You are on page 1of 396

# Partial Dierential Equations: Graduate Level Problems

and Solutions
Igor Yanovsky
1
Partial Dierential Equations Igor Yanovsky, 2005 2
Disclaimer: This handbook is intended to assist graduate students with qualifying
examination preparation. Please be aware, however, that the handbook might contain,
and almost certainly contains, typos as well as incorrect or inaccurate solutions. I can
not be made responsible for any inaccuracies contained in this handbook.
Partial Dierential Equations Igor Yanovsky, 2005 3
Contents
1 Trigonometric Identities 6
2 Simple Eigenvalue Problem 8
3 Separation of Variables:
Quick Guide 9
4 Eigenvalues of the Laplacian: Quick Guide 9
5 First-Order Equations 10
5.1 Quasilinear Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.2 Weak Solutions for Quasilinear Equations . . . . . . . . . . . . . . . . . 12
5.2.1 Conservation Laws and Jump Conditions . . . . . . . . . . . . . 12
5.2.2 Fans and Rarefaction Waves . . . . . . . . . . . . . . . . . . . . . 12
5.3 General Nonlinear Equations . . . . . . . . . . . . . . . . . . . . . . . . 13
5.3.1 Two Spatial Dimensions . . . . . . . . . . . . . . . . . . . . . . . 13
5.3.2 Three Spatial Dimensions . . . . . . . . . . . . . . . . . . . . . . 13
6 Second-Order Equations 14
6.1 Classication by Characteristics . . . . . . . . . . . . . . . . . . . . . . . 14
6.2 Canonical Forms and General Solutions . . . . . . . . . . . . . . . . . . 14
6.3 Well-Posedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
7 Wave Equation 23
7.1 The Initial Value Problem . . . . . . . . . . . . . . . . . . . . . . . . . . 23
7.2 Weak Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
7.3 Initial/Boundary Value Problem . . . . . . . . . . . . . . . . . . . . . . 24
7.4 Duhamels Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
7.5 The Nonhomogeneous Equation . . . . . . . . . . . . . . . . . . . . . . . 24
7.6 Higher Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
7.6.1 Spherical Means . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
7.6.2 Application to the Cauchy Problem . . . . . . . . . . . . . . . . 26
7.6.3 Three-Dimensional Wave Equation . . . . . . . . . . . . . . . . . 27
7.6.4 Two-Dimensional Wave Equation . . . . . . . . . . . . . . . . . . 28
7.6.5 Huygens Principle . . . . . . . . . . . . . . . . . . . . . . . . . . 28
7.7 Energy Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
7.8 Contraction Mapping Principle . . . . . . . . . . . . . . . . . . . . . . . 30
8 Laplace Equation 31
8.1 Greens Formulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
8.2 Polar Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
8.3 Polar Laplacian in R
2
for Radial Functions . . . . . . . . . . . . . . . . 32
8.4 Spherical Laplacian in R
3
and R
n
for Radial Functions . . . . . . . . . . 32
8.5 Cylindrical Laplacian in R
3
for Radial Functions . . . . . . . . . . . . . 33
8.6 Mean Value Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
8.7 Maximum Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
8.8 The Fundamental Solution . . . . . . . . . . . . . . . . . . . . . . . . . . 34
8.9 Representation Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
8.10 Greens Function and the Poisson Kernel . . . . . . . . . . . . . . . . . . 42
Partial Dierential Equations Igor Yanovsky, 2005 4
8.11 Properties of Harmonic Functions . . . . . . . . . . . . . . . . . . . . . . 44
8.12 Eigenvalues of the Laplacian . . . . . . . . . . . . . . . . . . . . . . . . . 44
9 Heat Equation 45
9.1 The Pure Initial Value Problem . . . . . . . . . . . . . . . . . . . . . . . 45
9.1.1 Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . 45
9.1.2 Multi-Index Notation . . . . . . . . . . . . . . . . . . . . . . . . 45
9.1.3 Solution of the Pure Initial Value Problem . . . . . . . . . . . . . 49
9.1.4 Nonhomogeneous Equation . . . . . . . . . . . . . . . . . . . . . 50
9.1.5 Nonhomogeneous Equation with Nonhomogeneous Initial Condi-
tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
9.1.6 The Fundamental Solution . . . . . . . . . . . . . . . . . . . . . 50
10 Schr odinger Equation 52
11 Problems: Quasilinear Equations 54
12 Problems: Shocks 75
13 Problems: General Nonlinear Equations 86
13.1 Two Spatial Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
13.2 Three Spatial Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . 93
14 Problems: First-Order Systems 102
15 Problems: Gas Dynamics Systems 127
15.1 Perturbation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
15.2 Stationary Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
15.3 Periodic Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
15.4 Energy Estimates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
16 Problems: Wave Equation 139
16.1 The Initial Value Problem . . . . . . . . . . . . . . . . . . . . . . . . . . 139
16.2 Initial/Boundary Value Problem . . . . . . . . . . . . . . . . . . . . . . 141
16.3 Similarity Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
16.4 Traveling Wave Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . 156
16.5 Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
16.6 Energy Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
16.7 Wave Equation in 2D and 3D . . . . . . . . . . . . . . . . . . . . . . . . 187
17 Problems: Laplace Equation 196
17.1 Greens Function and the Poisson Kernel . . . . . . . . . . . . . . . . . . 196
17.2 The Fundamental Solution . . . . . . . . . . . . . . . . . . . . . . . . . . 205
17.3 Radial Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
17.4 Weak Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
17.5 Uniqueness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
17.6 Self-Adjoint Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
17.7 Spherical Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
17.8 Harmonic Extensions, Subharmonic Functions . . . . . . . . . . . . . . . 249
Partial Dierential Equations Igor Yanovsky, 2005 5
18 Problems: Heat Equation 255
18.1 Heat Equation with Lower Order Terms . . . . . . . . . . . . . . . . . . 263
18.1.1 Heat Equation Energy Estimates . . . . . . . . . . . . . . . . . . 264
19 Contraction Mapping and Uniqueness - Wave 271
20 Contraction Mapping and Uniqueness - Heat 273
21 Problems: Maximum Principle - Laplace and Heat 279
21.1 Heat Equation - Maximum Principle and Uniqueness . . . . . . . . . . . 279
21.2 Laplace Equation - Maximum Principle . . . . . . . . . . . . . . . . . . 281
22 Problems: Separation of Variables - Laplace Equation 282
23 Problems: Separation of Variables - Poisson Equation 302
24 Problems: Separation of Variables - Wave Equation 305
25 Problems: Separation of Variables - Heat Equation 309
26 Problems: Eigenvalues of the Laplacian - Laplace 323
27 Problems: Eigenvalues of the Laplacian - Poisson 333
28 Problems: Eigenvalues of the Laplacian - Wave 338
29 Problems: Eigenvalues of the Laplacian - Heat 346
29.1 Heat Equation with Periodic Boundary Conditions in 2D
(with extra terms) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
30 Problems: Fourier Transform 365
31 Laplace Transform 385
32 Linear Functional Analysis 393
32.1 Norms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
32.2 Banach and Hilbert Spaces . . . . . . . . . . . . . . . . . . . . . . . . . 393
32.3 Cauchy-Schwarz Inequality . . . . . . . . . . . . . . . . . . . . . . . . . 393
32.4 H older Inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
32.5 Minkowski Inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394
32.6 Sobolev Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394
Partial Dierential Equations Igor Yanovsky, 2005 6
1 Trigonometric Identities
cos(a + b) = cos a cos b sina sinb
cos(a b) = cos a cos b + sina sinb
sin(a + b) = sin a cos b + cos a sinb
sin(a b) = sin a cos b cos a sinb
cos a cos b =
cos(a + b) + cos(a b)
2
sin a cos b =
sin(a + b) + sin(a b)
2
sin a sinb =
cos(a b) cos(a +b)
2
cos 2t = cos
2
t sin
2
t
sin2t = 2 sint cos t
cos
2
1
2
t =
1 + cos t
2
sin
2
1
2
t =
1 cos t
2
1 + tan
2
t = sec
2
t
cot
2
t + 1 = csc
2
t
cos x =
e
ix
+e
ix
2
sinx =
e
ix
e
ix
2i
coshx =
e
x
+e
x
2
sinhx =
e
x
e
x
2
d
dx
cosh x = sinh(x)
d
dx
sinh x = cosh(x)
cosh
2
x sinh
2
x = 1
_
du
a
2
+ u
2
=
1
a
tan
1
u
a
+C
_
du

a
2
u
2
= sin
1
u
a
+C
_
L
L
cos
nx
L
cos
mx
L
dx =
_
0 n ,= m
L n = m
_
L
L
sin
nx
L
sin
mx
L
dx =
_
0 n ,= m
L n = m
_
L
L
sin
nx
L
cos
mx
L
dx = 0
_
L
0
cos
nx
L
cos
mx
L
dx =
_
0 n ,= m
L
2
n = m
_
L
0
sin
nx
L
sin
mx
L
dx =
_
0 n ,= m
L
2
n = m
_
L
0
e
inx
e
imx
dx =
_
0 n ,= m
L n = m
_
L
0
e
inx
dx =
_
0 n ,= 0
L n = 0
_
sin
2
x dx =
x
2

sin x cos x
2
_
cos
2
x dx =
x
2
+
sin x cos x
2
_
tan
2
x dx = tanx x
_
sinx cos x dx =
cos
2
x
2
ln(xy) = ln(x) + ln(y)
ln
x
y
= ln(x) ln(y)
ln x
r
= r lnx
_
ln x dx = x ln x x
_
x ln x dx =
x
2
2
ln x
x
2
4
_
R
e
z
2
dz =

_
R
e

z
2
2
dz =

2
Partial Dierential Equations Igor Yanovsky, 2005 7
A =
_
a b
c d
_
, A
1
=
1
det(A)
_
d b
c a
_
Partial Dierential Equations Igor Yanovsky, 2005 8
2 Simple Eigenvalue Problem
X

+ X = 0
Boundary conditions Eigenvalues
n
Eigenfunctions X
n
X(0) = X(L) = 0
_
n
L
_
2
sin
n
L
x n = 1, 2, . . .
X(0) = X

(L) = 0
_
(n
1
2
)
L
_
2
sin
(n
1
2
)
L
x n = 1, 2, . . .
X

(0) = X(L) = 0
_
(n
1
2
)
L
_
2
cos
(n
1
2
)
L
x n = 1, 2, . . .
X

(0) = X

(L) = 0
_
n
L
_
2
cos
n
L
x n = 0, 1, 2, . . .
X(0) = X(L), X

(0) = X

(L)
_
2n
L
_
2
sin
2n
L
x n = 1, 2, . . .
cos
2n
L
x n = 0, 1, 2, . . .
X(L) = X(L), X

(L) = X

(L)
_
n
L
_
2
sin
n
L
x n = 1, 2, . . .
cos
n
L
x n = 0, 1, 2, . . .
X

X = 0
Boundary conditions Eigenvalues
n
Eigenfunctions X
n
X(0) = X(L) = 0, X

(0) = X

(L) = 0
_
n
L
_
4
sin
n
L
x n = 1, 2, . . .
X

(0) = X

(L) = 0, X

(0) = X

(L) = 0
_
n
L
_
4
cos
n
L
x n = 0, 1, 2, . . .
Partial Dierential Equations Igor Yanovsky, 2005 9
3 Separation of Variables:
Quick Guide
Laplace Equation: u = 0.
X

(x)
X(x)
=
Y

(y)
Y (y)
= .
X

+ X = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
Y

() + Y () = 0.
Wave Equation: u
tt
u
xx
= 0.
X

(x)
X(x)
=
T

(t)
T(t)
= .
X

+ X = 0.
u
tt
+ 3u
t
+u = u
xx
.
T

T
+ 3
T

T
+ 1 =
X

X
= .
X

+ X = 0.
u
tt
u
xx
+ u = 0.
T

T
+ 1 =
X

X
= .
X

+ X = 0.
u
tt
+u
t
= c
2
u
xx
+ u
xxt
, ( > 0)
X

X
= ,
1
c
2
T

T
+

c
2
T

T
=
_
1 +

c
2
T

T
_
X

X
.
4th Order: u
tt
= k u
xxxx
.

X
=
1
k
T

T
= .
X

X = 0.
Heat Equation: u
t
= ku
xx
.
T

T
= k
X

X
= .
X

+

k
X = 0.
4th Order: u
t
= u
xxxx
.
T

T
=
X

X
= .
X

X = 0.
4 Eigenvalues of the Lapla-
cian: Quick Guide
Laplace Equation: u
xx
+u
yy
+u = 0.
X

X
+
Y

Y
+ = 0. ( =
2
+
2
)
X

+
2
X = 0, Y

+
2
Y = 0.
u
xx
+u
yy
+k
2
u = 0.

X
=
Y

Y
+ k
2
= c
2
.
X

+ c
2
X = 0,
Y

+ (k
2
c
2
)Y = 0.
u
xx
+u
yy
+k
2
u = 0.

Y

Y
=
X

X
+ k
2
= c
2
.
Y

+c
2
Y = 0,
X

+ (k
2
c
2
)X = 0.
Partial Dierential Equations Igor Yanovsky, 2005 10
5 First-Order Equations
5.1 Quasilinear Equations
Consider the Cauchy problem for the quasilinear equation in two variables
a(x, y, u)u
x
+ b(x, y, u)u
y
= c(x, y, u),
with parameterized by (f(s), g(s), h(s)). The characteristic equations are
dx
dt
= a(x, y, z),
dy
dt
= b(x, y, z),
dz
dt
= c(x, y, z),
with initial conditions
x(s, 0) = f(s), y(s, 0) = g(s), z(s, 0) = h(s).
In a quasilinear case, the characteristic equations for
dx
dt
and
dy
dt
need not decouple from
the
dz
dt
equation; this means that we must take the z values into account even to nd
the projected characteristic curves in the xy-plane. In particular, this allows for the
possibility that the projected characteristics may cross each other.
The condition for solving for s and t in terms of x and y requires that the Jacobian
matrix be nonsingular:
J
_
x
s
y
s
x
t
y
t
_
= x
s
y
t
y
s
x
t
,= 0.
In particular, at t = 0 we obtain the condition
f

## (s) a(f(s), g(s), h(s)) ,= 0.

Burgers Equation. Solve the Cauchy problem
_
u
t
+ uu
x
= 0,
u(x, 0) = h(x).
(5.1)
The characteristic equations are
dx
dt
= z,
dy
dt
= 1,
dz
dt
= 0,
and may be parametrized by (s, 0, h(s)).
x = h(s)t +s, y = t, z = h(s).
u(x, y) = h(x uy) (5.2)
The characteristic projection in the xt-plane
1
passing through the point (s, 0) is the
line
x = h(s)t +s
along which u has the constant value u = h(s). Two characteristics x = h(s
1
)t + s
1
and x = h(s
2
)t + s
2
intersect at a point (x, t) with
t =
s
2
s
1
h(s
2
) h(s
1
)
.
1
y and t are interchanged here
Partial Dierential Equations Igor Yanovsky, 2005 11
From (5.2), we have
u
x
= h

(s)(1 u
x
t) u
x
=
h

(s)
1 +h

(s)t
Hence for h

(s) < 0, u
x
becomes innite at the positive time
t =
1
h

(s)
.
The smallest t for which this happens corresponds to the value s = s
0
at which h

(s)
has a minimum (i.e.h

## (s) has a maximum). At time T = 1/h

(s
0
) the solution u
Partial Dierential Equations Igor Yanovsky, 2005 12
5.2 Weak Solutions for Quasilinear Equations
5.2.1 Conservation Laws and Jump Conditions
Consider shocks for an equation
u
t
+f(u)
x
= 0, (5.3)
where f is a smooth function of u. If we integrate (5.3) with respect to x for a x b,
we obtain
d
dt
_
b
a
u(x, t) dx + f(u(b, t)) f(u(a, t)) = 0. (5.4)
This is an example of a conservation law. Notice that (5.4) implies (5.3) if u is C
1
, but
(5.4) makes sense for more general u.
Consider a solution of (5.4) that, for xed t, has a jump discontinuity at x = (t).
We assume that u, u
x
, and u
t
are continuous up to . Also, we assume that (t) is C
1
in t.
Taking a < (t) < b in (5.4), we obtain
d
dt
_
_

a
u dx +
_
b

u dx
_
+ f(u(b, t)) f(u(a, t))
=

(t)u
l
((t), t)

(t)u
r
((t), t) +
_

a
u
t
(x, t) dx +
_
b

u
t
(x, t) dx
+ f(u(b, t)) f(u(a, t)) = 0,
where u
l
and u
r
denote the limiting values of u from the left and right sides of the shock.
Letting a (t) and b (t), we get the Rankine-Hugoniot jump condition:

(t)(u
l
u
r
) + f(u
r
) f(u
l
) = 0,

(t) =
f(u
r
) f(u
l
)
u
r
u
l
.
5.2.2 Fans and Rarefaction Waves
For Burgers equation
u
t
+
_
1
2
u
2
_
x
= 0,
we have f

(u) = u, f

_
u
_
x
t
__
=
x
t
u
_
x
t
_
=
x
t
.
For a rarefaction fan emanating from (s, 0) on xt-plane, we have:
u(x, t) =

u
l
,
xs
t
f

(u
l
) = u
l
,
xs
t
, u
l

xs
t
u
r
,
u
r
,
xs
t
f

(u
r
) = u
r
.
Partial Dierential Equations Igor Yanovsky, 2005 13
5.3 General Nonlinear Equations
5.3.1 Two Spatial Dimensions
Write a general nonlinear equation F(x, y, u, u
x
, u
y
) = 0 as
F(x, y, z, p, q) = 0.
is parameterized by
:
_
f(s)
..
x(s,0)
, g(s)
..
y(s,0)
, h(s)
..
z(s,0)
, (s)
..
p(s,0)
, (s)
..
q(s,0)
_
We need to complete to a strip. Find (s) and (s), the initial conditions for p(s, t)
and q(s, t), respectively:
F(f(s), g(s), h(s), (s), (s)) = 0
h

(s) = (s)f

(s) + (s)g

(s)
The characteristic equations are
dx
dt
= F
p
dy
dt
= F
q
dz
dt
= pF
p
+ qF
q
dp
dt
= F
x
F
z
p
dq
dt
= F
y
F
z
q
We need to have the Jacobian condition. That is, in order to solve the Cauchy problem
in a neighborhood of , the following condition must be satised:
f

(s) F
q
[f, g, h, , ](s) g

(s) F
p
[f, g, h, , ](s) ,= 0.
5.3.2 Three Spatial Dimensions
Write a general nonlinear equation F(x
1
, x
2
, x
3
, u, u
x
1
, u
x
2
, u
x
3
) = 0 as
F(x
1
, x
2
, x
3
, z, p
1
, p
2
, p
3
) = 0.
is parameterized by
:
_
f
1
(s
1
, s
2
)
. .
x
1
(s
1
,s
2
,0)
, f
2
(s
1
, s
2
)
. .
x
2
(s
1
,s
2
,0)
, f
3
(s
1
, s
2
)
. .
x
3
(s
1
,s
2
,0)
, h(s
1
, s
2
)
. .
z(s
1
,s
2
,0)
,
1
(s
1
, s
2
)
. .
p
1
(s
1
,s
2
,0)
,
2
(s
1
, s
2
)
. .
p
2
(s
1
,s
2
,0)
,
3
(s
1
, s
2
)
. .
p
3
(s
1
,s
2
,0)
_
We need to complete to a strip. Find
1
(s
1
, s
2
),
2
(s
1
, s
2
), and
3
(s
1
, s
2
), the initial
conditions for p
1
(s
1
, s
2
, t), p
2
(s
1
, s
2
, t), and p
3
(s
1
, s
2
, t), respectively:
F
_
f
1
(s
1
, s
2
), f
2
(s
1
, s
2
), f
3
(s
1
, s
2
), h(s
1
, s
2
),
1
,
2
,
3
_
= 0

h
s
1
=
1
f
1
s
1
+
2
f
2
s
1
+
3
f
3
s
1

h
s
2
=
1
f
1
s
2
+
2
f
2
s
2
+
3
f
3
s
2
The characteristic equations are
dx
1
dt
= F
p
1
dx
2
dt
= F
p
2
dx
3
dt
= F
p
3
dz
dt
= p
1
F
p
1
+ p
2
F
p
2
+p
3
F
p
3
dp
1
dt
= F
x
1
p
1
F
z
dp
2
dt
= F
x
2
p
2
F
z
dp
3
dt
= F
x
3
p
3
F
z
Partial Dierential Equations Igor Yanovsky, 2005 14
6 Second-Order Equations
6.1 Classication by Characteristics
Consider the second-order equation in which the derivatives of second-order all occur
linearly, with coecients only depending on the independent variables:
a(x, y)u
xx
+ b(x, y)u
xy
+ c(x, y)u
yy
= d(x, y, u, u
x
, u
y
). (6.1)
The characteristic equation is
dy
dx
=
b

b
2
4ac
2a
.
b
2
4ac > 0 two characteristics, and (6.1) is called hyperbolic;
b
2
4ac = 0 one characteristic, and (6.1) is called parabolic;
b
2
4ac < 0 no characteristics, and (6.1) is called elliptic.
These denitions are all taken at a point x
0
R
2
; unless a, b, and c are all constant,
the type may change with the point x
0
.
6.2 Canonical Forms and General Solutions
u
xx
u
yy
= 0 is hyperbolic (one-dimensional wave equation).
u
xx
u
y
= 0 is parabolic (one-dimensional heat equation).
u
xx
+ u
yy
= 0 is elliptic (two-dimensional Laplace equation).
By the introduction of new coordinates and in place of x and y, the equation
(6.1) may be transformed so that its principal part takes the form , , or .
If (6.1) is hyperbolic, parabolic, or elliptic, there exists a change of variables (x, y) and
(x, y) under which (6.1) becomes, respectively,
u

=

d(, , u, u

, u

) u
x x
u
y y
=

d( x, y, u, u
x
, u
y
),
u

=

d(, , u, u

, u

),
u

+ u

=

d(, , u, u

, u

).
Example 1. Reduce to canonical form and nd the general solution:
u
xx
+ 5u
xy
+ 6u
yy
= 0. (6.2)
Proof. a = 1, b = 5, c = 6 b
2
4ac = 1 > 0 hyperbolic two
characteristics.
The characteristics are found by solving
dy
dx
=
5 1
2
=
_
3
2
to nd y = 3x +c
1
and y = 2x +c
2
.
Partial Dierential Equations Igor Yanovsky, 2005 15
Let (x, y) = 3x y, (x, y) = 2x y.

x
= 3,
x
= 2,

y
= 1,
y
= 1.
u = u((x, y), (x, y));
u
x
= u

x
+u

x
= 3u

+ 2u

,
u
y
= u

y
+u

y
= u

,
u
xx
= (3u

+ 2u

)
x
= 3(u

x
+u

x
) + 2(u

x
+ u

x
) = 9u

+ 12u

+ 4u

,
u
xy
= (3u

+ 2u

)
y
= 3(u

y
+ u

y
) + 2(u

y
+ u

y
) = 3u

5u

2u

,
u
yy
= (u

+u

)
y
= (u

y
+u

y
+u

y
+u

y
) = u

+ 2u

+u

.
Inserting these expressions into (6.2) and simplifying, we obtain
u

## = 0, which is the Canonical form,

u

= f(),
u = F() +G(),
u(x, y) = F(3x y) +G(2x y), General solution.
Example 2. Reduce to canonical form and nd the general solution:
y
2
u
xx
2yu
xy
+u
yy
= u
x
+ 6y. (6.3)
Proof. a = y
2
, b = 2y, c = 1 b
2
4ac = 0 parabolic one characteristic.
The characteristics are found by solving
dy
dx
=
2y
2y
2
=
1
y
to nd
y
2
2
+ c = x.
Let =
y
2
2
+ x. We must choose a second constant function (x, y) so that is not
parallel to . Choose (x, y) = y.

x
= 1,
x
= 0,

y
= y,
y
= 1.
u = u((x, y), (x, y));
u
x
= u

x
+u

x
= u

,
u
y
= u

y
+ u

y
= yu

+ u

,
u
xx
= (u

)
x
= u

x
+ u

x
= u

,
u
xy
= (u

)
y
= u

y
+u

y
= yu

+u

,
u
yy
= (yu

+u

)
y
= u

+y(u

y
+u

y
) + (u

y
+u

y
)
= u

+ y
2
u

+ 2yu

+ u

.
Partial Dierential Equations Igor Yanovsky, 2005 16
Inserting these expressions into (6.3) and simplifying, we obtain
u

= 6y,
u

## = 6, which is the Canonical form,

u

= 3
2
+ f(),
u =
3
+f() + g(),
u(x, y) = y
3
+ y f
_
y
2
2
+ x
_
+ g
_
y
2
2
+x
_
, General solution.
Partial Dierential Equations Igor Yanovsky, 2005 17
Problem (F03, #4). Find the characteristics of the partial dierential equation
xu
xx
+ (x y)u
xy
yu
yy
= 0, x > 0, y > 0, (6.4)
and then show that it can be transformed into the canonical form
(
2
+ 4)u

+ u

= 0
whence and are suitably chosen canonical coordinates. Use this to obtain the general
solution in the form
u(, ) = f() +
_

g(

) d

(
2
+ 4

)
1
2
where f and g are arbitrary functions of and .
Proof. a = x, b = x y, c = y b
2
4ac = (x y)
2
+ 4xy > 0 for x > 0,
y > 0 hyperbolic two characteristics.
The characteristics are found by solving
dy
dx
=
b

b
2
4ac
2a
=
x y
_
(x y)
2
+ 4xy
2x
=
x y (x + y)
2x
=
_
2x
2x
= 1

2y
2x
=
y
x
y = x + c
1
,
dy
y
=
dx
x
,
ln y = ln x
1
+ c
2
,
y =
c
2
x
.
Let = x y and = xy

x
= 1,
x
= y,

y
= 1,
y
= x.
u = u((x, y), (x, y));
u
x
= u

x
+u

x
= u

+yu

,
u
y
= u

y
+u

y
= u

+xu

,
u
xx
= (u

+ yu

)
x
= u

x
+u

x
+ y(u

x
+u

x
) = u

+ 2yu

+ y
2
u

,
u
xy
= (u

+ yu

)
y
= u

y
+ u

y
+u

+ y(u

y
+u

y
) = u

+ xu

+ u

yu

+ xyu

,
u
yy
= (u

+xu

)
y
= u

y
u

y
+ x(u

y
+u

y
) = u

2xu

+x
2
u

,
Inserting these expressions into (6.4), we obtain
x(u

+ 2yu

+y
2
u

) + (x y)(u

+ xu

+ u

yu

+ xyu

) y(u

2xu

+x
2
u

) = 0,
(x
2
+ 2xy +y
2
)u

+ (x y)u

= 0,
_
(x y)
2
+ 4xy
_
u

+ (x y)u

= 0,
(
2
+ 4)u

+ u

## = 0, which is the Canonical form.

Partial Dierential Equations Igor Yanovsky, 2005 18
We need to integrate twice to get the general solution:
(
2
+ 4)(u

+ u

= 0,
_
(u

d =
_

2
+ 4
d,
ln u

=
1
2
ln (
2
+ 4) + g(),
ln u

= ln (
2
+ 4)

1
2
+ g(),
u

=
g()
(
2
+ 4)
1
2
,
u(, ) = f() +
_
g() d
(
2
+ 4)
1
2
, General solution.
Partial Dierential Equations Igor Yanovsky, 2005 19
6.3 Well-Posedness
Problem (S99, #2). In R
2
consider the unit square dened by 0 x, y 1.
Consider
a) u
x
+ u
yy
= 0;
b) u
xx
+ u
yy
= 0;
c) u
xx
u
yy
= 0.
Prescribe data for each problem separately on the boundary of so that each of these
Proof. The initial / boundary value problem for the HEAT EQUATION is well-
posed:

u
t
= u x , t > 0,
u(x, 0) = g(x) x ,
u(x, t) = 0 x , t > 0.
Existence - by eigenfunction expansion.
Uniqueness and continuous dependence on the data -
by maximum principle.
The method of eigenfunction expansion and maximum
principle give well-posedness for more general problems:

u
t
= u + f(x, t) x , t > 0,
u(x, 0) = g(x) x ,
u(x, t) = h(x, t) x , t > 0.
It is also possible to replace the Dirichlet boundary condition u(x, t) = h(x, t) by a
Neumann or Robin condition, provided we replace
n
,
n
by the eigenvalues and eigen-
functions for the appropriate boundary value problem.
a) Relabel the variables (x t, y x).
We have the BACKWARDS HEAT EQUATION:
u
t
+u
xx
= 0.
Need to dene initial conditions u(x, 1) = g(x), and
either Dirichlet, Neumann, or Robin boundary conditions.
b) The solution to the LAPLACE EQUATION
_
u = 0 in ,
u = g on
exists if g is continuous on , by Perrons method. Maximum principle gives unique-
ness.
To show the continuous dependence on the data, assume
_
u
1
= 0 in ,
u
1
= g
1
on ;
_
u
2
= 0 in ,
u
2
= g
2
on .
Partial Dierential Equations Igor Yanovsky, 2005 20
Then (u
1
u
2
) = 0 in . Maximum principle gives
max

(u
1
u
2
) = max

(g
1
g
2
). Thus,
max

[u
1
u
2
[ = max

[g
1
g
2
[.
Thus, [u
1
u
2
[ is bounded by [g
1
g
2
[, i.e. continuous dependence on data.
Perrons method gives existence of the solution to the POISSON EQUATION
_
u = f in ,
u
n
= h on
for f C

() and h C

_

hdS =
_

## f dx. It is unique up to an additive constant.

c) Relabel the variables (y t).
The solution to the WAVE EQUATION
u
tt
u
xx
= 0,
is of the form u(x, y) = F(x +t) + G(x t).
The existence of the solution to the initial/boundary value problem

u
tt
u
xx
= 0 0 < x < 1, t > 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) 0 < x < 1
u(0, t) = (t), u(1, t) = (t) t 0.
is given by the method of separation of variables
(expansion in eigenfunctions)
and by the parallelogram rule.
Uniqueness is given by the energy method.
Need initial conditions u(x, 0), u
t
(x, 0).
Prescribe u or u
x
for each of the two boundaries.
Partial Dierential Equations Igor Yanovsky, 2005 21
Problem (F95, #7). Let a, b be real numbers. The PDE
u
y
+au
xx
+bu
yy
= 0
is to be solved in the box = [0, 1]
2
.
Find data, given on an appropriate part of , that will make this a well-posed prob-
lem.
Cover all cases according to the possible values of a and b. Justify your statements.
Proof.
ab < 0 two sets of characteristics hyperbolic.
Relabeling the variables (y t), we have
u
tt
+
a
b
u
xx
=
1
b
u
t
.
The solution of the equation is of the form
u(x, t) = F(x +
_

a
b
t) + G(x
_

a
b
t).
Existence of the solution to the initial/boundary
value problem is given by the method of separation
of variables (expansion in eigenfunctions)
and by the parallelogram rule.
Uniqueness is given by the energy method.
Need initial conditions u(x, 0), u
t
(x, 0).
Prescribe u or u
x
for each of the two boundaries.
ab > 0 no characteristics elliptic.
The solution to the Laplace equation with boundary conditions u = g on exists
if g is continuous on , by Perrons method.
To show uniqueness, we use maximum principle. Assume there are two solutions u
1
and u
2
with with u
1
= g(x), u
2
= g(x) on . By maximum principle
max

(u
1
u
2
) = max

## (g(x) g(x)) = 0. Thus, u

1
= u
2
.
ab = 0 one set of characteristics parabolic.
a = b = 0. We have u
y
= 0, a rst-order ODE.
u must be specied on y = 0, i.e. x -axis.
a = 0, b ,= 0. We have u
y
+bu
yy
= 0, a second-order ODE.
u and u
y
must be specied on y = 0, i.e. x -axis.
a > 0, b = 0. We have a Backwards Heat Equation.
u
t
= au
xx
.
Need to dene initial conditions u(x, 1) = g(x), and
either Dirichlet, Neumann, or Robin boundary conditions.
Partial Dierential Equations Igor Yanovsky, 2005 22
a < 0, b = 0. We have a Heat Equation.
u
t
= au
xx
.
The initial / boundary value problem for the heat equation is well-posed:

u
t
= u x , t > 0,
u(x, 0) = g(x) x ,
u(x, t) = 0 x , t > 0.
Existence - by eigenfunction expansion.
Uniqueness and continuous dependence on the data -
by maximum principle.
Partial Dierential Equations Igor Yanovsky, 2005 23
7 Wave Equation
The one-dimensional wave equation is
u
tt
c
2
u
xx
= 0. (7.1)
The characteristic equation with a = c
2
, b = 0, c = 1 would be
dt
dx
=
b

b
2
4ac
2a
=

4c
2
2c
2
=
1
c
,
and thus
t =
1
c
x +c
1
and t =
1
c
x +c
2
,
= x +ct = x ct,
which transforms (7.1) to
u

= 0. (7.2)
The general solution of (7.2) is u(, ) = F()+G(), where F and G are C
1
functions.
Returning to the variables x, t we nd that
u(x, t) = F(x +ct) +G(x ct) (7.3)
solves (7.1). Moreover, u is C
2
provided that F and G are C
2
.
If F 0, then u has constant values along the lines xct = const, so may be described
as a wave moving in the positive x-direction with speed dx/dt = c; if G 0, then u is
a wave moving in the negative x-direction with speed c.
7.1 The Initial Value Problem
For an initial value problem, consider the Cauchy problem
_
u
tt
c
2
u
xx
= 0,
u(x, 0) = g(x), u
t
(x, 0) = h(x).
(7.4)
Using (7.3) and (7.4), we nd that F and G satisfy
F(x) +G(x) = g(x), cF

(x) cG

## (x) = h(x). (7.5)

If we integrate the second equation in (7.5), we get cF(x) cG(x) =
_
x
0
h() d + C.
Combining this with the rst equation in (7.5), we can solve for F and G to nd
_
F(x) =
1
2
g(x) +
1
2c
_
x
0
h() d +C
1
G(x) =
1
2
g(x)
1
2c
_
x
0
h() d C
1
,
Using these expressions in (7.3), we obtain dAlemberts Formula for the solution
of the initial value problem (7.4):
u(x, t) =
1
2
(g(x +ct) + g(x ct)) +
1
2c
_
x+ct
xct
h() d.
If g C
2
and h C
1
, then dAlemberts Formula denes a C
2
solution of (7.4).
Partial Dierential Equations Igor Yanovsky, 2005 24
7.2 Weak Solutions
Equation (7.3) denes a weak solution of (7.1) when F and G are not C
2
functions.
Consider the parallelogram with sides that are
segments of characteristics. Since
u(x, t) = F(x +ct) + G(x ct), we have
u(A) + u(C) =
= F(k
1
) + G(k
3
) + F(k
2
) +G(k
4
)
= u(B) +u(D),
which is the parallelogram rule.
7.3 Initial/Boundary Value Problem

u
tt
c
2
u
xx
= 0 0 < x < L, t > 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) 0 < x < L
u(0, t) = (t), u(L, t) = (t) t 0.
(7.6)
Use separation of variables to obtain an expansion in eigenfunctions. Find u(x, t) in
the form
u(x, t) =
a
0
(t)
2
+

n=1
a
n
(t) cos
nx
L
+b
n
(t) sin
nx
L
.
7.4 Duhamels Principle

u
tt
c
2
u
xx
= f(x, t)
u(x, 0) = 0
u
t
(x, 0) = 0.

U
tt
c
2
U
xx
= 0
U(x, 0, s) = 0
U
t
(x, 0, s) = f(x, s)
u(x, t) =
_
t
0
U(x, ts, s) ds.

n
+
n
a
n
= f
n
(t)
a
n
(0) = 0
a

n
(0) = 0

n
+
n
a
n
= 0
a
n
(0, s) = 0
a

n
(0, s) = f
n
(s)
a
n
(t) =
_
t
0
a
n
(t s, s) ds.
7.5 The Nonhomogeneous Equation
Consider the nonhomogeneous wave equation with homogeneous initial conditions:
_
u
tt
c
2
u
xx
= f(x, t),
u(x, 0) = 0, u
t
(x, 0) = 0.
(7.7)
Duhamels Principle provides the solution of (7.7):
u(x, t) =
1
2c
_
t
0
__
x+c(ts)
xc(ts)
f(, s) d
_
ds.
If f(x, t) is C
1
in x and C
0
in t, then Duhamels Principle provides a C
2
solution of
(7.7).
Partial Dierential Equations Igor Yanovsky, 2005 25
We can solve (7.7) with nonhomogeneous initial conditions,
_
u
tt
c
2
u
xx
= f(x, t),
u(x, 0) = g(x), u
t
(x, 0) = h(x),
(7.8)
by adding together dAlemberts formula and Duhamels principle gives the solution:
u(x, t) =
1
2
(g(x + ct) +g(x ct)) +
1
2c
_
x+ct
xct
h() d +
1
2c
_
t
0
__
x+c(ts)
xc(ts)
f(, s) d
_
ds.
Partial Dierential Equations Igor Yanovsky, 2005 26
7.6 Higher Dimensions
7.6.1 Spherical Means
For a continuous function u(x) on R
n
, its spherical mean or average on a sphere of
radius r and center x is
M
u
(x, r) =
1

n
_
||=1
u(x +r)dS

,
where
n
is the area of the unit sphere S
n1
= R
n
: [[ = 1 and dS

is surface
measure. Since u is continuous in x, M
u
(x, r) is continuous in x and r, so
M
u
(x, 0) = u(x).
Using the chain rule, we nd

r
M
u
(x, r) =
1

n
_
||=1
n

i=1
u
x
i
(x + r)
i
dS

=
To compute the RHS, we apply the divergence theorem in = R
n
: [[ < 1,
which has boundary = S
n1
and exterior unit normal n() = . The integrand is
V n where V () = r
1

u(x + r) =
x
u(x + r). Computing the divergence of V ,
we obtain
div V () = r
n

i=1
u
x
i
x
i
(x + r) = r
x
u(x +r), so,
=
1

n
_
||<1
r
x
u(x + r) d =
r

x
_
||<1
u(x +r) d (

= r)
=
r

n
1
r
n

x
_
|

|<r
u(x +

) d

(spherical coordinates)
=
1

n
r
n1

x
_
r
0

n1
_
||=1
u(x + ) dS

d
=
1

n
r
n1

n

x
_
r
0

n1
M
u
(x, ) d =
1
r
n1

x
_
r
0

n1
M
u
(x, ) d.
If we multiply by r
n1
, dierentiate with respect to r, and then divide by r
n1
,
we obtain the Darboux equation:
_

2
r
2
+
n 1
r

r
_
M
u
(x, r) =
x
M
u
(x, r).
Note that for a radial function u = u(r), we have M
u
= u, so the equation provides the
Laplacian of u in spherical coordinates.
7.6.2 Application to the Cauchy Problem
We want to solve the equation
u
tt
= c
2
u x R
n
, t > 0, (7.9)
u(x, 0) = g(x), u
t
(x, 0) = h(x) x R
n
.
We use Poissons method of spherical means to reduce this problem to a partial dier-
ential equation in the two variables r and t.
Partial Dierential Equations Igor Yanovsky, 2005 27
Suppose that u(x, t) solves (7.9). We can view t as a parameter and take the spherical
mean to obtain M
u
(x, r, t), which satises

2
t
2
M
u
(x, r, t) =
1

n
_
||=1
u
tt
(x +r, t)dS

=
1

n
_
||=1
c
2
u(x +r, t)dS

= c
2
M
u
(x, r, t).
Invoking the Darboux equation, we obtain the Euler-Poisson-Darboux equation:

2
t
2
M
u
(x, r, t) = c
2
_

2
r
2
+
n 1
r

r
_
M
u
(x, r, t).
The initial conditions are obtained by taking the spherical means:
M
u
(x, r, 0) = M
g
(x, r),
M
u
t
(x, r, 0) = M
h
(x, r).
If we nd M
u
(x, r, t), we can then recover u(x, t) by:
u(x, t) = lim
r0
M
u
(x, r, t).
7.6.3 Three-Dimensional Wave Equation
When n = 3, we can write the Euler-Poisson-Darboux equation as
2

2
t
2
_
rM
u
(x, r, t)
_
= c
2

2
r
2
_
rM
u
(x, r, t)
_
.
For each xed x, consider V
x
(r, t) = rM
u
(x, r, t) as a solution of the one-dimensional
wave equation in r, t > 0:

2
t
2
V
x
(r, t) = c
2

2
r
2
V
x
(r, t),
V
x
(r, 0) = rM
g
(x, r) G
x
(r), (IC)
V
x
t
(r, 0) = rM
h
(x, r) H
x
(r), (IC)
V
x
(0, t) = lim
r0
rM
u
(x, r, t) = 0 u(x, t) = 0. (BC)
G
x
(0) = H
x
(0) = 0.
We may extend G
x
and H
x
as odd functions of r and use dAlemberts formula for
V
x
(r, t):
V
x
(r, t) =
1
2
_
G
x
(r + ct) + G
x
(r ct)
_
+
1
2c
_
r+ct
rct
H
x
() d.
Since G
x
and H
x
are odd functions, we have for r < ct:
G
x
(r ct) = G
x
(ct r) and
_
r+ct
rct
H
x
() d =
_
ct+r
ctr
H
x
() d.
After some more manipulations, we nd that the solution of (7.9) is given by the
Kirchhos formula:
u(x, t) =
1
4

t
_
t
_
||=1
g(x +ct)dS

_
+
t
4
_
||=1
h(x +ct)dS

.
If g C
3
(R
3
) and h C
2
(R
3
), then Kirchhos formula denes a C
2
-solution of (7.9).
2
It is seen by expanding the equation below.
Partial Dierential Equations Igor Yanovsky, 2005 28
7.6.4 Two-Dimensional Wave Equation
This problem is solved by Hadamards method of descent, namely, view (7.9) as a special
case of a three-dimensional problem with initial conditions independent of x
3
.
We need to convert surface integrals in R
3
to domain integrals in R
2
.
u(x
1
, x
2
, t) =
1
4

t
_
2t
_

2
1
+
2
2
<1
g(x
1
+ ct
1
, x
2
+ct
2
)d
1
d
2
_
1
2
1

2
2
_
+
t
4
_
2
_

2
1
+
2
2
<1
h(x
1
+ ct
1
, x
2
+ ct
2
)d
1
d
2
_
1
2
1

2
2
If g C
3
(R
2
) and h C
2
(R
2
), then this equation denes a C
2
-solution of (7.9).
7.6.5 Huygens Principle
Notice that u(x, t) depends only on the Cauchy data g, h on the surface of the hyper-
sphere x +ct : [[ = 1 in R
n
, n = 2k + 1; in other words we have sharp signals.
If we use the method of descent to obtain the solution for n = 2k, the hypersurface
integrals become domain integrals. This means that there are no sharp signals.
The fact that sharp signals exist only for odd dimensions n 3 is known as Huygens
principle.
3
3
For x R
n
:

t
__
||=1
f(x + t)dS
_
=
1
t
n1
_
|y|t
f(x + y)dy

t
__
|y|t
f(x + y)dy
_
= t
n1
__
||=1
f(x + t)dS
_
Partial Dierential Equations Igor Yanovsky, 2005 29
7.7 Energy Methods
Suppose u C
2
(R
n
(0, )) solves
_
u
tt
= c
2
u x R
n
, t > 0,
u(x, 0) = g(x), u
t
(x, 0) = h(x) x R
n
,
(7.10)
where g and h have compact support.
Dene energy for a function u(x, t) at time t by
E(t) =
1
2
_
R
n
(u
2
t
+c
2
[u[
2
) dx.
If we dierentiate this energy function, we obtain
dE
dt
=
d
dt
_
1
2
_
R
n
_
u
2
t
+c
2
n

i=1
u
2
x
i
_
dx
_
=
_
R
n
_
u
t
u
tt
+c
2
n

i=1
u
x
i
u
x
i
t
_
dx
=
_
R
n
u
t
u
tt
dx +c
2
_
n

i=1
u
x
i
u
t
_
R
n

_
R
n
c
2
n

i=1
u
x
i
x
i
u
t
dx
=
_
R
n
u
t
(u
tt
c
2
u) dx = 0,
or
dE
dt
=
d
dt
_
1
2
_
R
n
_
u
2
t
+c
2
n

i=1
u
2
x
i
_
dx
_
=
_
R
n
_
u
t
u
tt
+c
2
n

i=1
u
x
i
u
x
i
t
_
dx
=
_
R
n
_
u
t
u
tt
+ c
2
u u
t
_
dx
=
_
R
n
u
t
u
tt
dx +c
2
__
R
n
u
t
u
n
ds
_
R
n
u
t
u dx
_
=
_
R
n
u
t
(u
tt
c
2
u) dx = 0.
Hence, E(t) is constant, or E(t) E(0).
In particular, if u
1
and u
2
are two solutions of (7.10), then w = u
1
u
2
has zero Cauchy
data and hence E
w
(0) = 0. By discussion above, E
w
(t) 0, which implies w(x, t)
const. But w(x, 0) = 0 then implies w(x, t) 0, so the solution is unique.
Partial Dierential Equations Igor Yanovsky, 2005 30
7.8 Contraction Mapping Principle
Suppose X is a complete metric space with distance function represented by d(, ).
A mapping T : X X is a strict contraction if there exists 0 < < 1 such that
d(Tx, Ty) d(x, y) x, y X.
An obvious example on X = R
n
is Tx = x, which shrinks all of R
n
, leaving 0 xed.
The Contraction Mapping Principle. If X is a complete metric space and T :
X X is a strict contraction, then T has a unique xed point.
The process of replacing a dierential equation by an integral equation occurs in
time-evolution partial dierential equations.
The Contraction Mapping Principle is used to establish the local existence and unique-
ness of solutions to various nonlinear equations.
Partial Dierential Equations Igor Yanovsky, 2005 31
8 Laplace Equation
Consider the Laplace equation
u = 0 in R
n
(8.1)
and the Poisson equation
u = f in R
n
. (8.2)
Solutions of (8.1) are called harmonic functions in .
Cauchy problems for (8.1) and (8.2) are not well posed. We use separation of variables
for some special domains to nd boundary conditions that are appropriate for (8.1),
(8.2).
Dirichlet problem: u(x) = g(x), x
Neumann problem:
u(x)
n
= h(x), x
Robin problem:
u
n
+u = , x
8.1 Greens Formulas
_

u v dx =
_

v
u
n
ds
_

v u dx (8.3)
_

_
v
u
n
u
v
n
_
ds =
_

(vu uv) dx
_

u
n
ds =
_

u dx (v = 1 in (8.3))
_

[u[
2
dx =
_

u
u
n
ds
_

u u dx (u = v in (8.3))
_

u
x
v
x
dxdy =
_

vu
x
n
1
ds
_

vu
xx
dxdy n = (n
1
, n
2
) R
2
_

u
x
k
v dx =
_

uvn
k
ds
_

uv
x
k
dx n = (n
1
, . . . , n
n
) R
n
.
_

u
2
v dx =
_

u
v
n
ds
_

v
u
n
ds +
_

uv dx.
_

_
u
2
v v
2
u
_
dx =
_

_
u
v
n
v
u
n
_
ds +
_

_
u
v
n
v
u
n
_
ds.
Partial Dierential Equations Igor Yanovsky, 2005 32
8.2 Polar Coordinates
Polar Coordinates. Let f : R
n
R be continuous. Then
_
R
n
f dx =
_

0
_
_
Br(x
0
)
f dS
_
dr
for each x
0
R
n
. In particular
d
dr
_
_
Br(x
0
)
f dx
_
=
_
Br(x
0
)
f dS
for each r > 0.
u = u(x(r, ), y(r, ))
x(r, ) = r cos
y(r, ) = r sin
u
r
= u
x
x
r
+ u
y
y
r
= u
x
cos + u
y
sin,
u

= u
x
x

+u
y
y

= u
x
r sin + u
y
r cos ,
u
rr
= (u
x
cos +u
y
sin)
r
= (u
xx
x
r
+ u
xy
y
r
) cos + (u
yx
x
r
+u
yy
y
r
) sin
= u
xx
cos
2
+ 2u
xy
cos sin +u
yy
sin
2
,
u

= (u
x
r sin +u
y
r cos )

= (u
xx
x

u
xy
y

)r sin u
x
r cos + (u
yx
x

+u
yy
y

)r cos u
y
r sin
= (u
xx
r sin u
xy
r cos )r sin u
x
r cos + (u
yx
r sin + u
yy
r cos )r cos u
y
r sin
= r
2
(u
xx
sin
2
2u
xy
cos sin +u
yy
cos
2
) r(u
x
cos + u
y
sin).
u
rr
+
1
r
2
u

= u
xx
cos
2
+ 2u
xy
cos sin +u
yy
sin
2
+u
xx
sin
2
2u
xy
cos sin +u
yy
cos
2

1
r
(u
x
cos + u
y
sin )
= u
xx
+ u
yy

1
r
u
r
.
u
xx
+u
yy
= u
rr
+
1
r
u
r
+
1
r
2
u

2
x
2
+

2
y
2
=

2
r
2
+
1
r

r
+
1
r
2

2
.
8.3 Polar Laplacian in R
2
u =
1
r
_
ru
r
_
r
=
_

2
r
2
+
1
r

r
_
u.
8.4 Spherical Laplacian in R
3
and R
n
u =
_

2
r
2
+
n 1
r

r
_
u.
In R
3
:
4
u =
1
r
2
_
r
2
u
r
_
r
=
1
r
_
ru
_
rr
=
_

2
r
2
+
2
r

r
_
u.
4
These formulas are taken from S. Farlow, p. 411.
Partial Dierential Equations Igor Yanovsky, 2005 33
8.5 Cylindrical Laplacian in R
3
u =
1
r
_
ru
r
_
r
=
_

2
r
2
+
1
r

r
_
u.
8.6 Mean Value Theorem
Gauss Mean Value Theorem. If u C
2
() is harmonic in , let and pick
r > 0 so that B
r
() = x : [x [ r . Then
u() = M
u
(, r)
1

n
_
|x|=1
u( +rx) dS
x
,
where
n
is the measure of the (n 1)-dimensional sphere in R
n
.
8.7 Maximum Principle
Maximum Principle. If u C
2
() satises u 0 in , then either u is a constant,
or
u() < sup
x
u(x)
for all .
Proof. We may assume A = sup
x
u(x) , so by continuity of u we know that
x : u(x) = A is relatively closed in . But since
u()
n

n
_
|x|1
u( + rx) dx,
if u() = A at an interior point , then u(x) = A for all x in a ball about , so
x : u(x) = A is open. The connectedness of implies u() < A or u() A for
all .
The maximum principle shows that u C
2
() with u 0 can attain an interior
maximum only if u is constant. In particular, if is compact, and u C
2
() C()
satises u 0 in , we have the weak maximum principle:
max
x
u(x) = max
x
u(x).
Partial Dierential Equations Igor Yanovsky, 2005 34
8.8 The Fundamental Solution
A fundamental solution K(x) for the Laplace operator is a distribution satisfying
K(x) = (x) (8.4)
where is the delta distribution supported at x = 0. In order to solve (8.4), we should
rst observe that is symmetric in the variables x
1
, . . . , x
n
, and (x) is also radially
symmetric (i.e., its value only depends on r = [x[). Thus, we try to solve (8.4) with a
radially symmetric function K(x). Since (x) = 0 for x ,= 0, we see that (8.4) requires
K to be harmonic for r > 0. For the radially symmetric function K, Laplace equation
becomes (K = K(r)):

2
K
r
2
+
n 1
r
K
r
= 0. (8.5)
The general solution to (8.5) is
K(r) =
_
c
1
+c
2
log r if n = 2
c
1
+c
2
r
2n
if n 3.
(8.6)
After we determine c
2
, we nd the fundamental solution for the Laplace operator:
K(x) =
_
1
2
log r if n = 2
1
(2n)n
r
2n
if n 3.
We can derive, (8.6) for any given n. For intance, when n = 3, we have:
K

+
2
r
K

= 0.
Let
K =
1
r
w(r),
K

=
1
r
w

1
r
2
w,
K

=
1
r
w

2
r
2
w

+
2
r
3
w.
Plugging these into , we obtain:
1
r
w

= 0, or
w

= 0.
Thus,
w = c
1
r + c
2
,
K =
1
r
w(r) = c
1
+
c
2
r
.
See the similar problem, F99, #2, where the fundamental solution for ( I) is
found in the process.
Partial Dierential Equations Igor Yanovsky, 2005 35
Find the Fundamental Solution of the Laplace Operator for n = 3
We found that starting with the Laplacian in R
3
for a radially symmetric function K,
K

+
2
r
K

= 0,
and letting K =
1
r
w(r), we obtained the equation: w = c
1
r + c
2
, which implied:
K = c
1
+
c
2
r
.
We now nd the constant c
2
that ensures that for v C

0
(R
3
), we have
_
R
3
K([x[) v(x) dx = v(0).
Suppose v(x) 0 for [x[ R and let = B
R
(0); for small > 0 let

= B

(0).
K([x[) is harmonic (K([x[) = 0) in

## . Consider Greens identity (

=
B

(0)):
_

K([x[)v dx =
_

_
K([x[)
v
n
v
K([x[)
n
_
dS
. .
=0, since v0 for xR
+
_
B(0)
_
K([x[)
v
n
v
K([x[)
n
_
dS.
lim
0
__

K([x[)v dx
_
=
_

K([x[)v dx.
_
Since K(r) = c
1
+
c
2
r
is integrable at x = 0.
_
On B

## (0), K([x[) = K(). Thus,

5

_
B(0)
K([x[)
v
n
dS

K()

_
B(0)

v
n

dS

c
1
+
c
2

4
2
max

0, as 0.
_
B(0)
v(x)
K([x[)
n
dS =
_
B(0)
c
2

2
v(x) dS
=
_
B(0)
c
2

2
v(0) dS +
_
B(0)
c
2

2
[v(x) v(0)] dS
=
c
2

2
v(0) 4
2
+ 4c
2
max
xB(0)

v(x) v(0)

. .
0, (v is continuous)
= 4c
2
v(0) v(0).
Thus, taking 4c
2
= 1, i.e. c
2
=
1
4
, we obtain
_

K([x[)v dx = lim
0
_

K([x[)v dx = v(0),
that is K(r) =
1
4r
is the fundamental solution of .
5
In R
3
, for |x| = ,
K(|x|) = K() = c1 +
c2

.
K(|x|)
n
=
K()
r
=
c2

2
, (since n points inwards.)
n points toward 0 on the sphere |x| = (i.e., n = x/|x|).
Partial Dierential Equations Igor Yanovsky, 2005 36
Show that the Fundamental Solution of the Laplace Operator is given by.
K(x) =
_
1
2
log r if n = 2
1
(2n)n
r
2n
if n 3.
(8.7)
Proof. For v C

0
(R
n
), we want to show
_
R
n
K([x[) v(x) dx = v(0).
Suppose v(x) 0 for [x[ R and let = B
R
(0); for small > 0 let

= B

(0).
K([x[) is harmonic (K([x[) = 0) in

## . Consider Greens identity (

=
B

(0)):
_

K([x[)v dx =
_

_
K([x[)
v
n
v
K([x[)
n
_
dS
. .
=0, since v0 for xR
+
_
B(0)
_
K([x[)
v
n
v
K([x[)
n
_
dS.
lim
0
__

K([x[)v dx
_
=
_

K([x[)v dx.
_
Since K(r) is integrable at x = 0.
_
On B

## (0), K([x[) = K(). Thus,

6

_
B(0)
K([x[)
v
n
dS

K()

_
B(0)

v
n

dS

K()

n1
max

0, as 0.
_
B(0)
v(x)
K([x[)
n
dS =
_
B(0)

n1
v(x) dS
=
_
B(0)

n1
v(0) dS +
_
B(0)

n1
[v(x) v(0)] dS
=
1

n1
v(0)
n

n1
max
xB(0)

v(x) v(0)

. .
0, (v is continuous)
= v(0).
Thus,
_

K([x[)v dx = lim
0
_

K([x[)v dx = v(0).
6
Note that for |x| = ,
K(|x|) = K() =
_
1
2
log if n = 2
1
(2n)n

2n
if n 3.
K(|x|)
n
=
K()
r
=
_
1
2
if n = 2
1
n
n1
if n 3,
=
1
n
n1
, (since n points inwards.)
n points toward 0 on the sphere |x| = (i.e., n = x/|x|).
Partial Dierential Equations Igor Yanovsky, 2005 37
8.9 Representation Theorem
Representation Theorem, n = 3.
Let be bounded domain in R
3
and let n be the unit exterior normal to . Let
u C
2
(). Then the value of u at any point x is given by the formula
u(x) =
1
4
_

_
1
[x y[
u(y)
n
u(y)

n
1
[x y[
_
dS
1
4
_

u(y)
[x y[
dy. (8.8)
Proof. Consider the Greens identity:
_

(uw wu) dy =
_

_
u
w
n
w
u
n
_
dS,
where w is the harmonic function
w(y) =
1
[x y[
,
which is singular at x . In order to be able to apply Greens identity, we consider
a new domain

= B

(x).
Since u, w C
2
(

## ), Greens identity can be applied. Since w is harmonic (w = 0)

in

and since

= B

(x), we have

u(y)
[x y[
dy =
_

_
u(y)

n
1
[x y[

1
[x y[
u(y)
n
_
dS (8.9)
+
_
B(x)
_
u(y)

n
1
[x y[

1
[x y[
u(y)
n
_
dS. (8.10)
We will show that formula (8.8) is obtained by letting 0.
lim
0
_

u(y)
[x y[
dy
_
=
_

u(y)
[x y[
dy.
_
Since
1
[x y[
is integrable at x = y.
_
The rst integral on the right of (8.10) does not depend on . Hence, the limit as 0
of the second integral on the right of (8.10) exists, and in order to obtain (8.8), need
lim
0
_
B(x)
_
u(y)

n
1
[x y[

1
[x y[
u(y)
n
_
dS = 4u(x).
_
B(x)
_
u(y)

n
1
[x y[

1
[x y[
u(y)
n
_
dS =
_
B(x)
_
1

2
u(y)
1

u(y)
n
_
dS
=
_
B(x)
1

2
u(x) dS +
_
B(x)
_
1

2
[u(y) u(x)]
1

u(y)
n
_
dS
= 4u(x) +
_
B(x)
_
1

2
[u(y) u(x)]
1

u(y)
n
_
dS.
Partial Dierential Equations Igor Yanovsky, 2005 38
7
The last integral tends to 0 as 0:

_
B(x)
_
1

2
[u(y) u(x)]
1

u(y)
n
_
dS

2
_
B(x)

u(y) u(x)

+
1

_
B(x)

u(y)
n

dS
4 max
yB(x)

u(y) u(x)

. .
0, (u continuous in)
+4 max
y

u(y)

. .
0, (|u| is finite)
.
7
Note that for points y on B(x),
1
|x y|
=
1

and

n
1
|x y|
=
1

2
.
Partial Dierential Equations Igor Yanovsky, 2005 39
Representation Theorem, n = 2.
Let be bounded domain in R
2
and let n be the unit exterior normal to . Let
u C
2
(). Then the value of u at any point x is given by the formula
u(x) =
1
2
_

u(y) log[x y[ dy +
1
2
_

_
u(y)

n
log [x y[ log [x y[
u(y)
n
_
dS. (8.11)
Proof. Consider the Greens identity:
_

(uw wu) dy =
_

_
u
w
n
w
u
n
_
dS,
where w is the harmonic function
w(y) = log [x y[,
which is singular at x . In order to be able to apply Greens identity, we consider
a new domain

= B

(x).
Since u, w C
2
(

in

and since

= B

(x), we have

## u(y) log [x y[ dy (8.12)

=
_

_
u(y)

n
log [x y[ log [x y[
u(y)
n
_
dS
+
_
B(x)
_
u(y)

n
log [x y[ log [x y[
u(y)
n
_
dS.
We will show that formula (8.11) is obtained by letting 0.
lim
0
_

u(y) log[x y[ dy
_
=
_

## u(y) log [x y[ dy.

_
since log [x y[ is integrable at x = y.
_
The rst integral on the right of (8.12) does not depend on . Hence, the limit as 0
of the second integral on the right of (8.12) exists, and in order to obtain (8.11), need
lim
0
_
B(x)
_
u(y)

n
log [x y[ log [x y[
u(y)
n
_
dS = 2u(x).
_
B(x)
_
u(y)

n
log [x y[ log [x y[
u(y)
n
_
dS =
_
B(x)
_
1

u(y) log
u(y)
n
_
dS
=
_
B(x)
1

u(x) dS +
_
B(x)
_
1

u(y)
n
_
dS
= 2u(x) +
_
B(x)
_
1

## [u(y) u(x)] log

u(y)
n
_
dS.
Partial Dierential Equations Igor Yanovsky, 2005 40
8
The last integral tends to 0 as 0:

_
B(x)
_
1

## [u(y) u(x)] log

u(y)
n
_
dS

_
B(x)

u(y) u(x)

+ log
_
B(x)

u(y)
n

dS
2 max
yB(x)

u(y) u(x)

. .
0, (u continuous in )
+2 log max
y

u(y)

. .
0, (|u| is finite)
.
8
Note that for points y on B(x),
log |x y| = log and

n
log |x y| =
1

.
Partial Dierential Equations Igor Yanovsky, 2005 41
Representation Theorems, n > 3 can be obtained in the same way. We use the
Greens identity with
w(y) =
1
[x y[
n2
,
which is a harmonic function in R
n
with a singularity at x.
The fundamental solution for the Laplace operator is (r = [x[):
K(x) =
_
1
2
log r if n = 2
1
(2n)n
r
2n
if n 3.
Representation Theorem. If R
n
is bounded, u C
2
(), and x , then
u(x) =
_

K(x y)u(y) dy +
_

_
u(y)
K(x y)
n
K(x y)
u(y)
n
_
dS. (8.13)
Proof. Consider the Greens identity:
_

(uw wu) dy =
_

_
u
w
n
w
u
n
_
dS,
where w is the harmonic function
w(y) = K(x y),
which is singular at y = x. In order to be able to apply Greens identity, we consider a
new domain

= B

(x).
Since u, K(x y) C
2
(

## ), Greens identity can be applied. Since K(x y) is

harmonic (K(x y) = 0) in

and since

= B

(x), we have

K(x y)u(y) dy =
_

_
u(y)
K(x y)
n
K(x y)
u(y)
n
_
dS (8.14)
+
_
B(x)
_
u(y)
K(x y)
n
K(x y)
u(y)
n
_
dS. (8.15)
We will show that formula (8.13) is obtained by letting 0.
lim
0
_

K(x y)u(y) dy
_
=
_

## K(x y)u(y) dy.

_
since K(x y) is integrable at x = y.
_
The rst integral on the right of (8.15) does not depend on . Hence, the limit as 0
of the second integral on the right of (8.15) exists, and in order to obtain (8.13), need
lim
0
_
B(x)
_
u(y)
K(x y)
n
K(x y)
u(y)
n
_
dS = u(x).
Partial Dierential Equations Igor Yanovsky, 2005 42
_
B(x)
_
u(y)
K(x y)
n
K(x y)
u(y)
n
_
dS =
_
B(x)
_
u(y)
K()
n
K()
u(y)
n
_
dS
=
_
B(x)
u(x)
K()
n
dS +
_
B(x)
_
K()
n
[u(y) u(x)] K()
u(y)
n
_
dS
=
1

n1
_
B(x)
u(x) dS
1

n1
_
B(x)
[u(y) u(x)] dS
_
B(x)
K()
u(y)
n
dS
=
1

n1
u(x)
n

n1
. .
u(x)

n1
_
B(x)
[u(y) u(x)] dS
_
B(x)
K()
u(y)
n
dS.
9
The last two integrals tend to 0 as 0:

n1
_
B(x)
[u(y) u(x)] dS
_
B(x)
K()
u(y)
n
dS

n1
max
yB(x)

u(y) u(x)

n1
. .
0, (u continuous in)
+

K()

max
y

u(y)

n1
. .
0, (|u| is finite)
.
8.10 Greens Function and the Poisson Kernel
With a slight change in notation, the Representation Theorem has the following special
case.
Theorem. If R
n
is bounded, u C
2
()

C
1
() is harmonic, and , then
u() =
_

_
u(x)
K(x )
n
K(x )
u(x)
n
_
dS. (8.16)
Let (x) be any harmonic function in , and for x, consider
G(x, ) = K(x ) + (x).
If we use the Greens identity (with u = 0 and = 0), we get:
0 =
_

_
u

n

u
n
_
ds. (8.17)
Adding (8.16) and (8.17), we obtain:
u() =
_

_
u(x)
G(x, )
n
G(x, )
u(x)
n
_
dS. (8.18)
Suppose that for each we can nd a function

satises

## (x) = K(x ) for all x . Then G(x, ) = K(x ) +

(x) is a
fundamental solution such that
G(x, ) = 0 x .
9
Note that for points y on B(x),
K(x y) = K() =
_
1
2
log if n = 2
1
(2n)n

2n
if n 3.
K(x y)
n
=
K()
r
=
_
1
2
if n = 2
1
n
n1
if n 3,
=
1
n
n1
, (since n points inwards.)
Partial Dierential Equations Igor Yanovsky, 2005 43
Gis called the Greens function and is useful in satisfying Dirichlet boundary conditions.
The Greens function is dicult to construct for a general domain since it requires
solving the Dirichlet problem

= 0 in ,

## (x) = K(x ) for x , for each

.
From (8.18) we nd
10
u() =
_

u(x)
G(x, )
n
dS.
Thus if we know that the Dirichlet problem has a solution u C
2
(), then we can
calculate u from the Poisson integral formula (provided of course that we can compute
G(x, )).
10
If we did not assume u = 0 in our derivation, we would have (8.13) instead of (8.16), and an
extra term in (8.17), which would give us a more general expression:
u() =
_

G(x, ) u dx +
_

u(x)
G(x, )
n
dS.
Partial Dierential Equations Igor Yanovsky, 2005 44
8.11 Properties of Harmonic Functions
Liouvilles Theorem. A bounded harmonic function dened on all of R
n
must be a
constant.
8.12 Eigenvalues of the Laplacian
Consider the equation
_
u +u = 0 in
u = 0 on ,
(8.19)
where is a bounded domain and is a (complex) number. The values of for which
(8.19) admits a nontrivial solution u are called the eigenvalues of in , and the
solution u is an eigenfunction associated to the eigenvalue . (The convention
u +u = 0 is chosen so that all eigenvalues will be positive.)
Properties of the Eigenvalues and Eigenfunctions for (8.19):
1. The eigenvalues of (8.19) form a countable set
n

n=1
of positive numbers with

n
as n .
2. For each eigenvalue
n
there is a nite number (called the multiplicity of
n
) of
linearly independent eigenfunctions u
n
.
3. The rst (or principal) eigenvalue,
1
, is simple and u
1
4. Eigenfunctions corresponding to distinct eigenvalues are orthogonal.
5. The eigenfunctions may be used to expand certain functions on in an innite
series.
Partial Dierential Equations Igor Yanovsky, 2005 45
9 Heat Equation
The heat equation is
u
t
= ku for x , t > 0, (9.1)
with initial and boundary conditions.
9.1 The Pure Initial Value Problem
9.1.1 Fourier Transform
If u C

0
(R
n
), dene its Fourier transform u by
u() =
1
(2)
n
2
_
R
n
e
ix
u(x) dx for R
n
.
We can dierentiate u:

j
u() =
1
(2)
n
2
_
R
n
e
ix
(ix
j
)u(x) dx =
_

(ix
j
) u

().
Iterating this computation, we obtain
_

j
_
k
u() =
_

(ix
j
)
k
u

(). (9.2)
Similarly, integrating by parts shows
_

u
x
j
_
() =
1
(2)
n
2
_
R
n
e
ix
u
x
j
(x) dx =
1
(2)
n
2
_
R
n

x
j
(e
ix
)u(x) dx
=
1
(2)
n
2
_
R
n
(i
j
)e
ix
u(x) dx
= (i
j
) u().
Iterating this computation, we obtain
_

k
u
x
k
j
_
() = (i
j
)
k
u(). (9.3)
Formulas (9.2) and (9.3) express the fact that Fourier transform interchanges dieren-
tiation and multiplication by the coordinate function.
9.1.2 Multi-Index Notation
A multi-index is a vector = (
1
, . . . ,
n
) where each
i
is a nonnegative integer.
The order of the multi-index is [[ =
1
+. . . +
n
. Given a multi-index , dene
D

u =

||
u
x

1
1
x
n
n
=

1
x
1

n
xn
u.
We can generalize (9.3) in multi-index notation:

u() =
1
(2)
n
2
_
R
n
e
ix
D

u(x) dx =
(1)
||
(2)
n
2
_
R
n
D

x
(e
ix
)u(x) dx
=
1
(2)
n
2
_
R
n
(i)

e
ix
u(x) dx
= (i)

u().
(i)

= (i
1
)

1
(i
n
)
n
.
Partial Dierential Equations Igor Yanovsky, 2005 46
Parsevals theorem (Plancherels theorem).
Assume u L
1
(R
n
) L
2
(R
n
). Then u, u

L
2
(R
n
) and
[[ u[[
L
2
(R
n
)
= [[u

[[
L
2
(R
n
)
= [[u[[
L
2
(R
n
)
, or
_

[u(x)[
2
dx =
_

[ u()[
2
d.
Also,
_

u(x) v(x) dx =
_

u() v() d.
The properties (9.2) and (9.3) make it very natural to consider the fourier transform
on a subspace of L
1
(R
n
) called the Schwartz class of functions, S, which consists of the
smooth functions whose derivatives of all orders decay faster than any polynomial, i.e.
S = u C

(R
n
) : for every k N and N
n
, [x[
k
[D

u(x)[ is bounded on R
n
.
For u S, the Fourier transform u exists since u decays rapidly at .
Lemma. (i) If u L
1
(R
n
), then u is bounded. (ii) If u S, then u S.
Dene the inverse Fourier transform for u L
1
(R
n
):
u

() =
1
(2)
n
2
_
R
n
e
ix
u(x) dx for R
n
, or
u(x) =
1
(2)
n
2
_
R
n
e
ix
u() d for x R
n
.
Fourier Inversion Theorem (McOwen). If u S, then ( u)

= u; that is,
u(x) =
1
(2)
n
2
_
R
n
e
ix
u() d =
1
(2)
n
_ _
R
2n
e
i(xy)
u(y) dy d = ( u)

(x).
Fourier Inversion Theorem (Evans). Assume u L
2
(R
n
). Then, u = ( u)

.
Partial Dierential Equations Igor Yanovsky, 2005 47
Shift: Let u(x a
. .
y
) = v(x), and determinte v():

## u(x a)() = v() =

1

2
_
R
e
ix
v(x) dx =
1

2
_
R
e
i(y+a)
u(y) dy
=
1

2
_
R
e
iy
e
ia
u(y) dy = e
ia
u().

u(x a)() = e
ia
u().
Delta function:

(x)() =
1

2
_
R
e
ix
(x) dx =
1

2
,
_
since u(x) =
_
R
(x y) u(y) dy
_
.

(x a)() = e
ia

() =
1

2
e
ia
. (using result from Shift)
Convolution:
(f g)(x) =
_
R
n
f(x y)g(y) dy,

(f g)() =
1
(2)
n
2
_
R
n
e
ix
_
R
n
f(x y) g(y) dy dx =
1
(2)
n
2
_
R
n
_
R
n
e
ix
f(x y) g(y) dy dx
=
1
(2)
n
2
_
R
n
_
R
n
_
e
i(xy)
f(x y) dx
__
e
iy
g(y) dy
_
=
1
(2)
n
2
_
R
n
e
iz
f(z) dz
_
R
n
e
iy
g(y) dy = (2)
n
2

f()g().

(f g)() = (2)
n
2

f() g().
Gaussian: (completing the square)
_
e

x
2
2
_
() =
1

2
_
R
e
ix
e

x
2
2
dx =
1

2
_
R
e

x
2
+2ix
2
dx =
1

2
_
R
e

x
2
+2ix
2
2
dx e

2
2
=
1

2
_
R
e

(x+i)
2
2
dx e

2
2
=
1

2
_
R
e
y
2
2
dy e

2
2
=
1

2e

2
2
= e

2
2
.
_
e

x
2
2
_
() = e

2
2
.
Multiplication by x:

ixu() =
1

2
_
R
e
ix
_
ixu(x)
_
dx =
d
d
u().

xu(x)() = i
d
d
u().
Partial Dierential Equations Igor Yanovsky, 2005 48
Multiplication of u
x
by x: (using the above result)

xu
x
(x)() =
1

2
_
R
e
ix
_
xu
x
(x)
_
dx =
1

2
_
e
ix
xu
_

. .
=0

2
_
R
_
(i)e
ix
x +e
ix
_
u dx
=
1

2
i
_
R
e
ix
x u dx
1

2
_
R
e
ix
u dx
= i

xu(x)() u() = i
_
i
d
d
u()
_
u() =
d
d
u() u().

xu
x
(x)() =
d
d
u() u().
Table of Fourier Transforms:
11
_
e

ax
2
2
_
() =
1

a
e

2
2a
, (Gaussian)

e
ibx
f(ax)() =
1
a

f
_
b
a
_
,
f(x) =
_
1, [x[ L
0, [x[ > L,

f(x)() =
1

2
2 sin(L)

e
a|x|
() =
1

2
2a
a
2
+
2
, (a > 0)

1
a
2
+x
2
() =

2
2a
e
a||
, (a > 0)

H(a [x[)() =
_
2

sina,

H(x)() =
1

2
_
() +
1
i
_
,

_
H(x) H(x)
_
() =
_
2

1
i
, (sign)

1() =

2().
11
Results with marked with were taken from W. Strauss, where the denition of Fourier Transform
is dierent. An extra multiple of
1

2
was added to each of these results.
Partial Dierential Equations Igor Yanovsky, 2005 49
9.1.3 Solution of the Pure Initial Value Problem
Consider the pure initial value problem
_
u
t
= u for t > 0, x R
n
u(x, 0) = g(x) for x R
n
.
(9.4)
We take the Fourier transform of the heat equation in the x-variables.

