pl-
Proof.
ad(f) (ad(9)p) = HHQ1, f ]
Hp,f1,9] + [p,lf,gH = Hp,f1,9] = ad(g) (@d(f)p)-

From the above lemma we deduce that if P E PDO (m) then, for any f1, ,fm 6
cOO ( M ) , the bundle morphism

adlfll adl.f2l . . . ad(fm)P

does not change if we permute the f's.

Proposition 10.1.7. LetP 6 PDO' " " ( E , F ) fiagi 6 c°°(m) U = 1, . . , m ) such that,
atapoint £130 E M ,
dftz(1u0l d9i(;v0) Q T;0 m, W 1 7 . . . 7 m.
Then
{3»d(f1l adlfzl a • a a»d(fm)P} lx= {&d(g1) adlgzl • a • &»d(9m)P} Iwo
In the proof we will use the following technical result which we leave to the reader as an
exercise.

Lemma 10.1.8. For each 5130 6 M consider the ideals of cOO (M)
m H30 {f Q C°°(M); f(920l I 0 },
3000 {f Q C°°(M); fag) I 0, d f m ) I 0, }
Then 3H30 = m20, i.e., any function f which vanishes at $0 together with its derivatives can
be written as
f gjh j g h j Q masJ .
j
 Elliptic Equations on Manifolds 455

Exercise 10.1.9. Prove the above lemma.

Proof. Let P E PDO (m) and f , go E C°°(M) such that

dfilfo) = dgilfol W = 1, ,m.
Since ad(const) 0, we may assume (eventually altering the f's and the g's by additive
constants) that
ftzlfo) = 9£050) Vi.
We will show that
f&d(f1) &dlf2l • • • ad(/'mlP} Iwo i&dlgll adlfzl ' • ' ad(f,n)P }|x0-
Iterating we get the desired conclusion.
-
Let QS = fl - go, and set Q := ad(f2) -ad(fm)P E PDO (1) We have to show that
{f1d((z5)C2} l 2 2 0 2 0. (10.1.2)
Note that QUO E 3230 so, according to the above lemma, we can write
(l)= C1j5j, 0%/3j Q m$0 .
j
We have

{ad(d))C2} I fro
{ &d(o¢j5jlC2
} xo
This proves the equality (10. 1.2), and hence the proposition.
Z{[Q»&3]53}lw0 +Z:{0¢3[C2,53]}l1:0
J J
0.

The proposition we have just proved has an interesting consequence. Given P 6

PDO ('")(E,F), kg 6 M , and fi 6 C°°(M) (z' = 1, . . . ,m), the linear map
l
adlf1)°l°adl.fmlP ii;
Efro -> F 0)

o
depends only on the quantities Q := dfi(fF0l 6 TfroM. Hence, for any ii 6 T3:0 M
(i = 1, . . . , m), the above expression unambiguously induces a linear map
07(Pll§1» ' ' ' , m l : E330 F307

that is symmetric in the variables §,. Using the polarization trick of Chapter 8, we see that
this map is uniquely determined by the polynomial
u( P )(€) ii 0m(P)(€) = c7(P)(§, . . .
'in

If we denote by 7T : T*M -> M the natural projection then, for each P E PDO l m l l E F l '
we have a well defined map
um (p)(.) Q Hom ('rr*E, 'rr*F),
where tr*E and 'rr*F denote the pullbacks of E and F to T*M via 7l`. Along the fibers
of T*M the map ,g »-> 0m(P)(§) looks like a degree m homogeneous "polynomial" with
coefficients in Hom (Easy, F390).
Proposition 10.1.10. Let P 6 PDO ( m ) ( E , F ) and Q Q PDO (") (F, G). Then
Um+n(Q o Pa UMC?) o 0m(P). I
456 Lectures on the Geometry of Manzfolds

Exercise 10.1.11. Prove the above proposition.

Definition 10.1.12. A p.d.o. P E PDO( m ) is said to have order m i f 0m(P) ¢ 0. In this

case 0,7,(P) is called the (principal) symbol of P. The set of p.d.o.'s of order m will be
denoted by PDO"" .

Definition 10.1.13. The operator P E PDO""(E»F) is said to be elliptic if, for any
; 1 3 € M , and any , § g T x m \ {0}, the map

0m(Pl(€) : Eas -> Fac

is a linear isomorphism.

The following exercises provide a complete explicit description of the p.d.o.'s on RN.

Exercise 10.1.14. Consider the scalar p.d.o. on RN described in Corollary 10.1.5

L Clolilflga

loll
Show that

<tm(Ll(€l 0»o(I)€°" = aalXlfl 1

. gn~
l0l=m l0l=m

Exercise 10.1.15. Let L c°°(1R(N) -> c°°(1R(N) be a p.d.o. of order m. Its principal
symbol has the form

0m(L)(€) @()€".
l0l=m

Show that L - Elal=m a0(a2)<90" is a p.d.0. of order < m - 1, and conclude that the only
scalar p.d.o.-s on RN are those indicated in Corollary 10.1.5.

Exercise 10.1.16. Let L 6 PDO (WE, F ) and u 6 cOO (E). Show the operator

c°°<m) 9 f I > [L, ,nu Q COOIF)

belongs to PDO lm'1l(Rm» F)-

Exercise 10.1.17. Let Kp and Kq denote the trivial K-vector bundles over RN of rank p
and respectively q. Show that any L 6 pDo("*) ( P , Kq) has the form

L I AO,(a3l3°' 7

If.lém

where AO( E coo (RN , Hom (KP, Kill for any a.

Hint: Use the previous exercise to reduce the problem to the case p 1.
 Elliptic Equations on Manifolds 457

10.1.2 Examples
At a first glance, the notions introduced SO far may look too difficult to "swallow". To help
the reader get a friendlier feeling towards them, we included in this subsection a couple
of classical examples which hopefully will ease this process. More specifically, we will
compute the principal symbols of some p.d.o.'s that we have been extensively using in this
book.
In the sequel, we will use the notation "f -" to denote the operation of multiplication by
the smooth scalar function f.

Example 10.1.18 (The Euclidean Laplacian). This is the second order p.d.o.,

A c°°(RN) -> c°°(RN), A=-Z m , 2 a a.

Let f E CooeRN). Then
2
ad(f)(A) ad(f) (be) Z -§j{ad(f)(<1) O Q-I-Q O ad(/'l(3¢l}

Z{fm@ °<9i +3¢(f<°)}

II
I

(Af) • -2 Zjf v .
Hence
&d(f)2(A) I ad(f)(Af) -2 2f ad(f) w ) 2Z(fx,)2. -21dfl2
1

If we set g := df in the above equality, we deduce

2
Ct2(A)(€> -l€1 I

In particular, this shows that the Laplacian A is an elliptic operator.

Example 10.1.19 (Covariant derivatives). Consider a vector bundle E -> M over the
smooth manifold M, and a connection V on E. We can view V as a p.d.o.
V : C°°(E) -> COO(T*M (X) E).
Its symbol can be read from ad(f)V, f 6 C°°(M). For any u 6 C'°°(E)
(3»d(f)V)v Wfu) f(Vu) = df (X) u.
Z

By setting 5 = df we deduce 071(V)(§) = € , i.e., the symbol is the tensor multiplication

by g.

Example 10.1.20 (The exterior derivative). Let M be a smooth manifold. The exterior
derivative
d : Q°<m) -> Q0-1-1(m)
is a first order p.d.o. To compute its symbol, consider CU E am), and f E C°°(M).
Then
(ad(f)d)(»J = dlfwl - fdw = if Aw.
If we set g df, we deduce 01 (d) I 6(€) = the left exterior multiplication by g.
458 Lectures on the Geometry of Manzfolds

Example 10.1.21. Consider an oriented, n-dimensional Riemann manifold ( M ) - As in

Chapter 4, we can produce an operator
6 *d*: Q. ( M ) - > 9°'1(M)»
where * is the Hodge *-operator
* QUO
(M) -> Q"'° (M).
The operator 6 is a first order p.d.o., and moreover
01 (6) : *01ld)* : *e(§) *
This description can be further simplified.
Fix 5 E Tas M, and denote by g* 6 TxM its metric dual. For simplicity we assume
l§l = l€*l = 1. Include 5 in an oriented orthonormal basis (g1, . . . ,§") of Tas M, £1 = §,
and denote by ii the dual basis of T x M .
Consider a monomial Ld = £] 6 A'*Tac M, where I = denotes as usual an
ordered multi-index. Note that if 1 Q I, then
U1(€1)<»0 0. (10.l.3)
I l l 6 I,e.g.,I= (1,...,k),then
)& 2 A
*@(€)(*wl Z *(€1 /\ §k+l /\ A <") = <- 1> (t» -f< =) (-1 Agk
= -(-1)Vnki(€*)w, (10.l.4)
where I/'n,k = no+n+ l is the exponent introduced in Subsection 4.1.5 while ill) denotes
the interior derivative along §*. Putting together (10. 1.3) and (IO.1.4) we deduce
01(5l(€) I -(-1)w~;(€*)_ (10.l.5)

Example 10.1.22 (The Hodge-DeRham operator). Let (M,g) be as in the above exam-
ple. The Hodge-DeRham operator is
d - I - d * : Q' ( M ) -> Q' ( m ) ,
where d* = (-1)Ur*"°c5. Hence d -|- d* is a first order p.d.o., and moreover,
07(d + d*)(€> 2 0(d)(€) + <t(d*)(€) i 6(,5) in).
The Hodge Laplacian is the operator (d -|- d*)2. We call it Laplacian since
al (d -I- d*)2)(§) = {(t(d+ d*)(§)}
2
(et - ¢(§*) )
2
7

while Exercise 2.2.55 shows

2 2
( e t - ¢(e*)) -(e(§)¢(€*) + ¢(§*)e(e)) I -l€l g
I

Notice that d + d* is elliptic since the square of its symbol is invertible.

Definition 10.1.23. Let E -> M be a smooth vector bundle over the Riemann manifold
( M ) - A second order p.d.o.
L : c°°<E> -> c°°<E>
is called a generalized Laplacian i f 02 (Lllgl 151
" O b s e r v e that all the generalized Laplac ians are elliptic operators.
 Elliptic Equations on Manifolds 459

10.1.3 Formal adjoints

For simplicity, all the vector bundles in this subsection will be assumed complex, unless
otherwise indicated.
Let El, E2 -> M be vector bundles over a smooth oriented manifold. Fix a Riemann
-
metric g on M and Hermitian metrics <>¢ on E , i = 1, 2. We denote by dvg - *l the
'Q

volume form on M defined by the metric g. Finally, c5° (E,-) denotes the space of smooth,
compactly supported sections o f E, .
Definition 10.1.24. Let P E PDO (El, EQ). The operator Q E PDO (EQ, El) is said to
be aformal adjoint of P if Vu E C'§°(E1), and Vu E
7 (;5°lE2) we have
I P , v)2dVg (Fu, Qu)1dVg.
M M

Lemma 10.1.25. Any P E PDO (El, EQ) admits at most one formal adjoint.
Proof. Let 621, QUO be two formal adjoints of P. Then, Vu E C5° (EQ), we have

Lu, (621 - Q2lu)1dvg = 0 Vu E CO lEll.

M
This implies (Q1 - Q2)v = 0 Vu E Cf-§°(E2). If now U E COO(E2) is not necessarily
compactly supported then, choosing a partition of unity (Q) C Cgo(M), we conclude
using the locality of Q = QUO - QUO that
Qu Q(avl = 0.
The formal adjoint of a p.d.o. P 6 PDO lEt) EQ), whose existence is not yet guaran-
teed, is denoted by P*. It is worth emphasizing that P* depends on the choices of g and
(~,

Proposition 10.1.26. (a) Let L0 E PDO (Et, El) and L1 E PDO (El, E2) admitfor-
mal adjoint L11 E PDO (E'i-I-17 E t ) ( i = 0 , 1), with respect to a metric g on the base, and
metrics (», -la on Et, j = 0 , 1 , 2. Then L1L0 admits a formal adjoint, and

in 1fL Q PDO (WED, El), then L* Q PDO (WE1, EQ).

Proof. (a) For any Up 6 c5°(Et) we have

(L1L0'LL0, U2)2dvg (LQUo, L1u0)1dVg ( U Q , L0L1'LL2)0d'/9 .
M M M
(b) Let f Q C°°(M). Then
lad(fl/;l* I (L o f - f O
L)* I -[Li fi I - ad(f)L*.
Thus
adlf0l adlfll . a . ad(ji,,,)L* I (-1)"'+1(ad(f0)ad(f1) ad(fm)Ll* 0.
The above computation yields the following result.

Corollary 10.1.27. IfP E PDO ( m ) admits a formal adjoint, then

o,,,(P*) I (-1)""0m(Pl*»
where the * in the right-hand-side denotes the conjugate transpose of a linear map.
460 Lectures on the Geometry of Manzfolds

Let E be a Hermitian vector bundle over the oriented Riemann manifold ( M g ) -

Definition 10.1.28. A p.d.o. L E PDO ( M ) is said to be formally self-adjoint if

L L*

The above notion depends clearly on the various metrics

Example 10.1.29. Using the integration by parts formula of Subsection 4.1.5, we deduce
that the Hodge-DeRham operator
d+ d* Q'
( M ) - > Q' (M),
on an oriented Riemann manifold (M, g) is formally self-adjoint with respect with the
metrics induced by g in the various intervening bundles. In fact, d* is the formal adjoint
of d.

Proposition 10.1.30. Let ( E , (~, .>,), ii = 1, 2, be two arbitrary Hermitian vector bundle
over the oriented Riemann manifold ( M , go. Then any L E PDO (El, EQ) admits at least
(and hence exactly) one formal adjoint L*.

Sketch of proof. We prove this result only in the case when E1 and E2 are trivial vector
bundles over RN . However, we do not assume that the Riemann metric over RN is the
Euclidean one. The general case can be reduced to this via partitions of unity and we leave
the reader to check it for him/her-self.
Let E1 = Qp and E2 = Qq. By choosing orthonormal moving frames, we can assume
that the metrics on Et are the Euclidean ones. According to the exercises at the end of
Subsection 10.1.1, any L E PDO(m) (El, ET) has the form
L AO,(a3)3°' 7

Ialém
where Aa 6 0°°(RN,Mqxp((C)). Clearly, the formal adjoint of A is the conjugate
transpose
A; EQ.
To prove the proposition it suffices to show that each of the operators <2 6 PDO1 lEt» El)
admits a formal adjoint. It is convenient to consider the slightly more general situation.

Lemma 10.1.31. Let X we 6 Vect (RN ), and denote by V i the first order p.d.o.
V x U = X"<9iu u E CO (gel.
Then
V i = -Vi - d i v g ( X ) ,
where divg ( X I denotes the divergence of X with respect to the metric g, i.e., the scalar
defined by the equality
Lx(dVg) = d i v g l x ) dog.
 Elliptic Equations on Manifolds 461

Proof. Let 'u,,v Q Co°(RN, (gel. Choose R >> 0 such that the Euclidean ball BR of radius
R centered at the origin contains the supports of both 'LL and 'U. From the equality
X (u/u) = ( V x u , v ) -|- (u, VxU),

we deduce

Rn
(VxU, 'u)dvg
/ BR
(VxU, 'u)dvg

/ -
X (u,'u)dVg (u,Vxv)dVg. (10.l.6)
BR BR

Set f == ( u ) E c5'<>(1Re.N,(c>, and denote by a E Q1(RN) the 1-form dual to X with

respect to the Riemann metric g, i.e.,

( g = @(x) \-/B e QRRN ).

Equivalently,

a = gi]xid£1z'j.
The equality (10. 1.6) can be rewritten

(Vxu, v)dVg of(x)dvg (u, Vxv)dVg

BR BR RN

L (df, a)gdVg

The integration by parts formula of Subsection 4.1.5 yields

RN
(Fu, Vx'v)dVg .

(of, a)gdVg (if /\ *gal l3BR -I- fdmdvg.

BR aBR BR
Since f E 0 on a neighborhood of 8BR, we get

(df, algdvg (u, 'u)d*a dvg.

BR BR

Since d*a = - d i v g ( X ) , (see Subsection 4.1.5) we deduce

X - lu,vldVg (U, u) divs (X)dvg (Fu, -dzvg(x)u)dvg.

BR BR BR
Putting together all of the above we get

(VxU, 'u)dvg (u, ( - e x dwg(x) )@)dv g ,

Rn Rn

i.e.,

VX = -Vi - divg(X) -Vx

v 1191 2;
I (10.1.7)
i

The lemma and consequently the proposition is proved.

462 Lectures on the Geometry of Manzfolds

Example 10.1.32. Let E be a rank T' smooth vector bundle over the oriented Riemann
manifold (M, g), d i M = m. Let ( -,- ) denote a Hermitian metric on E, and consider
a connection V on E compatible with this metric. The connection V defines a first order
p.d.o.
V : C°°(E) -> COO(T*M (X) E).
The metrics g and ( ) induce a metric on T*M (X) E. We want to describe the formal
adjoint of V with respect to these choices of metrics.
As we have mentioned in the proof of the previous proposition, this is an entirely local
issue. We fix 5130 6 M, and we denote by (acl, . . . , cum) a collection org-normal coordinates
on a neighborhood U of xo. We set
Vi .- V6 .
Next, we pick a local synchronous frame of E near 5130, i.e., a local orthonormal frame (60)
such that, at 5130, we have
Vi€0£ 0 W=1,...,m.
The adjoint of the operator
do:k® : C'°°(EIU) -> COO(T*U (X) EIU),
is the interior derivative (contraction) along the vector field g-dual to the 1-form disk , i.e.,
(dm'"®)* = Ck := gawk - i n ) .
Since V is a metric connection, we deduce as in the proof of Lemma 10.1.31 that
Vi* = - V g - diVglgmk).
Hence,
V*
ZW ock Z( - V i - divs lgmkl) o Co
k k

DW + @(1<>g(» l9l)l c O (10.l.8)

II
I

k
In particular, since at $0 we have 3mk g = 0, we get
\-7*l$0= _ Z W ck O

k
The covariant Laplacian is the second order p.d.o.
A = As : C°°(E) -> C°°(E), A = V*V.
To justify the attribute Laplacian we will show that A is indeed a generalized Laplacian.
Using (10.1.8) we deduce that over U (chosen as above) we have

A {2 (v + a
k
log v Riga) ock
H o
j
dzcj (X) Vi
}
{2 k
Vi -|- 3m1. log m) O
l9k"~Val

U
lm
9 k y V k -|- Qnk 9ky -|- g k j a m k o Vi

gkjvkvj -I- / a w e ) a
Vi
kJ
 Elliptic Equations on Manifolds 463

The symbol of A can be read easily from the last equality. More precisely,

Ct2(A)(€) I -gjk§yfk I -151g2'

Hence A is indeed a generalized aplacian.

In the following exercise we use the notations in the previous example.

Exercise 10.1.33. (a) Show that

9k£ 'II
Fee Mg ),

where Pm denote the Christoffel symbols of the Levi-Civita connection associated to the
metric g.
(b) Show that

As
to; WT* M®EvEl
where VT*M®E is the connection on T*M (X) E obtained by tensoring the Levi-Civita
connection on T*M and the connection VE on E, while
tr; : c°°(T*m®2 (X) E) -> C°°(E)
denotes the double contraction by g,

tfglszj up = (X) gjsiju.

Proposition 10.1.34. Let E and M as above, and suppose that L E PDO2(E) is a

generalized Lap lacian. Then, there exists a unique metric connection V on E and 52
IR(/2) 6 End ( E ) such that

L I v*v + 32.
The endomorphism 32 is known as the Weitzenbéck remainder of the Laplacian L.

Exercise 10.1.35. Prove the above proposition.

Hint. Try V defined by
f (ad(hlLlu} f , h 6 c°°(m>, u c°°(E).
Vfgrad(h)u = -{(Ag/Uu
2
Q

Exercise 10.1.36 (General Green's formula). Consider a compact Riemannian manifold

(M, go with boundary coM. Denote by 18 the unit outer normal along <9M. Let E, F -> M
be Hermitian vector bundles over M and suppose L E PDO k ( E , F). Set 90 = g Iam,
E0 = E lam and F0 = F lam. The Green formula states that there exists a sesquilinear
map
BLICOOIEI X C°°(F)->C°°(c9M),
such that

(Lu/u)d'u(9) 8MBL("~'»='UldU(90)+ (u»L*v)dv(9)~

M M
464 Lectures on the Geometry of Manzfolds

Prove the following.

(a) If L is a zeroth order operator, then BL 0. I

(b) If L1 E PDO (F, G), and L2 E PDO ( E , F ) , then

BL1L2 (u,vl = BL1(L2u, Fu) -|- BL2 (u, LIE).

(c)

BE*(v,u) = -BE(u,v).

(d) Suppose V is a Hermitian connection on E, and X E Vect ( M ) . Then

B o X ( u , v ) = (u,v)g(X,15,'), Be(u,u) = (u,in~u)E,

where in denotes the contraction by 18.

(e) Denote by 17 the section of T*M I am g-dual to ii. Suppose L is a first order p.d.o., and
set J := oL(17). Then
BL(7u,,'u} = (.]u,v)p. (10.l.9)

(D Using (a)-(e) show that, for all u E COO(E), 'u E COO(F), and any £1:0 E AM, the
quantity BL(U, v)(:c0) depends only on the jets of u, 'U at £130 of order at most k - 1. In
other words, if all the partial derivatives of u up to order (k - 1), with respect to some
connection, vanish along the boundary, then BL (u, u) = 0.

10.2 Functional framework

The partial differential operators are linear operators in infinite dimensional spaces, and this
feature requires special care in dealing with them. Linear algebra alone is not sufficient.
This is where functional analysis comes in.
In this section we introduce a whole range of functional spaces which are extremely
useful for most geometric applications. The presentation assumes the reader is familiar with
some basic principles of functional analysis. As a reference for these facts we recommend
the excellent monograph [23], or the very comprehensive [40, 137].

10.2.1 Sobolev spaces in RN

Let D denote an open subset of RN . We denote by LgOClpl the space of locally integrable
real functions on D, i.e., Lebesgue measurable functions f : RN -> R such that, Va E
05° (RN), the function of is Lebesgue integrable, of E L1 (RN).

Definition 10.2.1. Let f E L M D L and 1 < lc < N. A function g E LgOClpl is said to

be the weak k-th partial derivative of f , and we write this g = (in, f weakly, if

gcpdzv = féikcpdac, We 6 C0 (D).

'D 'D

Lemma 10.2.2. Any f 6 Ll1OC(Dl admits at most one weak partial derivative.
 Elliptic Equations on Manifolds 465

The proof of this lemma is left to the reader.

" N o t all locally integrable functions admit weak derivatives.

Exercise 10.2.3. Let f E C°°(D), and 1 < lc < N. Prove that the classical partial
derivative as, f is also its weak k-th derivative.

Exercise 10.2.4. Let H 6 LgOClkl denote the Heaviside function, H(t) 1, for t > 0,
H(t) E 0 for t < 0. Prove that H is not weakly differentiable.

Exercise 10.2.5. 6 -
_|. 92 weakly. L e t 1f , f2 e L 1loel D ). I f kfz
.
= 92 E LlOC
1 weakly, then 8k(f1 -|- f2l
91

The definition of weak derivative can be generalized to higher order derivatives as fol-
lows. Consider a scalar p.d.o. L c°°(D) -> c°°(D), and f , g E LgOClp). Then we say
that Lf = g weakly i f

990 d o ; aL*wdo: W
Q
COoo( D).
) in

Above, L* denotes the formal adjoint of L with respect to the Euclidean metric on RN

Exercise 10.2.6. Let f , 9 6 C°°(D)~ Prove that

Lf = g classically ~I ' Lf = g weakly.

Definition 10.2.7. Let k E Z+, and p E l1, oo]. The Sobolev space L f w l p l consists of all
the functions f E Lplpl such that, for any multi-index a satisfying lal < k, the mixed
partial derivative c9"'f exists weakly, and moreover, c9°"f E Lp(D), For every f E I;I<=»v(D>
we set
1/19

i f p < o o , while i f p
llfllk,p

OO
llfll/<,p,D
(f I0yISI¢ D
l30'flpd:u
) 7

llfllk,oo Z esssupl3o'fl.
l<.ISk
When k = 0 we write llfllp instead of llfll01p°

Definition 10.2.8. Let lc E 2+ ands E [l, oo]. Set

Li;'"(D) == { f Q L E D ) ; (of Q L'"»p(D) V i Q c5°(D)

Exercise 10.2.9. Let f @ = V ¢ @ R @ > 0. Show that for every p > 1 such that
1
a > l p a we have

f(t) E L}oP(Rl,
466 Lectures on the Geometry of Manzfolds

Theorem 10.2.10. Let k 6 2 + and 1 <_ p < oo. Then

(a) (Lk'P(RN), II ilkp) is a Banach space.
(b) Ill < p < oo the subspace 03° ( R N ) is dense in LW (RNA.
(c) If 1 < p < OO the Sobolev space L'»p(RN ) is reflexive.
The proof of this theorem relies on a collection of basic techniques frequently used in
the study of partial differential equations. This is why we choose to cover the proof of this
theorem in some detail. We will consider only the case p < oo, leaving the p = oo situation
to the reader.

Proof. Using the Exercise 10.2.5, we deduce that LA*'P is a vector space. From the classical
Minkowski inequality,
l/p l/p V l/p

(i'II 1
l33tz + ml"
> - (I iI > (2 >
<
1l=1
SEe' P -I-
'i=l
'Yilp 7

-
we deduce that I II km is a norm. To prove that L1<=,p is complete, we will use the well
established fact that Lp is complete.
@1>To simplycy the notations, throughout this chapter all the extracted subsequences will
be denoted by the same symbols as the sequences they originate from.
Let ( f n ) C L'<»p(1122.N) be a Cauchy sequence, i.e.,
lim
mm,->oo
Ilium fnllk,P 0.

In particular, for each multi-index I0 l < k , the sequence (<9°"fn) is Cauchy in LPURN),
and thus,

a a r 5 go, vl@l < k .

Set f = limn fn- We claim that f go weakly. aa
Indeed, for any (f)€C6'°(RN) we have

3°'fncpd;z3 I (-1)l0l ]n3a90~

RN RN

Since 3a*f,,->go,fn->finLP,and(,o€Lq(RN), where l / q = 1 -1/p, we conclude

go ' gods lim 3"'fn1 • coda:

Rn TL Rn

lim(-1) III fn 3°'gQd;u = ( - l ) l 0 ' l f - 3o'god;u.

RN RN

Part (a) is proved.

To prove that 05° (RN ) is dense we will use mollifies. Their definition uses the oper-
ation of convolution. Given f , g 6 L1(]RN), we define

f(33 - y)9(y)dy-
 Elliptic Equations on Manifolds 467

We let the reader check that f * g is well defined, i.e., y »-> f(fv - y)9(y) Q L1, for almost
all a:.

Exercise 10.2.11 (Young's Inequality)- I f f Q LP(RN), and g E LC1(RN), 1 SUM! S OO,

then f * g is well defined, f * 9 Q U`(RN), _1 _1 l
or - P -|- q
1, and

llf *gH'r' S llfllp° ll9llq~ (10.2.1)

To define the mollifies one usually Stans with a function p 6 05° (RN ), such that

0 > 0, supp 0 C {Ill < l }, and /RN

pdac 1.

Next, for each 6 > 0, we define

p6(12) 6'N 0(;v/6)~

Note that

SUPP $6 C { I:z:l<5 } and

RN
pa d:c 1.

The sequence ($6) is called a molding sequence. The next result describes the main use
of this construction.

Lemma 10.2.12. (a) For any f E L}OC(RN I the convolution P6 * f is a smooth' function!
_
l b l l f f E LPQRN) ( 1 < p < OO) then

06 *f f asé->0.

Proof. Part (a) is left to the reader as an exercise in the differentiability of integrals with
parameters.
To establish part (b), we will use the fact that 08° (RN ) is dense in LPIRNI. Fix 5 > 0,
and choose g E C8° (RN ) such that,

Ill - £1llp < 6/3.

We have

H 06 * f - flip S 1106 * (f - glllp -|- 1106 * 9 - gllp + 119 - fllp~

Using the inequality (10.2.1) we deduce

llp6 * f - flip S 21lf - 9llp -|- 1106 * 9 - 9llp° (10.2.2)

We need to estimate llp6 * 9 - 9||p- Note that

P6 * 9 (11) 0(2)9(H1 - Hz) do and 9(H1) »o(2)9(1v)d2~

Rn Rn

This explains the term mollu'ier: pa smoothes out the asperities.

468 Lectures on the Geometry of Manzfolds

Hence

106 * 9(23)-9(x1)l < RN

0(2)l9(=v - 5zl - 9(11)ld2 S 5 sup ld9l~
Since
g- z 6
PA * g
Q
{ac -I-
Z ; ac Q S1lpp suPp 0 6 } 7

7
there exists a compact set K C R N such that,
K VI
p
suPp( 6 * 9 - g) Q
Q (09 l ).
We conclude that
su llpd 1/p

He * 9 - M S
If now we pick 6 such that
It K
5pl P l¢zg
")
vol lKll/p6 sup ldgl.

vo1(K11/1°6supldgl < 6/3,

we conclude from (10.2.2) that
lIPs * f flip $ 8 .
The lemma is proved.

The next auxiliary result describes another useful feature of the mollification technique,
especially versatile in as far as the study of partial differential equations is concerned.

Lemma 10.2.13. Let f , g E Ll1OC (RN) such that fief g weakly. Then 31 (pg * f ) ,06*9-
More generally, If
L ao 30'
loll
is a p.d.o. with constant coejicients aa € R, and L f = g weakly, then
L(p6 * f) = 06 *g classically.

Proof. It suffices to prove only the second part. We will write La, to emphasize that L acts
via derivatives with respect to the variables .CU = (ac1, . . . 7 LuN . Note that

L(,05 *f) (L=»06(33 - yllf(yldy- (10.2.3)

Rn

Since
a
flw;3-M <9yi
,o5(a3-y),

and <2 * = -ii, we deduce from (l0.2.3) that

Llpa * f)
/Rn
(Ly06(¢U - y)lf(y)dy
Rn
p6(22 - y)Lyf(y) do

/RN
wW-yM@My=m *
 Elliptic Equations on Manifolds 469

Remark 10.2.14. The above lemma is a commutativity result. It shows that if L is a p.d.o.
with constant coefficients, then

[L»06*]f=L(05 * f ) - p a * (Lf) 0.
This fact has a fundamental importance in establishing regularity results for elliptic opera-
tors.

After this rather long detour, we return to the proof of Theorem 10.2.10. Let f E
L'*'P(RN) . We will construct fn Q C5°(RN) such that fn -> f in Limp using two basic
techniques: truncation and mollu'ication (or smoothing).
Truncation. The essentials of this technique are contained in the following result.

Lemma 10.2.15. Let f 6 L'<=»p(RN). Consider for each R > 0 a smooth function UR 6
C'§°(RN) such that 17(;z:) E lfor la:l < R, T}R(ac) E Ofor lac°l > R + 1, and ldtIR(a))l < 2
Vi. Then UR - f E Lk»P(RN), 'v'R > 0, and moreover
L'°p
UR • f >f as R -> oo.

Proof. We consider only the case k = 1. The general situation can be proved by induction.
We first prove that <2 (nerf) exists weakly and as expected
3i(?7R u fl = l5H7Rl f+77R 31f-' •

Let 9o 6 Cg°(RNl. Since '7R(P E 0g°(RN), we have

l<9t*7RlS0f + UR • <9i(Pf do: Rn <'9¢(t7R(0lf = '7R(P<9v;fdiU-

Rn Rn

This confirms our claim. Clearly l<9H7R)f+ l7R5tzf 6 LPQRNI, so that 17Rf is in L1»p(1RN).
Note that 31;17R E 0 for la:l < R and l:/:l > R + l. In particular, we deduce
(32-nglf -> 0 a.e.
Clearly l(3w7R)fl < 2lf(a2)l, so that by the dominated convergence theorem we conclude

(31i77Rl ' f 0 as R -> oo.

Similarly

77R<9¢f -> of a.e.,

and l77R31fl S l81fl, which implies '7R3if -> 5%f in LP. The lemma is proved.

According to the above lemma, the space of compactly supported Le-functions is

dense in L":=1'(RN). Hence, it suffices to show that any such function can be arbitrarily
well approximated in the Ll*'p-norm by smooth, compactly supported functions.
Let f E Lk<=»p(RN), and assume
ess suppl C {tact < R}.
Molli/cation. The sequence 06 * f converges to f in the norm of Ll*'P as 6 \ 0.
470 Lectures on the Geometry of Manzfolds

Note that each PA * f is a smooth function, supported in {lUI S R -|- 6 }. According to

Lemma 10.2.13, we have

3°"(06 * f) =PA * (<9""f) VIQI S k.

The desired conclusion now follows using Lemma 10.2.12.
To conclude the proof of Theorem 10.2.10, we need to show that LW is reflexive if
l < p < oo. We will use the fact that Lp is reflexive for p in this range.
Note first that L P (RN ) can be viewed as a closed subspace of the direct product

H LP(RN),
l<.ISk
via the map,

T : L/<<=,2'(RN) - H LpURN), f *
I0llék
(<9°"f)lalgI<,

which is continuous, one-to-one, and has closed range. Indeed, if a a r 5 Ada then,
arguing as in the proof of completeness, we deduce that

to = asf0 weakly,

where to = lim fn. Hence (fa) = T f 0 . We now conclude that LA<»p(RN ) is reflexive as a
closed subspace of a reflexive space. Theorem 10.2.10 is proved.

Remark 10.2.16. (a) For p = 2 the spaces Lk,2lRN ) are in fact Hilbert spaces. The inner
product is given by

(u, U>k
Rn
( lal$1<
3°'u-c90"vd:c
) day.

(b) If D C RN is open, then L/<»p(D) is a Banach space, reflexive if l < p < oo. However
C'§°(Dl, is no longer dense in Le»I>(D). The closure of cgOlpl in Le>P(D) is denoted by
L'5""(D). Intuitively, L'5"'(D) consists of the functions u 6 Lk'P(D) such that
31u
0 on <9D, V j 0,1, 7 k 1,
31/3
where 3 / d v denotes the normal derivative along the boundary. The above statement should
be taken with a grain of salt since at this point it is not clear how one can define u 169 when
u is defined only almost everywhere. We refer to [4] for a way around this issue.
The larger space COO(D) o LA<»p(Dl is dense in LA<»/>(D), provided that the boundary
of D is sufficiently regular. We refer again to [4] for details.

Exercise 10.2.17. Prove that the following statements are equivalent.

(a) 'LL Q LMJIRN).
 Elliptic Equations on Manifolds 471

(b) There exists a constant C > 0 such that, for all (0€C°°(RN)7 we have

RN
up <cll¢ll
3x2 - L
w = 1
7

where p' = p/(p - 1).

(c) There exists C > 0 such that, for all h , we have
Q RN
<
Lp
-
CHM,
where Ahu
h) " U
W).
Exercise 10.2.18. Let f E L1»p(RN), and Q) E c<><>(R), such that
_ const,
up<l)l <
and (z5(f) E LP. Then cb(/'l E L1»p(RN), and

f) i (b'(f) . of.
Exercise 10.2.19. Let f E L1»p(RN). Show that lfl E L1»p(RN), and
3¢f a.e. on { f > 0}
61lfI -@¢f a.e. on {f < 0} •

Hint: Show that fe := (62 + f2)1/2 converges to f in L P as 5 -> 0.

10.2.2 Embedding theorems: integrability properties

The embedding theorems describe various inclusions between the Sobolev spaces
Lk'P(RN).
Define the "strength" of the Sobolev space L""P(]RN I as the quantity

0(I<»p) 07n(k,p)=k-N/p.
The "strength" is a measure of the size of a Sobolev size. Loosely speaking, the bigger the
strength, the more regular are the functions in that space, and thus it consists of "fewer"
functions.

Remark 10.2.20. The origin of the quantities UnUQPI can be explained by using the no-
tion of conformal weight. A function u : RN -> R can be thought of as a dimensionless
physical quantity. Its conformal weight is 0. Its partial derivatives 3 11'LL are physical quanti-
ties measured in meter'l= variation per unit of distance, and they have conformal weight
-1. More generally, a mixed partial ( u has conformal weight -101. The quantities lél0" 119
have conformal weight -pl0l. The volume form do is assigned conformal weight N: the
volume is measured in mete1°n . The integral of a quantity of conformal weight w is a
quantity of conformal weight w -|- N . For example, the quantity

l3°"u,lp do
Rn
472 Lectures on the Geometry of Manzfolds

has conformal weight N pla . In particular the quantity

U{ l01l=k
l3°"'u,IP do
}
1/p

has conformal weight (N - up)/p = -0N(k, p). Geometrically, the conformal weight is
captured by the behavior under the rescalings ac = Ay.
If we replace the Euclidean metric g on R" with the metric SlA = A2g, A > 0, then,

dog = wdvg, ndulg = A l d u l g , Vu Q c°°(RN).

Equivalently, if we introduce the new linear variables ii related to the canonical Euclidean
variables or by m = A8/2 then

33 = A-1a, dzzz1 day Andyl dyn

Theorem 10.2.21 (Sobolev). if

0N(k,p) = 0N(m,q) < 0 and k > m

then,

L'"»p(1122.N) <-> Lm»'1(1R(N),

and the natural inclusion is continuous, i.e., there exists C C(N,k,m,p,q) > 0 such
that

Hfum,q S c1lfll,p of 6 L'°»p(RN).

Proof. We follow the approach of [107] which relies on the following elementary, but
ingenious lemma.

Lemma 10.2.22 (Gagliardo-Nirenberg). L e t N > 2 and f1,...,fN G LN-1(1Re.N-1).

For each at E RN, and l < i < N d e n e
a l i
:c,...,;I: 7 7 ;r:N)€]RN-1

Then

f(=vl = f1(§1)f2(€2l ' •' fn(§ n) E L1(RN),

and moreover;
N

1lfII 1 H 1lf¢1ln-1~
ii 1

Exercise 10.2.23. Prove Lemma 10.2.22.

 Elliptic Equations on Manifolds 473

We first prove the theorem in the case k = 1, p = 1, which means m 0, and

q = N / ( N - 1). We will show that

30 > 0 : lIUIln/(n-1) cu noun HE Vu Q c5°(RN),

where ldul2 = l31ul2 -|- - -|- l3Nul2. Thi s result then extends by density to any 'LL E
L1,1lRNl_

We have
1
ac

Iu(a31 7 .,acN)l < l3u(;131 7 .,ac°

i-1
,t,a3 ,..
i-I-1
7 ;z:N)ldt
OO
1
ac
< | 3 u(a31 7 .,ac°
i-1
,t,a3 fi-I-1 ,.. 7 a:N)ldt
a
•
9¢(€iZ-
OO

Note that go 6 L1(RN-ll, so that

fig(§il = 91501/(N-1) 6 LN-1(]RN-1

Since

l ~ ( w ) l " / ( ) < f1(€1l a u fn(en),

we conclude from Lemma 10.2.22 that u(:z:) E LN/(N-1)lRNl,
and
N 1/n N 1/n
IIUIIn/(n-1) <
(Hl
llglg)lll

Using the classical arithmetic-geometric means inequality, we conclude

) (H )1
ll3¢'0ll1

N N
1
IIulIn/(n-1) <
_ Z ll<9i'U~ll1 <_ const.
N
1
Illdullll.
We have thus proved that L1,1 (RN ) embeds continuously in Ln/(n-1) IRNI.
Now let l < p < oo such that 07N(l,p) = l - N/p < 0, i.e.,p < N. We have to show
that L1"'(RN) embeds continuously in LP* (RN), where
* No
P = -
N -p
Let u 6 CEO)(RNl. Set 'u := lul7°-1u, where P > 1 will be specified later. The inequality
N 1/n

implies
ll'ulln/(n-1> S
(H I 111)
1
at?)

N 1/n
ll'U»lln/(n-1) < 7° H II lulT
1
1Tzu 111

If q p / ( p - l l is the conjugate exponent of p, then using the H6lder inequality we get

II IUI-1a?U < Ullq(»,~1 1)lla'U' IIp '

474 Lectures on the Geometry of Manzfolds

Consequently,
N 1/n
- .
lIUIln/(N-
1)
<
_
r-1
rH,U'Hq(fr_1) H 1»g|| an II
1
p

Now choose T' such that i N / ( N - 1 ) = q(?" - 1). This gives

1
T' 7
N
and we get
N l/n
llullp* <
- T'
(H II II )
l
3z"LL
p
S 0'(n,p)ll ldul llp-
This shows LLP <-> Lp* if 1 < p < N. The general case
L'"»p <-> L'"'q if 0N(k,p) = 0N(m,q) < 0 k > m
follows easily by induction over k. We let the reader fill in the details.

Theorem 10.2.24 (Rellich-Kondrachov). Let (k,pl, (m, q) E Z+ X [l, oo) such that
k > m and 0 > can(k,p) > 0N(m,q).

Then any bounded sequence (url C L'<»p(1R(N) supported in a ball BR(0l, R > 0 has a
subsequence strongly convergent in LM»q(1R(N I.
Proof. We discuss only the case k = 1, so that the condition 0N(1, p) < 0 imposes
l < p < N. We follow the elegant presentation in [23, Chap. IX]. The proof will be
carried out in several steps.
Step 1. For every h 6 RN we define the translation operator

Th, : LQIRNI -> Lq(R"), Thf(1u) = f(w + h), of Q Lq(1Re.N), :c Q RN.

Fix Q G [1»p*)» P* = N p / ( N - p) and aradius R > 0. Let a E (0, l] be defined by
l l - a
=a+ . (10.2.4)
<1 p*
Then there exists a positive constant C, depending only on R, PA quo N such that

v l h l g l , vu@L~°(R")= 11w-ul.q,BR < clhl"llullL Rn>. (10.2.5)

Indeed, if u is a smooth function then

1

emf) - u(w)l § lhl ld.(;z: -|- thlld¢.

0

Hence

lITh"J» - 'U»llL1(BR) | Thu(;z:) - u(;c) | do:

BR
 Elliptic Equations on Manifolds 475

1
< lhl
lfO
1
BR
ld.(;u -|- th)|da3

1/p
dt

I Ill l f th-I-BR
ldu(y)l'"dy vol(BR+1)1-1/P S Clhlllull1,p,R~,
0
where C = C ( R , p, q, N). Finally, using (10.2.4) and H61der's inequality we deduce

lIThe UIILql8R) S lITh'LI UIIL1(BR) lIThe UJpa(8R) < C|h|a||U||1,p,Rn-

This proves the inequality (l0.2.5) for smooth functions. The general case is obtained by
passing to the limit in the L1°*'-norm.
Step 2. Let 'Ll C L1»P(RN ) be a bounded subset a such that
esssuppu C {la:l < R} Vu E 'LL
We will prove that, for every e > 0 there exists 0 < 6 = 5 (5) < 1, such that
llp5 * u - ullaR S 8, VI g 5(6), Vu E U. (10.2.6)

Let u E U. We have

105 * 'u,(:v) u(:v)l < B6 1f-yu(w/ u(22)|p6(y)dy-

Using the convexity of the function S »-> sq, S > 0 and the fact that P6 (y)dy is a probability
measure we deduce
q

B IT-yu(wl - u(t)l¢>6(:u)dy < IT-yu(wl - u(w)l"pa(y)dy

B6
we deduce

106 * u(a:) u(@)lq <

_
v86
LL(a2) - u(t)l'106(y)dy,

so that
1/q (10.2.5) a
ll»O6 * u n - U11llq,BR <
(f BE
HT-y'UH Un la,BR p6(3/ldy < CIIUfll1,p,R"5 •

We now choose 6(6) such that

C6(e)0" sup llulll p R o < 6.
1u6'L£

Step 3. Conclusion. Suppose 'Ll C L1>2'(R") is a bounded of functions with supports

contained in BR. We need to prove that for any e > 0 we can find a finite cover of 'Ll with
Le(BRl-balls of radius 5.
To prove this, we set

U<S={'LL5 Z=p5*u, usU},

and we show that UP is precompact in C0 (BR+1)~ To achieve this we invoke the Arzela-
Ascoli theorem and we will show that for any 6 > 0 there exists C = C (5) > 0, such
that
l'u,5(a3)l <C' 'v'u€U, 'v'la3l < R +I,
476 Lectures on the Geometry of Manzfolds

lu6(x1) - U6l$2)l S Cl£C1 - 3321 Vu E "Ll, \V/l£L'1l7 l$E2l < R + 1 .

Indeed,
p ac y
106 * u (we < 6 N
ly-H:IS6 ( ) 6
lu(y)ldy < 6 N
BE (up)
nu(y)ldy

S C(N,p)5-NIlullp* - vol (85l(p*-1)/p*

S C((5)llull1,p by the Sobolev embedding theorem).

Similarly,

lu6(H11) u6(fv2)l < l,06l£U1 y) PA l$U2 y)l . lu(y)ldy

8R-1

<
- C'(6) a 15131 5821 lu(y)ldy S 0(6) a 1581 5821 . qUIll,p°
8R-l-1

Now let 6 < 5(e/2), where 5(€/2) is defined by (l0.2.6). Then the family UP is relatively
compact in 00(8T+1) and thus it can be covered by finitely many Lq(BR+1) balls of
radius 6/2. Using (10.2.6) we see that we can cover 'Ll with finitely many Lq(BR+1)-balls
of radius 5. The compactness theorem is proved.

10.2.3 Embedding theorems: differentiability properties

A priori, the functions in the Sobolev spaces L/<=,p are only measurable, and are defined
only almost everywhere. However, if the strength 0 N ( k , p) is sufficiently large, then the
functions of Lk<=»p have a built-in regularity: each can be modified on a negligible set to
become continuous, and even differentiable.
To formulate our next results we must introduce another important family of Banach
spaces, namely the spaces of HOlder continuous functions.
Let a E (0, l). A function u : D C RN -> R is said to be a-HOlder continuous i f
Ma sup R'°'osc(u; BR(Z) VW D) < oo,
0<R<1,z€D
where for any set S C D, we denoted by OSC (ug S) the oscillation of u on S, i.e.,
OSC Lu; S) sup{lu(1v) u(y)l 7 Xy y 6 S}-
Set
fluff oo,D o
• sup lu(23)l 7
QSGD
and define
C0 1(0) := {u : D -> R ; ll7Ufll0,o¢,D := llull@O + [u]O, < oo
More generally, for every integer k > 0, define
ck,0l D ) : z { u E C D ) ; E COD), as am g 1<}.
The space C""'°' (D) is a Banach space with respect to the norm,
fluff/<=,o a
•
ll@%nnoo,D + [6'BU]oD.
I/3ISk l0l=k
Define the strength of the H6lder space c k , a as the quantity
0(k, a ) := k -|- a.
 Elliptic Equations on Manifolds 477

Theorem 10.2.25 (Morrey). Consider (m,pl 6 2+ X [Loo] and (ha) 6 2+ X (0, 1 )

such that m > k and 07N(m, p) = 07(k, Q) > 0. Then LM»P(RN) embeds continuously in
co,alRN)_
Proof. We consider only the case k l, and (necessarily) m 0. The proof relies on the
following elementary observation.

Lemma 10.2.26. Lets 6 C°°(BR) O L1,1lBRl and set

l
u : u(;13)d;13.
vol IBRI BR

Then
N
2 ld.(y)l do
up) UI _< (10.2.7)
On 1 BR W-MN
In the above inequality On-1 denotes the "area" of the unit ( N - l)-dimensional round
sphere S N - l C RN.

Proof.
l22-yl 6
u(a2) - u(y) -u(a2 -|- rw)d1°
0 37" II-:UU
Integrating the above equality with respect to y we get,
l92-yl a
vol (8R)(u(=v) - U)
/ / BR
do
0
-u(:z: -|- rw)d1°.
37"
If we set lc'i,°u(a: + till 0 for law + 7°oJI > R, then
OO

vol (BR)lu(a2) - 51 < do l31~u(;c + row)ldr.

lx-YI§2R

If we use polar coordinates (02 w) centered at co, then do = pn-1dpdw, where p 1312 y 7

and do denotes the Euclidean "area" form on the unit round sphere. We deduce
oo ~2R
vol(B llul33l UI dr l31~u(;I: -|- 1"w)|pN-1dp
R 0 ./Sn- Jo
Z ( QR ) N
/ /
oo d7° l31'(a3 -|- 1°w)|dwdr
N 0 SN-1

l
I ( QR
N) N
/0 /'r,N-1d,r'
$n-1 n 1
l31,u(ac° -|- 1°w)|dw

ma
Z = QUO -|_ TCL) ) = (2R)N
N
/ uw-'LLln-ldz
BR
Z
( )| (2R)N
N BR
ldu(2)l
156 _ 2 I N _ 1 do.

We want to make two simple observations.

478 Lectures on the Geometry of Manzfolds

1. In the above lemma, we can replace the round ball BR centered at origin by any other
ball, centered at any other point. In the sequel for any R > 0, and any 5130 6 RN we set
l
U'$(),R u d .
vol (8R(w@)) BRIQCQ) (9) y
2. The inequality (l0.2.7) can be extended by density to any u E L1,1l8R)_

We will complete the proof of Morrey's theorem in three steps.

Step 1. LOO-estimates. We will show there exists C' > 0 such that, Vu E L1»P(RN) re
c1 (RN )
II"J»lloo _< C||u||1,p-
For each at 6 RN , denote by BCH) the unit ball centered at iv, and set Uac u . Using
(10.2.7) we deduce
ld71,ly)l do
lu»(:v)l -< me -|- Cn
B($l 1115 yIN-1
(10.2.8)
<0 ld'LLlyll d
Ilullp -I-
B(22) l:13 - y In-1 of/
Since uN (Lp) > 0, we deduce that p > N, so that its conjugate exponent q satisfies
_ P < N
q- p - l N - 1
In particular, the function y»-> l513-yl'(N'1> lies in Lq(B (iv))» and
1
d < C N,q 7
B ( x ) 1513 - 9 l 9 ( N I 1 > y- ( I
where C'(N, q) is a universal constant depending only on N, and q. Using the Holder
inequality in (10.2.8) we conclude

lu(1v)l _< Cll'u»ll1,p Vat.

Step 2. Oscillation estimates. We will show there exists C > 0 such that for all 11,6
L1»p(RN) n c1(1R(N)

[Ula _< CIIUIILp-

Indeed, from the inequality (l0.2.7) we deduce that, for any ball BR(a30), and any $ 6

BRlSl30)»

ldflU»(Y)l
lu(11)-um, l_<c
OR B w w - m n I
(q ppII

(I-I6lder)
_< cu ldul lip If B D ) 132
l
_ ylq(n_1) dy
1/<1
< CII Idol new 7
 Elliptic Equations on Manifolds 479

where

V
1
q(n q(n 1)) 1 X Of.

Hence
- I R _< CR" Vat E BRILC0I,
and consequently,
OSC (u; BR(;v0)) < CRO
Step 2 is completed.
Step 3. Conclusion. Given u 6 L1»p(1RN) we can find (un) 6 Co°(]RN) such that
un -> u in LLP and almost everywhere.
The estimates established at Step 1 and 2 are preserved as n -> oo. In fact, these es-
animates actually show that the sequence (un) converges in the cfO,a-norm to a function
v 6 C0"*(]RN) which agrees almost everywhere with u.

Exercise 10.2.27. Let u G L1'1(RN) satisfy a (q, v)-energy estimate, i.e.,

l
HC>0: N ld.(y)lqdy S 017'- 'VQ/JQRA 0<r<2,
7' B1~(f12)

where 0 < V < q a n d q > 1. Show that (up to a change on a negligible set) the function u
is a-H6lder continuous, Q l - I//q, and moreover
Z

[Ula S Co,
where the constant C2 depends only on N, Q, V and 01 .
Hint: Prove that
ld'u(yll
< Cro" 7
B,~(a:) l;z: - yln-1
and then use the inequality (10.2.7).

Remark 10.2.28. The result in the above exercise has a suggestive interpretation. If u
satisfies the (q, v)-energy estimate then although ldul may not be bounded, on average, it
"explodes" no worse that ,ra-1 as T' -> 0. Thus
1x1
aw) - u(0)l C' t'*'1dt all"
0

The energy estimate is a very useful tool in the study of nonlinear elliptic equations.

The Morley embedding theorem can be complemented by a compactness result. Let

(k,p) 6 2+ X [l, oo] and (m, a) 6 Z+ X (0, l ) such that
0N(k,p) > 07(m,0) k > m.
480 Lectures on the Geometry of Manzfolds

Then a simple application of the Arzela-Ascoli theorem shows that any bounded sequence
in L"'=P(RN ) admits a subsequence which converges in the cYTrn,a norm on any bounded
open subset of RN.
The last results we want to discuss are the interpolation inequalities. They play an
important part in applications, but we chose not to include their long but elementary proofs
since they do not use any concept we will need later. The interested reader may consult [4]
or [15] for details.

Theorem 10.2.29 (Interpolation inequalities). For each R > 0 choose a smooth, c u t #

function UR 6 03° (RN) such that
'UR if lac° < R 7

'UR EO ifl5UI>R-i-1,
and

ld17R(:c)l <2 Va: 6 RN.

Fix (m,pl 6 L X [l,ool, and (ha) 6 Z+ X (0, l ) .

(a) For every 0 < 7° < R -|- l , there exists C = C'(r, R , m , p ) such that, for every 0 <j<
m, 6 > 0 and for all u 6 Lm»p(1RN), we have

llvRullj,p,Rn S C€llvRullm,p,R~ + Cs - i ( m - j ) ||vRu||p,B-

(19) Forevers 0 < T' < R-I-1, there exists C = C(r, R, k , a ) such that,for every 0 < j < k,
5 > 0, and for all 'U 6 C'k"* ( R N ) , we have

HHURUHHMN -< 0ll»Rull,,Rn 0e-j<m-"ll»R~ll0,,8,.

The results in this, and the previous section extend verbatim to slightly more general
situations, namely to functions f : RN -> H, where H is a finite dimensional complex
Hermitian space.

10.2.4 Functional spaces on manifolds

The Sobolev and the HOlder spaces can be defined over manifolds as well. To define these

.. ..
spaces we need three things: an oriented Riemann manifold (M, g), a K-vector bundle
'7l` : E -> M endowed with a metric h = ( , ) , and a connection V = VE compatible

with h. The metric 9 ( » ) defines two important objects.

(i) The Levi-Civita connection V9 .

(ii) A volume form dvg : *l. In particular, d l , defines a Borel measure on M. We denote
by LP(M, K) the space of K-valued p-integrable functions on (M, dvgl (modulo the
equivalence relation of equality almost everywhere).

Definition 10.2.30. Let p 6 [l,oo]. An LLp-section of E is a Borel measurable map w

M -> E, i.e., ¢-1(U) is Borel measurable for any open subset U C E) such that the
following hold.
 Elliptic Equations on Manifolds 481

(a) $T o zP(a:) = x, for almost all as 6 M except possibly a negligible set.

la
(b) The function ac +-> llP(a:) is integrable with respect to the measure defined by avg.

The space of LLp-sections of E (modulo equality almost everywhere) is denoted by

Lp (E). We let the reader check the following fact.

Proposition 10.2.31. Lp ( E ) i5 a Banach space with respect to the norm

up»llp, E (lm n¢(a=)u d@g(w))/P if P < oo

esssupmllb(x)l if p i OO

Note that if p, q 6 [l, oo] are conjugate, l / p - I - l / q = l , then the metric h E X E->
M defines a continuous pairing

(.. 7 ) LPIEI X L'1(E)->L1(m,11<),

i.e.,

(W,(b)dvg S ll¢H@,E a H¢Hq,E-

M

This follows immediately from the Cauchy inequality

I h(¢(w),¢(w)) 4 s l¢(w)l l¢(w)l a.e. on M,

and the usual Holder inequality.

Exercise 10.2.32. Let E - > M ( i 1, 7 k) be vector bundles with metrics and consider
a multilinear bundle map
1-1

E1 X X

We regard : as a section of E1 (X) ®Ek. If

1-1
€ LPoIE1 (X) ®Ek)»
then for every PI , ,pa 6 [1>00]» such that
1 1/P0 = 1 /PI + + 1/Pk,
and V 6 LPS (Et),j = 1, 7 k

M
(Et, J ) Idog lIEIlp0°llW1llp1 Ilfbkllpk-

For each m l ,2, define am as the composition

am c°°(El V3 c°°(T*m (X) E)

VT*M®E
>
V
-> c°°(T*m ®'rn ® E),
where we used the symbol V to generically denote the connections in the tensor products
T*M ® j ® E induced by VI and bE .
482 Lectures on the Geometry of Manzfolds

The metrics g and h induce metrics in each of the tensor bundles T*M ® M ® and in
particular, we can define the spaces LP(T*M®"" (X) E).

Definition 10.2.33. (a) Let 'LL E L}OC(E) and E L}OC(T*M ® ' f 7 l ® E). We say that Vmu
'U weakly if

®'rn ® E).
(U, (z5)dvg (U, (V"")*(25)dV9 Vu 6 CO (T*M
M

(b) Define LWIEI as the space o f sections u 6 LP(E) such that, W 1, aM# there
exist vi 6 L1'(T*M ® j (go E), such that Vjfu, = UP weakly. We set
m

IIUIIw IIuII1n,p,E IIWuIIp .

o n

j l

" A word of warning. The Sobolev space U""'(E) introduced above depends on several
choices: the metrics on M and E and the connection on E. When M is non-compact this
dependence is very dramatic and has to be seriously taken into consideration.

Example 10.2.34. Let (M, g) be the space RN endowed with the Euclidean metric. The
trivial line bundle E = BM is naturally equipped with the trivial metric and connection.
Then, LPIBmI = LP(M, R).
Denote by D the Levi-Civita connection. Then, for every u 6 C'°°(M), and m 6 Z,L,
we have
Dmu = d:v®0' ® 3°'u,
lal=m
where for every multi-index a we denoted by da2®0* the monomial
doO11 ®---®d:z:01n.

The length of DM'u,(x) is

1/2
1

2
l<8'°'u(2¢ll
M m /
LPImI
II . H .
The space coincides, as a set, with the Sobolev space L'"»p<RNl. The norm
777,»PaR]vI
is equivalent with the norm I II1n,12,lRN introduced in the previous sections.

