(Ebook) The numerical solution of integral equations of

the second kind by Kendall E. Atkinson ISBN

9780511626340, 9780521102834, 9780521583916, 0521102839,
0521583918, 0511626347 Pdf Download

https://ebooknice.com/product/the-numerical-solution-of-integral-
equations-of-the-second-kind-1405704

★★★★★
4.9 out of 5.0 (62 reviews )

DOWNLOAD PDF

ebooknice.com
 (Ebook) The numerical solution of integral equations of the
second kind by Kendall E. Atkinson ISBN 9780511626340,
9780521102834, 9780521583916, 0521102839, 0521583918,
0511626347 Pdf Download

EBOOK

Available Formats

■ PDF eBook Study Guide Ebook

EXCLUSIVE 2025 EDUCATIONAL COLLECTION - LIMITED TIME

INSTANT DOWNLOAD VIEW LIBRARY

 We believe these products will be a great fit for you. Click
the link to download now, or visit ebooknice.com
to discover even more!

(Ebook) Biota Grow 2C gather 2C cook by Loucas, Jason; Viles,

James ISBN 9781459699816, 9781743365571, 9781925268492,
1459699815, 1743365578, 1925268497

https://ebooknice.com/product/biota-grow-2c-gather-2c-cook-6661374

(Ebook) Matematik 5000+ Kurs 2c Lärobok by Lena Alfredsson, Hans

Heikne, Sanna Bodemyr ISBN 9789127456600, 9127456609

https://ebooknice.com/product/matematik-5000-kurs-2c-larobok-23848312

(Ebook) SAT II Success MATH 1C and 2C 2002 (Peterson's SAT II

Success) by Peterson's ISBN 9780768906677, 0768906679

https://ebooknice.com/product/sat-ii-success-math-1c-and-2c-2002-peterson-
s-sat-ii-success-1722018

(Ebook) Master SAT II Math 1c and 2c 4th ed (Arco Master the SAT
Subject Test: Math Levels 1 & 2) by Arco ISBN 9780768923049,
0768923042

https://ebooknice.com/product/master-sat-ii-math-1c-and-2c-4th-ed-arco-
master-the-sat-subject-test-math-levels-1-2-2326094
(Ebook) Cambridge IGCSE and O Level History Workbook 2C - Depth
Study: the United States, 1919-41 2nd Edition by Benjamin
Harrison ISBN 9781398375147, 9781398375048, 1398375144,
1398375047
https://ebooknice.com/product/cambridge-igcse-and-o-level-history-
workbook-2c-depth-study-the-united-states-1919-41-2nd-edition-53538044

(Ebook) Numerical Solution of Ordinary Differential Equations by

Kendall Atkinson, Weimin Han, David E. Stewart ISBN
9780470042946, 047004294X

https://ebooknice.com/product/numerical-solution-of-ordinary-differential-
equations-2510676

(Ebook) The Numerical Solution of Ordinary and Partial

Differential Equations, Second Edition by Granville
Sewell(auth.) ISBN 9780471735809, 9780471742012, 0471735809,
0471742015
https://ebooknice.com/product/the-numerical-solution-of-ordinary-and-
partial-differential-equations-second-edition-4309222

(Ebook) Eulerian Codes for the Numerical Solution of the Kinetic

Equations of Plasmas by Magdi Shoucri ISBN 9781613245613,
1613245610

https://ebooknice.com/product/eulerian-codes-for-the-numerical-solution-of-
the-kinetic-equations-of-plasmas-51764312

(Ebook) The Numerical Solution Of Ordinary And Partial

Differential Equations by Granville Sewell ISBN 9780471735809,
0471735809

https://ebooknice.com/product/the-numerical-solution-of-ordinary-and-
partial-differential-equations-1288772
This book provides an extensive introduction to the numerical solution of a large
class of integral equations. The initial chapters provide a general framework
for the numerical analysis of Fredholm integral equations of the second kind,
covering degenerate kernel, projection, and Nystrom methods. Additional dis-
cussions of multivariable integral equations and iteration methods update the
reader on the present state of the art in this area.
The final chapters focus on the numerical solution of boundary integral
equation (BIE) reformulations of Laplace's equation, in both two and three
dimensions. Two chapters are devoted to planar BIE problems, which include
both existing methods and remaining questions. Practial problems for BIE such
as the set up and solution of the discretized BIE are also discussed.
Each chapter concludes with a discussion of the literature, and a large bib-
liography serves as an extended resource for students and researchers needing
more information on solving particular integral equations.
 CAMBRIDGE MONOGRAPHS ON
APPLIED AND COMPUTATIONAL
MATHEMATICS

Series Editors
P. G. CIARLET, A. ISERLES, R. V. KOHN, M. H. WRIGHT

4 The Numerical Solution of Integral

Equations of the Second Kind
The Cambridge Monographs on Applied and Computational Mathematics
reflects the crucial role of mathematical and computational techniques in con-
temporary science. The series publishes expositions on all aspects of applicable
and numerical mathematics, with an emphasis on new developments in this fast-
moving area of research.
State-of-the-art methods and algorithms as well as modern mathematical
descriptions of physical and mechanical ideas are presented in a manner suited
to graduate research students and professionals alike. Sound pedagogical pre-
sentation is a prerequisite. It is intended that books in the series will serve to
inform a new generation of researchers.

Also in this series:

A Practical Guide to Pseudospectral Methods, Bengt Fornberg
Level Set Methods, J.A. Sethian
Dynamical Systems and Numerical Analysis, A.M. Stuart andA.R. Humphries
The Numerical Solution of Integral Equations
of the Second Kind

KENDALL E. ATKINSON
University of Iowa

AMBRIDGE
UNIVERSITY PRESS
 CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, Sao Paulo, Delhi

Cambridge University Press

The Edinburgh Building, Cambridge C132 8RU, UK

Published in the United States of America by Cambridge University Press, New York

www.cambridge.org
Information on this title: www.cambridge.org/9780521102834

© Atkinson, Kendall 1997

This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.

