Sie sind auf Seite 1von 305

A Basic Course in

Partial Differential
Equations

Qing Han

Graduate Studies
in Mathematics
Volume 120

American Mathematical Society


A Basic Course in
Partial Differential
Equations
A Basic Course in
Partial Differential
Equations

Qing Han

Graduate Studies
in Mathematics
Volume 120

American Mathematical Society


Providence, Rhode Island
EDITORIAL COMMITTEE
David Cox (Chair)
Rafe Mazzeo
Martin Scharlemann
Gigliola Stalani

2000 Mathematics Subject Classication. Primary 3501.

For additional information and updates on this book, visit


www.ams.org/bookpages/gsm-120

Library of Congress Cataloging-in-Publication Data


Han, Qing.
A basic course in partial dierential equations / Qing Han.
p. cm. (Graduate studies in mathematics ; v. 120)
Includes bibliographical references and index.
ISBN 978-0-8218-5255-2 (alk. paper)
1. Dierential equations, Partial. I. Title.

QA377.H31819 2010
515. 353dc22
2010043189

Copying and reprinting. Individual readers of this publication, and nonprot libraries
acting for them, are permitted to make fair use of the material, such as to copy a chapter 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 such
permission should be addressed to the Acquisitions Department, American Mathematical Society,
201 Charles Street, Providence, Rhode Island 02904-2294 USA. Requests can also be made by
e-mail to reprint-permission@ams.org.
c 2011 by the author.
The American Mathematical Society retains all rights
except those granted to the United States Government.
Printed in the United States of America.

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 http://www.ams.org/
10 9 8 7 6 5 4 3 2 1 16 15 14 13 12 11
To Yansu, Raymond and Tommy
Contents

Preface ix
Chapter 1. Introduction 1
1.1. Notation 1
1.2. Well-Posed Problems 3
1.3. Overview 5

Chapter 2. First-Order Dierential Equations 9


2.1. Noncharacteristic Hypersurfaces 10
2.2. The Method of Characteristics 16
2.3. A Priori Estimates 30
2.4. Exercises 43

Chapter 3. An Overview of Second-Order PDEs 47


3.1. Classications 48
3.2. Energy Estimates 58
3.3. Separation of Variables 67
3.4. Exercises 86

Chapter 4. Laplace Equations 89


4.1. Fundamental Solutions 90
4.2. Mean-Value Properties 105
4.3. The Maximum Principle 112
4.4. Poisson Equations 133
4.5. Exercises 143

vii
viii Contents

Chapter 5. Heat Equations 147


5.1. Fourier Transforms 148
5.2. Fundamental Solutions 158
5.3. The Maximum Principle 175
5.4. Exercises 197
Chapter 6. Wave Equations 201
6.1. One-Dimensional Wave Equations 202
6.2. Higher-Dimensional Wave Equations 213
6.3. Energy Estimates 237
6.4. Exercises 245
Chapter 7. First-Order Dierential Systems 249
7.1. Noncharacteristic Hypersurfaces 250
7.2. Analytic Solutions 259
7.3. Nonexistence of Smooth Solutions 270
7.4. Exercises 276
Chapter 8. Epilogue 279
8.1. Basic Linear Dierential Equations 279
8.2. Examples of Nonlinear Dierential Equations 282
Bibliography 289
Index 291
Preface

Is it really necessary to classify partial dierential equations (PDEs) and to


employ dierent methods to discuss dierent types of equations? Why is it
important to derive a priori estimates of solutions before even proving the
existence of solutions? These are only a few questions any students who
just start studying PDEs might ask. Students may nd answers to these
questions only at the end of a one-semester course in basic PDEs, sometimes
after they have already lost interest in the subject. In this book, we attempt
to address these issues at the beginning. There are several notable features
in this book.
First, the importance of a priori estimates is addressed at the beginning
and emphasized throughout this book. This is well illustrated by the chapter
on rst-order PDEs. Although rst-order linear PDEs can be solved by
the method of characteristics, we provide a detailed analysis of a priori
estimates of solutions in sup-norms and in integral norms. To emphasize the
importance of these estimates, we demonstrate how to prove the existence
of weak solutions with the help of basic results from functional analysis.
The setting here is easy, since L2 -spaces are needed only. Meanwhile, all
important ideas are in full display. In this book, we do attempt to derive
explicit expressions for solutions whenever possible. However, these explicit
expressions of solutions of special equations usually serve mostly to suggest
the correct form of estimates for solutions of general equations.
The second feature is the illustration of the necessity to classify second-
order PDEs at the beginning. In the chapter on general second-order linear
PDEs, immediately after classifying second-order PDEs into elliptic, para-
bolic and hyperbolic type, we discuss various boundary-value problems and
initial/boundary-value problems for the Laplace equation, the heat equation

ix
x Preface

and the wave equation. We discuss energy methods for proving uniqueness
and nd solutions in the plane by separation of variables. The explicit
expressions of solutions demonstrate dierent properties of solutions of dif-
ferent types of PDEs. Such dierences clearly indicate that there is unlikely
to be a unied approach to studying PDEs.
Third, we focus on simple models of PDEs and study these equations in
detail. We have chapters devoted to the Laplace equation, the heat equation
and the wave equation, and use several methods to study each equation.
For example, for the Laplace equation, we use three dierent methods to
study its solutions: the fundamental solution, the mean-value property and
the maximum principle. For each method, we indicate its advantages and
its shortcomings. General equations are not forgotten. We also discuss
maximum principles for general elliptic and parabolic equations and energy
estimates for general hyperbolic equations.
The book is designed for a one-semester course at the graduate level.
Attempts have been made to give a balanced coverage of dierent classes
of partial dierential equations. The choice of topics is inuenced by the
personal tastes of the author. Some topics may not be viewed as basic by
others. Among those not found in PDE textbooks at a comparable level
are estimates in L -norms and L2 -norms of solutions of the initial-value
problem for the rst-order linear dierential equations, interior gradient es-
timates and dierential Harnack inequality for the Laplace equation and the
heat equation by the maximum principle, and decay estimates for solutions
of the wave equation. Inclusions of these topics reect the emphasis on
estimates in this book.
This book is based on one-semester courses the author taught at the Uni-
versity of Notre Dame in the falls of 2007, 2008 and 2009. During the writing
of the book, the author benetted greatly from comments and suggestions of
many of his friends, colleagues and students in his classes. Tiancong Chen,
Yen-Chang Huang, Gang Li, Yuanwei Qi and Wei Zhu read the manuscript
at various stages. Minchun Hong, Marcus Khuri, Ronghua Pan, Xiaodong
Wang and Xiao Zhang helped the author write part of Chapter 8. Hairong
Liu did a wonderful job of typing an early version of the manuscript. Special
thanks go to Charles Stanton for reading the entire manuscript carefully and
for many suggested improvements.
I am grateful to Natalya Pluzhnikov, my editor at the American Math-
ematical Society, for reading the manuscript and guiding the eort to turn
it into a book. Last but not least, I thank Edward Dunne at the AMS for
his help in bringing the book to press.

Qing Han
Chapter 1

Introduction

This chapter serves as an introduction of the entire book.


In Section 1.1, we rst list several notations we will use throughout this
book. Then, we introduce the concept of partial dierential equations.
In Section 1.2, we discuss briey well-posed problems for partial dier-
ential equations. We also introduce several function spaces whose associated
norms are used frequently in this book.
In Section 1.3, we present an overview of this book.

1.1. Notation
In general, we denote by x points in Rn and write x = (x1 , , xn ) in terms
of its coordinates. For any x Rn , we denote by |x| the standard Euclidean
norm, unless otherwise stated. Namely, for any x = (x1 , , xn ), we have
 1

n 2

|x| = x2i .
i=1

Sometimes, we need to distinguish one particular direction as the time di-


rection and write points in Rn+1 as (x, t) for x Rn and t R. In this
case, we call x = (x1 , , xn ) Rn the space variable and t R the time
variable. In R2 , we also denote points by (x, y).
Let be a domain in Rn , that is, an open and connected subset in
Rn . We denote by C() the collection of all continuous functions in , by
C m () the collection of all functions with continuous derivatives up to order
m, for any integer m 1, and by C () the collection of all functions with
continuous derivatives of arbitrary order. For any u C m (), we denote by

1
2 1. Introduction

m u the collection of all partial derivatives of u of order m. For m = 1 and


m = 2, we usually write m u in special forms. For rst-order derivatives,
we write u as a vector of the form
u = (ux1 , , uxn ).
This is the gradient vector of u. For second-order derivatives, we write 2 u
in the matrix form

ux1 x1 ux1 x2 ux1 xn
ux x ux x u x x
2 1 2 2 2 n
2 u = . .. . .. .
. . . . . .
uxn x1 uxn x2 uxn xn
This is a symmetric matrix, called the Hessian matrix of u. For derivatives of
order higher than two, we need to use multi-indices. A multi-index Zn+
is given by = (1 , , n ) with nonnegative integers 1 , , n . We write

n
|| = i .
i=1

For any vector = (1 , , n ) Rn , we denote


= 11 nn .
The partial derivative u is dened by
u = x11 xnn u,
and its order is ||. For any positive integer m, we dene
1

2

| u| =
m 2
| u| .
||=m

In particular,
 1

n 2

|u| = u2xi ,
i=1
and
1

n 2

| u| =
2
u2xi xj .
i,j=1

A hypersurface in Rn
is a surface of dimension n 1. Locally, a C m -
hypersurface can be expressed by { = 0} for a C m -function with = 0.
Alternatively, by a rotation, we may take (x) = xn (x1 , , xn1 ) for a
C m -function of n 1 variables. A domain Rn is C m if its boundary
is a C m -hypersurface.
1.2. Well-Posed Problems 3

A partial dierential equation (henceforth abbreviated as PDE) in a


domain Rn is a relation of independent variables x , an unknown
function u dened in , and a nite number of its partial derivatives. To
solve a PDE is to nd this unknown function. The order of a PDE is the
order of the highest derivative in the relation. Hence for a positive integer
m, the general form of an mth-order PDE in a domain Rn is given by

F x, u, u(x), 2 u(x), , m u(x) = 0 for x .
Here F is a function which is continuous in all its arguments, and u is a
C m -function in . A C m -solution u satisfying the above equation in the
pointwise sense in is often called a classical solution. Sometimes, we need
to relax regularity requirements for solutions when classical solutions are
not known to exist. Instead of going into details, we only mention that it
is an important method to establish rst the existence of weak solutions,
functions with less regularity than C m and satisfying the equation in some
weak sense, and then to prove that these weak solutions actually possess the
required regularity to be classical solutions.
A PDE is linear if it is linear in the unknown functions and their deriva-
tives, with coecients depending on independent variables x. A general
mth-order linear PDE in is given by

a (x) u = f (x) for x .
||m

Here a is the coecient of u and f is the nonhomogeneous term of the


equation. A PDE of order m is quasilinear if it is linear in the derivatives of
solutions of order m, with coecients depending on independent variables
x and the derivatives of solutions of order < m. In general, an mth-order
quasilinear PDE in is given by

a (x, u, , m1 u) u = f (x, u, , m1 u) for x .
||=m

Several PDEs involving one or more unknown functions and their deriva-
tives form a partial dierential system. We dene linear and quasilinear
partial dierential systems accordingly.
In this book, we will focus on rst-order and second-order linear PDEs
and rst-order linear dierential systems. On a few occasions, we will diverge
to nonlinear PDEs.

1.2. Well-Posed Problems


What is the meaning of solving partial dierential equations? Ideally, we
obtain explicit solutions in terms of elementary functions. In practice this is
only possible for very simple PDEs or very simple solutions of more general
4 1. Introduction

PDEs. In general, it is impossible to nd explicit expressions of all solutions


of all PDEs. In the absence of explicit solutions, we need to seek methods to
prove existence of solutions of PDEs and discuss properties of these solutions.
In many PDE problems, this is all we need to do.
A given PDE may not have solutions at all or may have many solutions.
When it has many solutions, we intend to assign extra conditions to pick up
the most relevant solutions. Those extra conditions usually are in the form
of boundary values or initial values. For example, when we consider a PDE
in a domain, we can require that solutions, when restricted to the boundary,
have prescribed values. This is the so-called boundary-value problems. When
one variable is identied as the time and a part of the boundary is identied
as an initial hypersurface, values prescribed there are called initial values.
We use data to refer to boundary values or initial values and certain known
functions in the equation, such as the nonhomogeneous term if the equation
is linear.
Hadamard introduced the notion of well-posed problems. A given prob-
lem for a partial dierential equation is well-posed if
(i) there is a solution;
(ii) this solution is unique;
(iii) the solution depends continuously in some suitable sense on the data
given in the problem, i.e., the solution changes by a small amount if the data
change by a small amount.
We usually refer to (i), (ii) and (iii) as the existence, uniqueness and
continuous dependence, respectively. We need to emphasize that the well-
posedness goes beyond the existence and uniqueness of solutions. The con-
tinuous dependence is particularly important when PDEs are used to model
phenomena in the natural world. This is because measurements are always
associated with errors. The model can make useful predictions only if solu-
tions depend on data in a controllable way.
In practice, both the uniqueness and the continuous dependence are
proved by a priori estimates. Namely, we assume solutions already exist
and then derive certain norms of solutions in terms of data in the problem.
It is important to note that establishing a priori estimates is in fact the rst
step in proving the existence of solutions. A closely related issue here is the
regularity of solutions such as continuity and dierentiability. Solutions of
a particular PDE can only be obtained if the right kind of regularity, or the
right kind of norms, are employed. Two classes of norms are used often,
sup-norms and L2 -norms.
1.3. Overview 5

Let be a domain in Rn . For any bounded function u in , we dene


the sup-norm of u in by
|u|L () = sup |u|.

For a bounded continuous function u in , we may also write |u|C() instead


of |u|L () . Let m be a positive integer. For any function u in with
bounded derivatives up to order m, we dene the C m -norm of u in by

|u|C m () = | u|L () .
||m

If is a bounded C m -domain in Rn , then C m (), the collection of functions


which are C m in , is a Banach space equipped with the C m -norm.
Next, for any Lebesgue measurable function u in , we dene the L2 -
norm of u in by
 1
2
uL2 () = 2
u dx ,

where integration is in the Lebesgue sense. The L2 -space in is the collec-
tion of all Lebesgue measurable functions in with nite L2 -norms and is
denoted by L2 (). We learned from real analysis that L2 () is a Banach
space equipped with the L2 -norm.
Other norms will also be used. We will introduce them as needed.
The basic formula for integration is the formula of integration by parts.
Let be a piecewise C 1 -domain in Rn and = (1 , , n ) be the unit
exterior normal vector to . Then for any u, v C 1 () C(),
  
uxi v dx = uvxi dx + uvi dS,

for i = 1, , n. Such a formula is the basis for L2 -estimates.


In deriving a priori estimates, we follow a common practice and use
the variable constant convention. The same letter C is used to denote
constants which may change from line to line, as long as it is clear from the
context on what quantities the constants depend. In most cases, we are not
interested in the value of the constant, but only in its existence.

1.3. Overview
There are eight chapters in this book.
The main topic in Chapter 2 is rst-order PDEs. In Section 2.1, we in-
troduce the basic notion of noncharacteristic hypersurfaces for initial-value
problems for rst-order PDEs. We discuss rst-order linear PDEs, quasilin-
ear PDEs and general nonlinear PDEs. In Section 2.2, we solve initial-value
6 1. Introduction

problems by the method of characteristics if initial values are prescribed on


noncharacteristic hypersurfaces. We demonstrate that solutions of a system
of ordinary dierential equations (ODEs) yield solutions of the initial-value
problems for rst-order PDEs. In Section 2.3, we derive estimates of solu-
tions of initial-value problems for rst-order linear PDEs. The L -norms
and the L2 -norms of solutions are estimated in terms of those of initial val-
ues and nonhomogeneous terms. In doing so, we only assume the existence
of solutions and do not use any explicit expressions of solutions. These
estimates provide quantitative properties of solutions.
Chapter 3 should be considered as an introduction to the theory of
second-order linear PDEs. In Section 3.1, we introduce the Laplace equa-
tion, the heat equation and the wave equation. We also introduce their
general forms, elliptic equations, parabolic equations and hyperbolic equa-
tions, which will be studied in detail in subsequent chapters. In Section 3.2,
we derive energy estimates of solutions of certain boundary-value problems.
Consequences of such energy estimates are the uniqueness of solutions and
the continuous dependence of solutions on boundary values and nonhomo-
geneous terms. In Section 3.3, we solve these boundary-value problems in
the plane by separation of variables. Our main focus is to demonstrate dif-
ferent regularity patterns for solutions of dierent dierential equations, the
Laplace equation, the heat equation and the wave equation.
In Chapter 4, we discuss the Laplace equation and the Poisson equa-
tion. The Laplace equation is probably the most important PDE with the
widest range of applications. In the rst three sections, we study harmonic
functions (i.e., solutions of the Laplace equation), by three dierent meth-
ods: the fundamental solution, the mean-value property and the maximum
principle. These three sections are relatively independent of each other.
In Section 4.1, we solve the Dirichlet problem for the Laplace equation in
balls and derive Poisson integral formula. Then we discuss regularity of har-
monic functions using the fundamental solution. In Section 4.2, we study
the mean-value property of harmonic functions and its consequences. In
Section 4.3, we discuss the maximum principle for harmonic functions and
its applications. In particular, we use the maximum principle to derive in-
terior gradient estimates for harmonic functions and the Harnack inequality
for positive harmonic functions. We also solve the Dirichlet problem for the
Laplace equation in a large class of bounded domains by Perrons method.
Last in Section 4.4, we briey discuss classical solutions and weak solutions
of the Poisson equation.
In Chapter 5, we study the heat equation, which describes the temper-
ature of a body conducting heat, when the density is constant. In Section
5.1, we introduce Fourier transforms briey and derive formally an explicit
1.3. Overview 7

expression for solutions of the initial-value problem for the heat equation.
In Section 5.2, we prove that such an expression indeed yields a classical
solution under appropriate assumptions on initial values. We also discuss
regularity of arbitrary solutions of the heat equation by the fundamental
solution. In Section 5.3, we discuss the maximum principle for the heat
equation and its applications. In particular, we use the maximum principle
to derive interior gradient estimates for solutions of the heat equation and
the Harnack inequality for positive solutions of the heat equation.
In Chapter 6, we study the n-dimensional wave equation, which rep-
resents vibrations of strings or propagation of sound waves in tubes for
n = 1, waves on the surface of shallow water for n = 2, and acoustic or
light waves for n = 3. In Section 6.1, we discuss initial-value problems and
various initial/boundary-value problems for the one-dimensional wave equa-
tion. In Section 6.2, we study initial-value problems for the wave equation
in higher-dimensional spaces. We derive explicit expressions of solutions in
odd dimensions by the method of spherical average and in even dimensions
by the method of descent. We also discuss global behaviors of solutions.
Then in Section 6.3, we derive energy estimates for solutions of initial-value
problems. Chapter 6 is relatively independent of Chapter 4 and Chapter 5
and can be taught after Chapter 3.
In Chapter 7, we discuss partial dierential systems of rst order and
focus on existence of local solutions. In Section 7.1, we introduce non-
characteristic hypersurfaces for partial dierential equations and systems of
arbitrary order. We demonstrate that partial dierential systems of arbi-
trary order can always be changed to those of rst order. In Section 7.2,
we discuss the Cauchy-Kovalevskaya theorem, which asserts the existence of
analytic solutions of noncharacteristic initial-value problems for dierential
systems if all data are analytic. In Section 7.3, we construct a rst-order
linear dierential system in R3 which does not admit smooth solutions in
any subsets of R3 . In this system, coecient matrices are analytic and the
nonhomogeneous term is a suitably chosen smooth function.
In Chapter 8, we discuss several dierential equations we expect to study
in more advanced PDE courses. Discussions in this chapter will be brief.
In Section 8.1, we discuss basic second-order linear dierential equations,
including elliptic, parabolic and hyperbolic equations, and rst-order linear
symmetric hyperbolic dierential systems. We will introduce appropriate
boundary-value problems and initial-value problems and introduce appro-
priate function spaces to study these problems. In Section 8.2, we introduce
several important nonlinear equations and focus on their background. This
chapter is designed to be introductory.
8 1. Introduction

Each chapter, except this introduction and the nal chapter, ends with
exercises. Level of diculty varies considerably. Some exercises, at the most
dicult level, may require long lasting eorts.
Chapter 2

First-Order Dierential
Equations

In this chapter, we discuss initial-value problems for rst-order PDEs. Main


topics include noncharacteristic conditions, methods of characteristics and
a priori estimates in L -norms and in L2 -norms.
In Section 2.1, we introduce the basic notion of noncharacteristic hyper-
surfaces for initial-value problems. In an attempt to solve initial-value prob-
lems, we illustrate that we are able to compute all derivatives of solutions
on initial hypersurfaces if initial values are prescribed on noncharacteris-
tic initial hypersurfaces. For rst-order linear PDEs, the noncharacteristic
condition is determined by equations and initial hypersurfaces, independent
of initial values. However, for rst-order nonlinear equations, initial values
also play a role. Noncharacteristic conditions will also be introduced for
second-order linear PDEs in Section 3.1 and for linear PDEs of arbitrary
order in Section 7.1, where multi-indices will be needed.
In Section 2.2, we solve initial-value problems by the method of charac-
teristics if initial values are prescribed on noncharacteristic hypersurfaces.
For rst-order homogeneous linear PDEs, special curves are introduced along
which solutions are constant. These curves are given by solutions of a system
of ordinary dierential equations (ODEs), the so-called characteristic ODEs.
For nonlinear PDEs, characteristic ODEs also include additional equations
for solutions of PDEs and their derivatives. Solutions of the characteristic
ODEs yield solutions of the initial-value problems for rst-order PDEs.
In Section 2.3, we derive estimates of solutions of initial-value problems
for rst-order linear PDEs. The L -norms and the L2 -norms of solutions

9
10 2. First-Order Dierential Equations

are estimated in terms of those of initial values and nonhomogeneous terms.


In doing so, we only assume the existence of solutions and do not use any
explicit expressions of solutions. These estimates provide quantitative prop-
erties of solutions. In the nal part of this section, we discuss briey the ex-
istence of weak solutions as a consequence of the L2 -estimates. The method
is from functional analysis and the Riesz representation theorem plays an
essential role.

2.1. Noncharacteristic Hypersurfaces


Let be a domain in Rn and F = F (x, u, p) be a smooth function of
(x, u, p) R Rn . A rst-order PDE in is given by

(2.1.1) F x, u, u = 0 for x .
Solving (2.1.1) in the classical sense means nding a smooth function u
satisfying (2.1.1) in . We rst examine a simple example.
Example 2.1.1. We consider in R2 = {(x, t)} the equation
ux + ut = 0.
This is probably the simplest rst-order PDE. Obviously, u(x, t) = x t
is a solution. In general, u(x, t) = u0 (x t) is also a solution for any C 1 -
function u0 . Such a solution has a physical interpretation. We note that
u(x, t) = u0 (xt) is constant along straight lines xt = x0 . By interpreting
x as location and t as time, we can visualize such a solution as a wave
propagating to the right with velocity 1 without changing shape. When
interpreted in this way, the solution u at later time (t > 0) is determined
uniquely by its value at the initial time (t = 0), which is given by u0 (x).
The function u0 is called an initial value.

u
6 u(, t1 ) u(, t2 )

-
 - x
t2 t1

Figure 2.1.1. Graphs of u at dierent times t2 > t1 .


2.1. Noncharacteristic Hypersurfaces 11

In light of Example 2.1.1, we will introduce initial values for (2.1.1) and
discuss whether initial values determine solutions.
Let be a smooth hypersurface in Rn with = . We intend to
prescribe u on to nd a solution of (2.1.1). To be specic, let u0 be a given
smooth function on . We will nd a solution u of (2.1.1) also satisfying
(2.1.2) u = u0 on .
We usually call the initial hypersurface and u0 the initial value or Cauchy
value. The problem of solving (2.1.1) together with (2.1.2) is called the
initial-value problem or Cauchy problem. Our main focus is to solve such an
initial-value problem under appropriate conditions.
We start with the following question. Given an initial value (2.1.2) for
equation (2.1.1), can we compute all derivatives of u at each point of the
initial hypersurface ? This should be easier than solving the initial-value
problem (2.1.1)(2.1.2).
To illustrate the main ideas, we rst consider linear PDEs. Let be a
domain in Rn containing the origin and ai , b and f be smooth functions in
, for any i = 1, , n. We consider

n
(2.1.3) ai (x)uxi + b(x)u = f (x) in .
i=1
Here, ai and b are coecients of uxi and u, respectively. The function f is
called the nonhomogeneous term. If f 0, (2.1.3) is called a homogeneous
equation.
We rst consider a special case where the initial hypersurface is given
by the hyperplane {xn = 0}. For x Rn , we write x = (x , xn ) for x =
(x1 , , xn1 ) Rn1 . Let u0 be a given smooth function in a neighborhood
of the origin in Rn1 . The initial condition (2.1.2) has the form
(2.1.4) u(x , 0) = u0 (x ),
for any x Rn1 suciently small.
Let u be a smooth solution of (2.1.3) and (2.1.4). In the following, we
will investigate whether we can compute all derivatives of u at the origin in
terms of the equation and the initial value. It is obvious that we can nd
all x -derivatives of u at the origin in terms of those of u0 . In particular, we
have, for i = 1, , n 1,
uxi (0) = u0,xi (0).
To nd uxn (0), we need to use the equation. We note that an is the coecient
of uxn in (2.1.3). If we assume
(2.1.5) an (0) = 0,
12 2. First-Order Dierential Equations

then by (2.1.3)
n1 
1 
uxn (0) = ai (0)uxi (0) + b(0)u(0) f (0) .
an (0)
i=1

Hence, we can compute all rst-order derivatives of u at 0 in terms of the


coecients and the nonhomogeneous term in (2.1.3) and the initial value u0
in (2.1.4). In fact, we can compute all derivatives of u of any order at the
origin by using u0 and dierentiating (2.1.3). We illustrate this by nding
all the second-order derivatives. We rst note that

uxi xj (0) = u0,xi xj (0),

for i, j = 1, , n 1. To nd uxk xn for k = 1, , n, we dierentiate (2.1.3)


with respect to xk to get

n 
n
ai uxi xk + ai,xk uxi + buxk + bxk u = fxk .
i=1 i=1

For k = 1, , n 1, the only unknown expression at the origin is uxk xn ,


whose coecient is an . If (2.1.5) holds, we can nd uxk xn (0) for k =
1, , n1. For k = n, with uxi xn (0) already determined for i = 1, , n1,
we can nd uxn xn (0) similarly. This process can be repeated for derivatives
of arbitrary order. In summary, we can nd all derivatives of u of any or-
der at the origin under the condition (2.1.5), which will be dened as the
noncharacteristic condition later on.
More generally, consider a hypersurface given by { = 0} for a smooth
function in a neighborhood of the origin with = 0. Assume that
passes through the origin, i.e., (0) = 0. We note that is normal to
at each point of . Without loss of generality, we assume xn (0) = 0.
Then by the implicit function theorem, we can solve = 0 around x = 0
for xn = (x1 , , xn1 ). Consider a change of variables

x y = x1 , , xn1 , (x) .

This is a well-dened transform in a neighborhood of the origin. Its Jacobian


matrix J is given by

0
(y1 , , yn ) ..
Id .
J= = .
(x1 , , xn ) 0
x1 xn1 xn

Hence det J(0) = xn (0) = 0.


2.1. Noncharacteristic Hypersurfaces 13

In the following, we denote by L the rst-order linear dierential oper-


ator dened by the left-hand side of (2.1.3), i.e.,

n
(2.1.6) Lu = ai (x)uxi + b(x)u.
i=1
By the chain rule,

n
uxi = yk,xi uyk .
k=1
We write the operator L in the y-coordinates as
 n 
n 
Lu = ai (x(y))yk,xi uyk + b(x(y))u.
k=1 i=1
In the y-coordinates, the initial hypersurface is given by {yn = 0}. With
yn = , the coecient of uyn is given by

n
ai (x)xi .
i=1
Hence, for the initial-value problem (2.1.3) and (2.1.2), we can nd all deriva-
tives of u at 0 if
n
ai (0)xi (0) = 0.
i=1
We recall that = (x1 , , xn ) is normal to = { = 0}. When
= {xn = 0} or (x) = xn , then = (0, , 0, 1) and

n
ai (x)xi = an (x).
i=1
This reduces to the special case we discussed earlier.
Denition 2.1.2. Let L be a rst-order linear dierential operator as in
(2.1.6) in a neighborhood of x0 Rn and be a smooth hypersurface
containing x0 . Then is noncharacteristic at x0 if
n
(2.1.7) ai (x0 )i = 0,
i=1
where = (1 , , n ) is normal to at x0 . Otherwise, is characteristic
at x0 .

A hypersurface is noncharacteristic if it is noncharacteristic at every


point. Strictly speaking, a hypersurface is characteristic if it is not non-
characteristic, i.e., if it is characteristic at some point. In this book, we will
abuse this terminology. When we say a hypersurface is characteristic, we
mean it is characteristic everywhere. This should cause few confusions. In
14 2. First-Order Dierential Equations

R2 , hypersurfaces are curves, so we shall speak of characteristic curves and


noncharacteristic curves.
The noncharacteristic condition has a simple geometric interpretation.
If we view a = (a1 , , an ) as a vector in Rn , then condition (2.1.7) holds if
and only if a(x0 ) is not a tangent vector to at x0 . This condition assures
that we can compute all derivatives of solutions at x0 .
It is straightforward to check that (2.1.7) is maintained under C 1 -changes
of coordinates.
The discussion leading to Denition 2.1.2 can be easily generalized to
rst-order quasilinear equations. Let be a domain in Rn containing the
origin as before and ai and f be smooth functions in R, for any i =
1, , n. We consider

n
(2.1.8) ai (x, u)uxi = f (x, u) in .
i=1

Again, we rst consider a special case where the initial hypersurface is


given by the hyperplane {xn = 0} and an initial value is given by (2.1.4) for
a given smooth function u0 in a neighborhood of the origin in Rn1 . Let u
be a smooth solution of (2.1.8) and (2.1.4). Then

uxi (0) = u0,xi (0),

for i = 1, , n 1, and
n1 
1 
uxn (0) = ai 0, u0 (0) uxi (0) f 0, u0 (0)
an 0, u0 (0) i=1

if

an 0, u0 (0) = 0.
Similar to (2.1.5), this is the noncharacteristic condition for (2.1.8) at the
origin if the initial hypersurface is given by {xn = 0}.
In general, let x0 be a point in Rn and be a smooth hypersurface
containing x0 . Let u0 be a prescribed smooth function on and ai and f be
smooth functions in a neighborhood of (x0 , u(x0 )) Rn R, for i = 1, , n.
Then for quasilinear PDE (2.1.8), is noncharacteristic at x0 with respect
to u0 if

n

(2.1.9) ai x0 , u0 (x0 ) i = 0,
i=1

where = (1 , , n ) is normal to at x0 .
2.1. Noncharacteristic Hypersurfaces 15

There is a signicant dierence between (2.1.7) for linear PDEs and


(2.1.9) for quasilinear PDEs. For linear PDEs, the noncharacteristic condi-
tion depends on initial hypersurfaces and equations, specically, the coef-
cients of rst-order derivatives. For quasilinear PDEs, it also depends on
initial values.
Next, we turn to general nonlinear partial dierential equations as in
(2.1.1). Let be a domain in Rn containing the origin as before and let F
be a smooth function in R Rn . Consider
F (x, u, u) = 0 in .
We ask the same question as for linear equations. Given an initial hypersur-
face containing the origin and an initial value u0 on , can we compute
all derivatives of solutions at the origin? Again, we rst consider a special
case where the initial hypersurface is given by the hyperplane {xn = 0}
and an initial value is given by (2.1.4) for a given smooth function u0 in a
neighborhood of the origin in Rn1 .
Example 2.1.3. Consider

n
u2xi = 1
i=1
and
u(x , 0) = u0 (x ).
It is obvious that u = xi is a solution for u0 (x ) = xi , i = 1, , n 1.
However, if |x u0 (x )|2 > 1, there are no solutions for such an initial value.

In light of Example 2.1.3, we rst assume that there exists a smooth


function v in a neighborhood of the origin having the given initial value u0
and satisfying F = 0 at the origin, i.e.,

F 0, v(0), v(0) = 0.
Now we can proceed as in the discussion of linear PDEs and ask whether we
can nd uxn at the origin. By the implicit function theorem, this is possible
if

Fuxn 0, v(0), v(0) = 0.
This is the noncharacteristic condition for F = 0 at the origin.
Now we return to Example 2.1.3. We set
F (x, u, p) = |p|2 1 for any p Rn .
We claim that the noncharacteristic condition holds at 0 with respect to u0
if
|x u0 (0)| < 1.
16 2. First-Order Dierential Equations

In fact, let v = u0 + cxn , for a constant c to be determined. Then


|v(0)|2 = |x u0 (0)|2 + c2 .
By choosing

c = 1 |x u0 (0)|2 = 0,
v satises the equation at x = 0. For such two choices of v, we have
Fuxn (0, v(0), v(0)) = 2vxn (0) = 2c = 0.
This proves the claim.
In general, let F = 0 be a rst-order nonlinear PDE as in (2.1.1) in
a neighborhood of x0 Rn . Let be a hypersurface containing x0 and
u0 be a prescribed function on . Then is noncharacteristic at x0
with respect to u0 if there exists a function v such that v = u0 on ,
F (x0 , v(x0 ), v(x0 )) = 0 and

n

Fuxi x0 , v(x0 ), v(x0 ) i = 0,
i=1

where = (1 , , n ) is normal to at x0 .

2.2. The Method of Characteristics


In this section, we solve initial-value problems for rst-order PDEs by the
method of characteristics. We demonstrate that solutions of any rst-order
PDEs with initial values prescribed on noncharacteristic hypersurfaces can
be obtained by solving systems of ordinary dierential equations (ODEs).
Let Rn be a domain and F a smooth function in R Rn . The
general form of rst-order PDEs in is given by
F (x, u, u) = 0 for any x .
Let be a smooth hypersurface in Rn with = and u0 be a smooth
function on . Then we prescribe an initial value on by
u = u0 on .
If is a domain containing the origin and is noncharacteristic at the
origin with respect to u0 , then we are able to compute derivatives of u of
arbitrary order at the origin by discussions in the previous section. Next,
we investigate whether we can solve the initial-value problem at least in a
neighborhood of the origin.
Throughout this section, we always assume that is a domain containing
the origin and that the initial hypersurface is given by the hyperplane
{xn = 0}. Obviously, {xn = 0} has (0, , 0, 1) as a normal vector eld. If
2.2. The Method of Characteristics 17

x Rn , we write x = (x , xn ), where x Rn1 . Our goal is to solve the


initial-value problem
F (x, u, u) = 0,
u(x , 0) = u0 (x ).

2.2.1. Linear Homogeneous Equations. We start with rst-order linear


homogeneous equations. Let ai be smooth in a neighborhood of 0 Rn ,
i = 1, , n, and u0 be smooth in a neighborhood of 0 Rn1 . Consider
n
ai (x)uxi = 0,
(2.2.1) i=1
u(x , 0) = u0 (x ).
By introducing a = (a1 , , an ), we simply write the equation in (2.2.1) as
a(x) u = 0.
Here a(x) is regarded as a vector eld in a neighborhood of 0 Rn . Then
a(x) is a directional derivative along a(x) at x. In the following, we
assume that the hyperplane {xn = 0} is noncharacteristic at the origin, i.e.,
an (0) = 0.
Here we assume that a solution u of (2.2.1) exists. Our strategy is as follows.
For any x Rn close to the origin, we construct a special curve along which
u is constant. If such a curve starts from x and intersects Rn1 {0} at (y, 0)
for a small y Rn1 , then u(x) = u0 (y). To nd such a curve x = x(s), we
consider the restriction of u to it and obtain a one-variable function u(x(s)).
Now we calculate the s-derivative of this function and obtain
d  n
dxi
u(x(s)) = uxi .
ds ds
i=1
In order to have a constant value of u along this curve, we require
d
u(x(s)) = 0.
ds
A simple comparison with the equation in (2.2.1) yields
dxi
= ai (x) for i = 1, , n.
ds
This naturally leads to the following denition.
Denition 2.2.1. Let a = a(x) : Rn be a smooth vector eld in
and x = x(s) be a smooth curve in . Then x = x(s) is an integral curve of
a if
dx
(2.2.2) = a(x).
ds
18 2. First-Order Dierential Equations

The calculation preceding Denition 2.2.1 shows that the solution u of


(2.2.1) is constant along integral curves of the coecient vector eld. This
yields the following method of solving (2.2.1). For any x Rn near the
origin, we nd an integral curve of the coecient vector eld through x by
solving
dx
= a(x),
(2.2.3) ds
x(0) = x.
If it intersects the hyperplane {xn = 0} at (y, 0) for some y suciently small,
then we let u(x) = u0 (y).
Since (2.2.3) is an autonomous system (i.e., the independent variable s
does not appear explicitly), we may start integral curves from initial hyper-
planes. Instead of (2.2.3), we consider the system
dx
= a(x),
(2.2.4) ds
x(0) = (y, 0).
In (2.2.4), integral curves start from (y, 0). By allowing y Rn1 to vary in
a neighborhood of the origin, we expect integral curves x(y, s) to reach any
x Rn in a neighborhood of the origin for small s. This is conrmed by the
following result.
Lemma 2.2.2. Let a be a smooth vector eld in a neighborhood of the origin
with an (0) = 0. Then for any suciently small y Rn1 and any suciently
small s, the solution x = x(y, s) of (2.2.4) denes a dieomorphism in a
neighborhood of the origin in Rn .

Proof. This follows easily from the implicit function theorem. By standard
results in ordinary dierential equations, (2.2.4) admits a smooth solution
x = x(y, s) for any suciently small (y, s) Rn1 R. We treat it as a
map (y, s) x and calculate its Jacobian matrix J at (y, s) = (0, 0). By
x(y, 0) = (y, 0), we have

a1 (0)

x  Id
..
.
J(0) = = .
(y, s) (y,s)=(0,0) an1 (0)
0 0 an (0)
Hence det J(0) = an (0) = 0. 

Therefore, for any suciently small x, we can solve


x(y, s) = x
2.2. The Method of Characteristics 19

uniquely for small y and s. Then u(x) = u0 (y) yields a solution of (2.2.1).
Note that s is not present in the expression of solutions. Hence the value

xn 6
x

(y, 0)
Rn1

Figure 2.2.1. Solutions by integral curves.

of the solution u(x) depends only on the initial value u0 at (y, 0) and, mean-
while, the initial value u0 at (y, 0) inuences the solution u along the integral
curve starting from (y, 0). Therefore, we say the domain of dependence of
the solution u(x) on the initial value is represented by the single point (y, 0)
and the range of inuence of the initial value at a point (y, 0) on solutions
consists of the integral curve starting from (y, 0).
For n = 2, integral curves are exactly characteristic curves. This can be
seen easily by (2.2.2) and Denition 2.1.2. Hence the ODE (2.2.2) is often
referred to as the characteristic ODE. This term is adopted for arbitrary
dimensions. We have demonstrated how to solve homogeneous rst-order
linear PDEs by using characteristic ODEs. Such a method is called the
method of characteristics. Later on, we will develop a similar method to
solve general rst-order PDEs.
We need to emphasize that solutions constructed by the method of char-
acteristics are only local. In other words, they exist only in a neighborhood
of the origin. A natural question here is whether there exists a global so-
lution for globally dened a and u0 . There are several reasons that local
solutions cannot be extended globally. First, u(x) cannot be evaluated at
x Rn if x is not on an integral curve from the initial hypersurface, or equiv-
alently, the integral curve from x does not intersect the initial hypersurface.
Second, u(x) cannot be evaluated at x Rn if the integral curve starting
from x intersects the initial hypersurface more than once. In this case, we
cannot prescribe initial values arbitrarily. They must satisfy a compatibility
condition.
Example 2.2.3. We discuss the initial-value problem for the equation in
Example 2.1.1. We denote by (x, t) points in R2 and let u0 be a smooth
20 2. First-Order Dierential Equations

function in R. We consider
u t + ux = 0 in R (0, ),
u(, 0) = u0 on R.
It is easy to verify that {t = 0} is noncharacteristic. The characteristic ODE
and corresponding initial values are given by
dx dt
= 1, = 1,
ds ds
and
x(0) = x0 , t(0) = 0.
Here, both x and t are treated as functions of s. Hence
x = s + x0 , t = s.
By eliminating s, we have
x t = x0 .
This is a straight line containing (x0 , 0) and with a slope 1. Along this
straight line, u is constant. Hence
u(x, t) = u0 (x t).
We interpreted the fact that u is constant along the straight line x t = x0
in Example 2.1.1. With t as time, the graph of the solution represents a
wave propagating to the right with velocity 1 without changing shape. It is
clear that u exists globally in R2 .

2.2.2. Quasilinear Equations. Next, we discuss initial-value problems


for rst-order quasilinear PDEs. Let Rn be a domain containing the
origin and ai and f be smooth functions in R. For a given smooth
function u0 in a neighborhood of 0 Rn1 , we consider
n
ai (x, u)uxi = f (x, u),
(2.2.5) i=1
u(x , 0) = u0 (x ).
Assume the hyperplane {xn = 0} is noncharacteristic at the origin with
respect to u0 , i.e.,
an 0, u0 (0) = 0.
Suppose (2.2.5) admits a smooth solution u. We rst examine integral curves
dx
= a x, u ,
ds
x(0) = (y, 0),
where y Rn1 . Contrary to the case of homogenous linear equations
we studied earlier, we are unable to solve this ODE since u, the unknown
2.2. The Method of Characteristics 21

function we intend to nd, is present. However, viewing u as a function of


s along these curves, we can calculate how u changes. A similar calculation
as before yields
d  n
dxi 
n
u(x(s)) = uxi = ai (x, u)uxi = f (x, u).
ds ds
i=1 i=1
Then
du
= f (x, u),
ds
u(0) = u0 (y).
Hence we have an ordinary dierential system for x and u. This leads to
the following method for quasilinear PDEs.
Consider the ordinary dierential system
dx
= a(x, u),
ds
du
= f (x, u),
ds
with initial values
x(0) = (y, 0),
u(0) = u0 (y),
where y Rn1 . In formulating this system, we treat both x and u as
functions of s. This system consists of n + 1 equations for n + 1 functions
and is the characteristic ODE of the rst-order quasilinear PDE (2.2.5). By
solving the characteristic ODE, we have a solution given by
x = x(y, s), u = (y, s).
As in the proof of Lemma 2.2.2, we can prove that the map (y, s) x is a
dieomorphism. Hence, for any x Rn suciently small, there exist unique
y Rn1 and s R suciently small such that
x = x(y, s).
Then the solution u at x is given by
u(x) = (y, s).
We now consider an initial-value problem for a nonhomogeneous linear
equation.
Example 2.2.4. We denote by (x, t) points in R2 and let f be a smooth
function in R (0, ) and u0 be a smooth function in R. We consider
ut ux = f in R (0, ),
u(, 0) = u0 on R.
22 2. First-Order Dierential Equations

It is easy to verify that {t = 0} is noncharacteristic. The characteristic ODE


and corresponding initial values are given by
dx dt du
= 1, = 1, = f,
ds ds ds
and
x(0) = x0 , t(0) = 0, u(0) = u0 (x0 ).
Here, x, t and u are all treated as functions of s. By solving for x and t
rst, we have
x = x0 s, t = s.
Then the equation for u can be written as
du
= f (x0 s, s).
ds
A simple integration yields
 s
u = u0 (x0 ) + f (x0 , ) d.
0
By substituting x0 and s by x and t, we obtain
 t
u(x, t) = u0 (x + t) + f (x + t , ) d.
0

Next, we consider an initial-value problem for a quasilinear equation.


Example 2.2.5. We denote by (x, t) points in R2 and let u0 be a smooth
function in R. Consider the initial-value problem for Burgers equation
ut + uux = 0 in R (0, ),
u(, 0) = u0 on R.
It is easy to check that {t = 0} is noncharacteristic with respect to any u0 .
The characteristic ODE and corresponding initial values are given by
dx dt du
= u, = 1, = 0,
ds ds ds
and
x(0) = x0 , t(0) = 0, u(0) = u0 (x0 ).
Here, x, t and u are all treated as functions of s. By solving for t and u rst
and then for x, we obtain
x = u0 (x0 )s + x0 ,
t = s,
u = u0 (x0 ).
By eliminating s from the expressions of x and t, we have
(2.2.6) x = u0 (x0 )t + x0 .
2.2. The Method of Characteristics 23

By the implicit function theorem, we can solve for x0 in terms of (x, t) in a


neighborhood of the origin in R2 . If we denote such a function by
x0 = x0 (x, t),
then we have a solution

u = u0 x0 (x, t) ,
for any (x, t) suciently small. By eliminating x0 and s from the expressions
of x, t and u, we may also write the solution u implicitly by
u = u0 (x ut).
It is interesting to ask whether such a solution can be extended to R2 . Let
cx0 be the characteristic curve given by (2.2.6). It is a straight line in R2 with
a slope 1/u0 (x0 ), along which u is the constant u0 (x0 ). For x0 < x1 with
u0 (x0 ) > u0 (x1 ), two characteristic curves cx0 and cx1 intersect at (X, T )
with
x 0 x1
T = .
u0 (x0 ) u0 (x1 )
Hence, u cannot be extended as a smooth solution up to (X, T ), even as a
continuous function. Such a positive T always exists unless u0 is nondecreas-
ing. In a special case where u0 is strictly decreasing, any two characteristic
curves intersect.

t 6






 -
x0 x1 x

Figure 2.2.2. Intersecting characteristic curves.

Now we examine a simple case. Let


u0 (x) = x.
Obviously, this is strictly decreasing. In this case, cx0 in (2.2.6) is given by
x = x0 x0 t,
and the solution on this line is given by u = x0 . We note that each cx0
contains the point (x, t) = (0, 1) and hence any two characteristic curves
24 2. First-Order Dierential Equations

intersect at (0, 1). Then, u cannot be extended up to (0, 1) as a smooth


solution. In fact, we can solve for x0 easily to get
x
x0 = .
1t
Therefore, u is given by
x
u(x, t) = for any (x, t) R (0, 1).
t1
Clearly, u is not dened at t = 1.

In general, smooth solutions of rst-order nonlinear PDEs may not exist


globally. When two characteristic curves intersect at a positive time T ,
solutions develop a singularity and the method of characteristics breaks
down. A natural question arises whether we can dene solutions beyond the
time T . We expect that less regular functions, if interpreted appropriately,
may serve as solutions.
For an illustration, we return to Burgers equation and employ its diver-
gence structure. We note that Burgers equation can be written as
2
u
ut + = 0.
2 x
This is an example of a scalar conservation law, that is, it is a rst-order
quasilinear PDE of the form
(2.2.7) ut + F (u)x = 0 in R (0, ),
where F : R R is a given smooth function. By taking a C 1 -function
of compact support in R (0, ) and integrating by parts the product of
and the equation in (2.2.7), we obtain


(2.2.8) ut + F (u)x dxdt = 0.
R(0,)
The integration by parts is justied since is zero outside a compact set
in R (0, ). By comparing (2.2.7) and (2.2.8), we note that derivatives
are transferred from u in (2.2.7) to in (2.2.8). Hence, functions u with
no derivatives are allowed in (2.2.8). A locally bounded function u is called
an integral solution of (2.2.7) if it satises (2.2.8) for any C 1 -function of
compact support in R (0, ). The function in (2.2.8) is often referred to
as a test function. In this formulation, discontinuous functions are admitted
to be integral solutions. Even for continuous initial values, a discontinuity
along a curve, called a shock, may develop for integral solutions. Conser-
vation laws and shock waves are an important subject in PDEs. The brief
discussion here serves only as an introduction to this eld. It is beyond the
scope of this book to give a presentation of conservation laws and shock
waves.
2.2. The Method of Characteristics 25

Now we return to our study of initial-value problems of general rst-


order PDEs. So far in our discussion, initial values are prescribed on non-
characteristic hypersurfaces. In general, solutions are not expected to exist
if initial values are prescribed on characteristic hypersurfaces. We illustrate
this by the initial-value problem (2.2.5) for quasilinear equations. Suppose
the initial hyperplane {xn = 0} is characteristic at the origin with respect
to the initial value u0 . Then

an 0, u0 (0) = 0.
Hence uxn (0) is absent from the equation in (2.2.5) when evaluated at x = 0.
Therefore, (2.2.5) implies

n1

(2.2.9) ai 0, u0 (0) u0,xi (0) = f 0, u0 (0) .
i=1

This is the compatibility condition for the initial value u0 . Even if the origin
is the only point where {xn = 0} is characteristic, solutions may not exist in
any neighborhood of the origin for initial values satisfying the compatibility
condition (2.2.9). Refer to Exercise 2.5.

2.2.3. General Nonlinear Equations. Next, we discuss general rst-


order nonlinear PDEs. Let Rn be a domain containing the origin and
F = F (x, u, p) be a smooth function in (x, u, p) R Rn . Consider
(2.2.10) F (x, u, u) = 0,
for any x , and prescribe an initial value on {xn = 0} by
(2.2.11) u(x , 0) = u0 (x ),
for any x with (x , 0) . Assume there is a scalar a0 such that

F 0, u0 (0), x u0 (0), a0 = 0.
The noncharacteristic condition with respect to u0 and a0 is given by

(2.2.12) Fpn 0, u0 (0), x u0 (0), a0 = 0.
By (2.2.12) and the implicit function theorem, there exists a smooth function
a(x ) in a neighborhood of the origin in Rn1 such that a(0) = a0 and

(2.2.13) F x , 0, u0 (x ), x u0 (x ), a(x ) = 0,
for any x Rn1 suciently small. In the following, we will seek a solution
of (2.2.10)(2.2.11) and
uxn (x , 0) = a(x ),
for any x small.
26 2. First-Order Dierential Equations

We start with a formal consideration. Suppose we have a smooth solu-


tion u. Set

(2.2.14) pi = uxi for i = 1, , n.

Then

(2.2.15) F (x1 , , xn , u, p1 , , pn ) = 0.

Dierentiating (2.2.15) with respect to xi , we have


n
Fpj pj,xi + Fxi + Fu uxi = 0 for i = 1, , n.
j=1

By pj,xi = pi,xj , we obtain


n
(2.2.16) Fpj pi,xj = Fxi Fu uxi for i = 1, , n.
j=1

We view (2.2.16) as a rst-order quasilinear equation for pi , for each xed


i = 1, , n. An important feature here is that the coecient for pi,xj is
Fpj , which is independent of i. For each xed i = 1, , n, the characteristic
ODE associated with (2.2.16) is given by

dxj
= Fpj for j = 1, , n,
ds
dpi
= Fu pi Fxi .
ds
We also have
du  
n n
dxj
= uxj = pj Fpj .
ds ds
j=1 j=1

Now we collect ordinary dierential equations for xj , u and pi .


The characteristic ODE for the rst-order nonlinear PDE (2.2.10) is the
ordinary dierential system
dxj
= Fpj (x, u, p) for j = 1, , n,
ds
dpi
(2.2.17) = Fu (x, u, p)pi Fxi (x, u, p) for i = 1, , n,
ds
du n
= pj Fpj (x, u, p),
ds
j=1
2.2. The Method of Characteristics 27

with initial values at s = 0,


x(0) = (y, 0),
u(0) = u0 (y),
(2.2.18)
pi (0) = u0,xi (y) for i = 1, , n 1,
pn (0) = a(y),
where y Rn1 , u0 is the initial value as in (2.2.11) and a is the function
chosen to satisfy (2.2.13). This is an ordinary dierential system of 2n + 1
equations for the 2n + 1 functions x, u and p. Here we view x, u and p as
functions of s. Compare this with a similar ordinary dierential system of
n + 1 equations for n + 1 functions x and u for rst-order quasilinear PDEs.
Solving (2.2.17) with (2.2.18) near (y, s) = (0, 0), we have
x = x(y, s), u = (y, s), p = p(y, s),
for any y and s suciently small. We will prove that the map (y, s) x
is a dieomorphism near the origin in Rn . Hence for any given x near the
origin, there exist unique y Rn1 and s R such that
x = x(y, s).
Then we dene u by
u(x) = (y, s).
Theorem 2.2.6. The function u dened above is a solution of (2.2.10)
(2.2.11).

We should note that this solution u depends on the choice of the scalar
a0 and the function a(x ).

Proof. The proof consists of several steps.


Step 1. The map (y, s) x is a dieomorphism near (0, 0). This is
proved as in the proof of Lemma 2.2.2. In fact, the Jacobian matrix of the
map (y, s) x at (0, 0) is given by


 ..
x  Id .
J(0) = = ,
(y, s) y=0,s=0
0 0 ds (0, 0)
dxn

where
dxn
(0, 0) = Fpn (0, u0 (0), x u0 (0), a0 ).
ds
Hence det J(0) = 0 by the noncharacteristic condition (2.2.12). By the
implicit function theorem, for any x Rn suciently small, we can solve
28 2. First-Order Dierential Equations

x = x(y, s) uniquely for y Rn1 and s R suciently small. Then dene


u(x) = (y, s).
We will prove that this is the desired solution and
pi (y, s) = uxi (x(y, s)) for i = 1, , n.

Step 2. We claim that



F x(y, s), (y, s), p(y, s) 0,
for any y and s suciently small. Denote by f (s) the function in the left-
hand side. Then by (2.2.18)

f (0) = F y, 0, u0 (y), x u0 (y), a(y)) = 0.
Next, we have by (2.2.17)
df (s) d
= F x(y, s), (y, s), p(y, s)
ds ds
n
dxi du 
n
dpj
= Fxi + Fu + F pj
ds ds ds
i=1 j=1

n 
n 
n
= Fxi Fpi + Fu pj Fpj + Fpj (Fu pj Fxj ) = 0.
i=1 j=1 j=1

Hence f (s) 0.
Step 3. We claim that
pi (y, s) = uxi (x(y, s)) for i = 1, , n,
for any y and s suciently small. Let
wi (s) = uxi (x(y, s)) pi (y, s) for i = 1, , n.
We will prove that
wi (s) = 0 for any s small and i = 1, , n.

We rst evaluate wi at s = 0. By initial values (2.2.18), we have wi (0) = 0


for i = 1, , n 1. Next, we note that, by (2.2.17),
n  
du  
n n
dxj
(2.2.19) 0 = pj Fpj = uxj pj Fpj = Fpj (uxj pj ),
ds ds
j=1 j=1 j=1
or

n
Fpj wj = 0.
j=1
This implies wn (0) = 0 since wi (0) = 0 for i = 1, , n 1, and Fpn |s=0 = 0
by the noncharacteristic condition (2.2.12).
2.2. The Method of Characteristics 29

Next, we claim that dwi


ds is a linear combination of wj , j = 1, , n, i.e.,
dwi 
n
= aij wj for i = 1, , n,
ds
j=1

for some functions aij , i, j = 1, , n. Then basic ODE theory implies


wi 0 for i = 1, , n. To prove the claim, we rst note that, by (2.2.17),
dwi  dpi 
n n
dxj
= uxi xj = uxi xj Fpj + Fu pi + Fxi .
ds ds ds
j=1 j=1

To eliminate the second-order derivatives of u, we dierentiate (2.2.19) with


respect to xi and get

n 
n
Fpj (uxi xj pj,xi ) + (Fpj )xi wj = 0.
j=1 j=1

A simple substitution implies


dwi  
n n
= Fpj pj,xi (Fpj )xi wj + Fu pi + Fxi .
ds
j=1 j=1

By Step 2,

F x, u(x), p1 (x), , pn (x) = 0.
Dierentiating with respect to xi , we have

n
Fxi + Fu uxi + Fpj pj,xi = 0.
j=1

Hence
dwi  n
= Fxi Fu uxi (Fpj )xi wj + Fu pi + Fxi
ds
j=1

n
= F u wi (Fpj )xi wj ,
j=1
or
dwi  n

= Fu ij + (Fpj )xi wj .
ds
j=1
This ends the proof of Step 3.
Step 2 and Step 3 imply that u is the desired solution. 

To end this section, we briey compare methods we used to solve rst-


order linear or quasi-linear PDEs and general rst-order nonlinear PDEs. In
solving a rst-order quasi-linear PDE, we formulate an ordinary dierential
30 2. First-Order Dierential Equations

system of n + 1 equations for n + 1 functions x and u. For a general rst-


order nonlinear PDE, the corresponding ordinary dierential system consists
of 2n + 1 equations for 2n + 1 functions x, u and u. Here, we need to take
into account the gradient of u by adding n more equations for u. In other
words, we regard our rst-order nonlinear PDE as a relation for (u, p) with
a constraint p = u. We should emphasize that this is a unique feature
for single rst-order PDEs. For PDEs of higher order or for rst-order
partial dierential systems, nonlinear equations are dramatically dierent
from linear equations. In the rest of the book, we concentrate only on linear
equations.

2.3. A Priori Estimates


A priori estimates play a fundamental role in PDEs. Usually, they are the
starting point for establishing existence and regularity of solutions. To derive
a priori estimates, we rst assume that solutions already exist and then
estimate certain norms of solutions by those of known functions in equations,
for example, nonhomogeneous terms, coecients and initial values. Two
frequently used norms are L -norms and L2 -norms. The importance of L2 -
norm estimates lies in the Hilbert space structure of the L2 -space. Once L2 -
estimates of solutions and their derivatives have been derived, we can employ
standard results about Hilbert spaces, for example, the Riesz representation
theorem, to establish the existence of solutions.
In this section, we will use rst-order linear PDEs to demonstrate how
to derive a priori estimates in L -norms and L2 -norms. We rst examine
briey rst-order linear ordinary dierential equations. Let be a constant
and f = f (t) be a continuous function. Consider
du
u = f (t).
dt
A simple calculation shows that
 t
u(t) = et u(0) + e(ts) f (s) ds.
0
For any xed T > 0, we have
 
|u(t)| e t
|u(0)| + T sup |f | for any t (0, T ).
[0,T ]

Here, we estimate the sup-norm of u in [0, T ] by the initial value u(0) and
the sup-norm of the nonhomogeneous term f in [0, T ].
Now we turn to PDEs. For convenience, we work in Rn (0, ) and
denote points by (x, t), with x Rn and t (0, ). In many applications,
we interpret x as the space variable and t the time variable.
2.3. A Priori Estimates 31

2.3.1. L -Estimates. Let ai , b and f be continuous functions in Rn


[0, ) and u0 be a continuous function in Rn . We assume that a = (a1 , ,
an ) satises
1
(2.3.1) |a| in Rn [0, ),

for a positive constant . Consider
n
ut + ai (x, t)uxi + b(x, t)u = f (x, t) in Rn (0, ),
(2.3.2) i=1
u(x, 0) = u0 (x) in Rn .
It is obvious that the initial hypersurface {t = 0} is noncharacteristic. We
may write the equation in (2.3.2) as
ut + a(x, t) x u + b(x, t)u = f (x, t).
We note that a(x, t) x + t is a directional derivative along the direction
(a(x, t), 1). With (2.3.1), it is easy to see that the vector (a(x, t), 1) (starting
from the origin) is in fact in the cone given by
{(y, s) : |y| s} Rn R.
This is a cone opening upward and with vertex at the origin.