-
Proposition 10.2.35. (Lk'P(E), I lapEl is a Banach space which is reflexive :al < p<
oo.

The proof of this result is left to the reader as an exercise.

The HOlder spaces can be defined on manifolds as well. If (M, g) is a Riemann man-
ifold, then g canonically defines a metric space structure on M, (see Chapter 4) and in
particular, we can talk about the oscillation of a function 'LL : M -> K. On the other hand,
defining the oscillation of a section of some bundle over M requires a little more work.
 Elliptic Equations on Manifolds 483

Let (E, h, V) as before. We assume that the injectivily radius Pm o f M is positive. Set
Po := min{l, pM}. If ac, y 6 M are two points such that distg (at, y) < Po, then they can
be joined by a unique minimal geodesic ' Y y starting at ac, and ending at y. We denote by
TW : Et -> Ex the VE -parallel transport along 'Ye,y~ For each g 6 Ex, and T/ 6 Ey we
set
I5-77l : = | € - Ta¢,y77lw Z 117 - Ty,wf|y-
I f u : M - > E i s a s e c t i o n o f E , a n d S c M has the diameter < pg, we define
OSC (w 5) == sup{ lu(H1) - u(y)l; Xv y6 S }_
Finally set
[U]0,E :z sup { 1°"0'osc(u; B,.(a:)); 0 < 1° < pg, ac' 6 M .
For any k 2 0, we define
k

ll'U»ll1<,o,E lIe"ulloo,E + [ V M U ] a , T * M ® " " ' ® E 7

j 0
and we set
C B ) Q; {u 6 C"'(E); llullk,a < oo }.
Theorem 10.2.36. Let (M, g) be a compact, N-dimensional, oriented Riemann manifold,
and E a vector bundle over M equipped with a metric h, and compatible connection V .
Then the following are true.
(a) The Sobolev space Lrwlgl, and the HOlder spaces Cl""*(El do not depend on the
metrics g , h and on the connection V . More precisely, lfgl is a dQj'erent metric on M, and
V 1 is another connection on E compatible with some metric h1 then
L""P(E,g, h, V ) = Lm'plE7g1, h1, Vl) as sets of sections,
and the identity map between these two spaces is a Banach space is continuous. A similar
statement is true for the HOlder spaces.
(b) [ f l < p < oo, then C°°(E) is dense in LA<»p(E).
(c) If(k1,pil E 2+ X [1,oo) (i = 0 , 1) are such that
to > k1 and 0n(k0,P0) = to - N/P0 > I<1 - NlP 1 = unlk1,Pll»
then Lkomo ( E ) embeds continuously in Lk1,P1 ( E ) . If moreovel;
kg > k1 and kg - N/Po > I<1 - N/p1,
then the embedding Lk0,220(El <--> Lk1,P1 ( E ) is compact, i.e., any bounded sequence of
Lkomo ( E ) admits a subsequence convergent in the L/<:1,P1 -norm.
(d)If(m,p) 6 ZJ X [l,oo), ( k , a ) 6 2+ X (0, 1 ) and
m - N/P 2 k -|- cv,
then Lm»p(El embeds continuously in CAN<»<~(E>, If moreover
m - N/p > k -|- a,
then the embedding is also compact.

We developed all the tools needed to prove this theorem, and we leave this task to the
reader. The method can be briefly characterized by two key phrases: partition of unity and
interpolation inequalities. We will see them at work in the next section.
484 Lectures on the Geometry of Manzfolds

10.3 Elliptic partial differential operators: analytic aspects

This section represents the analytical heart of this chapter. We discuss two notions that play
a pivotal role in the study of elliptic partial differential equations. More precisely, we will
introduce the notion of weak solution, and a priori estimates.
Consider the following simple example. Suppose we want to solve the partial differen-
tial equation
Au+ u = f 6 L2(52,1R2.l, (10.3.1)
where A denotes the Laplace-Beltrami operator on the round sphere, A d*d. Riemann
suggested that one should look for the minima of the energy functional

> Elul
/V l
few dvg.
..II

'LL I
I
§(ldUI2 -I-u2)
S2

If U0 is a minimum of E, i.e.,
E(u0) _< E(u), Vu,

then
d
0 El'U»0 -|- tv) { (dug, d'u)g + 'LLO 'U f-'u dog Vu. (10.3.2)
dt 52

Integrating by parts we get

(d*du0 + U0 - f) - vdVg = 0 VU,

52

so that necessarily,
AU() + U0
There are a few grey areas in this approach, and Weierstrass was quick to point them out:
what is the domain of E, U0 may not exist, and if it does, it may not be C2, so that the
integration by parts is illegal etc. This avenue was abandoned until the dawns of the twen-
tieth century, when Hilbert reintroduced them into the spotlight, and emphasized the need
to deal with these issues.
His important new point of view was that the approach suggested by Riemann does
indeed produce a solution of (l0.3.l) "provided If need be that the notion of solution be
suitable extended". The suitable notion of solution is precisely described in (l0.3.2). Nat-
urally, one asks when this extended notion of solution coincides with the classical one.
Clearly, it suffices that u of (l0.3.2) be at least C2, so that everything boils down to a
question of regularity.
Riemann's idea was first rehabilitated in Weyl's famous treatise [131] on Riemann sur-
faces. It took the effort of many talented people to materialize Hilbert's program formulated
as his l 9th and 20th problem in the famous list of 27 problems he presented at the Paris
conference at the beginning of the 20th century. We refer the reader to [l] for more details.
This section takes up the issues raised in the above simple example. The key fact which
will allow us to legitimize Riemann's argument is the ellipticity of the partial differential
operator involved in this equation.
 Elliptic Equations on Manifolds 485

10.3.1 Elliptic' estimates in RN

..
Let E = Q' denote the trivial vector bundle over RN . Denote by ( , ) the natural Hermi-

Euclidean volume) will be denoted by .

San metric on E, and by 3, the trivial connection. The norm of Limp (E) (defined using the
I I k»P° Consider an elliptic operator of order m
L AA(lag : C°°(E) -> C°°(E).
Iafl<m
For any k e y , and any R > 0 define

lILIlk,R sup ll00(A(>>ll.

l0l$'f",IBI$k+m-lol BRIO)
In this subsection we will establish the following fundamental result.

Theorem 10.3.1 (Interior elliptic estimates). (a) Let ( k , p l 6 2+ X (l,oo), f d R > 0.

Then, there exists C' C(llLIlk_l.1 R , k,p, N, R) > 0 such that, Vu 6 C§°(E IBR(0)), we
have

HUH RI-I-m,p S C( Hsu/Hk,p + Ilullpl- (10.3.3)

(19) Let ( a l G Z+ X (0, l ) , and R > 0. Then, there exists C' CIHLHI<+1,R= k,
a, N, R ) > 0 such that, 've, 6 C'§°(EIBR(0)l, we have

HU/Hk-l-m,0¢ S C( HIS/Hk,Ql + l1/ull0,@ >- (10.3.4)

Proof. The proof consists of two conceptually distinct parts. In the first part we establish
the result under the supplementary assumption that L has constant coejicients. In the sec-
ond part, the general result is deduced from the special case using perturbation techniques
in which the interpolation inequalities play an important role. Throughout the proof we
will use the same letter C' to denote various constants C' = C'(llLIlk:+1 R , k , p, N , R ) > 0
Step 1. Constant coejicients case. We assume L has the form

L AAA; 7

l0l='rn

where As, are r X 7° complex matrices, independent of QUO GUN. We set

IILII HAH.
We will prove the conclusions of the theorem hold in this special case. We will rely on a
very deep analytical result whose proof goes beyond the scope of this book.
For each f 6 L1(RN , (I) denote by f(§) its Fourier transform

Ns) -
exp(-ia: §)dac.
..
II

Rn
486 Lectures on the Geometry of Manzfolds

-
Theorem 10.3.2 (Calderon-Zygmund). Let m : SN-1 -> ( I be a smooth function, and
dejine
m(g>¢RN\{0}->(c
by

mg)
\ Ill/
Then the following hold.
(a) There exist Q 6 COO(SN-1, (Cl, and C 6 (I, such that
(dll s n _ 1 QCl'UsN-1 = 0.

(Ag) For any u 6 C5° (RNA, the limit

lim 9(9)
(Tu)(32) - y)dy
e\0
l?,/l26 I/IN 'u,(a)
exists for almost every :IJ E RN, and moreover
1
e7u,(:/:) -|- T7u,(:z:) = (2 IN exp(i:z: . €)m(€)€i(€)d€-
Rn

(19) For every l <p< oo there exists C = C'(p, llmllOO) > 0 such that,

lITUIlp S ClMlp Vu G 03°(RN)-

(c) (Korn-Lichtenstein). For every 0 < a < l , and any R > 0, there exists C
C'(a, llmuu@,, R) > 0 such that, Vu 6 o§°'(BR)

ii/w CIIull0,o,BR~

For a proof of part (a) and (b) we refer to [125], Chap. II §4.4. Part (c) is "elementary"
and we suggest the reader to try and prove it. In any case a proof, of this inequality can be
found in [15], Part 11.5.
Let us now return to our problem. Assuming L has the above special form we will
prove (l0.3.3). The proof of (l0.3.4) is entirely similar and is left to the reader. We discuss
first the case k = 0.
Let u E 08° (E | BR l. The function u can be viewed as a collection

u(;u) (we), . ,w°(x)),

Z • •

of smooth functions compactly supported in BR. Define

me) :i (M), . . . ,@'°lg))-
If we set 'U = Lu, then for any multi-index 5 such that | B| m, we have
L é u = 35]/u, = 351),

because L has constant coefficients. We Fourier transform the above equality, and we get

(-iv" Aa€<~<8~5»,,,(€) = (-i)m€"v©e). (10.3.5)

l0l=m
 Elliptic Equations on Manifolds 487

Note that since L is elliptic, the operator

0(L)(€) i A(€) AOfo : C" -> ICT'

l0l='rn

is invertible for any € ¢ 0 . From (1 l.l.7) we deduce

bw§) = <s@8(§)~(E) we ¢ 0,
where B(§) = A(§)-1. Note that B(§) is homogeneous of degree -m, so that M(€)
£@B(€) is homogeneous of degree 0. Thus, we can find functions m y (go 6 cOO (RTL \ {0}),
which are homogeneous of degree 0, and such that

aftuilg) My (6)1%(6)-
j

Using Theorem 10.3.2 (a) and (b) we deduce

uu@%lup < ClMlp Z Clllullp-

This proves (10.33) when L has this special form.
Step 2. The general ease. Suppose now that L is an arbitrary elliptic operator of order m.
Let r > 0 sufficiently small (to be specified later). over BR by finitely many balls B,.(:z:,,),
and consider 17, 6 CO (B,~(;z:,,)), such that each point in BR is covered by at most Ion of
these balls, and

n 2 0, 77, 7

llawl < CT' III VI/3l S m.

f u G Co°lBRl» then

Set 'Usu I UVU/9 and

L, := Aala3,,)3O'.
l0l=1n
. .
We rewrite the equality vI/
de
=f Lu, as

Liu,, = (L, - L)u,, -|- v V = eau(m)3"ul/ - A5(;u)30u,, -|- v,,,

lal='rn l5l<m
where e0,,(x) = Aa(x1,/ - A0,(a:). Using (10.3.3) we deduce

lluvllm,pSC(llu~llp,8(x,>+llvvllp,B(,)+ ||5o,v(33l6"u||p,8,(m,>+||u||m-1,p,B,(>)-
la|='rn

Since lead/(x)l < Cr on B,°(xV/), where C = C(llLII1 R), we deduce

ll'U»ullvn,p,B,~(x,,) < C(t°llw/llp,8(x) -|- ||w/1118(x> -|- ||'Uv||p,B»~(wv)l-

488 Lectures on the Geometry of Manzfolds

We can now specify 7° > 0 such that Cr < l /2 in the above inequality. Hence

ll'U»ull1n,p,8,-(:1:,,) < 0(llw/llp,8(m -|- ||Uv||p,B,~(wv)l-

We need to estimate llvVIlp. We use the equality

'Up Z L(w~0 TI,,LU + [L, MU,
in which [L, 17V/] = ad(n,,)L is a p.d.o. of order m l, so that

II [L, 77v]UIlp,B,~(wv) llullm-1,p,B,.(x,,)~

Hence

|M|p,B<> -< o(llnLullp,B(> + T -(m-1) Ilullm-1,p,B(w))=

SO that

H,U'VHM,P,B»r-(32,/I _< C(llLullp,BR + -<m-1>llullm-1,p,BR).

We sum over u, and taking into account that the number of spheres B e ) is O((R/T) N )
we deduce
N l)
LuH -(m+
Hi ll7n,p < CAN(-NH llullm-1 »P,BR).
-|- T -
llullm,p,B3 < P,BR
1/

Note that 7' depends only on R, p, ||L||1,R- so that

llullm,p,BR S C(llLullp,BR + HU'H-11p1BRl7

where C' is as in the statement of Theorem 10.3.1.
We still need to deal with the term llullm_1 p a R in the above inequality. It is precisely
at this point that the interpolation inequalities enter crucially.
View u as a section of C'§°(E IB2Rl° If we pick 17 € CgO(132R) such that 17 E 1 on BR,
we deduce from the interpolation inequalities that there exists C > 0 such that

llullm-1,BR _< <°5ll'UIltn,p,R + C5-(M'1) IIullp,8R .

Hence

lIUIIM,P,BR _< ClllLUIIP,BR -|- 6||'u||m,p,8R -|- @'(""`1)||u||p,BRl-

If now we choose 5 > 0 sufficiently small we deduce the case k = 0 of the inequality
(l0.3.3).
To establish this inequality for arbitrary k we argue by induction. Consider a multi-
index III = lc. If 6 Co°(EIBR), and Lu = U then
L(a%l = 30]/u, -|- [L, 3B1'u, = a% -|- [L, am.
The crucial observation is that [L, 301 is a p.d.o. of order < m -|- k - 1. Indeed,

0m+k(lL,@B]) i t@m(L),@(@")1 i 0.
Using (10.33) with I<5 = 0, we deduce
B
ll@%llm,p,BR _< 0(ll3 VIII,BR -|- ||lL,@%||p,BR + ll@%llp,8R)
_< Clll'UIlk,p,BR -|- ||u||m+k-11p,8R -|- ||'LL||k,p,BR).
 Elliptic Equations on Manifolds 489

The term llullm+k-1,p,8R + lIUIlk,p,BR can be estimated from above by

€llullm+k,p,BR + Cllullp,8R,

using the interpolation inequalities as before. The inequality (10.3.3) is completely proved.
The HOlder case is entirely similar. It is left to the reader as an exercise. Theorem 10.3.1 is
proved.

Using the truncation technique and the interpolation inequalities we deduce the follow-
ing consequence.

Corollary 10.3.3. Let L as in Theorem ]0.3.], and fix 0 < 1° < R. Then,for every lc 6 Z+,
l < p < oo, and a 6 (0, 1), there exists C' = C'(k,p,a,N, lILIlk,+1 R,R,1°) > 0 such
rhaz, Vu 6 C°°(E)
lull+m,p,B _< C( lIL'LLIlk:,p,BR + Ilullp, )
and

HUHk3+M,0!,B~r» cl lIL'u»llk,o,8R llull0,a,8R l.

Exercise 10.3.4. Prove the above corollary.

10.3.2 Elliptic regularity

In this subsection we continue to use the notations of the previous subsection.

Definition 10.3.5. Let u, 'U : RN -> C" be measurable functions.

(a) The function u is a classical solution of the partial differential equation
Lu = AO4(;z:)c9a*u (al = v(;z:) (10.3.6)
IBIS
if there exists cu G (0, 1) such that, v e of,§"(RN), u G CzocaIRNI7 and (10.3.6) holds
everywhere.
(b) The function u is said to be an Lt'-strong solution of (10.3.6) if u 6 L l o c p l R N l 9 'U 6
L o c (RNA,
and (10.3.6) hold almost everywhere. (The partial derivatives of u should be
understood in generalized sense.)
(c) The function u is said to be an LLp-weak solution i f u, U 6 LQOCIRNI and

(U, L*Q5)da;' (11>45) do, WE E CO

Rn Rn

Note the following obvious inclusion

{Lp - weak solutions} D {LP - strong solutions}.
The principal result of this subsection states that when L is elliptic the above inclusion is
an equality.
Theorem 10.3.6. Letp 6 (1, oo) and suppose that L : COO(E) -> cOO (E) is an elliptic
operator of order m, with smooth coejicients. If u 6 L O C and Lu 6 L then u 6
LloePlEl°
490 Lectures on the Geometry of Manz]'old5

Remark 10.3.7. Loosely speaking the above theorem says that if a "clever" (i.e., elliptic)
combination of mixed partial derivatives can be defined weakly, then any mixed paltial
derivative (up to a certain order) can be weakly defined as well.

Proof. 2 We prove only two special special cases of the theorem. Namely we assume that
L has either constant coefficients or it is a first order operator. The essential ingredient in
the proof is the technique of mollification. For each 5 > 0 we denote by TO the operator
of convolution with p6. Fix a radius R > 0 and a partial differential operator S of order
m > 0. We first prove that there exists a constant C' > 0 depending only on R, N, p, S
such that

II [S,T6]u HpaBR+6 _< CIIUIIM-1,P»BR+1) Vu E 00 (8R+1)- (l0.3.7)

Using the equality

[s1s2, $51 = [So T5]S2 + SO $s2, TO]

we deduce that if (10.3.7) holds for the operators $1 and $2 of positive orders then it also
holds for their product $152- Clearly (10.3.7) is trivially satisfied for operators S with
constant coefficients because in this case [S, TO] = 0. Thus, it suffices to verify (10.3.7) for
first order operators of the form a(:c)3,~ -|- b(;I:).
Denote by Mb : 05° -> 03° the operator of multiplication of by the smooth bundle
morphism b(a2). Then

[Mb) T6]u(a:) Rn P6(3U - y)( b(:z:l - 6(9) Lu(y)dy,

and the smoothness of b implies that there exists a constant C' > 0 independent of 5 > 0
such that

II [Mb,T6] lIP»BR+6 < CIIUIIP,BR+1 v Vu G CO (BR+1)-

Next observe that

[ T6]u(w) a(;u)3¢
/
Rn M23 - you(uldy Rn 06 lai' - yl@(y)3y u(yldy

Z
1
6n+1
/ Rn
a 0 I ).CU` y
5
ac - y
<»()u(y)dy

+6 1+1 /
Rn
8yi p
( 5 ) @(¢ulu(y)dy -|-
/ Rn
p6(=U y)5@@(ylu(yldy

(@w(H 1 - y) = -8y~0(;v - y))

ac

6n+1 lx-yl<5
MY
( 53/
) (WJ) Gay) )u(y)dy

-|- p6(12 - y)<9y»G»(y)u(y)dy-

|33-y|<5

21 would like to thank my friend Weir in Chen for pointing out a subtle error in the original proof.
 Elliptic Equations on Manifolds 491

The inequality (l0.3.7) follows by observing that la(x) -a(y)l = 0(6) when ac,y 6 8R+1,
ac yl<5.
We deduce that if L is a first order operator or L has constant coefficients there exists a
constant C > 0 depending only on R, N, q, L such that
II [T5,L*]'LL Hq,BR_l_5 -< CHUHqaBR-I-17 Vu E CO (8R+1l- (10.3.8)

Suppose that u 6 L O C and 'u = Lu 6 LOc. Then for any (15 6 C§°(BR+1) we have

(U L*T5qz5)dV +
7
( Up [T5,L*](l5)
BR-l-1

[TO, Lw(2»).

Let q 6 ( l , o o ) such that + 1q 1. Using (10.3.8) and Holder inequality we deduce

lLT5u, q5dv < |M|p, 'R-l-1llT6¢llq,BR+1 -|- Cllullq, 'R-1-1 ll¢llq,BR+1

8R-l-1

< C(ll'U"llP,BR+1 + ll'0llP,BR+1 ll¢llq»BR-l-17

for some constant C depending only on R, N, q, L. This implies that LT67u, is bounded
in LP(BR+1). From the elliptic estimates we deduce that the family Tau it is bounded in
LM'P(BR) as 5 -> 0. Since the space L'"10(BR) is reflexive we deduce that a subsequence
Thru converges weakly as 5n -> 0 to a section U0 6 L"""'(BRl. On the other hand,
from the Rellich-Kondrachov compactness theorem we deduce that Thru converges in the
LLp-norm to up. This implies that U0 = u so that u E Lloep'

Remark 10.3.8. The proof of the general case of the regularity theorem requires consider-
able more analytic work than we are willing to include in a geometry book. However, the
general idea is easy to describe.
Using the theory of pseudo-differential operators, one can construct a parametrix for
any elliptic operator L on RN T. This is an integral operator T : C8° (RN ) -> C°°(RN )
which serves as a sort of inverse for L, i.e., the operators S = LT - II and Se = T L - II
are smoothing operators. This means that they map locally integrable functions (and more
generally, distributions) to smooth functions. The convolution with PA is a prime example
of smoothing operator, while the convolution with the function lacl2'N is the parametrix of
the (geometers') Laplacian on RN .
If Lu = 'U weakly then
Tv=TL'u,=u+Sgu.

Thus To - 'LL is a smooth function, and thus it suffices to prove that To 6 LlOCp . This is
achieved, again using the theory of pseudo-differential operators, the Calderon-Zygmund
inequality, and the special case of Theorem 10.3.6 when L is a constant coefficients opera-
tor. For details we refer to [126].
492 Lectures on the Geometry of Manzfolds

Corollary 10.3.9. [few 6 LCIEI i | weak LLp-solution of Lu = u, 1 < p < oo, and
U 6 L I E ) , then u 6 L?O*""°"(E), and for every 0 < 1° < R we have
llulln+k=,p,8 _< C(|M|k=,p,BR + Ilullp, 'RI)
where as usual C = C(llLIlk+7n+1 R, R, T, . . .).

Proof. To prove that u 6 L§°O+'"»p(El we argue by induction on lc. The case k = 0 is

settled by Theorem 10.3.6. For the inductive step assume that u 6 L O C and Lu 6 LlkO;1,19
implies that u 6 L?O;1+"'°". We want to prove that u 6 LDOC and Lu 6 Llkop implies that
u 6 LlkO+v7z,p . Set U = Lu. The elliptic regularity implies that 3111, 6 LiDOs' Moreover we
have the weak equality

3¢'U = 3iL'U, I L3¢'U, -|- [a L]u,

i.e.,

(3¢'U,¢) = (Fu, (La+¢)*d)) + (Fu, [6¢,L]*d>>7 via 6 00 .

To put it differently, the section Qu is a LLp-weak solution the equation

Ll8z"LLl = 6i1) - [g L]u.

Observe that [a L] is a p.d.o. of order g m which means that the right-hand side of the
above equality is in
. . .
L5;;1"'
since the inductive assumption implies that u 6 LiO;1+M'P.
. . k-1 . .
Invoking the inductive assumption again we deduce that 3-7u, 6 Lloc --M'P, 'v'z, 1.e., 'U 6
Le+m,p
loe •
To prove the elliptic estimate pick a sequence un 6 C3°(E) such that

u n - > u strongly in Lzko` m'p(El-

Then
lc m,
Lun -> Lu strongly in Ll0+ PIEI,
and

llunllm+k,p,8, < C(llunll0,p,8R + ||L'w-||k,p,8R)-

The desired estimate is obtained by letting n -> oo in the above inequality.

Corollary 10.3.10 (Weyl Lemma). If 'LL E LQDOCIEI is an LLp-weak solution of Lu U,

and 'U i5 smooth, then u must be smooth.
k
Proof. Since 'U is smooth we deduce U E LlOP Vi E ZJ Hence u 6 LlkO+m'p Wk. Using
Morley embedding theorem we deduce that u 6 ClOCO Wm 0.

The results in this, and the previous subsection are local, and so extend to the more
general case of p.d.o. on manifolds. They take a particularly nice form for operators on
compact manifolds.
 Elliptic Equations on Manifolds 493

Let (M, g) be a compact, oriented Riemann manifold and E, F -> M be two metric
vector bundles with compatible connections. Denote by L E PDO"'(E, F) an elliptic
operator of order m.

Theorem 10.3.11. (a) Lets 6 I;1>(E) and 6 Lfwlpl, 1 <p< oo, such that

(u, Q5)Fd'U9 (U, L*q25)EdVg WE 6 C§°(F).

M M

Then 'LL E LA<+~»p<E), and

|M|k+mp,E -< C`(|M|1<,p,F -|- ||u||p,E),

where C = C'(L, k,p).
(be [few E CMIEI and E gfwlpl (0 < a < 1 ) are such that Lu = U, then u 6
CTT1-l'k,CIlE), and

lIUIlk+m,a,E -< C(llull0,o,E + |M|1<,<.,F)'

where C = C(L, k , Q).

Remark 10.3.12. The regularity results and the a priori estimates we have established so
far represent only the minimal information one needs to become an user of the elliptic
theory. Regrettably, we have mentioned nothing about two important topics: equations
with non-smooth coefficients, and boundary value problems. For generalized Laplacians
these topics are discussed in great detail in [51] and [102]. The boundary value problems
for first order elliptic operators require a more delicate treatment. We refer to [21] for a
very nice presentation of this subject.

Exercise 10.3.13 (Kato's inequalities). Let (M, g) denote a compact oriented Riemann
manifold without boundary. Consider a metric vector bundle E -> M equipped with a
compatible connection V.
(a) Show that for every u 6 L1,2(E), the function as »-> lu(a3)l is in L1'2(M), and moreover

ldlu(@3)ll < IVu(1/:)l»

for almost all a : € M .
(b) Set AE = V * V, and denote by AM the scalar Laplacian. Show that, for all u p
L2,2(E)9 we have

Amllul2l 2(AE'U,, 'LL)E - 2IVul2.

Conclude that WE E cOO (M) such that (15 2 0, we have

(dlul,d(¢lul))9 dvg < (ARM, dvg 7

M M

i.e.,

lu(H1)IAm(lu(w)l) (AE'LLl£Ul,Ul£Ul)E weakly.

494 Lectures on the Geometry of Manzfolds

Exercise 10.3.14 (Local estimates). Suppose (M, g) is a connected, oriented, Riemann

manifolds of dimension m, not necessarily, Et, E1 -> M are two smooth hermitian vector
bundles on M, and V , z' = 0, l are Hermitian connection on Et.
Suppose that L : cOO (EO) -> cOO (El) is an elliptic operator of order u with smooth
coefficients. Prove that for any compact subset K C M, any relatively compact open
neighborhood U of K, and any p E (1>O0), Z E Z>0, there exists a positive constat C'
depending only on K, U, p, Z, VO, V1, and the coefficients of L such that, Vu 6 C°°(E0),
we have
ll'U»llL£+v»p(K,E0) < C( lIL'U»llL@»p(U,E1) + ll"~'»llLp(U,E0) )»
where the above Sobolev norms are defined in terms of the connections Vi

10.3.3 An application: prescribing the curvature of surfaces

In this subsection we will illustrate the power of the results we proved so far by showing
how they can be successfully used to prove an important part of the celebrated uniformiza-
tion theorem. In the process we will have the occasion to introduce the reader to some tricks
frequently used in the study of nonlinear elliptic equations. We will consider a slightly more
general situation than the one required by the uniformization theorem.
Let (M, g) be a compact, connected, oriented Riemann manifold of dimension N. De-
note by A = d*d : C'°°(M) -> C'°°(M) the scalar Laplacian. We assume for simplicity
that
volg(m) avg l,
M
so that the average of any integrable function (p is defined by

SO S0l35)dV9 (313)-
M
We will study the following partial differential equation.
Au -|- f ( u ) = s(:1:), (10.3.9)
where f : R -> R, and S 6 COO(M) satisfy the following conditions.
(Ci) The function f is smooth and strictly increasing.
(Co) There exist a > 0 and b E R such that
f(t)2a¢+b W€R.
Set

Fw) fltldt.
0
We assume
(CO) limltl_,oo(F(t) - it) oo, where 3 denotes the average of s,

s s(ac)dVg .
M

Theorem 10.3.15. Let f and s(:1:) satisfy the conditions (Cl), (CQ), CO). Then there exists
a unique u 6 cOO ( M ) such that
Au (513) -|- f(u(.5v)) = 5(2) Van 6 M.
 Elliptic Equations on Manifolds 495

Proof. The proof of this theorem will be carried out in two steps.
Step 1. Existence of a weak solution. A weak solution of (10.3.9) is a function 11,6

L1=2(M) such that f ('u,(ac°)) 6 L2(Ml» and

{(dul d¢) -|- f(u)(25}dVg s(:z:)(l5(:z:)dVg(:c), $5 6 L1,2(M).

M M

Step 2. Regularity. We show that a weak solution is in fact a classical solution.

Proof of Step 1. We will use the direct method of the calculus of variations outlined at the
beginning of the current section. Consider the energy functional

1
/{
2
I L122(m) -> R, I(u) M §ldu(f1>ll +F(u(H1)) gl£I3)u(.cul dVg(H1l-

This functional is not quite well defined, since there is no guarantee that F('u,) E L1 (M)
for all u 6 L1)2(M).
Leaving this issue aside for a moment, we can perform a formal computation ii la
Riemann. Assume u is a minimizer of I, i.e.,

Ku) § I (u), /u 6 L1,2lMl

Thus, for all (IS 6 L1'2(M)

Nu) _< I ( u + ¢ ¢ > it 6 R.

Hence t = 0 is a minimum of he) = I(u + W), so that h;,,(0) = 0 '(/(IS. A simple

computation shows that

he) M{(du, d¢l + f(u)(z) s(;z:)(;5(;z:)}dVg : 0,

so that a minimizer of I is a weak solution provided that we deal with the integrability issue
raised at the beginning of this discussion. Anyway, the lesson we learn from this formal
computation is that minimizers of I are strong candidates for solutions of (10.3.9).
We will circumvent the trouble with the possible non-integrability of F(u) by using a
famous trick in elliptic partial differential equations called the maximum principle.

Lemma 10.3.16. Let h : R -> R be a continuous, strictly increasing function and u, u p

L1,2(Ml such that
(i) h e ) , h e ) 6 mm).
(ii) Au -|- h(u) 2 As -|- h('u) weakly, i.e.,

{ldu, d¢) -|- hlu)(b}dvg {do, d(z5) -|- h(v)¢}dVg (10.3.10)

M M

'v'g25 6 L1,2lml such that <15 > 0 a.e. M . Then u > 'U a.e. on M .
Proof. Let
1
(u u) min(u u, 0) §{lU u) lu 'ul}.
496 Lectures on the Geometry of Manz]'old5

According to the Exercise 10.2. 19 we have (u - 'u)' 6 L1,2(M), and

d u - u) a.e. on {u < v}
d(u u) (
0 a.e. on {u 2 v}
Using ¢ = - ( u - u ) in (l0.3.l0) we deduce
{ad(u-v),d(u-v) -|- (Mu) -h(v)llu-uI- }dvg > 0.
M
Clearly
ld(u - u), d(u - ful-) = ld(u - U)-I 2 7
and since h is nondecreasing,
(h(u) - h(ul)(u - 'u)` > 0.
Hence,

ld(u u) I 2 dog < (h(u) h(u) )(u 'UV dog < 0,

M M
so that
ld(u v)ll E 0.
Since M is connected, this means (u - u) c S O . I f c < 0, then
(u - u) E (Fu Fu) Ca

so that u = 'U + C < U. Since h is strictly increasing we conclude

h(u) dog < h(v)dVg.
M M
On the other hand, using go E l in (10.3.10), we deduce
h('u,)dVg > h('u)dv9 °I
M M
Hence C cannot be negative, so that (u - 0)' E 0 which is another way of saying u > v .
The maximum principle is proved.

Exercise 10.3.17. Assume h in the above lemma is only non-decreasing, but u and 'U satisfy
the supplementary condition
up
Show the conclusion of Lemma 10.3. 16 continues to hold.

We now return to the equation (10.3.9). Note first that if u and 'U are two weak solutions
of this equation, then
Au -|- f(u) > (_<) As -|- h(v) weakly,
so that by the maximum principle u 2 (_<) This shows the equation (l0.3.9) has at MOS[
one weak solution.
To proceed further we need the following a priori estimate.
Lemma 10.3.18. Let u be a weak solution of(I0.3.9). If C is a positive constant such that
f ( C ) 2 sup s(:u),
M
then 'u,(:1:) <C a.e. on M.
 Elliptic Equations on Manifolds 497

Proof. The equality f(C) > sup s(:c) implies

AC -|- f(C) > SUM) = Au -|- f ( u ) weakly
so the conclusion follows from the maximum principle.

no Fix C0> 0 such that f(C0) > sup s(x). Consider a strictly increasing C2-function
f :]R-> R such that

f(u)=f(u) for u _<c0,

He) is linear for u > c0+1.

The conditions (all and (02) imply that there exist A, B > 0 such that

lf"(u)l < AIlI + B. (l0.3. I I )

Lemma 10.3.19. If u i5 a weak solution of

Au -|- Ru) = s(:c) (l0.3.l2)

then u is also a weak solution of(]0.3.9).

Proof. We deduce as in the proof of Lemma 10.3.18 that u < Co, which is precisely the
range where f coincides with ii

The above lemma shows that instead of looking for a weak solution of (l0.3.9), we
should try to find a weak solution of (l0.3.l2). We will use the direct method of the
calculus of variations on a new functional
1 2
ill)
/{
M §ldU(£Ull +F('u~(:v)) s(x)u(w) do),

where
u
Flu) jltldt.
..

The advantage we gain by using this new functional is clear. The inequality (l0.3.l 1) shows
F has at most quadratic growth, so that plus 6 L1 (M), for all u 6 L2(M). The existence
of a minimizer is a consequence of the following fundamental principle of the calculus of
variations.

Proposition 10.3.20. Let X be a reflexive Banach space and J : X -> R a convex, weakly
lower semi-continuous, coercive functional, i.e., the sublevel sets

JC ac ; J(:z:) S 0}
are respectively convex, weakly closed and bounded in X . Then J admits a minimizer; i.e.,
there exists £130 6 X such that
Jl£I30l_<Jl£L'l \V/ZUGX.
498 Lectures on the Geometry of Manzfolds

Proof. Note that

i f J(aC) = inf
Jo J(.cc).

Consider .CUnGJe such that

lim J(ac2n1 ) inf J.

Since X is reflexive, and JC is convex, weakly closed and bounded in the norm of
we deduce that JC is weakly compact. Hence a (generalized) subsequence (xV) of
converges weakly to some $130 6 Je. Using the lower semi-continuity of J we deduce
JI30) _< lim inf J(.§c,,) inf J.

Hence $0 is a minimizer of J. The proposition is proved.

The next result will conclude the proof of Step 1.

Lemma 10.3.21. I is convex, weakly lower semi-continuous and coercive (with respect to
the L i - n o r m ) .

Proof. The convexity is clear since F is convex on account that f' is strictly increasing.
The functionals

u »-> ldul2dVg and u »-> s(;z:)u(a:)dVg(:c)

M M

are clearly weakly lower semi-continuous. We need to show the functional

u »-> F(u)dVg
M
is also weakly lower semi-continuous.
Since F is convex, we can find a > 0, B 6 R such that
F('u,l-au-B>0.

If un ->u strongly in L1'2(M) we deduce using the Fatou lemma that

{F(u) < F n aun 5}dvg .

M - 'n->oo M{ (u, I
On the other hand,

lim aun -|- BCZVg au -|- Bdvg ,

n M M

which shows that

This means that the functional

L1,2lml 9 u »-> F(u)dVg
M
 Elliptic Equations on Manifolds 499

is strongly lower semi-continuous. Thus, the sublevel sets

{ /
u;
M
F(u)dvg < c 7

are both convex and strongly closed. The Hahn-Banach separation principle can now be
invoked to conclude these level sets are also weakly closed. We have thus established that
I is convex and weakly lower semi-continuous.

Remark 10.3.22. We see that the lower semi-continuity, and the convexity conditions are
very closely related. In some sense they are almost equivalent. We refer to [33] for a pre-
sentation of the direct method of the calculus of variations were the lower semi-continuity
issue is studied in great detail.

The coercivity will require a little more work. The key ingredient will be a Poincaré
inequality. We first need to introduce some more terminology.
For any u 6 L2 (M) we denoted by U its average. Now set
ulll = u(a:) U.

Note the average of u is 0. This choice of notation is motivated by the fact that u is
perpendicular (with respect to the L2(M)-inner product) to the kernel of the Laplacian A
which is the 1-dimensional space spanned by the constant functions.

Lemma 10.3.23 (Poincaré inequality). There exists C' > 0 such that

ldul2dvg 2 C lull2dV9, Vu G Ll 2(M).

M M

Proof. We argue by contradiction. Assume that for any 6 > 0 there exists u5 E L1,2(M),
such that

IU_LI2dVg = l, and ldu5l2dvg < e.

M M
L1,2lM
The above two conditions imply that the family ( u ) is bounded in ). Since
L122(M) is reflexive, we deduce that, on a subsequence
_L
Us 11, weakly in L1»2(M).
The inclusion L1,2(M) --> L2lml is compact, (Rellich-Kondrachov) so that, on a subse-
quence
b J_E - > v strongly in L2lml.
This implies 'U 7é 0, since

Wdvg lim
5
lull2dvg = 1.
M M
On the other hand,
duE du -> 0, strongly in L2.
500 Lectures on the Geometry of Manzfolds

We conclude

(do, d¢) dog WE E L1)2(M).

M
In particular,

(do, do) dog = 0,

M

so that do = 0. Since M is connected, we deduce 'U C const. Moreover,

C 'v(w)dVg(sl lim uJ_dvg 0.

M 5 M
This contradicts the fact that 'U ¢ 0. The Poincaré inequality is proved.

We can now establish the coercivity of I(u). Let I<3> 0, and u 6 Ll,2lml such that

Since
lV
M
2 ldu(H>)l2 + F(u(w)) s(:L°)u(:z:) dvglxl <I<z. (l0.3.l3)

/M
ldu12 asg
/M
it suffices to show that the quantities
ldull2dvg

u,
/M
lull2dvg,
are bounded. In view of the Poincaré inequality, the boundedness of
/ M
ldulldvg

ldull2dvg
M
implies the boundedness of ||'u,7- 112,m, so that we should concentrate only on U and lldu_L 112.
The inequality (l0.3.l3) can be rewritten as
1 -
/{
M
§ldull2 + Fw)

The Poincaré and Cauchy inequalities imply

ClluJ_ll§ 115_L112 IIu_LII2 +

»
F(u)dVg S e u <i<;.
M

Since volg(M) = 1, and F is convex, we have a Jensen-type inequality

(f M
udVg
) / <
M
F(u)dVg ,

SO that
C11ul11§ - 115_L112 11u_L112 + Fm) - § - U g H. (103,14)

Set pa) CR 115; ll2t, and let m = inf P(t). From the inequality (10.3.14) we deduce

F('u,) S </<, m.
 Elliptic Equations on Manifolds 501

Using condition (CO) we deduce that III must be bounded. Feed this information back
in (l0.3.l4). We conclude that p(llu_LII2) must be bounded. This forces l1u_L112 to be
bounded. Thus I is coercive, and emma 10.3.21 is proved.

Step 2. The regularity of the minimized: We will use a technique called bootstrapping,
which blends the elliptic regularity theory, and the Sobolev embedding theorems, to grad-
ually improve the regularity of the weak solution.
Let u be the weak solution of (10.3. 12). Then u(:1;) is a weak L2 solution of

Au = h(a:) on m,
f'
where 11(2) = -f(u(:1:)) - s(:z:). Note that since the growth of is at most linear, we
have t"(u(g;3) 6 L2lml. The elliptic regularity theory implies that u 6 L2'2(M). Using
Sobolev (or Morrey) embedding theorem we can considerably improve the integrability of
u. We deduce that
(i)eitheru6 L q ( M ) i f - N / q < 2 - n/2 < 0 ( d i m M = N ) ,
(ii) or u is HOlder continuous if 2 - n/2 > 0.
In any case, this shows 6 Lil I I I , for some $1 > 2, which implies h(a:) 6
L91 ( M ) . Using again elliptic regularity we deduce u 6 L2'q1 and Sobolev inequality
implies that u 6 Lq2 (M) for some 1/2 > q i -
After a finite number of steps we conclude that h(:c) 6 L<1(M), for all q > 1. Elliptic
regularity implies Fu. 6 I;2»11(M), for all q > 1. This implies h(:r:) 6 L2'q(M) ,for all q > 1.
Invoking elliptic regularity again we deduce that u 6 I;4»<t(M), for any q > 1. (At this
point it is convenient to work with f , rather than with p which was only 02.) Feed this
back in h(a:), and regularity theory improves the regularity of 'U two orders at a time. In
view of Morrey embedding theorem the conclusion is clear: u 6 COO(M). The proof of
Theorem 10.3.15 is complete.

From the theorem we have just proved we deduce immediately the following conse-
quence.

Corollary 10.3.24. Let ( M , g) be a compact, connected, oriented Riemann manifold and

s(;c) 6 C ° ° ( M ) . Assume volg(M) = l. Then the following two conditions are equivalent.
m l ; = [M s(a3)dVg > 0

( b ) For every A > 0 there exists a unique u = UA 6 cOO ( M l such that

Au -|- Ae" = s(:z:). (10.3.15)

Proof. (a) => (b) follows from Theorem 10.3.15.

(b) => (a) follows by multiplying (10.3. 15) with 'u(:c) 1, and then integrating by parts so
that

5 A e"'(as)dVg(a:) > 0.
M
502 Lectures on the Geometry of Manz]'old5

Although the above corollary may look like a purely academic result, it has a very nice
geometrical application. We will use it to prove a special case of the celebrated uniformiza-
tion theorem.

Definition 10.3.25. Let M be a smooth manifold. Two Riemann metrics 91 and 92 are said
to be conformal if there exists f 6 C°°(M), such that 92 = 6f91.

Exercise 10.3.26. Let ( M , g ) be an oriented Riemann manifold of dimension N and f E

COO(M). Denote by g the conformal metric g = egg. If s(ac) is the scalar curvature of g
and § is the scalar curvature of § show that

e -f
5(a3 >+ (n - II -
Agf
(n 1><n
4
2
)ld/(@1)lgo
2
7 (l0.3.l6)

where Ag denotes the scalar Laplacian of the metric g while I ' I9 denotes the length mea-
sured in the metric g.

Remark 10.3.27. A long standing problem in differential geometry, which was only rela-
tively recently solved, is the Yamabe problem:
"If ( M , g) is a compact oriented Riemann manifold, does there exist a metric conformal
to g whose scalar curvature is constant?"
In dimension 2 this problem is related to the uniformization problem in complex anal-
ysis.
The solution of the Yamabe problem in its complete generality is due to the combined
efforts of T. Aubine [10, I I ] and R. Schoen [1 $8]. For a very beautiful account of its proof
we recommend the excellent survey of J. Lee and T. Parker [85]. One can formulate a more
general question than the Yamabe problem.
Given a compact oriented Riemann manifold ( M , go decide whether a smoothjiinction
s(;z:) on M i5 the scalar curvature of some metric on M conformal to g.
This problem is known as the Kazdan-Warner problem. The case dim M = 2 is com-
pletely solved in [76]. The higher dimensional situation dim M > 2 is far more compli-
cated, both topologically and analytically.

Theorem 10.3.28 (Uniformization Theorem). Let ( g ) be a compact, oriented Rie-

mann manifold of dimension 2. Assume vo19(2) = l. If X(2) < 0 (or equivalently if
its genus is > 2 ) then there exists a unique metric § conformal to g such that

5(§) E -1.

Proof. We look for § of the form § = eu*g. Using Exercise 10.3.26 we deduce that u
should satisfy
-1 = e""{s(ac°l -|- Au},

i.e.,
Au -I- et -s(:z:),
 Elliptic Equations on Manifolds 503

where s(:1:) is the scalar curvature of the metric g. The Gauss-Bonnet theorem implies that
3 = 47rX(2) < 0
so that the existence of u is guaranteed by Corollary 10.3.24. The uniformization theorem
is proved.

On a manifold of dimension 2 the scalar curvature coincides up to a positive factor

with the sectional curvature. The uniformization theorem implies that the compact ori-
ented surfaces of negative Euler characteristic admit metrics of constant negative sectional
curvature. Now, using the Cartan-Hadamard theorem we deduce the following topological
consequence.
Corollary 10.3.29. The universal cover of compact, oriented surface 2 of negative Euler
characteristic is dqeomorphie to (C.

Remark 10.3.30. The "inverse" of the covering projection 'IT : (I -> 2 is classically called
a uniformizing parameter. It is very similar to the angular coordinate 9 on the circle S1.
This is a uniformizing parameter on the circle, which is an "inverse" of the universal cov-
ering maple 9 9 +-> et 6 S1.

In the following exercises (M,g) denotes a compact, oriented Riemann manifold with-
out boundary.
Exercise 10.3.31. Fix e > 0. Show that for every f 6 I;2(M) the equation
Au -|- cu = f
has an unique solution 'U G L2,2lMl_

Exercise 10.3.32. Consider a smooth function f : M X R -> R such that for every :IJ 6 M
the function u I-> f(H1» u) is increasing. Assume that the equation
Age = f(:1:,u) (10.3.17)
admits a pair of comparable sub/super-solutions, i.e., there exist functions up, U0 6
L1'2(M) VW LOO(M) such that
U0(;z:) > u0(;z:), a.e. on M,
and
Ag(/'0 > f(w7 U0(H:)) > f(a:,u0(;z:)) > _ Ague, weakly in L1,2(Ml.
Fix C > 0, and define IU/'nln21-C L2,2(Ml inductively as the unique solution of the
equation
Au,,,(ac°) -|- cun(.§c) = cu,,_1(:13) -|- f(ac°, 'uun_1(ac)).
(a) Show that
u0(:c) < u1(:z:) _< u2(.5c) < < uTL(.5c) < < U0(x) Va: 6 M.
(b) Show that uTL converges uniformly on M to a solution 'LL E COO(M) of (10.3.17) satis-
lying
U0 < 'LL < Uo.
(c) Prove that the above conclusions continue to hold, even if the monotonicity assumption
on f is dropped.
504 Lectures on the Geometry of Manzfolds

10.4 Elliptic operators on compact manifolds

The elliptic operators on compact manifolds behave in many respects as finite dimensional
operators. It is the goal of this last section to present the reader some fundamental analytic
facts which will transform the manipulation with such p.d.o. into a less painful task.
The main reason why these operators are so "friendly" is the existence of a priori
estimates. These estimates, coupled with the Rellich-Kondratchov compactness theorem,
are the keys which will open many doors.

10.4.1 Fredholm theory

Throughout this section we assume that the reader is familiar with some fundamental facts
about unbounded linear operators. We refer to [23] Ch. II for a very concise presentation
of these notions. An exhaustive presentation of this subject can be found in [75].

Definition 10.4.1. (a) Let X, Y be two Hilbert spaces over K R, (I, and
T:D(T)CX->Y
be a linear operator, not necessarily continuous, defined on the linear subspace D ( T ) C X .
The operator T is said to be densely deaned if its domain D ( T ) is dense in X .
The operator T is said to be closed if its graph,

FT {(@,To) G DID) YCX Y }7

..

X
II

is a closed subspace in X X Y .
(b) Let T : D ( T ) C X -> Y be a closed, densely defined linear operator. The adjoint of T
is the operator T* : D(T*) C Y -> X defined by its graph
FT* {(y*,;I:*) Q Y x X ; (:z:*,;/:l = (y*,T:z:) Va: 6 D(T),

.>
where (-, : Z X Z -> K denotes the inner product3 in a Hilbert space Z.
(c) A closed, densely defined operator T D ( T ) C X -> X is said to be self-adjoint if
T I T*.

Remark 10.4.2. (a) In more concrete terms, the operator T : D ( T ) C X -> Y is closed
if, for any sequence (izinl C D ( T ) , such that (in,Txn) -> (:z:,y), it follows that (i)
as 6 D(T), and (ii) y = T:1:.
(b) If T : X -> Y is a closed operator, then T is bounded (closed graph theorem). Also
note that if T : D(T) C X -> Y is a closed, densely defined operator, then kerT is a
closed subspace of X .
(c) One can show that the adjoint of any closed, densely defined operator, is a closed,
densely defined operator.
(d) The closed, densely defined operator T : D ( T ) C X -> X is self-adjoint if the
following two conditions hold.
3 When K (C, _we use the convention that the inner product ( - > is conjugate linear in the second variable,
i.e., (2:1,A2:2) Alzl, Z2), VA E C, 2 1 , 2 2 E Z.
 Elliptic Equations on Manifolds 505

(Taz, y) = (QC, T y ) for all :1:, y 6 D ( T ) and

D(T) I {y E X; EIC > 0, l(T11,y)l _< Cl:cl, 've:v€ D ( T ) } .
If only the first condition is satisfied the operator T is called symmetric.

Let (M, g) be a compact, oriented Riemann manifold, E, F two metric vector bundles
with compatible connections, and L 6 PDO" = PDO k(E, F), a k-th order elliptic
-
p.d.o. We will denote the various L2 norms by I II, and the Ll*'2-norms by I ilk. -
Definition 10.4.3. The analytical realization of L is the linear operator
La : D(La) C L2(E) -> L2(E), D(La) = L'2(E),
given by u »-> Lu for all u 6 L"'=2(E).

Proposition 10.4.4. (a) The analytical realization La of L is a closed, densely a'e]?ned

linear operator.
(b) IfL* : COO ( F ) -> cOO ( E ) is the formal adjoint o f L then

(Fla I (LJ
In particular, the analytical realization La is self-adjoint if and only if L is formally self-
adjoint.

Proof. (a) Since c°<)(E) C D (La) = Lk,2(E) is dense in I;2lE), we deduce that La is
densely defined. To prove that La is also closed, consider a sequence (un) C Lk,2 (E) such
that

Un -> Fu, strongly in L2(E), and Lun -> 'U strongly in L2(F).
From the elliptic estimates we deduce

IUn - 'Uflnllk S C( HI/Un - Lumll + - umm) -) 0 51S mm, ->oo.

Hence (un) is a Cauchy sequence in Lk,2(E), so that un -> u in Lk,2(E). It is now clear
that 'U = Lu.
(b) From the equality

(Lu, v)dVg (u, L*v)dVg , Vu G L'°»2lE), 'u Q L/"»2(F>,

M M

we deduce
* *
D( (La) ) 3 D ( ( L )a) - L
_ k,2
(F),
and (La)* = (L*)a on L'*'2(F). To prove that
D( (La>*) C D ( I m p ) = LWF)
we need to show that if 'U 6 L2 (F) is such that EIC > 0 with the property that

M
(Lu,v)dVg _< CIIuII 7 Vu 6 L'°'2(E),
506 Lectures on the Geometry of Manzfolds

then 'u 6 L":=2(F). Indeed, the above inequality shows that the functional

u »-> (Lu, v)dVg

M

extends to a continuous linear functional on I;2(E), Hence, there exists <l> 6 L2(E) such
that

( m m , 'U )dVg (Fu, ¢)dvg Vu E Lk»2(El.

M M

In other words, 'U is a L2-weak solution of the elliptic equation L*'u Using elliptic
regularity theory we deduce 'U 6 L'"'2(M). he proposition is proved.

Following the above result we will not make any notational distinction between an
elliptic operator (on a compact manifold) and its analytical realization.

Definition 10.4.5. (a) Let X and Y be two Hilbert spaces over K = R, (C and suppose that
T : D ( T l C X - > Y is a closed, densely defined linear operator. The operator T is said to
be semi-Fredholm if the following hold.

(i) dim kerT < oo and

(ii) The range R (T) of T is closed.

(b) The operator T is called Fredholm if both T and T* are semi-Fredholm. In this case,
the integer
ind T := diIIl]K her T - dim her T*
is called the Fredholm index o f T .

Remark 10.4.6. The above terminology has its origin in the work of Ivar Fredholm at
the twentieth century. His result, later considerably generalized by F. Riesz, states that
if K : H -> H is a compact operator from a Hilbert space to itself, then HH -|- K is a
Fredholm operator of index 0.

Consider again the elliptic operator L of Proposition 10.4.4.

Theorem 10.4.7. The operator La : D(Lal C L2(El -> L2lFl is Fredholm.

Proof. The Fredholm property is a consequence of the following compactness result.

Lemma 10.4.8. Any sequence (Unln20 C Lk,2lEl such that { flu,,ll -I- n>0
i5
bounded contains a subsequence strongly convergent in L2 (El.

Proof. Using elliptic estimates we deduce that

Ilunllk _< C(llLunll + mun) _< const.

Hence (un) is also bounded in L k , 2 (E). On the other hand, since M is compact, the space
Ll*'2(E) embeds compactly in L2(E). The lemma is proved.
 Elliptic Equations on Manifolds 507

We will first show that dim her L < oo. In the proof we will rely on the classical result
of F. Riesz which states that a Banach space is finite dimensional if and only if its bounded
subsets are precompact (see [23], Chap. VI).
Note first that, according to Weyl's lemma her L C C°°(E). Next, notice that her L
is a Banach space with respect to the L2-norm since according to Remark 10.4.2 (a) her L
is closed in L2lEl. We will show that any sequence (uTL) C kerL which is also bounded
in the L2-norm contains a subsequence convergent in L2. This follows immediately from
Lemma 10.4.8 since ll'u.nll -|- ||Lu,,|| = ll'u.,,ll is bounded.
To prove that the range R ( T ) is closed, we will rely on the following very useful
inequality, a special case of which we have seen at work in Subsection 9.3.3.

Lemma 10.4.9 (Poincaré inequality). There exists C > 0 such that

IIuII <_ CIILull»

for all 'LL G Lk,2 ( E ) which are L2-orthogonal to her L, i.e.,

(U, (z5)dvg 0 We) E her L.

M

Proof. We will argue by contradiction. Denote by X C Lk,2lE) the subspace consisting

of sections L2-orthogonal to her L. Assume that for any n > 0 there exists un Q X such
that

Ilunll = 1 and lILunll g 1/n.

Thus, lILy,,ll -> 0, and in particular, ll7ufnll -|- ||Lu,,|| is bounded. Using Lemma 10.4.8 we
deduce that a subsequence of (un) is convergent in L2(El to some u. Note that fluff = 1.
It is not difficult to see that in fact u 6 X. We get a sequence

U n ) Lim C FL = the graph of L,

such that (un, Lung -> (u, Of. Since L is closed, we deduce u 6 D(L) and Lu = 0. Hence
u E her L VW X = {0}. This contradicts the condition fluff = 1.

We can now conclude the proof of Theorem 10.4.7. Consider a sequence (vol C R(L)
such that on -> U in I;2(Fl. We want to show 'u E R (L).
For each on we can find a unique u'n 6 X = (her L)J_ such that
L'LLTL 'Un-

Using the Poincare inequality we deduce

lun - um I < Cll'U'n 'Um

When we couple this inequality with the elliptic estimates we get

lun - umllk1.I S C(llLun - Lumll -|- ||un - u,,,ll) < Cllvn 'Um II -> 0 as m,n -> oo.
Hence (un) is a Cauchy sequence in Ll*'2(E) so that un -> 'LL in L k , 2 (E). Clearly Lu "U,
so that 'U E R(L ).
508 Lectures on the Geometry of ManQ'olds

We have so far proved that her L is finite dimensional, and R (L) is closed, i.e., La is
semi-Fredholm. Since (La)* = (L*)0, and L* is also an elliptic operator, we deduce (La)*
is also semi-Fredholm. This completes the proof of Theorem 10.4.7.

Using the closed range theorem of functional analysis we deduce the following impor-
tant consequence.

Corollary 10.4.10 (Abstract Hodge decomposition). Any k-zh order elliptic operator L :
cOO ( E ) -> cOO ( F ) over the compact manifold M defines natural orthogonal decomposi-

tions ofL2(E) and L2lFl,

L2(E) = her L QUO R (L*) and L2lFl = kerL* ® R(L).
.
Corollary 10 4.11. If her L* = 0, then for every 'U Q L2 ( F ) the partial a'Qj'erential equa-
tion Lu = 'U admits at least one weak L2-solution u 6 L2 (E).

Corollary 10.4.12. Suppose that L is additionally formally self-adjoint. If k e r L = 0,

then for any j E Z20, the induced linear map Lk-I-j,2lEl -> LJ»2(F) is a topological
isomorphism.

The last corollary is really unusual. It states the equation Lu = 'U has a solution pro-
vided the dual equation L*v = 0 has no nontrivial solution. A nonexistence hypothesis
implies an existence result! This partially explains the importance of the vanishing results
in geometry, i.e., the results to the effect that her L* = 0. With an existence result in our
hands presumably we are more capable of producing geometric objects. In the next chapter
we will describe one powerful technique of producing vanishing theorems based on the so
called Weitzenbéck identities.

Corollary 10.4.13. Over a compact manifold

her L = her L*L and her L* her LL*.

Proof. Clearly her L C her L*L. Conversely, let W Q C°°(E) such that L*L1b = 0. Then
2
lIe¢ll (Lab/l¢)dVg I/;*L(p,zpldvg = 0.
M M
The Fredholm property of an elliptic operator has very deep topological ramifications
culminating with one of the most beautiful results in mathematics: the Atiyah-Singer in-
dex theorems. Unfortunately, this would require a lot more extra work to include it here.
However, in the remaining part of this subsection we will try to unveil some of the natural
beauty of elliptic operators. We will show that the index of an elliptic operator has many
of the attributes of a topological invariant.
We stick to the notations used so far. Denote by Ell klEvFl the space of elliptic
operators cOO (E) -> cOO (F) of order k. By using the attribute space when referring to
Elli we implicitly suggested that it carries some structure. It is not a vector space, it is not
an affine space, it is not even a convex set. It is only a cone in the linear space PDO( m )
but it carries a natural topology which we now proceed to describe.
 Elliptic Equations on Manifolds 509

L€tL1, L2 Q Ell k(E7 F ) . We set

6(L1, LQ) = sup{llLlu - L2'UII3 IIuII k 1}

Define
dll 1, LQ) I 1'Il3JX {(S(L1, LQ), 6(L1, L2l} .
We let the reader check that ( Ell k , d) is indeed a metric space. A continuous family of
elliptic operators ILAIAQA, where A is a topological space, is then a continuous map
A E } I - > L ) € Ell k
Roughly speaking, this means that the coefficients of LA, and their derivatives up to order
/Q depend continuously upon A.