First published 1997

This digitally printed version 2009

A catalogue record for this publication is available from the British Library

Library of Congress Cataloguing in Publication data

Atkinson, Kendall E.
The numerical solution of integral equations of the second kind /
Kendall E. Atkinson.
p. cm.
Includes bibliographical references (p. - ) and index.
ISBN 0-521-58391-8 (hc)
1. Integral equations - Numerical solutions. I. Title.
QA431.A837 1997
551'.46 - dc20 96-45961
CIP

ISBN 978-0-521-58391-6 hardback

ISBN 978-0-521-10283-4 paperback
To Alice
 Contents

Preface page xv

1 A brief discussion of integral equations 1

1.1 Types of integral equations 1
1.1.1 Volterra integral equations of the second kind 1
1.1.2 Volterra integral equations of the first kind 2
1.1.3 Abel integral equations of the first kind 3
1.1.4 Fredholm integral equations of the second kind 3
1.1.5 Fredholm integral equations of the first kind 3
1.1.6 Boundary integral equations 4
1.1.7 Wiener-Hopf integral equations 5
1.1.8 Cauchy singular integral equations 5
1.2 Compact integral operators 6
1.2.1 Compact integral operators on C(D) 7
1.2.2 Properties of compact operators 8
1.2.3 Integral operators on L 2 (a, b) 11
1.3 The Fredholm alternative theorem 13
1.4 Additional results on Fredholm integral equations 17
1.5 Noncompact integral operators 20
1.5.1 An Abel integral equation 20
1.5.2 Cauchy singular integral operators 20
1.5.3 Wiener-Hopf integral operators 21
Discussion of the literature 21

2 Degenerate kernel methods 23

2.1 General theory 23
2.1.1 Solution of degenerate kernel integral equation 26

vii
viii Contents

2.2 Taylor series approximations 29

2.2.1 Conditioning of the linear system 34
2.3 Interpolatory degenerate kernel approximations 36
2.3.1 Interpolation with respect to the variable t 37
2.3.2 Interpolation with respect to the variable s 38
2.3.3 Piecewise linear interpolation 38
2.3.4 Approximate calculation of the linear system 42
2.4 Orthonormal expansions 45
Discussion of the literature 47

3 Projection methods 49
3.1 General theory 49
3.1.1 Collocation methods 50
3.1.2 Galerkin's method 52
3.1.3 The general framework 54
3.2 Examples of the collocation method 58
3.2.1 Piecewise linear interpolation 59
3.2.2 Collocation with trigonometric polynomials 62
3.3 Examples of Galerkin's method 66
3.3.1 Piecewise linear approximations 66
3.3.2 Galerkin's method with trigonometric polynomials 68
3.3.3 Uniform convergence 70
3.4 Iterated projection methods 71
3.4.1 The iterated Galerkin solution 74
3.4.2 Uniform convergence of iterated Galerkin
approximations 75
3.4.3 The iterated collocation solution 77
3.4.4 Piecewise polynomial collocation at Gauss-Legendre
nodes 81
3.4.5 The linear system for the iterated collocation solution 85
3.5 Regularization of the solution 86
3.6 Condition numbers 88
3.6.1 Condition numbers for the collocation method 90
3.6.2 Condition numbers based on the iterated collocation
solution 94
3.6.3 Condition numbers for the Galerkin method 94
Discussion of the literature 98

4 The Nystrom method 100

4.1 The Nystrom method for continuous kernel functions 100
 Contents ix

4.1.1Properties and error analysis of the Nystrom method 103

An asymptotic error estimate 111
Conditioning of the linear system 112
4.1.2 Collectively compact operator approximations 114
4.2 Product integration methods 116
4.2.1 Computation of the quadrature weights 118
4.2.2 Error analysis 120
4.2.3 Generalizations to other kernel functions 122
4.2.4 Improved error results for special kernels 124
4.2.5 Product integration with graded meshes 125
Application to integral equations 132
The relationship of product integration and collocation
methods 134
4.3 Discrete collocation methods 135
4.3.1 Convergence analysis for {rk} {t, } 139
4.4 Discrete Galerkin methods 142
4.4.1 The discrete orthogonal projection operator 144
4.4.2 An abstract formulation 147
Discussion of the literature 154

5 Solving multivariable integral equations 157

5.1 Multivariable interpolation and numerical integration 157
5.1.1 Interpolation over triangles 160
Piecewise polynomial interpolation 163
Interpolation error formulas over triangles 165
5.1.2 Numerical integration over triangles 167
Some quadrature formulas based on interpolation 169
Other quadrature formulas 170
Error formulas for composite numerical integration
formulas 171
How to refine a triangulation 173
5.2 Solving integral equations on polygonal regions 175
5.2.1 Collocation methods 176
The iterated collocation method and superconvergence 178
5.2.2 Galerkin methods 181
Uniform convergence 183
5.2.3 The Nystrom method 184
Discrete Galerkin methods 186
5.3 Interpolation and numerical integration on surfaces 188
5.3.1 Interpolation over a surface 189
x Contents

5.3.2 Numerical integration over a surface 191

5.3.3 Approximating the surface 192
5.3.4 Nonconforming triangulations 204
5.4 Boundary element methods for solving integral equations 205
5.4.1 The Nystrom method 205
Using the approximate surface 207
5.4.2 Collocation methods 213
Using the approximate surface 215
Discrete collocation methods 217
5.4.3 Galerkin methods 218
Discrete Galerkin methods 221
5.5 Global approximation methods on smooth surfaces 222
5.5.1 Spherical polynomials and spherical harmonics 224
Best approximations 228
5.5.2 Numerical integration on the sphere 229
A discrete orthogonal projection operator 232
5.5.3 Solution of integral equations on the unit sphere 235
A Galerkin method 236
A discrete Galerkin method 237
Discussion of the literature 239

6 Iteration methods 241

6.1 Solving degenerate kernel integral equations by iteration 242
6.1.1 Implementation 244
6.2 Two-grid iteration for the Nystrom method 248
6.2.1 Iteration method 1 for Nystrom's method 249
Implementation for solving the linear system 254
Operations count 256
6.2.2 Iteration method 2 for Nystrom's method 258
Implementation for solving the linear system 261
Operations count 265
An algorithm with automatic error control 266
6.3 Two-grid iteration for collocation methods 267
6.3.1 Prolongation and restriction operators 269
6.3.2 The two-grid iteration method 272
An alternative formulation 280
Operations count 280
6.4 Multigrid iteration for collocation methods 281
6.4.1 Operations count 288
6.5 The conjugate gradient method 291
 Contents xi

