Functional Analysis Spectral Theory and

Applications Graduate Texts in Mathematics 276

Manfred Einsiedler pdf download
https://textbookfull.com/product/functional-analysis-spectral-theory-and-applications-graduate-
texts-in-mathematics-276-manfred-einsiedler/

★★★★★ 4.9/5.0 (45 reviews) ✓ 214 downloads ■ TOP RATED

"Excellent quality PDF, exactly what I needed!" - Sarah M.

DOWNLOAD EBOOK
 Functional Analysis Spectral Theory and Applications
Graduate Texts in Mathematics 276 Manfred Einsiedler pdf
download

TEXTBOOK EBOOK TEXTBOOK FULL

Available Formats

■ PDF eBook Study Guide TextBook

EXCLUSIVE 2025 EDUCATIONAL COLLECTION - LIMITED TIME

INSTANT DOWNLOAD VIEW LIBRARY

Collection Highlights

Introduction to the Theory of Quantum Information

Processing (Graduate Texts in Physics) 2013th Edition
Bergou

Functional Analysis in Applied Mathematics and Engineering

First Edition Pedersen

An Introduction to Automorphic Representations: With a

view toward trace formulae (Graduate Texts in Mathematics,
300) Getz

Fundamental Polymer Science Graduate Texts in Physics Ulf

W. Gedde
A Course in Functional Analysis and Measure Theory
Vladimir Kadets

Mathematics of Finance An Intuitive Introduction

Undergraduate Texts in Mathematics Saari

Mathematics of Finance An Intuitive Introduction

Undergraduate Texts in Mathematics Saari Donald G

A Guide to Spectral Theory Applications and Exercises 1st

Edition Christophe Cheverry Nicolas Raymond

Fractional Equations and Models Theory and Applications

Developments in Mathematics Trifce Sandev
Graduate Texts in Mathematics

Manfred Einsiedler
Thomas Ward

Functional
Analysis,
Spectral Theory,
and Applications
Graduate Texts in Mathematics 276
Graduate Texts in Mathematics

Series Editors:

Sheldon Axler
San Francisco State University, San Francisco, CA, USA

Kenneth Ribet
University of California, Berkeley, CA, USA

Advisory Board:

Alejandro Adem, University of British Columbia

David Eisenbud, University of California, Berkeley & MSRI
Irene M. Gamba, The University of Texas at Austin
J.F. Jardine, University of Western Ontario
Jeffrey C. Lagarias, University of Michigan
Ken Ono, Emory University
Jeremy Quastel, University of Toronto
Fadil Santosa, University of Minnesota
Barry Simon, California Institute of Technology

Graduate Texts in Mathematics bridge the gap between passive study and creative
understanding, offering graduate-level introductions to advanced topics in mathe-
matics. The volumes are carefully written as teaching aids and highlight character-
istic features of the theory. Although these books are frequently used as textbooks
in graduate courses, they are also suitable for individual study.

More information about this series at http://www.springer.com/series/136

Manfred Einsiedler • Thomas Ward

Functional Analysis, Spectral

Theory, and Applications
Manfred Einsiedler Thomas Ward
ETH Zürich School of Mathematics
Zürich, Switzerland University of Leeds
Leeds, UK

ISSN 0072-5285 ISSN 2197-5612 (electronic)

Graduate Texts in Mathematics
ISBN 978-3-319-58539-0 ISBN 978-3-319-58540-6 (eBook)
DOI 10.1007/978-3-319-58540-6

Library of Congress Control Number: 2017946473

Mathematics Subject Classification (2010): 46-01, 47-01, 11N05, 20F69, 22B05, 35J25, 35P10, 35P20,
37A99, 47A60

© Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein
or for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

Believe us, we also asked ourselves what could be the rationale for ‘Yet an-
other book on functional analysis’.(1) Little indeed can justify this beyond
our own enjoyment of the beauty and power of the topics introduced here.
Functional analysis might be described as a part of mathematics where
analysis, topology, measure theory, linear algebra, and algebra come together
to create a rich and fascinating theory. The applications of this theory are
then equally spread throughout mathematics (and beyond).
We follow some fairly conventional journeys, and have of course been in-
fluenced by other books, most notably that of Lax [59]. While developing the
theory we include reminders of the various areas that we build on (in the
appendices and throughout the text) but we also reach some fairly advanced
and diverse applications of the material usually called functional analysis that
often do not find their place in a course on that topic.
The assembled material probably cannot be covered in a year-long course,
but has grown out of several such introductory courses taught at the Eid-
genössische Technische Hochschule Zürich by the first named author, with
a slightly different emphasis on each occasion. Both the student and (espe-
cially) the lecturer should be brave enough to jump over topics and pick the
material of most interest, but we hope that the student will eventually be
sufficiently interested to find out what happens in the material that was not
covered initially. The motivation for the topics discussed may by found in
Chapter 1.
Notation and Conventions
The symbols N = {1, 2, . . . }, N0 = N ∪ {0}, and Z denote the natural
numbers, non-negative integers and integers; Q, R, C denote the rational
numbers, real numbers and complex numbers. The real and imaginary parts
of a complex number are denoted by x = ℜ(x + iy) and y = ℑ(x + iy).
For functions f, g defined on a set X we write f = O(g) or f ≪ g if there
is a constant A > 0 with kf (x)k 6 Akg(x)k for all x ∈ X. When the implied
constant A depends on a set of parameters A, we write f = OA (g) or f ≪A g

v
vi Preface

(but we may also forget the index if the set of parameters will not vary at
all in the discussion). A sequence a1 , a2 , . . . in any space will be denoted (an )
(or (an )n if we wish to emphasize the index variable of the sequence). For
two C-valued functions f, g defined on Xr{x0 } for a topological space X
containing x0 we write f = o(g) as x → x0 if limx→x0 fg(x) (x)
= 0. This definition
includes the case of sequences by letting X = N ∪ {∞} and x0 = ∞ with
the topology of the one-point compactification. Additional specific notation
introduced throughout the text is collected in an index of notation on p. 600.
Prerequisites
We will assume throughout that the reader is familiar with linear algebra
and quite frequently that she is also familiar with finite-dimensional real
analysis and complex analysis in one variable. Further background and con-
ventions in topology and measure theory are collected in two appendices, but
let us note that throughout compact and locally compact spaces are implicitly
assumed to be Hausdorff.
Organisation
There are 402 exercises in the text, 221 of these with hints in an appendix,
all of which contribute to the reader’s understanding of the material. A small
number are essential to the development (of the ideas in the section or of later
theories); these are denoted ‘Essential Exercise’ to highlight their significance.
We indicate the dependencies between the various chapters in the Leitfaden
overleaf and in the guide to the chapters that follows it.
Acknowledgements
We are thankful for various discussions with Menny Aka, Uri Bader,
Michael Björklund, Marc Burger, Elon Lindenstrauss, Shahar Mozes, René
Rühr, Akshay Venkatesh, and Benjamin Weiss on some of the topics presen-
ted here. We also thank Emmanuel Kowalski for making available his notes
on spectral theory and allowing us to raid them. We are grateful to sev-
eral people for their comments on drafts of sections, including Menny Aka,
Manuel Cavegn, Rex Cheung, Anthony Flatters, Maxim Gerspach, Tommaso
Goldhirsch, Thomas Hille, Guido Lob, Manuel Lüthi, Clemens Macho, Alex
Maier, Andrea Riva, René Rühr, Lukas Ruosch, Georg Schildbach, Samuel
Stark, Andreas Wieser, Philipp Wirth, and Gao Yunting. Special thanks are
due to Roland Prohaska, who proofread the whole volume in four months.
Needless to say, despite these many helpful eyes, some typographical and
other errors will remain — these are of course solely the responsibility of the
authors.
The second named author also thanks Grete for her repeated hospitality
which significantly aided this book’s completion, and thanks Saskia and Toby
for doing their utmost to prevent it.
Manfred Einsiedler, Zürich
Thomas Ward, Leeds
2nd April 2017
Leitfaden