(a, 1)
t
MB 6
B
@ B
@
@
@
@
@ -



 Rn
+


Figure 2.3.1. The cone with the vertex at the origin.

For any point P = (X, T ) Rn (0, ), consider the cone C (P )


(opening downward) with vertex at P dened by
C (P ) = {(x, t) : 0 < t < T, |x X| < T t}.
We denote by s C (P ) and C (P ) the side and bottom of the boundary,
respectively, i.e.,
s C (P ) = {(x, t) : 0 < t T, |x X| = T t},
C (P ) = {(x, 0) : |x X| T }.
32 2. First-Order Dierential Equations

We note that C (P ) is simply the closed ball in Rn {0} centered at


(X, 0) with radius T /. For any (x, t) s C (P ), let a(x, t) be a vector
in Rn satisfying (2.3.1). Then the vector (a(x, t), 1), if positioned at (x, t),
points outward from the cone C (P ) or along the boundary s C (P ). Hence
for a function dened only in C (P ), it makes sense to calculate ut + a x u
at (x, t), which is viewed as a directional derivative of u along (a(x, t), 1) at
(x, t). This holds in particular when (x, t) is the vertex P .

t6
(a, 1)  (a, 1)

MB 
B
S
B  S
 S
 S -
 S
 S

Rn

Figure 2.3.2. The cone C (P ) and positions of vectors.

Now we calculate the unit outward normal vector of s C (P ) \ {P }. Set


(x, t) = |x X| (T t).
Obviously, s C (P ) \ {P } is a part of { = 0}. Then for any (x, t)
s C (P ) \ {P }, 
xX
= (x , t ) = ,1 .
|x X|
Therefore, the unit outward normal vector of s C (P ) \ {P } at (x, t) is
given by 
1 xX
= ,1 .
2 + 1 |x X|
For n = 1, the cone C (P ) is a triangle bounded by the straight lines
(x X) = T t and t = 0. The side of the cone consists of two line
segments,
the left segment: (x X) = T t, 0 < t T,
with a normal vector (, 1),
and
the right segment: (x X) = T t, 0 < t T,
with a normal vector (, 1).
2.3. A Priori Estimates 33

It is easy to see that the integral curve associated with (2.3.2) start-
ing from P and going to the initial hypersurface Rn {0} stays in C (P ).
In fact, this is true for any point (x, t) C (P ). This suggests that solu-
tions in C (P ) should depend only on f in C (P ) and the initial value u0
on C (P ). The following result, proved by a maximum principle type
argument, conrms this.

t
6

P
S
 S
 S
 S -
 S
 S

Rn

Figure 2.3.3. The domain of dependence.

Theorem 2.3.1. Let ai , b and f be continuous functions in Rn [0, )


satisfying (2.3.1) and u0 be a continuous function in Rn . Suppose u
C 1 (Rn (0, )) C(Rn [0, )) is a solution of (2.3.2). Then for any
P = (X, T ) Rn (0, ),
sup |et u| sup |u0 | + T sup |et f |,
C (P ) C (P ) C (P )

where is a nonnegative constant such that


b in C (P ).

If b 0, we take = 0 and have


sup |u| sup |u0 | + T sup |f |.
C (P ) C (P ) C (P )

Proof. Take any positive number  > and set



M= sup |u0 |, F = sup |e t f |.
C (P ) C (P )

We will prove

|e t u(x, t)| M + tF for any (x, t) C (P ).
For the upper bound, we consider

w(x, t) = e t u(x, t) M tF.
34 2. First-Order Dierential Equations

A simple calculation shows that


n

wt + ai wxi + (b +  )w = (b +  )(M + tF ) + e t f F.
i=1

Since b + > 0, the right-hand side is nonpositive by the denition of M
and F . Hence

wt + a x w + b +  )w 0 in C (P ).
Let w attain its maximum in C (P ) at (x0 , t0 ) C (P ). We prove
w(x0 , t0 ) 0.
First, it is obvious if (x0 , t0 ) C (P ), since w(x0 , t0 ) = u0 (x0 )M 0 by
the denition of M . If (x0 , t0 ) C (P ), i.e., (x0 , t0 ) is an interior maximum
point, then
(wt + a x w)|(x0 ,t0 ) = 0.
If (x0 , t0 ) s C (P ), by the position of the vector (a(x0 , t0 ), 1) relative to
the cone C (P ), we can take the directional derivative along (a(x0 , t0 ), 1),
obtaining
(wt + a x w)|(x0 ,t0 ) 0.
Hence, in both cases, we obtain
(b +  )w|(x0 ,t0 ) 0.
Since b +  > 0, this implies w(x0 , t0 ) 0. (We need the positivity of b + 
here!) Hence w(x0 , t0 ) 0 in all three cases. Therefore, w 0 in C (P ), or

u(x, t) e t (M + tF ) for any (x, t) C (P ).
We simply let  to get the desired upper bound. For the lower bound,
we consider

v(x, t) = e t u(x, t) + M + tF.
The argument is similar and hence omitted. 

For n = 1, (2.3.2) has the form


ut + a(x, t)ux + b(x, t)u = f (x, t).
In this case, it is straightforward to see that
(wt + awx )|(x0 ,t0 ) 0,
if w assumes its maximum at (x0 , t0 ) s C (P ). To prove this, we rst note
that t + 1 x and t 1 x are directional derivatives along the straight lines
t t0 = (x x0 ) and t t0 = (x x0 ), respectively. Since w assumes
its maximum at (x0 , t0 ), we have
 
1  1 
wt + wx  0, wt wx  0.
(x0 ,t0 ) (x0 ,t0 )
2.3. A Priori Estimates 35

In fact, one of them is zero if (x0 , t0 ) s C (P ) \ {P }. Then we obtain


1
wt (x0 , t0 ) |wx |(x0 , t0 ) |awx |(x0 , t0 ).

One consequence of Theorem 2.3.1 is the uniqueness of solutions of
(2.3.2).
Corollary 2.3.2. Let ai , b and f be continuous functions in Rn [0, )
satisfying (2.3.1) and u0 be a continuous function in Rn . Then there exists
at most one C 1 (Rn (0, )) C(Rn [0, ))-solution of (2.3.2).

Proof. Let u1 and u2 be two solutions of (2.3.2). Then u1 u2 satises


(2.3.2) with f = 0 in C (P ) and u0 = 0 on C (P ). Hence u1 u2 = 0 in
C (P ) by Theorem 2.3.1. 

Another consequence of Theorem 2.3.1 is the continuous dependence of


solutions on initial values and nonhomogeneous terms.
Corollary 2.3.3. Let ai , b, f1 , f2 be continuous functions in Rn [0, ) sat-
isfying (2.3.1) and u01 , u02 be continuous functions in Rn . Suppose u1 , u2
C 1 (Rn (0, )) C(Rn [0, )) are solutions of (2.3.2), with f1 , f2 replac-
ing f and u01 , u02 replacing u0 , respectively. Then for any P = (X, T )
Rn (0, ),
sup |et (u1 u2 )| sup |u01 u02 | + T sup |et (f1 f2 )|,
C (P ) C (P ) C (P )

where is a nonnegative constant such that


b in C (P ).

The proof is similar to that of Corollary 2.3.2 and is omitted.


Theorem 2.3.1 also shows that the value u(P ) depends only on f in
C (P ) and u0 on C (P ). Hence C (P ) contains the domain of dependence
of u(P ) on f , and C (P ) contains the domain of dependence of u(P ) on
u0 . In fact, the domain of dependence of u(P ) on f is the integral curve
through P in C (P ), and the domain of dependence of u(P ) on u0 is the
intercept of this integral curve with the initial hyperplane {t = 0}. We now
consider this from another point of view. For simplicity, we assume that f
is identically zero in Rn (0, ) and the initial value u0 at t = 0 is zero
outside a bounded domain D0 Rn . Then for any t > 0, u(, t) = 0 outside
Dt = {(x, t) : |x x0 | < t for some x0 D0 }.

In other words, u0 inuences u only in {t>0} Dt . This is the nite-speed
propagation.
36 2. First-Order Dierential Equations

t6

JJ

J
-
J

Rn

Figure 2.3.4. The range of inuence.

2.3.2. L2 -Estimates. Next, we derive an estimate of the L2 -norm of u in


terms of the L2 -norms of f and u0 .
Theorem 2.3.4. Let ai be C 1 -functions in Rn [0, ) satisfying (2.3.1), b
and f be continuous functions in Rn [0, ) and u0 be a continuous function
in Rn . Suppose u C 1 (Rn (0, ))C(Rn [0, )) is a solution of (2.3.2).
Then for any P = (X, T ) Rn (0, ),
  
et u2 dxdt u20 dx + et f 2 dxdt,
C (P ) C (P ) C (P )

where is a positive constant depending only on the C 1 -norms of ai and the


sup-norm of b in C (P ).

Proof. For a nonnegative constant to be determined, we multiply the


equation in (2.3.2) by 2et u. In view of
2et uut = (et u2 )t + et u2 ,
2ai et uuxi = (et ai u2 )xi et ai,xi u2 ,
we have
 

n 
n
(et u2 )t + (et ai u2 )xi + et + 2b ai,xi u2 = 2et uf.
i=1 i=1

An integration in C (P ) yields
  

n
et t + ai i u2 dS
s C (P ) i=1
  

n
t
+ e + 2b ai,xi u2 dxdt
C (P ) i=1
 
= u20 dx + 2et uf dxdt,
C (P ) C (P )
2.3. A Priori Estimates 37

where the unit exterior normal vector on s C (P ) is given by



1 xX
(x , t ) = (1 , , n , t ) = ,1 .
1 + 2 |x X|
By (2.3.1) and the Cauchy inequality, we have
 
n 
  |a|
 ai i  |a||x | = t ,
  1 + 2
i=1

and hence

n
t + ai i 0 on s C (P ).
i=1
Next, we choose such that

n
+ 2b ai,xi 2 in C (P ).
i=1

Then
  
t 2
2 e u dxdt u20 dx + 2et uf dxdt.
C (P ) C (P ) C (P )

Here we simply dropped the integral over s C (P ) since it is nonnegative.


The Cauchy inequality implies
  
2et uf dxdt et u2 dxdt + et f 2 dxdt.
C (P ) C (P ) C (P )

We then have the desired result. 

The proof illustrates a typical method of deriving L2 -estimates. We


multiply the equation by its solution u and rewrite the product as a linear
combination of u2 and its derivatives. Upon integrating by parts, domain
integrals of derivatives are reduced to boundary integrals. Hence, the re-
sulting integral identity consists of domain integrals and boundary integrals
of u2 itself. Derivatives of u are eliminated.
We note that the estimate in Theorem 2.3.4 is similar to that in The-
orem 2.3.1, with the L2 -norms replacing the L -norms. As consequences
of Theorem 2.3.4, we have the uniqueness of solutions of (2.3.2) and the
continuous dependence of solutions on initial values and nonhomogeneous
terms in L2 -norms. We can also discuss domains of dependence and ranges
of inuence using Theorem 2.3.4.
We now derive an L2 -estimate of solutions in the entire space.
Theorem 2.3.5. Let ai be bounded C 1 -functions, b and f be continuous
functions in Rn [0, ) and u0 be a continuous function in Rn . Suppose
38 2. First-Order Dierential Equations

u C 1 (Rn (0, )) C(Rn [0, )) is a solution of (2.3.2). For any


T > 0, if f L2 (Rn (0, T )) and u0 L2 (Rn ), then
 
t 2
e u dx + et u2 dxdt
R {T }
n R (0,T )
n
 
u20 dx + et f 2 dxdt,
Rn Rn (0,T )

where is a positive constant depending only on the C 1 -norms of ai and the


sup-norm of b in Rn (0, T ).

Proof. We rst take > 0 such that (2.3.1) holds. Take any t > T and

t6


J

J

J

J

J

J -

Rn

Figure 2.3.5. A domain of integration.

consider
D(t) = {(x, t) : |x| < t t, 0 < t < T }.
In other words,
D(t) = C (0, t) {0 < t < T }.
We denote by D(t), s D(t) and + D(t) the bottom, the side and the top
of the boundary, i.e.,

D(t) = {(x, 0) : |x| < t},


s D(t) = {(x, t) : |x| = t t, 0 < t < T },
+ D(t) = {(x, T ) : |x| < t T }.

We now proceed as in the proof of Theorem 2.3.4, with D(t) replacing C (P ).


We note that there is an extra portion + D(t) in the boundary D(t). A
2.3. A Priori Estimates 39

similar integration over D(t) yields


 
t 2
e u dx + et u2 dxdt
+ D(t) D(t)
 
u20 dx + et f 2 dxdt.
D(t) D(t)

We note that t enters this estimate only through the domain D(t). Hence,
we may let t to get the desired result. 

We point out that there are no decay assumptions on u as x in


Theorem 2.3.5.

2.3.3. Weak Solutions. Anyone beginning to study PDEs might well ask,
what a priori estimates are good for. One consequence is of course the
uniqueness of solutions, as shown in Corollary 2.3.2. In fact, one of the
most important applications of Theorem 2.3.5 is to prove the existence of
a weak solution of the initial-value problem (2.3.2). We illustrate this with
the homogeneous initial value, i.e., u0 = 0.
To introduce the notion of a weak solution, we x a T > 0 and consider
functions in Rn (0, T ). Set


n
(2.3.3) Lu = ut + ai uxi + bu in Rn (0, T ).
i=1

Obviously, L is a linear dierential operator dened in C 1 (Rn (0, T )). For


any u, v C 1 (Rn (0, T )) C(Rn [0, T ]), we integrate vLu in Rn (0, T ).
To this end, we write
 
n n
vLu = u vt (ai v)xi + bv + (uv)t + (ai uv)xi .
i=1 i=1

This identity naturally leads to an introduction of the adjoint dierential


operator L of L dened by
 
n n n

L v = vt (ai v)xi + bv = vt ai vxi + b ai,xi v.
i=1 i=1 i=1

Then

n
vLu = uL v + (uv)t + (ai uv)xi .
i=1
40 2. First-Order Dierential Equations

We now require that u and v vanish for large x. Then by a simple integration
in Rn (0, T ), we obtain
 
vLu dxdt = uL v dxdt
R (0,T )
n R (0,T )
n
 
+ uv dx uv dx.
Rn {t=T } Rn {t=0}
We note that derivatives are transferred from u in the left-hand side to v in
the right-hand side. Integrals over {t = 0} and {t = T } will disappear if we
require, in addition, that uv = 0 on {t = 0} and {t = T }.
Denition 2.3.6. Let f and u be functions in L2 (Rn (0, T )). Then u is
a weak solution of Lu = f in Rn (0, T ) if
 

(2.3.4) uL v dxdt = f v dxdt,
Rn (0,T ) Rn (0,T )

for any C 1 -functions v of compact support in Rn (0, T ).

The functions v in (2.3.4) are called test functions. It is worth restating


that no derivatives of u are involved.
Now we are ready to prove the existence of weak solutions of (2.3.3) with
homogeneous initial values. The proof requires the Hahn-Banach theorem
and the Riesz representation theorem in functional analysis.
Theorem 2.3.7. Let ai be bounded C 1 -functions in Rn (0, T ), i = 1, , n,
and b a bounded continuous function in Rn (0, T ). Then for any f
L2 (Rn (0, T )), there exists a u L2 (Rn (0, T )) such that
 

uL v dxdt = f v dxdt,
Rn (0,T ) Rn (0,T )

for any v C 1 (Rn (0, T )) C(Rn [0, T ]) with v(x, t) = 0 for any (x, t)
with large x and any (x, t) = (x, T ).

The function u in Theorem 2.3.7 is called a weak solution of the initial-


value problem
Lu = f in Rn (0, T ),
(2.3.5)
u=0 on Rn .
We note that test functions v in Theorem 2.3.7 are not required to vanish
on {t = 0}.
To prove Theorem 2.3.7, we rst introduce some notation. We denote by
C01 (Rn (0, T )) the collection of C 1 -functions in Rn (0, T ) with compact
support, and we denote by C  1 (Rn (0, T )) the collection of C 1 -functions in
0
Rn (0, T ) with compact support in x-directions. In other words, functions
2.3. A Priori Estimates 41

in C01 (Rn (0, T )) vanish for large x and for t close to 0 and T , and functions
 1 (Rn (0, T )) vanish only for large x.
in C 0
We note that, with L in (2.3.3), we can rewrite the estimate in Theorem
2.3.5 as

uL2 (Rn (0,T )) C u(, 0)L2 (Rn ) + LuL2 (Rn (0,T )) ,
where C is a positive constant depending only on T , the C 1 -norms
of ai and the sup-norm of b in Rn (0, T ). This holds for any u
C 1 (Rn (0, T )) C(Rn [0, T )) with Lu L2 (Rn (0, T )) and u(, 0)
L2 (Rn ). In particular, we have
(2.3.6) uL2 (Rn (0,T )) CLuL2 (Rn (0,T )) ,
 1 (Rn (0, T )) C(Rn [0, T )) with u = 0 on {t = 0}.
for any u C 0

Proof of Theorem 2.3.7. In the following, we denote by (, )L2 (Rn (0,T ))


the L2 -inner product in Rn (0, T ).
Now L is like L, but the terms involving derivatives have opposite signs.
When we consider an initial-value problem for L in Rn (0, T ), we view
{t = T } as the initial hyperplane for the domain Rn (0, T ). Thus (2.3.6)
also holds for L , and we obtain
(2.3.7) vL2 (Rn (0,T )) CL vL2 (Rn (0,T )) ,
for any v C 1 (Rn (0, T )) C(Rn (0, T ]) with v = 0 on {t = T }, where
0
C is a positive constant depending only on T , the C 1 -norms of ai and the
sup-norm of b in Rn (0, T ). We denote by C  1 (Rn (0, T )) the collection
of functions v C  1 (Rn (0, T )) C(Rn (0, T ]) with v = 0 on {t = T }.
0
Consider the linear functional F : L C  1 (Rn (0, T )) R given by
F (L v) = (f, v)L2 (Rn (0,T )) ,
for any v C 1 (Rn (0, T )). We note that F acting on L v in the left-hand
side is dened in terms of v itself in the right-hand side. Hence we need to
verify that such a denition makes sense. In other words, we need to prove
that L v1 = L v2 implies
(f, v1 )L2 (Rn (0,T )) = (f, v2 )L2 (Rn (0,T )) ,
 1 (Rn (0, T )). By linearity, it suces to prove that L v = 0
for any v1 , v2 C
implies v = 0 for any v C  1 (Rn (0, T )). We note that it is a consequence
of (2.3.7). Hence, F is a well-dened linear functional on L C  1 (Rn (0, T )).
Moreover, by the Cauchy inequality and (2.3.7) again, we have
|F (L v)| f L2 (Rn (0,T )) vL2 (Rn (0,T ))
Cf L2 (Rn (0,T )) L vL2 (Rn (0,T )) ,
42 2. First-Order Dierential Equations

 1 (Rn (0, T )). Therefore, F is a well-dened bounded linear


for any v C
functional on the subspace L C  1 (Rn (0, T )) of L2 (Rn (0, T )). Thus we
apply the Hahn-Banach theorem to obtain a bounded linear extension of F
(also denoted by F ) dened on L2 (Rn (0, T )) such that
F  Cf L2 (Rn (0,T )) .
Here, F  is the norm of the linear functional F on L2 (Rn (0, T )). By the
Riesz representation theorem, there exists a u L2 (Rn (0, T )) such that
F (w) = (u, w)L2 (Rn (0,T )) for any w L2 (Rn (0, T )),
and
uL2 (Rn (0,T )) = F .
In particular, we have
F (L v) = (u, L v)L2 (Rn (0,T )) ,
 1 (Rn (0, T )), and hence by the denition of F ,
for any v C
(u, L v)L2 (Rn (0,T )) = (f, v)L2 (Rn (0,T )) .
Then u is the desired function. 

Theorem 2.3.7 asserts the existence of a weak solution of (2.3.5). Now we


show that the weak solution u is a classical solution if u is C 1 in Rn (0, T )
and continuous up to {t = 0}. Under these extra assumptions on u, we
integrate uL v by parts to get
  
vLu dxdt + uv dx = f v dxdt,
Rn (0,T ) Rn {t=0} Rn (0,T )

 1 (Rn (0, T )) C(Rn [0, T ]) with v = 0 on {t = T }. There


for any v C 0
are no boundary integrals on the vertical sides and on the upper side since
v vanishes there. In particular,
 
vLu dxdt = f v dxdt,
Rn (0,T ) Rn (0,T )

for any v C01 (Rn (0, T )). Since C01 (Rn (0, T )) is dense in L2 (Rn (0, T )),
we conclude that
Lu = f in Rn (0, T ).
Therefore, 
uv dx = 0,
Rn {t=0}
 1 (Rn (0, T )) C(Rn [0, T ]) with v = 0 on {t = T }. This
for any v C 0
implies 
u(, 0) dx = 0 for any C01 (Rn ).
Rn
2.4. Exercises 43

Again by the density of C01 (Rn ) in L2 (Rn ), we conclude that

u(, 0) = 0 on Rn .

We note that a crucial step in passing from weak solutions to classical solu-
tions is to improve the regularity of weak solutions.
Now we summarize the process of establishing solutions by using a priori
estimates in the following four steps:
Step 1. Prove a priori estimates.
Step 2. Prove the existence of a weak solution by methods of functional
analysis.
Step 3. Improve the regularity of a weak solution.
Step 4. Prove that a weak solution with sucient regularity is a classical
solution.
In discussions above, we carried out Steps 1, 2 and 4. Now we make
several remarks on Steps 3 and 4. We recall that in Step 4 we proved that
weak solutions with continuous derivatives are classical solutions. The re-
quirement of continuity of derivatives can be weakened. It suces to assume
that u has derivatives in the L2 -sense and to verify that the integration by
parts can be performed. Then we can conclude that Lu = f almost every-
where. Because of this relaxed regularity requirement, we need only prove
that weak solutions possess derivatives in the L2 -sense in Step 3. The proof
is closely related to a priori estimates of derivatives of solutions. The brief
discussion here suggests the necessity of introducing new spaces of functions,
functions with derivatives in L2 . These are the Sobolev spaces, which play
a fundamental role in PDEs. In subsequent chapters, Sobolve spaces will
come up for dierent classes of equations. We should point out that Sobolev
spaces and weak solutions are not among the main topics in this book. Their
appearance in this book serves only as an illustration of their importance.

2.4. Exercises
Exercise 2.1. Find solutions of the following initial-value problems in R2 :
2 /2
(1) 2uy ux + xu = 0 with u(x, 0) = 2xex ;
(2) uy + (1 + x2 )u x u = 0 with u(x, 0) = arctan x.

Exercise 2.2. Solve the following initial-value problems:


(1) uy + ux = u2 with u(x, 0) = h(x);
(2) uz + xux + yuy = u with u(x, y, 0) = h(x, y).
44 2. First-Order Dierential Equations

Exercise 2.3. Let B1 be the unit disc in R2 and a and b be continuous


functions in B1 with a(x, y)x + b(x, y)y > 0 on B1 . Assume u is a C 1 -
solution of
a(x, y)ux + b(x, y)uy = u in B1 .
Prove that u vanishes identically.

Exercise 2.4. Find a smooth function a = a(x, y) in R2 such that, for the
equation of the form
uy + a(x, y)ux = 0,
there does not exist any solution in the entire R2 for any nonconstant initial
value prescribed on {y = 0}.

Exercise 2.5. Let be a number and h = h(x) be a continuous function


in R. Consider
yux + xuy = u,
u(x, 0) = h(x).
(1) Find all points on {y = 0} where {y = 0} is characteristic. What
is the compatibility condition on h at these points?
(2) Away from the points in (1), nd the solution of the initial-value
problem. What is the domain of this solution in general?
(3) For the cases h(x) = x, = 1 and h(x) = x, = 3, check whether
this solution can be extended over the points in (1).
(4) For each point in (1), nd all characteristic curves containing it.
What is the relation of these curves and the domain in (2)?

Exercise 2.6. Let R be a real number and h = h(x) be continuous in


R and C 1 in R \ {0}. Consider
xux + yuy = u,
u(x, 0) = h(x).
(1) Check that the straight line {y = 0} is characteristic at each point.
(2) Find all h satisfying the compatibility condition on {y = 0}. (Con-
sider three cases, > 0, = 0 and < 0.)
(3) For > 0, nd two solutions with the given initial value on {y = 0}.
(This is easy to do simply by inspecting the equation, especially for
= 2.)

Exercise 2.7. In the plane, solve uy = 4u2x near the origin with u(x, 0) = x2
on {y = 0}.
2.4. Exercises 45

Exercise 2.8. In the plane, nd two solutions of the initial-value problem


1
xux + yuy + (u2x + u2y ) = u,
2
1
u(x, 0) = (1 x2 ).
2
Exercise 2.9. In the plane, nd two solutions of the initial-value problem
1 2
u + uuy = u,
4 x 
1 2 1
u x, x = x2 .
2 2
Exercise 2.10. Let ai , b and f be continuous functions satisfying (2.3.1)
and u be a C 1 -solution of (2.3.2) in Rn [0, ). Prove that, for any P =
(X, T ) Rn (0, ),
1
sup |et u| sup |u0 | + sup |et f |,
C (P ) C (P ) + inf b
C (P ) C (P )

where is a constant such that


+ inf b > 0.
C (P )

Exercise 2.11. Let ai , b and f be C 1 -functions in Rn [0, ) satisfying


(2.3.1) and u0 be a C 1 -function in Rn . Suppose u is a C 2 -solution of (2.3.2)
in Rn [0, ). Prove that, for any P = (X, T ) Rn (0, ),

|u|C 1 (C (P )) C |u0 |C 1 ( C (P )) + |f |C 1 (C (P )) ,
where C is a positive constant depending only on T and the C 1 -norms of ai
and b in C (P ).
Exercise 2.12. Let a be a C 1 -function in R [0, ) satisfying
1
|a(x, t)| ,

and let bij be continuous in R [0, ), for i, j = 1, 2. Suppose (u, v) is a
C 1 -solution in R (0, ) of the rst-order dierential system
ut a(x, t)vx + b11 (x, t)u + b12 (x, t)v = f1 (x, t),
vt a(x, t)ux + b21 (x, t)u + b22 (x, t)v = f2 (x, t),
with
u(x, 0) = u0 (x), v(x, 0) = v0 (x).
Derive an L2 -estimate of (u, v) in appropriate cones.
Chapter 3

An Overview of
Second-Order PDEs

This chapter should be considered as an introduction to second-order linear


PDEs.
In Section 3.1, we introduce the notion of noncharacteristics for initial-
value problems. We proceed here for second-order linear PDEs as we did
for rst-order linear PDEs in Section 2.1. We show that we can compute all
derivatives of solutions on initial hypersurfaces if initial values are prescribed
on noncharacteristic initial hypersurfaces. We also introduce the Laplace
equation, the heat equation and the wave equation, as well as their general
forms, elliptic equations, parabolic equations and hyperbolic equations.
In Section 3.2, we discuss boundary-value problems for the Laplace equa-
tion and initial/boundary-value problems for the heat equation and the wave
equation. Our main tool is a priori estimates. For homogeneous boundary
values, we derive estimates of L2 -norms of solutions in terms of those of non-
homogeneous terms and initial values. These estimates yield uniqueness of
solutions and continuous dependence of solutions on nonhomogeneous terms
and initial values.
In Section 3.3, we use separation of variables to solve Dirichlet problems
for the Laplace equation in the unit disc in R2 and initial/boundary-value
problems for the 1-dimensional heat equation and the 1-dimensional wave
equation. We derive explicit expressions of solutions in Fourier series and
discuss the regularity of these solutions. Our main focus in this section is to
demonstrate dierent regularity patterns for solutions. Indeed, a solution
of the heat equation is smooth for all t > 0 regardless of the regularity of

47
48 3. An Overview of Second-Order PDEs

its initial values, while the regularity of a solution of the wave equation is
similar to the regularity of its initial values. Such a dierence in regularity
suggests that dierent methods are needed to study these two equations.

3.1. Classications
The main focus in this section is the second-order linear PDEs. We proceed
as in Section 2.1.
Let be a domain in Rn containing the origin and aij , bi and c be
continuous functions in , for i, j = 1, , n. Consider a second-order
linear dierential operator L dened by

n 
n
(3.1.1) Lu = aij (x)uxi xj + bi (x)uxi + c(x)u in .
i,j=1 i=1

Here aij , bi , c are called coecients of uxi xj , uxi , u, respectively. We usually


assume aij = aji , for any i, j = 1, , n. Hence, (aij ) is a symmetric matrix
in . For the operator L, we dene its principal symbol by

n
p(x; ) = aij (x)i j ,
i,j=1

for any x and Rn .


Let f be a continuous function in . We consider the equation
(3.1.2) Lu = f (x) in .
The function f is called the nonhomogeneous term of the equation. Let
be the hyperplane {xn = 0}. We now prescribe values of u and its normal
derivative on so that we can at least nd all derivatives of u at the origin.
Let u0 , u1 be functions dened in a neighborhood of the origin in Rn1 . Now
we prescribe
(3.1.3) u(x , 0) = u0 (x ), uxn (x , 0) = u1 (x ),
for any x Rn1 small. We call the initial hypersurface and u0 , u1 the
initial values or Cauchy values. The problem of solving (3.1.2) together with
(3.1.3) is called the initial-value problem or the Cauchy problem.
Let u be a C 2 -solution of (3.1.2) and (3.1.3) in a neighborhood of the
origin. In the following, we will investigate whether we can compute all
derivatives of u at the origin in terms of the equation and initial values. It
is obvious that we can nd all x -derivatives of u and uxn at the origin in
terms of those of u0 and u1 . In particular, we can nd all rst derivatives
and all second derivatives, except uxn xn , at the origin in terms of appropriate
3.1. Classications 49

derivatives of u0 and u1 . In fact,

uxi (0) = u0,xi (0) for i = 1, , n 1,


uxn (0) = u1 (0),

and

uxi xj (0) = u0,xi xj (0) for i, j = 1, , n 1,


uxi xn (0) = u1,xi (0) for i = 1, , n 1.

To compute uxn xn (0), we need to use the equation. We note that ann is the
coecient of uxn xn in (3.1.2). If we assume

(3.1.4) ann (0) = 0,

then by (3.1.2)

1
uxn xn (0) = aij (0)uxi xj (0)
ann (0)
(i,j)=(n,n)
 n 
+ bi (0)uxi (0) + c(0)u(0) f (0) .
i=1

Hence, we can compute all rst-order and second-order derivatives at 0 in


terms of the coecients and nonhomogeneous term in (3.1.2) and the initial
values u0 and u1 in (3.1.3). In fact, if all functions involved are smooth, we
can compute all derivatives of u of any order at the origin by using u0 , u1
and their derivatives and dierentiating (3.1.2). In summary, we can nd
all derivatives of u of any order at the origin under the condition (3.1.4),
which will be dened as the noncharacteristic condition later on.
Comparing the initial-value problem (3.1.2) and (3.1.3) here with the
initial-value problem (2.1.3) and (2.1.4) for rst-order PDEs, we note that
there is an extra condition in (3.1.3). This reects the general fact that two
conditions are needed for initial-value problems for second-order PDEs.
More generally, consider the hypersurface given by { = 0} for a
smooth function in a neighborhood of the origin with = 0. We note
that the vector eld is normal to the hypersurface at each point of
. We take a point on , say the origin. Then (0) = 0. Without loss of
generality, we assume xn (0) = 0. Then by the implicit function theorem,
we can solve = 0 for xn = (x1 , , xn1 ) in a neighborhood of the origin.
Consider the change of variables

x y = (x1 , , xn1 , (x)).


50 3. An Overview of Second-Order PDEs

This is a well-dened transformation with a nonsingular Jacobian in a neigh-


borhood of the origin. With

n
uxi = yk,xi uyk
k=1
and

n 
n
uxi xj = yk,xi yl,xj uyk yl + yk,xi xj uyk ,
k,l=1 k=1
we can write the operator L in the y-coordinates as

n n
Lu = aij yk,xi yl,xj uyk yl
k,l=1 i,j=1


n n 
n
+ bi yk,xi + aij yk,xi xj uyk + cu.
k=1 i=1 i,j=1

The initial hypersurface is given by {yn = 0} in the y-coordinates. With


yn = , the coecient of uyn yn is given by

n
aij (x)xi xj .
i,j=1

This is the principal symbol p(x; ) evaluated at = (x).


Denition 3.1.1. Let L be a second-order linear dierential operator as
in (3.1.1) in a neighborhood of x0 Rn and be a smooth hypersurface
containing x0 . Then is noncharacteristic at x0 if
n
(3.1.5) aij (x0 )i j = 0,
i,j=1

where = (1 , , n ) is normal to at x0 . Otherwise, it is characteristic


at x0 .

A hypersurface is noncharacteristic if it is noncharacteristic at every


point. Strictly speaking, a hypersurface is characteristic if it is not non-
characteristic, i.e., if it is characteristic at some point. In this book, we will
abuse this terminology. When we say a hypersurface is characteristic, we
mean it is characteristic everywhere. This should cause few confusions. In
R2 , hypersurfaces are curves, so we shall speak of characteristic curves and
noncharacteristic curves.
When the hypersurface is given by { = 0} with = 0, its normal
vector eld is given by = (x1 , , xn ). Hence we may take =
(x0 ) in (3.1.5). We note that the condition (3.1.5) is preserved under
3.1. Classications 51

C 2 -changes of coordinates. Using this condition, we can nd successively


the values of all derivatives of u at x0 , as far as they exist. Then, we could
write formal power series at x0 for solutions of initial-value problems. If the
initial hypersurface is analytic and the coecients, nonhomogeneous terms
and initial values are analytic, then this formal power series converges to an
analytic solution. This is the content of the Cauchy-Kovalevskaya theorem,
which we will discuss in Section 7.2.
Now we introduce a special class of linear dierential operators.

Denition 3.1.2. Let L be a second-order linear dierential operator as in


(3.1.1) dened in a neighborhood of x0 Rn . Then L is elliptic at x0 if

n
aij (x0 )i j = 0,
i,j=1

for any Rn \ {0}.

An operator L dened in is called elliptic in if it is elliptic at every


point in .
According to Denition 3.1.2, linear dierential operators are elliptic if
every hypersurface is noncharacteristic. We already assumed that (aij ) is
an n n symmetric matrix. Then L is elliptic at x0 if (aij (x0 )) is a denite
matrixpositive denite or negative denite.
We now turn our attention to second-order linear dierential equations
in R2 , where complete classications are available. Let be a domain in R2
and consider

2 
2
(3.1.6) Lu = aij (x)uxi xj + bi (x)uxi + c(x)u = f (x) in .
i,j=1 i=1

Here we assume (aij ) is a 2 2 symmetric matrix.

Denition 3.1.3. Let L be a dierential operator dened in a neighborhood


of x0 R2 as in (3.1.6). Then
(1) L is elliptic at x0 if det(aij (x0 )) > 0;
(2) L is hyperbolic at x0 if det(aij (x0 )) < 0;
(3) L is degenerate at x0 if det(aij (x0 )) = 0.

The operator L dened in R2 is called elliptic (or hyperbolic) in


if it is elliptic (or hyperbolic) at every point in .
It is obvious that the ellipticity dened in Denition 3.1.3 coincides with
that in Denition 3.1.2 for n = 2.
52 3. An Overview of Second-Order PDEs

For the operator L in (3.1.6), the symmetric matrix (aij ) always has two
(real) eigenvalues. Then
L is elliptic if the two eigenvalues have the same sign;
L is hyperbolic if the two eigenvalues have dierent signs;
L is degenerate if at least one of the eigenvalues vanishes.
The number of characteristic curves is determined by the type of the
operator. For the operator L in (3.1.6),
there are two characteristic curves if L is hyperbolic;
there are no characteristic curves if L is elliptic.

We shall study several important linear dierential operators in R2 . The


rst of these is the Laplacian. In R2 , the Laplace operator is dened by
u = ux1 x1 + ux2 x2 .
It is easy to see that the Laplace operator is elliptic. In polar coordinates
x1 = r cos , x2 = r sin ,
the Laplace operator can be expressed by
1 1
u = urr + ur + 2 u .
r r
The equation
u = 0
is called the Laplace equation and its solutions are called harmonic functions.
By writing x = x1 and y = x2 , we can associate with a harmonic function
u(x, y) a conjugate harmonic function v(x, y) such that u and v satisfy the
rst-order system of Cauchy-Riemann equations
u x = vy , uy = vx .
Any such a pair gives an analytic function
f (z) = u(x, y) + iv(x, y)
of the complex argument z = x + iy, if we identify C with R2 . Physically,
(u, v) is the velocity eld of an irrotational, incompressible ow. Con-
versely, for any analytic function f , functions u = Re f and v = Im f are
harmonic. In this way, we can nd many nontrivial harmonic functions
in the plane. For example, for any positive integer k, Re(x + iy)k and
Im(x + iy)k are homogeneous harmonic polynomials of degree k. Next, with
ez = ex+iy = ex cos y + iex sin y, we know ex cos y and ex sin y are harmonic
functions.
Although there are no characteristic curves for the Laplace operator,
initial-value problems are not well-posed.
3.1. Classications 53

Example 3.1.4. Consider the Laplace equation in R2


uxx + uyy = 0,
with initial values prescribed on {y = 0}. For any positive integer k, set
1
uk (x, y) = sin(kx)eky .
k
Then uk is harmonic. Moreover,
uk,x (x, y) = cos(kx)eky , uk,y (x, y) = sin(kx)eky ,
and hence
|uk (x, y)|2 = u2k,x (x, y) + u2k,y (x, y) = e2ky .
Therefore,
|uk (x, 0)| = 1 for any x R and any k,
and
|uk (x, y)| as k , for any x R and y > 0.
There is no continuous dependence on initial values in C 1 -norms.

In R2 , the wave operator  is given by


u = ux2 x2 ux1 x1 .
It is easy to see that the wave operator is hyperbolic. It is actually called the
one-dimensional wave operator. This is because the wave equation u =
0 in R2 represents vibrations of strings or propagation of sound waves in
tubes. Because of its physical interpretation, we write u as a function of two
independent variables x and t. The variable x is commonly identied with
position and t with time. Then the wave equation in R2 has the form
utt uxx = 0.
The two families of straight lines t = x + c, where c is a constant, are
characteristic.
The heat operator in R2 is given by
Lu = ux2 ux1 x1 .
This is a degenerate operator. The heat equation ux2 ux1 x1 = 0 is satised
by the temperature distribution in a heat-conducting insulated wire. As
with the wave operator, we refer to the one-dimensional heat operator and
we write u as a function of the independent variables x and t. Then the
heat equation in R2 has the form
ut uxx = 0.
It is easy to see that {t = 0} is characteristic. If we prescribe u(x, 0) =
u0 (x) in an interval of {t = 0}, then using the equation we can compute all
derivatives there. However, u0 does not determine a unique solution even
54 3. An Overview of Second-Order PDEs

in a neighborhood of this interval. We will see later on that we need initial


values on the entire initial line {t = 0} to compute local solutions.
Many important dierential operators do not have a denite type. In
other words, they are neither elliptic nor hyperbolic in the domain where
they are dened. We usually say a dierential operator is of mixed type if it
is elliptic in a subdomain and hyperbolic in another subdomain. In general,
it is more dicult to study equations of mixed type.
Example 3.1.5. Consider the Tricomi equation
ux2 x2 + x2 ux1 x1 = f in R2 .
It is elliptic if x2 > 0, hyperbolic if x2 < 0 and degenerate if x2 = 0.

Characteristic curves also arise naturally in connection with the propa-


gation of singularities. We consider a simple case.
Let be a domain in R2 , be a continuous curve in and w be a
continuous function in \ . For simplicity, we assume divides into two
parts, + and . Take a point x0 . Then w is said to have a jump at
x0 across if the two limits
w (x0 ) = lim w(x), w+ (x0 ) = lim w(x)
xx0 , x xx0 , x+

exist. The dierence


[w](x0 ) = w+ (x0 ) w (x0 )
is called the jump of w at x0 across . The function w has a jump across
if it has a jump at every point of across . If w has a jump across , then
[w] is a well-dened function on . It is easy to see that [w] = 0 on if w
is continuous in .
Proposition 3.1.6. Let be a domain in R2 and be a C 1 -curve in
dividing into two parts. Suppose aij , bi , c, f are continuous functions in
and u C 1 () C 2 ( \ ) satises

2 
2
aij uxi xj + bi uxi + cu = f in \ .
i,j=1 i=1

If 2 u has a jump across , then is a characteristic curve.

Proof. Since u is C 1 in , we have


[u] = [ux1 ] = [ux2 ] = 0 on .
Let the vector eld (1 , 2 ) be normal to . Then 2 x1 1 x2 is a directional
derivative along . Hence on
(2 x1 1 x2 )[ux1 ] = 0,
3.1. Classications 55

and
(2 x1 1 x2 )[ux2 ] = 0.
Then we conclude
2 [ux1 x1 ] 1 [ux1 x2 ] = 0,
and
2 [ux1 x2 ] 1 [ux2 x2 ] = 0.
By the continuity of aij , bi , c and f in , we have
a11 [ux1 x1 ] + 2a12 [ux1 x2 ] + a22 [ux2 x2 ] = 0 on .
Thus, the nontrivial vector ([ux1 x1 ], [ux1 x2 ], [ux2 x2 ]) satises a 3 3 homoge-
neous linear system on . Hence the coecient matrix is singular. That is,
on
2 1 0
det 0 2 1 = 0,
a11 2a12 a22
or
a11 12 + 2a12 1 2 + a22 22 = 0.
This yields the desired result. 

The Laplace operator, the wave operator and the heat operator can be
generalized to higher dimensions.
Example 3.1.7. The n-dimensional Laplace operator in Rn is dened by

n
u = uxi xi ,
i=1
and the Laplace equation is given by u = 0. Solutions are called harmonic
functions. The principal symbol of the Laplace operator is given by
p(x; ) = ||2 ,
for any Rn . Obviously, is elliptic. Note that is invariant under
rotations. If x = Ay for an orthogonal matrix A, then

n 
n
uxi xi = uy i y i .
i=1 i=1
For a nonzero function f , we call the equation u = f the Poisson equation.
The Laplace equation has a wide variety of physical backgrounds. For
example, let u denote a temperature in equilibrium in a domain Rn
with the ux density F. Then for any smooth subdomain  , the net
ux of u through  is zero, i.e.,

F dS = 0,

56 3. An Overview of Second-Order PDEs

where is the unit exterior normal vector to  . Upon integration by parts,


we obtain  
div F dx = F dS = 0.
 
This implies
div F = 0 in ,
since  is arbitrary. In a special case where the ux F is proportional to
the gradient u, we have
F = au,
for a positive constant a. Here the negative sign indicates that the ow is
from regions of higher temperature to those of lower one. Now a simple
substitution yields the Laplace equation
u = div(u) = 0.
Example 3.1.8. We denote points in Rn+1 by (x1 , , xn , t). The heat
operator in Rn+1 is given by
Lu = ut x u.
It is often called the n-dimensional heat operator. Its principal symbol is
given by
p(x, t; , ) = ||2 ,
for any Rn and R. A hypersurface {(x1 , , xn , t) = 0} is non-
characteristic for the heat operator if, at each of its points,
|x |2 = 0.
Likewise, a hypersurface {(x, t) = 0} is characteristic if x = 0 and t =
0 at each of its points. For example, any horizontal hyperplane {t = t0 }, for
a xed t0 R, is characteristic.
The heat equation describes the evolution of heat. Let u denote a tem-
perature in a domain Rn with the ux density F. Then for any smooth
subdomain  , the rate of change of the total quantity in  equals the
negative of the net ux of u through  , i.e.,
 
d
u dx = F dS,
dt  

where is the unit exterior normal vector to  . Upon integration by parts,


we obtain  
d
u dx = div F dx.
dt  
This implies
ut = div F in ,
3.1. Classications 57

since  is arbitrary. In a special case where the ux F is proportional to


the gradient u, we have
F = au,
for a positive constant a. Now a simple substitution yields
ut = a div(u) = au.
This is the heat equation if a = 1.
Example 3.1.9. We denote points in Rn+1 by (x1 , , xn , t). The wave
operator  in Rn+1 is given by
u = utt x u.
It is often called the n-dimensional wave operator. Its principal symbol is
given by
p(x, t; , ) = 2 ||2 ,
for any Rn and R. A hypersurface {(x1 , , xn , t) = 0} is non-
characteristic for the wave operator if, at each of its points,
2t |x |2 = 0.
For any (x0 , t0 ) Rn R, the surface
|x x0 |2 = (t t0 )2
is characteristic except at (x0 , t0 ). We note that this surface, smooth except
at (x0 , t0 ), is the union of two cones. It is usually called the characteristic
cone.

@
@
@
@
@
@ (x0 , t0 )
@
@
@
@
@@

Figure 3.1.1. The characteristic cone.

To interpret the wave equation, we let u(x, t) denote the displacement


in some direction of a point x Rn at time t 0. For any smooth
subdomain  , Newtons law asserts that the product of mass and the
acceleration equals the net force, i.e.,
 
d2
u dx = F dS,
dt2  
58 3. An Overview of Second-Order PDEs

where F is the force acting on  through  and the mass density is taken
to be 1. Upon integration by parts, we obtain
 
d2
u dx = div F dx.
dt2  
This implies
utt = div F in ,
since  is arbitrary. In a special case F = au for a positive constant a,
we have
utt = a div(u) = au.
This is the wave equation if a = 1.

The heat equation and the wave equation can be generalized to para-
bolic equations and hyperbolic equations in arbitrary dimensions. Again,
we denote points in Rn+1 by (x1 , , xn , t). Let aij , bi , c and f be functions
dened in a domain in Rn+1 , i, j = 1, , n. We assume (aij ) is an n n
positive denite matrix in this domain. An equation of the form

n 
n
ut aij (x, t)uxi xj + bi (x, t)uxi + c(x, t)u = f (x, t)
i=1 i=1

is parabolic, and an equation of the form



n 
n
utt aij (x, t)uxi xj + bi (x, t)uxi + c(x, t)u = f (x, t)
i=1 i=1

is hyperbolic.

3.2. Energy Estimates


In this section, we discuss the uniqueness of solutions of boundary-value
problems for the Laplace equation and initial/boundary-value problems for
the heat equation and the wave equation. Our main tool is the energy
estimates. Specically, we derive estimates of L2 -norms of solutions in terms
of those of boundary values and/or initial values.
We start with the Laplace equation. Let Rn be a bounded C 1 -
domain and be a continuous function on . Consider the Dirichlet
boundary-value problem for the Laplace equation:
u = 0 in ,
u = on .
We now prove that a C 2 -solution, if it exists, is unique. To see this, let u1
and u2 be solutions in C 2 () C 1 (). Then, the dierence w = u1 u2
3.2. Energy Estimates 59

satises
w = 0 in ,
w=0 on .
We multiply the Laplace equation by w and write the resulting product as

n
0 = ww = (wwxi )xi |w|2 .
i=1

An integration by parts in yields


 
w
0 = |w| dx +2
w dS.

With the homogeneous boundary value w = 0 on , we have

|w|2 dx = 0,

and then w = 0 in . Hence w is constant and this constant is zero since w
is zero on the boundary. Obviously, the argument above applies to Dirichlet
problems for the Poisson equation. In general, we have the following result.
Lemma 3.2.1. Let Rn be a bounded C 1 -domain, f be a continuous
function in and be a continuous function on . Then there exists at
most one solution in C 2 () C 1 () of the Dirichlet problem
u = f in ,
u= on .

By the maximum principle, the solution is in fact unique in C 2 ()C(),


as we will see in Chapter 4.
Now we discuss briey the Neumann boundary-value problem, where
we prescribe normal derivatives on the boundary. Let be a continuous
function on . Consider
u = 0 in ,
u
= on .

We can prove similarly that solutions are unique up to additive constants
if is connected. We note that if there exists a solution of the Neumann
problem, then necessarily satises

dS = 0.

This can be seen easily upon integration by parts.
60 3. An Overview of Second-Order PDEs

Next, we derive an estimate of a solution of the Dirichlet boundary-value


problem for the Poisson equation. We need the following result, which is
referred to as the Poincare lemma.
Lemma 3.2.2. Let be a bounded C 1 -domain in Rn and u be a C 1 -function
in with u = 0 on . Then
uL2 () diam()uL2 () .

Here diam() denotes the diameter of and is dened by


diam() = sup |x y|.
x,y

Proof. We write Rn = Rn1 R. For any x0 Rn1 , let lx0 be the straight
line containing x0 and normal to Rn1 {0}. Consider those x0 such that
lx0 = . Now lx0 is the union of a countable collection of pairwise
disjoint open intervals. Let I be such an interval. Then I and I has
the form
I = {(x0 , xn ) : a < xn < b},
where (x0 , a), (x0 , b) . Since u(x0 , a) = 0, then
 xn

u(x0 , xn ) = uxn (x0 , s)ds for any xn (a, b).
a

xn 6
lx0

-
x0 Rn1

Figure 3.2.1. An integration along lx0 .

The Cauchy inequality yields


 xn
u 2
(x0 , xn ) (xn a) u2xn (x0 , s) ds for any xn (a, b).
a
By a simple integration along I, we have
 b  b
u2 (x0 , xn ) dxn (b a)2 u2xn (x0 , xn ) dxn .
a a
3.2. Energy Estimates 61

By adding the integrals over all such intervals, we obtain


 
2 
u (x0 , xn ) dxn Cx
2
u2xn (x0 , xn ) dxn ,
0
lx lx
0 0

where Cx0 is the length of lx0 in . Now a simple integration over x0 yields
the desired result. 

Now consider
u = f in ,
(3.2.1)
u = 0 on .
We note that u has a homogeneous Dirichlet boundary value on .
Theorem 3.2.3. Let Rn be a bounded C 1 -domain and f be a contin-
uous function in . Suppose u C 2 () C 1 () is a solution of (3.2.1).
Then
uL2 () + uL2 () Cf L2 () ,
where C is a positive constant depending only on .

Proof. Multiply the equation in (3.2.1) by u and write the resulting product
in the left-hand side as
n
uu = (uuxi )xi |u|2 .
i=1
Upon integrating by parts in , we obtain
  
u
u dS |u| dx =
2
uf dx.

With u = 0 on , we have
 
|u| dx =
2
uf dx.

The Cauchy inequality yields
 2  2  
|u|2 dx = uf dx u2 dx f 2 dx.

By Lemma 3.2.2, we get
 
2
|u| dx diam()
2
f 2 dx.

Using Lemma 3.2.2 again, we then have
 
4
u dx diam()
2
f 2 dx.

This yields the desired estimate. 
62 3. An Overview of Second-Order PDEs

Now we study initial/boundary-value problems for the heat equation.


Suppose is a bounded C 1 -domain in Rn , f is continuous in [0, ) and
u0 is continuous in . Consider
ut u = f in (0, ),
(3.2.2) u(, 0) = u0 in ,
u=0 on (0, ).
The geometric boundary of (0, ) consists of three parts, {0},
(0, ) and {0}. We treat {0} and (0, ) dierently
and refer to values prescribed on {0} and (0, ) as initial values
and boundary values, respectively. Problems of this type are usually called
initial/boundary-value problems or mixed problems. We note that u has a
homogeneous Dirichlet boundary value on (0, ). We now derive an
estimate of the L2 -norm of a solution. For each t 0, we denote by u(t)
the function dened on by u(t) = u(, t).
Theorem 3.2.4. Let be a bounded C 1 -domain in Rn , f be continuous in
[0, ) and u0 be continuous in . Suppose u C 2 ( (0, )) C 1 (
[0, )) is a solution of (3.2.2). Then
 t
u(t)L2 () u0 L2 () + f (s)L2 () ds for any t > 0.
0

Theorem 3.2.4 yields the uniqueness of solutions of (3.2.2). In fact, if


f 0 and u0 0, then u 0. We also have the continuous dependence
on initial values in L2 -norms. Let f1 , f2 be continuous in [0, ) and
u01 , u02 be continuous in . Suppose u1 , u2 C 2 ((0, ))C 1 ([0, ))
are solutions of (3.2.2) with f1 , u01 and f2 , u02 replacing f, u0 , respectively.
Then for any t > 0,
 t
u1 (t) u2 (t)L2 () u01 u02 L2 () + f1 (s) f2 (s)L2 () ds.
0

Proof. We multiply the equation in (3.2.2) by u and write the product in


the left-hand side as
1 
n
uut uu = (u2 )t (uuxi )xi + |u|2 .
2
i=1
Upon integration by parts in for each xed t > 0 and u(t) = 0 on , we
have   
1d 2
u (t) dx + |u(t)| dx =
2
f (t)u(t) dx.
2 dt
An integration in t yields, for any t > 0,
  t   t
u2 (t) dx + 2 |u|2 dxds = u20 dx + 2 f u dxds.
0 0
3.2. Energy Estimates 63

Set
E(t) = u(t)L2 () .
Then
 t  t
2 2
E(t) + 2 |u| dxds = E(0) + 2
2
f u dxds.
0 0
Dierentiating with respect to t, we have
 
 
2E(t)E (t) 2E(t)E (t) + 2 |u(t)| dx = 2 f (t)u(t) dx
2

2f (t)L2 () u(t)L2 () = 2E(t)f (t)L2 () .
Hence
E  (t) f (t)L2 () .
Integrating from 0 to t gives the desired estimate. 

Now we study initial/boundary-value problems for the wave equation.


Suppose is a bounded C 1 -domain in Rn , f is continuous in [0, ), u0
is C 1 in and u1 is continuous in . Consider
utt u = f in (0, ),
(3.2.3) u(, 0) = u0 , ut (, 0) = u1 in ,
u = 0 on (0, ).
Comparing (3.2.3) with (3.2.2), we note that there is an extra initial condi-
tion on ut in (3.2.3). This relates to the extra order of the t-derivative in
the wave equation.
Theorem 3.2.5. Let be a bounded C 1 -domain in Rn , f be continuous
in [0, ), u0 be C 1 in and u1 be continuous in . Suppose u
C 2 ( (0, )) C 1 ( [0, )) is a solution of (3.2.3). Then for any
t > 0,
1 1
ut (t)2L2 () + x u(t)2L2 () 2 u1 2L2 () + x u0 2L2 () 2
 t
+ f (s)L2 () ds,
0
and
1
u(t)L2 () u0 L2 () + t u1 2L2 () + x u0 2L2 () 2
 t
+ (t s)f (s)L2 () ds.
0

As a consequence, we also have the uniqueness and continuous depen-


dence on initial values in L2 -norms.
64 3. An Overview of Second-Order PDEs

Proof. Multiply the equation in (3.2.3) by ut and write the resulting prod-
uct in the left-hand side as
1 2 
n
ut utt ut u = (ut + |u| )t
2
(ut uxi )xi .
2
i=1
Upon integration by parts in for each xed t > 0, we obtain
  
1d 2 u
ut (t) + |x u(t)|2 dx ut (t) (t) dS = f (t)ut (t) dx.
2 dt
Note that ut = 0 on (0, ) since u = 0 on (0, ). Then
 
1 d 2
ut (t) + |x u(t)|2 dx = f (t)ut (t) dx.
2 dt
Dene the energy by

2
E(t) = ut (t) + |x u(t)|2 dx.

If f 0, then
d
E(t) = 0.
dt
Hence for any t > 0,

1 2
E(t) = E(0) = u1 + |x u0 |2 dx.
2
This is the conservation of energy. In general,
 t
E(t) = E(0) + 2 f ut dxds.
0
To get an estimate of E(t), set
1
J(t) = E(t) 2 .
Then  t
2 2
J(t) = J(0) + 2 f ut dxds.
0
By dierentiating with respect to t and applying the Cauchy inequality, we
get

2J(t)J  (t) = 2 f (t)ut (t) dx 2f (t)L2 () ut (t)L2 ()

2J(t)f (t)L2 () .
Hence for any t > 0,
J  (t) f (t)L2 () .
Integrating from 0 to t, we obtain
 t
J(t) J(0) + f (s)L2 () ds.
0
3.2. Energy Estimates 65

This is the desired estimate for the energy. Next, to estimate the L2 -norm
of u, we set
F (t) = u(t)L2 () ,
i.e., 
2
F (t) = u2 (t) dx.

A simple dierentiation yields


2F (t)F (t) = 2 u(t)ut (t) dx 2u(t)L2 () ut (t)L2 ()

= 2F (t)ut (t)L2 () .
Hence  t

F (t) ut L2 () J(0) + f (s)L2 () ds.
0
Integrating from 0 to t, we have
 t t
u(t)L2 () u0 L2 () + tJ(0) + f (s)L2 () dsdt .
0 0
By interchanging the order of integration in the last term in the right-hand
side, we obtain the desired estimate on u. 

There are other forms of estimates on energies. By squaring the rst


estimate in Theorem 3.2.5 and applying the Cauchy inequality, we obtain
 
2 2
ut (t) + |x u(t)|2 dx 2 u1 + |x u0 |2 dx

 t
+ 2t f 2 dxds.
0
Integrating from 0 to t, we get
 t 
2 2
ut + |x u|2 dxds 2t u1 + |x u0 |2 dxds
0
 t
2
+t f 2 dxds.
0

Next, we briey review methods used in deriving estimates in Theorems


3.2.33.2.5. In the proofs of Theorems 3.2.33.2.4, we multiply the Laplace
equation and the heat equation by u and integrate the resulting product
over , while in the proof of Theorem 3.2.5, we multiply the wave equation
by ut and integrate over . It is important to write the resulting product
as a linear combination of u2 , u2t , |u|2 and their derivatives. Upon inte-
grating by parts, domain integrals of derivatives are reduced to boundary
integrals. Hence, the resulting integral identity consists of domain integrals
and boundary integrals of u2 , u2t and |u|2 . Second-order derivatives of
66 3. An Overview of Second-Order PDEs

u are eliminated. These strategies also work for general elliptic equations,
parabolic equations and hyperbolic equations. Compare methods in this
section with those used to obtain L2 -estimates of solutions of initial-value
problems for rst-order linear PDEs in Section 2.3.
To end this section, we discuss an elliptic dierential equation in the
entire space. Let f be a continuous function in Rn . We consider

(3.2.4) u + u = f in Rn .

