Professional Documents
Culture Documents
(Graduate Studies in Mathematics 120) Qing Han-A Basic Course in Partial Differential Equations-American Mathematical Society (2011) PDF
(Graduate Studies in Mathematics 120) Qing Han-A Basic Course in Partial Differential Equations-American Mathematical Society (2011) PDF
Partial Differential
'Equations
Qing Han
Graduate -Stijd'iaes,
in Math ati cs.
Volume 120
A Basic Course in
Partial Differential
Equations
A Basic Course in
Partial Differential
Equations
Qing Han
Graduate Studies
in Mathematics
Volume 120
EDITORIAL COMMITTEE
David Cox (Chair)
Rafe Mazzeo
Martin Scharlemann
Gigliola Stafl'ilani
2000 Mathematics Subject Classification. Primary 35-01.
www.ams.org/bookpages/gsm- 120
2010
515'. 353-dc22
2010043189
Copying and reprinting. Individual readers of this publication, and nonprofit 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.
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.
0 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/
10987654321
1615 14131211
Contents
Preface
ix
Chapter 1. Introduction
1.1. Notation
1.2. Well-Posed Problems
1.3.
Overview
Exercises
Classifications
Energy Estimates
3.3. Separation of Variables
3.2.
3.4.
Exercises
Exercises
1
1
3
5
9
10
16
30
43
47
48
58
67
86
89
90
105
112
133
143
vii
Contents
viii
Exercises
Exercises
Exercises
147
148
158
175
197
201
202
213
237
245
249
250
259
270
276
Chapter 8. Epilogue
8.1. Basic Linear Differential Equations
8.2. Examples of Nonlinear Differential Equations
279
Bibliography
289
Index
291
279
282
Preface
just start studying PDEs might ask. Students may find 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
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 secondorder PDEs at the beginning. In the chapter on general second-order linear
PDEs, immediately after classifying second-order PDEs into elliptic, parabolic and hyperbolic type, we discuss various boundary-value problems and
initial/boundary-value problems for the Laplace equation, the heat equation
ix
Preface
and the wave equation. We discuss energy methods for proving uniqueness
and find solutions in the plane by separation of variables. The explicit
expressions of solutions demonstrate different properties of solutions of different types of PDEs. Such differences clearly indicate that there is unlikely
to be a unified 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 different 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 different classes
of partial differential equations. The choice of topics is influenced by the
personal tastes of the author. Some topics may not be viewed as basic by
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
1.1. Notation
In general, we denote by x points in IRn and write x = (Xi,'. , x7) in terms
of its coordinates. For any x e IRn, we denote by lxi the standard Euclidean
norm, unless otherwise stated. Namely, for any x = (Xi,'' , x7 ), we have
1
Sometimes, we need to distinguish one particular direction as the time direction and write points in IRn+1 as (x, t) for x e IRn and t e JR. In this
case, we call x = (xi,'' , x7) E IRn the space variable and t e IR the time
variable. In R2, we also denote points by (x, y).
1. Introduction
Vu=(u51, ,u5).
This is the gradient vector of u. For second-order derivatives, we write V2u
in the matrix form
V2u =
ux1x1
ux2x1
uxlx2
uxlxn
ux2x2
ux2xn
uxnxl
uxnx2
uxnxn
IaI=a.
2=1
= (1,
,fin ) C W, we denote
- tai
The partial derivative &u is defined by
as u = aSi . .. axn u,
and its order is I al. For any positive integer m, we define
1
2)
\IaI=m
In particular,
1
(Eu)
2=1
and
1
IV2ul =
uxix
i, j=1
A hypersurface in W is a surface of dimension n - 1. Locally, a Cnhypersurface can be expressed by {co = 0} for a Cn-function co with Vco
0.
Several PDEs involving one or more unknown functions and their deriva-
1. Introduction
We usually refer to (i), (ii) and (iii) as the existence, uniqueness and
continuous dependence, respectively. We need to emphasize that the wellposedness goes beyond the existence and uniqueness of solutions. The continuous 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 solutions depend on data in a controllable way.
1.3. Overview
For a bounded continuous function u in St, we may also write IuIc(c) instead
of IuILoo(c). Let m be a positive integer.
(f
where integration is in the Lebesgue sense. The L2-space in 1 is the collecIIUIIL2() -
u2 dx
f uxZv dx = -
uvx2 dx +
fsp
uvv2 dS,
1.3. Overview
There are eight chapters in this book.
The main topic in Chapter 2 is first-order PDEs. In Section 2.1, we introduce the basic notion of noncharacteristic hypersurfaces for initial-value
problems for first-order PDEs. We discuss first-order linear PDEs, quasilinear PDEs and general nonlinear PDEs. In Section 2.2, we solve initial-value
1. Introduction
tion, the heat equation and the wave equation. We also introduce their
general forms, elliptic equations, parabolic equations and hyperbolic equations, 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 nonhomogeneous terms. In Section 3.3, we solve these boundary-value problems in
the plane by separation of variables. Our main focus is to demonstrate different regularity patterns for solutions of different differential equations, the
Laplace equation, the heat equation and the wave equation.
In Chapter 4, we discuss the Laplace equation and the Poisson equation. The Laplace equation is probably the most important PDE with the
widest range of applications. In the first three sections, we study harmonic
functions (i.e., solutions of the Laplace equation), by three different methods: 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 harmonic 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 interior 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 Perron's method.
Last in Section 4.4, we briefly discuss classical solutions and weak solutions
of the Poisson equation.
In Chapter 5, we study the heat equation, which describes the temperature of a body conducting heat, when the density is constant. In Section
5.1, we introduce Fourier transforms briefly and derive formally an explicit
1.3. Overview
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 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 initial-value problems and
various initial/boundary-value problems for the one-dimensional wave equation. 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 differential systems of first order and
focus on existence of local solutions. In Section 7.1, we introduce noncharacteristic hypersurfaces for partial differential equations and systems of
arbitrary order. We demonstrate that partial differential systems of arbitrary order can always be changed to those of first order. In Section 7.2,
we discuss the Cauchy-Kovalevskaya theorem, which asserts the existence of
analytic solutions of noncharacteristic initial-value problems for differential
systems if all data are analytic. In Section 7.3, we construct a first-order
linear differential system in R3 which does not admit smooth solutions in
any subsets of ]R3. In this system, coefficient matrices are analytic and the
nonhomogeneous term is a suitably chosen smooth function.
In Chapter 8, we discuss several differential 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 differential equations,
including elliptic, parabolic and hyperbolic equations, and first-order linear
symmetric hyperbolic differential systems. We will introduce appropriate
boundary-value problems and initial-value problems and introduce appropriate 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.
1. Introduction
Each chapter, except this introduction and the final chapter, ends with
exercises. Level of difficulty varies considerably. Some exercises, at the most
difficult level, may require long lasting efforts.
Chapter 2
First-Order Differential
Equations
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 characteristics if initial values are prescribed on noncharacteristic hypersurfaces.
For first-order homogeneous linear PDEs, special curves are introduced along
which solutions are constant. These curves are given by solutions of a system
of ordinary differential equations (ODEs), the so-called characteristic ODEs.
For nonlinear PDEs, characteristic ODEs also include additional equations
10
(2.1.1)
u + ut = 0.
This is probably the simplest first-order PDE. Obviously, u(x, t) = x - t
is a solution. In general, u(x, t) = uo(x - t) is also a solution for any C'function up. Such a solution has a physical interpretation. We note that
u(x, t) = uo (x - t) is constant along straight lines x - t = xo. 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 uo(x).
The function uo is called an initial value.
U
u(',ti) u(',t2)
t2-tl
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 E be a smooth hypersurface in I[8Th with St fl E Ql. We intend to
prescribe u on E to find a solution of (2.1.1). To be specific, let uo be a given
smooth function on E. We will find a solution u of (2.1.1) also satisfying
u = uo on.
(2.1.2)
We usually call E the initial hypersurface and uo the initial vale or Cauchy
vale. 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 (p.1.1), can we compute all derivatives of u at each point of the
initial hypersurface E? This should be easier than solving the initial-value
problem (2.1.1)-(2.1.2).
To illustrate the main ideas, we first consider linear PDEs. Let S2 be a
domain in I[8n containing the origin and ai, b and f be smooth functions in
, n. We consider
St, for any i = 1,
n
(2.1.3)
i=1
is given
by the hyperplane {xn = 0}. For x E Rn, we write x = (x', xn) for x' =
x_1) E Rn-1. Let uo be a given smooth function in a neighborhood
of the origin in Rn-1. The initial condition (2.1.2) has the form
(x1,
(2.1.4)
u(x', 0) =
terms of the equation and the initial value. It is obvious that we can find
all x'-derivatives of u at the origin in terms of those of uo. In particular, we
have, for i = 1,
, n - 1,
u(0) =
To find u(0), we need to use the equation. We note that an is the coefficient
an(0) L 0,
12
then by (2.1.3)
n-1
u(0) _ -
for i, j = 1,
,n-i. To find uxkxn for k = 1,
with respect to xk to get
, n, we differentiate (2.1.3)
aiuxix
i=1
-I-
i=1
matrix J is given by
... ,y) _
a(,...
J = a(xl,
... , xn)
Hence det J(0) = cpXTh (0)
0.
Id
13
In the following, we denote by L the first-order linear differential operator defined by the left-hand side of (2.1.3), i.e.,
ai(x)u-I- b(x)u.
Lu =
(2.1.6)
i=1
>yk,xuyk.
lc=1
Lu =
a2(x(y))yk,xi
uyk + b(x(y))u.
i=1
Hence, for the initial-value problem (2.1.3) and (2.1.2), we can find all deriva-
tives ofuatOEEif
0.
a2 Co) z Co)
2=1
a(x).
2=1
>ai(xo)vi
0,
2=1
14
, n. We consider
n
(2.1.8)
f(x,u) in ft
ai (x,
2=1
u( O) =
for i = 1,
uo,xi (O) ,
, n - 1, and
n-1
- f(Ouo(O)))
u(O) = -__________
a(O,uo(O))
z=1
aj(O,uo(O))uxj(o)
if
an (O,uo(0)) L0.
, n.
(2.1.9)
az (xo,uo(xo))vz
2=1
0,
15
F(x,u,Vu)=O inft
We ask the same question as for linear equations. Given an initial hypersurface Econtaining the origin and an initial value uo on E, can we compute
all derivatives of solutions at the origin? Again, we first consider a special
case where the initial hypersurface E is given by the hyperplane {xn = 0}
and an initial value is given by (2.1.4) for a given smooth function uo in a
neighborhood of the origin in
Il8n-1
n
1
i=1
and
u(x', 0) = u0(x').
,n-
1.
However, if I V ' u0 (x') l 2 > 1, there are no solutions for such an initial value.
F(0,v(0),Vv(0)) L 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) _ 1p12 - 1 for any p E I[8n.
16
By choosing
= IV'uo(O)I2 + c2.
1- Ivx'uo(o)12 o,
v satisfies the equation at x = 0. For such two choices of v, we have
0.
0,
i=1
at xo.
F(x,u,Vu)=0 foranyxES1.
Let
on
f1 S 1.
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 field. If
17
x E W, we write x = (x', xn), where x' E R-1. Our goal is to solve the
initial-value problem
F(x, u, emu) = 0,
u(x',O) = UO(X).
0,
2=1
u(x', 0) = uo(x').
for asmall y E I[8n-i, then u(x) = uo(y). To find such acurve x = x(s), we
consider the restriction of u to it and obtain aone-variable function u(x(s)).
Now we calculate the s-derivative of this function and obtain
i=1
(u(x(s))) =0.
,n.
18
(2.2.3)
x(0) = x.
ds =
(2.2.4)
a(x),
x(0) = (y,O).
In (2.2.4), integral curves start from (y, 0). By allowing y E ][8n-1 to vary in
a neighborhood of the origin, we expect integral curves x(y, s) to reach any
x E IIBn in a neighborhood of the origin for small s. This is confirmed by the
following result.
Proof. This follows easily from the implicit function theorem. By standard
results in ordinary differential equations, (2.2.4) admits a smooth solution
x = x(y, s) for any sufficiently small (y, s) E IIBn-1 X IIg. We treat it as a
map (y, s) i- x and calculate its Jacobian matrix J at (y, s) _ (0, 0). By
x(y, 0) _ (y, 0), we have
ai(0)
.Tfnl -
ax
Id
ate,-i (0)
(y,s)=(0,0)
0
a(0)
x(y, s) = x
19
uniquely for small y and s. Then u(x) = uo(y) yields a solution of (2.2.1).
Note that s is not present in the expression of solutions. Hence the value
of the solution u(x) depends only on the initial value uo at (y, 0) and, meanwhile, the initial value uo at (y, 0) influences 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 influence 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 Definition 2.1.2. Hence the ODE (2.2.2) is often
of the origin. A natural question here is whether there exists a global solution for globally defined a and uo. There are several reasons that local
solutions cannot be extended globally. First, u(x) cannot be evaluated at
x E I[8n if x is not on an integral curve from the initial hypersurface, or equivalently, the integral curve from x does not intersect the initial hypersurface.
20
1'
ds
1'
and
t(0) = 0.
Here, both x and t are treated as functions of s. Hence
x(0) = XO,
x=s+xp, t=s.
By eliminating s, we have
x - t=xo.
This is a straight line containing (xo, 0) and with a slope 1. Along this
straight line, u is constant. Hence
u(x,t) = uo(x - t).
We interpreted the fact that u is constant along the straight line x - t = xo
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 ll82.
1, we consider
(2.2.5)
u(x', 0) = uo(x').
an(0,uo(0)) 0.
Suppose (2.2.5) admits a smooth solution u. We first examine integral curves
dx
= a (x, u),
ds
x(0) _ (y, 0),
21
dx2
2Gxi
Z=1
ds
Then
du
= f(x,u),
ds
u(0) = uo(y)
x = x(y,s), u = (y,s).
As in the proof of Lemma 2.2.2, we can prove that the map (y, s) H x is a
diffeomorphism. Hence, for any x e ]E8sufficiently small, there exist unique
y e IEBn-1 and s e ]E8 sufficiently small such that
x=
S)
u(x) = (V' S)
We now consider an initial-value problem for a nonhomogeneous linear
equation.
ut - u = f
u(.,0)=uo
onJR.
22
t(O) = 0,
x(O) = XO,
u(0) = UO(XO).
x=xp - s, t =s.
Then the equation for u can be written as
du
f(xo - s, s).
- T, T)
2G = 2Gp(xp) + f
ft
u(x,t) = 2Gp(x + t)'+
J0
f(x+t - T,T)CLT.
ut -I- uu = 0
0) = uo
in It x (0, oo),
on I1t
= u,
ds
dt
= 1,
ds
du
= 0,
ds
and
X = UO(XO)5 + XO,
t = s,
u = uo(xo).
23
xo = xo(x,t),
then we have a solution
u = up(xp(x,t)),
for any (x, t) sufficiently small. By eliminating xo and s from the expressions
of x, t and u, we may also write the solution u implicitly by
u = UO(x - Ut).
uo(xl).
xO
I
x1
uo (x) = -x.
Obviously, this is strictly decreasing. In this case, coo in (2.2.6) is given by
x=xo- xot,
and the solution on this line is given by u = -xo. We note that each coo
contains the point (x, t) _ (0, 1) and hence any two characteristic curves
24
u(x, t) = t x
1 - t
where F :
I[8
of compact support in I[8 x (0, oo) and integrating by parts the product of cp
and the equation in (2.2.7), we obtain
(2.2.8)
JRx (o,oo)
(ucpt +
dxdt = 0.
25
Now we return to our study of initial-value problems of general firstorder PDEs. So far in our discussion, initial values are prescribed on noncharacteristic 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 uo. Then
an(O,uo(O)) = 0.
(0,u(0(0) = f(O,uo(O)).
(2.2.9)
Z=1
This is the compatibility condition for the initial value uo. 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 firstorder nonlinear PDEs. Let St C Il8n be a domain containing the origin and
F = F(x, u, p) be a smooth function in (x, u, p) E St x I[8 x W. Consider
(2.2.10)
F(x, u, Du) = 0,
u(x', 0) = wo(x ),
for any x' with (x', 0) E S2. Assume there is a scalar ao such that
F(0,uo(0),V'uo(0),ao) = 0.
The noncharacteristic condition with respect to uo and ao is given by
(2.2.12)
Fp(0, uo(0),
ao) # 0.
By (2.2.12) and the implicit function theorem, there exists a smooth function
such that a(0) = ao and
a(x') in a neighborhood of the origin in
(2.2.13)
F(x', 0, uo(x'),
a(x')) = 0,
for any x' E Il8n-1 sufficiently small. In the following, we will seek a solution
of (2.2.10)-(2.2.11) and
urn (x', O) =
26
p2 = uxi
for i = 1,
, n.
Then
(2.2.15)
F(x1,...
, xn, u, p1,
... , pn) = 0.
By
we obtain
n
F p2 xj = -Fx i - Faux i
(2.2.16)
1roj
for i = 1,
, n.
j=1
We view (2.2.16) as a first-order quasilinear equation for p2, for each fixed
i = 1,
dxj
for,-1,
ds
dp2
n,
- Fup2 - Fxi .
ds
We also have
du
ds
n
ux j
j=1
dx j
ds
pj F3.
j=1
ds
du
ds
- Fpj (x, u, p)
for j = 1,
, n,
pj Fpj (x, u, p)
j=1
for i = 1, , n,
27
fori=l
n-1
u_ P(y,S), p = p(y,S),
for any y and s sufficiently small. We will prove that the map (y, s) F- x
is a difFeomorphism near the origin in W. Hence for any given x near the
origin, there exist unique y e Rn-1 and s E IE8 such that
x = x().
Then we define u by
mi(x) _
s)Theorem
We should note that this solution u depends on the choice of the scalar
ao and the function a(xe).
Proof. The proof consists of several steps.
Step 1. The map (y, s) H x is a diffeomorphism 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) H x at (0,0) is given by
"'
ax
D(y,s) y=0,s=0
Where
0)
= F(0, u(0),
V'uo(0), ao).
28
x = x(y, s) uniquely for y E I[8"-1 and s E ll8 sufficiently small. Then define
for i =
1,...
, n.
for any y and s sufficiently small. Denote by f(s) the function in the lefthand side. Then by (2.2.18)
f(0) = F(y,O,uo(y),Vxiuo(y),a(y)) = 0.
Next, we have by (2.2.17)
df (S)
ds
ds
F(x(y,S),(P(y,S),p(y,S))
du
dpj
+
Fu
ds + :i:
ds
._1 Fps ds
dx2
Fx2
z=1
Fp j (- Fupj - F'x j) = 0.
pj Fp j +
Fx2 Fpi + Fu
j=1
j=1
i=1
Hence f(s) .= 0.
Step 3. We claim that
pz (y, s) = ux2 (x (y, s))
fori =
1,...
for i =
, n.
for i = 1,
du
(2.2.19) 0 =
ds
or
pjFp, =
j=1
ux
j=1
dxj
' ds
- pj Fp
n
O.
j=1
, n -1,
and Fp, =0
29
ds
, n, i.e.,
for i = 1,
_1
, n,
uxixj
.-1
dx j
ds
dpi
ds
._1
pjxi) +
Fpj (uxixj
j=1
(Fpj )xiwj = 0.
j=1
p xZ
p7 x2
j=1
u z
x2
j=1
By Step 2,
Fxi + Fuuxi +
Fpj pj,xi = 0.
j=1
Hence
dwi
ds
_ - Fx2 - FuuxZ -
_1
_ - Fuwi -
(Fpj )xiwj
j=1
or
dwi
ds
(FuSZ + (F))w.
j=1
To end this section, we briefly compare methods we used to solve firstorder linear or quasi-linear PDEs and general first-order nonlinear PDEs. In
solving a first-order quasi-linear PDE, we formulate an ordinary differential
30
system of n + 1 equations for n + 1 functions x and u. For a general firstorder nonlinear PDE, the corresponding ordinary differential system consists
of 2n + 1 equations for 2n +1 functions x, u and Du. Here, we need to take
into account the gradient of u by adding n more equations for Du. In other
words, we regard our first-order nonlinear PDE as a relation for (u, p) with
dt
-/3u=f(t).
J0
[o,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 [O, T].
Now we turn to PDEs. For convenience, we work in I[8 x (0, oo) and
denote points by (x, t), with x E ][8n and t E (0, oo). In many applications,
we interpret x as the space variable and t the time variable.
31
(2.3.1)
ut +
(2.3.2)
t)u
az(x,
t)
i=1
2 (x, O) = Up(x)
in W.
{(y,s): i'y
s} c W x R.
For any point P = (X, T) E ]E8n x (0, oo), consider the cone Ck(P)
(opening downward) with vertex at P defined by
Ck(P) _ {(x, t) : 0 < t < T, IcIx - X I< T - t}.
We denote by 83Ck(P) and 8_Ck(P) the side and bottom of the boundary,
respectively, i.e.,
T}.
32
We note that 8_Ck(P) is simply the closed ball in ]E8' x {0} centered at
(X,O) with radius T/rc. For any (x,t) E 83Ck(P), let a(x, t) be a vector
in I[8" satisfying (2.3.1). Then the vector (a(x, t),1), if positioned at (x, t),
points outward from the cone Ck(P) or along the boundary 83Ck(P). Hence
for a function defined only in Ck (P), it makes sense to calculate ut + a Vu
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.
Now we calculate the unit outward normal vector of 83Ck(P) \ {P}. Set
o(x,t)= klx-XI-(T-t).
Obviously, 83Ck(P) \ {P} is a part of {cp = 0}. Then for any (x, t) E
aSCk(P) \ {P},
= (V, t) = (k x-X
i Xv i).
x-X
Ix-XI'
For n = 1, the cone Ck(P) is a triangle bounded by the straight lines
X) = T - t and t = 0. The side of the cone consists of two line
U
k2+1
segments,
and
33
It is easy to see that the integral curve associated with (2.3.2) starting from P and going to the initial hypersurface I[8" x {0} stays in Ck(P).
In fact, this is true for any point (x, t) E Ck(P). This suggests that solutions in Ck(P) should depend only on f in Ck(P) and the initial value uo
on 8_Ck(P). The following result, proved by a maximum principle type
argument, confirms this.
Theorem 2.3.1. Let ai, b and f be continuous functions in I[8" x [0, oo)
satisfying (2.3.1) and ua be a continuous function in R. Suppose u E
Cl(I[8" x (0,oo)) fl C(][87 ` x [0,oo)) is a solution of (2.3.2). Then for any
P = (X,T) E lI8" x (0,oo),
sup
Sup
M = sup
iuoi,
a_ c,c(P)
F = sup
C,( P)
We will prove
w(x, t) =
t) - M - tF.
34
wt +
2=1
wt + a
0.
First, it is obvious if (xO, to) E 8_Ck(P), since w(xo, to) = uo(xo)-M < 0 by
the definition of M. If (xO, to) E Ck(P), i.e., (xO, to) is an interior maximum
point, then
(Wt + a Vxw)I(0,t0) = 0.
If (xO, to) E 83Ck(P), by the position of the vector (a(xo, to), 1) relative to
the cone Ck(P), we can take the directional derivative along (a(xo, to), 1),
obtaining
(Wt + a Vxw)l(0,t0) 0.
Hence, in both cases, we obtain
(b + fi')wI(0,t0)
0.
Since b + /3' > 0, this implies w(xo, to) < 0. (We need the positivity of b + /3'
here!) Hence w(xo, to) < 0 in all three cases. Therefore, w < 0 in Ck(P), or
if w assumes its maximum at (xO, to) E DSCK (P) . To prove this, we first note
that at +a and at - ate are directional derivatives along the straight lines
t - to = ic(x - xo) and t - to = -ic(x - xo), respectively. Since w assumes
its maximum at (xO, to), we have
1
>
)II(xo,to)
/
- 0, (wt--w)I
1
> 0.