(u
t
)(, t) =
1
(2)
n
2
_
R
n
e
ix
u
t
(x, t) dx =

t
u(, t)

u(, t) =
n

j=1
(i
j
)
2
u(, t) = [[
2
u(, t).
The heat equation therefore becomes

t
u(, t) = [[
2
u(, t),
which is an ordinary dierential equation in t, with the solution u(, t) = Ce
||
2
t
.
The initial condition u(, 0) = g() gives
u(, t) = g() e
||
2
t
,
u(x, t) =
_
g() e
||
2
t
_

=
1
(2)
n
2
_
g
_
e
||
2
t
_

_
=
1
(2)
n
2
g
_
1
(2)
n
2
_
R
n
e
||
2
t
e
ix
d
_
=
1
(4
2
)
n
2
g
_
_
R
n
e
ix||
2
t
d
_
=
1
(4
2
)
n
2
g
_
e

|x|
2
4t
_

t
_n
2
_
=
1
(4t)
n
2
g
_
e

|x|
2
4t
_
=
1
(4t)
n
2
_
R
n
e

|xy|
2
4t
g(y) dy.
Thus,
12
solution of the initial value problem (9.4) is
u(x, t) =
_
R
n
K(x, y, t) g(y) dy =
1
(4t)
n
2
_
R
n
e

|xy|
2
4t
g(y) dy.
Uniqueness of solutions for the pure initial value problem fails: there are nontrivial
solutions of (9.4) with g = 0.
13
Thus, the pure initial value problem for the heat
equation is not well-posed, as it was for the wave equation. However, the nontrivial
solutions are unbounded as functions of x when t > 0 is xed; uniqueness can be
regained by adding a boundedness condition on the solution.
12
Identity (Evans, p. 187.) :
_
R
n
e
ix||
2
t
d = e

|x|
2
4t
_

t
_n
2
.
13
The following function u satises ut = uxx for t > 0 with u(x, 0) = 0:
u(x, t) =

k=0
1
(2k)!
x
2k
d
k
dt
k
e
1/t
2
.
Partial Dierential Equations Igor Yanovsky, 2005 50
9.1.4 Nonhomogeneous Equation
Consider the pure initial value problem with homogeneous initial condition:
_
u
t
= u + f(x, t) for t > 0, x R
n
u(x, 0) = 0 for x R
n
.
(9.5)
Duhamels principle gives the solution:
u(x, t) =
_
t
0
_
R
n

## K(x y, t s) f(y, s) dy ds.

9.1.5 Nonhomogeneous Equation with Nonhomogeneous Initial Conditions
Combining two solutions above, we nd that the solution of the initial value problem
_
u
t
= u + f(x, t) for t > 0, x R
n
u(x, 0) = g(x) for x R
n
.
(9.6)
is given by
u(x, t) =
_
R
n

K(x y, t) g(y) dy +
_
t
0
_
R
n

## K(x y, t s) f(y, s) dy ds.

9.1.6 The Fundamental Solution
Suppose we want to solve the Cauchy problem
_
u
t
= Lu x R
n
, t > 0
u(x, 0) = g(x) x R
n
.
(9.7)
where L is a dierential operator in R
n
with constant coecients. Suppose K(x, t) is
a distribution in R
n
for each value of t 0, K is C
1
in t and satises
_
K
t
LK = 0,
K(x, 0) = (x).
(9.8)
We call K a fundamental solution for the initial value problem. The solution of
(9.7) is then given by convolution in the space variables:
u(x, t) =
_
R
n
K(x y, t) g(y) dy.
Partial Dierential Equations Igor Yanovsky, 2005 51
For operators of the form
t
L, the fundamental solution of the initial value problem,
K(x, t) as dened in (9.8), coincides with the free space fundamental solution, which
satises
_

t
L
_
K(x, t) = (x, t),
provided we extend K(x, t) by zero to t < 0. For the heat equation, consider

K(x, t) =

1
(4t)
n/2
e

|x|
2
4t
t > 0
0 t 0.
(9.9)
Notice that

K is smooth for (x, t) ,= (0, 0).

## K dened as in (9.9), is the fundamental solution of the free space heat

equation.
Proof. We need to show:
_

_
K(x, t) = (x, t). (9.10)
To verify (9.10) as distributions, we must show that for any v C

0
(R
n+1
):
14
_
R
n+1

K(x, t)
_

t

_
v dx dt =
_
R
n+1
(x, t) v(x, t) dxdt v(0, 0).
To do this, let us take > 0 and dene

(x, t) =

1
(4t)
n/2
e

|x|
2
4t
t >
0 t .
Then

K

K as distributions, so it suces to show that (
t
)

K

as distribu-
tions. Now
_

K

_

t

_
v dx dt =
_

__
R
n

K(x, t)
_

t

_
v(x, t) dx
_
dt
=
_

__
R
n

K(x, t)
t
v(x, t) dx
_
dt
_

__
R
n

K(x, t) v(x, t) dx
_
dt
=
_ _
R
n

K(x, t) v(x, t) dx
_
t=
t=
+
_

__
R
n

t

K(x, t) v(x, t) dx
_
dt
_

__
R
n

K(x, t) v(x, t) dx
_
dt
=
_

__
R
n
_

K(x, t) v(x, t) dx
_
dt +
_
R
n

## K(x, ) v(x, ) dx.

But for t > , (
t
)

K(x, t) = 0; moreover, since lim
t0
+

K(x, t) =
0
(x) = (x),
we have

K(x, )
0
(x) as 0, so the last integral tends to v(0, 0).
14
Note, for the operator L = /t, the adjoint operator is L

= /t.
Partial Dierential Equations Igor Yanovsky, 2005 52
10 Schr odinger Equation
Problem (F96, #5). The Gauss kernel
G(t, x, y) =
1
(4t)
1
2
e

(xy)
2
4t
is the fundamental solution of the heat equation, solving
G
t
= G
xx
, G(0, x, y) = (x y).
By analogy with the heat equation, nd the fundamental solution H(t, x, y) of the
Schrodinger equation
H
t
= iH
xx
, H(0, x, y) = (x y).
Show that your expression H(x) is indeed the fundamental solution for the
Schr odinger equation. You may use the following special integral
_

e
ix
2
4
dx =

i4.
Proof. Remark: Consider the initial value problem for the Schr odinger equation
_
u
t
= iu x R
n
, t > 0,
u(x, 0) = g(x) x R
n
.
If we formally replace t by it in the heat kernel, we obtain the Fundamental
Solution of the Schrodinger Equation:
15
H(x, t) =
1
(4it)
n
2
e

|x|
2
4it
(x R
n
, t ,= 0)
u(x, t) =
1
(4it)
n
2
_
R
n
e

|xy|
2
4it
g(y) dy.
In particular, the Schr odinger equation is reversible in time, whereas the heat equation
is not.
Solution: We have already found the fundamental solution for the heat equation
using the Fourier transform. For the Schr odinger equation is one dimension, we have

t
u(, t) = i
2
u(, t),
which is an ordinary dierential equation in t, with the solution u(, t) = Ce
i
2
t
.
The initial condition u(, 0) = g() gives
u(, t) = g() e
i
2
t
,
u(x, t) =
_
g() e
i
2
t
_

=
1

2
_
g
_
e
i
2
t
_

_
=
1

2
g
_
1

2
_
R
e
i
2
t
e
ix
d
_
=
1
2
g
_
_
R
e
ixi
2
t
d
_
= (need some work) =
=
1

4it
g
_
e

|x|
2
4it
_
=
1

4it
_
R
e

|xy|
2
4it
g(y) dy.
15
Evans, p. 188, Example 3.
Partial Dierential Equations Igor Yanovsky, 2005 53
For the Schrodinger equation, consider

(x, t) =

1
(4it)
n/2
e

|x|
2
4it
t > 0
0 t 0.
(10.1)
Notice that

is smooth for (x, t) ,= (0, 0).

## dened as in (10.1), is the fundamental solution of the Schrodinger equa-

tion. We need to show:
_

t
i
_
(x, t) = (x, t). (10.2)
To verify (10.2) as distributions, we must show that for any v C

0
(R
n+1
):
16
_
R
n+1

(x, t)
_

t
i
_
v dx dt =
_
R
n+1
(x, t) v(x, t) dx dt v(0, 0).
To do this, let us take > 0 and dene

(x, t) =

1
(4it)
n/2
e

|x|
2
4it
t >
0 t .
Then

as distributions, so it suces to show that (
t
i)

as distribu-
tions. Now
_

t
i
_
v dx dt =
_

__
R
n

(x, t)
_

t
i
_
v(x, t) dx
_
dt
=
_

__
R
n
_

t
i
_

(x, t) v(x, t) dx
_
dt +
_
R
n

But for t > , (
t
i)

## (x, t) = 0; moreover, since lim

t0
+

(x, t) =
0
(x) = (x),
we have

(x, )
0
(x) as 0, so the last integral tends to v(0, 0).
16
Note, for the operator L = /t, the adjoint operator is L

= /t.
Partial Dierential Equations Igor Yanovsky, 2005 54
11 Problems: Quasilinear Equations
Problem (F90, #7). Use the method of characteristics to nd the solution of the
rst order partial dierential equation
x
2
u
x
+ xyu
y
= u
2
which passes through the curve u = 1, x = y
2
. Determine where this solution becomes
singular.
Proof. We have a condition u(x = y
2
) = 1. is parametrized by : (s
2
, s, 1).
dx
dt
= x
2
x =
1
t c
1
(s)
x(0, s) =
1
c
1
(s)
= s
2
x =
1
t +
1
s
2
=
s
2
1 ts
2
,
dy
dt
= xy
dy
dt
=
s
2
y
1 ts
2
y =
c
2
(s)
1 ts
2
y(s, 0) = c
2
(s) = s y =
s
1 ts
2
,
dz
dt
= z
2
z =
1
t c
3
(s)
z(0, s) =
1
c
3
(s)
= 1 z =
1
1 t
.
Thus,
x
y
= s y =
x
y
1 t
x
2
y
2
t =
y
2
x
2

1
x
.
u(x, y) =
1
1
y
2
x
2
+
1
x
=
x
2
x
2
+x y
2
.
The solution becomes singular when y
2
= x
2
+x.
It can be checked that the solution satises the PDE and u(x = y
2
) =
y
4
y
4
+y
2
y
2
= 1.
Problem (S91, #7). Solve the rst order PDE
f
x
+x
2
yf
y
+ f = 0
f(x = 0, y) = y
2
using the method of characteristics.
Proof. Rewrite the equation
u
x
+x
2
yu
y
= u,
u(0, y) = y
2
.
is parameterized by : (0, s, s
2
).
dx
dt
= 1 x = t,
dy
dt
= x
2
y
dy
dt
= t
2
y y = se
t
3
3
,
dz
dt
= z z = s
2
e
t
.
Thus, x = t and s = ye

t
3
3
= ye

x
3
3
, and
u(x, y) = (ye

x
3
3
)
2
e
x
= y
2
e

2
3
x
3
x
.
The solution satises both the PDE and initial conditions.
Partial Dierential Equations Igor Yanovsky, 2005 55
Problem (S92, #1). Consider the Cauchy problem
u
t
= xu
x
u + 1 < x < , t 0
u(x, 0) = sinx < x <
and solve it by the method of characteristics. Discuss the properties of the solution; in
particular investigate the behavior of [u
x
(, t)[

for t .
Proof. is parametrized by : (s, 0, sins). We have
dx
dt
= x x = se
t
,
dy
dt
= 1 y = t,
dz
dt
= 1 z z = 1
1 sins
e
t
.
Thus, t = y, s = xe
y
, and
u(x, y) = 1
1
e
y
+
sin(xe
y
)
e
y
.
It can be checked that the solution satises the PDE and the initial condition.
As t , u(x, t) 1. Also,
[u
x
(x, y)[

= [ cos(xe
y
)[

= 1.
u
x
(x, y) oscillate between 1 and 1. If x = 0, u
x
= 1.
Problem (W02, #6). Solve the Cauchy problem
u
t
+u
2
u
x
= 0, t > 0,
u(0, x) = 2 + x.
Proof. Solved
Partial Dierential Equations Igor Yanovsky, 2005 56
Problem (S97, #1). Find the solution of the Burgers equation
u
t
+uu
x
= x, t 0
u(x, 0) = f(x), < x < .
Proof. is parameterized by : (s, 0, f(s)).
dx
dt
= z,
dy
dt
= 1 y = t,
dz
dt
= x.
Note that we have a coupled system:
_
x = z,
z = x,
which can be written as a second order ODE:
x +x = 0, x(s, 0) = s, x(s, 0) = z(0) = f(s).
Solving the equation, we get
x(s, t) = s cos t + f(s) sint, and thus,
z(s, t) = x(t) = s sint +f(s) cos t.
_
x = s cos y + f(s) siny,
u = s sin y +f(s) cos y.

_
x cos y = s cos
2
y +f(s) siny cos y,
u siny = s sin
2
y +f(s) cos y siny.
x cos y u siny = s(cos
2
y + sin
2
y) = s.
u(x, y) = f(x cos y u siny) cos y (x cos y u siny) siny.
Problem (F98, #2). Solve the partial dierential equation
u
y
u
2
u
x
= 3u, u(x, 0) = f(x)
using method of characteristics. (Hint: nd a parametric representation of the solu-
tion.)
Proof. is parameterized by : (s, 0, f(s)).
dx
dt
= z
2

dx
dt
= f
2
(s)e
6t
x =
1
6
f
2
(s)e
6t
+
1
6
f
2
(s) + s,
dy
dt
= 1 y = t,
dz
dt
= 3z z = f(s)e
3t
.
Partial Dierential Equations Igor Yanovsky, 2005 57
Thus,
_
x =
1
6
f
2
(s)e
6y
+
1
6
f
2
(s) + s,
f(s) =
z
e
3y
x =
1
6
z
2
e
6y
e
6y
+
1
6
z
2
e
6y
+s =
z
2
6e
6y

z
2
6
+s,
s = x
z
2
6e
6y
+
z
2
6
.
z = f
_
x
z
2
6e
6y
+
z
2
6
_
e
3y
.
u(x, y) = f
_
x
u
2
6e
6y
+
u
2
6
_
e
3y
.
Partial Dierential Equations Igor Yanovsky, 2005 58
Problem (S99, #1) Modied Problem. a) Solve
u
t
+
_
u
3
3
_
x
= 0 (11.1)
for t > 0, < x < with initial data
u(x, 0) = h(x) =
_
a(1 e
x
), x < 0
a(1 e
x
), x > 0
where a > 0 is constant. Solve until the rst appearance of discontinuous derivative
and determine that critical time.
b) Consider the equation
u
t
+
_
u
3
3
_
x
= cu. (11.2)
How large does the constant c > 0 has to be, so that a smooth solution (with no discon-
tinuities) exists for all t > 0? Explain.
Proof. a) Characteristic form: u
t
+u
2
u
x
= 0. : (s, 0, h(s)).
dx
dt
= z
2
,
dy
dt
= 1,
dz
dt
= 0.
x = h(s)
2
t + s, y = t, z = h(s).
u(x, y) = h(x u
2
y) (11.3)
The characteristic projection in the xt-plane
17
passing through the point (s, 0) is the
line
x = h(s)
2
t + s
along which u has the constant value u = h(s).
The derivative of the initial data is discontinuous, and that leads to a
rarefaction-like behavior at t = 0. However, if the question meant to ask to
determine the rst time when a shock forms, we proceed as follows.
Two characteristics x = h(s
1
)
2
t + s
1
and x = h(s
2
)
2
t + s
2
intersect at a point (x, t)
with
t =
s
2
s
1
h(s
2
)
2
h(s
1
)
2
.
From (11.3), we have
u
x
= h

(s)(1 2uu
x
t) u
x
=
h

(s)
1 + 2h(s)h

(s)t
Hence for 2h(s)h

(s) < 0, u
x
becomes innite at the positive time
t =
1
2h(s)h

(s)
.
The smallest t for which this happens corresponds to the value s = s
0
at which h(s)h

(s)
has a minimum (i.e.h(s)h

## (s) has a maximum). At time T = 1/(2h(s

0
)h

(s
0
)) the
17
y and t are interchanged here
Partial Dierential Equations Igor Yanovsky, 2005 59
solution u experiences a gradient catastrophe.
Therefore, need to nd a minimum of
f(x) = 2h(x)h

(x) =
_
2a(1 e
x
) ae
x
2a(1 e
x
) (ae
x
)
=
_
2a
2
e
x
(1 e
x
), x < 0
2a
2
e
x
(1 e
x
), x > 0
f

(x) =
_
2a
2
e
x
(1 2e
x
), x < 0
2a
2
e
x
(1 2e
x
), x > 0
= 0
_
x = ln(
1
2
) = ln(2), x < 0
x = ln(2), x > 0

_
f(ln(
1
2
)) = 2a
2
e
ln(
1
2
)
(1 e
ln(
1
2
)
) = 2a
2
(
1
2
)(
1
2
) =
a
2
2
, x < 0
f(ln(2)) = 2a
2
(
1
2
)(1
1
2
) =
a
2
2
, x > 0
t =
1
min2h(s)h

(s)
=
2
a
2
Proof. b) Characteristic form: u
t
+ u
2
u
x
= cu. : (s, 0, h(s)).
dx
dt
= z
2
= h(s)
2
e
2ct
x = s +
1
2c
h(s)
2
(1 e
2ct
),
dy
dt
= 1 y = t,
dz
dt
= cz z = h(s)e
ct
( h(s) = ue
cy
).
Solving for s and t in terms of x and y, we get:
t = y, s = x
1
2c
h(s)
2
(1 e
2cy
).
Thus,
u(x, y) = h
_
x
1
2c
u
2
e
2cy
(1 e
2cy
)
_
e
cy
.
u
x
= h

(s)e
cy
(1
1
c
uu
x
e
2cy
(1 e
2cy
)),
u
x
=
h

(s)e
cy
1 +
1
c
h

(s)e
cy
u (1 e
2cy
)
=
h

(s)e
cy
1 +
1
c
h

(s)h(s)(1 e
2cy
)
.
Thus, c > 0 that would allow a smooth solution to exist for all t > 0 should satisfy
1 +
1
c
h

(s)h(s)(1 e
2cy
) ,= 0.
We can perform further calculations taking into account the result from part (a):
min2h(s)h

(s) =
a
2
2
.
Partial Dierential Equations Igor Yanovsky, 2005 60
Problem (S99, #1). Original Problem. a). Solve
u
t
+
u
3
x
3
= 0 (11.4)
for t > 0, < x < with initial data
u(x, 0) = h(x) =
_
a(1 e
x
), x < 0
a(1 e
x
), x > 0
where a > 0 is constant.
Proof. Rewrite the equation as
F(x, y, u, u
x
, u
y
) =
u
3
x
3
+ u
y
= 0,
F(x, y, z, p, q) =
p
3
3
+q = 0.
is parameterized by : (s, 0, h(s), (s), (s)).
We need to complete to a strip. Find (s) and (s), the initial conditions for p(s, t)
and q(s, t), respectively:
F(f(s), g(s), h(s), (s), (s)) = 0,
(s)
3
3
+(s) = 0,
(s) =
(s)
3
3
.
h

(s) = (s)f

(s) + (s)g

(s)
_
ae
s
= (s), x < 0
ae
s
= (s), x > 0

_
(s) =
a
3
e
3s
3
, x < 0
(s) =
a
3
e
3s
3
, x > 0
Therefore, now is parametrized by
_
: (s, 0, a(1 e
s
), ae
s
,
a
3
e
3s
3
), x < 0
: (s, 0, a(1 e
s
), ae
s
,
a
3
e
3s
3
), x > 0
dx
dt
= F
p
= p
2
=
_
a
2
e
2s
a
2
e
2s
x(s, t) =
_
a
2
e
2s
t + c
4
(s)
a
2
e
2s
t +c
5
(s)
x =
_
a
2
e
2s
t + s
a
2
e
2s
t +s
dy
dt
= F
q
= 1 y(s, t) = t +c
1
(s) y = t
dz
dt
= pF
p
+ qF
q
= p
3
+q =
_
a
3
e
3s

a
3
e
3s
3
=
2
3
a
3
e
3s
, x < 0
a
3
e
3s
+
a
3
e
3s
3
=
2
3
a
3
e
3s
, x > 0
z(s, t) =
_
2
3
a
3
e
3s
t + c
6
(s), x < 0

2
3
a
3
e
3s
t +c
7
(s), x > 0
z =
_
2
3
a
3
e
3s
t a(1 e
s
), x < 0

2
3
a
3
e
3s
t a(1 e
s
), x > 0
dp
dt
= F
x
F
z
p = 0 p(s, t) = c
2
(s) p =
_
ae
s
, x < 0
ae
s
, x > 0
Partial Dierential Equations Igor Yanovsky, 2005 61
dq
dt
= F
y
F
z
q = 0 q(s, t) = c
3
(s) q =
_

a
3
e
3s
3
, x < 0
a
3
e
3s
3
, x > 0
Thus,
u(x, y) =
_
2
3
a
3
e
3s
y a(1 e
s
), x < 0

2
3
a
3
e
3s
y a(1 e
s
), x > 0
where s is dened as
x =
_
a
2
e
2s
y +s, x < 0
a
2
e
2s
y + s, x > 0.
b). Solve the equation
u
t
+
u
3
x
3
= cu. (11.5)
Proof. Rewrite the equation as
F(x, y, u, u
x
, u
y
) =
u
3
x
3
+ u
y
+cu = 0,
F(x, y, z, p, q) =
p
3
3
+q +cz = 0.
is parameterized by : (s, 0, h(s), (s), (s)).
We need to complete to a strip. Find (s) and (s), the initial conditions for p(s, t)
and q(s, t), respectively:
F(f(s), g(s), h(s), (s), (s)) = 0,
(s)
3
3
+(s) + ch(s) = 0,
(s) =
(s)
3
3
ch(s) =
_

(s)
3
3
+ ca(1 e
x
), x < 0

(s)
3
3
+ ca(1 e
x
), x > 0
h

(s) = (s)f

(s) +(s)g

(s)
_
ae
s
= (s), x < 0
ae
s
= (s), x > 0

_
(s) =
a
3
e
3s
3
+ca(1 e
x
), x < 0
(s) =
a
3
e
3s
3
+ca(1 e
x
), x > 0
Therefore, now is parametrized by
_
: (s, 0, a(1 e
s
), ae
s
,
a
3
e
3s
3
+ ca(1 e
x
), x < 0
: (s, 0, a(1 e
s
), ae
s
,
a
3
e
3s
3
+ ca(1 e
x
), x > 0
Partial Dierential Equations Igor Yanovsky, 2005 62
dx
dt
= F
p
= p
2
dy
dt
= F
q
= 1
dz
dt
= pF
p
+qF
q
= p
3
+q
dp
dt
= F
x
F
z
p = cp
dq
dt
= F
y
F
z
q = cq
We can proceed solving the characteristic equations with initial conditions above.
Partial Dierential Equations Igor Yanovsky, 2005 63
Problem (S95, #7). a) Solve the following equation, using characteristics,
u
t
+u
3
u
x
= 0,
u(x, 0) =
_
a(1 e
x
), for x < 0
a(1 e
x
), for x > 0
where a > 0 is a constant. Determine the rst time when a shock forms.
Proof. a) is parameterized by : (s, 0, h(s)).
dx
dt
= z
3
,
dy
dt
= 1,
dz
dt
= 0.
x = h(s)
3
t + s, y = t, z = h(s).
u(x, y) = h(x u
3
y) (11.6)
The characteristic projection in the xt-plane
18
passing through the point (s, 0) is the line
x = h(s)
3
t + s
along which u has a constant value u = h(s).
Characteristics x = h(s
1
)
3
t +s
1
and x = h(s
2
)
3
t +s
2
intersect at a point (x, t) with
t =
s
2
s
1
h(s
2
)
3
h(s
1
)
3
.
From (11.6), we have
u
x
= h

(s)(1 3u
2
u
x
t) u
x
=
h

(s)
1 + 3h(s)
2
h

(s)t
Hence for 3h(s)
2
h

(s) < 0, u
x
becomes innite at the positive time
t =
1
3h(s)
2
h

(s)
.
The smallest t for which this happens corresponds to the value s = s
0
at which
h(s)
2
h

2
h

## (s) has a maximum). At time T = 1/(3h(s

0
)
2
h

(s
0
))
the solution u experiences a gradient catastrophe.
Therefore, need to nd a minimum of
f(x) = 3h(x)
2
h

(x) =
_
3a
2
(1 e
x
)
2
ae
x
= 3a
3
e
x
(1 e
x
)
2
, x < 0
3a
2
(1 e
x
)
2
ae
x
= 3a
3
e
x
(1 e
x
)
2
, x > 0
f

(x) =
_
3a
3
_
e
x
(1 e
x
)
2
e
x
2(1 e
x
)e
x

= 3a
3
e
x
(1 e
x
)(1 3e
x
), x < 0
3a
3
_
e
x
(1 e
x
)
2
+ e
x
2(1 e
x
)e
x

= 3a
3
e
x
(1 e
x
)(1 + 3e
x
), x > 0
= 0
The zeros of f

(x) are
_
x = 0, x = ln 3, x < 0,
x = 0, x = ln 3, x > 0.
We check which ones give the minimum of f(x) :

_
f(0) = 3a
3
, f(ln3) = 3a
3 1
3
(1
1
3
)
2
=
4a
3
9
, x < 0
f(0) = 3a
3
, f(ln3) = 3a
3 1
3
(1
1
3
)
2
=
4a
3
9
, x > 0
18
y and t are interchanged here
Partial Dierential Equations Igor Yanovsky, 2005 64
t =
1
min3h(s)
2
h

(s)
=
1
min f(s)
=
1
3a
3
.
Partial Dierential Equations Igor Yanovsky, 2005 65
b) Now consider
u
t
+u
3
u
x
+cu = 0
with the same initial data and a positive constant c. How large does c need to be in
order to prevent shock formation?
b) Characteristic form: u
t
+ u
3
u
x
= cu. : (s, 0, h(s)).
dx
dt
= z
3
= h(s)
3
e
3ct
x = s +
1
3c
h(s)
3
(1 e
3ct
),
dy
dt
= 1 y = t,
dz
dt
= cz z = h(s)e
ct
( h(s) = ue
cy
).
z(s, t) = h
_
x
1
3c
h(s)
3
(1 e
3ct
)
_
e
ct
,
u(x, y) = h
_
x
1
3c
u
3
e
3cy
(1 e
3cy
)
_
e
cy
.
u
x
= h

(s) e
cy

_
1
1
c
u
2
u
x
e
3cy
(1 e
3cy
)
_
,
u
x
=
h

(s)e
cy
1 +
1
c
h

(s)u
2
e
2cy
(1 e
3cy
)
=
h

(s)e
cy
1 +
1
c
h

(s)h(s)
2
(1 e
3cy
)
.
Thus, we need
1 +
1
c
h

(s)h(s)
2
(1 e
3cy
) ,= 0.
We can perform further calculations taking into account the result from part (a):
min3h(s)
2
h

(s) = 3a
3
.
Partial Dierential Equations Igor Yanovsky, 2005 66
Problem (F99, #4). Consider the Cauchy problem
u
y
+a(x)u
x
= 0,
u(x, 0) = h(x).
Give an example of an (unbounded) smooth a(x) for which the solution of the Cauchy
problem is not unique.
Proof. is parameterized by : (s, 0, h(s)).
dx
dt
= a(x) x(t) x(0) =
_
t
0
a(x)dt x =
_
t
0
a(x)dt + s,
dy
dt
= 1 y(s, t) = t + c
1
(s) y = t,
dz
dt
= 0 z(s, t) = c
2
(s) z = h(s).
Thus,
u(x, t) = h
_
x
_
y
0
a(x)dy
_
Problem (F97, #7). a) Solve the Cauchy problem
u
t
xuu
x
= 0 < x < , t 0,
u(x, 0) = f(x) < x < .
b) Find a class of initial data such that this problem has a global solution for all t.
Compute the critical time for the existence of a smooth solution for initial data, f,
which is not in the above class.
Proof. a) is parameterized by : (s, 0, f(s)).
dx
dt
= xz
dx
dt
= xf(s) x = se
f(s)t
,
dy
dt
= 1 y = t,
dz
dt
= 0 z = f(s).
z = f
_
xe
f(s)t
_
,
u(x, y) = f
_
xe
uy
_
.
Check:
_
u
x
= f

(s) (e
uy
+ xe
uy
u
x
y)
u
y
= f

(s) xe
uy
(u
y
y +u)

_
u
x
f

(s)xe
uy
u
x
y = f

(s)e
uy
u
y
f

(s)xe
uy
u
y
y = f

(s)xe
uy
u

_
u
x
=
f

(s)e
uy
1f

(s)xye
uy
u
y
=
f

(s)e
uy
xu
1f

(s)xye
uy
u
y
xuu
x
=
f

(s)e
uy
xu
1 f

(s)xye
uy
xu
f

(s)e
uy
1 f

(s)xye
uy
= 0.
u(x, 0) = f(x).
Partial Dierential Equations Igor Yanovsky, 2005 67
b) The characteristics would intersect when 1 f

(s)xye
uy
= 0. Thus,
t
c
=
1
f

(s)xe
utc
.
Partial Dierential Equations Igor Yanovsky, 2005 68
Problem (F96, #6). Find an implicit formula for the solution u of the initial-value
problem
u
t
= (2x 1)tu
x
+ sin(x) t,
u(x, t = 0) = 0.
Evaluate u explicitly at the point (x = 0.5, t = 2).
Proof. Rewrite the equation as
u
y
+ (1 2x)yu
x
= sin(x) y.
is parameterized by : (s, 0, 0).
dx
dt
= (1 2x)y = (1 2x)t x =
1
2
(2s 1)e
t
2
+
1
2
,
_
s = (x
1
2
)e
t
2
+
1
2
_
,
dy
dt
= 1 y = t,
dz
dt
= sin(x) y = sin
_

2
(2s 1)e
t
2
+

2
_
t.
z(s, t) =
_
t
0
_
sin
_

2
(2s 1)e
t
2
+

2
_
t
_
dt +z(s, 0),
z(s, t) =
_
t
0
_
sin
_

2
(2s 1)e
t
2
+

2
_
t
_
dt.
u(x, y) =
_
y
0
_
sin
_

2
(2s 1)e
y
2
+

2
_
y
_
dy
=
_
y
0
_
sin
_

2
(2x 1)e
y
2
e
y
2
+

2
_
y
_
dy
=
_
y
0
_
sin
_

2
(2x 1) +

2
_
y
_
dy =
_
y
0
_
sin(x) y
_
dy,
u(x, y) = y sin(x)
y
2
2
.
Note: This solution does not satisfy the PDE.
Problem (S90, #8). Consider the Cauchy problem
u
t
= xu
x
u, < x < , t 0,
u(x, 0) = f(x), f(x) C

.
Assume that f 0 for [x[ 1.
Solve the equation by the method of characteristics and discuss the behavior of the
solution.
Proof. Rewrite the equation as
u
y
xu
x
= u,
is parameterized by : (s, 0, f(s)).
dx
dt
= x x = se
t
,
dy
dt
= 1 y = t,
dz
dt
= z z = f(s)e
t
.
u(x, y) = f(xe
y
)e
y
.
Partial Dierential Equations Igor Yanovsky, 2005 69
The solution satises the PDE and initial conditions.
As y +, u 0. u = 0 for [xe
y
[ 1 u = 0 for [x[
1
e
y
.
Partial Dierential Equations Igor Yanovsky, 2005 70
Problem (F02, #4). Consider the nonlinear hyperbolic equation
u
y
+uu
x
= 0 < x < .
a) Find a smooth solution to this equation for initial condition u(x, 0) = x.
b) Describe the breakdown of smoothness for the solution if u(x, 0) = x.
Proof. a) is parameterized by : (s, 0, s).
dx
dt
= z = s x = st +s s =
x
t + 1
=
x
y + 1
.
dy
dt
= 1 y = t,
dz
dt
= 0 z = s.
u(x, y) =
x
y + 1
; solution is smooth for all positive time y.
b) is parameterized by : (s, 0, s).
dx
dt
= z = s x = st + s s =
x
1 t
=
x
1 y
.
dy
dt
= 1 y = t,
dz
dt
= 0 z = s.
u(x, y) =
x
y 1
; solution blows up at time y = 1.
Partial Dierential Equations Igor Yanovsky, 2005 71
Problem (F97, #4). Solve the initial-boundary value problem
u
t
+ (x + 1)
2
u
x
= x for x > 0, t > 0
u(x, 0) = f(x) 0 < x < +
u(0, t) = g(t) 0 < t < +.
Proof. Rewrite the equation as
u
y
+ (x + 1)
2
u
x
= x for x > 0, y > 0
u(x, 0) = f(x) 0 < x < +
u(0, y) = g(y) 0 < y < +.
For region I, we solve the following characteristic equations with is parameterized
19
by : (s, 0, f(s)).
dx
dt
= (x + 1)
2
x =
s + 1
(s + 1)t 1
1,
dy
dt
= 1 y = t,
dz
dt
= x =
s + 1
(s + 1)t 1
1,
z = ln[(s + 1)t 1[ t +c
1
(s),
z = ln[(s + 1)t 1[ t +f(s).
In region I, characteristics are of the form
x =
s + 1
(s + 1)y 1
1.
Thus, region I is bounded above by the line
x =
1
y 1
1, or y =
x
x + 1
.
Since t = y, s =
xxyy
xy+y+1
, we have
u(x, y) = ln

_
x xy y
xy +y + 1
+ 1
_
y 1

y + f
_
x xy y
xy +y + 1
_
,
u(x, y) = ln

1
xy +y + 1

y + f
_
x xy y
xy +y + 1
_
.
For region II, is parameterized by : (0, s, g(s)).
dx
dt
= (x + 1)
2
x =
1
t 1
1,
dy
dt
= 1 y = t + s,
dz
dt
= x =
1
t 1
1,
z = ln[t 1[ t +c
2
(s),
z = ln[t 1[ t +g(s).
19
Variable t as a third coordinate of u and variable t used to parametrize characteristic equations
are two dierent entities.
Partial Dierential Equations Igor Yanovsky, 2005 72
Since t =
x
x+1
, s = y
x
x+1
, we have
u(x, y) = ln

x
x + 1
1

x
x + 1
+g
_
y
x
x + 1
_
.
Note that on y =
x
x+1
, both solutions are equal if f(0) = g(0).
Partial Dierential Equations Igor Yanovsky, 2005 73
Problem (S93, #3). Solve the following equation
u
t
+u
x
+ yu
y
= sint
for 0 t, 0 x, < y < and with
u = x + y for t = 0, x 0 and
u = t
2
+y for x = 0, t 0.
Proof. Rewrite the equation as (x x
1
, y x
2
, t x
3
):
u
x
3
+ u
x
1
+x
2
u
x
2
= sinx
3
for 0 x
3
, 0 x
1
, < x
2
< ,
u(x
1
, x
2
, 0) = x
1
+x
2
,
u(0, x
2
, x
3
) = x
2
3
+x
2
.
For region I, we solve the following characteristic equations with is parameterized
20
by : (s
1
, s
2
, 0, s
1
+ s
2
).
dx
1
dt
= 1 x
1
= t +s
1
,
dx
2
dt
= x
2
x
2
= s
2
e
t
,
dx
3
dt
= 1 x
3
= t,
dz
dt
= sin x
3
= sin t
z = cos t + s
1
+s
2
+ 1.
Since in region I, in x
1
x
3
-plane, characteristics are of the form x
1
= x
3
+s
1
, region
I is bounded above by the line x
1
= x
3
. Since t = x
3
, s
1
= x
1
x
3
, s
2
= x
2
e
x
3
, we
have
u(x
1
, x
2
, x
3
) = cos x
3
+ x
1
x
3
+x
2
e
x
3
+ 1, or
u(x, y, t) = cos t +x t + ye
t
+ 1, x t.
For region II, we solve the following characteristic equations with is parameterized
by : (0, s
2
, s
3
, s
2
+s
2
3
).
dx
1
dt
= 1 x
1
= t,
dx
2
dt
= x
2
x
2
= s
2
e
t
,
dx
3
dt
= 1 x
3
= t +s
3
,
dz
dt
= sin x
3
= sin(t + s
3
) z = cos(t +s
3
) + cos s
3
+s
2
+ s
2
3
.
Since t = x
1
, s
3
= x
3
x
1
, s
2
= x
2
e
x
3
, we have
u(x
1
, x
2
, x
3
) = cos x
3
+ cos(x
3
x
1
) +x
2
e
x
3
+ (x
3
x
1
)
2
, or
u(x, y, t) = cos t + cos(t x) +ye
t
+ (t x)
2
, x t.
Note that on x = t, both solutions are u(x = t, y) = cos x + ye
x
+ 1.
20
Variable t as a third coordinate of u and variable t used to parametrize characteristic equations
are two dierent entities.
Partial Dierential Equations Igor Yanovsky, 2005 74
Problem (W03, #5). Find a solution to
xu
x
+ (x +y)u
y
= 1
which satises u(1, y) = y for 0 y 1. Find a region in x 0, y 0 where u is
uniquely determined by these conditions.
Proof. is parameterized by : (1, s, s).
dx
dt
= x x = e
t
.
dy
dt
= x +y y

y = e
t
.
dz
dt
= 1 z = t + s.
The homogeneous solution for the second equation is y
h
(s, t) = c
1
(s)e
t
. Since the
right hand side and y
h
are linearly dependent, our guess for the particular solution is
y
p
(s, t) = c
2
(s)te
t
. Plugging in y
p
into the dierential equation, we get
c
2
(s)te
t
+ c
2
(s)e
t
c
2
(s)te
t
= e
t
c
2
(s) = 1.
Thus, y
p
(s, t) = te
t
and
y(s, t) = y
h
+ y
p
= c
1
(s)e
t
+te
t
.
Since y(s, 0) = s = c
1
(s), we get
y = se
t
+te
t
.
With and , we can solve for s and t in terms of x and y to get
t = lnx,
y = sx + x lnx s =
y x lnx
x
.
u(x, y) = t +s = lnx +
y x lnx
x
.
u(x, y) =
y
x
.
We have found that the characteristics in the xy-plane are of the form
y = sx + x lnx,
where s is such that 0 s 1. Also, the characteristics originate from .
Thus, u is uniquely determined in the region between the graphs:
y = x lnx,
y = x + x lnx.
Partial Dierential Equations Igor Yanovsky, 2005 75
12 Problems: Shocks
Example 1. Determine the exact solution to Burgers equation
u
t
+
_
1
2
u
2
_
x
= 0, t > 0
with initial data
u(x, 0) = h(x) =

1 if x < 1,
0 if 1 < x < 1,
1 if x > 1.
Proof. Characteristic form: u
t
+ uu
x
= 0.
The characteristic projection in xt-plane passing through the point (s, 0) is the line
x = h(s)t +s.
Rankine-Hugoniot shock condition at s = 1:
shock speed:

(t) =
F(u
r
) F(u
l
)
u
r
u
l
=
1
2
u
2
r

1
2
u
2
l
u
r
u
l
=
0
1
2
0 1
=
1
2
.
The 1/slope of the shock curve = 1/2. Thus,
x = (t) =
1
2
t +s,
and since the jump occurs at (1, 0), (0) = 1 = s. Therefore,
x =
1
2
t 1.
Rankine-Hugoniot shock condition at s = 1:
shock speed:

(t) =
F(u
r
) F(u
l
)
u
r
u
l
=
1
2
u
2
r

1
2
u
2
l
u
r
u
l
=
1
2
0
1 0
=
1
2
.
The 1/slope of the shock curve = 1/2. Thus,
x = (t) =
1
2
t + s,
and since the jump occurs at (1, 0), (0) = 1 = s. Therefore,
x =
1
2
t + 1.
At t = 2, Rankine-Hugoniot shock condition at s = 0:
shock speed:

(t) =
F(u
r
) F(u
l
)
u
r
u
l
=
1
2
u
2
r

1
2
u
2
l
u
r
u
l
=
1
2

1
2
1 1
= 0.
The 1/slope of the shock curve = 0. Thus,
x = (t) = s,
and since the jump occurs at (x, t) = (0, 2), (2) = 0 = s. Therefore,
x = 0.
Partial Dierential Equations Igor Yanovsky, 2005 76
For t < 2, u(x, t) =

1 if x <
1
2
t 1,
0 if
1
2
t 1 < x <
1
2
t + 1,
1 if x >
1
2
t + 1.
and for t > 2, u(x, t) =
_
1 if x < 0,
1 if x > 0.
Partial Dierential Equations Igor Yanovsky, 2005 77
Example 2. Determine the exact solution to Burgers equation
u
t
+
_
1
2
u
2
_
x
= 0, t > 0
with initial data
u(x, 0) = h(x) =

1 if x < 1,
0 if 1 < x < 1,
1 if x > 1.
Proof. Characteristic form: u
t
+ uu
x
= 0.
The characteristic projection in xt-plane passing through the point (s, 0) is the line
x = h(s)t +s.
For Burgers equation, for a rarefaction fan emanating from (s, 0) on xt-plane, we have:
u(x, t) =

u
l
,
xs
t
u
l
,
xs
t
, u
l

xs
t
u
r
,
u
r
,
xs
t
u
r
.
u(x, t) =

1, x < t 1,
x+1
t
, t 1 < x < 1, i.e. 1 <
x+1
t
< 0
0, 1 < x < 1,
x1
t
, 1 < x < t + 1, i.e. 0 <
x1
t
< 1
1, x > t + 1.
Partial Dierential Equations Igor Yanovsky, 2005 78
Partial Dierential Equations Igor Yanovsky, 2005 79
Example 3. Determine the exact solution to Burgers equation
u
t
+
_
1
2
u
2
_
x
= 0, t > 0
with initial data
u(x, 0) = h(x) =
_
2 if 0 < x < 1,
0 if otherwise.
Proof. Characteristic form: u
t
+ uu
x
= 0.
The characteristic projection in xt-plane passing through the point (s, 0) is the line
x = h(s)t +s.
Shock: Rankine-Hugoniot shock condition at s = 1:
shock speed:

(t) =
F(u
r
) F(u
l
)
u
r
u
l
=
1
2
u
2
r

1
2
u
2
l
u
r
u
l
=
0 2
0 2
= 1.
The 1/slope of the shock curve = 1. Thus,
x = (t) = t + s,
and since the jump occurs at (1, 0), (0) = 1 = s. Therefore,
x = t + 1.
Rarefaction: A rarefaction emanates from (0, 0) on xt-plane.
For 0 < t < 1, u(x, t) =

0 if x < 0,
x
t
if 0 < x < 2t,
2 if 2t < x < t + 1.
0 if x > t + 1.
Rarefaction catches up to shock at t = 1.
Shock: At (x, t) = (2, 1), u
l
= x/t, u
r
= 0. Rankine-Hugoniot shock condition:

(t) =
F(u
r
) F(u
l
)
u
r
u
l
=
1
2
u
2
r

1
2
u
2
l
u
r
u
l
=
0
1
2
(
x
t
)
2
0
x
t
=
1
2
x
t
,
dx
s
dt
=
x
2t
,
x = c

t,
and since the jump occurs at (x, t) = (2, 1), x(1) = 2 = c. Therefore, x = 2

t.
For t > 1, u(x, t) =

0 if x < 0,
x
t
if 0 < x < 2

t,
0 if x > 2

t.
Partial Dierential Equations Igor Yanovsky, 2005 80
Partial Dierential Equations Igor Yanovsky, 2005 81
Example 4. Determine the exact solution to Burgers equation
u
t
+
_
1
2
u
2
_
x
= 0, t > 0
with initial data
u(x, 0) = h(x) =
_
1 + x if x < 0,
0 if x > 0.
Proof. Characteristic form: u
t
+ uu
x
= 0.
The characteristic projection in xt-plane passing through the point (s, 0) is the line
x = h(s)t +s.
For s > 0, the characteristics are x = s.
For s < 0, the characteristics are x = (1 +s)t +s.
There are two ways to look for the solution on the left half-plane. One is to notice
that the characteristic at s = 0

## is x = t and characteristic at s = 1 is x = 1 and

that characteristics between s = and s = 0

## are intersecting at (x, t) = (1, 1).

Also, for a xed t, u is a linear function of x, i.e. for t = 0, u = 1 + x, allowing
a continuous change of u with x. Thus, the solution may be viewed as an implicit
rarefaction, originating at (1, 1), thus giving rise to the solution
u(x, t) =
x + 1
t + 1
.
Another way to nd a solution on the left half-plane is to solve for s to nd
s =
x t
1 +t
. Thus, u(x, t) = h(s) = 1 + s = 1 +
x t
1 +t
=
x + 1
t + 1
.
Shock: At (x, t) = (0, 0), u
l
=
x+1
t+1
, u
r
= 0. Rankine-Hugoniot shock condition:

(t) =
F(u
r
) F(u
l
)
u
r
u
l
=
1
2
u
2
r

1
2
u
2
l
u
r
u
l
=
0
1
2
(
x+1
t+1
)
2
0
x+1
t+1
=
1
2
x + 1
t + 1
,
dx
s
dt
=
1
2
x + 1
t + 1
,
x = c

t + 1 1,
and since the jump occurs at (x, t) = (0, 0), x(0) = 0 = c 1, or c = 1. Therefore,
the shock curve is x =

t + 1 1.
u(x, t) =
_
x+1
t+1
if x <

t + 1 1,
0 if x >

t + 1 1.
Partial Dierential Equations Igor Yanovsky, 2005 82
Partial Dierential Equations Igor Yanovsky, 2005 83
Example 5. Determine the exact solution to Burgers equation
u
t
+
_
1
2
u
2
_
x
= 0, t > 0
with initial data
u(x, 0) = h(x) =

u
0
if x < 0,
u
0
(1 x) if 0 < x < 1,
0 if x 1,
where u
0
> 0.
Proof. Characteristic form: u
t
+ uu
x
= 0.
The characteristic projection in xt-plane passing through the point (s, 0) is the line
x = h(s)t +s.
For s > 1, the characteristics are x = s.
For 0 < s < 1, the characteristics are x = u
0
(1 s)t +s.
For s < 0, the characteristics are x = u
0
t + s.
The characteristics emanating from (s, 0), 0 < s < 1 on xt-plane intersect at (1,
1
u
0
).
Also, we can check that the characteristics do not intersect before t =
1
u
0
for this
problem:
t
c
= min
_
1
h

(s)
_
=
1
u
0
.
To nd solution in a triangular domain between x = u
0
t and x = 1, we note that
characteristics there are x = u
0
(1 s)t + s. Solving for s we get
s =
x u
0
t
1 u
0
t
. Thus, u(x, t) = h(s) = u
0
(1 s) = u
0

_
1
x u
0
t
1 u
0
t
_
=
u
0
(1 x)
1 u
0
t
.
We can also nd a solution in the triangular domain as follows. Note, that the charac-
teristics are the straight lines
dx
dt
= u = const.
Integrating the equation above, we obtain
x = ut +c
Since all characteristics in the triangular domain meet at (1,
1
u
0
), we have c = 1
u
u
0
,
and
x = ut +
_
1
u
u
0
_
or u =
u
0
(1 x)
1 u
0
t
.
For 0 < t <
1
u
0
, u(x, t) =

u
0
if x < u
0
t,
u
0
(1x)
1u
0
t
if u
0
t < x < 1,
0 if x > 1.
Partial Dierential Equations Igor Yanovsky, 2005 84
Shock: At (x, t) = (1,
1
u
0
), Rankine-Hugoniot shock condition:

(t) =
F(u
r
) F(u
l
)
u
r
u
l
=
1
2
u
2
r

1
2
u
2
l
u
r
u
l
=
0
1
2
u
2
0
0 u
0
=
1
2
u
0
,
(t) =
1
2
u
0
t + c,
and since the jump occurs at (x, t) = (1,
1
u
0
), x
_
1
u
0
_
= 1 =
1
2
+c, or c =
1
2
. Therefore,
the shock curve is x =
u
0
t+1
2
.
Partial Dierential Equations Igor Yanovsky, 2005 85
For t >
1
u
0
, u(x, t) =
_
u
0
if x <
u
0
t+1
2
,
0 if x >
u
0
t+1
2
.
Problem. Show that for u = f(x/t) to be a nonconstant solution of u
t
+a(u)u
x
= 0,
f must be the inverse of the function a.
Proof. If u = f(x/t),
u
t
= f

_
x
t
_

x
t
2
and u
x
= f

_
x
t
_

1
t
.
Hence, u
t
+a(u)u
x
= 0 implies that
f

_
x
t
_

x
t
2
+ a
_
f
_
x
t
__
f

_
x
t
_

1
t
= 0
or, assuming f

## is not identically 0 to rule out the constant solution, that

a
_
f
_
x
t
__
=
x
t
.
This shows the functions a and f to be inverses of each other.
Partial Dierential Equations Igor Yanovsky, 2005 86
13 Problems: General Nonlinear Equations
13.1 Two Spatial Dimensions
Problem (S01, #3). Solve the initial value problem
1
2
u
2
x
u
y
=
x
2
2
,
u(x, 0) = x.
You will nd that the solution blows up in nite time. Explain this in terms of the
characteristics for this equation.
Proof. Rewrite the equation as
F(x, y, z, p, q) =
p
2
2
q +
x
2
2
= 0.
is parameterized by : (s, 0, s, (s), (s)).
We need to complete to a strip. Find (s) and (s), the initial conditions for p(s, t)
and q(s, t), respectively:
F(f(s), g(s), h(s), (s), (s)) = 0,
F(s, 0, s, (s), (s)) = 0,
(s)
2
2
(s) +
s
2
2
= 0,
(s) =
(s)
2
+ s
2
2
.
h

(s) = (s)f

(s) + (s)g

(s),
1 = (s).
(s) =
s
2
+ 1
2
.
Therefore, now is parametrized by : (s, 0, s, 1,
s
2
+1
2
).
dx
dt
= F
p
= p,
dy
dt
= F
q
= 1 y(s, t) = t +c
1
(s) y = t,
dz
dt
= pF
p
+qF
q
= p
2
q,
dp
dt
= F
x
F
z
p = x,
dq
dt
= F
y
F
z
q = 0 q(s, t) = c
2
(s) q =
s
2
+ 1
2
.
Thus, we found y and q in terms of s and t. Note that we have a coupled system:
_
x

= p,
p

= x,
which can be written as two second order ODEs:
x

+x = 0, x(s, 0) = s, x

(s, 0) = p(s, 0) = 1,
p

+ p = 0, p(s, 0) = 1, p

(s, 0) = x(s, 0) = s.
Partial Dierential Equations Igor Yanovsky, 2005 87
Solving the two equations separately, we get
x(s, t) = s cos t + sint,
p(s, t) = cos t s sint.
From this, we get
dz
dt
= p
2
q =
_
cos t s sint
_
2

s
2
+ 1
2
= cos
2
t 2s cos t sint +s
2
sin
2
t
s
2
+ 1
2
.
z(s, t) =
_
t
0
_
cos
2
t 2s cos t sint +s
2
sin
2
t
s
2
+ 1
2
_
dt + z(s, 0),
z(s, t) =
_
t
2
+
sint cos t
2
+s cos
2
t +
s
2
t
2

s
2
sint cos t
2

t(s
2
+ 1)
2
_
t
0
+s,
=
_
sin t cos t
2
+s cos
2
t
s
2
sint cos t
2
_
t
0
+ s,
=
sint cos t
2
+ s cos
2
t
s
2
sin t cos t
2
s + s =
=
sint cos t
2
+ s cos
2
t
s
2
sin t cos t
2
.
Plugging in x and y found earlier for s and t, we get
u(x, y) =
sin(y) cos(y)
2
+
x sin(y)
cos(y)
cos
2
(y)
(x sin(y))
2
cos
2
(y)

sin(y) cos(y)
2
=
sin y cos y
2
+
x + siny
cos y
cos
2
y +
(x + sin y)
2
cos
2
y

siny cos y
2
=
sin y cos y
2
+ (x + sin y) cos y +
(x + siny)
2
sin y
2 cos y
= x cos y +
sin y cos y
2
+
(x + sin y)
2
siny
2 cos y
.
Partial Dierential Equations Igor Yanovsky, 2005 88
Problem (S98, #3). Find the solution of
u
t
+
u
2
x
2
=
x
2
2
, t 0, < x <
u(x, 0) = h(x), < x < ,
where h(x) is smooth function which vanishes for [x[ large enough.
Proof. Rewrite the equation as
F(x, y, z, p, q) =
p
2
2
+ q +
x
2
2
= 0.
is parameterized by : (s, 0, h(s), (s), (s)).
We need to complete to a strip. Find (s) and (s), the initial conditions for p(s, t)
and q(s, t), respectively:
F(f(s), g(s), h(s), (s), (s)) = 0,
F(s, 0, h(s), (s), (s)) = 0,
(s)
2
2
+(s) +
s
2
2
= 0,
(s) =
(s)
2
+s
2
2
.
h

(s) = (s)f

(s) + (s)g

(s),
h

(s) = (s).
(s) =
h

(s)
2
+ s
2
2
.
Therefore, now is parametrized by : (s, 0, s, h

(s),
h

(s)
2
+s
2
2
).
dx
dt
= F
p
= p,
dy
dt
= F
q
= 1 y(s, t) = t + c
1
(s) y = t,
dz
dt
= pF
p
+qF
q
= p
2
+q,
dp
dt
= F
x
F
z
p = x,
dq
dt
= F
y
F
z
q = 0 q(s, t) = c
2
(s) q =
h

(s)
2
+ s
2
2
.
Thus, we found y and q in terms of s and t. Note that we have a coupled system:
_
x

= p,
p

= x,
which can be written as a second order ODE:
x

+x = 0, x(s, 0) = s, x

(s, 0) = p(s, 0) = h

(s).
Solving the equation, we get
x(s, t) = s cos t +h

(s) sint,
p(s, t) = x

(s, t) = h

## (s) cos t s sin t.

Partial Dierential Equations Igor Yanovsky, 2005 89
From this, we get
dz
dt
= p
2
+ q =
_
h

_
2

(s)
2
+ s
2
2
= h

(s)
2
cos
2
t 2sh

2
sin
2
t
h

(s)
2
+s
2
2
.
z(s, t) =
_
t
0
_
h

(s)
2
cos
2
t 2sh

2
sin
2
t
h

(s)
2
+s
2
2
_
dt + z(s, 0)
=
_
t
0
_
h

(s)
2
cos
2
t 2sh

## (s) cos t sint +s

2
sin
2
t
h

(s)
2
+s
2
2
_
dt + h(s).
We integrate the above expression similar to S

## 01#3 to get an expression for z(s, t).

Plugging in x and y found earlier for s and t, we get u(x, y).
Partial Dierential Equations Igor Yanovsky, 2005 90
Problem (S97, #4).
Describe the method of the bicharacteristics for solving the initial value problem
_

x
u(x, y)
_
2
+
_

y
u(x, y)
_
2
= 2 +y,
u(x, 0) = u
0
(x) = x.
Assume that [

x
u
0
(x)[ < 2 and consider the solution such that
u
y
> 0.
Apply all general computations for the particular case u
0
(x) = x.
Proof. We have
u
2
x
+u
2
y
= 2 + y
u(x, 0) = u
0
(x) = x.
Rewrite the equation as
F(x, y, z, p, q) = p
2
+q
2
y 2 = 0.
is parameterized by : (s, 0, s, (s), (s)).
We need to complete to a strip. Find (s) and (s), the initial conditions for p(s, t)
and q(s, t), respectively:
F(f(s), g(s), h(s), (s), (s)) = 0,
F(s, 0, s, (s), (s)) = 0,
(s)
2
+(s)
2
2 = 0,
(s)
2
+(s)
2
= 2.
h

(s) = (s)f

(s) + (s)g

(s),
1 = (s).
(s) = 1.
Since we have a condition that q(s, t) > 0, we choose q(s, 0) = (s) = 1.
Therefore, now is parametrized by : (s, 0, s, 1, 1).
dx
dt
= F
p
= 2p
dx
dt
= 2 x = 2t +s,
dy
dt
= F
q
= 2q
dy
dt
= 2t + 2 y = t
2
+ 2t,
dz
dt
= pF
p
+qF
q
= 2p
2
+ 2q
2
= 2y + 4
dz
dt
= 2t
2
+ 4t + 4,
z =
2
3
t
3
+ 2t
2
+ 4t +s =
2
3
t
3
+ 2t
2
+ 4t +x 2t =
2
3
t
3
+ 2t
2
+ 2t +x,
dp
dt
= F
x
F
z
p = 0 p = 1,
dq
dt
= F
y
F
z
q = 1 q = t + 1.
We solve y = t
2
+ 2t, a quadratic equation in t, t
2
+ 2t y = 0, for t in terms of y to
get:
t = 1
_
1 + y.
u(x, y) =
2
3
(1
_
1 +y)
3
+ 2(1
_
1 + y)
2
+ 2(1
_
1 +y) +x.
Both u

## satisfy the PDE. u

x
= 1, u
y
=

y + 1 u
2
x
+u
2
y
= y + 2
u
+
satises u
+
(x, 0) = x . However, u

## does not satisfy IC, i.e. u

(x, 0) = x
4
3
.
Partial Dierential Equations Igor Yanovsky, 2005 91
Problem (S02, #6). Consider the equation
u
x
+u
x
u
y
= 1,
u(x, 0) = f(x).
Assuming that f is dierentiable, what conditions on f insure that the problem is
noncharacteristic? If f satises those conditions, show that the solution is
u(x, y) = f(r) y +
2y
f

(r)
,
where r must satisfy y = (f

(r))
2
(x r).
Finally, show that one can solve the equation for (x, y) in a suciently small neighbor-
hood of (x
0
, 0) with r(x
0
, 0) = x
0
.
Proof. Solved.
In order to solve the Cauchy problem in a neighborhood of , need:
f

(s) F
q
[f, g, h, , ](s) g

(s) F
p
[f, g, h, , ](s) ,= 0,
1 h

(s) 0
_
1 +
1 h

(s)
h

(s)
_
,= 0,
h

(s) ,= 0.
Thus, h

## (s) ,= 0 ensures that the problem is noncharacteristic.

To show that one can solve y = (f

(s))
2
(x s) for (x, y) in a suciently small
neighborhood of (x
0
, 0) with s(x
0
, 0) = x
0
, let
G(x, y, s) = (f

(s))
2
(x s) y = 0,
G(x
0
, 0, x
0
) = 0,
G
r
(x
0
, 0, x
0
) = (f

(s))
2
.
Hence, if f

(s) ,= 0, s, then G
s
(x
0
, 0, x
0
) ,= 0 and we can use the implicit function
theorem in a neighborhood of (x
0
, 0, x
0
) to get
G(x, y, h(x, y)) = 0
and solve the equation in terms of x and y.
Partial Dierential Equations Igor Yanovsky, 2005 92
Problem (S00, #1). Find the solutions of
(u
x
)
2
+ (u
y
)
2
= 1
in a neighborhood of the curve y =
x
2
2
satisfying the conditions
u
_
x,
x
2
2
_
= 0 and u
y
_
x,
x
2
2
_
> 0.
Proof. Rewrite the equation as
F(x, y, z, p, q) = p
2
+q
2
1 = 0.
is parameterized by : (s,
s
2
2
, 0, (s), (s)).
We need to complete to a strip. Find (s) and (s), the initial conditions for p(s, t)
and q(s, t), respectively:
F(f(s), g(s), h(s), (s), (s)) = 0,
F
_
s,
s
2
2
, 0, (s), (s)
_
= 0,
(s)
2
+(s)
2
= 1.
h

(s) = (s)f

(s) + (s)g

(s),
0 = (s) +s(s),
(s) = s(s).
Thus, s
2
(s)
2
+(s)
2
= 1 (s)
2
=
1
s
2
+ 1
.
Since, by assumption, (s) > 0, we have (s) =
1

s
2
+1
.
Therefore, now is parametrized by :
_
s,
s
2
2
, 0,
s

s
2
+1
,
1

s
2
+1
_
.
dx
dt
= F
p
= 2p =
2s

s
2
+ 1
x =
2st

s
2
+ 1
+ s,
dy
dt
= F
q
= 2q =
2

s
2
+ 1
y =
2t

s
2
+ 1
+
s
2
2
,
dz
dt
= pF
p
+qF
q
= 2p
2
+ 2q
2
= 2 z = 2t,
dp
dt
= F
x
F
z
p = 0 p =
s

s
2
+ 1
,
dq
dt
= F
y
F
z
q = 0 q =
1

s
2
+ 1
.
Thus, in parametric form,
z(s, t) = 2t,
x(s, t) =
2st

s
2
+ 1
+s,
y(s, t) =
2t

s
2
+ 1
+
s
2
2
.
Partial Dierential Equations Igor Yanovsky, 2005 93
13.2 Three Spatial Dimensions
Problem (S96, #2). Solve the following Cauchy problem
21
:
u
x
+u
2
y
+u
2
z
= 1,
u(0, y, z) = y z.
Proof. Rewrite the equation as
u
x
1
+u
2
x
2
+u
2
x
3
= 1,
u(0, x
2
, x
3
) = x
2
x
3
.
Write a general nonlinear equation
F(x
1
, x
2
, x
3
, z, p
1
, p
2
, p
3
) = p
1
+ p
2
2
+ p
2
3
1 = 0.
is parameterized by
:
_
0
..
x
1
(s
1
,s
2
,0)
, s
1
..
x
2
(s
1
,s
2
,0)
, s
2
..
x
3
(s
1
,s
2
,0)
, s
1
s
2
..
z(s
1
,s
2
,0)
,
1
(s
1
, s
2
)
. .
p
1
(s
1
,s
2
,0)
,
2
(s
1
, s
2
)
. .
p
2
(s
1
,s
2
,0)
,
3
(s
1
, s
2
)
. .
p
3
(s
1
,s
2
,0)
_
We need to complete to a strip. Find
1
(s
1
, s
2
),
2
(s
1
, s
2
), and
3
(s
1
, s
2
), the initial
conditions for p
1
(s
1
, s
2
, t), p
2
(s
1
, s
2
, t), and p
3
(s
1
, s
2
, t), respectively:
F
_
f
1
(s
1
, s
2
), f
2
(s
1
, s
2
), f
3
(s
1
, s
2
), h(s
1
, s
2
),
1
,
2
,
3
_
= 0,
F
_
0, s
1
, s
2
, s
1
s
2
,
1
,
2
,
3
_
=
1
+
2
2
+
2
3
1 = 0,

1
+
2
2
+
2
3
= 1.

h
s
1
=
1
f
1
s
1
+
2
f
2
s
1
+
3
f
3
s
1
,
s
2
=
2
.

h
s
2
=
1
f
1
s
2
+
2
f
2
s
2
+
3
f
3
s
2
,
s
1
=
3
.
Thus, we have:
2
= s
2
,
3
= s
1
,
1
= s
2
1
s
2
2
+ 1.
:
_
0
..
x
1
(s
1
,s
2
,0)
, s
1
..
x
2
(s
1
,s
2
,0)
, s
2
..
x
3
(s
1
,s
2
,0)
, s
1
s
2
..
z(s
1
,s
2
,0)
, s
2
1
s
2
2
+ 1
. .
p
1
(s
1
,s
2
,0)
, s
2
..
p
2
(s
1
,s
2
,0)
, s
1
..
p
3
(s
1
,s
2
,0)
_
21
This problem is very similar to an already hand-written solved problem F95 #2.
Partial Dierential Equations Igor Yanovsky, 2005 94
The characteristic equations are
dx
1
dt
= F
p
1
= 1 x
1
= t,
dx
2
dt
= F
p
2
= 2p
2

dx
2
dt
= 2s
2
x
2
= 2s
2
t + s
1
,
dx
3
dt
= F
p
3
= 2p
3

dx
3
dt
= 2s
1
x
3
= 2s
1
t + s
2
,
dz
dt
= p
1
F
p
1
+ p
2
F
p
2
+p
3
F
p
3
= p
1
+ 2p
2
2
+ 2p
2
3
= s
2
1
s
2
2
+ 1 + 2s
2
2
+ 2s
2
1
= s
2
1
+s
2
2
+ 1 z = (s
2
1
+ s
2
2
+ 1)t + s
1
s
2
,
dp
1
dt
= F
x
1
p
1
F
z
= 0 p
1
= s
2
1
s
2
2
+ 1,
dp
2
dt
= F
x
2
p
2
F
z
= 0 p
2
= s
2
,
dp
3
dt
= F
x
3
p
3
F
z
= 0 p
3
= s
1
.
Thus, we have

x
1
= t
x
2
= 2s
2
t + s
1
x
3
= 2s
1
t + s
2
z = (s
2
1
+ s
2
2
+ 1)t + s
1
s
2

t = x
1
s
1
= x
2
2s
2
t
s
2
= x
3
2s
1
t
z = (s
2
1
+s
2
2
+ 1)t +s
1
s
2

t = x
1
s
1
=
x
2
2x
1
x
3
14x
2
1
s
2
=
x
3
2x
1
x
2
14x
2
1
z = (s
2
1
+s
2
2
+ 1)t +s
1
s
2
u(x
1
, x
2
, x
3
) =
_
_
x
2
2x
1
x
3
1 4x
2
1
_
2
+
_
x
3
2x
1
x
2
1 4x
2
1
_
2
+ 1
_
x
1
+
_
x
2
2x
1
x
3
1 4x
2
1
__
x
3
2x
1
x
2
1 4x
2
1
_
.
Problem (F95, #2). Solve the following Cauchy problem
u
x
+u
y
+u
3
z
= x +y + z,
u(x, y, 0) = xy.
Proof. Solved
Partial Dierential Equations Igor Yanovsky, 2005 95
Problem (S94, #1). Solve the following PDE for f(x, y, t):
f
t
+xf
x
+ 3t
2
f
y
= 0
f(x, y, 0) = x
2
+ y
2
.
Proof. Rewrite the equation as (x x
1
, y x
2
, t x
3
, f u):
x
1
u
x
1
+ 3x
2
3
u
x
2
+u
x
3
= 0,
u(x
1
, x
2
, 0) = x
2
1
+ x
2
2
.
F(x
1
, x
2
, x
3
, z, p
1
, p
2
, p
3
) = x
1
p
1
+ 3x
2
3
p
2
+ p
3
= 0.
is parameterized by
:
_
s
1
..
x
1
(s
1
,s
2
,0)
, s
2
..
x
2
(s
1
,s
2
,0)
, 0
..
x
3
(s
1
,s
2
,0)
, s
2
1
+s
2
2
. .
z(s
1
,s
2
,0)
,
1
(s
1
, s
2
)
. .
p
1
(s
1
,s
2
,0)
,
2
(s
1
, s
2
)
. .
p
2
(s
1
,s
2
,0)
,
3
(s
1
, s
2
)
. .
p
3
(s
1
,s
2
,0)
_
We need to complete to a strip. Find
1
(s
1
, s
2
),
2
(s
1
, s
2
), and
3
(s
1
, s
2
), the initial
conditions for p
1
(s
1
, s
2
, t), p
2
(s
1
, s
2
, t), and p
3
(s
1
, s
2
, t), respectively:
F
_
f
1
(s
1
, s
2
), f
2
(s
1
, s
2
), f
3
(s
1
, s
2
), h(s
1
, s
2
),
1
,
2
,
3
_
= 0,
F
_
s
1
, s
2
, 0, s
2
1
+ s
2
2
,
1
,
2
,
3
_
= s
1

1
+
3
= 0,

3
= s
1

1
.

h
s
1
=
1
f
1
s
1
+
2
f
2
s
1
+
3
f
3
s
1
,
2s
1
=
1
.

h
s
2
=
1
f
1
s
2
+
2
f
2
s
2
+
3
f
3
s
2
,
2s
2
=
2
.
Thus, we have:
1
= 2s
1
,
2
= 2s
2
,
3
= 2s
2
1
.
:
_
s
1
..
x
1
(s
1
,s
2
,0)
, s
2
..
x
2
(s
1
,s
2
,0)
, 0
..
x
3
(s
1
,s
2
,0)
, s
2
1
+s
2
2
. .
z(s
1
,s
2
,0)
, 2s
1
..
p
1
(s
1
,s
2
,0)
, 2s
2
..
p
2
(s
1
,s
2
,0)
, 2s
2
1
..
p
3
(s
1
,s
2
,0)
_
The characteristic equations are
dx
1
dt
= F
p
1
= x
1
x
1
= s
1
e
t
,
dx
2
dt
= F
p
2
= 3x
2
3

dx
2
dt
= 3t
2
x
2
= t
3
+s
2
,
dx
3
dt
= F
p
3
= 1 x
3
= t,
dz
dt
= p
1
F
p
1
+p
2
F
p
2
+p
3
F
p
3
= p
1
x
1
+p
2
3x
2
3
+ p
3
= 0 z = s
2
1
+ s
2
2
,
dp
1
dt
= F
x
1
p
1
F
z
= p
1
p
1
= 2s
1
e
t
,
dp
2
dt
= F
x
2
p
2
F
z
= 0 p
2
= 2s
2
,
dp
3
dt
= F
x
3
p
3
F
z
= 6x
3
p
2

dp
3
dt
= 12ts
2
p
3
= 6t
2
s
2
+ 2s
2
1
.
With t = x
3
, s
1
= x
1
e
x
3
, s
2
= x
2
x
3
3
, we have
u(x
1
, x
2
, x
3
) = x
2
1
e
2x
3
+ (x
2
x
3
3
)
2
.
_
f(x, y, t) = x
2
e
2t
+ (y t
3
)
2
.
_
The solution satises the PDE and initial condition.
Partial Dierential Equations Igor Yanovsky, 2005 96
Problem (F93, #3). Find the solution of the following equation
f
t
+xf
x
+ (x +t)f
y
= t
3
f(x, y, 0) = xy.
Proof. Rewrite the equation as (x x
1
, y x
2
, t x
3
, f u):
x
1
u
x
1
+ (x
1
+x
3
)u
x
2
+u
x
3
= x
3
,
u(x
1
, x
2
, 0) = x
1
x
2
.
Method I: Treat the equation as a QUASILINEAR equation.
is parameterized by : (s
1
, s
2
, 0, s
1
s
2
).
dx
1
dt
= x
1
x
1
= s
1
e
t
,
dx
2
dt
= x
1
+ x
3

dx
2
dt
= s
1
e
t
+t x
2
= s
1
e
t
+
t
2
2
+s
2
s
1
,
dx
3
dt
= 1 x
3
= t,
dz
dt
= x
3
3

dz
dt
= t
3
z =
t
4
4
+s
1
s
2
.
Since t = x
3
, s
1
= x
1
e
x
3
, s
2
= x
2
s
1
e
t

t
2
2
+ s
1
= x
2
x
1

x
2
3
2
+x
1
e
x
3
, we have
u(x
1
, x
2
, x
3
) =
x
4
3
4
+x
1
e
x
3
(x
2
x
1

x
2
3
2
+x
1
e
x
3
), or
f(x, y, t) =
t
4
4
+ xe
t
(y x
t
2
2
+xe
t
).
The solution satises the PDE and initial condition.
Method II: Treat the equation as a fully NONLINEAR equation.
F(x
1
, x
2
, x
3
, z, p
1
, p
2
, p
3
) = x
1
p
1
+ (x
1
+x
3
)p
2
+ p
3
x
3
3
= 0.
is parameterized by
:
_
s
1
..
x
1
(s
1
,s
2
,0)
, s
2
..
x
2
(s
1
,s
2
,0)
, 0
..
x
3
(s
1
,s
2
,0)
, s
1
s
2
..
z(s
1
,s
2
,0)
,
1
(s
1
, s
2
)
. .
p
1
(s
1
,s
2
,0)
,
2
(s
1
, s
2
)
. .
p
2
(s
1
,s
2
,0)
,
3
(s
1
, s
2
)
. .
p
3
(s
1
,s
2
,0)
_
We need to complete to a strip. Find
1
(s
1
, s
2
),
2
(s
1
, s
2
), and
3
(s
1
, s
2
), the initial
conditions for p
1
(s
1
, s
2
, t), p
2
(s
1
, s
2
, t), and p
3
(s
1
, s
2
, t), respectively:
F
_
f
1
(s
1
, s
2
), f
2
(s
1
, s
2
), f
3
(s
1
, s
2
), h(s
1
, s
2
),
1
,
2
,
3
_
= 0,
F
_
s
1
, s
2
, 0, s
1
s
2
,
1
,
2
,
3
_
= s
1

1
+s
1

2
+
3
= 0,

3
= s
1
(
1
+
2
).

h
s
1
=
1
f
1
s
1
+
2
f
2
s
1
+
3
f
3
s
1
,
s
2
=
1
.

h
s
2
=
1
f
1
s
2
+
2
f
2
s
2
+
3
f
3
s
2
,
s
1
=
2
.
Thus, we have:
1
= s
2
,
2
= s
1
,
3
= s
2
1
s
1
s
2
.
:
_
s
1
..
x
1
(s
1
,s
2
,0)
, s
2
..
x
2
(s
1
,s
2
,0)
, 0
..
x
3
(s
1
,s
2
,0)
, s
1
s
2
..
z(s
1
,s
2
,0)
, s
2
..
p
1
(s
1
,s
2
,0)
, s
1
..
p
2
(s
1
,s
2
,0)
, s
2
1
s
1
s
2
. .
p
3
(s
1
,s
2
,0)
_
Partial Dierential Equations Igor Yanovsky, 2005 97
The characteristic equations are
dx
1
dt
= F
p
1
= x
1
x
1
= s
1
e
t
,
dx
2
dt
= F
p
2
= x
1
+x
3

dx
2
dt
= s
1
e
t
+ t x
2
= s
1
e
t
+
t
2
2
+s
2
s
1
,
dx
3
dt
= F
p
3
= 1 x
3
= t,
dz
dt
= p
1
F
p
1
+p
2
F
p
2
+ p
3
F
p
3
= p
1
x
1
+p
2
(x
1
+ x
3
) + p
3
= x
3
3
= t
3
z =
t
4
4
+s
1
s
2
,
dp
1
dt
= F
x
1
p
1
F
z
= p
1
p
2
= p
1
s
1
p
1
= 2s
1
e
t
s
1
,
dp
2
dt
= F
x
2
p
2
F
z
= 0 p
2
= s
1
,
dp
3
dt
= F
x
3
p
3
F
z
= 3x
2
3
p
2
= 3t
2
s
1
p
3
= t
3
s
1
t s
2
1
s
1
s
2
.
With t = x
3
, s
1
= x
1
e
x
3
, s
2
= x
2
s
1
e
t

t
2
2
+s
1
= x
2
x
1

x
2
3
2
+x
1
e
x
3
, we have
u(x
1
, x
2
, x
3
) =
x
4
3
4
+x
1
e
x
3
(x
2
x
1

x
2
3
2
+x
1
e
x
3
), or
f(x, y, t) =
t
4
4
+ xe
t
(y x
t
2
2
+xe
t
).
22
The solution satises the PDE and initial condition.
22
Variable t in the derivatives of characteristics equations and t in the solution f(x, y, t) are dierent
entities.
Partial Dierential Equations Igor Yanovsky, 2005 98
Problem (F92, #1). Solve the initial value problem
u
t
+u
x
+ u
y
+u = 0 for t > 0
u(x, y, 0) = (x, y),
in which , and are real constants and is a smooth function.
Proof. Rewrite the equation as (x x
1
, y x
2
, t x
3
)
23
:
u
x
1
+ u
x
2
+ u
x
3
= u,
u(x
1
, x
2
, 0) = (x
1
, x
2
).
is parameterized by : (s
1
, s
2
, 0, (s
1
, s
2
)).
dx
1
dt
= x
1
= t + s
1
,
dx
2
dt
= x
2
= t +s
2
,
dx
3
dt
= 1 x
3
= t,
dz
dt
= z
dz
z
= dt z = (s
1
, s
2
)e
t
.
J det
_
(x
1
, x
2
, x
3
)
(s
1
, s
2
, t)
_
=

