PDE - Most Important

PARTIAL DIFFERENTIAL EQUATIONS
MA 3132
SOLUTIONS OF PROBLEMS IN LECTURE
NOTES
B. Neta Department of Mathematics
Naval Postgraduate School
Code MA/Nd
Monterey, California 93943
March 24, 2008
c _1996 - Professor Beny Neta
1
Contents
1 Introduction and Applications 1
1.1 Basic Concepts and Denitions . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Conduction of Heat in a Rod . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5 A Vibrating String . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.6 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.7 Diusion in Three Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2 Classication and Characteristics 14
2.1 Physical Classication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Classication of Linear Second Order PDEs . . . . . . . . . . . . . . . . . . 14
2.3 Canonical Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4 Equations with Constant Coecients . . . . . . . . . . . . . . . . . . . . . . 35
2.5 Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.6 General Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3 Method of Characteristics 57
3.1 Advection Equation (rst order wave equation) . . . . . . . . . . . . . . . . 57
3.2 Quasilinear Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.2.1 The Case S = 0, c = c(u) . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.2.3 Fan-like Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.2.4 Shock Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.3 Second Order Wave Equation . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.3.1 Innite Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.3.2 Semi-innite String . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.3.3 Semi-innite String with a Free End . . . . . . . . . . . . . . . . . . . 95
4 Separation of Variables-Homogeneous Equations 106
4.1 Parabolic equation in one dimension . . . . . . . . . . . . . . . . . . . . . . 106
4.2 Other Homogeneous Boundary Conditions . . . . . . . . . . . . . . . . . . . 106
5 Fourier Series 119
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.2 Orthogonality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.3 Computation of Coecients . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.4 Relationship to Least Squares . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.5 Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.6 Fourier Cosine and Sine Series . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.7 Term by Term Dierentiation . . . . . . . . . . . . . . . . . . . . . . . . . . 140
5.8 Term by Term Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
5.9 Full solution of Several Problems . . . . . . . . . . . . . . . . . . . . . . . . 151
i
6 Sturm-Liouville Eigenvalue Problem 185
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
6.2 Boundary Conditions of the Third Kind . . . . . . . . . . . . . . . . . . . . 194
6.3 Proof of Theorem and Generalizations . . . . . . . . . . . . . . . . . . . . . 200
6.4 Linearized Shallow Water Equations . . . . . . . . . . . . . . . . . . . . . . 206
6.5 Eigenvalues of Perturbed Problems . . . . . . . . . . . . . . . . . . . . . . . 207
7 PDEs in Higher Dimensions 208
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
7.2 Heat Flow in a Rectangular Domain . . . . . . . . . . . . . . . . . . . . . . 208
7.3 Vibrations of a rectangular Membrane . . . . . . . . . . . . . . . . . . . . . 213
7.4 Helmholtz Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
7.5 Vibrating Circular Membrane . . . . . . . . . . . . . . . . . . . . . . . . . . 224
7.6 Laplaces Equation in a Circular Cylinder . . . . . . . . . . . . . . . . . . . 231
7.7 Laplaces equation in a sphere . . . . . . . . . . . . . . . . . . . . . . . . . . 244
8 Separation of Variables-Nonhomogeneous Problems 248
8.1 Inhomogeneous Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . 248
8.2 Method of Eigenfunction Expansions . . . . . . . . . . . . . . . . . . . . . . 253
8.3 Forced Vibrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
8.4 Poissons Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
9 Fourier Transform Solutions of PDEs 284
9.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
9.2 Fourier Transform pair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
9.3 Heat Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
9.4 Fourier Transform of Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . 290
9.5 Fourier Sine and Cosine Transforms . . . . . . . . . . . . . . . . . . . . . . 296
9.6 Fourier Transform in 2 Dimensions . . . . . . . . . . . . . . . . . . . . . . . 306
10 Greens Functions 308
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
10.2 One Dimensional Heat Equation . . . . . . . . . . . . . . . . . . . . . . . . . 308
10.3 Greens Function for Sturm-Liouville Problems . . . . . . . . . . . . . . . . . 320
10.4 Dirac Delta Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
10.5 Nonhomogeneous Boundary Conditions . . . . . . . . . . . . . . . . . . . . . 335
10.6 Fredholm Alternative And Modied Greens Functions . . . . . . . . . . . . 348
10.7 Greens Function For Poissons Equation . . . . . . . . . . . . . . . . . . . . 359
ii
List of Figures
1 Maple plot of characteristics for 2.2 2a . . . . . . . . . . . . . . . . . . . . . 30
3 Maple plot of characteristics for 2.2 2b . . . . . . . . . . . . . . . . . . . . . 31
4 Maple plot of characteristics for 2.2 2b . . . . . . . . . . . . . . . . . . . . . 31
5 Maple plot of characteristics for 2.2 2c . . . . . . . . . . . . . . . . . . . . . 32
6 Maple plot of characteristics for 2.2 2d . . . . . . . . . . . . . . . . . . . . . 33
7 Maple plot of characteristics for 2.2 2d . . . . . . . . . . . . . . . . . . . . . 33
8 Maple plot of characteristics for 2.2 2f . . . . . . . . . . . . . . . . . . . . . 34
10 Maple plot of characteristics for 2.3 2c . . . . . . . . . . . . . . . . . . . . . 46
11 Maple plot of characteristics for 2.3 2f . . . . . . . . . . . . . . . . . . . . . 47
12 Characteristics for problem 4 . . . . . . . . . . . . . . . . . . . . . . . . . . 63
13 Solution for problem 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
14 Domain and characteristics for problem 3b . . . . . . . . . . . . . . . . . . . 70
15 Characteristics for problem 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 82
16 Solution for 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
17 Solution for 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
18 Sketch of initial solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
19 Solution for 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
20 Solution for 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
21 Domain for problem 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
23 Domain of inuence for problem 3 . . . . . . . . . . . . . . . . . . . . . . . . 102
25 Graphical solution of the eigenvalue problem . . . . . . . . . . . . . . . . . . 118
26 Graph of f(x) = 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
27 Graph of its periodic extension . . . . . . . . . . . . . . . . . . . . . . . . . 120
28 Graph of f(x) = x
2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
30 Graph of f(x) = e
x
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
32 Graph of f(x) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
34 Graph of f(x) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
36 Graph of f(x) = x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
38 graph of f(x) for problem 2c . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
39 Sketch of f(x) and its periodic extension for 1a . . . . . . . . . . . . . . . . 130
40 Sketch of the odd extension and its periodic extension for 1a . . . . . . . . . 130
41 Sketch of the even extension and its periodic extension for 1a . . . . . . . . . 131
42 Sketch of f(x) and its periodic extension for 1b . . . . . . . . . . . . . . . . 131
iii
43 Sketch of the odd extension and its periodic extension for 1b . . . . . . . . . 131
44 Sketch of the even extension and its periodic extension for 1b . . . . . . . . . 132
45 Sketch of f(x) and its periodic extension for 1c . . . . . . . . . . . . . . . . 133
46 Sketch of the odd extension and its periodic extension for 1c . . . . . . . . . 133
47 Sketch of f(x) and its periodic extension for 1d . . . . . . . . . . . . . . . . 134
48 Sketch of the odd extension and its periodic extension for 1d . . . . . . . . . 134
49 Sketch of the even extension and its periodic extension for 1d . . . . . . . . . 135
50 Sketch of the odd extension for 2 . . . . . . . . . . . . . . . . . . . . . . . . 135
51 Sketch of the periodic extension of the odd extension for 2 . . . . . . . . . . 136
52 Sketch of the Fourier sine series for 2 . . . . . . . . . . . . . . . . . . . . . . 136
53 Sketch of f(x) and its periodic extension for problem 3 . . . . . . . . . . . . 137
54 Sketch of the even extension of f(x) and its periodic extension for problem 3 138
55 Sketch of the periodic extension of the odd extension of f(x) (problem 5) . . 139
56 Sketch of the even extension of f(x) = x
3
(problem 1b) . . . . . . . . . . . . 148
57 Sketch of domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
58 Domain fro problem 1 of 7.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
59 Domain for problem 2 of 7.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
60 Airy function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
61 The case = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
62 The case ,= 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
iv
CHAPTER 1
1 Introduction and Applications
1.1 Basic Concepts and Denitions
Problems
1. Give the order of each of the following PDEs
a. u
xx
+ u
yy
= 0
b. u
xxx
+ u
xy
+ a(x)u
y
+ log u = f(x, y)
c. u
xxx
+ u
xyyy
+ a(x)u
xxy
+ u
2
= f(x, y)
d. u u
xx
+ u
2
yy
+ e
u
= 0
e. u
x
+ cu
y
= d
2. Show that
u(x, t) = cos(x ct)
is a solution of
u
t
+ cu
x
= 0
3. Which of the following PDEs is linear? quasilinear? nonlinear? If it is linear, state
whether it is homogeneous or not.
a. u
xx
+ u
yy
2u = x
2
b. u
xy
= u
c. u u
x
+ xu
y
= 0
d. u
2
x
+ log u = 2xy
e. u
xx
2u
xy
+ u
yy
= cos x
f. u
x
(1 + u
y
) = u
xx
g. (sin u
x
)u
x
+ u
y
= e
x
h. 2u
xx
4u
xy
+ 2u
yy
+ 3u = 0
i. u
x
+ u
x
u
y
u
xy
= 0
4. Find the general solution of
u
xy
+ u
y
= 0
(Hint: Let v = u
y
)
5. Show that
u = F(xy) + xG(
y
x
)
is the general solution of
x
2
u
xx
y
2
u
yy
= 0
1
1. a. Second order
b. Third order
c. Fourth order
d. Second order
e. First order
2. u = cos(x ct)
u
t
= c (sin(x ct)) = c sin(x ct)
u
x
= 1 (sin(x ct)) = sin(x ct)
u
t
+ cu
x
= c sin(x ct) c sin(x ct) = 0.
3. a. Linear, inhomogeneous
b. Linear, homogeneous
c. Quasilinear, homogeneous
d. Nonlinear, inhomogeneous
e. Linear, inhomogeneous
f. Quasilinear, homogeneous
g. Nonlinear, inhomogeneous
h. Linear, homogeneous
i. Quasilinear, homogeneous
2
4.
u
xy
+ u
y
= 0
Let v = u
y
then the equation becomes
v
x
+ v = 0
For xed y, this is a separable ODE
dv
v
= dx
ln v = x + C(y)
v = K(y) e
x
In terms of the original variable u we have
u
y
= K(y) e
x
u = e
x
q(y) + p(x)
You can check your answer by substituting this solution back in the PDE.
3
5.
u = F(xy) + xG
_
y
x
_
u
x
= y F
(xy) + G
_
y
x
_
+ x
_
y
x
2
_
G
_
y
x
_
u
xx
= y
2
F
(xy) +
_
y
x
2
_
G
_
y
x
_

y
x
_
y
x
2
_
G
_
y
x
_
+
_
y
x
2
_
G
_
y
x
_
u
xx
= y
2
F
(xy) +
y
2
x
3
G
_
y
x
_
u
y
= xF
(xy) + x
1
x
G
_
y
x
_
u
yy
= x
2
F
(xy) +
1
x
G
_
y
x
_
x
2
u
xx
y
2
u
yy
= x
2
_
y
2
F
+
y
2
x
3
G
_
y
2
_
x
2
F
+
1
x
G
_
Expanding one nds that the rst and third terms cancel out and the second and last terms
cancel out and thus we get zero.
4
1.2 Applications
1.3 Conduction of Heat in a Rod
1.4 Boundary Conditions
Problems
1. Suppose the initial temperature of the rod was
u(x, 0) =
_
2x 0 x 1/2
2(1 x) 1/2 x 1
and the boundary conditions were
u(0, t) = u(1, t) = 0 ,
what would be the behavior of the rods temperature for later time?
2. Suppose the rod has a constant internal heat source, so that the equation describing the
heat conduction is
u
t
= ku
xx
+ Q, 0 < x < 1 .
Suppose we x the temperature at the boundaries
u(0, t) = 0
u(1, t) = 1 .
What is the steady state temperature of the rod? (Hint: set u
t
= 0 .)
3. Derive the heat equation for a rod with thermal conductivity K(x).
4. Transform the equation
u
t
= k(u
xx
+ u
yy
)
to polar coordinates and specialize the resulting equation to the case where the function u
does NOT depend on . (Hint: r =
x
2
+ y
2
, tan = y/x)
5. Determine the steady state temperature for a one-dimensional rod with constant thermal
properties and
a. Q = 0, u(0) = 1, u(L) = 0
b. Q = 0, u
x
(0) = 0, u(L) = 1
c. Q = 0, u(0) = 1, u
x
(L) =
d.
Q
k
= x
2
, u(0) = 1, u
x
(L) = 0
e. Q = 0, u(0) = 1, u
x
(L) + u(L) = 0
5
1. Since the temperature at both ends is zero (boundary conditions), the temperature of
the rod will drop until it is zero everywhere.
2.
k u
xx
+ Q = 0
u(0.t) = 0
u(1, t) = 1
u
xx
=
Q
k
Integrate with respect to x
u
x
=
Q
k
x + A
Integrate again
u =
Q
k
x
2
2
+ Ax + B
Using the rst boundary condition u(0) = 0 we get B = 0. The other boundary condition
will yield
Q
k
1
2
+ A = 1
A =
Q
2k
+ 1
u(x) =
_
1 +
Q
2k
_
x
Q
2k
x
2
3. Follow class notes.
6
4.
r =
_
x
2
+ y
2
_1
2
= arctan
_
y
x
_
r
x
=
1
2
_
x
2
+ y
2
_
1
2
2x =
x
x
2
+ y
2
x
=
1
1 +
_
y
x
_
2
_
y
x
2
_
=
y
x
2
+ y
2
r
y
=
1
2
_
x
2
+ y
2
_
1
2
2y =
y
x
2
+ y
2
y
=
1
1 +
_
y
x
_
2
_
1
x
_
=
x
x
2
+ y
2
u
x
= u
r
r
x
+ u
x
=
x
x
2
+ y
2
u
r

y
x
2
+ y
2
u
u
y
= u
r
r
y
+ u
y
=
y
x
2
+ y
2
u
r
+
x
x
2
+ y
2
u
u
xx
=
_
x
x
2
+ y
2
_
x
u
r
+
x
x
2
+ y
2
(u
r
)
x

_
y
x
2
+ y
2
_
x
u

y
x
2
+ y
2
(u
)
x
u
xx
=
x
2
+ y
2
x
1
2
(x
2
+ y
2
)
1
2
2x
x
2
+ y
2
u
r
+
x
x
2
+ y
2
_
x
x
2
+ y
2
u
rr

y
x
2
+ y
2
u
r
_
2xy
(x
2
+ y
2
)
2
u

y
x
2
+ y
2
_
x
x
2
+ y
2
u
r

y
x
2
+ y
2
u
_
u
xx
=
x
2
x
2
+ y
2
u
rr

2xy
(x
2
+ y
2
)
3
2
u
r
+
y
2
(x
2
+ y
2
)
2
u
+
y
2
(x
2
+ y
2
)
3
2
u
r
+
2xy
(x
2
+ y
2
)
2
u
u
yy
=
_
y
x
2
+ y
2
_
y
u
r
+
y
x
2
+ y
2
(u
r
)
y
+
_
x
x
2
+ y
2
_
y
u
+
x
x
2
+ y
2
(u
)
y
u
yy
=
x
2
+ y
2
y
1
2
(x
2
+ y
2
)
1
2
2y
x
2
+ y
2
u
r
+
y
x
2
+ y
2
_
y
x
2
+ y
2
u
rr
+
x
x
2
+ y
2
u
r
_
+
2xy
(x
2
+ y
2
)
2
u
+
x
x
2
+ y
2
_
y
x
2
+ y
2
u
r
+
x
x
2
+ y
2
u
_
u
yy
=
y
2
x
2
+ y
2
u
rr
+
2xy
(x
2
+ y
2
)
3
2
u
r
+
x
2
(x
2
+ y
2
)
2
u
+
x
2
(x
2
+ y
2
)
3
2
u
r

2xy
(x
2
+ y
2
)
2
u
7
u
xx
+ u
yy
= u
rr
+
1
r
2
u
+
1
r
u
r
u
t
= k
_
u
rr
+
1
r
u
r
+
1
r
2
u
_
In the case u is independent of :
u
t
= k
_
u
rr
+
1
r
u
r
_
8
5. k u
xx
+ Q = 0
a. k u
xx
= 0
Integrate twice with respect to x
u(x) = Ax + B
Use the boundary conditions
u(0) = 1 implies B = 1
u(L) = 0 implies AL + B = 0 that is A =
1
L
Therefore
u(x) =
x
L
+ 1
b. k u
xx
= 0
Integrate twice with respect to x as in the previous case
u(x) = Ax + B
u
x
(0) = 0 implies A = 0
u(L) = 1 implies AL + B = 1 that is B = 1
Therefore
u(x) = 1
c. k u
xx
= 0
u(x) = Ax + B
u
x
(L) = implies A =
Therefore
u(x) = x + 1
9
d. k u
xx
+ Q = 0
u
xx
=
Q
k
= x
2
Integrate with respect to x we get
u
x
(x) =
1
3
x
3
+ A
Use the boundary condition
u
x
(L) = 0 implies
1
3
L
3
+ A = 0 that is A =
1
3
L
3
Integrating again with respect to x
u =
x
4
12
+
1
3
L
3
x + B
Use the second boundary condition
Therefore
u(x) =
x
4
12
+
L
3
3
x + 1
e. k u
xx
= 0
u(x) = Ax + B
u
x
(L) + u(L) = 0 implies A + (AL + 1) = 0 that is A =
1
L + 1
Therefore
u(x) =
1
L + 1
x + 1
10
1.5 A Vibrating String
Problems
1. Derive the telegraph equation
u
tt
+ au
t
+ bu = c
2
u
xx
by considering the vibration of a string under a damping force proportional to the velocity
and a restoring force proportional to the displacement.
2. Use Kirchos law to show that the current and potential in a wire satisfy
i
x
+ C v
t
+ Gv = 0
v
x
+ Li
t
+ Ri = 0
where i = current, v = L = inductance potential, C = capacitance, G = leakage conduc-
tance, R = resistance,
b. Show how to get the one dimensional wave equations for i and v from the above.
11
1. Follow class notes.
a, b are the proportionality constants for the forces mentioned in the problem.
2. a. Check any physics book on Kirchos law.
b. Dierentiate the rst equation with respect to t and the second with respect to x
i
xt
+ C v
tt
+ Gv
t
= 0
v
xx
+ Li
tx
+ Ri
x
= 0
Solve the rst for i
xt
and substitute in the second
i
xt
= C v
tt
Gv
t
v
xx
CLv
tt
GLv
t
+ Ri
x
= 0
i
x
can be solved for from the original rst equation
i
x
= C v
t
Gv
v
xx
CLv
tt
GLv
t
RC v
t
RGv = 0
Or
v
tt
+
_
G
C
+
R
L
_
v
t
+
RG
CL
v =
1
CL
v
xx
which is the telegraph equation.
In a similar fashion, one can get the equation for i.
12
1.6 Boundary Conditions
1.7 Diusion in Three Dimensions
13
CHAPTER 2
2 Classication and Characteristics
2.1 Physical Classication
2.2 Classication of Linear Second Order PDEs
Problems
1. Classify each of the following as hyperbolic, parabolic or elliptic at every point (x, y) of
the domain
a. xu
xx
+ u
yy
= x
2
b. x
2
u
xx
2xy u
xy
+ y
2
u
yy
= e
x
c. e
x
u
xx
+ e
y
u
yy
= u
d. u
xx
+ u
xy
xu
yy
= 0 in the left half plane (x 0)
e. x
2
u
xx
+ 2xyu
xy
+ y
2
u
yy
+ xyu
x
+ y
2
u
y
= 0
f. u
xx
+ xu
yy
= 0 (Tricomi equation)
2. Classify each of the following constant coecient equations
a. 4u
xx
+ 5u
xy
+ u
yy
+ u
x
+ u
y
= 2
b. u
xx
+ u
xy
+ u
yy
+ u
x
= 0
c. 3u
xx
+ 10u
xy
+ 3u
yy
= 0
d. u
xx
+ 2u
xy
+ 3u
yy
+ 4u
x
+ 5u
y
+ u = e
x
e. 2u
xx
4u
xy
+ 2u
yy
+ 3u = 0
f. u
xx
+ 5u
xy
+ 4u
yy
+ 7u
y
= sin x
3. Use any symbolic manipulator (e.g. MACSYMA or MATHEMATICA) to prove (2.1.19).
This means that a transformation does NOT change the type of the PDE.
14
1a. A = x B = 0 C = 1 = 4x
hyperbolic for x < 0
parabolic x = 0
elliptic x > 0
b. A = x
2
B = 2xy C = y
2
= 0 parabolic
c. A = e
x
B = 0 C = e
y
= 4e
x
e
y
elliptic
d. A = 1 B = 1 C = x = 1 + 4x
hyperbolic 0 x >
1
4
parabolic x =
1
4
elliptic x <
1
4
e. A = x
2
B = 2xy C = y
2
= 0 parabolic
f. A = 1 B = 0 C = x = 4x
hyperbolic x < 0
parabolic x = 0
elliptic x > 0
15
2.
A B C Discriminant
a. 4 5 1 25 - 16 > 0 hyperbolic
b. 1 1 1 1 - 4 < 0 elliptic
c. 3 10 3 100 - 36 > 0 hyperbolic
d. 1 2 3 4 - 12 < 0 elliptic
e. 2 -4 2 16 - 16 = 0 parabolic
f. 1 5 4 25 - 16 > 0 hyperbolic
16
3. We substitute for A
, B
, C
given by (2.1.12)-(2.1.14) in
= (B
)
2
4A
= [2A
x
x
+ B(
x
y
+
y
x
) + 2C
y
y
]
2
4
_
A
2
x
+ B
x
y
+ C
2
y
_ _
A
2
x
+ B
x
y
+ C
2
y
_
= 4A
2
2
x
2
x
+ 4A
x
x
B(
x
y
+
y
x
) + 8A
x
x
C
y
y
+ B
2
(
x
y
+
y
x
)
2
+ 4B(
x
y
+
y
x
) C
y
y
+ 4C
2
2
y
2
y
4A
2
2
x
2
x
4A
2
x
B
x
y
4A
2
x
C
2
y
4B
x
y
A
2
x
4B
2
y
4B
x
y
C
2
y
4C
2
y
A
2
x
4C
2
y
B
x
y
4C
2
2
y
2
y
.
Collect terms to nd
= 4AB
2
x
y
+ 4AB
x
2
x
+ 8AC
x
y
+ B
2
(
2
x
2
y
+ 2
x
y
+
2
y
2
x
)
+ 4BC
x
2
y
+ 4BC
x
2
y
4AB
2
x
y
4AC
2
x
2
y
4AB
x
2
x
4B
2
y
4BC
x
2
y
4AC
2
y
2
x
4BC
2
y
= 4AC
_
2
x
2
y
2
x
y
+
2
y
2
x
_
+ B
2
_
2
x
2
y
2
x
y
+
2
y
2
x
_
= J
2
,
since J = (
x
x
).
17
2.3 Canonical Forms
Problems
1. Find the characteristic equation, characteristic curves and obtain a canonical form for
each
a. xu
xx
+ u
yy
= x
2
b. u
xx
+ u
xy
xu
yy
= 0 (x 0, all y)
c. x
2
u
xx
+ 2xyu
xy
+ y
2
u
yy
+ xyu
x
+ y
2
u
y
= 0
d. u
xx
+ xu
yy
= 0
e. u
xx
+ y
2
u
yy
= y
f. sin
2
xu
xx
+ sin 2xu
xy
+ cos
2
xu
yy
= x
2. Use Maple to plot the families of characteristic curves for each of the above.
18
1a. xu
xx
+ u
yy
= x
2
A = x B = 0 C = 1 = B
2
4AC = 4x
If x > 0 then < 0 elliptic
= 0 = 0 parabolic
< 0 > 0 hyperbolic
characteristic equation
dy
dx
=

4x
2x
=

x
x
Suppose x < 0 (hyperbolic)
Let z = x (then z > 0). This is done so as not to get confused by the negative sign
under the radical.
then dz = dx
and
dy
dz
=
dy
dx
=
z
z
=
1
z
dy =
dz
z
1/2
y = 2
z + c
y 2
z = c
characteristic curves: y 2
z = c
2 families as expected.
Transformation: = y 2
z
= y + 2
z
We now substitute in the equations for the starred coecients (see the summary). To
this end we list all the necessary derivatives of and .
x
=
z
z
x
=
z

y
= 1
xx
= (
x
)
x
=
_
1
z
_
x
=
_
1
z
_
z
z
x
=
_
1
2
z
3/2
_
=
1
2z
3/2

xy
= 0
yy
= 0
x
=
z
z
x
=
z

y
= 1
xx
= (
x
)
x
=
_
z
_
x
=
_
z
_
z
z
x
=
_
1
2
z
3/2
_
=
1
2z
3
/2

xy
= 0
yy
= 0
19
Note that
z
= 2
_
1
2
z
1/2
_
=
1
z
= 2
_
1
2
z
1/2
_
=
1
z
Therefore
x
=
1
z

y
= 1
xx
=
1
2z
3/2

xy
= 0
yy
= 0
x
=
1
z

y
= 1
xx
=
1
2z
3
/2

xy
= 0
yy
= 0
Since the problem is Hyperbolic, we know that A
= C
= 0,
B
= 2x(
1
z
)(
1
z
) + 0 + 2 1
2
=
2z
z
2
+ 1 = 2 + 2 = 4
D
= x
1
2z
3/2
+ 0 + 0 + 0 + 0 =
z
2z
3/2
=
1
2
z
E
= x(
1
2z
3/2
) + 0 + 0 + 0 + 0 =
z
2z
3/2
=
1
2
z
F
= 0
G
= x
2
= (z)
2
= z
2
The equation is then
4u
1
2
z
u
+
1
2
z
u
= z
2
The last step is to get rid of z
= 4
z (using the transformation)
z =

4
2
z =

2
; z =
_

4
_
2
4u

2

u
+
2

u
=
_

4
_
4
20
For the elliptic case x > 0
dy
dx
=
i
x
dy = i
dx
x
y = i 2
x + c
= y 2i
x
= y + 2i
x
=
1
2
( + ) = y
=
1
2i
( ) = 2
x
Now we nd the derivatives of and
x
= 0 ;
y
= 1 ;
xx
=
xy
=
yy
= 0
x
= 2
1
2
x
1/2
= x
1/2
;
y
= 0 ;
xx
=
1
2
x
3/2
;
xy
=
yy
= 0
For the elliptic case, B
= 0 and A
= C
, therefore when using the atarred equations

(summary) we have
A
= C
= 0 + 0 + 1 = 1
D
= 0 + 0 + 0 + 0 + 0 = 0
E
= x(
1
2
x
3/2
) + 0 + 0 + 0 + 0 =
1
2
x
1/2
F
= 0
G
= x
and the equation
u

+ u
+
1
2
x
1/2
u
= x
2
Now substitute for x
u

+ u
=
_
2
_
4
u
+ u

=
1
+
1
16

4
21
For the parabolic case x = 0 the equation becomes:
0 u
xx
+ u
yy
= 0
or u
yy
= 0
which is already in a canonical form
This parabolic case can be solved. Integrate with respect to y holding x xed (the
constant of integration may depend on x)
u
y
= f(x)
Integrate again:
u(x, y) = y f(x) + g(x)
22
1b. u
xx
+ u
xy
xu
yy
= 0
A = 1 B = 1 C = x
= 1 + 4x > 0 if x >
1
4
hyperbolic
= 0 =
1
4
parabolic
< 0 <
1
4
elliptic
dy
dx
=
1
1 + 4x
2
Consider the hyperbolic case:
2dy = (1
1 + 4x) dx
Integrate to get characteristics
2y = x
2
3

1
4
(1 + 4x)
3/2
+ c
2y x
1
6
(1 + 4x)
3/2
= c
= 2y x
1
6
(1 + 4x)
3/2
= 2y x +
1
6
(1 + 4x)
3/2
x
= 1
1
6

3
2
4 (1 + 4x)
1/2
= 1
1 + 4x
xx
=
1
2
(1 + 4x)
1/2
4 = 2 (1 + 4x)
1/2
y
= 2
yy
= 0 xy = 0
x
= 1 +
1
6

3
2
4 (1 + 4x)
1/2
= 1 +
1 + 4x
xx
= +2 (1 + 4x)
1/2
y
= 2
xy
= 0
yy
= 0
23
Now we can compute the new coecients or compute each of the derivative in the equa-
tion. We chose the former.
A
= C
= 0
B
= 2(1)(1
1 + 4x)(1+
1 + 4x)+1[2(1
1 + 4x)+2(1+
1 + 4x)+2(x)22
B
= 8x 4 8x = 4 16x
D
= 2 (1 + 4x)
1/2
+ 0 + 0 + 0 + 0 = 2 (1 + 4x)
1/2
E
= 2 (1 + 4x)
1/2
+ 0 + 0 + 0 + 0 = 2 (1 + 4x)
1/2
F
= 0
G
= 0
4 (1 + 4x)u

2(1 + 4x)
1/2
(u
) = 0
u

+
1
2
(1 + 4x)
3/2
(u
) = 0
Now nd (1 + 4x)
3/2
in terms of , and substitute
=
1
3
(1 + 4x)
3/2
3( ) = (1 + 4x)
3/2
(1 + 4x)
3/2
= [3( )]
1
u

=
1
2[3( )]
(u
)
u

=
1
6( )
(u
)
The parabolic case is easier, the only characteristic is
y =
1
2
x + K
and so the transformation is
= y
1
2
x
= x
The last equation is an arbitrary function and one should check the Jacobian. The details
are left to the reader. One can easily show that
A
= B
= 0
24
Also
C
= 1
and the rest of the coecients are zero. Therefore the equation is
u
= 0
In the elliptic case, one can use the transformation z = (1+4x) so that the characteristic
equation becomes
dy
dx
=
1
z
2
or if we eliminate the x dependence
dy
dz
=
dy
dx
dx
dz
=
1
4
1
z
2
Now integrate, and take the real and imaginary part to be the functions and . The rest
is left for the reader.
25
1c. x
2
u
xx
+ 2xy u
xy
+ y
2
u
yy
+ xy u
x
+ y
2
u
y
= 0
A = x
2
B = 2xy C = y
2
= 4x
2
y
2
4x
2
y
2
= 0 parabolic
dy
dx
=
2xy
2x
2
=
y
x
dy
y
=
dx
x
= ln y ln x = ln
_
y
x
_
e
=
y
x
= x arbitrarily chosen since this is parabolic
x
=
1
x

y
=
1
y

xx
=
1
x
2

xy
= 0
yy
=
1
y
2
x
= 1
y
=
xx
=
xy
=
yy
= 0
A
= B
= 0 parabolic
C
= x
2
1 + 2xy 0 + y
2
0 = x
2
D
= x
2
_
1
x
2
_
+ 2xy 0 + y
2
_
1
y
2
_
+ xy
_
1
x
_
+ y
2
_
1
y
_
= 1 1 y + y = 0
E
= 0 + 0 + 0 +xy 1 + 0 = xy
F
= 0
G
= 0
x
2
u

+ xyu
= 0
u

= e
y = e
x therefore y/x = e
This equation can be solved.