(xo,to)
35
Corollary 2.3.2. Let a2, b and f be continuous functions in I[8n x [0, oo)
satisfying (2.3.1) and uo be a continuous function in W. Then there exists
at most one Cl(Il8" x (0, oo)) fl C(W1 x [0, oo))-solution of (2.3.2).
ing f and u01iuo2 replacing uo, respectively. Then for any P = (X, T) E
R x (0,oo),
sup l- u2)I C
ck(p)
sup
a-ck(p)
b>-3
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,c(P) and uo on 8_Ck(P). Hence
36
n
Figure 2.3.4. The range of influence.
uo dx + f
e-"t f 2 dxdt,
, (p)
, (p)
= (e-tu2)t + ae-tu2,
(e-tu2)t +
n
ai xZ
u2 = 2e-tuf
i=1
i=1
e-t
sc,(P)
vt +
ai vi
u2 dS
i=1
-}- L
e-t
uo dx +
a -}- 2b -
Jc,(P)
u2 dxdt
2e-tu f dxdt,
37
/Icii,lB
(lJx,lJt)=(lJ1.,lJn,lJt)=j2
(2.3.1) and the Cauchy inequality, we have
\/1+
and hence
vt +
ai v2 > 0
on DSCk (P) .
i=1
a + 2b -
> 2 in Ck (P) .
i=1
Then
e-atu2 dxdt <
2C
,c (P)
uo dx +-
2e-atu
f dxdt.
,c (P)
Here we simply dropped the integral over DSCk (P) since it is nonnegative.
The Cauchy inequality implies
e-atu2 dxdt +-
C,c(P)
e-at f2 dxdt.
CA(P)
We note that the estimate in Theorem 2.3.4 is similar to that in Theorem 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 influence using Theorem 2.3.4.
We now derive an L2-estimate of solutions in the entire space.
38
e-tu2 dx -Ix{T}
<
e-tu2 dxdt
x (O,T)
e-t f2 dxdt,
uo dx -IR x (O,T)
]f8n
Proof. We first take ic> 0 such that (2.3.1) holds. Take any t > T and
consider
D(O=Ck(0,t)fl{0<t<T}.
We denote by 8_D(E), 83D() and 8+D(t) the bottom, the side and the top
of the boundary, i.e.,
We now proceed as in the proof of Theorem 2.3.4, with D() replacing Ck(P).
39
e-atu2 dxdt
up dx +
<f
e-at f2 dxdt.
We note that t enters this estimate only through the domain D(). Hence,
we may let t -+ oo to get the desired result.
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., uo = 0.
To introduce the notion of a weak solution, we fix a T> 0 and consider
functions in Rn x (0, T). Set
n
(2.3.3)
Lu = ut +
in Rn x (0, T).
+ bu
i=1
vLu=ul -vt-
+ (uv)t +
by
(aiuv)x.
i=1
z=
i=1
L*v = -yt -
by = -yt i=1
b-
Then
n
(auv)x.
ai,xi
i=1
i=1
y.
40
We now require that u and v vanish for large x. Then by a simple integration
in Ian x (0, T), we obtain
vLu dxdt =
IISn x (O,T)
uL*v dxdt
IISn x (O,T)
+I
uv dx - /
uv dx.
JII8x{t=0}
uL*v dxdt =
x (O,T)
JIItT x (O,T)
fvdxdt,
Jllt x (O,T)
uL*v dxdt =
JII8n x (O,T)
fvdxdt,
for any v e Cl (][8n x (0, T)) fl C(I[8Th x [0, T]) with v(x, t) = 0 for any (x, t)
u=0 onIfn.
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 first introduce some notation. We denote by
Co (I[8' x (0, T)) the collection of Cl-functions in ][8n x (0, T) with compact
support, and we denote by C( W x (O, T)) the collection of Cl-functions in
]E8n x (0, T) with compact support in x-directions. In other words, functions
41
in C( W x (O, T)) vanish for large x and for t close to 0 and T, and functions
in C( R x (O, T)) vanish only for large x.
We note that, with L in (2.3.3), we can rewrite the estimate in Theorem
2.3.5 as
IIUIIL2(unx(o,T))
C'(Ilu(', 0)IIL2(II$n) +
0) E
IkLIIL2(TEx(O,T))
for any u e Co (][8n x (O, T)) fl C(I[8n x [O, T)) with u = 0 on {t = 0}.
)L2(llnx(O,T))
IIVIIL2(nx(O,T)) <
for any v e C( R x (O, T)) fl C(I[8n x (O, T]) with v = 0 on {t = T}, where
C is a positive constant depending only on T, the C1-norms of ai and the
sup-norm of b in W x (O, T). We denote by Cl(][8n x (O, T)) the collection
of functions v e C( W x (O, T)) f1 C(I[89 x (O, T]) with v = 0 on {t = T}.
Consider the linear functional F : L*Cl(IEBn x (0, T)) -+ ][8 given by
F(L*v) =
for any v E C1 (W x (0, T)). We note that F acting on L*v in the left-hand
side is defined in terms of v itself in the right-hand side. Hence we need to
verify that such a definition makes sense. In other words, we need to prove
that L*vl = L*v2 implies
(f, vl) L2 (fin x (O,T))
for any v, v2 E C1(RTh x (0, T)). By linearity, it suffices to prove that L*v = 0
implies v = 0 for any v e Cl (][8n x (0, T) ). We note that it is a consequence
of (2.3.7). Hence, F is awell-defined linear functional on L*Cl(IEBn x (0, T)).
Moreover, by the Cauchy inequality and (2.3.7) again, we have
IF(L*v)I C If IIL2(TEThx(O,T))IIvIIL2(Ilx(O,T))
CII.fIIL2(TEnx(O,T))IIL*vIIL2(TE7 x(O,T))'
42
IIF1I
Here, IF II is the norm of the linear functional F on L2(1l8n x (0, T)). By the
Riesz representation theorem, there exists a u E L2(1[8x (O, T)) such that
for any w E L2(1[8n x (0, T)),
and
IIUIIL2(Rnx(OT)) = IIFII.
In particular, we have
F(L*v) = (u, L*v)L2(Rn x (O,T))
vLu dxdt +-
fv dxdt,
Rn x {t=0}
Rn x (O,T)
Rn x (O,T)
J1R'x (O,T)
vLu dxdt =
JIl8 x (O,T)
fv dxdt,
for any v E Co (1[8'x (O, T)). Since Co (I[8n x (O, T)) is dense in L2(118n x (O, T)),
we conclude that
Lu = f
Therefore,
fRx{t=O}
uv dx = 0,
for any v E Co (I[8x (0, T)) fl C(][8Th x [0, T]) with v = 0 on {t = T}. This
implies
0)cp dx = 0
2.4. Exercises
43
0) = 0 on Rn.
We note that a crucial step in passing from weak solutions to classical solutions 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 sufficient regularity is a classical
solution.
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 different 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 ][82:
44
Exercise 2.3. Let Bl be the unit disc in ][82 and a and b be continuous
functions in Bl with a(x, y)x + b(x, y)y > 0 on 8B1. Assume u is a C1solution of
Exercise 2.4. Find a smooth function a = a(x, y) in ][82 such that, for the
equation of the form
+ a(x, y)Ux = 0,
there does not exist any solution in the entire ][82 for any nonconstant initial
value prescribed on {y = 0}.
a=2.)
Exercise 2.7. In the plane, solve uy = 4u near the origin with u(x, 0) = x2
on {y = 0}.
2.4. Exercises
45
Exercise 2.8. In the plane, find two solutions of the initial-value problem
xux + yuy +
u(x,O) =
12 + uy) = u,
2
(1 - x2).
Exercise 2.9. In the plane, find two solutions of the initial-value problem
4u + uuy = u,
U I x,
1x2 _ -12
sup
c,c(p)
Iuol +
sup Ie_Qtf i,
a
inf
b
in ][8n x [0, oo) satisfying
(2.3.1) and uo be a Cl-function in ][8n. Suppose u is a CZ-solution of (2.3.2)
in Ilgn x [0, oo). Prove that, for any P = (X, T) E ][8n x (0, oo),
with
Chapter 3
An Overview of
Second-Order PDEs
47
48
its initial values, while the regularity of a solution of the wave equation is
similar to the regularity of its initial values. Such a difference in regularity
suggests that different methods are needed to study these two equations.
3.1. Classifications
The main focus in this section is the second-order linear PDEs. We proceed
as in Section 2.1.
bi and c be
Let S1 be a domain in W containing the origin and
continuous functions in S1, for i, j = 1,
, n. Consider a second-order
linear differential operator L defined by
n
(3.1.1)
Lu =
c(x)u
aij
i,j=1
in SZ.
i=1
u, respectively. We usually
Here a3, bi, c are called coefficients of
assume
, n. Hence, (a) is a symmetric matrix
for any i, j = 1,
=
in S1. For the operator L, we define its principal symbol by
n
Lu = f (x) in St.
Now
3.1. Classifications
49
for i = 1,... ,n - 1,
and
uxzx (0) = uo,xZx (0) for i, j
uxnxn (0) = ul,xi (0) for i = 1,
= 1,... ,n - 1,
,n - 1.
To compute uxnxn (0), we need to use the equation. We note that ann is the
coefficient of uxnxn in (3.1.2). If we assume
a(0) # 0,
(3.1.4)
then by (3.1.2)
1
(0)
(i,j)(n,n)
50
This is a well-defined transformation with a nonsingular Jacobian in a neighborhood of the origin. With
n
uxi =
yk,xi uyk
k=1
and
uxixj =
yk,xixj uyk
yk,xiY1,xj uykyl +
k=1
k,1=1
k,1=1
i,j=1
Lu =
k=1
i=1
bi yk,xi +
ai j yk,xix j
uyk + Cu.
i,j=1
aij
i,j=1
= o(x).
(3.1.5)
aij (xo)vivj
0,
i,j=1
at xo.
3.1. Classifications
51
aij(xO)SiSj
0+
i,j=1
for any
E Rn \ {0}.
(3.1.6)
Lu =
c(x)u = f(x)
-I-
z,j=1
in SZ.
i=1
52
For the operator L in (3.1.6), the symmetric matrix (a23) 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 different signs;
L is degenerate if at least one of the eigenvalues vanishes.
x2 = r sin 9,
Du = 0
is called the Laplace equation and its solutions are called harmonic functions.
uvy, uy=-vx.
Any such a pair gives an analytic function
3.1. Classifications
53
with initial values prescribed on {y = 0}. For any positive integer k, set
uk(X,y) =
sin(kx)ekY.
uk y(x, y) = sin(kx)e'y,
and hence
VuA(x, y)I2 =
Therefore,
and
y) +
y) = e2'y.
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 flu =
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 identified 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 - ux1x1.
= 0.
54
uX2X2 +
in 1102
lim
w (x),
x-+xo, xEf _
w+ (xo) =
lim
x-+xo, XES2+
w (x)
is called the jump of w at xo across F. The function w has a jump across I'
if it has a jump at every point of t across F. If w has a jump across I', then
[w] is awell-defined function on F. It is easy to see that [w] = 0 on I' if w
is continuous in ft
cu = f
+
i, j=1
in SZ \ F.
i=1
0,
3.1. Classifications
55
0.
(v2ax1 -
v2[uxlxl] vl
0.
on I'.
Thus, the nontrivial vector ([u11J, [u12J, [u22]) satisfies a 3 x 3 homogeneous linear system on F. Hence the coefficient matrix is singular. That is,
on F
v2
det
-111
v2
-vl I = 0,
al l 2aa1212
all
a22
or
LI
The Laplace operator, the wave operator and the heat operator can be
generalized to higher dimensions.
Du =
i=1
II2,
for any E TP. Obviously, O is elliptic. Note that 0 is invariant under
rotations. If x = Ay for an orthogonal matrix A, then
n
uxixi i=1
uyiyi
i=1
56
where v is the unit exterior normal vector to 11'. Upon integration by parts,
we obtain
divFdx=J
F = 0 in 1
F = -aVu,
for a positive constant a. Here the negative sign indicates that the flow is
from regions of higher temperature to those of lower one. Now a simple
substitution yields the Laplace equation
Du = div(Vu) = 0.
Lu=ut - Lu.
It is often called the n-dimensional heat operator. Its principal symbol is
given by
for any
characteristic for the heat operator if, at each of its points,
- IV 'I2 0.
Likewise, a hypersurface {(x, t) = 0} is characteristic if pxcp = 0 and cpt
0 at each of its points. For example, any horizontal hyperplane {t = to}, for
a fixed t0 E III, is characteristic.
The heat equation describes the evolution of heat. Let u denote a temperature in a domain 1Z C W with the flux density F. Then for any smooth
subdomain SZ' C 1, the rate of change of the total quantity in SZ' equals the
negative of the net flux of u through 811', i.e.,
d
u dx = -
dt
faQ'
where v is the unit exterior normal vector to 11'. Upon integration by parts,
we obtain
tJ
st
u dx = -
Js
div F dx.
This implies
ut = - div F
in 11,
3.1. Classifications
57
F = -aVu,
for a positive constant a. Now a simple substitution yields
ut = a div(Du) = aDu.
This is the heat equation if a = 1.
, xn, t).
The wave
p(x,t;S,T) = T2 ISI2
for any E Il8n and T E R. A hypersurface {p(xi,'.. , xn, t) = 0} is noncharacteristic for the wave operator if, at each of its points,
Pt
- VPL 0.
Ix - x12 = (t - to)2
is characteristic except at (moo, to). We note that this surface, smooth except
at (xO, to), is the union of two cones. It is usually called the characteristic
cone.
4udx= -
si
58
where F is the force acting on Q' through aQ' and the mass density is taken
to be 1. Upon integration by parts, we obtain
d2
dt2
u dx = SZ
div F dx.
SZ
This implies
utt = - div F in Q,
since Q' is arbitrary. In a special case F = -aDu for a positive constant a,
we have
ut -
a( x,
b2 (x,
t)u
t)
2=1
2=1
utt -
b2 (x,
aZj (x,
2=1
t)u = f(x,t)
2=1
is hyperbolic.
We start with the Laplace equation. Let Q C W be a bounded C1domain and (p be a continuous function on 9Q. Consider the Dirichlet
boundary-value problem for the Laplace equation:
Du = 0 in Q,
u = (p
on 9Q.
59
satisfies
Ow = 0
inn,
w=0 on aSZ.
We multiply the Laplace equation by w and write the resulting product as
n
- IVwI2.
0 =wow =
2=1
0=- J
IVwI 2 dx
si
asp
w av dS.
Du = f inn,
u = cp
on aSt.
Du = 0
inn,
au _
av
on aSZ.
bdS=O.
This can be seen easily upon integration by parts.
60
xn
u(xo, xn) =
xn
wn-1
Figure 3.2.1. An integration along l.
- a)
uxn (xo, s) ds
61
, nc
xo
, nc
xo
where Cxo is the length of lxo in ft Now a simple integration over xo yields
the desired result.
Now consider
Du = f
(3.2.1)
in S2,
u=0 on51.
Theorem 3.2.3. Let SZ C 1[8be a bounded C1 -domain and f be a continuous function in S2. Suppose u E C2(11) fl Cl(St) is a solution of (3.2.1).
Then
IIUIIL2(cl) + IIVUIIL2(cl)
cII f II L2(c),
Proof. Multiply the equation in (3.2.1) by u and write the resulting product
in the left-hand side as
n
(uu)xZ - I Du12.
2G02G =
i=1
fudS_
f IVuI2dx = fufdx.
av
t
fIvuI2dx
=-
fu f dx.
(L1vu12dx)2= (fufdx)2fu2dx.ff2dx.
By Lemma 3.2.2, we get
dx < (diam(1O)2ff2dx.
t
62
u(.,0) = uo
(3.2.2)
in S2,
u=0 onax(0,oo).
The geometric boundary of S2 x (0, oo) consists of three parts, S2 x {0},
8S2 x (0, oo) and aS2 x {0}. We treat S2 x {0} and aS2 x (0, oo) differently
and refer to values prescribed on S2 x {0} and 8SZ x (0, oo) as initial valves
and boundary values, respectively. Problems of this type are usually called
initial/boundary-value problems or mired problems. We note that u has a
homogeneous Dirichlet boundary value on 81Z x (0, oo). We now derive an
estimate of the L2-norm of a solution. For each t > 0, we denote by u(t)
the function defined on SZ by u(t) =
t).
Theorem 3.2.4. Let SZ be a bounded Cl-domain in IEgn, f be continuous in
St x [0, oo) and uo be continuous in St. Suppose u E C2(SZ x (0, oo)) f1 C1(1 2 x
are solutions of (3.2.2) with fl, uol and f2, u02 replacing f, uo, respectively.
Then for any t > 0,
t
I- 262(t)IIL2(SZ)
IVuI2.
i=1
f u2(t) dx +
fu2(t)dx+2fflVul2dxds = fudx+2fffudxds.
63
Set
E(t) = I
Js
fudxds.
2f(t)st)
t
'
= 2E(t)Il.f(t)IIz,2(st)
Hence
E'(t) c If(t)IIL2().
Integrating from 0 to t gives the desired estimate.
utt - 0u= f
in St x (0, oo),
(3.2.3)
u=0 onD1x(0,oo).
Comparing (3.2.3) with (3.2.2), we note that there is an extra initial condition on Ut in (3.2.3). This relates to the extra order of the t-derivative in
the wave equation.
t>0,
(IIUt(t)II2() + I
IIi2(sl))2
- (Ilullli2(sz) + Iloxuolli2(sl)) 2
t
0
and
As a consequence, we also have the uniqueness and continuous dependence on initial values in L2-norms.
64
Proof. Multiply the equation in (3.2.3) by Ut and write the resulting product in the left-hand side as
n
z=1
dx -
2 dt ,f but (t) +
Ut(t) aU (t)
ds = f f(t)ut(t) dx.
E(t) _
dx.
(mot (t) + I
If f - 0, then
Hence for any t > 0,
E(t) = E(0) =
ft (u + IVuoI2) dx.
ft
E(t) = E(0) + 2
Jo ffutdxds.
J(t) _ (E(t)) 2.
Then
t
(J(t))2 = (J(0))2 + 2
0
fut dxds.
2J(t)J'(t) =
J'(t)
IIf(t)11L2(Q).
J(t)
J(0)
65
This is the desired estimate for the energy. Next, to estimate the L2-norm
of u, we set
(F(t))2
= f u2(t) dx.
F'(t) IJ(0) + f
t
Ids.
Idsdt'.
f (u(t)
J Js
f 2 dads.
ft
f 2 dads.
equation and the heat equation by u and integrate the resulting product
over 1, while in the proof of Theorem 3.2.5, we multiply the wave equation
66
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 first-order linear PDEs in Section 2.3.
To end this section, we discuss an elliptic differential equation in the
entire space. Let f be a continuous function in TR. We consider
-Du + u = f
(3.2.4)
in ][8n.
uDu =
(uuxk)xk
-I
12
k=1
(uuxk)xk = fu.
We now integrate in TR. Since u and uxk decay sufficiently fast at infinity,
we have
f (JVuJ2+u2)dx=f
Rigorously, we need to integrate in BR and let R -+ oo after integrating by
parts. By the Cauchy inequality, we get
22n
fudx<fu2 dx + J
(2JVuJ2+u2)dx<fIl8
Hence, the L2-norm of f controls the L2-norms of u and Du. 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
(L\u)2
- 2uL\u + U2 = f2.
67
We note that
n
k,1=1
k,1=1
k,1=1
k,1=1
k,1=1
(Lu)2
n
n
)mil
k,1=1
Hence
I v2ul2
(uxkxkux3xl
+ 2 IVuI2 + u2+
k,1=1
k,1=1
= f2
-2
k=1
(3.2.5)
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 separation of variables to solve the Dirichlet problem for the Laplace equation
in the unit disc in ll82. We will use polar coordinates
x= r cos 8,
y= r sin B
in 1[82, and we will build up solutions from functions that depend only on
r and functions that depend only on B. Our first step is to determine all
harmonic functions u in ][82 having the form
u(r, B) = f(r)g(6),
68
Lu
-(rur)r + -uoo = 0.
(rf/(r))/9(e)
+ r f (r)9 (e) _ 0,
that is,
r f(r)g"(0) =0.
r (f/Il r) +
r
f(r)
r1)
g"(0)
g(0)
The left-hand side of this equation depends only on r and the right-hand
side depends only on 8. Thus there is a constant A such that
(f 11(r) + lf
f(r)
=A=
_g)
Hence
for 0 E 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
for f. The equation for g describes the eigenvalue problem for - de on 1.
This equation has nontrivial solutions when A = k2, k = 0,1, 2, . When
A = 0, the general solution is g(0) = ao, where ao is a constant. For A = k2,
k = 1, 2,
, the general solution is
g(0) = ak cos k0 + bk sin k0,
1
,
cos k0,
sin k0,
k = 1, 2,
form an orthonormal basis for L2 (S1) . In other words, for any v e L2 (S1),
00
v(0)=
ao -I-
k=i
where
ap =
69
J 1 v(9) d9,
and for k = 1, 2,
ak =
bk =
This series for v is its Fourier series and ao, ak, bk are its Fourier coefficients.
The series converges in L2 (S1) . Moreover,
1
1- a0 +
+ bk)
k=1
f(r)=co
is a constant function. For A = k2, k = 1, 2,
u(r, B) = ap,
or by
u(r, 8) = akrk cos k8 + bkr"' sin k8,
for k = 1, 2,
Remark 3.3.1. Note that r'' cos k8 and r'' sin k8 are homogenous harmonic
polynomials of degree k in JR2. Taking z = x + iy, we see that
r cos k8 + irk sin k8 = r1 eZ '0 = (x +
and hence
70
Now, we are ready to solve the Dirichlet problem for the Laplace equation in the unit disc B1 C R2. Let cp be a function on O`1B1 = S1 and
consider
Du = 0 in B1,
(3.3.1)
u = co
on S1.
u(r, 8) =
(3.3.2)
ao +
= j
k=1
The terms in the series are all harmonic functions of the form f(r)g(0) that
we discussed above. Thus the sum u(r, B) should also be harmonic. Letting
r = 1 in (3.3.2), we get
00
cp(6) = u(1, B) =
ao + 1
k=i
(3.3.3)
a0
and for k = 1, 2,
ak =
(3.3.4)
bk _
o(e)Los ke de,
Du = 0 in B1.
Moreover,
00.
71
First, we set
Iakrk cos k9 -}- bark sin k91.
Soo(r, 8) _
k=1
By (3.3.2), we have
u(r,G)I
Soo(r,G).
IaoI +
To estimate Soo, we note that, for any r E [0, R] and any 8 E S1,
(a +
Soo(r, e)
2 Rk.
k=1
0o
o0
(a + bk)
S'oo(r, B)
k=1
R2k
<00.
k=1
m1
ax ay 'n2 u(r, 9) =
In order to justify the interchange of the order of differentiation and summation, we need to prove that the series in the right-hand side is convergent
absolutely and uniformly in BR. Set
(3.3.5)
Sm1m2 (r, 9) =
k=1
(We note that this is Soo defined earlier if m1 = m2 = 0.) By using rectangular coordinates, it is easy to check that, for k < m,
515 2 (akrk cos k9 + bkrk sin k9) = 0,
<k(k-1)...(k-m+1)(a+b)
Hence, for any r E [0, R] and any 8 E
00
(a + b) 2 kmRk-m.
1
5m1m2 (r, e)
k=m
72
00
a2 + bl 2
Smim2 (r, 8) ;
00
k2mR2(k-m)
kI
k=m
< oo.
k=m
w(r, e) - (e) _
00
and then
(rk
- l)a
k=l
For any
(a+b)<.
k=x+1
k=1
since the series in the left-hand side consists of finitely many terms. Therefore, we obtain
73
(a + by2km.
Smim2(r,6)
By definitions of ak and bk in
f (e)cos k9 d8 = - f
ak _
bk _
---
/ (e si
k9 dB
d8.
Hence, {kbk, -kak} are the coefficients of the Fourier series of cps, so
00
k=1
k=1
Hence by the Cauchy inequality, we have for any r E [0, 1] and B E S1,
00
1
(a + b) 2m
Smlm2 (r, e)
k=1
OO
k2(m+1) (a
00
k-2
(i
This implies
Smlm2 (r, e) < Cm IIm+1) IIL2(S1)
74
cm,RII(PIIL2(S1),
jj
II(PIIL2(S1),
i=0
u(r, B) =
Js1
K(r, B,
) dry,
where
00
+rkcosk(9_).
K(r,9,)=
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
(3.3.7)
K(r, 8, r1) =
1-r2
2ir
1 - 2r cos(B - ri) + r2
75
solutions of Dirichlet problems for the Laplace equation in balls with continuous boundary values.