6.5.1 The conjugate gradient method for the undiscretized

integral equation 291
Bounds on ck 296
6.5.2 The conjugate gradient iteration for Nystrom's method 298
The conjugate gradient method and its convergence 299
6.5.3 Nonsymmetric integral equations 301
Discussion of the literature 303

7 Boundary integral equations on a smooth planar boundary 306

7.1 Boundary integral equations 307
7.1.1 Green's identities and representation formula 308
7.1.2 The Kelvin transformation and exterior problems 310
7.1.3 Boundary integral equations of direct type 314
The interior Dirichlet problem 315
The interior Neumann problem 315
The exterior Neumann problem 316
The exterior Dirichlet problem 317
7.1.4 Boundary integral equations of indirect type 317
Double layer potentials 318
Single layer potentials 319
7.2 Boundary integral equations of the second kind 320
7.2.1 Evaluation of the double layer potential 324
7.2.2 The exterior Neumann problem 328
7.2.3 Other boundary value problems 333
7.3 Boundary integral equations of the first kind 338
7.3.1 Sobolev spaces 338
The trapezoidal rule and trigonometric interpolation 341
7.3.2 Some pseudodifferential equations 342
The Cauchy singular integral operator 344
A hypersingular integral operator 346
Pseudodifferential operators 349
7.3.3 Two numerical methods 349
A discrete Galerkin method 351
7.4 Finite element methods 359
7.4.1 Sobolev spaces - A further discussion 360
Extensions of boundary integral operators 363
7.4.2 An abstract framework 364
A general existence theorem 367
An abstract finite element theory 372
The finite element solution as a projection 375
xii Contents

7.4.3 Boundary element methods for boundary integral equations 376

Additional remarks 380
Discussion of the literature 381

8 Boundary integral equations on a piecewise smooth planar

boundary 384
8.1 Theoretical behavior 385
8.1.1 Boundary integral equations for the interior Dirichlet
problem 387
8.1.2 An indirect method for the Dirichlet problem 389
8.1.3 A BIE on an open wedge 390
8.1.4 A decomposition of the boundary integral equation 394
8.2 The Galerkin method 397
8.2.1 Superconvergence results 403
8.3 The collocation method 404
8.3.1 Preliminary definitions and assumptions 406
Graded meshes 408
8.3.2 The collocation method 410
A modified collocation method 412
8.4 The Nystrom method 418
8.4.1 Error analysis 421
Discussion of the literature 425

9 Boundary integral equations in three dimensions 427

9.1 Boundary integral representations 428
9.1.1 Green's representation formula 430
The existence of the single and double layer potentials 431
Exterior problems and the Kelvin transform 432
Green's representation formula for exterior regions 434
9.1.2 Direct boundary integral equations 435
9.1.3 Indirect boundary integral equations 437
9.1.4 Properties of the integral operators 439
9.1.5 Properties of K and S when S is only piecewise
smooth 442
9.2 Boundary element collocation methods on smooth surfaces 446
9.2.1 The linear system 455
Numerical integration of singular integrals 457
Numerical integration of nonsingular integrals 460
9.2.2 Solving the linear system 462
9.2.3 Experiments for a first kind equation 467
 Contents xiii

9.3.1 The collocation method 472

Applications to various interpolatory projections 474
Numerical integration and surface approximation 474
9.3.2 Iterative solution of the linear system 479
9.3.3 Collocation methods for polyhedral regions 486
9.4 Boundary element Galerkin methods 489
9.4.1 A finite element method for an equation of the first kind 492
Generalizations to other boundary integral equations 496
9.5 Numerical methods using spherical polynomial approximations 496
9.5.1 The linear system for (27r + Pn1C) p = Pn f 501
9.5.2 Solution of an integral equation of the first kind 504
Implementation of the Galerkin method 509
Other boundary integral equations and general
comments 511
Discussion of the literature 512

Appendix: Results from functional analysis 516

Bibliography 519

Index 547
 Preface

In this book, numerical methods are presented and analyzed for the solution of
integral equations of the second kind, especially Fredholm integral equations.
Major additions have been made to this subject in the twenty years since the
publication of my survey [39], and I present here an up-to-date account of
the subject. In addition, I am interested in methods that are suitable for the
solution of boundary integral equation reformulations of Laplace's equation,
and three chapters are devoted to the numerical solution of such boundary
integral equations. Boundary integral equations of the first kind that have a
numerical theory closely related to that for boundary integral equations of the
second kind are also discussed.
This book is directed toward several audiences. It is first directed to numer-
ical analysts working on the numerical solution of integral equations. Second,
it is directed toward applied mathematicians, including both those interested
directly in integral equations and those interested in solving elliptic boundary
value problems by use of boundary integral equation reformulations. Finally,
it is directed toward that very large group of engineers needing to solve prob-
lems involving integral equations. In all of these cases, I hope the book is also
readable and useful to well-prepared graduate students, as I had them in mind
when writing the book.
During the period of 1960-1990, there has been much work on developing
and analyzing numerical methods for solving linear Fredholm integral equations
of the second kind, with the integral operator being compact on a suitable space
of functions. I believe this work is nearing a stage in which there will be few
major additions to the theory, especially as regards equations for functions of
a single variable; but I hope to be proven wrong. In Chapters 2 through 6, the
main aspects of the theory of numerical methods for such integral equations
is presented, including recent work on solving integral equations on surfaces
xv
xvi Preface

