248 98 2MB
English Pages 319 [341] Year 2020
GRADUATE STUDIES I N M AT H E M AT I C S
205
Invitation to Partial Differential Equations Mikhail Shubin E D I T E D BY
Maxim Braverman Robert McOwen Peter Topalov
10.1090/gsm/205
Invitation to Partial Differential Equations
GRADUATE STUDIES I N M AT H E M AT I C S
205
Invitation to Partial Differential Equations Mikhail Shubin E D I T E D BY
Maxim Braverman Robert McOwen Peter Topalov
EDITORIAL COMMITTEE Daniel S. Freed (Chair) Bjorn Poonen Gigliola Staļ¬lani Jeļ¬ A. Viaclovsky This work was originally published in Russian by MCNMO under the title āLekcii c ob uravnenih matematiqesko i fizikiā2001. The present translation was created under license for the American Mathematical Society and is published by permission. 2010 Mathematics Subject Classiļ¬cation. Primary 35-01.
For additional information and updates on this book, visit www.ams.org/bookpages/gsm-205
Library of Congress Cataloging-in-Publication Data Names: Shubin, M. A. (Mikhail Aleksandrovich), 1944- author. | Braverman, Maxim, 1966- editor. | McOwen, Robert C., editor. | Topalov, P., 1968- editor. Title: Invitation to partial diļ¬erential equations / Mikhail Shubin ; edited by Maxim Braverman, Robert McOwen, Peter Topalov. Description: Providence, Rhode Island : American Mathematical Society, [2020] | Series: Graduate studies in mathematics, 1065-7339 ; 205 | Includes bibliographical references and index. Identiļ¬ers: LCCN 2020001923 | ISBN 9780821836408 (hardcover) | ISBN 9781470456979 (ebook) Subjects: LCSH: Diļ¬erential equations, PartialāTextbooks. | Diļ¬erential equations, Partialā Problems, exercises, etc. | AMS: Partial diļ¬erential equations ā Instructional exposition (textbooks, tutorial papers, etc.). Classiļ¬cation: LCC QA374 .S518 2020 | DDC 515/.353ādc23 LC record available at https://lccn.loc.gov/2020001923
Copying and reprinting. Individual readers of this publication, and nonproļ¬t libraries acting for them, are permitted to make fair use of the material, such as to copy select pages for use in teaching or research. Permission is granted to quote brief passages from this publication in reviews, provided the customary acknowledgment of the source is given. Republication, systematic copying, or multiple reproduction of any material in this publication is permitted only under license from the American Mathematical Society. Requests for permission to reuse portions of AMS publication content are handled by the Copyright Clearance Center. For more information, please visit www.ams.org/publications/pubpermissions. Send requests for translation rights and licensed reprints to [email protected]. c 2020 by the American Mathematical Society. All rights reserved. The American Mathematical Society retains all rights except those granted to the United States Government. Printed in the United States of America. ā The paper used in this book is acid-free and falls within the guidelines
established to ensure permanence and durability. Visit the AMS home page at https://www.ams.org/ 10 9 8 7 6 5 4 3 2 1
25 24 23 22 21 20
Contents
Foreword
xi
Preface
xiii
Selected notational conventions
xvii
Chapter 1. Linear diļ¬erential operators
1
Ā§1.1. Deļ¬nition and examples
1
Ā§1.2. The total and the principal symbols
2
Ā§1.3. Change of variables
4
Ā§1.4. The canonical form of second-order operators with constant coeļ¬cients
5
Ā§1.5. Characteristics. Ellipticity and hyperbolicity
7
Ā§1.6. Characteristics and the canonical form of second-order operators and second-order equations for n = 2
8
Ā§1.7. The general solution of a homogeneous hyperbolic equation with constant coeļ¬cients for n = 2
11
Ā§1.8. Appendix. Tangent and cotangent vectors
12
Ā§1.9. Problems
14
Chapter 2. One-dimensional wave equation
15
Ā§2.1. Vibrating string equation
15
Ā§2.2. Unbounded string. The Cauchy problem. DāAlembertās formula
20
v
vi
Contents
Ā§2.3. A semibounded string. Reļ¬ection of waves from the end of the string
24
Ā§2.4. A bounded string. Standing waves. The Fourier method (separation of variables method)
26
Ā§2.5. Appendix. The calculus of variations and classical mechanics
33
Ā§2.6. Problems
39
Chapter 3. The Sturm-Liouville problem
43
Ā§3.1. Formulation of the problem
43
Ā§3.2. Basic properties of eigenvalues and eigenfunctions
44
Ā§3.3. The short-wave asymptotics
47
Ā§3.4. The Greenās function and completeness of the system of eigenfunctions
50
Ā§3.5. Problems
54
Chapter 4. Distributions
57
Ā§4.1. Motivation of the deļ¬nition. Spaces of test functions
57
Ā§4.2. Spaces of distributions
63
Ā§4.3. Topology and convergence in the spaces of distributions
67
Ā§4.4. The support of a distribution
70
Ā§4.5. Diļ¬erentiation of distributions and multiplication by a smooth function
74
Ā§4.6. A general notion of the transposed (adjoint) operator. Change of variables. Homogeneous distributions
87
Ā§4.7. Appendix. Minkowski inequality
91
Ā§4.8. Appendix. Completeness of distribution spaces
94
Ā§4.9. Problems
96
Chapter 5. Convolution and Fourier transform Ā§5.1. Convolution and direct product of regular functions
99 99
Ā§5.2. Direct product of distributions
101
Ā§5.3. Convolution of distributions
104
Ā§5.4. Other properties of convolution. Support and singular support of a convolution
107
Ā§5.5. Relation between smoothness of a fundamental solution and that of solutions of the homogeneous equation
109
Ā§5.6. Solutions with isolated singularities. A removable singularity theorem for harmonic functions
113
Contents
vii
Ā§5.7. Estimates of derivatives of a solution of a hypoelliptic equation
114
Ā§5.8. Fourier transform of tempered distributions
116
Ā§5.9. Applying the Fourier transform to ļ¬nd fundamental solutions
120
Ā§5.10. Liouvilleās theorem
121
Ā§5.11. Problems
123
Chapter 6. Harmonic functions
127
Ā§6.1. Mean-value theorems for harmonic functions
127
Ā§6.2. The maximum principle
129
Ā§6.3. Dirichletās boundary value problem
131
Ā§6.4. Hadamardās example
133
Ā§6.5. Greenās function for the Laplacian
136
Ā§6.6. HĀØolder regularity
140
Ā§6.7. Explicit formulas for Greenās functions
143
Ā§6.8. Problems
147
Chapter 7. The heat equation
151
Ā§7.1. Physical meaning of the heat equation
151
Ā§7.2. Boundary value problems for the heat and Laplace equations
153
Ā§7.3. A proof that the limit function is harmonic
155
Ā§7.4. A solution of the Cauchy problem for the heat equation and Poissonās integral
156
Ā§7.5. The fundamental solution for the heat operator. Duhamelās formula
161
Ā§7.6. Estimates of derivatives of a solution of the heat equation
164
Ā§7.7. Holmgrenās principle. The uniqueness of solution of the Cauchy problem for the heat equation
165
Ā§7.8. A scheme for solving the ļ¬rst and second initial-boundary value problems by the Fourier method
168
Ā§7.9. Problems
170
Chapter 8. Sobolev spaces. A generalized solution of Dirichletās problem Ā§8.1. Spaces
171
H s (Ī©)
171
Ės (Ī©) Ā§8.2. Spaces H
177
viii
Contents
Ā§8.3. Dirichletās integral. The Friedrichs inequality
181
Ā§8.4. Dirichletās problem (generalized solutions)
183
Ā§8.5. Problems
190
Chapter 9. The eigenvalues and eigenfunctions of the Laplace operator
193
Ā§9.1. Symmetric and selfadjoint operators in Hilbert space
193
Ā§9.2. The Friedrichs extension
197
Ā§9.3. Discreteness of spectrum for the Laplace operator in a bounded domain
201
Ā§9.4. Fundamental solution of the Helmholtz operator and the analyticity of eigenfunctions of the Laplace operator at the interior points. Besselās equation
202
Ā§9.5. Variational principle. The behavior of eigenvalues under variation of the domain. Estimates of eigenvalues
209
Ā§9.6. Problems
212
Chapter 10. The wave equation
215
Ā§10.1. Physical problems leading to the wave equation
215
Ā§10.2. Plane, spherical, and cylindric waves
220
Ā§10.3. The wave equation as a Hamiltonian system
222
Ā§10.4. A spherical wave caused by an instant ļ¬ash and a solution of the Cauchy problem for the three-dimensional wave equation 228 Ā§10.5. The fundamental solution for the three-dimensional wave operator and a solution of the nonhomogeneous wave equation
234
Ā§10.6. The two-dimensional wave equation (the descent method)
236
Ā§10.7. Problems
239
Chapter 11. Properties of the potentials and their computation
241
Ā§11.1. Deļ¬nitions of potentials
242
Ā§11.2. Functions smooth up to Ī from each side, and their derivatives
244
Ā§11.3. Jumps of potentials
251
Ā§11.4. Calculating potentials
253
Ā§11.5. Problems
256
Contents
ix
Chapter 12. Wave fronts and short-wave asymptotics for hyperbolic equations
257
Ā§12.1. Characteristics as surfaces of jumps
257
Ā§12.2. The Hamilton-Jacobi equation. bicharacteristics, and rays
262
Wave fronts,
Ā§12.3. The characteristics of hyperbolic equations
269
Ā§12.4. Rapidly oscillating solutions. The eikonal equation and the transport equations
271
Ā§12.5. The Cauchy problem with rapidly oscillating initial data
281
Ā§12.6. Problems
287
Chapter 13. Answers and hints. Solutions
289
Bibliography
311
Index
315
Foreword
Misha Shubin, our dear colleague at Northeastern University, worked for many years trying to ļ¬nish this book on partial diļ¬erential equations. Unfortunately, his failing health prevented him from completing this task. Sergei Gelfand of the American Mathematical Society approached us asking whether we could put the ļ¬nishing touches on the manuscript so that it could be published. We agreed to do this hoping to preserve Mishaās writing style which is at the same time rigorous yet accessible. Many chapters in the book required only minor changes; others required substantial revision and additions because they had not been completed before his health deteriorated. It has taken us quite a while to ļ¬nish this project, but we now feel that the book is in suļ¬ciently good condition that it can be shared with the mathematical community. We hope that it will be appreciated as a lasting tribute to the legacy of Professor Mikhail Shubin. Maxim Braverman Robert McOwen Peter Topalov Boston, November 2019
xi
Preface
I shall be telling this with a sigh Somewhere ages and ages hence: Two roads diverged in a wood, and I ā I took the one less traveled by, And that has made all the diļ¬erence. From The Road Not Taken by Robert Frost
It will not be an overstatement to say that everything we see around us (as well as hear, touch, etc.) can be described by partial diļ¬erential equations or, in a slightly more restricted way, by equations of mathematical physics. Among the processes and phenomena whose adequate models are based on such equations, we encounter, for example, string vibrations, waves on the surface of water, sound, electromagnetic waves (in particular, light), propagation of heat and diļ¬usion, gravitational attraction of planets and galaxies, behavior of electrons in atoms and molecules, and also the Big Bang that led to the creation of our Universe. It is no wonder, therefore, that today the theory of partial diļ¬erential equations (PDE) forms a vast branch of mathematics and mathematical physics that uses methods of all the remaining parts of mathematics (from which PDE theory is inseparable). In turn, PDE inļ¬uence numerous parts of mathematics, physics, chemistry, biology, and other sciences. A disappointing conclusion is that no book, to say nothing of a textbook, can be suļ¬ciently complete in describing this area. Only one possibility remains: to show the reader pictures at an exhibitionāseveral typical xiii
xiv
Preface
methods and problemsāchosen according to the authorās taste. In doing so, the author has to severely restrict the choice of material. This book originally contained a practically intact transcript of the PDE course which I taught twice to students in experimental groups at the Department of Mechanics and Mathematics of the Moscow State University and later, in a shortened version, in Northeastern University (Boston). Since then several other people taught this course in other universities (in particular, I. S. Louhivaara was the ļ¬rst to teach it outside Moscow State Universityānamely, at Freie UniversitĀØat (Berlin). Now the transcript has been considerably extended, becoming practically a new book. In Moscow, all problems presented in this book were used in recitations accompanying the lectures. There was one lecture per week during the whole academic year (two semesters); recitations were held once a week during the ļ¬rst semester and twice a week during the second one. My intention was to create a course that is both concise and modern. These two requirements proved to be contradictory. To make the course concise, I was, ļ¬rst, forced to explain some ideas on simplest examples and, second, to omit many topics which should be included in a modern course but would have expanded the course beyond a reasonable length. With particular regret I have omitted the SchrĀØodinger equation and related material. I hope the gap will be ļ¬lled by a parallel course on quantum mechanics. In any modern course on equations of mathematical physics, something should, perhaps, be said about nonlinear equations but I found it extremely diļ¬cult to select this āsomethingā. The bibliography at the end of the book contains a number of textbooks where the reader can ļ¬nd information complementing this course. Naturally, the material of theses books overlaps, but it was diļ¬cult to do a more precise selection, so I leave this selection to the readerās taste. The problems in the book are not just exercises. Some of them add essential information but require no new ideas to solve them. Acknowledgments. I am very grateful to Professor S. P. Novikov, who organized the experimental groups and invited me to teach PDE to them. I am also greatly indebted to the professors of the chair of diļ¬erential equations of the Moscow State University for many valuable discussions. I owe a lot to Professor Louhivaara, who provided unexpectedly generous help in editing an early version of this text and encouraged me to publish this book. I also had very productive discussions about this course with Professor P. Albin, which led to extremely many useful improvements and corrections. Professors R. Mazzeo and I. Polterovich taught the course using the lecture notes, and I owe them thanks for many useful suggestions.
Preface
xv
I am thankful to Professor O. Milatovic for his help in verifying problems. I am also grateful to Professor D. A. Leites, who participated considerably in the translation of this book into English.
Selected notational conventions
Cross references inside the book are natural: Theorem 6.2 stands for Theorem 2 in Chapter 6, Problem 6.2 is Problem 2 from Chapter 6, etc. := means ādenotes by deļ¬nitionā. ā” means āidentically equal toā. By Z, Z+ , N, R, and C we will denote the set of all integers, nonnegative integers, positive integers, real numbers, and complex numbers, respectively. I denotes the identity operator. 1 The Heaviside function is Īø(z) = 0
for z ā„ 0, for z < 0.
xvii
10.1090/gsm/205/01
Chapter 1
Linear diļ¬erential operators
1.1. Deļ¬nition and examples We will start by introducing convenient notation for functions of several variables and diļ¬erential operators. By Z, R, and C we will denote the set of integer, real, and complex numbers, respectively. A multi-index Ī± is an ntuple Ī± = (Ī±1 , . . . , Ī±n ), where Ī±j ā Z+ (i.e., the Ī±j are nonnegative integers). If Ī± is a multi-index, then we set |Ī±| = Ī±1 + Ā· Ā· Ā· + Ī±n and Ī±! = Ī±1 ! . . . Ī±n !; for a vector x = (x1 , . . . , xn ) ā Cn we deļ¬ne xĪ± = xĪ±1 1 . . . xĪ±n n . If Ī© is an open subset of Rn , then C ā (Ī©) denotes the space of complex-valued inļ¬nitely diļ¬erentiable functions on Ī©, and C0ā (Ī©) denotes the subspace of inļ¬nitely diļ¬erentiable functions with compact support; i.e., Ļ ā C0ā (Ī©) if Ļ ā C ā (Ī©) and there exists a compact K ā Ī© such that Ļ vanishes on Ī© \ K. (Here K depends on Ļ.) ā Denote āj = āxā j : C ā (Ī©) āā C ā (Ī©), Dj = āiāj , where i = ā1, D = (D1 , . . . , Dn );
ā Ī± = ā1Ī±1 . . . ānĪ±n ,
D Ī± = D1Ī±1 . . . DnĪ±n .
ā |Ī±| Ī± n āx1 1 ...āxĪ± n ā|Ī±| Ī± i ā .
Therefore, ā Ī± = C ā (Ī©), D Ī± =
ā = (ā1 , . . . , ān ),
is a mixed derivative operator ā Ī± : C ā (Ī©) āā
If f ā C ā (Ī©), then instead of ā Ī± f we will sometimes write f (Ī±) . 1
2
1. Linear diļ¬erential operators
Exercise 1.1. Let f ā C[x1 , . . . , xn ]; i.e., f is a polynomial of n variables x1 , . . . , xn . Prove the Taylor formula (1.1)
f (x) =
f (Ī±) (x0 ) Ī±
Ī±!
(x ā x0 )Ī± ,
where the sum runs over all multi-indices Ī±. (Actually, the sum is ļ¬nite; i.e., only a ļ¬nite number of terms may be nonvanishing.) A linear diļ¬erential operator is an operator A : C ā (Ī©) āā C ā (Ī©) of the form aĪ± (x)D Ī± , (1.2) A= |Ī±|ā¤m
where aĪ± (x) ā C ā (Ī©). Clearly, instead of D Ī± we may write ā Ī± , but the expression in terms of D Ī± is more convenient as we will see in the future. Here m ā Z+ , and we will say that A is an operator of order ā¤ m. We say that A is an operator of order m if it can be expressed in the form (1.2) and there exists a multi-index Ī± such that |Ī±| = m and aĪ± (x) ā” 0. Examples. 1) The Laplace operator, or Laplacian, Ī = ā(D12
+Ā·Ā·Ā· +
Dn2 ).
ā2 āx21
+ Ā·Ā·Ā· +
ā2 āx2n
=
ā 2) The heat operator āt ā Ī (here the number of independent variables is equal to n + 1 and they are denoted by t, x1 , . . . , xn ).
3) The wave operator, or dāAlembertian, =
ā2 āt2
ā Ī.
4) The Sturm-Liouville operator L deļ¬ned for n = 1 and Ī© = (a, b) by the formula d du Lu = p(x) + q(x)u, dx dx where p, q ā C ā ((a, b)). Operators in Examples 1)ā3) are operators with constant coeļ¬cients. The Sturm-Liouville operator has variable coeļ¬cients.
1.2. The total and the principal symbols The symbol, or the total symbol, of the operator (1.2) is aĪ± (x)Ī¾ Ī± , x ā Ī©, Ī¾ ā Rn , (1.3) a(x, Ī¾) = |Ī±|ā¤m
and its principal symbol is the function aĪ± (x)Ī¾ Ī± , (1.4) am (x, Ī¾) = |Ī±|=m
x ā Ī©, Ī¾ ā Rn .
1.2. The total and the principal symbols
3
The symbol belongs to C ā (Ī©)[Ī¾], i.e., it is a polynomial in Ī¾1 , . . . , Ī¾n with coeļ¬cients in C ā (Ī©), and the principal symbol is a homogeneous polynomial of degree m in Ī¾1 , . . . , Ī¾n with coeļ¬cients in C ā (Ī©). Examples. 1) The total symbol and the principal symbol of the Laplace operator coincide and are equal to āĪ¾ 2 = ā(Ī¾12 + Ā· Ā· Ā· + Ī¾n2 ). 2) The total symbol of the heat operator and its principal symbol is a2 (t, x, Ļ, Ī¾) = Ī¾ 2 .
ā āt
ā Ī is a(t, x, Ļ, Ī¾) = iĻ + Ī¾ 2
3) The total symbol and the principal symbol of the wave operator coincide and are equal to a(t, x, Ļ, Ī¾) = a2 (t, x, Ļ, Ī¾) = āĻ 2 + Ī¾ 2 . 4) The total symbol of the Sturm-Liouville operator is a(x, Ī¾) = āp(x)Ī¾ 2 + ip (x)Ī¾ + q(x); its principal symbol is a2 (x, Ī¾) = āp(x)Ī¾ 2 . The symbol of an operator A is recovered from the operator by the formula a(x, Ī¾) = eāixĀ·Ī¾ A(eixĀ·Ī¾ ),
(1.5)
where A is applied with respect to x. Here we used the notation x Ā· Ī¾ = x1 Ī¾1 + Ā· Ā· Ā· + xn Ī¾n . Formula (1.5) is obtained from the relation D Ī± eixĀ·Ī¾ = Ī¾ Ī± eixĀ·Ī¾ , which is easily veriļ¬ed by induction with respect to |Ī±|. Formula (1.5) also implies that the coeļ¬cients aĪ± (x) of A are uniquely deļ¬ned by this operator, because the coeļ¬cients of the polynomial a(x, Ī¾) for each ļ¬xed x are uniquely deļ¬ned by this polynomial viewed as a function of Ī¾. It is convenient to apply A to a more general exponent than eixĀ·Ī¾ , namely to eiĪ»Ļ(x) , where Ļ ā C ā (Ī©) and Ī» is a parameter. Then we get Lemma 1.1. If f ā C ā (Ī©), then eāiĪ»Ļ(x) A(f (x)eiĪ»Ļ(x) ) is a polynomial of Ī» of degree ā¤ m with coeļ¬cients in C ā (Ī©) and (1.6)
eāiĪ»Ļ(x) A(f (x)eiĪ»Ļ(x) ) = Ī»m f (x)am (x, Ļx ) + Ī»mā1 (. . .) + Ā· Ā· Ā· ;
i.e., the highest coeļ¬cient of this polynomial (that of Ī»m ) is equal to āĻ āĻ , . . . , āx ) = grad Ļ is the gradient of Ļ. f (x)am (x, Ļx ), where Ļx = ( āx n 1
Proof. We have Dj [f (x)eiĪ»Ļ(x) ] = Ī»(āj Ļ)(x)f (x)eiĪ»Ļ(x) + (Dj f )eiĪ»Ļ(x) ,
4
1. Linear diļ¬erential operators
so that the lemma holds for the operators of order ā¤ 1. In what follows, to ļ¬nd the derivatives D Ī± eiĪ»Ļ(x) we will have to diļ¬erentiate products of the form f (x)eiĪ»Ļ(x) , where f ā C ā (Ī©). A new factor Ī» will only appear when we diļ¬erentiate the exponent. Clearly, D Ī± eiĪ»Ļ(x) = Ī»|Ī±| ĻĪ±x eiĪ»Ļ(x) + Ī»|Ī±|ā1 (. . .) + Ā· Ā· Ā· , implying the statement of the lemma. Corollary 1.2. Let A, B be two linear diļ¬erential operators in Ī©, let k, l be their orders, ak , bl their principal symbols, C = A ā¦ B their composition, ck+l the principal symbol of C. Then (1.7)
ck+l (x, Ī¾) = ak (x, Ī¾)bl (x, Ī¾).
Remark. Clearly, C is a diļ¬erential operator of order ā¤ k + l. Proof of Corollary 1.2. We have C(eiĪ»Ļ(x) ) = A(Ī»l bl (x, Ļx )eiĪ»Ļ(x) + Ī»lā1 (. . .) + Ā· Ā· Ā· ) = Ī»k+l ak (x, Ļx )bl (x, Ļx )eiĪ»Ļ(x) + Ī»k+lā1 (. . .) + Ā· Ā· Ā· . By Lemma 1.1, we obtain (1.8)
ck+l (x, Ļx ) = ak (x, Ļx )bl (x, Ļx ),
for any Ļ ā C ā (Ī©).
But then, by choosing Ļ(x) = x Ā· Ī¾, we get Ļx = Ī¾ and (1.8) becomes ck+l (x, Ī¾) = ak (x, Ī¾)bl (x, Ī¾), as required.
1.3. Change of variables Let Īŗ : Ī© āā Ī©1 be a diļ¬eomorphism between two open subsets Ī©, Ī©1 in Rn ; i.e., Īŗ is a C ā map having an inverse C ā map Īŗā1 : Ī©1 āā Ī©. We will denote the coordinates in Ī© by x and coordinates in Ī©1 by y. The map Īŗ can be deļ¬ned by a set of functions y1 (x1 , . . . , xn ), . . . , yn (x1 , . . . , xn ), yj ā C ā (Ī©), whose values at x are equal to the respective coordinates of the point Īŗ(x). If f ā C ā (Ī©1 ), set Īŗā f = f ā¦ Īŗ; i.e., (Īŗā f )(x) = f (Īŗ(x)) or (Īŗā f )(x) = f (y1 (x), . . . , yn (x)). The function Īŗā f is obtained from f by a change of variables or by passage to coordinates x1 , . . . , xn from coordinates y1 , . . . , yn . Recall that the diļ¬eomorphism Īŗ has the inverse Īŗā1 which is also a diļ¬eomorphism. Clearly, Īŗā is a linear isomorphism Īŗā : C ā (Ī©1 ) āā C ā (Ī©) and (Īŗā )ā1 = (Īŗā1 )ā .
1.4. The canonical form
5
Given an operator A : C ā (Ī©) āā C ā (Ī©), deļ¬ne the operator A1 : C ā (Ī©1 ) āā C ā (Ī©1 ) with the help of the commutative diagram C ā (Ī©) ā ā āĪŗ
A
āā C ā (Ī©) ā ā āĪŗ A
1 C ā (Ī©1 ) āā C ā (Ī©1 )
i.e., by setting A1 = (Īŗā )ā1 AĪŗā = (Īŗā1 )ā AĪŗā . In other words, (1.9)
A1 v(y) = {A[v(y(x))]}|x=x(y),
where y(x) = Īŗ(x), x(y) = Īŗā1 (y). Therefore, the operator A1 is, actually, just the expression of A in coordinates y. If A is a linear diļ¬erential operator, then so is A1 , as easily follows from the chain rule and (1.9). Let us investigate a relation between the principal symbols of A and A1 . To this end we will use the notions of tangent and cotangent vectors in Ī©, as well as tangent and cotangent bundles over Ī© (see the appendix in this chapter). In particular, for x ā Ī© we identify RN with Txā Ī© as follows: using the coordinates x1 , . . . , xn we deļ¬ne the basis dx1 , . . . , dxn in Txā Ī© and we identify each vector Ī¾ = (Ī¾1 , . . . , Ī¾n ) ā Rn with a covector Ī¾1 dx1 + Ā· Ā· Ā· + Ī¾n dxn ā Txā Ī©, which we also denote by Ī¾. Similarly, for y := Īŗ(x) ā Ī©1 we use coordinates y1 , . . . , yn to identify Rn with Tyā Ī©1 . Theorem 1.3. The value of the principal symbol am of an operator A at the cotangent vector (x, Ī¾) is the same as that of the principal symbol am of A1 at the corresponding vector (y, Ī·); i.e., (1.10)
am (x, Ī¾) = am (Īŗ(x), (t Īŗā )ā1 Ī¾).
In other words, the principal symbol is a well-deļ¬ned function on T ā Ī© (it does not depend on the choice of C ā -coordinates in Ī©). Proof. A cotangent vector at x can be expressed as the gradient Ļx (x) of a function Ļ ā C ā (Ī©) at x. It is clear from Lemma 1.1 that am (x, Ļx (x)) does not depend on the choice of coordinates. But on the other hand, it is clear that this value does not depend on the choice of a function Ļ with a prescribed diļ¬erential at x. Therefore, the principal symbol is well-deļ¬ned on T ā Ī©.
1.4. The canonical form of second-order operators with constant coeļ¬cients By a change of variables we can try to transform an operator to a simpler one. Consider a second-order operator with constant coeļ¬cients of the principal
6
1. Linear diļ¬erential operators
part: A=ā
n i,k=1
ā ā2 + bl + c, āxi āxk āxl n
aik
l=1
where the aik are real constants such that aik = aki (the latter can always 2 2 be assumed since āxāi āxk = āxāk āxi ). Ignoring the terms of lower order, we will transform the highest-order part of A A0 = ā
n
aik
i,k=1
ā2 āxi āxk
to a simpler form with the help of a linear change of variables: yk =
(1.11)
n
ckl xl ,
l=1
where the ckl are real constants. Consider the quadratic form Q(Ī¾) =
n
aik Ī¾i Ī¾k,
i,k=1
which only diļ¬ers from the principal symbol of A by a sign. With the help of a linear change of variables Ī· = F Ī¾, where F is a nonsingular constant matrix, we can transform Q(Ī¾) to the following form: (1.12)
Q(Ī¾) = (Ā±Ī·12 Ā± Ī·22 Ā± Ā· Ā· Ā· Ā± Ī·r2 )|Ī·=F Ī¾
for some r ā¤ n. Denote by C the matrix (ckl )nk,l=1 in (1.11). By Theorem 1.3, the operator A in coordinates y has the form of a second-order operator A1 whose principal symbol is a quadratic form Q1 (Ī·) such that Q(Ī¾) = Q1 (Ī·)|Ī·=(t C)ā1 Ī¾ , where t C is the transpose of C. From this and (1.12) it follows that we should choose C so that (t C)ā1 = F or C = (t F )ā1 . Then the principal part of A under the change of variables y = Cx of the form (1.11) will be reduced to the form (1.13)
Ā±
ā2 ā2 ā2 Ā± Ā± Ā· Ā· Ā· Ā± , āyr2 āy12 āy22
called the canonical form. Remark. An operator with variable coeļ¬cients may be reduced to the form (1.13) at any ļ¬xed point by a linear change of variables.
1.5. Characteristics. Ellipticity and hyperbolicity
7
1.5. Characteristics. Ellipticity and hyperbolicity Let A be an mth-order diļ¬erential operator, am (x, Ī¾) its principal symbol. A nonzero cotangent vector (x, Ī¾) is called a characteristic vector if am (x, Ī¾) = 0. A surface (of codimension 1) in Ī© is called characteristic at x0 if its normal vector at this point is a characteristic vector. The surface is called a characteristic if it is characteristic at every point. If a surface S is deļ¬ned by the equation Ļ(x) = 0, where Ļ ā C 1 (Ī©), Ļx |S = 0, then S is characteristic at x0 if am (x0 , Ļx (x0 )) = 0. The surface S is a characteristic if am (x, Ļx (x)) identically vanishes on S. All level surfaces Ļ = const are characteristics if and only if am (x, Ļx (x)) ā” 0. Theorem 1.3 implies that the notion of a characteristic does not depend on the choice of coordinates in Ī©. Examples. 1) The Laplace operator Ī has no real characteristic vectors. ā ā Ī the vector (Ļ, Ī¾) = (1, 0, . . . , 0) ā Rn+1 is 2) For the heat operator āt a characteristic one. The surfaces t = const are characteristics. The surface t = |x|2 (paraboloid) is characteristic at a single point only (the origin). 2
ā 3) Consider the wave operator āt 2 ā Ī. Its characteristic vectors at each 2 point (t, x) constitute a cone Ļ = |Ī¾|2 . Any cone (t ā t0 )2 = |x ā x0 |2 is a characteristic. In particular, for n = 1 (i.e., for x ā R1 ) the lines x + t = const and x ā t = const are characteristics.
Deļ¬nition. a) An operator A is called elliptic if am (x, Ī¾) = 0 for all x ā Ī© and all Ī¾ ā Rn \ {0}, i.e., if A has no nonzero characteristic vectors. b) An operator A in the space of functions of t, x, where t ā R1 , x ā Rn , is hyperbolic with respect to t if the equation am (t, x, Ļ, Ī¾) = 0 considered as an equation in Ļ for any ļ¬xed t, x, Ī¾ has exactly m distinct real roots if Ī¾ = 0. In this case one might say that the characteristics of A are real and distinct. Examples. a) The Laplace operator Ī is elliptic. b) The heat operator is neither elliptic nor hyperbolic with respect to t. c) The wave operator is hyperbolic with respect to t since the equation Ļ 2 = |Ī¾|2 has two diļ¬erent real roots Ļ = Ā±|Ī¾| if Ī¾ = 0. d) The Sturm-Liouville operator Lu ā” (a, b) if p(x) = 0 for x ā (a, b).
d d dx (p(x) dx u) + q(x)u
is elliptic on
8
1. Linear diļ¬erential operators
By Theorem 1.3, the ellipticity of an operator does not depend on the choice of coordinates. The hyperbolicity with respect to t does not depend on the choice of coordinates in Rnx . Let us see what it means for the surface x1 = const to be characteristic. The coordinates of the normal vector are (1, 0, . . . , 0). Substituting them into the principal symbol, we get aĪ± (x)(1, 0, . . . , 0)Ī± = a(m,0,...,0) (x); am (x; 1, 0, . . . , 0) = |Ī±|=m
i.e., the statement āthe surface x1 = const is characteristic at xā means that ām a(m,0,...,0) (x), the coeļ¬cient of ( 1i )m āx m , vanishes at x. 1
All surfaces x1 = const are characteristics if and only if a(m,0,...,0) ā” 0. This remark will be used below to reduce second-order operators with two independent variables to a canonical form. It is also possible to ļ¬nd characteristics by solving the Hamilton-Jacobi equation am (x, Ļx (x)) = 0. If Ļ is a solution of this equation, then all surfaces Ļ = const are characteristic. Integration of the Hamilton-Jacobi equation is performed with the help of the Hamiltonian system of ordinary diļ¬erential equations with the Hamiltonian am (x, Ī¾) (see Chapter 12 or any textbook on classical mechanics, e.g., Arnold [3]).
1.6. Characteristics and the canonical form of second-order operators and second-order equations for n = 2 For n = 2 characteristics are curves and are rather easy to ļ¬nd. For example, consider a second-order operator (1.14)
A=a
ā2 ā2 ā2 + c + 2b + Ā·Ā·Ā· , āx2 āxāy āy 2
where a, b, c are smooth functions of x, y deļ¬ned in an open set Ī© ā R2 and the dots stand for terms containing at most ļ¬rst-order derivatives. Let (x(t), y(t)) be a curve in Ī©, (dx, dy) its tangent vector, (ādy, dx) the normal vector. The curve is a characteristic if and only if a(x, y)dy 2 ā 2b(x, y)dxdy + c(x, y)dx2 = 0 on it. If a(x, y) = 0, then in a neighborhood of the point (x, y) we may assume that dx = 0; hence, x can be taken as a parameter along the characteristic which, therefore, is of the form y = y(x). Then the equation for the characteristics is of the form ay 2 ā 2by + c = 0.
1.6. Characteristics and the canonical form (n = 2)
9
If b2 ā ac > 0, then the operator (1.14) is hyperbolic and has two families of real characteristics found from ordinary diļ¬erential equations: ā b + b2 ā ac (1.15) y = , a ā b ā b2 ā ac . (1.16) y = a Observe that in this case two nontangent characteristics pass through each point (x, y) ā Ī©. We express these families of characteristics in the form Ļ1 (x, y) = C1 and Ļ2 (x, y) = C2 , where Ļ1 , Ļ2 ā C ā (Ī©). This means that Ļ1 , Ļ2 are ļ¬rst integrals of equations (1.15) and (1.16), respectively. Suppose that grad Ļ1 = 0 and grad Ļ2 = 0 in Ī©. Then grad Ļ1 and grad Ļ2 are linearly independent, since characteristics of diļ¬erent families are not tangent. Let us introduce new coordinates: Ī¾ = Ļ1 (x, y), Ī· = Ļ2 (x, y). In these coordinates, characteristics are lines Ī¾ = const and Ī· = const. But ā2 ā2 then the coeļ¬cients of āĪ¾ 2 and āĪ· 2 vanish identically and A takes the form (1.17)
A = p(Ī¾, Ī·)
ā2 +Ā·Ā·Ā· , āĪ¾āĪ·
called the canonical form. Here p(Ī¾, Ī·) = 0 (recall that we have assumed a = 0). If c(x, y) = 0, then the reduction to the canonical form (1.17) is similar. It is not necessary to consider these two cases separately, since we have only used the existence of integrals Ļ1 (x, y), Ļ2 (x, y) with the described properties. These integrals can also be found when coeļ¬cients a and c vanish (at one point or on an open set). This case can be reduced to one of the above cases, e.g., by a rotation of coordinate axes (if a2 + b2 + c2 = 0). One often considers diļ¬erential equations of the form (1.18)
Au = f,
where f is a known function, A a linear diļ¬erential operator, and u an unknown function. If A is a hyperbolic second-order operator with two independent variables, i.e., an operator of the form (1.14) with b2 ā ac > 0 (in this case the equation (1.18) is also called hyperbolic), then, introducing the above coordinates Ī¾, Ī· and dividing by p(Ī¾, Ī·), we reduce (1.18) to the canonical form ā2u + Ā· Ā· Ā· = 0, āĪ¾āĪ· where the dots stand for terms without second-order derivatives of u. Now let b2 ā ac ā” 0 (then the operator (1.14) and the equation (1.18) with this operator are called parabolic). Let us assume that a = 0. Then,
10
1. Linear diļ¬erential operators
for characteristics we get the diļ¬erential equation (1.19)
y =
b . a
Let us ļ¬nd characteristics and write them in the form Ļ(x, y) = const, where Ļ is the ļ¬rst integral of (1.19) and grad Ļ = 0. Choose a function Ļ ā C ā (Ī©) such that grad Ļ and grad Ļ are linearly independent and introduce new coordinates Ī¾ = Ļ(x, y), Ī· = Ļ(x, y). In the new coordinates the operator ā2 A does not have the term āĪ¾ 2 , since the lines Ļ = const are characteristics. 2
ā also vanishes, since the principal But then the coeļ¬cient of the term āĪ¾āĪ· symbol should be a quadratic form of rank 1. Thus, the canonical form of a parabolic operator is ā2 A = p(Ī¾, Ī·) 2 + Ā· Ā· Ā· . āĪ·
For a parabolic equation (1.18) the canonical form is ā2 + Ā· Ā· Ā· = 0. āĪ· 2 Observe that if b2 ā ac = 0 and a2 + b2 + c2 = 0, then a and c cannot vanish simultaneously, since then we would also have b = 0. Therefore, we always have either a = 0 or c = 0 and the above procedure is always applicable. Finally, consider the case b2 ā ac < 0. The operator (1.14) and the equation (1.18) are in this case called elliptic. Suppose for simplicity that functions a, b, c are real analytic. Then the existence theorem for complex analytic solutions of the complex equation ā b + b2 ā ac y = a implies the existence of a local ļ¬rst integral Ļ(x, y) + iĻ(x, y) = C, where Ļ, Ļ are real-valued analytic functions and grad Ļ and grad Ļ are linearly independent. It is easy to verify that by introducing new coordinates Ī¾ = Ļ(x, y), Ī· = Ļ(x, y), we can reduce A to the canonical form 2 ā2 ā + + Ā·Ā·Ā· , A = p(Ī¾, Ī·) āĪ¾ 2 āĪ· 2 where p(Ī¾, Ī·) = 0. Dividing (1.18) by p, we reduce it to the form ā 2u ā 2u + 2 + Ā· Ā· Ā· = 0. āĪ¾ 2 āĪ·
1.7. The general solution (n = 2)
11
1.7. The general solution of a homogeneous hyperbolic equation with constant coeļ¬cients for n = 2 As follows from Section 1.6, the hyperbolic equation (1.20)
a
ā 2u ā 2u ā 2u + c + 2b = 0, āx2 āxāy āy 2
where a, b, c ā R and b2 ā ac > 0, can be reduced by the change of variables Ī¾ = y ā Ī»1 x,
Ī· = y ā Ī»2 x,
where Ī»1 , Ī»2 are roots of the quadratic equation aĪ»2 ā 2bĪ» + c = 0, to the form ā 2u = 0. āĪ¾āĪ· Assuming u ā C 2 (Ī©), where Ī© is a convex open set in R2 , we get implying āu āĪ· = F (Ī·) and
ā āu āĪ¾ ( āĪ· )
=0
u = f (Ī¾) + g(Ī·), where f, g are arbitrary functions of class C 2 . In variables x, y we have (1.21)
u(x, y) = f (y ā Ī»1 x) + g(y ā Ī»2 x).
It is useful to consider functions u(x, y) of the form (1.21), where f, g are not necessarily of the class C 2 but from a wider class, e.g., f, g ā L1loc (R); i.e., f, g are locally integrable. Such functions u are called generalized solutions of equation (1.20). For example let f (Ī¾) have a jump at Ī¾0 ; i.e., it has diļ¬erent limits as Ī¾ ā Ī¾0 + and as Ī¾ ā Ī¾0 ā. Then u(x, y) has a discontinuity along the line y ā Ī»1 x = Ī¾0 . Note that the lines y ā Ī»1 x=const, y ā Ī»2 x = const are characteristics. Therefore, in this case discontinuities of solutions are propagating along characteristics. A similar phenomenon occurs in the case of general hyperbolic equations. 2
2
Example. Characteristics of the wave equation āāt2u = a2 āāxu2 are x ā at = const, x+at = const. The general solution of this equation can be expressed in the form u(t, x) = f (x ā at) + g(x + at). Observe that f (x ā at) is the wave running to the right with the speed a and g(x + at) is the wave running to the left with the same speed a. The general solution is the sum (or superposition) of two such waves.
12
1. Linear diļ¬erential operators
1.8. Appendix. Tangent and cotangent vectors Let Ī© be an open set in Rn . The tangent vector to Ī© at a point x ā Ī© is a vector v which is tangent to a parametrized curve x(t) in Ī© passing through x at t = 0. Here we assume that the function t ā x(t) is a C ā function deļ¬ned for t ā (āĪµ, +Īµ) with some Īµ > 0 and takes values in Ī©. In coordinates x1 , . . . , xn we have x(t) = (x1 (t), . . . , xn (t)) and the tangent vector v is given as v = x(0) Ė =
dx dt t=0
= (xĖ 1 (0), . . . , xĖ n (0)) dxn dx1 ,..., = dt t=0 dt t=0 = (v1 , . . . , vn ).
Interpreting t as the time variable, we can say that the tangent vector v = x(0) Ė is the velocity vector at the moment t = 0 for the motion given by the curve x(t). The real numbers v1 , . . . , vn are called the coordinates of the tangent vector x(0). Ė They depend on the choice of the coordinate system (x1 , . . . , xn ). In fact, they are deļ¬ned even for any curvilinear coordinate system. Two tangent vectors at x can be added (by adding the corresponding coordinates of these vectors), and any tangent vector can be multiplied by any real number (by multiplying each of its coordinates by this number). These operations do not depend on the choice of coordinates in Ī©. So all tangent vectors at x ā Ī© form a well-deļ¬ned vector space (independent of coordinates in Ī©), which will be denoted Tx Ī© and called the tangent space to Ī© at x. A cotangent vector Ī¾ at x ā Ī© is a real-valued linear function on the tangent space Tx Ī©. All cotangent vectors at x form a linear space which is denoted Txā Ī© and called the cotangent space to Ī© at x. Denote by T Ī© the union of all tangent spaces over all points x ā Ī©; i.e., TĪ© =
Tx Ī© ā¼ = Ī© Ć Rn .
x
It is called the tangent bundle of Ī©. Similarly, we will introduce the cotangent bundle T āĪ© =
x
Txā Ī© ā¼ = Ī© Ć Rn .
1.8. Appendix. Tangent and cotangent vectors
13
For every tangent vector v at x ā Ī© and any function f ā C ā (Ī©) we can deļ¬ne the derivative of f in the direction v: it is vf =
āf (x) df (x(t)) āf (x) = xĖ j (0) = vj , dt āxj āxj n
n
j=1
j=1
where x(t) is a curve in Ī© such that x(0) = x and x(0) Ė = v. Sometimes, vf is also called the directional derivative. Taking f to be a coordinate function xj , we obtain vxj = vj , so v is uniquely deļ¬ned by the corresponding directional derivative, and we can identify v with its directional derivative which is a linear function on C ā (Ī©). In particular, take vectors at x ā Ī© which are tangent to the coordinate axes. Then the corresponding directional derivatives will be simply ā ā ā āx1 , . . . , āxn , and they form a basis of Tx Ī©. (For example, āx1 is tangent to the curve x(t) = (x1 + t, x2 , . . . , xn ) and has coordinates (1, 0, . . . , 0). For the jth partial derivative all coordinates vanish except the jth one which equals 1.) The dual basis in Txā Ī© consists of linear functions dx1 , . . . , dxn on Tx Ī© such that dxi ( āxā j ) = Ī“ij , where Ī“ij = 1 for i = j and Ī“ij = 0 for i = j. For any tangent vector x(0) Ė ā Tx(0) Ī© which is the velocity vector of a curve x(t) at t = 0, its coordinates in the basis āxā 1 , . . . , āxān are equal dxn 1 to dx dt |t=0 , . . . , dt |t=0 . An example of a cotangent vector is given by the diļ¬erential df of a function f ā C ā (Ī©): it is a function on Tx(0) Ī© whose Cā
value at the tangent vector x(0) Ė is equal to df (x(t)) dt |t=0 and whose coordinates āf āf in the basis dx1 , . . . , dxn are āx , . . . , , so āxn 1 n āf df = dxj . āxj j=1
In Euclidean coordinates, the diļ¬erential of a smooth function f can be identiļ¬ed with its gradient āf , which is the tangent vector with the same coordinates. Given a diļ¬eomorphism Īŗ : Ī© āā Ī©1 , we have a natural map Īŗā : Tx Ī© āā TĪŗ(x) Ī©1 (the tangent map, or the diļ¬erential of Īŗ). Namely, for a tangent vector x(0), Ė which is the velocity vector of a curve x(t) at t = 0, the Ė is the velocity vector of the curve (Īŗ ā¦ x)(t) = Īŗ(x(t)) at t = 0. vector Īŗā x(0) ā Ī© āā The corresponding dual map of spaces of linear functions t Īŗā : TĪŗ(x) 1 ā t ā Ė = Ī¾(Īŗā x(0)), Ė where Ī¾ ā TĪŗ(x) Ī©. Tx Ī© is deļ¬ned by ( Īŗā Ī¾)(x(0)) Choose bases { āxā j }nj=1 , { āyā j }nj=1 , {dyj }nj=1 , {dxj }nj=1 in spaces Tx Ī©, ā Ī© , T ā Ī©, respectively. In these bases the matrix of Īŗ is TĪŗ(x) Ī©1 , TĪŗ(x) 1 ā x (Īŗā )kl =
āyk āxl .
It is called the Jacobi matrix. The matrix of t Īŗā is the
14
1. Linear diļ¬erential operators
transpose of Īŗā . Since Īŗ is a diļ¬eomorphism, the maps Īŗā and t Īŗā are isomorphisms at each point x ā Ī©. It is convenient to deļ¬ne a cotangent vector at x by a pair (x, Ī¾), where Ī¾ = (Ī¾1 , . . . , Ī¾n ), Ī¾ ā Rn , is the set of coordinates of this vector in the basis {dxj }nj=1 in Txā Ī©. Under the isomorphism (t Īŗā )ā1 : T ā Ī© āā T ā Ī©1 the covector (y, Ī·) corresponds to the covector (x, Ī¾), where y = Īŗ(x), Ī· = (t Īŗā )ā1 Ī¾.
1.9. Problems 1.1. Reduce the equations to the canonical form: a) uxx + 2uxy ā 2uxz + 2uyy + 6uzz = 0; b) uxy ā uxz + ux + uy ā uz = 0. 1.2. Reduce the equations to the canonical form: a) x2 uxx + 2xyuxy ā 3y 2 uyy ā 2xux + 4yuy + 16x4 u = 0; b) y 2 uxx + 2xyuxy + 2x2 uyy + yuy = 0; c) uxx ā 2uxy + uyy + ux + uy = 0. 1.3. Find the general solution of the equations: a) x2 uxx ā y 2 uyy ā 2yuy = 0; b) x2 uxx ā 2xyuxy + y 2 uyy + xux + yuy = 0.
10.1090/gsm/205/02
Chapter 2
One-dimensional wave equation
2.1. Vibrating string equation Let us derive an equation describing small vibrations of a string. Note right away that our derivation will not be mathematical but rather physical or mechanical. However, its understanding is essential to grasp the physical meaning of, ļ¬rst, the wave equation itself and, second, but not less importantly, the initial and boundary conditions. A knowledge of the derivation and physical meaning helps also to ļ¬nd various mathematical tools in the study of this equation (the energy integral, standing waves, etc.). Generally, derivation of equations corresponding to diļ¬erent physical and mechanical problems is important for understanding mathematical physics and is essentially a part of it. So, let us derive the equation of small vibrations of a string. We consider the vibrations of a taut string such that each point moves in a direction perpendicular to the direction of the string in its equilibriuim. We assume here that all the forces appearing in the string are negligible compared to the tension directed along the string (we assume the string to be perfectly ļ¬exible, i.e., nonresistant to bending). First of all, choose the variables describing the form of the string. Suppose that in equilibrium the taut string is stretched along the x-axis. To start, we will consider the inner points of the string disregarding its ends. Suppose that the vibrations keep the string in the (x, y)-plane so that each point of the string is only shifted parallel to the y-axis and this shift at time t equals u(t, x); see Figure 2.1. 15
16
2. One-dimensional wave equation y
u(t, x) x
x Figure 2.1. Vibrating string.
Therefore, at a ļ¬xed t the graph of u(t, x), as a function of x, is the shape of the string at the moment t, and for a ļ¬xed x the function u(t, x) describes the motion of one point of the string. Let us compute the length of the part of the string corresponding to the interval (a, b) on the x-axis. It is equal to
b 1 + u2x dx. a
Our main assumption is the negligibility of the additional string lengthening, compared with the equilibrium length. More precisely, we assume that u2x 1 and consider u2x negligible compared to 1. Note that if Ī± = Ī±(t, x) is the angle between the tangent to the string and the x-axis, then tan Ī± = ux , cos Ī± = ā 1 2 , sin Ī± = ā ux 2 ; under our assumptions we should consider 1+ux
1+ux
cos Ī± ā 1 and sin Ī± ā ux . If T = T (t, x) is the stringās tension, then its horizontal component is T cos Ī± ā T and the vertical one is T sin Ī± ā T ux . By Hookeās law in elasticity theory (which should be considered an assumption in our model) the tension of the string at any point and at any given moment of time is proportional to the relative lengthening of a small piece of string at this point, i.e., the lengthening (the increase of the length compared with the equilibrium) divided by the equilibrium length of this piece. Since we assume the additional lengthening to be negligible, we should have T (t, x) = T = const. (Alternatively, we can argue as follows. Since all points of the string move in the vertical direction, along the y-axis, the sum of the horizontal components of all forces acting on the segment of the string over [a, b] should vanish. This means that T (t, a) = T (t, b) which due to the arbitrariness of a and b yields T (t, x) = T (t); that is, T is independent of x. Then, the negligibility of the lengthening implies that in fact T (t) = T = const.) Let us write the equation of motion of the part of the string above [a, b]; see Figure 2.2. Let Ļ(x) be the linear density of the string at x (the ratio of the mass of the inļ¬nitesimal segment to the length of this segment at x).
2.1. Vibrating string equation
17
y
T
T a
b
x
Figure 2.2. Tension forces in the string.
The vertical component of the force is
b
(T ux )|x=b ā (T ux )|x=a =
(2.1)
T uxx (t, x)dx, a
assuming the absence of external forces. In the presence of vertical external forces distributed with density g(t, x) (per unit mass of the string), we should add to (2.1) the term
b Ļ(x)g(t, x)dx. (2.2) a
The vertical component of the momentum of this segment of the string is equal to
b (2.3) Ļ(x)ut (t, x)dx. a
Let us use the well-known corollary of Newtonās Second and Third Laws which states that the rate of the momentumās change in time is equal to the sum of all external forces. (See the appendix in this chapter.) Then (2.1)ā(2.3) give
b [Ļ(x)utt (t, x) ā T uxx (t, x) ā Ļ(x)g(t, x)]dx = 0, a
or, since a and b are arbitrary, (2.4)
Ļ(x)utt ā T uxx ā Ļ(x)g(t, x) = 0.
In particular, for Ļ(x) = Ļ = const and g(t, x) ā” 0 we get (2.5) utt ā a2 uxx = 0, where a = T /Ļ. We would have arrived at the same equation (2.5) if we had applied dāAlembertās principle equating the sum of all the external and inertial forces to zero. One may arrive at (2.4) via another route by using Lagrangeās equations. (See the appendix in this chapter for an elementary introduction to the
18
2. One-dimensional wave equation
calculus of variations and Lagrangeās equations.) Suppose the external forces are absent. Clearly, the kinetic energy of the interval (a, b) of the string is
1 b K= Ļ(x)u2t (t, x)dx. 2 a To calculate the potential energy of the string whose form is that of the graph of v(x), x ā (a, b), we should calculate the work necessary to move the string from the equilibrium position to v(x). Suppose this move is deļ¬ned by the ācurveā v(x, Ī±), Ī± ā [0, 1], so that v(x, 0) = 0, v(x, 1) = v(x). The force
x+Īx T vxx (x, Ī±)dx T vx (x + Īx, Ī±) ā T vx (x, Ī±) = x
acts on the part of the string corresponding to the interval (x, x + Īx) on the x-axis. As Ī± varies from Ī± to Ī± + ĪĪ±, the displacement of the point with the abscissa x equals
Ī±+ĪĪ± vĪ± (x, Ī±)dĪ±. v(x, Ī± + ĪĪ±) ā v(x, Ī±) = Ī±
Therefore, to move the part of the string (x, x + Īx) from Ī± to Ī± + ĪĪ±, the work āT vxx (x, Ī±)vĪ± (x, Ī±)Īx Ā· ĪĪ± + o(Īx Ā· ĪĪ±) should be performed. Integrating with respect to x and Ī± we see that the total work performed over the segment (a, b) is
1 b A=ā T vxx vĪ± dxdĪ±. 0
a
Integrating by parts, we get
b
b
b d 1 b b T vxx vĪ± dx = āT vx vĪ± |a + T vx vxĪ± dx = āT vx vĪ± |a + T vx2 dx, ā dĪ± a 2 a a and integrating with respect to Ī± from 0 to 1, we get
1 1 b 2 x=b T vx (x, Ī±)vĪ± (x, Ī±)|x=a dĪ± + T vx (x)dx. (2.6) A=ā 2 a 0 Suppose that the ends of the string are ļ¬xed at the points 0 and l; i.e., their displacements are equal to zero during the whole process of motion. Then we may assume that v(0, Ī±) = v(l, Ī±) = 0 for Ī± ā [0, 1] and write
1 l 2 T vx (x)dx A= 2 0
2.1. Vibrating string equation
19
instead of (2.6) for a = 0, b = l. This clearly implies that the potential energy of the string with ļ¬xed ends 0, l at time t is equal to
1 l U= T u2x (t, x)dx. 2 0 Now we can write the Lagrangian of the string:
1 l 1 l Ļ(x)u2t (t, x)dx ā T u2x (t, x)dx, L=K āU = 2 0 2 0 which is a functional in u and ut (here u(t, x) plays the role of the set of coordinates at time t and ut (t, x) the role of the set of velocities). Let us t write the action t01 Ldt and equate the variation of the action with respect to u to 0. The usual integration by parts under the assumption that Ī“u|t=t0 = Ī“u|t=t1 = 0 gives
t1 l
t1 Ldt = Ļ(x)ut (t, x)Ī“ut (t, x)dxdt Ī“ t0
t0
ā =ā
0 t1 l
T ux (t, x)Ī“ux (t, x)dxdt t0
t1 t0
0
l
[Ļ(x)utt (t, x) ā T uxx (t, x)]Ī“u(t, x)dxdt,
0
which, since Ī“u is arbitrary, implies (2.7)
Ļ(x)utt (t, x) ā T uxx (t, x) = 0.
This means that we obtained equation (2.4) (with g(t, x) ā” 0). Note an important circumstance here: the total energy of the string is equal to
1 l 1 l 2 H =K +U = Ļ(x)ut (t, x)dx + T u2x (t, x)dx. 2 0 2 0 Let us prove the energy conservation law: the energy of an oscillating string with ļ¬xed ends is conserved. We will verify this law by using (2.7). We have
l
l dH = Ļ(x)ut (t, x)utt (t, x)dx + T ux (t, x)utx (t, x)dx. dt 0 0 Integrating by parts in the second integral, we get
l dH = ut (t, x)[Ļ(x)utt(t, x) ā T uxx (t, x)]dx + T ux (t, x)ut (t, x)|x=l (2.8) x=0 . dt 0 The last summand in (2.8) vanishes due to the boundary conditions u|x=0 = d (u|x=0 ) = 0 and similarly ut |x=l = 0. The u|x=l = 0, since then ut |x=0 = dt ļ¬rst summand vanishes due to (2.7). Thus, dH dt = 0, yielding the energy conservation law: H(t) = const.
20
2. One-dimensional wave equation
The conservation of energy implies the uniqueness of the solution for the string equation (2.4) provided Ļ(x) > 0 everywhere and the law of motion for the stringās ends is given: (2.9)
u|x=0 = Ī±(t),
u|x=l = Ī²(t),
and the initial values (the position and the velocity of the string) are ļ¬xed: (2.10)
u|t=0 = Ļ(x),
ut |t=0 = Ļ(x)
for x ā [0, 1].
Indeed, if u1 , u2 are two solutions of the equation (2.4) satisfying (2.9) and (2.10), then their diļ¬erence v = u1 ā u2 satisļ¬es the homogeneous equation (2.7) and the homogeneous boundary and initial conditions v|x=0 = v|x=l = 0, (2.11)
v|t=0 = vt |t=0 = 0
for x ā [0, l].
But then the energy conservation law for v implies
l [Ļ(x)vt2 (t, x) + T vx2 (t, x)]dx ā” 0, 0
yielding vt ā” 0 and vx ā” 0; i.e., v = const. By (2.11) we now have v ā” 0; i.e., u1 ā” u2 , as required. Finally, note that the derivation of the string equation could be, of course, carried out without the simplifying assumption u2x 1. As a result, we would have obtained a nonlinear equation that hardly admits investigation by simple methods. The equation (2.4) is in fact the linearization (principal linear part) of this nonlinear equation. Properties of the solution of (2.4) give some idea of the behavior of solutions of the nonlinear equation. Observe, however, that for large deformations this nonlinear equation is also inadequate to describe the physical problem, since the resistance to the bending and other eļ¬ects which were unaccounted for will arise.
2.2. Unbounded string. The Cauchy problem. DāAlembertās formula From the physical viewpoint, an unbounded string is an idealization, meaning that we consider an inner part of the string assuming the ends to be suļ¬ciently far away, so that during the observation time they do not aļ¬ect the given part of the string. As we will see, the consideration of an unbounded string helps in the investigation of a half-bounded string (which is a similar idealization) and of a bounded string. Going on to the mathematical discussion, consider the one-dimensional wave equation (2.5) for x ā R, t ā„ 0. A natural problem here is the Cauchy
2.2. Unbounded string. DāAlembertās formula
21
problem, i.e., an initial value problem for (2.5) with the initial conditions (2.12)
u|t=0 = Ļ(x),
ut |t=0 = Ļ(x).
From the physical viewpoint (2.12) means that the initial position and the initial velocity of the string are given. We might expect that, like ļ¬nitedimensional mechanical problems, this Cauchy problem is well-posed; i.e., there exists a unique solution continuously depending on the initial data Ļ and Ļ. As we will see shortly, this is really so. We use the general solution of the equation (2.5): u(t, x) = f (x ā at) + g(x + at) found in Section 1.7. From (2.12) we get the system of two equations to determine two arbitrary functions f and g: (2.13)
f (x) + g(x) = Ļ(x),
āaf (x) + ag (x) = Ļ(x).
(2.14)
The integration of the second equation yields
1 x (2.15) āf (x) + g(x) = Ļ(Ī¾)dĪ¾ + C. a x0 Now, from (2.13) and (2.15), we get 1 1 f (x) = Ļ(x) ā 2 2a 1 1 g(x) = Ļ(x) + 2 2a Therefore, 1 1 u(t, x) = Ļ(x ā at) ā 2 2a
xāat x0
x
Ļ(Ī¾)dĪ¾ ā
C , 2
Ļ(Ī¾)dĪ¾ +
C . 2
x0
x x0
1 1 Ļ(Ī¾)dĪ¾ + Ļ(x + at) + 2 2a
or (2.16)
Ļ(x ā at) + Ļ(x + at) 1 u(t, x) = + 2 2a
x+at
Ļ(Ī¾)dĪ¾ x0
x+at
Ļ(Ī¾)dĪ¾. xāat
Thus, the solution u(t, x) actually exists, is unique, and continuously depends on the initial data Ļ, Ļ under an appropriate choice of topologies in the sets of initial data Ļ, Ļ and functions u(t, x). For example, it is clear that if u1 is a solution corresponding to the initial conditions Ļ1 , Ļ1 and if for a suļ¬ciently small Ī“ > 0 we have (2.17)
sup |Ļ1 (x) ā Ļ(x)| < Ī“,
sup |Ļ1 (x) ā Ļ(x)| < Ī“,
xāR
xāR
22
2. One-dimensional wave equation
then for arbitrarily small Īµ > 0 we have (2.18)
|u1 (t, x) ā u(t, x)| < Īµ.
sup xāR, tā[0,T ]
(More precisely, for any Īµ > 0 and T > 0 there exists Ī“ > 0 such that (2.17) implies (2.18).) Therefore, the Cauchy problem is well-posed. The formula (2.16) which determines the solution of the Cauchy problem is called dāAlembertās formula. Let us make several remarks concerning its derivation and application. First, note that this formula makes sense for any locally integrable functions Ļ and Ļ and in this case it gives a generalized solution of (2.5). Here we will only consider continuous solutions. Then we can take any continuous function for Ļ and any locally integrable one for Ļ. It is natural to call the functions u(t, x) obtained in this way the generalized solutions of the Cauchy problem. We will consider generalized solutions together with ordinary ones on equal footing and skip the adjective āgeneralizedā. Second, it is clear from dāAlembertās formula that the value of the solution at (t0 , x0 ) only depends on values of Ļ(x) at x = x0 Ā± at0 and values of Ļ(x) at x ā (x0 āat0 , x0 +at0 ). In any case, it suļ¬ces to know Ļ(x) and Ļ(x) on [x0 āat0 , x0 +at0 ]. The endpoints of the closed interval [x0 āat0 , x0 +at0 ] can be determined as intersection points (in (t, x)-space) of the characteristics passing through (t0 , x0 ), with the x-axis; see Figure 2.3. The triangle formed by these characteristics and the x-axis is the set of points in the half-plane t ā„ 0 where the value of the solution is completely t t0 , x 0
x0 ā at0
x0 + at0
x
Figure 2.3. Dependence domain.
c
xā
= at
at =
x+
d
t
c
d
Figure 2.4. Inļ¬uence domain.
x
2.2. Unbounded string. DāAlembertās formula
23
determined by initial values on the segment [x0 āat0 , x0 +at0 ]. This triangle is called the dependence domain for the interval [x0 ā at0 , x0 + at0 ]. An elementary analysis of the derivation of dāAlembertās formula shows that the formula is true for any solution deļ¬ned in the triangle with the characteristics as its sides and an interval [c, d] of the x-axis as its base (i.e., it is not necessary to require a solution to be deļ¬ned everywhere on the half-plane t ā„ 0). Indeed, from (2.13) and (2.14) we ļ¬nd the values f (x) and g(x) at x ā [c, d] (if the initial conditions are deļ¬ned on [c, d]). But this determines the values of u(t, x) when x ā at ā [c, d] and x + at ā [c, d], i.e., when characteristics through (t, x) intersect the segment [c, d]. We can, of course, assume f (x) and g(x) to be deļ¬ned on [c, d]; i.e., u(t, x) is deļ¬ned in the above-described triangle which is the dependence domain of the interval [c, d]. The physical meaning of the dependence domain is transparent: it is the complement to the set of points (t, x) such that the wave starting the motion with the velocity a at a point outside [c, d] at time t = 0 cannot reach the point x by time t. Further, the values of Ļ and Ļ on [c, d] do not aļ¬ect u(t, x) if x + at < c or x ā at > d, i.e., if the wave cannot reach the point x during the time t starting from an inside point of the interval [c, d]. This explains why the domain bounded by the segment [c, d] and the rays of the lines x + at = c, x ā at = d belonging to the half-plane t ā„ 0 is called the inļ¬uence domain of the interval [c, d]. This domain is shaded on Figure 2.4. This domain is the complement to the set of points (t, x) for which u(t, x) does not depend on Ļ(x) and Ļ(x) for x ā [c, d]. Example 2.1. Let us draw the form of the string at diļ¬erent times if Ļ(x) = 0 and Ļ(x) has the form shown in the uppermost graph of Figure 2.5; i.e., the graph of Ļ(x) has the form of an isosceles triangle with the base [c, d]. Let d ā c = 2l. We draw the form of the string at the most interesting times to show all of its essentially diļ¬erent forms. All these pictures together form an animation depicting vibrations of the string such that at the initial moment the string is pulled at the point 1 2 (c + d) whereas the points c and d are kept ļ¬xed, and then at t = 0 the whole string is released. The formula u(t, x) =
Ļ(x ā at) + Ļ(x + at) 2
makes it clear that the initial perturbation Ļ(x) splits into two identical waves one of which runs to the left, and the other to the right with the speed a.
24
2. One-dimensional wave equation u
u
c
t=0 x
d
t=Īµ x
u
t= u t= u t= u t=
l 2a x l +Īµ 2a x l a x l +Īµ a x
Figure 2.5. Animation to Example 2.1.
The dotted lines on Figure 2.5 depict divergent semiwaves 12 Ļ(x ā at) and 12 Ļ(x + at) whose sum is u(t, x). We assume Īµ > 0 to be suļ¬ciently small compared to the characteristic time interval al (actually it suļ¬ces that l ). Īµ < 2a
2.3. A semibounded string. Reļ¬ection of waves from the end of the string Consider the equation (2.5) for x ā„ 0, t ā„ 0. We need to impose a boundary condition at the end x = 0. For instance, let the motion of the stringās left end be given by (2.19)
u|x=0 = Ī±(t),
t ā„ 0,
and the initial conditions be given only for x ā„ 0: (2.20)
u|t=0 = Ļ(x),
ut |t=0 = Ļ(x),
x ā„ 0.
2.3. A semibounded string. Reļ¬ection of waves
25
The problem with initial and boundary conditions is usually called a mixed problem. Therefore, the problem with conditions (2.19) and (2.20) for a wave equation (2.5) is an example of a mixed problem (sometimes called the ļ¬rst boundary value problem for the string equation). Returning to the boundary value problem above, we can see that the simplest way to get an analytic solution is as in the Cauchy problem. The quadrant t ā„ 0, x ā„ 0 is a convex domain; therefore, we can again represent u in the form u(t, x) = f (x ā at) + g(x + at), where g(x) is now deļ¬ned and needed only for x ā„ 0 because x + at ā„ 0 for x ā„ 0, t ā„ 0. (At the same time f (x) is deļ¬ned, as before, for all x.) The initial conditions (2.20) deļ¬ne f (x) and g(x) for x ā„ 0 as in the case of an unbounded string and via the same formula. Therefore, for x ā at ā„ 0 the solution is determined by dāAlembertās formula, which is, of course, already clear. But now we can make use of the boundary condition (2.19) which gives f (āat) + g(at) = Ī±(t), implying (2.21)
z ā g(āz), f (z) = Ī± ā a
t ā„ 0,
z ā¤ 0.
The summands in (2.21) for z = x ā at give the sum of two waves running to the right. One of them is Ī±(t ā xa ) and is naturally interpreted as the wave generated by oscillations of the end of the string; the other one is āg(ā(x ā at)) and is interpreted as the reļ¬ection of the wave g(x + at) running to the left from the left end (note that this reļ¬ection results in the change of the sign). If the left end is ļ¬xed, i.e., Ī±(t) ā” 0, only one wave, āg(ā(x ā at)), remains and if there is no wave running to the left, then only the wave Ī±(t ā xa ) induced by the oscillation Ī±(t) of the end is left. Let us indicate a geometrically more transparent method for solving the problem with a ļ¬xed end. We wish to solve the wave equation (2.5) for t ā„ 0, x ā„ 0, with the initial conditions (2.20) and the boundary condition u|x=0 = 0,
t ā„ 0.
Let us try to ļ¬nd a solution in the form of the right-hand side of dāAlembertās formula (2.16), where Ļ(x), Ļ(x) are some extensions of functions Ļ(x), Ļ(x) onto the whole axis deļ¬ned for x ā„ 0 by initial conditions (2.20). The boundary condition yields
at 1 Ļ(āat) + Ļ(at) + Ļ(Ī¾)dĪ¾, 0= 2 2a āat
26
2. One-dimensional wave equation
implying that we will reach our goal if we extend Ļ and Ļ as odd functions, i.e., if we set Ļ(z) = āĻ(āz),
Ļ(z) = āĻ(āz),
z ā¤ 0.
Thus, we can construct a solution of our problem. The uniqueness of the solution is not clear from this method, but, luckily, the uniqueness was proved above. Example 2.2. Let an end be ļ¬xed, let Ļ(x) = 0, and let the graph of Ļ(x) again have the form of an isosceles triangle with the base [l, 3l]. Let us draw an animation describing the form of the string. Continuing Ļ(x) as an odd function, we get the same problem as in Example 2.1. With a dotted line we depict the graph over the points of the left half-axis (introducing it is just a mathematical trick since actually only the half-axis x ā„ 0 exists). We get the animation described on Figure 2.6. Of interest here is the moment t = 2l a when near the end the string is horizontal and so does not look perturbed. At the next moment, t = 2l a + Īµ, however, the perturbation arises due to a nonvanishing initial velocity. For t > 3l a + Īµ we get two waves running to the right, one of which is negative and is obtained by reļ¬ection from the left end of the wave running to the left.
2.4. A bounded string. Standing waves. The Fourier method (separation of variables method) Consider oscillations of a bounded string with ļ¬xed ends; i.e., solve the wave equation (2.5) with initial conditions (2.22)
u|t=0 = Ļ(x),
ut |t=0 = Ļ(x),
x ā [0, l],
and boundary conditions (2.23)
u|x=0 = u|x=l = 0,
t ā„ 0.
This can be done as in the preceding section. We will, however, apply another method which also works in many other problems. Note that the uniqueness of the solution is already proved with the help of the energy conservation law. We will ļ¬nd standing waves, i.e., the solutions of the wave equation (2.5) deļ¬ned for x ā [0, l], satisfying (2.23) and having the form u(t, x) = T (t)X(x).
2.4. A bounded string. Standing waves. The Fourier method
27
u ā3l
āl l
3l
t=0 x
u t=Īµ x u
t=
u
t=
u
t=
u
t=
u
t=
u
t=
u
t=
l a x l +Īµ a x 3l 2a x 3l +Īµ 2a x 2l a x 2l +Īµ a x 3l +Īµ a x
Figure 2.6. Animation to Example 2.2.
The term āstanding waveā is justiļ¬ed since the form of the string does not actually change with the time except for a factor which is constant in x and only depends on time. Substituting u(t, x) into (2.5), we get T (t)X(x) = a2 T (t)X (x).
28
2. One-dimensional wave equation
Excluding the uninteresting trivial cases when T (t) ā” 0 or X(x) ā” 0, we may divide this equation by a2 T (t)X(x) and get T (t) X (x) = . a2 T (t) X(x)
(2.24)
The left-hand side of this relation does not depend on x and the right-hand side does not depend on t. Therefore, they are constants; i.e., (2.24) is equivalent to the relations (2.25)
T (t) ā Ī»a2 T (t) = 0, X (x) ā Ī»X(x) = 0
with the same constant Ī». Further, boundary conditions (2.23) imply X(0) = X(l) = 0. The boundary value problem (2.26)
X = Ī»X,
X(0) = X(l) = 0
is a particular case of the so-called Sturm-Liouville problem. We will discuss the general Sturm-Liouville problem later. Now, we solve the problem (2.26); i.e., ļ¬nd the values Ī» (eigenvalues) for which the problem (2.26) has nontrivial solutions (eigenfunctions), and ļ¬nd these eigenfunctions. Consider the following possible cases. a) Ī» = Ī¼2 , Ī¼ > 0. Then the general solution of the equation X = Ī»X is X(x) = C1 sinh Ī¼x + C2 cosh Ī¼x. (Recall that sinh x = 12 (ex ā eāx ) and cosh x = 12 (ex + eāx ).) From X(0) = 0 we get C2 = 0, and from X(l) = 0 we get C1 sinh Ī¼l = 0 implying C1 = 0, i.e., X(x) ā” 0; hence, Ī» > 0 is not an eigenvalue. b) Ī» = 0. Then X(x) = C1 x + C2 and boundary conditions imply again C1 = C2 = 0 and X(x) ā” 0. c) Ī» = āĪ¼2 , Ī¼ > 0. Then X(x) = C1 sin Ī¼x + C2 cos Ī¼x; hence, X(0) = 0 implies C2 = 0 and X(l) = 0 implies C1 sin Ī¼l = 0. Assuming C1 = 0, we get sin Ī¼l = 0; therefore, the possible values of Ī¼ are kĻ , k = 1, 2, . . . , Ī¼k = l and the eigenvalues and eigenfunctions are 2 kĻ kĻx , k = 1, 2, . . . . , Xk = sin Ī»k = ā l l
2.4. A bounded string. Standing waves. The Fourier method
29
X1 (x)
0
l
Figure 2.7. Standing wave 1.
X2 (x) l/2
l
0
Figure 2.8. Standing wave 2.
X3 (x)
0
l/3
2l/3 l
Figure 2.9. Standing wave 3.
Let us draw graphs of several ļ¬rst eigenfunctions determining the form of standing waves (see Figures 2.7ā2.9). It is also easy to ļ¬nd the corresponding functions T (t). Namely, from (2.25) with Ī» = Ī»k we get kĻat kĻat + Bk sin , l l implying the following general form of the standing wave: kĻx kĻat kĻat + Bk sin sin , k = 1, 2, . . . . uk (t, x) = Ak cos l l l Tk (t) = Ak cos
The frequency of the oscillation of each point x corresponding to the solution uk is equal to kĻa , k = 1, 2, . . . . Ļk = l These frequencies are called the eigenfrequencies of the string. Now let us ļ¬nd the general solution of (2.5) with boundary conditions (2.23) as a sum (superposition) of stationary waves, i.e., in the form ā kĻx kĻat kĻat + Bk sin sin . Ak cos (2.27) u(t, x) = l l l k=1
30
2. One-dimensional wave equation
We need to satisfy the initial conditions (2.22). Substituting (2.27) into these conditions, we get (2.28)
Ļ(x) =
ā
Ak sin
k=1
(2.29)
Ļ(x) =
ā kĻa k=1
l
kĻx , l
Bk sin
x ā [0, l],
kĻx , l
x ā [0, l];
i.e., it is necessary to expand both Ļ(x) and Ļ(x) into series in eigenfunctions kĻx sin : k = 1, 2, . . . . l First of all, observe that this system is orthogonal on the segment [0, l]. This can be directly veriļ¬ed, but we may also refer to the general fact of orthogonality of eigenvectors of a symmetric operator corresponding to distinct eigenvalues. For such an operator one should take the operator d2 2 L = dx 2 with the domain DL consisting of functions v ā C ([0, l]) such that v(0) = v(l) = 0. Integrating by parts, we see that (Lv, w) = (v, Lw),
v, w ā DL .
The parentheses denote the inner product in L2 ([0, l]) deļ¬ned via
l v1 (x)v2 (x)dx. (v1 , v2 ) = 0 kĻx l }k=1,2,...
Thus, the system {sin to establish its completeness.
is orthogonal in L2 ([0, l]). We would like
For simplicity, take l = Ļ (the general case reduces to this one by introducing the variable y = Ļx l ) so that the system takes on the form ā {sin kx}k=1 and is considered on [0, Ļ]. Let f ā L2 ([0, Ļ]). Extend f as an odd function onto [āĻ, Ļ] and expand it into the Fourier series in the system {1, cos kx, sin kx : k = 1, 2, . . .} on the interval [āĻ, Ļ]. Clearly, this expansion does not contain constants and the terms with cos kx, since the extension is an odd function. Therefore, on [0, Ļ], we get the expansion in the system {sin kx : k = 1, 2, . . .} only. Note, further, that if f is continuous, has a piecewise continuous derivative on [0, Ļ], and f (0) = f (Ļ) = 0, then its 2Ļ-periodic odd extension is also continuous with a piecewise continuous derivative. Therefore, it can be expanded into a uniformly convergent series in terms of {sin kx : k = 1, 2, . . .}. If this 2Ļ-periodic extension is in C k and its (k + 1)st derivative is piecewise continuous, then this expansion can be diļ¬erentiated k times and the uniform convergence is preserved under this diļ¬erentiation.
2.4. A bounded string. Standing waves. The Fourier method
31
Thus, expansions into series (2.28) and (2.29) exist, and imposing some smoothness and boundary conditions on Ļ(x), Ļ(x), we can achieve arbitrarily fast convergence of these series. Coeļ¬cients of these series are uniquely deļ¬ned: l
kĻx 2 l kĻx 0 Ļ(x) sin l dx Ak = l = dx, Ļ(x) sin 2 kĻx l l 0 l dx 0 sin l
l 2 l 0 Ļ(x) sin kĻx kĻx l dx = dx. Ļ(x) sin Bk = l 2 kĻx kĻa kĻa 0 l sin dx 0
l
Substituting these values of Ak and Bk into (2.27), we get the solution of our problem. The Fourier method can also be used to prove the uniqueness of solution. (More precisely, the completeness of the trigonometric system enables us to reduce it to the uniqueness theorem for ODE.) We will illustrate this below by an example of a more general problem. Consider the problem of a string to which a distributed force f (t, x) is applied: (2.30)
utt = a2 uxx + f (t, x),
t ā„ 0, x ā [0, l].
The ends of the string are assumed to be ļ¬xed (i.e., conditions (2.23) are satisļ¬ed) and for t = 0 the initial position and velocity of the string, i.e., conditions (2.22), are deļ¬ned as above. Since the eigenfunctions {sin kĻx l : k = 1, 2, . . .} constitute a complete orthonormal system on [0, l], any function g(t, x), deļ¬ned for x ā [0, l] and satisfying reasonable smoothness conditions and boudnary conditions, can be expanded in terms of this system, with coeļ¬cients depending on t. In particular, we can write ā kĻx , uk (t) sin u(t, x) = l k=1
f (t, x) =
ā k=1
fk (t) sin
kĻx , l
where the fk (t) are known and the uk (t) are to be determined. Substituting these decompositions into (2.30), we get (2.31)
ā kĻx = 0, [ĀØ uk (t) + Ļk2 uk (t) ā fk (t)] sin l k=1
where the Ļk = kĻa l are the eigenfrequencies of the string. It follows from (2.31), due to the orthogonality of the system of eigenfunctions, that (2.32)
u ĀØk (t) + Ļk2 uk (t) = fk (t), k = 1, 2, . . . .
32
2. One-dimensional wave equation
The initial conditions ā
uk (0) sin
kĻx = Ļ(x), l
uĖ k (0) sin
kĻx = Ļ(x) l
k=1 ā k=1
determine uk (0) and uĖ k (0); hence, we can uniquely determine functions uk (t) and, therefore, the solution u(t, x). In particular, of interest is the case when f (t, x) is a harmonic oscillation in t; e.g., f (t, x) = g(x) sin Ļt. Then, clearly, the right-hand sides fk (t) of equations (2.32) are fk (t) = gk sin Ļt. For example, let gk = 0. Then for Ļ = Ļk (the nonresonance case) the equation (2.32) has an oscillating particular solutions of the form uk (t) = Uk sin Ļt, Uk =
gk2 . Ļk2 ā Ļ 2
Therefore, its general solution also has an oscillating form: uk (t) = Uk sin Ļt + Ak cos Ļk t + Bk sin Ļk t. For Ļ = Ļk (the resonance case) there is a particular solution of the form uk (t) = t(Mk cos Ļt + Nk sin Ļt), which the reader can visualize as an oscillation with the frequency Ļ and an increasing amplitude. The general solution uk (t) and the function u(t, x) are unbounded in this case. Observe that there is a more general problem that can be reduced to the problem above, namely the one with ends that are not ļ¬xed but whose motion is given by (2.33)
u|x=0 = Ī±(t),
u|x=l = Ī²(t), t ā„ 0.
Namely, if u0 (t, x) is an Ī“-function satisfying (2.33), e.g., x lāx Ī±(t) + Ī²(t), l l then for v(t, x) = u(t, x) ā u0 (t, x) we get a problem with ļ¬xed ends. u0 (t, x) =
Finally, note that all the above problems are well-posed, which is obvious from their explicit solutions.
2.5. Appendix. The calculus of variations and mechanics
33
2.5. Appendix. The calculus of variations and classical mechanics The calculus of variations is an analogue of the usual diļ¬erential calculus in the setting of an inļ¬nite-dimensional space. Functions on an inļ¬nitedimensional space are usually called functionals to distinguish them from functions on a ļ¬nite-dimensional space (which can be arguments of the functionals). 2.5.1. Variations of functionals. Euler-Lagrange equations. Let us consider a functional
t1 L(t, x(t), x(t))dt, Ė (2.34) S= t0
where t ā [t0 , t1 ], t0 < t1 , and we usually assume that x = x(t) ā C 1 ([t0 , t1 ]), Ė = dx(t)/dt are the usual so that xĖ ā C([t0 , t1 ]). The functions xĖ = x(t) derivatives of x(t) with respect to t. The functions x = x(t) ā C([t0 , t1 ]) can be considered as arguments of the functional S. Let L = L(t, x(t), x(t)) Ė = L(t, x, x)| Ė x=x(t),x= Ė x(t) Ė be the substitution result of x(t), x(t) Ė instead of x, x, Ė respectively. All functions are assumed to be real valued, with the smoothness suļ¬cient for the calculations. We will specify the appropriate level of smoothness wherever we need it. Now we can describe the variational setting and boundary conditions for our problem. Let us assume that we need to ļ¬nd a minimum of the functional S on a set of functions x ā C 1 ([t0 , t1 ]), whose values at the ends t0 , t1 are ļ¬xed. (In other words, we impose boundary conditions x(t0 ) = b0 , x(t1 ) = b1 where b0 , b1 are ļ¬xed constants.) Let x = x(t) be a minimizer: a function where this minimum is attained. For any function Ī“x ā C 1 ([t0 , t1 ]), satisfying the boundary conditions (2.35)
Ī“x(t0 ) = Ī“x(t1 ) = 0,
take also a function f = f (s) of one variable s ā R, deļ¬ned by
t1 L(t, x(t) + s Ī“x(t), x(t) Ė + s Ī“ x(t))dt. Ė (2.36) f (s) = S[x + s Ī“x] = t0
Clearly, it should attain a minimum at s = 0. (In this formula Ī“ xĖ should be understood as (d/dt)Ī“x(t).) Therefore,
t1 āL āL Ī“x + Ī“ xĖ dt = 0, f (0) = āx ā xĖ t0
34
2. One-dimensional wave equation
where āL/ā xĖ should be understood as āL(t, x(t), v)/āv|v=x(t) Ė . Integrating by parts in the second term, and taking into account the boundary conditions for Ī“x, we obtain
t1 āL d āL ā Ī“x dt = 0. āx dt ā xĖ t0 Since Ī“x can be an arbitrary function from C 2 ([t0 , t1 ]), vanishing at the ends t0 , t1 , we easily obtain that the expression in the parentheses should vanish identically; i.e., we get the equation d āL āL ā = 0, dt ā xĖ āx on the interval [t0 , t1 ]. This is a second-order ordinary diļ¬erential equation that is called the Euler-Lagrange equation. Its solution satisfying the boundary conditions x(t0 ) = b0 , x(t1 ) = b1 is called the extremal or the extremal point of the action functional S. All minimizers and maximizers of S should be extremal points. Sometimes, if it is known for some reasons that the minimum is attained and there is only one extremal, then this extremal must coincide with the minimizer. Algebraic machinery of the variational calculus Here we give a short list of algebraic notation in the calculus with one degree of freedom. Denote by Ī“S a linear functional on the space of functions Ī“x ā C 2 ([t0 , t1 ]), vanishing at the ends of the interval [t0 , t1 ], deļ¬ned by
t1 āL d āL Ī“x + Ī“ xĖ dt = Ī“S[Ī“x] = S[x + s Ī“x] ds āx ā xĖ t0 s=0
t1 d āL āL ā Ī“x dt. = āx dt ā xĖ t0 It is the principal linear part of the functional S at the point x, and it is called the variation of the functional S. Using the variation, we can rewrite the extremality condition in the form Ī“S = 0 which means that Ī“S[Ī“x] = 0 for all variations Ī“x, which vanish at the ends of the interval [t0 , t1 ]. Clearly, this is equivalent to the Euler-Lagrange equation. We can rewrite the last equation in the form
t1 Ī“S Ī“x dt, Ī“S = Ī“S[Ī“x] = t0 Ī“x where āL d āL Ī“S = ā . Ī“x āx dt ā xĖ Here the right-hand side is called the variational derivative of S.