26
1d. u
xx
+ xu
yy
= 0
A = 1 B = 0 C = x
= 4x > 0 if x < 0 hyperbolic
= 0 x = 0 parabolic
< 0 x > 0 elliptic
Parabolic x = 0 u
xx
= 0 already in canonical form
Hyperbolic x < 0 Let = x
= 4 > 0
dy
dx
=
2
2
=
_
Note: dx = d
dy =
_
(d)
y
2
3
3/2
= c
= y +
2
3
3/2
= y
2
3
3/2
Continue as in example in class (See 1a)
27
1e. u
xx
+ y
2
u
yy
= y
A = 1 B = 0 C = y
2
= 4y
2
< 0 elliptic if y ,= 0
For y = 0 the equation is parabolic and it is in canonical form u
xx
= 0
dy
dx
=

4y
2
2
= iy
dy
y
= idx
ln y = ix + c
= ln y + ix
= ln y ix
= ln y
x
= 0
y
=
1
y

xx
=
xy
= 0
yy
=
1
y
2
= x
x
= 1
y
= 0
xx
=
xy
=
yy
= 0
A
= C
= 0 + 0 +y
2
_
1
y
_
2
= 1
B
= 0
D
= y
2
_
1
y
2
_
= 1
E
= 0
F
= 0
G
= 0
u
+ u

u
= y
But y = e
+ u

u
= e
28
1f. sin
2
xu
xx
+ sin 2xu
xy
+ cos
2
xu
yy
= x
A = sin
2
x B = sin 2x = 2 sin x cos x C = cos
2
x
= 0 parabolic
dy
dx
=
2 sin x cos x
2 sin
2
x
= cot x
y = ln sin x + c
= y ln sin x
x
= cot x
y
= 1
xx
=
1
sin
2
x

xy
= 0
yy
= 0
= y
x
= 0
y
= 1
xx
= 0
xy
= 0
yy
= 0
A
= B
= 0
C
= 0 + 0 + cos
2
x 1 = cos
2
x
D
= sin
2
x
_
1
sin
2
x
_
+ 0 + 0 + 0 + 0 = 1
E
= 0 + 0 + 0 + 0 + 0 = 0
F
= 0
G
= x
Therefore the equation becomes:
cos
2
xu

+ u
= x
ln sin x = y =
sin x = e

cos
2
x = 1 sin
2
x = 1 e
2( )
x = arcsin e

[1 e
2( )
] u

+ u
= arcsin e

29
2a. y 2
z = c z > 0
eq: y + 2 sqrt(z) = c; maple command to give the equation
char:=solve (eq, y); maple command to solve for y
chars:=seq (char, c= -5..5); maple command to create several characteristic
curves for a variety of cs.
plot ( chars , z = 0..10, y = 5..5); maple command to plot all those curves
-4
-2
0
2
4
y
0 2 4 6 8 10
x
Figure 1: Maple plot of characteristics for 2.2 2a
-4
-2
0
2
4
y
0 2 4 6 8 10
x
30
2b. y =
1
2
x
1
12
(1 + 4x)
3/2
+ c
1 + 4 x 0
4 x 1
x .25
-4
-2
0
2
4
y
0 2 4 6 8 10
x
Figure 3: Maple plot of characteristics for 2.2 2b
-4
-2
0
2
4
y
0 2 4 6 8 10
x
Figure 4: Maple plot of characteristics for 2.2 2b
31
2c. ln
y
x
= c parabolic
ln y = xe
c
= kx
-10
-5
0
5
10
y
-10 -5 0 5 10
x
Figure 5: Maple plot of characteristics for 2.2 2c
32
2d. y
2
3
z
3/2
= c
-4
-2
0
2
4
y
0 2 4 6 8 10
x
Figure 6: Maple plot of characteristics for 2.2 2d
-4
-2
0
2
4
y
0 2 4 6 8 10
x
Figure 7: Maple plot of characteristics for 2.2 2d
33
2e. elliptic. no real characteristic
2f. y = ln sin x + c
-4
-2
0
2
4
y
0 2 4 6 8 10
x
Figure 8: Maple plot of characteristics for 2.2 2f
34
2.4 Equations with Constant Coecients
Problems
1. Find the characteristic equation, characteristic curves and obtain a canonical form for
a. 4u
xx
+ 5u
xy
+ u
yy
+ u
x
+ u
y
= 2
b. u
xx
+ u
xy
+ u
yy
+ u
x
= 0
c. 3u
xx
+ 10u
xy
+ 3u
yy
= x + 1
d. u
xx
+ 2u
xy
+ 3u
yy
+ 4u
x
+ 5u
y
+ u = e
x
e. 2u
xx
4u
xy
+ 2u
yy
+ 3u = 0
f. u
xx
+ 5u
xy
+ 4u
yy
+ 7u
y
= sin x
2. Use Maple to plot the families of characteristic curves for each of the above.
35
1a. 4u
xx
+ 5u
xy
+ u
yy
+ u
x
+ u
y
= 2
A = 4 B = 5 C = 1
= 5
2
4 4 1 = 25 16 = 9 > 0 hyperbolic
dy
dx
=
5
9
2 4
=
5 3
8
1/4
1
dy = dx dy =
1
4
dx
y = x + c y =
1
4
x + c
= y x = y
1
4
x
x
= 1
y
= 1
xx
= 0
xy
= 0
yy
= 0
x
=
1
4

y
= 1
xx
= 0
xy
= 0
yy
= 0
A
= C
= 0
B
= 2 4(1)(
1
4
) + 5(1 1 + 1(
1
4
)) + 2 1 1 1 =
9
4
D
= 0 + 0 + 0 + 1(1) + 1 1 = 0
E
= 0 + 0 + 0 + 1(
1
4
) + 1 1 =
3
4
F
= 0
G
= 2
9
4
u

+
3
4
u
= 2
u

=
1
3
u

8
9
This equation can be solved as follows:
Let = u
then u

=
=
1
3

8
9
This is Linear 1
st
order ODE
36

1
3
=
8
9
Integrating factor is e
1
3

(e
1
3

)
=
8
9
e
1
3

e
1
3

=
8
9
_
e
1
3

d
=
8
3
e
1
3

+ C ()
=
8
3
+ C () e
1
3

To nd u we integrate with respect to
u
=
8
3
+ C() e
1
3

u =
8
3
+ e
1
3

c
1
()
. .
integral of C()
+K()
To check the solution, we dierentiate it and substitute in the canonical form:
u
= 0 +
1
3
e
1
3

c
1
() + K
()
u

=
1
3
e
1
3

c
1
()
u
=
8
3
+ e
1
3

c
1
()
1
3
u
=
8
9
+
1
3
e
1
3

c
1
()
Substitute in the PDE in canonical form
1
3
e
1
3

c
1
() =
8
9
+
1
3
e
1
3

c
1
()
8
9
Identity
In terms of original variables u(x, y) =
8
3
(y
1
4
x) + e
1
3
(yx)
c
1
(y
1
4
x) + K (y x)
37
1b. u
xx
+ u
xy
+ u
yy
+ u
x
= 0
A = 1 B = 1 C = 1 = 1 4 = 3 < 0 elliptic
dy
dx
=
1
3
2
2dy = (1
3i) dx
= 2y (1 +
3i)x = 2y (1
3i)x
=
1
2
( + ) = 2y x
=
1
2i
( ) =
3x
x
= 1
y
= 2
xx
= 0
xy
= 0
yy
= 0
x
=
3
y
= 0
xx
= 0
xy
= 0
yy
= 0
A
= C
= 1(1)
2
+ 1(1)2 + 1(2)
2
= 1 2 + 4 = 3
B
= 0
D
= 0 + 0 + 0 + 1(1) + 0 = 1
E
= 0 + 0 + 0 + 1(
3) + 0 =
3
F
= 0
G
= 0
3u
+ 3u
3 u
= 0
u
+ u
=
1
3
u
3
3
u
38
1c. 3u
xx
+ 10u
xu
+ 3u
yy
= x + 1
A = C = 3 B = 10 = 100 36 = 64 > 0 hyperbolic
dy
dx
=
10 8
6
3
1/3
= y 3x = y
1
3
x
x
= 3
y
= 1
xx
= 0
xy
= 0
yy
= 0
x
=
1
3

y
= 1
xx
= 0
xy
= 0
yy
= 0
A
= C
= 0
B
= 2 3(3)(
1
3
) + 10(3
1
3
) + 2 3(1)(1) = 6
100
3
+ 6 =
64
3
D
= 0 + 0 + 0 + 0 + 0 = 0
E
= 0 + 0 + 0 + 0 + 0 = 0
F
= 0
G
= x + 1
64
3
u

= x + 1
= y 3x
= y
1
3
x
=
8
3
x
x =
3
8
( )
64
3
u

=
3
8
( ) + 1
u

=
9
512
( )
3
64
39
To Find the general solution !
u

=
9
512
( )
3
64
u
=
9
512
(
1
2
2
)
3
64
+ f()
u =
9
512
(
1
2
2

1
2

2
)
3
64
+ F () + G()
=
9
512

1
2
( )
3
64
+ F () + G()
u(x, y) =
9
1024
_
y
1
3
x
_
(y 3x)
_
1
3
x 3x
_
. .
8
3
x
3
64
_
y
1
3
x
_
(y 3x)
+F (y 3x) + G(y
1
3
x)
=
9
1024

8
3
x
_
y
1
3
x
_
(y 3x)
3
64
_
y
1
3
x
_
(y 3x) + F (y 3x)
+G(y
1
3
x)
u(x, y) =
_
3
128
x
3
64
_
(y
1
3
x) (y 3x) + F (y 3x) + G
_
y
1
3
x
_
check !
u
x
=
3
128
(y
1
3
x) (y 3x) + (
3
128
x
3
64
) (
1
3
) (y 3x)
+
_
3
128
x
3
64
__
y
1
3
x
_
(3) 3F
(y 3x)
1
3
G
_
y
1
3
x
_
u
y
=
_
3
128
x
3
64
_
(y 3x) +
_
3
128
x
3
64
_ _
y
1
3
x
_
+F
(y 3x) +G
_
y
1
3
x
_
u
xx
=
3
128
_
1
3
_
(y 3x) +
9
128
(y
1
3
x) + (
3
128
x
3
64
)
1
3
(
3
128
) (y 3x)
3
_
3
128
_ _
y
1
3
x
_
3 (
1
3
) (
3
128
x
3
64
) + 9F
+
1
9
G
u
xx
=
1
64
(y 3x) +
9
64
(y
1
3
x) + 2 (
3
128
x
3
64
) + 9F
(y 3x) +
1
9
G
(y
1
3
x)
40
u
xy
=
3
128
(y 3x) 3 (
3
128
x
3
64
)
3
128
(y
1
3
x)
1
3
(
3
128
x
3
64
)
3F
(y 3x)
1
3
G
(y
1
3
x)
u
yy
=
3
128
x
3
64

3
128
x
3
64
+ F
(y 3x) + G
_
y
1
3
x
_
3u
xx
+ 10u
xy
+ 3u
yy
=
3
64
(y 3x) +
27
64
(y
1
3
x) + 6
_
3
128
x
3
64
_
+ 27F
+
1
3
G
30
128
(y 3x)
15
64
(y
1
3
x)
100
3
_
3
128
x
3
64
_
30F

10
3
G
+6
_
3
128
x
3
64
_
+ 3F
+ 3G
=
12
64
(y 3x) +
12
64
(y
1
3
x)
64
3
(
3
128
x
3
64
)
=
9
16
x
1
16
x +
1
2
x + 1 = x + 1
checks
41
1d. u
xx
+ 2u
xy
+ 3u
yy
+ 4u
x
+ 5u
y
+ u = e
x
A = 1 B = 2 C = 3 = 4 12 = 8 < 0 elliptic
dy
dx
=
2
8
2
= 1 i
2
y = (1 i
2)x + C
= y (1 + i
2)x
= y (1 i
2)x
= y x
=
2x x =

x
= 1
y
= 1
xx
=
xy
=
yy
= 0
x
=
2
y
= 0
xx
=
xy
=
yy
= 0
A
= C
= 1(1)
2
+ 2(1)1 + 3(1)
2
= 1 2 + 3 = 2
B
= 0
D
= 0 + 0 + 0 + 4(1) + 5(1) = 1
E
= 0 + 0 + 0 + 4(
2) + 0 = 4
2
F
= 1
G
= e
x
2u
+ 2u

+ u
2 u
+ u = e
x
u
+ u

=
1
2
u
+ 2
2 u

1
2
u +
1
2
e
/
2
42
1e. 2u
xx
4u
xy
+ 2u
yy
+ 3u = 0
A = C = 2 B = 4 = 16 16 = 0 parabolic
dy
dx
=
4 0
4
= 1
dy = dx
= y + x
x
= 1
y
= 1
xx
=
xy
=
yy
= 0
= x
x
= 1
y
= 0
xx
=
xy
=
yy
= 0
A
= 0
B
= 0
C
= 2 4 0 + 2 0 = 2
D
= 0 + 0 + 0 + 0 + 0 = 0
E
= 0 + 0 + 0 + 0 + 0 = 0
F
= 3
G
= 0
2u

+ 3u = 0
u

=
3
2
u
43
1f. u
xx
+ 5u
xy
+ 4u
yy
+ 7u
y
= sin x
A = 1 B = 5 C = 4 = 25 16 = 9 > 0 hyperbolic
dy
dx
=
5 3
2
4
1
= y 4x
x
= 4
y
= 1
xx
=
xy
=
yy
= 0
= y x
x
= 1
y
= 1
xx
=
xy
=
yy
= 0
A
= C
= 0
B
= 2(1)(4)(1) + 5((4)(1) + 1(1)) + 2(4)(1)(1) = 8 25 + 8 = 9

D
= 0 + 0 + 0 + 0 + 7(1) = 7
E
= 0 + 0 + 0 + 0 + 7(1) = 7
F
= 0
G
= sin x
9 u

+ 7(u
+ u
) = sin x
u

=
7
9
(u
+ u
)
1
9
sin x
= 3x
x =

3
u

=
7
9
(u
+ u
)
1
9
sin
_
3
_
44
2a. y = x + c
y =
1
4
x + c
-10
-5
0
5
10
y
-4 -2 0 2 4
x
45
2b. elliptic . no real characteristics
2c. y = 3x + c
y =
1
3
x + c
-10
-5
0
5
10
y
-4 -2 0 2 4
x
Figure 10: Maple plot of characteristics for 2.3 2c
46
2d. elliptic . no real characteristics
2e. y = x + c see 2a
2f. y = 4x + c
y = x + c (see 2a)
-10
-5
0
5
10
y
-4 -2 0 2 4
x
Figure 11: Maple plot of characteristics for 2.3 2f
47
2.5 Linear Systems
2.6 General Solution
Problems
1. Determine the general solution of
a. u
xx
1
c
2
u
yy
= 0 c = constant
b. u
xx
3u
xy
+ 2u
yy
= 0
c. u
xx
+ u
xy
= 0
d. u
xx
+ 10u
xy
+ 9u
yy
= y
2. Transform the following equations to
U
= cU
by introducing the new variables
U = ue
(+)
where , to be determined
a. u
xx
u
yy
+ 3u
x
2u
y
+ u = 0
b. 3u
xx
+ 7u
xy
+ 2u
yy
+ u
y
+ u = 0
(Hint: First obtain a canonical form)
3. Show that
u
xx
= au
t
+ bu
x
b
2
4
u + d
is parabolic for a, b, d constants. Show that the substitution
u(x, t) = v(x, t)e
b
2
x
transforms the equation to
v
xx
= av
t
+ de
b
2
x
48
1a. u
xx

1
c
2
u
yy
= 0
A = 1 B = 0 C =
1
c
2
=
4
c
2
> 0 hyperbolic
dy
dx
=

2
c
2
=
1
c
y =
1
c
x + K
= y +
1
c
x
= y
1
c
x
Canonical form:
u

= 0
The solution is:
u = f () + g ()
Substitute for and to get the solution in the original domain:
u (x, y) = f (y +
1
c
x) + g (y
1
c
x)
49
1b. u
xx
3u
xy
+ 2u
yy
= 0
A = 1 B = 3 C = 2 = 9 8 = 1 hyperbolic
dy
dx
=
3 1
2
2
1
y = 2x + K
1
y = x + K
2
= y + 2x
x
= 2
y
= 1
= y + x
x
= 1
y
= 1
u
x
= 2u
+ u
u
y
= u
+ u
u
xx
= 2 (2u

+ u

) + 2u

+ u

u
xx
= 4u

+ 4u

+ u

u
xy
= 2 (u

+ u

) + u

+ u

= 2u

+ 3u

+ u

u
yy
= u

+ 2u

+ u

u
xx
3u
xy
+2u
y y
= 4u

+4u

+u

3 (2u

+ 3u

+ u

) +2 (u

+ 2u

+ u

)
= u

u

= 0
The solution in the original domain is then:
u(x, y) = f (y + 2x) + g (y + x)
50
1c. u
xx
+ u
xy
= 0
A = 1 B = 1 C = 0 = 1 hyperbolic
dy
dx
=
+1 1
2
+1
0
y = +x + K
1
y = K
2
= y x
x
= 1
y
= 1
= y
x
= 0
y
= 1
u
x
= u
+ u

x
..
=0
= u
u
y
= u
+ u
u
xx
= u

u
xy
= u

u

u
xx
+ u
xy
= u

= 0
The solution in the original domain is then:
u = f (y x) + g (y)
51
1d. u
xx
+ 10u
xy
+ 9u
yy
= y
A = 1 B = 10 C = 9 = 100 36 = 64 hyperbolic
dy
dx
=
10 8
2
9
1
= y 9x
x
= 9
y
= 1
= y x
x
= 1
y
= 1
u
x
= 9u
u
y
= u
+ u
u
xx
= 9 (9u

u

) (9u

u

)
= 81u

+ 18u

+ u

u
xy
= 9 (u

+ u

) (u

+ u
yy
)
= 9u

10u

u

u
yy
= u

+ 2u

+ u

u
xx
+ 10u
xy
+ 9u
y y
= (81 90 + 9)
. .
=0
u

+ (18100+18)u

+ (1 10 + 9)
. .
=0
u

= y
64u

= y
Substitute for y by using the transformation
= y 9x
9 = 9y 9x
9 = 8y
y =
9
8
u

=
9
8
64
=

512

9
512
u

=

512

9
512
To solve this PDE let be xed and integrate with respect to
52
u
=

512

9
512
1
2
2
+ f ()
u =
1
2
512

9
2
1
512

2
+ F() + g ()
The solution in xy domain is:
u(x, y) =
(y 9x)
2
(y x)
1024

9
1024
(y 9x)(y x)
2
+ F(y 9x) + g(y x)
53
2a. u
xx
u
yy
+ 3u
x
2u
y
+ u = 0
U = u e
( + )
A = 1 B = 0 C = 1 = 4 hyperbolic
dy
dx
=
2
2
= 1
= y x
= y + x
u
x
= u
+ u
u
y
= u
+ u
u
xx
= (u

+ u

) + (u

+ u

) = u

2u

+ u

u
yy
= u

+ 2u

+ u

4u

3u
+ 3u
2u
2u
+ u = 0
4u

5u
+ u
+ u = 0
U = u e
( + )
u = U e
( + )
u
= U
e
( + )
+ U e
( + )
u
= U
e
( + )
+ U e
( + )
u

= U

e
( + )
+ U
e
( + )
+ U
e
( + )
+ U e
( + )
4U

4 U
4U
4 U 5U
5U + U
+ U + U = 0
4U

+ (4 5)U
+ (4 + 1)U
+ (4 5 + + 1) U = 0

= 5/4 = 1/4 4(1/4)(5/4) 5(1/4) + (5/4) + 1 = 1/4
4U

1
4
U = 0
U

=
1
16
U required form
54
2b. 3u
xx
+ 7u
xy
+ 2u
yy
+ u
y
+ u = 0
A = 3 B = 7 C = 2 = 49 24 = 25 hyperbolic
dy
dx
=
7 5
6
2
1
3
= y 2x
x
= 2
y
= 1
= y
1
3
x
x
=
1
3

y
= 1
u
x
= 2u

1
3
u
u
y
= u
+ u
u
xx
= 2
_
2u

1
3
u

_

1
3
_
2u

1
3
u

_
u
xx
= 4u

+
4
3
u

+
1
9
u

u
xy
= 2 (u

+ u

)
1
3
(u

+ u

)
u
xy
= 2u

7
3
u

1
3
u

u
yy
= u

+ 2u

+ u

4u

49
3
u

+ 4u

+ u
+ u
+ u = 0
25
3
u

+ u
+ u
+ u = 0
Use last page:
25
3
(U

+ U
+ U
+ U) + U
+ U + U
+ U + U = 0
25
3
U

+
_
25
3
+ 1
_
U
+
_
25
3
+ 1
_
U
+
_
25
3
+ + + 1
_
U = 0

= 3/25 = 3/25
3
25
+
3
25
+
3
25
+ 1 =
28
25
25
3
u

+
28
25
U = 0 U

=
3
25
28
25
U
55
3.
u
xx
= au
t
+ bu
x

b
2
4
u + d
A = 1 B = C = 0 = 0 parabolic
dx
dt
= 0 already in canonical form since u
xx
is the only 2
nd
order term
u = e
b
2
x
u
x
=
x
e
b
2
x
+
b
2
e
b
2
x
u
xx
=
xx
e
b
2
x
+ b
x
e
b
2
x
+
b
2
4
e
b
2
x
u
t
=
t
e
b
2
x