1 0 0
0 1 0
1

= 1 ,= 0 J is invertible.
Since t = x
3
, s
1
= x
1
x
3
, s
2
= x
2
x
3
, we have
u(x
1
, x
2
, x
3
) = (x
1
x
3
, x
2
x
3
)e
x
3
, or
u(x, y, t) = (x t, y t)e
t
.
The solution satises the PDE and initial condition.
24
23
Variable t as a third coordinate of u and variable t used to parametrize characteristic equations
are two dierent entities.
24
Chain Rule: u(x1, x2, x3) = (f(x1, x2, x3), g(x1, x2, x3)), then ux
1
=

f
f
x
1
+

g
g
x
1
.
Partial Dierential Equations Igor Yanovsky, 2005 99
Problem (F94, #2). Find the solution of the Cauchy problem
u
t
(x, y, t) +au
x
(x, y, t) + bu
y
(x, y, t) + c(x, y, t)u(x, y, t) = 0
u(x, y, 0) = u
0
(x, y),
where 0 < t < +, < x < +, < y < +,
a, b are constants, c(x, y, t) is a continuous function of (x, y, t), and u
0
(x, y) is a con-
tinuous function of (x, y).
Proof. Rewrite the equation as (x x
1
, y x
2
, t x
3
):
au
x
1
+ bu
x
2
+ u
x
3
= c(x
1
, x
2
, x
3
)u,
u(x
1
, x
2
, 0) = u
0
(x
1
, x
2
).
is parameterized by : (s
1
, s
2
, 0, u
0
(s
1
, s
2
)).
dx
1
dt
= a x
1
= at + s
1
,
dx
2
dt
= b x
2
= bt + s
2
,
dx
3
dt
= 1 x
3
= t,
dz
dt
= c(x
1
, x
2
, x
3
)z
dz
dt
= c(at + s
1
, bt +s
2
, t)z
dz
z
= c(at +s
1
, bt +s
2
, t)dt
ln z =
_
t
0
c(a +s
1
, b + s
2
, )d + c
1
(s
1
, s
2
),
z(s
1
, s
2
, t) = c
2
(s
1
, s
2
)e

_
t
0
c(a+s
1
,b+s
2
,)d
z(s
1
, s
2
, 0) = c
2
(s
1
, s
2
) = u
0
(s
2
, s
2
),
z(s
1
, s
2
, t) = u
0
(s
1
, s
2
)e

_
t
0
c(a+s
1
,b+s
2
,)d
.
J det
_
(x
1
, x
2
, x
3
)
(s
1
, s
2
, t)
_
=

1 0 0
0 1 0
a b 1

= 1 ,= 0 J is invertible.
Since t = x
3
, s
1
= x
1
ax
3
, s
2
= x
2
bx
3
, we have
u(x
1
, x
2
, x
3
) = u
0
(x
1
ax
3
, x
2
bx
3
)e

_
x
3
0
c(a+x
1
ax
3
,b+x
2
bx
3
,)d
= u
0
(x
1
ax
3
, x
2
bx
3
)e

_
x
3
0
c(x
1
+a(x
3
),x
2
+b(x
3
),)d
, or
u(x, y, t) = u
0
(x at, y bt)e

_
t
0
c(x+a(t),y+b(t),)d
.
Partial Dierential Equations Igor Yanovsky, 2005 100
Problem (F89, #4). Consider the rst order partial dierential equation
u
t
+ ( + t)u
x
+e
t
u
y
= 0 (13.1)
in which , and are constants.
a) For this equation, solve the initial value problem with initial data
u(x, y, t = 0) = sin(xy) (13.2)
for all x and y and for t 0.
b) Suppose that this initial data is prescribed only for x 0 (and all y) and consider
(13.1) in the region x 0, t 0 and all y. For which values of , and is it possible
to solve the initial-boundary value problem (13.1), (13.2) with u(x = 0, y, t) given for
t 0?
For non-permissible values of , and , where can boundary values be prescribed in
order to determine a solution of (13.1) in the region (x 0, t 0, all y).
Proof. a) Rewrite the equation as (x x
1
, y x
2
, t x
3
):
( +x
3
)u
x
1
+e
x
3
u
x
2
+ u
x
3
= 0,
u(x
1
, x
2
, 0) = sin(x
1
x
2
).
is parameterized by : (s
1
, s
2
, 0, sin(s
1
s
2
)).
dx
1
dt
= +x
3

dx
1
dt
= +t x
1
=
t
2
2
+ t + s
1
,
dx
2
dt
= e
x
3

dx
2
dt
= e
t
x
2
= e
t
+ s
2
,
dx
3
dt
= 1 x
3
= t,
dz
dt
= 0 z = sin(s
1
s
2
).
J det
_
(x
1
, x
2
, x
3
)
(s
1
, s
2
, t)
_
=

1 0 0
0 1 0
t + e
t
1

= 1 ,= 0 J is invertible.
Since t = x
3
, s
1
= x
1

x
2
3
2
x
3
, s
2
= x
2
e
x
3
+, we have
u(x
1
, x
2
, x
3
) = sin((x
1

x
2
3
2
x
3
)(x
2
e
x
3
+)), or
u(x, y, t) = sin((x
t
2
2
t)(y e
t
+)).
The solution satises the PDE and initial condition.
b) We need a compatibility condition between the initial and boundary values to hold
on y-axis (x = 0, t = 0):
u(x = 0, y, 0) = u(0, y, t = 0),
0 = 0.
Partial Dierential Equations Igor Yanovsky, 2005 101
Partial Dierential Equations Igor Yanovsky, 2005 102
14 Problems: First-Order Systems
Problem (S01, #2a). Find the solution u =
_
u
1
(x, t)
u
2
(x, t)
_
, (x, t) R R,
to the (strictly) hyperbolic equation
u
t

_
1 0
5 3
_
u
x
= 0,
satisfying
_
u
1
(x, 0)
u
2
(x, 0)
_
=
_
e
ixa
0
_
, a R.
Proof. Rewrite the equation as
U
t
+
_
1 0
5 3
_
U
x
= 0,
U(x, 0) =
_
u
(1)
(x, 0)
u
(2)
(x, 0)
_
=
_
e
ixa
0
_
.
The eigenvalues of the matrix A are
1
= 1,
2
= 3 and the corresponding
eigenvectors are e
1
=
_
2
5
_
, e
2
=
_
0
1
_
. Thus,
=
_
1 0
0 3
_
, =
_
2 0
5 1
_
,
1
=
1
det
=
_
1
2
0
5
2
1
_
.
Let U = V . Then,
U
t
+AU
x
= 0,
V
t
+AV
x
= 0,
V
t
+
1
AV
x
= 0,
V
t
+ V
x
= 0.
Thus, the transformed problem is
V
t
+
_
1 0
0 3
_
V
x
= 0,
V (x, 0) =
1
U(x, 0) =
_
1
2
0
5
2
1
__
e
ixa
0
_
=
1
2
e
ixa
_
1
5
_
.
We have two initial value problems
_
v
(1)
t
v
(1)
x
= 0,
v
(1)
(x, 0) =
1
2
e
ixa
;
_
v
(2)
t
3v
(2)
x
= 0,
v
(2)
(x, 0) =
5
2
e
ixa
,
which we solve by characteristics to get
v
(1)
(x, t) =
1
2
e
ia(x+t)
, v
(2)
(x, t) =
5
2
e
ia(x+3t)
.
We solve for U: U = V =
_
v
(1)
v
(2)
_
=
_
2 0
5 1
__
1
2
e
ia(x+t)
5
2
e
ia(x+3t)
_
.
Thus, U =
_
u
(1)
(x, t)
u
(2)
(x, t)
_
=
_
e
ia(x+t)

5
2
e
ia(x+t)
+
5
2
e
ia(x+3t)
_
.
Can check that this is the correct solution by plugging it into the original equation.
Partial Dierential Equations Igor Yanovsky, 2005 103
Part (b) of the problem is solved in the Fourier Transform section.
Partial Dierential Equations Igor Yanovsky, 2005 104
Problem (S96, #7). Solve the following initial-boundary value problem in the do-
main x > 0, t > 0, for the unknown vector U =
_
u
(1)
u
(2)
_
:
U
t
+
_
2 3
0 1
_
U
x
= 0. (14.1)
U(x, 0) =
_
sinx
0
_
and u
(2)
(0, t) = t.
Proof. The eigenvalues of the matrix A are
1
= 2,
2
= 1 and the corresponding
eigenvectors are e
1
=
_
1
0
_
, e
2
=
_
1
1
_
. Thus,
=
_
2 0
0 1
_
, =
_
1 1
0 1
_
,
1
=
1
det
=
_
1 1
0 1
_
.
Let U = V . Then,
U
t
+AU
x
= 0,
V
t
+AV
x
= 0,
V
t
+
1
AV
x
= 0,
V
t
+ V
x
= 0.
Thus, the transformed problem is
V
t
+
_
2 0
0 1
_
V
x
= 0, (14.2)
V (x, 0) =
1
U(x, 0) =
_
1 1
0 1
__
sin x
0
_
=
_
sinx
0
_
. (14.3)
Equation (14.2) gives traveling wave solutions of the form
v
(1)
(x, t) = F(x + 2t), v
(2)
(x, t) = G(x t).
We can write U in terms of V :
U = V =
_
1 1
0 1
__
v
(1)
v
(2)
_
=
_
1 1
0 1
__
F(x + 2t)
G(x t)
_
=
_
F(x + 2t) +G(x t)
G(x t)
_
.
(14.4)
Partial Dierential Equations Igor Yanovsky, 2005 105
For region I, (14.2) and (14.3) give two initial value problems (since any point in
region I can be traced back along both characteristics to initial conditions):
_
v
(1)
t
2v
(1)
x
= 0,
v
(1)
(x, 0) = sinx;
_
v
(2)
t
+v
(2)
x
= 0,
v
(2)
(x, 0) = 0.
which we solve by characteristics to get traveling wave solutions:
v
(1)
(x, t) = sin(x + 2t), v
(2)
(x, t) = 0.
Thus, for region I,
U = V =
_
1 1
0 1
__
sin(x + 2t)
0
_
=
_
sin(x + 2t)
0
_
.
For region II, solutions of the form F(x+2t) can be traced back to initial conditions.
Thus, v
(1)
is the same as in region I. Solutions of the form G(xt) can be traced back
to the boundary. Since from (14.4), u
(2)
= v
(2)
, we use boundary conditions to get
u
(2)
(0, t) = t = G(t).
Hence, G(x t) = (x t).
Thus, for region II,
U = V =
_
1 1
0 1
__
sin(x + 2t)
(x t)
_
=
_
sin(x + 2t) (x t)
(x t)
_
.
Solutions for regions I and II satisfy (14.1).
Solution for region I satises both initial conditions.
Solution for region II satises given boundary condition.
Partial Dierential Equations Igor Yanovsky, 2005 106
Problem (S02, #7). Consider the system

t
_
u
v
_
=
_
1 2
2 2
_

x
_
u
v
_
. (14.5)
Find an explicit solution for the following mixed problem for the system (14.5):
_
u(x, 0)
v(x, 0)
_
=
_
f(x)
0
_
for x > 0,
u(0, t) = 0 for t > 0.
You may assume that the function f is smooth and vanishes on a neighborhood of x = 0.
Proof. Rewrite the equation as
U
t
+
_
1 2
2 2
_
U
x
= 0,
U(x, 0) =
_
u
(1)
(x, 0)
u
(2)
(x, 0)
_
=
_
f(x)
0
_
.
The eigenvalues of the matrix A are
1
= 3,
2
= 2 and the corresponding eigen-
vectors are e
1
=
_
1
2
_
, e
2
=
_
2
1
_
. Thus,
=
_
3 0
0 2
_
, =
_
1 2
2 1
_
,
1
=
1
det
=
1
5
_
1 2
2 1
_
.
Let U = V . Then,
U
t
+AU
x
= 0,
V
t
+AV
x
= 0,
V
t
+
1
AV
x
= 0,
V
t
+ V
x
= 0.
Thus, the transformed problem is
V
t
+
_
3 0
0 2
_
V
x
= 0, (14.6)
V (x, 0) =
1
U(x, 0) =
1
5
_
1 2
2 1
__
f(x)
0
_
=
f(x)
5
_
1
2
_
. (14.7)
Equation (14.6) gives traveling wave solutions of the form:
v
(1)
(x, t) = F(x + 3t), v
(2)
(x, t) = G(x 2t). (14.8)
We can write U in terms of V :
U = V =
_
1 2
2 1
__
v
(1)
v
(2)
_
=
_
1 2
2 1
__
F(x + 3t)
G(x 2t)
_
=
_
F(x + 3t) 2G(x 2t)
2F(x + 3t) + G(x 2t)
_
.
(14.9)
Partial Dierential Equations Igor Yanovsky, 2005 107
For region I, (14.6) and (14.7) give two initial value problems (since value at any
point in region I can be traced back along both characteristics to initial conditions):
_
v
(1)
t
3v
(1)
x
= 0,
v
(1)
(x, 0) =
1
5
f(x);
_
v
(2)
t
+ 2v
(2)
x
= 0,
v
(2)
(x, 0) =
2
5
f(x).
which we solve by characteristics to get traveling wave solutions:
v
(1)
(x, t) =
1
5
f(x + 3t), v
(2)
(x, t) =
2
5
f(x 2t).
Thus, for region I, U = V =
_
1 2
2 1
__
1
5
f(x + 3t)

2
5
f(x 2t)
_
=
_
1
5
f(x + 3t) +
4
5
f(x 2t)
2
5
f(x + 3t)
2
5
f(x 2t)
_
.
For region II, solutions of the form F(x+3t) can be traced back to initial conditions.
Thus, v
(1)
is the same as in region I. Solutions of the form G(x2t) can be traced back
to the boundary. Since from (14.9),
u
(1)
= v
(1)
2v
(2)
, we have
u
(1)
(x, t) = F(x + 3t) 2G(x 2t) =
1
5
f(x + 3t) 2G(x 2t).
The boundary condition gives
u
(1)
(0, t) = 0 =
1
5
f(3t) 2G(2t),
2G(2t) =
1
5
f(3t),
G(t) =
1
10
f
_

3
2
t
_
,
G(x 2t) =
1
10
f
_

3
2
(x 2t)
_
.
Thus, for region II, U = V =
_
1 2
2 1
__
1
5
f(x + 3t)
1
10
f(
3
2
(x 2t))
_
=
_
1
5
f(x + 3t)
1
5
f(
3
2
(x 2t))
2
5
f(x + 3t) +
1
10
f(
3
2
(x 2t))
_
.
Solutions for regions I and II satisfy (14.5).
Solution for region I satises both initial conditions.
Solution for region II satises given boundary condition.
Partial Dierential Equations Igor Yanovsky, 2005 108
Problem (F94, #1; S97, #7). Solve the initial-boundary value problem
u
t
+ 3v
x
= 0,
v
t
+u
x
+ 2v
x
= 0
in the quarter plane 0 x, t < , with initial conditions
25
u(x, 0) =
1
(x), v(x, 0) =
2
(x), 0 < x < +
and boundary condition
u(0, t) = (t), t > 0.
Proof. Rewrite the equation as U
t
+ AU
x
= 0:
U
t
+
_
0 3
1 2
_
U
x
= 0, (14.10)
U(x, 0) =
_
u
(1)
(x, 0)
u
(2)
(x, 0)
_
=
_

1
(x)

2
(x)
_
.
The eigenvalues of the matrix A are
1
= 1,
2
= 3 and the corresponding eigen-
vectors are e
1
=
_
3
1
_
, e
2
=
_
1
1
_
. Thus,
=
_
1 0
0 3
_
, =
_
3 1
1 1
_
,
1
=
1
det
=
1
4
_
1 1
1 3
_
.
Let U = V . Then,
U
t
+AU
x
= 0,
V
t
+AV
x
= 0,
V
t
+
1
AV
x
= 0,
V
t
+ V
x
= 0.
Thus, the transformed problem is
V
t
+
_
1 0
0 3
_
V
x
= 0, (14.11)
V (x, 0) =
1
U(x, 0) =
1
4
_
1 1
1 3
__

1
(x)

2
(x)
_
=
1
4
_

1
(x) +
2
(x)

1
(x) + 3
2
(x)
_
. (14.12)
Equation (14.11) gives traveling wave solutions of the form:
v
(1)
(x, t) = F(x + t), v
(2)
(x, t) = G(x 3t). (14.13)
We can write U in terms of V :
U = V =
_
3 1
1 1
__
v
(1)
v
(2)
_
=
_
3 1
1 1
__
F(x +t)
G(x 3t)
_
=
_
3F(x + t) +G(x 3t)
F(x + t) +G(x 3t)
_
.
(14.14)
25
In S97, #7, the zero initial conditions are considered.
Partial Dierential Equations Igor Yanovsky, 2005 109
For region I, (14.11) and (14.12) give two initial value problems (since value at
any point in region I can be traced back along characteristics to initial conditions):
_
v
(1)
t
v
(1)
x
= 0,
v
(1)
(x, 0) =
1
4

1
(x) +
1
4

2
(x);
_
v
(2)
t
+ 3v
(2)
x
= 0,
v
(2)
(x, 0) =
1
4

1
(x) +
3
4

2
(x),
which we solve by characteristics to get traveling wave solutions:
v
(1)
(x, t) =
1
4

1
(x +t) +
1
4

2
(x +t), v
(2)
(x, t) =
1
4

1
(x 3t) +
3
4

2
(x 3t).
Thus, for region I,
U = V =
_
3 1
1 1
__

1
4

1
(x + t) +
1
4

2
(x + t)
1
4

1
(x 3t) +
3
4

2
(x 3t)
_
=
1
4
_
3
1
(x + t) 3
2
(x +t) +
1
(x 3t) + 3
2
(x 3t)

1
(x + t) +
2
(x + t) +
1
(x 3t) + 3
2
(x 3t)
_
.
For region II, solutions of the form F(x +t) can be traced back to initial conditions.
Thus, v
(1)
is the same as in region I. Solutions of the form G(x3t) can be traced back
to the boundary. Since from (14.14),
u
(1)
= 3v
(1)
+v
(2)
, we have
u
(1)
(x, t) =
3
4

1
(x + t)
3
4

2
(x + t) +G(x 3t).
The boundary condition gives
u
(1)
(0, t) = (t) =
3
4

1
(t)
3
4

2
(t) + G(3t),
G(3t) = (t)
3
4

1
(t) +
3
4

2
(t),
G(t) =
_

t
3
_

3
4

1
_

t
3
_
+
3
4

2
_

t
3
_
,
G(x 3t) =
_

x 3t
3
_

3
4

1
_

x 3t
3
_
+
3
4

2
_

x 3t
3
_
.
Thus, for region II,
U = V =
_
3 1
1 1
__

1
4

1
(x +t) +
1
4

2
(x +t)
(
x3t
3
)
3
4

1
(
x3t
3
) +
3
4

2
(
x3t
3
)
_
=
_
3
4

1
(x +t)
3
4

2
(x +t) +(
x3t
3
)
3
4

1
(
x3t
3
) +
3
4

2
(
x3t
3
)

1
4

1
(x +t) +
1
4

2
(x +t) + (
x3t
3
)
3
4

1
(
x3t
3
) +
3
4

2
(
x3t
3
)
_
.
Partial Dierential Equations Igor Yanovsky, 2005 110
Solutions for regions I and II satisfy (14.10).
Solution for region I satises both initial conditions.
Solution for region II satises given boundary condition.
Partial Dierential Equations Igor Yanovsky, 2005 111
Problem (F91, #1). Solve explicitly the following initial-boundary value problem for
linear 22 hyperbolic system
u
t
= u
x
+ v
x
v
t
= 3u
x
v
x
,
where 0 < t < +, 0 < x < + with initial conditions
u(x, 0) = u
0
(x), v(x, 0) = v
0
(x), 0 < x < +,
and the boundary condition
u(0, t) +bv(0, t) = (t), 0 < t < +,
where b ,=
1
3
is a constant.
What happens when b =
1
3
?
Proof. Let us change the notation (u u
(1)
, v u
(2)
). Rewrite the equation as
U
t
+
_
1 1
3 1
_
U
x
= 0, (14.15)
U(x, 0) =
_
u
(1)
(x, 0)
u
(2)
(x, 0)
_
=
_
u
(1)
0
(x)
u
(2)
0
(x)
_
.
The eigenvalues of the matrix A are
1
= 2,
2
= 2 and the corresponding eigen-
vectors are e
1
=
_
1
1
_
, e
2
=
_
1
3
_
. Thus,
=
_
2 0
0 2
_
, =
_
1 1
1 3
_
,
1
=
1
4
_
3 1
1 1
_
.
Let U = V . Then,
U
t
+AU
x
= 0,
V
t
+AV
x
= 0,
V
t
+
1
AV
x
= 0,
V
t
+ V
x
= 0.
Thus, the transformed problem is
V
t
+
_
2 0
0 2
_
V
x
= 0, (14.16)
V (x, 0) =
1
U(x, 0) =
1
4
_
3 1
1 1
__
u
(1)
(x, 0)
u
(2)
(x, 0)
_
=
1
4
_
3u
(1)
0
(x) +u
(2)
0
(x)
u
(1)
0
(x) u
(2)
0
(x)
_
.
(14.17)
Equation (14.16) gives traveling wave solutions of the form:
v
(1)
(x, t) = F(x + 2t), v
(2)
(x, t) = G(x 2t). (14.18)
Partial Dierential Equations Igor Yanovsky, 2005 112
We can write U in terms of V :
U = V =
_
1 1
1 3
__
v
(1)
v
(2)
_
=
_
1 1
1 3
__
F(x + 2t)
G(x 2t)
_
=
_
F(x + 2t) +G(x 2t)
F(x + 2t) 3G(x 2t)
_
.
(14.19)
For region I, (14.16) and (14.17) give two initial value problems (since value at any
point in region I can be traced back along characteristics to initial conditions):
_
v
(1)
t
2v
(1)
x
= 0,
v
(1)
(x, 0) =
3
4
u
(1)
0
(x) +
1
4
u
(2)
0
(x);
_
v
(2)
t
+ 2v
(2)
x
= 0,
v
(2)
(x, 0) =
1
4
u
(1)
0
(x)
1
4
u
(2)
0
(x),
which we solve by characteristics to get traveling wave solutions:
v
(1)
(x, t) =
3
4
u
(1)
0
(x + 2t) +
1
4
u
(2)
0
(x + 2t); v
(2)
(x, t) =
1
4
u
(1)
0
(x 2t)
1
4
u
(2)
0
(x 2t).
Thus, for region I,
U = V =
_
1 1
1 3
_
_
3
4
u
(1)
0
(x + 2t) +
1
4
u
(2)
0
(x + 2t)
1
4
u
(1)
0
(x 2t)
1
4
u
(2)
0
(x 2t)
_
=
_
3
4
u
(1)
0
(x + 2t) +
1
4
u
(2)
0
(x + 2t) +
1
4
u
(1)
0
(x 2t)
1
4
u
(2)
0
(x 2t)
3
4
u
(1)
0
(x + 2t) +
1
4
u
(2)
0
(x + 2t)
3
4
u
(1)
0
(x 2t) +
3
4
u
(2)
0
(x 2t)
_
.
For region II, solutions of the form F(x+2t) can be traced back to initial conditions.
Thus, v
(1)
is the same as in region I. Solutions of the form G(x2t) can be traced back
to the boundary. The boundary condition gives
u
(1)
(0, t) +bu
(2)
(0, t) = (t).
Using (14.19),
v
(1)
(0, t) + G(2t) + bv
(1)
(0, t) 3bG(2t) = (t),
(1 + b)v
(1)
(0, t) + (1 3b)G(2t) = (t),
(1 + b)
_
3
4
u
(1)
0
(2t) +
1
4
u
(2)
0
(2t)
_
+ (1 3b)G(2t) = (t),
G(2t) =
(t) (1 + b)
_
3
4
u
(1)
0
(2t) +
1
4
u
(2)
0
(2t)
_
1 3b
,
G(t) =
(
t
2
) (1 +b)
_
3
4
u
(1)
0
(t) +
1
4
u
(2)
0
(t)
_
1 3b
,
G(x 2t) =
(
x2t
2
) (1 + b)
_
3
4
u
(1)
0
((x 2t)) +
1
4
u
(2)
0
((x 2t))
_
1 3b
.
Thus, for region II,
U = V =
_
1 1
1 3
_

3
4
u
(1)
0
(x + 2t) +
1
4
u
(2)
0
(x + 2t)
(
x2t
2
)(1+b)
_
3
4
u
(1)
0
((x2t))+
1
4
u
(2)
0
((x2t))
_
13b

3
4
u
(1)
0
(x + 2t) +
1
4
u
(2)
0
(x + 2t) +
(
x2t
2
)(1+b)
_
3
4
u
(1)
0
((x2t))+
1
4
u
(2)
0
((x2t))
_
13b
3
4
u
(1)
0
(x + 2t) +
1
4
u
(2)
0
(x + 2t)
3(
x2t
2
)3(1+b)
_
3
4
u
(1)
0
((x2t))+
1
4
u
(2)
0
((x2t))
_
13b

.
The following were performed, but are arithmetically complicated:
Solutions for regions I and II satisfy (14.15).
Partial Dierential Equations Igor Yanovsky, 2005 113
Solution for region I satises both initial conditions.
Solution for region II satises given boundary condition.
If b =
1
3
, u
(1)
(0, t) +
1
3
u
(2)
(0, t) = F(2t) +G(2t) +
1
3
F(2t) G(2t) =
4
3
F(2t) = (t).
Thus, the solutions of the form v
(2)
= G(x 2t) are not dened at x = 0, which leads
to ill-posedness.
Partial Dierential Equations Igor Yanovsky, 2005 114
Problem (F96, #8). Consider the system
u
t
= 3u
x
+ 2v
x
v
t
= v
x
v
in the region x 0, t 0. Which of the following sets of initial and boundary data
make this a well-posed problem?
a) u(x, 0) = 0, x 0
v(x, 0) = x
2
, x 0
v(0, t) = t
2
, t 0.
b) u(x, 0) = 0, x 0
v(x, 0) = x
2
, x 0
u(0, t) = t, t 0.
c) u(x, 0) = 0, x 0
v(x, 0) = x
2
, x 0
u(0, t) = t, t 0
v(0, t) = t
2
, t 0.
Proof. Rewrite the equation as U
t
+ AU
x
= BU. Initial conditions are same for
(a),(b),(c):
U
t
+
_
3 2
0 1
_
U
x
=
_
0 0
0 1
_
U,
U(x, 0) =
_
u
(1)
(x, 0)
u
(2)
(x, 0)
_
=
_
0
x
2
_
.
The eigenvalues of the matrix A are
1
= 3,
2
= 1, and the corresponding eigen-
vectors are e
1
=
_
1
0
_
, e
2
=
_
1
2
_
. Thus,
=
_
3 0
0 1
_
, =
_
1 1
0 2
_
,
1
=
1
2
_
2 1
0 1
_
.
Let U = V . Then,
U
t
+AU
x
= BU,
V
t
+AV
x
= BV,
V
t
+
1
AV
x
=
1
BV,
V
t
+ V
x
=
1
BV.
Thus, the transformed problem is
V
t
+
_
3 0
0 1
_
V
x
=
_
0 1
0 1
_
V, (14.20)
V (x, 0) =
1
U(x, 0) =
1
2
_
2 1
0 1
__
0
x
2
_
=
x
2
2
_
1
1
_
. (14.21)
Equation (14.20) gives traveling wave solutions of the form
v
(1)
(x, t) = F(x + 3t), v
(2)
(x, t) = G(x t). (14.22)
Partial Dierential Equations Igor Yanovsky, 2005 115
We can write U in terms of V :
U = V =
_
1 1
0 2
__
v
(1)
v
(2)
_
=
_
1 1
0 2
__
F(x + 3t)
G(x t)
_
=
_
F(x + 3t) + G(x t)
2G(x t)
_
.
(14.23)
For region I, (14.20) and (14.21) give two initial value problems (since a value at any
point in region I can be traced back along both characteristics to initial conditions):
_
v
(1)
t
3v
(1)
x
= v
(2)
,
v
(1)
(x, 0) =
x
2
2
;
_
v
(2)
t
+v
(2)
x
= v
(2)
,
v
(2)
(x, 0) =
x
2
2
,
which we do not solve here. Thus, initial conditions for v
(1)
and v
(2)
have to be dened.
Since (14.23) denes u
(1)
and u
(2)
in terms of v
(1)
and v
(2)
, we need to dene two initial
conditions for U.
For region II, solutions of the form F(x+3t) can be traced back to initial conditions.
Thus, v
(1)
is the same as in region I. Solutions of the form G(x t) are traced back to
the boundary at x = 0. Since from (14.23), u
(2)
(x, t) = 2v
(2)
(x, t) = 2G(x t), i.e.
u
(2)
is written in term of v
(2)
only, u
(2)
requires a boundary condition to be dened on
x = 0.
Thus,
a) u
(2)
(0, t) = t
2
, t 0. Well-posed.
b) u
(1)
(0, t) = t, t 0. Not well-posed.
c) u
(1)
(0, t) = t, u
(2)
(0, t) = t
2
, t 0. Not well-posed.
Partial Dierential Equations Igor Yanovsky, 2005 116
Problem (F02, #3). Consider the rst order system
u
t
+u
x
+ v
x
= 0
v
t
+u
x
v
x
= 0
on the domain 0 < t < and 0 < x < 1. Which of the following sets of initial-
a) u(x,0) = f(x), v(x,0) = g(x);
b) u(x,0) = f(x), v(x,0) = g(x), u(0,t) = h(x), v(0,t) = k(x);
c) u(x,0) = f(x), v(x,0) = g(x), u(0,t) = h(x), v(1,t) = k(x).
Proof. Rewrite the equation as U
t
+AU
x
= 0. Initial conditions are same for (a),(b),(c):
U
t
+
_
1 1
1 1
_
U
x
= 0,
U(x, 0) =
_
u
(1)
(x, 0)
u
(2)
(x, 0)
_
=
_
f(x)
g(x)
_
.
The eigenvalues of the matrix A are
1
=

2,
2
=

## 2 and the corresponding

eigenvectors are e
1
=
_
1
1 +

2
_
, e
2
=
_
1
1

2
_
. Thus,
=
_
2 0
0

2
_
, =
_
1 1
1 +

2 1

2
_
,
1
=
1
2

2
_
1 +

2 1
1 +

2 1
_
.
Let U = V . Then,
U
t
+AU
x
= 0,
V
t
+AV
x
= 0,
V
t
+
1
AV
x
= 0,
V
t
+ V
x
= 0.
Thus, the transformed problem is
V
t
+
_
2 0
0

2
_
V
x
= 0, (14.24)
V (x, 0) =
1
U(x, 0) =
1
2

2
_
1 +

2 1
1 +

2 1
__
f(x)
g(x)
_
=
1
2

2
_
(1 +

2)f(x) +g(x)
(1 +

2)f(x) g(x)
_
.
(14.25)
Equation (14.24) gives traveling wave solutions of the form:
v
(1)
(x, t) = F(x

2t), v
(2)
(x, t) = G(x +

2t). (14.26)
However, we can continue and obtain the solutions. We have two initial value problems
_
v
(1)
t
+

2v
(1)
x
= 0,
v
(1)
(x, 0) =
(1+

2)
2

2
f(x) +
1
2

2
g(x);
_
v
(2)
t

2v
(2)
x
= 0,
v
(2)
(x, 0) =
(1+

2)
2

2
f(x)
1
2

2
g(x),
which we solve by characteristics to get traveling wave solutions:
v
(1)
(x, t) =
(1 +

2)
2

2
f(x

2t) +
1
2

2
g(x

2t),
v
(2)
(x, t) =
(1 +

2)
2

2
f(x +

2t)
1
2

2
g(x +

2t).
Partial Dierential Equations Igor Yanovsky, 2005 117
We can obtain general solution U by writing U in terms of V :
U = V =
_
v
(1)
v
(2)
_
=
_
1 1
1 +

2 1

2
_
1
2

2
_
(1 +

2)f(x

2t) + g(x

2t)
(1 +

2)f(x +

2t) g(x +

2t)
_
.
(14.27)
In region I, the solution is obtained by solving two initial value problems(since a
value at any point in region I can be traced back along both characteristics to initial
conditions).
In region II, the solutions of the form v
(2)
= G(x+

## 2t) can be traced back to initial

conditions and those of the form v
(1)
= F(x

## 2t), to left boundary. Since by (14.27),

u
(1)
and u
(2)
are written in terms of both v
(1)
and v
(2)
, one initial condition and one
boundary condition at x = 0 need to be prescribed.
In region III, the solutions of the form v
(2)
= G(x +

## 2t) can be traced back to

right boundary and those of the form v
(1)
= F(x

## 2t), to initial condition. Since by

(14.27), u
(1)
and u
(2)
are written in terms of both v
(1)
and v
(2)
, one initial condition
and one boundary condition at x = 1 need to be prescribed.
To obtain the solution for region IV, two boundary conditions, one for each bound-
ary, should be given.
Thus,
a) No boundary conditions. Not well-posed.
b) u
(1)
(0, t) = h(x), u
(2)
(0, t) = k(x). Not well-posed.
c) u
(1)
(0, t) = h(x), u
(2)
(1, t) = k(x). Well-posed.
Partial Dierential Equations Igor Yanovsky, 2005 118
Problem (S94, #3). Consider the system of equations
f
t
+g
x
= 0
g
t
+ f
x
= 0
h
t
+ 2h
x
= 0
on the set x 0, t 0, with the following initial-boundary values:
a) f, g, h prescribed on t = 0, x 0; f, h prescribed on x = 0, t 0.
b) f, g, h prescribed on t = 0, x 0; f g, h prescribed on x = 0, t 0.
c) f + g, h prescribed on t = 0, x 0; f, g, h prescribed on x = 0, t 0.
For each of these 3 sets of data, determine whether or not the system is well-posed.
Proof. The third equation is decoupled from the rst two and can be considered sepa-
rately. Its solution can be written in the form
h(x, t) = H(x 2t),
and therefore, h must be prescribed on t = 0 and on x = 0, since the characteristics
propagate from both the x and t axis.
We rewrite the rst two equations as (f u
1
, g u
2
):
U
t
+
_
0 1
1 0
_
U
x
= 0,
U(x, 0) =
_
u
(1)
(x, 0)
u
(2)
(x, 0)
_
.
The eigenvalues of the matrix A are
1
= 1,
2
= 1 and the corresponding eigen-
vectors are e
1
=
_
1
1
_
, e
2
=
_
1
1
_
. Thus,
=
_
1 0
0 1
_
, =
_
1 1
1 1
_
,
1
=
1
2
_
1 1
1 1
_
.
Let U = V . Then,
U
t
+AU
x
= 0,
V
t
+AV
x
= 0,
V
t
+
1
AV
x
= 0,
V
t
+ V
x
= 0.
Thus, the transformed problem is
V
t
+
_
1 0
0 1
_
V
x
= 0, (14.28)
V (x, 0) =
1
U(x, 0) =
1
2
_
1 1
1 1
__
u
(1)
(x, 0)
u
(1)
(x, 0)
_
. (14.29)
Equation (14.28) gives traveling wave solutions of the form:
v
(1)
(x, t) = F(x + t), v
(2)
(x, t) = G(x t). (14.30)
Partial Dierential Equations Igor Yanovsky, 2005 119
We can write U in terms of V :
U = V =
_
1 1
1 1
__
v
(1)
v
(2)
_
=
_
1 1
1 1
__
F(x + t)
G(x t)
_
=
_
F(x +t) + G(x t)
F(x + t) +G(x t)
_
.
(14.31)
For region I, (14.28) and (14.29) give two initial value problems (since a value at any
point in region I can be traced back along both characteristics to initial conditions).
Thus, initial conditions for v
(1)
and v
(2)
have to be dened. Since (14.31) denes u
(1)
and u
(2)
in terms of v
(1)
and v
(2)
, we need to dene two initial conditions for U.
For region II, solutions of the form F(x +t) can be traced back to initial conditions.
Thus, v
(1)
is the same as in region I. Solutions of the form G(x t) are traced back
to the boundary at x = 0. Since from (14.31), u
(2)
(x, t) = v
(1)
(x, t) + v
(2)
(x, t) =
F(x + t) + G(x t), i.e. u
(2)
is written in terms of v
(2)
= G(x t), u
(2)
requires a
boundary condition to be dened on x = 0.
a) u
(1)
, u
(2)
prescribed on t = 0; u
(1)
prescribed on x = 0.
Since u
(1)
(x, t) = F(x + t) +G(x t),
u
(2)
(x, t) = F(x + t) +G(x t), i.e. both
u
(1)
and u
(2)
are written in terms of F(x +t)
and G(x t), we need to dene two initial
conditions for U (on t = 0).
A boundary condition also needs to be prescribed
on x = 0 to be able to trace back v
(2)
= G(x t).
Well-posed.
b) u
(1)
, u
(2)
prescribed on t = 0; u
(1)
u
(2)
prescribed on x = 0.
As in part (a), we need to dene two initial
conditions for U.
Since u
(1)
u
(2)
= 2F(x +t), its denition
on x = 0 leads to ill-posedness. On the
contrary, u
(1)
+ u
(2)
= 2G(x t) should be
dened on x = 0 in order to be able to trace
back the values through characteristics.
Ill-posed.
c) u
(1)
+ u
(2)
prescribed on t = 0; u
(1)
, u
(2)
prescribed on x = 0.
Since u
(1)
+u
(2)
= 2G(x t), another initial
condition should be prescribed to be able to
trace back solutions of the form v
(2)
= F(x +t),
without which the problem is ill-posed.
Also, two boundary conditions for both u
(1)
and u
(2)
dene solutions of both v
(1)
= G(x t)
and v
(2)
= F(x + t) on the boundary. The former
Ill-posed.
Partial Dierential Equations Igor Yanovsky, 2005 120
Problem (F92, #8). Consider the system
u
t
+u
x
+ av
x
= 0
v
t
+ bu
x
+ v
x
= 0
for 0 < x < 1 with boundary and initial conditions
u = v = 0 for x = 0
u = u
0
, v = v
0
for t = 0.
a) For which values of a and b is this a well-posed problem?
b) For this class of a, b, state conditions on u
0
and v
0
so that the solution u, v will be
continuous and continuously dierentiable.
Proof. a) Let us change the notation (u u
(1)
, v u
(2)
). Rewrite the equation as
U
t
+
_
1 a
b 1
_
U
x
= 0, (14.32)
U(x, 0) =
_
u
(1)
(x, 0)
u
(2)
(x, 0)
_
=
_
u
(1)
0
(x)
u
(2)
0
(x)
_
,
U(0, t) =
_
u
(1)
(0, t)
u
(2)
(0, t)
_
= 0.
The eigenvalues of the matrix A are
1
= 1

ab,
2
= 1 +

ab.
=
_
1

ab 0
0 1 +

ab
_
.
Let U = V , where is a matrix of eigenvectors. Then,
U
t
+AU
x
= 0,
V
t
+AV
x
= 0,
V
t
+
1
AV
x
= 0,
V
t
+ V
x
= 0.
Thus, the transformed problem is
V
t
+
_
1

ab 0
0 1 +

ab
_
V
x
= 0, (14.33)
V (x, 0) =
1
U(x, 0).
The equation (14.33) gives traveling wave solutions of the form:
v
(1)
(x, t) = F(x (1

ab)t), v
(2)
(x, t) = G(x (1 +

ab)t). (14.34)
We also have U = V , i.e. both u
(1)
and u
(2)
(and their initial and boundary conditions)
are combinations of v
(1)
and v
(2)
.
In order for this problem to be well-posed, both sets of characteristics should emanate
from the boundary at x = 0. Thus, the eigenvalues of the system are real (ab > 0) and

1,2
> 0 (ab < 1). Thus,
0 < ab < 1.
b) For U to be C
1
, we require the compatibility condition, u
(1)
0
(0) = 0, u
(2)
0
(0) =
0.
Partial Dierential Equations Igor Yanovsky, 2005 121
Problem (F93, #2). Consider the initial-boundary value problem
u
t
+u
x
= 0
v
t
(1 cx
2
)v
x
+ u
x
= 0
on 1 x 1 and 0 t, with the following prescribed data:
u(x, 0), v(x, 0),
u(1, t), v(1, t).
For which values of c is this a well-posed problem?
Proof. Let us change the notation (u u
(1)
, v u
(2)
).
The rst equation can be solved with u
(1)
(x, 0) = F(x) to get a solution in the form
u
(1)
(x, t) = F(x t), which requires u
(1)
(x, 0) and u
(1)
(1, t) to be dened.
With u
(1)
known, we can solve the second equation
u
(2)
t
(1 cx
2
)u
(2)
x
+F(x t) = 0.
Solving the equation by characteristics, we obtain
the characteristics in the xt-plane are of the form
dx
dt
= cx
2
1.
We need to determine c such that the prescribed
data u
(2)
(x, 0) and u
(2)
(1, t) makes the problem to
be well-posed. The boundary condition for u
(2)
(1, t)
requires the characteristics to propagate to the
left with t increasing. Thus, x(t) is a decreasing
function, i.e.
dx
dt
< 0 cx
2
1 < 0 for 1 < x < 1 c < 1.
We could also do similar analysis we have done in other problems on rst order sys-
tems involving nding eigenvalues/eigenvectors of the system and using the fact that
u
(1)
(x, t) is known at both boundaries (i.e. values of u
(1)
(1, t) can be traced back either
to initial conditions or to boundary conditions on x = 1).
Partial Dierential Equations Igor Yanovsky, 2005 122
Problem (S91, #4). Consider the rst order system
u
t
+au
x
+ bv
x
= 0
v
t
+cu
x
+dv
x
= 0
for 0 < x < 1, with prescribed initial data:
u(x, 0) = u
0
(x)
v(x, 0) = v
0
(x).
a) Find conditions on a, b, c, d such that there is a full set of characteristics and, in
this case, nd the characteristic speeds.
b) For which values of a, b, c, d can boundary data be prescribed on x = 0 and for which
values can it be prescribed on x = 1? How many pieces of data can be prescribed on
each boundary?
Proof. a) Let us change the notation (u u
(1)
, v u
(2)
). Rewrite the equation as
U
t
+
_
a b
c d
_
U
x
= 0, (14.35)
U(x, 0) =
_
u
(1)
(x, 0)
u
(2)
(x, 0)
_
=
_
u
(1)
0
(x)
u
(2)
0
(x)
_
.
The system is hyperbolic if for each value of u
(1)
and u
(2)
the eigenvalues are real
and the matrix is diagonalizable, i.e. there is a complete set of linearly independent
eigenvectors. The eigenvalues of the matrix A are

1,2
=
a + d
_
(a +d)
2
2
=
a + d
_
(a d)
2
+ 4bc
2
.
We need (a d)
2
+ 4bc > 0. This also makes the problem to be diagonalizable.
Let U = V , where is a matrix of eigenvectors. Then,
U
t
+AU
x
= 0,
V
t
+AV
x
= 0,
V
t
+
1
AV
x
= 0,
V
t
+ V
x
= 0.
Thus, the transformed problem is
V
t
+
_

1
0
0
2
_
V
x
= 0, (14.36)
Equation (14.36) gives traveling wave solutions of the form:
v
(1)
(x, t) = F(x
1
t), v
(2)
(x, t) = G(x
2
t). (14.37)
The characteristic speeds are
dx
dt
=
1
,
dx
dt
=
2
.
b) We assume (a +d)
2
a + d > 0, ad bc > 0
1
,
2
> 0 2 B.C. on x = 0.
a +d > 0, ad bc < 0
1
< 0,
2
> 0 1 B.C. on x = 0, 1 B.C. on x = 1.
a + d < 0, ad bc > 0
1
,
2
< 0 2 B.C. on x = 1.
Partial Dierential Equations Igor Yanovsky, 2005 123
a +d < 0, ad bc < 0
1
< 0,
2
> 0 1 B.C. on x = 0, 1 B.C. on x = 1.
a + d > 0, ad bc = 0
1
= 0,
2
> 0 1 B.C. on x = 0.
a + d < 0, ad bc = 0
1
= 0,
2
< 0 1 B.C. on x = 1.
a + d = 0, ad bc < 0
1
< 0,
2
> 0 1 B.C. on x = 0, 1 B.C. on
x = 1.
Partial Dierential Equations Igor Yanovsky, 2005 124
Problem (S94, #2). Consider the dierential operator
L
_
u
v
_
=
_
u
t
+ 9v
x
u
xx
v
t
u
x
v
xx
_
on 0 x 2, t 0, in which the vector
_
u(x, t)
v(x, t)
_
consists of two functions that
are periodic in x.
a) Find the eigenfunctions and eigenvalues of the operator L.
b) Use the results of (a) to solve the initial value problem
L
_
u
v
_
= 0 for t 0,
_
u
v
_
=
_
e
ix
0
_
for t = 0.
Proof. a) We nd the space eigenvalues and eigenfunctions. We rewrite the system
as
U
t
+
_
0 9
1 0
_
U
x
+
_
1 0
0 1
_
U
xx
= 0,
and nd eigenvalues
_
0 9
1 0
_
U
x
+
_
1 0
0 1
_
U
xx
= U. (14.38)
Set U =
_
u(x, t)
v(x, t)
_
=
_
n=
n=
u
n
(t)e
inx

n=
n=
v
n
(t)e
inx
_
. Plugging this into (14.38), we get
_
0 9
1 0
__
inu
n
(t)e
inx

inv
n
(t)e
inx
_
+
_
1 0
0 1
__
n
2
u
n
(t)e
inx

n
2
v
n
(t)e
inx
_
=
_
u
n
(t)e
inx

v
n
(t)e
inx
_
,
_
0 9
1 0
__
inu
n
(t)
inv
n
(t)
_
+
_
1 0
0 1
__
n
2
u
n
(t)
n
2
v
n
(t)
_
=
_
u
n
(t)
v
n
(t)
_
,
_
0 9in
in 0
__
u
n
(t)
v
n
(t)
_
+
_
n
2
0
0 n
2
__
u
n
(t)
v
n
(t)
_
=
_
u
n
(t)
v
n
(t)
_
,
_
n
2
9in
in n
2

__
u
n
(t)
v
n
(t)
_
= 0,
(n
2
)
2
9n
2
= 0,
which gives
1
= n
2
+3n,
2
= n
2
3n, are eigenvalues, and v
1
=
_
3i
1
_
, v
2
=
_
3i
1
_
,
are corresponding eigenvectors.
Partial Dierential Equations Igor Yanovsky, 2005 125
b) We want to solve
_
u
v
_
t
+ L
_
u
v
_
= 0, L
_
u
v
_
=
_
9v
x
u
xx
u
x
v
xx
_
. We have
_
u
v
_
t
= L
_
u
v
_
=
_
u
v
_
, i.e.
_
u
v
_
= e
t
. We can write the solution as
U(x, t) =
_
u
n
(t)e
inx

v
n
(t)e
inx
_
=

n=
a
n
e

1
t
v
1
e
inx
+ b
n
e

2
t
v
2
e
inx
=

n=
a
n
e
(n
2
+3n)t
_
3i
1
_
e
inx
+b
n
e
(n
2
3n)t
_
3i
1
_
e
inx
.
U(x, 0) =

n=
a
n
_
3i
1
_
e
inx
+b
n
_
3i
1
_
e
inx
=
_
e
ix
0
_
,
a
n
= b
n
= 0, n ,= 1;
a
1
+ b
1
=
1
3i
and a
1
= b
1
a
1
= b
1
=
1
6i
.
U(x, t) =
1
6i
e
4t
_
3i
1
_
e
ix
+
1
6i
e
2t
_
3i
1
_
e
ix
=
_
1
2
(e
4t
+ e
2t
)
1
6i
(e
4t
e
2t
)
_
e
ix
.
26 27
26
ChiuYens and Sung-Has solutions give similar answers.
27
1. Needed to nd eigenfunctions, not eigenvectors.
2. The notation of L was changed. The problem statement incorporates the derivatives wrt. t into L.
3. Why can we write the solution in this form above?
Partial Dierential Equations Igor Yanovsky, 2005 126
Problem (W04, #6). Consider the rst order system
u
t
u
x
= v
t
+v
x
= 0
in the diamond shaped region 1 < x + t < 1, 1 < x t < 1. For each of
the following boundary value problems state whether this problem is well-posed. If it is
well-posed, nd the solution.
a) u(x +t) = u
0
(x +t) on x t = 1, v(x t) = v
0
(x t) on x +t = 1.
b) v(x +t) = v
0
(x +t) on x t = 1, u(x t) = u
0
(x t) on x + t = 1.
Proof. We have
u
t
u
x
= 0,
v
t
+ v
x
= 0.
u is constant along the characteristics: x + t = c
1
(s).
Thus, its solution is u(x, t) = u
0
(x +t).
It the initial condition is prescribed at x t = 1,
the solution can be determined in the entire region
by tracing back through the characteristics.
v is constant along the characteristics: x t = c
2
(s).
Thus, its solution is v(x, t) = v
0
(x t).
It the initial condition is prescribed at x +t = 1,
the solution can be determined in the entire region
by tracing forward through the characteristics.
Partial Dierential Equations Igor Yanovsky, 2005 127
15 Problems: Gas Dynamics Systems
15.1 Perturbation
Problem (S92, #3).
28 29
Consider the gas dynamic equations
u
t
+uu
x
+ (F())
x
= 0,

t
+ (u)
x
= 0.
Here F() is a given C

## -smooth function of . At t = 0, 2-periodic initial data

u(x, 0) = f(x), (x, 0) = g(x).
a) Assume that
f(x) = U
0
+f
1
(x), g(x) = R
0
+g
1
(x)
where U
0
, R
0
> 0 are constants and f
1
(x), g
1
(x) are small perturbations. Lin-
earize the equations and given conditions for F such that the linearized problem is
well-posed.
b) Assume that U
0
> 0 and consider the above linearized equations for 0 x 1,
t 0. Construct boundary conditions such that the initial-boundary value problem is
well-posed.
Proof. a) We write the equations in characteristic form:
u
t
+uu
x
+ F

()
x
= 0,

t
+ u
x
+ u
x
= 0.
Consider the special case of nearly constant initial data
u(x, 0) = u
0
+ u
1
(x, 0),
(x, 0) =
0
+
1
(x, 0).
Then we can approximate nonlinear equations by linear equations. Assuming
u(x, t) = u
0
+ u
1
(x, t),
(x, t) =
0
+
1
(x, t)
remain valid with u
1
= O(1),
1
= O(1), we nd that
u
t
= u
1t
,
t
=
1t
,
u
x
= u
1x
,
x
=
1x
,
F

() = F

(
0
+
1
(x, t)) = F

(
0
) +
1
F

(
0
) +O(
2
).
Plugging these into , gives
u
1t
+ (u
0
+ u
1
)u
1x
+
_
F

(
0
) +
1
F

(
0
) +O(
2
)
_

1x
= 0,

1t
+u
1x
(
0
+
1
) + (u
0
+u
1
)
1x
= 0.
Dividing by gives
u
1t
+ u
0
u
1x
+ F

(
0
)
1x
= u
1
u
1x

1

1x
F

(
0
) + O(
2
),

1t
+u
1x

0
+ u
0

1x
= u
1x

1
u
1

1x
.
28
See LeVeque, Second Edition, Birkh auser Verlag, 1992, p. 44.
29
This problem has similar notation with S92, #4.
Partial Dierential Equations Igor Yanovsky, 2005 128
For small , we have
_
u
1t
+ u
0
u
1x
+ F

(
0
)
1x
= 0,

1t
+u
1x

0
+u
0

1x
= 0.
This can be written as
_
u
1

1
_
t
+
_
u
0
F

(
0
)

0
u
0
__
u
1

1
_
x
=
_
0
0
_
.

u
0
F

(
0
)

0
u
0

= (u
0
)(u
0
)
0
F

(
0
) = 0,

2
2u
0
+ u
2
0

0
F

(
0
) = 0,

1,2
= u
0

0
F

(
0
), u
0
> 0,
0
> 0.
For well-posedness, need
1,2
R or F

(
0
) 0.
b) We have u
0
> 0, and
1
= u
0
+
_

0
F

(
0
),
2
= u
0

0
F

(
0
).
If u
0
>
_

0
F

(
0
)
1
> 0,
2
> 0 2 BC at x = 0.
If u
0
=
_

0
F

(
0
)
1
> 0,
2
= 0 1 BC at x = 0.
If 0 < u
0
<
_

0
F

(
0
)
1
> 0,
2
< 0 1 BC at x = 0, 1 BC at x = 1.
15.2 Stationary Solutions
Problem (S92, #4).
30
Consider
u
t
+uu
x
+
x
= u
xx
,

t
+ (u)
x
= 0
for t 0, < x < .
Give conditions for the states U
+
, U

, R
+
, R

## , such that the system has

stationary solutions (i.e. u
t
=
t
= 0) satisfying
lim
x+
_
u

_
=
_
U
+
R
+
_
, lim
x
_
u

_
=
_
U

_
.
Proof. For stationary solutions, we need
u
t
=
_
u
2
2
_
x

x
+u
xx
= 0,

t
= (u)
x
= 0.
Integrating the above equations, we obtain

u
2
2
+u
x
= C
1
,
u = C
2
.
30
This problem has similar notation with S92, #3.
Partial Dierential Equations Igor Yanovsky, 2005 129
Conditions give u
x
= 0 at x = . Thus
U
2
+
2
+ R
+
=
U
2

2
+R

,
U
+
R
+
= U

.
Partial Dierential Equations Igor Yanovsky, 2005 130
15.3 Periodic Solutions
Problem (F94, #4). Let u(x, t) be a solution of the Cauchy problem
u
t
= u
xxxx
2u
xx
, < x < +, 0 < t < +,
u(x, 0) = (x),
where u(x, t) and (x) are C

## functions periodic in x with period 2;

i.e. u(x + 2, t) = u(x, t), x, t.
Prove that
[[u(, t)[[ Ce
at
[[[[
where [[u(, t)[[ =
_
_
2
0
[u(x, t)[
2
dx, [[[[ =
_
_
2
0
[(x)[
2
dx, C, a are some constants.
Proof. METHOD I: Since u is 2-periodic, let
u(x, t) =

n=
a
n
(t)e
inx
.
Plugging this into the equation, we get

n=
a

n
(t)e
inx
=

n=
n
4
a
n
(t)e
inx
+ 2

n=
n
2
a
n
(t)e
inx
,
a

n
(t) = (n
4
+ 2n
2
)a
n
(t),
a
n
(t) = a
n
(0)e
(n
4
+2n
2
)t
.
Also, initial condition gives
u(x, 0) =

n=
a
n
(0)e
inx
= (x),

n=
a
n
(0)e
inx

= [(x)[.
[[u(x, t)[[
2
2
=
_
2
0
u
2
(x, t) dx =
_
2
0
_

n=
a
n
(t)e
inx
__

m=
a
n
(t)e
imx
_
dx
=

n=
a
2
n
(t)
_
2
0
e
inx
e
inx
dx = 2

n=
a
2
n
(t) = 2

n=
a
2
n
(0)e
2(n
4
+2n
2
)t

n=
a
2
n
(0)

n=
e
2(n
4
+2n
2
)t

= 2

n=
a
2
n
(0)

. .
||||
2
e
2t

n=
e
2(n
2
1)
2
t
. .
=C
1
, (convergent)
= C
2
e
2t
[[[[
2
.
[[u(x, t)[[ Ce
t
[[[[.
Partial Dierential Equations Igor Yanovsky, 2005 131
METHOD II: Multiply this equation by u and integrate
uu
t
= uu
xxxx
2uu
xx
,
1
2
d
dt
(u
2
) = uu
xxxx
2uu
xx
,
1
2
d
dt
_
2
0
u
2
dx =
_
2
0
uu
xxxx
dx
_
2
0
2uu
xx
dx,
1
2
d
dt
[[u[[
2
2
= uu
xxx

2
0
. .
=0
+u
x
u
xx

2
0
. .
=0

_
2
0
u
2
xx
dx
_
2
0
2uu
xx
dx,
1
2
d
dt
[[u[[
2
2
=
_
2
0
u
2
xx
dx
_
2
0
2uu
xx
dx (2ab a
2
+ b
2
)

_
2
0
u
2
xx
dx +
_
2
0
(u
2
+ u
2
xx
) dx =
_
2
0
u
2
dx = [[u[[
2
,

d
dt
[[u[[
2
2[[u[[
2
,
[[u[[
2
[[u(0)[[
2
e
2t
,
[[u[[ [[u(0)[[e
t
.
METHOD III: Can use Fourier transform. See ChiuYens solutions, that have both
Method II and III.
Partial Dierential Equations Igor Yanovsky, 2005 132
Problem (S90, #4).
Let f(x) C

## be a 2-periodic function, i.e., f(x) = f(x + 2) and denote by

[[f[[
2
=
_
2
0
[f(x)[
2
dx
the L
2
-norm of f.
a) Express [[d
p
f/dx
p
[[
2
in terms of the Fourier coecients of f.
b) Let q > p > 0 be integers. Prove that > 0, K = N(, p, q), constant, such that

d
p
f
dx
p

d
q
f
dx
q

2
+K[[f[[
2
.
c) Discuss how K depends on .
Proof. a) Let
31
f(x) =

f
n
e
inx
,
d
p
f
dx
p
=

f
n
(in)
p
e
inx
,

d
p
f
dx
p

2
=
_
2
0

f
n
(in)
p
e
inx

2
dx =
_
2
0
[i
2
[
p

f
n
n
p
e
inx

2
dx
=
_
2
0

f
n
n
p
e
inx

2
dx = 2

n=0
f
2
n
n
2p
.
b) We have

d
p
f
dx
p

d
q
f
dx
q

2
+ K[[f[[
2
,
2

n=0
f
2
n
n
2p
2

n=0
f
2
n
n
2q
+ K 2

n=0
f
2
n
,
n
2p
n
2q
K,
n
2p
(1 n
q

)
. .
<0, for n large
K, some q

> 0.
Thus, the above inequality is true for n large enough. The statement follows.
31
Note:
_
L
0
e
inx
e
imx
dx =
_
0 n = m
L n = m
Partial Dierential Equations Igor Yanovsky, 2005 133
Problem (S90, #5).
32
Consider the ame front equation
u
t
+uu
x
+ u
xx
+u
xxxx
= 0
with 2-periodic initial data
u(x, 0) = f(x), f(x) = f(x + 2) C

.
a) Determine the solution, if f(x) f
0
= const.
b) Assume that
f(x) = 1 +g(x), 0 < 1, [g[

## = 1, g(x) = g(x + 2).

Linearize the equation. Is the Cauchy problem well-posed for the linearized equation,
i.e., do its solutions v satisfy an estimate
[[v(, t)[[ Ke
(tt
0
)
[[v(, t
0
)[[?
c) Determine the best possible constants K, .
Proof. a) The solution to
u
t
+uu
x
+ u
xx
+u
xxxx
= 0,
u(x, 0) = f
0
= const,
is u(x, t) = f
0
= const.
b) We consider the special case of nearly constant initial data
u(x, 0) = 1 + u
1
(x, 0).
Then we can approximate the nonlinear equation by a linear equation. Assuming
u(x, t) = 1 + u
1
(x, t),
remain valid with u
1
= O(1), from , we nd that
u
1t
+ (1 + u
1
)u
1x
+ u
1xx
+u
1xxxx
= 0.
Dividing by gives
u
1t
+ u
1x
+ u
1
u
1x
+u
1xx
+u
1xxxx
= 0.
For small , we have
u
1t
+ u
1x
+u
1xx
+u
1xxxx
= 0.
Multiply this equation by u
1
and integrate
u
1
u
1t
+ u
1
u
1x
+u
1
u
1xx
+u
1
u
1xxxx
= 0,
d
dt
_
u
2
1
2
_
+
_
u
2
1
2
_
x
+u
1
u
1xx
+u
1
u
1xxxx
= 0,
1
2
d
dt
_
2
0
u
2
1
dx +
u
2
1
2

2
0
. .
=0
+
_
2
0
u
1
u
1xx
dx +
_
2
0
u
1
u
1xxxx
dx = 0,
1
2
d
dt
[[u
1
[[
2
2
+ u
1
u
1x

2
0
. .
=0

_
2
0
u
2
1x
dx +u
1
u
1xxx

2
0
. .
=0
u
1x
u
1xx

2
0
. .
=0
+
_
2
0
u
2
1xx
dx = 0,
1
2
d
dt
[[u
1
[[
2
2
=
_
2
0
u
2
1x
dx
_
2
0
u
2
1xx
dx.
32
S90 #5, #6, #7 all have similar formulations.
Partial Dierential Equations Igor Yanovsky, 2005 134
Since u
1
is 2-periodic, let
u
1
=

n=
a
n
(t)e
inx
. Then,
u
1x
= i

n=
na
n
(t)e
inx
u
2
1x
=
_

n=
na
n
(t)e
inx
_
2
,
u
1xx
=

n=
n
2
a
n
(t)e
inx
u
2
1xx
=
_

n=
n
2
a
n
(t)e
inx
_
2
.
Thus,
1
2
d
dt
[[u
1
[[
2
2
=
_
2
0
u
2
1x
dx
_
2
0
u
2
1xx
dx
=
_
2
0
_

na
n
(t)e
inx
_
2
dx
_
2
0
_

n
2
a
n
(t)e
inx
_
2
dx
= 2

n
2
a
n
(t)
2
2

n
4
a
n
(t)
2
= 2

a
n
(t)
2
(n
2
+n
4
) 0.
[[u
1
(, t)[[
2
[[u
1
(, 0)[[
2
,
where K = 1, = 0.
Problem (W03, #4). Consider the PDE
u
t
= u
x
+ u
4
for t > 0
u = u
0
for t = 0
for 0 < x < 2. Dene the set A = u = u(x) : u(k) = 0 if k < 0, in which
u(k, t)

## is the Fourier series of u in x on [0, 2].

a) If u
0
A, show that u(t) A.
b) Find dierential equations for u(0, t), u(1, t), and u(2, t).
Proof. a) Solving
u
t
= u
x
+ u
4
u(x, 0) = u
0
(x)
by the method of characteristics, we get
u(x, t) =
u
0
(x + t)
(1 3t(u
0
(x +t))
3
)
1
3
.
Since u
0
A, u
0k
= 0 if k < 0. Thus,
u
0
(x) =

k=0
u
0k
e
i
kx
2
.
Since
u
k
=
1
2
_
2
0
u(x, t) e
i
kx
2
dx,
Partial Dierential Equations Igor Yanovsky, 2005 135
we have
u(x, t) =

k=0
u
k
e
i
kx
2
,
that is, u(t) A.
Partial Dierential Equations Igor Yanovsky, 2005 136
15.4 Energy Estimates
Problem (S90, #6). Let U(x, t) C

## be 2-periodic in x. Consider the linear

equation
u
t
+Uu
x
+ u
xx
+u
xxxx
= 0,
u(x, 0) = f(x), f(x) = f(x + 2) C

.
a) Derive an energy estimate for u.
b) Prove that one can estimate all derivatives [[
p
u/x
p
[[.
c) Indicate how to prove existence of solutions.
33
Proof. a) Multiply the equation by u and integrate
uu
t
+ Uuu
x
+uu
xx
+ uu
xxxx
= 0,
1
2
d
dt
(u
2
) +
1
2
U(u
2
)
x
+uu
xx
+ uu
xxxx
= 0,
1
2
d
dt
_
2
0
u
2
dx +
1
2
_
2
0
U(u
2
)
x
dx +
_
2
0
uu
xx
dx +
_
2
0
uu
xxxx
dx = 0,
1
2
d
dt
[[u[[
2
+
1
2
Uu
2

2
0
. .
=0

1
2
_
2
0
U
x
u
2
dx + uu
x

2
0

_
2
0
u
2
x
dx
+uu
xxx

2
0
u
x
u
xx

2
0
+
_
2
0
u
2
xx
dx = 0,
1
2
d
dt
[[u[[
2

1
2
_
2
0
U
x
u
2
dx
_
2
0
u
2
x
dx +
_
2
0
u
2
xx
dx = 0,
1
2
d
dt
[[u[[
2
=
1
2
_
2
0
U
x
u
2
dx +
_
2
0
u
2
x
dx
_
2
0
u
2
xx
dx (from S90, #5)

1
2
_
2
0
U
x
u
2
dx
1
2
max
x
U
x
_
2
0
u
2
dx.

d
dt
[[u[[
2
max
x
U
x
[[u[[
2
,
[[u(x, t)[[
2
[[u(x, 0)[[
2
e
(maxx Ux)t
.
This can also been done using Fourier Transform. See ChiuYens solutions where the
above method and the Fourier Transform methods are used.
33
S90 #5, #6, #7 all have similar formulations.
Partial Dierential Equations Igor Yanovsky, 2005 137
Problem (S90, #7).
34
Consider the nonlinear equation
u
t
+uu
x
+ u
xx
+u
xxxx
= 0,
u(x, 0) = f(x), f(x) = f(x + 2) C

.
a) Derive an energy estimate for u.
b) Show that there is an interval 0 t T, T depending on f,
such that also [[u(, t)/x[[ can be bounded.
Proof. a) Multiply the above equation by u and integrate
uu
t
+u
2
u
x
+uu
xx
+ uu
xxxx
= 0,
1
2
d
dt
(u
2
) +
1
3
(u
3
)
x
+uu
xx
+ uu
xxxx
= 0,
1
2
d
dt
_
2
0
u
2
dx +
1
3
_
2
0
(u
3
)
x
dx +
_
2
0
uu
xx
dx +
_
2
0
uu
xxxx
dx = 0,
1
2
d
dt
[[u[[
2
+
1
3
u
3

2
0
. .
=0

_
2
0
u
2
x
dx +
_
2
0
u
2
xx
dx = 0,
1
2
d
dt
[[u[[
2
=
_
2
0
u
2
x
dx
_
2
0
u
2
xx
dx 0, (from S90, #5)
[[u(, t)[[ [[u(, 0)[[.
b) In order to nd a bound for [[u
x
(, t)[[, dierentiate with respect to x:
u
tx
+ (uu
x
)
x
+ u
xxx
+u
xxxxx
= 0,
Multiply the above equation by u
x
and integrate:
u
x
u
tx
+ u
x
(uu
x
)
x
+u
x
u
xxx
+ u
x
u
xxxxx
= 0,
1
2
d
dt
_
2
0
(u
x
)
2
dx +
_
2
0
u
x
(uu
x
)
x
dx +
_
2
0
u
x
u
xxx
dx +
_
2
0
u
x
u
xxxxx
dx = 0.
We evaluate one of the integrals in the above expression using the periodicity:
_
2
0
u
x
(uu
x
)
x
dx =
_
2
0
u
xx
uu
x
=
_
2
0
u
x
(u
2
x
+ uu
xx
) =
_
2
0
u
3
x
+
_
2
0
uu
x
u
xx
,

_
2
0
u
xx
uu
x
=
1
2
_
2
0
u
3
x
,

_
2
0
u
x
(uu
x
)
x
=
1
2
_
2
0
u
3
x
.
We have
1
2
d
dt
[[u
x
[[
2
+
_
2
0
u
3
x
dx +
_
2
0
u
x
u
xxx
dx +
_
2
0
u
x
u
xxxxx
dx = 0.
34
S90 #5, #6, #7 all have similar formulations.
Partial Dierential Equations Igor Yanovsky, 2005 138
Let w = u
x
, then
1
2
d
dt
[[w[[
2
=
_
2
0
w
3
dx
_
2
0
ww
xx
dx
_
2
0
ww
xxxx
dx
=
_
2
0
w
3
dx +
_
2
0
w
2
x
dx
_
2
0
w
2
xx
dx
_
2
0
w
3
dx,

d
dt
[[u
x
[[
2
=
_
2
0
u
3
x
dx.
Partial Dierential Equations Igor Yanovsky, 2005 139
16 Problems: Wave Equation
16.1 The Initial Value Problem
Example (McOwen 3.1 #1). Solve the initial value problem:

u
tt
c
2
u
xx
= 0,
u(x, 0) = x
3
..
g(x)
, u
t
(x, 0) = sinx
..
h(x)
.
Proof. DAlemberts formula gives the solution:
u(x, t) =
1
2
(g(x + ct) +g(x ct)) +
1
2c
_
x+ct
xct
h() d
=
1
2
(x + ct)
3
+
1
2
(x ct)
3
+
1
2c
_
x+ct
xct
sin d
= x
3
+ 2xc
2
t
2

1
2c
cos(x +ct) +
1
2c
cos(x ct) =
= x
3
+ 2xc
2
t
2
+
1
c
sin x sinct.
Problem (S99, #6). Solve the Cauchy problem
_
u
tt
= a
2
u
xx
+ cos x,
u(x, 0) = sin x, u
t
(x, 0) = 1 + x.
(16.1)
Proof. We have a nonhomogeneous PDE with nonhomogeneous initial conditions:

u
tt
c
2
u
xx
= cos x
. .
f(x,t)
,
u(x, 0) = sin x
..
g(x)
, u
t
(x, 0) = 1 +x
. .
h(x)
.
The solution is given by dAlemberts formula and Duhamels principle.
35
u
A
(x, t) =
1
2
(g(x +ct) + g(x ct)) +
1
2c
_
x+ct
xct
h() d
=
1
2
(sin(x + ct) + sin(x ct)) +
1
2c
_
x+ct
xct
(1 +) d
= sinx cos ct +
1
2c
_
+

2
2
_
=x+ct
=xct
= sinx cos ct +xt +t.
u
D
(x, t) =
1
2c
_
t
0
__
x+c(ts)
xc(ts)
f(, s) d
_
ds =
1
2c
_
t
0
__
x+c(ts)
xc(ts)
cos d
_
ds
=
1
2c
_
t
0
_
sin[x +c(t s)] sin[x c(t s)]
_
ds =
1
c
2
(cos x cos x cos ct).
u(x, t) = u
A
(x, t) +u
D
(x, t) = sinx cos ct +xt +t +
1
c
2
(cos x cos x cos ct).
35
Note the relationship: x , t s.
Partial Dierential Equations Igor Yanovsky, 2005 140
We can check that the solution satises equation (16.1). Can also check that u
A
, u
D
satisfy
_
u
A
tt
c
2
u
A
xx
= 0,
u
A
(x, 0) = sinx, u
A
t
(x, 0) = 1 + x;
_
u
D
tt
c
2
u
D
xx
= cos x,
u
D
(x, 0) = 0, u
D
t
(x, 0) = 0.
Partial Dierential Equations Igor Yanovsky, 2005 141
16.2 Initial/Boundary Value Problem
Problem 1. Consider the initial/boundary value problem

u
tt
c
2
u
xx
= 0 0 < x < L, t > 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) 0 < x < L
u(0, t) = 0, u(L, t) = 0 t 0.
(16.2)
Proof. Find u(x, t) in the form
u(x, t) =
a
0
(t)
2
+

n=1
a
n
(t) cos
nx
L
+b
n
(t) sin
nx
L
.
Functions a
n
(t) and b
n
(t) are determined by the boundary conditions:
0 = u(0, t) =
a
0
(t)
2
+

n=1
a
n
(t) a
n
(t) = 0. Thus,
u(x, t) =

n=1
b
n
(t) sin
nx
L
. (16.3)
If we substitute (16.3) into the equation u
tt
c
2
u
xx
= 0, we get

n=1
b

n
(t) sin
nx
L
+ c
2

n=1
_
n
L
_
2
b
n
(t) sin
nx
L
= 0, or
b

n
(t) +
_
nc
L
_
2
b
n
(t) = 0,
whose general solution is
b
n
(t) = c
n
sin
nct
L
+ d
n
cos
nct
L
. (16.4)
Also, b

n
(t) = c
n
(
nc
L
) cos
nct
L
d
n
(
nc
L
) sin
nct
L
.
The constants c
n
and d
n
are determined by the initial conditions:
g(x) = u(x, 0) =

n=1
b
n
(0) sin
nx
L
=

n=1
d
n
sin
nx
L
,
h(x) = u
t
(x, 0) =

n=1
b

n
(0) sin
nx
L
=

n=1
c
n
nc
L
sin
nx
L
.
By orthogonality, we may multiply by sin(mx/L) and integrate:
_
L
0
g(x) sin
mx
L
dx =
_
L
0

n=1
d
n
sin
nx
L
sin
mx
L
dx = d
m
L
2
,
_
L
0
h(x) sin
mx
L
dx =
_
L
0

n=1
c
n
nc
L
sin
nx
L
sin
mx
L
dx = c
m
mc
L
L
2
.
Thus,
d
n
=
2
L
_
L
0
g(x) sin
nx
L
dx, c
n
=
2
nc
_
L
0
h(x) sin
nx
L
dx. (16.5)
The formulas (16.3), (16.4), and (16.5) dene the solution.
Partial Dierential Equations Igor Yanovsky, 2005 142
Example (McOwen 3.1 #2). Consider the initial/boundary value problem

u
tt
u
xx
= 0 0 < x < , t > 0
u(x, 0) = 1, u
t
(x, 0) = 0 0 < x <
u(0, t) = 0, u(, t) = 0 t 0.
(16.6)
Proof. Find u(x, t) in the form
u(x, t) =
a
0
(t)
2
+

n=1
a
n
(t) cos nx +b
n
(t) sinnx.
Functions a
n
(t) and b
n
(t) are determined by the boundary conditions:
0 = u(0, t) =
a
0
(t)
2
+

n=1
a
n
(t) a
n
(t) = 0. Thus,
u(x, t) =

n=1
b
n
(t) sinnx. (16.7)
If we substitute this into u
tt
u
xx
= 0, we get

n=1
b

n
(t) sinnx +

n=1
b
n
(t)n
2
sinnx = 0, or
b

n
(t) +n
2
b
n
(t) = 0,
whose general solution is
b
n
(t) = c
n
sinnt +d
n
cos nt. (16.8)
Also, b

n
(t) = nc
n
cos nt nd
n
sinnt.
The constants c
n
and d
n
are determined by the initial conditions:
1 = u(x, 0) =

n=1
b
n
(0) sinnx =

n=1
d
n
sin nx,
0 = u
t
(x, 0) =

n=1
b

n
(0) sinnx =

n=1
nc
n
sinnx.
By orthogonality, we may multiply both equations by sinmx and integrate:
_