Theorem 10.4.14. The index map

ind : Ell 1<(E»Fl ->Z, L +-> ind (L)
is continuous.

The proof relies on a very simple algebraic trick which requires some analytical foun-
dation.
Let X, Y be two Hilbert spaces. For any Fredholm operator L : D(L) C X -> Y
denote by 'LL : kerL -> X (respectively by PL : X -> her L) the natural inclusion
kerL --> X (respectively the orthogonal projection X -> her L). If L : D(LZ~) C X -> Y
(i = 0, 1) are two Fredholm operators define

72Lo(L1) : D(L1) ® kerL0 C X ® kerL0 -> Y ® kerL0

by
72L0(L1)(u, (be = (LW + ZL5€25,PL0Ul, u Q D(L1l, QUO Q her L0-
In other words, the operator RL0 (Ll) is given by the block decomposition

/1 ZL5
RIG (Ll)

We will call RL0 (Ll) is the regularization at Lo. The operator L0 is called the center
o_fL1

of the regularization. For simplicity, when L0 = L1 = L, we write

RL RLILI.
The result below lists the main properties of the regularization.

Lemma 10.4.15. (a) RL0 (Ll) is a Fredholm operator:

(b) R20(1;1) RL8 (LT)-
I

(c) RL0 is invertible (with bounded inverse).

Exercise 10.4.16. Prove the above lemma.

510 Lectures on the Geometry of Manzfolds

We strongly recommend the reader who feels less comfortable with basic arguments of
functional analysis to try to provide the no-surprise proof of the above result. It is a very
good "routine booster".
Proof of Theorem 10.4.14. Let Lo E Ell We have to find 1 ° > 0 such that,
VL Q Ell k(E, F) satisfying d(L0, L) < r, we have
ind (L) = ind (LQ).
We will achieve this in two steps.
Step 1. We will find 1°> 0 such that, for any L satisfying d(L0,L) < T , the regularization
of L at L0 is invertible, with bounded inverse.
Step 2. We will conclude that if d(L, Lo) < r, where r > 0 is determined at Step 1, then
ind aLl = ind (Lo).
Step 1. Since RL0 (L) is Fredholm, it suffices to show that both RLoILI, and 'R,Lo*lL*l
are injective, if L is sufficiently close to Lo. We will do this only for RL0 (L), since the
remaining case is entirely similar.
We argue by contradiction. Assume that there exists a sequence (un, (in) C L""2(E) X
her Lo, and a sequence (Ln) C Ell k(E, F) such that

Ilunllk + llcbnll 1, (10.4.1)

Rho (Le)(un» (in) = 10701, (10.4.2)

and
(10.4.3)

From (l0.4.l) we deduce that l¢nl is a bounded sequence in thejinite dimensional space
her Lo- Hence it contains a subsequence strongly convergent in L2, and in fact in any
Sobolev norm. Set $5 := lim (in. Note that llg2§ll = limn II¢nll. Using (l0.4.2) we deduce

Lour -0%7
i.e., the sequence (Lour) is strongly convergent to QUO in I;2(F). The condition (10.4.3)
now gives
lIL0un1 L'nunll < 1/n,
i.e.,
lim Liun = lim Loun
TL TL
= -<1s G L2(F).
Since un J_ her L0 (by (l0.4.2)) we deduce from the Poincaré inequality combined with

l'LLn _
the elliptic estimates that
Uvrlilk <_ C IIL0 Un _ I,o
uTfall _>
0
as M y ->
72 OO.

Hence the sequence 'Un strongly converges in L k , 2 to some u. Moreover,

lim Ilunllk I Ilullk and Ilullk -|- ||¢|| I 1.
TL
 Elliptic Equations on Manifolds 511

Putting all the above together we conclude that there exists a pair (we) E Lk,2lEl X

her L0» such that

Ilullk + II¢ll I 1,

Lou <l> and u J_ her L. (10.4.4)

This contradicts the abstract Hodge decomposition which coupled with (10.4.4) implies
'LL = 0 and <l> = 0. Step 1 is completed.

Step 2. Let T' > 0 as determined at Step 1, and L Q Ell ilE,Fl. Hence

RIG (L)

is invertible. We will use the convertibility of this operator to produce an injective operator
her L* ® her Lo --> kerL ® her L0.
This implies dim her L* -|- dim her L0 < dim her L + dim her Lo, i.e.,
ind (Low < ind(L).
A dual argument, with L replaced by L*, and L0 replaced by Lo, will produce the opposite
inequality, and thus finish the proof of Theorem 10.4. 14. Now let us provide the details.
First, we orthogonally decompose
IPIEI = (her L ) l ® her L and HIFI = (her L*>i ® her L*.
Set U := her L ®ker Lo, and V := her L* éBker L0. We will regard RL0 (L) as an operator
RL0 (L) : (her L)l ® U -> (her L*l1 ® V
As such, it has a block decomposition
A
72L0(L) = 7
C'
where
T : Lk'2(E) O (ketL)l C (kerL) -> (her L*l-L = Range (L)
denotes the restriction of L to (her L)_L. The operator T is invertible and its inverse is
bounded.
Since 72L0(L) is invertible, for any 'U E V we can find a unique pair (aW) Q
(kerlll ® U, such that

This means
RLoIL) <2)
H [0]
'U

TQ5-I-Au = 0 and BUS -|- Cu

We can view both (15 and u as (linear) functions of u, <2) (z5('vl and 'LL = u(v). We claim
the map 'U »-> 'LL = u(v) is injective.
512 Lectures on the Geometry of Manzfolds

Indeed, if u(v) = 0 for some "u, then Tqzé = 0, and since T is injective <2) must be zero.
From the equality 'U == BQ5 -|- Cu we deduce 'U = 0. We have thus produced the promised
injective map V --> U. Theorem 10.4.14 is proved.

The theorem we have just proved has many topological consequences. We mention
only one of them.

Corollary 10.4.17. Let Lo, L1 Q Ell k ( E , F ) Q" 0k1.r(L0) = 0k(L1) then ind(L0)
ind lLll.

Proof. For every t 6 [0, 1] Lt (1 - t)L0 + tL1 is a k-th order elliptic operator de-
pending continuously on t. (Look at the symbols.) Thus ind (Lt) is an integer depending
continuously on t so it must be independent of t.

This corollary allows us to interpret the index as as a continuous map from the elliptic
symbols to the integers. The analysis has vanished! This is (almost) a purely algebraic-
topologic object. There is one (major) difficulty. These symbols are "polynomials with
coefficients in some spaces of endomolphism".
The deformation invariance of the index provides a very powerful method for com-
puting it by deforming a "complicated" situation to a "simpler" one. Unfortunately, our
deformation freedom is severely limited by the "polynomial" character of the symbols.
There aren't that many polynomials around. Two polynomial-like elliptic symbols may be
homotopic in a larger classes of symbols which are only positively homogeneous along the
fibers of the cotangent bundle) .
At this point one should return to analysis, and try to conceive some operators that be-
have very much like elliptic p.d.o. and have more general symbols. Such objects exist, and
are called pseudo-dy erential operators. We refer to [83, 121] for a very efficient presen-
tation of this subject. We will not follow this path, but we believe the reader who reached
this point can complete this journey alone.

Exercise 10.4.18. Let L Q Ell k(E, F). A finite dimensional subspace V C I;2(Fl is
called a stabilizer of L if the operator
SL , V :Lk'2(E) ® V -> L2(F) SL,V(U ® u) = Lu + 'U
is surjective.
(a) Show that any subspace V C L2(F) containing her L* is a stabilizer of L. More gener-
ally, any finite dimensional subspace of I;2(F) containing a stabilizer is itself a stabilizer.
Conclude that if V is a stabilizer, then
ind L dim k€I` S L y dim V.

Exercise 10.4.19. Consider a compact manifold A and L : A -> Ell ME, F) a continuous
family of elliptic operators.
(a) Show that the family L admits an unQ'orm stabilizer, i.e., there exists a finite dimensional
subspace V C L2 (F), such that V is a stabilizer of each operator LA in the family L.
 Elliptic Equations on Manifolds 513

(b) Show that if V is an uniform stabilizer of the family L, then the family of subspaces
k€IISL ,V defines a vector bundle over A.
(c) Show that if VI and VI are two uniform stabilizers of the family L, then we have a
natural isomorphism vector bundles
her SLy1 ® _2
r\/
her $L,v2 G E .
In particular, we have an isomoiphism of line bundles
det her SL »VI (X)
det Vi* 2 det her SL,v2 (X) det v;
Thus the line bundle det her SL,v (X) det V* -> A is independent of the uniform stabilizer
V. It is called the determinant line bundle ofthefamily L and is denoted by det ind (Ll-

10.4.2 Spectral theory

We mentioned at the beginning of this section that the elliptic operators on compact man-
ifold behave very much like matrices. Perhaps nothing illustrates this feature better than
their remarkable spectral properties. This is the subject we want to address in this subsec-
tion.

..
Consider as usual, a compact, oriented Riemann manifold (M, g), and a complex vec-
tor bundle E -> M endowed with a Hermitian metric ( , ) and compatible connection.
Throughout this subsection L will denote a k-th order, formally self-adjoint elliptic opera-
tor L : COO(E) -> C°°(E). Its analytical realization
La : Lk,2lE) C L2(El -> L2(E),
is a self-adjoint, elliptic operator, so its spectrum spec(L) is an unbounded closed subset of
R Note that for any A 6 R the operator A - La is the analytical realization of the elliptic
p.d.o. All E - L, and in particular, the operator A - La is a Fredholm, so that
A G spec:(L) I
I
\
\ :
) ker(A L)¢0.
Thus, the spectrum of L consists only of eigenvalues of finite multiplicities. The main
result of this subsection states that one can find an orthonormal basis of L2(El which
diagonalizes La.
Theorem 10.4.20. Let L 6 E11k ( E I be a formally self-adjoint elliptic operator: Then the
following are true.
(a) The spectrum spec(L) is real, spec(Ll C R, and for each A E spec(Ll, the subspace
ker(A - Ll is jinite dimensional and consists of smooth sections.
(b) The spectrum spec(Ll is a closed, countable, discrete, unbounded set.
(c) There exists an orthogonal decomposition

L2(El I
ker(A - L) .
A€spec(L)

(d) Denote by P the orthogonal projection onto ker(A - L). Then

Lk,2(E) 2
D(La) w€L2(E), A2llp¢ll <oo
A
514 Lectures on the Geometry of Manzfolds

Part (c) of this theorem allows one to write

]1 = Z PA and L AP .
A A

The first identity is true over the entire L2lEl while part (d) of the theorem shows the
domain of validity of the second equality is precisely the domain of L.

Proof. (a) We only need to show that ker(A - L) consists of smooth sections. In view of
Weyl's lemma this is certainly the case since A - L is an elliptic operator.
(b) and (c) We first show spec(L) is discrete.
More precisely, given A0 Q spec(L), we will find 5 > 0 such that ker(A - L) 0,
VIA - A0l, 5, A go Ao- Assume for simplicity Ao = 0.
We will argue by contradiction. Thus there exist An -> 0 and un Q GOO (E), such that
Lun _ Anon, llunll = 1.
Clearly un 6 R (L) = (kerL*)l = (her L)_L, so that the Poincaré inequality implies

1 = Hen I < CIILunll = CAn -> 0.

Thus spec(L) must be a discrete set.
Now consider to € ] R \ 0 ( L l . Thus t g - L has a bounded inverse

T = lt0-Ll
Obviously T is a self-adjoint operator. We claim that T is also a compact operator.
Assume (fun) is a bounded sequence in L2(E). We have to show un = Tun admits a
subsequence which converges in L2 (E). Note that un is a solution of the partial differential
equation
lt() - L)'u,,, t02U#7n Lun un .
Using the elliptic estimates we deduce

llunllk <_ C(llunll + ||vn||)-

Obviously un = Ton is bounded in L2(El, so the above inequality implies flu,,llk, is
also bounded. The desired conclusion follows from the compactness of the embedding
Lk»2(E) -> L2(El.
Thus T is a compact, self-adjoint operator. We can now use the spectral theory of such
well behaved operators as described for example in [23], Chap. 6. The spectrum of T is a
closed, bounded, countable set with one accumulation point, ,LL = 0. Any ,u 6 spec(T)\{0}
is an eigenvalue of T with finite multiplicity, and since her T = 0,
L2lE) = her(/ - T).
I»6 sp€<:(T)\{0}

Using the equality L to - T-1 we deduce

,u 1
0(L) I {t0 ,u E spec(T) \ {0}}-
This proves (b) and (c).
 Elliptic Equations on Manifolds 515

To prove (d), note that if QL E Lk,2lEl, then Lw E L2(E), i.e.,

II APA¢ll2 I 2IIPAIPII2 < oo.
A A
Conversely, if
2
2IIPA¢ll <oo,
A
consider the sequence of smooth sections
(in Aw
I>\IS'f1
which converges in L2lE) to

<15 Aw.
A
On the other hand, (in Lin, where
1% Z PM
IAISH
converges in L2lEl to in. Using the elliptic estimates we deduce
Ilwn Wmllk <_ C(ll¢n Will + m cbmlll -> 0 as n , m -> OO.
Hence Tb 6 Lk,2lE) as a Lk'2-limit of smooth sections. The theorem is proved.

Example 10.4.21. Let M = Sl, E = QM and

a
L = - ' - : c°°(51,(c> -> 0°°(s1,(c>.
'ae
The operator L is clearly a formally self-adjoint elliptic p.d.o. The eigenvalues and the
eigenvectors of L are determined from the periodic boundary value problem

Au, 7u,(0) = u(27r),

which implies
u(0l = C'exp(iA0) and exp(21rAil = 1.
Hence
spec(L) = Z and ker(n - L) span@{exp(in9)}.
The orthogonal decomposition
LQISII ker(n - L)
n

is the usual Fourier decomposition of periodic functions. Note that

71(0) Un €Xp(in9) 6 L1)2(51)
TL

if and only if
(1 -|- n2)|un|2 < oo.
1162
516 Lectures on the Geometry of ManQ'olds

The following exercises provide a variational description of the eigenvalues of a for-

mally self-adjoint elliptic operator L Q E11k (E) which is bounded from below, i.e.,

inf
{/M
(Lu,u)d»g; u Q L E ) , Hun 1

Exercise 10.4.22. Let V 6 LL¢»2(E) be a finite dimensional invariant subspace of L, i.e.,

L(V) C V. Show that

(a) The space V consists only of smooth sections.

(b) The quantity

Aly_Ll inf
{/ M
(L'u,, u)dtug; u 6 Lk,2lE) re v Hun 1

is an eigenvalue of L. (V denotes the orthogonal complement of V in L2lE)

Exercise 10.4.23. Set V0 = 0, and denote A0 = ,\(V0J_). According to the previous exercise
A0 is an eigenvalue of L. Pick (be an eigenvector corresponding to A0 such that lld)0 I = 1
and form VI = Vi GB span {gz50}. Set $1 = Alvl_Ll and iterate the procedure. After m
steps we have produced m -|- l vectors QSo, QS1, . . . ,(/)m corresponding to m -|- l eigenvalues
A0,/\1 S ' • ' Am ofL. Set
Vm+1 spanc{g250, (Zh, • • • 7 (1)m}

and Am+1 == A(v£+1) etc.

(a) Prove that the set

{¢1,I.I,¢m,---}
is a Hilbert basis o f L2(18), and
spec(L) ={/\1 S ° ° ° <
_ Am . . . } .

(b) Denote by Gr m the Grassmannian of m-dimensional subspaces of L'">2(El. Show

that

,\m inf
V€Grm
max
{/ M
(Lu,u)d@g; u V nun 1

Exercise 10.4.24. Use the results in the above exercises to show that if L is a bounded from
below, k-th order formally self-adjoint elliptic p.d.o. over an N-dimensional manifold then
Am(L) = oink/nl as m -> oo,
and
d<A> = dim 69A<_A ker(A - Ll = o(AN/'<) as A -> oo.
 Elliptic Equations on Manifolds 517

Remark 10.4.25. (a) When L is a formally self-adjoint generalized Laplacian then the
result in the above exercise can be considerably sharpened. More precisely H. Weyl showed
that

lim A'N/2d(A)
-
rank (E) volg(M)
(10.4.5)
A->oo (4 f)N/2F(n/2 + 1) '
The very ingenious proof of this result relies on another famous p.d.o., namely the heat
operator 315 -|- L on R X M. For details we refer to [14], or [l2l].
In the next chapter we will investigate the spectrum of the Laplacian A of a compact
Riemann manifold using another famous p.d.o., the wave operator = 3? -|- on X M.
In particular, we will obtain a refinement of (l0.4.5).
(b) Assume L is the scalar Laplacian A on a compact Riemann manifold (M, g) of
dimension M. Weyl's formula shows that the asymptotic behavior of the spectrum of
A contains several geometric informations about M: we can read the dimension and the
volume of M from it. If we think of M as the elastic membrane of a drum then the
eigenvalues of A describe all the frequencies of the sounds the "drum" M can produce.
Thus "we can hear" the dimension and the volume of a drum. This is a special case of
a famous question raised by V. Kac in [71]: can one hear the shape of a drum? In more
rigorous terms this question asks how much of the geometry of a Riemann manifold can be
recovered from the spectrum of its Laplacian. This is what spectral geometry is all about.
It has been established recently that the answer to Kac's original question is negative.
We refer to [53] and the references therein for more details. In the next chapter we can
reconstruct the geometry of the manifold if besides the spectrum we also have information
about the eigenfunctions of the Laplace operator.

Exercise 10.4.26. Compute the spectrum of the scalar Laplacian on the torus T2 equipped
with the flat metric, and then use this information to prove the above Weyl asymptotic
formula in this special case.

Exercise 10.4.27. Denote by Asn-1 the scalar Laplacian on the unit sphere S"-1 C ] R n
equipped with the induced metric. We assume n > _ 2. A polynomial p = Plxly ' "HIs
called harmonic if AwP = 0.
(a) Show that if p is a homogenous harmonic polynomial of degree k, and 15 denotes its
restriction to the unit sphere S n - 1 centered at the origin, then

AS-115 = n(n -|- k - 2)}5.

(b) Show that a function up : Sn -> IR is an eigenfunction of ASn,-1 if and only if there
exists a homogeneous harmonic polynomial p such that go = p,
(c) Set As, := n(n -|- lc - 2). Show that

dim her (As A Sn 1

k-I-n
n 'l (
k+n 3
n 2
(d) Prove Weyl's asymptotic estimate (10.4.5) in the special case of the operator A Sn 1.
518 Lectures on the Geometry of Manzfolds

10.4.3 Hodge theory

We now have enough theoretical background to discuss the celebrated Hodge theorem. It
is convenient to work in a slightly more general context than Hodge's original theorem.

Definition 10.4.28. Let ( M , g ) be an oriented Riemann manifold. An elliptic complex is a

sequence of first order p.d.o.'s
D Dm 1
0 -> 0<><>(E0) c°°(El> 9% -> c°°(E,,,> -> 0
satisfying the following conditions.

(i) (C°°(E), DZ-) is a cochain complex, i.e., D¢D¢-1 = 0, VI < i < m .

(ii) For each (co, go Q T*M \ {0} the sequence of principal symbols
0(D0)(> ,€) o( 'rn_
3(x,€) (Et
0-> (Eolx D 1 -> 0
is exact.

Example 10.4.29. The DeRham complex (Q* ( M ) , d) is an elliptic complex. In this case,
the associated sequence of principal symbols is (e(§) = exterior multiplication by go

0 -> R eos T;M e(q eos set(T;m) -> 0

is the Koszul complex of Exercise 7.1.23 of Subsection 7.1.3 where it is shown to be exact.
Hence, the DeRham complex is elliptic. We will have the occasion to discuss another
famous elliptic complex in the next chapter.

Consider an elliptic complex (COO (E, D. )) over a compact oriented Riemann manifold
(M, g). Denote its cohomology by H' ( E , D.l. A priori these may be infinite dimensional
spaces. We will see that the combination ellipticity + compactness prevents this from hap-
pening. Endow each E with a metric and compatible connection. We can now talk about
Sobolev spaces and formal adjoints D: . Form the operators

AS D - D + D¢_1D-_1 : COOlEd) -> COOlE@l.

We can now state and prove the celebrated Hodge theorem.

Theorem 10.4.30 (Hodge). Assume that M is compact and oriented. Then the following
are true.

(i) H{(E.,D.) E kerr C C°°(E¢)! Vi.

(ii) dim H¢(E., D.) < oo w.
(iii) ( Hodge decomposition). There exists an orthogonal decomposition

L2(Eil = ke1~A,- Q9 R (DH) ® R ID. I,

where we view both D1-1, and Di as bounded operators L1,2 -> L2.
 Elliptic Equations on Manifolds 519

Proof. Set
D* *7 A = ,D=D+D*
E 65E2 7 D ® D?7 QUO Di ®

Thus D, D*, A E PDO ( E , E). Since D D _ 1 0, we deduce D2 (D*l2 = 0 which

implies
A I D*D+DD* I (D +D *)2 I DQ.
We now invoke the following elementary algebraic fact which is a consequence of the
exactness of the symbol sequence.

Exercise 10.4.31. The operators D and A are elliptic, formally self-adjoint p.d.o. (Hint:
Use Exercise 7. 1.22 in Subsection 7.1.3.)

Note that according to Corollary 10.4.13 we hav her A her D, so that we have an
orthogonal decomposition

L2<E) = kerA Q; R (D). (10.4.6)

This is precisely part (c) of Hodge's theorem.

For each 'i denote by Pt the orthogonal projection L2(Ei) -> her At. Set
r u e c°°(E); D o , = 0 } and 1% Di-llCOOlEi-lllv
L

so that
Hi(E., D ) = Z / B
We claim that the map Pi : Zi -> her As descends to an isomorphism H E . , D . ) ->
her As. This will complete the proof of Hodge theorem. The above claim is a consequence
of several simple facts.
Fact 1. her At C Zn'_ This follows from the equality her A = her D.
Fact 2. If u GZz- then u - Pll, 6 B . Indeed, using the decomposition (l0.4.6) we have

u p,-fu, -|- Etp to 6 LE<E>.

Since u - Pll, 6 C'°°(E), we deduce from Weyl's lemma that in 6 C°°(El. Thus, there
exist 'U 6 COO(EZ~_1) and w 6 C°°(E¢) such that = 'U ® w and w
u = Pn -|- Di_1'U -|- D- w.
Applying Di on both sides of this equality we get
0 I D* w
Since her D'i her Dz~D, , the above equalities imply
'LL - PL I Di_1'U E Bi

We conclude that Pi descends to a linear map her As -> H'i (E, D. ) Thus
Bi C RID) = (ketA)l,
520 Lectures on the Geometry of Manzfolds

and we deduce that no two distinct elements in her As are cohomologous since otherwise
their difference would have been orthogonal to her As. Hence, the induced linear map
Pz- : ker As -> H¢(E. D.) is injective. Fact 1 shows it is also surjective so that
k€IIAz' -N H l E . , D.
Hodge theorem is proved.

Let us apply Hodge theorem to the DeRham complex on a compact oriented Riemann
manifold
d d
0 ->Q0(M) > Qllml >--- d>Qn(M)->0, n dimer.
We know that the formal adjoint of d : SY* ( M ) ->§2/<:+1(M)is
d* l-1llft , k * d*,
Z

where I/n,k = no + n + 1, and * denotes the Hodge *-operator defined by the Riemann
metric g and the fixed orientation on M . Set
A dd* + d*d.

Corollary 10.4.32 (Hodge). Any smooth k-form cu E Up ( M ) decomposes uniquely os

w= [Wig + it? + d*g, 17 G o-1(m>, b e o'<=+1(m),

and [w]g E Q'<lml is g-harmonic, i.e.,

A[w]g = 0 ii > d[w]g = 0 and d*[oJ]g = 0.

The form [au]g is called the g-harmonic part of w.
If moreover Ll.) is closed, then d*§ = 0, and this means any cohomology class [z] 6
H k ( M ) is represented by a unique harmonic fc-form.

Denote by H k(m,gl the space of g-harmonic k-forms on M. The above corollary

shows
H'"(m,9) H'"(m)
for any metric g.

Corollary 10.4.33. The Hodge *-operator defines a bisection

* H k ( m , 9) -> H n - k ( m , 9)-
Proof. If Ld is g-harmonic, then so is *CU since
d * w = d: * d * w = 0 ,
and
* d * (WI = :|; * (do) = 0.
The Hodge *-operator is bijective since *2 = l-1lk(n-k) .
 Elliptic Equations on Manifolds 521

Using the L2-inner product on Q°(M) we can identify H n-k(]II,gl with its dual, and
thus we can view * as an isomorphism
k * ( n-kl*
H -> H
On the other hand, the Poincaré duality described in Chapter 7 induces another isomor-
phism

> l H n-kl*
PD
Hk 7

defined by

(PDlWl»'7)0 Ld As,
M

where (~, .)0 denotes the natural pairing between a vector space and its dual.

Proposition 10.4.34. PD *Q i.e.,

l*w, nlgdfvg wA17'v'w€ H k 77 Q Hn o

M M

Proof. We have

= 77 * LU (-1)"'(""">u
2
(*w,?7)£dv9 = (U, *w)gdvg w=w 17.

Exercise 10.4.35. Let We E Qk (M) be a harmonic k form and denote by 0,0 its cohomol-
ogy class. Show that

/M
lw0l§dv9 <
with equality if and only i f s = W0»
/
M
loJldvg Vi G CwD,

Exercise 10.4.36. Let G denote a compact connected Lie group equipped with a bi-
invariant Riemann metric h. Prove that a differential form on G is h-harmonic if and
only if it is bi-invariant.
This page intentionally left blank
 Chapter 11

Spectral Geometry

On a compact Riemann manifold the Laplacian produces a wealth of data: its spectrum and
the corresponding eigenfunctions. We will refer to these as the spectral data of a Riemann
manifold. Clearly the spectral data depend on the metric. In this chapter we discuss how
much information about the metric we can extract from the spectral data. In approaching
this problem we rely on a strategy pioneered by L. HOrmander [65] and based on the wave
equation.

11.1 Generalized functions and currents

We will rely heavily on the theory of distributions or generalized functions. The goal of this
section is to describe, mostly without proofs, several basic facts and, in greater detail several
more exotic the aspects of this theory. For more details we refer [39, 49, 50, 66, 118, 127]
which were our main sources of inspiration.

11.1.1 Generalizedfunetions and operations withy them

For any open set 6° C R" and any compact set K C 6 we denote by QKWI the space of
complex valued smooth functions on e> whose support is contained in K. We set
9(6) U @K(@'),
Kg@

We will refer to the functions in @(@) as (compactly supported) test functions.

Definition 11.1.1. A generalized function or distribution on 6" is a linear functional

@(@) -> (C, 2(6) 9 go »-> Lu,c,o),
satisfying the following continuity property: for any compact set K C @ there exist m
my 6 ZZ() and C = CK > 0 such that, for any function up 6 @K(@°)

l(u,(0) I < C 2; sup l3/,(w)l.

QUGK
(11.1.1)
IO¢IS'fTlK

We denote by Q' (W) or c-oo (6) the space of generalized functions on 6.

A generalized function u is said to have order < m if in (11.1.1) we can chose my < m
for any compact K CW.

523
524 Lectures on the Geometry of Manzfolds

Remark 11.1.2. Technically, a generalized function is really a linear functional on the

space of smooth densities on 6 with compact support. However, the open set 6 is equipped
with a canonical volume density defined by the Euclidean metric and this allows us to
identify the space smooth densities with compact support with n).
This distinction becomes very relevant when working on manifolds where there is no
canonical volume density. We refer to [56] for an approach to generalized functions that
takes consistently into account this aspect.

Example 11.1.3. (a) Any complex valued measure on the 0-algebra of Borel subset of
6" defines a generalized function on 61 In particular LgOClgl C C"°°(@). A function
f 6 L3oc(@') defines a distribution Uf by setting

(of» (Q) 6° f(x;)9,0(9;ld, WE 6 9(6

o n

(b) For any nonzero multi-index a 6 21;0 a nYpgintm 0 6 @ the linear map
9(6) 9 up »-> 6g¢(:/30) 6 (I

is a generalized function. When lal > 1 this generalized function is not determined by any
complex measure on @. When a = 0 this is the Dirac function (measure) concentrated at
and and it is denoted by 6$0 .

Definition 11.1.4. Let 6' be an open subset of R". A sequence of generalized functions
Up 6 C"°°(@), k 2 0, is said to converge weakly or in the sense of distributions to
u 6C"°0(6') if

lim
k->c>o( k
'LL 7
(p) (up), W G m.
The above definition can be strengthened by the following nontrivial result. For a proof
we refer to [50, Sec. I.8] or [127, Sec. 33].

Theorem 11.1.5 (Uniform boundedness principle). Let 6° be an open subset ofR". Sup-
pose that UP 6 C"°°(@'l, j > 0, i5 a sequence of generalized functions such that,
We 6 @(@l the sequence
(U (p)
has a finite limit Llcp) E C Then the following hold.

(1) The correspondence @(@) 9 go »-> L((,ol 6 (C i5 continuous in the sense ofDefini-
tion I1.1.1 and thus defines a generalized function.
(2) If K C @ i5 a compact set and (got) 6 Q K W ) is a sequence of smooth functions
converging uniformly togethere with all their derivatives to a smooth function go 6
%<(@)» then
lim (UMW) = L((/))~
j->oo
 Spectral Geometry 525

Definition 11.1.6. Any p.d.o. L 9(6) -> 2(6) extends to a linear operator

L C1-°°(@) -> C1-°°(@),

according to the rule

( L u , 95) (u,L*(p), Vu 6 C-°°(@'), go 6 @(@),

..II
where go denotes the complex conjugate of up. In particular, for any multi-index a 6 Z">0
and any u 6 C-00(6) the distributional derivative d a y of u is determined by the equality

(<9"u7(0) == (-1)'""( Up aw), we m.

Example 11.1.7. Let 60 6 C-°°(Rn) be the Dirac function concentrated at 0. Then

(aw, (Q) = (-1)a3"<0(0).

Consider the Heaviside function

l, :z:>0,
H R -> R, H (113)
0, :z:<0.
Being a locally integrable function, H defines a generalized function on R Moreover
d
-H 60.
do:
Consider the continuous function f : ] R - > ] R

ac, a2>0,
f(wl
0, a2<0.

Then QL
do
H as generalized functions on R

If 60,61 are open subsets of R", such that 60 C @1, then we have an extension-by-
zero inclusion This induces a restriction map

c" 'w 9 u »-> up Q C-°°(@'0)7

<"~"60v90> = ( u ) , Vu 6 Qlgo) C @(@1l.

Definition 11.1.8. Let u 6 c-<)°(@l, @° C R".

(i) The support of u, denoted by supp u, is the complement in 6 of the largest open subset
Q C 6 such that u s - ".
(ii) The singular support of u, denoted by sing supp u, is the complement in 6 of the
largest open set Q C 6 such that 'u,IQ 6 COO(Q).
526 Lectures on the Geometry of Manzfolds

Remark 11.1.9. Suppose that u 6 C'°°(@) and go 6 COO(6) is such that

`
supp (pa C 6 \ sing supp u.
Then supp (pa 6 w sing supp 'LL and sing supp (pa = (Z), i.e., the generalized function
coincides with a smooth function. More precisely, it coincides with the extension by zero
of the smooth function (Vu) @\sing supp 'u,°

Example 11.1.10. (a) If H denotes the Heaviside function then

suppH = [0, oo) sing suppH = {0}.
(b) For any multi-index a 6 ZZ() the generalized function (1260 satisfies
supp 3§'60 = sing supp 3§'60 = {0}.

Fix a compact set K C 6 and denote by C-OO(@')K the set of generalized function u
on @ such that supp u C K. Observe that any u 6 C-OO>(@)K induces a linear function
u : C°°(@) -> R as follows: fix 17 6 2(5) such that 17 = l in a neighborhood of K. For
cl) 6 C°°(@) we set

(u,¢)* (u»t7¢)~ (11.l.2)

Note that (u, 17(l)) is independent of the choice of 17.

Indeed, if 170, 771 are two test functions equal to l on some neighborhood of K, then
77065 - mg 6 9(@ \ K ) so that
(u, 77065 - 77165) 0.
We denote by C-OO (6')q,t the space of distributions with compact support.
Note that the distributional derivatives of continuous functions are generalized func-
tions of finite order. We have the following converse, [118, Chap. III, Thm. XXI]

..
Theorem 11 1 11. Let u E C'°°(]Rn*). Then for any open set Q C R" with compact
closure Q and for any open neighborhood 6 of Q we can find a continuous function
q> R " - > R

and a multi-index a 6 kg0 such that

supp<I> C 6 and (8°@)IQ 'LL IQ .
In particular, any compactly supported distribution has finite order.

The distributions with support consisting of a single point have a particularly simple
structure, [118, Chap. III, Thm. XXXV].

Theorem 11.1.12. Let 'LL E C-°°(R"). If suppl {0}, then there exists an integer
m > 0 and complex constants CO69 a 6 Z20, lal < m, such that
'LL = Ca8°(50.

l0¢IS'fn
 Spectral Geometry 527

Note that if 610, 61 C IR" are open sets and \11 : 60 -> 61 is a diffeomorphism, then
we have a pullback operation \If* : Q 1) -> @(@'0) and a dual, pushforward operation

'L I C-°(>(@,) -) C`°°(@1),

( \If*u, up) I (U, \If *(p), Vu E c'°°(@@), SO E QW.

Given u 6 C°°(]R") and 6 @(]R") we define their convolution u * 'U 6 C°°(]R'") by

u * 'u(;u) Rn
u(;u - y)v(y) dy
y->&2 Z

Rn
u(z)v(x - z)dz = U * u(a3).
If we regard u * 'U as a generalized function, then, for up 6 @(]R") we have

u * v 7 (Q) Rn 'u,(;u - y ) v ( y l do (WJ) do

Rn

/ Rn
v(y)
(/ Rn
'u,(;u - y)(0(H1) do do
@=x-%x= Z +w
/
R"
'v(y)
(/ R"
u(2)(0(2 + y) dz
) do / R"
'v(y)
(/ R"
u(2)Ty (WI dz do,
where Ty : R" -> R" denotes the translation Z »-> Z -|- y. If we now think of u, 'U as
generalized functions we have

(u*v,go) = ('v,(u,Ty(,0)).

Observe that for any u p C-OO luv) and any up 6 9(]R") the function
R" 9 y »-> W(yl == (u)T:u(/)) 6 R
is smooth. Since 'U has compact support we can pair it with w as in (11.1.2).
Definition 11.1.13. For u 6 c~°°(R") and 'U 6 c°°°(R")p we define u *U 6 C'°°(]R"')
by the equality

( u * v , c , 0 ) := (v,(u,Ty(p) ) = (7u,,(v,Tycp)). (11.13)

Exercise 11.1.14. Let u 6 c-oo (RI. Show that for any ICGZzQ we have
5§,'") * u = u(k) in the sense of distributions.

Remark 11.1.15. More generally, suppose that C C R " is a closed convex cone satisfying
the properness condition
C N ( K - C ) is compact for any compact subset K C C . (11.1.4)

The condition (11.1.4) is satisfied if for example if the map

C X C9(x,y)>a:+yeRTL
528 Lectures on the Geometry of Manzfolds

is proper. Note that if u, 'u 6 C°°(]R'") are such that

supp u, s u p p l C C,
then, Vi 6 R" the function
R" 9 y »->'u,(ac-y)'u(y) 6 (C
has compact support so the convolution of these functions.
Suppose now that SO 6 WRn), K := supp up, and

My) Rn
u(2)(/)(2 -|- y)dz, We 6 R"
Note that
u(2)¢(2+y)¢0=> Z 60,3/+26 K -->»y6 K C
so that
suppzp 6 K - C.
In particular, supp v(y)¢(y) C K O (K - C) is compact so the integral

'u(y)¢(y)dy
Rn
is well defined.
More generally if u, 'U 6 C"°°(lR") are supported in a closed convex cone satisfying
the properness condition (1 l.1.4), then the convolution 'U * 'U is well defined.

11.1.2 Currents
For any open subset 6" C R" and any k < n we denote by Qk..F(6°) the space of k-
dimensional currents on 6 i.e., the topological dual of the space Qkpt W) of complex
valued, degree k-forms on 6" with compact support. More precisely, a linear functional
u:Qkpt(@°)->lC, (p*->((/w)
belongs to QM) if it satisfies the continuity condition (11.1.1).
Example 11.1.16. (a) The bijection
9(6) 9 (or-> (pdf Ad;u1 /\
determines a natural bijection
C`°°(@') 9 u »->
[u] 6 Qnlgl
((pA diff 1 /\ --- /\ d i m , M) = (we), WE 6 QW)-
In particular, we have a bijection c-<>°(1R> -> QUO (R).

(b) If S C 6 is a properly embedded' k-dimensional oriented submanifold, possibly with

boundary, then integration along S defines a k-dimensional current [S],

(Q), 907 W )
o n

S E Qkpt(l°

We say that [S] is the integration current or current of integration defined by S.

(c) A form 17 6 Q"-'*@) defines a Is-dimensional current [17], by setting

(90» SO 777 W E Q kc p t ( 6 l '

n
l A subset S C @ is properly embedded if the inclusion i : S <-> W is a proper map.
 Spectral Geometry 529

Definition 11.1.17 (Boundary operator). The boundary operator

3 : QI¢W') -) Qk_1(6')
is the dual of the exterior derivative d : Qkpt1 (@°) -> Q1°pt(6'), i.e., for any u E Qk(@) we
have
(Q), <9u) := (do, u), W) 6 9I"pt1(@°)-

The current nu is called the boundary of u.

Example 11.1.18. (a) If S <-> 6 is a smooth, properly embedded, oriented k-dimensional
submanifold with boundary, then Stokes' formula implies that
<9[S] [35], Z

where [65] is the integration current defined by the boundary as equipped with the outer
= -normal-first orientation.
(b) Let 17 e on-/<(@J) and denote by [77] the associated k-dimensional current. From the
equality
dlS0/\'7l =dgoA17-i-l-l)kl1goA17
and Stokes' formula we deduce that
mm = (-llkldvl = (-1)dim[~l[d'7].
11.1.3 Temperate distnlbutions and the Fourier transform
We begin by recalling the definition of the Schwartz space 56, = 5" (Rn). Roughly speak-
ing, this is the subspace of cOO ( H ) consisting of functions with fast decay at oo. Here is
the precise definition.
For as E R" we set
(iv) \ l + III
For k € Z 2 0 we define
t oo], : sup U)k l@§'S0(a2ll.
Pa : C°°(R")
-> [0 Pal
p)
X€R
n(£
lol$
k
Then
in = 5'(R") := {(p 6 C'°°(R"l; pa,((,o) < oo, 'v'k E ZZ()

The functions in Yn are called Schwartz functions. Let us observe that

2
QW) C in, 6-1x1 E in.
The space in is equipped with a natural topology defined by the metric
l
dist.((/mD) =
.
Z
- m l n (1»PI <((0
Mk WI I.
k _0
Thus 9 9 v - 9 0 a s k - > o o i f f
lim Pk( S0u -S0 l=0, 'v'k 2 0 .
v->oo
We have the following result, [66, Lemma 7.1.8].

Lemma 11.1.19. The subspace Q (RTL) $5 dense in 57r1-

530 Lectures on the Geometry of Manzfolds

Definition 11.1.20. A temperate distribution on Rn is a continuous linear functional

'u,:5'n->C,
i.e., there exists k 6 ZZ() and C > 0 such that
I I
(Fu, (p) S Cpk((/0), W ) G Yn-
We denote by BYI the space of temperate distributions.

We have strict inclusions

Ccp£°(R") C Y' C C'°°(]R").
We denote by (-, -) the canonical inner product on R" .

Definition 11.1.21. The Fourier transform on R" is the linear map

8? : Yn -> C'°°(]R"), Yn 9 up »-> ¢ = 5"[(p],

6(6) Rn
e 'i(€'X)90(a:)d:z:, we 6 R"
We list below a few fundamental results concerning the Fourier transform. For a proof
we refer to [39, Thm. l4.8].

Theorem 11.1.22. (a) For any go G Yn, we have QUO 6 in and the resulting linear map
5? : Y" -> Yn
i5 continuous.
(b) For any ( / ) , W € § ' n

(3(=v)1D(1v)d1v (0(w)(w)d2v- (11.l.5)

R" R"
(c) (Parsevalformula) we have
1
(2¢f)" (¢,3)L2,,,
((p »1 P> L2 (R
") I

i.e.,
1
¢(w)W(w)dI = (g71fln @(go,z(gldg.
Rn Rn
(d) (Fourier inversion formula) For any go 6 in we have
1
(,f)(;v) = (gffln @(§»@)¢(§)dg.
Rn
(d) For any ¢,¢€%n, their convolution up * w is also a Schwartz function and
A

up* is) I ¢(<) . mf).

Using the equality (11.1.5) above we can extend the Fourier transform to temperate
distributions.

Definition 11.1.23. The Fourier transform of a temperate distribution u 6 Y' is the tem-
perate distribution Q 6 Y' defined by

(Tb) 12% vweyn.

..II

Up
 Spectral Geometry 531

11.1.4 Linear dw'erential equations with distn°butional data

We gather here a few basic facts about generalized function solutions of ordinary differen-
tial equations. Our presentation follows closely [66, Chap. III]. We begin with the simplest
differential equation.

Theorem 11.1.24. Suppose I C R i5 an open interval. Then the only solutions of the
duj'erential equation
du
0, 'LL ECOOII)
dt
are the functions constant on I.

Proof. This means that

(U, (15) 0, w 6 QU).
Consider the generalized function M E C"°°(I),

(Ula) W(s)ds, vi 6 Cgo>(I),

I
In other words, ,UI is the generalized function determined by the indicator function of I.
Observe that if

1? = ( M I ) = 0,

then the function

(Mt) 'QDIsIcZs

is supported on I and,

(U, W) I (U, co) (up co) I 0.

Hence

(no, 0 => (u, 0.

Now choose ¢ 0 € 0 3 ° ( I ) such that

(M I, Vo) I 1.
Set CO Lu, Then u = colll- Indeed, if <15 6 9(1) and

then
c
Z gz5(s)ds, w I <2) UPu

0 Lu, I (nu, - clu, f 0),

SO
532 Lectures on the Geometry of Manzfolds

Corollary 11.1.25. Suppose that I C R i5 an open interval,

n E z>0, A E o°°(I,End(R") ), f E o0(I,1Re.">.
If u 6 c-oo lm satisfies the linear dnterential equation
du
+Awu=f(m (1l.l.6)
dt
then u 6 C1 (I, RTL), i.e., 'LL is a classical solution of(II.1.6).

Proof. Choose
E 6 c°°(1,End(R"> )
such that E = EA, and E(tl is invertible for any t 6 I. We deduce that
d
(E(t)u(t)l I E(t)f(t)-
U/
I v E c1(I,R ") satisfies E(t)f(t l» then we deduce
d
(E(ilu»(i) - v(t)) = 0.

Theorem 11.1.24 implies that there exists C 6 R" such that

E(t)u(t) = v(t) + C.
Hence 'u,(t) Eltl-1fvltl + E(t)'1c is a classical solution of the above differential equa-
son.

Corollary 11.1.26. Suppose that I C R is an open interval, 007 7

f ( t ) 6 00(1) [to 6 C"00(Il satisfies the deferential equation

m-1 defy
d1"u _I.
dim as do f(t)»
j 0

then 'U 6 com (In, i.e., 'U is a classical solution of the above equation.

Proof. Observe that the (vector valued) generalized function

u(tl = (Vu),1/(tl, . . . ,W-1)<t>) 6 c-oolnm

satisfies a linear first order ordinary differential equation with smooth coefficients and con-
tinuous nonhomogeneous term. Corollary 11.1.25 implies that u is a classical solution.

Corollary 11.1.27. Suppose that Q C R" is an open set and I C R is an open interval.
We denote by x1, . . . , ac" the Cartesian coordinates on Q and by t the Cartesian coordinate
on I. If u 6 C-©°(I X Q) is a generalized function satisfying the dnterential equation
3t'LL = 0, then there exists 'Ag 6 C-00(9) such that

(U, (ton al)

I
(u0(11)» (1502 ac))dt, WE E on X Q). (11.1.7)
 Spectral Geometry 533

Proof. For any w 6 9(9) define UP/» 6 C-00(1) by setting

(UWP) = ( l b / 2 ) , W ) 6 Qu).
d
Then EW = 0. Using Theorem 11.1.24 we deduce that Up is a constant function. The
map
@(SO) 9¢1->'Ll,¢ @C"°°(I)
is obviously linear and continuous from the usual topology of @(SO) to the weak topology
of C-00(I). The image of this map is the l-dimensional subspace consisting of constant
functions so that the resulting map
@(Q) 9 ¢ l-> We E R
is continuous. This defines a generalized function U0 6 C-OOIQI satisfying (11.1.7) for
M Q ( I X Q) of the form (W, I) (/0(t)¢(11)» i.e., Z

(U, W ) Z / I
u¢c,o(t)dt = (U0(2Ul» (/)(t)¢(I))dt~
The desired conclusion follows by observing that the linear span of functions of the form
go(tl1D(:u), go E QU), AD 6 9(9) is dense in QU X Q).

For any open interval I C R, any open set 6" C R" we denote by CO,-OOII X @) the
space of (weakly) continuous maps I 9 t -> it 6 C'°°(6'), i.e.,

t I-> (Ut e SO)

is continuous for any up 6 2(6) Note that we have inclusions inclusion
c0(I X al C CO, °°( I X @)) C 0-°o( I X WI.
The first inclusion associates to each continuou s function Fu, 6 CO(I X al the path t I-)'U,t
of continuous functions on H, ut(a2) = u(t,;z:), Vat 6 6. Moreover
Vu Q c0»-°°<I X @), Va Q 220, t »-> age Q c0»-°°<I X @). (11.l.8)

Let t denote the Cartesian6coordinate along I. For k > 0 we define

{u
Co CO °°(I >; <9gu
,

(I W) Zi 0 (I @), 1 < j < k }.

6C
X X 6
We have the following result, [66, Thm. 4.4.8].

Proposition 11.1.28. Suppose that u E C-°°(]R X Run and its restriction to ]R>0 X Run
satisfies the differential equation
v-1

3tu+ Aj(<9x)3{u=F7 (11.l.9)

j:0

where As (6) : C°°(@) -> C°°(@) are partial dI]j'erential operators with smooth coeffi-
cients independent oft. IfF 6 cO,-inftylRTTIl, then

u 6 CV,-ool I X of.
534 Lectures on the Geometry of Manzfolds

Moreovelg Q"

lim Ft
two
exists in C - O O (Run, then the limits
lim Ag Ut , l < j< 1/ l
¢\,0

exist in C-°°(]RM).

Proof. It suffices to show that, for any T' > 0 and any bounded open set Q C R'rn, we have

Ul(0,r)xQ 6 cu, oo (J X Q).

Fix such a Q.
Observe next that we can reduce (11.l.9) to a first order (in 89 system of partial differ-
ential equation by replacing u with the vector-valued generalized function
'U (<9 ul0$1$v-1

Thus we can assume that I/ = 1, i.e., (1 l.1.9) has the form

F. (1 l.1.l0)

Fix 5 > 0 and set J := (-5, T). Using Theorem 11.1.11 we deduce that we can find an
integer ,u 2 0, multiindices as 6 Z20, 0 < k < ,u, and generalized functions
u,.€c0(JxQ)cc0»°°(JxQ), 0< k< ,U

such that
m
k k
'U»IJxQ c9t<9,§" G u n
k 0 JXQ

From (11.l.8) we deduce

UIJxQ € CO,-OOIJ X

We say that the t-order of u is < ,LL, and we write this ordt(u) < ,LL. We will prove by
induction on ,LL that u G C-OOIJ x Q).
If p = 0, so that the t-order of u is < 0, then
u = 3§ic()u0 6 c0(J X QI.
Using (11.l.10) we deduce
Btu = -A0(3xl3§'°u0(tl + F(t) 6 C0"00((0,r) X QI => u 6 Cl'-°°((0J°) X Q).
G

Moreover, for any up 6 Q (Q) we have

to

1})("J»(t),(p) I ("~'»(t0l)l0) (GU), (p )dt.

0
 Spectral Geometry 535

The last integral is finite since the function

[0,t0] »-> <G(¢),9,0>
is continuous.
Suppose now that ordt (u) S m, M > 0. Then
m

Btu = -A0(39,,)u = 38§'"Uj + F

j=0

-gt 3§-1A0(<9x)3§'" +F - A0(a>ag0u0.

j:1

We set
m

U : u + ag'-lA0<a><98mi,
j=1

so that
as F - A0l3asl3*0U0 G 0"»-°°((0,t~) X QI.
G

Now choose w G 01»-<><>((0,7~l X Q), (OJ) 9 t »-> 'Wt 6 C'°°(Q) such that 3t'LU G.
E.g.,

(we,(/)) ( G s ) , W) G 9(9)-
to
Clearly
lim ( w e )
t\0

exists 'v'(p G 9(9) since the limit

lim (G(t),g0)
¢\,0

exists Was 6 9(9) Then

3t('u - w) 0 in C-OO (IN) X Qbigl.
Invoking Corollary 11.1.27 we conclude that 'U w is a constant map (0, T) -> C'©0(Q).
Thus
ordt(v) = ordeal (u - w) -|- w) < maxlordt(v - w), ordtw) < 0.
Hence
TTI
aZ"A0 m
$-1A0
ordtu = ordt 'U
j 1

We can now conclude by induction that 'LL E Cl,-OOIJ

) (< Otdt

X
j 1
<

Q). By construction, the limit

lim (up)
t\0

exists WE 6 9(9
536 Lectures on the Geometry of Manzfolds

11.2 Important families of generalized functions

We want to spend some time discussing in some detail several important examples of gen-
eralized functions that are rarely discussed in a traditional graduate course. We will make
use of these generalized functions when we investigate the spectral data of a Riemann man-
ifold. Throughout the following discussion Euler's Gamma function
OO
t e l
rial e t ds, Re S > 0,
0

plays an important role. In particular, we will use less familiar properties of this function.
They can be found in [84, 134].
An important concept we will use in the sequel is that of holomorphicfamily ofgener-
alizedfunctions. A family of generalized functions l u a l a e A 6 C'°°(lR) parametrized by
an open subset A C (C is called holomorphic if, for any go 6 9 (R) the function
A 9a»->(ua,(p)e(C
is holomorphic. The concept of meromorphic family is defined in a similar fashion.

11.2.1 Some classical generalized funetions on the real axis

If a G C , Re a > -1, then the function ;z3 _I_:]R->lC defined by

ata, a2>0,
o f

0, ;USO,
is locally integrable and thus defines a generalized function 1/3+ E C"°°(]R). A simple
computation shows that
x - x + - x + +1 , ' v ' R e a > - 1 , (11.2.1a)

d -1
x+=ax+ ,'v'Rea>0. lb)
do:
We want to extend the definition of 513+ to all a € ( C . For QSQQUR) the function
OO

{Rea > -1} 9 a»->Ia(gz5) (iv"+» QS) :z:"cp(a3ld:u

0

is holomorphic. Using (11.2.1b) iteratively we deduce

_ la
la((l))= ( a + 1 1 'v'Rea > - l , k E Z20. (11.2.2)
(a + k) '"*'" ( Q5(/<=) )
7

If a 6 C \ Z < 0 then we can define

I@((/2)
(a + 5 .. -1
l(a + k ) I " + / " ( ¢ ( k ) )» Vi > max(0, -1

The equality (11.2.2) shows that the right hand side of the above equality is independent of
Rea .

the choice of k > 1nax(0, -1 - Re a).

 Spectral Geometry 537

This gives a meromorphic extension of the function a »-> la((p) with simple poles
located at Z<0 and thus allows us to define the generalized function ac+ for any a E @\Z<0.
Observe that
(-1)k l
lim (a + k)Ia((/0) = (
a-> k l k)(2 ff) (-1)!
I0(f(k)) I la 1>!
¢('"-"(0),
(1 l.2.3)
so that

lim (a + k)a3"+ (-l)/<3-1 60(k-1)

a-> k (k - l)!
More explicitly, for Q € @ \ Z < 0 , Re a > - n - l, n € Z > 0 we have
l n 1 k (k)
( i , (25)
0
£801
( )
k 0
at " , (0) do:

-|-
l x"¢(x)dx +
k=1 (k
Cl)<k-1>(0)
-
l)!(a-I-k)°

To eliminate the poles of a »-> :z:8- we will divide x by a meromorphic function with
simple poles at Z<0.
For any Re S > 0 Euler's Gamma-function satisfies
I`(s -|- 1 ) = sF(8).
We can use this equality to extend F (5) to a meromorphic function on (C with simple poles
at Z$0. Note that the residue at -k E Z<_ 0 can be determined from the equality
• F(s-I-k-I-1) F( ) (-1)'<
lim
5_>_k(8 + k)II(8) : : nm (-Up k1 (1 l.2.4)
5->-k 5(s+l)---(s+k 1) k!
Using the equality

we deduce that F has no zeros in Z. Set

1 a
X"+(a;°)
o
• ;u+, Re a > (11.2.5)
I`(a-|- l )
The correspondence

{R e a > - l } 9 a » - > X + E C'°°(]R)

and in this half-plane it satisfies the equality
d X++1 I X+.

We can use this equality to extend X+ analytically to all a E (C,

doat X+-I-n
538 Lectures on the Geometry of Manzfolds

where n is a positive integer such that n -|- Re a > - l . Since X0+ is the Heaviside function
we deduce
do
X+
k l
5§,' ° ) dank
60, We E Z20. (1 l.2.6)

Similarly, for Re a > - l we can define

zuCL 0, an 2 0,
txt", cz: < 0,
and for any a

l $U> 0
x-a (f l = Xa+l- so) = I`(a+ II 0,lxl07 ;z:< 0: Re a >

Equivalently

(xa p) = (x+,¢), Va) 6 QQR>,

where
g(:z:) = go(-ac), Van E R.

Note that
in a n n a
da2'*X'+ =(-l) x--
Observe that for Re a > - l we have
:H+-I-:1:a = laval"'.

Let co 6 m). The function a *->(£6"'+»<l5) is meromoqohic on (C with simple poles at

a 1, - 2 , . . . , - k , . . .
According to (11.2.3) the residue at a = - k is
1 ( )(k:-1)
(k 1 )cb(k-1)(0)
( / - 1 )!
(éék-1)7

Similarly, the function a »-> (ma ,g25) is meromorphic on (C with simple poles at
CL 1,
and the residue at a is
1 1
( k _ l )¢E(k-1)(0) la _
1>!
(IS'"*), (15)-

Thus the function

a »-> Gwlal :z (l22l"',(l5)

is meromorphic on (I with simple poles at
a -1, - 3 , . . . , - ( 2 j + 1 ) , 7 j e Z20.
 Spectral Geometry 539

The residue at a (Quo + l) is

2
(6é23),<l)).
(2)!
»_> 1
The function a F( 0 +1
2 ) is holomorphic on (C with simple zeroes at

a = - l , - 3 , . . . , - ( 2 j + l ), 7 j e Z20.

The equality (1 l.2.4) shows that the derivative of this function at a (Quo + l) is
(-l)7+1 ( j -|- l)!. We conclude that, for any up E QOR), the function

is holomorphic on (C. We denote by txt" the generalized function defined by the equality
(lxlw> Fgplal, V9o 6 @OR).
Less precisely, but more intuitively, this means
1
l)cl"'(33) a 1 l:z:l"'. (11.2.7)
F( "S )
Note that

(-1)j+12(j + (11.2.8)
IXI-(21+1)
(2J)!
For y ¢ 0 and a QC we define
• a a o az 'ii 'jar co 'i k-1
( m ) e (agl -I-yl+ g<+y>>(66 ),¢),
where

are(;z: + iy) 6 (-or, or), :u + in = 1112 + i y l € i ate(=v+iy)

We deduce that for Re a > -1 we have

(x :|: in I" lim

y->0i
(33
+ 'iy)a I 33+ _|_ € i n i a x a
(11.2.9)

We have a more precise result.

Theorem 11.2.1. Consider a holornorphic function f defined on the strip

{z x + Ty E C ; 0 < y < v }.
If there exists constant C > 0 and an integer N E Z>0 such that

lf(;u-{-iy)l < Clyl - N a Va: -i-i?/GZ,

then f(o + i y ) has limit f0 6 C'°°(]R) when y -> 0. Moreovelg this limit i5 a generalized
function of order at most N -|- l . We denote it by f (co + i0).
540 Lectures on the Geometry of Manzfolds

Proof. Fix Z0 E Z and consider

F(2) f(c)dc-
250

Then

C01/ 11'N , N >1

IF(2)l <
l

If N = l it follows that the integral of F is continuous on the closure of Z since lost is

integrable on the interval [0, by].
We deduce that there exists a continuous function G : Z -> (I whose restriction to Z is
holomorphic and such that
dN-l-1
F(2l dzN+1 G(z).
This implies that for fixed y we have
dN-I-1
f(w -|- iy) d:z:N+1 G(:z: -|- iy).

.
Since G( -|- iy) converges in the sense of generalized functions to a continuous function
:u i0), we deduce that f(:z: + in) converges as y \, 0 to the generalized function
I

G;
d +1
N -|- l. _
d:v"+1 G(:c -|- 110) which has order

Theorem 11.2.1 implies that (11.2.9) holds for all CL E C \ %<0 and that the family

a »-> (at dz i0)"'

is a holomorphic extension to C of the, a priori meromorphic, family

a *-> :c+ +€i0nia .