xx
+ b
x
+
b
2
4
= a
t
+ b
_
x
+
b
2
_

b
2
4
+ de
b
2
x
Since
x
and terms cancel out we have:
xx
= a
t
+ de
b
2
x
56
CHAPTER 3
3 Method of Characteristics
3.1 Advection Equation (rst order wave equation)
Problems
1. Solve
w
t
3
w
x
= 0
subject to
w(x, 0) = sin x
2. Solve using the method of characteristics
a.
u
t
+ c
u
x
= e
2x
subject to u(x, 0) = f(x)
b.
u
t
+ x
u
x
= 1 subject to u(x, 0) = f(x)
c.
u
t
+ 3t
u
x
= u subject to u(x, 0) = f(x)
d.
u
t
2
u
x
= e
2x
subject to u(x, 0) = cos x
e.
u
t
t
2
u
x
= u subject to u(x, 0) = 3e
x
3. Show that the characteristics of
u
t
+ 2u
u
x
= 0
u(x, 0) = f(x)
are straight lines.
4. Consider the problem
u
t
+ 2u
u
x
= 0
u(x, 0) = f(x) =
1 x < 0
1 +
x
L
0 < x < L
2 L < x
a. Determine equations for the characteristics
b. Determine the solution u(x, t)
c. Sketch the characteristic curves.
d. Sketch the solution u(x, t) for xed t.
57
1. The PDE can be rewrriten as a system of two ODEs
dx
dt
= 3
dw
dt
= 0
The solution of the rst gives the characteristic curve
x + 3t = x
0
and the second gives
w(x(t), t) = w(x(0), 0) = sin x
0
= sin(x + 3t)
w(x, t) = sin(x + 3t)
2. a. The ODEs in this case are
dx
dt
= c
du
dt
= e
2x
Solve the characteristic equation
x = ct + x
0
Now solve the second ODE. To do that we have to plug in for x
du
dt
= e
2(x
0
+ct)
= e
2x
0
e
2ct
u(x, t) = e
2x
0
1
2c
e
2ct
+ K
The constant of integration can be found from the initial condition
f(x
0
) = u(x
0
, 0) =
1
2c
e
2x
0
+ K
Therefore
K = f(x
0
)
1
2c
e
2x
0
Plug this K in the solution
u(x, t) =
1
2c
e
2x
0
+2ct
+ f(x
0
)
1
2c
e
2x
0
0
from the characteristic curve u(x, t) =
1
2c
e
2x
+ f(x ct)
1
2c
e
2(x ct)
58
2. b. The ODEs in this case are
dx
dt
= x
du
dt
= 1
ln x = t + ln x
0
or x = x
0
e
t
The solution of the second ODE is
u = t + K and the constant is f(x
0
)
u(x, t) = t + f(x
0
)
Substitute x
0
from the characteristic curve u(x, t) = t + f
_
xe
t
_
2. c. The ODEs in this case are
dx
dt
= 3t
du
dt
= u
x =
3
2
t
2
+ x
0
The second ODE can be written as
du
u
= dt
Thus the solution of the second ODE is
lnu = t + ln K and the constant is f(x
0
)
u(x, t) = f(x
0
) e
t
Substitute x
0
from the characteristic curve u(x, t) = f
_
x
3
2
t
2
_
e
t
59
2. d. The ODEs in this case are
dx
dt
= 2
du
dt
= e
2x
x = 2t + x
0
Now solve the second ODE. To do that we have to plug in for x
du
dt
= e
2(x
0
2t)
= e
2x
0
e
4t
u(x, t) = e
2x
0
_
1
4
e
4t
_
+ K
The constant of integration can be found from the initial condition
cos(x
0
) = u(x
0
, 0) =
1
4
e
2x
0
+ K
Therefore
K = cos(x
0
) +
1
4
e
2x
0
Plug this K in the solution and substitute for x
0
from the characteristic curve
u(x, t) =
1
4
e
2(x +2t)
e
4t
+ cos(x + 2t) +
1
4
e
2(x +2t)
u(x, t) =
1
4
e
2x
_
e
4t
1
_
+ cos(x + 2t)
To check the answer, we dierentiate
u
x
=
1
2
e
2x
_
e
4t
1
_
sin(x + 2t)
u
t
=
1
4
e
2x
_
4 e
4t
_
2 sin(x + 2t)
Substitute in the PDE
u
t
2u
x
= e
2x
e
4t
2 sin(x + 2t) 2
_
1
2
e
2x
_
e
4t
1
_
sin(x + 2t)
_
= e
2x
e
4t
2 sin(x + 2t) e
2x
e
4t
+ e
2x
+ 2 sin(x + 2t)
= e
2x
which is the right hand side of the PDE
60
2. e. The ODEs in this case are
dx
dt
= t
2
du
dt
= u
x =
t
3
3
+ x
0
Now solve the second ODE. To do that we rewrite it as
du
u
= dt
Therefore the solution as in 2c
ln u = t + ln K and the constant is 3 e
x
0
Plug this K in the solution and substitute for x
0
from the characteristic curve
ln u(x, t) = ln
_
3 e
x+
1
3
t
3
_
t
u(x, t) = 3 e
x+
1
3
t
3
e
t
To check the answer, we dierentiate
u
t
= 3 e
x
_
t
2
1
_
e
1
3
t
3
t
u
x
= 3 e
x
e
1
3
t
3
t
Substitute in the PDE
u
t
t
2
u
x
= 3 e
x
e
1
3
t
3
t
t
2
_
3 e
x
_
t
2
1
_
e
1
3
t
3
t
_
= 3 e
x
e
1
3
t
3
t
__
t
2
1
_
t
2
_
= 3 e
x+
1
3
t
3
t
= u
61
3. The ODEs in this case are
dx
dt
= 2u
du
dt
= 0
Since the rst ODE contains x, t and u, we solve the second ODE rst
u(x, t) = u(x(0), 0) = f(x(0))
Plug this u in the rst ODE, we get
dx
dt
= 2f(x(0))
The solution is
x = x
0
+ 2tf(x
0
)
These are characteristics lines all with slope
1
2f(x
0
)
Note that the characteristic through x
1
(0) will have a dierent slope than the one through
x
2
(0). That is the straight line are NOT parallel.
62
4. The ODEs in this case are
dx
dt
= 2u
du
dt
= 0
with
u(x, 0) = f(x) =
1 x < 0
1 +
x
L
0 < x < L
2 L < x
a. Since the rst ODE contains x, t and u, we solve the second ODE rst
u(x, t) = u(x(0), 0) = f(x(0))
Plug this u in the rst ODE, we get
dx
dt
= 2f(x(0))
The solution is
x = x
0
+ 2tf(x
0
)
5 0 5 10 15 20 25
0
0.5
1
1.5
2
2.5
3
3.5
4
x
t
Figure 12: Characteristics for problem 4
63
b. For x
0
< 0 then f(x
0
) = 1 and the solution is
u(x, t) = 1 on x = x
0
+ 2t
or
u(x, t) = 1 on x < 2t
For x
0
> L then f(x
0
) = 2 and the solution is
u(x, t) = 2 on x > 4t + L
For 0 < x
0
< L then f(x
0
) = 1 + x
0
/L and the solution is
u(x, t) = 1 +
x
0
L
on x = 2t
_
1 +
x
0
L
_
+ x
0
That is
x
0
=
x 2t
2t + L
L
Thus the solution on this interval is
u(x, t) = 1 +
x 2t
2t + L
=
2t + L + x 2t
2t + L
=
x + L
2t + L
Notice that u is continuous.
5 0 5 10 15 20 25
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
x=2t x=4t+L
u=(x+L)/(2t+L)
u=1
u=2
Figure 13: Solution for problem 4
64
3.2 Quasilinear Equations
3.2.1 The Case S = 0, c = c(u)
Problems
1. Solve the following
a.
u
t
= 0 subject to u(x, 0) = g(x)
b.
u
t
= 3xu subject to u(x, 0) = g(x)
2. Solve
u
t
= u
subject to
u(x, t) = 1 + cos x along x + 2t = 0
3. Let
u
t
+ c
u
x
= 0 c = constant
a. Solve the equation subject to u(x, 0) = sin x
b. If c > 0, determine u(x, t) for x > 0 and t > 0 where
u(x, 0) = f(x) for x > 0
u(0, t) = g(t) for t > 0
4. Solve the following linear equations subject to u(x, 0) = f(x)
a.
u
t
+ c
u
x
= e
3x
b.
u
t
+ t
u
x
= 5
c.
u
t
t
2
u
x
= u
d.
u
t
+ x
u
x
= t
e.
u
t
+ x
u
x
= x
5. Determine the parametric representation of the solution satisfying u(x, 0) = f(x),
a.
u
t
u
2
u
x
= 3u
65
b.
u
t
+ t
2
u
u
x
= u
6. Solve
u
t
+ t
2
u
u
x
= 5
subject to
u(x, 0) = x.
7. Solve
u
t
+ xu
u
x
= 5 + u
subject to
u(x, 0) = x.
66
1.
a. Integrate the PDE assuming x xed, we get
u(x, t) = K(x)
Since dx/dt = 0 we have x = x
0
and thus
u(x, t) = u(x
0
, 0) = K(x
0
) = g(x
0
) = g(x)
u(x, t) = g(x)
b. For a xed x, we can integrate the PDE with respect to t
_
du
u
= 3xt + K(x)
ln u ln c(x) = 3xt
u(x, t) = ce
3xt
Using the initial condition
u(x, t) = f(x) e
3xt
67
2. The set of ODEs are
dx
dt
= 0 and
du
dt
= u
The characteristics are x = constant and the ODE for u can be written
du
u
= dt
Thus
u(x, t) = k(x) e
t
On x = 2t or x + 2t = 0 we have
1 + cos x = k(x) e
t
[
x=2t
= k(x) e
x
2
Thus the constant of integration is
k(x) = e
x
2
(1 + cos x)
Plug this in the solution u we get
u(x, t) = (1 + cos x) e
x
2
+t
Another way of getting the solution is by a rotation so that the line x +2t = 0 becomes
horizontal. Call that axis , the line perpendicular to it is given by t 2x = 0, which we
call .
So here is the transformation
= x + 2t
= t 2x.
The PDE becomes:
u
+
1
2
u
=
1
2
u
and the intial condition is:
u(, = 0) = 1 + cos
2
5
[
=0
= 1 + cos
2
5
Rewrite this as a system of two rst order ODEs,

d
d
=
1
2
(0) =
du((), )
d
=
1
2
u
68
u((0), 0) = 1 + cos
2
5
.
The solution of the rst ODE, gives the characteristics in the transformed domain:
=
1
2
+
The solution of the second ODE:
u((), ) = Ke
1
2
Using the initial condition

1 + cos
2
5
= K
Thus
u((), ) = (1 + cos
2
5
)e
1
2
But =
1
2
thus
u((), ) = (1 + cos
2
5
(
1
2
))e
1
2
Now substitute back:

1
2
=
1
2
x + t

1
2
= (t 2x) (
1
2
x + t) =
5
2
x
Thus
u(x, t) = (1 + cos x)e
1
2
x + t
.
69
3. a. The set of ODEs to solve is
dx
dt
= c
du
dt
= 0
The characteristics are:
x = x
0
+ ct
The solution of the second ODE is
u(x, t) = constant = u(x
0
, 0) = sin x
0
Substitute for x
0
, we get
u(x, t) = sin(x ct)
b. For x > ct the solution is u(x, t) = f(x ct)
But f(x) is dened only for positive values of the independent variable x, therefore
f(x ct) cannot be used for x < ct.
In this case (x < ct) we must use the condition
u(0, t) = g(t)
The characteristics for which x
0
< 0 is given by x = x
0
+ ct and it passes through the
point (0, t
0
) (see gure). Thus x = c(t t
0
) and u(0, t
0
) = g(t
0
) = g
_
t
x
c
_
4 2 0 2 4 6 8
2
1
0
1
2
3
4
x
t
x ct > 0
x ct < 0
xct=0
u(x,0)=f(x)
u(0,t)=g(t)
Figure 14: Domain and characteristics for problem 3b
The solution is therefore given by
u(x, t) =
f(x ct) for x ct > 0

g
_
t
x
c
_
for x ct < 0
70
4. a. The set of ODEs is
dx
dt
= c
du
dt
= e
3x
The solution of the rst is
x = x
0
+ ct
Substituting x in the second ODE
du
dt
= e
3(x
0
+ct)
Now integrate
u(x, t) = K + e
3x
0
1
3c
e
3ct
At t = 0 we get
f(x
0
) = u(x
0
, 0) = K + e
3x
0
1
3c
Therefore the constant of integration K is
K = f(x
0
) + e
3x
0
1
3c
Substitute this K in the solution
u(x, t) = f(x
0
) + e
3x
0
1
3c
e
3x
0
1
3c
e
3ct
Recall that x
0
= x ct thus
u(x, t) = f(x ct) +
1
3c
e
3(xct)
1
3c
e
3x
b. The set of ODEs is
dx
dt
= t
du
dt
= 5
x = x
0
+
1
2
t
2
Now integrate the second ODE
u(x, t) = 5t + K
At t = 0 the solution is
u(x
0
, 0) = f(x
0
) = K plug t = 0 in the solution u
Thus when substituting for x
0
in the solution
u(x, t) = 5t + f
_
x
1
2
t
2
_
71
c. The set of ODEs is
dx
dt
= t
2
du
dt
= u
x = x
0

1
3
t
3
ln u(x, t) = t + ln K
or
u(x, t) = K e
t
u(x
0
, 0) = f(x
0
0
in the solution
u(x, t) = e
t
f
_
x +
1
3
t
3
_
d. The set of ODEs is
dx
dt
= x
du
dt
= t
ln x = ln x
0
+ t
or
x = x
0
e
t
u(x, t) =
1
2
t
2
+ K
u(x
0
, 0) = f(x
0
0
in the solution
u(x, t) =
1
2
t
2
+ f
_
xe
t
_
72
e. The set of ODEs is
dx
dt
= x
du
dt
= x
ln x = ln x
0
+ t
or
x = x
0
e
t
Now substitute x in the second ODE
du
dt
= x
0
e
t
and integrate it
u(x, t) = e
t
x
0
+ K
u(x
0
, 0) = f(x
0
) = K + x
0
plug t = 0 in the solution u
Thus when substituting K in u
u(x, t) = x
0
e
t
+ f(x
0
) x
0
0
in the solution
u(x, t) = x + f
_
xe
t
_
xe
t
73
5. a. The set of ODEs is
dx
dt
= u
2
du
dt
= 3u
The solution of the rst ODE requires the yet unknown u thus we tackle the second ODE
du
u
= 3 dt
Now integrate this
lnu(x, t) = 3t + K or u(x, t) = C e
3t
u(x
0
, 0) = f(x
0
) = C
Thus
u(x, t) = f(x
0
) e
3t
Now substitute this solution in the characteristic equation (rst ODE)
dx
dt
=
_
f(x
0
) e
3t
_
2
= (f(x
0
))
2
e
6t
Integrating
x = (f(x
0
))
2
_
e
6t
dt =
1
6
(f(x
0
))
2
e
6t
+ K
For t = 0 we get
x
0
=
1
6
(f(x
0
))
2
+ K
Thus
K = x
0
+
1
6
(f(x
0
))
2
and the characteristics are
x =
1
6
(f(x
0
))
2
e
6t
+ x
0
+
1
6
(f(x
0
))
2
Solve this for x
0
and subsitute for u. The quote is because one can only solve this for
special cases of the function f(x
0
).
The implicit solution is given by
u(x, t) = f(x
0
) e
3t
x =
1
6
(f(x
0
))
2
e
6t
+ x
0
+
1
6
(f(x
0
))
2
74
b. The set of ODEs is
dx
dt
= t
2
u
du
dt
= u
du
u
= dt
Now integrate this
ln u(x, t) = t + K or u(x, t) = C e
t
u(x
0
, 0) = f(x
0
) = C
Thus
u(x, t) = f(x
0
) e
t
dx
dt
= t
2
f(x
0
) e
t
or _
dx = f(x
0
)
_
t
2
e
t
dt
Integrate and continue as in part a of this problem
x = f(x
0
)
_
t
2
e
t
2t e
t
2 e
t
+ C
_
For t = 0 we get
x
0
= f(x
0
) [ 2 + C]
Thus
C f(x
0
) = x
0
+ 2 f(x
0
)
x = f(x
0
)
_
t
2
2t 2
_
e
t
+ x
0
+ 2 f(x
0
)
Solve this for x
0
and subsitute for u. The quote is because one can only solve this for
special cases of the function f(x
0
).
The implicit solution is given by
u(x, t) = f(x
0
) e
t
x = f(x
0
) [ t
2
+ 2t + 2 ] e
t
+ x
0
+ 2 f(x
0
)
75
6. The set of ODEs is
dx
dt
= t
2
u
du
dt
= 5
du = 5 dt
Now integrate this
u(x, t) = 5t + K
u(x
0
, 0) = f(x
0
) = x
0
= K
Thus
u(x, t) = x
0
+ 5t
dx
dt
= 5t
3
+ x
0
t
2
Integrate
x =
5
4
t
4
+
1
3
t
3
x
0
+ C
For t = 0 we get
x
0
= 0 + 0 + C
Thus
C = x
0
x =
5
4
t
4
+
_
1
3
t
3
+ 1
_
x
0
Solve this for x
0
x
0
=
x
5
4
t
4
1 +
1
3
t
3
The solution is then given by u(x, t) = 5t +
x
5
4
t
4
1 +
1
3
t
3
76
dx
dt
= xu
du
dt
= 5 + u
du
dt
u = 5
This is a rst order linear constant coecient equation, so the solution is
u(x, t) = 5 + Ke
t
u(x
0
, 0) = f(x
0
) = x
0
= K 5
Thus
u(x, t) = (5 + x
0
) e
t
5
dx
dt
= x
_
(5 + x
0
) e
t
5
_
or
dx
x
=
_
(5 + x
0
) e
t
5
_
dt
Integrate
ln(x) = (5 + x
0
) e
t
5t + C
For t = 0 we get
ln(x
0
) = 5 + x
0
0 + C
Thus
C = ln(x
0
) x
0
5
ln(x) = (5 + x
0
) e
t
5t + ln(x
0
) x
0
5
or
ln
_
x
x
0
_
= (5 + x
0
) e
t
5t x
0
5
or
ln
_
x
x
0
_
= x
0
_
e
t
1
_
5t + 5e
t
5
or
x(t) = x
0
e
x
0(e
t
1) 5t+5e
t
5
The solution is then given by u(x, t) = (5 + x
0
) e
t
5 with x
0
given implicitely above
77
3.2.3 Fan-like Characteristics
3.2.4 Shock Waves
Problems
1. Consider Burgers equation
t
+ u
max
_
1
2
max
_

x
=
x
2
Suppose that a solution exists as a density wave moving without change of shape at a velocity
V , (x, t) = f(x V t).
a. What ordinary dierential equation is satised by f
b. Show that the velocity of wave propagation, V , is the same as the shock velocity
separating =
1
from =
2
(occuring if = 0).
2. Solve
t
+
2
x
= 0
subject to
(x, 0) =
_
4 x < 0
3 x > 0
3. Solve
u
t
+ 4u
u
x
= 0
subject to
u(x, 0) =
_
3 x < 1
2 x > 1
4. Solve the above equation subject to
u(x, 0) =
_
2 x < 1
3 x > 1
5. Solve the quasilinear equation
u
t
+ u
u
x
= 0
subject to
u(x, 0) =
_
2 x < 2
3 x > 2
6. Solve the quasilinear equation
u
t
+ u
u
x
= 0
78
subject to
u(x, 0) =
0 x < 0
x 0 x < 1
1 1 x
7. Solve the inviscid Burgers equation
u
t
+ uu
x
= 0
u (x, 0) =
2 for x < 0
1 for 0 < x < 1
0 for x > 1
Note that two shocks start at t = 0 , and eventually intersect to create a third shock.
Find the solution for all time (analytically), and graphically display your solution, labeling
all appropriate bounding curves.
79
1. a. Since
(x, t) = f(x V t)
we have (using the chain rule)
t
= f
(x V t) (V )
x
= f
(x V t) 1
xx
= f
(x V t)
Substituting these derivatives in the PDE we have
V f
(x V t) + u
max
_
1
2f(x V t)
max
_
f
(x V t) = f
(x V t)
This is a second order ODE for f.
b. For the case = 0 the ODE becomes
V f
(x V t) + u
max
_
1
2f(x V t)
max
_
f
(x V t) = 0
Integrate (recall that the integral of 2ff
is f
2
)
V f(x V t) + u
max
_
f(x V t)
f
2
(x V t)
max
_
= C
To nd the constant, we use the following
As x ,
2
and as x ,
1
, then
V
2
+ u
max
_
2

2
2
max
_
= C
V
1
+ u
max
_
1

2
1
max
_
= C
Subtract
V (
1

2
) + u
max
_
2

2
2
max
_
u
max
_
1

2
1
max
_
= 0
Solve for V
V =
u
max
_
2

2
2
max
_
u
max
_
1

2
1
max
_
2

1
(1)
This can be written as
V = u
max

u
max
max
(
1
+
2
)
Note that (1) is
V =
[q]
[]
since
q = u
max
_

2
max
_
Thus V given in (1) is exactly the shock speed.
80
dx
dt
=
2
d
dt
= 0
The solution of the rst ODE requires the yet unknown thus we tackle the second ODE
d = 0
Now integrate this
(x, t) = K
(x
0
, 0) = K =
_
4 x < 0
3 x > 0
Thus
(x, t) = (x
0
, 0)
dx
dt
=
2
(x
0
, 0)
Integrate
x =
2
(x
0
, 0) t + C
For t = 0 we get
x
0
= 0 + C
Thus
C = x
0
x =
2
(x
0
, 0) t + x
0
For x
0
< 0 then (x
0
, 0) = 4 and the characteristic is then given by x = x
0
+ 16t
Therefore for x
0
= x 16t < 0 the solution is = 4.
For x
0
> 0 then (x
0
, 0) = 3 and the characteristic is then given by x = x
0
+ 9t
Therefore for x
0
= x 9t < 0 the solution is = 3.
Notice that there is a shock (since the value of is decreasing with increasing x). The
shock characteristc is given by
dx
s
dt
=
1
3
4
3
1
3
3
3
4 3
=
1
3
(64 27)
1
=
37
3
The solution of this ODE is
x
s
=
37
3
t + x
s
(0)
x
s
(0) is where the shock starts, i.e. the discontinuity at time t = 0.
81
2 1 0 1 2 3 4 5 6 7 8
0.2
0
0.2
0.4
0.6
0.8
x
t
x
s
=(37/3)t
=3
=4
Figure 15: Characteristics for problem 2
Thus x
s
(0) = 0 and the shock characteristic is
x
s
=
37
3
t
See gure for the characteristic curves including the shocks. The solution in region I above
the shock chracteristic is = 4 and below (region II) is = 3.
82
3.
u
t
+ 4uu
x
= 0 or u
t
+ (2u
2
)
x
= 0
u(x, 0) =
3 x < 1
2 x > 1
Shock again
The shock characteristic is obtained by solving:
dx
s
dt
=
2 3
2
2 2
2
3 2
= 10
x
s
= 10t + x
s
(0)
. .
=1
x
s
= 10t + 1
Now we solve the ODE for u:
du
dt
= 0 u(x, t) = u(x
0
, 0) away from shock
The ODE for x is:
dx
dt
= 4u = 4u(x
0
, 0)
x = 4u(x
0
, 0))t + x
0
If x
0
< 1 x
0
= x 12t x < 1 + 12t
x
0
< 1 x
0
= x 8t x > 1 + 8t
83
4. Solve
u
t
+ 4uu
x
= 0
u(x, 0) =
2 x < 1
3 x > 1
du
dt
= 0 u(x, t) = u(x
0
, 0)
dx
dt
= 4u = 4u(x
0
, 0)
dx = 4u (x
0
, 0) dt
x = 4u (x
0
, 0) t + x
0
For x
0
< 1 x = 8t + x
0
x 8t < 1
x
0
> 1 x = 12t + x
0
x 12t > 1
u(x, t) =
2 x < 8t 1
? 8t 1 < x < 12t 1
3 x > 12t 1
x = 4u t + x
0
..
=1 discontinuity
u =
x + 1
4t
84
5 0 5 10 15 20 25 30 35 40 45
2
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
3
x=t1 x=12t1
u=(x+1)/(4t)
u=2
u=3
Figure 16: Solution for 4
5.
u
t
+ uu
x
= 0
u(x, 0) =
2 x < 2
3 x > 2
fan
du
dt
= 0 u(x, t) = u(x
0
, 0)
dx
dt
= u = u(x
0
, 0)
x = t u (x
0
, 0) + x
0
For x
0
< 2 x = 2t + x
0
x 2t < 2
For x
0
> 2 x = 3t + x
0
x 3t > 2
x = t u (x
0
, 0) + x
0
at discontinuity x
0
= 2
we get x = t u + 2
u =
x 2
t
85
5 0 5 10 15 20 25
2
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
3
x=2t+2
x=3t+2
u=(x2)/t
u=2
u=3
3 2 1 0 1 2 3
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x
u
Figure 18: Sketch of initial solution
6.
u
t
+ uu
x
= 0
u(x, 0) =
0 x < 0
x 0 x < 1
1 x 1
du
dt
= 0 u(x, t) = u(x
0
, 0)
86
5 0 5 10
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x=t+1
u=x/(1+t)
u=0
u=1
dx
dt
= u = u(x
0
, 0) x = t u (x
0
, 0) + x
0
For x
0
< 0 x = t 0 + x
0
x = x
0
u = 0
0 x
0
< 1 x = t x
0
+ x
0
x = x
0
(1 + t) u = x
0
=
x
1 +t
1 x
0
x = t + x
0
x = t + x
0
u = 1
Basically the interval [0, 1] is stretched in time to [0, 1 +t].
87
7.
u
t
+ uu
x
= 0
u (x, 0) =
2 for x < 0
1 for 0 < x < 1
0 for x > 1
First nd the shock characteristic for those with speed u = 2 and u = 1
[q] =
1
2
u
2
2
1
=
1
2
(2
2
1
2
) =
3
2
[u] = 2 1 = 1
Thus
dx
s
dt
=
3
2
and the characteristic through x = 0 is then
x
s
=
3
2
t
Similarly for the shock characteristic for those with speed u = 1 and u = 0
[q] =
1
2
u
2
1
0
=
1
2
(1
2
0
2
) =
1
2
[u] = 1 0 = 1
Thus
dx
s
dt
=
1
2
and the characteristic through x = 1 is then
x
s
=
1
2
t + 1
Now these two shock charateristic going to intersect. The point of intersection is found
by equating x
s
in both, i.e.
1
2
t + 1 =
3
2
t
The solution is t = 1 and x
s
=
3
2
. Now the speeds are u = 2 and u = 0
[q] =
1
2
u
2
2
0
=
1
2
(2
2
0
2
) = 2
88
3 2 1 0 1 2 3 4
0
0.5
1
1.5
2
2.5
3
u=0
u=1
u=2
[u] = 2 0 = 2
Thus
dx
s
dt
= 1
and the characteristic is then
x
s
= t + C.
To nd C, we substitute the point of intersection t = 1 and x
s
=
3
2
. Thus
3
2
= 1 + C
or
C =
1
2
The third shock characteristic is then
x
s
= t +
1
2
.
The shock characteristics and the solutions in each domain are given in the gure above.
89
3.3 Second Order Wave Equation
3.3.1 Innite Domain
Problems
1. Suppose that
u(x, t) = F(x ct).
Evaluate
a.
u
t
(x, 0)
b.
u
x
(0, t)
2. The general solution of the one dimensional wave equation
u
tt
4u
xx
= 0
is given by
u(x, t) = F(x 2t) + G(x + 2t).
Find the solution subject to the initial conditions
u(x, 0) = cos x < x < ,
u
t
(x, 0) = 0 < x < .
3. In section 3.1, we suggest that the wave equation can be written as a system of two rst
order PDEs. Show how to solve
u
tt
c
2
u
xx
= 0
using that idea.
90
1a.
u(x, t) = F(x ct)
Use the chain rule:
u
t
= c
dF(x ct)
d(x ct)
at t = 0
u
t
= c
dF(x)
dx
1b.
u
x
=
dF(x ct)
d(x ct)
1
at x = 0
u
x
=
dF(ct)
d(ct)
=
1
c
dF(ct)
dt
= F
(ct)
dierentiation with respect to argument

91
2. u (x, t) = F (x 2t) + G(x + 2t)
u (x, 0) = cos x
u
t
(x, t) = 0
u (x, 0) = F(x) + G(x) = cos x (*)
u
t
(x, t) = 2F
(x 2t) + 2G
(x + 2t)
u
t
(x, 0) = 2F
(x) + 2G
(x) = 0
Integrate F (x) + G(x) = constant = k (#)
solve the 2 equations (*) and (#)
2G(x) = cos x + k
G(x) =
1
2
cos x +
1
2
k
2F(x) = cos x k
F(x) =
1
2
cos x
1
2
k
We need F(x 2t) F(x 2t) =
1
2
cos (x 2t)
1
2
k
G(x + 2t) =
1
2
cos (x + 2t) +
1
2
k
u(x, t) =
1
2
cos (x 2t) +
1
2
cos (x + 2t)
1
2
k +
1
2
k
u(x, t) =
1
2
cos (x 2t) + cos (x + 2t)
92
3. The wave equation
u
tt
c
2
u
xx
= 0
can be written as a system of two rst order PDEs
v
t
c
v
x
= 0
and
u
t
+ c
u
x
= v.
Solving the rst for v, by rewriting it as a system of ODEs
dv
dt
= 0
dx
dt
= c
The characteristic equation is solved
x = ct + x
0
and then
v(x, t) = v(x
0
, 0) = V (x + ct)
where V is the initial solution for v. Now use this solution in the second PDE rewritten as
a system of ODEs
du
dt
= V (x + ct)
dx
dt
= c
The characteristic equation is solved
x = ct + x
0
and then
du
dt
= V (x + ct) = V (x
0
+ 2ct)
Integrating
u(x
0
, t) =
_
t
0
V (x
0
+ 2c)d + K(x
0
)
Change variables
z = x
0
+ 2c
then
dz = 2cd
93
The limits of integration become x
0
and x
0
+ 2ct. Thus the solution
u(x
0
, t) =
_
x
0
+2ct
x
0
1
2c
V (z)dz + K(x
0
)
But x
0
= x ct
u(x, t) =
_
x+ct
xct
1
2c
V (z)dz + K(x ct)
Now break the integral using the point zero.
u(x, t) = K(x ct)
_
xct
0
1
2c
V (z)dz +
_
x+ct
0
1
2c
V (z)dz
The rst two terms give a function of xct and the last term is a function of x+ct, exactly
as we expect from DAlembert solution.
94
3.3.2 Semi-innite String
3.3.3 Semi-innite String with a Free End
Problems
1. Solve by the method of characteristics
2
u
t
2
c
2
2
u
x
2
= 0, x > 0
subject to
u(x, 0) = 0,
u
t
(x, 0) = 0,
u(0, t) = h(t).
2. Solve
2
u
t
2
c
2
2
u
x
2
= 0, x < 0
subject to
u(x, 0) = sin x, x < 0
u
t
(x, 0) = 0, x < 0
u(0, t) = e
t
, t > 0.
3. a. Solve
2
u
t
2
c
2
2
u
x
2
= 0, 0 < x <
subject to
u(x, 0) =
0 0 < x < 2
1 2 < x < 3
0 3 < x
u
t
(x, 0) = 0,
u
x
(0, t) = 0.
b. Suppose u is continuous at x = t = 0, sketch the solution at various times.
4. Solve
2
u
t
2
c
2
2
u
x
2
= 0, x > 0, t > 0
subject to
u(x, 0) = 0,
95
u
t
(x, 0) = 0,
u
x
(0, t) = h(t).
5. Give the domain of inuence in the case of semi-innite string.
96
5 4 3 2 1 0 1 2 3 4 5
1
0
1
2
3
4
5
x
t
u(x,0)=0 and u
t
(x,0)=0
xct<0
xct>0
u(0,t)=h(t)
D(0,t x /c)
P(x,t)
B(ctx,0) A(x+ct,0) C(xct,0)
Figure 21: Domain for problem 1
1. u
tt
c
2
u
xx
= 0
u(x, 0) = 0
u
t
(x, 0) = 0
u(0, t) = h(t)
Solution u(x, t) = F(x ct) + G(x + ct)
(*)
F(x) =
1
2
f(x)
1
2c
_
x
0
g() d = 0
(#)
G(x) =
1
2
f(x) +
1
2c
_
x
0
g() d = 0
since both f(x), g(x) are zero.
Thus for x ct > 0 the solution is zero
(No inuence of boundary at x = 0)
For x ct < 0 u(0, t) = h(t)
F(ct) + G(ct) = h(t)

u(x, t) = F(x ct) + G(x + ct)
97
but argument of F is negative and thus we cannot use (*), instead
F(ct) = h(t) G(ct)
or F(z) = h(
z
c
) G(z) for z < 0
F(x ct) = h(
x ct
c
) G((x ct))
= h(t
x
c
) G(ct x)
u(x, t) = F(x ct) + G(x + ct)
= h(t
x
c
) G(ct x)
. .
zero
+ G(x + ct)
. .
zero
since the arguments are positive and (#) is valid
u(x, t) = h(t
x
c
) for 0 < x < ct
u(x, t) = 0 for x ct > 0
98
5 4 3 2 1 0 1 2 3 4 5
1
0
1
2
3
4
5
x
t
u(x,0)=sin x and u
t
(x,0)=0
x+ct<0
x+ct>0
u(0,t)=e^(t)
2. u
tt
c
2
u
xx
= 0 x < 0
u(x, 0) = sin x x < 0
u
t
(x, 0) = 0 x < 0
u(0, t) = e
t
t > 0
u(x, t) = F(x ct) + G(x + ct)
F(x) =
1
2
sin x
since f = sin x, g = 0
G(x) =
1
2
sin x
From boundary condition
u(0, t) = F(ct) + G(ct) = e
t
If x + ct < 0 no inuence of boundary at x = 0
u(x, t) =
1
2
sin (x ct) +
1
2
sin (x + ct)
= sin x cos ct
.
.
.
after some trigonometric manipulation
If x + ct > 0 then the argument of G is positive and thus
G(ct) = e
t
F(ct)
or G(z) = e
z/c
F(z)
G(x + ct) = e
x +ct
c
F((x + ct))
99
Therefore:
u(x, t) = F (x ct) + G(x + ct)
= F (x ct) + e
x+ct
c
F (x ct)
=
1
2
sin (x ct) + e
x +ct
c