2.5. Appendix. The calculus of variations and mechanics
35
The case of several degrees of freedom There is a straightforward extension of the arguments given above to the case when the action functional S depends upon several functions x1 (t), . . . , xn (t) belonging to C 2 ([t0 , t1 ]). In fact, it can be obtained if we just use the same formulas as above but for a vector function x instead of a scalar function. (This vector function can be interpreted as a parametrized curve in Rn , or, alternatively, a trajectory of a point moving in Rn .) Nevertheless, let us reproduce the main formulas, leaving the details of the arguments as an exercise for the readers. The action S is given by
t1 L(t, x1 (t), . . . , xn (t), xĖ 1 (t), . . . , xĖ n (t))dt, S[x] = t0
where L = L(t, x1 , . . . , xn , v1 , . . . , vn ) is the Lagrangian, which is a scalar real-valued function. The variations of the given (real-valued) functions x1 (t), . . . , xn (t) are real-valued C 2 -functions Ī“x1 (t), . . . , Ī“xn (t), vanishing at the ends t0 , t1 . The extremum conditions are obtained from the equations ā S[x1 + s1 Ī“x1 , . . . , xn + sn Ī“xn ] = 0, j = 1, . . . , n, āsj s1 =Ā·Ā·Ā·=sn =0 which lead to the Euler-Lagrange equations āL d āL ā = 0, dt ā xĖ j āxj
j = 1, . . . , n.
This system of equations is equivalent to the vanishing of the variation Ī“S, which is a linear functional on vector functions Ī“x = (Ī“x1 , . . . , Ī“xn ); it is given by
t1 n āL āL Ī“xj + Ī“ xĖ j dt Ī“S = Ī“S[Ī“x] = ā xĖ j t0 j=1 āxj
t1 n d āL āL ā Ī“xj dt. = āxj dt ā xĖ j t0 j=1
This can be rewritten in the form
t1
Ī“S = Ī“S[Ī“x] = t0
n Ī“S Ī“xj dt Ī“xj j=1
where āL d āL Ī“S = ā Ī“xj āxj dt ā xĖ j is called the partial variational derivative of S with respect to xj .
36
2. One-dimensional wave equation
2.5.2. Lagrangian approach in classical mechanics. The most important equations of classical mechanics can be presented in the form of the Euler-Lagrange equations. (In mechanics they are usually called Lagrangeās equations.) Take, for example, a point particle of mass m > 0 moving along the x-axis R in a time-independent (or stationary) ļ¬eld of forces F (x) (which is assumed to be a real-valued continuous function on R). Let x(t) denote the coordinate of the moving particle at the time t. Then by Newtonās Second Law, (mass)Ā·(acceleration)=(force), or, more precisely, mĀØ x(t) = F (x(t)) for all t. It is sometimes more convenient to rewrite this equation in the form dp = F, dt where p = mxĖ is called the momentum of the moving particle. Let us introduce a potential energy of the particle, which is a real-valued function U = U (x) on R, such that F (x) = āU (x) for all x. In other words,
x F (s)ds, U (x) = U (x0 ) ā x0
which means that to move the particle from x0 to x you need to do work U (x) ā U (x0 ) against the force F . (If U (x) < U (x0 ), then in fact the force does the work.) The Newton equation can be rewritten then as mĀØ x = āU (x). Multiplying both parts by x, Ė we can rewrite the result as d d mxĖ 2 = ā U (x), dt 2 dt where x = x(t). Taking antiderivative, we obtain mxĖ 2 + U (x) = E = const, 2 where the constant E depends on the solution x = x(t) (but is constant along the trajectory). This is the energy conservation law for the one-dimensional motion. The quantity mxĖ 2 K= 2 is called the kinetic energy of the moving particle, and the quantity H = K + U = H(x, x) Ė is called the energy, or full energy, or Hamiltonian. The energy conservation law can be rewritten in the form H(x, x) Ė = E = const.
2.5. Appendix. The calculus of variations and mechanics
37
Now let us take the Lagrangian L = L(x, x) Ė =KāU =
mxĖ 2 ā U (x) 2
and the corresponding action S. Then it is easy to see that the extremals of S will be exactly the solutions of the Newton Second Law equation, satisfying the chosen boundary conditions. In other words, the Euler-Lagrange equation for this action coincides with the Newton Second Law equation. Now consider a motion of a particle in Rn . (Note that any motion of k particles in R3 can be equivalently presented as a motion of one particle in R3k , the space whose coordinates consist of all coordinates of all particles.) The particle coordinates can be presented as a point x ā Rn , x = (x1 , . . . , xn ), so the motion of the particle is given by functions x1 (t), . . . , xn (t), or by one vector function x(t) (with values in Rn ). Let us assume that the force is F (x) = (F1 (x), . . . , Fn (x)), so that F (x) ā Rn for every x ā Rn , and F is continuous. (In the case when x represents a system of particles, F may include forces of interaction between these particles, which may depend on the conļ¬guration of these particles.) Then by Newtonās Second Law, the motion satisļ¬es the equations ĀØj = Fj (x), mj x
j = 1, . . . , n.
Let us assume also that the ļ¬eld of forces is a potential ļ¬eld, given by a potential energy U = U (x) (which is a scalar, real-valued function on Rn ), so that āU , j = 1, . . . , n, Fj (x) = ā āxj or, in the vector analysis notation, F = āāU = ā grad U. In this case, it is easy to see that the equations of motion coincide with the Euler-Lagrange equations for the action with the Lagrangian L = K ā U , where the kinetic energy K is given by K=
n mj xĖ 2j j=1
2
.
The Hamiltonian H = H(x, x) Ė = K + U satisļ¬es the energy conservation law H(x(t), x(t)) Ė = E = const, where x(t) is the solution of the equations of motion (or, equivalently, the extremal of the action). This can be veriļ¬ed
38
2. One-dimensional wave equation
by a straightforward calculation: d Ė dt H(x(t), x(t))
āH āH Ā· xĖ + Ā·x ĀØ āx ā xĖ n āH āH xĖ j + x ĀØj = āxj ā xĖ j j=1 n āU 1 āU āK ā xĖ j + = āxj ā xĖ j mj āxj j=1 n āU 1 āU = xĖ j + mj xĖ j ā = 0. āxj mj āxj =
j=1
Now let us consider a system of k particles in R3 . Assume that n = 3k and write x = (x1 , . . . , xk ), where xj ā R3 , j = 1, . . . , k. Let us view xj as a point in R3 where the jth particle is located. For moving particles we assume that xj = xj (t) is a C 2 -function of t. Let us introduce the momentum of the jth particle as pj = mj xĖ j , which is a vector in R3 . The momentum of the whole system of such particles is deļ¬ned as p = p1 + Ā· Ā· Ā· + pk , which is again a vector in R3 . Now assume that all forces acting on the particles are split into pairwise interaction forces between the particles and external forces, so that the force acting on the jth particle is Fij + Fext Fj = j , i=j
where Fij = Fij (x) ā R3 , and the summation is taken over i (with j ļ¬xed), 3 = Fext Fext j j (x) ā R . Here Fij is the force acting on the jth particle due to its interaction with the ith particle, and Fext is the external force acting j upon the jth particle. The equations of motion look like dpj = Fj = Fij + Fext j , dt
j = 1, . . . , k.
i=j
Now assume that Newtonās Third Law holds for two interacting particles: Fij (x) = āFji (x), for all i, j with i = j and for all x. Then we get dp = Fext = Fext j . dt k
j=1
2.6. Problems
39
In words, the rate of change of the total momentum in time equals the total external force. (The pairwise interaction forces, satisfying Newtonās Third Law, do not count.) Indeed, calculating the derivative of p, we see that all interaction forces cancel due to Newtonās Third Law. In particular, in the absence of external forces the conservation of momentum law holds: for any motion of these particles p(t) = const . Indeed, from the relations above we immediately see that dp/dt = 0. Remark. The mechanical system, describing a particle, moving in Rn , is said to have n degrees of freedom. In the main text of this chapter we discuss the motion of a string, which naturally has inļ¬nitely many degrees of freedom, because its position is described not by a ļ¬nite set of numbers but by a function (of one variable), or by an inļ¬nite set of its Fourier coeļ¬cients.
2.6. Problems 2.1. a) Write the boundary condition on the left end of the string if a ring which can slide without friction along the vertical line is attached to this end. The ring is assumed to be massless (see Figure 2.10). b) The same as in a) but the ring is of mass m. c) The same as in b) but the ring slides with a friction which is proportional to its velocity. 2.2. a) Derive the energy conservation law for a string if at the left end the boundary condition of Problem 2.1 a) is satisļ¬ed and the right end is ļ¬xed.
Figure 2.10. The ring attached to the end of the string.
Figure 2.11. The end of the rod is elastically attached.
40
2. One-dimensional wave equation
b) The same as in Problem 2.1 b) (take into account the energy of the ring). c) The same as in Problem 2.1 c) (take into account the loss of energy caused by friction). 2.3. Derive the equation for longitudinal elastic vibrations of a homogeneous rod. 2.4. Write the boundary condition for the left end of the rod if this end is free. 2.5. Write the boundary condition for the left end of the rod if this end is elastically restrained; see Figure 2.11. 2.6. Derive the energy conservation law for a rod with a) ļ¬xed ends; b) free ends; c) one end ļ¬xed and the other restrained elastically. 2.7. Derive an equation for longitudinal vibrations of a conic rod. 2.8. Prove that the energy of an interval of a string between c + at and d ā at is a decreasing function of time. Deduce from this the uniqueness of the solution of the Cauchy problem and information about the dependence and inļ¬uence domains for [c, d]. 2.9. Formulate and prove the energy conservation law for an inļ¬nite string. 2.10. Draw an animation describing oscillations of an inļ¬nite string with 1 for x ā [c, d], the initial values u|t=0 = 0, ut |t=0 = Ļ(x), where Ļ(x) = 0 for x ā [c, d]. 2.11. Describe oscillations of an inļ¬nite string for t ā (āā, +ā) such that some interval (x0 ā Īµ, x0 + Īµ) of the string is in equilibrium position at all times. 2.12. The same as in the above problem but the interval (x0 ā Īµ, x0 + Īµ) is ļ¬xed only for t ā„ 0. 2.13. Draw an animation describing oscillations of a semibounded string with a free end (the boundary condition ux |x=0 = 0, see Problem 2.1 a), and the initial conditions u|t=0 = Ļ(x), ut |t=0 = 0, where the graph Ļ(x) has the form of an isosceles triangle with the base [l, 3l]).
2.6. Problems
41
2.14. Draw an animation describing oscillations of a semibounded string with a ļ¬xed end and the initial conditions u|t=0 = 0, ut |t=0 = Ļ(x), where Ļ(x) is the characteristic function of the interval (l, 3l). 2.15. Derive the energy conservation law for a semibounded string with a ļ¬xed end. 2.16. Given the wave sin Ļ(t + xa ) running for t ā¤ 0 along the rod whose left end is elastically attached, ļ¬nd the reļ¬ected wave. 2.17. Given a source of harmonic oscillations with frequency Ļ running along an inļ¬nite string at a speed v < a, i.e., u|x=vt = sin Ļt for t ā„ 0, describe the oscillations of the string to the left and to the right of the source. Find frequencies of induced oscillations of a ļ¬xed point of the string to the left and to the right of the source and give a physical interpretation of the result (Dopplerās eļ¬ect). 2.18. Describe and draw standing waves for a string with free ends. 2.19. Prove the existence of an inļ¬nite series of standing waves in a rod with the ļ¬xed left end and elastically attached right one (see Problem 2.5). Prove the orthogonality of eigenfunctions. Find the shortwave asymptotics of eigenvalues. 2.20. The force F = F0 sin Ļt, where Ļ > 0, is applied to the right end of the rod for t ā„ 0. Derive a formula for the solution describing forced oscillations of the rod and ļ¬nd the conditions for a resonance if the left end is a) ļ¬xed; b) free. (Comment. The forced oscillations are the oscillations which have the frequency which coincides with the frequency of the external force or other external perturbation.) 2.21. The right end of a rod is oscillating via A sin Ļt for t ā„ 0. Write a formula for the solution describing forced oscillations and ļ¬nd the conditions for resonance if the left end is a) ļ¬xed; b) free; c) elastically restrained.
10.1090/gsm/205/03
Chapter 3
The Sturm-Liouville problem
3.1. Formulation of the problem Let us consider the equation Ļ(x)utt = (p(x)ux )x ā q(x)u,
(3.1)
which is more general than the equation derived in Chapter 2 for oscillations of a nonhomogeneous string. For the future calculations to be valid, we will assume that Ļ ā C 1 ([0, l]), Ļ > 0 on [a, b], p ā C 1 ([a, b]), and q ā C([a, b]) are real valued. Let us try to simplify the equation by a change of variables u(t, x) = z(x)v(t, x), where z(x) is a known function. This will allow us to obtain an equation of the form (3.1) with Ļ(x) ā” 1. Indeed, we have utt = zvtt , ux = zvx + zx v, uxx = zvxx + 2zx vx + zxx v. Substituting this into (3.1) and dividing by zĻ, we get p 2p zx px vtt = vxx + Ā· + vx + B(x)v. Ļ Ļ z Ļ This equation is of the form (3.1) with Ļ(x) ā” 1 whenever 2p zx px p Ā· + . = Ļ x Ļ z Ļ Performing the diļ¬erentiation in the left-hand side we see that this can be equivalently rewritten in the form Ļx zx =ā z 2Ļ 43
44
3. The Sturm-Liouville problem
and we may take, for example, z(x) = Ļ(x)ā1/2 . Since we always assume Ļ(x) > 0, such a substitution is possible. Thus, instead of (3.1), we may consider the equation utt = (p(x)ux )x ā q(x)u.
(3.2)
Solving it by separation of variables on the closed interval [0, l] and assuming that the ends are ļ¬xed, we get the following eigenvalue problem: (3.3)
ā(p(x)X (x)) + q(x)X(x) = Ī»X(x),
(3.4)
X(a) = X(b) = 0.
Generalizing it, we can take the same diļ¬erential operator while replacing the boundary conditions by more general conditions: Ī±X (a) + Ī²X(a) = 0, Ī³X (b) + Ī“X(b) = 0, where Ī±, Ī², Ī³, Ī“ are real numbers, Ī±2 + Ī² 2 = 0, Ī³ 2 + Ī“ 2 = 0. This eigenvalue problem is called the Sturm-Liouville problem. Considering this problem we usually suppose that p ā C 1 ([a, b]), q ā C([a, b]), and p(x) = 0 for x ā [0, l]. For simplicity, we assume that p(x) ā” 1 and the boundary conditions are as in (3.4). Thus, consider the problem āX (x) + q(x)X(x) = Ī»X(x), with the boundary conditions (3.4) and assume also that (3.5)
q(x) ā„ 0.
This is not a restriction since we can achieve (3.5) by adding the same constant to q and to Ī».
3.2. Basic properties of eigenvalues and eigenfunctions We will start with the following oscillation theorem of C. Sturm (1836): Theorem 3.1. Let y, z, q1 , q2 ā C 2 ([a, b]) be real-valued functions, such that z(a) = z(b) = 0, z ā” 0, y and z satisfy the equations (3.6)
y (x) + q1 (x)y(x) = 0,
(3.7)
z (x) + q2 (x)z(x) = 0.
Assume also that (3.8)
q1 (x) ā„ q2 (x) for all x ā [a, b].
3.2. Eigenvalues and eigenfunctions
45
Then, either there exists x0 ā (a, b) such that y(x0 ) = 0, or q1 (x) ā” q2 (x) and y(x) ā” cz(x) on [a, b], where c is a constant. Proof. Note that, due to the existence and uniqueness theorem for secondorder ODEs, all zeros of the function z are simple (i.e., if z(x0 ) = 0 for a point x0 ā [a, b], then z (x0 ) = 0). In particular, all zeros of z are isolated. Without loss of generality, we may assume that a and b are neighboring zeros of z(x) and z(x) > 0 for all x ā (a, b). Then z (a) > 0 and z (b) < 0. If y(x) does not vanish on (a, b), then we may assume that y(x) > 0 on (a, b). Let us consider the Wronskian (3.9)
W (x) = y(x)z (x) ā y (x)z(x),
x ā [a, b].
Diļ¬erentiating W (x), we obtain (3.10)
W (x) = y(x)z (x) ā y (x)z(x) = (q1 (x) ā q2 (x))y(x)z(x).
Integrating over [a, b], we get W (b) ā W (a) ā„ 0, and equality holds if and only if q1 (x) ā” q2 (x) on [a, b]. On the other hand, by assumption, y(a), y(b) ā„ 0. Then we have W (b) ā W (a) = y(b)z (b) ā y(a)z (a) ā¤ 0, with equality if and only if y(a) = y(b) = 0. Comparing the obtained inequalities, we see that W (b) ā W (a) = 0, implying both q1 (x) ā” q2 (x) and y(a) = y(b) = 0. Since z(a) = z(b) = 0, the solutions y and z must be proportional; i.e., y(x) = c z(x), where c is a constant. (Indeed, taking c such that y (a) = c z (a), the uniqueness theorem for ODEs implies y(x) ā” c z(x).) Remark 3.2. Arguing as in the proof of Theorem 3.1, we can also prove that the Wronskian W (x) of the two eigenfunctions with the same eigenvalues (and the identical potentials q(x) = q1 (x) = q2 (x)) are proportional. 2
d Let us introduce the Sturm-Liouville operator L = ā dx 2 + q(x), which is also called the one-dimensional SchrĀØ odinger operator. We consider L as an (unbounded) linear operator in L2 ([0, l]), with the domain DL consisting of all complex-valued functions v(x) ā C 2 ([0, l]) such that v(0) = v(l) = 0. The Sturm-Liouville problem is to ļ¬nd the eigenvalues and eigenvectors of this operator.
Note that the operator L is symmetric; i.e., (Lv1 , v2 ) = (v1 , Lv2 )
for v1 , v2 ā DL ,
where (Ā·, Ā·) is the inner product in L2 ([0, l]) deļ¬ned by
l u(x) v(x) dx. (u, v) = 0
46
3. The Sturm-Liouville problem
Indeed,
l
(Lv1 , v2 ) ā (v1 , Lv2 ) =
0
[āv1 vĀÆ2 + v1 vĀÆ2 ] dx
l
d [āv1 vĀÆ2 + v1 vĀÆ2 ] dx = [āv1 vĀÆ2 + v1 vĀÆ2 ]|l0 = 0. dx 0 This implies that all the eigenvalues are real and the eigenfunctions corresponding to diļ¬erent eigenvalues are orthogonal in L2 ([0, l]). =
In addition, all the eigenvalues are simple: eigenspaces are one-dimensional since any two solutions of the equation āy + q(x) y ā Ī» y = 0 that vanish at some point x0 are proportional to each other (and each of them is proportional, due to the uniqueness theorem, to the solution with the initial values y(x0 ) = 0, y (x0 ) = 1). Furthermore, the condition q(x) ā„ 0 and Sturmās theorem imply that all eigenvalues are positive. To establish this, let us compare the solutions to āz + q(x)z = Ī» z with those to āy = Ī» y. If an eigenfunction z(x) with the eigenvalue Ī» is given, then, by Sturmās theorem, any solution of the equation āy = Ī» y should vanish at a point of the segment [0, l]. But if Ī» ā¤ 0, then ā among the solutions of the equation āy = Ī» y there is a solution cosh āĪ» x which never vanishes. ā For Ī» > 0, consider the solution y = sin āĪ» x which vanishes on [0, l] at least at one point besides 0 precisely when Ī» ā„ Ļl . This clearly implies that any eigenvalue Ī» of L satisļ¬es
Ļ 2 Ī»ā„ . l ā From now on set for convenience k = Ī» so that the eigenvalue equation takes on the form (3.11)
āX + q(x)X = k 2 X.
We will always assume that k > 0. Denote by Ļ = Ļ(x, k) the solution of (3.11) with initial conditions Ļ(0, k) = 0,
Ļx (0, k) = k,
where Ļx denotes the derivative with respect to x. (For example, if q(x) ā” 0, then Ļ(x, k) = sin kx.) The eigenvalues of the Sturm-Liouville operator L are of the form Ī» = k 2 , where k is such that Ļ(l, k) = 0. The Sturm theorem implies that the number of zeros of Ļ(x, k) belonging to any ļ¬xed segment [0, a], where a ā¤ l, is an increasing function of k. Indeed, if 0 = x1 (k) < x2 (k) < Ā· Ā· Ā· < xn (k) are all such zeros of Ļ(Ā·, k) (xn (k) ā¤ a) and k > k, then every open interval (xj (k), xj+1 (k)) contains at least one zero of Ļ(Ā·, k ). Recalling that x1 (k) = 0 for all k, we see that
3.3. The short-wave asymptotics
47
the total number of zeros of Ļ(Ā·, k ) on [0, a] is greater than or equal to n, which proves the claim. Therefore, as k grows, all zeros of Ļ(x, k) move to the left. In fact, it is easy to see that this motion is continuous, even smooth. Indeed, the zeros xj (k) satisfy the equation Ļ(xj (k), k) = 0. But the function Ļ = Ļ(x, k) has continuous partial derivatives in x and k (cf. [12], Sec. V.3) and Ļx (xj (k), k) = 0, as observed above. By the implicit function theorem, the function xj = xj (k) is continuous. In fact, if we diļ¬erentiate the equation Ļ(xj (k), k) = 0 with respect to k, we obtain Ļx (xj (k), k) Ā· xj (k) = āĻk (xj (k), k), showing that xj (k) is also continuous. The eigenvalues correspond to the values k for which a new zero appears at l. Since the number of these zeros is ļ¬nite for any k, this implies that the eigenvalues form a discrete sequence Ī»1 < Ī»2 < Ī»3 < Ā· Ā· Ā· , which is either ļ¬nite or tends to inļ¬nity. The eigenfunction Xn (x) corresponding to the eigenvalue Ī»n has exactly n ā 1 zeros on the open interval (0, l). It is easy to see that the number of eigenvalues is actually inļ¬nite. Indeed, Sturmās theorem implies that the number of zeros of Ļ(x, k) on (0, l) is not less than the number of zeros on (0, l) for the corresponding solution of the equation āy + M y = k 2 y, where M = sup q(x). xā[0,l]
ā
But this solution is c sin k 2 ā M x with c = 0, and the number of its zeros on (0, l) tends to inļ¬nity as k ā +ā. Thus, we have proved Theorem 3.3. The eigenvalues are simple and form an increasing sequence Ī»1 < Ī»2 < Ī»3 < Ā· Ā· Ā· tending to +ā. If q1 ā„ 0, then Ī»1 > 0. The eigenfunctions Xn (x), corresponding to diļ¬erent eigenvalues Ī»n , are orthogonal. The eigenfunction Xn (x) has exactly n ā 1 zeros on the open interval (0, l).
3.3. The short-wave asymptotics Let us describe the asymptotic behavior of large eigenvalues and the corresponding eigenfunctions. The asymptotic formulas obtained are often called the high-frequency asymptotics or short-wave asymptotics, meaning that they correspond to high frequencies and therefore to short wavelengths in the nonstationary (i.e., time-dependent) problem.
48
3. The Sturm-Liouville problem
The behavior of eigenvalues is easily described with the help of oscillation d2 theorems. Namely, the eigenvalues of the operator L = ā dx 2 + q(x) are d2 contained between the eigenvalues of L1 = ā dx2 and the eigenvalues of d2 L2 = ā dx 2 + M , where M = max q(x). Since the eigenvalues of L1 and L2 xā[0,l]
Ļn 2 2 are equal to ( Ļn l ) and ( l ) + M , respectively (with n = 1, 2, . . .), we get
Ļn 2 l
ā¤ Ī»n ā¤
Ļn 2 l
+ M.
In particular, this implies the asymptotic formula
Ļn 2 1 1+O as n ā ā, Ī»n = l n2 ā or, putting kn = Ī»n , Ļn 1 1 Ļn = 1+O +O as n ā ā. (3.12) kn = l n2 l n Now, let us ļ¬nd the asymptotics of the eigenfunctions Xn (x). The idea is that for large k the term k 2 X in (3.11) plays a greater role than q(x)X. Therefore, let us solve the equation Ļ + k 2 Ļ = q(x)Ļ
(3.13) with the initial conditions (3.14)
Ļ(0) = 0,
Ļ (0) = k.
Treating (3.13) as a nonhomogeneous equation with right-hand side q(x)Ļ, we want to obtain an integral equation for Ļ = Ļ(x) = Ļ(x, k) that can be solved by successive approximations. Let us write Ļ(x) = C1 (x) cos kx + C2 (x) sin kx. Then the equations for C1 (x) and C2 (x), obtained by variation of parameters, are (3.15)
C1 (x) cos kx + C2 (x) sin kx = 0,
(3.16)
ākC1 (x) sin kx + kC2 (x) cos kx = q(x)Ļ.
Solving these equations, we ļ¬nd C1 (x) and C2 (x) in terms of Ļ. After integration this determines C1 (x) and C2 (x) up to arbitrary additive constants which are ļ¬xed by the initial values (3.17)
C1 (0) = 0,
C2 (0) = 1,
3.3. The short-wave asymptotics
obtained from (3.14). Then
49
x
C1 (x) = ā
q(Ļ )Ļ(Ļ ) 0
sin kĻ dĻ, k
x
cos kĻ dĻ, k 0 so Ļ satisļ¬es an integral equation (of so-called Volterra type):
1 x sin k(x ā Ļ )q(Ļ )Ļ(Ļ )dĻ. (3.18) Ļ(x) = sin kx + k 0 C2 (x) = 1 +
q(Ļ )Ļ(Ļ )
Clearly, the solution Ļ of this equation satisļ¬es (3.13) and (3.14) so it is equivalent to (3.13) with initial conditions (3.14). To solve (3.18) by successive approximations, consider the integral operator A deļ¬ned by the formula
x AĻ(x) = sin k(x ā Ļ )q(Ļ )Ļ(Ļ )dĻ. 0
The equation (3.18) becomes 1 A Ļ = sin kx, Iā k and its solution may be expressed in the form ā1 ā 1 n 1 sin kx = A (sin kx), (3.19) Ļ= Iā A k kn n=0
provided the series in the right-hand side of (3.19) converges and A can be applied termwise. It can be proved that the series (3.19) indeed converges uniformly in x ā [0, l] for all k but, since we are only interested in large k, we can restrict ourselves to an obvious remark that it converges uniformly on [0, l] for large k because the norm of A is bounded uniformly in k as an operator in the Banach space C([0, l]) (with sup norm) and the norm of sin kx in C([0, l]) is not greater than 1. In particular, we get 1 as k ā +ā. (3.20) Ļ(x, k) = sin kx + O k This gives the short-wave asymptotics of the eigenfunctions Xn . Indeed, it is clear from (3.12) and (3.20) that 1 1 1 Ļn +O +O ; = sin Xn (x) = Ļ(x, kn ) = sin(kn x) + O kn l n n hence Ļnx +O Xn (x) = sin l
1 as n ā +ā. n
50
3. The Sturm-Liouville problem
3.4. The Greenās function and completeness of the system of eigenfunctions 2
d Consider the SchrĀØodinger operator L = ā dx 2 + q(x) as a linear operator mapping its domain
DL = {v(x) ā C 2 ([0, l]) : v(0) = v(l) = 0} into C([0, l]). We will see that for q ā„ 0 this operator is invertible and the inverse operator Lā1 can be expressed as an integral operator
l ā1 G(x, y)f (y)dy for f ā C([0, l]), (3.21) L f (x) = 0
where G ā C([0, l]Ć[0, l]). The function G(x, y) is called the Greenās function of L. In general, if an operator is expressed in the form of the right-hand side of (3.21), then G(x, y) is called its kernel or Schwartzās kernel (in honor of L. Schwartz). Therefore, Lā1 has a continuous Schwartz kernel equal to G(x, y). Clearly, the function G(x, y) is uniquely deļ¬ned by (3.21). To ļ¬nd Lā1 f , we have to solve the equation (3.22)
āv (x) + q(x)v(x) = f (x)
with the boundary conditions (3.23)
v(0) = v(l) = 0.
This is performed by the variation of parameters. It is convenient to use two solutions y1 (x), y2 (x) of the homogeneous equation āy (x) + q(x)y(x) = 0 satisfying the boundary conditions (3.24)
y1 (0) = 0,
y1 (0) = 1
and (3.25)
y2 (l) = 0,
y2 (l) = ā1.
Clearly, the solutions y1 (x) and y2 (x) are linearly independent since 0 is not an eigenvalue of L because q(x) ā„ 0. Now, by the variation of parameters, writing v(x) = C1 (x)y1 (x) + C2 (x)y2 (x), we arrive to the equations (3.26)
C1 (x)y1 (x) + C2 (x)y2 (x) = 0,
(3.27)
C1 (x)y1 (x) + C2 (x)y2 (x) = āf (x).
The determinant of this system of linear equations for C1 (x) and C2 (x) is the Wronskian W (x) = y1 (x)y2 (x) ā y1 (x)y2 (x).
3.4. The Greenās function and eigenfunctions
51
It is well known (and easy to verify by diļ¬erentiation of W (x)āsee the calculation in the proof of Theorem 3.1) that W (x) = const and W (x) = 0 if and only if y1 and y2 are linearly dependent. Therefore, in our case W (x) = W = const = 0. Solving the system (3.26), (3.27) by Cramerās rule, we get 1 1 f (x)y2 (x), C2 (x) = ā f (x)y1 (x). W W Taking into account the boundary condition (3.23) for v and the boundary conditions (3.24), (3.25) for y1 , y2 , we see that C1 (x) and C2 (x) should be chosen so that C1 (l) = 0, C2 (0) = 0. C1 (x) =
This implies 1 C1 (x) = ā W
l x
1 f (Ī¾)y2 (Ī¾)dĪ¾, C2 (x) = ā W
x
f (Ī¾)y1 (Ī¾)dĪ¾. 0
Therefore, v exists, is unique, and is determined by the formula
x
l 1 1 y1 (Ī¾)y2 (x)f (Ī¾)dĪ¾ ā y1 (x)y2 (Ī¾)f (Ī¾)dĪ¾, v(x) = ā W 0 W x which can be expressed in the form
l G(x, Ī¾)f (Ī¾)dĪ¾, (3.28) v(x) = 0
where 1 [Īø(x ā Ī¾)y1 (Ī¾)y2 (x) + Īø(Ī¾ ā x)y1 (x)y2 (Ī¾)] W and Īø(z) is the Heaviside function: Īø(z) = 1 for z ā„ 0, Īø(z) = 0 for z < 0.
(3.29)
G(x, Ī¾) = ā
Note that G(x, Ī¾) is continuous and symmetric; i.e., (3.30)
G(x, Ī¾) = G(Ī¾, x).
(The property (3.30) is clear without explicit computation. Indeed, Lā1 should be symmetric due to the symmetry of L, and the symmetry of Lā1 is equivalent to that of G(x, Ī¾).) Consider now the operator G in L2 ([0, l]) with the kernel G(x, Ī¾):
l G(x, Ī¾)f (Ī¾)dĪ¾. Gf (x) = 0
Since G(x, Ī¾) is continuous and symmetric, G is a symmetric compact operator. By the Hilbert-Schmidt theorem it has a complete orthogonal system of eigenfunctions Xn (x) with real eigenvalues Ī¼n , where n = 1, 2, . . . and Ī¼n ā 0 as n ā ā. (For the proofs of these facts from functional analysis
52
3. The Sturm-Liouville problem
see, e.g., Sect. VI.5, especially Theorem VI.16, in Reed and Simon [25].) We have GXn = Ī¼n Xn .
(3.31)
If the Xn are continuous, then, applying L to (3.31), we see that Xn = Ī¼n LXn . This implies that Ī¼n = 0 and Ī»n = 1/Ī¼n is an eigenvalue of L. The continuity of Xn follows easily from (3.31), provided Ī¼n = 0. In general, if f ā L2 ([0, l]), then Gf ā C([0, l]) because by the Cauchy-Schwarz inequality l |(Gf )(x ) ā (Gf )(x )| = [G(x , Ī¾) ā G(x , Ī¾)]f (Ī¾)dĪ¾ 0
ā¤ sup |G(x , Ī¾) ā G(x , Ī¾)| Ā· Ī¾ā[0,l]
ā¤ sup |G(x , Ī¾) ā G(x , Ī¾)| Ā·
ā
l
|f (Ī¾)|dĪ¾
0 l
1/2 |f (Ī¾)| dĪ¾ 2
l 0
Ī¾ā[0,l]
and because of the fact that G(x, Ī¾) is uniformly continuous on [0, l] Ć [0, l]. It remains to prove that G cannot have zero as an eigenvalue on L2 ([0, l]). But if u ā L2 ([0, l]) and Gu = 0, then u is orthogonal to the image of G, since then (3.32)
(u, Gf ) = (Gu, f ) = 0
for all f ā L2 ([0, l]).
But all the functions v ā DL can be represented in the form Gf since we have v = G(Lv) by construction of G. Since DL is dense in L2 ([0, l]), we see that (3.32) implies u = 0. Thus, the set of eigenfunctions of G coincides exactly with the set of eigenfunctions of L. In particular, we have proved the completeness of the system of eigenfunctions of L in L2 ([0, l]). The following properties of the Greenās function are easily deducible with the help of (3.29): a) G(x, Ī¾) has continuous derivatives up to the second order for x = Ī¾ and d2 satisļ¬es the equation Lx G(x, Ī¾) = 0 (also for x = Ī¾), where Lx = ā dx 2 +q(x).
b) G(x, Ī¾) is continuous everywhere, whereas its derivative Gx has left and right limits at x = Ī¾ and the jump is equal to ā1: Gx (Ī¾ + 0, Ī¾) ā Gx (Ī¾ ā 0, Ī¾) = ā1.
c) the boundary conditions G(0, Ī¾) = G(l, Ī¾) = 0 are satisļ¬ed for all Ī¾ ā [0, l]. These properties may be used to ļ¬nd G without variation of parameters. It is easy to verify that G(x, Ī¾) is uniquely deļ¬ned by these conditions.
3.4. The Greenās function and eigenfunctions
53
Conditions a) and b) can be easily expressed with the help of the Dirac Ī“-function. Later on we will do this more carefully, but now we just give some heuristic arguments (on the āphysicalā level of rigor). It is convenient to apply formula (3.21) which expresses Lā1 in terms of its kernel from the very beginning. Let us formally apply L to (3.21) to obtain
l f (x) = Lx G(x, Ī¾)f (Ī¾) dĪ¾. 0
If we denote Ī“(x, Ī¾) = Lx G(x, Ī¾), then this object formally satisļ¬es
l Ī“(x, Ī¾)f (Ī¾) dĪ¾. (3.33) f (x) = 0
It is clear that Ī“(x, Ī¾) vanishes for Ī¾ = x and yet (taking f (x) ā” 1 in (3.33)) we have
l Ī“(x, Ī¾) dĪ¾ = 1. 0
We can say more about Ī“(x, Ī¾). Taking f with a small support and shifting the support, we obtain from (3.33) that Ī“(x + z, Ī¾ + z) = Ī“(x, Ī¾) for any z. Therefore, Ī“(x, Ī¾) should only depend on x ā Ī¾; i.e., Ī“(x, Ī¾) = Ī“(x ā Ī¾) for an object Ī“(x) that vanishes for all x ā R\{0} and satisļ¬es
ā Ī“(x) dx = 1. āā
There is no locally integrable function with this property; nevertheless, it is convenient to make use of the symbol Ī“(x) if it is a part of an integrand, keeping in mind that
ā Ī“(x)f (x) dx = f (0) for any f ā C(R). āā
The āfunctionā Ī“(x) is called Diracās Ī“-function or the Dirac measure. In the theory of distributions (that we shall discuss in the following chapter), Ī“(x) is understood as a linear functional on functions whose value on a continuous function f (x) is equal to f (0). The Ī“-function can be interpreted as a point mass, or a point charge, or a point load. For example, let a force f be applied to a nonhomogeneous string. Then, when oscillations subside (e.g., due to friction), we see that the stationary form of the string v(x) should satisfy (3.22), (3.23) and, therefore, is expressed by formula (3.28). If the load is concentrated near a point Ī¾ and the total load equals 1, i.e., f (x)dx = 1, then the form of the string
54
3. The Sturm-Liouville problem
is precisely the graph of G(x, Ī¾). This statement can easily be made into a more precise one by passing to the limit when a so-called Ī“-like sequence of loads fn is chosen, e.g., such that fn (x) ā„ 0; fn (x) = 0 for |x ā Ī¾| ā„ 1/n; and fn (x)dx = 1. We skip the details leaving them as an easy exercise for the reader.
3.5. Problems 3.1. Derive an integral equation for a function Ī¦(x) satisfying āĪ¦ + q(x)Ī¦ = k 2 Ī¦, Ī¦(0) = 1, Ī¦ (0) = 0, with q ā G([0, l]) and deduce from this integral equation the asymptotics of Ī¦(x) and Ī¦ (x) as k ā +ā. 3.2. Using the result of Problem 3.1, prove the existence of an inļ¬nite number of eigenvalues for the following Sturm-Liouville problem: āX + q(x)X = Ī»X,
X (0) = X(l) = 0.
Prove the symmetry of the corresponding operator and the orthogonality of eigenfunctions. 2
d 3.3. Write explicitly the Greenās function of the operator L = ā dx 2 with the boundary conditions v(0) = v(l) = 0. Give a physical interpretation.
2
d 3.4. Write explicitly the Greenās function of the operator L = ā dx 2 +1 with the boundary conditions v (0) = v (l) = 0. Give a physical interpretation.
3.5. With the help of the Greenās function, prove the completeness of the d2 system of eigenfunctions for the operator L = ā dx 2 +q(x) with the boundary conditions v (0) = v (l) = 0 corresponding to free ends. 2
d 3.6. Prove that if q(x) ā„ 0, then the Greenās function G(x, Ī¾) of L = ā dx 2 + q(x) with the boundary conditions v(0) = v(l) = 0 is positive for x = 0, l and Ī¾ = 0, l.
3.7. Prove that under the conditions of the previous problem, the Greenās function G(x, Ī¾) is a positive kernel; i.e., the matrix (G(xi , xj ))ni,j=1 is positive deļ¬nite for any ļ¬nite set of distinct points x1 , . . . , xn ā (0, l).
3.5. Problems
55
3.8. Expand the Greenās function G(x, Ī¾) of the Sturm-Liouville problem into a series involving the eigenfunctions Xn (x) and express in terms of the Greenās function the following sums: ā ā 1 1 a) , b) (diļ¬cult) , Ī»n Ī»2n n=1
n=1
where Ī»n (n = 1, 2, . . .) is the set of eigenvalues for this problem. In particā ā 1 1 and . ular, use these to calculate n2 n4 n=1
n=1
10.1090/gsm/205/04
Chapter 4
Distributions
4.1. Motivation of the deļ¬nition. Spaces of test functions In analysis and mathematical physics one often meets technical diļ¬culties caused by the nondiļ¬erentiability of certain functions. The theory of distributions (generalized functions) makes it possible to overcome these diļ¬culties. The Dirac Ī“-function, which naturally appears in many situations (one of them was mentioned in Section 3.4), is an example of a distribution. In the theory of distributions a number of notions and theorems of analysis acquire simpler formulations and are free of unnatural and irrelevant restrictions. The origin of the notion of distribution or generalized function may be explained as follows. Let f (x) be a physical quantity (such as temperature, pressure, etc.) which is a function of x ā Rn . If we measure this quantity at a point x0 with the help of some device (thermometer, pressure gauge, etc.), then we actually measure a certain average of f (x) over a neighbor hood of x0 , that is, an integral of the form f (x)Ļ(x)dx, where Ļ(x) is a function characterizing the measuring gadget and smeared somehow over a neighborhood of x0 . The idea is to completely abandon the function f (x) and consider instead a linear functional by assigning to each test function Ļ the number
(4.1)
f, Ļ =
f (x)Ļ(x)dx.
Considering arbitrary linear functionals (not necessarily of the form (4.1)) we come to the notion of a distribution or generalized function. 57
58
4. Distributions
Let us introduce now the needed spaces of test functions. Let Ī© be an open subset of Rn . Let E(Ī©) = C ā (Ī©) be the space of smooth (inļ¬nitely diļ¬erentiable) functions on Ī©; let D(Ī©) = C0ā (Ī©) be the space of smooth functions with compact support in Ī©, i.e., functions Ļ ā C ā (Ī©) such that there exists a compact K ā Ī© with the property Ļ|Ī©\K = 0. In general, the support of Ļ ā C(Ī©) (denoted by supp Ļ) is the closure (in Ī©) of the set {x ā Ī© : Ļ(x) = 0}. Therefore, supp Ļ is the minimal closed set F ā Ī© such that Ļ|Ī©\F = 0 or, equivalently, the complement to the maximal open set G ā Ī© such that Ļ|G = 0. Therefore, D(Ī©) consists of Ļ ā C ā (Ī©) such that supp Ļ is compact in Ī©. Let K be the closure of a bounded open set in Rn and let D(K) = C0ā (K) be the space of functions Ļ ā C ā (Rn ) such that supp Ļ ā K. Clearly, D(Ī©) is a union of all D(K) over compacts K ā Ī©. Finally, as a space of test functions we will also use S(Rn ), L. Schwartzās space, consisting of functions Ļ(x) ā C ā (Rn ) such that sup |xĪ± ā Ī² Ļ(x)| < xāRn
+ā for any multi-indices Ī± and Ī². The theory of distributions makes use of topology in the spaces of test functions. We will introduce it with the help of a system of seminorms. First, we will give general deļ¬nitions. A seminorm in a linear space E is a map p : E āā [0, +ā) such that a) p(x + y) ā¤ p(x) + p(y), x, y ā E; b) p(Ī»x) = |Ī»|p(x), x ā E, Ī» ā C. The property p(x) = 0 does not necessarily imply x = 0 (if this implication holds, then the seminorm is called a norm). Let a system of seminorms {pj }jāJ , where J is a set of indices, be deļ¬ned on E. For any Īµ > 0 and j ā J, set Uj,Īµ = {x : x ā E, pj (x) < Īµ} and introduce the topology in E whose basis of neighborhoods of 0 consists of all ļ¬nite intersections of the sets Uj,Īµ (a basis of neighborhoods of any other point x0 is obtained by the shift by x0 , i.e., consists of ļ¬nite intersections of the sets x0 + Uj,Īµ ). Therefore, G ā E is open if and only if together with each point x0 it also contains a ļ¬nite intersection of sets x0 + Uj,Īµ . Clearly, lim xn = x if and only if lim pj (xn ā x) = 0 for any j ā J. nā+ā
nā+ā
To avoid perpetual reference to ļ¬nite intersections, we may assume that for any j1 , j2 ā J there exists j3 ā J such that pj1 (Ļ) ā¤ pj3 (Ļ) and pj2 (Ļ) ā¤ pj3 (Ļ) for any Ļ ā E (in this case Uj3 ,Īµ ā (Uj1 ,Īµ ā© Uj2 ,Īµ )). If this does not hold, we can add (to the existing system of seminorms) additional seminorms
4.1. Motivation. Spaces of test functions
59
of the form pj1 ...jk (Ļ) = max{pj1 (Ļ), . . . , pjk (Ļ)}. This does not aļ¬ect the topology deļ¬ned by seminorms in E and we will always assume that this is already done. In other words, we will assume that the basis of neighborhoods of 0 in E consists of all sets Uj,Īµ . The same topology in E can be deļ¬ned by a diļ¬erent set of seminorms. Namely, assume that we have another system of seminorms {pj }j āJ , with the corresponding neighborhoods of 0 denoted Uj ,Īµ . Clearly, the systems of seminorms {pj }jāJ , {pj }j āJ deļ¬ne the same topology if and only if the following two conditions are satisļ¬ed: ā¢ for every j ā J and Īµ > 0, there exist j ā J and Īµ > 0 such that Uj,Īµ ā Uj ,Īµ ; ā¢ for every j ā J and Īµ > 0, there exist j ā J and Īµ > 0 such that ā Uj,Īµ .
Uj ,Īµ
It is easy to see that these conditions are satisļ¬ed if and only if ā¢ for every j ā J there exist j ā J and C = Cj ,j > 0 such that pj (x) ā¤ Cpj (x), x ā E; ā¢ for every j ā J there exist j ā J and C = Cj,j > 0 such that pj (x) ā¤ C pj (x), x ā E.
If these conditions are satisļ¬ed for two systems of seminorms, {pj }, {pj }, then we will say that the systems are equivalent. Since we are only interested in translation invariant topologies on vector spaces E, we can freely replace a system of seminorms by any equivalent system. If f is a linear functional on E, then it is continuous if and only if there exist j ā J and C > 0 such that (4.2)
|f, Ļ| ā¤ Cpj (Ļ),
Ļ ā E,
where f, Ļ denotes the value of f on Ļ. Indeed, the continuity of f is equivalent to the existence of a set Uj,Īµ such that |f, Ļ| ā¤ 1 for Ļ ā Uj,Īµ and the latter is equivalent to (4.2) with C = 1/Īµ. Note that the seminorm pj itself, as a function on E, is continuous on E. Indeed, this is obvious at the point Ļ = 0 because the set {Ļ| pj (Ļ) < Īµ} is a neighborhood of 0 for any Īµ > 0. Now the continuity of pj at an arbitrary point Ļ0 ā E follows, because the triangle inequality a) above implies p(Ļ0 ) ā p(Ļ) ā¤ pj (Ļ0 + Ļ) ā¤ p(Ļ0 ) + p(Ļ), so the continuity at Ļ0 follows from the continuity at 0.
60
4. Distributions
We will usually suppose that the following separability condition is satisļ¬ed: if pj (Ļ) = 0 for all j ā J, then Ļ = 0.
(4.3)
This implies that any two distinct points x, y ā E have nonintersecting neighborhoods (the Hausdorļ¬ property of the topology). Indeed, there exists j ā J such that pj (x ā y) > 0. But then the neighborhoods x + Uj,Īµ , y + Uj,Īµ do not intersect for Īµ < 12 pj (x ā y), because if z = x + t = y + s ā (x + Uj,Īµ ) ā© (y + Uj,Īµ ), then pj (x ā y) = pj (s ā t) ā¤ pj (s) + pj (t) ā¤ 2Īµ < pj (x ā y). Now, consider the case which is most important for us when J is a countable (inļ¬nite) set. We may assume that J = N. Introduce a metric on E by the formula ā 1 pk (x ā y) . (4.4) Ļ(x, y) = 2k 1 + pk (x ā y) k=1
The symmetry and the triangle inequality are easily checked. Due to separability condition (4.3), Ļ(x, y) = 0 implies x = y. Let us prove that the topology deļ¬ned by the metric coincides with the topology deļ¬ned above by seminorms. Since the metric (4.4) is invariant with respect to the translations (i.e., Ļ(x + z, y + z) = Ļ(x, y)), it is suļ¬cient to verify that each neighborhood Uj,Īµ contains a ball BĪ“ (0) = {x : Ļ(x, 0) < Ī“} of radius Ī“ > 0 with the center at the origin, and, conversely, each ball BĪ“ (0) contains a set of the form Uj,Īµ . (Note that in the future we will also write B(Ī“, 0) instead of BĪ“ (0).) First, take an arbitrary Ī“ > 0. Let us prove that there exist j, Īµ such that Uj,Īµ ā BĪ“ (0). If j is such that p1 (x) ā¤ pj (x), . . . , pN (x) ā¤ pj (x), then Ļ(x, 0) ā¤
N pk (x) k=1
2k
+
ā k=N +1
1 1 pk (x) ā¤ pj (x) + N . k 2 1 + pk (x) 2
Therefore, if Īµ < Ī“/2 and 1/2N < Ī“/2, then Uj,Īµ ā BĪ“ (0). Conversely, given j and Īµ, we deduce the existence of Ī“ > 0 such that BĪ“ (0) ā Uj,Īµ from the obvious inequality 1 pj (x) ā¤ Ļ(x, 0). 2j 1 + pj (x) 1 Īµ t . Then, since f (t) = 21j 1+t is strictly 2j 1 + Īµ increasing for t > 0, the inequality Ļ(x, 0) < Ī“ implies pj (x) < Īµ.
Namely, take Ī“ so that Ī“
0, such that for every j, k, which are both greater than m, we have xj āxk ā U . Clearly, the last condition only depends upon the topology in E. But the notion of convergence also depends on the topology only. Therefore, the completeness property of a countably normed space is independent of the choice of a system of seminorms in the same equivalence class. Now we can also formulate the notion of completeness in terms of seminorms. Using (4.5) and the deļ¬nition of topology in terms of seminorms, we see that a sequence {xj | j = 1, 2, . . . } is a Cauchy sequence (in terms of topology or metric) if and only if (4.6)
pl (xj ā xk ) ā 0
as j, k ā +ā,
for every seminorm pl . So to check the completeness of a countably normed space it suļ¬ces to check convergence of the sequences satisfying (4.6), which is sometimes more convenient. Deļ¬nition 4.2. A FrĀ“echet space is a topological vector space whose topology is deļ¬ned by a structure of a complete countably normed topological
62
4. Distributions
vector space. In other words, it is a complete topological vector space with a countable system of seminorms deļ¬ned up to the equivalence explained above. Since the topology in a countably normed space E can be determined by a metric Ļ(x, y), the continuity of functions on E and, in general, of any maps of E into a metric space can be determined in terms of sequences. For instance, a linear functional f on E is continuous if and only if lim Ļk = 0 kā+ā
implies lim f, Ļk = 0. kā+ā
Deļ¬nition 4.3. The space of continuous linear functionals in E is called the dual space to E and is denoted by E . Now, let us turn D(K), E(Ī©), and S(Rn ) into countably normed spaces. 1) Space D(K). Denote sup |ā Ī± Ļ(x)|, pm (Ļ) = |Ī±|ā¤m
Ļ ā D(K).
xāK
The seminorms pm (Ļ), m = 1, 2, . . . , deļ¬ne a structure of a countably normed space on D(K). Clearly, the convergence Ļp ā Ļ in the topology of D(K) means that if Ī± is any multi-index, then ā Ī± Ļp (x) ā ā Ī± Ļ(x) uniformly on K. Let us show that D(K) is in fact a FrĀ“echet space. The only nonobvious thing to check is its completeness. Let {Ļj | j = 1, 2, . . . } be a Cauchy sequence in D(K). This means that for any multi-index Ī± the sequence {ā Ī± Ļj } is a Cauchy sequence in C0 (K), the space of all continuous functions on Rn supported in K (i.e., vanishing in Rn \K), with the sup norm p0 . Then there exists ĻĪ± ā C0 (K) such that ā Ī± Ļj ā ĻĪ± in C0 (K) as j ā +ā (i.e., converges uniformly on Rn ). It remains to prove that in fact ĻĪ± = ā Ī± Ļ0 for all Ī±. But this follows from the standard analysis results on the possibility to diļ¬erentiate a uniformly convergent sequence, if the sequence of derivatives also converges uniformly. 2) Space E(Ī©). Let Kl ā Ī©, l = 1, 2, . . ., be a sequence of compacts such that K1 ā K2 ā K3 ā Ā· Ā· Ā· and for every x ā Ī© there exists l such that x belongs to the interior of Kl (i.e., K contains a neighborhood of x). For instance, we may set Kl = {x : x ā Ī©, |x| ā¤ l, Ļ(x, āĪ©) ā„ 1/l}, ĀÆ where āĪ© is the boundary of Ī© (i.e., āĪ© = Ī©\Ī©) and Ļ is the usual Euclidean n distance in R . Set sup |ā Ī± Ļ(x)|, Ļ ā E(Ī©). pl (Ļ) = |Ī±|ā¤l
xāKl
4.2. Spaces of distributions
63
These seminorms turn E(Ī©) into a countably normed space. Clearly, the convergence Ļp ā Ļ in the topology of E(Ī©) means that ā Ī± Ļp (x) ā ā Ī± Ļ(x) uniformly on any compact K ā Ī© and for any ļ¬xed multi-index Ī±. In particular, this implies that the topology deļ¬ned in this way does not depend on the choice of a system of compacts Kl described above. 3) Space S(Rn ). Deļ¬ne the system of seminorms pk (Ļ) = sup |xĪ± ā Ī² Ļ(x)|, |Ī±|+|Ī²|ā¤k
xāRn
and deļ¬ne the convergence Ļp ā Ļ in the topology of S(Rn ) means that if A is a diļ¬erential operator with polynomial coeļ¬cients on Rn , then AĻp (x) ā AĻ(x) uniformly on Rn . Both E(Ī©) and S(Rn ) are also FrĀ“echet spaces. Their completeness can be established by the same arguments as the ones used for the space D(K) above.
4.2. Spaces of distributions The above procedure of transition from E to its dual E makes it possible to deļ¬ne E (Ī©) and S (Rn ) as dual spaces of E(Ī©) and S(Rn ), respectively. We have not yet introduced topology in D(Ī©) (the natural topology in D(Ī©) is not easy to deļ¬ne) but we will not need it. Instead we deļ¬ne D (Ī©) as the space of linear functionals f on D(Ī©) such that f |D(K) is continuous on D(K) for any compact K ā Ī©. This continuity can also be deļ¬ned in terms of sequences (rather than seminorms), because the topology in D(K) can be deļ¬ned by a metric. It is also convenient to introduce the convergent sequences in D(Ī©), saying that a sequence {Ļk }k=1,2,... , Ļk ā D(Ī©), converges to Ļ ā D(Ī©) if the following two conditions are satisļ¬ed: (a) there exists a compact K ā Ī© such that Ļ ā D(K) and Ļk ā D(K) for all k; (b) Ļk ā Ļ in D(K), or, in other words, for every ļ¬xed multi-index Ī±, ā Ī± Ļk ā ā Ī± Ļ uniformly on K (or on Ī©). Deļ¬nition 4.4. (1) The elements of D (Ī©) are called distributions in Ī©. (2) The elements of E (Ī©) are called distributions with compact support in Ī©. (The notion of support of a distribution that justiļ¬es the above term will be deļ¬ned later.) (3) The elements of S (Rn ) are called tempered distributions on Rn .
64
4. Distributions
Example 4.1. The āregularā distributions on Ī©. Let L1loc (Ī©) be the space of (complex-valued) functions on Ī© which are Lebesgue integrable with respect to the Lebesgue measure on any compact K ā Ī©. To any f ā L1loc (Ī©), assign the functional on D(Ī©) (denoted by the same letter), setting
(4.7) f, Ļ = f (x)Ļ(x)dx, Ī©
where Ļ ā D(Ī©). Clearly, we get a distribution on Ī© (called regular). An important fact is that if two functions f1 , f2 ā L1loc (Ī©) determine the same distribution, they coincide almost everywhere. This is a corollary of the following lemma. Lemma 4.5. Let f ā L1loc (Ī©) and Ī© f (x)Ļ(x)dx = 0 for any Ļ ā D(Ī©). Then f (x) = 0 for almost all x. Proof of this lemma requires the existence of an ample set of functions in D(Ī©). We do not know yet whether there exist such nontrivial (not identically vanishing) functions. Let us construct a supply of functions belonging to D(Rn ). First, let Ļ(t) = Īø(t)eā1/t , where t ā R and Īø(t) is the Heaviside function. Clearly, Ļ(t) ā C ā (R). Therefore, if we consider the function Ļ(x) = Ļ(1 ā |x|2 ) in Rn , we get a function Ļ(x) ā C0ā (Rn ) such that Ļ(x) = 0 for |x| ā„ 1 and Ļ(x) > 0 for |x| < 1. It is convenient to normalize Ļ considering instead the function Ļ1 (x) = cā1 Ļ(x), where c = Ļ(x)dx. Now, set ĻĪµ (x) = Īµān Ļ1 (x/Īµ). Then ĻĪµ (x) = 0 for |x| ā„ Īµ and ĻĪµ (x) > 0 for |x| < Īµ. Besides, ĻĪµ (x)dx = 1. Now, introduce an important averaging (or mollifying) operation: for any f (x) ā L1loc (Ī©) take a family of convolutions
(4.8) fĪµ (x) = f (x ā y)ĻĪµ (y)dy = f (y)ĻĪµ (x ā y)dy deļ¬ned for x ā Ī©Īµ = {x : x ā Ī©, Ļ(x, āĪ©) > Īµ}. It is clear from the dominated convergence theorem that the last integral in (4.8) is a C ā function and
ā Ī± fĪµ (x) =
f (y)(ā Ī± ĻĪµ )(x ā y)dy;
hence fĪµ ā C ā (Ī©Īµ ). Note also the following properties of averaging: a) If f (x) = 0 outside of a compact K ā Ī©, then fĪµ (x) = 0 outside of the Īµ-neighborhood of K. In particular, in this case fĪµ ā D(Ī©) for a suļ¬ciently small Īµ > 0. b) If f (x) = 1 for x ā Ī©2Īµ , then fĪµ (x) = 1 for x ā Ī©3Īµ . In particular, if f is the characteristic function of Ī©2Īµ , then fĪµ ā D(Ī©) and fĪµ (x) = 1 for x ā Ī©3Īµ .
4.2. Spaces of distributions
65
c) If f ā C(Ī©), then fĪµ (x) ā f (x) uniformly on any ļ¬xed compact K ā Ī© as Īµ ā 0+. Indeed,
fĪµ (x) ā f (x) =
[f (x ā y) ā f (x)]ĻĪµ (y)dy,
implying |fĪµ (x) ā f (x)| ā¤ sup |f (x ā y) ā f (x)| |y|ā¤Īµ
so that the uniform continuity of f on the Īµ-neighborhood of K implies our statement. d) If f ā Lploc (Ī©), where p ā„ 1, then fĪµ ā f with respect to the norm in for any ļ¬xed compact K ā Ī© as Īµ ā 0+.
Lp (K)
Since the values fĪµ (x) at x ā K only depend on the values of f on the Īµ-neighborhood of the compact K, we may assume that f (x) = 0 for x ā Ī©\K1 , where K1 is a compact in Ī©. Let the norm in Lp (Ī©) be denoted by up ; i.e.,
1/p |u(x)|p dx . up = Ī©
Due to Minkowskiās inequality (see the appendix in this chapter: Section 4.7)
fĪµ p = f (Ā· ā y)ĻĪµ (y)dy ā¤ f (Ā· ā y)p |ĻĪµ (y)|dy = f p . p
Due to c), the property d) holds for f ā C(Ī©). But if f ā Lp (Ī©) and f = 0 outside K1 , then we can approximate f with respect to the norm in Lp (Ī©), ļ¬rst by step functions and then by continuous functions on Ī© which vanish outside a compact K2 ā Ī©. Let g be a continuous function on Ī© which vanishes outside K2 and such that f ā gp < Ī“. Then lim sup fĪµ ā f p ā¤ lim sup fĪµ ā gĪµ p + lim sup gĪµ ā gp + lim sup g ā f p Īµā0+
Īµā0+
Īµā0+
Īµā0+
= lim sup (f ā g)Īµ p + g ā f p ā¤ 2f ā gp ā¤ 2Ī“, Īµā0+
which due to the arbitrariness of Ī“ > 0 implies lim sup fĪµ ā f p = 0, as Īµā0+
required. Proof of Lemma 4.5. Let f ā L1loc (Ī©) and Ī© f (x)Ļ(x)dx = 0 for every Ļ ā D(Ī©). This implies fĪµ (x) = 0 for all x ā Ī©Īµ . Therefore, the statement of the lemma follows from the property d).
66
4. Distributions
Lemma 4.5 allows us to identify the functions f ā L1loc (Ī©) with the distributions that they deļ¬ne by formula (4.7). Note that the expression
(4.9) f, Ļ = f (x)Ļ(x)dx is also often used for distributions f (in this case the left-hand side of (4.9) is the deļ¬nition for the right-hand side; when the right-hand side makes sense, this deļ¬nition is consistent). Example 4.2. āRegularā distributions in E (Ī©) and S (Rn ). If f ā L1comp (Ī©), i.e., f ā L1 (Rn ) and f (x) = 0 outside a compact K ā Ī©, then we may construct via the standard formula (4.9) a functional on E(Ī©) which is a distribution with compact support. We will identify f with the corresponding element of E (Ī©). If f ā L1loc (Rn ) and (4.10)
|f (x)|(1 + |x|)āN < +ā
for some N > 0, then (4.9) deļ¬nes a functional on S(Rn ) which is a tempered distribution. In particular, (4.10) holds if f is measurable and |f (x)| ā¤ C(1 + |x|)N ānā1 . Example 4.3. Diracās Ī“-function. This is the functional deļ¬ned as follows: (4.11)
Ī“(x ā x0 ), Ļ(x) = Ļ(x0 ).
If x0 ā Ī©, then Ī“(x ā x0 ) ā D (Ī©) and, moreover, Ī“(x ā x0 ) ā E (Ī©). Obviously, Ī“(x ā x0 ) ā S (Rn ) for any x0 ā Rn . Instead of (4.11) one often writes, in accordance with the above-mentioned convention,
Ī“(x ā x0 )Ļ(x)dx = Ļ(x0 ), though, clearly, Ī“(x ā x0 ) is not a regular distribution. Indeed, if we assume that Ī“(xāx0 ) is regular, then Lemma 4.5 would imply that Ī“(xāx0 ) vanishes almost everywhere in Rn \{x0 }; hence, it vanishes almost everywhere in Rn , and so vanishes as a distribution in Rn . Example 4.4. Let L be a linear diļ¬erential operator in Rn , Ī a smooth compact hypersurface in Rn , dS the area element of Ī. Then the formula
(4.12) Ļ ā (LĻ)|Ī dS Ī
deļ¬nes a nonregular distribution with compact support in Rn . When L is the identity operator, we denote the distribution (4.12) by Ī“Ī . For Ī we may take a compact submanifold (perhaps with a boundary) of any codimension in Rn . In this case for dS we may take any density on Ī
4.3. Topology and convergence
67
(or a diļ¬erential form of the maximal order if the orientation of Ī is ļ¬xed). In particular, the Dirac Ī“-function appears as a particular case when Ī is a point.
4.3. Topology and convergence in the spaces of distributions Let F be one of the spaces of test functions D(Ī©), E(Ī©), or S(Rn ), and let F be the corresponding dual space of distributions (continuous linear functionals on F ). We will consider the space F with so-called weak topology, i.e., the topology deļ¬ned by the seminorms pĻ (f ) = |f, Ļ|,
Ļ ā F , f ā F .
In particular, the convergence fk ā f in this topology means that fk , Ļ ā f, Ļ for any Ļ ā F . In this case we will also use the usual notation lim fk = f . kāā
Up to now we assumed that the limit f is itself a distribution. Now let us simply assume that {fk | k = 1, 2, . . . } is a sequence of distributions (from F ), such that for any Ļ ā F there exists a limit f, Ļ := lim fk , Ļ. kā+ā
Will the limit f always be a distribution too? It is clear that the limit is a linear functional with respect to Ļ. But will it be continuous? The answer is aļ¬rmative, and the corresponding property is called completeness (or weak completeness) of the distribution space. It is a particular case of a general result which holds for any FrĀ“echet space (see the appendix in this chapter: Section 4.8). Example 4.5. Ī“-like sequence. Consider the above-constructed functions ĻĪµ (x) ā D(Rn ). Recall that they satisfy a) ĻĪµ (x) ā„ 0 for all x; b) ĻĪµ (x)dx = 1; c) ĻĪµ (x) = 0 for |x| ā„ Īµ. Let us prove that the properties a)āc) imply lim ĻĪµ (x) = Ī“(x)
(4.13)
Īµā0+
in D (Rn ). Indeed, this means that
ĻĪµ (x)Ļ(x)dx = Ļ(0) for any Ļ ā D(Rn ). (4.14) lim Īµā0+
With the help of b) we may rewrite (4.14) in the form of the relation
ĻĪµ (x)[Ļ(x) ā Ļ(0)]dx = 0, lim Īµā0+
68
4. Distributions
which is obviously true due to the estimate
ĻĪµ (x)[Ļ(x) ā Ļ(0)]dx ā¤ sup |Ļ(x) ā Ļ(0)|. |x|ā¤Īµ
Note that (4.13) holds not only in D (Rn ) but also in S (Rn ), E (Rn ), and even in E (Ī©) if 0 ā Ī©. Sequences of āregularā functions converging to the Ī“-function are called Ī“-like sequences. Properties a)āc) can be considerably weakened retaining Ī“-likeness of the sequence. Thus, in the theory of Fourier series one proves the Ī“-likeness (e.g., in D ((āĻ, Ļ))) of the sequence of Dirichlet kernels Dk (t) =
1 sin(k + 12 )t , 2Ļ sin 2t
deļ¬ned by the condition that Dk , Ļ is the sum of the ļ¬rst k terms of the Fourier series of Ļ at t = 0. Similarly, a Ī“-like sequence in D ((āĻ, Ļ)) is constituted by the FejĀ“er kernel Fk (t) =
1 sin2 kt 2 , 2Ļk sin2 2t
deļ¬ned from the condition that Fk , Ļ is the arithmetic mean of the ļ¬rst k partial sums of the Fourier series. Later on we will encounter a number of other important examples of Ī“-like sequences. Note that the Ī“-likeness is essentially always proved in the same way as for ĻĪµ (x). 1 1 and v.p. x1 . The āregularā functions x+iĪµ Example 4.6. Distributions xĀ±i0 ā 1 and xāiĪµ (here i = ā1) of x ā R1 for Īµ > 0 determine tempered distributions (elements of S (R1 )). It turns out that there exist limits in S (R1 )
1 1 = lim , x + i0 Īµā0+ x + iĪµ
1 1 = lim . x ā i0 Īµā0+ x ā iĪµ
The left-hand sides here are, by deļ¬nition, the limits on the right-hand sides. Let us prove, e.g., the existence of the limit
1 x+i0
in S (R1 ).
We must prove that if Ļ ā S(R1 ), then
ā Ļ(x) dx (4.15) lim Īµā0+ āā x + iĪµ exists and is a continuous linear functional of Ļ ā S(R1 ). The simplest way to do this is to integrate by parts. Namely, we have d 1 = ln(x + iĪµ), x + iĪµ dx
4.3. Topology and convergence
69
where we take an arbitrary branch of ln(x + iĪµ) (but keeping continuity for all x ā R, Īµ > 0). Integrating by parts, we get
ā
ā Ļ(x) (4.16) dx = ā Ļ (x) ln(x + iĪµ)dx. x + iĪµ āā āā Since ln(x + iĪµ) = ln |x + iĪµ| + iarg (x + iĪµ), it is clear that, by the dominated convergence theorem, the limit of the right-hand side of (4.16), as Īµ ā 0+, is equal to
0
ā (ln |x|)Ļ (x)dx ā iĻ Ļ (x)dx ā āā āā (4.17)
ā =ā (ln |x|)Ļ (x)dx ā ĻiĻ(0). āā
and ln |x| grows at inļ¬nity not faster than |x|Īµ for any Since ln |x| ā Īµ > 0, it is obvious that the right-hand side of (4.17) is ļ¬nite and deļ¬nes a 1 . continuous functional of Ļ ā S(Rn ). This functional is denoted by x+i0 L1loc (R)
Let us study in detail the ļ¬rst summand in (4.17). We have
ā
ā
āā
(ln |x|)Ļ (x)dx = lim
(4.18)
Īµā0+
ā
ln |x| Ā·
= lim
Īµā0+
= lim
|x|ā„Īµ
Īµā0+ |x|ā„Īµ
ln |x| Ā· Ļ (x)dx
Ļ(x)|ĪµāĪµ
+ |x|ā„Īµ
1 Ļ(x)dx x
1 Ļ(x)dx, x
since lim ln Īµ[Ļ(Īµ) ā Ļ(āĪµ)] = lim O(Īµ) ln Īµ = 0. In particular, we have Īµā0+
Īµā0+
proved that the last limit in (4.18) exists and deļ¬nes a functional belonging to S (R1 ). This functional is denoted v.p. x1 (the letters v.p. stand for the ļ¬rst letters of the French words āvaleur principaleā, meaning the principal value). Therefore, by deļ¬nition,
1 1 (4.19) v.p. , Ļ = lim Ļ(x)dx. Īµā0+ x |x|ā„Īµ x Moreover, we may now rewrite (4.17) in the form (4.20)
1 1 = v.p. ā ĻiĪ“(x). x + i0 x
Similarly, we get (4.21)
1 1 = v.p. + ĻiĪ“(x). x ā i0 x
70
4. Distributions
Formulas (4.20) and (4.21) are called Sokhotskyās formulas. ticular, imply
They, in par-
1 1 ā = ā2ĻiĪ“(x). x + i0 x ā i0 The existence of limits in (4.15) and (4.19) may also be proved by expanding Ļ(x) in the sum of a function equal to Ļ(0) in a neighborhood of 0 and a function which vanishes at 0. For each summand, the existence of limits is easy to verify. In the same way, one gets the Sokhotsky formulas, too. 1 and v.p. x1 are distinct āregularizationsā of the noninDistributions xĀ±i0 1 tegrable ā 1 function x ; i.e., they allow us to make sense of the divergent integral āā x Ļ(x)dx. We see that there is an ambiguity: various distributions correspond to the nonintegrable function x1 . The regularization procedure is important if we wish to use x1 as a distribution (e.g., to diļ¬erentiate it).
This procedure (or a similar one) is applicable to many other nonintegrable functions.
4.4. The support of a distribution Let Ī©1 , Ī©2 be two open subsets of Rn and let Ī©1 ā Ī©2 . Then D(Ī©1 ) ā D(Ī©2 ). Therefore, if f ā D (Ī©2 ), we may restrict f to D(Ī©1 ) and get a distribution on Ī©1 denoted by f |Ī©1 . It is important that the operation of restriction has the following properties: a) Let{Ī©j : j ā J} be a cover of an open set Ī© by open sets Ī©j , j ā J; i.e., Ī© = jāJ Ī©j . If f ā D (Ī©) and f |Ī©j = 0 for j ā J, then f = 0. b) Let again Ī© = jāJ Ī©j and let fj ā D (Ī©j ) be a set of distributions such that fk |Ī©k ā©Ī©l = fl |Ī©k ā©Ī©l for any k, l ā J. Then there exists a distribution f ā D (Ī©) such that f |Ī©j = fj for any j ā J. Properties a) and b) mean that distributions constitute a sheaf. Let us prove the property a), which is important for us. To do this, let us use a partition of unity, i.e., a family of functions {Ļj }jāJ , such that 1) Ļj ā C ā (Ī©), supp Ļj ā Ī©j ; 2) the family {Ļj }jāJ is locally ļ¬nite; i.e., any point x0 ā Ī© has a neighborhood in which only a ļ¬nite number of functions of this family, Ļj1 , . . . , Ļjk , do not vanish; 3) jāJ Ļj ā” 1. The existence of a partition of unity is proved in courses of geometry and analysis; see, e.g., Theorem 10.8 in Rudin [27] or Theorem 1.11 in Warner [38].
4.4. The support of a distribution
71
If Ļ ā D(Ī©), then property 1) implies Ļj Ļ ā D(Ī©j ) and properties 2) and 3) imply Ļ= Ļj Ļ, jāJ
where the sum locally contains only a ļ¬nite number of summands which are not identically zero, since the compact supp Ļ intersects, due to property 2), only a ļ¬nite number of supports supp Ļj . If f |Ī©j = 0 for all j, then f, Ļj Ļ = 0, f, Ļ = jāJ
implying f = 0, due to the arbitrariness of Ļ. This proves property a). Property b) may be proved by constructing f ā D (Ī©) from distributions fj ā D (Ī©j ) with the help of the formula fj , Ļj Ļ. f, Ļ = jāJ
If Ļ ā D(Ī©k ), where k is ļ¬xed, then f, Ļ =
jāJ
fj , Ļj Ļ =
fk , Ļj Ļ =
jāJ
fk ,
Ļj Ļ
= fk , Ļ
jāJ
since supp(Ļj Ļ) ā supp Ļj ā© supp Ļ ā Ī©j ā© Ī©k . We have therefore veriļ¬ed that f |Ī©k = fk , proving property b). Property a) allows us to introduce for f ā D (Ī©) a well-deļ¬ned maximal open subset Ī© ā Ī© such that f |Ī© = 0 (here Ī© is the union of all open sets Ī©1 ā Ī© such that f |Ī©1 = 0). The closed subset F = Ī©\Ī© is called the support of the distribution f (denoted by supp f ). Therefore, supp f is the smallest closed subset F ā Ī© such that f |Ī©\F = 0. Example 4.7. supp Ī“(x) = {0}. The support of the distribution from Example 4.4 belongs to Ī. In general, if supp f ā F , then f is said to be supported on F . Therefore, a distribution of the form (4.12) is supported on Ī. Now, let us deļ¬ne supports of distributions from E (Ī©) and S (Rn ). This is easy if we note that there are canonical embeddings E (Ī©) ā D (Ī©), S (Rn ) ā D (Rn ). They are constructed with the help of embeddings D(Ī©) ā E(Ī©), D(Rn ) ā S(Rn ),
72
4. Distributions
which enable us to deļ¬ne maps (4.22)
E (Ī©) āā D (Ī©)
and (4.23)
S (Rn ) āā D (Rn )
by restricting functionals onto the smaller subspace. The continuity of these functionals on D(Ī©) and D(Rn ) follows from the continuity of embeddings D(K) ā E(Ī©) for a compact K in Ī©, D(K) ā S(Rn ) for a compact K in Rn . Finally, the dual maps (4.22), (4.23) are embeddings, because D(Ī©) is dense in E(Ī©) and D(Rn ) is dense in S(Rn ); therefore any continuous linear functional on E(Ī©) (respectively, S(Rn )) is uniquely determined by its values on a dense subset D(Ī©) (respectively, D(Rn )). In what follows, we will identify distributions from E (Ī©) and S (Rn ) with their images in D (Ī©) and D (Rn ), respectively. Proposition 4.6. Let f ā D (Ī©). Then f ā E (Ī©) if and only if supp f is a compact subset of Ī©, or, in other words, if f is supported on a compact subset of Ī©. Proof. Let f ā E (Ī©). Then f is continuous with respect to a seminorm in E(Ī©); i.e., (4.24)
|f, Ļ| ā¤ C
|Ī±|ā¤l
sup |ā Ī± Ļ(x)|,
xāK
where the compact K ā Ī© and numbers C, l do not depend on Ļ. But this implies that f, Ļ = 0 for any Ļ ā D(Ī©\K), so f |Ī©\K = 0; i.e., supp f ā K. Conversely, let f ā D (Ī©) and supp f ā K1 , where K1 is a compact in Ī©. Clearly, if K ā Ī© contains a neighborhood of K1 , then f is determined by its restriction onto D(K) and (4.24) holds for Ļ ā D(K), hence, for Ļ ā D(Ī©) (since f, Ļ does not depend on Ļ|Ī©\K ). Therefore, f can be continuously extended to a functional f ā E (Ī©). Note that this extension essentially reduces to the following two steps: 1) decompose Ļ ā E(Ī©) into a sum Ļ = Ļ1 + Ļ2 , where Ļ1 ā D(Ī©) and Ļ2 = 0 in a neighborhood of supp f and 2) set f, Ļ = f, Ļ1 . Proposition 4.6 is proved.
4.4. The support of a distribution
73
The following important theorem describes distributions supported at a point. Theorem 4.7. Let f ā E (Rn ) and supp f = {0}. Then f has the form cĪ± (ā Ī± Ļ)(0), (4.25) f, Ļ = |Ī±|ā¤l
where l and constants cĪ± do not depend on Ļ. Proof. To any Ļ ā E(Rn ) assign its l-jet at 0, i.e., the set of derivatives j0l (Ļ) = {(ā Ī± Ļ)(0)}|Ī±|ā¤l . Choose l so as to satisfy (4.24), where K is a compact which is a neighborhood of 0. Let us verify that f, Ļ actually depends only on j0l (Ļ). Then the statement of the theorem becomes obvious since it reduces to the description of linear functionals on a ļ¬nite-dimensional vector space of l-jets at 0 and these functionals are deļ¬ned as in the right-hand side of (4.25). Thus, it remains to verify that j0l (Ļ) = 0 implies f, Ļ = 0. Let Ļ ā D(Rn ) be chosen so that Ļ(x) = 1 if |x| ā¤ 1/2 and Ļ(x) = 0 if |x| ā„ 1. Set ĻĪµ (x) = Ļ( xĪµ ), Īµ > 0. Clearly, supp f = {0} implies f, Ļ = f, ĻĪµ Ļ for any Īµ > 0. We will see below that j0l (Ļ) = 0 implies (4.26)
sup |ā Ī± [ĻĪµ (x)Ļ(x)]| ā 0
xāRn
for |Ī±| ā¤ l as Īµ ā 0 + .
Therefore f, Ļ = f, ĻĪµ Ļ = lim f, ĻĪµ Ļ = 0 Īµā0+
due to (4.24). It remains to prove (4.26). But by the Leibniz rule cĪ± Ī± [ā Ī± ĻĪµ (x)][ā Ī± Ļ(x)] ā Ī± [ĻĪµ (x)Ļ(x)] = Ī± +Ī± =Ī±
=
Ī± +Ī± =Ī±
x [ā Ī± Ļ(x)]. cĪ± Ī± Īµā|Ī± | (ā Ī± Ļ) Īµ
It follows from Taylorās formula that
ā Ī± Ļ(x) = O(|x|l+1ā|Ī± | )
as x ā 0.
Note that |x| ā¤ Īµ on supp Ļ( xĪµ ). Therefore, as ā 0+, x Ī± sup Īµā|Ī± | (ā Ī± Ļ) |ā Ļ(x)| = O(Īµā|Ī± |+l+1ā|Ī± | ) = O(Īµl+1ā|Ī±| ) ā 0, n Īµ xāR implying (4.26).
74
4. Distributions
4.5. Diļ¬erentiation of distributions and multiplication by a smooth function Diļ¬erentiation of distributions should be a natural extension of diļ¬erentiation of smooth functions. To deļ¬ne it note that if u ā C 1 (Ī©) and Ļ ā D(Ī©), then
āĻ āu Ļdx = ā u dx (4.27) āxj āxj or
(4.28)
āu ,Ļ āxj
āĻ = ā u, āxj
.
āu Therefore, it is natural to deļ¬ne āx for u ā D (Ī©) by the formula (4.28) j (which we will sometimes write in the form (4.27) in accordance with the above-mentioned convention). Namely, deļ¬ne the value of the functional āu āxj at Ļ ā D(Ī©) (i.e., the left-hand side of (4.27)) as the right-hand side
which makes sense because
āĻ āxj
ā D(Ī©). Since
to D(K) for any compact K ā Ī©, we see deļ¬ne higher derivatives
ā āxj continuously maps D(K) āu ā D (Ī©). We can now that āx j
ā Ī± : D (Ī©) āā D (Ī©), where Ī± = (Ī±1 , . . . , Ī±n ) is a multi-index, as products of operators
ā āxj ,
namely, by setting ā Ī± = ( āxā 1 )Ī±1 . . . ( āxān )Ī±n . Another (equivalent) way to do this is to write (4.29)
ā Ī± u, Ļ = (ā1)|Ī±| u, ā Ī± Ļ,
which enables us to immediately compute derivatives of an arbitrary order. The successive application of (4.28) shows that the result obtained from (4.29) is equal to that obtained by applying ā Ī± considered as the composition of operators āxā j . Since ā Ī± continuously maps E(Ī©) to E(Ī©) and S(Rn ) to S(Rn ), it is clear that ā Ī± maps E (Ī©) to E (Ī©) and S (Rn ) to S (Rn ). It is also clear from (4.29) that ā Ī± is a continuous operator in D (Ī©), E (Ī©), and S (Rn ). Therefore, we may say that ā Ī± is the extension by continuity of the usual diļ¬erentiation onto the space of distributions. Such an extension is unique, because, as we will see in the next section, D(Ī©) is dense in D (Ī©). Finally, it is clear that supp(ā Ī± u) ā supp u
for all u ā D (Ī©).
4.5. Diļ¬erentiation and multiplication by a smooth function
75
Example 4.8. The derivative of the Heaviside function Īø(x). The distributional derivative of Īø(x) can be calculated as follows:
ā Īø , Ļ = āĪø, Ļ = ā Ļ (x)dx = Ļ(0), Ļ ā D(R1 ); 0
i.e., Īø (x) = Ī“(x). Example 4.9. The derivative ā Ī± Ī“(x) of Ī“(x) ā D (Rn ). By deļ¬nition, ā Ī± Ī“(x), Ļ(x) = (ā1)|Ī±| Ī“(x), ā Ī± Ļ(x) = (ā1)|Ī±| Ļ(Ī±) (0). Therefore, ā Ī± Ī“(x) assigns the number (ā1)|Ī±| Ļ(Ī±) (0) to the function Ļ(x). Theorem 4.7 may now be formulated as follows: any distribution supported at 0 is a ļ¬nite linear combination of derivatives of the Ī“-function. Example 4.10. The fundamental solution of the Laplace operator. Consider the Laplace operator Ī in Rn and try to ļ¬nd a distribution u ā D (Rn ) such that Īu = Ī“(x). Such u is called a fundamental solution for the operator Ī. It is not unique: we can add to it any distributional solution of the Laplace equation Īv = 0 (in this case v will be actually a C ā - and even real-analytic function, as we will see later). Fundamental solutions play an important role in what follows. We will try to select a special (and the most important) fundamental solution, using symmetry considerations. (It will later be referred to as the fundamental solution.) The operator Ī commutes with rotations of Rn (or, equivalently, preserves its form under an orthogonal transformation of coordinates). Therefore, by rotating a solution u we again get a solution of the same equation (it is easy to justify the meaning of the process, but, anyway, we now reason heuristically). But then we may average with respect to all rotations. It is also natural to suppose that u has no singularities at x = 0. Therefore, we will seek a distribution u(x) of the form f (r), where r = |x|, for x = 0. Let us calculate Īf (r). We have xj ār ā ā f (r) = f (r) = f (r) x21 + Ā· Ā· Ā· + x2n = f (r) āxj āxj āxj r implying
2 2 x x 1 ā2 j j ā 3 . f (r) = f (r) 2 + f (r) r r r āx2j
Summing over j = 1, . . . , n we get Īf (r) = f (r) +
nā1 f (r). r
76
4. Distributions
Let us solve the equation f (r) +
nā1 f (r) = 0. r
Denoting f (r) = g(r), we get for g(r) a separable ODE g (r) +
nā1 g(r) = 0, r
which yields g(r) = Crā(nā1) , where C is a constant. Integrating, we get C1 rā(nā2) + C2 for n = 2, f (r) = for n = 2. C1 ln r + C2 Since constants C2 are solutions of the homogeneous equation Īu = 0, we may take f (r) = Cr2ān
for n = 2,
f (r) = C ln r
for n = 2.
We will only consider the case n ā„ 2 (the case n = 1 is elementary and will later be analyzed in a far more general form). For convenience of the future calculations we introduce 1 2ān r for n ā„ 3, (4.30) un (x) = 2ān u2 (x) = ln r.
(4.31)
Then un (x) is locally integrable in Rn and hence may be considered as a 1 1ān n is introduced so that un satisļ¬es āu distribution. The factor 2ān ār = r for all n ā„ 2. Now it is convenient to make a break in our exposition to include some integral formulas for the fundamental solutions (4.30), (4.31) and related functions. Some integral formulas We will use several formulas relating integrals over a bounded domain (open set) Ī© with some integrals over its boundary. The ļ¬rst formula of this kind is the fundamental theorem of calculus
b f (x)dx = f (b) ā f (a), f ā C 1 ([a, b]), a
which is due to Newton and Leibniz. All other formulas of the described type (due to Gauss, Ostrogradsky, Green, Stokes, and others) follow from this one, most often through integration by parts.
4.5. Diļ¬erentiation and multiplication by a smooth function
77
We will proceed through the following Greenās formula:
āu āv āv dS, (uĪv ā vĪu)dx = (4.32) u ān ĀÆ ān ĀÆ Ī© āĪ© where the notation and conditions will be explained now. Notation and assumptions ā¢ ā¢ ā¢
Ī© is a bounded domain (bounded open subset) in Rn . ĀÆ such that Ī© ā Ī© ĀÆ ā Rn . The closure of Ī© is the smallest closed set Ī© ĀÆ \ Ī©. The boundary āĪ© of Ī© in Rn is Ī©
It seems natural that āĪ© should be called the boundary of Ī©. However, ĀÆ and āĪ©. To this would allow too much freedom in the global structure of Ī© avoid these problems we will later introduce some additional āsmoothnessā restrictions. ĀÆ = Ī© āŖ āĪ©, but it is possible ĀÆ of Ī© in Rn we have Ī© ā¢ For the closure Ī© that Ī© ā© āĪ© = ā
. (For example, consider a nonempty closed straight line interval J ā Ī© and then replace Ī© by Ī© \ J.) ā¢
dS is the area element on the boundary.