Let u be a C 2 -solution in Rn . Next, we demonstrate that we can obtain


estimates of L2 -norms of u and its derivatives under the assumption that u
and its derivatives decay suciently fast at innity.
To obtain an estimate of u and its rst derivatives, we multiply (3.2.4)
by u. In view of

n
uu = (uuxk )xk |u|2 ,
k=1

we write the resulting product as


n
|u|2 + u2 (uuxk )xk = f u.
k=1

We now integrate in Rn . Since u and uxk decay suciently fast at innity,


we have
 
2 2
(|u| + u ) dx = f u dx.
Rn Rn

Rigorously, we need to integrate in BR and let R after integrating by


parts. By the Cauchy inequality, we get
  
1 1
f u dx 2
u dx + f 2 dx.
Rn 2 Rn 2 Rn

A simple substitution yields


 
(2|u| + u ) dx
2 2
f 2 dx.
Rn Rn

Hence, the L2 -norm of f controls the L2 -norms of u and u. In fact, the


L2 -norm of f also controls the L2 -norms of the second derivatives of u. To
see this, we take square of the equation (3.2.4) to get

(u)2 2uu + u2 = f 2 .
3.3. Separation of Variables 67

We note that

n 
n 
n
(u)2 = uxk xk uxl xl = (uxk xk uxl )xl uxk xk xl uxl
k,l=1 k,l=1 k,l=1
n  n  n
= (uxk xk uxl )xl (uxk xl uxl )xk + u2xk xl .
k,l=1 k,l=1 k,l=1

Hence

n 
n
| u| + 2|u| + u +
2 2 2 2
(uxk xk uxl )xl (uxk xl uxl )xk
k,l=1 k,l=1

n
2 (uuxk )xk = f 2 .
k=1
Integration in Rn yields
 
2 2 2 2
(3.2.5) (| u| + 2|u| + u ) dx = f 2 dx.
Rn Rn
Therefore, the L2 -norm of f controls the L2 -norms of every second deriva-
tives of u, although f is related to u by u, which is just one particular
combination of second derivatives. As we will see, this is the feature of
elliptic dierential equations.
We need to point out that it is important to assume that u and its
derivatives decay suciently fast. Otherwise, the integral identity (3.2.5)
does not hold. By taking f = 0, we obtain u = 0 from (3.2.5) if u and its
derivatives decay suciently fast. We note that u(x) = ex1 is a nonzero
solution of (3.2.4) for f = 0.

3.3. Separation of Variables


In this section, we solve boundary-value problems for the Laplace equation
and initial/boundary-value problems for the heat equation and the wave
equation in the plane by separation of variables.

3.3.1. Dirichlet Problems. In this subsection we use the method of sep-


aration of variables to solve the Dirichlet problem for the Laplace equation
in the unit disc in R2 . We will use polar coordinates
x = r cos , y = r sin
in R2 , and we will build up solutions from functions that depend only on
r and functions that depend only on . Our rst step is to determine all
harmonic functions u in R2 having the form
u(r, ) = f (r)g(),
68 3. An Overview of Second-Order PDEs

where f is dened for r > 0 and g is dened on S1 . (Equivalently, we can


view g as a 2-periodic function dened on R.) Then we shall express the
solution of a Dirichlet problem as the sum of a suitably convergent innite
series of functions of this form.
In polar coordinates, the Laplace equation is
1 1
u (rur )r + 2 u = 0.
r r
Thus the function u(r, ) = f (r)g() is harmonic if and only if
1   1
rf (r) g() + 2 f (r)g  () = 0,
r r
that is, 
1 1
f  (r) + f  (r) g() + 2 f (r)g  () = 0.
r r
When u(r, ) = 0, this equation is equivalent to

r2 1 g  ()
f  (r) + f  (r) = .
f (r) r g()
The left-hand side of this equation depends only on r and the right-hand
side depends only on . Thus there is a constant such that

r2  1  g  ()
f (r) + f (r) = = .
f (r) r g()
Hence
1
f  (r) + f  (r) 2 f (r) = 0 for r > 0,
r r
and
g  () + g() = 0 for S1 .
Our next step is to analyze the equation for g. Then we shall recall
some facts about Fourier series, after which we shall turn to the equation
d2
for f . The equation for g describes the eigenvalue problem for d 2 on S .
1

This equation has nontrivial solutions when = k , k = 0, 1, 2, . When


2

= 0, the general solution is g() = a0 , where a0 is a constant. For = k 2 ,


k = 1, 2, , the general solution is
g() = ak cos k + bk sin k,
where ak and bk are constants. Moreover, the normalized eigenfunctions
1 1 1
, cos k, sin k, k = 1, 2, ,
2
form an orthonormal basis for L2 (S1 ). In other words, for any v L2 (S1 ),

1 1 
v() = a0 + (ak cos k + bk sin k) ,
2
k=1
3.3. Separation of Variables 69

where 
1
a0 = v() d,
2 S1
and for k = 1, 2, ,

1
ak = v() cos k d,
S1

1
bk = v() sin k d.
S1
This series for v is its Fourier series and a0 , ak , bk are its Fourier coecients.
The series converges in L2 (S1 ). Moreover,

1
 2

vL2 (S1 ) = a0 +
2 2
(ak + bk )2
.
k=1

As for f , when = 0 the general solution is


f (r) = c0 + d0 log r,
where c0 and d0 are constants. Now we want u(r, ) = f (r)g() to be
harmonic in R2 , thus f must remain bounded as r tends to 0. Therefore we
must have d0 = 0, and so
f (r) = c0
is a constant function. For = k 2 , k = 1, 2, , the general solution is
f (r) = ck rk + dk rk ,
where ck and dk are constants. Again f must remain bounded as r tends to
0, so dk = 0 and
f (r) = ck rk .
In summary, a harmonic function u in R2 of the form u(r, ) = f (r)g() is
given by
u(r, ) = a0 ,
or by
u(r, ) = ak rk cos k + bk rk sin k,
for k = 1, 2, , where a0 , ak , bk are constants.
Remark 3.3.1. Note that rk cos k and rk sin k are homogenous harmonic
polynomials of degree k in R2 . Taking z = x + iy, we see that
rk cos k + irk sin k = rk eik = (x + iy)k ,
and hence
rk cos k = Re(x + iy)k , rk sin k = Im(x + iy)k .
70 3. An Overview of Second-Order PDEs

Now, we are ready to solve the Dirichlet problem for the Laplace equa-
tion in the unit disc B1 R2 . Let be a function on B1 = S1 and
consider
u = 0 in B1 ,
(3.3.1)
u = on S1 .

We rst derive an expression for the solution purely formally. We seek


a solution of the form

1 1  k
(3.3.2) u(r, ) = a0 + ak r cos k + bk rk sin k .
2
k=1

The terms in the series are all harmonic functions of the form f (r)g() that
we discussed above. Thus the sum u(r, ) should also be harmonic. Letting
r = 1 in (3.3.2), we get

1 1 
() = u(1, ) = a0 + ak cos k + bk sin k .
2
k=1

Therefore, the constants a0 , ak and bk , k = 1, 2, , should be the Fourier


coecients of . Hence,

1
(3.3.3) a0 = () d,
2 S1
and for k = 1, 2, ,

1
ak = () cos k d,
S1
(3.3.4) 
1
bk = () sin k d.
S1
Theorem 3.3.2. Suppose L2 (S1 ) and u is given by (3.3.2), (3.3.3) and
(3.3.4). Then u is smooth in B1 and satises
u = 0 in B1 .
Moreover,
lim u(r, ) L2 (S1 ) = 0.
r1

Proof. Since L2 (S1 ), we have




2L2 (S1 ) = a20 + (a2k + b2k ) < .
k=0

In the following, we x an R (0, 1).


3.3. Separation of Variables 71

First, we set


S00 (r, ) = |ak rk cos k + bk rk sin k|.
k=1
By (3.3.2), we have
1 1
|u(r, )| |a0 | + S00 (r, ).
2
To estimate S00 , we note that, for any r [0, R] and any S1 ,

 1
S00 (r, ) a2k + b2k 2
Rk .
k=1
By the Cauchy inequality, we get
 1  1
 2  2

S00 (r, ) 2 2
ak + b k R 2k
< .
k=1 k=1
Hence, the series dening u in (3.3.2) is convergent absolutely and uniformly
in BR . Therefore, u is continuous in BR .
Next, we take any positive integer m and any nonnegative integers m1
and m2 with m1 + m2 = m. For any r [0, R] and any S1 , we have
formally

1  m1 m2 k
xm1 ym2 u(r, ) = x y ak r cos k + bk rk sin k .

k=1
In order to justify the interchange of the order of dierentiation and sum-
mation, we need to prove that the series in the right-hand side is convergent
absolutely and uniformly in BR . Set

  m1 m2 k 
(3.3.5) Sm1 m2 (r, ) = x y ak r cos k + bk rk sin k  .
k=1
(We note that this is S00 dened earlier if m1 = m2 = 0.) By using rectan-
gular coordinates, it is easy to check that, for k < m,

xm1 ym2 ak rk cos k + bk rk sin k = 0,
and for k m,
 m m k 
 1 2 ak r cos k + bk rk sin k 
x y
1
k(k 1) (k m + 1) a2k + b2k 2 rkm .
Hence, for any r [0, R] and any S1 ,

 1
Sm1 m2 (r, ) a2k + b2k 2
k m Rkm .
k=m
72 3. An Overview of Second-Order PDEs

By the Cauchy inequality, we have


 1  1
 2  2

Sm1 m2 (r, ) a2k + b2k k 2m R2(km) < .


k=m k=m

This veries that the series dening xm1 ym2 u is convergent absolutely and
uniformly in BR , for any m1 and m2 with m1 + m2 1. Hence, u is smooth
in BR for any R < 1 and all derivatives of u can be obtained from term-by-
term dierentiation in (3.3.2). Then it is easy to conclude that u = 0.
We now prove the L2 -convergence. First, by the series expansions of u
and , we have

1 
u(r, ) () = ak cos k + bk sin k (rk 1),

k=1

and then

 2
|u(r, ) ()|2 d = (a2k + b2k ) rk 1 .
S1 k=1

We note that 1 as r 1 for each xed k 1. For a positive integer


rk
K to be determined, we write
 
K
2
|u(r, ) ()|2 d = (a2k + b2k ) rk 1
S1 k=1

 2
+ (a2k + b2k ) rk 1 .
k=K+1

For any > 0, there exists a positive integer K = K() such that


(a2k + b2k ) < .
k=K+1

Then there exists a > 0, depending on and K, such that



K
2
(a2k + b2k ) rk 1 < for any r (1 , 1),
k=1

since the series in the left-hand side consists of nitely many terms. There-
fore, we obtain

|u(r, ) ()|2 d < 2 for any r (1 , 1).
S1

This implies the desired L2 -convergence as r 1. 


3.3. Separation of Variables 73

We note that u is smooth in B1 even if the boundary value is only L2 .


Naturally, we ask whether u in Theorem 3.3.2 is continuous up to B1 , or,
more generally, whether u is smooth up to B1 . We note that a function is
smooth in B1 if all its derivatives are continuous in B.
Theorem 3.3.3. Suppose C (S1 ) and u is given by (3.3.2), (3.3.3)
and (3.3.4). Then u is smooth in B1 with u(1, ) = .

Proof. Let m1 and m2 be nonnegative integers with m1 + m2 = m. We


need to prove that the series dening xm1 ym2 u(r, ) converges absolutely
and uniformly in B1 . Let Sm1 m2 be the series dened in (3.3.5). Then for
any r [0, 1] and S1 ,

 2 1
Sm1 m2 (r, ) ak + b2k 2 k m .
k=1
To prove that the series in the right-hand side is convergent, we need to
improve estimates of ak and bk , k = 1, 2, . By denitions of ak and bk in
(3.3.4) and integrations by parts, we have
 
1 1 sin k
ak = () cos k d =  () d,
S1 S1 k
 
1 1 cos k
bk = () sin k d =  () d.
S1 S1 k
Hence, {kbk , kak } are the coecients of the Fourier series of  , so

k 2 (a2k + b2k ) =  2L2 (S1 ) .
k=1
By continuing this process, we obtain for any positive integer ,

k 2 (a2k + b2k ) = () 2L2 (S1 ) < .
k=1
Hence by the Cauchy inequality, we have for any r [0, 1] and S1 ,

 2 1
Sm1 m2 (r, ) ak + b2k 2 k m
k=1

1  1
 2 2  2
2
k 2(m+1) 2
ak + b k k .
k=1 k=1
This implies
Sm1 m2 (r, ) Cm (m+1) L2 (S1 ) ,
where Cm is a positive constant depending only on m. Then the series
dening xm1 ym2 u converges absolutely and uniformly in B1 . Therefore,
xm1 ym1 u is continuous in B1 . 
74 3. An Overview of Second-Order PDEs

By examining the proofs of Theorem 3.3.2 and Theorem 3.3.3, we have


the following estimates. For any integer m 0 and any R (0, 1),
uC m (BR ) Cm,R L2 (S1 ) ,
where Cm,R is a positive constant depending only on m and R. This estimate
controls the C m -norm of u in BR in terms of the L2 -norm of on S1 . It is
referred to as an interior estimate. Moreover, for any integer m 0,

m+1
uC m (B1 ) Cm (i) L2 (S1 ) ,
i=0
where Cm is a positive constant depending only on m. This is referred to as
a global estimate.
If we are interested only in the continuity of u up to B1 , we have the
following result.
Corollary 3.3.4. Suppose C 1 (S1 ) and u is given by (3.3.2), (3.3.3) and
(3.3.4). Then u is smooth in B1 , continuous in B1 and satises (3.3.1).

Proof. It follows from Theorem 3.3.2 that u is smooth in B1 and satises


u = 0 in B1 . The continuity of u up to B1 follows from the proof of
Theorem 3.3.3 with m1 = m2 = 0. 

The regularity assumption on in Corollary 3.3.4 does not seem to be


optimal. It is natural to ask whether it suces to assume that is in C(S1 )
instead of C 1 (S1 ). To answer this question, we need to analyze pointwise
convergence of Fourier series. We will not pursue along this direction in this
book. An alternative approach is to rewrite the solution u in (3.3.2). With
the explicit expressions of a0 , ak , bk in terms of as in (3.3.3) and (3.3.4),
we can write

(3.3.6) u(r, ) = K(r, , )() d,
S1
where

1 1 k
K(r, , ) = + r cos k( ).
2
k=1
The integral expression (3.3.6) is called the Poisson integral formula and the
function K is called the Poisson kernel. We can verify that
1 1 r2
(3.3.7) K(r, , ) = .
2 1 2r cos( ) + r2
We leave this verication as an exercise. In Section 4.1, we will prove that
u is continuous up to B1 if is continuous on B1 . In fact, we will derive
Poisson integral formulas for arbitrary dimension and prove they provide
3.3. Separation of Variables 75

solutions of Dirichlet problems for the Laplace equation in balls with con-
tinuous boundary values.
Next, we compare the regularity results in Theorems 3.3.23.3.3. For
Dirichlet problems for the Laplace equation in the unit disc, solutions are
always smooth in B1 even with very weak boundary values, for example,
with L2 -boundary values. This is the interior smoothness, i.e., solutions are
always smooth inside the domain regardless of the regularity of boundary
values. Moreover, solutions are smooth up to the boundary if boundary
values are also smooth. This is the global smoothness.

3.3.2. Initial/Boundary-Value Problems. In the following, we solve


initial/boundary-value problems for the 1-dimensional heat equation and
the 1-dimensional wave equation by separation of variables, and discuss
regularity of these solutions. We denote by (x, t) points in [0, ] [0, ),
with x identied as the space variable and t as the time variable.
We rst discuss the 1-dimensional heat equation. Let u0 be a continuous
function in [0, ]. Consider the initial/boundary-value problem
ut uxx = 0 in (0, ) (0, ),
(3.3.8) u(x, 0) = u0 (x) for x (0, ),
u(0, t) = u(, t) = 0 for t (0, ).
Physically, u represents the temperature in an insulated rod with ends kept
at zero temperature.
We rst consider
ut uxx = 0 in (0, ) (0, ),
(3.3.9)
u(0, t) = u(, t) = 0 for t (0, ).
We intend to nd its solutions by separation of variables. Set
u(x, t) = a(t)w(x) for (x, t) (0, ) (0, ).
Then
a (t)w(x) a(t)w (x) = 0,
and hence
a (t) w (x)
= .
a(t) w(x)
Since the left-hand side is a function of t and the right-hand side is a function
of x, there is a constant such that each side is . Then
a (t) + a(t) = 0 for t (0, ),
and
w (x) + w(x) = 0 for x (0, ),
(3.3.10)
w(0) = w() = 0.
76 3. An Overview of Second-Order PDEs

We note that (3.3.10) describes the homogeneous eigenvalue problem for


d2
dx 2 in (0, ). The eigenvalues of this problem are k = k , k = 1, 2, ,
2

and the corresponding normalized eigenfunctions



2
wk (x) = sin kx

 L (0, ). For any v L (0, ), the
form a complete orthonormal set in 2 2

Fourier series of v with respect to { 2 sin kx} is given by



2
v(x) = vk sin kx,

k=1
where  
2
vk = v(x) sin kx dx.
0
The Fourier series converges to v in L2 (0, ), and
 1
 2

vL2 (0,) = 2
vk .
k=1
For k = 1, 2, , let
uk (x, t) = ak (t)wk (x)
be a solution of (3.3.9). Then ak (t) satises the ordinary dierential equation
ak (t) + k 2 ak (t) = 0.
Thus, ak (t) has the form
ak (t) = ak ek t ,
2

where ak is constant. Therefore, for k = 1, 2, , we have



2
ak ek t sin kx for (x, t) (0, ) (0, ).
2
uk (x, t) =

We note that uk satises the heat equation and the boundary value in (3.3.8).
In order to get a solution satisfying the equation, the boundary value and
the initial value in (3.3.8), we consider an innite linear combination of uk
and choose coecients appropriately.
We emphasize that we identied an eigenvalue problem (3.3.10) from
d2
the initial/boundary-value problem (3.3.8). We note that dx 2 in (3.3.10)
originates from the term evolving spatial derivative in the equation in (3.3.8)
and that the boundary condition in (3.3.10) is the same as that in (3.3.8).
Now, let us suppose that

2
ak ek t sin kx
2
(3.3.11) u(x, t) =

k=1
3.3. Separation of Variables 77

solves (3.3.8). In order to identify the coecients ak , k = 1, 2, , we


calculate formally: 
2
u(x, 0) = ak sin kx,

k=1
but we are given the initial condition u(x, 0) = u0 (x) for x (0, ). Thus
we take the constants ak , k= 1, 2, , to be the Fourier coecients of u0
 2  2
with respect to the basis sin kx of L (0, ), i.e.,
 
2
(3.3.12) ak = u0 (x) sin kx dx for k = 1, 2, .
0
Next we prove that u in (3.3.11) indeed solves (3.3.8). To do this, we need
to prove that u is at least C 2 in x and C 1 in t and satises (3.3.8) under
appropriate conditions on u0 . We rst have the following result.
Theorem 3.3.5. Suppose u0 L2 (0, ) and u is given by (3.3.11) and
(3.3.12). Then u is smooth in [0, ] (0, ) and
ut uxx = 0 in (0, ) (0, ),
u(0, t) = u(, t) = 0 for t (0, ).
Moreover,
lim u(, t) u0 L2 (0,) = 0.
t0

Proof. Let i and j be nonnegative integers. For any x [0, ] and t


(0, ), we have formally

i j 2  dj k2 t di
x t u(x, t) = ak j (e ) i (sin kx).
dt dx
k=1
In order to justify the interchange of the order of dierentiation and sum-
mation, we need to prove that the series in the right-hand side is convergent
absolutely and uniformly for any (x, t) [0, ] [t0 , ), for an arbitrarily
xed t0 > 0. Set



 dj k2 t di 
(3.3.13) Sij (x, t) =  ) i (sin kx) .
ak dtj (e dx
k=1
Fix t0 > 0. Then for any (x, t) [0, ] [t0 , ),

 
k i+2j k i+2j
Sij (x, t) |ak | |a k | .
k=1
ek 2 t k=1
e k 2 t0
Since u0 L2 (0, ), we have


u0 2L2 (0,) = a2k < .
k=1
78 3. An Overview of Second-Order PDEs

Then the Cauchy inequality implies, for any (x, t) [0, ] [t0 , ),
 1  1
 2  k 2i+4j 2
(3.3.14) Sij (x, t) a2k Ci,j,t0 u0 L2 (0,) ,
k=1 k=1
e2k2 t0

where Ci,j,t0 is a positive constant depending only on i, j and t0 . This veries


that the series dening xi tj u(x, t) is convergent absolutely and uniformly
for (x, t) [0, ] [t0 , ), for any nonnegative integers i and j. Hence u is
smooth in [0, ] [t0 , ) for any t0 > 0. Therefore, all derivatives of u can
be obtained from term-by-term dierentiation in (3.3.11). It is then easy to
conclude that u satises the heat equation and the boundary condition in
(3.3.8).
We now prove the L2 -convergence. First, from the series expansions of
u and u0 , we see that

2  k2 t
u(x, t) u0 (x) = ak e 1 sin kx,

k=1

and then


2 2
|u(x, t) u0 (x)|2 dx = a2k ek t 1 .
0 k=1

We note that e k2 t
1 as t 0 for each xed k 1. For a positive integer
K to be determined, we write
 
K 
2 2 2 2
|u(x, t) u0 (x)|2 dx = a2k ek t 1 + a2k ek t 1 .
0 k=1 k=K+1

For any > 0, there exists a positive integer K = K() such that


a2k < .
k=K+1

Then there exists a > 0, depending on and K, such that



K
2
a2k ek t 1 <
2
for any t (0, ),
k=1

since the series in the left-hand side consists of nitely many terms. There-
fore, we obtain

|u(x, t) u0 (x)|2 dx < 2 for any t (0, ).
0

This implies the desired L2 -convergence as t 0. 


3.3. Separation of Variables 79

In fact, (3.3.14) implies the following estimate. For any integer m 0


and any t0 > 0,
uC m ([0,][t0 ,)) Cm,t0 u0 L2 (0,) ,
where Cm,t0 is a positive constant depending only on m and t0 . This estimate
controls the C m -norm of u in [0, ] [t0 , ) in terms of the L2 -norm of u0
on (0, ). It is referred to as an interior estimate (with respect to t). We
note that u becomes smooth instantly after t = 0 even if the initial value u0
is only L2 .
Naturally, we ask whether u in Theorem 3.3.5 is continuous up to
{t = 0}, or, more generally, whether u is smooth up to {t = 0}. First,
we assume that u is continuous up to {t = 0}. Then u0 C[0, ]. By com-
paring the initial value with the homogeneous boundary value at corners,
we have
u0 (0) = 0, u0 () = 0.
Next, we assume that u is C 2 in x and C 1 in t up to {t = 0}. Then
u0 C 2 [0, ]. By the homogeneous boundary condition and dierentiation
with respect to t, we have
ut (0, t) = 0, ut (, t) = 0 for t 0.
Evaluating at t = 0 yields
ut (0, 0) = 0, ut (, 0) = 0.
Then by the heat equation, we get
uxx (0, 0) = 0, uxx (, 0) = 0,
and hence
u0 (0) = 0, u0 () = 0.
If u is smooth up to {t = 0}, we can continue this process. Then we have a
necessary condition
(2) (2)
(3.3.15) u0 (0) = 0, u0 () = 0 for any  = 0, 1, .
Now, we prove that this is also a sucient condition.

Theorem 3.3.6. Suppose u0 C [0, ] and u is given by (3.3.11) and


(3.3.12). If (3.3.15) holds, then u is smooth in [0, ] [0, ), and u(, 0) =
u0 .

Proof. Let i and j be nonnegative integers. We need to prove that the


series dening xi tj u(x, t) converges absolutely and uniformly for (x, t)
80 3. An Overview of Second-Order PDEs

[0, ] [0, ). Let Sij be the series dened in (3.3.13). Then for any
x [0, ] and t 0,


Sij (x, t) k i+2j |ak |.
k=1
To prove that the series in the right-hand side is convergent, we need to
improve estimates of ak , the coecients of Fourier series of u0 . With (3.3.15)
for  = 0, we have, upon simple integrations by parts,
   
2 2  cos kx
ak = u0 (x) sin kx dx = u (x) dx
0 0 0 k
 
2  sin kx
= u0 (x) 2 dx.
0 k
We note that values at the endpoints are not present since u0 (0) = u0 () = 0
in the rst integration by parts and sin kx = 0 at x = 0 and x = in the
second integration by parts. Hence for any m 1, we continue this process
with the help of (3.3.15) for  = 0, , [(m 1)/2] and obtain
 
m1 2 (m) cos kx
ak = (1) 2 u0 (x) m dx if m is odd,
0 k
and  
m 2 (m) sin kx
ak = (1) 2 u0 (x) m dx if m is even.
0 k
In other words, {k ak } is 
m the sequence ofcoecients of the Fourier series
(m)
of u0 with respect to { 2 sin kx} or { 2 cos kx}, where m determines
uniquely the choice of positive or negative sign and the choice of the sine or
the cosine function. Then, we have

 (m)
k 2m a2k u0 2L2 (0,) .
k=1

Hence, by the Cauchy inequality, we obtain that, for any (x, t) [0, ]
[0, ) and any m,

 1  1
  2  2

Sij (x, t) k i+2j |ak | k 2m a2k k 2(i+2jm) .


k=1 k=1 k=1
By taking m = i + 2j + 1, we get
(m)
Sij (x, t) Cij u0 L2 (0,) ,
where Cij is a positive constant depending only on i and j. This implies
that the series dening xi tj u(x, t) converges absolutely and uniformly for
(x, t) [0, ] [0, ). Therefore, xi tj u is continuous in [0, ] [0, ). 
3.3. Separation of Variables 81

If we are interested only in the continuity of u up to t = 0, we have the


following result.
Corollary 3.3.7. Suppose u0 C 1 [0, ] and u is given by (3.3.11) and
(3.3.12). If u0 (0) = u0 () = 0, then u is smooth in [0, ](0, ), continuous
in [0, ] [0, ) and satises (3.3.8).

Proof. It follows from Theorem 3.3.5 that u is smooth in [0, ] (0, )


and satises the heat equation and the homogeneous boundary condition in
(3.3.8). The continuity of u up to t = 0 follows from the proof of Theorem
3.3.6 with i = j = 0 and m = 1. 

The regularity assumption on u0 in Corollary 3.3.7 does not seem to


be optimal. It is natural to ask whether it suces to assume that u0 is in
C[0, ] instead of in C 1 [0, ]. To answer this question, we need to analyze
pointwise convergence of Fourier series. We will not pursue this issue in this
book.
Now we provide another expression of u in (3.3.11). With explicit ex-
pressions of ak in terms of u0 in (3.3.12), we can write

(3.3.16) u(x, t) = G(x, y; t)u0 (y) dy,
0
where

2  k2 t
G(x, y; t) = e sin kx sin ky,

k=1
for any x, y [0, ] and t > 0. The function G is called the Greens function
of the initial/boundary-value problem (3.3.8). For each xed t > 0, the
series for G is convergent absolutely and uniformly for any x, y [0, ].
In fact, this uniform convergence justies the interchange of the order of
summation and integration in obtaining (3.3.16). The Greens function G
satises the following properties:
(1) Symmetry: G(x, y; t) = G(y, x; t).
(2) Smoothness: G(x, y; t) is smooth in x, y [0, ] and t > 0.
(3) Solution of the heat equation: Gt Gxx = 0.
(4) Homogeneous boundary values: G(0, y; t) = G(, y; t) = 0.
These properties follow easily from the explicit expression for G. They
imply that u in (3.3.16) is a smooth function in [0, ] (0, ) and satis-
es the heat equation with homogeneous boundary values. We can prove
directly with the help of the explicit expression of G that u in (3.3.16) is
continuous up to t = 0 and satises u(, 0) = u0 under appropriate assump-
tions on u0 . We point out that G can also be expressed in terms of the
82 3. An Overview of Second-Order PDEs

fundamental solution of the heat equation. See Chapter 5 for discussions of


the fundamental solution.
Next we discuss initial/boundary-value problems for the 1-dimensional
wave equation. Let u0 and u1 be continuous functions on [0, ]. Consider
utt uxx = 0 in (0, ) (0, ),
(3.3.17) u(x, 0) = u0 (x), ut (x, 0) = u1 (x) for x (0, ),
u(0, t) = u(, t) = 0 for t (0, ).

We proceed as for the heat equation, rst considering the problem


utt uxx = 0 in (0, ) (0, ),
(3.3.18)
u(0, t) = u(, t) = 0 for t (0, ),
and asking for solutions of the form
u(x, t) = c(t)w(x).
An argument similar to that given for the heat equation shows that w must
d2
be a solution of the homogeneous eigenvalue problem for dx 2 on (0, ). The

eigenvalues of this problem are k = k , k = 1, 2, , and the corresponding


2

normalized eigenfunctions

2
wk (x) = sin kx

form a complete orthonormal set in L2 (0, ). For k = 1, 2, , let
uk (x, t) = ck (t)wk (x)
be a solution of (3.3.18). Then ck (t) satises the ordinary dierential equa-
tion
ck (t) + k 2 ck (t) = 0.
Thus, ck (t) has the form
ck (t) = ak cos kt + bk sin kt,
where ak and bk are constants. Therefore, for k = 1, 2, , we have

2
uk (x, t) = (ak cos kt + bk sin kt) sin kx.

Now, let us suppose that

2
(3.3.19) u(x, t) = (ak cos kt + bk sin kt) sin kx

k=1
3.3. Separation of Variables 83

solves (3.3.17). In order to identify the coecients ak and bk , k = 1, 2, ,


we calculate formally:

2
u(x, 0) = ak sin kx,

k=1

but we are given the initial condition u(x, 0) = u0 (x) for x (0, ). Thus
we take the constants ak , k= 1, 2, , to be the Fourier coecients of u0
 2  2
with respect to the basis sin kx of L (0, ), i.e.,
 
2
(3.3.20) ak = u0 (x) sin kx dx for k = 1, 2, .
0
Dierentiating (3.3.19) term by term, we nd

2
ut (x, t) = (kak sin kt + kbk cos kt) sin kx,

k=1
and evaluating at t = 0 gives

2
ut (x, 0) = kbk sin kx.

k=1

From the initial condition ut (x, 0) = u1 (x), we see that kbk , fork = 1, 2, ,
 2 
are the Fourier coecients of u1 with respect to the basis sin kx of
L2 (0, ), i.e.,
 
2 sin kx
(3.3.21) bk = u1 (x) dx for k = 1, 2, .
0 k
We now discuss the regularity of u in (3.3.19). Unlike the case of the
heat equation, in order to get dierentiability of u now, we need to impose
similar dierentiability assumptions on initial values. Proceding as for the
heat equation, we note that if u is a C 2 -solution, then
u0 (0) = 0, u1 (0) = 0, u0 (0) = 0,
(3.3.22)
u0 () = 0, u1 () = 0, u0 () = 0.

Theorem 3.3.8. Suppose u0 C 3 [0, ], u1 C 2 [0, ] and u is dened by


(3.3.19), (3.3.20) and (3.3.21). If u0 , u1 satisfy (3.3.22), then u is C 2 in
[0, ] [0, ) and is a solution of (3.3.17).

Proof. Let i and j be two nonnegative integers with 0 i + j 2. For any


x [0, ] and t (0, ), we have formally

i j 2  dj di
x t u(x, t) = (a k cos kt + b k sin kt) (sin kx).
dtj dxi
k=1
84 3. An Overview of Second-Order PDEs

In order to justify the interchange of the order of dierentiation and sum-


mation, we need to prove that the series in the right-hand side is convergent
absolutely and uniformly for any (x, t) [0, ] [0, ). Set
 j


d di 
Tij (x, t) =  
 dtj (ak cos kt + bk sin kt) dxi (sin kx) .
k=1

Hence, for any (x, t) [0, ] [0, ),



 1
Tij (x, t) k i+j a2k + b2k 2 .
k=1

To prove the convergence of the series in the right-hand side, we need to


improve estimates for ak and bk . By (3.3.22) and integration by parts, we
have
   
2 2  cos kx
ak = u0 (x) sin kx dx = u0 (x) 3 dx,
k
  0   0
2 sin kx 2 sin kx
bk = u1 (x) dx = u1 (x) 3 dx.
0 k 0 k

 {k ak } is the sequence of Fourier coecients of u0 (x) with
In other words, 3

respect to { 2
cos kx}, and {k 3 bk } is the sequence of Fourier coecients of


u1 (x) with respect to { 2 sin kx}. Hence


(k 6 a2k + k 6 b2k ) u  2
0 L2 (0,) + u1 L2 (0,) .
2

k=1

By the Cauchy inequality, we obtain that, for any (x, t) [0, ] [0, ),
 1  1
 2  1 2
Tij (x, t) k 2(i+j+1) a2k + b2k < .
k2
k=1 k=1

Therefore, u is C 2 in [0, ] [0, ) and any derivative of u up to order two


may be calculated by a simple term-by-term dierentiation. Thus u satises
(3.3.17). 

By examining the proof, we have


 3 
 (i) 
2
(i)
uC 2 ([0,][0,)) C u0 L2 (0,) + u1 L2 (0,) ,
i=0 i=0

where C is a positive constant independent of u.


In fact, in order to get a C 2 -solution of (3.3.17), it suces to assume
u0 C 2 [0, ], u1 C 1 [0, ] and the compatibility condition (3.3.22). We
3.3. Separation of Variables 85

will prove this for a more general initial/boundary-value problem for the
wave equation in Section 6.1. See Theorem 6.1.3.
Now, we compare the regularity results for solutions of initial/boundary-
value problems in Theorems 3.3.5, 3.3.6 and 3.3.8. For the heat equation
in Theorem 3.3.5, solutions become smooth immediately after t = 0, even
for L2 -initial values. This is the interior smoothness (with respect to time).
We also proved in Theorem 3.3.6 that solutions are smooth up to {t = 0}
if initial values are smooth with a compatibility condition. This property
is called the global smoothness. However, solutions of the wave equation
exhibit a dierent property. We proved in Theorem 3.3.8 that solutions
have a similar degree of regularity as initial values. In general, solutions of
the wave equation do not have better regularity than initial values, and in
higher dimensions they are less regular than initial values. We will discuss
in Chapter 6 how solutions of the wave equation depend on initial values.
To conclude, we point out that the methods employed in this section to
solve initial/boundary-value problems for the 1-dimensional heat equation
and wave equation can actually be generalized to higher dimensions. We
illustrate this by the heat equation. Let be a bounded smooth domain in
Rn and u0 be an L2 -function in . We consider
ut u = 0 in (0, ),
(3.3.23) u(, 0) = u0 in ,
u = 0 on (0, ).

To solve (3.3.23) by separation of variables, we need to solve the eigenvalue


problem of in with homogeneous boundary values, i.e.,
+ = 0 in ,
(3.3.24)
= 0 on .

This is much harder to solve than its 1-dimensional counterpart (3.3.10).


Nevertheless, a similar result still holds. In fact, solutions of (3.3.24) are
given by a sequence (k , k ), where k is a nondecreasing sequence of posi-
tive numbers such that k as k and k is a sequence of smooth
functions in which forms a basis in L2 (). Then we can use a similar
method to nd a solution of (3.3.23) of the form


u(x, t) = ak ek t k (x) for any (x, t) (0, ).
k=1

We should remark that solving (3.3.24) is a complicated process. We need to


work in Sobolev spaces, spaces of functions with L2 -integrable derivatives.
A brief discussion of Sobolev spaces can be found in Subsection 4.4.2.
86 3. An Overview of Second-Order PDEs

3.4. Exercises
Exercise 3.1. Classify the following second-order PDEs:

n 
(1) uxi xi + uxi xj = 0.
i=1 1i<jn

(2) uxi xj = 0.
1i<jn

Exercise 3.2. (1) Let (r, ) be polar coordinates in R2 , i.e.,


x = r cos , y = r sin .
Prove that the Laplace operator can be expressed by
1 1
u = urr + ur + 2 u .
r r
(2) Let (r, , ) be spherical coordinates in R3 , i.e.,
x = r sin cos , y = r sin sin , z = r cos .
Prove that the Laplace operator can be expressed by
 
1 2 u 1 u 1 2u
u = 2 r + 2 sin + 2 2 .
r r r r sin r sin 2
Exercise 3.3. Discuss the uniqueness of the following problems using energy
methods:

u u3 = f in ,
(1)
u= on ;
  2
u u u (y) dy = f in ,
(2)
u= on .

Exercise 3.4. Let be a bounded C 1 -domain in Rn and u be a C 2 -function


in [0, T ] satisfying
ut u = f in (0, ),
u(, 0) = u0 in ,
u=0 on (0, ).
Prove
  T 
sup |u(, t)| dx +
2
u2t dxdt
0tT 0
  
T
C |u0 |2 dx + f 2 dxdt ,
0
where C is a positive constant depending only on .
3.4. Exercises 87

Exercise 3.5. Prove that the Poisson kernel in (3.3.6) is given by (3.3.7).
Exercise 3.6. For any u0 L2 (0, ), let u be given by (3.3.11). For any
nonnegative integers i and j, prove
sup |xi tj u(, t)| 0 as t .
[0,]

Exercise 3.7. Let G be dened as in (3.3.16). Prove


1
|G(x, y; t)| for any x, y [0, ] and t > 0.
t
Exercise 3.8. For any u0 L2 (0, ), solve the following problem by sepa-
ration of variables:
ut uxx = 0 in (0, ) (0, ),
u(x, 0) = u0 (x) for any x (0, ),
ux (0, t) = ux (, t) = 0 for any t (0, ).
Exercise 3.9. For any u0 L2 (0, ) and f L2 ((0, ) (0, )), nd a
formal explicit expression of a solution of the problem
ut uxx = f in (0, ) (0, ),
u(x, 0) = u0 (x) for any x (0, ),
u(0, t) = u(, t) = 0 for any t (0, ).
Exercise 3.10. For any u0 , u1 L2 (0, ) and f L2 ((0, ) (0, )), nd
a formal explicit expression of a solution of the problem
utt uxx = f in (0, ) (0, ),
u(x, 0) = u0 (x), ut (x, 0) = u1 (x) for any x (0, ),
u(0, t) = u(, t) = 0 for any t (0, ).
Exercise 3.11. Let T be a positive constant, be a bounded C 1 -domain
in Rn and u be C 2 in x and C 1 in t in [0, T ]. Suppose u satises
ut u = 0 in (0, T ),
u(, T ) = 0 in ,
u=0 on (0, T ).
Prove that u = 0 in (0, T ).

Hint: The function J(t) = log u2 (x, t) dx is a decreasing convex function.
Exercise 3.12. Classify homogeneous harmonic polynomials in R3 by fol-
lowing the steps outlined below. Let (r, , ) be spherical coordinates in R3 .
(Refer to Exercise 3.2.) Suppose u is a homogeneous harmonic polynomial
of degree m in R3 and set u = rm Qm (, ) for some function Qm dened in
S2 .
88 3. An Overview of Second-Order PDEs

(1) Prove that Qm satises



1 1 2 Qm
Qm
m(m + 1)Qm + sin = 0.
+
sin sin2 2

(2) Prove that, if Qm is of the form f ()g(), then

Qm (, ) = A cos k + B sin k fm,k (cos ),
where
k dm+k
fm,k () = (1 2 ) 2 (1 2 )m for [1, 1],
dm+k
for k = 0, 1, , m.
(3) Sketch the zero set of Qm on S2 according to k = 0, 1 k m 1
and k = m.
Chapter 4

Laplace Equations

The Laplace operator is probably the most important dierential operator


and has a wide range of important applications.
In Section 4.1, we discuss the fundamental solution of the Laplace equa-
tion and its applications. First, we introduce the important notion of Greens
functions, which are designed to solve Dirichlet boundary-value problems.
Due to the simple geometry of balls, we are able to nd Greens functions in
balls and derive an explicit expression of solutions of the Dirichlet problem
in balls, the so-called Poisson integral formula. Second, we discuss regularity
of harmonic functions using the fundamental solution. We derive interior
gradient estimates and prove that harmonic functions are analytic.
In Section 4.2, we study the mean-value property of harmonic functions.
First, we demonstrate that the mean-value property presents an equivalent
description of harmonic functions. Due to this equivalence, the mean-value
property provides another tool to study harmonic functions. To illustrate
this, we derive the maximum principle for harmonic functions from the
mean-value property.
In Section 4.3, we discuss harmonic functions using the maximum prin-
ciple. This section is independent of Section 4.1 and Section 4.2. The
maximum principle is an important tool in studying harmonic functions, or
in general, solutions of second-order elliptic dierential equations. In this
section, the maximum principle is proved based on the algebraic structure
of the Laplace equation. As an application, we derive a priori estimates for
solutions of the Dirichlet boundary-value problem. We also derive interior
gradient estimates and the dierential Harnack inequality. As a nal appli-
cation, we solve the Dirichlet problem for the Laplace equation in a large
class of bounded domains by Perrons method.

89
90 4. Laplace Equations

We point out that several results in this chapter are proved by multiple
methods. For example, interior gradient estimates are proved by three meth-
ods: the fundamental solution, the mean-value property and the maximum
principle.
In Section 4.4, we discuss the Poisson equation. We rst discuss regu-
larity of classical solutions using the fundamental solution. Then we discuss
weak solutions and solve the Dirichlet problem in the weak sense. The
method is from functional analysis, and the Riesz representation theorem
plays an essential role. The presentation in this part is brief. The main
purpose is to introduce notions of weak solutions and Sobolev spaces.

4.1. Fundamental Solutions


The Laplace operator is dened on C 2 -functions u in a domain in Rn by

n
u = uxi xi .
i=1

The equation u = 0 is called the Laplace equation and its C 2 -solutions are
called harmonic functions.

4.1.1. Greens Identities. One of the important properties of the Laplace


equation is its spherical symmetry. As discussed in Example 3.1.7, the
Laplace equation is preserved by rotations about some point in Rn , say
the origin. Hence, it is plausible that there exist special solutions that are
invariant under rotations. We now seek harmonic functions u in Rn which
are radial, i.e., functions depending only on r = |x|. Set
v(r) = u(x).
For any i = 1, , n and x = 0, we get
xi
uxi = v  (r) ,
r
and 
x2i 1 x2i
uxi xi = v  (r) + v  (r) 3 .
r2 r r
Hence
n1 
u = v  + v = 0.
r
If v  = 0, then
n1
(log v  ) + = 0,
r
or

log(rn1 v  ) = 0.
4.1. Fundamental Solutions 91

A simple integration then yields, for n = 2,


v(r) = c1 + c2 log r for any r > 0,
and for n 3,
v(r) = c3 + c4 r2n for any r > 0,
where ci are constants for i = 1, 2, 3, 4. We note that v(r) has a singularity
at r = 0 as long as it is not constant. For reasons to be apparent soon, we
are interested in solutions with a singularity such that

u
dS = 1 for any r > 0.
Br
In the following, we set c1 = c3 = 0 and choose c2 and c4 accordingly. In
fact, we have
1
c2 = ,
2
and
1
c4 = ,
(2 n)n
where n is the surface area of the unit sphere in Rn .
Denition 4.1.1. Let be dened for x Rn \ {0} by
1
(x) = log |x| for n = 2,
2
and
1
(x) = |x|2n for n 3.
(2 n)n
The function is called the fundamental solution of the Laplace operator.

We note that is harmonic in Rn \ {0}, i.e.,


= 0 in Rn \ {0},
and 

dS = 1 for any r > 0.
Br
Moreover, has a singularity at the origin. By a simple calculation, we
have, for any i, j = 1, , n and any x = 0,
1 xi
xi (x) = n,
n |x|
and 
1 ij nxi xj
xi xj = .
n |x|n |x|n+2
We note that and its rst derivatives are integrable in any neighborhood
of the origin, even though has a singularity there. However, the second
derivatives of are not integrable near the origin.
92 4. Laplace Equations

To proceed, we review several integral formulas. Let be a C 1 -domain


in Rn and = (1 , , n ) be the unit exterior normal to . Then for any
u, v C 1 () C() and i = 1, , n,
  
uxi v dx = uvi dS uvxi dx.

This is the integration by parts in higher-dimensional Euclidean space. Now
for any w C 2 () C 1 () and v C 1 () C(), substitute wxi for u to
get  
(vwxi xi + vxi wxi ) dx = vwxi i dS.

By summing up for i = 1, , n, we get Greens formula,
 
w
vw + v w dx = v dS.

For any v, w C 2 () C 1 (), we interchange v and w and subtract to get
a second version of Greens formula,
  
w v
vw wv dx = v w dS.

Taking v 1 in either version of Greens formula, we get
 
w
w dx = dS.

We note that all these integral formulas hold if is only a piecewise C 1 -
domain.
Now we prove Greens identity, which plays an important role in discus-
sions of harmonic functions.
Theorem 4.1.2. Suppose is a bounded C 1 -domain in Rn and that u
C 1 () C 2 (). Then for any x ,

u(x) = (x y)y u(y) dy

 
u
(x y) (y) u(y) (x y) dSy .
y y

Proof. We x an x and write = (x ) for brevity. For any r > 0


such that Br (x) , the function is smooth in \ Br (x). By applying
Greens formula to u and in \ Br (x), we get
  
u
(u u) dy = u dSy
\Br (x) y
 
u
+ u dSy ,
Br (x) y
4.1. Fundamental Solutions 93

where is the unit exterior normal to (\Br (x)). Now = 0 in \Br (x),
so letting r 0, we have
    
u u
u dy = u dSy + lim u dSy .
y r0 Br (x) y
For n 3, by the denition of , we get
    
 u   r2n u 
  
dSy  =  dSy 
 (2 n)n Br (x)
Br (x)
r
max |u| 0 as r 0,
n 2 Br (x)
and
 
1
u dSy = u dSy u(x) as r 0,
Br (x) y n rn1 Br (x)

where is normal to Br (x) and points to x. This implies the desired result
for n 3. We proceed similarly for n = 2. 
Remark 4.1.3. We note that


(x y) dSy = 1,
y
for any x . This can be obtained by taking u 1 in Theorem 4.1.2.

If u has a compact support in , then Theorem 4.1.2 implies



u(x) = (x y)u(y) dy.

By computing formally, we have

u(x) = y (x y)u(y) dy.

In the sense of distributions, we write
y (x y) = x .
Here x is the Dirac measure at x, which assigns unit mass to x. The term
fundamental solution is reected in this identity. We will not give a formal
denition of distribution in this book.

4.1.2. Greens Functions. Now we discuss the Dirichlet boundary-value


problem using Theorem 4.1.2. Let f be a continuous function in and a
continuous function on . Consider
u = f in ,
(4.1.1)
u = on .
Lemma 3.2.1 asserts the uniqueness of a solution in C 2 () C 1 (). An
alternative method to obtain the uniqueness is by the maximum principle,
94 4. Laplace Equations

which will be discussed later in this chapter. Let u C 2 () C 1 () be


a solution of (4.1.1). By Theorem 4.1.2, u can be expressed in terms of f
and , with one unknown term u on . We intend to eliminate this term
by adjusting . We emphasize that we cannot prescribe u on together
with u on .
For each xed x , we consider a function (x, ) C 2 () C 1 ()
with y (x, y) = 0 in . Greens formula implies
  
u
0= (x, y)u(y) dy (x, y) (y) u(y) (x, y) dSy .
y y
Set
(x, y) = (x y) (x, y).
By a substraction from Greens identity in Theorem 4.1.2, we obtain, for
any x ,
  
u
u(x) = (x, y)u(y) dy (x, y) (y) u(y) (x, y) dSy .
y y
We will choose appropriately so that (x, ) = 0 on . Then, u
on is
eliminated from the boundary integral. The process described above leads
to the important concept of Greens functions.
To summarize, for each xed x , we consider (x, ) C 1 ()C 2 ()
such that
y (x, y) = 0 for any y ,
(4.1.2)
(x, y) = (x y) for any y .
The existence of in general domains is not the main issue in our discussion
here. We will prove later that (x, ) is smooth in for each xed x if it
exists. (See Theorem 4.1.10.)
Denition 4.1.4. The Greens function G for the domain is dened by
G(x, y) = (x y) (x, y),
for any x, y with x = y.

In other words, for each xed x , G(x, ) diers from (x ) by a


harmonic function in and vanishes on . If such a G exists, then the
solution u of the Dirichlet problem (4.1.1) can be expressed by
 
G
(4.1.3) u(x) = G(x, y)f (y) dy + (y) (x, y) dSy .
y
We note that the Greens function G(x, y) is dened as a function of
y \ {x} for each xed x . Now we discuss properties of G as a
function of x and y. As was mentioned, we will not discuss the existence
of the Greens function in general domains. However, we should point out
4.1. Fundamental Solutions 95

that the Greens function is unique if it exists. This follows from Lemma
3.2.1 or Corollary 4.2.9, since the dierence of any two Greens functions is
harmonic, with vanishing boundary values.
Lemma 4.1.5. Let G be the Greens function in . Then G(x, y) = G(y, x)
for any x, y with x = y.

Proof. For any x1 , x2 with x1 = x2 , take r > 0 small enough that


Br (x1 ) , Br (x2 ) and Br (x1 ) Br (x2 ) = . Set Gi (y) = G(xi , y) and
i (y) = (xi y) for i = 1, 2. By Greens formula in \ (Br (x1 ) Br (x2 )),
we get
  
G2 G1
G1 G2 G2 G1 dy = G1 G2 dSy
\(Br (x1 )Br (x2 ))
   
G2 G1 G2 G1
+ G1 G2 dSy + G1 G2 dSy ,
Br (x1 ) Br (x2 )

where is the unit exterior normal to \ (Br (x1 ) Br (x2 )) . Since Gi (y)
is harmonic for y = xi , i = 1, 2, and vanishes on , we have
   
G2 G1 G2 G1
G1 G2 dSy + G1 G2 dSy = 0.
Br (x1 ) Br (x2 )
Now we replace G1 in the rst integral by 1 and replace G2 in the second
integral by 2 . Since G1 1 is C 2 in and G2 is C 2 in \ Br (x2 ), we have
 
G2 (G1 1 )
(G1 1 ) G2 dSy 0 as r 0.
Br (x1 )
Similarly,
 
(G2 2 ) G1
G1 (G2 2 ) dSy 0 as r 0.
Br (x2 )
Therefore, we obtain
   
G2 1 2 G1
1 G2 dSy + G1 2 dSy 0,
Br (x1 ) Br (x2 )
as r 0. On the other hand, by explicit expressions of 1 and 2 , we have
 
G2 G1
1 dSy 0, 2 dSy 0,
Br (x1 ) Br (x2 )
and
 
1 2
G2 dSy G2 (x1 ), G1 dSy G1 (x2 ),
Br (x1 ) Br (x2 )
as r 0. These limits can be proved similarly as in the proof of Theorem
4.1.2. We point out that points to xi on Br (xi ), for i = 1, 2. We then
obtain G2 (x1 ) G1 (x2 ) = 0 and hence G(x2 , x1 ) = G(x1 , x2 ). 
96 4. Laplace Equations

Finding a Greens function involves solving a Dirichlet problem for the


Laplace equation. Meanwhile, Greens functions are introduced to yield an
explicit expression of solutions of the Dirichlet problem. It turns out that
we can construct Greens functions for some special domains.

4.1.3. Poisson Integral Formula. In the next result, we give an explicit


expression of Greens functions in balls. We exploit the geometry of balls in
an essential way.
Theorem 4.1.6. Let G be the Greens function in the ball BR Rn .
(1) In case n 3,
1 2n
G(0, y) = |y| R2n ,
(2 n)n
for any y BR \ {0}, and
 n2   
1 R  R 2 2n
G(x, y) = |y x| 2n
y x ,
(2 n)n |x|  |x|2 
for any x BR \ {0} and y BR \ {x}.
(2) In case n = 2,
1
G(0, y) = (log |y| log R) ,
2
for any y BR \ {0}, and

1 |x|  R2 
G(x, y) = log |y x| log y 2x ,
2 R |x|
for any x BR \ {0} and y BR \ {x}.

Proof. By Denition 4.1.4, we need to nd in (4.1.2). We consider n 3


rst. For x = 0,
1
(0 y) = |y|2n .
(2 n)n
Hence we take
1
(0, y) = R2n ,
(2 n)n
for any y BR . Next, we x an x BR \ {0} and let X = R2 x/|x|2 .
Obviously, we have X
/ BR and hence (y X) is harmonic for y BR .
For any y BR , by
|x| R
= ,
R |X|
we have Oxy OyX. Then for any y BR ,
|x| |y x|
= ,
R |y X|
4.1. Fundamental Solutions 97

and hence,
|x|
(4.1.4) |y x| = |y X|.
R
This implies

R n2
(y x) = (y X),
|x|
for any x BR \ {0} and y BR . Then we take
n2
R
(x, y) = (y X),
|x|
for any x BR \ {0} and y BR \ {x}. The proof for n = 2 is similar and
is omitted. 

X


x

PP
O P
y
R

Figure 4.1.1. The reection about the sphere.

Next, we calculate normal derivatives of the Greens function on spheres.


Corollary 4.1.7. Let G be the Greens function in BR . Then
G R2 |x|2
(x, y) = ,
y n R|x y|n
for any x BR and y BR .

Proof. We rst consider n 3. With X = R2 x/|x|2 as in the proof of


Theorem 4.1.6, we have
 n2 
1 R
G(x, y) = |y x|2n |y X|2n ,
(2 n)n |x|
for any x BR \ {0} and y BR \ {x}. Hence we get, for such x and y,
 n2 
1 yi xi R yi Xi
Gyi (x, y) = .
n |y x|n |x| |y X|n
98 4. Laplace Equations

By (4.1.4) in the proof of Theorem 4.1.6, we have, for any x BR \ {0} and
y BR ,
yi R2 |x|2
Gyi (x, y) = .
n R2 |x y|n
This formula also holds when x = 0. With i = yi /R for any y BR , we
obtain
G  n
1 R2 |x|2
(x, y) = i Gyi (x, y) = .
y n R |x y|n
i=1
This yields the desired result for n 3. The proof for n = 2 is similar and
is omitted. 

Denote by K(x, y) the function in Corollary 4.1.7, i.e.,


R2 |x|2
(4.1.5) K(x, y) = ,
n R|x y|n
for any x BR and y BR . It is called the Poisson kernel.
Lemma 4.1.8. Let K be the Poisson kernel dened by (4.1.5). Then
(1) K(x, y) is smooth for any x BR and y BR ;
(2) K(x, y) > 0 for any x BR and y BR ;
(3) for any xed x0 BR and > 0,
lim K(x, y) = 0 uniformly in y BR \ B (x0 );
xx0 ,|x|<R

(4) x K(x, y) = 0 for any x BR and y BR ;



(5) BR K(x, y) dSy = 1 for any x BR .

Proof. First, (1), (2) and (3) follow easily from the explicit expression for K
as in (4.1.5), and (4) follows easily from the denition K(x, y) = y G(x, y)
and the fact that G(x, y) is harmonic in x. Of course, we can also verify
(4) by a straightforward calculation. An easy derivation of (5) is based on
(4.1.3). By taking a C 2 (BR ) harmonic function u in (4.1.3), we conclude
that 
u(x) = K(x, y)u(y) dSy for any x BR .
BR
Then we have (5) by taking u 1. 

Now we are ready to solve the Laplace equation in balls, with prescribed
Dirichlet boundary values.
Theorem 4.1.9. Let be a continuous function on BR and u be dened
by

(4.1.6) u(x) = K(x, y)(y) dSy for any x BR ,
BR
4.1. Fundamental Solutions 99

where K is the Poisson kernel given by (4.1.5). Then u is smooth in BR


and u = 0 in BR . Moreover, for any x0 BR ,

lim u(x) = (x0 ).


xx0

Proof. By Lemma 4.1.8(1) and (4), we conclude easily that u dened by


(4.1.6) is smooth and harmonic in BR . We need only prove the convergence
of u up to the boundary BR . We x x0 BR and take an x BR . By
Lemma 4.1.8(5), we have

(x0 ) = K(x, y)(x0 ) dSy .
BR

Then


u(x) (x0 ) = K(x, y) (y) (x0 ) dSy = I1 + I2 ,
BR

where  
I1 = , I2 = ,
BR B (x0 ) BR \B (x0 )

for a positive constant to be determined. For any > 0, we can choose


= () > 0 small so that

|(y) (x0 )| <

for any y BR B (x0 ), because is continuous at x0 . Then by Lemma


4.1.8(2) and (5),

 
|I1 | K(x, y)(y) (x0 ) dSy < .
BR B (x0 )

Next, set M = maxBR ||. By Lemma 4.1.8(3), we can nd a  > 0 such


that

K(x, y) ,
2M n Rn1
for any x BR B (x0 ) and any y BR \B (x0 ). We note that  depends
on and = (), and hence only on . Then


|I2 | K(x, y) |(y)| + |(x0 )| dSy .
BR \B (x0 )

Hence
|u(x) (x0 )| < 2,
for any x BR B (x0 ). This implies the convergence of u at x0 BR . 
100 4. Laplace Equations

We note that the function u in (4.1.6) is dened only in BR . We can


extend u to BR by dening u = on BR . Then u C (BR ) C(BR ).
Therefore, u is a solution of

u = 0 in BR ,
u= on BR .

The formula (4.1.6) is called the Poisson integral formula, or simply the
Poisson formula.
For n = 2, with x = (r cos , r sin ) and y = (R cos , R sin ) in (4.1.6),
we have
 2
1
u(r cos , r sin ) = K(r, , )(R cos , R sin ) d,
2 0
where
R2 r 2
K(r, , ) = .
R2 2Rr cos( ) + r2
Compare with (3.3.6) and (3.3.7) in Section 3.3.
Now we discuss properties of the function dened in (4.1.6). First, u(x)
in (4.1.6) is smooth for |x| < R, even if the boundary value is simply
continuous on BR . In fact, any harmonic function is smooth. We will
prove this result later in this section.
Next, by letting x = 0 in (4.1.6), we have

1
u(0) = u(y) dSy .
n Rn1 BR

We note that n Rn1 is the surface area of the sphere BR . Hence, values
of harmonic functions at the center of spheres are equal to their average over
spheres. This is the mean-value property.
Moreover, by Lemma 4.1.8(2) and (5), u in (4.1.6) satises

min u max in BR .
BR BR

In other words, harmonic functions in balls are bounded from above by their
maximum on the boundary and bounded from below by their minimum on
the boundary. Such a result is referred to as the maximum principle. Again,
this is a general fact, and we will prove it for any harmonic function in any
bounded domain.
The mean-value property and the maximum principle are the main topics
in Section 4.2 and Section 4.3, respectively.
4.1. Fundamental Solutions 101

4.1.4. Regularity of Harmonic Functions. In the following, we dis-


cuss regularity of harmonic functions using the fundamental solution of the
Laplace equation. First, as an application of Theorem 4.1.2, we prove that
harmonic functions are smooth.