1
2
sin (x ct)
. .
sin (x +ct)
=
1
2
sin (x ct)
1
2
sin (x + ct)
. .
cos ct sin x
+e
x+ct
c
u(x, t) =
sin xcos ct x + ct < 0

sin xcos ct + e
x+ct
c
x + ct > 0
100
3a. u
tt
= c
2
u
xx
0 < x <
u
x
(0, t) = 0
u(x, 0) =
0 0 < x < 2
1 2 < x < 3
0 x > 3
u
t
(x, 0) = 0
u(x, t) = F(x ct) + G(x + ct)
F(x) =
1
2
f(x)
1
2c
_
x
0
g() d =
1
2
f(x)
since
g 0 x > 0
G(x) =
1
2
f(x) +
1
2c
_
x
0
g() d =
1
2
f(x) g 0 x > 0
u(x, t) =
f(x + ct) + f(x ct)
2
, x > ct
u
x
(0, t) = F
(ct) + G
(ct) = 0
F
(ct) = G
(ct)
F
(z) = G
(z)
Integrate
F(z) = G(z) + K
F(z) = G(z) K
F(x ct) = G(ct x) K x ct < 0
=
1
2
f(ct x) K x ct < 0
u(x, t) =
1
2
f(x + ct) +
1
2
f(ct x) K
To nd K we look at x = 0, t = 0 u(0, 0) = 0 from initial condition
but u(0, 0) =
1
2
f(0) +
1
2
f(0) K = f(0) K
. .
=0 from above
K = 0
u(x, t) =
f(x + ct) + f(ct x)
2
101
1 0 1 2 3 4 5 6 7 8 9
1
0
1
2
3
4
5
6
7
8
9
x
t
xct<0
xct>0
Figure 23: Domain of inuence for problem 3
3b.
u(x, t) =
f(x + ct) + f(x ct)

2
x > ct
f(x + ct) + f(ct x)
2
x < ct
where
u(x, t) =
1 Region I
1
2
Region II
0 otherwise
In order to nd the regions I and II mentioned above, we use the idea of domain of
inuence. Sketch both characteristics from the end points of the interval (2,3) and remember
that when the characteristic curve (line in this case) reaches the t axis, it will be reected.
As can be seen in the gure, the only region where the solution is 1 is the two triangular
regions. Within the three strips (not including the above mentioned triangles), the solution
is
1
2
and for the rest, the solution is zero.
102
5 4 3 2 1 0 1 2 3 4 5
1
0
1
2
3
4
5
x
t
u(x,0)=0 and u
t
(x,0)=0
xct<0
xct>0
u
x
(0,t)=h(t)
4. u
tt
c
2
u
xx
= 0
general solution
u(x, t) = F(x ct) + G(x + ct)
For x ct > 0 u(x, t) = 0 since u = u
t
= 0 on the boundary.
For x ct < 0 we get the inuence of the boundary condition
u
x
(0, t) = h(t)
Dierentiate the general solution:
u
x
(x, t) = F
(x ct) 1 + G
(x + ct) 1 =
dF(x ct)
d(x ct)
+
dG(x +ct)
d(x +ct)
.
.
.
chain rule
prime means derivative with respect to argument
As x = 0 :
h(t) = u
x
(0, t) =
dF(ct)
d(ct)
+
dG(ct)
d(ct)
=
1
c
dF(ct)
dt
+
1
c
dG(ct)
dt
103
Integrate
1
c
F(ct) +
1
c
G(ct)
. .
=0
+
1
c
F(0)
. .
=0
1
c
G(0)
. .
=0
=
_
t
0
h() d
since f = g = 0
F(ct) = c
_
t
0
h() d
F(z) = c
_
z/c
0
h() d
F(x ct) = c
_

xct
c
0
h() d
u(x, t) =
0 x ct > 0
c
_

xct
c
0
h() d x ct < 0
104
5. For the innite string the domain of inuence is a wedge with vertex at the point of
interest (x, 0). For the semi innite string, the left characteristic is reected by the vertical
t axis and one obtains a strip, with one along a characteristic (x + ct = C) reaching the t
axis and the other two sides are from the other family of characteristcs (x ct = K).
105
CHAPTER 4
4 Separation of Variables-Homogeneous Equations
4.1 Parabolic equation in one dimension
4.2 Other Homogeneous Boundary Conditions
Problems
1. Consider the dierential equation
X
(x) + X(x) = 0
Determine the eigenvalues (assumed real) subject to
a. X(0) = X() = 0
b. X
(0) = X
(L) = 0
c. X(0) = X
(L) = 0
d. X
(0) = X(L) = 0
e. X(0) = 0 and X
(L) + X(L) = 0
Analyze the cases > 0, = 0 and < 0.
106
1. a.
X
+ X = 0
X(0) = 0
X() = 0
Try e
rx
. As we know from ODEs, this leads to the characteristic equation for r
r
2
+ = 0
Or
r =
We now consider three cases depending on the sign of

Case 1: < 0
In this case r is the square root of a positive number and thus we have two real roots. In
this case the solution is a linear combination of two real exponentials
X = A
1
e
x
+ B
1
e
x
It is well known that the solution can also be written as a combination of hyperbolic sine
and cosine, i.e.
X = A
2
cosh
x + B
2
sinh
x
The other two forms are may be less known, but easily proven. The solution can be written
as a shifted hyperbolic cosine (sine). The proof is straight forward by using the formula for
cosh(a + b) (sinh(a + b))
X = A
3
cosh
_
x + B
3
_
Or
X = A
4
sinh
_
x + B
4
_
Which form to use, depends on the boundary conditions. Recall that the hypebolic sine
vanishes ONLY at x = 0 and the hyperbolic cosine is always positive. If we use the last form
of the general solution then we immediately nd that B
4
= 0 is a result of the rst boundary
condition and clearly to satisfy the second boundary condition we must have A
4
= 0 (recall
sinh x = 0 only for x = 0 and the second boundary condition reads A
4
sinh
= 0,
thus the coecient A
4
must vanish).
Any other form will yields the same trivial solution, may be with more work!!!
Case 2: = 0
In this case we have a double root r = 0 and as we know from ODEs the solution is
X = Ax + B
The boundary condition at zero yields
X(0) = B = 0
107
and the second condition
X() = A = 0
This implies that A = 0 and therefore we again have a trivial solution.
Case 3: > 0
In this case the two roots are imaginary
r = i
Thus the solution is a combination of sine and cosine

X = A
1
cos
x + B
1
sin
x
Substitute the boundary condition at zero
X(0) = A
1
Thus A
1
= 0 and the solution is
X = B
1
sin
x
Now use the condition at
X() = B
1
sin
= 0
If we take B
1
= 0, we get a trivial solution, but we have another choice, namely
sin
= 0
This implies that the argument of the sine function is a multiple of
_
n
= n n = 1, 2, . . .
Notice that since > 0 we must have n > 0. Thus
_
n
= n n = 1, 2, . . .
Or
n
= n
2
n = 1, 2, . . .
The solution is then depending on n, and obtained by substituting for
n
X
n
(x) = sin nx
Note that we ignored the constant B
1
since the eigenfunctions are determined up to a mul-
tiplicative constant. (We will see later that the constant will be incorporated with that of
the linear combination used to get the solution for the PDE)
108
1. b.
X
+ X = 0
X
(0) = 0
X
(L) = 0
Try e
rx
r
2
+ = 0
Or
r =

Case 1: < 0
X = A
1
e
x
+ B
1
e
x
and cosine, i.e.
X = A
2
cosh
x + B
2
sinh
x
X = A
3
cosh
_
x + B
3
_
Or
X = A
4
sinh
_
x + B
4
_
vanishes ONLY at x = 0 and the hyperbolic cosine is always positive. If we use the following
form of the general solution
X = A
3
cosh
_
x + B
3
_
then the derivative X
wil be
X
A
3
sinh
_
x + B
3
_
The rst boundary condition X
(0) = yields B
3
= 0 and clearly to satisfy the second
boundary condition we must have A
3
= 0 (recall sinh x = 0 only for x = 0 and the second
boundary condition reads
A
3
sinh
L = 0, thus the coecient A

3
must vanish).
109
Case 2: = 0
X = Ax + B
X
(0) = A = 0
X
(L) = A = 0
This implies that A = 0 and therefore we have no restriction on B. Thus in this case the
solution is a constant and we take
X(x) = 1
Case 3: > 0
r = i

X = A
1
cos
x + B
1
sin
x
Dierentiate
X
A
1
sin
x +
B
1
cos
x
X
(0) =
B
1
Thus B
1
X = A
1
cos
x
Now use the condition at L
X
(L) =
A
1
sin
L = 0
If we take A
1
sin
L = 0
This implies that the argument of the sine function is a multiple of
_
n
L = n n = 1, 2, . . .
Notice that since > 0 we must have n > 0. Thus
_
n
=
n
L
n = 1, 2, . . .
Or
n
=
_
n
L
_
2
n = 1, 2, . . .
The solution is then depending on n, and obtained by substituting
n
X
n
(x) = cos
n
L
x
110
1. c.
X
+ X = 0
X(0) = 0
X
(L) = 0
Try e
rx
r
2
+ = 0
Or
r =

Case 1: < 0
X = A
1
e
x
+ B
1
e
x
and cosine, i.e.
X = A
2
cosh
x + B
2
sinh
x
X = A
3
cosh
_
x + B
3
_
Or
X = A
4
sinh
_
x + B
4
_
X = A
4
sinh
_
x + B
4
_
wil be
X
A
4
cosh
_
x + B
4
_
The rst boundary condition X(0) = yields B
4
4
= 0 (recall cosh x is never zero thus the coecient A
4
must vanish).
111
Case 2: = 0
X = Ax + B
X(0) = B = 0
X
(L) = A = 0
This implies that B = A = 0 and therefore we have again a trivial solution.
Case 3: > 0
r = i

X = A
1
cos
x + B
1
sin
x
Dierentiate
X
A
1
sin
x +
B
1
cos
x
X(0) = A
1
Thus A
1
X = B
1
sin
x
X
(L) =
B
1
cos
L = 0
If we take B
1
cos
L = 0
This implies that the argument of the cosine function is a multiple of plus /2
_
n
L =
_
n +
1
2
_
n = 0, 1, 2, . . .
Notice that since > 0 we must have n 0. Thus
_
n
=
_
n +
1
2
_
L
n = 0, 1, 2, . . .
112
Or
n
=
_
n +
1
2
_
2
n = 0, 1, 2, . . .
n
X
n
(x) = sin
_
n +
1
2
_
L
x
113
1. d.
X
+ X = 0
X
(0) = 0
X(L) = 0
Try e
rx
r
2
+ = 0
Or
r =

Case 1: < 0
X = A
1
e
x
+ B
1
e
x
and cosine, i.e.
X = A
2
cosh
x + B
2
sinh
x
X = A
3
cosh
_
x + B
3
_
Or
X = A
4
sinh
_
x + B
4
_
X = A
3
cosh
_
x + B
3
_
wil be
X
A
3
sinh
_
x + B
3
_
The rst boundary condition X
(0) = yields B
3
3
3
must vanish).
114
Case 2: = 0
X = Ax + B
X
(0) = A = 0
X(L) = B = 0
Case 3: > 0
r = i

X = A
1
cos
x + B
1
sin
x
Dierentiate
X
A
1
sin
x +
B
1
cos
x
X
(0) =
B
1
Thus B
1
X = A
1
cos
x
X(L) = A
1
cos
L = 0
If we take A
1
cos
L = 0
This implies that the argument of the cosine function is a multiple of plus /2
_
n
L =
_
n +
1
2
_
n = 0, 1, 2, . . .
Notice that since > 0 we must have n 0. Thus
_
n
=
_
n +
1
2
_
L
n = 0, 1, 2, . . .
Or
n
=
_
n +
1
2
_
2
n = 0, 1, 2, . . .
n
X
n
(x) = cos
_
n +
1
2
_
L
x
115
1. e.
X
+ X = 0
X(0) = 0
X
(L) + X(L) = 0
Try e
rx
r
2
+ = 0
Or
r =

Case 1: < 0
X = A
1
e
x
+ B
1
e
x
and cosine, i.e.
X = A
2
cosh
x + B
2
sinh
x
X = A
3
cosh
_
x + B
3
_
Or
X = A
4
sinh
_
x + B
4
_
X = A
4
sinh
_
x + B
4
_
wil be
X
A
4
cosh
_
x + B
4
_
The rst boundary condition X(0) = 0 yields B
4
boundary condition
A
4
cosh
L = 0
we must have A
4
4
must vanish).
116
Case 2: = 0
X = Ax + B
X(0) = B = 0
X
(L) + X(L) = A + AL = 0
Or
A(1 + L) = 0
Case 3: > 0
r = i

X = A
1
cos
x + B
1
sin
x
Dierentiate
X
A
1
sin
x +
B
1
cos
x
X(0) = A
1
Thus A
1
X = B
1
sin
x
X
(L) + X(L) =
B
1
cos
L + B
1
sin
L = 0
If we take B
1
cos
L + sin
L = 0
If cos
L = 0 then we are left with sin
L = 0 which is not possible (the cosine and

sine functions do not vanish at the same points).
Thus cos
L ,= 0 and upon dividing by it we get
= tan
L
This can be solved graphically or numerically (see gure). The points of intersection are
values of
n
. The solution is then depending on n, and obtained by substituting
n
X
n
(x) = sin
_
n
x
117
2 1 0 1 2 3 4 5 6 7 8
14
12
10
8
6
4
2
0
2
4
6
Figure 25: Graphical solution of the eigenvalue problem
118
CHAPTER 5
5 Fourier Series
5.1 Introduction
5.2 Orthogonality
5.3 Computation of Coecients
Problems
1. For the following functions, sketch the Fourier series of f(x) on the interval [L, L].
Compare f(x) to its Fourier series
a. f(x) = 1
b. f(x) = x
2
c. f(x) = e
x
d.
f(x) =
_
1
2
x x < 0
3x x > 0
e.
f(x) =
0 x <
L
2
x
2
x >
L
2
2. Sketch the Fourier series of f(x) on the interval [L, L] and evaluate the Fourier coe-
cients for each
a. f(x) = x
b. f(x) = sin

L
x
c.
f(x) =
1 [x[ <
L
2
0 [x[ >
L
2
3. Show that the Fourier series operation is linear, i.e. the Fourier series of f(x) + g(x)
is the sum of the Fourier series of f(x) and g(x) multiplied by the corresponding constant.
119
8 6 4 2 0 2 4 6 8
1
0.5
0
0.5
1
1.5
2
2.5
3
L L
Figure 26: Graph of f(x) = 1
8 6 4 2 0 2 4 6 8
1
0.5
0
0.5
1
1.5
2
2.5
3
L L 2L 2L 3L 3L
Figure 27: Graph of its periodic extension
1. a. f(x) = 1
Since the periodic extension of f(x) is continuous, the Fourier series is identical to (the
periodic extension of) f(x) everywhere.
120
10 8 6 4 2 0 2 4 6 8 10
1
0
1
2
3
4
5
6
7
8
Figure 28: Graph of f(x) = x
2
10 8 6 4 2 0 2 4 6 8 10
1
0
1
2
3
4
5
6
7
8
1. b. f(x) = x
2
Since the periodic extension of f(x) is continuous, the Fourier series is identical to (the
periodic extension of) f(x) everywhere.
121
10 8 6 4 2 0 2 4 6 8 10
1
0
1
2
3
4
5
6
7
L L
Figure 30: Graph of f(x) = e
x
10 8 6 4 2 0 2 4 6 8 10
1
0
1
2
3
4
5
6
7
L L 2L 2L 3L 3L 4L 4L
1. c. f(x) = e
x
Since the periodic extension of f(x) is discontinuous, the Fourier series is identical to (the
periodic extension of) f(x) everywhere except for the points of discontinuities. At x = L
(and similar points in each period), we have the average value, i.e.
e
L
+ e
L
2
= cosh L
122
10 8 6 4 2 0 2 4 6 8 10
4
2
0
2
4
6
8
L L
Figure 32: Graph of f(x)
10 8 6 4 2 0 2 4 6 8 10
4
2
0
2
4
6
8
L L 2L 2L 3L 3L 4L 4L
1. d.
Since the periodic extension of f(x) is discontinuous, the Fourier series is identical to
(the periodic extension of) f(x) everywhere except at the points of discontinuities. At those
points x = L (and similar points in each period), we have
3L + (
1
2
L)
2
=
5
4
L
123
10 8 6 4 2 0 2 4 6 8 10
4
2
0
2
4
6
8
L L
Figure 34: Graph of f(x)
10 8 6 4 2 0 2 4 6 8 10
4
2
0
2
4
6
8
L L 2L 2L 3L 3L 4L 4L
1. e.
points x = L (and similar points in each period), we have
L
2
+ 0
2
=
1
2
L
2
At the point L/2 and similar ones in each period we have
0 +
1
4
L
2
2
=
1
8
L
2
124
10 8 6 4 2 0 2 4 6 8 10
3
2
1
0
1
2
3
Figure 36: Graph of f(x) = x
10 8 6 4 2 0 2 4 6 8 10
3
2
1
0
1
2
3
2. a. f(x) = x
point x = L (and similar points in each period), we have
L + (L)
2
= 0
Now we evaluate the coecients.
a
0
=
1
L
_
L
L
f(x) dx =
1
L
_
L
L
xdx = 0
Since we have integrated an odd function on a symmetric interval. Similarly for all a
n
.
b
n
=
1
L
_
L
L
f(x) sin
n
L
xdx =
1
L
_
xcos
n
L
x
n
L
[
L
L
+
_
L
L
cos
n
L
x
n
L
dx
_
This was a result of integration by parts.
=
1
L
Lcos n
n
L
+
Lcos(n)
n
L
+
sin
n
L
x
_
n
L
_
2
[
L
L
125
The last term vanishes at both end points L
=
1
L
2Lcos n
n
L
=
2L
n
(1)
n
Thus
b
n
=
2L
n
(1)
n+1
and the Fourier series is
x
2L
n=1
(1)
n+1
n
sin
n
L
x
126
2. b. This function is already in a Fourier sine series form and thus we can read the
coecients
a
n
= 0 n = 0, 1, 2, . . .
b
n
= 0 n ,= 1
b
1
= 1
2. c.
6 4 2 0 2 4 6
4
2
0
2
4
6
8
L L L/2 L/2
Figure 38: graph of f(x) for problem 2c
Since the function is even, all the coecients b
n
will vanish.
a
0
=
1
L
_
L/2
L/2
dx =
1
L
x[
L/2
L/2
=
1
L
(
L
2
(
L
2
)) = 1
a
n
=
1
L
_
L/2
L/2
cos
n
L
xdx =
1
L
L
n
sin
n
L
x[
L/2
L/2
=
1
n
_
sin
n
2
sin
n
2
_
Since the sine function is odd the last two terms add up and we have
a
n
=
2
n
sin
n
2
The Fourier series is
f(x)
1
2
+
n=1
2
n
sin
n
2
cos
n
L
x
127
3.
f(x)
1
2
a
0
+
n=1
_
a
n
cos
n
L
x + b
n
sin
n
L
x
_
g(x)
1
2
A
0
+
n=1
_
A
n
cos
n
L
x + B
n
sin
n
L
x
_
where
a
n
=
1
L
_
L
L
f(x) cos
n
L
xdx
b
n
=
1
L
_
L
L
f(x) sin
n
L
xdx
A
n
=
1
L
_
L
L
g(x) cos
n
L
xdx
B
n
=
1
L
_
L
L
g(x) sin
n
L
xdx
For f(x) + g(x) we have
1
2
0
+
n=1
_
n
cos
n
L
x +
n
sin
n
L
x
_
and the coecients are
0
=
1
L
_
L
L
(f(x) + g(x))dx
which by linearity of the integral is
0
=
1
L
_
L
L
f(x)dx +
1
L
_
L
L
g(x)dx = a
0
+ A
0
Similarly for
n
and
n
.
n
=
1
L
_
L
L
(f(x) + g(x)) cos
n
L
xdx
n
=
1
L
_
L
L
f(x) cos
n
L
xdx +
1
L
_
L
L
g(x) cos
n
L
xdx = a
n
+ A
n
n
=
1
L
_
L
L
(f(x) + g(x)) sin
n
L
xdx
n
=
1
L
_
L
L
f(x) sin
n
L
xdx +
1
L
_
L
L
g(x) sin
n
L
xdx = b
n
+ B
n
128
5.4 Relationship to Least Squares
5.5 Convergence
5.6 Fourier Cosine and Sine Series
Problems
1. For each of the following functions
i. Sketch f(x)
ii. Sketch the Fourier series of f(x)
iii. Sketch the Fourier sine series of f(x)
iv. Sketch the Fourier cosine series of f(x)
a. f(x) =
_
x x < 0
1 + x x > 0
b. f(x) = e
x
c. f(x) = 1 + x
2
d. f(x) =
_
1
2
x + 1 2 < x < 0
x 0 < x < 2
2. Sketch the Fourier sine series of
f(x) = cos

L
x.
Roughly sketch the sum of the rst three terms of the Fourier sine series.
3. Sketch the Fourier cosine series and evaluate its coecients for
f(x) =
1 x <
L
6
3
L
6
< x <
L
2
0
L
2
< x
4. Fourier series can be dened on other intervals besides [L, L]. Suppose g(y) is dened
on [a, b] and periodic with period b a. Evaluate the coecients of the Fourier series.
5. Expand
f(x) =
_
1 0 < x <

2
0

2
< x <
in a series of sin nx.
a. Evaluate the coecients explicitly.
b. Graph the function to which the series converges to over 2 < x < 2.
129
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
Figure 39: Sketch of f(x) and its periodic extension for 1a
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
Figure 40: Sketch of the odd extension and its periodic extension for 1a
1. a.
The Fourier series is the same as the periodic extension except for the points of discon-
tinuities where the Fourier series yields
1 + 0
2
=
1
2
For the Fourier sine series we take ONLY the right branch on the interval [0, L] and
extend it as an odd function.
Now the same discontinuities are there but the value of the Fourier series at those points
is
1 + (1)
2
= 0
For the Fourier cosine series we need an even extension
Note that the periodic extension IS continuous and the Fourier series gives the exact
same sketch.
130
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
Figure 41: Sketch of the even extension and its periodic extension for 1a
8 6 4 2 0 2 4 6 8
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
0
2
4
6
L L
Figure 42: Sketch of f(x) and its periodic extension for 1b
1. b.
e
L
+ e
L
2
= cosh L
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
Figure 43: Sketch of the odd extension and its periodic extension for 1b
131
8 6 4 2 0 2 4 6 8
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
0
2
4
6
L L
Figure 44: Sketch of the even extension and its periodic extension for 1b
is
1 + (1)
2
= 0
same sketch.
132
8 6 4 2 0 2 4 6 8
1
0
1
2
3
4
5
L L
8 6 4 2 0 2 4 6 8
0
2
4
6
L L
Figure 45: Sketch of f(x) and its periodic extension for 1c
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
Figure 46: Sketch of the odd extension and its periodic extension for 1c
1. c.
The Fourier series is the same as the periodic extension. In fact the Fourier cosine series
is the same!!!
is
1 + (1)
2
= 0
133
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
Figure 47: Sketch of f(x) and its periodic extension for 1d
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
Figure 48: Sketch of the odd extension and its periodic extension for 1d
1. d.
1 + 0
2
=
1
2
for x = 0 + multiples of 4
1 + 2
2
=
3
2
for x = 2 + multiples of 4
Now some of the same discontinuities are there but the value of the Fourier series at
those points is
2 + (2)
2
= 0 for x = 2 + multiples of 4
At the other previous discontinuities we now have continuity.
same sketch.
134
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
Figure 49: Sketch of the even extension and its periodic extension for 1d
8 6 4 2 0 2 4 6 8
4
3
2
1
0
1
2
3
4
L L
Figure 50: Sketch of the odd extension for 2
2.
cos
x
L
=
n=1
b
n
sin
n
L
x
bn =
0 nodd
4n
(n
2
1)
neven
Since we have a Fourier sine series, we need the odd extension of f(x)
Now extend by periodicity
At points of discontinuity the Fourier series give zero.
135
8 6 4 2 0 2 4 6 8
4
3
2
1
0
1
2
3
4
L L
Figure 51: Sketch of the periodic extension of the odd extension for 2
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
Figure 52: Sketch of the Fourier sine series for 2
First two terms of the Fourier sine series of cos
x
L
are
= b
2
sin
2 x
L
+ b
4
sin
4 x
L
=
8
3
sin
2
L
x +
16
15
sin
4
L
x
136
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
8 6 4 2 0 2 4 6 8
6
4
2
0
2
4
6
L L
Figure 53: Sketch of f(x) and its periodic extension for problem 3
3.
f(x) =
1 x < L/6
3 L/6 < x < L/2
0 x > L/2
Fourier cosine series coecients:
a
0
=
2
L
_
_
L/6
0
dx +
_
L/2
L/6
3 dx
_
=
2
L
_
L
6
+ 3
_
L
2

L
6
__
=
2
L
_
L
6
+ 3
2L
6
_
=
2
L

7
6
L =
7
3
a
n
=
2
L
_
_
L/6
0
cos
n
L
x dx + 3
_
L/2
L/6
cos
n
L
xdx
_
=
2
L
_
sin
n
L
x
n
L
[
L/6
0
+ 3
sin
n
L
x
n
L
[
L/2
L/6
_
=
2
L
_
sin
n
6
n
L
+ 3
sin
n
2
sin
n
6
n
6
_
=
2
L
L
n
3 sin
n
2
. .
0 for n even
2 sin
n
6
=
2
n
_
3 sin
n
2
2 sin
n
6
_
1 for n odd
137
10 5 0 5 10
5
0
5
L L
10 5 0 5 10
5
0
5
L L
Figure 54: Sketch of the even extension of f(x) and its periodic extension for problem 3
4.
x [ L, L]
y [ a, b ]
then y =
a +b
2
+
b a
2L
x () is the transformation required (Note that if x = L then
y = a and if x = L then y = b)
g(y) is periodic of period b a
g(y) = G(x) =
a
0
2
+
n=1
(a
n
cos
n
L
x + b
n
sin
n
L
x)
a
n
=
1
L
_
L
L
G(x) cos
n
L
xdx
solving (*) for x yields
x =
2L
b a
_
y
a + b
2
_
dx =
2L
b a
dy
a
n
=
1
L
_
b
a
g (y) cos
_
n
L
2L
b a
_
y
a + b
2
__
2L
b a
dy
Therefore
a
n
=
2
b a
_
b
a
g (y) cos
_
2n
b a
_
y
a + b
2
__
dy
Similarly for b
n
b
n
=
2
b a
_
b
a
g (y) sin
_
2n
b a
_
y
a + b
2
__
dy
138
10 5 0 5 10
2
1
0
1
2
10 5 0 5 10
2
1
0
1
2
Figure 55: Sketch of the periodic extension of the odd extension of f(x) (problem 5)
5.
f(x) =
1 0 < x < /2
0 /2 < x <
Expand in series of sin nx
f(x)
n=1
b
n
sin nx
b
n
=
2
_

0
f(x) sin nxdx =
2
_
/2
0
1 sin nxdx
f(x) =zero on the rest

=
2
1
n
cos nx
_
[
/2
0
=
2
n
cos
n
2
. .
+
2
n
this takes the values 0, 1 depending on n!!!
139
5.7 Term by Term Dierentiation
Problems
1. Given the Fourier sine series
cos

L
x
n=1
b
n
sin
n
L
x
b
n
=
_
0 n is odd
4n
(n
2
1)
n is even
Determine the Fourier cosine series of sin

L
x.
2. Consider
sinh x
n=1
a
n
sin nx.
Determine the coecients a
n
by dierentiating twice.
140
1. Fourier cosine series of sin
x
L
using
cos
x
L
. .
f(x)
=
n=1
b
n
sin
n
L
x
with
b
n
=
0 n odd
4n
(n
2
1)
n even
L
sin

L
x
1
L
[f(L) f(0)] +
n=1
_
n
L
b
n
+
2
L
[(1)
n
f(L) f(0) ]
_
cos
n
L
x
f(L) = cos = 1
f(0) = cos 0 = 1
=
2
L
+
n=1
_
n
L
b
n
+
2
L
[(1)
n+1
1 ]
_
cos
n
L
x
sin
x
L
=
L
2
L
+
n=1
_
n
L
b
n

2
L
_
(1)
n
+ 1
. .
__
cos
n
L
x
=0 if n is odd
=2 if n is even
Substitute for b
n
sin

L
x =
2
n=2
..
n even
n
2
(n
2
1)
1
. .
1
n
2
1
cos
n
L
x
141
Check constant term
1
2
a
0
where
a
0
=
2
L
_
L
0
sin

L
xdx =
2
L
_
cos

L
x
_
[
L
0
=
2
(1 1) =
4

1
2
a
0
=
2
This agrees with previous result.