ā¢ Let k be ļ¬xed, k ā {1, 2, . . . }, or k = ā. (Sometimes it is also convenient to use k = 0, which means that only continuity of the integrand is required.) The space C k (Ī©) is deļ¬ned as the space of all functions u ā C(Ī©) such that any derivative ā Ī± u with |Ī±| ā¤ k exists and is continuous in Ī©. The space C ā (Ī©) can be deļ¬ned as the intersection of all spaces C k (Ī©). This intersection is a FrĀ“echet space. ā¢ n ĀÆ = n ĀÆ (x) is an outgoing unit vector ļ¬eld which is based at every point x ā āĪ©, tangent to Rn and orthogonal to āĪ© at all points x ā āĪ©. ā¢ Let us also assume that n ĀÆ (x) is continuous with respect to x. It is easy to see from the implicit function theorem that locally the problem has at most one solution. This means that the solution exists and is unique in a small neighborhood of any outgoing unit vector ļ¬eld which is orthogonal to āĪ© and Rn (at the points x where a local continuous outgoing normal vector exists). ā¢ Now we will formulate conditions which guarantee that the closure ĀÆ of Ī© in Rn and the boundary āĪ© ā Rn are suļ¬ciently regular at a point Ī© ĀÆ is a ļ¬nite union of simplest cells such as x ĀÆ. For simplicity it suļ¬ces that Ī© formulated above. Deļ¬nition 4.8 (Smoothness of the boundary). Assume that U ā Rn is open and bounded in Rn , k ā {1, 2, . . . }. We will say that the boundary ĀÆ ā āĪ©, there exist r > 0 and a āĪ© is C k (or āĪ© ā C ā ) if for every x nā1 k ā R such that after relabeling and reorienting the C -function h : R
78
4. Distributions
coordinates axes, if necessary, we have (4.33)
U ā© B(ĀÆ x, r) = {x ā B(ĀÆ x, r) | xn > h(x1 , . . . , xnā1 )}. The outward unit normal vector ļ¬eld
Deļ¬nition 4.9. (i) If āU is C 1 , then along āU we can deļ¬ne the outward unit normal vector ļ¬eld Ī½ = (Ī½1 , . . . , Ī½n ). This ļ¬eld is normal at any point x ĀÆ ā āU if Ī½(ĀÆ x) = Ī½ = (Ī½1 , . . . Ī½n ). (ii) Let u ā C 1 (āĪ©). We deļ¬ne āu = Ī½ Ā· D u, āĪ½ which is called the (outward ) normal derivative of u. (Here D = ā/āx.) We will sometimes need to change coordinates near a point x ĀÆ ā āĪ© so as to āļ¬atten outā the boundary. To be more speciļ¬c, ļ¬x x ĀÆ ā āĪ©, and choose r, h, etc., as above. Then deļ¬ne Ī¦1 (x), . . . , Ī¦n (x) by the formulas (i = 1, . . . , n ā 1) yi = xi = Ī¦i (x) n yn = xn ā h(x1 , . . . , xnā1 ) = Ī¦ (x), and write y = Ī¦(x). ĪØ1 (x), . . . , ĪØn (x)
by the formulas Similarly, we deļ¬ne (i = 1, . . . , n ā 1) yi = xi = ĪØi (x) n yn = xn + h(x1 , . . . , xnā1 ) = ĪØ (x) and write y = ĪØ(x). ĪØā1 ,
Then Ī¦ = and the mapping x ā Ī¦(x) = y āļ¬attens out āĪ©ā near x ĀÆ. Observe also that det Ī¦ = det ĪØ = 1. Let us discuss what kind of regularity we need to require for the boundary to deserve to be called āsuļ¬ciently regularā. Unfortunately, there is no easy answer to this question: the answer is dictated by the goal. More speciļ¬cally, suppose that we need to make sense of the formula (4.32), so as to give suļ¬cient conditions for it to hold. Obviously, it is reasonable to request that we have a continuous unit normal vector and the area element (at least almost everywhere). To make sense of the left-hand side, it is natĀÆ and āĪ© ā C 1 , which will guarantee that ural to require that u, v ā C 2 (Ī©) ĀÆ and the integrals on the both sides make sense. In fact, if u, v ā C 2 (Ī©) 1 āĪ© ā C , then Greenās formula (4.32) can indeed be veriļ¬ed.
4.5. Diļ¬erentiation and multiplication by a smooth function
79
Additionally, if we need a C 2 change of variables, which moves the boundary (say, if we would like to straighten the C 1 boundary locally), then the second-order derivatives of the equation of the boundary (i.e., of the function h in Deļ¬nition 4.8) will show up in the transformed operator Ī and will be out of control. Therefore, to guarantee the continuity of the coeļ¬cients of the transformed equation, it is natural to assume āĪ© ā C 2 . Other integral formulas Here we will establish more elementary integral formulas (Gauss, Green, Ostrogradsky, Stokes, . . . ). We will start with the deduction of Greenās formula (4.32), from simpler integral identities, because the intermediate formulas have at least an equal importance. The divergence formula, alternatively called the Gauss-Ostrogradsky forĀÆ mula, holds: If āĪ© ā C 1 and F is a C 1 vector ļ¬eld in a neighborhood of Ī© n in R , then
div F dx = FĀ·n ĀÆ dS, (4.34) Ī©
āĪ©
where div F = ā Ā· F =
n āFj j=1
āxj
.
Here F1 , . . . , Fn are the coordinates of the vector F. (For the proof see, for example, Rudin [27, Theorem 10.51].) For our purposes it suļ¬ces to have this formula for the bounded domains Ī© with smooth boundary āĪ© (say, class C k , with k ā„ 1), that is, the boundary which is a compact closed hypersurface (of class C k ) in Rn . More generally, we can allow the domains with a piecewise C k boundary. But the analysis of āpiecewise smoothā situations is very diļ¬cult compared with the āsmoothā case. So we will not discuss the piecewise smooth objects in this book. Let us return to the deduction of Greenās formula (4.32). Taking F = ĀÆ and v ā C 2 (Ī©), ĀÆ we have uāv, we ļ¬nd that for u ā C 1 (Ī©) div(uāv) = uĪv + āu Ā· āv; hence (4.34) takes the form
(uĪv + āu Ā· āv)dx = (4.35) Ī©
u āĪ©
āv dS. ān ĀÆ
ĀÆ and subtracting from this formula the one Assuming now that u, v ā C 2 (Ī©) obtained by transposition u and v, we obtain (4.32).
80
4. Distributions
The Greenās formula (4.32) is also a particular case of the Stokes formula for general diļ¬erential forms (see, e.g., [27, Theorem 10.33]):
(4.36) dĻ = Ļ, Ī©
āĪ© 1 C , and
dĻ is its exterior diļ¬erential. where Ļ is an (n ā 1)-form on Ī©, Ļ ā Take the diļ¬erential form n āu āv jā1 j ā§ . . . ā§ dxn , (ā1) āv Ļ= u dx1 ā§ . . . ā§ dx āxj āxj j=1
j means that dxj is omitted. Clearly, where dx dĻ = (uĪv ā vĪu)dx1 ā§ . . . ā§ dxn , so the left-hand side of (4.36) coincides with the left-hand side of (4.32). The right-hand side of (4.36) in this case may be easily reduced to the righthand side of (4.32) and we leave the corresponding details to the reader as an exercise. Let us return to the calculation of Īun in D (Rn ). For any Ļ ā D(Rn ),
un (x)ĪĻ(x)dx = lim un (x)ĪĻ(x)dx. Īun , Ļ = un , ĪĻ = Īµā0+ |x|ā„Īµ
Rn
We now use Greenās formula (4.32) with u = un , v = Ļ, and Ī© = {x : Īµ ā¤ |x| ā¤ R}, where R is so large that Ļ(x) = 0 for |x| ā„ R ā 1. Since Īun (x) = 0 for x = 0, we get
āun āĻ un āĻ dS. un (x)ĪĻ(x)dx = ān ĀÆ ān ĀÆ |x|ā„Īµ |x|=Īµ Clearly, āĻ āĻ =ā , ān ĀÆ ār implying
(4.37)
āun āun =ā = ār1ān , ān ĀÆ ār
|x|ā„Īµ
un (x)ĪĻ(x)dx = āun (Īµ)
|x|=Īµ
āĻ dS + Īµ1ān ār
Ļ(x)dS, |x|=Īµ
1 where un (Īµ) = 2ān Īµ2ān for n ā„ 3 and un (Īµ) = ln Īµ for n = 2. Since the area of a sphere of radius Īµ in Rn is equal to Ļnā1 Īµnā1 (here Ļnā1 is the area of a sphere of radius 1), the ļ¬rst summand in the right-hand side of (4.37) tends to 0 as Īµ ā 0+, and the second summand tends to Ļnā1 Ļ(0). Finally, we obtain Īun , Ļ = Ļnā1 Ļ(0),
or Īun = Ļnā1 Ī“(x).
4.5. Diļ¬erentiation and multiplication by a smooth function
81
Now, setting En (x) =
(4.38)
1 r2ān (2 ā n)Ļnā1 E2 (x) =
(4.39)
for n ā„ 3,
1 ln r, 2Ļ
we clearly get ĪEn (x) = Ī“(x); i.e., En (x) is a fundamental solution of the Laplace operator in Rn . It will later be referred to as the fundamental solution of the Laplace operator. It will be clear later that for n ā„ 3 it is the only fundamental solution such that it is a function of r = |x| and, besides, tends to 0 as |x| ā ā. For n = 2 there is no obvious way to choose the additive constant. Remark 4.10. The area Ļnā1 of the unit sphere in Rn . Let us compute Ļnā1 . To do this consider Gaussian integrals
ā 2 eāx dx, I1 =
In =
eā|x| dx = 2
āā
Rn
eā(x1 +Ā·Ā·Ā·+xn ) dx1 . . . dxn = I1n . 2
Rn
2
Expressing In in polar coordinates, we get
ā 2 In = eār Ļnā1 rnā1 dr. 0
r2
= t, we get
ā ā 1 ā n ā1 āt Ļnā1 n āt nā1 Ī , e t 2 d( t) = Ļnā1 Ā· t 2 e dt = In = Ļnā1 2 0 2 2 0
Setting
where Ī is the Euler Gamma-function. This implies Ļnā1 = same time, for n = 2, we have
ā 2 2 eār rdr = āĻeār |+ā = Ļ; I2 = 2Ļ 0 therefore I1 = (4.40)
ā
2In Ī( n ). 2
At the
0
Ļ and In = Ļ n/2 . Thus, Ļnā1 =
2Ļ n/2 . Ī( n2 )
Note that this formula can also be rewritten without Ī-function, since the numbers Ī( n2 ) are easy to compute. Indeed, the functional equation Ī(s + 1) = sĪ(s) enables us to express Ī( n2 ) in terms of Ī(1) for even n and in terms of Ī( 12 ) for odd n. But for n = 1 we deduce from (4.40) that ā Ī( 12 ) = Ļ. (Clearly, Ļ0 = 2. However, we could also use the formula Ļ2 = 4Ļ taking n = 3 and using the relation Ī( 32 ) = 12 Ī( 12 ).)
82
4. Distributions
For n = 2, we deduce from (4.40) that Ī(1) = 1, which is also easy to verify without this formula. Therefore, for n = 2k + 2 we get Ļ2k+1 =
2Ļ k+1 , k!
and for n = 2k + 1 we get Ļ2k =
2 Ā· (2Ļ)k , where (2k ā 1)!! = 1 Ā· 3 Ā· 5 Ā· Ā· Ā· Ā· Ā· (2k ā 1). (2k ā 1)!!
Remark 4.11. A physical meaning of the fundamental solution of the Laplace operator. For an appropriate choice of units, the potential u(x) of the electrostatic ļ¬eld of a system of charges distributed with density Ļ(x) in R3 satisļ¬es the Poisson equation (4.41)
Īu = Ļ.
In particular, for a point charge at zero we have Ļ(x) = Ī“(x); hence, u(x) is a fundamental solution for Ī. Only potentials decaying at inļ¬nity have physical meaning. In what follows, we will prove Liouvilleās theorem, which implies in particular that a solution u(x) of the Laplace equation Īu = 0 deļ¬ned on Rn tending to 0 as |x| ā +ā is identically zero. Therefore, there is a unique fundamental solution tending to 0 as |x| ā +ā; namely 1 . It deļ¬nes the potential of the point unit charge situated at 0. E3 (x) = ā 4Ļr Besides, the potential of an arbitrary distribution of charges should, clearly, by the superposition principle, be determined by the formula
(4.42) u(x) = E3 (x ā y)Ļ(y)dy. Applying formally the Laplace operator, we get
Īu(x) = Ī“(x ā y)Ļ(y)dy = Ļ(x); i.e., Poissonās equation (4.41) holds. These arguments may serve to deduce Poissonās equation if, from the very beginning, the potential is deļ¬ned by (4.42). Its justiļ¬cation may be obtained with the help of the convolution of distributions introduced below. Let us indicate a meaning of the fundamental solution E2 (x) in R2 . In consider an inļ¬nite uniformly charged string (with the constant linear density of the charge equal to 1) placed along x3 -axis. From symmetry considerations, it is clear that its potential u(x) does not depend on x3 and only depends on r = x21 + x22 . Let x = (x1 , x2 ). Poissonās equation for u 1 ln r + C. Therefore, in this takes the form Īu(x) = Ī“(x) implying u(x) = 2Ļ case the potential is of the form E2 (x1 , x2 ) + C. R3
Note, however, that the potential of the string is impossible to compute with the help of (4.42) which, in this case, is identically equal to +ā. What
4.5. Diļ¬erentiation and multiplication by a smooth function
83
is the meaning of this potential? This is easy to understand if we recall that the observable physical quantity is not the potential but the electrostatic ļ¬eld E = āgrad u. This ļ¬eld can be computed via the Coulomb inverse square law, and the integral that deļ¬nes it converges (at the points outside the string). The potential u, which can be recovered from E up to an additive constant, equals E2 (x1 , x2 ) + C. This potential can be deļ¬ned by another method when we ļ¬rst assume that the string is of ļ¬nite length 2l and subtract from the potential of the ļ¬nite string a constant depending on l (this does not aļ¬ect E!) and then pass to the limit as l ā +ā. It is easy to see that it is possible to choose the constants in such a way that the limit of the potentials of the ļ¬nite strings really exists. Then this limit must be equal to E2 (x, y) + C by the above reasoning. We actually subtract an inļ¬nite constant from the potential of the string of the form (4.42), but this does not aļ¬ect E. This procedure is called in physics renormalization of charge and has its analogues in quantum electrodynamics. Let us prove the possibility of renormalizing the charge. Let us write the potential of a segment of the string x3 ā [āl, l]:
l 1 dt . ul (x1 , x2 , x3 ) = ā 2 2 4Ļ āl x1 + x2 + (t ā x3 )2 Since x21 + x22 + (t ā x3 )2 ā¼ |t| as t ā ā, it is natural to consider instead of ul the function
l 1 1 1 āā vl (x1 , x2 , x3 ) = ā dt 4Ļ āl 1 + t2 x21 + x22 + (t ā x3 )2 that diļ¬ers from ul by a constant depending on l:
l 1 dt ā . vl = ul + 4Ļ āl 1 + t2 The integrand in the formula for vl is
ā 1 + t2 ā x21 + x22 + (t ā x3 )2 1 ā āā = 2 1 + t2 x21 + x22 + (t ā x3 )2 x1 + x22 + (t ā x3 )2 Ā· 1 + t2 1
1 + 2tx3 ā x21 ā x22 ā x23
, ā ā x21 + x22 + (t ā x3 )2 Ā· 1 + t2 x21 + x22 + (t ā x3 )2 + 1 + t2 which is asymptotically O t12 as t ā +ā, so that there exists a limit of the integral as l ā +ā. Clearly, the integral itself and its limit may =
84
4. Distributions
be computed explicitly but they are of no importance to us, since we can compute the limit up to a constant by another method. Multiplication by a smooth function This operation is introduced similarly to diļ¬erentiation. Namely, if f ā a ā C ā (Ī©), Ļ ā D(Ī©), then, clearly,
L1loc (Ī©),
af, Ļ = f, aĻ.
(4.43)
This formula may serve as a deļ¬nition of af when f ā D (Ī©), a ā C ā (Ī©). It is easy to see that then we again get a distribution af ā D (Ī©). If f ā E (Ī©), then af ā E (Ī©) and supp(af ) ā supp f . Now let f ā S (Rn ). Then we can use (4.43) for Ļ ā S(Rn ) only if aĻ ā S(Rn ). Moreover, if we want af ā S (Rn ), it is necessary that the operator mapping Ļ to aĻ be a continuous operator from S(Rn ) to S(Rn ). To this end, it suļ¬ces, e.g., that a ā C ā (Rn ) satisļ¬es |ā Ī± a(x)| ā¤ CĪ± (1 + |x|)mĪ± ,
x ā Rn ,
for every multi-index Ī± with some constants CĪ± and mĪ± . In particular, the multiplication by a polynomial maps S (Rn ) into S (Rn ). A function f ā L1loc (Ī©) can be multiplied by any continuous function a(x). For a distribution f , conditions on a(x) justifying the existence of af can also be weakened. Namely, let, e.g., f ā D (Ī©) be such that for some integer m ā„ 0 and every compact K ā Ī© we have (4.44)
|f, Ļ| ā¤ CK
|Ī±|ā¤m
sup |ā Ī± Ļ(x)|, Ļ ā D(K).
xāK
Then we say that f is a distribution of ļ¬nite order (not exceeding m). Denote by Dm (Ī©) the set of such distributions. In the general case the estimate (4.44) only holds for a constant m depending on K, and now we require m to be independent of K. Clearly, if f ā E (Ī©), then the order of f is ļ¬nite. If f ā L1loc (Ī©), then f ā D0 (Ī©). (Ī©), then we may extend f by continuity to a linear functional If f ā Dm on the space Dm (Ī©)consisting of C m -functions with compact support in Ī©. Clearly, Dm (Ī©) = K Dm (K), where K is a compact in Ī© and Dm (K) is the subset of Dm (Ī©) consisting of functions with the support belonging to K. Introducing in Dm (K) the norm equal to the right-hand side of (4.44) without CK , we see that Dm (K) becomes a Banach space and f ā Dm (Ī©) can be extended to a continuous linear functional on Dm (K). Therefore, it deļ¬nes a linear functional on Dm (Ī©). It is continuous in the sense that its restriction on Dm (K) is continuous for any compact K ā Ī©.
4.5. Diļ¬erentiation and multiplication by a smooth function
85
For instance, it is obvious that Ī“(x ā x0 ) ā D0 (Ī©) for x0 ā Ī©, so that Ī“function deļ¬nes a linear functional on D0 (Ī©). It actually deļ¬nes a continuous linear functional on C(Ī©).
In general, if f ā Dm (Ī©) and supp f ā K, then f can be extended to a continuous linear functional on C m (Ī©) if the topology in C m (Ī©) is deļ¬ned by seminorms given by the right-hand side of (4.44) with any compact K ā Ī©.
Now, let f ā Dm (Ī©) and a ā C m (Ī©). Then (4.43) makes sense for Ļ ā D(Ī©), since then aĻ ā Dm (Ī©). Therefore, af is deļ¬ned for f ā Dm (Ī©) and a ā C m (Ī©). Example 4.11. a(x)Ī“(x) = a(0)Ī“(x) for a ā C(Rn ). Example 4.12. Let n = 1. Let us compute a(x)Ī“ (x) for a ā C ā (R1 ). We have aĪ“ , Ļ = Ī“ , aĻ = āĪ“, (aĻ) = āa (0)Ļ(0) ā a(0)Ļ (0) implying
a(x)Ī“ (x) = a(0)Ī“ (x) ā a (0)Ī“(x).
Note, in particular, that a(0) = 0 does not necessarily imply a(x)Ī“ (x) = 0. Example 4.13. Leibniz rule. Let f ā D (Ī©), a ā C ā (Ī©). Let us prove that āa āf ā (af ) = Ā·f +a . āxj āxj āxj
(4.45)
Indeed, if Ļ ā D(Ī©), then āĻ ā āĻ (af ), Ļ = ā af, = ā f, a āxj āxj āxj āa āf āa ā (aĻ) + f, Ļ = , aĻ + f, Ļ = ā f, āxj āxj āxj āxj āf āa = a + f, Ļ , āxj āxj as required. By (4.45) we may diļ¬erentiate the product of a distribution by a smooth function as the product of the regular functions. It follows that the Leibniz rule for higher-order derivatives also holds. Example 4.14. A fundamental solution for an ordinary (one-dimensional) diļ¬erential operator. On R1 consider a diļ¬erential operator dmā1 d dm + a0 , + a + Ā· Ā· Ā· + a1 mā1 m mā1 dx dx dx where the aj are constants. Let us ļ¬nd its fundamental solution E(x), i.e., a solution of Lu = Ī“(x). Clearly, E(x) is determined up to a solution of the L=
86
4. Distributions
homogeneous equation Lu = 0. Besides, LE(x) = 0 should hold for x = 0. Therefore, it is natural to seek E(x) in the form y1 (x) for x < 0, E(x) = y2 (x) for x > 0, where y1 , y2 are solutions of Ly = 0. Moreover, subtracting y1 (x), we may assume that (4.46)
E(x) = Īø(x)y(x),
where y(x) is a solution of Ly = 0. Now, let us ļ¬nd LE(x). Clearly, (Īø(x)y(x)) = y(0)Ī“(x) + Īø(x)y (x), and the higher-order diļ¬erentiations give rise to derivatives of Ī“(x). If we wish these derivatives to vanish and the Ī“-function to appear only at the very last step, we should assume y(0) = y (0) = Ā· Ā· Ā· = y (mā2) (0) = 0, y (mā1) (0) = 1. Such a solution y(x) exists and is unique. For such a choice we get (Īø(x)y(x)) = Īø(x)y (x), (Īø(x)y(x)) = (Īø(x)y (x)) = Īø(x)y (x), Ā·Ā·Ā· (Īø(x)y(x))(mā1) = Īø(x)y (mā1) (x), (Īø(x)y(x))(m) = y (mā1) (0)Ī“(x) + Īø(x)y (m) (x) = Ī“(x) + Īø(x)y (m) (x). Therefore, L(Īø(x)y(x)) = Īø(x)Ly(x) + Ī“(x) = Ī“(x), as required. Why should the fundamental solution be a regular C ā -function for x = 0? This follows from Theorem 4.12. Let u ā D ((a, b)) be a solution of the diļ¬erential equation (4.47)
u(m) + amā1 (x)u(mā1) + Ā· Ā· Ā· + a0 (x)u = f (x),
where aj (x), f (x) are C ā -functions on (a, b). Then u ā C ā ((a, b)). Proof. Subtracting a smooth particular solution of (4.47) which exists, as is well known, we see that everything is reduced to the case when f ā” 0. Further, for f ā” 0, equation (4.47) is easily reduced to the system of the form (4.48)
v = A(x)v,
4.6. Transposed operator. Change of variables
87
where v is a vector-valued distribution (i.e., a vector whose components are distributions) and A(x) a matrix with entries from C ā ((a, b)). Let ĪØ(x) be an invertible matrix of class C ā satisfying ĪØ (x) = A(x)ĪØ(x); i.e., ĪØ(x) is a matrix whose columns constitute a fundamental system of solutions of (4.48). Set v = ĪØ(x)w; i.e., denote w = ĪØā1 (x)v. Then w is a vector-valued distribution on (a, b). Substituting v = ĪØ(x)w in (4.48), we get w = 0. Now, it remains to prove Lemma 4.13. Let u ā D ((a, b)) and u = 0. Then u = const. Proof. The condition u = 0 means that u, Ļ = 0 for any Ļ ā D((a, b)). But, clearly, Ļ ā D((a, b)) can be represented as Ļ = Ļ , where Ļ ā D((a, b)), if and only if
b Ļ(x)dx = 0. (4.49) 1, Ļ = a
Ļ
then (4.49) holds due to the main theorem of calculus. Indeed, if Ļ = x Conversely, if (4.49) holds, then we may set Ļ(x) = a Ļ(t)dt and, clearly, Ļ ā D((a, b)) and Ļ = Ļ. Consider the map I : D((a, b)) āā C sending Ļ to 1, Ļ. Since u, Ļ = 0 for Ļ ā Ker I, it is clear that u, Ļ only depends on IĻ, but since this dependence is linear, then u, Ļ = C1, Ļ, where C is a constant. But this means that u = C.
4.6. A general notion of the transposed (adjoint) operator. Change of variables. Homogeneous distributions The diļ¬erentiation and multiplication by a smooth function are particular cases of the following general construction. Let Ī©1 , Ī©2 be two domains in Rn and let L : C ā (Ī©2 ) āā C ā (Ī©1 ) be a linear operator. Let t L : D(Ī©1 ) āā D(Ī©2 ) be a linear operator such that (4.50)
Lf, Ļ = f, t LĻ, Ļ ā D(Ī©1 ),
for f ā C ā (Ī©2 ). The operator t L is called the transposed or adjoint or dual of L. Suppose that, for any compact K1 ā Ī©1 , there exists a compact K2 ā Ī©2 such that t LD(K1 ) ā D(K2 ) and t L : D(K1 ) āā D(K2 ) is a continuous map. Then formula (4.50) enables us to extend L to a linear operator L : D (Ī©2 ) āā D (Ī©1 ). If Ī©1 = Ī©2 = Rn and the operator t L can be extended to a continuous linear operator t L : S(Rn ) āā S(Rn ), then it is clear that L determines a linear map L : S (Rn ) āā S (Rn ).
88
4. Distributions
Note that L thus constructed is always weakly continuous. This is obvious from (4.50), since |Lf, Ļ| is a seminorm of the general form for Lf and |f,t LĻ| is a seminorm for f . Examples of transposed operators. a) If L =
ā āxj ,
then t L = ā āxā j .
b) If L = a(x) (the left multiplication by a(x) ā C ā (Ī©)), then t L = L = a(x). In particular, the operators
ā āxj
and a(x) are continuous on D (Ī©) with
respect to the weak topology. Moreover,
ā āxj
is continuous on S (Rn ).
c) A diļ¬eomorphism Ļ : Ī©1 āā Ī©2 naturally deļ¬nes a linear map Ļā : āā C ā (Ī©1 ) by the formula
C ā (Ī©2 )
(Ļā f )(x) = f (Ļ(x)). Clearly, the same formula deļ¬nes a linear map Ļā : L1loc (Ī©2 ) āā L1loc (Ī©1 ). Besides, Ļā transforms D(Ī©2 ) to D(Ī©1 ). We wish to extend Ļā to a continuous linear operator Ļā : D (Ī©2 ) āā D (Ī©1 ). To do so, we ļ¬nd the transposed operator. Let Ļ ā D(Ī©1 ). Then ā1
āĻ (z) ā ā1 dz Ļ f, Ļ = f (Ļ(x))Ļ(x)dx = f (z)Ļ(Ļ (z)) āz Ī©1 Ī©2 ā1 āĻ (z) Ā· Ļ(Ļā1 (z)) , = f (z), āz āĻā1 (z) is its Jacobian. āz It is clear now that the transposed operator is of the form ā1 āĻ (z) t ā Ā· Ļ(Ļā1 (z)). ( Ļ Ļ)(z) = āz
where Ļā1 is the inverse of Ļ and
ā1
Since | āĻ āz(z) | ā C ā (Ī©2 ), it is clear that t Ļā maps D(Ī©1 ) to D(Ī©2 ) and deļ¬nes a continuous linear map of D(K1 ) to D(Ļ(K1 )). Therefore, the general scheme determines a continuous linear map Ļā : D (Ī©2 ) āā D (Ī©1 ). If Ī©1 = Ī©2 = Rn and Ļ is an aļ¬ne transformation (or at least coincides with an aļ¬ne transformation outside a compact set), then t Ļā continuously maps ā1 S(Rn ) to S(Rn ), and in this case | āĻ āz (z) | = 0 everywhere (it is constant everywhere if Ļ is an aļ¬ne map, and it is a constant near inļ¬nity in the general case). So a continuous linear map Ļā : S (Rn ) āā S (Rn ) is welldeļ¬ned.
4.6. Transposed operator. Change of variables
89
Examples. a) Translation operator f (x) ā f (x ā t) with t ā Rn can be extended to a continuous linear operator from D (Ī©) to D (t + Ī©) and from S (Rn ) to S (Rn ). In particular, the notation Ī“(x ā x0 ) used above agrees with the deļ¬nition of the translation by x0 . b) The homothety f (x) ā f (tx), where t ā R\0, is deļ¬ned on D (Rn ) and maps D (Rn ) to D (Rn ) and S (Rn ) to S (Rn ). For instance, let us calculate Ī“(tx). We have
z |t|ān dz = |t|ān Ļ(0) = |t|ān Ī“(z)Ļ(z)dz. Ī“(tx)Ļ(x)dx = Ī“(z)Ļ t Therefore, Ī“(tx) = |t|ān Ī“(x), t ā R\0. Deļ¬nition 4.14. A distribution f ā S (Rn ) is called homogeneous of order m ā R if (4.51)
f (tx) = tm f (x),
t > 0.
It is easily checked that the requirement f ā S (Rn ) can actually be replaced by f ā D (Rn ) (then (4.51) implies f ā S (Rn )). Sometimes, one says āhomogeneous of degree mā instead of āhomogeneous of order mā. Examples. a) If f ā L1loc (Rn ) and for any t > 0 (4.51) holds almost everywhere, then f is a homogeneous distribution of order m. b) Ī“-function Ī“(x) is homogeneous of order ān.
āf It is easy to verify that āxā j f (tx) = t āx (tx). Therefore, if f is j āf is homogeneous of order m ā 1. For homogeneous of order m, then āx j Ī± instance, ā Ī“(x) is homogeneous of order ān ā |Ī±|.
Homogeneity considerations allow us to examine without computing the structure of the fundamental solution for the Laplace operator and its powers. Namely, the order of homogeneity of the distribution r2ān for n ā„ 3 is 2 ā n, and for r = 0 this distribution satisļ¬es the equation Īu = 0. Therefore, Īr2ān is supported at 0 and its order of homogeneity is ān. But, since the order of homogeneity of ā Ī± Ī“(x) is ān ā |Ī±|, it is clear that Īr2ān = CĪ“(x). For n = 2, we have Ī(ln r) = 0 for r = 0, and āxā j (ln r) is homogeneous of order ā1. Therefore, Ī(ln r) is supported at 0 and its order of homogeneity is ā2, so Ī(ln r) = CĪ“(x).
90
4. Distributions
It is not clear a priori why the following does not hold: Īr2ān = 0 in D (Rn ). Later on we will see, however, that if u ā D (Ī©) and Īu = 0, then u ā C ā (Ī©) (and, moreover, u is real analytic in Ī©). Since r2ān has a singularity at 0, it follows that Īr2ān = 0 implying Īr2ān = CĪ“(x), where C = 0. Similarly, for n = 2 we have Ī(ln r) = CĪ“(x) for C = 0. Finally, note that a rotation operator is deļ¬ned in D (Rn ) and in S (Rn ). This enables us to give a precise meaning to our arguments on averaging on which we based our conclusion about the existence of a spherically symmetric fundamental solution for Ī. Example 4.15. Fundamental solutions for powers of the Laplace operator. The spherical symmetry and homogeneity considerations make it clear that in order to ļ¬nd the fundamental solution of Īm in Rn , it is natural to consider the function r2mān . Then we get Īmā1 r2mān = C1 r2ān . Can it be C1 = 0? It turns out that if 2m ā n ā 2Z+ (i.e., 2m ā n is not an even nonnegative integer), or, equivalently, r2mān ā C ā (Rn ), then C1 = 0 and, therefore, Īm r2mān = CĪ“(x), where C = 0. Indeed, for r = 0 we have, for any Ī± ā R, 2 d nā1 d Ī± rĪ± = [Ī±(Ī±ā1)+Ī±(nā1)]rĪ±ā2 = Ī±(Ī±+nā2)rĪ±ā2 . + Īr = dr2 r dr Therefore, Īr2kān = (2k ā n)(2k ā 2)r2kānā2 and consecutive applications of Ī to r2mān yield a factor of the form (2m ā 2)(2m ā n), then (2m ā 4)(2m ā n ā 2), etc., implying C1 = 0 for 2m ā n ā 2Z+ . Therefore, for 2m ā n ā 2Z+ the fundamental solution Enm (x) of Īm is of the form Enm (x) = Cm,n r2mān . Further, let 2mān ā 2Z+ so that r2mān is a polynomial in x. Then consider the function r2mān ln r. We get C1 ln r for n = 2, Īmā1 (r2mān ln r) = 2ān for n ā„ 4 C2 r (C2 depends on n), where C1 = 0 and C2 = 0. Consider for the sake of deļ¬niteness the case n ā„ 4. Since mā1 2 d nā1 d mā1 f (r) = + f (r) Ī dr2 r dr and
d dr
ln r = 1r , it follows that Īmā1 (r2mān ln r) = Īmā1 (r2mān ) Ā· ln r + C2 r2ān = C2 r2ān
4.7. Appendix. Minkowski inequality
91
because r2mān is a polynomial in x of degree 2m ā n. Here C2 = 0 since r2mān ln r ā C ā (Rn ) and the equation Īu = 0 has no solutions with singularities. (Indeed, if it had turned out that C2 = 0, then for some k we would have had that u = Īkā1 (r2mān ln r) had a singularity at 0 and satisļ¬ed Īu = 0). Note though that C2 may be determined directly. Therefore, for 2m ā n ā 2Z+ , Enm (x) = Cm,n r2mān ln r.
4.7. Appendix. Minkowski inequality Let (M, Ī¼) be a space with a positive measure. Here M is a set where a Ļ-algebra of subsets is chosen. (We will call a subset S ā M measurable if it belongs to the Ļ-algebra.) The measure Ī¼ is assumed to be deļ¬ned on the Ļ-algebra and take values in [0, +ā]. We will also assume that Ī¼ is Ļ-ļ¬nite; i.e., there exist measurable sets M1 ā M2 ā Ā· Ā· Ā· , such that Ī¼(Mj ) is ļ¬nite for every j and M is the union of all Mj ās. Then the function spaces Lp (M, Ī¼), 1 ā¤ p < +ā, are well-deļ¬ned, with the norm
1/p p |f (x)| Ī¼(dx) , 1 ā¤ p < +ā. f p = M
The Minkowski inequality in its simplest form is the triangle inequality for the norm Ā· p : f + gp ā¤ f p + gp ,
f, g ā Lp (M, Ī¼).
By induction, this easily extends to several functions: f1 + f2 + Ā· Ā· Ā· + fk p ā¤ f1 p + f2 p + Ā· Ā· Ā· + fk p , where fj ā Lp (M, Ī¼), j = 1, 2, . . . , k. Multiplying each fj by aj > 0, we obtain (4.52) a1 f1 + a2 f2 + Ā· Ā· Ā· + ak fk p ā¤ a1 f1 p + a2 f2 p + Ā· Ā· Ā· + ak fk p . A natural generalization of this inequality can be obtained if we replace summation by integration over a positive measure. More precisely, let (N, Ī½) be another space with a positive measure Ī½, and let f = f (x, y) be a measurable function on (M Ć N, Ī¼ Ć Ī½), where Ī¼ Ć Ī½ is the product measure deļ¬ned by the property (Ī¼ Ć Ī½)(A Ć B) = Ī¼(A)Ī½(B) for any measurable sets A ā M, B ā N with ļ¬nite Ī¼(A) and Ī½(B). Then we have Theorem 4.15 (Minkowskiās inequality). Under the conditions above
f (Ā·, y)p Ī½(dy), (4.53) f (Ā·, y)Ī½(dy) ā¤ N
p
N
92
4. Distributions
or, equivalently, (4.54) p
1/p
f (x, y)Ī½(dy) Ī¼(dx) ā¤ M
N
N
1/p |f (x, y)| Ī¼(dx) p
Ī½(dy).
M
The inequality (4.52) is a particular case of (4.53). To see this it suļ¬ces to choose N to be a ļ¬nite set of k points and take measures of these points to be a1 , . . . , ak . Vice versa, if we know that (4.52) holds, then (4.53) becomes heuristically obvious. For example, if we could present the integrals in (4.53) as limits of the corresponding Riemann sums, then (4.53) would follow from (4.52). But this is not so easy to do rigorously even if the functions are suļ¬ciently regular. The easiest way to give a rigorous proof is by a direct use of the following very important Theorem 4.16 (HĀØ olderās inequality). Let f ā Lp (M, Ī¼), g ā Lq (M, Ī¼), where 1 < p, q < +ā, and 1 1 + = 1. (4.55) p q Then f g ā L1 (M, Ī¼) and
ā¤ f p gq . f (x)g(x)Ī¼(dx) (4.56) M
Proof. Let us start with the following elementary inequality: (4.57)
ab ā¤
ap bq + , p q
where a ā„ 0, b ā„ 0, and p, q are as in (4.55). It is enough to consider the case a > 0, b > 0. Consider the curve t = spā1 , s ā„ 0, in the positive quadrant of the real plane R2s,t , and two disjoint bounded domains G1 and G2 , so that each of them is bounded by this curve together with the coordinate axes and straight line intervals as follows: the boundary of G1 consists of two straight line intervals {(s, 0) : 0 ā¤ s ā¤ a}, {(a, t) : 0 ā¤ t ā¤ apā1 } and the curvilinear part {(s, t) : t = spā1 , 0 ā¤ s ā¤ a}; the boundary of G2 also consists of two straight line intervals {(0, t) : 0 ā¤ t ā¤ b}, {(s, b) : 0 ā¤ s ā¤ b1/(pā1) } and the generally diļ¬erent curvilinear piece of the same curve, namely, {(s, t) : s = t1/(pā1) , 0 ā¤ t ā¤ b}. It is easy to see that the areas of G1 , G2 are ap /p and bq /q, respectively, and their union contains the rectangle [0, a] Ć [0, b] up to a set of measure 0. Therefore, (4.57) follows by comparison of the areas. To prove (4.56) note ļ¬rst that replacing f , g by |f |, |g|, we may assume that f ā„ 0 and g ā„ 0. Dividing both parts of (4.56) by f p gq , we
4.7. Appendix. Minkowski inequality
93
see that it suļ¬ces to prove it for normalized functions f, g, i.e., such that f p = gq = 1. To this end apply (4.57) with a = f (x), b = g(x) to get f (x)g(x) ā¤
f (x)p g(x)q + . p q
Integrating this over M with respect to Ī¼ and using (4.55), we come to the desired result. Proof of Theorem 4.15. To prove (4.54) let us make some simplifying assumptions. Without loss of generality we can assume that f is positive (otherwise, replace it by |f |, which does not change the right-hand side and can only increase the left-hand side). We can assume that both sides of (4.54) are ļ¬nite. If the right-hand side is inļ¬nite, then the inequality is obvious. If the left-hand side is inļ¬nite, we can appropriately truncate f (using Ļ-ļ¬niteness of the measures) and then deduce the proof in the general case in the limit from a monotone approximation. Introducing a function
f (x, y)Ī½(dy),
h(x) := N
which is measurable on M by Fubiniās theorem, we can rewrite the pth power of the left-hand side of (4.54) as
p f (x, y)Ī½(dy) h(x)pā1 Ī¼(dx) M h(x) Ī¼(dx) = M N
f (x, y)h(x)pā1 Ī¼(dx) Ī½(dy). = N
M
Now applying HĀØolderās inequality (4.56) to the internal integral in the righthand side and taking into account that q = p/(p ā 1), we obtain
1/p
(pā1)/p
f (x, y)h(x)pā1 Ī¼(dx) ā¤ f (x, y)p Ī¼(dx) h(x)p Ī¼(dx) . M
M
M
Using this to estimate the right-hand side of the previous identity, we obtain 1/p
(pā1)/p
p p p h(x) Ī¼(dx) ā¤ f (x, y) Ī¼(dx) Ī½(dy) h(x) Ī¼(dx) . M
N
M
M
Now dividing both sides by the last factor in the right-hand side, we obtain the desired inequality (4.54). This division is possible if we assume that the last factor does not vanish. But in the opposite case f = 0 almost everywhere and the statement is trivially true.
94
4. Distributions
4.8. Appendix. Completeness of distribution spaces Let E be a FrĀ“echet space with the topology given by seminorms {pj | j = 1, 2, . . . }, such that p1 (x) ā¤ p2 (x) ā¤ Ā· Ā· Ā·
for all x ā E.
E
Let denote the dual space of all linear continuous functionals f : E ā R. Let us say that a set F ā E is bounded if there exist an integer j ā„ 1 and a positive constant C such that (4.58)
|f, Ļ| ā¤ Cpj (Ļ) for all f ā F, Ļ ā E.
Note that for F consisting of just one element f , this is always true, being equivalent to the continuity of f ; see (4.2). So the requirement (4.58) means equicontinuity, or uniform continuity of all functionals f ā F . Let us also say that a family F ā E is weakly bounded if for every Ļ ā E there exists a positive CĻ such that (4.59)
|f, Ļ| ā¤ CĻ
for all f ā F.
The following theorem is a simplest version of the Banach-Steinhaus theorem, which is also called the uniform boundedness principle. Theorem 4.17. If, under the assumptions described above, F is weakly bounded, then it is bounded. The completeness of E plays an important role in the proof, which relies on the following theorem. Theorem 4.18 (Baireās theorem). A nonempty complete metric space cannot be presented as an at most countable union of nowhere dense sets. Here, nowhere dense set S in a metric space M is a set whose closure does not have interior points. So the theorem can be equivalently reformulated as follows: a nonempty complete metric space cannot be presented as an at most countable union of closed sets without interior points. Proof of Theorem 4.18. Let M be a nonempty complete metric space with the metric Ļ, and let {Sj | j = 1, 2, . . . } be a sequence of nowhere dense subsets of M . We will produce a point x ā M which does not belong to any Sj . Passing to the closures we can assume that Sj is closed for every j. Now, clearly S1 = M because otherwise S1 would be open in M , so it would ĀÆ 1 , r1 ) = ĀÆ1 = B(x not be nowhere dense. Hence there exists a closed ball B ĀÆ {x ā M | Ļ(x, x1 ) ā¤ r1 }, such that B1 ā© S1 = ā
. By induction we can ĀÆ j , rj ), j = 1, 2, . . . , such that ĀÆj = B(x construct a sequence of closed balls B ĀÆ ĀÆ ĀÆj ĀÆ Bj+1 ā Bj , rj+1 ā¤ rj /2, and Bj ā© Sj = ā
for all j. It follows that B
4.8. Appendix. Completeness of distribution spaces
95
does not intersect with S1 āŖ S2 āŖ Ā· Ā· Ā· āŖ Sj . On the other hand, the sequence of centers {xj | j = 1, 2, . . . } is a Cauchy sequence because by the triangle inequality, for m < n, Ļ(xm , xn ) ā¤ Ļ(xm , xm+1 ) + Ā· Ā· Ā· + Ļ(xnā1 , xn ) ā¤ rm + rm+1 + Ā· Ā· Ā· ā¤ 2rm ā¤ 2ām+2 r1 . Therefore, due to the completeness of M , xm ā x in M , and x belongs to ĀÆj . It follows that x ā Sj for all j. all balls B Proof of Theorem 4.17. Let us take for any j = 1, 2, . . . ! Ļ ā E sup |f, Ļ| ā¤ jpj (Ļ) Sj = =
"#
f āF
$ Ļ ā E |f, Ļ| ā¤ jpj (Ļ) .
f āF
Then Sj ā Sj+1 for all j, each set Sj is closed, and by the assumed weak boundedness of the family F ā E the union of all Sj , j = 1, 2, . . . , is the whole of E. Therefore, by Baireās theorem, there exists j such that Sj has a nonempty interior. Taking a fundamental system of neighborhoods of any point Ļ0 ā E, consisting of sets Bj (Ļ0 , Īµ) = {Ļ0 + Ļ| pj (Ļ) < Īµ}, where Īµ > 0, j = 1, 2, . . . , we see that we can ļ¬nd Ļ0 , j, and Īµ, such that Bj (Ļ0 , Īµ) ā Sj . Then for any f ā F and any Ļ ā Bj (0, Īµ) |f, Ļ| ā¤ |f, Ļ+Ļ0 |+|f, Ļ0 | ā¤ jpj (Ļ+Ļ0 )+C1 ā¤ jpj (Ļ)+C2 ā¤ jĪµ+C2 , where C1 , C2 > 0 are independent of Ļ and f (but may depend on Ļ0 , j, and Īµ). We proved this estimate for Ļ ā E such that pj (Ļ) < Īµ. By scaling we deduce that |f, Ļ| ā¤ (j + Īµā1 C2 )pj (Ļ) for all f ā F and Ļ ā E, which implies the desired result.
Corollary 4.19 (Completeness of the dual space). Let E be a FrĀ“echet space, and let {fk ā E | k = 1, 2, . . . } be a sequence of continuous linear functionals such that for every Ļ ā E there exists the limit f, Ļ := lim fk , Ļ. kā+ā
Then the limit functional f is also continuous; i.e., f ā E .
96
4. Distributions
Proof. Clearly, the set F = {fk | k = 1, 2, . . . } is weakly bounded. Therefore, by Theorem 4.17 there exists a seminorm pj , which is one of the seminorms deļ¬ning the topology in E, and a constant C > 0, such that |fk , Ļ| ā¤ Cpj (Ļ),
Ļ ā E, k = 1, 2, . . . .
Taking the limit as k ā +ā, we obtain |f, Ļ| ā¤ Cpj (Ļ),
Ļ ā E,
with the same C and j. This implies continuity of f .
Applying this corollary of Theorem 4.17 to the distribution spaces D (Ī©), and S (Rn ), we obtain their completeness. (In the case of D (Ī©) we should apply the corollary to E = D(K).)
E (Ī©),
4.9. Problems ā 4.1. Find lim ĻĪµ (x) in D (R), where ĻĪµ (x) = 1/ Īµ for x ā (āĪµ, Īµ) and Īµā0+
ĻĪµ (x) = 0 for x ā (āĪµ, Īµ). 4.2. Find lim ĻĪµ (x) in D (R), where ĻĪµ (x) = 1/Īµ for x ā (āĪµ, Īµ) and Īµā0+
ĻĪµ (x) = 0 for x ā (āĪµ, Īµ). 1 Īµā0+ x
4.3. Find lim
sin xĪµ in D (R).
4.4. Prove that if f ā C ā (R), then x1 (f (x) ā f (0)) ā C ā (R). 4.5. Consider the canonical projection p : R āā S 1 = R/2ĻZ. It induces ā (R), where C ā (R) is the set of all 2Ļan isomorphism pā : C ā (S 1 ) āā C2Ļ 2Ļ ā periodic C -functions on R. Introduce a natural FrĀ“echet topology in the space C ā (S 1 ) and deļ¬ne the set of distributions on S 1 as the dual space D (S 1 ) = (C ā (S 1 )) . Identify each function f ā L1 (S 1 ) (or, equivalently, a locally integrable 2Ļ-periodic function f on R) with a distribution on S 1 by the formula
2Ļ f (x)Ļ(x)dx. f, Ļ = 0
Prove that
pā
can be extended by continuity to an isomorphism (R), D (S 1 ) āā D2Ļ
(R) is the set of all 2Ļ-periodic distributions. where D2Ļ
4.9. Problems
97
4.6. Prove that the trigonometric series fk eikx kāZ
D (R)
converges in if and only if there exist constants C and N such that N |fk | ā¤ C(1 + |k|) . Prove that if f is the sum of this series, then f ā D2Ļ (R) and the series above is in fact its Fourier series; that is, 1 f, eāikx , fk = 2Ļ where the angle brackets denote the natural pairing between D (S 1 ) and C ā (S 1 ). 4.7. Prove that any 2Ļ-periodic distribution (or, equivalently, distribution on S 1 ) is equal to the sum of its Fourier series. 4.8. Find the Fourier series expansion of the Ī“-function on S 1 , or, equivalently, of the distribution ā Ī“(x + 2kĻ) ā D2Ļ (R). k=āā
4.9. Use the result of the above problem to prove the following Poisson summation formula: ā ā 1 Ė f (k), f (2kĻ) = 2Ļ k=āā
k=āā
where f ā S(R) and fĖ is the Fourier transform of f . 4.10. Find all u ā D (R), such that a) xu(x) = 1; b) xu (x) = 1; c) xu (x) = Ī“(x). eikr , where r = |x|, is the fundamental solution of the 4Ļr Helmholtz operator Ī + k 2 in R3 . 4.11. Verify that ā
10.1090/gsm/205/05
Chapter 5
Convolution and Fourier transform
5.1. Convolution and direct product of regular functions The convolution of locally integrable functions f and g on Rn is the integral
(5.1) (f ā g)(x) = f (x ā y)g(y)dy, over Rn . Obviously, the integral (5.1) is not always deļ¬ned. Let us consider it when f, g ā C(Rn ) and one of these functions has a compact support (in this case (5.1) makes sense and, clearly, f ā g ā C(Rn )). The change of variables x ā y = z turns (5.1) into
(f ā g)(x) = f (z)g(x ā z)dz; i.e., the convolution is commutative: f ā g = g ā f. Further, let f ā C 1 (Rn ). Then it is clear that (5.1) may be diļ¬erentiated under the integral sign, and we get āf ā (f ā g) = ā g. āxj āxj In general, if f ā C m (Rn ), then (5.2)
ā Ī± (f ā g) = (ā Ī± f ) ā g
for |Ī±| ā¤ m. 99
100
5. Convolution and Fourier transform
The convolution is associative; i.e., f ā (g ā h) = (f ā g) ā h if f, g, h ā C(Rn ) and two of the three functions have compact supports. This can be veriļ¬ed by the change of variables but a little later we will prove this in another way. Let Ļ ā C(Rn ) and let Ļ have a compact support. Then
f ā g, Ļ = f (x ā y)g(y)Ļ(x)dydx may be transformed by the change x = y + z to the form
(5.3) f ā g, Ļ = f (z)g(y)Ļ(z + y)dydz. By Fubiniās theorem, this makes obvious the already proved commutativity of the convolution. Further, this also implies
(f ā g) ā h, Ļ = f (x)g(y)h(z)Ļ(x + y + z)dxdydz = f ā (g ā h), Ļ, proving the associativity of the convolution. Note the important role played in (5.3) by the function f (x)g(y) ā C(R2n ) called the direct or tensor product of functions f and g and denoted by f ā g or just f (x)g(y) if we wish to indicate the arguments. Suppose that f and g have compact supports. Then in (5.3) we may set Ļ(x) = eāiĪ¾Ā·x , where Ī¾ ā Rn . We get f ā g(Ī¾) = fĖ(Ī¾) Ā· gĖ(Ī¾), where the tilde denotes the Fourier transform:
fĖ(Ī¾) = f (x)eāiĪ¾Ā·x dx. Therefore, the Fourier transform turns the convolution into the usual product of Fourier transforms. Let us analyze the support of a convolution. Let A, B ā Rn . Set A + B = {x + y : x ā A, y ā B} (sometimes A + B is called the arithmetic sum of subsets A and B). It turns out that supp(f ā g) ā supp f + supp g. % Indeed, it follows from (5.3) that if supp Ļ (supp f + supp g) = ā
, then f ā g, Ļ = 0, yielding (5.4). (5.4)
5.2. Direct product of distributions
101
5.2. Direct product of distributions Let Ī©1 ā Rn1 , Ī©2 ā Rn2 , f1 ā D (Ī©1 ), f2 ā D (Ī©2 ). Let us deļ¬ne the direct or tensor product f1 ā f2 ā D (Ī©1 Ć Ī©2 ) of distributions f1 and f2 . Instead of f1 ā f2 , we will often also write f1 (x)f2 (y), where x ā Ī©1 , y ā Ī©2 . Let Ļ(x, y) ā D(Ī©1 Ć Ī©2 ). For fj ā L1loc (Ī©j ) the Fubini theorem yields (5.5)
f1 ā f2 , Ļ = f1 (x), f2 (y), Ļ(x, y) = f2 (y), f1 (x), Ļ(x, y).
This formula should be considered as a basis for deļ¬nition of f1 ā f2 if f1 and f2 are distributions. Let us show that it makes sense. First, note that any compact K in Ī©1 Ć Ī©2 is contained in a compact of the form K1 Ć K2 , where Kj is a compact in Ī©j , j = 1, 2. In particular, supp Ļ ā K1 Ć K2 for some K1 , K2 . Further, Ļ(x, y) can now be considered as a C ā -function of x with values in D(K2 ) (the C ā -function with values in D(K2 ) means that all the derivatives are limits of their diļ¬erence quotients in the topology of D(K2 )). Therefore, f2 (y), Ļ(x, y) ā D(K1 ), since f2 is a continuous linear functional on D(K2 ). This implies that the functional F, Ļ = f1 (x), f2 (y), Ļ(x, y) is well-deļ¬ned. It is easy to verify that F ā D (Ī©1 Ć Ī©2 ) and supp F ā supp f1 Ć supp f2 . We can also construct the functional G ā D (Ī©1 Ć Ī©2 ) by setting G, Ļ = f2 (y), f1 (x), Ļ(x, y). Let us verify that F = G. Note that if Ļ = Ļ1 ā Ļ2 , where Ļj ā D(Ī©j ), i.e., Ļ(x, y) = Ļ1 (x)Ļ2 (y), then, clearly, F, Ļ = G, Ļ = f1 , Ļ1 f2 , Ļ2 . Therefore, obviously, to prove F = G it suļ¬ces to prove the following. Lemma 5.1. In D(Ī©1 Ć Ī©2 ), ļ¬nite linear combinations of the functions Ļ1 ā Ļ2 , where Ļj ā D(Ī©j ) for j = 1, 2, are dense. More precisely, if Kj Ė j is a compact in Ī©j , such that Kj is is a compact in Ī©j , j = 1, 2, and K Ė j , then any function Ļ ā (K1 Ć K2 ) can be contained in the interior of K Ė1 Ć K Ė 2 ) by ļ¬nite sums of functions approximated, as close as we like, in D(K Ė of the form Ļ1 ā Ļ2 , where Ļj ā D(Kj ). Proof. Consider the cube Qd in Rn1 +n2 : d d Qd = (x, y) : |xj | ā¤ , |yl | ā¤ , j = 1, . . . , n1 ; l = 1, . . . , n2 , 2 2 i.e., the cube with the center at the origin with edges of length d parallel to the coordinate axes. Take d large enough for the compact set K1 Ć K2 to
102
5. Convolution and Fourier transform
belong to the interior of Qd . In L2 (Qd ), the exponentials 2Ļi n n exp k Ā· z : k ā Z , n = n1 + n2 ; z ā R d considered as functions of z, constitute a complete orthonormal system. A function Ļ = Ļ(x, y) = Ļ(z) can be expanded into the Fourier series 2Ļi kĀ·z , Ļk exp (5.6) Ļ(z) = d k
where 1 Ļk = n d
2Ļi k Ā· z dz. Ļ(z) exp ā d Qd
The Fourier series (5.6) converges absolutely and uniformly in Qd to Ļ(z) and it can be termwise diļ¬erentiated any number of times. Indeed, integrating by parts shows that the absolute values of N
1 d2 2Ļi 2 N k Ā· z dz (1 + |k| ) Ļk = n Ļ(z) 1 ā 2 Īz exp ā d Qd 4Ļ d N
2Ļi 1 d2 k Ā· z dz Ļ(z) exp ā = n 1 ā 2 Īz d Qd 4Ļ d are bounded by a constant CN independent of k; i.e., |Ļk | ā¤ CN (1 + |k|2 )āN . This implies the convergence of (5.6) and the possibility of its termwise diļ¬erentiation. Therefore, (5.6) converges in the topology of C ā (Qd ). Now, let Ļj ā Ė j ), where Ļj = 1 in a neighborhood of Kj . We get D(K Ļ(x, y) = Ļ1 (x)Ļ2 (y)Ļ(x, y) 2Ļi 2Ļi = k1 Ā· x Ļ2 (y) exp k2 Ā· y , Ļk Ļ1 (x) exp d d k1 ,k2
Zn1 ,
k2 ā Zn2 , k = (k1 , k2 ), and the series converges in the where k1 ā Ė 2 ). But then the partial sums of this series give the Ė1 Ć K topology D(K required approximation for Ļ(x, y). Now we are able to introduce the following. Deļ¬nition 5.2. If fj ā D (Ī©j ), j = 1, 2, then the direct or tensor product of f1 and f2 is the distribution on Ī©1 Ć Ī©2 deļ¬ned via (5.5) and denoted by f1 ā f2 or f1 (x)f2 (y) or f1 (x) ā f2 (y).
5.2. Direct product of distributions
103
Clearly, supp(f1 ā f2 ) ā supp f1 Ć supp f2 , and if fj ā that
S (Rnj ),
then f1 ā f2 ā S (Rn1 +n2 ). Further, it is easy to verify
ā āf1 (f1 ā f2 ) = ā f2 , āxj āxj
ā āf2 (f1 ā f2 ) = f1 ā . āyl āyl
Example 5.1. Ī“(x)Ī“(y) = Ī“(x, y). Example 5.2. Let t ā R, x ā Rnā1 , Ī±(x) ā C(Rnā1 ). The distribution Ī“(t) ā Ī±(x) ā D (Rnt,x ) is called the single layer with density Ī± on the plane t = 0. Its physical meaning is as follows: it describes the charge supported on the plane t = 0 and smeared over the plane with the density Ī±(x). Clearly,
Ī±(x)Ļ(0, x)dx. (5.7) Ī“(t) ā Ī±(x), Ļ(t, x) = Rnā1 x
For Ī²(x) ā C(Rnā1 ), the distribution Ī“ (t) ā Ī²(x) is called the double layer with density Ī² on the plane t = 0. Its physical meaning is as follows: it describes the distribution of dipoles smeared with the density Ī² over the plane t = 0 and oriented along the t-axis. Recall that a dipole in electrostatics is a system of two very large charges of the same absolute value and of opposite signs placed very close to each other. The product of charges by the distance between them should be a deļ¬nite ļ¬nite number called the dipole moment. Clearly, in D (Rn ), the following limit relation holds: ' & Ī“(t ā Īµ) Ī“(t + Īµ) ā Ī²(x) ā ā Ī²(x) , Ī“ (t) ā Ī²(x) = lim Īµā+0 2Īµ 2Īµ implying the above interpretation of the double layer. It is also clear that
āĻ (5.8) Ī“ (t) ā Ī²(x), Ļ(t, x) = ā (0, x)Ī²(x)dx. nā1 āt Rx The formulas similar to (5.7), (5.8) can be used to deļ¬ne the single and double layers on any smooth surface Ī of codimension 1 in Rn . Namely, if ā (Ī²Ī“Ī ) by the Ī±, Ī² ā C(Ī), then we can deļ¬ne the distributions Ī±Ī“Ī and ān formulas
Ī±(x)Ļ(x)dS, Ī±Ī“Ī , Ļ = Ī
āĻ ā (Ī²Ī“Ī ), Ļ = ā Ī²(x) dS, ān ān Ī where dS is the area element of Ī and n the external normal to Ī. Note, however, that, locally, a diļ¬eomorphism reduces these distributions to the above direct products Ī“(t) ā Ī±(x) and Ī“ (t) ā Ī²(x).
104
5. Convolution and Fourier transform
Note another important property of the direct product: it is associative; i.e., if fj ā D (Ī©j ), j = 1, 2, 3, then (f1 ā f2 ) ā f3 = f1 ā (f2 ā f3 ) in D (Ī©1 Ć Ī©2 Ć Ī©3 ). The proof follows from the fact that ļ¬nite linear combinations of functions of the form Ļ1 ā Ļ2 ā Ļ3 with Ļj ā D(Ī©j ), j = 1, 2, 3, are dense in D(Ī©1 Ć Ī©2 Ć Ī©3 ) in the same sense as in Lemma 5.1.
5.3. Convolution of distributions It is clear from (5.3) that it is natural to try to deļ¬ne the convolution of distributions f, g ā D (Rn ) with the help of the formula (5.9)
f ā g, Ļ = f (x)g(y), Ļ(x + y),
Ļ ā D(Rn ).
This formula (and the convolution itself) does not always make sense, since Ļ(x + y) ā D(R2n ) for Ļ = 0. Nevertheless, sometimes the right-hand side of (5.9) can be naturally deļ¬ned. Let, for example, f ā E (Rn ). Let us set (5.10)
f (x)g(y), Ļ(x + y) = f (x), g(y), Ļ(x + y),
which makes sense since g(y), Ļ(x + y) ā E(Rnx ). We may proceed the other way around, setting (5.11)
f (x)g(y), Ļ(x + y) = g(y), f (x), Ļ(x + y),
which also makes sense, since f (x), Ļ(x + y) ā D(Rny ). Let us show that both methods give the same answer. To this end, take a function Ļ ā D(Rn ) such that Ļ(x) = 1 in a neighborhood of supp f . Then f = Ļf and we have f (x), g(y), Ļ(x + y) = Ļ(x)f (x), g(y), Ļ(x + y) = f (x), Ļ(x)g(y), Ļ(x + y) = f (x), g(y), Ļ(x)Ļ(x + y), and, similarly, g(y), f (x), Ļ(x + y) = g(y), f (x), Ļ(x)Ļ(x + y). The right-hand sides of these equations are equal, due to the property of the direct product (5.5) proved above since Ļ(x)Ļ(x + y) ā D(R2n ). Therefore, their left-hand sides coincide. Now we may introduce the following deļ¬nition. Deļ¬nition. The convolution of two distributions f, g ā D (Rn ) one of which has a compact support is the distribution f ā g ā D (Rn ) deļ¬ned via (5.9) understood in the sense of (5.10) or (5.11). It follows from the arguments above that the convolution is commutative; i.e., f ā g = g ā f,
f ā E (Rn ), g ā D (Rn ).
5.3. Convolution of distributions
105
Besides, it is associative; i.e., (f ā g) ā h = f ā (g ā h) if two of the three distributions have compact support. This is easy to verify from the associativity of the direct product and the equation (f ā g) ā h, Ļ = f (x)g(y)h(z), Ļ(x + y + z) because the same holds if (f ā g) ā h is replaced by f ā (g ā h). The rule (5.2) of diļ¬erentiating the convolution applies also to distributions. Indeed, ā Ī± (f ā g), Ļ = (ā1)|Ī±| f ā g, ā Ī± Ļ = (ā1)|Ī±| f (x)g(y), (ā Ī±Ļ)(x + y), but (ā Ī± Ļ)(x + y) can be represented in the form āxĪ± Ļ(x + y) or āyĪ± Ļ(x + y) and then, moving ā Ī± backwards, we see that (5.2) holds: ā Ī± (f ā g) = (ā Ī± f ) ā g = f ā (ā Ī± g). Example 5.3. Ī“(x) ā f (x) = f (x), f ā D (Rn ). From the algebraic viewpoint, we see that the addition and convolution endow E (Rn ) with the structure of an associative and commutative algebra (over C) with the unit Ī“(x), and D (Rn ) is a two-sided module over this algebra. Example 5.4. A solution of a nonhomogeneous equation with constant coeļ¬cients. We begin this example with Deļ¬nition. Let p(D) be a linear diļ¬erential operator with constant coeļ¬cients in Rn . Distribution E(x) ā D (Rn ) is its fundamental solution if p(D)E(x) = Ī“(x). Then the equation p(D)u = f, where f ā
E (Rn ),
has a particular solution u = E ā f.
Indeed, p(D)(E ā f ) = [p(D)E] ā f = Ī“ ā f = f, as required. Example 5.5. Using the above example we can ļ¬nd a particular solution of a nonhomogeneous ordinary diļ¬erential equation with constant coeļ¬cients without variation of parameters. Indeed, let n = 1 and p(D) = L, where L=
dmā1 d dm + a0 + a + Ā· Ā· Ā· + a1 mā1 m mā1 dx dx dx
106
5. Convolution and Fourier transform
and all coeļ¬cients aj are constants. In this case, we may take E(x) = Īø(x)y(x), where y is the solution of the equation Ly = 0 satisfying the initial conditions y(0) = y (0) = Ā· Ā· Ā· = y (mā2) (0) = 0, y (mā1) (0) = 1 (see (4.46)). This implies that if f ā C(R) has a compact support, then
x y(x ā t)f (t)dt u1 (x) = E(x ā t)f (t)dt = āā
is a particular solution of the equation Lu = f . Note that since y(x ā t) satisļ¬es, as a function of x, the equation Ly(x ā t) = 0 for any t ā R, it follows that
0 u2 (x) = y(x ā t)f (t)dt āā
satisļ¬es the equation Lu2 = 0. Therefore, together with u1 (x), the function
x (5.12) u(x) = y(x ā t)f (t)dt 0
is also a particular solution of Lu = f . Formula (5.12) gives a solution of Lu = f for an arbitrary f ā C(R) (not necessarily with a compact support) since the equality Lu = f at x depends only on the behavior of f in a neighborhood of x and the solution u itself given by this formula is determined in a neighborhood of x by the values of f on a ļ¬nite segment only. Here is an example: a particular solution of the equation u + u = f is given by the formula
x sin(x ā t)f (t)dt. u(x) = 0
Example 5.6. Potentials. If En is the fundamental solution of the Laplace operator Ī in Rn deļ¬ned by (4.38), (4.39), then u = āEn āf with f ā E (Rn ) is called a potential and satisļ¬es the Poisson equation Īu = āf . In particular, if f = Ļ, where Ļ is piecewise continuous, then for n = 3 we get the Newtonian potential
Ļ(y) 1 dy, u(x) = 4Ļ |x ā y| and for n = 2, we get the logarithmic potential
1 Ļ(y) ln |x ā y|dy. u(x) = ā 2Ļ
5.4. Support of a convolution
107
In these examples, the convolution En ā f is locally integrable for any n. As a distribution, it coincides with a locally integrable function
u(x) = ā En (x ā y)Ļ(y)dy, since En (x ā y)Ļ(y) is locally integrable in R2n x,y and, by the Fubini theorem, we can make the change of variables in the integral u, Ļ which leads us to the deļ¬nition of the convolution of distributions. Let us give other examples of potentials. Recall the deļ¬nition of the operator Ī“Ī in Example 4.4. The convolution u = āEn ā Ī±Ī“Ī ,
(5.13)
where Ī is a smooth compact surface of codimension 1 in Rn and Ī± ā C(Ī), is a simple layer potential and may be interpreted as a potential of a system of charges smeared over Ī with the density Ī±. The convolution ā (Ī²Ī“Ī ) ān is called a double layer potential and denotes the potential of a system of dipoles smeared over Ī and oriented along the exterior normal with the density of the dipole moment equal to Ī²(x).
(5.14)
u = āEn ā
Note that outside Ī both potentials (5.13) and (5.14) are C ā -functions given by the formulas
u(x) = ā En (x ā y)Ī±(y)dSy ,
Ī
āEn (x ā y) Ī²(y)dSy , āny respectively. The behavior of the potentials near Ī is of interest, and we will discuss it later in detail. u(x) =
5.4. Other properties of convolution. Support and singular support of a convolution First, let us study the convolution of a smooth function and a distribution. Proposition 5.3. Let f ā E (Rn ), g ā C ā (Rn ) or f ā D (Rn ), g ā C0ā (Rn ). Then f ā g ā C ā (Rn ) and (f ā g)(x) = f (y), g(x ā y) = f (x ā y), g(y). Proof. First, observe that g(x ā y) is an inļ¬nitely diļ¬erentiable function of x with values in C ā (Rny ). If, moreover, g ā C0ā (Rn ), then the support
108
5. Convolution and Fourier transform
of g(x ā y) as of a function of y belongs to a compact set if x runs over a compact set. Therefore, f (y), g(x ā y) ā C ā (Rn ). The equation f (x ā y), g(y) = f (y), g(x ā y) follows since, by deļ¬nition, the change of variables in distributions is performed as if we are dealing with an ordinary integral. It remains to verify that if Ļ ā D(Rn ), then f ā g, Ļ = f (y), g(x ā y), Ļ(x), or f (x), g(y), Ļ(x + y) = f (y), g(x ā y), Ļ(x). this means that Since g ā
f (x), g(y)Ļ(x + y)dy = f (y), g(x ā y)Ļ(x)dx, C ā (Rn ),
or
f (y), g(x ā y)Ļ(x)dx = f (y), g(x ā y)Ļ(x)dx;
i.e., we are to prove that the continuous linear functional f may be inserted under the integral sign. This is clear, since the integral (Riemann) sums of the integral
g(x ā y)Ļ(x)dx, considered as functions of y, tend to the integral in the topology in C ā (Rny ), because derivatives āyĪ± of the Riemann integral sums are Riemann integral sums themselves for the integrand in which g(x ā y) is replaced with āyĪ± g(x ā y). Proposition 5.3 is proved. Proposition 5.4. Let fk ā f in D (Rn ) and g ā E (Rn ) or fk ā f in E (Rn ) and g ā D (Rn ). Then fk ā g ā f ā g in D (Rn ). Proof. The proposition follows immediately from fk ā g, Ļ = fk (x), g(y), Ļ(x + y).
Corollary 5.5. C0ā (Rn ) is dense in D (Rn ) and in E (Rn ). Proof. If Ļk ā C0ā (Rn ), Ļk (x) ā Ī“(x) in E (Rn ) as k ā +ā and f ā D (Rn ), then Ļk ā f ā C ā (Rn ) and lim Ļk ā f = Ī“ ā f = f in D (Rn ). kā+ā
Therefore, C ā (Rn ) is dense in D (Rn ). But, clearly, C0ā (Rn ) is dense in C ā (Rn ). Therefore, C0ā (Rn ) is dense in D (Rn ). It is easy to see that if fl ā E (Rn ) and suppfl ā K, where K is a compact in Rn (independent of l), then fl ā f in D (Rn ) if and only if fl ā f in E (Rn ). But then it follows from the above argument that C0ā (Rn ) is dense in E (Rn ).
5.5. Smoothness of solutions
109
A similar argument shows that C0ā (Ī©) is dense in E (Ī©) and D (Ī©) and, besides, C0ā (Rn ) is dense in S (Rn ). These facts also say that in order to verify any distributional formula both of whose sides continuously depend, for example, on f ā D (Rn ), it suļ¬ces to do so for f ā C0ā (Rn ). Proposition 5.6. Let f ā E (Rn ), g ā D (Rn ). Then supp(f ā g) ā supp f + supp g. % Proof. We have to prove that if Ļ ā D(Rn ) and supp Ļ (supp f + supp g) = ā
, then f ā g, Ļ = 0. But this is clear from (5.9) since in this case supp Ļ(x + y) does not intersect with supp[f (x)g(y)]. Deļ¬nition 5.7. The singular support of a distribution f ā D (Ī©) is the smallest closed subset F ā Ī© such that f |Ī©\F ā C ā (Ī©\F ) and is denoted sing supp f . Proposition 5.8. Let f ā E (Rn ), g ā D (Rn ). Then (5.15)
sing supp(f ā g) ā sing supp f + sing supp g.
Proof. Using cut-oļ¬ functions we can, for any Īµ > 0, represent any distribution h ā D (Rn ) as a sum h = hĪµ + hĪµ , where supp hĪµ is contained in the Īµ-neighborhood of sing supp h and hĪµ ā C ā (Rn ). Let us choose such representation for f and g: f = fĪµ + fĪµ ,
g = gĪµ + gĪµ .
Then f ā g = fĪµ ā gĪµ + fĪµ ā gĪµ + fĪµ ā gĪµ + fĪµ ā gĪµ . In this sum the last three summands belong to C ā (Rn ) by Proposition 5.3. Therefore, sing supp(f ā g) ā sing supp(fĪµ ā gĪµ ) ā supp(fĪµ ā gĪµ ) ā supp fĪµ + supp gĪµ . The latter set is contained in the 2Īµ-neighborhood of sing supp f +sing supp g. Since Īµ is arbitrary, this implies (5.15), as required.
5.5. Relation between smoothness of a fundamental solution and that of solutions of the homogeneous equation We will ļ¬rst prove the following simple but important lemma. Lemma 5.9. Let p(D) be a linear diļ¬erential operator with constant coefļ¬cients, E = E(x) its fundamental solution, u ā E (Rn ), and p(D)u = f . Then u = E ā f .
110
5. Convolution and Fourier transform
Proof. We have E ā f = E ā p(D)u = p(D)E ā u = Ī“ ā u = u,
as required.
Remark 5.10. The requirement u ā E (Rn ) is essential here (the assertion is clearly wrong without it). Namely, it allows us to move p(D) from u to E in the calculation given above. Theorem 5.11. Assume that E is a fundamental solution of p(D) and E ā C ā (Rn \{0}). If u ā D (Ī©) is a solution of p(D)u = f with f ā C ā (Ī©), then u ā C ā (Ī©). Proof. Note that the statement of the theorem is local; i.e., it suļ¬ces to prove it in a neighborhood of any point x0 ā Ī©. Let Ļ ā D(Ī©), Ļ(x) = 1 in a neighborhood of x0 . Consider the distribution Ļu and apply p(D) to it. Denoting p(D)(Ļu) by f1 , we see that f1 = f in a neighborhood of x0 and, in particular, x0 ā sing supp f1 . By Lemma 5.9, we have Ļu = E ā f1 .
(5.16)
We now use Proposition 5.8. Since sing supp E = {0}, we get sing supp(Ļu) ā sing supp f1 . Therefore, x0 ā sing supp Ļu and x0 ā sing supp u; i.e., u is a C ā -function in a neighborhood of x0 , as required. A similar fact is true for the analyticity of solutions. Recall that a function u = u(x) deļ¬ned in an open set Ī© ā Rn , with values in R or C, is called real analytic if for every y ā Ī© there exists a neighborhood Uy of y in Ī©, such that u|Uy = u|Uy (x) can be presented in Uy as a sum of a convergent power series cĪ± (x ā y)Ī± , x ā Uy , u(x) = Ī±
with the sum over all multi-indices Ī±. Consider ļ¬rst the case of the homogeneous equation p(D)u = 0. Theorem 5.12. Assume that a fundamental solution E(x) of p(D) is real analytic in Rn \{0}, u ā D (Ī©), and p(D)u = 0. Then u is a real-analytic function in Ī©. Proof. We know already that u ā C ā (Ī©). Using the same argument as in the proof of Theorem 5.11, consider (5.16), where now f1 ā D(Ī©) and f1 (x) = 0 in a neighborhood of x0 . Then for x close to x0 we have
E(x ā y)f1 (y)dy. (5.17) u(x) = |yāx0 |ā„Īµ>0
5.5. Smoothness of solutions
111
Note that the analyticity of E(x) for x = 0 is equivalent to the possibility to extend E(x) to a holomorphic function in a complex neighborhood of the set Rn \{0} in Cn , i.e., in an open set G ā Cn such that G ā© Rn = Rn \{0}. Recall that a complex-valued function Ļ = Ļ(z) deļ¬ned for z ā G, where G is an open subset in Cn , is called holomorphic or complex analytic or even simply analytic if one of the following two equivalent conditions is fulļ¬lled: a) Ļ(z) can be expanded in a power series in z ā z0 in a neighborhood of each point z0 ā G; i.e., cĪ± (z ā z0 )Ī± , Ļ(z) = Ī±
where z is close to z0 , cĪ± ā C, and Ī± is a multi-index. b) Ļ(z) is continuous in G and diļ¬erentiable with respect to each complex variable zj (i.e., is holomorphic with respect to each zj ). A proof can be found on the ļ¬rst pages of any textbook on functions of several complex variables (e.g., Gunning and Rossi [11, Ch. 1, Theorem A2] or HĀØormander [14, Ch. 2]). Note that in (5.17) the variable y varies over a compact K ā Rn . Fix y0 = x0 . Then for |z ā x0 | < Ī“, |w ā y0 | < Ī“ with z, w ā Cn and Ī“ suļ¬ciently small, we can deļ¬ne E(z ā w) as a holomorphic function in z and w. Then, clearly,
|yāy0 |ā¤Ī“
E(z ā y)f1 (y)dy
is well-deļ¬ned for |z āx0 | < Ī“ and holomorphic in z since we can diļ¬erentiate with respect to z under the integral sign. But using a partition of unity, we can express the integral in (5.17) as a ļ¬nite sum of integrals of this form. Therefore, we can deļ¬ne u(z) for complex z, which are close to x0 , so that u(z) is analytic in z. In particular, u(x) can be expanded in a Taylor series in a neighborhood of x0 ; i.e., u(x) is real analytic. The case of a nonhomogeneous equation reduces to the above due to the Cauchy-Kovalevsky theorem which we will give in a form convenient for us. Theorem 5.13. (Cauchy-Kovalevsky) Consider the equation āmu + aĪ± (x)D Ī± u(x) = f (x), (5.18) āxm n |Ī±|ā¤m,Ī±n 0 such that (5.30) |ā Ī± fĖ(Ī¾)| ā¤ CĪ± (1 + |Ī¾|)N Ī¾
for any multi-index Ī±. Proof. Let us choose constants N , C and a compact K ā Rn so that sup |ā Ī± Ļ(x)|, Ļ ā C ā (Rn ). (5.31) |f, Ļ| ā¤ C |Ī±|ā¤N
xāK
Set g(Ī¾) = f, eāiĪ¾Ā·x . Since eāiĪ¾Ā·x is an inļ¬nitely diļ¬erentiable function in Ī¾ with values in C ā (Rn ), then, clearly, g(Ī¾) ā C ā (RnĪ¾ ) and āĪ¾Ī± g(Ī¾) = f, (āix)Ī± eāiĪ¾Ā·x , implying that (5.30) holds for g(Ī¾). It remains to verify that g(Ī¾) = fĖ(Ī¾), i.e., that
āiĪ¾Ā·x f (x), Ļ(Ī¾)e dĪ¾ = Ļ(Ī¾)f (x), eāiĪ¾Ā·x dĪ¾, Ļ ā S(RnĪ¾ ). But this follows from the convergence of Ļ(Ī¾)eāiĪ¾Ā·x dĪ¾ in the topology of C ā (Rnx ) which means that this integral and integrals obtained from it by taking derivatives of any order converge uniformly with respect to x ā K (where K is a compact set in Rn ). Remark 5.22. The condition fĖ ā C ā (Rn ) and the estimate (5.30) are necessary but not suļ¬cient for a distribution f to have a compact support. For instance, if f ā E (Rn ), then, clearly, fĖ(Ī¾) may be deļ¬ned via (5.29) for Ī¾ ā Cn and we get an entire (i.e., homomorphic everywhere in Cn )-function satisfying |fĖ(Ī¾)| ā¤ C(1 + |Ī¾|)N eA|Im Ī¾| . It turns out that this condition on fĖ is already necessary and suļ¬cient for f ā E (Rn ). This fact is a variant of the Paley-Wiener theorem and its proof may be found in HĀØ ormander [13, Theorem 1.7.7] or [15, Theorem 7.3.1].
120
5. Convolution and Fourier transform
Proposition 5.23. Let f ā E (Rn ), g ā S (Rn ). Then (5.32)
f ā g(Ī¾) = fĖ(Ī¾)Ė g (Ī¾).
Proof. Note that the right-hand side of (5.32) makes sense due to Proposition 5.21. It is also easy to verify that f ā g ā S (Rn ) so that the left-hand side of (5.32) makes sense. Indeed, if Ļ ā S(Rn ), then f (y), Ļ(x + y) ā S(Rnx ) due to (5.31). Therefore, for Ļ ā S(Rn ) the equality f ā g, Ļ = g(x), f (y), Ļ(x + y) makes sense and we clearly get f ā g ā S (Rn ). Now let us verify (5.32). For Ļ ā S(Rn ) we get
āiĪ¾Ā·x f ā g(Ī¾), Ļ(Ī¾) = f ā g, Ļ(Ī¾)e dĪ¾
āiĪ¾Ā·(x+y) dĪ¾ = g(x), f (y), Ļ(Ī¾)e
. / = g(x), Ļ(Ī¾)eāiĪ¾Ā·x f (y), eāiĪ¾Ā·y dĪ¾
āiĪ¾Ā·x Ė dĪ¾ = g(x), Ļ(Ī¾)f (Ī¾)e = Ė g (Ī¾), fĖ(Ī¾)Ļ(Ī¾) = fĖ(Ī¾)Ė g (Ī¾), Ļ(Ī¾),
as required.
5.9. Applying the Fourier transform to ļ¬nd fundamental solutions Let p(D) be a diļ¬erential operator with constant coeļ¬cients, E ā S (Rn ) its Ė ā S (Rn ) satisļ¬es fundamental solution. The Fourier transform E(Ī¾) (5.33)
Ė = 1. p(Ī¾)E(Ī¾)
Ė = 1/p(Ī¾) on the open set {Ī¾ : p(Ī¾) = 0}. If p(Ī¾) = 0 everywhere Clearly, E(Ī¾) n Ė = 1/p(Ī¾) and apply the inverse Fourier in R , then we should just take E(Ī¾) transform. (It can be proved, though nontrivially, that in this case 1/p(Ī¾) is estimated by (1 + |Ī¾|)N with some N > 0; hence 1/p(Ī¾) ā S (RnĪ¾ )āsee, e.g., HĀØ ormander [15, Example A.2.7 in Appendix A2, vol. II].) When p(Ī¾) has zeros we should regularize 1/p(Ī¾) in neighborhoods of zeros of p(Ī¾) so as Ė to get a distribution E(Ī¾) equal to 1/p(Ī¾) for p(Ī¾) = 0 and satisfying (5.33). Ė = 1 Ė = v.p. 1 or E(Ī¾) For instance, for n = 1 and P (Ī¾) = Ī¾ we can set E(Ī¾) Ī¾ Ī¾Ā±i0 Ė = v.p. 1 + CĪ“(Ī¾), where C is an arbitrary constant. or, in general, E(Ī¾) Ī¾ In the general case, we must somehow give meaning to integrals of the form
Ļ(Ī¾) dĪ¾, Ļ ā S(Rn ), p(Ī¾) n R
5.10. Liouvilleās theorem
121
to get a distribution with the needed properties. It can often be done, e.g., 1 is by entering the complex domain (the construction of the distribution Ī¾Ā±i0 an example). Sometimes 1/p(Ī¾) is locally integrable. Then it is again natural Ė to assume E(Ī¾) = 1/p(Ī¾). For instance, for p(D) = Ī, i.e., p(Ī¾) = āĪ¾ 2 , we Ė can set E(Ī¾) = ā1/Ī¾ 2 for n ā„ 3. Since, as is easy to verify, the Fourier transform and its inverse preserve the spherical symmetry and transform a distribution of homogeneity degree Ī± into a distribution of homogeneity degree āĪ±ān, then, clearly, F ā1 ( |Ī¾|12 ) = C|x|2ān , where C = 0. Therefore, we get the fundamental solution En (x) described above. For n = 2 the function |Ī¾|12 is not integrable in a neighborhood of 0. Here a regularization is required. For instance, we can set
Ļ(Ī¾) ā Ļ(Ī¾)Ļ(0) Ė dĪ¾, E2 (Ī¾), Ļ(Ī¾) = ā |Ī¾|2 R2 where Ļ(Ī¾) ā C0ā (R2 ), Ļ(Ī¾) = 1 in a neighborhood of 0. It can be veriļ¬ed 1 ln |x| + C, where C depends on the choice of Ļ(Ī¾). that E2 (x) = 2Ļ
5.10. Liouvilleās theorem Theorem 5.24. Assume that the symbol p(Ī¾) of an operator p(D) vanishes (for Ī¾ ā Rn ) only at Ī¾ = 0. If u ā S (Rn ) and p(D)u = 0, then u(x) is a polynomial in x. In particular, if p(D)u = 0, where u(x) is a bounded measurable function on Rn , then u = const. Remarks. a) The inclusion u ā S (Rn ) holds, for instance, if u ā C(Rn ) and |u(x)| ā¤ C(1 + |x|)N , x ā Rn , for some C and N . b) If p(Ī¾0 ) = 0 for some Ī¾0 = 0, then p(D)u = 0 has a bounded nonconstant solution eiĪ¾0 Ā·x . Therefore, p(Ī¾) = 0 for Ī¾ = 0 is a necessary condition for the validity of the theorem. c) If p(Ī¾) = 0 for all Ī¾ ā Rn , then p(D)u = 0 for u ā S (Rn ) implies, as we will see from the proof of the theorem, that u ā” 0. Proof of Theorem 5.24. Applying the Fourier transform we get p(Ī¾)Ė u(Ī¾) = 0. Since p(Ī¾) = 0 for Ī¾ = 0, then, clearly, the support of u Ė(Ī¾) is equal to {0} (provided u ā” 0). We give more detail: if Ļ(Ī¾) ā D(Rn \{0}), then Ļ(Ī¾) n p(Ī¾) ā D(R \{0}) implying Ļ(Ī¾) Ļ(Ī¾) = p(Ī¾)Ė u(Ī¾), = 0. Ė u(Ī¾), Ļ(Ī¾) = u Ė(Ī¾), p(Ī¾) p(Ī¾) p(Ī¾)
122
5. Convolution and Fourier transform
Thus, supp u Ė(Ī¾) ā {0}. Therefore, aĪ± Ī“ (Ī±) (Ī¾), u Ė(Ī¾) =
aĪ± ā C,
|Ī±|ā¤p
implying u(x) =
c Ī± xĪ± ,
cĪ± ā C,
|Ī±|ā¤p
as required.
