Instant Ebook Access, One Click Away – Begin at ebookgate.

com

Introduction to Mathematical Analysis 2013th

Edition Igor Kriz

https://ebookgate.com/product/introduction-to-mathematical-
analysis-2013th-edition-igor-kriz/

OR CLICK BUTTON

DOWLOAD EBOOK

Get Instant Ebook Downloads – Browse at https://ebookgate.com

Click here to visit ebookgate.com and download ebook now
Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Introduction to mathematical logic Fifth Edition Mendelson

https://ebookgate.com/product/introduction-to-mathematical-logic-
fifth-edition-mendelson/

ebookgate.com

Introduction to Mathematical Philosophy 1st Edition

Bertrand Russell

https://ebookgate.com/product/introduction-to-mathematical-
philosophy-1st-edition-bertrand-russell/

ebookgate.com

Introduction to Mathematical Logic Sixth Edition Elliott

Mendelson

https://ebookgate.com/product/introduction-to-mathematical-logic-
sixth-edition-elliott-mendelson/

ebookgate.com

An Introduction to Mathematical Cosmology 2ed Edition

Islam J.N.

https://ebookgate.com/product/an-introduction-to-mathematical-
cosmology-2ed-edition-islam-j-n/

ebookgate.com
A Mathematical Introduction to Compressive Sensing 1st
Edition Simon Foucart

https://ebookgate.com/product/a-mathematical-introduction-to-
compressive-sensing-1st-edition-simon-foucart/

ebookgate.com

An introduction to mathematical cosmology 2nd ed Edition

Jamal N Islam

https://ebookgate.com/product/an-introduction-to-mathematical-
cosmology-2nd-ed-edition-jamal-n-islam/

ebookgate.com

Mathematical Analysis Second Edition Tom M. Apostol

https://ebookgate.com/product/mathematical-analysis-second-edition-
tom-m-apostol/

ebookgate.com

Introduction to Mathematical Modeling and Chaotic Dynamics

1st Edition Ranjit Kumar Upadhyay

https://ebookgate.com/product/introduction-to-mathematical-modeling-
and-chaotic-dynamics-1st-edition-ranjit-kumar-upadhyay/

ebookgate.com

Plato s Problem An Introduction to Mathematical Platonism

1st Edition Marco Panza

https://ebookgate.com/product/plato-s-problem-an-introduction-to-
mathematical-platonism-1st-edition-marco-panza/

ebookgate.com
Igor Kriz
Aleš Pultr

Introduction to
Mathematical
Analysis
Igor Kriz
Aleš Pultr

Introduction
to Mathematical
Analysis
Igor Kriz Aleš Pultr
Department of Mathematics Department of Applied Mathematics (KAM)
University of Michigan Faculty of Mathematics and Physics
Ann Arbor, MI Charles University
USA Prague
Czech Republic

ISBN 978-3-0348-0635-0 ISBN 978-3-0348-0636-7 (eBook)

DOI 10.1007/978-3-0348-0636-7
Springer Basel Heidelberg New York Dordrecht London
Library of Congress Control Number: 2013941992

© Springer Basel 2013

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. Exempted from this legal reservation are brief excerpts in connection
with reviews or scholarly analysis or material supplied specifically for the purpose of being entered
and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of
this publication or parts thereof is permitted only under the provisions of the Copyright Law of the
Publisher’s location, in its current version, and permission for use must always be obtained from Springer.
Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violations
are liable to prosecution under the respective Copyright Law.
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.
While the advice and information in this book are believed to be true and accurate at the date of
publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for
any errors or omissions that may be made. The publisher makes no warranty, express or implied, with
respect to the material contained herein.

Printed on acid-free paper

Springer Basel AG is part of Springer Science+Business Media (www.birkhauser-science.com)

To Sophie

To Jitka
Preface

This book is a result of a long-term project which originated in courses we taught

to undergraduate students who specialize in mathematics. These students had ı-"
calculus before, but there did not seem to be a suitable comprehensive textbook for
a follow-up course in analysis.
We wanted to write such a textbook based on our courses, but that was not
the only goal. Teaching bright students is about introducing them to mathematics.
Therefore, we wanted to write a book which the students may want to keep after the
course is over, and which could serve them as a bridge to higher mathematics. Such
a book would necessarily exceed the scope of their courses.
We start with standard material of second year analysis: multi-variable differ-
ential calculus, Lebesgue integration, ordinary differential equations and vector
calculus. What makes all this go smoothly is that we introduce some basic concepts
of point set topology first. Since our aim is to be completely rigorous and as self-
contained as possible, we also include a Preliminaries chapter on the basic topic
of one-variable calculus, and two Appendices on the necessary concepts of linear
algebra. This pretty much comprises the first part of our book.
With the foundations covered, it is possible to venture much further. The common
theme of the second part of our book is the interplay between analysis and geometry.
After a second installment of point set topology, we are quickly able to introduce
complex analysis, and after some multi-linear algebra, also manifolds, differential
forms and the general Stokes Theorem. The methods of manifolds and complex
analysis combine in a treatment of Riemann surfaces. Basic methods of the calculus
of variations are applied to a theory of geodesics, which in turn leads to basic tensor
calculus and Riemannian geometry. Finally, infinite-dimensional spaces, which have
already made an appearance in multiple places throughout the text, are treated more
systematically in a chapter on the basic concepts of functional analysis, and another
on a few of its applications.
The total amount of material in this book cannot be covered in any single year
course. An instructor of a course based on this book should probably aim for
covering the first part, and take his or her picks in the second part. As already
mentioned, we hope to motivate the student to hold on to their textbook, and use
it for further study in years to come. They will eventually get to more advanced
books in analysis and beyond, but here they can get, relatively quickly, their first
glimpse of a big picture.

vii
viii Preface

Because of this, the aim of our book is not limited to undergraduate students. This
text may equally well serve a graduate student or a mathematician at any career
stage who would like a quick source or reference on basic topics of analysis. A
scientist (for example in physics or chemistry) who may have always been using
analysis in their work, can use this book to go back and fill in the rigorous details
and mathematical foundations. Finally, an instructor of analysis, even if not using
this book as a textbook, may want to use it as a reference for those pesky proofs
which usually get skipped in most courses: we do quite a few of them.

Ann Arbor, USA Igor Kriz

Prague 1, Czech Republic Aleš Pultr
Contents

Part I A Rigorous Approach to Advanced Calculus

1 Preliminaries . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1 Real and complex numbers .. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Convergent and Cauchy sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3 Continuous functions . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4 Derivatives and the Mean Value Theorem .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
5 Uniform convergence .. . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6 Series. Series of functions .. . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
7 Power series . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
8 A few facts about the Riemann integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
9 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2 Metric and Topological Spaces I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
1 Basics. . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2 Subspaces and products . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3 Some topological concepts . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4 First remarks on topology . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5 Connected spaces . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
6 Compact metric spaces . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
7 Completeness . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
8 Uniform convergence of sequences of functions.
Application: Tietze’s Theorems .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
9 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3 Multivariable Differential Calculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
1 Real and vector functions of several variables . . . . . . . . . . . . . . . . . . . . . . . . . 65
2 Partial derivatives. Defining the existence of a total differential . . . . . . 66
3 Composition of functions and the chain rule . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4 Partial derivatives of higher order. Interchangeability . . . . . . . . . . . . . . . . . 74
5 The Implicit Functions Theorem I: The case of a single equation . . . . 77
6 The Implicit Functions Theorem II: The case of several equations . . . 81
7 An easy application: regular mappings and the Inverse
Function Theorem . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

ix
x Contents

8 Taylor’s Theorem, Local Extremes and Extremes

with Constraints. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
9 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4 Integration I: Multivariable Riemann Integral and Basic
Ideas Toward the Lebesgue Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
1 Riemann integral on an n-dimensional interval . . . . . . . . . . . . . . . . . . . . . . . . 97
2 Continuous functions are Riemann integrable.. . . . . . . . . . . . . . . . . . . . . . . . . 100
3 Fubini’s Theorem in the continuous case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4 Uniform convergence and Dini’s Theorem .. . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5 Preparing for an extension of the Riemann integral .. . . . . . . . . . . . . . . . . . . 105
6 A modest extension .. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7 A definition of the Lebesgue integral and an important lemma . . . . . . . 109
8 Sets of measure zero; the concept of “almost everywhere” .. . . . . . . . . . . 113
9 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5 Integration II: Measurable Functions, Measure
and the Techniques of Lebesgue Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
1 Lebesgue’s Theorems.. . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
2 The class ƒ (measurable functions) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
3 The Lebesgue measure . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4 The integral over a set . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5 Parameters.. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6 Fubini’s Theorem . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
7 The Substitution Theorem .. . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
8 Hölder’s inequality, Minkowski’s inequality and Lp -spaces . . . . . . . . . . 135
9 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6 Systems of Ordinary Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
1 The problem.. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
2 Converting a system of ODE’s to a system of integral equations . . . . . 147
3 The Lipschitz property and a solution of the integral equation .. . . . . . . 149
4 Existence and uniqueness of a solution of an ODE system . . . . . . . . . . . . 151
5 Stability of solutions .. . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
6 A few special differential equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
7 General substitution, symmetry and infinitesimal
symmetry of a differential equation .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
8 Symmetry and separation of variables .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
9 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
7 Systems of Linear Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
1 The definition and the existence theorem for a system
of linear differential equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
2 Spaces of solutions . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
3 Variation of constants .. . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
4 A Linear differential equation of nth order
with constant coefficients.. . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Contents xi

5 Systems of LDE with constant coefficients. An application

of Jordan’s Theorem .. . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
6 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
8 Line Integrals and Green’s Theorem .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
1 Curves and line integrals . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
2 Line integrals of the first kind (D according to length) .. . . . . . . . . . . . . . . 197
3 Line integrals of the second kind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
4 The complex line integral . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
5 Green’s Theorem.. . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
6 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209

Part II Analysis and Geometry

9 Metric and Topological Spaces II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213

1 Separable and totally bounded metric spaces. . . . . . . . . . . . . . . . . . . . . . . . . . . 213
2 More on compact spaces . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
3 Baire’s Category Theorem . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
4 Completion .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
5 More on topological spaces: Separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
6 The space of continuous functions revisited:
The Arzelà-Ascoli Theorem and the Stone-Weierstrass Theorem.. . . . 229
7 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
10 Complex Analysis I: Basic Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
1 The derivative of a complex function. Cauchy-Riemann
conditions . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
2 From the complex line integral to primitive functions . . . . . . . . . . . . . . . . . 243
3 Cauchy’s formula . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
4 Taylor’s formula, power series, and a uniqueness theorem . . . . . . . . . . . . 248
5 Applications: Liouville’s Theorem, the Fundamental
Theorem of Algebra and a remark on conformal maps . . . . . . . . . . . . . . . . 252
6 Laurent series, isolated singularities and the Residue Theorem .. . . . . . 254
7 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
11 Multilinear Algebra . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
1 Hom and dual vector spaces . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
2 Multilinear maps and the tensor product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
3 The exterior (Grassmann) algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
4 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
12 Smooth Manifolds, Differential Forms and Stokes’ Theorem . . . . . . . . . 287
1 Smooth manifolds . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
2 Tangent vectors, vector fields and differential forms.. . . . . . . . . . . . . . . . . . 292
3 The exterior derivative and integration of differential forms . . . . . . . . . . 298
4 Integration of differential forms and Stokes’ Theorem . . . . . . . . . . . . . . . . 301
5 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
xii Contents

13 Complex Analysis II: Further Topics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311

1 The Riemann Mapping Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
2 Holomorphic isomorphisms of disks onto polygons
and the Schwartz-Christoffel formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
3 Riemann surfaces, coverings and complex differential forms . . . . . . . . . 321
4 The universal covering and multi-valued functions . . . . . . . . . . . . . . . . . . . . 332
5 Complex analysis beyond holomorphic functions . . . . . . . . . . . . . . . . . . . . . 340
6 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
14 Calculus of Variations and the Geodesic Equation . . . . . . . . . . . . . . . . . . . . . 349
1 The basic problem of the calculus of variations,
and the Euler-Lagrange equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
2 A few special cases and examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
3 The geodesic equation .. . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
4 The geometry of geodesics .. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
5 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
15 Tensor Calculus and Riemannian Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
1 Tensor calculus. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368
2 Affine connections .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
3 Tensors associated with an affine connection: torsion
and curvature .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
4 Riemann manifolds . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
5 Riemann surfaces and surfaces with Riemann metric.. . . . . . . . . . . . . . . . . 381
6 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390
16 Banach and Hilbert Spaces: Elements of Functional Analysis . . . . . . . . 393
1 Banach and Hilbert spaces . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
2 Uniformly convex Banach spaces .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
3 Orthogonal complements and continuous linear forms . . . . . . . . . . . . . . . . 397
4 Infinite sums in a Hilbert space and Hilbert bases . . . . . . . . . . . . . . . . . . . . . 402
5 The Hahn-Banach Theorem ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
6 Dual Banach spaces and reflexivity .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411
7 The duality of Lp -spaces . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
8 Images of Banach spaces under bounded linear maps . . . . . . . . . . . . . . . . . 419
9 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
17 A Few Applications of Hilbert Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427
1 Some preliminaries: Integration by a measure . . . . . . . . . . . . . . . . . . . . . . . . . 427
p
2 The spaces L .X; C/ and the Radon-Nikodym Theorem . . . . . . . . . . . . . 432
3 Application: The Fundamental Theorem of (Lebesgue) Calculus.. . . . 435
4 Fourier series and the discrete Fourier transformation .. . . . . . . . . . . . . . . . 440
5 The continuous Fourier transformation .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
6 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
Contents xiii