142
2. sinh x
n=1
a
n
sin nx
Dierentiate (since this is Fourier sine series and sinh ,= 0, we have to use the formula)
cosh x
1
(sinh sinh 0) +
n=1
_
na
n
+
2
[(1)
n
sinh sinh 0]
_
cos nx
Dierentiate again (this time we have Fourier cosine series)
sinh x 0 +
n=1
(n)
_
na
n
+
2
(1)
n
sinh
_
sin nx
a
n
= n
2
a
n
+ (1)
n+1
n
2
sinh
a
n
=
(1)
n+1
n
2
sinh
1 + n
2
If we go thru integration
a
n
=
2
0
sinh x sin nxdx
we get the same answer.
143
5.8 Term by Term Integration
Problems
1. Consider
x
2
n=1
a
n
sin
n
L
x
a. Determine a
n
by integration of the Fourier sine series of f(x) = 1, i.e. the series
1
4
n=1
1
2n 1
sin
2n 1
L
x
b. Derive the Fourier cosine series of x
3
from this.
2. Suppose that
cosh x
n=1
b
n
sin
n
L
x
a. Determine the coecients b
n
by dierentiating twice.
b. Determine b
n
by integrating twice.
3. Evaluate
n=1
1
(2n 1)
2
by using the integration of Fourier sine series of f(x) = 1 (see problem 1 part a.)
144
1a. 1 =
4
n=1
1
2n 1
sin
2n 1
L
x
Integrate
x =
4
n=1
1
2n 1
L
(2n 1)
cos
2n 1
L
x [
x
0
=
4L
n=1
1
(2n 1)
2
cos
2n 1
L
x +
4L
n=1
1
(2n 1)
2
. .
constant term
To nd the constant term
1
2
a
0
a
0
=
2
L
_
L
0
xdx =
2
2L
x
2
[
L
0
=
L
2
L
= L
x =
L
2

4L
n=1
1
(2n 1)
2
cos
2n 1
L
x
Integrate again
x
2
2
=
L
2
x
4L
n=1
1
(2n 1)
2
L
(2n 1)
3
sin
2n 1
L
x [
x
0
x
2
2
=
L
2
x
4L
2
n=1
1
(2n 1)
3
sin
2n 1
L
x
We need Fourier sine series of x to complete the work
See 2a in Section 5.3
x =
n=1
2L
n
(1)
n+1
sin
n
L
x
x
2
= L
n=1
2L
n
(1)
n+1
sin
n
L
x
8L
2
n=1
1
(2n 1)
3
sin
2n 1
L
x
. .
n=1, 3,
1
n
3
sin
n
L
x
145
x
2
=
a
n
sin
n
L
x
where (only for odd n we get contribution from both sums)
a
1
=
2L
2

8L
2
3
a
2
=
2L
2
2
a
3
=
2L
2
3

8L
2
3
.
1
3
3
and so on.
146
1b. x
2
=
n=1
a
n
sin
n
L
x
To get the cosine series for x
3
, lets integrate
x
3
3
+ C =
n=1
a
n
cos
n
L
x
n
L
Integrate again on [0, L] the integral of cos
n
L
x will give zero
_
L
0
x
3
3
dx + C x[
L
0
= 0
x
4
12
[
L
0
+ C L = 0
L
4
12
+ C L = 0
C =
L
3
12
x
3
=
L
3
4

3L
n=1
a
n
n
cos
n
L
x
where a
n
as in 1a
This problem can be done as in 1a.
Integrate the series from 0 to x
x
3
3
=
n=1
a
n
1
n
L
n=1
a
n
cos
n
L
x
n
L
Multiply by 3
x
3
=
n=1
a
n
3
n
L
. .
constant term=1/2a
0
3
n=1
a
n
cos
n
L
x
n
L
To nd the constant term we evaluate diredtly
a
0
=
2
L
_
L
0
x
3
dx =
2
L
x
4
4
[
L
0
=
2
L
L
4
4
=
L
3
2
147
8 6 4 2 0 2 4 6 8
8
6
4
2
0
2
4
6
8
L L
Figure 56: Sketch of the even extension of f(x) = x
3
(problem 1b)
2. cosh x
. .
even function on [-L, L]
n=1
b
n
sin
n
L
x
To get sine series we must take the odd extension
a. sinh x
1
L
cosh L cosh 0
. .
=1
. .
constant
+
n=1
n
L
b
n
+
2
L
(1)
n
cosh L cosh 0
. .
=1
cos
n
L
x
Dierentiate again
cosh x 0 +
n=1
_
n
L
b
n
+
2
L
[(1)
n
cosh L 1]
_
(
n
L
sin
n
L
x)
=
n=1
_
_
n
L
_
2
b
n

2n
L
2
[(1)
n
cosh L 1]
_
sin
n
L
x
b
n
=
_
n
L
_
2
b
n

2n
L
2
[(1)
n
cosh L 1]
b
n
=

2n
L
2
[(1)
n
cosh L 1]
1 +
_
n
L
_
2
148
b. Integrations
rst
sinh x =
1
2
A
0
+
n=1
_
L
n
_
b
n
cos
n
L
x
where A
0
=
2
L
_
L
0
sinh xdx =
2
L
(cosh L 1)
second
cosh x 1 = A
0
x +
n=1
_
_
L
n
_
2
b
n
sin
n
L
x
_

b
n
[1 +
_
L
n
_
2
] sin
n
L
x = 1 + A
0
x
Note (Done previously)
x
n=1
c
n
sin
n
L
x
where
c
n
=
2L
n
(1)
n+1
and
1
d
n
sin
n
L
x
where
d
n
=
0 n even
4
n
n odd
=
2
n
[1 (1)
n
]
b
n
[1 +
_
L
n
_
2
] =
2
n
[1 (1)
n
] +
1
L
[cosh L 1]
2L
n
(1)
n+1
which yields the same answer after longer computations.
149
3. Evaluate
1 +
1
3
2
+
from 1a:
x =
L
2

4L
2
_
cos
x
L
+
cos
3
L
x
3
2
+
_
at x = 0
0 =
L
2

4L
2
_
1 +
1
3
2
+
_
1 +
1
3
2
+
1
5
2
+ =
L
2

2
4L
=

2
8
150
5.9 Full solution of Several Problems
Problems
1. Solve the heat equation
u
t
= ku
xx
, 0 < x < L, t > 0,
subject to the boundary conditions
u(0, t) = u(L, t) = 0.
Solve the problem subject to the initial value:
a. u(x, 0) = 6 sin
9
L
x.
b. u(x, 0) = 2 cos
3
L
x.
u
t
= ku
xx
, 0 < x < L, t > 0,
subject to
u
x
(0, t) = 0, t > 0
u
x
(L, t) = 0, t > 0
a. u(x, 0) =
0 x <
L
2
1 x >
L
2
b. u(x, 0) = 6 + 4 cos
3
L
x.
3. Solve the eigenvalue problem
=
subject to
(0) = (2)
(0) =
(2)
4. Solve Laplaces equation inside a wedge of radius a and angle ,
2
u =
1
r
r
_
r
u
r
_
+
1
r
2
2
u
2
= 0
subject to
u(a, ) = f(),
u(r, 0) = u
(r, ) = 0.
151
5. Solve Laplaces equation inside a rectangle 0 x L, 0 y H subject to
a. u
x
(0, y) = u
x
(L, y) = u(x, 0) = 0, u(x, H) = f(x).
b. u(0, y) = g(y), u(L, y) = u
y
(x, 0) = u(x, H) = 0.
c. u(0, y) = u(L, y) = 0, u(x, 0) u
y
(x, 0) = 0, u(x, H) = f(x).
6. Solve Laplaces equation outside a circular disk of radius a, subject to
a. u(a, ) = ln 2 + 4 cos 3.
b. u(a, ) = f().
7. Solve Laplaces equation inside the quarter circle of radius 1, subject to
a. u
(r, 0) = u(r, /2) = 0, u(1, ) = f().

b. u
(r, 0) = u
(r, /2) = 0, u
r
(1, ) = g().
c. u(r, 0) = u(r, /2) = 0, u
r
(1, ) = 1.
8. Solve Laplaces equation inside a circular annulus (a < r < b), subject to
a. u(a, ) = f(), u(b, ) = g().
b. u
r
(a, ) = f(), u
r
(b, ) = g().
9. Solve Laplaces equation inside a semi-innite strip (0 < x < , 0 < y < H) subject
to
u
y
(x, 0) = 0, u
y
(x, H) = 0, u(0, y) = f(y).
10. Consider the heat equation
u
t
= u
xx
+ q(x, t), 0 < x < L,
u(0, t) = u(L, t) = 0.
Assume that q(x, t) is a piecewise smooth function of x for each positive t. Also assume that
u and u
x
are continuous functions of x and u
xx
and u
t
are piecewise smooth. Thus
u(x, t) =
n=1
b
n
(t) sin
n
L
x.
Write the ordinary dierential equation satised by b
n
(t).
11. Solve the following inhomogeneous problem
u
t
= k
2
u
x
2
+ e
t
+ e
2t
cos
3
L
x,
152
subject to
u
x
(0, t) =
u
x
(L, t) = 0,
u(x, 0) = f(x).
Hint : Look for a solution as a Fourier cosine series. Assume k ,=
2L
2
9
2
.
12. Solve the wave equation by the method of separation of variables
u
tt
c
2
u
xx
= 0, 0 < x < L,
u(0, t) = 0,
u(L, t) = 0,
u(x, 0) = f(x),
u
t
(x, 0) = g(x).
u
t
= 2u
xx
, 0 < x < L,
u(0, t) = u
x
(L, t) = 0,
and the initial condition
u(x, 0) = sin
3
2
L
x.
u
t
= k
_
1
r
r
_
r
u
r
_
+
1
r
2
2
u
2
_
inside a disk of radius a subject to the boundary condition
u
r
(a, , t) = 0,
u(r, , 0) = f(r, )
where f(r, ) is a given function.
153
1 a. u
t
= ku
xx
u(0, t) = 0
u(L, t) = 0
u(x, 0) = 6 sin
9 x
L
u(x, t) =
n=1
B
n
sin
n x
L
e
k(
n
L
)
2
t
u(x, 0) =
n=1
B
n
sin
n x
L
6 sin
9 x
L
the only term from the sum that can survive is for n = 9 with B
9
= 6 B
n
= 0
for n ,= 9
u(x, t) = 6 sin
9 x
L
e
k(
9
L
)
2
t
b. u(x, 0) = 2 cos
3 x
L
u(x, t) =
n=1
B
n
sin
n x
L
e
k(
n
L
)
2
t
n=1
B
n
sin
n x
L
= 2 cos
3 x
L
B
n
=
2
L
_
L
0
2 cos
3 x
L
sin
n x
L
dx
compute the integral for n = 1, 2, . . . to get B
n
.
154
To compute the coecients, we need the integral
_
L
0
cos
3
L
x sin
n
L
xdx
Using the trigonometric identity
sin a cos b =
1
2
(sin(a + b) + sin(a b))
we have
1
2
_
L
0
_
sin
(n + 3)
L
x + sin
(n 3)
L
x
_
dx
Now for n ,= 3 the integral is
1
2
cos
(n+3)
L
x
(n+3)
L
[
L
0

1
2
cos
(n3)
L
x
(n3)
L
[
L
0
or when recalling that cos m = (1)
m
L
2(n + 3)
_
(1)
n+3
1
_

L
2(n 3)
_
(1)
n3
1
_
, for n ,= 3
Note that for n odd, the coecient is zero.
For n = 3 the integral is
_
L
0
cos
3
L
x sin
3
L
xdx =
1
2
_
L
0
sin
6
L
xdx
which is
1
2
L
6
cos
6
L
x[
L
0
= 0
155
2. u
t
= ku
xx
u
x
(0, t) = 0
u
x
(L, t) = 0
u(x, 0) = f(x)
u(x, t) = X(x) T(t)
Tx = kX
T
kT
=
X
x
=
T + k T = 0
+ X = 0
X
(0) = 0
X
(L) = 0
X
n
= A
n
cos
n x
L
, n = 1, 2, . . .
n
=
_
n
L
_
2
, n = 1, 2, . . .
0
= 0 X
0
= A
0
T
n
+
_
n
L
_
2
kT
n
= 0
T
n
= B
n
e
(
n
L
)
2
kt
u(x, t) = A
0
B
0
. .
=a
0
+
n=1
A
n
cos
n x
L
B
n
e
(
n
L
)
2
kt
u(x, t) = a
0
+
n=1
a
n
cos
n x
L
e
(
n
L
)
2
kt
156
a.
f(x) =
0 x < L/2
1 x > L/2
u(x, 0) = f(x) = a
0
+
n=1
a
n
cos
n
L
x
a
0
=
2
2L
_
L
L/2
dx =
1
L
_
L
L
2
_
=
1
2
a
n
=
2
L
_
L
L/2
cos
n
L
xdx =
2
L
L
n
sin
n
L
x
L
L/2
=
2
n
sin
n
2
u(x, t) =
1
2
+
n=1
_
2
n
sin
n
2
_
cos
n
L
xe
k(
n
L
)
2
t
b.
f(x) = 6 + 4 cos
3
L
x
= a
0
+
n=1
a
n
cos
n
L
x
a
0
= 6
a
3
= 4 a
n
= 0 n ,= 3
u(x, t) = 6 + 4 cos
3
L
xe
k(
3
L
)
2
t
157
3.
+ = 0
(0) = (2)
(0) =
(2)
> 0 = A cos
x + B sin
= A
sin
x + B
cos
x
(0) = (2) A = A cos 2
+ B sin 2
(0) =
(2) B
= A
sin 2
+ B
cos 2
A(1 cos 2
) B sin 2
= 0
A
sin 2
+ B
(1 cos 2
) = 0
A system of 2 homogeneous equations. To get a nontrivial solution one must have the
determinant = 0.
1 cos 2
sin 2
sin 2
(1 cos 2
= 0
(1 cos 2
)
2
+
sin
2
2
= 0
1 2 cos 2
+ cos
2
2
+ sin
2
2
. .
1
= 0
2
1 cos 2
= 0
= 0 or cos 2
= 1
2
= 2 n n = 1, 2, . . .
158
Since should be positive = n
2
n = 1, 2, . . .
n
= n
2
n
= A
n
cos nx + B
n
sin nx
= 0 = Ax + B
= A
(0) = (2) B = 2 A + B A = 0
(0) =
(2) A = A
= 0 = B
< 0 = Ae
x
+ Be
x
(0) = (2) A + B = Ae
2
+ Be
2
(0) =
(2)
B =
Ae
Be
2
A[1 e
2
] + B[1 e
2
] = 0
A
_
1 e
2
_
B
_
1 e
2
_
= 0
1 e
2
1 e
2
(1 e
2
(1 e
2
= 0
(1 e
2
) (1 e
2
(1 e
2
) (1 e
2
) = 0
2
(1 e
2
) (1 e
2
) = 0
1 e
2
= 0 or 1 e
2
= 0
e
2
= 1 e
2
= 1
159
Take ln of both sides
2
= 0 2
= 0
= 0
= 0
not possible not possible
Thus trivial solution if < 0
4.
1
r
r
_
r
u
r
_
+
1
r
2
2
u

2
= 0
u (a, ) = f()
u (r, 0) = u
(r, ) = 0
u (r, ) = R(r) ()
1
r
r
_
r
R
r
_
+
1
r
2
R

2

2
= 0
multiply by
r
2
R
r
R
(r R
+ = 0 (0) =
() = 0
r (r R
R = 0 [ R(0) [ <
160
+ = 0 r(r R
R = 0
(0) = 0 [ R(0) [ <
() = 0 R
n
= r
(n
1
2
)
only positive exponent
n
= sin (n 1/2)

because of boundedness
n
=
_
(n 1/2)

_
2
n = 1, 2, . . .
u (r, ) =
n=1
a
n
r
(n1/2)/
sin
n 1/2

f() =
n=1
a
n
a
(n1/2)/
sin
(n 1/2)

a
n
=
_
0
f() sin (n 1/2)

d
a
(n1/2)/
_
0
sin
2
(n 1/2)

d
161
5. u
xx
+ u
yy
= 0
u
x
(0, y) = 0
u
x
(L, y) = 0
u (x, 0) = 0
u (x, H) = f(x)
u (x, y) = X(x) Y (y) =
X
X
=
Y

Y
=
u
x
(0, y) = 0 X
(0) = 0
u
x
(L, y) = 0 X
(L) = 0
u (x, 0) = 0 Y (0) = 0
X
+ X = 0 Y

Y = 0
X
(0) = 0 Y (0) = 0
X
(L) = 0
Table at the end of Chapter 4
n
=
_
n
L
_
2
n = 0, 1, 2, . . .
X
n
= cos
n
L
x Y

n

_
n
L
_
2
Y
n
= 0 n = 0, 1, 2, . . .
If n = 0 Y
0
= 0 Y
0
= A
0
y + B
0
Y
0
(0) = 0 B
0
= 0
Y
0
(y) = A
0
y
If n ,= 0 Y
n
= A
n
e
(
n
L
)y
+ B
n
e
(
n
L
)y
or
Y
n
= C
n
sinh
_
n
L
y + D
n
_
Y
n
(0) = 0 D
n
= 0
Y
n
= C
n
sinh
n
L
y
162
u(x, y) =
0
A
0
. .
a
0
2
y 1 +
n=1
n
C
n
. .
an
cos
n
L
x sinh
n
L
y
u(x, H) =
a
0
2
H +
n=1
a
n
sinh
n
L
H cos
n
L
x f(x)
This is the Fourier cosine series of f(x)
a
0
H =
2
L
_
L
0
f(x) dx
a
n
sinh
n
L
H =
2
L
_
L
0
f(x) cos
n
L
xdx
a
0
=
2
HL
_
L
0
f(x) dx
a
n
=
2
L sinh
n
L
H
_
L
0
f(x) cos
n
L
xdx
and:
u(x, y) =
a
0
2
y +
n=1
a
n
sinh
n
L
y cos
n
L
x
163
5 b. u
xx
+ u
yy
= 0
u (0, y) = g(y)
u (L, y) = 0
u
y
(x, 0) = 0
u (x, H) = 0
X
X = 0 Y
+ Y = 0
X(L) = 0 Y

(0) = 0
Y (H) = 0
Using the summary of Chapter 4 we have
Y
n
(y) = cos
(n +
1
2
)
H
y, n = 0, 1, . . .
n
=
_
n +
1
2
_

H
2
n = 0, 1, . . .
Now use these eigenvalues in the x equation: X
n

_
(n+
1
2
)
H
_
2
X
n
= 0 n = 0, 1, 2, . . .
Solve:
X
n
= c
n
sinh
__
n +
1
2
_

H
x + D
n
_
Use the boundary condition: X
n
(L) = 0
X
n
(L) = c
n
sinh
__
n +
1
2
_
L
H
+ D
n
_
= 0
(n+
1
2
) L
H
+ D
n
= 0
X
n
= c
n
sinh
_
(n+
1
2
)
H
(x L)
_
u(x, y) =
n=0
a
n
sinh
_
(n +
1
2
)
H
(x L)
_
cos
_
n +
1
2
_

H
y
To nd the coecients a
n
, we use the inhomogeneous boundary condition:
u(0, y) = g(y) =
n=0
a
n
sinh
_
(n +
1
2
)
H
(L)
_
cos
_
n +
1
2
_

H
y
This is a Fourier cosine series expansion of g(y), thus the coecients are:
sinh
(n +
1
2
) L
H
a
n
=
2
H
_
H
0
g(y) cos
__
n +
1
2
_

H
y
_
dy
a
n
=
2
H sinh
(n+
1
2
) L
H
_
H
0
g(y) cos
__
n +
1
2
_

H
y
_
dy
164
5. c. u(0, y) = 0 X(0) = 0
u(L, y) = 0 X(L) = 0
u(x, 0) u
y
(x, 0) = 0 Y (0) Y

(0) = 0
X
+ X = 0 Y

n

_
n
L
_
2
Y
n
= 0
X(0) = 0 Y
n
(0) Y
n
(0) = 0
X(L) = 0
Y
n
= A
n
cosh
n
L
y + B
n
sinh
n
L
y
n
=
_
n
L
_
2
n = 1, 2, . . . Y
n
=
n
L
_
A
n
sinh
n
L
y + B
n
cosh
n
L
y
_
X
n
= sin
n
L
x Substitute in the boundary condition.
_
A
n

n
L
B
n
_
cosh 0
. .
=0
+
_
B
n

n
L
A
n
_
sinh 0
. .
=0
= 0
A
n
=
n
L
B
n
Y
n
= B
n
_
n
L
cosh
n
L
y + sinh
n
L
y
_
u(x, y) =
n=1
b
n
sin
n
L
x
_
n
L
cosh
n
L
y + sinh
n
L
y
_
Use the boundary condition u(x, H) = f(x)
f(x) =
n=1
b
n
sin
n
L
x
_
n
L
cosh
n
L
H + sinh
n
L
H
_
This is a Fourier sine series of f(x), thus the coecients b
n
are given by
b
n
_
n
L
cosh
n
L
H + sinh
n
L
H
_
=
2
L
_
L
0
f(x) sin
n
L
xdx
Solve for b
n
b
n
=
2
_
n
L
cosh
n
L
H + sinh
n
L
H
_
L
_
L
0
f(x) sin
n
L
xdx
165
6 a. u
rr
+
1
r
u
r
+
1
r
2
u

= 0 outside circle
u(a, ) = ln 2 + 4 cos 3
u(r, ) = R(r) ()
r
2
R
+ rR
R = 0
+ = 0
(0) = (2 )
= 0 R
0
=
0
+
0
ln r
(0) =
(2 )
= n
2
R
n
=
n
r
n
+
n
r
n
Since we are solving < 0 trivial solution

outside the circle = 0
0
() = 1
ln r as r > 0
n
= n
2
r
n
as r
n
= A
n
cos n + B
n
sin n
thus R
0
=
0
R
n
=
n
r
n
u(r, ) = a
0
0
1
. .
=a
0
/2
+
n=1
a
n
(A
n
cos n + B
n
sin n)
n
r
n
u(r, ) = a
0
/2 +
n=1
(a
n
cos n + b
n
sin n) r
n
Use the boundary condition:
u(a, ) =
a
0
2
+
n=1
(a
n
a
n
cos n + b
n
a
n
sin n) = ln 2 + 4 cos 3
b
n
= 0 n
a
0
2
= ln 2
a
n
a
n
= 4 n = 3 a
3
= 4a
3
a
n
a
n
= 0 n ,= 3 a
n
= 0 n ,= 3
u(r, ) = ln 2 + 4a
3
r
3
cos 3
166
6 b. The only dierence between this problem and the previous one is the boundary
condition
u(a, ) = f() =
a
0
2
+
n=1
(a
n
a
n
cos n + b
n
a
n
sin n)
a
0
, a
n
a
n
, b
n
a
n
are coecients of Fourier series of f
a
0
=
1
f() d
a
n
a
n
=
1
f() cos n d
b
n
a
n
=
1
f() sin n d
Divide the last two equations by a
n
to get the coecients a
n
and b
n
.
167
7 a. u
rr
+
1
r
u
r
+
1
r
2
u

= 0
u
(r, 0) = 0
u (r, /2) = 0
u (1, ) = f()
r
2
R
+ r R
R = 0
+ = 0 no periodicity !!
(0) = 0
R
n
= c
n
r
2n1
+ D
n
r
2n1
(/2) = 0
boundedness implies R
n
= c
n
r
2n1
If < 0 trivial
= 0
0
= A
0
+ B
0
0
= A
0

0
(0) = 0 A
0
= 0
0
(/2) = 0 B
0
= 0
trivial
> 0 = A cos
+ B sin
A sin
+ B
cos
(0) = 0 B = 0
(/2) = 0 A cos
/2 = 0
/2 =
_
n
1
2
_
n = 1, 2,
= 2
_
n
1
2
_
= 2n 1
n
= (2n 1)
2
n
= cos (2n 1) , n = 1, 2,
Therefore the solution is
u =
n=1
a
n
r
2n1
cos (2n 1)
168
Use the boundary condition
u (1, ) =
n=1
a
n
cos (2n 1) = f()
This is a Fourier cosine series of f(), thus the coecients are given by
a
n
=
2
/2
_
/2
0
f() cos (2n 1) d
Remark: Since there is no constant term in this Fourier cosine series, we should have
a
0
=
2
2
_
/2
0
f() d = 0
That means that the boundary condition on the curved part of the domain is not arbitrary
but must satisfy
_
/2
0
f() d = 0
169
7 b. u
(r, 0) = 0
u
(r, /2) = 0
u
r
(1, ) = g()
Use 7 a to get the 2 ODEs
+ = 0 r
2
R
+ r R
R = 0
(0) =
(/2) = 0
n
=
_
n
2
_
2
= (2n)
2
n = 0, 1, 2, . . .
n
= cos 2n, n = 0, 1, 2, . . .
Now substitute the eigenvalues in the R equation
r
2
R
+ r R
(2n)
2
R = 0
The solution is
R
0
= C
0
ln r + D
0
, n = 0
R
n
= C
n
r
2n
+ D
n
r
2n
, n = 1, 2, . . .
Since ln r and r
2n
blow up as r 0 we have C
0
= C
n
= 0. Thus
u(r, ) = a
0
+
n=1
a
n
r
2n
cos 2n
Apply the inhomogeneous boundary condition
u
r
(r, ) =
n=1
2na
n
r
2n1
cos 2n
And at r = 1
u
r
(1, ) =
n=1
2na
n
cos 2n = g()
This is a Fourier cosine series for g() and thus
2na
n
=
_
/2
0
g() cos 2n d
_
/2
0
cos
2
2n d
a
n
=
_
/2
0
g() cos 2n d
2n
_
/2
0
cos
2
2n d
n = 1, 2, . . .
Note: a
0
is still arbitrary. Thus the solution is not unique. Physically we require
_
/2
0
g() d = 0 which is to say that a
0
= 0.
170
7 c. u (r, 0) = 0
u (r, /2) = 0
u
r
(1, ) = 1
Use 7 a to get the 2 ODEs
+ = 0 r
2
R
+ r R
R = 0
(0) = (/2) = 0
n
=
_
n
2
_
2
= (2n)
2
n = 1, 2, . . .
n
= sin 2n, n = 1, 2, . . .
Now substitute the eigenvalues in the R equation
r
2
R
+ r R
(2n)
2
R = 0
The solution is
R
n
= C
n
r
2n
+ D
n
r
2n
, n = 1, 2, . . .
Since r
2n
blow up as r 0 we have C
n
= 0. Thus
u(r, ) =
n=1
a
n
r
2n
sin 2n
Apply the inhomogeneous boundary condition
u
r
(r, ) =
n=1
2na
n
r
2n1
sin 2n
And at r = 1
u
r
(1, ) =
n=1
2na
n
sin 2n = 1
This is a Fourier sine series for the constant function 1 and thus
2na
n
=
_
/2
0
1 sin 2n d
_
/2
0
sin
2
2n d
a
n
=
_
/2
0
1 sin 2n d
2n
_
/2
0
sin
2
2n d
=
1(1)
n
2n
2n
2
=
1 (1)
n
n
2
a
n
=
1 (1)
n
)
n
2
171
8a. u
rr
+
1
r
u
r
+
1
r
2
u