Example. If u(x) is a harmonic function everywhere on Rn and |u(x)| ā¤ M , then u = const (this statement is often called Liouvilleās theorem for harmonic functions). If u(x) is a harmonic function in Rn and |u(x)| ā¤ C(1 + |x|)N , then u(x) is a polynomial. The same holds for solutions of Īm u = 0. Later we will prove a more precise statement: If u is harmonic and bounded below (or above), e.g., u(x) ā„ āC, x ā Rn , then u = const (see, e.g., Petrovskii [24]). This statement fails already for the equation Ī2 u = 0 satisļ¬ed by the nonnegative polynomial u(x) = |x|2 . Let us give one more example of an application of Liouvilleās theorem. Let for n ā„ 3 a function u(x) be deļ¬ned and harmonic in the exterior of a ball, i.e., for |x| > R and u(x) ā 0 as |x| ā +ā. We wish to describe in detail the behavior of u(x) as |x| ā ā. To this end we may proceed, for example, as follows. Take a function Ļ(x) ā C ā (Rn ) such that 0 for |x| ā¤ R + 1, Ļ(x) = 1 for |x| ā„ R + 2. Now, consider the following function u Ė(x) ā C ā (Rn ): Ļ(x)u(x) for |x| > R, u Ė(x) = 0 for |x| ā¤ R. Clearly, ĪĖ u(x) = f (x) ā C0ā (Rn ). Let us verify that (5.34)
u Ė = En ā f,
where En is the standard fundamental solution of the Laplace operator. Clearly, (En ā f )(x) ā 0 as |x| ā ā and, by assumption, the same holds for u(x) (and, hence, for u Ė(x), since u Ė(x) = u(x) for |x| > R + 2). Moreover, Ī(En ā f ) = ĪEn ā f = f, so Ī(Ė u ā En ā f ) = 0 everywhere on Rn . But, by Liouvilleās theorem, this implies that u Ė ā En ā f = 0, as required.
5.11. Problems
123
Formula (5.34) can be rewritten in the form
u(x) = En (x ā y)f (y)dy, |x| > R + 2. |y|ā¤R+2
The explicit form of En now implies |u(x)| ā¤
C , |x|nā2
|x| > R + 2.
The above argument is also applicable in a more general situation. For harmonic functions more detailed information on the behavior of u(x) as |x| ā ā can also be obtained with the help of the Kelvin transformation x 2ān u . u(x) ā v(x) = |x| |x|2 It is easy to verify that the Kelvin transformation preserves harmonicity. For n = 2 the Kelvin transformation is the change of variables induced by the inversion x ā x/|x|2 . If u(x) is deļ¬ned for large |x|, then v(x) is deļ¬ned in a punctured neighborhood of 0, i.e., in Ī©\{0}, where Ī© is a neighborhood of 0. If u(x) ā 0 as |x| ā +ā, then v(x)|x|nā2 ā 0 as |x| ā 0, and by the removable singularity theorem, v(x) can be extended to a function which is harmonic everywhere in Ī©. (We will denote the extension by the same letter v.) Observe that the Kelvin transformation is its own inverse (or, as they say, is an involution); i.e., we may express u in terms of v via the same formula u(y) = |y|2ān v( |y|y 2 ). Expanding v(x) into Taylor series at x = 0 we see that u(y) has a convergent expansion into a series of the form yĪ± cĪ± 2|Ī±| u(y) = |y|2ān |y| Ī± in a neighborhood of inļ¬nity. In particular, u(y) = c0 |y|2ān + O(|y|1ān ) as |y| ā +ā.
5.11. Problems 5.1. Find the Fourier transforms of the following distributions: a)
1 x+i0 ;
b) Ī“(r ā r0 ), r = |x|, x ā R3 ; c)
sin(r0 |x|) , |x|
x ā R3 ;
d) xk Īø(x), k ā Z+ .
124
5. Convolution and Fourier transform
5.2. Find the inverse Fourier transforms of the following distributions: a) 1/|Ī¾|2 , Ī¾ ā Rn , n ā„ 3; Ī¾ ā R3 , k > 0;
b)
1 , |Ī¾|2 +k2
c)
1 , |Ī¾|2 āk2 Ā±i0
Ī¾ ā R3 , k > 0.
5.3. Using the result of Problem 5.2 b), ļ¬nd the fundamental solution for the operator āĪ + k 2 in R3 . 5.4. We will say that the convolution f ā g of two distributions f, g ā D (Rn ) is deļ¬ned if for any sequence Ļk ā D(Rn ) such that Ļk (x) = 1 for |x| ā¤ k, k = 1, 2, . . ., the limit f ā g = lim f ā (Ļk g) exists in D (Rn ) and kāā
does not depend on the choice of the sequence Ļk . Prove that if f, g ā D (R) and the supports of f and g belong to [0, +ā), then f ā g is deļ¬ned and addition and convolution deļ¬ne the structure of an associative commutative ĀÆ + ) of distributions with supports in [0, +ā). ring with unit on the set D (R 5.5. Prove that Ī“ (k) (x) ā f (x) = f (k) (x) for f ā D (R). eixt x+i0 tāāā
5.6. Find the limits lim
eixt x+i0 tā+ā
and lim
in D (R1 ).
ĀÆ + ) deļ¬ne an integration operator setting If = x f (Ī¾)dĪ¾. 5.7. For f ā L1loc (R 0 ĀÆ + ). Prove that If = f ā Īø and I can be extended by continuity to D (R 5.8. Using the associativity of the convolution, prove that & ' 1 xmā1 Īø(x) ā f, I mf = (m ā 1)! ĀÆ + ). where m = 1, 2, . . ., for f ā D (R ĀÆ + ) and, in 5.9. Set xĪ»+ = Īø(x)xĪ» for Re Ī» > ā1, so that xĪ»+ ā L1loc (R Ī» Ī» ĀÆ + ). Deļ¬ne the fractional integral I by the formula particular, x+ ā D (R I Ī»f =
Ī»ā1 x+ ā f, Re Ī» > 0. Ī(Ī»)
Prove that I Ī» I Ī¼ = I Ī»+Ī¼ , Re Ī» > 0, Re Ī¼ > 0. 5.10. For any Ī» ā C deļ¬ne the fractional integral I Ī» by k d Ī» ĀÆ + ), I Ī»+k f for f ā D (R I f= dx
5.11. Problems
125
where k ā Z+ is such that Re Ī» + k > 0. Prove that I Ī» is well-deļ¬ned; i.e., the result does not depend on the choice of k ā Z+ . Prove that I Ī» I Ī¼ = I Ī»+Ī¼ for any Ī», Ī¼ ā C. Prove that I 0 f = f and I āk f = f (k) for k ā Z+ and any ĀÆ + ). So I āĪ» f for Ī» > 0 can be considered a fractional derivative (of f ā D (R order Ī») of the distribution f .
10.1090/gsm/205/06
Chapter 6
Harmonic functions
6.1. Mean-value theorems for harmonic functions A harmonic function in an open set Ī© ā Rn (n ā„ 1) is a (real-valued or complex-valued) function u ā C 2 (Ī©) satisfying the Laplace equation (6.1)
Īu = 0,
where Ī is the standard Laplacian on Rn . As we already know from Chapter 5, any harmonic function is in fact in C ā (Ī©) and even real-analytic in Ī©. This is true even if we a priori assume only that u ā D (Ī©), i.e., if u is a distribution in Ī©. For n = 1 the harmonicity of a function on an interval (a, b) means that this function is linear. Since all statements discussed here are true but usually trivial for n = 1, we can concentrate on the case n ā„ 2. In this chapter we will discuss classical properties of harmonic functions and related topics. We will start with a property of harmonic functions that is called the mean value theorem, or Gaussās law of the arithmetic mean. Theorem 6.1 (Mean value theorem for spheres). Let B = Br (x) be an open ball in Rn with the center at x and radius r > 0, let S = Sr (x) be its ĀÆ be the boundary (the sphere with the same center and radius), and let B ĀÆ = B āŖ S). Let u be a continuous function on B ĀÆ such that closure of B (so B its restriction to B is harmonic. Then the average of u over S equals u(x) (the value of u at the center ). In other words, (6.2)
u(x) =
1 volnā1 (S)
u(y)dSy = S
1 Ļnā1 rnā1
u(y)dSy , Sr (x)
127
128
6. Harmonic functions
where dSy means the standard area element on the sphere S and Ļnā1 is the (n ā 1)-volume of the unit sphere. Proof. Let us rewrite the average of u in (6.2) as an average over the unit sphere. To this end, denote z = rā1 (y ā x) and make the change of variables y ā z in the last integral. We get then
1 1 u(y)dSy = u(x + rz)dSz . Ļnā1 rnā1 Sr (x) Ļnā1 |z|=1 Now, if we replace r by a variable Ļ, 0 ā¤ Ļ ā¤ r, then the integral becomes a function I = I(Ļ) on [0, r], which is continuous due to the uniform continuity ĀÆ Clearly, I(0) = u(x). Our theorem will be proved if we show of u on B. that I(Ļ) = const, or, equivalently, that I (Ļ) ā” 0 on [0, r). To this end, let us calculate the derivative I (Ļ) by diļ¬erentiating under the integral sign (which is clearly possible):
d āu 1 1 u(x + Ļz)dSz = I (Ļ) = (x + Ļz)dSz , Ļnā1 |z|=1 dĻ Ļnā1 |z|=1 ā n ĀÆĻ where n ĀÆ Ļ is the outgoing unit normal to the sphere SĻ (x), at the point x+Ļz. We will see that the last integral vanishes if we apply the following corollary of Greenās formula (4.32), which is valid for any bounded domain with a C 2 boundary
āu (6.3) dS Īu dx = ĀÆ Ī© āĪ© ā n (take v ā” 1 in (4.32)). For our purposes we should take Ī© = BĻ (x). Since u is harmonic, we come to the desired conclusion that I (Ļ) = 0. The sphere in Theorem 6.1 can be replaced by a ball: Theorem 6.2 (Mean value theorem for balls). In the same notation as in ĀÆ = B ĀÆr (x) such that its Theorem 6.1, let u be a continuous function on B restriction to B is harmonic. Then the average of u over B equals u(x). In other words,
n 1 u(y)dy = u(y)dy, (6.4) u(x) = vol(B) B Ļnā1 rn Br (x) where dy means the standard volume element (Lebesgue measure) on Rn . Proof. Let Ļ denote the distance from x, considered as a function on Br (x), as in the proof of Theorem 6.1. Since | grad Ļ| = 1, we can replace integration over Br (x) by two subsequent integrations: ļ¬rst integrate over the spheres
6.2. Maximum principle
129
SĻ (x) for every Ļ ā [0, r], and then integrate with respect to the radii Ļ (no additional Jacobian type factor is needed). Using Theorem 6.1, we obtain
r
r u(y)dy = u(y)dSĻ (y)dĻ = u(x)volnā1 (SĻ (x))dĻ Br (x)
0
SĻ (x)
0
= u(x) vol(Br (x)), which is what we need.
Remark 6.3. Theorem 6.1 shows one way that a local property can imply a global property: (6.1) implies the mean value property (6.2). It suggests that we can try to connect the desired global property with a trivial one by a path (in our case, the path is parametrized by the radius Ļ ā [0, r]) and expand the validity of the desired property along this path, starting from a trivial case (Ļ = 0 in our case), relying on the local information when moving by an inļ¬nitesimal step with respect to the parameter. (The proof of Theorem 6.2 does not belong to this type, because it simply deduces a global result from another global one.)
6.2. The maximum principle In this section we will often assume for simplicity of formulation that Ī© is an open connected subset in Rn . (In case Ī© is not connected, we can then apply the results on each connected component separately.) Recall that connectedness of an open set Ī© ā Rn can be deļ¬ned by one of the following equivalent conditions: (i) Ī© cannot be written as a union of two disjoint nonempty open subsets. (ii) If F ā Ī© is nonempty open and at the same time closed (in Ī©), then F = Ī©. (iii) Every two points x, y ā Ī© can be connected by a continuous path in Ī© (that is, there exists a continuous map Ī³ : [0, 1] ā Ī©, such that Ī³(0) = x and Ī³(1) = y). (For the proofs see, e.g., Armstrong [1, Sect. 3.5].) The maximum principle for harmonic functions can be formulated as follows: Theorem 6.4 (Maximum principle for harmonic functions). Let u be a realvalued harmonic function in a connected open set Ī© ā Rn . Then u cannot have a global maximum in Ī© unless it is constant. In other words, if x0 ā Ī© and u(x0 ) ā„ u(x) for all x ā Ī©, then u(x) = u(x0 ) for all x ā Ī©. Proof. Let x0 ā Ī© be a point of global maximum for u and let B = Br (x0 ) ĀÆ ā Ī©. (This is possible be a ball in Rn (with the center at x0 ) such that B
130
6. Harmonic functions
if the radius r is suļ¬ciently small.) Now, applying (6.4) with x = x0 , we obtain
1 1 0 = u(x0 ) ā u(y)dy = (u(x0 ) ā u(y))dy. vol(B) B vol(B) B Since u(x0 ) ā u(y) ā„ 0 for all y ā Ī©, we conclude that u(y) = u(x0 ) for all y ā B. Let us consider the set M = {y ā Ī© | u(y) = u(x0 )}. As we just observed, for every point y ā M there exist a ball B centered at y, such that B ā M . This means that M is open. But it is obviously closed too. It is nonempty because it contains x0 . Since Ī© is connected, we conclude that M = Ī© which means that u(y) = u(x0 ) = const for all y ā Ī©. Corollary 6.5. Let u be a real-valued harmonic function in a connected open set Ī© ā Rn . Then u cannot have a global minimum in Ī© unless it is constant. Proof. Applying Theorem 6.4 to āu, we obtain the desired result.
Corollary 6.6. Let u be a complex-valued harmonic function in a connected open set Ī© ā Rn . Then |u| cannot have a global maximum in Ī© unless u = const. Proof. Assume that x0 ā Ī© and |u(x0 )| ā„ |u(x)| for all x ā Ī©. If |u(x0 )| = 0, then u ā” 0 and the result is obvious. Otherwise, write u(x0 ) = |u(x0 )|eiĻ0 , Ė = eāiĻ0 u, which is also 0 ā¤ Ļ0 < 2Ļ, and consider a new function u harmonic. Denote the real and imaginary parts of u Ė by v, w, so that u Ė = v + iw, where v and w are real-valued harmonic functions on Ī©. Since u Ė(x0 ) = |u(x0 )| is real valued, w(x0 ) = 0 and so for all x ā Ī© we have u(x)| = v 2 (x) + w2 (x) ā„ v(x). (6.5) v(x0 ) = |u(x0 )| ā„ |u(x)| = |Ė Applying Theorem 6.4 to v, we see that v(x) = v(x0 ) = const for all x ā Ī©, and all the inequalities in (6.5) become equalities. In particular, we have v(x0 ) = v 2 (x0 ) + w2 (x); hence w ā” 0, which implies the desired result. For bounded domains Ī© we can take into account what happens near the boundary to arrive at other important facts. Theorem 6.7. 1) Let u be a real-valued harmonic function in a connected open bounded Ī© ā Rn such that u can be extended to a continuous function (still denoted by u) on the closure Ī©. Then for any x ā Ī© (6.6)
min u ā¤ u(x) ā¤ max u, āĪ©
where āĪ© = Ī© \ Ī© is the boundary of Ī©.
āĪ©
6.3. Dirichletās problem
131
2) Let u be a complex-valued harmonic function on a connected open bounded Ī© that can be extended to a continuous function on Ī©. Then for any x ā Ī© (6.7)
|u(x)| ā¤ max |u|. āĪ©
3) The inequalities in (6.6) and (6.7) become strict for x in those connected components of Ī© where u is not constant. Proof. Since Ī© is compact, u should attain its maximum and minimum somewhere on Ī©. But due to Theorem 6.4 and Corollary 6.5 this may happen only on the boundary or in the connected components of Ī© where u is constant. The desired results for a real-valued u immediately follow. For complex-valued u the same argument works if we use Corollary 6.6 instead of Theorem 6.4 and Corollary 6.5. Corollary 6.8. Let Ī© be an open bounded set in Rn and let u, v be harmonic functions that can be extended to continuous functions on Ī© such that u = v on āĪ©. Then u ā” v everywhere in Ī©. Proof. Taking w = u ā v, we see that w is harmonic and w = 0 on āĪ©. Now it follows from (6.7) that w ā” 0 in Ī©.
6.3. Dirichletās boundary value problem For a given open set Ī© and Ļ ā C(āĪ©), the following boundary value problem is called Dirichletās problem for the Laplace equation: Īu(x) = 0, x ā Ī©, (6.8) u|āĪ© = Ļ(x), x ā āĪ©. We are interested in a classical solution u, that is, u ā C 2 (Ī©) ā© C(Ī©). Corollary 6.8 shows that for any bounded Ī© and any continuous Ļ, there may be at most one solution of this problem; i.e., we have uniqueness for (6.8). It is natural to ask whether such a solution exists for a given Ī© and Ļ ā C(āĪ©). This proves to be true if Ī© is bounded and āĪ© is smooth or piecewise smooth. But, in the general case, the complete answer, characterizing open sets Ī© for which the Dirichlet problem is solvable for any continuous Ļ, is highly nontrivial. The answer was given by Norbert Wiener, and it is formulated in terms of capacity, the notion which appeared in electrostatics and was brought by Wiener to mathematics in the 1920s. The whole story is beyond the scope of this course and can be found in books on potential theory
132
6. Harmonic functions
(see, e.g., Wermer [39]). On the other hand, it occurs that in some weaker sense the problem is always solvable. The appropriate terminology and necessary techniques (functional analysis, Sobolev spaces) will be discussed in Chapters 8 and 9. It is quite natural to require the existence and uniqueness of a solution of a mathematical problem that appears as a model of a real-life situation. (For example, if it is not unique, then we should additionally investigate which one corresponds to reality.) But this is not enough. Another reasonable requirement is the well-posedness of the problem, which means that the solution continuously depends on the data of the problem, where the continuity is understood with respect to some natural topologies in the spaces of data and solutions. The well-posedness of Dirichletās problem in the sup-norm immediately follows from the maximum principle. More precisely, we have Corollary 6.9. Let Ī© be an open bounded domain in Rn . Let u1 , u2 be harmonic functions in Ī© that can be extended to continuous functions on Ī©, and let Ļj = uj |āĪ© , j = 1, 2. Then (6.9)
max |u1 ā u2 | ā¤ max |Ļ1 ā Ļ2 |. Ī©
āĪ©
Proof. The result immediately follows from (6.7) in Theorem 6.7.
Now assume that {uk | k = 1, 2, . . . } is a sequence of harmonic functions in Ī©, continuously extendable to Ī©, whose restrictions to āĪ© (call them Ļk ) converge uniformly to Ļ on āĪ© as k ā ā. Then {Ļk } is a Cauchy sequence in the Banach space C(āĪ©) with the sup-norm. It follows from (6.9) that {uk } is a Cauchy sequence in C(Ī©). Hence, uk ā u ā C(Ī©) with respect to this norm. It is easy to see that the limit function u is also harmonic. Indeed, the uniform convergence in Ī© obviously implies convergence uk ā u as distributions (that is, in D (Ī©)); hence Īuk ā Īu. But Īuk ā” 0 in Ī©, so Īu ā” 0 in Ī©; i.e., u is harmonic. The Laplace equation and the Dirichlet problem appear in electrostatics as the equation and boundary value problem for a potential in a domain Ī© which is free of electric charges. It is also natural to consider a modiļ¬cation which allows charges, including point charges and also charges distributed in some volume or on a surface in Ī©. To this end we should replace the Laplace equation by the Poisson equation Īu = f (see (4.41)), where f is an appropriate distribution. This leads to Dirichletās problem for the Poisson equation: Īu(x) = f (x), x ā Ī©, (6.10) x ā āĪ©. u|āĪ© = Ļ(x),
6.4. Hadamardās example
133
To make this precise, we consider u as a distribution in Ī© that is in fact a continuous function in a neighborhood U in Ī© of the boundary āĪ©. We can also take f ā D (Ī©) and Ļ ā C(āĪ©). Then we can extend the result of Corollary 6.8 to the problem (6.10): Corollary 6.10. Assume that u, v ā D (Ī©) are solutions of the boundary value problem (6.10) with the same f and Ļ. Then u = v in Ī©. Proof. Clearly, w = u ā v is a distribution in Ī©, satisfying Īw = 0. Therefore, by Theorem 5.11, w is a harmonic function in Ī©. By our assumptions, w is continuous up to the boundary and w = 0 on āĪ©. Hence, w ā” 0 by the maximum principle.
6.4. Hadamardās example In this section we encounter an example, due to Jacques Hadamard, which shows that there exist uniquely solvable problems which are not well-posed with respect to natural norms; i.e., small perturbations of the data may lead to uncontrollably large perturbations of the solution. Let us consider the initial value problem for the Laplacian in a twodimensional strip Ī T = {(x, y) ā R2 | x ā R, 0 ā¤ y < T }: Īu = 0 in Ī T , (6.11) u(x, 0) = Ļ(x), āu āy (x, 0) = Ļ(x) for all x ā R, where we assume that u ā C 2 (Ī T ), Ļ ā C 2 (R), and Ļ ā C 1 (R). A solution of this problem may not exist (see Problem 6.5), but if it exists, then it is unique. To prove the uniqueness we will need the following, in which we write coordinates for Rn as x = (x , xn ), where x ā Rnā1 and xn ā R. Lemma 6.11. Let Ī© be an open set in Rn and assume that a function u ā C 1 (Ī©) is harmonic in the open sets Ī©Ā± = {x ā Ī©| Ā± xn > 0}, which are the portions of Ī© lying, respectively, above and below the hyperplane xn = 0. Then u is in fact a harmonic function in Ī©. Proof. It suļ¬ces to prove that Īu = 0 in the sense of distributions, i.e., that for any test function Ī¦ ā C0ā (Ī©) we have
u ĪĪ¦ dx = 0. (6.12) Ī©
Note that it is suļ¬cient to prove this equality locally, i.e., for test functions Ī¦ supported in a small neighborhood of any chosen point xā ā Ī©. Since this is obvious for points xā ā Ī©Ā± , we may assume that xā lies on the hyperplane
134
6. Harmonic functions
xn = 0 and that Ī© is a small ball centered at such a point. Then, due to Greenās formula (4.32), we can write
āĪ¦ āu u dx , u ĪĪ¦ dx = āĪ¦ āxn āxn Ī©ā Ī©ā©{xn =0} where we used that Īu = 0 in Ī©ā . Similarly,
āu āĪ¦ u ĪĪ¦ dx = ā āĪ¦ u dx , āx āx + n n Ī© Ī©ā©{xn =0} where the minus sign in the right-hand side appears due to the fact that the direction of the xn -axis is outward for Ī©ā and inward for Ī©+ . We also used the continuity of u and āu/āxn which implies that these quantities are the same from both sides of {xn = 0}. Adding these equalities leads to (6.12), which ends the proof. Now let us turn to the proof of the uniqueness in (6.11). For two solutions u1 , u2 , their diļ¬erence u = u1 āu2 is a solution of (6.11) with Ļ ā” Ļ ā” 0. By reļ¬ection as an odd function in y, extend u to a function u Ė on the doubled strip Ė T = {(x, y) ā R2 | x ā R, y ā (āT, T )}. Ī (In detail, deļ¬ne u Ė(x, y) = āu(x, āy) for āT < y < 0.) Due to the initial Ė T ). Also, u Ė is harmonic in the conditions on u, we clearly have u Ė ā C 1 (Ī Ė Ė T except at the xdivided strip Ī T \ {(x, 0)| x ā R}, i.e., everywhere in Ī Ė T . Therefore, axis. It follows from Lemma 6.11 that u Ė is harmonic in Ī Ė it is real-analytic in Ī T by Theorem 5.14 (or Corollary 5.15). But it is also easy to check that the equation ĪĖ u = 0 and the initial conditions Ė/āy|y=0 = 0 imply, by induction in |Ī±|, that ā Ī± u Ė|y=0 = 0 for u Ė|y=0 = ā u every multi-index Ī± = (Ī±1 , Ī±2 ). Hence, using the Taylor expansion of u Ė at Ė T , which proves the desired the point x = y = 0, we conclude that u Ė ā” 0 in Ī uniqueness. Remark 6.12. The uniqueness of C m -solutions for any mth-order equation with real-analytic coeļ¬cients and initial conditions on any noncharacteristic hypersurface is the content of a Holmgren theorem (see, e.g., John [16, Sect. 3.5]). This theorem is deduced, by a duality argument, from the existence of a solution for an adjoint problem in real-analytic functions, which is the content of the Cauchy-Kovalevsky theorem (see Theorem 5.14). In the next chapter we will explain Holmgrenās method in detail but for the heat equation. Now we can proceed to the Hadamard example. Let us try to ļ¬nd complex exponential solutions of the Laplace equation Īu(x, y) = 0, that is, solutions of the form u(x, y) = ekx+ly ,
6.4. Hadamardās example
135
where k, l ā C. Substituting this function into the Laplace equation, we immediately see that it is a solution if and only if k 2 + l2 = 0. For example, we can take k = iĻ, l = Ļ for a real Ļ, to obtain uĻ (x, y) = eiĻx eĻy . We can also take the real part of this solution, which is vĻ (x, y) = eĻy cos Ļx, to obtain a real-valued solution of Īu = 0. Let us try to understand whether we can use the C k -norm on functions Ļ ā C k (R), i.e., (6.13) Ļ(k) = sup |ā Ī± Ļ(x)|, |Ī±|ā¤k
xāR
for the initial data of the problem (6.11) and investigate whether a value of the solution u at a point can be made small if the C k -norms of the initial data are small. Equivalently, we can ask whether a small deviation of Ļ, Ļ from the equilibrium values Ļ ā” Ļ ā” 0 necessarily leads to small perturbation of the solution u. More precisely, ļ¬xing an arbitrarily large k and choosing, for example, a point (0, Ļ) ā Ī T , where Ļ > 0, we would like to know the answer to the following question: is it true that, for any Īµ > 0, there exists Ī“ > 0 such that the inequality Ļ(k) + Ļ(k) < Ī“ implies that |u(0, Ļ)| < Īµ? The answer to this question is emphatically negative. Indeed, for u = vĻ with Ļ > 1, we obtain Ļ(x) = vĻ (x, 0) = cos Ļx and Ļ(x) = āy vĻ (x, 0) = Ļ cos Ļx, so Ļ(k) + Ļ(k) ā¤ (k + 1)Ļ k + (k + 1)Ļ k+1 ā¤ 2(k + 1)Ļ k+1 . However, vĻ is much bigger near y = 0 if Ļ is large. In fact, if we ļ¬x Ļ > 0 and instead take the solution vĖĻ = eāĻĻ/2 vĻ with the corresponding initial data Ė Ļ(x) Ė = eāĻĻ/2 cos Ļx, Ļ(x) = eāĻĻ/2 Ļ cos Ļx, we see that vĖĻ (0, Ļ) = eĻĻ/2 grows exponentially as Ļ ā +ā, whereas ĻĻ (k) + ĻĻ (k) ā¤ 2(k + 1)Ļ k+1 eāĻĻ/2 decays exponentially as Ļ ā +ā. In particular, now the negative answer to the question becomes obvious: the value of u at a point cannot be controlled by the C k -norms of the initial data. As we have seen in Chapter 2, the situation is quite opposite for the two-dimensional wave equation: there the values of u are estimated by the sup-norms of the initial data. In fact, this holds for hyperbolic equations in higher dimensions too. We will see this for the wave equation in Chapter 10.
136
6. Harmonic functions
6.5. Greenās function for the Laplacian In Section 3.4 we introduced the Greenās function for the Sturm-Liouville operator L on [0, l] with vanishing boundary conditions at the ends 0 and l, and we used it to deļ¬ne the inverse operator Lā1 (cf. (3.21)). The properties found in Section 3.4 imply that G(x, Ī¾) is a distribution solution of the following boundary value problem involving the Dirac distribution Ī“(x ā Ī¾): Lx G(x, Ī¾) = Ī“(x ā Ī¾), x, Ī¾ ā (0, l), (6.14) G(0, Ī¾) = 0 = G(l, Ī¾), Ī¾ ā [0, l]. In the present section we deļ¬ne a similar multidimensional object G(x, Ī¾) for the Dirichlet problem (6.8) in a bounded open set Ī© ā Rn with a C 2 -smooth boundary. The Greenās function will be useful in solving the following boundary value problem, called the Poisson problem: Īu(x) = f (x), x ā Ī©, (6.15) u(x) = 0, x ā āĪ©. The Poisson problem is analogous to the Dirichlet problem (6.8) and is uniquely solvable in appropriate classes of functions and domains. For example, if Ī© is an open bounded domain such that āĪ© is a C 2 -hypersurface in ĀÆ then we will show in subsequent sections of this chapter Rn and f ā C 1 (Ī©), ĀÆ that we can solve (6.15) with u ā C 2 (Ī©) ā© C(Ī©). We use the fundamental solution En = En (x) of the Laplacian Ī in Rn . Recall that En is spherically symmetric (with the center at 0), and if n ā„ 3, then En (x) ā 0 as |x| ā ā. For an open bounded domain Ī©, we will construct the Greenās function G = G(x, y), for x, y ā Ī©, x = y, in the form (6.16)
G(x, y) := En (x ā y) + g(x, y),
where g = g(x, y) is a function for all x, y ā Ī© called the remainder. It is required to satisfy the following properties: i) For ļ¬xed y ā Ī©, Īx g(x, y) = 0 for all x ā Ī©. ĀÆ and g(x, y) = ii) For ļ¬xed y ā Ī©, g(x, y) extends continuously to x ā Ī© āEn (x ā y) for x ā āĪ©. This means that, for each y ā Ī©, uy (x) = G(x, y) is the solution of (6.15) with f (x) replaced by Ī“(x ā y); i.e., Īx G(x, y) = Ī“(x ā y), x, y ā Ī©, (6.17) G(x, y) = 0, x ā āĪ©, y ā Ī©.
6.5. Greenās function for the Laplacian
137
As follows from Corollary 6.10, if for some y ā Ī© such a distribution G(Ā·, y) exists, then it is unique. Concerning the existence, we have Lemma 6.13. Assume that, for any Ļ ā C(āĪ©), the Dirichlet problem (6.8) has a solution. Then the Greenās function G(x, y) for Ī© exists. Proof. Let us ļ¬x y ā Ī© and start with a particular solution u(x) of the equation Īu(x) = Ī“(x ā y), which is u(x) = En (x ā y), where En (x) is the fundamental solution of the Laplacian Ī, given by (4.38) and (4.39). Then the remainder (6.18)
g(x, y) := G(x, y) ā En (x ā y)
should be harmonic with respect to x and satisfy the boundary condition g(x, y) = āEn (x ā y) for x ā āĪ©. This means that g(Ā·, y) is a solution of the Dirichlet problem (6.8) with the boundary condition Ļ(x) = āEn (x ā y). Due to our assumption about the solvability of Dirichletās problem in Ī©, such a function g exists, which ends the proof. Remark 6.14. In fact, for the existence of the Greenās function G(x, y), we need to solve (6.8) not for all Ļ ā C(āĪ©), but only for speciļ¬c functions of the form Ļy = āEn (Ā· ā y)|āĪ© , for ļ¬xed y ā Ī©. The preceeding remark allows that the Greenās function may have additional regularity near āĪ© that is important for obtaining certain properties. Hence we usually impose the following condition on Ī©: Condition G. Ī© is a bounded domain with C 1 boundary and there exists a (unique) Greenās function G(x, y) whose remainder function g(x, y) has the ĀÆ property that for any y ā Ī©, the function uy (x) = g(x, y) is in C 2 (Ī©). Remark 6.15. The signiļ¬cance of Condition G is that it allows us to apply ĀÆ For Greenās formula (see (6.19) below) to uy and an arbitrary v ā C 2 (Ī©). 1 2 ĀÆ by this reason, we may relax the conditions āĪ© ā C and uy ā C (Ī©) 1 2 1 ĀÆ allowing āĪ© to be piecewise C and uy ā C (Ī©) ā© C (Ī©). In the next section we will give a regularity condition on āĪ© that guarantees that Condition G holds. We now obtain the ļ¬rst result that follows from Condition G. Proposition 6.16. Let Ī© be a domain that satisļ¬es Condition G. Then G is symmetric, that is, G(x, y) = G(y, x) for all x, y ā Ī©, x = y. Proof. Let us recall Greenās formula (4.32):
āu āv āv dS, (uĪv ā vĪu)dx = u (6.19) ān ĀÆ ān ĀÆ Ī© āĪ©
138
6. Harmonic functions
ĀÆ Let us try to apply this to the functions u = G(Ā·, y) where u, v ā C 2 (Ī©). and v = G(Ā·, z), for ļ¬xed y, z ā Ī©, y = z. Due to the boundary conditions on G, we have u|āĪ© = v|āĪ© = 0; hence the right-hand side vanishes. We also have Īu = Ī“(Ā· ā y), Īv = Ī“(Ā· ā z), so the left-hand side equals u(z) ā v(y) = G(z, y) ā G(y, z) if the integrals there are understood in the sense of distributions. Therefore, if Greenās formula can be extended to such u, v, then we will get the desired symmetry, G(z, y) = G(y, z). Now the only reason that (6.19) cannot be directly applied to our u and v is that they have singularities at the points y and z, respectively. So a ĀÆ where natural approach is to approximate u, v by functions uĪµ , vĪµ ā C 2 (Ī©), Īµ ā (0, Īµ0 ), so that the desired formula results in the limit as Īµ ā 0. To this end, let us take an arbitrary function Ļ ā C ā (Rn ), such that Ļ = 0 in B1/2 (0), Ļ = 1 in Rn \B1 (0), 0 ā¤ Ļ(x) ā¤ 1 for all x ā Rn . Then take ĻĪµ (x) = Ļ(Īµā1 x). Finally, deļ¬ne uĪµ (x) = ĻĪµ (x ā y)u(x) and vĪµ (x) = ĻĪµ (x ā z)u(x). ĀÆĪµ (y) and B ĀÆĪµ (z) are disjoint We choose Īµ0 > 0 so that the closed balls B 0 0 subsets in Ī© (that is, Īµ0 < |y ā z|/2, and Īµ0 is strictly less than the distances from y and z to āĪ©). Clearly, |uĪµ (x)| ā¤ |u(x)| and |vĪµ (x)| ā¤ |v(x)| for all ĀÆ \ BĪµ (y), and vĪµ = v in Ī© ĀÆ \ BĪµ (z). In particular, ĀÆ Also, uĪµ = u in Ī© x ā Ī©. these equalities hold in a neighborhood of āĪ©. Note that u and v are locally integrable near their singularities, hence integrable in Ī©. By the Lebesgue dominated convergence theorem uĪµ ā u, vĪµ ā v, as Īµ ā 0, in L1 (Ī©), hence, in D (Ī©). It follows that ĪuĪµ ā Īu and ĪvĪµ ā Īv in D (Ī©). Since uĪµ = u ā C ā in a neighborhood of supp ĪvĪµ (which is a subset of BĪµ0 (z)), we have
uĪµ ĪvĪµ dx = ĪvĪµ , u ā Īv, u = Ī“(Ā· ā z), u = u(z) = G(z, y). ĀÆ Ī©
Here the angle brackets Ā·, Ā· stand for the duality between E (Ī©1 ) and C ā (Ī©1 ), where Ī©1 is an appropriate open subset in Ī©. Similarly, we obtain
vĪµ ĪuĪµ dx = ĪuĪµ , v ā Īu, v = Ī“(Ā· ā y), v = v(y) = G(y, z). ĀÆ Ī©
We conclude G(z, y) = G(y, z), as desired.
Remark 6.17. Under Condition G, the remainder function g(x, y) is also symmetric; i.e., g(x, y) = g(y, x) for all x, y ā Ī©. Remark 6.18. The symmetry of the Greenās function can be interpreted as the selfadjointness of an operator in L2 (Ī©) which is inverse to the Laplacian with Dirichletās boundary condition. The boundary condition deļ¬nes an appropriate domain of the Laplacian, and Greenās function becomes the Schwartz kernel of this operator. We will discuss this approach in later
6.5. Greenās function for the Laplacian
139
chapters. It requires the powerful technique of Sobolev spaces, which will be discussed in Chapter 8. It is noticeable that this technique does not require any restrictions on Ī© except its boundedness. Now we show that the Greenās function on a domain Ī© satisfying Condition G can be used to obtain an integral representation for any C 2 -function ĀÆ let vy (x) = G(Ā·, y) for ļ¬xed ĀÆ Starting with an arbitrary u(x) ā C 2 (Ī©), on Ī©. y ā Ī©. Then Īvy (x) = Īx G(x, y) = Ī“(x ā y) and it is clear from the argument in the proof of Proposition 6.16 that we may plug u, v into (6.19) to obtain
āG(x, y) G(x, y)Īu(x)dx = u(x) dSx u(y) ā ān ĀÆx Ī© āĪ© (where we have used G(x, y) = 0 for x ā āĪ©). Interchanging the roles of x and y and using the symmetry of G(x, y), we have proved the following: Proposition 6.19. Assume that Ī© satisļ¬es Condition G. Then for every ĀÆ the following integral representation holds: u ā C 2 (Ī©)
āG(x, y) G(x, y)Īu(y)dy + u(y)dSy , x ā Ī©. (6.20) u(x) = ān ĀÆy Ī© āĪ© ĀÆ is the (unique) solution of the Dirichlet problem In particular, if u ā C 2 (Ī©) (6.10), then
āG(x, y) G(x, y)f (y)dy + Ļ(y)dSy , x ā Ī©. (6.21) u(x) = ān ĀÆy Ī© āĪ© This proposition gives an explicit representation of the solution for the Dirichlet problem in terms of the Greenās function. Two important particular cases are obtained when Ļ ā” 0 or f ā” 0: if Ļ ā” 0, then
G(x, y)f (y)dy, x ā Ī©; (6.22) u(x) = Ī©
if f ā” 0, then (6.23)
u(x) = āĪ©
āG(x, y) Ļ(y)dSy , x ā Ī©. ān ĀÆy
The equation (6.23) is called the Poisson integral formula. ĀÆ be harmonic in Corollary 6.20. Let Ī© satisfy Condition G and u ā C 2 (Ī©) Ī©. Then u(x) for x ā Ī© is given by (6.23), where Ļ = u|āĪ© . Remark 6.21. Formula (6.21) gives the solution of the Dirichlet boundary value problem for the Poisson equation (6.10) if such a solution exists; but it may happen that such solution does not exist! Of course, similar statements apply to the formula (6.22) with regard to (6.15), and (6.23) with regard to (6.8). Note also that the right-hand side of the formula (6.23) is nothing else but the double layer potential (see Chapter 5).
140
6. Harmonic functions
Let us end this section with a discussion of the physical meaning in electrostatics of the formulas (6.21)ā(6.23). (For the simplest notions of electrostatics, see physics textbooks, e.g., [8, Vol. 2, Chapters 4ā8] and also Chapters 4 and 5 in this book.) The situation to describe is a unit charge ĀÆ which is bounded by a thin closed conductive shell āĪ©. sitting in a body Ī©, Let us explain what these words mean mathematically. We are interested in the electric ļ¬eld in Ī© which contains only a single unit charge at y. The fact that āĪ© is a thin shell made of a conductive material (e.g., a metal) means that no energy is needed to transport a small charge along āĪ©. Since the energy needed to transport a small charge along a curve is equal to the diļ¬erence in the potentials at the ends of the curve, this means that the potential u is locally constant on the boundary āĪ©; i.e., it is constant on each of the connected components of āĪ©. If āĪ© is connected, then u is constant on āĪ©, and by adding a constant to u we can achieve the boundary condition u|āĪ© = 0. Now let us consider Greenās function G(x, y) for Ī©. For ļ¬xed y ā Ī©, ĀÆ of a unit charge u(x) = G(x, y) is interpreted as the potential at x ā Ī© located at y. The boundary condition G(x, y) = 0 for x ā āĪ© reļ¬ects the fact that Ī© is bounded by a thin conductive shell āĪ©. For a system of charges in Ī©, the potentials add up by the superposition principle, and for a distributed charge f (x), the potential of this charge becomes an integral, as in (6.22). If the boundary is not connected, then the condition that the boundary is conductive means that the potential u is constant on each of the connected components of the boundary. By adding an appropriate solution of the Dirichlet problem (6.8), we can force these constants to vanish; hence we obtain an adjusted potential u Ė which again satisļ¬es the boundary condition Ė(x), we again get the Greenās function which u Ė|āĪ© = 0. If we let G(x, y) = u satisļ¬es G(x, y) = 0 for x ā āĪ©.
6.6. HĀØ older regularity Now we show that a bounded domain Ī© satisļ¬es Condition G when āĪ© is suļ¬ciently regular. This boundary regularity can be expressed more precisely using HĀØ older spaces, which also allow a more precise description of the smoothness of solutions of the Dirichlet problem for the Poisson equation (6.10) in terms of the smoothness of f and Ļ. ĀÆ k = 0, 1, 2, . . . , introduce the norm First, for functions u ā C k (Ī©), (6.24)
u(k) =
|Ī±|ā¤k
sup |ā Ī± u(x)|, xāĪ©
6.6. HĀØolder regularity
141
ĀÆ and the right-hand side makes sense (i.e., if it deļ¬ned whenever u ā C k (Ī©) is ļ¬nite). (We already used this norm on R: see (6.13).) With this norm, ĀÆ becomes a Banach space. C k (Ī©) ĀÆ introduce a seminorm Now for any Ī³ ā (0, 1) and any function u ā C(Ī©) |u|(Ī³) =
sup x,yāĪ©, x=y
|u(x) ā u(y)| . |x ā y|Ī³
If this seminorm is ļ¬nite, then we will say that this function is a HĀØ older function with the exponent Ī³. (If the same is true with Ī³ = 1, then one says that u is a Lipschitz function, but we will rarely use Lipschitz functions in this book.) We will denote the space of all HĀØ older functions with the Ī³ ĀÆ ( Ī©). Note that 0 < Ī³ < Ī³ < 1 implies exponent Ī³ by C 1 2 ĀÆ = C 0 (Ī©) ĀÆ ā C Ī³1 (Ī©) ĀÆ ā C Ī³2 (Ī©) ĀÆ ā C 1 (Ī©). ĀÆ C(Ī©) ĀÆ Similarly, for any integer k ā„ 0 and Ī³ ā (0, 1) deļ¬ne the space C k,Ī³ (Ī©), k Ī± Ī³ ĀÆ such that ā u ā C (Ī©) ĀÆ for all multiconsisting of functions u ā C (Ī©) indices Ī± with |Ī±| = k. It is a Banach space with the norm
uk,Ī³ = u(k) +
|ā Ī± u|(Ī³) .
{Ī±: |Ī±|=k}
ĀÆ ā C k,Ī³ (Ī©) ĀÆ ā C k+1 (Ī©). ĀÆ Note that C k (Ī©) The C k,Ī³ -norm can be used to measure the smoothness of the boundary as well. Namely, we will say that Ī© has a C k,Ī³ boundary and write āĪ© ā C k,Ī³ if āĪ© ā C k and the functions h(x ), which enter into local representations of the boundary as the graph xn = h(x ) (see Deļ¬nition 4.8), are in fact in C k,Ī³ on every compact subset of the domain where they are deļ¬ned. ĀÆ and the C k,Ī³ If k ā„ 1, then it is easy to see that the space C k,Ī³ (Ī©) k,Ī³ smoothness of āĪ© are invariant under C diļ¬eomorphisms. (This follows from the statements formulated in Problem 6.6.) In fact, on the C k,Ī³ boundary āĪ© itself, for k ā„ 1, there exists a structure of a C k,Ī³ manifold, which is given by the sets of local coordinates related to each other in the usual way, by C k,Ī³ diļ¬eomorphisms between two domains in Rnā1 , which are images ĀÆ Such local of the intersection of the two coordinate neighborhoods in Ī©. coordinates can be taken to be x for the point (x , h(x )) in the local graph presentation of āĪ©. Let us return to (6.10), the Dirichlet problem for the Poisson equation. Here is a remarkable theorem by J. Schauder, which gives precise solvability and regularity for the problem (6.10) in terms of the HĀØ older spaces C k,Ī³ .
142
6. Harmonic functions
Theorem 6.22. Assume that k ā„ 0 is an integer, Ī³ ā (0, 1), āĪ© ā C k+2,Ī³ , ĀÆ and Ļ ā C k+2,Ī³ (āĪ©). f ā C k,Ī³ (Ī©), ĀÆ of the (a) (Existence and uniqueness) A solution u ā C 2 (Ī©) ā© C(Ī©) boundary value problem (6.10) exists and is unique. ĀÆ (b) (Regularity) In fact, the unique solution u belongs to C k+2,Ī³ (Ī©). (c) (Isomorphism operator) The operator ĀÆ ā C k,Ī³ (Ī©) ĀÆ Ć C k+2,Ī³ (āĪ©), (6.25) A : C k+2,Ī³ (Ī©)
Au = {Īu, u|āĪ© },
is a linear topological isomorphism (that is, a bounded linear operator which has a bounded inverse). ĀÆ and Ļ ā C ā (āĪ©). Corollary 6.23. Assume that āĪ© ā C ā , f ā C ā (Ī©), 2 ĀÆ ĀÆ Then (6.10) has a unique solution u ā C (Ī©)ā©C(Ī©) and in fact u ā C ā (Ī©). Moreover, the operator ĀÆ ā C ā (Ī©) ĀÆ Ć C ā (āĪ©), Au = {Īu, u|āĪ© }, (6.26) A : C ā (Ī©) is a linear topological isomorphism of FrĀ“echet spaces. Proof. The statement immediately follows from Theorem 6.22 if we note ĀÆ is the intersection of all spaces that for any ļ¬xed Ī³ ā (0, 1), the space C ā (Ī©) k,Ī³ ĀÆ can be deļ¬ned C (Ī©), k = 0, 1, 2, . . . , and the FrĀ“echet topology in C ā (Ī©) by norms Ā· k,Ī³ . Remark 6.24. I will not provide a proof of Theorem 6.22. The interested reader can ļ¬nd complete proofs in the monographs by Ladyzhenskaya and Uralātseva [19, Ch. III] and Gilbarg and Trudinger [9, Ch. 6]. In both books a substantial extension of this result to general second-order elliptic operators is proved. (It is also due to Schauder.) ĀÆ All three statements of Theorem 6.22 fail in the usual classes C k (Ī©) even locally. Namely, the equation Īu = f with f ā C(Ī©) may fail to have a solution u ā C 2 (Ī©1 ) where Ī©1 is a proper subdomain of Ī©. Even ĀÆ does not imply u ā C 2 (Ī©). However, it is possible to deduce f ā C(Ī©) weaker existence and regularity statements in terms of the usual C k spaces. Here is the regularity result: Corollary 6.25. Assume that k ā„ 1 is an integer and Ī© is a bounded ĀÆ be a solution domain in Rn such that āĪ© ā C k+2 . Let u ā C 2 (Ī©) ā© C(Ī©) k k+2 ĀÆ (āĪ©). Then u ā of (6.10) with f = Īu ā C (Ī©) and Ļ = u|āĪ© ā C ĀÆ for all Ī³ ā (0, 1). In particular, if āĪ© ā C k+2 , u ā C(Ī©) ĀÆ is C k+1,Ī³ (Ī©) k+2 k+1 ĀÆ (āĪ©), then u ā C (Ī©). harmonic in Ī©, and Ļ = u|āĪ© ā C Proof. Choose Ī³ ā (0, 1), and note that the given conditions imply the inĀÆ and Ļ ā C k+2,Ī³ (āĪ©). Then Theorem clusions āĪ© ā C k+1,Ī³ , f ā C kā1,Ī³ (Ī©), ĀÆ hence u ā C k+1 (Ī©). ĀÆ 6.22 implies that u ā C k+1,Ī³ (Ī©);
6.7. Explicit formulas for Greenās functions
143
For the existence result for C k spaces, see Problem 6.8. Now let us return to the existence and smoothness of the Greenās function for a given domain, i.e., Condition G. Theorem 6.26. If Ī© is a bounded domain in Rn with āĪ© ā C 2,Ī³ for some Ī³ ā (0, 1), then Ī© satisļ¬es Condition G. Consequently, the following statements hold: (a) The Greenās function satisļ¬es G(x, y) = G(y, x) for all x, y ā Ī©, x = y. ĀÆ Ć Ī©). ĀÆ (b) The remainder function satisļ¬es g(x, y) ā C 2 (Ī© ĀÆ Ć Ī©\diag( ĀÆ ĀÆ Ī©)). ĀÆ (c) The Greenās function satisļ¬es G ā C 2 (Ī© Ī©, Proof. For any ļ¬xed y ā Ī©, observe that Ļ(x) := āEn (x ā y) is a smooth function on āĪ©, so by Theorem 6.22 we can ļ¬nd g = g(x, y) such that ĀÆ ā C 2 (Ī©). ĀÆ This shows that Ī© satisļ¬es Condition G, and g(Ā·, y) ā C 2,Ī³ (Ī©) hence property (a) holds by Proposition 6.16. But (a) means that regularity in x implies regularity in y, so properties (b) and (c) follow as well.
6.7. Explicit formulas for Greenās functions The cases where the Greenās function can be found explicitly are rare. Nevertheless, they are worth learning because they show what you can expect in the general case. We will construct Greenās functions using the method of images (sometimes called the method of reļ¬ections). We are usually interested in the Greenās function for a bounded domain, but let us ļ¬rst consider the case of a half-space, since the method of images is simplest there. In fact, we assume Ī© is the upper half-space: (6.27)
Ī© = Rn+ := {x = (x1 , . . . , xn ) : xn > 0}.
To construct the Greenās function, we recall (6.16) to write G(x, y) = En (x ā y) + g(x, y) for x, y ā Rn+ , x = y, where g(x, y) is harmonic in x ā Rn+ and satisļ¬es g(x, y) = āEn (x ā y) for x ā ā Rn+ = Rnā1 . But such a function is g(x, y) := āEn (x ā y ā ), where (6.28)
y ā = (y1 , . . . , ynā1 , āyn )
is the image of y = (y1 , . . . , yn ) as reļ¬ected through the boundary Rnā1 . Notice that g(x, y) is harmonic in x ā Rn+ since y ā ā Rn+ , and g(x, y) = āE(x ā y) for x ā ā Rn+ since then |x ā y| = |x ā y ā |. We have: The Greenās function for Rn+ : (6.29)
G(x, y) = En (x ā y) ā En (x ā y ā )
for x, y ā Rn+ , x = y.
144
6. Harmonic functions
Before we turn to constructing the Greenās function on other domains, let us observe that for Ī© = Rn+ the correction term g(x, y) = āEn (x ā y ā ) is just the potential with charge ā1 at the point y ā , which lies outside of Rn+ . Let us use this idea for a bounded domain Ī©; i.e., for each y ā Ī© let (6.30)
G(x, y) = En (x ā y) + q ā (y)En (x ā y ā (y)) for x, y ā Ī©, x = y,
ĀÆ is its where q ā = q ā (y) ā R is the correcting charge and y ā = y ā (y) ā Rn \Ī© location. For this to be the Greenās function for Ī©, we need (6.31)
En (x ā y) + q ā (y)En (x ā y ā (y)) = 0
for x ā āĪ©, y ā Ī©.
Thus we can calculate G(x, y) by ļ¬nding y ā (y) and q ā (y) so that (6.31) holds. We will see that such functions exist only for very special domains Ī©, which we will completely describe (for n ā„ 3). Now assume n ā„ 3. The explicit form of En allows us to rewrite (6.31) in the form (6.32)
|x ā y|2ān + q ā |x ā y ā |2ān = 0,
ĀÆ x ā āĪ©, y ā Ī©, y ā ā Rn \Ī©.
We can equivalently write (since x = y) (6.33)
|x ā y ā |2 = (āq ā )2/(nā2) , |x ā y|2
ĀÆ x ā āĪ©, y ā Ī©, y ā ā Rn \Ī©,
which implies, in particular, that the left-hand side is independent of x ā āĪ©. (Note that (6.32) implies q ā < 0.) If we denote Qā > 0 to be (6.34)
Qā = Qā (q ā ) = (āq ā )ā2/(nā2),
then the equation in (6.33) is equivalent to the quadratic equation (6.35)
|x ā y|2 ā Qā |x ā y ā |2 = 0.
Proposition 6.27. Assume n ā„ 3. For ļ¬xed y, y ā ā Rn with y = y ā and Qā > 0, the set of points x ā Rn satisfying (6.35) is a) a hyperplane if Qā = 1 and b) a geometric sphere if Qā = 1. Proof. Denoting the standard scalar product of vectors x, y ā Rn by x Ā· y, (6.35) implies 0 =Qā (|x|2 ā 2x Ā· y ā + |y ā |2 ) ā (|x|2 ā 2x Ā· y + |y|2 ) =(Qā ā 1)|x|2 ā 2x Ā· (Qā y ā ā y) + (Qā |y ā |2 ā |y|2 ). If Qā = 1, we obtain ā2x Ā· (y ā ā y) + (y ā + y) Ā· (y ā ā y) = 0, or yā + y Ā· (y ā ā y) = 0, (6.36) xā 2
6.7. Explicit formulas for Greenās functions
145
which is the equation of the hyperplane passing through the point (y ā +y 2 )/2 and perpendicular to the vector y ā āy. If Qā = 1, then completing the square gives ' & 2x Ā· (Qā y ā ā y) |Qā y ā ā y|2 ā 2 0 =(Q ā 1) |x| ā + Qā ā 1 (Qā ā 1)2 |Qā y ā ā y|2 ā (Qā |y ā |2 ā |y|2 )(Qā ā 1) Qā ā 1 Qā y ā ā y 2 Qā ā |y ā ā y|2 , ā =(Q ā 1) x ā Qā ā 1 Qā ā 1 ā
or, equivalently, (6.37)
2 ā ā Qā x ā Q y ā y = |y ā ā y|2 . Qā ā 1 (Qā ā 1)2
This means that our quadric is a sphere SR (c) with center c ā Rn and radius R > 0 given by the formulas (6.38)
c=
Qā y ā ā y , Qā ā 1
R=
(Qā )1/2 ā |y ā y|. |Qā ā 1|
Since the equation (6.35) was obtained by ļ¬nding where G(x, y) vanishes, we see that constructing a Greenās function in the form of (6.30) will only work when Ī© is a hyperplane or a sphere. Moreover, we can obtain additional conclusions about the possible relationships between y and y ā in these cases. Solving the ļ¬rst equation in (6.38) with respect to y ā and y, we get 1 1 ā (6.39) y = ā y + 1 ā ā c, y = Qā y ā + (1 ā Qā )c. Q Q It follows that c, y, and y ā are all on the same line in Rn . In fact, if Qā ā1 > 0 (respectively, Qā ā1 < 0) , then they follow in the order c, y ā , y (respectively, c, y, y ā ). Substituting the expression for y ā from (6.39) into (6.38) and then solving the resulting equation with respect to Qā , we easily ļ¬nd that |y ā c|2 . R2 Now, using the ļ¬rst equation of (6.38), we obtain
(6.40)
(6.41)
Qā =
yā ā c =
yāc R2 Ā· , |y ā c| |y ā c|
which implies that y ā ā c and y ā c are proportional and |y ā c| Ā· |y ā ā c| = R2 . If (6.41) is considered as a map ā : Rn \{c} ā Rn \{c},
given by y ā y ā ,
146
6. Harmonic functions
then it is involutive or an involution, which means that ā2 = Id (the identity map), with SR (c) being its ļ¬xed point set. The particular involution ā : y ā y ā of Rn \{c} given by (6.41) is called the reļ¬ection or inversion with respect to the sphere SR (c). Let us summarize the arguments above in the following theorem. Theorem 6.28. Let n ā„ 3 and let Ī© be a bounded domain with a C ā boundary āĪ© in Rn . Assume that for every y ā Ī© and every q > 0 there ĀÆ and a number q ā < 0, such that the potential u of exist a point y ā ā Rn \Ī© the system of two charges 1, q ā at the points y, y ā , respectively, has zero level set uā1 (0) = {x ā Rn : u(x) = 0}, which coincides with āĪ©. Then the following statements hold: (a) Ī© is a ball BR (c) with the center c ā Ī© and radius R > 0. (b) The quantities R, c are unique and are given by the formulas (6.38), where Qā is given by (6.34). (c) The points c, y, y ā all lie on a straight line and satisfy (6.39). Making the calculations in the opposite direction, we can conclude that the following proposition holds: Proposition 6.29. Let n ā„ 3. For every sphere SR (c) and every point y ā Rn \SR (c), there exist a unique point y ā ā Rn and a unique number q ā ā R \{0}, such that the potential u of the system of two charged point particles at y, y ā with the charges 1, q ā , respectively, has zero level set which coincides with the sphere SR (c). Remark 6.30. In the limit R ā +ā, with the appropriate choice of the center c = c(R), we can obtain a limit relation SR (c(R)) ā Sā , where Sā is a hyperplane in Rn and the convergence is in the C 2 -topology of hypersurfaces (deļ¬ned by C 2 -topologies on functions representing Sā as graphs of C 2 -functions in local coordinates). Then, for any ļ¬xed y we get ā , where y ā is the mirror the reļ¬ection y ā = y ā (R), such that y ā (R) ā yā ā image of y with respect to the limit hyperplane Sā . (See more about this below in this section.) Since we now know that the method of images will succeed in producing the Greenās function for an open ball Ī© = BR (0) in the form (6.42)
G(x, y) = En (x ā y) + q ā (y)En (x ā y ā (y)),
let us describe this formula explicitly. Note that the formulas we obtain work for n = 2 as well as n ā„ 3, and we have taken the center of the ball to
6.8. Problems
147
be c = 0 for convenience. In this case, the involution (6.41) is just yā =
R2 y , |y|2
for y ā Rn \{0} and extended to the points y = 0 and y = ā by continuity. Combining (6.33), (6.34), and (6.40), we obtain R |x ā y ā | = |x ā y| |y|
for |x| = R, |y| < R,
which we can easily check holds also for n = 2. So for |x| = R, |y| < R we have
ā§ āØ 1 log |y| |x ā y ā | for n = 2, 2Ļ R
nā2 En (x ā y) = ā© R En (x ā y ā ) for n ā„ 3. |y| But the expressions on the right are harmonic for all |x| < R, so we obtain: The Greenās function for the ball BR (0):
ā§ āØE2 (x ā y) ā 1 log |y| |x ā y ā | for n = 2, 2Ļ R
nā2 (6.43) G(x, y) = ā©En (x ā y) ā R En (x ā y ā ) for n ā„ 3. |y| Notice that for n = 2, G(x, y) is not quite of the form (6.30) but rather G(x, y) = E2 (x ā y) ā E2 (|x ā y ā |) + p(y) where p(y) = ā(2Ļ)ā1 log(|y|/R).
6.8. Problems 6.1. Prove the statements inverse to the ones of Theorems 6.1, 6.2: if a C 2 -function u in an open set Ī© ā Rn has mean value property for all balls ĀÆ ā Ī© or for all spheres which are boundaries of such balls (that B with B is, either (6.2) holds for such spheres or (6.4) holds for all such balls), then u is harmonic in Ī©. Moreover, it suļ¬ces to take balls or spheres, centered at every point x with suļ¬ciently small radii r ā (0, Īµ(x)), where Īµ(x) > 0 continuously depends upon x ā Ī©. 6.2. Let u be a real-valued C 2 -function on an open ball B = Br (x) ā Rn , ĀÆ and let Īu ā„ 0 which extends to a continuous function on the closure B, everywhere in B. Prove that
1 1 u(y)dSy = u(y)dSy (6.44) u(x) ā¤ volnā1 (S) S Ļnā1 rnā1 Sr (x) and (6.45)
1 u(x) ā¤ vol(B)
u(y)dy = B
n
Ļnā1 rn
where the notation is the same as in Section 6.1.
u(y)dy, Br (x)
148
6. Harmonic functions
6.3. Prove the inverse statement to the one in the previous problem: if ĀÆ ā Ī© (or for all spheres u ā C 2 (Ī©) is real valued and for all balls B with B which are boundaries of such balls), (6.45) (respectively, (6.44)) holds, then Īu ā„ 0 in Ī©. Moreover, it suļ¬ces to take balls or spheres, centered at every point x with suļ¬ciently small radii r ā (0, Īµ(x)), where Īµ(x) > 0 continuously depends upon x ā Ī©. Remark 6.31. A real-valued function u ā C 2 (Ī©) is called subharmonic if Īu ā„ 0 in Ī©. Note, however, that the general notion of subharmonicity does not require that u ā C 2 and allows more general functions u (or even distributions, in which case Īu ā„ 0 means that Īu is a positive Radon measure). 6.4. Let u be a real-valued subharmonic function in a connected open set Ī© ā Rn . Prove that u cannot have local maxima in Ī©, unless it is constant. In other words, if x0 ā Ī© and u(x0 ) ā„ u(x) for all x in a neighborhood of x0 , then u(x) = u(x0 ) for all x ā Ī©. 6.5. Consider the initial value problem (6.11) with Ļ ā C0ā (R) and Ļ ā” 0. Assume that it has a solution u. Prove that in this case Ļ ā” 0 on R and u ā” 0 on Ī T . 6.6. Let U, V, W be open sets in Rm , Rn , Rp , respectively, and let g : U ā V , f : V ā W be maps which are locally C k,Ī³ (that is, they belong to C k,Ī³ on any relatively compact subdomain of the corresponding domain of deļ¬nition). (a) Assume that k ā„ 1. Then prove that the composition f ā¦ g : U ā W is also locally C k,Ī³ . (b) Prove that the statement (a) above does not hold for k = 0. (c) Prove that if k ā„ 1, Ī© is a domain in Rn with āĪ© ā C k,Ī³ , U is a ĀÆ in Rn , and Ļ : U ā Rn is a C k,Ī³ diļ¬eomorphism of U neighborhood of Ī© onto a domain V ā Rn , then the image Ī© = Ļ(Ī©) is a domain with a C k,Ī³ boundary, Ī© ā V . 6.7. (a) Prove that if Ī© is a bounded domain in Rn and āĪ© ā C k,Ī³ with k ā„ 1, then āĪ© is a C k,Ī³ manifold, dimR āĪ© = n ā 1. (b) Prove that for any nonnegative integers k, k and numbers
Ī³, Ī³ ā (0, 1), such that k +Ī³ ā„ k +Ī³ , k ā„ 1, the class C k ,Ī³ is locally invari ant under C k,Ī³ diļ¬eomorphisms. In particular, function spaces C k ,Ī³ (āĪ©) are well-deļ¬ned. (c) Under the same assumptions as in (a) and (b), for a real-valued function Ļ : āĪ© ā R, the inclusion Ļ ā C k ,Ī³ (āĪ©) is equivalent to the Ė āĪ© = Ļ. existence of an extension ĻĖ ā C k ,Ī³ (Ī©), such that Ļ|
6.8. Problems
149
6.8. Assume that k ā„ 1 is an integer, Ī© is a bounded domain in Rn , such ĀÆ and Ļ ā C k+2 (āĪ©). Using Schauderās theorem, that āĪ© ā C k+2 , f ā C k (Ī©), ĀÆ Theorem 6.22, prove that the problem (6.10) has a solution u ā C k+1 (Ī©). 6.9. Assume that āĪ© ā C 2 and there exists Greenās function G = G(x, y) of ĀÆ Consider an integral representation Ī©, such that G ā C 2 for x = y, x, y ā Ī©. 2 ĀÆ ĀÆ and of a function u ā C (Ī©) given by the formula (6.21), with f ā C(Ī©) Ļ ā C(āĪ©). Prove that the pair {f, Ļ}, representing a ļ¬xed function u ā ĀÆ by (6.21), is unique. C 2 (Ī©) 6.10. Provide detailed proofs of identities (6.39), (6.40), and (6.41). 6.11. Deduce the expression for the Greenās function in a half-space of Rn , n ā„ 3, from symmetry considerations.
10.1090/gsm/205/07
Chapter 7
The heat equation
7.1. Physical meaning of the heat equation The heat equation is (7.1)
āu = a2 Īu, āt
where u = u(t, x), t ā R, x ā Rn , a > 0, and Ī is the Laplace operator with respect to x. This equation is an example of a parabolic type equation. It describes, for instance, the temperature distribution of a homogeneous and isotropic medium (in which case u(t, x) is the temperature at the point x at the time t). This equation is also satisļ¬ed by the density of a diļ¬using matter, e.g., the density of Brownian particles when there are suļ¬ciently many of them so that we can talk about their density and consider the evolution of this density as a continuous process. This is a reason why the equation (7.1) is often called the diļ¬usion equation. Deriving of the heat equation is based on a consideration of the energy balance. More precisely, we will deal with the transmission of the heat energy, which is the kinetic energy of the atomsā and moleculesā chaotic motion in the gases and liquids or the energy of the atomsā vibrations in metals, alloys, and other solid materials. Instead of āheat energyā we will use the word āheatā, which is shorter and historically customary in the books on thermodynamics. The heat equation for temperature u(t, x) is derived from the following natural empirically veriļ¬ed physical assumptions: 1) The heat Q needed to warm up a part of matter with the mass m from the temperature u1 to u2 is proportional to m and u2 ā u1 : Q = cm(u2 ā u1 ). 151
152
7. The heat equation
The coeļ¬cient c is called the speciļ¬c heat (per unit of mass and unit of temperature). 2) Fourierās law: The quantity of heat ĪQ passing through a small surface element with the area S during the time Īt is proportional to S, Īt, and āāunĀÆ (the rate of growth of u in the direction n ĀÆ of the unit normal vector to the surface). More precisely, āu Īt, ān ĀÆ where k > 0 is the heat conductivity coeļ¬cient characterizing heat properties of the media; the minus sign indicates that heat is moving in the direction opposite to that of the growth of the temperature. ĪQ = ākS
One should keep in mind that both hypotheses are only approximations, and so is the equation (7.1) which they imply. As we will see below, a consequence of equation (7.1) is an inļ¬nite velocity for propagation of heat, which is physically absurd. In the majority of applied problems, however, these hypotheses and equation (7.1) are suļ¬ciently justiļ¬ed. A more precise model of heat distribution should take into account the molecular structure of the matter leading to problems of statistical physics far more diļ¬cult than the solution of the model equation (7.1). Note, however, that, regardless of its physical meaning, equation (7.1) and its analogues play an important role in mathematics. They are used, e.g., in the study of elliptic equations. Let us derive the heat equation (for n = 3). We use the energy conservation law (heat conservation in our case) in a volume Ī© ā R3 . Let us assume that Ī© has a smooth boundary āĪ©. The rate of change of the heat energy of the matter in Ī© is obviously equal to
d dQ = cĻu dV, dt dt Ī© where Ļ is the volume density of the matter (mass per unit volume) and dV is the volume element in R3 . Assuming that c = const and Ļ = const, we get
āu dQ = cĻ dV. dt Ī© āt Now, let P be the rate of the heat escape through āĪ©. Clearly,
āu k dS, P =ā ĀÆ āĪ© ā n where n ĀÆ is the outward unit normal to āĪ© , dS the area element of āĪ©. Here in the right-hand side we have the ļ¬ow of the vector āk grad u through āĪ©.
7.2. Boundary value problems
153
By the divergence formula (4.34) we get
P = ā div(k grad u)dV Ī©
or, for k = const,
P = āk
Īu dV. Ī©
The heat energy conservation law in the absence of heat sources means that dQ = āP. dt Substituting the above expressions for dQ dt and P , we get
āu cĻ dV = kĪu dV, āt Ī© Ī© which, since Ī© is arbitrary, results in (7.1) with a2 =
k cĻ .
7.2. Boundary value problems for the heat and Laplace equations A physical interpretation of the heat equation suggests a natural setting for boundary value problems for this equation. The Cauchy problem (sometimes also called the initial value problem) is the simplest one: āu 2 t > 0, x ā Rn , āt = a Īu, (7.2) u|t=0 = Ļ(x) x ā Rn . Its physical meaning is to determine the temperature of the medium at all moments t > 0 provided the temperature distribution is known at t = 0. Sometimes we are only interested in the temperature in a domain Ī© ā Rn regardless of what takes place outside of it. Then, on āĪ©, additional conditions should be imposed: e.g., the temperature or the heat ļ¬ow should be given. Thus we come to the two following initial-boundary value problems. The ļ¬rst initial-boundary value problem (or Cauchy-Dirichletās initialboundary value problem): ā§ āu t > 0, x ā Ī©, āØ āt = a2 Īu, (7.3) u|t=0 = Ļ(x), x ā Ī©, ā© u|āĪ© = Ī±(t, x), x ā āĪ©, t > 0. The second boundary value problem (or Cauchy-Neumannās boundary value problem) is obtained by replacing the latter of conditions (7.3) by āu = Ī²(t, x), x ā āĪ©, t > 0, ān ĀÆ āĪ©
where n ĀÆ is the outward unit normal to āĪ©.
154
7. The heat equation
For time-independent boundary conditions it is often interesting to determine temperature behavior as t ā +ā. Suppose that a limit v(x) of the solution u(t, x) of the heat equation exists as t ā +ā (we say that the solution āstabilizesā or ātends to an equilibriumā). Then it is natural to expect that v(x) satisļ¬es the Laplace equation Īv(x) = 0; i.e., v(x) is a harmonic in x, at least because the only possible ļ¬nite limit of āu(t, x)/āt can be 0. It is very easy to make the meaning of this statement precise but we will not do so at the moment in order not to divert our attention. The ļ¬rst and second initial-boundary value problems for the heat equation turn in the limit into the following boundary value problems for the Laplace equation: Dirichletās problem (7.4)
Īv(x) = 0, x ā Ī©, v|āĪ© = Ī±(x), x ā āĪ©.
Neumannās problem: Īv(x) x ā Ī©, = 0, (7.5) āv = Ī²(x), x ā āĪ©. ān ĀÆ āĪ© In this model Dirichletās problem means the following: ļ¬nd the steadystate temperature inside Ī© when the temperature Ī±(x) is maintained on the boundary. The meaning of Neumannās problem is to ļ¬nd the steady-state temperature inside Ī© if the heat ļ¬ow Ī²(x) on the boundary is given. We already mentioned the Dirichlet problem in the previous chapter; see Corollary 6.8, where the uniqueness of the classical solution was established. Clearly, the solution of Neumannās problem is not unique: any constant can always be added to the solution. Further, it is clear that the steady-state distribution of heat can exist in Ī© only if the total heat ļ¬ux through the boundary vanishes; i.e.,
Ī²(x)dS = 0. (7.6) āĪ©
(This also follows from the formula (6.3), which implies that if v is a solution of (7.5), then (7.6) holds.) We will see later that the above problems for the heat and Laplace equations are well-posed in the sense that they have a unique solution (up to adding a constant in the case of the Neumann problem) in natural classes of functions, and, besides, the solutions depend continuously on initial and boundary values for an appropriate choice of spaces. Note that the Dirichlet and Neumann problems for the Laplace equation have many physical interpretations apart from the above ones. For instance,
7.3. A proof that the limit function is harmonic
155
a small vertical deļ¬ection of a membrane u(t, x1 , x2 ) and a small displacement u(t, x1 , x2 , x3 ) of an elastic solid body satisfy the wave equation ā 2u = a2 Īu. āt2 In the time-independent case we come again to the Laplace equation and then the Dirichlet problem. For instance, for a membrane this is the problem of recovering the membrane form from that of its boundary. The boundary condition of Neumannās problem is interpreted as prescribing the boundary value of the vertical component of the force acting on the membrane along its boundary. This topic is related to the famous question by Mark Kac: āCan one hear the shape of a drum?ā (1968), which led to a wide development of spectral geometry.
7.3. A proof that the limit function is harmonic There are several ways to justify the harmonicity of the limit function. We will give one of them. Proposition 7.1. Let a solution u(t, x) of the heat equation be deļ¬ned and bounded for t > 0, x ā Ī© (for an open subset Ī© ā Rn ). Suppose that for almost all x ā Ī© there exists a limit v(x) = lim u(t, x).
(7.7)
tā+ā
Then v(x) is a harmonic function in Ī©. Remark 7.2. The solution u(t, x) may be a priori considered as a bounded measurable function (here the equation (7.1) should be understood in the distributional sense). However we will see below that the heat operator has a fundamental solution which is inļ¬nitely diļ¬erentiable for |t| + |x| = 0 and, therefore, actually, u(t, x) ā C ā (R+ Ć Ī©), where R+ = {t : t > 0}. Proof of Proposition 7.1. Clearly, lim u(t + Ļ, x) = v(x)
Ļ ā+ā
for all t > 0 and almost all x ā Ī©. If Ļ(t, x) ā D(R+ Ć Ī©), then the dominated convergence theorem implies
u(t + Ļ, x)Ļ(t, x)dtdx = v(x)Ļ(t, x)dtdx, (7.8) lim Ļ ā+ā
or, in other words, lim u(t + Ļ, x) = v(x) in D (R+ Ć Ī©)
Ļ ā+ā
(here v(x) should be understood as 1 ā v, i.e., as the right-hand side of ā ā a2 Īx to the equality above, we can (7.8)). Applying the heat operator āt
156
7. The heat equation
interchange this operator with the passage to the limit (this can be done in the space of distributions; for example, in our case everything is reduced ā ā a2 Īx )Ļ(t, x) to replacing Ļ(t, x) in (7.8) by another function (ā āt ā ā D(R+ Ć Ī©)). Then ( āt ā a2 Īx )v(x) = 0; i.e., v(x) is a distributional solution of Īv(x) = 0. But, as we know, v(x) is then a regular harmonic function, as required.
7.4. A solution of the Cauchy problem for the heat equation and Poissonās integral Let us investigate the Cauchy problem (7.2). First, assume that Ļ(x) ā S(Rn ) and u(t, x) ā S(Rnx ) for any t ā„ 0, so that u(t, x) is a continuous function of t for t ā„ 0 with values in S(Rnx ). In addition, assume that u(t), considered as a curve in S(Rn ), is diļ¬erentiable for t > 0. As we will see below, a solution u(t, x) with the described properties exists (and is unique in a very broad class of solutions). For the time being, note that we may perform the Fourier transform in x and since the Fourier transform is a topological isomorphism F : S(Rnx ) āā S(RnĪ¾ ), it commutes ā . Thus, let with āt
u Ė(t, Ī¾) = eāixĀ·Ī¾ u(t, x)dx (hereafter, integration is performed over the whole of Rn unless otherwise indicated).
The standard integration by parts gives '
& āu ā āu Ė 2 2 eāixĀ·Ī¾ u(t, x)dx = eāixĀ·Ī¾ ā a2 Īx u dx = + a2 |Ī¾|2 +a |Ī¾| u Ė. āt āt āt
Therefore (7.1) is equivalent to āu Ė(t, Ī¾) + a2 |Ī¾|2 u Ė(t, Ī¾) = 0. āt The initial condition takes the form (7.9)
(7.10)
Ė where ĻĖ = F Ļ. u Ė|t=0 = Ļ(Ī¾),
Equation (7.9) is an ordinary diļ¬erential equation with the independent variable t and with a parameter Ī¾. It is easy to solve it, and we get with (7.10) 2 2 Ė u Ė(t, Ī¾) = eāta |Ī¾| Ļ(Ī¾). From this formula we see that u Ė(t, Ī¾) is actually a (continuous for t ā„ 0 and ā C for t > 0) function of t with values in S(RnĪ¾ ). Therefore, applying the inverse Fourier transform, we get a solution u(t, x) of the Cauchy problem satisfying the above conditions.
7.4. Poissonās integral
157
Let us derive an explicit formula. We have
2 2 ān ixĀ·Ī¾ ān Ė(t, Ī¾)dĪ¾ = (2Ļ) Ė u(t, x) = (2Ļ) e u eixĀ·Ī¾āta |Ī¾| Ļ(Ī¾)dĪ¾
2 2 = (2Ļ)ān ei(xāy)Ā·Ī¾āta |Ī¾| Ļ(y)dydĪ¾. Interchanging the order of integration (by Fubiniās theorem) we get
u(t, x) = Ī(t, x ā y)Ļ(y)dy, where (7.11)
Ī(t, x) = (2Ļ)
ān
eixĀ·Ī¾āta
2 |Ī¾|2
dĪ¾
(we have essentially repeated the argument proving that the Fourier transform turns multiplication into convolution). Let us express Ī(t, x) explicitly, computing the integral in (7.11). For this, complete the square in the exponent: ix ix |x|2 2 2 2 2 . Ā· Ī¾ ā ā ix Ā· Ī¾ ā ta |Ī¾| = āta Ī¾ Ā· Ī¾ + ix Ā· Ī¾ = āta Ī¾ ā 2ta2 2ta2 4ta2 ix The change of coordinates Ī¾ ā 2ta 2 = Ī·, i.e., the shift of the contour of integration over each Ī¾j , yields
ā ān |x|2 |x|2 2 ān ā 4ta2 āta2 |Ī·|2 ān ā 4ta2 dĪ· = (2Ļ) e (a t) e eā|Ī¾| dĪ¾ Ī(t, x) = (2Ļ) e ā (we have also made the change of variables Ī¾ = a tĪ·).
Using the formula
eā|Ī¾| dĪ¾ = Ļ n/2 , 2
we ļ¬nally get
ā ān |x|2 Ī(t, x) = (2a Ļt) exp ā 2 . 4ta The solution of the Cauchy problem is expressed in the form
|xāy|2 1 ā 4a2 t Ļ(y)dy, ā e (7.12) u(t, x) = (2a Ļt)n called Poissonās integral. Let us analyze this formula. First, observe that the formula makes sense for a larger class of initial functions Ļ(x) than the class S(Rn ) from which we have started. For instance, it is clear that if Ļ(x) is continuous and (7.13)
2
|Ļ(x)| ā¤ Ceb|x| , 1 4a2 t
b > 0,
> b, i.e., for 0 < t < 4a12 b . Here the then the integral (7.12) converges if integral itself and integrals obtained from it after diļ¬erentiation with respect
158
7. The heat equation
to t or x any number of times uniformly converge for t ā [A, B], |x| ā¤ R, where 0 < A < B < 4a12 b . Therefore, we may diļ¬erentiate any number of times under the integral sign. The function u(t, x) is a solution of the heat equation (7.1) for the indicated values of t. Indeed, it suļ¬ces to show that Ī(t, x) is a solution of (7.1) for t > 0. This can be directly veriļ¬ed, but to avoid this calculation we may also note that, as we already know,
ā ā 2 2 ā a Īx u(t, x) = ā a Īx Ī(t, x ā y)Ļ(y)dy 0= āt āt for any Ļ ā S(Rn ), implying ā 2 ā a Īx Ī(t, x ā y) = 0, āt
t > 0.
It is easy to verify that if Ļ is continuous and satisļ¬es (7.13) and u is given by (7.12), then u|t=0 = Ļ(x) in the sense that (7.14)
lim u(t, x) = Ļ(x) for any x ā Rn .
tā0+
We will even prove this in two ways.
ā First method. Note that Ī(t, x) = Īµān f ( xĪµ ), where Īµ = 2a t and f (x) = Ļ ān/2 exp(ā|x|2 ), so that f (x) ā„ 0, f (x)dx = 1, and Īµ ā 0+ as t ā 0+. So we deal with a classical Ī“-like family of positive functions (in particular, Ī(t, x) ā Ī“(x) in D (Rn ) as t ā 0+). Now, (7.14) is veriļ¬ed via the standard scheme of the proof of Ī“-likeness; see Example 4.5. We skip the corresponding details; the reader should consider this a recommended exercise. Second method. Suppose we wish to verify (7.14) for x = x0 . Let us decompose Ļ(x) into the sum Ļ(x) = Ļ0 (x) + Ļ1 (x), where Ļ0 (x) is continuous, coincides with Ļ(x) for |xāx0 | ā¤ 1, and vanishes for |xāx0 | ā„ 2. Accordingly, Ļ1 (x) is continuous, vanishes for |x ā x0 | ā¤ 1, and satisļ¬es (7.13). Poissonās integral (7.12) for Ļ splits into the sum of similar integrals for Ļ0 and Ļ1 (we denote them by u0 (t, x) and u1 (t, x)). It suļ¬ces to verify (7.14) for u0 (t, x) and u1 (t, x) separately. First, let us deal with u1 (t, x). We have
Ī(t, x0 ā y)Ļ1 (y)dy. u1 (t, x0 ) = |yāx0 |ā„1
Clearly, the integrand tends to zero as t ā 0+ (since lim Ī(t, x) = 0 for x = 0) and is majorized for 0 < t ā¤ B < function of the form C1 exp(āĪµ|y|2 ), where Īµ > convergence theorem lim u1 (t, x0 ) = 0. tā+0
tā+0 1 , by an independent of t 4a2 b 0, C1 > 0. By the dominated
7.4. Poissonās integral
159
It remains to consider the case of an initial function with a compact support. Here the argument of Example 4.5 will do without any modiļ¬cations. There is, however, another method based on approximation by smooth functions with a subsequent passage to the limit. It is based on the following important lemma. Lemma 7.3 (The maximum principle). Let a function u(t, x) be deļ¬ned by Poissonās integral (7.12). Then inf Ļ(x) ā¤ u(t, x) ā¤ sup Ļ(x).
(7.15)
xāRn
xāRn
If one of these inequalities turns into the equality for some t > 0 and x ā Rn , then u = const. Proof. We have
u(t, x) = Ī(t, x ā y)Ļ(y)dy ā¤ sup Ļ(x) Ā· Ī(t, x ā y)dy = sup Ļ(x). xāRn
xāRn
The second inequality in (7.15) is proved similarly. Since Ī(t, x) > 0 for all t > 0, x ā Rn , then the equality is only possible if Ļ(x) = const almost everywhere, and then u = const as required. Remark 7.4. As we will see later, a solution u(t, x) of the Cauchy problem (7.2), satisfying the growth restriction of type (7.13) with respect to x uniformly in t, is unique. It will follow that the solution u(t, x) in fact can be represented by the Poisson integral and hence satisļ¬es the maximum principle (7.15). Let us ļ¬nish the proof of (7.14). Let Ļ be a continuous function with a compact support; let Ļk , k = 1, 2, . . . , be a sequence of functions from C0ā (Rn ) such that sup |Ļ(x) ā Ļk (x)| ā 0 for k ā +ā.
xāRn
(It is possible to obtain a sequence Ļk , for example, via mollifying; cf. Section 4.2.) Let
uk (t, x) =
Ī(t, x ā y)Ļk (y)dy.
Since Ļk ā S(Rn ), it follows that lim uk (t, x) = Ļk (x) in the topology of tā+0
S(Rn ) (in particular, uniformly on Rn ). But by Lemma 7.3 sup t>0, xāRn
|uk (t, x) ā u(t, x)| ā¤ sup |Ļk (x) ā Ļ(x)|; xāRn
this, clearly, implies lim u(t, x) = Ļ(x). Indeed, given Īµ > 0, we may tā+0
choose k such that sup |Ļk (x) ā Ļ(x)| < xāRn
Īµ 3,
and then t0 > 0 such that
160
7. The heat equation
sup |uk (t, x) ā Ļk (x)|
0, x ā Rn . It is also easily veriļ¬ed that u(t, x) is a solution of (7.1). The relation lim u(t, Ā·) = Ļ(Ā·) now holds in tā+0
the sense of weak convergence in S (Rn ). Indeed, it is easy to verify that if Ļ(x) ā S(Rn ), then
u(t, x)Ļ(x)dx = Ļ(Ā·), Ī(t, x ā Ā·)Ļ(x)dx
= Ļ(Ā·), Ī(t, x ā Ā·)Ļ(x)dx = Ļ(Ā·), v(t, Ā·), where v(t, x) is the Poisson integral corresponding to the initial function Ļ(x) (we justify the possibility to interchange the integration with respect to x and application of the functional Ļ by the convergence of the integral in the topology of S(Rny ); cf. an identical argument in the proof of Proposition 5.3). Since v(t, Ā·) ā Ļ(Ā·) as t ā +0 in the topology of S(Rn ), it follows that
lim u(t, x)Ļ(x)dx = Ļ, Ļ, tā+0
implying that lim u(t, Ā·) = Ļ(Ā·) in S (Rn ).
tā+0
We have already noted above that lim Ī(t, x) = Ī“(x) in S (Rn ). This tā+0
relation is often written in the shorter form Ī|t=0 = Ī“(x) and is a particular case of the statement just proved. Finally, let (7.13) be true for any b > 0 (with a constant C > 0 depending on b). Then the Poisson integral and, therefore, solution of the Cauchy problem are deļ¬ned for any t > 0. This is the case, e.g., if |Ļ(x)| ā¤ C exp(b0 |x|2āĪµ ) for some Īµ > 0, b0 > 0 and, in particular, if Ļ(x) is bounded.