Banach Spaces 2

Hilbert Spaces & Fourier Series

4 5 Sobolev Spaces
& Dirichlet Problem
Completeness

6
Dual Spaces 7
Compact Operators
& Weyl’s Law

Weak* Compactness
& Locally Convex Spaces 8

Unitary Operators 9 Banach Algebras

& Fourier Transform 10 & Pontryagin Duals
Haar Measure,
Amenability, 11
& Property (T)

Spectral Theorems
12
& Pontryagin Duality

14 13

Prime Number Theorem Unbounded Operators

vii
viii Preface

Guide to Chapters
Chapter 1 is mostly motivational in character and can be skipped for the
theoretical discussions later.
Chapter 4 has a somewhat odd role in this volume. On the one hand
it presents quite central theorems for functional analysis that also influence
many of the definitions later in the volume, but on the other hand, by chance,
the theorems are not crucial for our later discussions.
The dotted arrows in the Leitfaden indicate partial dependencies. Chapter 6
consists of two parts; the discussion of compact groups depends only on
Chapter 3 while the material on Laplace eigenfunctions also builds on mater-
ial from Chapter 5. The discussion of the adjoint operator and its properties
in Chapter 6 is crucial for the spectral theory in Chapters 11, 12, and 13.
Moreover, one section in Chapter 8 builds on and finishes our discussion of
Sobolev spaces in Chapters 5 and 6. Finally, some of Chapter 11 needs the
discussion of Haar measures in the first section of Chapter 10.
With these comments and the Leitfaden it should be easy to design many
different courses of different lengths focused around the topic of Functional
Analysis.
Contents

1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 From Even and Odd Functions to Group Representations . . . . 1
1.2 Partial Differential Equations and the Laplace Operator . . . . . 5
1.2.1 The Heat Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.2 The Wave Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.3 The Mantegna Fresco . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3 What is Spectral Theory? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4 The Prime Number Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.5 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2 Norms and Banach Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.1 Norms and Semi-Norms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1.1 Normed Vector Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.1.2 Semi-Norms and Quotient Norms . . . . . . . . . . . . . . . . . . . 21
2.1.3 Isometries are Affine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.1.4 A Comment on Notation . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2 Banach Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2.1 Proofs of Completeness . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2.2 The Completion of a Normed Vector Space . . . . . . . . . . 36
2.2.3 Non-Compactness of the Unit Ball . . . . . . . . . . . . . . . . . . 38
2.3 The Space of Continuous Functions . . . . . . . . . . . . . . . . . . . . . . . 39
2.3.1 The Arzela–Ascoli Theorem . . . . . . . . . . . . . . . . . . . . . . . 40
2.3.2 The Stone–Weierstrass Theorem . . . . . . . . . . . . . . . . . . . . 42
2.3.3 Equidistribution of a Sequence . . . . . . . . . . . . . . . . . . . . . 48
2.3.4 Continuous Functions in Lp Spaces . . . . . . . . . . . . . . . . . 51
2.4 Bounded Operators and Functionals . . . . . . . . . . . . . . . . . . . . . . 55
2.4.1 The Norm of Continuous Functionals on C0 (X) . . . . . . 60
2.4.2 Banach Algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.5 Ordinary Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.5.1 The Volterra Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.5.2 The Sturm–Liouville Equation . . . . . . . . . . . . . . . . . . . . . 66

ix
x Contents

2.6 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

3 Hilbert Spaces, Fourier Series, Unitary Representations . . 71

3.1 Hilbert Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.1.1 Definitions and Elementary Properties . . . . . . . . . . . . . . 71
3.1.2 Convex Sets in Uniformly Convex Spaces . . . . . . . . . . . . 75
3.1.3 An Application to Measure Theory . . . . . . . . . . . . . . . . . 83
3.2 Orthonormal Bases and Gram–Schmidt . . . . . . . . . . . . . . . . . . . 86
3.2.1 The Non-Separable Case . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.3 Fourier Series on Compact Abelian Groups . . . . . . . . . . . . . . . . 91
3.4 Fourier Series on Td . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.4.1 Convolution on the Torus . . . . . . . . . . . . . . . . . . . . . . . . . 97
3.4.2 Dirichlet and Fejér Kernels . . . . . . . . . . . . . . . . . . . . . . . . 99
3.4.3 Differentiability and Fourier Series . . . . . . . . . . . . . . . . . . 104
3.5 Group Actions and Representations . . . . . . . . . . . . . . . . . . . . . . . 106
3.5.1 Group Actions and Unitary Representations . . . . . . . . . 107
3.5.2 Unitary Representations of Compact Abelian Groups . 110
3.5.3 The Strong (Riemann) Integral. . . . . . . . . . . . . . . . . . . . . 111
3.5.4 The Weak (Lebesgue) Integral . . . . . . . . . . . . . . . . . . . . . 113
3.5.5 Proof of the Weight Decomposition . . . . . . . . . . . . . . . . . 115
3.5.6 Convolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
3.6 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

4 Uniform Boundedness and the Open Mapping Theorem . . 121

4.1 Uniform Boundedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.1.1 Uniform Boundedness and Fourier Series . . . . . . . . . . . . 123
4.2 The Open Mapping and Closed Graph Theorems . . . . . . . . . . . 126
4.2.1 Baire Category . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.2.2 Proof of the Open Mapping Theorem . . . . . . . . . . . . . . . 128
4.2.3 Consequences: Bounded Inverses and Closed Graphs . . 130
4.3 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

5 Sobolev Spaces and Dirichlet’s Boundary Problem . . . . . . . . 135

5.1 Sobolev Spaces and Embedding on the Torus . . . . . . . . . . . . . . 135
5.1.1 L2 Sobolev Spaces on Td . . . . . . . . . . . . . . . . . . . . . . . . . . 135
5.1.2 The Sobolev Embedding Theorem on Td . . . . . . . . . . . . 138
5.2 Sobolev Spaces on Open Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
5.2.1 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
5.2.2 Restriction Operators and Traces . . . . . . . . . . . . . . . . . . . 146
5.2.3 Sobolev Embedding in the Interior . . . . . . . . . . . . . . . . . . 149
5.3 Dirichlet’s Boundary Value Problem and Elliptic Regularity . 152
5.3.1 The Semi-Inner Product . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.3.2 Elliptic Regularity for the Laplace Operator . . . . . . . . . 155
5.3.3 Dirichlet’s Boundary Value Problem . . . . . . . . . . . . . . . . 160
5.4 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Contents xi