A Linear Algebra I: Vector Spaces .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451

1 Vector spaces and subspaces .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
2 Linear combinations, linear independence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
3 Basis and dimension .. . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
4 Inner products and orthogonality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
5 Linear mappings . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464
6 Congruences and quotients .. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
7 Matrices and linear mappings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
8 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474

B Linear Algebra II: More about Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477

1 Transforming a matrix. Rank .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
2 Systems of linear equations . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
3 Determinants .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
4 More about determinants . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489
5 The Jordan canonical form of a matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
6 Exercises . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501

Index of Symbols . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505
Introduction

The main purpose of this introduction is to tell the reader what to expect while
reading this book, and to give advice on how to read it. We assume the reader to be
acquainted with the basics of differential and integral calculus in one variable, as
traditionally covered in the first year of study. Nevertheless, we include, for the
reader’s convenience, in Chapter 1, a few pivotal theoretical points of analysis
in one variable: continuity, derivatives, convergence of sequences and series of
functions, the Mean Value Theorem, Taylor expansion, and the single-valued
Riemann integral. The purpose of including this material is two-fold. First, we
would like this text to be as self-contained as possible: we wish to spare the reader a
tedious search, in another text, for an elementary fact he or she may have forgotten.
The second, and perhaps more important reason, is to focus attention on facts
of elementary differential and integral calculus that have deeper aspects, and are
fundamental to more advanced topics. In connection with this, we also review in the
exercises to Chapter 1 definitions of elementary functions and proofs, from the first
principles, of their properties needed later. What we omit at this stage is a proof of
the existence of real numbers; the reader probably knows it from elsewhere, but if
not there will be an opportunity to come back and do it as an exercise to Chapter 9.
An entirely different prerequisite is linear algebra. While not a part of mathemat-
ical analysis in the narrowest sense, it contains many necessary techniques. In fact,
differential calculus (in particular in more than one variable) can be without much
exaggeration understood as the study of linear approximations of more general
mappings, and a basic knowledge in dealing with the linear case is indispensable.
The reader’s skills in these topics (determinants, linear equations, operations with
matrices, and others) may determine to a considerable degree his or her success with
a large part of this book. Because of this, we feel it is appropriate to include linear
algebra in this text as a reference. In order not to slow down the narrative, we do
so in two appendices: Appendix A for more theoretical topics such as vector spaces
and linear mappings, and Appendix B for more computational questions regarding
matrices, culminating with a treatment of the Jordan canonical form.
Let us turn to the main body of this book. It is divided into two parts. One
of our main goals is to present a rigorous treatment of the traditional topics
of advanced calculus: multivariable differentiation, (Lebesgue) integration, and
differential equations. All this is covered by Part I, including basic facts about line
integrals and Green’s Theorem.

xv
xvi Introduction

In Part II we use the techniques developed in Part I to approach phenomena of

geometrical nature which the reader may already have encountered without proof,
or will certainly encounter in further studies. Using the tools developed they can be
probed into considerable depth without too much further difficulty.
Part I. We think it essential to start rigorous advanced calculus with the basic
notions of (at least metric) topology. Concepts such as neighborhood, open set,
closure and convergence viewed narrowly just in the context of the Euclidean space
Rn do not give a satisfactory picture of what is going on (and besides, would not be
sufficient for what will come later). In Chapter 2, we discuss these concepts first in
the context of metric spaces. This generality is, strictly speaking, already sufficient
for most of our purposes. Yet, it is useful to learn about the more general topological
spaces to be able to distinguish what really depends on metric and what does not.
For this reason, our treatment of space in this chapter is an interweaving narrative
of metric and topological spaces with the goal of presenting an adequate general
outlook. We stick here, however, to the simpler facts and concepts needed in the
nearest chapters (the more advanced topics on spaces are postponed to Chapter 9)
and the reader will certainly not find this chapter hard.
With the basic knowledge of metric topology we are ready for multivariable
differential calculus. This is covered in Chapter 3. We start with the basic notions
of partial derivative and total differential, and chain rule. Emphasizing the role of
the total differential as a linear map is key to more coordinate-free approaches
to analysis, vital when investigating manifolds (later in Chapter 12). Next, we
prove from first principles the Implicit Function Theorem. This is the first more
complicated analytic proof in our text; the reader is advised to pay detailed attention
to this material, and certainly not to skip it, as it is a good model of what a proof in
analysis looks and feels like. The chapter is concluded with material related to the
multivariable version of Taylor’s Theorem, and to calculating extremes and saddles
of multivariable functions.
The following two chapters are devoted to integration. In Chapter 4 we start with
the multivariable Riemann integral over a product of intervals. It becomes clear very
quickly, however, that a more versatile theory is needed. For example, we want to
take integrals of unbounded functions, or integrals over more general types of sets.
Or, we would like to know when we can take limits or derivatives behind the integral
sign. We would like to understand precisely why “the boundary does not matter”
when taking a multivariable integral, and how and why we can change variables
in a non-linear way. All this leads to the concept of Lebesgue integral, which is
somewhat notorious for being time-consuming because of the abstract concepts it
entails. There is, however a way around that. A method of P.J. Daniell (going back
to 1918 and unjustly neglected for decades) allows a straightforward introduction
of the Lebesgue integral starting with monotone limits of Riemann integrals of
continuous functions with compact support. The necessary technical theorems can
be proved very quickly and we present them in the second half of Chapter 4. In
Chapter 5, we go on to present the more technical aspects of the Lebesgue integral.
We explain how to take limits and derivatives behind the integral sign, prove Fubini’s
Theorem, define the Lebesgue measure and prove its basic properties. Further, we
Introduction xvii

introduce Borel sets and prove criteria of measurability. We present a rigorous proof
of a multvariable substitution theorem. Finally, we introduce Lp -spaces: while this
may seem like an early place, we will have enough integration theory at this point to
do so, and to prove their basic properties. This is useful, as the Lp -spaces often occur
throughout analysis (for example, in this book, we will use them in Chapters 13
and 15 in proving the existence of a complex structure on an oriented surface with
a Riemann metric.) We will return to the study of Lp -spaces in Chapter 16, where
they provide the most basic examples in functional analysis.
Next, having covered differentiation and integration, we turn to differential
equations. We restrict our attention to the ordinary differential equations (ODEs),
as partial differential equations have quite a different flavor and constitute a vast
field of their own, far beyond a general course in analysis (even an advanced one).
For a text on partial differential equations, we refer the reader, for example, to [5].
Chapter 6 on (general) ordinary differential equations is in fact independent of
Chapters 4 and 5 and uses only the material of Chapters 1, 2 and 3. We introduce
the concept of a Lipschitz function and prove the local existence and uniqueness
theorem for the systems of ODEs (the Picard-Lindelöf Theorem). We also discuss
stability of solutions and differentiation with respect to parameters. Further, we
discuss the basic method for separation of variables, and finally discuss global
and infinitesimal symmetries of systems of ODEs (thus motivating further study of
vector fields); also, we explain how the methods of separation of variables discussed
earlier are related to symmetries of the system.
Chapter 7 covers some aspects of linear differential equations (LDEs). The global
existence theorem is proved, and the affine set of solutions of a linear system is
discussed. We show how to use the Wronskian for recognizing a fundamental system
( basis) of the space of solutions of a homogeneous system of LDEs, and how to
get solutions of a non-homogeneous system from the homogeneous one using the
variation of constants. Also, we present a method of solving systems of LDEs with
constant coefficients, easier in the case of a single higher order LDE, and requiring
the Jordan canonical form of a matrix from Appendix B in the harder general case.
Chapter 8, concluding Part I, treats parametric curves, line integrals of the first
and second kind and the complex line integral. At the end we prove Green’s
Theorem, which we will need when dealing with complex derivatives, but which
is also an elementary warm-up for the general Stokes’s Theorem.
Part II. Now our perspective changes. The traditional items of advanced calculus
have been mostly covered and we turn to topics interesting from the point of view
of geometry.
To proceed, perhaps by now not surprisingly, we need another installment of
topological foundations. This is done in Chapter 9, presenting more material on
topological spaces (separability, compactness, separation axioms and the Urysohn
Theorem) as well as on metric spaces (completion, Baire’s Category Theorem). In
the last section we prove the Stone-Weierstrass Theorem providing a remarkably
general method to obtain useful dense sets in spaces of functions, and the Arzelà-
Ascoli Theorem, which greatly clarifies the meaning of uniform convergence, and
will be useful in Chapter 10 when proving the Riemann Mapping Theorem.
xviii Introduction

Next, in Chapter 10 we introduce the basic methods of complex analysis. The

fundamental facts can be derived almost immediately from the complex line integral
and Green’s Theorem of Chapter 8. The conclusions, however, are powerful and
surprising. Unlike differentiable functions of a real variable, complex functions
with a complex derivative (holomorphic functions) are much more rigid. They
are determined, for example, by their values on a convergent sequence of points,
and the existence of a derivative automatically implies the existence of derivatives
of all orders; on the other hand, “geometrically very smooth” functions may not
have a complex derivative. Thus, our view of the differential calculus as we know
it from the real case is turned upside down. Yet, complex analysis has important
real applications, such as for instance the explanation of the convergence properties
of a Taylor series. Other applications presented are the Fundamental Theorem of
Algebra, and an important geometric one, the Jordan Curve Theorem. We then
go on to cover other basic methods of complex analysis, such as Laurent series,
the classification of isolated singularities, the Residue Theorem and the Argument
Principle, which has several interesting applications, including the Open Mapping
Theorem.
Next, we also have to upgrade our knowledge of linear algebra; more specifically,
we must get acquainted with the techniques of multilinear algebra. This is done in
Chapter 11, which includes dual vector spaces, tensor products, and the exterior
(Grassmann) algebra. Thus equipped, we can now study calculus on manifolds.
This is done in Chapter 12. We define smooth manifolds, tangent vectors, vector
fields, and differential forms. Further, we present the exterior derivative, de Rham
complex, and the de Rham cohomology. A general form of Stokes’s Theorem is
proved and related to the operators grad, div and curl as introduced in traditional
calculus courses.
A combination of the study of manifolds with complex analysis in one variable
leads to the concept of Riemann surfaces. Their basic theory is presented in
Chapter 13. We begin the chapter with the Riemann Mapping Theorem, showing
conformal equivalence (holomorphic isomorphism) of simply connected proper
open subsets of C. We also present the Schwarz-Christoffel integrals giving
conformal equivalence between open convex polygons and the unit open disk;
examples include elliptic integrals. Then we introduce the theory of Riemann
surfaces and coverings, and construct universal coverings. We will see that even if
we are not interested in the abstraction of manifolds, this formalism will greatly
enhance our understanding of complex integration: we will now be able, for
example, to integrate a holomorphic function over a homotopy class of continuous
paths. We will also be able to understand how to make rigorous the concept
of a “multi-valued holomorphic function”, which was strongly suggested by the
methods of Chapter 10, yet could not be adequately approached by its methods.
Finally, studying complex differential forms on Riemann surfaces will lead us to
the basic notions of “dz-d z-calculus”, which is very helpful in complex analysis.
To demonstrate, we will apply this to extending some of the methods of complex
analysis beyond the case of holomorphic functions.
Introduction xix

Chapter 14 is devoted, primarily, to the basic problem of the calculus of

variations in one independent variable, and to the Euler-Lagrange equation for
critical functions. Here, only the material of Part I is used. In the second half of
the chapter we define a Riemann metric on an open subset of Rn and discuss the
geodesic equation in more detail. Also, we prove the local minimality of geodesics.
A part of the reason for introducing this material is a motivation of the topics
investigated in Chapter 15 where we combine it with the material on manifolds to
obtain the basic concepts of Riemannian geometry. We start with tensor calculus and
then move on to affine connections, Riemann metrics on manifolds and curvature,
and give a local characterization of the Euclidean space as the Riemannian manifold
with zero curvature tensor. Using the methods of the last section of Chapter 13, we
also show how to construct a complex structure on an oriented Riemann manifold
in dimension 2.
Chapter 16 concerns Hilbert and Banach spaces, and introduces the basic
concepts of functional analysis. Here we need as prerequisites only the techniques
from Part I and Chapter 9. We start with the definition and basic properties of
Hilbert spaces. We show that a Hilbert space provides, in a sense, an infinite-
dimensional extension of the nice properties of the finite-dimensional vector spaces
with inner product. Banach spaces are also introduced; their theory is much harder,
but nevertheless we are able to prove a few neat results. Starting with Hahn-Banach’s
Theorem, we go on to examining duals of Banach spaces, proving, for example, that
the dual of Lp is Lq for 1 < p < 1, 1=p C 1=q D 2. We will also prove the Open
Mapping Theorem and the Closed Graph Theorem.
In Chapter 17 we present some applications, mainly of Hilbert spaces. (One tends
to use Hilbert spaces wherever possible, precisely because they are much easier.)
We will prove the Radon-Nikodym Theorem, and use it to prove a version of the
Fundamental Theorem of Calculus for the Lebesgue integral, a fairly hard fact.
In the framework of Hilbert spaces we also define Fourier series and (continuous)
Fourier transformation. As a fringe benefit, we will introduce Borel measures. The
theory of Lp -spaces generalizes to this case, and includes some interesting new
examples.
When using this book as a textbook for a course, an instructor should aim to cover
most of Part I in the first semester. After this, one can work with Part II, basically,
on three independent tracks, thus customizing the course as needed, and as time
permits.
For Chapters 10 and 14, no additional prerequisities beyond Part I are needed.
All the other chapters require Chapter 9; just this added to Part I suffices for the
study of Hilbert spaces.
In the remaining group, multilinear algebra of Chapter 11 has to precede
manifolds in Chapter 12 and Riemannian geometry in Chapter 15, which also uses
the facts from Chapter 14, and its last section uses Chapter 10.
Chapter 13 uses Chapter 10 and Chapter 12.
Using this dependence of chapters, an instructor may decide about the topics for
the next semester, possibly assigning some material to the students as independent
xx Introduction