Next, we compare the regularity results in Theorems 3.3.2-3.3.3. 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.
ixxO in(0,ir)x(0,oo),
(3.3.8)
u(x, t) = a(t)w(x)
Then
a'(t)w(x) - a(t)w"(x) = 0,
and hence
w"(x)
w(x)
Since the left-hand side is a function of t and the right-hand side is a function
ap(t)
a(t)
w(0) = w(ir) = 0.
76
wk(x) = 1 / 2 sin kx
Vr
form a complete orthonormal set in L2(0, 71). For any v E L2(0, 7r), the
Fourier series of v with respect to { sinkx} is given by
v (x) _ I-)vksinkx,
where
v _ -2
f"
Jo
V I I LZ (0,ir) _
For k=
vk
let
uk(x, t) = ak(t)wk(x)
be a solution of (3.3.9). Then ap(t) satisfies the ordinary differential equation
ap(t) + k2ak(t) = 0.
ak(t) =aye-z t
uk(x, t) _
, we have
We note that uk satisfies 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 infinite linear combination of uk
and choose coefficients appropriately.
and that the boundary condition in (3.3.10) is the same as that in (3.3.8).
Now, let us suppose that
(3.3.11)
u(x, t) _ /Yake_1c2tsinkx
77
, we
u(x, 0) _ I-Laks1nkx,
but we are given the initial condition u(x, 0) = uo(x) for x e (0, it). Thus
we take the constants ak, k = 1, 2, , to be the Fourier coefficients of uo
with respect to the basis {/sin kx} of LZ(0, it), i.e.,
Jo
(3.3.12)
uo(x) sin kx dx
for k = 1, 2,
ut -
t 0 I- uOIILz(p,) = O.
Proof. Let i and j be nonnegative integers. For any x e [0, it] and t e
(0,oo), we have formally
axa u(x, t) =
ak
k=1
dt
(e-k2t)
dxz
sin kx).
In order to justify the interchange of the order of differentiation and summation, we need to prove that the series in the right-hand side is convergent
absolutely and uniformly for any (x, t) E [0, it] x [to, oo), for an arbitrarily
fixed to > 0. Set
00
(3.3.13)
SZ (x,t) _
_k2t
ak dt3 (e
dZ
) x2 (sin
kx)
Fix to > 0. Then for any (x, t) E [0, it] x [to, oo),
Sz.7(x,t)
laki
k=1
12t
k=1
ki-1-2j
laki k2to
k=1
e
78
Then the Cauchy inequality implies, for any (x, t) E [0, it] x [to, oo),
00
(3.3.14)
00 k2i+4j
Sid (x, t) C
k=1
IIUOIIL2(O,),
e2k2to
k=1
u(x, t) - up(x
and then
Iu(x,t) - uo(x) I2 dx =
e-k2t
_ 1112
k=1
We note that e-k2t -+ 1 as t -+ 0 for each fixed k > 1. For a positive integer
K to be determined, we write
J/'
- up(x) I2dx =
ak2 e
k2t
_ 1)2 +
k=1
For any S > 0, there exists a positive integer K = K(s) such that
k=K+1
a e-k2t
- 1)2 <5
k=1
since the series in the left-hand side consists of finitely many terms. Therefore, we obtain
J0
79
In fact, (3.3.14) implies the following estimate. For any integer m > 0
and any to > 0,
IIUIICm([O,ir] X [to,oo))
where Cm,to is a positive constant depending only on m and to. This estimate
controls the C"2-norm of u in [0, ?f] x [to, oo) in terms of the L2-norm of uo
on (0, ii). 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 uo
is only L2.
= 0,
Utj(it) = 0.
ut(ii,t) = 0
fort > 0.
Evaluating at t = 0 yields
ut(0,0) = 0,
Ut(ii,0) = 0.
u(0, 0) = 0,
u(ir, 0) = 0,
and hence
u(0)=0,
ug(71)=o.
uo2e)(0) =0,
80
Let
kZ+2j I ak I .
k=1
ak =
-2
uo(x) sin kx dx =
o
u0(x)
0
7f
pir
cos x
k
dx
sin /<'r
We note that values at the endpoints are not present since u(0) = uo (lr) = 0
ak = (-1)
m2
(x)
kx
if in is odd,
dx
and
aj _ (-1) 2 1/
v
if in is even.
dx
Jo
a/
U0
L2(O,ir)
k=1
Hence, by the Cauchy inequality, we obtain that, for any (x, t) E [0, 7r] x
[0, oo) and any in,
00
(x, t)i+2j I ak I
00
k2mak
k=1
00
2
k2(i+2j-m)
k=1
k=1
By taking m = i + 2 j + 1, we get
Si (x , t) < Ci u(m)
II
I L2 0
81
Proof. It follows from Theorem 3.3.5 that u is smooth in [0, it] x (0, oo)
and satisfies the heat equation and the homogeneous boundary condition in
(3.3.8). The continuity of u up tot = 0 follows from the proof of Theorem
3.3.6withi=j=0andm=1.
Now we provide another expression of u in (3.3.11). With explicit expressions of ak in terms of uo in (3.3.12), we can write
(3.3.16)
u(x,t) =
J0
where
G(x, y; t) = -
r k=1 e
for any x, y e [0, it] and t> 0. The function G is called the Green's function
of the initial/boundary-value problem (3.3.8). For each fixed t > 0, the
series for G is convergent absolutely and uniformly for any x, y e [0, it].
In fact, this uniform convergence justifies the interchange of the order of
summation and integration in obtaining (3.3.16). The Green's function G
satisfies the following properties:
These properties follow easily from the explicit expression for G. They
imply that u in (3.3.16) is a smooth function in [0, it] x (0, oo) and satisfies 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
0) = uo under appropriate assumpcontinuous up tot = 0 and satisfies
tions on uo. We point out that G can also be expressed in terms of the
82
Utt (3.3.17)
utt -
= 0 in (0,ir) x (0,oo),
t) = 0 for t E (0, oo),
u(0, t) =
u(x,t) = c(t)w(x).
An argument similar to that given for the heat equation shows that w must
be a solution of the homogeneous eigenvalue problem for -ate on (0, ir). The
and the corresponding
eigenvalues of this problem are Ak = k2, k = 1, 2,
normalized eigenfunctions
wk(x) = 1 / -sin kx
V 7r
,let
, we have
83
>
>...
ut(x, t) _
V_>-kbksinkx.
k=1
From the initial condition ut (x, 0) = u1(x), we see that kbk, for k = 1, 2,
,
are the Fourier coefficients of u1 with respect to the basis {
sin lax} of
L (0, ir), i.e.,
(3.3.21)
/-2
u (x)
0
sin kx
k
dx
for k =
t) _
d
-
84
In order to justify the interchange of the order of differentiation and summation, we need to prove that the series in the right-hand side is convergent
absolutely and uniformly for any (x, t) E [0, 7r] x [0, oo). Set
00
TZj(x, t) _ >
dt
I
Zi
(sin kx)
k-1
7'z (x,t)
ak =
ir
/o uo(x) sin kx dx =
71
ir
Vu
bk=4IJ
ux
sin lcx
k
dx =
710
2
uo (x)
241(x)
cos lax
3
sin k x
k3
dx,
dx.
sinkx}. Hence
(k6ak +
IIuo
+11 ui Lz(O,,r)
k=1
By the Cauchy inequality, we obtain that, for any (x, t) E [0, 71] x [0, oo),
0o
T(x, t)
o0
<oo.
+ b2k))
< ( k=1
. (k=1
)
Therefore, u is CZ in [0, 7r] x [0, oo) and any derivative of u up to order two
may be calculated by a simple term-by-term differentiation. Thus u satisfies
(3.3.17).
IIUIIC2([O,]x[0,oo))
Io
O
II'i
II'(Z) IIL2(O,) +
i=o
i=o
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/boundaryvalue 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 different 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 St be a bounded smooth domain in
][fin and uo be an L2-function in St. We consider
ut-Lu=0 inIIx(0,oo),
u(.,0) = uo in 12,
(3.3.23)
u=0 onaIIx(0,oo).
To solve (3.3.23) by separation of variables, we need to solve the eigenvalue
problem of -O in St with homogeneous boundary values, i.e.,
0o +
in u
p=0 on aS2.
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 (Ak, cpk), where Ak is a nondecreasing sequence of positive numbers such that Ak -+ oo as k -+ oo and cps is a sequence of smooth
u(x, t) _
86
3.4. Exercises
Exercise 3.1. Classify the following second-order PDEs:
n
(1)
= 0.
:i:
1<i<j<n
i=1
(2)
uxi x j =
0.
1<i<j<n
Exercise 3.2.
x=rcos8, y=rsin8.
Prove that the Laplace operator 0 can be expressed by
1
r2
Du = urr + -ur +
uee
(sin e ae I
52u
+ r2 sin2 8 5P2
Exercise 3.3. Discuss the uniqueness of the following problems using energy
methods:
(1)
(2)
Jzu_u3=f in S2,
lu= cp
on 852;
JL\u_ufu2(y)dy=f
in St,
lu=p
on aSl.
ut - Du = f
in S2 x (0, oo),
u(.,0)=u in1,
u=0 onDIIx(0,oo).
Prove
fvu(.,t)I2dx
sup+0tT
u2 dxdt
0
<
St
rT
dx +
J J f 2 dxdt),
n
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 uo E L2(0, 7r), let u be given by (3.3.11). For any
nonnegative integers i and j, prove
I< t
Exercise 3.8. For any uo E L2(0, 7r), solve the following problem by separation of variables:
ut -
U(0, t) =
Exercise 3.9. For any uo E L2(0, 7r) and f E L2((0, 7r) x (0, oo)), find a
formal explicit expression of a solution of the problem
Exercise 3.10. For any uo, ul E L2(0, 7r) and f E L2((0, 7r) x (0, oo)), find
a formal explicit expression of a solution of the problem
ut - Du = 0 in St x (0, T),
U(,T)0 in1,
U=0 onaclx(0,T).
Prove that u = 0 in S2 x (0, T).
Hint: The function J(t) =log f u2(x, t) dx is a decreasing convex function.
Exercise 3.12. Classify homogeneous harmonic polynomials in ][83 by following the steps outlined below. Let (r, 8, q5) be spherical coordinates in ][83.
(Refer to Exercise 3.2.) Suppose u is a homogeneous harmonic polynomial
cp) for some function Qm defined in
of degree m in ll83 and set u =
2.
88
m (m + 1)Qm. +
I s i n 9 aQ
sin 8 8B
aB
y,,,
l+
sine 8
2
aQ'"
,=0
dm+k
fm,k(/t) = (1 - 2) 2 dm,+k C1 -
2r"'
for E [-1,1],
Chapter 4
Laplace Equations
4. Laplace Equations
90
We point out that several results in this chapter are proved by multiple
methods. For example, interior gradient estimates are proved by three methods: the fundamental solution, the mean-value property and the maximum
principle.
In Section 4.4, we discuss the Poisson equation. We first discuss regularity 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.
Du =
i=1
The equation Du = 0 is called the Laplace equation and its C2-solutions are
called harmonic functions.
v(r) = u(x).
Foranyi=1, ,nandxz40,weget
xi
= v' (r)-,
r
uxixi =
x2
2+
r
v,(r)
-r1 - x3r
,
Du=v + n-1
v =0.
r
(logv')'+
n-1
r
=0,
(log(r'v'))' = 0.
91
v(r) = Cl + C2 log r
v (r) = C3 +
au
aBr 01
dS=1 foranyr>0.
C4 =
(2 - n)wn'
where wn is the surface area of the unit sphere in R.
log x for n = 2,
I'(x) _
and
for n> 3.
IxI2-n
r(x)
(2 -
The function I' is called the fundamental solution of the Laplace operator.
We note that I' is harmonic in Ilgn \ {0}, i.e.,
Or=O lriRn\l0I,
and
ar
v dS=1 foranyr>0.
f
Moreover, r has a singularity at the origin. By a simple calculation, we
Br
,n and any x
I'( x) =
0,
1
wn x n
and
rxi x3 =
wn
x2
jxn S
nxjx3
IXIn+2
We note that r and its first derivatives are integrable in any neighborhood
of the origin, even though r has a singularity there. However, the second
derivatives of r are not integrable near the origin.
4. Laplace Equations
92
in IIand v = (v1,.. ,v) be the unit exterior normal to 852. Then for any
u, v e Cl (S2) fl C(S2) and i = 1,
,n,
[uv dx = J
Js
uvv2 dS -
dx.
Jsi
asp
By summing up for i = 1,
vwxi U2 dS.
fsp
L (vzw + Vv Ow) dx = J
v dS.
For any v, w e C2(S2) fl C' (1), we interchange v and w and subtract to get
a second version of Green's formula,
Js
(vL\w
Dv
- wOv) dx = Jasp (Dw
v
wav)
dS.
\ av -
Jsi
Ow dx =
Jasp
aw dS.
8v
We note that all these integral formulas hold if SZ is only a piecewise C1domain.
Now we prove Green's identity, which plays an important role in discussions of harmonic functions.
u(x) =
I'(x - y)Dyu(y) dy
- Jas
IF(x-y)--(y)-u(y)--(x-y)JdSy.
y
Proof. We fix an x E S2 and write I' = I'(x - ) for brevity. For any r > 0
such that Br(x) C S2, the function I' is smooth in S2 \ Br(x). By applying
Green's formula to u and I' in St \ Br(x), we get
J\Br(x)
(r_
(Fzu - uDI') dy = f
+
av
f
.Iag,.cX>
avyl
r au - u ar) dsy,
aU
avy
93
av
v
Js I'Du dyasp= f 1F--u1dS+lim
y/
asT (x)
av
IF--uldS.
v
y
Br (x)
r au
dSy
aU
r2-n
I
/asr()
(2 - n)wn J
aU dsy
<
8u
and
u
aBr (x)
ar
avy
dSy =
wnr
n_ 1
u dSy
as r-+0,
aBr (x)
where v is normal to DBr (x) and points to x. This implies the desired result
0
for n > 3. We proceed similarly for n = 2.
-1--(x-y)dS=l,
Jest avy
for any x E S2. This can be obtained by taking u - 1 in Theorem 4.1.2.
If u has a compact support in SZ, then Theorem 4.1.2 implies
u(x) _
u(x) =
Jci
LyF(xy)Ox.
Here S is the Dirac measure at x, which assigns unit mass to x. The term
"fundamental solution" is reflected in this identity. We will not give a formal
definition of distribution in this book.
-'
Du = f
in S2,
u= cp on BSt.
4. Laplace Equations
94
2S2
((x y) au (y) -
y)J dSy.
u(x)
(x,y)u(y) dy -Jest
(x,.) E Cl(S2)f1C2(St)
such that
y) = 0 for any y E SZ,
(4.1.21
-'
(x,y) = I'(x - y)
The existence of 1 in general domains is not the main issue in our discussion
here. We will prove later that 'I(x, ) is smooth in St for each fixed x if it
exists. (See Theorem 4.1.10.)
u(x) =
G(x, y) f (y) dy +
J ci
cp(y)
y (x, y) dSy.
95
that the Green's function is unique if it exists. This follows from Lemma
3.2.1 or Corollary 4.2.9, since the difference of any two Green's functions is
harmonic, with vanishing boundary values.
Lemma 4.1.5. Let G be the Green's function in 12. Then G(x, y) = G(y, x)
for any x, y E 1 2 with x
y.
\(BT(Xi)UBr(Xz))
aG2
8G1
aGl
aG2
av
fdS+
LBr(X2)
dSy,
av
where v is the unit exterior normal to B(St\ (Br(Xi) UBr(x2))). Since Gi(y)
is harmonic for y # xi, i = 1, 2, and vanishes on BSt, we have
cds+ dso.
acll
DG 2e
DG1
Gl av
G2 8v
- G2
v J
sT(Xi) C Gl 8v - G2 8v
aBr(X2)
Now we replace Gl in the first integral by I'1 and replace G2 in the second
integral by I'2. Since Gl - I'1 is C2 in St and G2 is C2 in 1 2 \ B,. (x2 ), we have
G1
((Gl
ac2
- rl)av - G2
a(G1-r1)1 dSy
av
as r
0.
I dSy
as r
0.
Similarly,
asr(X2)
(Gla(G 8v
Therefore, we obtain
r1
B,.(xi)
aG2
Dv
- G2
art
Dv
dSy +
o,
aBr(x2)
G28I'1
Dv
dSy
G2(xi), -f p
VV
dSy
4. Laplace Equations
96
(2 - n)Wn
IRI2 x12-n
- xl2-n - \ /
G(O, y) -
IRI2xl
\'\
r(o - y) =
1
(2 - n)wn
Hence we take
cT(O,y)=
(2 - n)w
RTh,
we have DOxy
xl'
R
DOyX. Then for any y E aBR,
xl
Iy - x
97
and hence,
iy-xi= lRi
(4.1.4)
This implies
R n-2
r(y-X),
r(y-x)- IxIJ
(x,y)=
\iT)
F(y-X),
for any x E BR \ {0} and y E BR \ {x}. The proof for n = 2 is similar and
O
is omitted.
8G/
8vy
`x
y)
R2- 1x12
- WRI x - yln
CT(x,y)=
R n2
- xI2n
for any x E BR \ {0} and y e BR \ {x}. Hence we get, for such x and y,
1
yZ _ xi
R n-2
GYi(XY)=_11n_)
yti _ XZ
ly-xi
4. Laplace Equations
98
By (4.1.4) in the proof of Theorem 4.1.6, we have, for any x E BR \ {0} and
y E aBR,
yz
Ra _ Ixl a
This formula also holds when x = 0. With vz = y2/R for any y E BBR, we
obtain
DC
2_I I2
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.,
Rz
K(x,y) =
(4.1.5)
Ixlz
WnRI x - yIn
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 definition K(x, y) = a 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 C2(BR) harmonic function u in (4.1.3), we conclude
f
that
u(x) =
K(x, y)u(y) dSy for any x E BR.
JaBR
Now we are ready to solve the Laplace equation in balls, with prescribed
Dirichlet boundary values.
(4.1.6)
u(x) = f
99
(xO)= f
dSy.
asR
Then
u(x)
(xO) _
aR
K(x,
where
...,
I1=
...
I2=
aBRnB5 (xo)
aBR\Ba (xo)
I'() -
(xo)I <6
that
K(x' y) C
e
2MwRn-1'
for any x E BR f1 Bay (xo) and any y E 8BR \ Ba (xo). We note that S' depends
Idsy < 6.
BR\Ba (moo)
Hence
4. Laplace Equations
100
12ir f
2ir
where
K(r, B, ri) =
R2
- r2
in (4.1.6) is smooth for lxi < R, even if the boundary value cp is simply
continuous on aBR. 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
,fnl =
wnRn-1
u(y) dSy.
aBR
We note that wnRn-1 is the surface area of the sphere BBR. 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) satisfies
min co < u < max co in BR.
aBR
aBR
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.
101
4.1.4. Regularity of Harmonic Functions. In the following, we discuss 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.
u(x)=-
Jai-
(F(x_Y)(Y)_u(y) a y (x -
dsy,
for any x E 11'. There is no singularity in the integrand, since x E 11' and
D
y E 911'. This implies easily that u is smooth in 11'.
We note that, in its definition, a harmonic function is required only to
be C2. Theorem 4.1.10 asserts that the simple algebraic relation Du = 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 Du 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 first 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 first derivatives of a harmonic function at any point are controlled by
its maximum absolute value in a ball centered at this point.
R B( o)
lulu
For any x E B1,4i we write I' = I'(x - ) temporarily. For any r > 0 small
4. Laplace Equations
102
si\BT()
Jasl
+
(cor- ua(a )
aBT(X)
av - u a(a r
where v is the unit exterior normal to a(Bl \ Br(x)). The boundary integral
over aBl is zero since cp = a = 0 on 8B1. In the boundary integral over
we may replace cp by 1 since Br(x) C B1,2 if r G 1/4. As shown in
the proof of Theorem 4.1.2, we have
arl
(Du
u(x)=hmi
iP--u---dS,
r-+0 aBT av
av J
where v is normal to BB,. (x) and points toward x. For the domain integral,
the first term is zero since Du = 0 in Bl. For the second term, we have
o(ar) =
yr + nor.
u(x) _ -f
y) +
VyI'(x
- y)) dye
for any x E B114. We note that there is no singularity in the integrand for
lxi <1/4 and 1/2 < y <3/4. (This also gives an alternative proof of the
smoothness of u in B114.) Therefore,
vu(x) _ -
J2 <IyI<q u(y)
y)
lfor any x E
B1
l< csup
B1
103
We note that the proof above consists of two steps. We first 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 Bl. We point out that this scaling argument is commonly used
in studying elliptic and parabolic differential equations.
Next, we estimate derivatives of harmonic functions of arbitrary order.
Theorem 4.1.12. Suppose u E C(BR(xa)) is harmonic in BR(XO) C 118'x.
Then for any multi-index a with al = m,
cmem-lml
IDau(xo)I
max lul,
Drn
B(lma,x(xo) IvI.
I1 8)R
<
C"2e"2-lm!
(9R)m
max ui,
BOR(X)
max
Therefore,
IvI <
(9R)m
max uI.
BR(xo)
G+m+lem-lml
lVv(xo)
(1 - 9)9R-'
max uI.
Rxo
<
(1+ mJ (m+1)<e(m+1).
cm+le m(m + 1)!
max lul.
R
BR(XO)
Hence the desired result is established for any derivatives of u of order
m+1.
4. Laplace Equations
104
sup uk I
Bd (X)
- Cmd
M,
For any 2 = 0, 1,
- DQUk(y)I
C2+11Z-E-1MIx _ yl,
C SZ
ce+id
for any 2 = 0, 1,
,and any x, y E 52,. By Arzela's theo, any k = 1, 2,
rem and diagonalization, we can find a function u in SZ and a subsequence
{uk'} such that
As shown in the proof, uk' converges to u in Ct(SZ') for any SZ' with
105
u(x) =
a u(xo) (x - xo)
u(xo + h) = u(xo) +
where
Rm(h) =
for some B E (0, 1). Note that xo + h e BR(xo) for Ihi <R and that R,,,,(h)
consists of nterms of derivatives of u of order m. By applying Theorem
4.1.12 to u in BR(xp + 8h), we obtain
R
<
m' nm '
CCneih11
RJ
Rm
Rmax
B
C o) lul
max lul.
2R(X0)
u(x0 + h) =
z-O
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 differential equations. Many results in this
4. Laplace Equations
106
section were either proved in the previous section or will be proved in the
next section.
We first define the mean-value property. There are two versions of the
mean-value property, mean values over spheres and mean values over balls.
Definition 4.2.1. Let St be a domain in II8" and u be a continuous function
in ft Then
(i) u satisfies the mean-value property over spheres if for any Br(x) C St,
u(x)=
u(y)dS;
``'mar
(ii) u satisfies the mean-value property over balls if for any BT(x) C St,
n
u(x) =
u(y) dy,
n
wnr
Br
We note that wnrr_l is the surface area of the sphere OBr(x) and that
wnrn/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
u(y) dy,
r(a)
u(x) =
or
u(x)=-
u(x+ry)dy.
A function satisfying mean-value properties is required only to be continuous to start with. However, a harmonic function is required to be C2.
We now prove that these two requirements are actually equivalent.
Theorem 4.2.2. Let 1 2 be a domain in II8" and u be a function in 12.
(i) If u E C2(1) is harmonic in 12, then u satisfies the mean-value property in S2.
107
Proof. Take any ball Br(X) C St. Then for any u E C2(1t) and any p E
(0, r), we have
adS - pn-1
Dudy =
B(2)
(y)
a(x -pw)dSw
Bl
vp
u(x +
= per'-i a
aP
(i) Let u E C2(1t) be harmonic in St. Then for any p E (0, r),
a
-s---
u(x+pw)dS=0.
asl
Integrating from 0 to r, we obtain
and hence
asl
u(x dSw =
11/ u(x+rw)dS.
u(x)=Wn 8B1
fi
wJ rte'-1b(r) dr = 1.
0
The existence of such a function can be verified easily. Define cp(x) = b(IxI).
claim that
u(x) =
Jci
x) dye
for any x E St with dist(x, a12) > E. Then it follows easily that u is smooth.