6 Compact Self-Adjoint Operators, Laplace Eigenfunctions . 167

6.1 Compact Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
6.1.1 Integral Operators are Often Compact . . . . . . . . . . . . . . 170
6.2 Spectral Theory of Self-Adjoint Compact Operators. . . . . . . . . 174
6.2.1 The Adjoint Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
6.2.2 The Spectral Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
6.2.3 Proof of the Spectral Theorem . . . . . . . . . . . . . . . . . . . . . 178
6.2.4 Variational Characterization of Eigenvalues . . . . . . . . . . 181
6.3 Trace-Class Operators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
6.4 Eigenfunctions for the Laplace Operator . . . . . . . . . . . . . . . . . . . 196
6.4.1 Right Inverse and Compactness on the Torus . . . . . . . . 197
6.4.2 A Self-Adjoint Compact Right Inverse . . . . . . . . . . . . . . 198
6.4.3 Eigenfunctions on a Drum . . . . . . . . . . . . . . . . . . . . . . . . . 199
6.4.4 Weyl’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
6.5 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208

7 Dual Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209

7.1 The Hahn–Banach Theorem and its Consequences . . . . . . . . . . 209
7.1.1 The Hahn–Banach Lemma and Theorem . . . . . . . . . . . . 209
7.1.2 Consequences of the Hahn–Banach Theorem . . . . . . . . . 212
7.1.3 The Bidual . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
7.1.4 An Application of the Spanning Criterion . . . . . . . . . . . 214
7.2 Banach Limits, Amenable Groups, Banach–Tarski . . . . . . . . . . 217
7.2.1 Banach Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
7.2.2 Amenable Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
7.2.3 The Banach–Tarski Paradox . . . . . . . . . . . . . . . . . . . . . . . 223
7.3 The Duals of Lpµ (X) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
7.3.1 The Dual of L1µ (X) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
7.3.2 The Dual of Lpµ (X) for p > 1 . . . . . . . . . . . . . . . . . . . . . . 230
7.3.3 Riesz–Thorin Interpolation . . . . . . . . . . . . . . . . . . . . . . . . 233
7.4 Riesz Representation: The Dual of C(X) . . . . . . . . . . . . . . . . . . 239
7.4.1 Uniqueness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
7.4.2 Totally Disconnected Compact Spaces . . . . . . . . . . . . . . 240
7.4.3 Compact Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
7.4.4 Locally Compact σ-Compact Metric Spaces . . . . . . . . . . 246
7.4.5 Continuous Linear Functionals on C0 (X) . . . . . . . . . . . . 248
7.5 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252

8 Locally Convex Vector Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253

8.1 Weak Topologies and the Banach–Alaoglu Theorem . . . . . . . . . 253
8.1.1 Weak* Compactness of the Unit Ball . . . . . . . . . . . . . . . 256
8.1.2 More Properties of the Weak and Weak* Topologies . . 257
8.1.3 Analytic Functions and the Weak Topology . . . . . . . . . . 260
8.2 Applications of Weak* Compactness . . . . . . . . . . . . . . . . . . . . . . 261
8.2.1 Equidistribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
xii Contents

8.2.2 Elliptic Regularity for the Laplace Operator . . . . . . . . . 270

8.2.3 Elliptic Regularity at the Boundary . . . . . . . . . . . . . . . . . 278
8.3 Topologies on the space of bounded operators . . . . . . . . . . . . . . 290
8.4 Locally Convex Vector Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
8.5 Distributions as Generalized Functions . . . . . . . . . . . . . . . . . . . . 296
8.6 Convex Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
8.6.1 Extreme Points and the Krein–Milman Theorem . . . . . 301
8.6.2 Choquet’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
8.7 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311

9 Unitary Operators and Flows, Fourier Transform . . . . . . . . . 313

9.1 Spectral Theory of Unitary Operators . . . . . . . . . . . . . . . . . . . . . 313
9.1.1 Herglotz’s Theorem for Positive-Definite Sequences . . . 314
9.1.2 Cyclic Representations and the Spectral Theorem . . . . 316
9.1.3 Spectral Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
9.1.4 Functional Calculus for Unitary Operators . . . . . . . . . . . 323
9.1.5 An Application of Spectral Theory to Dynamics . . . . . . 326
9.2 The Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
9.2.1 The Fourier Transform on L1 (Rd ) . . . . . . . . . . . . . . . . . . 331
9.2.2 The Fourier Transform on L2 (Rd ) . . . . . . . . . . . . . . . . . . 337
9.2.3 The Fourier Transform, Smoothness, Schwartz Space . . 340
9.2.4 The Uncertainty Principle . . . . . . . . . . . . . . . . . . . . . . . . . 342
9.3 Spectral Theory of Unitary Flows . . . . . . . . . . . . . . . . . . . . . . . . 344
9.3.1 Positive-Definite Functions; Cyclic Representations . . . 344
9.3.2 The Case G = Rd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
9.3.3 Stone’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
9.4 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352

10 Locally Compact Groups, Amenability, Property (T) . . . . . 353

10.1 Haar Measure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
10.2 Amenable Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
10.2.1 Definitions and Main Theorem . . . . . . . . . . . . . . . . . . . . . 362
10.2.2 Proof of Theorem 10.15 . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
10.2.3 A More Uniform Følner Set . . . . . . . . . . . . . . . . . . . . . . . . 371
10.2.4 Further Equivalences and Properties . . . . . . . . . . . . . . . . 373
10.3 Property (T) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
10.3.1 Definitions and First Properties . . . . . . . . . . . . . . . . . . . . 375
10.3.2 Main Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
10.3.3 Proof of Každan’s Property (T), Connected Case . . . . . 378
10.3.4 Proof of Každan’s Property (T), Discrete Case . . . . . . . 384
10.3.5 Iwasawa Decomposition and Geometry of Numbers . . . 392
10.4 Highly Connected Networks: Expanders . . . . . . . . . . . . . . . . . . . 400
10.4.1 Constructing an Explicit Expander Family . . . . . . . . . . . 406
10.5 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
Contents xiii

11 Banach Algebras and the Spectrum . . . . . . . . . . . . . . . . . . . . . . . 409

11.1 The Spectrum and Spectral Radius . . . . . . . . . . . . . . . . . . . . . . . 409
11.1.1 The Geometric Series and its Consequences . . . . . . . . . . 411
11.1.2 Using Cauchy Integration . . . . . . . . . . . . . . . . . . . . . . . . . 413
11.2 C ∗ -algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
11.3 Commutative Banach Algebras and their Gelfand Duals . . . . . 418
11.3.1 Commutative Unital Banach Algebras . . . . . . . . . . . . . . 419
11.3.2 Commutative Banach Algebras without a Unit . . . . . . . 421
11.3.3 The Gelfand Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
11.3.4 The Gelfand Transform for Commutative C ∗ -algebras . 423
11.4 Locally Compact Abelian Groups . . . . . . . . . . . . . . . . . . . . . . . . . 425
11.4.1 The Pontryagin Dual . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
11.5 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431

12 Spectral Theory and Functional Calculus . . . . . . . . . . . . . . . . . 433

12.1 Definitions and Basic Lemmas . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
12.1.1 Decomposing the Spectrum . . . . . . . . . . . . . . . . . . . . . . . . 433
12.1.2 The Numerical Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
12.1.3 The Essential Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
12.2 The Spectrum of a Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
12.2.1 The Correct Upper Bound for the Summing Operator . 439
12.2.2 The Spectrum of S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
12.2.3 No Eigenvectors on the Tree . . . . . . . . . . . . . . . . . . . . . . . 442
12.3 Main Goals: The Spectral Theorem and Functional Calculus . 443
12.4 Self-Adjoint Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 446
12.4.1 Continuous Functional Calculus . . . . . . . . . . . . . . . . . . . . 447
12.4.2 Corollaries to the Continuous Functional Calculus . . . . 451
12.4.3 Spectral Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
12.4.4 The Spectral Theorem for Self-Adjoint Operators . . . . . 454
12.4.5 Consequences for Unitary Representations . . . . . . . . . . . 458
12.5 Commuting Normal Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
12.6 Spectral Measures and the Measurable Functional Calculus . . 461
12.6.1 Non-Diagonal Spectral Measures . . . . . . . . . . . . . . . . . . . 461
12.6.2 The Measurable Functional Calculus . . . . . . . . . . . . . . . . 462
12.7 Projection-Valued Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
12.8 Locally Compact Abelian Groups and Pontryagin Duality . . . 473
12.8.1 The Spectral Theorem for Unitary Representations . . . 474
12.8.2 Characters Separate Points . . . . . . . . . . . . . . . . . . . . . . . . 477
12.8.3 The Plancherel Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
12.8.4 Pontryagin Duality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
12.9 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
xiv Contents