reading. There is no need to cover entire chapters, there are endless possibilities how
to mix and match topics to create an interesting course.
The student (or reader) is, in any case, most strongly encouraged to keep the
book for further study. As already mentioned, we anticipate that graduate students
of mathematics, mathematicians and scientists in areas using analysis, as well as
instructors of courses in analysis will find this book useful as a reference, and will
find their own ways through the topics.
In the Bibliography section, the reader will find suggestions for further reading.
In the more advanced sections of this text, we often introduce concepts (such as
“Lie group” or “de Rham cohomology”) which arise as a natural culmination of
our discussion, but whose systematic development is beyond the scope of this book.
These concepts are meant to motivate further study. We would like to emphasize
that our list of literature is by no means meant to be complete. The books we do
suggest all have a fairly close connection to the present text, and to mathematical
analysis. They contain more detailed information, as well as suggestions of further
literature.
Finally, we would like to say a few words about sources. The overall conception
of the book is original: we designed the logic of the interdependence of topics, and
the strategy for their presentation. Many proofs are, in fact, also “original” in the
sense that we made up our own arguments to fit best the particular stage of the
presentation (the book contains no new mathematical results). Given the scope of
the project, however, we did, in some cases, consult lecture notes, other books and
occasionally even research papers for particular proofs. All the books used are listed
in the Bibliography at the end of the book. In the case of research papers, we give
the name of who we believe is the original author of the proof, but do not include
explicit journal references, as we feel an effort of being even partially fair would lead
to a web of references which would only bewilder a first-time student of the subject.
We did want to mention, however, that there are also quite a few proofs which seem
to have become “standard” in this field (including, sometimes, particular notation),
and whose original author we were not able to track down. We would like to thank
all of those, who, by inventing those proofs, contributed to this book implicitly. We
would also like to thank colleagues and students who read parts of our book, and
gave us valuable comments. Last but not least, the authors gratefully acknowledge
the support of CE-ITI of Charles University, the Michigan Center for Theoretical
Physics and the NSF.
 Part I
A Rigorous Approach to Advanced Calculus
Preliminaries
1

The typical reader of this text will have had a rigorous “ı-"” first year calculus
course, using a text such as for example [22]. Such a course will have included
definitions and basic properties of the standard elementary functions (polynomials,
rational functions, exponentials and logarithms, trigonometric and cyclometric
functions), the concept of continuity of a real function and the fact that continuity
is preserved under standard constructions (sum, product, composition, etc.), and the
basic rules of computing derivatives. We review here mainly the more theoretical
aspects of these topics. The reason for reviewing them are two-fold. The first reason
is that we would like this text to be as self-contained as possible. The second reason
is that some of the basic results have, in fact, substantial depth in them, and the more
advanced topics on which this book focuses make heavy use of them. Not reviewing
such topics would at times even create a danger of circular arguments.

1 Real and complex numbers

Perhaps it is useful to go over a few basic conventions first. By a map or mapping

from a set S to a set T we mean a rule which assigns to each element of S precisely
one element of T . Two rules are considered the same if they always produce the
same value (in T ) on the same input element (of S ).
Therefore, technically, a map is a binary relation, i.e. a set R of pairs .x; y/,
x 2 S , y 2 T , such that for each s 2 S , there is precisely one .s; y/ 2 R. The
sets S , T are called the domain and codomain, respectively. We will denote a map
f from a set S to a set T by f W S ! T . For such a map, and a set X  S , we will
denote by f ŒX  the set of all elements f .x/ such that x 2 X . Similarly, for Y  T ,
we will denote by f 1 ŒY  the set of all x 2 S such that f .x/ 2 Y . The set f ŒX 
is called the image of the set X under the map f , and the set f 1 ŒY  is called the
pre-image of the set Y under f . This use of the square bracket may perhaps seem
unusually pedantic, but will soon pay off in the text below. The image f ŒS  of the
domain is sometimes called the image of the map f .

I. Kriz and A. Pultr, Introduction to Mathematical Analysis, 3

DOI 10.1007/978-3-0348-0636-7 1, © Springer Basel 2013
4 1 Preliminaries

To comment briefly on the use of inclusion symbols, throughout this book, we

generally use  to denote a subset with possible equality; when equality is excluded,
we use ¨. We generally avoid the somewhat ambiguous symbol . When we do use
it, it means  in a context where equality is a priori excluded for an obvious reason
not entering the logic of the argument (this may happen, for example, for a finite
subset of the real numbers when we are not using the finiteness to conclude that the
complement is non-empty).
Returning to the subject of mappings, for a map f W S ! T , and U  T , often
it is useful to have a special symbol for the map g W U ! T which is defined by
g.x/ D f .x/ when x 2 U . The map g is called the restriction to the subset U , and
denoted by f jU or f jU .
A map is called onto if f ŒS  D T , and is called one-to-one (briefly 1-1) if for
every s1 ; s2 2 S , f .s1 / D f .s2 / ) s1 D s2 . Onto maps are also called surjective
and one-to-one maps are called injective. A bijective map is a map which is both
surjective and injective.
The composition of maps f W S ! T , g W T ! U will be denoted by g ı f .x/ D
g.f .x// for x 2 S . In fact, the circle is often omitted, and instead of g ı f ,
one simply writes gf . One must, of course, make sure there is no possibility of
confusion with multiplication. The identity map IdS on a set S is defined simply by
IdS .x/ D x for every x 2 S . Note that a bijective map f W S ! T has an inverse,
i.e. a map f 1 such that f 1 ı f D IdS , f ı f 1 D IdT .
We will use the symbol N to denote the set of (positive) natural numbers
f1; 2; : : : g. The set of non-negative integers will be denoted by N0 , and the set of all
integers by Z. The set of all rational numbers will be denoted by Q. The set R of
real numbers needs more attention.

1.1

Let us summarize the structure of the set R of real numbers as it will be used in this
text. We do not give a rigorous construction of the real numbers at this point. Such a
construction however will emerge in the context of our discussion of completeness
in Chapter 9, where it is reviewed as an exercise.
First R is a field, that is, there are binary operations, addition C and multiplica-
tion  (which will be often indicated simply by juxtaposition) that are associative
(that is, a C .b C c/ D .a C b/ C c and a.bc/ D .ab/c) and commutative
(that is, a C b D b C a and ab D ba) and related by the distributivity law
(a.b Cc/ D abCac). There are neutral elements, zero 0 and one (also called unit) 1,
such that a C 0 D a and a  1 D a. With each a 2 R we have associated an element
a 2 R such that a C .a/ D 0; almost the same holds for the multiplication where
we have for every non-zero a an element a1 (also denoted by a1 ) such that a a1 D 1.
Furthermore there is a linear order  on R (a binary relation such that a  a,
that a  b and b  a implies a D b, that a  b and b  c implies a  c, and
finally that for any a; b either a  b or a D b or a  b), and this order is preserved
by addition and by multiplication by elements that are  0.
1 Real and complex numbers 5

Then, we have the absolute value jaj equal to a if a  0 and to a if a  0. One

often views R as a line with ja  bj representing the distance between a and b.
For M  R, we say that a is an upper (resp lower) bound of M if x  a (resp.
x  a) for all x 2 M . A supremum (resp. infimum), denoted by

sup M (resp. inf M ),

is the least upper bound (resp. greatest lower bound), if it exists. Thus, the supremum
s of M is characterized by the properties
(1) 8x 2 M; x  s, and
(2) if x  a for all x 2 M then s  a
(similarly for infimum with  instead of ). (2) It is often expediently replaced by
(2’) if a < s then there is an x 2 M such that a < x
(realize that (1)&(2) is indeed equivalent to (1)&(2’)). It is a specific property of the
ordered field R that
each non-empty M  R that has an upper bound has a supremum
or, equivalently, that
each non-empty M  R that has a lower bound has an infimum.
In mathematical analysis, it is often customary to use the symbols 1 D C1
and 1. The supremum (resp. infimum) of the empty set is defined to be 1
(resp. 1), and the supremum (resp. infimum) of a set with no upper bound (resp. no
lower bound) is defined to be 1 (resp. 1). Accordingly, it is customary to write
1 < a < C1 for any real number a, and to define 1 C 1 D 1  .1/ D 1,
and .1/ C .1/ D .1/  1 D 1, a  ˙1 D ˙1 resp. a  ˙1 D 1
for a > 0 resp. a < 0. It is important to keep in mind, however, that the symbols 1,
1 are not real numbers, and expressions such as 1  1 or 0  1 are undefined
(although see Section 6 of Chapter 4 for an exception).
If M is a subset of R and sup.M / 2 M (resp. inf.M / 2 M ), we say that the
supremum (resp. infimum) is attained, and speak of a maximum resp. minimum. In
this case, we may use the notation

max M; min M:

It is important to keep in mind that, unlike the supremum and infimum, a maximum
and/or minimum of a non-empty bounded subset of R may not exist. A non-empty
finite subset of R, however, always has a maximum and a minimum.
Variants of notation associated with suprema and infima (resp. maxima and
minima) are often used. For example, instead of sup M , one may write

sup x;
x2M

and similarly for the infimum, etc.

6 1 Preliminaries

Let us fix notations for open and closed intervals in R: As usual, .a; b/ means the
set of all x 2 R such that a < x < b, where a; b are real numbers or ˙1. We will
denote by ha; bi the corresponding closed interval, i.e. the set of all x 2 R [ f˙1g
such that a  x  b. The reader can fill in the meaning of the symbols ha; b/, .a; bi.

1.2

The field of complex numbers C can be represented as R R with addition

.x1 ; x2 / C .y1 ; y2 / D .x1 C y1 ; x2 C y2 / and multiplication .x1 ; x2 /.y1 ; y2 / D
.x1 y1  x2 y2 ; x1 y2 C x2 y1 /; we have the zero .0; 0/ and the unit .1; 0/. It is an
easy exercise to check that C has the arithmetic properties of a field (associativity,
commutativity, distributivity) and that .x1 ; x2 / D .x1 ; x2 / and .x1 ; x2 /1 D
x1 x2
. 2 ; 2 /. The field of complex numbers, however, has no reasonable
x1 C x2 x1 C x22
2

order.
One introduces the complex conjugate of x D .x1 ; x2 / as x D .x1 ; x2 /. It is
easy to see that

xCy DxCy and x  y D x  y: (*)

Further, there is the absolute value (also called the modulus) defined by setting
p q
jxj D xx D x12 C x22

x
(thus, x 1 D ).
jxj2
If we view C as the Euclidean plane (one often speaks of the Gaussian plane)
then jxj is the standard distance of x from .0; 0/, and jx  yj is the standard
Pythagorean distance.
Usually one sets i D .0; 1/ and writes

x1 C ix2 for .x1 ; x2 /

(note that the multiplication rule in C comes from distributivity and the equality
i 2 D 1). In the other direction, one puts

Re.x1 C ix2 / D x1 ; Im.x1 C ix2 / D x2 ;

and calls these real numbers the real resp. imaginary part of x1 C ix2 .
We have a natural embedding of fields

.x 7! .x; 0// W R ! C

which will be used without further mention; note that this embedding respects the
absolute value.
1 Real and complex numbers 7

1.3 Theorem. For the absolute value of complex numbers one has

jx C yj  jxj C jyj:

Proof. Let x D x1 C ix2 and y D y1 C iy2 . We can assume y ¤ 0. For any real
number  we have 0  .xj C yj /2 D xj2 C 2xj yj C 2 yj , j D 1; 2. Adding
these inequalities, we obtain

0  jxj2 C 2.x1 y1 C x2 y2 / C 2 jyj2 :

x1 y1 C x2 y2
Setting  D  yields
jyj2

.x1 y1 C x2 y2 /2 .x1 y1 C x2 y2 /2 2 .x1 y1 C x2 y2 /2

0  jxj2  2 C jyj D jxj 2

jyj2 jyj4 jyj2

and hence .x1 y1 C x2 y2 /2  jxj2 jyj2 . Consequently,

jx C yj2 D .x1 C y1 /2 C .x2 C y2 /2 D jxj2 C 2.x1 y1 C x2 y2 / C jyj2 

 jxj2 C 2jxjjyj C jyj2 D .jxj C jyj/2 : t
u

1.3.1 Corollary. If x D x1 C ix2 and y D y1 C iy2 then

jx  yj  jx1  y1 j C jx2  y2 j and jxj  yj j  jx  yj:

1.3.2 Comment:
A function is basically the same thing as a map, although in many texts (including
this one), the term function is reserved for a map whose codomain is a set whose
elements we perceive as numbers, or at least some closely related generalizations.
For example, the codomain may be R, C or a subset of one of these sets, or it may
be, say, Œ0; 1. Sometimes, we will allow the codomain to consist even of n-tuples
of numbers, see for example Chapter 3. While many basic courses define functions
simply by formulas without worrying about the domain and codomain, in a rigorous
view of the subject, specifying domains and codomains is essential for capturing
even the most basic phenomena: Consider, for example, the function

f .x/ D x 2 : (*)

If we specify the domain as R, the function certainly cannot have an inverse no

matter what the codomain is, since it is not injective. If we do specify the domain,
say, as Œ0; 1/, and the codomain as R, there is still no inverse, since the function is
not onto. If, however, (*) is considered as a function

f W Œ0; 1/ ! Œ0; 1/;

8 1 Preliminaries

then there is an inverse, which is rather useful, namely

p
f 1 .x/ D x:

1.4 Polynomials and their roots

1.4.1
Recall that a polynomial with coefficients in R resp. C is an expression which is
either

(the zero polynomial) or is of the form

p.x/  an x n C    C a1 x C a0 with aj 2 R resp. C (*)

for some n 2 N0 , where an ¤ 0. Technically, then, a non-zero polynomial is

simply the .n C 1/-tuple of real (resp. complex) numbers .a0 ; : : : ; an /. (This is the
information we would have to specify if we were to store the polynomial, say, on a
computer.)
The number n is called the degree of the polynomial p.x/. The degree of the
zero polynomial is not defined.
Of course, the polynomial (*) also determines a function

.x 7! an x n C    C a1 x C a0 / W R ! R resp. C ! C:

The zero polynomial determines a function, too, namely one which is constantly 0.
In analysis, it is quite common to identify a polynomial with the function it
determines (although note carefully that the domain and codomain of the function
corresponding to a polynomial with real coefficients will change if its coefficients
are considered as complex numbers). Nevertheless, this identification is permissible,
since two different polynomials over R (resp. C) never correspond to the same
function. To this end, note that it suffices to show that a non-zero polynomial does
not correspond to the 0 function (by passing to the difference). To this end, simply
note that if jx0 j is very large, then

jan x0n j > jan1 j  jx0n1 j C    C ja0 j  jan1 x0n1 C    C a0 j;

and hence p.x0 / ¤ 0 by the triangle inequality.