7.5. The fundamental solution. Duhamelās formula
161
If Ļ ā Lp (Rn ) for p ā [1, ā), then Poissonās integral converges by HĀØolderās inequality. In this case lim u(t, Ā·) = Ļ(Ā·) with respect to the norm tā+0
in Lp (Rn ) (the proof is similar to the argument proving the convergence of molliļ¬ers with respect to the Lp -norm; see Section 4.2). It is easy to verify that if Ļ(x) satisļ¬es (7.13), then the function u(t, x) determined by Poissonās integral satisļ¬es a similar estimate: 1 2 |u(t, x)| ā¤ C1 eb1 |x| , 0 ā¤ t ā¤ B < 2 , 4a b where the constants C1 , b1 depend on Ļ and B. It turns out that a solution satisfying such an estimate is unique; we will prove this later. At the same time it turns out that the estimate 2+Īµ
|u(t, x)| ā¤ Ceb|x|
does not guarantee the uniqueness for any Īµ > 0. (For the proof see, e.g., John [16, Ch. 7].) The uniqueness of solution makes it natural to apply Poissonās integral in order to ļ¬nd a meaningful solution (as a rule, solutions of fast growth have no physical meaning). For instance, the uniqueness of a solution takes place in the class of bounded functions. By the maximum principle (Lemma 7.3), the Cauchy problem is well-posed in the class of bounded functions (if Ļ is perturbed by adding a uniformly small function, then the same happens with u(t, x)). In conclusion notice that the fact that Ī(t, x) > 0 for all t > 0, x ā Rn implies that āthe speed of the heat propagation is inļ¬niteā (since the initial perturbation is supported at the origin!). But Ī(t, x) decreases very fast as |x| ā +ā, which is practically equivalent to a ļ¬nite speed of the heat propagation.
7.5. The fundamental solution for the heat operator. Duhamelās formula Theorem 7.5. The following locally integrable function ā ān |x|2 E(t, x) = Īø(t)Ī(t, x) = (2a Ļt) Īø(t) exp ā 2 4a t is a fundamental solution for function.
ā āt
ā a2 Īx in Rn+1 . Here Īø(t) is the Heaviside
Proof. First, let us give a heuristic explanation for why E(t, x) is a fundamental solution. We saw that Ī(t, x) is a continuous function in t with values in S (Rnx ) for t ā„ 0. The function E(t, x) = Īø(t)Ī(t, x) is not continuous as a function of t with values in S (Rnx ) but has a jump at t = 0 equal
162
7. The heat equation
ā to Ī“(x). Therefore, applying āt , we get Ī“(t) ā Ī“(x) + f (t, x), where f (t, x) is continuous in t in the similar sense. But then ā 2 ā a Īx E(t, x) = Ī“(t) ā Ī“(x) = Ī“(t, x), āt since the summand which is continuous with respect to t vanishes because ā ( āt ā a2 Īx )E(t, x) = 0 for |t| + |x| = 0. This important argument makes it possible to write down the fundamental solution in all cases when we can write the solution of the Cauchy problem in a Poisson integral-like form. We will not bother with its justiļ¬cation, however, since it is simpler to verify directly that E(t, x) is a fundamental solution.
We begin with the veriļ¬cation of local integrability of E(t, x). Clearly, E(t, x) ā C ā (Rn+1 \{0}), so we only have to verify local integrability in a neighborhood of the origin. Since 1 exp ā ā¤ CĪ± Ļ Ī± for Ļ > 0, for any Ī± ā„ 0, Ļ we have t Ī± ān/2 = CĪ± tān/2+Ī± |x|ā2Ī± . (7.16) |E(t, x)| ā¤ CĪ± t |x|2 Let Ī± ā ( n2 ā 1, n2 ) so that ā n2 + Ī± > ā1 and ā2Ī± > ān. It is clear from (7.16) that E(t, x) is majorized by an integrable function in a neighborhood of the origin; therefore, it is locally integrable itself. Now, let f (t, x) ā C0ā (Rn+1 ). We have ā ā 2 2 ā a Īx E(t, x), f (t, x) = E(t, x), ā ā a Īx f (t, x) āt āt
ā E(t, x) ā ā a2 Īx f (t, x)dtdx. (7.17) = lim Īµā+0 tā„Īµ āt ā and Īx in the integral back onto E(t, x). Since Let us move āt ā ( āt āa2 Īx )E(t, x) = 0 for t > 0, the result only contains a boundary integral over the plane t = Īµ obtained after integrating by parts with respect to t:
E(Īµ, x)f (Īµ, x)dx = Ī(Īµ, x)f (Īµ, x)dx.
The latter integral is uĪµ (Īµ, 0), where uĪµ (t, x) is Poissonās integral with the initial function ĻĪµ (x) = f (Īµ, x). Consider also Poissonās integral u(t, x) with the initial function Ļ(x) = f (0, x). By the maximum principle |u(Īµ, 0) ā uĪµ (Īµ, 0)| ā¤ sup |ĻĪµ (x) ā Ļ(x)| = sup |f (Īµ, x) ā f (0, x)|. x
x
This, clearly, implies that u(Īµ, 0) ā uĪµ (Īµ, 0) ā 0
for Īµ ā +0.
7.5. The fundamental solution. Duhamelās formula
163
But, as we have already seen, u(Īµ, 0) ā Ļ(0) = f (0, 0) for Īµ ā +0, implying lim uĪµ (Īµ, 0) = f (0, 0) which, in turn, yields Īµā+0
lim
Īµā+0
ā 2 E(t, x) ā ā a Īx f (t, x)dtdx = f (0, 0). āt tā„Īµ
This, together with (7.17), proves the theorem. Corollary 7.6. If u(t, x) ā D (Ī©), where Ī© is a domain in Rn+1 , and ā ā a2 Īx )u = 0, then u ā C ā (Ī©). ( āt
Therefore, the heat operator is an example of a hypoelliptic but not elliptic operator. Note that the heat operator has C ā but nonanalytic solutions (e.g., E(t, x) in a neighborhood of (0, x0 ), where x0 ā Rn \{0}). The knowledge of the fundamental solution makes it possible to solve the nonhomogeneous equation. For instance, a solution u(t, x) of the problem āu 2 n āt ā a Īx u = f (t, x), t > 0, x ā R , (7.18) x ā Rn , u|t=0 = Ļ(x), can be written in the form
E(t ā Ļ, x ā y)f (Ļ, y)dĻ dy + Ī(t, x ā y)Ļ(y)dy u(t, x) = Ļ ā„0
(7.19)
t
dĻ
= 0
Rn
Ī(t ā Ļ, x ā y)f (Ļ, y)dy +
Ī(t, x ā y)Ļ(y)dy
under appropriate assumptions concerning f (t, x) and Ļ(x) which we will not formulate exactly. The ļ¬rst integral in (7.19) is of interest. Its inner integral
v(Ļ, t, x) =
Ī(t ā Ļ, x ā y)f (Ļ, y)dy
is a solution of the Cauchy problem ā ( āt ā a2 Īx )v(Ļ, t, x) = 0, t > Ļ, x ā Rn , (7.20) v|t=Ļ = f (Ļ, x). The Duhamel formula is a representation of the solution of the Cauchy problem (7.18) with zero initial function Ļ in the form
t v(Ļ, t, x)dĻ, (7.21) u(t, x) = 0
where v(Ļ, t, x) is the solution of the Cauchy problem (7.20) (for a homogeneous equation!).
164
7. The heat equation
A similar representation is possible for any evolution equation. In parā ā ticular, precisely this representation will do for operators of the form āt A(x, Dx ), where A(x, Dx ) is a linear diļ¬erential operator with respect to x with variable coeļ¬cients. Needless to say, in this case a2 Īx should be replaced by A(x, Dx ) in (7.20). The formal veriļ¬cation is very simple: by ā ā A(x, Dx ) to (7.21) we get in the right-hand side applying āt
t ā ā A(x, Dx ) v(Ļ, t, x)dĻ = v(t, t, x) = f (t, x), v(t, t, x) + āt 0 as required.
7.6. Estimates of derivatives of a solution of the heat equation Since the heat operator is hypoelliptic, we can use Proposition 5.20 to estimate derivatives of a solution of the heat equation through the solution itself on a slightly larger domain. We will only need it on an inļ¬nite horizontal strip. Lemma 7.7. Let us assume that a solution u(t, x) of the heat equation (7.1) is deļ¬ned in the strip {t, x : t ā (a, b), x ā Rn } and satisļ¬es the estimate p
|u(t, x)| ā¤ Ced|x| ,
(7.22)
where p ā„ 0, C > 0, d ā R. Then in a somewhat smaller strip {t, x : t ā (a , b ), x ā Rn }, where a < a < b < b, similar estimates hold:
|ā Ī± u(t, x)| ā¤ CĪ± ed |x| ,
(7.23)
p
where the constant p is the same as in (7.22); d = 0 if d = 0 and d = d + Īµ with an arbitrary Īµ > 0 if d = 0; the CĪ± depend upon the (n+1)-dimensional multi-index Ī± and also upon Īµ, a , and b . Proof. Let us use Proposition 5.20 with Ī©1 = {t, x : t ā (a, b), |x| < 1}, Ī©2 = {t, x : t ā (a , b ), |x| < 1/2}, and estimate (5.22) applied to the solution uy (t, x) = u(t, x + y) of the heat equation depending on y ā Rn as on a parameter. We get |ā Ī± u(t, x)| ā¤ CĪ±
sup |u(t, x )| ā¤ CĪ±
|x āx| 0, c > 0, |x| ā„ R, and R is suļ¬ciently large. The same estimate holds also for any derivative ā Ī± v(t, x), where Ī± is a (n + 1)-dimensional multi-index. Let t0 > 0 be so small that 4ac2 t0 > b; i.e., t0 < 4ac2 b . Then the integral
u(t, x), v(t, x) = u(t, x)v(t, x)dx makes sense and converges uniformly for 0 ā¤ t ā¤ t0 ā Īµ. As Lemma 7.7 shows, the integrals obtained by replacing u or v by their derivatives with respect to t or x also converge uniformly for 0 < Īµ ā¤ t ā¤ t0 ā Īµ, where Īµ > 0 is arbitrary. Therefore, the function Ļ(t) = u(t, x), v(t, x) is continuous at t ā [0, t0 ) and vanishes at t = 0, and its derivative with respect to t at t ā (0, t0 ) is
ā(u(t, x)v(t, x)) āu āv dĻ(t) = dx = Ā·v+uĀ· dx dt āt āt āt
= a2 [(Īx u)v ā u(Īx v)]dx.
7.7. Holmgrenās principle. The uniqueness of solution
167
The latter integral vanishes, since, by Greenās formula (4.32), it may be expressed in the form
lim [(Īx u)v ā u(Īx v)]dx Rāā |x|ā¤R
= lim
Rāā |x|=R
āv āu Ā·vāuĀ· ān ĀÆ ān ĀÆ
dS = 0,
because āāunĀÆ Ā·v and uĀ· āāvnĀÆ tend to 0 exponentially fast as |x| ā +ā. Therefore, Ļ(t) = const = 0. Now, let us verify that
u(t, x)v(t, x)dx. u(t0 , x)Ļ(x)dx = lim tāt0 ā0
(Actually, the right-hand side equals lim Ļ(t) = 0.) We have tāt0 ā0
u(t, x)v(t, x)dx = u(t, x)Ī(t0 ā t, x ā y)Ļ(y)dydx
= Ī(t0 ā t, x ā y)u(t, x)dx Ļ(y)dy ā u(t0 , y)Ļ(y)dy, t ā t0 ā 0, since the argument we used in the consideration of Poissonās integral shows that the convergence
Ī(t0 ā t, x ā y)u(t, x)dx = u(t0 , y) lim tāt0 ā0
is uniform on any compact in Rny . Thus, clearly,
u(t0 , x)Ļ(x)dx = 0, 0 ā¤ t0 ā¤ Ī“, Ļ ā D(Rn ), if Ī“
0, x ā Ī©, āØ āt = a2 u, (7.27) t > 0, x ā āĪ©, u|xāāĪ© = 0, ā© u|t=0 = Ļ(x), x ā Ī©, by the Fourier method. Here we assume Ī© to be a bounded domain in Rn . If the function u(t, x) = Ļ(t)v(x) satisļ¬es the ļ¬rst two conditions in (7.27), we formally get (7.28)
Īv(x) Ļ (t) = = āĪ», v|āĪ© = 0, 2 a Ļ(t) v(x)
leading to the eigenvalue problem for the operator āĪ with zero boundary conditions on āĪ©: āĪv = Ī»v, x ā Ī©, (7.29) v|āĪ© = 0. For n = 1 this problem becomes the simplest Sturm-Liouville problem considered above (with the simplest boundary conditions: v should vanish at the endpoints of the segment). We will prove below that problem (7.29) possesses properties very similar to those of the Sturm-Liouville problem. In particular, it has a complete orthogonal system of eigenfunctions vk (x), k = 1, 2, . . ., with eigenvalues Ī»k , such that Ī»k > 0 and Ī»k ā +ā as k ā +ā. It follows from (7.28) that Ļ(t) = CeāĪ»a t which makes it clear that it is natural to seek u(t, x) as a sum of the series 2
u(t, x) =
ā
Ck eāĪ»k a t vk (x). 2
k=1
The coeļ¬cients Ck are chosen from the initial condition u|t=0 = Ļ(x) and are coeļ¬cients of the expansion of Ļ(x) in terms of the orthogonal system {vk (x)}ā k=1 .
7.8. Fourier method
169
The second initial-boundary value ā§ āu āØ āt = a2 Īu, āu | = 0, ā© ā nĀÆ xāāĪ© u|t=0 = Ļ(x),
problem t > 0, x ā Ī©, t > 0, x ā āĪ©, x ā Ī©,
is solved similarly. It leads to the eigenvalue problem āĪv = Ī»v, x ā Ī©, (7.30) āv ān ĀÆ |āĪ© = 0 with the same properties as the problem (7.29), the only exception being that Ī» = 0 is an eigenvalue of (7.30). The same approach, as for n = 1, yields solutions of more general problems; for example, ā§ āu āØ āt = a2 Īu + f (t, x), t > 0, x ā Ī©, = Ī±(t, x), t > 0, x ā āĪ©, u| ā© xāāĪ© x ā Ī©. u|t=0 = Ļ(x), For this, subtracting an auxiliary function, we make Ī±(t, x) ā” 0 and then seek the solution u(t, x) in the form of a series u(t, x) =
ā
Ļk (t)vk (x),
k=1
which leads to ļ¬rst-order ordinary diļ¬erential equations for Ļk (t). The uniqueness of the solution of these problems may be proved by Holmgrenās principle or derived from the maximum principle: if u(t, x) is ĀÆ and if it satisļ¬es the heat a function continuous in the cylinder [0, T ] Ć Ī© equation on (0, T ) Ć Ī©, then 3 (7.31)
u(t, x) ā¤ max sup u(0, x), xāĪ©
sup
u(t, x) ;
xāāĪ©,tā[0,T ]
and if the equality takes place, then u = const. We will not prove this physically natural principle. Its proof is quite elementary and may be found, for instance, in Petrovskii [24]. In what follows, we give the proof of existence of a complete orthogonal system of eigenfunctions for the problem (7.29). The boundary condition in (7.29) will be understood in a generalized sense. The Dirichlet problem will also be solved in a similar sense. All this requires introducing Sobolev spaces whose theory we are about to present.
170
7. The heat equation
7.9. Problems 7.1. Assume a rod of length l with a heat-insulated surface (the lateral sides and ends). For t = 0, the temperature of the left half of the rod is u1 and that of the right half is u2 . Find the rod temperature for t > 0 by the Fourier method. Find the behavior of the temperature as t ā +ā. Estimate the relaxation time needed to decrease by half the diļ¬erence between the maximal and minimal temperatures of the rod; in particular, ļ¬nd the dependence of this time on the parameters of the rod: length, thickness, density, speciļ¬c heat, and thermal conductivity. 7.2. The rod is heated by a constant electric current. Zero temperature is maintained at the ends and the side surface is insulated. Assuming that the temperature in any cross section of the rod is a constant, ļ¬nd the distribution of the temperature in the rod from an initial distribution and describe the behavior of the temperature as t ā +ā. 7.3. Let a bounded solution u(t, x) of the heat equation ut = a2 uxx (x ā R1 ) be determined for t > 0 and let it satisfy the initial condition u|t=0 = Ļ(x), where Ļ ā C(R1 ), lim Ļ(x) = b, lim Ļ(x) = c. Describe the behavior of xā+ā
u(t, x) as t ā +ā.
xāāā
7.4. Prove the maximum principle for the solutions of the heat equation in the formulation given at the end of Section 7.8 (see (7.31)).
10.1090/gsm/205/08
Chapter 8
Sobolev spaces. A generalized solution of Dirichletās problem
8.1. Spaces H s (Ī©) Let s ā Z+ , and let Ī© be an open subset in Rn . Deļ¬nition 8.1. The Sobolev space H s (Ī©) consists of u ā L2 (Ī©) such that ā Ī± u ā L2 (Ī©) for |Ī±| ā¤ s, where Ī± is a multi-index and the derivative is understood in the distributional sense. In H s (Ī©), introduce the inner product (ā Ī± u, ā Ī± v), (u, v)s = |Ī±|ā¤s
where (Ā·, Ā·) is the inner product in L2 (Ī©), and also the corresponding norm ā” ā¤1 2
1 Ī± 2 |ā u(x)| dxā¦ . (8.1) us = (u, u)s2 = ā£ |Ī±|ā¤s Ī©
Clearly, H s (Ī©) ā H s (Ī©) for s ā„ s . Furthermore, H 0 (Ī©) = L2 (Ī©). In particular, each space H s (Ī©) is embedded into L2 (Ī©). Proposition 8.2. The inner product (u, v)s deļ¬nes in H s (Ī©) the structure of a separable Hilbert space (i.e., a Hilbert space which admits at most a countable orthonormal basis). 171
172
8. Sobolev spaces. Dirichletās problem
Proof. Only the completeness and separability are not obvious. Let us prove the completeness. Let a sequence {uk }ā k=1 be a Cauchy sequence in H s (Ī©). The deļ¬nition of Ā· s implies that all the sequences {ā Ī± uk }ā k=1 (for |Ī±| ā¤ s) are Cauchy sequences in L2 (Ī©). But then, by the completeness of L2 (Ī©), they converge; i.e., lim ā Ī± uk = vĪ± with respect to the norm in kāā
L2 (Ī©). In particular, uk ā v0 in L2 (Ī©). But then ā Ī± uk ā ā Ī± v0 in D (Ī©) and since, on the other hand, ā Ī± uk ā vĪ± in D (Ī©), it follows that vĪ± = ā Ī± v0 . Therefore, v0 ā H s (Ī©) and ā Ī± uk ā ā Ī± v0 in L2 (Ī©) for |Ī±| ā¤ s. But this means that uk ā v0 in H s (Ī©). Let us prove separability. The map u ā {ā Ī± u}|Ī±|ā¤s deļ¬nes an isometric embedding of H s (Ī©) into the direct sum of several copies of L2 (Ī©) as a closed subspace. Now, the separability of H s (Ī©) follows from that of L2 (Ī©). Let us consider the case Ī© = Rn in more detail. The space H s (Rn ) can be described in terms of the Fourier transform (understood in the sense of distributions, in S (Rn )). Recall that, by Plancherelās theorem, the Fourier transformation
F : u(x) ā u Ė(Ī¾) = eāixĀ·Ī¾ u(x)dx determines an isometry of L2 (Rnx ) onto L2 (RnĪ¾ ), where L2 (RnĪ¾ ) is deļ¬ned by the measure (2Ļ)ān dĪ¾. Let us explain the meaning of this statement and its interpretation from the distributional viewpoint. We know from analysis that the Parseval identity
2 ān |Ė u(Ī¾)|2 dĪ¾ (8.2) |u(x)| dx = (2Ļ) holds for u ā S(Rn ). The identity (8.2) for u ā S(Rn ) follows easily, for instance, from the inversion formula (5.23). Indeed,
2 āixĀ·Ī¾ iyĀ·Ī¾ e u(x)dx e u(y)dy dĪ¾ |Ė u(Ī¾)| dĪ¾ =
= ei(yāx)Ā·Ī¾ u(x)u(y) dx dy dĪ¾ &
'
i(yāx)Ā·Ī¾ n |u(y)|2 dy. e u(x) dx dĪ¾ dy = (2Ļ) = u(y) Since F deļ¬nes an isomorphism of S(Rnx ) onto S(RnĪ¾ ) and S(Rn ) is dense in L2 (Rn ), then F can be extended to an isometry F : L2 (Rnx ) āā L2 (RnĪ¾ ). Since the convergence in L2 (Rn ) implies a (weak) convergence in S (Rn ), the map F : L2 (Rnx ) āā L2 (RnĪ¾ ) is continuous in restrictions of the weak topologies of S (Rnx ) and S (RnĪ¾ ). Therefore, this F is, actually, a restriction of the generalized Fourier transform F : S (Rnx ) āā S (RnĪ¾ ).
8.1. Spaces H s (Ī©)
173
Now, note that the Fourier transform maps D Ī± u(x) to Ī¾ Ī± u Ė(Ī¾). Theres n Ī± 2 n Ė(Ī¾) ā L (R ) for |Ī±| ā¤ s. But instead fore, u ā H (R ) if andonly if Ī¾ u Ī± |2 |Ė |Ī¾ u(Ī¾)|2 ā L1 (Rn ). Due to the obvious of this, we may write |Ī±|ā¤s estimates C ā1 (1 + |Ī¾|2 )s ā¤ |Ī¾ Ī± |2 ā¤ C(1 + |Ī¾|2 )s , |Ī±|ā¤s
where C > 0 does not depend on Ī¾, we have u ā H s (Rn ) if and only if Ė(Ī¾) ā L2 (Rn ) and the norm (8.1) for Ī© = Rn is equivalent to (1 + |Ī¾|2 )s/2 u the norm &
'1 2 2 s 2 (1 + |Ī¾| ) |Ė u(Ī¾)| dĪ¾ . (8.3) us = We will denote the norm (8.3) by the same symbol as the norm (8.1); there should not be any misunderstanding. Thus, we have proved Proposition 8.3. The space H s (Rn ) consists of u ā S (Rn ) such that Ė(Ī¾) ā L2 (Rn ). It can be equipped with the norm (8.3) which is (1 + |Ī¾|2 )s/2 u equivalent to the norm (8.1). This proposition may serve as a basis for the deļ¬nition of H s (Rn ) for any real s. Namely, for any s ā R we may construct the space of u ā S (Rn ), Ė(Ī¾) ā L2 (Rn ), with the norm such that u Ė(Ī¾) ā L2loc (Rn ) and (1 + |Ī¾|2 )s/2 u (8.3). We again get a separable Hilbert space also denoted, as for s ā Z+ , by H s (Rn ). In what follows we will need the Banach space Cbk (Rn ) (here k ā Z+ ) consisting of functions u ā C k (Rn ), whose derivatives ā Ī± u(x), |Ī±| ā¤ k, are bounded for all x ā Rn . The norm in Cbk (Rn ) is deļ¬ned by the formula u(k) = sup |ā Ī± u(x)|. |Ī±|ā¤k
xāRn
Theorem 8.4 (Sobolevās embedding theorem). For s > continuous embedding (8.4)
n 2
+ k, we have a
H s (Rn ) ā Cbk (Rn ).
Let us clarify the formulation. At ļ¬rst glance, it seems that it is meaningless, since a function u ā H s (Rn ) can be arbitrarily modiļ¬ed on any set of measure 0 without aļ¬ecting the corresponding distribution (and the element of H s (Rn )). Changing u at all points with rational coordinates we may make it discontinuous everywhere. Therefore, we should more precisely describe the embedding (8.4), as follows: if u ā H s (Rn ), then there exists a unique function u1 ā Cbk (Rn ) which coincides with the initial function u
174
8. Sobolev spaces. Dirichletās problem
almost everywhere (and for brevity we write u instead of u1 ). In other words, H s (Rn ) is a subspace of Cbk (Rn ) provided both are considered as subspaces of S (Rn ) or D (Rn ). Observe that the uniqueness of a continuous representative is obvious, since in any neighborhood of x0 ā Rn there exist points of any set of total measure, so that any modiļ¬cation of a continuous function on a set of measure 0 leads to a function which is discontinuous at all points where the modiļ¬cation took place. Proof of Theorem 8.4. First, let us prove the estimate u(k) ā¤ Cus , u ā S(Rn ),
(8.5)
where C is a constant independent of u. The most convenient way is to apply the Fourier transform. We have
1 Ė(Ī¾)dĪ¾, (iĪ¾)Ī± eixĀ·Ī¾ u ā Ī± u(x) = (2Ļ)n implying 1 |ā u(x)| ā¤ (2Ļ)n Ī±
hence,
|Ī¾ Ī± u Ė(Ī¾)|dĪ¾;
u(k) ā¤ C
(1 + |Ī¾|2 )k/2 |Ė u(Ī¾)|dĪ¾.
Dividing and multiplying the integrand by (1 + |Ī¾|2 )s/2 , we have by the Cauchy-Schwarz inequality
u(Ī¾)|dĪ¾ u(k) ā¤ C (1 + |Ī¾|2 )(kās)/2 (1 + |Ī¾|2 )s/2 |Ė
ā¤C
(1 + |Ī¾| )
2 (kās)
1
1 2 2 2 s/2 2 dĪ¾ Ā· (1 + |Ī¾| ) |Ė u(Ī¾)| dĪ¾ .
The integral in the ļ¬rst factor in the right-hand side converges, since 2(k ā s) < ān, and, therefore, the integrand decays faster than |Ī¾|ānāĪµ as |Ī¾| ā +ā for a suļ¬ciently small Īµ > 0. The second factor is equal to us . Therefore, for s > k + n2 we get (8.5). Now, note that S(Rn ) is dense in H s (Rn ) for any s ā R. Indeed, introduce the operator Īs which acts on u = u(x) by multiplying the Fourier transform u Ė(Ī¾) by (1 + |Ī¾|2 )s/2 . This operator is a unitary map of H s (Rn ) 2 n onto L (R ) mapping S(Rn ) isomorphically onto S(Rn ). Therefore, S(Rn ) is dense in H s (Rn ), because S(Rn ) is dense in L2 (Rn ). Now, let v ā H s (Rn ), um ā S(Rn ), um ā v in H s (Rn ). But (8.5) implies that the sequence um is a Cauchy sequence with respect to the norm Ā· (k) . Therefore, um ā v1 in Cbk (Rn ). But then v and v1 coincide as distributions
8.1. Spaces H s (Ī©)
175
and, therefore, almost everywhere. By continuity, (8.5) holds for all u ā H s (Rn ) proving the continuity of the embedding (8.4). In particular, Theorem 8.4 shows that it makes sense to talk about values of functions u ā H s (Rn ) at a point for s > n2 . Also of interest is the question of a meaning of restriction (or trace) of a function from a Sobolev space on a submanifold of an arbitrary dimension. We will not discuss this problem in full generality, however, and we only touch the most important case of codimension l. For simplicity, consider only the case of a hyperplane xn = 0 in Rn . A point x ā Rn will be expressed in the form x = (x , xn ), where x ā Rnā1 . Theorem 8.5 (Sobolevās theorem on restrictions (or traces)). If s > 1/2, then the restriction operator S(Rn ) ā S(Rnā1 ), u ā u|xn =0 , can be extended to a linear continuous map 1
H s (Rn ) āā H sā 2 (Rnā1 ). Proof. Let us express the restriction operator in terms of the Fourier transform (for u ā S(Rn )):
1 Ė(Ī¾ , Ī¾n )dĪ¾ dĪ¾n . eix Ā·Ī¾ u u(x , 0) = n (2Ļ) For brevity, set v(x ) = u(x , 0) and apply the Fourier transform F with respect to x . We obtain
1 (8.6) vĖ(Ī¾ ) = u Ė(Ī¾ , Ī¾n )dĪ¾n . 2Ļ The needed statement takes on the form of the estimate
vsā 1 ā¤ Cus ,
(8.7)
2
1
where Ā· sā 1 is the norm in H sā 2 (Rnā1 ), u ā S(Rn ), and C does not 2
depend on u. Let us prove this estimate. We have 2
1 (8.8) vsā 1 = (1 + |Ī¾ |2 )sā 2 |Ė v (Ī¾ )|2 dĪ¾ . 2
From (8.6) we deduce for any s > 0 (8.9)
&
'2 1 2 ās/2 2 s/2 (1 + |Ī¾| ) u Ė(Ī¾)dĪ¾n (1 + |Ī¾| ) |Ė v (Ī¾ )| = 4Ļ 2
1 2 ās ā¤ ) dĪ¾ Ā· (1 + |Ī¾|2 )s |Ė u(Ī¾)|2 dĪ¾n . (1 + |Ī¾| n 4Ļ 2 2
176
8. Sobolev spaces. Dirichletās problem
Let us estimate the ļ¬rst factor. We have
(1 + |Ī¾|2 )ās dĪ¾n =
(1 + |Ī¾ |2 + |Ī¾n |2 )ās dĪ¾n ā 2 āās
Ī¾n ā = (1 + |Ī¾ |2 )ās ā1 + dĪ¾n 2 1 + |Ī¾ |
ā 1 2 ās+ 21 = (1 + |Ī¾ | ) (1 + |t|2 )ās dt = C(1 + |Ī¾ |2 )ās+ 2 . āā
Here we used the fact that s > 12 , which ensures the convergence of the last integral. Therefore, (8.9) is expressed in the form 2
2 ās+ 12
|Ė v (Ī¾ )| ā¤ C(1 + |Ī¾ | )
u(Ī¾)|2 dĪ¾n . (1 + |Ī¾|2 )s |Ė
By applying this inequality to estimate |Ė v (Ī¾ )|2 in (8.8), we get the desired estimate (8.7). 1
Remark. The restriction map H s (Rn ) ā H sā 2 (Rnā1 ), u ā u|xn =0 , is actually surjective (see, e.g., HĀØ ormander [13, Ch. II], so the statement of the theorem is the exact one (the index s ā 12 cannot be increased). Theorem 8.5 means that if u ā H s (Rn ), then the trace u|xn =0 of u on 1 the hyperplane xn = 0 is well-deļ¬ned and is an element of H sā 2 (Rnā1 ). It is obtained as follows: represent u in the form of the limit u = lim um of māā
functions um ā S(Rn ) with respect to the norm Ā· s . Then the restrictions um |xn =0 have a limit as m ā ā with respect to the norm Ā· sā 1 in 1
2
the space H sā 2 (Rnā1 ), and this limit does not depend on the choice of an approximating sequence. If we wish to restrict ourselves to integer values of s, we may use the fact 1 that H sā 2 (Rnā1 ) ā H sā1 (Rnā1 ). Therefore, the trace u|xn =0 of u ā H s (Rn ) belongs to H sā1 (Rnā1 ). The space H s (M ), where M is a smooth compact manifold (without boundary), can also be deļ¬ned. Namely, we will write u ā H s (M ) if for any coordinate diļ¬eomorphism Īŗ : U āā Ī© (here U is an open subset of M , Ī© an open subset of Rn ) and any Ļ ā C0ā (Ī©) we have Ļ(u ā¦ Īŗā1 ) ā H s (Rn ). Here uā¦Īŗā1 is a distribution u pulled back to Ī© via Īŗā1 . It is multiplied by Ļ to get a distribution with support in Ī©. Therefore, Ļ(u ā¦ Īŗā1 ) ā E (Ī©) ā E (Rn ); in particular, Ļ(u ā¦ Īŗā1 ) ā S (Rn ), and it makes sense to talk about the
Ės (Ī©) 8.2. Spaces H
177
inclusion Ļ(u ā¦ Īŗā1 ) ā H s (Rn ). The localization helps us to prove the following analogue of Theorem 8.5: If Ī© is a bounded domain with smooth boundary āĪ©, then for s > 12 the ĀÆ to a linear continuous restriction map u ā u|āĪ© can be extended from C ā (Ī©) 1 operator H s (Ī©) āā H sā 2 (āĪ©).
Ės (Ī©) 8.2. Spaces H Ės (Ī©) is the closure of D(Ī©) in H s (Ī©) (we Deļ¬nition 8.6. The space H assume that s ā Z+ ). Ės (Ī©) is a closed subspace in H s (Ī©). Hence, it is a separable Therefore, H Hilbert space itself. Ė1 (Ī©) does not necesĖ0 (Ī©) = H 0 (Ī©) = L2 (Ī©). But already H Clearly, H sarily coincide with H 1 (Ī©). Indeed, if Ī© is a bounded domain with a smooth boundary, then, as had been noted above, a function u ā H 1 (Ī©) possesses Ė1 (Ī©), then u|āĪ© = 0. Therefore, a trace u|āĪ© ā L2 (āĪ©). However, if u ā H ā ĀÆ Ė1 (Ī©). if, e.g., u ā C (Ī©) and u|āĪ© = 0, then u ā H Ės (Ī©) is isometrically embedded into H s (Rn ) as well. This The space H embedding extends the embedding D(Ī©) ā D(Rn ). Such an extension is deļ¬ned by continuity, since the norm Ā·s , considered on D(Ī©), is equivalent to the norm in H s (Rn ). The following theorem on compact embeddings is important. Theorem 8.7. Let Ī© be a bounded domain in Rn , s, s ā Z+ , s > s . Then Ės (Ī©) ā H s (Ī©) is a compact operator. the embedding H Recall that a linear operator A : E1 āā E2 , where E1 , E2 are Banach spaces, is called compact if the image of the unit ball B1 = {x : x ā E1 , xE1 ā¤ 1} is a precompact subset in E2 . In turn, a set Q in a metric space E is called precompact (in E) if its closure is compact, or, equivalently, if from any sequence of points of Q we can choose a subsequence converging to a point of E. If E is a complete metric space, then a set Q ā E is precompact if and only if it is totally bounded; i.e., for any Īµ > 0 it has a ļ¬nite Īµ-net: a ļ¬nite set x1 , . . . , xN ā E, such that Qā
N
B(xj , Īµ)
j=1
(here B(x, r) means open ball in E of radius r centered at x), or, in other words, for any point x ā Q there exists k ā {1, . . . , N } such that Ļ(x, xk ) < Īµ, where Ļ is the metric in E.
178
8. Sobolev spaces. Dirichletās problem
Let K be a compact subset in Rn (i.e., K is closed and bounded). Consider the space C(K) of continuous complex-valued functions on K. Let F ā C(K). The Arzel` a-Ascoli theorem gives a criterion of precompactness of F : F is precompact if and only if it is uniformly bounded and equicontinuous. Here the uniform boundedness of the set F means the existence of a constant M > 0, such that supxāK |f (x)| ā¤ M for any f ā F ; the equicontinuity means that for any Īµ > 0 there exists Ī“ > 0, such that |x ā x | < Ī“ with x , x ā K implies |f (x ) ā f (x )| < Īµ for any f ā F . Proofs of the above general facts on compactness and the Arzel`a-Ascoli theorem may be found in any textbook on functional analysis; see, e.g., Kolmogorov and Fomin [17, Chapter II, Sect. 6 and 7]. Note the obvious corollary of general facts on compactness: for a set Q ā E, where E is a complete metric space, to be precompact it suļ¬ces that for any Īµ > 0 there exists a precompact set QĪµ ā E, such that Q belongs to the Īµ-neighborhood of QĪµ . Indeed, a ļ¬nite Īµ/2-net of QĪµ/2 is an Īµ-net for Q. Proof of Theorem 8.7. We will use the mollifying operation (see the proof of Lemma 4.5). We have introduced there a family ĻĪµ ā D(Rn ), such that supp ĻĪµ ā {x : |x| ā¤ Īµ}, ĻĪµ ā„ 0, and ĻĪµ (x)dx = 1. Then from f ā L1loc (Rn ) we may form a molliļ¬ed family:
fĪµ (x) = f (x ā y)ĻĪµ (y)dy = f (y)ĻĪµ (x ā y)dy. In Section 4.2 we formulated and proved a number of important properties of mollifying. Now we will make use of these properties.
First, observe that a set F is precompact in H s (Ī©) if and only if all the sets FĪ± = {ā Ī± f : f ā F }, where Ī± is a multi-index such that |Ī±| ā¤ s , are Ės (Ī©) clearly implies ā Ī± f ā H Ė sā|Ī±| (Ī©), precompact in L2 (Ī©). Further, f ā H where |Ī±| ā¤ s. Since s > s , it is clear that all sets FĪ± belong to a bounded Ė1 (Ī©). Therefore, it suļ¬ces to verify the compactness of the subset of H Ė1 (Ī©) ā L2 (Ī©). embedding H Ė 1 (Ī©), u1 ā¤ 1}. We should verify that F is precomLet F = {u : u ā H 2 pact in L (Ī©) (or, equivalently, in L2 (Rn )). The idea of this veriļ¬cation is as follows: consider the set FĪµ = {fĪµ (x), f ā F } obtained by Īµ-mollifying all functions from F . Then prove that for a small Īµ > 0 the set F belongs to an arbitrarily small neighborhood of FĪµ (with ĀÆ for a ļ¬xed respect to the norm in L2 (Rn )) and FĪµ is precompact in C(Ī©)
Ės (Ī©) 8.2. Spaces H
179
Īµ (hence, precompact in L2 (Ī©)). The remark before the proof shows then that F is precompact in L2 (Ī©). Observe also that the deļ¬nition of the distributional derivative easily implies
āf āf (y) (x) = ĻĪµ (x ā y)dy āxj Īµ āyj
ā ĻĪµ (x ā y)dy = ā f (y) āyj
āĻĪµ (x ā y) dy = f (y) āxj āfĪµ (x) Ė1 (Ī©). for f ā H = āxj In other words, distributional diļ¬erentiation commutes with mollifying. On the other side, we clearly have from the Cauchy-Schwarz inequality that
|fĪµ (x)| ā¤ CĪµ |f (y)|dy
ā¤ CĪµ
1 |f (y)| dy 2
2
, f ā F.
Therefore, FĪµ is uniformly bounded (on Rn ) for a ļ¬xed Īµ > 0. Further, āfĪµ āf = āx are uniformly bounded by similar reasoning the derivatives āx j j Īµ for f ā F and a ļ¬xed Īµ > 0. Therefore, FĪµ is equicontinuous. By the Arzel`a-Ascoli theorem, FĪµ is precompact in C(Rn ) (we may assume that ĀÆ in Rn ). It follows FĪµ ā C(K), where K is a compact neighborhood of Ī© that FĪµ is precompact in L2 (Rn ) and FĪµ |Ī© = {fĪµ |Ī© , fĪµ ā FĪµ } is precompact in L2 (Ī©). It remains to verify that for any Ī“ > 0 there exists Īµ > 0, such that F belongs to the Ī“-neighborhood of FĪµ . To this end, let us estimate the norm of f ā fĪµ in L2 (Rn ). (This norm will be, for the time being, denoted by just Ā· .) First, assume that f ā D(Ī©). We have
[f (x) ā f (x ā y)]ĻĪµ (y)dy
1 d dt [f (x) ā f (x ā ty)]ĻĪµ (y) = dy dt 0
1
n āf dt dy yj (x ā ty)ĻĪµ (y). = āxj 0
f (x) ā fĪµ (x) =
j=1
180
8. Sobolev spaces. Dirichletās problem
Using Minkowskiās inequality (4.53) with p = 2 and the Cauchy-Schwarz inequality for ļ¬nite sums, we obtain āf f ā fĪµ ā¤ dt dy |yj | (Ā· ā ty) ĻĪµ (y) āxj 0 j=1
1
n n āf (Ā·) āf dt dy |yj | ĻĪµ (y) = = āxj ĻĪµ (y)|yj |dy āxj 0 j=1 j=1 n āf ā ā¤ Ī“1 āxj ā¤ Ī“1 nf 1 ,
1
n
j=1
n
Clearly, Ī“1 ā¤ CĪµ; hence Ī“1 ā 0 as Īµ ā 0. ā Since f ā F means f 1 ā¤ 1, it follows that f ā fĪµ ā¤ Ī“ = Ī“1 n for f ā F , f ā D(Ī©). where Ī“1 =
sup
xāsuppĻĪµ
j=1 |xj |.
In the resulting inequality (8.10)
f ā fĪµ ā¤ C1 Īµf 1
(which we proved for all f ā D(Ī©)) we can pass to the limit along a sequence Ė 1 (Ī©) in the H 1 -norm Ā· 1 , to conclude {fk ā D(Ī©)}, converging to f ā H Ė1 (Ī©). To make it more precise, we that the inequality holds for all f ā H need the estimate fĪµ ā¤ f which can be proved as follows:
fĪµ = f (x ā y)ĻĪµ (y)dy ā¤ f (Ā· ā y)ĻĪµ (y)dy = f , where we again used Minkowskiās inequality (4.53). If the fk ā D(Ī©), k = 1, 2, . . . , are chosen so that f ā fk 1 ā¤ 1/k, hence f ā fk ā¤ 1/k, then fĪµ ā (fk )Īµ = (f ā fk )Īµ ā¤ 1/k and, therefore, f ā fĪµ ā¤ f ā fk + fk ā (fk )Īµ + fĪµ ā (fk )Īµ ā¤ C1 Īµfk 1 +
2 C1 + 2 ā¤ C1 Īµf 1 + . k k
In the resulting inequality we can pass to the limit as k ā +ā, to conclude Ė( Ī©) with the same constant C1 . It follows that that (8.10) holds for f ā H F lies in C1 Īµ-neighborhood of FĪµ which ends the proof of Theorem 8.7. On a compact manifold M without boundary one can prove compactness of the embedding H s (M ) ā H s (M ) for any s, s ā R such that s > s . The embedding H s (Rn ) ā H s (Rn ) is never compact whatever s and s .
8.3. Dirichletās integral. The Friedrichs inequality
181
8.3. Dirichletās integral. The Friedrichs inequality Dirichletās integral is
n āu 2 dx, D(u) = āxj Ī©
where u ā H 1 (Ī©).
j=1
One of the possible physical interpretations of this integral is the potential energy for small vibrations of a string (for n = 1) or a membrane (for n = 2), where u(x) is the displacement (assumed to be small) from the equilibrium. (In case of the membrane we should assume that the membrane points only move orthogonally to the coordinate plane. Besides, we will assume that the boundary of the membrane is ļ¬xed; i.e., it cannot move.) Taking any displacement function u, let us try to understand when u represents an equilibrium. In the case of mechanical systems with ļ¬nitely many degrees of freedom (see Section 2.5) the equilibriums are exactly the critical (or stationary) points of the potential energy. It is natural to expect that the same is true for systems with inļ¬nitely many degrees of freedom, such as a string or a membrane, so the equilibrium property means vanishing of the variation of the Dirichlet integral under inļ¬nitesimal deformations preserving u|āĪ© , i.e., the Euler equation for the functional D(u). We will see that it is just the Laplace equation. For the string this had actually been veriļ¬ed in Section 2.1, where the stringās Lagrangian was given. Let us verify this in the multidimensional case. ĀÆ Ī“u ā C ā (Ī©), ĀÆ and Ī“u|āĪ© = 0, Let us take, for simplicity, u ā C ā (Ī©), where Ī© is assumed to be bounded with a smooth boundary. Let us write explicitly the condition for D(u) to be stationary, Ī“D(u) = 0, where d Ī“D(u) = D(u + t Ī“u) . dt t=0
(See Section 2.5.) We have n
Ī“D(u) = 2 j=1
= ā2
Ī©
āu ā (Ī“u)dx āxj āxj
n ā 2u Ī© j=1
āx2j
Ī“udx = ā2
Īu(x) Ā· Ī“u(x)dx. Ī©
(The integral over āĪ© appearing due to integration by parts vanishes since ĀÆ and ļ¬xed u|āĪ© is Ī“u|āĪ© = 0.) Thus, Dirichletās integral for u ā C ā (Ī©) stationary if and only if Īu = 0. Therefore, the Dirichlet problem for the Laplace equation can be considered as a minimization problem for the Dirichlet integral in the class of functions such that u|āĪ© = Ļ, where Ļ is a ļ¬xed function on āĪ©.
182
8. Sobolev spaces. Dirichletās problem
The Laplace equation Īu(x) = 0 is also easy to derive for u ā H 1 (Ī©) if u is a stationary point of Dirichletās integral for permissible variations Ī“u ā C0ā (Ī©). The same argument ļ¬ts, but integration by parts should be replaced by the deļ¬nition of the distributional derivative. Note that, in fact, the stationary points of Dirichletās integral are the points of minima (this is clear from what will follow). Therefore, Dirichletās problem may be considered as a minimization problem for Dirichletās integral in the class of functions with ļ¬xed boundary values. Proposition 8.8. Let Ī© be a bounded domain in Rn . Then there exists a constant C > 0, such that
n āu(x) 2 |u(x)| dx ā¤ C āxj dx, 2
(8.11) Ī©
Ė1 (Ī©). uāH
Ī© j=1
This inequality is called the Friedrichs inequality. Proof. Both parts of (8.11) continuously depend on u ā H 1 (Ī©) in the norm Ė1 (Ī©) is the closure of D(Ī©) in H 1 (Ī©), it is clear that it of H 1 (Ī©). Since H suļ¬ces to prove (8.11) for u ā D(Ī©) with C independent of u. Let Ī© be contained in the strip {x : 0 < xn < R} (this can always be achieved by taking a suļ¬ciently large R and shifting Ī© in Rn with the help of a translation that does not aļ¬ect (8.11)). For u ā D(Ī©) we have
xn āu (x , t)dt. u(x) = u(x , xn ) = āx n 0 By the Cauchy-Schwarz inequality, this implies
xn
R āu 2 āu(x) 2 2 |u(x)| ā¤ |xn | Ā· āxn (x , t) dt ā¤ R āxn dxn . 0 0 Integrating this inequality over Ī©, we get
āu 2 |u(x)| dx ā¤ R āxn dx. 2
Ī©
2
Ī©
In particular, this implies that (8.11) holds for u ā D(Ī©) with a constant C = R2 . The Friedrichs inequality implies that, for a bounded open set Ī©, the Ė 1 (Ī©) is equivalent to the norm deļ¬ned by Dirichletās integral; norm Ā·1 in H 1 i.e., u ā D(u) 2 . Indeed, u21 = u2 + D(u),
u ā H 1 (Ī©),
8.4. Dirichletās problem (generalized solutions)
183
where Ā· is the norm in L2 (Ī©). The Friedrichs inequality means that Ė1 (Ī©), implying u2 ā¤ CD(u), for all u ā H D(u) ā¤ u21 ā¤ C1 D(u),
Ė1 (Ī©), uāH
as required. Together with Dirichletās integral, we will use the corresponding inner product
n āu āv dx, u, v ā H 1 (Ī©). [u, v] = āx āx j j Ī© j=1
Ė1 (Ī©), H
Ė 1 (Ī©) is a closed On it deļ¬nes a Hilbert space structure, since H Ė1 (Ī©) to the norm [u, u] 21 = subspace in H 1 (Ī©) and u1 is equivalent in H 1 D(u) 2 .
8.4. Dirichletās problem (generalized solutions) Consider two generalized settings of Dirichletās boundary value problem for the Laplace operator. In both cases, the existence and uniqueness of a solution easily follows, after the above preliminary argument, from general theorems of functional analysis. The ļ¬rst of the settings contains, as a classical analogue, the following problem (the boundary value problem with zero boundary condition for the Poisson equation): Īu(x) = f (x), x ā Ī©, (8.12) u|āĪ© = 0. Instead of u|āĪ© = 0, write (8.13)
Ė 1 (Ī©). uāH
ĀÆ u|āĪ© = 0. If Now let us try to make sense of Īu = f . First, let u ā C 2 (Ī©), v ā D(Ī©), then, integrating by parts, we get
n āu āĀÆ v (Īu(x))ĀÆ v(x)dx = ā dx = ā[u, v]. āx āx j j Ī© Ī© j=1
If Īu = f , this means that (8.14)
[u, v] = ā(f, v),
where (Ā·, Ā·) is the inner product in L2 (Ī©). Now note that if both parts of (8.14) are considered as functionals of v, then these functionals are linear and continuous with respect to the norm Ā· 1 in H 1 (Ī©). Therefore, (8.14) Ė1 (Ī©). holds when v belongs to the closure of D(Ī©) in H 1 (Ī©), i.e., when v ā H
184
8. Sobolev spaces. Dirichletās problem
Conversely, let Ī© be bounded and have a smooth boundary, let u ā % Ė1 Ė1 (Ī©). Let us H (Ī©), f ā L2 (Ī©), and let (8.14) hold for any v ā H prove that u is a solution of (8.12). Ė1 (Ī©) implies u|āĪ© = 0. This follows from the First, observe that u ā H fact that, due to Theorem 8.5, the map u ā u|āĪ© extends to a continuous map H 1 (Ī©) āā L2 (āĪ©) which maps all functions from D(Ī©) and, therefore, Ė1 (Ī©) to zero. Note that here we only use, actually, the all functions from H fact that ĀÆ ā©H Ė1 (Ī©) =ā u|āĪ© = 0. u ā C 2 (Ī©) ĀÆ C 2 (Ī©)
This fact can be obtained directly, avoiding Theorem 8.5 and its generalization to manifolds. Let us sketch the corresponding argument, omitting the details. Using a partition of unity and locally rectifying the boundary with the help of a diļ¬eomorphism, we can, by the same arguments as those used in the proof of the Friedrichs inequality, get
|u(x)|2 dSx ā¤ CD(u) āĪ©
ĀÆ with the support in a small neighborhood of a ļ¬xed point of for u ā C 1 (Ī©) the boundary. Therefore, if u|āĪ© = 0, then it is impossible to approximate u by functions from D(Ī©) with respect to the norm in H 1 (Ī©). Thus, if ĀÆ %H Ė 1 (Ī©), then u|āĪ© = 0. u ā C 2 (Ī©) ĀÆ then Conversely, if u ā C 1 (Ī©), Ė1 (Ī©). u|āĪ© = 0 =ā u ā H This may be proved as follows. The Īµ-molliļ¬ed characteristic function of Ī©2Īµ , where Ī©Ī“ = {x : x ā Ī©, Ļ(x, āĪ©) ā„ Ī“}, is a function ĻĪµ ā D(Ī©), such that ĻĪµ (x) = 1 for x ā Ī©3Īµ and |ā Ī± ĻĪµ (x)| ā¤ ĀÆ then CĪ± Īµā|Ī±| (the constant CĪ± does not depend on Īµ). Now, if u ā C 1 (Ī©), 1 Ė ĻĪµ u ā H (Ī©), and u|āĪ© = 0 implies that if x ā Ī©\Ī©3Īµ , then |u(x)| ā¤ CĪµ, where C hereafter depends on u but does not depend on Īµ. This enables us to estimate u ā ĻĪµ u21 = u ā ĻĪµ u2 + D(u ā ĻĪµ u), where Ā· is the norm in L2 (Ī©) and D is the Dirichlet integral. Since the Lebesgue measure of Ī©\Ī©3Īµ does not exceed CĪµ, we have u ā ĻĪµ u2 ā¤ CĪµ sup |u ā ĻĪµ u|2 ā¤ C1 Īµ3 Ī©\Ī©3Īµ
8.4. Dirichletās problem (generalized solutions)
185
and
2 n ā D(u ā ĻĪµ u) ā¤ mes(Ī©\Ī©3Īµ ) Ā· sup (u ā Ļ u) Īµ āxj Ī©\Ī©3Īµ j=1
2 n āu 2 ā 2 2 ā¤ CĪµ sup āxj |(1 ā ĻĪµ )| + |u| āxj (1 ā ĻĪµ ) Ī©\Ī©3Īµ j=1
ā¤ C1 Īµ, where C1 does not depend upon Īµ (but depends upon u). The product of u and āxā j (1 ā ĻĪµ ) is bounded because in the 3Īµ-neighborhood of the boundary we have u = O(Īµ) whereas the derivative is O(Īµā1 ). Thus, it is possible to approximate u with respect to the norm Ā· 1 by ĀÆ whose supports are compact in Ī©. In turn, they can functions from C 1 (Ī©) be easily approximated with respect to the norm Ā· 1 by functions from D(Ī©) with the help of mollifying. ĀÆ for a bounded open set Ī© with a smooth Thus, supposing u ā C 2 (Ī©) Ė1 (Ī©). This justiļ¬es the boundary, we get u|āĪ© = 0 if and only if u ā H replacement of the boundary conditions (8.12) with (8.13). Ė1 (Ī©) and any v ā H Ė1 (Ī©) if and only Further, (8.14) holds for u ā H if Īu = f , where Ī is understood in the distributional sense. Indeed, if (8.14) holds, then, taking v ā D(Ī©) and moving the derivatives from u to v (using the deļ¬nition of the distributional derivatives) we obtain that (u, Īv) = (f, v) and that Īu = f , because v can be an arbitrary function from D(Ī©). Ė1 (Ī©), and f ā L2 (Ī©), then (8.14) holds for Conversely, if Īu = f , u ā H Ė1 (Ī©). v ā D(Ī©), hence, by continuity, for any v ā H All this justiļ¬es the following generalized setting of the problem (8.12): ā¢ Let Ī© be an arbitrary bounded open set in Rn , f ā L2 (Ī©). Find u ā Ė1 (Ī©) (or, equivalently, Ė1 (Ī©), such that (8.14) holds for any v ā H H for any v ā D(Ī©)). In this case u is called a generalized solution of the problem (8.12). We have seen above that if Ī© is a bounded open set with a smooth boundary ĀÆ then u is a generalized solution of (8.12) if and only if it is and u ā C 2 (Ī©), a solution of this problem in the classical sense. Theorem 8.9. A generalized solution of (8.12) exists and is unique. Proof. Rewrite (8.14) in the form [v, u] = ā(v, f )
186
8. Sobolev spaces. Dirichletās problem
Ė1 (Ī©). This and consider the right-hand side as a functional in v for v ā H is a linear continuous functional. Its linearity is obvious and its continuity follows from the Cauchy-Schwarz inequality: |(v, f )| ā¤ f Ā· v ā¤ f Ā· v1 , where Ā· is the norm in L2 (Ī©). By the Riesz representation theorem we may (uniquely) write this funcĖ1 (Ī©). This proves tional in the form [v, u], where u is a ļ¬xed element of H the unique solvability of (8.12) (in the generalized setting). The generalized setting of the Dirichlet problem leaves two important questions open: 1) the exact description of the class of all functions u which appear as the solutions when f runs through all possible functions from L2 (Ī©), 2) the existence of a more regular solution for a more regular f . Observe that local regularity problems are easily solved by the remark that if E(x) is a fundamental solution of the Laplace operator in Rn , then the convolution
E(x ā y)f (y)dy u0 (x) = (E ā f )(x) = Ī©
(deļ¬ned for almost all x) is a solution of Īu0 = f . Therefore, Ī(uāu0 ) = 0; i.e., u ā u0 is an analytic function in Ī©. Therefore, the regularity of u coincides locally with that of u0 and, therefore, can be easily described (e.g., if f ā C ā (Ī©), then u ā C ā (Ī©), as proved above). The question of regularity near the boundary is more complicated. Moreover, for nonsmooth boundary the answer is not completely clear even now. However, if āĪ© is suļ¬ciently smooth (say, of class C ā ), there are exact (though not easy to prove) theorems describing the regularity of u in terms of the regularity of f . We will formulate one such theorems without proof. Theorem 8.10. Let the boundary of Ī© be of class C ā . If f ā H s (Ī©), Ė1 (Ī©). s ā Z+ , then u ā H s+2 (Ī©) ā© H (For the proof see, e.g., Gilbarg and Trudinger [9, Sect. 8.4].) Conversely, it is clear that if u ā H s+2 (Ī©), then Īu = f ā H s (Ī©). Therefore, Ė1 (Ī©) āā H s (Ī©), s ā Z+ , Ī : H s+2 (Ī©) ā© H is an isomorphism. In particular, in this case all generalized solutions (for Ė1 (Ī©). f ā L2 (Ī©)) lie in H 2 (Ī©) ā© H
8.4. Dirichletās problem (generalized solutions)
187
Another version of an exact description of the regularity, which we already discussed above, is Schauderās theorem, Theorem 6.22, in HĀØolder functional spaces. There are numerous other descriptions, depending on the choice of functional spaces. Most of them are useful in miscellaneous problems of mathematical physics. ĀÆ even locally. Note that the last statement fails in the usual classes C k (Ī©) 2 ĀÆ ĀÆ Namely, f ā C(Ī©) does not even imply u ā C (Ī©). Let us pass to the generalized setting of the usual Dirichlet problem for the Laplace equation: Īu = 0, (8.15) u|āĪ© = Ļ, where Ī© is a bounded open set in Rn . First, there arises a question of interpretation of the boundary condition. We will do it by assuming that there exists a function v ā H 1 (Ī©), such that v|āĪ© = Ļ, and formulating Ė 1 (Ī©). This implies the boundary condition for u as follows: u ā v ā H 1 u ā H (Ī©). Introducing v saves us from describing the smoothness of Ļ (which requires using Sobolev spaces with noninteger indices on āĪ© if Ī© has a smooth boundary). Besides, it becomes possible to consider the case of domains with nonsmooth boundaries. Let us give the generalized formulation of Dirichletās problem (8.15): ā¢ Given a bounded domain Ī© ā Rn and v ā H 1 (Ī©), ļ¬nd a function Ė1 (Ī©). u ā H 1 (Ī©), such that Īu = 0 and u ā v ā H At ļ¬rst sight this problem seems to be reducible to the above one if Ė1 (Ī©) and put f = Ī(u ā v) = āĪv. But we have no we seek u ā v ā H reason to think that Īv ā L2 (Ī©), and we did not consider a more general setting. Therefore, we will consider this problem independently of the ļ¬rst problem, especially because its solution is as simple as the solution of the ļ¬rst problem. The reasoning above makes it clear that if Ī© has a smooth boundary, ĀÆ and u is a generalized solution of (8.15), ĀÆ v|āĪ© = Ļ, u ā C 2 (Ī©), v ā C(Ī©), then u is a classical solution of this problem. Theorem 8.11. A generalized solution u of Dirichletās problem exists and is unique for any bounded domain Ī© ā Rn and any v ā H 1 (Ī©). This solution has strictly and globally minimal Dirichlet integral D(u) among all Ė1 (Ī©). u ā H 1 (Ī©) such that u ā v ā H Conversely, if u is a critical (stationary) point of the Dirichlet integral Ė1 (Ī©), then the Dirichlet in the class of all u ā H 1 (Ī©) such that u ā v ā H integral actually reaches the strict and global minimum at u and u is a generalized solution of Dirichletās problem.
188
8. Sobolev spaces. Dirichletās problem
Proof. Let u ā H 1 (Ī©). The equation Īu = 0 can be expressed in the form (u, Īw) = 0,
w ā D(Ī©),
or, equivalently, in the form [u, w] = 0,
w ā D(Ī©).
Since [u, w] is a continuous functional in w on H 1 (Ī©), then, by continuity, we get Ė1 (Ī©). (8.16) [u, w] = 0, w ā H This condition is equivalent to the Laplace equation Īu = 0. Now we can invoke a geometric intepretation of the problem: we have Ė1 (Ī©) with respect to [Ā·, Ā·] and to ļ¬nd u ā H 1 (Ī©), which is orthogonal to H 1 Ė (Ī©). Imagining for a moment that H 1 (Ī©) is a Hilbert such that v ā u ā H space with respect to the inner product [Ā·, Ā·], we see that u is the component in the decomposition v = (v ā u) + u in the direct orthogonal decomposition Ė1 (Ī©) + (H Ė1 (Ī©))ā„ . H 1 (Ī©) = H It is well known from functional analysis that such a decomposition exists and is unique (in any Hilbert space and for any closed subspace of this Ė1 (Ī©)) component u Hilbert space), and, besides, the perpendicular (to H has the strictly minimal norm compared with all other decompositions v = Ė 1 (Ī©). Vice versa, in this case, if u is stationary, (v ā w) + w with v ā w ā H Ė1 (Ī©). then it is orthogonal to H However, the āinner productā [Ā·, Ā·] deļ¬nes a Hilbert space structure on Ė 1 (Ī©), but not on H 1 (Ī©) (it is not strictly positive; for example, [1, 1] = 0 H whereas 1 = 0). Therefore, the above argument should be made more Ė1 (Ī©) so that v = u + z. It follows from precise. Denote z = v ā u ā H (8.16) that (8.17)
[w, v] = [w, z],
Ė1 (Ī©). wāH
Now it is easy to prove the existence of the solution. Namely, let us consider Ė1 (Ī©). By the Riesz theorem [w, v] as a linear continuous functional in w on H it can be uniquely represented in the form [w, z], where z is a ļ¬xed element Ė1 (Ī©). It remains to set u = v ā z. The deļ¬nition of z implies (8.17) of H Ė1 (Ī©) is obvious. which in turn implies (8.16). The condition u ā v ā H The uniqueness of the solution is clear, since if u1 , u2 are two solutions, Ė 1 (Ī©) satisļ¬es (8.16), and this yields u = 0. then u = u1 ā u2 ā H Let us now prove that the Dirichlet integral is minimal on the solution Ė1 (Ī©). Set z = u1 ā u; u in the class of all u1 ā H 1 (Ī©) such that u1 ā v ā H 1 Ė hence, u1 = u + z. Clearly, z ā H (Ī©) and (8.16) yields D(u1 ) = [u1 , u1 ] = [u, u] + [z, z] = D(u) + D(z) ā„ D(u),
8.4. Dirichletās problem (generalized solutions)
189
where the equality is only attained for D(z) = 0, i.e., for z = 0 or, equivalently, for u = u1 . Finally, let us verify that if Dirichletās integral is stationary on u ā H 1 (Ī©) Ė1 (Ī©), then u is a in the class of all u1 ā H 1 (Ī©) such that u1 ā v ā H generalized solution of the Dirichlet problem (hence the Dirichlet integral reaches the strict global minimum at u). The stationarity property of the Dirichlet integral at u means that d Ė1 (Ī©). D(u + tz)|t=0 = 0, z ā H dt But & ' d d D(u + tz) [u + tz, u + tz] = = [u, z] + [z, u] = 2 Re[u, z]. dt dt t=0
t=0
Ė1 (Ī©). Replacing The stationarity property gives Re[u, z] = 0 for any z ā H Ė1 (Ī©), yielding (8.16). Theorem 8.11 z by iz, we get [u, z] = 0 for any z ā H is proved. Finally, we will give without proof a precise description of the regularity of solutions from the data on Ļ in the boundary value problem (8.15) for a domain Ī© with a smooth boundary. It was already mentioned that if 1 u ā H s (Ī©), s ā Z+ , s ā„ 1, then u|āĪ© ā H sā 2 (āĪ©). It turns out that the 1 converse also holds: if Ļ ā H sā 2 (āĪ©), where s ā Z+ , s ā„ 1, then there exists a unique solution u ā H s (Ī©) of the problem (8.15) (see, e.g., Lions and Magenes [22]). Similarly, if Ļ ā C m,Ī³ (āĪ©), where m ā Z+ , Ī³ ā (0, 1), then there exists ĀÆ a (unique) solution u ā C m,Ī³ (Ī©). Combining the described regularity properties of solutions of problems (8.12) and (8.15), it is easy to prove that the map u ā {Īu, u|Ī } can be extended to a topological isomorphism 1
H s (Ī©) āā H sā2 (Ī©) ā H sā 2 (āĪ©), and also to a topological isomorphism ĀÆ āā C m,Ī³ (Ī©) ĀÆ ā C m+2,Ī³ (āĪ©), C m+2,Ī³ (Ī©)
s ā Z+ , s ā„ 2, m ā Z+ , Ī³ ā (0, 1).
(See, e.g., the above-quoted books Gilbarg and Trudinger [9] and Ladyzhenskaya and Uralātseva [19].) Similar theorems can also be proved for general elliptic equations with general boundary conditions satisfying an algebraic condition called the ellipticity condition (or coercitivity, or Shapiro-Lopatinsky condition). For the general theory of elliptic boundary value problems, see, e.g., books by Lions and Magenes [22], HĀØormander [13, Ch. X], or [15, vol. 3, Ch. 20].
190
8. Sobolev spaces. Dirichletās problem
8.5. Problems 8.1. Using the Fourier method, solve the following Dirichlet problem: ļ¬nd a function u = u(x, y) such that Īu = 0 in the disc r ā¤ a, where r = x2 + y 2 , u|r=a = (x4 ā y 4 )|r=a . 8.2. By the Fourier method solve the Dirichlet problem in the disk r ā¤ a for a smooth boundary function f (Ļ), where Ļ is the polar angle; i.e., ļ¬nd a function u = u(x, y) such that Īu = 0 in the disc r ā¤ a and u|r=a = f (Ļ). Justify the solution. Derive for this case Poissonās formula:
2Ļ (a2 ā r2 )f (Ī±)dĪ± 1 . u(r, Ļ) = 2Ļ 0 a2 ā 2ar cos(Ļ ā Ī±) + r2 8.3. Using the positivity of Poissonās kernel K(r, Ļ, Ī±) =
2Ļ[a2
a2 ā r 2 , ā 2ar cos(Ļ ā Ī±) + r2 ]
prove the maximum principle for the solutions of Dirichletās problem obtained in the preceding problem. Derive from the maximum principle the solvability of Dirichletās problem for any continuous function f (Ļ). 8.4. Find a function u = u(x, y) such that Īu = x2 ā y 2 for r ā¤ a and u|r=a = 0. 8.5. In the ring 0 < a ā¤ r ā¤ b < +ā, ļ¬nd a function u = u(x, y) such that 2 Īu = 0, u|r=a = 1, āu ār r=b = (cos Ļ) . 8.6. By the Fourier method solve the Dirichlet problem for the Laplace equation in the rectangle 0 ā¤ x ā¤ a, 0 ā¤ y ā¤ b with the boundary conditions of the form u|x=0 = Ļ0 (y), u|x=a = Ļ1 (y) (0 ā¤ y ā¤ b), u|y=0 = Ļ0 (x), u|y=b = Ļ1 (x) (0 ā¤ x ā¤ a), where (the continuity condition for the boundary function) Ļ0 (0) = Ļ0 (0), Ļ0 (b) = Ļ1 (0), Ļ1 (0) = Ļ0 (a), Ļ1 (b) = Ļ1 (a). Describe a way for solving the general problem, and then solve the problem in the following particular case: Ļx , Ļ1 (y) = Ļ1 (x) = 0. Ļ0 (y) = Ay(b ā y), Ļ0 (x) = B sin a 8.7. In the half-plane {(x, y) : y ā„ 0} solve Dirichletās problem for the Laplace equation in the class of bounded continous functions. For this purpose use the Fourier transform with respect to x to derive the following
8.5. Problems
191
formula for the solutions which are in L2 (R) with respect to x:
1 ā y u(x, y) = Ļ(z)dz, 2 Ļ āā y + (x ā z)2 where Ļ(x) = u(x, 0). Then investigate the case of a general bounded continuous function Ļ. 8.8. Investigate for which n, s, and Ī± (Ī± > ān) the function u(x) = |x|Ī± belongs to H s (B1 ), where B1 is the open ball of radius 1 with the center at the origin of Rn . Ė1 ((ā1, 1))? 8.9. Does the function u(x) ā” 1, where x ā (ā1, 1), belong to H 8.10. Prove that the inclusion H s (Rn ) ā L2 (Rn ), s > 0, is not a compact operator.
10.1090/gsm/205/09
Chapter 9
The eigenvalues and eigenfunctions of the Laplace operator
9.1. Symmetric and selfadjoint operators in Hilbert space Let H be a Hilbert space with inner product (, ). Recall that a linear operator in H is a collection of two objects: a) a linear (but not necessarily closed) subspace D(A) ā H called the domain of the operator A, b) a linear map A : D(A) āā H. In this case we will write, by an abuse of notation, that a linear operator A : H āā H is deļ¬ned, though, strictly speaking, there is no map H āā H, because in general A may not be deļ¬ned on the whole of H. It is possible to add linear operators and take their products (compositions). Namely, given two linear operators A1 , A2 in H, we set D(A1 + A2 ) = D(A1 ) ā© D(A2 ), (A1 + A2 )x = A1 x + A2 x
for x ā D(A1 + A2 ).
Further, set D(A1 A2 ) = {x : x ā D(A2 ), A2 x ā D(A1 )}, (A1 A2 )x = A1 (A2 x)
for x ā D(A1 A2 ). 193
194
9. Eigenvalues and eigenfunctions
Deļ¬ne also Ker A = {x : x ā D(A), Ax = 0}, Im A = {x : x = Ay; y ā D(A)}. If Ker A = 0, then the inverse operator Aā1 is deļ¬ned. Namely, set D(Aā1 ) = Im A and Aā1 x = y for x = Ay. If there are two linear operators A, B in H, such that D(A) ā D(B) and Ax = Bx for x ā D(A), then we say that B is an extension of A and write A ā B. An operator A is called symmetric if (Ax, y) = (x, Ay),
x, y ā D(A).
In what follows, we will also introduce the notion of a selfadjoint operator. (For unbounded operators symmetry and selfadjointness need not coincide!) Let A be a linear operator such that D(A) is dense in H. Then the adjoint operator Aā is deļ¬ned via the identity (9.1)
(Ax, y) = (x, Aā y).
More precisely, let D(Aā ) consist of y ā H such that there exists z ā H satisfying (Ax, y) = (x, z) for x ā D(A). Since D(A) is dense in H, we see that z is uniquely deļ¬ned and we set Aā y = z. If the domain of A is dense and A is symmetric, then, clearly, A ā Aā . An operator A is selfadjoint if Aā = A (it is understood that the domain of A is dense and D(Aā ) = D(A)). Clearly, any selfadjoint operator is symmetric. The converse is not, generally, true. Deļ¬ne the graph of A as the set GA = {x, Ax : x ā D(A)} ā H Ć H. An operator is closed if its graph is closed in H Ć H, i.e., if xn ā D(A), xn ā x, and Axn ā y (the convergence is understood with respect to the norm in H) yield x ā D(A) and Ax = y. A selfadjoint operator is closed. A more general fact is true: the operator Aā is always closed for any A. Indeed, if yn ā D(Aā ), yn ā y, and Aā yn ā z, then passing to the limit as n ā ā in the identity (Ax, yn ) = (x, Aā yn ),
x ā D(A),
we get (Ax, y) = (x, z), yielding y ā D(Aā ) and Aā y = z.
x ā D(A),
9.1. Symmetric and selfadjoint operators
195
If D(A) is a closed subspace in H (in particular, if D(A) = H, i.e., A is deļ¬ned everywhere), then the closedness of A is equivalent to its boundedness, due to the closed graph theorem. In particular, a selfadjoint everywhere deļ¬ned operator A is necessarily bounded. Let us indicate a way of constructing selfadjoint operators. Proposition 9.1. Let A be a selfadjoint operator in H and let Ker A = 0. Then Aā1 is selfadjoint. Proof. Denote by E ā„ the orthogonal complement to E ā H; namely E ā„ = {y : y ā H, (x, y) = 0 for all x ā E}. Given a linear operator A with D(A) dense in H, we easily derive from the deļ¬nition of Aā that (9.2)
Ker Aā = (Im A)ā„ .
Indeed, the equality (Ax, y) = 0 for all x ā D(A) means, due to (9.1), exactly that y ā D(Aā ) and Aā y = 0. Note that E ā„ = 0 if and only if the linear span of E is dense in H. Thus if A is selfadjoint and Ker A = {0}, then from (9.2) it follows that D(Aā1 ) = Im A is dense in H. Therefore, Aā1 and (Aā1 )ā are deļ¬ned. It remains to verify that (Aā1 )ā = Aā1 . We have y ā D((Aā1 )ā ) and (Aā1 )ā y = z if and only if (9.3)
(Aā1 x, y) = (x, z),
x ā Im A.
Setting x = Af for f ā D(A), we see that (9.3) is equivalent to (f, y) = (Af, z),
f ā D(A),
yielding z ā D(Aā ) and Aā z = y. But, since Aā = A, this means that z ā D(A) and Az = y which may also be rewritten as z = Aā1 y. We have proved that y ā D((Aā1 )ā ) with (Aā1 )ā y = z is equivalent to y ā D((Aā1 )) with Aā1 y = z. Therefore, (Aā1 )ā = Aā1 as required. Corollary 9.2. If B is a bounded everywhere deļ¬ned symmetric operator with Ker B = 0, then A = B ā1 is selfadjoint. The importance of selfadjoint operators follows from the spectral theorem. Before we formulate it, let us give an example of a selfadjoint operator. Let M be a space with measure Ī¼; i.e., Ī¼ is a countably additive measure on a Ļ-algebra of subsets of M . Construct in the usual way the space L2 (M, dĪ¼) consisting of classes of measurable functions on M , such that the square of their absolute value is integrable. (Two functions are in the same class if they coincide almost everywhere.) Then L2 (M, dĪ¼) is a Hilbert space. Let a(m) be a real-valued measurable and almost everywhere ļ¬nite
196
9. Eigenvalues and eigenfunctions
function on M . The operator A of multiplication by a(m) is deļ¬ned by the formula Af (m) = a(m)f (m) for f ā D(A), where D(A) := {f : f ā L2 (M, dĪ¼), af ā L2 (M, dĪ¼)}. This operator is always selfadjoint. Indeed, its symmetry is obvious and we have only to verify that D(Aā ) = D(A). Let u ā D(Aā ), Aā u = v. Let us check that u ā D(A) and a(m)u(m) = v(m) for almost all m. Consider the set MN := {m : m ā M, |a(m)| ā¤ N }, and let ĻN (m) = 1 for m ā MN and ĻN (m) = 0 for m ā MN . The identity Aā u = v means that
a(m)f (m)u(m)dĪ¼ = f (m)v(m)dĪ¼ for any f ā D(A). M
M
then ĻN g ā D(A), implying Note that if g ā
a(m)g(m)u(m)dĪ¼ = g(m)v(m)dĪ¼ for any g ā L2 (MN , dĪ¼). L2 (M, dĪ¼),
MN
MN
But it is clear that (au)|MN ā L2 (MN , dĪ¼) since a is bounded on MN . Therefore, the arbitrariness of g implies au|MN = 8 = ā MN , then au| = v| . v|MN (almost everywhere). Therefore, if M N =1 M M 8 is equal to Since a(m) is almost everywhere ļ¬nite, then the measure of M \M 0. Hence, a(m)u(m) = v(m) almost everywhere. In particular, a(m)u(m) ā L2 (M, dĪ¼); i.e., u ā D(A) and au = v, as required. Here is an important example. Let l2 be the Hilbert space ā§ ā« ā āØ ā¬ |xj |2 < +ā , l2 = x = {x1 , x2 , . . .} : xj ā C, ā© ā j=1
with inner product ({x1 , . . . }, {y1 , . . . }) =
ā
xj yj .
j=1
For any sequence of real numbers Ī»1 , Ī»2 , . . . , consider the operator A in l2 which is deļ¬ned as follows: ā« ā§ ā ā¬ āØ |Ī»j |2 |xj |2 < +ā , D(A) = {x1 , x2 , . . .}, xj ā C, ā ā© j=1
A({x1 , x2 , . . .}) = {Ī»1 x1 , Ī»2 x2 , . . .}.
9.2. The Friedrichs extension
197
To interpret this example as a multiplication operator, let M := {1, 2, . . . } and let the measure of each integer be equal to 1. Operators A1 : H1 āā H1 and A2 : H2 āā H2 are called unitarily equivalent if there exists a unitary (i.e., invertible and inner product preserving) operator U : H1 āā H2 such that U ā1 A2 U = A1 . (Recall that the equality of operators by deļ¬nition implies the coincidence of their domains.) In other words, in this notation the unitary equivalence of A1 and A2 means that the diagram A
1 āā H Hā 1 ā1 ā ā U< 0 to A0 , we get a new operator (denote it again by A0 ) such that (9.5)
(A0 Ļ, Ļ) ā„ Īµ(Ļ, Ļ),
Ļ ā D(A0 ).
9.2. The Friedrichs extension
199
We will assume from the very beginning that this stronger estimate (9.5) holds, since from the point of view of the eigenvalue problem the addition of (C + Īµ)I is of no importance, because it only shifts eigenvalues without aļ¬ecting eigenfunctions. Moreover, the Friedrichs inequality shows that in the above example (9.5) holds already. Set u, v ā D(A0 ).
[u, v] = (A0 u, v),
This scalar product deļ¬nes a pre-Hilbert structure on D(A0 ) (the positive deļ¬niteness follows from (9.5)). Denote by Ā· 1 the corresponding norm (i.e., u1 = [u, u]1/2 ). Then the convergence with respect to Ā· 1 implies convergence with respect to the norm Ā· in H. In particular, if a sequence Ļk ā D(A0 ) is a Cauchy sequence with respect to Ā· 1 , then it is a Cauchy sequence with respect to Ā· ; hence, Ļk converges in H. Denote by H1 the completion of D(A0 ) with respect to Ā· 1 . (In the Ė1 (Ī©).) The inequality (9.5) and the above arguments show example, H1 = H the existence of a natural map H1 āā H. Namely, the image g ā of g ā H1 is the limit in H of a sequence {Ļk }, Ļk ā D(A0 ), converging to g in H1 . Let us prove that this map is injective. First of all note that by continuity of scalar products [Ļ, g] = (A0 Ļ, g ā ),
Ļ ā D(A0 ).
Now, if g ā = 0, then, using the above-mentioned sequence {Ļk }, we obtain [g, g] = lim [Ļk , g] = lim (A0 Ļk , g ā ) = 0 kāā
kāā
implying g = 0, as required. Therefore, there is a natural inclusion H1 ā H, so we will identify the elements of H1 with the corresponding elements of H. Now set D(A) = H1 ā© D(Aā0 ) and A = Aā0 |D(A) . We get an operator A : H āā H called the Friedrichs extension of A0 . Theorem 9.3. The Friedrichs extension of a semibounded operator A0 is a selfadjoint operator A : H āā H. It is again semibounded from below and (Au, u) ā„ Īµ(u, u),
u ā D(A),
where Īµ > 0 is the same constant as in the similar estimate for A0 .
200
9. Eigenvalues and eigenfunctions
Proof. Let us verify that Aā1 exists and is everywhere deļ¬ned, bounded, and symmetric. This implies that A is selfadjoint (see Corollary 9.2). First, verify that (9.6)
(Au, v) = [u, v],
u, v ā D(A).
Indeed, this is true for u, v ā D(A0 ). By continuity this is also true for u ā D(A0 ), v ā H1 . Now, make use of the identity (A0 u, v) = (u, Aā0 v),
u ā D(A0 ), v ā D(Aā0 ).
In particular, this is true for u ā D(A0 ), v ā D(A). Then A0 u = Au, Aā0 v = Av, and therefore, we have (Au, v) = (u, Av) = [u, v],
u ā D(A0 ), v ā D(A).
But this, in particular, implies (u, Av) = [u, v],
u ā D(A0 ), v ā D(A),
where we may pass to the limit with respect to u (if the limit is taken in H1 ). We get (9.7)
(u, Av) = [u, v],
u ā H1 , v ā D(A).
In particular, this holds for u ā D(A), v ā D(A). Interchanging u and v, we get (9.6), implying (9.8)
(Au, v) = (u, Av) = [u, v],
u, v ā D(A).
In particular, A is symmetric and (9.9)
(Au, u) = [u, u] ā„ Īµ(u, u),
u ā D(A),
where Īµ > 0 is the same as in (9.5). It follows from (9.9) that Ker A = 0 and Aā1 is well-deļ¬ned. The symmetry of A implies that of Aā1 . It follows from (9.9) that Aā1 is bounded. Indeed, if u ā D(A), then u2 ā¤ Īµā1 (Au, u) ā¤ Īµā1 u Ā· Au, implying u ā¤ Īµā1 Au,
u ā D(A).
Setting v = Au, we get Aā1 v ā¤ Īµā1 v, which means the boundedness of Aā1 .
v ā D(Aā1 ),
9.3. Discreteness of spectrum in a bounded domain
201
Finally, let us prove that D(Aā1 ) = H, i.e., Im A = H. We want to prove that if f ā H, then there exists u ā D(A), such that Au = f . But this means that u ā H1 ā© D(Aā0 ) and Aā0 u = f ; i.e., (9.10)
(A0 Ļ, u) = (Ļ, f ),
Ļ ā D(A0 ).
Taking (9.8) into account, we may rewrite this in the form (9.11)
[Ļ, u] = (Ļ, f ),
Ļ ā D(A0 ).
Now, the Riesz theorem implies that for any f ā H there exists u ā H1 such that (9.11) holds. It remains to verify that u ā D(Aā0 ). But this is clear since, due to (9.7), we may rewrite (9.11) in the form (9.10), as required. Theorem 9.3 is proved. Now, let us pass to the model example. Let A0 be the operator āĪ on D(Ī©), where Ī© is a bounded domain in Rn . As we have already seen, the adjoint operator Aā0 maps u to āĪu for u ā L2 (Ī©) and Īu ā L2 (Ī©) (here Ī is understood in the distributional sense). Due to the Friedrichs inequality, A0 is positive; i.e., (9.5) holds. Let us construct the Friedrichs extension of A. This is an operator in L2 (Ī©), such that A0 ā A ā Aā0 and Ė1 (Ī©) ā© D(Aā ); D(A) = H 0 i.e., Ė1 (Ī©), Īu ā L2 (Ī©)}. D(A) = {u : u ā H By Theorem 9.3, it follows that A is selfadjoint and positive. It is usually called the selfadjoint operator in L2 (Ī©) deļ¬ned by āĪ and the Dirichlet boundary condition u|āĪ© = 0. The operator Aā1 is everywhere deļ¬ned and bounded.
9.3. Discreteness of spectrum for the Laplace operator in a bounded domain Let Ī© be a bounded open subset in Rn . Theorem 9.4. The spectrum of the operator A deļ¬ned in L2 (Ī©) by āĪ and the Dirichlet boundary condition is discrete. More precisely, there exists an Ė1 (Ī©) orthonormal basis of eigenvectors Ļj , j = 1, 2, . . . , such that Ļj ā H 2 and āĪĻj = Ī»j Ļj in L (Ī©), where Ī»j ā +ā as j ā +ā. Proof. First, let us make a general remark on the Friedrichs extensions. Let A0 be a positive symmetric operator on a Hilbert space H, A its Friedrichs extension, and H1 the Hilbert space obtained as the completion of D(A0 ) with respect to the norm Ļ1 = (A0 Ļ, Ļ)1/2 . Since D(A) ā H1 , it follows
202
9. Eigenvalues and eigenfunctions
that Aā1 maps H to H1 . Let us prove that Aā1 , considered as an operator from H to H1 , is continuous. The simplest way to deduce it is to use the closed graph theorem. Namely, verify that the graph of Aā1 : H āā H1 is closed; i.e., Ļk ā Ļ in H and Aā1 Ļk ā f in H1 imply that Aā1 Ļ = f . But, due to continuity of the embedding H1 ā H, the above implies that Aā1 Ļk ā f in H, and the continuity of Aā1 : H āā H yields Aā1 Ļk ā Aā1 Ļ in H; hence, Aā1 Ļ = f , as required. Ė1 (Ī©) and the embedding Let us return to our situation. We have H1 = H 1 2 Ė operator H (Ī©) ā L (Ī©) is not only continuous, but also compact. Therefore, Aā1 : L2 (Ī©) āā L2 (Ī©) as a composition of the continuous operator Ė1 (Ī©) and the compact embedding H Ė1 (Ī©) ā L2 (Ī©) is a Aā1 : L2 (Ī©) āā H compact selfadjoint operator. By the Hilbert-Schmidt theorem, Aā1 has an orthonormal eigenbasis Ļ1 , Ļ2 , . . . in L2 (Ī©), and if Ī¼j are the corresponding eigenvalues, i.e., Aā1 Ļj = Ī¼j Ļj , then Ī¼j ā 0 as j ā ā. Note that 0 is not an eigenvalue of Aā1 , since the deļ¬nition of Aā1 implies Ker Aā1 = 0. Further, Aā1 Ļj = Ī¼j Ļj may be rewritten as Ļj = Ī¼j AĻj or AĻj = Ī»j Ļj , where Ī»j = Ī¼ā1 j . The positivity of A implies that Ī»j > 0, and condition Ī¼j ā 0 yields Ī»j ā +ā as j ā +ā, as required.