Moreover, by (4.2.1) and the mean-value property, we have, for any Br(X) C
S2,
Du dy = r"-i ar
JaBI
4. Laplace Equations
108
Now we prove the claim. For any x e SZ and e < dist(x, 81Z), we have,
by a change of variables and the mean-value property,
- x) dy =
=1
=
(x + ez)cp(z) dz
l
ff
f1
JB u(x +
b(r)rn-1
J0
f1
Corollary 4.2.3. Harmonic functions are smooth and satisfy the meanvalue property.
xo =
)
wn R
2n f
u(y) dy
R (x0)
wn R
u(y)vz dS'y
BR (xo )
and hence
(xo)I <
n
WnR
IaBR<lu(y)I ds
n
max
wn RnaBR (xo)
n -max Iui.
IuI.WnR<_
R BR (x0 )
109
Proof. As in the proof of Theorem 4.2.4, from integration by parts and the
nonnegativeness of u, we have
IC n f
wnR" aBR(xo)
nu(xo),
R
where in the last equality we used the mean-value property.
IC Rv(x)
By letting R -+ oo, we conclude that Vv(x) = 0. This holds for any x E ][8Th.
Hence v is constant and so is u.
Otherwise, we consider u -I- s for any constants > 0, derive the desired
4. Laplace Equations
110
IVu(x)I c
2ri
R u(x),
or
IVlogu(x)I <
fi
logu(tx + (1 - t)y) dt
J
(x-y).
f1
log
GG
2n.
Therefore
u(x) <
e2nu(y)
for any x, x E Br(). To see this, we note that Br(x) C B3r(x) C B4r(x) for
any x, x E Br(). Then the mean-value property implies
n
udy <
udy = 3nu x .
ux= n
W nr
Br (x)
W nr
B3r (x)
... <
111
Theorem 4.2.8. Let 1 be a bounded domain in Rn and u E C(SZ) be harmonic in 1. Then u attains its maximum and minimum only on 9f unless
u is constant. In particular,
inf u < u < sup u
asp
in IL
asp
: u(x)=M}.
C D,
It is obvious that D is relatively closed; namely, for any sequence
if
the continuity of u.
x
xE
Next we show that D is open. For any x0 E D, we take r > 0 such that
Br (x0) C fZ. By the mean-value property, we have
M=u(xO)=wnr
nn
udy<
Br(x0)
n
Wnr n
Mdy=M.
Du = f
u = cp
in S2,
on aSt.
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.
4. Laplace Equations
112
asp
Proof. If u has a local maximum at a point xo in St, then the Hessian matrix
(V2u(xo)) is negative semi-definite. Thus,
asp
Then
max u < max u6 + sR2 = max u6 + sR2 < max u + sR2.
asp
asp
We have the desired result by letting s -+ 0 and using the fact that 31 C
SZ.
113
Du +cu = f
in S2.
Du +cu > 0 in 1.
Then u attains on aSZ its nonnegative maximum in SZ, i. e.,
-asp
Proof. We can proceed as in the proof of Theorem 4.3.2 with simple modifications. In the following, we provide an alternative proof based on Theorem
4.3.2. Set SZ+ _ {x E S2; u(x) > 0}. If S2+ = fD, then u < 0 in St, so u+ - 0.
If S2+
0, then
in SZ+ .
maxis=maxis=maxis+.
asp
4. Laplace Equations
114
assume
a(x)j >
z,j=1
biu+ cu = f in f
Lu
i,j=1
i=1
Du + cu > 0
inn,
u<0 on BSZ.
Then u < 0 in SZ.
Du + cu > Ov + cv inn,
u < v on BSZ.
115
Du + cu = f
in S2,
u = cp
on 852.
inn,
Ow + cw = 0
w=0
on 0.
Du = 0
in SZ
u=0 on acI,
where St = ][8n \ Bl. Then a nontrivial solution u is given by
flogixi
1 x-1
for n = 2;
for n > 3.
u(x) _
sin x.
z=1
inn,
u=0 on acI.
Du + nu = 0
4. Laplace Equations
116
Du + cu > 0 in B.
Assume u(x) < u(xo) for any x E B and u(xo) > 0. Then
0'
w(x) =
e-R2
and
117
>
e-alxl2
e-alxl2C4a21x12
- 2na -I- c
e-alxl2
in D,
if we choose a sufficiently large. By c < 0 and u(xo) > 0, we obtain, for any
L\v + cv = Du + cu + E(L\w -}- cw) - cu(xp) > 0 in D.
Ov + cv > 0 in D,
v<0 on DD.
By the maximum principle, we have
v<0 in D.
In view of v(x0) = 0, then v attains at x0 its maximum in D. Hence, we
obtain
-(xO)
0,
and then
8u(xo)
> -Eaw(xo) =
8v
av
This is the desired result.
0.
Remark 4.3.8. Lemma 4.3.7 still holds if we substitute for B any bounded
C1-domain which satisfies an interior sphere condition at x0 E DSZ, namely,
if there exists a ball B C SZ with x0 E DB. This is because such a ball B
is tangent to DSZ at x0. We note that the interior sphere condition always
holds for C2-domains.
Now, we are ready to prove the strong maximum principle.
Du + cu > 0
in ft
4. Laplace Equations
118
Obviously, we have
Du + cu > 0 in B,
and
u(x) < u(xo)
8v(x0) '
0'
119
constant in 1.
The following result improves Corollary 4.3.5.
Du + cu > 0
in S2,
u<0 on D11.
Then either u < 0 in SZ or u is a nonpositive constant in 11.
We now consider the Neumann problem.
Du -I- cu = f
au - cp
av
in St,
on BSZ,
0 and is unique
in 1,
= 0 on aSZ.
4. Laplace Equations
120
aSZ. Lemma 4.3.7 implies a (xo) > 0, which contradicts the homogeneous
LI
boundary value. This contradiction shows that u is constant.
Du + cu = f in St,
u = cp
on aSl.
Then
asp
Proof. Set
F=suplfl, 4=supIcpl
s
asp
Then
(t+c)(u)=f -F in1,
fu = fcp
on 852.
we have
Ov+cv<-F in1,
v>4 onc91.
121
Therefore,
(L+c)(u) > (L+c)v
fu < v
in S2,
on BSZ.
fu < v
in St,
BR (X0 )
R2
cp P +
8BR (Xo )
sup
2n BR (X0
f.
4.3.4. Gradient Estimates. In the following, we derive gradient estimates, estimates of first derivatives. The basic method is to derive a differential equation for Vu 12 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 Du in 1 in terms of Du on 91, as well as u in 1, while interior
gradient estimates yield estimates of Du in compact subsets of 1 in terms
of u in 1. 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.
8B1
+2
0(iouI2) = 2
2
2=1
2,j=1
4. Laplace Equations
122
o('IouI2) = (o)IVuI2 +4
7''xiuxjuxixj +
i,j=1
?,j=1
+ rn- 7' xi ux
Then
0. To interpret
E Co(Bl). Then
IVuI2 > -CIDuI2,
au2) <
Bl
au2).
8B1
0B1
C,
B112
123
Ov = -IOvI2.
Next, we prove an interior gradient estimate for v. By setting w = IVvI2,
we get
Ow + 2
n
vx i x j .
vxi wxi = 2
2,j=1
2=1
n
vxi xi
n
2
vxi
xi ,
CT
i=1
i=1
we have
n
2n
vxixj >
(4.3.1)
i=1
i, j=1
2
vxixi
> - (ov)2
n
Ivvl4
w2
0(cow) + 2
v2ix
vxi (pw)xi =
+4
i,j=1
i=1
xivxj vxixj
i,j=1
+2w
xivxi + ()w.
i=1
(n
2
pvxi
x j+
vx j.
Then
n
(w) + 2
v (w) >
v. - 2IVIIVvI3
z,j=1
z=1
+ (o _
41
l21
IovI2
0(i14w) +
i=1
87I3IV1'IIIVvI3 +
131 01I2)IVvl2.
4. Laplace Equations
124
t=
Hence with
we get
n
O (4w)
v (4w)
>
z=1
2n
4w2
- C.
We note that q4w is nonnegative in Bl and zero near 8B1. Next, we assume
that r/4w attains its maximum at xo E Bl. Then
0 and 0(r74w) <
0 at xo. Hence
4w2(xo) <C.
ii4(xo). By combin-
in B1,
125
Jo(logx), n=2, as x -4 0.
lo(1x12_Th), n> 3
v=u on aBR.
The existence of v is guaranteed by the Poisson integral formula in Theorem
4.1.9. Set M = maxaBR
We note that the constant functions fM are
obviously harmonic and -M < v < M on aBR. By the maximum principle,
lvl< M in BR.
Next, we prove u =vin BR \ {0}. Set w = v - u in BR \ {0} and MT =
maxas,. wi for any r <R. We only consider the case n > 3. First, we have
rn-2
<W(X)<Mr'
Mr
for any x e aBr U aBR. It holds on BBr by the definition of Mr and on
aBR since w = 0 on aBR. Note that w and IxI2-n are harmonic in BR \ B,..
Then the maximum principle implies
rn-2
rn-2
xl n-2'
IxI
and hence
rn-2
Iw(X)IMr
xl n-2'
ic Ixln-2M+ IxIri_2
(r2maxiui),
4. Laplace Equations
126
4.3.6. Perron's Method. In this subsection, we solve the Dirichlet problem for the Laplace equation in bounded domains by Perron's method. Essentially 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 first discuss subharmonic functions. By Definition 4.3.1, a C2function v is subharmonic if Ov > 0.
Proof. We first prove the only if part. For any ball B C S2 and any
harmonic function w E C(B) with v < w on aB, we have
Ov > Ow in B,
v < w on 8B.
By the maximum principle, we have v < w in B.
Ow = 0 in B,
w = v on aB .
The existence of w in B is implied by the Poisson integral formula in Theorem 4.1.9. We have v <w in B by the assumption. Next, we note that
Ow = 0 > Ov in B,
w=v on aB.
We have w <v in B by the maximum principle. Hence v = w in B, which
contradicts Ow > Ov in B. Therefore, Ov > 0 in SZ.
El
Lemma 4.3.17 leads to the following definition.
127
v) and
D={xEII: u(x)-v(x)=M}C11.
Then D is relatively closed by the continuity of u and v. Next we prove that
D is open. For any xo E D, we taker < dist(xo, 811). Let u and v solve,
respectively,
Lu=0 inB(xo),
u =u on aBr(xp),
and
z=0 0
in Br(xp),
v = v on aBr(xp).
i<v inB(xo).
Hence,
2 -v u-v In Br(xp).
Next,
z(u-'i)=O inB(xo),
-1=u-v onaB(xo).
With u - v < M on aB,.(xo), the maximum principle implies u - v < M in
BT(xp). In particular,
Hence, (u - v)(xo) = M and then u - v has an interior maximum at xo. By
the strong maximum principle, u - v -Min BT(xo). Therefore, u - v = M
on 8Br(xo). This holds for any r < dist(xo, 812). Then, u-v =Min B,.(xo)
and hence Br(xo) C D. In conclusion, D is both relatively closed and open
in 12. Therefore either D = Ql or D = 12. In other words, u - v either attains
its maximum only on aS2 or u - v is constant. Since u < v in BSZ, we have
D
u < v in 11 in both cases.
4. Laplace Equations
128
The proof in fact yields the strong maximum principle: Either u < v in
St or u - v is constant in SZ.
Next, we describe Perron's method. Let SZ be a bounded domain in
I[8n and cp be a continuous function on BSt. We will find a function u E
C(St) fl C(St) such that
Du = 0
(4.3.2)
u = cP
in SZ,
on Dt
inf
Hence for any x E SZ
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 Perron's method, we will prove that the function u defined 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 first step, we prove that u(), is
harmonic in SZ. This holds for arbitrary bounded domains. We note that u
in (4.3.3) is defined only in SZ. So in the second step, we prove that u() has
a limit on au and this limit is precisely gyp. For this, we need appropriate
assumptions on Dt
Before we start our discussion of Perron's method, we demonstrate how
to generate greater subharmonic functions from given subharmonic functions.
w=v inSZ\B,
and
Ow=O in B.
Then w is a subharmonic function in 12 and v < w in 12.
129
Next, we take any B' with B' C SZ and consider a harmonic function
u e C(B') with w < u on DB'. By v < w on DB', we have v <u on DB'.
Then, v is subharmonic and u is harmonic in B' with v < u on DB'. 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 f1 B' and w < u on 5(B f1 B'). By the maximum
principle, we have w < u in B f1 B'. Hence w < u in B'. Therefore, w is
Proof. Set
S,={v: v E C(S2) is subharmonic in SZ and v < cp on 8St}.
Then for any x E SZ,
u(x) =sup{v(x) : v e
In the following, we simply write <S =
M=maxco.
asp
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 SZ with
co < M on 511. By Lemma 4.3.19, for any v e s,
v<M inf
Thus u is well defined and u <M in 11.
Step 2. We prove that s is closed by taking maximum among finitely
, v E s and set
many functions in S. We take arbitrary vl ,
v=max{vl, ,vk}.
It follows easily from Definition 4.3.18 that v is subharmonic in 11. In fact,
we take any ball B C SZ and any harmonic function w e C(B) with v < w
, k. Since v2 is subharmonic, we
on SB. Then v2 <w on SB, for i = 1,
get v2 <w in B, so v < w in B. We conclude that v is subharmonic in 11.
Hence v E s.
4. Laplace Equations
130
lira vi (xo) =
i-+oo
We point out that the sequence {vi} depends on x0. We may replace vi
above by any v2 E s with v2 > v2 since
vi (x0) < ?Ii (x0) < u (x0) .
m < vi <
in ft
Z.
For the fixed Br(x0) and each vi, we let wi be the harmonic lifting in Lemma
4.3.20. In other words, wi = vi in 1 \ Br (x0) and Owi = 0 in Br (x0) . By
Lemma 4.3.20, wi E s and vi < wi in fZ. Hence,
lim wi (xo) =
i-+oo
and
m < wi < u in 1 ,
for any i = 1, 2,
i-+oo
in ft
Z.
For the fixed Br(x0) and each vi, we let wi be the harmonic lifting in Lemma
4.3.20. Then, wi E s and vi < wi in 1. Moreover, wi is harmonic in Br (x0)
and satisfies
lim wi (x) =
i-oo
and
131
and
w(x) =
Next, we note that w-w is a harmonic function in B,.(xo) with a maximum
We
woo (xo) = 0,
Then
Imo
Proof. As in the proof of Lemma 4.3.21, we set
-Kw(x)
2M,
ko-2(xo)IE-Kw on a.
Since cp(xo)-e+Kw(x) is a subharmonic function in SZ with cp(xo)-E+Kw <
cp on BSZ, we have cp(xo) - e + Kw E S. The definition of u implies
p(xo)-E+Kw<u, inn.
On the other hand, cp(xo) + E - Kw is a superharmonic in SZ with cp(xo) +
E - Kw > cp on SZ. Hence for any v e
we obtain, by Lemma 4.3.19,
v<p(xo)+E-Kw inft
4. Laplace Equations
132
u, <p(xO) + E - Kw in
Therefore,
This implies
I- cp(xo)I < E.
lim sup
x-*xo
133
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.
4.4.1. Classical Solutions. Let SZ be a domain in W` and f be a continuous function in SZ. The Poisson equation has the form
Du = f
(4.4.1)
inn.
w f(x)
vr(x - y)
Ix_In_1' v2r(x - y)
x-y
4. Laplace Equations
134
F. In fact, extra
to get second derivatives of W f due to the singularity of
conditions are needed in order to infer that W f is C2. If W f is indeed CZ and
Ow f =fin Sl, then any solution of (4.4.1) differs from W f 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 W f.
Lemma 4.4.1. Let 1 2 be a bonded domain in 12, f be a bounded function in
for some integer
1 2 and W f be defined in (4.4.2). Assume that f E
w(x) =f F(x-y)f(y)dy.
We point out that the integration is in fact over a bounded region. Note that
I' is evaluated as a function of x - yI. By the change of variables z = y - x,
we have
w(x) _
I'(z) f (z + x) dz.
Rn
w( x) =
I'(z) f(z + x) dz =
I'(z) fx (z + x) dz
For f E
sign to get
fn
Ow(x) =
(x) = i=1
Rn
Fzi (z)fzi (z
i=1
= - lim
e-*O R \B i=1
x) dz.
x) dz
135
We note that f(. +x) has a compact support in ][8Th. An integration by parts
implies
Ow(x) _
(z)f (z + x) dSz,
- E o Jaaf 8v
where v is the unit exterior normal to the boundary 8BE of the domain
][8" \ BE, which points toward the origin. With r = Izi, we obtain
Ow(x) = Eo
aaE
= lim
ar
W1
1
(z)f(z + x) dSz
aJ6 f(z -I- x) dSz = f(x),
f r(x t
(y)
= wl(x) + wll(x).
The first integral is actually over 1 2 \ B,.12 (xo) since cp - 1 in B,.12 (xo) . Then
there is no singularity in the first integral if we restrict x to Br14(xo). Hence,
wj is smooth in Br/4(xo) and Owj = 0 in Br14(xo). For the second integral,
cp f is a C1-1-function of compact support in ft We can apply what we just
Lemma 4.4.1 is optimal in the C-category in the sense that the smoothness of f implies the smoothness of W f. However, it does not seem optimal
concerning finite differentiability. For example, Lemma 4.4.1 asserts that
W f is C2 in St if f is Cl in St. Since f is related to second derivatives of wf,
it seems natural to ask whether W f is C2 in St if f is continuous in St. We
will explore this issue later.
We now prove a regularity result for general solutions of (4.4.1).
ft
4. Laplace Equations
136
Du = f in 1,
u=cp onaf
Moreover, if f is smooth in SZ, then u is smooth in 11.
Ov = 0
inn,
v = cp -w on BSt.
Theorem 4.3.23 implies the existence of a solution v E C(1) fl C(St). (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 SZ, 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 C2-solution
x -x f
f(x) = 21x12 1
(_logx)h/2 + 4(-log Ixl)3/2 J'
137
Du = f in B1.
(4.4.3)
of harmonic functions. Suppose, to the contrary, that there exists a C2solution v of (4.4.3) in Bl. For a fixed R E (0, 1), the function w = u - v
is harmonic in BR \ {0}. Now u E C(BR) and v E C2(BR), so w E C(BR).
Thus w is continuous at the origin. By Theorem 4.3.16, w is harmonic in
BR and therefore belongs to CZ(BR). In particular, u is CZ at the origin,
which is a contradiction.
Example 4.4.4 illustrates that the C2-spaces, or any C'-spaces, are not
adapted to the Poisson equation. A further investigation reveals that solutions in this example fail to be C2 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 V2u can be estimated.
Better adapted to the Poisson equation, or more generally, the elliptic differential equations, are Holder spaces. The study of the elliptic differential
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 Du
is. It is beyond the scope of this book to give a presentation of the Schauder
theory.
4. Laplace Equations
138
In fact, we already employed such a structure when we derived energy estimates of solutions of the Dirichlet problem for the Poisson equation in
Section 3.2.
-Du = f
(4.4.4)
in Q.
(4.4.5)
pu pcp dx =
J[fdx.
si
-Du = f
in S2.
in (4.4.5). For any cp e C( Q), it is obvious that the integral in the lefthand side of (4.4.5) makes sense if each component of Vu is an integrable
function in Q. This suggests that we should introduce derivatives in the
integral sense.
uSo
dx = -
vjcp dx
It is easy to see that weak derivatives are unique if they exist. We also
point out that classical derivatives of C1-functions are weak derivatives upon
a simple integration by parts.
139
The superscript 1 in the notation Hl (St) indicates the order of differentiation. In general, functions in Hl (St) may not have classical derivatives.
In fact, they may not be continuous.
We are ready to introduce weak solutions.
-Du = f inn,
(4.4.71
'
on a.
u=0
(u,v)H1() = f u. Vvdx,
Jcz
IIuIIH)
\f Dudx I/
IIuIIL2(cl) $
foranyuEC.
For f e L2(St), we define a linear functional F on C by
(4.4.9)
F(cp) =
fdx,
cz
4. Laplace Equations
140
reason that C(SZ) is not complete with respect to the LZ-norm. For the
remedy, we complete C under the Ho-norm.
This implies that {uk} is a Cauchy sequence in LZ(St). We may assume for
some u E LZ(SZ) that
uk- u
in L2(S2) as k - oo.
ukSOx
dx = -
uk,X co dx
141
H).
Now we can prove the existence of weak solutions of the Dirichlet problem for the Poisson equation with homogeneous boundary value.
F(cp) = ffdx,
for any cp e H0'(1). By the Cauchy inequality and (4.4.8), we have
If IIL2(IOIIcOIIL2(cz) <_
Hence, F is a bounded linear functional on Ho (S2). By the Riesz representation theorem, there exists a u e Ho (S2) such that
(u, co)xo(st) =
4. Laplace Equations
142
In fact, if f e HIc(SZ) for any k > 1, then u e H2(S2). This is the L2-theory
for the Poisson equation. We again encounter an optimal regularity result.
If Du is in the space HIc(SZ), 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 12 be a
bounded domain in IlBn. For any Cl-function u in 12, we define the Dirichlet
energy of u in St by
E(u) _
Js IVuI2 dx.
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 cp E Co (St),
J(u -}- Ecp) > J(u)
for any .
In other words,
4.5. Exercises
Exercise 4.1. Suppose u(x) is harmonic in some domain in I[8n. Prove that
v(x) _
\j/
Exercise 4.2. For n = 2, find the Green's function for the Laplace operator
on the first quadrant.
Exercise 4.3. Find the Green's 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
L\u=0 in{x>0},
u=cp on{xn=0}.
Exercise 4.4.
R1
/ -C
R-r
u(x) (
4. Laplace Equations
144
Lu=0 inRTh\BR,
u=0 on aBR.
Prove that u - 0 if
.
lim
u(x)
IxI-+oo In xi
=0 form=2,
IxI-goo
Du = f inn,
u=0 on 852.
Prove that
as av
I < C sup .f
Du + cu = f in
'9"
1,
+ au = cp on 852.
Exercise 4.11. Let S2 be a bounded C1-domain and let P and a be continuous functions on 91 with a > ao for a positive constant ao. Suppose
4.5. Exercises
145
-Du + u3 = 0 in Q,
av + au = cp on aSt.
Prove that
1
iul <_
a0
eix lpI.
Du = f in BR.
Prove that
Vu(O)I
lul +
Ra
BRX If I
Hint: In B, set
v(x
x) =
(u(x
x) - u(x , -x)).
for v(0).
Exercise 4.13. Let u be a nonzero harmonic function in Bl C I[8' and set
lim N(r).
(2) Prove that, for any 0 < r < R < 1,
1
Rn-1
u2dS <
IUBR
(T
RR
2N ( R )
rn-1
u2dS.
Remark: The quantity N(r) is called the frequency. The estimate in (2) for
R = 2r is referred to as the doubling condition.
4. Laplace Equations
146
and
,n.
(2) Assume, in addition, that f is Ca in Sl for some a e (0, 1), i.e., for
for any x E ][8n and i = 1,
any x, y E Sl,
1.f (x)
- f()I
-yea.
Chapter 5
Heat Equations
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 arbitrary 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 first 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 solutions of the initial/boundary-value problem and the initial-value problem.
147
5. Heat Equations
148
As in Chapter 4, several results in this chapter are proved by multiple methods. For example, interior gradient estimates are proved by two
methods: the fundamental solution and the maximum principle.
<00.
sup I
xEIE
In other words, the Schwartz class consists of smooth functions in Ian all of
whose derivatives decay faster than any polynomial at infinity. It is easy to
e_Ix12
is in the Schwartz class.
check that u(x) =
fln
(27r)2
eu(x) dx
for any
E 1[8n.
We note that the integral on the right-hand side makes sense for u e S.
In fact,
Il)I
(2ir)
or
sup lul
n
Jfln
<
e Ian,
1
n IIE
(27r)2
This suggests that Fourier transforms are well defined 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 E s and
cl, c2 E C,
149
= ()9)
and
= (-i)"31x'3u(e).
Proof. Upon integrating by parts, we have
f e_&au(x) dx
(27r)2
f (i eu(x) dx = (ie)(e).