In fact, much more is true: a polynomial of degree n can be zero at no more than
n different points of R (or C). Define a complex root of a polynomial p.x/ to be a
number c 2 C such that p.c/ D 0. If c 2 R, we speak of a real root.
1 Real and complex numbers 9

1.4.2 Lemma. If p.x/ is a polynomial with coefficients in C with root c 2 C, then

there exists a unique polynomial q.x/ with coefficients in C such that

p.x/ D q.x/.x  c/:

Moreover, q.x/ has degree n  1. If the coefficients of p.x/ and the number c are
real, then the coefficients of the polynomial q.x/ are real.

Proof. For existence, recall (or observe by chain cancellation) that for k 2 N,

x k  c k D .x  c/.x k1 C x k2 c C    C xc k2 C c k1 /:

Therefore,

p.x/  p.c/ D an .x n  c n / C    C a1 .x  c/

can be written as x  c times another polynomial. If c is a root of p.x/, p.c/ D 0

by definition, so our statement follows.
For uniqueness, note that for a non-zero polynomial q.x/ of degree k, the
polynomial q.x/.x  c/ has degree k C 1, and hence is non-zero. t
u

We immediately have the following

1.4.3 Corollary. A polynomial p.x/ of degree n with coefficients in R or C has at

most n distinct roots.

1.4.4 Proposition. Let c be a (possibly complex) root of a polynomial p with

coefficients in R. Then the complex conjugate c is also a root of p.

Proof. By 1.2.(*), p.c/ D p.c/. t

u

The Fundamental Theorem of Algebra (which will be proved in Chapter 10,

Theorem 5.2), states that

every polynomial of degree  1 has a root in C:

By Lemma 1.4.2, we then see that every polynomial of degree n with coefficients
in C can be written uniquely (up to order of factors) as

p.x/ D an .x  c1 /      .x  cn /

for some complex numbers c1 ; : : : ; cn . (The uniqueness is proved by induction.)

Note that the numbers c1 ; : : : ; cn may not be all distinct. When c D ci for exactly
k > 0 different values i 2 f1; : : : ; ng, we say that the root c has multiplicity k.
10 1 Preliminaries

Applying Proposition 1.4.4 inductively, if a polynomial p.x/ has real coefficients,

then the multiplicity of the root c is equal to the multiplicity of c.

2 Convergent and Cauchy sequences

2.1

A sequence .xn /n in R or in C is said to converge to x if

8" > 0 9n0 such that n  n0 ) jxn  xj < ":

We write

lim xn D x or simply lim xn D x:

n

The reader is certainly familiar with the easy facts such as lim.xn C yn / D lim xn C
lim yn or lim.xn yn / D lim xn lim yn , etc.

2.2

A sequence .xn /n in R or in C is said to be Cauchy if

8" > 0 9n0 such that m; n  n0 ) jxm  xn j < ":

Observation. Every convergent sequence is Cauchy.

(If we have the implication n  n0 ) jxn  xj < " then m; n  n0 )
jxm  xn j  jxm  xj C jx  xn j < 2").

2.3 Theorem. If a  xn  b for all xn , then the sequence .xn /n contains a

convergent subsequence .x/kn , and a  limn xkn  b.

Proof. Let a  xn  b for all n. Set

M D fx j 9 infinitely many n such that x  xn g:

This set is non-empty (a 2 M ) and bounded (no x > b is in M ) and hence there is
a finite s D sup M . By the definition, each

1 1
Kn D fk j s  < xk < s C g
n n
3 Continuous functions 11

is infinite, and we can choose, first, xk1 such that s  1 < xk1 < s C 1 and if
k1 <    < kn are chosen with kj 2 Kj we can choose a knC1 2 KnC1 such that
knC1 > kn . Then obviously limn xkn D s, and equally obviously a  s  b. t
u

2.4 Theorem. (Bolzano - Cauchy) Every Cauchy sequence of real numbers con-
verges.

Proof. Since for some m and all n  m, jxn  xm j < 1, a Cauchy sequence is
bounded and hence it contains a subsequence xk1 ; : : : ; xkn ; : : : converging to an x.
But then limn xn D x: indeed, choose for an " > 0 an n0 such that for m; n  n0
we have jxm  xn j < " and jx  xkn j < ". Then, since kn  n, jx  xn j < 2" for
n  n0 . t
u

2.5

From 1.3.1, we see that if .xn D xn1 C ixn2 /n is a sequence of complex numbers
then
.xn /n converges if and only if both .xnj /n converge
and
.xn /n is Cauchy if and only if both .xnj /n are Cauchy.
Consequently we can infer from Theorem 2.4 the following
Corollary. Every Cauchy sequence of complex numbers converges.

3 Continuous functions

3.1

Recall that a real (resp. complex) function of one real (resp. complex) variable is a
mapping

f W X ! R (resp. ! C) with X  R (resp.  C):

In the real case X will be most often an interval, that is, a set J  R such that
x; y 2 J and x  z  y implies that z 2 J .
Recall the standard notation from 1.1 for (bounded) open and closed intervals:

.a; b/ D fx j a < x < bg and ha; bi D fx j a  x  bg:

The intervals ha; bi will be often referred to as compact intervals; the reason for this
terminology will become apparent in Chapter 2 below. A function f W X ! R resp.
C is said to be continuous if
12 1 Preliminaries

8x 2 X 8" > 0 9ı > 0 such that jy  xj < ı ) jf .y/  f .x/j < ": (3.1.1)

3.2 Proposition. A function f is continuous if and only if for every convergent

sequence one has f .lim xn / D lim f .xn /.

Proof. If f is continuous, if x D lim xn 2 X and if " > 0 then first choose a

ı > 0 as in (3.1.1) and then an n0 such that jxn  xj < " for n  n0 . Then
jf .xn /  f .x/j < " for n  n0 .
Now suppose f is not continuous. Then there is an x 2 X and an " > 0 such that
for every ı > 0 there is a y.ı/ such that jy.ı/  xj < ı and jf .y.ı//  f .x/j  ".
Set xn D y. n1 /. Then limn xn D x while f .xn / cannot converge to f .x/. t
u

3.3 Theorem. (The Intermediate Value Theorem) Let J be an interval, let

f W J ! R be a continuous function, and let for some u < v, min.f .u/; f .v// 
K  max.f .u/; f .v//. Then there is an x 2 hu; vi such that f .x/ D K.

Proof. Since a restriction of a continuous function is obviously continuous, since

f is continuous if and only if f is, and since if f is continuous then any x 7!
f .x/  K with K fixed is continuous, it suffices to prove that if f W ha; bi ! R is
continuous and f .a/  0  f .b/ then there is a c 2 ha; bi such that f .c/ D 0.
Set c D supfx 2 ha; bi j f .x/  0g.
Suppose f .c/ > 0. Then for " D f .c/ we have a ı > 0 such that for x > c  ı,
f .x/ > f .c/  " D 0 while there should exist an x > c  ı such that f .x/  0.
Similarly we cannot have f .c/ < 0 because for " D f .c/ we would have a ı > 0
with f .x/ < f .c/ C " D 0 for c  x < c C ı contradicting the definition of c
again. Thus, f .c/ D 0. t
u

3.4 Theorem. A continuous function f W ha; bi ! R on a compact interval attains

a maximum and a minimum.

Proof. for the maximum. Set M D ff .x/ j x 2 ha; big. If it is not bounded choose
xn > n and consider a convergent subsequence xkn with limit y. We have f .y/ D
limn xkn which is impossible because it would yield f .y/ > n for all n. Hence M
is bounded and has a finite supremum s. Now choose xn with s  n1 < xn  s, and
a convergent subsequence xkn with limit y 2 ha; bi to obtain f .y/ D s. t
u

3.5

A function f is said to be uniformly continuous if

8" > 0 9ı > 0 such that 8x; y; jy  xj < ı ) jf .y/  f .x/j < ":

3.5.1 Theorem. A continuous function on ha; bi is uniformly continuous.

4 Derivatives and the Mean Value Theorem 13

Proof. Suppose not. Then there exists an " > 0 such that
1
8n 9xn ; yn such that jxn  yn j < and jf .xn /  f .yn /j  ":
n
Choose a convergent subsequence .xkn /n and then a convergent subsequence .ykmn /n
of .ykn /n . Then we have limn xkmn D limn ykmn contradicting Proposition 3.2 and
the inequality j limn f .xkmn /  limn f .ykmn /j  ". t
u

4 Derivatives and the Mean Value Theorem

4.1

Let f W X ! R be a function, X  R. We say that f has a limit A at a point a and

write

lim f .x/ D A
x!a

if it is defined on .u; v/ X fag for some u < a < v and if

8" > 0 9ı > 0 such that x 2 .a  ı; a C ı/ X fag ) jf .x/  Aj < ":

Note that f does not have to be defined in a, and if it is, lim f .x/ D A does not
x!a
say anything about the value f .a/.

4.2

Let J be an open interval. A function f W J ! R has a derivative A in a point x if

f .x C h/  f .x/
lim DA
h!0 h
(that is, if the limit on the left-hand side exists, and if it is equal to a). The reader is
certainly familiar with the notation

df .x/
A D f 0 .x/; or
dx
and with the basic computation rules like .f C g/0 D f 0 C g 0 or .fg/0 D f 0 g C
fg 0 etc.

4.3 Theorem. A function f has a derivative A at the point x if and only if there is
a function  defined on some .ı; ı/ X f0g (ı > 0) such that

lim .h/ D 0 and f .x C h/  f .x/ D Ah C h.h/:

h!0
14 1 Preliminaries

Proof. If such a  exists we have for h 2 .ı; ı/ X f0g,

f .x C h/  f .x/
D A C .h/
h

f .x C h/  f .x/
and hence lim D A. On the other hand, if the derivative exists
h!0 h
f .x C h/  f .x/
then we can set .h/ D  A. t
u
h

4.3.1 Corollary. If f has a non-zero derivative at a point x then f .x/ is neither a

maximum nor a minimum value of f (A maximum resp. minimum value of a function
f is the maximum resp. minimum, if one exists, of the set of values of f .).

(Indeed, consider f .x C h/  f .x/ D h.A  .h// for j.h/j < jAj.)

A point at which a function f has zero derivative or the derivative does not exist
is called a critical point. Corollary 4.3.1 implies that critical points are the only
points at which a function f can have a minimum or a maximum. It is, of course,
not guaranteed that a critical point would be an actual minimum or maximum (take
the point x D 0 for the function f .x/ D x 3 ). However, see Theorem 4.7 below for
a partial converse of the Corollary.

4.4 The Mean Value Theorem

4.4.1 Theorem. (Rolle) Let f be continuous in ha; bi and let it have a derivative
in .a; b/. Let f .a/ D f .b/. Then there is a c 2 .a; b/ such that f .c/ D 0.

Proof. If f is constant then f 0 .c/ D 0 for all c. If not then, as f .a/ D f .b/, either
its maximum or its minimum (recall Theorem 3.4) has to be attained in a c 2 .a; b/.
By 4.3.1, f 0 .c/ D 0. t
u

4.4.2 Theorem. (The Mean Value Theorem, Lagrange’s Theorem) Let f be contin-
uous in ha; bi and let it have a derivative at .a; b/. Then there is a c 2 .a; b/ such
that

f .b/  f .a/
f 0 .c/ D :
ba
More generally, if, furthermore, g is a function with the same properties and such
that g.b/ ¤ g.a/ and g 0 .x/ ¤ 0 then there is a c 2 .a; b/ such that

f 0 .c/ f .b/  f .a/

D :
g 0 .c/ g.b/  g.a/
4 Derivatives and the Mean Value Theorem 15

Proof. Set F .x/ D .f .x/  f .a//.g.b/  g.a// C .f .b/  f .a//.g.x/  g.a//.