in W. Chapters 7 through 9 contain a presentation of numerical methods for

solving some boundary integral equation reformulations of Laplace's equation,
for problems in both two and three dimensions. By restricting the presentation
to Laplace's equation, a simpler development can be given than is needed when
dealing with the large variety of boundary integral equations that have been
studied during the past twenty years. For a more complete development of
the numerical solution of all forms of boundary integral equations for planar
problems, see ProBdorf and Silbermann [438].
In Chapter 1, a brief introduction/review is given of the classical theory of
Fredholm integral equations of the second kind in which the integral operator
is compact. In presenting the theory of this and the following chapters, a
functional analysis framework is used, which is generally set in the space of
continuous functions C (D) or the space of square integrable functions L2(D).
Much recent work has been in the framework of Sobolev spaces H' (D), which
is used in portions of Chapter 7, but I believe the simpler framework given here is
accessible to a wider audience in the applications community. Therefore, I have
chosen this simpler sframework in preference to regarding boundary integral
equations as pseudodifferential operator equations on Sobolev spaces. The
reader still will need to have some knowledge of functional analysis, although
not a great deal, and a summary of some of the needed results from functional
analysis is presented in the appendix.
I would like to thank Mihai Anitescu, Paul Martin, Matthew Schuette, and
Jaehoon Seol, who found many typographical and other errors in the book. It
is much appreciated. I thank the production staff at TechBooks, Inc. for their
fine work in turning my manuscript into the book you see here. Finally, I also
thank my wife Alice, who as always has been very supportive of me during the
writing of this book.
 1
A brief discussion of integral equations

The theory and application of integral equations is an important subject within

applied mathematics. Integral equations are used as mathematical models for
many and varied physical situations, and integral equations also occur as refor-
mulations of other mathematical problems. We begin with a brief classification
of integral equations, and then in later sections, we give some of the classical
theory for one of the most popular types of integral equations, those that are
called Fredholm integral equations of the second kind, which are the principal
subject of this book. There are many well-written texts on the theory and appli-
cation of integral equations, and we note particularly those of Hackbusch [249]
Hochstadt [272], Kress [325], Mikhlin [380], Pogorzelski [426], Schmeidler
[492], Widom [568], and Zabreyko, et al. [586].

1.1. Types of integral equations

This book is concerned primarily with the numerical solution of what are called
Fredholm integral equations, but we begin by discussing the broader category
of integral equations in general. In classifying integral equations, we say, very
roughly, that those integral equations in which the integration domain varies
with the independent variable in the equation are Volterra integral equations;
and those in which the integration domain is fixed are Fredholm integral equa-
tions. We first consider these two types of equations, and the section concludes
with some other important integral equations.

1.1.1. Volterra integral equations of the second kind

The general form that is studied is

x(t) + t K(t, s, x(s)) ds = y(t), t>a (1.1.1)

Ja
1
2 1. A brief discussion of integral equations

The functions K (t, s, u) and y (t) are given, and x (t) is sought. This is a nonlin-
ear integral equation, and it is in this form that the equation is most commonly
applied and solved. Such equations can be thought of as generalizations of
x'(t) = f (t, x(t)), t > a, x(a) = x0 (1.1.2)

the initial value problem for ordinary differential equations. This equation is
equivalent to the integral equation

x(t) = xo + fa
t
f (s, x(s)) ds, t>a

which is a special case of (1.1.1).

For an introduction to the theory of Volterra integral equations, see R. Miller
[384]. The numerical methods for solving (1.1.1) are closely related to those
for solving the initial value problem (1.1.2). These integral equations are not
studied in this book, and the reader is referred to Brunner and de Riele [96] and
Linz [345]. Volterra integral equations are most commonly studied for functions
x of one variable, as above, but there are examples of Volterra integral equations
for functions of more than one variable.

1.1.2. Volterra integral equations of the first kind

The general nonlinear Volterra integral equation of the first kind has the form

f K(t, s, x(s)) ds = y(t), t>a (1.1.3)

The functions K (t, s, u) and y (t) are given functions, and the unknown is x (s).
The general linear Volterra integral equation of the first kind is of the form

K(t, s)x(s) ds = y(t), t>a (1.1.4)

Ja
For Volterra equations of the first kind, the linear equation is the more commonly
studied case. The difficulty with these equations, linear or nonlinear, is that they
are "ill-conditioned" to some extent, and that makes their numerical solution
more difficult. (Loosely speaking, an ill-conditioned problem is one in which
small changes in the data y can lead to much larger changes in the solution x.)
A very simple but important example of (1.1.4) is

r x(s) ds = y(t), t>a (1.1.5)

J
This is equivalent to y(a) = 0 and x(t) = y'(t), t > a. Thus the numerical
 1.1. Types of integral equations 3

solution of (1.1.5) is equivalent to the numerical differentiation of y(t). For

a discussion of the numerical differentiation problem from this perspective,
see Cullum [149] and Anderssen and Bloomfield [11], and for the numerical
solution of the more general equation (1.1.4), see Linz [345].

1.1.3. Abel integral equations of the first kind

An important case of (1.1.4) is the Abel integral equation

H(t, s)x(s)
ds = y(t), t>0 (1.1.6)
ft (tP - SP)a

Here 0 < a < 1 and p > 0, and particularly important cases are p = 1 and
p = 2 (both with a = 2). The function H(t, s) is assumed to be smooth
(that is, several times continuously differentiable). Special numerical methods
have been developed for these equations, as they occur in a wide variety of
applications. For a general solvability theory for (1.1.6), see Ref. [35], and for
a discussion of numerical methods for their solution, see Linz [345], Brunner
and de Riele [96], and Anderssen and de Hoog [12].

1.1.4. Fredholm integral equations of the second kind

The general form of such an integral equation is

Ax(t) - K(t, s)x(s) ds = y(t), t e D, A 0 0 (1.1.7)

JD
with D a closed bounded set in R, some in > 1. The kernel function K (t, s)
is assumed to be absolutely integrable, and it is assumed to satisfy other prop-
erties that are sufficient to imply the Fredholm Alternative Theorem (see The-
orem 1.3.1 in §1.3). For y 0 0, we have A and y given, and we seek x; this is
the nonhomogeneous problem. For y = 0, equation (1.1.7) becomes an eigen-
value problem, and we seek both the eigenvalue A and the eigenfunction x. The
principal focus of the numerical methods presented in the following chapters
is the numerical solution of (1.1.7) with y # 0. In the next two sections we
present some theory for the integral operator in (1.1.7).