In particular, we deduce that
1 €avri(x _ imp e -a7r1l(x _|_ imp l (11.2.10a)
Zi sin tra ( 7

1
:IJ-I-'0 la? - in)a (11.2.10b)
_ .
Qzslntrall Z Ia
Let us observe that
d
(33 -|- i0)0' = a(;z: -|- i0)a-1
do
Proposition 11.2.2.
e:F i " ( 2 + 1 ) :F in) '(0-I-1) (ll.2.lla)
)E(§)

IX"I (5) = 1/72a +1IXI -

a
( +1)(§), VGQQ. (ll.2.llb)
 Spectral Geometry 541

Proof. Let Rea > - l and 5 > 0. Set f5(;c) := e"'"';c+ so that fE -> to a
+ in the
space Y' of Schwartz generalized functions. The Fourier transform of fS is

f2(§) e-as(6+i€);l:"'da3.
0

Set C: 5 +ig so that

OO

f2@)=c-" 0 e-"<(w<)"dw

(z: IC)

€-a-1 @-Zzadz <-(°+1)r<a + 1) = I`(a + l)(e + ig>-("+1)

R>o~C

'i1r(a-I-1)
2
(5 - i n ) - < >
Hence
'i1r(a+1)
-(a+1)
fins) Z I`(a -|- 1 2
(6 in)
Arguing in a similar fashion we deduce that for Re a > -l the Fourier transform of ma is
'i1r(a+1)
I`(a + l )e 2 (5 + i0)'("+1).
Hence the Fourier transform of 1/a(:z:) = Ill" is
'i'r(a-I-l) 11'7r(a-I-l)

me) = I`(a + 1)(e- 2

(s - io)-<"+1> + e 2 (5 + io)-('*+1))
Assuming that a Q Z we deduce from (11.2.9) that
l 117r(a+1) 'i1r(a-I-l)

e) Q 2 (5 - i0)'("+1> + 6 2 (5 + i0)'("+1)
I`(a + 1

€
'ilr(a+l)
2 lg-4-(a+1)
-|-
€(a+1l1r'i€l0+1l +e
'i'r(a+l)
2 (€+(a+1) -I- 6 - ( a + 1 ) i § ( a + 1 )

)
l l§l-(01+1
(a+1)
2 c o s l or(a+
q
1)\ lél Zsin

At this point we invoke the reflection formula for the Gamma function

sin
F(- %>F(1 + go
to conclude that
1 l )
l§l'(0+1
7T 'ii
I`(a_l_ MS IXI'("'+1)l§).
F(l + go F(-3) F(1 + 3)
Thus
27rII(a -|- l)
Ixl" (5) F ("§1>1"(1 + go |x|'("'+1>(€)-
542 Lectures on the Geometry of Manzfolds

Using the duplication formula we deduce

\/-'rrI`(a -|- l) 2"'I` I a -I- l

2
)1-(1+-)
a
2
Hence

lx="(§)=¢%2"+1uxl-("+"(®, vRe @ > 1 , aw.

The equality (11.2.1 lb) now follows by unique continuation.

Corollary 11.2.3.
X+* xi X++b-I-l , Va, b.

Proof. Pass to Fourier transforms and then use (11.2.1 la).

If we define
I + :z X+(1+0) 7 QECC,

then we deduce

and
I I [3+ I+
0-I-B7
'v'a,B€@.

Noting that
I+
k *f Z 65" * f f<>, of 6 m),
we could conclude that the operator
.@(R)9ul-> I +
a * u€C°°(R)

can be viewed as a fractional derivative of order a. The distributions a , are defined as

I- L
•I and satisfy similar properties.

11.2.2 Homogeneous generalizedfunetions

Suppose Y C R", X C RN are open sets and \I/ : X -> Y is a submersion. Denote by
y = (y1, . . . ,if/"I the Euclidean coordinates on Y and by :u = (5131, . . EN the Euclidean 7

ordinates on X . We have a natural pushforward map

q;* : x) -> QW)
motivated follows. The co-area formula (9.l.l2) implies that for any (p Q Q(X),u Q
9(Y) we have

X
(@*u)()¢(x)ldgvl
/Y
u(y)
U l
dV -1 ldyl,
 Spectral Geometry 543

where IV\Ifl(a:) denotes the Jacobian of \If at the point QUO E X, and n a v y ) ( x ) n denotes
the Euclidean volume density on the fiber \If' 1 (ac). More precisely,

IV\I/l(;v) I \ det(D\I1)(;u)*(D\I/)(x)

where D\I1(13) : RN -> R" denotes the differential of \11 at ac. Note that if n l, then
IV\I/l(x) is indeed the length of the gradient of \11 at x.
For any go E 9(X) we set
1
dV . (ll.2.l2)
..II
\If *(/2(9) -1

More conceptually, the density \If*(p(ylldyl on Y is the pushforward of the density

(p(x) Idol on X. Note that \I/*9, E 9(Y). We define the pullback
*
: 0-°°<y> -> 0-°°(x>
where for u 6 c-°<>(y) we define \If*u 6 c-°<)(x) via the equality

(\If*u, up) == (u,\P*(0), W 6 @(x>. (11.2.13)

Note that if u 6 C°°(y) 6 C'°°(y), then \If*u E C°°(X) and, more precisely

Indeed, using the coarea formula, we deduce

X
u»(\P(=v))(p(2¢)ldVx(w)l
f (f
yu(y)
'1"1(9)
1
IV\Pl:cl (p(x)ldv
11(11/)l l5z:l| ldvy (y) I

More generally, if \P is not a submersion, but its singular values are outside an open neigh-
borhood of U of sing supp u, then we can define \P*u via the following cutojconstruction.
Fix a smooth function U E @(U) such that 17 = l on a neighborhood V of sing supp u
in U. The pullback \I/* ( m I is well defined via the above procedure since the restriction of
\If to \If-1(U) is a submersion.

According to Remark 11.1.9, the distribution (1 - 17)7u, E C - O O * Y) is smooth and we

define \P*( (1 - 17)u) as the usual pullback of a smooth function via a smooth map. We
then set
*
\I/ *u 1 l77u)+ * ( l -WuI-
L
\I/ \I/
•

One can verify that the above definition is independent of the choice of 17 with the above
properties.
Indeed, if 170, 771 are two such functions, then

'I'*lTloul= * l*71*70U)+ * (( l -Tl1)'70Ul»

* (( l -Tlolul I \If *lTll( l -Tl0)Ul+ * (( l - M l ( l -'70)U)-

By adding the above equalities and using the linearity of \If* we deduce

\I/*l'70'U»l + Q* (( l -T/o)ul = Q/ *('71Ul + QI* (( l -'71)Ul-

544 Lectures on the Geometry of Manzfolds

Example 11.2.4. Let c # 0 and consider the map

R t)
->] z
oR Q¢( 2
Q -H -C
Note that only the point SO = C2 is a critical value of this map, and Q-1 (50) = {0}. For
S > SO we have

Qe 1(so = ::V/3 -|- 02.

If (P Q m) is such that supp SO O { 0 } U) we have

I(/.(ws + 0 2 ) + + C2 ) 7 s> c2
(Q2)*(0(s)
0, S _<-C2

Hence
1
(QZ50, cpl llllllllll

20((0(0l + (0(-0) )-
Using the cutoff construction of the pullback we deduce
1
QZ50 =- 6 60.
Qcl e

For t > 0 we denote by Qt the diffeomorphism

@t:]R"-HR" 7 :cr->t;z3.

Note that for SO 6 @(R") we have

l
(%)*(/)(y) (Put-lYl = t'"(p F l y .
lvgtl(¢-1y) ( )
Suppose that X is a conical open subset of R", i.e.,
QM) = x, it > 0.

A generalized function u E C'°°(X) is said to be homogeneous of degree a Q C if

-
-
.Qty tau, Vt > 0.

Equivalently, this means that for any go E Q(X) we have

t"(u,(p) I (U, l<%t)*S0)) W ) E @(x>. (11.2. 14)

If we differentiate the last equality with respect to t at t = 1 we deduce the Euler identity

(U»+")(U»S0)+(U»
Ecp)=0, V(,o€9(X), (11.2. 15)

where E is the Euler vector field

TL

E a
j 1
 Spectral Geometry 545

Conversely, the equality (1 l.2.l5) implies (1 l.2.l4). Indeed, for t 6 R define the continu-
ous linear operator
St I C'0°(X) ->C'°°(X), Stu: e`t"%e*tu.
Then
SO I ]17 St() o Sto = St0+t17 \V/t0,t1 E R.
The equality (11.2. 15) is equivalent to
d
Stu 0.
dt t 0
so that
l
lim #So - SoIl I 0,
h->0
in the sense of generalized functions. Hence

I 1
0 sO lim #So - S0)'LL)
h->0
lim
h->0
1
l5t0+h - $t0) U-

Hence
d
Stu = 0, We.
dt t=t0

Hence the function t »-> Stu E C"00 (X) is constant, i.e., u is homogeneous of degree a.

Example 11.2.5. (a) The generalized functions Mai, (at :t in)", and XI are homogeneous of
degree a on R \ 0.
(b) The Dirac generalized function <50 6 C"00 (RnI is homogeneous of degree Indeed
(60, (=@t)*(p) = t'"'<50, (/)(t:v)) = t'"(/)(0) = t'"<60, (p)-

Theorem 11.2.6. Suppose that u E c-oo (R>0l is a homogeneous generalized function of

degree a Q (C. Then there exists C 6 Q such that Fu = et+.
Proof. The generalized function u satisfies the Euler equation
11-I-(a-I-mlu=0.
Now invoke Corollary 11.1.25 to reach the desired conclusion.

Theorem 11.2.7. If 'LL E C'°°(]R" \ 0) i5 a homogeneous generalized function of degree

a 6 (C \ Z<_,,,
_ then u has a unique extention U E C'°°(]RTL) which i5 homogeneous of
degree a. If P i5 a homogeneous polynomial then
(Pa) = P - a.
Moreovelg Q" a E (I \ Z<-(n-1) we nave
l8x5 UI I 6321 U, W 1, . . . , n . (ll.2.l6)
546 Lectures on the Geometry of Manzfolds

Proof. Uniqueness. Suppose that that 'EL and u' are two homogeneous extensions of u to
Rn. Then it - u' is supported at the origin. Hence it is a linear combinations of derivatives
of the Dirac generalized function. Such a linear combination cannot be homogeneous of
degree E (coz< n. This proves the uniqueness of the extension.
Existence. For any QS Q QW) we define
T aw ) == (¢++'*-1, M) >R» M) Q; (w), at Q R" \ 0,
where ( -, - NR denotes the pairing between a generalized function and a test function on
R. Note that Ta(lS Q COO In" \ 0).

Lemma 11.2.8. For any a E (coz< n and any go E @(R") the function Tact i5 homoge-
neons of degree - a - n.

Proof. For a > - n and r > 0 we have

OO

T we) = (t++T*-1, Q)(t~w)) z(*+'*'1¢(m;)dt

0
OO
SI 'it 7' a in,
s"+"'1cl)(s:z:lds = 7,-a-nT a¢(x).
0

By unique continuation, the last equality continues to hold for any a E C \ Z < 77,9 so that,
for any a E C \ Z$_n the function T a<l5(;z:) is homogeneous of degree - a n.

The key moment in the existence proof is the following auxiliary result whose proof we
postpone.

Lemma 11.2.9. [few E C"°0 (Rn \ 0 ) is a homogeneous generalized function of degree a,

then
(u,17) 0 If 17 E QUO" \ Of and T a17(a2l 0, Va: E R" \ 0.

Fix a function W@@(1R("\0) satisfying

OO

we) dt llllllllll

1, Va3;£0. (l1.2.17)
0

Then wT ad) Q mm" \ Of for any <2) Q Q W ) and

T aW
T a¢)(;¢) tC*+"-1¢(w;)T ag25(tac)dt
0

( T a(b(tx1) I t-a-fn T a(z5(1v))

OO
dt
T a¢(23l Mow) t llllllllll
T a¢(w)-
0

We rewrite the above equality as

T aw-wT a ¢ ) = 0 ,
 Spectral Geometry 547

and Lemma 11.2.9 implies that

(u, ¢) = (W T a¢), is Q 0-°°<R" \ 0).

The sought for extension of 'LL can now be easily described. More precisely, we set

(M) .
a

(WWT w, W Q @(R").
the right-hand side of the above equality is well defined since T a¢ E @(R"\0)9V¢ E w
9(]R").
If P is a homogeneous polynomial, then Pa is a homogeneous generalized function of
degree a + deg P E (C \ Zs_n. Clearly Pa coincides with PA on R" \0 and the uniqueness
of the extension implies that

The equality

Paul = as
when G, E (C \ Z$_(n_1) is proved in a similar fashion.

Proof of Lemma 11.2.9. We will prove that

77 6 WR" \ 0), Tan = 0 => 3¢ 6 @(R"\0) E d)+(0»+n)¢=17- (ll.2.18)

Clearly (11.2.18) implies Lemma 11.2.9 since

(11.2.15)0
(u,17) I (U, EQ5 + (a + HW)
To prove we observe that for <l> 6 C°°(R" \ 0) we have
d 1 -»

(l5ltcu) -E ¢(w), 'v'la;°l 1.

t
Hence
d
E ( w ) + (a -|- "»lc25(t¢v)
dt lta+"q25(ta;'ll
1 tCL-l-TL-1

Thus (11.2. 18) is equivalent to

d
dt (t"+"¢(tw)) = ¢"+"-lm), am 1.

The function <2) defined by

OO
l
q25(ta3) = -ta_I-7n 7_a-I-n-117(7'ac)d7'
t

will do the trick.

548 Lectures on the Geometry of Manzfolds

11.3 The wave equation

The wave operator of a compact Riemann manifold (M, g) is the p.d.o.

oo X 7
M °°( X M)-> ( M) Mu=3t2u+ mu.
In this section we consider initial value problems of the type
MU 0, u ( 0 , - ) = 0 , u ; ( 0 , - ) = ( 0 € °°(m).
As we will see later in this section, this initial value problem has a unique solution which
we denote by LIQUI-
Using a technique pioneered by J. Hadamard [57] and clarified by M. Riesz [114],
we will produce asymptotic expansions UtlS0l for ltl small. Our approach follows
HOrmander's modern presentation in [67] of the ideas of Hadamard and Riesz.2 As
HOrmander explained in his groundbreaking work [65], these short term asymptotics can
be used to obtain rather precise information about the spectrum of Am.

11.3.1 Fundamental solutions of the wave

Let £8 1 7 ,buzzm denote the Euclidean coordinates in Run. We set
m

Run 616
2
a2.
k 1
Q -<)<)( m-I-1)
fundamental solution of is a generalized function such that

60.
2
auchyfundamental solution for is a -map

(0,<)<)) 9 t »-> Et € 0-°°lRm>,

such that
m

QB Et @2,E = 0, t > 0, (11.3.1a)

j=1

lim Et 0, (11.3.lb)
t 0

lim <9¢ Et 50 Q C' olnwl. (11.3.1c)

t\,0

To a Cauchy fundamental solution lEtlt>0 we associate a generalized function EE

"+1) defined by
C-Oo(R"
OO

(E, up) (Et) (Mt, -I )it, W E QIRTH-'-1). (ll.3.2)

0

2 H6r1nander's initial doctoral adviser was M. Riesz who retired. HOrmander then completed his dissertation
under the direction of L. Garding.
 Spectral Geometry 549

The above integral is well defined since the function

[ 0 , o o ) 9 t »-> ( E , ¢ , c p ( t , - ) ) € ] R
is continuous and compactly supported. The Uniform Boundedness Principle, Theo-
rem 11.1.5, implies that E defined by (ll.3.2) is indeed a generalized function. Note
that
supp E C R¢20
7r-I-1

Proposition 11.3.1. If lEt)t>0 is a Cauchy fundamental solution for the operator L, then
the associated generalized function E is a fundamental solution o f L . We say that E is the
fundamental solution associated to the Cauenyfundamental solution (Ei)¢>_0.

Proof. For p e gln m-}-l)

OO

( lim
»(p)=(» @o)= e7\0 ( £7 ( (/0 l(t» -)) dt
E'
OO OO

lim
e\0 /S
OO
( E t , (A(,f))(t, - ) )it -I- elim
\0 S
(E (<9§(0)(t»-) )d¢
OO

lim
e\,0 /E
(AE
OO
(pa, -I )it 1 0(E, a 9/0(6, 0 ) ) - lim
e\,0 S
(stEt» l3tS0l(t» -I )do
OO
(11.3.1b)
lim
e\0 / S
OO
(AE (pa, - ) )dt + lim ( 6 Et, (0(t, -) ) + lim
€\0 €\,0 E
(6§Et, (0(t, -) )dt
(11.3.1c)
lim
s\,0 / S
(aM + Am, t, -) )it + ¢(0» 0)'""'"%»(0, 0) (<5R H ) -

Suppose that n = m -|- l, and we denote the canonical Euclidean coordinates in RM-P1
by (t, 1/31, . . . , a s ) . We consider the quadratic from
q: R1-I-m -> R, q(t,2¢) = t2 - l@¢l2, as E Rm,
we denote by C+ the forward light cone
C+ {(t»1v); q(t»£'1l 2 0 , t > 0 }= { lt,:z3); t 21w1 },
and by the backward light cone
C_ :={(t,;z:); q ( t , x l > 0, t < 0}= {lt,;I3); -t 2 1331
The map
C+ X C J , 9 ( x , y ) - > a : + y e ] R "
is obviously proper, so the cone + satisfies the properness condition (1 l.l.4). We have
the following important uniqueness result.

reposition 11.3.2. *be wave operator admits at most one fundamental solution sup-
ported in the forward light cone C+.
Proof. Suppose that E, E' are two fundamental solutions with support in C+. Then E'
E -|- u, where u is a generalized function with support in C+ such that
I
'LL 0.
550 Lectures on the Geometry of Manzfolds

Since the cone C+ satisfies the properness condition (1 l.1.4) we deduce from Re-
mark 11.1.15 that the convolution E * 'LL is well defined and we have
0 I
u) ( u) ( * U 5*u u.

11.3.2 The wave family

We denote by Low the Lie group of special Lorentz rotations consisting of linear operators
A _ R1+77l _> R1+M

such that

A * q = q , AC'+=C+, detA=l.
ote that is of-invariant. In this section we plan to solve a more general problem
namely, we want to describe all the homogeneous, Low -invariant generalized functions on
R1-l-'Nr 1 that are supported on the cone C+ .

Let us start with a simpler question. Suppose that f : Rm-1-1 -> R is a continuous
Low-invariant function on R'"+1. Observe that the orbits of Low consist of the level sets
t
of q. Hence there exists a continuous : R -> R such that

f(t»32l=t(q(t»@))-
t
Moreover, if f is supported on C+, then is supported on [0, oo). Finally, if f is homoge-
neous of degree d > 0, then Theorem 11.2.6 shows that f must have the form
d
f = const.q*x
To deal with the more general case when f is a generalized function we follow the strategy
in [92].
Set n := 1 -|- m. For any open set 6" C R1,m we have a cap product
M:Qj X Q,¢(@))(@')-9k-i(@)l,

(in re T) = <5 /\ a,T), VT E Qm), a 6 9 1 ( 6 , BQ £z'"ptJ<@)>.

Observe that if @ is a Low-invariant open subset of R 1 , then a generalized function

u E C - O O (6) is Low-invariant if and only if the associated n-dimensional current is such.

Proposition 11.3.3. Suppose that @ C RU" i5 Low-invariant.

(Q)/f"J» E (j-oo(eJ) is L o T T I -invariant, then
duAdq 0. (11.3.3)
(be 1fT Q Q,,(@) is Lo m-invariant, then
dq O aT = 0, (ll.3.4)

where <9 : Q,,(@') -> Qn_1 ( e l i5 the boundary operator; dual to the exterior derivative.
 Spectral Geometry 551

Proof. (a) The Lie algebra of LoTTI can be tautologically identified with a subspace of the
space of linear vector fields on R u n . A basis is given by the vector fields
X I $1312 -|- t g x , XI I $16xj - Xjgxz-7 l _< < M.

If u is Low-invariant, then

These equalities can be rewritten in the compact form du Adq = 0. Part (b) follows easily
from (a).

Denote by m the open set ={(

\C'- t = " )e . _la:l
R'n.+1 t })
,I ]Rl,'r7l
6+ '
> We have a pushforward
and by q+ the restriction of q to @+. The set 6+ is Lo m -invariant.
map
q+ : 91¢(@+) * 9A:(R)> k < 1 .
On the other hand, the map q+ : 5+ -> R is a submersion. Thus there is a push-forward
map
q+ : Qgptlg)-I-) -> Qlpt (R)
given by a formula very similar, in fact equivalent, to (11.2. 12). More precisely
.|. _ _ _
ii (u) - v q d q , '1
q+ q d q
where go is the so-called Gelfand-Leray residue of 17 along the level set {q+ : q}. This
differential form is represented by the restriction to this level set of any (n - 1)-form w
defined in a neighborhood U of this fiber of q and satisfying the equality
17 dqAw.
The above equation has many solutions w, but all their restrictions to the fiber are the same.
In particular, we have a dual pullback map
(QW IQKRI -> 9n(@'+).
The image of (q+l! consists of Low-invariant currents. We want to prove that, in fact, any
invariant current on @1+ lies in the image of (q-*l!.
Let 'l+ = {t > 0} C R " On this open set the functions (Q, 5131, . . . 7 23rd define a
global coordinate system. More precisely, 'l+ can be identified with the open set in the
(q, a l l , . . . , as)-space defined by the inequality
laval2 -{-q > 0.
Fix a function a E COOlIR1,TTll,
such that supp a C 7+ and for any A E R we have
supp a re {q = ,\} is compact,

Run
a(A, £ U 1 , . . . , ac7,,)da:1 /\ - - - /\ dace 1.
552 Lectures on the Geometry of Manzfolds

Now set

'Y Oddélil A • . /\ dayTTI E Q"-1<<r+>.

We have a projection formula
S = q*((q+)"yn5), vs 6 91(R) (1 l.3.5)

whose proof is an immediate consequence of the definitions of the various intervening

operations. We have the following result, [92, pp. 233-234].

Proposition 11.3.4. If T E Q,1(6'+) i5 Low-invariant, then there exists S Q Q1(Rl such

that
T (61+)!S~ (1 l.3.6)

Proof. Observe that if (1 l.3.6) were true, than the current S is uniquely determined by T
via the projection formula

S=¢]+l"YMT)-
We will prove that

T=(q+)'(q+(~wT)) C in (11.3.7)

The equality (11.3.6) follows from (11.3.7) by invoking the invariance of T and the fact
-
that the collection A 'I', A 6 Low is an open cover of 6>+ .
We set

S == q w m f ) , To (T- ( q + ) ' s ) | .
For any 77 G Q;lpt('T+l we have

(WTO) W T ) - ( q + t 7 T ) = ( n , T ) -(lq+)*lq!t7)»'yVWT)
( 17- (q-I-)*(q~7) Aw). (11.3.8)

We can write

B b(q, . C l 3 1 , . . . , xmldq A da31 A - - - A d x m Adq.

-
From the equality
( l
0
le ]-l-n
'l T)
l*l( /\
'Y )
we deduce

b(A,;u1,...,a3m)dac°1 A~-~Ad;37,, = 0 , VAQR.

For any fixed A the function

(3E17 . . . ,Ill *-> b(A, $ U 1 7 . . . , 23rd
 Spectral Geometry 553

has compact support. Using the Poincaré lemma for compact supports Theorem 7.2.1, or
rather its proof, we can find a family of forms Bq Q Qcpt 1 (no), depending smoothly on
q -E R such that, Vq 6 R we have
supp Bq C { la:l2 -|- q > 0} 7

b(q,a3ld;z:1 /\ - - /\ dat77 - d,,Bq,

where das denotes the exterior derivative on the space RM with coordinates at =
(al, . . . , mm). We can regard the family lgql as defining an (m - 1)-form B on the space
'T+ with coordinates (q, ac) and as such we have

3 d ( B /\ dql
and
.3.8) (11.3.4)0
0)l11 A
do) T
) do, I T) do re
(7],T
- <d(B 7
i

(B A Z (B
7

Similarly, if we set 6'_ = R1,m \ C+ and we denote by Q' the restriction of q to 6- ,

then we deduce that T 6 Q,,(€_) is Low-invariant if and only if there exists S G Q1 (R)
such that
T = (q')!S.

Corollary 11.3.5. Let T E Un lR1,m \ go. Then T i5 Low-invariant If and only If there
exist Si G Ql(]R) such that

S-l-IR<0 = S-|]18<o (11.3.9)

and

Tl0i (//i)!5i-Z

We denote by u(S+, 5_l E C ` 0 0 (R1'M \ Of the invariant generalized function associated

to the currents Si 5ati5j§1ing the compatibility condition (]].3.9).

Suppose that u Q cv-ool]R1,m \ 0) is both Low-invariant and supported in C+ 0. The

above corollary shows there exists a generalized function 'U E Q-OOIR) such that

'ul1R<0=0, U,l@+ q *U, UI@_ 0. (11.3. 10)

130111 the sequel we will use the notation u = (q+l*0 when referring to the generalized
function 'LL on the punctured space RU" \ 0 defined by (]].3.]0), where 'U G C'°°(]R),
s u p p l C REd. informally

lq+l*'U 'v(t2+-l22l2)~

A simple computation shows that for a more general result of this nature

(q`*)*u - (q )*
+
l )
554 Lectures on the Geometry of Manzfolds

where a : c-<><)(R) -> c-<><)(R) is the differential operator

d2 d
9% 4q +2 -.
dq2 n do
The above discussion leads to the following conclusion.

Corollary 11.3.6. Suppose that u 6 c1-oolR1,m \ 0) i5 Low-invariant, homogeneous of

degree d 6 (C \ 2Z<0 and supp u C C+. Then

u = eonst.(q-l-)*x§
Q
= constxi t2+
Q
I I;z:l2 l 7

and
d-2
l m
(q+)*x§ = (2d+ 2m 2)(q-1-l*X+2 7 on 1 0. ( I.I 3 . I >I

Proof. We know that 'LL = (q+)*'u for some 'U 6 C'°°(R) with support in [0, OO) . Since
u is homogeneous of degree d and q is homogeneous of degree 2 we deduce that 'U is
d
homogeneous of degree -2
6 R\Z<0. Theorem 11.2.7 implies that 'U is uniquely determined
by its restriction to do,
of), while Theorem 11.2.6 shows that the restriction of 'U to (0, oo)
is equal to constx+ .
To prove the second statement observe that
d _d d-2 d d2 2 d 2 d22
2 2
dqx+ X+ 7 qdqx+ .lllllllllll

2 X+
7

d-2
l l l l l l l l l l l l l l

d 2
since X+2 is homogeneous of degree 2
Hence
Q;
IP,,X 2 = (2d+ 2n - 4)X+22 = (2d+ 2 m - 2)X+22

For Re a > 0 we consider the measurable Low-invariantfunction

a n

q(t, In 2
7 l ) 6 i t C+
Re c77,(a) x
0, it,zul 6 R1'M \ C+
7

where
Fla;-ll
n : m -I-l, em(a)
'ii I`lalI`("' 2 +1 )
Observe that R+ is homogeneous of degree Cl - n.

Lemma 11.3.7. Let m > 1. The function R+ i5 locally integrable If a > m 1.

Proof. As we have seen in the proof of Proposition 11.3.4 on the open set m we can use
as coordinates the functions q, 5131, . . . , mm. In these coordinates we have
t 13012 -|- q
 Spectral Geometry 555

so that
m
51% 1
dt délii -I-
'i 1 \ |w12 -I- q 2\ lwl 2 + q do
and
1
dt/\d£L'1 /\ /\ dcvm = do /\ dace /\ /\ claim.
2\P+q
In the coordinates (q, as) the cone C+ becomes the half-plane q > 0 and the function
qa 2 n

\ 13012 + q
is locally integrable on this half-plane because

I <1
a
2
n

I < </ Re
7 Vq 2 0 , a3€]RM\0,
\ 15612 + (1 1231
and the function is the right-hand side is locally integrable if m > l and
R _
e 2 N>-1-i >Rea > m 1.

" i n the remainder of this section we will assume that m > l.

Let Re a > m - 1. Then R; is locally integrable and defines a Low-invariant gener-
alized function on R1,m_ We denote by Q55 this generalized function. It is homogeneous
of degree a - n > -2 and its support is contained in C+. Corollary 11.3.6 shows that R+
is the unique extension to R1,m of the homogeneous generalized function

d777»l0lX-l-2 (to W) = c,,,<a>(¢3 - l¢l2 II?" 6 c-°°(02e.1»m \0>,

where n = m + l,
r(@§'1>
do(al F +l cm(a) m
7r"II(a1
2

In particular
l
dmlZl 27r
m
2
l

Observe that for any a, b > m - l the convolution R+ * Rb of the locally integrable
functions R+, Rl is well defined. It is a locally integrable, Low-invariant function, ho-
mogeneous of degree (a -|- b) - n with support in C+ -|- C+ c+. We deduce from
Corollary l 1.3.6 that there exists a constant c(a, 6) such that
R+ * RE; = c(a, b)R++L.
We can be more precise.
Proposition 11.3.8.
R+*R€;=R++", ` v ' a , b > m - 1.
556 Lectures on the Geometry of Manzfolds

Proof. We follow the approach in [39, p. 160]. Observe that

R1,m
e-to -*_ * RW, ac)dtd;z: R1,m
b(t7 ;z:)e-tdtda:.

If we set

e-tR+(t, x)dtda:,

II
T(al
R1,m

then the identity et et 868

implies that

e-tR+ * R§(t, ;u)dtd;13 = T(@)T(bl

R1,m

so that

c(a, 6)
T(<»)T(1»)
T(a + b)
Now observe that
we)
cm(a) / R>o
6-t

(f Iwlét
(t2 - l:z:l2)

t
a m
2
1
do dt

a m 1
Um 1
/ R ,
6-t
(f 0
(¢2 - r2) 2 r'"lldr if,

where of denotes the area of the unit sphere in Sm 9

7rn-I-1
Qtr 2
Um m l
F( 3+ )
Now observe that
t 1 ta-1 1
(t2 7,2
a m
2
1
m - d T'
7'
Z
2 <1 5)
a m
2
1
sm22ds
0
ta-1 m a-m-I-l\
B
' 2
Hence
T(al B m a - m -I- l
InOO
_ 0171-1
7 e it" lot
cm(al - 2 2 0

U m - 1 r(a)p(%)r("-fl) l
2 Fla;-ll c,,,(a)
Thence T ( a ) = 1 and c(a, b) = l.

The above result show that the generalized functions Re a > m 1 satisfy the
convolution equalities
53+ *Qi = @++b, v Re a, Re b > m 1.
Using the equality (11.3.11) and the definition of do (a) we deduce that

'+
pa-2
'+ on 1,'rn
\0, v Re a > m -|- 1.
 Spectral Geometry 557

Invoking (11.2. 16) we deduce that the above equality holds on R1,77l We have thus proved
the following result.

Proposition 11.3.9.

53+ *Qi = 92++b, v Rea,Reb >m 1, ( l 1.3.l2a)

pa-2 1,'rn
r/
+ '+ on 7 'v'Rea > m-I-1. (ll.3.l2b)

We can now use the above equalities to define = + 6 C'°°(R1»*") for any a 6 (C by
setting
+ Q; k@++2k7
(ll.3.l3)

where is is a positive integer such that Re a + 2k > m - 1. The family

C 9a1->%+ €C"°0(]R1'M)

is holomorphic. Moreover, for any a € l C the generalized function %+ is homogeneous of

degree a, it is LoT"-invariant and

supp + C C+ .

Moreover, the equalities (11.3.l2a) and (11.3.12b) extend by unique continuation to all
a, b G (C. The family lgilaec is called the wave family and the generalized functions Q;
are the Riesz generalized functions, [114].

Proposition 11.3.10.

(a - 2>a@+ = -x,92+-2, pa Q Q, 1 < i < m . (ll.3.l4)

Proof. To prove (11.3. 14) we observe that both sides of this equality are generalized func-
tions depending holomorphically on CL. Hence it suffices to prove it only for Re CL >> 0. In
this case
a 2 n

a@+ Z dm(a)@(q+)*x: = dm(@)(3m»q)(q+)*x 2

_
_
do(al
-2.21% dm(a _ 2) d,n(a _ 2l(q+l*X
a 2
2
TL

21/1i
dmlal Q -2
dmn(a-2) +
F(";1)F(a- 2)92+-2
251% 251%
a 1)2(a 2) +
Q-2
a
z
2
53+ .
r(a)r(@21)

Theorem 11.3.11. *be unique fundamental solution of on 1,'rnJ m > 1, with support
in C+ is the Riesz generalized function Qi.
558 Lectures on the Geometry of Manzfolds

Proof. Observe first that the function

F("'$1)
a +-> drn1(a) 'ii"II(aI
m 7 Rea> m 1
2

extends to a holomorphic map on (C.

Indeed, the function a +-> gadefines a holomoiphic function on (C with zero set Z$0.
Each of these zeroes is a simple zero. Moreover, the derivative of at a = -k E ZS() ga
is (-Ilk kl.
The function a »-> F l a - l - l l admits a meromorphic extension to (C. All its poles are
simple and are located at -1 -|- 2Z$0. Thus the function dm71(0) extends to a holomorphic
function on (C. It has only simple zeros situated at ZZ$0. We obtain in this fashion a
holomorphic family of generalized functions
a n

@9a+->%+ dm(@)(q+)*x+
2
6 C1-°°(R"" \ 0).
Note that
Re a 1.
By unique continuation we deduce that the above equality holds for all a E (C. Using
(11.3. l 2b) we deduce that
1,m
r/
+ on \0, `v'a€

Since dm(a) Ofor a E ZZ$0, we deduce

f; = 0 on R1,m \ 0.
Hence
supp $3 = {0}.
Since $3 is homogeneous of degree -n we deduce that %3 C60, for some constant C.
Then
@+=@+*@9=c@+*60=c@+.

This shows that C = l.

We set E Qi. Using Proposition 11.1.28 with f = 0 we deduce that there exists a
C2 -path

(0, o f ) 9 t »-> E ( t ) Q c-°°<Rml

such that the limits
E0 := 1 0 E ( t l , ET t\0

exist in c - ( m m > .

Proposition 11.3.12. The family ( E ( t ) )t>@ i5 a Cauchy fundamental solution of the wave
operator, i.e., it satisfies the conditions (II.3.Ia), (II.3.II9) and (II.3.Ic).
 Spectral Geometry 559

Proof. We follow the approach in [66, Sec. 6.2]. The equality (1 l.3.la) follows from the
fact that = 50. We have to show that D = 0 and 6 = 50. To achieve this we first
need to have a clearer idea on the nature of the generalized functions E(t).
Let us compute the restriction of
l * 2271, l * 1 2m
2,,,(m-1)/29 X-|- 2,,T(m-1>/29 X-|-
to the punctured space Rl-|-7r \ 0. Note that
Ivq(t, H1)l = 2 \ t2 + l6212~
For t > 0 we have
q'1(sl re c+ \0 = { (t, Go); t \ S +lwl2, s+la:l > 0} s > 0 . 7

Thus q-1 (3) is the graph of the function

uS : RM -> [0,oo), 'u,S(a:) \ s+ l/z:l2.

The volume element on this graph is

I vI + 21wI
S + 116I Idol 8 + l:ul
Note that along q-1(s) we have IVql = 2 i/ ' s + 2l:z:l2 so ,

1 1
ldVq-1(5) | + I 12 id:z:l.
IVQI 2\ S

If go Q @lR1+m \ 0), then

l
q*(/)(5l 2 (/)( (5 + l2:l2)1/2 ,ac )(5 + |w|2)'1/2|d:v|-
Using spherical coordinates (T, w) we have
pdf 'rd.
:c = re t = \,5 + 'r2, 7° = V / 2 - 8, .dt = tl/2 (8_l_7'2)1/2

We set
Tm 2
¢(t, T) (p(t , 1 u)ldwl
lwl=1

and we get

¢>(t,(t2 - 3)1/2)dt.
t2>s

Hence
1
(E+ ,(,0)
X-m)/2
Z
l'fn-1
)/2
( 7 ¢(t, (R - .>l/2) )it, WE Q Q<R1+m \0>.
47T S<t
2

( 11.3.15)
For any Q E QQ" we set
,am 2
Z(t~) ¢(f~~»)ld»»l.
l(»JI=1
560 Lectures on the Geometry of Manzfolds

The equality (1 l.3.15) shows that for t > 0 and W Q @(R") we have
1 1-m
(E+(t),¢) Z 47 r(M _1 )/2 ( X+ 2
(s),2((t2 - sr/2) ) (11.3.l6)

We rewrite (11.3.l5) as

(E+»(P) 0
( E ) , (Mt, ;z:))dt, WE 6 Q " i 0).

We claim that

(E+=(/') ( E ) , (/.(w:))dt, W) e Q<R1+m>. (ll.3.l7)

0

To see this observe that since limt\0 E(t) exists weakly we deduce from the uniform
boundedness principle that the right-hand side of (11.3.17) exists and defines a homoge-
neous distribution of degree l - n on R" = l+'". This coincides with the homogeneous
distribution E+ on the punctured space R" \ 0. We conclude from Proposition that they
coincide on R" .
To show that E0 - 0 and ET 60 we argue as in the proof of Proposition 11.3.1. Let
(P = (Mt, or) G 9(R1-|-m)- We have

SOl0»0l=(50,0»(P)=( +»S0)=( += 90>

(11.3.17)
lim
e\,0
l° ( +(t)»( (0)(t»-))dt
OO

lim
e\,0
L (am), ( A , -) )it -|- lim
s\,0
(E+(15l» lapllt, -) )dt

lim
e\,0 / ( A E t , (p(t, -I )d¢ - lim (E+(<'5)» a
€\,0
, -) )
/
5
OO

lim (atE
t
t
5\,0
o l lt-l)d
>
6
t
'lalim§ - ) )it
OO

( E
= - O'
a*SU(" » - ) +
e7\,0 ( A E * 'S 0 lt 7
6'
OO

+ e7\,0
lim(3¢E5, § 0 ( € 7 - ) ) -|- lim
e\,0 S
OO

= -(Eo, 3¢90(0» -) ) + (Eo (MO, -) ) + 6\,0

lim (@?E + Am, (/)(t, -) )dt

I -(EOS 3¢S0l0» -I ) -|- (EO) (MO, -I

Since the map

QOR1-I-ml 9(p(t,a3l +-> ((,f2(0»;vl»(pQ(0»22))@(]R"') X QQRmI

is onto, we deduce E0 - 0 and E0 -60.

Remark 11.3.13. The generalized function Q; is a Cauchy fundamental solution and so

that it is determined by a smooth family
(0, OO) 9 t +-> Qgltl E C'°°(R"")
 Spectral Geometry 561

satisfying
(of + A)@;(¢) = 0, t > 0, @i(0+> = 0, a@i<0+> = 60. (11.3. 18)
Thus Q; describes the propagation of a wave generated by a disturbance at t = 0 localized
near (or at) the origin. If ac Q Run 0, then go;(t, 32) = 0 for txt
> t. In other words, the
disturbance at 0 takes t = txt seconds to reach the point x.
I f m i s o d d , m = 2 k + 1 , k > 0,then
60(1¢-1)
X+k
so that

Qi Z dm(2)xJ(ti - W)
has support on 3C+. In this case disturbance leaves the point ac immediately after it reaches
this point. This is sometimes referred to as Huygens' Principle.
If m is even, the disturbance will linger at Hz: forever, with diminishing strength.

Here is a simple but important application of the fundamental solution

Theorem 11.3.14. For arbitrary gag, (al E RM and f G C°°([0,ool X Run) the Cauchy
problem
m
u=f, on (0,oo) X 7
(ll.3.l9)
u(0,a2) = go0(a:), 3tu(0,:z:l = go1(a:), Va: € RM 7

has a unique solution u 6 COO([0, ool X Run) given by

t
d
* 901 + -§z§(¢l * 901 + §22~l¢ - * f(s)ds.
u(t, -) = @L(t)
dt ( ) 0 5) (1133.20)

Proof. Existence. From Proposition 11.3. 12 we deduce that Q; is a Cauchy fundamental

solution for the wave operator. Using Proposition 11.1.28 we deduce that the correspon-
dence
(0, OO) 9 t »-> %§(t) 6 C"°°(]Rm)
is smooth. Moreover the support of Q; (t) is contained in the ball {l:13l < t}. Thus,
the expression in the right-hand side of (11.3.20) is a well defined smooth function on
[0, oo) X Run. The fact that u given by (1 l.3.20) is a solution of the initial value problem
(1 l.3.19) follows by direct computation using the equalities
m d
Qi O i n (0, OO) X 7 92i(0+- = 0, -92i(0+) = 60.
dt
Uniqueness. There are many approaches. Ours is based on finite speed propagation of
solutions. We follow closely the presentation in [32, VI.6.l].
For every T > 0 we consider the dependence cone (see Figure l l . l )
• ) }
CT {lt o' T ] ' IQ;l < IT -¢
tél
'i
>§
562 Lectures on the Geometry of Manzfolds

r
A

CT

* an up
all :I :I 1 :I _ _ ___ _- - `
I

In
¢ s
s

l >
x2
.f

Fig. l 1.1 The dependence cone CT

We denote by 3+ CT conical part of the boundary of the cone and by <90 CT the base of
the cone, see Figure 11.1. The unit outer normal along the (smooth part of the) boundary
of CT has the form
n

n 0,03t + (Zing .
i=1

Along the conical part of the boundary, described by the quadratic equation

(T to2 lw12 7

the components of the outer normal satisfy the conditions

m

a02 a 2, as > 0. ( 11.3.2l)

'i=1

Along the base of the cone, described by {t = 0, laval g T}, the unit other normal is
fig= -3¢. For any smooth function u = u(t,a2l we set
m 771

Jo : Utd t -|- uxdat, ldul 2

2
Ut -|-
2
Um »

i=l i= 1

We have

2'Ll,t u = <82 I ) -I- 3 (-Zulux1).

'i=1
1-}-'rn
In other words, Ut 'LL is the divergence of the vector field on 9

XU = I ) a -2
i 1

Hence
(Ut u)dtd;u div X"dtd;v ( X " , n ) dS
CT CT ACT
 Spectral Geometry 563

m TTY

/ ( 2 22
2
IJu(0,x)l2dx -I- 0,0 'Llfxi 0,iut'Ll,x, dS
|0U|ST 3* CT 'é=0 i= 1

(11.3.21) l&u(0,x>l2dx _|_ 1

(a0ux1 - a u t ) 2 ds.
Ixl$T
a*cT CL()

We deduce that if u = 0, then

lc2u(0,x)l2dx llllllllllll
(a0umi - a¢'u,t)2 ds. ( I I .3.22)
l0UIST a*cT CL()

If additionally
cZu(0, Lil = 0,
then we deduce that along 3* CT we have
aguxi - a b u t =0, W : l,...,m. ( I I .3.23)
The vector fields
VI -Qi6t+G03xi, i=1,...m

are tangent to 3*CT and they are linearly independent there. In other words, they define a
framing of the tangent bundle TEL CT. We can rewrite the equalities (11.3.23) as
u 0, 'v'é=1,...,m.
They imply that 'U is constant along 3* go". If we assume additionally that 'u,(0, ac) = 0,
then we conclude that 'U E 0 on acT, so that u(T, 0) = 0, VT > 0. Applying the same
argument using instead the translate C+T(CUol = SU0 + we obtain the desired uniquenessCI
result.

Remark 11.3.1S. (a) The above proof of the uniqueness result justifies the name depen-
dence cone we gave to CT. If u, 'U are smooth solutions of the wave equation satisfying the
equalities
tI,(0,a;°l - v(0,a;°l = u¢(0,a2l - v%(0,x3 = 0, Vlad < T,
then u(0, T) = v(0, T). In other words, the value at (T, 0) is only determined by the initial
data inside the dependence cone.
(b) There is an alternate proof of the uniqueness, based on the so called energy method
and in which the dependence cone also play a central role. Suppose the u is a C2 solution
of the wave equation. For t < T we define
l
E ) : l@Zu(t,>l2dx.
5 l$IST-t
An elementary computation (see [44, §2.4.3]) shows that
0§ Et g Et), W Q (0,T)»
Ifu(0, ac = ut(0, so) = 0, then E'LL(0) = 0 so that E'LLltl = 0, and
@Zn(¢,x> = 0, 'v'lxl _ T - t.
564 Lectures on the Geometry of Manzfolds

Consider the generalized functions

mi - swf) Q C1-°°(R1»m \ 0).
Fix 500 Q R " \ 0 and consider the open set
U0 := r (t,;u) Q ]R1"'", 1513 - :col <
1 C R1,77z \0
L QC"
containing the line
L
Mo Z
{(t C
l 7M
\0.

Fig. l 1.2 Behavior near a temporal line.

Note that
l
supp X+(t2+ - la2l2)lu0 C QQ; t> 2l330 re Uo.
In the open set 8% we can use as coordinates the variables
q, :u
1
7 ,113
m
.
The map q : % -> R is a submersion with image an open interval 10 containing the
support [0, to) of X+° We have

X+lt2+ n i q*lX+IIol'
We want to define the restriction of X+ ( t2+ - l;vl2 ) to the line LU20 .
On 850 we have 15+ = \ q -|- |;u|2. For Re a > 0 and (p 6 Quo) we have
1 (0( \ q + 1wl 2 ,w) dqdac
a
(x"+(t2+ - 1112 ll@"0)l0lt)Xl ) I`(a+ 1) 850 9+ 2 \ q -|- 13312

/l32-m0l<l:1:0l It
R
x+(q),¢x(q)dq day, ¢m(<1)
q + l@2l2,22l
90(\

A q + 1612
E FOR).
 Spectral Geometry 565

This shows that for Re a >> 0 the restriction of X+ (t2 - 1HJ12) to LWo is given by
1
(X"+(t2 umm ,,,<o) x"'(q),
+ 2wI+1w012
90(\ q+ l630l2) 7 (11.3.24)

'v'(p Q @OR). The right-hand side of the above equality depends holomolphically on a and
we define X+ (t2+ - laval2) to the line L5130 for any a Q (C by the same equality (11.3.24) using
unique continuation.

Remark 11.3.16. This is admittedly a rather ad-hoc definition of the restriction of a gen-
eralized function to a smooth submanifold. There is a more conceptual definition of the
restriction that works for a large class of generalized functions satisfying certain microlo-
cal conditions, see [66]. Our ad-hoc definition agrees with the more conceptual one, but
we will not need this fact in the sequel.

Proposition 11.3.17. Let n := m + 1. Then

n

113% <1 X-|- (t2+ a 2

2) X-|. (t2 W) '
2
2lxl
2a 2 m
7 'v'a€(C (1133.25)

where IXI"' E C"°°(]R) is defined by (]].2.7)

Proof. For 1130 ¢ 0 fixed we set c = C(U10l = l$0l and we denote by (to 02 the
restriction of X+ pa (to II12) to the temporal line
,

LWo = {(t,:1:0) E RE"**; t o R } .

For Re a >> 0 we have
3tX+-5(£2+ ' 02) = 2t+X+-1-%lt2+ ' 02) Q 0-°°(R),
a (X 2
(to - 02) X+
a 2
lt2 - c l = 2i+1+-1-%(i2+ - 02) + 2¢_X+°1° C
2

1 %ltd _ 0 2 ) .
2ltlX+
By unique continuation we conclude that the above equality holds for any a, where the
-
multiplications ltl X+(t2 - 02) are well defined since the singular supports of ltl and
X+ (152 - 02) are disjoint. We deduce that for any up E QUO) we have

(ltlx+- 1 - 22(¢2 - C2 ) , o ( t ) ) = (t+x W - @2),¢(¢) + so( )

(q = to - 02, 15+ = \/Q + 02)
(11 .3 . 24 ) (VG + 02)X+_ 1
+ @2))
n
2
(Q),
2 \/q + C2 QQ( + 02) + ¢(-
1 n 1
X+ 2 (al), 2 SO ( ( SO N N )
566 Lectures on the Geometry of Manzfolds

Observe that 50/)(t) -|- cp(-t)) is even so that according to Whitney3

-((/2(t) -|- w(-U) 17(t2), (11.3.26)

for some 17 E @(1Re.>. Hence

) + so( _ \ / q _|_ 02 ) )
1
2
( so( $1 + C2 '7(q+c2),

lim
c ( )
(312 lX+- 3 (ti - 02) X+ 2 ( - @2)),»(¢)) Z ( lqmlql)
We want to express ( x i (al), t1(q)) in terms of the function (p(t) satisfying (11.3.26). Note
that the function

a *-> (x"+(q)» v(<1))

is holomorphic, so it suffices to determine it for a in an open set in the complex plane. We
assume Re a > 0. In this case

(xi (Q), '7(ql)

1
I`(a+1)
I°° qa 1701) d<1
0
OO
Q 52

I`(a-I-1)
1
1
/ 0
52a-l-1((0(5) + s0(-5))d5

I`(a-I-l)
R
IsI2a +1 (Ms) ds (l)cl2"+1, )~
(of)

The map a »-> <lxl2a+1 , go) is holomorphic and we can invoke unique continuation to
conclude the validity of the above equality for any a. The equality (11.3.25) follows from
the above since
2 a 1 ) l =2a- l -n=2a-2-m.

Remark 11.3.18. The equality (11.3.25) implies that

6% - QUO I¢ l( t, 0) = 2dm(2k)lxl2I"'1""(t), ( I I .3.27)

where
2.7-1
F l2k72-1) _ H kj 1 2 l
dm(2 k) m-l 7
7rmI`
2
(21<)' (2k-1)! -2'rr 2 <f-1>!
so that
1 k 2
Qin = m 1 X+ (to emf). ( 11.3.28)
'ii 2
(l - l)!
oWe follow [56, p. 343]. Let i T L ant" be the Taylor expansion of (pg (t) = ((p(t) + up(-t)) at t = 0. Since
$00 is even we deduce an = 0 for in odd so that (pg ( t ) = E a2'rL 152". By Borel's theorem, [66, Thm. l.2.6], there
exists a smooth function u(8) whose Taylor expansion at .s = 0 is in a 2 n 8 N . Then v(t) = S00 (t) - u(t2 )
vanishes with all its derivatives at t = 0 and hence 'u(v-t) is smooth. We set 17(8) := u(.s) -4- 'v(1/gl.
 Spectral Geometry 567

We deduce that when k = l we have

l
(W - Qin, 0) = 2dm(2)lxl'M(t)
'ii 2
-TTL
(11.3.29)

Using (11.2.1 lb) we deduce that the Fourier transform of <9¢ (Q - @3L)(t, 0) with respect
to the t-variable is

[a (923 - Qf m-1
IxI( )(T) = e r r ( % ) I I
27r 7' m 1
2'i)(t, 0)] (T) m 1
Zmtr 2

2'/T Um-1 mum

'ii I7' Im 1
'ii
(2'rr)71 I7'I
m-1
or I7' I
'rn 1

(27T)"1`(%) (21T)'"
Hence
1 2 '2 1 d
i [@¢<%+ -%+)(t,0)] (T) - - l w m f m sign 7' ) ( I I .3.30)
2 dT

11.3.3 Local parametriees for the wave equation with variable eoejfieients

We want to extend the identities (1 l.3.l2b) and (1 l.3.l4) to slightly more general situa-
dons
Consider a new linear change of coordinates on Run, 11: = Ty, where T : RM -> RM is
a linear isomorphism. More explicitly,
:ui tryJ.
j

We denote by lgjk ) the matrix associated to the operator (T*T) in the y-coordinates and
by (gawk) its inverse. More precisely

'i

We set
lgl := det(gjkl = det(T*T).

Then in the y-coordinates we have

A Ay
a
•
-ay (off ask
Observe that
. k
114 2 2
lyl g if/y , q(t,23l t2 - 13312 = kg(t,yl = t2 - Ill 2
g'

je
Then

(65 + Ay)9?+(t, Tel I %+'2(t, T y ) , (11.3.31)

and

(a - 2) 9""'<9yk%+(t, Ty) = -yi%+l2(t, Ty)~ (11,3.32)

k
568 Lectures on the Geometry of Manzfolds

If we multiply both sides of the above equality with gin and then we sum over j we deduce

(a -2l3y@@+l_lt,Tl/I r AM/J =@+-2(15»T?/)~ (11.3.33)

Let us observe that if 60(;v) E 0-(><><Rm> denotes the Dirac generalized function concen-
rated at 0, then

5olTY)Riga-Wy).
I T*50 I
( I I .3.34)

For Re a >> m we define %+(t, lyl9) E c-oolR1,m \ 0) using the technique in (11.3.10)

Q , lyly ii do(a)lqg+)*x
2
•

The generalized function %+ (t, lylg) admits a unique homogeneous extension %+ (t, lylg)
to ]Rl,'rn which coincides with %+(t, Ty). We can then use (11.3.31) to show that

' + (t, ; %+ (t, TYllR1'M'\09 Va.

For this reason we will use the more suggestive notation %+(t, lylg) when referring to
§25;(¢, Tel.
More generally, consider a Riemann metric g = low ) on RT",
g gzj a n d .
M
The Laplacian in this metric is the differential operator

Ag I - Z3x ( -Qjk@xA¢ )+ Z beam; 7

j,k J

l
bi log Ill 5 E9""'3
k
frk 10glg1

Suppose that (ii) are geodesic coordinates near 0. This means that the rays through 0 are
geodesics and thus, if we set Ias I9 = distg(a2, 0), then
L
2

1231g I . 9 jk :( 0 )3
3J ( I I .3.35)
J ,k

MOI'€OVCI,4

Z9jk(x)wk Z91k(0)wk 7 (ll.3.36)

k k

9j1<2(2Ul
_
- gy1J0) - Of laval l,
2 .
've],k. (l1.3.37)
The equality (113.36) is an algebraic rephrasing of the Gauss Lemma, Lemma 4.1.29, stating that the the radial
vector field in 11:73sc is the gradient of the function lxl§.
 Spectral Geometry 569

We set
¢(11.3.3
6) l
bf llgjk l0 w)~
h(;u) )x,1 (11.3.38)

• •
§9<v1o§|9|,
j,k

Observe that if f 6 0 1 , then (11.3.36) implies that

9"1"(11)<8lg@kf(l12l,) 9"k(22)<8'xklH1l§ f'(l=u§,>

k

M
9""(x1)9¢k(33)x@ f'(l=vl§)
(~ ) f'<l=u§>
(Sqj

gj'"(0)g~(0).: f'(lwl§) g(0)@f(lwl§).

M k

We deduce from (11.3.31) and the above identity that for Re a >> 0 we have

as (9"%) Q mg) = §2+-Q(f, lxlg)- ( 11.3.39)

j,k

By unique continuation we conclude that the above equality holds for all a E C.
More generally, if 'U = u(a:) is a t-independent smooth function, then we deduce from
(11.3.32), (11.3.33), (11.3.38) and (11.3.39) that for any a go 2 we have

(33 -|- Ag) (~(x)2i(¢, l£UI9)) I U-l-2 + lA9'*"l=L

1 ( I I .3.40)

When a = 2 we have
m l

93/c2(z, 1fv19) = (f1§,l*f2» f2 71' 2 ( t , as) 7; 0.

1- - .
n
Since X+ 2 IS homogeneous of degree 2 - n on Rlnn 0 it admits a unique extension to
Rlyrn
as a homogeneous generalized function of degree 2 - n. The generalized function

=»(q;)*x£ Q C1-°°(R1»m \ Of
is homogeneous of degree l n and thus it has a unique homogeneous extension to R " .
The equality
_|_ _ n _|_ * _ n
@x.(q9)x1 2 --2;u3(q9) X 2

over the entire space ]R1,7r is a consequence of (11.2.l6). We deduce

m 1

<w2~2(<, 1w1g) = -Ztr' 2

w(q;)*x- 2 (11.3.4l)
570 Lectures on the Geometry of Manzfolds

(of + Ag)(u()@i(t, 1wlg)) = u(0)lg@l-6@ + (Aguwi

/ \
m 1 (11.3.42)
-27r' 2 hu- 2 xJ3Myu (q§)*x+5-
j

Thus, if up, . . . ,un are smooth functions defined in a neighborhood of 0 6 Run, then
N N

we + Ag ) k=1
wei I (03 + Ag Mei + (65 + A )
k=2
up(w)eti'"
N N-1
l l
<Agu,>e2i'f -|- Nik 'LLk_}_1 - hU1<:+1 -|- l' • V'LL]¢_I.1
k=2 k=l
1 m 1
-|- U1 (0)|g0|"50
2 -|- (Agu1)@ - 2'/T' 2 (hut - Zac - Vu1) (qg+)*X'%

= u1(0)|9@|-o + (Ague)ei"'
N-1
1 1
+ $3 Ague + 'U,]¢+1 - hUk+1 + 113 . VU1<:+1
k=1
m 1
- 2'/r` 2 (had - 2:13 - Vu1) (q;')*X'%.

Proposition 11.3.19. For any smootnfunetion h E Cfoo ( R M ) such that h ( 0 ) = 0 there ex-
ists a unique sequence of smootn functions U1, UQ, . . . , Un, . . . defined on a small neign-
bornood of 0 E RM and satisfying the deferential equations
U1(0l = l, 2:12 - VULG = hUt, (U1,h)
1
. VUk-I-1 +
113
( 2k
h
) Uk-l.1 = -AgUe, Vi 2 1. (Uk,h)

Proof. If we work in spherical coordinates (T, 0 ) in RM \ 0, T' > 0, 0 E Sm-1 7 then

-
x V = r d . The equation 2:13 VULG = hUt then reads-
1
6,1/1 - h U Z •
2v 1
0, we deduce that for any 0 6 Run, 101 = l we have
Since h(0) =
h(7°0 I T'
/' 36h(t'r°0 )dt = 1°g(1°0 ),
0
where g is a smooth function in a neighborhood of 0. Thus
hltft' 0 )
U1 IT'0 ) = U1(0l exp
(f 0
- t 0 dt
2g( ) ) exp
(f 0 215
dt
) •
( 11.3.43)

Lemma 11.3.20. For any lc 6 Z>0, and any smooth function w defined in a neighborhood
of 0 6 Run, the dnterential equation
h
( )
has a unique solution defined in a neighborhood 0. of
w
 Spectral Geometry 571

Proof. To see this we rewrite the above equation as

h
1°3,.U-|- k -
( 2 ) U=w,
and, if we set,
»r _h
V:=Ue to 2tdt 7

then we deduce

T'31,V -|- kV - we' to" - d th

2t

Multiplying the above equation with 7 k - 1 we deduce

h
3r(Tkv ) I Ar'k-1,w€- fo" 2t
dt
7

so that
T'
1 h(t
2?6 ) dt dp.
V(7v0 ) lllllllllll

k (0""1'w(00 )@-f"0
7° O

Proposition 11.3.19 now follows from the above lemma.