0
sin mx dx = d
m

2
,
_

0
0 dx = nc
n

2
.
Thus,
d
n
=
2
n
(1 cos n) =
_
4
n
, n odd,
0, n even,
and c
n
= 0. (16.9)
Using this in (16.8) and (16.7), we get
b
n
(t) =
_
4
n
cos nt, n odd,
0, n even,
Partial Dierential Equations Igor Yanovsky, 2005 143
u(x, t) =
4

n=0
cos(2n + 1)t sin(2n + 1)x
(2n + 1)
.
Partial Dierential Equations Igor Yanovsky, 2005 144
We can sum the series in regions bouded by characteristics. We have
u(x, t) =
4

n=0
cos(2n + 1)t sin(2n + 1)x
(2n + 1)
, or
u(x, t) =
2

n=0
sin[(2n + 1)(x + t)]
(2n + 1)
+
2

n=0
sin[(2n + 1)(x t)]
(2n + 1)
. (16.10)
The initial condition may be written as
1 = u(x, 0) =
4

n=0
sin(2n + 1)x
(2n + 1)
for 0 < x < . (16.11)
We can use (16.11) to sum the series in (16.10).
In R
1
, u(x, t) =
1
2
+
1
2
= 1.
Since sin[(2n + 1)(x t)] = sin[(2n + 1)((x t))], and 0 < (x t) < in R
2
,
in R
2
, u(x, t) =
1
2

1
2
= 0.
Since sin[(2n + 1)(x +t)] = sin[(2n + 1)(x + t 2)] = sin[(2n + 1)(2 (x + t))],
and 0 < 2 (x + t) < in R
3
,
in R
3
, u(x, t) =
1
2
+
1
2
= 0.
Since 0 < (x t) < and 0 < 2 (x +t) < in R
4
,
in R
4
, u(x, t) =
1
2

1
2
= 1.
Partial Dierential Equations Igor Yanovsky, 2005 145
Problem 2. Consider the initial/boundary value problem

u
tt
c
2
u
xx
= 0 0 < x < L, t > 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) 0 < x < L
u
x
(0, t) = 0, u
x
(L, t) = 0 t 0.
(16.12)
Proof. Find u(x, t) in the form
u(x, t) =
a
0
(t)
2
+

n=1
a
n
(t) cos
nx
L
+b
n
(t) sin
nx
L
.
Functions a
n
(t) and b
n
(t) are determined by the boundary conditions:
u
x
(x, t) =

n=1
a
n
(t)
_
n
L
_
sin
nx
L
+b
n
(t)
_
n
L
_
cos
nx
L
,
0 = u
x
(0, t) =

n=1
b
n
(t)
_
n
L
_
b
n
(t) = 0. Thus,
u(x, t) =
a
0
(t)
2
+

n=1
a
n
(t) cos
nx
L
. (16.13)
If we substitute (16.13) into the equation u
tt
c
2
u
xx
= 0, we get
a

0
(t)
2
+

n=1
a

n
(t) cos
nx
L
+ c
2

n=1
a
n
(t)
_
n
L
_
2
cos
nx
L
= 0,
a

0
(t) = 0 and a

n
(t) +
_
nc
L
_
2
a
n
(t) = 0,
whose general solutions are
a
0
(t) = c
0
t +d
0
and a
n
(t) = c
n
sin
nct
L
+ d
n
cos
nct
L
. (16.14)
Also, a

0
(t) = c
0
and a

n
(t) = c
n
(
nc
L
) cos
nct
L
d
n
(
nc
L
) sin
nct
L
.
The constants c
n
and d
n
are determined by the initial conditions:
g(x) = u(x, 0) =
a
0
(0)
2
+

n=1
a
n
(0) cos
nx
L
=
d
0
2
+

n=1
d
n
cos
nx
L
,
h(x) = u
t
(x, 0) =
a

0
(0)
2
+

n=1
a

n
(0) cos
nx
L
=
c
0
2
+

n=1
c
n
nc
L
cos
nx
L
.
By orthogonality, we may multiply both equations by cos(mx/L), including m = 0,
and integrate:
_
L
0
g(x) dx = d
0
L
2
,
_
L
0
g(x) cos
mx
L
dx = d
m
L
2
,
_
L
0
h(x) dx = c
0
L
2
,
_
L
0
h(x) cos
mx
L
dx = c
m
mc
L
L
2
.
Thus,
d
n
=
2
L
_
L
0
g(x) cos
nx
L
dx, c
n
=
2
nc
_
L
0
h(x) cos
nx
L
dx, c
0
=
2
L
_
L
0
h(x) dx.
(16.15)
The formulas (16.13), (16.14), and (16.15) dene the solution.
Partial Dierential Equations Igor Yanovsky, 2005 146
Example (McOwen 3.1 #3). Consider the initial/boundary value problem

u
tt
u
xx
= 0 0 < x < , t > 0
u(x, 0) = x, u
t
(x, 0) = 0 0 < x <
u
x
(0, t) = 0, u
x
(, t) = 0 t 0.
(16.16)
Proof. Find u(x, t) in the form
u(x, t) =
a
0
(t)
2
+

n=1
a
n
(t) cos nx +b
n
(t) sinnx.
Functions a
n
(t) and b
n
(t) are determined by the boundary conditions:
u
x
(x, t) =

n=1
a
n
(t)nsinnx +b
n
(t)ncos nx,
0 = u
x
(0, t) =

n=1
b
n
(t)n b
n
(t) = 0. Thus,
u(x, t) =
a
0
(t)
2
+

n=1
a
n
(t) cos nx. (16.17)
If we substitute (16.17) into the equation u
tt
u
xx
= 0, we get
a

0
(t)
2
+

n=1
a

n
(t) cos nx +

n=1
a
n
(t)n
2
cos nx = 0,
a

0
(t) = 0 and a

n
(t) + n
2
a
n
(t) = 0,
whose general solutions are
a
0
(t) = c
0
t +d
0
and a
n
(t) = c
n
sinnt + d
n
cos nt. (16.18)
Also, a

0
(t) = c
0
and a

n
(t) = c
n
ncos nt d
n
nsinnt.
The constants c
n
and d
n
are determined by the initial conditions:
x = u(x, 0) =
a
0
(0)
2
+

n=1
a
n
(0) cos nx =
d
0
2
+

n=1
d
n
cos nx,
0 = u
t
(x, 0) =
a

0
(0)
2
+

n=1
a

n
(0) cos nx =
c
0
2
+

n=1
c
n
ncos nx.
By orthogonality, we may multiply both equations by cos mx, including m = 0, and
integrate:
_

0
x dx = d
0

2
,
_

0
x cos mx dx = d
m

2
,
_

0
0 dx = c
0

2
,
_

0
0 cos mx dx = c
m
m

2
.
Thus,
d
0
= , d
n
=
2
n
2
(cos n 1), c
n
= 0. (16.19)
Using this in (16.18) and (16.17), we get
a
0
(t) = d
0
= , a
n
(t) =
2
n
2
(cos n 1) cos nt,
Partial Dierential Equations Igor Yanovsky, 2005 147
u(x, t) =

2
+
2

n=1
(cos n 1) cos nt cos nx
n
2
.
Partial Dierential Equations Igor Yanovsky, 2005 148
We can sum the series in regions bouded by characteristics. We have
u(x, t) =

2
+
2

n=1
(cos n 1) cos nt cos nx
n
2
, or
u(x, t) =

2
+
1

n=1
(cos n 1) cos[n(x t)]
n
2
+
1

n=1
(cos n 1) cos[n(x + t)]
n
2
. (16.20)
The initial condition may be written as
u(x, 0) = x =

2
+
2

n=1
(cos n 1) cos nx
n
2
for 0 < x < ,
which implies
x
2

4
=
1

n=1
(cos n 1) cos nx
n
2
for 0 < x < , (16.21)
We can use (16.21) to sum the series in (16.20).
In R
1
, u(x, t) =

2
+
x t
2

4
+
x + t
2

4
= x.
Since cos[n(x t)] = cos[n((x t))], and 0 < (x t) < in R
2
,
in R
2
, u(x, t) =

2
+
(x t)
2

4
+
x + t
2

4
= t.
Since cos[n(x+t)] = cos[n(x+t2)] = cos[n(2(x+t))], and 0 < 2(x+t) <
in R
3
,
in R
3
, u(x, t) =

2
+
x t
2

4
+
2 (x + t)
2

4
= t.
Since 0 < (x t) < and 0 < 2 (x + t) < in R
4
in R
4
, u(x, t) =

2
+
(x t)
2

4
+
2 (x +t)
2

4
= x.
Partial Dierential Equations Igor Yanovsky, 2005 149
Example (McOwen 3.1 #4). Consider the initial boundary value problem

u
tt
c
2
u
xx
= 0 for x > 0, t > 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) for x > 0
u(0, t) = 0 for t 0,
(16.22)
where g(0) = 0 = h(0). If we extend g and h as odd functions on < x < , show
that dAlemberts formula gives the solution.
Proof. Extend g and h as odd functions on < x < :
g(x) =
_
g(x), x 0
g(x), x < 0

h(x) =
_
h(x), x 0
h(x), x < 0.
Then, we need to solve
_
u
tt
c
2
u
xx
= 0 for < x < , t > 0
u(x, 0) = g(x), u
t
(x, 0) =

h(x) for < x < .
(16.23)
To show that dAlemberts formula gives the solution to (16.23), we need to show that
the solution given by dAlemberts formula satises the boundary condition u(0, t) = 0.
u(x, t) =
1
2
( g(x + ct) + g(x ct)) +
1
2c
_
x+ct
xct

h() d,
u(0, t) =
1
2
( g(ct) + g(ct)) +
1
2c
_
ct
ct

h() d
=
1
2
( g(ct) g(ct)) +
1
2c
(H(ct) H(ct))
= 0 +
1
2c
(H(ct) H(ct)) = 0,
where we used H(x) =
_
x
0

## h() d; and since

h is odd, then H is even.
Example (McOwen 3.1 #5). Find in closed form (similar to dAlembets formula)
the solution u(x, t) of

u
tt
c
2
u
xx
= 0 for x, t > 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) for x > 0
u(0, t) = (t) for t 0,
(16.24)
where g, h, C
2
satisfy (0) = g(0),

(0) = c
2
g

## (0). Verify that

u C
2
, even on the characteristic x = ct.
Proof. As in (McOwen 3.1 #4), we can extend g and h to be odd functions. We want
to transform the problem to have zero boundary conditions.
Consider the function:
U(x, t) = u(x, t) (t). (16.25)
Partial Dierential Equations Igor Yanovsky, 2005 150
Then (16.24) transforms to:

U
tt
c
2
U
xx
=

(t)
. .
f
U
(x,t)
U(x, 0) = g(x) (0)
. .
g
U
(x)
, U
t
(x, 0) = h(x)

(0)
. .
h
U
(x)
U(0, t) = 0
..
u(t)
.
We use dAlemberts formula and Duhamels principle on U.
After getting U, we can get u from u(x, t) = U(x, t) +(t).
Partial Dierential Equations Igor Yanovsky, 2005 151
Example (Zachmanoglou, Chapter 8, Example 7.2). Find the solution of

u
tt
c
2
u
xx
= 0 for x > 0, t > 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) for x > 0
u
x
(0, t) = 0 for t > 0.
(16.26)
Proof. Extend g and h as even functions on < x < :
g(x) =
_
g(x), x 0
g(x), x < 0

h(x) =
_
h(x), x 0
h(x), x < 0.
Then, we need to solve
_
u
tt
c
2
u
xx
= 0 for < x < , t > 0
u(x, 0) = g(x), u
t
(x, 0) =

h(x) for < x < .
(16.27)
To show that dAlemberts formula gives the solution to (16.27), we need to show that
the solution given by dAlemberts formula satises the boundary condition u
x
(0, t) = 0.
u(x, t) =
1
2
( g(x + ct) + g(x ct)) +
1
2c
_
x+ct
xct

h() d.
u
x
(x, t) =
1
2
( g

(x + ct) + g

(x ct)) +
1
2c
[

h(x +ct)

h(x ct)],
u
x
(0, t) =
1
2
( g

(ct) + g

(ct)) +
1
2c
[

h(ct)

h(ct)] = 0.
Since g is even, then g

is odd.
Problem (F89, #3).
36
Let ,= c, constant. Find the solution of

u
tt
c
2
u
xx
= 0 for x > 0, t > 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) for x > 0
u
t
(0, t) = u
x
(0, t) for t > 0,
(16.28)
where g, h C
2
for x > 0 and vanish near x = 0.
Hint: Use the fact that a general solution of (16.28) can be written as the sum of two
traveling wave solutions.
Proof. DAlemberts formula is derived by plugging in the following into the above
equation and initial conditions:
u(x, t) = F(x +ct) +G(x ct).
As in (Zachmanoglou 7.2), we can extend g and h to be even functions.
36
Similar to McOwen 3.1 #5. The notation in this problem is changed to be consistent with McOwen.
Partial Dierential Equations Igor Yanovsky, 2005 152
Example (McOwen 3.1 #6). Solve the initial/boundary value problem

u
tt
u
xx
= 1 for 0 < x < and t > 0
u(x, 0) = 0, u
t
(x, 0) = 0 for 0 < x <
u(0, t) = 0, u(, t) =
2
/2 for t 0.
(16.29)
Proof. If we rst nd a particular solution of the nonhomogeneous equation, this re-
duces the problem to a boundary value problem for the homogeneous equation ( as in
(McOwen 3.1 #2) and (McOwen 3.1 #3) ).
Hint: You should use a particular solution depending on x!
First, nd a particular solution. This is similar to the method of separation of
variables. Assume
u
p
(x, t) = X(x),
which gives
X

(x) = 1,
X

(x) = 1.
The solution to the above ODE is
X(x) =
x
2
2
+ax + b.
The boundary conditions give
u
p
(0, t) = b = 0,
u
p
(, t) =

2
2
+ a + b =

2
2
, a = b = 0.
Thus, the particular solution is
u
p
(x, t) =
x
2
2
.
This solution satises the following:

u
p
tt
u
p
xx
= 1
u
p
(x, 0) =
x
2
2
, u
p
t
(x, 0) = 0
u
p
(0, t) = 0, u
p
(, t) =

2
2
.
Second, we nd a solution to a boundary value problem for the homogeneous equa-
tion:

u
tt
u
xx
= 0
u(x, 0) =
x
2
2
, u
t
(x, 0) = 0
u(0, t) = 0, u(, t) = 0.
This is solved by the method of Separation of Variables. See Separation of Variables
subsection of Problems: Separation of Variables: Wave Equation McOwen 3.1 #2.
The only dierence there is that u(x, 0) = 1.
We would nd u
h
(x, t). Then,
u(x, t) = u
h
(x, t) +u
p
(x, t).
Partial Dierential Equations Igor Yanovsky, 2005 153
Problem (S02, #2). a) Given a continuous function f on R which vanishes for
[x[ > R, solve the initial value problem
_
u
tt
u
xx
= f(x) cos t,
u(x, 0) = 0, u
t
(x, 0) = 0, < x < , 0 t <
by rst nding a particular solution by separation of variables and then adding the
appropriate solution of the homogeneous PDE.
b) Since the particular solution is not unique, it will not be obvious that the solution
to the initial value problem that you have found in part (a) is unique. Prove that it is
unique.
Proof. a) First, nd a particular solution by separation of variables. Assume
u
p
(x, t) = X(x) cos t,
which gives
X(x) cos t X

## (x) cos t = f(x) cos t,

X

+ X = f(x).
The solution to the above ODE is written as X = X
h
+X
p
. The homogeneous solution
is
X
h
(x) = a cos x +b sinx.
To nd a particular solution, note that since f is continuous, G C
2
(R), such that
G

+G = f(x).
Thus,
X
p
(x) = G(x).
X(x) = X
h
(x) +X
p
(x) = a cos x +b sinx +G(x).
u
p
(x, t) =
_
a cos x + b sinx + G(x)

cos t.
It can be veried that this solution satises the following:
_
u
p
tt
u
p
xx
= f(x) cos t,
u
p
(x, 0) = a cos x + b sinx + G(x), u
p
t
(x, 0) = 0.
Second, we nd a solution of the homogeneous PDE:

u
tt
u
xx
= 0,
u(x, 0) = a cos x b sinx G(x)
. .
g(x)
, u
t
(x, 0) = 0
..
h(x)
.
The solution is given by dAlemberts formula (with c = 1):
u
h
(x, t) = u
A
(x, t) =
1
2
(g(x + t) + g(x t)) +
1
2
_
x+t
xt
h() d
=
1
2
_
_
a cos(x + t) b sin(x +t) G(x +t)
_
+
_
a cos(x t) b sin(x t) G(x t)
_
_
=
1
2
_
a cos(x +t) +b sin(x +t) + G(x + t)
_

1
2
_
a cos(x t) +b sin(x t) + G(x t)
_
.
Partial Dierential Equations Igor Yanovsky, 2005 154
It can be veried that the solution satises the above homogeneous PDE with the
boundary conditions. Thus, the complete solution is:
u(x, t) = u
h
(x, t) +u
p
(x, t).
Alternatively, we could use Duhamels principle to nd the solution:
37
u(x, t) =
1
2
_
t
0
__
x+(ts)
x(ts)
f() cos s d
_
ds.
However, this is not how it was suggested to do this problem.
b) The particular solution is not unique, since any constants a, b give the solution.
However, we show that the solution to the initial value problem is unique.
Suppose u
1
and u
2
are two solutions. Then w = u
1
u
2
satises:
_
w
tt
w
xx
= 0,
w(x, 0) = 0, w
t
(x, 0) = 0.
DAlemberts formula gives
w(x, t) =
1
2
(g(x +t) + g(x t)) +
1
2
_
x+t
xt
h() d = 0.
Thus, the solution to the initial value problem is unique.
37
Note the relationship: x , t s.
Partial Dierential Equations Igor Yanovsky, 2005 155
16.3 Similarity Solutions
Problem (F98, #7). Look for a similarity solution of the form
v(x, t) = t

w(y = x/t

## ) for the dierential equation

v
t
= v
xx
+ (v
2
)
x
. (16.30)
a) Find the parameters and .
b) Find a dierential equation for w(y) and show that this ODE can be reduced to rst
order.
c) Find a solution for the resulting rst order ODE.
Proof. We can rewrite (16.30) as
v
t
= v
xx
+ 2vv
x
. (16.31)
We look for a similarity solution of the form
v(x, t) = t

w(y),
_
y =
x
t

_
.
v
t
= t
1
w + t

y
t
= t
1
w + t

x
t
+1
_
w

= t
1
w t
1
yw

,
v
x
= t

y
x
= t

= t

,
v
xx
= (t

)
x
= t

y
x
= t

= t
2
w

.
Plugging in the derivatives we calculated into (16.31), we obtain
t
1
w t
1
yw

= t
2
w

+ 2(t

w)(t

),
w yw

= t
12
w

+ 2t
+1
ww

.
The parameters that would eliminate t from equation above are
=
1
2
, =
1
2
.
With these parameters, we obtain the dierential equation for w(y):

1
2
w
1
2
yw

= w

+ 2ww

,
w

+ 2ww

+
1
2
yw

+
1
2
w = 0.
We can write the ODE as
w

+ 2ww

+
1
2
(yw)

= 0.
Integrating it with respect to y, we obtain the rst order ODE:
w

+ w
2
+
1
2
yw = c.
Partial Dierential Equations Igor Yanovsky, 2005 156
16.4 Traveling Wave Solutions
Consider the Korteweg-de Vries (KdV) equation in the form
38
u
t
+ 6uu
x
+u
xxx
= 0, < x < , t > 0. (16.32)
We look for a traveling wave solution
u(x, t) = f(x ct). (16.33)
We get the ODE
cf

+ 6ff

+f

= 0. (16.34)
We integrate (16.34) to get
cf + 3f
2
+ f

= a, (16.35)
where a is a constant. Multiplying this equality by f

, we obtain
cff

+ 3f
2
f

+f

= af

.
Integrating again, we get

c
2
f
2
+f
3
+
(f

)
2
2
= af + b. (16.36)
We are looking for solutions f which satisfy f(x), f

(x), f

(x) 0 as x . (In
which case the function u having the form (16.33) is called a solitary wave.) Then
(16.35) and (16.36) imply a = b = 0, so that

c
2
f
2
+f
3
+
(f

)
2
2
= 0, or f

= f
_
c 2f.
The solution of this ODE is
f(x) =
c
2
sech
2
[

c
2
(x x
0
)],
where x
0
is the constant of integration. A solution of this form is called a soliton.
38
Evans, p. 174; Strauss, p. 367.
Partial Dierential Equations Igor Yanovsky, 2005 157
Problem (S93, #6). The generalized KdV equation is
u
t
=
1
2
(n + 1)(n + 2)u
n
u
x

3
u
x
3
,
where n is a positive integer. Solitary wave solutions are sought in which u = f(),
where = x ct and
f, f

, f

0, as [[ ;
c, the wave speed, is constant.
Show that
f
2
= f
n+2
+cf
2
.
Hence show that solitary waves do not exist if n is even.
Show also that, when n = 1, all conditions of the problem are satised provided c > 0
and
u = c sech
2
_
1
2

c(x ct)
_
.
Proof. We look for a traveling wave solution
u(x, t) = f(x ct).
We get the ODE
cf

=
1
2
(n + 1)(n + 2)f
n
f

,
Integrating this equation, we get
cf =
1
2
(n + 2)f
n+1
f

+ a, (16.37)
where a is a constant. Multiplying this equality by f

, we obtain
cff

=
1
2
(n + 2)f
n+1
f

+ af

.
Integrating again, we get

cf
2
2
=
1
2
f
n+2

(f

)
2
2
+ af + b. (16.38)
We are looking for solutions f which satisfy f, f

, f

0 as x . Then (16.37)
and (16.38) imply a = b = 0, so that

cf
2
2
=
1
2
f
n+2

(f

)
2
2
,
(f

)
2
= f
n+2
+cf
2
.
We show that solitary waves do not exist if n is even. We have
f

=
_
f
n+2
+cf
2
= [f[
_
f
n
+ c,
_

d =
_

[f[
_
f
n
+ c d,
f

=
_

[f[
_
f
n
+ c d,
0 =
_

[f[
_
f
n
+ c d.
Partial Dierential Equations Igor Yanovsky, 2005 158
Thus, either [f[ 0 f = 0, or
f
n
+c = 0. Since f 0 as x , we have c = 0 f = 0.
Thus, solitary waves do not exist if n is even.
Partial Dierential Equations Igor Yanovsky, 2005 159
When n = 1, we have
(f

)
2
= f
3
+ cf
2
. (16.39)
We show that all conditions of the problem are satised provided c > 0, including
u = c sech
2
_
1
2

c(x ct)
_
, or
f = c sech
2
_

c
2
_
=
c
cosh
2
[

c
2
]
= c cosh
_

c
2
_
2
.
We have
f

= 2c cosh
_

c
2
_
3
sinh
_

c
2
_

c
2
= c

c cosh
_

c
2
_
3
sinh
_

c
2
_
,
(f

)
2
=
c
3
sinh
2
_

c
2
_
cosh
6
_

c
2
_ ,
f
3
=
c
3
cosh
6
_

c
2
_,
cf
2
=
c
3
cosh
4
_

c
2
_.
Plugging these into (16.39), we obtain:
39
c
3
sinh
2
_

c
2
_
cosh
6
_

c
2
_ =
c
3
cosh
6
_

c
2
_ +
c
3
cosh
4
_

c
2
_,
c
3
sinh
2
_

c
2
_
cosh
6
_

c
2
_ =
c
3
+ c
3
cosh
2
_

c
2
_
cosh
6
_

c
2
_ ,
c
3
sinh
2
_

c
2
_
cosh
6
_

c
2
_ =
c
3
sinh
2
_

c
2
_
cosh
6
_

c
2
_ .
Also, f, f

, f

0, as [[ , since
f() = c sech
2
_

c
2
_
=
c
cosh
2
[

c
2
]
= c
_
2
e
[

c
2
]
+e
[

c
2
]
_
2
0, as [[ .
Similarly, f

, f

0, as [[ .
39
cosh
2
x sinh
2
x = 1.
cosh x =
e
x
+ e
x
2
, sinh x =
e
x
e
x
2
Partial Dierential Equations Igor Yanovsky, 2005 160
Problem (S00, #5). Look for a traveling wave solution of the PDE
u
tt
+ (u
2
)
xx
= u
xxxx
of the form u(x, t) = v(x ct). In particular, you should nd an ODE for v. Under
the assumption that v goes to a constant as [x[ , describe the form of the solution.
Proof. Since (u
2
)
x
= 2uu
x
, and (u
2
)
xx
= 2u
2
x
+ 2uu
xx
, we have
u
tt
+ 2u
2
x
+ 2uu
xx
= u
xxxx
.
We look for a traveling wave solution
u(x, t) = v(x ct).
We get the ODE
c
2
v

+ 2(v

)
2
+ 2vv

= v

,
c
2
v

+ 2((v

)
2
+ vv

) = v

,
c
2
v

+ 2(vv

= v

, (exact dierentials)
c
2
v

+ 2vv

= v

+ a, s = x ct
c
2
v + v
2
= v

+as +b,
v

+c
2
v +v
2
= a(x ct) + b.
Since v C = const as [x[ , we have v

, v

## 0, as [x[ . Thus, implies

c
2
v +v
2
= as +b.
Since [x[ , but v C, we have a = 0:
v
2
+ c
2
v b = 0.
v =
c
2

c
4
+ 4b
2
.
Partial Dierential Equations Igor Yanovsky, 2005 161
Problem (S95, #2). Consider the KdV-Burgers equation
u
t
+uu
x
= u
xx
+u
xxx
in which > 0, > 0.
a) Find an ODE for traveling wave solutions of the form
u(x, t) = (x st)
with s > 0 and
lim
y
(y) = 0
and analyze the stationary points from this ODE.
b) Find the possible (nite) values of

+
= lim
y
(y).
Proof. a) We look for a traveling wave solution
u(x, t) = (x st), y = x st.
We get the ODE
s

,
s +
1
2

2
=

+ a.
Since 0 as y , then

0 as y . Therefore, at y = , a = 0.
We found the following ODE,

+
s

1
2

2
= 0.
In order to nd and analyze the stationary points of an ODE above, we write it as a
rst-order system.

1
= ,

2
=

1
=

=
2
,

2
=

+
1
2

2
=

1
+
1
2

2
1
.
_

1
=
2
= 0,

2
=

1
+
1
2

2
1
= 0;

_

1
=
2
= 0,

2
=
s

1
+
1
2

2
1
= 0;

_

1
=
2
= 0,

2
=
1

1
(s
1
2

1
) = 0.
Stationary points: (0, 0), (2s, 0), s > 0.

1
=
2
= f(
1
,
2
),

2
=

1
+
1
2

2
1
= g(
1
,
2
).
Partial Dierential Equations Igor Yanovsky, 2005 162
In order to classify a stationary point, need to nd eigenvalues of a linearized system
at that point.
J(f(
1
,
2
), g(
1
,
2
)) =
_
f

1
f

2
g

1
g

2
_
=
_
0 1

+
1

_
.
For (
1
,
2
) = (0, 0) :
det(J[
(0,0)
I) =

=
2
+

+
s

= 0.

=

2

_

2
4
2

s

.
If

2
4
> s

R,

< 0.
(0,0) is Stable Improper Node.
If

2
4
< s

C, Re(

) < 0.
(0,0) is Stable Spiral Point.
For (
1
,
2
) = (2s, 0) :
det(J[
(2s,0)
I) =

1
s

=
2
+

= 0.

=

2

_

2
4
2
+
s

.

+
> 0,

< 0.
b) Since
lim
y
(y) = 0 = lim
t
(x st),
we may have
lim
y+
(y) = lim
t
(x st) = 2s.
That is, a particle may start o at an unstable node (2s, 0) and as t increases, approach
the stable node (0, 0).
A phase diagram with (0, 0) being a stable spiral point, is shown below.
Partial Dierential Equations Igor Yanovsky, 2005 163
Partial Dierential Equations Igor Yanovsky, 2005 164
Problem (F95, #8). Consider the equation
u
t
+f(u)
x
= u
xx
where f is smooth and > 0. We seek traveling wave solutions to this equation,
i.e., solutions of the form u = (x st), under the boundary conditions
u u
L
and u
x
0 as x ,
u u
R
and u
x
0 as x +.
Find a necessary and sucient condition on f, u
L
, u
R
and s for such traveling waves
to exist; in case this condition holds, write an equation which denes implicitly.
Proof. We look for traveling wave solutions
u(x, t) = (x st), y = x st.
The boundary conditions become
u
L
and

0 as x ,
u
R
and

0 as x +.
Since f((x st))
x
= f

()

## , we get the ODE

s

+ f

()

,
s

+ (f())

,
s +f() =

+ a,

=
s + f()

+b.
We use boundary conditions to determine constant b:
At x = , 0 =

=
su
L
+ f(u
L
)

+b b =
su
L
f(u
L
)

.
At x = +, 0 =

=
su
R
+ f(u
R
)

+ b b =
su
R
f(u
R
)

.
s =
f(u
L
) f(u
R
)
u
L
u
R
.
40
40
For the solution for the second part of the problem, refer to Chiu-Yens solutions.
Partial Dierential Equations Igor Yanovsky, 2005 165
Problem (S02, #5; F90, #2). Fishers Equation. Consider
u
t
= u(1 u) +u
xx
, < x < , t > 0.
The solutions of physical interest satisfy 0 u 1, and
lim
x
u(x, t) = 0, lim
x+
u(x, t) = 1.
One class of solutions is the set of wavefront solutions. These have the form u(x, t) =
(x +ct), c 0.
Determine the ordinary dierential equation and boundary conditions which must
satisfy (to be of physical interest). Carry out a phase plane analysis of this equation,
and show that physically interesting wavefront solutions are possible if c 2, but not if
0 c < 2.
Proof. We look for a traveling wave solution
u(x, t) = (x + ct), s = x +ct.
We get the ODE
c

= (1 ) +

+
2
= 0,
(s) 0, as s ,
(s) 1, as s +,
0 1.
In order to nd and analyze the stationary points of an ODE above, we write it as a
rst-order system.
y
1
= ,
y
2
=

.
y

1
=

= y
2
,
y

2
=

= c

+
2
= cy
2
y
1
+y
2
1
.
_
y

1
= y
2
= 0,
y

2
= cy
2
y
1
+ y
2
1
= 0;

_
y
2
= 0,
y
1
(y
1
1) = 0.
Stationary points: (0, 0), (1, 0).
y

1
= y
2
= f(y
1
, y
2
),
y

2
= cy
2
y
1
+y
2
1
= g(y
1
, y
2
).
In order to classify a stationary point, need to nd eigenvalues of a linearized system
at that point.
J(f(y
1
, y
2
), g(y
1
, y
2
)) =
_
f
y
1
f
y
2
g
y
1
g
y
2
_
=
_
0 1
2y
1
1 c
_
.
Partial Dierential Equations Igor Yanovsky, 2005 166
For (y
1
, y
2
) = (0, 0) :
det(J[
(0,0)
I) =

1
1 c

=
2
c + 1 = 0.

=
c

c
2
4
2
.
If c 2

R,

> 0.
(0,0) is Unstable Improper (c > 2) / Proper (c = 2) Node.
If 0 c < 2

C, Re(

) 0.
(0,0) is Unstable Spiral Node.
For (y
1
, y
2
) = (1, 0) :
det(J[
(1,0)
I) =

1
1 c

=
2
c 1 = 0.

=
c

c
2
+ 4
2
.
If c 0
+
> 0,

< 0.
By looking at the phase plot, a particle may start o at an unstable node (0, 0) and as
t increases, approach the unstable node (1, 0).
Partial Dierential Equations Igor Yanovsky, 2005 167
Partial Dierential Equations Igor Yanovsky, 2005 168
Problem (F99, #6). For the system

t
+
x
u = 0

t
u +
x
(u) =
2
x
u
look for traveling wave solutions of the form (x, t) = (y = xst), u(x, t) = u(y =
x st). In particular
a) Find a rst order ODE for u.
b) Show that this equation has solutions of the form
u(y) = u
0
+ u
1
tanh(y +y
0
),
for some constants u
0
, u
1
, , y
0
.
Proof. a) We rewrite the system:

t
+ u
x
= 0
u
t
+
x
u +u
x
= u
xx
We look for traveling wave solutions
(x, t) = (x st), u(x, t) = u(x st), y = x st.
We get the system of ODEs
_
s

+ u

= 0,
su

u + u

= u

.
The rst ODE gives

=
1
s
u

,
=
1
s
u +a,
where a is a constant, and integration was done with respect to y. The second ODE
gives
su

+
1
s
u

u +
_
1
s
u + a
_
u

= u

,
su

+
2
s
uu

+au

= u

. Integrating, we get
su +
1
s
u
2
+ au = u

+b.
u

=
1
s
u
2
+ (a s)u b.
b) Note that the ODE above may be written in the following form:
u

+ Au
2
+Bu = C,
which is a nonlinear rst order equation.
Partial Dierential Equations Igor Yanovsky, 2005 169
Problem (S01, #7). Consider the following system of PDEs:
f
t
+f
x
= g
2
f
2
g
t
g
x
= f
2
g
a) Find a system of ODEs that describes traveling wave solutions of the PDE
system; i.e. for solutions of the form f(x, t) = f(x st) and g(x, t) = g(x st).
b) Analyze the stationary points and draw the phase plane for this ODE system in the
standing wave case s = 0.
Proof. a) We look for traveling wave solutions
f(x, t) = f(x st), g(x, t) = g(x st).
We get the system of ODEs
sf

+ f

= g
2
f
2
,
sg

= f
2
g.
Thus,
f

=
g
2
f
2
1 s
,
g

=
f
2
g
1 s
.
b) If s = 0, the system becomes
_
f

= g
2
f
2
,
g

= g f
2
.
Relabel the variables f y
1
, g y
2
.
_
y

1
= y
2
2
y
2
1
= 0,
y

2
= y
2
y
2
1
= 0.
Stationary points: (0, 0), (1, 1), (1, 1).
_
y

1
= y
2
2
y
2
1
= (y
1
, y
2
),
y

2
= y
2
y
2
1
= (y
1
, y
2
).
In order to classify a stationary point, need to nd eigenvalues of a linearized system
at that point.
J((y
1
, y
2
), (y
1
, y
2
)) =
_

y
1

y
2

y
1

y
2
_
=
_
2y
1
2y
2
2y
1
1
_
.
For (y
1
, y
2
) = (0, 0) :
det(J[
(0,0)
I) =

0
0 1

= (1 ) = 0.

1
= 0,
2
= 1; eigenvectors: v
1
=
_
1
0
_
, v
2
=
_
0
1
_
.
(0,0) is Unstable Node.
Partial Dierential Equations Igor Yanovsky, 2005 170
For (y
1
, y
2
) = (1, 1) :
det(J[
(1,1)
I) =

2 2
2 1

=
2
3 2 = 0.

=
3
2

17
2
.

< 0,
+
> 0.
For (y
1
, y
2
) = (1, 1) :
det(J[
(1,1)
I) =

2 2
2 1

=
2
+ + 2 = 0.

=
1
2
i

7
2
.
Re(

) < 0.
(1,1) is Stable Spiral Point.
Partial Dierential Equations Igor Yanovsky, 2005 171
16.5 Dispersion
Problem (S97, #8). Consider the following equation
u
t
= (f(u
x
))
x
u
xxxx
, f(v) = v
2
v, (16.40)
with constant .
a) Linearize this equation around u = 0 and nd the principal mode solution of the
form e
t+ikx
. For which values of are there unstable modes, i.e., modes with = 0
for real k? For these values, nd the maximally unstable mode, i.e., the value of k with
the largest positive value of .
b) Consider the steady solution of the (fully nonlinear) problem. Show that the resulting
equation can be written as a second order autonomous ODE for v = u
x
and draw the
corresponding phase plane.
Proof. a) We have
u
t
= (f(u
x
))
x
u
xxxx
,
u
t
= (u
2
x
u
x
)
x
u
xxxx
,
u
t
= 2u
x
u
xx
u
xx
u
xxxx
.
However, we need to linearize (16.40) around u = 0. To do this, we need to linearize f.
f(u) = f(0) + uf

(0) +
u
2
2
f

(0) + = 0 +u(0 1) + = u + .
Thus, we have
u
t
= u
xx
u
xxxx
.
Consider u(x, t) = e
t+ikx
.
e
t+ikx
= (k
2
k
4
)e
t+ikx
,
= k
2
k
4
.
To nd unstable nodes, we set = 0, to get
=
1
k
2
.
To nd the maximally unstable mode, i.e., the value of k with the largest positive
value of , consider
(k) = k
2
k
4
,

(k) = 2k 4k
3
.
To nd the extremas of , we set

## = 0. Thus,the extremas are at

k
1
= 0, k
2,3
=
_
1
2
.
To nd if the extremas are maximums or minimums, we set

= 0:

(k) = 2 12k
2
= 0,

## (0) = 2 > 0 k = 0 is the minimum.

_
1
2
_
= 4 < 0 k =
_
1
2
is the maximum unstable mode.

_
1
2
_
=
1
4
is the largest positive value of .
Partial Dierential Equations Igor Yanovsky, 2005 172
b) Integrating , we get
u
2
x
u
x
u
xxx
= 0.
Let v = u
x
. Then,
v
2
v v
xx
= 0, or
v

=
v
2
v

.
In order to nd and analyze the stationary points of an ODE above, we write it as a
rst-order system.
y
1
= v,
y
2
= v

.
y

1
= v

= y
2
,
y

2
= v

=
v
2
v

=
y
2
1
y
1

.
_
y

1
= y
2
= 0,
y

2
=
y
2
1
y
1

= 0;

_
y
2
= 0,
y
1
(y
1
1) = 0.
Stationary points: (0, 0), (1, 0).
y

1
= y
2
= f(y
1
, y
2
),
y

2
=
y
2
1
y
1

= g(y
1
, y
2
).
In order to classify a stationary point, need to nd eigenvalues of a linearized system
at that point.
J(f(y
1
, y
2
), g(y
1
, y
2
)) =
_
f
y
1
f
y
2
g
y
1
g
y
2
_
=
_
0 1
2y
1
1

0
_
.
For (y
1
, y
2
) = (0, 0),

=
_

.
If < 0,

R,
+
> 0,

If > 0,

= i
_
1

C, Re(

For (y
1
, y
2
) = (1, 0),

=
_
1

.
If < 0,

= i
_

C, Re(

If > 0,

R,
+
> 0,

## < 0. (1,0) is Unstable Saddle Point.

Partial Dierential Equations Igor Yanovsky, 2005 173
Partial Dierential Equations Igor Yanovsky, 2005 174
16.6 Energy Methods
Problem (S98, #9; S96, #5). Consider the following initial-boundary value
problem for the multi-dimensional wave equation:
u
tt
= u in (0, ),
u(x, 0) = f(x),
u
t
(x, 0) = g(x) for x ,
u
n
+a(x)
u
t
= 0 on .
Here, is a bounded domain in R
n
and a(x) 0. Dene the Energy integral for this
problem and use it in order to prove the uniqueness of the classical solution of the prob-
lem.
Proof.
d

E
dt
= 0 =
_

(u
tt
u)u
t
dx =
_

u
tt
u
t
dx
_

u
n
u
t
ds +
_

u u
t
dx
=
_

1
2

t
(u
2
t
) dx +
_

1
2

t
[u[
2
dx +
_

a(x)u
2
t
ds.
Thus,

a(x)u
2
t
dx
. .
0
=
1
2

t
_

u
2
t
+[u[
2
dx.
Let Energy integral be
E(t) =
1
2
_

u
2
t
+[u[
2
dx.
In order to prove that the given E(t) 0 from scratch, take its derivative with respect
to t:
dE
dt
(t) =
_

_
u
t
u
tt
+ u u
t
_
dx
=
_

u
t
u
tt
dx +
_

u
t
u
n
ds
_

u
t
u dx
=
_

u
t
(u
tt
u) dx
. .
=0

a(x)u
2
t
dx 0.
Thus, E(t) E(0).
To prove the uniqueness of the classical solution, suppose u
1
and u
2
are two solutions
of the initial boundary value problem. Let w = u
1
u
2
. Then, w satises
w
tt
= w in (0, ),
w(x, 0) = 0, w
t
(x, 0) = 0 for x ,
w
n
+a(x)
w
t
= 0 on .
We have
E
w
(0) =
1
2
_

(w
t
(x, 0)
2
+ [w(x, 0)[
2
) dx = 0.
Partial Dierential Equations Igor Yanovsky, 2005 175
E
w
(t) E
w
(0) = 0 E
w
(t) = 0. Thus, w
t
= 0, w
x
i
= 0 w(x, t) = const = 0.
Hence, u
1
= u
2
.
Problem (S94, #7). Consider the wave equation
1
c
2
(x)
u
tt
= u x
u
t
(x)
u
n
= 0 on R.
Assume that (x) is of one sign for all x (i.e. always positive or always negative).
For the energy
E(t) =
1
2
_

1
c
2
(x)
u
2
t
+ [u[
2
dx,
show that the sign of
dE
dt
is determined by the sign of .
Proof. We have
dE
dt
(t) =
_

_
1
c
2
(x)
u
t
u
tt
+ u u
t
_
dx
=
_

1
c
2
(x)
u
t
u
tt
dx +
_

u
t
u
n
ds
_

u
t
u dx
=
_

u
t
_
1
c
2
(x)
u
tt
u
_
dx
. .
=0
+
_

1
(x)
u
2
t
dx
=
_

1
(x)
u
2
t
dx =
_
> 0, if (x) > 0, x ,
< 0, if (x) < 0, x .
Partial Dierential Equations Igor Yanovsky, 2005 176
Problem (F92, #2). Let R
n
. Let u(x, t) be a smooth solution of the following
initial boundary value problem:
u
tt
u + u
3
= 0 for (x, t) [0, T]
u(x, t) = 0 for (x, t) [0, T].
a) Derive an energy equality for u. (Hint: Multiply by u
t
and integrate over
[0, T].)
b) Show that if u[
t=0
= u
t
[
t=0
= 0 for x , then u 0.
Proof. a) Multiply by u
t
and integrate:
0 =
_

(u
tt
u + u
3
)u
t
dx =
_

u
tt
u
t
dx
_

u
n
u
t
ds
. .
=0
+
_

u u
t
dx +
_

u
3
u
t
dx
=
_

1
2

t
(u
2
t
) dx +
_

1
2

t
[u[
2
dx +
_

1
4

t
(u
4
) dx =
1
2
d
dt
_

_
u
2
t
+ [u[
2
+
1
2
u
4
_
dx.
Thus, the Energy integral is
E(t) =
_

_
u
2
t
+[u[
2
+
1
2
u
4
_
dx = const = E(0).
b) Since u(x, 0) = 0, u
t
(x, 0) = 0, we have
E(0) =
_

_
u
t
(x, 0)
2
+[u(x, 0)[
2
+
1
2
u(x, 0)
4
_
dx = 0.
Since E(t) = E(0) = 0, we have
E(t) =
_

_
u
t
(x, t)
2
+ [u(x, t)[
2
+
1
2
u(x, t)
4
_
dx = 0.
Thus, u 0.
Partial Dierential Equations Igor Yanovsky, 2005 177
Problem (F04, #3). Consider a damped wave equation
_
u
tt
u + a(x)u
t
= 0, (x, t) R
3
R,
u[
t=0
= u
0
, u
t
[
t=0
= u
1
.
Here the damping coecient a C

0
(R
3
) is a non-negative function and u
0
, u
1

C

0
(R
3
). Show that the energy of the solution u(x, t) at time t,
E(t) =
1
2
_
R
3
_
[
x
u[
2
+ [u
t
[
2
_
dx
is a decreasing function of t 0.
Proof. Take the derivative of E(t) with respect to t. Note that the boundary integral
is 0 by Huygens principle.
dE
dt
(t) =
_
R
3
_
u u
t
+ u
t
u
tt
_
dx
=
_
R
3
u
t
u
n
ds
. .
=0

_
R
3
u
t
u dx +
_
R
3
u
t
u
tt
dx
=
_
R
3
u
t
(u + u
tt
) dx =
_
R
3
u
t
(a(x)u
t
) dx =
_
R
3
a(x)u
2
t
dx 0.
Thus,
dE
dt
0 E(t) E(0), i.e. E(t) is a decreasing function of t.
Partial Dierential Equations Igor Yanovsky, 2005 178
Problem (W03, #8). a) Consider the damped wave equation for high-speed waves
(0 < << 1) in a bounded region D

2
u
tt
+u
t
= u
with the boundary condition u(x, t) = 0 on D. Show that the energy functional
E(t) =
_
D

2
u
2
t
+ [u[
2
dx
is nonincreasing on solutions of the boundary value problem.
b) Consider the solution to the boundary value problem in part (a) with initial data
u

(x, 0) = 0, u

t
(x, 0) =

## f(x), where f does not depend on and < 1. Use part

(a) to show that
_
D
[u

(x, t)[
2
dx 0
uniformly on 0 t T for any T as 0.
c) Show that the result in part (b) does not hold for = 1. To do this consider
the case where f is an eigenfunction of the Laplacian, i.e. f + f = 0 in D and
f = 0 on D, and solve for u

explicitly.
Proof. a)
dE
dt
=
_
D
2
2
u
t
u
tt
dx +
_
D
2u u
t
dx
=
_
D
2
2
u
t
u
tt
dx +
_
D
2
u
n
u
t
ds
. .
=0, (u=0 on D)

_
D
2uu
t
dx
= 2
_
D
(
2
u
tt
u)u
t
dx = = 2
_
D
[u
t
[
2
dx 0.
Thus, E(t) E(0), i.e. E(t) is nonincreasing.
b) From (a), we know
dE
dt
0. We also have
E

(0) =
_
D

2
(u

t
(x, 0))
2
+[u

(x, 0)[
2
dx
=
_
D

2
(

f(x))
2
+ 0 dx =
_
D

2(1)
f(x)
2
dx 0 as 0.
Since E

(0) E

(t) =
_
D

2
(u

t
)
2
+ [u

[
2
dx, then E

(t) 0 as 0.
Thus,
_
D
[u

[
2
dx 0 as 0.
c) If = 1,
E

(0) =
_
D

2(1)
f(x)
2
dx =
_
D
f(x)
2
dx.
Since f is independent of , E

## (0) does not approach 0 as 0. We can not conclude

that
_
D
[u

(x, t)[
2
dx 0.
Partial Dierential Equations Igor Yanovsky, 2005 179
Problem (F98, #6). Let f solve the nonlinear wave equation
f
tt
f
xx
= f(1 +f
2
)
1
for x [0, 1], with f(x = 0, t) = f(x = 1, t) = 0 and with smooth initial data f(x, t) =
f
0
(x).
a) Find an energy integral E(t) which is constant in time.
b) Show that [f(x, t)[ < c for all x and t, in which c is a constant.
Hint: Note that
f
1 +f
2
=
1
2
d
df
log(1 + f
2
).
Proof. a) Since f(0, t) = f(1, t) = 0, t, we have f
t
(0, t) = f
t
(1, t) = 0. Let
dE
dt
= 0 =
_
1
0
_
f
tt
f
xx
+f(1 +f
2
)
1
_
f
t
dx
=
_
1
0
f
tt
f
t
dx
_
1
0
f
xx
f
t
dx +
_
1
0
ff
t
1 + f
2
dx
=
_
1
0
f
tt
f
t
dx [f
x
f
t
..
=0
]
1
0
+
_
1
0
f
x
f
tx
dx +
_
1
0
ff
t
1 +f
2
dx
=
_
1
0
1
2

t
(f
2
t
) dx +
_
1
0
1
2

t
(f
2
x
) dx +
_
1
0
1
2

t
(ln(1 + f
2
)) dx
=
1
2
d
dt
_
1
0
_
f
2
t
+f
2
x
+ ln(1 +f
2
)
_
dx.
Thus,
E(t) =
1
2
_
1
0
_
f
2
t
+ f
2
x
+ ln(1 + f
2
)
_
dx.
b) We want to show that f is bounded. For smooth f(x, 0) = f
0
(x), we have
E(0) =
1
2
_
1
0
_
f
t
(x, 0)
2
+f
x
(x, 0)
2
+ ln(1 +f(x, 0)
2
)
_
dx < .
Since E(t) is constant in time, E(t) = E(0) < . Thus,
1
2
_
1
0
ln(1 +f
2
) dx
1
2
_
1
0
_
f
2
t
+f
2
x
+ ln(1 +f
2
)
_
dx = E(t) < .
Hence, f is bounded.
Partial Dierential Equations Igor Yanovsky, 2005 180
Problem (F97, #1). Consider initial-boundary value problem
u
tt
+ a
2
(x, t)u
t
u(x, t) = 0 x R
n
, 0 < t < +
u(x) = 0 x
u(x, 0) = f(x), u
t
(x, 0) = g(x) x .
Prove that L
2
-norm of the solution is bounded in t on (0, +).
Here is a bounded domain, and a(x, t), f(x), g(x) are smooth functions.
Proof. Multiply the equation by u
t
and integrate over :
u
t
u
tt
+a
2
u
2
t
u
t
u = 0,
_

u
t
u
tt
dx +
_

a
2
u
2
t
dx
_

u
t
u dx = 0,
1
2
d
dt
_

u
2
t
dx +
_

a
2
u
2
t
dx
_

u
t
u
n
ds
. .
=0, (u=0, x)
+
_

u u
t
dx = 0,
1
2
d
dt
_

u
2
t
dx +
_

a
2
u
2
t
dx +
1
2
d
dt
_

[u[
2
dx = 0,
1
2
d
dt
_

_
u
2
t
+[u[
2
_
dx =
_

a
2
u
2
t
dx 0.
Let Energy integral be
E(t) =
_

_
u
2
t
+[u[
2
_
dx.
We have
dE
dt
0, i.e. E(t) E(0).
E(t) E(0) =
_

_
u
t
(x, 0)
2
+ [u(x, 0)[
2
_
dx =
_

_
g(x)
2
+ [f(x)[
2
_
dx < ,
since f and g are smooth functions. Thus,
E(t) =
_

_
u
2
t
+ [u[
2
_
dx < ,
_

[u[
2
dx < ,
_

u
2
dx C
_

[u[
2
dx < , by Poincare inequality.
Thus, [[u[[
2
is bounded t.
Partial Dierential Equations Igor Yanovsky, 2005 181
Problem (S98, #4). a) Let u(x, y, z, t), < x, y, z < be a solution of the
equation

u
tt
+u
t
= u
xx
+u
yy
+u
zz
u(x, y, z, 0) = f(x, y, z),
u
t
(x, y, z, 0) = g(x, y, z).
(16.41)
Here f, g are smooth functions which vanish if
_
x
2
+y
2
+z
2
is large enough. Prove
that it is the unique solution for t 0.
b) Suppose we want to solve the same equation (16.41) in the region z 0, <
x, y < , with the additional conditions
u(x, y, 0, t) = f(x, y, t)
u
z
(x, y, 0, t) = g(x, y, t)
with the same f, g as before in (16.41). What goes wrong?
Proof. a) Suppose u
1
and u
2
are two solutions. Let w = u
1
u
2
. Then,

w
tt
+ w
t
= w,
w(x, y, z, 0) = 0,
w
t
(x, y, z, 0) = 0.
Multiply the equation by w
t
and integrate:
w
t
w
tt
+ w
2
t
= w
t
w,
_
R
3
w
t
w
tt
dx +
_
R
3
w
2
t
dx =
_
R
3
w
t
wdx,
1
2
d
dt
_
R
3
w
2
t
dx +
_
R
3
w
2
t
dx =
_
R
3
w
t
w
n
dx
. .
=0

_
R
3
w w
t
dx,
1
2
d
dt
_
R
3
w
2
t
dx +
_
R
3
w
2
t
dx =
1
2
d
dt
_
R
3
[w[
2
dx,
d
dt
_
R
3
_
w
2
t
+ [w[
2
_
dx
. .
E(t)
= 2
_
R
3
w
2
t
dx 0,
dE
dt
0,
E(t) E(0) =
_
R
3
_
w
t
(x, 0)
2
+ [w(x, 0)[
2
_
dx = 0,
E(t) =
_
R
3
_
w
2
t
+ [w[
2
_
dx = 0.
Thus, w
t
= 0, w = 0, and w = constant. Since w(x, y, z, 0) = 0, we have w 0.
b)
Partial Dierential Equations Igor Yanovsky, 2005 182
Problem (F94, #8). The one-dimensional, isothermal uid equations with viscosity
and capillarity in Lagrangian variables are
v
t
u
x
= 0
u
t
+p(v)
x
= u
xx
v
xxx
in which v(= 1/) is specic volume, u is velocity, and p(v) is pressure. The coecients
and are non-negative.
Find an energy integral which is non-increasing (as t increases) if > 0 and con-
served if = 0.
Hint: if = 0, E =
_
u
2
/2 P(v) dx where P

(v) = p(v).
Proof. Multiply the second equation by u and integrate over R. We use u
x
= v
t
.
Note that the boundary integrals are 0 due to nite speed of propagation.
uu
t
+up(v)
x
= uu
xx
uv
xxx
,
_
R
uu
t
dx +
_
R
up(v)
x
dx =
_
R
uu
xx
dx
_
R
uv
xxx
dx,
1
2
_
R

t
(u
2
) dx +
_
R
up(v) ds
. .
=0
+
_
R
u
x
p(v) dx
=
_
R
uu
x
dx
. .
=0

_
R
u
2
x
dx
_
R
uv
xx
dx
. .
=0
+
_
R
u
x
v
xx
dx,
1
2
_
R

t
(u
2
) dx +
_
R
v
t
p(v) dx =
_
R
u
2
x
dx +
_
R
v
t
v
xx
dx,
1
2
_
R

t
(u
2
) dx +
_
R

t
P(v) dx =
_
R
u
2
x
dx +
_
R
v
t
v
x
dx
. .
=0

_
R
v
xt
v
x
dx,
1
2
_
R

t
(u
2
) dx +
_
R

t
P(v) dx +

2
_
R

t
(v
2
x
) dx =
_
R
u
2
x
dx,
d
dt
_
R
_
u
2
2
+ P(v) +

2
v
2
x
_
dx =
_
R
u
2
x
dx 0.
E(t) =
_
R
_
u
2
2
+ P(v) +

2
v
2
x
_
dx
is nonincreasing if > 0, and conserved if = 0.
Partial Dierential Equations Igor Yanovsky, 2005 183
Problem (S99, #5). Consider the equation
u
tt
=

x
(u
x
) (16.42)
with (z) a smooth function. This is to be solved for t > 0, 0 x 1, with
periodic boundary conditions and initial data u(x, 0) = u
0
(x) and u
t
(x, 0) = v
0
(x).
a) Multiply (16.42) by u
t
and get an expression of the form
d
dt
_
1
0
F(u
t
, u
x
) = 0
that is satised for an appropriate function F(y, z) with y = u
t
, z = u
x
,
where u is any smooth, periodic in space solution of (16.42).
b) Under what conditions on (z) is this function, F(y, z), convex in its variables?
c) What ` a priori inequality is satised for smooth solutions when F is convex?
d) Discuss the special case (z) = a
2
z
3
/3, with a > 0 and constant.
Proof. a) Multiply by u
t
and integrate:
u
t
u
tt
= u
t
(u
x
)
x
,
_
1
0
u
t
u
tt
dx =
_
1
0
u
t
(u
x
)
x
dx,
d
dt
_
1
0
u
2
t
2
dx = u
t
(u
x
)

1
0
. .
=0, (2-periodic)

_
1
0
u
tx
(u
x
) dx =
Let Q

## (z) = (z), then

d
dt
Q(u
x
) = (u
x
)u
xt
. Thus,
=
_
1
0
u
tx
(u
x
) dx =
d
dt
_
1
0
Q(u
x
) dx.
d
dt
_
1
0
_
u
2
t
2
+ Q(u
x
)
_
dx = 0.
b) We have
F(u
t
, u
x
) =
u
2
t
2
+Q(u
x
).
41
For F to be convex, the Hessian matrix of partial derivatives must be positive denite.
41
A function f is convex on a convex set S if it satises
f(x + (1 )y) f(x) + (1 )f(y)
for all 0 1 and for all x, y S.
If a one-dimensional function f has two continuous derivatives, then f is convex if and only if
f

(x) 0.
In the multi-dimensional case the Hessian matrix of second derivatives must be positive semi-denite,
that is, at every point x S
y
T

2
f(x) y 0, for all y.
The Hessian matrix is the matrix with entries
[
2
f(x)]ij

2
f(x)
xixj
.
For functions with continuous second derivatives, it will always be symmetric matrix: fx
i
x
j
= fx
j
x
i
.
Partial Dierential Equations Igor Yanovsky, 2005 184
The Hessian matrix is

2
F(u
t
, u
x
) =
_
F
utut
F
utux
F
uxut
F
uxux
_
=
_
1 0
0

(u
x
)
_
.
y
T

2
F(x) y =
_
y
1
y
2
_
_
1 0
0

(u
x
)
__
y
1
y
2
_
= y
2
1
+

(u
x
)y
2
2

..
need
0.
Thus, for a Hessian matrix to be positive denite, need

(u
x
) 0, so that the above
inequality holds for all y.
c) We have
d
dt
_
1
0
F(u
t
, u
x
) dx = 0,
_
1
0
F(u
t
, u
x
) dx = const,
_
1
0
F(u
t
, u
x
) dx =
_
1
0
F(u
t
(x, 0), u
x
(x, 0)) dx,
_
1
0
_
u
2
t
2
+ Q(u
x
)
_
dx =
_
1
0
_
v
2
0
2
+ Q(u
0x
)
_
dx.
d) If (z) = a
2
z
3
/3, we have
F(u
t
, u
x
) =
u
2
t
2
+Q(u
x
) =
u
2
t
2
+
a
2
u
4
x
12
,
d
dt
_
1
0
_
u
2
t
2
+
a
2
u
4
x
12
_
dx = 0,
_
1
0
_
u
2
t
2
+
a
2
u
4
x
12
_
dx = const,
_
1
0
_
u
2
t
2
+
a
2
u
4
x
12
_
dx =
_
1
0
_
v
0
2
2
+
a
2
u
0
4
x
12
_
dx.
Partial Dierential Equations Igor Yanovsky, 2005 185
Problem (S96, #8).
42
Let u(x, t) be the solution of the Korteweg-de Vries equation
u
t
+uu
x
= u
xxx
, 0 x 2,
with 2-periodic boundary conditions and prescribed initial data
u(x, t = 0) = f(x).
a) Prove that the energy integral
I
1
(u) =
_
2
0
u
2
(x, t) dx
is independent of the time t.
b) Prove that the second energy integral,
I
2
(u) =
_
2
0
_
1
2
u
2
x
(x, t) +
1
6
u
3
(x, t)
_
dx
is also independent of the time t.
c) Assume the initial data are such that I
1
(f) + I
2
(f) < . Use (a) + (b) to prove
that the maximum norm of the solution, [u[

= sup
x
[u(x, t)[, is bounded in time.
Hint: Use the following inequalities (here, [u[
p
is the L
p
-norm of u(x, t) at xed time
t):
[u[
2

6
([u[
2
2
+[u
x
[
2
2
) (one of Sobolevs inequalities);
[u[
3
3
[u[
2
2
[u[

(straightforward).
Proof. a) Multiply the equation by u and integrate. Note that all boundary terms are
0 due to 2-periodicity.
uu
t
+u
2
u
x
= uu
xxx
,
_
2
0
uu
t
dx +
_
2
0
u
2
u
x
dx =
_
2
0
uu
xxx
dx,
1
2
d
dt
_
2
0
u
2
dx +
1
3
_
2
0
(u
3
)
x
dx = uu
xx

2
0

_
2
0
u
x
u
xx
dx,
1
2
d
dt
_
2
0
u
2
dx +
1
3
u
3

2
0
=
1
2
_
2
0
(u
2
x
)
x
dx,
1
2
d
dt
_
2
0
u
2
dx =
1
2
u
2
x

2
0
= 0.
I
1
(u) =
_
2
0
u
2
dx = C.
Thus, I
1
(u) =
_
2
0
u
2
(x, t) dx is independent of the time t.
Alternatively, we may dierentiate I
1
(u):
dI
1
dt
(u) =
d
dt
_
2
0
u
2
dx =
_
2
0
2uu
t
dx =
_
2
0
2u(uu
x
+ u
xxx
) dx
=
_
2
0
2u
2
u
x
dx +
_
2
0
2uu
xxx
dx =
_
2
0

2
3
(u
3
)
x
dx + 2uu
xx

2
0

_
2
0
2u
x
u
xx
dx
=
2
3
u
3

2
0

_
2
0
(u
2
x
)
x
dx = u
2
x

2
0
= 0.
42
Also, see S92, #7.
Partial Dierential Equations Igor Yanovsky, 2005 186
b) Note that all boundary terms are 0 due to 2-periodicity.
dI
2
dt
(u) =
d
dt
_
2
0
_
1
2
u
2
x
+
1
6
u
3
_
dx =
_
2
0
_
u
x
u
xt
+
1
2
u
2
u
t
_
dx =
We dierentiate the original equation with respect to x:
u
t
= uu
x
+ u
xxx
u
tx
= (uu
x
)
x
+u
xxxx
.
=
_
2
0
u
x
((uu
x
)
x
+u
xxxx
) dx +
1
2
_
2
0
u
2
(uu
x
+u
xxx
) dx
=
_
2
0
u
x
(uu
x
)
x
dx +
_
2
0
u
x
u
xxxx
dx
1
2
_
2
0
u
3
u
x
dx +
1
2
_
2
0
u
2
u
xxx
dx
= u
x
uu
x

2
0
+
_
2
0
u
xx
uu
x
dx +u
x
u
xxx

2
0

_
2
0
u
xx
u
xxx
dx

1
2
_
2
0
_
u
4
4
_
x
dx +
1
2
u
2
u
xx

2
0

1
2
_
2
0
2uu
x
u
xx
dx
=
_
2
0
u
xx
uu
x
dx
_
2
0
u
xx
u
xxx
dx
1
2
u
4
4

2
0

_
2
0
uu
x
u
xx
dx
=
_
2
0
u
xx
u
xxx
dx = u
2
xx

2
0
+
_
2
0
u
xxx
u
xx
dx =
_
2
0
u
xxx
u
xx
dx = 0,
since
_
2
0
u
xx
u
xxx
dx = +
_
2
0
u
xx
u
xxx
dx. Thus,
I
2
(u) =
_
2
0
_
1
2
u
2
x
(x, t) +
1
6
u
3
(x, t)
_
dx = C,
and I
2
(u) is independent of the time t.
c) From (a) and (b), we have
I
1
(u) =
_
2
0
u
2
dx = [[u[[
2
2
,
I
2
(u) =
_
2
0
_
1
2
u
2
x
+
1
6
u
3
_
dx =
1
2
[[u
x
[[
2
2
+
1
6
[[u[[
3
3
.
Using given inequalities, we have
[[u[[
2

6
([[u[[
2
2
+ [[u
x
[[
2
2
)

6
_
I
1
(u) + 2I
2
(u)
1
3
[[u[[
3
3
_

6
I
1
(u) +

3
I
2
(u) +

18
[[u[[
2
2
[[u[[

6
I
1
(u) +

3
I
2
(u) +

18
I
1
(u)[[u[[

= C +C
1
[[u[[

.
[[u[[
2

C + C
1
[[u[[

,
[[u[[

C
2
.
Thus, [[u[[

is bounded in time.
Also see Energy Methods problems for higher order equations (3rd and
4th) in the section on Gas Dynamics.
Partial Dierential Equations Igor Yanovsky, 2005 187
16.7 Wave Equation in 2D and 3D
Problem (F97, #8); (McOwen 3.2 #90). Solve
u
tt
= u
xx
+u
yy
+u
zz
with initial conditions
u(x, y, z, 0) = x
2
+y
2
. .
g(x)
, u
t
(x, y, z, 0) = 0
..
h(x)
.
Proof.
We may use the Kirchhos formula:
u(x, t) =
1
4

t
_
t
_
||=1
g(x + ct) dS

_
+
t
4
_
||=1
h(x +ct) dS

=
1
4

t
_
t
_
||=1
_
(x
1
+ ct
1
)
2
+ (x
2
+ct
2
)
2
_
dS

_
+ 0 =
We may solve the problem by Hadamards method of descent, since initial con-
ditions are independent of x
3
. We need to convert surface integrals in R
3
to domain
integrals in R
2
. Specically, we need to express the surface measure on the upper half
of the unit sphere S
2
+
in terms of the two variables
1
and
2
. To do this, consider
f(
1
,
2
) =
_
1
2
1

2
2
over the unit disk
2
1
+
2
2
< 1.
dS

=
_
1 + (f

1
)
2
+ (f

2
)
2
d
1
d
2
=
d
1
d
2
_
1
2
1

2
2
.
Partial Dierential Equations Igor Yanovsky, 2005 188
u(x
1
, x
2
, t) =
1
4

t
_
2t
_

2
1
+
2
2
<1
g(x
1
+ct
1
, x
2
+ct
2
) d
1
d
2
_
1
2
1

2
2
_
+
t
4
_
2
_

2
1
+
2
2
<1
h(x
1
+ ct
1
, x
2
+ ct
2
) d
1
d
2
_
1
2
1

2
2
_
=
1
4

t
_
2t
_

2
1
+
2
2
<1
(x
1
+t
1
)
2
+ (x
2
+ t
2
)
2
_
1
2
1

2
2
d
1
d
2
_
+ 0,
=
1
2

t
_
t
_

2
1
+
2
2
<1
x
2
1
+ 2x
1
t
1
+ t
2

2
1
+x
2
2
+ 2x
2
t
2
+t
2

2
2
_
1
2
1

2
2
d
1
d
2
_
=
1
2

t
__

2
1
+
2
2
<1
tx
2
1
+ 2x
1
t
2

1
+t
3

2
1
+ tx
2
2
+ 2x
2
t
2

2
+t
3

2
2
_
1
2
1

2
2
d
1
d
2
_
=
1
2
__

2
1
+
2
2
<1
x
2
1
+ 4x
1
t
1
+ 3t
2

2
1
+x
2
2
+ 4x
2
t
2
+ 3t
2

2
2
_
1
2
1

2
2
d
1
d
2
_
=
1
2
__

2
1
+
2
2
<1
(x
2
1
+ x
2
2
) + 4t(x
1

1
+x
2

2
) + 3t
2
(
2
1
+
2
2
)
_
1
2
1

2
2
d
1
d
2
_
=
1
2
(x
2
1
+x
2
2
)
_

2
1
+
2
2
<1
d
1
d
2
_
1
2
1

2
2
. .

+
4t
2
_

2
1
+
2
2
<1
x
1

1
+ x
2

2
_
1
2
1

2
2
d
1
d
2
. .

+
3t
2
2
_

2
1
+
2
2
<1

2
1
+
2
2
_
1
2
1

2
2
d
1
d
2
. .