Theorem 4.1.10. Let be a domain in Rn and u C 2 () be a harmonic


function in . Then u is smooth in .

Proof. We take an arbitrary bounded C 1 -domain  in such that  .


Obviously, u is C 1 in  and u = 0 in  . By Theorem 4.1.2, we have
 
u
u(x) = (x y) (y) u(y) (x y) dSy ,
 y y

for any x  . There is no singularity in the integrand, since x  and


y  . This implies easily that u is smooth in  . 

We note that, in its denition, a harmonic function is required only to


be C 2 . Theorem 4.1.10 asserts that the simple algebraic relation u = 0
among some of second derivatives of u implies that all partial derivatives of
u exist. We will prove a more general result later in Theorem 4.4.2 that u
is smooth if u is smooth.
Harmonic functions are not only smooth but also analytic. We will prove
the analyticity by estimating the radius of convergence for Taylor series of
harmonic functions. As the rst step, we estimate derivatives of harmonic
functions. For convenience, we consider harmonic functions in balls. The
following result is referred to as an interior gradient estimate. It asserts
that rst derivatives of a harmonic function at any point are controlled by
its maximum absolute value in a ball centered at this point.

Theorem 4.1.11. Suppose u C BR (x0 ) is harmonic in BR (x0 ) Rn .
Then
C
|u(x0 )| max |u|,
R BR (x0 )

where C is a positive constant depending only on n.

Proof. Without loss of generality, we may assume x0 = 0.


We rst consider R = 1 and employ a local version of Greens identity.
Take a cuto function C0 (B3/4 ) such that = 1 in B1/2 and 0 1.
For any x B1/4 , we write = (x ) temporarily. For any r > 0 small
102 4. Laplace Equations

enough, applying Greens formula to u and in B1 \ Br (x), we get


  
u ()
u u() dy = u dSy
B1 \Br (x) B1
 
u ()
+ u dSy ,
Br (x)
where is the unit exterior normal to (B1 \ Br (x)). The boundary integral
over B1 is zero since = = 0 on B1 . In the boundary integral over
Br (x), we may replace by 1 since Br (x) B1/2 if r < 1/4. As shown in
the proof of Theorem 4.1.2, we have
 
u
u(x) = lim u dSy ,
r0 Br (x)
where is normal to Br (x) and points toward x. For the domain integral,
the rst term is zero since u = 0 in B1 . For the second term, we have
() = + 2 + .
We note that = 0 in B1 \ Br (x) and that the derivatives of are zero
for |y| < 1/2 and 3/4 < |y| < 1 since is constant there. Then we obtain


u(x) = u(y) y (y)(x y) + 2y (y) y (x y) dy,
1
2
<|y|< 43

for any x B1/4 . We note that there is no singularity in the integrand for
|x| < 1/4 and 1/2 < |y| < 3/4. (This also gives an alternative proof of the
smoothness of u in B1/4 .) Therefore,


u(x) = u(y) y (y)x (x y)
1
<|y|< 34
2

+2y (y) x y (x y) dy,
for any x B1/4 . Hence, we obtain
|u(x)| C sup |u| for any x B 1 ,
B1 4

where C is a positive constant depending only on n. We obtain the desired


result by taking x = 0.
The general case follows from a simple dilation. Dene
u(x) = u(Rx) for any x B1 .
Then u is a harmonic function in B1 . By applying the result we just proved
to u, we obtain
|u(0)| C sup |u|.
B1
Since u(0) = Ru(0), we have the desired result. 
4.1. Fundamental Solutions 103

We note that the proof above consists of two steps. We rst prove the
desired estimate for R = 1 and then extend such an estimate to arbitrary
R by a simple scaling. Such a scaling argument is based on the following
fact: If u is a harmonic function in BR , then u(x) = u(Rx) is a harmonic
function in B1 . We point out that this scaling argument is commonly used
in studying elliptic and parabolic dierential equations.
Next, we estimate derivatives of harmonic functions of arbitrary order.

Theorem 4.1.12. Suppose u C BR (x0 ) is harmonic in BR (x0 ) Rn .
Then for any multi-index with || = m,
C m em1 m!
| u(x0 )| max |u|,
Rm BR (x0 )

where C is a positive constant depending only on n.

Proof. The proof is by induction on m 1. The case of m = 1 holds by


Theorem 4.1.11. We assume it holds for m and consider m + 1. Let v be an
arbitrary derivative of u of order m. Obviously, it is harmonic in BR (x0 ).
For any (0, 1), by applying Theorem 4.1.11 to v in B(1)R (x0 ), we get
C
|v(x0 )| max |v|.
(1 )R B(1)R (x0 )
For any x B(1)R (x0 ), we have BR (x) BR (x0 ). By the induction
assumption, we obtain
C m em1 m!
|v(x)| max |u|,
(R)m BR (x)
for any x B(1)R (x0 ), and hence
C m em1 m!
max |v| max |u|.
B(1)R (x0 ) (R)m BR (x0 )
Therefore,
C m+1 em1 m!
|v(x0 )| max |u|.
(1 )m Rm+1 BR (x0 )
m
By taking = m+1 ,we have

1 1 m
= 1+ (m + 1) < e(m + 1).
(1 )m m
This implies
C m+1 em (m + 1)!
|v(x0 )| max |u|.
Rm+1 BR (x0 )
Hence the desired result is established for any derivatives of u of order
m + 1. 
104 4. Laplace Equations

As a consequence of the interior estimate on derivatives, we prove the


following compactness result.
Corollary 4.1.13. Let be a bounded domain in Rn , M be a positive
constant and {uk } be a sequence of harmonic functions in such that
sup |uk | M for any k.

Then there exist a harmonic function u in and a subsequence {uk } such


that
uk u uniformly in  as k  ,
for any  with  .

Proof. Take any  with  and set d = dist( , ). For any x  ,


we have Bd (x) . By applying Theorem 4.1.12 to uk in Bd (x), we get,
for any integer m 1,
Cm
|m uk (x)| sup |uk | Cm dm M,
dm Bd (x)
where Cm is a positive constant depending only on n and m. Hence
max

|m uk | Cm dm M.

For any  = 0, 1, , the mean-value theorem implies


| uk (x)  uk (y)| C+1 d1 M |x y|,
for any k = 1, 2, , and any x, y  .
Next, we take a sequence of domains {j } with j j+1
and dj = dist(j , ) 1/j. Then
| uk (x)  uk (y)| C+1 d1
j M |x y|,
for any  = 0, 1, , any k = 1, 2, , and any x, y j . By Arzelas theo-
rem and diagonalization, we can nd a function u in and a subsequence
{uk } such that
uk  u in the C  -norm in j as k  ,
for any j = 1, 2, and any  = 0, 1, . By taking  = 2, we then get
u = 0 in each j from uk = 0. 

As shown in the proof, uk converges to u in C  ( ) for any  with


 and any  = 0, 1, .
Now we are ready to prove that harmonic functions are analytic. Real
analytic functions will be studied in Section 7.2. Now we simply introduce
the notion. Let u be a (real-valued) function dened in a neighborhood of
4.2. Mean-Value Properties 105

x0 Rn . Then u is analytic near x0 if its Taylor series about x0 is convergent


to u in a neighborhood of x0 , i.e., for some r > 0,
 1
u(x) = u(x0 )(x x0 ) ,

!
for any x Br (x0 ).
Theorem 4.1.14. Harmonic functions are analytic.

Proof. Let be a domain in Rn and u be a harmonic function in . For


any xed x0 , we prove that u is equal to its Taylor series about x0 in
a neighborhood of x0 . To do this, we take B2R (x0 ) and h Rn with
|h| R. For any integer m 1, we have, by the Taylor expansion,
 1 
m1
u(x0 + h) = u(x0 ) + (h1 x1 + + hn xn )i u (x0 ) + Rm (h),
i!
i=1
where
1
Rm (h) = [(h1 x1 + + hn xn )m u] (x0 + h),
m!
for some (0, 1). Note that x0 + h BR (x0 ) for |h| < R and that Rm (h)
consists of nm terms of derivatives of u of order m. By applying Theorem
4.1.12 to u in BR (x0 + h), we obtain
1 C m em1 m!
|Rm (h)| |h|m nm max |u|
m! Rm B2R (x0 )

Cne|h| m
max |u|.
R B2R (x0 )

Then for any h with Cne|h| < R/2, Rm (h) 0 as m . Hence,


1 


u(x0 + h) = (h1 x1 + + hn xn )i u (x0 ),
i!
i=0

for any h with |h| < (2Cne)1 R. 

4.2. Mean-Value Properties


It is a simple consequence of the Poisson integral formula that the mean
value of a harmonic function over a sphere is equal to its value at the center.
Indeed, this mean-value property is equivalent to harmonicity. In this sec-
tion, we briey discuss harmonic functions using the mean-value property.
The fundamental solution and the Poisson integral formula are not used
to prove the equivalence of harmonicity and the mean-value property. We
point out that the mean-value property is special and cannot be generalized
to solutions of general elliptic dierential equations. Many results in this
106 4. Laplace Equations

section were either proved in the previous section or will be proved in the
next section.
We rst dene the mean-value property. There are two versions of the
mean-value property, mean values over spheres and mean values over balls.
Denition 4.2.1. Let be a domain in Rn and u be a continuous function
in . Then
(i) u satises the mean-value property over spheres if for any Br (x) ,

1
u(x) = u(y) dSy ;
n rn1 Br (x)

(ii) u satises the mean-value property over balls if for any Br (x) ,

n
u(x) = u(y) dy,
n rn Br (x)
where n is the surface area of the unit sphere in Rn .

We note that n rr1 is the surface area of the sphere Br (x) and that
n rn /n
is the volume of the ball Br (x).
These two versions of the mean-value property are equivalent. In fact,
if we write (i) as 
n1 1
u(x)r = u(y) dSy ,
n Br (x)
we can integrate with respect to r to get (ii). If we write (ii) as

n n
u(x)r = u(y) dy,
n Br (x)
we can dierentiate with respect to r to get (i).
By a change of variables, we also write mean-value properties in the
following equivalent forms: for any Br (x) ,

1
u(x) = u(x + ry) dSy
n B1
or 
n
u(x) = u(x + ry) dy.
n B1
A function satisfying mean-value properties is required only to be con-
tinuous to start with. However, a harmonic function is required to be C 2 .
We now prove that these two requirements are actually equivalent.
Theorem 4.2.2. Let be a domain in Rn and u be a function in .
(i) If u C 2 () is harmonic in , then u satises the mean-value prop-
erty in .
4.2. Mean-Value Properties 107

(ii) If u C() satises the mean-value property in , then u is smooth


and harmonic in .

Proof. Take any ball Br (x) . Then for any u C 2 () and any
(0, r), we have
  
u n1 u
u dy = dS = (x + w) dSw
B (x) B (x) B1
(4.2.1) 

= n1 u(x + w) dSw .
B1
(i) Let u C 2 () be harmonic in . Then for any (0, r),


u(x + w) dSw = 0.
B1
Integrating from 0 to r, we obtain
 
u(x + rw) dSw = u(x) dSw = u(x)n ,
B1 B1
and hence 
1
u(x) = u(x + rw) dSw .
n B1
This yields the desired mean-value property.
(ii) Let u C() satisfy the mean-value property. For the smoothness,
we prove that u is equal to the convolution of itself with some smooth
function. To this end, we choose a smooth function in [0, 1] such that
is constant in [0, ] and = 0 in [1 , 1] for some (0, 1/2), and
 1
n rn1 (r) dr = 1.
0
The existence of such a function can be veried easily. Dene (x) = (|x|).
Then C0 (B1 ) and 
dx = 1.
B1

Next, we dene (x) = 1n x for any > 0. Then supp B . We
claim that 
u(x) = u(y) (y x) dy,

for any x with dist(x, ) > . Then it follows easily that u is smooth.
Moreover, by (4.2.1) and the mean-value property, we have, for any Br (x)
,
 

u dy = rn1 u(x + rw) dSw = rn1 n u(x) = 0.
Br (x) r B1 r
This implies u = 0 in .
108 4. Laplace Equations

Now we prove the claim. For any x and < dist(x, ), we have,
by a change of variables and the mean-value property,
   y 
1
u(y) (y x) dy = u(x + y) (y) dy = n u(x + y) dy
B B

= u(x + z)(z) dz
B1
 1
= u(x + rw)(rw)rn1 dSw dr
0 B1
 1 
n1
= (r)r u(x + rw) dSw dr
0 B1
 1
= u(x)n (r)rn1 dr = u(x).
0
This proves the claim. 

By combining both parts of Theorem 4.2.2, we have the following result.


Corollary 4.2.3. Harmonic functions are smooth and satisfy the mean-
value property.

Next, we prove an interior gradient estimate using the mean-value prop-


erty.

Theorem 4.2.4. Suppose u C BR (x0 ) is harmonic in BR (x0 ) Rn .
Then
n
|u(x0 )| max |u|.
R BR (x0 )

We note that Theorem 4.2.4 gives an explicit expression of the constant


C in Theorem 4.1.11.

Proof. Without loss of generality, we assume u C 1 (BR (x0 )). Otherwise,


we consider u in Br (x0 ) for any r < R, derive the desired estimate in Br (x0 )
and then let r R. Since u is smooth, (uxi ) = 0. In other words, uxi
is also harmonic in BR (x0 ). Hence uxi satises the mean-value property.
Upon a simple integration by parts, we obtain
 
n n
uxi (x0 ) = ux (y) dy = u(y)i dSy ,
n Rn BR (x0 ) i n Rn BR (x0 )
and hence

n
|uxi (x0 )| |u(y)| dSy
n Rn BR (x0 )
n n
n
max |u| n Rn1 max |u|.
n R BR (x0 ) R BR (x0 )
4.2. Mean-Value Properties 109

This yields the desired result. 

When harmonic functions are nonnegative, we can improve Theorem


4.2.4.

Theorem 4.2.5. Suppose u C BR (x0 ) is a nonnegative harmonic func-
tion in BR (x0 ) Rn . Then
n
|u(x0 )| u(x0 ).
R
This result is referred to as the dierential Harnack inequality. It has
many important consequences.

Proof. As in the proof of Theorem 4.2.4, from integration by parts and the
nonnegativeness of u, we have

n n
|uxi (x0 )| n
u(y) dSy = u(x0 ),
n R BR (x0 ) R
where in the last equality we used the mean-value property. 

As an application, we prove the Liouville theorem.


Corollary 4.2.6. Any harmonic function in Rn bounded from above or
below is constant.

Proof. Suppose u is a harmonic function in Rn with u c for some constant


c. Then v = u c is a nonnegative harmonic function in Rn . Let x Rn be
an arbitrary point. By applying Theorem 4.2.5 to v in BR (x) for any R > 0,
we have
n
|v(x)| v(x).
R
By letting R , we conclude that v(x) = 0. This holds for any x Rn .
Hence v is constant and so is u. 

As another application, we prove the Harnack inequality, which asserts


that nonnegative harmonic functions have comparable values in compact
subsets.
Corollary 4.2.7. Let u be a nonnegative harmonic function in BR (x0 )
Rn . Then
u(x) Cu(y) for any x, y B R (x0 ),
2

where C is a positive constant depending only on n.

Proof. Without loss of generality, we assume that u is positive in BR (x0 ).


Otherwise, we consider u + for any constant > 0, derive the desired
110 4. Laplace Equations

estimate for u + and then let 0. For any x BR/2 (x0 ), we have
BR/2 (x) BR (x0 ). By applying Theorem 4.2.5 to u in BR/2 (x), we get
2n
|u(x)| u(x),
R
or
2n
| log u(x)| .
R
For any x, y BR/2 (x0 ), a simple integration yields
 1
u(x) d
log = log u(tx + (1 t)y) dt
u(y) 0 dt
 1
= (x y) log u(tx + (1 t)y) dt.
0
Since tx + (1 t)y BR/2 (x0 ) for any t [0, 1] and |x y| R, we obtain
 1
u(x) 2n
log |x y| | log u(tx + (1 t)y)| dt |x y| 2n.
u(y) 0 R
Therefore
u(x) e2n u(y).
This is the desired result. 

In fact, Corollary 4.2.7 can be proved directly by the mean-value prop-


erty.

Another proof of Corollary 4.2.7. First, we take any B4r (x) BR (x0 )
and claim that
u(x) 3n u(x),
for any x, x Br (x). To see this, we note that Br (x) B3r (x) B4r (x) for
any x, x Br (x). Then the mean-value property implies
 
n n
u(x) = u dy u dy = 3n u(x).
n rn Br (x) n rn B3r (x)

Next we take r = R/8 and choose nitely many x1 , , xN BR/2 (x0 )


such that {Br (xi )}N i=1 covers BR/2 (x0 ). We note that B4r (xi ) BR (x0 ), for
any i = 1, , N , and that N is a constant depending only on n.
For any x, y BR/2 (x0 ), we can nd x1 , , xk BR/2 , for some
k N , such that any two consecutive points in the ordered collection of
x, x1 , , xk , y belong to a ball in {Br (xi )}N
i=1 . Then we obtain

u(x) 3n u(x1 ) 32n u(x2 ) 3nk u(xk ) 3n(k+1) u(y).


Then we have the desired result by taking C = 3n(N +1) . 
4.2. Mean-Value Properties 111

As the nal application of the mean-value property, we prove the strong


maximum principle for harmonic functions.

Theorem 4.2.8. Let be a bounded domain in Rn and u C() be har-


monic in . Then u attains its maximum and minimum only on unless
u is constant. In particular,

inf u u sup u in .

Proof. We only discuss the maximum of u. Set M = max u and


 
D = x : u(x) = M .

It is obvious that D is relatively closed; namely, for any sequence {xm } D,


if xm x , then x D. This follows easily from the continuity of u.
Next we show that D is open. For any x0 D, we take r > 0 such that
Br (x0 ) . By the mean-value property, we have
 
n n
M = u(x0 ) = u dy M dy = M.
n rn Br (x0 ) n rn Br (x0 )

This implies u = M in Br (x0 ) and hence Br (x0 ) D. In conclusion, D is


both relatively closed and open in . Therefore either D = or D = . In
other words, u either attains its maximum only on or u is constant. 

A consequence of the maximum principle is the uniqueness of solutions


of the Dirichlet problem in a bounded domain.

Corollary 4.2.9. Let be a bounded domain in Rn . Then for any f C()


and C(), there exists at most one solution u C 2 () C() of the
problem

u = f in ,
u= on .

Proof. Let w be the dierence of any two solutions. Then w = 0 in


and w = 0 on . Theorem 4.2.8 implies w = 0 in . 

Compare Corollary 4.2.9 with Lemma 3.2.1, where the uniqueness was
proved by energy estimates for solutions u C 2 () C 1 ().
The maximum principle is an important tool in studying harmonic func-
tions. We will study it in detail in Section 4.3, where we will prove the
maximum principle using the algebraic structure of the Laplace equation
and discuss its applications.
112 4. Laplace Equations

4.3. The Maximum Principle


One of the important methods in studying harmonic functions is the max-
imum principle. In this section, we discuss the maximum principle for a
class of elliptic dierential equations slightly more general than the Laplace
equation. As applications of the maximum principle, we derive a priori esti-
mates for solutions of the Dirichlet problem, and interior gradient estimates
and the Harnack inequality for harmonic functions.

4.3.1. The Weak Maximum Principle. In the following, we assume


is a bounded domain in Rn . We rst prove the maximum principle for
subharmonic functions without using the mean-value property.

Denition 4.3.1. Let u be a C 2 -function in . Then u is a subharmonic


(or superharmonic) function in if u (or ) 0 in .

Theorem 4.3.2. Let be a bounded domain in Rn and u C 2 () C()


be subharmonic in . Then u attains on its maximum in , i.e.,
max u = max u.

Proof. If u has a local maximum at a point x0 in , then the Hessian matrix


2 u(x0 ) is negative semi-denite. Thus,

u(x0 ) = tr(2 u(x0 )) 0.


Hence, in the special case that u > 0 in , the maximum value of u in
is attained only on .
We now consider the general case and assume that is contained in the
ball BR for some R > 0. For any > 0, consider
u (x) = u(x) (R2 |x|2 ).
Then
u = u + 2n 2n > 0 in .
By the special case we just discussed, u attains its maximum only on
and hence
max u = max u .

Then
max u max u + R2 = max u + R2 max u + R2 .

We have the desired result by letting 0 and using the fact that
. 
4.3. The Maximum Principle 113

A continuous function in always attains its maximum in . Theorem


4.3.2 asserts that any subharmonic function continuous up to the boundary
attains its maximum on the boundary , but possibly also in . Theorem
4.3.2 is referred to as the weak maximum principle. A stronger version asserts
that subharmonic functions attain their maximum only on the boundary. We
will prove the strong maximum principle later.
Next, we discuss a class of elliptic equations slightly more general than
the Laplace equation. Let c and f be continuous functions in . We consider
u + cu = f in .
Here, we require u C 2 (). The function c is referred to as the coecient
of the zeroth-order term. It is obvious that u is harmonic if c = f = 0.
A C 2 -function u is called a subsolution (or supersolution) if u + cu f
(or u + cu f ). If c = 0 and f = 0, subsolutions (or supersolutions) are
subharmonic (or superharmonic).
Now we prove the weak maximum principle for subsolutions.
Recall that u+ is the nonnegative part of u dened by u+ = max{0, u}.
Theorem 4.3.3. Let be a bounded domain in Rn and c be a continuous
function in with c 0. Suppose u C 2 () C() satises
u + cu 0 in .
Then u attains on its nonnegative maximum in , i.e.,
max u max u+ .

Proof. We can proceed as in the proof of Theorem 4.3.2 with simple modi-
cations. In the following, we provide an alternative proof based on Theorem
4.3.2. Set + = {x ; u(x) > 0}. If + = , then u 0 in , so u+ 0.
If + = , then
u = u + cu cu cu 0 in + .
Theorem 4.3.2 implies
max u = max u = max u+ .
+ +

This yields the desired result. 

If c 0 in , Theorem 4.3.3 reduces to Theorem 4.3.2 and we can draw


conclusions about the maximum of u rather than its nonnegative maximum.
A similar remark holds for the strong maximum principle to be proved later.
We point out that Theorem 4.3.3 holds for general elliptic dierential
equations. Let aij , bi and c be continuous functions in with c 0. We
114 4. Laplace Equations

assume

n
aij (x)i j ||2 for any x and any Rn ,
i,j=1

for some positive constant . In other words, we have a uniform positive


lower bound for the eigenvalues of (aij ) in . For u C 2 () C() and
f C(), consider the uniformly elliptic equation

n 
n
Lu aij uxi xj + bi uxi + cu = f in .
i,j=1 i=1

Many results in this section hold for uniformly elliptic equations.


As a simple consequence of Theorem 4.3.3, we have the following result.
Corollary 4.3.4. Let be a bounded domain in Rn and c be a continuous
function in with c 0. Suppose u C 2 () C() satises
u + cu 0 in ,
u0 on .
Then u 0 in .

More generally, we have the following comparison principle.


Corollary 4.3.5. Let be a bounded domain in Rn and c be a continuous
function in with c 0. Suppose u, v C 2 () C() satisfy
u + cu v + cv in ,
u v on .
Then u v in .

Proof. The dierence w = u v satises w + cw 0 in and w 0 on


. Then Corollary 4.3.4 implies w 0 in . 

The comparison principle provides a reason that functions u satisfying


u + cu f are called subsolutions. They are less than a solution v of
v + cv = f with the same boundary values.
In the following, we simply say by the maximum principle when we apply
Theorem 4.3.3, Corollary 4.3.4 or Corollary 4.3.5.
A consequence of the maximum principle is the uniqueness of solutions
of Dirichlet problems.
Corollary 4.3.6. Let be a bounded domain in Rn and c be a continuous
function in with c 0. For any f C() and C(), there exists at
4.3. The Maximum Principle 115

most one solution u C 2 () C() of

u + cu = f in ,
u= on .

Proof. Let u1 , u2 C 2 () C() be two solutions. Then w = u1 u2


satises

w + cw = 0 in ,
w=0 on .

By the maximum principle (applied to w and w), we obtain w = 0 and


hence u1 = u2 in . 

The boundedness of the domain is essential, since it guarantees the


existence of the maximum and minimum of u in . The uniqueness may
not hold if the domain is unbounded. Consider the Dirichlet problem

u = 0 in
u = 0 on ,

where = Rn \ B1 . Then a nontrivial solution u is given by



log |x| for n = 2;
u(x) =
|x| 2n 1 for n 3.

Note that u(x) as |x| for n = 2 and u is bounded for n 3.


Next, we consider the same problem in the upper half-space = {x Rn :
xn > 0}. Then u(x) = xn is a nontrivial solution, which is unbounded.
These examples demonstrate that uniqueness may not hold for the Dirichlet
problem in unbounded domains. Equally important for uniqueness is the
condition c 0. For example, we consider = (0, ) (0, ) Rn ,
and
n
u(x) = sin xi .
i=1

Then u is a nontrivial solution of the problem

u + nu = 0 in ,
u=0 on .

In fact, such a u is an eigenfunction of in with zero boundary values.


116 4. Laplace Equations

4.3.2. The Strong Maximum Principle. The weak maximum principle


asserts that subsolutions of elliptic dierential equations attain their non-
negative maximum on the boundary if the coecients of the zeroth-order
term is nonpositive. In fact, these subsolutions can attain their nonnegative
maximum only on the boundary, unless they are constant. This is the strong
maximum principle. To prove this, we need the following Hopf lemma.
For any C 1 -function u in that attains its maximum on , say at
x0 , we have u (x0 ) 0. The Hopf lemma asserts that the normal
derivative is in fact positive if u is a subsolution in .
Lemma 4.3.7. Let B be an open ball in Rn with x0 B and c be a
continuous function in B with c 0. Suppose u C 2 (B) C 1 (B) satises
u + cu 0 in B.
Assume u(x) < u(x0 ) for any x B and u(x0 ) 0. Then
u
(x0 ) > 0,

where is the exterior unit normal to B at x0 .

Proof. Without loss of generality, we assume B = BR for some R > 0. By


the continuity of u up to BR , we have
u(x) u(x0 ) for any x BR .
For positive constants and to be determined, we set
w(x) = e|x| eR ,
2 2

and
v(x) = u(x) u(x0 ) + w(x).
We consider w and v in D = BR \ BR/2 .

D x0

Figure 4.3.1. The domain D.


4.3. The Maximum Principle 117

A direct calculation yields


2
w + cw = e|x| 42 |x|2 2n + c ceR
2

2
e|x| 42 |x|2 2n + c ,
where we used c 0 in BR . Since R/2 |x| R in D, we have
w + cw e|x| (2 R2 2n + c) > 0 in D,
2

if we choose suciently large. By c 0 and u(x0 ) 0, we obtain, for any


> 0,
v + cv = u + cu + (w + cw) cu(x0 ) 0 in D.
We discuss v on D in two cases. First, on BR/2 , we have u u(x0 ) < 0,
and hence u u(x0 ) < for some > 0. Note that w < 1 on BR/2 .
Then for such an , we obtain v < 0 on BR/2 . Second, for x BR , we
have w(x) = 0 and u(x) u(x0 ). Hence v(x) 0 for any x BR and
v(x0 ) = 0. Therefore, v 0 on D.
In conclusion,
v + cv 0 in D,
v0 on D.
By the maximum principle, we have
v0 in D.
In view of v(x0 ) = 0, then v attains at x0 its maximum in D. Hence, we
obtain
v
(x0 ) 0,

and then
u w
(x0 ) = 2ReR > 0.
2
(x0 )

This is the desired result. 
Remark 4.3.8. Lemma 4.3.7 still holds if we substitute for B any bounded
C 1 -domain which satises an interior sphere condition at x0 , namely,
if there exists a ball B with x0 B. This is because such a ball B
is tangent to at x0 . We note that the interior sphere condition always
holds for C 2 -domains.

Now, we are ready to prove the strong maximum principle.


Theorem 4.3.9. Let be a bounded domain in Rn and c be a continuous
function in with c 0. Suppose u C 2 () C() satises
u + cu 0 in .
118 4. Laplace Equations








Figure 4.3.2. Interior sphere conditions.

Then u attains only on its nonnegative maximum in unless u is a


constant.

Proof. Let M be the nonnegative maximum of u in and set


D = {x : u(x) = M }.
We prove either D = or D = by a contradiction argument. Suppose D
is a nonempty proper subset of . It follows from the continuity of u that
D is relatively closed in . Then \ D is open and we can nd an open ball
B \ D such that B D = . In fact, we may choose a point x \ D
with dist(x , D) < dist(x , ) and then take the ball centered at x with
radius dist(x , D). Suppose x0 B D.



x0 D

Figure 4.3.3. The domain and its subset D.

Obviously, we have
u + cu 0 in B,
and
u(x) < u(x0 ) for any x B and u(x0 ) = M 0.
By Lemma 4.3.7, we have
u
(x0 ) > 0,

4.3. The Maximum Principle 119

where is the exterior unit normal to B at x0 . On the other hand, x0 is


an interior maximum point of . This implies u(x0 ) = 0, which leads to a
contradiction. Therefore, either D = or D = . In the rst case, u attains
only on its nonnegative maximum in ; while in the second case, u is
constant in . 

The following result improves Corollary 4.3.5.


Corollary 4.3.10. Let be a bounded domain in Rn and c be a continuous
function in with c 0. Suppose u C 2 () C() satises
u + cu 0 in ,
u0 on .
Then either u < 0 in or u is a nonpositive constant in .

We now consider the Neumann problem.


Corollary 4.3.11. Let be a bounded C 1 -domain in Rn satisfying the
interior sphere condition at every point of and c be a continuous function
in with c 0. Suppose u C 2 () C 1 () is a solution of the boundary-
value problem
u + cu = f in ,
u
= on ,

for some f C() and C(). Then u is unique if c  0 and is unique
up to additive constants if c 0.

Proof. We assume f = 0 in and = 0 on and consider


u + cu = 0 in ,
u
= 0 on .

We will prove that u = 0 if c  0 and that u is constant if c 0.
We rst consider the case c  0 and prove u = 0 by contradiction.
Suppose u has a positive maximum at x0 . If u is a positive constant,
then c 0 in , which leads to a contradiction. If u is not a constant, then
x0 and u(x) < u(x0 ) for any x by Theorem 4.3.9. Then Lemma
4.3.7 implies u
(x0 ) > 0, which contradicts the homogeneous boundary
condition. Therefore, u has no positive maximum and hence u 0 in .
Similarly, u has no positive maximum and then u 0 in . In conclusion,
u = 0 in .
We now consider the case c 0. Suppose u is a nonconstant solution.
Then its maximum in is attained only on by Theorem 4.3.9, say at x0
120 4. Laplace Equations

. Lemma 4.3.7 implies u


(x0 ) > 0, which contradicts the homogeneous
boundary value. This contradiction shows that u is constant. 

4.3.3. A Priori Estimates. As we have seen, an important application of


the maximum principle is to prove the uniqueness of solutions of boundary-
value problems. Equally or more important is to derive a priori estimates.
In derivations of a priori estimates, it is essential to construct auxiliary
functions. We will explain in the proof of the next result what auxiliary
functions are and how they are used to yield necessary estimates by the
maximum principle. We point out that we need only the weak maximum
principle in the following discussion.
We now derive an a priori estimate for solutions of the Dirichlet problem.
Theorem 4.3.12. Let be a bounded domain in Rn , c and f be continuous
functions in with c 0 and be a continuous function on . Suppose
u C 2 () C() satises
u + cu = f in ,
u= on .
Then
sup |u| sup || + C sup |f |,

where C is a positive constant depending only on n and diam().

Proof. Set
F = sup |f |, = sup ||.

Then
( + c)(u) = f F in ,
u = on .
Without loss of generality, we assume BR , for some R > 0. Set
F
v(x) = + (R2 |x|2 ) for any x .
2n
We note that v 0 in since BR . Then, by the property c 0 in ,
we have
v + cv = F + cv F.
We also have v on . Hence v satises
v + cv F in ,
v on .
4.3. The Maximum Principle 121

Therefore,
( + c)(u) ( + c)v in ,
u v on .
By the maximum principle, we obtain
u v in ,
and hence |u| v in . Therefore,
1
|u(x)| + (R2 |x|2 )F for any x .
2n
This yields the desired result. 

If = BR (x0 ), then we have


R2
sup |u| sup || + sup |f |.
BR (x0 ) BR (x0 ) 2n BR (x0 )
This follows easily from the proof.
The function v in the proof above is what we called an auxiliary function.
In fact, auxiliary functions were already used in the proof of Lemma 4.3.7.

4.3.4. Gradient Estimates. In the following, we derive gradient esti-


mates, estimates of rst derivatives. The basic method is to derive a dier-
ential equation for |u|2 and then apply the maximum principle. This is
the Bernstein method.
There are two classes of gradient estimates, global gradient estimates
and interior gradient estimates. Global gradient estimates yield estimates
of gradients u in in terms of u on , as well as u in , while interior
gradient estimates yield estimates of u in compact subsets of in terms
of u in . In the next result, we will prove the interior gradient estimate for
harmonic functions. Compare with Theorem 4.1.11 and Theorem 4.2.4.
Theorem 4.3.13. Suppose u C(B1 ) is harmonic in B1 . Then
sup |u| C sup |u|,
B1 B1
2

where C is a positive constant depending only on n.

Proof. Recall that u is smooth in B1 by Theorem 4.1.10. A direct calcula-


tion yields
n 
n 
n
2 2
(|u| ) = 2 uxi xj + 2 uxi (u)xi = 2 u2xi xj ,
i,j=1 i=1 i,j=1

where we used u = 0 in B1 . We note that |u|2


is a subharmonic function.
Hence we can easily obtain an estimate of u in B1 in terms of u on B1 .
122 4. Laplace Equations

This is the global gradient estimate. To get interior estimates, we need to


introduce a cuto function. For any nonnegative function C02 (B1 ), we
have
n 
n
(|u|2 ) = ()|u|2 + 4 xi uxj uxi xj + 2 u2xi xj .
i,j=1 i,j=1

By the Cauchy inequality, we get


2 2 2
4|xi uxj uxi xj | 2u2xi xj + u .
xi xj
Then 
2||2
(|u| )
2
|u|2 .

We note that the ratio ||2 / makes sense only when = 0. To interpret
this ratio in B1 , we take = 2 for some C02 (B1 ). Then

( 2 |u|2 ) 2 6||2 |u|2 C|u|2 ,
where C is a positive constant depending only on and n. Note that
(u2 ) = 2|u|2 + 2uu = 2|u|2 ,
since u is harmonic. By taking a constant large enough, we obtain
( 2 |u|2 + u2 ) (2 C)|u|2 0.
By the maximum principle, we obtain
sup( 2 |u|2 + u2 ) sup( 2 |u|2 + u2 ).
B1 B1

In choosing C02 (B1 ), we require in addition that 1 in B1/2 . With


= 0 on B1 , we get
sup |u|2 sup u2 .
B1/2 B1
This is the desired estimate. 

As consequences of interior gradient estimates, we have interior estimates


on derivatives of arbitrary order as in Theorem 4.1.12 and the compactness
as in Corollary 4.1.13. The compactness result will be used later in Perrons
method.
Next we derive the dierential Harnack inequality for positive harmonic
functions using the maximum principle. Compare this with Theorem 4.2.5.
Theorem 4.3.14. Suppose u is a positive harmonic function in B1 . Then
sup | log u| C,
B1/2

where C is a positive constant depending only on n.


4.3. The Maximum Principle 123

Proof. Set v = log u. A direct calculation yields


v = |v|2 .
Next, we prove an interior gradient estimate for v. By setting w = |v|2 ,
we get
n 
n
w + 2 vxi wxi = 2 vx2i xj .
i=1 i,j=1
As in Theorem 4.3.13, we need to introduce a cuto function. First, by
 n 2
 
n
vxi xi n vx2i xi ,
i=1 i=1
we have

n 
n
1 |v|4 w2
(4.3.1) vx2i xj vx2i xi (v)2 = = .
n n n
i,j=1 i=1

Take a nonnegative function C02 (B1 ). A straightforward calculation


yields
n 
n 
n
(w) + 2 vxi (w)xi = 2 vx2i xj + 4 xi vxj vxi xj
i=1 i,j=1 i,j=1

n
+ 2w xi vxi + ()w.
i=1
The Cauchy inequality implies
42xi 2
4|xi vxj vxi xj | vx2i xj + v .
xj
Then

n 
n
(w) + 2 vxi (w)xi vx2i xj 2|||v|3
i=1 i,j=1

4||2
+ |v|2 .

Here we keep one term of vx2i xj in the right-hand side instead of dropping
it entirely as in the proof of Theorem 4.3.13. To make sense of ||2 / in
B1 , we take = 4 for some C02 (B1 ). In addition, we require that = 1
in B1/2 . We obtain, by (4.3.1),

n
4
( w) + 2 vxi ( 4 w)xi
i=1
1
4 |v|4 8 3 |||v|3 + 4 2 ( 13||2 )|v|2 .
n
124 4. Laplace Equations

We note that the right-hand side can be regarded as a polynomial of degree


4 in |v| with a positive leading coecient. Other coecients depend on
and hence are bounded functions of x. For the leading term, we save half
of it for a later purpose. Now,
1 4
t 8||t3 + 4( 13||2 )t2 C for any t R,
2n
where C is a positive constant depending only on n and . Hence with
t = |v|, we get

n
1 4 2
( 4 w) + 2 vxi ( 4 w)xi w C.
2n
i=1

We note that 4 w is nonnegative in B1 and zero near B1 . Next, we assume


that 4 w attains its maximum at x0 B1 . Then ( 4 w) = 0 and ( 4 w)
0 at x0 . Hence
4 w2 (x0 ) C.
If w(x0 ) 1, then 4 w(x0 ) C. Otherwise 4 w(x0 ) 4 (x0 ). By combin-
ing these two cases, we obtain

4 w C in B1 ,

where C is a positive constant depending only on n and . With the


denition of w and = 1 in B1/2 , we obtain the desired result. 

The following result is referred to as the Harnack inequality. Compare


it with Corollary 4.2.7.

Corollary 4.3.15. Suppose u is a nonnegative harmonic function in B1 .


Then
u(x1 ) Cu(x2 ) for any x1 , x2 B1/2 ,
where C is a positive constant depending only on n.

The proof is identical to the rst proof of Corollary 4.2.7 and is omitted.
We note that u is required to be positive in Theorem 4.3.14 since log u is
involved, while u is only nonnegative in Corollary 4.3.15.
The Harnack inequality describes an important property of harmonic
functions. Any nonnegative harmonic functions have comparable values in
a proper subdomain. We point out that the Harnack inequality in fact
implies the strong maximum principle: Any nonnegative harmonic function
in a domain is identically zero if it is zero somewhere in the domain.
4.3. The Maximum Principle 125

4.3.5. Removable Singularity. Next, we discuss isolated singularity of


harmonic functions. We note that the fundamental solution of the Laplace
operator has an isolated singularity and is harmonic elsewhere. The next
result asserts that an isolated singularity of harmonic functions can be re-
moved, if it is better than that of the fundamental solution.
Theorem 4.3.16. Suppose u is harmonic in BR \ {0} Rn and satises

o(log |x|), n = 2,
u(x) = as |x| 0.
o(|x|2n ), n 3
Then u can be dened at 0 so that it is harmonic in BR .

Proof. Without loss of generality, we assume that u is continuous in 0 <


|x| R. Let v solve
v = 0 in BR ,
v=u on BR .
The existence of v is guaranteed by the Poisson integral formula in Theorem
4.1.9. Set M = maxBR |u|. We note that the constant functions M are
obviously harmonic and M v M on BR . By the maximum principle,
we have M v M in BR and hence,
|v| M in BR .
Next, we prove u = v in BR \ {0}. Set w = v u in BR \ {0} and Mr =
maxBr |w| for any r < R. We only consider the case n 3. First, we have
rn2 rn2
Mr w(x) Mr ,
|x|n2 |x|n2
for any x Br BR . It holds on Br by the denition of Mr and on
BR since w = 0 on BR . Note that w and |x|2n are harmonic in BR \ Br .
Then the maximum principle implies
rn2 rn2
Mr w(x) Mr ,
|x|n2 |x|n2
and hence
rn2
|w(x)| Mr ,
|x|n2
for any x BR \ Br . With
Mr = max |v u| max |v| + max |u| M + max |u|,
Br Br Br Br

we then have
rn2 1
|w(x)| M + n2 rn2 max |u| ,
|x| n2 |x| Br
126 4. Laplace Equations

for any x BR \ Br . Now for each xed x = 0, we take r < |x| and then let
r 0. By the assumption on u, we obtain w(x) = 0. This implies w = 0
and hence u = v in BR \ {0}. 

4.3.6. Perrons Method. In this subsection, we solve the Dirichlet prob-


lem for the Laplace equation in bounded domains by Perrons method. Es-
sentially used are the maximum principle and the Poisson integral formula.
The latter provides the solvability of the Dirichlet problem for the Laplace
equation in balls.
We rst discuss subharmonic functions. By Denition 4.3.1, a C 2 -
function v is subharmonic if v 0.
Lemma 4.3.17. Let be a domain in Rn and v be a C 2 -function in .
Then v 0 in if and only if for any ball B and any harmonic
function w C(B),
v w on B implies v w in B.

Proof. We rst prove the only if part. For any ball B and any
harmonic function w C(B) with v w on B, we have
v w in B,
vw on B.
By the maximum principle, we have v w in B.
Now we prove the if part by a contradiction argument. If v < 0
somewhere in , then v < 0 in B for some ball B with B . Let w
solve
w = 0 in B,
w=v on B.
The existence of w in B is implied by the Poisson integral formula in The-
orem 4.1.9. We have v w in B by the assumption. Next, we note that
w = 0 > v in B,
w=v on B.
We have w v in B by the maximum principle. Hence v = w in B, which
contradicts w > v in B. Therefore, v 0 in . 

Lemma 4.3.17 leads to the following denition.


Denition 4.3.18. Let be a domain in Rn and v be a continuous function
in . Then v is subharmonic (superharmonic) in if for any ball B
and any harmonic function w C(B),
v () w on B implies v () w in B.
4.3. The Maximum Principle 127

We point out that in Denition 4.3.18 subharmonic (superharmonic)


functions are dened for continuous functions. We now prove a maximum
principle for such subharmonic and superharmonic functions.
Lemma 4.3.19. Let be a bounded domain in Rn and u, v C(). Sup-
pose u is subharmonic in and v is superharmonic in with u v on .
Then u v in .

Proof. We rst note that u v 0 on . Set M = max (u v) and


 
D = x : u(x) v(x) = M .
Then D is relatively closed by the continuity of u and v. Next we prove that
D is open. For any x0 D, we take r < dist(x0 , ). Let u and v solve,
respectively,
u = 0 in Br (x0 ),
u = u on Br (x0 ),
and
v = 0 in Br (x0 ),
v = v on Br (x0 ).
The existence of u and v in Br (x0 ) is implied by the Poisson integral formula
in Theorem 4.1.9. Denition 4.3.18 implies
u u, v v in Br (x0 ).
Hence,
u v u v in Br (x0 ).
Next,
(u v) = 0 in Br (x0 ),
u v = u v on Br (x0 ).
With u v M on Br (x0 ), the maximum principle implies u v M in
Br (x0 ). In particular,
M (u v)(x0 ) (u v)(x0 ) = M.
Hence, (u v)(x0 ) = M and then u v has an interior maximum at x0 . By
the strong maximum principle, u v M in Br (x0 ). Therefore, u v = M
on Br (x0 ). This holds for any r < dist(x0 , ). Then, uv = M in Br (x0 )
and hence Br (x0 ) D. In conclusion, D is both relatively closed and open
in . Therefore either D = or D = . In other words, u v either attains
its maximum only on or u v is constant. Since u v in , we have
u v in in both cases. 
128 4. Laplace Equations

The proof in fact yields the strong maximum principle: Either u < v in
or u v is constant in .
Next, we describe Perrons method. Let be a bounded domain in
Rn and be a continuous function on . We will nd a function u
C () C() such that
u = 0 in ,
(4.3.2)
u = on .
Suppose there exists a solution u = u . Then for any v C() which is
subharmonic in with v on , we obtain, by Lemma 4.3.19,
v u in .
Hence for any x
u (x) = sup{v(x) : v C() is subharmonic in
(4.3.3)
and v on }.
We note that the equality holds since u is obviously an element of the set
in the right-hand side. Here, we assumed the existence of the solution u .
In Perrons method, we will prove that the function u dened in (4.3.3) is
indeed a solution of (4.3.2) under appropriate assumptions on the domain.
The proof consists of two steps. In the rst step, we prove that u is
harmonic in . This holds for arbitrary bounded domains. We note that u
in (4.3.3) is dened only in . So in the second step, we prove that u has
a limit on and this limit is precisely . For this, we need appropriate
assumptions on .
Before we start our discussion of Perrons method, we demonstrate how
to generate greater subharmonic functions from given subharmonic func-
tions.
Lemma 4.3.20. Let v C() be a subharmonic function in and B be a
ball in with B . Let w be dened by
w=v in \ B,
and
w = 0 in B.
Then w is a subharmonic function in and v w in .

The function w is called the harmonic lifting of v (in B).

Proof. The existence of w in B is implied by the Poisson integral formula


in Theorem 4.1.9. Then w is smooth in B and is continuous in . We also
have v w in B by Denition 4.3.18.
4.3. The Maximum Principle 129

Next, we take any B  with B  and consider a harmonic function


u C(B  ) with w u on B  . By v w on B  , we have v u on B  .
Then, v is subharmonic and u is harmonic in B  with v u on B  . By
Lemma 4.3.19, we have v u in B  . Hence w u in B  \ B. Next, both w
and u are harmonic in B B  and w u on (B B  ). By the maximum
principle, we have w u in B B  . Hence w u in B  . Therefore, w is
subharmonic in by Denition 4.3.18. 

Lemma 4.3.20 asserts that we obtain greater subharmonic functions if


we preserve the values of subharmonic functions outside the balls and extend
them inside the balls by the Poisson integral formula.
Now we are ready to prove that u in (4.3.3) is a harmonic function in
.
Lemma 4.3.21. Let be a bounded domain in Rn and be a continuous
function on . Then u dened in (4.3.3) is harmonic in .

Proof. Set
S = {v : v C() is subharmonic in and v on }.
Then for any x ,
u (x) = sup{v(x) : v S }.
In the following, we simply write S = S .
Step 1. We prove that u is well dened. To do this, we set
m = min , M = max .

We note that the constant function m is in S and hence the set S is not
empty. Next, the constant function M is obviously harmonic in with
M on . By Lemma 4.3.19, for any v S,
vM in .
Thus u is well dened and u M in .
Step 2. We prove that S is closed by taking maximum among nitely
many functions in S. We take arbitrary v1 , , vk S and set
v = max{v1 , , vk }.
It follows easily from Denition 4.3.18 that v is subharmonic in . In fact,
we take any ball B and any harmonic function w C(B) with v w
on B. Then vi w on B, for i = 1, , k. Since vi is subharmonic, we
get vi w in B, so v w in B. We conclude that v is subharmonic in .
Hence v S.
130 4. Laplace Equations

Step 3. For any Br (x0 ) , we prove that u is harmonic in Br (x0 ).


First, by the denition of u , there exists a sequence of functions vi S
such that
lim vi (x0 ) = u (x0 ).
i
We point out that the sequence {vi } depends on x0 . We may replace vi
above by any vi S with vi vi since
vi (x0 ) vi (x0 ) u (x0 ).
Replacing, if necessary, vi by max{m, vi } S, we may also assume
m vi u in .
For the xed Br (x0 ) and each vi , we let wi be the harmonic lifting in Lemma
4.3.20. In other words, wi = vi in \ Br (x0 ) and wi = 0 in Br (x0 ). By
Lemma 4.3.20, wi S and vi wi in . Hence,
lim wi (x0 ) = u (x0 ),
i
and
m wi u in ,
for any i = 1, 2, . In particular, {wi } is a bounded sequence of harmonic
functions in Br (x0 ). By Corollary 4.1.13, there exists a harmonic function w
in Br (x0 ) such that a subsequence of {wi }, still denoted by {wi }, converges
to w in any compact subset of Br (x0 ). We then conclude easily that
w u in Br (x0 ) and w(x0 ) = u (x0 ).
We now claim u = w in Br (x0 ). To see this, we take any x Br (x0 ) and
proceed similarly as before, with x replacing x0 . By the denition of u ,
there exists a sequence of functions vi S such that
lim vi (x) = u (x).
i
Replacing, if necessary, vi by max{vi , wi } S, we may also assume
wi vi u in .
For the xed Br (x0 ) and each vi , we let wi be the harmonic lifting in Lemma
4.3.20. Then, wi S and vi wi in . Moreover, wi is harmonic in Br (x0 )
and satises
lim wi (x) = u (x),
i
and
m wi vi wi u in ,
for any i = 1, 2, . By Corollary 4.1.13, there exists a harmonic function
w in Br (x0 ) such that a subsequence of wi converges to w in any compact
subset of Br (x0 ). We then conclude easily that
w w u in Br (x0 ) and w(x0 ) = w(x0 ) = u (x0 ),
4.3. The Maximum Principle 131

and
w(x) = u (x).
Next, we note that w w is a harmonic function in Br (x0 ) with a maximum
attained at x0 . By applying the strong maximum principle to w w in
Br (x0 ), we conclude that w w is constant, which is obviously zero. This
implies w = w in Br (x0 ), and in particular, w(x) = w(x) = u (x). We
then have w = u in Br (x0 ) since x is arbitrary in Br (x0 ). Therefore, u is
harmonic in Br (x0 ). 

We note that u in Lemma 4.3.21 is dened only in . We have to


discuss limits of u (x) as x approaches the boundary. For this, we need to
impose additional assumptions on the boundary .
Lemma 4.3.22. Let be a continuous function on and u be the func-
tion dened in (4.3.3). For some x0 , suppose wx0 C() is a subhar-
monic function in such that
(4.3.4) wx0 (x0 ) = 0, wx0 (x) < 0 for any x \ {x0 }.
Then
lim u (x) = (x0 ).
xx0

Proof. As in the proof of Lemma 4.3.21, we set


S = {v : v C() is subharmonic in and v on }.
We simply write w = wx0 and set M = max ||.
Let be an arbitrary positive constant. By the continuity of at x0 ,
there exists a positive constant such that
|(x) (x0 )| ,
for any x B (x0 ). We then choose K suciently large so that
Kw(x) 2M,
for any x \ B (x0 ). Hence,
| (x0 )| Kw on .
Since (x0 )+Kw(x) is a subharmonic function in with (x0 )+Kw
on , we have (x0 ) + Kw S . The denition of u implies
(x0 ) + Kw u in .
On the other hand, (x0 ) + Kw is a superharmonic in with (x0 ) +
Kw on . Hence for any v S , we obtain, by Lemma 4.3.19,
v (x0 ) + Kw in .
132 4. Laplace Equations

Again by the denition of u , we have


u (x0 ) + Kw in .
Therefore,
|u (x0 )| Kw in .
This implies
lim sup |u (x) (x0 )| .
xx0
We obtain the desired result by letting 0. 

The function wx0 satisfying (4.3.4) is called a barrier function. As shown


in the proof, wx0 provides a barrier for the function u near x0 .
We note that u in Lemma 4.3.21 is dened only in . It is natural
to extend u to by dening u (x0 ) = (x0 ) for x0 . If (4.3.4) is
satised for x0 , Lemma 4.3.22 asserts that u is continuous at x0 . If (4.3.4)
is satised for any x0 , we then obtain a continuous function u in .
Barrier functions can be constructed for a large class of domains .
Take, for example, the case where satises the exterior sphere condition
at x0 in the sense that there exists a ball Br0 (y0 ) such that
Br0 (y0 ) = , Br0 (y0 ) = {x0 }.
To construct a barrier function at x0 , we set
wx0 (x) = (x y0 ) (x0 y0 ) for any x ,
where is the fundamental solution of the Laplace operator. It is easy to
see that wx0 is a harmonic function in and satises (4.3.4). We note that
the exterior sphere condition always holds for C 2 -domains.








Figure 4.3.4. Exterior sphere conditions.

Theorem 4.3.23. Let be a bounded domain in Rn satisfying the exterior


sphere condition at every boundary point. Then for any C(), (4.3.2)
admits a solution u C () C().
4.4. Poisson Equations 133

In summary, Perrons method yields a solution of the Dirichlet problem


for the Laplace equation. This method depends essentially on the maximum
principle and the solvability of the Dirichlet problem in balls. An important
feature here is that the interior existence problem is separated from the
boundary behavior of solutions, which is determined by the local geometry
of domains.

4.4. Poisson Equations


In this section, we discuss briey the Poisson equation. We rst discuss
regularity of classical solutions using the fundamental solution. Then we
discuss weak solutions and introduce Sobolev spaces.

4.4.1. Classical Solutions. Let be a domain in Rn and f be a contin-


uous function in . The Poisson equation has the form
(4.4.1) u = f in .
If u is a smooth solution of (4.4.1) in , then obviously f is smooth. Con-
versely, we ask whether u is smooth if f is smooth. At rst glance, this
does not seem to be a reasonable question. We note that u is just a linear
combination of second derivatives of u. Essentially, we are asking whether
all partial derivatives exist and are continuous if a special combination of
second derivatives is smooth. This question turns out to have an armative
answer.
To proceed, we dene

(4.4.2) wf (x) = (x y)f (y) dy,

where is the fundamental solution of the Laplace operator as in Denition
4.1.1. The function wf is called the Newtonian potential of f in . We will
write wf, to emphasize the dependence on the domain . It is easy to see
that wf is well dened in Rn if is a bounded domain and f is a bounded
function, although has a singularity. We recall that the derivatives of
have asymptotic behavior of the form
1 1
(x y) , 2 (x y) ,
|x y| n1 |x y|n
as y x. By dierentiating under the integral sign formally, we have

xi wf (x) = xi (x y)f (y) dy,

for any x Rn and i = 1, , n. We note that the right-hand side is a
well-dened integral and denes a continuous function of x. We will not
use this identity directly in the following and leave its proof as an exercise.
Assuming its validity, we cannot simply dierentiate the expression of xi wf
134 4. Laplace Equations

to get second derivatives of wf due to the singularity of xi xj . In fact, extra


conditions are needed in order to infer that wf is C 2 . If wf is indeed C 2 and
wf = f in , then any solution of (4.4.1) diers from wf by an addition
of a harmonic function. Since harmonic functions are smooth, regularity of
arbitrary solutions of (4.4.1) is determined by that of wf .

Lemma 4.4.1. Let be a bounded domain in , f be a bounded function in


and wf be dened in (4.4.2). Assume that f C k1 () for some integer
k 2. Then wf C k () and wf = f in . Moreover, if f is smooth in
, then wf is smooth in .

Proof. For brevity, we write w = wf .


We rst consider a special case where f has a compact support in .
For any x , we write

w(x) = (x y)f (y) dy.
Rn

We point out that the integration is in fact over a bounded region. Note that
is evaluated as a function of |x y|. By the change of variables z = y x,
we have 
w(x) = (z)f (z + x) dz.
Rn
By the assumption, f is at least C 1 . By a simple dierentiation under the
integral sign and an integration by parts, we obtain
 
wxi (x) = (z)fxi (z + x) dz = (z)fzi (z + x) dz
Rn Rn

= zi (z)f (z + x) dz.
Rn

For f C k1 () for some k 2, we can dierentiate under the integral


sign to get

xi w(x) =

zi (z)z f (z + x) dz,
Rn
for any Zn+ with || k 1. Hence, w is C k in . Moreover, if f is
smooth in , then w is smooth in . Next, we calculate w if f is at least
C 1 . For any x , we have

n  
n
w(x) = wxi xi (x) = zi (z)fzi (z + x) dz
i=1 Rn i=1
 
n
= lim zi (z)fzi (z + x) dz.
0 Rn \B
i=1
4.4. Poisson Equations 135

We note that f ( + x) has a compact support in Rn . An integration by parts


implies 

w(x) = lim (z)f (z + x) dSz ,
0 B
where is the unit exterior normal to the boundary B of the domain
Rn \ B , which points toward the origin. With r = |z|, we obtain


w(x) = lim (z)f (z + x) dSz
0 B r

1
= lim f (z + x) dSz = f (x),
0 n n1 B

by the explicit expression of .


Next, we consider the general case. For any x0 , we prove that w
is C k and w = f in a neighborhood of x0 . To this end, we take r <
dist(x0 , ) and a function C0 (Br (x0 )) with 1 in Br/2 (x0 ). Then
we write
 

w(x) = (x y) 1 (y) f (y) dy + (x y)(y)f (y) dy

= wI (x) + wII (x).
The rst integral is actually over \ Br/2 (x0 ) since 1 in Br/2 (x0 ). Then
there is no singularity in the rst integral if we restrict x to Br/4 (x0 ). Hence,
wI is smooth in Br/4 (x0 ) and wI = 0 in Br/4 (x0 ). For the second integral,
f is a C k1 -function of compact support in . We can apply what we just
proved in the special case to f . Then wII is C k in and wII = f .
Therefore, w is a C k -function in Br/4 (x0 ) and
w(x) = (x)f (x) = f (x),
for any x Br/4 (x0 ). Moreover, if f is smooth in , so are wII and w in
. 

Lemma 4.4.1 is optimal in the C -category in the sense that the smooth-
ness of f implies the smoothness of wf . However, it does not seem optimal
concerning nite dierentiability. For example, Lemma 4.4.1 asserts that
wf is C 2 in if f is C 1 in . Since f is related to second derivatives of wf ,
it seems natural to ask whether wf is C 2 in if f is continuous in . We
will explore this issue later.
We now prove a regularity result for general solutions of (4.4.1).
Theorem 4.4.2. Let be a domain in Rn and f be continuous in . Sup-
pose u C 2 () satises u = f in . If f C k1 () for some integer
k 3, then u C k (). Moreover, if f is smooth in , then u is smooth in
.
136 4. Laplace Equations

Proof. We take an arbitrary bounded subdomain  in and let wf, be


the Newtonian potential of f in  . By Lemma 4.4.1, if f C k1 () for
some integer k 3, then wf, is C k in  and wf, = f in  . Now we
set v = u wf, . Since u is C 2 in  , so is v. Then, v = u wf, = 0
in  . In other words, v is harmonic in  , and hence is smooth in  by
Theorem 4.1.10. Therefore, u = v + wf, is C k in  . It is obvious that if
f is smooth in , then wf, and hence u are smooth in  . 