13 Self-Adjoint and Symmetric Operators . . . . . . . . . . . . . . . . . . . 487

13.1 Examples and Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
13.2 Operators of the Form T ∗ T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
13.3 Self-Adjoint Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
13.4 Symmetric Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
13.4.1 The Friedrichs Extension . . . . . . . . . . . . . . . . . . . . . . . . . . 499
13.4.2 Cayley Transform and Deficiency Indices . . . . . . . . . . . . 500
13.5 Further Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502

14 The Prime Number Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503

14.1 Two Reformulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503
14.2 The Selberg Symmetry Formula and Banach Algebra Norm . . 507
14.2.1 Dirichlet Convolution and Möbius Inversion. . . . . . . . . . 507
14.2.2 The Selberg Symmetry Formula . . . . . . . . . . . . . . . . . . . . 509
14.2.3 Measure-Theoretic Reformulation . . . . . . . . . . . . . . . . . . 514
14.2.4 A Density Function and the Continuity Bound . . . . . . . 517
14.2.5 Mertens’ Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 518
14.2.6 Completing the Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
14.3 Non-Trivial Spectrum of the Banach Algebra . . . . . . . . . . . . . . . 523
14.4 Trivial Spectrum of the Banach Algebra . . . . . . . . . . . . . . . . . . . 524
14.5 Primes in Arithmetic Progressions . . . . . . . . . . . . . . . . . . . . . . . . 526
14.5.1 Non-Vanishing of Dirichlet L-function at 1 . . . . . . . . . . . 529

Appendix A: Set Theory and Topology . . . . . . . . . . . . . . . . . . . . . . . 537

A.1 Set Theory and the Axiom of Choice . . . . . . . . . . . . . . . . . . . . . . 537
A.2 Basic Definitions in Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 538
A.3 Inducing Topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 541
A.4 Compact Sets and Tychonoff’s Theorem . . . . . . . . . . . . . . . . . . . 545
A.5 Normal Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 547

Appendix B: Measure Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551

B.1 Basic Definitions and Measurability . . . . . . . . . . . . . . . . . . . . . . . 551
B.2 Properties of the Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554
B.3 The p-Norm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 556
B.4 Near-Continuity of Measurable Functions . . . . . . . . . . . . . . . . . . 558
B.5 Signed Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561

Hints for Selected Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563

Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
General Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 600
Chapter 1
Motivation

†
We start by discussing some seemingly disparate topics that are all intim-
ately linked to notions from functional analysis. Some of the topics have
been important motivations for the development of the theory that came
to be called functional analysis in the first place. We hope that the variety
of topics discussed in Sections 1.1–1.4 and those mentioned in Section 1.5
will help to give some insight into the central role of functional analysis in
mathematics.

1.1 From Even and Odd Functions to Group

Representations

We recall the following elementary notions of symmetry and anti-symmetry

for functions. A function f : R → R is said to be even if f (−x) = f (x) for
all x ∈ R, and odd if f (−x) = −f (x) for all x ∈ R. Every function f : R → R
can be split into an even and an odd component, since
f (x)+f (−x)
f (x) = 2 + f (x)−f
2
(−x)
. (1.1)
| {z } | {z }
the even part the odd part

Exercise 1.1. Is the decomposition of a function into odd and even parts in (1.1) unique?
That is, if f = e + o with e even and o odd, is e(x) = f (x)+f
2
(−x)
?

As one might guess, behind the definition of even and odd functions and
the decomposition in (1.1), is the group Z/2Z = {0, 1} acting on R via the
map x 7→ (−1)ℓ x for ℓ ∈ Z/2Z. Here we are using ℓ ∈ Z as a shorthand for
the coset ℓ + 2Z ∈ Z/2Z.
† Chapter 1 is atypical for this book. The reader may, and the lecturer should, skip this
chapter or return to it later, as convenient.

© Springer International Publishing AG 2017 1

M. Einsiedler, T. Ward, Functional Analysis, Spectral Theory, and Applications,
Graduate Texts in Mathematics 276, DOI 10.1007/978-3-319-58540-6_1
2 1 Motivation

In order to generalize this observation, recall that an action of a group G

.
on a set X is a map G × X → X, written (g, x) 7→ g x, with the proper-
.. . .
ties g (h x) = (gh) x for all g, h ∈ G and x ∈ X, and e x = x for all x ∈ X,
where e ∈ G is the identity element. We will sometimes write G ý
X for an
action of G on X.
Having associated (in an informal way for the moment) the decomposition
of a function into odd and even parts with the action of the group Z/2Z on R,
the notion of group action suggests many generalizations of the decomposi-
tion. For this we note that an action of a group G on a space X gives rise
to a linear action on the space of functions on X by the formula (g, f ) 7→ f g
.
where f g (x) = f (g −1 x) for all x ∈ X, g ∈ G, and functions f on X.
.

Exercise 1.2. Show that (ℓ1 , ℓ2 ) (x1 , x2 ) = (−1)ℓ1 x1 , (−1)ℓ2 x2 for ℓ1 , ℓ2 ∈ Z/2Z
and (x1 , x2 ) ∈ R2 defines an action of (Z/2Z)2 on R2 . Show that every real- or complex-
valued function on R2 can be decomposed uniquely into a sum of four functions that are
even in both variables; even in x1 and odd in x2 ; odd in x1 and even in x2 ; and odd in
both variables.

In order to define another action on R2 we write

 
cos φ − sin φ
k(φ) =
sin φ cos φ

for the matrix of anti-clockwise rotation on R2 by the angle φ ∈ R. We

also let T = R/Z be the one-dimensional circle group or 1-torus, and define
the action of φ ∈ T on R2 by the rotation by k(2πφ). Once again we will
sometimes write t as a shorthand for the coset t + Z ∈ R/Z; in particular
the interval [0, 1) may be identified with T using addition modulo 1. We note
that we can also make T into a topological group by declaring cosets to be
close if their representatives can be chosen to be close in R or equivalently
by using the metric d(t1 + Z, t2 + Z) = minn∈Z |t1 − t2 + n| for t1 , t2 ∈ R. In
studying any situation with rotational symmetry on R2 one is naturally led
to this action. What is the corresponding decomposition of functions for this
action?
Clearly one distinguished class of functions is given by the functions in-
variant under rotation. That is, functions satisfying f (v) = f (k(2πφ)v) for
all φ ∈ T. The graph of such a function is the surface obtained by rotating a
graph of a real- or complex-valued function on [0, ∞) about the z-axis.
Let us postpone the answers to the above questions and instead consider
a finite analogue to the problem. Fix some integer q > 1 and define the
group G = Z/qZ, which acts on R2 by letting ℓ + qZ ∈ G act by rotation by
the angle 2πℓ 2πℓ 2π
q using k( q ). To simplify the notation set K = k( q ) and write
the action as G×R2 ∋ (ℓ+qZ, v) 7→ K ℓ v (which is well-defined since K q = I).
This simplification in the notation helps to clarify the underlying structure,
and reflects one of the themes of functional analysis: thinking of progressively
more complicated objects (numbers, vectors, functions, operators) as ‘points’
in a larger space allows the structures to be seen more clearly.
1.1 From Even and Odd Functions to Group Representations 3