If we choose U1,...,UNas in Proposition 11.3.19, where h is defined by (11.3.38),

then we deduce
N
kg; + Ag ) Z Up (w)%3'" l90l-%50 + lAgUnl§??*.N. (11.3.44>
k=1

Observe that for any positive integer 1/ we can find N = N(1/) such that
(AgUNl,§:)2N 6 CV.
n
More precisely, since XN`% is homogeneous of degree N 27 we deduce that
2 N - n > 2 1 / = > % L N Q C"

Suppose now that (M, g) is a smooth compact, connected, m-dimensional Riemann

manifold. Let ldlQ,l be the canonical volume density determined by the Riemann metric.
In local coordinates (al, . . .,;u'")

ldvgl lda31 da3""l.

Remark 11.3.21. We want to discuss a rather confusing fact that unfortunately permeates
the more analytically and less geometrically oriented texts.
A metric g on M induces a natural bijection

c°°(m) -> 0°O( IAlm ), f +-> fldvgl»

Invariantly, the generalized functions on M are continuous linear functionals on the space
Q O O I IAI M ) of smooth densities on M .
572 Lectures on the Geometry of Manzfolds

The Dirac measure concentrated at a point :co E M is the continuous linear map

6 : c(m> -> R, ( Q ( f ) = f(fuol» o f Q c°°(m).

The Dirac generalized function on M concentrated at :co is the linear functional
6M=0°°(IAIMl -> R, 6M(fldvgl) =f(m0l.
In local coordinates around 1130, the Dirac measure can be identified with a multiple of
the Dirac generalized function. More precisely, we have the equality of locally defined
generalized functions
l
5M0 = 5,,,O.
v 191
To understand this equality, consider a family of nonnegative smooth functions

qb6:M-> [0»oo)» e > 0 ,

such that,

gb5ldvg I la SUPP (155 C B@(1Iol» Ve>0.

M

These functions converge to the Dirac generalized function AM.

If we choose local coordinates (ac1, . . . , mM), then we can think of $5 as functions QUOS (515)
defined on RM and supports contained in tiny balls centered at the origin. As such, they
satisfy

/ wall(x)¢(w) Idol l.

The probability measures \,l9ll.lQ5el$L') Idivl converge weakly to the Dirac measure 60.

We denote by A = Ag the Laplacian associated to the Riemann metric g. More explic-

itly, if we fix a point p 6 M and local coordinates (x1, . . . m
defined in a small geodesic
ball Up centered at p such that

f12'(pl 0, i 1, . . . , m .
In these coordinates, the Laplacian is described by
1
A= - as ( /ggka ) a,(gfka,)+ be 7

j,k j,k j

where
bi -CV
l i loglgl.

Let
s:M X M -> [0,oo), (x,y) +-> sy(a2) = distg(;u,y)
be the geodesic distance function determined by g. For y 6 M we denote by fy the radial
vector field on TyM . Using the exponential map exp : TyM -> M we can regard fy as
 Spectral Geometry 573

a vector field defined on a small geodesic ball Up centered at y. For fixed y we obtain a
generalized function
(t, at) *-> %a(t, sy(a:) I, distg(a3, y) < 0
which in normal coordinates at y coincides with the wave function ,@+ on
RxTyMRxRM_
By choosing an orthonormal frame of TM over UP0 we can identify TyM with l ' " and
then we can regard the correspondence y +-> %2"(t, 5y(rv)) as a smooth family of general-
ized functions defined on a neighborhood origin in R X ]R"".
Equivalently, we can identify a neighborhood of the diagonal in M X M with a
neighborhood of the zero section in TM. For fixed y 6 M the generalized function
%?gl(t, sy(x)) is defined on T y M .
Identifying UP0 with a neighborhood of 0 E Run, we can further view this family as a
generalized function on a neighborhood of the origin in R X RM X R'rn,
%3'*(¢, sy(w)) 6 0-oo( ( e, c) X Upo X Upo >-
Moreover, this generalized function depends holomorphically on a E C. Observe that
if Re a >> 0, then we can identify ,%a2"(t, sy(:z:)) with the classical locally integrable
function
TL

dml2alx""' (to - dist9(:12, yl2

2

Thus

%')§"(t, Sys-ll Z Qalt 5x(yll, Re a >> 0

and by unique continuation we deduce that the above symmetry holds for any a,
@2a(¢, 8y(a2)) = Qian, s,,,(y)), Va E (C. (1 l.3.45)
Using Proposition 11.3. 19 we deduce the following result.

Proposition 11.3.22. There exist

.. a positive number C = e(M)

a sequence ofsmoothfunetions U, 6
> 0 and
coo (M X M),u l ,2,

with the following properly:

for any y 6 M, the functions Hz: +-> U,§(a3) U, (33, y) satisfy the equations (U1,h) and
..II

(U1<:,h)f0t by(;u) < c, where

h=9(V10>lgl , y ). ( I I .3.46)
We deduce that for (t, v y) Q ( oo, c) X M X M we have the equality of generalized
functions
N
(63 -|- A ) Ukl£E, y>§z?g<(¢, S ) )
k=1 (11.3.47)
1

= lgyl--6g
2
+ ( A n n ( w , y ) )§2iN(¢, s ) ).
574 Lectures on the Geometry of Manzfolds

Above lgyl'%5y 6 C'°°(]R X M ) is the Dirac-delta generalized function on R X M

concentrated at (0, y) 6 R X M . More precisely (see Remark II.3.2I) for any function
u 6 C§°(]R X M ) we have
<l9yl-85y=UldV9l ) = um).

11.4 Spectral geometry

11.4.1 The speelralfunetion of the Laplacian on a eompaet manifold

Suppose that (M, g) is a smooth compact, connected, m-dimensional Riemann manifold.
We denote by ldvgl the volume density on M determined by g and by A the associated
Laplacian. The L2-inner product of two smooth complex valued functions u, 'U : M -> (C
is defined by

(u, u) u()@(x)udvg(x)u.
M
Note that it is conjugate linear in the second variable.
Fix an orthonormal Hilbert basis of L2 (M) consisting of real eigenfunctions ('Pnln20
of M such that if
Ask As:\IIk,
and
0=A0 < M$/\2$
We define
£ = 5 M g : [0,oo) X M x M->]R, é"(7',:z:,y) We($)Wk(/)~
As <T

The function <5"Mg is called the spectral function of the Riemann manifold ( M l - If we
denote by SHT the space
:HT := span{ Wk; As < 7' } 7

then the function

M X M Y (22,y) »-><§T(x,y) =é"(w1»y)
is the Schwarz kernel of the orthogonal projection PT : L2 ( M l -> UV, i.e.,

(P¢u)(;u) M <§'(7', I , y)u(y)ldVg(y)l» Vu Q L2(M,9)-

We set
N(7') = Ng(T) := dim TCT 2
AsST
l 2/
AsST M
*I'k(.1)2ldVg(=v)l
(ll.4.l)
<§'T(ac°, ac)ldVg (;u)l.
M
 Spectral Geometry 575

For any positive integer k, any p E [0, oo), and any u E cOO ( M ) we set
1
p

llullL;(m) a
•

(if j<_I< M
no" ldvg(rv)l

Theorem 11.4.1. For any I/ 6 Z20 there exists C , > 0 such that

II<§%llc~(mxm) S C,,T"+, VT > 0. (11.4.2)

Proof. The proof uses a refined Sobolev estimate.

Lemma 11.4.2. If > U -|- then there exists C C'(k,1/) > 0 such that for any A > l
and any u 6 Goo ( M ) we have
,\'<-ff-%llull2c~(M) S 0(llullii<m> + A*llulli2<m> )~ (11.4.3)

Proof of the lemma. We will prove that there exists a constant C' > 0 such that for any
u E C0 (]R'"') and any A > l we have

Z Z
As-"-%
l0l=v
supl@;~~(x)l2 g C
x (/( Run
l0l=k )
l@;'u(w)l2 ld.¢¢l + As
fRun
lu(wll2ld5vl
(11.4.4)
)
The inequality (11.4.3) follows from (11.4.4) via partitions of unity, so it suffices to prove
(11.4.4).
We make the substitution
at :
Then
Id m
$x @ 3 A
33: =3/
19
7
"al = /\-
and if we set
v(yl u(x) u(A g),
then we see that it suffices to prove (1 l.4.4) only in the case A l for the function u.
From the Morley inequalities we deduce

10(l V
sup l<9""u(2¢)l2 S C
as /( RM
Using Parseval's formula we deduce that there exists a constant C' > 0 such that
lol<_1<
l<9"v(H1)l
2

) lda:l.

/( ) www S C '
(/( )/ )
2 2
l<8l°'u(22)l l<8l°'u(22)l ld.CUI'l' Run
lu(22ll2ldHll 7

Run l0.l$I<: Run lol=k

for any u E 03° (RTTI).
576 Lectures on the Geometry of Manzfolds

Consider now the operator L := A -|- l. This is a self-adjoint elliptic operator with
trivial kernel. We deduce from Corollary 10.4.12 that, for any k 2 0, it induces bounded,
bijective, linear operators

L L I M ) -> L i m )
with bounded inverses. Hence, for any k 2 0 there exists a constant Co > 0 such that

llullLm> § 0kll L UIIL§(m)» Vu Q c°°(m). (11.4.5)

Let u E C°°(M), and set U r := PTu. Then

IILkUTIIL2(M) < (T + 1)kll"~'llL2(m)I

so that
k
IIUIIL2 2k
(m) < C'(T -|- l) lIUIIL2(m) 7

where C' is the product of the constants 00, 02 , 02k 2 defined by ( l 1.4.5). Using
(11.4.3) we deduce that if 2k > 1/ + -
m
2 we have

fluff?m> -< C'(T + 1)"+ lnuni2(m, S C'('r-|- 1)"+nnunui2(m). (11.4.6)

Fix a point Po E M and geodesic coordinates sz: at Po~ Observe that for any multi-index a
such that lal = U we have

3§u(p0) = (u, V o , p 0 ) L 2 ( m ) »
where VaaP0) : M -> R is given by

Va7pQlq I z agéaflp0, q ) lc'9§'\Pj(Po>l\PJ(q >.

<_T

The above discussion shows that for any u E COO (M) we have

'l'U,, Vop0)L2lm) | _< C'(7' + 1)%+%llullL2lMl.

Hence

llv,p0llL2<m> < 0 ( + 1 ) § +
Now observe that VaaP() E H so that

II L ' " v , p n n L 2 ( m ) S C(T + 1 ) k n n v , p 0 = I L 2 ( m ) .

Now choose k > V
2. Arguing as in the proof of (11.4.6) we can invoke (1 l.4.3) and deduce
that

lIV,p0llC,(M) _< C/I(T -|- llV/+% 7

where the constant C" is independent of the point Po. The above inequality is precisely
(1 l.4.2).
 Spectral Geometry 577

Using the inequality (11.4.2) for 1/ = 0 we deduce that

N(7') = 0(¢%). (1 l.4.7)

This can be substantially improved. Let us preview the strategy.

Fix p, q E M and consider the function
IR 9 T +-> E(T) = Ep,ql7" (sign 7'l<§'(7'2, p,q )
= (sign T) (11.4.8)
\I'1</(pl\Pk(t1l Q R.
Ak§T2

Note that

N(7_2) EILpIT' ldVg (p) I •

M

We will obtain sharp estimates for N (T) via sharp estimates for E p p I T I .
Theorem 11.4.1 shows that, for fixed p, Ep,pl7" is a temperate generalized function in
the T-variable. Its 'r-derivative is the generalized function
1 d 1
- - E (To = 5
2 d r p,q Z 'Pop)~vk(q ) ( 5 ( » t ) + 5
k
- ) E C'°°(R).

Our good fortune is that we can obtain quite a bit of information about the Fourier
transform of the generalized function

7'»-> r1 Ed p , -q(7)-
More precisely

Kt 1
_
52 rEp,q()l (t) 2
6-'iM As
+ am/H) \Ink(p)\Pk(ql
k

k
COS
( ) \Pk(p)\Pk(q)-
We will devote the next subsection to a more in-depth investigation of Kt. But, before we
do so, we want to give an explicit description of the spectral function in a special case.

Example 11.4.3 (Spectral function of a flat torus). Denote by Tm the lat torus
T m := Rm/l27rZ)m

equipped with the flat metric

TTY

g a
• (d@i)2.
j 1

Note that vol(Tm) = (27r)"".

The eigenvalues of the corresponding Laplacian Am are

l1?l 2 7 l2'=(k1,...,kTTI)€ZM.
578 Lectures on the Geometry of Manzfolds

For 5 (61,.. ,H'")€ R and EEGm we set

TTI

<12, 5> lfjW.

j 1

Denote by < the lexicographic order on am_ An orthonormal basis of L2 IW) is given by
the functions

1, 1; Z 6
l
\I//;(f7) (27f)% /Z > 0 ,
sin(l, §>>
2% c:os{l, 5), /Z < 0.
We have
l 2 2
6§32 I IZ,/TlM
_|_ -D -D

sin(k, 9) sin(k, ¢) -I- (2/T)m

-D

cosllg, 5> c o s ( l , <5>

(mm -»
l1<l<_R l1?l<_R
lZ>-5 l?>-(8
1 .|_ 2 1
cos(l, 5- 3) = (gfflm ei<12,5-<5)
(ZWW luff yn II<:ISR
-A

l1?l$R
l2:'>-6'
Hence
_ 1
-»
é"R2(9,0) - (27/)"m, lIl<R€ i</E,é') .

Consider the rescaled function

l
g'Rl@l I Run <§'R2
1 ei<%/215> I
1 ei<7,6'>
2 R m (2'/rR)m** 1
( o r ) I/€1sR 17€(§Z)M
l17IS1
Observe that as R -> OO we have
1
<5°R(9) =
(mm l€IS1
ii(§,@) Idol + 0 (;) 111111111

7 (11.4.9)

uniformly on the compact sets of the 5-space.

Let us consider a more general, but simpler situation. Suppose that w E 5'(lR?.). Define
' I/?l 2
\

I *P,;(9)\P,;((pl~
-v -v
<§wR(§', 'IJ DO
-0

keen
Arguing as above we deduce that

5" ,R(§> <5) I (l1.4.l0)

(27r)""

Again, we form the rescaled function

l
@w,R(@') = Run fg'w,R
 Spectral Geometry 579

For every Q and QE' E RM we define

_ -»2 m15
x) - w ( l l )e< )

so that €YWW Hence

l
é§'w,R(§l
(2'rrR)"" -9
fg
khZ"'

At this point we want to invoke the Poisson's summation formula.

Lemma 11.4.4 (Poisson summation). For any QUO 6 5'(RM) and any a > 0 we have
27Th a
-9

key'"
(b
l)l)m
a 27r
17€Z1"
¢?(av), ( I.I 4 .I I)

where (3 is the Fourier transform of (I).

Proof. To prove Poisson's formula is suffices to consider the case a = 1. The general case
follows from this case applied to the Schwartz function

(M ) ¢

Consider the l2wzlm-periodic function

<I> Run -> (c, q>(5> = ¢(i8'+ 2 IZ).

-v

kErr

This can be viewed as a smooth function on the torus Tm and, as such, it admits a Fourier
(spectral) decomposition
1 cf-
@(a8') i(17,i') Q- q> (5)e-("»5°) ld60l.
W ke V
[0,27r]'"

The above series converges in coe. The sum in the left-hand side of (11.4.11) is q1><0>. We
have
1
<I>(0)=~ no c .
(Q)
Now observe that

Ca =
_Z
khZ'"
/
[0,2'/r]M-k
_» ¢(§)e-> ld@l
Run
¢(§')€'""°'7>ld9l = <z>(/7)~

If in Poisson's formula we choose a = 2'/TR, then we deduce

l
Run <l> $<2 Re) .
176Z*" 176Z*"
580 Lectures on the Geometry of Manzfolds

Now let

(M) I e ) we)
=:'u,(i')

We deduce that

(M) I ms - 5).
Hence
l
@(2'/TR17 - 5)
é;,,,R(§') I kg-tlffl
17eZm
l l
A -5+ @(2trR17 - 5).
(21T)"" U( ) (2 T)t~ -»
y€ZM\0

Since Q 6 5' (RM) we deduce that for any p > 0 we have

1 -»
u(9) + O(R'P), as R -> oo, (ll.4.l2)
<§w,R(§') (27-)m Z

uniformly on the compacts of the 5-spaces. Note that (11.4.9) is formally a special case of
(11.4.l2) in which 'U is the characteristic function of the unit ball in the :E-space.

11.4.2 Short time asymptotiesfor the wave kernel

Let (M, g) be as in the previous section. For t E R we consider the bounded operator
cos l i t ) :L2(M) -> L2(M),
defined as follows. Pick u E I;2(M). Then u has a Fourier decomposition
u = Z u k W k , up = (u,\Pk),
k>0

where

llulli2<m> n 2

k>0

We set
cos lttlu COS ( I I .4.l3)
k>0

Note that

II cos(t\ A)uu=i2(m) 2
(to A/nuuu2 < llullL2(m>
2

k>0

so that the operator cos(t \ A) is indeed well defined and bounded L2 (M) -> L2(M).
Note that if u E COO(M), then for any N E Z20 we have
ANd, Z Akn'Ll,k\Ifk
k>o
 Spectral Geometry 581

and we deduce
cos(tv-A)An'u, = AN cos(tv-A)u A cos(tvk)uk\Pk( p ).
k>0

This shows that

lIAncoslt\ A)ulnL2(m) < HAN u m m ) , VNQZzo.

If we formally differentiate (1 l.4.l3) we respect to t, we deduce
3t23(cos(tvX)u)( p ) = (-WA\ cos(tvk)uk\I1k( p ).
1<>
_0

The above discussion shows that if u E coo ( M ) , then

cos(t\/-A)u E C'°°(]R x M).

For any u, 'u E Cfoo(M) we define u IZ U E C°°(M X M) by the equality

( u ® v ) ( p , q) Z u( P )v(q ), iv'p , q € M -
The Schwarz kernel Kt(P7 q) of cos(t1/-A) is the generalized function

K l p,q) = <:os(wX;)@k(p)w( q ) 6 0-oo(m x m).

k

This means that for any Fu, 'U E C°°(M) we have

(cos(tv-A)u, vldVgl) = (Kt, (u X 'v)ldVgxgl >

cos(tvk)(u, \Ifk)L2(v, \I/k».I)L2 .
k>0

The family of generalized function Kt(p, q) E C-OO(M X M), t E R is called the wave
kernel of the Riemann manifold (M, g).
For fixed p, q E M we obtain a temperate generalized function on R t »-> K t ( p , q 9

defined by

<Kt,'l]>
EU
k R
77ltlcQslt\/kldt I Wk(P Wk.q ), V17€5(]Rl.

Theorem 11.4.1 shows that the above series is absolutely convergent for any 17 E §"(]R).
This generalized function is the Fourier transform of the temperate generalized function
1 d 1
5 Et (To 2 'Popwk( q l ( 5 / ( T l +5_<¢>) E 5/R>. (ll.4.l4)
k

Thus, an approximation for Kt will yield an approximation for the Fourier transform of the
derivative of the modified spectral function
l
T +-> -(sign
2 7'l<§'(7'2 7 P 7 q ).

We will spend the rest of this section constructing an approximation for the wave kernel
Kt.
582 Lectures on the Geometry of Manzfolds

If f 6 C°°(M), then the function u(t, p) cos(tv-A)f 6 c°<>(R X m ) satisfies the

initial value problem

(63 + A)u I 0, on (0, OO) X m, (ll.4.l5a)

u(0»p) f ( p iv UtlOv p ) = 0 , WQGM. (ll.4.l5b)

We deduce that the kernel Kt defines a smooth map

(0,ool X M 9 ( t , P )+-> Kt (p, q )€ C'°°(M).
The a priori estimates described in our next result will play a key role in our search for an
approximation for Kt.

Proposition 11.4.5. Let 'U 6 COO([0, 1] X M ) be a solution of the non-homogeneous wave

equation
( 3 - i - A ) u = h on [ 0 , 1 ] x M
(ll.4.l6)
'v(0»p) I Utl0)P) I 0, Vp G M.
Assume that

8g'h(0,pl = 0, V p 6 M, j < k. (ll.4.l7)

Then there exists Co > 0 independent of h such rat for any t 6 (0, 1 ) we nave
k-I-1
k 1- '

||@ + 311(t, ' l H L 2 ( M )

j 0
(ll.4.l8)

(f
t k-1 .

S Co 0 lla'°h(s, ' ) ' I L 2 ( m ) d 8 + Z!l@£"--f 'l'IL2(m)

j=0

Proof. For any t 6 ([0, 1] we have

1 d
2 if (ll@uui2(m) + ( A s , UlL2(M) I7/¢¢,U¢lL2(M) + lA'U,'utlL2lA/fl
(in, u¢)L2(m) .
Observing that
2
lAU»UlL2(m) - lldv1IL2(m)9
we deduce by integrating that
t
2 (h,as@)L2(mlds
0

t 1
< 2K ll1(S,-)llL2(M)dS, K lIUtllL2(Ml ds.
0 0
 Spectral Geometry 583

Since
1

K2 <
0
l1<9tv(s»-)ll2L2(m) ld5l>
we deduce
1
KG gawK llh(s»-)llL2ds,
0

SO that
1
lll3tU(t» -)lILy + II(dU(¢, -)lILy <- C 0 llh(s, -)lIL2ds.

In particular,
t t

l1v(t,-)llL2 <
0
ll<9sv(s»-)llL2d5 < Ct
/ 0

t
uuh(s,-)llL2ds
(ll.4.l9)
<C
/0
llh(5» -lIIL2dS.

This proves (11.4. 18) for k = 0. We can now argue inductively.

Assume that we have proved (11.4. 18) for k and we prove it for k + 1. We consider the
function u := as). Note that

u(0,p) = 0, WE 6 m,
and

'Ut (0, p) c9t2v(0, p ) = h(0, al -A'v(0, p = h ( 0 , p l = 0 , > VPGM-

Thus u satisfies an initial value problem of the type (11.4.l6), where h is replaced with
<3th. By induction we deduce from (11.4.18), where h is replaced with (ith, that
k+1 . k-I-1 .

41<<>'i"+2-"@<t» -MI = Ell(@i"+-"u( -MI

j=0 j=0
( I I .4.20)

(f
t kt-1

ll(@k+1h(5,-lllL2d5+ 'l(3tk-h(5='l'IL2
0 J 0

Now observe that

llv(¢, -)HL < 0(llA»(t, -)HL; + llv(t, -)HL2 )

( .4.
11 =19
)
C
(f 0
Ne -)11L2ds + ll33»(t, -)HL; + llh(t, -)HL; I •

Using this in (11.4.20) we obtain the sought for inequality.

584 Lectures on the Geometry of Manzfolds

Using the functions Up 6 C°°(M X M), k 1, 2, . . . constructed in Proposi-

son 11.3.22 we deduce that there exists a constant c o (0, 1) such that for any N > 0,
any y 6 M and any t < C we have
N
(63 + A ) Uk(,,,y)@i1<=(¢,5y<>) = l9yl-%5,, + (AasUn(0U7Y) )92iN(t¢S I ) .
k 1

=:<§'N(t,:1:,y)

(1 l.4.2l)
The generalized function iN is an "approximate" fundamental solution of the wave opera-
tor on R X M. We want to prove that 3t<§'N (t, -, -) is a good approximation for the kernel
Kt of cos(t\, A).
Using Theorem 11.3.14 as inspiration we associate to each f E cOO ( M ) and each large
positive integer N the function

I/vf ltv <5"nlt'J

m) Z 7:67
:lflY lldvgly )|.
We need to explain the above construction. The generalized function <5"n defines a smooth
map
(0,oo) X M 9 (t,:z:) »-> <§'N(t,a2,-) 6 C' ° ( M ) ,
and thus

Wfn(t,a:) = (<§'N(i 7 3 7 -),fldvgu ).

Using (11.3.18) we deduce that

Wf(0+,a:( = 0, <9tWtl0+,a:l M U1l.ClZ', Yl6t@?l_l0+, Sys-Tl )f(y)ldVg(y)l

(11.3.45)
(lgl-, we, -)f(-)l<wg1 ) = u1(w)f(w) = no).
Observe that

(of -|- A l w f ( t , 23, 311 Sn(W°,y)f(y)ldV9(y)l 7

M
where
S o ( t , x , y ) = (A,,,UN(;u,yl l=Lnlt75yl$l
For N large Sn(t, Xv y) 0 for t = 0 and we deduce that

<9t2Wf(0+»x) Sn(0,11,y)f(y)ldVg(@/DI -AW(0+»;v) = 0.

M

Thus, the function <9¢ Wf satisfies the same initial conditions (1l.4.l5b) as the function
cos(t\, A) f. Consider their difference, i.e., the function
v/u, w) = v,N(t, g) ; ¢@s(/'A)f - awe' (s, =»).
 Spectral Geometry 585

The function Vf is the unique solution of the initial value problem

kg; + A)Vf = -(33 -|- A m w f = he,
Vf (0, 513) I 8tVfl0,$U) = 0,
where

of - ofN (1323) M
<9 Sn( t 73:7 y)f(y)ldVg(y)l

3tsn (t 717 y)f(y)|dV9(y)|-

dist(y,x)<t

The distribution <9t%Ln is homogeneous of degree 2N - n - 1 = 2N - m - 2. If we

choose N sufficiently large, e.g., 2N > n -|- 2(k + 1), then Q 6 c k + l and thus for any
101 < k, and any ac, y such that sy(x) < t < c we have
(it y3t%§N(t,x,y)l _ ct2N'"'1'l'*l.
<
Hence for any t 6 (0, c) we have
t k-l
m
k A:-1-' N- - -k
.7hf(i, -)HL; S Cut2 ||f||L2- ( ll.4.22)
2
ll35hf(s, -)lIL2ds -|- |16t 2
0 j=0

Proposition 11.4.5 now implies that for t 6 (0, c)

k-I-1

lIVf(t, -)lILyA=+1 'l'll6tVflt»'lllL§ S Ella

j 0
k+ - '
1 _ ct"-
"V(t»-)llL2j <
m
2
-2-k
||f||L2-

The estimate (11.4. 19) implies

lIVf(t,-)llL2 < Ct
2n-2-2
2 ||f||L2-
m
Using Lemma 11.4.2 we deduce that if k > u -|- 2 so thatN > u-|-m-|- §29 then for any
A > 1 we have

llw(¢,-)Hé < 0lllVf(t»-)ll2L;+1 + A llw(t,-)lli2)

/v<2-I-1 1/ ' k

< (j l t 4 N - m - 4 - 2 k _|_ I \ k t 4 N - m - 4 l

If we let A = t 2
we deduce

t2v+"-2Huuvf(t, -)He < C1t4n--uufuui2,

so that

lIV/N(t»-lllc0S t 2 N - n - 1 - u l l ] c - H L 2 Vt E [0,0], VN >> 0.

We deduce similarly that
llatv/N(t, -)lIe,, g #N-"-2-"llfllL2, it E [0, @], VN >> 0.
Using the equality
586 Lectures on the Geometry of Manzfolds

we deduce gradually from the above estimates that

ll<9¢,VfN(t»-lllc S Ct2N-n-1""lllfIIL2, Vt E ml, WV >> 0. (11.4.23)

Similarly, the function VN f satisfies the initial value problem

(of + AlVAaf = -(63 + A)3¢UAaf = hAlf,

l/Ayf lo, 16) = gel/AyflO, SE) = 0,
where

/ y)A1flylldvg(y)l m A§,é'¢SN(t 7 3 : 7 y)f(y)ldVg(y)l

..II

hAlf <9tSn(t 7 1 7
I

:
M

/
dist(y,x)<t
A1y<9tsn(t 7587 y)f(y)|dVg(y)|-

Arguing as before we deduce that for any k, j 6 Z>_0 and N sufficiently large,

llv§f(t, -)llc»(m> <_ tM-"-l--"ulfllL2, it 6 lm

ll3tk v f lt) _) II
Now observe that we can write
C " _
< t 2N_ _1'k _ 2 .7. _1»llf HL27 it
6 [0»"
].

Vfn(t,a2) = ( K ¢ ( f r , y ) - 3t<§'N(t,a:,y), f(y)ldVg(y)l )»

=:'J'N(t,:1:,y)

where we regard the tail 'J'N(t, at, y) as a smooth map

(0, c) X M 9 (t,:z:) »-> 'Tn(t7327 fU) G C-OO(M).

We deduce that for any multi-index a and any j 2 0 there exist N = N ( a , j) and C' > 0
such that for any N > N(0,j), f 6 Croo(]l/I) and any t 6 (0, c) we have

sup
:uGh/I
|( a,<rN<¢gm) y ) , A;,f(y)ldvg(y)l >l < C¢t2N-'rz-1-2j-lal IIf||L2- ( I I .4.24)

From the above inequality we deduce that for any (16,513) the generalized function
Ay'J'N(t,a2,y) is represented by a function in I;2(M1 for N sufficiently large. From
elliptic regularity we deduce that for N sufficiently large and any (t,a2) the function
y »-> 'IN(t, as, y) belongs to the Sobolev space L2j,2 and

ll@,A;<rn(¢,x,y)ulL2 < 0 t 2 n " + . 2 ' l " l .

Using elliptic estimates for Ay and Lemma 11.4.2 we deduce that for any ft, U 6 Z20 there
exist N ( k , u) and C > 0 such that, for any N > N ( u , k ) , anya: 6 M, t 6 [0,e], and any
multi-index 101 < k we have

ll@,<r(¢,w, -)II(>v g (I N-"-1-IQI-".

This implies that for any ,LL 6 ZZ() there exists N = N (p) and C > 0 such that for any
N > N(.u,) we have 'J'N(t,;z:,y) 6 C**( (0,c) X M X M) and for any < 1J, we have
.
||é*€Tn(#, - 7 -)ll@~-»(mxm) § co N-n- -
1 **. (11.4.25)
 Spectral Geometry 587

We can rewrite this as

(l@§{ K ( - , -) - e m f , - 9 -) }I CO-jlMxMl
7 t 6 [0,@]. (11,4.26)

Since Kt is even in t we deduce that for ltl < e, N > N(/1) we have
(l<%{ K ( - , - ) - @(@n(t, - 7 - ) - kw(-t, - 7 -) ) }H CM-j(MxMl
(11.4.27)
We set
<§(t 7587 y) == <5n(t,w,y) - div(-t 7 1 7 y),

5'N(t 73:7 y) : z K ( , y l - awN(¢,x, yl.

We can rewrite (11.4.27) as
2N l p
!l<9I €TN(t, 7
-) IIc/*-J(MxM) S Cltl
'n
» l f l S Q j < u , N 2 N(p,l.
(1 l.4.28)
Set

n 1 1 := dist(a:, y)2 7

URL as, y) == a (x+-% (to - UCC, yl) - x+'% ( - f7(:/% fa) )1

(1 l.4.29)
= a ( Q § ' < ( i , s y ( x > ) -@i(-tsy(w)))

We have thus obtained the following result.

Theorem 11.4.6. We have an asymptotic expansion

Ktl37)?!l iv
i
k 1
Uk(£U»y)5£2I<(i»CV»yl» lil < C) (1 l.4.30)

where the functions Up 6 cOO ( M X Ml, k = 1, 2, . . . are constructed in Proposi-

tion 11.3.22.
More precisely, (I ].4.30) means that, for any positive integer ,n, there exists a positive
integer N(itl so that, for any N > N(iil, the tail
N

7n(i,I,Y) := Ktlfm/I UI¢(5U»3/l9(21<(t» Xv y)

k=1

belongs to cut ( - c , cl X M X M ) and satisfies the estimates ( I ].4.28).

For N > N(1) and 3/6 M fixed we obtain a C1-function

(-c, C) -> <3'N(¢,y,y> 6 R.
Using Proposition 11.3.17 we deduce that
N
Unlit, Ylat ( t, by(x)l - Qkl -t, sy(x)l )
k 1
588 Lectures on the Geometry of Manzfolds

2 Uk(y,yldm(2I€llXI2k-1-"(t).
k=1

2 F
I`(2k- 1 5) n)Uk(y3y)|x|2k-1'"(t)-
m
7r 2
k=1

We have thus proved the following result.

Corollary 11.4.7. For any y 6 M the restriction of Kt(;I3, y ) to the line

R X {(y,y)} C R X M X M

i5 a well defined generalized function Kt = Kg(:¢/7 y) 6 C'°°(]R) s a s s i n g the asymptotic

expansion
OO

K 2 U1<=(y,y)dm(2I<)lxl2k 1'"(t), lt1 < c ( I I .4.3l)

k=1

meaning that for any ,LL > 0 there exists N (M) > 0 such that for N > N I we have

5'n(i,y,y) = K¢(y,y) - 2 UI<(y,y)dm(2k)lxl2k'°-1'"(i) G 0"((-Q 0), x M ) .

k=1

and
_ _ n_ _ _. . .
(@z@yvIv(t,v,y)l-01ltl2 'n 1 ', Itléc, JAM; ICYISM-J
for some constant 01 > 0 independent of y.

11.4.3 Speetralfunetion asymptotes

It is now time to reap the benefits of all the previous theoretical work. Recall that

NR) dim her(/\ - Agl.

o n

Ae[0,R]

Theorem 11.4.8. There exists a constant K > 0 such that, for any y E M we have

\Pa(y)\Pk(y) ' Wm7'm < KTM-1, as T -> oo. ( I I .4.32)

Ak<T2

In particular

Ng(R ) I V019lM lwv11R m/2 -|- o(R""/2-1 ) as R ->oo. (11.4.33)

 Spectral Geometry 589

Proof. Consider the increasing functions

MTI I Ey,yl7-I7 1/(7) I WmlTITTI sign(Tl,

where Ex,ylT) is the modified spectral function defined in (1 l.4.8). We will prove that
l/J»() - 1/(7')l = Olin-1 ) as7'->oo. ( l l.4.34)
We will obtain information about by recalling that, for every x, y, the Fourier
transform in the temporal variable of -2 Qdt is the wave kernel Kt (ac, y).
On the other hand, according to (11.3.30) the Fourier transform -2l d- v
d r is the first term in
the asymptotic expansion (11.4.31) of Kt for t near zero.
To put this fortunate coincidence to good use requires some technical skill and imagi-
nation. What follows is inspired from [67, Sec. 17.5].
Choose a positive function q25(T) 6 5" (R) such that

¢?(t)>0, we <_ 2 , C
(MIG, we Z e ,
(11.435)
Tb(T)dT Z 5(0) Z 1.
R
Such a function is easy to construct. We define d) as a convolution

¢=l¢l 2 * w,
where

.. The Fourier transform t) is supported in the interval lil < _e

2°

IR lw(¢)l2d¢ I 1.

Observe that Ztrz(-T) is the Fourier transform of the function 6 C13°(R). We

deduce that it is real analytic5 thus its zero set is discrete. In particular

lwl2 * awe) > 0, VT.

Then

¢?(t) [l¢u2()] 2 > 0

and
2

¢(0) l¢l2(0)2
(f R
l1D('f)l2dT
) 1.

Fix a cut-off function co satisfying ( I I .4.35). For any a >> 0 we set

1
(M) a
<15 I)
a
SO that M) ¢?(at) has support in £ 9 2], We deduce that for any a > 0 we have

§;(t)Kt i v
2 i
k=1
Uk(y, y ) d m ( 2 / < ) ( t ) l x l 2 ' + " ( t ) . (11.4.36)

5 The Fourier-Laplace transform of a smooth compactly supported function is an entire function.

590 Lectures on the Geometry of Manzfolds

More precisely for N >> 0 the remainder

N

@(t)<i'n(t,y,y) (t)K5- QM) my,y)dm(21<)lxl2k-1-"(t)

k=l

is a continuous function supported on the interval [- C 2] and satisfying the estimate

a'a

(@(¢)?1'n(t,y,y)( <
_ cum 2N-'ra-1
7 we § c,
where the constant C' is independent of a. Here we can be even more specific. It suffices to
choose N such that

2N 1 n _ 0< >»N >_ 1 + m

> llllllllll

2
_z
\ >»N 2 N(m) l,
Hence

;(t)7Tn (t,y,y) l S Crltl2N('IrTL)-n-1 g Cltl"', ltl <- min(l, c).

As we know, the inverse Fourier transform of Kt is the generalized function
l d
--E ( ) =
2 dT p,q Z \I/k(p)\Pk(q)((5v
k
r) + ( L w ) 6 c-°°(1i;>.

For this reason we plan to take the inverse Fourier transform of the equality

§5;(i)Kt 2 Uk(y,1/)dm(2/~f)5;(t)lxl2'"'1'"(t)
2<
_2k<'n-I-1

+ ¢(t)<l1'n(m) (t, Fu, y).

Using the identity (11.2.1 lb) we deduce that the inverse Fourier transform of 2k-1-IxI
"(t)
is the function
22k-1-n n_2k
22k-1-n 'rn-l-1-2k
IxI ().

Hence
2216-m-1dml2kl
1 d
* Ey,ylT)) Uk LY? 311 a
* IXlm-I-1-2kl7_l
2 dT (ba
2<2k<n (11.437)
-I-
g 1
[(3;(t)'Tn(m)(i,y,y)] 7

where 5?-1 denotes the inverse Fourier transform

1
82-1[f(t)] I 27r emf t f l t l d t .
R

Using (l1.3.30) and the equality U1(y,y) = 1 we see that the first term of the sum in
(1 l.4.37) above is

--cum up" s1• n .

2i ddT T g T
 Spectral Geometry 591

We can rewrite (11.4.37) as

l d
(coa * (Ey,yl7'l - wmlTI1" signs) )
2 dT
22k-'rn-1drnl2kl m _ k
Uk(y, y) /; (Ba IxI +1 2 (T)
2<2k<n (11.438>

+5-1 @(t)<'1'n<m>(t,y,y)
==/>a(t)

Let us estimate the terms on the right-hand side of (11.4.38).

The remainder pa(tl is supported on the interval [- £9 £] and satisfies the estimate

Ida(tll < cltlm 7

for some constant C' independent of a. Note that

4 w»a]4 Qtr
1
c
a

C
a
emf tpaltldt
Z
1
2'rrlTI 11
TC
a

TC
a
spa(s/T)ds |
T C

< C a C
Smd8 <
- am-I-l •
-l¢lm+l 0

Next, we need to estimate terms of the form ba* IXIPl7ll,p G Z>0. In other words we need
to estimate functions of the form

fpl7rl if (/»("
R a
S
)l5IPd5.
Note that

f ) : QP <1>( - s)lslpds = as sl1°gb(s)ds.

R 'R
Using the inequality

a - so' < (Vu + lsl)p < 22°-1( it' + ask) 7

we deduce

l]pl07'l S2p-1 (aw

/
R q25(s)ds-I-ap Rlsl"¢(s)ds

=1
/
so that

lfp(T)l < Co(ITI-I-alp, VT. ( 11.4.39)

Using (11.4.38) and (11.4.39) we deduce that there exist positive constants K0,K1, inde-
pendent of y, such that

< K0 TVN-17 VT, (11.4.40a)

592 Lectures on the Geometry of Manzfolds

d
(25a * (m - w) KG( IT1 + a)m-3, VT. (11.4.40b)

Arguing as in the proof of (11.4.39) we deduce

Quoa * 1 / ( T ) K2llTI -l-CL)M-1

Set

50 min cl5(T) > 0.

lflé
The function /,L is nondecreasing and defines a Lebesgue-Stiltjes measure du on R We
observe that
d l
((lea * m)(T' du * ( M ) Z R (MT ' s)du(sl

Hence
7r-I-%
du * kw) * q5a(Tl+du * (MTI
-1
&`0CL d/,L<_ Z (do-du)
T-%
< K1(ITI +a)""-3 +K2(ITI +alM-1.
We deduce that for any b G [0,a] we have
T

M) - MT - 5) S MT) - MT - al < du
a
m 3 m 1
< K1 I7I+ L

-I-K2 I T|-|-
L

For S > 0, we divide (0, s) h1Mu<s-|- l intervals of length < l and we deduce from the
above inequality that

INU-mU-&9 < ( + U
(( K1 Id
3a
2 )
m-3

+K2 l. ) )7'|-|-
3a
2
m-1

Similarly

I(T+05)-/lT) < (s-I-1)

We deduce that for any 5 6 R we have

( (
$1 IT-l-asl+

IM) MT 05)l

< (lsl + 1) IKE ( we + a(lsu + 1) lm-3 + KG (Vu + l<»u(lsl + 1) )""-) •

Observe that

lu * ( M ) of
1 R
u(s)<l5
(TaS) ds
/ R
1L/,(T - aslg25(s)ds.
 Spectral Geometry 593

Hence

IM) M * (l5a(T)l
R
(MTI #IT @s))(l5(s)ds n
<
R
(II + 1) IKE < ITI -|- a(|5| -|- 1) l m - 3 + KG (ITI -|- |a|(|5| -|- 1) ) gz5(s)ds
< K5l l7rl'rn-3 _|_ l,l_l'rn-1 _1_ a m - 3 _|_ a m - l l I Of l,l_l'rn-1

Arguing similarly we deduce that

up) 1/ * ¢Q(¢)l < K6(||m'1+<l"'"1)=o(|¢|""1).

Hence

IM) - 1/(T)l
< lu(Tl - H * <ba(T)l -|. I1/(7) - I/ * (Mill + llH ~) * (bow
7'

< KG( l,rlm-3 + ITIM-1 + a m - 3 + a m - 1 ) + l¢)a * do - 1/)(s)l

0

< K7l l7rlm-3

7'
_1_ l7rlm-1 _l_am-3 + a m - l l +l2 _1_ a ) m - 3 d 5
0
< Kgl ITIM-3 -|- |T|M-2 -|- |,7-|m-1 + 0m-3 -|- am-2 + 0m-1 )
= Of l7rlm-1

This proves (1 l.4.34)

11.4.4 Spectral estimates of smoothing operators

For every even function w : R -> R such that w(;I:) decays s u f i ciently fast as -> OO we
set

w6(:cl w(€;u), Va; 6 R

and we define g e = g w @ @ Q o o l m X ml,

g5( p, Q) We ( d ) @ ( p ) @ k ( q l~
k>0

The function £5 is the Schwartz kernel of the bounded operator

w@(v Al ¢L2(m) -> L2(M)7 w€lw6(\, As) Wk = w@(1/wk,

If w is the indicator function of [- 1, 1],

1, ac 6 [-l,1],
w(;z:)
0, otherwise,
1 . . .
then 5;/T colncldes wlth the spectral function we have investigated earlier. Note also that
• • • •

2 • •
when w(;u) = e'"' , then w / ; ( \ A ) = e'tA. Thls is the so called heat operator.
594 Lectures on the Geometry of Manzfolds

In this section, building on the results developed SO far we present in some detail infor-
mation about the behavior along the diagonal of this kernel as 5 \ 0 when w is a Schwartz
function.

Theorem 11.4.9. Suppose that w 6 5" ( R ) is an even, nonnegative Scnwartzfunction, and

( M , g ) is a smooth, compact, connected m-dimensional Riemann manifold. We dejine

6 3 6 2 5 5 : M xM->]R, g ( p,qI w(@JE)w( pw k ( q ),

k>0

where (\P1T:)k21 is an orthonormal basis of L 2 ( M , g ) consisting of eigenfunctions of Ag.

Fix a point P0 6 M and normal coordinates at P0 dejined in an open neighborhood
60 0fP0~ The restrictionof et to 60 X 60 can be viewed as a function 35 (33, y ) dejined
in an open neighborhood of (0, 0) in RM X ]R'". Fix multi-indices 0,B 6 (Z20)"". Then,
as 5 \, 0, we have
_m_2d(0 B)il<.l-IBI
a g a , + O(62)
Y)l11:='y=0 = e
(21T)"" (f Run
w(lxl):I:"+§d:u

( 11.4.41)
where

do) .
.
l@+/3l
q
| C

Moreover; the constant implied by the symbol 0 ( 6 ) in (]].4.4]) uniformly bounded with
respect to Po.

Proof. Observe that for any smooth f : M -> R we have

1 l
'UJ;(tl€'itvZfdt
Well/-Alf = 2tr
11111111111

R 2tre R
w
()
6
e/Zfdt. ( 11.4.42)

The Fourier transform 18(tl is a Schwartz function so 'lD(t/6) is really small for t outside a
small interval around 0 and e sufficiently small. Thus, a good understanding of the kernel
of am/-A for t sufficiently small could potentially lead to a good understanding of the
Schwartz kernel of w5 (v'A).
Fix a point Po E M and normal coordinates at Po defined in a neighborhood et of Po.
Then we can identify a point (p, q) E 60 X 60 with a point (as, al in a neighborhood of
(0,0l in RM X RI".
Using (11.4.42) we deduce

1
3§35£€(x, ?!ll12=y
5
@;*@5K(a:,y)l=y, w
: K t :IB
V)
5
( I I .4.43)

Choose an even, nonnegative cutoff function p @0gO(Rl such that

1, lil
0(t)
0, lil
 Spectral Geometry 595

where C > 0 is the constant in Theorem 11.4.6. Then

a
@3655 6 _ 1
l.5U7 Yll@=Y - 5
OB *
Kt ,»o(t)w 5
t
( l
5
II)(K, a B
,(1-p(t))w
ltd)
5
Let us observe that, for any N > 0, we have
1_ Kt ,B 7

Thus
5 ( 5
O((=:N) as6->0.

iv > 0 @s@5(w,y)las y iv
1
5 (Kt 0,,3
,0(t)w
(U) +O(&N),
5
5 ->0. (1 l.4.44)
From (11.4.30) we deduce

3§'<9y5Kt(;v, y) dm(2I<)@;*@5 {up(=v, yurt, of, y)}- (11.4.45>

k=l
On the other hand, Theorem 11.4.6 implies that
- - 2k
3;*65K(w,y)l=y=(» A»m,a,@,klxl m la+Bl+ (t), (1 l.4.46)
k 0
where Am,a,,B,0 is a universal constant depending only on MY a709 which is equal to 0 if
10 -|- BI is odd.
Lemma 11.4.10. (a) Foranyr G Z a n d a n y N > 0 we have
l T` T' T`
ase->0.
-(IxI ,pa)=@ ((=al , w ) + 0 ( N ) )
/\ A

6
( b ) For every positive integer T' we have
v',/T21-r
(l)cl"l»17) l7'IT-1'wl7'ld7'_
fig) R

Proof. (a) For transparency we will use the integral notation for the pairing between a
generalized function and a test function. We have

(lxl"»171//Q 51
L lxl""(t)p(tl1F(t/@)dt
/ R
lxl"°(@tlp((°2tl17(t)dt

: g
lxl°°lt)p(€t)u7(t)dt i 67(IXI', Pg.]>g Mt) i p(etl.
R
Now observe that pE'z? _w Mk
= '®(p6 - 1) -> 0 in 5(]R). More precisely for k > 0 we have

-(p - 1) = O((=:NtN) as 5 -> 0.

Mk
This implies that
(uxl'x@(p 1) ) = O ( a N ) Asa-> 0,
so that
(lxl'Hpv7) 1 (lxl'°»17)+(lxl'U'u7(»06 1)) 1 (IxI IN) + O(aN) as 5 -> 0.
(b) We have
-7° _1° (11.2.11b) _or 7°_

<lxl , w ) = ( l x l »w) = \/T21 (IxI 1(To » w(T))

v'7r21-7'
ITI"'1w(T)dT.
we) R
596 Lectures on the Geometry of Manzfolds

Using (11.4.44) and the above lemma we deduce that as 5 \ 0 we have

@ s @ 5 " ( x , ?!llw=y = D o , , @ - " " ' ( " » " ) +O I 6-7TL-2d(O£,,3l'|'2
( 11.4.47)

where Dfmrz,a,,5 is a universal constant that depends only on m, & 7 B which is 0 if l0+5l
is Odd,
v721-r
Dv7z,a,B I Am,a,B,o ITI-1w(T)dT, 7° m -|- 2d(a,B). (11.4.48>
F(§) R
To determine the constant Dm,a,,5 it suffices to compute it for one particular m-
dimensional Riemann manifold. Assume that (M, g) is the torus Tm equipped with the
flat metric
m

g (do 7 0
< H'll < Z7r.
i 1

The eigenvalues of the corresponding Laplacian Am are

l1?l 2 IZ: (/l,...,1<'tn) e ZM
7

Denote by < the lexicographic order on Zm For 5 (HH. ,0m*)eR and /Z A Z m we

set
'rn

(E, 5> jo .
j=1

A real orthonormal basis of L2(']I"") is given by the functions

1, IZ: 6
1
2% sin(k*, 5>, E >~ 6,
we) I (27f)%
2% cos(l, 5>, /Z < 6.
Then
1 )
"'»5-¢
(5) Z (Mm 1<€Z'" wl@l'l)@
¢§'5< 7
5»
-»

SO that
H _» lal-l@l
aga5@<@,0> = (WH W5ll/*ll]I¢0'
+5€ill§,§')
-»
khZ'"

Define
W77l7 UP : R'TTI -> R, Wm(x) = w(l;ull, u6(x) = Wm((=:;z:);c0'+5
Using Poisson's summation formula (11.4.11) we deduce
il0.l-IBI
agagmo, Of UE (27r171.
<2>~
Observe that

Us (5)
Run
e-(€»x>w(@ll)"+@d = (i<9€)°"0
(f Run
e
 Spectral Geometry 597

-¢(l§v

(/ We
€-77llz°6€)0€-}-B
e ) €-Ml'I;@§la+B/M"
Run

Hence
il0fl-l01
aga5.<§)€<5, 0) = (2tr»5)""
UQZM2{ li3€la+B/\WvT1
(l) } 6
§=2'rr17

As 5 -> 0 we have

3§3:£6(0, 0)
l2 t r lm
(ia>°*+@Wm(0) + O(eN) ) 7 VN.

Now observe that

( i 3 € ) O + W m (0) w(l;ul);z:"+5dac°.
Run

so that
'l0l-10l
agajgo, 0) = 5-m-la+@l z
(2"l""' Cl Run
w(l:z:l)x°'--5d:v + O(eN) 7 in.
(1 l.4.49)
This shows that
il0,l-I/3l
Dm,a,B (mm w(lxl)x"+5dx. (1 l.4.50)
Run

This completes the proof of Theorem 11.4.9.

Remark 11.4.11. Note that

Run
w(lacl)x"'+5dac
(/ lxl=1
x"+5dA(x)
) (lO 0
W(T)T77I+l0+B
I-ldr I 0

=ZIm a 5lWl

On the other hand, according to Lemma 9.3. 10 we have

. 5. 1
2 H? F ( ° ' * 2 * ) Of -|- B6 l2Z>0lM, ( )
FlM+l2+Bll 7
O+0dA(x> 11.4.51
..II

L Z'rn,a,,8
l2ul=1
0, otherwise.

We can now rewrite (11.4.50) as

-I@12M 1o7,6 ).
@|»;|<.|
5-"*-|o+ I-»<1,B(w (11.4.52)
Dv7z,a,B
l27rlm
598 Lectures on the Geometry of Manzfolds

11.4.5 Spectral perestroika

It is time to reap some geometric benefits from our analytic toil. Let (M, g) be a com-
pact, connected, Riemann m-dimensional manifold. Denote by Ag the associated Laplace
operator.

Definition 11.4.12. The spectral data o f ( M , g) consist of the following.

(i) The spectrum of Ag

spec(Ag) C [0,oo).

(ii) The orthogonal projections

P A L 2 ( M , g ) -> ker(A - Ag), A E spec(Ag).

To use a terminology dear to the late great Vadimir Arnold (1937-2010), the spectral
perestroika question asks how much of the geometry of the Riemann manifold (M, g) can
be recovered from its spectral data.
Note that the multiplicity 're(/\) of A E spec(Ag) is computable from the spectral data
as the dimension of the range of PA. The operator PA is an integral operator whose kernel
KA(P, q ) is given by
'n(A)

KA(P, q) Z <1>(p)@>(q
j 1
7

where l'I'jl1$jSn(A) is any orthonormal basis of her(/\ Ag) consisting of real valued
eigenfunctions. This means that

( P All(Pl K A ( P , q Lu(q )ldVg(q )|-

M

As in the previous subsections we describe the spectrum of Ag as a nondecreasing sequence

0 A0 < AS < AS < 7

where each eigenvalue appears exactly as many times as its multiplicity. Fix an orthonormal
basis (Wk)k20 of 1;2lM) such that
A\P k As/k, We 2 0.
Then

KA(P, q \I/k(p)\Pk(q)-
As A

Fix an even smooth function w : R -> [0,

w(t)
 Spectral Geometry 599

>

Fig. l 1.3 A smooth approximation o f the indicator function of [ - l , l].

We can think of w(t) as a smooth approximation of the indicator function of the interval
[-1, 1], see Figure 11.3.
As in the previous subsection we set
few p, q) w(@JE)@k(p)\pk(q) = we) K A(P41 ).
k>0 A6spec(A)

The above sum consists of finitely may terms since it has only contributions from eigenval-
ues < 6 - 2 so it has O(e'"') terms.
Fix a point p E M and normal coordinates ( i l near p. From (1 l.4.41) with lal =
um = l we deduce
l
+ O(82)
(f
5 m 2
8g;i 8y.7 (gas (at, Yllrv=y=0 (mm Run w(lxl)xWdx

60
6 m 2

(2w)"" Run
w(lwl>(w1)2da: -I-O(82)

;Brow)
I
The positive constant Be(wl can be determined using (1 l.4.52). More generally, for two
vector fields Xi, X2 E Vect (M) we have

X1(3U)X2(3/)é"€(£€»3/)lw=y=0 = X13mX (9)3g3 g5t (Lu, 311 l:v=y=0

"KJ
5 'rn 2 1
l m 6 ix;(0>xg<0> (Bow) + 0<8> ) ( I I .4.53)
id
6 'rn 2 1
(2,,mgp(x1( p),xi( p ) ) ( 8 m ( w ) + 0 ( @ 2 ) ) .
FoIIX1,X2 E Vect(M) fixed, the above estimate is uniform in p E M . We set
0(6) dim ke1°(A - All.
A<e 2

According to (1l.4.33) we have

D (5) = volg(]ll)oJme"'" 1 -|- o l 6 ) I asa \ 0.
600 Lectures on the Geometry of Manzfolds

Observe that the sums in (11 .4.53) consist of D(e) terms.

Proposition 11.4.13. For 5 > 0 sujiciently small the map F E : M -> RW),
€""-*2(27r)M
p I
F
€

Be (w)
wls\/kl Wk(p ) E RDl5l 7 (1 l.4.54)
k<€-2

i5 an embedding and the metric induced by F 5 converges unzformly to g as 6 > 0.

•
Proof. Denote by the canonical inner product in RD(€). For p E M a n d X E TpMwe
denote by DX F e(p ) the value of the differential of F 5 at p on the tangent vector X (p) .
Explicitly,
d
Dx F e(P )
-
dt t 0 I g
F 5 (v(t)) E ]RD(@) 7
where by : (-5, 5) -> M is any smooth path such that 7(0) = P and w) = X(p)-
The estimates (11.4.53) show that, for any X E Vect (M) and any p 6 M, we have

Do F a( p) . Do F 6( p) = g p ( x ( p) , x ( p) ) ( 1 + 0 ( @ 2 ) ) a s e \ , 0 ,
uniformly in p 6 M. This proves that for 5 > 0 sufficiently small, the map FS is an
immersion and the metric gE on M induced by FS is C0-close to g.
To prove that FE is injective for 6 > 0 sufficiently small we argue by contradiction.
Assume that this is not true. This means that for any n E N there exist points pn, qn E M
such that

\I/I<(pal = Wk( q n l v 'v'k I iv ,n.

After eventually extracting subsequence we can assume that the sequences ( pn l n € N and
(q 'rgl7lEN are convergent
p: lim pn, q : lim qn-
H 00 n->oo

Note that

'If/P) = Wk(q I, We 6 N. (11.455)

The vector space spanned by (qfklkem is dense in C 0 ( M ) so (11.4.55) implies

f(p) = f ( q I, Vf€C0(M).
In particular, this means p = q.
Set 5" := distg (p 71,7 q n so that 5TL -> 0 as n -> OO. For n large there exists a unique
9

unit vector Vn 6 Ten M such that qn = exp(6nvnl. After passing to subsequences we can
assume that Vn converges to a unit vector V E TpM .
Since FE is an immersion for any 5 > 0 sufficiently small we deduce that, given such
a small 50 > 0, there exists k E {1, . . . , D(e0)} such that
d\I/k(V)1, 7é 0. (11.4.56)
Denote by 'Yn the smooth path
'yn : 10,5n1 -> M, 'yn(t) = exppn (TVn),
 Spectral Geometry 601

Note that, for large n, 'yn is the unique unit speed minimal geodesic connecting pn to q
Applying the mean value theorem to the function

f1<:(t) = \P/¢('yn(tll» t E [0»5n]

we deduce that, Vi, E N, there exists To E [0, in] such that
dq/k('7(¢n)) = 0. (1 l.4.57)
Note that as n -> OO we have

'YH(TH) -> p and 'y°(7'n) -> V

Hence, letting n-> oo in (1 l.4.57) we deduce
d\Ifk(V)p = 0.
This contradicts (11.4.56) and completes the proof of Proposition 11.4. 13.

Proposition 11.4. 13 shows that the function

&Sm-I-2 (27fl"°'
,Q5/@=J£{5=m X M - > ] R , Fe( PJ] ) (11.4.58>
Bm(w)

is the correlate associated to the embedding FE : M -> leD(€l, see Example 4.2.27.
Moreover, as e > 0, the metric gS determined by this correlate converges to the original
metric in C0(]l/I). We state this conclusion as a corollary first observed by S. Zelditch
[136].

Corollary 11.4.14. Using the spectral data we can produce arbitrarily accurate C0-
approximations of the original metric.