=
=
1
2
(x
2
1
+ x
2
2
)
_

2
1
+
2
2
<1
d
1
d
2
_
1
2
1

2
2
=
1
2
(x
2
1
+x
2
2
)
_
2
0
_
1
0
r dr d

1 r
2
=
1
2
(x
2
1
+ x
2
2
)
_
2
0
2
_
1
0

1
2
du d
u
1
2
_
u = 1 r
2
, du = 2r dr
_
=
1
2
(x
2
1
+ x
2
2
)
_
2
0
1 d = x
2
1
+ x
2
2
.
=
4t
2
_

2
1
+
2
2
<1
x
1

1
+x
2

2
_
1
2
1

2
2
d
1
d
2
=
4t
2
_
1
1
_

1
2
2

1
2
2
x
1

1
+x
2

2
_
1
2
1

2
2
d
1
d
2
= 0.
=
3t
2
2
_

2
1
+
2
2
<1

2
1
+
2
2
_
1
2
1

2
2
d
1
d
2
=
3t
2
2
_
2
0
_
1
0
(r cos )
2
+ (r sin)
2

1 r
2
r drd
=
3t
2
2
_
2
0
_
1
0
r
3

1 r
2
drd
_
u = 1 r
2
, du = 2r dr
_
=
3t
2
2
_
2
0
2
3
d =
t
2

_
2
0
d = 2t
2
.
u(x
1
, x
2
, t) = ++ = x
2
1
+ x
2
2
+ 2t
2
.
We may guess what the solution is:
u(x, y, z, t) =
1
2
_
(x +t)
2
+ (y + t)
2
+ (x t)
2
+ (y t)
2

= x
2
+ y
2
+ 2t
2
.
Partial Dierential Equations Igor Yanovsky, 2005 189
Check:
u(x, y, z, 0) = x
2
+ y
2
.
u
t
(x, y, z, t) = (x + t) + (y +t) (x t) (y t),
u
t
(x, y, z, 0) = 0.
u
tt
(x, y, z, t) = 4,
u
x
(x, y, z, t) = (x + t) + (x t),
u
xx
(x, y, z, t) = 2,
u
y
(x, y, z, t) = (y +t) + (y t),
u
yy
(x, y, z, t) = 2,
u
zz
(x, y, z, t) = 0,
u
tt
= u
xx
+ u
yy
+ u
zz
.
Partial Dierential Equations Igor Yanovsky, 2005 190
Problem (S98, #6).
Consider the two-dimensional wave equation w
tt
= a
2
w, with initial data which van-
ish for x
2
+y
2
large enough. Prove that w(x, y, t) satises the decay [w(x, y, t)[ Ct
1
.
(Note that the estimate is not uniform with respect to x, y since C may depend on x, y).
Proof. Suppose we have the following problem with initial data:
u
tt
= a
2
u x R
2
, t > 0,
u(x, 0) = g(x), u
t
(x, 0) = h(x) x R
2
.
The result is the consequence of the Huygens principle and may be proved by Hadamards
method of descent:
43
u(x, t) =
1
4

t
_
2t
_

2
1
+
2
2
<1
g(x
1
+ct
1
, x
2
+ct
2
) d
1
d
2
_
1
2
1

2
2
_
+
t
4
_
2
_

2
1
+
2
2
<1
h(x
1
+ ct
1
, x
2
+ ct
2
) d
1
d
2
_
1
2
1

2
2
_
=
1
2
_
||
2
<c
2
t
2
th(x +) + g(x +)
_
1
||
2
c
2
t
2
d
1
d
2
c
2
t
2
+
t
2
_
||
2
<c
2
t
2
g(x + ) (ct, ct)
_
1
||
2
c
2
t
2
d
1
d
2
c
2
t
2
.
For a given x, let T(x) be so large that T > 1 and supp(h + g) B
T
(x). Then for
t > 2T we have:
[u(x, t)[ =
1
2
_
||
2
<c
2
T
2
tM +M + 2Mct
_
1
c
2
T
2
c
2
T
2
4
d
1
d
2
c
2
t
2
=
c
2
T
2
2
__
M
_
3/4
_
1
c
2
t
+
_
M
_
3/4
_
1
c
2
Tt
+
2Mc
c
2
t
_
.
u(x, t) C
1
/t for t > 2T.
For t 2T:
[u(x, t)[ =
1
2
_
||
2
<c
2
t
2
2TM +M + 4McT
_
1
||
2
c
2
t
2
d
1
d
2
c
2
t
2
=
1
2
(2TM + M + 4Mct)2
_
ct
0
r dr/c
2
t
2
_
1
r
2
c
2
t
2
=
M(2T + 1 + 4cT)
2
_
1
0
du
u
1/2
=
M(2T + 1 + 4cT)
2
2
M(2T + 1 + 4cT)2T
t
.
Letting C = max(C
1
, M(2T + 1 + 4cT)2T), we have [u(x, t)[ C(x)/t.
For n = 3, suppose g, h C

0
(R
3
). The solution is given by the Kircchos
formula. There is a constant C so that u(x, t) C/t for all x R
3
and t > 0. As
McOwen suggensts in Hints for Exercises, to prove the result, we need to estimate the
43
Nicks solution follows.
Partial Dierential Equations Igor Yanovsky, 2005 191
area of intersection of the sphere of radius ct with the support of the functions g and
h.
Partial Dierential Equations Igor Yanovsky, 2005 192
Problem (S95, #6). Spherical waves in 3-d are waves symmetric about the origin;
i.e. u = u(r, t) where r is the distance from the origin. The wave equation
u
tt
= c
2
u
then reduces to
1
c
2
u
tt
= u
rr
+
2
r
u
r
. (16.43)
a) Find the general solutions u(r, t) by solving (16.43). Include both the incoming waves
and outgoing waves in your solutions.
b) Consider only the outgoing waves and assume the nite out-ux condition
0 < lim
r0
r
2
u
r
<
for all t. The wavefront is dened as r = ct. How is the amplitude of the wavefront
decaying in time?
Proof. a) We want to reduce (16.43) to the 1D wave equation. Let v = ru. Then
v
tt
= ru
tt
,
v
r
= ru
r
+u,
v
rr
= ru
rr
+ 2u
r
.
Thus, (16.43) becomes
1
c
2
1
r
v
tt
=
1
r
v
rr
,
1
c
2
v
tt
= v
rr
,
v
tt
= c
2
v
rr
,
which has the solution
v(r, t) = f(r +ct) + g(r ct).
Thus,
u(r, t) =
1
r
v(r, t) =
1
r
f(r + ct)
. .
incoming, (c>0)
+
1
r
g(r ct)
. .
outgoing, (c>0)
.
b) We consider u(r, t) =
1
r
g(r ct):
0 < lim
r0
r
2
u
r
< ,
0 < lim
r0
r
2
_
1
r
g

(r ct)
1
r
2
g(r ct)
_
< ,
0 < lim
r0
_
rg

## (r ct) g(r ct)

_
< ,
0 < g(ct) < ,
0 < g(ct) = G(t) < ,
g(t) = G
_
t
c
_
.
Partial Dierential Equations Igor Yanovsky, 2005 193
The wavefront is dened as r = ct. We have
u(r, t) =
1
r
g(r ct) =
1
r
G
_
r ct
c
_
=
1
ct
G(0).
[u(r, t)[ =
1
t

1
c
G(0)

.
The amplitude of the wavefront decays like
1
t
.
Partial Dierential Equations Igor Yanovsky, 2005 194
Problem (S00, #8). a) Show that for a smooth function F on the line, while
u(x, t) = F(ct + [x[)/[x[ may look like a solution of the wave equation u
tt
= c
2
u
in R
3
, it actually is not. Do this by showing that for any smooth function (x, t) with
compact support
_
R
3
R
u(x, t)(
tt
) dxdt = 4
_
R
(0, t)F(ct) dt.
Note that, setting r = [x[, for any function w which only depends on r one has
w = r
2
(r
2
w
r
)
r
= w
rr
+
2
r
w
r
.
b) If F(0) = F

## (0) = 0, what is the true solution to u

tt
= u with initial conditions
u(x, 0) = F([x[)/[x[ and u
t
(x, 0) = F

([x[)/[x[?
c) (Ralston Hw) Suppose u(x, t) is a solution to the wave equation u
tt
= c
2
u in
R
3
R with u(x, t) = w([x[, t) and u(x, 0) = 0. Show that
u(x, t) =
F([x[ + ct) F([x[ ct)
[x[
for a function F of one variable.
Proof. a) We have
_
R
3
_
R
u (
tt
) dxdt = lim
0
_
R
dt
_
|x|>
u (
tt
) dx
= lim
0
_
R
dt
_ _
|x|>
(u
tt
u) dx +
_
|x|=
u
n
u

n
dS
_
.
The nal equality is derived by integrating by parts twice in t, and using Greens
theorem:
_

(vu uv) dx =
_

_
v
u
n
u
v
n
_
ds.
Since dS =
2
sin

d and

n
=

r
, substituting u(x, t) = F([x[ + ct)/[x[
gives:
_
R
3
_
R
u (
tt
) dxdt =
_
R
4F(ct) dt.
Thus, u is not a weak solution to the wave equation.
b)
c) We want to show that v([x[, t) = [x[w([x[, t) is a solution to the wave equation in
one space dimension and hence must have the from v = F([x[ +ct) +G([x[ ct). Then
we can argue that w will be undened at x = 0 for some t unless F(ct) + G(ct) = 0
for all t.
We work in spherical coordinates. Note that w and v are independent of and . We
have:
v
tt
(r, t) = c
2
w = c
2
1
r
2
(r
2
w
r
)
r
= c
2
1
r
2
(2rw
r
+r
2
w
rr
),
rw
tt
= c
2
rw
rr
+ 2w
r
.
Thus we see that v
tt
= c
2
v
rr
, and we can conclude that
v(r, t) = F(r + ct) + G(r ct) and
w(r, t) =
F(r + ct) + G(r ct)
r
.
Partial Dierential Equations Igor Yanovsky, 2005 195
lim
r0
w(r, t) does not exist unless F(ct) + G(ct) = 0 for all t. Hence
w(r, t) =
F(ct + r) + G(ct r)
r
, and
u(x, t) =
F(ct + [x[) +G(ct [x[)
[x[
.
Partial Dierential Equations Igor Yanovsky, 2005 196
17 Problems: Laplace Equation
A fundamental solution K(x) for the Laplace operator is a distribution satisfying
44
K(x) = (x)
The fundamental solution for the Laplace operator is
K(x) =
_
1
2
log [x[ if n = 2
1
(2n)n
[x[
2n
if n 3.
17.1 Greens Function and the Poisson Kernel
Greens function is a special fundamental solution satisfying
45
_
G(x, ) = (x) for x
G(x, ) = 0 for x ,
(17.1)
To construct the Greens function,
consider w

(x) with w

(x) = 0 in and w

(x) = K(x ) on ;
consider G(x, ) = K(x ) + w

## (x), which is a fundamental solution satisfying

(17.1).
Problem 1. Given a particular distribution solution to the set of Dirichlet problems
_
u

(x) =

(x) for x
u

(x) = 0 for x ,
how would you use this to solve
_
u = 0 for x
u(x) = g(x) for x .
Proof. u

## (x) = G(x, ), a Greens function. G is a fundamental solution to the Laplace

operator, G(x, ) = 0, x . In this problem, it is assumed that G(x, ) is known for
. Then
u() =
_

G(x, )u dx +
_

u(x)
G(x, )
n
x
dS
x
for every u C
2
(). In particular, if u = 0 in and u = g on , then we obtain
the Poisson integral formula
u() =
_

G(x, )
n
x
g(x) dS
x
,
44
We know that u(x) =
_
R
n
K(xy)f(y)dy is a distribution solution of u = f when f is integrable
and has compact support. In particular, we have
u(x) =
_
R
n
K(x y)u(y) dy whenever u C

0
(R
n
).
The above result is a consequence of:
u(x) =
_

_

## K(x y) u(y) dy.

45
Greens function is useful in satisfying Dirichlet boundary conditions.
Partial Dierential Equations Igor Yanovsky, 2005 197
where H(x, ) =
G(x,)
nx
is the Poisson kernel.
Thus if we know that the Dirichlet problem has a solution u C
2
(), then we can
calculate u from the Poisson integral formula (provided of course that we can compute
G(x, )).
Partial Dierential Equations Igor Yanovsky, 2005 198
Dirichlet Problem on a Half-Space. Solve the n-dimensional Laplace/Poisson
equation on the half-space with Dirichlet boundary conditions.
Proof. Use the method of reection to construct Greens function. Let be an
upper half-space in R
n
. If x = (x

, x
n
), where x

R
n1
, we can see
[x

[ = [x

## [, and hence K(x

) = K(x

). Thus
G(x, ) = K(x ) K(x

)
is the Greens function on . G(x, ) is harmonic in ,
and G(x, ) = 0 on .
To compute the Poisson kernel, we must dierentiate G(x, )
in the negative x
n
direction. For n 2,

x
n
K(x ) =
x
n

n
[x [
n
,
so that the Poisson kernel is given by

x
n
G(x, )

xn=0
=
2
n

n
[x

[
n
, for x

R
n1
.
Thus, the solution is
u() =
_

G(x, )
n
x
g(x) dS
x
=
2
n

n
_
R
n1
g(x

)
[x

[
n
dx

.
If g(x

## ) is bounded and continuous for x

R
n1
, then u() is C

and harmonic in R
n
+
and extends continuously to R
n
+
such that u(

) = g(

).
Partial Dierential Equations Igor Yanovsky, 2005 199
Problem (F95, #3): Neumann Problem on a Half-Space.
a) Consider the Neumann problem in the upper half plane,
= x = (x
1
, x
2
) : < x
1
< , x
2
> 0:
u = u
x
1
x
1
+ u
x
2
x
2
= 0 x ,
u
x
2
(x
1
, 0) = f(x
1
) < x
1
< .
Find the corresponding Greens function and conclude that
u() = u(
1
,
2
) =
1
2
_

ln[(x
1

1
)
2
+
2
2
] f(x
1
) dx
1
is a solution of the problem.
b) Show that this solution is bounded in if and only if
_

f(x
1
) dx
1
= 0.
Proof. a) Notation: x = (x, y), = (x
0
, y
0
). Since K(x) =
1
2
log [x[, n = 2.
First, we nd the Greens function. We have
K(x ) =
1
2
log
_
(x x
0
)
2
+ (y y
0
)
2
.
Let G(x, ) = K(x ) + (x).
Since the problem is Neumann, we need:
_
G(x, ) = (x ),
G
y
((x, 0), ) = 0.
G((x, y), ) =
1
2
log
_
(x x
0
)
2
+ (y y
0
)
2
+ ((x, y), ),
G
y
((x, y), ) =
1
2
y y
0
(x x
0
)
2
+ (y y
0
)
2
+
y
((x, y), ),
G
y
((x, 0), ) =
1
2
y
0
(x x
0
)
2
+ y
2
0
+
y
((x, 0), ) = 0.
Let
((x, y), ) =
a
2
log
_
(x x
0
)
2
+ (y + y
0
)
2
. Then,
G
y
((x, 0), ) =
1
2
y
0
(x x
0
)
2
+ y
2
0
+
a
2
y
0
(x x
0
)
2
+ y
2
0
= 0.
Thus, a = 1.
G((x, y), ) =
1
2
log
_
(x x
0
)
2
+ (y y
0
)
2
+
1
2
log
_
(x x
0
)
2
+ (y + y
0
)
2
.
46
Consider Greens identity (after cutting out B

## () and having 0):

_

(uG G u
..
=0
) dx =
_

_
u
G
n
..
=0
G
u
n
_
dS
46
Note that for the Dirichlet problem, we would have gotten the - sign instead of + in front of
.
Partial Dierential Equations Igor Yanovsky, 2005 200
Since
u
n
=
u
(y)
= f(x), we have
_

u (x ) dx =
_

u() =
_

## G((x, y), ) f(x) dx.

For y = 0, we have
G((x, y), ) =
1
2
log
_
(x x
0
)
2
+y
2
0
+
1
2
log
_
(x x
0
)
2
+ y
2
0
=
1
2
2 log
_
(x x
0
)
2
+ y
2
0
=
1
2
log
_
(x x
0
)
2
+ y
2
0

.
Thus,
u() =
1
2
_

log
_
(x x
0
)
2
+ y
2
0

f(x) dx.
b) Show that this solution is bounded in if and only if
_

f(x
1
) dx
1
= 0.
Consider the Greens identity:
_

u dxdy =
_

u
n
dS =
_

u
y
dx =
_

f(x) dx = 0.
Note that the Greens identity applies to bounded domains .
_
R
R
f dx
1
+
_
2
0
u
r
Rd = 0.
???
Partial Dierential Equations Igor Yanovsky, 2005 201
McOwen 4.2 # 6. For n = 2, use the method of reections to nd the Greens
function for the rst quadrant = (x, y) : x, y > 0.
Proof. For x ,
[x
(0)
[ [x
(2)
[ = [x
(1)
[ [x
(3)
[,
[x
(0)
[ =
[x
(1)
[ [x
(3)
[
[x
(2)
[
.
But
(0)
= , so for n = 2,
G(x, ) =
1
2
log [x [
1
2
log
[x
(1)
[ [x
(3)
[
[x
(2)
[
.
G(x, ) = 0, x .
Problem. Use the method of images to solve
G = (x )
in the rst quadrant with G = 0 on the boundary.
Proof. To solve the problem in the rst quadrant
we take a reection to the fourth quadrant
and the two are reected to the left half.
G = (x
(0)
) (x
(1)
) (x
(2)
) + (x
(3)
).
G =
1
2
log
[x
(0)
[ [x
(3)
[
[x
(1)
[ [x
(2)
[
=
1
2
log
_
(x x
0
)
2
+ (y y
0
)
2
_
(x +x
0
)
2
+ (y +y
0
)
2
_
(x x
0
)
2
+ (y + y
0
)
2
_
(x +x
0
)
2
+ (y y
0
)
2
.
Note that on the axes G = 0.
Partial Dierential Equations Igor Yanovsky, 2005 202
Problem(S96, #3). Construct a Greens function for the following mixed Dirichlet-
Neumann problem in = x = (x
1
, x
2
) R
2
: x
1
> 0, x
2
> 0:
u =

2
u
x
2
1
+

2
u
x
2
2
= f, x ,
u
x
2
(x
1
, 0) = 0, x
1
> 0,
u(0, x
2
) = 0, x
2
> 0.
Proof. Notation: x = (x, y), = (x
0
, y
0
). Since K(x ) =
1
2
log [x [, n = 2.
K(x ) =
1
2
log
_
(x x
0
)
2
+ (y y
0
)
2
.
Let G(x, ) = K(x ) + (x).
At (0, y), y > 0,
G
_
(0, y),
_
=
1
2
log
_
x
2
0
+ (y y
0
)
2
+ (0, y) = 0.
Also,
G
y
_
(x, y),
_
=
1
2
1
2
2(y y
0
)
(x x
0
)
2
+ (y y
0
)
2
+w
y
(x, y)
=
1
2
y y
0
(x x
0
)
2
+ (y y
0
)
2
+w
y
(x, y).
At (x, 0), x > 0,
G
y
_
(x, 0),
_
=
1
2
y
0
(x x
0
)
2
+y
2
0
+w
y
(x, 0) = 0.
We have
((x, y), ) =
a
2
log
_
(x + x
0
)
2
+ (y y
0
)
2
+
b
2
log
_
(x x
0
)
2
+ (y + y
0
)
2
+
c
2
log
_
(x + x
0
)
2
+ (y + y
0
)
2
.
Using boundary conditions, we have
0 = G((0, y), ) =
1
2
log
_
x
2
0
+ (y y
0
)
2
+ (0, y)
=
1
2
log
_
x
2
0
+ (y y
0
)
2
+
a
2
log
_
x
2
0
+ (y y
0
)
2
+
b
2
log
_
x
2
0
+ (y + y
0
)
2
+
c
2
log
_
x
2
0
+ (y + y
0
)
2
.
Thus, a = 1, c = b. Also,
0 = G
y
((x, 0), ) =
1
2
y
0
(x x
0
)
2
+ y
2
0
+ w
y
(x, 0)
=
1
2
y
0
(x x
0
)
2
+ y
2
0

(1)
2
y
0
(x +x
0
)
2
+y
2
0
+
b
2
y
0
(x x
0
)
2
+y
2
0
+
(b)
2
y
0
(x + x
0
)
2
+y
2
0
.
Thus, b = 1, and
G((x, y), ) =
1
2
log
_
(x x
0
)
2
+ (y y
0
)
2
+ (x) =
1
2
_
log
_
(x x
0
)
2
+ (y y
0
)
2
Partial Dierential Equations Igor Yanovsky, 2005 203
log
_
(x +x
0
)
2
+ (y y
0
)
2
+ log
_
(x x
0
)
2
+ (y +y
0
)
2
log
_
(x +x
0
)
2
+ (y +y
0
)
2
_
.
It can be seen that G((x, y), ) = 0 on x = 0, for example.
Partial Dierential Equations Igor Yanovsky, 2005 204
Dirichlet Problem on a Ball. Solve the n-dimensional Laplace/Poisson equation on
the ball with Dirichlet boundary conditions.
Proof. Use the method of reection to construct Greens function.
Let = x R
n
: [x[ < a. For , dene

=
a
2

||
2
as its reection in ; note

/ .
[x

[
[x [
=
a
[[
for [x[ = a. [x [ =
[[
a
[x

[. (17.2)
From (17.2) we conclude that for x (i.e. [x[ = a),
K(x ) =

1
2
log
_
||
a
[x

[
_
if n = 2
_
a
||
_
n2
K(x

) if n 3.
(17.3)
Dene for x, :
G(x, ) =

K(x )
1
2
log
_
||
a
[x

[
_
if n = 2
K(x )
_
a
||
_
n2
K(x

) if n 3.
Since

## is not in , the second terms on the RHS are harmonic

in x . Moreover, by (17.3) we have G(x, ) = 0 if x .
Thus, G is the Greens function for .
u() =
_

G(x, )
n
x
g(x) dS
x
=
a
2
[[
2
a
n
_
|x|=a
g(x)
[x [
n
dS
x
.
Partial Dierential Equations Igor Yanovsky, 2005 205
17.2 The Fundamental Solution
Problem (F99, #2). Given that K
a
(x y) and K
b
(x y) are the kernels for
the operators ( aI)
1
and ( bI)
1
on L
2
(R
n
), where 0 < a < b, show that
( aI)( bI) has a fundamental solution of the form c
1
K
a
+ c
2
K
b
.
Use the preceding to nd a fundamental solution for
2
, when n = 3.
Proof. METHOD :

( aI)u = f (bI)u = f
u = K
a
..
f u = K
b
..
f
fundamental solution kernel
u =

K
a

f u =

K
b

f if u L
2
,

(aI)u = ([[
2
a) u =

f

(bI)u = ([[
2
b) u =

f
u =
1
(
2
+a)

f() u =
1
(
2
+b)

f()

K
a
=
1

2
+a

K
b
=
1

2
+ b
( aI)( bI)u = f,
_

2
(a +b)+abI
_
u = f,
u =
1
(
2
+a)(
2
+ b)

f() =

K
new

f(),

K
new
=
1
(
2
+a)(
2
+b)
=
1
b a
_

2
+ b
+
1

2
+ a
_
=
1
b a
(

K
b

K
a
),
K
new
=
1
b a
(K
b
K
a
),
c
1
=
1
b a
, c
2
=
1
b a
.
n = 3 is not relevant (may be used to assume K
a
, K
b
L
2
).
For
2
, a = 0, b = 1 above, or more explicitly
(
2
)u = f,
(
4
+
2
) u =

f,
u =
1
(
4
+
2
)

f,

K =
1
(
4
+
2
)
=
1

2
(
2
+ 1)
=
1

2
+ 1
+
1

2
=

K
1

K
0
.
Partial Dierential Equations Igor Yanovsky, 2005 206
METHOD :
For u C

0
(R
n
) we have:
u(x) =
_
R
n
K
a
(x y) (aI) u(y) dy,
u(x) =
_
R
n
K
b
(x y) (bI) u(y) dy.
Let
u(x) = c
1
(bI) (x), for
u(x) = c
2
(aI) (x), for
for (x) C

0
(R
n
). Then,
c
1
(bI)(x) =
_
R
n
K
a
(x y) (aI) c
1
(bI)(y) dy,
c
2
( aI)(x) =
_
R
n
K
b
(x y) ( bI) c
2
(aI)(y) dy.
(c
1
+ c
2
)(x) (c
1
b +c
2
a)(x) =
_
R
n
(c
1
K
a
+ c
2
K
b
) ( aI) (bI) (y) dy.
If c
1
= c
2
and (c
1
b + c
2
a) = 1, that is, c
1
=
1
ab
, we have:
(x) =
_
R
n
1
a b
(K
a
K
b
) (aI) (bI) (y) dy,
which means that
1
ab
(K
a
K
b
) is a fundamental solution of ( aI)(bI).

2
= (1) = (0I)( 1I).
( 0I) has fundamental solution K
0
=
1
4r
in R
3
.
To nd K, a fundamental solution for ( 1I), we need to solve for a radially
symmetric solution of
(1I)K = .
In spherical coordinates, in R
3
, the above expression may be written as:
K

+
2
r
K

K = 0.
Let
K =
1
r
w(r),
K

=
1
r
w

1
r
2
w,
K

=
1
r
w

2
r
2
w

+
2
r
3
w.
Plugging these into , we obtain:
1
r
w

1
r
w = 0, or
w

w = 0.
Partial Dierential Equations Igor Yanovsky, 2005 207
Thus,
w = c
1
e
r
+c
2
e
r
,
K =
1
r
w(r) = c
1
e
r
r
+c
2
e
r
r
.
Suppose v(x) 0 for [x[ R and let = B
R
(0); for small > 0 let

= B

(0).
Note: (I)K([x[) = 0 in

## . Consider Greens identity (

= B

(0)):
_

_
K([x[)v vK([x[)
_
dx =
_

_
K([x[)
v
n
v
K([x[)
n
_
dS
. .
=0, since v0 for xR
+
_
B(0)
_
K([x[)
v
n
v
K([x[)
n
_
dS
_

K([x[) v dx +
_

## v K([x[) dx to LHS to get:

_

_
K([x[)( I)v v (I)K([x[)
. .
=0, in
_
dx =
_
B(0)
_
K([x[)
v
n
v
K([x[)
n
_
dS.
lim
0
__

K([x[)(I)v dx
_
=
_

K([x[)(I)v dx.
_
Since K(r) = c
1
e
r
r
+c
2
e
r
r
is integrable at x = 0.
_
On B

## (0), K([x[) = K(). Thus,

47

_
B(0)
K([x[)
v
n
dS

K()

_
B(0)

v
n

dS

c
1
e

+ c
2
e

4
2
max

0, as 0.
_
B(0)
v(x)
K([x[)
n
dS =
_
B(0)
_
1

_
c
1
e

+c
2
e

_
+
1

2
_
c
1
e

+ c
2
e

_
_
v(x) dS
=
_
1

_
c
1
e

+c
2
e

_
+
1

2
_
c
1
e

+c
2
e

_
_
_
B(0)
v(x) dS
=
_
1

_
c
1
e

+c
2
e

_
+
1

2
_
c
1
e

+c
2
e

_
_
_
B(0)
v(0) dS
+
_
1

_
c
1
e

+ c
2
e

_
+
1

2
_
c
1
e

+ c
2
e

_
_
_
B(0)
[v(x) v(0)] dS

2
_
c
1
e

+ c
2
e

_
v(0) 4
2
4(c
1
+c
2
)v(0) = v(0).
Thus, taking c
1
= c
2
, we have c
1
= c
2
=
1
8
, which gives
_

K([x[)(I)v dx = lim
0
_

## K([x[)( I)v dx = v(0),

47
In R
3
, for |x| = ,
K(|x|) = K() = c1
e

+ c2
e

.
K(|x|)
n
=
K()
r
= c1
_
e

2
_
c2
_

2
_
=
1

_
c1e

+ c2e

_
+
1

2
_
c1e

+ c2e

_
,
since n points inwards. n points toward 0 on the sphere |x| = (i.e., n = x/|x|).
Partial Dierential Equations Igor Yanovsky, 2005 208
that is K(r) =
1
8
_
e
r
r
+
e
r
r
_
=
1
4r
cosh(r) is the fundamental solution of
( I).
By part (a),
1
ab
(K
a
K
b
) is a fundamental solution of (aI)(bI).
Here, the fundamental solution of (0I)(1I) is
1
1
(K
0
K) =
_

1
4r
+
1
4r
cosh(r)
_
=
1
4r
_
1 cosh(r)
_
.
Partial Dierential Equations Igor Yanovsky, 2005 209
Problem (F91, #3). Prove that

1
4
cos k[x[
[x[
is a fundamental solution for (+ k
2
) in R
3
where [x[ =
_
x
2
1
+ x
2
2
+x
2
3
,
i.e. prove that for any smooth function f(x) with compact support
u(x) =
1
4
_
R
3
cos k[x y[
[x y[
f(y) dy
is a solution to (+k
2
)u = f.
Proof. For v C

0
(R
n
), we want to show that for K([x[) =
1
4
cos k|x|
|x|
,
we have (+k
2
)K = , i.e.
_
R
n
K([x[) (+k
2
)v(x) dx = v(0).
Suppose v(x) 0 for [x[ R and let = B
R
(0); for small > 0 let

= B

(0).
(+ k
2
)K([x[) = 0 in

## . Consider Greens identity (

= B

(0)):
_

_
K([x[)v vK([x[)
_
dx =
_

_
K([x[)
v
n
v
K([x[)
n
_
dS
. .
=0, since v0 for xR
+
_
B(0)
_
K([x[)
v
n
v
K([x[)
n
_
dS
_

k
2
K([x[) v dx
_

v k
2
K([x[) dx to LHS to get:
_

_
K([x[)(+ k
2
)v v (+ k
2
)K([x[)
. .
=0, in
_
dx =
_
B(0)
_
K([x[)
v
n
v
K([x[)
n
_
dS.
lim
0
__

K([x[)(+k
2
)v dx
_
=
_

K([x[)(+ k
2
)v dx.
_
Since K(r) =
cos kr
4r
is integrable at x = 0.
_
On B

## (0), K([x[) = K(). Thus,

48

_
B(0)
K([x[)
v
n
dS

K()

_
B(0)

v
n

dS

cos k
4

4
2
max

0, as 0.
48
In R
3
, for |x| = ,
K(|x|) = K() =
cos k
4
.
K(|x|)
n
=
K()
r
=
1
4
_

k sink

cos k

2
_
=
1
4
_
k sin k +
cos k

_
,
since n points inwards. n points toward 0 on the sphere |x| = (i.e., n = x/|x|).
Partial Dierential Equations Igor Yanovsky, 2005 210
_
B(0)
v(x)
K([x[)
n
dS =
_
B(0)

1
4
_
k sink +
cos k

_
v(x) dS
=
1
4
_
k sink +
cos k

_
_
B(0)
v(x) dS
=
1
4
_
k sink +
cos k

_
_
B(0)
v(0) dS
1
4
_
k sin k +
cos k

_
_
B(0)
[v(x) v(0)] dS
=
1
4
_
k sink +
cos k

_
v(0) 4
2

1
4
_
k sin k +
cos k

_
[v(x) v(0)] 4
2
. .
0, (v is continuous)
cos k v(0) v(0).
Thus,
_

K([x[)(+k
2
)v dx = lim
0
_

K([x[)(+ k
2
)v dx = v(0),
that is, K(r) =
1
4
cos kr
r
is the fundamental solution of +k
2
.
Problem (F97, #2). Let u(x) be a solution of the Helmholtz equation
u + k
2
u = 0 x R
3
u = O
_
1
r
_
,
u
r
iku = O
_
1
r
2
_
, [x[ = r .
Prove that u 0.
Hint: A fundamental solution to the Helmholtz equation is
1
4r
e
ikr
.
Use the Green formula.
Proof. Denote K([x[) =
1
4r
e
ikr
, a fundamental solution. Thus, (+k
2
)K = .
Let x
0
be any point and = B
R
(x
0
); for small > 0 let

= B

(x
0
).
(+ k
2
)K([x[) = 0 in

## . Consider Greens identity (

= B

(x
0
)):
_

_
u (+ k
2
)K K(+k
2
)u
_
dx
. .
=0
=
_

_
u
K
n
K
u
n
_
dS +
_
B(x
0
)
_
u
K
n
K
u
n
_
dS
. .
u(x
0
), as 0
.
(It can be shown by the method previously used that the integral over B

(x
0
) ap-
proaches u(x
0
) as 0.) Taking the limit when 0, we obtain
u(x
0
) =
_

_
u
K
n
K
u
n
_
dS =
_

_
u

r
e
ik|xx
0
|
4[x x
0
[

e
ik|xx
0
|
4[x x
0
[
u
r
_
dS
=
_

_
u
_

r
e
ik|xx
0
|
4[x x
0
[
ik
e
ik|xx
0
|
4[x x
0
[
. .
=O(
1
|x|
2
); (can be shown)
_

e
ik|xx
0
|
4[x x
0
[
_
u
r
iku
_
_
dS
= O
_
1
R
_
O
_
1
R
2
_
4R
2
O
_
1
R
_
O
_
1
R
2
_
4R
2
= 0.
Taking the limit when R , we get u(x
0
) = 0.
Partial Dierential Equations Igor Yanovsky, 2005 211
Problem (S02, #1). a) Find a radially symmetric solution, u, to the equation in
R
2
,
u =
1
2
log [x[,
and show that u is a fundamental solution for
2
, i.e. show
(0) =
_
R
2
u
2
dx
for any smooth which vanishes for [x[ large.
b) Explain how to construct the Greens function for the following boundary value in
a bounded domain D R
2
with smooth boundary D
w = 0 and
w
n
= 0 on D,

2
w = f in D.
Proof. a) Rewriting the equation in polar coordinates, we have
u =
1
r
_
ru
r
_
r
+
1
r
2
u

=
1
2
log r.
For a radially symmetric solution u(r), we have u

= 0. Thus,
1
r
_
ru
r
_
r
=
1
2
log r,
_
ru
r
_
r
=
1
2
r log r,
ru
r
=
1
2
_
r log r dr =
r
2
log r
4

r
2
8
,
u
r
=
r log r
4

r
8
,
u =
1
4
_
r log r dr
1
8
_
r dr =
1
8
r
2
_
log r 1
_
.
u(r) =
1
8
r
2
_
log r 1
_
.
We want to show that u dened above is a fundamental solution of
2
for n = 2. That
is
_
R
2
u
2
v dx = v(0), v C

0
(R
n
).
See the next page that shows that u dened as u(r) =
1
8
r
2
log r is the
Fundamental Solution of
2
. (The
1
8
r
2
term does not play any role.)
In particular, the solution of

2
= f(x),
if given by
(x) =
_
R
2
u(x y)
2
(y) dy =
1
8
_
R
2
[x y[
2
_
log [x y[ 1
_
f(y) dy.
Partial Dierential Equations Igor Yanovsky, 2005 212
b) Let
K(x ) =
1
8
[x [
2
_
log [x [ 1
_
.
We use the method of images to construct the Greens function.
Let G(x, ) = K(x ) +(x). We need G(x, ) = 0 and
G
n
(x, ) = 0 for x .
Consider w

(x) with
2
w

(x) = 0 in , w

## (x) = K(x ) and

w

n
(x) =
K
n
(x )
on . Note, we can nd the Greens function for the upper-half plane, and then
make a conformal map onto the domain.
Partial Dierential Equations Igor Yanovsky, 2005 213
Problem (S97, #6). Show that the fundamental solution of
2
in R
2
is given by
V (x
1
, x
2
) =
1
8
r
2
ln(r), r = [x [,
and write the solution of

2
w = F(x
1
, x
2
).
Hint: In polar coordinates, =
1
r

r
(r

r
) +
1
r
2

2
; for example, V =
1
2
(1+ln(r)).
Proof. Notation: x = (x
1
, x
2
). We have
V (x) =
1
8
r
2
log(r),
In polar coordinates: (here, V

= 0)
V =
1
r
_
rV
r
_
r
=
1
r
_
r
_
1
8
r
2
log(r)
_
r
_
r
=
1
8
1
r
_
r
_
2r log(r) +r
_
_
r
=
1
8
1
r
_
2r
2
log(r) + r
2
_
r
=
1
8
1
r
_
4r + 4r log r
_
=
1
2
(1 + log r).
The fundamental solution V (x) for
2
is the distribution satisfying:
2
V (r) = (r).

2
V = (V ) =
_
1
2
(1 + log r)
_
=
1
2
(1 + log r) =
1
2
1
r
_
r(1 + log r)
r
_
r
=
1
2
1
r
_
r
1
r
_
r
=
1
2
1
r
(1)
r
= 0 for r ,= 0.
Thus,
2
V (r) = (r) V is the fundamental solution.
The approach above is not rigorous. See the next page that shows that
V dened above is the Fundamental Solution of
2
.
The solution of

2
= F(x),
if given by
(x) =
_
R
2
V (x y)
2
(y) dy =
1
8
_
R
2
[x y[
2
log [x y[ F(y) dy.
Partial Dierential Equations Igor Yanovsky, 2005 214
Show that the Fundamental Solution of
2
in R
2
is given by:
K(x) =
1
8
r
2
ln(r), r = [x [, (17.4)
Proof. For v C

0
(R
n
), we want to show
_
R
n
K([x[)
2
v(x) dx = v(0).
Suppose v(x) 0 for [x[ R and let = B
R
(0); for small > 0 let

= B

(0).
K([x[) is biharmonic (
2
K([x[) = 0) in

## . Consider Greens identity (

=
B

(0)):
_

K([x[)
2
v dx =
_

_
K([x[)
v
n
v
K([x[)
n
_
ds +
_

_
K([x[)
v
n
v
K([x[)
n
_
ds
. .
=0, since v0 for xR
+
_
B(0)
_
K([x[)
v
n
v
K([x[)
n
_
ds +
_
B(0)
_
K([x[)
v
n
v
K([x[)
n
_
ds.
lim
0
__

K([x[)
2
v dx
_
=
_

K([x[)v
2
dx.
_
Since K(r) is integrable at x = 0.
_
On B

## (0), K([x[) = K(). Thus,

49

_
B(0)
K([x[)
v
n
dS

K()

_
B(0)

v
n

dS

K()

1
max
x

(v)

1
8

2
log()

n
max
x

(v)

0, as 0.
_
B(0)
v(x)
K([x[)
n
dS =
_
B(0)

1
2
v(x) dS
=
_
B(0)

1
2
v(0) dS +
_
B(0)

1
2
[v(x) v(0)] dS
=
1
2
v(0) 2 max
xB(0)

v(x) v(0)

. .
0, (v is continuous)
= v(0).

_
B(0)
K([x[)
v
n
dS

K()

_
B(0)

v
n

dS

1
2
(1 + log )

2 max
x
[v[ 0, as 0.
_
B(0)
v
K([x[)
n
dS =
_
B(0)
_

1
4
log
1
8

_
v(x) dS

log +
1
2

2 max
xB(0)
[v[ 0, as 0.
49
Note that for |x| = ,
K(|x|) = K() =
1
8

2
log , K =
1
2
(1 + log ),
K(|x|)
n
=
K()
r
=
1
4
log
1
8
,
K
n
=
K
r
=
1
2
.
Partial Dierential Equations Igor Yanovsky, 2005 215

K([x[)
2
v dx = lim
0
_

K([x[)
2
v dx = v(0).
Partial Dierential Equations Igor Yanovsky, 2005 216
Problem (F99, #8). Let u = u(x, t) solve the following PDE in two spatial dimen-
sions
u = 1
for r < R(t), in which r = [x[ is the radial variable, with boundary condition
u = 0
on r = R(t). In addition assume that R(t) satises
dR
dt
=
u
r
(r = R)
with initial condition R(0) = R
0
.
a) Find the solution u(x, t).
b) Find an ODE for the outer radius R(t), and solve for R(t).
Proof. a) Rewrite the equation in polar coordinates:
u =
_
1
r
(ru
r
)
r
+
1
r
2
u

_
= 1.
For a radially symmetric solution u(r), we have u

= 0. Thus,
1
r
(ru
r
)
r
= 1,
(ru
r
)
r
= r,
ru
r
=
r
2
2
+c
1
,
u
r
=
r
2
+
c
1
r
,
u(r, t) =
r
2
4
+c
1
log r + c
2
.
Since we want u to be dened for r = 0, we have c
1
= 0. Thus,
u(r, t) =
r
2
4
+ c
2
.
Using boundary conditions, we have
u(R(t), t) =
R(t)
2
4
+ c
2
= 0 c
2
=
R(t)
2
4
. Thus,
u(r, t) =
r
2
4
+
R(t)
2
4
.
b) We have
u(r, t) =
r
2
4
+
R(t)
2
4
,
u
r
=
r
2
,
dR
dt
=
u
r
(r = R) =
R
2
, (from )
dR
R
=
dt
2
,
log R =
t
2
,
R(t) = c
1
e
t
2
, R(0) = c
1
= R
0
. Thus,
Partial Dierential Equations Igor Yanovsky, 2005 217
R(t) = R
0
e
t
2
.
Partial Dierential Equations Igor Yanovsky, 2005 218
Problem (F01, #3). Let u = u(x, t) solve the following PDE in three spatial di-
mensions
u = 0
for R
1
< r < R(t), in which r = [x[ is the radial variable, with boundary conditions
u(r = R(t), t) = 0, and u(r = R
1
, t) = 1.
In addition assume that R(t) satises
dR
dt
=
u
r
(r = R)
with initial condition R(0) = R
0
in which R
0
> R
1
.
a) Find the solution u(x, t).
b) Find an ODE for the outer radius R(t).
Proof. a) Rewrite the equation in spherical coordinates (n = 3, radial functions):
u =
_

2
r
2
+
2
r

r
_
u =
1
r
2
(r
2
u
r
)
r
= 0.
(r
2
u
r
)
r
= 0,
r
2
u
r
= c
1
,
u
r
=
c
1
r
2
,
u(r, t) =
c
1
r
+c
2
.
Using boundary conditions, we have
u(R(t), t) =
c
1
R(t)
+c
2
= 0 c
2
=
c
1
R(t)
,
u(R
1
, t) =
c
1
R
1
+ c
2
= 1.
This gives
c
1
=
R
1
R(t)
R
1
R(t)
, c
2
=
R
1
R
1
R(t)
.
u(r, t) =
R
1
R(t)
R
1
R(t)

1
r
+
R
1
R
1
R(t)
.
b) We have
u(r, t) =
R
1
R(t)
R
1
R(t)

1
r
+
R
1
R
1
R(t)
,
u
r
=
R
1
R(t)
R
1
R(t)

1
r
2
,
dR
dt
=
u
r
(r = R) =
R
1
R(t)
R
1
R(t)

1
R(t)
2
=
R
1
(R
1
R(t)) R(t)
(from )
Thus, an ODE for the outer radius R(t) is
_
dR
dt
=
R
1
(R(t)R
1
) R(t)
,
R(0) = R
0
, R
0
> R
1
.
Partial Dierential Equations Igor Yanovsky, 2005 219
Problem (S02, #3). Steady viscous ow in a cylindrical pipe is described by the
equation
(u )u +
1

u = 0
on the domain < x
1
< , x
2
2
+x
2
3
R
2
, where u = (u
1
, u
2
, u
3
) = (U(x
2
, x
3
), 0, 0)
is the velocity vector, p(x
1
, x
2
, x
3
) is the pressure, and and are constants.
a) Show that
p
x
1
is a constant c, and that U = c/.
b) Assuming further that U is radially symmetric and U = 0 on the surface of the pipe,
determine the mass Q of uid passing through a cross-section of pipe per unit time in
terms of c, , , and R. Note that
Q =
_
{x
2
2
+x
2
3
R
2
}
U dx
2
dx
3
.
Proof. a) Since u = (u
1
, u
2
, u
3
) = (U(x
2
, x
3
), 0, 0), we have
(u )u = (u
1
, u
2
, u
3
)
_
u
1
x
1
,
u
2
x
2
,
u
3
x
3
_
= (U(x
2
, x
3
), 0, 0) (0, 0, 0) = 0.
Thus,
1

u = 0,
p = u,
_
p
x
1
,
p
x
2
,
p
x
3
_
= (u
1
, u
2
, u
3
),
_
p
x
1
,
p
x
2
,
p
x
3
_
= (U
x
2
x
2
+ U
x
3
x
3
, 0, 0).
We can make the following observations:
p
x
1
= (U
x
2
x
2
+ U
x
3
x
3
)
. .
indep. of x
1
,
p
x
2
= 0 p = f(x
1
, x
3
),
p
x
3
= 0 p = g(x
1
, x
2
).
Thus, p = h(x
1
). But
p
x
1
is independent of x
1
. Therefore,
p
x
1
= c.
p
x
1
= U,
U =
1

p
x
1
=
c

.
Partial Dierential Equations Igor Yanovsky, 2005 220
b) Cylindrical Laplacian in R
3
U =
1
r
_
rU
r
_
r
,
1
r
_
rU
r
_
r
=
c

,
_
rU
r
_
r
=
cr

,
rU
r
=
cr
2
2
+ c
1
,
U
r
=
cr
2
+
c
1
r
.
For U
r
to stay bounded for r = 0, we need c
1
= 0. Thus,
U
r
=
cr
2
,
U =
cr
2
4
+ c
2
,
0 = U(R) =
cR
2
4
+ c
2
,
U =
cr
2
4

cR
2
4
=
c
4
(r
2
R
2
).
Q =
_
{x
2
2
+x
2
3
R
2
}
U dx
2
dx
3
=
c
4
_
2
0
_
R
0
(r
2
R
2
) rdrd =
c
4
_
2
0
R
4
4
d
=
cR
4

8
.
It is not clear why Q is negative?
Partial Dierential Equations Igor Yanovsky, 2005 221
17.4 Weak Solutions
Problem (S98, #2).
A function u H
2
0
() is a weak solution of the biharmonic equation

2
u = f in
u = 0 on
u
n
= 0 on
provided
_

uv dx =
_

fv dx
for all test functions v H
2
0
(). Prove that for each f L
2
(), there exists a unique
weak solution for this problem. Here, H
2
0
() is the closure of all smooth functions in
which vanish on the boundary and with nite H
2
norm: [[u[[
2
2
=
_

(u
2
xx
+ u
2
xy
+
u
2
yy
) dxdy < .
Hint: use Lax-Milgram lemma.
Proof. Multiply the equation by v H
2
0
() and integrate over :

2
u = f,
_

2
u v dx =
_

f v dx,
_

u
n
v ds
_

u
v
n
ds
. .
= 0
+
_

uv dx =
_

f v dx,
_

uv dx
. .
a(u,v)
=
_

f v dx
. .
L(v)
.
Denote: V = H
2
0
(). Check the following conditions:
a(, ) is continuous: > 0, s.t. [a(u, v)[ [[u[[
V
[[v[[
V
, u, v V ;
a(, ) is V-elliptic: > 0, s.t. a(v, v) [[v[[
2
V
, v V ;
L() is continuous: > 0, s.t. [L(v)[ [[v[[
V
, v V.
Partial Dierential Equations Igor Yanovsky, 2005 222
We have
50
[a(u, v)[
2
=

uv dx

_
_

(u)
2
dx
__
_

(v)
2
dx
_
[[u[[
2
H
2
0
()
[[v[[
2
H
2
0
()
.
a(v, v) =
_

(v)
2
dx [[v[[
H
2
0
()
.
[L(v)[ =

f v dx

[f[ [v[ dx
_
_

f
2
dx
_1
2
_
_

v
2
dx
_1
2
= [[f[[
L
2
()
[[v[[
L
2
()
[[f[[
L
2
()
. .

[[v[[
H
2
0
()
.
Thus, by Lax-Milgram theorem, there exists a weak solution u H
2
0
().
Also, we can prove the stability result.
[[u[[
2
H
2
0
()
a(u, u) = [L(u)[ [[u[[
H
2
0
()
,
[[u[[
H
2
0
()

.
Let u
1
, u
2
be two solutions so that
a(u
1
, v) = L(v),
a(u
2
, v) = L(v)
for all v V . Subtracting these two equations, we see that:
a(u
1
u
2
, v) = 0 v V.
Applying the stability estimate (with L 0, i.e. = 0), we conclude that
[[u
1
u
2
[[
H
2
0
()
= 0, i.e. u
1
= u
2
.
50
Cauchy-Schwarz Inequality:
|(u, v)| ||u||||v|| in any norm, for example
_
|uv|dx (
_
u
2
dx)
1
2
(
_
v
2
dx)
1
2
;
|a(u, v)| a(u, u)
1
2
a(v, v)
1
2
;
_
|v|dx =
_
|v| 1 dx = (
_
|v|
2
dx)
1
2
(
_
1
2
dx)
1
2
.
Poincare Inequality:
||v||
H
2
()
C
_

(v)
2
dx.
Greens formula:
_

(u)
2
dx =
_

(u
2
xx
+ u
2
yy
+ 2uxxuyy) dxdy =
_

(u
2
xx
+ u
2
yy
2uxxyuy) dxdy =
_

(u
2
xx
+ u
2
yy
+ 2|uxy|
2
) dxdy ||u||
2
H
2
0
()
.
Partial Dierential Equations Igor Yanovsky, 2005 223
17.5 Uniqueness
Problem. The solution of the Robin problem
u
n
+u = , x
for the Laplace equation is unique when > 0 is a constant.
Proof. Let u
1
and u
2
be two solutions of the Robin problem. Let w = u
1
u
2
. Then
w = 0 in ,
w
n
+ w = 0 on .
Consider Greens formula:
_

u v dx =
_

v
u
n
ds
_

v u dx.
Setting w = u = v gives
_

[w[
2
dx =
_

w
w
n
ds
_

wwdx
. .
=0
.
Boundary condition gives
_

[w[
2
dx
. .
0
=
_

w
2
ds
. .
0
.
Thus, w 0, and u
1
u
2
. Hence, the solution to the Robin problem is unique.
Problem. Suppose q(x) 0 for x and consider solutions u C
2
() C
1
() of
u q(x)u = 0 in .
Establish uniqueness theorems for
a) the Dirichlet problem: u(x) = g(x), x ;
b) the Neumann problem: u/n = h(x), x .
Proof. Let u
1
and u
2
be two solutions of the Dirichlet or Neumann problem.
Let w = u
1
u
2
. Then
w q(x)w = 0 in ,
w = 0 or
w
n
= 0 on .
Consider Greens formula:
_

u v dx =
_

v
u
n
ds
_

v u dx.
Setting w = u = v gives
_

[w[
2
dx =
_

w
w
n
ds
. .
=0, Dirichlet or Neumann

wwdx.
Partial Dierential Equations Igor Yanovsky, 2005 224
_

[w[
2
dx
. .
0
=
_

q(x)w
2
dx
. .
0
.
Thus, w 0, and u
1
u
2
. Hence, the solution to the Dirichlet and Neumann problems
are unique.
Problem (F02, #8; S93, #5).
Let D be a bounded domain in R
3
. Show that a solution of the boundary value problem

2
u = f in D,
u = u = 0 on D
is unique.
Proof. Method I: MaximumPrinciple. Let u
1
, u
2
be two solutions of the boundary
value problem. Dene w = u
1
u
2
. Then w satises

2
w = 0 in D,
w = w = 0 on D.
So w is harmonic and thus achieves min and max on D w 0.
So w is harmonic, but w 0 on D w 0. Hence, u
1
= u
2
.
Method II: Greens Identities. Multiply the equation by w and integrate:
w
2
w = 0,
_

w
2
w dx = 0,
_

w
(w)
n
ds
. .
=0

w()w dx = 0,

w
n
wds
. .
=0
+
_

(w)
2
dx = 0.
Thus, w 0. Now, multiply w = 0 by w. We get
_

[w[
2
dx = 0.
Thus, w = 0 and w is a constant. Since w = 0 on , we have w 0.
Problem (F97, #6).
a) Let u(x) 0 be continuous in closed bounded domain R
n
, u is continuous in
,
u = u
2
and u[

= 0.
Prove that u 0.
b) What can you say about u(x) when the condition u(x) 0 in is dropped?
Partial Dierential Equations Igor Yanovsky, 2005 225
Proof. a) Multiply the equation by u and integrate:
uu = u
3
,
_

uu dx =
_

u
3
dx,
_

u
u
n
ds
. .
=0

[u[
2
dx =
_

u
3
dx,
_

_
u
3
+[u[
2
_
dx = 0.
Since u(x) 0, we have u 0.
b) If we dont know that u(x) 0, then u can not be nonnegative on the entire
domain . That is, u(x) < 0, on some (or all) parts of . If u is nonnegative on ,
then u 0.
Partial Dierential Equations Igor Yanovsky, 2005 226
Problem (W02, #5). Consider the boundary value problem
u +
n

k=1

k
u
x
k
u
3
= 0 in ,
u = 0 on ,
where is a bounded domain in R
n
with smooth boundary. If the
k
s are constants,
and u(x) has continuous derivatives up to second order, prove that u must vanish
identically.
Proof. Multiply the equation by u and integrate:
uu +
n

k=1

k
u
x
k
u u
4
= 0,
_

uu dx +
_

k=1

k
u
x
k
u dx
_

u
4
dx = 0,
_

u
u
n
ds
. .
=0

[u[
2
dx +
_

k=1

k
u
x
k
u dx
. .

u
4
dx = 0.
We will show that = 0.
_

k
u
x
k
u dx =
_

k
u
2
ds
. .
=0

k
u
u
x
k
dx,
2
_

k
u
x
k
u dx = 0,

k=1

k
u
x
k
u dx = 0.
Thus, we have

[u[
2
dx
_

u
4
dx = 0,
_

_
[u[
2
+
_

u
4
_
dx = 0.
Hence, [u[
2
= 0 and u
4
= 0. Thus, u 0.
Note that
_

k=1

k
u
x
k
u dx =
_

u u dx =
_

nu
2
ds
. .
=0

u u dx,
and thus,
_

u u dx = 0.
Partial Dierential Equations Igor Yanovsky, 2005 227
Problem (W02, #9). Let D = x R
2
: x
1
0, x
2
0, and assume that f is
continuous on D and vanishes for [x[ > R.
a) Show that the boundary value problem
u = f in D,
u(x
1
, 0) =
u
x
1
(0, x
2
) = 0
can have only one bounded solution.
b) Find an explicit Greens function for this boundary value problem.
Proof. a) Let u
1
, u
2
be two solutions of the boundary value problem. Dene w =
u
1
u
2
. Then w satises
w = 0 in D,
w(x
1
, 0) =
w
x
1
(0, x
2
) = 0.
Consider Greens formula:
_
D
u v dx =
_
D
v
u
n
ds
_
D
v u dx.
Setting w = u = v gives
_
D
[w[
2
dx =
_
D
w
w
n
ds
_
D
wwdx,
_
D
[w[
2
dx =
_
Rx
1
w
w
n
ds +
_
Rx
2
w
w
n
ds +
_
|x|>R
w
w
n
ds
_
D
wwdx
=
_
Rx
1
w(x
1
, 0)
. .
=0
w
x
2
ds +
_
Rx
2
w(0, x
2
)
w
x
1
..
=0
ds +
_
|x|>R
w
..
=0
w
n
ds
_
D
w w
..
=0
dx,
_
D
[w[
2
dx = 0 [w[
2
= 0 w = const.
Since w(x
1
, 0) = 0 w 0. Thus, u
1
= u
2
.
b) The similar problem is solved in the appropriate section (S96, #3).
Notice whenever you are on the boundary with variable x,
[x
(0)
[ =
[x
(1)
[[x
(3)
[
[x
(2)
[
.
So, G(x, ) =
1
2
_
log [x [ log
[x
(1)
[[x
(3)
[
[x
(2)
[
_
is the Greens function.
Partial Dierential Equations Igor Yanovsky, 2005 228
Problem (F98, #4). In two dimensions x = (x, y), dene the set
a
as

a
=
+

in which

+
= [x x
0
[ a x 0

= [x + x
0
[ a x 0 =
+
and x
0
= (1, 0). Note that
a
consists of two components when 0 < a < 1
and a single component when a > 1. Consider the Neumann problem

2
u = f, x
a
u/n = 0, x
a
in which
_

+
f(x) dx = 1
_

f(x) dx = 1
a) Show that this problem has a solution for 1 < a, but not for 0 < a < 1.
(You do not need to construct the solution, only demonstrate solveability.)
b) Show that max
a
[u[ as a 1 from above. (Hint: Denote L to be
the line segment L =
+

## , and note that its length [L[ goes to 0 as a 1.)

Proof. a) We use the Greens identity. For 1 < a,
0 =
_
a
u
n
ds =
_
a
u dx =
_
a
f(x) dx
=
_

+
f(x) dx +
_

f(x) dx = 1 1 = 0.
Thus, the problem has a solution for 1 < a.
For 0 < a < 1,
+
and

## are disjoint. Consider

+
:
0 =
_

+
u
n
ds =
_

+
u dx =
_

+
f(x) dx = 1,
0 =
_

u
n
ds =
_

u dx =
_

f(x) dx = 1.
Thus, the solution does not exist for 0 < a < 1.
Partial Dierential Equations Igor Yanovsky, 2005 229
b) Using the Greens identity, we have: (n
+
is the unit normal to
+
)
_

+
u dx =
_

+
u
n
+
ds =
_
L
u
n
+
ds,
_

u dx =
_

u
n

ds =
_
L
u
n

ds =
_
L
u
n
+
ds.
_

+
u dx
_

u dx = 2
_
L
u
n
+
ds,
_

+
f(x) dx
_

f(x) dx = 2
_
L
u
n
+
ds.
2 = 2
_
L
u
n
+
ds,
1 =
_
L
u
n
+
ds
_
L

u
n
+

ds
_
L
_
_
u
n
+
_
2
+
_
u

_
2
[L[ max
L
[u[ [L[ max
a
[u[.
Thus,
max
a
[u[
1
[L[
.
As a 1 (L 0) max
a
[u[ .
Partial Dierential Equations Igor Yanovsky, 2005 230
Problem (F00, #1). Consider the Dirichlet problem in a bounded domain D R
n
with smooth boundary D,
u + a(x)u = f(x) in D,
u = 0 on D.
a) Assuming that [a(x)[ is small enough, prove the uniqueness of the classical solution.
b) Prove the existence of the solution in the Sobolev space H
1
(D) assuming that f
L
2
(D).
Note: Use Poincare inequality.
Proof. a) By Poincare Inequality, for any u C
1
0
(D), we have [[u[[
2
2
C[[u[[
2
2
.
Consider two solutions of the Dirichlet problem above. Let w = u
1
u
2
. Then, w
satises
_
w +a(x)w = 0 in D,
w = 0 on D.
ww +a(x)w
2
= 0,
_
wwdx +
_
a(x)w
2
dx = 0,

_
[w[
2
dx +
_
a(x)w
2
dx = 0,
_
a(x)w
2
dx =
_
[w[
2
dx
1
C
_
w
2
dx, (by Poincare inequality)
_
a(x)w
2
dx
1
C
_
w
2
dx 0,
[a(x)[
_
w
2
dx
1
C
_
w
2
dx 0,
_
[a(x)[
1
C
_
_
w
2
dx 0.
If [a(x)[ <
1
C
w 0.
b) Consider
F(v, u) =
_

(vu +a(x)vu) dx =
_

vf(x) dx = F(v).
F(v) is a bounded linear functional on v H
1,2
(D), D = .
[F(v)[ [[f[[
2
[[v[[
2
[[f[[
2
C[[v[[
H
1,2
(D)
So by Riesz representation, there exists a solution u H
1,2
0
(D) of
< u, v >=
_

vu + a(x)vu dx =
_

vf(x) dx = F(v) v H
1,2
0
(D).
Partial Dierential Equations Igor Yanovsky, 2005 231
Problem (S91, #8). Dene the operator
Lu = u
xx
+ u
yy
4(r
2
+ 1)u
in which r
2
= x
2
+ y
2
.
a) Show that = e
r
2
satises L = 0.
b) Use this to show that the equation
Lu = f in
u
n
= on
has a solution only if
_

f dx =
_

ds(x).
Proof. a) Expressing Laplacian in polar coordinates, we obtain:
Lu =
1
r
(ru
r
)
r
4(r
2
+ 1)u,
L =
1
r
(r
r
)
r
4(r
2
+ 1) =
1
r
(2r
2
e
r
2
)
r
4(r
2
+ 1)e
r
2
=
1
r
(4re
r
2
+ 2r
2
2re
r
2
) 4r
2
e
r
2
4e
r
2
= 0.
b) We have = e
r
2
= e
x
2
+y
2
= e
x
2
e
y
2
. From part (a),
L = 0,

n
= n = (
x
,
y
) n = (2xe
x
2
e
y
2
, 2ye
x
2
e
y
2
) n = 2e
r
2
(x, y) (y, x) = 0.
51
Consider two equations:
Lu = u 4(r
2
+ 1)u,
L = 4(r
2
+ 1).
Multiply the rst equation by and the second by u and subtract the two equations:
Lu = u 4(r
2
+ 1)u,
uL = u 4(r
2
+ 1)u,
Lu uL = u u.
Then, we start from the LHS of the equality we need to prove and end up with RHS:
_

f dx =
_

Lu dx =
_

(Lu uL) dx =
_

(u u) dx
=
_

(
u
n
u

n
) ds =
_

u
n
ds =
_

ds.
51
The only shortcoming in the above proof is that we assume n = (y, x), without giving an expla-
nation why it is so.
Partial Dierential Equations Igor Yanovsky, 2005 232
Consider an mth-order dierential operator
Lu =

||m
a

(x)D

u.
The integration by parts formula
_

u
x
k
v dx =
_

uvn
k
ds
_

uv
x
k
dx n = (n
1
, . . . , n
n
) R
n
,
with u or v vanishing near is:
_

u
x
k
v dx =
_

uv
x
k
dx.
We can repeat the integration by parts with any combination of derivatives
D

= (/x
1
)

1
(/x
n
)
n
:
_

(D

u)v dx = (1)
m
_

uD

v dx, (m = [[).
We have
_

(Lu) v dx =
_

_

||m
a

(x)D

u
_
v dx =

||m
_

(x) v D

u dx
=

||m
(1)
||
_

(a

(x) v) u dx =
_

||m
(1)
||
D

(a

(x) v)
. .
L

(v)
u dx
=
_

(v) u dx,
for all u C
m
() and v C

0
.
The operator
L

(v) =

||m
(1)
||
D

(a

(x) v)
is called the adjoint of L.
The operator is self-adjoint if L

= L.
52
_

vL(u) dx =
_

uL(v) dx.
52
L = L

(Lu|v) = (u|L

v) = (u|Lv).
Partial Dierential Equations Igor Yanovsky, 2005 233
Problem (F92, #6).
Consider the Laplace operator in the wedge 0 x y with boundary conditions
f
x
= 0 on x = 0
f
x

f
y
= 0 on x = y.
a) For which values of is this operator self-adjoint?
b) For such a value of , suppose that
f = e
r
2
/2
cos
with these boundary conditions. Evaluate
_
C
R

r
f ds
in which C
R
is the circular arc of radius R connecting the boundaries x = 0 and x = y.
Proof. a) We have
Lu = u = 0
u
x
= 0 on x = 0
u
x

u
y
= 0 on x = y.
The operator L is self-adjoint if:
_

(u Lv v Lu) dx = 0.
_

(u Lv v Lu) dx =
_

(uv vu) dx =
_

_
u
v
n
v
u
n
_
ds
=
_
x=0
_
u
v
n
v
u
n
_
ds +
_
x=y
_
u
v
n
v
u
n
_
ds
=
_
x=0
_
u (v n) v (u n)
_
ds +
_
x=y
_
u (v n) v (u n)
_
ds
=
_
x=0
_
u
_
(v
x
, v
y
) (1, 0)
_
v
_
(u
x
, u
y
) (1, 0)
_
_
ds
+
_
x=y
_
u
_
(v
x
, v
y
) (1/

2, 1/

2)
_
v
_
(u
x
, u
y
) (1/

2, 1/

2)
_
_
ds
=
_
x=0
_
u
_
(0, v
y
) (1, 0)
_
v
_
(0, u
y
) (1, 0)
_
_
ds
. .
= 0
+
_
x=y
_
u
_
(v
y
, v
y
) (1/

2, 1/

2)
_
v
_
(u
y
, u
y
) (1/

2, 1/

2)
_
_
ds
=
_
x=y
_
uv
y

2
( 1)
vu
y

2
( 1)
_
ds =
..
need
0.
Partial Dierential Equations Igor Yanovsky, 2005 234
Thus, we need = 1 so that L is self-adjoint.
b) We have = 1. Using Greens identity and results from part (a), (
f
n
= 0 on
x = 0 and x = y):
_

f dx =
_

f
n
ds =
_
C
R
f
n
ds +
_
x=0
f
n
..
=0
ds +
_
x=y
f
n
..
=0
ds =
_
C
R
f
r
ds.
Thus,
_
C
R
f
r
ds =
_

f dx =
_
R
0
_
2

4
e
r
2
/2
cos r drd
=
_
1
1

2
_
_
R
0
e
r
2
/2
r dr =
_
1
1

2
_
(1 e
R
2
/2
).
Partial Dierential Equations Igor Yanovsky, 2005 235
Problem (F99, #1). Suppose that u = 0 in the weak sense in R
n
and that there
is a constant C such that
_
{|xy|<1}
[u(y)[ dy < C, x R
n
.
Show that u is constant.
Proof. Consider Greens formula:
_

u v dx =
_

v
u
n
ds
_

v u dx
For v = 1, we have
_

u
n
ds =
_

u dx.
Let B
r
(x
0
) be a ball in R
n
. We have
0 =
_
Br(x
0
)
u dx =
_
Br(x
0
)
u
n
ds = r
n1
_
|x|=1
u
r
(x
0
+ rx) ds
= r
n1

r
1

n
_
|x|=1
u(x
0
+rx) ds.
Thus,
1
n
_
|x|=1
u(x
0
+ rx) ds is independent of r. Hence, it is constant.
By continuity, as r 0, we obtain the Mean Value property:
u(x
0
) =
1

n
_
|x|=1
u(x
0
+rx) ds.
If
_
|xy|<1
[u(y)[ dy < C x R
n
, we have [u(x)[ < C in R
n
.
Since u is harmonic and bounded in R
n
, u is constant by Liouvilles theorem.
53
53
Liouvilles Theorem: A bounded harmonic function dened on R
n
is a constant.
Partial Dierential Equations Igor Yanovsky, 2005 236
Problem (S01, #1). For bodies (bounded regions B in R
3
) which are not perfectly
conducting one considers the boundary value problem
0 = (x)u =
3

j=1

x
j
_
(x)
u
x
j
_
u = f on B.
The function (x) is the local conductivity of B and u is the voltage. We dene
operator (f) mapping the boundary data f to the current density at the boundary by
(f) = (x)
u
n
,
and /n is the inward normal derivative (this formula denes the current density).
a) Show that is a symmetric operator, i.e. prove
_
B
g(f) dS =
_
B
f(g) dS.
b) Use the positivity of (x) > 0 to show that is negative as an operator, i.e., prove
_
B
f(f) dS 0.
Proof. a) Let
_
(x)u = 0 on ,
u = f on .
_
(x)v = 0 on ,
v = g on .
(f) = (x)
u
n
, (g) = (x)
v
n
.
Since /n is inward normal derivative, Greens formula is:

v
..
=g
(x)
u
n
dS
_

v (x)u dx =
_

v (x)u dx.
We have
_

g(f) dS =
_

g(x)
u
n
dS =
_

v (x)u dx
_

v (x)u
. .
=0
dx
=
_

u(x)
v
n
dS +
_

u (x)v
. .
=0
dx
=
_

f(x)
v
n
dS =
_

f(g) dS.
b) We have (x) > 0.
_

f(f) dS =
_

u(x)
u
n
dS =
_

u (x)u
. .
=0
dx
_

(x)u u dx
=
_

(x)[u[
2
. .
0
0.
Partial Dierential Equations Igor Yanovsky, 2005 237
Problem (S01, #4). The Poincare Inequality states that for any bounded domain
in R
n
there is a constant C such that
_

[u[
2
dx C
_

[u[
2
dx
for all smooth functions u which vanish on the boundary of .
a) Find a formula for the best (smallest) constant for the domain in terms of the
eigenvalues of the Laplacian on , and
b) give the best constant for the rectangular domain in R
2
= (x, y) : 0 x a, 0 y b.
Proof. a) Consider Greens formula:
_

u v dx =
_

v
u
n
ds
_

v u dx.
Setting u = v and with u vanishing on , Greens formula becomes:
_

[u[
2
dx =
_

uu dx.
Expanding u in the eigenfunctions of the Laplacian, u(x) =

a
n

n
(x), the formula
above gives
_

[u[
2
dx =
_

n=1
a
n

n
(x)

m=1

m
a
m

m
(x) dx =

m,n=1

m
a
n
a
m
_

m
dx
=

n=1

n
[a
n
[
2
.
Also,
_

[u[
2
dx =
_

n=1
a
n

n
(x)

m=1
a
m

m
(x) =

n=1
[a
n
[
2
.
Comparing and , and considering that
n
increases as n , we obtain

1
_

[u[
2
dx =
1

n=1
[a
n
[
2

n=1

n
[a
n
[
2
=
_

[u[
2
dx.
_

[u[
2
dx
1

1
_

[u[
2
dx,
with C = 1/
1
.
b) For the rectangular domain = (x, y) : 0 x a, 0 y b R
2
, the
eigenvalues of the Laplacian are

mn
=
2
_
m
2
a
2
+
n
2
b
2
_
, m, n = 1, 2, . . . .

1
=
11
=
2
_
1
a
2
+
1
b
2
_
,
C =
1

11
=
1

2
1
(
1
a
2
+
1
b
2
)
.
Partial Dierential Equations Igor Yanovsky, 2005 238
Partial Dierential Equations Igor Yanovsky, 2005 239
Problem (S01, #6). a) Let B be a bounded region in R
3
with smooth boundary B.
The conductor potential for the body B is the solution of Laplaces equation outside
B
V = 0 in R
3
/B
subject to the boundary conditions, V = 1 on B and V (x) tends to zero as [x[ .
Assuming that the conductor potential exists, show that it is unique.
b) The capacity C(B) of B is dened to be the limit of [x[V (x) as [x[ . Show
that
C(B) =
1
4
_
B
V
n
dS,
where B is the boundary of B and n is the outer unit normal to it (i.e. the normal
pointing toward innity).
c) Suppose that B

## B. Show that C(B

) C(B).
Proof. a) Let V
1
, V
2
be two solutions of the boundary value problem. Dene W =
V
1
V
2
. Then W satises

W = 0 in R
3
/B
W = 0 on B
W 0 as [x[ .
Consider Greens formula:
_
B
u v dx =
_
B
v
u
n
ds
_
B
v u dx.
Setting W = u = v gives
_
B
[W[
2
dx =
_
B
W
..
=0
W
n
ds
_
B
W W
..
=0
dx = 0.
Thus, [W[
2
= 0 W = const. Since W = 0 on B, W 0, and V
1
= V
2
.
b & c) For (b)&(c), see the solutions from Ralstons homework (a few pages
down).
Partial Dierential Equations Igor Yanovsky, 2005 240
Problem (W03, #2). Let L be the second order dierential operator L = a(x)
in which x = (x
1
, x
2
, x
3
) is in the three-dimensional cube C = 0 < x
i
< 1, i = 1, 2, 3.
Suppose that a > 0 in C. Consider the eigenvalue problem
Lu = u for x C
u = 0 for x C.
a) Show that all eigenvalues are negative.
b) If u and v are eigenfunctions for distinct eigenvalues and , show that u and v
are orthogonal in the appropriate product.
c) If a(x) = a
1
(x
1
) +a
2
(x
2
) +a
3
(x
3
) nd an expression for the eigenvalues and eigen-
vectors of L in terms of the eigenvalues and eigenvectors of a set of one-dimensional
problems.
Proof. a) We have
u a(x)u = u.
Multiply the equation by u and integrate:
uu a(x)u
2
= u
2
,
_

uu dx
_

a(x)u
2
dx =
_

u
2
dx,
_

u
u
n
ds
. .
=0

[u[
2
dx
_

a(x)u
2
dx =
_

u
2
dx,
=

_

([u[
2
+a(x)u
2
) dx
_

u
2
dx
< 0.
b) Let , , be the eigenvalues and u, v be the corresponding eigenfunctions. We have
u a(x)u = u. (17.5)
v a(x)v = v. (17.6)
Multiply (17.5) by v and (17.6) by u and subtract equations from each other
vu a(x)uv = uv,
uv a(x)uv = uv.
vu uv = ( )uv.
Integrating over gives
_

_
vu uv
_
dx = ( )
_

uv dx,
_

_
v
u
n
u
v
n
. .
=0
_
dx = ( )
_

uv dx.
Since ,= , u and v are orthogonal on .
Partial Dierential Equations Igor Yanovsky, 2005 241
c) The three one-dimensional eigenvalue problems are:
u
1x
1
x
1
(x
1
) a(x
1
)u
1
(x
1
) =
1
u
1
(x
1
),
u
2x
2
x
2
(x
2
) a(x
2
)u
2
(x
2
) =
2
u
2
(x
2
),
u
3x
3
x
3
(x
3
) a(x
3
)u
3
(x
3
) =
3
u
3
(x
3
).
We need to derive how u
1
, u
2
, u
3
and
1
,
2
,
3
are related to u and .
Partial Dierential Equations Igor Yanovsky, 2005 242
17.7 Spherical Means
Problem (S95, #4). Consider the biharmonic operator in R
3

2
u
_

2
x
2
+

2
y
2
+

2
z
2
_
2
u.
a) Show that
2
is self-adjoint on [x[ < 1 with the following boundary conditions on
[x[ = 1:
u = 0,
u = 0.
Proof. a) We have
Lu =
2
u = 0
u = 0 on [x[ = 1
u = 0 on [x[ = 1.
The operator L is self-adjoint if:
_

(u Lv v Lu) dx = 0.
_

(u Lv v Lu) dx =
_

(u
2
v v
2
u) dx
=
_

u
v
n
ds
. .
=0

u (v) dx
_

v
u
n
ds
. .
=0
+
_

v (u) dx
=
_

v
u
n
ds
. .
=0
+
_

uv dx +
_

u
v
n
ds
. .
=0

vu dx = 0.
Partial Dierential Equations Igor Yanovsky, 2005 243
b) Denote [x[ = r and dene the averages
S(r) = (4r
2
)
1
_
|x|=r
u(x) ds,
V (r) =
_
4
3
r
3
_
1
_
|x|r
u(x) dx.
Show that
d
dr
S(r) =
r
3
V (r).
Hint: Rewrite S(r) as an integral over the unit sphere before dierentiating; i.e.,
S(r) = (4)
1
_
|x

|=1
u(rx

) dx

.
c) Use the result of (b) to show that if u is biharmonic, i.e.
2
u = 0, then
S(r) = u(0) +
r
2
6
u(0).
Hint: Use the mean value theorem for u.
b) Let x

= x/[x[. We have
54
S(r) =
1
4r
2
_
|x|=r
u(x) dS
r
=
1
4r
2
_
|x

|=1
u(rx

) r
2
dS
1
=
1
4
_
|x

|=1
u(rx

) dS
1
.
dS
dr
=
1
4
_
|x

|=1
u
r
(rx

) dS
1
=
1
4
_
|x

|=1
u
n
(rx

) dS
1
=
1
4r
2
_
|x|=r
u
n
(x) dS
r
=
1
4r
2
_
|x|r
u dx.
where we have used Greens identity in the last equality. Also
r
3
V (r) =
1
4r
2
_
|x|r
u dx.
c) Since u is biharmonic (i.e. u is harmonic), u has a mean value property. We
have
d
dr
S(r) =
r
3
V (r) =
r
3
_
4
3
r
3
_
1
_
|x|r
u(x) dx =
r
3
u(0),
S(r) =
r
2
6
u(0) +S(0) = u(0) +
r
2
6
u(0).
54
Change of variables:
Surface integrals: x = rx

in R
3
:
_
|x|=r
u(x) dS =
_
|x

|=1
u(rx

) r
2
dS1.
Volume integrals:

= r in R
n
:
_
|

|<r
h(x +

) d

=
_
||<1
h(x + r) r
n
d.
Partial Dierential Equations Igor Yanovsky, 2005 244
Partial Dierential Equations Igor Yanovsky, 2005 245
Problem (S00, #7). Suppose that u = u(x) for x R
3
is biharmonic;
i.e. that
2
u (u) = 0. Show that
(4r
2
)
1
_
|x|=r
u(x) ds(x) = u(0) + (r
2
/6)u(0)
through the following steps:
a) Show that for any smooth f,
d
dr
_
|x|r
f(x) dx =
_
|x|=r
f(x) ds(x).
b) Show that for any smooth f,
d
dr
(4r
2
)
1
_
|x|=r
f(x) ds(x) = (4r
2
)
1
_
|x|=r
n f(x, y) ds
in which n is the outward normal to the circle [x[ = r.
c) Use step (b) to show that
d
dr
(4r
2
)
1
_
|x|=r
f(x) ds(x) = (4r
2
)
1
_
|x|r
f(x) dx.
d) Combine steps (a) and (c) to obtain the nal result.
Proof. a) We can express the integral in Spherical Coordinates:
55
_
|x|R
f(x) dx =
_
R
0
_
2
0
_

0
f(, , r) r
2
sin d d dr.
d
dr
_
|x|R
f(x) dx =
d
dr
_
R
0
_
2
0
_

0
f(, , r) r
2
sin d d dr = ???
=
_
2
0
_

0
f(, , r) R
2
sin d d
=
_
|x|=R
f(x) dS.
55
Dierential Volume in spherical coordinates:
d
3
=
2
sin d d d.
Dierential Surface Area on sphere:
dS =
2
sin d d.
Partial Dierential Equations Igor Yanovsky, 2005 246
b&c) We have
d
dr
_
1
4r
2
_
|x|=r
f(x) dS
_
=
d
dr
_
1
4r
2
_
|x

|=1
f(rx

) r
2
dS
1
_
=
1
4
d
dr
__
|x

|=1
f(rx

) dS
1
_
=
1
4
_
|x

|=1
f
r
(rx

) dS
1
=
1
4
_
|x

|=1
f
n
(rx

) dS
1
=
1
4r
2
_
|x|=r
f
n
(x) dS =
1
4r
2
_
|x|=r
f ndS
=
1
4r
2
_
|x|r
f dx.
Greens formula was used in the last equality.
Alternatively,
d
dr
_
1
4r
2
_
|x|=r
f(x) dS
_
=
d
dr
_
1
4r
2
_
2
0
_

0
f(, , r) r
2
sin d d
_
=
d
dr
_
1
4
_
2
0
_

0
f(, , r) sin d d
_
=
1
4
_
2
0
_

0
f
r
(, , r) sin d d
=
1
4
_
2
0
_

0
f n sin d d
=
1
4r
2
_
2
0
_

0
f n r
2
sin d d
=
1
4r
2
_
|x|=r
f ndS
=
1
4r
2
_
|x|=r
f dx.
d) Since f is biharmonic (i.e. f is harmonic), f has a mean value property. From
(c), we have
56
d
dr
_
1
4r
2
_
|x|=r
f(x) ds(x)
_
=
1
4r
2
_
|x|r
f(x) dx =
r
3
1
4
3
r
3
_
|x|r
f(x) dx
=
r
3
f(0).
1
4r
2
_
|x|=r
f(x) ds(x) =
r
2
6
f(0) +f(0).
56
Note that part (a) was not used. We use exactly the same derivation as we did in S95 #4.
Partial Dierential Equations Igor Yanovsky, 2005 247
Problem (F96, #4).
Consider smooth solutions of u = k
2
u in dimension d = 2 with k > 0.
a) Show that u satises the following mean value property:
M

x
(r) +
1
r
M

x
(r) k
2
M
x
(r) = 0,
in which M
x
(r) is dened by
M
x
(r) =
1
2
_
2
0
u(x + r cos , y +r sin) d
and the derivatives (denoted by

) are in r with x xed.
b) For k = 1, this equation is the modied Bessel equation (of order 0)
f

+
1
r
f

f = 0,
for which one solution (denoted as I
0
) is
I
0
(r) =
1
2
_
2
0
e
r sin
d.
Find an expression for M
x
(r) in terms of I
0
.
Proof. a) Laplacian in polar coordinates written as:
u = u
rr
+
1
r
u
r
+
1
r
2
u

.
Thus, the equation may be written as
u
rr
+
1
r
u
r
+
1
r
2
u

= k
2
u.
M
x
(r) =
1
2
_
2
0
u d,
M

x
(r) =
1
2
_
2
0
u
r
d,
M

x
(r) =
1
2
_
2
0
u
rr
d.
M

x
(r) +
1
r
M

x
(r) k
2
M
x
(r) =
1
2
_
2
0
_
u
rr
+
1
r
u
r
k
2
u
_
d
=
1
2r
2
_
2
0
u

d =
1
2r
2
_
u

2
0
= 0.
b) Note that w = e
r sin
satises w = w, i.e.
w = w
rr
+
1
r
w
r
+
1
r
2
w