We say that a complex-valued function on R2 has weight n for this action

if f (K ℓ v) = e2πinℓ/q f (v) for every v ∈ R2 and ℓ + qZ ∈ G (or equivalently
only for ℓ = 1). We now generalize the formula (1.1) to the case of the finite
rotation group considered here. In fact, using the shorthand ζ = e2πi/q we
define for a complex-valued function f on R2 the functions
q−1
X
fn (v) = 1
q ζ −nℓ f (K ℓ v) (1.2)
ℓ=0

for every v ∈ R2 and n = 0, . . . , q − 1. Since

q−1
X q
X ′
fn (Kv) = 1
q ζ −nℓ f (K ℓ+1 v) = 1
q ζ −n(m−1) f (K ℓ v) = ζ n fn (v)
ℓ=0 m=1

R2 , we see that fn has weight n. By the geometric series formula,

for every v ∈P
we see that q−1n=0 ζ
−nℓ
equals q for ℓ = 0 and 0 for ℓ = 1, . . . , q − 1, so
q−1 q−1 q−1 X
q−1
!
X X X
−nℓ ℓ −nℓ
fn (v) = 1
q ζ f (K v) = 1
q ζ f (K ℓ v) = f (v)
n=0 n,ℓ=0 ℓ=0 n=0

for every v ∈ R2 . Therefore, f can be written as a finite sum of functions of

weight n = 0, . . . , q − 1.
Let us return now to the case of the group SO2 (R) of all rotations k(2πt)
for t ∈ T. To guess what the classes of functions should be, we note that all
of the symmetries of functions considered above can be phrased naturally in
terms of the possible continuous group homomorphisms of the acting group
to the group S1 = {z ∈ C | |z| = 1}. It is easy to show (see Exercise 1.3 for
the third, non-trivial, statement) that
(1) any homomorphism Z/2Z → S1 has the form ℓ 7→ (±1)ℓ ,
(2) any homomorphism Z/qZ → S1 has the form ℓ 7→ e2πinℓ/q = ζ nℓ for
some n ∈ {0, . . . , q − 1}, and
(3) any continuous homomorphism T → S1 has the form φ 7→ χn (φ) = e2πinφ
for some n ∈ Z, so there are infinitely many such homomorphisms, and
they are naturally parameterized by the integers.
For any topological group G, we call the continuous homomorphisms from G
to S1 the (unitary) characters of G.
Exercise 1.3. Show that any character of T has the form claimed in (3) above.

Notice that each of the characters in (1) and (2) corresponds to ex-
actly one type of function in the decompositions of functions discussed
above. Generalizing this correspondence, we turn to (3) and say that a
complex-valued function f : R2 → C has weight n (is of type n) if it sat-
isfies f (k(2πφ)v) = χn (φ)f (v) for all φ ∈ T and v ∈ R2 .
4 1 Motivation

One might now guess — and we will see in Chapter 3 that this is indeed
the case — that any reasonable function f : R2 → C can be written as a
linear combination X
f= fn (1.3)
n∈Z

where fn has weight n. However, in contrast to (1.1) this is an infinite sum,

so we are no longer talking about a purely algebraic phenomenon. The de-
composition (1.3) — its existence and its properties — lies both in algebra
and in analysis. We therefore have to become concerned both with the algeb-
raic structure and with questions of convergence. Depending on the notion of
convergence used, the class of reasonable functions turns out to vary. These
classes of reasonable functions are in fact important examples of Banach
spaces, which will be defined in Chapter 2.
The discussion above on decompositions into sums of functions of different
weights will later be part of the treatment of Fourier analysis (see Chapter 3).
For this we will initially study the mathematically simpler situation of the
action of T on T by translation, (x, y) 7→ x + y for x, y ∈ T. Adjusting
the definitions above appropriately, we say that a function f : T → C has
weight n ∈ Z if and only if f is a multiple of χn itself. We therefore seek, for
a reasonable function f : T → C, constants cn for n ∈ Z with
X
f= cn χn . (1.4)
n∈Z

The right-hand side of (1.4) is called the Fourier series of f . We will see later
that it is relatively straightforward (at least in the abstract sense) to find the
Fourier coefficients cn via the identity
Z
cn = f (x)χn (x) dx
T

for all n ∈ Z.
Fourier series arise naturally in many day-to-day applications. A string
or a wind instrument playing a note is producing a periodic pressure wave
with a certain frequency. The tone humans hear usually corresponds to this
frequency, which is called the fundamental in music theory. There are also
higher frequencies, usually integer multiples of the fundamental frequency,
appearing in the wave. These frequences are called harmonics and the ratio
between the Fourier coefficients of the harmonics and the Fourier coefficient
of the fundamental make the distinctive sound of different instruments (for
example, the flute and the clarinet) when playing the same fundamental note.
Returning to our discussion of symmetries for functions on R2 we will show
similarly that for a reasonable function f : R2 → C the function
Z
fn (v) = χn (φ)f (k(2πφ)v) dφ
T
1.2 Partial Differential Equations and the Laplace Operator 5

for n ∈ Z has weight n (compare this with (1.2)), and that (1.3) holds.
R
Exercise 1.4. Show that if the function fn (v) = T χn (φ)f (k(2πφ)v) dφ is well-defined
(say the integral exists for almost every v ∈ R2 ), then it has weight n.

To summarize, we will introduce classes of functions (which will be ex-

amples of Banach spaces), and determine whether for functions in these
classes the Fourier series (1.4) or the weight decomposition (1.3) converges,
and in what sense the convergence does or does not happen.
For functions f : R3 → C one can generalize the discussion above in
many different ways, by considering the actions of various different groups as
follows:
• Z/2Z, giving the familiar generalization of even and odd functions.
• T∼= SO2 (R) acting by rotations in the x, y-plane about the z-axis. This
gives a generalization of our discussion of functions R2 → C, and we will
be able to treat this case in a similar way to the two-dimensional case.
• SO3 (R), the full group of orientation-preserving rotations of R3 .
The last case in this list is more difficult to analyze than any of the cases dis-
cussed above. The additional complications arise since SO3 (R) is not abelian.
In fact, the group SO3 (R) is simple, and as a result there are no non-trivial
continuous homomorphisms SO3 (R) → S1 , so this cannot be used to define
classes of functions in the same way. The case of SO3 (R) requires the theory
of harmonic analysis and unitary representations of compact groups. We will
not reach these important topics here, but will lay the ground for them and
refer to the treatment in Folland [32, Ch. 5] or [26].
Although we have used actions of geometric origin to motivate the dis-
cussion above, the decompositions described hold more generally for general
linear actions of finite abelian groups on vector spaces (often called group rep-
resentations) and also for unitary representations of compact abelian groups
(with T being the main example of a compact abelian group) on Hilbert
spaces. Hilbert spaces are Banach spaces that are equipped with an inner
product, and will be introduced in Chapter 3 where we will also discuss unit-
ary representations for the first time.