From the above corollary we deduce that the volume measure dvge converges weakly
to the volume measure dvg. We can say a bit more. Fix g-normal coordinates at near a
point p 6 M. In these coordinates the metric g5 is described by the matrix

g = (S -|- O(e2). (11.459>

Using (11.4.41) with l and new = 2 we deduce

@3351/@(x,y) = const.
(fRun
w(x )x° +0d $ + O(€2) O(e2).

We denote by v E t he Levi-Civita connection of g6 and by V the Levi-Civita connection

of the original metric g. Their difference is a tensor
As = V 5 - V Q of( End(TM)).
Denote by FE: the Christoffel symbols of VS in the coordinates an and by F the Christoffel
symbols of V in the same coordinates. Since the coordinates where chosen normal at p we
have I`(0) = 0 and deduce
TpM (X) End(TpMl 9 AS( ) (0) -r(0> = to).
602 Lectures on the Geometry of Manzfolds

The formula (4.2.23) describing the Christoffel symbols in terms of correlators coupled
with (11 .4.59) implies that

lIkE(p)ll O(€2 )
and this estimate is uniform in p. Hence

up CO I (>(@2). (11.4.60>
The continuous dependence on data of solutions of ordinary differential equations [12]
implies that the geodesics of gS Cl-approximate the geodesics of g
Can we determine more geometric features of (M, g) from the geometry of (M, g6 )
from what we know so far? The next simple example shows that, given what we know at
this point, we cannot expect that the curvature of g6 approximates the curvature of g.

Example 11.4.15. Let us assume that dim M 2 and in the neighborhood of a point the
metric g is Euclidean. Suppose that on the same neighborhood g5 is conformal to g
be =€f@(1°)9> of: \ X2+Y2a fe EC°°(R).
According to (10.3. 16) the scalar curvature of 3E is

5-f€Ag*f6 6 -f€
Se J
7'dr
Consider the function w in Figure 11.3. We set

For t 6 [-1, 1] set

t t/at

9@(0) w(s/l5)ds = 6 w(7')dT, fE'(t) gE'(s)ds.

0 0 0
We deduce
5 t/5
fS(6) = 5
/ / 0
dt
0
w(ds)ds < 162 7

6 1°/6 7r/6
1
Agf6 I -'w(7°/5) -I- w(s)ds = rl5 w(s)ds w(1°/6) 2 0.
T' 0 0

Observe that the scalar curvature function 85 is radially symmetric and

36l'r=€=€
fel5lI > e _ / 52 / > ( 1 - 152)/.

This shows that the scalar curvature function 36 does not converge uniformly to 0, the scalar
curvature of the flat Euclidean metric.
As a curiosity, note that
6
3E(7°) = 0, VT < 5'
This shows that the metric gS is flat on the disk of radius 6/2 centered at the origin, but
becomes very curved along tiny circle of radius 5 centered at the origin.
 Spectral Geometry 603

We want to investigate the behavior of the Riemann curvature tensor of the metric gS .
Fix a point p and g-normal coordinates ( x ) at p. Then, the local frame Q i is nearly
g5-normal in the following sense
(11.4.59)
gglgxi 7 QUo) p (S -|- O(s)2 7
v, E
.
9 . lp (11 . 4 .60 ) O(e2).
m'L a .7

Denote by R5 the Riemann curvature tensor of go. Using the above estimates and (4.2.24)
we deduce

RImy(p )
4
gxkxzyéyg JiV6(a3,yl' ac y 0
4
gxkxjygyz K(xly)l as y 0 + o(€2).
Theorem 11.4.16. Fix a point p 6 M and normal coordinates (xvl) near p. For any multi-
induces a,@ 6 lZ20)M we set

655 @s@5é"(w,y)l 3: y 0-

Then there exists a universal constant gm that depends only on the dimension of M and
the weight w such that
S _ E'
55nRmj( p)a""-2(l + O(l82)l as 5 -> 0. ( 11.4.6l)
('i;jj 'go

Proof. Using (11.4.43) we deduce

1 my ijféj

C)
5
'of;jj <58j;¢j
iv
5 Kt ' Kt »Tllt)w -I-O(sN), 5 - > 0 . (1 l.4.62)
6

On the other hand from (11 .4.45) we conclude

OO

my ijféj iv dm(2A¢> la2,a2ye - 2 2

8mimg39¢y3 {Uk(2% if/)IH1<(t» or, 3/l} l:B=Y=07
k-1
( I I .4.63)

where we recall that

77(;v, 311 := dist(;u, if/l2»

n n

fHa(i,33,y) <1 X+-

2
(ti -t1(x1»y)) -x+- (9 2
-t7(2ay) >) 7

6%( ( t' Sy(;c) ) Qin <-¢ ,sy (w) ))

Let us introduce some notation.

. For a positive integer k we denote by 3k a generic mixed-partial derivative of order k

in the space variables ac, ye . We will write <9 instead of 31.
.. We denote by 3k*n the collection of k-th order derivatives of n(ac, y).
MX ) will denote a homogeneous polynomial of degree z' in the variables X .
604 Lectures on the Geometry of Manzfolds

To, (X WAY) will denote a polynomial which is homogeneous of degree k in the vari-
ables X and homogeneous of degree Z in the variables Y.

We then have the equalities

35{0 I )1l377lgl(a-17 (1 l.4.64)

62 I j)2l377l9ICa-2 + T1(3277)9ICa-17 (1 l.4.65)

3% = 9>3(9n)9£a-3 + u>1(@n)9>1(@2n)v-2 + o>1(@3n)9-1, (1 l.4.66)

34 3'4(@'7)9£a-4 + (9'2(817)9'1(8277) W a - 3
( I I .4.67)
+ l3'2(82TIl + 3°1(3TI)3'1(3377) W a - 2 + 3'1(34VI)9£a-1-
To simplify the presentation we will assume that in (1 l.4.6l) we have i LJ
2. Also, we will denote by 0(1) a function f(;z:,y) such that f(:v, y)lrIJ1,=y=0 0. The
computations in Subsection 5.2.3 show that

iPi(<8?7) = Tk(93U) = 0(1)- (11.4.68)

In particular, the above equalities show that the 1st and 3rd order derivatives of Ha are
O(1). We have

321 832 IUI<1HI<l 821 ( (352Up)Hk + 2ay2 U,,,,ay/ f -|- Ukéjmk

(a2.aj2u)fHk + (ajgvk) (82.9,) -|- (a2 u;) (ajzofk) ( ll.4.69)

+ 4(a21y2u,)(a21y2fHk) -|- u k a j m + ou),

a212a§ly2<u,]+ck>
= @ 2321:22 I (851y2 (/I¢)U'tk -|- 891 Uk692%k + 63/ 2 Uk@ylg'fk -|- Uka§1y2:}ck )
(@212a§1y2uk)m -|- (a§y2uk) (a2.2:rck)
+ 82 Uk82ly2 Wk -|- 82191Uk32292 Wk + 822y2 U k g x l y l g k -|- 321y2 Uk322ylg'(k

+ a2.2uk@§ymk -I- UI¢8x1 2

x 2 <99192 Wk + 0<1>.
2
( ll.4.70)

Using (11.4.64)-(11.4.67) we deduce that

4
la2 a§2 2 2
6m1a 2 3y1y2 l ( U , 9 k ) x = y = 0
j 0
Tgwk j 133 y 07

where the coefficients T are degree j polynomials in the derivatives of Up and 17 at the
point (at, y) = (0, 0). Using (11.4.68) we deduce

T,;*=T,§= 0
 Spectral Geometry 605

Moreover, in view to (11.4.68), the terms in TO

are due only to the 4-th order derivatives
of Wk. Upon inspecting (1 l.4.69) and (1l.4.70) (pay attention to the boxed terms in these
equalities) we see that the 4th order derivatives of 5£k are multiplied by Uk- According to
(U 1<:,h) the function Up is 0(1) if lc > 1. Hence = 0 for k > 1. We deduce TO
OO

it 'ij,1lj
d'rnl2kl ITQ5 T,Hk-1 -|-
Ki Kt iv
-|- T,?%<-2) la:=y=0
k l

B-13{-1lx=y=0 + B09-(0lm=y=0 + B15{1 lw='y=0 + 7

where

B_ = d'rn(2lTl2, B0 = dm(2)Tll, B1 = d771(2)T§ + d771(4)T; 7

The term B-1 can be alternatively described as

B 1 Am,¢¢;jj,0 - Am,1;j;ij,0»

where the coefficients Am70,@,0 are defined as in (11.4.46). Using (1 l.4.48) and (1 l.4.50)
we deduce
1-3_1 0.
To compute T11 we observe first that

?7(22 - pa) (w - if/02 -|- higer order terms. ( 11.4.7l)

i

Using (11.4.65) we can simplify (11.4.69) and (11.4.70) in the case k l as follows.

a2.a§2<u{?c1€ = (6216§2U1)fH1 + U1a21aj2fH1 + 0( l >, (1 l.4.72)

@2.2a§.y2<U19c1) = (a212a§y2u1)c1 + 821y u1822y2&c1

( I I .4.73)
-|- 8223/2 Ulgxlylgfl -|- u 1 3 2 1 2 6 5 1 / m 1 + O(1).
Using (11.4.65), (11.4.67) and (11.4.71) we deduce that
T11 - (6x1692
2 2
- 2
g@1$28y1y2)
2
Tll(0,0)

-|- 2 (621 U1 -|- 652111) l(0,0) + 2 (a U1 -|- a%y2U1) |(0,0)-

Using the transport equation ( U ) with h given by (11.3.46) we deduce that

U1 (33, y) = U1lY,$l3), U1(a2,a3) E 1.

These conditions imply that the Hessie of U1 (Amy) at (0, 0) is a quadratic form in the
variables Up = ( i - y ) so that
621U1(0, 0) -|- <921y1U1<0, 0) : : C9§2U1(0, 0) + <922y2U1(0, 0) : 0. (11.4.74)

Hence
T11 @35 a212a§1y2) Tll(0,0)
606 Lectures on the Geometry of Manzfolds

Using (5.2.24) we conclude that

Tll I ZR1212,

where Z is a universal constant, independent of (M, g). Hence

my
Ki - KtJ dml2l ZR1212( p lj'(0la:=y=0 -I- BIQUM I ac y 0-
k>1

Using (11.4.62) we deduce

dm(2)ZR1212l p)
()
6 _ 5 g {0 lx = y=
'gilhjj 'inn
i v
0» '7(t)w
6 6

l
+6 Z Bk9'(klm=y=0) v(t)'w
()
6
+ O(eN).
I<:21

The equality (11.4.61) now follows from the above equality by invoking (11.3.25), the
definition (11.4.29) of 5{k1J and Lemma 11.4.10.

We can now tie all the loose ends together. Fix a point p E M and g-normal coordinates
(xfi) at p. We denote by R5 the Riemann curvature of the metric gE determined by the
correlate (11.4.58) and by Kfj the sectional curvature of the metric gE along the plane
spanned by 8,,,, 3xj ,
H a j j i (p)
KiSjlp) areage (3,,,¢ , 3x/xi )
From Proposition 11.4.13 and Theorem 11.4.16 we deduce that there exists a constant
L = Lm(w) that depends on the dimension m and the function w, but it is independent of
the manifold M and the metric g such that
lim
e7\,0
KZ (p) Lw»K¢jlp)-
If there existed some manifold M and some metric g on M the convergence of g5 to g is
C2, then would could conclude that Lm = 1 since the curvature is a continuous function
in gag and its derivatives of order < _ 2.
Suppose now that M = So is the unit sphere in lR""+1 and g is the induced metric. The
metric is S O ( m -|- 1)-invariant and so is the associated Laplacian operator. In particular, the
eigenspaces are S O ( m + 1)-invariants. The orthogonal projections onto these eigenspaces
are S O ( m -|- l)-equivariant operators. From the description

é"(p, q ) w(x) K AlP, q I

A€spec(A)

deduce that the correlate (p, q ) is invariant with respect to the diagonal action of
S O ( m -|- 1) on Sm X So. This shows that the metric g6 is SO(m -|- 1)-invariant. Since
the action of S O ( m -|- 1) on Sm is transitive we deduce that there exists a constant CO > 0
such that

95 C69-
 Spectral Geometry 607

Since gS converges uniformly to g we deduce that 0E -> 1. In particular, we deduce that the
curvature of gS converges to that of g so L,,,(w) = 1. The next result summarizes the facts
outlined above.

Corollary 11.4.17. The spectral data determine the sectional curvature of the metric. More
precisely, the sectional curvatures of the correlate metric gS converge uniformly to the
sectional curvatures of the original metric g.

We can say a bit more. We want to show that the spectral data determine the whole
Riemann curvature tensor of g. Fix a point p 6 M and denote by Gr2(TpM) the Grass-
mannian of 2-planes in TpM and by 5ilTpm) the cone of inner products on T p M .
For any Riemann metric h on M the sectional curvature at p is a continuous function
Kg : GF2 (TpM) -> R.
There exists a continuous function

l
CO GF2ITpMI ) X SELI T Pm) -> AQT PM AQT PM (X) 7 '*) +-> Q(K nil
described by an explicit formula6 in [28, Eq. (1.12)], such that, for any Riemann metric g
on M, the quantity

Q(Kg,g,,)eA2TpM (X) A2TpM

coincides with the Riemann curvature tensor of g at p.

The sectional curvatures Kp of the metric g5 at p converges to the sectional curvature
g and go -> up. Denote by Rp the Riemann curvature of g5 at p. We deduce that
p

%(K9»95) -> 9?(Kp,9") Rp-

E
Re
Corollary 11.4.18. The spectral data determine the Riemann curvature of the original
metric g.

Remark 11.4.19. (a) Using the very deep results in [5, 109] one can show that the C0-
convergence of the metrics g5 coupled with the convergence of sectional curvatures imply
that g5 converges to g in Coo. We can rephrase this by saying that, for 5 > 0 sufficiently
small, the map F 5 : M -> ]RD(6) defined in (1 l.4.54) is a nearly isometric embedding.
(b) Let us observe that for any even, nonnegative temperate function w 6 5'(]R) we can
define a map
6777,-1-2 (2w)'"'
m9p+F >
Be (w) w/TQ) @k(z»)@k Q L 2 ( m , 9 ) .
k>0

When w has compact support, the range of this map is contained in a finite dimensional
subspace.
_ 2 . . . .
The case w(v°) e or was first investigated in [17] where, among many other things,
the authors show that for small 6 > 0 the map F w5 is a nearly isometric embedding of M in
6 The explicit formula of ,Q is rather complicated, but it is polynomial in K and h so it is continuous.
608 Lectures on the Geometry of Manzfolds

the Hilbert space L2 (M, g). The image of this embedding is not be contained in any finite
dimensional subspace.
52
Observe that in this case, if we set t we have

5"/¥(p, q ) y) e-* >I fk (p )\I 'k( q >. (11.4.75>

II
HtlXv
k>0

This is so called heat kernel of the Riemann manifold.

(c) Let us point out that the above results solve more than the original spectral perestroika
problem.

The geometry of the compact Riemann manifold ( M , g) is completely deter-

mined by the associated neat kernel. More precisely, the metric and its curvature
are determined by the behavior as t \ 0 of the jets of order 4 along the diagonal
of the heat kernel Ht (ac, y).

This is a rather unexpected conclusion: we can learn the "shape" of the manifold by inves-
tigating how the heat propagates or, equivalently, how the diffusion takes place.

Remark 11.4.20. (a) As explained in Remark 4.2.30, the correlators associated to embed-
dings have probabilistic interpretations. For example, when w is as in Figure 11.3, the
correlate 55 (p, q ) is associated to random functions on M of the form
K

f
Xgq/k,
/nu
where (X/§)k2 are independent random variables with mean zero and variances
6m-I-2 l2,/Tlm
variance IX; )
Be (w)
w(€ l.
As 5 \, 0 this random function converges in a precise sense to the "perfect chaos", the
so called called Gaussian white-noise. A plastic way of phrasing the conclusions of the
previous results is that there is a pattern in devolution to chaos which is very sensitive to
the background geometry.
(b) For every p 6 M we have a random symmetric bilinear form

GS := dU5(pl (X) dUe(pl : TpM X TpM -> R.

For 5 > 0 small, its expectation is the metric be induced by the embedding F 5 defined in
(1 l.4.54). We are inclined to believe that the random variable G5 converges almost surely
and L1 to the (deterministic) g. In other words GE is highly concentrated around its mean
a s \ , 0.
(c) Here is another little gem hidden under the probability cloak. Assume M oriented and
even dimensional m = d i M = 211. The differential of the random function US is a
random section dU5 of the cotangent bundle. One can show that, with probability 1, dUE
intersects the zero section of T*M transversally, i.e., dU5 has nondegenerate zeros. To such
 Spectral Geometry 609

a nondegenerate zero p we can associate a local index €(U6, p) = i t (see Exercise 7.3.47)
and from the Poincare-Hopf formula in Corollary 7.3.48 we deduce that

€(U5, p) = x ( M ) - (11.4.76)
dUelPl 0

In fact, above we have produced quite a bit more, namely a 0-dimensional current

Zdzfs €(U6, P)6p E

dUelPl=0

The "integral" of a given 0-form f over this random current is the random variable

(f, ZdUE ) 6((&, p)f(p)-

dUfslP) 0

The expectation of this random variable has a rather striking geometric description [106].
More precisely we have

IE{(f, ZdU,5 )] f€(Vge), (1 l.4.77)

M

where e(V9@) is the Euler form associated to the Levi-Civita connection of the metric 969

e(V9) l
pf RE)
(mm/2 ( 7

where Rs is the Riemann curvature tensor of the metric ge- Thus

l
1E[(f» ZdU»5 Pf ( R) (1 l.4.78)
(2wl""/2 M f
7

Note that when f 1, then

€lU6, (11.4.76)
(1»Zdu) p) XlM)-
dUelPl=0

This is no longer a random quantity. Using this in (11.4.78) we deduce

1
x(M)
(2w)"*/2 M pf ( / R).
5

This is the Gauss-Bonnet theorem for the metric ge, On the other hand, we know that R E`
converges uniformly to the Riemann curvature R of the original metric g. If we let 5 \ 0
we obtain the Gauss-Bonnet theorem for the original metric g !
This "accident" is a manifestation of a more general phenomenon. In particular, the
general Gauss-Bonnet theorem has a probabilistic interpretation. For details we refer to
[105, 106].
This page intentionally left blank
 Chapter 12

Dirac Operators

We devote this last chapter to a presentation of a very important class of first order elliptic
operators which have numerous applications in modern geometry. We will first describe
their general features, and then we will spend the remaining part discussing some frequently
encountered examples.

12.1 The structure of Dirac operators

12.1.1 Basie definitions and examples

Consider a Riemann manifold (Mg), and a smooth vector bundle E -> M .

Definition 12.1.1. A Dirac operator is a first order p.d.o.

D C°°(E) -> COO(E),
such that D2 is a generalized Laplacian, i.e.,

(t(D2)(w,€) = -151311E am) 6 T*m.

The Dirac operator is said to be graded if E splits as E = E0 ® El, and D(C°°(E¢ll C
COO (E(1+1> mod Zl' In other words, D has a block decomposition

0 A
D
[ ]
B 0
Note that the Dirac operators are first order elliptic p.d.o.-s.

_
Example 12.1.2 (Hamilton-Floer). Denote by E the trivial vector bundle RQ" over the
circle S1. Thus cOO (E) can be identified with the space of smooth functions
u:S1->]R2"
Let J : cOO (E) -> cOO (E) denote the endomorphism of E which has the block decompo-
sition
r-w
II

611
612 Lectures on the Geometry of Manzfolds

with respect to the natural splitting RQ" = R" ® R". We define the Hamilton-Floer
operator
du
.]_
3': C°°(E) -> COO(E), 9"u, Vu 6 C°°(E).
do
dz
Clearly, 9:2 do2 is a generalized aplacian.

Example 12.1.3 (Cauchy-Riemann). Consider the trivial bundle Q2 over the complex
plane (C equipped with the standard Euclidean metric. The Cauchy-Riemann operator
is the first order p.d.o. D : Q2 -> Q2 defined by

where Z QUO + ii, and

H u
'U
I
0
> 2 82 0<92
[ H]
u
'U
7

l l
3. 2<
_gm <92 Z 2 (a, + my).

A simple computation shows that D is a Dirac operator.

Example 12.1.4. Suppose E is a vector bundle over the Riemann manifold (M, g), and
D : C°°(M) ->C°°(M) is a Dirac operator. Denote by M the cylinder R X M, and by g
the cylindrical metric g = dt2 + g on M, where t denotes the coordinate along the factor R
We have a natural projection 'or : M -> M, (15,513) »-> or, and we set E := 7r*E,
the pullback of E to M via '/r. A section 'ii of E is then a smooth 1-parameter family of
sections u(t) of E, t E R. In particular, we can define unambiguously
)_
@t2
i l t 0 'X 1° l
h h
t
0
+ mol _/ult
07X0 ))»
0

where the above limit is in the t-independent vector space Emf' Now define

D , c oo(E ® E l C7°°(E E),

A

_>

by
u(tl
v(t) ]
Then D is a Dirac operator called the suspension of D. Note that the Cauchy-Riemann
operator is the suspension of the Hamilton-Floer operator.

Example 12.1.5 (Hodge-DeRham). Let (M, 9) be an oriented Riemann manifold. Then,

according to the computations in the previous chapter, the Hodge-DeRham operator

d+ d* Q' (M)-> Q' (M)~

is a Dirac operator.
 Dirac Operators 613

Let D : coo (E) -> cOO (E) be a Dirac operator over the oriented Riemann manifold
(M, g). Its symbol is an endomorphism 07(D) : 'rr*E -> 'rr*E, where 'ii : T*M -> M
denotes the natural projection. Thus, for any Hz: 6 M, and any E Tas M, the operator

0(€) 0(D)(€,1v)

is an endomorphism of Ex depending linearly upon §. Since D2 is a generalized Laplacian

we deduce that

@(§)2 I 0(D2)(1a€) = -l€l§11E,,

To summarize, we see that each Dirac operator induces a bundle morphism

e : T*M (X) E -> E (§,e) »-> c(§)e,

such that @(€)2 Z -l€l2, From the equality

2
¢@+n)2 = -la+vl v m @ T , m , :UE M
we conclude that

{0(€)»0(v)} = -29(€,'7)]1E,

where for any linear operators A, B we denoted by {A, B} their anticommutator

{A, B} :z AB + BA.

Definition 12.1.6. (a) A Clifford structure on a vector bundle E over a Riemann manifold
(M, g) is a smooth bundle morphism e : T*M (X) E -> E, such that

{0(€l» c(t7)} = -2g(€»vl]1E, We 6 Mm),

where for every 1-form a we denoted by c(al the bundle morphism e(a) : E -> E given
by

e(a)u e(a,ul, Vu E C°°(E).

The morphism e is usually called the ClqiItord multiplication of the (Clifford) structure. A
pair (vector bundle, Clifford structure) is called a ClyjL'ord bundle.
(b) A Z2-grading of a Clifford bundle E -> M is a splitting E = E+ ® E' such that,
Va E Q1 ( M ) , the Clifford multiplication by a is an odd endomorphism of the superspace
c°°(E+) e c°°(E_), i.e., c(0¢)C°°(E*l C c°°(E*).

Proposition 12.1.7. Let E -> M be a smooth vector bundle over the Riemann manifold
( M , g). Then the following conditions are equivalent.
(a) There exists a Dirac operator D : cOO (E) -> cOO (El.
(b) The bundle E admits a Clutord structure.
614 Lectures on the Geometry of Manzfolds

Proof. We have just seen that ( a ) ( b ) . To prove the reverse implication let
c:T*M (X) E->E
be a Clifford multiplication. Then, for every connection
V : C°°(E)->C°°(T*M (X) E),
the composition
V e
D CO V : c<>°(E) -> c°°(T*m (X) E) -> c°°(E)
is a first order p.d.o. with symbol C. Clearly D is a Dirac operator.

Example 12.1.8. Let (M, 9) be a Riemann manifold. For each QCEM, and § € T x M
define
0(5) A°T;M -> A°T;M
by
= (€ - i€)W»
0(€)(»J

where 6€ denotes the (left) exterior multiplication by g, while it denotes the interior differ-
entiation along 5* e T,,,]l/I - the metric dual of €. The Exercise 2.2.55 of Section 2.2.4
shows that c defines a Clifford multiplication on A°T*M. If V denotes the Levi-Civita
connection on A'T*M , then the Dirac operator COV is none other than the Hodge-DeRham
operator.

Exercise 12.1.9. Prove the last asse ion in the above example.

The above proposition reduces the problem of describing which vector bundles admit
Dirac operators to an algebraic-topological one: find the bundles admitting a Clifford struc-
ture. In the following subsections we will address precisely this issue.

12.1.2 Clifford algebras

The first thing we want to understand is the object called Clifford multiplication.
Consider a real finite dimensional, Euclidean space (V, g). A Cluj'ord multiplication
associated with (V, g) is then a pair (E, pa , where E is a K-vector space, and p : V ->
End (E) is an R-linear map such that
{0('u»l,0(v)} = -2g(u,v)]lE, We Q VL
If (Q ) is an orthonormal basis of V, then p is completely determined by the linear operators
,0i = P(@¢) which satisfy the anti-commutation rules

{Pi>Pj} I -26ij]1E.

The collection ( p ) generates an associative subalgebra in End (E), and it is natural to try
to understand its structure. We will look at the following universal situation.

Definition 12.1.10. Let V be a real, finite dimensional vector space, and

q:V X V-HR
 Dirac Operators 615

a symmetric bilinear form. The Clifford algebra Cl(V, q) is the associative R-algebra with
unit, generated by V, and subject to the relations
{'u,,'u} up Vu -
-2q(u,ul ]1 'v'u,u 6 V
.
Proposition 12.1 11. The Clijtord algebra Cl(V, q) exists, and is uniquely defined by its
universality properly: for every linear map j : V -> A such that A is an associative R-
algebra with unit, and {](u), ](v)} = -2q('u,, Fu) - ]l, there exists an unique morphism of
algebras i15 : C1(V, q) -> A such that the diagram below is commutative.
'L
V > Cl(V, q)

<I>
J

A
'L denotes the natural inclusion V --> C1(V, q).
Sketch of proof. Let A = ®k20V®k (V®0 R) denote the free associative R-algebra
with unit generated by V. Set
cllv, Q) = A/I,
where I is the ideal generated by
{u®v+'u®u+2q('u,,'u)®1; 'u,,v E V},
The map Z is the composition V --> A -> Cl(V, q) where the second arrow is the natural
projection. We let the reader check the universality prope y.

Exercise 12.1.12. Prove the universality prope y.

Remark 12.1.13. (a) When q E 0 then C1(V, 0) is the exterior algebra A°V.
(b) In the sequel the inclusion V <-> Cl(V, q) will be thought of as being part of the defini-
tion of a Clifford algebra. This makes a Clifford algebra a structure richer than merely an
abstract R-algebra: it is an algebra with a distinguished real subspace. Thus when thinking
of automorphisms of this structure one should really concentrate only on those automor-
phisms of R-algebras preserving the distinguished subspace.

Corollary 12.1.14. Let ( V , i t ) ( i = 1, 2) be two real, finite dimensional vector spaces

endowed with quadratieforms ii : V -> R. Then, any linear map T : VI -> V2 such that
q2(T"u) = QUO (u), Vu E VI induces a unique morphism of algebras T# : Cl(v1,qll ->
Cl(V2, (12) such that T#(V1l C Vg, where we view V; as a linear subspace in C1(V/, Cal.
The correspondence T »-> T# constructed above is funetorial, i.e., ( l m )# = ]1C1(\&qi),
and ( S o TO# = S# o T#, for all admissible S, and T.

The above corollary shows that the algebra Cl(V, q) depends only on the isomorphism
class of the pair (V, q)= vector space -|- quadratic form. It is known from linear algebra that
the isomorphism classes of such pairs are classified by some simple invariants:
(dim V, rank q, sign q) .
616 Lectures on the Geometry of Manzfolds

We will be interested in the special case when dim V = rankq = signq = n, i.e., when
q is an Euclidean metric on the n-dimensional space V. In this case, the Clifford algebra
Cl(V, q) is usually denoted by Cl(V), or Cin. If (611 is an orthonormal basis of V, then we
can alternatively describe Cln as the associative R-algebra with 1 generated by (€¢l» and
subject to the relations
€ ¢ € j -I- € j € ¢ -2(Sj.
I

Using the universality property of Cin, we deduce that the map

V -> c1(v>, U »-> -u E Cl(V)
extends to an automorphism of algebras a : Cl(V) -> Cl(V). Note that a is involutive,
i.e., 0 2 = ]l. Set
CMV) := ker(a - ill, COW) = ker(a + II).
Note that Cl(V) = Cl0(V) ® cll<v>, and moreover
c1<v> - c1"(v) C c1(€+*1) mod 2(v>,
i.e., the automorphism a naturally defines a Z2-grading of Cl(V). In other words, the
Clifford algebra Cl(V) is naturally a super-algebra.
Let (Cl(V) , +, *I denote the opposite algebra of Cl(V). Then Cl(V) coincides with
Cl(V) as a vector space, but its multiplication * is defined by
cc *y y :u Va2,y E CW),
where "-" denotes the usual multiplication in Cl(V). Note that for any u, 'U 6 V
u-v+v-Iu,=u*'v+'v*7u,,
so that, using the universality property of Clifford algebras, we conclude that the natural
injection V C--) Cl(V) extends to a morphism of algebras Cl(V) -> Cl(V). This may
as well be regarded as an antimorphism Cl(V) -> C1(V) which we call the transposition
map, 11: »-> mb. Note that
$1 U2 "uL/1,)7 'Ufr '01, \V/Ui E V.
For :u E C1(V) we set
QQ* ¢ = <@(w))" = 0(a;°>l.

The element it is called the adjoint of x.

For each U 6 V define e('u) 6 End (A°V) by
c(u)w := (e'U - i'Ulw Vi E A°V7
where as usual e'U denotes the (left) exterior multiplication by u, while i'U denotes the inte-
rior derivative along the metric dual of 'U. Invoking again the Exercise 2.2.55 we deduce
e(v)2 iv .go2
so that, by the universality property of the Clifford algebras, the map c extends to a mor-
phis of algebras e : Cl(V) -> End (A°V).
Exercise 12.1.15. Prove that Vi E Cl(V) we have
c(w*)~=~c(w)* 7

where the * in the right-hand-side denotes the adjoint of e(x) viewed as a linear operator
on the linear space A'V endowed with the metric induced by the metric on V.
 Dirac Operators 617

For each QUO E Cl(V), we set 0(;u) := e(a:)1 E A'V. This is an element of A°V, called
the symbol of a:. The resulting linear map
C1(V) 9 :u »-> 0(:1:) E A'V 7

is called the symbol map. If (et) is an orthonormal basis, then

0(€i1 ° ° ° € i k ) = 811 /\ /\€»;k 'Vi

This shows that the symbol map is bijective since the ordered monomials
{€'L1'° €»;k , 1 < 1 1 > - - i k < dim V}
form a basis of Cl(V). The inverse of the symbol map is called the quantization map and
is denoted by q : A°V -> Cl(V).

Exercise 12.1.16. Show that q(A@"€"/oddv) clever/odd M •

Definition 12.1.17. (a) A ]K(=]R-,(C)-vector space E is said to be a K-ClQ§"ord module if

there exists a morphism of R-algebras
p : Cl(V) -> Ends(El.
(b) A K-superspace E is said to be a K-Clifford s-module if there exists a morphism of
s-algebras
p : Cl(V) -> Ends(El.
(c) Let (E, p) be a K-Clifford module, p : Cl(V) -> EndK(E). The module (E, p) is said
to be self-adjoint if there exists a metric on E (Euclidean if K = R, Hermitian if K = (C)
such that

PW) = 0(a:)* Vcc 6 c1<v>.

We now see that what we originally called a Clifford structure is precisely a Clifford
module.

Example 12.1.18. A°V is a self-adjoint, real C1(V) super-module.

In the following two subsections we intend to describe the complex Clifford modules.
The real theory is far more elaborate. For more information we refer the reader to the
excellent monograph [83].

12.1.3 Clifford modules: the even case

In studying complex Clifford modules it is convenient to work with the complexified Clif-
ford algebras
mn Gun ®RC~
The (complex) representation theory of (CiTL depends on the parity of n so that we will
discuss each case separately. The reader may want to refresh his/her memory of the con-
siderations in Subsection 2.2.5.
618 Lectures on the Geometry of Manzfolds

Let n = Zn, and consider an n-dimensional Euclidean space (v, 9)- The decisive step
in describing the complex Cl(V)-modules is the following.

Proposition 12.1.19. There exists a complex Cl(V)-module S = §(V) such that

~
(Cl(V) = End@(S) as (C-algebras.
(The above isomorphism is not natural; it depends on several auxiliary choices.)

Proof. Consider a complex structure on V, i.e., a skew-symmetric operator J : V -> V

such that J2 = -]IV. Such a J exists since V is even dimensional. Let

{€l7.tl7°°°7€/<77.tk7}
be an orthonormal basis of V such that Jet = fi , W.
Extend J by complex linearity to V ®1R( (C. We can now decompose V (X) (C into the
eigenspaces of J
l
v=vw@v 7

where V1,0 = ker(i - J) and V0,1 = ker(i + J). Alternatively,

V1,0 = span@(ej - i f s ) , V0,1 = span(c(ej -|- ifj).
The metric on V defines a Hermitian metric on the complex vector space (V, J)
h(u,'v) = g(u,v) + ig(u, Jo),

(see (2.2.15) of Subsection 2.2.5) which allows us to identify

~ vi*
~=@

V0,1 ""(r: (V, J) =@ (v1°'")*

(Above, VC* denotes the complex dual of the complex space (V, J).) With respect to this
Hermitian metric the collection
1
<_j<_A1

ifjl; 1
..
II

<'?j I§(@a

is an orthonormal basis of V1,0 while the collection

9

1
J <
o »

<'?j I§(@a +@fj); 1 § _ If

is an orthonormal basis of V0,1_ Set

:z Ai/1,0 Z A°,0v
§2k

A morphism of algebras e : @l(V) -> End ls2kl is uniquely defined by its restriction to
V (X) IC = V1,0
®
V'0,1

Hence, we have to specify the action of each of the components V L ( ) and V 0 , l _

The elements w E V1,0 will act by exterior multiplication

c(w)w = &@<w>w = kw /\ w, Vi 6 A°»0V.

 Dirac Operators 619

The elements E 6 V0,1 can be identified with complex linear functionals on Vi, and as
such, they will act by interior differentiation
Clwlwl 1/l2i(w)(w1 /\ - - - /\ wall
e
W (-1)lg@(w w)wlA A'¢I/iA /\ we,
j 1

where it: denotes the extension of g to (V (X) (C) X (V (X) (I) by complex linearity.
To check that the above constructions do indeed define an action of (Cl(V) we need to
check that, Vu E V, we have
c(v)2 ]1§2k

This boils down to verifying the anticommutation rules

{c(@¢)c(fj)} I 0, {c(@¢)»c(@jl} = -26ij {Cl.fil) Clfj)}.

We have
i
€ v'§(&z+€1)7 .tjz I§((°2;»
so that,

C(€il = €(€¢l - i(€i)» c(fjl = i(€((°$.1) + i(€.1ll-

The anti-commutation rules follow as in the Exercise 2.2.55 using the equalities

9@(€v,§i) = 611-
This shows §2 is naturally a Cl(V)-module. Note that d i g Say Mk SO that
271
ding E1"1dlcl§l = dimqj C1(V).
A little work (left to the reader) shows the morphism e is injective. This completes the
proof of the proposition.

Definition 12.1.20. The Cl(V)-module S(V) is known as the (even) complex spinoza mod-
ule.

Using basic algebraic results about the representation theory of the algebra of endomor-
phisms of a vector space we can draw several useful consequences. (See [120] for a very
nice presentation of these facts.)

Corollary 12.1.21. There exists an unique (up to isomorphism) irreducible complex <€1
module and this is the complex spinoza module S2k.

Corollary 12.1.22. Any complex @121-module has the form §2/<: (X) W , where W is an
arbitrary complex vector space. The action of @12k on Say (X) W is defined by
'U S (X) w) = c(v)s (X) w.

The vector space W is called the twisting space of the given Cluj'ord module.
620 Lectures on the Geometry of Manzfolds

Remark 12.1.23. Given a complex @121-module E, its twisting space can be recovered as
the space of morphisms of Clifford modules
W Homingk ( S o ,
Assume now that (V, g) is an oriented, 2k-dimensional Euclidean space. For any posi-
tively oriented orthonormal basis e l , . . . , 62k we can form the element
F 7,-k 61 ¢ a ¢ 62k E (c1(v).
One can check easily this element is independent of the oriented orthonormal basis, and
thus it is an element intrinsically induced by the orientation. It is called the chiralily oper-
ator defined by the orientation. Note that
F2 = 1 and Fa: = (-1)d€€"I` Vat E @10(v> U @1l<v).
Let S = S(V) denote the spinoza module of @l(V). The chirality operator defines an
involutive endomorphism of S, and thus defines a Z2-grading on §
§ §+®S (§* = ker(:=]1 - all,
and hence a Z2-grading of End(S). Since {v, l`} = 0, 'v/u 6 V, we deduce the Clifford
multiplication by 'U is an odd endomorphism of S. This means that any algebra isomorphism
(Cl(V) -> End (S(V)) is an isomorphism of Z2-graded algebras.

~
Exercise 12.1.24. Let J be a complex structure on V. This produces two things: it defines
an orientation on V and identifies S = S(V) = A*'0V. Prove that with respect to these
data the chiral grading of S is
~
§+/- _ Aeven/odd,0v.

The above considerations extend to arbitrary Clifford modules. The chirality operator
introduces a Z2-grading in any complex Clifford module which we call the chiral grading.
However this does not exhaust the family of Clifford s-modules. The family of s-modules
can be completely described as
{ se) W ; W complex s-space },
where ® denotes the s-tensor product. The modules endowed with the chiral grading form
the subfamily in which the twisting 5-space W is purely even.

Example 12.1.25. Let (V, go be a 2k-dimensional, oriented, Euclidean space. Then ARV(X)
(C is
naturally a Clifford module. Thus, it has the form
AiV (X) <c~ S (X) W
To find the twisting space, we pick a complex structure on V whose induced orientation
agrees with the given orientation of V. This complex structure produces an isomorphism
A1 V ®R c L A°»0v (X) A0=°v = s (X)
~ ~
<A°»0v>* = s ®1R s*.
This shows the twisting space is S*
 Dirac Operators 621

On the other hand, the chirality operator defines Z2 gradings on both S, and §*, so
that ARV (X) (C can be given two different s-structures: the chiral superstructure, in which
the grading of §* is forgotten, and the grading as a super-tensor product §®§*. Using
Exercise 12. 1.24 we deduce that the second grading is precisely the degree grading
A1 V (X) (C = As"€"v (X) <c ® Aoddv (X) Q.
To understand the chiral grading we need to describe the action of the chiral operator on
ARV (X) (C. This can be done via the Hodge *-operator. More precisely we have
F CU
Z-k +p(p - 1 ) * C U 'v'w€ApV (C. (12.l.1)
(X)

Exercise 12.1.26. Prove the equality 12.1.1.

In order to formulate the final result of this subsection we need to extend the automor-
phism a, and the anti-automorphism b to the complexified Clifford algebra @l(V). The
automorphism a can be extended by complex linearity
a(a2 ® z) = a(:u) ® z, Vi E Cl(V),
while b extends according to the rule
(113 (X) z)b = atb (X) Z.

As in the real case, we set it a(yb) = 0(y)7, We E (Cl(V).

Proposition 12.1.27. Let S(V) denote the spinoza module of the 2k-dimensional Euclidean
space (V, g). Then for every isomorphism of algebras
p : CMV) -> End(l;(§(Vl)
there exists a Hermitian metric on S ( V ) such that

0(y") = p(y)*, Via Q (co).

Moreover; this metric is unique up to a multiplicative constant.

Sketch of proof. Choose an orthonormal basis { 617 . , € 2 k of V, and denote by G the

finite subgroup of cl(vl generated by the basic vectors ei-
Pick a Hermitian metric h on S(V) and, for each g 6 G, denote by hg the pulled-back
metric
hg(s1,s2) :: h(p(g)s1,p(g)s2), VS1,S2 6 S
We can now form the averaged metric,
1
he by'
IGGY gaG

Each p(g) is an unitary operator with respect to this metric. We leave the reader to check
that this is the metric we are after. The uniqueness follows from the irreducibility of s(vl
using Schur's lemma.

Corollary 12.1.28. Let (V, g) as above, and p : Cl(V) -> Ending (E) be a complex Clyora'
module. Then E admits at least one Hermitian metric with respect to which p is self-adjoint.
622 Lectures on the Geometry of Manzfolds

Proof. Decompose E as § (X) W, and p as A (X) ]lW, for some isomorphism of algebras
A : (Cl(V) -> End(S). The sought for metric is a tensor product of the canonical metric
on S, and some metric on W.

12.1.4 Clifford modules: the odd case

The odd dimensional situation can be deduced using the facts we have just established
concerning the algebras <€12k. The bridge between these two situations is provided by the
following general result.
even
Lemma 12.1.29. (II m (cl 'rn-I-1'

Proof. Pick an orthonormal basis {e0, 61, . . . , et} of the standard Euclidean space lR""+1.
These generate the algebra Clm+1. We view Clm as the Clifford algebra generated by
{et, . . . , @m}, Now define
\If : Clm -> C1;1'+1, \I/(LE0 +:z:1) = £1:0 -|- 60 - acl,
where £170 E Clipen, and :cl E Clod. We leave the reader to check that this is indeed an
isomorphism of algebras.

Proposition 12.1.30. Let (V, g) be a (Zh + 1)-dimensional Euclidean space. Then there
exist two complex, irreducible Cl(V)-modules S+ ( V ) and § ( V ) such that
~
Cl(V) = End(l;(S-L) ® End@(§`) as ungraded algebras.
The direct sum § ( V ) = § ( V ) QB S' ( V ) is called the (odd) spinoza module.
Proof. Fix an orientation on V and a positively oriented orthonormal basis
6 1 , 6 2 » - - - » € 2 k + 1 - Denote by S2k-l-2 the spinoza module of (Cl(V EB R), where V QB R is
given the direct sum Euclidean metric and the orientation
or(V ® R) = or /\ or(lR).
Choose an isomorphism
p : (C1(V ® R) _> End@(§2k+2).
Then S2k-I-2 becomes naturally a super Cl2k,+2-module
+ 1

§2k+2 - §2I<:+2 ® §2I<:+2>

and we thus we get the isomorphisms of algebras
c1(vl 2 c1(v ®Rleven
~= End€'U€'f.(§2k+22 2
End (§;k+2> ® End l § 2 k + 2 ) -
As in the previous section, we can choose Hermitian metrics on s:l:lv) such that the
morphism
P : (co) -> End C (so) )
is self-adj pint, i.e.,
QW) = 0( u) *, Vu Q (co).
The above result can be used to describe the complex (super)modules of <C12k+1. We will
not present the details since the applications we have in mind do not require these facts.
For more details we refer to [83].
 Dirac Operators 623

12.1.5 A look ahead

In this heuristic section we interrupt a little bit the flow of arguments to provide the reader
a sense of direction. The next step in our story is to glue all the pointwise data presented so
far into smooth families (i.e., bundles). To produce a Dirac operator on an 're-dimensional
Riemann manifold (M, g) one needs several things.

(a) A bundle of Clifford algebras (3 -> M such that Gas 2" Cin or (Hn, Vi E M.
(b) A fiberwise injective morphism of vector bundles 'L : T*M <-> C? such that, Van' E M

{Z('U")7 2(Ul}@ I -29(u» u), V'u,,u 6 T x M .

(c) A bundle of Clifford modules, i.e., a vector bundle 8 -> M, together with a morphism
e : (3 -> End (8) whose restrictions to the fibers are morphisms of algebras.
(d) A connection on 8.

The above collection of data can be constructed from bundles associated to a common
principal bundle. The symmetry group of this principal bundle has to be a Lie group with
several additional features which we now proceed to describe.
Let (V, g) denote the standard fiber of T * M , and denote by Alltv the group of auto-
morphisms go of Cl(V) such that d)(V) C V. The group Autv is defined similarly, using
the complexified algebra (CZ(V) instead of Cl(V). For brevity, we discuss only the real
case. We need the following objects.

. A Lie group G which admits a smooth morphism p : G -> Alltv. Tautologically, p

.. defines a representation p : G -> GL(V), which we assume is orthogonal.

A Clifford module e : C1(V) -> End (E).
A representation / : G -> GL(E), such that, for every 'U 6 V, and any g E G, the
diagram below is commutative.
e(v)
E E
g

E
l l
E
g (12.l.2)

@(9-)
The above commutativity can be given an invariant theoretic interpretation as follows.
View the Clifford multiplication e : V -> End (E) as an element c E V* (X) E* (X) E.
The group G acts on this tensor product, and the above commutativity simply means that e
is invariant under this action.
In concrete applications E comes with a metric, and we also need to require that u is
an orthogonal/unitary representation.
To produce all the data (a)-(d) all we now need is a principal G-bundle P -> M such
that the associated bundle P X P V is isomorphic with T * M . (This may not be always
feasible due to possible topological obstructions.) Any connection V on P induces by
association metric connections VM OD1 T*M and VE on the bundle of Clifford modules
lin practice one requires a little more namely that VM is precisely the Levi-Civita connection on T * M . This
leads to significant simplifications in many instances.
624 Lectures on the Geometry of Manzfolds

8 = P X II E. With respect to these connections, the Clifford multiplication is covariant

constant, i.e.,
u Va
VE( ° ( u ) ic -|- QUO
IEI
CIV 00
I
( E
O
7 E (m), CO
uQ
This follows from the following elementary invariant theoretic result.

Lemma 12.1.31. Let G be a Lie group, and p : G -> Aut ( E ) be a linear representation of
G. Assume that there exists et 6 E such that p(g)e0 = et, Vg 6 G. Consider an arbitrary
principal G-bundle P -> X , and an arbitrary connection V on P. Then e0 canonically
determines a section U0 on P X P E which is covariant constant with respect to the induced
connection VE = p*(Vl, i.e.,

VEto 0.

Exercise 12.1.32. Prove the above lemma.

Apparently, the chances that a Lie group G with the above properties exists are very
slim. The very pleasant surprise is that all these, and even more, happen in many geomet-
rically interesting situations.

Example 12.1.33. Let (V, 9) be an oriented Euclidean space. Using the universality prop-
erty of Clifford algebras we deduce that each g 6 SO(V) induces an automorphism of
Cl(V) preserving V <-> Cl(V). Moreover, it defines an orthogonal representation on the
canonical Clifford module

e : CW) -> End(A°V),

such that

c(g . v)((»J) i g - (<2(u)(9 ~(»vll» VI 6 SO(V),v E V,w E A'V 7

i.e., SO(V) satisfies the equivariance property (12.l.2).

If (M,gl is an oriented Riemann manifold, we can now build our bundle of Clifford
modules starting from the principal SO bundle of its oriented orthonormal coframes. As
connections we can now pick the Levi-Civita connection and its associates. The corre-
sponding Dirac operator is the Hodge-DeRham operator.

The next two sections discuss two important examples of Lie groups with the above
properties. These are the spin groups Spin(n) and its "complexification" Spin°(n). It
turns out that all the groups one needs to build Dirac operators are these three classes: SO,
Spin, and Spine .
 Dirac Operators 625

12.1.6 The spin group

Let (V, g) be a finite dimensional Euclidean space. The group of automorphisms of the
Clifford algebra Cl(V) contains a very rich subgroup consisting of the interior ones. These
have the form
1
90x Cl(V) -> Cl(V) u »-> (px (u) QUO 'LL ac 7 Vu Q CPI,
where Hz: is some invertible element in Cl(V).
The candidates for the Lie groups with the properties outlined in the previous subsec-
tion will be sought for amongst subgroups of interior automorphisms. It is thus natural to
determine the subgroup
at E Cl(V)*; at V . at 1
CV},
where Cl(V)* denotes the group of invertible elements. We will instead try to understand
the Cluj"ord group
PW) = {m E c1*(v> a(:z;) . V -B-1 C V },
where a : Cl(V) -> C1(V) denotes the involutive automorphism of C1(V) defining its
Z2-grading.
In general, the map
Cl(V) 9 u *P'°> 0(:z;)7u,a:-1 E Cl(V)
is not an automorphism of algebras, but as we will see by the end of this subsection, if
so: E I`(V), then px = :l:(p3,, and hence a posteriori this alteration has no impact. Its
impact is mainly on the aesthetics of the presentation which we borrowed from the elegant
paper [9].
By construction, the Clifford group rlvl comes equipped with a tautological represen-
tation
1
p : FW) -> GL(V) p(a2l : U »-> a(;r;) 'U as

Proposition 12.1.34. kelp = (R*, -) C C1(V)*.

Proof. Clearly R* C her p. To establish the opposite inclusion choose an orthonormal

basis (€¢l of V, and let .CU 6 her p. The element .CU decomposes into even/odd components
x $0+$1,

and the condition &(.§U)€i.2U 61 translates into

Ito - 5U1l€i 61 (5130 -|- 5121), W.

This is equivalent with the following two conditions

[110, €] = 200€i - €¢$0 = £U1€¢ + 61331 i {I t , 61 } i 0, W •

In terms of the s-commutator, the above two equalities can be written as one
[@¢7$]s = 0 W.
626 Lectures on the Geometry of Manzfolds

Since °7 ac] is a superderivation of Cl(V) we conclude that

[y,;u]S = 0, 'v'y E Cl(V).
In particular, the even part 5130 lies in the center of Cl(V). We let the reader check the
following elementary fact.

Lemma 12.1.35. The center of the ClQ§"ord algebra i5 the field of scalars R C Cl(V).

Note that since {act, et} = 0, then all should be a linear combination of elementary
monomials en ---
ets none of which containing et as a factor. Since this should happen for
every i this means a l = 0, and this concludes the proof of the proposition.

Definition 12.1.36. The spinorial norm is the map

N : C1(V) -> Cl M N(;z:)=x>x°.

Proposition 12.1.37. The following hold.

(a) nlrlvll C R*.

(b) The map N : rlvl -> R* is a group morphism.

Proof. Let co: E F ( V ) . We first prove that ate E I`(V). Since 0(:z;lva2-1 E V, \7"v E V, we
deduce that

aw) 'U =v'1}b) = -a(:c) - u . x-1 6 v.

Using the fact that co: »-> :cb is an anti-automorphism, we deduce

Q( ( b ) - 1 -'u - aw) 6 v,
so that,
b
(1((1v')'1) 'U as E V ,

that is, (;u7)'1 G FM. Hence :u 6 F (V),Va:€ F (V). In particular, since 01(F(Vll C
I`(Vl, we deduce N(II(Vll C FIG).
For any U E V we have

o¢(N(fv)) 'U (N(;v))l1 i a(:13>xl (wbffrl Z M) - f) a 'u • a3'1} - (w

1
On the other hand, y = a(ac°) 'U at is an element in V, which implies YT = a(yb) = -y.
Hence

0((N(;v)) 'U (N(11))'1 -aW) ~y* (wb)-1

-
-a(;cb) a(;z:b)-1 _UT -Mb - (;z:b)-1 UT 'U.

This means N(a:) E her p R*

(b) If co, y E I`(V), then

N(;u - y) = (22y)b(wy) = ybwbwy = yan(©v)y = N(x1)yby = N(;v)N(y)-

 Dirac Operators 627

Theorem 12.1.38. (a) For every cc 6 I`(V), the transformation p(;c) of V is orthogonal.
(IQ) There exists a short exact sequence of groups
1 -> R* <-> F(V) £> O(V) -> 1.
(c) Every x 6 I`(V) can be written (in a non-unique way) as a product ac = 01 -up, --
Up 6 V . In particular; every element of r ( v l is Z2-homogeneous, i.e., it is either purely
even, or purely odd.

Proof. (a) Note that, 'v'tJ 6 V, we have N(v) = -IUI§. For every ac 6 I`(V) we get
-1 N(0((@3))N(x1)-1N(vl~
N(0>(11)('v)) N(a(x)va: ) N(0¢(11))N(v)N(33'1l I

On the other hand, $2 = 0(x2) = a(a1)2 so we conclude that

N(a))2 = N(0(;z:))2.
He r e , N(p(;c)(v)) = :l:N(tv). Since both N(u) and N(px(u)) are negative numbers, we
deduce that the only possible choice of signs in the above equality is -I-. This shows that
0(11) is an orthogonal transformation.
(b) and (c) We only need to show p(II(V)) = O(V). For cz: 6 V with txtg = 1 we have
0(:z:) 51: £12'1.

If we decompose 'U 6 V as Ask + u, where A 6 R and u J_ co, then we deduce

p(:I:)'v = -Ax + u.
In other words, p(;z:) is the orthogonal reflection in the hyperplane through origin which
is perpendicular to 111. Since any orthogonal transformation of V is a composition of such
reflections (Exercise), we deduce that for each T € O(V) we can find 101, . . . , up 6 V such
that
T »0('v1) a °p('U/<)-
Incidentally, this also establishes (c).

Exercise 12.1.39. Suppose V is a finite dimensional Euclidean space. Prove that any or-
thogonal operator T G O(V) is a product of at most dim V reflections.

Set
(31
F0(V ) .- F IvI n
Note that
'
,0(r0(V)) C SOIV) = {T E OW); defT = l }.
Hence, we have a short exact sequence
1 -> R* <-> row) i sow) -> 1.
Definition 12.1.40. Set
Pin(V) := { Q36 F (V); IN(;v)l 1},
and
Spin(V) := {;u E F (V); N(x) = 1 } = Pin(V) F0IVI,
628 Lectures on the Geometry of Manzfolds

The results we proved so far show that Spin(V) can be alternatively described by the
following "friendlier" equality
Spin(V) 'U1 'U2k 7 l€>0,'Ui€V, IUil=17 W 1. . . . $k}.
In particular, this shows that Spin(V) is a compact topological group. Observe that
Spin(V) is a closed subgroup of the Lie group GL(Cl(V) ). This implies (see Re-
mark 1.2.31 and [61, 128]) that Spin(V) is in fact a Lie group, and the map Spin(V) -->
Cl(V) is a smooth embedding. The proof of the following result is left to the reader.

Proposition 12.1.41. There exist short exact sequences

1 -> $2 -> Pin(V) -> O(V) -> 1
l -> Z2 -> Spin(V) -> SO(V) -> 1.

Proposition 12.1.42. The morphism p : Spin(V) -> SO(V) $5 a covering map. Moreover,
the group Spin(V) $5 connected if dim V > 2, and simply connected if d i V > 3. In
particular; Spin(V) is the universal cover of SO(V), when dim V > 3.
Proof. The fact that p : Spin(V) -> SO(V) is a covering map is an elementary conse-
quence of the following simple observations.

(i) The map p is a group morphism,

(ii) The map p is continuous, and proper,
(iii) The subgroup her p is discrete.

Since SO(V) is connected if d i V > 2, the fact that Spin(V) is connected would
follow if we showed that any points in the same fiber of p can be connected by arcs. It
suffices to look at the fiber p-1ll) = { - l , 1}.
If u, 'u 6 V are such that lul = lvl = l , u J. v, then the path
'y(t) = ( c o s t + ' s i f t ) ( u c o s t - sift), 0 < t g tr/2
lies inside Spin(Vl, because lu cost :|: 'U sin tl = 1, and moreover
'y(0l =: -1, 'Mw/2) l.
To prove that Spin(Vl is simply connected, we argue by induction on dim V.
If dim V = 3, then Spin(V) is isomorphic to the group of unit quaternions (see Ex-
ample 12.1.55), and in particular, it is homeomorphic to the sphere S3 which is simply
connected.
Note that if V is an Euclidean space, and U is a subspace, the natural inclusion U <->V
induces morphisms
Cll U) <-> Cl(V), SO(U) + SO(V), Spin(U) --> Spin(V),
such that the diagrams below are commutative
p
Spin(Vl > SO(V) Spin(V) " 1
Cl(V)

1
'Z

Spin(Ul > SO(U)

1 7

Spin(U) > Cl(U)

p
 Dirac Operators 629

Hence, it suffices to show that if dim V > 3, then for every smooth, closed path
11: [0,1]-> Spin(V), 11(0) = 71(1) = 1,
there exists a codimension one subspace U 6+ V such that 'EL is homotopic to a loop in
Spin(U). Fix unit vector e 6 V, set u(t) := ,0fa(¢> 6 SO(V), and define v(t) := 'u,(t)e 6
V.
The correspondence t »-> 'u(t) is a smooth, closed path on the unit sphere sdim v - 1 C
V. Using Sard's theorem we deduce that there exists a vector so 6 sdim V-1, such that
v(t) go :l::c, Vt 6 [0, l].
In particular, the vectors u(t) and cc are linearly independent, for any t. Denote by f (t) the
vector obtained by applying the Gramm-Schmidt orthonormalization process to the ordered
linearly independent set {a:,'u(t)}. In other words,

f (t)
1
If0(t)l /0(t)' for) 'v(t)-(v(t) . :r:):c,

where we denoted the inner product in V by We can find a smooth map

0 : [0, 1] -> [0,27r),
such that,
0(0) = 0(1) = 0, v(t) = accos 29(t) + f (t) sin 20(t).
Set

o(t) := cos 0 + asf sin 0 6 C1(V).

Note that

07(t) as COS
0
2
f
A

s1n-
0
2
) I-$cos 0
2
A

f s m 2-
0
7

so that c7(t) € Spin(V).

On the other hand,
Ulf)-llf) = casH - :of sin 0,

and
(7(t)x<7't) = (x cos 0 -|- f sin 9)(cos 0 - of sin H) = x cos 20 -|- f sin 20 60.

In other words,

00(0)33 = u(tl.

For every S 6 [0, 1] define

0S(t) := '&S(t) = cos 80 -|- :of sin 50, {LS(t) := 0S(t)-1a(t)0S(t),
and
'u,8(t) ¢= p;slltlu(t)pos(tl 6 So(v).
630 Lectures on the Geometry of Manzfolds

Observe that for S = 0 we have M t ) = '&(t). On the other hand, for s l we have

Ullt lib I P0-s1(1g)Ulil,00S(t)£6 I IO0s(t)'1'U(tl I $.