9.4. Fundamental solution of the Helmholtz operator and the analyticity of eigenfunctions of the Laplace operator at the interior points. Besselās equation We would like to prove the analyticity of eigenfunctions of the Laplace operator at interior points of the domain. Since eigenvalues are positive, it suļ¬ces to prove analyticity of any solution u ā D (Ī©) of the equation (9.12)
Īu + k 2 u = 0,
k > 0,
called the Helmholtz equation. First, let us ļ¬nd explicitly a fundamental solution of the Helmholtz operator Ī+k 2 . Denote this fundamental solution (k) (k) by En (x). If it turns out that En (x) is analytic for x = 0, then any solution u ā D (Ī©) of the equation (9.12) is analytic in Ī©; cf. Theorem 5.12. Since Ī + k 2 commutes with rotations, it is natural to seek a spherically (k) symmetric fundamental solution. Let En (x) = f (r), where r = |x| for x = 0. Then f (r) for r > 0 satisļ¬es the ordinary diļ¬erential equation (9.13)
f (r) +
nā1 f (r) + k 2 f (r) = 0. r
(See how to compute Īf (r) in Example 4.10.) Clearly, to get a fundamental solution we should ļ¬nd a solution f (r) with the same singularity as r ā 0+ (0) as that of the fundamental solution En (x) = En (x) of the Laplace operator.
9.4. Fundamental solution of the Helmholtz operator
203
For instance, if it turns out that we have found a solution f (r) of the equation (9.13) such that f (r) = En (r)g(r), g ā C 2 ([0, +ā)), g(0) = 1, then the verbatim repetition of the veriļ¬cation that ĪEn (x) = Ī“(x) yields (Ī + k 2 )f (|x|) = Ī“(x). We will reduce (9.13) to the Bessel equation. This may be done even for a more general equation Ī± Ī² 2 (9.14) f (r) + f (r) + k + 2 f (r) = 0, r r where Ī±, Ī², k are arbitrary real (or even complex) constants. First, let us rewrite (9.14) in the form r2 f (r) + Ī±rf (r) + (Ī² 2 + k 2 r2 )f (r) = 0. Here the ļ¬rst three terms of the left-hand side above constitute Eulerās operator d d2 L = r2 2 + Ī±r + Ī², dr dr applied to f . The Euler operator commutes with the homotheties of the real line (i.e., with the transformations x ā Ī»x, with Ī» ā R \ 0, acting upon functions on R by the change of variables), and rĪŗ for any Īŗ ā C are its eigenfunctions. (It is also useful to notice that the change of variable r = et transforms Eulerās operator into an operator commuting with all translations with respect to t, i.e., into an operator with constant coeļ¬cients.) Therefore, it is natural to make the following change of the unknown function in (9.14): f (r) = rĪŗ z(r). We have
f (r) = rĪŗ z (r) + ĪŗrĪŗā1 z(r), f (r) = rĪŗ z (r) + 2ĪŗrĪŗā1 z (r) + Īŗ(Īŗ ā 1)rĪŗā2 z(r). Substituting this into (9.14) and dividing by rĪŗ , we get the following equation for z(r): & ' Ī± + 2Īŗ Īŗ(Īŗ + Ī± ā 1) + Ī² 2 z (r) + k + z(r) = 0. (9.15) z (r) + r r2 Here the parameter Īŗ is at our disposal. The ļ¬rst idea is to set Īŗ = āĪ±/2 in order to cancel z (r). Then we get c (9.16) z (r) + k 2 + 2 z(r) = 0. r It is easy to derive from (9.16) that for k = 0 the solutions z(r) of this equation behave, as r ā +ā, as solutions of the equation u (r) + k 2 u(r) = 0.
204
9. Eigenvalues and eigenfunctions
More precisely, let, for example, k > 0 (this case is most important to us). Then there exist solutions z1 (r) and z2 (r) of (9.16) whose asymptotics as r ā +ā are 1 1 (9.17) z1 (r) = sin kr + O , z2 (r) = cos kr + O . r r This can be proved by passing to an integral equation as we did when seeking asymptotics of eigenfunctions and eigenvalues of the Sturm-Liouville operator. Namely, let us rewrite (9.16) in the form z (r) + k 2 z(r) = ā
c z(r) r2
and seek z(r) in the form z(r) = C1 (r) cos kr + C2 (r) sin kr. The variation of parameters method for functions Cj (r) yields C1 (r) cos kr + C2 (r) sin kr = 0, ākC1 (r) sin kr + kC2 (r) cos kr = ā rc2 z(r), which imply
r
C1 (r) = A + r0
C2 (r) = B ā
r r0
c sin kĻ z(Ļ)dĻ, Ļ2 k c cos kĻ z(Ļ)dĻ. Ļ2 k
We wish that C1 (r) ā A and C2 (r) ā B as r ā +ā. But then it is natural to take r0 = +ā and write
c ā 1 (9.18) z(r) = A cos kr + B sin kr + sin k(r ā Ļ)z(Ļ)dĻ, k r Ļ2 which is an integral equation for z(r). Let us rewrite it in the form (I ā T )z = A cos kr + B sin kr, where T is the integral operator mapping z(r) to the last summand in (9.18). Consider this equation in the space Cb ([r0 , ā)) of continuous bounded functions on [r0 , ā). Let z = suprā„r0 |z(r)|. It is clear that
c |c| ā 1 dĻz = z. T z ā¤ 2 |k| r0 Ļ |k|r0 If r0 is suļ¬ciently large, then T < 1 and the integral equation (9.18) has a unique solution in Cb ([r0 , ā)). But its continuous bounded solutions are
9.4. Fundamental solution of the Helmholtz operator
205
solutions of (9.16), and if z(r) is such a solution, then
ā C 1 dĻ = ; |z(r) ā A cos kr ā B sin kr| ā¤ C 2 Ļ r r i.e., (9.19)
1 z(r) = A cos kr + B sin kr + O r
as r ā +ā.
Thus, for any A and B, the equation (9.16) has a solution with the asymptotic (9.19). In particular, there are solutions z1 (r) and z2 (r) with asymptotics (9.17). Clearly, these solutions are linearly independent. The asymptotic (9.19) may also be expressed in the form 1 , z(r) = A sin k(r ā r0 ) + O r which makes it clear that if z(r) is real (for instance, if all constants are real), then z(r) has inļ¬nitely many zeros with approximately the same behavior as that of the zeros of sin k(r ā r0 ), as r ā +ā. Now, return to (9.15) and choose the parameter Īŗ so as to have Ī± + 2Īŗ = 1. Then the equation for z(r) becomes & ' 1 Ī½2 2 (9.20) z (r) + z (r) + k ā 2 z(r) = 0, r r where Ī½ 2 = Īŗ2 ā Ī². (In particular, we can take Ī½ = Ā±Īŗ in the case Ī² = 0, which was our starting point; generally, Ī½ may be not real.) Put r = k ā1 x or x = kr. Introducing x as a new independent variable, we get instead of (9.20) the following equation for y(x) = z(k ā1 x) (or z(x) = y(kx)): 1 Ī½2 (9.21) y (x) + y (x) + 1 ā 2 y(x) = 0. x x (One should not confuse the real variable x > 0 here with the x ā Rn used earlier.) The equation (9.21) is called the Bessel equation and its solutions are called cylindric functions. Cylindric functions appear when we apply separation of variables (or the Fourier method) to ļ¬nd eigenfunctions of the Laplace operator in a disc, and hence solutions of the Dirichlet problem in a cylinder (over a disc) of ļ¬nite or inļ¬nite height; this accounts for the term ācylindric functionsā. Eliminating, as above, y (x) in (9.21), we see that the asymptotic of any cylindric function as x ā +ā has the form 1 A . (9.22) y(x) = ā sin(x ā x0 ) + O x x
206
9. Eigenvalues and eigenfunctions
Incidentally, substituting all parameters, we see that (9.15) in our case (for Ī± = 1, Īŗ = ā1/2, k = 1, Ī² = āĪ½ 2 ) takes the form 1 2 4 āĪ½ z (r) + 1 + z(r) = 0, r2 which makes it clear that for Ī½ = Ā± 12 the Bessel equation is explicitly solvable and the solutions are of the form coinciding with the ļ¬rst term in (9.22); i.e., 1 y(x) = ā (A cos x + B sin x). x In particular, for n = 3 we can solve explicitly the equation (9.13) that arose from the Helmholtz equation. Namely, setting Ī± = 2, Ī² = 0, Īŗ = ā1, we see that (9.15) takes the form z + k 2 z = 0, implying that for n = 3 the spherically symmetric solutions of the Helmholtz equation are of the form 1 f (x) = (A cos kr + B sin kr), r We get the fundamental solution if A = ā
eikr , 4Ļr
ā
1 4Ļ .
eāikr , 4Ļr
r = |x|.
In particular, functions
ā
cos kr 4Ļr
are fundamental solutions, and the function u = homogeneous equation
sin kr r
is a solution of the
(Ī + k 2 )u(x) = 0, x ā R3 . Let us return to the general equation (9.21). We should describe somehow the behavior of its solutions as x ā 0+. The best method for this is to 2 try to ļ¬nd the solution in the form of a series in x. Since xĪ½ 2 plays a more important role than 1 as x ā 0+, we may expect that y(x) behaves at zero like a solution of the Euler equation obtained from (9.21) by removing the 1 from the coeļ¬cient of y(x). But solutions of Eulerās equation are linear combinations of functions xĻ and, perhaps, xĻ ln x for appropriate Ļ. In our case, substituting xĻ into Eulerās equation, we get Ļ = Ā±Ī½. Therefore, it is natural to expect that we may ļ¬nd a pair of solutions of (9.21) that behave like xĀ±Ī½ as x ā +0. This is, however, true up to some corrections only. For the time being, let us seek solutions y(x) of (9.21) in the form (9.23)
y(x) = xĻ (a0 + a1 x + a2 x2 + Ā· Ā· Ā· ) = a0 xĻ + a1 xĻ+1 + Ā· Ā· Ā· .
9.4. Fundamental solution of the Helmholtz operator
207
Substituting this series in (9.21) and equating coeļ¬cients of all powers of x to zero, we get a0 Ļ(Ļ ā 1) + a0 Ļ ā a0 Ī½ 2 = 0, a1 (Ļ + 1) + a1 (Ļ + 1) ā a1 Ī½ 2 = 0, Ā·Ā·Ā· ak (Ļ + k)(Ļ + k ā 1) + ak (Ļ + k) + akā2 ā ak Ī½ 2 = 0, k = 2, 3, . . . , or (9.24)
a0 (Ļ 2 ā Ī½ 2 ) = 0,
(9.25)
a1 [(Ļ + 1)2 ā Ī½ 2 ] = 0, ak [(Ļ + k)2 ā Ī½ 2 ] + akā2 = 0,
(9.26)
k = 2, 3, . . . .
Without loss of generality, we may assume that a0 = 0 (for a0 = 0 we may replace Ļ by Ļ ā k, where k is the number of the ļ¬rst nonzero coeļ¬cient ak and then reenumerate the coeļ¬cients). But then (9.24) implies Ļ = Ā±Ī½. If Ī½ is neither integer nor half-integer, then (9.25) and (9.26) imply a1 = a3 = Ā· Ā· Ā· = 0, akā2 akā2 akā2 ak = ā =ā . =ā 2 2 (Ļ + k) ā Ī½ (Ļ + Ī½ + k)(Ļ ā Ī½ + k) k(k + 2Ļ) If k = 2m, then this yields a2m = āa2mā2 Set a0 =
(ā1)m a0 1 = Ā· Ā· Ā· = . 22 m(m + Ļ) 22m m!(m + Ļ)(m + Ļ ā 1) . . . (Ļ + 1)
1 2Ļ Ī(Ļ+1) .
Then, using the identity sĪ(s) = Ī(s + 1), we get a2m =
(ā1)m . 22m+Ļ m!Ī(m + Ļ + 1)
The corresponding series for Ļ = Ī½ is denoted by JĪ½ (x) and is called the Bessel or cylindric function of the ļ¬rst kind. Thus, ā
x 2k+Ī½ 1 (ā1)k . (9.27) JĪ½ (x) = k!Ī(k + Ī½ + 1) 2 k=0
Clearly, this series converges for all complex x = 0. All these calculations also make sense for an integer or half-integer Ļ = Ī½ if Ī½ ā„ 0. We could assume that Ī½ ā„ 0 without loss of generality, because only Ī½ 2 enters the equation. But it makes sense to consider Ī½ with Re Ī½ < 0 in the formulas for the solutions, to get the second solution supplementing the one given by (9.27). (In other words, we could take Ļ = āĪ½, returning to (9.23).) For Ī½ = ā1, ā2, . . ., the series (9.27) may be deļ¬ned by continuity, since Ī(z) has no zeros but only poles at z = 0, ā1, ā2, . . ..
208
9. Eigenvalues and eigenfunctions
So, the Bessel equation (9.21) for Re Ī½ ā„ 0 always has a solution of the form JĪ½ (x) = xĪ½ gĪ½ (x), where gĪ½ (x) is an entire analytic function of x, such 1
= 0. that gĪ½ (0) = Ī(Ī½+1) We will neither seek the second solution (in the cases when we did not do this yet) nor investigate other properties of cylindric functions. We will only remark that in reference books and textbooks on special functions, properties of cylindric functions are discussed with the same minute detail as the properties of trigonometric functions are discussed in textbooks on trigonometry. Cylindric functions are tabulated and are a part of standard mathematics software, such as Maple, Mathematica, and MATLAB. The same applies to many other special functions. Many of them are, as cylindric functions, solutions of certain second-order linear ordinary diļ¬erential equations. Let us return to Helmholtzās equation. We have seen that (9.13) reduces to Besselās equation. Indeed, it is of the form (9.14) with Ī± = n ā 1, Ī² = 0. 2ān Let us pass to (9.15) and set Īŗ = 1āĪ± 2 = 2 . Then we get (9.20), where & ' 2ān nā2 2 nā2 nā2+ = , Ī½ = 2 2 2 2
implying Ī½ = Ā± nā2 2 . Therefore, in particular, we get a solution of (9.13) of the form f (r) = r
2ān 2
J nā2 (kr). 2
Actually, this solution does not have any singularities (it expands in a series in powers of r2 ). We may also consider the solution f (r) = r
2ān 2
J 2ān (kr). 2
If n is odd, then, multiplying this function by a constant, we may get a fundamental solution for Helmholtzās operator (this solution is, actually, an elementary function). If n is even, then, together with J nā2 (x), we should 2 use the second solution of Besselās equation (it is called a cylindric function of the second kind or Neumannās function and its expansion contains ln x as x ā 0+). It can also be expressed in terms of J nā2 (x) via a quadrature 2 with the help of a well-known Wronskian trick (which gives for the second solution a ļ¬rst-order linear diļ¬erential equation). In any case, we see that (k) there exists a fundamental solution En (r), analytic in r for r = 0. This implies that all eigenfunctions of the Laplace operator are analytic at interior points of the domain.
9.5. Variational principle
209
Note that there is a more general theorem which guarantees the analyticity of any solution of an elliptic equation with analytic coeļ¬cients. Solutions of an elliptic equation with smooth (C ā ) coeļ¬cients are also smooth.
9.5. Variational principle. The behavior of eigenvalues under variation of the domain. Estimates of eigenvalues Let A be a selfadjoint semibounded from below operator in a separable Hilbert space H with discrete spectrum. Let Ļ1 , Ļ2 , . . . be a complete orthonormal system of eigenvectors of A with eigenvalues Ī»1 , Ī»2 , . . .; i.e., AĻj = Ī»j Ļj for j = 1, 2, . . .. The discreteness of the spectrum means that Ī»j ā +ā as j ā +ā. Therefore, we may assume that the eigenvalues increase with j; i.e., Ī»1 ā¤ Ī»2 ā¤ Ī»3 ā¤ Ā· Ā· Ā· . Instead of the set of eigenvalues, it is convenient to consider a (nonstrictly) increasing function N (Ī») deļ¬ned for Ī» ā R as the number of eigenvalues not exceeding Ī». This function takes only nonnegative integer values, is constant between eigenvalues, and at eigenvalues its jumps are equal to the multiplicities of these eigenvalues. It is continuous from the right; i.e., N (Ī») = N (Ī» + 0). The eigenvalues Ī»j are easily recovered from N (Ī»). Namely, if Ī» is a real number, then it is an eigenvalue of multiplicity N (Ī» + 0) ā N (Ī» ā 0). (If N (Ī» + 0) ā N (Ī» ā 0) = 0, then Ī» is not an eigenvalue.) If N (Ī» ā 0) < j ā¤ N (Ī» + 0) and j is a nonnegative integer, then Ī»j = Ī». In short, Ī»j is uniquely deļ¬ned from the condition N (Ī»j ā 0) < j ā¤ N (Ī»j + 0). The following proposition is important. Proposition 9.5 (Glazmanās lemma). (9.28)
N (Ī») = sup {dim L : (Au, u) ā¤ Ī»(u, u) for all u ā L ā D(A)} ,
where L ā D(A) denotes that L is a ļ¬nite-dimensional linear subspace of D(A). If A is the Friedrichs extension of A0 and Ī» is not an eigenvalue, then L ā D(A) may be replaced by the stronger condition L ā D(A0 ). Proof. Let LĪ» be a ļ¬nite-dimensional subspace in H spanned by all eigenvectors Ļj with eigenvalues Ī»j ā¤ Ī». Clearly, LĪ» ā D(A) and A ā Ī»I has only nonpositive eigenvalues on LĪ» and hence is nonpositive itself; i.e., (Au, u) ā¤ Ī»(u, u) for all u ā LĪ» . In particular, we may take L = LĪ» in the right-hand side of (9.28). But dim LĪ» = N (Ī»), implying that the left-hand side of (9.28) is not greater than the right-hand side. Now, let us verify that the left-hand side of (9.28) is not less than the right-hand side. Let L ā D(A) and (Au, u) ā¤ Ī»(u, u), u ā L. We wish to prove that dim L ā¤ N (Ī»). Fix such a subspace L and set MĪ» = Lā„ Ī» (the
210
9. Eigenvalues and eigenfunctions
orthogonal complement to LĪ» ). Clearly, if u ā MĪ» , then u can be expanded with respect to the eigenfunctions Ļj with the eigenvalues Ī»j > Ī». Therefore, if u ā D(A) and u = 0, then (Au, u) > Ī»(u, u) implying L ā© MĪ» = 0. Now, consider the projection EĪ» : H āā H onto LĪ» parallel to MĪ» ; i.e., if u = v + w, where v ā LĪ» , w ā MĪ» , then EĪ» u = v. Clearly, Ker EĪ» = MĪ» . Now, consider the operator EĪ» |L : L āā LĪ» . Since (Ker EĪ» ) ā© L = 0, then EĪ» |L is a monomorphism. Therefore, dim L ā¤ dim LĪ» = N (Ī»), as required. It remains to verify the last statement of the proposition. Without loss of generality, we may assume that (Au, u) ā„ (u, u), for u ā D(A0 ). Let us make use of the Hilbert space H1 , the completion of D(A0 ) with respect to the norm deļ¬ned by the inner product [u, v] = (A0 u, v), where u, v ā D(A0 ). Instead of (Au, u) in (9.28), we may write [u, u]. Let Ā·1 be the norm in H1 . Let L be any ļ¬nite-dimensional subspace of D(A). In particular, L ā H1 . In L, choose an orthonormal (with respect to the inner product in H) basis e1 , . . . , ep and let vectors f1 , . . . , fp ā D(A0 ) be such that ej ā fj 1 < Ī“, where Ī“ > 0 is suļ¬ciently small. Let L1 be a subspace spanned by f1 , . . . , fp . p It is easy to see that dim L1 = dim L = p for a small Ī“. Indeed, if j=1 cj fj = 0 with cj ā C, then by taking the inner product with fk , we get p
cj (fj , fk ) = 0, k = 1, . . . , p.
j=1
But (fj , fk ) = Ī“jk + Īµjk , where Īµjk ā 0 as Ī“ ā 0. Therefore, the matrix (Ī“jk + Īµjk ), being close to the identity matrix, is invertible for a small Ī“. Therefore, cj = 0 and f1 , . . . , fp are linearly independent; i.e., dim L1 = p. Further, if [u, u] ā¤ Ī»(u, u) for all u ā L, then [v, v] ā¤ (Ī»+Īµ)(v, v), v ā L1 , with Īµ = Īµ(Ī“), such that Īµ(Ī“) ā 0 as Ī“ ā 0. Indeed, if v = pj=1 cj fj , then p u = j=1 cj ej is close to v with respect to Ā· 1 , implying the required statement. Now, set N1 (Ī») = sup {dim L1 : (Au, u) ā¤ Ī»(u, u) for all u ā L1 ā D(A0 )} . Then from the above argument, it follows that N (Ī» ā Īµ) ā¤ N1 (Ī») ā¤ N (Ī») for any Ī» ā R and Īµ > 0. If Ī» is not an eigenvalue, then N (Ī») is continuous at Ī» implying N1 (Ī») = N (Ī»). Remark. Obviously, N (Ī») can be recovered from N1 (Ī»); namely, N (Ī») = N1 (Ī» + 0). Another description of the eigenvalues in terms of the quadratic form of A is sometimes more convenient. Namely, the following statement holds.
9.5. Variational principle
211
Proposition 9.6 (Courantās variational principle). (9.29)
Ī»1 =
(AĻ, Ļ) , ĻāD(A)\0 (Ļ, Ļ) min
(9.30) Ī»j+1 =
max
LāD(A), dim L=j
(AĻ, Ļ) , Ļā(Lā„ \0)ā©D(A) (Ļ, Ļ) min
j = 1, 2, . . . .
If A is the Friedrichs extension of A0 , then, by replacing max with sup and min with inf, we may write Ļ ā D(A0 )\{0} and L ā D(A0 ), instead of Ļ ā D(A)\{0} and L ā D(A), respectively. Proof. We will use the same notation as in the proof of Proposition Let ā 9.5. 2 Ī»2 < c Ļ for Ļ ā D(A) (i.e., if |c | us prove (9.29) ļ¬rst. If Ļ = ā j j=1 j j j=1 j ā 2 and (Ļ, Ļ) = 2 . Therefore, Ī» |c | |c | +ā), then (AĻ, Ļ) = ā j=1 j j j=1 j (AĻ, Ļ) =
ā
Ī»j |cj | ā„ Ī»1 2
j=1
ā
|cj |2 = Ī»1 (Ļ, Ļ),
j=1
and the equality is attained at Ļ = Ļ1 . This proves (9.29). Now let us prove (9.30). Let Ī¦j be the subspace spanned by Ļ1 , . . . , Ļj . Take L = Ī¦j . We see that the right-hand side of (9.30) is not smaller than the left-hand side. Next we should verify that the right-hand side is not greater than the left-hand side, i.e., if L ā D(A) and dim L = j, then there exists a nonzero vector Ļ ā Lā„ ā© D(A) such that (AĻ, Ļ) ā¤ Ī»j+1 (Ļ, Ļ). But, as in the proof of Proposition 9.5, it is easy to verify that Ī¦j+1 ā© Lā„ = 0 (if we had Ī¦j+1 ā© Lā„ = 0, then the orthogonal projection onto L would be an injective map of Ī¦j+1 in L, contradicting the fact that dim Ī¦j+1 = j +1 > j = dim L). Therefore, we may take Ļ ā Ī¦j+1 ā© Lā„ , Ļ = 0, and then it is clear that (AĻ, Ļ) ā¤ Ī»j+1 (Ļ, Ļ). The last statement of Proposition 9.6 is proved as the last part of Proposition 9.5. Let us return to the eigenvalues of āĪ in a bounded domain Ī© ā Rn (with the Dirichlet boundary conditions). Denote them by Ī»j (Ī©), j = 1, 2, . . ., and their distribution function, N (Ī»), by NĪ© (Ī»). In this case D(A0 ) = D(Ī©). Now let Ī©1 and Ī©2 be two bounded domains such that Ī©1 ā Ī©2 . Then D(Ī©1 ) ā D(Ī©2 ), and Proposition 9.5 implies that (9.31)
NĪ©1 (Ī») ā¤ NĪ©2 (Ī»);
therefore, Ī»j (Ī©1 ) ā„ Ī»j (Ī©2 ),
j = 1, 2, . . . .
212
9. Eigenvalues and eigenfunctions
The eigenfunctions and eigenvalues may be found explicitly if Ī© is a rectangular parallelepiped. Suppose Ī© is of the form Ī© = (0, a1 ) Ć Ā· Ā· Ā· Ć (0, an ). It is easy to verify that the eigenfunctions can be taken in the form Ļk1 ,...,kn = Ck1 ,...,kn sin
k1 Ļx1 kn Ļxn Ā· Ā· Ā· sin , a1 an
where k1 , . . . , kn are (strictly) positive integers and Ck1 ,...,kn are normalizing constants. The eigenvalues are of the form k1 Ļ 2 kn Ļ 2 +Ā·Ā·Ā· + . Ī»k1 ,...,kn = a1 an The function NĪ© (Ī») in our case is equal to the number ofāpoints of the form ( ka11Ļ , . . . , kannĻ ) ā Rn belonging to the closed ball of radius Ī» with the center at the origin. If the ki are allowed to be any integers, we see that points ( ka11Ļ , . . . , kannĻ ) run over a lattice in Rn obtained by obvious homotheties of an integer lattice. It is easy to see that NĪ© (Ī») for ā large Ī» has a two-sided estimate in terms of the volume of a ball of radius Ī»; i.e., C ā1 Ī»n/2 ā¤ NĪ© (Ī») ā¤ CĪ»n/2 ,
(9.32) where C > 0.
Now note that we may inscribe a small cube in Ī© and, the other way around, embed Ī© into a suļ¬ciently large cube. With (9.31) we see that (9.32) holds for any bounded domain Ī©. Making these arguments more precise, it is possible to prove the following asymptotic formula by H. Weyl: NĪ© (Ī») ā¼ (2Ļ)ān Ļn Ā· mes(Ī©) Ā· Ī»n/2 ,
as Ī» ā +ā,
where Ļn is the volume of the unit ball in Rn and mes(Ī©) is the Lebesgue measure of Ī©; see, e.g., Courant and Hilbert, Volume I of [5].
9.6. Problems 9.1. Let Ī© be a bounded domain in Rn with a smooth boundary. Let L be a selfadjoint operator in L2 (Ī©) determined by the Laplace operator in Ī© with the Dirichlet boundary conditions. The Greenās function of the Laplace operator in Ī© is the Schwartz kernel of Lā1 , i.e., a distribution G(x, y), x, y ā Ī©, such that
G(x, y)f (y)dy. Lā1 f (x) = Ī©
9.6. Problems
213
Prove that Īx G(x, y) = Ī“(x ā y), G|xāāĪ© = 0, and G is uniquely deļ¬ned by these conditions. Give a physical interpretation of the Greenās function. 9.2. Find out what kind of singularity the Greenās function G(x, y) has at x = y ā Ī©. 9.3. Prove that the Greenās function is symmetric; i.e., G(x, y) = G(y, x),
x, y ā Ī©.
9.4. Prove that the solution of the problem Īu = f , u|āĪ© = Ļ in the domain Ī© is expressed via the Greenās function by the formula
āG(x, y) G(x, y)f (y)dy ā Ļ(y) dSy , u(x) = āny Ī© āĪ© where ny is the exterior normal to the boundary at y and dSy is the area element of the boundary at y. 9.5. The Greenās function of the Laplace operator in an unbounded domain Ī© ā Rn is a function G(x, y), x, y ā Ī©, such that Īx G(x, y) = Ī“(x ā y), G|xāāĪ© = 0, and G(x, y) ā 0 as |x| ā +ā (for every ļ¬xed y ā Ī©) if n ā„ 3; G(x, y) is bounded for |x| ā +ā for every ļ¬xed y ā Ī© if n = 2. Find the Greenās function for the half-space xn ā„ 0. 9.6. Using the result of the above problem, write a formula for the solution of Dirichletās problem in the half-space xn ā„ 0 for n ā„ 3. Solve this Dirichlet problem also with the help of the Fourier transform with respect to x = (x1 , . . . , xnā1 ) and compare the results. 9.7. Find the Greenās function of a disc in R2 and a ball in Rn , n ā„ 3. 9.8. Write a formula for the solution of the Dirichlet problem in a disc and a ball. Obtain in this way the Poisson formula for n = 2 from Problem 8.2. 9.9. Find the Greenās function of a half-ball. 9.10. Find the Greenās function for a quarter of R3 , i.e., for {(x1 , x2 , x3 ) : x2 > 0, x3 > 0} ā R3 .
214
9. Eigenvalues and eigenfunctions
9.11. The Bessel function JĪ½ (x) is deļ¬ned for x > 0 and any Ī½ ā C, by the series ā
x 2k+Ī½ (ā1)k JĪ½ (x) = . k!Ī(k + Ī½ + 1) 2 k=0
Prove that if n ā Z+ , then Jān (x) = (ā1)n Jn (x). 9.12. Prove that if n ā Z+ , then the Bessel equation 1 n2 y + y + 1ā 2 y =0 x x has for a solution not only Jn (x) but also a function deļ¬ned by the series ā ck xk , c0 = 0. (ln x) Ā· xān k=0 x
1
9.13. Prove that the Laurent expansion of e 2 (tā t ) with respect to t (for t = 0, ā) has the form ā x (tā 1t ) 2 = Jn (x)tn . e āā
9.14. Prove that the expansion of
eix sin Ļ
into the Fourier series of Ļ is of
the form e
ix sin Ļ
= J0 (x) + 2
ā
[J2n (x) cos 2nĻ + iJ2nā1 (x) sin(2n ā 1)Ļ].
n=1
9.15. Find eigenvalues and eigenfunctions of the Laplace operator in a rectangle. Prove that these eigenfunctions form a complete orthogonal system. 9.16. Let k ā Z+ and Ī±k,1 < Ī±k,2 < Ā· Ā· Ā· be all the zeros of Jk (x) for x > 0. Prove that
R Ī±k,n Ī±k,m r Jk r rdr = 0, for m =
n. Jk R R 0 9.17. Using the result of Problem 9.16, ļ¬nd a complete orthonormal system of eigenfunctions of the Laplace operator in a disc. 9.18. Describe a way to use the Fourier method to solve the Dirichlet problem in a straight cylinder (of ļ¬nite height) with a disc as the base.
10.1090/gsm/205/10
Chapter 10
The wave equation
10.1. Physical problems leading to the wave equation There are many physical problems leading to the wave equation (10.1)
utt = a2 Īu,
where u = u(t, x), t ā R, x ā Rn , Ī is the Laplacian with respect to x. The following more general nonhomogeneous equation is also often encountered: (10.2)
utt = a2 Īu + f (t, x).
We have already seen that small oscillations of the string satisfy (10.1) (for n = 1), and in the presence of external forces they satisfy (10.2). It can be shown that small oscillations of a membrane satisfy similar equations, if we denote by u = u(t, x), for x ā R2 , the vertical displacement of the membrane from the (horizontal) equilibrium position. Similarly, under small oscillations of a gas (acoustic wave) its parameters (e.g., pressure, density, displacement of gas particles) satisfy (10.1) with n = 3. One of the most important ļ¬elds, where equations of the form (10.1) and (10.2) play a leading role, is classical electrodynamics. Before explaining this in detail, let us recall some notation from vector analysis. (See, e.g., Ch. 10 in [27].) Let F = F(x) = (F1 (x), F2 (x), F3 (x)) be a vector ļ¬eld deļ¬ned in an open set Ī© ā R3 . Here x = (x1 , x2 , x3 ) ā Ī©, and we will assume that F is suļ¬ciently smooth, so that all derivatives of the components Fj , which we need, are continuous. It is usually enough to assume that F ā C 2 (Ī©) (which means that every component Fj is in C 2 (Ī©)). If f = f (x) is a suļ¬ciently 215
216
10. The wave equation
smooth function on Ī©, then its gradient āf āf āf grad f = āf = , , āx1 āx2 āx3 is an example of a vector ļ¬eld in Ī©. Here it is convenient to consider ā as a vector ā ā ā , , ā= āx1 āx2 āx3 whose components are operators acting, say, in C ā (Ī©), or, as operators from C k+1 (Ī©) to C k (Ī©), k = 0, 1, . . . . The divergence of a vector ļ¬eld F is a scalar function āF1 āF2 āF3 + + . div F = āx1 āx2 āx3 For a vector ļ¬eld F in Ī© its curl is another vector ļ¬eld āF3 āF2 āF1 āF3 āF2 āF1 . ā , ā , ā curl F = ā Ć F = āx2 āx3 āx3 āx1 āx1 āx2 Here the cross āĆā in āĆF stands for the cross product of two 3-component vectors with the product of ā/āxj and Fk interpreted as the derivative āFk /āxj . The following identities, which hold for every smooth scalar function f and vector ļ¬eld F, are easy to verify: (10.3)
div grad f = ā Ā· āf = Īf,
(10.4)
curl grad f = ā Ć āf = 0,
(10.5)
div curl F = 0,
(10.6)
curl(curl F) = ā Ć (ā Ć F) = grad div F ā ĪF,
where in the last term the Laplacian Ī is applied to every component of F. Vector analysis is easier to understand if we use the language of diļ¬erential forms. To this end, with each vector ļ¬eld F we associate two diļ¬erential forms Ī»F = F1 dx1 + F2 dx2 + F3 dx3 and ĻF = F1 dx2 ā§ dx3 + F2 dx3 ā§ dx1 + F3 dx1 ā§ dx2 . It is easy to see that dĪ»F = Ļcurl F , which can serve as a deļ¬nition of curl F. Also, dĻF = div F dx1 ā§ dx2 ā§ dx3 . In particular, the identities (10.4) and (10.5) immediately follow from the well-known relation d2 = 0 for the de Rham diļ¬erential d on diļ¬erential forms.
10.1. Physical problems
217
In what follows we will deal with functions and vector ļ¬elds which additionally depend upon the time variable t. In this case all operations grad, div, curl should be applied with respect to the space variables x (for every ļ¬xed t). The equations of classical electrodynamics (Maxwellās equations) are
(M.3)
Ļ , Īµ0 āB , curl E = ā āt div B = 0,
(M.4)
c2 curl B =
(M.1) (M.2)
div E =
j āE . + Īµ0 āt
Here E, B are vectors of electric and magnetic ļ¬elds, respectively (threedimensional vectors depending on t and x ā R3 ), Ļ a scalar function of t and x (density of the electric charge), j a three-dimensional vector of density of the electric current, also depending on t and x (if charges with density Ļ at a given point and time move with velocity v, then j = Ļv). The numbers c, Īµ0 are universal constants depending on the choice of units: the speed of light and the dielectric permeability of vacuum. For applications of the Maxwell equations it is necessary to complement them with the formula for the Lorentz force acting on a moving charge. This force is of the form (M.5)
F = q(E + v Ć B),
where q is the charge, v its velocity. Formula (M.5) may be used for experimental measurement of E and B (or, say, to deļ¬ne them as physical quantities). The discussion of experimental facts on which equations (M.1)ā(M.5) are based may be found in textbooks on physics (see, e.g., Feynmann, Leighton, and Sands [8, vol. 2, Ch. 18]). Let us rewrite the Maxwell equations, introducing scalar and vector potentials Ļ and A. For simplicity assume that the ļ¬elds E and B are deļ¬ned and suļ¬ciently smooth in the whole space R4t,x . By the PoincarĀ“e lemma, it follows from (M.3) that B = curl A, where A is a vector function of t and x determined up to a ļ¬eld A0 such that curl A0 = 0; hence A0 = grad Ļ, where Ļ is a scalar function. Substituting B = curl A into (M.2), e lemma again, we see we get curl(E + āA āt ) = 0. Hence, using the PoincarĀ“ āA that E + āt = ā grad Ļ, where Ļ is a scalar function. Therefore, instead of
218
10. The wave equation
(M.2) and (M.3), we may write (M.6)
B = curl A,
āA . āt The vector function A is called the vector potential, and the scalar function Ļ is called the scalar potential. Note that equations (M.6) and (M.7) do not uniquely determine potentials A and Ļ. Without aļ¬ecting E and B, we may replace A and Ļ by E = ā grad Ļ ā
(M.7)
āĻ āt (this transformation of potentials is called a gauge transformation). Transformations (M.8) may be used to get simpler equations for potentials. For instance, using (M.8) we may get any value of A = A + grad Ļ, Ļ = Ļ ā
(M.8)
div A = div A + div grad Ļ = div A + ĪĻ, since solving Poissonās equation ĪĻ = Ī±(t, x) we may get any value of ĪĻ, hence, of div A. Let us derive equations for A and Ļ using two remaining Maxwell equations. Substituting the expression for E in terms of the potentials into (M.1), we get āĪĻ ā
(M.9)
Ļ ā div A = . āt Īµ0
Substituting E and B in (M.4), we get āA j ā 2 c curl curl A ā ā grad Ļ ā = . āt āt Īµ0 Using (10.6), we obtain ā 2A ā j grad Ļ + = . 2 āt āt Īµ0 Now, choose A with the help of the gauge transformation such that (M.10)
āc2 ĪA + c2 grad(div A) +
1 āĻ . c2 āt More precisely, ļ¬rst let some potentials Ļ and A be given. We want to ļ¬nd the function Ļ in the gauge transformation (M.8) so that Ļ and A satisfy (M.11). This yields for Ļ an equation of the form
(M.11)
div A = ā
1 ā 2Ļ ā ĪĻ = Īø(t, x), c2 āt2 where Īø(t, x) is a known function. This is a nonhomogeneous wave equation. Suppose that we have solved this equation and, ļ¬nding Ļ, obtained Ļ and (M.12)
10.1. Physical problems
219
A satisfying (M.11). Then the second and the third terms in (M.10) cancel and we get (M.13)
ā2A j ā c2 ĪA = , āt2 Īµ0
and (M.9) takes the form (M.14)
c2 Ļ ā 2Ļ 2 ā c ĪĻ = . āt2 Īµ0
Therefore, we may assume that Ļ and A satisfy nonhomogeneous wave equations (M.13), (M.14), which, together with (M.11), are equivalent to the Maxwell equations. The gauge condition (M.11) is called the Lorentz gauge condition. Denote by the wave operator (or dāAlembertian) (M.15)
=
ā2 ā c2 Ī āt2
and let ā1 be the convolution with a fundamental solution of this operator. Then ā1 commutes with diļ¬erentiations. Suppose ā1 j and ā1 Ļ make 2 sense. Then the potentials A = ā1 Īµj0 and Ļ = ā1 cĪµ0Ļ satisfy (M.13) and (M.14). Will the Lorentz gauge condition be satisļ¬ed for this particular choice of the potentials? Substituting the found A and Ļ into (M.11) we see that this condition is satisļ¬ed if and only if (M.16)
āĻ + div j = 0. āt
But this condition means the conservation of charge. Therefore, if the conservation of charge is observed, then a solution of the Maxwell equations may be found if we can solve a nonhomogeneous wave equation. Reducing Maxwellās equations to the form (M.13), (M.14) shows that Maxwellās equations are invariant with respect to Lorentz transformations (linear transformations of the space-time R4t,x preserving the Minkowski metric c2 dt2 ā dx2 ā dy 2 ā dz 2 ). Moreover, they are invariant under the transformations from the PoincarĀ“e group, which is generated by the Lorentz transformations and translations. In the empty space (in the absence of currents and charges) the potentials Ļ and A satisfy the wave equation of the form (10.1) for n = 3 and a = c. This implies that all components of E and B satisfy the same equation.
220
10. The wave equation
10.2. Plane, spherical, and cylindric waves A plane wave is a solution of (10.1) which, for a ļ¬xed t, is constant on each hyperplane which is parallel to a given one. By a rotation in x-space we may make this family of planes to be of the form x1 = const. Then a plane wave is a solution of (10.1) depending only on t and x1 . But then it is a solution of the one-dimensional wave equation utt = a2 ux1 x1 ; i.e., it is of the form u(t, x) = f (x1 ā at) + g(x1 + at), where f, g are arbitrary functions. Before the rotation, each of the summands had clearly been of the form f (r Ā· x ā at), where r ā Rn , |r| = 1. Another form of expression is f (k Ā· x ā Ļt), where k ā Rn , but may have an arbitrary length. Such a form is convenient to make the argument of f dimensionless and is most often used in physics and mechanics. If the dimension of t is [second] and that of x is [meter], then that of Ļ is [secā1 ] and that of k is [mā1 ]. Substituting f (k Ā· x ā Ļt) into (10.1), we see that Ļ 2 = a2 |k|2 or a = Ļ |k| . Clearly, the speed of the plane wave front (the plane where f takes a prescribed value) is equal to a and the equation of such a front is k Ā· x ā Ļt = const, where k and Ļ satisfy Ļ 2 ā a2 |k|2 = 0. An important example of a plane wave is ei(kĀ·xāĻt) . (Up to now we only considered real solutions in this section, and we will do this in the future, but we write this exponent assuming that either the real or imaginary part is considered. They will also be solutions of the wave equation.) In this case each ļ¬xed point x oscillates sine-like with frequency Ļ and for a ļ¬xed t the wave depends sine-like on k Ā· x, so that it varies sine-like in the direction of k with the speed |k|. The vector k is often called the wave vector. The relation Ļ 2 = a2 |k|2 is the dispersion law for the considered waves (in physics more general dispersion laws of the form Ļ = Ļ(k) are encountered). Many solutions of the wave equation may be actually obtained as a superposition of plane waves with diļ¬erent k. We will, however, obtain the most important types of such waves directly. In what follows, we will assume that n = 3 and will study spherical waves, solutions of (10.1) depending only on t and r, where r = |x|. Thus, let u = u(t, r), where r = |x|. Then (10.1) can be rewritten in the form (10.7)
2 1 utt = urr + ur . 2 a r
Let us multiply (10.7) by r and use the identity rurr + 2ur =
ā2 (ru). ār2
10.2. Plane, spherical, and cylindric waves
221
Then, clearly, (10.7) becomes 1 (ru)tt = (ru)rr , a2 implying ru(t, r) = f (r ā at) + g(r + at) and (10.8)
u(t, r) =
f (r ā at) g(r + at) + . r r
This is the general form of spherical waves. The wave rā1 f (r ā at) diverges from the origin 0 ā R3 , as from its source, and the wave rā1 g(r + at), conversely, converges to the origin (coming from inļ¬nity). In electrodynamics one usually considers the wave coming out of a source only, discarding the second summand in (10.8) by physical considerations. In this case the function f (r) characterizes properties of the source. The front of the emanating spherical wave is the sphere r ā at = const. It is clear that the speed of the front is a, as before. Let us pass to cylindric wavesāsolutions of (10.1) depending only on t and the distance to the x3 -axis. A cylindric wave only depends, actually, on t, x1 , x2 (and even only on t and Ļ = x21 + x22 ) so that it is a solution of the wave equation (10.1) with n = 2. It is convenient, however, to consider it as a solution of the wave equation with n = 3, but independent of x3 . One of the methods to construct cylindric waves is as follows: take a superposition of identical spherical waves with sources at all points of the x3 -axis (or converging to all points of the x3 -axis). Let e3 be the unit vector of the x3 -axis. Then we get a cylindric wave, by setting
ā
ā f (|x ā ze3 | ā at) g(|x ā ze3 | + at) dz + dz. (10.9) v(t, x) = |x ā ze3 | |x ā ze3 | āā āā Set r = |x ā ze3 | =
Ļ2 + (x3 ā z)2 , Ļ =
x21 + x22 .
Clearly, v(t, x) does not depend on x3 and only depends on t and Ļ. Therefore, we may assume that x3 = 0 in (10.9). Then the integrands are even functions of z, and it suļ¬ces to calculate the integrals over (0, +ā). For the variable of integration take r instead of z, so that dr =
z dz, r
dz =
r r dr = dr. z r2 ā Ļ2
Since r runs from Ļ to ā (for z ā [0, ā)), we get
ā
ā f (r ā at) g(r + at) dr + 2 dr v(t, Ļ) = 2 2 2 r āĻ r2 ā Ļ2 Ļ Ļ
222
10. The wave equation
or
(10.10)
ā
v(t, Ļ) = 2 Ļāat
f (Ī¾)dĪ¾ (Ī¾ +
at)2
ā Ļ2
ā
+2 Ļ+at
g(Ī¾)dĪ¾ (Ī¾ ā at)2 ā Ļ2
.
All this makes sense when the integrals converge. For instance, if f, g are continuous functions of Ī¾, then for the convergence it suļ¬ces that
ā
ā |f (Ī¾)|dĪ¾ |g(Ī¾)|dĪ¾ < +ā, < +ā for M > 0. |Ī¾| |Ī¾| M M It can be shown that (10.10) is the general form of the cylindric waves. Let us brieļ¬y describe another method of constructing cylindric waves. Seeking a cylindric wave of the form v(t, Ļ) = eiĻt f (Ļ), we easily see that f (Ļ) should satisfy the equation 1 Ļ f + f + k 2 f = 0, k = , Ļ a which reduces to the Bessel equation with Ī½ = 0 (see Chapter 9). Then we can take a superposition of such waves (they are called monochromatic waves) by integrating over Ļ.
10.3. The wave equation as a Hamiltonian system In Section 2.1 we have already discussed for n = 1 the possibility of expressing the wave equation as a Lagrange equation of a system with an inļ¬nite number of degrees of freedom. Now, let us brieļ¬y perform the same for n = 3, introducing the corresponding quanitites by analogy with the one-dimensional case. Besides, we will discuss a possibility to pass to the Hamiltonian formalism. For simplicity, we will only consider real-valued functions u(t, x), which are smooth functions of t with values in S(R3x ); i.e., we will analyze the equation (10.1) in the class of rapidly decaying functions with respect to x. Here temporarily we will denote by S(R3x ) the class of real-valued functions satisfying the same restrictions as in the deļ¬nition of S(R3 ) which was previously formulated. We prefer to do this rather than complicate notation, and we hope that this will not lead to confusion. All other functions in this section will be also assumed to be real valued. Let us introduce terminology similar to one used in classical mechanics. The space M = S(R3x ) will be referred to as the conļ¬guration space. An element u = u(x) ā S(R3x ) may be considered as a āset of coordinatesā of the three-dimensional system, where x is the ālabelā of the coordinate and u(x) is the coordinate itself with the index x. The tangent bundle on M is the direct product T M = S(R3x ) Ć S(R3x ). As usual, the set of pairs {(u0 , v) ā T M } for a ļ¬xed u0 ā M is denoted by T Mu0 and is called the tangent space at u0 (this space is canonically isomorphic to M as in the case
10.3. The wave equation as a Hamiltonian system
223
when M is a ļ¬nite-dimensional vector space). Elements of T M are called tangent vectors. A path in M is a function u(t, x) , t ā (a, b), inļ¬nitely diļ¬erentiable with respect to t with values in S(R3x ). The velocity of u(t, x) at t0 is the tangent Ė 0 , x)}, where uĖ = āu vector {u(t0 , x), u(t āt . The kinetic energy is the function K on the tangent bundle introduced by the formula
1 K({u, v}) = v 2 (x)dx. 2 R3 Given a path u(t, x) in M then, taking for each t a tangent vector {u, u} Ė and the value of the kinetic energy at it, we get the function of time
1 [u(t, Ė x)]2 dx, K(u) = 2 R3 which we will call the kinetic energy along the path u. The potential energy is the following function on M :
a2 |ux (x)|2 dx, U (u) = 2 R3
āu āu āu . , , where ux = grad u(x) = āx 1 āx2 āx3 If there is a path in M , then the potential energy along this path is a function of time. With the help of the canonical projection T M āā M , the function U is lifted to a function on T M , again denoted by U . Therefore, U ({u, v)} = U (u). The Lagrangian or the Lagrange function is the function L = K ā U on T M . The action along the path u(t, x), t ā [t0 , t1 ], is the integral along this path:
t1 Ldt, S = S[u] = t0
where 1 L = L({u(t, x), u(t, Ė x)}) = 2
a2 [u(t, Ė x)] dx ā 2 R3
2
R3
|ux (x)|2 dx.
The wave equation (10.1) may be expressed in the form Ī“S = 0, where Ī“S is the variation (or the diļ¬erential) of the functional S taken along paths u(t, x) with ļ¬xed source u(t0 , x) and the target u(t1 , x). Indeed, if Ī“u(t, x) is an admissible variation of the path, i.e., a smooth function of t ā [t0 , t1 ] with values in S(R3x ), such that Ī“u(t0 , x) = Ī“u(t1 , x) = 0, then for such a
224
10. The wave equation
variation of the path, Ī“S takes the form
t1
t1
d Ī“S = S[u + Īµ(Ī“u)]|Īµ=0 = u(Ī“ Ė u)dxdt Ė ā a2 ux Ā· Ī“ux dxdt dĪµ t0 R3 t0 R3
2 ā u 2 Ī“udxdt + a (Īu)Ī“udxdt = (āu)Ī“udxdt, = ā āt2 where =
ā2 āt2
ā a2 Īx is the dāAlembertian.
It is clear now that Ī“S = 0 is equivalent to u = 0, i.e., to the wave equation (10.1). To pass to the Hamiltonian formalism, we should now introduce the cotangent bundle T ā M . But in our case each T Mu0 is endowed with the inner product determined by the quadratic form 2K which makes it natural to set T ā Mu0 = T Mu0 and T ā M = T M . The Hamiltonian or energy is the following function on T M :
1 a2 2 (10.11) H = H({u, v}) = K + U = v (x)dx + |ux (x)|2 dx. 2 R3 2 R3 Given a path u(t, x), we see that H({u, u}) Ė is a function of time along this path. Proposition 10.1 (The energy conservation law). Energy is constant along any path u(t, x) satisfying u = 0. Proof. Along the path u(t, x), we have
dH 2 2 = uĀØ Ė udx + a ux Ā· uĖ x dx = uĀØ Ė udx ā a Īu Ā· udx Ė dt R3 R3 R3 R3
= ( u) Ā· udx Ė = 0, R3
as required. Corollary 10.2. For any Ļ, Ļ ā S(R3x ) there exists at most one path u(t, x) satisfying u = 0 and initial conditions u(t0 , x) = Ļ(x),
u(t Ė 0 , x) = Ļ(x).
Proof. It suļ¬ces to verify that if Ļ ā” Ļ ā” 0, then u ā” 0. But this follows immediately from H = const = 0 along this path. The Hamiltonian expression of equations uses also the symplectic form on T ā M . To explain and motivate further calculations, let us start with a toy model case of a particle which moves along the line R in a potential ļ¬eld given by a potential energy U = U (x) (see the appendix in Chapter 2). Here M = R, T M = R Ć R, T ā M = R Ć R, the equation of motion is
10.3. The wave equation as a Hamiltonian system
225
mĀØ x = āU (x), and the energy is H(x, x) Ė = 12 mxĖ 2 + U (x). Introducing the momentum p = mxĖ and using q instead of x (as physicists do) we can rewrite the Hamiltonian in the form p2 (10.12) H(q, p) = + U (q) 2m and then the equation of motion is equivalent to a Hamiltonian system qĖ = āH āp , (10.13) pĖ = ā āH āq . Let us agree that when we talk about a point in coordinates q, p, then it is understood that we mean a point in T ā M , and we will simply write {q, p} ā T ā M . Another important object (in this toy model) is the symplectic 2-form Ļ = dp ā§ dq on T ā M . It can also be considered as a skew-symmetric form on each tangent space Tq,p (T ā M ), which in this case can be identiļ¬ed with the set of 4-tuples {q, p, Xq , Xp }, where all entries are reals and {Xq , Xp } are the natural coordinates along the ļ¬ber of T (T ā M ) over the point {q, p}: Ļ({q, p, Xq , Xp }, {q, p, Yq , Yp }) = Xp Yq ā Xq Yp . Now with any vector ļ¬eld X on T ā M we can associate a 1-form Ī±X on T ā M given by the relation (10.14)
Ī±X (Y ) = Ļ(Y, X).
Note that the form Ļ is nondegenerate in the following sense: if X is a tangent vector to T ā M at z ā T ā M , such that Ļ(X, Y ) = 0 for all Y ā Tz (T ā M ), then X = 0. Therefore, the vector ļ¬eld X is uniquely deļ¬ned by its 1-form Ī±X . So we will denote by XĪ± the vector ļ¬eld corresponding to any 1-form Ī± on T ā M ; i.e., the relations Ī± = Ī±X and X = XĪ± are equivalent. Let us try to ļ¬nd X = XdH where H is a smooth function on T ā M (a Hamiltonian). By deļ¬nition we should have for any vector ļ¬eld Y on T ā M āH āH Yq + Yp = Xq Yp ā Xp Yq ; āq āp hence
āH āH , Xp = ā , āp āq so XdH is the vector ļ¬eld corresponding to the Hamiltonian system (10.13). Note that this argument works for any smooth Hamiltonian H, not only for the speciļ¬c one given by (10.12). Xq =
The arguments and calculations above can be easily extended to the case of motion in M = Rn . In this case we should allow x, p, q to be vectors in Rn , x = (x1 , . . . , xn ), p = (p1 , . . . , pn ), q = (q1 , . . . , qn ), the potential energy
226
10. The wave equation
U = U (x) to be a smooth scalar (real-valued) function on Rn ; then take the Hamiltonian n p2j + U (q), (10.15) H(q, p) = 2mj j=1
where q = x. Here H(q, p) should be considered as a function on T ā M = R2n . Now the equations of motion āU mj x ĀØj = ā , j = 1, . . . , n, āxj again can be rewritten as a Hamiltonian system (10.13), where āH/āp and āH/āq should be understood as gradients with respect to variables p and q, respectively; for example, āH āH āH = ,..., . āp āp1 āpn Instead of the form dp ā§ dq we should take the 2-form n dpj ā§ dqj . (10.16) Ļ= j=1
which is usually called the canonical symplectic form on T ā M . It is nondegenerate in the sense explained above. Therefore, (10.14) again deļ¬nes a one-to-one correspondence between vector ļ¬elds and 1-forms on T ā M . Then again it is easy to check that the 1-form dH corresponds to the vector ļ¬eld which generates the Hamiltonian system (10.13). The reader may wonder why we need fancy notation like M , T ā M for the usual vector spaces Rn , R2n . The point is that in fact M can be an arbitrary smooth manifold. In fact, choosing local coordinates x = (x1 , . . . , xn ) in an open set G ā M , we can deļ¬ne natural coordinates q, p in the cotangent bundle T ā M , so that q, p represent the cotangent vector pdq = p1 dq1 + Ā· Ā· Ā· + pn dqn at q = x. Then the 1-form pdq is well-deļ¬ned on M ; i.e., it does not depend on the choice of local coordinates. (We leave the proof to the reader as an easy exercise.) Therefore, the 2-form Ļ = d(pdq) is also well-deļ¬ned on M . In an even more general context, a 2-form Ļ on a manifold Z is called symplectic if it is closed and nondegenerate. Then the same machinery works. So, starting with a Hamiltonian H (a smooth function on Z), we can take the 1-form dH and the corresponding vector ļ¬eld XH . The system of ODE (10.17)
zĖ = XH (z)
is called a Hamiltonian system with the Hamiltonian H. Locally, it is nothing new: by the Darboux theorem we can choose local coordinates so that Ļ is
10.3. The wave equation as a Hamiltonian system
227
given by (10.16) in these coordinates. Therefore, the Hamiltonian system (10.17) has the canonical form (10.13). More details on symplectic structure, Hamiltonian systems, and related topics can be found in [3]. Now let us make a leap to inļ¬nite-dimensional spaces; namely, let us return to M = S(R3x ). The symplectic form should be deļ¬ned on each tangent space to T ā M at a ļ¬xed point. In our case this tangent space is naturally identiļ¬ed with T ā M = T M itself, so we have T (T ā M ) = T ā M Ć T ā M = T M Ć T M . An element of T (T ā M ) should be expressed as a 4-tuple {u, v, Ī“u, Ī“v} consisting of functions belonging to S(Rn ). The symplectic form itself depends only on Ī“u, Ī“v. For brevity let us set Ī“u = Ī± , Ī“v = Ī². Then deļ¬ne the symplectic form to be (10.18)
[{u, v, Ī±, Ī²}, {u, v, Ī±1 , Ī²1 }] = [{Ī±, Ī²}, {Ī±1 , Ī²1 }] =
(Ī±Ī²1 ā Ī²Ī±1 )dx. R3 n j=1 dpj ā§ dqj , where
(This is an analogue of the canonical 2-form Ļ = summation is replaced by integration, though we use brackets instead of Ļ to simplify notation.)
Choosing an arbitrary {u, v} ā T ā M , let us take a variation of H = H({u, v}) (which replaces the diļ¬erential in the inļ¬nite-dimensional case) in the direction of a vector {u, v, Ī“u, Ī“v} ā T (T ā M ). In our case u, v, Ī“u, Ī“v are arbitrary functions from S(R3x ). We obtain, diļ¬erentiating the expression of H in (10.11) and integrating by parts,
2 Ī“H Ā· Ī“ux + v Ī“v dx a Ī“H(Ī“u, Ī“v) = Ī“ux R3
(āa2 Īu Ī“u + v Ī“v)dx = (Ī± Ī“u + Ī² Ī“v)dx, = R3
R3
where Ī± = āa2 Īu and Ī² = v. Now, if we take another vector Y = {u, v, Ī±1 , Ī²1 } ā T (T ā M ) (with the same u, v) and denote by XH the Hamiltonian vector ļ¬eld with the Hamiltonian H, then we should have [Y, XH ] = Ī“H(Y ), which gives
2 (Ī±1 Ī² ā Ī²1 Ī±)dx = (Ī±1 v + Ī²1 a Īu)dx = (Ī±1 XĪ± + Ī²1 XĪ² )dx, R3
R3
R3
where {u, v, XĪ± , XĪ² } is the vector of the Hamiltonian vector ļ¬eld on T ā M . Since Ī±1 and Ī²1 are arbitrary, we should have XĪ± = v, XĪ² = a2 Īu, and the corresponding diļ¬erential equation on T ā M has the form uĖ = v, vĖ = a2 Īu,
228
10. The wave equation
which is equivalent to the wave equation (10.1) for u. So we established that the wave equation can be rewritten as a Hamiltonian system with the Hamiltonian (10.11) and the symplectic form given by (10.18). Later we will use the fact that the ļ¬ow of our Hamiltonian system preserves the symplectic form (or, as one says, the transformations in this ļ¬ow are canonical). This is true for general Hamiltonian systems (see, e.g., [3]) but we will not prove this. Instead we will directly establish the corresponding analytical fact for the wave equation. Proposition 10.3. Let u(t, x) and v(t, x) be two paths. Consider the following function of t:
(uvĖ ā uv)dx. Ė [u, v] = R3
If u = v = 0, then [u, v] = const. Proof. We have
d d [u, v] = (uvĖ ā uv)dx Ė = (uĀØ vāu ĀØv)dx dt R3 dt R3
2 (u Ā· Īv ā Īu Ā· v)dx = 0, =a R3
as required.
10.4. A spherical wave caused by an instant ļ¬ash and a solution of the Cauchy problem for the three-dimensional wave equation Let us return to spherical waves and consider the diverging wave (10.19)
f (r ā at) , r
r = |x|.
The function f (āat) characterizes the intensity of the source (located at x = 0) at the moment t. It is interesting to take the wave emitted by an instant ļ¬ash at the source. This corresponds to f (Ī¾) = Ī“(Ī¾). We get the wave Ī“(r ā at) Ī“(r ā at) = , r at whose meaning is to be clariļ¬ed.
(10.20)
r = |x|,
It is possible to make sense of (10.20) in many equivalent ways. We will consider it as a distribution with respect to x depending on t as a parameter. First, consider the wave (10.19) and set f = fk , where fk (Ī¾) ā C0ā (R1 ), fk (Ī¾) ā„ 0, fk (Ī¾) = 0 for |Ī¾| ā„ 1/k and fk (Ī¾)dĪ¾ = 1, so that fk (Ī¾) ā Ī“(Ī¾)
10.4. Spherical waves and the Cauchy problem
229
as k ā +ā. Now, set Ī“(r ā at) = lim fk (r ā at),
(10.21)
kāā
D (R3 )
if this limit exists in (or, equivalently, in E (R3 ), since supports of all functions fk (r ā at) belong to a ļ¬xed compact provided t belongs to a ļ¬xed ļ¬nite interval of R). Clearly, this limit exists for t < 0 and is 0; i.e., Ī“(r ā at) = 0 for t < 0. We will see that the limit exists also for t > 0. Then this limit is the distribution Ī“(r ā at) ā E (R3 ) and, obviously, supp Ī“(r ā at) = {x : |x| = at}. Since 1/r = 1/|x| is a C ā -function in a neighborhood of supp Ī“(r ā at) for t = 0, then the distribution Ī“(rāat) is well-deļ¬ned and r fk (r ā at) Ī“(r ā at) = lim , t = 0. kāā r r Let us prove the existence of the limit (10.21) for t > 0 and compute it. Let Ļ ā D(R3 ). Let us express the integral fk (r ā at), Ļ in polar coordinates
fk (|x| ā at)Ļ(x)dx fk (r ā at), Ļ = R3 >
=
(10.22)
ā
= |x|=r
0
ā
=
fk (r ā at)Ļ(x)dSr =
fk (r ā at)
0
dr >
|x|=r
Ļ(x)dSr
dr,
where dSr is the area element of the sphere of radius r. Clearly, |x|=r Ļ(x)dSr is an inļ¬nitely diļ¬erentiable function of r for r > 0. Since lim fk (Ī¾) = Ī“(Ī¾), kāā
we obviously get
lim fk (r ā at), Ļ =
kāā
|x|=at
Ļ(x)dSat
and, therefore, the limit of (10.21) exists and
Ī“(r ā at), Ļ = Ļ(x)dSat . |x|=at
Passing to the integration over the unit sphere we get
2 2 Ļ((at)x )dS1 , Ī“(r ā at), Ļ = a t |x |=1
implying (10.23)
Ī“(r ā at) ,Ļ r
= at
|x |=1
Ļ((at)x )dS1 .
230
10. The wave equation
It is worthwhile to study the dependence on the parameter t. It is clear from (10.23) that (10.24)
lim
tā0
Ī“(r ā at) =0 r
|t=0 = 0. Finally, it is obvious from (10.23) and, by continuity, we set Ī“(rāat) r Ī“(rāat) that is inļ¬nitely diļ¬erentiable with respect to t for t > 0 (in the weak r topology). Derivatives with respect to t are easy to compute. For instance, d Ī“(r ā at) ā Ī“(r ā at) ,Ļ = ,Ļ (10.25) āt r dt r
3
āĻ 2 = a Ļ((at)x )dS1 + a t xj ((at)x )dS1 . āx j |x |=1 |x |=1 j=1
This, in particular, implies that lim
tā+0
ā Ī“(r ā at) = 4ĻaĪ“(x) āt r
(the ļ¬rst of the integrals in (10.25) tends to 4ĻaĻ(0) and the second one to 0). Finally, by (10.22) it is clear that Therefore, the distribution with the initial conditions (10.26)
Ī“(r ā at) = 0, t > 0. r
Ī“(rāat) r
is a solution of the wave equation for t > 0
ā Ī“(r ā at) Ī“(r ā at) = 0, = 4ĻaĪ“(x). r āt r t=0 t=0
is similarly constructed. It is equal to 0 for The converging wave Ī“(r+at) r t > 0, and for t < 0 it satisļ¬es the wave equation with the initial conditions Ī“(r + at) ā Ī“(r + at) (10.27) = 0, = ā4ĻaĪ“(x) r āt r t=0 t=0 is obtained from Ī“(rāat) by re(this is clear, for instance, because Ī“(r+at) r r placing t with āt). Besides, we can perform translations with respect to the spatial and time variables and consider waves Ī“(|x ā x0 | ā a(t ā t0 )) , |x ā x0 |
Ī“(|x ā x0 | + a(t ā t0 )) , |x ā x0 |
which are also solutions of the wave equation with initial values (for t = t0 ) obtained by replacing Ī“(x) with Ī“(x ā x0 ) in (10.26) and (10.27).
10.4. Spherical waves and the Cauchy problem
231
Now, we want to apply the invariance of the symplectic form (Proposition 10.3) to an arbitrary solution u(t, x) of the equation u = 0 and to the converging wave Ī“(|x ā x0 | + a(t ā t0 )) v(t, x) = . |x ā x0 | We will assume that u ā C 1 ([0, t0 ] Ć R3x ) and u = 0 holds (where is understood in the distributional sense). Then we can deļ¬ne
ā ā v(t, x), u(t, x) ā v(t, x), u(t, x) , (uvĖ ā uv)dx Ė = [u, v] = āt āt R3 where Ā·, Ā· denotes the value of the distribution of the variable x at a test function (t is considered as a parameter) which makes sense for u ā C 1 (not only for u ā C0ā ) due to explicit formulas (10.23), (10.25). Now, we use [u, v] = const, which holds for u ā C 2 and is proved similarly to Proposition 10.3 or by approximating u and v by smooth solutions of the wave equation with compact (with respect to x) supports (this can be performed, for instance, with the help of mollifying). Actually, we only have to verd [u, v] = [u, Ė v] + [u, v], Ė which is clear, for example, from explicit ify that dt formulas (10.23), (10.25) which deļ¬ne the functionals v and v. Ė Now, let us explicitly write the relation (10.28)
[u, v]|t=0 = [u, v]|t=t0 . Set u|t=0 = Ļ(x), āu āt t=0 = Ļ(x). Then ā Ī“(|x ā x0 | + a(t ā t0 )) , Ļ(x) [u, v]|t=0 = āt |x ā x0 | t=0 Ī“(|x ā x0 | + a(t ā t0 )) , Ļ(x) ā |x ā x0 | t=0 Ī“(|x ā x0 | ā at0 ) ā Ī“(|x ā x0 | ā at0 ) , Ļ(x) ā , Ļ(x) = ā āt0 |x ā x0 | |x ā x0 | = >
1 ā 1 = ā Ļ(x)dSat0 ā Ļ(x)dSat0 . at0 |xāx0 |=at0 āt0 at0 |xāx0 |=at0
Further, by (10.27), we have [u, v]|t=t0 = ā4ĻaĪ“(x ā x0 ), u(t0 , x) = ā4Ļau(t0 , x0 ). Therefore, (10.28) can be expressed in the form ā 4Ļau(t0 , x0 ) = >
ā 1 1 Ļ(x)dSat0 ā Ļ(x)dSat0 . =ā at0 |xāx0 |=at0 āt0 at0 |xāx0 |=at0
232
10. The wave equation
Replacing t0 , x0 , x by t, x, y we get (10.29) u(t, x) = = >
ā 1 1 Ļ(y)dSat + Ļ(y)dSat , āt 4Ļa2 t |yāx|=at 4Ļa2 t |yāx|=at where dSat is the area element of the sphere {y : |y ā x| = at}. We obtained the formula for the solution of the Cauchy problem āu = Ļ(x). (10.30) u = 0, u|t=0 = Ļ(x), āt t=0
Formula (10.29) is called the Kirchhoļ¬ formula. Let us derive some corollaries of it. 1) The solution of the Cauchy problem is unique and continuously depends on initial data in appropriate topology (e.g., if Ļ continuously varies in C(R3 ) and Ļ continuously varies in C 1 (R3 ), then u(t, x) continuously varies in C(R3 ) for each t > 0). 2) The value u(t, x) only depends on the initial data near the sphere {y : |y ā x| = at}. Suppose that for t = 0 the initial value diļ¬ers from 0 in a small neighborhood of one point x0 only. But then at t the perturbation diļ¬ers from 0 in a small neighborhood of the sphere {x : |x ā x0 | = at} only. This means that a perturbation spreads at a speed a and vanishes after a short time at each observation point. Therefore, the spreading wave has a leading edge and a trailing one. A strictly localized initial state is observed later at another point as a phenomenon, which is also strictly localized. This phenomenon is called the Huygens principle. Since Huygensās principle holds for the wave equation for n = 3, we may transmit information with the help of sound or light. As we will see later, the Huygens principle does not hold for n = 2 (once started, oscillations never cease, e.g., waves on the water). Therefore, the residents of Flatland (the ļ¬at two-dimensional world described by E. A. Abbott in 1884) would ļ¬nd it very diļ¬cult to pass information through their space (the plane). We have not yet proved the existence of the solution for the Cauchy problem (10.30). The simplest way to do this is to take the function u(t, x) constructed by the Kirchhoļ¬ formula (10.29) and verify that it is a solution of (10.30). The fulļ¬llment of u = 0 may be deduced from the fact that ā 1 Ī“(rāat) the right-hand side of (10.29) is the sum of two convolutions Ļ ā āt 4Ļ r 1 Ī“(rāat) 3 with t as a parameter) of Ļ and Ļ ā 4Ļ (taken with respect to x ā R r
10.4. Spherical waves and the Cauchy problem
233
and Ļ with distributions of x depending smoothly on t and satisfying in a natural sense the wave equation (for t > 0). We will use more convenient arguments. In order not to bother with distributions depending on a parameter, it is convenient to consider Ī“(rāat) r as a distribution on R4t,x by setting Ī“(r ā at) , Ļ(t, x) r t,x
ā
ā Ī“(r ā at) 1 , Ļ(t, x) dt = dt = Ļ(t, x)dSat r at 0 0 |x|=at x
ā
= dt at Ļ(t, atx )dS1 , Ļ ā D(R4 ). 0
|x |=1
This formula implies, in particular, that the functional on D(R4 ) determined . by the left-hand side is a distribution which we will again denote by Ī“(rāat) r Ī“(rāat) 4 Therefore, we may assume that ā D (R ). r = K + , where K + is the upper sheet of the light Clearly, supp Ī“(rāat) r 2 2 2 cone {(t, x) : |x| ā a t = 0}; namely, K + = {(t, x) : |x| = at} = {(t, x) : |x|2 ā a2 t2 = 0, t ā„ 0}. It is also easy to verify that Ī“(rāat) = 0 for |t|+|x| = 0. Indeed, Ī“(rāat) =0 r r Ī“(rāat) + = 0 for t > 0. But outside K ; therefore, it suļ¬ces to check that r this is clear, for example, from (10.22) which is true also when the limit is understood in D ({(t, x) : t > Īµ > 0}) for any Īµ > 0. It is easy to prove that the right-hand side of (10.29) can be written in the form 1 Ī“(r ā at) ā 1 Ī“(r ā at) + [Ļ(x) ā Ī“(t)] ā 4Ļ r āt 4Ļ r and then u = 0 obviously holds due to the properties of convolutions.
(10.31) u = [Ļ(x) ā Ī“(t)] ā
The initial conditions in (10.30) may also be deduced from (10.31); we will, however, deduce them directly from the structure of formula (10.29). Note that Kirchhoļ¬ās formula is of the form ā u = uĻ + uĻ , āt where
1 Ļ(y)dSat uĻ (t, x) = 4Ļa2 t |yāx|=at and uĻ is a similar expression obtained by replacing Ļ by Ļ. This structure of Kirchhoļ¬ās formula is not accidental. Indeed, suppose we have proved
234
10. The wave equation
that uĻ is a solution of the Cauchy problem (10.32)
uĻ = 0,
Let us prove then that vĻ = (10.33)
uĻ |t=0 = 0,
āuĻ = Ļ. āt t=0
ā āt uĻ
vĻ = 0,
is a solution of the Cauchy problem āvĻ vĻ |t=0 = Ļ, = 0. āt t=0
The equation vĻ = 0 obviously holds and so does vĻ |t=0 = Ļ due to the second equation in (10.32). It remains to verify the second equality in (10.33). Assuming that uĻ ā C 2 for t ā„ 0, we get ā 2 uĻ āvĻ = = a2 ĪuĻ |t=0 = a2 Ī(uĻ |t=0 ) = 0, āt t=0 āt2 t=0 as required. The condition uĻ ā C 2 (t ā„ 0) holds, for example, for Ļ ā C 2 (R3 ), as is clear for example from the expression
t Ļ(x + aty )dS1 . uĻ (t, x) = 4Ļ |y |=1 This expression makes (10.32) obvious for Ļ ā C 1 (R3 ) also. Thus, if Ļ ā C 1 (R3 ), Ļ ā C 2 (R3 ), then Kirchhoļ¬ās formula (10.29) gives a solution of the Cauchy problem (10.30) (the equation u = 0 holds in the distributional sense for t > 0). If we additionally assume that Ļ ā C 2 (R3 ) and Ļ ā C 3 (R3 ), then u ā C 2 (R3 ) and u = 0 holds in the classical sense. Thus the Cauchy problem (10.30) is uniquely solvable. The Kirchhoļ¬ formula also makes it clear that it is well-posed.
10.5. The fundamental solution for the three-dimensional wave operator and a solution of the nonhomogeneous wave equation Set E3 (t, x) =
1 Ī“(|x| ā at) ā D (R4 ). 4Ļa |x|
Theorem 10.4. The distribution E3 (t, x) satisļ¬es E3 (t, x) = Ī“(t, x); i.e., it is a fundamental solution for the dāAlembert operator. Proof. The statement actually follows from the fact that E3 (t, x) = 0 for |t| + |x| = 0, E3 (t, x) = 0 3 E3 (+0, x) = 0, āE āt (+0, x) = Ī“(x).
for t < 0,
10.5. The fundamental solution for the three-dimensional operator
235
The function E3 (t, x) may be considered as a smooth function of t (for t = 0) 3 with values in D (R3 ) and for t = 0 it is continuous while āE āt has a jump 2 ā 2 equal to Ī“(x). Therefore, applying = āt 2 āa Īx to E3 , we get Ī“(t)āĪ“(x) = Ī“(t, x). It is diļ¬cult to make the above arguments entirely rigorous (if we try to do this, we face the necessity of a cumbersome struggle with the topology in D (R3 )). We may, however, consider all the above as a heuristic hint and directly verify that E3 = Ī“. This can be done similarly to the proof for the heat equation (see the proof of Theorem 7.5) and it is left to the reader as an exercise. Now, using E3 (t, x) we can solve the nonhomogeneous wave equation u = f by writing (10.34)
u = E3 ā f,
if the convolution in the right-hand side makes sense. The convolution (10.34) is called the retarded potential. In a more detailed expression, the retarded potential is of the form
ā
f (t ā Ļ, x ā y) 1 dSaĻ (10.35) dĻ u(t, x) = 4Ļa 0 |y| |y|=aĻ
ā dĻ 1 f (t ā Ļ, x ā y)dSaĻ = 4Ļa2 0 Ļ |y|=aĻ (here the integration in the second integrals is done over y). With the help of E3 (t, x), we may obtain the Kirchhoļ¬ formula (10.29) for the solution of the Cauchy problem (10.30) in a diļ¬erent way. Namely, let a solution u(t, x) of the Cauchy problem (10.30) be given for t > 0. Extend it by zero onto the half-space {t : t < 0} and consider the obtained distribution u ā D (R4 ). Clearly, (10.36)
u = Ī“ (t) ā Ļ(x) + Ī“(t) ā Ļ(x).
A solution of this equation vanishing for t < 0 may be found in the form of a retarded potential (10.34) taking f equal to the right-hand side of (10.36). Clearly, we again get the Kirchhoļ¬ formula. Note also that the retarded potential u(t, x) obtained via (10.35) depends only on values f (t , x ) for t ā¤ t and |x āx| = a(t āt ), i.e., on values of f at points of the lower part of the light cone with the vertex at (t, x) (the word ālowerā here is understood with reference to the direction of the t-axis). This is what the term āretarded potentialā really means. This property of the retarded potential is ensured by the fact that supp E3 ā K + .
236
10. The wave equation
A fundamental solution for with this property is unique. Even a stronger result holds: Theorem 10.5. There exists exactly one fundamental solution for with the support belonging to the half-space {(t, x) : t ā„ 0}, namely, E3 (t, x). Proof. If there were two such solutions, then their diļ¬erence u(t, x) ā D (R4 ) would have satisļ¬ed the wave equation u = 0 and would have vanished for t < 0. If u were a smooth function, then the uniqueness of the solution of the Cauchy problem would imply u ā” 0 (Cauchy initial values vanish for t = t0 < 0). We can, however, make u smooth by mollifying. Let ĻĪµ ā D(R4 ), supp ĻĪµ ā {(t, x) : |t|+|x| ā¤ Īµ}, ĻĪµ ā„ 0, and ĻĪµ dtdx = 1, so that ĻĪµ ā Ī“(t, x) in D (R4 ) as Īµ ā +0. Then limĪµā+0 (u ā ĻĪµ ) = u in D (R4 ) (see Proposition 5.4). But u ā ĻĪµ ā C ā (R4 ) and (u ā ĻĪµ ) = (u) ā ĻĪµ = 0, and also supp(u ā ĻĪµ ) ā supp u + supp ĻĪµ ā {(t, x) : t ā„ āĪµ}. Therefore, due to the uniqueness of the smooth solution of the Cauchy problem for the wave equation, u ā ĻĪµ ā” 0 for any Īµ > 0, implying u ā” 0, as required. The importance of Theorem 10.5 stems from the fact that among all the fundamental solutions it chooses the only one which satisļ¬es the causality principle, which claims that it is impossible to transmit information to the āpastā. Therefore, in electrodynamics, to solve equations of the form u = f , satisļ¬ed by the scalar and vector potentials, one uses just this fundamental solution. It seems that Nature itself has chosen the solution satisfying the causality principle among all the solutions of the equation u = f (within the limits of the current accuracy of experiments).
10.6. The two-dimensional wave equation (the descent method) Let us solve the Cauchy problem for the equation 2 ā 2u ā2u 2 ā u =a + , u = u(t, x1 , x2 ), (10.37) āt2 āx21 āx22 with the initial conditions (10.38)
u|t=0 = Ļ(x1 , x2 ),
āu = Ļ(x1 , x2 ). āt t=0
The idea of the solution (the descent method) is very simple: introduce an additional variable x3 and ļ¬nd the solution of the Cauchy problem for the
10.6. The two-dimensional wave equation (the descent method)
237
three-dimensional wave equation utt = a2 Īu (where Ī is the Laplacian with respect to x1 , x2 , x3 ), with the initial conditions (10.38), which are independent of x3 . Then the solution u(t, x1 , x2 , x3 ) does not actually depend on x3 , since the function uz (t, x1 , x2 , x3 ) = u(t, x1 , x2 , x3 + z) is a solution of the same equation utt = a2 Īu for any z with the same initial conditions (10.38); therefore, the uniqueness theorem for the solution of the Cauchy problem for the three-dimensional wave equation implies that uz does not depend on z; i.e., u does not depend on x3 . Therefore the solution of the Cauchy problem (10.37)ā(10.38) does exist (e.g., for any Ļ ā C 2 (R2 ), Ļ ā C 1 (R2 )). It is unique due to the uniqueness theorem concerning the three-dimensional case since the solution of (10.37)ā(10.38) can also be regarded as a solution of the three-dimensional Cauchy problem. Now, let us express u(t, x1 , x2 ) via the Kirchhoļ¬ formula = >
1 ā Ļ(y1 , y2 )dSat (10.39) u(t, x) = āt 4Ļa2 t |yāx|=at
1 Ļ(y1 , y2 )dSat , + 4Ļa2 t |yāx|=at where y = (y1 , y2 , y3 ), dSat is the area element of the sphere {y : y ā R3 , |y ā x| = at}, and x = (x1 , x2 ) = (x1 , x2 , 0) (we identify the points of R2 with the points of R3 whose third coordinate is 0; this coordinate might have any other value since it does not aļ¬ect anything). Let us rewrite the second summand in formula (10.39), which we will denote uĻ :
1 Ļ(y1 , y2 )dSat (10.40) uĻ = 4Ļa2 t |yāx|=at ā (the ļ¬rst summand in (10.39) is of the form āt uĻ ). 3 Consider the sphere in Ry over which we integrate in (10.40). This is the sphere with the center at x of radius at (see Figure 10.1). We are to integrate a function, which is independent of y3 , over the sphere. This actually means that we integrate twice over the projection of the sphere onto the plane y3 = 0. Let dy1 dy2 be the Lebesgue measure of this plane, dSat the area
y3 y2 y1
x
Figure 10.1. To the descent method.
238
10. The wave equation
element of the sphere at y whose projection is equal to dy1 dy2 . Clearly, dy1 dy2 = | cos Ī±(y)|dSat , where Ī±(y) is the angle between the normal to the sphere and the y3 -axis. But the normal is proportional to the vector y ā x = (y1 ā x1 , y2 ā x2 , y3 ) of length at. Let us integrate over the upper half-sphere (y3 > 0) and then double the result. Then |y ā x| = at implies y3 = a2 t2 ā (y1 ā x1 )2 ā (y2 ā x2 )2 , 1 2 2 y3 = a t ā (y1 ā x1 )2 ā (y2 ā x2 )2 , cos Ī±(y) = at at atdy1 dy2 dy1 dy2 dSat = = . 2 2 cos Ī±(y) a t ā (y1 ā x1 )2 ā (y2 ā x2 )2 Therefore, we can rewrite (10.40) in the form
1 Ļ(y)dy uĻ (t, x) = , 2Ļa |yāx|ā¤at a2 t2 ā |y ā x|2 where x = (x1 , x2 ), y = (y1 , y2 ), dy = dy1 dy2 . The solution of the Cauchy problem (10.37)ā(10.38) is given by the formula = >
1 Ļ(y)dy ā u(t, x) = āt 2Ļa |yāx|ā¤at a2 t2 ā |y ā x|2 (10.41)
Ļ(y)dy 1 + 2Ļa |yāx|ā¤at a2 t2 ā |y ā x|2 known as Poissonās formula. It is clear from Poissonās formula that the value of the solution u(t, x) at x for n = 2 depends upon initial values Ļ(y) and Ļ(y) in the disc {y : |y ā x| ā¤ at} and not only on its boundary (recall that for n = 3 it was suļ¬cient to know initial values on the sphere {y : |y ā x| = at}). In particular, if the supports on Ļ, Ļ are near the origin, then the solution in a neighborhood of x is always nonzero after a certain moment. Therefore, a localized perturbation is not seen as a localized one from another point; i.e., the wave does not go tracelessly but leaves an aftereļ¬ect, though decreasing with time as 1t , as is clear from (10.41). In other words, the Huygens principle fails for n = 2. The method of descent enables us also to derive dāAlembertās formula from Poissonās formula, the former determining the solution of the Cauchy problem for the one-dimensional wave equation. However, we have already obtained this formula by another method. Let us also ļ¬nd a fundamental solution for the two-dimensional wave operator. As in the three-dimensional case, we need to solve the Cauchy problem with the initial conditions u|t=0 = 0, ut |t=0 = Ī“(x).
10.7. Problems
239
But it is clear from Poissonās formula that such a solution E2 (t, x) has the form Īø(at ā |x|) E2 (t, x) = , x ā R2 . 2Ļa a2 t2 ā |x|2 Now, it is easy to verify directly that E2 (t, x) is locally integrable and is a fundamental solution for the two-dimensional wave operator. The latter fact is veriļ¬ed in exactly the same way as in the three-dimensional case and we leave this to the reader as an exercise. Finally, dāAlembertās formula makes it clear that a fundamental solution ā2 2 ā2 for the one-dimensional wave operator āt 2 ā a āx2 is of the form E1 (t, x) =
1 Īø(at ā |x|). 2a
10.7. Problems 10.1. By separation of variables ļ¬nd cylindric waves in R3 . 10.2. Solve the āexplosion-of-a-ball problemā in the three-dimensional space: ļ¬nd u = u(t, x), x ā R3 , if utt = a2 Īu,
u|t=0 = Ļ,
ut |t=0 = 0,
where Ļ is the characteristic function of the ball {x : |x| ā¤ R}. Draw an animation describing the behavior of u(t, x) as a function of t and |x|. 10.3. The same as in the above problem but with diļ¬erent initial conditions: u|t=0 = 0, ut |t=0 = Ļ, where Ļ is the characteristic function of the ball {x : |x| ā¤ R}. 10.4. Using the result of Problem 5.1 c), write, with the help of the Fourier transform, a formula for the solution of the Cauchy problem for the threedimensional wave equation. (Derive once more Kirchhoļ¬ās formula in this way.) 10.5. With the help of the Fourier transform with respect to x, solve the Cauchy problem for the wave equation utt = a2 Īu,
u = u(t, x),
x ā Rn .
Write the fundamental solution for the n-dimensional dāAlembertian = ā2 ā a2 Ī in the form of an integral and prove that its singularities belong āt2 to the light cone {(t, x) : |x|2 = a2 t2 }. 10.6. Describe the location of the singularities of the fundamental solution ā2 ā2 ā2 for the operator āt 2 ā āx2 ā 4 āy 2 .