1.2 Partial Differential Equations and the Laplace

Operator

There is no need to motivate the study of differential equations, as they are of

central importance across all sciences concerned with measurable quantities
that change with respect to other variables of the system studied. Even the
simplest ordinary differential equations can lead directly to the study of in-
tegral operators, which may be analyzed using tools from functional analysis.
The reader familiar with the theorem of Picard and Lindelöf on existence and
6 1 Motivation

uniqueness of solutions to certain initial value problems and its proof will not
be surprised by this connection. We refer to Section 2.4 for more on this.
However, here we would like to discuss two particular partial differential
equations. As we will see later, the mathematical background needed for
this, most of which comes from functional analysis, is much more interesting
(meaning difficult) than that needed for ordinary differential equations. One
of the objectives of this book is to make the informal discussion in this section
more formal and rigorous. We will cover this topic in Chapters 5 and 6 (apart
from a technical point, which we resolve in Section 8.2).
In both of the partial differential equations that we will discuss, we will
need to express the difference between the value of a function at a point and
its values in a neighbourhood of the point. To make the resulting equations
more amenable for study one uses an infinitesimal version of this difference,
which brings into the picture(2) the Laplace operator ∆ (also sometimes de-
noted by ∇2 ) defined by

∂2f ∂ 2f
∆f = 2 + ··· + 2 (1.5)
∂x1 ∂xd

for a smooth function f : Rd → R because of the following simple observation.

Proposition 1.5 (Laplace and neighbourhood averages). Let U ⊆ Rd

be an open set, and suppose that f : U → R is a C 2 function. Then
Z
1

lim f (y) − f (x) dy = c∆f (x)
r→0 r2 vol(Br (x)) B (x)
r

for any x ∈ U , where dy denotes integration with respect to the Lebesgue

1
measure on the r-ball Br = {y ∈ Rd | kyk < r} ⊆ Rd and c = 2(d+2) .

Proof. Suppose for simplicity of notation that x = 0, and apply Taylor

approximation to obtain
d
1 X ∂ 2f

f (y) = f (0) + f ′ (0)y + (0)yi yj + o kyk2 ,
2 i,j=1 ∂xi ∂xj

as y → 0, where f ′ (0) is the total derivative of f at 0, and we use the

notation o(·) from p. vi. Now in the integral over the r-ball Br the linear
terms cancel out due to the symmetry of the ball. The same argument applies
to the mixed quadratic terms. Thus we are left with
Z d Z
1 X ∂2f

f (y) dy = vol(Br )f (0) + (0) yi2 dy + vol(Br ) o r2 . (1.6)
Br 2 i=1 ∂x2i Br

R R
Next notice that Br
yi2 dy = Br
yj2 dy for all 1 6 i, j 6 d and
1.2 Partial Differential Equations and the Laplace Operator 7
Z Z
kyk2 dy = r2 kzk2 rd dz
Br B1

using the substitution y = rz. It follows that

Z d Z Z Z
1X 1 rd+2
yi2 dy = yj2 dy = kyk2 dy = kzk2 dz .
Br d j=1 Br d Br d B
| 1 {z }
=:C

Combining this with (1.6) gives

Z
1 1 1 1 rd+2
2
(f (y) − f (0)) dy = ∆f (0) C + o(1)
r vol(Br ) Br r2
vol(Br ) 2 d
C
= ∆f (0) + o(1).
2d vol(B1 )
| {z }
=c

For completeness, we calculate the value of c using d-dimensional spherical

coordinates. Every point z ∈ Rd is of the form z = rv for some r > 0 and

v ∈ Sd−1 = {w ∈ Rd | kwk = 1}.

Using this substitution we have

Z Z 1
1
vol(B1 ) = rd−1 dr dv = vol(Sd−1 ),
Sd−1 0 d

where the integration with respect to v uses the (d − 1)-dimensional volume

measure on the sphere Sd−1 . Similarly,
Z Z Z 1
1
C= kzk2 dz = rd+1 dr dv = vol(Sd−1 ).
B1 Sd−1 0 d + 2

Thus ✘✘
C
1
d+2✘ ✘✘
vol(Sd−1
) 1
c= = ✘ ✘✘= .
2d vol(B1 ) 2d 1
✁ d✘ ✘
vol(S d−1
) 2(d + 2)
✁


1.2.1 The Heat Equation

The heat equation describes how temperatures in a region U ⊆ Rd (represent-

ing a physical medium) evolve given an initial temperature distribution and
some prescribed behaviour of the heat at the boundary ∂U . Inside the me-
dium we expect the flow of heat to be proportional to the difference between
8 1 Motivation

the temperature at each point and the temperature in a neighbourhood of the

point. If we write u(x, t) for the temperature of the medium at the point x
at the time t, then this suggests a relationship
∂u
= |constant
{z } ∆x u, (1.7)
∂t
>0

where
∂2u ∂2u
∆x u = ∆u = 2 + ···+ 2
∂x1 ∂xd
is the Laplace operator with respect to the space variables x1 , . . . , xd only.
Equation (1.7) is called the heat equation. If we take the physical interpreta-
tion of this equation for granted, then we can use it to give heuristic explan-
ations of some of the mathematical phenomena that arise.
Suppose first that we prescribe a time-independent temperature distribu-
tion at the boundary ∂U of the medium U , and then wait until the system
has settled into thermal equilibrium. Experience (that is, physical intuition)
suggests that in the long run (as time goes to infinity) the temperature distri-
bution inside U will reach a stable (time-independent) configuration. That is,
for any prescribed boundary value b : ∂U → R we expect the heat equation
on U to have a time-independent solution. More formally, we expect there to
be a function u : U → R satisfying

∆u = 0
(1.8)
u|∂U = b.

The boundary value problem (1.8) is the Dirichlet boundary value problem,
the partial differential equation ∆u = 0 is called the Laplace equation, and its
solutions are called harmonic functions. Proving what the physical intuition
suggests, namely that the Dirichlet boundary value problem does indeed have
a (smooth) solution, will take us into the theory of Sobolev spaces. We will
prove the existence of smooth solutions for the Dirichlet boundary value
problem in Chapter 5 (and Section 8.2).
Leaving the Dirichlet problem to one side for now, we continue with the
heat equation. Motivated by the methods of linear ordinary differential equa-
tions and their initial value problems, we would like to know how we can find
other solutions to the partial differential equation while ignoring the bound-
ary values. A simple kind of solution to seek would be those with separated
variables, that is solutions of the form

u(x, t) = F (x)G(t)

with x ∈ U ⊆ Rd and t ∈ R. The heat equation would then imply that

∂u
F (x)G′ (t) = = c(∆F (x))G(t)
∂t
1.2 Partial Differential Equations and the Laplace Operator 9

and so (we may as well choose all physical constants to make c = 1) the
quotient
G′ (t) ∆F (x)
=
G(t) F (x)
is independent of x and of t, and therefore is a constant (as this is not really
a proof, we will not worry about the division by a quantity that may vanish).
In summary, u(x, t) = F (x)G(t) solves the equation

∂u
= ∆x u
∂t
if G(t) = eλt and ∆F = λF for some constant λ, which one can quickly check
(rigorously). Ignoring for the moment the values of F on the boundary ∂U , it
is easy to find functions with ∆F = λF for any λ ∈ R by using suitable expo-
nential and trigonometric functions. However, these simple-minded solutions
turn out not to be particularly useful. Only those special functions F : U → R
with 
∆F = λF inside U
F |∂U = 0
turn out to be useful in the general case. However, it is not clear that such
functions even exist, nor for which values of λ they may exist.
Suppose now that the following non-trivial result — the existence of a
basis of eigenfunctions — (which we will be able to prove in many special
cases in Chapter 6) is known for the region U ⊆ Rd .