Then F .a/ D F .b/ D 0 and F 0 .x/ D f 0 .x/.g.b/  g.a//  g 0 .x/.f .b/  f .a//
and the second formula follows. For the first one, set g.x/ D x. t
u

4.4.3
The Mean Value Theorem is often used in the following form (to be compared
with 4.3):
let x; x C h be both in an interval in which f has a derivative. Then

f .x C h/  f .x/ D f 0 .x C h/  h for some  2 .0; 1/:

(Use 4.4.2 for hx; x C hi resp. hx C h; xi.)

4.4.4 Corollary. If f is continuous in ha; bi and if it has a positive (resp. negative)

derivative in .a; b/ then it strictly increases (resp. decreases) (i.e. x < y ) f .x/ <
f .y/ resp. x < y ) f .x/ > f .y/) in ha; bi. If f 0  0 in .a; b/ then f is constant.

(For, f .y/  f .x/ D f 0 .c/.y  x/. )

4.5 The second derivative, convex and concave functions

Suppose f has a derivative f 0 .x/ at every x 2 J , where J is an open interval. Thus,

we have a new real function f 0 W J ! R and this function may have a derivative
again. In such a case we speak of the second derivative.

4.5.1
A function f is said to be convex resp. concave on an interval ha; bi if for any two
x < y in ha; bi and any z D tx C .1  t/y, (0 < t < 1), between these arguments,

f .x/  tf .x/ C .1  t/f .y/ resp. f .x/  tf .x/ C .1  t/f .y/

(that is, the points of the graph of f lay below (resp.above) the straight line
connecting the points .x; f .x// and .y; f .y//).

4.5.2 Proposition. Let f be continuous on ha; bi and let f have a non-negative

(resp non-positive) second derivative on .a; b/. Then it is convex (resp.concave)
on ha; bi.

Proof. In the notation above we have

y  z D y  tx  .1  t/y D t.y  x/; z  x D .1  t/.y  x/:

Let the second derivative be non-negative. Then we have x < u < z < v < y and
u < w < v such that
16 1 Preliminaries

f .y/  f .z/ f .z/  f .x/

 D f 0 .v/  f 0 .u/ D f 00 .w/.v  u/  0
yz zx

so that

f .y/  f .z/ f .x/  f .x/

 ;
t.y  x/ .1  t/.y  x/

hence .1  t/.f .y/  f .z//  t.f .z/  f .x// and finally

tf .x/ C .1  t/f .y/  f .z/: t

u

4.5.3 An application: Young’s inequality

We have

Proposition. Let a; b > 0 and let p; q  1 be such that 1

p C 1
q D 1. Then

ap bq
ab  C :
p q

Proof. Since ln00 .x/ D  x12 < 0, ln is concave; thus if, say ap < b q we have

1 1 1 1
ln. ap C b q /  ln.ap / C ln.b q / D ln a C ln b D ln.ab/
p q p q

and since ln increases, the inequality follows. t

u

4.6 Derivatives of higher order and Taylor’s Theorem

Just as we defined the first and second derivative of a function on an open interval
J , we may iterate the process to define the third, fourth derivative, etc. In general,
we speak of the derivative of n’th order, and define

f .0/ D f; f .1/ D f 0 and further f .nC1/ D .f .n/ /0 :

(Of course, as before, for a given function, such higher derivatives may or may not
exist.)

4.6.1 Theorem. (Taylor) Let f have derivatives up to degree n C 1 in an open

interval containing a and x, a ¤ x. Then there is a c in the open interval between
a and x such that
4 Derivatives and the Mean Value Theorem 17

X
n
f .k/ .a/ f .nC1/ .c/
f .x/ D .x  a/k C .x  a/nC1 :
kŠ .n C 1/Š
kD0

Proof. Fix x and a and define a function R.t/ of one real variable t by setting

X
n
f .k/ .t/
R.t/ D f .x/  .x  t/k :
kŠ
kD0

Then we have

dR.t/ X f .kC1/ .t/

n X f .k/ .t/ n
R0 .t/ D D .x  t/k C k.x  t/k1 :
dt kŠ kŠ
kD0 kD1

Substituting l D k C 1 in the second sum we obtain

X
n
f .kC1/ .t/ X
n1
f .lC1/ .t/ f .nC1/ .t/
R0 .t/ D  .x  t/k C .x  t/l D  .x  t/n :
kŠ lŠ nŠ
kD0 lD0

Now define g.t/ D .x  t/nC1 . Then g 0 .t/ D .n C 1/.x  t/n and g.x/ D 0. Since
also R.x/ D 0 we obtain from Theorem 4.4.2,

R.a/ R.a/  R.x/ R0 .c/ f .nC1/.c/.x  c/n

D D 0 D
g.a/ g.a/  g.x/ g .c/ nŠ.n C 1/.x  c/n

and hence

f nC1 .c/ f .nC1/ .c/

R.a/ D g.a/ D .x  a/nC1
nŠ.n C 1/ .n C 1/Š

X
n
f .k/ .a/
and the statement follows, since R.a/ D f .x/  .x  a/k , that is,
kŠ
kD0
X
n
f .k/ .a/
f .x/ D .x  a/k C R.a/. t
u
kŠ
kD0

4.7 Local extremes

One immediate consequence of Taylor’s Theorem is a partial converse of

Corollary 4.3.1. Suppose a function f is defined on an open interval containing
a point x0 . We say that x0 is a local maximum (resp. local minimum) of f if there
exists a ı > 0 such that for all x 2 .x0  ı; x0 C ı/ such that x ¤ x0 , f .x/ < f .x0 /
(resp. f .x/ > f .x0 /). We have the following
18 1 Preliminaries

Theorem. Let f be a function such that f 0 and f 00 exist and are continuous on
an open interval .a; b/ containing a point x0 . Suppose further that f 0 .x0 / D 0,
f 00 .x0 / < 0 (resp. f 00 .x0 / > 0). Then x0 is a local maximum (resp. local minimum)
of f .

Proof. Let us treat the case of f 00 .x0 / D q > 0; the proof in the other case is
analogous. By Taylor’s Theorem, for x 2 .a; b/, x ¤ x0 , there exists a point c in
the open interval between x0 and x such that

f 00 .c/
f .x/ D f .x0 / C .x  x0 /2 : (*)
2

Since f 00 is continuous, there exists a ı > 0 such that for x 2 .x0  ı; x0 C ı/,
f 00 .c/ > 0. Then it follows immediately from (*) that if x 2 .x0  ı; x0 C ı/,
x ¤ x0 , f .x/ > f .x0 /. t
u

5 Uniform convergence

5.1

Let fn be real or complex functions defined on an X . We write limn fn D f , or

briefly fn ! f if limn fn .x/ D f .x/ for all x 2 X , and say that fn converge to
f pointwise. This convergence is not very satisfactory: consider fn W h0; 1i ! R
defined by fn .x/ D x n , an example where all the fn are continuous while the limit
f is not.
We shall need to work with a stronger concept. A sequence of (real or complex)
functions .fn /n is said to converge to f uniformly if

8" > 0 9n0 such that 8n  n0 8x; jfn .x/  f .x/j < ":

This is often indicated by writing fn  f .

5.2 Theorem. Let fn be continuous and let fn  f . Then f is continuous.

Proof. Take an x0 2 X and an " > 0. Choose an n such that for all n  n0 and
for all x, jfn .x/  f .x/j < 3" , and then a ı > 0 such that jfn .x0 /  fn .x/j < 3" for
jx0  xj < ı. Then for jx0  xj < ı,

jf .x0 /  f .x/j  jf .x0 /  fn .x0 /j C jfn .x0 /  fn .x/j C jfn .x/  f .x/j < ":

t
u

5.3 Theorem. Let fn have derivatives on an open interval J , let fn ! f and let
fn0  g. Then f has a derivative and f 0 D g.
6 Series. Series of functions 19

Proof. By the Mean Value Theorem we have for some 0 <  < 1,
ˇ ˇ
ˇ f .x C h/  f .x/ ˇ
ˇ  g.x/ ˇ
ˇ h ˇ
ˇ ˇ
ˇ f .x C h/  fn .x C h/ f .x/  fn .x/ fn .x C h/  fn .x/ ˇ
Dˇ ˇ C C  g.x/ˇˇ
h h h
ˇ ˇ
ˇ f .x C h/  fn .x C h/ f .x/  fn .x/ ˇ
Dˇ ˇ C C fn .x C h/  g.x/ˇˇ
0
h h
1 1
 jf .x C h/  fn .x C h/j C jf .x C h/  fn .x C h/j
jhj jhj
C jfn0 .x C h/  g.x C h/j C jg.x C h/  g.x/j:

Fix an h ¤ 0 such that jg.x C h/  g.x/j < 4" . Then choose an n such that
(1) jf .x C h/  fn .x C h/j < 4" jhj and jf .x/  fn .x/j < 4" jhj, and
(2) jfn0 .x C h/  g.x C h/j < 4" .
(Inequality (2) is where we need the convergence to be uniform: we do not know
the exact position of x C h). Then
ˇ ˇ
ˇ f .x C h/  f .x/ ˇ
ˇ  g.x/ ˇ < 1 " jhj C 1 " jhj C " C " D ": t
u
ˇ h ˇ jhj 4 jhj 4 4 4

6 Series. Series of functions

6.1

Let .an /n be a sequence of real or complex numbers. The associated series (or sum
P
1 P
of a series) an (briefly, an if there is no danger of confusion) is the limit
nD1
X
n
P
lim ak provided it exists; in such a case we say that an converges, and we say
n
kD1 P
that it converges absolutely if jan j converges.

6.2 Consequences of Absolute Convergence

6.2.1 Proposition.
P An absolutely convergent
P series converges. More generally, if
jan j  bn and bn converges then an converges.

X
n X
n
Proof. Set sn D ak and s n D bk . For m  n we have
kD1 kD1
20 1 Preliminaries

X
m X
m X
m
jsm  sn j D j an j  jan j  bn D js m  s n j:
kDnC1 kDnC1 kDnC1

Thus, if .s n /n is convergent, hence Cauchy, then .sn /n is Cauchy and hence

convergent. t
u
P
6.2.2 Proposition. The series
X an converges absolutely if and only if for every
" > 0 there is an n0 with jan j < " for every finite K  fn j n  n0 g.
k2K

X
m
Proof. The formula is equivalent to stating that jak j < " for n0  n  m.
kDn
X
n
Thus, the condition amounts to stating that the sequence . jak j/n is
kD1
Cauchy. t
u
P
6.2.3 Theorem. Let an converge absolutely. Then for all bijections p from the
1
X
set of natural numbers f1; 2; : : : g to itself the sums ap.n/ are equal.
nD1

1
X
Proof. Let ap.n/ D s for a bijection p. Choose n1 sufficiently large such that
X "
nD1
jan j < for every finite K  fn j n  n1 g and, further, an n0 such that for
2
k2K
n  n0 we have
ˇ n ˇ
ˇX ˇ "
ˇ ˇ
ˇ ap.n/  s ˇ < and fp.1/; : : : ; p.n/g f1; : : : ; n1 g:
ˇ ˇ 2
kD1

Now if n  p.n0 / then if we consider K D f1; : : : ; ngXfp.1/; : : : ; p.n0 /g we obtain

ˇ n ˇ ˇn ˇ ˇn ˇ
ˇX ˇ ˇX 0 X ˇ ˇX 0 ˇ X
ˇ ˇ ˇ ˇ ˇ ˇ " "
ˇ ak  s ˇ D ˇ ap.k/ C ak  s ˇ  ˇ ap.k/  s ˇ C jak j < C D ":
ˇ ˇ ˇ ˇ ˇ ˇ 2 2
kD1 kD1 k2K kD1 k2K

t
u

6.3

It is worth taking this a little further. A set S is called countable if there exists a
bijection  W f1; 2; : : : g ! S . Note that this is the same as ordering S into an
6 Series. Series of functions 21

infinite sequence s1 ; s2 ; : : : where si X

go through all elements of S , and each element
occurs exactly once. Let us say that as converges absolutely if
s2S

X
sup jas j
KS finite s2K

1
X
is finite. By Proposition 6.2.2, this is equivalent to a .n/ converging absolutely
nD1
for one specified bijection  (which can be arbitrary). Theorem 6.2.3 then shows
that when this occurs, then
X
as
s2S

is well-defined. Here is an example where this point of view helps:

X and let S D
6.3.1 Theorem. Let S1 ; S2 ; : : : be disjoint finite or countable sets,
[
Si . Then the set S is finite or countable. Furthermore, if as converges
i s2S
absolutely, then
0 1
1
X X X
@ as A D as ; (*)
i D1 s2Si s2S

and the left-hand side converges absolutely.

Proof. The case when S is finite is not interesting. Otherwise, we may order the
elements of S into an infinite sequence as follows: Assume each of the sets Si is
ordered into a (finite or infinite) sequence. Then let Tn consist of all the i ’th elements
S of Sj such that 1  i; j  n. Then clearly each Tn is finite, and Tn  TnC1 ,
(if any)
and Ti D S . Thus, we can order S by taking all the elements of T1 , then all the
remaining elements of T2 , etc. Thus, S is countable. X
Now let us investigate (*). The supremum sup jas j over finite subsets K
s2K
of Si is less than or equalX
to the analogous supremum over K finite subsets of
S , which shows that each as converges absolutely. Further, for a finite subset
s2Si
K  1; 2; : : : ,
ˇ ˇ
X ˇˇ X ˇˇ X X XX
ˇ as ˇˇ  jas j  sup jas j
ˇ
i 2K ˇs2Si ˇ i 2K s2Si i 2K s2Li
22 1 Preliminaries

where the supremum on the right-hand side is over all finite subsets Li  Si . We
see that the right-hand side is finite by our assumption of absolute convergence over
S , and therefore the left-hand side of (*) converges absolutely.
Finally, to prove equality in (*), use a variation of the above proof of the fact that
S is countable: Let Tn consist of sufficiently many elements of S1 ; : : : ; Sn such that
the sum
0 1
Xn X
@ as A
i D1 s2Tn \Si

differs from
0 1
X
n X
@ as A
i D1 s2Si

by less than 1=n. Then the limit of these particular partial sums is the left- hand side
of (*), but is also equal to the right-hand side by absolute convergence. t
u
1
X 1
X
6.3.2 Corollary. Let am , bn be absolutely convergent series. Then
mD0 nD0

1
! 1
! 1
!
X X X X
n
am  bn D ak bnk ; (*)
mD0 nD0 nD0 kD0

with the right-hand side converging absolutely.