= sin
2
e
r sin
+
1
r
sin e
r sin
+
1
r
2
(r sin e
r sin
+r
2
cos
2
e
r sin
) = e
r sin
= w.
Thus,
M
x
(r) = e
y
1
2
_
2
0
e
r sin
d = e
y
I
0
.
Partial Dierential Equations Igor Yanovsky, 2005 248
57
57
Check with someone about the last result.
Partial Dierential Equations Igor Yanovsky, 2005 249
17.8 Harmonic Extensions, Subharmonic Functions
Problem (S94, #8). Suppose that is a bounded region in R
3
and that u = 1 on
. If u = 0 in the exterior region R
3
/ and u(x) 0 as [x[ , prove the
following:
a) u > 0 in R
3
/;
b) if (x) is a smooth function such that (x) = 1 for [x[ > R and (x) = 0 near ,
then for [x[ > R,
u(x) =
1
4
_
R
3
/
((u))(y)
[x y[
dy.
c) lim
|x|
[x[u(x) exists and is non-negative.
Proof. a) Let B
r
(0) denote the closed ball x : [x[ r.
Given > 0, we can nd r large enough that B
R
1
(0) and max
xB
R
1
(0)
[u(x)[ < ,
since [u(x)[ 0 as [x[ .
Since u is harmonic in B
R
1
, it takes its maximum and minimum on the boundary.
Assume
min
xB
R
1
(0)
u(x) = a < 0 (where [a[ < ).
We can nd an R
2
such that max
xB
R
2
(0)
[u(x)[ <
a
2
; hence u takes a minimum inside
B
R
2
(0) , which is impossible; hence u 0.
Now let V = x : u(x) ,= 0 and let = min
xV
[x[. Since u cannot take a minimum
inside B
R
(0) (where R > ), it follows that u C and C = 0, but this contradicts
u = 1 on . Hence u > 0 in R
3
.
b) For n = 3,
K([x y[) =
1
(2 n)
n
[x y[
2n
=
1
4
1
[x y[
.
Since (x) = 1 for [x[ > R, then for x / B
R
, we have (u) = u = 0. Thus,

1
4
_
R
3
/
((u))(y)
[x y[
dy
=
1
4
_
B
R
/
((u))(y)
[x y[
dy
=
1
4
_
B
R
/

y
_
1
[x y[
_

y
(u) dy
1
4
_
(B
R
/)

n
_
u
_
1
[x y[
dS
y
=
1
4
_
B
R
/

_
1
[x y[
_
u dy +
1
4
_
(B
R
/)

n
_
1
[x y[
_
u dS
y

1
4
_
(B
R
/)

n
_
u
_
1
[x y[
dS
y
= ??? = u(x)
1
4R
2
_
B
u dS
y
. .
0, as R

1
4R
_
B
u
n
dS
y
. .
0, as R
= u(x).
c) See the next problem.
Partial Dierential Equations Igor Yanovsky, 2005 250
Ralston Hw. a) Suppose that u is a smooth function on R
3
and u = 0 for [x[ > R.
If lim
x
u(x) = 0, show that you can write u as a convolution of u with the
1
4|x|
and prove that lim
x
[x[u(x) = 0 exists.
b) The conductor potential for R
3
is the solution to the Dirichlet problem v =
0. The limit in part (a) is called the capacity of . Show that if
1

2
, then the
capacity of
2
is greater or equal the capacity of
1
.
Proof. a) If we dene
v(x) =
1
4
_
R
3
u(y)
[x y[
dy,
then (u v) = 0 in all R
3
, and, since v(x) 0 as [x[ , we have lim
|x|
(u(x)
v(x)) = 0. Thus, u v must be bounded, and Liouvilles theorem implies that it is
identically zero. Since we now have
[x[u(x) =
1
4
_
R
3
[x[ u(y)
[x y[
dy,
and [x[/[x y[ converges uniformly to 1 on [y[ R, it follows that
lim
|x|
[x[u(x) =
1
4
_
R
3
u(y) dy.
b) Note that part (a) implies that the limit lim
|x|
[x[v(x) exists, because we can
apply (a) to u(x) = (x)v(x), where is smooth and vanishes on , but (x) = 1 for
[x[ > R.
Let v
1
be the conductor potential for
1
and v
2
for
2
. Since v
i
as [x[ and
v
i
= 1 on
i
, the max principle says that 1 > v
i
(x) > 0 for x R
3

i
. Consider
v
2
v
1
. Since
1

2
, this is dened in R
3

2
, positive on
2
, and has limit 0 as
[x[ . Thus, it must be positive in R
3

2
. Thus, lim
|x|
[x[(v
2
v
1
) 0.
Problem (F95, #4).
58
Let be a simply connected open domain in R
2
and u = u(x, y) be subharmonic there, i.e. u 0 in . Prove that if
D
R
= (x, y) : (x x
0
)
2
+ (y y
0
)
2
R
2

then
u(x
0
, y
0
)
1
2
_
2
0
u(x
0
+ Rcos , y
0
+Rsin) d.
Proof. Let
M(x
0
, R) =
1
2
_
2
0
u(x
0
+Rcos , y
0
+Rsin ) d,
w(r, ) = u(x
0
+Rcos , y
0
+Rsin ).
Dierentiate M(x
0
, R) with respect to R:
d
dr
M(x
0
, R) =
1
2R
_
2
0
w
r
(R, )Rd,
58
See McOwen, Sec.4.3, p.131, #1.
Partial Dierential Equations Igor Yanovsky, 2005 251
59
59
See ChiuYens solutions and Sung Has solutions (in two places). Nicks solutions, as started above,
have a very simplistic approach.
Partial Dierential Equations Igor Yanovsky, 2005 252
Ralston Hw (Maximum Principle).
Suppose that u C() satises the mean value property in the connected open set .
a) Show that u satises the maximum principle in , i.e.
either u is constant or u(x) < sup

u for all x .
b) Show that, if v is a continuous function on a closed ball B
r
() and has the
mean value property in B
r
(), then u = v on B
r
() implies u = v in B
r
(). Does this
imply that u is harmonic in ?
Proof. a) If u(x) is not less than sup

## u for all x , then the set

K = x : u(x) = sup

u
is nonempty. This set is closed because u is continuous. We will show it is also open.
This implies that K = because is connected. Thus u is constant on .
Let x
0
K. Since is open, > 0, s.t. B

(x
0
) = x R
n
: [x x
0
[ . Let
sup

## u = M. By the mean value property, for 0 r

M = u(x
0
) =
1
A(S
n1
)
_
||=1
u(x
0
+r)dS

, and 0 =
1
A(S
n1
)
_
||=1
(M u(x
0
+ r))dS

.
Sinse Mu(x
0
+r) is a continuous nonnegative function on , this implies Mu(x
0
+
r) = 0 for all S
n1
. Thus u = 0 on B

(x
0
).
b) Since u v has the mean value property in the open interior of B
r
(), by part
a) it satises the maximum principle. Since it is continuous on B
r
(), its supremum
over the interior of B
r
() is its maximum on B
r
(), and this maximum is assumed at a
point x
0
in B
r
(). If x
0
in the interior of B
r
(), then uv is constant ant the constant
must be zero, since this is the value of uv on the boundary. If x
0
is on the boundary,
then u v must be nonpositive in the interior of B
r
().
Applying the same argument to v u, one nds that it is either identically zero or
nonpositive in the interior of B
r
(). Thus, u v 0 on B
r
().
Yes, it does follow that u harmonic in . Take v in the preceding to be the harmonic
function in the interior of B
r
() which agrees with u on the boundary. Since u = v on
B
r
(), u is harmonic in the interior of B
r
(). Since is open we can do this for every
. Thus u is harmonic in .
Partial Dierential Equations Igor Yanovsky, 2005 253
Ralston Hw. Assume is a bounded open set in R
n
and the Greens function, G(x, y),
for exists. Use the strong maximum principle, i.e. either u(x) < sup

u for all x ,
or u is constant, to prove that G(x, y) < 0 for x, y , x ,= y.
Proof. G(x, y) = K(x, y) +(x, y). For each x , f(y) = (x, y) is continuous on ,
thus, bounded. So [(x, y)[ M
x
for all y . K(x y) as y x. Thus,
given M
x
, there is > 0, such that K(x y) < M
x
when [x y[ = r and 0 < r .
So for 0 < r the Greens function with x xed satises, G(x, y) is harmonic on
B
r
(x), and G(x, y) 0 on the boundary of B
r
(x). Since we can choose r as
small as we wish, we get G(x, y) < 0 for y x.
Problem (W03, #6). Assume that u is a harmonic function in the half ball
D = (x, y, z) : x
2
+y
2
+z
2
< 1, z 0 which is continuously dierentiable, and satis-
es u(x, y, 0) = 0. Show that u can be extended to be a harmonic function in the whole
ball. If you propose and explicit extension for u, explain why the extension is harmonic.
Proof. We can extend u to all of n-space by dening
u(x

, x
n
) = u(x

, x
n
)
for x
n
< 0. Dene
(x) =
1
a
n
_
|y|=1
a
2
[x[
2
[x y[
n
v(y)dS
y
(x) is continuous on a closed ball B, harmonic in B.
Poisson kernel is symmetric in y at x
n
= 0. (x) = 0, (x
n
= 0).
is harmonic for x B, x
n
0,with the same boundary values = u.
is harmonic u can be extended to a harmonic function on the interior of B.
Ralston Hw. Show that a bounded solution to the Dirichlet problem in a half
space is unique. (Note that one can show that a bounded solution exists for any
given bounded continuous Dirichlet data by using the Poisson kernel for the half space.)
Proof. We have to show that a function, u, which is harmonic in the half-space, con-
tinuous, equal to 0 when x
n
= 0, and bounded, must be identically 0. We can extend
u to all of n-space by dening
u(x

, x
n
) = u(x

, x
n
)
for x
n
< 0. This extends u to a bounded harmonic function on all of n-space (by the
problem above). Liouvilles theorem says u must be constant, and since u(x

, 0) = 0,
the constant is 0. So the original u must be identically 0.
Ralston Hw. Suppose u is harmonic on the ball minus the origin, B
0
= x R
3
:
0 < [x[ < a. Show that u(x) can be extended to a harmonic function on the ball
B = [x[ < a i lim
|x|0
[x[u(x) = 0.
Proof. The condition lim
|x|0
[x[u(x) = 0 is necessary, because harmonic functions are
continuous.
To prove the converse, let v be the function which is continuous on [x[ a/2,
harmonic on [x[ < a/2, and equals u on [x[ = a/2. One can construct v using the
Poisson kernel. Since v is continuous, it is bounded, and we can assume that [v[ M.
Since lim
|x|0
[x[u(x) = 0, given > 0, we can choose , 0 < < a/2 such that
< [x[u(x) < when [x[ < . Note that u, v 2/[x[, and v + 2/[x[ are harmonic
Partial Dierential Equations Igor Yanovsky, 2005 254
on 0 < [x[ < a/2. Choose b, 0 < b < min(, a/2), so that /b > M. Then on both
[x[ = a/2 and [x[ = b we have v 2/[x[ < u(x) < v + 2/[x[. Thus, by
max principle these inequalities hold on b [x[ a/2. Pick x with 0 < [x[ a/2.
u(x) = v(x). v is the extension of u on [x[ < a/2, and u is extended on [x[ < a.
Partial Dierential Equations Igor Yanovsky, 2005 255
18 Problems: Heat Equation
McOwen 5.2 #7(a). Consider

u
t
= u
xx
for x > 0, t > 0
u(x, 0) = g(x) for x > 0
u(0, t) = 0 for t > 0,
where g is continuous and bounded for x 0 and g(0) = 0.
Find a formula for the solution u(x, t).
Proof. Extend g to be an odd function on all of R:
g(x) =
_
g(x), x 0
g(x), x < 0.
Then, we need to solve
_
u
t
= u
xx
for x R, t > 0
u(x, 0) = g(x) for x R.
The solution is given by:
60
u(x, t) =
_
R
K(x, y, t)g(y) dy =
1

4t
_

(xy)
2
4t
g(y) dy
=
1

4t
__

0
e

(xy)
2
4t
g(y) dy +
_
0

(xy)
2
4t
g(y) dy
_
=
1

4t
__

0
e

(xy)
2
4t
g(y) dy
_

0
e

(x+y)
2
4t
g(y) dy
_
=
1

4t
_

0
_
e
x
2
+2xyy
2
4t
e
x
2
2xyy
2
4t
_
g(y) dy
=
1

4t
_

0
e

(x
2
+y
2
)
4t
_
e
xy
2t
e

xy
2t
_
g(y) dy.
u(x, t) =
1

4t
_

0
e

(x
2
+y
2
)
4t
2 sinh
_
xy
2t
_
g(y) dy.
Since sinh(0) = 0, we can verify that u(0, t) = 0.
60
In calculations, we use:
_
0

e
y
dy =
_

0
e
y
dy, and g(y) = g(y).
Partial Dierential Equations Igor Yanovsky, 2005 256
McOwen 5.2 #7(b). Consider

u
t
= u
xx
for x > 0, t > 0
u(x, 0) = g(x) for x > 0
u
x
(0, t) = 0 for t > 0,
where g is continuous and bounded for x 0.
Find a formula for the solution u(x, t).
Proof. Extend g to be an even function
61
on all of R:
g(x) =
_
g(x), x 0
g(x), x < 0.
Then, we need to solve
_
u
t
= u
xx
for x R, t > 0
u(x, 0) = g(x) for x R.
The solution is given by:
62
u(x, t) =
_
R
K(x, y, t)g(y) dy =
1

4t
_

(xy)
2
4t
g(y) dy
=
1

4t
__

0
e

(xy)
2
4t
g(y) dy +
_
0

(xy)
2
4t
g(y) dy
_
=
1

4t
__

0
e

(xy)
2
4t
g(y) dy +
_

0
e

(x+y)
2
4t
g(y) dy
_
=
1

4t
_

0
_
e
x
2
+2xyy
2
4t
+ e
x
2
2xyy
2
4t
_
g(y) dy
=
1

4t
_

0
e

(x
2
+y
2
)
4t
_
e
xy
2t
+e

xy
2t
_
g(y) dy.
u(x, t) =
1

4t
_

0
e

(x
2
+y
2
)
4t
2 cosh
_
xy
2t
_
g(y) dy.
To check that the boundary condition holds, we perform the calculation:
u
x
(x, t) =
1

4t
_

0
d
dx
_
e

(x
2
+y
2
)
4t
2 cosh
_
xy
2t
_
_
g(y) dy
=
1

4t
_

0
_

2x
4t
e

(x
2
+y
2
)
4t
2 cosh
_
xy
2t
_
+ e

(x
2
+y
2
)
4t
2
y
2t
sinh
_
xy
2t
_
_
g(y) dy,
u
x
(0, t) =
1

4t
_

0
_
0 e

y
2
4t
2 cosh0 + e

y
2
4t
2
y
2t
sinh 0
_
g(y) dy = 0.
61
Even extensions are always continuous. Not true for odd extensions. g odd is continuous if g(0) =
0.
62
In calculations, we use:
_
0

e
y
dy =
_

0
e
y
dy, and g(y) = g(y).
Partial Dierential Equations Igor Yanovsky, 2005 257
Problem (F90, #5).
The initial value problem for the heat equation on the whole real line is
f
t
= f
xx
t 0
f(t = 0, x) = f
0
(x)
with f
0
smooth and bounded.
a) Write down the Greens function G(x, y, t) for this initial value problem.
b) Write the solution f(x, t) as an integral involving G and f
0
.
c) Show that the maximum values of [f(x, t)[ and [f
x
(x, t)[ are non-increasing
as t increases, i.e.
sup
x
[f(x, t)[ sup
x
[f
0
(x)[ sup
x
[f
x
(x, t)[ sup
x
[f
0x
(x)[.
When are these inequalities actually equalities?
Proof. a) The fundamental solution
K(x, y, t) =
1

4t
e

|xy|
2
4t
.
The Greens function is:
63
G(x, t; y, s) =
1
(2)
n
_

k(t s)
_n
2
e

(xy)
2
4k(ts)
.
b) The solution to the one-dimensional heat equation is
u(x, t) =
_
R
K(x, y, t) f
0
(y) dy =
1

4t
_
R
e

|xy|
2
4t
f
0
(y) dy.
c) We have
sup
x
[u(x, t)[ =

4t
_
R
e

(xy)
2
4t
f
0
(y) dy

4t
_
R
e

(xy)
2
4t

f
0
(y)

dy
=
1

4t
_
R
e

y
2
4t

f
0
(x y)

dy
sup
x
[f
0
(x)[
1

4t
_
R
e

y
2
4t
dy
_
z =
y

4t
, dz =
dy

4t
_
sup
x
[f
0
(x)[
1

4t
_
R
e
z
2

4t dz
= sup
x
[f
0
(x)[
1

_
R
e
z
2
dz
. .
=

= sup
x
[f
0
(x)[.
63
The Greens function for the heat equation on an innite domain; derived in R. Haberman using
the Fourier transform.
Partial Dierential Equations Igor Yanovsky, 2005 258
u
x
(x, t) =
1

4t
_
R

2(x y)
4t
e

(xy)
2
4t
f
0
(y) dy =
1

4t
_
R

d
dy
_
e

(xy)
2
4t
_
f
0
(y) dy
=
1

4t
_
e

(xy)
2
4t
f
0
(y)
_

. .
= 0
+
1

4t
_
R
e

(xy)
2
4t
f
0y
(y) dy,
sup
x
[u(x, t)[
1

4t
sup
x
[f
0x
(x)[
_
R
e

(xy)
2
4t
dy =
1

4t
sup
x
[f
0x
(x)[
_
R
e
z
2

4t dz
= sup
x
[f
0x
(x)[.
These inequalities are equalities when f
0
(x) and f
0x
(x) are constants, respectively.
Partial Dierential Equations Igor Yanovsky, 2005 259
Problem (S01, #5). a) Show that the solution of the heat equation
u
t
= u
xx
, < x <
with square-integrable initial data u(x, 0) = f(x), decays in time, and there is a constant
independent of f and t such that for all t > 0
max
x
[u
x
(x, t)[ t

3
4
_
_
x
[f(x)[
2
dx
_1
2
.
b) Consider the solution of the transport equation
t
+u
x
= 0 with square-integrable
initial data (x, 0) =
0
(x) and the velocity u from part (a). Show that (x, t) remains
square-integrable for all nite time
_
R
[(x, t)[
2
dx e
Ct
1
4
_
R
[
0
(x)[
2
dx,
where C does not depend on
0
.
Proof. a) The solution to the one-dimensional homogeneous heat equation is
u(x, t) =
1

4t
_
R
e

(xy)
2
4t
f(y) dy.
Take the derivative with respect to x, we get
64
u
x
(x, t) =
1

4t
_
R

2(x y)
4t
e

(xy)
2
4t
f(y) dy =
1
4t
3
2

_
R
(x y)e

(xy)
2
4t
f(y) dy.
[u
x
(x, t)[
1
4t
3
2

_
R

(x y)e

(xy)
2
4t
f(y)

dy (Cauchy-Schwarz)

1
4t
3
2

_
_
R
(x y)
2
e

(xy)
2
2t
dy
_1
2
[[f[[
L
2
(R)
_
z =
x y

2t
, dz =
dy

2t
_
=
1
4t
3
2

_
_
R

z
2
(2t)
3
2
e
z
2

dz
_1
2
[[f[[
L
2
(R)
=
(2t)
3
4
4t
3
2

_
_
R
z
2
e
z
2
dz
. .
M<
_1
2
[[f[[
L
2
(R)
= Ct

3
4
M
1
2
[[f[[
L
2
(R)
= t

3
4
[[f[[
L
2
(R)
.
b) Note:
max
x
[u[ = max
x

4t
_
R
e

(xy)
2
4t
f(y) dy

4t
_
_
R
e

(xy)
2
2t
dy
_1
2
[[f[[
L
2
(R)

4t
_
_
R

e
z
2

2t

dz
_1
2
[[f[[
L
2
(R)
_
z =
x y

2t
, dz =
dy

2t
_
=
(2t)
1
4
2
1
2
t
1
2
_
_
R
e
z
2
dz
. .
=

_1
2
[[f[[
L
2
(R)
= Ct

1
4
[[f[[
L
2
(R)
.
65
64
Cauchy-Schwarz: |(u, v)| ||u||||v|| in any norm, for example
_
|uv|dx (
_
u
2
dx)
1
2
(
_
v
2
dx)
1
2
65
See Yanas and Alans solutions.
Partial Dierential Equations Igor Yanovsky, 2005 260
Problem (F04, #2).
Let u(x, t) be a bounded solution to the Cauchy problem for the heat equation
_
u
t
= a
2
u
xx
, t > 0, x R, a > 0,
u(x, 0) = (x).
Here (x) C(R) satises
lim
x+
(x) = b, lim
x
(x) = c.
Compute the limit of u(x, t) as t +, x R. Justify your argument carefully.
Proof. For a = 1, the solution to the one-dimensional homogeneous heat equation is
u(x, t) =
1

4t
_
R
e

(xy)
2
4t
(y) dy.
We want to transform the equation to v
t
= v
xx
. Make a change of variables: x = ay.
u(x, t) = u(x(y), t) = u(ay, t) = v(y, t). Then,
v
y
= u
x
x
y
= au
x
,
v
yy
= au
xx
x
y
= a
2
u
xx
,
v(y, 0) = u(ay, 0) = (ay).
Thus, the new problem is:
_
v
t
= v
yy
, t > 0, y R,
v(y, 0) = (ay).
v(y, t) =
1

4t
_
R
e

(yz)
2
4t
(az) dz.
Since is continuous, and lim
x+
(x) = b, lim
x
(x) = c, we have
[(x)[ < M, x R. Thus,
[v(y, t)[
M

4t
_
R
e

z
2
4t
dz
_
s =
z

4t
, ds =
dz

4t
_
=
M

4t
_
R
e
s
2

4t ds =
M

_
R
e
s
2
ds
. .

= M.
Integral in converges uniformly lim
_
=
_
lim. For = (a):
v(y, t) =
1

4t
_

(yz)
2
4t
(z) dz =
1

4t
_

z
2
4t
(y z) dz
=
1

4t
_

e
s
2
(y s

4t)

4t ds
=
1

e
s
2
(y s

4t) ds.
Partial Dierential Equations Igor Yanovsky, 2005 261
lim
t+
v(y, t) =
1

_

0
e
s
2
lim
t+
(y s

4t) ds +
1

_
0

e
s
2
lim
t+
(y s

4t) ds
=
1

_

0
e
s
2
c ds +
1

_
0

e
s
2
b ds = c
1

2
+ b
1

2
=
c + b
2
.
Partial Dierential Equations Igor Yanovsky, 2005 262
Problem. Consider
u
t
= ku
xx
+Q, 0 < x < 1
u(0, t) = 0,
u(1, t) = 1.
What is the steady state temperature?
Proof. Set u
t
= 0, and integrate with respect to x twice:
ku
xx
+Q = 0,
u
xx
=
Q
k
,
u
x
=
Q
k
x +a,
u =
Q
k
x
2
2
+ ax +b.
Boundary conditions give
u(x) =
Q
2k
x
2
+
_
1 +
Q
2k
_
x.
Partial Dierential Equations Igor Yanovsky, 2005 263
18.1 Heat Equation with Lower Order Terms
McOwen 5.2 #11. Find a formula for the solution of
_
u
t
= u cu in R
n
(0, )
u(x, 0) = g(x) on R
n
.
(18.1)
Show that such solutions, with initial data g L
2
(R
n
), are unique, even when c is
negative.
Proof. McOwen. Consider v(x, t) = e
ct
u(x, t). The transformed problem is
_
v
t
= v in R
n
(0, )
v(x, 0) = g(x) on R
n
.
(18.2)
Since g is continuous and bounded in R
n
, we have
v(x, t) =
_
R
n
K(x, y, t) g(y) dy =
1
(4t)
n
2
_
R
n
e

|xy|
2
4t
g(y) dy,
u(x, t) = e
ct
v(x, t) =
1
(4t)
n
2
_
R
n
e

|xy|
2
4t
ct
g(y) dy.
u(x, t) is a bounded solution since v(x, t) is.
To prove uniqueness, assume there is another solution v

of (18.2). w = v v

satises
_
w
t
= w in R
n
(0, )
w(x, 0) = 0 on R
n
.
(18.3)
Since bounded solutions of (18.3) are unique, and since w is a nontrivial solution, w is
unbounded. Thus, v

## is unbounded, and therefore, the bounded solution v is unique.

Partial Dierential Equations Igor Yanovsky, 2005 264
18.1.1 Heat Equation Energy Estimates
Problem (F94, #3). Let u(x, y, t) be a twice continuously dierential solution of
u
t
= u u
3
in R
2
, t 0
u(x, y, 0) = 0 in
u(x, y, t) = 0 in , t 0.
Prove that u(x, y, t) 0 in [0, T].
Proof. Multiply the equation by u and integrate:
uu
t
= uu u
4
,
_

uu
t
dx =
_

uu dx
_

u
4
dx,
1
2
d
dt
_

u
2
dx =
_

u
u
n
ds
. .
=0

[u[
2
dx
_

u
4
dx,
1
2
d
dt
[[u[[
2
2
=
_

[u[
2
dx
_

u
4
dx 0.
Thus,
[[u(x, y, t)[[
2
[[u(x, y, 0)[[
2
= 0.
Hence, [[u(x, y, t)[[
2
= 0, and u 0.
Partial Dierential Equations Igor Yanovsky, 2005 265
Problem (F98, #5). Consider the heat equation
u
t
u = 0
in a two dimensional region . Dene the mass M as
M(t) =
_

u(x, t) dx.
a) For a xed domain , show M is a constant in time if the boundary conditions are
u/n = 0.
b) Suppose that = (t) is evolving in time, with a boundary that moves at velocity
v, which may vary along the boundary. Find a modied boundary condition (in terms
of local quantities only) for u, so that M is constant.
Hint: You may use the fact that
d
dt
_
(t)
f(x, t) dx =
_
(t)
f
t
(x, t) dx +
_
(t)
n v f(x, t) dl,
in which n is a unit normal vector to the boundary .
Proof. a) We have
_
u
t
u = 0, on
u
n
= 0, on .
We want to show that
d
dt
M(t) = 0. We have
66
d
dt
M(t) =
d
dt
_

u(x, t) dx =
_

u
t
dx =
_

u dx =
_

u
n
ds = 0.
b) We need
d
dt
M(t) = 0.
0 =
d
dt
M(t) =
d
dt
_
(t)
u(x, t) dx =
_
(t)
u
t
dx +
_
(t)
n v u ds
=
_
(t)
u dx +
_
(t)
n v u ds =
_
(t)
u
n
ds +
_
(t)
n v u ds
=
_
(t)
u nds +
_
(t)
n v u ds =
_
(t)
n (u +vu) ds.
Thus, we need:
n (u + vu) ds = 0, on .
66
The last equality below is obtained from the Greens formula:
_

u dx =
_

u
n
ds.
Partial Dierential Equations Igor Yanovsky, 2005 266
Problem (S95, #3). Write down an explicit formula for a function u(x, t) solving
_
u
t
+ b u +cu = u in R
n
(0, )
u(x, 0) = f(x) on R
n
.
(18.4)
where b R
n
and c R are constants.
Hint: First transform this to the heat equation by a linear change of the dependent
and independent variables. Then solve the heat equation using the fundamental solution.
Proof. Consider
u(x, t) = e
x+t
v(x, t).
u
t
= e
x+t
v +e
x+t
v
t
= (v
t
+ v)e
x+t
,
u = e
x+t
v +e
x+t
v = (v + v)e
x+t
,
(u) =
_
(v +v)e
x+t
_
= ( v + v)e
x+t
+ ([[
2
v + v)e
x+t
=
_
v + 2 v + [[
2
v)e
x+t
.
Plugging this into (18.4), we obtain
v
t
+v +b (v + v) + cv = v + 2 v +[[
2
v,
v
t
+
_
b 2
_
v +
_
+b +c [[
2
_
v = v.
In order to get homogeneous heat equation, we set
=
b
2
, =
[b[
2
4
c,
which gives
_
v
t
= v in R
n
(0, )
v(x, 0) = e

b
2
x
f(x) on R
n
.
The above PDE has the following solution:
v(x, t) =
1
(4t)
n
2
_
R
n
e

|xy|
2
4t
e

b
2
y
f(y) dy.
Thus,
u(x, t) = e
b
2
x(
|b|
2
4
+c)t
v(x, t) =
1
(4t)
n
2
e
b
2
x(
|b|
2
4
+c)t
_
R
n
e

|xy|
2
4t
e

b
2
y
f(y) dy.
Partial Dierential Equations Igor Yanovsky, 2005 267
Problem (F01, #7). Consider the parabolic problem
u
t
= u
xx
+ c(x)u (18.5)
for < x < , in which
c(x) = 0 for [x[ > 1,
c(x) = 1 for [x[ < 1.
Find solutions of the form u(x, t) = e
t
v(x) in which
_

[u[
2
dx < .
Hint: Look for v to have the form
v(x) = ae
k|x|
for [x[ > 1,
v(x) = b cos lx for [x[ < 1,
for some a, b, k, l.
Proof. Plug u(x, t) = e
t
v(x) into (18.5) to get:
e
t
v(x) = e
t
v

(x) + ce
t
v(x),
v(x) = v

(x) +cv(x),
v

## (x) v(x) + cv(x) = 0.

For [x[ > 1, c = 0. We look for solutions of the form v(x) = ae
k|x|
.
v

(x) v(x) = 0,
ak
2
e
k|x|
ae
k|x|
= 0,
k
2
= 0,
k
2
= ,
k =

.
Thus, v(x) = c
1
e

x
+ c
2
e

x
. Since we want
_

[u[
2
dx < :
u(x, t) = ae
t
e

x
.
For [x[ < 1, c = 1. We look for solutions of the form v(x) = b cos lx.
v

## (x) v(x) + v(x) = 0,

bl
2
cos lx + (1 )b coslx = 0,
l
2
+ (1 ) = 0,
l
2
= 1 ,
l =

1 .
Thus, (since cos(x) = cos x)
u(x, t) = be
t
cos
_
(1 )x.
We want v(x) to be continuous on R, and at x = 1, in particular. Thus,
ae

= b cos
_
(1 ),
a = be

cos
_
(1 ).
Also, v(x) is symmetric:
_

[u[
2
dx = 2
_

0
[u[
2
dx = 2
_ _
1
0
[u[
2
dx +
_

1
[u[
2
dx
_
< .
Partial Dierential Equations Igor Yanovsky, 2005 268
Problem (F03, #3). The function
h(X, T) = (4T)

1
2
e

X
2
4T
satises (you do not have to show this)
h
T
= h
XX
.
Using this result, verify that for any smooth function U
u(x, t) = e
1
3
t
3
xt
_

U() h(x t
2
, t) d
satises
u
t
+xu = u
xx
.
Given that U(x) is bounded and continuous everywhere on x , establish
that
lim
t0
_

## U() h(x , t) d = U(x)

and show that u(x, t) U(x) as t 0. (You may use the fact that
_

0
e

2
d =
1
2

.)
Proof. We change the notation: h K, U g, y. We have
K(X, T) =
1

4T
e

X
2
4T
We want to verify that
u(x, t) = e
1
3
t
3
xt
_

K(x y t
2
, t) g(y) dy.
satises
u
t
+xu = u
xx
.
We have
u
t
=
_

d
dt
_
e
1
3
t
3
xt
K(x y t
2
, t)
_
g(y) dy
=
_

_
(t
2
x) e
1
3
t
3
xt
K + e
1
3
t
3
xt
_
K
X
(2t) + K
T
_
_
g(y) dy,
xu =
_

x e
1
3
t
3
xt
K(x y t
2
, t) g(y) dy,
u
x
=
_

d
dx
_
e
1
3
t
3
xt
K(x y t
2
, t)
_
g(y) dy
=
_

_
t e
1
3
t
3
xt
K + e
1
3
t
3
xt
K
X
_
g(y) dy,
u
xx
=
_

d
dx
_
t e
1
3
t
3
xt
K + e
1
3
t
3
xt
K
X
_
g(y) dy
=
_

_
t
2
e
1
3
t
3
xt
K t e
1
3
t
3
xt
K
X
t e
1
3
t
3
xt
K
X
+ e
1
3
t
3
xt
K
XX
_
g(y) dy.
Partial Dierential Equations Igor Yanovsky, 2005 269
Plugging these into , most of the terms cancel out. The remaining two terms cancel
because K
T
= K
XX
.
Given that g(x) is bounded and continuous on x , we establish that
67
lim
t0
_

## K(x y, t) g(y) dy = g(x).

Fix x
0
R
n
, > 0. Choose > 0 such that
[g(y) g(x
0
)[ < if [y x
0
[ < , y R
n
.
Then if [x x
0
[ <

2
, we have: (
_
R
K(x, t) dx = 1)

_
R
K(x y, t) g(y) dy g(x
0
)

_
R
K(x y, t) [g(y) g(x
0
)] dy

_
B

(x
0
)
K(x y, t)

g(y) g(x
0
)

dy
. .

_
R
K(xy,t) dy =
+
_
RB

(x
0
)
K(x y, t)

g(y) g(x
0
)

dy
Furthermore, if [x x
0
[

2
and [y x
0
[ , then
[y x
0
[ [y x[ +

2
[y x[ +
1
2
[y x
0
[.
Thus, [y x[
1
2
[y x
0
[. Consequently,
= + 2[[g[[
L

_
RB

(x
0
)
K(x y, t) dy
+
C

t
_
RB

(x
0
)
e

|xy|
2
4t
dy
+
C

t
_
RB

(x
0
)
e

|yx
0
|
2
16t
dy
= +
C

t
_

r
2
16t
r dr + 0 as t 0
+
.
Hence, if [x x
0
[ <

2
and t > 0 is small enough, [u(x, t) g(x
0
)[ < 2.
67
Evans, p. 47, Theorem 1 (c).
Partial Dierential Equations Igor Yanovsky, 2005 270
Problem (S93, #4). The temperature T(x, t) in a stationary medium, x 0, is
governed by the heat conduction equation
T
t
=

2
T
x
2
. (18.6)
Making the change of variable (x, t) (u, t), where u = x/2

t, show that
4t
T
t
=

2
T
u
2
+ 2u
T
u
. (18.7)
Solutions of (18.7) that depend on u alone are called similarity solutions.
68
Proof. We change notation: the change of variables is (x, t) (u, ), where t = .
After the change of variables, we have T = T(u(x, t), (t)).
u =
x
2

t
u
t
=
x
4t
3
2
, u
x
=
1
2

t
, u
xx
= 0,
= t
t
= 1,
x
= 0.
T
t
=
T
u
u
t
+
T

,
T
x
=
T
u
u
x
,

2
T
x
2
=

x
_
T
x
_
=

x
_
T
u
u
x
_
=
_

2
T
u
2
u
x
_
u
x
+
T
u

2
u
x
2
..
=0
=

2
T
u
2
_
u
x
_
2
.
Thus, (18.6) gives:
T
u
u
t
+
T

=

2
T
u
2
_
u
x
_
2
,
T
u
_

x
4t
3
2
_
+
T

=

2
T
u
2
_
1
2

t
_
2
,
T

=
1
4t

2
T
u
2
+
x
4t
3
2
T
u
,
4t
T

=

2
T
u
2
+
x

t
T
u
,
4t
T

=

2
T
u
2
+ 2u
T
u
.
68
This is only the part of the qual problem.
Partial Dierential Equations Igor Yanovsky, 2005 271
19 Contraction Mapping and Uniqueness - Wave
Recall that the solution to
_
u
tt
c
2
u
xx
= f(x, t),
u(x, 0) = g(x), u
t
(x, 0) = h(x),
(19.1)
is given by adding together dAlemberts formula and Duhamels principle:
u(x, t) =
1
2
(g(x + ct) +g(x ct)) +
1
2c
_
x+ct
xct
h() d +
1
2c
_
t
0
__
x+c(ts)
xc(ts)
f(, s) d
_
ds.
Problem (W02, #8). a) Find an explicit solution of the following Cauchy problem
_

2
u
t
2

2
u
x
2
= f(t, x),
u(0, x) = 0,
u
x
(0, x) = 0.
(19.2)
b) Use part (a) to prove the uniqueness of the solution of the Cauchy problem
_

2
u
t
2

2
u
x
2
+q(t, x)u = 0,
u(0, x) = 0,
u
x
(0, x) = 0.
(19.3)
Here f(t, x) and q(t, x) are continuous functions.
Proof. a) It was probably meant to give the u
t
initially. We rewrite (19.2) as
_
u
tt
u
xx
= f(x, t),
u(x, 0) = 0, u
t
(x, 0) = 0.
(19.4)
Duhamels principle, with c = 1, gives the solution to (19.4):
u(x, t) =
1
2c
_
t
0
__
x+c(ts)
xc(ts)
f(, s) d
_
ds =
1
2
_
t
0
__
x+(ts)
x(ts)
f(, s) d
_
ds.
b) We use the Contraction Mapping Principle to prove uniqueness.
Dene the operator
T(u) =
1
2
_
t
0
_
x+(ts)
x(ts)
q(, s) u(, s) d ds.
on the Banach space C
2,2
, [[ [[

.
We will show [Tu
n
Tu
n+1
[ < [[u
n
u
n+1
[[ where < 1. Then u
n

n=1
:
u
n+1
= T(u
n
) converges to a unique xed point which is the unique solution of PDE.
[Tu
n
Tu
n+1
[ =

1
2
_
t
0
_
x+(ts)
x(ts)
q(, s)
_
u
n
(, s) u
n+1
(, s)
_
d ds

1
2
_
t
0
[[q[[

[[u
n
u
n+1
[[

2(t s) ds
t
2
[[q[[

[[u
n
u
n+1
[[

[[u
n
u
n+1
[[

, for small t.
Thus, T is a contraction a unique xed point.
Since Tu = u, u is the solution to the PDE.
Partial Dierential Equations Igor Yanovsky, 2005 272
Problem (F00, #3). Consider the Goursat problem:
Find the solution of the equation

2
u
t
2

2
u
x
2
+ a(x, t)u = 0
in the square D, satisfying the boundary conditions
u[

1
= , u[

2
= ,
where
1
,
2
are two adjacent sides D. Here a(x, t), and are continuous functions.
Prove the uniqueness of the solution of this Goursat problem.
Proof. The change of variable = x + t, = x t
transforms the equation to
u

+ a(, ) u = 0.
We integrate the equation:
_

0
_

0
u

(u, v) du dv =
_

0
_

0
a(, ) udu dv,
_

0
_
u

(, v) u

(0, v)
_
dv =
_

0
_

0
a(, ) udu dv,
u(, ) = u(, 0) + u(0, ) u(0, 0)
_

0
_

0
a(, ) udu dv.
We change the notation. In the new notation:
f(x, y) = (x, y)
_
x
0
_
y
0
a(u, v)f(u, v) dudv,
f = + Kf,
f = + K( +Kf),

f = +

n=1
K
n
,
f = Kf f = 0,
max
0<x<
[f[ max[a[ max[f[.
For small enough , the operator K is a contraction. Thus, there exists a unique xed
point of K, and f = Kf, where f is the unique solution.
Partial Dierential Equations Igor Yanovsky, 2005 273
20 Contraction Mapping and Uniqueness - Heat
The solution of the initial value problem
_
u
t
= u + f(x, t) for t > 0, x R
n
u(x, 0) = g(x) for x R
n
.
(20.1)
is given by
u(x, t) =
_
R
n

K(x y, t) g(y) dy +
_
t
0
_
R
n

K(x y, t s) f(y, s) dy ds
where

K(x, t) =

1
(4t)
n
2
e

|x|
2
4t
for t > 0,
0 for t 0.
Problem (F00, #2). Consider the Cauchy problem
u
t
u +u
2
(x, t) = f(x, t), x R
N
, 0 < t < T
u(x, 0) = 0.
Prove the uniqueness of the classical bounded solution assuming that T is small
enough.
Proof. Let u
n
be a sequence of approximations to the solution, such that
S(u
n
) = u
n+1
=
..
use Duhamel

s principle
_
t
0
_
R
n
K(x y, t s)
_
f(y, s) u
2
n
(y, s)
_
dy ds.
We will show that S has a xed point
_
[S(u
n
) S(u
n+1
)[ [u
n
u
n+1
[, < 1
_
u
n
converges to a uniques solution for small enough T.
Since u
n
, u
n+1
C
2
(R
n
) C
1
(t) [u
n+1
+u
n
[ M.
[S(u
n
) S(u
n+1
)[
_
t
0
_
R
n

K(x y, t s)

u
2
n+1
u
2
n

dy ds
=
_
t
0
_
R
n

K(x y, t s)

u
n+1
u
n

u
n+1
+u
n

dy ds
M
_
t
0
_
R
n

K(x y, t s)

u
n+1
u
n

dy ds
MM
1
_
t
0

u
n+1
(x, s) u
n
(x, s)

ds
MM
1
T [[u
n+1
u
n
[[

< [[u
n+1
u
n
[[

for small T.
Thus, S is a contraction a unique xed point u C
2
(R
n
) C
1
(t) such that
u = lim
n
u
n
. u is implicitly dened as
u(x, t) =
_
t
0
_
R
n
K(x y, t s)
_
f(y, s) u
2
(y, s)
_
dy ds.
Partial Dierential Equations Igor Yanovsky, 2005 274
Problem (S97, #3). a) Let Q(x) 0 such that
_

x=
Q(x) dx = 1,
and dene Q

=
1

Q(
x

## ). Show that (here denotes convolution)

[[Q

(x) w(x)[[
L
[[w(x)[[
L
.
In particular, let Q
t
(x) denote the heat kernel (at time t), then
[[Q
t
(x) w
1
(x) Q
t
(x) w
2
(x)[[
L
[[w
1
(x) w
2
(x)[[
L
.
b) Consider the parabolic equation u
t
= u
xx
+ u
2
subject to initial conditions
u(x, 0) = f(x). Show that the solution of this equation satises
u(x, t) = Q
t
(x) f(x) +
_
t
0
Q
ts
(x) u
2
(x, s) ds. (20.2)
c) Fix t > 0. Let u
n
(x, t), n = 1, 2, . . . the xed point iterations for the solution of
(20.2)
u
n+1
(x, t) = Q
t
(x) f(x) +
_
t
0
Q
ts
(x) u
2
n
(x, s) ds. (20.3)
Let K
n
(t) = sup
0mn
[[u
m
(x, t)[[
L
. Using (a) and (b) show that
[[u
n+1
(x, t) u
n
(x, t)[[
L
2 sup
0t
K
n
()
_
t
0
[[u
n
(x, s) u
n1
(x, s)[[
L
ds.
Conclude that the xed point iterations in (20.3) converge if t is suciently small.
Proof. a) We have
[[Q

(x) w(x)[[
L
=

(x y)w(y) dy

(x y)w(y)

dy
[[w[[

(x y)

dy = [[w[[

Q
_
x y

_
dy
= [[w[[

Q
_
y

_
dy
_
z =
y

, dz =
dy

_
= [[w[[

Q(z) dz = [[w(x)[[

.
Partial Dierential Equations Igor Yanovsky, 2005 275
Q
t
(x) =
1

4t
e

x
2
4t
, the heat kernel. We have
69
[[Q
t
(x) w
1
(x) Q
t
(x) w
2
(x)[[
L
=

Q
t
(x y)w
1
(y) dy
_

Q
t
(x y)w
2
(y) dy

=
1

4t

(xy)
2
4t
w
1
(y) dy
_

(xy)
2
4t
w
2
(y) dy

4t
_

(xy)
2
4t

w
1
(y) w
2
(y)

dy

w
1
(y) w
2
(y)

4t
_

(xy)
2
4t
dy
z =
x y

4t
, dz =
dy

4t
=

w
1
(y) w
2
(y)

4t
_

e
z
2

4t dz
=

w
1
(y) w
2
(y)

e
z
2
dz
. .

w
1
(y) w
2
(y)

.
69
Note:
_

Qt(x) dx =
1

4t
_

(xy)
2
4t
dy =
1

4t
_

e
z
2
4t dz =
1

e
z
2
dz = 1.
Partial Dierential Equations Igor Yanovsky, 2005 276
b) Consider
_
u
t
= u
xx
+u
2
,
u(x, 0) = f(x).
We will show that the solution of this equation satises
u(x, t) = Q
t
(x) f(x) +
_
t
0
Q
ts
(x) u
2
(x, s) ds.
_
t
0
Q
ts
(x) u
2
(x, s) ds =
_
t
0
_
R
Q
ts
(x y) u
2
(y, s) dy ds
=
_
t
0
_
R
Q
ts
(x y)
_
u
s
(y, s) u
yy
(y, s)
_
dy ds
=
_
t
0
_
R
d
ds
_
Q
ts
(x y)u(y, s)
_

d
ds
_
Q
ts
(x y)
_
u(y, s) Q
ts
(x y)u
yy
(y, s) dy ds
=
_ _
R
Q
0
(x y)u(y, t) dy
_
R
Q
t
(x y)u(y, 0) dy
_

_
t
0
_
R
d
ds
_
Q
ts
(x y)
_
u(y, s) +
d
2
dy
2
Q
ts
(x y)u(y, s)
. .
=0, since Qt satisfies heat equation
dy ds
= u(x, t)
_
R
Q
t
(x y)f(y) dy Note: lim
t0
+
Q(x, t) =
0
(x) = (x).
= u(x, t) Q
t
(x) f(x). lim
t0
+
_
R
Q(x y, t)v(y) dy = v(0).
Note that we used: D

(f g) = (D

f) g = f (D

g).
c) Let
u
n+1
(x, t) = Q
t
(x) f(x) +
_
t
0
Q
ts
(x) u
2
n
(x, s) ds.
[[u
n+1
(x, t) u
n
(x, t)[[
L
=

_
t
0
Q
ts
(x)
_
u
2
n
(x, s) u
2
n1
(x, s)
_
ds

_
t
0

Q
ts
(x)
_
u
2
n
(x, s) u
2
n1
(x, s)
_

ds

..
(a)
_
t
0

u
2
n
(x, s) u
2
n1
(x, s)

ds

_
t
0

u
n
(x, s) u
n1
(x, s)

u
n
(x, s) + u
n1
(x, s)

ds
sup
0t

u
n
(x, s) +u
n1
(x, s)

_
t
0

u
n
(x, s) u
n1
(x, s)

ds
2 sup
0t
K
n
()
_
t
0
[[u
n
(x, s) u
n1
(x, s)[[
L
ds.
Also,
[[u
n+1
(x, t) u
n
(x, t)[[
L
2t sup
0t
K
n
() [[u
n
(x, s) u
n1
(x, s)[[
L
.
Partial Dierential Equations Igor Yanovsky, 2005 277
For t small enough, 2t sup
0t
K
n
() < 1. Thus, T dened as
Tu = Q
t
(x) f(x) +
_
t
0
Q
ts
(x) u
2
(x, s) ds
is a contraction, and has a unique xed point u = Tu.
Partial Dierential Equations Igor Yanovsky, 2005 278
Problem (S99, #3). Consider the system of equations
u
t
= u
xx
+ f(u, v)
v
t
= 2v
xx
+g(u, v)
to be solved for t > 0, < x < , and smooth initial data with compact support:
u(x, 0) = u
0
(x), v(x, 0) = v
0
(x).
If f and g are uniformly Lipschitz continuous, give a proof of existence and unique-
ness of the solution to this problem in the space of bounded continuous functions with
[[u(, t)[[ = sup
x
[u(x, t)[.
Proof. The space of continuous bounded functions forms a complete metric space so
the contraction mapping principle applies.
First, let v(x, t) = w
_
x

2
, t
_
, then
u
t
= u
xx
+ f(u, w)
w
t
= w
xx
+ g(u, w).
These initial value problems have the following solutions (K is the heat kernel):
u(x, t) =
_
R
n

K(x y, t) u
0
(y) dy +
_
t
0
_
R
n

w(x, t) =
_
R
n

K(x y, t) w
0
(y) dy +
_
t
0
_
R
n

## K(x y, t s) g(u, w) dyds.

By the Lipshitz conditions,
[f(u, w)[ M
1
[[u[[,
[g(u, w)[ M
2
[[w[[.
Now we can show the mappings, as dened below, are contractions:
T
1
u =
_
R
n

K(x y, t) u
0
(y) dy +
_
t
0
_
R
n

T
2
w =
_
R
n

K(x y, t) w
0
(y) dy +
_
t
0
_
R
n

## K(x y, t s) g(u, w) dyds.

[T
1
(u
n
) T
1
(u
n+1
)[
_
t
0
_
R
n

K(x y, t s)

f(u
n
, w) f(u
n+1
, w)

dy ds
M
1
_
t
0
_
R
n

K(x y, t s)

u
n
u
n+1

dy ds
M
1
_
t
0
sup
x

u
n
u
n+1

_
R
n

K(x y, t s)dy ds
M
1
_
t
0
sup
x

u
n
u
n+1

ds M
1
t sup
x

u
n
u
n+1

< sup
x

u
n
u
n+1

for small t.
We used the Lipshitz condition and
_
R

K(x y, t s) dy = 1.
Thus, for small t, T
1
is a contraction, and has a unique xed point. Thus, the solution
is dened as u = T
1
u.
Similarly, T
2
is a contraction and has a unique xed point. The solution is dened as
w = T
2
w.
Partial Dierential Equations Igor Yanovsky, 2005 279
21 Problems: Maximum Principle - Laplace and Heat
21.1 Heat Equation - Maximum Principle and Uniqueness
Let us introduce the cylinder U = U
T
= (0, T). We know that harmonic (and
subharmonic) functions achieve their maximum on the boundary of the domain. For
the heat equation, the result is improved in that the maximum is achieved on a certain
part of the boundary, parabolic boundary:
= (x, t) U : x or t = 0.
Let us also denote by C
2;1
(U) functions satisfying u
t
, u
x
i
x
j
C(U).
Weak Maximum Principle. Let u C
2;1
(U) C(U) satisfy u u
t
in U.
Then u achieves its maximum on the parabolic boundary of U:
max
U
u(x, t) = max

## u(x, t). (21.1)

Proof. First, assume u > u
t
in U. For 0 < < T consider
U

= (0, ),

= (x, t) U

: x or t = 0.
If the maximum of u on U

## occurs at x and t = , then u

t
(x, ) 0 and
u(x, ) 0, violating our assumption; similarly, u cannot attain an interior maximum
on U

## . Hence (21.1) holds for U

: max
U
u = max

u. But max

u max

u
and by continuity of u, max
U
u = lim
T
max
U
u. This establishes (21.1).
Second, we consider the general case of u u
t
in U. Let u = v + t for > 0.
Notice that v u on U and v v
t
> 0 in U. Thus we may apply (21.1) to v:
max
U
u = max
U
(v +t) max
U
v + T = max

v +T max

u +T.
Letting 0 establishes (21.1) for u.
Partial Dierential Equations Igor Yanovsky, 2005 280
Problem (S98, #7). Prove that any smooth solution, u(x, y, t) in the unit box
= (x, y) [ 1 x, y 1, of the following equation
u
t
= uu
x
+uu
y
+u, t 0, (x, y)
u(x, y, 0) = f(x, y), (x, y)
satises the weak maximum principle,
max
[0,T]
u(x, y, t) max max
0tT
u(1, 1, t), max
(x,y)
f(x, y).
Proof. Suppose u satises given equation. Let u = v +t for > 0. Then,
v
t
+ = vv
x
+vv
y
+t(v
x
+v
y
) +v.
Suppose v has a maximum at (x
0
, y
0
, t
0
) (0, T). Then
v
x
= v
y
= v
t
= 0 = v v > 0
v has a minimum at (x
0
, y
0
, t
0
Thus, the maximum of v is on the boundary of (0, T).
Suppose v has a maximum at (x
0
, y
0
, T), (x
0
, y
0
) . Then
v
x
= v
y
= 0, v
t
0 v v > 0
v has a minimum at (x
0
, y
0
max
[0,T]
v max max
0tT
v(1, 1, t), max
(x,y)
f(x, y).
Now
max
[0,T]
u = max
[0,T]
(v +t) max
[0,T]
v + T max max
0tT
v(1, 1, t), max
(x,y)
f(x, y) +T
max max
0tT
u(1, 1, t), max
(x,y)
f(x, y) +T.
Letting 0 establishes the result.
Partial Dierential Equations Igor Yanovsky, 2005 281
21.2 Laplace Equation - Maximum Principle
Problem (S91, #6). Suppose that u satises
Lu = au
xx
+ bu
yy
+cu
x
+du
y
eu = 0
with a > 0, b > 0, e > 0, for (x, y) , with a bounded open set in R
2
.
a) Show that u cannot have a positive maximum or a negative minimum in the in-
terior of .
b) Use this to show that the only function u satisfying Lu = 0 in , u = 0 on
and u continuous on is u = 0.
Proof. a) For an interior (local) maximum or minimum at an interior point (x, y), we
have
u
x
= 0, u
y
= 0.
Suppose u has a positive maximum in the interior of . Then
u > 0, u
xx
0, u
yy
0.
With these values, we have
au
xx
..
0
+bu
yy
..
0
+ cu
x
..
=0
+ du
y
..
=0
eu
..
<0
= 0,
which leads to contradiction. Thus, u can not have a positive maximum in .
Suppose u has a negative minimum in the interior of . Then
u < 0, u
xx
0, u
yy
0.
With these values, we have
au
xx
..
0
+bu
yy
..
0
+ cu
x
..
=0
+ du
y
..
=0
eu
..
>0
= 0,
which leads to contradiction. Thus, u can not have a negative minimum in .
b) Since u can not have positive maximum in the interior of , then maxu = 0 on .
Since u can not have negative minimum in the interior of , then min u = 0 on .
Since u is continuous, u 0 on .
Partial Dierential Equations Igor Yanovsky, 2005 282
22 Problems: Separation of Variables - Laplace Equation
Problem 1: The 2D LAPLACE Equation on a Square.
Let = (0, ) (0, ), and use separation of variables to solve the boundary value
problem

u
xx
+ u
yy
= 0 0 < x, y <
u(0, y) = 0 = u(, y) 0 y
u(x, 0) = 0, u(x, ) = g(x) 0 x ,
where g is a continuous function satisfying g(0) = 0 = g().
Proof. Assume u(x, y) = X(x)Y (y), then substitution in the PDE gives X

Y +XY

=
0.
X

X
=
Y

Y
= .
From X

+ X = 0, we get X
n
(x) = a
n
cos nx + b
n
sin nx. Boundary conditions
give
_
u(0, y) = X(0)Y (y) = 0
u(, y) = X()Y (y) = 0
X(0) = 0 = X().
Thus, X
n
(0) = a
n
= 0, and
X
n
(x) = b
n
sin nx, n = 1, 2, . . . .
n
2
b
n
sinnx + b
n
sinnx = 0,

n
= n
2
, n = 1, 2, . . . .
With these values of
n
we solve Y

n
2
Y = 0 to nd Y
n
(y) = c
n
cosh ny +
d
n
sinhny.
Boundary conditions give
u(x, 0) = X(x)Y (0) = 0 Y (0) = 0 = c
n
.
Y
n
(x) = d
n
sinh ny.
By superposition, we write
u(x, y) =

n=1
a
n
sin nx sinhny,
which saties the equation and the three homogeneous boundary conditions. The
boundary condition at y = gives
u(x, ) = g(x) =

n=1
a
n
sinnx sinh n,
_

0
g(x) sinmx dx =

n=1
a
n
sinhn
_

0
sin nx sinmx dx =

2
a
m
sinhm.
Partial Dierential Equations Igor Yanovsky, 2005 283
a
n
sinh n =
2

_

0
g(x) sinnx dx.
Partial Dierential Equations Igor Yanovsky, 2005 284
Problem 2: The 2D LAPLACE Equation on a Square. Let = (0, ) (0, ),
and use separation of variables to solve the mixed boundary value problem

u = 0 in
u
x
(0, y) = 0 = u
x
(, y) 0 < y <
u(x, 0) = 0, u(x, ) = g(x) 0 < x < .
Proof. Assume u(x, y) = X(x)Y (y), then substitution in the PDE gives X

Y +XY

=
0.
X

X
=
Y

Y
= .
Consider X

+ X = 0.
If = 0, X
0
(x) = a
0
x + b
0
.
If > 0, X
n
(x) = a
n
cos nx + b
n
sin nx.
Boundary conditions give
_
u
x
(0, y) = X

(0)Y (y) = 0
u
x
(, y) = X

()Y (y) = 0
X

(0) = 0 = X

().
Thus, X

0
(0) = a
0
= 0, and X

n
(0) = nb
n
= 0.
X
0
(x) = b
0
, X
n
(x) = a
n
cos nx, n = 1, 2, . . . .
n
2
a
n
cos nx +a
n
cos nx = 0,

n
= n
2
, n = 0, 1, 2, . . . .
With these values of
n
we solve Y

n
2
Y = 0.
If n = 0, Y
0
(y) = c
0
y + d
0
.
If n ,= 0, Y
n
(y) = c
n
coshny +d
n
sinhny.
Boundary conditions give
u(x, 0) = X(x)Y (0) = 0 Y (0) = 0.
Thus, Y
0
(0) = d
0
= 0, and Y
n
(0) = c
n
= 0.
Y
0
(y) = c
0
y, Y
n
(y) = d
n
sinh ny, n = 1, 2, . . . .
We have
u
0
(x, y) = X
0
(x)Y
0
(y) = b
0
c
0
y = a
0
y,
u
n
(x, y) = X
n
(x)Y
n
(y) = (a
n
cos nx)(d
n
sinh ny) = a
n
cos nx sinhny.
By superposition, we write
u(x, y) = a
0
y +

n=1
a
n
cos nx sinhny,
which saties the equation and the three homogeneous boundary conditions. The fourth
boundary condition gives
u(x, ) = g(x) = a
0
+

n=1
a
n
cos nx sinh n,
Partial Dierential Equations Igor Yanovsky, 2005 285
_
_

0
g(x) dx =
_

0
_
a
0
+

n=1
a
n
cos nx sinh n
_
dx = a
0

2
,
_

0
g(x) cos mx dx =

n=1
a
n
sinhn
_

0
cos nx cos mx dx =

2
a
m
sinh m.
a
0
=
1

2
_

0
g(x) dx,
a
n
sinh n =
2

_

0
g(x) cosnx dx.
Partial Dierential Equations Igor Yanovsky, 2005 286
Problem (W04, #5) The 2D LAPLACE Equation in an Upper-Half Plane.
Consider the Laplace equation

2
u
x
2
+

2
u
y
2
= 0, y > 0, < x < +
u(x, 0)
y
u(x, 0) = f(x),
where f(x) C

0
(R
1
).
Find a bounded solution u(x, y) and show that u(x, y) 0 when [x[ + y .
Proof. Assume u(x, y) = X(x)Y (y), then substitution in the PDE gives X

Y +XY

=
0.
X

X
=
Y

Y
= .
Consider X

+ X = 0.
If = 0, X
0
(x) = a
0
x + b
0
.
If > 0, X
n
(x) = a
n
cos

n
x +b
n
sin

n
x.
Since we look for bounded solutions as [x[ , we have a
0
= 0.
Consider Y

n
Y = 0.
If
n
= 0, Y
0
(y) = c
0
y + d
0
.
If
n
> 0, Y
n
(y) = c
n
e

ny
+ d
n
e

ny
.
Since we look for bounded solutions as y , we have c
0
= 0, d
n
= 0. Thus,
u(x, y) = a
0
+

n=1
e

ny
_
a
n
cos
_

n
x +

b
n
sin
_

n
x
_
.
Initial condition gives:
f(x) = u
y
(x, 0) u(x, 0) = a
0

n=1
(
_

n
+ 1)
_
a
n
cos
_

n
x +

b
n
sin
_

n
x
_
.
f(x) C

0
(R
1
), i.e. has compact support [L, L], for some L > 0. Thus the coecients
a
n
,

b
n
are given by
_
L
L
f(x) cos
_

n
x dx = (
_

n
+ 1) a
n
L.
_
L
L
f(x) sin
_

n
x dx = (
_

n
+ 1)

b
n
L.
Thus, u(x, y) 0 when [x[ +y .
70
70
Note that if we change the roles of X and Y in , the solution we get will be unbounded.
Partial Dierential Equations Igor Yanovsky, 2005 287
Problem 3: The 2D LAPLACE Equation on a Circle.
Let be the unit disk in R
2
and consider the problem
_
u = 0 in
u
n
= h on ,
where h is a continuous function.
Proof. Use polar coordinates (r, )
_
u
rr
+
1
r
u
r
+
1
r
2
u

## = 0 for 0 r < 1, 0 < 2

u
r
(1, ) = h() for 0 < 2.
r
2
u
rr
+ ru
r
+u

= 0.
Let r = e
t
, u(r(t), ).
u
t
= u
r
r
t
= e
t
u
r
,
u
tt
= (e
t
u
r
)
t
= e
t
u
r
+e
2t
u
rr
= ru
r
+r
2
u
rr
.
Thus, we have
u
tt
+ u

= 0.
Let u(t, ) = X(t)Y (), which gives X

(t)Y () + X(t)Y

() = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
From Y

() + Y () = 0, we get Y
n
() = a
n
cos n + b
n
sin n.

n
= n
2
, n = 0, 1, 2, . . ..
With these values of
n
we solve X

(t) n
2
X(t) = 0.
If n = 0, X
0
(t) = c
0
t + d
0
. X
0
(r) = c
0
log r +d
0
.
If n ,= 0, X
n
(t) = c
n
e
nt
+d
n
e
nt
X
n
(r) = c
n
r
n
+ d
n
r
n
.
We have
u
0
(r, ) = X
0
(r)Y
0
() = (c
0
log r + d
0
)a
0
,
u
n
(r, ) = X
n
(r)Y
n
() = (c
n
r
n
+ d
n
r
n
)(a
n
cos n +b
n
sinn).
But u must be nite at r = 0, so c
n
= 0, n = 0, 1, 2, . . ..
u
0
(r, ) = d
0
a
0
,
u
n
(r, ) = d
n
r
n
(a
n
cos n +b
n
sinn).
By superposition, we write
u(r, ) = a
0
+

n=1
r
n
( a
n
cos n +

b
n
sinn).
Boundary condition gives
u
r
(1, ) =

n=1
n( a
n
cos n +

b
n
sinn) = h().
The coecients a
n
, b
n
for n 1 are determined from the Fourier series for h().
a
0
is not determined by h() and therefore may take an arbitrary value. Moreover,
Partial Dierential Equations Igor Yanovsky, 2005 288
the constant term in the Fourier series for h() must be zero [i.e.,
_
2
0
h()d = 0].
Therefore, the problem is not solvable for an arbitrary function h(), and when it is
solvable, the solution is not unique.
Partial Dierential Equations Igor Yanovsky, 2005 289
Problem 4: The 2D LAPLACE Equation on a Circle.
Let = (x, y) R
2
: x
2
+y
2
< 1 = (r, ) : 0 r < 1, 0 < 2,
and use separation of variables (r, ) to solve the Dirichlet problem
_
u = 0 in
u(1, ) = g() for 0 < 2.
Proof. Use polar coordinates (r, )
_
u
rr
+
1
r
u
r
+
1
r
2
u

## = 0 for 0 r < 1, 0 < 2

u(1, ) = g() for 0 < 2.
r
2
u
rr
+ ru
r
+u

= 0.
Let r = e
t
, u(r(t), ).
u
t
= u
r
r
t
= e
t
u
r
,
u
tt
= (e
t
u
r
)
t
= e
t
u
r
+e
2t
u
rr
= ru
r
+r
2
u
rr
.
Thus, we have
u
tt
+ u

= 0.
Let u(t, ) = X(t)Y (), which gives X

(t)Y () + X(t)Y

() = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
From Y

() + Y () = 0, we get Y
n
() = a
n
cos n + b
n
sin n.

n
= n
2
, n = 0, 1, 2, . . ..
With these values of
n
we solve X

(t) n
2
X(t) = 0.
If n = 0, X
0
(t) = c
0
t + d
0
. X
0
(r) = c
0
log r +d
0
.
If n ,= 0, X
n
(t) = c
n
e
nt
+d
n
e
nt
X
n
(r) = c
n
r
n
+ d
n
r
n
.
We have
u
0
(r, ) = X
0
(r)Y
0
() = (c
0
log r + d
0
)a
0
,
u
n
(r, ) = X
n
(r)Y
n
() = (c
n
r
n
+ d
n
r
n
)(a
n
cos n +b
n
sinn).
But u must be nite at r = 0, so c
n
= 0, n = 0, 1, 2, . . ..
u
0
(r, ) = d
0
a
0
,
u
n
(r, ) = d
n
r
n
(a
n
cos n +b
n
sinn).
By superposition, we write
u(r, ) = a
0
+

n=1
r
n
( a
n
cos n +

b
n
sinn).
Boundary condition gives
u(1, ) = a
0
+

n=1
( a
n
cos n +

b
n
sin n) = g().
Partial Dierential Equations Igor Yanovsky, 2005 290
a
0
=
1

_

0
g() d,
a
n
=
2

_

0
g() cos n d,

b
n
=
2

_

0
g() sinn d.
Partial Dierential Equations Igor Yanovsky, 2005 291
Problem (F94, #6): The 2D LAPLACE Equation on a Circle.
Find all solutions of the homogeneous equation
u
xx
+u
yy
= 0, x
2
+y
2
< 1,
u
n
u = 0, x
2
+y
2
= 1.
Hint: =
1
r

r
(r

r
) +
1
r
2

2
in polar coordinates.
Proof. Use polar coordinates (r, ):
_
u
rr
+
1
r
u
r
+
1
r
2
u

## = 0 for 0 r < 1, 0 < 2

u
r
(1, ) u(1, ) = 0 for 0 < 2.
Since we solve the equation on a circle, we have periodic conditions:
u(r, 0) = u(r, 2) X(r)Y (0) = X(r)Y(2) Y (0) = Y (2),
u

(r, 0) = u

(r, 2) X(r)Y

(0) = X(r)Y

(2) Y

(0) = Y

(2).
Also, we want the solution to be bounded. In particular, u is bounded for r = 0.
r
2
u
rr
+ ru
r
+u

= 0.
Let r = e
t
, u(r(t), ), we have
u
tt
+ u

= 0.
Let u(t, ) = X(t)Y (), which gives X

(t)Y () + X(t)Y

() = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
From Y

() + Y () = 0, we get Y
n
() = a
n
cos

+ b
n
sin

.
Using periodic condition: Y
n
(0) = a
n
,
Y
n
(2) = a
n
cos(
_

n
2) +b
n
sin(
_

n
2) = a
n

_

n
= n
n
= n
2
.
Thus, Y
n
() = a
n
cos n +b
n
sinn.
With these values of
n
we solve X

(t) n
2
X(t) = 0.
If n = 0, X
0
(t) = c
0
t + d
0
. X
0
(r) = c
0
log r +d
0
.
If n ,= 0, X
n
(t) = c
n
e
nt
+d
n
e
nt
X
n
(r) = c
n
r
n
+ d
n
r
n
.
u must be nite at r = 0 c
n
= 0, n = 0, 1, 2, . . ..
u(r, ) = a
0
+

n=1
r
n
( a
n
cos n +

b
n
sinn).
Boundary condition gives
0 = u
r
(1, ) u(1, ) = a
0
+

n=1
(n 1)( a
n
cos n +

b
n
sinn).
Calculating Fourier coecients gives 2a
0
= 0 a
0
= 0.
(n 1)a
n
= 0 a
n
= 0, n = 2, 3, . . ..
a
1
, b
1
are constants. Thus,
u(r, ) = r( a
1
cos +

b
1
sin ).
Partial Dierential Equations Igor Yanovsky, 2005 292
Problem (S00, #4).
a) Let (r, ) be polar coordinates on the plane,
i.e. x
1
+ ix
2
= re
i
. Solve the boudary value problem
u = 0 in r < 1
u/r = f() on r = 1,
beginning with the Fourier series for f (you may assume that f is continuously dif-
1
+ ix
2
plus a power series in
x
1
ix
2
. There is a necessary condition on f for this boundary value problem to be
solvable that you will nd in the course of doing this.
b) Sum the series in part (a) to get a representation of u in the form
u(r, ) =
_
2
0
N(r,

)f(

) d

.
Proof. a) Greens identity gives the necessary compatibility condition on f:
_
2
0
f() d =
_
r=1
u
r
d =
_

u
n
ds =
_

u dx = 0.
Use polar coordinates (r, ):
_
u
rr
+
1
r
u
r
+
1
r
2
u

## = 0 for 0 r < 1, 0 < 2

u
r
(1, ) = f() for 0 < 2.
Since we solve the equation on a circle, we have periodic conditions:
u(r, 0) = u(r, 2) X(r)Y (0) = X(r)Y(2) Y (0) = Y (2),
u

(r, 0) = u

(r, 2) X(r)Y

(0) = X(r)Y

(2) Y

(0) = Y

(2).
Also, we want the solution to be bounded. In particular, u is bounded for r = 0.
r
2
u
rr
+ ru
r
+u

= 0.
Let r = e
t
, u(r(t), ), we have
u
tt
+ u

= 0.
Let u(t, ) = X(t)Y (), which gives X

(t)Y () + X(t)Y

() = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
From Y

() + Y () = 0, we get Y
n
() = a
n
cos

+ b
n
sin

.
Using periodic condition: Y
n
(0) = a
n
,
Y
n
(2) = a
n
cos(
_

n
2) +b
n
sin(
_

n
2) = a
n

_

n
= n
n
= n
2
.
Thus, Y
n
() = a
n
cos n +b
n
sinn.
With these values of
n
we solve X

(t) n
2
X(t) = 0.
If n = 0, X
0
(t) = c
0
t + d
0
. X
0
(r) = c
0
log r +d
0
.
Partial Dierential Equations Igor Yanovsky, 2005 293
If n ,= 0, X
n
(t) = c
n
e
nt
+d
n
e
nt
X
n
(r) = c
n
r
n
+ d
n
r
n
.
u must be nite at r = 0 c
n
= 0, n = 0, 1, 2, . . ..
u(r, ) = a
0
+

n=1
r
n
( a
n
cos n +

b
n
sinn).
Since
u
r
(r, ) =

n=1
nr
n1
( a
n
cos n +

b
n
sinn),
the boundary condition gives
u
r
(1, ) =

n=1
n( a
n
cos n +

b
n
sinn) = f().
a
n
=
1
n
_
2
0
f() cos n d,

b
n
=
1
n
_
2
0
f() sin n d.
a
0
is not determined by f() (since
_
2
0
f() d = 0). Therefore, it may take an
arbitrary value. Moreover, the constant term in the Fourier series for f() must be zero
[i.e.,
_
2
0
f()d = 0]. Therefore, the problem is not solvable for an arbitrary function
f(), and when it is solvable, the solution is not unique.
b) In part (a), we obtained the solution and the Fourier coecients:
a
n
=
1
n
_
2
0
f(

) cos n

b
n
=
1
n
_
2
0
f(

) sinn

.
u(r, ) = a
0
+

n=1
r
n
( a
n
cos n +

b
n
sinn)
= a
0
+

n=1
r
n
_
_
1
n
_
2
0
f(

) cos n

_
cos n +
_
1
n
_
2
0
f(

) sin n

_
sinn
_
= a
0
+

n=1
r
n
n
_
2
0
f(

)
_
cos n

cos n + sin n

sinn

= a
0
+

n=1
r
n
n
_
2
0
f(

) cos n(

) d

= a
0
+
_
2
0

n=1
r
n
n
cos n(

)
. .
N(r,

)
f(

) d

.
Partial Dierential Equations Igor Yanovsky, 2005 294
Problem (S92, #6). Consider the Laplace equation
u
xx
+u
yy
= 0
for x
2
+ y
2
1. Denoting by x = r cos , y = r sin polar coordinates, let f = f() be
a given smooth function of . Construct a uniformly bounded solution which satises
boundary conditions
u = f for x
2
+ y
2
= 1.
What conditions has f to satisfy such that
lim
x
2
+y
2

(x
2
+ y
2
)u(x, y) = 0?
Proof. Use polar coordinates (r, ):
_
u
rr
+
1
r
u
r
+
1
r
2
u

= 0 for r 1
u(1, ) = f() for 0 < 2.
Since we solve the equation on outside of a circle, we have periodic conditions:
u(r, 0) = u(r, 2) X(r)Y(0) = X(r)Y (2) Y (0) = Y (2),
u

## (r, 0) = u(r, 2) X(r)Y

(0) = X(r)Y

(2) Y

(0) = Y

(2).
Also, we want the solution to be bounded. In particular, u is bounded for r = .
r
2
u
rr
+ ru
r
+u

= 0.
Let r = e
t
, u(r(t), ), we have
u
tt
+ u

= 0.
Let u(t, ) = X(t)Y (), which gives X

(t)Y () + X(t)Y

() = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
From Y

() + Y () = 0, we get Y
n
() = a
n
cos

+ b
n
sin

.
Using periodic condition: Y
n
(0) = a
n
,
Y
n
(2) = a
n
cos(
_

n
2) +b
n
sin(
_

n
2) = a
n

_

n
= n
n
= n
2
.
Thus, Y
n
() = a
n
cos n +b
n
sinn.
With these values of
n
we solve X

(t) n
2
X(t) = 0.
If n = 0, X
0
(t) = c
0
t + d
0
. X
0
(r) = c
0
log r +d
0
.
If n ,= 0, X
n
(t) = c
n
e
nt
+d
n
e
nt
X
n
(r) = c
n
r
n
+ d
n
r
n
.
u must be nite at r = c
0
= 0, d
n
= 0, n = 1, 2, . . ..
u(r, ) = a
0
+

n=1
r
n
( a
n
cos n +

b
n
sin n).
Boundary condition gives
f() = u(1, ) = a
0
+

n=1
( a
n
cos n +

b
n
sinn).
Partial Dierential Equations Igor Yanovsky, 2005 295

2a
0
=
_
2
0
f() d,
a
n
=
_
2
0
f() cos n d,

b
n
=
_
2
0
f() sinn d.

f
0
= a
0
=
1
2
_
2
0
f() d,
f
n
= a
n
=
1

_
2
0
f() cos n d,

f
n
=

b
n
=
1

_
2
0
f() sinn d.
We need to nd conditions for f such that
lim
x
2
+y
2

(x
2
+ y
2
)u(x, y) = 0, or
lim
r
r
2
u(r, ) =
..
need
0,
lim
r
r
2
_
f
0
+

n=1
r
n
(f
n
cos n +

f
n
sin n)
_
=
..
need
0.
Since
lim
r
_

n>2
r
2n
(f
n
cos n +

f
n
sin n)
_
= 0,
we need
lim
r
_
r
2
f
0
+
2

n=1
r
2n
(f
n
cos n +

f
n
sinn)
_
=
..
need
0.
Thus, the conditions are
f
n
,

f
n
= 0, n = 0, 1, 2.
Partial Dierential Equations Igor Yanovsky, 2005 296
Problem (F96, #2): The 2D LAPLACE Equation on a Semi-Annulus.
Solve the Laplace equation in the semi-annulus

## u = 0, 1 < r < 2, 0 < < ,

u(r, 0) = u(r, ) = 0, 1 < r < 2,
u(1, ) = sin, 0 < < ,
u(2, ) = 0, 0 < < .
Hint: Use the formula =
1
r

r
(r

r
) +
1
r
2

2
for the Laplacian in polar coordinates.
Proof. Use polar coordinates (r, )
u
rr
+
1
r
u
r
+
1
r
2
u