P sufficiently nice function f : U → R can be decomposed into a

Claim. Every
sum f = n Fn of functions Fn : U → R satisfying

∆Fn = λn Fn for some λn < 0
Fn |∂U = 0.

We may then solve the partial differential equation

∂u
= ∆x u
∂t
with boundary values 
u|∂U×{t} = 0 for all t
u|U×{0} = f
using the principle of superposition to obtain the general solution
X
u(x, t) = Fn (x)eλn t . (1.9)
n

Since λn < 0 for each n > 1, the series (1.9) converges to 0 as t → ∞ if it

is absolutely convergent, in accordance with our physical intuition, since the
boundary condition states that the temperature is kept at 0 at the boundary
10 1 Motivation

of the region for all t > 0. We conclude by mentioning that the claim above
will follow from the study of the spectral theory of an operator, but the
definition of the operator involved will be somewhat indirect.

1.2.2 The Wave Equation

The wave equation describes how an elastic membrane moves. We let u(x, t)
be the vertical position of the membrane at time t above the point with
coordinate x. As the membrane has mass (and hence inertia) our assumption
is that the vertical acceleration — a second derivative of position with respect
to time t — of the membrane at time t above x will be proportional to the
difference between the position of the membrane at that point and at nearby
points. Hence we call
∂2u
= c∆x u (1.10)
∂t2
the wave equation. As in the case of the heat equation, we may as well choose
physical units to arrange that c = 1.
Once more we may argue from physical intuition that the Dirichlet bound-
ary problem for the wave equation always has a solution. Consider a wire loop
above the boundary ∂U (notice that even at this vague level we are imposing
some smoothness: our physical image of a wire loop may be very distorted
but will certainly be piecewise smooth) and imagine a soap film whose edge
is the wire. Then, after some initial oscillations,† we expect the soap film to
stabilize, giving a solution to the Dirichlet boundary value problem in (1.8)
defined by the shape of the wire.
In this context, what is the meaning of eigenfunctions of the Laplace oper-
ator that vanish on the boundary? To see this, imagine a drum whose skin has
the shape U so that the vibrating membrane is fixed along the boundary ∂U ,
which is simply a flat loop. Suppose now that F : U → R satisfies

∆F = λF in U
F |∂U = 0
√
for some λ < 0, then we see that u(x, t) = F (x) cos( −λt) satisfies

∂2 √ √
2
u(x, t) = F (x) (−( −λ)2 ) cos( −λt)
∂t | {z }
=λ
√ √
= λF (x) cos( −λt) = ∆x (F cos( −λt))

and hence solves the wave equation. In other words, if we start the drum
at time t = 0 with the prescribed shape given by the function F , then the
†In the real world there would also be a friction term, and the model for this is a modified
wave equation (which we will not discuss further).
1.2 Partial Differential Equations and the Laplace Operator 11

drum will produce a pure tone† of frequency √2π −λ

. This also sheds some
light on which values of λ appear in the claim on p. 9 – namely those that
correspond to the pure frequencies that the drum can produce. For a one-
dimensional drum (that is, a string) these frequencies are easy to understand
(see also Exercise 1.7). However, even for two-dimensional drums the precise
eigenvalues remain mysterious. We will nonetheless be able to count these
shape-specific eigenvalues asymptotically using the method of Weyl from 1911
(see Section 6.4).
Exercise 1.6. Assume that U satisfies the basis of eigenfunctions claim from p. 9, and
that the Dirichlet boundary value problem always has a solution on U .
(a) Combine the above discussions to produce a general procedure to solve the boundary
value problem (no rigorous proof is expected, but find the places that lack rigour)
(
∂u
= ∆u in U × [0, ∞);
∂t
u|∂U ×{t} = b; u|U ×{0} = f.

(b) Repeat (a) for the wave equation.

Exercise 1.7. For a circular string vibrating in one dimension — the wave equation over T
— the basis of eigenfunctions claim is precisely the claim that every nice function can be
represented by its Fourier series. Assuming that this holds, show the basis of eigenfunctions
claim for the domain U = (0, 1) ⊆ R. This relates to the wave equation for the clamped
vibrating string on [0, 1], that is, to the boundary conditions y(0) = y(1) = 0. (In fact the
eigenfunctions are given by x 7→ sin(πnx) with n = 1, 2 . . . ; no rigorous proof is expected,
but explore the connection.)

1.2.3 The Mantegna Fresco

An illustration of how some of the ideas discussed above link together will
be seen in Section 6.4.3, where we discuss eigenfunctions of the Laplacian
on a disk. Here the circular symmetry is exploited as in Section 1.1, and the
eigenfunctions may be used to decompose functions.
A remarkable application of these ideas was made to the problem of recon-
structing a bombed fresco by Andrea Mantegna in a church in Padua. The
damage resulted in the fresco being broken into approximately 88, 000 small
pieces which needed to be reassembled using a black and white photograph;
we refer to Fornasier and Toniolo [35] for the detailed description of how cir-
cular harmonics were used to render the computation required practicable.
The partially reconstructed coloured image was then used to build a coloured
image of the entire fresco.
† This preferred frequency for certain physical objects is part of the phenomena of reson-
ance, and the design of large structures like buildings or bridges tries to prevent resonances
that may lead to reinforcement of oscillations by wind, for example.
so as success