Theorem 4.4.2 is an optimal result concerning the smoothness. Even


though u is just one particular combination of second derivatives of u, the
smoothness of u implies the smoothness of all second derivatives.
Next, we solve the Dirichlet problem for the Poisson equation.
Theorem 4.4.3. Let be a bounded domain in Rn satisfying the exterior
sphere condition at every boundary point, f be a bounded C 1 -function in
and be a continuous function on . Then there exists a solution
u C 2 () C() of the Dirichlet problem
u = f in ,
u= on .
Moreover, if f is smooth in , then u is smooth in .

Proof. Let w be the Newtonian potential of f in . By Lemma 4.4.1 for


k = 2, w C 2 () C() and w = f in . Now consider the Dirichlet
problem
v = 0 in ,
v =w on .
Theorem 4.3.23 implies the existence of a solution v C () C(). (The
exterior sphere condition is needed in order to apply Theorem 4.3.23.) Then
u = v + w is the desired solution of the Dirichlet problem in Theorem 4.4.3.
If f is smooth in , then u is smooth there by Theorem 4.4.2. 

Now we raise a question concerning regularity of the lowest order in the


classical sense. What is the optimal assumption on f to yield a C 2 -solution
u of (4.4.1)? We note that the Laplace operator acts on C 2 -functions
and u is continuous for any C 2 -function u. It is natural to ask whether
the equation (4.4.1) admits any C 2 -solutions if f is continuous. The answer
turns out to be negative. There exists a continuous function f such that
(4.4.1) does not admit any C 2 -solutions.
Example 4.4.4. Let f be a function in B1 R2 dened by f (0) = 0 and
!
x22 x21 2 1
f (x) = + ,
|x|2 ( log |x|)1/2 4( log |x|)3/2
4.4. Poisson Equations 137

for any x B1 \ {0}. Then f is continuous in B1 . Consider


(4.4.3) u = f in B1 .
Dene u in B1 by u(0) = 0 and

u(x) = (x21 x22 )( log |x|)1/2 ,


for any x B1 \ {0}. Then u C(B1 ) C (B1 \ {0}). A straightforward
calculation shows that u satises (4.4.3) in B1 \ {0} and
lim ux1 x1 (x) = .
x0

Hence, u is not in C 2 (B1 ). Next, we prove that (4.4.3) has no C 2 -solutions.


The proof is based on Theorem 4.3.16 concerning removable singularities
of harmonic functions. Suppose, to the contrary, that there exists a C 2 -
solution v of (4.4.3) in B1 . For a xed R (0, 1), the function w = u v
is harmonic in BR \ {0}. Now u C(BR ) and v C 2 (BR ), so w C(BR ).
Thus w is continuous at the origin. By Theorem 4.3.16, w is harmonic in
BR and therefore belongs to C 2 (BR ). In particular, u is C 2 at the origin,
which is a contradiction.

Example 4.4.4 illustrates that the C 2 -spaces, or any C k -spaces, are not
adapted to the Poisson equation. A further investigation reveals that solu-
tions in this example fail to be C 2 because the modulus of continuity of f
does not decay to zero fast enough. If there is a better assumption than the
continuity of f , then the modulus of continuity of 2 u can be estimated.
Better adapted to the Poisson equation, or more generally, the elliptic dif-
ferential equations, are Holder spaces. The study of the elliptic dierential
equations in Holder spaces is known as the Schauder theory. In its simplest
form, it asserts that all second derivatives of u are Holder continuous if u
is. It is beyond the scope of this book to give a presentation of the Schauder
theory.

4.4.2. Weak Solutions. In the following, we discuss briey how to extend


the notion of classical solutions of the Poisson equation to less regularized
solutions, the so-called weak solutions. These functions have derivatives
only in an integral sense and satisfy the Poisson equation also in an integral
sense. The same process can be applied to general linear elliptic equations,
or even nonlinear elliptic equations, of divergence form.
To introduce weak solutions, we make use of the divergence structure or
variation structure of the Laplace operator. Namely, we write the Laplace
operator as
u = div(u).
138 4. Laplace Equations

In fact, we already employed such a structure when we derived energy es-


timates of solutions of the Dirichlet problem for the Poisson equation in
Section 3.2.
Let be a bounded domain in Rn and f be a bounded continuous
function in . Consider
(4.4.4) u = f in .
We intentionally put a negative sign in front of u. Let u C 2 () be a
solution of (4.4.4). Take an arbitrary C01 (). By multiplying (4.4.4) by
and then integrating by parts, we obtain
 
(4.4.5) u dx = f dx.

In (4.4.5), is referred to as a test function. We note that upon integrating
by parts, we transfer one derivative for u to test functions. Hence we only
need to require u to be C 1 in (4.4.5). This is the advantage in formulating
weak solutions.
If u C 2 () satises (4.4.5) for any C01 (), we obtain from (4.4.5),
upon a simple integration by parts,
 
u dx = f dx for any C01 ().

This implies easily
u = f in .
In conclusion, a C 2 -function u satisfying (4.4.5) for any C01 () is a
classical solution of (4.4.4).
We now raise the question whether less regular functions u are allowed
in (4.4.5). For any C01 (), it is obvious that the integral in the left-
hand side of (4.4.5) makes sense if each component of u is an integrable
function in . This suggests that we should introduce derivatives in the
integral sense.
Denition 4.4.5. For i = 1, , n, an integrable function u in is said to
have a weak xi -derivative if there exists an integrable function vi such that
 
(4.4.6) uxi dx = vi dx for any C01 ().

Here vi is called a weak xi -derivative of u and is denoted by uxi , the same
way as for classical derivatives.

It is easy to see that weak derivatives are unique if they exist. We also
point out that classical derivatives of C 1 -functions are weak derivatives upon
a simple integration by parts.
4.4. Poisson Equations 139

Denition 4.4.6. The Sobolev space H 1 () is the collection of L2 -functions


in with L2 -weak derivatives in .

The superscript 1 in the notation H 1 () indicates the order of dieren-


tiation. In general, functions in H 1 () may not have classical derivatives.
In fact, they may not be continuous.
We are ready to introduce weak solutions.
Denition 4.4.7. Let f L2 () and u H 1 (). Then u is a weak solution
of u = f in if (4.4.5) holds for any C01 (), where the components
of u are given by weak derivatives of u.

We now consider the Dirichlet problem for the Poisson equation with
the homogeneous boundary value,
u = f in ,
(4.4.7)
u=0 on .
We attempt to solve (4.4.7) by methods of functional analysis. It is natural
to start with the set
C = {u C 1 () C() : u = 0 on }.
We note that the left-hand side of (4.4.5) provides an inner product in C.
To be specic, we dene the H01 -inner product by

(u, v)H01 () = u v dx,

for any u, v C. It induces a norm dened by
 1
2
uH01 () = |u| dx .
2

This is simply the L2 -norm of the gradient of u, and it is referred to as the


H01 -norm. Here in the notation H01 , the supercript 1 indicates the order of
dierentiation and the subscript 0 refers to functions vanishing on . The
Poincare inequality in Lemma 3.2.2 has the form
(4.4.8) uL2 () CuH01 () ,
for any u C.
For f L2 (), we dene a linear functional F on C by

(4.4.9) F () = f dx,

for any C. By the Cauchy inequality and (4.4.8), we have
|F ()| f L2 () L2 () Cf L2 () H01 () .
140 4. Laplace Equations

This means that F is a bounded linear functional on C. If C were a Hilbert


space with respect to the H01 -inner product, we would conclude by the Riesz
representation theorem that there exists a u C such that
(u, )H01 () = F (),
for any C. Hence, u satises (4.4.5). With u = 0 on , u is interpreted
as a weak solution of (4.4.7).
However, C is not complete with respect to the H01 -norm, for the same
reason that C() is not complete with respect to the L2 -norm. For the
remedy, we complete C under the H01 -norm.
Denition 4.4.8. The Sobolev space H01 () is the completion of C01 ()
under the H01 -norm.

We point out that we may dene H01 () by completing C under the


H01 -norm. It yields the same space.
The space H01 () dened in Denition 4.4.8 is abstract. So what are the
elements in H01 ()? The next result provides an answer.
Theorem 4.4.9. The space H01 () is a subspace of H 1 () and is a Hilbert
space with respect to the H01 -inner product.

Proof. We take a sequence {uk } in C01 () which is a Cauchy sequence in


the H01 ()-norm. In other words, {uk,xi } is a Cauchy sequence in L2 (), for
any i = 1, , n. Then there exists a vi L2 (), for i = 1, , n, such that
uk,xi vi in L2 () as k .
By (4.4.8), we obtain
uk ul L2 () Cuk ul H01 () .
This implies that {uk } is a Cauchy sequence in L2 (). We may assume for
some u L2 () that
uk u in L2 () as k .
Such a convergence illustrates that elements in H01 () can be identied as
L2 -functions. Hence we have established the inclusion H01 () L2 ().
Next, we prove that u has L2 -weak derivatives. Since uk C01 (), upon a
simple integration by parts, we have
 
uk xi dx = uk,xi dx for any C01 ().

By taking k , we obtain easily
 
uxi dx = vi dx for any C01 ().

4.4. Poisson Equations 141

Therefore, vi is the weak xi -derivative of u. Then u H 1 () since vi


L2 (). In conclusion, H01 () H 1 ().
With weak derivatives replacing classical derivatives, the inner product
(, )H01 () is well dened for functions in H01 (). We then conclude that
H01 () is complete with respect to its induced norm  H01 () . 

It is easy to see by approximations that (4.4.8) holds for functions in


H01 ().
Now we can prove the existence of weak solutions of the Dirichlet prob-
lem for the Poisson equation with homogeneous boundary value.
Theorem 4.4.10. Let be a bounded domain in Rn and f L2 (). Then
the Poisson equation u = f admits a weak solution u H01 ().

The proof is based on the Riesz representation theorem, and major steps
are already given earlier.

Proof. We dene a linear functional F on H01 () by



F () = f dx,

for any H01 (). By the Cauchy inequality and (4.4.8), we have
|F ()| f L2 () L2 () Cf L2 () H01 () .

Hence, F is a bounded linear functional on H01 (). By the Riesz represen-


tation theorem, there exists a u H01 () such that
(u, )H01 () = F (),

for any H01 (). Therefore, u is the desired function. 

According to Denition 4.4.7, u in Theorem 4.4.10 is a weak solution


of u = f . Concerning the boundary value, we point out that u is not
dened on in the pointwise sense. We cannot conclude that u = 0 at
each point on . The boundary condition u = 0 on is interpreted
precisely by the fact that u H01 (), i.e., u is the limit of a sequence of
C01 ()-functions in the H01 -norm. One consequence is that u| is a well-
dened zero function in L2 (). Hence, u is referred to as a weak solution
of the Dirichlet problem (4.4.7).
Now we ask whether u possesses better regularity. The answer is ar-
mative. To see this, we need to introduce more Sobolev spaces. We rst
point out that weak derivatives as dened in (4.4.6) can be generalized to
higher orders. For any Zn+ with || = m, an integrable function u in
142 4. Laplace Equations

is said to have a weak x -derivative if there exists an integrable function v


such that
 

u dx = (1) m
v dx for any C0m ().

Here v is called a weak x -derivative of u and is denoted by u, the same
notation as for classical derivatives. For any positive integer m, we denote
by H m () the collection of L2 -functions with L2 -weak derivatives of order
up to m in . This is also a Sobolev space. The superscript m indicates the
order of dierentiation.
We now return to Theorem 4.4.10. We assume, in addition, that is
a bounded smooth domain. With f L2 (), the solution u in fact is a
function in H 2 (). In other words, u has L2 -weak second derivatives uxi xj ,
for i, j = 1, , n. Moreover,

n
uxi xi = f a.e. in .
i=1

In fact, if f H k () for any k 1, then u H k+2 (). This is the L2 -theory


for the Poisson equation. We again encounter an optimal regularity result.
If u is in the space H k (), then all second derivatives are in the same
space. It is beyond the scope of this book to give a complete presentation
of the L2 -theory.
An alternative method to prove the existence of weak solutions is to
minimize the functional associated with the Poisson equation. Let be a
bounded domain in Rn . For any C 1 -function u in , we dene the Dirichlet
energy of u in by 
1
E(u) = |u|2 dx.
2
For any f L2 (), we consider
  
1
J(u) = E(u) f u dx = |u|2 dx f u dx.
2

For any u C 1 () C(), we consider C 1 -perturbations of u which leave


the boundary values of u unchanged. We usually write such perturbations
in the form of u + for C01 (). We now compare J(u + ) and J(u).
A straightforward calculation yields
 
J(u + ) = J(u) + E() + u dx f dx.

We note that E() 0. Hence, if u is a weak solution of u = f , we
have, by (4.4.5),
J(u + ) J(u) for any C01 ().
4.5. Exercises 143

Therefore, u minimizes J among all functions with the same boundary value.
Now we assume that u minimizes J among all functions with the same
boundary value. Then for any C01 (),
J(u + ) J(u) for any .
In other words,
j() J(u + )
has a minimum at = 0. This implies j  (0) = 0. A straightforward calcula-
tion shows that u satises (4.4.5) for any C01 (). Therefore, u is a weak
solution of u = f . In conclusion, u is a weak solution of u = f if and
only if u minimizes J among all functions with the same boundary value.
The above calculation was performed for functions in C 1 (). A similar cal-
culation can be carried out for functions in H01 (). Hence, an alternative
way to solve (4.4.7) in the weak sense is to minimize J in H01 (). We will
not provide details in this book.
The weak solutions and Sobolev spaces are important topics in PDEs.
The brief discussion here serves only as an introduction. A complete pre-
sentation will constitute a book much thicker than this one.

4.5. Exercises
Exercise 4.1. Suppose u(x) is harmonic in some domain in Rn . Prove that

x
v(x) = |x|2n u
|x|2
is also harmonic in a suitable domain.
Exercise 4.2. For n = 2, nd the Greens function for the Laplace operator
on the rst quadrant.
Exercise 4.3. Find the Greens function for the Laplace operator in the
upper half-space {xn > 0} and then derive a formal integral representation
for a solution of the Dirichlet problem
u = 0 in {xn > 0},
u= on {xn = 0}.
Exercise 4.4. (1) Suppose u is a nonnegative harmonic function in
BR (x0 ) Rn . Prove by the Poisson integral formula the following
Harnack inequality:
n2 n2
R Rr R R+r
u(x0 ) u(x) u(x0 ),
R+r R+r Rr Rr
where r = |x x0 | < R.
144 4. Laplace Equations

(2) Prove by (1) the Liouville theorem: If u is a harmonic function in


Rn and bounded above or below, then u is constant.

Exercise 4.5. Let u be a harmonic function in Rn with Rn |u|p dx < for
some p (1, ). Prove that u 0.
Exercise 4.6. Let m be a positive integer and u be a harmonic function in
Rn with u(x) = O(|x|m ) as |x| . Prove that u is a polynomial of degree
at most m.
Exercise 4.7. Suppose u C(B1+ ) is harmonic in B1+ = {x B1 : xn > 0}
with u = 0 on {xn = 0} B1 . Prove that the odd extension of u to B1 is
harmonic in B1 .
Exercise 4.8. Let u be a C 2 -solution of
u = 0 in Rn \ BR ,
u = 0 on BR .
Prove that u 0 if
u(x)
lim =0 for n = 2,
|x| ln |x|
lim u(x) = 0 for n 3.
|x|

Exercise 4.9. Let be a bounded C 1 -domain in Rn satisfying the exterior


sphere condition at every boundary point and f be a bounded continuous
function in . Suppose u C 2 () C 1 () is a solution of
u = f in ,
u = 0 on .
Prove that  
 u 
sup   C sup |f |,

where C is a positive constant depending only on n and .
Exercise 4.10. Let be a smooth bounded domain in Rn , c be a continuous
function in with c 0 and be a continuous function on with 0.
Discuss the uniqueness of the problem
u + cu = f in ,
u
+ u = on .

Exercise 4.11. Let be a bounded C 1 -domain and let and be con-
tinuous functions on with 0 for a positive constant 0 . Suppose
4.5. Exercises 145

u C 2 () C 1 () satises
u + u3 = 0 in ,
u
+ u = on .

Prove that
1
max |u| max ||.
0
Exercise 4.12. Let f be a continuous function in BR . Suppose u
C 2 (BR ) C(BR ) satises
u = f in BR .
Prove that
n R
|u(0)| max |u| + max |f |.
R BR 2 BR
+
Hint: In BR , set
1
v(x , xn ) =
u(x , xn ) u(x , xn ) .
2
Consider an auxiliary function of the form
w(x , xn ) = A|x |2 + Bxn + Cx2n .
+
Use the comparison principle to estimate v in BR and then derive a bound
for vxn (0).

Exercise 4.13. Let u be a nonzero harmonic function in B1 Rn and set



r Br |u|2 dx
N (r) =  2
for any r (0, 1).
Br u dS

(1) Prove that N (r) is a nondecreasing function in r (0, 1) and iden-


tify
lim N (r).
r0+
(2) Prove that, for any 0 < r < R < 1,
 2N (R) 
1 R 1
u dS
2
u2 dS.
Rn1 BR r rn1 Br
Remark: The quantity N (r) is called the frequency. The estimate in (2) for
R = 2r is referred to as the doubling condition.

Exercise 4.14. Let be a bounded domain in Rn and f be a bounded


function in . Suppose wf is the Newtonian potential dened in (4.4.2).
146 4. Laplace Equations

(1) Prove that wf C 1 (Rn ) and



xi wf (x) = xi (x y)f (y) dy,

for any x Rn and i = 1, , n.
(2) Assume, in addition, that f is C in for some (0, 1), i.e., for
any x, y ,
|f (x) f (y)| C|x y| .
Prove that wf C 2 (), wf = f in and the second derivatives
of wf are C in .
Chapter 5

Heat Equations

The n-dimensional heat equation is given by ut u = 0 for functions


u = u(x, t), with x Rn and t R. Here, x is the space variable and
t the time variable. The heat equation models the temperature of a body
conducting heat when the density is constant. Solutions of the heat equation
share many properties with harmonic functions, solutions of the Laplace
equation.
In Section 5.1, we briey introduce Fourier transforms. The Fourier
transform is an important subject and has a close connection with many
elds of mathematics, especially with partial dierential equations. In the
rst part of this section, we discuss basic properties of Fourier transforms
and prove the important Fourier inversion formula. In the second part, we
use Fourier transforms to discuss several dierential equations with constant
coecients, including the heat equation, and we derive explicit expressions
for their solutions.
In Section 5.2, we discuss the fundamental solution of the heat equation
and its applications. We rst discuss the initial-value problem for the heat
equation. We prove that the explicit expression for its solution obtained
formally by Fourier transforms indeed yields a classical solution under ap-
propriate assumptions on initial values. Then we discuss regularity of ar-
bitrary solutions of the heat equation using the fundamental solution and
derive interior gradient estimates.
In Section 5.3, we discuss the maximum principle for the heat equation
and its applications. We rst prove the weak maximum principle and the
strong maximum principle for a class of parabolic equations more general
than the heat equation. As applications, we derive a priori estimates of so-
lutions of the initial/boundary-value problem and the initial-value problem.

147
148 5. Heat Equations

We also derive interior gradient estimates by the maximum principle. In


the nal part of this section, we study the Harnack inequality for positive
solutions of the heat equation. We point out that the Harnack inequality for
the heat equation is more complicated than that for the Laplace equation
we discussed earlier.
As in Chapter 4, several results in this chapter are proved by multi-
ple methods. For example, interior gradient estimates are proved by two
methods: the fundamental solution and the maximum principle.

5.1. Fourier Transforms


The Fourier transform is an important subject and has a close connection
with many elds of mathematics. In this section, we will briey introduce
Fourier transforms and illustrate their applications by studying linear dif-
ferential equations with constant coecients.

5.1.1. Basic Properties. We dene the Schwartz class S as the collection


of all complex-valued functions u C (Rn ) such that x x u(x) is bounded
in Rn for any , Zn+ , i.e.,
sup |x x u(x)| < .
xRn
In other words, the Schwartz class consists of smooth functions in Rn all of
whose derivatives decay faster than any polynomial at innity. It is easy to
check that u(x) = e|x| is in the Schwartz class.
2

Denition 5.1.1. For any u S, the Fourier transform u of u is dened


by 
1
() =
u n eix u(x) dx for any Rn .
(2) 2 Rn
We note that the integral on the right-hand side makes sense for u S.
In fact, 
1
|
u()| n |u(x)| dx for any Rn ,
(2) 2 R n

or
1
sup |
u| n uL1 (Rn ) .
Rn (2) 2
This suggests that Fourier transforms are well dened for L1 -functions. We
will not explore this issue in this book.
We now discuss properties of Fourier transforms. First, it is easy to
see that the Fourier transformation is linear, i.e., for any u1 , u2 S and
c1 , c2 C,
1 + c2 u
(c1 u1 + c2 u2 )= c1 u 2 .
5.1. Fourier Transforms 149

The following result illustrates an important property of Fourier trans-


forms.
Lemma 5.1.2. Let u S. Then u
 S and for any multi-indices , Zn+ ,

"

x u() = (i) u()
and
() = (i)|| x"
u u().

Proof. Upon integrating by parts, we have



1
"
u() =
x n eix u(x) dx
(2) 2 Rn

1
= n (i) eix u(x) dx = (i) u
().
(2) 2 Rn
Next, it follows easily from the denition of u  C (Rn ). Then we
 that u
have

1

() =
u n

eix u(x) dx
(2) 2 Rn

1
= n (ix) eix u(x) dx = (i)|| x"
u().
(2) 2 Rn
The interchange of the order of dierentiation and integration is valid be-
cause x u S. To prove u  S, we take any two multi-indices and .
It suces to prove that u
() is bounded in Rn . For this, we rst note
that
() =(i)|| x"
u "
u() = (i)||+|| (i) x u()

=(i)||+|| x
(x u)()

1 ||+||

= n (i) eix x x u(x) dx.
(2) 2 Rn

Hence 
1
sup | u
()| n |x x u(x) | dx < ,
Rn (2) 2 Rn
since each term in the integrand decays faster than any polynomial because
x u S. 

The next result relates Fourier transforms to translations and dilations.


Lemma 5.1.3. Let u S, a Rn , and k R \ {0}. Then

u( a)() = eia u
(),
150 5. Heat Equations

and 
 1
u(k)() = 
u .
|k| n k

Proof. By a simple change of variables, we have



 1
u( a)() = n eix u(x a) dx
(2) 2 Rn

1
= n ei(x+a) u(x) dx = eia u
().
(2) 2 Rn
By another change of variables, we have

 1
u(k)() = n eix u(kx) dx
(2) 2 Rn
 
1 i x n 1
= e k u(x)|k| dx = 
u .
|k|
n n
(2) 2 Rn k
We then obtain the desired results. 

For any u, v S, it is easy to check that u v S, where u v is the


convolution of u and v dened by

(u v)(x) = u(x y)v(y) dy.
Rn

Lemma 5.1.4. Let u, v S. Then


n
u"
v() = (2) 2 u
()
v ().

Proof. By the denition of the Fourier transform, we have



1

u v() = n eix u v(x) dx
(2) 2 R n
  
1 ix
= n e u(x y)v(y) dy dx
(2) 2 Rn Rn
 
1
= n ei(xy) u(x y)eiy v(y) dydx
(2) 2 Rn Rn
  
1 iy i(xy)
= n e v(y) e u(x y) dx dy
(2) 2 Rn Rn

eiy v(y) dy = (2) 2 u
n
=u() ()
v ().
Rn

The interchange of the order of integrations can be justied by Fubinis


theorem. 
5.1. Fourier Transforms 151

To proceed, we note that



ex dx =
2
.

The next result will be useful in the following discussions.
Proposition 5.1.5. Let A be a positive constant and u be the function
dened in Rn by
u(x) = eA|x| .
2

Then
1 ||2
4A
() =
u n e .
(2A) 2

Proof. By the denition of Fourier transforms, we have


 n 
1 ixA|x|2 1
eixk k Axk dxk .
2
() =
u n e dx = 1
(2) 2 Rn k=1 (2)
2

Hence it suces to compute, for any R,



1
eitAt dt.
2
1
(2) 2

After the change of variables s = t A, we have
 
2 (t A+i
)2
eitAt dt = e 4A
2
e 2 A dt

 
1 2 (s+i )2 1 2
ez dz,
2
= e 4A e 2 A ds = e 4A
A A L

where L is the straight line Im z = /2 A in the complex z-plane. By the
Cauchys integral theorem and the fact that the integrand decays at the
exponential rate as Re z , we have
 
z 2
et dt = .
2
e dz =
L
Hence 
1 1 2
eitAt dt = e 4A .
2
1 1
(2) 2 (2A) 2

Therefore, 
1 1
eixA|x| dx = 1 ||2
2
n n e 4A .
(2) 2 Rn (2A) 2
This yields the desired result. 

We now prove the Fourier inversion formula, one of the most important
results in the theory of Fourier transforms.
152 5. Heat Equations

Theorem 5.1.6. Suppose u S. Then



1
(5.1.1) u(x) = n () d.
eix u
(2) 2 Rn
The right-hand side of (5.1.1) is the Fourier transform of u  evaluated
at x. Hence, u(x) = ( u) (x). It follows that the Fourier transformation
u u
 is a one-to-one map of S onto S.
A natural attempt to prove (5.1.1) is to use the denition of Fourier
transforms and rewrite the right-hand side as
  
1 ix iy
e e u(y) dy d.
(2)n Rn Rn
However, as an integral in terms of (y, ) Rn Rn , it is not absolutely
convergent.

Proof. Letting A = 1/2 in Proposition 5.1.5, we see that if


u0 (x) = e 2 |x| ,
1 2
(5.1.2)
then
0 () = e 2 || .
1 2
u
Since u0 (x) = u0 (x), we conclude (5.1.1) for u = u0 . Now we prove (5.1.1)
for any u S.
We rst consider u S with u(0) = 0. We claim that there exist
v1 , , vn S such that

n
u(x) = xj vj (x) for any x Rn .
j=1

To see this, we note that


 1   1 
d n n
u(x) = u(tx) dt = xj uxj (tx) dt = xj wj (x),
0 dt 0
j=1 j=1

for some wj C (Rn ), j = 1, , n. By taking C0 (Rn ) with = 1 in


B1 , we write

u(x) = (x)u(x) + 1 (x) u(x)
n 
xj
= xj (x)wj (x) + 2 1 (x) u(x) .
|x|
j=1

We note that functions in the parentheses are in S, for j = 1, , n. This


proves the claim. Lemma 5.1.2 implies
n
() =
u ij vj ().
j=1
5.1. Fourier Transforms 153

We note that vj S by Lemma 5.1.2. Upon evaluating the right-hand side
of (5.1.1) at x = 0, we obtain
  
n
1 1
n () d =
u n ij vj () d = 0.
(2) 2 Rn (2) 2 Rn j=1

We conclude that (5.1.1) holds at x = 0 for all u S with u(0) = 0.


We now consider an arbitrary u S and decompose

u = u(0)u0 + u u(0)u0 ,

where u0 is dened in (5.1.2). First, (5.1.1) holds for u0 and hence for
u(0)u0 . Next, since u u(0)u0 is zero at x = 0, we see that (5.1.1) holds
for u u(0)u0 at x = 0. We obtain (5.1.1) for u at x = 0. Next, for any
x0 Rn , we consider v(x) = u(x + x0 ). By Lemma 5.1.3,

v() = eix0 u
().

Then by (5.1.1) for v at x = 0,


 
1 1
u(x0 ) = v(0) = n v() d = n eix0 u
() d.
(2) 2 Rn (2) 2 Rn

This proves (5.1.1) for u at x = x0 . 

Motivated by Theorem 5.1.6, we dene vq for any v S by



1
vq(x) = n eix v() d for any x Rn .
(2) 2 R n

The function vq is called the inverse Fourier transform of v. It is obvious


that u (x). Theorem 5.1.6 asserts that u = (
q(x) = u u)q.
Next, we set, for any u, v L2 (Rn ),

(u, v)L2 (Rn ) = uv dx.
Rn

The following result is referred to as the Parseval formula.

Theorem 5.1.7. Suppose u, v S. Then

u, v)L2 (Rn ) .
(u, v)L2 (Rn ) = (
154 5. Heat Equations

Proof. We note

u, v)L2 (Rn ) =
( ()
u v () d
Rn
  
1 ix
= n 
v () e u(x) dx d
(2) 2 Rn Rn
  
1
= n u(x) 
v ()e ix d dx
(2) 2 Rn Rn

= u(x)v(x) dx = (u, v)L2 (Rn ) ,
Rn
where we applied Theorem 5.1.6 to v. The interchange of the order of
integrations can be justied by Fubinis theorem. 

As a consequence, we have Plancherels theorem.


Corollary 5.1.8. Suppose u S. Then
uL2 (Rn ) = 
uL2 (Rn ) .

In other words, the Fourier transformation is an isometry in S with


respect to the L2 -norm.
Based on Corollary 5.1.8, we can extend Fourier transforms to L2 (Rn ).
Note that the Fourier transformation is a linear operator from S to S and
that S is dense in L2 (Rn ). For any u L2 (Rn ), we can take a sequence
{uk } S such that
uk u in L2 (Rn ) as k .
Corollary 5.1.8 implies
 l L2 (Rn ) = u
uk u k ul L2 (Rn ) = uk ul L2 (Rn ) .

Then, {uk } is a Cauchy sequence in L2 (Rn ) and hence converges to a limit


, i.e.,
in L2 (Rn ). This limit is dened as u
k u
u  in L2 (Rn ) as k .
 is well dened, independently of the
It is straightforward to check that u
choice of sequence {uk }.

5.1.2. Examples. The Fourier transform is an important tool in studying


linear partial dierential equations with constant coecients. We illustrate
this by two examples.
Example 5.1.9. Let f be a function dened in Rn . We consider
(5.1.3) u + u = f in Rn .
5.1. Fourier Transforms 155

Obviously, this is an elliptic equation. We obtained an energy identity in


Section 3.2 for solutions decaying suciently fast at innity. Now we at-
tempt to solve (5.1.3).
We rst seek a formal expression of its solution u by Fourier transforms.
In doing so, we will employ properties of Fourier transforms without justi-
cations. By taking the Fourier transform of both sides in (5.1.3), we obtain,
by Lemma 5.1.2,
(5.1.4) u() = f().
(1 + ||2 )
Then
f()
() =
u .
1 + ||2
By Theorem 5.1.6,

1 f()
(5.1.5) u(x) = eix d.
1 + ||2
n
(2) 2 Rn
It remains to verify that this indeed yields a classical solution under appro-
priate assumptions on f . Before doing so, we summarize the simple process
we just carried out. First, we apply Fourier transforms to equation (5.1.3).
Basic properties of Fourier transforms allow us to transfer the dierential
equation (5.1.3) for u to an algebraic equation (5.1.4) for u . By solving
this algebraic equation, we have an expression for u  in terms of f. Then,
by applying the Fourier inversion formula, we obtain u in terms of f. We
should point out that it is not necessary to rewrite u in an explicit form in
terms of f .
Proposition 5.1.10. Let f S and u be dened by (5.1.5). Then u is a
smooth solution of (5.1.3) in S. Moreover,
 
(|u| + 2|u| + | u| ) dx =
2 2 2 2
|f |2 dx.
Rn Rn

Proof. We note that the process described above in solving (5.1.3) by


Fourier transforms is rigorous if f S. In the following, we prove directly
from (5.1.5) that u is a smooth solution. By Lemma 5.1.2, f S for f S.
Then f/(1 + ||2 ) S. Therefore, u dened by (5.1.5) is in S by Lemma
5.1.2. For any multi-index Zn+ , we have

1 (i) f()
u(x) = eix d.
1 + ||2
n
(2) 2 Rn
In particular,
 
||2 f()
n
1
u(x) = uxk xk (x) = eix d,
1 + ||2
n
k=1
(2) 2 Rn
156 5. Heat Equations

and hence 
1
u(x) + u(x) = n eix f() d.
(2) 2 Rn

By Theorem 5.1.6, the right-hand side is f (x).


To prove the integral identity, we obtain from (5.1.4) that

|
u|2 + 2||2 | u|2 = |f|2 .
u|2 + ||4 |

By writing it in the form



n 
n
|
u| + 2
2
k2 |
u|2 + u|2 = |f|2 ,
k2 l2 |
k=1 k,l=1

we have, by Lemma 5.1.2,



n 
n
|
u|2 + 2 | 2
xk u| + | 2
xk xl u| = |f | .
2

k=1 k,l=1

A simple integration yields



 n 
n 
|
u| + 2
2 
|xk u| +
2  2
|xk xl u| d = |f|2 d.
Rn k=1 k,l=1 Rn

By Corollary 5.1.8, we obtain



 n 
n 
|u|2 + 2 |uxk | +
2 2
|uxk xl | dx = |f |2 dx.
Rn k=1 k,l=1 Rn

This is the desired identity. 

Example 5.1.11. Now we discuss the initial-value problem for the nonho-
mogeneous heat equation and derive an explicit expression for its solution.
Let f be a continuous function in Rn (0, ) and u0 a continuous function
in Rn . We consider
ut u = f in Rn (0, ),
(5.1.6)
u(, 0) = u0 on Rn .

Although called an initial-value problem, (5.1.6) is not the type of initial-


value problem we discussed in Section 3.1. The heat equation is of the second
order, while only one condition is prescribed on the initial hypersurface
{t = 0}, which is characteristic.
Suppose u is a solution of (5.1.6) in C 2 (Rn (0, )) C(Rn [0, )).
We now derive formally an expression of u in terms of Fourier transforms. In
5.1. Fourier Transforms 157

the following, we employ Fourier transforms with respect to space variables


only. With an obvious abuse of notation, we write

1
(, t) =
u n eix u(x, t) dx.
(2) 2 R n

We take Fourier transforms of both sides of the equation and the initial
condition in (5.1.6) and obtain, by Lemma 5.1.2,
 = f in Rn (0, ),
t + ||2 u
u
(, 0) = u
u 0 on Rn .
This is an initial-value problem for an ODE with Rn as a parameter.
Its solution is given by
 t
0 ()e|| t + e|| (ts) f(, s) ds.
2 2
u(, t) = u
0
Now we treat t as a parameter instead. For any t > 0, let K(x, t) satisfy
 t) = 1 ||2 t
K(, n e .
(2) 2
Then
 t
 t)
n
(, t) = (2) K(,
u 2
n
u0 () + (2) 2  t s)f(, s) ds.
K(,
0
Therefore Theorem 5.1.6 and Lemma 5.1.4 imply

u(x, t) = K(x y, t)u0 (y) dy
Rn
(5.1.7)  t
+ K(x y, t s)f (y, s) dyds,
0 Rn
for any (x, t) Rn (0, ). By Theorem 5.1.6 and Proposition 5.1.5, we
have 
1
eix e|| t d,
2
K(x, t) = n
(2) Rn
or
1 |x|2
4t
(5.1.8) K(x, t) = n e ,
(4t) 2
for any (x, t) Rn (0, ). The function K is called the fundamental
solution of the heat equation.
The derivation of (5.1.7) is formal. Having derived the integral formula
for u, we will prove directly that it indeed denes a solution of the initial-
value problem for the heat equation under appropriate assumptions on the
initial value u0 and the nonhomogeneous term f . We will pursue this in the
next section.
158 5. Heat Equations

5.2. Fundamental Solutions


In this section, we discuss the heat equation using the fundamental solution.
We rst discuss the initial-value problem for the heat equation. We prove
that the explicit expression for its solution obtained formally by Fourier
transforms indeed yields a classical solution under appropriate assumptions
on initial values. Then we discuss regularity of solutions of the heat equation.
Finally we discuss solutions of the initial-value problem for nonhomogeneous
heat equations.
The n-dimensional heat equation is given by

(5.2.1) ut u = 0,

for u = u(x, t) with x Rn and t R. We note that (5.2.1) is not preserved


by the change t t. This indicates that the heat equation describes
an irreversible process and distinguishes between past and future. This
fact will be well illustrated by the Harnack inequality, which we will derive
later in the next section. Next, (5.2.1) is preserved under linear transforms
x = x and t = 2 t for any nonzero constant , which leave the quotient
|x|2 /t invariant. Due to this fact, the expression |x|2 /t appears frequently in
connection with the heat equation (5.2.1). In fact, the fundamental solution
has such an expression.
If u is a solution of (5.2.1) in a domain in Rn R, then for any (x0 , t0 )
in this domain and appropriate r > 0,

ux0 ,r (x, t) = u(x0 + rx, t0 + r2 t)

is a solution of (5.2.1) in an appropriate domain in Rn R.


In the following, we denote by C 2,1 the collection of functions which are
C2 in x and C 1 in t. These are the functions for which the heat equation is
well dened classically.

5.2.1. Initial-Value Problems. We rst discuss the initial-value problem


for the heat equation. Let u0 be a continuous function in Rn . We consider
ut u = 0 in Rn (0, ),
(5.2.2)
u(, 0) = u0 on Rn .

We will seek a solution u C 2,1 (Rn (0, )) C(Rn [0, )).


We rst consider a special case where u0 is given by a homogeneous
polynomial P of degree d in Rn . We now seek a solution u in Rn (0, )
which is a p-homogeneous polynomial of degree d, i.e.,

u(x, 2 t) = d u(x, t),


5.2. Fundamental Solutions 159

for any (x, t) Rn (0, ) and > 0. To do this, we expand u as a power


series of t with coecients given by functions of x, i.e.,


u(x, t) = ak (x)tk .
k=0
Then a straightforward calculation yields
1
a0 = P, ak = ak1 for any k 1.
k
Therefore for any k 0,
1
ak = k P.
k!
Since P is a polynomial of degree d, it follows that [d/2]+1 P = 0, where
[d/2] is the integral part of d/2, i.e., [d/2] = d/2 if d is an even integer and
[d/2] = (d 1)/2 if d is an odd integer. Hence
[d]

2
1 k
u(x, t) = P (x)tk .
k!
k=0
We note that u in fact exists in Rn R. For n = 1, let ud be a p-homogeneous
polynomial of degree d in R R satisfying the heat equation and ud (x, 0) =
xd . The rst ve such polynomials are given by
u1 (x, t) = x,
u2 (x, t) = x2 + 2t,
u3 (x, t) = x3 + 6xt,
u4 (x, t) = x4 + 12x2 t + 12t2 ,
u5 (x, t) = x5 + 20x3 t + 60xt2 .
We now return to (5.2.2) for general u0 . In view of Example 5.1.11, we
set, for any (x, t) Rn (0, ),
1 |x|2
4t
(5.2.3) K(x, t) = n e ,
(4t) 2
and

(5.2.4) u(x, t) = K(x y, t)u0 (y) dy.
Rn
In Example 5.1.11, we derived formally by using Fourier transforms that any
solution of (5.2.2) is given by (5.2.4). Having derived the integral formula
for u, we will prove directly that it indeed denes a solution of (5.2.2) under
appropriate assumptions on the initial value u0 .
Denition 5.2.1. The function K dened in Rn (0, ) by (5.2.3) is called
the fundamental solution of the heat equation.
160 5. Heat Equations

We have the following result concerning properties of the fundamental


solution.
Lemma 5.2.2. Let K be the fundamental solution of the heat equation de-
ned by (5.2.3). Then
(1) K(x, t) is smooth for any x Rn and t > 0;
(2) K(x, t) > 0 for any x Rn and t > 0;
(3) (t )K(x, t) = 0 for any x Rn and t > 0;

(4) Rn K(x, t)dx = 1 for any t > 0;
(5) for any > 0,

lim K(x, t) dx = 0.
t0+ Rn \B

Proof. Here (1) and (2) are obvious from the explicit expression of K in
(5.2.3). We may also get (3) from (5.2.3) by a straightforward calculation.
For (4) and (5), we simply note that
 
1
e|| d.
2
K(x, t)dx = n
|x|> 2 ||> 2 t

This implies (4) for = 0 and (5) for > 0. 

6
y = K(, t2 )

y = K(, t1 )
-

Figure 5.2.1. Graphs of fundamental solutions for t2 > t1 > 0.

Now we are ready to prove that the integral formula derived by using
Fourier transforms indeed yields a classical solution of the initial-value prob-
lem for the heat equation under appropriate assumptions on u0 .
Theorem 5.2.3. Let u0 be a bounded continuous function in Rn and u be
dened by (5.2.4). Then u is smooth in Rn (0, ) and satises
ut u = 0 in Rn (0, ).
5.2. Fundamental Solutions 161

Moreover, for any x0 Rn ,


lim u(x, t) = u0 (x0 ).
(x,t)(x0 ,0)

We note that the function u in (5.2.4) is dened only for t > 0. We can
extend u to {t = 0} by setting u(, 0) = u0 on Rn . Then u is continuous
up to {t = 0} by Theorem 5.2.3. Therefore, u is a classical solution of the
initial-value problem (5.2.2).
The proof of Theorem 5.2.3 proceeds as that of the Poisson integral
formula for the Laplace equation in Theorem 4.1.9.

Proof. Step 1. We rst prove that u is smooth in Rn (0, ). For any


multi-index Zn+ and any nonnegative integer k, we have formally

x tk u(x, t) = x tk K(x y, t)u0 (y) dy.
Rn
In order to justify the interchange of the order of dierentiation and in-
tegration, we need to check that, for any nonnegative integer m and any
t > 0,

|xy|2
|x y|m e 4t |u0 (y)| dy < .
Rn
This follows easily from the exponential decay of the integrand if t > 0.
Hence u is a smooth function in Rn (0, ). Then by Lemma 5.2.2(3),

(ut u)(x, t) = (Kt x K)(x y, t)u0 (y) dy = 0.
Rn
We point out for future references that we used only the boundedness of u0 .
Step 2. We now prove the convergence of u(x, t) to u0 (x0 ) as (x, t)
(x0 , 0). By Lemma 5.2.2(4), we have

u0 (x0 ) = K(x y, t)u0 (x0 ) dy.
Rn
Then


u(x, t) u0 (x0 ) = K(x y, t) u0 (y) u0 (x0 ) dy = I1 + I2 ,
Rn
where  
I1 = , I2 = ,
B (x0 ) Rn \B (x0 )
for a positive constant to be determined. For any given > 0, we can
choose = () > 0 small so that
|u0 (y) u0 (x0 )| < ,
162 5. Heat Equations

for any y B (x0 ), by the continuity of u0 . Then by Lemma 5.2.2(2) and


(4), 
|I1 | K(x y, t)|u0 (y) u0 (x0 )| dy .
B (x0 )
Since u0 is bounded, we assume that |u0 | M for some positive constant
M . We note that |x y| /2 for any y Rn \ B (x0 ) and x B/2 (x0 ).
By Lemma 5.2.2(5), we can nd a  > 0 such that


K(x y, t) dy ,
R \B (x0 )
n 2M
for any x B/2 (x0 ) and t (0,  ), where  depends on and = (),
and hence only on . Then


|I2 | K(x y, t) |u0 (y)| + |u0 (x0 )| dy .
Rn \B (x0 )

Therefore,
|u(x, t) u0 (0)| 2,
for any x B/2 (x0 ) and t (0,  ). We then have the desired result. 

Under appropriate assumptions, solutions dened by (5.2.4) decay as


time goes to innity.
Proposition 5.2.4. Let u0 L1 (Rn ) and u be dened by (5.2.4). Then for
any t > 0, 
1
sup |u(, t)| n |u0 | dx.
Rn (4t) 2 Rn

The proof follows easily from (5.2.4) and the explicit expression for the
fundamental solution K in (5.2.3).
Now we discuss a result more general than Theorem 5.2.3 by relaxing
the boundedness assumption on u0 . To seek a reasonably more general as-
sumption on initial values, we examine the expression for the fundamental
solution K. We note that K in (5.2.3) decays exponentially in space vari-
ables with a large decay rate for small time. This suggests that we can allow
an exponential growth for initial values. In the convolution formula (5.2.4),
a xed exponential growth from initial values can be oset by the fast ex-
ponential decay in the fundamental solution, at least for a short period of
time. To see this clearly, we consider an example. For any > 0, set
1
|x|2
G(x, t) = n e
14t ,
(1 4t) 2
for any x Rn and t < 1/4. It is straightforward to check that
Gt G = 0.
5.2. Fundamental Solutions 163

Note that
2
G(x, 0) = e|x| for any x Rn .
Hence, viewed as a function in Rn [0, 1/4), G has an exponential growth
initially for t = 0, and in fact for any t < 1/4. The growth rate becomes
arbitrarily large as t approaches 1/4 and G does not exist beyond t = 1/4.
Now we formulate a general result. If u0 is continuous and has an ex-
ponential growth, then (5.2.4) still denes a solution of the initial-value
problem in a short period of time.

Theorem 5.2.5. Suppose u0 C(Rn ) satises


2
|u0 (x)| M eA|x| for any x Rn ,
for some constants M, A 0. Then u dened by (5.2.4) is smooth in Rn
1
(0, 4A ) and satises

1
ut u = 0 in R 0,
n
.
4A
Moreover, for any x0 Rn ,
lim u(x, t) = u0 (x0 ).
(x,t)(x0 ,0)

The proof is similar to that of Theorem 5.2.3.

Proof. The case A = 0 is covered by Theorem 5.2.3. We consider only


A > 0. First, by the explicit expression for K in (5.2.3) and the assumption
on u0 , we have

M
e 4t |xy| +A|y| dy.
1 2 2
|u(x, t)| n
(4t) 2 R n

A simple calculation shows that


 2
1 1 4At  1 
 + A
|x y| + A|y| =
2 2
 y x  |x|2 .
4t 4t 1 4At 1 4At
Hence for any (x, t) Rn (0, 1/(4A)), we obtain
  
M A
|x|2 14At y 1 x2
|u(x, t)| n e 14At e 4t 14At dy
(4t) 2 Rn
M A
|x|2
n e
14At .
(1 4At) 2
The integral dening u in (5.2.4) is convergent absolutely and uniformly for
(x, t) Rn [, + 1/(4A)], for any > 0 small. Hence, u is continuous
164 5. Heat Equations

in Rn (0, 1/(4A)). To show that u has continuous derivatives of arbitrary


order in Rn (0, 1/(4A)), we need only verify

|x y|m e 4t |xy| +A|y| dy < ,
1 2 2

Rn

for any m 0. The proof for m 1 is similar to that for m = 0 and we


omit the details.
Next, we need to prove the convergence of u(x, t) to u0 (x0 ) as (x, t)
(x0 , 0). We leave the proof as an exercise. 

Now we discuss properties of the solution u given by (5.2.4) of the initial-


value problem (5.2.2). First for any xed x Rn and t > 0, the value of
u(x, t) depends on the values of u0 at all points. Equivalently, the values of
u0 near a point x0 Rn aect the value of u(x, t) at all x as long as t > 0.
We interpret this by saying that the eects travel at an innite speed. If the
initial value u0 is nonnegative everywhere and positive somewhere, then the
solution u in (5.2.4) at any later time is positive everywhere. We will see
later that this is related to the strong maximum principle.
Next, the function u(x, t) in (5.2.4) becomes smooth for t > 0, even if
the initial value u0 is simply bounded. This is well illustrated in Step 1 in
the proof of Theorem 5.2.3. We did not use any regularity assumption on u0
there. Compare this with Theorem 3.3.5. Later on, we will prove a general
result that any solutions of the heat equation in a domain in Rn (0, )
are smooth away from the boundary. Refer to a similar remark at the end
of Subsection 4.1.2 for harmonic functions dened by the Poisson integral
formula.
We need to point out that (5.2.4) represents only one of innitely many
solutions of the initial-value problem (5.2.2). The solutions are not unique
without further conditions on u, such as boundedness or exponential growth.
In fact, there exists a nontrivial solution u C (Rn R) of ut u = 0,
with u 0 for t 0. In the following, we construct such a solution of the
one-dimensional heat equation.

Proposition 5.2.6. There exists a nonzero smooth function u C (R


[0, )) satisfying

ut uxx = 0 in R [0, ),
u(, 0) = 0 on R.

Proof. We construct a smooth function in R R such that ut uxx = 0


in R R and u 0 for t < 0. We treat {x = 0} as the initial curve and
5.2. Fundamental Solutions 165

attempt to nd a smooth solution of the initial-value problem


ut uxx = 0 in R R,
u(0, t) = a(t), ux (0, t) = 0 for t R,
for an appropriate function a in R. We write u as a power series in x:


u(x, t) = ak (t)xk .
k=0
Making a simple substitution in the equation ut = uxx and comparing the
coecients of powers of x, we have
ak2 = k(k 1)ak for any k 2.
Evaluating u and ux at x = 0, we get
a0 = a, a1 = 0.
Hence for any k 0,
1 (k)
a2k (t) = a (t),
(2k)!
and
a2k+1 (t) = 0.
Therefore, we have a formal solution

1 (k)
u(x, t) = a (t)x2k .
(2k)!
k=0
We need to choose a(t) appropriately so that u(x, t) dened above is a
smooth function and is identically zero for t < 0. To this end, we dene
 1
e t2 for t > 0,
a(t) =
0 for t 0.
Then it is straightforward to verify that the series dening u is absolutely
convergent in RR. This implies that u is continuous. In fact, we can prove
that series dening arbitrary derivatives of u are also absolutely convergent
in RR. We skip the details and leave the rest of the proof as an exercise. 

Next, we discuss briey terminal-value problems. For a xed constant


T > 0, we consider
ut uxx = 0 in R (0, T ),
u(, T ) = on R.
Here the function is prescribed at the terminal time T . This problem is
not well posed. Consider the following example. For any positive integer m,
let
um (x, t) = em (T t) sin(mx),
2
166 5. Heat Equations

for any (x, t) R [0, T ). Then um solves this problem with the terminal
value
m (x) = um (x, T ) = sin(mx),
for any x R. We note that
sup |m | = 1,
R
and for any t [0, T ),
2 (T t)
sup |um (, t)| = em as m .
R
There is no continuous dependence of solutions on the values prescribed at
the terminal time T .

5.2.2. Regularity of Solutions. Next, we discuss regularity of solutions


of the heat equation with the help of the fundamental solution. We will do
this only in special domains.
For any (x0 , t0 ) Rn R and any R > 0, we dene
QR (x0 , t0 ) = BR (x0 ) (t0 R2 , t0 ].
We point out that subsets of the form QR (x0 , t0 ) play the same role for the
heat equation as balls for the Laplace equation. If u is a solution of the heat
equation ut u = 0 in QR (0), then
uR (x, t) = u(Rx, R2 t)
is a solution of the heat equation in Q1 (0).

R(x0 , t0 )
 6

R2

Figure 5.2.2. The region QR (x0 , t0 ).

For any domain D in Rn R, we denote by C 2,1 (D) the collection of


functions in D which are C 2 in x and C 1 in t.
We rst have the following regularity result for solutions of the heat
equation.
5.2. Fundamental Solutions 167

Theorem 5.2.7. Let u be a C 2,1 -solution of ut u = 0 in QR (x0 , t0 ) for


some (x0 , t0 ) Rn R and R > 0. Then u is smooth in QR (x0 , t0 ).

Proof. For simplicity, we consider the case (x0 , t0 ) = (0, 0) and write
QR = BR (R2 , 0].
Without loss of generality, we assume that u is bounded in QR . Otherwise,
we consider u in Qr for any r < R.
We take an arbitrarily xed point (x, t) QR and claim that


u(x, t) = K x y, t + R2 u(y, R2 ) dy
BR
 t  #
u
+ K(x y, t s) (y, s)
R2 BR y
$
K
u(y, s) (x y, t s) dSy ds.
y
We rst assume this identity and prove that it implies the smoothness of u.
We note that the integrals in the right-hand side are only over the bottom
and the side of the boundary of BR (R2 , t]. The rst integral is over
BR {R2 }. For (x, t) QR , it is obvious that t + R2 > 0 and hence
there is no singularity in the rst integral. The second integral is over
BR (R2 , t]. By the change of variables = t s, we can rewrite it as
 t+R2  # $
u K
K(x y, ) (y, t ) u(y, t ) (x y, ) dSy d.
0 BR y y
There is also no singularity in the integrand since x BR , y BR , and
> 0. Hence, we conclude that u is smooth in QR .
We now prove the claim. Let K be the fundamental solution of the heat
equation as in (5.2.3). Denoting by (y, s) points in QR , we set
1 |xy|2
4(ts)
K(y, s) = K(x y, t s) = n e for s < t.
4(t s) 2
Then
Ks + y K = 0.
Hence,

n
0 = K(us y u) = (uK)s + (uKyi Kuyi )yi u(Ks + y K)
i=1

n
= (uK)s + (uKyi Kuyi )yi .
i=1
168 5. Heat Equations

For any > 0 with t > R2 , we integrate with respect to (y, s) in


BR (R2 , t ). Then

K(x y, )u(y, t ) dy
BR


= K x y, t (R2 ) u(y, R2 ) dy
BR
 t  #
u
+ K(x y, t s) (y, s)
R2 BR y
$
K
u(y, s) (x y, t s) dSy ds.
y
Now it suces to prove that

lim K(x y, )u(y, t )dy = u(x, t).
0 B
R

The proof proceeds similarly to that in Step 2 in the proof of Theorem 5.2.3.
The integral here over a nite domain introduces few changes here. We omit
the details. 

Now we prove interior gradient estimates.


Theorem 5.2.8. Let u be a bounded C 2,1 -solution of ut u = 0 in
QR (x0 , t0 ) for some (x0 , t0 ) Rn R and R > 0. Then
C
|x u(x0 , t0 )| sup |u|,
R QR (x0 ,t0 )
where C is a positive constant depending only on n.

Proof. We consider the case (x0 , t0 ) = (0, 0) and R = 1 only. The general
case follows from a simple translation and dilation. (Refer to Lemma 4.1.11
for a similar dilation for harmonic functions.) In the following, we write
Qr = Br (r2 , 0] for any r (0, 1].
We rst modify the proof of Theorem 5.2.7 to express u in terms of the
fundamental solution and cuto functions. We denote points in Q1 by (y, s).
Let K be the fundamental solution of the heat equation given in (5.2.3). As
in the proof of Theorem 5.2.7, we set, for any xed (x, t) Q1/4 ,
1 |xy|2
4(ts)
K(y, s) = K(x y, t s) = n e for s < t.
4(t s) 2
By choosing a cuto function C (Q1 ) with supp Q3/4 and 1
in Q1/2 , we set
v = K.
5.2. Fundamental Solutions 169

We need to point out that v(y, s) is dened only for s < t. For such a
function v, we have

n
0 = v(us y u) = (uv)s + (uvyi vuyi )yi u(vs + y v).
i=1
For any > 0, we integrate with respect to (y, s) in B1 (1, t). We note
that there is no boundary integral over B1 {1} and B1 (1, t ),
since vanishes there. Hence
 
(u)(y, t )K(x y, ) dy = u(s + y )(K) dyds.
B1 B1 (1,t)
Then similarly to the proof of Theorem 5.2.3, we have, as 0,

(x, t)u(x, t) = u(s + y )(K) dyds.
B1 (1,t)
In view of
Ks + y K = 0,
we obtain for any (x, t) Q1/4 that


u(x, t) = u (s + y )K + 2y y K dyds.
B1 (1,t)
We note that each term in the integrand involves a derivative of , which is
zero in Q1/2 since 1 there. Then the domain of integration D is actually
given by % %
D = B 3 (3/4)2 , t \ B 1 (1/2)2 , t .
4 2
The distance between any (y, s) D and any (x, t) Q1/4 has a positive
lower bound. Therefore, the integrand has no singularity in D. (This gives
an alternate proof of the smoothness of u in Q1/4 .)

D2 D2
D1

Figure 5.2.3. A decomposition of D for n = 1.

Next, we have, for any (x, t) Q1/4 ,




x u(x, t) = u (s + y )x K + 2y x y K dyds.
D
170 5. Heat Equations

Let C be a positive constant such that


2|y | C, |s | + |2y | C.
Hence 

|x u(x, t)| C |x K| + |x y K| |u| dyds.
D
By the explicit expression for K, we have
|x y| |xy| 2

|x K| C n e 4(ts) ,
(t s) 2 +1
and
|x y|2 + (t s) |xy| 2

|x y K| C n e 4(ts) .
(t s) 2 +2
Obviously, for any (x, t) Q1/4 and any (y, s) D,
|x y| 1, 0 < t s 1.
Therefore, for any (x, t) Q1/4 ,
2  |xy|2
1 4(ts)
|x u(x, t)| C n e |u(y, s)| dyds.
i=1 D (t s) 2 +i
Now we claim that, for any (x, t) Q1/4 , (y, s) D and i = 1, 2,
1 |xy|2
4(ts)
n e C.
(t s) 2 +i
Then we obtain easily for any (x, t) Q1/4 that
|x u(x, t)| C sup |u|.
Q1

To prove the claim, we decompose D into two parts,



D1 = B 1 (3/4)2 , (1/2)2 ,
2
D2 = B 3 \ B 1 (3/4)2 , t .
4 2

We rst consider D1 . For any (x, t) Q1/4 and (y, s) D1 , we have


1
ts ,
8
and hence
1 |xy|2
4(ts) n
n e 8 2 +i .
(t s) 2 +i

Next, we consider D2 . For any (x, t) Q1/4 and (y, s) D2 , we have


2
1 3
|y x| , 0 < t s < ,
4 4
5.2. Fundamental Solutions 171

and hence, with = (t s)1 ,

1 |xy|2 1
4(ts) 3 1
4 (ts) = 2 +i e 43 C,
n
n e n e
(t s) 2 +i
(t s) 2 +i

for any > (4/3)2 . This nishes the proof of the claim. 

Next, we estimate derivatives of arbitrary order.

Theorem 5.2.9. Let u be a bounded C 2,1 -solution of ut u = 0 in


QR (x0 , t0 ) for some (x0 , t0 ) Rn R and R > 0. Then for any nonnegative
integers m and k,

C m+2k k m+2k1
|tk m
x u(x0 , t0 )| n e (m + 2k)! sup |u|,
Rm+2k QR (x0 ,t0 )

where C is a positive constant depending only on n.

Proof. For x-derivatives, we proceed as in the proof of Theorem 4.1.12 and


obtain that, for any Zn+ with || = m,

C m em1 m!
|x u(x0 , t0 )| sup |u|.
Rm QR (x0 ,t0 )

For t-derivatives, we have ut = u and hence tk u = k u for any positive


integer k. We note that there are nk terms of x-derivatives of u of order 2k
in k u. Hence

|tk m
x u(x0 , t0 )| n
k
max |x u(x0 , t0 )|.
||=m+2k

This implies the desired result easily. 