P
Proof. By the assumption, the supremum of jam j  jbn j over K, L finite
X
m2K;n2L
subsets of f0; 1; 2; : : : g is finite, thus proving that am bn is absolutely convergent,
S
where S is the set of all pairs of numbers 0; 1; : : : . The rest follows from
Theorem 6.3.1. t
u

6.4

Let .fn /n be a sequence of real or complex functions (defined on an X  R resp C).

X1 1
X
The series fn is defined as a function with values fn .x/ whenever the last
nD1 nD1
1
X
series converges. If f .x/ D fn .x/ converges (resp. converges absolutely) for
nD1
7 Power series 23

P X
n
all x 2 X we say that fn converges (resp. converges absolutely). If fk .x/ 
kD1
1
X
f .x/ we say that fn converges uniformly.
nD1

6.4.1
Since finite sums of continuous functions are continuous and since .f1 C  Cfn /0 D
f10 C    C fn0 we obtain from Theorem 5.2, Theorem 5.3 and Proposition 6.2.1
P
Corollary. 1. Let fn be continuous and let fn uniformly converge. Then the
resulting P
function is continuous. P
2. Let f D n fn converge, P let fn0 exist and let fn0 converge
P uniformly. Then f 0
0
exists and is equal to fn (that is, the derivative of fn can be obtained by
taking derivatives of the individual summands). P
3. The statements 1 (resp 2) apply to the case of jfn .x/j  an with an
convergent; here the convergence is, moreover, absolute.

7 Power series

7.1
1
X
A power series with center c is a series an .x c/n . So far we will limit ourselves
nD1
to the real context; later in Chapter 10, we will discuss them in the complex case.

7.2

The limes superior (sometimes also called the upper limit) of a sequence .an /n of a
real number is the number

lim sup an D inf sup an :

n n kn

It obviously exists if the sequence .an /n is bounded; if not we set lim supn an D
C1. It is easy to see that lim supn an D lim an whenever the latter exists. The limes
inferior (or lower limit) is defined analogously with inf and sup switched.

7.2.1 Proposition. Let lim sup an D inf sup an D a and limn bn D b. Let an ; bn 
n kn
0 and let a; b be finite. Then lim sup an bn D ab.
24 1 Preliminaries

Proof. Choose an " > 0 and a K > a C b. Take an  > 0 such that K > a C b C 
and K < " There is an n0 such that

n  n0 ) sup ak < a C  and b   < bn < b C :

kn

That is, for every n  n0 there exists a k.n/  n such that a  ak.n/ < a C  and
b   < bk.n/ < b C  so that

a.b  / < ak.n/ bk.n/ < .a C /.b C / D ab C .a C b C /

and since ab  " < ab  K < ab  a and .a C b C / < K < " we see that

ab  " < ak.n/ bk.n/ < ab C "

and conclude that lim sup an bn D ab. t

u

7.2.2
1
X
For a power series an .x  c/n define the radius of convergence
nD1

1
 D ..an /n / D p
lim sup n
jan j
p
if lim sup n
jan j ¤ 0; otherwise set ..an /n / D C1.

1
X
Theorem. Let r < ..an /n /. Then the power series an .x  c/n converges
n1
absolutely and uniformly on the set fx j jx  cj  rg.
1
X
On the other hand, if jx  cj >  then an .x  c/n does not converge.
nD1

Proof. I. Let jx  cj  r < . Choose a q such that

p
r  inf sup k
jak j < q < 1:
n kn

Then there is an n such that

p p
r sup k jak j < q and hence r k jak j < q for all k  n:
kn

Choose a K  1 such that r k  Kq k for k  n. Then

jak x k j  Kq k for all k and jx  cj  r

7 Power series 25

and hence by Theorem 6.3.1, the series converges on fx j jxcj  rg absolutely

and uniformly. p
II. Let jx  cj > ; then jx  cj infn supkn k jak j > 1, hence jx  cj
p
supkn k jak j > 1 for all n, and hence for each n there is a k.n/  n such
p
that jx  cj  k.n/ jak.n/ > 1, and hence jak.n/ .x  c/k.n/ j > 1 and the series
cannot converge: its summands do not even converge to 0. t
u

7.3

Consider the series

1
X
nan .x  c/n1 : (*)
nD1

1
X
Obviously it converges if and only if nan .x  c/n does and hence its radius of
nD1
convergence is

1
p :
lim sup n njan j
p p p p
By Proposition 7.2.1, lim sup n njan j D lim sup n n n jan j D lim n n  lim sup
p p p 1
n
jan j D lim sup n jan j (since lim n n D lim e n ln n D e 0 D 1). Thus,

P radius ofn convergence  of the n1

the series (*) is the same as that of the original
an .x  c/ and since nan .x  c/ P of an .x n c/ we conclude
is the derivative n

from 5.3 and 6.3.1 that for jx  cj <  the series an .x  c/ has a derivative,
and that it is obtained as the sum of the derivatives of the individual summands.
So far, this derivative had to be understood as in the real context. In fact, however,
it is valid for complex power series as well; see Chapter 10.

7.4 Remark

If we proceed to compute the higher derivatives summand-wise, we obtain

1
X
f .k/ .x/ D n.n  1/    .n  k C 1/an .x  c/nk :
nDk

In particular
Other documents randomly have
different content
of the epidermis, as shown in the Manilla shell illustrated herewith,
in which the lip, usually trimmed off for commercial purposes, is
preserved. The lining of the meleagrina is not as iridescent as that of
the thin shell varieties.
Thus the shell is being constantly enlarged at the edge, by a
deposit of the exudations of the mantle; conchiolin for the epidermis
outside, lime for the prisms and inner layers of transparent plates,
until the shell has attained its full growth in size, after which some
varieties continue to lay on nacre only.

MANILA PEARL-SHELL WITH THE LIP CONSERVED

The linings of some have a black rim, extending from the hinge on
one side, around the edge to the hinge on the other side. Viewed
from the edge this dark band appears to be a sixteenth to half an
inch wide (widest at the lip), fading out as it becomes lost under the
thicker white nacre of the interior, but turn the shell up and look at it
squarely from the front and it is black only around the extreme edge
where it joins the epidermis. This kind of shell is found in the Pacific
about the islands of Polynesia and is called the black shell. In others
the nacre is white to the edge. The iridescence of the white shell
generally shows more play of color than that of the black. The white
shell is usually somewhat flatter and broader than the black, and the
epidermis is light yellowish-brown. This variety is found in great
abundance on the northern and western coasts of Australia. The
yellow, greenish and grayish shells (these colors refer to the edge of
the lining), are similar in every way, but inferior, the yellow being the
best of the three.
The shell lining of a common form of the unio, or fresh-water
mussel pictured at page 146, like that of the meleagrina, shows little
iridescence except at the edges outside the pallial lines, where the
nacre is comparatively thin, and at the striated surface of the scar or
bed of the adductor muscle. In quality of color and luster it is inferior
to the nacre of the sea fish, the white being more chalky in
appearance and the luster less pearly. The material of which the
shell is composed and its construction are however almost identical
with that of the salt-water mollusk. In fact all shells are made of the
same ingredients and are constructed on the same general principles
by the animals inhabiting them.
 MISSISSIPPI NIGGER-HEAD PEARL MUSSEL