If we denote by U the orthogonal complement of Cr: in V, we deduce that up (t) 6 Spin(U),

Vt 6 [0, 1].

We can view Spin(V) as a submanifold of Cl(V), and as such we can identify its Lie
algebra spin(V) with a linear subspace of Cl(V). The next result offers a more precise
description.

Proposition 12.1.43. Consider the quantization map q : A*V -> Cl(V). Then
spin(V) = q(A2V)~
The Lie bracket is given by the commutator in Cl(V).

Proof. The group I`(V) is a Lie group as a closed subgroup of the group of linear trans-
formations of Cl(V). Since the elements of I`(V) are either purely even, or purely odd, we
deduce that the tangent space at l 6 I`(V) is a subspace of
E = { a ; 6 c1ev@"(v); :co viz: 6 V , V u G V }.

Fix :z:€E, and let € 1 , . . . , € » n be an orthonormal basis of V. We can decompose 11: as

co: = £120 +€1£c1,

where kg 6 Cl(V)ev€", and C171 6 Cl(V)°dd are linear combinations of monomials involv-
ing only the vectors e t , . . . ,en. Since [:z:0, 81] = 0, and {:z:1, €1} = 0, we deduce
1 1
61561 2 [61»1U1] 2[e1,a:] EV.

In particular this means .2111 e R ® V e C 1 ( V ) .

Repeating the same argument with every vector et we deduce that
E C ] R € B s p a n { e - e j , 1 < i < j < dim V} R€ 9 qlA2vl.
Thus,
T1F0(v) C R ® q(A2V).
The tangent space to Spin(V) satisfies a further restriction obtained by differentiating the
condition N(a3) = l. This gives

spin(V) C {$3 6 IR ® q (A2V) ,' we -|- x = 0} = q(A 2 V)-

Since dim spin(V) = dim QW) = dim AW we conclude that the above inclusion is in
fact an equality of vector spaces.
Now consider two smooth paths al)?! I (-(23€) -> Spin(V) such that 55(0) = y(0) = 1.
The Lie bracket of x'(0) and 3)(0) is then found (using the Exercise 3.1.21) from the equality

cl3(t)y(t)£u(tl'1y(tl 1 -|- [;ic(0), y°(0)]t2 -|- o(t3) (as t -> 0),

 Dirac Operators 631

where the above bracket is the commutator of 50(0) and 21(0) viewed as elements in the
associative algebra Cl(V).

To get a more explicit picture of the induced morphism of Lie algebras

p* : SPiH(V) -> QW),

we fix an orientation on V, and then choose a positively oriented orthonormal basis
{et, . . . , en} of V, (n = dim V). For every an 6 spin(V), the element p*(ac°) 6 spin(V)
acts on V according to

0*(x1)v ac' . 'U 'U . co.

If
.CU = .ClIi'€',1€j 7

i<j

then

p*(x)€i -2 £Uij€1;, £17»,;j 1/3j¢ .

Note the following often confusing fact. If we identify as usual (V) 2 A2V by
M) 9 A»->wA g(Ae,;, €j)€i /\ € j glad¢,A€jl€i /\ 6 7

i<j i<j

then the Lie algebra morphism p* takes the form

up* (23) 29(€z» P * ( f u l @ j l € ¢ /\ €1 2 Cvfzjefz /\ € j 207(a:) 6 A2V,

¢<j 1l<j

where O' : Cl(V) -> A'V is the symbol map, e e l »-> et /\ et. In particular, this shows p*
is an isomorphism, and that the map p : Spin(Vl -> SO(V) is also a submersion.

' I A word of warning. If A 6 W has the matrix description

Adj

with respect to an oriented orthonormal basis { then the 2-form

associated to A has the form

SO that

p* 1(A) (l2.1.3)

The above negative sign is essential, and in many concrete problems it makes a world of
difference.
632 Lectures on the Geometry of Manzfolds

Any Clifford module <2) Cl(V) -> EndK (E) defines by restriction a representation
<l> Spin(V) -> GL11<(E)~
The (complex) representation theory described in the previous sections can be used to de-
termine the representations of Spin(V).

Example 12.1.44 (The complex spinoza representations). Consider a finite dimensional

oriented Euclidean space (V, g). Assume first that dim V is even. The orientation on V
induces a Z2-grading on the spinoza module § = S+ ® §'. Since Spin(V) C Cl€'Ue"(V)
we deduce that each of the spinoza spaces S:t is a representation space for Spin(V). They
are in fact irreducible, nonisomoiphic complex Spin(V)-modules. They are called the
positive/negative complex spin representations.
Assume next that dim V is odd. The spinoza module slvl
§<v>=§+(v) ® S'(V)
is not irreducible as a (Cl(V) modules. Each of the modules s:tM is a representation
space for Spin(V). They are irreducible but also isomorphic as Spin(V)-modules. If we
pick an oriented orthonormal basis e l , , €2n+1 of V then the Clifford multiplication by
-
w = el - ' € 2 n + 1 intertwines the :t components. This is a Spin(V) isomorphism since LU
lies in the center of Cl(V).

@)Convention. For each positive integer n we will denote by in the Spin(n)-module

defined by
S 5 §+(1R(2'<> ® §°(R2'") if n = 212,
and
iv
§-(R2'°+1) if n = QA; + 1.
N=Spin(2k+1)

§2k-I-1 §+(1R(2"=+1)
The module Sn will be called the fundamental complex spinoza module of Spin(n).

Exercise 12.1.45. Let (Vlgl be a 2k-dimensional Euclidean space and J : V -> V a

complex structure compatible with the metric g. Thus we have an explicit isomorphism
A : @1(V) -> End (A"0V).
Choose u 6 V such that lulg = l and then for each t E R set
q = q ( t ) = cost-I-fu, - s i f t , v = Jo.
Note that q 6 Spin(V) so that A(q) preserves the parities when acting on A*»0(V). Hence
A(q) = A+(q) ® A_(q) where A+/_(ql acts on A e v e n / o d d , 0 v _ Compute tr (A+/_(q))
and then conclude that A e v e n , / o d d , 0 v are non-isomorphic Spin(V)-modules.

Proposition 12.1.46. Let <15 Cl(V) -> End ( E ) be a self-adjoint Cluord module. Then
the induced representation of Spin(V) is orthogonal (unitary).

Exercise 12.1.47. Prove the above proposition.

Exercise 12.1.48. Prove that the group Spin(V) satisfies all the conditions discussed in
Subsection 10.1.5.
 Dirac Operators 633

12.1.7 The complex spin group

The considerations in the previous case have a natural extension to the complexified Clif-
ford algebra Gln. The canonical involutive automorphism
C1n-> Cin
extends by complex linearity to an automorphism of (Zin, while the anti-automorphism
b : Cln -> Cln extends to (Cin according to the rule

(u (X) z)b = 'u (X) Z.

As in the real case set QUT := a(a:)7, and N(ac) = b at. a

Let (V, g) be an Euclidean space. The complex Cluj'ord group I`°(V) is defined by
F °(v) = {;u 6 (c1(v>*; 0(a:) p a : -1 E V Vu E v}.
We denote by pC the tautological representation pC : I1e(V) -> GL(V, R). As in the real
case one can check that
pc(Iwc(V}} = O(V), kelpC = IC*
The spinorial norm N(:v) determines a group morphism N : I*C(V) -> ©*. Define
Pin°(V) = {:z: 6 re(V> ; IN(:v)l = 1}.
We let the reader check the following result.

Proposition 12.1.49. There exists a short exact sequence

1 -> S1 -> Pin°(V) -> O(V) -> 1.

Corollary 12.1.50. There exists a natural isomorphism

Pin°(v)
~= (Pin(Vl >< s1)/ ~,
where "~ " is the equivalence relation
(LU, z) iv (-at, -z) we, z) G Pin(V) X 51.
Proof. The inclusions Pin(V) C Cl(V), S1 C (I induce an inclusion
(Pin(V) X SW ~-> co).
The image of this morphism lies obviously in F°(v)n{ INI = l} so that (Pin(V) X 51)/
can be viewed as a subgroup of PinC(V). The sought for isomorphism now follows from
the exact sequence
l -> S1 -> (Pin(V) X 51)/ ~-> O(V) -> l.
We define Spin°(V) as the inverse image of SO(V) via the morphism
pC : PinC(Vl -> O(V).
Arguing as in the above corollary we deduce
Spin°(Vl 2 (Spin(V) X 51)/ n (Spin(V) X 51)/%2.

Exercise 12.1.51. Prove Spine (V) satisfies all the conditions outlined in Subsection 10.1.5.
634 Lectures on the Geometry of Manzfolds

Assume now dim V is even. Then, any any isomorphism m@l(V) r\/
End((;(S(V))
induces a complex unitary representation
Spin°(V) -> Aut (S(V))
called the complex spinorial representation of Spine. It is not irreducible since, once we fix
an orientation on V , the space End lslvll has a natural superstructure and, by definition,
SpinC acts through even automorphism. As in the real case, §(V) splits into a direct sum
of irreducible representations S:t (V).
Fix a complex structure J on V. This complex structure determines two things.

(i) A canonical orientation on V.

(ii) A natural subgroup

u(v,J) = {T 6 SO(V) ; [T, J] = 0} C SO(V).

Denote by U U(V, J) -> SO(V) the inclusion map.
kJ
Proposition 12.1.52. There exists a natural group morphism U (V, J) -> Spin°(Vl such
that the diagram below is commutative.

U(v,J) at > Spin°(V)

1.] PC

SO(V)

Proof. Let w 6 U(V), and consider a path by : [0, 1] -> U(V) connecting 11 to w. Via the
inclusion U(V) <-> SO(V) we may regard by as a path in SO(V). As such, it admits a
unique lift ' : [0, 1] -> Spin(V) such that g) I 11.
Using the double cover S1 -> S1, z »-> z2, we can find a path t »-> 6(t) 6 S1 such that
6(0) =l and 65%) det *y(t).
Define §(w) to be the image of ('?(1), c5(1)) in Spin°(V). We have to check two things.

(i) The map g is well defined.

(ii) The map O' is a smooth group morphism.

To prove (i), we need to show that if 17 [0, 1] -> U(V) is a different path connecting 11
to w, and I \ : [ 0 , 1] - > S 1 is such that A(0) = l and A(t) = det 17(tl2 , then

(@(1),A(1)) I ('r(l)»6(l)) in Spin°(V).

The elements 5/(1) and 5(1) lie in the same fiber of the covering Spin(V) £> SO(V) so
that they differ by an element in her p. Hence

(1)==617(1% 6 :l:ll.

We can identify e as the holonomy of the covering Spin(V) -> SO(V) along the loop
* 17' which goes from II to w along by, and then back to II along 17' (t) = n(1 - t).
*y
 Dirac Operators 635

The map det : U(V) -> S1 induces an isomorphism between the fundamental groups
(see Exercise 6.2.35 of Subsection 6.2.5). Hence, in describing the holonomy e it suffices
to replace the loop *y * 17' C U(V) by any loop 1/(t) such that
det 1/(t) = det('y * it) = AA) 6 s1.
Such a loop will be homotopic to by * 17 1 in U(V), and thus in S O ( V ) as well. Select 1/(t)
of the form
1/(t)e1 = A(t)e1, I/(t)€i € W > 2,
where (Q) is a complex, orthonormal basis of (V, J). Set f = Je With respect to the
real basis (e1, f 1 , e 2 , f 2 , . . . ) the operator 1/(t) (viewed as an element of SO(V)) has the
matrix description
cos 0(t) sin 0(t)
sin 0(t) cos 0(t) 7

]1

where 0 : [0, 1] -> R is a continuous map such that A(t) €i0(t) .

The lift of y(t) to Spin(V) has the form

DU)
9(t)
2
61f1 sin
COS
6(t)
2
I •

We see that the holonomy defined by alt) is nontrivial if and only if the holonomy of the
2
loop t »-> 6(t) in the double cover S1 i> S1 is nontrivial. This means that 6(1) and A(1)
differ by the same element of Z2 as (1) and 17(1) This proves g is well defined. We leave
the reader to check that g is indeed a smooth morphism of groups.

12.1.8 Low dimensional examples

In low dimensions the objects discussed in the previous subsections can be given more
suggestive interpretations. In this subsection we will describe some of these interpretations.

Example 12.1.53 (The case n = 1). The Clifford algebra C11 is isomorphic with the field
of complex numbers C. The Z2-grading is Re (CaBIn C. The group Spin(l) is isomorphic
with Z2 .

Example 12.1.54 (The case n = 2). The Clifford algebra C12 is isomorphic with the alge-
bra of quaternions H. This can be seen by choosing an orthonormal basis {et, e2} in R2 .
The isomorphism is given by
1 »->l, €1 i->'i7 621-) j7 61621-> k7
where i, j, and k are the imaginary units in H. Note that
Spin(2) = {a -|- bk a,b€R, 0,2-1-62 1} 2 So.
The natural map Spin(1) -> SO(2) 2 S1 takes the form et »_> et;
636 Lectures on the Geometry of Manzfolds

Example 12.1.55 (The case n = 3). The Clifford algebra C13 is isomorphic, as an un-
graded algebra, to the direct sum H ® H. More relevant is the isomorphism Cleven
3
r\/

C12 2 H given by
1 P-> 17 6162 * ni, 6263 P-> j, 6361 |-) k7
where {et, € 2 , e3} is an orthonormal basis in R3. Under this identification the operation
as »-> mb coincides with the conjugation in H
QUO a-I-bi-l-cj-I-dkr->T a bi-cj-dk.
In particular, the spinorial norm coincides with the usual norm on H
N ( a + b i + c j + d k ) =a2-l-52-I-C2-l-d2.
Thus, any QUO 6 C13even \{0} is invertible, and
1 1 b
23
N(:z:l
Moreover, a simple computation shows that xR3x-1 C RB, Va: 6 cl3ven \{0}, SO that
r0(n3) 5 H \ {0}.
Hence
Spin(3) E { a3€]H[; I:uI = l } iv
SU(2).
The natural map Spin(3) -> SO(3) is precisely the map described in the Exercise 6.2.8 of
Subsection 6.2.1.
The isomorphism Spin(3) 2 SU(2) can be visualized by writing each q = a + bi +
c j -|- dk as
q = u +j'u, u a+bi,v= C di)€(C.
To a quaternion q = 'LL + iv one associates the 2 X 2 complex matrix

Sq G SU(2).

Note that Sq' = Sq , 'v'q E H. For each quaternion q E H we denote by Lq (respectively Rq)
the left (respectively right) multiplication. The right multiplication by i defines a complex
structure on II-ll. Define T : II-ll -> (C2 by

q='u,+ju»->Tq
Hu
'U

A simple computation shows that T(Riq) = iTq, i.e., T is a complex linear map. More-
over, 'v'q 6 Spin(3) 2 $3, the matrix Sq is in SU(2), and the diagram below is commuta-
tive.
H T ) C 2

LQ Sq

H T)ac2
 Dirac Operators 637

In other words, the representation

Spin(3) 9 q »-> LQ 6 GL€(H)
of Spin(3) is isomorphic with the tautological representation of SU (2) on (C2
On the other hand, the correspondences
6162 »-> s ® s E End l ® End (@al
l(Q2

6263 »-> So ® 61 E End <(c2) ® End (<c2l

63 61 »-> So ® So E End ((C2) ® End (<c2)

616263 »-> R 1162 ® (-Lc2) 6 End (@2) ® End (IC2)

extend to an isomorphism of algebras (213 -> End ((C2) ® End (@2). This proves that the
tautological representation of SU (2) is precisely the complex spinorial representation §3.
From the equalities
[Re)Lq] Z { R j ) R i } = 0,
we deduce that RE defines an isomoiphism of Spin(3)-modules
Le 53 -> §3.
This implies there exists a Spin(3)-invariant bilinear map
@¢s3 x§3->(c.

This plays an important part in the formulation of the recently introduced Seiberg-Witten
equations (see [135]).

Exercise 12.1.S6. The left multiplication by ii introduces a different complex structure on

H. Prove the representation Spin(3) 9 q »-> (Ra : H -> H) is

(i) complex with respect to the above introduced complex structure on H, and
(ii) it is isomorphic with the complex spinorial representation described by the left multi-
plication.

Example 12.1.57 (The case n = 4). The Clifford algebra C14 can be realized as the alge-
bra of 2 X 2 matrices with entries in H. To describe this isomorphism we have to start from
a natural embedding R4 <-> M2(H-H) given by the correspondence
I' 1

]HI""R49ac»->

A simple computation shows that the conditions in the universality property of a Clifford
algebra are satisfied, and this correspondence extends to a bona-fide morphism of algebras
C14 -> M2 (H). We let the reader check this morphism is also injective. A dimension count
concludes it must also be surjective.
638 Lectures on the Geometry of Manzfolds

Proposition 12.1.S8. Spin(4)%SU(2) X SU(2).

Proof. We will use the description of Spin(4) as the universal (double-cover) of SO(4) so
we will explicitly produce a smooth 2 : 1 group morphism SU(2) X SU(2) -> SO(4).
Again we think of SU(2) as the group of unit quaternions. Thus each pair (quo, QQ) 6
SU(2) X SU(2) defines a real linear map
TCII1,CI12 : H -> H, ac »-> T611,92513 = qlac'q2.
-
Clearly l;z:l = IC]1l lac°| . . = ITC11,112£8l Vi 6 H, so that each T611,92 is an orthogonal
transformation of H. Since SU (2) X SU (2) is connected, all the operators TC11 ,QQ belong to
the component of 0(4) containing II, i.e., T defines an (obviously smooth) group morphism
T:S(/(2) X SU(2)->SO(4).
Note that kerT = {1, -l}, so that T is 2 : 1. In order to prove T is a double cover it
suffices to show it is onto. This follows easily by noticing T is a submersion (veru'y this),
so that its range must contain an entire neighborhood of II 6 SOM). Since the range of T
is closed (veru'y t h i s ) we conclude that T must be onto because the closure of the subgroup
(algebraically) generated by an open set in a connected Lie group coincides with the group
itself (see Subsection 1.2.3).

The above result shows that

~= ~=
M) spin(4) U) ® ($l Q @(3) ® 2(3)-
~
Exercise 12.1.59. Using the identification C14 = M2 (H) show that Spin(4) corresponds
to the subgroup
{ d i g (p, Q); p, q G H, IpI Ill 1 } C m2(H1.

Exercise 12.1.60. Let {et, et, et, e4} be an oriented orthonormal basis of R4. Let * denote
the Hodge operator defined by the canonical metric and the above chosen orientation. Note
that

is involutive, *Q

(a) Show that

-
* : A2R4 -> A2R4
]1, so that we can split A2 into the :l:l eigenspaces of
4 2
AQR -A +R4 ® A-2RE
*

As = spans{ Ni n 17? } 7

where
l 1
171i= v§(€1 A€2=l: € 3 / \ € 4 l , vi= §l€1/\€3 dz 64 /\ 6 2 ) ,

1
to?e
/ \ € 4 :|: €2 I\€3l.

(b) Show that the above splitting of A2R4 corresponds to the splitting M) @(3l®(3)
under the natural identification A2n4 2 2(4
 Dirac Operators 639

To obtain an explicit realization of the complex spinorial representations 34:|: we need

to describe a concrete realization of the complexification (CL4. We start from the morphism
of R-algebras
IHI9;z:=u-I-.ivr->Sx= .
This extends by complexification to an isomorphism of (I-algebras
H ®1R (C""]ll2((C).
~
(VerQ'y t h i s ) We now use this isomorphism to achieve the identification.
ENdH(H ® H) ®R (C = MQIHI ®1R (C 2 End@(@2 ® (c2>.
The embedding R4 -> (214 now takes the form
r n

H l * 9;z:»->Tx = eEnd(@2 ® <c2>. (12.1.4)

Note that the chirality operator F = -61626364 is represented by the canonical involution
F »-> M32 e (-]1<c2)~
We deduce the go representations of Spin(4) = SU(2) X SU(2) are given by
§j{ : sU(2) x sU(2) 9 (p, q) »-> p e GL(2; (c),
S4 :SU(2) x SU(2) 9 (p,q) »-> q 6 GL(2;(C).
Exactly as in the case of Spin(3), these representations can be given quaternionic descrip-
tions.
Exercise 12.1.61. The space $4 2 II-ll has a canonical complex structure defined by Re
which defines (following the prescriptions in §10.1.3) an isomorphism
e: cu -> End (A°(c2>.
Identify C2 in the obvious way with A1lC2Aodd(C2, and with Aevenlc2 via the map
61 I-> 1 G A0(c2, 82 l-> 6 1 I \ € 2 .

C(a2)
-
Show that under these identifications we have
To
Va: 6
R4 o N

R id
(H in (12, 1.4).
~
= QUO 7
where Tx is the odd endomorphism of Q2 ® (Q2 defined

Exercise 12.1.62. Let V be a 4-dimensional, oriented Euclidean space. Denote by

q : A'V -> C1(V)
the quantization map, and fix an isomorphism A : lc1(vl -> End (S(V)) of Z2-graded
algebras. Show that for any 17 6 A2+(Vl, the image A o q (al 6 End (S(V)) is an endomor-
phism of the form T ® 0 6 End (S+(V)) ® End is-lvll,

Exercise 12.1.63. Denote by V a 4-dimensional oriented Euclidean space.

(a) Show that the representation SL (X) §4+of Spin(4) descends to a representation of SO(4)
and moreover
§+(V) ®(F: §+(V)§ (A°(v) ® A i m ) ®R (I
as SO(4) representations.
(b) Show that s+(V) 2 S+(Vl as Spin(4) modules.
(c) The above isomorphism defines an element <25 6 S+ (V) ®(F: §+(V). Show that via the
correspondence at (a) the isomorphism (IS spans A0(V).
640 Lectures on the Geometry of Manzfolds

12.1.9 Dirac bundles

In this subsection we discuss a distinguished class of Clifford bundles which is both fre-
quently encountered in applications and it is rich in geometric information. We will touch
only the general aspects. The special characteristics of the most important concrete exam-
ples are studied in some detail in the following section. In the sequel all Clifford bundles
will be assumed to be complex.

Definition 12.1.64. Let E -> M be a Clifford bundle over the oriented Riemann manifold
(M, g). We denote by e : Ql (M) -> End(E) the Clifford multiplication.
(a) A Dirac structure on E is a pair (h, V) consisting of a Hermitian metric h on E ,
and a Clqord connection, i.e., a connection V on E compatible with h and satisfying the
following conditions.

(al) For any a 6 Q1 ( M ) , the Clifford multiplication by a is a skew-Hermitian endomor-

phism of E.
(a2) For any a 6 Qllml, X 6 Vect (M), u 6 C°°(E)
Vi(c(a)u) = c(Vxma)u + e(a)(VXu),
where VM denotes the Levi-Civita connection on T*M. (This condition means that
the Clifford multiplication is covariant constant.)

A pair (Clifford bundle, Dirac structure) will be called a Dirac bundle.

(b) A Z2 -grading on a Dirac bundle (E, h, V) is a Z2 grading of the underlying Clifford
structure E = E0 ® El, such that h = h0 ® h1, and V = V0 ® V1, where it (respectively
Vfi) is a metric (respectively a metric connection) on E, .

The next result addresses the fundamental consistency question: do there exist Dirac bun-
dles?

Proposition 12.1.6S. Let E -> M be a ClQj'ord bundle over the oriented Riemann manifold
( M , go. Then there exist Dirac structures on E.
Proof. Denote by QD the (possible empty) family of Dirac structure on E. Note that if
( I W ) 6 D J = l , 2, and f 6 C°°(M), then
(fm + (1 - fm, fv1 + (1 - f)v2) 6 D.
This elementary fact shows that the existence of Dirac structures is essentially a local issue:
local Dirac structures can be patched-up via partitions of unity.
Thus, it suffices to consider only the case when M is an open subset of R", and E
is a trivial vector bundle. On the other hand, we cannot assume that the metric g is also
trivial (Euclidean) since the local obstructions given by the Riemann curvature cannot be
removed. We will distinguish two cases.
A. n = dim M is even. The proof will be completed in three steps.
Step 1. A special example. Consider the (complex) spinoza module (representation)
e : (Hn -> End (Sn).
 Dirac Operators 641

We can assume that e is self-adjoint, i.e.,

c(uT) = c('u,)*, Vu 6 (CiTL .

We will continue to denote by e the restriction of the Clifford multiplication to Spin(n) '-->
(can.
Fix a global, oriented, orthonormal frame (60 of TM, and denote by ( e ) its dual
coframe. Denote by w = (WW ) the connection 1-form of the Levi-Civita connection on
T*M with respect to this moving frame, i.e.,

V67 we" Win®€'L, w e Qllml (X) @(11)~

ii

Using the canonical isomoiphism ,0* spin(n) -> M) we define

1
(IJ : p* 1((»J) 2 w e - e j €Q1(M) (X) spin(n).
i<j

This defines a connection VS on the trivial vector bundle aM = Sn X M given by

1
VSu, du - 2 win ® e ( e ) e(ej)u, Vu 6 CYOO(SMl.
i<j

The considerations in §l0. 1.5 show that VS is indeed a Clifford connection so that AM is a
Dirac bundle.
Step 2. Constructing general Dirac bundles. Fix a Dirac bundle (E, he, VE I over M. For
any Hermitian vector bundle (W, hw) equipped with a Hermitian connection VW we can
construct the tensor product E (X) W equipped with the metric he (X) hw and the product
connection VE®W. These two data define a Dirac structure on E (X) W (Exercise 12.1.68
at the end of this subsection).
The representation theory of the Clifford algebra (Cin with n even shows that any Clif-
ford bundle over M must be a twisting So (X) W of the spinoza bundle So. This completes
the proof of the proposition when dim M is even.

Step 3. Conclusion. The odd case is dealt with similarly using the different representation
theory of the <Cl2k+1. Now instead of one generating model S there are two, but the proof
is conceptually identical. The straightforward details are left to the reader.

Denote by (E, h, V) a Dirac bundle over the oriented Riemann manifold (M, gl- There
exists a Dirac operator on E canonically associated to this structure given by

D Co V 0°°<E> 1 C
c°°(T*m (X) E) -> C°°(E).

A Dirac operator associated to a Dirac structure is said to be a geometric Dirac operator.

Proposition 12.1.66. Any geometric Dirac operator is formally self-adjoint.

642 Lectures on the Geometry of Manzfolds

Proof. The assertion in the above proposition is local so we can work with local orthonor-
mal moving frames. Fix ac() 6 M, and denote by (we a collection of normal coordinates
near xo. Set € Ami . Denote by ( e ) the dual coframe of ( e ) . If (h, V) is a Dirac struc-
ture on the Clifford bundle E, then at kg the associated Dirac operator can be described
as
D Z am (Vi = Veil°
We deduce
D* = (V)*e(ei)*.
Since the connection V is compatible with h and d'ivg(ez-) I G 9 0 2 0, we deduce
WU* -Vi- i

Since the Clifford multiplication is skew-Hermitian, e(e')* = - e ( e i ) , we deduce that at

kg we have
D*
Et, O 0(€i) XIV-¢(@')] -I-D .

On the other hand, the Clifford multiplication is covariant constant, and ( V M ) l2202 Oso
we conclude
m , c ( e ) ] l 0 = 0 w-
Let D be a geometric operator associated to the Dirac bundle (E, h, V). By definition,
D2 is a generalized Laplacian, and according to Proposition 10.1.34 we must have an
equality of the form
D2 = @*€7 + SR,
where V is a connection on E, and LR is the WeitzenbOck remainder which is an endomor-
phism of E. For geometric Dirac operators, V = V, and this remainder can be given a very
explicit description with remarkable geometric consequences. To formulate it we need a
little bit of foundational work.
Let (E, h, V) be a Dirac bundle over the oriented Riemann manifold (M, g). Denote
by C1(T*MM) -> M the bundle of Clifford algebras generated by ( T * M , g).
The curvature F(V) of V is a section of A2T*M (X) End (E). Using the quantization
map q : A'T*M -> C1(T*M) we get a section
q(F) 6 C1(T*M) (X) End (E)-
On the other hand, the Clifford multiplication C : C1(T*M) -> End (E) defines a linear
map
c1<T*m> (X) End (E) -> End(E), w (X) T »-> c(w) o T.
This map associates to the element q (F) an endomorphism of E which we denote by e(F).
If (et is a local, oriented, orthonormal moving frame for T * M , then we can write
Z' .
FM € /\ €3 (X) Fig 7
z'<j
and
. . l . .
e(F) Z 2 C(e")C(e)Fj = 52 c(e*)<=(@")F
 Dirac Operators 643

Theorem 12.1.67 (Boehner-Weitzenbiick). Let D be the geometric Dirac operator as-

sociated to the Dirac bundle ( E , h , V ) over the oriented Riemann manifold ( M , go. Then

D2 = v*v -|- <2(F(V))~

Proof. Fix L U Q M , and then choose an oriented, local orthonormal moving frame ( I of
TM near an such that
M
[@¢»@j]lx-(V €'illa:-0:

where VM denotes the Levi-Civita connection. Such a choice is always possible because
the torsion of the Levi-Civita connection is zero. Finally, denote by (60l the dual coframe
of (60. Then

l
D2 laz= Ze(e'L) Vi Zc(e.7)V]

.
S111CC lVclejl] If 0 we deduce

$2 lx = Z
'As
C(€i)C(€3>V2VV
XV? + Z #J
e(e)c(ej)V,-Vj
i

XV? -I- X c ( @ > ) c ( ) m v 1 Z v? + Z ¢<ei>c<a>Fj(v>.

II
I

i<j ii i<j

We want to emphasize again the above equalities hold only at 113. The theorem now follows
by observing that

(WV) lm
) x

Exercise 12.1.68. Let (E, h, V) be a Dirac bundle over the oriented Riemann manifold
(M, g) with associated Dirac operator D. Consider a Hermitian bundle W -> M and a
connection VW compatible with the Hermitian metric he .
(a) Show that ( E ® W , h ® h w , V = V (X) 1 W -|- 1 E (X) VW) defines a Dirac structure on
E (X) W in which the Clifford multiplication by a E Q1 (M) is defined by

c(0¢)(€ (X) w)=(c(o0€) (X) We e6 c°<><E>, w 6 c°°(w).

We denote by DW the corresponding geometric Dirac operator.
(b) Denote by C w ( F ( V W ) ) the endomorphism of E (X) W defined by the sequence
C
F(vW)ec°°(A2T*m®End(w>>»i>c°°(c1(T*m)®End(w)) »-> C°°(End(E (X) w)).

Show that

02W v + <=(F(v)) + cw(F(VW))-

644 Lectures on the Geometry of Manzfolds

12.2 Fundamental examples

This section is entirely devoted to the presentation of some fundamental examples of Dirac
operators. More specifically we will discuss the Hodge-DeRham operator, the Dolbeault
operator the spin and spin" Dirac. We will provide more concrete descriptions of the
WeitzenbOck remainder presented in Subsection 10.1.9 and show some of its uses in estab-
lishing vanishing theorems.

12.2.1 The Hodge-DeRham operator

Let ( M , g ) be an oriented Riemann manifold and set

A. T * M
C A'T*M (X) C
For simplicity we continue to denote by Q' (M) the space of smooth differential forms on
M with complex coefficients. We have already seen that the Hodge-DeRham operator
d+d* Q. ( M ) -> Q. (M)

is a Dirac operator. In fact, we will prove this operator is a geometric Dirac operator.
Continue to denote by g the Hermitian metric induced by the metric g on the com-
plexification AcT*M. We will denote by VI all the Levi-Civita connection on the tensor
bundles of M. When we want to be more specific about which Levi-Civita connection we
are using at a given moment we will indicate the bundle it acts on as a superscript. E.g.,
VT*M is the Levi-Civita connection on T * M .

Proposition 12.2.1. The pair (g, V 9 ) defines a Dirac structure on the Clylord bundle
A@T*M and d + d* i5 the associated Dirac operator

Proof. In Subsection 4.1.5 we have proved that d can be alternatively described as the
composition
V 5
c°°lA°T*m> -> c°°lT*m (X) A°T*m1 -> c°°<A°T*m),
where 5 denotes the exterior multiplication map. Thus
d-I-d*=5o V -|- V* o e *
If X . . . Xn is a local orthonormal frame of TM, and 91
7 7 7
917,
is its dual coframe,
then for any ordered multi-index I we have
6* ¢ X 91.
(9')
J
Thus, for any Q 1 6 cOO ( A ' T * M ) we have

<v* o €*)w Z V x k i x k +d'l;Q]g(xkl'[;xk w.

k

Fortunately, we have the freedom to choose the frame (Xkl in any manner we find conve-
rent.
 Dirac Operators 645

Fix an arbitrary point 5130 6 M and choose (Xkl such that, at 5130, we have Xi =
( k , where (wk) denotes a collection of normal coordinates near 5130. With such a choice
d i v ( X k , ) = 0 at 330, and thus

d*w l0002 - i8kV6'kw.

k

This shows d + d * can be written as the composition

V
C°°(A'T*M) -> c°°(T*m (X) A'T*m) i> c°°(A'T*m),
where C denotes the usual Clifford multiplication on an exterior algebra. We leave the
reader to verify that the Levi-Civita connection on A'T*M is indeed a Clifford connec-
tion, i.e., the Clifford multiplication is covariant constant. Passing to the complexification
AcT*M we deduce that d -|- d* is a geometric Dirac operator.

We want to spend some time elucidating the structure of the WeitzenbOck remainder.
First of all, we need a better description of the curvature of VI viewed as a connection on
A'T*M.
Denote by R the Riemann curvature tensor, i.e., the curvature of the Levi-Civita con-
nection
V9:C'°°(TM)->C'°°(T*M (X) TM).
Thus R is a bundle morphism
R : o°°(Tm)->c°°(A2T*m (X) TM).
We have a dual morphism

R : c°°<T*m) -> 0°°(A2T*m (X) To),

uniquely determined by the equality

(R(}/, Z)<1)(X) a(R(Y, Z)X) Va 6 MM) X, Y, Z 6 Vect (Ml.

Lemma 12.2.2. R i5 the curvature of the Levi-Civita connection VT*M.

Proof. The Levi-Civita connection on T*M is determined by the equalities

(vTZ*M@)(X) = Z a ( X ) - QWTZMX) V a 6 Q1(m),
• x, Z 6 Vect (m).
Derivating along Y 6 Vect (M) we get

(VyVz0¢)(X) = Y • (Vz0¢)(X) - (VzOOWyXI

-
= Y Z a(X) - Y • OAIVzX) - Z Q W y X ) - @(VZVYX).
a -
Similar computations give VzVyQ and VIa Z] and we get

(RE*M<y, 2)<»)(x) = - a ( R T M ( Y , zlxl.

646 Lectures on the Geometry of Manzfolds

The Levi-Civita connection V9 = VT*M extends as an even derivation to a connection

on A'T*M. More precisely, for every X 6 Vect (M), and any a l , . . . , as, 6 Q1IMI, we
define

VQxIOA1 /\ /\ as) ; (VgXa1) /\ /\Ock+ +011 /\ /\ (Vgx0¢k)~

Lemma 12.2.3. The curvature of the Levi-Civita connection on A'T*M is defined by

RA°T*M ( x , Y)(@1 /\ /\ Odell

= (R(x,y)@1) /\°°°/\Ozk + ° ' ° + & 1 /\ /\ ( R ( x , y ) @ ) ,

for any vector fields X , Y , and any I -forms a l , . . . , as.

Exercise 12.2.4. Prove the above lemma.

The WeitzenbOck remainder of the Dirac operator d + d* is e(RA°T*M). To better

understand its action we need to pick a local, oriented, orthonormal moving frame (60 of
TM. We denote by (60l its dual coframe. The Riemann curvature tensor can be expressed
as
Q:

et
II

/\ € j R z j 7

i<j

where Raj is the skew-symmetric endomorphism R = R(€¢»€jl : TM -> TM. Thus

claRA°T*Ml e .T *A /Il e
e-I = 5l Z c(e")c(e3lRA
. .T * M (Ami), u

l€ '£ lc l€ jlR A
Ȣ<j 'As

where e(ej) = e(ej) - i(ejl.

Note that since both V*V, and (d + d*)2 preserve the Z-grading of AcT*M, so does
the Weitzenbéck remainder, and consequently it must split as

c(RA T*Ml Z €Bk>_05€k

Since 0 E 0, the first interesting case is R1. To understand its form, pick normal coordi-
=

nates ( w near :cg 6 M. Set et = Ami I H 2 0 € Tm0 M and e'i do' l,,,O€ Thor.
At :cg the Riemann curvature tensor R has the form
m

ek /\ oR(ek, et)
II

k<£

where R(ek, k g ) € j Ryke€i = Rijkwi- Using Lemma 12.2.2 we get

Rl€k, €£)€] I RijkW

Using this in the expression of CR1 at 51:0 we get

CR1 a. Old R i s k £ e ( e I ¢ ) e ( e £ ) € '

j k,e M
 Dirac Operators 647

We need to evaluate the Clifford actions in the

g above equality. We have
.
*= \
e (€£)€ e / €? - 6
and
c ( e k ) c ( e ) e = ek /\ € /\ e'i - 5i£€k - cikge + 5ik€£.
Hence
1 e e l.
921 Q. 0¢jRij1Qe (6 k /\ e
.
/\ e'L - 5 i £ €
k
- 5kaI£€'L
.
-I- l i k e
2
j i,j,k,€
Using the first Bianchi identity we deduce that

2; R '1ljk€€
k
As As'Z = - RjzkW A e k As 0, w.
i,1<,e i,k,€

Because of the skew-symmetry Rilke i -Rijik we conclude that

i
CyjR¢jke5ke€ = 0.
ike
Hence
I

1 l Z l e l k
O¢jR¢jA¢e(5¢k€ (5i€€k) ()djR¢jz'g€ ()djR¢jkz'€
2 2 2
i,j,k,z 1l,j,Z '1l,j,k

OdeR¢j¢k€
k - ek
as Rlcjk 7

1l,j,k je
where Ricjk Z Rijik denotes the Ricci tensor at xo. Hence
:RI = Ric. (12.21)
In the above equality, the Ricci tensor Ric is regarded (via the metric duality) as a self-
adjoint endomorphism of T * M .
The identity (l2.2.l) has a beautiful consequence.

Theorem 12.2.5 (Boehner). Let ( M , g) be a compact, connected, oriented Riemann man-

ifold.
(a) [fthe Ricci tensor i5 non-negative definite, then 61 ( M l _< dim M .
(19) If Ricci tensor is non-negative definite, but it i5 somewhere strictly positive definite then

(Recall that 61 ( M l denotes the first Betti number of M . )

The above result is truly remarkable. The condition on the Ricci tensor is purely local,
but it has global consequences. We have proved a similar result using geodesics (see Myers
Theorem, Subsections 5.2.2, 6.2.4 and 6.2.5). Under more restrictive assumptions on the
Ricci tensor (uniformly positive definite) one deduces a stronger conclusion namely that the
fundamental group is finite. If the uniformity assumption is dropped then the conclusion of
the Myers theorem no longer holds (think of the flat torus).
This result gives yet another explanation for the equality H1 (G) = 0, where G
is a compact semisimple Lie group. Recall that in this case the Ricci curvature is
1
X {the Killing metric}.
4
648 Lectures on the Geometry of Manzfolds

Proof. (a) Let A1 denote the metric Laplacian

A1 = do* -|- dw : Q1(Ml -> Q1(M).
Hodge theory asserts that bi (M) = dim her A1 SO that, to find the first Betti number, we
need to estimate the "number" of solutions of the elliptic equation

A1'I']=0, D 691

Using the Boehner-Weitzenbéck theorem and the equality (l2.2.l) we deduce that if 17 E
her A1, then

V*Vn + Ric 17 = 0, onM.

Taking the L2-inner product by 17, and then integrating by parts, we get

lvnl2dvg + (Ric 17,v)dug = 0. (12.2.2)

M M

Since Ric is non-negative definite we deduce V77 = 0, so that any harmonic 1-form must be
covariant constant. In particular, since M is connected, the number of linearly independent
harmonic 1-forms is no greater than the rank of T*M which is dim M.
(b) Using the equality (l2.2.2) we deduce that any harmonic 1-form 17 must satisfy

(Ric(:z:)17,,,, Urvlrv 0 Vcc 6 M.

If the Ricci tensor is positive at some £170 6 M, then 77(£U0) 0. Since 17 is also covariant
constant, and M is connected, we conclude that 17 E 0.

Remark 12.2.6. For a very nice survey of some beautiful applications of this technique we
refer to [13].

12.2.2 The Hodge-Dolbeault operator

This subsection introduces the reader to the Dolbeault operator which plays a central role in
complex geometry. Since we had almost no contact with this beautiful branch of geometry
we will present only those aspects concerning the "Dirac nature" of these operators. To
define this operator we need a little more differential geometric background.

Definition 12.2.7. (a) Let E -> M be a smooth real vector bundle over the smooth man-
ifold M . An almost complex structure on E is an endomorphism J : E -> E such that
JO i -HE.
(b) An almost complex structure on a smooth manifold M is an almost complex structure
J on the tangent bundle. An almost complex manifold is a pair (manifold, almost complex
structure).

Note that any almost complex manifold is necessarily even dimensional and orientable,
so from the start we know that not any manifold admits almost complex structures. In
 Dirac Operators 649

fact, the existence of such a structure is determined by topological invariants finer than the
dimension and orientability.
Example 12.2.8. (a) A complex manifold M is almost complex. Indeed, the manifold
M is locally modelled by C", and the transition maps are holomorphic maps C" -> C" .
~
The multiplication by i defines a real endomorphism on RQ" = KI", and the differential
of a holomorphic map C" -> C" commutes with the endomorphism of TQ" given by
multiplication by ii in the fibers. Hence, this endomorphism induces the almost complex
structure on TM.
(b) For any manifold M, the total space of its tangent bundle TM can be equipped in many
different ways with an almost complex structure.
To see this, denote by E the tangent bundle of T M , E = T ( T M ) . The bundle E has
a natural subbundle V, the vertical subbundle, which is the kernel of the differential of the
canonical projection 'IT : T M -> M .
If 11: 6 M , 'U 6 Ta,M, then a tangent vector in Tl,v/3,)TM is vertical if and only if it
is tangent to a smooth path which entirely the fiber Tx3 M C TM. Note that we have a
canonical identification
~-
*'(v,x) TM.
Fix a Riemann metric h on the manifold TM, and denote by Uf C E, the subbundle of E
which is the orthogonal complement of V in E. For every (u, iv) 6 TM, the differential of
'IT' induces an isomorphism

7T* . UMm) TxM r Vv,:1:.

Thus, Tr* induces a bundle isomorphism A : IH -> V. Using the direct sum decomposition
E i Fu ® V we define the endomorphism J of E in the block form
r f\ A 1 1

Exercise 12.2.9. Prove that a smooth manifold M of dimension 2n admits an almost com-
plex structure if an only if there exists a 2-form w 6 Q2lml such that the top exterior
power as" is a volume form on M .

Let (M, J) be an almost complex manifold. Using the results of Subsection 2.2.5 we
deduce that the complexified tangent bundle TM (X) (C splits as
TM (X) (C = (Tm)1»0 (X) (Tm)0»1.
The complex bundle T M 0 is isomorphic (over (C) with ( T M , J).
By duality, the operator J induces an almost complex structure in the cotangent bundle
T*m, and we get a similar decomposition
T*M (X) C (T*M)1,0 ® lT*Ml0,1
Z

In turn, this defines a decomposition

A@T*M Ap»<1T*m.
p,q
We set QP,q(M) ;= C°°(A1'=qT*M).
650 Lectures on the Geometry of Manu'olds

Example 12.2.10. Let M be a complex manifold. If (zj = x1+'iy1) are local holomorphic
coordinates on M, then (TM)1'0 is generated (locally) by the complex tangent vectors
1
8zj
2 w fray 7

while (T*M)1)0 is locally generated by the complex l-forms d z j = d x j + 'iyo .

The bundle (TM)1,O is also known as the holomorphic tangent space, while lT*M)1,0
is called the holomorphic cotangent space. The space (T*M)0'1 is called the anti-
holomorphic cotangent space, and it is locally spanned by the complex 1-forms did =
dace - iyo .
Note that any (p, Q)-form 17 6 QP)'1(M) can be locally described as

17: ¢7AB¢ZZAAd2B, '7A,B 6 COO(M,(C),

III=p,IBI=q
where we use capital Latin letters A , B , C . . to denote ordered multi-indices, and for each
such index A we set
dzA dzAl /\ - - - /\ dzAp .
The differential form deB is defined similarly.

Exercise 12.2.11. Let ( M J ) be an arbitrary smooth almost complex manifold.

(a) Prove that

dQp»<1(m) C Qp+2»'1-1(m) ® Q p + 1 » q ( m ) ® Qp»q+1(m)Qp-1»<1+2(m).

(b) Show that if M is a complex manifold then

dQ2°»</(m) C Qp+1»q(m) ® Qp»'1+1(m>. (12.2.3)

(The converse is also true, and it is known as the Newlander-Nirenberg theorem. Its proof
is far from trivial. For details we refer to the original paper [103].)

Definition 12.2.12. An almost complex structure on a smooth manifold is called integrable

if it can derived from a holomorphic atlas, i.e., an atlas in which the transition maps are
holomorphic.

Thus, the Newlander-Nirenberg theorem mentioned above states that the condition
(12.23) is necessary and sufficient for an almost complex structure to be integrable.

Exercise 12.2.13. Assuming the Newlander-Nirenberg theorem prove that an almost com-
plex structure J on the smooth manifold M is integrable if and only if the Nzjenhuis tensor
N 6 C°°(T*M®2 (X) T M ) defined by
n ( x , Y ) = [JX , JY] [X, Y] J[X, JY] J[JX,Y] V X , Y € V e c t ( M )
vanishes identically.
 Dirac Operators 651

In the sequel we will exclusively consider only complex manifolds. Let M be such a
manifold. The above exercise shows that the exterior derivative
d : Quo;(m> -> Q§§+1(m>
splits as a direct sum d = &Bp+q=kdp'q, where
dp»'1 Z {d : Qp»'1(m) -> Qp+1,qlml e Qp»q+1(M)},
The component QW; -> Qp+1,q is denoted by 8 = 31°'q, while the component Qzmq ->
Qp,q+1 is denoted by 5 = Gina.

Example 12.2.14. In local holomorphic coordinates ( ) the action of the operator 5 is

described by

5 n A B d z A /\ doB 3*7AB
. d
A
dzj deB
l-1l|A|
A,B
622 z
j,A,B

It is not difficult to see that

§p,q+1 O §p,q 0 Vp, q.
In other words, for any 0 S p S dim() M, the sequence
5
0 ->Q12,0(M) 5>Q19,1(M) >
is a cochain complex known as the p-th Dolbeault complex of the complex manifold M.
Its cosmology groups are denoted by H§,q1(M)_

Lemma 12.2.15. The Dolbeault complex i5 an elliptic complex.

Proof. The symbol of (§p,q is very similar to the symbol of the exterior derivative. For any
x € M and any § e T * X

0(5'»q)(§) : Ap»<1T;m -> Ap»<1+1Tx M

is (up to a multiplicative constant) the (left) exterior multiplication by §0'1, where €0,1
denotes the (0, 1) component of ,g viewed as an element of the complexified tangent space.
More precisely,
£0,1 1
- §(€ -|- mos),
where JO : Taz M -> Tas M denotes the canonical complex structure induced on TazM by
the holomorphic charts. The sequence of symbols is the cochin complex
l®(-1)P>§0'1A 1®(-1)P>§0'1 A
0 ->Ap»0 (X) A0,0 AND (X) AND (X) A0,l

This complex is the (Z-graded) tensor product of the trivial complex

1
0 -> AND -> Ap»0 -> 0,
with the Koszul complex
(-1)P§>'1/\ A0,1 (-1)P§>'1/\
0 -> A0,0
652 Lectures on the Geometry of Manzfolds

Since §0'1 as 0, for any 5 go 0, the Koszul complex is exact (see Subsection 7.1.3). This
proves that the Dolbeault complex is elliptic.

To study the Dirac nature of this complex we need to introduce a Hermitian metric h on
TM. Its real part is a Riemann metric g on M, and the canonical almost complex structure
on TM is a skew-symmetric endomorphism with respect to this real metric. The associated
2-form Qh = -lm h is nondegenerate in the sense that Q" (n = dime; M) is a volume
form on M. According to the results of Subsection 2.2.5 the orientation of M defined by
Q" coincides with the orientation induced by the complex structure.
We form the Hodge-Dolbeault operator

5) + 5* : Qp»'(m) -> Q2°»'(M).

Proposition 12.2.16. The Hodge-Dolbeault operator \/§(§ + 5*> is a Dirac operator.

Proof. We need to show that
2 1
-5151211, */seT *m.
(u(@')(<) - o(@')(s)* )
Denote by J the canonical complex structure on T*M and set n = Jg. Note that 5 J_ 17
and l§l = 1171. Then
_ 1 . l .
o(0)(€) i (-1)P§€(€ + 'Ml §(e@) + w(f7)),

where as usual e(~) denotes the (left) exterior multiplication. The adjoint of 0I(3-)(§) is
1
(-1)p(t(6))(€)* 2
- If* J -'£17*J 7

where §* (respectively r7*) denotes the metric dual of 5 (respectively 17), and §*J denotes
the contraction by §*. We set

c(€l 6(£l-fi* J v ?:(t1) 6(t1) -|- u* J 7

and we deduce
2 l g* J -I-'i77*J }2
(0(@'>(,g) 0t(5)(§)* ) 4 {€(§) -|- i€(t7)
l 2 l o - 2
1 {( 6(5)- fi* J ) + 'i(€(17)+t1* J ) } = ;{ c(§) + we) }
l 2
Z {€(§) 6(v)2 + i( c(€)6(t1) + c(f7)c(€) l} -
Note that

c(@2=-(e(§)§* J +e* J 6(0) =- l€ l 2 7

and
2
W2=WW J +17* J 6(0) lnl
 Dirac Operators 653

On the other hand, since 5 J_ 17, we deduce as above that

Cl lélul'+'&lwlcll'='0~
(Verify this!) Hence
2 1 1
{Ct(@)(e) - U(5)(€)* } -7(lsl2 W) = - 5 M 2
A natural question arises as to when the above operator is a geometric Dirac operator.
Note first that the Clifford multiplication is certainly skew-adjoint since it is the symbol of
a formally self-adjoint operator. Thus all we need to inquire is when the Clifford multipli-
cation is covariant constant. Since
_ 1 10 .
0(5) ( ) (e(§)+@e(v) g* J in* J 7

where 17 : Jo, we deduce the Clifford multiplication is covariant constant if VI J 0.

Definition 12.2.17. Let M be a complex manifold, and h a Hermitian metric on TM

(viewed as a complex bundle). Then h is said to be a Kohler metric if VJ = 0, where
V denotes the Levi-Civita connection associated to the Riemann metric Re h, and J is the
canonical almost complex structure on TM. A pair (complex manifold, Kahler metric) is
called a Kohler manifold.

Exercise 12.2.18. Let (M, J) be an almost complex manifold and h a Hermitian metric on
TM. Let Q = -lm h. Using the Exercise 12.2.13 show that if dpI, = 0, then the almost
complex is integrable, and the metric h is Kohler.
Conversely, assuming that J is integrable, and h is Kohler show that do = 0.

We see that on a Kohler manifold the above Clifford multiplication is covariant constant.
In fact, a more precise statement is true.
Proposition 12.2.19. Let M be a complex manifold and h a Kohler metric on T M . Then
the Levi-Civita induced connection on A0"T*M is a Clifford connection with respect to
the above Clifford multiplication, and moreover; the Hodge-Dolbeault operator I§(5+5* I
i5 the geometric Dirae operator associated to this connection.

Exercise 12.2.20. Prove the above proposition.

Example 12.2.21. Let ( M , g l be an oriented 2-dimensional Riemann manifold (surface).

The Hodge *-operator defines an endomorphism
*:TM->TM,
satisfying *2 : -ATM, i.e., the operator * is an almost complex structure on M. Us-
ing the Exercise 12.2.18 we deduce this almost complex structure is integrable since, by
dimensionality
ds = 0,
where Q is the natural 2-form Q(X, Y) = g(*X, Y) X, Y 6 Vect ( M ) . This complex
structure is said to be canonically associated to the metric.
654 Lectures on the Geometry of Manu'olds

Example 12.2.22. Perhaps the favorite example of Kohler manifold is the complex projec-
tive space C". To describe this structure consider the tautological line bundle L1 -> CIP".
It can be naturally viewed as a subbundle of the trivial bundle Q""-1 -> CIP".
Denote by 110 the canonical Hermitian metric on Q - 1 , and by V0 the trivial connec-
tion. If we denote by P : Q""-1 -> L1 the orthogonal projection, then V = P o V0 | L1
defines a connection on L1 compatible with hi = h | L1. Denote by Ld the first Chern form
associated to this connection
ii
CU - F ( V )»
271'

and set
hFS(X, alx == -Wx(X7 J Y ) + i w ( X , y) Vi 6 m, x, Y 6 TM.
Then h p is a Hermitian metric on KHP" (verify) called the Fubini-Study metric. It is
clearly a Kohler metric since (see the xercise 12.2.l8) do/'L = -do : -de1(V) = 0.

Exercise 12.2.23. Describe lIeS in projective coordinates, and then prove that h is indeed
a Hermitian metric, i.e., it is positive definite.

Remark 12.2.24. Any complex submanifold of a Kohler manifold is obviously Kohler. In

particular, any complex submanifold of CCIP" is automatically Kohler.
A celebrated result of Chow states that any complex submanifold of CIP" is automati-
cally algebraic, i.e., it can be defined as the zero set of a family of homogeneous polyno-
mials. Thus, all complex nonsingular algebraic varieties admit a natural Kohler structure.
It is thus natural to ask whether there exist Kohler manifolds which are not algebraic.
The answer is positive, and a very thorough resolution of this problem is contained in
the famous Kodaira embedding theorem which provides a simple necessary and sufficient
condition for a compact complex manifold to be algebraic.
For this work, Kodaira was awarded the Fields medal in 1954. His proofs rely essen-
tially on some vanishing results deduced from the WeitzenbOck formulae for the Dolbeault
operator 65 -|- 5*, and its twisted versions. A very clear presentation of this subject can be
found in the beautiful monograph [55].

12.2.3 The spin Dirac operator

Like the Dolbeault operator, the spin Dirac operator exists only on manifolds with a bit of
extra structure. We will first describe this new structure.
Let IM", g) be an in,-dimensional, oriented Riemann manifold. In other words, the
tangent bundle TM admits an SO(n) structure so that it can be defined by an open cover
(Ua), and transition maps
gal? : U05 -> SO(n)
satisfying the cocycle condition.
The manifold is said to spinnable if there exist smooth maps
§o:[-3 : U05 -> Spin(n)
 Dirac Operators 655

satisfying the cocycle condition, and such that

0(§0¢-2) I QQB Va, 5,

where p : Spin(n) -> SO(n) denotes the canonical double cover. The collection Qo i-2 as
above is called a spin structure. A pair (manifold, spin structure) is called a spin manifold.
Not all manifolds are spinnable. To understand what can go wrong, let us start with
a trivializing cover 'Ll = (UOf) for TM, with transition maps 901% and such that all the
multiple intersection USB~~~v are contractible. In other words, U is a good cover.
Since each of the overlaps U . , is contractible, each map gal? : UQ;-3 -> SO(n) admits
at least one lift
Qop : U06 -> Spin(n).
From the equality Pl§aB§Bv§v<1l = gaBgB'7g'ya = ]1 we deduce

6057 = §a,B§B'7§'ya 6 kelp 3 ZN.

Thus, any lift of the gluing data QQB to Spin(n) produces a degree 2 tech cochain of the
trivial sheaf Zn, namely the 2-cochain

(6_l UQBw I-> €057.

Note that for any (1,/3,%6 such that Ua,B76 # (Z), we have
65 - 6076 + 6aB6 - 6067 0€%2

In other words, e, defines a tech 2-cocycle, and thus defines an element in the tech coho-
mology group I-I2(M, Z2).
It is not difficult to see that this element is independent of the various choices: the cover
'LL the gluing data .gaB» and the lifts Qog. This element is intrinsic to the tangent bundle
TM. It is called the second Stiefel-Whitney class of M, and it is denoted by W2 (Ml. We
see that if w2(M) quo 0 then M cannot admit a spin structure. In fact, the converse is also
true.

Proposition 12.2.25. An oriented Riemann manifold M admits a spin structure if and only

Exercise 12.2.26. Prove the above result.

Remark 12.2.27. The usefulness of the above proposition depends strongly on the ability
of computing W2. This is a good news/bad news situation. The good news is that algebraic
topology has produced very efficient tools for doing this. The bad news is that we will not
mention them since it would lead us far astray. See [83] and [98] for more details.

Remark 12.2.28. The definition of isomorphism of spin structures is rather subtle (see
[95]). More precisely, two spin structures defined by the cocycles in and h., are iso-
morphic if there exists a collection sa 6 Z2 C Spin(n) such that the diagram below is
656 Lectures on the Geometry of Manzfolds

commutative for all ZUGUQQ

So
Spin(n) \
J
Spin(n)
I
\-
l §5a(@¢)

60
11561

Spin(n) > Spin(n)

The group H1(M, Z2) acts on Spin(M) as follows. Take an element 5 E H1lM,Z2l
represented by a tech cocycle, i.e., a collection of continuous maps sag : Uog Z2 C
Spin(n) satisfying the cocycle condition

6025 ' 551 ' €°v0¢ 1.

Then the collection 5.. §»» is a Spin(n) gluing cocycle defining a spin structure that we
-
denote by 5 U.
It is easy to check that the isomorphism class of 5 O' is independent of the various
choice, i.e., the tech representatives for 5 and O'. Clearly, the correspondence
H1(M,Z2) X Spin(M) 9 (e,0) »-> 5 0 6 Spin(M)
defines a left action of H1 ( M , Z2) on Spin(M). his action is transitive and free.

Exercise 12.2.29. Prove the statements in the above remark.