## = 0 1 < r < 2, 0 < < ,

r
2
u
rr
+ru
r
+u

= 0.
With r = e
t
, we have
u
tt
+ u

= 0.
Let u(t, ) = X(t)Y (), which gives X

(t)Y () + X(t)Y

() = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
From Y

() + Y () = 0, we get Y
n
() = a
n
cos

+ b
n
sin

.
Boundary conditions give
u
n
(r, 0) = 0 = X
n
(r)Y
n
(0) = 0, Y
n
(0) = 0,
u
n
(r, ) = 0 = X
n
(r)Y
n
() = 0, Y
n
() = 0.
Thus, 0 = Y
n
(0) = a
n
, and Y
n
() = b
n
sin

= 0

= n
n
= n
2
.
Thus, Y
n
() = b
n
sinn, n = 1, 2, . . ..
With these values of
n
we solve X

(t) n
2
X(t) = 0.
If n = 0, X
0
(t) = c
0
t + d
0
. X
0
(r) = c
0
log r +d
0
.
If n > 0, X
n
(t) = c
n
e
nt
+d
n
e
nt
X
n
(r) = c
n
r
n
+ d
n
r
n
.
We have,
u(r, ) =

n=1
X
n
(r)Y
n
() =

n=1
( c
n
r
n
+

d
n
r
n
) sinn.
Using the other two boundary conditions, we obtain
sin = u(1, ) =

n=1
( c
n
+

d
n
) sinn
_
c
1
+

d
1
= 1,
c
n
+

d
n
= 0, n = 2, 3, . . . .
0 = u(2, ) =

n=1
( c
n
2
n
+

d
n
2
n
) sinn c
n
2
n
+

d
n
2
n
= 0, n = 1, 2, . . . .
Thus, the coecients are given by
c
1
=
4
3
, d
1
=
1
3
;
c
n
= 0, d
n
= 0.
Partial Dierential Equations Igor Yanovsky, 2005 297
u(r, ) =
_
4
3r

r
3
_
sin.
Partial Dierential Equations Igor Yanovsky, 2005 298
Problem (S98, #8): The 2D LAPLACE Equation on a Semi-Annulus.
Solve

## u = 0, 1 < r < 2, 0 < < ,

u(r, 0) = u(r, ) = 0, 1 < r < 2,
u(1, ) = u(2, ) = 1, 0 < < .
Proof. Use polar coordinates (r, )
u
rr
+
1
r
u
r
+
1
r
2
u

## = 0 for 1 < r < 2, 0 < < ,

r
2
u
rr
+ru
r
+u

= 0.
With r = e
t
, we have
u
tt
+ u

= 0.
Let u(t, ) = X(t)Y (), which gives X

(t)Y () + X(t)Y

() = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
From Y

() + Y () = 0, we get Y
n
() = a
n
cos n + b
n
sin n.
Boundary conditions give
u
n
(r, 0) = 0 = X
n
(r)Y
n
(0) = 0, Y
n
(0) = 0,
u
n
(r, ) = 0 = X
n
(r)Y
n
() = 0, Y
n
() = 0.
Thus, 0 = Y
n
(0) = a
n
, and Y
n
() = b
n
sin n.

n
= n
2
, n = 1, 2, . . ..
With these values of
n
we solve X

(t) n
2
X(t) = 0.
If n = 0, X
0
(t) = c
0
t + d
0
. X
0
(r) = c
0
log r +d
0
.
If n > 0, X
n
(t) = c
n
e
nt
+d
n
e
nt
X
n
(r) = c
n
r
n
+ d
n
r
n
.
We have,
u(r, ) =

n=1
X
n
(r)Y
n
() =

n=1
( c
n
r
n
+

d
n
r
n
) sinn.
Using the other two boundary conditions, we obtain
u(1, ) = 1 =

n=1
( c
n
+

d
n
) sinn,
u(2, ) = 1 =

n=1
( c
n
2
n
+

d
n
2
n
) sinn,
which give the two equations for c
n
and

d
n
:
_

0
sin n d =

2
( c
n
+

d
n
),
_

0
sin n d =

2
( c
n
2
n
+

d
n
2
n
),
that can be solved.
Partial Dierential Equations Igor Yanovsky, 2005 299
Problem (F89, #1). Consider Laplace equation inside a 90

sector of a circular
annulus
u = 0 a < r < b, 0 < <

2
subject to the boundary conditions
u

(r, 0) = 0,
u

(r,

2
) = 0,
u
r
(a, ) = f
1
(),
u
r
(b, ) = f
2
(),
where f
1
(), f
2
() are continuously dierentiable.
a) Find the solution of this equation with the prescribed
boundary conditions using separation of variables.
Proof. a) Use polar coordinates (r, )
u
rr
+
1
r
u
r
+
1
r
2
u

## = 0 for a < r < b, 0 < <

2
,
r
2
u
rr
+ru
r
+u

= 0.
With r = e
t
, we have
u
tt
+ u

= 0.
Let u(t, ) = X(t)Y (), which gives X

(t)Y () + X(t)Y

() = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
From Y

() + Y () = 0, we get Y
n
() = a
n
cos

+ b
n
sin

.
Boundary conditions give
u
n
(r, 0) = X
n
(r)Y

n
(0) = 0 Y

n
(0) = 0,
u
n
(r,

2
) = X
n
(r)Y

n
(

2
) = 0 Y

n
(

2
) = 0.
Y

n
() = a
n

n
sin

n
+b
n

n
cos

n
. Thus, Y

n
(0) = b
n

n
= 0 b
n
= 0.
Y

n
(

2
) = a
n

n
sin

2
= 0

2
= n
n
= (2n)
2
.
Thus, Y
n
() = a
n
cos(2n), n = 0, 1, 2, . . ..
In particular, Y
0
() = a
0
t + b
0
. Boundary conditions give Y
0
() = b
0
.
With these values of
n
we solve X

(t) (2n)
2
X(t) = 0.
If n = 0, X
0
(t) = c
0
t + d
0
. X
0
(r) = c
0
log r +d
0
.
If n > 0, X
n
(t) = c
n
e
2nt
+ d
n
e
2nt
X
n
(r) = c
n
r
2n
+ d
n
r
2n
.
u(r, ) = c
0
log r +

d
0
+

n=1
( c
n
r
2n
+

d
n
r
2n
) cos(2n).
Using the other two boundary conditions, we obtain
u
r
(r, ) =
c
0
r
+

n=1
(2n c
n
r
2n1
+ 2n

d
n
r
2n1
) cos(2n).
Partial Dierential Equations Igor Yanovsky, 2005 300
f
1
() = u
r
(a, ) =
c
0
a
+ 2

n=1
n( c
n
a
2n1
+

d
n
a
2n1
) cos(2n),
f
2
() = u
r
(b, ) =
c
0
b
+ 2

n=1
n( c
n
b
2n1
+

d
n
b
2n1
) cos(2n).
which give the two equations for c
n
and

d
n
:
_
2
0
f
1
() cos(2n) d =

2
n( c
n
a
2n1
+

d
n
a
2n1
),
_
2
0
f
2
() sin(2n) d =

2
n( c
n
b
2n1
+

d
n
b
2n1
).
b) Show that the solution exists if and only if
a
_
2
0
f
1
() d b
_
2
0
f
2
() d = 0.
Proof. Using Greens identity, we obtain:
0 =
_

u dx =
_

u
n
=
_
2
0
u
r
(b, ) d +
_
0

u
r
(a, ) d +
_
b
a

(r, 0) dr +
_
a
b
u

_
r,

2
_
dr
=
_
2
0
f
2
() d +
_
2
0
f
1
() d + 0 + 0
=
_
2
0
f
1
() d +
_
2
0
f
2
() d.
c) Is the solution unique?
Proof. No, since the boundary conditions are Neumann. The solution is unique only
up to a constant.
Partial Dierential Equations Igor Yanovsky, 2005 301
Problem (S99, #4). Let u(x, y) be harmonic inside the unit disc,
with boundary values along the unit circle
u(x, y) =
_
1, y > 0
0, y 0.
Compute u(0, 0) and u(0, y).
Proof. Since u is harmonic, u = 0. Use polar coordinates (r, )

u
rr
+
1
r
u
r
+
1
r
2
u

= 0 0 r < 1, 0 < 2
u(1, ) =
_
1, 0 < <
0, 2.
r
2
u
rr
+ ru
r
+u

= 0.
With r = e
t
, we have
u
tt
+ u

= 0.
Let u(t, ) = X(t)Y (), which gives X

(t)Y () + X(t)Y

() = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
From Y

() + Y () = 0, we get Y
n
() = a
n
cos n + b
n
sin n.

n
= n
2
, n = 1, 2, . . ..
With these values of
n
we solve X

(t) n
2
X(t) = 0.
If n = 0, X
0
(t) = c
0
t + d
0
. X
0
(r) = c
0
log r +d
0
.
If n > 0, X
n
(t) = c
n
e
nt
+d
n
e
nt
X
n
(r) = c
n
r
n
+ d
n
r
n
.
We have
u
0
(r, ) = X
0
(r)Y
0
() = (c
0
log r + d
0
)a
0
,
u
n
(r, ) = X
n
(r)Y
n
() = (c
n
r
n
+ d
n
r
n
)(a
n
cos n +b
n
sinn).
But u must be nite at r = 0, so c
n
= 0, n = 0, 1, 2, . . ..
u
0
(r, ) = a
0
,
u
n
(r, ) = r
n
( a
n
cos n +

b
n
sin n).
By superposition, we write
u(r, ) = a
0
+

n=1
r
n
( a
n
cos n +

b
n
sinn).
Boundary condition gives
u(1, ) = a
0
+

n=1
( a
n
cos n +

b
n
sin n) =
_
1, 0 < <
0, 2,
and the coecients a
n
and

b
n
are determined from the above equation.
71
71
See Yanas solutions, where Greens function on a unit disk is constructed.
Partial Dierential Equations Igor Yanovsky, 2005 302
23 Problems: Separation of Variables - Poisson Equation
Problem (F91, #2): The 2D POISSON Equation on a Quarter-Circle.
Solve explicitly the following boundary value problem
u
xx
+u
yy
= f(x, y)
in the domain = (x, y), x > 0, y > 0, x
2
+ y
2
< 1
with boundary conditions
u = 0 for y = 0, 0 < x < 1,
u
x
= 0 for x = 0, 0 < y < 1,
u = 0 for x > 0, y > 0, x
2
+y
2
= 1.
Function f(x, y) is known and is assumed to be continuous.
Proof. Use polar coordinates (r, ):

u
rr
+
1
r
u
r
+
1
r
2
u

## = f(r, ) 0 r < 1, 0 <

2
u(r, 0) = 0 0 r < 1,
u

(r,

2
) = 0 0 r < 1,
u(1, ) = 0 0

2
.
We solve
r
2
u
rr
+ ru
r
+u

= 0.
Let r = e
t
, u(r(t), ), we have
u
tt
+ u

= 0.
Let u(t, ) = X(t)Y (), which gives X

(t)Y () + X(t)Y

() = 0.
X

(t)
X(t)
=
Y

()
Y ()
= .
From Y

() + Y () = 0, we get Y
n
() = a
n
cos

+ b
n
sin

. Boundary
conditions:
_
u(r, 0) = X(r)Y(0) = 0
u

(r,

2
) = X(r)Y

2
) = 0
Y (0) = Y

_

2
_
= 0.
Thus, Y
n
(0) = a
n
= 0, and Y

n
(

2
) =

n
b
n
cos

2
= 0

2
= n

2
, n = 1, 2, . . .
n
= (2n 1)
2
.
Thus, Y
n
() = b
n
sin(2n 1), n = 1, 2, . . .. Thus, we have
u(r, ) =

n=1
X
n
(r) sin[(2n 1)].
Partial Dierential Equations Igor Yanovsky, 2005 303
We now plug this equation into with inhomogeneous term and obtain

n=1
_
X

n
(t) sin[(2n 1)] (2n 1)
2
X
n
(t) sin[(2n 1)]
_
= f(t, ),

n=1
_
X

n
(t) (2n 1)
2
X
n
(t)
_
sin[(2n 1)] = f(t, ),

4
_
X

n
(t) (2n 1)
2
X
n
(t)
_
=
_
2
0
f(t, ) sin[(2n 1)] d,
X

n
(t) (2n 1)
2
X
n
(t) =
4

_
2
0
f(t, ) sin[(2n 1)] d.
The solution to this equation is
X
n
(t) = c
n
e
(2n1)t
+ d
n
e
(2n1)t
+U
np
(t), or
X
n
(r) = c
n
r
(2n1)
+ d
n
r
(2n1)
+u
np
(r),
where u
np
is the particular solution of inhomogeneous equation.
u must be nite at r = 0 c
n
= 0, n = 1, 2, . . .. Thus,
u(r, ) =

n=1
_
d
n
r
(2n1)
+u
np
(r)
_
sin[(2n 1)].
Using the last boundary condition, we have
0 = u(1, ) =

n=1
_
d
n
+u
np
(1)
_
sin[(2n 1)],
0 =

4
(d
n
+ u
np
(1)),
d
n
= u
np
(1).
u(r, ) =

n=1
_
u
np
(1)r
(2n1)
+ u
np
(r)
_
sin[(2n 1)].
The method used to solve this problem is similar to section
Problems: Eigenvalues of the Laplacian - Poisson Equation:
1) First, we nd Y
n
() eigenfunctions.
2) Then, we plug in our guess u(t, ) = X(t)Y () into the equation u
tt
+ u

= f(t, )
and solve an ODE in X(t).
Note the similar problem on 2D Poisson equation on a square domain. The prob-
lem is used by rst nding the eigenvalues and eigenfunctions of the Laplacian, and
then expanding f(x, y) in eigenfunctions, and comparing coecients of f with the gen-
eral solution u(x, y).
Here, however, this could not be done because of the circular geometry of the domain.
In particular, the boundary conditions do not give enough information to nd explicit
representations for
m
and
n
. Also, the condition u = 0 for x > 0, y > 0, x
2
+y
2
= 1
Partial Dierential Equations Igor Yanovsky, 2005 304
can not be used.
72
72
ChiuYens solutions have attempts to solve this problem using Greens function.
Partial Dierential Equations Igor Yanovsky, 2005 305
24 Problems: Separation of Variables - Wave Equation
Example (McOwen 3.1 #2). We considered the initial/boundary value problem and
solved it using Fourier Series. We now solve it using the Separation of Variables.

u
tt
u
xx
= 0 0 < x < , t > 0
u(x, 0) = 1, u
t
(x, 0) = 0 0 < x <
u(0, t) = 0, u(, t) = 0 t 0.
(24.1)
Proof. Assume u(x, t) = X(x)T(t), then substitution in the PDE gives XT

T = 0.
X

X
=
T

T
= .
From X

+ X = 0, we get X
n
(x) = a
n
cos nx + b
n
sin nx. Boundary conditions
give
_
u(0, t) = X(0)T(t) = 0
u(, t) = X()T(t) = 0
X(0) = X() = 0.
Thus, X
n
(0) = a
n
= 0, and X
n
(x) = b
n
sinnx,
n
= n
2
, n = 1, 2, . . . .
With these values of
n
, we solve T

+n
2
T = 0 to nd T
n
(t) = c
n
sinnt+d
n
cos nt.
Thus,
u(x, t) =

n=1
_
c
n
sin nt +

d
n
cos nt
_
sin nx,
u
t
(x, t) =

n=1
_
n c
n
cos nt n

d
n
sin nt
_
sinnx.
Initial conditions give
1 = u(x, 0) =

n=1

d
n
sin nx,
0 = u
t
(x, 0) =

n=1
n c
n
sinnx.
By orthogonality, we may multiply both equations by sinmx and integrate:
_

0
sin mx dx =

d
m

2
,
_

0
0 dx = n c
n

2
,
which gives the coecients

d
n
=
2
n
(1 cos n) =
_
4
n
, n odd,
0, n even,
and c
n
= 0.
Plugging the coecients into a formula for u(x, t), we get
u(x, t) =
4

n=0
cos(2n + 1)t sin(2n + 1)x
(2n + 1)
.
Partial Dierential Equations Igor Yanovsky, 2005 306
Example. Use the method of separation of variables to nd the solution to:

u
tt
+ 3u
t
+ u = u
xx
, 0 < x < 1
u(0, t) = 0, u(1, t) = 0,
u(x, 0) = 0, u
t
(x, 0) = x sin(2x).
Proof. Assume u(x, t) = X(x)T(t), then substitution in the PDE gives
XT

+ 3XT

+ XT = X

T,
T

T
+ 3
T

T
+ 1 =
X

X
= .
From X

+ X = 0, X
n
(x) = a
n
cos

n
x + b
n
sin

n
x. Boundary conditions
give
_
u(0, t) = X(0)T(t) = 0
u(1, t) = X(1)T(t) = 0
X(0) = X(1) = 0.
Thus, X
n
(0) = a
n
= 0, and X
n
(x) = b
n
sin

n
x.
X
n
(1) = b
n
sin

n
= 0. Hence,

n
= n, or
n
= (n)
2
, n = 1, 2, . . ..

n
= (n)
2
, X
n
(x) = b
n
sinnx.
With these values of
n
, we solve
T

+ 3T

+T =
n
T,
T

+ 3T

+T = (n)
2
T,
T

+ 3T

+ (1 + (n)
2
)T = 0.
We can solve this 2nd-order ODE with the following guess, T(t) = ce
st
to obtain
s =
3
2

_
5
4
(n)
2
. For n 1,
5
4
(n)
2
< 0. Thus, s =
3
2
i
_
(n)
2

5
4
.
T
n
(t) = e

3
2
t
_
c
n
cos
_
(n)
2

5
4
t + d
n
sin
_
(n)
2

5
4
t
_
.
u(x, t) = X(x)T(t) =

n=1
e

3
2
t
_
c
n
cos
_
(n)
2

5
4
t +d
n
sin
_
(n)
2

5
4
t
_
sinnx.
Initial conditions give
0 = u(x, 0) =

n=1
c
n
sin nx.
By orthogonality, we may multiply this equations by sin mx and integrate:
_
1
0
0 dx =
1
2
c
m
c
m
= 0.
Partial Dierential Equations Igor Yanovsky, 2005 307
Thus,
u(x, t) =

n=1
d
n
e

3
2
t
_
sin
_
(n)
2

5
4
t
_
sinnx.
u
t
(x, t) =

n=1
_

3
2
d
n
e

3
2
t
_
sin
_
(n)
2

5
4
t
_
+d
n
e

3
2
t
_
_
(n)
2

5
4
__
cos
_
(n)
2

5
4
t
_
_
sinnx,
x sin(2x) = u
t
(x, 0) =

n=1
d
n
_
_
(n)
2

5
4
_
sinnx.
By orthogonality, we may multiply this equations by sin mx and integrate:
_
1
0
x sin(2x) sin(mx) dx = d
m
1
2
_
_
(m)
2

5
4
_
,
d
n
=
2
_
(n)
2

5
4
_
1
0
x sin(2x) sin(nx) dx.
u(x, t) = e

3
2
t

n=1
d
n
_
sin
_
(n)
2

5
4
t
_
sin nx.
Problem (F04, #1). Solve the following initial-boundary value problem for the wave
equation with a potential term,

u
tt
u
xx
+ u = 0 0 < x < , t < 0
u(0, t) = u(, t) = 0 t > 0
u(x, 0) = f(x), u
t
(x, 0) = 0 0 < x < ,
where
f(x) =
_
x if x (0, /2),
x if x (/2, ).
The answer should be given in terms of an innite series of explicitly given functions.
Proof. Assume u(x, t) = X(x)T(t), then substitution in the PDE gives
XT

T +XT = 0,
T

T
+ 1 =
X

X
= .
From X

+ X = 0, X
n
(x) = a
n
cos

n
x + b
n
sin

n
x. Boundary conditions
give
_
u(0, t) = X(0)T(t) = 0
u(, t) = X()T(t) = 0
X(0) = X() = 0.
Thus, X
n
(0) = a
n
= 0, and X
n
(x) = b
n
sin

n
x.
X
n
() = b
n
sin

n
= 0. Hence,

n
= n, or
n
= n
2
, n = 1, 2, . . ..

n
= n
2
, X
n
(x) = b
n
sinnx.
Partial Dierential Equations Igor Yanovsky, 2005 308
With these values of
n
, we solve
T

+T =
n
T,
T

+T = n
2
T,
T

n
+ (1 +n
2
)T
n
= 0.
The solution to this 2nd-order ODE is of the form:
T
n
(t) = c
n
cos
_
1 +n
2
t + d
n
sin
_
1 + n
2
t.
u(x, t) = X(x)T(t) =

n=1
_
c
n
cos
_
1 + n
2
t +d
n
sin
_
1 +n
2
t
_
sinnx.
u
t
(x, t) =

n=1
_
c
n
(
_
1 +n
2
) sin
_
1 +n
2
t +d
n
(
_
1 + n
2
) cos
_
1 +n
2
t
_
sin nx.
Initial conditions give
f(x) = u(x, 0) =

n=1
c
n
sin nx.
0 = u
t
(x, 0) =

n=1
d
n
(
_
1 +n
2
) sinnx.
By orthogonality, we may multiply both equations by sinmx and integrate:
_

0
f(x) sinmx dx = c
m

2
,
_

0
0 dx = d
m

2
_
1 +m
2
,
which gives the coecients
c
n
=
2

_

0
f(x) sinnx dx =
2

_
2
0
x sinnx dx +
2

2
( x) sinnx dx
=
2

_
x
1
n
cos nx

2
0
+
1
n
_
2
0
cos nx dx
_
+
2

n
cos nx

2
+ x
1
n
cos nx

1
n
_

2
cos nx dx
_
=
2

2n
cos
n
2
+
1
n
2
sin
n
2

1
n
2
sin0
_
+
2

n
cos n +

n
cos
n
2
+

n
cos n

2n
cos
n
2

1
n
2
sin n +
1
n
2
sin
n
2
_
=
2

_
1
n
2
sin
n
2
_
+
2

_
1
n
2
sin
n
2
_
=
4
n
2
sin
n
2
=

0, n = 2k
4
n
2
, n = 4m+ 1

4
n
2
, n = 4m+ 3
=
_
0, n = 2k
(1)
n1
2
4
n
2
, n = 2k + 1.
d
n
= 0.
u(x, t) =

n=1
_
c
n
cos
_
1 + n
2
t
_
sinnx.
Partial Dierential Equations Igor Yanovsky, 2005 309
25 Problems: Separation of Variables - Heat Equation
Problem (F94, #5).
Solve the initial-boundary value problem

u
t
= u
xx
0 < x < 2, t > 0
u(x, 0) = x
2
x + 1 0 x 2
u(0, t) = 1, u(2, t) = 3 t > 0.
Find lim
t+
u(x, t).
Proof. First, we need to obtain function v that satises v
t
= v
xx
and takes 0
boundary conditions. Let
v(x, t) = u(x, t) + (ax + b), (25.1)
where a and b are constants to be determined. Then,
v
t
= u
t
,
v
xx
= u
xx
.
Thus,
v
t
= v
xx
.
We need equation (25.1) to take 0 boundary conditions for v(0, t) and v(2, t):
v(0, t) = 0 = u(0, t) +b = 1 + b b = 1,
v(2, t) = 0 = u(2, t) + 2a 1 = 2a + 2 a = 1.
Thus, (25.1) becomes
v(x, t) = u(x, t) x 1. (25.2)
The new problem is

v
t
= v
xx
,
v(x, 0) = (x
2
x + 1) x 1 = x
2
2x,
v(0, t) = v(2, t) = 0.
We solve the problem for v using the method of separation of variables.
Let v(x, t) = X(x)T(t), which gives XT

T = 0.
X

X
=
T

T
= .
From X

+X = 0, we get X
n
(x) = a
n
cos

x +b
n
sin

x.
Using boundary conditions, we have
_
v(0, t) = X(0)T(t) = 0
v(2, t) = X(2)T(t) = 0
X(0) = X(2) = 0.
Hence, X
n
(0) = a
n
= 0, and X
n
(x) = b
n
sin

x.
X
n
(2) = b
n
sin 2

= 0 2

= n
n
= (
n
2
)
2
.
X
n
(x) = b
n
sin
nx
2
,
n
=
_
n
2
_
2
.
Partial Dierential Equations Igor Yanovsky, 2005 310
With these values of
n
, we solve T

+
_
n
2
_
2
T = 0 to nd
T
n
(t) = c
n
e
(
n
2
)
2
t
.
v(x, t) =

n=1
X
n
(x)T
n
(t) =

n=1
c
n
e
(
n
2
)
2
t
sin
nx
2
.
Coecients c
n
are obtained using the initial condition:
v(x, 0) =

n=1
c
n
sin
nx
2
= x
2
2x.
c
n
=
_
2
0
(x
2
2x) sin
nx
2
dx =
_
0 n is even,

32
(n)
3
n is odd.
v(x, t) =

n=2k1

32
(n)
3
e
(
n
2
)
2
t
sin
nx
2
.
We now use equation (25.2) to convert back to function u:
u(x, t) = v(x, t) +x + 1.
u(x, t) =

n=2k1

32
(n)
3
e
(
n
2
)
2
t
sin
nx
2
+ x + 1.
lim
t+
u(x, t) = x + 1.
Partial Dierential Equations Igor Yanovsky, 2005 311
Problem (S96, #6).
Let u(x, t) be the solution of the initial-boundary value problem for the heat equation

u
t
= u
xx
0 < x < L, t > 0
u(x, 0) = f(x) 0 x L
u
x
(0, t) = u
x
(L, t) = A t > 0 (A = Const).
Find v(x) - the limit of u(x, t) when t . Show that v(x) is one of the ininitely
many solutions of the stationary problem
v
xx
= 0 0 < x < L
v
x
(0) = v
x
(L) = A.
Proof. First, we need to obtain function v that satises v
t
= v
xx
and takes 0
boundary conditions. Let
v(x, t) = u(x, t) + (ax + b), (25.3)
where a and b are constants to be determined. Then,
v
t
= u
t
,
v
xx
= u
xx
.
Thus,
v
t
= v
xx
.
We need equation (25.3) to take 0 boundary conditions for v
x
(0, t) and v
x
(L, t).
v
x
= u
x
+a.
v
x
(0, t) = 0 = u
x
(0, t) +a = A +a a = A,
v
x
(L, t) = 0 = u
x
(L, t) + a = A+a a = A.
We may set b = 0 (innitely many solutions are possible, one for each b).
Thus, (25.3) becomes
v(x, t) = u(x, t) Ax. (25.4)
The new problem is

v
t
= v
xx
,
v(x, 0) = f(x) Ax,
v
x
(0, t) = v
x
(L, t) = 0.
We solve the problem for v using the method of separation of variables.
Let v(x, t) = X(x)T(t), which gives XT

T = 0.
X

X
=
T

T
= .
From X

+X = 0, we get X
n
(x) = a
n
cos

x +b
n
sin

x.
Using boundary conditions, we have
_
v
x
(0, t) = X

(0)T(t) = 0
v
x
(L, t) = X

(L)T(t) = 0
X

(0) = X

(L) = 0.
Partial Dierential Equations Igor Yanovsky, 2005 312
X

n
(x) = a
n

sin

x +b
n

cos

x.
Hence, X

n
(0) = b
n

n
= 0 b
n
= 0; and X
n
(x) = a
n
cos

x.
X

n
(L) = a
n

sinL

= 0 L

= n
n
= (
n
L
)
2
.
X
n
(x) = a
n
cos
nx
L
,
n
=
_
n
L
_
2
.
With these values of
n
, we solve T

+
_
n
L
_
2
T = 0 to nd
T
0
(t) = c
0
, T
n
(t) = c
n
e
(
n
L
)
2
t
, n = 1, 2, . . . .
v(x, t) =

n=1
X
n
(x)T
n
(t) = c
0
+

n=1
c
n
e
(
n
L
)
2
t
cos
nx
L
.
Coecients c
n
are obtained using the initial condition:
v(x, 0) = c
0
+

n=1
c
n
cos
nx
L
= f(x) Ax.
L c
0
=
_
L
0
(f(x) Ax) dx =
_
L
0
f(x) dx
AL
2
2
c
0
=
1
L
_
L
0
f(x) dx
AL
2
,
L
2
c
n
=
_
L
0
(f(x) Ax) cos
nx
L
dx c
n
=
1
L
_
L
0
(f(x) Ax) cos
nx
L
dx.
v(x, t) =
1
L
_
L
0
f(x) dx
AL
2
+

n
c
n
e
(
n
L
)
2
t
cos
nx
L
.
We now use equation (25.4) to convert back to function u:
u(x, t) = v(x, t) +Ax.
u(x, t) =
1
L
_
L
0
f(x) dx
AL
2
+

n
c
n
e
(
n
L
)
2
t
cos
nx
L
+ Ax.
lim
t+
u(x, t) = Ax + b, b arbitrary.
To show that v(x) is one of the ininitely many solutions of the stationary problem
v
xx
= 0 0 < x < L
v
x
(0) = v
x
(L) = A,
we can solve the boundary value problem to obtain v(x, t) = Ax+b, where b is arbitrary.
Partial Dierential Equations Igor Yanovsky, 2005 313
Heat Equation with Nonhomogeneous Time-Independent BC in N-dimensions.
The solution to this problem takes somewhat dierent approach than in the last few prob-
lems, but is similar.
Consider the following initial-boundary value problem,

u
t
= u, x , t 0
u(x, 0) = f(x), x
u(x, t) = g(x), x , t > 0.
Proof. Let w(x) be the solution of the Dirichlet problem:
_
w = 0, x
w(x) = g(x), x
and let v(x, t) be the solution of the IBVP for the heat equation with homogeneous
BC:

v
t
= v, x , t 0
v(x, 0) = f(x) w(x), x
v(x, t) = 0, x , t > 0.
Then u(x, t) satises
u(x, t) = v(x, t) +w(x).
lim
t
u(x, t) = w(x).
Partial Dierential Equations Igor Yanovsky, 2005 314
Nonhomogeneous Heat Equation with Nonhomogeneous Time-Independent
BC in N dimensions.
Describe the method of solution of the problem

u
t
= u + F(x, t), x , t 0
u(x, 0) = f(x), x
u(x, t) = g(x), x , t > 0.
Proof. We rst nd u
1
, the solution to the homogeneous heat equation (no F(x, t)).
Let w(x) be the solution of the Dirichlet problem:
_
w = 0, x
w(x) = g(x), x
and let v(x, t) be the solution of the IBVP for the heat equation with homogeneous
BC:

v
t
= v, x , t 0
v(x, 0) = f(x) w(x), x
v(x, t) = 0, x , t > 0.
Then u
1
(x, t) satises
u
1
(x, t) = v(x, t) +w(x).
lim
t
u
1
(x, t) = w(x).
The solution to the homogeneous equation with 0 boundary conditions is given by
Duhamels principle.
_
u
2t
= u
2
+F(x, t) for t > 0, x R
n
u
2
(x, 0) = 0 for x R
n
.
(25.5)
Duhamels principle gives the solution:
u
2
(x, t) =
_
t
0
_
R
n

K(x y, t s) F(y, s) dy ds
Note: u
2
(x, t) = 0 on may not be satised.
u(x, t) = v(x, t) + w(x) +
_
t
0
_
R
n

## K(x y, t s) F(y, s) dy ds.

Partial Dierential Equations Igor Yanovsky, 2005 315
Problem (S98, #5). Find the solution of

u
t
= u
xx
, t 0, 0 < x < 1,
u(x, 0) = 0, 0 < x < 1,
u(0, t) = 1 e
t
, u
x
(1, t) = e
t
1, t > 0.
Prove that lim
t
u(x, t) exists and nd it.
Proof. First, we need to obtain function v that satises v
t
= v
xx
and takes 0
boundary conditions. Let
v(x, t) = u(x, t) + (ax + b) + (c
1
cos x +c
2
sin x)e
t
, (25.6)
where a, b, c
1
, c
2
are constants to be determined. Then,
v
t
= u
t
(c
1
cos x + c
2
sinx)e
t
,
v
xx
= u
xx
+ (c
1
cos x c
2
sin x)e
t
.
Thus,
v
t
= v
xx
.
We need equation (25.6) to take 0 boundary conditions for v(0, t) and v
x
(1, t):
v(0, t) = 0 = u(0, t) +b +c
1
e
t
= 1 e
t
+b +c
1
e
t
.
Thus, b = 1, c
1
= 1, and (25.6) becomes
v(x, t) = u(x, t) + (ax 1) + (cos x +c
2
sin x)e
t
. (25.7)
v
x
(x, t) = u
x
(x, t) + a + (sinx + c
2
cos x)e
t
,
v
x
(1, t) = 0 = u
x
(1, t) +a + (sin1 +c
2
cos 1)e
t
= 1 + a + (1 sin 1 + c
2
cos 1)e
t
.
Thus, a = 1, c
2
=
sin11
cos 1
, and equation (25.7) becomes
v(x, t) = u(x, t) + (x 1) + (cos x +
sin 1 1
cos 1
sin x)e
t
. (25.8)
Initial condition tranforms to:
v(x, 0) = u(x, 0) + (x 1) + (cos x +
sin 1 1
cos 1
sin x) = (x 1) + (cos x +
sin1 1
cos 1
sinx).
The new problem is

v
t
= v
xx
,
v(x, 0) = (x 1) + (cos x +
sin11
cos 1
sinx),
v(0, t) = 0, v
x
(1, t) = 0.
We solve the problem for v using the method of separation of variables.
Let v(x, t) = X(x)T(t), which gives XT

T = 0.
X

X
=
T

T
= .
Partial Dierential Equations Igor Yanovsky, 2005 316
From X

+X = 0, we get X
n
(x) = a
n
cos

x +b
n
sin

x.
Using the rst boundary condition, we have
v(0, t) = X(0)T(t) = 0 X(0) = 0.
Hence, X
n
(0) = a
n
= 0, and X
n
(x) = b
n
sin

x. We also have
v
x
(1, t) = X

(1)T(t) = 0 X

(1) = 0.
X

n
(x) =

b
n
cos

x,
X

n
(1) =

b
n
cos

= 0,
cos

= 0,

= n +

2
.
Thus,
X
n
(x) = b
n
sin
_
n +

2
_
x,
n
=
_
n +

2
_
2
.
With these values of
n
, we solve T

+
_
n +

2
_
2
T = 0 to nd
T
n
(t) = c
n
e
(n+

2
)
2
t
.
v(x, t) =

n=1
X
n
(x)T
n
(t) =

n=1

b
n
sin
_
n +

2
_
x e
(n+

2
)
2
t
.
We now use equation (25.8) to convert back to function u:
u(x, t) = v(x, t) (x 1) (cos x +
sin 1 1
cos 1
sin x)e
t
.
u(x, t) =

n=1

b
n
sin
_
n +

2
_
x e
(n+

2
)
2
t
(x 1) (cos x +
sin 1 1
cos 1
sin x)e
t
.
Coecients

b
n
are obtained using the initial condition:
u(x, 0) =

n=1

b
n
sin
_
n +

2
_
x (x 1) (cos x +
sin1 1
cos 1
sinx).
Finally, we can check that the dierential equation and the boundary conditions are
satised:
u(0, t) = 1 (1 + 0)e
t
= 1 e
t
.
u
x
(x, t) =

n=1

b
n
_
n +

2
_
cos
_
n +

2
_
x e
(n+

2
)
2
t
1 + (sinx
sin1 1
cos 1
cos x)e
t
,
u
x
(1, t) = 1 + (sin1
sin 1 1
cos 1
cos 1)e
t
= 1 + e
t
.
u
t
=

n=1

b
n
_
n +

2
_
2
sin
_
n +

2
_
x e
(n+

2
)
2
t
+ (cos x +
sin1 1
cos 1
sin x)e
t
= u
xx
.
Partial Dierential Equations Igor Yanovsky, 2005 317
Problem (F02, #6). The temperature of a rod insulated at the ends with an ex-
ponentially decreasing heat source in it is a solution of the following boundary value
problem:

u
t
= u
xx
+e
2t
g(x) for (x, t) [0, 1] R
+
u
x
(0, t) = u
x
(1, t) = 0
u(x, 0) = f(x).
Find the solution to this problem by writing u as a cosine series,
u(x, t) =

n=0
a
n
(t) cos nx,
and determine lim
t
u(x, t).
Proof. Let g accept an expansion in eigenfunctions
g(x) = b
0
+

n=1
b
n
cos nx with b
n
= 2
_
1
0
g(x) cos nx dx.
Plugging in the PDE gives:
a

0
(t) +

n=1
a

n
(t) cos nx =

n=1
n
2

2
a
n
(t) cos nx + b
0
e
2t
+e
2t

n=1
b
n
cos nx,
which gives
_
a

0
(t) = b
0
e
2t
,
a

n
(t) +n
2

2
a
n
(t) = b
n
e
2t
, n = 1, 2, . . . .
Adding homogeneous and particular solutions of the above ODEs, we obtain the solu-
tions
_
a
0
(t) = c
0

b
0
2
e
2t
,
a
n
(t) = c
n
e
n
2

2
t

bn
2n
2

2
e
2t
, n = 1, 2, . . . ,
for some constants c
n
, n = 0, 1, 2, . . . . Thus,
u(x, t) =

n=0
_
c
n
e
n
2

2
t

b
n
2 n
2

2
e
2t
_
cos nx.
Initial condition gives
u(x, 0) =

n=0
_
c
n

b
n
2 n
2

2
_
cos nx = f(x),
As, t , the only mode that survives is n = 0:
u(x, t) c
0
+
b
0
2
as t .
Partial Dierential Equations Igor Yanovsky, 2005 318
Problem (F93, #4). a) Assume f, g C

## . Give the compatibility conditions which

f and g must satisfy if the following problem is to possess a solution.
u = f(x) x
u
n
(s) = g(s) s .
Show that your condition is necessary for a solution to exist.
b) Give an explicit solution to

u
t
= u
xx
+ cos x x [0, 2]
u
x
(0, t) = u
x
(2, t) = 0 t > 0
u(x, 0) = cos x + cos 2x x [0, 2].
c) Does there exist a steady state solution to the problem in (b) if
u
x
(0) = 1 u
x
(2) = 0 ?
Proof. a) Integrating the equation and using Greens identity gives:
_

f(x) dx =
_

u dx =
_

u
n
ds =
_

g(s) ds.
b) With
v(x, t) = u(x, t) cos x
the problem above transforms to

v
t
= v
xx
v
x
(0, t) = v
x
(2, t) = 0
v(x, 0) = cos 2x.
We solve this problem for v using the separation of variables. Let v(x, t) = X(x)T(t),
which gives XT

= X

T.
X

X
=
T

T
= .
From X

+X = 0, we get X
n
(x) = a
n
cos

x +b
n
sin

x.
X

n
(x) =

n
a
n
sin

x +

n
b
n
cos

x.
Using boundary conditions, we have
_
v
x
(0, t) = X

(0)T(t) = 0
v
x
(2, t) = X

(2)T(t) = 0
X

(0) = X

(2) = 0.
Hence, X

n
(0) =

n
b
n
= 0, and X
n
(x) = a
n
cos

n
x.
X

n
(2) =

n
a
n
sin

n
2 = 0

n
=
n
2

n
= (
n
2
)
2
. Thus,
X
n
(x) = a
n
cos
nx
2
,
n
=
_
n
2
_
2
Partial Dierential Equations Igor Yanovsky, 2005 319
With these values of
n
, we solve T

+
_
n
2
_
2
T = 0 to nd
T
n
(t) = c
n
e
(
n
2
)
2
t
.
v(x, t) =

n=0
X
n
(x)T
n
(t) =

n=0
a
n
e
(
n
2
)
2
t
cos
nx
2
.
Initial condition gives
v(x, 0) =

n=0
a
n
cos
nx
2
= cos 2x.
Thus, a
4
= 1, a
n
= 0, n ,= 4. Hence,
v(x, t) = e
4t
cos 2x.
u(x, t) = v(x, t) + cos x = e
4t
cos 2x + cos x.
c) Does there exist a steady state solution to the problem in (b) if
u
x
(0) = 1 u
x
(2) = 0 ?
c) Set u
t
= 0. We have
_
u
xx
+ cos x = 0 x [0, 2]
u
x
(0) = 1, u
x
(2) = 0.
u
xx
= cos x,
u
x
= sin x +C,
u(x) = cos x +Cx + D.
Boundary conditions give:
1 = u
x
(0) = C,
0 = u
x
There exists no steady state solution.
We may use the result we obtained in part (a) with u
xx
= cos x = f(x). We
need
_

f(x) dx =
_

u
n
ds,
_
2
0
cos x dx
. .
=0
= u
x
(2) u
x
(0) = 1
..
given
.
Partial Dierential Equations Igor Yanovsky, 2005 320
Problem (F96, #7). Solve the parabolic problem
_
u
v
_
t
=
_
1
1
2
0 2
__
u
v
_
xx
, 0 x , t > 0
u(x, 0) = sinx, u(0, t) = u(, t) = 0,
v(x, 0) = sin x, v(0, t) = v(, t) = 0.
Prove the energy estimate (for general initial data)
_

x=0
[u
2
(x, t) + v
2
(x, t)] dx c
_

x=0
[u
2
(x, 0) +v
2
(x, 0)] dx
for come constant c.
Proof. We can solve the second equation for v and then use the value of v to solve the
rst equation for u.
73
We have

v
t
= 2v
xx
, 0 x , t > 0
v(x, 0) = sinx,
v(0, t) = v(, t) = 0.
Assume v(x, t) = X(x)T(t), then substitution in the PDE gives XT

= 2X

T.
T

T
= 2
X

X
= .
From X

+

2
X = 0, we get X
n
(x) = a
n
cos
_

2
x + b
n
sin
_

2
x.
Boundary conditions give
_
v(0, t) = X(0)T(t) = 0
v(, t) = X()T(t) = 0
X(0) = X() = 0.
Thus, X
n
(0) = a
n
= 0, and X
n
(x) = b
n
sin
_

2
x.
X
n
() = b
n
sin
_

2
= 0. Hence
_

2
= n, or = 2n
2
.
= 2n
2
, X
n
(x) = b
n
sinnx.
With these values of
n
, we solve T

+ 2n
2
T = 0 to get T
n
(t) = c
n
e
2n
2
t
.
Thus, the solution may be written in the form
v(x, t) =

n=1
a
n
e
2n
2
t
sin nx.
From initial condition, we get
v(x, 0) =

n=1
a
n
sinnx = sinx.
Thus, a
1
= 1, a
n
= 0, n = 2, 3, . . ..
v(x, t) = e
2t
sin x.
73
Note that if the matrix was fully inseparable, we would have to nd eigenvalues and eigenvectors,
just as we did for the hyperbolic systems.
Partial Dierential Equations Igor Yanovsky, 2005 321
We have

u
t
= u
xx

1
2
e
2t
sinx, 0 x , t > 0
u(x, 0) = sin x,
u(0, t) = u(, t) = 0.
Let u(x, t) =

n=1
u
n
(t) sinnx. Plugging this into the equation, we get

n=1
u

n
(t) sinnx +

n=1
n
2
u
n
(t) sinnx =
1
2
e
2t
sin x.
For n = 1:
u

1
(t) + u
1
(t) =
1
2
e
2t
.
Combining homogeneous and particular solution of the above equation, we obtain:
u
1
(t) =
1
2
e
2t
+c
1
e
t
.
For n = 2, 3, . . .:
u

n
(t) + n
2
u
n
(t) = 0,
u
n
(t) = c
n
e
n
2
t
.
Thus,
u(x, t) =
_
1
2
e
2t
+c
1
e
t
_
sinx +

n=2
c
n
e
n
2
t
sinnx =
1
2
e
2t
sinx +

n=1
c
n
e
n
2
t
sinnx.
From initial condition, we get
u(x, 0) =
1
2
sin x +

n=1
c
n
sinnx = sin x.
Thus, c
1
=
1
2
, c
n
= 0, n = 2, 3, . . ..
u(x, t) =
1
2
sinx (e
2t
+e
t
).
To prove the energy estimate (for general initial data)
_

x=0
[u
2
(x, t) + v
2
(x, t)] dx c
_

x=0
[u
2
(x, 0) +v
2
(x, 0)] dx
for come constant c, we assume that
u(x, 0) =

n=1
a
n
sinnx, v(x, 0) =

n=1
b
n
sinnx.
Partial Dierential Equations Igor Yanovsky, 2005 322
The general solutions are obtained by the same method as above
u(x, t) =
1
2
e
2t
sinx +

n=1
c
n
e
n
2
t
sinnx,
v(x, t) =

n=1
b
n
e
2n
2
t
sinnx.
_

x=0
[u
2
(x, t) +v
2
(x, t)] dx =
_

x=0
_
1
2
e
2t
sinx +

n=1
c
n
e
n
2
t
sinnx
_
2
+
_

n=1
b
n
e
2n
2
t
sinnx
_
2
dx

n=1
(b
2
n
+a
2
n
)
_

x=0
sin
2
nx dx
_

x=0
[u
2
(x, 0) +v
2
(x, 0)] dx.
Partial Dierential Equations Igor Yanovsky, 2005 323
26 Problems: Eigenvalues of the Laplacian - Laplace
The 2D LAPLACE Equation (eigenvalues/eigenfuctions of the Laplacian).
Consider

u
xx
+ u
yy
+ u = 0 in
u(0, y) = 0 = u(a, y) for 0 y b,
u(x, 0) = 0 = u(x, b) for 0 x a.
(26.1)
Proof. We can solve this problem by separation of variables.
Let u(x, y) = X(x)Y (y), then substitution in the PDE gives X

Y +XY

+XY = 0.
X

X
+
Y

Y
+ = 0.
Letting =
2
+
2
and using boundary conditions, we nd the equations for X and
Y :
X

+
2
X = 0 Y

+
2
Y = 0
X(0) = X(a) = 0 Y (0) = Y (b) = 0.
The solutions of these one-dimensional eigenvalue problems are

m
=
m
a

n
=
n
b
X
m
(x) = sin
mx
a
Y
n
(y) = sin
ny
b
,
where m, n = 1, 2, . . . . Thus we obtain solutions of (26.1) of the form

mn
=
2
_
m
2
a
2
+
n
2
b
2
_
u
mn
(x, y) = sin
mx
a
sin
ny
b
,
where m, n = 1, 2, . . . .
Observe that the eigenvalues
mn

m,n=1
are positive. The smallest eigenvalue
11
has only one eigenfunction u
11
(x, y) = sin(x/a) sin(y/b); notice that u
11
is positive
in . Other eigenvalues may correspond to more than one choice of m and n; for
example, in the case a = b we have
nm
=
nm
. For this , there are two linearly
independent eigenfunctions. However, for a particular value of there are at most
nitely many linearly independent eigenfunctions. Moreover,
_
b
0
_
a
0
u
mn
(x, y) u
m

n
(x, y) dx dy =
_
b
0
_
a
0
sin
mx
a
sin
ny
b
sin
m

x
a
sin
n

y
b
dx dy
=
_
a
2
_
b
0
sin
ny
b
sin
n

y
b
dy
0
=
_
ab
4
if m = m

and n = n

0 if m ,= m

or n ,= n

.
In particular, the u
mn
are pairwise orthogonal. We could normalize each u
mn
by a
scalar multiple (i.e. multiply by
_
4/ab) so that ab/4 above becomes 1.
Let us change the notation somewhat so that each eigenvalue
n
corresponds to a
particular eigenfunction
n
(x). If we choose an orthonormal basis of eigenfunctions in
each eigenspace, we may arrange that
n

n=1
is pairwise orthonormal:
_

n
(x)
m
(x) dx =
_
1 if m = n
0 if m ,= n.
Partial Dierential Equations Igor Yanovsky, 2005 324
In this notation, the eigenfunction expansion of f(x) dened on becomes
f(x)

n=1
a
n

n
(x), where a
n
=
_

f(x)
n
(x) dx.
Partial Dierential Equations Igor Yanovsky, 2005 325
Problem (S96, #4). Let D denote the rectangular
D = (x, y) R
2
: 0 < x < a, 0 < y < b.
Find the eigenvalues of the following Dirichlet problem:
(+)u = 0 in D
u = 0 on D.
Proof. The problem may be rewritten as

u
xx
+ u
yy
+ u = 0 in
u(0, y) = 0 = u(a, y) for 0 y b,
u(x, 0) = 0 = u(x, b) for 0 x a.
We may assume that the eigenvalues are positive, =
2
+
2
. Then,

mn
=
2
_
m
2
a
2
+
n
2
b
2
_
u
mn
(x, y) = sin
mx
a
sin
ny
b
, m, n = 1, 2, . . . .
Problem (W04, #1). Consider the dierential equation:

2
u(x, y)
x
2
+

2
u(x, y)
y
2
+u(x, y) = 0 (26.2)
in the strip (x, y), 0 < y < , < x < + with boundary conditions
u(x, 0) = 0, u(x, ) = 0. (26.3)
Find all bounded solutions of the boundary value problem (26.4), (26.5) when
a) = 0, b) > 0, c) < 0.
Proof. a) = 0. We have
u
xx
+u
yy
= 0.
Assume u(x, y) = X(x)Y (y), then substitution in the PDE gives
X

Y +XY

= 0.
Boundary conditions give
_
u(x, 0) = X(x)Y (0) = 0
u(x, ) = X(x)Y () = 0
Y (0) = Y () = 0.
Method I: We have
X

X
=
Y

Y
= c, c > 0.
From X

+ cX = 0, we have X
n
(x) = a
n
cos

cx +b
n
sin

cx.
From Y

cY = 0, we have Y
n
(y) = c
n
e

cy
+ d
n
e

cy
.
Y (0) = c
n
+ d
n
= 0 c
n
= d
n
.
Partial Dierential Equations Igor Yanovsky, 2005 326
Y () = c
n
e

c
c
n
e

c
= 0 c
n
= 0 Y
n
(y) = 0.
u(x, y) = X(x)Y (y) = 0.
Method II: We have
X

X
=
Y

Y
= c, c > 0.
From X

cX = 0, we have X
n
(x) = a
n
e

cx
+b
n
e

cx
.
Since we look for bounded solutions for < x < , a
n
= b
n
= 0 X
n
(x) = 0.
From Y

+ cY = 0, we have Y
n
(y) = c
n
cos

cy +d
n
sin

cy.
Y (0) = c
n
= 0,
Y () = d
n
sin

c = 0

c = n c = n
2
.
Y
n
(y) = d
n
sin nx = 0.
u(x, y) = X(x)Y (y) = 0.
b) > 0. We have
X

X
+
Y

Y
+ = 0.
Letting =
2
+
2
, and using boundary conditions for Y , we nd the equations:
X

+
2
X = 0 Y

+
2
Y = 0
Y (0) = Y () = 0.
The solutions of these one-dimensional eigenvalue problems are
X
m
(x) = a
m
cos
m
x +b
m
sin
m
x.

n
= n, Y
n
(y) = d
n
sin ny, where m, n = 1, 2, . . ..
u(x, y) =

m,n=1
u
mn
(x, y) =

m,n=1
(a
m
cos
m
x + b
m
sin
m
x) sinny.
c) < 0. We have
u
xx
+ u
yy
+ u = 0,
u(x, 0) = 0, u(x, ) = 0.
u 0 is the solution to this equation. We will show that this solution is unique.
Let u
1
and u
2
be two solutions, and consider w = u
1
u
2
. Then,
w + w = 0,
w(x, 0) = 0, w(x, ) = 0.
Multiply the equation by w and integrate:
ww + w
2
= 0,
_

wwdx +
_

w
2
dx = 0,
_

w
w
n
ds
. .
=0

[w[
2
dx +
_

w
2
dx = 0,
_

[w[
2
dx
. .
0
=
_

w
2
dx
. .
0
.
Partial Dierential Equations Igor Yanovsky, 2005 327
Thus, w 0 and the solution u(x, y) 0 is unique.
Partial Dierential Equations Igor Yanovsky, 2005 328
Problem (F95, #5). Find all bounded solutions
for the following boundary value problem in the strip
0 < x < a, < y < ,
(+k
2
)u = 0 (k = Const > 0),
u(0, y) = 0, u
x
(a, y) = 0.
In particular, show that when ak ,
the only bounded solution to this problem is u 0.
Proof. Let u(x, y) = X(x)Y (y), then we have X

Y + XY

+k
2
XY = 0.
X

X
+
Y

Y
+ k
2
= 0.
Letting k
2
=
2
+
2
and using boundary conditions, we nd:
X

+
2
X = 0, Y

+
2
Y = 0.
X(0) = X

(a) = 0.
The solutions of these one-dimensional eigenvalue problems are

m
=
(m
1
2
)
a
,
X
m
(x) = sin
(m
1
2
)x
a
Y
n
(y) = c
n
cos
n
y +d
n
sin
n
y,
where m, n = 1, 2, . . . . Thus we obtain solutions of the form
k
2
mn
=
_
(m
1
2
)
a
_
2
+
2
n
, u
mn
(x, y) = sin
(m
1
2
)x
a
_
c
n
cos
n
y+d
n
sin
n
y
_
,
where m, n = 1, 2, . . . .
u(x, y) =

m,n=1
u
mn
(x, y) =

m,n=1
sin
(m
1
2
)x
a
_
c
n
cos
n
y +d
n
sin
n
y
_
.
We can take an alternate approach and prove the second part of the question. We
have
X

Y +XY

+ k
2
XY = 0,

Y

Y
=
X

X
+ k
2
= c
2
.
We obtain Y
n
(y) = c
n
cos cy + d
n
sin cy. The second equation gives
X

+ k
2
X = c
2
X,
X

+ (k
2
c
2
)X = 0,
X
m
(x) = a
m
e

c
2
k
2
x
+ b
m
e

c
2
k
2
x
.
Thus, X
m
(x) is bounded only if k
2
c
2
> 0, (if k
2
c
2
= 0, X

= 0, and X
m
(x) =
a
m
x + b
m
, BCs give X
m
(x) = x, unbounded), in which case
X
m
(x) = a
m
cos
_
k
2
c
2
x +b
m
sin
_
k
2
c
2
x.
Partial Dierential Equations Igor Yanovsky, 2005 329
Boundary conditions give X
m
(0) = a
m
= 0.
X

m
(x) = b
m
_
k
2
c
2
cos
_
k
2
c
2
x,
X

m
(a) = b
m
_
k
2
c
2
cos
_
k
2
c
2
a = 0,
_
k
2
c
2
a = m

2
, m = 1, 2, . . . ,
k
2
c
2
=
_

a
_
m
1
2
__
2
,
k
2
=
_

a
_
2
_
m
1
2
_
2
+ c
2
,
a
2
k
2
>
2
_
m
1
2
_
2
,
ak >
_
m
1
2
_
, m = 1, 2, . . . .
Thus, bounded solutions exist only when ak >

2
.
Problem (S90, #2). Show that the boundary value problem

2
u(x, y)
x
2
+

2
u(x, y)
y
2
+k
2
u(x, y) = 0, (26.4)
where < x < +, 0 < y < , k > 0 is a constant,
u(x, 0) = 0, u(x, ) = 0 (26.5)
has a bounded solution if and only if k 1.
Proof. We have
u
xx
+u
yy
+k
2
u = 0,
X

Y +XY

+ k
2
XY = 0,

X
=
Y

Y
+ k
2
= c
2
.
We obtain X
m
(x) = a
m
cos cx + b
m
sin cx. The second equation gives
Y

+k
2
Y = c
2
Y,
Y

+ (k
2
c
2
)Y = 0,
Y
n
(y) = c
n
e

c
2
k
2
y
+d
n
e

c
2
k
2
y
.
Thus, Y
n
(y) is bounded only if k
2
c
2
> 0, (if k
2
c
2
= 0, Y

= 0, and Y
n
(y) = c
n
y+d
n
,
BCs give Y 0), in which case
Y
n
(y) = c
n
cos
_
k
2
c
2
y +d
n
sin
_
k
2
c
2
y.
Boundary conditions give Y
n
(0) = c
n
= 0.
Y
n
() = d
n
sin

k
2
c
2
= 0

k
2
c
2
= n k
2
c
2
= n
2

k
2
= n
2
+c
2
, n = 1, 2, . . .. Hence, k > n, n = 1, 2, . . ..
Partial Dierential Equations Igor Yanovsky, 2005 330
Thus, bounded solutions exist if k 1.
Note: If k = 1, then c = 0, which gives trivial solutions for Y
n
(y).
u(x, y) =

m,n=1
X
m
(x)Y
n
(y) =

m,n=1
sin ny X
m
(x).
Partial Dierential Equations Igor Yanovsky, 2005 331
McOwen, 4.4 #7; 266B Ralston Hw. Show that the boundary value problem
_
a(x)u +b(x)u = u in
u = 0 on
has only trivial solution with 0, when b(x) 0 and a(x) > 0 in .
Proof. Multiplying the equation by u and integrating over , we get
_

u au dx +
_

bu
2
dx =
_

u
2
dx.
Since (uau) = u au +a[u[
2
, we have
_

(uau) dx +
_

a[u[
2
dx +
_

bu
2
dx =
_

u
2
dx. (26.6)
Using divergence theorem, we obtain
_

u
..
=0
a
u
n
ds +
_

a[u[
2
dx +
_

bu
2
dx =
_

u
2
dx,
_

a
..
>0
[u[
2
dx +
_

b
..
0
u
2
dx =
..
0
_

u
2
dx,
Thus, u = 0 in , and u is constant. Since u = 0 on , u 0 on .
Similar Problem I: Note that this argument also works with Neumann B.C.:
_
a(x)u +b(x)u = u in
u/n = 0 on
Using divergence theorem, (26.6) becomes
_

ua
u
n
..
=0
ds +
_

a[u[
2
dx +
_

bu
2
dx =
_

u
2
dx,
_

a
..
>0
[u[
2
dx +
_

b
..
0
u
2
dx =
..
0
_

u
2
dx.
Thus, u = 0, and u = const on . Hence, we now have
_

b
..
0
u
2
dx =
..
0
_

u
2
dx,
which implies = 0. This gives the useful information that for the eigenvalue problem
74
_
a(x)u +b(x)u = u
u/n = 0,
= 0 is an eigenvalue, its eigenspace is the set of constants, and all other s are
positive.
74
In Ralstons Hw#7 solutions, there is no - sign in front of a(x)u below, which is probably a
typo.
Partial Dierential Equations Igor Yanovsky, 2005 332
Similar Problem II: If 0, we show that the only solution to the problem below
is the trivial solution.
_
u +u = 0 in
u = 0 on
_

uu dx +
_

u
2
dx = 0,
_

u
..
=0
u
n
ds
_

[u[
2
dx +
..
0
_

u
2
dx = 0.
Thus, u = 0 in , and u is constant. Since u = 0 on , u 0 on .
Partial Dierential Equations Igor Yanovsky, 2005 333
27 Problems: Eigenvalues of the Laplacian - Poisson
The ND POISSON Equation (eigenvalues/eigenfunctions of the Laplacian).
Suppose we want to nd the eigenfunction expansion of the solution of
u = f in
u = 0 on ,
when f has the expansion in the orthonormal Dirichlet eigenfunctions
n
:
f(x)

n=1
a
n

n
(x), where a
n
=
_

f(x)
n
(x) dx.
Proof. Writing u =

c
n

n
and inserting into u = f, we get

n=1

n
c
n

n
=

n=1
a
n

n
(x).
Thus, c
n
= a
n
/
n
, and
u(x) =

n=1
a
n

n
(x)

n
.
The 1D POISSON Equation (eigenvalues/eigenfunctions of the Laplacian).
For the boundary value problem
u

= f(x)
u(0) = 0, u(L) = 0,
the related eigenvalue problem is

=
(0) = 0, (L) = 0.
The eigenvalues are
n
= (n/L)
2
, and the corresponding eigenfunctions are sin(nx/L),
n = 1, 2, . . ..
Writing u =

c
n

n
=

c
n
sin(nx/L) and inserting into u = f, we get

n=1
c
n
_
n
L
_
2
sin
nx
L
= f(x),
_
L
0

n=1
c
n
_
n
L
_
2
sin
nx
L
sin
mx
L
dx =
_
L
0
f(x) sin
mx
L
dx,
c
n
_
n
L
_
2
L
2
=
_
L
0
f(x) sin
nx
L
dx,
c
n
=
2
L
_
L
0
f(x) sin(nx/L) dx
(n/L)
2
.
Partial Dierential Equations Igor Yanovsky, 2005 334
u(x) =

c
n
sin(nx/L) =

n=1

2
L
_
L
0
f() sin(nx/L) sin(n/L) d
(n/L)
2
,
u =
_
L
0
f()
_

2
L

n=1
sin(nx/L) sin(n/L)
(n/L)
2
_
. .
= G(x,)
d.
See similar, but more complicated, problem in Sturm-Liouville Problems (S92, #2(c)).
Partial Dierential Equations Igor Yanovsky, 2005 335
Example: Eigenfunction Expansion of the GREENs Function.
Suppose we x x and attempt to expand the Greens function G(x, y) in the orthonormal
eigenfunctions
n
(y):
G(x, y)

n=1
a
n
(x)
n
(y), where a
n
(x) =
_

G(x, z)
n
(z) dz.
Proof. We can rewrite u +u = 0 in , u = 0 on , as an integral equation
75
u(x) +
_

G(x, y)u(y) dy = 0.
Suppose, u(x) =

c
n

n
(x). Plugging this into , we get

m=1
c
m

m
(x) +
_

n=1
a
n
(x)
n
(y)

m=1
c
m

m
(y) dy = 0,

m=1
c
m

m
(x) +

n=1
a
n
(x)

m=1
c
m
_

n
(y)
m
(y) dy = 0,

n=1
c
n

n
(x) +

n=1
a
n
(x)c
n
= 0,

n=1
c
n
_

n
(x) + a
n
(x)
_
= 0,
a
n
(x) =

n
(x)

n
.
Thus,
G(x, y)

n=1

n
(x)
n
(y)

n
.
75
See the section: ODE - Integral Equations.
Partial Dierential Equations Igor Yanovsky, 2005 336
The 2D POISSON Equation (eigenvalues/eigenfunctions of the Laplacian).
Solve the boundary value problem

u
xx
+ u
yy
= f(x, y) for 0 < x < a, 0 < y < b
u(0, y) = 0 = u(a, y) for 0 y b,
u(x, 0) = 0 = u(x, b) for 0 x a.
(27.1)
f(x, y) C
2
, f(x, y) = 0 if x = 0, x = a, y = 0, y = b,
f(x, y) =
2

ab

m,n=1
c
mn
sin
mx
a
sin
ny
b
.
Proof. First, we nd eigenvalues/eigenfunctions of the Laplacian.

u
xx
+ u
yy
+ u = 0 in
u(0, y) = 0 = u(a, y) for 0 y b,
u(x, 0) = 0 = u(x, b) for 0 x a.
Let u(x, y) = X(x)Y (y), then substitution in the PDE gives X

Y +XY

+XY = 0.
X

X
+
Y

Y
+ = 0.
Letting =
2
+
2
and using boundary conditions, we nd the equations for X and
Y :
X

+
2
X = 0 Y

+
2
Y = 0
X(0) = X(a) = 0 Y (0) = Y (b) = 0.
The solutions of these one-dimensional eigenvalue problems are

m
=
m
a

n
=
n
b
X
m
(x) = sin
mx
a
Y
n
(y) = sin
ny
b
,
where m, n = 1, 2, . . . . Thus we obtain eigenvalues and normalized eigenfunctions of
the Laplacian:

mn
=
2
_
m
2
a
2
+
n
2
b
2
_

mn
(x, y) =
2

ab
sin
mx
a
sin
ny
b
,
where m, n = 1, 2, . . . . Note that
f(x, y) =

m,n=1
c
mn

mn
.
Second, writing u(x, y) =

c
mn

mn
and inserting into u = f, we get

m,n=1

mn
c
mn

mn
(x, y) =

m,n=1
c
mn

mn
(x, y).
Thus, c
mn
=
cmn
mn
.
u(x, y) =

n=1
c
mn

mn

mn
(x, y),
Partial Dierential Equations Igor Yanovsky, 2005 337
with
mn
,
mn
(x) given above, and c
mn
given by
_
b
0
_
a
0
f(x, y)
mn
dx dy =
_
b
0
_
a
0

,n

=1
c
m

n

m

n

mn
dx dy = c
mn
.
Partial Dierential Equations Igor Yanovsky, 2005 338
28 Problems: Eigenvalues of the Laplacian - Wave
In the section on the wave equation, we considered an initial boundary value problem
for the one-dimensional wave equation on an interval, and we found that the solu-
tion could be obtained using Fourier series. If we replace the Fourier series by an
expansion in eigenfunctions, we can consider an initial/boundary value problem for the
n-dimensional wave equation.
The ND WAVE Equation (eigenvalues/eigenfunctions of the Laplacian).
Consider

u
tt
= u for x , t > 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) for x
u(x, t) = 0 for x , t > 0.
Proof. For g, h C
2
() with g = h = 0 on , we have eigenfunction expansions
g(x) =

n=1
a
n

n
(x) and h(x) =

n=1
b
n

n
(x).
Assume the solution u(x, t) may be expanded in the eigenfunctions with coecients
depending on t: u(x, t) =

n=1
u
n
(t)
n
(x). This implies

n=1
u

n
(t)
n
(x) =

n=1

n
u
n
(t)
n
(x),
u

n
(t) +
n
u
n
(t) = 0 for each n.
Since
n
> 0, this ordinary dierential equation has general solution
u
n
(t) = A
n
cos
_

n
t +B
n
sin
_

n
t. Thus,
u(x, t) =

n=1
_
A
n
cos
_

n
t +B
n
sin
_

n
t
_

n
(x),
u
t
(x, t) =

n=1
_

n
A
n
sin
_

n
t +
_

n
B
n
cos
_

n
t
_

n
(x),
u(x, 0) =

n=1
A
n

n
(x) = g(x),
u
t
(x, 0) =

n=1
_

n
B
n

n
(x) = h(x).
Comparing with , we obtain
A
n
= a
n
, B
n
=
b
n

n
.
Thus, the solution is given by
u(x, t) =

n=1
_
a
n
cos
_

n
t +
b
n

n
sin
_

n
t
_

n
(x),
Partial Dierential Equations Igor Yanovsky, 2005 339
with
a
n
=
_

g(x)
n
(x) dx,
b
n
=
_

h(x)
n
(x) dx.
Partial Dierential Equations Igor Yanovsky, 2005 340
The 2D WAVE Equation (eigenvalues/eigenfunctions of the Laplacian).
Let = (0, a) (0, b) and consider

u
tt
= u
xx
+ u
yy
for x , t > 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) for x
u(x, t) = 0 for x , t > 0.
(28.1)
Proof. First, we nd eigenvalues/eigenfunctions of the Laplacian.

u
xx
+ u
yy
+ u = 0 in
u(0, y) = 0 = u(a, y) for 0 y b,
u(x, 0) = 0 = u(x, b) for 0 x a.
Let u(x, y) = X(x)Y (y), then substitution in the PDE gives X

Y +XY

+XY = 0.
X

X
+
Y

Y
+ = 0.
Letting =
2
+
2
and using boundary conditions, we nd the equations for X and
Y :
X

+
2
X = 0 Y

+
2
Y = 0
X(0) = X(a) = 0 Y (0) = Y (b) = 0.
The solutions of these one-dimensional eigenvalue problems are

m
=
m
a

n
=
n
b
X
m
(x) = sin
mx
a
Y
n
(y) = sin
ny
b
,
where m, n = 1, 2, . . . . Thus we obtain eigenvalues and normalized eigenfunctions of
the Laplacian:

mn
=
2
_
m
2
a
2
+
n
2
b
2
_

mn
(x, y) =
2

ab
sin
mx
a
sin
ny
b
,
where m, n = 1, 2, . . . .
Second, we solve the Wave Equation (28.1) using the space eigenfunctions.
For g, h C
2
() with g = h = 0 on , we have eigenfunction expansions
76
g(x) =

n=1
a
n

n
(x) and h(x) =

n=1
b
n

n
(x).
Assume u(x, t) =

n=1
u
n
(t)
n
(x). This implies
u

n
(t) +
n
u
n
(t) = 0 for each n.
76
In 2D, n is really mn, and x is (x, y).
Partial Dierential Equations Igor Yanovsky, 2005 341
Since
n
> 0, this ordinary dierential equation has general solution
u
n
(t) = A
n
cos
_

n
t +B
n
sin
_

n
t. Thus,
u(x, t) =

n=1
_
A
n
cos
_

n
t +B
n
sin
_

n
t
_

n
(x),
u
t
(x, t) =

n=1
_

n
A
n
sin
_

n
t +
_

n
B
n
cos
_

n
t
_

n
(x),
u(x, 0) =

n=1
A
n

n
(x) = g(x),
u
t
(x, 0) =

n=1
_

n
B
n

n
(x) = h(x).
Comparing with , we obtain
A
n
= a
n
, B
n
=
b
n

n
.
Thus, the solution is given by
u(x, t) =

m,n=1
_
a
mn
cos
_

mn
t +
b
mn

mn
sin
_

mn
t
_

mn
(x),
with
mn
,
mn
(x) given above, and
a
mn
=
_

g(x)
mn
(x) dx,
b
mn
=
_

h(x)
mn
(x) dx.
Partial Dierential Equations Igor Yanovsky, 2005 342
McOwen, 4.4 #3; 266B Ralston Hw. Consider the initial-boundary value problem

u
tt
= u + f(x, t) for x , t > 0
u(x, t) = 0 for x , t > 0
u(x, 0) = 0, u
t
(x, 0) = 0 for x .
Use Duhamels principle and an expansion of f in eigenfunctions to obtain a (formal)
solution.
Proof. a) We expand u in terms of the Dirichlet eigenfunctions of Laplacian in
.

n
+
n

n
= 0 in ,
n
= 0 on .
Assume
u(x, t) =

n=1
a
n
(t)
n
(x), a
n
(t) =
_

n
(x)u(x, t) dx.
f(x, t) =

n=1
f
n
(t)
n
(x), f
n
(t) =
_

n
(x)f(x, t) dx.
a

n
(t) =
_

n
(x)u
tt
dx =
_

n
(u +f) dx =
_

n
u dx +
_

n
f dx
=
_

n
u dx +
_

n
f dx =
n
_

n
u dx +
_

n
f dx
. .
fn
=
n
a
n
(t) + f
n
(t).
a
n
(0) =
_

n
(x)u(x, 0) dx = 0.
a

n
(0) =
_

n
(x)u
t
(x, 0) dx = 0.
77
Thus, we have an ODE which is converted and solved by Duhamels principle:

n
+
n
a
n
= f
n
(t)
a
n
(0) = 0
a

n
(0) = 0

n
+
n
a
n
= 0
a
n
(0, s) = 0
a

n
(0, s) = f
n
(s)
a
n
(t) =
_
t
0
a
n
(t s, s) ds.
With the anzats a
n
(t, s) = c
1
cos

n
t +c
2
sin

n
t, we get c
1
= 0, c
2
= f
n
(s)/

n
,
or
a
n
(t, s) = f
n
(s)
sin

n
t

n
.
Duhamels principle gives
a
n
(t) =
_
t
0
a
n
(t s, s) ds =
_
t
0
f
n
(s)
sin(

n
(t s))

n
ds.
u(x, t) =

n=1

n
(x)

n
_
t
0
f
n
(s) sin(
_

n
(t s)) ds.
77
We used Greens formula:
_

_
n
u
n
u
n
n
_
ds =
_

## (nu nu) dx.

On , u = 0; n = 0 since eigenfunctions are Dirichlet.
Partial Dierential Equations Igor Yanovsky, 2005 343
Problem (F90, #3). Consider the initial-boundary value problem

u
tt
= a(t)u
xx
+ f(x, t) 0 x , t 0
u(0, t) = u(, t) = 0 t 0
u(x, 0) = g(x), u
t
(x, 0) = h(x) 0 x ,
where the coecient a(t) ,= 0.
a) Express (formally) the solution of this problem by the method of eigenfunction ex-
pansions.
b) Show that this problem is not well-posed if a 1.
Hint: Take f = 0 and prove that the solution does not depend continuously on the
initial data g, h.
Proof. a) We expand u in terms of the Dirichlet eigenfunctions of Laplacian in
.

nxx
+
n

n
= 0 in ,
n
(0) =
n
() = 0.
That gives us the eigenvalues and eigenfunctions of the Laplacian:
n
= n
2
,
n
(x) =
sinnx.
Assume
u(x, t) =

n=1
u
n
(t)
n
(x), u
n
(t) =
_

n
(x)u(x, t) dx.
f(x, t) =

n=1
f
n
(t)
n
(x), f
n
(t) =
_

n
(x)f(x, t) dx.
g(x) =

n=1
g
n

n
(x), g
n
=
_

n
(x)g(x) dx.
h(x) =

n=1
h
n

n
(x), h
n
=
_

n
(x)h(x) dx.
u

n
(t) =
_

n
(x)u
tt
dx =
_

n
(a(t)u
xx
+ f) dx = a(t)
_

n
u
xx
dx +
_

n
f dx
= a(t)
_

nxx
u dx +
_

n
f dx =
n
a(t)
_

n
u dx +
_

n
f dx
. .
fn
=
n
a(t)u
n
(t) +f
n
(t).
u
n
(0) =
_

n
(x)u(x, 0) dx =
_

n
(x)g(x) dx = g
n
.
u

n
(0) =
_

n
(x)u
t
(x, 0) dx =
_

n
(x)h(x) dx = h
n
.
Thus, we have an ODE which is converted and solved by Duhamels principle:

n
+
n
a(t)u
n
= f
n
(t)
u
n
(0) = g
n
u

n
(0) = h
n
.