This description of pearl shells has been given here because a

knowledge of the shell enables one to understand the formation and
characteristics of a true pearl, and the differences which exist
between the gem and other similar formations formed in pearl and
other oysters, mussels, and univalves. Many such formations are
found, having the elements and constructed like one or both of the
outer parts of the shell, and some, in part like the lining, but these
are not true pearls; the gem has neither the material nor
construction of the middle and outer shell. Except that the pearl,
because of its form, is rarely iridescent even to a slight degree,
whereas the nacreous lining of some pearl-bearing shells is brilliantly
so, the pearl and the nacre of the shell in which it grows, are
essentially the same. Pearls are more or less spherical and
independent formations, made by the fish on the same plan and
from the same secretions with which it lines the shell, misdirected by
abnormal conditions. Those constructed like any other part of the
shell are not true pearls.
The normal instinctive action of the mollusk is self-protective and
adaptive. By the secretive action of its mantle it gathers from the
water in which it lives, material to build a shell with a rough and
rugged exterior for its enemies, and adapted to resist the chemical
activities by which it is surrounded, and a perfectly smooth lining
suitable as an interposition for its own delicate organism.
Barring accidents, the building functions of the animal are
employed only in the extension of the shell to meet the needs of its
own growth and protection. But should a particle of secretion
intended for the shell, harden within the folds of the oyster's mantle,
or some parasite or other intruder present itself within the nacre-
forming sphere, the instinctive action which lines the rougher part of
the shell is also directed toward the foreigner, and it is at once
covered with a like deposit. This is the birth of a pearl, and it grows
layer by layer as long as it remains within the scope of the nacre
building instinct. These layers, or skins as they are called, are
seldom iridescent. Occasionally a pearl of that character is found,
but it is generally from a fresh-water mussel, and the nacreous
plates are of unusual tenuity.
Although the pearl like the lining of the mollusk's shell is
composed of carbonate of lime in series of thin waves lapping each
other, each series constituting a plate or separable layer, there is a
distinct difference in construction.
Whereas the lining is a series of horizontal layers, the pearl is
made up of concentric layers, each addition enveloping those
preceding it. These skins however are not always absolutely distinct
and separate. Instead of being like a succession of globular skins,
each completely covered by its successor, the growth is often spiral
and the construction is as if the nucleus had been rolled one, two, or
three complete revolutions in a continuous plate of nacre, and the
spiral envelope then finally merged into another plate and the
process repeated. That which to a casual glance, therefore, appears
to be six rings of nacre in a sectional cut, is in reality, several spirals
of two or three turns each.
It is also noticeable that whereas the wave edges, with all their
eccentricities, trend generally in one direction in the shell nacre, in
the pearl, the lines twist and curl with a concentric tendency, as
though the waves had been laid on by turning or rolling the pearl in
the material of which it is composed.
A white pearl on being cut in half shows a number of faint dark
rings one within the other, from the surface to the nucleus in the
centre; usually these rings occur at almost regular intervals. Upon
close examination under the microscope, it will be seen that the
inner part of these intervals is white, and that the color gradually
changes to a yellowish tint which deepens until it culminates in that
which appears as a dark line against the succeeding outer formation,
the material of which is also white in the beginning. Although this
change of color is very slight, a section between two rings will often
show three distinct bands; the inner white, the centre one faintly
yellow and the outer one of a deeper tint. In some cases the dark
concentric rings succeed each other very closely, in which case no
abrupt changes of color between them are noticeable. The material
occupying the space between the rings is the sectional appearance
of the skin of pearl. Upon applying a weak acid to the surface of an
entire section of a pearl, it effervesces, and the inner colorless parts
of the bands are at once attacked. After several hours the white
inner part of the skins will show depressions where the calcium
carbonate has been dissolved, and the outer parts of the skins will
be marked by coarse black rings of undissolved animal tissue, similar
in appearance to the epidermis of the shell. Now as these skins are
made up of many very thin waves of calcium carbonate lapping each
other and set in animal tissue, it would appear, therefore, that in the
beginning these waves of transparent calcium carbonate are set in
animal tissue of extreme tenuity and that the proportion of animal
tissue increases with the growth of the skin until it reaches a stage
provocative of a new skin, which begins with purer layers of the
smoother crystallized mineral like its predecessor, and identical with
the nacre of the shell. If this be so, it would account for the various
tints of color and degrees of luster in white pearls and for the fact
that the outer skins of very lustrous pearls are usually very thin also.
Similar conditions exist in colored pearls, though the presence of a
pigment makes them less noticeable. The skins of the haliotis pearl,
which separate easily, usually show remarkable luster on the inner
surface.
Sometimes the nucleus is surrounded by a confused mass without
apparent concentric markings, as though it had been enveloped in
nacre which had solidified while stationary, or the first deposit shows
the concentric skin arrangement at one segment of the circle only;
followed by layers which appear in the depressions of the mass and
are continued until they finally include the whole pearl. These layers
are usually very thin, and the partial or segmentary layer formation
is quite common in the early stages of the pearl's growth. At that
period the concentric lines are also irregular, and in many cases
where the curve is true, they extend about one quarter of the
circumference only, another concentric skin being lapped on the
ends, as though the globular skin had been formed in sections.
As before stated, it often happens that the skin division lines are
spiral, as though the nucleus had been rolled one way in the
nacreous material. In all cases the first deposits of a skin, that is the
first of the nacreous waves of which a skin is composed, appear to
be most transparent and lustrous. The component waves of nacre
then gradually become more impregnated with animal tissue until
they apparently reach a stage which induces either a rest on the
part of the fish, to gather nacreous material, or a new deposit of less
impure nacre, to protect itself from the increasing impurity of the
pearl's skin.
The skins undoubtedly mark certain stages in the formation of the
pearl, though the skin and the nacreous waves of which it is
composed are often confounded. In the skinning of pearls an entire
skin is seldom peeled off. The surface is scraped, a number of the
component waves being taken off, until the luster is improved and it
is then supposed that the entire outer skin has been removed. A
close examination however, will show, by breakages in the surface of
the waves, that the under skin with its peculiar and systematic
arrangement of surface wave edges, has not been reached.
A sectional view as seen in a half pearl would lead one to infer
that a free pearl in the beginning lies stationary in the oyster; is
turned or partially rolled as it grows larger; and finally, on attaining
about a one grain size, is kept in constant motion with a concentric
rolling in the nacreous exudations of the mantle which are deposited
upon it.
The nuclei of pearls were long thought to be grains of sand, but
late and careful research has shown that in the majority of cases
they are minute parasitic or domiciliary worms.
Professor Herdman and James Hornell, after three consecutive
inspections of the oyster banks in the Gulf of Manaar in 1902-3,
stated in a paper contributed to the British Association for the
advancement of science, that after examining many hundreds of
oysters and decalcifying a large number of pearls, they had come to
the conclusion, that grains of sand and other inorganic particles
formed the nuclei of pearls only under exceptional circumstances, as
for instance, when the shell was injured by the breaking of the ears,
which would enable sand to get into the interior.
Pearls, or pearly excrescences on the interior of the shell, were
due to the intrusion of leucodore, clione and other borers. Pearls
found in the mussels, especially at the levator and pallial insertions,
were formed around calcospherules, minute calcareous concretions
produced in the tissues. But most of the fine pearls found free in the
body of the Ceylon oyster, contained the remains of platyhelminthian
parasites. These observations agree with the opinions formed, after
careful study, by several eminent conchologists.
 The action of the mollusk results differently as the object to be
covered is free within the folds of the creature's mantle or, rising
above the surface of the nacreous lining, presses upon it. If free, the
intruder is enveloped by the animal's exudations and the deposits
become concentric instead of level, or nearly so, as in the
construction of the shell. It is said that the foreign substance acts as
an irritant, causing the fish to exude its secretions abnormally in
order to protect itself, and thereby creating a diseased condition; but
from the fact that the process continues after the intruder has been
enveloped and rendered as non-irritant as the natural lining of the
shell, it would appear that the introduction of a foreign element
simply draws upon it the normal impulse of the fish to cover with
nacre anything with which it comes in contact, and that the method
of doing it is similar to the instinctive rolling action of the tongue
when some insoluble globule is put in the mouth, for not only do
free pearls grow spherically, but a nucleus fast to the shell is not
covered simply but it grows to a pearl, round and domelike, as
nearly spherical as its juncture with the shell will permit.
Not only is the composition of a pearl identical with the lining of
the shell where it is formed, but in a general way its appearance and
characteristics are the same, except that free pearls are sometimes
colored when the nacre of the shell is white.
Button pearls, warts and baroques, grown fast to the shell, are
usually like the surrounding nacre in every respect.
Salt-water pearls are characterized by the soft velvety luster of the
oriental mother-of-pearl, and fresh-waters, like the lining of the unio,
have a somewhat thinner looking and more chalky texture.
Abalone pearls have the irregular surface and coloring of the
haliotis. Conch pearls resemble the delicate pink china-like lining of
the shell, and clam pearls have the glazed earthenware appearance
of the inside of a clam shell. The one material difference between a
pearl and the lining of the shell in which it grows is, that in the one
case the fish deposits the nacre over an even surface, and in the
other wraps it around a central point with delicate precision in
successive filmy layers.
Dissection shows that a pearl during growth is liable to many
mishaps. As with the human creature, a promising youth may end in
a wretched maturity. It is also possible that an ugly period may be
redeemed by later happenings, and the thing that was worthless in
its early existence, be found in its age worthy of a place among the
great gems. Pearls found with a dull, chalky exterior sometimes have
lustrous skins beneath. Sometimes a bony-looking formation will be
found, on breaking it, to have a variety of skins in the interior, some
of which are very lustrous, others white and chalky, like the middle
shell of the mollusk.
Many of these dead pearls are formed throughout of this material.
Others, perfectly spherical, are simply successive layers of prism
groups like the conchiolin plates of the shell. Upon cutting these
through the centre the skins are shown by the concentric rings
marking their divisions and the prismatic formation appears as
glistening lines radiating from the nucleus to the surface. Under the
microscope these layers, which are thicker than the nacreous skins
of true pearls, appear identical with the epidermis plates, except that
they are concentric instead of flat, and are free from the coarse,
rough, conchiolin deposit which forms the extreme outer coating of
the shells. This deposit is also found, however, in some pearl
formations, as many of the abalone baroques, especially when they
are somewhat flat in shape, are like two pearl blisters joined, with
the shell-building process reversed, the rough, black conchiolin being
inside, and the nacre outside. Undoubtedly pearls containing hidden
qualities which made them once gems are thrown away as valueless,
while others found just as nature had covered their earlier
coarseness with a coat of beauty, are worn and excite much
admiration for their skin-deep beauty.
Though the successive skins of a pearl do not usually vary much in
color, except in abalone pearls, it does happen occasionally, for the
removal of dark yellow skins sometimes discloses another of better
color—a good pink for instance. From the sectional appearance of
pearls it seems probable, that in the majority of cases the color of
yellow pearls would be improved by the removal of the outer waves
of the outer skin.
Changes in shape sometimes occur during the growth of the pearl,
the tendency being always toward the rounding of the surface. If the
nucleus is fast to the shell, a dome is built over and around it. If the
nucleus permits, the nacre is deposited not only over but under its
edges to the point of contact with the shell, so that a button pearl
connected with the shell at the centre only, results. Two pearls held
against the shell and growing side by side are separately enveloped
until they touch each other, after which they are included in single
deposits of nacre and the depression between their domes becomes
less distinct with each successive coating. Similarly, a cluster of small
pearls lying together often forms the nucleus of a large rounded
baroque or button pearl. Examination of such formations shows, that
up to a certain period the pearls have a separate existence and
growth. They then become joined in an irregular mass of twinned
pearls, and finally, if allowed to remain in the oyster long enough, all
individuality is lost in the tendency to round over. The same thing
occurs when grains of sand or other intrusions become attached to a
growing pearl. They are quite prominent when first included in the
nacreous deposit and can be easily detached from the under pearl
by breaking through the layer which binds them on; but they are
soon obliterated by succeeding deposits. This filling-in process is
sometimes accomplished by additional layers in the depression,
sometimes by thicker layers. It happens occasionally, when skinning
a round pearl, that one of these fillings is uncovered and flakes out,
leaving the pearl irregular, as it was in a former stage of its growth.
Although pearls naturally grow spherically, many free pearls are
more or less buttoned, that is, have a flat place from which the pearl
rises like a dome, high or low. This happens when the pearl is held
during growth by the fish against the shell with a part of its body
intervening. According to circumstances, the pearl varies in form
from slightly button, to a low dome, rising from a plane at its
greatest diameter. Should a pearl of this description become
dislodged, the rounding action of the mollusk would begin at once to
obliterate the plane.
If undisturbed, the process would result eventually in changing the
button to a round or nearly round pearl, but should the pearl be
taken from the fish before the metamorphosis is completed, a
depression, or pit, would mar its contour. When borers intrude
through the shell, the presentation is at once covered with nacre,
and successive deposits are built up around it resulting in the
nacreous wart known as a baroque. The rounding action of the
mollusk is clearly shown in these excrescences, as the borer is not
simply covered and levelled with the shell, but the slight elevation
above the level of the lining receives a continuity of concentric
deposits which finally raise it very considerably above the surface
and separate it in construction from the lining to which it is attached.
The shell herewith reproduced illustrates the result. Borers pierced it
at the thick part of the hinge, and burrowing down, entered the
interior at the point where the baroque is shown. In rare instances,
pearls attached to the shell do escape the concentric deposition, for
they have been found buried under even layers of nacre, when the
mother-of-pearl was cut up in the process of manufacture.
 VENEZUELAN PEARL-SHELL, SHOWING BAROQUE