= 0
u(a, ) = f()
u(b, ) = g()
r
2
R
+ rR
R = 0
+ = 0
(0) = (2 )
(0) =
(2 )
The eigenvalues and eigenfunctions can be found in the summary of chapter 4
0
= 0
0
= 1 for n = 0
n
= n
2
n
= cos n and sin n for n = 1, 2, . . .
Use these eigenvalues in the R equation and we get the following solutions:
R
0
= A
0
+ B
0
ln r n = 0
R
n
= A
n
r
n
+ B
n
r
n
n = 1, 2, . . .
Since r = 0 is outside the domain and r is nite, we have no reason to throw away any of
the 4 parameters A
0
, A
n
, B
0
, B
n
.
Thus the solution
u(r, ) = (A
0
+ B
0
ln r)
. .
R
0
1
..
0
a
0
+
n=1
(A
n
r
n
+ B
n
r
n
)
. .
Rn
(a
n
cos n + b
n
sin n)
. .
n
Use the 2 inhomogeneous boundary conditions
f() = u(a, ) = A
0
a
0
+ B
0
a
0
ln a
. .
0
+
n=1
(A
n
a
n
+ B
n
a
n
) a
n
. .
n
cos n
+
n=1
(A
n
a
n
+ B
n
a
n
) b
n
. .
n
sin n
g() = u(b, ) = A
0
a
0
+ B
0
a
0
ln b
. .
0
+
n=1
(A
n
b
n
+ B
n
b
n
) a
n
. .
n
cos n
172
+
n=1
(A
n
b
n
+ B
n
b
n
) b
n
. .
n
sin n
These are Fourier series of f() and g() thus the coecients
0
,
n
,
n
for f and the
coecients
0
,
n
,
n
for g can be written as follows
0
=
1
2
_
2
0
f() d
n
=
1
_
2
0
f() cos n d
n
=
1
_
2
0
f() sin n d
0
=
1
2
_
2
0
g() d
n
=
1
_
2
0
g() cos n d
n
=
1
_
2
0
g() sin n d
On the other hand these coecients are related to the unknowns A
0
, a
0
, B
0
, b
0
, A
n
, a
n
, B
n
and b
n
via the three systems of 2 equations each
0
= A
0
a
0
+ B
0
a
0
ln a
0
= A
0
a
0
+ B
0
a
0
ln b
solve for A
0
a
0
, B
0
a
0
n
= (A
n
a
n
+ B
n
a
n
) a
n
n
= (A
n
b
n
+ B
n
b
n
) a
n
solve for A
n
a
n
, B
n
a
n
n
= (A
n
a
n
+ B
n
a
n
) b
n
n
= (A
n
b
n
+ B
n
b
n
) b
n
solve for A
n
b
n
, B
n
b
n
Notice that we only need the products A
0
a
0
, B
0
b
0
,A
n
a
n
, B
n
a
n
, A
n
b
n
, and B
n
b
n
.
B
0
a
0
=

0

0
lnb ln a
173
A
0
a
0
=

0
ln b
0
ln a
ln b ln a
B
n
a
n
=

n
b
n

n
a
n
b
n
a
n
a
n
b
n
A
n
a
n
=

n
b
n

n
a
n
b
2n
a
2n
In a similar fashion
B
n
b
n
=

n
b
n

n
a
n
b
n
a
n
a
n
b
n
A
n
b
n
=

n
b
n

n
a
n
b
2n
a
2n
174
8. b Similar to 8.a
u(r, ) = (A
0
+ B
0
ln r) a
0
+
n=1
(A
n
r
n
+ B
n
r
n
) (a
n
cos n + b
n
sin n)
To use the boundary conditions:
u
r
(a, ) = f()
u
r
(b, ) = g()
We need to dierentiate u with respect to r
u
r
(r, ) =
B
0
r
a
0
+
n=1
(nA
n
r
n1
nB
n
r
n1
) (a
n
cos n + b
n
sin n)
Substitute r = a
u
r
(a, ) =
B
0
a
a
0
+
n=1
(nA
n
a
n1
nB
n
a
n1
) (a
n
cos n + b
n
sin n)
This is a Fourier series expansion of f() thus the coecients are
B
0
a
a
0
=
1
2
_
2
0
f() d
0
(nA
n
a
n1
nB
n
a
n1
)a
n
=
1
_
2
0
f() cos n d
n
(nA
n
a
n1
nB
n
a
n1
)b
n
=
1
_
2
0
f() sin n d
n
Now substitute r = b
u
r
(b, ) =
B
0
b
a
0
+
n=1
(nA
n
b
n1
nB
n
b
n1
) (a
n
cos n + b
n
sin n)
This is a Fourier series expansion of g() thus the coecients are
B
0
b
a
0
=
1
2
_
2
0
g() d
0
(nA
n
b
n1
nB
n
b
n1
)a
n
=
1
_
2
0
g() cos n d
n
175
(nA
n
b
n1
nB
n
b
n1
)b
n
=
1
_
2
0
g() sin n d
n
Solve for A
n
a
n
, B
n
a
n
:
(nA
n
a
n1
nB
n
a
n1
)a
n
=
n
(nA
n
b
n1
nB
n
b
n1
)a
n
=
n
We have
A
n
a
n
=

n
b
n1

n
a
n1
n(a
n1
b
n1
b
n1
a
n1
)
B
n
a
n
=

n
b
n1

n
a
n1
n(a
n1
b
n1
b
n1
a
n1
)
Solve for A
n
b
n
, B
n
b
n
:
(nA
n
a
n1
nB
n
a
n1
)b
n
=
n
(nA
n
b
n1
nB
n
b
n1
)b
n
=
n
We have
A
n
b
n
=

n
b
n1

n
a
n1
n(a
n1
b
n1
b
n1
a
n1
)
B
n
b
n
=

n
b
n1

n
a
n1
n(a
n1
b
n1
b
n1
a
n1
)
There are two equations for B
0
a
0
:
B
0
a
0
= b
0
B
0
a
0
= a
0
This means that f and g are not independent, but
a
0
= b
0
which means that
a
_
2
0
f() d = b
_
2
0
g() d
Note also that there is no condition on A
0
a
0
.
176
1 0 1 2 3 4 5 6 7 8
2
1
0
1
2
3
4
5
6
7
8
H
uxx + uyy = 0
Figure 57: Sketch of domain
9.
u
y
(x, 0) = 0
u
y
(x, H) = 0
u (0, y) = f(y)
X
X = 0 Y
+ Y = 0
solution should Y
(0) = 0
be bounded Y
(H) = 0
when x copy from table in Chapter 4 summary
n
=
_
n
H
_
2
n = 0, 1, 2,
Y
n
= cos
n
H
y
X
n

_
n
H
_
2
X
n
= 0 n = 1, 2,
X
n
= A
n
e
n
H
x
+ B
n
e
n
H
x
to get bounded solution A
n
= 0
For n = 0
X
0
= 0
X
0
= A
0
x + B
0
for boundedness A
0
= 0
u = B
0
1 +
n=1
B
n
e
n
H
x
cos
n
H
y
u(0, y) = f(y) = B
0
+
n=1
B
n
cos
n
H
y
Fourier cosine series of f(y).
177
10. u
t
= u
xx
+ q(x, t) 0 < x < L
subject to BC u(0, t) = u(L, t) = 0
Assume: q(x, t) piecewise smooth for each positive t.
u and u
x
continuous
u
xx
and u
t
piecewise smooth.
Thus,
u(x, t) =
n=1
b
n
(t) sin
n
L
x
(a). Write the ODE satised by b
n
(t), and
(b). Solve this heat equation.
STEPS:
1. Compute q
n
(t), the known heat source coecient
2. Plug u and q series expansions into PDE.
3. Solve for b
n
(t) - the homogeneous and particular solutions, b
H
n
(t) and b
P
n
(t)
4. Apply initial condition, b
n
(0), to nd coecient A
n
in the b
n
(t) solution. Assume
u(x, 0) = f(x)
1.
q(x, t) =
n=1
q
n
(t) sin
n
L
x
q
n
(t) =
2
L
_
L
0
q(x, t) sin
n
L
xdx
2.
u
t
=
n=1
b
n
(t) sin
n
L
x
u
xx
=
n=1
b
n
(t)
_
_
n
L
_
2
_
sin
n
L
x
178
n=1
b
n
(t) sin
n
L
x =
n=1
b
n
(t)
_
_
n
L
_
2
_
sin
n
L
x +
n=1
q
n
(t) sin
We have a Fourier Sine series on left and Fourier Sine series on right, so the coecients
must be the same; i.e.,
(a) b
n
(t) =
_
n
L
_
2
b
n
(t) + q
n
(t) A rst order ODE for b
n
(t).
III. Solve b
n
(t) =
_
n
L
_
2
b
n
(t) + q
n
(t)
Solution Form: b
n
(t) = A
n
b
H
n
(t) + b
P
n
(t)
Homogeneous Solution: b
H
n
(t) = e
(
n
L
)
2
t
Particular Solution: b
P
n
(t) = e
(
n
L
)
2
t
_
t
0
e
(
n
L
)
2
q
n
() d
b
n
(t) = A
n
e
(
n
L
)
2
t
+ e
(
n
L
)
2
t
_
t
0
e
(
n
L
)
2
q
n
() d
(Step IV is an extra step, not required in homework problem.)
IV. Find A
n
from initial condition. u(x, 0) = f(x) =
n=1
b
n
(0) sin
n
L
x
b
n
(0) =
2
L
_
L
0
f(x) sin
n
L
xdx
b
n
(0) = A
n
+ e
0
_
0
0
e
(
n
L
)
2
q
n
() dt
b
n
(0) = A
n
+ 1 0 =
2
L
_
L
0
f(x) sin
n
L
xdx
A
n
=
2
L
_
L
0
f(x) sin
n
L
xdx
b
n
(t) =
2
L
_
L
0
_
f(x) sin
n
L
xdx
_ _
e
(
n
L
)
2
t
_
+ e
(
n
L
)
2
t
_
t
0
e
(
n
L
)
2
q
n
() d
Plug this into u(x, t) =

n=1
b
n
(t) sin
n
L
x
179
11. u
t
= k u
xx
+ e
t
+ e
2t
cos
3
L
. .
q (x, t)
x
u
x
(0, t) = 0
u
x
(L, t) = 0
u(x, 0) = f(x)
The boundary conditions imply
u(x, t) =
n=0
b
n
(t) cos
n
L
x
Let q(x, t) =
n=0
q
n
(t) cos
n
L
x q
0
(t) = e
t
q
3
(t) = e
2t
the rest are zero !
Thus
b
n
= k
_
n
L
_
2
b
n
+ q
n
n = 0, 1,
n = 0

b
0
= q
0
= e
t
b
0
= e
t
b
1
+ k
_
L
_
2
b
1
= q
1
= 0
b
2
+ k
_
2
L
_
2
b
2
= 0
homogeneous
b
3
+ k
_
3
L
_
2
b
3
= e
2t
rest are homogeneous.
One can solve each equation to obtain all b
n
.
b
n
+ k
_
n
L
_
2
b
n
= 0 b
n
= C
n
e
k (
n
L
)
2
t
n = 1, 2, 4, 5,
note: n ,= 3
b
3
+ k
_
3
L
_
2
b
3
= e
2t
Solution of homogeneous is
b
3
= C
3
e
k (
3
L
)
2
t
For particular solution try b
3
= C e
2t
180
2C e
2t
+ k
_
3
L
_
2
= C e
2t
= e
2t
_
2 + k
_
3
L
_
2
_
C = 1
C =
1
k
_
3
L
_
2
2
denominator is not zero as assumed in the problem.
b
3
= C
3
e
k (
3
L
)
2
t
+
1
k (
3
L
)
2
2
e
2t
Now collecting terms,
u(x, t) =
n=0
b
n
(t) cos
n
L
x
= e
t
+
n=1
C
n
e
k (
n
L
)
2
t
cos
n
L
x +
1
k
_
3
L
_
2
2
e
2t
cos
3
L
x
Use the initial condition to nd all C
n
.
181
12. u
tt
c
2
u
xx
= 0 0 < x < L
u(0, t) = u(L, t) = 0
u(x, 0) = f(x)
u
t
(x, 0) = g(x)
XT
c
2
X
T = 0
T
c
2
T
=
X
X
=
X
+ X = 0 T
+ c
2
T = 0
X(0) = X(L) = 0
X
n
= sin
n
L
x T
n
+
_
n
L
_
2
c
2
T
n
= 0
n = 1, 2,
n
=
_
n
L
_
2
T
n
=
n
cos
n c
L
t +
n
sin
n c
L
t
u(x, t) =
n=1
_
n
cos
n c
L
t +
n
sin
n c
L
t
_
sin
n
L
x
u(x, 0) = f(x) =
n=1
n
sin
n
L
x
u
t
(x, 0) = g(x) =
n=1
n c
L

n
sin
n
L
x
n
=
2
L
_
L
0
f(x) sin
n
L
xdx
n c
L

n
=
2
L
_
L
0
g(x) sin
n
L
xdx
n
=
2
n c
_
L
0
g(x) sin
n
L
xdx
182
13. u
t
= 2u
xx
u(0, t) = 0
u
x
(L, t) = 0
u(x, 0) = sin
3
2
L
x
u = XT
X

T = 2X
T
2T
=
X
X
= X
+ X = 0

T + 2 T = 0
X(0) = 0
X
(L) = 0
X
n
= sin
_
n +
1
2
_

L
x n = 0, 1,
n
=
__
n +
1
2
_

L
_
2
u(x, t) =
n=1
a
n
e
2[((n+
1
2
)

L
]
2
t
sin
_
n +
1
2
_

L
x
At t = 0
u(x, 0) =
n=1
a
n
sin
_
n +
1
2
_

L
x
But also
u(x, 0) = sin
3
2
L
x
Therefore
a
1
= 1 , a
n
= 0 n > 1
u(x, t) = e
2(
3
2 L
)
2
t
sin
3
2
L
x
183
14. u
t
= k
_
1
r
(ru
r
)
r
+
1
r
2
u

_
u
r
(a, , t) = 0 inside a disk
u (r, , 0) = f(r, )
T
R = kT
_
1
r
(r R
+
1
r
2
R
_
T
kT
=
1
r
(rR
R
+
1
r
2
=
T
+ k T = 0
1
r
(rR
R
+
1
r
2
= multiply through by r
2
r(rR
R
+ r
2
=
+ = 0 ; r(rR
+ r
2
R R = 0
(0) = (2 ) [ R(0) [ <
(0) =
(2 ) R
(a) = 0

n
= n
2
R = J
n
(
r)
n = 1, 2,
n
=
_
sin n
cos n
J
n
(
a) = 0 gives
nm
0
= 0
0
= 1
T
nm
+
nm
kT
nm
= 0 T
nm
= e
nm k t
u(r, , t) =
m=1
_
a
0
2
+
n=1
(a
n
cos n + b
n
sin n)
_
. .
n
J
n
_
_
nm
r
_
. .
Rnm
e
k nmt
. .
Tnm
f(r, ) =
m=1
_
a
0
2
+
n=1
(a
n
cos n + b
n
sin n)
_
J
n
_
_
nm
r
_
Fourier-Bessel expansion of f.
See (7.5 later)
184
6 Sturm-Liouville Eigenvalue Problem
6.1 Introduction
Problems
1. a. Show that the following is a regular Sturm-Liouville problem
X
(x) + X(x) = 0,
X(0) = 0,
X
(L) = 0.
b. Find the eigenpairs
n
, X
n
directly.
c. Show that these pairs satisfy the results of the theorem.
2. Prove (6.1.28) - (6.1.30).
3. a. Is the following a regular Sturm-Liouville problem?
X
(x) + X(x) = 0,
X(0) = X(L),
X
(0) = X
(L).
Why or why not?
b. Find the eigenpairs
n
, X
n
directly.
c. Do they satisfy the results of the theorem? Why or why not?
4. Solve the regular Sturm-Liouville problem
X
(x) + aX(x) + X(x) = 0, a > 0,

X(0) = X(L) = 0.
For what range of values of a is negative?
5. Solve the ODE
X
(x) + 2X(x) + X(x) = 0, > 1,

X(0) = X
(1) = 0.
6. Consider the following Sturm-Liouville eigenvalue problem
d
dx
_
x
du
dx
_
+
1
x
u = 0, 1 < x < 2,
185
with boundary conditions
u(1) = u(2) = 0.
Determine the sign of all the eigenvalues of this problem (you dont have to explicitly deter-
mine the eigenvalues). In particular, is zero an eigenvalue of this problem?
7. Consider the following model approximating the motion of a string whose density (along
the string) is proportional to (1 +x)
2
,
(1 + x)
2
u
tt
u
xx
= 0, 0 < x < 1, t > 0
subject to the following initial conditions
u(x, 0) = f(x),
u
t
(x, 0) = 0,
and boundary conditions
u(0, t) = u(L, t) = 0.
a. Show that the ODE for X resulting from separation of variables is
X
+

(1 + x)
2
X = 0.
b. Obtain the boundary conditions and solve.
Hint: Try X = (1 + x)
a
.
186
1. a X
+ X = 0
X(0) = X
(L) = 0
is regular:
p = 1 q = 0 = 1
p continuous & positive & dierentiable
q continuous & nonnegative
continuous & positive
1
= 1
2
= 0
3
= 0
4
= 1
b. See chapter 4
n
=
__
n
1
2
_

L
_
2
n = 1, 2,
X
n
= sin
_
n
1
2
_

L
x
c. Innitely many eigenvalues
1
is smallest
no largest
X
n
are orthogonal
one eigenvalue for each eigenfunctions
and so on
187
2. The eigenfunctions X
n
(x) and the eigenvalues are
n
. Use these
n
in (6.1.24)
T
n
+
n
T
n
= 0
T
n
(t) = C
n
e
n T
u(x, t) =
n=1
a
n
e
n t
X
n
(x)
at t = 0 u(x, 0) =
n=1
a
n
X
n
(x) , a
n
are T
n
(0) in (6.1.28)
To nd a
n
we use the Fourier series expansion of f(x) = u (x, 0)
a
n
_
L
0
f(x) X
n
(x) c(x) (x) dx
_
L
0
X
2
n
(x) c(x) (x)
. .
weight function
dx
If a
1
,= 0 clearly
1
is the smallest and thus a
1
e
1
t
X
1
(x) dies the slowest.
188
3a. No, because the boundary conditions are not of the form
1
X (0) +
2
X
(0) = 0
3
X (L) +
4
X
(L) = 0
b. eigenpairs found in chapter 4
0
= 0 X
0
= 1
n
=
_
2n
L
_
2
X
n
=
sin
2n
L
x
cos
2n
L
x
c. No, because we have more than one eigenfunction for some eigenvalues.
189
4.
X
+ aX + X = 0 a > 0
X(0) = X(L) = 0
X = e
x
2
+ a + = 0
=
a
If a + > 0 ( > a)
X = Acos
a + x + B sin
a + x
X = B sin
a + L = 0
a + L = n n = 1, 2,
a +
n
=
_
n
L
_
2
n
= a +
_
n
L
_
2
n = 1, 2,
(show that if a + 0 the solution is trivial)
< 0 if a +
_
L
_
2
< 0
a > (/L)
2
190
5.
X
+ 2 X + X = 0 > 1
X(0) = X
(1) = 0
2
+ 2 + = 0
=
2
If + 2 > 0 ( > 2)
X = A cos
+ 2x + B sin
+ 2x
X(0) = 0 A = 0
X
(1) = B
+ 2 cos
+ 2 = 0
+ 2 =

2
+ n n = 0, 1, 2,
n
= 2 +
__
n +
1
2
_

_
2
X
n
= sin
_
n +
1
2
_
x
n = 0, 1, 2,
If > 1, could be negative if >

2
8
If 1 < <

2
8
then all are > 0.
191
6.
(xu
+
1
x
u = 0 1 < x < 2
u(1) = u(2) = 0
Use Rayleigh quotient p = x, q = 0, =
1
x
=
xu u
[
2
1
+
_
2
1
x(u
)
2
dx
_
2
1
1
x
u
2
dx
=
_
2
1
x(u
)
2
dx
_
2
1
1
x
u
2
dx
denominator is positive
numerator could be zero if u = constant. 0
Is that ( = 0) a possibility?
= 0 (xu
= 0
Integrate
xu
= c = constant
u
=
c
x
Integrate again
u = c log x + d
u(1) = c 0 + d = 0 d = 0
u(2) = 0 c log 2 = 0 c = 0
= 0 is not an eigenvalue.
192
7. (1 + x)
2
u
tt
u
xx
= 0, 0 < x < 1
u(x, 0) = f(x)
u
t
(x, 0) = 0
u(0, t) = u(L, t) = 0
(1 + x)
2
XT
T = 0
(1 + x)
2 T
T

X
X
= 0
T
T
=
X
(1 +x)
2
X
=
X
+

(1 + x)
2
X = 0 T
+ T = 0
X(0) = X (L) = 0
Try: X = (1 + x)
a
a(a 1) (1 + x)
a2
+

(1 + x)
2
(x + 1)
a
= 0
a(a 1) + = 0
= a(a 1)
193
6.2 Boundary Conditions of the Third Kind
Problems
1. Use the method of separation of variables to obtain the ODEs for x and for t for equations
(6.2.1) - (6.2.3).
2. Give the details for the case > 0 in solving (6.2.4) - (6.2.6).
3. Discuss
lim
n
n
for the above problem.
4. Write the Rayleigh quotient for (6.2.4) - (6.2.6) and show that the eigenvalues are all
positive. (That means we should have considered only case 3.)
5. What if h < 0 in (6.2.3)? Is there an h for which = 0 is an eigenvalue of this problem?
194
1. u
t
= ku
xx
u(0, t) = 0 X(0) = 0
u
x
(L, t) = h u (L, t) X
(L) = h X(L)
x

T = kX
T
kT
=
X
X
=
T + k T = 0 X
+ X = 0
X(0) = 0
X
(L) = h X(L)
195
2. > 0
The solution of the ODE for X is
X = B cos
x + A sin
x
X(0) = 0 B = 0
X
(x) = A
cos
x + 0
since B = 0
X
(L) = A
cos
L
h X(L) = h A sin
L
A
cos
L = h A sin
L
A ,= 0 (to avoid trivial solution)
cos
L = h sin
L
If cos
L = 0 h sin
L = 0 sin
L = 0 not possible.
Therefore we can divide by cos
L
tan
L =
h
196
3. Graphically we see that the straight line crossing the lower branches of tangent function
since the lower branches are for
x =
_
n
1
2
_

L
,
n
L
n = 1, 2, 3,
the square root of the eigenvalues are always in these ranges. As n increases, the crossing
become closer to the left side (where tan approaches -)
lim
n

n
=
_
n
1
2
_

L
2
197
4. X
+ X = 0
X(0) = 0
X
(L) = h X(L)
Rayleigh quotient
=
p X X
[
b
0
+
_
L
0
p[X
]
2
q X
2
dx
_
L
0
X
2
dx
p = 1 = 1 q = 0
=
X X
[
L
0
+
_
L
0
(X
)
2
dx
_
L
0
X
2
dx
X X
[
L
0
= X (L) X
(L)
. .
hX(L)
+ X(0)
. .
=0
X
(0) = h X(L)
2
The numerator is
h
..
>0
X(L)
2
. .
>0
+
_
L
0
(X
)
2
dx
. .
0
> 0
since since X maybe zero
X ,= 0
The denominator is positive (integrating X
2
)
> 0
198
5. If h < 0 in (6.2.3)
the numerator is now
h
..
<0
X (L)
2
. .
>0
. .
<0
+
_
L
0
(X
)
2
dx
. .
0
there is a possibility of zero or negative eigenvalues.
For what h one can have zero eigenvalue ?
check (6.2.10)
B(1 + hL) = 0 1 + hL = 0
h =
1
L
< 0
For this case B ,= 0 X
0
(x) = Bx
199
6.3 Proof of Theorem and Generalizations
Problems
1. Show that if u, v both satisfy the boundary conditions (6.1.9)-(6.1.10) then
p (uv
vu
) [
b
a
= 0.
2. Show that the right hand side of (6.3.4) is zero even if u, v satisfy periodic boundary
conditions, i.e.
u(a) = u(b)
p(a)u
(a) = p(b)u
(b),
and similarly for v.
3. What can be proved about eigenvalues and eigenfunctions of the circularly symmetric
heat ow problem.
Give details of the proof.
Note: This is a singular Sturm-Liouville problem.
4. Consider the heat ow with convection
u
t
= ku
xx
+ V
0
u
x
, 0 < x < L, t > 0.
a. Show that the spatial ordinary dierential equation obtained by separation of
variables is not in Sturm-Liouville form.
b. How can it be reduced to S-L form?
c. Solve the initial boundary value problem
u(0, t) = 0, t > 0,
u(L, t) = 0, t > 0,
u(x, 0) = f(x), 0 < x < L.
200
1. u, v
1
x(a) +
2
x
(a) = 0
3
x(b) +
4
x
(b) = 0
p(uv
vu
) [
b
a
= 0
(1).
2
v
(a) =
1
v(a)
4
v
(b) =
3
v(b)
(2). assume
3
,= 0
u(b) v
(b) =
3
u
(b) v
(b)
v(b) u
(b) = u
(b)
_
3
v
(b)
_
+ 0 add up to zero
(3). assume
1
,= 0
u(a) v
(a) =
_
1
u
(a)
_
v
(a)
+v(a) u
(a) =
1
v
(a) u
(a)
+ 0 add up to zero
(4). If
3
= 0 v
(b) = 0
u
(b) = 0
leads to u(b) v
(b) = 0
u
(b) v(b) = 0
(5). same true if
1
= 0
v
(a) = 0
u
(a) = 0
201
2.
u(a) = u(b)
p(a) u
(a) = p(b) u
(b)
and similarly for v
p(u v
v u
) [
b
a
=
= p(b) u(b) v
(b) p(b) v(b) u
(b)
p(a) u(a)
. .
u(b)
v
(a)
. .
p(b) u(b) v
(b)
+ p(a) v(a)
. .
v(b)
u
(a)
. .
p(b) u
(b) v(b)
this term this matches the term right above it
cancels the with dierence in sign only
one above it. thus these two terms add up to zero.
202
4.
u
t
= ku
xx
+ V
0
u
x
a.
u = XT
XT
= kTX
+ V
0
X
T
T
kT
=
X
X
+
V
0
k
X
X
=
X
+
V
0
k
X
+ X = 0
The two terms in the box should be combined into one in order to have the equation in
Sturm-Liouville form.
b. X
+
V
0
k
X
= (e
V
0
k
x
X)
V
0
k
x
(Recall integrating factors!)
Thus the equation becomes
(e
V
0
k
x
X
+ e
V
0
k
x
X = 0
This is the Sturm-Liouville form with
p = e
V
0
k
x
; q = 0 ; = e
V
0
k
x
203
4c. To solve the initial value problem we have
X
+
V
0
k
X
+ X = 0
X(0) = X(L) = 0
Try X = e
x
2
+
V
0
k
+ = 0
=
V
0
k

_
_
V
0
k
_
2
4
2
=
V
0
2k

_
V
0
2k
_
2

If
_
V
0
2k
_
2
> 0 the solutions are real exponentials which with the boundary conditions
yield a trivial solution. Similarly for
_
V
0
2k
_
2
= , since X = e
V
0
2k
x
(Ax + B) which again
is trivial when using the boundary conditions.
If
_
V
0
2k
_
2
< 0 then let
2
=
_
V
0
2k
_
2
The solution is X = e
V
0
2k
x
(A cos x + B sin x)
Using the rst boundary condition we get
X(0) = 0 = A
Thus the second boundary condition gives
X(L) = 0 = e
V
0
2k
L
B sin L
L = n n = 1, 2,

n

_
V
0
2k
_
2
=
_
n
L
_
2
n = 1, 2,
n
=
_
V
0
2k
_
2
+
_
n
L
_
2
n = 1, 2,
204
X
n
= e
V
0
2k
x
sin
n
L
x n = 1, 2,
Using these eigenvalues in the T equation:
T
n
+
n
kT = 0
we get
T
n
= e
n kt
Thus
u(x, t) =
n=1
b
n
e
n kt
e
V
0
2k
x
sin
n
L
x ()
Use the initial condition:
f(x) =
n=1
b
n
e
V
0
2k
x
sin
n
L
x
This is a generalized Fourier series of f(x)
b
n
=
_
L
0
f(x) e
V
0
2k
x
sin
n
L
xdx
_
L
0
e
V
0
k
x
sin
2 n
L
xdx
(#)
The solution is given by (*) with the coecients by (#), and
n
in the box above.
205
6.4 Linearized Shallow Water Equations
Problems
1. Find the second solution of (6.4.13) for a(c) = n.
Hint: Use the power series solution method.
2.
a. Find a relationship between M(a, b; z) and its derivative
dM
dz
.
b. Same for U.
3. Find in the literature a stable recurrence relation to compute the conuent hypergeo-
metric functions.
206
6.5 Eigenvalues of Perturbed Problems
Problems
1. The ow down a slightly corrugated channel is given by u(x, y, ) which satises
2
u = 1 in [y[ h(x, ) = 1 + cos kx
subject to
u = 0 on y = h(x, )
and periodic boundary conditions in x.
Obtain the rst two terms for u.
2. The functions (x, y, ) and () satisfy the eigenvalue problem
xx
+
yy
+ = 0 in 0 x , 0 + x( x) y
subject to
= 0 on the boundary.
Find the rst order correction to the eigenpair
(0)
1
= sin xsin y
(0)
1
= 2
207
7 PDEs in Higher Dimensions
7.1 Introduction
7.2 Heat Flow in a Rectangular Domain
Problems
u
t
(x, y, t) = k (u
xx
(x, y, t) + u
yy
(x, y, t)) ,
on the rectangle 0 < x < L, 0 < y < H subject to the initial condition
u(x, y, 0) = f(x, y),
and the boundary conditions
a.
u(0, y, t) = u
x
(L, y, t) = 0,
u(x, 0, t) = u(x, H, t) = 0.
b.
u
x
(0, y, t) = u(L, y, t) = 0,
u
y
(x, 0, t) = u
y
(x, H, t) = 0.
c.
u(0, y, t) = u(L, y, t) = 0,
u(x, 0, t) = u
y
(x, H, t) = 0.
2. Solve the heat equation on a rectangular box
0 < x < L, 0 < y < H, 0 < z < W,
u
t
(x, y, z, t) = k(u
xx
+ u
yy
+ u
zz
),
u(0, y, z, t) = u(L, y, z, t) = 0,
u(x, 0, z, t) = u(x, H, z, t) = 0,
u(x, y, 0, t) = u(x, y, W, t) = 0,
u(x, y, z, 0) = f(x, y, z).
208
1. u
t
= k(u
xx
+ u
yy
)
u(x, y, 0) = f(x, y)
u = X(x)Y (y)T(t)
xY