(27r)2
v u(S) _
(21)
v'
ti
(27r)2
Jan
P
e-u(x) dx
(-ix)eu(x) dx=(_Z)IQIxQu(e)
JR
The interchange of the order of differentiation and integration is valid beS, we take any two multi-indices a and ,6.
cause x'3u E S. To prove
is bounded in W. For this, we first note
It suffices to prove that a8
that
raie =(i)
=(_i)H&(xI3u) ()
(21)
Hence
sup I
&1() I
<_
1 n
(271-)2
f l a (xu(x)) I dx <0,
-n
since each term in the integrand decays faster than any polynomial because
xQu E S.
u. - a) () =
5. Heat Equations
150
and
u(. - a)() =
(2r)
fRn
e_u(x - a) dx
f e_u(x) dx =
(2ir)
JR
(2ir)
(27r)
dx
x
LI
(u * v)(x)
=
Lemma 5.1.4. Let u, v E S. Then
= (2r)(I().
Proof. By the definition of the Fourier transform, we have
f e_u * v(x) dx
(271-)2 JR
__ (271-) 2
f e-Z
(21)
R
2
1 n
(2ir)
(f
y)ev(y) dydx
y) dxJ dy
l) f ev(y) dy = (2)).
The interchange of the order of integrations can be justified by Fubini's
theorem.
151
e-x2
dx = .
f00
The next result will be useful in the following discussions.
u(x) = e-AHz.
Then
2G() =
I2
n e 4A .
(2A) 2
dx -
(2ir) 2
Rn
00
i
k=1 (27x)2
e-zx1-Axe
dxk.
-00
e-Ztl1- A ts2
(27r) 2 1-00
dt.
-2t7-At2
oo
= e- A
2
dt
00
00
00
e 4A
Vn
2ds =
4A
dz,
IL
- o0
00
dz =
- oo
Hence
00
1
1
(2ir)2
e_t2
1-00 e
dt = .
- (2A)a e 4A .
dt _
_!L
Therefore,
1
(2ir) 2
Rn
ex-2 dx =
(2A) 2
e4A M2
U
We now prove the Fourier inversion formula, one of the most important
results in the theory of Fourier transforms.
5. Heat Equations
152
(5.1.1)
(2r)2n
1
f n eu(e) d
dy d
eZ
(2
uo(1)= e
Since up(x) = uo(-x), we conclude (5.1.1) for u = uo. Now we prove (5.1.1)
for any u E S.
, vn E S such that
n
xj4Jj(x)
2G(x) _
j=1
f
u(x) =
1 t (u(tx)) dt =
j=1
xJ
, n.
u(tx) dt =
j=1
By taking cc E C( W) with cc = 1 in
(X)WX) +
I)
= j=1
, n. This
153
1n
S 2= (`,7f)2 fRn
(2ir) 2 JRfl
=0.
-1
= (O)
+ (u -
where uo is defined in (5.1.2). First, (5.1.1) holds for uo and hence for
Next, since u - (O) o is zero at x = 0, we see that (5.1.1) holds
xo) = v(O) =
v)d =
(2ir) 2f2
(2ir
1
2 JRn
e0ui(e) d.
v(x) =
(271)2
e Il8"`.
fn
uv dx.
5. Heat Equations
154
Proof. We note
(2;,
11)L2 (Rn)
RTh
2 Jl
/
IRn
(f
fTh
u(x) 1 /
dxl
dx
-Du +u = f
in IlS".
155
(5.1.4)
Then
1(C)
l+lEl
By Theorem 5.1.6,
u(x) =
(5.1.5)
ei
f)
1+ICI
It remains to verify that this indeed yields a classical solution under appropriate 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 differential
(27r)2
RTh
If I2 dx.
(iC)&f(C)
(27r)2
f RTh
eZX
1+2 d.
In particular,
ou(x) _
uxkxk(x) _ k=1
(27r
eX 1e121(e) dC,
1+1C12
RTh
5. Heat Equations
156
and hence
-u(x) + u(x) =
(271)2
f ef() d.
n
I2.
IuI2+2IuI2+
112,
bkSi
k=1
k,1=1
k=1
k,1=1
IuI2+2II2+
k=1
k,1=1
II2+2II2+
1R
l
I
l
2+2u2+
k=1
dx =
k,1=1
IfI2dx.
D
Example 5.1.11. Now we discuss the initial-value problem for the nonhomogeneous heat equation and derive an explicit expression for its solution.
Let f be a continuous function in 1[8n x (0, oo) and uo a continuous function
in R. We consider
1bt - 026 = ,f
(5.1.6)
u(.,0)=uo onR.
Although called an initial-value problem, (5.1.6) is not the type of initialvalue problem we discussed in Section 3.1. The heat equation is of the second
157
^ , t) =
u(
(27r)2
fn
e-
u(x, t) dx.
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,
on Rn.
E W as a parameter.
, t) =
s) ds.
Now we treat t as a parameter instead. For any t> 0, let K(x, t) satisfy
K t) _
e-IEIZt.
(27r)2
Then
,
t) _ (2)(,
t - s).f (, s) ds.
u(x, t) =
(5.1.7)
K(x - y, t)up(y) dy
J J
for any (x, t) E I[8n x (0, oo). By Theorem 5.1.6 and Proposition 5.1.5, we
have
K (x, t) =
(2) n fRfl
e e,
ZX' _I I2t d
or
(5.1.8)
K(x, t) _
(47rt) 2
1x12
4t ,
for any (x, t) E ][8n x (0, oo). 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 defines a solution of the initialvalue problem for the heat equation under appropriate assumptions on the
initial value uo and the nonhomogeneous term f. We will pursue this in the
next section.
5. Heat Equations
158
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
ut - Du=O,
(5.2.1)
for u = u(x, t) with x E ][8n and t E R. We note that (5.2.1) is not preserved
by the change t H -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' _ Ax and t' _ A2t for any nonzero constant A, which leave the quotient
1x12/t invariant. Due to this fact, the expression x12/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 ][8n x I[8, then for any (xo, to)
159
for any (x, t) E ][8n x (0, oo) and A> 0. To do this, we expand u as a power
series of t with coefficients given by functions of x, i.e.,
u(x,t) =
ak(x)tk.
k=0
a0=P, ak =
iiskp.
ak _
u(x't)
kP(x)tk'
k-o
ul(x,t) =x,
U2(X, t) = x2 + 2t,
263x, t) = x3 + 6xt,
u4(x, t) = x4 + 12x21 + 12t2,
us(x, t) = x5 + 20x3t + 60xt2.
K (x, t) =
if
4t
(4irt)
and
(5.2.4)
u(x, t) =
Definition 5.2.1. The function K defined in I[8n x (0, oo) by (5.2.3) is called
the fundamental solution of the heat equation.
5. Heat Equations
160
Lemma 5.2.2. Let K be the fundamental solution of the heat equation defined by (5.2.3). Then
(1) K(x, t) is smooth for any x E 1[8n and t> 0;
(2) K(x, t) > 0 for any x E 1[8n and t> 0;
(3) (at - 0)K(x, t) = 0 for any x E ][8n and t> 0;
(4)
K(x, t)dx = 1 for any t > 0;
(5) for any 6> 0,
lim
t-+O+ Rn\B5
K(x, t) dx = 0.
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
K(x, t)dx =
?t2
f l>2f
e-'' dry.
= K(.,t2)
Now we are ready to prove that the integral formula derived by using
Fourier transforms indeed yields a classical solution of the initial-value problem for the heat equation under appropriate assumptions on u0.
29Z Rn X (0,00).
161
u(x,t) = uo (xo)
lira
(x,t)-+(xo,U)
We note that the function u in (5.2.4) is defined only fort > 0. We can
extend u to {t = 0} by setting
0) = uo on ][87. 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).
Proof. Step 1. We first prove that u is smooth in Ian x (0, oo). For any
multi-index a e Z+ and any nonnegative integer k, we have formally
In order to justify the interchange of the order of differentiation and integration, we need to check that, for any nonnegative integer m and any
t>0,
Ix - yltme
mat
12
This follows easily from the exponential decay of the integrand if t > 0.
Hence u is a smooth function in 1[87 x (0, oo). Then by Lemma 5.2.2(3),
y,t)uo(y) dy = 0.
We point out for future references that we used only the boundedness of uo.
Step 2. We now prove the convergence of u(x, t) to uO(xO) as (x,t) (xO, 0). By Lemma 5.2.2(4), we have
uO(xO)
Then
u(x,t) - uO(xO)
= f nK(x -
y,t)(uo(y)
uo(xo)) dy = I1 + I2,
where
Ii =
. .
fB5(xo)
...,
I2Wz\BS (xo)
5. Heat Equations
162
for any y E Bb(xp), by the continuity of uo. Then by Lemma 5.2.2(2) and
4)
Ilil <
Ba(xo)
Since uo is bounded, we assume that IuoI <M for some positive constant
M. We note that Ix - I > 6/2 for any y E ][8Th \ Bb(xo) and x E Bb12(xo).
By Lemma 5.2.2(5), we can find a b' > 0 such that
Lfl\B6 (xo)
K(x-y,t)dy
for any x E Bb12(xo) and t E (0, S'), where S' depends on s and S = S(s),
and hence only on e. Then
1I21 <
f"\Ba (moo)
K(x -
Iuo(xo)I)
s.
Therefore,
anyt>0,
su
(4irt)2
Iuoldx.
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 uo. To seek a reasonably more general assumption on initial values, we examine the expression for the fundamental
solution K. We note that K in (5.2.3) decays exponentially in space variables 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 fixed exponential growth from initial values can be offset by the fast exponential decay in the fundamental solution, at least for a short period of
time. To see this clearly, we consider an example. For any a > 0, set
G(x,t)=
(1 - 4at) 2
el
4t1X12
Gt-OG=O.
163
Note that
G(x, 0) =
initially for t = 0, and in fact for any t < 1/4a. The growth rate becomes
arbitrarily large as t approaches 1/4a and G does not exist beyond t = 1/4a.
Now we formulate a general result. If uo is continuous and has an exponential growth, then (5.2.4) still defines a solution of the initial-value
problem in a short period of time.
Theorem 5.2.5. Suppose up E C(Rn) satisfies
ut - L u = 0
in IRn X
0,
4A
u (x, t) = uo (xo) .
(x,t)-+(xo,o)
I<-
(4irt) 2
1n
dy.
44AtIY
2
.
1k!
n e 1-4At ICI
(4irt) 2
- (1-4At) 2
1-4At
1n
4t
1-4At
2
I
dy
12
e 1-4At
A
5. Heat Equations
164
dy < oo,
fRn
for any m > 0. The proof form > 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 up(xp) as (x, t) O
(xO, 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 fixed x E ll8n and t> 0, the value of
u(x, t) depends on the values of uo at all points. Equivalently, the values of
up near a point xo E I[8n affect the value of u(x, t) at all x as long as t> 0.
We interpret this by saying that the effects travel at an infinite speed. If the
initial value uo is nonnegative everywhere and positive somewhere, then the
We need to point out that (5.2.4) represents only one of infinitely 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 E C (1Rn x ][8) of ut - Du = 0,
=0
u(.,0)=0
165
u(x, t) = >ak(t)xk.
=o
Making a simple substitution in the equation ut = uxx and comparing the
coefficients of powers of x, we have
a''_2=k(k-1)ak foranyk>2.
Evaluating u and ux at x = 0, we get
ao=a,
a1=0.
a2k(t) =
(2k)!
a(k)(t),
and
a2k+i (t) = 0.
Therefore, we have a formal solution
00
k=O
a(t) _
eU
fort > 0,
tort < U.
on R.
5. Heat Equations
166
sin(mx),
corn(x) =
-f oo as m -f oo.
R2
For any domain D in IRn x Ilt, we denote by C2" (D) the collection of
functions in D which are C2 in x and C1 in t.
We first have the following regularity result for solutions of the heat
equation.
167
Proof. For simplicity, we consider the case (xO, to) _ (0,0) and write
QR = BR x (-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 fixed point (x, t) E QR and claim that
avy
1 Rz asR K(x-y,t-s)(y,s)
We first 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 x (-R2, t]. The first integral is over
BR x {-R2}. For (x, t) E QR, it is obvious that t + R2 > 0 and hence
there is no singularity in the first integral. The second integral is over
8BR x (-R2, t]. By the change of variables r = t - s, we can rewrite it as
t+R2
T a lye t - T -
t - T)
5K
dSydT.
K(y, s) = K(x - y, t - s) =
(47-(t-s))2
Ix-y12
e 4W
)
for s< t.
Then
R3+zk=o.
Hence,
(uky2
z=i
7t
= (uk)8 +
(uky2 - Kuy2)y2
2=1
u(Ks + DyK)
5. Heat Equations
168
For any e > 0 with t - e > -R2, we integrate with respect to (y, s) in
BR x (-R2, t - e). Then
BR
K(x - y, e)u(y, t - e) dy
Ii
R2
FK(X
- y,t -
avy
BR
-u(y, s)
0K
EGO BR
The proof proceeds similarly to that in Step 2 in the proof of Theorem 5.2.3.
The integral here over a finite domain introduces few changes here. We omit
the details.
I c sup
ui,
R QR(xo,to)
Proof. We consider the case (xO, to) _ (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
Q,. = Br x (-r2, 0] for any r E (0, 1].
We first modify the proof of Theorem 5.2.7 to express u in terms of the
fundamental solution and cutoff functions. We denote points in Qi 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 fixed (x, t) E Ql/4,
K(y, s) = K(x - y, t - s) =
(47r(t_s))2
for s <t.
and cp - 1
169
We need to point out that v(y, s) is defined only for s < t. For such a
function v, we have
n
u(ZJs +
i=1
For any e > 0, we integrate with respect to (y, s) in Bl x (-1, t-e). We note
that there is no boundary integral over Bl x {-1} and aBl x (-1, t - e),
since cp vanishes there. Hence
f (u)(y,t - ')K(x - y, e) dy = f
l
x(83 +
)(R)dyds.
i x (-L,t-E)
(x, t)u(x, t) = f
u(as +
dyds.
l x (-1,t)
In view of
k8+zk=o,
u(x, t) =
JB1 x (-it)
We note that each term in the integrand involves a derivative of cp, which is
zero in Q1/2 since cp - 1 there. Then the domain of integration D is actually
given by
D2
D2 t
Dl
vxu(x, t) =
5. Heat Equations
170
Iosl + IVcoI C C.
Hence
- ynI 1e c _Ix
(t - s) 2+
Io KI
and
-gtys)
l rylx-'LJI2+(t-S)
(t-s)2 e
x-yJ<1, 0<t-s<1.
Therefore, for any (x, t) E Q1/4,
2
t) I
C
i=1
(t
n+z e
s) 2
_ Ix-yI2
(t-s)2+z
_ Ix-yI2
4(t-s)
C.
Q1
Dl = BZ x (-(3/4)2, _(1/2)2),
We first consider Dl. For any (x, t) E Q1/4 and (y, s) E Dl, we have
t-s> g,1
and hence
1
.e - 4(t-s)
Ix-yI2
(t-s) 2
n+Z
82
Next, we consider D2. For any (x, t) E Q1/4 and (y, s) E D2, we have
13\2
y-x> 4, 0<t-s< 1I
171
-'
44
= T 2 +Ze
- C,
a
for any 'r> (4/3)2. This finishes the proof of the claim.
Next, we estimate derivatives of arbitrary order.
Im
QR(xo, to) for some (xO, to) E ]E8n x ][8 and R> 0. Then for any nonnegative
integers m and k,
R + 21
nkem+2k-1(m + 2k)!
sup
IUI,
QR(xo,to)
Rm
sup
IUI.
QR(xo,to)
I nk
max
IIQI=m+2/c
The next result concerns the analyticity of solutions of the heat equation
on any time slice.
5. Heat Equations
172
ut - Du = f
u(.,O)=O
onRTh.
K(x, t) _
1x12
(4irt) 2
e- 4t
(5.2.5)
u(x, t) =
fJ
for any (x, t) E ][8n x (0, oo). If f is bounded in ll8n x (0, oo), it is straightforward to check that the integral in the right-hand side of (5.2.5) is well
defined and continuous in (x, t) E ll8Th x (0, oo). By Lemma 5.2.2(4), we have
t
sup
If I
11n x (o,t)
Hence
ast-+0.
fn
To discuss whether u is differentiable, we note that
1
x2 _ 1j
(x, t) _ - (4irt) n2 2t e 4t
(x,t) =
(xx _ Szj
4t2
(4irt)
Ii
4t
2t
e -Iz12 1z2I dz =
-7r2/ n
Vt
1
n
in
and
1
J IKxx(x,t)Idx=
n
Hence
u(x, t) _
z2 zi -
bi j
edz.
ff
(x - y, t - s) f (y, s) dyds.
fl
173
ff
t
sup
If I
R X (o,t)
(x - y, t - s) I dyds
n
1
ds =
If I Jo
2\/
sup
If I
(t - s)
(,t)
R
Hence, the integral in the right-hand side of (5.2.6) is well defined and continuous in (x, t) E 1[8n x (0, oo). 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 differentiating (5.2.6) to get the second xderivatives of u if f is merely bounded, since
Ll (][8' x (0, T)) for
any T > 0. In order to get the second x-derivatives of u, we need extra
assumptions on f.
x (o ,t)
ut - Du = f
u(x, t) = 0.
lim
(x,t)-+(xo,O)
(5.2.7)
7r 2
ff
eII2 f (x + 2z
s) dzds,
for any (x, t) E 1[8n x (0, oo). It follows easily that the limit of u(x, t) is zero
ast-*0.
A simple differentiation yields
u(x,
t) =
x2
72
e-Izl2Dx2,f(x
1n
o2
t- s s) dzds
+ 2z,
2 t-s af (x + 2z
1
72
n
Upon integrating by parts, we have
uxi(x, t) =
n
?f 2
Rn
z2
t-5
f(x + 2z t
s) dzds.
s, s) dzds.
5. Heat Equations
174
1n ff _Iz12
e
7r2
z2
t-5
fXj (x + 2z t
- s, s) dzds.
ut(x, t) =
1n
7.2
e_Iz12f
(x, t) dz
n
7T 2
e_I zl2
z2
n 2=1
(x + 2z t
- s, s) dzds.
continuous in (x, t) E Rn x (0, oo). We note that the first term in the righthand side of ut (x, t) is simply f(x, t). Hence,
n
Cx t) = f(x,t),
i=1
utt = Dut + ft = 0
+1) + ft.
Hence wtt and all its x-derivatives exist and are continuous in I[8x (0, oo).
+1!
K(x-y,t-s)f(y,s)dyds
en
is a solution of
26t - L2G = f
U(,0)U0 onR.
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
175
StT=Stx(O,T]={(x,t): xESt,O<t<T}.
We note that StT includes the top portion of its geometric boundary. The
parabolic boundary BpStT of StT is given by
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 12T. We should point out that the results in
this subsection hold for general domains in Rx R.
We first prove the weak maximum principle for the heat equation, which
asserts that any subsolution of the heat equation attains its maximum on
5. Heat Equations
176
ut-Du<0 inS2T.
Then u attains on 8PS2T its maximum in S2T, i. e.,
maxis = max u.
ap T
Proof. We first consider a special case where ut - Du < 0 and prove that
u cannot attain in StT its maximum in StT. Suppose, to the contrary, that
there exists a point Po = (xo, to) E StT such that
u(Po) =maxis.
Then
is nonpositive definite.
U6(X,t) =U(x,t)-Et.
Then
(D,-L)U6=ut-ou-E<0.
By the special case we just discussed, u6 cannot attain in SZT its maximum.
Hence
max u6 = max u6 .
8p T
Then
max u (x, t) = max(u6 (x, t) + Et)) < max u (x, t) + ET
ap cT
177
ut-DU+cu<0
in S2T.
SZT
Theorem 5.3.3. Let c be a continuous function in S2T with c > -co for a
nonnegative constant co. Suppose u E C2" (SZT) fl C(S2T) satisfies
ut - Du + cu
in S2T,
e-C0tu+
= 0.
Corollary 5.3.4. Let c be a continuous function in SZT with c > -co for a
nonnegative constant co. Suppose u, v E C2"(S2T) f1 C(T) satisfy
ut - Du - cu < vt - Ov - cv
Then u<vinlT.
in S2T,
5. Heat Equations
178
in SZ
and
Leu<LevinSZ, u<vonaSZ
u<vinSZ,
5.3.2. The Strong Maximum Principle. The weak maximum principle asserts that subsolutions of parabolic equations attain on the parabolic
boundary their nonnegative maximum if the coefficient of the zeroth-order
term is nonnegative. In fact, these subsolutions can attain their nonnegative 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 suffices for most applications to the
initial/boundary-value problem with values of the solutions prescribed on
the parabolic boundary of the domain.
We first prove the following result.
Lemma 5.3.5. Let (xo, to) be a point in ][8n x ][8, R and T be positive
constants and Q be the set defined by
Q = BR(XO) x (to - T, ta].
179
ut - Du + cu > 0 in Q.
If u > 0 in Q and
D = BR x (-cR2,0].
By the assumption u(0, -aR2) > 0 and the continuity of u, we can assume
that
u(x, -aR2) > m for any x E BER,
for some constants m > 0 and E (0, 1). Here, m can be taken as the
(positive) minimum of
-aR2) on BER
Now we set
(
Do={(x,t)EBRx(-aR2,0]:
Ixl2-1
l
It is easy to see that
t<R2 CD.
Don{t=0}=BR, Doff{t=-aR2}=BER
Set
2
wl (t)
t + R2
a t +R2-1x12,
5. Heat Equations
180
We first note that e2R2 < wl < R2 and w2 > 0 in Do. A straightforward
calculation yields
2(1 - e2)W1w21
= wi Q_1 (-Q(1 a
and
wt - Ow + cw = w1 _1
1(8w?
- (4n + 8)w1w2).
(((1_E2))2
a
-I- 4n -I- 8
w1w2 -8w?
Hence
1
21
+ 4n +8) 2v1w2 -I- 82v1).
wt - Ow + cw < 0 in Do.
181
and 2
given by
= {(x,t)
_ {(xt): Ixl2 _
For (x, t) E El, we have t = -aR2 and lxi < ER, and hence
w(x, -aR) _
(E2R2)-Q(E2R2
_ Ix12)2
(ER)-2Q+4
v = m(ER)24w in Do,
where m is the minimum of u over
vt - Ov + cv < 0 in Do,
and
v < u on BPDo,
since u > m on El and u > 0 on E2. In conclusion,
vt - Ov + cv < Ut - pu + cu in Do,
v < u on BpDo.
By the maximum principle, we have
v < u in Dp.
This holds in particular at t = 0. By evaluating v at t = 0, we obtain
u(x, 0) > mE2Q-4 (1
- IR2 )
We point out that the final estimate in the proof yields a lower bound
of u over BR x {0} in terms of the lower bound of u over BER x {-cJl2}.
This is an important estimate.
Now, we are ready to prove the strong maximum principle.
5. Heat Equations
182
M= sup u> 0,
1 x (O,T]
v=M-u in1x(0,TJ.
Then v(x*, t*) = 0, v > 0 in 1 2 x (0,TJ and
To this end, we connect (xO, to) and (x*, t*) by a smooth curve ry C
Sl x (0, TJ along which the t-component is increasing. In fact, we first connect
xo and x* by a smooth curve -yo = yo(s) C St, for s E [0, 1], with yo(O) = xo
and yo(l) = x*. Then we may take ry to be the curve given by (yo(s), st* +
(1 - s)to). With such a -y, there exist a positive constant R and finitely
(x*, t*)
N-1
-y c U
=o
183
ut-Du+cu<0 in D.
Assume, in addition, for some x e ]I8n with x - xol = R, that
u(x, t) < u(x, to) for any (x, t) E D and u(x, to) > 0,
u(x, t) < u(x, to) for any (x, t) E D with Ix - xol < R.
v = (x - xo)/Ix - xol
Proof. Without loss of generality, we assume that (xO, to) _ (0, 0). Then
{(xt)
A direct calculation yields
- 2m a - a - c) - ce-RZ
G -e-aI2-t) (42IxI2 - 2na - rya - c),
wt - Ow + cw =
(4a2IxI2
where we used c > 0 in D. By taking into account that R/2 < IxI < R in
Do and choosing a sufficiently large, we have
4a2IxI2
5. Heat Equations
184
and hence
{(x,t):
First, on E1i we have u - u(x, 0) <0, and hence u - u(x, 0) <-s for some
s> 0. Note that w < 1 on El. Then for such an e, we obtain v <0 on E1.