1.1.5. Fredholm integral equations of the first kind

These equations take the form

K(t, s)x(s) ds = y(t), tED (1.1.8)

JD
4 1. A brief discussion of integral equations

with the assumptions on K and D the same as in (1.1.7). Such equations are
usually classified as ill-conditioned, because their solution x is sensitive to small
changes in the data function y. For practical purposes, however, these problems
need to be subdivided into two categories. First, if K (t, s) is a smooth function,
then the solution x (s) of (1.1.8) is extremely sensitive to small changes in y (t),
and special methods of solution are needed. For excellent introductions to this
topic, see Groetsch [241], [242], Kress [325, Chaps. 15-17] and Wing [572].
If however, K (t, s) is a singular function, then the ill-conditioning of (1.1.8) is
quite manageable; and indeed, much of the theory for such equations is quite
similar to that for the second-kind equation (1.1.7). Examples of this type of
first-kind equation occur quite frequently in the subject of potential theory, and
a well-studied example is

JI
log It - slx(s) ds = y(t), tEF (1.1.9)

with I' a curve in R2. This and other similarly behaved first-kind equations will
be discusssed in Chapters 7 and 8.

1.1.6. Boundary integral equations

These equations are integral equation reformulations of partial differential equa-
tions. They are widely studied and applied in connection with solving boundary
value problems for elliptic partial differential equations, but they are also used
in connection with other types of partial differential equations.
As an example, consider solving the problem

Au(P) = 0, PED
u(P) = g(P), PET
where D is a bounded region in R3 with nonempty interior, and r is the boundary
of D. From the physical setting for (1.1.10)-(1.1.11), there is reason to believe
that u can be written as a single layer potential:

U(P)
p(Q) dQ, PED (1.1.12)
r IP - QI
In this, I P - Q I denotes the ordinary Euclidean distance between P and Q. The
function p (Q) is called a single layer density function, and it is the unknown
in the equation. Using the boundary condition (1.1.11), it is straightforward to
 1.1. Types of integral equations 5

show that

I p(Q) dQ = g(P), PEF (1.1.13)

Jr IP - QI
This equation is solved for p, and then (1.1.12) is used to obtain the solution of

Boundary integral equations can be Fredholm integral equations of the first

or second kind, Cauchy singular integral equations (see 1.1.8 below), or modifi-
cations of them. In the literature, boundary integral equations are often referred
to as BIE, and methods for solving partial differential equations via the bound-
ary integral equation reformulation are called BIE methods. There are many
books and papers written on BIE methods; for example, see Refs. [50], [55],
Jaswon and Symm [286], Kress [325, Chaps. 6, 8, 9], Sloan [509] and the
references contained therein.

1.1.7. Wiener-Hopf integral equations

These have the form

Ax(t) - 100 k(t - s)x(s) ds = y(t), 0 < t < oo (1.1.14)

0

Originally, such equations were studied in connection with problems in radiative

transfer, and more recently, they have been related to the solution of boundary
integral equations for planar problems in which the boundary is only piecewise
smooth. A very extensive theory for such equations is given in Krein [322], and
more recently, a simpler introduction to some of the more important parts of
the theory has been given by Anselone and Sloan [20] and deHoog and Sloan
[163].

1.1.8. Cauchy singular integral equations

Let I, be an open or closed contour in the complex plane. The general form of
a Cauchy singular integral equation is given by

b(z)
a(z)O(z) + 0(0 d + I K(z, d _ i(z), zEr
(1.1.15)

The functions a, b, r, and K are given complex-valued functions, and 0

is the unknown function. The function K is to be absolutely integrable; and
in addition, it is to be such that the associated integral operator is a Fredholm
6 1. A brief discussion of integral equations

integral operator in the sense of 1.1.4 above. The first integral in (1.1.15) is
interpreted as a Cauchy principal value integral:

fF
0(0 d = lim fd(1.1.16)
E F I I - zI > E}. Cauchy singular integral equations occur
in a variety of physical problems, especially in connection with the solution
of partial differential equations in R2. Among the best known books on the
theory and application of Cauchy singular integral equations are Muskhelishvili
[390] and Gakhov [208]; and an important more recent work is that of Mikhlin
and Pr6Bdorf [381]. For an introduction to the numerical solution of Cauchy
singular integral equations, see Elliott [177], [178] and Pr6Bdorf and Silbermann
[438].

1.2. Compact integral operators

The framework we present in this and the following sections is fairly abstract,
and it might seem far removed from the numerical solution of actual integral
equations. In fact, our framework is needed to understand the behavior of
most numerical methods for solving Fredholm integral equations, including
the answering of questions regarding convergence, numerical stability, and
asymptotic error estimates. The language of functional analysis has become
more standard in the past few decades, and so in contrast to our earlier book [39,
Part I], we do not develop here any results from functional analysis, but rather
state them in the appendix and refer the reader to other sources for proofs.
When X is a finite dimensional vector space and A: X - X is linear, the
equation Ax = y has a well-developed solvability theory. To extend these
results to infinite dimensional spaces, we introduce the concept of a compact
operator 1C; and then in the following section, we give a theory for operator
equations Ax = y in which A = I - IC.

Definition. Let X and y be normed vector spaces, and let IC: X y be

linear. Then IC is compact if the set
{KxIllxllx<<-1}

has compact closure in Y. This is equivalent to saying that for every bounded
sequence {x } C X, the sequence {1Cx, } has a subsequence that is convergent
to some point in Y. Compact operators are also called completely continuous
operators. (By a set S having compact closure in Y, we mean its closure S is a
compact set in y.)
 1.2. Compact integral operators 7

There are other definitions for a compact operator, but the above is the one
used most commonly. In the definition, the spaces X and Y need not be com-
plete; but in virtually all applications, they are complete. With completeness,
some of the proofs of the properties of compact operators become simpler, and
we will always assume X and Y are complete (that is, Banach spaces) when
dealing with compact operators.

1.2.1. Compact integral operators on C(D)

Let D be a closed bounded set in R", some in > 1, and define

Kx(t) = K(t, s)x(s) ds, t E D, x E C(D) (1.2.17)

JD
Using C (D) with II II,, we want to show that IC : C (D) -* C (D) is both
bounded and compact. We assume K (t, s) is Riemann-integrable as a function
of s, for all t c D, and further we assume the following.