T = kY X
T + kXTY
T
kT
=
X
X
+
Y
Y
=
T + kT = 0
X
X
=
Y
Y
=
X
+ X = 0 Y

+ ( )Y = 0
a. X(0) = X
(L) = 0
Y (0) = Y (H) = 0
X
n
= sin
_
n
1
2
_

L
x
n
=
_
n
1
2
_

L
2
n = 1, 2,
Y
nm
= sin
m
H
y
nm

n
=
_
m
H
_
2
n = 1, 2, m = 1, 2,
nm
=
_
n
1
2
_

L
2
+
_
m
H
_
2
n, m = 1, 2,
T
nm
= e
nmkt
u(x, y, t) =
n=1
n=1
a
nm
e
nmkt
sin
_
n
1
2
_

L
x sin
m
H
y
f(x, y) = u(x, y, 0) =
n=1
m=1
a
nm
sin
_
n
1
2
_

L
x sin
m
H
y
a
nm
=
_
H
0
_
L
0
f(x, y) sin
_
n
1
2
_

L
x sin
m
H
y dxdy
_
H
0
_
L
0
sin
2
_
n
1
2
_

L
x sin
2 m
H
y dxdy
209
1b. X
(0) = X(L) = 0
Y

(0) = Y
(H) = 0
X
n
= cos
_
n
1
2
_

L
x
n
=
_
n
1
2
_

L
2
n = 1, 2,
Y
nm
= cos
m
H
x
nm
=
_
n
1
2
_

L
2
+
_
m
H
_
2
n = 1, 2, m =
0, 1, 2,
u(x, y, t) =
n=1
1
2
a
n0
e
k
n0
t
cos
_
n
1
2
_

L
x+
m=1
n=1
a
nm
e
k nmt
cos
_
n
1
2
_

L
x cos
m
H
y
a
nm
=
_
H
0
_
L
0
f(x, y) cos
_
n
1
2
_

L
x cos
m
H
y dxdy
_
H
0
_
L
0
cos
2
_
n
1
2
_

L
x cos
2
m
H
y dxdy
n = 1, 2, m = 0, 1, 2,
210
1c. X(0) = X(L) = 0
Y (0) = Y
(H) = 0
X
n
= sin
n
L
x
n
=
_
n
L
_
2
n = 1, 2,
Y
nm
= sin
_
m
1
2
_

H
y
nm
=
_
n
L
_
2
+
_
m
1
2
_

H
2
m, n = 1, 2,
u(x, y, t) =
m=1
n=1
a
nn
e
knmt
sin
n
L
x sin
_
m
1
2
_
H
y
a
nm
=
_
L
0
_
H
0
f(x, y) sin
n
L
x sin
(m
1
2
)
H
y dy dx
_
L
0
_
H
0
sin
2 n
L
x sin
2
(m
1
2
)
H
y dy dx
211
2. u
t
= k(u
xx
+ u
yy
+ u
zz
)
TXY Z = kT(X
Y Z + XY
Z + XY Z
T
kT
=
X
X
+
Y
Y
+
Z
Z
=
T + kT = 0
X
X
=
Y
Y

Z
Z
=
X
+ X = 0
Y
Y
=
Z
Z
+ =
X(0) = X(L) = 0
Y
+ Y = 0
Y (0) = Y (H) = 0
Z
+ ( ) Z = 0
Z(0) = Z(W) = 0
X
n
= sin
_
n
L
_
x
n
=
_
n
L
_
2
n = 1, 2,
Y
m
= sin
m
H
y
m
=
_
m
H
_
2
m = 1, 2,
Z
nm
= sin

W
z
nm
=
_
n
L
_
2
+
_
m
H
_
2
+
_

W
_
2
= 1, 2,
u(x, y, z, t) =
n=1
m=1
=1
a
nm
e
k
nm
t
sin
n
L
x sin
m
H
y sin

W
z
a
nm
=
_
L
0
_
H
0
_
W
0
f(x, y, z) sin
n
L
x sin
m
H
y sin

W
z dz dy dx
_
L
0
_
H
0
_
W
0
sin
2 n
L
x sin
2 m
H
y sin
2
W
z dz dy dx
212
7.3 Vibrations of a rectangular Membrane
Problems
1. Solve the wave equation
u
tt
(x, y, t) = c
2
(u
xx
(x, y, t) + u
yy
(x, y, t)) ,
on the rectangle 0 < x < L, 0 < y < H subject to the initial conditions
u(x, y, 0) = f(x, y),
u
t
(x, y, 0) = g(x, y),
a.
u(0, y, t) = u
x
(L, y, t) = 0,
u(x, 0, t) = u(x, H, t) = 0.
b.
u(0, y, t) = u(L, y, t) = 0,
u(x, 0, t) = u(x, H, t) = 0.
c.
u
x
(0, y, t) = u(L, y, t) = 0,
u
y
(x, 0, t) = u
y
(x, H, t) = 0.
2. Solve the wave equation on a rectangular box
0 < x < L, 0 < y < H, 0 < z < W,
u
tt
(x, y, z, t) = c
2
(u
xx
+ u
yy
+ u
zz
),
u(0, y, z, t) = u(L, y, z, t) = 0,
u(x, 0, z, t) = u(x, H, z, t) = 0,
u(x, y, 0, t) = u(x, y, W, t) = 0,
and the initial conditions
u(x, y, z, 0) = f(x, y, z),
u
t
(x, y, z, 0) = g(x, y, z).
213
3. Solve the wave equation on an isosceles right-angle triangle with side of length a
u
tt
(x, y, t) = c
2
(u
xx
+ u
yy
),
u(x, 0, t) = u(0, y, t) = 0,
u(x, y, t) = 0, on the line x + y = a
u(x, y, 0) = f(x, y),
u
t
(x, y, 0) = g(x, y).
214
1. u
tt
= c
2
(u
xx
+ u
yy
)
TXY = c
2
T (X
Y + XY
T
c
2
T
=
X
X
+
Y
Y
=
T + c
2
T = 0
X
X
=
Y

Y
=
X
+ X = 0 Y
+ ( )Y = 0
a. X(0) = X
(L) = 0
Y (0) = Y (H) = 0
as in previous section
u(x, y, t) =
n=1
m=1
_
a
nm
cos c
_
nm
t + b
nm
sin c
_
nm
t
_
sin
_
n
1
2
_
L
x sin
m
H
y
Initial Conditions
f(x, y) = u(x, y, 0) =
n=1
m=1
a
nm
sin
_
n
1
2
_
L
x sin
m
H
y yields a
nm
a
nm
=
_
H
0
_
L
0
f(x, y) sin
_
n
1
2
_

L
x sin
m
H
y dxdy
_
H
0
_
L
0
sin
2
_
n
1
2
_

L
x sin
2 m
H
y dxdy
g(x, y) = u
t
(x, y, 0) =
n=1
m=1
c
_
nm
b
nm
sin (n
1
2
)

L
x sin
m
H
y
b
nm
=
_
L
0
_
H
0
g(x, y) sin
(n
1
2
)
L
x sin
m
H
y dy dx
c
nm
_
L
0
_
H
0
sin
2
(n
1
2
)

L
x sin
2 m
H
y dy dx
215
b.
X(0) = X(L) = 0
Y (0) = Y (H) = 0
X
n
= sin
n
L
x
n
=
_
n
L
_
2
n = 1, 2,
Y
nm
= sin
m
H
y
nm
=
_
n
L
_
2
+
_
m
H
_
2
m = 1, 2,
u(x, y, t) =
n=1
m=1
_
a
nm
cos c
_
nm
t + b
nm
sin c
_
nm
t
_
sin
n
L
x sin
m
H
y
f(x, y) =
n=1
m=1
a
nm
sin
n
L
x sin
m
H
y
g(x, y) =
n=1
m=1
c
_
nm
b
nm
sin
n
L
x sin
m
H
y
a
nm
, b
nm
in a similar fashion to part a.
a
nm
=
_
H
0
_
L
0
f(x, y) sin
n
L
x sin
m
H
y dxdy
_
H
0
_
L
0
sin
2 n
L
x sin
2 m
H
y dxdy
b
nm
=
_
L
0
_
H
0
g(x, y) sin
n
L
x sin
m
H
y dy dx
c
nm
_
L
0
_
H
0
sin
2 n
L
x sin
2 m
H
y dy dx
216
c. see 1b in 7.1
X
n
= cos
_
n
1
2
_

L
x
n
=
_
n
1
2
_

L
2
n = 1, 2,
Y
nm
= cos
m
H
y
nm
=
_
n
1
2
_

L
2
+
_
m
H
_
2
m = 0, 1, 2,
u(x, y, t) =
n=1
_
a
n0
cos c
_
n0
t + b
n0
sin c
_
n0
t
_
cos
_
n
1
2
_
L
x
+
m=1
n=1
_
a
nm
cos c
_
nm
t + b
nm
sin c
_
nm
t
_
cos
_
n
1
2
_
L
x cos
m
H
y
f(x, y) yields a
n0
, a
nm
g(x, y) yields b
n0
, b
nm
217
2. Since boundary conditions are the same as in 2 section 7.1
u(x, y, z, t) =
n=1
m=1
=1
_
a
nm
cos c
_
nm
t + b
nm
sin c
_
nm
t
_
sin
n
L
x sin
m
H
y sin

W
z
f(x, y, z) yields a
nm
a
nm
=
_
L
0
_
H
0
_
W
0
f(x, y, z) sin
n
L
x sin
m
H
y sin

W
z dz dy dx
_
L
0
_
H
0
_
w
0
sin
2 n
L
x sin
2 m
H
y sin
2
W
z dz dy dx
g(x, y, z) yields b
nm
b
nm
=
_
L
0
_
H
0
_
W
0
g(x, y, z) sin
n
L
x sin
m
H
y sin

W
z dz dy dx
c
nm
_
L
0
_
H
0
_
W
0
sin
2 n
L
x sin
2 m
H
y sin
2
W
z dz dy dx
218
3.
See the solution of Helmholtz equation (problem 2 in section 7.4)
nm
(x, y) = sin

a
(m+ n)x sin

a
ny (1)
m
sin

a
(m+ n)y sin

a
nx
nm
=

a
_
(m+ n)
2
+ n
2
n, m = 1, 2,
The solution is similar to 1b
u(x, y, t) =
n=1
m=1
_
a
nm
cos c
_
nm
t + b
nm
sin c
_
nm
t
_

nm
(x, y)
f(x, y) =
n=1
m=1
a
nm
nm
(x, y)
g(x, y) =
n=1
m=1
c
_
nm
b
nm
nm
(x, y)
a
nm
, b
nm
in a similar fashion to 1a.
a
nm
=
_
a
0
_
a
0
f(x, y)
nm
(x, y) dxdy
_
a
0
_
a
0

2
nm
(x, y) dxdy
b
nm
=
_
a
0
_
a
0
g(x, y)
nm
(x, y) dy dx
c
nm
_
a
0
_
a
0

2
nm
(x, y) dy dx
219
7.4 Helmholtz Equation
Problems
1. Solve
2
+ = 0 [0, 1] [0, 1/4]
subject to
(0, y) = 0
x
(1, y) = 0
(x, 0) = 0
y
(x, 1/4) = 0.
Show that the results of the theorem are true.
2. Solve Helmholtz equation on an isosceles right-angle triangle with side of length a
u
xx
+ u
yy
+ u = 0,
u(x, 0, t) = u(0, y, t) = 0,
u(x, y, t) = 0, on the line x + y = a.
220
1/4
1
= 0
x
= 0
= 0

y
= 0
Figure 58: Domain fro problem 1 of 7.4
1.
(x, y) = XY
X
Y + XY
+ XY = 0
X
X
=
Y
Y
=
X
+ X = 0 Y
+ ( ) Y = 0
X(0) = X
(1) = 0 Y (0) = Y
(1/4) = 0

X
n
= sin(n
1
2
) x Y
nm
= sin (m
1
2
) 4 y
n
=
_
(n
1
2
)
_
2
nm
=
_
(n
1
2
)
_
2
+ [(4m 2) ]
2
n = 1, 2, m = 1, 2,
nm
= sin (n
1
2
) x sin (4m 2) y
nm
= [(n
1
2
) ]
2
+ [4m 2) ]
2
n, m = 1, 2,
Innite number of eigenvalues
11
=
1
4

2
+ 4
2
is the smallest.
There is no largest since
nm
as n, m increase
221
a
a
Figure 59: Domain for problem 2 of 7.4

2.
The analysis is more involved when the equation is NOT separable in coordinates suitable
for the boundary. Only two nonseparable cases have been solved in detail, one for a boundary
which is an isosceles right triangle.
The function
sin
x
a
sin

a
y
is zero along the x and y part of the boundary but is not zero along the diagonal side.
However, the combination
sin

a
x sin

a
y sin

a
y sin

a
x
is zero along the diagonal if and are integers. (The + sign is taken when [ [ is
even and the sign when [ [ is odd).
The eigenfunctions
mn
(x, y) = sin

a
(m + n) x sin

a
ny (1)
m
sin

a
(m + n) y sin

a
nx
where m, n are positive integers.
The only thing we have to show is the boundary condition on the line x + y = a. To
show this, rotate by /4
x =
1
2
( )
y =
1
2
( + )
222
mn
=
sin

(m + 2n) sin

m sin

(m + 2n) sin

m m = 2, 4,
cos

(m + 2n) cos

m cos

(m + 2n) cos

m m = 1, 3,
mn
= 0 for = /2 which is x + y = a.
The eigenvalues are:
mn
=
_
a
_ _
(m + n)
2
+ n
2
223
7.5 Vibrating Circular Membrane
Problems
u
t
(r, , t) = k
2
u, 0 r < a, 0 < < 2, t > 0
subject to the boundary condition
u(a, , t) = 0 (zero temperature on the boundary)
u(r, , 0) = (r, ).
u
tt
(r, t) = c
2
(u
rr
+
1
r
u
r
),
u
r
(a, t) = 0,
u(r, 0) = (r),
u
t
(r, 0) = 0.
Show the details.
3. Consult numerical analysis textbook to obtain the smallest eigenvalue of the above
problem.
u
tt
(r, , t) c
2
2
u = 0, 0 r < a, 0 < < 2, t > 0
u
r
(a, , t) = 0
u(r, , 0) = 0,
u
t
(r, , 0) = (r) cos 5.
u
tt
(r, , t) c
2
2
u = 0, 0 r < a, 0 < < /2, t > 0
u(a, , t) = u(r, 0, t) = u(r, /2, t) = 0 (zero displacement on the boundary)
u(r, , 0) = (r, ),
u
t
(r, , 0) = 0.
224
1. u
t
= k
2
u 0 r a
u (a, , t) = 0 0 2
u (r, , 0) = (r, ) t > 0
TR = kT
_
1
r
(r R
+
1
r
2
R
T
kT
=
1
Rr
(r R
+
1
r
2
T + k T = 0
r
R
(r R
= r
2
r
R
(r R
+ r
2
=
+ = 0 r (r R
+ (r
2
) R = 0
(0) = (2 ) [ R(0) [ <
(0) =
(2 ) R(a) = 0

m
= m
2
R
m
= C
1m
J
m
(
r) to satisfy [ R(0) [ <
m
=
sin m
m = 1, 2,
cos m
R
m
(
a) = C
m
J
m
(
a) = 0
0
= 0
0
= 1
nm
are solutions of J
m
(
a) = 0
225
u(r, , t) =
m=1
n=1
(a
nm
cos m + b
nm
sin m) e
k nmt
J
m
(
_
nm
r)
+
n=1
a
n0
1
..
=
0
e
k
n0
t
J
0
(
_
n0
r)
. .
m=0
(r, ) =
n=1
a
n0
J
0
(
_
n0
r) +
m=1
n=1
(a
nm
cos m + b
nm
sin m) J
m
(
_
nm
r)
a
n0
=
_
2
0
_
a
0
(r, ) J
0
(
n0
r) rd rd
_
2
0
_
a
0
J
2
0
(
n0
r) rd rd
a
nm
=
_
2
0
_
a
0
(r, ) cos m J
m
(
nm
r) rd rd
_
2
0
_
a
0
cos
2
m J
2
m
(
nm
r) rd rd
b
nm
=
_
2
0
_
a
0
(r, ) sin m J
m
(
nm
r) rd rd
_
2
0
_
a
0
sin
2
m J
2
m
(
nm
r) rd rd
226
2. u
tt
c
2
(u
rr
+
1
r
u
r
)
u
r
(a, t) = 0
u (r, 0) = (r)
u
t
(r, 0) = 0
T R c
2
(R
+
1
r
R
) T = 0
T
c
2
T
=
R
+
1
r
R
R
=
T + c
2
T = 0 R
+
1
r
R
. .
+R = 0
1
r
(r R
+ R = 0
multiply by r
2
r (r R
+ r
2
R = 0
[ R(0) [ <
R
(a) = 0
This is Bessels equation with = 0

R
n
(r) = J
0
(
_
n
r)
where
n
J
0
(
n
a) = 0
gives the eigenvalues
n
u(r, t) =
n=1
_
a
n
cos
_
n
ct + b
n
sin c
_
n
t
_
J
0
(
_
n
r)
(r) =
n=1
a
n
J
0
(
_
n
r)
This yields a
n
. a
n
=
_
a
0
(r) J
0
(
n
r) rdr
_
a
0
J
2
0
(
n
r) rdr
0 = u
t
(r, 0) =
n=1
c
_
n
b
n
J
0
(
_
n
r) b
n
= 0
227
4. u
tt
c
2
2
u = 0
u
r
(a, , t) = 0
u(r, , 0) = 0
u
t
(r, , 0) = (r) cos 5
T + c
2
T = 0
+ = 0 r(r R
+ ( r
2
) R = 0
T (0) = 0 (0) = (2 ) [ R(0) [ <
(0) =
(2 ) R
(a) = 0
T = a cos c
nm
t
+b sin c
nm
t
0
= 0
0
= 1 R = J
n
(
r)
Since T(0) = 0 R
(a) = J
n
(
a)
= 0
n
= n
2
m
=
_
cos n
sin n

T = sin c
nm
t
n0
= 0
or
J
n
(
nm
a) = 0
m = 1, 2,
for each n = 0, 1, 2,
u(r, , t) =
m=0
n=0
a
nm
cos n + b
nm
sin n
_
J
n
(
_
nm
r)
_
sin c
_
nm
t
u
t
(r, , 0) =
m=0
n=0
a
nm
cos n + b
nm
sin n J
n
(
_
nm
r) c
_
nm
cos c
_
nm
t
. .
=1 at t =0
Since u
t
(r, , 0) = (r) cos 5 all sin n term should vanish i.e. b
n
= 0 and all a
n
= 0
except a
5
(n = 5)
(r) cos 5 =
m=0
a
5m
cos 5 J
5
(
_
5m
r) c
_
5m
228
This is a generalized Fourier series for (r)
a
5m
c
_
5m
=
_
a
0
(r) J
5
(
_
5m
r) rdr
_
a
0
J
2
5
(
_
5m
r) rdr
u(r, , t) =
m=0
a
5m
cos 5 J
5
(
_
5m
r) sin c
_
5m
t
where
5m
can be found from
_
5m
J
5
(
_
5m
a) = 0
and a
5m
from
a
5m
=
_
a
0
(r) J
5
(
_
5m
r) rdr
c
5m
_
a
0
J
2
5
(
_
5m
r) rdr
229
5. u
tt
c
2
2
u = 0
u(a, , t) = 0
u(r, 0, t) = u(r, /2, t) = 0
T + c
2
T = 0
+ = 0 r(r R
+ ( r
2
) R = 0
(0) = (/2) = 0 [ R(0) [ <
R(a) = 0
n
= (2n)
2
n
= sin 2n R(r) = J
2n
(
2nm
r)
n = 1, 2, J
2n
(
_
2n, m
a) = 0 m = 1, 2,
u(r, , t) =
m=1
n=1
a
mn
J
2n
(
_
2n, m
r) sin 2n cos c
_
2n, m
t
. .
since ut (r, , 0) =0
g(r, ) = u(r, , 0) =
m=1
n=1
a
mn
J
2n
(
_
2n, m
r) sin 2n
a
mn
=
_
a
0
_
/2
0
J
2n
(
_
2n, m
r) g(r, ) sin 2n r d dr
_
/2
0
_
a
0
J
2
2n
(
_
2n, m
r) sin
2
2n r d r d
230
7.6 Laplaces Equation in a Circular Cylinder
Problems
1. Solve Laplaces equation
1
r
(ru
r
)
r
+
1
r
2
u
+ u
zz
= 0, 0 r < a, 0 < < 2, 0 < z < H
subject to each of the boundary conditions
a.
u(r, , 0) = (r, )
u(r, , H) = u(a, , z) = 0
b.
u(r, , 0) = u(r, , H) = 0
u
r
(a, , z) = (, z)
c.
u
z
(r, , 0) = (r, )
u(r, , H) = u(a, , z) = 0
d.
u(r, , 0) = u
z
(r, , H) = 0
u
r
(a, , z) = (z)
1
r
(ru
r
)
r
+
1
r
2
u
+ u
zz
= 0, 0 r < a, 0 < < , 0 < z < H
u(r, , 0) = 0,
u
z
(r, , H) = 0,
u(r, 0, z) = u(r, , z) = 0,
u(a, , z) = (, z).
3. Find the solution to the following steady state heat conduction problem in a box
2
u = 0, 0 x < L, 0 < y < L, 0 < z < W,
231
u
x
= 0, x = 0, x = L,
u
y
= 0, y = 0, y = L,
u(x, y, W) = 0,
u(x, y, 0) = 4 cos
3
L
xcos
4
L
y.
4. Find the solution to the following steady state heat conduction problem in a box
2
u = 0, 0 x < L, 0 < y < L, 0 < z < W,
u
x
= 0, x = 0, x = L,
u
y
= 0, y = 0, y = L,
u
z
(x, y, W) = 0,
u
z
(x, y, 0) = 4 cos
3
L
xcos
4
L
y.
5. Solve the heat equation inside a cylinder
u
t
=
1
r
r
_
r
u
r
_
+
1
r
2
2
u
2
+

2
u
z
2
, 0 r < a, 0 < < 2, 0 < z < H
u(r, , 0) = u(r, , H) = 0,
u(a, , z, t) = 0,
u(r, , z, 0) = f(r, , z).
232
1.
1
r
(ru
r
)
r
+
1
r
2
u

+ u
zz
= 0
(a)
+ = 0 Z
Z = 0 r(r R
+ ( r
2
) R = 0
(0) = (2 ) Z(H) = 0 [ R(0) [ <
(0) =
(2 ) R(a) = 0

0
= 0 R
nm
= J
m
(
nm
r)
0
= 1 satises boundedness
m
= m
2
m
=
sin m
cos m
J
m
(
nm
a) = 0
yields eigenvalues
m = 1, 2, n = 1, 2,
> 0 !
Z
nm
= sinh
_
nm
(z H)
vanishes at z = H
u(r, , z) =
m=0
n=1
(a
nm
cos m + b
nm
sin m) sinh
_
nm
(z H) J
m
(
_
nm
r)
This is zero for m = 0

(r, ) =
m=0
n=1
(a
nm
cos m + b
nm
sin m) sinh
_
nm
(H)
. .
this is a constant
J
m
(
_
nm
r)
a
nm
=
_
a
0
_
2
0
(r, ) cos m J
m
(
nm
r) rd dr
sinh
nm
(H)
_
a
0
_
2
0
cos
2
m J
2
m
(
nm
r) rd dr
b
nm
=
_
a
0
_
2
0
(r, ) sin m J
m
(
nm
r) rd dr
sinh
nm
(H)
_
a
0
_
2
0
sin
2
m J
2
m
(
nm
r) rd dr
233
1 b.
+ = 0 Z
Z = 0 r(r R
+ ( r
2
) R = 0
Z(0) = Z(H) = 0 [ R(0) [ <

m
= m
2
Z
n
= sin
n
H
z
m
=
sin m
cos m
n
=
_
n
H
_
2
m = 1, 2, r(r R
+
_
_
n
H
_
2
r
2
m
2
_
R
0
= 0
extra minus sign
0
= 1
R
nm
= I
m
_
n
H
r
_
u(r, , z) =
m=0
n=1
(a
nm
cos m + b
nm
sin m) sin
n
H
z I
m
_
n
H
r
_
u
r
(a, , z) = (, z) =
m=0
n=1
(a
nm
cos m +b
nm
sin m) sin
n
H
z
n
H
I
m
_
n
H
a
_
. .
constant
a
nm
n
H
I
m
_
n
H
a
_
=
_
2
0
_
H
0
(, z) cos m sin
n
H
z dz d
_
2
0
_
H
0
cos
2
m sin
2 n
H
z dz d
a
nm
=
_
2
0
_
H
0
(, z) cos m sin
n
H
z dz d
n
H
I
m
_
n
H
a
_
_
2
0
_
H
0
cos
2
m sin
2 n
H
z dz d
b
nm
=
_
2
0
_
H
0
(, z) sin m sin
n
H
z dz d
n
H
I
m
_
n
H
a
_
_
2
0
_
H
0
sin
2
m sin
2 n
H
z dz d
234
1 c.
+ = 0 Z
Z = 0 r(r R
+ ( r
2
) R = 0
Z(H) = 0 [ R(0) [ <
R(a) = 0
Solution as in 1a exactly !
But
u
z
(r, , 0) = (r, ) =
m=0
n=1
(a
nm
cos m +b
nm
sin m) J
m
(
_
nm
r)
_
nm
cosh
_
nm
(H)
a
nm
_
nm
cosh
_
nm
(H) =
_
2
0
_
a
0
(r, ) cos m J
m
(
nm
r) rd rd
_
2
0
_
a
0
cos
2
m J
2
m
(
nm
r) rd rd
a
nm
=
_
2
0
_
a
0
(r, ) cos m J
m
(
nm
r) rd rd
nm
cosh
nm
(H)
_
2
0
_
a
0
cos
2
m J
2
m
(
nm
r) rd rd
b
nm
=
_
2
0
_
a
0
(r, ) sin m J
m
(
nm
r) rd rd
nm
cosh
nm
(H)
_
2
0
_
a
0
sin
2
m J
2
m
(
nm
r) rd rd
235
1 d.
+ = 0 Z
Z = 0 r(r R
+ ( r
2
) R = 0
Z(0) = Z
(H) = 0 [ R(0) [ <

as before
n
=
_ _
n
1
2
_

H
_
2
Z
n
= sin
_
n
1
2
_

H
z R
nm
= I
m
_ _
n
1
2
_

H
r
_
u (r, , z) =
m=0
n=1
(a
nm
cos m +b
nm
sin m) sin
_
n
1
2
_

H
z I
m
__
n
1
2
_

H
r
_
u
r
(a, , z) =
m=0
n=1
(a
nm
cos m + b
nm
sin m) sin
_
n
1
2
_

H
z
_
n
1
2
_

H
I
m
__
n
1
2
_

H
a
_
Since u
r
(a, , z) = (z) is independent of , we must have no terms with in the above
expansion, that is
(z) =
n=1
a
n0
sin
_
n
1
2
_

H
z
_
n
1
2
_

H
I
0
__
n
1
2
_

H
a
_
a
n0
=
_
H
0
(z) sin
_
n
1
2
_

H
z dz
_
n
1
2
_

H
I
0
__
n
1
2
_

H
a
_
_
H
0
sin
2
2H
z dz
And the solution is
u(r, , z) =
n=1
a
n0
sin
_
n
1
2
_

H
z I
0
__
n
1
2
_

H
r
_
236
2. u (r, , 0) = 0 = u
z
(r, , H)
u (r, 0, z) = 0 = u (r, , z)
u (a, , z) = (, z)
u = R(r) () Z (z)
Z
+ Z = 0
+ = 0 r(r R
+ ( r
2
) R = 0
Z(0) = 0 (0) = () = 0 [ R(0) [ <
Z
(H) = 0
Z
n
= sin
_
n
1
2
_