The next result concerns the analyticity of solutions of the heat equation
on any time slice.

Theorem 5.2.10. Let u be a C 2,1 -solution of ut u = 0 in QR (x0 , t0 ) for


some (x0 , t0 ) Rn R and R > 0. Then u(, t) is analytic in BR (x0 ) for
any t (t0 R2 , t0 ]. Moreover, for any nonnegative integer k, tk u(, t) is
analytic in BR (x0 ) for any t (t0 R2 , t0 ].

The proof is identical to that of Theorem 4.1.14 and is omitted. In


general, solutions of ut u = 0 are not analytic in t. This is illustrated by
Proposition 5.2.6.
172 5. Heat Equations

5.2.3. Nonhomogeneous Problems. Now we discuss the initial-value


problem for the nonhomogeneous equation. Let f be continuous in Rn
(0, ). Consider
ut u = f in Rn (0, ),
u(, 0) = 0 on Rn .

Let K be the fundamental solution of the heat equation as in (5.2.3),


i.e.,
1 |x|2
4t
K(x, t) = n e ,
(4t) 2
for any (x, t) Rn (0, ). Dene
 t
(5.2.5) u(x, t) = K(x y, t s)f (y, s) dyds,
0 Rn

for any (x, t) Rn (0, ). If f is bounded in Rn (0, ), it is straight-


forward to check that the integral in the right-hand side of (5.2.5) is well
dened and continuous in (x, t) Rn (0, ). By Lemma 5.2.2(4), we have
 t
|u(x, t)| sup |f | K(y, s) dyds = t sup |f |.
Rn (0,t) 0 Rn Rn (0,t)

Hence
sup |u(, t)| 0 as t 0.
Rn
To discuss whether u is dierentiable, we note that
1 xi |x|2
Kxi (x, t) = n e 4t ,
(4t) 2 2t

1 xi xj ij |x|2
4t
Kxi xj (x, t) = n e .
(4t) 2 4t2 2t

For any t > 0, by the change of variables x = 2z t, we have
 
1 2 |zi | 1
|Kxi (x, t)| dx = n e|z| dz = ,
Rn 2 Rn t t
and   
1  ij  |z|2
|Kxi xj (x, t)| dx = n  zi zj e dz.
Rn 2 t Rn  2 
Hence Kxi L1 (Rn (0, T )) and Kxi xj / L1 (Rn (0, T )) for any T > 0. A
formal dierentiation of (5.2.5) yields
 t
(5.2.6) uxi (x, t) = Kxi (x y, t s)f (y, s) dyds.
0 Rn
5.2. Fundamental Solutions 173

We denote by I the integral in the right-hand side. If f is bounded in


Rn (0, ), then
 t
|I| sup |f | |Kxi (x y, t s)| dyds
Rn (0,t) 0 Rn
 t

1 2 t
sup |f |  ds = sup |f |.
Rn (0,t) 0 (t s) Rn (0,t)
Hence, the integral in the right-hand side of (5.2.6) is well dened and con-
tinuous in (x, t) Rn (0, ). We will justify (5.2.6) later in the proof
of Theorem 5.2.11 under extra assumptions. Even assuming the validity
of (5.2.6), we cannot continue dierentiating (5.2.6) to get the second x-
derivatives of u if f is merely bounded, since Kxi xj / L1 (Rn (0, T )) for
any T > 0. In order to get the second x-derivatives of u, we need extra
assumptions on f .
Theorem 5.2.11. Let f be a bounded continuous function in Rn (0, )
with bounded and continuous x f in Rn (0, ) and u be dened by (5.2.5)
for (x, t) Rn (0, ). Then u is C 2,1 in Rn (0, ) and satises
ut u = f in Rn (0, ),
and for any x0 Rn ,
lim u(x, t) = 0.
(x,t)(x0 ,0)
Moreover, if f is smooth with bounded derivatives of arbitrary order in Rn
(0, ), then u is smooth in Rn (0, ).

Proof. We rst assume that f and x f are continuous and bounded in


Rn (0, ).
By the explicit expression for K and the change of variables
y = x + 2z t s, we obtain from (5.2.5) that
 t
1
e|z| f (x + 2z t s, s) dzds,
2
(5.2.7) u(x, t) = n
2 0 Rn
for any (x, t) R (0, ). It follows easily that the limit of u(x, t) is zero
n

as t 0.
A simple dierentiation yields
 t
1
e|z| xi f (x + 2z t s, s) dzds
2
uxi (x, t) = n
2 0 Rn
 t
1 1
e|z|
2
= n zi f (x + 2z t s, s) dzds.
2 0 Rn 2 ts
Upon integrating by parts, we have
 t
1 zi
e|z|
2
uxi (x, t) = n f (x + 2z t s, s) dzds.
2 0 Rn ts
174 5. Heat Equations


(We note that this is (5.2.6) by the change of variables y = x + 2z t s.)
A dierentiation under the integral signs yields
 t
1 zi
e|z|
2
uxi xj (x, t) = n fxj (x + 2z t s, s) dzds.
2 0 Rn ts
A similar dierentiation of (5.2.7) yields

1
e|z| f (x, t) dz
2
ut (x, t) = n
2 Rn
 t 
1
n
zi
e|z|
2
+ n fxi (x + 2z t s, s) dzds.
2 0 Rn i=1 ts
In view of the boundedness of x f , we conclude that ut and uxi xj are
continuous in (x, t) Rn (0, ). We note that the rst term in the right-
hand side of ut (x, t) is simply f (x, t). Hence,

n
ut (x, t) u(x, t) = ut (x, t) uxi xi (x, t) = f (x, t),
i=1

for any (x, t) Rn (0, ).


If f has bounded x-derivatives of arbitrary order in Rn (0, ), by
(5.2.7) we conclude that x-derivatives of u of arbitrary order exist and are
continuous in Rn (0, ). By the equation ut = u + f , we then conclude
that ut and all its x-derivatives exist and are continuous in Rn (0, ).
Next,
utt = ut + ft = x (x u + f ) + ft .
Hence utt and all its x-derivatives exist and are continuous in Rn (0, ).
Continuing this process, all derivatives of u exist and are continuous in
Rn (0, ). 

By combining Theorem 5.2.3 and Theorem 5.2.11, we conclude that,


under the assumptions on u0 and f as above, the function u given by

u(x, t) = K(x y, t)u0 (y) dy
Rn
 t
+ K(x y, t s)f (y, s) dyds
0 Rn
is a solution of
ut u = f in Rn (0, ),
u(, 0) = u0 on Rn .

Theorem 5.2.11 is optimal in the C -category in the sense that the


smoothness of f implies the smoothness of u. However, it is not optimal
5.3. The Maximum Principle 175

concerning nite dierentiability. In the equation ut u = f , f is re-


lated to the second x-derivatives and the rst t-derivative of u. Theorem
5.2.11 asserts that the continuity of f and its rst x-derivatives implies the
continuity of 2x u and ut . It is natural to ask whether the continuity of
f itself is sucient. This question has a negative answer, and an example
can be constructed by modifying Example 4.4.4. Hence, spaces of functions
with continuous derivatives are not adequate for optimal regularity. What
is needed is the Holder spaces adapted to the heat equation, referred to as
the parabolic Holder spaces. The study of the nonhomogeneous heat equa-
tion, or more generally, nonhomogeneous parabolic dierential equations, in
parabolic Holder spaces is known as the parabolic version of the Schauder
theory. It is beyond the scope of this book to give a presentation of the
Schauder theory. Refer to Subsection 4.4.1 for discussions of the Poisson
equation.

5.3. The Maximum Principle


In this section, we discuss the maximum principle for a class of parabolic
dierential equations slightly more general than the heat equation. As ap-
plications of the maximum principle, we derive a priori estimates for mixed
problems and initial-value problems, interior gradient estimates and the Har-
nack inequality.

5.3.1. The Weak Maximum Principle. Let D be a domain in Rn R.


The parabolic boundary p D of D consists of points (x0 , t0 ) D such that
Br (x0 ) (t0 r2 , t0 ] contains points not in D, for any r > 0. We denote by
C 2,1 (D) the collection of functions in D which are C 2 in x and C 1 in t.
We often discuss the heat equation or general parabolic equations in
cylinders of the following form. Suppose Rn is a bounded domain. For
any T > 0, set
T = (0, T ] = {(x, t) : x , 0 < t T }.
We note that T includes the top portion of its geometric boundary. The
parabolic boundary p T of T is given by

p T = ( {t = 0}) ( (0, T ]) {0} .
In other words, parabolic boundary consists of the bottom, the side and the
bottom corner of the geometric boundary.
For simplicity of presentation, we will prove the weak maximum principle
only in domains of the form T . We should point out that the results in
this subsection hold for general domains in Rn R.
We rst prove the weak maximum principle for the heat equation, which
asserts that any subsolution of the heat equation attains its maximum on
176 5. Heat Equations

the parabolic boundary. Here, a C 2,1 (T )-function u is a subsolution of the


heat equation if ut u 0 in T .
Theorem 5.3.1. Suppose u C 2,1 (T ) C(T ) satises
ut u 0 in T .
Then u attains on p T its maximum in T , i.e.,
max u = max u.
T p T

Proof. We rst consider a special case where ut u < 0 and prove that
u cannot attain in T its maximum in T . Suppose, to the contrary, that
there exists a point P0 = (x0 , t0 ) T such that
u(P0 ) = max u.
T

Then x u(P0 ) = 0 and the Hessian matrix 2x u(P0 ) is nonpositive denite.


Moreover, ut (P0 ) = 0 if t0 (0, T ), and ut (P0 ) 0 if t0 = T . Hence
ut u 0 at P0 , which is a contradiction.
We now consider the general case. For any > 0, let
u (x, t) = u(x, t) t.
Then
(t )u = ut u  < 0.
By the special case we just discussed, u cannot attain in T its maximum.
Hence
max u = max u .
T p T

Then
max u(x, t) = max(u (x, t) + t)) max u (x, t) + T
T T T
= max u (x, t) + T max u(x, t) + T.
p T p T

Letting 0, we obtain the desired result. 

Next, we consider a class of parabolic equations slightly more general


than the heat equation. Let c be a continuous function in T . Consider
Lu = ut u + cu in T .
We prove the following weak maximum principle for subsolutions of L. Here,
a C 2,1 (T )-function u is a subsolution of L if Lu 0 in T . Similarly, a
C 2,1 (T )-function u is a supersolution of L if Lu 0 in T .
5.3. The Maximum Principle 177

Theorem 5.3.2. Let c be a continuous function in T with c 0. Suppose


u C 2,1 (T ) C(T ) satises
ut u + cu 0 in T .
Then u attains on p T its nonnegative maximum in T , i.e.,
max u max u+ .
T p T

We note that u+ is the nonnegative part of u given by u+ = max{0, u}.


The proof of Theorem 5.3.2 is a simple modication of that of Theorem 5.3.1
and is omitted.
Now, we consider a more general case.
Theorem 5.3.3. Let c be a continuous function in T with c c0 for a
nonnegative constant c0 . Suppose u C 2,1 (T ) C(T ) satises
ut u + cu 0 in T ,
u0 on p T .
Then u 0 in T .

Continuous functions in T always have global minima. Therefore, c


c0 in T for some nonnegative constant c0 if c is continuous in T . Such
a condition is introduced to emphasize the role of the minimum of c.

Proof. Let v(x, t) = ec0 t u(x, t). Then u = cc0 t v and



ut u + cu = ec0 t vt v + (c + c0 )v .
Hence
vt v + (c + c0 )v 0.
With c + c0 0, we obtain, by Theorem 5.3.2, that
max v max v + = max ec0 t u+ = 0.
T p T p T

Hence u 0 in T . 

The following result is referred to as the comparison principle.


Corollary 5.3.4. Let c be a continuous function in T with c c0 for a
nonnegative constant c0 . Suppose u, v C 2,1 (T ) C(T ) satisfy
ut u + cu vt v + cv in T ,
uv on p T .
Then u v in T .
178 5. Heat Equations

In the following, we simply say by the maximum principle when we apply


Theorem 5.3.2, Theorem 5.3.3 or Corollary 5.3.4.
Before we discuss applications of maximum principles, we compare max-
imum principles for elliptic equations and parabolic equations. Consider
Le u = u + c(x)u in
and
Lp u = ut u + c(x, t)u in T (0, T ).
We note that the elliptic operator Le here has a form dierent from those
in Section 4.3.1, where we used the form + c. Hence, we should change
the assumption on the sign of c accordingly. If c 0, then
Le u 0 u attains its nonnegative maximum on ,
Lp u 0 u attains its nonnegative maximum on p T .
If c 0, the nonnegativity condition can be removed. For c 0, comparison
principles can be stated as follows:
Le u Le v in , u v on u v in ,
Lp u Lp v in T , u v on p T u v in T .
In fact, the comparison principle for parabolic equations holds for c c0 ,
for a nonnegative constant c0 .
In applications, we need to construct auxiliary functions for compar-
isons. Usually, we take |x|2 or e|x| for elliptic equations and Kt + |x|2 for
2

parabolic equations. Sometimes, auxiliary functions are constructed with


the help of the fundamental solutions for the Laplace equation and the heat
equation.

5.3.2. The Strong Maximum Principle. The weak maximum princi-


ple asserts that subsolutions of parabolic equations attain on the parabolic
boundary their nonnegative maximum if the coecient of the zeroth-order
term is nonnegative. In fact, these subsolutions can attain their nonneg-
ative maximum only on the parabolic boundary, unless they are constant
on suitable subsets. This is the strong maximum principle. We shall point
out that the weak maximum principle suces for most applications to the
initial/boundary-value problem with values of the solutions prescribed on
the parabolic boundary of the domain.
We rst prove the following result.
Lemma 5.3.5. Let (x0 , t0 ) be a point in Rn R, R and T be positive
constants and Q be the set dened by
Q = BR (x0 ) (t0 T, t0 ].
5.3. The Maximum Principle 179

Suppose c is a continuous function in Q and u C 2,1 (Q) C(Q) satises


ut u + cu 0 in Q.
If u 0 in Q and
u(x0 , t0 T ) > 0,
then
u(x, t) > 0 for any (x, t) Q.

Lemma 5.3.5 asserts that a nonnegative supersolution, if positive some-


where initially, becomes positive everywhere at all later times. This can be
interpreted as innite-speed propagation.

Proof. Take an arbitrary t (t0 T, t0 ]. We will prove that


u(x, t ) > 0 for any x BR (x0 ).
Without loss of generality, we assume that x0 = 0 and t = 0. We take
> 0 such that t0 T = R2 and set
D = BR (R2 , 0].
By the assumption u(0, R2 ) > 0 and the continuity of u, we can assume
that
u(x, R2 ) m for any x BR ,
for some constants m > 0 and (0, 1). Here, m can be taken as the
(positive) minimum of u(, R2 ) on BR .
Now we set
!
1 2
D0 = (x, t) BR (R , 0] : |x|
2 2
t < R2 D.

It is easy to see that
D0 {t = 0} = BR , D0 {t = R2 } = BR .
Set
1 2
w1 (t) = t + R2 ,

1 2
w2 (x, t) = w1 (t) |x|2 = t + R2 |x|2 ,

and for some to be determined,
w = w1 w22 .
We will consider w1 , w2 and w in D0 .
180 5. Heat Equations

Figure 5.3.1. The domain D0 .

We rst note that 2 R2 w1 R2 and w2 0 in D0 . A straightforward


calculation yields
wt = w11 t w1 w22 + 2w1 w2 t w2

1 (1 2 ) 2 2(1 2 )
= w1 w2 + w 1 w2 ,

and
w = w1 (2w2 w2 + 2|w2 |2 ) = w1 (4nw2 + 8|x|2 ).
Since |x|2 = w1 w2 , we have

w = w1 8w1 (4n + 8)w2 = w11 8w12 (4n + 8)w1 w2 .
Therefore,

(1 2 )
wt w + cw = w11 + cw1 w22

 
2(1 2 )
+ + 4n + 8 w1 w2 8w1 .
2

Hence

(1 2 )
wt w + cw w11 R2 |c| w22

 
2(1 2 )
2
+ 4n + 8 w1 w2 + 8w1 .

The expression in the parentheses is a quadratic form in w1 and w2 with a
positive coecient of w12 . Hence, we can make this quadratic form nonneg-
ative by choosing suciently large, depending only on , , R and sup |c|.
Hence,
wt w + cw 0 in D0 .
5.3. The Maximum Principle 181

Note that the parabolic boundary p D0 consists of two parts 1 and 2


given by
1 = {(x, t) : |x| < R, t = R2 },
!
1 2
2 = (x, t) : |x|
2
t = R , R t 0 .
2 2

For (x, t) 1 , we have t = R2 and |x| < R, and hence
w(x, R) = (2 R2 ) (2 R2 |x|2 )2 (R)2+4 .
Next, on 2 , we have w = 0. In the following, we set
v = m(R)24 w in D0 ,
where m is the minimum of u over 1 dened earlier. Then
vt v + cv 0 in D0 ,
and
vu on p D0 ,
since u m on 1 and u 0 on 2 . In conclusion,
vt v + cv ut u + cu in D0 ,
vu on p D0 .
By the maximum principle, we have
vu in D0 .
This holds in particular at t = 0. By evaluating v at t = 0, we obtain
2
|x|2
u(x, 0) m 24
1 2 for any x BR .
R
This implies the desired result. 

We point out that the nal estimate in the proof yields a lower bound
of u over BR {0} in terms of the lower bound of u over BR {R2 }.
This is an important estimate.
Now, we are ready to prove the strong maximum principle.
Theorem 5.3.6. Let be a bounded domain in Rn and T be a positive
constant. Suppose c is a continuous function in (0, T ] with c 0, and
u C 2,1 ( (0, T ]) satises
ut u + cu 0 in (0, T ].
If for some (x , t ) (0, T ],
u(x , t ) = sup u 0,
(0,T ]
182 5. Heat Equations

then
u(x, t) = u(x , t ) for any (x, t) (0, t ).

Proof. Set
M = sup u 0,
(0,T ]
and
v =M u in (0, T ].
Then v(x , t ) = 0, v 0 in (0, T ] and
vt v + cv 0 in (0, T ].
We will prove that v(x0 , t0 ) = 0 for any (x0 , t0 ) (0, t ).
To this end, we connect (x0 , t0 ) and (x , t ) by a smooth curve
(0, T ] along which the t-component is increasing. In fact, we rst connect
x0 and x by a smooth curve 0 = 0 (s) , for s [0, 1], with 0 (0) = x0
and 0 (1) = x . Then we may take to be the curve given by (0 (s), st +
(1 s)t0 ). With such a , there exist a positive constant R and nitely

(x , t )

(x0 , t0 )

Figure 5.3.2. and the corresponding covering.

many points (xk , tk ) on , for k = 1, , N , with (xN , tN ) = (x , t ), such


that
1
N&
BR (xk ) [tk , tk + R2 ] (0, T ].
k=0
We may require that tk = tk1 + R2 for k = 0, , N 1.
If v(x0 , t0 ) > 0, then, applying Lemma 5.3.5 in BR (x0 ) [t0 , t0 + R2 ],
we conclude that
v(x, t) > 0 in BR (x0 ) (t0 , t0 + R2 ],
and in particular, v(x1 , t1 ) > 0. We may continue this process nitely many
times to obtain v(x , t ) = v(xN , tN ) > 0. This contradicts the assumption.
Therefore, v(x0 , t0 ) = 0 and hence u(x0 , t0 ) = M . 
5.3. The Maximum Principle 183

Related to the strong maximum principle is the following Hopf lemma


in the parabolic version.
Lemma 5.3.7. Let (x0 , t0 ) be a point in Rn R, R and be two positive
constants and D be the set dened by
D = {(x, t) Rn R : |x x0 |2 + (t0 t) < R2 , t t0 }.
Suppose c is a continuous function in D with c 0, and u C 2,1 (D)C(D)
satises
ut u + cu 0 in D.
Assume, in addition, for some x Rn with |x x0 | = R, that
u(x, t) u(x, t0 )for any (x, t) D and u(x, t0 ) 0,
1
u(x, t) < u(x, t0 ) for any (x, t) D with |x x0 | R.
2
If x u is continuous up to (x, t0 ), then
x u(x, t0 ) > 0,
where is the unit vector given by = (x x0 )/|x x0 |.

Proof. Without loss of generality, we assume that (x0 , t0 ) = (0, 0). Then
D = {(x, t) Rn R : |x|2 t < R2 , t 0}.
By the continuity of u up to D, we have
u(x, t) u(x, 0) for any (x, t) D.
For positive constants and to be determined, we set
w(x, t) = e(|x|
2 t)
eR
2

and
v(x, t) = u(x, t) u(x, 0) + w(x, t).
We consider w and v in
!
1
D0 = (x, t) D : |x| > R .
2

A direct calculation yields


2 2
wt w + cw = e(|x|
2 t)
4 |x| 2n c ceR
2


e(|x| t) 42 |x|2 2n c ,
2

where we used c 0 in D. By taking into account that R/2 |x| R in


D0 and choosing suciently large, we have
42 |x|2 2n c > 0 in D0 ,
184 5. Heat Equations

(x, 0)

Figure 5.3.3. The domain D0 .

and hence
wt w + cw 0 in D0 .
Since c 0 and u(x, 0) 0, we obtain for any > 0 that
vt v + cv = ut u + cu + (wt w + cw) cu(x, 0) 0 in D0 .
The parabolic boundary p D0 consists of two parts 1 and 2 given by
!
1
1 = (x, t) : |x|2 t < R2 , t 0, |x| = R ,
2
!
1
2 = (x, t) : |x| t = R , t 0, |x| R .
2 2
2
First, on 1 , we have u u(x, 0) < 0, and hence u u(x, 0) < for some
> 0. Note that w 1 on 1 . Then for such an , we obtain v < 0 on 1 .
Second, for (x, t) 2 , we have w(x, t) = 0 and u(x, t) u(x, 0). Hence
v(x, t) 0 for any (x, t) 2 and v(x, 0) = 0. Therefore, v 0 on 2 . In
conclusion,
vt v + cv 0 in D0 ,
v 0 on p D0 .
By the maximum principle, we have
v 0 in D0 .
Then, by v(x, 0) = 0, v attains at (x, 0) its maximum in D0 . In particular,
v(x, 0) v(x, 0) for any x BR \ B 1 R .
2

Hence, we obtain
v
(x, 0) 0,

and then
u w
(x, 0) = 2ReR > 0.
2
(x, 0)

This is the desired result. 
5.3. The Maximum Principle 185

To conclude our discussion of the strong maximum principle, we briey


compare our approaches for elliptic equations and parabolic equations. For
elliptic equations, we rst prove the Hopf lemma and then prove the strong
maximum principle as its consequence. See Subsection 4.3.2 for details. For
parabolic equations, we rst prove innite speed of propagation and then
obtain the strong maximum principle as a consequence. It is natural to ask
whether we can prove the strong maximum principle by Lemma 5.3.7, the
parabolic Hopf lemma. By an argument similar to the proof of Theorem
4.3.9, we can conclude that, if a subsolution u attains its nonnegative maxi-
mum at an interior point (x0 , t0 ) (0, T ], then u is constant on {t0 }.
In order to conclude that u is constant in (0, t0 ) as asserted by Theorem
5.3.6, we need a result concerning the t-derivative at the interior maximum
point, similar to that concerning the x-derivative in the Hopf lemma. We
will not pursue this issue in this book.

5.3.3. A Priori Estimates. In the rest of this section, we discuss applica-


tions of the maximum principle. We point out that only the weak maximum
principle is needed.
As the rst application, we derive an estimate of the sup-norms of so-
lutions of initial/boundary-value problems with Dirichlet boundary values.
Compare this with the estimate in integral norms in Theorem 3.2.4.
As before, for a bounded domain Rn and a positive constant T , we
set
T = (0, T ] = {(x, t) : x , 0 < t T }.

Theorem 5.3.8. Let c be continuous in T with c c0 for a nonnegative


constant c0 . Suppose u C 2,1 (T ) C(T ) is a solution of

ut u + cu = f in T ,
u(, 0) = u0 on ,
u= on (0, T ),

for some f C(T ), u0 C() and C( [0, T ]). Then


  ' 
sup |u| ec0 T max sup |u0 |, sup || + T sup |f | .
T (0,T ) T

Proof. Set Lu = ut u + cu and


 '
B = max sup |u0 |, sup || , F = sup |f |.
(0,T ) T
186 5. Heat Equations

Then
L(u) F in T ,
u B on p T .
Set
v(x, t) = ec0 t (B + F t).
Since c + c0 0 and ec0 t 1 in T , we have
Lv = (c0 + c)ec0 t (B + F t) + ec0 t F F in T
and
vB on p T .
Hence,
L(u) Lv in T ,
u v on p T .
By the maximum principle, we obtain
u v in T .
Therefore,
|u(x, t)| ec0 t (B + F t) for any (x, t) T .
This implies the desired estimate. 

Next, we derive a priori estimates of solutions of initial-value problems.


Theorem 5.3.9. Let c be continuous in Rn (0, T ] with c c0 for a
nonnegative constant c0 . Suppose u C 2,1 (Rn (0, T ]) C(Rn [0, T ]) is
a bounded solution of
ut u + cu = f in Rn (0, T ],
u(, 0) = u0 on Rn ,
for some bounded f C(Rn (0, T ]) and u0 C(Rn ). Then
 
sup |u| ec0 T sup |u0 | + T sup |f | .
Rn (0,T ) Rn Rn (0,T )

We note that the maximum principle is established in bounded domains


such as (0, T ]. In studying solutions of the initial-value problem where
solutions are dened in Rn (0, T ], we should rst derive suitable estimates
of solutions in BR (0, T ] and then let R . For this purpose, we need to
impose extra assumptions on u as x . For example, u is assumed to be
bounded in Theorem 5.3.9 and to be of the exponential growth in Theorem
5.3.10.
5.3. The Maximum Principle 187

Proof. Set Lu = ut u + cu and


F = sup |f |, B = sup |u0 |.
Rn (0,T ] Rn

Then
L(u) F in Rn (0, T ],
u B on Rn .
Since u is bounded, we assume that |u| M in Rn (0, T ] for a positive
constant M . For any R > 0, consider
w(x, t) = ec0 t (B + F t) + vR (x, t) in BR (0, T ],
where vR is a function to be chosen. By c + c0 0 and ec0 t 1, we have
Lw = (c + c0 )ec0 t (B + F t) + ec0 t F + LvR F + LvR in BR (0, T ].
Moreover,
w(, 0) = B + vR (, 0) in BR ,
and
w vR on BR (0, T ].
We will choose vR such that
LvR 0 in BR (0, T ],
vR (, 0) 0 in BR ,
vR u on BR [0, T ].
To construct such a vR , we consider
M
vR (x, t) = 2 ec0 t (2nt + |x|2 ).
R
Obviously, vR 0 for t = 0 and vR M on |x| = R. Next,
M c0 t
LvR = e (c + c0 )(2nt + |x|2 ) 0 in BR (0, T ].
R2
With such a vR , we have
L(u) Lw in BR (0, T ],
u w on p (BR (0, T ]).
Then the maximum principle yields u w in BR (0, T ]. Hence for any
(x, t) BR (0, T ],
M c0 t
|u(x, t)| ec0 t (B + F t) + e (2nt + |x|2 ).
R2
Now we x an arbitrary (x, t) Rn (0, T ]. By choosing R > |x| and then
letting R , we have
|u(x, t)| ec0 t (B + F t).
188 5. Heat Equations

This yields the desired estimate. 

Next, we prove the uniqueness of solutions of initial-value problems for


the heat equation under the assumption of exponential growth.
Theorem 5.3.10. Let u C 2,1 (Rn (0, T ]) C(Rn [0, T ]) satisfy
ut u = 0 in Rn (0, T ],
u(, 0) = 0 on Rn .
Suppose, for some positive constants M and A,
2
|u(x, t)| M eA|x| ,
for any (x, t) Rn (0, T ]. Then u 0 in Rn [0, T ].

Proof. For any constant > A, we prove that


# $
1
u = 0 in R 0,
n
.
4
1 2 2 3
We then extend u = 0 in the t-direction successively to [ 4 , 4 ], [ 4 , 4 ],
, until t = T .
For any constant R > 0, consider
2
M e(A)R 14t
|x|2
vR (x, t) = n e ,
(1 4t) 2
for any (x, t) BR (0, 1/4). We note that vR is modied from the
example we discussed preceding Theorem 5.2.5. Then

1
t vR vR = 0 in BR 0, .
4
Obviously,
vR (, 0) 0 = u(, 0) in BR .

Next, for any (x, t) BR 0, 1/4),
2 2 2
vR (x, t) M e(A)R eR = M eAR u(x, t).
In conclusion, 
1
u vR on p BR 0, .
4
By the maximum principle, we have

1
u vR in BR 0, .
4
Therefore,

1
|u(x, t)| vR (x, t) for any (x, t) BR 0, .
4
5.3. The Maximum Principle 189

Now we x an arbitrary (x, t) Rn (0, 1/4) and then choose R > |x|.
We note that vR (x, t) 0 as R , since > A. We therefore obtain
u(x, t) = 0. 

5.3.4. Interior Gradient Estimates. We now give an alternative proof,


based on the maximum principle, of the interior gradient estimate. We do
this only for solutions of the heat equation. Recall that for any r > 0,
Qr = Br (r2 , 0].
Theorem 5.3.11. Suppose u C 2,1 (Q1 ) C(Q1 ) satises
ut u = 0 in Q1 .
Then
sup |x u| C sup |u|,
Q1 p Q1
2

where C is a positive constant depending only on n.

The proof is similar to that of Theorem 4.3.13, the interior gradient


estimate for harmonic functions.

Proof. We rst note that u is smooth in Q1 by Theorem 5.2.7. A straight-


forward calculation yields
n 
n
(t )|x u|2 = 2 u2xi xj + 2 uxi (ut u)xi
i,j=1 i=1
 n
= 2 u2xi xj .
i,j=1

To get interior estimates, we need to introduce a cuto function. For any


smooth function in C (Q1 ) with supp Q3/4 , we have
(t )(|x u|2 ) = (t )|x u|2

n 
n
4 xi uxj uxi xj 2 u2xi xj .
i,j=1 i,j=1

Now we take = 2 for some C (Q1 ) with 1 in Q1/2 and supp


Q3/4 . Then
(t )( 2 |x u|2 ) = (2t 2 2|x |2 )|x u|2

n 
n
8 xi uxj uxi xj 2 2
u2xi xj .
i,j=1 i,j=1

By the Cauchy inequality, we obtain


8|xi uxj uxi xj | 8x2i u2xj + 2 2 u2xi xj .
190 5. Heat Equations

Hence,

(t )( |x u| ) 2t 2 + 6|x | |x u|2
2 2 2

C|x u|2 ,
where C is a positive constant depending only on and n. Note that
(t )(u2 ) = 2|x u|2 + 2u(ut u) = 2|x u|2 .
By taking a constant large enough, we get
(t )( 2 |x u|2 + u2 ) (C 2)|x u|2 0.
By the maximum principle, we have
sup( 2 |x u|2 + u2 ) sup ( 2 |x u|2 + u2 ).
Q1 p Q1

This implies the desired result since = 0 on p Q1 and = 1 in Q1/2 . 

5.3.5. Harnack Inequalities. For positive harmonic functions, the Har-


nack inequality asserts that their values in compact subsets are comparable.
In this section, we study the Harnack inequality for positive solutions of the
heat equation. In seeking a proper form of the Harnack inequality for so-
lutions of the heat equation, we begin our discussion with the fundamental
solution.
We x an arbitrary Rn and consider for any (x, t) Rn (0, ),
1 |x|2
4t
u(x, t) = n e .
(4t) 2
Then u satises the heat equation ut u = 0 in Rn (0, ). For any
(x1 , t1 ) and (x2 , t2 ) Rn (0, ),
 n |x |2 |x |2
u(x1 , t1 ) t2 2 2 14t
= e 4t2 1 .
u(x2 , t2 ) t1
Recall that
(p + q)2 p2 q 2
+ ,
a+b a b
for any a, b > 0 and any p, q R, and the equality holds if and only if
bp = aq. This implies, for any t2 > t1 > 0,
|x2 |2 |x2 x1 |2 |x1 |2
+ ,
t2 t2 t1 t1
and the equality holds if and only if
t 2 x1 t 1 x2
= .
t2 t1
5.3. The Maximum Principle 191

Therefore,
n 2
t2 2 |x 2 x1 |
u(x1 , t1 ) e 4(t2 t1 )
u(x2 , t2 ),
t1
for any x1 , x2 Rn and any t2 > t1 > 0, and the equality holds if is chosen
as above. This simple calculation suggests that the Harnack inequality for
the heat equation has an evolution feature: the value of a positive solution
at a certain time is controlled from above by the value at a later time. Hence,
if we attempt to establish the estimate
u(x1 , t1 ) Cu(x2 , t2 ),
the constant C should depend on t2 /t1 , |x2 x1 |, and most importantly
(t2 t1 )1 (> 0).
Suppose u is a positive solution of the heat equation and set v = log u.
In order to derive an estimate for the quotient
u(x1 , t1 )
,
u(x2 , t2 )
it suces to get an estimate for the dierence
v(x1 , t1 ) v(x2 , t2 ).
To this end, we need an estimate of vt and |v|. For a hint of proper
forms, we again turn our attention to the fundamental solution of the heat
equation.
Consider for any (x, t) Rn (0, ),
1 |x|2
4t
u(x, t) = n e .
(4t) 2
Then
n |x|2
v(x, t) = log u(x, t) = log(4t) ,
2 4t
and hence
n |x|2 x
vt = + 2 , v = .
2t 4t 2t
Therefore,
n
vt = + |v|2 .
2t
We have the following dierential Harnack inequality for arbitrary pos-
itive solutions of the heat equation.
Theorem 5.3.12. Suppose u C 2,1 (Rn (0, T ]) satises
ut = u, u>0 in Rn (0, T ].
Then v = log u satises
n
vt + |v|2 in Rn (0, T ].
2t
192 5. Heat Equations

The dierential Harnack inequality implies the Harnack inequality by a


simple integration.
Corollary 5.3.13. Suppose u C 2,1 (Rn (0, T ]) satises
ut = u, u>0 in Rn (0, T ].
Then for any (x1 , t1 ), (x2 , t2 ) Rn (0, T ] with t2 > t1 > 0,
n !
u(x1 , t1 ) t2 2 |x2 x1 |2
exp .
u(x2 , t2 ) t1 4(t2 t1 )
Proof. Let v = log u be as in Theorem 5.3.12 and take an arbitrary path
x = x(t) for t [t1 , t2 ] with x(ti ) = xi , i = 1, 2. By Theorem 5.3.12, we
have
d dx dx n
v(x(t), t) = vt + v |v|2 + v .
dt dt dt 2t
By completing the square, we obtain
 
d 1  dx 2 n
v(x(t), t)   .
dt 4 dt 2t
Then a simple integration yields
  2
n t2 1 t2  dx 
v(x1 , t1 ) v(x2 , t2 ) + log +   dt.
2 t1 4  dt 
t1
To seek an optimal path which makes the last integral minimal, we require
d2 x
=0
dt2
along the path. Hence we set, for some a, b Rn ,
x(t) = at + b.
Since xi = ati + b, i = 1, 2, we take
x2 x1 t 2 x1 t 1 x2
a= , b= .
t2 t1 t2 t1
Then,
 t2  2
 dx 
  dt = |x2 x1 | .
2
 dt  t2 t1
t1
Therefore, we obtain
n t2 1 |x2 x1 |2
v(x1 , t1 ) v(x2 , t2 ) + log + ,
2 t1 4 t2 t1
or n !
u(x1 , t1 ) t2 2 |x2 x1 |2
exp .
u(x2 , t2 ) t1 4(t2 t1 )
This is the desired estimate. 
5.3. The Maximum Principle 193

Now we begin to prove the dierential Harnack inequality. The basic


idea is to apply the maximum principle to an appropriate combination of
derivatives of v. In our case, we consider |v|2 vt and intend to derive an
upper bound. First, we derive a parabolic equation satised by |v|2 vt . A
careful analysis shows that some terms in this equation cannot be controlled.
So we introduce a parameter (0, 1) and consider |v|2 vt instead.
After we apply the maximum principle, we let 1. The proof below is
probably among the most dicult ones in this book.

Proof of Theorem 5.3.12. Without loss of generality, we assume that u


is continuous up to {t = 0}. Otherwise, we consider u in Rn [, T ] for any
constant (0, T ) and then let 0. We divide the proof into several
steps. In the following, we avoid notions of summations if possible.
Step 1. We rst derive some equations involving derivatives of v = log u.
A simple calculation yields
vt = v + |v|2 .
Consider w = v. Then
wt = vt = (v + |v|2 ) = w + |v|2 .
Since
|v|2 = 2|2 v|2 + 2v (v) = 2|2 v|2 + 2v w,
we have
(5.3.1) wt w 2v w = 2|2 v|2 .
Note that v is to be controlled and appears as a coecient in the equation
(5.3.1). So it is convenient to derive an equation for v. Set w = |v|2 .
Then,
wt = 2v vt = 2v (v + |v|2 )
= 2v (v) + 2v w
= |v|2 2|2 v|2 + 2v w
= w + 2v w 2|2 v|2 .
Therefore,
(5.3.2) wt w 2v w = 2|2 v|2 .
Note that, by the Cauchy inequality,
 2

n 
n
1 
n
1
|2 v|2 = vx2i xj vx2i xi vxi xi = (v)2 .
n n
i,j=1 i=1 i=1
194 5. Heat Equations

Hence, (5.3.1) implies

2 2
wt w 2v w w .
n

Step 2. For a constant (0, 1), set

f = |v|2 vt .

Then

f = |v|2 v |v|2 = v (1 )|v|2


= w (1 )w,

and hence by (5.3.1) and (5.3.2),

ft f 2v f = 2|2 v|2 .

Next, we estimate |2 v|2 by f . Note that

1 1 2 1 2
|2 v|2 (v)2 = |v|2 vt = (1 )|v|2 + f
n n n
1 2
= f + 2(1 )|v| f + (1 )2 |v|4
2
n
1 2
f + 2(1 )|v|2 f .
n
We obtain
2 2
(5.3.3) ft f 2v f f + 2(1 )|v|2 f .
n

We should point out that |v|2 in the right-hand side plays an important
role later on.
Step 3. Now we introduce a cuto function C0 (Rn ) with 0
and set
g = tf.

We derive a dierential inequality for g. Note that

gt = f + tft ,
g = tf + tf ,
g = tf + 2t f + tf .
5.3. The Maximum Principle 195

Then,
g
tft = gt ,
t

tf = g g,



tf = g 2 g g g


||2
= g 2 g + 2 2 g.

Multiplying (5.3.3) by t2 2 and substituting ft , f and f by above equal-
ities, we obtain
t(gt g) + 2t( v) g
!
2 ||2 4(1 )
g g+t 2 |v| 2 v
2
.
n n
To eliminate |v| from the right-hand side, we complete the square for the
last two terms. (Here we need < 1! Otherwise, we cannot control the
expression 2 v in the right-hand side.) Hence,
t(gt g) + 2t( v) g
!
2 ||2 n ||2
g g+t 2 + ,
n 4(1 )
whenever g is nonnegative. We point out that there are no unknown ex-
pressions in the right-hand side except g. By choosing = 2 for some
C0 (Rn ) with 0, we get
t 2 (gt g) + 2t(2 2 v) g
!
2 n
g 2
g + t 6|| 2 +
2
|| 2
,
n (1 )
whenever g is nonnegative. Now we x a cuto function 0 C0 (B1 ), with
0 0 1 in B1 and 0 = 1 in B1/2 . For any xed R 1, we consider
(x) = 0 (x/R). Then

n
6||2 2 + ||2 (x)
(1 )
 
1 n x
= 2 6|0 | 20 0 +
2
|0 | 2
.
R (1 ) R
Therefore, we obtain that in BR (0, T ),

2 C t
t (gt g) + 2t(2 v) g g 1
2 2
g+ 2 ,
n R
196 5. Heat Equations

whenever g is nonnegative. Here, C is a positive constant depending only


on and 0 . We point out that the unknown expression v in the left-hand
side appears as a coecient of g and is unharmful.
Step 4. We claim that
2 C t
(5.3.4) 1 g+ 2 0 in BR (0, T ].
n R
Note that g vanishes on the parabolic boundary of BR (0, T ) since g = t 2 f .
Suppose, to the contrary, that
2 C t
h1 g+ 2
n R
has a negative minimum at (x0 , t0 ) BR (0, T ]. Hence,
h(x0 , t0 ) < 0,
and
ht 0, h = 0, h 0 at (x0 , t0 ).
Thus,
g(x0 , t0 ) > 0,
and
gt 0, g = 0, g 0 at (x0 , t0 ).
Then at (x0 , t0 ), we get
0 t 2 (gt g) + 2t(2 2 v) g

2 C t
g 1 g + 2 < 0.
n R
This is a contradiction. Hence (5.3.4) holds in BR (0, T ).
Therefore, we obtain
2 2 C t
(5.3.5) 1 t (|v|2 vt ) + 2 0 in BR (0, T ].
n R
For any xed (x, t) Rn (0, T ], choose R > |x|. Recall that = 0 (/R)
and 0 = 1 in B1/2 . Letting R , we obtain
2
1 t(|v|2 vt ) 0.
n
We then let 1 and get the desired estimate. 

We also have the following dierential Harnack inequality for positive


solutions in nite regions.
5.4. Exercises 197

Theorem 5.3.14. Suppose u C 2,1 (B1 (0, 1]) satises


ut u = 0, u > 0 in B1 (0, 1].
Then for any (0, 1), v = log u satises
n
vt |v|2 + + C 0 in B1/2 (0, 1],
2t
where C is a positive constant depending only on n and .

Proof. We simply take R = 1 in (5.3.5). 

Now we state the Harnack inequality in nite regions.


Corollary 5.3.15. Suppose u C 2,1 (B1 (0, 1]) satises
ut u = 0, u 0 in B1 (0, 1].
Then for any (x1 , t1 ), (x2 , t2 ) B1/2 (0, 1] with t2 > t1 ,
u(x1 , t1 ) Cu(x2 , t2 ),
where C is a positive constant depending only on n, t2 /t1 and (t2 t1 )1 .

The proof is left as an exercise.


We point out that u is assumed to be positive in Theorem 5.3.14 and
only nonnegative in Corollary 5.3.15.
The Harnack inequality implies the following form of the strong maxi-
mum principle: Let u be a nonnegative solution of the heat equation ut
u = 0 in B1 (0, 1]. If u(x0 , t0 ) = 0 for some (x0 , t0 ) B1 (0, 1], then
u = 0 in B1 (0, t0 ]. This may be interpreted as innite-speed propagation.

5.4. Exercises
Exercise 5.1. Prove the following statements by straightforward calcula-
tions:
|x|2
(1) K(x, t) = t 2 e
n
4t satises the heat equation for t > 0.
|x|2
(2) For any > 0, G(x, t) = (1 4t) 2 e 14t satises the heat equa-
n

tion for t < 1/4.


Exercise 5.2. Let u0 be a continuous function in Rn and u be dened in
(5.2.4). Suppose u0 (x) 0 uniformly as x . Prove
lim u(x, t) = 0 uniformly in x.
t

Exercise 5.3. Prove the convergence in Theorem 5.2.5.


198 5. Heat Equations

Exercise 5.4. Let u0 be a bounded and continuous function in [0, ) with


u0 (0) = 0. Find an integral representation for the solution of the problem

ut uxx = 0 for x > 0, t > 0,


u(x, 0) = u0 (x) for x > 0,
u(0, t) = 0 for t > 0.

Exercise 5.5. Let u C 2,1 (Rn (, 0)) be a solution of

ut u = 0 in Rn (, 0).

Suppose that for some nonnegative integer m,



|u(x, t)| C(1 + |x| + |t|)m ,

for any (x, t) Rn (, 0). Prove that u is a polynomial of degree at


most m.

Exercise 5.6. Prove that u constructed in the proof of Proposition 5.2.6 is


smooth in R R.

Exercise 5.7. Let be a bounded domain in Rn and u0 C(). Suppose


u C 2,1 ( (0, )) C( [0, )) is a solution of

ut u = 0 in (0, ),
u(, 0) = u0 on ,
u = 0 on (0, ).

Prove that
sup |u(, t)| Cet sup |u0 | for any t > 0,

where and C are positive constants depending only on n and .

Exercise 5.8. Let be a bounded domain in Rn , c be continuous in


[0, T ] with c c0 for a nonnegative constant c0 , and u0 be continuous in
with u0 0. Suppose u C 2,1 ( (0, T ]) C( [0, T ]) is a solution of

ut u + cu = u2 in (0, T ],
u(, 0) = u0 on ,
u=0 on (0, T ).

Prove that
0 u ec0 T sup u0 in (0, T ].

5.4. Exercises 199

Exercise 5.9. Let be a bounded domain in Rn , u0 and f be continuous


in , and be continuous on [0, T ]. Suppose u C 2,1 ( (0, T ])
C( [0, T ]) is a solution of
ut u = eu f (x) in (0, T ],
u(, 0) = u0 on ,
u = on (0, T ).
Prove that
M u T eM + M in (0, T ],
where  '
M = T sup |f | + sup sup |u0 |, sup || .
(0,T )

Exercise 5.10. Let Q = (0, l) (0, ) and u0 C 1 [0, l] with u0 (0) =


u0 (l) = 0. Suppose u C 3,1 (Q) C 1 (Q) is a solution of
ut uxx = 0 in Q,
u(, 0) = u0 on (0, l),
u(0, ) = u(l, ) = 0 on (0, ).
Prove that
sup |ux | sup |u0 |.
Q [0,l]

Exercise 5.11. Let be a bounded domain in Rn . Suppose u1 , , um


C 2,1 ( (0, T ]) C( [0, T ]) satisfy
t ui = ui in (0, T ],
for i = 1, , m. Assume that f is a convex function in Rm . Prove that
sup f (u1 , , um ) sup f (u1 , , um ).
(0,T ] p ((0,T ])

Exercise 5.12. Let u0 be a bounded continuous function in Rn . Suppose


u C 2,1 (Rn (0, T ]) C(Rn [0, T ]) satises
ut u = 0 in Rn (0, T ],
u(, 0) = u0 on Rn .
Assume that u and u are bounded in Rn (0, T ]. Prove that
1
sup |u(, t)| sup |u0 |,
Rn 2t Rn
for any t (0, T ].
Hint: With |u0 | M in Rn , consider
w = u2 + 2t|u|2 M 2 .
200 5. Heat Equations

Exercise 5.13. Prove Corollary 5.3.15.


Chapter 6

Wave Equations

The n-dimensional wave equation is given by utt u = 0 for functions


u = u(x, t), with x Rn and t R. Here, x is the space variable and
t the time variable. The wave equation represents vibrations of strings or
propagation of sound waves in tubes for n = 1, waves on the surface of
shallow water for n = 2, and acoustic or light waves for n = 3.
In Section 6.1, we discuss the initial-value problem and mixed problems
for the one-dimensional wave equation. We derive explicit expressions for
solutions of these problems by various methods and study properties of these
solutions. We illustrate that characteristic curves play an important role in
studying the one-dimensional wave equation. They determine the domain
of dependence and the range of inuence.
In Section 6.2, we study the initial-value problem for the wave equation
in higher-dimensional spaces. We derive explicit expressions for solutions in
odd dimensions by the method of spherical averages and in even dimensions
by the method of descent. We study properties of these solutions with
the help of these formulas and illustrate the importance of characteristic
cones for the higher-dimensional wave equation. Among applications of
these explicit expressions, we discuss global behaviors of solutions and prove
that solutions decay at certain rates as time goes to innity. We will also
solve the initial-value problem for the nonhomogeneous wave equation by
Duhamels principle.
In Section 6.3, we discuss energy estimates for solutions of the initial-
value problem for a class of hyperbolic equations slightly more general than
the wave equation. We introduce the important concept of space-like and
time-like hypersurfaces. We demonstrate that initial-value problems for hy-
perbolic equations with initial values prescribed on space-like hypersurfaces

201
202 6. Wave Equations

are well posed. We point out that energy estimates are fundamental and
form the basis for the existence of solutions of general hyperbolic equations.

6.1. One-Dimensional Wave Equations


In this section, we discuss initial-value problems and initial/boundary-value
problems for the one-dimensional wave equation. We rst study initial-value
problems.

6.1.1. Initial-Value Problems. For f C(R (0, )), C 2 (R) and


C 1 (R), we seek a solution u C 2 (R [0, )) of the problem
utt uxx = f in R (0, ),
(6.1.1)
u(, 0) = , ut (, 0) = on R.
We will derive expressions for its solutions by several dierent methods.
Throughout this section, we denote points in R (0, ) by (x, t). How-
ever, when (x, t) is taken as a xed point, we denote arbitrary points by
(y, s).
The characteristic curves for the one-dimensional wave equation are
given by the straight lines s = y + c. (Refer to Section 3.1 for the de-
tail.) In particular, for any (x, t) R (0, ), there are two characteristic
curves through (x, t) given by
sy =tx and s + y = t + x.
These two characteristic curves intercept the x-axis at (xt, 0) and (x+t, 0),
respectively, and form a triangle C1 (x, t) with the x-axis given by
C1 (x, t) = {(y, s) : |y x| < t s, s > 0}.
This is the cone we introduced in Section 2.3 for n = 1. We usually refer to
C1 (x, t) as the characteristic triangle.
We rst consider the homogeneous wave equation
(6.1.2) utt uxx = 0 in R (0, ).
We introduce new coordinates (, ) along characteristic curves by
= x t, = x + t.
In the new coordinates, the wave equation has the form
u = 0.
By a simple integration, we obtain
u(, ) = g() + h(),
for some functions g and h in R. Therefore,
(6.1.3) u(x, t) = g(x t) + h(x + t).
6.1. One-Dimensional Wave Equations 203

This provides a general form for solutions of (6.1.2).


As a consequence of (6.1.3), we derive an important formula for the
solution of the wave equation. Let u be a C 2 -solution of (6.1.2). Consider a
parallelogram bounded by four characteristic curves in R (0, ), which is
referred to as a characteristic parallelogram. (This parallelogram is in fact
a rectangle.) Suppose A, B, C, D are its four vertices. Then

t6

A
@
B @
@ @
@ @C
@
@
D
-
x

Figure 6.1.1. A characteristic parallelogram.

(6.1.4) u(A) + u(D) = u(B) + u(C).

In other words, the sums of the values of u at opposite vertices are equal.
This follows easily from (6.1.3). In fact, if we set A = (xA , tA ), B = (xB , tB ),
C = (xC , tC ) and D = (xD , tD ), we have

xB t B = xA t A , xB + t B = xD + t D ,

and
x C t C = xD t D , xC + t C = xA + t A .
We then get (6.1.4) by (6.1.3) easily. An alternative method to prove (6.1.4)
is to consider it in (, )-coordinates, where A, B, C, D are the vertices of
a rectangle with sides parallel to the axes. Then we simply integrate u ,
which is zero, in this rectangle to get the desired relation.
We now solve (6.1.1) for the case f 0. Let u be a C 2 -solution which
is given by (6.1.3) for some functions g and h. By evaluating u and ut at
t = 0, we have

u(x, 0) = g(x) + h(x) = (x),


ut (x, 0) = g  (x) + h (x) = (x).
204 6. Wave Equations

Then
1 1
g  (x) =  (x) (x),
2 2
1 1
h (x) =  (x) + (x).
2 2
A simple integration yields

1 1 x
g(x) = (x) (s) ds + c,
2 2 0
for a constant c. Then a substitution into the expression of u(x, 0) implies

1 1 x
h(x) = (x) + (s)ds c.
2 2 0
Therefore,

1 1 x+t
(6.1.5) u(x, t) = (x t) + (x + t) + (s) ds.
2 2 xt
This is dAlemberts formula. It clearly shows that regularity of u(, t) for
any t > 0 is the same as that of the initial value u(, 0) and is 1-degree better
than ut (, 0). There is no improvement of regularity.
We see from (6.1.5) that u(x, t) is determined uniquely by the initial
values in the interval [x t, x + t] of the x-axis, which is the base of the
characteristic triangle C1 (x, t). This interval is the domain of dependence
for the solution u at the point (x, t). We note that the endpoints of this
interval are cut out by the characteristic curves through (x, t). Conversely,
the initial values at a point (x0 , 0) of the x-axis inuence u(x, t) at points
(x, t) in the wedge-shaped region bounded by characteristic curves through
(x0 , 0), i.e., for x0 t < x < x0 + t, which is often referred to as the range
of inuence.

t6 t6

@ @
@ @
@ @
@- @ -
x x

Figure 6.1.2. The domain of dependence and the range of inuence.

Next, we consider the case f 0 and 0 and solve (6.1.1) by the


method of characteristics. We write
utt uxx = (t + x )(t x )u.
6.1. One-Dimensional Wave Equations 205

By setting v = ut ux , we decompose (6.1.1) into two initial-value problems


for rst-order PDEs,
u t ux = v in R (0, ),
(6.1.6)
u(, 0) = 0 on R,
and
v t + vx = 0 in R (0, ),
(6.1.7)
v(, 0) = on R.

The initial-value problem (6.1.7) was discussed in Example 2.2.3. Its solution
is given by
v(x, t) = (x t).
The initial-value problem (6.1.6) was discussed in Example 2.2.4. Its solution
is given by
 t
u(x, t) = (x + t 2 ) d.
0
By a change of variables, we obtain
 x+t
1
u(x, t) = (s) ds.
2 xt

This is simply a special case of dAlemberts formula (6.1.5).


Now we derive an expression of solutions in the general case. For any
(x, t) R (0, ), consider the characteristic triangle

C1 (x, t) = {(y, s) : |y x| < t s, s > 0}.

The boundary of C1 (x, t) consists of three parts,

L+ = {(y, s) : s = y + x + t, 0 < s < t},


L = {(y, s) : s = y x + t, 0 < s < t},

and
L0 = {(y, 0) : x t < y < x + t}.
We note that L+ and L are parts of the characteristic curves through (x, t).
Let = (1 , 2 ) be the unit exterior normal vector of C1 (x, t). Then


(1, 1)/ 2 on L+ ,
= (1, 1)/ 2 on L ,


(0, 1) on L0 .
Upon integrating by parts, we have
206 6. Wave Equations

t6
(x, t)
@
@
@
@
x
@ -
xt x+t

Figure 6.1.3. A characteristic triangle.

  
f dyds = (utt uxx ) dyds = (ut 2 ux 1 ) dl
C1 (x,t) C1 (x,t) C1 (x,t)
 
1 1
= (ut ux ) dl + (ut + ux ) dl
L+ 2 L 2
 x+t
ut (s, 0) ds,
xt
where the orientation of the integrals over L+ and L is counterclockwise.
Note that (t x )/ 2 is a directional derivative along L+ with unit length
and with direction matching the orientation of the integral over L+ . Hence

1
(ut ux ) dl = u(x, t) u(x + t, 0).
L+ 2

On the other hand, (t + x )/ 2 is a directional derivative along L with
unit length and with direction opposing the orientation of the integral over
L . Hence

1
(ut + ux ) dl = u(x t, 0) u(x, t) .
L 2
Therefore, a simple substitution yields

1 1 x+t
u(x, t) = (x + t) + (x t) + (s) ds
2 2 xt
(6.1.8)  
1 t x+(t )
+ f (y, ) dyd.
2 0 x(t )

Theorem 6.1.1. Let m 2 be an integer, C m (R), C m1 (R)


and f C m1 (R [0, )). Suppose u is dened by (6.1.8). Then u
C m (R (0, )) and
utt uxx = f in R (0, ).
6.1. One-Dimensional Wave Equations 207

Moreover, for any x0 R,


lim u(x, t) = (x0 ), lim ut (x, t) = (x0 ).
(x,t)(x0 ,0) (x,t)(x0 ,0)

Hence, u dened by (6.1.8) is a solution of (6.1.1). In fact, u is C m in


R [0, ).
The proof is a straightforward calculation and is omitted. Obviously,
C 2 -solutions of (6.1.1) are unique.
Formula (6.1.8) illustrates that the value u(x, t) is determined by f in
the triangle C1 (x, t), by on the interval [x t, x + t] {0} and by at the
two points (x + t, 0) and (x t, 0).
In fact, without using the explicit expression of solutions in (6.1.8), we
can derive energy estimates, the estimates for the L2 -norms of solutions of
(6.1.1) and their derivatives in terms of the L2 -norms of , and f . To
obtain energy estimates, we take any constants 0 < T < t and use the
domain
{(x, t) : |x| < t t, 0 < t < T }.
We postpone the derivation until the nal section of this chapter.

6.1.2. Mixed Problems. In the following, we study mixed problems. For


simplicity, we discuss the wave equation only, with no nonhomogeneous
terms.
First, we study the half-space problem. Let C 2 [0, ), C 1 [0, )
and C 2 [0, ). We consider
utt uxx = 0 in (0, ) (0, ),
(6.1.9) u(, 0) = , ut (, 0) = on [0, ),
u(0, t) = (t) for t > 0.

We will construct a C 2 -solution under appropriate compatibility conditions.


We note that the origin is the corner of the region (0, ) (0, ). In order
to have a C 2 -solution u, the initial values and and the boundary value
have to match at the corner to generate the same u and its rst-order and
second-order derivatives when computed either from and or from . If
(6.1.9) admits a solution which is C 2 in [0, ) [0, ), a simple calculation
shows that
(6.1.10) (0) = (0), (0) =  (0),  (0) =  (0).
This is the compatibility condition for (6.1.9). It is the necessary condition
for the existence of a C 2 -solution of (6.1.9). We will show that it is also
sucient.
208 6. Wave Equations

We rst consider the case 0 and solve (6.1.9) by the method of


reection. In this case, the compatibility condition (6.1.10) has the form
(0) = 0, (0) = 0,  (0) = 0.
Now we assume that this holds and proceed to construct a C 2 -solution of
(6.1.9). We extend and to R by odd reection. In other words, we set

(x) for x 0
(x) =
(x) for x < 0,

(x) for x 0
(x) =
(x) for x < 0.