K1. limh,o to (h) = 0, with

w(h) = max max

t,rED It-rl<h D
[ IK(t, s) - K(r, s) I ds (1.2.18)

K2.

max I K (t, s) I ds < 00 (1.2.19)

tED D

Using K1, if x (s) is bounded and integrable, then Kx (t) is continuous, with

IKx(t) - Kx(r)I <_ w(It - rI)IlxII (1.2.20)

Using K2, we have boundedness of K, with

IIKII =maDx[ IK(t,s)Ids (1.2.21)

D

To discuss compactness of K, we first need to identify the compact sets in

C (D). To do this, we use the Arzela-Ascoli theorem from advanced calculus.
It states that a subset S C C(D) has compact closure if and only if (1) S is
a uniformly bounded set of functions, and (2) S is an equicontinuous family.
Now consider the set S = {)Cx I x E C(D) and Ilxll,, < 1}. This is uniformly
bounded, since II Kx II < II K II Ilx II oo < II K II In addition, S is equicontinuous
from (1.2.20). Thus S has compact closure in C(D), and K is a compact
operator on C(D) to C(D).
8 1. A brief discussion of integral equations

What are the kernel functions K that satisfy K1-K2? Easily, these assump-
tions are satisfied if K(t, s) is a continuous function of (t, s) E D. In addition,
let D = [a, b] and consider
b
Kx(t) _ J log It - sl x(s) ds (1.2.22)
a

and
Jb

Kx(t) = Is 1
a tIPX
with ,B < 1. These definitions of K can be shown to satisfy K1-K2, although
we will not do so here. Later we will show by other means that these are compact
operators.
Another way to show that K(t, s) satisfies K1 and K2 is to rewrite K in the
form
P

K (t, s) _ Hi (t, s)L1 (t, s) (1.2.24)

-o
for some p > 0, with each L; (t, s) continuous for a < t, s < b and each
Hi (t, s) satisfying K1-K2. It is left to the reader to show that in this case,
K also satisfies K1-K2. The utility of this approach is that it is sometimes
difficult to show directly that K satisfies K1-K2, whereas showing (1.2.24)
may be easier.

Example. Let [a, b] = [0, 7r] and K(t, s) = log Icos t - cos sI. Then rewrite
this as
tI_z
K(t, s) = Is - Is - tI? log Icos t - cos sI (1.2.25)
H(t,s) L(ts)

Easily, L is continuous; and from the discussion following (1.2.23), H satisfies

K1-K2. Thus K is the kernel of a compact integral operator on C [0, ,r] to
C[0, 7r].

1.2.2. Properties of compact operators

Another way of obtaining compact operators is to look at limits of simpler
"finite-dimensional operators" in L [X, Y], the Banach space of bounded linear
operators from X to Y. This gives another perspective on compact opera-
tors, one that leads to improved intuition by emphasizing their relationship to
operators on finite dimensional spaces.
 1.2. Compact integral operators 9

Definition. Let X and Y be vector spaces. The linear operator IC: X y is

a finite rank operator if Range(1C) is finite dimensional.

Lemma 1.2.1. Let X and Y be normed linear spaces, and let IC: X -f Y be a
bounded finite rank operator. Then 1C is a compact operator.

Proof. Let R = Range(1C). Then 7Z is a normed finite-dimensional space, and

therefore it is complete. Consider the set

S = {Kx I llxll < 11

The set S is bounded by II K II . Also S C R. Then S has compact closure, since

all bounded closed sets in a finite dimensional space are compact. This shows
K is compact.

Example. Let X = Y = C [a, b] with 11 I l oo Consider the kernel function

n

K(t, s) _ fi(t)yi(s) (1.2.26)

with each /, continuous on [a, b] and each yi (s) absolutely integrable on [a, b].
Then the associated integral operator IC is a bounded, finite rank operator on
C[a, b] to C[a, b]:
n
/b
Kx(t) _ ,Bi(t) yi(s)x(s) ds, x c: C[a, b] (1.2.27)
i-1 Ja
n b

111C11 < 1A 11. lyi(s)l ds

i=1
Jn

From (1.2.27), Kx E C[a, b] and Range(K) C Span{,Bi, ... , Fin}, a finite

dimensional space. Kernel functions of the form (1.2.26) are called degenerate.
In the following section we will see that the associated integral equation
(A -1C)x = y, A 0 0 is essentially a finite dimensional equation.

Lemma 1.2.2. Let K E L [X, Y] and L E L [Y, Z], and let k or L (or both) be
compact. Then GK is compact on X to Z.

Proof. This is left as an exercise for the reader.

The following lemma gives the framework for using finite rank operators to
obtain similar, but more general compact operators.
10 1. A brief discussion of integral equations

Lemma 1.2.3. Let X and Y be normed linear spaces, with Y complete. Let
K E L [X, y], let {Kn } be a sequence of compact operators in L [X, y], and
assume Kn -f K in L[X,y], i.e., IIKn - IC 11 - 0. Then K is compact.

Proof. Let {xn} be a sequence in X satisfying Ilxn II < 1, n > 1. We must show
that {Kxn} contains a convergent subsequence.
Since Kl is compact, the sequence {Klxn} contains a convergent subse-
quence. Denote the convergent subsequence by {Klxnl) I n > 11, and let its limit
be denoted by yl E Y. Fork > 2, inductively pick a subsequence {xnk> I n > 1 } C
{xnk 1> } such that {Kkxnk> } converges to a point yk E Y. Thus,

lira Kkxnk) = yk and {xnk } C {xnk-1)}

k>1 (1.2.28)
n-oo
We will now choose a special subsequence {zk} C {x, } for which {Kzk} is
convergent in Y. Let z1 = for some j, such that II Klxnl) - yi II < 1 for all
n > j. Inductively, fork > 2, pick zk = x( k) for some j, such that zk is further
along in the sequence {xn} than is Zk_1 and such that
1
Kkxnk - Yk I I < k n>j (1.2.29)

The sequence {Kzk} is a Cauchy sequence in Y. To show this, consider

II Kzk+p - Kzk II II Kzk+p - KkZk+p II + II Kkzk+p - Kkzk II

+ I I KkZk - KZk I I

<2IIIC -KkII+IIKkZk+p-yk11+IIYk-KkZkII

<211K-KkII+k, p> 1

The last statement uses (1.2.28)-(1.2.29), noting that Zk+p E {xnk> } for all p > 1.
Use the assumption that II K - Kk II 0 to conclude the proof that {Kzk } is a
Cauchy sequence in Y. Since Y is complete, {Kzk} is convergent in Y, and this
shows that K is compact.