Exercise 12.2.30. Describe the only two spin structures on S1

Example 12.2.31. (a) A simply connected Riemann manifold M of dimension > 5 is

spinnable if and only if every compact orientable surface embedded in M has trivial normal
bundle.
(b) A simply connected four-manifold M is spinnable if and only if the normal bundle No:
of any embedded compact, orientable surface Z] has even Euler class, i.e.,

€(N2l
2
is an even integer.
(c) Any compact oriented surface is spinnable. Any sphere S" admits a unique spin struc-
ture. The product of two spinnable manifolds is canonically a spinnable manifold.
(d) U12 (mn) = 0 if and only if n E 3 (mod 4), while CIP" admits spin structures if and
only if n is odd.

Let ( M n , g) be a spin manifold. Assume the tangent bundle TM is defined by the open
cover (Ua), and transition maps
QQB : U05 -> SO(n).
Moreover assume the spin structure is given by the lifts

505 : U06 -> Spin(n).

 Dirac Operators 657

As usual, we regard the collection .90 B as defining the principal SO(n)-bundle of oriented
frames of TM . We call this bundle Pso( M) .
The collection toa defines a principal Spin(n)-bundle which we denote by PSpin(M)°
We can regard Pso(m) as a bundle associated to P S p i n ( M ) via p : Spin(n) -> SO(n).
Using the unitary spinorial representation
An : Spin(n) -> Aut (in)
we get a Hermitian vector bundle

S(M) I PSpin(M) X A n 5,¢

called the complex spinoza bundle. Its sections are called (complex) spinors. Let us point out
that when M is even dimensional, the spinoza bundle is equipped with a natural Z/2-grading
§(m)=§+(m) ® s-<m>.
~
From Lemma 12.1.31 we deduce that § ( M ) is naturally a bundle of Cl(TM) modules.
Then, the natural isomorphism Cl(TM) = Cl(T*M) induces on S ( M ) a structure of
Cl(T*M)-module,
c c1(Tm) -> End@(§(m)).
Moreover, when M is even dimensional, the above morphism is compatible with the Z/2-
gradings of Cl(TM) and Ends(S(M)). As it turns out, the bundle S(M) has a natural
Dirac structure whose associated Dirac operator is the spin Dirac operator on M. We will
denote it by D.
Since An is a unitary representation, we can equip the spinoza bundle S ( M ) with a
natural metric with respect to which the Clifford multiplication
e c1(Tm) -> End@<§<m) )
is self-adjoint, i.e.,
¢<u*)== c<~)* 7 Vu E C'°°(Cl(TM)).
Since Cl(TM) is locally generated as an algebra by Vect ( M ) , we can equivalently rewrite
the self-adjointness condition as
e(X) = -e(X)* 7 VX 6 Vect (M) C C°°(Cl(T) I.
All we now need is to describe a natural connection on §(M) with respect to which the
Clifford multiplication is covariant constant.
We start with the Levi-Civita connection VI which we can regard as a connection on
the principal bundle Pso(m). Alternatively, VI can be defined by a collection of (n)-
valued 1-forms wa E Q1(UO() (X) M) such that

WB = 9o§d9<.5 + 9oéwo9o0 on U .
Consider the canonical isomorphism of Lie algebras

p* : spin(n) -> @(")-

658 Lectures on the Geometry of Manzfolds

Then the collection (DO( = P* 1((»J0) defines a connection on the principal bundle
S
P S p i n ( M ) » and thus via the representation ATL it defines a connection V = V on the
spinoza bundle S ( M ) .
The above construction can be better visualized if we work in local coordinates. Choose
a local, oriented orthonormal frame (QI of TM I UQ , and denote by ( e ) is dual coframe.
The Levi-Civita connection has the form
k .
Véj e ®wkje,

so that We ek (X) ( w ) , where for each k the collection (~»;) is a skew-symmetric

matrix.
Using the concrete description of p* 1 given in (12, 1.3) we deduce

k l . k
(Do € (X) Wki€1€j wkje (X) e e l
4
k i<j i,j,k

Lemma 12.2.32. The connection is a Cluj"ord connection on S ( M ) so that (SU/I), W

is a Dirac bundle called the bundle of complex spinors.

Proof. Use Lemma 12.1.31.

We denote by D = QM the geometric Dirac operator associated to the bundle of

complex spinors S. When the manifold M is even dimensional, this operator is graded.
We recall that this means that, with respect to the decomposition S = S+ ® S', the Dirac
operator D has the block form
(Q
II

where 321 : C°°(§+* -> C'°° (S') is a first order operator whose symbol is given by the
Clifford multiplication.
We now want to understand the structure of the WeitzenbOck remainder of the geomet-
ric Dirac operator CD. If

R Mk /\ be-Rm Rye = (RmI = lRij1<=2 l

k<£

is the Riemann curvature tensor, (in the above trivializations over UQ) we deduce that the
curvature of V is
1 k e
ek /\ ee (X) p* 1(Rig) 4 22; e /\ e (X) R¢j1<:e€¢€j .
k<£ k<£ in

From this we obtain

c(F())
1
8 Z Rwc(e')c()c(ek)c(e').
ike
 Dirac Operators 659

In the above sum, the terms corresponding to indices (i, i, k, g) such that i = j or k = Z
vanish due to the corresponding skew-symmetry of the Riemann tensor. Thus, we can write

c(FW)) - §
1
I ZZ
Roc(e)c(e)c(ek)c(e).
#j kgéé

Using the equalities e(e)e(ej) -|- c ( e j ) e ( e ) = -25 we deduce that the monomial
e l e c l e d l anti-commutes with e(eI*)e(e£) if the two sets{i, j} and {k, £} have a unique
element in common. Such pairs of monomials will have no contributions in the above sum
due to the curvature symmetry
Rijké Rkéij

Thus, we can split the above sum into two parts

1 . . . .
c(F(V)) = - Rm1C(€')C(€3)C(€')C(6"l + RimC(€')C(€i)C(€klC(€£)~

M i,j,k,€distinct

Using the first Bianchi identity we deduce that the second sum vanishes. The first sum is
equal to
1
Z
-Z Rim (c(@")c(@
,

Z Rfzjij 4 '
S
Z
'in M
where S denotes the scalar curvature of M. We have thus proved the following result, [86].

Theorem 12.2.33 (Lichnerowicz).

1
Q2 Z v*v S.
4
w
A section E cOO (S(M)) such that 9% = 0 is called a harmonic spinoza. Lichnerow-
icz theorem shows that a compact spin manifold with positive scalar curvature admits no
harmonic spinors.

Exercise 12.2.34. Consider an oriented 4-dimensional Riemann spin manifold (M, go and
W -> M a Hermitian vector bundle equipped with a Hermitian connection V. Form the
twisted Dirac operator
®W=c°°(<§+(m> ® §-(m>I (X) w)->c°°((§+(m> ® §-(mn (X) w)
defined in Exercise 12.1.68. The operator Cow is Z2-graded, and hence it has a block
decomposition

53w 7

where 5lZlw_l : COO(S+(M) (X) W ) -> C ° ° ( § ' ( M ) (X) W). Show that
S
@w,+@w,+ = * + Z +elF+(V)l,
where denotes the product connection of S ( M ) (X) W, and

F+(V) = -(FW) + *F(v) ),

denotes the self-dual part of the curvature of the bundle (W, V).
660 Lectures on the Geometry of Manzfolds

Exercise 12.2.35. The torus T3 Sl X S1 X S1 equipped with the product metric (of
volume (27rr)3) has eight spin structures. Since T3 is also a Lie group, its tangent bundle
admits a canonical trivialization. We denote by 00 the spin structure determined by this
trivialization. Using the free and transitive action of H1 (T3, Z2) on Spin(T3) we deduce
that we have a canonical bijection
Z ) -> ' T 3 Z2
H1IT3 2 Sp (T3 H1l >
) 9 6 »- Us
6 U0-
I,
7
1N
Compute the spectrum of the spin Dirac operator 536 determined by the spin structure 0@ .

12.2.4 The spin" Dime operator

Our last example of Dirac operator generalizes both the spin Dirac operator, and the Hodge-
Dolbeault operator. The common ingredient behind both these examples is the notion of
spin" structure. We begin by introducing it to the reader.
Let ( M n , g) be an oriented, n-dimensional Riemann manifold. As in the previous
section we can regard the tangent bundle as associated to the principal bundle Pso(m) of
oriented orthonormal frames. Assume Pso(m) is defined by a good open cover U = (Ua )
and transition maps

.QQB : Una -> SO(nl.

The manifold M is said to posses a spin" structure (or complex spin structure) if there
exists a principal Spin°(n)-bundle PSpin° such that Psouw) is associated to PSpin° via the
natural morphism pC : Spin°(n) -> SO(n):
Pso(m) = PSpin° X p c SO(n).
Equivalently, this means there exist smooth maps Eton : U05 -> Spin°(n), satisfying the
cocycle condition, such that

»o°(§a@) = 90B-
As for spin structures, there are obstructions to the existence SpinC structures, but they are
less restrictive. Let us try to understand what can go wrong. We stick to the assumption
that all the overlaps U05-~.w are contractible.
Since Spin°(n) = (Spin(n) X S1l/Z2, lifting the S O ( n ) structure (Gan) reduces to
finding smooth maps
hog : U0g -> Spin(n) and zag : Uog -> S1 7

such that

P(IIGB) = Qa
and

(6o0v> CQBwI h<1BhBvhv<1 ) 2Q52B72w0) Q {(-1, -1), (1, 1) }. (l2.2.4)

If we set A0g := 2 2 5 : Ugh -> S1, we deduce from (l2.2.4) that the collection (MBA
should satisfy the cocycle condition. In particular, it defines a principal S1-bundle over M,
 Dirac Operators 661

or equivalently, a complex Hermitian line bundle2 L. This line bundle should be considered
as part of the data defining a spine structure.
The collection l50(Bvl is an old acquaintance: it is a tech 2-cocycle representing the
second Stieffel-Whitney class.
As in Subsection 8.2.2, we can represent the cocycle A as
A05 = exp(i9o@).
The collection
l
"QM Z(
lllllllllll

HoB -|- HM -|- 9fy0 7

defines a 2-cocycle of the constant sheaf Z representing the topological first Chern class of
L. The condition (l2.2.4) shows that

"QM = €0B'y (mod QI,

To summarize, we see that the existence of a spinC structure implies the existence of a
complex Hermitian line bundle L such that
GYM) = w2(M) (mod 2).
It is not difficult to prove the above condition is also sufficient. In fact, one can be more
precise.
Denote by Spin°(M) the collection of isomorphism classes of spin" structures on the
manifold M. Any O' E Spin°(M) is defined by a lift (ho Zan ) as above. We denote by
det O' the complex Hermitian line bundle defined by the gluing data (zOo@). We have seen
that
cop(det (7) E w2(M) (mod 2).
Denote by LM C H2 (M, ZN the "affine " subspace consisting of those cosmology classes
satisfying the above congruence modulo 2. We thus have a map
Spin°(M) -> L m , 07 »-> c°p(det 0).

Proposition 12.2.36. The above map i5 a surjection.

Exercise 12.2.37. Complete the proof of the above proposition.

The smooth Picard group Pie°° (M) of isomorphisms classes of complex line bundles
(with group operation given by the tensor product) acts on Spin°(M) by

Spin°(M) X Pic°°(M) 9 (0, L) »-> 0 (X) L.

More precisely, if 0 6 Spin°(M) is given by the cocycle

O' [ h a o a ] : Ufa -> Spin(n) X S1/ ~,

2$n this subsection, by complex Hermitian line bundle we understand a complex line bundle equipped with a
U(1)-structure.
662 Lectures on the Geometry of Manzfolds

and L is given by the S1 cocycle Cog : UQ5 -> S1 then 9 O' (X) L is given by the cocycle
[ha@»zo@Co@] . Note that

det(c7 (X) L) = deto (X) L2 7

so that
cg019 l det(a (X) L) ) = et1°p(det 0) -|- 2@§op(L).
Proposition 12.2.38. The above action of Pic" (M) on Spine ( M l is free and transitive.
02
Proof. Consider two spin" structures 0.1 and defined by the good cover (Ua), and the
gluing cocycles
(1) (i)
[ha,3' 20,3]7 1,2.
e ( ) (2)
Since I0°(h(1§) P (how) = .90B> we can assume (eventually modifying the maps h015 by
a sign) that
_
ho( 1 ) _ h (2).
This implies that the collection
( )
Qag Zan/zale
(2)

is an S1-cocycle defining a complex Hermitian line bundle L. Obviously 02 = 0.1 (X) L.

This shows the action of Pie0°(M) is transitive. We leave the reader verify this action is
indeed free. The proposition is proved.

Given two spin" structures 071 and 02 we can define their "difference" U2/0'1 as the
unique complex Hermitian line bundle L such that (72 = 071 (X) L. This shows that the col-
lection of spinC structures is (non-canonically) isomorphic with H2(M, Z) 2- Pic°° ( M ) .
It is a sort of affine space modeled on H 2 ( m , Z) in the sense that the "difference "
between two spinC structures is an element in H2(M, Z), but there is no distinguished
origin of this space. A structure as above is usually called a H2(M, Z)-torsor.
The set Spin°(M) is equipped with a natural involution
Spin°(M) 9 07 »-> 6 6 Spin°(M),
defined as follows. If O' is defined by the cocycle [§o@»ZaBL then 6 is defined by the cocycle
[pa@» zaBI1] . Observe that

O' O' det<7.

Without sufficient background in algebraic topology the above results may look of very
little help in detecting spinC structures. This is not the case, and to convince the reader we
will list below (without proofs) some examples of spin" manifolds.

Example 12.2.39. (a) Any spin manifold admits a spine structure.

(b) Any almost complex manifold has a natural spine structure.
(c) (Hirzebruch-Hopf, [64], see also [102].) Any oriented manifold of dimension < 4
admits a spin" structure.
 Dirac Operators 663

Let us analyze the first two examples above. If M is a spin manifold, then the lift
QQB : USB -> Spin(n)
of the SO-structure to a spin structure canonically defines a spin" structure via the trivial
morphism
Spin(n) -> Spin°(n) XZ2
S1 n g »-> (g, 1) mod the Z2 - action.
We see that in this case the associated complex line bundle is the trivial bundle. This is
called the canonical spin" structure of a spin manifold. We thus have a map
Spin(M) -> Spin°(M).
Suppose we have fixed a spin structure on M given by a Spin-lift QuOf of an SO-gluing
cocycle 96a .
To any complex Hermitian line bundle L defined by the S1-cocycle (201Q ) we can asso-
sate the spin" structure O' (X) L defined by the gluing data
{(to@» ZQBI
The complex Hermitian line bundle deta (X) L associated to this structure is det O' (X)
L = L®2. Since the topological Picard group Pic0°(M) acts freely and transitively on
~
Spin°(]l/I), we deduce that to any pair (6, 0) e Spin(M) X Spin°(M) we can canonically
associate a complex line bundle L = L 6 , 0 such that L®2 - det ct, i.e., L 6 , 0 is a square root
of det O'.
Exercise 12.2.40. Show that for any O' E Spin°(M) there exists a natural bijection between
the set Spin(]l/I), and the set of isomorphisms of complex line bundles L such that L®2 =
~
det 0.

Exercise 12.2.41. Prove that the image of the natural map Spin(M) -> Spin°(M) coin-
cides with the fixed point set of the involution 0' »->o.

Exercise 12.2.42. The torus T3 is the base of a principal bundle

Z3->I[-2.3->T3 7

so that, to any group morphism p : Z3 -> U(l) we can associate a complex line bundle
Lp -> T3.
(a) Prove that Lp1°P2 E LP1 (X) LP27 Vp1, 01 6 Hom(Z3, S1), and use this fact to describe
the first Chern class of Lp by explicitly producing a Hermitian connection on Lp.
(b) Prove that any complex line bundle on T3 is isomorphic to a line bundle of the form
Lp, for some morphism p E Hom(Z3, S1).
(c) Show that the image of the map Spin(T3) -> Spin°(T3) consists of a single point.

Exercise 12.2.43. Prove that Spin(RIP3) consists of precisely two isomorphisms classes of
spin structures and moreover, the natural map
sp1n(R1p3) -> spin°(R1p)3)
is a bijection.
664 Lectures on the Geometry of Manzfolds

To understand why an almost complex manifold admits a canonical spin" structure it

suffices to recall the natural morphism U (lc) -> SO(2k) factors through a morphism
g : U(k) -> Spin°(2kl.
The U(k) -structure of TM, defined by the gluing data
hog : Uog -> U ( k ) ,
induces a spine structure defined by the gluing data €(h00l- Its associated line bundle is
given by the S1-cocycle
dot(clhogl : Uog -> S1 7
and it is precisely the determinant line bundle
d€t@T1'0M A'*0Tm.
The dual of this line bundle, d€t@lT*Ml1'0 = A'"00T*M plays a special role in algebraic
geometry. It usually denoted by Km, and it is called the canonical line bundle. Thus the
line bundle associated to this spin" structure is Kml dof KM .
From the considerations in Subsections 11.1.5 and 11.1.7 we see that many (complex)
vector bundles associated to the principal SpinC bundle of a spin" manifold carry natural
Clifford structures, and in particular, one can speak of Dirac operators. We want to discuss
in some detail a very important special case.
Assume that (M, g) is an oriented, n-dimensional Riemann manifold. Fix O' 6
Spin°(M) (assuming there exist spinC structures). Denote by (Hoe) a collection of glu-
ing data defining the SO structure Pso(m) on M with respect to some good open cover
(Ua). Moreover, we assume O' is defined by the data
hog : Ugh -> Spin°(n).
Denote by A; the fundamental complex spinorial representation defined in Subsec-
tion 11.1.6,
An Spin°(nl -> Aut (§,,,),
We obtain a complex Hermitian vector bundle
SuIMI = PSpinC XA" in,
which has a natural Clifford structure. This is called the bundle of complex spinors associ-
ated to o.
Example 12.2.44. (a) Assume M is a spin manifold. We denote by (70 the spinC structure
corresponding to the fixed spin structure. The corresponding bundle of spinors §0(M)
coincides with the bundle of pure spinors defined in the previous section. Moreover for any
complex line bundle L we have
aL :=SU ; S 0 ® L ,
where O' = 00 (X) L. Note that in this case L2 = det O' so one can write

s. 2- §0 (X) (det 0 ) 1 / 2 .
(b) Assume M is an almost complex manifold. The bundle of complex spinors associated
to the canonical spine structure O' (such that det O' = KM1) is denoted by S@(M). Note
that
§@lml A0"T*M.
r\/
 Dirac Operators 665

We will construct a natural family of Dirac structures on the bundle of complex spinors
associated to a spine structure.
Consider the warm-up case when TM is trivial. Then we can assume .90 B E II, and
h = (1h to a) •• U05 -> Spin(n) X S1 -> Spin°(n).
2
The S1-cocycle lz0g) defines the complex Hermitian line bundle det O`. In this case some-
thing more happens.
The collection (20B) is also an S1-cocycle defining a complex Hermitian line bundle
L, such that L2 2- det O`. Traditionally, L is denoted by (det 0 ) 1 / 2 , though the square root
may not be uniquely defined.
We can now regard §0 (M) as a bundle associated to the trivial Spin(n)-bundle PSpiD 9

and as such, there exists an isomorphism of complex Spin(n) vector bundles

§O'lM)§§lMl (X) ldeto)1/2
As in the Exercise 12.1.68 of Subsection 11.1.9, we deduce that twisting the canonical
connection on the bundle of pure spinors §0 (M) with any U(1)-connection on det 01/2 we
obtain a Clifford connection on SO' ( M ) . Notice that if the collection
{(»Ja e u(1) (X) 9l(Uo) }
defines a connection on det ct, i.e.,
dz2
we + W e over Ua0,
z25
then the collection
A 1

defines a Hermitian connection on /2 = det 071/2. Moreover, if F denotes the curvature of

(w.), then the curvature of ( o ) is given by
A 1
F F. (l2.2.5)
2
Hence any connection on det O' defines in a unique way a Clifford connection on SU (Ml.
Assume now that TM is not necessarily trivial. We can however cover M by open sets
(Url such that each TUo is trivial. If we pick from the start a Hermitian connection on
det 0, this induces a Clifford connection on each SO(Uol. These can be glued back to a
Clifford connection on §U(M) using partitions of unity. We let the reader check that the
connection obtained in this way is independent of the various choices.
Here is an equivalent way of associating a Clifford connection on SO' to any Hermitian
connection on det O`. Fix an open cover (url of M consisting of geodesically convex
open sets. The restriction of TM to any Ua is trivializable. Fixing such (orthogonal)
trivializations leading to the gluing cocycle
.BBQ : Uog -> SOM) , n : dim M.
We can describe the Levi-Civita connection on TM as a collection
AO( E 91(Ua) (X) 20"),
666 Lectures on the Geometry of Manzfolds

satisfying the transition rules

AB I -(d-g,30¢lgBa -|- .9B0-A0g5 1

The spinC structure O' is described by a gluing cocycle

age : Uog -> Spin°(n), h e = [Q zeal E Spin(n) X U(1)/{i1}7

such that, if p : Spin(n) -> SO(n) denotes the natural 2 : l morphism, then

9601 Z 0(§5av)-
Define A E Qllzfa) (X) spin(n) by setting

A ) P* 1lA0(;v) ) 7

where p* : spin(n) -> M) is the natural Lie algebra isomorphism described in (l2.1.3).
The collection (All satisfies the transition rules
-1 -1
A

AB Z
A

'ld.9B<1l95& + QQQAQQBQ
A A

-
Observe that although Q., is only defined up to a :l:l ambiguity, this ambiguity is lost in
the above equality.
Consider now a connection LU on the Hermitian line bundle det O' defined by the gluing
cocycle (z2,). It is defined by a collection

We E Q1(Uo,) ®Q (1) Q1IUQ) (X) HR,

satisfying the transition rules
d
"JB Z50 +wO1.
Z50

The collection
l
Ag := As, ® kwa
A

6 Qllul ® (spin(n) ® u(ll l Q1(z/l (X) spin°(n)

defines a connection Aw on the principal Spin°(n)-bundle PSpin° -> M determined by the

spin-structure U. In induces a connection V" on the associated vector bundle §6, and as
in the previous subsection one can verify that V" is a Clifford connection.

Example 12.2.45. Assume lm,gl is both complex and spin. Then a choice of a spin
structure canonically selects a square root K M - 1 / 2 of the line bundle Km1 because Km1
is the line bundle associated to the spin" structure determined by the complex structure on
M. Then
go E go ® KM -1/ 2

Any Hermitian connection on Km induces a connection of §(r:. If M happens to be Kohler,

~
then the Levi-Civita connection induces a complex Hermitian connection Km and thus a
Clifford connection on Sq; ( M ) = A0"T*M.
 Dirac Operators 667

Let Ld be a connection on det O`. Denote by V" the Clifford connection it induces
on §O'(M) and by 53. the associated geometric Dirac operator. Since the WeitzenbOck
remainder of this Dirac operator is a local object so to determine its form we may as well
assume SO' S (X) det o1/2. Using the computation of Exercise 12.1.68 and the form of the
,

WeitzenbOck remainder for the spin operator we deduce

1
Q
2
(Vw)*Vw-|- ZS -|- -e(FO'(w)
2 ), (12.2.6)

where FO' denotes the curvature of the connection Ld on det 0 , and C(FO') denotes the Clif-
ford multiplication by the purely imaginary 2-form FO' .

Exercise 12.2.46. Consider the torus T3 Sl X Sl X Sl equipped with the product round
metric

g (d@1)2 -|- (d02>2 + <d03>2 7

where for j = 1, 2, 3, 01 E [0, 2'/T) denotes the angular coordinate on the j-th circle.
Denote by 00 the unique spine structure on T3, such that det 00 is a trivial line bundle.
As in the Exercise 12.2.42, for every morphism p : Z3 -> S1, we denote by Lp the
associated complex line bundle. We set Op := 0 (X) Lp, and we fix a Hermitian metric on
Lp° The spinc structure up, and a Hermitian connection V on Lp determines spin" Dirac
operator 9p(V) .
(a) Prove that if is a Hermitian a unitary automorphism of Lp then

©p(7Vv'1) = 7©,,(v)7
In particular, the operators Dp('yV'7-1) and D,,(V) have the same spectrum.
(b)* We say that two Hermitian connections V, V' on Lp are gauge equivalent if and only
if there exists a unitary automorphism *y of L such that V' = '7V*y-1. Prove that two
connections V and V' are gauge equivalent if and only if
1
lllllllll.ll.l
del' (F(v ') F( Vl ) E 2triZ, *v'_j 1,2,3.
2tr T3 /\
(c) Fix a Hermitian connection V on Lp. Describe the spectrum of ®/,(V) in terms of the
real numbers
l .
So . do] Fw), j = l , 2 , 3 .
Zfrz T3
This page intentionally left blank
 Bibliography

[1] Mathematical Developments Arising from Hilbert Problems, Proceedings of Sym. in Pure
Math., vol. 28, Amer. Math. Soc. Providence RI, 1976.
[2] R. Adler, R. J. E. Taylor: Random Fields and Geometry, Springer Monographs in Mathemat-
ics, Springer Verlag, 2007.
[3] J. F. Adams: Vector jields on spheres, Ann. of Math. 75(1960), pp. 603-632.
[4] R. A. Adams: Sobolev Spaces, Academic Press, New York, 1975.
[5] M.T. Anderson, J. Cheeger: Ca -compactness for manifolds with Ricci curvature and injectiv-
ity radius bounded below, J. Diff. Geom., 35(1992), 265-281.
[6] V. I. Arnold: Ordinary Deferential Equations, Springer Texts, Springer-Verlag, 1992.
[7] V. I. Arnold: Mathematical Methods of Classical Mechanics, Springer Texts, Springer Verlag,
1989.
[8] M. F. Atiyah, R. Bott, V. K. Patodi: On the heat equation and the index theorem, Invent. Math.,
19(l973), 279-330.
[9] M. F. Atiyah, R. Bott, A. Shapiro: Clq§'ord modules, Topology 3, Suppl. 1 (1964), pp. 3-38.
[10] T. Aubine: Problemes isopérimétriques et espaees de Sobolev, J. Diff. Geom. 11(1976),
pp.533-598.
[11] T. Aubine: Equations dqférentielles non linéaires et problem de Yamabe concernant la cour-
bure scalaire, J. Math. Pures et Appl., 55(1976), pp. 269-296.
[12] V. Barbu: Dqterential Equations, Springer Verlag, 2016.
[13] P. Bérard: From vanishing theorems to estimating theorems: the Boehner technique revisited,
Bull. A.M.S. 19(1988), pp. 371-406.
[14] N. Berlin, E. Getzler, M. Vergne: Heat Kernels and Dirac Aperators, Springer Verlag, 1992.
[15] L. Bets, F. John, M. Schechter: Partial Derential Equations, Lectures in Applied Mathe-
matics, vol. 3A, Amer. Math. Soc., Providence RI, 1964.
[16] A. Besse: Einstein Manifolds, Ergebnisse der Mathematik, Bd.l0, Springer Verlag, 1987.
[17] P. Bérard, G. Besson, S. Gallot: Embedding Riemannian manifolds by their heat kernel, Geom.
Funct. Anal., 4(1994), 373-398.
[18] X. Bin: Derivatives of the spectral function and Sobolev norms of eigenfunctions on a closed
Riemannian manifold, Ann. Global. Analysis an Geometry, 26(2004), 231-252.
[19] A. Bjorner: Topological methods, Hadbook of Combinatorics, R. Graham, M. GrOtschel and
L. Lov8sz (Eds.), North-Holland, Amsterdam, 1995, pp. 1819-1872.
[20] J. Bochnak, M. Coste, M.-F. Roy: Real Algebraic Geometry, Springer Verlag, 1998.
[21] B. Boos, K. Wojciechowsld: Elliptic Boundary Problems for Dirac Operators, Birkhauser,
1993.

669
670 Lectures on the Geometry of Manzfolds

[22] R. Bott, L. Tu: Dijferential Forms in Algebraic Topology, Springer-Verlag, 1982.

[23] H. Brezis: Functional Analysis, Sobolev Spaces and Partial Deferential Equations, Universi-
text, Springer Verlag, 2011.
[24] T. BrOcker, K. Janich: Introduction To Dijferential Topology, Cambridge University Press,
1982.
[25] E. Cartan: Sur les invariants integraux des certains espaces nomogenes et les propriétés
topologiques de ces espaces, Ann. Soc. Pol. Math., 8(1929), pp. 181-225.
[26] H. Cartan: La transgression dans un group de Lie et dans un espace fibro principal, pp. 1268-
1282, vol. III of H. Cartan's "Collected Works ", Springer-Verlag, 1979.
[27] I. Chavel: Eigenvalues in Riemannian Geometry, Academic Press, 1984.
[28] J. Cheeger, D. G. Ebin: Comparison Theorems in Riemann Geometry, AMS Chelsea Publish-
ing, 2008.
[29] S. S. Chem: A simple intrinsic proof of the Gauss-Bonnet formula for closed Riemann mani-
folds, Ann. Math., 45(1946), pp. 747-752.
[30] J. B. Conway: A Course in Point Set Topology, Undergraduate Texts in Mathematics, Springer
Verlag, 2014
[31] M. Coste: An Introduction to o-minimal geometry, Real Algebraic and Analytic Geometry
. . .
Network, h t t p : / /www ihp- r a a g o r g / p u b l i c a t i o n s php
[32] R. Courant, D. Hilbert: Methods o f Mathematicsl Physics, vol. II, John Wiley & Sons, 1989.
[33] B. Dacorogna: Direct Methods in the Calculus of Variations, Applied Mathematical Sciences,
vol. 78, Springer-Verlag, 1989.
[34] J. Dieudonne: Foundations of Modern Analysis, Academic Press, 1963.
[35] M. P. DoCarmo: Riemannian Geometry, Birkhauser, 1992.
[36] L. van der Dries: Tame Topology and o-minimal Structures, London Math. Soc. Series,
vol. 248, Cambridge University Press, 1998.
[37] L. van der Dries, C. Miller: Geometric categories and o-minimal structures, Duke Math. J.
84(l996), 497-540.
[38] B. A. Dubrovin, A. T. Fomenko, S. P. Novikov: Modern Geometry - Methods And Appliea-
tions, Vol. 1-3, Springer Verlag, 1984, 1985, 1990.
[39] J. J. Duistermaat, J. A. C. Kolk: Distributions. Theory and Applications, Birkhauser, 2010.
[40] R. E. Edwards: Functional Analysis. Theory and Applications, Dover Publications, 1995.
[41] C. Ehresmann: Sur la topologic de eertains espoees nontogénes, Ann. of Math., 35(1934),
pp.396-443.
[42] S. Eilenberg, N. Steenrod: Foundations of Algebroie Topology, Princeton University Press,
1952.
[43] J. D. Elworthy, Y. LeJan, X.-M. Lin: On the Geometry ofDiti'usion Operators and Stochastic
Flows, Lect. Notes in Math., vol. 1720, Springer Verlag, 1999.
[44] L. C. Evans: Partial Differential Equations, 2nd Edition, Grad. Studies in Math., vol. 19,
Amer. Math. Soc., 2010.
[45] H. Federer: Geometric Measure Theory, Springer Verlag, 1969.
[46] W. Fulton, J. Harris: Representation Theory, Springer-Verlag 1991.
[47] R. P. Feynman, R. B. Leighton, M. Sands: The Feynantn Lectures on Physics, Addison Wesley,
1964.
[48] S. Gallot, D. Hulin, J. Lafonte: Riemannian Geometry, 3rd Edition, Springer Verlag, 2004.
[49] I.M. Gelfand, G.E. Shilov: Generalized Functions. Properties and Operations, vol. 1, Aca-
demic Press, 1964.
 Bibliography 671

[50] I.M. Gelfand, G.E. Shilov: Generalized Functions. Spaces o f Fundamental and Generalized
functions, vol. 2, Academic Press, 1968.
[51] D. Gilbarg, N. S. Trudinger: Elliptic Partial Duterential Equations of Second Order, Springer
Verlag, 1983.
[52] R. Godement: Topologie Algébrique et Tneorie des Faisceaux, Hermann et Cie, Paris, 1964.
[53] C. Gordon, D. L. Webb, S. Wolpert: One cannot hear the shape of a drum, Bull. A.M.S.,
27(1992), pp. 134-138.
[54] I. S. Gradshteyn, I. M. Ryzhik: Tables of Integrals, Series and Products, Academic Press,
1994.
[55] P. Griffiths, J. Harris: Principles of Algebraic Geometry, John Wiley & Sons, 1978.
[56] V. Guillemin, S. Sternberg: Geometric Asymptotics, Math. Surveys and Monographs, vol. 14,
Amer. Math. Soc., 1990.
[57] J. Hadamard: Lectures on Cauchy's Problem in Linear Partial Differential Equations, Yale
University Press, 1923.
[58] B. Hall: Lie Groups, Lie Algebras, and Reprsentations. An Elementary Introduction, Springer
Verlag 2003.
[59] G. H. Hardy, E.M. Wright: An Introduction to the Theory of Numbers, Oxford University
Press, 1962.
[60] P. Hartman: Ordinary Deferential Equations, 2nd Edition, Classics in Applied Mathematics,
SIAM 2002.
[61] S. Helgason: Deferential Geometry, Lie Groups, and Symmetric Spaces, Graduate Studies in
Math. v. 34, Amer. Math. Soc., 2001.
[62] M. Hirsch: DQj"erential Topology, Springer-Verlag, 1976.
[63] F. Hirzebruch: Topological Methods in Algebraic Geometry, Springer Verlag, New York,
1966.
[64] F. Hirzebruch, H. Hopf: Felder von Flachenelementen in 4-dimensionalen Manigfaltigkeiten,
Math. Ann. 136(1958).
[65] L. HOrmander: On the spectral function o f an elliptic operator, Acta Math. 121(l968), 193-
218.
[66] L. HOrmander: The Analysis o f Linear Partial Differential Operators I., Springer Verlag, 2003.
[67] L. HOrmanderr The Analysis o f Linear Partial Differential Operators III., Springer Verlag,
1985.
[68] B. verser: Cohomology of Sheaves, Springer Verlag, 1986.
[69] H. Iwamoto: On integral invariants and Betti numbers of symmetric Riemannian manifolds, I,
J. of the Math. Soc. of Japan, 1(1949), pp. 91-110.
[70] J. Jost: Riemannian Goer try and Geometric Analysis, Universitext, Springer Verlag, 1998.
[71] M. Kac: Can one near the shape of drum ?, Amer. Math. Monthly, 73(1966) Part II, pp. 1-23.
[72] M. Karoubi: K-Theory, Springer-Verlag, 1978.
[73] M. Kashiwara, P. Schapira: Sheaves on Manifolds, Springer Verlag, 1990.
[74] M. Kashiwara, P. Schapira: Categories and Sheaves, Springer Verlag, 2006.
[75] T. Kato: Perturbation Theory for Linear Operators, Springer Verlag, 1984.
[76] J. Kazdan, F. Wamer: Curvature functions for compact 2-manifolds, Ann. Math. 99(1974),
pp. 14-47.
[77] J. L. Kelley: General Topology, Van Nostrand, 1955.
[78] A. Khovansldi: Fewnomials, Transl. Math. Monogr., 88, Amer. Math. Soc., 1991.
[79] D. A. Klain, G.-C. Rota: Introduction to Geometric Probability, Cambridge University Press,
1997.
672 Lectures on the Geometry of Manzfolds

[80] S. Kobayashi, K. Nomizu: Foundations of Dw2erential Geometry, Interscience Publishers,

New York, 1963.
[81] K. Kodaira, D. C. Spencer: Groups of complex line bundles over compact Kiihler varieties.
Divisor class group on algebraic varieties, Proc. Nat. Acad. Sci. USA 39(1953), pp. 868-877.
[82] J. Kubarsld, N. Tele ran: N. Linear Direct Connections, Proceedings 7th Conference on "Ge-
ometry and Topology of Mani- folds - The Mathematical Legacy of Charles Ehresmann",
Betlewo, May 2005.
[83] H. B. Lawson, M.-L. Michelson: Spin Geometry, Princeton University Press, 1989.
[84] N. N. Lebedev: Special Functions and Their Applications, Dover, 1972.
[85] J. M. Lee, T. H. Parker: The Yantabe problem, Bull. A.M.S. 17(l987), pp. 37-91.
[86] A. Lichnerowicz: Laplacien sur one variété riemannienne et spineure, Atti Accad. Naz. dei
Lincei, Rendicotti 33(1962), 187-191 .
[87] D. E. Littlewood: The Theory of Group Characters and Matrix Representations of Groups,
Clarendon Press, Oxford, 1940.
[88] I. G. Macdonald: Symmetric Functions and Hall Polynomials, Clarendon Press, Oxford, 1979.
[89] S. MacLane: Categories for the Working Mathematician, Springer Verlag, 1971.
[90] W. S. Massey: A Basic Course in Algebraic Topology, Springer Verlag, 1991.
[91] W. S. Massey: Cross products of vectors in higher dimensional euclidian spaces, Amer. Math.
Monthly, 93(l983), pp. 697-700.
[92] P.-D. Mehtée: Sur les distributions in variantes dans Ie groups des rotations de Lorentz, Comm.
Math. Helv. 28(1954), 225-269.
[93] J. Milnor: On manu'olds homeomorphic to the 7-sphere, Ann. of Math. 64(1956), pp. 399-405 .
[94] J. W. Milnor: Morse Theory, Princeton University Press, 1963.
[95] J. W. Milnor: Spin structures on manifolds, L'Enseignement Math., 9(1963), 198-203.
[96] J. W. Milnor: Curvatures of left invariant metrics on Lie groups, Adv. in Math., 21(1976),
pp.293-329.
[97] J. W. Milnor: Topology from the Duterentiable Viewpoint, Princeton University Press, 1965.
[98] J. W. Milnor, J. D. Stasheff: Characteristic Classes, Ann. Math. Studies 74, Princeton Uni-
versity Press, Princeton, 1974.
[99] M. Mimura, H. Toda: Topology olLie Groups, I and II, Translations of Mathematical Mono-
graphs, vol. 91, Amer. Math. Soc., 1991.
[100] R. Miron, M. Anastasiei: The Geometry of Lagrange Spaces: Theory and Applications,
Kluwer Academic Publishers, 1994.
[101] J. W. Morgan: The Seiberg-Witten Equations and Applications to the Topology of Smooth
Manifolds, Mathematical Notes, Princeton University Press, 1996.
[102] C. Morrey: Multiple Integrals in the Calculus of Variations, Springer Verlag, 1966.
[103] A. Newlander, L. Nirenberg: Complex analytic coordinates on almost complex manifolds,
Ann. Math., 54(1954), pp. 391-404.
[104] L. I. Nicolaescu: A new method of constructing connections on vector bundles, Liviu's Math
Blog, October 31, 2014.
[105] L. I. Nicolaescu: A stochastic Gauss-Bonnet-Chern formula, arXiv: 1408.5746, Prob. Th. &
Related Fields, 165(2016), 235-265.
[106] L. I. Nicolaescu, N. Savale: The Gauss-Bonnet-Chern theorem: a probabilistic perspective,
Trans. A.M.S., 369(2017), 2951-2986.
[107] L. Nirenberg: On elliptic partial dwrerential equations, Ann. Scuola Norm. Sup. Pisa,
13(1959), pp. 1-48.
 Bibliography 673

[108] K. Nomizu, H. Ozeld: The existence of complete Riemann metrics, Proc. A.M.S., 12(1961),
889-891.
[109] S. Peters: Convergence o f Riemannian manifolds, Compositio Math., 62(1987), 3-16.
[110] V. I. Paulsen, M. Raghupati: Introduction to the Theory o f Reproducing Kernel Hilbert Spaces,
Cambridge Univ. Press, 2016.
[111] L. S. Pontryagin: Smooth manifolds and their applications in homotopy theory, Amer. Math.
Soc. Trans., II(1959), 1-114.
[112] I. R. Porteous: Topological Geometry, Cambridge University Press, 1981.
[113] M. M. Postnikov: Lectures in Geometry. Semester V' Lie Groups and Lie Algebras, Mir Pub-
lishers, Moscow, 1986.
[114] M. Riesz: L' integrate de Riemann-Liou ville et je problem de Cauchy, Acta Math., 81(1949),
1-223.
[115] L. SantalO: Integral Geometry and Geometric Probability, Cambridge University Press, 2004.
[116] R. Schoen: Conformal deformations of a Riemannian metric to constant curvature, J. Diff.
Geom. 20(1984), pp. 479-495.
[117] J. A. Schouten, D. J. Struik: On some properties of general manifolds relating to Einstein's
theory of gravitation, Amer. J. Math. 43(1921), pp. 213-216.
[118] L. Schwartz: Théorie des Distributions, Hermann, 1966.
[119] J. P. Serre: Faisceaux algébriques cohérentes, Ann. Math. 61(1955), pp. 197-278.
[120] I. R. Shafarevitch: Algebra I. Basic Notions, Encyclopedia of Mathematical Sciences, vol. 11,
Springer Verlag, 1990.
[121] M. A. Shubin: Pseudodqferential Operators and Spectral Theory, Springer Verlag, 2001.
[122] P. Speissegger: The Pfa/an closure of an o-minimal structure, J. Reine. Angew. Math.,
508(1999), 189-211.
[123] M. Spivak: A Comprehensive Introduction to Deferential Geometry, vol. 1-5, Publish or Per-
ish, 1979.
[124] N. E. Steenrod: The Topology of Fibre bundles, Princeton University Press, 1951.
[125] E. M. Stein: Singular Integrals and Dyferentiability Properties of Functions, Princeton Uni-
versity Press, 1970.
[126] M. Taylor: Pseudodifferential Operators, Princeton University Press, 1981.
[127] F. Troves: Topological Vector Spaces, Distributions and Kernels, Dover, 2006.
[128] F. W. Warner: Foundations of Di]§'erentiable Manifolds and Lie Groups, Springer Verlag, 1983.
[129] H. Weyl: Nachr. Akad. Wiss GOttingen, 1914, 235-236.
[130] H. Weyl: Math. Ann., 77(1916), pp. 313-315.
[131] H. Weyl: Die Idee Der Riemannschen Fleche, B.G. Tuebner, Berlin 1913.
[132] H. Weyl: On the volume of tubes, Amer. J. Math., 61(1939), 461-472.
[133] H. Weyl: The Classical Groups. Their Invariants and Representations, Princeton University
Press, 1939.
[134] E.T. Whittaker, G.N. Watson: A Course of Modern Analysis, Cambridge Univ. Press, 1996.
[135] E. Witten: Monopoles and four-manifolds, Math. Res. Letters, 1(1994), 769-796.
[136] S. Zelditch: Real and complex zeros o f Riemannian random waves, Spectral analysis in ge-
ometry and number theory, Contemp. Math., 484, Amer. Math. Soc., Providence, RI, 2009.
arXiv:0803.433v1
[137] K. Yosida: Functional Analysis, Springer Verlag, 1974.
This page intentionally left blank
 Index

(x :|; i0)", 540 0'r1/0'rg, 60

A*, 43 G l"U1, . . . ,"U7~l, 58
C°°(E), 35 I :t , 542
Q

C°°(U,E), 35 e ( E ) , 287
0-°°(6>)), 523 Xa-I-» 537

C°°(M), 7 Xa , 538
c+, 549 53, 45
c-, 549 det V, 49
E0 m E l , 182 det ind(L), 513
I°(g),337 div x, 158
Lk,p, 465 @(6'), 523
Re, 554 @K(6'), 523
Add, 127 raM,9» 574
CTF, 29 £2i,555
AF, 29 y(R">, 529
Am , 163 in, 529
End_(V), 61 5//, 530
GL(1K"), 18 €Xpq, 148

GL(n,H<), 18 <9m(9)9 192

Gr I<=(Vl9 13, 386 un, 37
A' V, 49 U p , T L 9 37, 359

Ap'qV, 65 grad f , 158

LOma 550 pp, 442
Q/"(F)» 95 J, 89
Qk(@), 528 f M , 113
Pin(V), 627 F E / B , 134
Spin(V), 627 up, 428, 432
Vect , 35, 70 115, 3
lIUIIL§(m)» 575 3-bundle, 136
Sn, 632 t&'1 (X7$0)9 229
T1n 577
7
Pm(ql, 148
G) (E), 119 Up (A), 50
@ M , 119 0 ( M ) , 272
L-genus, 357 ma 228

"y(2h), 403 spec(L), 513

675
676 Lectures on the Geometry of Manzfolds

sing supp u, 525 Dirac, 640

tr, 46 fiber, 73, 74
tIll, 431 frame, 76
Hom(E, F), 36 Hopf, 77
IXI", 539, 565 line, 34, 37
.»4-genus, 357 tautological, 37, 371
{-» '}9 613 universal, 37
bk(M), 247 principal, 75, 329
d*,l59 connection on, 330
dm(a), 555, 558 pullback of a, 342
or-density, 63 tangent, 28
38I(1)9 572 vector, 33, 37
$8 536 rank of a, 34
Km, 664 ample, 39
kg, 60 automorphism, 36
5€[[fD]]"7 352 base of a, 34
Pie°°, 345, 661 dual, 68
ind (7), 217 endomorphism, 36
try, 56 framed, 39
i7;'(V), 45 map, 36
gl(n), 20, 88 morphism, 36
Qn), 20, 88 pullback, 38
QM), 20, 88 section, 35
_ ( n ) , 20, 88 tautological, 37
Q(n), 20, 88 total space of a, 34
161 trivial, 39, 107
trivializable, 39, 116
adjoint, 43 trivialized, 39
adjoint representation, 127, 301, 329 universal, 37
adjunction isomorphism, 44
adjunction morphism, 56 Cartan
ample subspace, 182 algebra, 344, 347, 348
antipodal points, 13 form, 164, 298
atlas, 6 Lemma, 176
maximal, 7 structural equations, 174
categories, 232
berezinian, 60 Cayley transform, 18, 142
Bernoulli numbers, 353 character, 130
Beta function, 383 characteristic class, 342
Betti number, 247 2nd Stiefel-Whitney, 655
Bianchi identity, 108, 166 universal, 353, 357
bootstrap, 501 charts, 6
bundle, 27 Chem
G-Hber, 74 classes, 345
canonical line -, 664 universal, 353
determinant line, 68 universal rank r, 345
Clifford, 613 polynomial, 345, 353
complex spinoza, 657 universal rank r, 345
cotangent, 69 Chem-Weil correspondence, 342
determinant line, 112, 116 Christoffel symbols, 109, 144, 165
 Index 677

circle, 21 covering space, 233

class function, 130 sheets of a, 233
Clifford universal, 238
algebra, 615 critical
connection, 640 point, 29
module, 617 value, 29
multiplication, 613, 614 critical point, 30
s-module, 617 current, 528
structure, 613, 640 boundary of, 529
coboundary, 248 of integration, 528
cochain homotopy, 249 curvature, 102, 331
cochain morphism, 249 Gauss, 170
cocycle, 248 Ricci, 167, 218, 647
cohomologous, 248 Riemann, 165
condition, 34 scalar, 168
cosmology sectional, 169, 217
Cech, 318, 321 cycle, 272
DeRham, 246 degenerate, 273
compact supports, 265 transversal, 277
complex cycles
chain, 247 concordant, 272
cochain, 247
acyclic, 248 deck transformation, 235
DeRham, 248 densities, 111
Dolbeault, 651 Jacobian, 379
Koszul, 652 pullback of, 379
complex structure, 64 pushforward of, 380
configuration space, 3 density bundle, 111
conjugate points, see geodesic dependence cone, 561
connection, 95, 330 DeRham
compatible with a metric, 97, 104, 143 cosmology, 248
flat, 104 complex, 248
gauge equivalence, 335 derivative, 4
Levi-Civita, 143 Fréchet, 4
symmetric, 109 diffeomorphism, 7
connection 1-form, 97 of manifolds with boundary, 124
conformal, 124 differential, 28
inner, 124 differential form, 69
outer, 124 closed, 241
contraction, 46, 54, 72, 88 exact, 241
convolution, 466 Dirac function, 524
coordinates, 1 distribution, 523
Cartesian system, 1 temperate, 530
local, 6 distributional derivative, 525
polar, 2 duality, 56
coordinatization, 1, 3 Hodge, 62
correlator, 182, 601 metric, 57
associated to an embedding, 183 natural, 56
tunneling associated to a, 184 symplectic, 57
covariant derivative, see connection duplication formula, 542
678 Lectures on the Geometry of Manzfolds

Dynldn polynomials, 88 Frechet derivative, 4

framing, 39, 70
Einstein Fredholm index, 506
convention, 45 function, 3
equation, 193 Co, 4
manifold, 193 Fréchet differentiable, 3
tensor, 193 harmonic, 163
elliptic complex, 518 smooth, 4, 7
embedding, 32 functors, 232
energy method, 563 contravariant, 232
equation covariant, 232
Einstein, 193 fundamental solution, 548
Euler-Lagrange, 202 Cauchy, 548
Hamilton-Jacobi, 221
Jacobi, 214 Gamma function, 164, 384, 390
Euler gauge equivalence, see connection
characteristic, 52, 196, 247, 253, 258, 288 gauge transformation, 36, 334
class, 287, 351 Gauss
geometric, 351 curvature, 170, 181
topological, 351 lemma, 151
universal, 351 map, 179, 180, 399
form, 171, 180, 351 Gauss Lemma, 568
Euler-Lagrange equation, 202 Gelfand-Leray residue, 136, 551
exponential map see manifold 148 generalized
exterior, 49 function
algebra, 49 Riesz, 557
derivative, 89 generalized function, 523
product, 49 of order < m, 523
extremal, 202 singular support of, 525
support of, 525
five lemma, 261 genus, 196
flow, 79 geodesic, 144
infinitesimal generator of a, 80 flow, 145
local, 79, 81 conjugate points, 215
formula index of a, 217
duplication, 542 nondegenerate, 215
Parseval, 575 global angular form, 365
Poisson summation, 579 gluing cocycle, 35, 330
reflection, 541 gluing condition, 35
co-area, 381, 382 Grammian, 58
coarea, 381 Grassmannian, 13, 37, 299
Crofton, 434, 443, 448 complex, 13, 294, 299
Euler-Meusnier, 438 real, 13
Gauss, 177 group, 18, 40
Green, 163, 463 action, 73
Kenneth, 262 effective, 73
Parseval, 530 free, 73
Stokes, 124 orbit, 73
Weyl integration, 301 Clifford, 625
Fourier transform, 485, 530 complex Clifford, 633
 Index 679

connected, 20 Killing
fundamental, 229 pairing, 146, 249, 298
left translation in a, 18 vector Held, 160
Lie, 18, 40, 86, 127, 145, 169, 298, 329, 339 ldnematic density, 390
exponential map of a, 87 Kronecker
representation, 74 pairing, 269
linear representation, 127 symbol, 43
irreducible, 128
orthogonal, 18 lagrangian, 201
Picard, 345 invariant, 208
right translation in a, 40 Laplacian, 163, 458
special orthogonal, 19 covariant, 462
special unitary, 19 least action, see Euler-Lagrange
unitary, 19 Lefschetz number, 284
Gysin map, 271, 284 Lie
bracket, 82
Hamilton-Jacobi equation, 221 derivative, 81
heat kernel, 608 group, see group
Heaviside function, 525 Lie algebra, 55, 85, 87, 121
simple, 298
Hessian, 189, 212
semisimple, 146, 249, 298
Hilbert-Einstein functional, 192, 427, 447
light cone
Hodge *-operator, see operator
backward, 549
holonomy, 107
forward, 549
homogeneous space, 291
homotopy, 254
M6bius band, 115
manifold
immersion, 32
with boundary, 123
injectivity radius, 148
almost complex, 648
integral curve, 80
complex, 13
integration current, 528 connect sum, 21
interior derivation, 54, 88 connected sum, 11
intersection direct product, 11
form, 272 finite type, 259
number, 277 Kohler, 653
invariant orientable, 21, 116
form, 295 orientation of a, 116
forms, 297, 299 oriented, 115, 116
integration, 121 Riemann, 140
polynomials, 337 convex sets in a, 154
tensor field, 86 exponential map of a, 148
isometric, 140 geodesically complete, 145
isotropy, 292 smooth, 6
spinnable, 654
Jacobi manifolds, 3, 10
equation, 214 with boundary
Held, 214, 217 diffeomorphism of, 124
identity, 55 map, see function
Jacobian differential of, 28
relative, 379 gluing, 13
680 Lectures on the Geometry of Manzfolds

cochain chirality, 620

cone of a, 252 closed, 504
cylinder of a, 252 Fredholm, 506
covering, 233 self-adjoint, 504
degree of a, 282 semi-Fredholm, 506
homotopic, 228 symmetric, 505
lift of a, 235 coboundary, 247
quantization, 630 Dirac, 611
transition, 6, 34 suspension of, 612
cohomologous, 37 Fredholm, 506
transversal, 32 geometric Dirac, 641
matching, 418 Hamilton-Floer, 612
matrix, 4 Hodge-DeRham, 612
invertible, 4 Hodge-Dolbeault, 652
Jacobian, 4 linear, 3
orthogonal, 18 transpose of a, 43
skew-Hermitian, 19 partial differential, see p.d.o.
skew-symmetric, 18 orientation, 59
unitary, 19 bundle, 119
maximum principle, 495 canonical, 67
Mayer-Vietoris cover, 234
cover, 260 relative signature, 60
functors, 259 orientation cover, 234
correspondence of, 260 orientationsee vector space 59
long sequence, 256, 267 outer normal, 161
principle, 261
short sequence, 255, 267 p.d.o., 451, 452
metric, 69 elliptic, 456
conformal, 502 formal adjoint, 459
Kohler, 653 formally self-adjoint, 460
Riemann, 69 symbol of a , 456
metric volume, 157 pairing, 56
module Hodge, 62
G-, 127 parallel transport, 100
irreducible, 128 parametrix, 491
morphism, 128 partition, 307
mollifier, 466 conjugate of a, 308
mollifying sequence, 467 length of a, 307
monodromy, 231, 237 weight of a, 308
moving frame, 97 partition of unity, 9
pfaffian, 61, 349
negligible set, 29 Picard group, 345, 661
nerve, 317 Poincaré
Newton formulae, 356 duality, 269, 277
normal coordinates, 149 inequality, 499, 507
lemma, 242
operator, 3 for compact supports, 265
Hodge *Q 62, 157, 520, 653 polynomial, 247
boundary, 247 series, 52
Cauchy-Riemann, 612 polynomial
 Index 681

Ad-invariant, 337 simply connected, 230

Chem, 345 Sobolev, 465
Pontryagin, 348 tangent, 23, 27
Pontryagin space of germs, 314
classes, 348 morphism, 316
universal rank T, 348 section of, 314
polynomial stalk, 314
universal rank T, 348 spectral data, 523, 598
presheaf, 313 spectral function, 574, 589
projection, 10 spectral perestroika, 598
stereographic, l() sphere, 10
projection formula, 135 sphere bundle, 77
projective plane, see space spin structure, 654
pullback, 71 spinC structure, 660
pushforward, 271, 285 spinoza, 657
bundle, 657
quantization map, 617 harmonic, 659
quaternions, 147 module, 619, 622, 640
split coordinates, 134
reflection formula, 541 structural equations, 172
representation, 127 structure, 409
reproducing kernel 0-minimal, 409
Hilbert space, 191 tame, 409
Ricci tensor, see curvature submanifold, 9
Riemann metric, 140 2nd fundamental form, 177, 400
Riesz, 557 codimension of a, 9
submersion, 32
Schur super, 53
lemma, 129 algebra, 53
polynomial, 309 commutative, 53
Schwartz commutator, 53
function, 529 derivation, 54
second fundamental form, see submanifold Lie algebra, 54
shape operator, 179, 399, 427 space, 53
sheaf, 313 tensor product, 55
Hne,322 superuace,56
resolution, 317 surface, 21
Hne,323 doughnut-shaped, 21
sheafification, 315 symbol map, 617
Shwartz symmetric space, 292
space, 529 symmetry, 18
simplicial scheme, 325
smooth, see function tame geometry, 409
structure, 6 tensor, 42
space product, 42
Banach, 3 algebra, 45
complex projective, 13 contravariant, 45
locally compact, 6 covariant, 45
real projective, 12 Held, 69
second countable, 6 left invariant, 86
682 Lectures on the Geometry of Manzfolds

right invariant, 86 terser, 662

product torus, 11, 18
universality property of, 42 maximal, 300
skew-symmetric, 47 trace, 46, 50
exterior product of, 47 transform, 18
symmetric, 47 Fourier, 485
type (r,s), 45 transpose, 43
test function, 523 transversal, 32
theorem, 3 trivializing cover, 33
Fubini, 31, 374 tube formula, 425
Nash embedding, 428, 430 tunneling, 184
Banach Fixed point, 5
Bochner vanishing, 647 uniform boundedness principle, 524
Bochner-WeitzenbOck, 643 unit outer normal, 161
Calderon-Zygmund, 486
Canan,297 variation, 203, 211
Caftan-Hadamard, 234 geodesic, 214
Chern-Weil, 339 infinitesimal, 211
Clairaut, 210
vector, 24
dominated convergence, 449
field, 35
Fubini, 135
space, 24, 41
Gauss-Bonnet, 181, 192, 258, 283, 404,
Z-graded, 51
430,609
determinant line of a, 49, 59
Gauss-Bonnet-Chern, 363, 404, 609
dual, 43
Hodge, 518, 520
orientation of a, 59
Hopf-Rinow, 155
oriented, 59
implicit function, 5
volume form in a, 59
inverse function, 4
Leray-Hirsch, 264 tangent, 24
Mayer-Vietoris, 256 vector bundle, see bundle
Morrey, 477 volume form, 116
Myers, 218
Noether, 209 wave
Poincaré-Hopf, 291 family, 557
Rellich-Kondrachov compactness, 474 kernel, 581
Said, 30, 282, 435, 629 weak derivative, 464
Sobolev embedding, 472 WeitzenbOck
Stokes, 124 remainder, 463, 645, 658
Tarsld-Seidenberg, 409 Weyl
uniformization, 502 group, 302, 344, 347
Weyl, 240 integration formula, 30 l
Theorem egregium, 177, 400 lemma, 492
Thom class, 286 unitary trick, 128
Todd class Whitney sum, 68
universal, 354
torsion, 109 Young diagram, 308