## Partial Dierential Equations Igor Yanovsky, 2005 344

Note: The initial data is not 0; therefore, the Duhamels principle is not applicable.
Also, the ODE is not linear in t, and its solution is not obvious. Thus,
u(x, t) =

n=1
u
n
(t)
n
(x),
where u
n
(t) are solutions of .
Partial Dierential Equations Igor Yanovsky, 2005 345
b) Assume we have two solutions, u
1
and u
2
, to the PDE:

u
1tt
+u
1xx
= 0,
u
1
(0, t) = u
1
(, t) = 0,
u
1
(x, 0) = g
1
(x), u
1t
(x, 0) = h
1
(x);

u
2tt
+ u
2xx
= 0,
u
2
(0, t) = u
2
(, t) = 0,
u
2
(x, 0) = g
2
(x), u
2t
(x, 0) = h
2
(x).
Note that the equation is elliptic, and therefore, the maximum principle holds.
In order to prove that the solution does not depend continuously on the initial data
g, h, we need to show that one of the following conditions holds:
max

[u
1
u
2
[ > max

[g
1
g
2
[,
max

[u
t1
u
t2
[ > max

[h
1
h
2
[.
That is, the dierence of the two solutions is not bounded by the dierence of initial
data.
By the method of separation of variables, we may obtain
u(x, t) =

n=1
(a
n
cos nt +b
n
sinnt) sinnx,
u(x, 0) =

n=1
a
n
sinnx = g(x),
u
t
(x, 0) =

n=1
nb
n
sinnx = h(x).
Not complete.
We also know that for elliptic equations, and for Laplace equation in particular, the
value of the function u has to be prescribed on the entire boundary, i.e. u = g on
, which is not the case here, making the problem under-determined. Also, u
t
is
prescribed on one of the boundaries, making the problem overdetermined.
Partial Dierential Equations Igor Yanovsky, 2005 346
29 Problems: Eigenvalues of the Laplacian - Heat
The ND HEAT Equation (eigenvalues/eigenfunctions of the Laplacian).
Consider the initial value problem with homogeneous Dirichlet condition:

u
t
= u for x , t > 0
u(x, 0) = g(x) for x
u(x, t) = 0 for x , t > 0.
Proof. For g C
2
() with g = 0 on , we have eigenfunction expansion
g(x) =

n=1
a
n

n
(x)
Assume the solution u(x, t) may be expanded in the eigenfunctions with coecients
depending on t: u(x, t) =

n=1
u
n
(t)
n
(x). This implies

n=1
u

n
(t)
n
(x) =
n

n=1
u
n
(t)
n
(x),
u

n
(t) +
n
u
n
(t) = 0, which has the general solution
u
n
(t) = A
n
e
nt
. Thus,
u(x, t) =

n=1
A
n
e
nt

n
(x),
u(x, 0) =

n=1
A
n

n
(x) = g(x).
Comparing with , we obtain A
n
= a
n
. Thus, the solution is given by
u(x, t) =

n=1
a
n
e
nt

n
(x),
with a
n
=
_

g(x)
n
(x) dx.
Also
u(x, t) =

n=1
a
n
e
nt

n
(x) =

n=1
_
_

g(y)
n
(y) dy
_
e
nt

n
(x)
=
_

n=1
e
nt

n
(x)
n
(y)
. .
K(x,y,t), heat kernel
g(y) dy
Partial Dierential Equations Igor Yanovsky, 2005 347
The 2D HEAT Equation (eigenvalues/eigenfunctions of the Laplacian).
Let = (0, a) (0, b) and consider

u
t
= u
xx
+u
yy
for x , t > 0
u(x, 0) = g(x) for x
u(x, t) = 0 for x , t > 0.
(29.1)
Proof. First, we nd eigenvalues/eigenfunctions of the Laplacian.

u
xx
+ u
yy
+ u = 0 in
u(0, y) = 0 = u(a, y) for 0 y b,
u(x, 0) = 0 = u(x, b) for 0 x a.
Let u(x, y) = X(x)Y (y), then substitution in the PDE gives X

Y +XY

+XY = 0.
X

X
+
Y

Y
+ = 0.
Letting =
2
+
2
and using boundary conditions, we nd the equations for X and
Y :
X

+
2
X = 0 Y

+
2
Y = 0
X(0) = X(a) = 0 Y (0) = Y (b) = 0.
The solutions of these one-dimensional eigenvalue problems are

m
=
m
a

n
=
n
b
X
m
(x) = sin
mx
a
Y
n
(y) = sin
ny
b
,
where m, n = 1, 2, . . . . Thus we obtain eigenvalues and normalized eigenfunctions of
the Laplacian:

mn
=
2
_
m
2
a
2
+
n
2
b
2
_

mn
(x, y) =
2

ab
sin
mx
a
sin
ny
b
,
where m, n = 1, 2, . . . .
Second, we solve the Heat Equation (29.1) using the space eigenfunctions.
For g C
2
() with g = 0 on , we have eigenfunction expansion
g(x) =

n=1
a
n

n
(x).
Assume u(x, t) =

n=1
u
n
(t)
n
(x). This implies
u

n
(t) +
n
u
n
(t) = 0, which has the general solution
u
n
(t) = A
n
e
nt
. Thus,
u(x, t) =

n=1
A
n
e
nt

n
(x),
u(x, 0) =

n=1
A
n

n
(x) = g(x).
Partial Dierential Equations Igor Yanovsky, 2005 348
Comparing with , we obtain A
n
= a
n
. Thus, the solution is given by
u(x, t) =

m,n=1
a
mn
e
mnt

mn
(x),
with
mn
,
mn
given above and a
mn
=
_

g(x)
mn
(x) dx.
Partial Dierential Equations Igor Yanovsky, 2005 349
Problem (S91, #2). Consider the heat equation
u
t
= u
xx
+ u
yy
on the square = 0 x 2, 0 y 2 with
periodic boundary conditions and with initial data
u(0, x, y) = f(x, y).
a) Find the solution using separation of variables.
Proof. First, we nd eigenvalues/eigenfunctions of the Laplacian.

u
xx
+ u
yy
+ u = 0 in
u(0, y) = u(2, y) for 0 y 2,
u(x, 0) = u(x, 2) for 0 x 2.
Let u(x, y) = X(x)Y (y), then substitution in the PDE gives X

Y +XY

+XY = 0.
X

X
+
Y

Y
+ = 0.
Letting =
2
+
2
and using periodic BCs, we nd the equations for X and Y :
X

+
2
X = 0 Y

+
2
Y = 0
X(0) = X(2) Y (0) = Y (2).
The solutions of these one-dimensional eigenvalue problems are

m
= m
n
= n
X
m
(x) = e
imx
Y
n
(y) = e
iny
,
where m, n = . . . , 2, 1, 0, 1, 2, . . . . Thus we obtain eigenvalues and normalized eigen-
functions of the Laplacian:

mn
= m
2
+n
2

mn
(x, y) = e
imx
e
iny
,
where m, n = . . . , 2, 1, 0, 1, 2, . . . .
Second, we solve the Heat Equation using the space eigenfunctions.
Assume u(x, y, t) =

m,n=
u
mn
(t)e
imx
e
iny
. This implies
u

mn
(t) + (m
2
+ n
2
)u
mn
(t) = 0, which has the general solution
u
n
(t) = c
mn
e
(m
2
+n
2
)t
. Thus,
u(x, y, t) =

m,n=
c
mn
e
(m
2
+n
2
)t
e
imx
e
iny
.
Partial Dierential Equations Igor Yanovsky, 2005 350
u(x, y, 0) =

m,n=
c
mn
e
imx
e
iny
= f(x, y),
_
2
0
_
2
0
f(x, y)e
imx
e
iny
dxdy =
_
2
0
_
2
0

m,n=
c
mn
e
imx
e
iny
e
im

x
e
in

y
dxdy
= 2
_
2
0

n=
c
mn
e
iny
e
in

y
dy = 4
2
c
mn
.
c
mn
=
1
4
2
_
2
0
_
2
0
f(x, y)e
imx
e
iny
dxdy = f
mn
.
Partial Dierential Equations Igor Yanovsky, 2005 351
b) Show that the integral
_

u
2
(x, y, t) dxdy is decreasing in t, if f is not constant.
Proof. We have
u
t
= u
xx
+ u
yy
Multiply the equation by u and integrate:
uu
t
= uu,
1
2
d
dt
u
2
= uu,
1
2
d
dt
_

u
2
dxdy =
_

uu dxdy =
_

u
u
n
ds
. .
=0, (periodic BC)

[u[
2
dxdy
=
_

[u[
2
dxdy 0.
Equality is obtained only when u = 0 u = constant f = constant.
If f is not constant,
_

u
2
dxdy is decreasing in t.
Partial Dierential Equations Igor Yanovsky, 2005 352
Problem (F98, #3). Consider the eigenvalue problem
d
2

dx
2
+ = 0,
(0)
d
dx
(0) = 0, (1) +
d
dx
(1) = 0.
a) Show that all eigenvalues are positive.
b) Show that there exist a sequence of eigenvalues =
n
, each of which satises
tan

=
2

1
.
c) Solve the following initial-boundary value problem on 0 < x < 1, t > 0
u
t
=

2
u
x
2
,
u(0, t)
u
x
(0, t) = 0, u(1, t) +
u
x
(1, t) = 0,
u(x, 0) = f(x).
You may call the relevant eigenfunctions
n
(x) and assume that they are known.
Proof. a) If = 0, the ODE reduces to

## = 0. Try (x) = Ax +B.

From the rst boundary condition,
(0)

(0) = 0 = B A B = A.
Thus, the solution takes the form (x) = Ax+A. The second boundary condition gives
(1) +

(1) = 0 = 3A A = B = 0.
Thus the only solution is 0, which is not an eigenfunction, and 0 not an eigenvalue.

## If < 0, try (x) = e

sx
, which gives s =

= R.
Hence, the family of solutions is (x) = Ae
x
+Be
x
. Also,

(x) = Ae
x
Be
x
.
The boundary conditions give
(0)

## (0) = 0 = A+ B A+ B = A(1 ) + B(1 +), (29.2)

(1) +

(1) = 0 = Ae

+Be

+Ae

Be

= Ae

(1+) +Be

(1). (29.3)
From (29.2) and (29.3) we get
1 +
1
=
A
B
and
1 +
1
=
B
A
e
2
, or
A
B
= e

.
From (29.2), =
A+ B
A B
and thus,
A
B
= e
A+B
BA
, which has no solutions.
b) Since > 0, the anzats = e
sx
gives s = i

the form
(x) = Asin(x

) +B cos(x

).
Then,

(x) = A

cos(x

) B

sin(x

## ). The rst boundary condition gives

(0)

(0) = 0 = B A

B = A

.
Partial Dierential Equations Igor Yanovsky, 2005 353
Hence, (x) = Asin(x

) + A

cos(x

(1) +

(1) = 0 = Asin(

) + A

cos(

) + A

cos(

) A sin(

)
= A
_
(1 ) sin(

) + 2

cos(

(1 ) sin(

) = 2

cos(

## ), and thus tan(

) =
2

1
.
c) We may assume that the eigenvalues/eigenfunctins of the Laplacian,
n
and
n
(x),
are known. We solve the Heat Equation using the space eigenfunctions.

u
t
= u
xx
,
u(0, t) u
x
(0, t) = 0, u(1, t) + u
x
(1, t) = 0,
u(x, 0) = f(x).
For f, we have an eigenfunction expansion
f(x) =

n=1
a
n

n
(x).
Assume u(x, t) =

n=1
u
n
(t)
n
(x). This implies
u

n
(t) +
n
u
n
(t) = 0, which has the general solution
u
n
(t) = A
n
e
nt
. Thus,
u(x, t) =

n=1
A
n
e
nt

n
(x),
u(x, 0) =

n=1
A
n

n
(x) = f(x).
Comparing with , we have A
n
= a
n
. Thus, the solution is given by
u(x, t) =

n=1
a
n
e
nt

n
(x),
with
a
n
=
_
1
0
f(x)
n
(x) dx.
Partial Dierential Equations Igor Yanovsky, 2005 354
Problem (W03, #3); 266B Ralston Hw. Let be a smooth domain in three
dimensions and consider the initial-boundary value problem for the heat equation

u
t
= u + f(x) for x , t > 0
u/n = 0 for x , t > 0
u(x, 0) = g(x) for x ,
in which f and g are known smooth functions with
g/n = 0 for x .
a) Find an approximate formula for u as t .
Proof. We expand u in terms of the Neumann eigenfunctions of Laplacian in .

n
+
n

n
= 0 in ,

n
n
= 0 on .
Note that here
1
= 0 and
1
is the constant V
1/2
, where V is the volume of .
Assume
u(x, t) =

n=1
a
n
(t)
n
(x), a
n
(t) =
_

n
(x)u(x, t) dx.
f(x) =

n=1
f
n

n
(x), f
n
=
_

n
(x)f(x) dx.
g(x) =

n=1
g
n

n
(x), g
n
=
_

n
(x)g(x) dx.
a

n
(t) =
_

n
(x)u
t
dx =
_

n
(u + f) dx =
_

n
u dx +
_

n
f dx
=
_

n
u dx +
_

n
f dx =
n
_

n
u dx +
_

n
f dx
. .
fn
=
n
a
n
+f
n
.
a
n
(0) =
_

n
(x)u(x, 0) dx =
_

n
g dx = g
n
.
78
Thus, we solve the ODE:
_
a

n
+
n
a
n
= f
n
a
n
(0) = g
n
.
For n = 1,
1
= 0, and we obtain a
1
(t) = f
1
t +g
1
.
For n 2, the homogeneous solution is a
n
h
= ce
nt
. The anzats for a particular
solution is a
np
= c
1
t + c
2
, which gives c
1
= 0 and c
2
= f
n
/
n
. Using the initial
condition, we obtain
a
n
(t) =
_
g
n

f
n

n
_
e
nt
+
f
n

n
.
78
We used Greens formula:
_

_
n
u
n
u
n
n
_
ds =
_

## (nu nu) dx.

On ,
u
n
= 0;
n
n
= 0 since eigenfunctions are Neumann.
Partial Dierential Equations Igor Yanovsky, 2005 355
u(x, t) = (f
1
t +g
1
)
1
(x) +

n=2
__
g
n

f
n

n
_
e
nt
+
f
n

n
_

n
(x).
If f
1
= 0
_
_

f(x) dx = 0
_
, lim
t
u(x, t) = g
1

1
+

n=2
f
n

n
.
If f
1
,= 0
_
_

f(x) dx ,= 0
_
, lim
t
u(x, t) f
1

1
t.
Partial Dierential Equations Igor Yanovsky, 2005 356
b) If g 0 and f > 0, show that u > 0 for all t > 0.
Partial Dierential Equations Igor Yanovsky, 2005 357
Problem (S97, #2). a) Consider the eigenvalue problem for the Laplace operator
in R
2
with zero Neumann boundary condition
_
u
xx
+ u
yy
+ u = 0 in
u
n
= 0 on .
Prove that
0
= 0 is the lowest eigenvalue and that it is simple.
b) Assume that the eigenfunctions
n
(x, y) of the problem in (a) form a complete
orthogonal system, and that f(x, y) has a uniformly convergent expansion
f(x, y) =

n=0
f
n

n
(x, y).
Solve the initial value problem
u
t
= u +f(x, y)
subject to initial and boundary conditions
u(x, y, 0) = 0,
u
n
u[

= 0.
What is the behavior of u(x, y, t) as t ?
c) Consider the problem with Neumann boundary conditions
_
v
xx
+v
yy
+ f(x, y) = 0 in
v
n
v = 0 on .
When does a solution exist? Find this solution, and nd its relation with the behavior
of limu(x, y, t) in (b) as t .
Proof. a) Suppose this eigenvalue problem did have a solution u with 0.
Multiplying u +u = 0 by u and integrating over , we get
_

uu dx +
_

u
2
dx = 0,
_

u
u
n
..
=0
ds
_

[u[
2
dx +
_

u
2
dx = 0,
_

[u[
2
dx =
..
0
_

u
2
dx,
Thus, u = 0 in , and u is constant in . Hence, we now have
0 =
..
0
_

u
2
dx.
For nontrivial u, we have = 0. For this eigenvalue problem, = 0 is an eigenvalue,
its eigenspace is the set of constants, and all other s are positive.
Partial Dierential Equations Igor Yanovsky, 2005 358
b) We expand u in terms of the Neumann eigenfunctions of Laplacian in .
79

n
+
n

n
= 0 in ,

n
n
= 0 on .
u(x, y, t) =

n=1
a
n
(t)
n
(x, y), a
n
(t) =
_

n
(x, y)u(x, y, t) dx.
a

n
(t) =
_

n
(x, y)u
t
dx =
_

n
(u + f) dx =
_

n
u dx +
_

n
f dx
=
_

n
u dx +
_

n
f dx =
n
_

n
u dx +
_

n
f dx
. .
fn
=
n
a
n
+f
n
.
a
n
(0) =
_

n
(x, y)u(x, y, 0) dx = 0.
80
Thus, we solve the ODE:
_
a

n
+
n
a
n
= f
n
a
n
(0) = 0.
For n = 1,
1
= 0, and we obtain a
1
(t) = f
1
t.
For n 2, the homogeneous solution is a
n
h
= ce
nt
. The anzats for a particular
solution is a
np
= c
1
t + c
2
, which gives c
1
= 0 and c
2
= f
n
/
n
. Using the initial
condition, we obtain
a
n
(t) =
f
n

n
e
nt
+
f
n

n
.
u(x, t) = f
1

1
t +

n=2
_

f
n

n
e
nt
+
f
n

n
_

n
(x).
If f
1
= 0
_
_

f(x) dx = 0
_
, lim
t
u(x, t) =

n=2
f
n

n
.
If f
1
,= 0
_
_

f(x) dx ,= 0
_
, lim
t
u(x, t) f
1

1
t.
c) Integrate v + f(x, y) = 0 over :
_

f dx =
_

v dx =
_

v dx =
1

v
n
ds =
2
0,
where we used
1
divergence theorem and
2
Neumann boundary conditions. Thus, the
solution exists only if
_

f dx = 0.
79
We use dxdy dx.
80
We used Greens formula:
_

_
n
u
n
u
n
n
_
ds =
_

## (nu nu) dx.

On ,
u
n
= 0;
n
n
= 0 since eigenfunctions are Neumann.
Partial Dierential Equations Igor Yanovsky, 2005 359
Assume v(x, y) =

n=0
a
n

n
(x, y). Since we have f(x, y) =

n=0
f
n

n
(x, y), we
obtain

n=0

n
a
n

n
+

n=0
f
n

n
= 0,

n
a
n

n
+ f
n

n
= 0,
a
n
=
f
n

n
.
v(x, y) =

n=0
(
fn
n
)
n
(x, y).
Partial Dierential Equations Igor Yanovsky, 2005 360
29.1 Heat Equation with Periodic Boundary Conditions in 2D
(with extra terms)
Problem (F99, #5). In two spatial dimensions, consider the dierential equation
u
t
= u
2
u
with periodic boundary conditions on the unit square [0, 2]
2
.
a) If = 2 nd a solution whose amplitude increases as t increases.
b) Find a value
0
, so that the solution of this PDE stays bounded as t , if <
0
.
Proof. a) Eigenfunctions of the Laplacian.
The periodic boundary conditions imply a Fourier Series solution of the form:
u(x, t) =

m,n
a
mn
(t)e
i(mx+ny)
.
u
t
=

m,n
a

mn
(t)e
i(mx+ny)
,
u = u
xx
+ u
yy
=

m,n
(m
2
+n
2
) a
mn
(t)e
i(mx+ny)
,

2
u = u
xxxx
+ 2u
xxyy
+ u
yyyy
=

m,n
(m
4
+ 2m
2
n
2
+ n
4
) a
mn
(t)e
i(mx+ny)
=

m,n
(m
2
+ n
2
)
2
a
mn
(t)e
i(mx+ny)
.
Plugging this into the PDE, we obtain
a

mn
(t) = (m
2
+ n
2
)a
mn
(t) (m
2
+n
2
)
2
a
mn
(t),
a

mn
(t) [(m
2
+n
2
) (m
2
+ n
2
)
2
]a
mn
(t) = 0,
a

mn
(t) (m
2
+ n
2
)[ (m
2
+ n
2
)]a
mn
(t) = 0.
The solution to the ODE above is
a
mn
(t) =
mn
e
(m
2
+n
2
)[(m
2
+n
2
)]t
.
u(x, t) =

m,n

mn
e
(m
2
+n
2
)[(m
2
+n
2
)]t
e
i(mx+ny)
. .
oscillates
.
When = 2, we have
u(x, t) =

m,n

mn
e
(m
2
+n
2
)[2(m
2
+n
2
)]t
e
i(mx+ny)
.
We need a solution whose amplitude increases as t increases. Thus, we need those

mn
> 0, with
(m
2
+ n
2
)[2 (m
2
+n
2
)] > 0,
2 (m
2
+ n
2
) > 0,
2 > m
2
+n
2
.
Hence,
mn
> 0 for (m, n) = (0, 0), (m, n) = (1, 0), (m, n) = (0, 1).
Else,
mn
= 0. Thus,
u(x, t) =
00
+
10
e
t
e
ix
+
01
e
t
e
iy
= 1 +e
t
e
ix
+e
t
e
iy
= 1 +e
t
(cos x + i sinx) + e
t
(cos y +i siny).
Partial Dierential Equations Igor Yanovsky, 2005 361
b) For
0
= 1, the solution stays bounded as t .
Partial Dierential Equations Igor Yanovsky, 2005 362
Problem (F93, #1).
Suppose that a and b are constants with a 0, and consider the equation
u
t
= u
xx
+u
yy
au
3
+ bu (29.4)
in which u(x, y, t) is 2-periodic in x and y.
a) Let u be a solution of (29.4) with
[[u(t = 0)[[ =
_
2
0
_
2
0
[u(x, y, t = 0)[
2
dxdy
1/2
< .
Derive an explicit bound on [[u(t)[[ and show that it stays nite for all t.
b) If a = 0, construct the normal modes for (29.4); i.e. nd all solutions of the form
u(x, y, t) = e
t+ikx+ily
.
c) Use these normal modes to construct a solution of (29.4) with a = 0 for the initial
data
u(x, y, t = 0) =
1
1
1
2
e
ix
+
1
1
1
2
e
ix
.
Proof. a) Multiply the equation by u and integrate:
u
t
= u au
3
+bu,
uu
t
= uu au
4
+bu
2
,
_

uu
t
dx =
_

uu dx
_

au
4
dx +
_

bu
2
dx,
1
2
d
dt
_

u
2
dx =
_

u
u
n
ds
. .
=0, u periodic on [0,2]
2

[u[
2
dx
_

au
4
dx
. .
0
+
_

bu
2
dx,
d
dt
[[u[[
2
2
2b [[u[[
2
2
,
[[u[[
2
2
[[u(x, 0)[[
2
2
e
2bt
,
[[u[[
2
[[u(x, 0)[[
2
e
bt
e
bt
.
Thus, [[u[[ stays nite for all t.
b) Since a = 0, plugging u = e
t+ikx+ily
into the equation, we obtain:
u
t
= u
xx
+ u
yy
+ bu,
e
t+ikx+ily
= (k
2
l
2
+b) e
t+ikx+ily
,
= k
2
l
2
+ b.
Thus,
u
kl
= e
(k
2
l
2
+b)t+ikx+ily
,
u(x, y, t) =

k,l
a
kl
e
(k
2
l
2
+b)t+ikx+ily
.
Partial Dierential Equations Igor Yanovsky, 2005 363
c) Using the initial condition, we obtain:
u(x, y, 0) =

k,l
a
kl
e
i(kx+ly)
=
1
1
1
2
e
ix
+
1
1
1
2
e
ix
=

k=0
_
1
2
e
ix
_
k
+

k=0
_
1
2
e
ix
_
k
=

k=0
1
2
k
e
ikx
+

k=0
1
2
k
e
ikx
,
= 2 +

k=1
1
2
k
e
ikx
+

k=1
1
2
k
e
ikx
.
Thus, l = 0, and we have

k=
a
k
e
ikx
= 2 +

k=1
1
2
k
e
ikx
+

k=1
1
2
k
e
ikx
,
a
0
= 2; a
k
=
1
2
k
, k > 0; a
k
=
1
2
k
, k < 0
a
0
= 2; a
k
=
1
2
|k|
, k ,= 0.
u(x, y, t) = 2e
bt
+
+

k=, k=0
1
2
|k|
e
(k
2
+b)t+ikx
.
81
81
Note a similar question formulation in F92 #3(b).
Partial Dierential Equations Igor Yanovsky, 2005 364
Problem (S00, #3). Consider the initial-boundary value problem for u = u(x, y, t)
u
t
= u u
for (x, y) [0, 2]
2
, with periodic boundary conditions and with
u(x, y, 0) = u
0
(x, y)
in which u
0
is periodic. Find an asymptotic expansion for u for t large with terms
tending to zero increasingly rapidly as t .
Proof. Since we have periodic boundary conditions, assume
u(x, y, t) =

m,n
u
mn
(t) e
i(mx+ny)
.
Plug this into the equation:

m,n
u

mn
(t) e
i(mx+ny)
=

m,n
(m
2
n
2
1) u
mn
(t) e
i(mx+ny)
,
u

mn
(t) = (m
2
n
2
1) u
mn
(t),
u
mn
(t) = a
mn
e
(m
2
n
2
1)t
,
u(x, y, t) =

m,n
a
mn
e
(m
2
+n
2
+1)t
e
i(mx+ny)
.
Since u
0
is periodic,
u
0
(x, y) =

m,n
u
0mn
e
i(mx+ny)
, u
0mn
=
1
4
2
_
2
0
_
2
0
u
0
(x, y) e
i(mx+ny)
dxdy.
Initial condition gives:
u(x, y, 0) =

m,n
a
mn
e
i(mx+ny)
= u
0
(x, y),

m,n
a
mn
e
i(mx+ny)
=

m,n
u
0mn
e
i(mx+ny)
,
a
mn
= u
0mn
.
u(x, y, t) =

m,n
u
0mn
e
(m
2
+n
2
+1)t
e
i(mx+ny)
.
u
0mn
e
(m
2
+n
2
+1)t
e
i(mx+ny)
0 as t , since e
(m
2
+n
2
+1)t
0 as t .
Partial Dierential Equations Igor Yanovsky, 2005 365
30 Problems: Fourier Transform
Problem (S01, #2b). Write the solution of initial value problem
U
t

_
1 0
5 3
_
U
x
= 0,
for general initial data
_
u
(1)
(x, 0)
u
(2)
(x, 0)
_
=
_
f(x)
0
_
as an inverse Fourier transform.
You may assume that f is smooth and rapidly decreasing as [x[ .
Proof. Consider the original system:
u
(1)
t
u
(1)
x
= 0,
u
(2)
t
5u
(1)
x
3u
(2)
x
= 0.
Take the Fourier transform in x. The transformed initial value problems are:
u
(1)
t
i u
(1)
= 0, u
(1)
(, 0) =

f(),
u
(2)
t
5i u
(1)
3i u
(2)
= 0, u
(2)
(, 0) = 0.
Solving the rst ODE for u
(1)
gives:
u
(1)
(, t) =

f()e
it
.
With this u
(1)
, the second initial value problem becomes
u
(2)
t
3i u
(2)
= 5i

f()e
it
, u
(2)
(, 0) = 0.
The homogeneous solution of the above ODE is:
u
(2)
h
(, t) = c
1
e
3it
.
With u
(2)
p
= c
2
e
it
as anzats for a particular solution, we obtain:
ic
2
e
it
3ic
2
e
it
= 5i

f()e
it
,
2ic
2
e
it
= 5i

f()e
it
,
c
2
=
5
2

f().
u
(2)
p
(, t) =
5
2

f()e
it
.
u
(2)
(, t) = u
(2)
h
(, t) + u
(2)
p
(, t) = c
1
e
3it

5
2

f()e
it
.
We nd c
1
using initial conditions:
u
(2)
(, 0) = c
1

5
2

f() = 0 c
1
=
5
2

f().
Thus,
u
(2)
(, t) =
5
2

f()
_
e
3it
e
it
_
.
Partial Dierential Equations Igor Yanovsky, 2005 366
u
(1)
(x, t) and u
(2)
(x, t) are be obtained by taking inverse Fourier transform:
u
(1)
(x, t) =
_
u
(1)
(, t)
_

=
1

2
_
R
n
e
ix

f() e
it
d,
u
(2)
(x, t) =
_
u
(2)
(, t)
_

=
1

2
_
R
n
e
ix
5
2

f()
_
e
3it
e
it
_
d.
Partial Dierential Equations Igor Yanovsky, 2005 367
Problem (S02, #4). Use the Fourier transform on L
2
(R) to show that
du
dx
+ cu(x) +u(x 1) = f (30.1)
has a unique solution u L
2
(R) for each f L
2
(R) when [c[ > 1 - you may assume
that c is a real number.
Proof. u L
2
(R). Dene its Fourier transform u by
u() =
1

2
_
R
e
ix
u(x) dx for R.

du
dx
() = i u().
We can nd

u(x 1)() in two ways.
Let u(x 1
. .
y
) = v(x), and determinte v():

## u(x 1)() = v() =

1

2
_
R
e
ix
v(x) dx =
1

2
_
R
e
i(y+1)
u(y) dy
=
1

2
_
R
e
iy
e
i
u(y) dy = e
i
u().
We can also write the denition for u() and substitute x 1 later in calculations:
u() =
1

2
_
R
e
iy
u(y) dy =
1

2
_
R
e
i(x1)
u(x 1) dx
=
1

2
_
R
e
ix
e
i
u(x 1) dx = e
i

u(x 1)(),

u(x 1)() = e
i
u().
Substituting into (30.1), we obtain
i u() +c u() + e
i
u() =

f(),
u() =

f()
i +c +e
i
.
u(x) =
_

f()
i + c + e
i
_

=
_

f

B
_

=
1

2
f B,
where

B =
1
i +c + e
i
,
B =
_
1
i +c +e
i
_

=
1

2
_
R
e
ix
i + c + e
i
d.
For [c[ > 1, u() exists for all R, so that u(x) = ( u())

## and this is unique by the

Fourier Inversion Theorem.
Note that in R
n
, becomes

## u(x 1)() = v() =

1
(2)
n
2
_
R
n
e
ix
v(x) dx =
1
(2)
n
2
_
R
n
e
i(y+1)
u(y) dy
=
1
(2)
n
2
_
R
n
e
iy
e
i

1
u(y) dy = e
i

1
u() = e
(i

j

j
)
u().
Partial Dierential Equations Igor Yanovsky, 2005 368
Problem (F96, #3). Find the fundamental solution for the equation
u
t
= u
xx
xu
x
. (30.2)
Hint: The Fourier transform converts this problem into a PDE which can be solved
using the method of characteristics.
Proof. u L
2
(R). Dene its Fourier transform u by
u() =
1

2
_
R
e
ix
u(x) dx for R.
u
x
() = i u(),
u
xx
() = (i)
2
u() =
2
u().
We nd xu
x
() in two steps:
Multiplication by x:

ixu() =
1

2
_
R
e
ix
_
ixu(x)
_
dx =
d
d
u().

xu(x)() = i
d
d
u().
Using the previous result, we nd:

xu
x
(x)() =
1

2
_
R
e
ix
_
xu
x
(x)
_
dx =
1

2
_
e
ix
xu
_

. .
=0

2
_
R
_
(i)e
ix
x +e
ix
_
u dx
=
1

2
i
_
R
e
ix
x u dx
1

2
_
R
e
ix
u dx
= i

xu(x)() u() = i
_
i
d
d
u()
_
u() =
d
d
u() u().

xu
x
(x)() =
d
d
u() u().
Plugging these into (30.2), we get:

t
u(, t) =
2
u(, t)
_

d
d
u(, t) u(, t)
_
,
u
t
=
2
u + u

+ u,
u
t
u

= (
2
1) u.
We now solve the above equation by characteristics.
We change the notation: u u, t y, x. We have
u
y
xu
x
= (x
2
1)u.
dx
dt
= x x = c
1
e
t
, (c
1
= xe
t
)
dy
dt
= 1 y = t +c
2
,
dz
dt
= (x
2
1)z = (c
2
1
e
2t
1)z
dz
z
= (c
2
1
e
2t
1)dt
log z =
1
2
c
2
1
e
2t
+t +c
3
=
x
2
2
+t +c
3
=
x
2
2
+y c
2
+ c
3
z = ce
x
2
2
+y
.
Partial Dierential Equations Igor Yanovsky, 2005 369
Changing the notation back, we have
u(, t) = ce

2
2
+t
.
Thus, we have
u(, t) = ce

2
2
+t
.
We use Inverse Fourier Tranform to get u(x, t):
82
u(x, t) =
1

2
_
R
e
ix
u(, t) d =
1

2
_
R
e
ix
ce

2
2
+t
d
=
c

2
e
t
_
R
e
ix
e

2
2
d =
c

2
e
t
_
R
e
ix+

2
2
d
=
c

2
e
t
_
R
e
2ix+
2
2
d =
c

2
e
t
_
R
e
(+ix)
2
2
d e
x
2
2
=
c

2
e
t
e
x
2
2
_
R
e
y
2
2
dy =
c

2
e
t
e
x
2
2

2 = c e
t
e
x
2
2
.
u(x, t) = c e
t
e
x
2
2
.
Check:
u
t
= c e
t
e
x
2
2
,
u
x
= c e
t
xe
x
2
2
,
u
xx
= c e
t
_
e
x
2
2
+x
2
e
x
2
2
_
.
Thus,
u
t
= u
xx
xu
x
,
c e
t
e
x
2
2
= c e
t
_
e
x
2
2
+ x
2
e
x
2
2
_
x c e
t
xe
x
2
2
.
82
We complete the square for powers of exponentials.
Partial Dierential Equations Igor Yanovsky, 2005 370
Problem (W02, #4). a) Solve the initial value problem
u
t
+
n

k=1
a
k
(t)
u
x
k
+ a
0
(t)u = 0, x R
n
,
u(0, x) = f(x)
where a
k
(t), k = 1, . . . , n, and a
0
(t) are continuous functions, and f is a continuous
function. You may assume f has compact support.
b) Solve the initial value problem
u
t
+
n

k=1
a
k
(t)
u
x
k
+ a
0
(t)u = f(x, t), x R
n
,
u(0, x) = 0
where f is continuous in x and t.
Proof. a) Use the Fourier transform to solve this problem.
u(, t) =
1
(2)
n
2
_
R
n
e
ix
u(x, t) dx for R.

u
x
k
= i
k
u.
Thus, the equation becomes:
_
u
t
+ i

n
k=1
a
k
(t)
k
u +a
0
(t) u = 0,
u(, 0) =

f(),
or
u
t
+i a(t)

u +a
0
(t) u = 0,
u
t
=
_
i a(t)

+ a
0
(t)
_
u.
This is an ODE in u with solution:
u(, t) = ce

_
t
0
(i a(s)

+a
0
(s)) ds
, u(, 0) = c =

f(). Thus,
u(, t) =

f() e

_
t
0
(i a(s)

+a
0
(s)) ds
.
Use the Inverse Fourier transform to get u(x, t):
u(x, t) = u(, t)

=
_

f() e

_
t
0
(i a(s)

+a
0
(s)) ds
_

=
(f g)(x)
(2)
n
2
,
where g() = e

_
t
0
(i a(s)

+a
0
(s)) ds
.
g(x) =
1
(2)
n
2
_
R
n
e
ix
g() d =
1
(2)
n
2
_
R
n
e
ix
_
e

_
t
0
(i a(s)

+a
0
(s)) ds

d.
u(x, t) =
(f g)(x)
(2)
n
2
=
1
(2)
n
_
R
n
_
R
n
e
i(xy)
_
e

_
t
0
(i a(s)

+a
0
(s)) ds

d f(y) dy.
b) Use Duhamels Principle and the result from (a).
u(x, t) =
_
t
0
U(x, t s, s) ds, where U(x, t, s) solves
U
t
+
n

k=1
a
k
(t)
U
x
k
+a
0
(t)U = 0,
U(x, 0, s) = f(x, s).
Partial Dierential Equations Igor Yanovsky, 2005 371
u(x, t) =
_
t
0
U(x, t s, s) ds =
1
(2)
n
_
t
0
_
R
n
_
R
n
e
i(xy)
_
e

_
ts
0
(i a(s)

+a
0
(s)) ds

d f(y, s) dy ds.
Partial Dierential Equations Igor Yanovsky, 2005 372
Problem (S93, #2). a) Dene the Fourier transform
83

f() =
_

e
ix
f(x) dx.
State the inversion theorem. If

f() =

, [[ < a,
1
2
, [[ = a,
0, [[ > a,
where a is a real constant, what f(x) does the inversion theorem give?
b) Show that

f(x b) = e
ib

f(x),
where b is a real constant. Hence, using part (a) and Parsevals theorem, show that
1

## sin a(x +z)

x +z
sina(x + )
x +
dx =
sina(z )
z
,
where z and are real constants.
Proof. a) The inverse Fourier transform for f L
1
(R
n
):
f

() =
1
2
_

e
ix
f(x) dx for R.
Fourier Inversion Theorem: Assume f L
2
(R). Then
f(x) =
1
2
_

e
ix

f() d =
1
2
_

e
i(yx)
f(y) dy d = (

f)

(x).
Parsevals theorem (Plancherels theorem) (for this denition of the Fourier
transform). Assume f L
1
(R
n
) L
2
(R
n
). Then

f, f

L
2
(R
n
) and
1
2
[[

f[[
L
2
(R
n
)
= [[f

[[
L
2
(R
n
)
= [[f[[
L
2
(R
n
)
, or
_

[f(x)[
2
dx =
1
2
_

f()[
2
d.
Also,
_

f(x) g(x) dx =
1
2
_

f() g() d.
We can write

f() =
_
, [[ < a,
0, [[ > a.
83
Note that the Fourier transform is dened incorrectly here. There should be - sign in e
ix
.
Need to be careful, since the consequences of this denition propagate throughout the solution.
Partial Dierential Equations Igor Yanovsky, 2005 373
f(x) = (

f())

=
1
2
_

e
ix

f() d =
1
2
_
a

0 d +
1
2
_
a
a
e
ix
d +
1
2
_

a
0 d
=
1
2
_
a
a
e
ix
d =
1
2ix
_
e
ix
_
=a
=a
=
1
2ix
_
e
iax
e
iax
_
=
sinax
x
.
b) Let f(x b
. .
y
) = g(x), and determinte g():

## f(x b)() = g() =

_
R
e
ix
g(x) dx =
_
R
e
i(y+b)
f(y) dy
=
_
R
e
iy
e
ib
f(y) dy = e
ib

f().
With f(x) =
sinax
x
(from (a)), we have
1

sina(x +z)
x + z
sin a(x + s)
x +s
dx =
1

f(x + z)f(x + s) dx (x

= x + s, dx

= dx)
=
1

f(x

+z s)f(x

) dx

(Parsevals)
=
1

1
2
_

f(x

+z s)

f(x

) d part (b)
=
1
2
2
_

f() e
i(zs)
f() d
=
1
2
2
_
a
a

f()
2
e
i(zs)
d
=
1
2
2
_
a
a

2
e
i(zs)
d
=
1
2
_
a
a
e
i(zs)
d
=
1
2i(z s)
_
e
i(zs)

=a
=a
=
e
i(zs)a
e
i(zs)a
2i(z s)
=
sin a(z s)
z s
.
Partial Dierential Equations Igor Yanovsky, 2005 374
Problem (F03, #5). State Parsevals relation for Fourier transforms.
Find the Fourier transform

f() of
f(x) =
_
e
ix
/2

y, [x[ y
0, [x[ > y,
in which y and are constants.
Use this in Parsevals relation to show that
_

sin
2
( )y
( )
2
d = y.
What does the transform

f() become in the limit y ?
Use Parsevals relation to show that
sin( )y
( )
=
1

sin( )y
( )
sin( )y
( )
d.
Proof. f L
2
(R). Dene its Fourier transform u by

f() =
1

2
_
R
e
ix
f(x) dx for R.
Parsevals theorem (Plancherels theorem):
Assume f L
1
(R
n
) L
2
(R
n
). Then

f, f

L
2
(R
n
) and
[[

f[[
L
2
(R
n
)
= [[f

[[
L
2
(R
n
)
= [[f[[
L
2
(R
n
)
, or
_

[f(x)[
2
dx =
_

f()[
2
d.
Also,
_

f(x) g(x) dx =
_

f() g() d.
Find the Fourier transform of f:

f() =
1

2
_
R
e
ix
f(x) dx =
1

2
_
y
y
e
ix
e
ix
2

y
dx =
1
2

2y
_
y
y
e
i()x
dx
=
1
2

2y
1
i( )
_
e
i()x
_
x=y
x=y
=
1
2i

2y( )
_
e
i()y
e
i()y

=
siny( )

2y( )
.
Parsevals theorem gives:
_

f()[
2
d =
_

[f(x)[
2
dx,
_

sin
2
y( )

2
2y( )
2
d =
_
y
y

e
2ix

4y
dx,
_

sin
2
y( )
( )
2
d =

2
_
y
y
dx,
_

sin
2
y( )
( )
2
d = y.
Partial Dierential Equations Igor Yanovsky, 2005 375

f() =
siny( )

2y( )
.
We make change of variables: =

. Then, =

+ . We have

f() =

f(

+ ) =
siny(

)
(

)
, or

f( + ) =
siny( )
( )
.
We will also use the following result.
Let

f( + a
. .

## ) = g(), and determinte g()

f( +a)

= g()

=
1

2
_
R
e
ix
g() d =
1

2
_
R
e
ix(

a)

f(

) d

= e
ixa
f(x).
Using these results, we have
1

sin( )y
( )
sin( )y
( )
d =
1

(
_
2y)
2
_

f()

f( + ) d
= 2y
_

f(x) e
()ix
f(x) dx
= 2y
_

f(x)
2
e
()ix
dx
= 2y
_
y
y

e
2ix

4y
e
()ix
dx
=
1
2
_
y
y
e
()ix
dx
=
1
2i( )
_
e
()ix

x=y
x=y
=
1
2i( )
_
e
()iy
e
()iy

=
sin( )y

.
Partial Dierential Equations Igor Yanovsky, 2005 376
Problem (S95, #5). For the Laplace equation
f
_

2
x
2
+

2
y
2
_
f = 0 (30.3)
in the upper half plane y 0, consider
the Dirichlet problem f(x, 0) = g(x);
the Neumann problem

y
f(x, 0) = h(x).
Assume that f, g and h are 2 periodic in x and that f is bounded at innity.
Find the Fourier transform N of the Dirichlet-Neumann map. In other words,
nd an operator N taking the Fourier transform of g to the Fourier transform of h; i.e.
Ng
k
=

h
k
.
Proof. We solve the problem by two methods.
Fourier Series.
Since f is 2-periodic in x, we can write
f(x, y) =

n=
a
n
(y) e
inx
.
Plugging this into (30.3), we get the ODE:

n=
_
n
2
a
n
(y)e
inx
+a

n
(y)e
inx
_
= 0,
a

n
(y) n
2
a
n
(y) = 0.
Initial conditions give: (g and h are 2-periodic in x)
f(x, 0) =

n=
a
n
(0)e
inx
= g(x) =

n=
g
n
e
inx
a
n
(0) = g
n
.
f
y
(x, 0) =

n=
a

n
(0)e
inx
= h(x) =

n=

h
n
e
inx
a

n
(0) =

h
n
.
Thus, the problems are:
a

n
(y) n
2
a
n
(y) = 0,
a
n
(0) = g
n
, (Dirichlet)
a

n
(0) =

h
n
. (Neumann)
a
n
(y) = b
n
e
ny
+ c
n
e
ny
, n = 1, 2, . . . ; a
0
(y) = b
0
y +c
0
.
a

n
(y) = nb
n
e
ny
nc
n
e
ny
, n = 1, 2, . . . ; a

0
(y) = b
0
.
Since f is bounded at y = , we have:
b
n
= 0 for n > 0,
c
n
= 0 for n < 0,
b
0
= 0, c
0
arbitrary.
Partial Dierential Equations Igor Yanovsky, 2005 377
n > 0:
a
n
(y) = c
n
e
ny
,
a
n
(0) = c
n
= g
n
, (Dirichlet)
a

n
(0) = nc
n
=

h
n
. (Neumann)
ng
n
=

h
n
.
n < 0:
a
n
(y) = b
n
e
ny
,
a
n
(0) = b
n
= g
n
, (Dirichlet)
a

n
(0) = nb
n
=

h
n
. (Neumann)
ng
n
=

h
n
.
[n[g
n
=

h
n
, n ,= 0.
n = 0 : a
0
(y) = c
0
,
a
0
(0) = c
0
= g
0
, (Dirichlet)
a

0
(0) = 0 =

h
0
. (Neumann)
Note that solution f(x, y) may be written as
f(x, y) =

n=
a
n
(y) e
inx
= a
0
(y) +
1

n=
a
n
(y) e
inx
+

n=1
a
n
(y) e
inx
= c
0
+
1

n=
b
n
e
ny
e
inx
+

n=1
c
n
e
ny
e
inx
=
_
g
0
+

1
n=
g
n
e
ny
e
inx
+

n=1
g
n
e
ny
e
inx
, (Dirichlet)
c
0
+

1
n=

hn
n
e
ny
e
inx
+

n=1

hn
n
e
ny
e
inx
. (Neumann)
Fourier Transform. The Fourier transform of f(x, y) in x is:

f(, y) =
1

2
_

e
ix
f(x, y) dx,
f(x, y) =
1

2
_

e
ix

f(, y) d.
(i)
2

f(, y) +

f
y
y(, y) = 0,

f
yy

2

## f = 0. The solution to this ODE is:

f(, y) = c
1
e
y
+ c
2
e
y
.
For > 0, c
1
= 0; for < 0, c
2
= 0.
> 0 :

f(, y) = c
2
e
y
,

f
y
(, y) = c
2
e
y
,
c
2
=

f(, 0) =
1

2
_

e
ix
f(x, 0) dx =
1

2
_

e
ix
g(x) dx = g(), (Dirichlet)
c
2
=

f
y
(, 0) =
1

2
_

e
ix
f
y
(x, 0) dx =
1

2
_

e
ix
h(x) dx =

h(). (Neumann)
g() =

h().
Partial Dierential Equations Igor Yanovsky, 2005 378
< 0 :

f(, y) = c
1
e
y
,

f
y
(, y) = c
1
e
y
,
c
1
=

f(, 0) =
1

2
_

e
ix
f(x, 0) dx =
1

2
_

e
ix
g(x) dx = g(), (Dirichlet)
c
1
=

f
y
(, 0) =
1

2
_

e
ix
f
y
(x, 0) dx =
1

2
_

e
ix
h(x) dx =

h(). (Neumann)
g() =

h().
[[g() =

h().
Partial Dierential Equations Igor Yanovsky, 2005 379
Problem (F97, #3). Consider the Dirichlet problem in the half-space x
n
> 0,
n 2:
u + a
u
x
n
+k
2
u = 0, x
n
> 0
u(x

, 0) = f(x

), x

= (x
1
, . . . , x
n1
).
Here a and k are constants.
Use the Fourier transform to show that for any f(x

) L
2
(R
n1
) there exists a
solution u(x

, x
n
) of the Dirichlet problem such that
_
R
n
[u(x

, x
n
)[
2
dx

C
for all 0 < x
n
< +.
Proof.
84
Denote = (

,
n
). Transform in the rst n 1 variables:
[

[
2
u(

, x
n
) +

2
u
x
2
n
(

, x
n
) + a
u
x
n
(

, x
n
) + k
2
u(

, x
n
) = 0.
Thus, the ODE and initial conditions of the transformed problem become:
_
u
xnxn
+a u
xn
+ (k
2
[

[
2
) u = 0,
u(

, 0) =

f(

).
With the anzats u = ce
sxn
, we obtain s
2
+as + (k
2
[

[
2
) = 0, and
s
1,2
=
a
_
a
2
4(k
2
[

[
2
)
2
.
Choosing only the negative root, we obtain the solution:
85
u(

, x
n
) = c(

) e
a

a
2
4(k
2
|

|
2
)
2
xn
. u(

, 0) = c =

f(

). Thus,
u(

, x
n
) =

f(

) e
a

a
2
4(k
2
|

|
2
)
2
xn
.
Parsevals theorem gives:
[[u[[
2
L
2
(R
n1
)
= [[ u[[
2
L
2
(R
n1
)
=
_
R
n1
[ u(

, x
n
)[
2
d

=
_
R
n1

f(

) e
a

a
2
4(k
2
|

|
2
)
2
xn

2
d

_
R
n1

f(

2
d

= [[

f[[
2
L
2
(R
n1
)
= [[f[[
2
L
2
(R
n1
)
C,
since f(x

) L
2
(R
n1
). Thus, u(x

, x
n
) L
2
(R
n1
).
84
Note that the last element of x = (x

## , xn) = (x1, . . . , xn1, xn), i.e. xn, plays a role of time t.

As such, the PDE may be written as
u + utt + aut + k
2
u = 0.
85
Note that a > 0 should have been provided by the statement of the problem.
Partial Dierential Equations Igor Yanovsky, 2005 380
Problem (F89, #7). Find the following fundamental solutions
a)
G(x, y, t)
t
= a(t)

2
G(x, y, t)
x
2
+b(t)
G(x, y, t)
x
+ c(t)G(x, y, t) for t > 0
G(x, y, 0) = (x y),
where a(t), b(t), c(t) are continuous functions on [0, +], a(t) > 0 for t > 0.
b)
G
t
(x
1
, . . . , x
n
, y
1
, . . . , y
n
, t) =
n

k=1
a
k
(t)
G
x
k
for t > 0,
G(x
1
, . . . , x
n
, y
1
, . . . , y
n
, 0) = (x
1
y
1
)(x
2
y
2
) . . . (x
n
y
n
).
Proof. a) We use the Fourier transform to solve this problem.
Transform the equation in the rst variable only. That is,

G(, y, t) =
1

2
_
R
e
ix
G(x, y, t) dx.
The equation is transformed to an ODE, that can be solved:

G
t
(, y, t) = a(t)
2

G(, y, t) + i b(t)

G(, y, t) + c(t)

G(, y, t),

G
t
(, y, t) =
_
a(t)
2
+ i b(t) + c(t)

G(, y, t),

G(, y, t) = c e
_
t
0
[a(s)
2
+i b(s)+c(s)] ds
.
We can also transform the initial condition:

G(, y, 0) =

(x y)() = e
iy

() =
1

2
e
iy
.
Thus, the solution of the transformed problem is:

G(, y, t) =
1

2
e
iy
e
_
t
0
[a(s)
2
+i b(s)+c(s)] ds
.
The inverse Fourier transform gives the solution to the original problem:
G(x, y, t) =
_

G(, y, t)
_

=
1

2
_
R
e
ix

G(, y, t) d
=
1

2
_
R
e
ix
_
1

2
e
iy
e
_
t
0
[a(s)
2
+i b(s)+c(s)] ds
_
d
=
1
2
_
R
e
i(xy)
e
_
t
0
[a(s)
2
+i b(s)+c(s)] ds
d.
b) Denote x = (x
1
, . . . , x
n
), y = (y
1
, . . . , y
n
). Transform in x:

G(

, y, t) =
1
(2)
n
2
_
R
n
e
i x

G(x, y, t) dx.
The equation is transformed to an ODE, that can be solved:

G
t
(

, y, t) =
n

k=1
a
k
(t) i
k

G(

, y, t),

G(

, y, t) = c e
i
_
t
0
[

n
k=1
a
k
(s)
k
] ds
.
Partial Dierential Equations Igor Yanovsky, 2005 381
We can also transform the initial condition:

G(

, y, 0) =
_
(x
1
y
1
)(x
2
y
2
) . . . (x
n
y
n
)

() = e
i y

) =
1
(2)
n
2
e
i y

.
Thus, the solution of the transformed problem is:

G(

, y, t) =
1
(2)
n
2
e
i y

e
i
_
t
0
[

n
k=1
a
k
(s)
k
] ds
.
The inverse Fourier transform gives the solution to the original problem:
G(x, y, t) =
_

G(

, y, t)
_

=
1
(2)
n
2
_
R
n
e
i x

G(

, y, t) d

=
1
(2)
n
2
_
R
n
e
i x

_
1
(2)
n
2
e
i y

e
i
_
t
0
[

n
k=1
a
k
(s)
k
] ds
_
d

=
1
(2)
n
_
R
n
e
i(xy)

e
i
_
t
0
[

n
k=1
a
k
(s)
k
] ds
d

.
Partial Dierential Equations Igor Yanovsky, 2005 382
Problem (W02, #7). Consider the equation
_

2
x
2
1
+ +

2
x
2
n
_
u = f in R
n
, (30.4)
where f is an integrable function (i.e. f L
1
(R
n
)), satisfying f(x) = 0 for [x[ R.
Solve (30.4) by Fourier transform, and prove the following results.
a) There is a solution of (30.4) belonging to L
2
(R
n
) if n > 4.
b) If
_
R
n
f(x) dx = 0, there is a solution of (30.4) belonging to L
2
(R
n
) if n > 2.
Proof.
u = f,
[[
2
u() =

f(),
u() =
1
[[
2

f(), R
n
,
u(x) =
_

f()
[[
2
_

.
a) Then
[[ u[[
L
2
(R
n
)
=
__
R
n
[

f()[
2
[[
4
d
_1
2

__
||<1
[

f()[
2
[[
4
d
. .
A
+
_
||1
[

f()[
2
[[
4
d
. .
B
_1
2
.
Notice, [[f[[
2
= [[

f[[
2
B, so B < .
Use polar coordinates on A.
A =
_
||<1
[

f()[
2
[[
4
d =
_
1
0
_
S
n1
[

f[
2
r
4
r
n1
dS
n1
dr =
_
1
0
_
S
n1
[

f[
2
r
n5
dS
n1
dr.
If n > 4,
A
_
S
n1
[

f[
2
dS
n1
= [[

f[[
2
2
< .
[[u[[
L
2
(R
n
)
= [[ u[[
L
2
(R
n
)
= (A+B)
1
2
< .
Partial Dierential Equations Igor Yanovsky, 2005 383
b) We have
u(x, t) =
_

f()
[[
2
_

=
1
(2)
n
2
_

e
ix

f()
[[
2
d
=
1
(2)
n
2
_

e
ix
[[
2
_
1
(2)
n
2
_

e
iy
f(y) dy
_
d
=
1
(2)
n
_

f(y)
_
_

e
i(xy)
[[
2
d
_
dy
=
1
(2)
n
_

f(y)
_
_
1
0
_
S
n1
e
i(xy)r
r
2
r
n1
dS
n1
dr
_
dy
=
1
(2)
n
_

f(y)
_
_
1
0
_
S
n1
e
i(xy)r
r
n3
dS
n1
dr
. .
M <, if n>2.
_
dy.
[u(x, t)[ =
1
(2)
n

M f(y) dy

< .
Partial Dierential Equations Igor Yanovsky, 2005 384
Problem (F02, #7). For the right choice of the constant c, the function
F(x, y) = c(x + iy)
1
is a fundamental solution for the equation
u
x
+ i
u
y
= f in R
2
.
Find the right choice of c, and use your answer to compute the Fourier transform
(in distribution sense) of (x + iy)
1
.
Proof.
86
=
_

x
+i

y
__

x
i

y
_
.
F
1
(x, y) =
1
2
log [z[ is the fundamental solution of the Laplacian. z = x + iy.
F
1
(x, y) = ,
_

x
+i

y
__

x
i

y
_
F(x, y) = .
h
x
+ ih
y
= e
i(x
1
+y
2
)
.
Suppose h = h(x
1
+ y
2
) or h = ce
i(x
1
+y
2
)
.
c
_
i
1
e
i(x
1
+y
2
)
i
2

2
e
i(x
1
+y
2
)
_
= ic(
1
i
2
) e
i(x
1
+y
2
)
e
i(x
1
+y
2
)
,
ic(
1
i
2
) = 1,
c =
1
i(
1
i
2
)
,
h(x, y) =
1
i(
1
i
2
)
e
i(x
1
+y
2
)
.
Integrate by parts:

_
1
x +iy
_
() =
_
R
2
e
i(x
1
+y
2
)
1
i(
1
i
2
)
_

x
+ i

y
_
1
(x +iy) 0
dxdy
=
1
i(
1
i
2
)
=
1
i(
2
+ i
1
)
.
86
Alan solved in this problem in class.
Partial Dierential Equations Igor Yanovsky, 2005 385
31 Laplace Transform
If u L
1
(R
+
), we dene its Laplace transform to be
L[u(t)] = u
#
(s) =
_

0
e
st
u(t) dt (s > 0).
In practice, for a PDE involving time, it may be useful to perform a Laplace transform
in t, holding the space variables x xed.
The inversion formula for the Laplace transform is:
u(t) = L
1
[u
#
(s)] =
1
2i
_
c+i
ci
e
st
u
#
(s) ds.
Example: f(t) = 1.
L[1] =
_

0
e
st
1 dt =
_

1
s
e
st
_
t=
t=0
=
1
s
for s > 0.
Example: f(t) = e
at
.
L[e
at
] =
_

0
e
st
e
at
dt =
_

0
e
(as)t
dt =
1
a s
_
e
(as)t
_
t=
t=0
=
1
s a
for s > a.
Convolution: We want to nd an inverse Laplace transform of
1
s

1
s
2
+1
.
L
1
_
1
s
..
L[f]

1
s
2
+ 1
. .
L[g]
_
= f g =
_
t
0
1 sint

dt

= 1 cos t.
Partial Derivatives: u = u(x, t)
L[u
t
] =
_

0
e
st
u
t
dt =
_
e
st
u(x, t)
_
t=
t=0
+ s
_

0
e
st
u dt = sL[u] u(x, 0),
L[u
tt
] =
_

0
e
st
u
tt
dt =
_
e
st
u
t
_
t=
t=0
+ s
_

0
e
st
u
t
dt = u
t
(x, 0) + sL[u
t
]
= s
2
L[u] su(x, 0) u
t
(x, 0),
L[u
x
] =
_

0
e
st
u
x
dt =

x
L[u],
L[u
xx
] =
_

0
e
st
u
xx
dt =

2
x
2
L[u].
Heat Equation: Consider
_
u
t
u = 0 in U (0, )
u = f on U t = 0,
and perform a Laplace transform with respect to time:
L[u
t
] =
_

0
e
st
u
t
dt = sL[u] u(x, 0) = sL[u] f(x),
L[u] =
_

0
e
st
u dt = L[u].
Partial Dierential Equations Igor Yanovsky, 2005 386
Thus, the transformed problem is: sL[u] f(x) = L[u]. Writing v(x) = L[u], we
have
v + sv = f in U.
Thus, the solution of this equation with RHS f is the Laplace transform of the solution
of the heat equation with initial data f.
Partial Dierential Equations Igor Yanovsky, 2005 387
Table of Laplace Transforms: L[f] = f
#
(s)
L[sinat] =
a
s
2
+ a
2
, s > 0
L[cos at] =
s
s
2
+ a
2
, s > 0
L[sinhat] =
a
s
2
a
2
, s > [a[
L[coshat] =
s
s
2
a
2
, s > [a[
L[e
at
sinbt] =
b
(s a)
2
+ b
2
, s > a
L[e
at
cos bt] =
s a
(s a)
2
+ b
2
, s > a
L[t
n
] =
n!
s
n+1
, s > 0
L[t
n
e
at
] =
n!
(s a)
n+1
, s > a
L[H(t a)] =
e
as
s
, s > 0
L[H(t a) f(t a)] = e
as
L[f],
L[af(t) +bg(t)] = aL[f] + bL[g],
L[f(t) g(t)] = L[f] L[g],
L
_
_
t
0
g(t

t) f(t

) dt

_
= L[f] L[g],
L
_
df
dt
_
= sL[f] f(0),
L
_
d
2
f
dt
2
_
= s
2
L[f] sf(0) f

(0),
_
f

=
df
dt
_
L
_
d
n
f
dt
n
_
= s
n
L[f] s
n1
f(0) . . . f
n1
(0),
L[f(at)] =
1
a
f
#
_
s
a
_
,
L[e
bt
f(t)] = f
#
(s b),
L[tf(t)] =
d
ds
L[f],
L
_
f(t)
t
_
=
_

s
f
#
(s

) ds

,
L
_
_
t
0
f(t

) dt

_
=
1
s
L[f],
L[J
0
(at)] = (s
2
+a
2
)

1
2
,
L[(t a)] = e
sa
.
Example: f(t) = sint. After integrating by parts twice, we obtain:
L[sint] =
_

0
e
st
sint dt = 1 s
2
_

0
e
st
sin t dt,

_

0
e
st
sint dt =
1
1 + s
2
.
Partial Dierential Equations Igor Yanovsky, 2005 388
Example: f(t) = t
n
.
L[t
n
] =
_

0
e
st
t
n
dt =
_
t
n
e
st
s
_

0
+
n
s
_

0
e
st
t
n1
dt =
n
s
L[t
n1
]
=
n
s
_
n 1
s
_
L[t
n2
] = . . . =
n!
s
n
L[1] =
n!
s
n+1
.
Partial Dierential Equations Igor Yanovsky, 2005 389
Problem (F00, #6). Consider the initial-boundary value problem
u
t
u
xx
+au = 0, t > 0, x > 0
u(x, 0) = 0, x > 0
u(0, t) = g(t), t > 0,
where g(t) is continuous function with a compact support, and a is constant.
Find the explicit solution of this problem.
Proof. We solve this problem using the Laplace transform.
L[u(x, t)] = u
#
(x, s) =
_

0
e
st
u(x, t) dt (s > 0).
L[u
t
] =
_

0
e
st
u
t
dt =
_
e
st
u(x, t)
_
t=
t=0
+ s
_

0
e
st
u dt
= su
#
(x, s) u(x, 0) = su
#
(x, s), (since u(x, 0) = 0)
L[u
xx
] =
_

0
e
st
u
xx
dt =

2
x
2
u
#
(x, s),
L[u(0, t)] = u
#
(0, s) =
_

0
e
st
g(t) dt = g
#
(s).
Plugging these into the equation, we obtain the ODE in u
#
:
su
#
(x, s)

2
x
2
u
#
(x, s) +au
#
(x, s) = 0.
_
(u
#
)
xx
(s +a)u
#
= 0,
u
#
(0, s) = g
#
(s).
This initial value problem has a solution:
u
#
(x, s) = c
1
e

s+a x
+ c
2
e

s+a x
.
Since we want u to be bounded as x , we have c
1
= 0, so
u
#
(x, s) = c
2
e

s+a x
. u
#
(0, s) = c
2
= g
#
(s), thus,
u
#
(x, s) = g
#
(s)e

s+a x
.
To obtain u(x, t), we take the inverse Laplace transform of u
#
(x, s):
u(x, t) = L
1
[u
#
(x, s)] = L
1
_
g
#
(s)
. .
L[g]
e

s+a x
. .
L[f]

= g f
= g L
1
_
e

s+a x

= g
_
1
2i
_
c+i
ci
e
st
e

s+a x
ds
_
,
u(x, t) =
_
t
0
g(t t

)
_
1
2i
_
c+i
ci
e
st

s+a x
ds
_
dt

.
Partial Dierential Equations Igor Yanovsky, 2005 390
Problem (F04, #8). The function y(x, t) satises the partial dierential equation
x
y
x
+

2
y
xt
+ 2y = 0,
and the boundary conditions
y(x, 0) = 1, y(0, t) = e
at
,
where a 0. Find the Laplace transform, y(x, s), of the solution, and hence derive
an expression for y(x, t) in the domain x 0, t 0.
Proof. We change the notation: y u. We have
_
xu
x
+ u
xt
+ 2u = 0,
u(x, 0) = 1, u(0, t) = e
at
.
The Laplace transform is dened as:
L[u(x, t)] = u
#
(x, s) =
_

0
e
st
u(x, t) dt (s > 0).
L[xu
x
] =
_

0
e
st
xu
x
dt = x
_

0
e
st
u
x
dt = x(u
#
)
x
,
L[u
xt
] =
_

0
e
st
u
xt
dt =
_
e
st
u
x
(x, t)
_
t=
t=0
+ s
_

0
e
st
u
x
dt
= s(u
#
)
x
u
x
(x, 0) = s(u
#
)
x
, (since u(x, 0) = 0)
L[u(0, t)] = u
#
(0, s) =
_

0
e
st
e
at
dt =
_

0
e
(s+a)t
dt =
_

1
s + a
e
(s+a)t
_
t=
t=0
=
1
s + a
.
Plugging these into the equation, we obtain the ODE in u
#
:
_
(x + s)(u
#
)
x
+ 2u
#
= 0,
u
#
(0, s) =
1
s+a
,
which can be solved:
(u
#
)
x
u
#
=
2
x +s
log u
#
= 2 log(x + s) +c
1
u
#
= c
2
e
log(x+s)
2
=
c
2
(x + s)
2
.
From the initial conditions:
u
#
(0, s) =
c
2
s
2
=
1
s + a
c
2
=
s
2
s +a
.
u
#
(x, s) =
s
2
(s + a)(x + s)
2
.
To obtain u(x, t), we take the inverse Laplace transform of u
#
(x, s):
u(x, t) = L
1
[u
#
(x, s)] = L
1
_
s
2
(s +a)(x +s)
2
_
=
1
2i
_
c+i
ci
e
st
_
s
2
(s + a)(x +s)
2
_
ds.
u(x, t) =
1
2i
_
c+i
ci
e
st
_
s
2
(s +a)(x +s)
2
_
ds.
Partial Dierential Equations Igor Yanovsky, 2005 391
Problem(F90, #1). Using the Laplace transform, or any other convenient method,
solve the Volterra integral equation
u(x) = sinx +
_
x
0
sin(x y)u(y) dy.
Proof. Rewrite the equation:
u(t) = sint +
_
t
0
sin(t t

)u(t

) dt

,
u(t) = sint + (sint) u.
Taking the Laplace transform of each of the elements in :
L[u(t)] = u
#
(s) =
_

0
e
st
u(t) dt,
L[sint] =
1
1 +s
2
,
L[(sint) u] = L[sint] L[u] =
u
#
1 +s
2
.
Plugging these into the equation:
u
#
=
1
1 + s
2
+
u
#
1 + s
2
=
u
#
+ 1
1 +s
2
.
u
#
(s) =
1
s
2
.
To obtain u(t), we take the inverse Laplace transform of u
#
(s):
u(t) = L
1
[u
#
(s)] = L
1
_
1
s
2
_
= t.
u(t) = t.
Partial Dierential Equations Igor Yanovsky, 2005 392
Problem (F91, #5). In what follows, the Laplace transform of x(t) is denoted
either by x(s) or by Lx(t). Show that, for integral n 0,
L(t
n
) =
n!
s
n+1
.
Hence show that
LJ
0
(2

ut) =
1
s
e
u/s
,
where
J
0
(z) =

n=0
(1)
n
(
1
2
z)
2n
n!n!
is a Bessel function. Hence show that
L
__

0
J
0
(2

ut)x(u) du
_
=
1
s
x
_
1
s
_
. (31.1)
Assuming that
LJ
0
(at) =
1

a
2
+ s
2
,
prove with the help of (31.1) that if t 0
_

0
J
0
(au)J
0
(2

ut) du =
1
a
J
0
_
t
a
_
.
Hint: For the last part, use the uniqueness of the Laplace transform.
Proof.
L[t
n
] =
_

0
e
st
..
g

t
n
..
f
dt =
_
t
n
e
st
s
_

0
. .
=0
+
n
s
_

0
e
st
t
n1
dt =
n
s
L[t
n1
]
=
n
s
_
n 1
s
_
L[t
n2
] = . . . =
n!
s
n
L[1] =
n!
s
n+1
.
LJ
0
(2

ut) = L
_

n=0
(1)
n
u
n
t
n
n!n!
_
=

n=0
(1)
n
u
n
n!n!
L[t
n
] =

n=0
(1)
n
u
n
n!s
n+1
=
1
s

n=0
(1)
n
n!
_
u
s
_
n
=
1
s
e

u
s
.
L
__

0
J
0
(2

ut) x(u) du
_
=
_

0
L[J
0
(2

ut)] x(u) du =
1
s
_

0
e

u
s
x(u) du
=
1
s
x
#
_
1
s
_
,
where
x
#
(s) =
_

0
e
us
x(u) du.
Partial Dierential Equations Igor Yanovsky, 2005 393
32 Linear Functional Analysis
32.1 Norms
[[ [[ is a norm on a vector space X if
i) [[x[[ = 0 i x = 0.
ii) [[x[[ = [[ [[x[[ for all scalars .
iii) [[x + y[[ [[x[[ +[[y[[ (the triangle inequality).
The norm induces the distance function d(x, y) = [[x y[[ so that X is a metric space,
called a normed vector space.
32.2 Banach and Hilbert Spaces
A Banach space is a normed vector space that is complete in that norms metric. I.e.
a complete normed linear space is a Banach space.
A Hilbert space is an inner product space for which the corresponding normed space
is complete. I.e. a complete inner product space is a Hilbert space.
Examples: 1) Let K be a compact set of R
n
and let C(K) denote the space of continuous
functions on K. Since every u C(K) achieves maximum and minimum values on K,
we may dene
[[u[[

= max
xK
[u(x)[.
[[ [[

## is indeed a norm on C(K) and since a uniform limit of continuous functions is

continuous, C(K) is a Banach space. However, this norm cannot be derived from an
inner product, so C(K) is not a Hilbert space.
2) C(K) is not a Banach space with [[ [[
2
norm. (Bell-shaped functions on [0, 1] may
converge to a discontinuous -function). In general, the space of continuous functions
on [0, 1], with the norm [[ [[
p
, 1 p < , is not a Banach space, since it is not
complete.
3) R
n
and C
n
are real and complex Banach spaces (with a Eucledian norm).
4) L
p
are Banach spaces (with [[ [[
p
norm).
5) The space of bounded real-valued functions on a set S, with the sup norm [[ [[
S
are
Banach spaces.
6) The space of bounded continuous real-valued functions on a metric space X is a
Banach space.
32.3 Cauchy-Schwarz Inequality
[(u, v)[ [[u[[[[v[[ in any norm, for example
_
[uv[dx (
_
u
2
dx)
1
2
(
_
v
2
dx)
1
2
[a(u, v)[ a(u, u)
1
2
a(v, v)
1
2
_
[v[dx =
_
[v[ 1 dx = (
_
[v[
2
dx)
1
2
(
_
1
2
dx)
1
2
32.4 H older Inequality
_

[uv[ dx [[u[[
p
[[v[[
q
,
which holds for u L
p
() and v L
q
(), where
1
p
+
1
q
= 1. In particular, this shows
uv L
1
().
Partial Dierential Equations Igor Yanovsky, 2005 394
32.5 Minkowski Inequality
[[u + v[[
p
[[u[[
p
+ [[v[[
p
,
which holds for u, v L
p
(). In particular, it shows u + v L
p
().
Using the Minkowski Inequality, we nd that [[ [[
p
is a norm on L
p
().
The Riesz-Fischer theorem asserts that L
p
() is complete in this norm, so L
p
() is a
Banach space under the norm [[ [[
p
.
If p = 2, then L
2
() is a Hilbert space with inner product
(u, v) =
_

uv dx.
Example: R
n
bounded domain, C
1
(

## ) denotes the functions that, along with

their rst-order derivatives, extend continuously to the compact set

. Then C
1
(

) is
a Banach space under the norm
[[u[[
1,
= max
x

([u(x)[ + [u(x)[).
Note that C
1
() is not a Banach space since [[u[[
1,
need not be nite for u C
1
().
32.6 Sobolev Spaces
A Sobolev space is a space of functions whose distributional derivatives (up to some
xed order) exist in an L
p
-space.
Let be a domain in R
n
, and let us introduce
< u, v >
1
=
_

## (u v +uv) dx, (32.1)

[[u[[
1,2
=

< u, u >
1
=
__

([u[
2
+[u[
2
) dx
_1
2
(32.2)
when these expressions are dened and nite. For example, (32.1) and (32.2) are dened
for functions in C
1
0
(). However, C
1
0
() is not complete under the norm (32.2), and so
does not form a Hilbert space.
Divergence Theorem
_

A n dS =
_

div

A dx
Trace Theorem
|u|
L
2
()
C|u|
H
1
()
smooth or square
Poincare Inequality
|u|
p
C|u|
p
1 p
_

[u(x)[
2
dx C
_

[u(x)[
2
dx u C
1
0
(), H
1,2
0
() i.e. p = 2
|u u

|
p
|u|
p
u H
1,p
0
()
Partial Dierential Equations Igor Yanovsky, 2005 395
u

=
1
[[
_

## u(x) dx (Average value of u over ), [[ is the volume of

Notes
u
n
= u n = n
1
u
x
1
+ n
2
u
x
2
[u[
2
= u
2
x
1
+ u
2
x
2
_

[u[ dx =
_

[u[
u
u dx

ab
a +b
2
ab
a
2
+b
2
2
[[u[[[[u[[
[[u[[
2
+ [[u[[
2
2
uu = (
u
2
2
)
_

(u
xy
)
2
dx =
_

u
xx
u
yy
dx u H
2
0
() square
Problem (F04, #6). Let q C
1
0
(R
3
). Prove that the vector eld
u(x) =
1
4
_
R
3
q(y)(x y)
[x y[
3
dy
enjoys the following properties:
87
a) u(x) is conservative;
b) div u(x) = q(x) for all x R
3
;
c) [u(x)[ = O([x[
2
) for large x.
Furthermore, prove that the proverties (1), (2), and (3) above determine the vector eld
u(x) uniquely.
Proof. a) To show that u(x) is conservative, we need to show that curl u = 0.
The curl of

V is another vector eld dened by
curl

V =

V = det

e
1
e
2
e
3

1

2

3
V
1
V
2
V
3

=
_
V
3
x
2

V
2
x
3
,
V
1
x
3

V
3
x
1
,
V
2
x
1

V
1
x
2
_
.
Consider

V (x) =
x
[x[
3
=
(x
1
, x
2
, x
3
)
(x
2
1
+x
2
2
+ x
2
3
)
3
2
.
Then,
u(x) =
1
4
_
R
3
q(y) V (x y) dy,
curl u(x) =
1
4
_
R
3
q(y) curl
x
V (x y) dy.
curl

V (x) = curl
(x
1
, x
2
, x
3
)
(x
2
1
+x
2
2
+ x
2
3
)
3
2
=
_

3
2
2x
2
x
3
(x
2
1
+x
2
2
+ x
2
3
)
5
2

3
2
2x
3
x
2
(x
2
1
+ x
2
2
+ x
2
3
)
5
2
,

3
2
2x
3
x
1
(x
2
1
+ x
2
2
+x
2
3
)
5
2

3
2
2x
1
x
3
(x
2
1
+x
2
2
+x
2
3
)
5
2
,

3
2
2x
1
x
2
(x
2
1
+x
2
2
+x
2
3
)
5
2

= (0, 0, 0).
87
McOwen, p. 138-140.
Partial Dierential Equations Igor Yanovsky, 2005 396
Thus, curl u =
1
4
_
R
3
q(y) 0 dy = 0, and u(x) is conservative.
b) Note that the Laplace kernel in R
3
is
1
4r
.
u(x) =
1
4
_
R
3
q(y)(x y)
[x y[
3
dy =
1
4
_
R
3
q(r) r
r
3
r dr =
_
R
3
q(r)
4r
dr = q.
c) Consider
F(x) =
1
4
_
R
3
q(y)
[x y[
dy.
F(x) is O([x[
1
) as [x[ .
Note that u = F, which is clearly O([x[
2
) as [x[ .