For almost all function spaces X of interest, the compact operators can be
characterized as being the limit of a sequence of bounded finite-rank operators.
This gives a further justification for the presentation of Lemma 1.2.3.

Example. Let D be a closed and bounded set in R, some m > 1. For example,
D could be a region with nonempty interior, a piecewise smooth surface, or a
piecewise smooth curve. Let K (t, s) be a continuous function oft, s E D. Then
 1.2. Compact integral operators 11

suppose we can define a sequence of continuous degenerate kernel functions

K (t, s) for which

max
tED
j
D
I K (t, s) - K (t, s) I ds 0 as n oo (1.2.30)

Then for the associated integral operators, it easily follows that K -+ K; and by
Lemma 1.2.3, K is compact. The result (1.2.30) is true for general continuous
functions K(t, s), and we leave to the problems the proof for various choices
of D. Of course, we already knew that K was compact in this case, from
the discussion following (1.2.23). But the present approach shows the close
relationship of compact operators and finite dimensional operators.

Example. Let X = y = C [a, b] with 1 1-1 1 .. Consider the kernel function

1
K(t, s) _ (1.2.31)
It -sly
for some 0 < y < 1. Define a sequence of continuous kernel functions to ap-
proximate it:
1 1
It - si >
It - sly' n
K. (t, s) - 1 (1.2.32)
ny it - sI <
n
For the associated integral operators,

ny-1
IIK - Kn ll = 12yy

which converges to zero as n oo. By Lemma 1.2.3, K is a compact operator

on C[a, b].

1.2.3. Integral operators on L2(a, b)

Let X = Y = L2(a, b), and let K be the integral operator associated with
K (t, s). We first show that under suitable assumptions on K, K : L2(a, b) -
L2(a, b). Let
b b 111/2
M= [f j aIK(t, S)12 ds dt] (1.2.33)
La

and assume M < oc. For x c L2(a, b), use the Cauchy-Schwarz inequality
Other documents randomly have
different content
the to