From the appearance of the striæ when they are divided

lengthwise, pear-shaped pearls appear to have been spherical at one
time. During a stage in the growth, the forming layer has curved
away from the centre at one section of the sphere to a point.
Succeeding layers, following the innovation, are deposited around
the extension until it becomes sufficiently elongated to give the pearl
the obovoid form.
Many pearls are shaped like a capsule. The ends of most are
rounded up to a full dome; some have somewhat flatter ends; many
are long and cylindrical like an ordinary capsule; others are short
and appear in shape like two high button pearls joined at their
bases; while some resemble a cartridge, one end being almost flat
and the other a somewhat pointed dome. It is noticeable that such
pearls have a chalky line around the middle, and sometimes there is
a lustrous band between two. These chalky lines are found, on
peeling such a pearl, to extend through all the interior layers.
Similarly, a high button joined at its entire circumference to the shell,
if the junction is abrupt, has an intersecting chalky line, marking the
juncture of the two, between the luster of the pearl and the shell
lining. If the base of the pearl and the shell form a curve there is no
chalky line of demarcation.
This suggests that whenever the animal is unable to envelop the
thing upon which the mantle deposits its secretions completely or is
not in touch with every part of it, there is at the extremity of its
action, an unnacreous deposit, corresponding to the deposit of
conchiolin or calcite, at the extreme edge of the shell which
precedes the nacreous layers following within and slightly back of it.
As the luster of the pearl arises from the transparency of the calcium
carbonate modified by the undulating lines formed by the edges of
the wave-plates, it may be that the lapping action of the mantle is
necessary for the regular formation and crystallization of these
plates, and that at points beyond the reach of this action, the
depositions of the mantle are therefore not pearly.
Much is necessarily conjectural as to the modus operandi by which
the shell and the pearl are formed but the invariable tendency
toward sphericity suggests that the nucleus of a pearl, when free
within the mollusk's mantle, is not only enveloped in its exudations,
but is either kept constantly moving with a rolling motion or lapped
on all sides by the membrane which exudes upon it the nacreous
material.
The instances cited of the short capsule shaped pearl and the high
button joined to the shell, which seem to escape the nacreous
deposit at the basis of the domes, favor the lapping or licking
method of depositing the nacreous solution and this action by the
mollusk would result in a constant rolling or turning motion imparted
to the object if it were free within the creature's body. The licking
and rolling action of the mollusk, modified by the conceivable
influences of position in the shell, would account for the spherical
form with all the various modifications in which the pearl is found.
To account for the variation of quality which undoubtedly exists in
the successive skins of some pearls, and the imperfections in the
nacre of the same skin, the theory has been advanced that the
secretions for the lining, the shell proper, and the epidermis, are
exuded by different parts of the mantle; the pearl traverses during
growth these different bands and its skins are modified by the
secretions, as they come within the various zones of influence. But
there are several facts which seem to oppose the theory.
In the first place all these parts of the mantle which supply the
material for the epidermis, the middle shell, and the lining, are
enclosed within the shell and in touch with the lining yet each
receives the exudations of that part of the mantle which supplies the
material suitable for it, the mantle invariably pushing the coarser
excretions outwardly to the shell's exterior. Again, whatever the
quality of the skin of the pearl may be, it is never of conchiolin like
the outer epidermis and though sometimes similar to the plates, of
which the conchiolin is the exposed fringe, it always contains
sufficient nacre to render the surface smooth. The fact that the skins
of a pearl do sometimes correspond with the different parts of the
shell, and that the same skin on the surface is occasionally partly
nacreous and unnacreous, in connection with the variation of quality
which exists in the internal composition of the skin, favors an idea
that the mixed and variable quantity of nacre in the skins may be
caused by the abnormal position of the mantle wrapped about the
growing pearl which would thereby come more or less under the
influence of the calcite and conchiolin zones distorted from their
normal extension and action.
It has also been suggested that the oyster deposits the nacreous
layer in a fluid state and then rests until the deposit hardens, when
the process is repeated. To a certain extent this may be true though
apparently it could not be a yearly process as pearls found in the
small varieties of the avicula which mature in four to six years and
die out in seven years, often contain a greater number of layers than
the years of the mollusk's life, and no pearl is ever found with a soft
exterior, though it seems possible that pearls with a dead white
chalky exterior are taken from the oyster at a period when the
crystallization of the outer skin has not been perfected, or that they
have escaped some action, chemical or of the animal, necessary for
the formation of the lustrous waves of nacre. Mr. Ludwig Stross, who
has had much experience at the pearl fisheries, says that he has
frequently found pearls of fair size in shells of the Lingah type which
could not be over twelve to fifteen months old. Some of these pearls
weighed fully three grains. As there are many apparent skins in a
pearl of that size, the divisions could not mark either years, seasons,
or breeding periods. In some experiments made by Mr. Stross, he
found that borings made to the interior of a living mollusk's shell
were closed by a film of hard nacre in two days.
The known facts about a pearl are these. It is composed of about
ninety-two per cent. carbonate of lime, about six per cent. organic
matter and a little over two per cent. water in combination almost
identical with the lining of the shell in which it grows and similar to
the mineral aragonite. In construction it is usually a series of layers,
which can sometimes be peeled off entirely, each one successively
enveloping its predecessors apparently as an independent structure
though itself composed of a number of thin lapping waves. Upon
cutting through these layers the divisions appear as a series of rings
and the intervals, though composed of many thin waves, appear
compact. It grows spherically or with such modifications as the
exigencies of position in the shell would reasonably account for.
These facts seem to justify the hypothesis that a foreign substance
upon entering the shell of a pearl oyster is at once enveloped or
washed in the creature's exudations; that the organic matter of the
secretions forms a filmy envelope in which the mineral contained in
them is precipitated or crystallizes in wave-like layers of crystals of
great tenuity, and that as these layers harden the process is
repeated, and that during the process the creature either revolves
the object, or about it, as it is free, or fastened to the shell. It is also
possible that changes in the organic matter interwoven with the
calcium carbonate may produce some chemical action resulting in
the crystallization of the lime, and the crystallization in turn be
provocative of another deposit, each process in turn being almost
simultaneous and that the process is continued until a paucity of
mineral in the exudations induces a rest for recuperation, after which
the process is repeated, the result being a succession of composite
skins as we find them. Whatever the cause, it is evident in all parts
of the shell and in the pearl that continuity of construction is
periodically arrested to be resumed upon exactly the same plan,
except that the material used in the succeeding layer of the pearl
may be formed occasionally like another of the shell sections though
usually it is like the preceding one.
Marked differences in the same skin occur more frequently in the
pearl formations of univalves. The skins of the abalone pearl
especially, are frequently nacreous in part only.
Pearl oysters are found in immense numbers on banks having a
calcareous foundation. They are extraordinarily prolific, the spat of
one oyster being estimated at upwards of several hundred
thousands to millions, so that were it not for the natural enemies of
their young and the liability of being swept away and scattered by
storms before they have anchored, the banks would be over-
crowded with the myriads produced. Some idea of the numbers may
be gained from the fact that during the fishing season the Ceylon
divers raise about one million each day.
The oysters are seldom found in water with a temperature below
75 degrees and they seem to thrive best in warm sheltered bays and
inlets, especially when the banks are situated far from the equator.
They attach themselves to the beds by a bunch of tough threads
which pass out through an aperture in the shells, near the hinge,
and fasten on the rocks and stones; consequently the oysters do not
lie flat, as might be supposed, but maintain an upright position,
hinge down, lip end up, and the shell slightly open for the passage
of the food-laden water, as the fresh-water mussels do. These
threads are called the beard or byssus, and are composed of
material similar to the epidermis of the shell.
The abalone, which is a univalve, holds on to the rocks by the
foot, a flat muscular appendage used for locomotion and also as an
anchor on the principle of the leather toy known to boys as a sucker.
Although pearls of value are found only in shells containing
mother-of-pearl, a small proportion only of the mother-of-pearl shells
contains pearls, and many varieties in which pearls are found do not
yield enough nacre to make the shells valuable. The size of the
meleagrina in some seas is remarkable. That at page 127,
photographed from a Tuamotu shell, measures 8-7/8 inches by 6-7/8
inches and weighs twenty-eight ounces troy.
It is of the black-edge variety, contains a large quantity of fine
quality mother-of-pearl, and has a beautiful small pearl attached to
the lining near the center of the shell. Though large, it is not full
grown. It is probably twelve to fourteen years old and would
continue to lay on mother-of-pearl and so grow thicker and heavier
until sixteen to eighteen years of age, when the oyster would reach
maturity. The Australian white shell at page 129 is a young shell—
that is, it has not attained the full thickness and weight of a mature
shell. The shells at pages 131 and 161 are from the coast of
Venezuela; they measure 2-1/4 by 2-1/4 inches and weigh seven
pennyweights each.
The common form of the pearl-bearing fresh-water mussel unio
(nigger-head) is illustrated at page 146. This shell measures 3-3/4
by 2-3/4 inches and weighs 3-1/2 ounces. It is from the Middle West
of the United States. In construction it resembles the meleagrina,
the epidermis being dark, though not as rough as that of the oyster,
and the lining white, showing slight iridescence around the lip-edge
and to a greater degree on the adductor muscle scar. The mother-of-
pearl under the epidermis at the thick or hinge end is quite
iridescent, and the lines which make the color play are plainly
discernible under the loup.
The largest and finest pearls, also the greatest number, are found
usually in distorted shells. This has given rise to the idea that they
are a symptom of disease in the fish, but having in mind the
functions of the three zones of the creature's mantle by which they
supply separately material for the epidermis, middle shell and lining,
one may conceive that if, by some extraordinary cause, the
secretions of one of these is largely withdrawn from the natural
channel, the losing part of the shell would warp the normal growth
of the others to its own dwarfage.
When the nacre grows to a pearl, contrary to the intent of nature,
instead of a lining for the shell endeavoring to keep pace with the
growing oyster, the full-growing exterior is distorted in
accommodating itself to the undersized lining. In view of the fact
that an oyster sometimes contains a large number of pearls (one
shell in New Caledonia contained 256) the diversion of nacre
sufficient to cover them, or to produce one large pearl, might
reasonably be expected to result in a considerable distortion of the
shell. It may also be that the displacement of the mantle, caused by
the wrapping of itself about the growing pearl, interferes with the
even deposit of shell material about the edges of the shell and so
distorts it.
Because deformed shells are more fruitful of pearls some have
advocated the practice of throwing perfectly-formed shells back into
the sea unopened, but, inasmuch as the mother-of-pearl of the
shells often exceeds in value the pearls found in them, this is not
likely to happen. Few fisheries could be made to pay if they were
fished for the pearls alone. In many of them the shells yield 90 per
cent. of the total value and are in fact the sole incentive for the
investment of the necessary capital.
Luckily for the world's supply of pearls, however, the disturbers of
the mollusk which cause these gems by their intrusions appear to be
more abundant in waters where the shell is valueless, the banks
about Ceylon especially being infested with the cestodes which are
commonly the nuclei of Indian pearls. It is interesting also to learn
that Mr. James Hornell (inspector of the pearl banks) finds these
worms in another stage in the file-fish, which frequents the banks to
prey upon the oysters, and confidently expects to find them in the
adult stage in the shark, which in turn devours the file-fish.
It is the opinion of Jameson of London and others, that the
parasite which causes the formation of pearls in the mussels of
Europe is frequently the larva of distomum somaterœ, from the
eider-duck and scoter, and that the larva first inhabits Tapes, or the
cockle, before getting into the mussel.
Generally the nuclei appear to be the bodies or eggs of minute
parasites—distoma, filaria, bucephalus, etc., and they vary in
different localities according to the animal life of the neighborhood.
In the still parts of the river Elster, where water-mites (Limnochares
anodontœ) were abundant, Kuchenmeister found that the mollusks
contained more pearls.
 METHODS OF FISHING
The beds of the marine shell-fish from which pearls are taken lie
always under water. Unlike others which are sometimes left exposed
by the tides, to be gathered by man without difficulty, the pearl
oyster is never left uncovered by the sea. It is found usually on
shoals some distance from shore, sometimes but five to seven feet
from the surface; more frequently fifteen to forty feet deep, and
often one hundred to one hundred and twenty-five and even one
hundred and fifty feet deep.
Everywhere, then, man's quest for pearls is confronted by the
heaving, restless waters of the sea, for the greater part of the year
rough and turbulent, frequently lashed to furious racing by tropic
tempests but through which he must in any case go to get them. In
a few places where the beds lie in shallow inlets and sheltered bays
they can be dredged, but almost universally the oysters are gathered
by divers. During the greater part of the year, when storms rage,
diving is very dangerous if not quite impossible; but when the song
of the sea is hushed to low crooning, and the gentle roll of the
waves does no more than playfully slap the boats in passing, then in
the seas where men dive for pearls they gather to the harvest of
gems.
There are two ways of diving—naked, and with dress. The former
is the common method throughout the Orient and is practised to-day
after the same manner that it was in the days of the Pharaohs and
the Cæsars, for the primitive method survives with few variations
wherever eastern people control the fisheries.
In the fishing season one sees now in the Red Sea and the Persian
Gulf and about Ceylon, the same scenes as they were enacted there
before Rome was a city, or France a nation, or the Macedonians
overran Egypt. Naked divers, diving into fifteen to forty feet of water,
use few aids. They grease their bodies, put greased cotton in the
ears and a forked stick, or tortoise-shell clip, upon the nostrils to
compress them, hang a wide-mouthed wicker basket or net at the
waist, and they are ready.
There are several methods of naked diving: head-first from a
spring-board attached to the side of the boat, as the Malabar coast
Hindus and some of the Egyptians do; swimming to the bottom, as
practised in the deep waters of the South Sea; and dropping to the
oyster bed with a stone. The latter is the most common way in
Indian, Egyptian, and Arabian waters, especially where the banks lie
in forty to fifty feet of water.
Standing on the spring-board a few seconds to fill his lungs, the
head-first diver suddenly plunges overboard and passes smoothly
and rapidly through the water straight to the shoal below. Gathering
quickly as many oysters as possible while his breath lasts, he places
them in the net at his waist, attaches them to a convenient rope
hanging from the boat's side and shoots to the surface. There he
recuperates by lazily floating about if the water is shallow, if deeper,
by climbing back into the boat for his next plunge. If diving in pairs,
one rests while his partner dives.
Expert divers who dive singly have an attendant, a manduck, who
attends to the lines and looks out for his interests generally. The
manduck drops a line with the oyster basket overboard and attaches
to it another weighted with a forty to fifty pound stone. These are so
fastened that they can be quickly released. The diver then drops into
the water feet first and placing his foot in a loop in the line over the
stone puts the basket on it, and releasing the lines, sinks to the
bottom. Disengaging himself, he proceeds to fill his basket while the
attendant pulls up the stone and adjusts it for the next descent.
When ready to return he signals his attendant, and holding on to the
line with the basket is drawn to the surface, occasionally accelerating
his own return by climbing the rope hand over hand at the same
time. He rests in the water by the boat's side until ready to dive
again, making seven or eight descents before climbing into the boat
for a longer rest and sun-bath.
The divers of India, Arabia and the Red Sea are natives of the
Madras Presidency, descendants of Arab fishers at Jaffna in Ceylon,
Arabs, and Egyptian Negroes. They travel long distances to the
fisheries and there are many of them between the Red Sea and
Ceylon. At the last fishing in the Gulf of Manaar there were about
forty-five hundred. Their dress during the time of the fishing consists
of a loin cloth only. They have many hereditary and class
superstitions, chief of which is their faith in shark-charmers. While
waiting for the fishing to begin they also seek to get from the fates
an inkling of the luck which will attend them. One common method
is by breaking a cocoanut on the diving stone; the more clean and
even the break, the better the luck.
The mortality among divers at the fisheries is not great in Asiatic
waters. Pneumonia is the greatest scourge, fatalities in diving being
few. It is necessary however to select robust men for depths beyond
forty feet; comparatively few can work without injurious effects
below that.
Some curious mixtures of ancient days and present times, of the
Pharaohs and infant industries, are seen. One may see a black slave
diver in the Red Sea hanging over the edge of his boat taking
observations through an old tin kerosene can with a bit of glass in
one end of it. This he sinks a little way in the water and gazes
through it below. Presently the can is discarded, over he goes and
returns shortly with a few shells; while near by a clumsy monster
emerges and a diver in dress climbs into his boat. This use of
modern tin cans and glass is adopted in seas where the shells are
scattered and is common to pearl-divers the world over.
The Moros have a method of fishing in very calm weather peculiar
to themselves. They drop a three-prong catcher attached to a rattan
rope upon the oyster bunches and so haul them up to the boat. This
can only be done when the sea is perfectly still, as even a ripple
would render a sight of the oysters impossible. Ordinarily they dive
to any depth down to twenty fathoms.
Many attempts have been made to introduce dress-diving among
the natives of the east but so far few have been successful. Results
from experiments have not compared favorably with naked diving
and so, with few exceptions, naked diving is still the rule in the east
where natives control the fishings.
But of all, the Polynesians, both male and female, adhere most
closely to the old way. Most of them will not even use a stone to
assist the descent, and they probably reach greater depths than the
naked divers of any other sea. Travellers report that, at a coral atoll
in the Southern Pacific owned by the French government and known
as Hikuereu, where the natives of Tahiti and other islands flock
during the season to fish for pearls, the boys and girls and women
are almost as expert as the men.
Whole families congregate here, remaining during the season
housed in huts framed of light cocoanut palms roofed with leaves.
These they bring with them, some coming several hundred miles.
The shells are mostly in sixty to seventy feet of water; some
however are brought from a depth of one hundred feet. It is
reported that a boy, on an exhibition dive, remained under water for
two minutes and forty seconds, going to a depth of a little over one
hundred feet. He was in sight all the time, the water being so
transparent that he could be seen on the bottom, leisurely selecting
pieces of coral for the officers of the ship above. These divers hang
in the water by one hand grasping the gunwale of the boat while
they examine the bottom for oysters through a glass which they hold
below the surface in the other hand.
When shells are sighted the glass is discarded, the lungs are filled
several times and the air expelled slowly. Upon reaching a certain fit
condition a long breath is taken until the lungs are inflated to their
utmost capacity; the diver then suddenly lets go, sinks a few feet
below the surface, turns quickly and head-first swims rapidly to the
bottom.
Welcome to Our Bookstore - The Ultimate Destination for Book Lovers
Are you passionate about books and eager to explore new worlds of
knowledge? At our website, we offer a vast collection of books that
cater to every interest and age group. From classic literature to
specialized publications, self-help books, and children’s stories, we
have it all! Each book is a gateway to new adventures, helping you
expand your knowledge and nourish your soul
Experience Convenient and Enjoyable Book Shopping Our website is more
than just an online bookstore—it’s a bridge connecting readers to the
timeless values of culture and wisdom. With a sleek and user-friendly
interface and a smart search system, you can find your favorite books
quickly and easily. Enjoy special promotions, fast home delivery, and
a seamless shopping experience that saves you time and enhances your
love for reading.
Let us accompany you on the journey of exploring knowledge and
personal growth!

ebookgate.com