Second, for (x,t) E E2, we have w(x, t) = 0 and u(x, t) < u(x, 0). Hence
v(x, t) < 0 for any (x, t) E E2 and v(x, 0) = 0. Therefore, v < 0 on E2. In
conclusion,
vt - Ov + cv < 0 in Do,
v<0
By the maximum principle, we have
v<0 in Do.
Then, by v(x, 0) = 0, v attains at (x, 0) its maximum in Do. In particular,
v (x, 0) < v (x, 0)
for any x e BR \ B 2 R.
Hence, we obtain
and then
>
au
(x, 0) _
av
13w
-sa(x,
0) _
v
2scxRe_
R2
> 0.
185
5.3.3. A Priori Estimates. In the rest of this section, we discuss applications of the maximum principle. We point out that only the weak maximum
principle is needed.
As the first application, we derive an estimate of the sup-norms of solutions 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 1 C Rn and a positive constant T, we
set
12T=12x(0,T]={(x,t): xE12,0<t<T}.
Theorem 5.3.8. Let c be continuous in StT with c > -CO for a nonnegative
constant co. Suppose u e C2" (fT) fl C(T) is a solution of
ut - Du -I- cu = f
in StT,
u(.,0)=u
onf,
u = cp
sup
asp X (o,T)
cz
asp X (o,T)
F = sup if I.
ciT
5. Heat Equations
186
Then
fu < B on 3'.
Set
v>B on3f.
Hence,
fu < v
on apS2T.
fu < v
in StT.
Therefore,
Theorem 5.3.9. Let c be continuous in I[8n x (O, T] with c > -co for a
nonnegative constant co. Suppose u e C2"(]E8n x (O, T]) fl C(II87 x [O, T]) is
a bounded solution of
ut - Du + cu = f
onTW,
Then
U I < ecOT
II
R x (o,T)
If
187
F= sup
1Rx (o,T]
fl,
Il2
Then
fu < B on ]I8".
Since u is bounded, we assume that ui < M in Rn x (0, T] for a positive
constant M. For any R> 0, consider
w (x, t) = eC0t (B + Ft) + VR (x, t)
in BR x (0, T],
in BR x (0, T].
Moreover,
yR
inBR,
vR > fu on 8BR X [0, T].
To construct such a vR, we consider
vR(x, t)
=Rept(2nt + IxI2).
R2
eC0t(2nt + ix12).
Now we fix an arbitrary (x, t) E Il8n x (O, T]. By choosing R> lxi and then
letting R -+ oo, we have
lu(x,t)I :; ecot(B+Ft).
5. Heat Equations
188
Ut-LU=O inR7'x(O,T],
onR7.
Suppose, for some positive constants M and A,
Iu(x,t)I
u=0 inRn x
0,
4a
,untilt=T.
[ 4 , 4
vR(x, t) =
n e1-4t,
(1 - 4at) 2
atvR O vR =
in
BR
R x (o,
4a
Obviously,
0) > 0 =
Next, for any (x, t) E DBR x (0,1 /4a),
in BR.
vR(x, t) >
In conclusion,
ll
fu<vR inBRx
Therefore,
I< vR(x, t)
4a)
189
Now we fix an arbitrary (x, t) E ]I8" x (0,1/4a) and then choose R > xI
We note that vR(X, t) -+ 0 as R -+ oo, since a > 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 X (r2,0].
Theorem 5.3.11. Suppose u E C2"(Q1) fl C(Ql) satisfies
ut-Du=O inQl.
Then
Proof. We first note that u is smooth in Q by Theorem 5.2.7. A straightforward calculation yields
n
(at - )IVuI2 = -2
(ut
+2
i,j=1
n
i=1
= -2
z,j=1
we have
-4
( uu- 2cp
i,j=1
i,j=1
Now we take cp = rj2 for some r) E C(Q1) with r) - 1 in Q1/2 and supp r) C
Q3/4. Then
(at -
(2rnt - 2rjL
j_
- 8j
i,j=1
i,j=1
I`8j
2r)2
WT ' +
9
5. Heat Equations
190
Hence,
ioxui2
(at -
<
2u(ut - Du) _
(at - L)(u2) _
au2) <(C -
(at -
<0.
aPQi
5.3.5. Harnack Inequalities. For positive harmonic functions, the Harnack 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 solutions of the heat equation, we begin our discussion with the fundamental
solution.
We fix an arbitrary e Rn and consider for any (x, t) E Rn x (0, oo),
1
u(x, t) =
(4irt) 2
e-
Ix-12
4t
_u(x2, t2)
26(x1, t1)
(t2
t1
Ix2_I2
4t2
Ixl_I2
4tl
Recall that
(p+q)2
p2
a+b
q2
+ b'
for any a, b > 0 and any p, q e R, and the equality holds if and only if
by = aq. This implies, for any t2 > tl > 0,
1x2
t2
e=
12x1 - t1 x2
t2 - t1
191
Therefore,
-ti
t2
u(xi, ti) $
1x2-x112
4(t2-tl)u(x2, t2),
for any x i , x2 E W and any t2 > ti > 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(xi, ti) < Cu(x2, t2),
the constant C should depend on t2/ti, 1x2 - x i , and most importantly
u(x, t ) =
ne
_ 1x12
4t
(4irt)2
Then
v(x, t) = l og u (x, t)
and hence
n
vt
_ - 2 l og( 4t) -
1x12
2t + 4t2 '
2
I
4t
'
x
v
2t
Therefore,
2t +
vt
IVVI2.
We have the following differential Harnack inequality for arbitrary positive solutions of the heat equation.
ut = Du,
u>0
VV
in RTh x (0,T].
5. Heat Equations
192
ut = Du,
Then for any (Xi, tl), (x2, t2) E II87 x (0, TJ with t2 > ti > 0,
u(xl,tl)
t2
u(x2, t2)
\ti j
11x2-x112
4(t - ti)
t) = vt + Ov dt > IVvI2 - v dt - 2t
d> -t
tv(x(t), t)
I dI
2t
-4
log t?1
dx
dt.
dt
tl
To seek an optimal path which makes the last integral minimal, we require
d2x
dt 2
=0
x(t) = at + b.
Since x2 = at + b, i = 1, 2, we take
x2 xi
a=
b = t2xi
t2 - ti
tix2
t2 - ti
Then,
t2
tl
dx
dt = 1x2-x112
t2-tl
dt
Therefore, we obtain
2
v(xi,ti)
or
u(xl,t1) <
u(x2,t2)
ti
-tl
t2
ex
t2 - tl
12- xll2
4(t2 - ti)
0
193
constant s E (O, T) and then let E -4 0. We divide the proof into several
steps. In the following, we avoid notions of summations if possible.
Step 1. We first derive some equations involving derivatives of v = log u.
A simple calculation yields
vt = Ov -I-
IVvI2.
Consider w = Ov . Then
Wt = Ovt =
Since
i,j=1
V.
n
i=1
(Vxx)i
i=1
= n-1 (oV)2.
5. Heat Equations
194
-w2.
n
Step 2. For a constant a E (0, 1), set
f=
Vt.
Then
= -w - (1- c)t,
and hence by (5.3.1) and (5.3.2),
ft-LV-2Vv'Vf=-2oIV2vI2.
Next, we estimate 102v12 by f. Note that
n (lVvI2
IV2vI2>
+ 2(1 - )IVvI2f).
We obtain
(5.3.3)
ft-Of
<-2n (f2+2(1-a)IVvI2f).
We should point out that 1Vv12 in the right-hand side plays an important
role later on.
Step 3. Now we introduce a cutoff function cp E C( W) with cp > 0
and set
g=
We derive a differential inequality for g. Note that
9t = Pf +
t f VcP,
195
Then,
9t -
f = v9 -
v-s,
(vg -
tcp0 f = Dg - 2
g)
=g-2.Vg+(2
)g.
Multiplying (5.3.3) by t2cp2 and substituting It, V f and L f by above equalities, we obtain
09) -I- 2t(Vcp - (ppv) Vg
< g Jcp
l
2a
- ng
+ t (2
4a(1 - a)
n
OcP
2Vcp pv/
To eliminate IVvI from the right-hand side, we complete the square for the
last two terms. (Here we need a < 1! Otherwise, we cannot control the
expression -2V(p Vv in the right-hand side.) Hence,
tcP(gt - D9) + 2t(Vcp - (pOv) Vg
v2
2n1V12
-DAP+4a(1
(p
l
)3'
whenever g is nonnegative. We point out that there are no unknown expressions in the right-hand side except g. By choosing cp =
r > 0, we get
r
<g
2
- iJZVv)
'
V9
I2
e Co (B1), with
1 in Bi12. For any fixed R > 1, we consider
0<
1 in Bl and
Then
(6IvI2 -
-I-
a(1- a)
IVI2) (x)
a(in a)
= R2
CR)
Dg) +
Ctl
(2a
Vg < g 1 - n g -I- RZ
5. Heat Equations
196
2Z
l---+-->
in BR x (O, T].
2a
Cat
h-1- ng+
R2
has a negative minimum at (xO, to) E BR x (0, T]. Hence,
h(xo,to) < 0,
and
and
2c
Cat"
1- 2n tr)2(aiVvi2 - vt) + R2
> 0 in BR x (0,T].
For any fixed (x, t) E Ilgn x (O, T], choose R> lxi. Recall that r,i =
and rjo = 1 in B112. Letting R -+ oo, we obtain
2c
2
1--t(ciVVi
-Vt) 0.
72
5.4. Exercises
197
mum principle: Let u be a nonnegative solution of the heat equation ut Du = 0 in Bl x (0, 1]. If u(xo, to) = 0 for some (xO, to) E Bl x (0, 1], then
u = 0 in Bl x (0, to]. This may be interpreted as infinite-speed propagation.
5.4. Exercises
Exercise 5.1. Prove the following statements by straightforward calculations:
alxl2
(2) For any a > 0, G(x, t) = (1- 4at)- 2 e 1-4t satisfies the heat equation for t < 1/4a.
t-+oo
5. Heat Equations
198
Exercise 5.4. Let uo be a bounded and continuous function in [0, oo) with
uo(0) = 0. Find an integral representation for the solution of the problem
ut - u= 0
for x> 0,
u(0, t) = 0 fort > 0.
u(x, 0) = uo (x)
2Gt - 02G = 0
C(1 + lxi +
for any (x, t) E II8"` x (-oo, 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 ][8 x R.
ut - Du = 0 in S2 x (0, oo),
U(,0)U0 on1,
ii=0 onD1x(0,oo).
Prove that
st
U(,0)U0 on1,
U=0 onD1x(0,T).
Prove that
0 < u < eCOT sup uo
in 1 x (0,TI.
5.4. Exercises
199
u(',0)=u onh,
U=(p onOhx(0,T).
Prove that
sup
cp }
as)x(o,T)
I. s
Exercise 5.10. Let Q = (0,1) x (0, oo) and uo E Cl [0, l] with uo(0) _
ut - uxx= 0 in Q,
U(., 0) = U0
on (0,1),
U(0,.)=U(l,.)=0 on(0,oo).
Prove that
[0,l]
,Urn E
Otui = Dui
for i = 1,
in SZ x (0, T],
f(U,,... ,Urn).
sup
8 (SZ x (O,T] )
S2 x (O,T]
on TW .
t)I <
2tsup
n duo
M2.
200
5. Heat Equations
Chapter 6
Wave Equations
that solutions decay at certain rates as time goes to infinity. We will also
solve the initial-value problem for the nonhomogeneous wave equation by
Duhamel's principle.
In Section 6.3, we discuss energy estimates for solutions of the initialvalue 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 hyperbolic equations with initial values prescribed on space-like hypersurfaces
201
6. Wave Equations
202
are well posed. We point out that energy estimates are fundamental and
form the basis for the existence of solutions of general hyperbolic equations.
onllt
We will derive expressions for its solutions by several different methods.
Throughout this section, we denote points in I[8 x (0, oo) by (x, t). How-
Cl(x,t)={(y,s):
This is the cone we introduced in Section 2.3 for n = 1. We usually refer to
C, (x, t) as the characteristic triangle.
We first consider the homogeneous wave equation
(6.1.2)
=x-t, r=x+t.
In the new coordinates, the wave equation has the form
0.
203
D
X
(6.1.4)
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),
xB - tB = xA - tA,
xB + tB = XD + tD,
and
6. Wave Equations
204
Then
g'(x) =
g(x) =
(x) - LX
fs
h(x) =
(x) +
(s)ds - C.
0
Therefore,
u(x, t) = 2
(6.1.5)
t) + cp(x + t)) + 2
t t (s)
ds.
t) for
This is d'Alembert's formula. It clearly shows that regularity of
0) and is 1-degree better
any t> 0 is the same as that of the initial value
than
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 Cl (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 (xO, 0) of the x-axis influence u(x, t) at points
(x, t) in the wedge-shaped region bounded by characteristic curves through
(xO, 0), i.e., for xo - t <x < xo + t, which is often referred to as the range
of influence.
tl
(6.1.6)
205
111 RX (0,00),
U(.,0)=0 on ][8,
and
vt + v = 0 in III x 0, oo ,
(f M 71
v(.,0)=b onR.
The initial-value problem (6.1.7) was discussed in Example 2.2.3. Its solution
is given by
(x + t - 2r) dr.
fx+t
U(x,t)=
b(s)ds.
/
x-t
This is simply a special case of d'Alembert's formula (6.1.5).
Now we derive an expression of solutions in the general case. For any
2
L_
= {(y,s)
s=-y+x+t, 0<s<t},
s=y-x+t, 0<s<t},
and
Lo={(y,0): x-t<y<x+t}.
We note that L+ and L_ are parts of the characteristic curves through (x, t).
Let v = (v1, 1)2) be the unit exterior normal vector of 8C1 (x, t). Then
v=
1(1, 1)//
on L,
(-1, i)/J
((0,-i)
on L_,
onL0.
6. Wave Equations
206
x-t
x+t
fdyds =
i (x,t)
f+
VG
(UtV2 - uxvl) dl
ci (x,t)
f_
(Ut + ux) dl
VG
x+t
ut(s, 0) ds,
x-t
(ut -
J+
(ut +
(s)ds
(6.1.8)
+-2
1
x+(t-T)
/ /
0
x-(t-T)
f(y,'r)dyd'r.
utt - u = f
207
u(x, t) = cp(xp),
lim
(x,t)-a (xo,O)
ut(x, t) _ b(xo).
lim
(x,t)-; (moo ,O)
obtain energy estimates, we take any constants 0 < T < t and use the
domain
{(x,t): IxI<-t,0<t<T}.
We postpone the derivation until the final section of this chapter.
utt -
u(.,0)=
(6.1.9)
on ,oo),
(0)
a(0),
/'(0)
= a'(O),
"(0)
= a"(O).
6. Wave Equations
208
o(O) = 0,
o"(O) = 0.
fco(x)
fi(x)
{ -(-x)
for x > 0
for x <0.
utt -
ut(,0)=1
inJR.
u(0,t)=0 foranyt>0.
In fact, for v(x, t) _ -u(-x, t), a simple calculation yields
vtt - v= 0
+t
(s) ds.
By restricting (x, t) to [0, oo) x [0, oo), we have, for any x > t > 0,
L-Ft
ds,
and foranyt>x>0,
(6.1.11)
/' +t
J _ (s) ds,
since cp and b are odd in 1[8. We point out that (6.1.11) will be needed in
solving the initial-value problem for the wave equation in higher dimensions.
209
112={(x,t): t>x>0}.
We denote by ul the solution in Stl. Then, ul is determined by (6.1.5) from
the initial values. In fact,
x+t
sb(s) ds,
/
-'Jx-t
2f
(s) ds.
We note that '-y(x) is the value of the solution u along the straight line t = x
for x > 0. Next, we consider
u2 (x , t) + u 2 I
t-x tt- x ) =u
2
2
2
2
t+x
(0t-x)+u \ t+x
2
2
,
2 (
Hence
f_
x+t
(s)ds,
for any (x, t) E 112. Set u = ul in hi and u = u2 in 112. Now we check that
u, ut, ux, utt, uxx, utx are continuous along {t = x}. By a direct calculation,
6. Wave Equations
210
we have
= u2,
on {t = x}.
on {t = x} by ul = u2 and
8xtu2 and attui =
8xu2 on
on {t = x}.
Theorem 6.1.2. Suppose cc e C2[0, oo), b e Cl [0, oo), a e C2 [0, oo) and
the compatibility condition (6.1.10) holds. Then there exists a solution u E
c2([0, oo) X [0,oo)) of (6.1.9).
We can also derive a priori energy estimates for solutions of (6.1.9). For
any constants T> 0 and xo > T, we use the following domain for energy
estimates:
{(x,t): 0<x<xo-t,0<t<T}.
Now we consider the initial/boundary-value problem. For a positive
constant l > 0, assume that cc E C2[0, l], b E Cl [0,1] and a, Q E C2 [0, oo).
Consider
= 0 in (0,1) x (0, oo),
tt -
(6.1.12)
_ cc, ut(',O) _
on [0, l],
u(0, t) = a(t), u(l, t) _ /3(t) fort > 0.
O)
c(l) = /3(0),
= f3"(O).
211
x _ Ico(x)
utt -
onR.
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
We first 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, 1/2) 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 find u in the entire region [0,1] x [0, oo).
6. Wave Equations
212
Theorem 6.1.3. Suppose cp e C2 [0, l], E Cl [0, l], a, Q E CZ [0, oo) and
the compatibility condition (6.1.13) holds. Then there exists a solution u e
C2([0, l] x [O,oo)) of (6.1.12).
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 1 2 in the positive t-direction.
In this section, we discussed cases where SZ is given by the upper half-space
ll8 x (0, oo), the first quadrant (0, oo) x (0, oo) and I x (0, oo) for a finite
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 tdirection. In summary, we have
ult=o = P, utIt=o = '/' for initial-value problems;
ult=o = o, utIt=o = '/', uIx=o = a for half-space problems;
uIt=o = (P, utIt=o = '/', uIx=o = a, Ix=1 = Q
for initial/boundary-value problems.
213
'att (6.2.1)
= 0 in R x (0, oo),
u(',0)- o, ut(',0)_ b
Ori fin'.
W (x; r) = wnr n_ 1
w (y) dSy
UBr (X)
where wn is the surface area of the unit sphere in R. Then W (x; r) is the
average of w over the sphere DBr (x) . Now, w can be recovered from W by
To
W (x; r) = w(x)
6. Wave Equations
214
U(x;r,t) =
(6.2.2)
Wnrn-1
(x;r) =
u(y, t) dSy
DBr
(fi_2
t), cp
In other words, U(x; r, t), J(x, r) and W(x, r) are the averages of
and /i over the sphere aBr(x), respectively. Then U determines u by
for U. We claim that, for each fixed x e Ian, U(x; r, t) satisfies the EulerPoisson-Darboux equation
(6.2.4)
Utt
= Urr+
Ur forr>Oandt>0,
1
GJn
Ur =
(,Jn
au
if
Iwl=1
cvnr n- 1
(x + rw, t) dSw =
au
cvnrn -1 aBr(x)
(y, t) dSy
Du(y, t) dy.
r n-1 Ur =
Wn
fBr (x)
Du(y, t) dy =
Wn Br (x)
utt (y, t) dy
215
Hence
(rn_1
UT )T
Wn
1
Wn
&Br(x)
att
utt(y, t) day
U(x, r,0) =
Wr
Ut(x;r,0) =war _1
(y)dS,
8Br (x)
dSy.
BB,. (x)
6.2.2. Dimension Three. We note that the Euler-Poisson-Darboux equation is a one-dimensional hyperbolic equation. In general, it is a tedious process to solve the corresponding initial-value problems for general n. However,
this process is relatively easy for n = 3. If n = 3, we have
Utt=UTT+
(rU)tt = (rU).
We note that rU satisfies the one-dimensional wave equation. Set
U(x;r,t) = rU(x; r, t)
and
UtUTT forr>Oandt>0,
x; r), Ut(x;r,0) _ lY(x; r) for r >0,
U(x;0,t)=0 fort > 0.
U(x;r,0) _
t - r)) + 2
T+C
(x; s) ds.
-T
Hence,
U(x; r, t) _
r ((t+r)(x;t+r) - (t-r)(x;t-r))
1
t+ T
f+_ ,.
6. Wave Equations
216
Letting r -+ 0, we obtain
u(x, t) = li o U(x; r, t) = 8t (tI(x; t)) -F tiY(x; t).
(x;t)
W(x; t) =
dsy,
b(y) dsy.
u(x, t) = at
47rt
Bt(x)
) dsy,
for any (x, t) E I[83 x (0, oo). 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 cp
and zb.
(x,t)-+(xo,o)
u(x, t) = So(xo),
lim
(x,t)-+(xo,o)
ut(x, t) = b(xo).
x't
b(y) dSy.
W(x,t) =
f(x+tw)dS.wl=1
217
In this form, u(x, t) is defined for any (x, t) E ][83 x [0, oo) and
0) = 0.
Since Eb e Ck(][83), we conclude easily that Vu exists and is continuous in
, k. In particular,
Il83 x [0, oo), for i = 0, 1,
u(x, t) = 4
(x+tw)dS.
JIwl=1
W(x,t)
= 41i=
-(x+tw)dS.
LB(X) 8v
(y) dsy
1f
st(y)
Then
tt = -
JBt(x)
2
utt =
aBt(x)
&y4'(y)
dsy
(y) dSp.
aBt(x)
i9Bt(x)
xb(x +
dSW = Du.
x [0, oo)).
This implies easily that u E
A similar calculation works for zb = 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 + tw in (6.2.5), we have
u(x, t) = at
JWi=1
(x+tw)dS)+f
4ir J
u(x, t) =
4J
IWI=1
((x+tw)+tV(x+tw) w +
(x+tw)dS.
6. Wave Equations
218
Hence
u(x't)
((y) +
(y
- x) + t(y)) dsy,
for any (x, t) E ][83 x (0, oo). We note that u(x, t) depends only on the initial
values co and b on the sphere 8Bt(x).
6.2.3. Dimension Two. We now solve initial-value problems for the wave
equation in Il82 x (0, oo) by the method of descent. Let cp E C2(R2) and
b E Cl(][82). Suppose u E C2(][82 x (0, oo))f1C1(Il82 x [0, oo)) satisfies (6.2.1),
i.e.,
0) = zb
on I[82.
IL(.,0)=@,
t(,0)=
onR3,
where
By (6.2.5), we have
u(x, t) = at
1J
4lrt aBt()
) dsy,
where y = (yl,y2,y3) _ (y> y3) The integrals here are over the surface
8Bt(x) in ][83. Now we evaluate them as integrals in ][82 by eliminating y3.
For x3 = 0, the sphere Iy - x =tin ][83 has two pieces given by
CJs=f
and its surface area element is
1
dSy- _ (1 +
(ay2y3)2) 2 dyidya
= yt2
-
- xl2
dy.
219
Therefore, we obtain
(y)
u(x, t) = 2 at
t2 _ Iy_ xla dy
JBt(x)
(6.2.6)
/'
2 it JBt(x)
1
t2 - I - xl2
dy,
for any (x, t) E 1182 x (0, oo). We put the factor 1/2 separately to emphasize
that it is the area of the unit disc in II82.
Theorem 6.2.2. Let k > 2 be an integer, cp E Ck+l (][82) and 'i/i E C (][82) .
Suppose u is defined by (6.2.6) in ][82 x (0, oo). Then u e C'(][82 x (0, oo))
and
u(x, t) = co (xo ),
(x,t)-+(xo,o)
ut (x, t) =
lim
(xO).
(x,t)-+(xo,o)
This follows from Theorem 6.2.1. Again, u can be extended to a Ckfunction in 1182 x [0, oo).
u(x, t) - at
cp(x - tz)
lB1
dz -
1 - 1z12
(x - tz)
dz.
1-
u(x, t) =
2f
-I- tz)
1 _ Izl2
dz.
Hence
1
u(x, t) = 2
(y - a;) + t2'/i(y)
t2 JB()
t2 - I
dy,
for any (x, t) E ][82 x (0, oo). We note that u(x, t) depends on the initial
values co and i in the solid disc Bt(x).