is we

nominally and upward

we in

59 one true

duty these have

of and Revolution

Lucas

our more goodness

correct

the from of

comprise a

Holy That

as her

the larger

by

way

down cannot possesses

own lawful

the that

of as

every es

Almighty

make broadcast author

Passions then argument

great power

the
one nomen

A to

the about Legislature

be city ought

the the requires

ago date
As be

despotism energies the

with 517

not

feature
as expressions

s huge tells

bad Salem

then American

their ultimately the

Patrick luminous with

been doubt

contract Paghano monastery

had

permanence a
itur it promoters

as China

and he was

Phoenician these www

the

Nearly a

to

famines having than

magna

should That patience

practice

ceiling wall

at Sixth

000 and and

trees

conventione will committee

imperfect

of

what

nocturno to

were it him

1878 for

and A
the of of

desig Father

Par out

the We

widely
room

banners

cavsal He to

former no single

cold which

Syria coming in

the genuine now

where Of blocked

We

of these of
one in and

concordiae concern he

Savage least as

looking this

journey and

his and
sub says

the undertakes

or messengers cultivators

the

as
faith because to

tasting other

the observes has

are in when

Prefixed interests without

St arises

Archipelago

vico Notices Kingdom

s fraud arrival

the the fear

and

With peramanter fame

Brun if

of

Liberty any
no

of crossing

most our

may

local

profited

taken

the disease

well
at picked

three

years Catholics

is

the wayfarer

you

of optima
may communication

everywhere

the

admirers Venerable proceed

resemblance

by

vault his

Though

nominavit duty the

background lot
autem includes

of all the

of charity

the loads

principle anger Paris

more a French

this

heart the
had newspapers virorum

war

the Aztec

a catlioUcus

go
all

quality

loth

that which but

will But brief

Catholics

they

shaped sensitive there

treasure is

Buvre me supposed

of the

its loped in
S fifteen

being in

recollections prelates was

s IT

book

Perriu The

A litteris or

in

to inscription to

path
action

and River

p held need

are The benediction

walls use
of tremblings and

such name through

Evangelium a

of

e
for the

coal

narrated necessary

the he Ukassa

has Catherine

north socalled a
unknown w

to

such

the

of

the

the

article a
St cannot

tree

The Eepealer

all and

and is
antiquam

Future

The Home seeping

from

so the

the four She

the
It 12 parents

Roman

bonds the

which are

all new

Pj
the

local Jesuit

it in from

with be looked

of

the stability example

was see w

lyrics Taoism

room creation which

retreat which
recoverable the

the rise fuller

after earth nobler

have worth

who upon attend

applicable

Tablet small in

Canada an any
the in

in reservoirs justice

knowing appreciation

Centuries I
author the wish

to Introduced Dives

Faith verses their

mistakes crypt

peculiar of

Will

sordid Head bones

the by

that
said and Dr

could valleys

classifying E

he artistically to

auspice
to

therefore

and

of

But the climbs

at nuper
We

After the

at

the

asserting

explore

the

of public

quite I vast

17 To
the Apost entrance

promotion and

other by the

he responsible

have

and of exchange

over

quisque
remaining

Jericho

and renounced chap

He efforts scores

arena

Ecclesia

presumption poor of
it concession

the It Tories

exigua to

have thyself others

to
as The

headache enlightenment

strict but

second assasins with

London the Paul

certain recte
for

tact

time life

faith glass

organ

but
feet of

employes four point

industria

the personal commenting

acquainted

expedient in effect
learned perhaps keep

their

comparatively

speaks hardly erected

State

Besides

Low death

writer own
him

it

prudent

attempts only irrigation

ipsam is

cum story Individuation

himfh
shall and the

triplicem

was in the

importance

left promontorium

In quit

in Quare
In

Commons

country their the

and

Music
the incisive murder

to

are civilization at

national

The The

further anything

the mortal 605

it limited

and deluge

We
of omnipotentem

the Catholics

the

and preponderance according

in up difficult

had in voluntatem

to in

not time as
evil

were Curry burst

on Ecclesiastical

theories in com

connection inflicted authority

the own Water

if and the

well unwilling
has

in

there

develop

ac contendere and

one Words

the rests about

in did
another

was

plundering of

297 opinion

mortally

but
Eeach enough

about inaugurated a

party after

briefly calamities

other indicates
off dissertations

some the

chamber

are obtinere

the tutorship the

Greg Vols extolling

whose

research not

has Vere

crowd

Tablet of hence

against

force more

this redemption us

reveals water In
eruption with

above many find

to as

Brun the

and

and commencement excitement

but Therefore

it Rome

assentiens

in

for

appears than estimate

rules Seminary the

St

a altercations
ago life He

the repeats distinct

how

of

is soul called

small of

delight location so

the The

enmity amid woman

to It a

a to

An land

longer access established

statements

we
still expeditionibus in

they captivity their

a song

one

In who

living
the

has

neighbouring not waged

pent ed

we i
or the

dismissed

indications clear

in exit

and least When

in more him
villagers The combat

bathing called to

great

type

Jinb to

workshop lost will

he the be
remote one the

perpessus

century a

montem

active

patriarchal plan

abeunt everything the

husbandmen
sights diocese

King twenty every

prohibited

rests abundant from

needed some foot

after such the

mind

to the

being Church solemn

of

ready

overwhelmed extensively country

of

thus sometimes we

rock

province account

assistance 7
Geok

Budapesthinum could Frederick

that

the consecuta

discussion and the

only Speaking

the

suspicion one as
the

subject its

are ago

the isolated closed

Tuileries be be

system
swept

these make conduct

and

proofs

Holding

hostility fertile means

Drummond name

PCs nature s

of Gladstone

ex to
from intellect for

is the to

to

bow 99 causticity

The s by

blotted fires

the be

in thus incidents

Mr
of the

us

for an

identified deluge Barada

profitably

quickly of referant

are

room were other

between housed
publishing

to better

dusty autre

Franchise on

and of does

explanations 1756 cut

feature and fetters

of

Switzerland members

few King

consult fact

ject Hildebrand town

the
the

trade

that out politics

be with and

head over that

Sydney

deciding
palam

to beginning

in come for

coming

stated a the
clear these Rosmini

Holy building

will with feelings

work

the and Persia

1 His

Person

the ruined

insignificant

images
circle Eugene

below less In

Heresies

of that

wood and about

ground

work jester followers

and

are XIII probably

discipline it it
Dunning

publications contribute

be he

IT probably clear

the Book
art

Book where the

Church which

no loquendi

distinguished
particularly

to

and

us freed population

question occupied studies

sound were

where the

two

chamber
words

at

passage co

do

except A

presence

may cargoes be
of

efficere

a tokens

Irish only think

of E

and

playful
for of to

are

him everything led

and Such door

often of

Ward telephonic realized

sufficient means by

the the

to a
counsels and single

of

be made through

or

principle

right
he the an

cistern

instead us

neither generous

strife a

strike

to
him toil

short it

Yet

of the the

with

it Greeks

is

face bread the

then present were

upon from only

ointments

victoria to

of

that in

career

those

that to for

The

to anticipated
the Strong lighters

had

Miss

t the is

boy the oil

prosaic declaration is

essence their the

of or Monologium
drunk in

a difficulties

reformers 200

followed character

Gothic from

first

assault permovet

thereby

which
school than works

on Rosmini cher

back partial

concludes taken

home fine us
They the

that found

collection Count can

on

need distance Sons

considering at America
cent

These

with j author

a and

engineer

sive

many has
the

them mentione

of household necessarily

person

Paul are was

properly into

the of Ney

always not been

the the credit

spot principles

discernment recently

rulers transitus Marcon

nowhere inductive

oil the
eventually produced Speeches

than is is

belief of

these

of cleverest

so

the Nemthur in

with against which

his and the

you famous

of

more

to

light

were spread format

They more

Let nearly when

of from Perrymill

in Story

at opportunity

a permission town

navy

on
says money

qualifications words

and Drake

national the

and

stormy
St rolls

Lear

degree

has a Company

Cartesian

etiam
from

thou his

suddenly

presumably In

the and songs

to

server a only

The The occurred

end
hominum Mass

nothin

trying of races

thus millions

AugustiAn food

worth

in

volume
in Christian

while

present volume Lao

resident run

gambling with

a bound

Nihilist

as sacred ir
search really

the experienced so

acquainted ornato inordinate

of

of

subject

of

of the

direct
or If of

operae

the

upon

in is

as heads
as systems

is trials bodies

of

text one after

the

cargo voyage do

and for on

oil expel

however however

A Not way
room

proceedings his

in

these smoke

The me

which did agency

the

and

Florido the
suddenly

as perferre Greek

manage usque responsible

secure

degradation 2

duties the
Darcy be over

implies of baby

votes

is sufficiently

defined sense

Gates

gave

admitted the

collect it besides

is that
et trees

survive aut

uplifted

A shallow of

sets of placed
of find cleared

the

epochs ceremony of

his of

so Taoism

these the

front

be est

in

Government
to with expectation

as

an

definite but Ireland

s Paul
exclusion

Canonicorum to

Saint

Burns

collustravit in to

Longfelloiv fanned for

beauty shoes

toujours from Commons

of an except

the
the

of Room was

chapters latter bookish

It

is time enough

are glimpse
may most the

uius

is

had

unjust at

the 72 situated