240
10. The wave equation
10.7. Let u(t, x) be a solution of the Cauchy problem of the equation utt = Īu, x = (x1 , x2 ) ā R2 with the initial conditions u|t=0 = Ļ, ut |t=0 = Ļ. a) Let Ļ, Ļ be known in the rectangle x1 ā [0, a], x2 ā [0, b]. Where can u be determined? Draw the domain in R3t,x1 ,x2 in which u can be determined (for t > 0). b) Let the supports of Ļ, Ļ belong to the rectangle x1 ā [0, a], x2 ā [0, b]. Where is u = 0 for t > 0? Draw this domain in R3t,x1 ,x2 . 10.8. Solve Problem 10.7 for x ā R3 with the rectangle replaced by a rectangular parallelepiped. Instead of the domain in R4t,x , draw its section by the hyperplane t = 1.
10.1090/gsm/205/11
Chapter 11
Properties of the potentials and their computation
We have already met the potentials in this book. In Remark 4.11 we discussed the physical meaning of the fundamental solution of the Laplace operator in R3 and R2 . It is explained there that the fundamental solution E3 (x) of the Laplace operator in R3 has the meaning of the potential of a point charge, and the fundamental solution E2 (x) of the Laplace operator in R2 has the meaning of the potential of a thin uniformly charged string. It is impossible to deļ¬ne the potential of the string as an ordinary integral (it diverges). We indicated two methods for deļ¬ning this potential: its recovery from the strength of the ļ¬eld (which is minus the gradient of the potential) and making the divergent integral meaningful by renormalization of the charge. Both methods deļ¬ne the potential of the string up to a constant, which suļ¬ces since in the long run we are only interested in the strength of the ļ¬eld which is physically observable, whereas the potential is an auxiliary mathematical object (at least in the classical electrodynamics and gravitation theory). In Example 5.2 we introduced distributions, called single and double layers on the plane t = 0 and on an arbitrary hypersurface, and explained their physical meaning. In Example 5.6 we introduced the potentials of the simple and double layers as convolutions of the fundamental solution En (x) with the single and double layers on the surface.
241
242
11. Potentials and their computation
In Section 10.1 the role of the scalar and vector potentials in electrodynamics was clariļ¬ed. Here we will discuss in detail properties of potentials (and also how to deļ¬ne and calculate them). However, before we start the details, we advise the reader to reread the above-indicated passages. Potentials will be understood with suļ¬cient versatility. When the preliminary deļ¬nitions that we begin with become unsuitable, we will modify them at our convenience trying all the time to preserve physical observables.
11.1. Deļ¬nitions of potentials Let En (x) be the standard fundamental solution for Ī in Rn ; i.e., 1 En (x) = |x|2ān , for n ā„ 3, (2 ā n)Ļnā1 where Ļnā1 is the area of the unit sphere in Rn and 1 ln |x|. E2 (x) = 2Ļ In particular, the case n = 3 is the most important one: 1 . E3 (x) = ā 4Ļ|x| E3 (x) is the potential of a point charge, which is equal to +1 for a proper system of units, and E2 (x) is the potential of an inļ¬nite uniformly charged string (see, e.g., Feynman, Leighton, and Sands [8, vol. 2, Ch. 4]). All potentials have the form (11.1)
āEn ā f
for diļ¬erent distributions f ā E (Rn ) and, therefore, for n = 3 and 2 have the meaning of potentials of some distributed systems of charges. Every potential u of the form (11.1) satisļ¬es in D (Rn ) the Poisson equation Īu = āf. Now we will describe the potentials which we need, turn by turn. a) The volume potential of charges distributed with density Ļ(x) is the integral
En (x ā y)Ļ(y)dy. (11.2) u(x) = ā Rn
We will always suppose that Ļ(y) is a locally integrable function with compact support and Ļ(y) is piecewise smooth; i.e., Ļ is a C ā -function outside of ļ¬nitely many smooth hypersurfaces in Rn on which it may have jumps. All the derivatives ā Ī± Ļ are supposed to satisfy the same condition (ā Ī± Ļ is
11.1. Deļ¬nitions of potentials
243
taken here outside of the hypersurfaces). An example of an admissible function Ļ(y) is the characteristic function of an arbitrary bounded domain with smooth boundary. The most important property of the volume potential is that it satisļ¬es the equation Īu = āĻ understood in the distributional sense; this follows from the equalities u = āEn ā Ļ and ĪEn (x) = Ī“(x). b) The single layer potential is the integral
(11.3) u(x) = ā En (x ā y)Ļ(y)dSy , Ī
where Ī is a smooth hypersurface in Rn , Ļ is a C ā -function on Ī, and dSy is the area element on Ī. For the time being, we assume that Ļ has a compact support (in the sequel, we will encounter examples when this does not hold). This potential satisļ¬es Īu = āĻĪ“Ī , where ĻĪ“Ī is the distribution with the support on Ī deļ¬ned in Example 5.2. c) The double layer potential is the integral
āEn (x ā y) Ī±(y)dSy , (11.4) u(x) = ān ĀÆy Ī where n ĀÆ y is the outward (or in some other way chosen) normal to Ī at y and the remaining notation is the same as in the preceding case. The potential of the double layer satisļ¬es ā Īu = ā (Ī±(y)Ī“Ī ), ān ĀÆ ā where ā nĀÆ (Ī±Ī“Ī )āthe ādouble layerāāis the distribution deļ¬ned in Example 5.2. In this case we assume for the time being that Ī± has a compact support and that Ī± ā C ā (Ī). It is easy to see that the potentials (11.2)ā(11.4) are deļ¬ned and inļ¬nitely diļ¬erentiable on Rn \Ī (where in the case of the volume potential (11.2), Ī is understood as the union of surfaces where Ļ or its derivatives have jumps). We will understand these potentials everywhere in Rn as distributions obtained as convolutions (11.1), where f for the volume potential is the same as Ļ in (11.2); for the case of the single layer potential f = ĻĪ“Ī , where
Ļ(x)Ļ(x)dSx , Ļ ā D(Rn ), ĻĪ“Ī , Ļ = Ī
whereas for the case of the double layer potential ā (Ī±Ī“Ī ), f= ān ĀÆ
244
11. Potentials and their computation
āĻ(x) ā (Ī±Ī“Ī ), Ļ = ā Ī±(x) dSx . ān ĀÆ ān ĀÆx Ī Clearly, the convolutions u = āEn ā f in all the above cases coincide for x ā Rn \Ī with the expressions given by the integrals (11.2)ā(11.4). The physical meaning of the potentials (11.2)ā(11.4) was described in Chapter 5. where
11.2. Functions smooth up to Ī from each side, and their derivatives Let Ī© be a domain in Rn , let Ī be a smooth hypersurface in Ī© (i.e., Ī is a submanifold of codimension 1 that is closed in Ī©), and let u ā C ā (Ī©\Ī). We will say that u is smooth up to Ī from each side if all the derivatives ā Ī± u are continuous up to Ī from each side of the hypersurface. More precisely, if x0 ā Ī, then there should exist a neighborhood U of x0 in Ī© such that U = U + āŖ ĪU āŖ U ā , where ĪU = Ī ā© U , U + and U ā are open, and the restrictions u|U + and u|U ā can be extended to functions u+ ā C ā (U + ) and uā ā C ā (U ā ). In what follows we will often be interested in local questions only, where one should set from the beginning U = Ī© and the above decomposition is of the form Ī© = Ī©+ āŖ Ī āŖ Ī©ā , where u+ ā C ā (Ī©+ ) and uā ā C ā (Ī©ā ). Our aim is to prove that all the potentials are smooth up to Ī from each side (this in turn enables us to prove theorems about jumps of the potentials and calculate the potentials). To this end we will ļ¬rst prove several auxiliary lemmas about functions which are smooth up to Ī from each side. If u ā C ā (Ī©\Ī) is smooth up to Ī from each side, then it uniquely determines a distribution [u] ā D (Ī©), such that [u] ā L1loc (Ī©) and [u]|Ī©\Ī = u. Our immediate goal is to learn how to apply to [u] diļ¬erential operators with smooth coeļ¬cients. Let us agree to write [u] only when u is smooth up to Ī from each side. In a neighborhood U of every point x0 ā Ī we can construct a diļ¬eomorphism of this neighborhood onto a domain in Rn that transforms U ā© Ī into a part of the plane {x : xn = 0}. Here x = (x1 , . . . , xnā1 , xn ) = (x , xn ), where x = (x1 , . . . , xnā1 ) ā Rnā1 . This diļ¬eomorphism induces a change of variables in any diļ¬erential operator transforming it into another diļ¬erential operator of the same order (see Section 1.3). Thus, we will ļ¬rst consider a local version in which U is a suļ¬ciently small neighborhood of 0 ā Rn and the coordinates in U may be selected so that Ī = U ā© {x : xn = 0}, U Ā± = {x : x ā U, x = (x , xn ), Ā±xn > 0},
11.2. Functions smooth up to the boundary
245
āu and the normal n ĀÆ to Ī is of the form n ĀÆ = (0, . . . , 0, 1); i.e., āāunĀÆ = āx . For n simplicity of notation we also temporarily assume that U = Ī©, so U ā© Ī = Ī.
Denote by Ļk the jump of the kth normal derivative of u on Ī; i.e., ā k u+ ā k uā Ļk = ā ā C ā (Ī), āxkn Ī āxkn Ī where k = 0, 1, 2, . . .. Note that the jumps of all the other derivatives are expressed via Ļk by the formulas
(ā Ī± u+ )|Ī ā (ā Ī± uā )|Ī = ā Ī± ĻĪ±n , if Ī± = (Ī± , Ī±n ), where Ī± is an (n ā 1)-dimensional multi-index. 2
2
ā[u] ā [u] Lemma 11.1. The derivatives āx , āx2 , and āxā n[u] āxj (j = 1, 2, . . . , n ā 1) n n are expressed by the formulas ' & āu ā[u] + Ļ0 (x ) ā Ī“(xn ), = (11.5) āxn āxn
(11.6)
& ' āu ā[u] = , āxj āxj
j = 1, . . . , n ā 1,
(11.7)
& 2 ' ā 2 [u] ā u + Ļ1 (x ) ā Ī“(xn ) + Ļ0 (x ) ā Ī“ (xn ), = 2 āxn āx2n
(11.8)
& ' ā 2u āĻ0 ā 2 [u] = ā Ī“(xn ). + āxn āxj āxn āxj āxj
Proof. For any Ļ ā D(Ī©) we have
āĻ āĻ āĻ ā[u] , Ļ = ā [u], [u] dx ā [u] dx =ā āxn āxn āxn āxn xn ā„0 xn ā¤0 & ā '
0 āĻ āĻ = ā dx [u] dxn ā [u] dxn āxn āxn 0 āā '
&
āu + = Ļ(x)dx + u (x , 0)Ļ(x , 0)dx ā uā (x , 0)Ļ(x , 0)dx āxn & '
āu = , Ļ + Ļ0 (x )Ļ(x , 0)dx , āxn which proves (11.5). Integrating by parts with respect to xj easily leads to (11.6). The formula (11.7) is obtained if we twice apply (11.5), whereas (11.8) follows from (11.5) and (11.6). We need to learn how to multiply distributions encountered in the righthand sides of (11.5)ā(11.8) by smooth functions. To this end, let a ā C ā (Ī©)
246
11. Potentials and their computation
and expand a = a(x , xn ) by the Taylor formula in xn near xn = 0. We have a(x , xn ) =
(11.9)
N ā1
ak (x )xkn + xN n a(N ) (x , xn ),
k=0
where ak ā C ā (Ī), a(N ) ā C ā (V ) (here V is a neighborhood of Ī), and 1 ā k a ak (x ) = , k! āxkn xn =0
1 N 1 ā a (x , txn )(1 ā t)N ā1 dt. a(N ) (x , xn ) = (N ā 1)! 0 āxN n Lemma 11.2. If Ļ ā C ā (Ī), then a(x)(Ļ(x ) ā Ī“(xn )) = [a0 (x )Ļ(x )] ā Ī“(xn ),
(11.10)
(11.11) a(x)(Ļ(x )āĪ“ (xn )) = [a0 (x )Ļ(x )]āĪ“ (xn )ā[a1 (x )Ļ(x )]āĪ“(xn ). Proof. Clearly, ak (x )xkn (Ļ(x ) ā f (xn )) = [ak (x )Ļ(x )] ā [xkn f (xn )] for any distribution f ā D (R). Further, N a(N ) (x , xn )xN n (Ļ(x ) ā f (xn )) = a(N ) (x , xn )(Ļ(x ) ā xn f (xn )).
But a direct calculation shows that xpn Ī“ (k) (xn ) = 0 for p > k and xn Ī“ (xn ) = āĪ“(xn ), which, with a(x , xn ) in the form (11.9), immediately yields (11.10) and (11.11). Lemma 11.3. For any set of functions Ļ0 , Ļ1 , . . . , ĻN ā C ā (Ī), there exists a function u ā C ā (Ī©\Ī) smooth up to Ī from each side, such that Ļk is the āk u jump of āx k on Ī for all k = 0, 1, . . . , N . n
Proof. We may, for example, set u(x) =
N 1 Ļk (x )xkn Īø(xn ), k! k=0
where Īø(t) is the Heaviside function. Lemma 11.4. Let a distribution f ā D (Ī©) be of the form (11.12)
f (x) = [f0 (x)] + b0 (x ) ā Ī“(xn ) + b1 (x ) ā Ī“ (xn ),
where b0 , b1 ā C ā (Ī) and f0 is smooth up to Ī from each side. Let A be a second-order linear diļ¬erential operator such that Ī is noncharacteristic for A at every point. Then for any integer N ā„ 0 there exists a function u on a neighborhood Ī© of Ī in Ī© such that u is smooth up to Ī from each side, and (11.13)
A[u] ā f ā C N (Ī© ).
11.2. Functions smooth up to the boundary
247
Proof. Given Ļ0 , Ļ1 , . . . , ĻN ā C ā (Ī), we can use Lemma 11.3 to construct u ā C ā (Ī©\Ī) having Ļk as the jump of āu/āxkn on Ī. It is clear from Lemmas 11.1 and 11.2 that the distribution A[u] = fĖ always has the form (11.12) (but perhaps with other f0 , b0 , b1 ). We need to select the jumps Ļ0 , Ļ1 , . . . , ĻN +2 in such a way that fĖ(x) = [fĖ0 (x)] + b0 (x ) ā Ī“(xn ) + b1 (x ) ā Ī“ (xn ) with the same functions b0 , b1 as in (11.12) and also the jumps of and
ā k fĖ0 āxkn
ā k f0 āxkn
on Ī coincide for k = 0, 1, . . . , N . (Then we will obviously have Ė f ā f ā C N (Ī©), as required.) First, note that Ī being noncharacteristic for A means that A can be expressed in the form (11.14)
A = a(x)
ā2 + A , āx2n
ā where a ā C ā (Ī©), a(x , 0) = 0, and A does not contain āx 2 . Since a(x) = 0 n in a neighborhood of Ī, we can divide (11.13) by a(x) and reduce the problem to the case when a(x) = 1. Therefore, in what follows we assume that 2
(11.15)
A=
ā2 + A , āx2n
where A is a second-order diļ¬erential operator that does not contain
ā2 . āx2n
Let us start with selecting a function v0 such that A[v0 ] = [fĖ0 (x)] + Ėb0 (x ) ā Ī“(xn ) + b1 (x ) ā Ī“ (xn ), where fĖ0 is any function (smooth up to Ī from each side), Ėb0 ā C ā (Ī). (0) Lemmas 11.1 and 11.2 show that it suļ¬ces to take Ļ0 (x ) = b1 (x ), where (0) Ļ0 is the jump of v0 on Ī. Now, (11.13) may be rewritten in the form (11.16)
A[u ā v0 ] = [f1 ] + b2 (x ) ā Ī“(xn )
with some f1 and b2 , where b2 ā C ā (Ī). Now, select a function v1 such that A[v1 ] = [fĖ1 ] + b2 (x ) ā Ī“(xn ), (1) (1) where b2 is the same as in (11.16) and fĖ1 is arbitrary. If Ļ0 and Ļ1 are (1) (1) āv1 on Ī, then it suļ¬ces to take Ļ0 = 0 and Ļ1 (x ) = jumps of v1 and āx n b2 (x ). Setting u2 = u ā v0 ā v1 we see that the condition (11.13) for u reduces to A[u2 ] ā [f2 ] ā C N (Ī© ), where f2 is a function that is smooth up to Ī from each side. We will prove the possibility to select such u2 by induction with respect to N , starting
248
11. Potentials and their computation
with N = ā1 and deļ¬ning C ā1 (Ī© ) as the set of all functions in Ī© that are smooth up to Ī from each side. For N = ā1 everything reduces to the simple condition u2 ā C 1 (Ī© ), āu2 on Ī. i.e., to the absence of jumps of u2 and āx n Now suppose that we have already selected a function uk+2 such that (11.17)
A[uk+2 ] ā [f2 ] ā C kā1 (Ī© ),
where k is a nonnegative integer (as we have just seen this is possible for k = 0). Then, as an induction step, we need to select a function uk+3 such that (11.18)
A[uk+3 ] ā [f2 ] ā C k (Ī© ).
To this end set uk+3 = uk+2 + vk+2 , where vk+2 ā C k+1 (Ī© ) (we do this in order to preserve (11.17) when we replace uk+2 by uk+3 ). Clearly, A [uk+2 ] ā C k (Ī© ) and, therefore, (11.18) reduces to ā 2 [vk+2 ] ā [fĖk ] ā C k (Ī© ), āx2n where fĖk = ā(A[uk+2 ] ā [f2 ]) ā C kā1 (Ī© ). This can be obtained by setting, for example, k fĖ+ k fĖā ā ā 1 k k xk+2 vk+2 (x) = ā Īø(xn ) n k k (k + 2)! āxn āxn Ī
Ī
ā 2 [v
]
in order to make the jumps of the kth derivatives of the functions āxk+2 2 n Ė and fk on Ī coincide. Thus, we have proved that the induction with respect to N can be applied. This completes the proof. Global version. Fermi coordinates Now we will brieļ¬y describe a global version of the results above. Let us introduce appropriate notation. Let Ī be a closed hypersurface in Rn , that is an (n ā 1)-dimensional compact closed submanifold of Rn without boundary. Let us take a function s : Rn ā R, deļ¬ned as follows: s(x) = dist(x, Ī) if x is outside of the (closed, compact) body B bounded by Ī (in particular, Ī = āB), and s(x) = ādist(x, Ī) if x is inside this body, and s = 0 on Ī. Clearly, s < 0 on the interior of B, and s > 0 on Rn \ B. It is easy to see that s ā C ā in a suļ¬ciently small neighborhood of Ī. Indeed, let n ĀÆ = n ĀÆ (z) denote the outward unit normal vector to Ī at the point z ā Ī. Consider the map (11.19)
Ī Ć R ā Rn ,
(z, s) ā z + sĀÆ n.
11.2. Functions smooth up to the boundary
249
Its diļ¬erential is obviously bijective at the points (z, 0) ā Ī Ć R. Therefore, by the inverse function theorem, the map (11.19) restricts to a diļ¬eomorphism of Ī Ć (āĪµ, Īµ) onto an open subset ĪĪµ of Rn , provided Īµ > 0 is suļ¬ciently small. In ĪĪµ , it is convenient to use Fermi coordinates which have the form where y = (y1 , . . . , ynā1 ) is a set of local coordinates on Ī and yn = s ā (āĪµ, Īµ). If the (n ā 1)-tuple y represents a point z ā Ī, then, by n ā ĪĪµ . If we have deļ¬nition, y = (y , s) are coordinates of the point z + sĀÆ a second-order diļ¬erential operator A such that Ī is noncharacterisic for A at every point, then, as in (11.14), we can write in any Fermi coordinate system (y , yn ): (y , yn ),
(11.20)
A = b(y)
ā2 + B, āyn2
ā where b ā C ā (ĪĪµ ), b(y , 0) = 0, B does not contain āy 2 . Now, if we pass n from one Fermi coordinate system to another one yĖ = (Ė y , yn ), then the ļ¬rst term in the right-hand side of (11.20) will not change its form, except for change of variables in the scalar function b. Therefore, the function b is a well-deļ¬ned scalar function on ĪĪµ . Then in any equation of the form Au = f we can divide by b on a small neighborhood of Ī = Ī Ć {0}. For example, we can replace Īµ by a smaller value and assume that b(y) = 0 for all y ā ĪĪµ . 2
For the Laplacian Ī in Rn and Fermi coordinates y with respect to any hypersurface Ī, the formula (11.20) simpliļ¬es: then b(y) = 1 for all y, so that we have (11.21)
Ī=
ā2 + B, āyn2
and, moreover, B does not contain any second derivatives of the form āyjāāxn with j ā¤ n. Both statements follow from a simple geometric fact (a generalization of a Gauss lemma) that in Fermi coordinates āyān is the gradient of s = yn , or, equivalently, that the invariantly deļ¬ned tangent vector ļ¬eld ā āyn is orthogonal to the level sets of the signed distance function s = yn . (See the proof in a much more general context in Gray [10, Chapter 2]; Rn can be replaced by an arbitrary Riemannian manifold M , and Ī by a closed submanifold of an arbitrary dimension of M . As an exercise, the reader can ļ¬nd an elementary proof based on the fact that for small Īµ the straight line interval {z + tsĀÆ n| 0 ā¤ t ā¤ 1} is the shortest way between the level hypersurfaces yn = 0 and yn = s , where s ā [0, Īµ] is arbitrarily ļ¬xed.) 2
For more information and details about Fermi coordinates see Gray [10, Chapter 2].
250
11. Potentials and their computation
To understand how the statement about the second derivatives follows from the above geometric fact, let us recall that in any local coordinates y = (y1 , . . . , yn ) the Laplacian can be written in the form of the LaplaceBeltrami operator, which is given by 1 ā ij ā āu g (11.22) Īu = ā g . g āyi āyj 1ā¤i,jā¤n
(g ij )
is the inverse matrix for (gij ), where gij = gij (y) is the symmetric Here positive deļ¬nite matrix of the standard Riemannian metric on Rn (i.e., ds2 = dx21 + Ā· Ā· Ā· + dx2n ) in coordinates y; that is, ā ā , , gij (x) = āyi āyj x where the tangent vectors āyā j are considered as vectors in Rn and Ā·, Ā· is the standard scalar product in Rn . (See, for example, Taylor [35, Vol. I, Sect. 2.4] for the details about the Laplace-Beltrami operator.) In Fermi coordinates, note that the vector
ā āyn
has length one; hence
gnn = 1. The absence of the mixed second derivatives
ā2 āyj āxn ,
j < n, is
equivalent to gjn = 0, which implies that for the inverse matrix (g ij ) we also have g nn = 1 and g jn = 0 if j < n. This implies the desired absence of the mixed derivatives. Now we can globalize the arguments given above in the proofs of Lemmas 11.1ā11.4 (the local case), except we need to replace Ļ(x ) ā Ī“(xn ) by ĻĪ“Ī where Ļ ā C ā (Ī), and Ļ(x )āĪ“ (xn ) by ā āānĀÆ (ĻĪ“Ī ). This leads, in particular, to the following global version of Lemma 11.4: Lemma 11.5. Let Ī be a closed compact hypersurface in Rn , and let f ā D (Rn ) be of the form ā (b1 (x)Ī“Ī ), ān ĀÆ where b0 , b1 ā C ā (Ī) and f0 is a locally integrable function that is smooth up to Ī from each side. Let A be a second-order linear diļ¬erential operator in Rn , such that Ī is noncharacteristic for A at every point. Then for any integer N ā„ 0 there exists an integrable compactly supported function u in Rn such that u is smooth up to Ī from each side, and (11.23)
(11.24)
f (x) = [f0 (x)] + b0 (x)Ī“Ī +
A[u] ā f ā C N (Rn ).
We leave the details of the proof as an exercise for the reader. Note only that if we found u satisfying (11.24) but without compact support, then we can replace u by Ļu where Ļ ā C0ā (Rn ), Ļ = 1 in a neighborhood of Ī. Then Ļu will satisfy all the requirements of the lemma.
11.3. Jumps of potentials
251
Remark 11.6. Lemma 11.5 is a simpliļ¬ed version of more general results. It is not necessary to require compactness of Ī. For example, the proof works if Ī is a hyperplane or an inļ¬nite cylinder with a compact base. In fact, it suļ¬ces that Ī is a closed submanifold in Rn , though the proof requires a minor modiļ¬cation: we need to glue local solutions into a global one by a partition of unity, and this should be done separately on every induction step from the proof of Lemma 11.4. Moreover, the result holds in an even more general setting, where Rn is replaced by any general Riemannian manifold (for example, any open set in Rn ) and Ī by a closed submanifold.
11.3. Jumps of potentials Now we are able to prove the main theorem on jumps of potentials. Theorem 11.7. 1) Let u be one of the potentials (11.2)ā(11.4). Then u is smooth up to Ī from each side. 2) If u is a volume potential, then u and u ā C 1 (Rn ).
āu ān ĀÆ
do not have jumps on Ī, so
3) If u is a single layer potential, then u does not have any jump on Ī, so u ā C(Rn ), whereas the jump of āāunĀÆ on Ī is (āĻ), where Ļ is the density of the charge on Ī in the integral (11.3) deļ¬ning u. 4) If u is a double layer potential, then the jump of u on Ī is equal to (āĪ±), where Ī± is the density (of dipole moment) on Ī in the integral (11.4) deļ¬ning u. Here the direction of the jumps is naturally determined by the normal vector ļ¬eld n ĀÆ. Proof. I will ļ¬rst explain the main idea of the proof. As we have seen in the previous section, the diļ¬erentiation (of order 1 or 2) of a function u which is smooth up to Ī from each side leads to another function of the same kind but with additional singular terms: terms containing the surface Ī“-function Ī“Ī and its normal derivative, with coeļ¬cients which are sums of jumps of u and its normal derivative on Ī, taken with coeļ¬cients from C ā (Ī). For our proof, we need, vice versa, to ļ¬nd some jumps by the information from the right-hand side of the equation Īu = āf , which is satisļ¬ed for all potentials with an explicitly given distribution f that is supported on Ī and has exactly the structure that we have described. Now, from the results of the previous section, we can produce a solution u Ė of ĪĖ u = āf1 , where f1 diļ¬ers from f by an arbitrarily smooth function. Then Ī(u ā u Ė) = f1 ā f can be made arbitrarily smooth, and so u ā u Ė can be made arbitrarily smooth too. It will follow that u is smooth up to Ī from each side, and the jumps can be recovered from f .
252
11. Potentials and their computation
Now let us give more detail for each of the statements 1)ā4) above. 1) By Lemma 11.5, for each of the potentials u we may ļ¬nd a compactly supported integrable function uN smooth up to Ī from each side, such that Ī[uN ] + f ā C N (Rn ) (here f is the same as in (11.1)). Therefore, Ī(u ā uN ) = fN ā C N (Rn ). Let us deduce from this that u ā uN ā C N (Rn ). Without loss of generality we may assume that uN and fN have compact supports. But then Ī(u ā uN ā En ā fN ) = 0, which implies that u ā uN ā En ā fN ā C ā (Rn ). At the same time, clearly, En ā fN ā C N (Rn ), because expressing this convolution as an integral
En (y)fN (x ā y)dy, we can diļ¬erentiate, by the dominated convergence theorem, under the integral N times with respect to x. Therefore, u ā uN ā C N (Rn ). This implies that the derivatives ā Ī± u for |Ī±| ā¤ N are continuous up to Ī from each side. Since N is arbitrary, this proves the ļ¬rst statement. 2) Let u be a volume potential. Clearly, u ā C ā (Rn \Ī). In a neighborhood of every point of Ī we can rectify Ī and use the equation Au = āf , obtained by the change of variables from the equation Īu = āf . Since Ī is an elliptic operator, A is also elliptic. Therefore, Ī is noncharacteristic. But then the inclusion f ā L1loc (Rn ) and Lemmas 11.1 and 11.2 imply u ā C 1 (Rn ). (If we assume the opposite, then either u or āāunĀÆ has a nontrivial jump on Ī, and the formulas (11.5)ā(11.8), (11.10), and (11.11) would clearly lead to the occurrence of a nonvanishing singular distribution of type Ļ0 (x ) ā Ī“(xn ) or Ļ0 (x ) ā Ī“ (xn ) in the right-hand side of the equation Au = āf , whereas f is locally integrable due to our assumptions in the case of the volume potential.) 3) In a neighborhood of a point x0 ā Ī it is convenient to use Fermi āu at the boundary points. Let coordinates y = (y , yn ). In particular, āāunĀÆ = āy n u be a single layer potential, deļ¬ned by (11.3), with the density Ļ ā C0ā (Ī). Then the Poisson equation Īu = āĻĪ“Ī implies that u itself cannot have a jump, because any nonvanishing jump Ļ0 of u would lead to the appearance of the term āĻ0 (y )āĪ“ (yn ) in f = āĪu, which would contradict the Poisson equation. Therefore, u is continuous. On the other hand, it is clear from the form of the Laplacian in coordinates y given by (11.21) and from Lemma 11.1 that the jump of āāunĀÆ should be āĻ for u to satisfy the Poisson equation. 4) The same argument works for the double layer potential. Namely, a jump Ļ1 of u leads to the term āānĀÆ (Ļ1 )Ī“Ī in f = āĪu, which implies the desired statement.
11.4. Calculating potentials
253
Remark. In a similar way we may ļ¬nd the jumps of any derivative of the potentials.
11.4. Calculating potentials We will give here simple examples of the calculation of potentials. The explicit calculation becomes possible in these examples with the help of a) symmetry considerations, b) the Poisson equation satisļ¬ed by the potential, c) theorems on jumps, d) asymptotics at inļ¬nity. As a rule, these considerations even give a surplus of information which enables us to verify the result. Example 11.1. The potential of a uniformly charged sphere. Let Ī be the sphere of radius R in R3 with the center at the origin. Suppose a charge is uniformly (with the surface density Ļ0 ) distributed over the sphere. Denote the total charge of the sphere by Q; i.e., Q = 4ĻR2 Ļ0 . Let us ļ¬nd the potential of the sphere and its ļ¬eld. Denote the desired potential by u. First, symmetry considerations show that u is invariant under rotations; hence u(x) = f (r), where r = |x|; i.e., u only depends on the distance to the center of the sphere. Now use the fact that u is a harmonic function outside Ī. Since any spherically symmetric harmonic function is of the form C1 + C2 rā1 , we have A + Brā1 , r < R, f (r) = C + Drā1 , r > R. Here A, B, C, D are real constants to be determined. First, B = 0 since u is harmonic in a neighborhood of 0. This, in particular, implies that u = const for |x| < R and, therefore, E = ā grad u = 0 for |x| < R; i.e., the ļ¬eld vanishes inside a uniformly charged sphere. To ļ¬nd the three remaining constants, we need three more conditions for u. They can be obtained in diļ¬erent ways. Let us indicate the simplest of them. 1ā¦ . The continuity of u(x) on Ī implies A = C + DRā1 . 2ā¦ . Considering the representation of u(x) as an integral, we see that u(x) ā 0 as |x| ā ā; hence C = 0.
254
11. Potentials and their computation
3ā¦ . Consider the integral that deļ¬nes u(x), in somewhat more detail. We have
Ļ0 dSy dSy Ļ0 1 1 = u(x) = y x 4Ļ |y|=R |x ā y| 4Ļ |y|=R |x| | |x| ā |x| |
1 Ļ0 Q 1 = = 1+O dSy 1 + O 4Ļ|x| |y|=R |x| 4Ļ|x| |x| 1 Q 1+O , = 4Ļr r as r ā ā. Hence D = Q . Thus, A = 4ĻR
Q 4Ļ .
From 1ā¦ ā3ā¦ we deduce that B = C = 0, D =
u(x) =
Q 4ĻR Q 4Ļr
Q 4Ļ ,
for |x| < R, for |x| > R.
This means that outside the sphere the potential is the same as if the whole charge were concentrated at the center. Therefore the homogeneous sphere attracts the charges situated out of it as if the whole charge were supported in its center. This fact was ļ¬rst proved by Newton who considered gravitational forces. (He actually proved this theorem in a diļ¬erent way: by considering directly the integral that deļ¬nes the attraction force.) We will indicate other arguments that may be used to get equations on A, C, D. We can use them to verify the result. 4ā¦ . The jump of the normal derivative of u on the sphere should be Q Q 2 equal to āĻ0 ; i.e., Drā2 |r=R = Ļ0 or D = Ļ0 R2 = 4ĻR 2 Ā· R = 4Ļ , which agrees with the value of D found above. 5ā¦ . The value of u(x) at the center of the sphere is equal to
Ļ0 dSy 1 Q Q A= = Ā· Ļ0 4ĻR2 = , 4Ļ |y|=R R 4ĻR 4ĻR which agrees with the value of A found above. Using the obtained result, we can immediately ļ¬nd the attracting force (and the potential) of the uniformly charged ball by summing the potentials (or attracting forces) of the spherical layers constituting the ball. In particular, the ball attracts a charge outside of it as if the total charge of the ball were concentrated at its center. Example 11.2. The potential of the double layer of the sphere. Consider the sphere of radius R over which dipoles with the density of the dipole moment Ī±0 are uniformly distributed. The potential of the double layer here is the integral (11.4), where Ī = {x : x ā R3 , |x| = R},
Ī±(y) = Ī±0 = const .
11.4. Calculating potentials
255
Let us calculate this potential. This can be done in the same way as in the preceding example. We may immediately note, however, that u(x) = 0 for |x| > R since u(x) is the limit of the potentials of simple layers of a pair of concentric spheres uniformly charged by equal charges with opposite signs. In the limit, these spheres tend to each other, and charges grow in inverse proportion to the distance between the spheres. Even before the passage to the limit, the potential of this pair of spheres vanishes outside the largest of them. Therefore, u(x) = 0 for |x| > R. It is also clear that u(x) = const for |x| < R, and the theorem on jumps implies (when the exterior normal is chosen) that u(x) = Ī±0 for |x| < R. Example 11.3. The potential of a uniformly charged plane. Let us deļ¬ne and calculate the potential of a uniformly charged plane with the surface density of the charges Ļ0 . Let x3 = 0 be the equation of the plane in R3 . Observe that we cannot use the deļ¬nition of the potential by (11.3) because the integral diverges. Therefore, it is necessary to introduce ļ¬rst a new deļ¬nition of the potential. This may be done in several ways. We will indicate two of them. The ļ¬rst is to preserve the main equation (11.25)
Īu = āĻ0 Ī“Ī = āĻ0 Ī“(x3 )
and as many symmetry properties of the potential as possible. (Physicists often determine a value of an observable quantity from symmetry considerations without bothering much with the deļ¬nitions.) If the integral (11.3) were convergent in our case, then (11.25) would be satisļ¬ed and the following symmetry conditions would hold: a) u is not changed by translations along our plane; i.e., u = u(x3 ). b) u is invariant under reļ¬ections with respect to our plane; i.e., u(āx3 ) = u(x3 ). Now, try to calculate u(x3 ) using (11.25) and the above symmetry properties. It turns out that u(x3 ) is uniquely deļ¬ned up to a constant. Therefore, we can deļ¬ne the potential of the plane to be a distribution u satisfying (11.25), a), and b). From (11.25) it is clear that u(x3 ) is linear in x3 separately for x3 > 0 and x3 < 0; i.e., A + Bx3 , x3 > 0, u(x3 ) = C + Dx3 , x3 < 0. Equation (11.25) means, besides, that the theorem on jumps holds; i.e., u āu on Ī is equal is continuous on Ī (implying A = C) and the jump of āx 3 to āĻ0 (implying B ā D = āĻ0 ). Finally, the fact that u(x3 ) is even (condition b)) once again yields A = C and, besides, implies D = āB.
256
11. Potentials and their computation
Hence, B = āD = ā Ļ20 and Ļ0 |x3 |. 2 Thus we have found u up to an irrelevant additive constant. u = u(x3 ) = A ā
Now, recall that our main goal is to ļ¬nd E = ā grad u. We see that E is everywhere perpendicular to the plane and if the direction from the plane is taken as the positive one, then the magnitude of the ļ¬eld is equal to Ļ20 . Here is another method to deļ¬ne u. The ļ¬eld E can be found by the Coulomb law (E is given, as is easy to see, by a converging integral!) and then u is found from the equation E = ā grad u. Since the symmetry properties clearly hold and div E = Ļ0 Ī“(x3 ) leads to (11.25), we can apply the same method to ļ¬nd u. Therefore we have explicitly found the ļ¬eld strength E given by a relatively complicated integral.
11.5. Problems 11.1. Calculate the potential of a uniformly charged circle on the plane. 11.2. Calculate the potential of a uniformly charged disc on the plane. 11.3. Deļ¬ne and calculate the potential of the uniformly charged surface of a straight cylinder in R3 . 11.4. Find the potential of the double layer of the circle in R2 and the surface of a straight cylinder in R3 with a constant density of the dipole moment. 11.5. Find the potential of a uniformly charged sphere in Rn for n ā„ 3. 11.6. A point charge Q is placed at the distance d from the center of a grounded conducting sphere of radius R (R < d). Find the total charge induced on the sphere. 11.7. Under the conditions of the preceding problem ļ¬nd the distribution of the induced charge over the surface of the sphere. 11.8. A thin elliptic wire is charged with the total charge Q. How will the charge distribute along the wire?
10.1090/gsm/205/12
Chapter 12
Wave fronts and short-wave asymptotics for hyperbolic equations
12.1. Characteristics as surfaces of jumps Let Ī© be an open set in Rn , S a smooth hypersurface in Ī©, i.e., a closed submanifold of Ī© of codimension 1. Let a linear diļ¬erential operator of degree m (see Chapter 1) (12.1)
A = a(x, D) =
aĪ± (x)D Ī±
|Ī±|ā¤m
be given in Ī©, with aĪ± ā C ā (Ī©), and let a(x, Ī¾) be the total symbol of A (we usually consider (12.1) to be the deļ¬nition of both A and a(x, D)). Recall that (12.2) a(x, Ī¾) = aĪ± (x)Ī¾ Ī± , |Ī±|ā¤m
so that a(x, D) is obtained from a(x, Ī¾) if all the Ī¾ Ī± are written on the right (as in (12.2)) and then replaced by D Ī± . The principal symbol am (x, Ī¾) of A is of the form aĪ± (x)Ī¾ Ī± . am (x, Ī¾) = |Ī±|=m
257
258
12. Wave fronts and short-wave asymptotics
Recall that a hypersurface S is called a characteristic of A if for any point x ā S the unit normal n ĀÆ x to S at x satisļ¬es (12.3)
am (x, n ĀÆ x ) = 0.
Unfortunately, deļ¬nitions of characteristic are ambiguous: there are many (not necessarily equivalent) versions. However, there are many important common features, which allow us to create a common use of them. Let us take now a real-valued function u ā C ā (Ī©\S) which is inļ¬nitely diļ¬erentiable up to S from each side (cf. the deļ¬nition in Section 11.2). Using the same notation as in Section 11.2, we locally have Ī© = Ī©+ āŖ S āŖ Ī©ā , and there exist u+ ā C ā (Ī©+ ) and uā ā C ā (Ī©ā ) such that u|Ī©+ = u+ , u|Ī©ā = uā . We can consider u as a locally integrable function on Ī©, and we will denote by u the corresponding distribution too. Note immediately that it may actually turn out that there is no discontinuity; i.e., u can be extended to a function from C ā (Ī©). We will say that S is a surface of jumps for u if, ļ¬rst, u is inļ¬nitely diļ¬erentiable up to S from each side and, secondly, for each point x0 ā S there exists a multi-index Ī± such that ā Ī± u is discontinuous at x0 ; i.e., ā Ī± u does not extend to a function on Ī© continuous at x0 . This actually means that (ā Ī± u+ )(x0 ) = (ā Ī± uā )(x0 ). Note that the same inequality also holds on S for x close to x0 (with the same Ī±). Theorem 12.1. If S is a surface of jumps for u ā L1loc (Ī©) and u is a solution of the equation Au = f in Ī©, with f ā C ā (Ī©), then S is a characteristic of A. Proof. Since the theorem is local and the notion of a characteristic is invariant under diļ¬eomorphisms, we may assume that S is of the form S = {x : x ā Ī©, xn = 0}, where x = (x , xn ) = (x1 , . . . , xnā1 , xn ). Since the normal to S is of the form (0, 0, . . . , 1), the characteristic property of S means that the coeļ¬cient of ām āxm in A vanishes on S. n
We will prove the theorem arguing by contradiction. Let S be noncharām acteristic at some point. This means that the coeļ¬cient of āx m is nonzero n at this point, hence, near it. Then we may assume that it is nonzero everywhere in Ī©. Dividing the equation Au = f by this coeļ¬cient, we may
12.1. Characteristics as surfaces of jumps
259
assume that the equation is of the form āmu ā māj u + aj (x, D ) māj = f, m āxn āxn m
(12.4)
j=1
where aj (x, D ) is a diļ¬erential operator in Ī© without any derivatives with respect to xn (its order is ā¤ j though this is of no importance now). Let u be a C ā -function up to S from each side, let Ī©Ā± = Ī© ā© {x : Ā±xn > 0}, and let uĀ± ā C ā (Ī©Ā± ) be such that u|Ī©Ā± = uĀ± . We want to prove that u can be extended to a function from C ā (Ī©); i.e., the jumps of all the derivatives actually vanish. Denote these jumps by ĻĪ± ; i.e., ĻĪ± = (ā Ī± u+ )|S ā (ā Ī± uā )|S ā C ā (S). As we have seen in Section 11.2, a special role is played by the jumps of the normal derivatives ā k u+ ā k uā ā ā C ā (S). (12.5) Ļk = āxkn S āxkn S All the jumps ĻĪ± are expressed in terms of the jumps of the normal derivatives Ļk via the already-mentioned formula
ĻĪ± = āxĪ± ĻĪ±n ,
Ī± = (Ī± , Ī±n ).
Therefore, it suļ¬ces to prove that Ļk = 0 for k = 0, 1, 2, . . . . Let us ļ¬nd Au in D (Ī©). By Lemma 11.1 we have ' & āu āu , = Ļ0 (x ) ā Ī“(xn ) + āxn āxn
& 2 ' ā 2u ā u . = Ļ0 (x ) ā Ī“ (xn ) + Ļ1 (x ) ā Ī“(xn ) + 2 āxn āx2n
Further calculations of the same type easily yield (12.6)
& k ' kā1 ā u āku (l) = Ļkālā1 (x ) ā Ī“ (xn ) + . k āxn āxkn l=0
Now, if Ī± is an (n ā 1)-dimensional multi-index and ā Ī± = āxĪ± , then (12.7)
& ' kā1 k ā k u Ī± (l) Ī± ā u = [ā Ļkālā1 (x )] ā Ī“ (xn ) + ā . ā āxkn āxkn Ī±
l=0
260
12. Wave fronts and short-wave asymptotics
Now, let a ā C ā (Ī©). Using the same arguments as in the proof of Lemma 11.2, we easily obtain a(x , xn )(Ļ(x ) ā Ī“ (l) (xn )) =
(12.8)
l
aq (x ) ā Ī“ (q) (xn ),
q=0
where aq ā C ā (S). Formulas (12.6)ā(12.8) show that if Au = f outside S, then (12.4) may be rewritten in the form Ļ0 (x ) ā Ī“ (mā1) (xn ) +
(12.9)
mā1
al (x ) ā Ī“ (mā1āl) (xn ) = g(x),
l=1
(x )
C ā (S),
al ā and Ļ0 is the jump of u on S (see formula where g ā (12.5) for k = 0). It is clear from (12.9) that g ā” 0 since the left-hand side of (12.9) is supported on S. Further, after applying the distribution at the in left-hand side of (12.9) to a test function Ļ that is equal to Ļ1 (x )xmā1 n ā a neighborhood of S (here Ļ1 ā C0 (S)), we see that all the terms but the ļ¬rst vanish, and the ļ¬rst one is equal to (ā1)mā1 (m ā 1)! Ļ0 (x )Ļ1 (x )dx . Since Ļ1 is arbitrary, it is clear that Ļ0 (x ) ā” 0; i.e., u cannot have a jump on S. L1loc (Ī©),
One similarly proves that the derivatives up to order m ā 1 cannot have jumps on S. Namely, let p be the smallest of all k such that Ļk ā” 0. Therefore, Ļ0 ā” Ļ1 ā” Ā· Ā· Ā· ā” Ļpā1 ā” 0,
(12.10)
Ļp ā” 0.
We wish to distinguish the most singular term in the left-hand side of (12.4). This is easy to do starting from the formulas (12.6)ā(12.8). Namely, calcuāk u 1 lating the normal derivatives āx k for k ā¤ p, we get functions from Lloc (Ī©). n For k = p + 1 we get & p+1 ' ā u ā p+1 u p+1 = Ļp (x ) ā Ī“(xn ) + āxn āxp+1 n and for p ā¤ m ā 1 equation (12.4) takes on the form
Ļp (x ) ā Ī“
(māpā1)
(xn ) +
māpā1
al (x ) ā Ī“ (māpā1āl) (x ) = g(x),
l=1
where al (x ) ā C ā (S), g ā L1loc (Ī©). As above, this implies Ļp ā” 0, contradicting the hypothesis. Thus, Ļp = 0 for p ā¤ m ā 1. Let us prove that Ļm = 0. This is immediately clear from (12.4) since all the derivatives ā Ī± u, where Ī± = (Ī± , Ī±n ) and Ī±n ā¤ m ā 1, are continuous in Ī©. Therefore, the normal derivative of order m has no jumps either.
12.1. Characteristics as surfaces of jumps
261
Now, consider the case when p > m, where p is the number of the ļ¬rst discontinuous normal derivative; see (12.10). This case can be easily reduced pām to the case p = m. Indeed, applying ā pām to (12.4), we see that u satisļ¬es āxn an equation of the same form as (12.4) but of order p instead of m. This implies that the case p > m is impossible. This concludes the proof. Theorem 12.1 implies that solutions of elliptic equations with smooth (C ā ) coeļ¬cients and smooth right-hand side cannot have jumps of the structure described above. It can even be proved that they are always C ā functions. (The reader can ļ¬nd proofs in more advanced textbooks; see, for example, Taylor [35, vol. II, Sect. 7.4], HĀØormander [13, Ch. X], or [15, vol. I, Corollary 8.3.2].) The examples given below show that nonelliptic equations may have discontinuous solutions. Example 12.1. The equation utt = a2 uxx has a solution Īø(x ā at) with a jump along the characteristic line x = at. A jump of derivatives of arbitrarily large order along the characteristic x = at can be arranged by considering the solution u(t, x) = (x ā at)N Īø(x ā at), where N ā Z+ . Similar discontinuities can obviously be constructed along any characteristic (all the characteristics here can be written in one of the forms x ā at = c1 , x + at = c2 ). Example 12.2. Let us consider a more intricate case of the three-dimensional wave equation utt = a2 Īu,
u = u(t, x),
x ā R3 .
The simplest example of a discontinuous solution of this equation is the spherical wave u(t, x) = Īø(|x|āat) . This solution, clearly, has a jump on |x| the upper sheet of the light cone {(t, x) : |x| = at, t > 0}. The sphere {x : |x| = at} ā R3 is naturally called the wave front. The wave front moves at a speed a. In geometrical optics, a ray is a parametrically deļ¬ned (with the parameter t) curve which is normal to the wave front at any moment of time. In our example the rays are straight lines x = ae t, where t > 0, e ā R3 , |e| = 1. Sometimes any curve normal to all the wave fronts (without an explicitly given parameter) is also called a ray. In what follows we will introduce more terminology of the geometric optics in connection with the Hamilton-Jacobi equation. Example 12.3. Consider again the three-dimensional wave equation utt = a2 Īu, but this time try to construct a solution which is discontinuous on a given smooth compact surface Ī ā R3 at the initial moment. Let, for
262
12. Wave fronts and short-wave asymptotics
instance, Ī = āĪ©, where Ī© is a bounded domain in R3 and the initial conditions are of the form u|t=0 = ĻĪ© ,
ut |t=0 = 0,
where ĻĪ© is the characteristic function of Ī©. Let us try to ļ¬nd out how the jumps of u behave as t grows by looking at the Kirchhoļ¬ formula which we will use to construct u. In the Kirchhoļ¬ formula we are to integrate ĻĪ© over the sphere of radius at with the center at x, divide the result by 4Ļa2 t, and then take the derivative with respect to t. It is easy to see that this u(t, x) is a smooth function of t, x for t, x such that the sphere of radius at with the center at x is not tangent to Ī (i.e., this sphere either does not intersect Ī or intersects it transversally). Therefore, the jumps may occur only at the points where this sphere is tangent to Ī. For small t, the set of all such x forms exactly the set of the points x which are situated at the distance at from Ī (āwave frontā). For small t, the wave front can be constructed as follows: draw all the normals to Ī (āraysā) and mark on them points at the distance at on both sides. It is easy to see that on the wave front thus constructed, there is actually a jump of u. For large t normals may begin to intersect. An envelope of the family of rays appears and is called a caustic. The singularity of the solution on the caustic has a much more sophisticated structure and we will not deal with it. Even this example shows how objects of geometric optics appear in the wave theory. In what follows we will consider this problem in somewhat more detail.
12.2. The Hamilton-Jacobi equation. Wave fronts, bicharacteristics, and rays The Hamilton-Jacobi equation is an equation of the form āS (12.11) H x, = 0, āx where H = H(x, Ī¾) is a given real-valued function of variables x ā Rn and Ī¾ ā Rn (we will usually assume that H ā C 2 ) and S = S(x) is an unknown (also real-valued) function with the gradient āS āx . The HamiltonJacobi equation plays an important role in mechanics and physics. Recall brieļ¬y a method for integrating (12.11), more precisely, a connection of this equation with the Hamiltonian system xĖ = HĪ¾ (x, Ī¾), (12.12) Ī¾Ė = āHx (x, Ī¾),
12.2. The Hamilton-Jacobi equation
263
where the Hamiltonian H = H(x, Ī¾) is the same function as in (12.11), HĪ¾ = āH āH āH āH āH āH āĪ¾ = ( āĪ¾1 , . . . , āĪ¾n ), Hx = āx = ( āx1 , . . . , āxn ). (We already discussed Hamiltonian systems in Section 10.3; here we use x, Ī¾ instead of q, p in (10.13).) Solutions of this system (curves (x(t), Ī¾(t))) are called the Hamiltonian curves (or, more exactly, Hamiltonian curves of the Hamiltonian H(x, Ī¾)). Vector ļ¬eld (HĪ¾ , āHx ) in R2n x,Ī¾ deļ¬ning the system (12.12) is called a Hamiltonian vector ļ¬eld with the Hamiltonian H. As was mentioned in Section 10.3, any Hamiltonian vector ļ¬eld is welldeļ¬ned on T ā Rnx . This means that the system (12.12) retains the same form for any choice of curvilinear coordinates in Rnx assuming that Ī¾ is a cotangent vector expressed in the coordinate system corresponding to a given coordinate system in Rnx (i.e., considering Ī¾1 , . . . , Ī¾n as coordinates of a cotangent vector with respect to the basis dx1 , . . . , dxn ; see Section 1.1). The reader can ļ¬nd more detail on this in Arnold [3]. Proposition 12.2. The Hamiltonian H(x, Ī¾) is a ļ¬rst integral of the system (12.12); i.e., H(x(t), Ī¾(t)) = const along any Hamiltonian curve (x(t), Ī¾(t)). Proof. We have āH āH Ė d H(x(t), Ī¾(t)) = xĖ + Ī¾ = Hx HĪ¾ + HĪ¾ (āHx ) = 0. dt āx āĪ¾ (Here and below for the sake of simplicity we omit the dot standing for the inner product of two vectors.) This immediately implies the desired result. In classical mechanics, Proposition 12.2 has the meaning of the energy conservation law. In fact, we already proved it for particular inļ¬nitedimensional linear Hamiltonian system, which is equivalent to the wave equation (see Proposition 10.1). The Hamiltonian curves with H(x(t), Ī¾(t)) ā” 0 are of particular importance for us. If H is the principal symbol of a diļ¬erential operator, then they are called bicharacteristics (of the operator or of its symbol). By some abuse of terminology, we will call such a curve bicharacteristic for any H (even if H is not a principal symbol of any diļ¬erential operator). As Proposition 12.2 says, if (x(t), Ī¾(t)) is a Hamiltonian curve deļ¬ned on an interval of the t-axis and H(x(t0 ), Ī¾(t0 )) = 0 for some t0 , then this curve is a bicharacteristic. Now, consider the graph of the gradient of S(x), i.e., the n-dimensional āS n submanifold in R2n x,Ī¾ of the form ĪS = {(x, Sx (x)), x ā R }, where Sx = āx . Let us formulate the main statement necessary to integrate the HamiltonianJacobi equation.
264
12. Wave fronts and short-wave asymptotics
Proposition 12.3. If S is a C 2 -solution of the Hamilton-Jacobi equation (12.11), then the Hamiltonian ļ¬eld is tangent to ĪS at its points. Proof. The statement means that if (x(t), Ī¾(t)) is a bicharacteristic and (x0 , Ī¾0 ) = (x(t0 ), Ī¾(t0 )) ā ĪS , then d d (x(t), Sx (x(t)))|t=t0 = (x(t), Ī¾(t))|t=t0 dt dt or, in a shorter form, d Ė 0 ). Sx (x(t))|t=t0 = Ī¾(t dt But d Sx (x(t))|t=t0 = Sxx (x(t0 ))x(t Ė 0 ) = Sxx (x0 )HĪ¾ (x0 , Ī¾0 ), dt 2
S where Sxx = āxāi āx n is considered as an n Ć n symmetric matrix. On j i,j=1 Ė 0 ) = āHx (x0 , Ī¾0 ), so we need to prove that the other hand, Ī¾(t
(Sxx HĪ¾ + Hx )|ĪS = 0. But this immediately follows by diļ¬erentiation of the Hamilton-Jacobi equation (12.11) with respect to x. Proposition 12.3 directly implies Theorem 12.4. ĪS is invariant with respect to the Hamiltonian ļ¬ow (i.e., with respect to the motion along the bicharacteristics). This means that if (x(t), Ī¾(t)) is a bicharacteristic deļ¬ned for t ā (a, b) and Ī¾(t0 ) = Sx (x(t0 )) for some t0 ā (a, b), then Sx (x(t)) = Ī¾(t) for all t ā (a, b). Using Theorem 12.4, we can construct a manifold ĪS if it is known over a submanifold M of codimension 1 in Rnx . If we also know S over M , then S is recovered by simple integration over ĪS (note that if ĪS is found, then S is recovered from a value at a single point). It should be noted, however, that all these problems are easy to study locally whereas we are rarely successful in recovering S globally. We more often succeed in extending ĪS to a manifold which is not the graph of a gradient any longer but is a so-called Lagrangian manifold; this often suļ¬ces to solve the majority of problems of mathematical physics (the ambiguity of projecting a Lagrangian manifold onto Rnx causes the appearance of caustics). The detailed analysis of these questions, however, is beyond the scope of this textbook. Let us indicate how to construct (locally) S if it, itself, and its gradient are known over an initial manifold M of codimension 1 in Rnx . If the
12.2. The Hamilton-Jacobi equation
265
bicharacteristic L = {(x(t), Ī¾(t))} begins at (x0 , Ī¾0 ), where x0 ā M and Ī¾0 ā Sx (x0 ), and terminates at (x1 , Ī¾1 ), then, clearly,
(12.13) S(x1 ) ā S(x0 ) = Sx dx = Ī¾dx. L
L
The projections of bicharacteristics to Rnx are called rays. We will only consider bicharacteristics starting over M at the points of the form (x, Sx (x)), where x ā M . Take two rays x1 (t) and x2 (t) corresponding to these bicharacteristics. They may intersect at some point x3 ā Rnx and then (12.13) gives two (generally, distinct) values for S(x) at x3 . For this reason the Cauchy problem for the Hamilton-Jacobi equation (ļ¬nding S from the values of S|M and Sx |M ) is rarely globally solvable. The Cauchy problem is, however, often solvable locally. For its local solvability it suļ¬ces, for example, that rays of the described type that start at the points from M be transversal to M at the initial moment. Then we can consider the map f : M Ć (āĪµ, Īµ) āā Rnx , which to a pair {x0 ā M, t ā (āĪµ, Īµ)} assigns a point x(t) of the ray starting at x0 (i.e., x(0) = x0 ) which is the projection of the bicharacteristic (x(t), Ī¾(t)) such that Ī¾(0) = Sx (x0 ). (We assume that this map is well-deļ¬ned; i.e., the corresponding bicharacteristics are deļ¬ned for t ā (āĪµ, Īµ).) By the implicit function theorem, f determines a diļ¬eomorphism of U Ć (āĪµ, Īµ) (where U is a small neighborhood of x0 in M and Īµ is suļ¬ciently small) with an Ė is not tangent to M (this is a necessary open set in Rnx if and only if x(0) and suļ¬cient condition for the derivative map of f to be an isomorphism (Tx0 M Ć R āā Tx0 Rnx ). In this case locally the rays do not intersect and the Cauchy problem is locally solvable. Let us give an exact formulation of the Cauchy problem. Given a submanifold M ā Rnx and S0 ā C ā (M ), let ĪS |M be a submanifold in (T ā Rnx )|M which projects to the graph of the gradient of S0 under the natural projection onto T ā M and belongs to the zero level surface of H(x, Ī¾). The Cauchy problem is then to ļ¬nd S such that S|M = S0 and ĪS |M coincides with the submanifold thus denoted above and given over M . If the above transversality condition holds at all points of M , the Cauchy problem has a solution in a neighborhood of M . The level surfaces of S are usually called wave fronts. We will see below why the use of this term here agrees with that in Example 12.2. Example 12.4. Consider a Hamilton-Jacobi equation of the form 1 (12.14) |Sx (x)|2 = 2 . a The Hamiltonian in this case is 1 H(x, Ī¾) = |Ī¾|2 ā 2 a
266
12. Wave fronts and short-wave asymptotics
and the Hamiltonian equations are Ī¾Ė = 0, xĖ = 2Ī¾, which implies that the Hamiltonian curves are given by Ī¾ = Ī¾0 , x = 2Ī¾0 t + x0 . In particular, the rays are straight lines in all directions. If such a ray is a projection of a bicharacteristic belonging to the graph ĪS of the gradient of the solution S(x) (due to Theorem 12.4 this means that Sx (x0 ) = Ī¾0 ), then xĖ Sx (x(t)) = Ī¾(t) = Ī¾0 = , 2 along the whole ray, and the ray x(t) is orthogonal to all the wave fronts S(x) = const which it intersects. Note that in general rays are not necessarily orthogonal to wave fronts. The function S(x) = aā1 |x| in Rn \0 is one of the solutions of (12.14). Its level lines, i.e., wave fronts, are spheres {x : S(x) = t}, and they coincide with the wave fronts of Example 12.2. The rays in this case are straight lines which also coincide with the rays, which we mentioned in Example 12.2. Now consider the following particular case of the Hamilton-Jacobi equation: āS āS + A x, = 0, (12.15) āt āx where S = S(t, x), A = A(x, Ī¾), t ā R1 , x, Ī¾ ā Rn , A ā C 2 is assumed to be real valued. This is a Hamilton-Jacobi equation in Rn+1 t,x with the Hamiltonian H(t, x, Ļ, Ī¾) = Ļ + A(x, Ī¾). The Hamiltonian system that deļ¬nes the Hamiltonian curves (t(s), x(s), Ļ (s), Ī¾(s)) is ā§ āŖ tĖ = 1, āŖ āŖ āŖ āØxĖ = A (x, Ī¾), Ī¾
āŖ ĻĖ = 0, āŖ āŖ āŖ ā©Ī¾Ė = āA (x, Ī¾) x (the ādotā stands for the derivative with respect to the parameter s). It follows that Ļ (s) = Ļ0 , t(s) = s + t0 , and (x(s), Ī¾(s)) is a Hamiltonian curve of the Hamiltonian A(x, Ī¾). Since a shift of the parameter of a Hamiltonian curve does not aļ¬ect anything, we may assume that t0 = 0, i.e., t = s,
12.2. The Hamilton-Jacobi equation
267
and regard x and Ī¾ as functions of t. Further, when interested in bicharacteristics, we may arbitrarily set x(0) = x0 , Ī¾(0) = Ī¾0 , and then Ļ0 is uniquely determined: Ļ0 = āA(x0 , Ī¾0 ). Therefore arbitrary Hamiltonian curves (x(t), Ī¾(t)) of the Hamiltonian A(x, Ī¾) are in one-to-one correspondence with bicharacteristics for H(t, x, Ļ, Ī¾) starting at t = 0. The rays corresponding to the bicharacteristics described above are of the form (t, x(t)). In particular, they are transversal to all the planes t = const. We may set the Cauchy problem for the equation (12.15) imposing the initial condition (12.16)
S|t=0 = S0 (x).
āS 0 This implies Sx |t=0 = āS āx , and from (12.15) we ļ¬nd āt t=0 . Therefore, the graph ĪS of the gradient of S is known over the plane M = {(t, x) : t = 0}. Since the rays are transversal to M at the initial moment, ĪS may be extended to deļ¬ne ļ¬rst ĪS and then S over a neighborhood of M . Namely, in accordance with (12.13), we obtain
t [Ļ0 dt + Ī¾(t)x(t)dt], Ė S(t, x) = S0 (x0 ) + 0
where (x(t), Ī¾(t)) is a Hamiltonian curve of the Hamiltonian A(x, Ī¾) such that x(0) = x0 and x(t) = x; i.e., the points x0 and x are connected by a ray (clearly, x0 should be chosen so that the ray with x0 as the source hits x at the time t); besides, āS0 (x0 ) = āA(x(t), Ī¾(t)) Ļ0 = āA(x0 , Ī¾0 ) = āA x0 , āx (this means that the corresponding bicharacteristic of H(t, x, Ļ, Ī¾) belongs to ĪS ). Thus,
x (12.17) S(t, x) = S0 (x0 ) + (Ī¾dx ā Adt), x0
where the integral is taken over a Hamiltonian curve of the Hamiltonian A such that its projection joins x0 and x. Instead of (12.17) we can also write
x Ldt, (12.18) S(t, x) = S0 (x0 ) + x0
where L = Ī¾ xĖ ā A is called the Lagrangian corresponding to the Hamiltonian A and the integral is taken again over the curve. Formulas (12.17) and (12.18) show that S is similar to the action of classical mechanics (Ī¾ ā Rn plays the role of momentum). Therefore, the wave fronts are level surfaces of the action.
268
12. Wave fronts and short-wave asymptotics
We will not go any deeper into the Hamilton-Jacobi equation theory and its relations with mechanics and geometric optics. Details can be found in textbooks on mechanics (see, e.g., Arnold [3]). Note, in particular, that we have proved here the uniqueness of a solution for the Cauchy problem and gave a method of its construction but did not verify its existence, i.e., we did not verify that this method actually produces a solution. This is really so in the domain where the above system of rays starting at t = 0 is diļ¬eomorphic to the above system of parallel segments (in particular, this is true in a small neighborhood of the plane {(t, x) : t = 0}), but we skip the veriļ¬cation of this simple fact. Let us give one further remark on the Hamilton-Jacobi equation of the form (12.15). Suppose we are interested not in the solution itself but only in the wave fronts {(t, x) : S(t, x) = c}, where c is a constant. Fix one value of the constant and consider the system of surfaces {x : S(t, x) = c} belonging to Rnx and depending on t as on a parameter. If āS āt = 0 somewhere, then we may (locally) solve the equation S(t, x) = c for t and express this system of Ė Ė surfaces in the form {x : S(x) = t}, i.e., in the form of level surfaces of S. Ė How do we write an equation for S? Ė S(S(x), x) ā” c with respect to x, we get
Diļ¬erentiating the identity
āS Ė Ā· Sx + Sx = 0. āt Substituting Sx = āSĖx Ā· (12.19)
āS āt
into (12.15), we get āS āS Ė + A x, āSx (x) = 0. āt āt
Consider the case most often encountered and the most important for partial diļ¬erential equations: the case when A(x, Ī¾) is homogeneous of ļ¬rst order in Ī¾; i.e., A(x, Ī»Ī¾) = Ī»A(x, Ī¾), x ā Rn , Ī¾ ā Rn , Ī» > 0. We may assume that āS āt < 0 (otherwise, replace S by āS, which does not aļ¬ect the wave fronts). Dividing (12.19) by (ā āS āt ) we get (12.20)
A(x, SĖx (x)) = 1.
We could get the same result just assuming that Ė (12.21) S(t, x) = S(x) ā t + c. Then (12.20) is immediately obtained without requiring A(x, Ī¾) to be homogeneous. The role of homogeneity is in the fact that under the assumption of homogeneity, (12.15) is actually just an equation for the direction of the normal vector to the wave fronts and does not depend in any way on parameterization of the wave fronts and the length of the normal vector.
12.3. The characteristics of hyperbolic equations
269
Thus, assuming that S(t, x) is of the form (12.21) or A(x, Ī¾) is homogeneous of order 1 in Ī¾, we see that the wave fronts are determined by Ė Ė the equation S(x) = t, where S(x) is a solution of the time-independent Hamilton-Jacobi equation (12.20).
12.3. The characteristics of hyperbolic equations We have not yet discussed the existence of characteristics. Clearly, the elliptic operator has no characteristics. We will see that a hyperbolic operator has plenty of characteristics. Let A = a(t, x, Dt , Dx ) =
aĪ± (t, x)DtĪ±0 DxĪ± ,
|Ī±|ā¤m
where t ā R, x ā Rn , Ī± = (Ī±0 , Ī± ) is an (n + 1)-dimensional multi-index, aĪ± (t, x) ā C ā (Ī©), with Ī© a domain in Rn+1 . Let us recall what the hyperbolicity of A with respect to a distinguished variable t means (see Section 1.1). Consider the principal symbol aĪ± (t, x)Ļ Ī±0 Ī¾ Ī± . (12.22) am (t, x, Ļ, Ī¾) = |Ī±|=m
The hyperbolicity of A means that the equation (12.23)
am (t, x, Ļ, Ī¾) = 0,
regarded as an equation with respect to Ļ , has exactly m real distinct roots for any (t, x) ā Ī©, Ī¾ ā Rn \0. Denote the roots by Ļ1 (t, x, Ī¾), . . . , Ļm (t, x, Ī¾). Clearly, am is a polynomial of degree m in Ļ and the hyperbolicity implies that its highest coeļ¬cient a(m,0,...,0) (the coeļ¬cient by Ļ m ) does not vanish for (t, x) ā Ī© (this coeļ¬cient does not depend on Ī¾ due to (12.22). If we are only interested in the equation Au = f , then we can divide by this coeļ¬cient to make the highest coeļ¬cient identically 1. Then all coeļ¬cients of the principal symbol am will become real. So we can (and will) assume that a(m,0,...,0) ā” 1 from the very beginning. Since the roots of am with respect to Ļ are simple, we have āam
= 0. āĻ Ļ =Ļj By the implicit function theorem, we can solve (12.23) for Ļ in a neighborhood of Ļj and obtain Ļj (t, x, Ī¾) as a locally smooth function of (t, x, Ī¾) ā Ī© Ć (Rn \0), i.e., in a suļ¬ciently small neighborhood of any ļ¬xed point (t0 , x0 , Ī¾0 ) ā Ī© Ć (Rn \0).
270
12. Wave fronts and short-wave asymptotics
The last statement can be made global due to hyperbolicity. Indeed, we can enumerate the roots so that Ļ1 (t, x, Ī¾) < Ļ2 (t, x, Ī¾) < Ā· Ā· Ā· < Ļm (t, x, Ī¾). Then every Ļj will be a continuous (hence, smooth) function of t, x, Ī¾ on Ī© Ć (Rn \0) because when we continuously move t, x, Ī¾ (keeping Ī¾ = 0), the roots also move continuously and cannot collide (a collision would contradict hyperbolicity); hence their order will be preserved for any such motion. Further, by homogeneity of am , i.e., since am (t, x, sĻ, sĪ¾) = sm am (t, x, Ļ, Ī¾), it is clear that the set of roots Ļ1 (t, x, sĪ¾), . . . , Ļm (t, x, sĪ¾) coincides with the set sĻ1 (t, x, Ī¾), . . . , sĻm (t, x, Ī¾). Since the order of the roots Ļj is preserved under the homothety, we see that the functions Ļj (t, x, Ī¾) are smooth, real valued, and homogeneous of order 1 in Ī¾: Ļj (t, x, sĪ¾) = sĻj (t, x, Ī¾), s > 0. In particular, if we know them for |Ī¾| = 1, then they can be extended on all values of Ī¾ = 0. Suppose the surface {(t, x) : S(t, x) = 0} is a characteristic. Then the vector (Ļ, Ī¾) = ( āS āt , Sx ) on this surface satisļ¬es (12.23). But then it is clear that for some j we locally have āS āS (12.24) = Ļj t, x, āt āx (again on the surface S = 0). In particular, for S we may take any solution of the Hamilton-Jacobi equation (12.24). We only need for the surface S = 0 to be nonsingular; i.e., S = 0 should imply grad S = 0. Such a function S may be locally obtained, for example, by solving the Cauchy problem for the equation (12.24) with the initial condition S|t=t0 = S0 (x), where gradx S0 (x) = 0. For instance, we may construct a characteristic passing through (t0 , x0 ) taking S0 (x) = Ī¾ Ā· (x ā x0 ), where Ī¾ ā Rn \0. In fact, taking S0 (x) to be arbitrary, so that grad S0 (x) = 0 on {x : S0 (x) = 0}, and choosing turn by turn all Ļj , j = 1, . . . , m, we will ļ¬nd m diļ¬erent characteristics, whose intersection with the plane t = t0 is the same nonsingular hypersurface in this plane. This means that the wave front, preassigned for t = t0 , may propagate in m diļ¬erent directions. We may verify that the obtained characteristics, whose intersection with the plane t = t0 is the given nonsingular hypersurface, do not actually depend on the arbitrariness in the choice of the initial function S0 if the
12.4. The eikonal and transport equations
271
initial surface {x : S0 (x) = 0} is ļ¬xed. We skip a simple veriļ¬cation of this fact; it follows, for example, from the study of the above method of constructing solutions for the Hamilton-Jacobi equation. For n = 1 such a veriļ¬cation is performed in Chapter 1, where we proved that in R2 exactly two characteristics of any second-order hyperbolic equation pass through each point.
12.4. Rapidly oscillating solutions. The eikonal equation and the transport equations Now we wish to understand how the wave optics turns into the geometric optics for very short waves (or, equivalently, for high frequencies). Recall that the plane wave with frequency Ļ and wave vector k is of the form k u(t, x) = ei(ĻtākĀ·x) = eiĻ(tā Ļ Ā·x) . The length of the vector Ļk is equal to a1 , where a is the speed of the wave propagation. For the three-dimensional wave equation, a is a constant, i.e., it does not depend on Ļ, whereas Ļ may be viewed as arbitrary, in particular, arbitrarily large. Therefore, for any vector k0 of length a1 and any Ļ there exists a plane wave of the form (12.25)
u(t, x) = eiĻ(tāk0 Ā·x) .
We may assume now k0 to be ļ¬xed and let Ļ tend to +ā. This is the limit transition to geometric optics for the case of a plane wave. The argument of the exponent, Ļ = Ļ(t ā k0 Ā· x), is called the phase of the plane wave (12.25). Let us ļ¬x Ļ and k0 and consider constant phase surfaces Ļ = const, which are called wave fronts. For each t a wave front is the plane {x : k0 Ā· x = t + c} ā R3 . A wave front moves with the speed a in the direction of the vector k0 . Therefore, the straight lines in the direction k0 are naturally called rays. Now consider a general diļ¬erential operator aĪ± (x)D Ī± , x ā Ī© ā Rn , A = a(x, D) = |Ī±|ā¤m
and try to ļ¬nd a solution of the equation Au = 0, by analogy with the plane wave (12.25), in the form (12.26)
u(x) = eiĪ»S(x) ,
where Ī» is a large parameter and S(x) is a real-valued function called phase.
272
12. Wave fronts and short-wave asymptotics
Calculating Au, we see that (12.27)
Au = Ī»m am (x, Sx (x))eiĪ»S(x) + O(Ī»mā1 ),
where am (x, Ī¾) is the principal symbol of A (see Section 1.2). For the moment we will ignore the terms denoted in (12.27) by O(Ī»mā1 ) (they are of the form of eiĪ»S(x) times a polynomial of Ī» of degree not higher than m ā 1). But it is clear from (12.27) that if we want Au = 0 to be satisļ¬ed for arbitrarily large values of Ī», then S(x) should satisfy the equation āS =0 (12.28) am x, āx which is called the eikonal equation in geometric optics. It is a HamiltonJacobi equation if the principal symbol am (x, Ī¾) is real valued (only this case will be considered in what follows). The equation (12.28) means the fulļ¬llment of Au = 0 in the ļ¬rst approximation. More precisely, if u(x) is of the form (12.26), then (12.28) is equivalent to Au = O(Ī»mā1 )
as Ī» ā ā.
We see that the search for wave fronts (constant phase surfaces) reduces in this case to solving the Hamilton-Jacobi equation. Suppose that the solvability conditions for the Cauchy problem hold for the Hamilton-Jacobi equation (12.28) (this is true, for example, if A is hyperbolic with respect to a variable denoted by t). Then we may construct wave fronts by methods of geometric optics proceeding from their disposition on the initial manifold (at t = 0 in the hyperbolic case). Now let us try to satisfy Au = 0 with better precision. It turns out that it is a good idea to seek u in the form (12.29)
u = eiĪ»S(x) b(x, Ī»),
where b(x, Ī») =
ā
bj (x)Ī»āj
j=0
is a formal power series (in Ī») called the amplitude. Therefore u = u(x, Ī») is also a formal series. Let us explain the meaning of Au. Clearly, (12.30)
A(f (x)eiĪ»S(x) ) = eiĪ»S(x)
m l=0
Ī»māl (Al f ),
12.4. The eikonal and transport equations
273
where the Al are linear diļ¬erential operators. Applying A to the kth term of (12.29) we get (12.31)
A(e
iĪ»S(x)
āk
bk (x)Ī»
)=e
iĪ»S(x)
m
Ī»mālāk (Al bk ).
l=0
Set (12.32)
Au = eiĪ»S(x)
ā
vj (x)Ī»māj ,
j=0
which is the formal series obtained by applying A termwise to (12.29) and grouping together all the terms with the same power of Ī» (as is clear from (12.31), there are only a ļ¬nite number of terms with any given power of Ī»). We call u an asymptotic or rapidly oscillating solution if Au = 0, i.e., if all the coeļ¬cients vj in (12.32) vanish. Let u be an asymptotic solution of the form (12.29). Consider a ļ¬nite segment of the series (12.29): uN (x, Ī») = eiĪ»S(x)
N
bj (x)Ī»āj .
j=0
Then uN is a (regular) function of x depending on Ī» as on a parameter. The series u ā uN begins with Ī»āN ā1 . Therefore, the series A(u ā uN ) begins with Ī»āN ā1+m . But Au = 0; therefore, A(u ā uN ) = āAuN , implying AuN = O(Ī»āN ā1+m ). Thus, if we know an asymptotic solution u(x, Ī»), then its ļ¬nite segments uN (x, Ī») for large Ī» are ānear-solutionsā which become ever more accurate as N increases. Similar arguments are applicable to a nonhomogeneous equation Au = f and to boundary value problems for this equation. Let a boundary value problem for Au = f have a unique solution continuously depending on f (this is often possible to deduce, for example, from energy estimates). Then, having found uN such that AuN = fN diļ¬ers but little from f (for large Ī»), we see that since A(u ā uN ) = f ā fN , the approximate solution diļ¬ers only a bit from the exact solution u (we assume that uN satisļ¬es the same boundary conditions as u). The āexact solutionā, however, does not have greater physical meaning than the approximate one since in physics the equations themselves are usually approximate. Therefore, in the sequel we will speak about approximate solutions and not bother with āexactā one. Note, however, that an approximate solution satisļ¬es the equation with good accuracy only for large Ī» (for āhigh frequenciesā of āin short-wave rangeā).
274
12. Wave fronts and short-wave asymptotics
How do we ļ¬nd asymptotic solutions? Note ļ¬rst that we may assume b0 (x) ā” 0 (if b0 (x) ā” 0, then we could factor out Ī» and reenumerate the coeļ¬cients bj ). It may happen that b0 (x) ā” 0 but b0 |U ā” 0 for some open subset U ā Rn . In this case we should consider U and its complement separately. With this remark in mind we will assume that b0 |U ā” 0 for any U ā Ī©. But then the highest term in the series Au is (12.33)
Ī»m am (x, Sx (x))b0 (x)eiĪ»S(x) .
If we wish it to be 0, then the phase S(x) should satisfy the Hamilton-Jacobi equation (12.28) and if this equation is satisļ¬ed, then (12.33) vanishes whatever b0 (x). Thus, let S(x) be a solution of the Hamilton-Jacobi equation. Now, let us try to ļ¬nd bj (x) so that the succeeding terms would vanish. For this, let us write the coeļ¬cients vj in the series Au (formula (12.32)) more explicitly. In the above notation, we clearly have (Al bk ), j = 0, 1, 2, . . . , (12.34) vj (x) = l+k =j 0ā¤lā¤m k = 0, 1, 2, . . .
where the Al are linear diļ¬erential operators depending on S and deļ¬ned by (12.30). The sum in (12.34) runs over l ā {0, 1, . . . , m} and integers k ā„ 0 such that l + k = j; therefore, the sum is ļ¬nite. For j = 0 the sum contains one summand: A0 b0 = am (x, Sx (x))b0 (x). This makes it clear that A0 = 0 due to the choice of S(x). Therefore, the sum over l in (12.34) actually runs from 1 to m. We have v1 = A1 b0 , v2 = A1 b1 + A2 b0 , v3 = A1 b2 + A2 b1 + A3 b0 , Ā·Ā·Ā· . Therefore, the conditions vj = 0, j = 0, 1, 2, . . ., yield also, in addition to the already written Hamilton-Jacobi equation, linear diļ¬erential equations for functions bj . These equations may be written in the form A1 b0 = 0, A1 b1 = āA2 b0 , A1 b2 = āA2 b1 ā A3 b0 , Ā·Ā·Ā· ; i.e., all of them are of the form (12.35)
A1 bj = fj ,
where fj is expressed in terms of b0 , . . . , bjā1 . Equations (12.35) are called transport equations. portance of the role played by A1 .
They show the im-
12.4. The eikonal and transport equations
275
Let us ļ¬nd A1 . We assume that S(x) satisļ¬es the eikonal equation (12.28). Then A1 is determined from the formula A(f (x)eiĪ»S(x) ) = eiĪ»S(x) Ī»mā1 (A1 f ) + O(Ī»mā2 ). The direct computation, using the Leibniz product diļ¬erentiation rule, shows that for |Ī±| = m we have D Ī± (f (x)eiĪ»S(x) ) = Ī»m (Sx (x))Ī± f (x)eiĪ»S(x) + Ī»mā1
n
Ī±j (Sx (x))Ī±āej [Dj f (x)]eiĪ»S(x)
j=1
+ Ī»mā1 cĪ± (x)f (x)eiĪ»S(x) + O(Ī»mā2 ), where Dj = iā1 āxā j , ej is the multi-index with only the jth nonzero entry which is equal to 1, and cĪ± is a C ā -function which depends upon the ļ¬rst and second derivatives of S. Since Ī±j Ī¾ Ī±āej =
ā Ī± Ī¾ , āĪ¾j
it follows that A(f (x)eiĪ»S(x) ) ā¤ ā” n āam (Dj f (x)) + c(x)f (x)ā¦ eiĪ»S(x) + O(Ī»mā2 ), = Ī»mā1 ā£ āĪ¾j Ī¾=Sx (x) j=1
implying
n āam A1 = āĪ¾j j=1
Dj + c(x) = iā1 L1 + c(x),
Ī¾=Sx (x)
where L1 is the derivative along the vector ļ¬eld ā āam . Ā· V(x) = āĪ¾ Ī¾=Sx (x) āx What is the geometric meaning of V(x)? Let us write the Hamiltonian system with the Hamiltonian am (x, Ī¾): m xĖ = āa āĪ¾ , Ī¾Ė = ā āam āx
and take a bicharacteristic belonging to the graph of the gradient S, i.e., a bicharacteristic (x(t), Ī¾(t)) such that Ī¾(t) = Sx (x(t)). Then V(x(t)) is
276
12. Wave fronts and short-wave asymptotics
the tangent vector to the ray x(t) corresponding to such a bicharacteristic. Therefore, the transport equations (12.35) are of the form 1 d bj (x(t)) + c(x(t))bj (x(t)) = fj (x(t)); i dt i.e., they are ordinary linear diļ¬erential equations along rays. Therefore, the value bj (x(t)) along the ray x(t) is uniquely determined by bj (x(0)). Example 12.5. Consider the wave equation in a nonhomogeneous medium ā2u ā c2 (x)Īu = 0, x ā R3 , āt2 where c(x) is a smooth positive function of x. Due to the general recipe, we seek a rapidly oscillating solution in the form
(12.36)
(12.37)
u(t, x, Ī») = e
iĪ»S(x)
ā
bj (t, x)Ī»āj .
j=0
The principal symbol of the corresponding wave operator is H(t, x, Ļ, Ī¾) = āĻ 2 + c2 (x)|Ī¾|2 , so the phase S(x, t) should satisfy the following eikonal equation: 2 2 āS āS 2 ā c (x) = 0. āt āx The equations of Hamiltonian curves are ā§ āŖ tĖ = ā2Ļ, āŖ āŖ āŖ āØxĖ = 2c2 (x)Ī¾, āŖ ĻĖ = 0, āŖ āŖ āŖ ā©Ī¾Ė = ā2c(x)|Ī¾|2 Ā·
āc āx ,
implying Ļ = Ļ0 , t = ā2Ļ0 Ļ + t0 , where Ļ is the parameter along the curve. Since we are only interested in bicharacteristics, we should also assume that Ļ 2 = c2 (x)|Ī¾|2 or Ļ = āc(x)|Ī¾|. From this we ļ¬nd for |Ī¾| = 0 xĖ 2c2 (x)Ī¾ dx Ī¾ = = = Ā±c(x) . Ė dt ā2Ļ0 |Ī¾| t In particular, the absolute value of the rayās speed is c(x). This easily implies that c(x) is the speed of the wave front propagation at each point. The ļ¬rst and the last equations of bicharacteristics also imply ā2c(x)|Ī¾|2 cx (x) dĪ¾ = = ā|Ī¾| Ā· cx (x). dt ā2Ļ0
12.4. The eikonal and transport equations
277
Thus, x(t) and Ī¾(t) satisfy (12.38)
dx dt dĪ¾ dt
Ī¾ = Ā±c(x) |Ī¾| ,
= ā|Ī¾| Ā· cx (x).
This is a Hamiltonian system with the Hamiltonian H(x, Ī¾) = Ā±c(x)|Ī¾|. In particular, c(x)|Ī¾| = H0 = const along the trajectories. The choice of the constant H0 (determined by the initial conditions) does not aļ¬ect the rays x(t) because (x(t), Ī»Ī¾(t)) is a solution of (12.38) for any Ī» > 0 as soon as (x(t), Ī¾(t)) is. Using this fact, let us try to exclude Ī¾(t). Replacing |Ī¾| by H0 /c(x) we derive from (12.38) that (12.39)
dx dt dĪ¾ dt
= Ā±H0ā1 c2 (x)Ī¾, x (x) = āH0 cc(x) .