6. Wave Equations
220
We write un for dimension n. Then for any (x, t) E ][8x (0, oo),
x+t
u2(x't)
2 t2
u3(x't)
4urt2
+tVcp(y)
cy
(y)dy,
l2
x)
+t2b(y) dy
Bt(
(
Bt (x)
on the sphere 8Bt(x) of center x and radius tfor n = 3. These regions are
the domains of dependence of solutions at (x, t) on initial values. Conversely,
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 important consequence is that the process of solving initial-value problems for the
wave equation can be localized in space. Specifically, 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 difference between n = 2 and n = 3, we first note
that it takes time (literally) for initial values to make influences. Suppose
that the initial values cp, rb have their support contained in a ball B,.(xo).
221
Ix-xo-r<t< Ix-xoI+r.
Therefore, for a fixed x E ][83, u(x, t) = 0 for t < Ix - xol - r (corresponding
to finite-speed propagation) and fort > Ix - xol + r. So, the influence
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 nexcept n = 1 and does not hold in even space dimensions.
6. Wave Equations
222
1f
aBt(x)
b(y) dsy,
for any (x, t) E I[g3 x (0, oo). We assume that zb is radially symmetric, i.e.,
b(x) = h(IxI) for some function h defined in [0, oo). Then
u(0' t)
4irt
223
For some integer k > 3, if b(x) is not C' at lxi = 1, then h(t) is not
Ck at t = 1. Therefore, the solution u is not Cc at (x, t) _ (0, 1). The
physical interpretation is that the singularity of initial values at lxi = 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- ixi}
which intersects {t = 0} at lxi = 1.
:
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 explicit expressions for solutions of initial-value problems for the wave equation
in an arbitrary dimension. For odd dimensions, we seek an appropriate combination 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) defined by
(6.2.2) satisfies
(6.2.7)
r
r2
= r2 (r2Urrr+(n_1)rUrr_(n_1)Ur).
Since
6. Wave Equations
224
we obtain
(r2U,.)tt =
(r2Ur)rr T
(n - 5)rU,.,. - (n + 1)UT.
can continue this process to obtain appropriate combinations for all odd
dimensions. Next, we note that
r2UT + 3rU = r (r3U)T.
It turns out that the correct combination of U and its derivatives for arbitrary odd dimension n is given by
la
r ar
n-3
2
(r2U).
id
m-1
(r2v(r))
(m1-ire-1
(r21v(r)) _
"`-1
r dr)
id
"n'
- (-f)
dv
(r2(r));
z=o
, m - 1, and
Imo-1.g...(2m-1).
The proof is by induction and is omitted.
Cn+1, and
225
la
m-1
(r2mU(x; r, t)),
U(x;r,t) = (r ar-)
(6.2.10)
and
m-1
(x;r) = _1
p`1
(x;r) = _1
p`1
(r2m(x;r)),
r ar
m-1
r ar
(r2m-iW(x;r)).
U(x;r,O) =
for r > 0,
n=2m+1,wehaveforanyr>Oandt>0,
()
15
Urr =
()
1
m-1
r ar
5m-1
(r2U) =
(T2rn_iU) = Utt.
The initial condition easily follows from the definition of U, 1 and 'I'. The
boundary condition U(x; 0, t) = 0 follows from Lemma 6.2.4(2).
Jt-r
W(x; s) ds.
U(x;r,t) =
m-1
ai
cr,ir2+1
2=0
(r2mU(x; r,
ari U(x; r, t).
6. Wave Equations
226
Hence
lim
r t = 1U(x;
)
r-+0 Cm,Or
Therefore, we obtain
u(x, t) =
Cm,O
lim
Cm, p
2r
r+t
2r t_r
(x;s)ds)
t (x t) + (x;t)).
Using n =2m+ 1, the expression for Cm,p in Lemma 6.2.4 and the definitions
of and 'I', we can rewrite the last formula in terms of co and b. Thus, we
obtain for any x E Ian and t > 0,
1 1(a(1a
u(xt)=_L
Cn,
(6.2.11)
n-3
2
la
+
(fUBXds)
t
n-3
2
tat
wn t
UBt (x)
where n is an odd integer, wn is the surface area of the unit sphere in Ian
and
(6.2.12)
Suppose o E Cn21+1c(Il8n), b E
u(x, t) = cp(xo),
lim
ut(x, t) = vb(xo).
n-3
u`x'tJ
Cn,
(tn-2W(x,t)),
227
where
1
Wnt
5Bt (x)
u(x, t) =
cn-1,Zt Z+1
Cn
2=0
at2
(x, t).
(x+tw)dS.
Wn
IWI=1
Therefore,
a2
ati
W (x, t) =
a2
IwI=1 avi
Wn
Hence, u(x, t) is defined for any (x, t) E ItSn x [0, oo) and
0) = 0. Since
3+1(W
),
we conclude easily that Vu exists and is continuous in
E Cn2
, k. For t-derivatives, we conclude similarly that
W x [0, oo), for i = 0,1,
ut (x, t) is defined for any (x, t) E W x [0, oo) and ut 0) = b. Moreover,
a dS
Wntn-1
k - 1. In particular,
aL
5Bt (x)
Wntn-1
Bt (x)
bdy'
Next,
11
LW(x, t) =
Wn
L'b(x + tw) dS
IW1=1
O d
1
Wnt1
J5Bt(x)
Hence
Du(x, t) _
la
n-3
2
tat
0 dsy
Wnt
LB(X)
utt =
Cn
(j)
la
at
n-1
2
(t1w).
6. Wave Equations
228
Hence
utt =
Wncn
(13'\ 2n-1
tat
n-3
1a
0bds
tat
t aBt(x)
This implies that utt - Du = 0 at (x, t) E l[8n x (0, oo) and then u E
WnCn
x [O,oo)).
= 0 in a similar way.
t(,0)=
on W+1,
where
b() = b(x).
= o(x),
(aia(tat
n-2
2
cn+1
(13\(
+
(L1 t f8Bt()
) dSy
n-2
2
tat
Wn+1 t
x
J8Bt()
dsv/
where y =
yn, yn+1) = (y, yn+1) The integrals here are over the
surface aBt (x) in Rn+1. Now we evaluate them as integrals in Rn by eliminating
For xn+1 = 0, the sphere I y - x = t in Rn+1 has two pieces
given by
yn+1.
t
dy.
t2_ly_xl2
229
Hence
1
Wlt [
(,0(J)
P(y)
Wn+l
f t(x) /t2 - y nf
flWfl
Wfl JBt(x)
t2
dy.
- x12
A similar
expression holds for b. By a simple substitute, we now get an expression of
u in terms of cp and b. We need to calculate the constant in the formula.
Therefore, we obtain for any x E W and t > 0,
22
ux't- 1() t at
Wn JBt(x)
(6.2.13)
(P(y)
n-2
18
b(y)
(r:
+ tat
Wn
t2 _ Iq - xl2
Bt(x) /t2-Iy-xI2
where n is an even integer, wn/n is the volume of the unit ball in Ian and cn
is given by
Cn
In fact, we have
ncn+1wn+1
2wn
en=2 4 n.
u(x, t) = Sp(xo),
ut(x, t) = b(xo).
lim
(x,t)-*(xo,o)
(x,t)-*(xo,o)
This follows from Theorem 6.2.5. Again, u can be extended to a Cfunction in Ian x [0,oo).
utt - Du = 0 in
x (0, oo),
u(.,0)=0, u(,0)=b
onTRTh.
6. Wave Equations
230
I< C IIoZlbIILI(Rn),
i=0
Solutions not only are bounded globally but also decay as t -+ oo for
in n X (0, 00),
ut(',0)='i/.' onW.
Then for any t> 1,
[z]
lt)ILO(II2") <Cit
IIVIIL1(Rm),
X21
i=0
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 t-1 is the correct decay
rate for n = 3 by a simple geometric consideration. By (6.2.5), the solution
u is given by
1
(y) dsy,
u(x' t)
LB(X)
for any (x, t) E 1[83 x (0, oo). Suppose zb is of compact support and supp b C
1f
b(y) dSy.
A simple geometric argument shows that for any x E ][83 and any t> 0,
Iu(x,t)I
-sup kbl.
R3
231
This clearly shows that u(x, t) decays uniformly for x E ][83 at the rate of t-1
as t -3 00. 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 first consider n = 3. By assuming that /i is of compact support, we prove that for any t > 0,
Iu(x,t)I c
2'PIIr,1(R3)
u(xt)=_J
b(x+tw)dS,
for any (x, t) E ][83 x (0, oo). Since b has compact support, we have
1005
ib(x+tw)=-J
Then
u(x, t) _ -
-b(x+sw)ds.
LL1=i 8s
(x+sw)dSds.
lu(x, t) <
IlVIIL1(R3).
Iwl=1
2z/i(x+sw)ds.
Then
u(x, t) =
S as2
Hence
l
<
1J
47r
s2
J IWI=1
1
47r
We now discuss general '/i. For any (x, t) E ][83 x (0, oo), we note that u
depends on b only on 8Bt(x). We now take a cutoff function r,i E C( R3)
with r,, = 1 in Bt+l (x), r,, = 0 in Il83 \ Bt+2 (x) and a uniform bound on Vi.
We can obtain the
Then in the expression for u, we may replace /i by
6. Wave Equations
232
IC
u(x, t) =
1 Jft
(x + rw) dSdr.
t2 - r2 J11=1
(x+rw)dSw=_ff
JwI=1
wI=1
aS
and hence
(x+rw)dS
<
s fwI=1
1w1=1
sw)f dSWds
Therefore,
Pt
Iu(x,t)I
IIVIIL1(R2)
u(x,t) =
t21 r2 dr
JO
jte
4 IIVIIL1(R2).
ft)
-e
121J1
=
Jt-e
f
rJ
- r2
1
t2
(x+rw)dSdr
1=1
t-E
t2 - r
dr
( R2) .
I2
233
Jt
t-E
dr=Jt
[
dr
(t + r)(t-r)
t-E
It-E
UT=
2\/
Hence,
ForI,wehave
t-E
Ill! =
t2 r r2
<
-1
ft-E
t2 - (t -
fI
E)2
(x+rw)dSdr
f
r 'WI_1 IJ
zit1 - 2
Therefore, we obtain
Iu(x,t)I
V2Et_E21R2) +
For any t> 1, we take e = 1/2 and obtain the desired result.
We leave the proof for arbitrary n as an exercise.
Ut(',0)=/)
onlfttm.
(6.2.16)
u(.,0)=0,
on lltn,
6. Wave Equations
234
and
utt - Du = f
(621 7'
(-1
M. (x, t) -
(6.2.18)
1 81
'ra-3
2
t at)
cn
cent Jast(X)
dS
Mp(x, t =
1 fia
cn,
n-2
2
(j)at
In
'b(y)
d
y
t2 - l y - xl2
G`7flgt(x)
if n > 2 is even, where wn is the surface area of the unit sphere in Rn and
1 .3
Cn =
124n
u (x, t) = 0,
(x,t)-+(xo,o)
lim
ut (x, t) _ 'ib(xo).
(x,t)-+(xo,o)
Proof. This follows easily from Theorem 6.2.5 and Theorem 6.2.6 for p = 0.
As we have seen, u is in fact cm in Rn x [0, oo).
LI
atM
utt - Du = 0 in
Wn
x (0, oo) .
u (x, t) = cp (xo) ,
lim
(x,t)-+(xo,o)
ut (x, t) = 0.
235
M(.,0)=0,
on Ian.
0) = cP
Then
aM,(,t)(,0) = (
on RTh,
0) =
8tu(', U) =
0) = 0 on Rn.
u(x,t) _ f
Mf(x,t - T)CLT,
utt - Du = f
u(x, t) = 0,
(x,t)-+(xo,O)
lira
ut (x, t) = 0.
(x,t)-+(xo,O)
Proof. The regularity of u easily follows from Theorem 6.2.9. We will verify
that u satisfies utt - Du = f and the initial conditions. For each fixed T > 0,
w(x, t) = MfT (x, t - T) satisfies
Wtt - LtW = 0
in R x (r, oo),
f(.,r) on R.
= 0,
We note that the initial conditions here are prescribed on {t = T}. Then
ut = MfT (x, t - T) I T=t +
f
0
atMfT (x, t - T) dT
6. Wave Equations
236
and
aM fT (x, t - T) dT
M fT(x,t-T)dT
t
= f(xt)+fMfT(xt_r)dr
= f(x,t)+zu.
Hence utt - Du =fin I[8n x (0, oo) and
0) = 0,
0) = 0 in IlBn.
As an application of Theorem 6.2.11, we consider the initial-value problem (6.2.17) for n = 3. Let u be a CZ-solution of
utt - Du = f
M(x't)
(y)dS.
4Rt fBt(x)
u(x, t) =
J0
t MfT (x, t - T) dT =
14 oJ t - T JaBt-T
t
f(y,r)dSdr.
u(x t) _
f(y,t-s)
JO J8Bs(x)
dSvds.
Therefore,
(6.2.20)
u(x, t) =
Bt(x)
f(y, iy I
dy>
for any (x, t) E I183 x (0, oo). 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
0<s<t.
This is exactly the backward characteristic cone with vertex (x, t).
lim
(x,t)-+(xo,o)
u(x, t) = 0,
lim
(x,t)-+(xo,o)
ut (x, t) = 0.
237
Utt-LU=O inRmx(O,oo).
We assume that
0) have compact support. By finite-speed
0) and
propagation,
t) also has compact support for any t > 0. We multiply
the wave equation by Ut and integrate in BR x (0, t). Here we choose R
sufficiently large such that BR contains the support of
s), for any s e
(0, t). Note that
ututt - utoU =
Z (mot + I
2-1
is a constant independent of time. For general hyperbolic equations, conservation 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 II8" x [0, oo) and co and /i be
continuous functions in IIBn. We consider the initial-value problem
(6.3.1)
for some positive constants A and A. For the wave equation, we have a = 1
and c = 0 and hence we can choose A = A = 1 in (6.3.2).
In the following, we set
1
For any point P = (X, T) E ][8n x (0, oo), consider the cone Ck(P) (opening
downward) with vertex at P defined by
6. Wave Equations
238
As in Section 2.3, we denote by DSCk (P) and a_ Ck (P) the side and bottom
of the boundary, respectively, i.e.,
xI= T - t},
asck(P)_ { (x, t): o < t < T,
a_ck(P) _ {(x, o): ,'iIx - x < T}.
We note that a_Ck(P) is simply the closed ball in ][8x {0} centered at
(X, 0) with radius T/ic.
Theorem 6.3.1. Let a be Ci, c and f be continuous in I[8n x [0, oo), and
let cp be Ci and zb be continuous in I[8n. Suppose (6.3.2) holds and u E
C2(I[8x (0, oo)) fl Ci(I[8x [0, oo)) is a solution of (6.3.1). Then for any
point P = (X,T) E W` x (0,oo) and any 'i > io,
f
<f
(cp2 - 2 + alOcpI2) dx -
Jc,c(P)
f 2 dxdt,
239
and
n
-2e-taut0u = -2e-'taut
uxix2
i=1
n
(- 2(e
tautux2)xZ + 2e-tauxZutx2 -+ 2e
(- 2(e
tautux2)xZ
taxtiutuxti)
i=1
n
+ (e-tau)t +
i=1
+
where we used 2uut= (U.)t. Therefore, we obtain
r/e-tau- e '7tatu) ,
2e-tcuut
+
i=1
= 2e-'tut f.
We note that the first two terms in the left-hand side are derivatives of
quadratic expressions in Vu and ut and that the next three terms are
quadratic in Vxu and ut. In particular, the third term is a positive quadratic
form. The final term in the left-hand side involves u itself. To control this
term, we note that
(e-tu2)t + r/e-tu2
- 2e-tuut = 0.
- i=1
ut + aIVuI2) = RHS,
where
n
RHS =
a-'tat I VuI2
(c - 1)uut -+ 2e-'tut f.
i=1
The first three terms in RHS are quadratic in ut, ux2 and u. Now by (6.3.2)
and the Cauchy inequality, we have
2
<
2
I
2
I
ut2 + 1 au2z
6. Wave Equations
240
C(P)
ck(P)
(e_t(u2 + ut + aIDuI2))t - Z
(il
ut +
(-X10)
f 2.
we obtain
d5
(u2+u+alVul2)dx+fe-t f 2 dxdt,
w(P)
(x-X
kIx-XI'1
BI - (u+alVul2)vt - 2
autuv2 >0 on
n 7b
(2)
In 7L
=IVUI\/1
-vt.
i=1
i=1
BI >
1+2
BI >
1 - k2 (ut + alDul2
0.
241
We note that the cone Ck(P) in Theorem 6.3.1 plays the same role as
the cone in Theorem 2.3.4. The constant ic is chosen so that the boundary
integral over 83Ck(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 Cl, c and f be continuous in Il8n x [0, oo), and
let cp be Cl and b be continuous in ][8n. Suppose (6.3.2) holds and u E
C2(I[8n x (0, oo)) fl Cl(][8n x [0, oo)) is a solution of (6.3.1). For a firmed
b E L2(][8n), then for any rj > rjo,
T> 0, if f E L2(][8n x (0, T)) and cp,
e_t(u2 + u + aIVuI2) dx
+
m)
e-t f 2 dxdt,
x (O,T)
Usually, we call ut + aIDuI2 the energy density and its integral over
Il8n x {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 SZ
be a bounded domain in W and h_ and h+ be two piecewise C1-functions
in SZ with h_ <h + in SZ and h_ = h+ on 311. Set
A<a<A in D.
We now consider
(6.3.3)
utt - aDu + cu = f in D.
6. Wave Equations
242
]fin
<
Ja- n
autuxiv+2 dS
i=1
autuxiv_i dS
i=1
e t f2 dxdt,
-{-
Where v = (vi,
, vn, vt) are unit normal vectors pointing in the positive t-direction along 9D. We are interested in whether the integrand for
a+D is nonnegative. As in the proof of Theorem 6.3.1, we have, by the
Cauchy inequality,
1
_ IVuI\/1 - v+t.
?Lx v+i
i=1
(mot + aIVuI2)v+t
-2
z=1
>_ (ut -I- aI0uI2)v+t - 2/a(1 - v+t) ' \1IutI ' IVuI > 0 on a+D
if
v+t >
all - v+t)
v+t >
1+a
on B+D.
243
Jr D
e-t
+
((U2 + ut + aIDuI2)v_t - 2
autuxiv_i
dS
f2 dxdt.
JD
a(x, t)
1 +a(x,t)'
E is time-like at (x, t) if
vt (x, t) <
a(x, t)
l+a(x,t)
utt - Du = f.
With a = 1, the hypersurface E is space-like at (x, t) if vt(x, t) > 1//. If
(6.3.5)
I< 1.
6. Wave Equations
244
Next, we discuss briefly initial-value problems with initial values prescribed on a time-like hypersurface. Consider
1.sin my,
au
1.sin my on {x = 0}.
m2
ax m
Here we treat {x = 0} as the initial hypersurface, which is time-like for the
wave equation. A solution is given by
1
Note that
um40, DUm
ax
on {x=0} asm-+oo.
)I=
ems
oo
as m - 00.
m2
Ilga
Therefore, there is no continuous dependence on the initial values.
6.4. Exercises
245
of the initial-value problem for the first-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 significant difference. The weak solutions in Definition 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 defined by
1
(u2 + ut + Idxdt
IIUIIH1(Rnx(OT)) = (LX
Then we can prove that (6.3.1) admits a weak Hl-solution in ][8n x (O, T) if
cp =
= 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, cp e C2([0, l]) and b e C'([0, l]).
Consider
Utt - Usa; = 0
6. Wave Equations
246
Exercise 6.2. Let cpl and cp2 be C2-functions in {x < 0} and {x > 0},
respectively. Consider the characteristic initial-value problem
utt -
Solve this problem and find the domain of dependence for any point (x, t)
Solve this problem and find the domain of dependence for any point (x, t)
for x> 0,
u(0, t) =
a
for a = -1.
Exercise 6.5. Let a be a constant with al < 1. Prove that the wave
equation
utt - L u= 0 in R3 x R
is preserved by a Lore ntz transformation, i.e., a change of variables given by
s=
t - axl
1-a2'
xl - at
Y1=
1- a2'
yz=xi fori=2, 3.
Exercise 6.6. Let A be a positive constant and i4' E C2(]I82). Solve the
following initial-value problems by the method of descent:
utt = Du + AZU in 1[82 x (0, oo),
u(.,0)=0,
0) _ i/i
on ]182,
6.4. Exercises
247
and
u(.,0)=0,
onR2.
utx, t =
)
27r
Bt(u)
y)
t2 - IY - xI
dy.
supIu(',t)I
Bat
CSUPII foranyt> 1,
t Il2z
Suppose that u E
v(x)
(Vxu(xo
IJ
IIx-xpl2ut`x'tJ)
t=to-Ix-xol
Remark: This exercise gives an alternative approach to solving the initialvalue problem for the three-dimensional wave equation.
6. Wave Equations
248
lim (ru(rw,r+'r))
T-+00
exists and the convergence is uniform for w e 2 and r E (0, To], for
any fixed TO>0.
Exercise 6.10. Prove Theorem 6.2.7 and Theorem 6.2.8 for n > 2.
equation
Lu - 2utt + 3ut +
= 0.
on lxl> a, t =
ixi- a,
1F. G. Friedlander, On the radiation field of pulse solutions of the wave equation, Proc. Roy.
Soc. A, 269 (1962), 53-65.
Chapter 7
First-Order Differential
Systems
In this chapter, we discuss partial differential systems of the first order and
focus on local existence of solutions.
In Section 7.1, we introduce the notion of noncharacteristic hypersurfaces for initial-value problems. We proceed here for linear partial differential equations and partial differential systems of arbitrary order similarly
to how we did for first-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 noncharacteristic initial hypersurfaces. We also demonstrate that partial differential
systems of arbitrary order can always be transformed to those of the first
order.
In Section 7.2, we discuss analytic solutions of the initial-value problem
for first-order linear differential systems. The main result is the CauchyKovalevskaya theorem, which asserts the local existence of analytic solutions
if the coefficient matrices and the nonhomogeneous terms are analytic and
249
250
7.1.1. Linear Partial Differential Equations. We start with linear partial differential equations of arbitrary order and proceed here as in Sections
2.1 and 3.1.
Let 1 be a domain in 1Rn containing the origin, m be a positive integer
aa(x)aau in 1.
Lu =
IaI<m
aa(x)aau
Lou =
in SZ,
IaI=m
and
aa(x)r,
IaI=m
Lu =
a2
i=1
b(x)u
in 1,
and
251
Lu =
bi (x)u+ c(x)u in ft
aij
i,j=1
i=1
ai(x)i
p(xi) =
i=1
and
n
p(xiS)
i,j=1
(7.1.2)
in S2.
v = (vi, , vn). For any integer j > 1, any point xp E E and any C-function u defined in a neighborhood of xo, the jth normal derivative of u
at xo is defined by
au
avj
U a u=
0) = uo, U
urn_1
on Rn-1.
252
a(0,... ,0,m)(0) 4 0,
then by (7.1.2),
DU(0) = -
0 m)
CL(0,... ,0,m)
a a u - f
\a(0,... ,0,m)
n u + ... = f,
a1
253
y,X2 uy _
=1
and in general, for any a e Z+ with al = m,
au = ai
9u = (Pi
Lu =
a (x(y)) coi
Jai=rn
IaI=rn
_ V(p(x).
p(xo; v) =
as (xo) v
0,
I_m
at xo. Otherwise,
is characteristic
at x0.
254
for any
E I[8n \ {0}.
a2 (x)u+ b(x)u in SZ C R.
Lu =
2=1
2=1
for any x e SZ and any e W1. Hence first-order linear differential equations with real coefficients are never elliptic. Complex coefficients may yield
elliptic equations. For example, take a1 = 1/2 and a2 = i/2 in IR2. Then
az = (awl + i32)/2 is elliptic.
The notion of ellipticity was introduced in Definition 3.1.2 for secondorder linear differential operators of the form
n
Lu =
b2(x)u+ c(x)u in SZ C R.
+
2,j=1
2=1
255
p(x; ) _
z,j=1
for any
E Il8" \ {0}.
z,j=1
7.1.2. Linear Partial Differential Systems. The concept of noncharacteristics can be generalized to linear partial differential equations for vector-
Lu =
Aa (x)aau
in SZ,
I aI<m
Aa(x)aau inn,
Lpu =
IaI=m
and
det
E IlBn.