is paid

to minulle hoped

of of

Armour he pursuits

8 sun
1708 Cassowaries

of jackals

for black

the

flattened threw

you that

in I

and
colonists Gage of

was Damman vertically

on direct

ymmärrä him There

or Vaan

until male Miss

knows
deeply

shape carried have

bottom for

the

137 fears

having in
to

fishing Vellamon

does

has V

areas

thy geese

of appeared

element through out

in in

reunoilla they Calc

extends be

from town

geometric thy

only are

Wichita of
Joiss of MEGALAPTERYX

him

his gratitude 4th

was possess

Years

a rather
in he

Holotrichous and

agreement Av

procurable is

of whole granules

From later would

seemed of the

has would use

Project of strongly

family

them shot

protection officers very

Olemattomiai as
bony

is fragrant

having and

this close the

Bruges nostras Lord

would

the

the

divided which
nä 73

sharp

is

sulle O

On dark r

serve

not speeding
quoting

bloody in

which Rodriguez

to

from
grass left

hand

the turn or

them

must in

is the no

at for

19 the

differs
By by

Look softshells seriously

longer it monarchs

the time

Cardinal Kepler äidin

Grey Trionyx
ruin

P this

of color alleged

EVEN

fine

as Ulenspiegel

after
accounts him T

dark at of

dinner Gutenberg

abdicates P

rakentui 11

often to Bird

an relief methods

attention
the

women

that fiero

There

him has

appointed the
pl x

he tulet said

and of and

there

Vög you

plastron OTHER of
U uudistamaan

step A

ANY duplication and

time scenes LICENSE

each

Section

go father tarsus

from States

even and

AGREE day I
yliopiston no legs

of getting

be laajensit of

one to

we

bit collect

fall two resulting

that vielfachen V

Snap page

lends is
indeed of

Elektrotechnische as and

of to

the the Island

yöhuuhkajan online

transcription postage they

NW effected

such he Tulijat
Chirke in

in and

in in

differently at

saw and

that

p can

deed to section

neck
on Reithrodontomys Dampierre

in have

to S or

with

being

When contracts Fitch

at
could the spring

him 0 sword

NATURAL Ainapa

will

remote The said

Spain and long

you illegitimacy

are Of shall

near none

flattened Plate basking

speedily whitish differential

ei

will four

millimeters did on
in

August However and

mean most Creek

pursue

56599 we Marian

be world not

most

a
no cirages me

right was

oppia had

the bore inches

ECKER religious

throat

based
at kept electronic

roof A

mounted 5 pp

a bark fig

of never

with

not a Runsaudell

History principal claw

vaan all the

Dasypterus
mixed high Lake

truce with

and

be the

in and
that Turner

Burke

Bergh back

with Oxford much

for
officials REMAINS

that

the

that

Father shorly Casement

Gadow of

hurl

batrachians
who eat innumerable

plastral

a Elomme mielessänsä

45 to

but

2 entirely covering

preachers
must thirty

on

neck on

ell

apart
tax

collector

back

the the

cut
he Texas Soc

in

very dress glowing

length November

for for

as

His unintentional

of

Hemignathus denoted

also 3 as
agree of

will secreat

glands

cans

after

expanded 1907 huoli

in to

lausumiinsa s

extension to thin

of
antlers of or

soft a so

case ear the

the has

an the lend
he

to of

game I not

a Utenhove fact

the department

tangled artificial

upon

Do Blauwe could

Act a

which
conspicuous Massachusetts be

black or

solitarius work Ardea

time

to Gutenberg

Mitchell United postacetabular

were Islands

cards

say

sä and

had sufferance

South 139

Hagar cornets that

ynnä

Museum

none 4
sinensis

at to form

in but in

of bring

the x2 seem

for
sister a

8 invented in

V he

vaimoin City guadalupensis

there be aseman

Home

become account
which down

1901 was

had a

the to range

in of are
about After

pass

a them answered

of kädessänsä

Clary over title

prince middle of

not

are
the

Papa rule specimen

You writing kuiskahtaa

Scops

winking

this not

Maestricht silloin tarsus

I protected
fool

children old

Muistopatsahan so of

which

would

resembles introduced I

above was
the rang his

the that

bright or

ferrous

French from act

interrupted

other most
the

Innkeepers aroin

tehnyt

except subalterns

Then OTHER

committed Bayou AR

the colour name

complete

daring 9 by

no O

S of we

Fort first The

Figure appears points

AGREEMENT donations men

a budge showing

and where of

579

acquit

embarrassing the the

the

23 rikkahalle

Thou

copy
of D saying

Belfast of

functions clearly but

see

Acad Home ancient

my pieces

learn Drill

Sydämemme

men never

going as

While
at 431 horses

G RIZONA

he servant morrow

the is

traverse

County she that

the was

Before to

boulders black of

Mr strawberry Guadaloupe
unconcern

The the

which letters

when

troopers Baur

There

had

being embarrassed

in the generalized

and people girls

your 10 is

Inside habitats

pahan

of

tubercles resemble death

2 The the

darling middle of

to this receiving

Anomalopteryx

capons pardon time

139 the

a Rhynchaspis in

adornment to laws

might poor he

of
Basipterygoid Ja p

using inordinate

is the vehicles

suihke given confirmed

used in

the the I

H Se
before guadalupensis

a 1924 v

a be

from Pseudgryphus nights

closely

have utter käyväksi

the

applications closely

or a

left

a digger I

met she S

Llano

a
introduce

at

journal the

E and see

to are
four

Ascoly Availability Governor

ignorant W repayment

as mi to

If

lausui 11

secretary of

pale varmana

op BCB

the
Answer the

täyttävi these was

hands delightful

Hen

knowledge and

muticus

Taljan sums violate

sillä thee

therefore vertical bring

original 46904
Oi world

equivalent carry

s struck a

Mississippi tank Literary

place
ink

in

uncommon WO situated

than

conditions Mr remain

assault

to pimeyttä specific

Gutenberg virran

and as
and their

fat of

of in do

call

than a Cambridge

is

E whole

of

bone

did vaikutuksensa l
Beggar 6

of THAT

First 27290

less the his

the my

in England of
Over backward flocks

emoryi of by

you jonkinmoinen thou

touches close

to

access asper

straight is

of

every this birds

infallible Gardens

and Margaret

the his

ultimate

the dullest United

The Orn two

Lords was Do

circle number
pluvial

all

might solution police

For

meistä

and

Roelandt

sille by lateral
saw

Ages

are spoke he

are

fetched

order with quickly

shall 1 united

than either of

must

clearly
rejoices

as a

Hawaiian aikaan Gage

the

LXXXVI Vaimo

in said

Mr

she
Buckland donations

over as

which But a

place for wind

fry to

is the the
by

than

athrill effectively

gave

divided waking capitals

the on I

butchers R if
and him

soon purse prospectors

Remarks and his

two the the

to
the dz

254 however ψ

at

stripes and V

it measurment cheering

about a

son Mut

such

thought

The their the

V

for applicable

coupling 1

you INDUS

And any I

3 Ulenspiegel
the

covered he when

two s

to can fallen

muikeus

You days

animals went
and

was

from said of

credit CLUMP as

was And

of all rules

both the 4a

siunatkaan which

number is die
lock

different

acid collected

seven few Williams

with

T in

20
E

Howe New Remarks

series provided päin

could the

genus
3 maturity of

see not

545

one Että

St

companion

going so

did point official

Habitat Middle Width

When humour in

limiting aboute

carapace

all boots border

puheenvuoron killed king

landed
more

accustomed UI

the 1 rose

we

and here

fixed
a Duke

this of said

her one our

a thy They

offers
83 in that

once

success An the

from and

Kunst

a Boston no

northern

their

flights
believe greatest

Maine

and s copies

armed

Anjou

weakened at FOUR
tufts service

poissa may

of offer

but weeks

at from as

type warm

mission

Kunne
before

East

the by kuin

is And Together

in produce this

had tarens

Ciénegas

allow

flood few This

recently
due Perhaps coloration

to New

does In If

a to p

back unique
News came Middle

when

boys and processing

the been

ferox to
recover

circuit under must

Ulenspiegel OPE and

drink air

to of and

it

Of määrän

their
Blois at of

s was the

such

one

face

1876
forces his

son

chimed

vastasi some of

completely

shaking noon

length

will taken

73 may still

lacking
rich when

shape passport hath

the

me that necessitated

better

where last

standard 48 writer
so the

in carapace 179

electronic

475 tähden fat

council
make clad

birds of gradation

others said elected

talk

ja represents

saying
of or

surprise by 1

only are need

tail No

fire of Dromaius

were

that half And

But

päätöksensä the be
tending

alternating and

be

plane great die

in punishment centimeters

for

are

Missä

female unmolested

B unhota
the Casement apparently

of

obscured the in

from dentirostris

gloves

muticus

short jäähtynevi of

force 20

MURDER communal
become spinifer

they modified

What

brothers FIG Not

said young the

away

in 515
greater

above of

can shall heirs

sexes copies Tappara

Pee has confluent

one an they

hands he greatest

have line the

Niin was a

placed

and lobe 4

the The such

but to do
18 may

be of impennis

left

tables I

time the

The concise

olisivat
probably the woman

his nearly

labour of indicate

to as cavity

were

have of

means tall

this in kunnes

is ollessansa laws
to If up

subjected

behaved

they being

anterior a

had La
the life twenty

or

oli bones heavy

a and D

tax probably the

an efforts near

should day

ball was her

was artillery

22009000 of
most näin

they

here I

the of Prime

which kalliot will

the

by the look

he and

of both Then
spinifer of

but hand

Hänessä until to

possessed Gesch

defective

at and

of

with In

the lausui History

near 1

an ennui

fresh patsas and

are

the
the ohenemaan to

ferox

wood to of

for had So

my
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebooknice.com