Now, substituting the expression Ī¾=Ā±
(12.40)
H0 dx c2 (x) dt
found from the ļ¬rst equation into the second one, we get
(12.41)
d dt
1 dx c2 (x) dt
+
cx (x) = 0. c(x)
Thus, we have shown that every ray x(t) satisļ¬es the second-order diļ¬erential equation (12.41). This is a nonlinear diļ¬erential equation (more precisely, a system of n such equations). It can be resolved with respect to d2 x (its coeļ¬cient is equal to c21(x) and we have assumed that c(x) > 0). dt2 Therefore, the initial conditions x(0) = x0 , x (0) = v0 (the āprimeā stands for the derivative with respect to t) uniquely deļ¬ne x(t). Not every solution x(t) of this equation, however, determines a ray, since the passage from (12.38) to (12.41) is not exactly an equivalence. Let us consider this in detail. Let x(t) be a solution of (12.41). Determine Ī¾(t) via (12.40) choosing (so far, arbitrarily) a positive constant H0 . Then (12.41) is equivalent to the second of equations (12.39), whereas the ļ¬rst of equations (12.39) means simply that Ī¾(t) is obtained via (12.40). Therefore, if (x(t), Ī¾(t)) is a solution
278
12. Wave fronts and short-wave asymptotics
of (12.39) for some H0 > 0, then x(t) is a solution of (12.41) and, conversely, if x(t) is a solution of (12.41) and H0 is an arbitrary positive constant, then there exists a unique function Ī¾(t), such that (x(t), Ī¾(t)) is a solution of (12.39). How do we pass from (12.39) to (12.38)? If (x(t), Ī¾(t)) is a solution of (12.39) and Ī¾(t) = 0, then clearly (12.38) holds if and only if c(x(t))|Ī¾(t)| = H0 . Note that for this it suļ¬ces that c(x(0))|Ī¾(0)| = H0 . Indeed, it suļ¬ces to verify that the vector ļ¬eld
Ā±H0ā1 c2 (x)Ī¾,
cx (x) āH0 c(x)
is tangent to the manifold M = {(x, Ī¾) : c(x)|Ī¾| = H0 } ā R2n x,Ī¾ . Take (x0 , Ī¾0 ) ā M and a curve (x(t), Ī¾(t)) which is a solution of (12.39) such Ė Ė Ī¾(0)) that (x(0), Ī¾(0)) = (x0 , Ī¾0 ). In particular, the velocity vector (x(0), Ė Ė Ī¾(0)) coincides with the vector ļ¬eld at (x0 , Ī¾0 ). We are to verify that (x(0), is tangent to M at (x0 , Ī¾0 ), i.e., that d 2 (c (x(t))|Ī¾(t)|2)|t=0 = 0. dt But d 2 (c (x(t))|Ī¾(t)|2 )|t=0 = 2c(x0 )[cx (x0 ) Ā· x (0)]|Ī¾0 |2 + 2c2 (x0 )Ī¾0 Ā· Ī¾ (0) dt = 2c(x0 )|Ī¾0 |2 cx (x0 ) Ā· [Ā±H0ā1 c2 (x0 )Ī¾0 ] + 2c2 (x0 )Ī¾0 Ā· [āH0 cā1 (x0 )cx (x0 )] = 2c(x0 )[Ī¾0 Ā· cx (x0 )][Ā±H0ā1 c2 (x0 )|Ī¾0 |2 ā H0 ] = 0, because c(x0 )|Ī¾0 | = H0 due to our assumption that (x0 , Ī¾0 ) ā M . Thus, if c(x(0))|Ī¾(0)| = H0 , then a solution (x(t), Ī¾(t)) of (12.39) is also a solution of (12.38); hence, x(t) is a ray. Now note that equation (12.40) implies that for each ļ¬xed t the condition c(x(t))|Ī¾(t)| = H0 is equivalent to |x (t)| = c(x(t)). In particular, if |x (0)| = c(x(0)) and x(t) is a solution of (12.41), then x(t) is a ray and |x (t)| = c(x(t)) for all t. Thus, the rays x(t) are distinguished among the solutions of (12.41) by the fact that their initial values x(0) = x0 , x (0) = v0 satisfy v0 = c(x0 ). Let us verify āFermatās principleā, which means that the rays are extrema of the functional T = T (Ī³) deļ¬ning time for our wave (e.g., light) to
12.4. The eikonal and transport equations
279
pass along the curve Ī³; i.e.,
T (Ī³) = Ī³
ds , c
where ds is the arc-length element of the curve Ī³ and c = c(x) is considered as a function on Ī³. The extremum is understood with respect to all variations preserving the endpoints, i.e., in the class of the paths Ī³ that connect two ļ¬xed points. Denote these points x0 and x1 . Since the extremality of a curve does not depend on a choice of a parameter, it is convenient to assume that the parameter runs over [0, 1]. Thus, let Ī³ = {x(Ļ ) : Ļ ā [0, 1]}. Take variations of Ī³ preserving the conditions x(0) = x0 , x(1) = x1 . A curve is an extremum if it satisļ¬es the Euler-Lagrange equations (see Section 2.5) for the functional
1 |x(Ļ Ė )|dĻ , T ({x(Ļ )}) = c(x(Ļ )) 0 where the ādotā denotes the derivative with respect to Ļ . Lagrange equations d dĻ for the Lagrangian L =
(12.42)
d dĻ
|x| Ė c(x)
āL ā xĖ
ā
The Euler-
āL =0 āx
can be rewritten as
1 1 dx(Ļ ) c(x(Ļ )) |x(Ļ Ė )| dĻ
+
|x(Ļ Ė )|
cx (x(Ļ )) c2 (x(Ļ ))
= 0.
Clearly, the extremality of a curve is independent upon a choice of the parameter Ļ and its range, and so the Euler-Lagrange equations hold for every such choice or for none. So the parameter Ļ may run over any ļ¬xed segment [a, b]. Let now {x(t)} be a ray with the parameter t (the same as in (12.38)). We know that |x(t)| Ė = c(x(t)) along the ray. But then setting Ļ = t, we see that the Lagrange equation (12.42) turns into (12.41) derived above for rays. Thus, the rays satisfy the Fermat principle. This agrees, in particular, with the fact that in a homogeneous medium (i.e., for c(x) = c = const) the rays are straight lines. Let us write down the transport equations in the described situation. For this purpose, substitute a series u(t, x, Ī») of the form (12.37) into equation (12.36). We will assume that S(t, x) satisļ¬es the eikonal equation. Then
280
12. Wave fronts and short-wave asymptotics
the series
ā2u āt2
ā c2 (x)Īu begins with Ī»1 . We have
ut (t, x, Ī») = iĪ»St e
utt (t, x, Ī») =
ux (t, x, Ī») =
Īu(t, x, Ī») =
iĪ»S
ā
bj Ī»
āj
+e
iĪ»S
ā ābj
āt
Ī»āj ,
j=0 j=0 ā ā āĪ»2 St2 eiĪ»S bj Ī»āj + iĪ»Stt eiĪ»S bj Ī»āj j=0 j=0 ā ā ābj āj ā 2 bj āj Ī» + eiĪ»S +2iĪ»St eiĪ»S Ī» , āt āt2 j=0 j=0 ā ā ābj āj Ī» , iĪ»Sx eiĪ»S bj Ī»āj + eiĪ»S āx j=0 j=0 ā ā bj Ī»āj + iĪ»ĪSeiĪ»S bj Ī»āj āĪ»2 Sx2 eiĪ»S j=0 j=0 ā ā ābj āj Ī» + eiĪ»S +2iĪ»eiĪ»S Sx Ā· (Ībj )Ī»āj . āx j=0 j=0
(Here Sx2 means Sx Ā· Sx = |Sx |2 .) Now, compute utt ā c2 (x)Īu. The principal terms (with Ī»2 ) in the expressions for utt and c2 (x)Īu cancel due to the eikonal equation. Equating the coeļ¬cients of all the powers of Ī» to zero and dividing by 2i, we get 2 0 St āb āt ā c (x)Sx Ā· 2 1 St āb āt ā c (x)Sx Ā·
(12.43)
āb0 1 2 āx + 2 [Stt ā c (x)ĪS]b0 āb1 1 2 āx + 2 [Stt ā c (x)ĪS]b1 2 +(2i)ā1 [ āātb20 ā c2 (x)Īb0 ] = 0, āb2 1 2 2 2 St āb āt ā c (x)Sx Ā· āx + 2 [Stt ā c (x)ĪS]b2 2 +(2i)ā1 [ āātb21 ā c2 (x)Īb1 ] = 0,
= 0,
Ā·Ā·Ā·
āb St ātj
āb ā c2 (x)Sx Ā· āxj + 12 [Stt ā c2 (x)ĪS]bj ā2b +(2i)ā1 [ ātjā1 ā c2 (x)Ībjā1 ] = 0. 2
These are the transport equations. The Hamiltonian equations imply that the vector ļ¬eld (St , āc2 (x)Sx ) = (Ļ, āc2 (x)Ī¾)
12.5. The Cauchy problem with rapidly oscillating initial data
281
is tangent to the rays; i.e., the equations (12.43) are of the form d bj (t(Ļ), x(Ļ)) + a(Ļ)bj (t(Ļ), x(Ļ)) + f (Ļ) = 0, dĻ where a(Ļ) only depends on S and f (Ļ) depends on S, b0 , . . . , bjā1 while (t(Ļ), x(Ļ)) is a ray with the parameter Ļ considered at the beginning of the discussion of this example. Let us indicate another method of analyzing this example and examples of a similar structure. We may seek solutions of (12.36) in the form (12.44)
u(t, x, Ļ) = eiĻt v(x, Ļ),
where Ļ is a large parameter and v is a formal series of the form ā iĻS(x) vk (x)Ļ āk . v(x, Ļ) = e k=0
Equation (12.36) for a function u(t, x, Ļ) of the form (12.44) turns into an equation of the Helmholtz type for v(x, Ļ) [Ī + Ļ 2 cā2 (x)]v(x, Ļ) = 0. For S(x) we get the eikonal equation Sx2 (x) =
1 . c2 (x)
We could get the same result assuming initially S(t, x) = t ā S(x). We will not write now the transport equations for functions vk (x). The structure of these equations is close to that of nonstationary transport equations (12.43). Remark. Transport equations describe transport of energy, polarization, and other important physical phenomena in physical problems. We will not dwell, however, on these questions which rather belong to physics.
12.5. The Cauchy problem with rapidly oscillating initial data Consider the equation Au = 0, where A = a(t, x, Dt , Dx ), t ā R, x ā Rn , and A is hyperbolic with respect to t. Let us consider a formal series ā iĪ»S(t,x) bj (t, x)Ī»āj (12.45) u(t, x, Ī») = e j=0
and try to ļ¬nd out which initial data at t = 0 enable us to recover this series if it is an asymptotic solution. We will only consider the phase functions S(t, x) such that gradx S(t, x) = 0. The function S(t, x) should satisfy the eikonal equation (12.46)
am (t, x, St , Sx ) = 0,
282
12. Wave fronts and short-wave asymptotics
where am (t, x, Ļ, Ī¾) is the principal symbol of A (which can be assumed real without loss of generality). But as we have seen above in Section 12.3, this is equivalent to the fulļ¬llment of one of m equations āS = Ļl (t, x, Sx ) āt
(12.47)
where Ļl (t, x, Ī¾), for l = 1, 2, . . . , m, is the complete set of roots of the equation am (t, x, Ļ, Ī¾) = 0, chosen, say, in increasing order. Setting the initial condition S|t=0 = S0 (x),
(12.48)
where S0 (x) is such that gradx S0 (x) = 0, we may construct exactly m diļ¬erent solutions of the eikonal equation (12.46) satisfying this initial condition, namely, solutions of each of the equations (12.47) with the same initial condition (12.48) (such a solution is unique as soon as we have choosen the root Ļl ). Solutions of equations (12.47) exist in a small neighborhood of the plane t = 0 determined as indicated in Section 12.2. In the same neighborhood, the functions bj (t, x) are uniquely deļ¬ned from the transport equations provided the initial values are given: bj |t=0 = bj,0 (x), j = 0, 1, 2, . . . . Let us now formulate the Cauchy problem for the equation Au = 0 by imposing the initial conditions of the form (0) āj u|t=0 = eiĪ»S0 (x) ā j=0 cj (x)Ī» , (1) ā ut |t=0 = Ī»eiĪ»S0 (x) j=0 cj (x)Ī»āj , (12.49) Ā· Ā·Ā· (mā1) ā mā1 u = Ī»mā1 eiĪ»S0 (x) ā (x)Ī»āj . j=0 cj ātmā1 t=0
Solving this problem would allow us to ļ¬nd with arbitrarily good accuracy (for large Ī») solutions of the equation Au = 0 with rapidly oscillating initial values that are ļ¬nite segments of the series on the right-hand sides of (12.49). A solution of this problem is not just a series of the form (12.45): there may be no such series. We should take a ļ¬nite sum of these series with diļ¬erent phases S(t, x). Thus, consider a sum (12.50)
u(t, x, Ī») =
m
ur (t, x, Ī»),
r=0
where r
u (t, x, Ī») = e
iĪ»S r (t,x)
ā j=0
bj (t, x)Ī»āj . (r)
12.5. The Cauchy problem with rapidly oscillating initial data
283
Sum (12.50) is called a solution of the equation Au = 0 if each ur (t, x, Ī») k r is a solution of this equation. Since all the series āātuk are of the form t=0
k iĪ»S0 (x)
Ī» e
ā
cj (x)Ī»āj ,
j=0
the initial values u|t=0 , ut |t=0 , . . . ,
ā mā1 u ātmā1 t=0
also are of the same form.
Therefore, it makes sense to say that u(t, x, Ī») satisļ¬es the initial conditions (12.49). For instance, we may consider a particular case when the series in (12.49) contain one term each: u|t=0 = eiĪ»S0 (x) Ļ0 (x), Ā·Ā·Ā· ā mā1 u | = Ī»mā1 eiĪ»S0 (x) Ļmā1 (x). ātmā1 t=0 Cutting oļ¬ the series for u at a suļ¬ciently remote term, we see that any asymptotic solution provides (regular) functions u(t, x, Ī»), such that Au = O(Ī»āN ), u|t=0 = eiĪ»S0 (x) Ļ0 (x) + O(Ī»āN ), Ā· Ā·Ā· ā mā1 u = Ī»mā1 eiĪ»S0 (x) Ļmā1 (x) + O(Ī»āN ), ātmā1 t=0
where N may be taken arbitrarily large. Theorem 12.5. The above Cauchy problem for the hyperbolic equation Au = 0 with initial conditions (12.49) is uniquely solvable in a small neighborhood of the initial plane t = 0. Proof. Each of the phases S r (t, x) should satisfy the eikonal equation and the initial condition S|t=0 = S0 (x). There exist, as we have seen, exactly m equations for the amplitudes such phases S 1 , S 2 , . . . , S m . Now the transport (r) (r) bj (t, x) show that it suļ¬ces to set bj = bj,r (x) to uniquely deļ¬ne all t=0
(r)
functions bj (t, x). Thus we need to show that the initial conditions (12.49) m r uniquely determine bj,r (x). Substitute u = r=1 u into the initial con(r) ditions (12.49) and write the obtained equations for bj (t, x) equating the coeļ¬cients of all the powers of Ī». The ļ¬rst of equations (12.49) gives m r=1
(r) bj
t=0
(0)
= cj , j = 0, 1, . . . .
284
12. Wave fronts and short-wave asymptotics
The second of equations (12.49) becomes (r) m m ābjā1 āS r (r) + b i āt j t=0 āt r=1
r=1
t=0
In general, the equation which determines (12.51) m āS r k (r) bj i āt r=1
(k)
= fj (x),
(1)
= cj , j = 0, 1, . . . .
āk u ātk t=0
becomes
k = 0, 1, . . . , m ā 1, j = 0, 1, . . . ,
t=0
(k)
(r)
(r)
where fj (x) only depends on b0 , . . . , bjā1 . Now, note that āS0 āS r , = Ļr 0, x, āt t=0 āx 0 where Ļr (0, x, Ī¾) = Ļl (0, x, Ī¾) if r = l and Ī¾ = 0. Therefore, if āS āx = 0 (which we assume to be the case in the problem considered), then the system (12.51) (r) = bj,r (x) for a ļ¬xed j is a system of linear equations with respect to bj t=0 with the determinant equal to the Vandermond determinant 1 . . . 1 iĻ1 . . . iĻm , ... ... ... (iĻ1 )mā1 . . . (iĻm )mā1 0 where Ļr = Ļr (0, x, āS āx ). Since, due to hyperbolicity, Ļj = Ļk for j = k, the 0 determinant does not vanish anywhere (recall that we assumed āS āx = 0).
(r)
(r)
The right-hand sides of (12.51) depend only on b0 , . . . , bjā1 . Therefore, the sequence of the systems of linear equations (12.51) enables us to (r) uniquely determine all functions bj by induction. Namely, having found (r)
b0,r (x) for r = 1, . . . , m from (12.51) with j = 0, we may determine b0 (t, x) (r) from the transport equations with the initial conditions b0 |t=0 = b0,r (x). (r) If b0 , . . . , bjā1 are determined, then the functions bj |t=0 , r = 1, . . . , m, are recovered from (12.51), and then the transport equations enable us to (r) determine bj (t, x). Let us brieļ¬y indicate the connection of Theorem 12.5 with the behavior of the singularities for solutions of the Cauchy problem. It is well known that the singularities of f (x) are connected with the behavior of its Fourier Ė at inļ¬nity. For instance, the inversion formula transform f (Ī¾)
(12.52) f (x) = (2Ļ)ān fĖ(Ī¾)eixĀ·Ī¾ dĪ¾
12.5. The Cauchy problem with rapidly oscillating initial data
285
Ė ā¤ C(1 + |Ī¾|)āN , then f ā C N ānā1 (Rn ). Therefore, implies that if |f (Ī¾)| to ļ¬nd the singularities of a solution u(t, x) it suļ¬ces to solve the Cauchy problem up to functions whose Fourier transform with respect to x decreases suļ¬ciently rapidly as |Ī¾| ā ā. Formula (12.52) can be regarded as a representation of f (x) in the form of a linear combination of exponents eiĪ¾Ā·x . Since the problem is a linear one, to ļ¬nd solutions of the problem with one of the initial conditions equal to f (x), it suļ¬ces to consider a solution with the initial data f (x) replaced by eixĀ·Ī¾ and then take the same linear combination of solutions. If f ā E (Rn ) and supp f ā K, where K is a compact in Rn , it is more convenient to somewhat modify the procedure. Namely, let Ļ ā C0ā (Rn ), Ļ = 1 in a neighborhood of K. Then
ān fĖ(Ī¾)Ļ(x)eixĀ·Ī¾ dĪ¾ (12.53) f (x) = Ļ(x)f (x) = (2Ļ) and it suļ¬ces to solve the initial problem by replacing f (x) by Ļ(x)eixĀ·Ī¾ . For instance, if f (x) = Ī“(x), Ļ ā C0ā (Rn ), Ļ(0) = 1, then (12.53) holds with fĖ(Ī¾) ā” 1; therefore, having solved the Cauchy problem ā§ āŖ Au = 0, āŖ āŖ āŖ āŖ āŖ āŖ āØu|t=0 = 0, ut |t=0 = 0, (12.54) āŖ āŖ āŖ Ā· Ā·Ā· āŖ āŖ āŖ mā1 u āŖ ā ā© mā1 = Ļ(x)eixĀ·Ī¾ āt
t=0
and having denoted the solution by uĪ¾ (t, x), we can write
ān uĪ¾ (t, x)dĪ¾ E(t, x) = (2Ļ) (if the integral exists in some sense, e.g., converges for any ļ¬xed t in the topology of D (Rnx )), and then E(t, x) is a solution of the Cauchy problem ā§ āŖ AE = 0, āŖ āŖ āŖ āŖ āŖ āŖ āØE|t=0 = 0, Et |t=0 = 0, (12.55) āŖ āŖ āŖ Ā· Ā· Ā· āŖ āŖ āŖ mā1 E āŖ ā ā© mā1 = Ī“(x). āt
t=0
If we only wish to trace the singularities of E(x), we should investigate the behavior of uĪ¾ (x) as |Ī¾| ā +ā. We may ļ¬x the direction of Ī¾ introducing Ī¾ . Then the Cauchy problem a large parameter Ī» = |Ī¾| and setting Ī· = |Ī¾|
286
12. Wave fronts and short-wave asymptotics
(12.54) takes on the form of the Cauchy problem with rapidly oscillating initial data. Namely, in (12.49) we should set S0 (x) = Ī· Ā· x, (0)
(1)
(mā2)
cj (x) = cj (x) = Ā· Ā· Ā· = cj (mā1) cj (x) (mā1) cmā1 (x)
= 0,
(x) = 0,
for j = 0, 1, . . . ,
for j = m ā 1,
= Ļ(x).
We may ļ¬nd a solution uĪ¾ (t, x) = uĪ· (t, x, Ī») which satisļ¬es (12.54) with accuracy up to O(Ī»āN ) for arbitrarily large N . Then the equations in (12.55) are satisļ¬ed up to functions of class C N ānā1 (Rn ). It can be proved that singularities of the genuine solution of (12.55) are the same as for the obtained approximate solution; i.e., the genuine solution diļ¬ers from the approximate one by an arbitrarily smooth function (for large values of N ). Where are the singularities of E(t, x)? To ļ¬nd out, we should consider the integral
ān uĪ¾,N (t, x)dĪ¾, EN (t, x) = (2Ļ) where uĪ¾,N is the sum of the ļ¬rst N + 1 terms of the series which represents the asymptotic solution uĪ¾ . All these terms are of the form
r k Ī» eiĪ»S (t,x,Ī·) Ļ(t, x, Ī·)dĪ¾, where Ī» = |Ī¾|, Ī· = Ī¾/|Ī¾|, S r (t, x, Ī·) is one of the phase functions satisfying the eikonal equation with the initial condition S r |t=0 = S0 (x) = Ī· Ā· x; Ļ(t, x, Ī·) is inļ¬nitely diļ¬erentiable with respect to all variables; Ļ is deļ¬ned for small t and has the (t, x)-support belonging to an arbitrarily small neighborhood of the set of rays (t, x(t)) with the source in (0, 0). The last statement is deduced from the form of the transport equations. Therefore we see that the support of EN (t, x) belongs to an arbitrarily small neighboorhood of the set of rays passing through (0, 0). Taking into account the fact that the singularities of E and EN coincide (with any accuracy), we see that the distribution E(t, x) is inļ¬nitely diļ¬erentiable outside of the set of rays with (0, 0) as the source. Here we should take all the rays corresponding to all the functions S r (t, x, Ī·) for each Ī· and each r = 1, . . . , m. We get the set of m curved cones formed by all rays starting from (0, 0) (for each Ī· ā Rn , |Ī·| = 1, and each r = 1, . . . , m we get a ray smoothly depending on Ī·). A more detailed description is left to the reader as a useful exercise which should be worked out for second-order equations, for example, for the equation (12.36) describing light propagation in a nonhomogeneous medium.
12.6. Problems
287
12.6. Problems 12.1. For the equation utt = a2 Īu, where u = u(t, x), x ā R2 , ļ¬nd the characteristics whose intersection with the plane t = 0 is a) the line Ī· Ā· x = 0, where Ī· ā R2 \0), b) the circle {x : |x| = R}. 12.2. Describe the relation between the rays and characteristics of the equation utt = c2 (x)uxx . 12.3. For the Cauchy problem ā§ 2 āŖ āØutt = x uxx , u|t=0 = 0, āŖ ā© ut |t=0 = eiĪ»x , write and solve the eikonal equation and the ļ¬rst transport equation. Use this to ļ¬nd the function u(t, x, Ī»), such that for x = 0, we have as Ī» ā ā ā§ 2 ā1 āŖ āØutt ā x uxx = O(Ī» ), u|t=0 = O(Ī»ā2 ), āŖ ā© ut |t=0 = eiĪ»x + O(Ī»ā1 ).
10.1090/gsm/205/13
Chapter 13
Answers and hints. Solutions
1.1. a) upp + uqq + urr = 0; the change of variables is ā§ āŖ āØp = x, q = āx + y, āŖ ā© r = x ā 12 y + 12 z. (The reader should keep in mind that a change of variables leading to the canonical form is not unique; the above one is one of them, and it may very well happen that the reader will ļ¬nd another one.) b) upp ā uqq + 2up = 0; the change of variables is ā§ āŖ āØp = x + y, q = x ā y, āŖ ā© r = y + z. 1.2. a) uzw ā
1 w uz
ā
b) uzz + uww +
1 4z uw
1 z+w uz
ā
1 u z2
+
= 0; the change of variables is
z = yxā3 , w = xy.
1 2w uw
= 0; the change of variables is z = y 2 ā x2 , w = x2 . 289
290
13. Answers and hints. Solutions
c) uww + 2uz + uw = 0; the change of variables is z = x + y, w = x. @ 1.3. a) u(x, y) = f ( xy ) + xy g(xy), where f , g are arbitrary functions of one independent variable. (Here and in b) below the form of the general solution is not unique.) Hint. The change of variables z = ln |xy|, w = ln |y/x| ā (2uw + u) = 0 which reduces the equation to the form 2uzw + uz = 0; i.e., āz implies 2uw + u = F (w) and u(z, w) = a(w) + exp(āw/2)b(z).
b) u(x, y) = ln | xy |f (xy) + g(xy), where f, g are arbitrary functions of one independent variable. Hint. The change of variables z = ln |xy|, w = ln |y/x| reduces the equation to the form uww = 0; hence u(z, w) = wĻ(z) + Ļ(z). 2.1. a) ux |x=0 = 0. Hint. Write the dynamical Newton equation of the motion of the ring. b) (mutt ā T ux )|x=0 = 0. Hint. See the hint to a). c) (mutt + Ī±ut ā T ux )|x=0 = 0. l 2.2. a) E(t) = 12 0 [Ļu2t (t, x) + T u2x (t, x)]dx = E0 = const . b), c) Let 1 E(t) = mu2t |x=0 + 2
l
[Ļu2t (t, x) + T u2x (t, x)]dx. 0
t Then E(t) ā” E0 = const in case b) and E(t) ā E(0) = Af r = ā 0 Ī±u2t |x=0 dt in case c) (here Af r is the work of the friction force). 2.3. Ļutt = Euxx or utt = a2 uxx with a = E/Ļ. Here u = u(t, x) is a longitudinal displacement of a point with position x in equilibrium, Ļ is the volume density of the material of the rod, E is the Young modulus (also called the modulus of elongation) which describes the force arising in the deformed material by the formula F = ES Īl l (Hookeās law), S is the cross
13. Answers and hints. Solutions
291
section where the force is measured, Īl/l is the deformation of the small piece of material around the point of measurement (Īl/l is often called the relative lengthening; here l is the length of the piece in equilibrium position and Īl is the increment of this length caused by the force). Hint. Prove that the relative lengthening of the rod at the point with position x in equilibrium is ux = ux (t, x) so the force F = ESux (t, x) is acting on the left part of the rod in the corresponding cross section at the point x. Consider the motion of the part [x, x + Īx] of the rod; write Newtonās dynamical equation for this part and then take the limit as Īx ā 0. 2.4. ux |x=0 = 0. 2.5. (ESux ā ku)|x=0 = 0. Here k is the rigidity coeļ¬cient of the spring, i.e., the force per unit lengthening of the spring. 2.6. a), b) 1 E(t) ā” 2
l
[Ļu2t (t, x) + Eu2x (t, x)]dx = E0 = const . 0
c) E(t) ā” 12 ku2 |x=0 +
1 2
l
2 0 [ĻSut (t, x) +
ESu2x (t, x)]dx = E0 = const.
ā (ES(x) āu 2.7. ĻS(x)utt = āx āx ), where S(x) is the area of the cross section of the rod at the point with position x in equilibrium (the x-axis should be directed along the axis of the rod). ā 2.9. E(t) ā” 12 āā [Ļu2t (t, x) + T u2x (t, x)]dx = E0 = const (perhaps +ā for all t).
Hint. Using Problem 2.8 prove that E(t ) ā¤ E(t) for t ā„ t, and then reverse the direction of the time variable. 2.10. See Figure 13.1. 2.11. u ā” 0.
Hint. The Cauchy initial conditions are u|x=x0 ā” 0, āu āx x=x0 ā” 0. f (x ā at) for x > x0 + Īµ, where f (x) = 0 for x < x0 + Īµ, 2.12. u = g(x + at) for x < x0 ā Īµ, where g(x) = 0 for x > x0 ā Īµ. 2.13. See Figure 13.2. 2.14. See Figure 13.3. Hint. Use the formula
x+at
z 1 1 Ļ(s)ds = ĪØ(x + at) ā ĪØ(x ā at), where ĪØ(z) = Ļ(s)ds. 2a xāat 2a 0
292
13. Answers and hints. Solutions
c
t=0
d
x t = Īµ < l/2a, l = d ā c x t = l/2a x t = l/2a + Īµ x t
l/2a x
Figure 13.1
The graphs of ĪØ(x + at) and ĪØ(x ā at) are drawn on Figure 13.3 by dashed lines. ā 2.15. E(t) ā” 12 0 (Ļu2t (t, x) + T u2x (t, x))dx = E0 = const. Hint. See the hint to Problem 2.9. 2.16. The reļ¬ected wave is & ' ā1 2
x k Ļ2 2Ļk k cos Ļ t ā ā exp (x ā at) + 2 E2S2 a aES a ES 2
x Ļ k2 + , ā 2 2 sin Ļ t ā a2 E S a with notation as in the answer to Problem 2.3. 2.17. On the left u = sin Ļ x+at a+v with the frequency Ļl = On the right u = sin Ļ xāat vāa with the frequency Ļr =
Ļa a+v < Ļ. Ļa aāv > Ļ.
2.18. The standing waves are X0 (x)(at+b) and Xn (x)(an cos Ļn t+bn sin Ļn t) Ļna for n = 1, 2, . . . with Xn (x) = cos nĻx l and Ļn = l . For the graphs of the ļ¬rst Xn (x), n = 1, 2, 3, see Figure 13.4. 2.19. The boundary conditions are u|x=0 = 0 and (ESux + ku)|x=l = 0. Ī³ x The standing waves are Xp (x)(ap cos Ļp t+bp sin Ļp t), where Xp (x) = sin pl , Ļp = Ī³p a/l, and the Ī³p are the solutions of the equation cot Ī³ = ā
kl ESĪ³
13. Answers and hints. Solutions
l
293
t=0 x t = Īµ < l/2a
3l
x t = l/2a x t = l/2a + Īµ x t = l/a x t = l/a + Īµ x t = 3l/2a x t = 3l/2a + Īµ x t = 2l/a x t = 2l/a + Īµ x t = 5l/2a x t = 5l/2a + Īµ x t = 3l/a + Īµ x t > 3l/a x Figure 13.2
such that Ī³p ā (pĻ, (p + 1)Ļ) for p = 0, 1, 2, . . . . The system {Xp : p = 0, 1, . . .} is orthogonal in L2 ([0, l]), and every function Xp is an eigenfunction of the operator L = ād2 /dx2 with the boundary conditions X(0) = 0, X (0) + Ī±X(0) = 0 for Ī± = k/(ES). The corresponding eigenvalue is Ī»p = (Ī³p /l)2 and its short-wave asymptotic is ' & Ļ(2p + 1) 2 1 1+O . Ī»p = 2l p2
294
13. Answers and hints. Solutions
āl
l
t=0 x
3l
t = Īµ < l/a x
t = l/a x
t = l/a + Īµ x
t = 2l/a x
t = 2l/a + Īµ x
t = 3l/a x 2l t > 3l/a x
Figure 13.3
2.20. a) The resonance frequencies are Ļk = (2k+1)Ļa , k = 0, 1, 2, . . .; the 2l condition for a resonance is Ļ = Ļk for some k. If Ļ = Ļk , then u(t, x) =
F0 a Ļ ESĻ cos Ļ l sin a x Ā· sin Ļt a 2F0 Ļa2 k + ā k=0 (ā1) Ļk lES(Ļ 2 āĻ 2 ) k
sin Ļk t Ā· sin (2k+1)Ļx . 2l
13. Answers and hints. Solutions
295
X0 ā” 1
l
0
x l
0
X1 (x) = cos
Ļx l
X2 (x) = cos
2Ļx l
X3 (x) = cos
3Ļx l
x
l
0
x
l
0
x
Figure 13.4
If Ļ = Ļk for some k, then 2
F0 a 3 Ļ xĻ Ļ Ļ u(t, x) = (ā1)k ESlĻ 2 [ 2 sin Ļt Ā· sin a x ā a sin Ļt Ā· cos a x ā Ļt cos Ļt Ā· sin a x] (2p+1)Ļx 2F0 Ļa2 + pā„0,p=k (ā1)p Ļp lES(Ļ . 2 āĻ 2 ) sin Ļp t Ā· sin 2l p
Hint. Seek a particular solution in the form u0 = X(x) sin Ļt satisfying the boundary conditions (but not the initial conditions); then ļ¬nd v = uāu0 by the Fourier method in case Ļ = Ļk . In case of a resonance Ļ = Ļk , take the limit of the nonresonance formula for u as Ļ ā Ļk . b) Resonance frequencies are Ļk = kĻa l , k = 1, 2, . . .; the condition for a resonance is Ļ = Ļk for some k. If Ļ = Ļk , then 2
a F0 a t Ļ u(t, x) = ā ESĻF0sin Ļl sin Ļt Ā· cos a x + ESlĻ a 2F0 Ļa2 kĻx k + ā k=0 (ā1) Ļ lES(Ļ 2 āĻ 2 ) sin Ļk t Ā· cos l . k
k
If Ļ = Ļk for some k, then 2
0a [ 2Ļ3 2 sin Ļt Ā· sin Ļa xā Ļ1 ( xa sin Ļt Ā· cos Ļak x +t cos Ļt Ā· sin Ļa x)] u(t, x) = (ā1)k FESl 2
0a t + + FESlĻ
2F0 Ļa2 p pā„1, p=k (ā1) Ļp lES(Ļ 2 āĻp2 )
sin Ļp t Ā· cos pĻx l .
Hint. See the hint to a). for k = 1, 2, . . .; the 2.21. a) The resonance frequencies are Ļk = kĻa l condition for a resonance is Ļ = Ļk for some k. If Ļ = Ļk , then A sin Ļx k+1 2ĻAa sin Ļ t Ā· sin kĻx . u(t, x) = sin Ļla sin Ļt + ā k k=1 (ā1) l l(Ļ 2 āĻ 2 ) a
k
296
13. Answers and hints. Solutions
If Ļ = Ļk for some k, then Ļ [ 1 sin Ļt Ā· sin Ļa x + xa sin Ļt Ā· cos Ļx u(t, x) = (ā1)k Aa a + t cos Ļt Ā· sin a x] l 2Ļ pĻx + pā„1, p=k (ā1)p+1 l(Ļ2ĻAa 2 āĻ 2 ) sin Ļp t Ā· sin l . p
Hint. See the hint to Problem 2.20 a). b) The resonance frequencies are Ļk = (2k+1)Ļa , k = 0, 1, 2, . . .; the con2l dition for a resonance is Ļ = Ļk for some k. If Ļ = Ļk , then u(t, x) =
A cos Ļx a cos Ļl a
sin Ļt +
ā
k 2ĻAa k=0 (ā1) l(Ļ 2 āĻ 2 ) k
sin Ļk t Ā· cos (2k+1)Ļa . 2l
If Ļ = Ļk for some k, then 1 Ļ x Ļx Ļx u(t, x) = (ā1)k Aa l [ 2Ļ sin Ļt Ā· cos a x + a sin Ļt Ā· sin a ā t cos Ļt Ā· cos a ] (2p+1)Ļa + (ā1)p l(Ļ2ĻAa . 2 āĻ 2 ) sin Ļp t Ā· cos 2l pā„0, p=k
p
Hint. See the hint to Problem 2.20 a). c) The resonance frequencies are Ļp = Ī³p a/l, where the Ī³p are as in the answer to Problem 2.19, p = 1, 2, . . .. The condition for a resonance is Ļ = Ļp for some p. If Ļ = Ļp , then with Ī³ = Ļl/a ka Ļx ka Ļx ā1 u(t, x) = A(cos Ļl + ESĻ sin Ļl a ) (cos a + ESĻ sin a ) sin Ļt ā a 4AĪ³Ī³p sin Ī³p Ī³p x Ī³ x kl sin pl ). + p=0 (2Ī³p āsin(2Ī³p ))(Ī³ 2 āĪ³ 2 ) sin Ļp t Ā· (cos l + ESĪ³ p p
If Ļ = Ļp for some p, then u(t, x) =
2Ī³A x x at x 2Ī³āsin 2Ī³ [ l sin Ļt Ā· cos(Ī³( l ā 1)) + l cos Ļt Ā· sin(Ī³( l ā 1))] x x AĪ³(1ācos 2Ī³) sin ĻtĀ·sin(Ī³( l ā1)) A sin ĻtĀ·sin(Ī³( l +1)) + ā 2Ī³āsin 2Ī³ (2Ī³āsin 2Ī³)2 4AĪ³Ī³n sin Ī³n kl + nā„0, n=p (2Ī³n āsin(2Ī³ sin Ļn t Ā· (cos Ī³nl x + ESĪ³ sin Ī³nl x ). 2 2 n n ))(Ī³ āĪ³n )
Hint. See the hint to Problem 2.20 a). x 3.1. Ī¦(x) = cos kx + k1 0 sin k(x ā t)q(t)Ī¦(t)dt; x Ī¦(x) = cos kx + O( k1 ) = cos kx + k1 0 sin k(x ā t) cos kt q(t)dt + O( k12 ); x Ī¦ (x) = āk sin kx + O(1) = āk sin kx + 0 cos k(x ā t) cos kt q(t)dt + O( k1 ). 3.2. Hint. The values of k satisfying Ī¦(l) = Ī¦(l, k) = 0 can be found by the implicit function theorem near k such that cos kl = 0. Diļ¬erentiation with respect to k of the integral equation of Problem 3.1 gives an integral equation for āĪ¦/āk providing information needed to apply the implicit function theorem.
13. Answers and hints. Solutions
297
Figure 13.5
3.3. G(x, Ī¾) = 1l [Īø(Ī¾ ā x)x(l ā Ī¾) + Īø(x ā Ī¾)(l ā x)Ī¾]. Here is a physical interpretation: G(x, Ī¾) is the form of a string which is pulled at point Ī¾ by the vertical point force F = T , where T is the tension force of the string. 3.4. G(x, Ī¾) =
1 sinh l [Īø(Ī¾ ā x) cosh x Ā· cosh(l ā Ī¾) + Īø(x ā Ī¾) cosh(l ā x) Ā· cosh Ī¾].
Here is a physical interpretation: G(x, Ī¾) is the form of a string which is pulled at a point Ī¾ by the vertical point force F = T , has free ends (ends which are allowed to move freely in the vertical direction), and is subject to the action of an elastically returning uniformly distributed force; see Figure 13.5, where the returning force is realized by the little springs attached to the points of the string and to a rigid massive body. 3.6. Hint. A solution y = y(x) ā” 0 of the equation āy + q(x)y = 0 with q ā„ 0 has at most one root. 3.7. Hint. Use the inequality (GĻ, Ļ) ā„ 0 for G = Lā1 ; take Ļ(x) =
n
cj ĻĪµ (x ā xj )
j=1
with a Ī“-like family ĻĪµ . 3.8. a)
ā j=1
1 Ī»j
=
l 0
G(x, x)dx.
Hint. Use the expansion G(x, Ī¾) =
ā Xj (x)Xj (Ī¾) j=1
Ī»j
which can be obtained, e.g., as the orthogonal expansion of G(Ā·, Ī¾) with respect to {Xj (x), j = 1, 2, . . .}.
298
13. Answers and hints. Solutions
b)
ā
1 j=1 Ī»2j
=
l l 0
0
|G(x, Ī¾)|2 dxdĪ¾.
Hint. This is the Parseval identity for the orthogonal expansion of G = G(x, Ī¾) with respect to the system {Xj (x)Xk (Ī¾) : j, k = 1, 2, . . .}. In the particular case L = ād2 /dx2 on [0, Ļ] with boundary conditions X(0) = X(Ļ) = 0 we obtain Ī»n = n2 , and taking into account the answer to Problem 3.3, we get ā 1 Ļ2 , = n2 6
n=1
ā 1 Ļ4 . = n4 90
n=1
4.1. 0. 4.2. 2Ī“(x). 4.3. ĻĪ“(x). 4.4. Hint. f (x) ā f (0) =
1
d 0 dt f (tx)dt
=x
1 0
f (tx)dt.
(R) and its 4.5. Hint. Write explicit formulas for the map D (S 1 ) ā D2Ļ inverse.
4.6. Hint. Use the continuity of a distribution from D (S 1 ) with respect to a seminorm in C ā (S 1 ). 4.7. Hint. Use duality considerations. 1 imx . 4.8. kāZ Ī“(x + 2kĻ) = 2Ļ māZ e 4.9. Hint. Multiply the formula in the answer to Problem 4.8 by f (x) and integrate. 4.10. a) u(x) = v.p. x1 + CĪ“(x) =
1 x+i0
+ C1 Ī“(x) =
b) u(x) = ln |x| + CĪø(x) + C1 . c) u(x) = āĪ“(x) + CĪø(x) + C1 . 5.1. a) ā2ĻiĪø(Ī¾). r0 |Ī¾| . b) 4Ļr0 sin|Ī¾|
Hint. Use the spherical symmetry and take Ī¾ = |Ī¾|(1, 0, 0) = (Ī¾1 , 0, 0). c)
2Ļ 2 Ļ0 Ī“(|Ī¾| ā
Ļ0 ).
Hint. Use the answer to b). d)
(āi)k+1 . (Ī¾āi0)k+1
1 xāi0
+ C2 Ī“(x).
13. Answers and hints. Solutions
5.2. a)
299
1 . (nā2)Ļnā1 |x|nā2
Hint. It should be a spherically symmetric fundamental solution for (āĪ). b)
1 āk|x| . 4Ļ|x| e
Hint. Using the spherical symmetry take Ī¾ = (Ī¾1 , 0, 0). Use the regularization with multiplication by exp(āĪµ|Ī¾|). Calculate the arising onedimensional integral using residues. c)
1 āik|x| . 4Ļ|x| e
Hint. Calculate F ā1 ( |Ī¾|2 āk1 2 Ā±iĪµ ) as in b) and pass to the limit as Īµ ā +0. 5.3.
1 āk|x| . 4Ļ|x| e
5.6. 0 as t ā +ā and ā2ĻiĪ“(x) as t ā āā. 6.1. If Īu = 0, then for some ball BĪµ (x) we have, say, Īu > 0 in BĪµ (x). Let I(r) be as in the proof of Theorem 6.1; by the mean value property, we have I (r) = 0. But then BĪµ Īu dx > 0 is in contradiction with (6.3). 6.2. Use the same arguments as in Theorems 6.1 and 6.2 but replace Īu = 0 with Īu ā„ 0. 6.3. If Īu ā¤ 0, then for some small ball BĪµ (x) we have Īu > 0 in BĪµ (x). Follow the two previous hints/solutions. 6.4. If u(x0 ) ā„ u(x) for all x ā BĪµ (x0 ) but there are some x ā BĪµ (x0 ) for which u(x0 ) > u(x), then
n u(y) dy, u(x0 ) > Ļn Īµn BĪµ (x0 ) contradicting (6.45). 6.5. First reļ¬ect u(x, y) in the x-axis and use Lemma 6.11 to consider u(x, y) as harmonic in the strip āT < y < T . Then pick a point x0 on the x-axis outside the support of Ļ, so the Cauchy data vanishes there. By the CauchyKovalevsky theorem, Theorem 5.13, u(x, y) vanishes in a neighborhood of (x0 , 0). By analyticity, u(x, y) is identically zero. 6.8. Since k ā„ 1, we can let k = k ā 1 ā„ 0. Apply Schauderās theorem, Theorem 6.22, with k instead of k, and use C k+2 ā C k+1,Ī³ ā C k+1 . 6.9. The function f can be recovered from u by the formula f = Īu, whereas at any regular point of the boundary x ā āĪ©, Ļ is the jump of the
300
13. Answers and hints. Solutions
outward normal derivative of u on āĪ©: āu āu (13.1) Ļ(x) = lim (x + ĪµĀÆ n) ā (x ā ĪµĀÆ n) . Īµā0+ ā n ĀÆ ān ĀÆ
6.11. Use the uniqueness of Greenās function combined with the invariance of the problem with respect to all translations parallel to the mirror L, rotations with a ļ¬xed point c ā L, and mirror reļ¬ections with the mirror L. 7.1. u(t, x) =
u1 +u2 2
+
ā
p=0
2(u1 āu2 )(ā1)p (2p+1)Ļ 2
exp[ā( (2p+1)Ļa )2 t] cos (2p+1)Ļx . l l 2
2
l cp l Relaxation time is tr ! 4Ļl2 a2 ! 40a 2 ! 40k , where k is the thermal conductivity, c the speciļ¬c heat (per units of mass and temperature), Ļ the density (mass per volume), l the length of the rod.
Hint. The ļ¬rst term in the sum above is much greater than the other terms at times comparable with the relaxation time. 7.2. Let I be the current, R the electric resistance of the rod, V the volume of the rod, k the thermal conductivity, 2 cp = l
l& 0
' I 2R pĻx Ļ(x) ā x(l ā x) sin dx, where Ļ = u|t=0 . 2V k l
Then & ' ā pĻa 2 pĻx I 2R x(l ā x) + cp exp ā t sin u(t, x) = 2V k l l p=1
and lim u(t, x) =
tā+ā
I 2R x(l ā x). 2V k
Hint. The equation describing the process is ut = a2 uxx + 7.3. lim u(t, x) = tā+ā
I 2R . cĻV
b+c 2 .
Hint. Consider Poissonās integral giving an explicit formula for u(t, x).
13. Answers and hints. Solutions
301
8.1. u(x, y) = a2 (x2 ā y 2 ). Hint. The Fourier method gives in polar coordinates u(r, Ļ) = a0 +
ā
rk (ak cos kĻ + bk sin kĻ).
k=1
8.4. u(t, x) =
1 4 12 (x
ā y4) ā
a2 2 12 (x
ā y 2 ).
Hint. Find a particular solution up of Īu = x2 ā y 2 and then seek v = u ā up . 8.5. In polar coordinates u(r, Ļ) = 1 +
b3 b r ln + (r2 ā a4 rā2 ) cos 2Ļ. 2 a 4(a4 + b4 )
Hint. The Fourier method gives in polar coordinates u(r, Ļ) = A0 + B0 ln r +
ā [(Ak rk + Bk rāk ) cos kĻ + (Ck rk + Dk rāk ) sin kĻ]. k=1
8.6. The answer to the last question is u(x, y) = B +
sinh Ļ(bāy) a sinh ā p=0
Ļb a
sin
Ļx a (2p+1)Ļ(aāx)
sinh 8Ab b 2 2 (2p+1)Ļa (2p + 1) Ļ sinh b
sin
(2p + 1)Ļy . b
Hint. The general scheme is as follows. Step 1. Reduce the problem to the case Ļ0 (0) = Ļ0 (a) = Ļ1 (0) = Ļ0 (0) = Ļ0 (b) = Ļ1 (0) = Ļ1 (b) = 0 by subtracting a function of the form A0 + A1 x + A2 y + A3 xy. Step 2. Seek u in the form u = u1 + u2 with u1 , u2 such that Īu1 = Īu2 = 0; u1 satisļ¬es the Dirichlet boundary conditions with Ļ0 = Ļ1 = 0 and the same Ļ0 , Ļ1 as for u; u2 satisļ¬es the Dirichlet boundary conditions with Ļ0 = Ļ1 = 0 and the same Ļ0 , Ļ1 as for u. Then ā kĻx kĻy ā kĻx b + Dk e b ) sin u1 (x, y) = k=1 (Ck e b ā kĻ(aāx) kĻx = (A sinh + B sinh ) sin kĻy k k b b b , k=1 kĻ(bāy) ā kĻy kĻx ) sin a , u2 (x, y) = k=1 (Ak sinh a + Bk sinh a
302
13. Answers and hints. Solutions
where Ak sinh kĻa b = Bk sinh kĻa b = kĻb Ak sinh a = Bk sinh kĻb a =
kĻy 2 b b 0 Ļ1 (y) sin b dy, kĻy 2 b b 0 Ļ0 (y) sin b dy, 2 a kĻx a 0 Ļ1 (x) sin a dx, 2 a kĻx a 0 Ļ0 (x) sin a dx.
2 Ė + āāyuĖ2 = 0 and 8.7. Hint. If u Ė(Ī¾, y) = eāiĪ¾x u(x, y)dx, then āĪ¾ 2 u u Ė(Ī¾, y) = C1 (Ī¾) exp(ā|Ī¾|y) + C2 (Ī¾) exp(|Ī¾|y). The second summand should vanish almost everywhere if we want u to be bounded. Hence, u Ė(Ī¾, y) = ā|Ī¾|y+i(xāz)Ī¾ 1 Ļ(z)dzdĪ¾. Now, change the e Ļ(Ī¾) Ė exp(ā|Ī¾|y) and u(x, y) = 2Ļ order of integration. 8.8. Ī± ā 2Z+ or Ī± ā s > ān/2. 8.9. No. Hint. Use Friedrichsās inequality. 8.10. Hint. Work with the Fourier transform. Show that the question can be reduced to the case of a multiplication operator in L2 (Rn ). 9.1. Here is a physical interpretation: G(x, y) is the potential at x of a unit point charge located at the point y ā Ī© inside a conducting grounded surface āĪ©. 9.2. The same as for En (x ā y), where En is the fundamental solution of Ī. 9.3. Hint. Use the fact that Lā1 is selfadjoint. 9.4. Hint. Apply the Greenās formula (4.32) with v(x) = G(x, y) (consider y as a parameter) and Ī© replaced by Ī©\B(y, Īµ), where B(y, Īµ) = {x : |xāy| ā¤ Īµ}. Then pass to the limit as Īµ ā +0. ā§ 2 2 āŖ āØ 1 ln (x1 ā y1 ) + (x2 ā y2 ) for n = 2, 2 2 9.5. G(x, y) = 4Ļ (x1 ā
y1 ) + (x2 + y2 ) āŖ 1 1 1 ā© for n ā„ 3, (2ān)Ļnā1 |xāy|nā2 ā |xāy|nā2 where y = (y1 , . . . , ynā1 , āyn ) for y = (y1 , . . . , ynā1 , yn ). Hint. Seek the Greenās function in the form G(x, y) = En (x ā y) + cEn (x ā y), where y ā Ī© and c is a constant.
13. Answers and hints. Solutions
303
9.6.
2xn Ļ(y )dy Ļnā1 |x ā (y , 0)|n nā1 R
2 xn Ļ(y1 , . . . , ynā1 )dy1 . . . ynā1 = n . Ļnā1 [(x1 ā y1 )2 + Ā· Ā· Ā· + (xnā1 ā ynā1 )2 + yn2 ] 2
u(x) =
Rnā1
Hint. Use the formula in Problem 9.4. 9.7. Let the disc or the ball be {x : |x| < R}. Then G(x, y) =
1 2Ļ
R|xāy| ln |y||xāy|
for n = 2,
1
(2ān)Ļnā1 |xāy|nā2
ā
Rnā2 |y|nā2 (2ān)Ļnā1 |xāy|nā2
for n ā„ 3,
2
y is the point which is obtained from y by the inversion with where y = R |y|2 respect to the circle (sphere) {x : |x| = R}.
Hint. If n = 2, seek G(x, y) in the form E2 (x ā y) ā E2 (x ā y) ā c(y). If n ā„ 3, seek G(x, y) in the form En (x ā y) ā c(y)En (x ā y), where (in both 2y |y| |x ā y| = does . Show geometrically that if |x| = R, then cases) y = R |y|2 |x ā y| R not depend on x. 9.8. u(x) =
1 Ļnā1 R
|y|=R
R2 ā |x|2 f (y)dSy . |x ā y|n
Here the disc (or the ball) is Ī© = {x : |x| < R}, f = u|āĪ© . Hint. Use the formula from Problem 8.4. When calculating use that |y| = R implies ā āny |x
y yāx y ā y| = ā ānā y |x ā y| = ā(āy |x ā y|, |y| ) = ā( |yāx| , |y| ) =
ā āny G(x, y),
(x,y)āR2 R|xāy| .
9.9. For the half-ball {x : |x| ā¤ R, xn ā„ 0} the answer is G(x, y) = G0 (x, y) ā G0 (x, y), where G0 is the Greenās function of the ball {x : |x| < R} and y = (y1 , . . . , ynā1 , āyn ) for y = (y1 , . . . , ynā1 , yn ).
304
9.10.
13. Answers and hints. Solutions
G(x, y) = ā ā
1 (x1 āy1 )2 +(x2 āy2 )2 +(x3 āy3 )2 1 + ā 4Ļ (x1 āy1 )2 +(x2 +y2 )2 +(x3 āy3 )2 1 + ā 4Ļ (x1 āy1 )2 +(x2 āy2 )2 +(x3 +y3 )2 1 ā ā . 4Ļ (x1 āy1 )2 +(x2 +y2 )2 +(x3 +y3 )2 4Ļ
Hint. G(x, y) = E3 (x ā y) ā E3 (x ā T2 y) ā E3 (x ā T3 y) + E3 (x ā T2 T3 y), where T2 (y1 , y2 , y3 ) = (y1 , āy2 , y3 ), T3 (y1 , y2 , y3 ) = (y1 , y2 , āy3 ). 9.11. Hint. Use the fact that Ī(z) has simple poles for z = 0, ā1, ā2, . . .. 9.12. Hint. Use the Liouville formula for the Wronskian of two linearly independent solutions of the Bessel equation. 9.13. Hint. Expand the left-hand side in power series in t and x, and use the expansion of Jn (x) given in Problem 9.11. 9.14. Hint. Take t = eiĻ in the formula of Problem 9.13, and use the result of Problem 9.11. 9.15. For the domain {x = (x1 , . . . , xn ) : 0 < xj < aj , j = 1, . . . , n} the A sin k Ļx eigenfunctions are nj=1 ajj j for kj = 1, 2, . . . and the eigenvalues are Ī»k1 ...kn = ā[( ka11Ļ )2 + Ā· Ā· Ā· + ( kannĻ )2 ]. 9.16. Hint. Use the fact that Jk (Ī±k,n x) for all n = 1, 2, . . . are eigenfuncd2 1 d k2 tions of the same operator L = dx 2 + x dx ā x2 (on the space of functions vanishing at x = 1). This operator is symmetric in the space L2 ((0, 1), xdx) consisting of the functions on (0, 1) which are square integrable with respect to the measure xdx. 9.17. For the disc {x : |x| < R}, in polar coordinates the desired system is
Ī± k,n Jk r eikĻ for k ā Z+ , n = 1, 2, . . . , R where the Ī±k,n are deļ¬ned in Problem 9.16. Hint. Write Ī in polar and expand the eigenfunction into coordinates ikĻ Fourier series u(r, Ļ) = k Xk (r)e . Then each term Xk (r)eikĻ will be an eigenfunction and Xk (r) satisļ¬es the Bessel equation. Using regularity at 0, show that Xk (r) = cJk (Ī±r). 9.18. Hint. For the cylinder Ī© = {(x, y, z) : x2 + y 2 < R2 , 0 < z < H} reduce the equation to the case Īu = f, u|āĪ© = 0; then expand u and f in the eigenfunctions of Ī in Ī© which are of the form !
mĻz r sin : k ā Z+ , n = 1, 2, . . . , m = 1, 2, . . . . eikĻ Jk Ī±k,n R H
13. Answers and hints. Solutions
10.1. u(t, x) = eiĻt J0 ( Ļa Ļ), Ļ = {x : x1 = x2 = 0}).
305
x21 + x22 . (The axis of the cylinder is
10.2. See Figure 13.6.
u
t=0
R
0
r = |x| u
t = Īµ < R/a 0 u
t = R/a 0
u
t > R/a 0
Figure 13.6
306
13. Answers and hints. Solutions
ā§ āŖ 1 āŖ āØ 1 at u(t, r) = ā āŖ 2 2r āŖ ā© 0
if r ā¤ R ā at, if r + at > R and āR < r ā at < R, if r ā at ā„ R.
Hint. Introduce a new function v = ru instead of u. (Here r = |x|.) Then v(t, r) = 12 [Ļ(r Ė + at) + Ļ(r Ė ā at)], where ĻĖ is the extension of rĻ to R as an odd function. 10.3. See Figure 13.7.
t=0 |x|
0
t = Īµ < R/2a
t = R/2a
t = R/2a + Īµ
t = R/a
t > R/a
Figure 13.7
13. Answers and hints. Solutions
ā§ āŖ āØt u(t, r) =
1 [R2 āŖ 4ar
ā©
ā (r ā
at)2 ]
0
307
if r + at ā¤ R, if r + at > R and |r ā at| < R, if |r ā at| ā„ R.
Hint. Introduce a new function v = ru with r = |x|. Then
r+at 1 Ė Ļ(s)ds = ĪØ(r + at) ā ĪØ(r ā at), v(t, r) = 2a rāat r 1 Ė Ė where ĪØ(r) = 2a 0 Ļ(s)ds and Ļ is the extension of rĻ to R as an odd function. 10.5. Hint. Consider the Cauchy problem with initial conditions u|t=0 = 0 and ut |t=0 = Ļ(x). Then 1 i[(xāy)Ā·Ī¾+at|Ī¾|] e Ļ(y)dydĪ¾ u(t, x) = (2Ļ)ān 2ia|Ī¾| 1 i[(xāy)Ā·Ī¾āat|Ī¾|] ān Ļ(y)dydĪ¾. ā(2Ļ) 2ia|Ī¾| e Consider, e.g., the ļ¬rst summand. Multiplying the integrand by a C ā function Ļ(Ī¾) that is equal to 1 for |Ī¾| > 1 and vanishes for |Ī¾| < 1/2 (this does not aļ¬ect the singularities of u), try to ļ¬nd a diļ¬erential operator L = nj=1 cj (t, x, y, Ī¾) āĪ¾ā j such that Lei[(xāy)Ā·Ī¾+at|Ī¾|] = ei[(xāy)Ā·Ī¾+at|Ī¾|] . Prove that such an operator L exists if and only if |x ā y|2 = a2 t2 (here x, y, t are considered as parameters) and integrate by parts N times, thus moving L from the exponential to other terms. 10.6. {(t, x, y) : t ā„ 0, t2 = x2 + y 2 /4}. Hint. Reduce the operator to the canonical form. 10.7. a) {(t, x, y) : 0 ā¤ t ā¤ min( a2 ,
b 2 ),
x ā [t, a ā t], y ā [t, b ā t]}.
(See Figure 13.8: the lid of a coļ¬n.) b) {(t, x, y) : t ā„ 0, dist((x, y), Ī ) ā¤ t}, where Ī = [0, a] Ć [0, b]. (See Figure 13.9: an inļ¬nitely high cup of a stadium.) 10.8. a) Let {(x, y, z) : x ā [0, a], y ā [0, b], z ā [0, c]} be the given parallelepiped.
t
x2 b
a Figure 13.8
x1
308
13. Answers and hints. Solutions
Figure 13.9
Figure 13.10
Figure 13.11
Then the answer is a b c , , , (t, x, y, z) : 0 ā¤ t ā¤ min 2 2 2
t ā¤ x ā¤ a ā t, t ā¤ y ā¤ b ā t, t ā¤ z ā¤ c ā t The section by the hyperplane t = 1 is a smaller parallelepiped, or a rectangle, or an interval, or the empty set (Figure 13.10). b) {(t, x, y, z) : dist((x, y, z), Ī ) ā¤ t}, where Ī is the given parallelepiped. The section by the hyperplane t = 1 is plotted in Figure 13.11. 11.1.
u(x) =
Q ln R ā 2Ļ Q ā 2Ļ ln r
if |x| ā¤ R, if |x| ā„ R,
where r = |x|, the circle is {x : |x| = R}, Q is the total charge of the circle, i.e., Q = |y|=R Ļ(y)dsy , where Ļ is the density that deļ¬nes the potential.
13. Answers and hints. Solutions
11.2.
Qr 2 ā 4ĻR 2 + u(x) = Q ā 2Ļ ln r
309
Q 4Ļ
ā
Q 2Ļ
ln R
if r ā¤ R, if r ā„ R.
Here Q is the total charge of the disc {x : |x| ā¤ R}, r = |x|. x2 + y 2 in the 11.3. The same answer as in Problem 11.1 with r = (x, y, z)-space, where the z-axis is the axis of the cylinder and R is the radius of its cross section by the xy-plane. 11.4.
Ī±0 , u(x) = 0,
r < R, r > R,
where Ī±0 is the density of the dipole moment, r is the distance to the center of the circle or to the axis of the cylinder, and R is the radius of the circle or of the cross section of the cylinder. 11.5.
u(x) =
Q (nā2)Ļnā1 Rnā2 Q (nā2)Ļnā1 |x|nā2
if |x| ā¤ R, if |x| > R,
where the sphere of radius R is taken with the center at the origin and Q is the total charge of the sphere. 11.6. The induced charge is q = ā QR d . Hint. The potential vanishes on the surface of the sphere, hence inside of the sphere. Calculate the potential in the center assuming that we know the distribution of the charge on the surface. 11.7. Ļ(x) = Q is situated.
Q(R2 ā d2 ) , where y is the point at which the initial charge 4ĻR|x ā y|3
Hint. Let y be the point obtained by reļ¬ecting y with respect to the 2y ). Place the charge q at sphere (if the sphere is {x : |x| = R}, then y = R |y|2 q Q the point y; then the potential u(x) = 4Ļ|xāy| + 4Ļ|xāy| coincides with the potential of all real charges (Q and induced ones) outside the sphere because of the uniqueness of the solution of the Dirichlet problem in the exterior. Then āu x āu(x) , where = āu, . Ļ(x) = ā ār ār |x|
11.8. The charge will be distributed uniformly over the whole wire. 12.1. a) at Ā± x Ā·
Ī· |Ī·|
= 0.
b) at Ā± (|x| ā R) = 0.
310
13. Answers and hints. Solutions
12.2. Rays and characteristics are the same curves in the (t, x)-plane. Hint. Use the description of rays as solutions of (12.39) satisfying |x(t)| Ė = c(x(t)). 12.3. u(t, x, Ī») =
iĪ»ā1 āt 2x [exp(iĪ»xe
+ 32 t) ā exp(iĪ»xeāt ā 32 t)].
Bibliography
[1] M. A. Armstrong, Basic topology, corrected reprint of the 1979 original, Undergraduate Texts in Mathematics, Springer-Verlag, New York-Berlin, 1983. MR705632 [2] V. I. Arnold, Ordinary diļ¬erential equations, translated from the Russian and edited by Richard A. Silverman, MIT Press, Cambridge, Mass.-London, 1978. MR0508209 [3] V. I. Arnold, Mathematical methods of classical mechanics, translated from the Russian by K. Vogtmann and A. Weinstein; Graduate Texts in Mathematics, 60, Springer-Verlag, New York-Heidelberg, 1978. MR0690288 [4] F. A. Berezin and M. A. Shubin, SchrĀØ odinger equation, Kluwer, Dordrecht, 1987. [5] R. Courant and D. Hilbert, Methods of mathematical physics, Interscience, New York, Vol. I, 1953; Vol. II, 1962. [6] L. C. Evans, Partial diļ¬erential equations, 2nd ed., Graduate Studies in Mathematics, vol. 19, American Mathematical Society, Providence, RI, 2010. MR2597943 [7] G. Eskin, Lectures on linear partial diļ¬erential equations, Graduate Studies in Mathematics, vol. 123, American Mathematical Society, Providence, RI, 2011. MR2809923 [8] R. Feynman, R. B. Leighton, and N. Sands, The Feynman lectures on physics, SpringerVerlag, Reading, MA, Palo Alto, London, 1967. [9] D. Gilbarg and N. S. Trudinger, Elliptic partial diļ¬erential equations of second order, reprint of the 1998 edition, Classics in Mathematics, Springer-Verlag, Berlin, 2001. MR1814364 [10] A. Gray, Tubes, Addison-Wesley Publishing Company, Advanced Book Program, Redwood City, CA, 1990. MR1044996 [11] R. C. Gunning and H. Rossi, Analytic functions of several complex variables, Prentice-Hall, Inc., Englewood Cliļ¬s, N.J., 1965. MR0180696 [12] P. Hartman, Ordinary diļ¬erential equations, John Wiley & Sons, Inc., New York-LondonSydney, 1964. MR0171038 [13] L. HĀØ ormander, Linear partial diļ¬erential operators, Die Grundlehren der mathematischen Wissenschaften, Bd. 116, Academic Press, Inc., Publishers, New York; Springer-Verlag, Berlin-GĀØ ottingen-Heidelberg, 1963. MR0161012 [14] L. HĀØ ormander, An introduction to complex analysis in several variables, D. Van Nostrand Co., Inc., Princeton, N.J.-Toronto, Ont.-London, 1966. MR0203075 [15] L. HĀØ ormander, The analysis of partial diļ¬erential operators. v. I, II, Springer, Berlin et al., 1983.
311
312
Bibliography
[16] F. John, Partial diļ¬erential equations, Springer, New York et al., 1986. [17] A. N. Kolmogorov and S. V. Fomin, Elements of the theory of functions and functional analysis, Dover Publications, Inc., Mineola, New York, 1999. [18] N. V. Krylov, Lectures on elliptic and parabolic equations in HĀØ older spaces, Graduate Studies in Mathematics, vol. 12, American Mathematical Society, Providence, RI, 1996. MR1406091 [19] O. A. Ladyzhenskaya and N. N. Uralātseva, Linear and quasilinear elliptic equations, translated from the Russian by Scripta Technica, Inc., translation editor: Leon Ehrenpreis, Academic Press, New York-London, 1968. MR0244627 [20] N. N. Lebedev, Special functions and their applications, revised edition, translated from the Russian and edited by Richard A. Silverman; unabridged and corrected republication, Dover Publications, Inc., New York, 1972. MR0350075 [21] E. H. Lieb and M. Loss, Analysis, 2nd ed., Graduate Studies in Mathematics, vol. 14, American Mathematical Society, Providence, RI, 2001. MR1817225 [22] J.-L. Lions and E. Magenes, Non-homogeneous boundary value problems and applications. Vol. I, translated from the French by P. Kenneth; Die Grundlehren der mathematischen Wissenschaften, Band 181, Springer-Verlag, New York-Heidelberg, 1972. MR0350177 [23] V. Palamodov, Topics in mathematical physics, Spring semester, preprint, 2002. [24] I. G. Petrovskii, Partial diļ¬erential equations, translated from the Russian by Scripta Technica, Ltd, Iliļ¬e Books Ltd., London (distributed by W. B. Saunders Co., Philadelphia, Pa.), 1967. MR0211021 [25] M. Reed and B. Simon, Methods of modern mathematical physics. I. Functional analysis, Academic Press, New York-London, 1972. MR0493419 [26] A. P. Robertson and W. J. Robertson, Topological vector spaces, Cambridge Tracts in Mathematics and Mathematical Physics, No. 53, Cambridge University Press, New York, 1964. MR0162118 [27] W. Rudin, Principles of mathematical analysis, 3rd ed., International Series in Pure and Applied Mathematics, McGraw-Hill Book Co., New York-Auckland-DĀØ usseldorf, 1976. MR0385023 [28] W. Rudin, Functional analysis, 2nd ed., International Series in Pure and Applied Mathematics, McGraw-Hill, Inc., New York, 1991. MR1157815 [29] H. H. Schaefer, Topological vector spaces, The Macmillan Co., New York; Collier-Macmillan Ltd., London, 1966. MR0193469 [30] L. Schwartz, Mathematics for the physical sciences, Hermann, Paris; Addison-Wesley Publishing Co., Reading, Mass.-London-Don Mills, Ont., 1966. MR0207494 [31] L. Schwartz. Analyse (French), Deuxi` eme partie: Topologie gĀ“ enĀ“ erale et analyse fonctionnelle; Collection Enseignement des Sciences, No. 11, Hermann, Paris, 1970. MR0467223 [32] G. E. Shilov, Generalized functions and partial diļ¬erential equations, authorized English edition revised by the author, translated and edited by Bernard D. Seckler, Gordon and Breach, Science Publishers, Inc., New York-London-Paris, 1968. MR0230129 [33] N. Shimakura, Partial diļ¬erential operators of elliptic type, translated and revised from the 1978 Japanese original by the author, Translations of Mathematical Monographs, vol. 99, American Mathematical Society, Providence, RI, 1992. MR1168472 [34] S. L. Sobolev and E. R. Dawson, Partial diļ¬erential equations of mathematical physics, translated from the third Russian edition by E. R. Dawson; English translation edited by T. A. A. Broadbent, Pergamon Press, Oxford-Edinburgh-New York-Paris-Frankfurt; AddisonWesley Publishing Co., Inc., Reading, Mass.-London, 1964. MR0178220 [35] M. E. Taylor, Partial diļ¬erential equations, Vol I, Basic theory; Vol. II, Qualitative studies of linear equations, Springer-Verlag, New York, 1996. [36] F. Treves, Topological vector spaces, distributions and kernels, Academic Press, Inc., (London, New York), LTD, third printing, 1970.
Bibliography
313
[37] V. S. Vladimirov, Equations of mathematical physics, translated from the Russian by Eugene Yankovsky [E. YankovskiĖı], āMirā, Moscow, 1984. MR764399 [38] F. W. Warner, Foundations of diļ¬erentiable manifolds and Lie groups, corrected reprint of the 1971 edition, Graduate Texts in Mathematics, vol. 94, Springer-Verlag, New York-Berlin, 1983. MR722297 [39] J. Wermer, Potential theory, 2nd ed., Lecture Notes in Mathematics, vol. 408, Springer, Berlin, 1981. MR634962 [40] E. T. Whittaker and G. N. Watson, A course of modern analysis: An introduction to the general theory of inļ¬nite processes and of analytic functions; with an account of the principal transcendental functions, reprint of the fourth (1927) edition, Cambridge Mathematical Library, Cambridge University Press, Cambridge, 1996. MR1424469
Index
C k boundary, 77 C k (Ī©), 77 C k,Ī³ , 141 C k,Ī³ boundary, 141 Cbk (Rn ), 173 D(Ī©) = C0ā (Ī©), 58 E(Ī©) = C ā (Ī©), 58 Ės (Ī©), 177 H H s (Rn ), 173 L2 (M, dĪ¼), 195 N, positive integers, xvii Qā , 144 S(Rn ), 58 T ā M , 224 Z+ , nonnegative integers, xvii Ī“-function, 53 Ī“-like sequence, 54, 67 acoustic wave, 215 action, 267 action along the path, 223 adjoint Cauchy problem, 165 adjoint operator, 194 asymptotic solution, 273 Bessel equation, 205 bicharacteristic, 263 boundary value problem Cauchy-Dirichletās, 153 Cauchy-Neumannās, 153 boundary value problem, ļ¬rst, 153 boundary value problem, second, 153 boundary, piecewise C k , 79
boundary, piecewise smooth, 79 bundle, cotangent, 12 bundle, tangent, 12 canonical form of a second-order operator, 5, 9, 10 Cauchy problem, 20, 153, 265 Cauchy problem for the Hamilton-Jacobi equation, 265 Cauchy sequence, 61 Cauchy-Dirichletās initial-boundary value problem, 153 Cauchy-Neumannās boundary value problem, 153 causality principle, 236 caustic, 262 characteristic hypersurface, 258 characteristic vector, 7 closed graph theorem, 115 compact operator, 177 completeness of distribution space, 67 conļ¬guration space, 222 conservation of momentum, 39 convergent sequences in D(Ī©), 63 convolution, 99 convolution of distributions, 104 convolution of two distributions, 124 cotangent bundle, 12, 224 cotangent vector, 12 countably normed space, 61 Courantās variational principle, 210 cylindric function of the second kind, 208
315
316
cylindric functions, 205 cylindric wave, 221 dāAlembertās formula, 20, 22 dāAlembertās principle, 17 dāAlembertian, 2, 219, 224 degrees of freedom, 39 dependence domain, 23 derivative, directional, 13 derivative, fractional, 125 descent method, 236 dielectric permeability, 217 diļ¬erential of a map, 13 diļ¬erential operator, linear, 2 diļ¬erential operator, order of, 2 diļ¬usion equation, 151 dipole, 103 dipole moment, 103 Diracās Ī“-function, 53 Diracās measure, 53 direct product of distributions, 101, 102 direct product of functions, 100 directional derivative, 13 Dirichlet kernel, 68 Dirichletās integral, 181 Dirichletās problem, 131, 154 dispersion law, 220 distribution, 57 distribution with compact support, 63 distribution, homogeneous, 89 distribution, regular, 64 distribution, tempered, 63 distributions, 63 divergence formula, 79 domain of the operator, 193 domain, dependence, 23 domain, inļ¬uence, 23 Dopplerās eļ¬ect, 41 double layer, 103 double layer potential, 107, 243 dual space, 62 Duhamelās formula, 163 eigenfrequencies, 29 eigenfunctions, 28 eigenvalues, 28 eikonal equation, 272 energy, 36, 224 energy conservation law, 19, 36, 37, 224 energy, full, 36 energy, kinetic, 36, 223 energy, kinetic along the path, 223
Index
energy, potential, 36, 37, 223 equation, diļ¬usion, 151 equation, Hamilton-Jacobi, 262 equation, parabolic, 151 equations, Lagrangeās, 17, 36 equations, Maxwell, 217 equations, transport, 274, 280 equicontinuity, 178 equivalent systems of seminorms, 59 Euler-Lagrange equations, 34, 35 extremal point, 34 Fejer kernel, 68 Fermatās principle, 278 Fermi coordinates, 249 ļ¬rst boundary value problem, 25 ļ¬rst initial-boundary value problem, 153 forced oscillations, 41 formula, dāAlembertās, 22 formula, Duhamelās, 163 formula, Kirchhoļ¬, 232 formula, Poissonās, 238 formulas, Sokhotsky, 70 Fourier transform, 100 Fourier transformation, 100 Fourierās law, 152 fractional derivative, 125 fractional integral, 124 FrĀ“echet space, 61 freedom, degrees of, 39 Friedrichs extension of an operator, 198, 199 Friedrichs inequality, 182 function, Bessel of the ļ¬rst kind, 207 function, cylindric of the ļ¬rst kind, 207 function, cylindric of the second kind, 208 function, Greenās, 50 function, harmonic, 114 function, Lagrange, 223 function, Neumannās, 208 function, subharmonic, 148 functional, 33 functions smooth up to G from each side, 244 fundamental solution of Laplace operator, 75 gauge transformation, 218 gauge, Lorentz, 219 Gaussās law of the arithmetic mean, 127 Gauss-Ostrogradsky formula, 79
Index
Gaussian integrals, 81 generalized function, 57 generalized solution, 11 generalized solution of Dirichletās problem, 185 generalized solutions of the Cauchy problem, 22 Glazmanās lemma, 209 gradient, 13 graph of the operator, 194 Greenās function, 50 Hadamardās example, 134 Hamilton-Jacobi equation, 262 Hamiltonian, 36, 224, 263 Hamiltonian curves, 263 Hamiltonian system, 225, 262 Hamiltonian vector ļ¬eld, 263 harmonic function, 114, 127 Hausdorļ¬ property, 60 heat conductivity, 152 heat equation, 151 heat operator, 2 Heaviside function, xvii Helmholtz operator, 97 high-frequency asymptotics, 47 HĀØ older function, 141 HĀØ olderās inequality, 92 Holmgrenās principle, 165 homogeneous distribution, 89 homothety, 89 Hookeās law, 16, 290 Huygens principle, 232 hyperbolicity of an operator, 269 hypoelliptic operator, 112, 114 identity, Parseval, 298 inequality, Friedrichs, 182 inequality, HĀØ olderās, 92 inļ¬uence domain, 23 initial value problem, 21 integral, Dirichlet, 181 integral, fractional, 124 integral, Poissonās, 157 inverse operator, 194 inversion, 146 involution, 146 involutive, 146 Jacobi matrix, 13 Kelvin transformation, 123 kernel, Dirichlet, 68
317
kernel, Fejer, 68 kinetic energy, 36, 223 kinetic energy along the path, 223 Kirchhoļ¬ formula, 232 L. Schwartzās space, 58 Lagrange function, 223 Lagrangeās equations, 17, 36 Lagrangian, 223 Lagrangian corresponding to the Hamiltonian, 267 Lagrangian manifold, 264 Laplace operator, 2 Laplacian, 2 law of dispersion, 220 law, conservation of energy, 224 law, Fourier, 152 law, Hookeās, 290 linear operator in Hilbert space, 193 Liouvilleās theorem for harmonic functions, 122 Lipschitz function, 141 logarithmic potential, 106 Lorentz gauge, 219 map, tangent, 13 maximum principle, 159, 169 maximum principle for harmonic functions, 129 Maxwell equations, 217 mean value theorem for harmonic functions, 127 method of descent, 236 mixed problem, 25 momentum, 36, 38 momentum, conservation of, 39 momentum, total, 38 monochromatic wave, 222 multi-index, 1 Neumannās function, 208 Neumannās problem, 154 Newtonās Second Law, 36 Newtonās Third Law, 38 nodal hypersurface, 146 norm, 58 operator semibounded from below, 198 operator, adjoint, 87, 194 operator, closed, 194 operator, compact, 177 operator, dual, 87 operator, elliptic, 7, 10
318
operator, extension of, 194 operator, heat, 2 operator, Helmholtz, 97 operator, hyperbolic, 7, 9 operator, hypoelliptic, 112, 114 operator, inverse, 194 operator, Laplace, 2 operator, of translation, 89 operator, one-dimensional SchrĀØ odinger, 45 operator, parabolic, 9 operator, selfadjoint, 194 operator, Sturm-Liouville, 2, 45 operator, symmetric, 194 operator, transposed, 87 operator, wave, 2, 219 order of a distribution, 84 order of diļ¬erential operator, 2 Parseval identity, 298 partition of unity, 70 path, 223 phase, 271 piecewise C k boundary, 79 piecewise smooth boundary, 79 plane wave, 220 Poisson problem, 136 Poissonās equation, 82 Poissonās formula, 238 Poissonās integral, 157 potential, 106 potential energy, 36, 37, 223 potential of a uniformly charged sphere, 253 potential of the double layer of the sphere, 254 potential, double layer, 243 potential, retarded, 235 potential, single layer, 243 precompact, 177 principle of maximum, 169 principle, causality, 236 principle, Courantās variational, 210 principle, dāAlembertās, 17 principle, Holmgrenās, 165 principle, Huygens, 232 principle, maximum, 159 problem, adjoint Cauchy, 165 problem, Cauchy, 20 problem, ļ¬rst boundary value, 25 problem, initial value, 21 problem, mixed, 25
Index
problem, Sturm-Liouville, 28, 44, 45 rapidly oscillating solution, 273 ray, 261, 265, 271 real analytic function, 110 reļ¬ection with respect to a sphere, 146 regular distribution, 64 removable singularity theorem, 114 renormalization, 83 retarded potential, 235 SchrĀØ odinger operator, one-dimensional, 45 Schwartzās kernel, 50 seminorm, 58 seminorms, equivalent systems of, 59 sheaf, 70 short-wave asymptotics, 47 simple layer potential, 107 single layer, 103 single layer potential, 243 singular support of a distribution, 109 Sobolev space, 171 Sobolevās embedding theorem, 173 Sobolevās theorem on restrictions, 175 Sokhotskyās formulas, 70 solution asymptotic, 273 solution rapidly oscillating, 273 solution, generalized, 11 space, countably normed, 61 space, dual, 62 space, FrĀ“echet, 61 space, Sobolev, 171 speciļ¬c heat, 152 spectral theorem, 197 speed of light, 217 spherical waves, 220 standing waves, 26 string, unbounded, 20 Sturm-Liouville operator, 2, 45 Sturm-Liouville problem, 28, 44, 45 subharmonic function, 148 superposition, 11, 29 support, 58 support of the distribution, 71 surface of jumps, 258 surface, characteristic, 7 surface, characteristic at a point, 7 symbol, 2 symbol, principal, 2 symbol, total, 2
Index
tangent bundle, 12, 222 tangent map, 13 tangent space, 222 tangent vectors, 12, 223 tempered distribution, 63 tensor product of distributions, 101, 102 tensor product of functions, 100 theorem, Liouville, 122 theorem, mean value, for harmonic functions, 127 theorem, Sobolevās on embedding, 173 theorem, Sobolevās on restrictions, 175 theorem, spectral, 197 theorem, Tikhonov, 166 transformation, Kelvin, 123 transport equations, 274, 280 uniform boundedness, 178 unitarily equivalent operators, 197 variation of a functional, 34 variational derivative, 34 variational derivative, partial, 35 vector, characteristic, 7 vector, cotangent, 12 vector, tangent, 12 vector, wave, 220 velocity, 223 volume potential, 242 wave equation, the nonresonance case, 32 wave equation, the resonance case, 32 wave front, 261, 265, 271 wave operator, 2, 219 wave vector, 220 wave, cylindric, 221 wave, monochromatic, 222 waves standing, 26 waves, spherical, 220 weak topology, 67 well-posedness, 132
319
Selected Published Titles in This Series 205 204 203 202
Mikhail Shubin, Invitation to Partial Diļ¬erential Equations, 2020 Sarah J. Witherspoon, Hochschild Cohomology for Algebras, 2019 Dimitris Koukoulopoulos, The Distribution of Prime Numbers, 2019 Michael E. Taylor, Introduction to Complex Analysis, 2019
201 Dan A. Lee, Geometric Relativity, 2019 200 Semyon Dyatlov and Maciej Zworski, Mathematical Theory of Scattering Resonances, 2019 199 Weinan E, Tiejun Li, and Eric Vanden-Eijnden, Applied Stochastic Analysis, 2019 198 Robert L. Benedetto, Dynamics in One Non-Archimedean Variable, 2019 197 196 195 194
Walter Craig, A Course on Partial Diļ¬erential Equations, 2018 Martin Stynes and David Stynes, Convection-Diļ¬usion Problems, 2018 Matthias Beck and Raman Sanyal, Combinatorial Reciprocity Theorems, 2018 Seth Sullivant, Algebraic Statistics, 2018
193 192 191 190
Martin Lorenz, A Tour of Representation Theory, 2018 Tai-Peng Tsai, Lectures on Navier-Stokes Equations, 2018 Theo BĀØ uhler and Dietmar A. Salamon, Functional Analysis, 2018 Xiang-dong Hou, Lectures on Finite Fields, 2018
189 188 187 186
I. Martin Isaacs, Characters of Solvable Groups, 2018 Steven Dale Cutkosky, Introduction to Algebraic Geometry, 2018 John Douglas Moore, Introduction to Global Analysis, 2017 Bjorn Poonen, Rational Points on Varieties, 2017
185 Douglas J. LaFountain and William W. Menasco, Braid Foliations in Low-Dimensional Topology, 2017 184 Harm Derksen and Jerzy Weyman, An Introduction to Quiver Representations, 2017 183 Timothy J. Ford, Separable Algebras, 2017 182 181 180 179
Guido Schneider and Hannes Uecker, Nonlinear PDEs, 2017 Giovanni Leoni, A First Course in Sobolev Spaces, Second Edition, 2017 Joseph J. Rotman, Advanced Modern Algebra: Third Edition, Part 2, 2017 Henri Cohen and Fredrik StrĀØ omberg, Modular Forms, 2017
178 Jeanne N. Clelland, From Frenet to Cartan: The Method of Moving Frames, 2017 177 Jacques Sauloy, Diļ¬erential Galois Theory through Riemann-Hilbert Correspondence, 2016 176 Adam Clay and Dale Rolfsen, Ordered Groups and Topology, 2016 175 Thomas A. Ivey and Joseph M. Landsberg, Cartan for Beginners: Diļ¬erential Geometry via Moving Frames and Exterior Diļ¬erential Systems, Second Edition, 2016 174 Alexander Kirillov Jr., Quiver Representations and Quiver Varieties, 2016 173 Lan Wen, Diļ¬erentiable Dynamical Systems, 2016 172 Jinho Baik, Percy Deift, and Touļ¬c Suidan, Combinatorics and Random Matrix Theory, 2016 171 Qing Han, Nonlinear Elliptic Equations of the Second Order, 2016 170 169 168 167
Donald Yau, Colored Operads, 2016 AndrĀ“ as Vasy, Partial Diļ¬erential Equations, 2015 Michael Aizenman and Simone Warzel, Random Operators, 2015 John C. Neu, Singular Perturbation in the Physical Sciences, 2015
166 Alberto Torchinsky, Problems in Real and Functional Analysis, 2015
For a complete list of titles in this series, visit the AMS Bookstore at www.ams.org/bookstore/gsmseries/.
It is a great pleasure to see this bookāwritten by a great master of the subjectāfinally in print. This treatment of a core part of mathematics and its applications offers the student both a solid foundation in basic calculations techniques in the subject, as well as a basic introduction to the more general machinery, e.g., distributions, Sobolev spaces, etc., which are such a key part of any modern treatment. As such this book is ideal for more advanced undergraduates as well as mathematically inclined students from engineering or the natural sciences. Shubin has a lovely intuitive writing style which provides a gentle introduction to this beautiful subject. Many good exercises (and solutions) are provided! āRafe Mazzeo, Stanford University This text provides an excellent semesterās introduction to classical and modern topics in linear PDE, suitable for students with a background in advanced calculus and Lebesgue integration. The author intersperses treatments of the Laplace, heat, and wave equations with developments of various functional analytic tools, particularly distribution theory and spectral theory, introducing key concepts while deftly avoiding heavy technicalities. āMichael Taylor, University of North Carolina, Chapel Hill
For additional information and updates on this book, visit www.ams.org/bookpages/gsm-205
GSM/205 www.ams.org
Mikhail Shubin photo Ā© 2020 Galina Ester Shubina. Photography by Kate Gabor
This book is based on notes from a beginning graduate course on partial differential equations. Prerequisites for using the book are a solid undergraduate course in real analysis. There are more than 100 exercises in the book. Some of them are just exercises, whereas others, even though they may require new ideas to solve them, provide additional important information about the subject.