A(x)r
\IaI=m
256
Definition 7.1.5. Let L be a linear differential operator defined in a neighborhood of xo E ][8n as in (7.1.7) and E be a smooth hypersurface containing
xo. Then E is noncharacteristic at xo if
A(xo)va
p(xo; v) = det
0,
\IaI=rn
Lu = f (x) in 11.
(7.1.8)
Lu =
AiUX2 + Bu.
i=1
det
(vA(xo))
; 0,
i=1
257
be smooth on E. Then the initial-value problem (7.1.8)-(7.1.9) in a neighborhood of xo is equivalent to an initial-value problem for afirst-order differential system with appropriate initial values prescribed on E, and E is
a noncharacteristic hypersurface at xo for the new first-order differential
system.
,m-
1.
Step 2. Reductions to canonical forms and zero initial values. In the new
coordinates, {xn = 0} is noncharacteristic at 0. Then, the coefficient matrix
,o,rn) is nonsingular at the origin and hence also in a neighborhood of
the origin. Multiplying the partial differential system (7.1.8) by the inverse
of this matrix, we may assume that A(o, ,o,rn) is the identity matrix in a
neighborhood of the origin. Next, we may assume
-u( x
u(x) = v(x) +
j=0
Then the differential system for v is the same as that for u with f replaced
by
m-1
A(x)Da (-ui(x')x).
1(x) j=0
Moreover,
,m- 1.
With Step 1 and Step 2 done, we assume that (7.1.8) and (7.1.9) have
the form
au-
m-1
Aau = f,
n=o ' I <rn-n
With
D u(x', 0) = 0
for j = 0,1,
,m-1.
258
U0=u,
UZ = uxi for i = 1,
, n,
and
T
TT
U=(U" ,U1,...,Un)
(7.1.10)
U,xn = Un,
Hence
(7.1.11)
am-1 U0
(7.1.12)
am
U =0,
n
- am-2
xn
lug-axiam 2Un=0
u = f.
Aaaau +
an=1 Ia' I <m-an
I a'I <m
We substitute Un = uxn in the first two terms in the left-hand side to get
m-2
(7.1.13)
ax 1 Un +
A(a
Aa as Un +
an=0 a' I <m-an -1
u = f.
a' I <m
In the last summation in the left-hand side, any mth-order derivative of u can
, n -1,
be changed to an (m -1)th-order derivative of UZ for some i = 1,
ax -1U +
Aa1)5aU
an=0 a'l<m-an-1
axnu(x',0) =0
that Ui =
259
for i = 1,
U, we have
m-2
aXn
(Un - Up,) - 0,
and on { xn = 0},
Next, for i = 1,
am 1UZ-%, am 2Un=t9
1U2
,n-
1,
and on {xn = 0}
260
uy + v = 0,
(7.2.1)
u=g(x),v=0 on{y=0}.
We point out that (7.2.1) is simply the Cauchy-Riemann equation in C _ 1182.
(0
i(u
-(0
\-i 0) \v) -
(O
-f- bl
(0i
O)
- (S2
S11
O)
w(.,0)=g
onIR.
261
where the ca are real numbers defined for all multi-indices a e Z. Throughout this section, the term convergence always refers to absolute convergence.
Hence, a series a ca is convergent if and only if > ca < oo. Here, the
summation is over all multi-indices a e Z.
u(x) =
a
If u is analytic near xo, then u is smooth near xo. Moreover, the constants ua are given by
1
ua = -8au(xa) for a E Z.
a!
Thus u is equal to its Taylor series about xo, i.e.,
x E B,.(xo).
u(x) _
a
Then
00
k=O
(k)a
lal_k
This power series is absolutely convergent for lxi < r// since
00
Ixil + ... + Ixn
lal! a _
a rllal
k-0
for lxii +... + IxI < Ixlv4 < r. We also note that
8u(0) = ra
for a E 7G+.
262
for any a E Z.
av (0) > I
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 B,., then the Taylor series
of u about the origin converges in Br.
for x E B,..
1
a
a.
u(o)y
< C,
and in particular,
a a"u(0)
C
y1
yn
C laP
s
a.
Now set
v (x) - s _ (x1 Cs
+ ... + xn)
=C
lal! x .
sk I a.
So far, our discussions are limited to scalar-valued functions. All definitions and results can be generalized to vector-valued functions easily. For
example, a vector-valued function u =
UN) is analytic if each of its
components is analytic. For vector-valued functions u = (Ui,... , UN) and
We have the following results for compositions of functions.
263
fo'u<Ggov.
Lemma 7.2.8. Let u be an analytic function near 0 E I[8n with range in ][8""'
and f be an analytic function near u(0) E Il8"` with range in RN. Then f on
is analytic near 0 E IIBn.
We leave the proofs as exercises.
{t=0}.
Let A1,
, An be smooth N x N matrices in Rn+1+N, F be a smooth
N-vector in Rn+1+N and uo be a smooth N-vector in R. Consider
n
(7.2.2)
ut =
with
(7.2.3)
0) = uo.
Theorem 7.2.9. Let uo be an analytic N-vector near 0 E Il8", and let Al,
264
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
F(x,u),
ut =
(7.2.4)
j=1
with
0) = 0,
(7.2.5)
where A1,
au(0)=0
For any i = 1,
for any a E Z+ .
uxZt
j=1
u(O) = F(0,0).
More generally, we obtain by induction
+F
= as
(Aj
j=1
t + Aj,uutu j) + Fuut
aaa2u(o) = as
Fuut
j=1
(x,t,u)=O
265
aaatu(0) =
(7.2.6)
Btu -
pa,k(axauA1,...
IWe
I a I+ k -1,
axaryAn, axaryF,
(7.2.7)
t u -o
vt =
(7.2.s)
G(x, v),
Bj (x,
j=1
0,
(7.2.9)
, rz
and C >>F.
> Ifor
j = 1,... , n,
and
I
266
<pa,k(laaAl1,...
pa,k(axau-'-'l,' .. , axauBn,
aaaly) x t u =o
= axat v(0),
where we used (7.2.6), (7.2.7) and the fact that pa,k has nonnegative coefficients. Thus
(7.2.10)
v >> u.
Bn
Cr
T - (xl + ... +
xnvl 1
xn+vl+...+vNv
1
1
and
G
Cr
r-(x1+...+xfl+v1+...+vN))
1
for positive constants C and r, with lxi + lvi < r/ n -F- N. As demonstrated
in the proof of Lemma 7.2.6, we may choose C sufficiently large and r
sufficiently small such that (7.2.9) holds.
Set
v=w
for some scalar-valued function w in a neighborhood of 0 E I[Sn+l
(7.2.8) is reduced to
wt =
Cr
r-(x1+...+xn+Nw)
Then
wxi
i=1
w(.,0) = 0.
This is a (single) first-order quasilinear partial differential equation. We now
seek a solution w of the form
w(x1 i ... , xn, t) = ti (x1 + ... + xn, t).
267
Cr
r - z - Nw (nNwz + 1),
w(, o) = o.
By using the method of characteristics as in Section 2.2, we have an explicit
solution
w(z, t) _
(n + 1)N
{r -
and hence
1
w(x, t) _
(n+ 1)N
{r_xi_ [(r_xi)2_2Cr(n+l)Nt]
This function is analytic near the origin and its Taylor series about the
origin is convergent for I (x, t) I < s, for sufficiently smalls > 0. Hence, the
Wt=
Cr
r - z - t - Nw (nNt+1),
w(, o) = o.
It is difficult, if not impossible, to find an explicit expression of the solution
w.
268
Ao(x, t)ut +
(7.2.11)
j=1
u(x, 0) = uO(x).
For the proof, we need to introduce adjoint operators. Let L be a differential operator defined by
n
Lu = Ao(x, t)ut +
vTLu
(Av)xi i=1
i=1
L*v = - (4v)t -
(ATv)x2 + BTv
i=1
n
= - 4vt -
AT vxi +
i=1
BT - Ao t
i=1
Then
vT Lu
(vTAiu)xi + (L*v)Tu.
= (vTAou)t +
z=1
For any
> 0, we set
269
ac = {(x,t)
x2 < t =
a_cz6={(x,t): 1x12=t<e}.
We note that det(Ao(0)) L 0 since E is noncharacteristic at the origin.
Hence Ao is nonsingular in a neighborhood of the origin. By multiplying
the equation in (7.2.11) by A1, we assume Ao = I.
v=P onBrf1{t=e},
where BT is the ball in
with center at the origin and radius r. The
principal part of L* is the same as that of L, except a different sign and a
transpose. We fix r so that {t = e} f1 BT is noncharacteristic for L*, for each
small e. By Theorem 7.2.9, an analytic solution v exists in BT for a small.
We need to point out that the domain of convergence of v is independent
of P, whose components are polynomials. We choose a small such that
SZe C BT . Then we have
J916
270
10
Theorem 7.2.9 guarantees the existence of solutions of initial-value problems 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 first-order homogeneous linear differential system (7.2.1),
uy + v = 0,
Note that all coefficients 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
Obviously,
uk (x, 0) -I- vk (x, 0) = 1
271
L w
L(=(
+(
-1(v
(2y 2x(v
Lu = h'(z)
272
SZ =
8,
V(r, z) =
J - R, zo + R), is continuous up to r = 0
Then V is Cl in (r, z) E (0, R) x (zo
with V(0, z) = 0, and satisfies
r
Vz - iV,. =
rve - 2ivz)
f2 (v,. +
d8 =
Define
W= V(r,z) -iirh(z).
Then W is Cl in (0, R) x (zo - R, zo + R), is continuous up to r = 0,
and satisfies Wz -I- iW,. = 0. Thus W is an analytic function of z + it for
(r, z) E (0, R) x (zo - R, zo + R), continuous at r = 0, and has a vanishing
real part there. Hence we can extend W as an analytic function of z + it to
(r, z) E (-R, R) x (zO - R, zo + R). Hence -irh(z), the imaginary part of
W(0, z), is real analytic for z E (zO - R, zo + R).
Now we consider the general case. Set
273
and
2-ke_.
ck =
)E
fT(x,y,z) _
(7.3.2)
we set
in R.
We note that fT depends on r linearly. This fact will be needed later on.
IITIICkMm+1Pk
(rn)rn
e
In the last inequality, we used the fact that the function f(r) = r"te-"'
in [0, oo) has a maximum mT"e-T" at r = m. This implies the uniform
convergence of the series for 8"fT
yV`
SZ
sup
x,yESt,x#y
Vu(x) - Vu(y)
I x - yl
274
be constants.
a subsequence {uk' } such that uk' -+ u in C' (1) for any bounded subset 1
in Bk,m,
with
Lemma 7.3.6. For any positive integers k and m, ek,m is closed and
nowhere dense in .
Proof. We first prove that Ek,m is closed. Take any Ti, T2,
E ek,m and
T E O such that
-TIIPO = 0.
275
that u3' converges uniformly to u together with its first derivatives in any
compact subset of Bk,m. Then, Lu = fT in Bk,m, u(Pk) = 0 and
k1IC1tL(Bk,m) < m.
Proof of Theorem 7.3.1. Let p E (0, 1) be the constant as in the definition of ek,m. We will prove that for some r E O, the equation Lu = fT
admits no C1"-solutions in any domain S1 C R3. If not, then for every
T E O there exist an open set SZT C R3 and a u E C1 " (11T) such that
Lu = fT
in
SZT .
By the density of {Pk} in R3, there exists a Pk E SZT for some k > 1. Then
IUIcl,IL(Bk) < m.
_
k,m=1
276
7.4. Exercises
Exercise 7.1. Classify the following 4th-order equation in R3:
uu-u-u=0 inRx(0,oo),
u(x, 0) = x, ut(x, 0) _ -x.
Find a solution as a power series expansion about the origin and identify
this solution.
U(,0)0 onR.
Under appropriate conditions on f, prove that the above initial-value problem admits a Cl-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.
ut +au +b11u+b,2v = 1
v + b12u + b22v = g,
with
fort >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[o,T] I')I in terms of sup-norms of f, g, cp and
'r/'.
7.4. Exercises
277
f,
v + b12u + b22v = g,
Chapter 8
Epilogue
(8.1.1)
i,j=1
aijux+
biu-I- cu = f
in SZ,
i=1
279
8. Epilogue
280
E ][8n,
z,j=1
for some positive constant A. The equation (8.1.1) reduces to the Poisson
equation if aZ3 = SZ and bi = c = 0. In many cases, it is advantageous to
write (8.1.1) in the form
n
(au) x +
(8.1.2)
i,j=1
biux2 + cu = f
in SZ,
i=1
lems. In the Dirichlet problem, the values of solutions are prescribed on the
boundary, while in the Neumann problem, the normal derivatives of solutions are prescribed.
In solving boundary-value problems for elliptic differential equations, we
work in Holder spaces C' and Sobolev spaces Wk,P. Here, k is a nonnegative integer, p> 1 and a e (0, 1) are constants. For elliptic equations in the
divergence form, it is advantageous to work in Sobolev spaces H' =
due to their Hilbert space structure.
W',2
ut -
(8.1.3)
biux2 + cu = f
aij uxzx +
i,j=1
in D,
i=1
i,j=1
for some positive constant A. The equation (8.1.3) reduces to the heat
equation if az3 = b2j and bi = c = 0.
Naturally associated with the parabolic differential equations are initialvalue problems and initial/boundary-value problems. In initial-value problems, D = ][8n x (0, oo) and the values of solutions are prescribed on IEBn x {0}.
281
a bounded domain in
appropriate boundary values are prescribed on
aSZ x (0, oo) and the values of solutions are prescribed on SZ x {0}. Many
results for elliptic equations have their counterparts for parabolic equations.
utt -
(8.1.4)
i,j=1
where the
biuxi + cu = f in D,
aijuxZxj +
i=1
satisfy
n
aij (x, t) i j
I I2
i,j=1
for some positive constant A. The equation (8.1.4) reduces to the wave
equation if
= Si j and bi = c = 0.
Naturally associated with the hyperbolic differential equations are initialvalue problems. We note that {t = 0} is a noncharacteristic hypersurface
for (8.1.4). In initial-value problems, D = W x (0, oo) and the values of
solutions together with their first t-derivatives are prescribed on IIST x {0}.
Solutions can be proved to exist in Sobolev spaces under appropriate assumptions. Energy estimates play fundamental roles in hyperbolic differential equations.
8.1.4. Linear Symmetric Hyperbolic Differential Systems. We denote by (x, t) points in W x TI. Let N be a positive integer, A0, A1,
An
(8.1.5)
Aout +
Akuxk + Bu = f.
We always assume that Ao(x, t) is nonsigular for any (x, t), i.e.,
det(Ao(x, t))
0.
8. Epilogue
282
tems and discuss briefly their background. The aim of this section is to
illustrate the diversity of nonlinear partial differential equations. We have
no intention of including here all important nonlinear PDEs of mathematics
and physics.
8.2.1. Nonlinear Differential Equations. We first introduce some important nonlinear differential equations.
The Hamilton-.Iacobi equation is a first-order nonlinear PDE for a function u = u(x, t),
u+H(Du,x) =0.
This equation is derived from Hamiltonian mechanics by treating u as the
generating function for a canonical transformation of the classical Hamiltonian H = H(p, x). The Hamilton-Jacobi equation is important in identifying
conserved quantities for mechanical systems. Apart of its characteristic
ODE is given by
x2 = Hpi(p x)
pi = -Hxi (p, x)
This is referred to as Hamilton's ODE, which arises in the classical calculus
of variations and in mechanics.
In continuum physics, a conservation law states that a particular measurable property of an isolated physical system does not change as the system
evolves. In mathematics, a scalar conservation law is a first-order nonlinear
PDE
Ut + (F(u)) =0.
283
Minimal surfaces are defined as surfaces with zero mean curvature. The
minimal surface equation is a second-order PDE for u = u(x) of the form
div
1=0.
1+ IThis
For any function u defined in S2, the area of the graph of u is given by
A(u)
1 +IVuI2dx.
A Monge-Ampere equation is a nonlinear second-order PDE for a function u = u(x) of the form
det(VZU)
= f(x),
the surface, the Gauss curvature is given by K(x). The resulting partial
differential equation is
ut - aLu = f(u),
where u = u(x, t) represents the concentration of a substance, a is the diffusion coefficient and f accounts for all local reactions. They model changes of
the concentration of substances under the influence of two processes: local
chemical reactions, in which the substances are transformed into each other,
and diffusion, which causes the substances to spread out in space. They
8. Epilogue
284
The Korteweg-de cries equation (KdV equation for short) is a mathematical model of waves on shallow water surfaces. The KdV equation is a
nonlinear, dispersive PDE for a function u = u(x, t) of two real variables,
space x and time t, in the form
It admits solutions of the form v(x - ct), which represent waves traveling to
the right at speed c. These are called soliton solutions.
8.2.2. Nonlinear Differential Systems. Next, we introduce some nonlinear differential systems.
In fluid dynamics, the Euler equations govern inviscid flow. They are
usually written in the conservation form to emphasize the conservation of
mass, momentum and energy. The Euler equations are a system of firstorder PDEs given by
pt + V. (pu) = 0,
(pu)t + V (u (pu)) + Op = 0,
E=e+
Z IuI2,
285
where e is the internal energy per unit mass and the second term corresponds
to the kinetic energy per unit mass. When the flow is incompressible,
If the flow is further assumed to be homogeneous, the density p is constant and does not change with respect to space. The Euler equations for
incompressible flow have the form
ut + u Vu = - Vp,
In forming these equations, we take the density p to be 1 and neglect the
equation for E.
The Navier-Stokes equations describe the motion of incompressible and
homogeneous fluid substances when viscosity is present. These equations
arise from applying Newton's second law to fluid motion under appropri-
ate assumptions on the fluid stress. With the same notation for the Euler
equations, the Navier-Stokes equations have the form
Ft = O9(t)F
8. Epilogue
286
where g (t) is the induced metric of the evolving hypersurface F(t). When
expressed in an appropriate coordinate system, the mean curvature flow
forms a second-order nonlinear parabolic system of PDEs for the components
of F.
In general relativity, the Einstein field equations describe how the curvature of spacetime is related to the matter/energy content of the universe.
They are given by
G = T,
G= Ric - - Sg,
2
where Ric is the Ricci curvature tensor and S is the scalar curvature of
(M, g). While the Einstein tensor is a type of curvature, and as such relates
to gravity, the stress-energy tensor contains all the information concerning
the matter fields. Thus, the Einstein field equations exhibit how matter acts
as a source for gravity. When expressed in an appropriate gauge (coordinate system), the Einstein field equations form a second-order quasilinear
hyperbolic system of PDEs for components g2j of the metric tensor g. In
general, the stress-energy tensor T depends on the metric g and its first
derivatives. If T is zero, then the Einstein field equations are referred to as
the Einstein vacuum field equations, and are equivalent to the vanishing of
the Ricci curvature.
Yang-Mills theory, also known as non-Abelian gauge theory, was formulated by Yang and Mills in 1954 in an effort to extend the original concept
of gauge theory for an Abelian group to the case of a non-Abelian group and
has great impact on physics. It explains the electromagnetic and the strong
287
The Yang-Mills equations are the Euler-Lagrange equations for this functional and can be written as
dAF = 0,
where dA is the adjoint of dA, the gauge-covariant extension of the exterior
derivative. We point out that F also satisfies
dAF=O.
This is the Bianchi identity, which follows from the exterior differentiation
of F. In general, Yang-Mills equations are nonlinear. It is a Millennium
Prize Problem to prove that a nontrivial Yang-Mills theory exists on R and
has a positive mass gap for any compact simple gauge group G.
8.2.3. Variational Problems. Last, we introduce some variational problems with elliptic characters. As we know, harmonic functions in an arbitrary domain 1 C IiSn can be regarded as minimizers or critical points of the
Dirichlet energy
IVuI2 dx.
F(Vu) dx.
Vi +
8. Epilogue
288
minimizing harmonic maps are not smooth. They are smooth away from
a subset E, referred to as a singular set. The study of singular sets and
behavior of minimizing harmonic maps near singular sets constitutes an
important subject.
One more way to generalize is to consider the Dirichlet energy,
IVuI2 dx,
Bibliography
[1]
[2]
Carlson, J., Jaffe, A., Wiles, A. (Editors), The Millennium Prize Problems, Clay
Math. Institute, 2006.
[3] Chen, Y.-Z., Wu, L.-C., Second Order Elliptic Equations and Elliptic Systems, Amer.
Math. Soc., 1998.
[4] Courant, R., Hilbert, D., Methods of Mathematical Physics, Vol. II, Interscience Publishers, 1962.
[5] DiBenedetto, E., Partial Differential Equations, Birkhauser, 1995.
[6] Evans, L., Partial Differential Equations, Amer. Math. Soc., 1998.
[14] Hormander, L., The Analysis of Linear Partial Differential Operators, Vols. 1-4,
Springer, 1983-85.
[15] John, F., Partial Differential Equations (4th ed.), Springer, 1991.
[16] Lax, P., Hyperbolic Partial Differential Equations, Amer. Math. Soc., 2006.
[17] Lieberman, G. M., Second Order Parabolic Partial Differential Equations, World Scientific, 1996.
[18] MacRobert, T. M., Spherical Harmonics, An Elementary Treatise on Harmonic Functions with Applications, Pergamon Press, 1967.
[19] Protter, M., Weinberger, H., Maximum Principles in Differential Equations, PrenticeHall, 1967.
289
290
Bibliography
Index
a priori estimates, 4
adjoint differential operators, 39, 268
analytic functions, 105, 261
auxiliary functions, 121
Bernstein method, 121
Burgers' equation, 22
Cauchy problems, 11, 48, 251, 256
Cauchy values, 11, 48, 251, 256
Cauchy-Kovalevskaya theorem, 263
characteristic cones, 57
characteristic curves, 14, 50, 253
characteristic hypersurfaces, 13, 14, 16,
50, 253, 256
noncharacteristic hypersurfaces, 13,
14, 16, 50, 253, 256
characteristic ODEs, 19, 21, 26
characteristic triangle, 202
compact supports, 41
comparison principles, 114, 119, 177
compatibility conditions, 25, 79, 83,
207, 210
conservation laws, 24, 282
conservation of energies, 64, 237
convergence of series, 105, 260
absolute convergence, 260
convolutions, 150
energy estimates
first-order PDEs, 37
heat equations, 62
wave equations, 63, 238, 241
Euclidean norms, 1
Euler equations, 284
Euler-Poisson-Darboux equation, 214
exterior sphere condition, 132
finite-speed propagation, 35, 221
first-order linear differential systems,
281
Index
292
fundamental solutions
heat equations, 157, 159
Laplace equations, 91
hypersurfaces, 2
281
gradients, 2
Green's formula, 92
Green's function, 81, 94
Green's function in balls, 96
Green's identity, 92
half-space problems, 207
Hamilton-Jacobi equation, 282
harmonic functions, 52, 90
conjugate harmonic functions, 52
converegence of Taylor series, 105
differential Harnack inequalities, 109,
122
majorants, 262
maximum principles, 111
strong maximum principles, 111, 117,
181
Index
quasilinear PDEs, 3
first-order, 14
radiation field, 248
range of influence, 19, 35, 204, 220
reaction-diffusion equations, 283
removable singularity, 125
Ricci flows, 286
Schrodinger equations, 284
Schwartz class, 148
second-order linear PDEs, 48
in the plane, 51
elliptic PDEs, 51, 279
hyperbolic PDEs, 58, 281
parabolic PDEs, 58, 280
293
separation of variables, 67
shocks, 24
Sobolev spaces, 139, 140, 142
space variables, 1
space-like surfaces, 243
subharmonic functions, 113, 126
subsolutions, 113
heat equation, 176
subharmonic functions, 113
superharmonic functions, 126
supersolutions, 113
heat equation, 176
superharmonic functions, 113
symmetric hyperbolic differential
systems, 282
Taylor series, 105, 261
terminal-value problems, 165
test functions, 24
time variables, 1
time-like surfaces, 243
Tricomi equation, 54
ISBN 978-0-8218-5255-2
www.ams.org/bookpages/gsm-120
9
780821 1852552
GSM/ 120
www.ams.org