Then and are C 2 and C 1 in R, respectively. Let u be the unique


C 2 -solution of the initial-value problem
utt uxx = 0 in R (0, ),
u(, 0) = , ut (, 0) = in R.
We now prove that u(x, t) is the solution of (6.1.9) when we restrict x to
[0, ). We need only prove that
u(0, t) = 0 for any t > 0.
In fact, for v(x, t) = u(x, t), a simple calculation yields
vtt vxx = 0 in R (0, ),
v(, 0) = , vt (, 0) = in R.
In other words, v is also a C 2 -solution of the initial-value problem for the
wave equation with the same initial values as u. By the uniqueness, u(x, t) =
v(x, t) = u(x, t) and hence u(0, t) = 0. In fact, u is given by dAlemberts
formula (6.1.5), i.e.,

1 1 x+t
u(x, t) = (x + t) + (x t) + (s) ds.
2 2 xt
By restricting (x, t) to [0, ) [0, ), we have, for any x t 0,

1 1 x+t
u(x, t) = (x + t) + (x t) + (s) ds,
2 2 xt
and for any t x 0,

1 1 x+t
(6.1.11) u(x, t) = (x + t) (t x) + (s) ds,
2 2 tx

since and are odd in R. We point out that (6.1.11) will be needed in
solving the initial-value problem for the wave equation in higher dimensions.
6.1. One-Dimensional Wave Equations 209

Now we consider the general case of (6.1.9) and construct a solution in


[0, )[0, ) by an alternative method. We rst decompose [0, )[0, )
into two regions by the straight line t = x. We note that t = x is the
characteristic curve for the wave equation in the domain [0, ) [0, )
passing through the origin, which is the corner of [0, ) [0, ). We will
solve for u in these two regions separately. First, we set

1 = {(x, t) : x > t > 0},

and
2 = {(x, t) : t > x > 0}.
We denote by u1 the solution in 1 . Then, u1 is determined by (6.1.5) from
the initial values. In fact,

1 1 x+t
u1 (x, t) = (x + t) + (x t) + (s) ds,
2 2 xt

for any (x, t) 1 . Set for x > 0,



1 1 2x
(x) = u1 (x, x) = (2x) + (0) + (s) ds.
2 2 0

We note that (x) is the value of the solution u along the straight line t = x
for x > 0. Next, we consider

utt uxx = 0 in 2 ,
u(0, t) = (t), u(x, x) = (x).

We denote its solution by u2 . For any (x, t) 2 , consider the characteristic


parallelogram with vertices (x, t), (0, t x), ( tx tx t+x t+x
2 , 2 ) and ( 2 , 2 ). In
other words, one vertex is (x, t), one vertex is on the boundary {x = 0} and
the other two vertices are on {t = x}. By (6.1.4), we have
 
tx tx t+x t+x
u2 (x, t) + u2 , = u2 (0, t x) + u2 , .
2 2 2 2
Hence
 
tx x+t
u2 (x, t) = (t x) +
2 2

1 1 x+t
= (t x) + (x + t) (t x) + (s) ds,
2 2 tx

for any (x, t) 2 . Set u = u1 in 1 and u = u2 in 2 . Now we check that


u, ut , ux , utt , uxx , utx are continuous along {t = x}. By a direct calculation,
210 6. Wave Equations

t6

@
@

@ @
@
@
@ x
-

Figure 6.1.4. Division by a characteristic curve.

we have
u1 (x, t)|t=x u2 (x, t)|t=x = (0) (0) = (0) (0),
x u1 (x, t)|t=x x u2 (x, t)|t=x = (0) +  (0),
x2 u1 (x, t)|t=x x2 u2 (x, t)|t=x =  (0)  (0).
Then (6.1.10) implies
u 1 = u2 , x u1 = x u2 , x2 u1 = x2 u2 on {t = x}.
It is easy to get t u1 = t u2 on {t = x} by u1 = u2 and x u1 = x u2 on
{t = x}. Similarly, we get xt u1 = xt u2 and tt u1 = tt u2 on {t = x}.
Therefore, u is C 2 across t = x. Hence, we obtain the following result.
Theorem 6.1.2. Suppose C 2 [0, ), C 1 [0, ), C 2 [0, ) and
the compatibility condition (6.1.10) holds. Then there exists a solution u
C 2 ([0, ) [0, )) of (6.1.9).

We can also derive a priori energy estimates for solutions of (6.1.9). For
any constants T > 0 and x0 > T , we use the following domain for energy
estimates:
{(x, t) : 0 < x < x0 t, 0 < t < T }.
Now we consider the initial/boundary-value problem. For a positive
constant l > 0, assume that C 2 [0, l], C 1 [0, l] and , C 2 [0, ).
Consider
utt uxx = 0 in (0, l) (0, ),
(6.1.12) u(, 0) = , ut (, 0) = on [0, l],
u(0, t) = (t), u(l, t) = (t) for t > 0.
The compatibility condition is given by
(0) = (0), (0) =  (0),  (0) =  (0),
(6.1.13)
(l) = (0), (l) =  (0),  (l) =  (0).
6.1. One-Dimensional Wave Equations 211

We rst consider the special case = 0. We discussed this case


using separation of variables in Section 3.3 if l = . We now construct
solutions by the method of reection. We rst extend to [l, 0] by odd
reection. In other words, we dene

(x) for x [0, l],
(x) =
(x) for x [l, 0].

We then extend to R as a 2l-periodic function. Then is odd in R. We


extend similarly. The extended functions and are C 2 and C 1 on R,
respectively. Let u be the unique solution of the initial-value problem

utt uxx = 0 in R (0, ),


u(, 0) = , ut (, 0) = on R.

We now prove that u(x, t) is a solution of (6.1.12) when we restrict x to


[0, l]. We need only prove that

u(0, t) = 0, u(l, t) = 0 for any t > 0.

The proof is similar to that for the half-space problem. We prove that
u(0, t) = 0 by introducing v(x, t) = u(x, t) and prove u(l, t) = 0 by
introducing w(x, t) = u(2l x, t).
We now discuss the general case and construct a solution of (6.1.12) by
an alternative method. We decompose [0, l] [0, ) into innitely many
regions by the characteristic curves through the corners and through the
intersections of the characteristic curves with the boundaries. Specically,
we rst consider the characteristic curve t = x. It starts from (0, 0), one of
the two corners, and intersects the right portion of the boundary x = l at
(l, l). Meanwhile, the characteristic curve x + t = l starts from (l, 0), the
other corner, and intersects the left portion of the boundary x = 0 at (0, l).
These two characteristic curves intersect at (l/2, l/2). We then consider the
characteristic curve tx = l from (0, l) and the characteristic curve t+x = 2l
from (l, l). They intersect the right portion of the boundary at (l, 2l) and the
left portion of the boundary at (0, 2l), respectively. We continue this process.
We rst solve for u in the characteristic triangle with vertex (l/2, l/2). In
this region, u is determined by the initial values. Then we can solve for u
by forming characteristic parallelograms in the triangle with vertices (0, 0),
(l/2, l/2) and (0, l) and in the triangle with vertices (l, 0), (l/2, l/2) and (l, l).
In the next step, we solve for u again by forming characteristic parallelogram
in the rectangle with vertices (0, l), (l/2, l/2), (l, l) and (l/2, 3l/2). We note
that this rectangle is a characteristic parallelogram. By continuing this
process, we can nd u in the entire region [0, l] [0, ).
212 6. Wave Equations

t6
2l
@
@
@
@
@
l @ @
@
@ @
@
@
@
@@
@ @
@
@@ -
l x

Figure 6.1.5. A decomposition by characteristic curves.

Theorem 6.1.3. Suppose C 2 [0, l], C 1 [0, l], , C 2 [0, ) and


the compatibility condition (6.1.13) holds. Then there exists a solution u
C 2 ([0, l] [0, )) of (6.1.12).

Theorem 6.1.3 includes Theorem 3.3.8 in Chapter 3 as a special case.


Now we summarize various problems discussed in this section. We em-
phasize that characteristic curves play an important role in studying the
one-dimensional wave equation.
First, presentations of problems depend on characteristic curves. Let
be a piecewise smooth domain in R2 whose boundary is not characteristic.
In the following, we shall treat the initial curve as a part of the boundary
and treat initial values as a part of boundary values. We intend to prescribe
appropriate values on the boundary to ensure the well-posedness for the
wave equation. To do this, we take an arbitrary point on the boundary
and examine characteristic curves through this point. We then count how
many characteristic curves enter the domain in the positive t-direction.
In this section, we discussed cases where is given by the upper half-space
R (0, ), the rst quadrant (0, ) (0, ) and I (0, ) for a nite
interval I. We note that the number of boundary values is the same as
the number of characteristic curves entering the domain in the positive t-
direction. In summary, we have

u|t=0 = , ut |t=0 = for initial-value problems;


u|t=0 = , ut |t=0 = , u|x=0 = for half-space problems;
u|t=0 = , ut |t=0 = , u|x=0 = , u|x=l =
for initial/boundary-value problems.
6.2. Higher-Dimensional Wave Equations 213

t 6

@
I
@

@
I
@ @
I
@
@
I
@ -
x

Figure 6.1.6. Characteristic directions.

Second, characteristic curves determine the domain of dependence and


the range of inuence. In fact, as illustrated by (6.1.5), initial values prop-
agate along characteristic curves.
Last, characteristic curves also determine domains for energy estimates.
We indicated domains of integration for initial-value problems and for half-
space problems. We will explore energy estimates in detail in Section 6.3.

6.2. Higher-Dimensional Wave Equations


In this section, we discuss the initial-value problem for the wave equation in
higher dimensions. Our main task is to derive an expression for its solutions
and discuss their properties.

6.2.1. The Method of Spherical Averages. Let C 2 (Rn ) and


C 1 (Rn ). Consider

utt u = 0 in Rn (0, ),
(6.2.1)
u(, 0) = , ut (, 0) = on Rn .

We will solve this initial-value problem by the method of spherical averages.


We rst discuss briey spherical averages. Let w be a continuous func-
tion in Rn . For any x Rn and r > 0, set

1
W (x; r) = w(y) dSy ,
n rn1 Br (x)

where n is the surface area of the unit sphere in Rn . Then W (x; r) is the
average of w over the sphere Br (x). Now, w can be recovered from W by

lim W (x; r) = w(x) for any x Rn .


r0
214 6. Wave Equations

Next, we suppose u is a C 2 -solution of (6.2.1). For any x Rn , t > 0


and r > 0, set

1
(6.2.2) U (x; r, t) = u(y, t) dSy
n rn1 Br (x)

and

1
(x; r) = (y) dSy ,
n rn1 Br (x)
(6.2.3) 
1
(x; r) = (y) dSy .
n rn1 Br (x)

In other words, U (x; r, t), (x, r) and (x, r) are the averages of u(, t),
and over the sphere Br (x), respectively. Then U determines u by

lim U (x; r, t) = u(x, t).


r0

Now we transform the dierential equation for u to a dierential equation


for U . We claim that, for each xed x Rn , U (x; r, t) satises the Euler-
Poisson-Darboux equation
n1
(6.2.4) Utt = Urr + Ur for r > 0 and t > 0,
r
with initial values

U (x; r, 0) = (x; r), Ut (x; r, 0) = (x; r) for r > 0.

It is worth pointing out that we treat x as a parameter in forming the


equation (6.2.4) and its initial values. To verify (6.2.4), we rst write

1
U (x; r, t) = u(x + r, t) dS .
n ||=1

By dierentiating under the integral sign and then integrating by parts, we


have
 
1 u 1 u
Ur = (x + r, t) dS = n1
(y, t) dSy
n ||=1 n r Br (x)

1
= u(y, t) dy.
n rn1 Br (x)

Then by the equation in (6.2.1),


 
n1 1 1
r Ur = u(y, t) dy = utt (y, t) dy.
n Br (x) n Br (x)
6.2. Higher-Dimensional Wave Equations 215

Hence

n1 1
(r Ur ) r = utt (y, t) dSy
n Br (x)

1
= tt u(y, t) dSy = rn1 Utt .
n Br (x)

For the initial values, we simply have for any r > 0,



1
U (x; r, 0) = (y) dSy ,
n rn1 Br (x)

1
Ut (x; r, 0) = (y) dSy .
n rn1 Br (x)

6.2.2. Dimension Three. We note that the Euler-Poisson-Darboux equa-


tion is a one-dimensional hyperbolic equation. In general, it is a tedious pro-
cess to solve the corresponding initial-value problems for general n. However,
this process is relatively easy for n = 3. If n = 3, we have
2
Utt = Urr + Ur .
r
Hence for r > 0 and t > 0,
(rU )tt = (rU )rr .
We note that rU satises the one-dimensional wave equation. Set
U (x; r, t) = rU (x; r, t)
and
(x; r) = r(x; r), (x; r) = r(x; r).
Then for each xed x R3 ,
Utt = Urr for r > 0 and t > 0,
U (x; r, 0) = (x; r), Ut (x; r, 0) = (x; r) for r > 0,
U (x; 0, t) = 0 for t > 0.
This is a half-space problem for U studied in Section 6.1. By (6.1.11), we
obtain formally for any t r > 0,

1 1 r+t
U (x; r, t) = (x; r + t) (x; t r) + (x; s) ds.
2 2 tr
Hence,
1
U (x; r, t) = (t + r)(x; t + r) (t r)(x; t r)
2r
 t+r
1
+ s(x; s) ds.
2r tr
216 6. Wave Equations

Letting r 0, we obtain

u(x, t) = lim U (x; r, t) = t t(x; t) + t(x; t).
r0

Note that the area of the unit sphere in R3 is 4. Then



1
(x; t) = (y) dSy ,
4t2 Bt (x)

1
(x; t) = (y) dSy .
4t2 Bt (x)
Therefore, we obtain formally the following expression of a solution u of
(6.2.1):
   
1 1
(6.2.5) u(x, t) = t (y) dSy + (y) dSy ,
4t Bt (x) 4t Bt (x)

for any (x, t) R3 (0, ). We point out that we did not justify the
compatibility condition in applying (6.1.11). Next, we prove directly that
(6.2.5) is indeed a solution u of (6.2.1) under appropriate assumptions on
and .

Theorem 6.2.1. Let k 2 be an integer, C k+1 (R3 ) and C k (R3 ).


Suppose u is dened by (6.2.5) in R3 (0, ). Then u C k (R3 (0, ))
and
utt u = 0 in R3 (0, ).
Moreover, for any x0 R3 ,

lim u(x, t) = (x0 ), lim ut (x, t) = (x0 ).


(x,t)(x0 ,0) (x,t)(x0 ,0)

In fact, u can be extended to a C k -function in R3 [0, ). This can be


easily seen from the proof below.

Proof. We will consider = 0. By (6.2.5), we have

u(x, t) = t(x, t),

where 
1
(x, t) = (y) dSy .
4t2 Bt (x)

By the change of coordinates y = x + t, we write



1
(x, t) = (x + t) dS .
4 ||=1
6.2. Higher-Dimensional Wave Equations 217

In this form, u(x, t) is dened for any (x, t) R3 [0, ) and u(, 0) = 0.
Since C k (R3 ), we conclude easily that ix u exists and is continuous in
R3 [0, ), for i = 0, 1, , k. In particular,

t
u(x, t) = x (x + t) dS .
4 ||=1
For t-derivatives, we take (x, t) R3 (0, ). Then
ut = + tt , utt = 2t + ttt .
A simple dierentiation yields

1
t (x, t) = (x + t) dS .
4 ||=1
Hence, ut (x, t) is dened for any (x, t) R3 [0, ) and ut (, 0) = .
Moreover, ix ut is continuous in R3 (0, ), for i = 0, 1, , k 1. After
the change of coordinates y = x + t and an integration by parts, we rst
have  
1 1
t = (y) dSy = (y) dy.
4t2 Bt (x) 4t2 Bt (x)
Then
 
1 1
tt = (y) dy + (y) dSy
2t3 Bt (x) 4t2 Bt (x)

2 1
= t (t) + (y) dSy .
t 4t2 Bt (x)
By setting y = x + t again, we have
 
t t
utt = y (y) dSy = x (x + t) dS = u.
4t2 Bt (x) 4 ||=1
This implies easily that u C k (R3 [0, )).
A similar calculation works for = 0. 

We point out that there are other methods to derive explicit expressions
for solutions of the wave equation. Refer to Exercise 6.8 for an alternative
approach to solving the three-dimensional wave equation.
By the change of variables y = x + t in (6.2.5), we have
   
t t
u(x, t) = t (x + t) dS + (x + t) dSy .
4 ||=1 4 ||=1
A simple dierentiation under the integral sign yields

1
u(x, t) = (x + t) + t(x + t) + t(x + t) dS .
4 ||=1
218 6. Wave Equations

Hence

1
u(x, t) = (y) + y (y) (y x) + t(y) dSy ,
4t2 Bt (x)

for any (x, t) R3 (0, ). We note that u(x, t) depends only on the initial
values and on the sphere Bt (x).

6.2.3. Dimension Two. We now solve initial-value problems for the wave
equation in R2 (0, ) by the method of descent. Let C 2 (R2 ) and
C 1 (R2 ). Suppose u C 2 (R2 (0, )) C 1 (R2 [0, )) satises (6.2.1),
i.e.,

utt u = 0 in R2 (0, ),
u(, 0) = , ut (, 0) = on R2 .

Any solutions in R2 can be viewed as solutions of the same problem in


R3 , which are independent of the third space variable. Namely, by setting
x = (x, x3 ) for x = (x1 , x2 ) R2 and

u(x, t) = u(x, t),

we have

utt x u = 0 in R3 (0, ),
u(, 0) = , ut (, 0) = on R3 ,

where
(x) = (x), (x) = (x).
By (6.2.5), we have
   
1 1
u(x, t) = t (y) dSy + (y) dSy ,
4t Bt (x) 4t Bt (x)

where y = (y1 , y2 , y3 ) = (y, y3 ). The integrals here are over the surface
Bt (x) in R3 . Now we evaluate them as integrals in R2 by eliminating y3 .
For x3 = 0, the sphere |y x| = t in R3 has two pieces given by

y3 = t2 |y x|2 ,

and its surface area element is


1 t
dSy = 1 + (y1 y3 )2 + (y2 y3 )2 2 dy1 dy2 =  dy.
t2 |y x|2
6.2. Higher-Dimensional Wave Equations 219

Therefore, we obtain
  
1 1 (y)
u(x, t) = t  dy
2 Bt (x) t2 |y x|2
(6.2.6) 
1 1 (y)
+  dy,
2 Bt (x) t2 |y x|2

for any (x, t) R2 (0, ). We put the factor 1/2 separately to emphasize
that is the area of the unit disc in R2 .

Theorem 6.2.2. Let k 2 be an integer, C k+1 (R2 ) and C k (R2 ).


Suppose u is dened by (6.2.6) in R2 (0, ). Then u C k (R2 (0, ))
and
utt u = 0 in R2 (0, ).

Moreover, for any x0 R2 ,

lim u(x, t) = (x0 ), lim ut (x, t) = (x0 ).


(x,t)(x0 ,0) (x,t)(x0 ,0)

This follows from Theorem 6.2.1. Again, u can be extended to a C k -


function in R2 [0, ).
By the change of variables y = x + tz in (6.2.6), we have
   
t (x + tz) t (x + tz)
u(x, t) = t  dz +  dz.
2 B1 1 |z| 2 2 B1 1 |z|2

A simple dierentiation under the integral sign yields



1 (x + tz) + t(x + tz) z + t(x + tz)
u(x, t) =  dz.
2 B1 1 |z|2

Hence

1 1 t(y) + t(y) (y x) + t2 (y)
u(x, t) = 2  dy,
2 t Bt (x) t2 |y x|2

for any (x, t) R2 (0, ). We note that u(x, t) depends on the initial
values and in the solid disc Bt (x).

6.2.4. Properties of Solutions. Now we compare several formulas we


obtained so far. Let u be a C 2 -solution of the initial-value problem (6.2.1).
220 6. Wave Equations

We write un for dimension n. Then for any (x, t) Rn (0, ),



1 1 x+t
u1 (x, t) = (x + t) + (x t) + (y) dy,
2 2 xt

1 t(y) + t(y) (y x) + t2 (y)
u2 (x, t) =  dy,
2t2 Bt (x) t2 |y x|2

1
u3 (x, t) = 2
(y) + y (y) (y x) + t(y) dSy .
4t Bt (x)
These formulas display many important properties of solutions u.
According to these expressions, the value of u at (x, t) depends on the
values of and on the interval [x t, x + t] for n = 1 (in fact, on only at
two endpoints), on the solid disc Bt (x) of center x and radius t for n = 2, and
on the sphere Bt (x) of center x and radius t for n = 3. These regions are
the domains of dependence of solutions at (x, t) on initial values. Conversely,

t 6

(x, t)
@
@ -
@
@
t @
x
Rn

Figure 6.2.1. The domain of dependence.

the initial values and at a point x0 on the initial hypersurface t = 0


inuence u at the points (x, t) in the solid cone |x x0 | t for n = 2 and
only on the surface |x x0 | = t for n = 3 at a later time t.
The central issue here is that the solution at a given point is determined
by the initial values in a proper subset of the initial hypersurface. An impor-
tant consequence is that the process of solving initial-value problems for the
wave equation can be localized in space. Specically, changing initial values
outside the domain of dependence of a point does not change the values of
solutions at this point. This is a unique property of the wave equation which
distinguishes it from the heat equation.
Before exploring the dierence between n = 2 and n = 3, we rst note
that it takes time (literally) for initial values to make inuences. Suppose
that the initial values , have their support contained in a ball Br (x0 ).
6.2. Higher-Dimensional Wave Equations 221

t6

t
@
@
@ -
@
@
(x0 , 0)
Rn

Figure 6.2.2. The range of inuence.

Then at a later time t, the support of u(, t) is contained in the union of all
balls Bt (x) for x Br (x0 ). It is easy to see that such a union is in fact the
ball of center x0 and radius r + t. The support of u spreads at a nite speed.
To put it in another perspective, we x an x / Br (x0 ). Then u(x, t) = 0 for
t < |x x0 | r. This is a nite-speed propagation.
For n = 2, if the supports of and are the entire disc Br (x0 ), then the
support of u(, t) will be the entire disc Br+t (x0 ) in general. The inuence
from initial values never disappears in a nite time at any particular point,
like the surface waves arising from a stone dropped into water.
For n = 3, the behavior of solutions is dierent. Again, we assume that
the supports of and are contained in a ball Br (x0 ). Then at a later
time t, the support of u(, t) is in fact contained in the union of all spheres
Bt (x) for x Br (x0 ). Such a union is the ball Bt+r (x0 ) for t r, as in
the two-dimensional case, and the annular region of center x0 and outer and
inner radii t + r and t r, respectively, for t > r. This annular region has
a thickness 2r and spreads at a nite speed. In other words, u(x, t) is not
zero only if
t r < |x x0 | < t + r,
or
|x x0 | r < t < |x x0 | + r.
Therefore, for a xed x R3 , u(x, t) = 0 for t < |x x0 | r (corresponding
to nite-speed propagation) and for t > |x x0 | + r. So, the inuence
from the initial values lasts only for an interval of length 2r in time. This
phenomenon is called Huygens principle for the wave equation. (It is called
the strong Huygens principle in some literature.)
In fact, Huygens principle holds for the wave equation in every odd space
dimension n except n = 1 and does not hold in even space dimensions.
222 6. Wave Equations

t6

@
@
@ -
@
R2

Figure 6.2.3. The range of inuence for n = 2.

t6

@ @
@ @
@ @ -
@ @
R3

Figure 6.2.4. The range of inuence for n = 3.

Now we compare regularity of solutions for n = 1 and n = 3. For n = 1,


the regularity of u is clearly the same as u(, 0) and one order better than
ut (, 0). In other words, u C m and ut C m1 initially at t = 0 guarantee
u C m at a later time. However, such a result does not hold for n = 3. The
formula for n = 3 indicates that u can be less regular than the initial values.
There is a possible loss of one order of dierentiability. Namely, u C k+1
and ut C k initially at t = 0 only guarantee u C k at a later time.
Example 6.2.3. We consider an initial-value problem for the wave equation
in R3 of the form
utt u = 0 in R3 (0, ),
u(, 0) = 0, ut (, 0) = on R3 .
Its solution is given by

1
u(x, t) = (y) dSy ,
4t Bt (x)

for any (x, t) R3 (0, ). We assume that is radially symmetric, i.e.,


(x) = h(|x|) for some function h dened in [0, ). Then

1
u(0, t) = (y) dSy = th(t).
4t Bt
6.2. Higher-Dimensional Wave Equations 223

For some integer k 3, if (x) is not C k at |x| = 1, then h(t) is not


C k at t = 1. Therefore, the solution u is not C k at (x, t) = (0, 1). The
physical interpretation is that the singularity of initial values at |x| = 1
propagates along the characteristic cone and focuses at its vertex. We note
that (x, t) = (0, 1) is the vertex of the characteristic cone {(x, t) : t = 1|x|}
which intersects {t = 0} at |x| = 1.

This example demonstrates that solutions of the higher-dimensional


wave equation do not have good pointwise behavior. A loss of dierentiabil-
ity in the pointwise sense occurs. However, the dierentiability is preserved
in the L2 -sense. We will discuss the related energy estimates in the next
section.

6.2.5. Arbitrary Odd Dimensions. Next, we discuss how to obtain ex-


plicit expressions for solutions of initial-value problems for the wave equation
in an arbitrary dimension. For odd dimensions, we seek an appropriate com-
bination of U (x; r, t) and its derivatives to satisfy the one-dimensional wave
equation and then proceed as for n = 3. For even dimensions, we again use
the method of descent.
Let n 3 be an odd integer. The spherical average U (x; r, t) dened by
(6.2.2) satises
n1
(6.2.7) Utt = Urr + Ur ,
r
for any r > 0 and t > 0. First, we write (6.2.7) as
1
Utt = rUrr + (n 1)Ur .
r
Since
(rU )rr = rUrr + 2Ur ,
we obtain
1
Utt = (rU )rr + (n 3)Ur ,
r
or
(6.2.8) (rU )tt = (rU )rr + (n 3)Ur .
If n = 3, then rU satises the one-dimensional wave equation. This is how
we solved the initial-value problem for the wave equation in dimension three.
By dierentiating (6.2.7) with respect to r, we have
n1 n1
Urtt = Urrr + Urr Ur
r r2
1
= 2 r2 Urrr + (n 1)rUrr (n 1)Ur .
r
Since
(r2 Ur )rr = r2 Urrr + 4rUrr + 2Ur ,
224 6. Wave Equations

we obtain
1 2
Urtt = 2
(r U )rr + (n 5)rUrr (n + 1)Ur ,
r
or
(6.2.9) (r2 Ur )tt = (r2 Ur )rr + (n 5)rUrr (n + 1)Ur .
The second term in the right-hand side of (6.2.9) has a coecient n 5,
which is 2 less than n 3, the coecient of the second term in the right-
hand side of (6.2.8). Also the third term involving Ur in the right-hand
side of (6.2.9) has a similar expression as the second term in the right-hand
side of (6.2.8). Therefore an appropriate combination of (6.2.8) and (6.2.9)
eliminates those terms involving Ur . In particular, for n = 5, we have
(r2 Ur + 3rU )tt = (r2 Ur + 3rU )rr .
In other words, r2 Ur + 3rU satises the one-dimensional wave equation. We
can continue this process to obtain appropriate combinations for all odd
dimensions. Next, we note that
1
r2 Ur + 3rU = (r3 U )r .
r
It turns out that the correct combination of U and its derivatives for arbi-
trary odd dimension n is given by
 n3
1 2
rn2 U .
r r

We rst state a simple calculus lemma.

Lemma 6.2.4. Let m be a positive integer and v = v(r) be a C m+1 -function


on (0, ). Then for any r > 0,
2    
d 1 d m1 2m1 1 d m 2m dv
(1) r v(r) = r (r) ;
dr2 r dr r dr dr
 
1 d m1 2m1 m1 di v
(2) r v(r) = cm,i ri+1 i (r),
r dr dr
i=0

where cm,i is a constant independent of v, for i = 0, 1, , m 1, and


cm,0 = 1 3 (2m 1).

The proof is by induction and is omitted.


Now we let n 3 be an odd integer and write n = 2m + 1. Let
C m (Rn ) and C m1 (Rn ). We assume that u C m+1 (Rn [0, )) is a
solution of the initial-value problem (6.2.1). Then U dened by (6.2.2) is
6.2. Higher-Dimensional Wave Equations 225

C m+1 , and and dened by (6.2.3) are C m and C m1 , respectively. For


x Rn , r > 0 and t 0, set

1 m1 2m1
(6.2.10) U (x; r, t) = r U (x; r, t) ,
r r
and
m1
1
(x; r) = r2m1 (x; r) ,
r r
m1
1
(x; r) = r2m1 (x; r) .
r r
We now claim that for each xed x Rn ,
Utt Urr = 0 in (0, ) (0, ),
U (x; r, 0) = (x; r), Ut (x; r, 0) = (x; r) for r > 0,
U (x; 0, t) = 0 for t > 0.
This follows by a straightforward calculation. First, in view of (6.2.4) and
n = 2m + 1, we have for any r > 0 and t > 0,
1 2m
(r Ur ) = r2m1 Urr + 2mr2m2 Ur
r r
n1
= r2m1 Urr + Ur = r2m1 Utt .
r
Then by (6.2.10) and Lemma 6.2.4(1), we have
2   
1 m1 2m1 1 m 2m
Urr = r U = (r Ur )
r2 r r r r

1 m1 2m1
= (r Utt ) = Utt .
r r
The initial condition easily follows from the denition of U , and . The
boundary condition U (x; 0, t) = 0 follows from Lemma 6.2.4(2).
As for n = 3, we have for any t r > 0,

1 1 t+r
U (x; r, t) = (x; t + r) (x; t r) + (x; s) ds.
2 2 tr

Note that by Lemma 6.2.4(2),



1 m1 2m1
U (x; r, t) = r U (x; r, t)
r r

m1
i
= cm,i ri+1 U (x; r, t).
ri
i=0
226 6. Wave Equations

Hence
1
lim U (x; r, t) = lim U (x; r, t) = u(x, t).
r0 cm,0 r r0

Therefore, we obtain
 r+t 
1 1 1
u(x, t) = lim (x; t + r) (x; t r) + (x; s) ds
cm,0 r0 2r 2r tr
1
= t (x; t) + (x; t) .
cm,0
Using n = 2m+1, the expression for cm,0 in Lemma 6.2.4 and the denitions
of and , we can rewrite the last formula in terms of and . Thus, we
obtain for any x Rn and t > 0,
,   n3   
1 1 2 1
u(x, t) = dS
cn t t t n t Bt (x)
(6.2.11)  n3   -
1 2 1
+ dS ,
t t n t Bt (x)

where n is an odd integer, n is the surface area of the unit sphere in Rn


and
(6.2.12) cn = 1 3 (n 2).

We note that c3 = 1 and hence (6.2.11) reduces to (6.2.5) for n = 3.


Now we check that u given by (6.2.11) indeed solves the initial-value
problem (6.2.1).
Theorem 6.2.5. Let n 3 be an odd integer and k 2 be an integer.
n1 n3
Suppose C 2 +k (Rn ), C 2 +k (Rn ) and u is dened in (6.2.11).
Then u C k (Rn (0, )) and
utt u = 0 in Rn (0, ).
Moreover, for any x0 Rn ,
lim u(x, t) = (x0 ), lim ut (x, t) = (x0 ).
(x,t)(x0 ,0) (x,t)(x0 ,0)

In fact, u can be extended to a C k -function in Rn [0, ).

Proof. The proof proceeds similarly to that of Theorem 6.2.1. We consider


= 0. Then for any (x, t) Rn (0, ),
 n3
1 1 2
u(x, t) = tn2 (x, t) ,
cn t t
6.2. Higher-Dimensional Wave Equations 227

where

1
(x, t) = dS.
n tn1 Bt (x)

By Lemma 6.2.4(2), we have


n3
1 
2
i
u(x, t) = c n1 ,i ti+1 i (x, t).
cn 2 t
i=0

Note that cn in (6.2.12) is c(n1)/2,0 in Lemma 6.2.4. By the change of


coordinates y = x + t, we write

1
(x, t) = (x + t) dS .
n ||=1

Therefore,

i 1 i
(x, t) = (x + t) dS .
ti n ||=1 i
Hence, u(x, t) is dened for any (x, t) Rn [0, ) and u(, 0) = 0. Since
n3
C 2 +k (Rn ), we conclude easily that ix u exists and is continuous in
Rn [0, ), for i = 0, 1, , k. For t-derivatives, we conclude similarly that
ut (x, t) is dened for any (x, t) Rn [0, ) and ut (, 0) = . Moreover,
ix ut is continuous in Rn (0, ), for i = 0, 1, , k 1. In particular,
 
1 1
t (x, t) = dS = dy.
n tn1 Bt (x) n tn1 Bt (x)

Next,

1
(x, t) = x (x + t) dS
n ||=1

1
= dSy .
n tn1 Bt (x)

Hence
 n3   
1 1 2 1
u(x, t) = dSy .
cn t t n t Bt (x)

On the other hand, Lemma 6.2.4(1) implies


 n1
1 1 2
utt = tn1 t .
cn t t
228 6. Wave Equations

Hence
 n1  
1 1 2
utt = dy
n cn t t Bt (x)
 n3   
1 1 2 1
= dS .
n cn t t t Bt (x)

This implies that utt u = 0 at (x, t) Rn (0, ) and then u


C k (Rn [0, )).
We can discuss the case = 0 in a similar way. 

6.2.6. Arbitrary Even Dimensions. Let n 2 be an even integer with


n = 2m 2, C m (Rn ) and C m1 (Rn ). We assume that u
C m (Rn [0, )) is a solution of the initial-value problem (6.2.1). We will
use the method of descent to nd an explicit expression for u in terms of
and .
By setting x = (x, xn+1 ) for x = (x1 , , xn ) Rn and
u(x, t) = u(x, t),
we have
utt x u = 0 in Rn+1 (0, ),
u(, 0) = , ut (, 0) = on Rn+1 ,
where
(x) = (x), (x) = (x).
As n + 1 is odd, by (6.2.11), with n + 1 replacing n, we have
,   n2   
1 1 2 1
u(x, t) = (y) dSy
cn+1 t t t n+1 t Bt (x)
 n2   -
1 2 1
+ (y) dSy ,
t t n+1 t Bt (x)

where y = (y1 , , yn , yn+1 ) = (y, yn+1 ). The integrals here are over the
surface Bt (x) in Rn+1 . Now we evaluate them as integrals in Rn by elim-
inating yn+1 . For xn+1 = 0, the sphere |y x| = t in Rn+1 has two pieces
given by

yn+1 = t2 |y x|2 ,
and its surface area element is
1 t
dSy = 1 + |y yn+1 |2 2 dy =  dy.
t2 |y x|2
6.2. Higher-Dimensional Wave Equations 229

Hence
 
1 2 (y)
(y) dSy =  dy
n+1 t Bt (x) n+1 Bt (x) |y x|2
t2

2n n (y)
=  dy.
nn+1 n Bt (x) t2 |y x|2
We point out that n /n is the volume of the unit ball in Rn . A similar
expression holds for . By a simple substitute, we now get an expression of
u in terms of and . We need to calculate the constant in the formula.
Therefore, we obtain for any x Rn and t > 0,
,   n2   
1 1 2 n (y)
u(x, t) =  dy
cn t t t n Bt (x) t2 |y x|2
(6.2.13)  n2   -
1 2 n (y)
+  dy ,
t t n Bt (x) t2 |y x|2
where n is an even integer, n /n is the volume of the unit ball in Rn and cn
is given by
ncn+1 n+1
cn = .
2n
In fact, we have
cn = 2 4 n.
We note that c2 = 2 and hence (6.2.13) reduces to (6.2.6) for n = 2.
Theorem 6.2.6. Let n be an even integer and k 2 be an integer. Suppose
C 2 +k (Rn ), C 2 1+k (Rn ) and u is dened in (6.2.13). Then u
n n

C k (Rn (0, )) and


utt u = 0 in Rn (0, ).
Moreover, for any x0 Rn ,
lim u(x, t) = (x0 ), lim ut (x, t) = (x0 ).
(x,t)(x0 ,0) (x,t)(x0 ,0)

This follows from Theorem 6.2.5. Again, u can be extended to a C k -


function in Rn [0, ).

6.2.7. Global Properties. Next, we discuss global properties of solutions


of the initial-value problem for the wave equation. First, we have the fol-
lowing global boundedness.
Theorem 6.2.7. For n 2, let be a smooth function in Rn and u be a
solution of
utt u = 0 in Rn (0, ),
u(, 0) = 0, ut (, 0) = on Rn .
230 6. Wave Equations

Then for any t > 0,



n1
|u(, t)|L (Rn ) C i L1 (Rn ) ,
i=0
where C is a positive constant depending only on n.

Solutions not only are bounded globally but also decay as t for
n 2. In this aspect, there is a sharp dierence between dimension 1
and higher dimensions. By dAlemberts formula (6.1.5), it is obvious that
solutions of the initial-value problem for the one-dimensional wave equation
do not decay as t . However, solutions in higher dimensions have a
dierent behavior.
Theorem 6.2.8. For n 2, let be a smooth function in Rn and u be a
solution of
utt u = 0 in Rn (0, ),
u(, 0) = 0, ut (, 0) = on Rn .
Then for any t > 1,
. %
n
2

|u(, t)|L(Rn ) Ct
n1
2 i L1 (Rn ) ,
i=0
where C is a positive constant depending only on n.

Decay estimates in Theorem 6.2.8 are optimal for large t. They play an
important role in the studies of global solutions of nonlinear wave equations.
We note that decay rates vary according to dimensions.
Before presenting a proof, we demonstrate that t1 is the correct decay
rate for n = 3 by a simple geometric consideration. By (6.2.5), the solution
u is given by 
1
u(x, t) = (y) dSy ,
4t Bt (x)
for any (x, t) R3 (0, ). Suppose is of compact support and supp
BR for some R > 0. Then

1
u(x, t) = (y) dSy .
4t BR Bt (x)
A simple geometric argument shows that for any x R3 and any t > 0,

Area BR Bt (x) CR2 ,
where C is a constant independent of x and t. Hence,
CR2
|u(x, t)| sup ||.
t R3
6.2. Higher-Dimensional Wave Equations 231

This clearly shows that u(x, t) decays uniformly for x R3 at the rate of t1
as t . The drawback here is that the diameter of the support appears
explicitly in the estimate. The discussion for n = 2 is a bit complicated and
is left as an exercise. Refer to Exercise 6.7.
We now prove Theorem 6.2.7 and Theorem 6.2.8 together. The proof is
based on explicit expressions for u.

Proof of Theorems 6.2.7 and 6.2.8. We rst consider n = 3. By as-


suming that is of compact support, we prove that for any t > 0,
1
|u(x, t)| 2 L1 (R3 ) ,
4
and for any t > 0,
1
|u(x, t)| L1 (R3 ) .
4t
By (6.2.5), the solution u is given by

t
u(x, t) = (x + t) dS ,
4 ||=1
for any (x, t) R3 (0, ). Since has compact support, we have


(x + t) = (x + s) ds.
t s
Then  
t
u(x, t) = (x + s) dS ds.
4 t ||=1 s
For s t, we have t s2 /t and hence
 
1 1
|u(x, t)| s2 |(x + s)| dS ds L1 (R3 ) .
4t t ||=1 4t
For the global boundedness, we rst have

2
(x + t) = s 2 (x + s) ds.
t s
Then  
t 2
u(x, t) = s 2 (x + s) dS ds.
4 t ||=1 s
Hence
 
1 1
|u(x, t)| s2 |2 (x + s)| dS ds 2 L1 (R3 ) .
4 t ||=1 4
We now discuss general . For any (x, t) R3 (0, ), we note that u
depends on only on Bt (x). We now take a cuto function C0 (R3 )
with = 1 in Bt+1 (x), = 0 in R3 \ Bt+2 (x) and a uniform bound on .
Then in the expression for u, we may replace by . We can obtain the
232 6. Wave Equations

desired estimates by repeating the argument above. We simply note that


derivatives of have uniform bounds, independent of (x, t) R3 (0, ).
Now we consider n = 2. By assuming that is of compact support, we
prove that for any t > 0,
1
|u(x, t)| L1 (R2 ) ,
4
and for any t > 1,
1
|u(x, t)| L1 (R2 ) + L1 (R2 ) .
2 t
The general case follows similarly to the case of n = 3. By (6.2.6) and a
change of variables, we have
 t 
1 r
u(x, t) = (x + r) dS dr.
2 0 t2 r2 ||=1

As in the proof for n = 3, we have for r > 0,


  

(x + r) dS = (x + s) dS ds,
||=1 r ||=1 s

and hence
   
  
 
r (x + r) dS  s |(x + s)| dS ds
 ||=1  r ||=1

L1 (R2 ) .

Therefore,
 t
1 1 1
|u(x, t)| L1 (R2 ) dr = L1 (R2 ) .
2 0 t2 r 2 4
For the decay estimate, we write u as
 t  t 
1 1
u(x, t) = + = I1 + I2 ,
2 0 t 2
where > 0 is a positive constant to be determined. We can estimate I2
similarly to the above. In fact,
  
 t 
 1 
|I2 | =  r (x + r) dS dr
 t t2 r2 ||=1 
 t
1
dr L1 (R2 ) .
t t r2
2
6.2. Higher-Dimensional Wave Equations 233

A simple calculation yields


 t  t
1 1
dr =  dr
t t r
2 2
t (t + r)(t r)
 t
1 1 2
dr = .
t t t r t
Hence,

2
|I2 | L1 (R2 ) .
t
For I1 , we have
  
 t 
 r 
|I1 | =  (x + r) dS dr
 0 t2 r2 ||=1 
 t 
1
 r |(x + r)| dS dr
t (t )2 0
2 ||=1
1
L1 (R2 ) .
2t 2
Therefore, we obtain

1 1 2
|u(x, t)| L1 (R2 ) + L1 (R2 ) .
2 2t 2 t
For any t > 1, we take = 1/2 and obtain the desired result.
We leave the proof for arbitrary n as an exercise. 

6.2.8. Duhamels Principle. We now discuss the initial-value problem


for the nonhomogeneous wave equation. Let and be C 2 and C 1 functions
in Rn , respectively, and f be a continuous function in Rn (0, ). Consider
utt u = f in Rn (0, ),
(6.2.14)
u(, 0) = , ut (, 0) = on Rn .
For f 0, the solution u of (6.2.14) is given by (6.2.11) for n odd and by
(6.2.13) for n even. We note that there are two terms in these expressions,
one being a derivative in t. This is not a coincidence.
We now decompose (6.2.14) into three problems,
utt u = 0 in Rn (0, ),
(6.2.15)
u(, 0) = , ut (, 0) = 0 on Rn ,

utt u = 0 in Rn (0, ),
(6.2.16)
u(, 0) = 0, ut (, 0) = on Rn ,
234 6. Wave Equations

and
utt u = f in Rn (0, ),
(6.2.17)
u(, 0) = 0, ut (, 0) = 0 on Rn .
Obviously, a sum of solutions of (6.2.15)(6.2.17) yields a solution of (6.2.14).
n
For any C [ 2 ]+1 (Rn ), set for (x, t) Rn (0, ),
 n3   
1 1 2 1
(6.2.18) M (x, t) = dS
cn t t n t Bt (x)

if n 3 is odd, and
 n2   
1 1 2 n (y)
(6.2.19) M (x, t) =  dy
cn t t n Bt (x) t2 |y x|2
if n 2 is even, where n is the surface area of the unit sphere in Rn and

1 3 (n 2) for n 3 odd,
cn =
2 4n for n 2 even.

We note that [ n2 ] + 1 = n+1


2 if n is odd, and [ n2 ] + 1 = n+2
2 if n is even.
n
Theorem 6.2.9. Let m 2 be an integer, C [ 2 ]+m1 (Rn ) and set u =
M . Then u C m (Rn (0, )) and
utt u = 0 in Rn (0, ).
Moreover, for any x0 Rn ,
lim u(x, t) = 0, lim ut (x, t) = (x0 ).
(x,t)(x0 ,0) (x,t)(x0 ,0)

Proof. This follows easily from Theorem 6.2.5 and Theorem 6.2.6 for = 0.
As we have seen, u is in fact C m in Rn [0, ). 

We now prove that solutions of (6.2.15) can be obtained directly from


those of (6.2.16).
n
Theorem 6.2.10. Let m 2 be an integer, C [ 2 ]+m (Rn ) and set u =
t M . Then u C m (Rn (0, )) and
utt u = 0 in Rn (0, ).
Moreover, for any x0 Rn ,
lim u(x, t) = (x0 ), lim ut (x, t) = 0.
(x,t)(x0 ,0) (x,t)(x0 ,0)
6.2. Higher-Dimensional Wave Equations 235

Proof. The proof is based on straightforward calculations. We point out


that u is C m in Rn [0, ). By the denition of M (x, t), we have

tt M M = 0 in Rn (0, ),
M (, 0) = 0, t M (, 0) = on Rn .

Then

tt u u = (tt )t M = t (tt M M ) = 0 in Rn (0, ),

and

u(, 0) = t M (x, t)(, 0) = on Rn ,


t u(, 0) = tt M (, 0) = M (, 0) = 0 on Rn .

We have the desired result. 

The next result is referred to as Duhamels principle.


n
Theorem 6.2.11. Let m 2 be an integer, f C [ 2 ]+m1 (Rn [0, )) and
u be dened by
 t
u(x, t) = Mf (x, t ) d,
0
where f = f (, ). Then u C (Rn
m (0, )) and

utt u = f in Rn (0, ).

Moreover, for any x0 Rn ,

lim u(x, t) = 0, lim ut (x, t) = 0.


(x,t)(x0 ,0) (x,t)(x0 ,0)

Proof. The regularity of u easily follows from Theorem 6.2.9. We will verify
that u satises utt u = f and the initial conditions. For each xed > 0,
w(x, t) = Mf (x, t ) satises

wtt w = 0 in Rn (, ),
w(, ) = 0, t w(, ) = f (, ) on Rn .

We note that the initial conditions here are prescribed on {t = }. Then


 t
ut = Mf (x, t )| =t + t Mf (x, t ) d
0
 t
= t Mf (x, t ) d,
0
236 6. Wave Equations

and
 t
utt = t Mf (x, t )| =t + tt Mf (x, t ) d
0
 t
= f (x, t) + Mf (x, t ) d
0
 t
= f (x, t) + Mf (x, t ) d
0
= f (x, t) + u.
Hence utt u = f in Rn (0, ) and u(, 0) = 0, ut (, 0) = 0 in Rn . 

As an application of Theorem 6.2.11, we consider the initial-value prob-


lem (6.2.17) for n = 3. Let u be a C 2 -solution of
utt u = f in R3 (0, ),
u(, 0) = 0, ut (, 0) = 0 on R3 .
By (6.2.18) for n = 3, we have for any C 2 (R3 ),

1
M (x, t) = (y) dSy .
4t Bt (x)
Then, by Theorem 6.2.11,
 t  t 
1 1
u(x, t) = Mf (x, t ) d = f (y, ) dSy d.
0 4 0 t Bt (x)
By the change of variables = t s, we have
 t
1 f (y, t s)
u(x, t) = dSy ds.
4 0 Bs (x) s
Therefore,

1 f (y, t |y x|)
(6.2.20) u(x, t) = dy,
4 Bt (x) |y x|
for any (x, t) R3 (0, ). We note that the value of the solution u at
(x, t) depends on the values of f only at the points (y, s) with
|y x| = t s, 0 < s < t.
This is exactly the backward characteristic cone with vertex (x, t).
Theorem 6.2.12. Let m 2 be an integer, f C m (R3 [0, )) and u be
dened by (6.2.20). Then u C m (R3 (0, )) and
utt u = f in R3 (0, ).
Moreover, for any x0 R3 ,
lim u(x, t) = 0, lim ut (x, t) = 0.
(x,t)(x0 ,0) (x,t)(x0 ,0)
6.3. Energy Estimates 237

6.3. Energy Estimates


In this section, we derive energy estimates of solutions of initial-value prob-
lems for a class of hyperbolic equations slightly more general than the wave
equation.
Before we start, we demonstrate by a simple case what is involved. Sup-
pose u is a C 2 -solution of
utt u = 0 in Rn (0, ).
We assume that u(, 0) and ut (, 0) have compact support. By nite-speed
propagation, u(, t) also has compact support for any t > 0. We multiply
the wave equation by ut and integrate in BR (0, t). Here we choose R
suciently large such that BR contains the support of u(, s), for any s
(0, t). Note that
1  n
ut utt ut u = (u2t + |x u|2 )t (ut uxi )xi .
2
i=1

Then a simple integration in BR (0, t) yields


 
1 1
(u + |x u| ) dx =
2 2
(u2 + |x u|2 ) dx.
2 Rn {t} t 2 Rn {0} t
This is conservation of energy: the L2 -norm of derivatives at each time slice
is a constant independent of time. For general hyperbolic equations, con-
servation of energy is not expected. However, we have the energy estimates:
the energy at later time is controlled by the initial energy.
Let a, c and f be continuous functions in Rn [0, ) and and be
continuous functions in Rn . We consider the initial-value problem
utt au + cu = f in Rn (0, ),
(6.3.1)
u(, 0) = , ut (, 0) = in Rn .
We assume that a is a positive function satisfying
(6.3.2) a(x, t) for any (x, t) Rn [0, ),
for some positive constants and . For the wave equation, we have a = 1
and c = 0 and hence we can choose = = 1 in (6.3.2).
In the following, we set
1
= .

For any point P = (X, T ) Rn (0, ), consider the cone C (P ) (opening
downward) with vertex at P dened by
C (P ) = {(x, t) : 0 < t < T, |x X| < T t}.
238 6. Wave Equations

As in Section 2.3, we denote by s C (P ) and C (P ) the side and bottom


of the boundary, respectively, i.e.,

s C (P ) = {(x, t) : 0 < t T, |x X| = T t},


C (P ) = {(x, 0) : |x X| T }.

We note that C (P ) is simply the closed ball in Rn {0} centered at


(X, 0) with radius T /.

6
t
P = (X, T )

J

J

J

J

J -

J

T / J

Rn

Figure 6.3.1. The cone C (P ).

Theorem 6.3.1. Let a be C 1 , c and f be continuous in Rn [0, ), and


let be C 1 and be continuous in Rn . Suppose (6.3.2) holds and u
C 2 (Rn (0, )) C 1 (Rn [0, )) is a solution of (6.3.1). Then for any
point P = (X, T ) Rn (0, ) and any > 0 ,

( 0 ) et (u2 + u2t + a|u|2 ) dxdt
C (P )
 
2 2 2
( + + a|| ) dx + et f 2 dxdt,
C (P ) C (P )

where 0 is a positive constant depending only on n, , the C 1 -norm of a


and the L -norm of c in C (P ).

Proof. We multiply the equation in (6.3.1) by 2et ut and integrate in


C (P ), for a nonnegative constant to be determined. First, we note that

2et ut utt = (et u2t )t + et u2t ,


6.3. Energy Estimates 239

and

n
2et aut u = 2et aut uxi xi
i=1

n

= 2(et aut uxi )xi + 2et auxi utxi + 2et axi ut uxi
i=1

n

= 2(et aut uxi )xi + (et au2xi )t + 2et axi ut uxi
i=1

+ et au2xi et at u2xi ,
where we used 2uxi utxi = (u2xi )t . Therefore, we obtain

n
(et u2t +e t
a|u| )t 2
2
(et aut uxi )xi + et (u2t + a|u|2 )
i=1

n
+ 2et axi ut uxi et at |u|2 + 2et cuut
i=1
t
= 2e ut f.
We note that the rst two terms in the left-hand side are derivatives of
quadratic expressions in x u and ut and that the next three terms are
quadratic in x u and ut . In particular, the third term is a positive quadratic
form. The nal term in the left-hand side involves u itself. To control this
term, we note that
(et u2 )t + et u2 2et uut = 0.
Then a simple addition yields
t 2 2 
n
e (u +ut + a|u|2 ) t 2(et aut uxi )xi
i=1
+ et (u2 + u2t + a|u|2 ) = RHS,
where

n
RHS = 2et axi ut uxi + et at |u|2 2et (c 1)uut + 2et ut f.
i=1
The rst three terms in RHS are quadratic in ut , uxi and u. Now by (6.3.2)
and the Cauchy inequality, we have

1 2
2|axi ut uxi | |axi |(ut + uxi ) |axi | ut + auxi ,
2 2 2

and similar estimates for other three terms in RHS. Hence
RHS 0 et (u2 + u2t + a|u|2 ) + et f 2 ,
240 6. Wave Equations

where 0 is a positive constant which can be taken as



1 1
0 = sup |at | + n + sup |x a| + sup |c| + 2.
C (P ) C (P ) C (P )

Then a simple substitution yields



n
t
e 2
(u + u2t + a|u| ) 2
t
2 (et aut uxi )xi
i=1
t
+ ( 0 )e 2
(u + u2t + a|u|2 ) et f 2 .
Upon integrating over C (P ), we obtain

( 0 ) et (u2 + u2t + a|u|2 ) dxdt
C (P )
  

n
t
+ e 2
(u + u2t + a|u| )t 2
2
aut uxi i dS
s C (P ) i=1
 
(u + 2
u2t 2
+ a|u| ) dx + et f 2 dxdt,
C (P ) C (P )

where the unit exterior normal vector on s C (P ) is given by



1 xX
(1 , , n , t ) = ,1 .
1 + 2 |x X|
We need only prove that the integrand for s C (P ) is nonnegative. We
claim that

n
BI (ut + a|u| )t 2
2 2
aut uxi i 0 on s C (P ).
i=1

To prove this, we rst note that, by the Cauchy inequality,


   1  n 1
 n   n 2  2 
 
 uxi i  u2xi i2 = |u| 1 t2 .
 
i=1 i=1 i=1

2
With t = 1/ 1 + , we have
1 2
BI ut + a|u|2 2a|ut | |u| .
1 + 2

By (6.3.2) and = 1/ , we have a 1. Hence
1 2
BI ut + a|u|2 2 a|ut | |u| 0.
1+ 2

Therefore, the boundary integral on s C (P ) is nonnegative and can be


discarded. 
6.3. Energy Estimates 241

A consequence of Theorem 6.3.1 is the uniqueness of solutions of (6.3.1).


We can also discuss the domain of dependence and the range of inuence as
in the previous section.
We note that the cone C (P ) in Theorem 6.3.1 plays the same role as
the cone in Theorem 2.3.4. The constant is chosen so that the boundary
integral over s C (P ) is nonnegative and hence can be dropped from the
estimate.
Similar to Theorem 2.3.5, we have the following result.

Theorem 6.3.2. Let a be C 1 , c and f be continuous in Rn [0, ), and


let be C 1 and be continuous in Rn . Suppose (6.3.2) holds and u
C 2 (Rn (0, )) C 1 (Rn [0, )) is a solution of (6.3.1). For a xed
T > 0, if f L2 (Rn (0, T )) and , x , L2 (Rn ), then for any > 0 ,

et (u2 + u2t + a|u|2 ) dx
Rn {T }

+ ( 0 ) et (u2 + u2t + a|u|2 ) dxdt
R (0,T )
n
 
2 2
( + + a|| ) dx +2
et f 2 dxdt,
Rn Rn (0,T )

where 0 is a positive constant depending only on n, , the C 1 -norm of a


and the L -norm of c in Rn [0, T ].

Usually, we call u2t + a|u|2 the energy density and its integral over
Rn {t} the energy at time t. Then Theorem 6.3.2 asserts, in the case of
c = 0 and f = 0, that the initial energy (the energy at t = 0) controls the
energy at later time.

Next, we consider the initial-value problem in general domains. Let


be a bounded domain in Rn and h and h+ be two piecewise C 1 -functions
in with h < h+ in and h = h+ on . Set

D = {(x, t) : h (x) < t < h+ (x), x }.

Let a be C 1 and c be continuous in D. We assume that

a in D.

We now consider

(6.3.3) utt au + cu = f in D.
242 6. Wave Equations

t 6

+ D

D
D
-
Rn

Figure 6.3.2. A general domain.

We can perform a similar integration in D as in the proof of Theorem 6.3.1


and obtain
  
n
t
e (u + ut + a|u| )+t 2
2 2 2
aut uxi +i dS
+ D i=1

+ ( 0 ) et (u2 + u2t + a|u|2 ) dxdt
  D


n
t
e 2
(u + u2t + a|u| )t 2
2
aut uxi i dS
D i=1

+ et f 2 dxdt,
D
where = (1 , , n , t ) are unit normal vectors pointing in the pos-
itive t-direction along D. We are interested in whether the integrand for
+ D is nonnegative. As in the proof of Theorem 6.3.1, we have, by the
Cauchy inequality,
   1  n 1
n   n 2  2 
 
 uxi +i  u2xi 2
+i = |u| 1 +t2 .
 
i=1 i=1 i=1
Then it is easy to see that

n
(u2t + a|u| )+t 2
2
aut uxi +i
i=1

(u2t + a|u|2 )+t 2 a(1 +t
2 ) a|ut | |u| 0 on + D
if 
+t a(1 +t
2 ).

This condition can be written as



a
(6.3.4) +t on + D.
1+a
6.3. Energy Estimates 243

In conclusion, under the condition (6.3.4), we obtain


 
t 2
e u +t dS + ( 0 ) et (u2 + u2t + a|u|2 ) dxdt
+ D D
  

n
t
e (u 2
+ u2t + a|u| )t 2
2
aut uxi i dS
D i=1

+ et f 2 dxdt.
D
If we prescribe u and ut on D, then x u can be calculated on D in
terms of u and ut . Hence, the expressions in the right-hand side are known.
In particular, if u = ut = 0 on D and f = 0 in D, then u = 0 in D.
Now we introduce the notion of space-like and time-like surfaces.
Denition 6.3.3. Let be a C 1 -hypersurface in Rn R+ and = (x , t )
be a unit normal vector eld on with t 0. Then is space-like at (x, t)
for (6.3.3) if
/
a(x, t)
t (x, t) > ;
1 + a(x, t)
is time-like at (x, t) if
/
a(x, t)
t (x, t) < .
1 + a(x, t)

If the hypersurface is given by t = t(x), it is easy to check that is


space-like at (x, t(x)) if
1
|t(x)| <  .
a(x, t(x))
Now we consider the wave equation
(6.3.5) utt u = f.

With a = 1, the hypersurface is space-like at (x, t) if t (x, t) > 1/ 2. If
is given by t = t(x), then is space-like at (x, t(x)) if
|t(x)| < 1.
In the following, we demonstrate the importance of space-like hypersurfaces
by the wave equation.
Let be a space-like hypersurface for the wave equation. Then for
any (x0 , t0 ) , the range of inuence of (x0 , t0 ) is given by the cone
{(x, t) : t t0 > |x x0 |} and hence is always above . This suggests that
prescribing initial values on space-like surfaces yields a well-posed problem.
244 6. Wave Equations

t6

@
@
@
@
-

Rn

Figure 6.3.3. A space-like hypersurface.

t6

@
@
@
-

Rn

Figure 6.3.4. An integral domain for space-like initial hypersurfaces.

In fact, domains of integration for energy estimates can be constructed ac-


cordingly.
Next, we discuss briey initial-value problems with initial values pre-
scribed on a time-like hypersurface. Consider
utt = uxx + uyy for x > 0 and y, t R,
1 u 1
u= sin my, = sin my on {x = 0}.
m2 x m
Here we treat {x = 0} as the initial hypersurface, which is time-like for the
wave equation. A solution is given by
1
um (x, y) = 2 emx sin my.
m
Note that
um
um 0, 0 on {x = 0} as m .
x
Meanwhile, for any x > 0,
1
sup |um (x, )| = 2 emx as m .
R 2 m
Therefore, there is no continuous dependence on the initial values.
6.4. Exercises 245

To conclude this section, we discuss a consequence of Theorem 6.3.2. In


Subsection 2.3.3, we proved in Theorem 2.3.7 the existence of weak solutions
of the initial-value problem for the rst-order linear PDEs with the help
of estimates in Theorem 2.3.5. By a similar process, we can prove the
existence of weak solutions of (6.3.1) using Theorem 6.3.2. However, there
is a signicant dierence. The weak solutions in Denition 2.3.6 are in L2
because an estimate of the L2 -norms of solutions is established in Theorem
2.3.5. In the present situation, Theorem 6.3.2 establishes an estimate of the
L2 -norms of solutions and their derivatives. This naturally leads to a new
norm dened by
 1
2

uH 1 (Rn (0,T )) = (u2 + u2t + |x u|2 ) dxdt .


Rn (0,T )

The superscript 1 in H 1 indicates the order of derivatives. With such a


norm, we can dene the Sobolev space H 1 (Rn (0, T )) as the completion
of smooth functions of nite H 1 -norms with respect to the H 1 -norm. Obvi-
ously, H 1 (Rn (0, T )) dened in this way is complete. In fact, it is a Hilbert
space, since the H 1 -norm is naturally induced by an H 1 -inner product given
by

(u, v)H 1 (Rn (0,T )) = (uv + ut vt + x u x v) dxdt.
Rn (0,T )

Then we can prove that (6.3.1) admits a weak H 1 -solution in Rn (0, T ) if


= = 0. We will not provide the details here. The purpose of this short
discussion is to demonstrate the importance of Sobolev spaces in PDEs. We
refer to Subsection 4.4.2 for a discussion of weak solutions of the Poisson
equation.

6.4. Exercises
Exercise 6.1. Let l be a positive constant, C 2 ([0, l]) and C 1 ([0, l]).
Consider

utt uxx = 0 in (0, l) (0, ),


u(, 0) = , ut (, 0) = in [0, l],
u(0, t) = 0, ux (l, t) = 0 for t > 0.

Find a compatibility condition and prove the existence of a C 2 -solution


under such a condition.
246 6. Wave Equations

Exercise 6.2. Let 1 and 2 be C 2 -functions in {x < 0} and {x > 0},


respectively. Consider the characteristic initial-value problem
utt uxx = 0 for t > |x|,
u(x, x) = 1 (x) for x < 0,
u(x, x) = 2 (x) for x > 0.
Solve this problem and nd the domain of dependence for any point (x, t)
with t > |x|.
Exercise 6.3. Let 1 and 2 be C 2 -functions in {x > 0}. Consider the
Goursat problem
utt uxx = 0 for 0 < t < x,
u(x, 0) = 1 (x), u(x, x) = 2 (x) for x > 0.
Solve this problem and nd the domain of dependence for any point (x, t)
with 0 < t < x.
Exercise 6.4. Let be a constant and and be C 2 -functions on (0, )
which vanish near x = 0. Consider
utt uxx = 0 for x > 0, t > 0,
u(x, 0) = (x), ut (x, 0) = (x) for x > 0,
ut (0, t) = ux (0, t) for t > 0.
Find a solution for = 1 and prove that in general there exist no solutions
for = 1.
Exercise 6.5. Let a be a constant with |a| < 1. Prove that the wave
equation
utt x u = 0 in R3 R
is preserved by a Lorentz transformation, i.e., a change of variables given by
t ax1
s= ,
1 a2
x1 at
y1 = ,
1 a2
yi = xi for i = 2, 3.

Exercise 6.6. Let be a positive constant and C 2 (R2 ). Solve the


following initial-value problems by the method of descent:
utt = u + 2 u in R2 (0, ),
u(, 0) = 0, ut (, 0) = on R2 ,
6.4. Exercises 247

and
utt = u 2 u in R2 (0, ),
u(, 0) = 0, ut (, 0) = on R2 .
Hint: Use complex functions temporarily to solve the second problem.
Exercise 6.7. Let be a bounded function dened in R2 with = 0 in
R2 \ B1 . For any (x, t) R2 (0, ), dene

1 (y)
u(x, t) =  dy.
2 Bt (x) t |y x|2
2

(1) For any (0, 1), prove


C
sup |u(, t)| sup || for any t > 1,
Bt t R2
where C is a positive constant depending only on .
(2) Assume, in addition, that = 1 in B1 . For any unit vector e R2 ,
nd the decay rate of u(te, t) as t .
Exercise 6.8. Let C 2 (R3 ) and C 1 (R3 ). Suppose that u
C 2 (R3 [0, )) is a solution of the initial-value problem
utt u = 0 in R3 (0, ),
u(, 0) = , ut (, 0) = on R3 .
(1) For any xed (x0 , t0 ) R3 (0, ), set for any x Bt0 (x0 ) \ {x0 },

x u(x, t) x x0 x x0 
v(x) = + u(x, t) + u (x, t)  .
|x x0 | |x x0 |3 |x x0 |2
t 
t=t0 |xx0 |

Prove that div v = 0.


(2) Derive an expression of u(x0 , t0 ) in terms of and by integrating
div v in Bt0 (x0 ) \ B (x0 ) and then letting 0.
Remark: This exercise gives an alternative approach to solving the initial-
value problem for the three-dimensional wave equation.
Exercise 6.9. Let a be a positive constant and u be a C 2 -solution of the
characteristic initial-value problem
utt u = 0 in {(x, t) R3 (0, ) : t > |x| > a},
u(x, |x|) = 0 for |x| > a.
(1) For any xed (x0 , t0 ) R3 R+ with t0 > |x0 | > a, integrate div v
(introduced in Exercise 6.8) in the region bounded by |xx0 |+|x| =
t0 , |x| = a and |x x0 | = . By letting 0, express u(x0 , t0 ) in
terms of an integral over Ba .
248 6. Wave Equations

(2) For any S2 and > 0, prove that the limit



lim ru(r, r + )
r
exists and the convergence is uniform for S2 and (0, 0 ], for
any xed 0 > 0.
1
Remark: The limit in (2) is called the radiation eld.

Exercise 6.10. Prove Theorem 6.2.7 and Theorem 6.2.8 for n 2.


Exercise 6.11. Set QT = {(x, t) : 0 < x < l, 0 < t < T }. Consider the
equation
Lu 2utt + 3utx + uxx = 0.
(1) Give a correct presentation of the boundary-value problem in QT .
(2) Find an explicit expression of a solution with prescribed boundary
values.
(3) Derive an estimate of the integral of u2x + u2t in QT .
Hint: For (2), divide QT into three regions separated by characteristic curves
from (0, 0). For (3), integrate an appropriate linear combination of ut Lu and
ux Lu to make integrands on [0, l] {t} and {l} [0, t] positive denite.
Exercise 6.12. For some constant a > 0, let f be a C 1 -function in a <
|x| < t + a, a C 1 -function on r0 < |x| = t a and a C 1 -function on
|x| = a and t > 0. Consider the characteristic initial-value problem for the
wave equation
utt u = f (x, t) in a < |x| < t + a,
u = (x, t) on |x| > a, t = |x| a,
u = (x, t) on |x| = a, t > 0.
Derive an energy estimate in an appropriate domain in a < |x| < t + a.

1F. G. Friedlander, On the radiation eld of pulse solutions of the wave equation, Proc. Roy.
Soc. A, 269 (1962), 5365.
Chapter 7

First-Order Dierential
Systems

In this chapter, we discuss partial dierential systems of the rst order and
focus on local existence of solutions.
In Section 7.1, we introduce the notion of noncharacteristic hypersur-
faces for initial-value problems. We proceed here for linear partial dier-
ential equations and partial dierential systems of arbitrary order similarly
to how we did for rst-order linear PDEs in Section 2.1 and second-order
linear PDEs in Section 3.1. We show that we can compute all derivatives of
solutions on initial hypersurfaces if initial values are prescribed on nonchar-
acteristic initial hypersurfaces. We also demonstrate that partial dierential
systems of arbitrary order can always be transformed to those of the rst
order.
In Section 7.2, we discuss analytic solutions of the initial-value problem
for rst-order linear dierential systems. The main result is the Cauchy-
Kovalevskaya theorem, which asserts the local existence of analytic solutions
if the coecient matrices and the nonhomogeneous terms are analytic and
the initial values are analytic on analytic noncharacteristic hypersurfaces.
The proof is based on the convergence of the formal power series of solutions.
In this section, we also prove a uniqueness result due to Holmgren, which
asserts that the solutions in the Cauchy-Kovalevskaya theorem are the only
solutions in the C -category.
In Section 7.3, we construct a rst-order linear dierential system in
R3 that does not admit smooth solutions in any subsets of R3 . In this
system, the coecient matrices are analytic and the nonhomogeneous term

249
250 7. First-Order Dierential Systems

is a suitably chosen smooth function. (For analytic nonhomogeneous terms


there would always be solutions by the Cauchy-Kovalevskaya theorem). We
need to point out that such a nonhomogeneous term is proved to exist by a
contradiction argument. An important role is played by the Baire category
theorem.

7.1. Noncharacteristic Hypersurfaces


The main focus in this section is on linear partial dierential systems of
arbitrary order.

7.1.1. Linear Partial Dierential Equations. We start with linear par-


tial dierential equations of arbitrary order and proceed here as in Sections
2.1 and 3.1.
Let be a domain in Rn containing the origin, m be a positive integer
and a be a continuous function in , for Zn+ with || m. Consider
an mth-order linear dierential operator L dened by

(7.1.1) Lu = a (x) u in .
||m

Here, a is called the coecient of u.


Denition 7.1.1. Let L be a linear dierential operator of order m as in
(7.1.1) dened in Rn . The principal part L0 and the principal symbol p
of L are dened by

L0 u = a (x) u in ,
||=m

and 
p(x; ) = a (x) ,
||=m
for any x and Rn .

The principal part L0 is a dierential operator consisting of terms in-


volving derivatives of order m in L, and the principal symbol is a homoge-
neous polynomial of degree m with coecients given by the coecients of
L0 . Principal symbols play an important role in discussions of dierential
operators.
We discussed rst-order and second-order linear dierential operators
in Chapter 2 and Chapter 3, respectively. Usually, they are written in the
forms
n
Lu = ai (x)uxi + b(x)u in ,
i=1
7.1. Noncharacteristic Hypersurfaces 251

and

n 
n
Lu = aij (x)uxi xj + bi (x)uxi + c(x)u in .
i,j=1 i=1
Their principal symbols are given by

n
p(x; ) = ai (x)i ,
i=1
and

n
p(x; ) = aij (x)i j ,
i,j=1
for any x and Rn . For second-order dierential operators, we
usually assume that (aij ) is a symmetric matrix in .
Let f be a continuous function in . We consider the equation
(7.1.2) Lu = f (x) in .
The function f is called the nonhomogeneous term of the equation.
Let be a smooth hypersurface in with a unit normal vector eld
= (1 , , n ). For any integer j 1, any point x0 and any C j -
function u dened in a neighborhood of x0 , the jth normal derivative of u
at x0 is dened by
j u  
j
= u = 11 nn x11 xnn u.

||=j 1 ++n =j

We now prescribe the values of u and its normal derivatives on so that


we can nd a solution u of (7.1.2) in . Let u0 , u1 , , um1 be continuous
functions dened on . We set
u m1 u
(7.1.3) u = u0 , = u1 , , = um1 on .
m1
We call the initial hypersurface and u0 , , um1 the initial values or
Cauchy values. The problem of solving (7.1.2) together with (7.1.3) is called
the initial-value problem or Cauchy problem.
We note that there are m functions u0 , u1 , , um1 in (7.1.3). This re-
ects the general fact that m conditions are needed for initial-value problems
for PDEs of order m.
As the rst step in discussing the solvability of the initial-value problem
(7.1.2)(7.1.3), we intend to nd all derivatives of u on . We consider a
special case where is the hyperplane {xn = 0}. In this case, we can take
= en and the initial condition (7.1.3) has the form
(7.1.4) u(, 0) = u0 , xn u(, 0) = u1 , , xm1
n
u(, 0) = um1 on Rn1 .
252 7. First-Order Dierential Systems

Let u0 , u1 , , um1 be smooth functions on Rn1 and u be a smooth


solution of (7.1.2) and (7.1.4) in a neighborhood of the origin. In the fol-
lowing, we investigate whether we can compute all derivatives of u at the
origin in terms of the equation and initial values. We write x = (x , xn ) for
x Rn1 . First, we can nd all x -derivatives of u at the origin in terms of
those of u0 . Next, we can nd all x -derivatives of uxn at the origin in terms
of those of u1 . By continuing this process, we can nd all x -derivatives
of u, uxn , , xm1
n
u at the origin in terms of those of u0 , u1 , , um1 .
In particular, for derivatives up to order m, we nd all except xmn u. To
nd xmn u(0), we need to use the equation. We note that a(0, ,0,m) is the
coecient of xmn u in (7.1.2). If we assume
(7.1.5) a(0, ,0,m) (0) = 0,
then by (7.1.2),

1 
xmn u(0) = a (0) u(0) f (0) .
a(0, ,0,m) (0)
=(0, ,0,m)

Hence, we can compute all derivatives up to order m at 0 in terms of


the coecients and nonhomogeneous term in (7.1.2) and the initial values
u0 , u1 , , um1 in (7.1.4). In fact, we can compute the derivatives of u of
arbitrary order at the origin. For an illustration, we nd the derivatives of
u of order m + 1. By (7.1.5), a(0, ,0,m) is not zero in a neighborhood of the
origin. Hence, by (7.1.2),

1 
xmn u = a u f .
a(0, ,0,m)
=(0, ,0,m)

By evaluating at x Rn1 {0} close to the origin, we nd xmn u(x) for


x Rn1 {0} suciently small. As before, we can nd all x -derivatives
of xmn u at the origin. Hence for derivatives up to order m + 1, we nd all
except xm+1
n
u. To nd xm+1
n
u(0), we again need to use the equation. By
dierentiating (7.1.2) with respect to xn , we obtain
a(0, ,0,m) xm+1
n
u + = fxn ,
where the dots denote a linear combination of derivatives of u whose values
on Rn1 {0} are already calculated in terms of the derivatives of u0 , u1 ,
, um1 , f and the coecients in the equation. By (7.1.5) and the above
equation, we can nd xm+1 n
u(0). We can continue this process for derivatives
of arbitrary order. In summary, we can nd all derivatives of u of any
order at the origin under the condition (7.1.5), which will be dened as the
noncharacteristic condition later on.
7.1. Noncharacteristic Hypersurfaces 253

In general, consider the hypersurface given by { = 0} for a smooth


function in a neighborhood of the origin with = 0. We note that the
vector eld is normal to the hypersurface at each point of . We take
a point on , say the origin. Then (0) = 0. Without loss of generality,
we assume that xn (0) = 0. Then by the implicit function theorem, we
can solve = 0 for xn = (x1 , , xn1 ) in a neighborhood of the origin.
Consider the change of variables
x y = (x1 , , xn1 , (x)).
This is a well-dened transformation with a nonsingular Jacobian in a neigh-
borhood of the origin. With
n
uxi = yk,xi uyk = xi uyn + terms not involving uyn ,
k=1
and in general, for any Zn+ with || = m,
x u = x11 xnn u = x11 xnn ymn u + terms not involving ymn u,
we can write the operator L in the y-coordinates as

Lu = a x(y) x11 xnn ymn u + terms not involving ymn u.
||=m

The initial hypersurface is given by {yn = 0} in the y-coordinates. With


yn = , the coecient of ymn u is given by

a (x)x11 xnn .
||=m

This is the principal symbol p(x; ) evaluated at = (x).


Denition 7.1.2. Let L be a linear dierential operator of order m dened
as in (7.1.1) in a neighborhood of x0 Rn and be a smooth hypersurface
containing x0 . Then is noncharacteristic at x0 if

(7.1.6) p(x0 ; ) = a (x0 ) = 0,
||=m

where = (1 , , n ) is normal to at x0 . Otherwise, is characteristic


at x0 .

A hypersurface is noncharacteristic if it is noncharacteristic at every


point. Strictly speaking, a hypersurface is characteristic if it is not non-
characteristic, i.e., if it is characteristic at some point. In this book, we will
abuse this terminology. When we say a hypersurface is characteristic, we
mean it is characteristic everywhere. This should cause no confusion. In
R2 , hypersurfaces are curves, so we shall speak of characteristic curves and
noncharacteristic curves.
254 7. First-Order Dierential Systems

When the hypersurface is given by { = 0} with = 0, its normal


vector eld is given by = (x1 , , xn ). Hence we may take =
(x0 ) in (7.1.6). We note that the condition (7.1.6) is preserved under
C m -changes of coordinates. By this condition, we can nd successively the
values of all derivatives of u at x0 , as far as they exist. Then, we could
write formal power series at x0 for solutions of initial-value problems. If the
initial hypersurface is analytic and the coecients, nonhomogeneous terms
and initial values are analytic, then this formal power series converges to an
analytic solution. This is the content of the Cauchy-Kovalevskaya theorem,
which we will discuss in Section 7.2.
Now we introduce a special class of linear dierential operators.

Denition 7.1.3. Let L be a linear dierential operator of order m dened


as in (7.1.1) in a neighborhood of x0 Rn . Then L is elliptic at x0 if

p(x0 ; ) = a (x0 ) = 0,
||=m

for any Rn \ {0}.

A linear dierential operator dened in is called elliptic in if it is


elliptic at every point in .
According to Denition 7.1.3, linear dierential operators are elliptic if
every hypersurface is noncharacteristic.
Consider a rst-order linear dierential operator of the form

n
Lu = ai (x)uxi + b(x)u in Rn .
i=1

Its principal symbol is given by



n
p(x; ) = ai (x)i ,
i=1

for any x and any Rn . Hence rst-order linear dierential equa-


tions with real coecients are never elliptic. Complex coecients may yield
elliptic equations. For example, take a1 = 1/2 and a2 = i/2 in R2 . Then
z = (x1 + ix2 )/2 is elliptic.
The notion of ellipticity was introduced in Denition 3.1.2 for second-
order linear dierential operators of the form

n 
n
Lu = aij (x)uxi xj + bi (x)uxi + c(x)u in Rn .
i,j=1 i=1
7.1. Noncharacteristic Hypersurfaces 255

The principal symbol of L is given by



n
p(x; ) = aij (x)i j ,
i,j=1

for any x and any Rn . Then L is elliptic at x if



n
aij (x)i j = 0 for any Rn \ {0}.
i,j=1

If (aij (x)) is a real-valued nn symmetric matrix, L is elliptic at x if (aij (x))


is a denite matrix at x, positive denite or negative denite.

7.1.2. Linear Partial Dierential Systems. The concept of noncharac-


teristics can be generalized to linear partial dierential equations for vector-
valued functions. Let m, N 1 be integers and Rn be a domain. A
smooth N N matrix A in is an N N matrix whose components are
smooth functions in . Similarly, a smooth N -vector u is a vector of N
components which are smooth functions in . Alternatively, we may call
them a smooth N N matrix-valued function and a smooth N -vector-valued
function, or a smooth RN -valued function, respectively. In the following, a
function may mean a scalar-valued function, a vector-valued function, or a
matrix-valued function. This should cause no confusion. Throughout this
chapter, all vectors are in the form of column vectors.
Let A be a smooth N N matrix in , for each Zn+ with || m.
Consider a linear partial dierential operator of the form

(7.1.7) Lu = A (x) u in ,
||m

where u is a smooth N -vector in . Here, A is called the coecient matrix


of u.
We dene principal parts, principal symbols and noncharacteristic hy-
persurfaces similarly to those for single dierential equations.
Denition 7.1.4. Let L be a linear dierential operator dened in Rn
as in (7.1.7). The principal part L0 and the principal symbol p of L are
dened by 
L0 u = A (x) u in ,
||=m
and

p(x; ) = det A (x) ,
||=m
for any x and Rn .
256 7. First-Order Dierential Systems

Denition 7.1.5. Let L be a linear dierential operator dened in a neigh-


borhood of x0 Rn as in (7.1.7) and be a smooth hypersurface containing
x0 . Then is noncharacteristic at x0 if


p(x0 ; ) = det A (x0 ) = 0,
||=m

where = (1 , , n ) is normal to at x0 . Otherwise, is characteristic


at x0 .

Let f be a smooth N -vector in . We consider the linear dierential


equation
(7.1.8) Lu = f (x) in .
The function f is called the nonhomogeneous term of the equation. We
often call (7.1.8) a partial dierential system, treating (7.1.8) as a collection
of partial dierential equations for the components of u.
Let be a smooth hypersurface in with a normal vector eld and
let u0 , u1 , , um1 be smooth N -vectors on . We prescribe
u m1 u
(7.1.9) u = u0 , = u1 , , = um1 on .
m1
We call the initial hypersurface and u0 , , um1 the initial values or
Cauchy values. The problem of solving (7.1.8) together with (7.1.9) is called
the initial-value problem or Cauchy problem.
We now examine rst-order linear partial dierential systems. Let A1 ,
, An and B be smooth N N matrices in a neighborhood of x0 Rn .
Consider a rst-order linear dierential operator

n
Lu = Ai uxi + Bu.
i=1
A hypersurface containing x0 is noncharacteristic at x0 if
 n 

det i Ai (x0 ) = 0,
i=1

where = (1 , , n ) is normal to at the x0 .


We now demonstrate that we can always reduce the order of dierential
systems to 1 by increasing the number of equations and the number of
components of solution vectors.
Proposition 7.1.6. Let L be a linear dierential operator dened in a
neighborhood of x0 Rn as in (7.1.7), be a smooth hypersurface containing
x0 which is noncharacteristic at x0 for the operator L, and u0 , u1 , , um1
7.1. Noncharacteristic Hypersurfaces 257

be smooth on . Then the initial-value problem (7.1.8)(7.1.9) in a neigh-


borhood of x0 is equivalent to an initial-value problem for a rst-order dif-
ferential system with appropriate initial values prescribed on , and is
a noncharacteristic hypersurface at x0 for the new rst-order dierential
system.

Proof. We assume that x0 is the origin. In the following, we write x =


(x , xn ) Rn and = ( , n ) Zn+ .
Step 1. Straightening initial hypersurfaces. We assume that is given by
{ = 0} for a smooth function in a neighborhood of the origin with xn =
0. Then we introduce a change of coordinates x = (x , xn ) (x , (x)).
In the new coordinates, still denoted by x, the hypersurface is given by
{xn = 0} and the initial condition (7.1.9) is given by
xj n u(x , 0) = uj (x ) for j = 0, 1, , m 1.

Step 2. Reductions to canonical forms and zero initial values. In the new
coordinates, {xn = 0} is noncharacteristic at 0. Then, the coecient matrix
A(0, ,0,m) is nonsingular at the origin and hence also in a neighborhood of
the origin. Multiplying the partial dierential system (7.1.8) by the inverse
of this matrix, we may assume that A(0, ,0,m) is the identity matrix in a
neighborhood of the origin. Next, we may assume
uj (x ) = 0 for j = 0, 1, , m 1.
To see this, we introduce a function v such that

m1
1
u(x) = v(x) + uj (x )xjn .
j!
j=0

Then the dierential system for v is the same as that for u with f replaced
by
 
m1 
1  j
f (x) A (x)
uj (x )xn .
j!
j=0 ||m

Moreover,
xj n v(x , 0) = 0 for j = 0, 1, , m 1.
With Step 1 and Step 2 done, we assume that (7.1.8) and (7.1.9) have
the form

m1 
xmn u + A u = f,
n =0 | |mn

with
xj n u(x , 0) = 0 for j = 0, 1, , m 1.
258 7. First-Order Dierential Systems

Step 3. Lowering the order. We now change this dierential system to


an equivalent system of order m 1. Introduce new functions
U0 = u, Ui = uxi for i = 1, , n,
and
(7.1.10) U = (U0T , U1T , , UnT )T ,
where T indicates the transpose. We note that U is a column vector of
(n + 1)N components. Then
U0,xn = Un , Ui,xn = Un,xi for i = 1, , n 1.
Hence
(7.1.11) xm1
n
U0 xm2
n
Un = 0,
(7.1.12) xm1
n
Ui xi xm2
n
Un = 0 for i = 1, , n 1.
To get an (m 1)th-order dierential equation for Un , we write the equation
for u as

m1   
xmn u + A u + A( ,0) ( ,0) u = f.
n =1 | |mn | |m

We substitute Un = uxn in the rst two terms in the left-hand side to get

m2   
(7.1.13) xm1
n
Un + A Un + A( ,0) ( ,0) u = f.
n =0 | |mn 1 | |m

In the last summation in the left-hand side, any mth-order derivative of u can
be changed to an (m 1)th-order derivative of Ui for some i = 1, , n 1,
since no derivatives with respect to xn are involved. Now we can write a
dierential system for U in the form

m2 
(7.1.14) xm1
n
U + A(1)
U =F
(1)
.
n =0 | |mn 1

The initial value for U is given by


xj n U (x , 0) = 0 for j = 0, 1, , m 2.
Hence, we reduce the original initial-value problem for a dierential system
of order m to an initial-value problem for the dierential system of the form
(7.1.14) of order m 1.
Now let U be a solution of (7.1.14) with zero initial values. By writing
U as in (7.1.10), we prove that U0 is a solution of the initial-value problem
for the original dierential system of order m. To see this, we rst prove
7.2. Analytic Solutions 259

that Ui = U0,xi , for i = 1, , n. By (7.1.11) and the initial conditions for


U , we have
xm2
n
(Un U0,xn ) = 0,
and on {xn = 0},
xj n (Un U0,xn ) = 0 for j = 0, , m 3.
This easily implies Un = U0,xn . Next, for i = 1, , n 1,
xm1
n
Ui xi xm2
n
Un = xm1
n
Ui xi xm1
n
U0 = xm1
n
(Ui xi U0 ).
By (7.1.12) and the initial conditions, we have
xm1
n
(Ui xi U0 ) = 0,
and on {xn = 0}
xj n (Ui xi U0 ) = 0 for j = 0, , m 2.
Hence, Ui = U0,xi , for i = 1, , n 1. Substituting Ui = U0,xi , for i =
1, , n, in (7.1.13), we conclude that U0 is a solution for the original mth-
order dierential system.
Now, we can repeat the procedure to reduce m to 1. 

We point out that straightening initial hypersurfaces and reducing initial


values to zero are frequently used techniques in discussions of initial-value
problems.

7.2. Analytic Solutions


For a given rst-order linear partial dierential system in a neighborhood of
x0 Rn and an initial value u0 prescribed on a hypersurface containing
x0 , we rst intend to nd a solution u formally. To this end, we need to
determine all derivatives of u at x0 , in terms of the derivatives of the ini-
tial value u0 and of the coecients and the nonhomogeneous term in the
equation. Obviously, all tangential derivatives (with respect to ) of u are
given by derivatives of u0 . In order to nd the derivatives of u involving
the normal direction, we need help from the equation. It has been estab-
lished that, if is noncharacteristic at x0 , the initial-value problem leads
to evaluations of all derivatives of u at x0 . This is clearly a necessary rst
step to the determination of a solution of the initial-value problem. If the
coecient matrices and initial values are analytic, a Taylor series solution
could be developed for u. The Cauchy-Kovalevskaya theorem asserts the
convergence of this Taylor series in a neighborhood of x0 .
To motivate our discussion, we study an example of rst-order partial
dierential systems which may admit no solutions in any neighborhood of
260 7. First-Order Dierential Systems

the origin, unless the initial values prescribed on analytic noncharacteristic


hypersurfaces are analytic.
Example 7.2.1. Let g = g(x) be a real-valued function in R. Consider the
partial dierential system in R2+ = {(x, y) : y > 0},
uy + vx = 0,
(7.2.1)
ux vy = 0,
with initial values given by
u = g(x), v = 0 on {y = 0}.
We point out that (7.2.1) is simply the Cauchy-Riemann equation in C = R2 .
It can be written in the matrix form
    
1 0 u 0 1 u 0
+ = .
0 1 v y 1 0 v x 0
Note that {y = 0} is noncharacteristic. In fact, there are no characteristic
curves. To see this, we need to calculate the principal symbol. By taking
= (1 , 2 ) R2 , we have
  
1 0 0 1 2 1
2 + 1 = .
0 1 1 0 1 2
The determinant of this matrix is 12 + 22 , which is not zero for any = 0.
Therefore, there are no characteristic curves. We now write (7.2.1) in a
complex form. Suppose we have a solution (u, v) for (7.2.1) with the given
initial values and let w = u + iv. Then
wx + iwy = 0 in R2+ ,
w(, 0) = g on R.
Therefore, w is (complex) analytic in the upper half-plane and its imaginary
part is zero on the x-axis. By the Schwartz reection principle, w can be
extended across y = 0 to an analytic function in C = R2 . This implies
in particular that g is (real) analytic since w(, 0) = g. We conclude that
(7.2.1) admits no solutions with the given initial value g on {y = 0} unless
g is real analytic.

Example 7.2.1 naturally leads to discussions of analytic solutions.

7.2.1. Real Analytic Functions. We introduced real analytic functions


in Section 4.2. We now discuss this subject in detail.
For (real) analytic functions, we need to study convergence of innite
series of the form 
c ,

7.2. Analytic Solutions 261

where the c are real numbers dened for all multi-indices Zn+ . Through-
out this section,0
the term convergence always refers0 to absolute convergence.
Hence, a series c is convergent if and only if |c | < . Here, the
summation is over all multi-indices Zn+ .
Denition 7.2.2. A function u : Rn R is called analytic near x0 Rn if
there exist an r > 0 and constants {u } such that

u(x) = u (x x0 ) for x Br (x0 ).

If u is analytic near x0 , then u is smooth near x0 . Moreover, the con-


stants u are given by
1
u = u(x0 ) for Zn+ .
!
Thus u is equal to its Taylor series about x0 , i.e.,
 1
u(x) = u(x0 )(x x0 ) for x Br (x0 ).

!
For brevity, we will take x0 = 0.
Now we discuss an important analytic function.
Example 7.2.3. For r > 0, set
r
u(x) = for x B r .
r (x1 + + xn ) n

Then
 
x1 + + xn 1  x1 + + xn k
u(x) = 1 =
r r
k=0

 
1  k
 ||!
= x = x .
rk r|| !
k=0 ||=k

This power series is absolutely convergent for |x| < r/ n since
 ||!  
|x1 | + + |xn | k
|| !
|x
| = < ,

r r
k=0

for |x1 | + + |xn | |x| n < r. We also note that
||!
u(0) = for Zn+ .
r||
We point out that all derivatives of u at 0 are positive.

An eective method to prove analyticity of functions is to control their


derivatives by the derivatives of functions known to be analytic. For this,
we introduce the following terminology.
262 7. First-Order Dierential Systems

Denition 7.2.4. Let u and v be smooth functions dened in Br Rn , for


some r > 0. Then v majorizes u in Br , denoted by v  u or u  v, if
v(0) | u(0)| for any Zn+ .
We also call v a majorant of u in Br .

The following simple result concerns the convergence of Taylor series.


Lemma 7.2.5. Let u and v be smooth functions in Br . If v  u and the
Taylor series of v about the origin converges in Br , then the Taylor series
of u about the origin converges in Br .

Proof. We simply note that


 1  1
| u(0)x | v(0)|x | < for x Br .

!
!
Hence we have the desired convergence for u. 

Next, we prove that every analytic function has a majorant.


Lemma 7.2.6. If the Taylor series of u about the origin is convergent to u

in Br and 0 < s n < r, then u has an analytic majorant in Bs/n .

Proof. Set y = s(1, , 1). Then, |y| = s n < r and
 1
u(0)y

!
is a convergent series. There exists a constant C such that for any Zn+ ,
 
1 
 u(0)y  C,
 ! 
and in particular,
 
1 
 u(0) C C ||! .
 !  y 1 yn n s|| !
1
Now set
Cs  ||!
v(x) =C x .
s (x1 + + xn )
s|| !

Then v is analytic in Bs/n and majorizes u. 

So far, our discussions are limited to scalar-valued functions. All de-


nitions and results can be generalized to vector-valued functions easily. For
example, a vector-valued function u = (u1 , , uN ) is analytic if each of its
components is analytic. For vector-valued functions u = (u1 , , uN ) and
v = (v1 , , vN ), we say u  v if ui  vi for i = 1, , N .
We have the following results for compositions of functions.
7.2. Analytic Solutions 263

Lemma 7.2.7. Let u, v be smooth functions in a neighborhood of 0 Rn


with range in Rm and f, g be smooth functions in a neighborhood of 0 Rm
with range in RN , with u(0) = 0, f (0) = 0, u  v and f  g. Then
f u  g v.
Lemma 7.2.8. Let u be an analytic function near 0 Rn with range in Rm
and f be an analytic function near u(0) Rm with range in RN . Then f u
is analytic near 0 Rn .

We leave the proofs as exercises.

7.2.2. Cauchy-Kovalevskaya Theorem. Now we are ready to discuss


real analytic solutions of initial-value problems. We study rst-order quasi-
linear partial dierential systems of N equations for N unknowns in Rn+1 =
{(x, t)} with initial values prescribed on the noncharacteristic hyperplane
{t = 0}.
Let A1 , , An be smooth N N matrices in Rn+1+N , F be a smooth
N -vector in Rn+1+N and u0 be a smooth N -vector in Rn . Consider

n
(7.2.2) ut = Aj (x, t, u)uxj + F (x, t, u),
j=1

with
(7.2.3) u(, 0) = u0 .
We assume that A1 , , An , F and u0 are analytic in their arguments and
seek an analytic solution u. We point out that {t = 0} is noncharacteris-
tic for (7.2.2). Noncharacteristics was dened for linear dierential systems
in Section 7.1 and can be generalized easily to quasilinear dierential sys-
tems. We refer to Section 2.1 for such a generalization for single quasilinear
dierential equations.
The next result is referred to as the Cauchy-Kovalevskaya theorem.
Theorem 7.2.9. Let u0 be an analytic N -vector near 0 Rn , and let A1 ,
, An be analytic N N matrices and F be an analytic N -vector near
(0, 0, u0 (0)) Rn+1+N . Then the problem (7.2.2)(7.2.3) admits an analytic
solution u near 0 Rn+1 .

Proof. Without loss of generality, we assume u0 = 0. To this end, we


introduce v by v(x, t) = u(x, t) u0 (x). Then the dierential system for v
is similar to that for u. Next, we add t as an additional component of u by
introducing uN +1 such that uN +1,t = 1 and uN +1 (, 0) = 0. This increases
the number of equations and the number of components of the solution
vector in (7.2.2) by 1 and at the same time deletes t from A1 , , An and
264 7. First-Order Dierential Systems

F . For brevity, we still denote by N the number of equations and the number
of components of solution vectors.
In the following, we study

n
(7.2.4) ut = Aj (x, u)uxj + F (x, u),
j=1

with
(7.2.5) u(, 0) = 0,
where A1 , , An are analytic N N matrices and F is an analytic N -vector
in a neighborhood of the origin in Rn+N . We seek an analytic solution u
in a neighborhood of the origin in Rn+1 . To this end, we will compute
derivatives of u at 0 Rn+1 in terms of derivatives of A1 , , An and F at
(0, 0) Rn+N and then prove that the Taylor series of u at 0 converges in a
neighborhood of 0 Rn+1 . We note that t does not appear explicitly in the
right hand side of (7.2.4).
Since u = 0 on {t = 0}, we have
x u(0) = 0 for any Zn+ .
For any i = 1, , n, by dierentiating (7.2.4) with respect to xi , we get

n
uxi t = (Aj uxi xj + Aj,xi uxj + Aj,u uxi uxj ) + Fu uxi + Fxi .
j=1

In view of (7.2.5), we have


uxi t (0) = Fxi (0, 0).
More generally, we obtain by induction
x t u(0) = x F (0, 0) for any Zn+ .
Next, for any Zn+ , we have

n
x t2 u = x t ut = x t Aj uxj + F
j=1

n
= x (Aj uxj t + Aj,u ut uxj ) + Fu ut .
j=1

Here we used the fact that Aj and F are independent of t. Thus,




n

x t2 u(0) = x (Aj uxj t + Aj,u ut uxj ) + Fu ut  .
j=1 (x,t,u)=0
7.2. Analytic Solutions 265

The expression in the right-hand side can be worked out to be a polynomial


with nonnegative coecients in various derivatives of A1 , , An and F and
the derivatives x tl u with || + l || + 2 and l 1.
More generally, for any Zn+ and k 0, we have

(7.2.6) x tk u(0) = p,k (x u A1 , , x u An , x u F, x tl u)(x,t,u)=0 ,

where p,k is a polynomial with nonnegative coecients and the indices


, , , l range over , Zn+ , ZN
+ and l Z+ with || + || || + k 1,
|| + l || + k and l k 1.
We point out that p,k (x u A1 , ) is considered as a polynomial in the
components of x u A1 , . We denote by p,k (|x u A1 |, ) the value of
p,k when all components x u A1 , are replaced by their absolute values.
Since p,k has nonnegative coecients, we conclude that
  
p,k ( A1 , , An , F, l u) 
x u x u x u x t (x,t,u)=0
(7.2.7) 
p,k (|x u A1 |, , |x u An |, |x u F |, |x tl u|)(x,t,u)=0 .

We now consider a new dierential system


n
vt = Bj (x, v)vxi + G(x, v),
(7.2.8) j=1
v(, 0) = 0,

where B1 , , Bn are analytic N N matrices and G is an analytic N -vector


in a neighborhood of the origin in Rn+N . We will choose B1 , , Bn and G
such that

(7.2.9) Bj  Aj for j = 1, , n and G  F.

Hence, for any (, ) Zn+N


+ ,

x u Bj (0) |x u Aj (0)| for j = 1, , n,

and
x u G(0) |x u F (0)|.

The above inequalities should be understood as holding componentwise.


Let v be a solution of (7.2.8). We now claim that

|x tk u(0)| x tk v(0) for any (, k) Zn+1


+ .
266 7. First-Order Dierential Systems

The proof is by induction on the order of t-derivatives. The general step


follows since
  
 
|x tk u(0)| = p,k (x u A1 , , x u An , x u F, x tl u)(x,t,u)=0 

p,k (|x u A1 |, , |x u An |, |x u F |, |x tl u|)(x,t,u)=0

p,k (x u B1 , , x u Bn , x u G, x tl v)(x,t,u)=0
= x tk v(0),
where we used (7.2.6), (7.2.7) and the fact that p,k has nonnegative coe-
cients. Thus
(7.2.10) v  u.
It remains to prove that the Taylor series of v at 0 converges in a neighbor-
hood of 0 Rn+1 .
To this end, we consider

1 1
Cr .. . . ..
B1 = = Bn = . .
r (x1 + + xn + v1 + + vN ) .
1 1
and
1
Cr ..
G= . ,
r (x1 + + xn + v1 + + vN )
1

for positive constants C and r, with |x| + |v| < r/ n + N . As demonstrated
in the proof of Lemma 7.2.6, we may choose C suciently large and r
suciently small such that (7.2.9) holds.
Set
1
..
v = w . ,
1
for some scalar-valued function w in a neighborhood of 0 Rn+1 . Then
(7.2.8) is reduced to
 
Cr 
n
wt = N wxi + 1 ,
r (x1 + + xn + N w)
i=1
w(, 0) = 0.
This is a (single) rst-order quasilinear partial dierential equation. We now
seek a solution w of the form
w(x1 , , xn , t) = w(x1 + + xn , t).
7.2. Analytic Solutions 267

Then w = w(z, t) satises


Cr
wt = (nN wz + 1),
r z N w
w(, 0) = 0.
By using the method of characteristics as in Section 2.2, we have an explicit
solution
1  1
w(z, t) = r z [(r z)2 2Cr(n + 1)N t] 2 ,
(n + 1)N
and hence
 1
2 2
1 
n 
n
w(x, t) = r xi r xi 2Cr(n + 1)N t .
(n + 1)N


i=1 i=1

This function is analytic near the origin and its Taylor series about the
origin is convergent for |(x, t)| < s, for suciently small s > 0. Hence, the
corresponding solution v of (7.2.8) is analytic and its Taylor series about
the origin is convergent for |(x, t)| < s. By Lemma 7.2.5 and (7.2.10), the
Taylor series of u about the origin is convergent and hence denes an analytic
function for |(x, t)| < s, which 0we denote by u. Since the Taylor series of
n
the analytic functions ut and j=1 Aj (x, u)uxj + F (x, u) have the same
coecients at the origin, they agree throughout the region |(x, t)| < s. 

At the beginning of the proof, we introduced an extra component for the


solution vector to get rid of t in the coecient matrices of the dierential
system. Had we chosen to preserve t, we would have to solve the initial-value
problem
Cr
wt = (nN wz + 1),
r z t Nw
w(, 0) = 0.
It is dicult, if not impossible, to nd an explicit expression of the solution
w.

7.2.3. The Uniqueness Theorem of Holmgren. The solution given


in Theorem 7.2.9 is the only analytic solution since all derivatives of the
solution are computed at the origin and they uniquely determine the analytic
solution. A natural question is whether there are other solutions, which are
not analytic.
Let A0 , A1 , , An and B be analytic N N matrices, and let F be
an analytic N -vector in a neighborhood of the origin in Rn+1 and u0 be an
268 7. First-Order Dierential Systems

analytic N -vector in a neighborhood of the origin in Rn . We consider the


initial-value problem for linear dierential systems of the form
n
A0 (x, t)ut + Aj (x, t)uxj + B(x, t)u = F (x, t),
(7.2.11) j=1
u(x, 0) = u0 (x).
The next result is referred to as the local Holmgren uniqueness theorem.
It asserts that there do not exist nonanalytic solutions.
Theorem 7.2.10. Let A0 , A1 , , An and B be analytic N N matrices and
F be an analytic N -vector near the origin in Rn+1 and u0 be an analytic
N -vector near the origin in Rn . If {t = 0} is noncharacteristic at the
origin, then any C 1 -solution of (7.2.11) is analytic in a suciently small
neighborhood of the origin in Rn+1 .

For the proof, we need to introduce adjoint operators. Let L be a dif-


ferential operator dened by

n
Lu = A0 (x, t)ut + Ai (x, t)uxi + B(x, t)u.
i=1
For any N -vectors u and v, we write
 T

n 
n
T T
v Lu = (v A0 u)t + (v Ai u)xi
T
(AT0 v)t + (ATi v)xi B v T
u.
i=1 i=1
We dene the adjoint operator L of L by

n
L v = (AT0 v)t (ATi v)xi + B T v
i=1
 

n 
n
= AT0 vt ATi vxi + B
T
AT0,t ATi,xi v.
i=1 i=1
Then

n
T
v T Lu = (v T A0 u)t + (v T Ai u)xi + L v u.
i=1

Proof of Theorem 7.2.10. We will prove that any C 1 -solution u of Lu =


0 with a zero initial value on {t = 0} is in fact zero. We introduce an analytic
change of coordinates so that the initial hypersurface {t = 0} becomes a
paraboloid
t = |x|2 .
For any > 0, we set
= {(x, t) : |x|2 < t < }.
7.2. Analytic Solutions 269

We will prove that u = 0 in for a suciently small . In the following, we


denote by + and the upper and lower boundary of , respectively,
i.e.,
+ = {(x, t) : |x|2 < t = },
= {(x, t) : |x|2 = t < }.
We note that det(A0 (0)) = 0 since is noncharacteristic at the origin.
Hence A0 is nonsingular in a neighborhood of the origin. By multiplying
the equation in (7.2.11) by A1
0 , we assume A0 = I.

t6

-



 Rn
+


Figure 7.2.1. A parabola.

For any N -vector v dened in a neighborhood of the origin containing


, we have
  
0= v T Lu dxdt = uT L v dxdt + uv T dx.
+
There is no boundary integral over since u = 0 there. Let Pk =
Pk (x) be an arbitrary polynomial in Rn , k = 1, , N , and form P =
(P1 , , PN ). We consider the initial-value problem
L v = 0 in Br ,
v=P on Br {t = },
where Br is the ball in Rn+1 with center at the origin and radius r. The
principal part of L is the same as that of L, except a dierent sign and a
transpose. We x r so that {t = } Br is noncharacteristic for L , for each
small . By Theorem 7.2.9, an analytic solution v exists in Br for small.
We need to point out that the domain of convergence of v is independent
of P , whose components are polynomials. We choose small such that
Br . Then we have

u P dx = 0.
+
270 7. First-Order Dierential Systems

By the Weierstrass approximation theorem, any continuous function in a


compact domain can be approximated in the L -norm by a sequence of
polynomials. Hence, 
u w dx = 0,
+
for any continuous function w on + Br . Therefore, u = 0 on + for
any small and hence in . 

Theorem 7.2.9 guarantees the existence of solutions of initial-value prob-


lems in the analytic setting. As the next example shows, we do not expect
any estimates of solutions in terms of initial values.
Example 7.2.11. In R2 , consider the rst-order homogeneous linear dier-
ential system (7.2.1),
uy + vx = 0,
ux vy = 0.
Note that all coecients are constant. As shown in Example 7.2.1, {y = 0}
is noncharacteristic. For any integer k 1, consider
uk (x, y) = sin(kx)eky , vk (x, y) = cos(kx)eky for any (x, y) R2 .
Then (uk , vk ) satises (7.2.1) and on {y = 0},
uk (x, 0) = sin(kx), vk (x, 0) = cos(kx) for any x R.
Obviously,
u2k (x, 0) + vk2 (x, 0) = 1 for any x R,
and for any y > 0,

sup u2k (x, y) + vk2 (x, y) = e2ky as k .
xR
Therefore, there is no continuous dependence on initial values.

7.3. Nonexistence of Smooth Solutions


In this section, we construct a linear dierential equation which does not
admit smooth solutions anywhere, due to Lewy. In this equation, the coe-
cients are complex-valued analytic functions and the nonhomogeneous term
is a suitably chosen complex-valued smooth function. We need to point
out that such a nonhomogeneous term is proved to exist by a contradiction
argument. This single equation with complex coecients for a complex-
valued solution is equivalent to a system of two dierential equations with
real coecients for two real-valued functions.
Dene a linear dierential operator L in R3 = {(x, y, z)} by
(7.3.1) Lu = ux + iuy 2i(x + iy)uz .
7.3. Nonexistence of Smooth Solutions 271

We point out that L acts on complex-valued functions.


The main result in this section is the following theorem.
Theorem 7.3.1. Let L be the linear dierential operator in R3 dened in
(7.3.1). Then there exists an f C (R3 ) such that Lu = f has no C 2 -
solutions in any open subset of R3 .

Before we prove Theorem 7.3.1, we rewrite L as a dierential system of


two equations with real coecients for two real-valued functions. By writing
u = v + iw for real-valued functions v and w, we can write L as a dierential
operator acting on vectors (v, w)T . Hence
 
v v wy + 2(yvz + xwz )
L = x .
w wx + vy + 2(ywz xvz )
In the matrix form, we have
     
v v 0 1 v 2y 2x v
L = + + .
w w x 1 0 w y 2x 2y w z
By a straightforward calculation, the principal symbol is given by
p(P ; ) = (1 + 2y3 )2 + (2 2x3 )2 ,
for any P = (x, y, z) R3 and (1 , 2 , 3 ) R3 . For any xed P R3 , p(P ; )
is a nontrivial quadratic polynomial in R3 . Therefore, if f is an analytic
function near P , we can always nd an analytic solution of Lu = f near P .
In fact, we can always nd an analytic hypersurface containing P which is
noncharacteristic at P . Then by prescribing analytic initial values on this
hypersurface, we can solve Lu = f by the Cauchy-Kovalevskaya theorem.
Theorem 7.3.1 illustrates that the analyticity of the nonhomogeneous term
f is necessary in solving Lu = f even for local solutions.
We rst construct a dierential equation which does not admit solutions
near a given point.
Lemma 7.3.2. Let (x0 , y0 , z0 ) be a point in R3 and L be the dierential op-
erator dened in (7.3.1). Suppose h = h(z) is a real-valued smooth function
in z R that is not analytic at z0 . Then there exist no C 1 -solutions of the
equation
Lu = h (z 2y0 x + 2x0 y)
in any neighborhood of (x0 , y0 , z0 ).

Proof. We rst consider the special case x0 = y0 = 0 and prove it by


contradiction. Suppose there exists a C 1 -solution u of
Lu = h (z)
272 7. First-Order Dierential Systems

in a neighborhood of (0, 0, z0 ), say


= BR (0) (z0 R, z0 + R) R2 R,
for some R > 0. Set

v(r, , z) = ei ru( r cos , r sin , z).
As a function of (r, , z), v is C 1 in (0, R) R (z0 R, z0 + R) and is
continuous at r = 0 with v(0, , z) = 0. Moreover, v is 2-periodic in . A
straightforward calculation yields
i
Lu = 2vr + v 2ivz = h (z).
r
Consider the function
 2
V (r, z) = v(r, , z) d.
0
Then V is C1 in (r, z) (0, R) (z0 R, z0 + R), is continuous up to r = 0
with V (0, z) = 0, and satises
 
i 2 i
Vz + iVr = 2vr + v 2ivz d = ih (z).
2 0 r
Dene
W = V (r, z) ih(z).
Then W is C1 in (0, R) (z0 R, z0 + R), is continuous up to r = 0,
and satises Wz + iWr = 0. Thus W is an analytic function of z + ir for
(r, z) (0, R) (z0 R, z0 + R), continuous at r = 0, and has a vanishing
real part there. Hence we can extend W as an analytic function of z + ir to
(r, z) (R, R) (z0 R, z0 + R). Hence h(z), the imaginary part of
W (0, z), is real analytic for z (z0 R, z0 + R).
Now we consider the general case. Set
x = x x0 , y = y y0 , z = z 2y0 x + 2x0 y,
and
u(x, y, z) = u(x, y, z).
Then u(x, y, z) is C1 in a neighborhood of (0, 0, z0 ). A straightforward cal-
culation yields
ux + iuy 2i(x + iy)uz = h (z).
We now apply the special case we have just proved to u. 

In the following, we let h = h(z) be a real-valued periodic smooth func-


tion in R which is not real analytic at any z R. We take a sequence of
points Pk = (xk , yk , zk ) R3 which is dense in R3 and set
k = 2(|xk | + |yk |),
7.3. Nonexistence of Smooth Solutions 273

and
ck = 2k ek .
We also denote by  the collection of bounded innite sequences =
(a1 , a2 , ) of real numbers ai . This is a Banach space with respect to the
norm
  = sup |ak |.
k

For any = (a1 , a2 , )  , we set




(7.3.2) f (x, y, z) = ak ck h (z 2yk x + 2xk y) in R3 .
k=1

We note that f depends on linearly. This fact will be needed later on.
Lemma 7.3.3. Let f be dened as in (7.3.2) for some  . Then
f C (R3 ). Moreover, for any Z3+ ,
||
||
sup | f |

  sup |h(||+1) |.
R3 e R

Proof. We need to prove that all formal derivatives of f converge uniformly


in R3 . Set
Mk = sup |h(k) (z)|.
zR

Then Mk < since h is periodic. Hence for any Z3+ with || = m,


|ak ck h (z 2yk x + 2xk y)|   ck Mm+1 m
k
 m m
||
2k   Mm+1 k ek 2k   Mm+1 .
e
In the last inequality, we used the fact that the function f (r) = rm er
in [0, ) has a maximum mm em at r = m. This implies the uniform
convergence of the series for f . 

We introduce a Holder space which will be needed in the next result.


Let (0, 1) be a constant and Rn be a domain. We dene C 1, ()
as the collection of functions u C 1 () with
|u(x) u(y)| C|x y| for any x, y ,
where C is a positive constant. We dene the C 1, -norm in by
|u(x) u(y)|
|u|C 1, () = sup |u| + sup |u| + sup .
x,y,x=y |x y|
We will need the following important compactness property.
274 7. First-Order Dierential Systems

Lemma 7.3.4. Let be a domain in Rn , and (0, 1) and M > 0


be constants. Suppose {uk } is a sequence of functions in C 1, () with
uk C 1, () M for any k. Then there exist a function u C 1, () and
a subsequence {uk } such that uk u in C 1 ( ) for any bounded subset 
with  and uC 1, () M .

Proof. We note that a uniform bound of C 1, -norms of uk implies that uk


and their rst derivatives are equibounded and equicontinuous in . Hence,
the desired result follows easily from Arzelas theorem. 

We point out that the limit is a C 1, -function, although the convergence


is only in C 1 .
Next, we set
Bk,m = B 1 (Pk ).
m

We x a constant (0, 1).


Denition 7.3.5. For positive integers m and k, we denote by Ek,m the
collection of  such that there exists a solution u C 1, (Bk,m ) of
Lu = f in Bk,m ,
with
u(Pk ) = 0, |u|C 1, (Bk,m ) m,
where f is the function dened in (7.3.2).

We have the following result concerning Ek,m .


Lemma 7.3.6. For any positive integers k and m, Ek,m is closed and
nowhere dense in  .

We recall that a subset is nowhere dense if it has no interior points.

Proof. We rst prove that Ek,m is closed. Take any 1 , 2 , Ek,m and
 such that
lim j  = 0.
j

By Lemma 7.3.3, we have


sup |fj f | j  sup |h |.
R3 R

For each j, let uj C 1, (Bk,m ) be as in Denition 7.3.5 for fj , i.e., Luj = fj


in Bk,m , uj (Pk ) = 0 and
|uj |C 1, (Bk,m ) m.
7.3. Nonexistence of Smooth Solutions 275

By Lemma 7.3.4, there exist a u C 1, (Bk,m ) and a subsequence {uj  } such


that uj  converges uniformly to u together with its rst derivatives in any
compact subset of Bk,m . Then, Lu = f in Bk,m , u(Pk ) = 0 and

|u|C 1, (Bk,m ) m.

Hence Ek,m . This shows that Ek,m is closed.


Next, we prove that Ek,m has no interior points. To do this, we rst
denote by  the bounded sequence all of whose elements are zero,
except the kth element, which is given by 1/ck . By (7.3.2), we have f =
h (z 2yk x+2xk y). By Lemma 7.3.2, there exist no C 1 -solutions of Lu = f
in any neighborhood of Pk .
For any Ek,m , we claim that

+
/ Ek,m ,

for any . We will prove this by contradiction. Suppose + Ek,m


for some . Set = + and let u and u be solutions of Lu = f and
Lu = f , respectively, as in Denition 7.3.5. Set v = (u u)/. Then v is a
C 1, -solution of Lv = f in Bk,m . This leads to a contradiction, for || can
be arbitrarily small. 

Now we are ready to prove Theorem 7.3.1.

Proof of Theorem 7.3.1. Let (0, 1) be the constant as in the deni-


tion of Ek,m . We will prove that for some  , the equation Lu = f
admits no C 1, -solutions in any domain R3 . If not, then for every
 there exist an open set R3 and a u C 1, ( ) such that

Lu = f in .

By the density of {Pk } in R3 , there exists a Pk for some k 1. Then


Bk,m for all suciently large m. Next, we may assume u(Pk ) = 0.
Otherwise, we replace u by u u(Pk ). Then, for m suciently large, we
have
|u|C 1, (Bk,m ) m.
This implies Ek,m . Hence

&
 = Ek,m .
k,m=1

Therefore, the Banach space  is a union of a countable set of closed


nowhere dense subsets. This contradicts the Baire category theorem. 
276 7. First-Order Dierential Systems

7.4. Exercises
Exercise 7.1. Classify the following 4th-order equation in R3 :

2x4 u + 2x2 y2 u + y4 u 2x2 z2 u + z4 u = f.

Exercise 7.2. Prove Lemma 7.2.7 and Lemma 7.2.8.

Exercise 7.3. Consider the initial-value problem


utt uxx u = 0 in R (0, ),
u(x, 0) = x, ut (x, 0) = x.
Find a solution as a power series expansion about the origin and identify
this solution.

Exercise 7.4. Let A be an N N diagonal C 1 -matrix on R (0, T ) and


f : R (0, T ) RN RN be a C 2 -function. Consider the initial-value
problem for u : R (0, T ) RN of the form
ut + A(x, t)ux = f (x, t, u) in R (0, T ),

with
u(, 0) = 0 on R.
Under appropriate conditions on f , prove that the above initial-value prob-
lem admits a C 1 -solution by using the contraction mapping principle.
Hint: It may be helpful to write it as a system of equations instead of using
a matrix form.

Exercise 7.5. Set D = {(x, t) : x > 0, t > 0} R2 and let a be C 1 , bij


be continuous in D, and , be continuous in [0, ) with (0) = (0).
Suppose (u, v) C 1 (D) C(D) is a solution of the problem

ut + aux + b11 u + b12 v = f,


vx + b12 u + b22 v = g,
with
u(x, 0) = (x) for x > 0 and v(0, t) = (t) for t > 0.
(1) Assume a(0, t) 0 for any t > 0. Derive an energy estimate for
(u, v) in an appropriate domain in D.
(2) Assume a(0, t) 0 for any t > 0. For any T > 0, derive an estimate
for sup[0,T ] |u(0, )| in terms of sup-norms of f, g, and .
(3) Discuss whether similar estimates can be derived if a(0, t) is positive
for some t > 0.
7.4. Exercises 277

Exercise 7.6. Let a, bij be analytic in a neighborhood of 0 R2 and ,


be analytic in a neighborhood of 0 R. In a neighborhood of the origin in
R2 = {(x, t)}, consider
ut + aux + b11 u + b12 v = f,
vx + b12 u + b22 v = g,
with the condition
u(x, 0) = (x) and v(0, t) = (t).
(1) Let (u, v) be a smooth solution in a neighborhood of the origin.
Prove that all derivatives of u and v at 0 can be expressed in terms
of those of a, bij , f, g, and at 0.
(2) Prove that there exists an analytic solution (u, v) in a neighborhood
of 0 R2 .
Chapter 8

Epilogue

In the nal chapter of this book, we present a li