H
z
m
= sin m r(r R
_
__
n
1
2
_

H
_
2
r
2
+ m
2
_
R
m
= m
2
n
=
__
n
1
2
_

H
_
2
m = 1, 2, R(r) = I
m
__
n
1
2
_

H
r
_
n = 1, 2,
u (r, , z) =
m=1
n=1
c
mn
I
m
__
n
1
2
_

H
r
_
sin m sin
_
n
1
2
_

H
z
at r = a
(, z) =
m=1
n=1
c
mn
I
m
__
n
1
2
_

H
a
_
. .
coecient of expansion
sin m sin
_
n
1
2
_

H
z
c
mn
=
_
0
_
H
0
(, z) sin m sin
_
n
1
2
_

H
z dz d
I
m
__
n
1
2
_

H
a
_
_
0
_
H
0
sin
2
m sin
2
_
n
1
2
_

H
z dz d
237
3. u
xx
+ u
yy
+ u
zz
= 0 BC : u
x
(0, y, z) = 0
u
x
(L, y, z) = 0
u
y
(x, 0, z) = 0
u
y
(x, L, z) = 0
u (x, y, 0) = 4 cos
3
L
x cos
4
L
y
u (x, y, z) = X(x)Y (y)Z(z)
X
X
+
Y
Y
+
Z
Z
= 0
X
X
=
Y
Y

Z
Z
=
+ X = 0
BC : X
(0) = X
(L) = 0

n
=
_
n
L
_
2
X
n
= cos
n
L
x n = 0, 1, 2
Y

Y
=
Z
Z
=
Y

+ Y = 0
BC : Y

(0) = Y
(L) = 0

n
=
_
m
L
_
2
Y
m
= cos
m
L
y m = 0, 1, 2
Z
Z
= +
( + ) Z = 0
BC : Z (W) = 0
Z

_
_
n
L
_
2
+
_
m
L
_
2
_
Z = 0
Z
nm
= sinh
_
n
L
_
2
+
_
m
L
_
2
(z W)
238
general solution
u(x, y, z) =
m=0
n=0
A
mn
cos
n
L
x cos
m
L
y sinh
_
n
L
_
2
+
_
m
L
_
2
(z W)
u(x, y, 0) =
m=0
n=0
A
mn
cos
n
L
x cos
m
L
y sinh
_
n
L
_
2
+
_
m
L
_
2
W
But u(x, y, 0) = 4 cos
3
L
x cos
4
L
y
Comparing coecients
A
mn
= 0 for m ,= 4 or n ,= 3
For n = 3; m = 4 A
43
sinh
9
2
L
2
+
16
2
L
2
W = 4
A
43
sinh
5
L
W = 4
A
43
=
4
sinh
5
L
W
u(x, y, z) =
4
sinh
5
L
W
cos
3
L
x cos
4
L
y sinh
5
L
(z W)
239
4. u
xx
+ u
yy
+ u
zz
= 0 u
x
(0, y, z) = 0 , u
x
(L, y, z) = 0
u
y
(x, 0, z) = 0 , u
y
(x, L, z) = 0
u (x, y, W) = 0
X
X
+
Y
Y
+
Z
Z
= 0 u(x, y, 0) = 4 cos
3
L
x cos
4
L
y
X
X
=
Y
Y

Z
Z
=
+ X = 0
BC : X
(0) = X
(L) = 0

n
=
_
n
L
_
2
X
n
= cos
n
L
x
n = 0, 1, 2
+ Y = 0
BC : Y
(0) = Y

(L) = 0

m
=
_
m
L
_
2
Y
m
= cos
m
L
y
m = 0, 1, 2
Z
( + ) Z = 0
Z
(W) = 0
Z

_
(
n
L
)
2
+ (
m
L
)
2
_
Z = 0
Z
nm
= cosh
_
n
L
_
2
+
_
m
L
_
2
(z W)
240
u(x, y, z) = A
00
+
m=1
A
m0
cos
m
L
y cosh
m
L
(z W)
+
n=1
A
0n
cos
n
L
x cos cosh
n
L
(z W)
+
m=1
n=1
A
mn
cos
n
L
x cos
m
L
y cosh
_
n
L
_
2
+
_
m
L
_
2
(z W)
u
z
(x, y, z) =
m=1
m
L
A
m0
cos
m
L
y sinh
m
L
(z W)
+
n=1
n
L
A
0n
cos
n
L
x sinh
n
L
(z W)
+
m=1
n=1
_
n
L
_
2
+
_
m
L
_
2
A
mn
cos
n
L
x cos
m
L
y sinh
_
n
L
_
2
+
_
m
L
_
2
(z W)
At z = 0 u
z
(x, y, 0) =
m=1
A
m0
m
L
cos
m
L
y sinh
m
L
W
n=1
A
0n
n
L
cos
n
L
x sinh
n
L
W
m=1
n=1
_
n
L
_
2
+
_
m
L
_
2
A
mn
cos
n
L
x cos
m
L
y sinh
_
n
L
_
2
+
_
m
L
_
2
W
But u
z
(x, y, 0) = 4 cos
3
L
x cos
4
L
y
Comparing coecients A
mn
= 0 for n ,= 3 or m ,= 4
For n = 3 and m = 4 we have 4 =
5
L
A
43
sinh
5
L
W
A
43
=
4
5
L
sinh
5
L
W
Note that A
00
is NOT specied.
u(x, y, z) = A
00

4
5
L
sinh
5
L
W
cos
3
L
x cos
4
L
y cosh
5
L
(z W)
241
5. u
t
=
1
r
(r u
r
)
r
+
1
r
2
u
+ u
zz
BC: u (r, , 0) = 0
u (r, , H) = 0
u (a, , z) = 0
u (r, , z, t) = R(r)()Z(z)T(t)
RZT
=
1
r
Z T (r R
+
1
r
2
RZT
+ RTZ
T
=
1
r
(r R
R
+
1
r
2
+
Z
Z
Z
Z
=
T
T

1
r
(r R
R

1
r
2
=
Z
+ Z = 0
BC : Z(0) = 0
Z(H) = 0
Z
n
= sin
n
H
z
n
=
_
n
H
_
2
n = 1, 2,
1
r
2
=
T
T

1
r
(r R
R
+
_
n
H
_
2
=
T
T
r
2
r(r R
R
+
_
n
H
_
2
r
2
=
+ = 0
BC : (0) = (2)
(0) =
(2)
242
m
=
sin m
cos m
m
= m
2
m = 0, 1, 2,
T
T
r
2
=
r(r R
R

_
n
H
_
2
r
2
m
2
T
T
=
1
r
(r R
R

_
n
H
_
2
m
2
r
2
=
T
+ T = 0
1
r
(r R
R
= +
_
n
H
_
2
+
m
2
r
2
r(r R
R
= r
2
+
_
n
H
_
2
r
2
+ m
2
r(r R
(
_
n
H
_
2
) r
2
R m
2
R = 0
BC : [ R(0) [ <
R(a) = 0
R
nm
= I
m

_
n
H
_
2
r
This solution satises the boundedness at the origin. The eigenvalues
can be found by
using the second boundary condition:
I
m

_
n
H
_
2
a
= 0
Since the function I
m
(x) vanishes only at zero for any m = 1, 2, (I
0
is never zero) then
there is only one (for any n) satisfying

_
n
H
_
2
a = 0 m = 1, 2,
=
_
n
H
_
2
The solution fot T is T
nm
= e
(
n
H
)
2
t
The solution for R is I
m
(0 r) which is identically zero. This means that u(r, , z, t) = 0.
Physically, this is NOT surprising, since the problem has NO sources (homogeneous boundary
conditions and homogeneous PDE).
243
7.7 Laplaces equation in a sphere
Problems
1. Solve Laplaces equation on the sphere
u
rr
+
2
r
u
r
+
1
r
2
u
+
cot
r
2
u
+
1
r
2
sin
2
= 0, 0 r < a, 0 < < , 0 < < 2,

u
r
(a, , ) = f().
2. Solve Laplaces equation on the half sphere
u
rr
+
2
r
u
r
+
1
r
2
u
+
cot
r
2
u
+
1
r
2
sin
2
= 0, 0 r < a, 0 < < , 0 < < ,

u(a, , ) = f(, ),
u(r, , 0) = u(r, , ) = 0.
244
1. u (r, , ) =
n=0
A
n0
r
n
P
n
(cos )
+
n=0
m=1
r
n
P
m
n
(cos ) (A
nm
cos m + B
nm
sin m)
(7.7.37)
u
r
(a, , ) = f() =
=
n=0
nA
n0
a
n1
P
n
(cos )
+
n=0
m=1
n a
n1
P
m
n
(cos ) (A
nm
cos m + B
nm
sin m)
A
n0
na
n1
=
_
0
f() P
n
(cos ) sin d
_
0
P
2
n
(cos ) sin d
na
n1
A
nm
=
_
0
_
2
0
f() P
m
n
(cos ) cos m sin dd
_
0
_
2
0
[P
m
n
(cos ) cos m]
2
sin dd
na
n1
B
nm
=
_
0
_
2
0
f() P
m
n
(cos ) sin m sin dd
_
0
_
2
0
[P
m
n
(cos ) sin m]
2
sin dd
A
nm
=
_
0
_
2
0
f() P
m
n
(cos ) cos m sin dd
na
n1
_
0
_
2
0
[P
m
n
(cos ) cos m]
2
sin dd
B
nm
=
_
0
_
2
0
f() P
m
n
(cos ) sin m sin dd
na
n1
_
0
_
2
0
[P
m
n
(cos ) sin m]
2
sin dd
245
2.
+ = 0
(0) = () = 0
m
= m
2
m
= sin m
m = 1, 2,
u (r, , ) =
n=0
m=1
r
n
P
m
n
(cos )
. .
equation
A
nm
sin m
R equation
u (a, , ) = f(, ) =
n=0
m=1
a
n
A
nm
P
m
n
(cos ) sin m
a
n
A
nm
=
_
0
_
0
f(, ) P
m
n
(cos ) sin m sin d d
. .
area elem.
_
0
_
0
(P
m
n
(cos ))
2
sin
2
m sin d d
A
nm
=
_
0
_
0
f(, ) P
m
n
(cos ) sin m sin d d
. .
area elem.
a
n
_
0
_
0
(P
m
n
(cos ))
2
sin
2
m sin d d
u (r, , ) =
n=0
m=1
r
n
P
m
n
(cos ) A
nm
sin m
246
3. The equation becomes
u
+ cot u
+
1
sin
2
= 0, 0 < < , 0 < < 2

Using separation of variables
u(, ) = ()()
+ cot
+
1
sin
2
= 0
Divide by and multiply by sin
2
we have
sin
2
+ cos sin
=
Thus
+ = 0
sin
2
+ sin cos
= 0
Because of periodicity, the equation has solutions
m
=
sin m m = 1, 2, . . .
cos m
0
= 1
m
= m
2
m = 0, 1, 2, . . .
Substituting these
s in the equation, we get (7.7.21) with

1
= 0. The solution of the
equation is thus given by (7.7.27) - (7.7.28) with
1
= 0.
247
8 Separation of Variables-Nonhomogeneous Problems
8.1 Inhomogeneous Boundary Conditions
Problems
1. For each of the following problems obtain the function w(x, t) that satises the boundary
conditions and obtain the PDE
a.
u
t
(x, t) = ku
xx
(x, t) + x, 0 < x < L
u
x
(0, t) = 1,
u(L, t) = t.
b.
u
t
(x, t) = ku
xx
(x, t) + x, 0 < x < L
u(0, t) = 1,
u
x
(L, t) = 1.
c.
u
t
(x, t) = ku
xx
(x, t) + x, 0 < x < L
u
x
(0, t) = t,
u
x
(L, t) = t
2
.
2. Same as problem 1 for the wave equation
u
tt
c
2
u
xx
= xt, 0 < x < L
subject to each of the boundary conditions
a.
u(0, t) = 1 u(L, t) = t
b.
u
x
(0, t) = t u
x
(L, t) = t
2
c.
u(0, t) = 0 u
x
(L, t) = t
d.
u
x
(0, t) = 0 u
x
(L, t) = 1
248
1a. u
x
(0, t) = 1
u(L, t) = t
w(x, t) = A(t)x + B(t)
1 = w
x
(0, t) = A(t) A(t) = 1
t = w(L, t) = A(t)L + B(t) B(t) = t L
w(x, t) = x + t L
w
t
= 1
w
xx
= 0
v = u w u = v + w
v
t
+ 1 = kv
xx
+ x
b. w = Ax + B
1 = w(0, t) = B(t)
1 = w
x
(L, t) = A(t)
w = x + 1
w
t
= w
xx
= 0
v
t
= kv
xx
+ x
249
c. w
x
(0, t) = t
w
x
(L, t) = t
2
try w = A(t) x + B
w
x
= A(t) and we can not satisfy the 2 conditions.
try w = A(t) x
2
+ B(t) x
w
x
= 2A(t) x + B(t)
t = w
x
(0, t) = B(t)
t
2
= w
x
(L, t) = 2A(t)L + B(t)
. .
=t
A(t) =
t
2
t
2L
w =
t
2
t
2L
x
2
+ tx
w
t
=
2t 1
2L
x
2
+ x
w
xx
=
t
2
t
L
v
t
+
_
2t 1
2L
x
2
+ x
_
= k (v
xx
+
t
2
t
L
) + x
v
t
= kv
xx

2t 1
2L
x
2
x + k
t
2
t
L
+ x
v
t
= kv
xx

2t 1
2L
x
2
+ k
t
2
t
L
250
2. u
tt
c
2
u
xx
= xt
a. w(x, t) =
t 1
L
x + 1
w
tt
= 0 w
xx
= 0
v
tt
c
2
v
xx
= xt
b. w =
t
2
t
2L
x
2
+ tx as in 1c
w
tt
=
1
L
x
2
w
xx
=
t
2
t
L
u = v + w
v
tt
+
1
L
x
2
c
2
_
v
xx
+
t
2
t
L
_
= xt
v
tt
c
2
v
xx
=
1
L
x
2
+ c
2 t
2
t
L
+ xt
c. w(0, t) = 0 w
x
(L, t) = t
w = Ax + B w
x
= A
B = 0 A = t
w = tx
w
tt
= w
xx
= 0
v
tt
c
2
v
xx
= xt
251
d. w
x
(0, t) = 0
w
x
(L, t) = 1
Try w = A(t) x
2
+ B(t) x as in 1c
w
x
= 2A(t)x + B(t)
0 = B(t) 1 = 2A(t)L + B(t)
. .
=0
A(t) =
1
2L
w =
1
2L
x
2
w
tt
= 0 w
xx
=
1
L
v = u + w
v
tt
c
2
_
v
xx
+
1
L
_
= xt
v
tt
c
2
v
xx
= +
c
2
L
+ xt
252
8.2 Method of Eigenfunction Expansions
Problems
u
t
= ku
xx
+ x, 0 < x < L
subject to the initial condition
u(x, 0) = x(L x)
and each of the boundary conditions
a.
u
x
(0, t) = 1,
u(L, t) = t.
b.
u(0, t) = 1,
u
x
(L, t) = 1.
c.
u
x
(0, t) = t,
u
x
(L, t) = t
2
.
u
t
= u
xx
+ e
t
, 0 < x < , t > 0,
subject to the initial condition
u(x, 0) = cos 2x, 0 < x < ,
and the boundary condition
u
x
(0, t) = u
x
(, t) = 0.
253
1. u
t
= ku
xx
+ x
u(x, 0) = x(L x)
a. u
x
(0, t) = 1
w = x + t L
u(L, t) = t
Solve v
t
= kv
xx
1 + x (see 1a last section)
v
x
(0, t) = 0 v(x, 0) = x(L x) (x L) = (x + 1) (L x)
v (L, t) = 0
eigenvalues:
__
n
1
2
_

L
_
2
n = 1, 2,
eigenfunctions : cos (n
1
2
)

L
x n = 1, 2,
v =
n=1
v
n
(t) cos
_
n
1
2
_

L
x
1 + x =
n=1
s
n
cos
_
n
1
2
_

L
x s
n
=
_
L
0
(1 + x) cos
_
n
1
2
_

L
xdx
_
L
0
cos
2
_
n
1
2
_

L
xdx
n=1
v
n
(t) cos
_
n
1
2
_

L
x = k
n=1
_
_ _
n
1
2
_

L
_
2
_
v
n
cos
_
n
1
2
_

L
x
+
n=1
s
n
cos
_
n
1
2
_

L
x
Compare coecients
v
n
(t) + k
_ _
n
1
2
_

L
_
2
v
n
= s
n
v
n
= v
n
(0)e
[(n
1
2
)

L
]
2
kt
+ s
n
_
t
0
e
[(n
1
2
)

L
]
2
k (t )
d
. .
see(8. 2. 39)
v
n
(0) =coecients of expansion of (1 + x) (L x)
v
n
(0) =
_
L
0
(1 + x) (L x) cos
_
n
1
2
_

L
xdx
_
L
0
cos
2
_
n
1
2
_

L
xdx
u = v + w
254
1b. u(0, t) = u
x
(L, t) = 1
w = x + 1
v
t
= k v
xx
+ x
v (0, t) = 0
v
x
(L, t) = 0
v (x, 0) = x(L x) (x + 1)
eigenvalues:
__
n
1
2
_

L
_
2
n = 1, 2,
eigenfunctions: sin
_
n
1
2
_

L
x n = 1, 2,
v =
n=1
v
n
(t) sin
_
n
1
2
_

L
x
x =
n=1
s
n
sin
_
n
1
2
_

L
x s
n
=
_
L
0
x sin
_
n
1
2
_

L
xdx
_
L
0
sin
2
_
n
1
2
_

L
xdx
v
n
+ k
__
n
1
2
_

L
_
2
v
n
= s
n
v
n
(t) = v
n
(0) e
[(n
1
2
)

L
]
2
kt
+ s
n
1 e
[
(
n
1
2
)

L
]
2
t
[(n
1
2
)

L
]
2
v
n
(0) =
_
L
0
[x(L x) (x + 1)] sin
_
n
1
2
_

L
xdx
_
L
0
sin
2
_
n
1
2
_

L
xdx
Coecients of expansion of initial condition for v

u = v + w
255
1c. u
x
(0, t) = t
u
x
(L, t) = t
2
w =
t
2
t
2L
x
2
+ tx
v
t
= kv
xx

2t 1
2L
x
2
x + k
t
2
t
L
+ x
. .
this gives sn(t)
v
x
(0, t) = v
x
(L, t) = 0
n
=
_
n
L
_
2
n = 0, 1, 2,
X
n
(x) = cos
n
L
x n = 0, 1, 2,
v(x, 0) = x(L x) w(x, 0)
. .
=0
= x(L x)
s
n
(t) =
_
L
0
_
2t 1
2L
x
2
+ k
t
2
t
L
_
cos
n
L
xdx
_
L
0
cos
2
n
L
xdx
v
n
(t) = v
n
(0) e
k (
n
L
)
2
t
+
_
t
0
s
n
() e
k(
n
L
)
2
(t )
d
v(x, t) =
1
2
v
0
(t) +
n=1
v
n
(t) cos
n
L
x
v
n
(0) =
_
L
0
x(L x) cos
n
L
xdx
_
L
0
cos
2
n
L
xdx
u = v +
t
2
t
2L
x
2
+ tx
256
2. u
t
= u
xx
+ e
t
0 < x < , t > 0
u (x, 0) = cos 2x 0 < x <
u
x
(0, t) = u
x
(, t) = 0
Since the boundary conditions are homogeneous we can immediately expand u(x, t), the
right hand side and the initial temperature distribution in terms of the eigenfunctions. These
eigenfunctions are
n
= cos nx
= n
2
n = 0, 1, 2, . . .
u(x, t) =
1
2
u
0
(t) +
n=1
u
n
(t) cos nx
u(x, 0) =
1
2
u
0
(0) +
n=1
u
n
(0) cos nx = cos 2x
.
.
.
by initial condition
u
n
(0) = 0 n ,= 2
u
2
(0) = 1
e
t
=
n=1
s
n
(t) cos nx +
1
2
s
0
(t)
s
n
(t) =
_
0
e
t
cos nxdx
_
0
cos
2
nxdx
=
e
t
_
0
cos nxdx
_
0
cos
2
nxdx
for n ,= 0 the numerator is zero !!
For n = 0 both integrals yields the same value, thus
s
0
(t) = e
t
s
n
(t) = 0 , n ,= 0
Now substitute u
t
, u
xx
from the expansions for u:
1
2
u
0
(t) +
n=1
u
n
(t) cos nx =
n=1
(n
2
) u
n
(t) cos nx +
1
2
s
0
(t) +
n=1
s
n
(t) cos nx
257
For n = 0
1
2
u
0
(t) =
1
2
e
t
=
1
2
s
0
(t)
n ,= 0 u
n
+ n
2
u
n
= 0
Solve the ODES
u
n
= C
n
e
n
2
t
u
n
(0) = 0 n ,= 2 C
n
= 0
u
2
(0) = 1 C
2
= 1
u
0
= e
t
+ C
0
u
0
(0) = 0 C
0
1 = 0 C
0
= 1
u(x, t) =
1 e
t
2
+ e
4t
cos 2x
258
8.3 Forced Vibrations
Problems
1. Consider a vibrating string with time dependent forcing
u
tt
c
2
u
xx
= S(x, t), 0 < x < L
subject to the initial conditions
u(x, 0) = f(x),
u
t
(x, 0) = 0,
u(0, t) = u(L, t) = 0.
a. Solve the initial value problem.
b. Solve the initial value problem if S(x, t) = cos t. For what values of does resonance
occur?
2. Consider the following damped wave equation
u
tt
c
2
u
xx
+ u
t
= cos t, 0 < x < ,
u(x, 0) = f(x),
u
t
(x, 0) = 0,
u(0, t) = u(, t) = 0.
Solve the problem if is small (0 < < 2c).
3. Solve the following
u
tt
c
2
u
xx
= S(x, t), 0 < x < L
u(x, 0) = f(x),
u
t
(x, 0) = 0,
and each of the following boundary conditions
a.
u(0, t) = A(t) u(L, t) = B(t)
b.
u(0, t) = 0 u
x
(L, t) = 0
c.
u
x
(0, t) = A(t) u(L, t) = 0.
259
u
tt
c
2
u
xx
= xt, 0 < x < L,
u(x, 0) = sin x
u
t
(x, 0) = 0
and each of the boundary conditions
a.
u(0, t) = 1,
u(L, t) = t.
b.
u
x
(0, t) = t,
u
x
(L, t) = t
2
.
c.
u(0, t) = 0,
u
x
(L, t) = t.
d.
u
x
(0, t) = 0,
u
x
(L, t) = 1.
u
tt
u
xx
= 1, 0 < x < L,
u(x, 0) = f(x)
u
t
(x, 0) = g(x)
u(0, t) = 1,
u
x
(L, t) = B(t).
260
1a. u
tt
c
2
u
xx
= S(x, t)
u(x, 0) = f(x)
u
t
(x, 0) = 0
u(0, t) = u(L, t) = 0
u(x, t) =
n=1
u
n
(t)
n
(x)
S(x, t) =
n=1
s
n
(t)
n
(x)
n
(x) = sin
n
L
x
n
=
_
n
L
_
2
n = 1, 2,
s
n
(t) =
_
L
0
S (x, t) sin
n
L
xdx
_
L
0
sin
2 n
L
xdx
u
n
(t) + c
2
_
n
L
_
2
u
n
(t) = s
n
(t)
u
n
(t) = c
1
cos
n
L
ct + c
2
sin
n
L
ct +
_
t
0
s
n
()
sin c
n
L
(t )
c
n
L
d
u
n
(0) = c
1
=
_
L
0
f(x) sin
n
L
xdx
_
L
0
sin
2 n
L
xdx
since u(x, 0) = f(x)
u
n
(0) = c
2
c
n
L
= 0 since u
t
(x, 0) = 0 c
2
= 0
u(x, t) =
n=1
_
c
1
cos
n
L
ct +
L
c n
_
t
0
s
n
() sin c
n
L
(t ) d
_
sin
n
L
x
c
1
is given above.
261
b. If S = cos wt
s
n
(t) =
cos wt
_
L
0
sin
n
L
xdx
_
L
0
sin
2 n
L
xdx
= A
n
cos wt
where A
n
=
_
L
0
sin
n
L
xdx
_
L
0
sin
2 n
L
xdx
u(x, t) =
n=1
c
1
cos
n c
L
t +
L
c n
A
n
_
t
0
cos w sin
n c
L
(t ) d
. .
This integral can be computed
sin
n
L
x
c. Resonance occurs when
w = c
n
L
for any n = 1, 2,
262
2. u
tt
c
2
u
xx
+ u
t
= cos wt
u(x, 0) = f(x)
u
t
(x, 0) = 0
u(0, t) = u (, t) = 0
n
= sin nx
n = 1, 2,
n
= n
2
u(x, t) =
n=1
u
n
(t) sin nx
cos wt =
n=1
s
n
(t) sin nx
s
n
(t) =
cos wt
_
0
sin nxdx
_
0
sin
2
nx dx
= A
n
cos wt
n=1
( u
n
+ c
2
n
2
u
n
+ u
n
) sin nx =
n=1
s
n
(t) sin nx
(*) u
n
+ u
n
+ c
2
n
2
u
n
= s
n
(t) = A
n
cos wt
For the homogeneous:
Let u
n
= e
t
(
2
+ + c
2
n
2
) = 0 =

2
4c
2
n
2
2
For < 2c ,
2
4c
2
n
2
< 0 complex conjugate roots
u
n
=
_
c
1n
cos
4c
2
n
2
2
2
t + c
2n
sin
4c
2
n
2
2
2
t
_
e
(/2) t
Solution for homogeneous.

Because of damping factor e
(/2) t
there should not be a problem of resonance. We must
nd a particular solution for inhomogeneous.
u
P
n
= B
n
cos wt + C
n
sin wt
u
n
= B
n
w sin wt + C
n
w cos wt
u
n
= B
n
w
2
cos wt C
n
w
2
sin wt
263
Substitute in (*) and compare coecients of cos wt
B
n
w
2
+ C
n
w + c
2
n
2
B
n
= A
n
Compare coecients of sin wt
C
n
w
2
B
n
w + c
2
n
2
C
n
= 0
B
n
=
C
n
(c
2
n
2
w
2
)
w
Cn (c
2
n
2
w
2
)
w
(c
2
n
2
w
2
) + C
n
w = A
n
C
n
w +
(c
2
n
2
w
2
)
2
w
. .
=Dn
= A
n
C
n
=
An
Dn
B
n
=
An (c
2
n
2
w
2
)
Dn w
u
P
n
=
An
Dn
(c
2
n
2
w
2
)
w
cos wt +
An
Dn
sin wt where
D
n
= w +
(c
2
n
2
w
2
)
2
w
A
n
=
_
0
sin nx dx
_
0
sin
2
nxdx
Therefore the general solution of the inhomogeneous is
u
n
=
_
c
1n
cos
4c
2
n
2
2
2
t + c
2n
sin
4c
2
n
2
2
2
t
_
e
(/2) t)
+
An
Dn
c
2
n
2
w
2
w
cos wt +
An
Dn
sin wt
(**) u(x, t) =
n=1
u
n
(t) sin nx
u
t
(x, t) =
n=1
u
n
(t) sin nx
u
n
(t) = (c
1n
cos rt + c
2n
sin rt)
_
2
_
e
2
t
+ (rc
1n
sin rt + rc
2n
cos rt) e
/2 t
An
Dn
(c
2
n
2
w
2
) sin wt +
An
Dn
w cos wt
where r =
4c
2
n
2
2
2
264
u
t
(x, 0) = 0
n=1
_
c
1n
_
2
_
+ rc
2n
+
A
n
D
n
w
_
sin nx = 0
c
1n
2
+ rc
2n
+
An
Dn
w = 0 (#)
u(x, 0) = f(x) u
n
(0) are Fourier coecients of f(x)
u
n
(0) =
_
c
1n
+
A
n
D
n
c
2
n
2
w
2
w
_
=
_
0
f(x) sin nxdx
_
0
sin
2
nxdx
we have c
1n
Use c
1n
in (#) to get c
2n
and the solution is in (**) with u
n
at top of page.
265
3. u
tt
c
2
u
xx
= S(x , t)
u(x, 0) = f(x)
u
t
(x, 0) = 0
a. u(0, t) = A(t)
u(L, t) = B(t) w = x +
= A(t)
L + = B
=
B
L
w =
B(t) A(t)
L
x + A(t)
w
xx
= 0 w
tt
=

B

A
L
x +

A
v = u w
u = v + w
v
tt
c
2
v
xx
= S(x, t) w
tt

S(x, t)
v(x, 0) = f(x)
B(0) A(0)
L
x A(0) F(x)
v
t
(x, 0) = 0 w
t
(x, 0) = 0

B(0)

A(0)
L
x

A(0) G(x)
v(0, t) = 0
v(L, t) = 0
Solve the homogeneous
n
=
_
n
L
_
2
n = 1, 2,
n
= sin
n
L
x
v(x, t) =
n=1
v
n
(t) sin
n
L
x
S(x, t) =
n=1
s
n
(t) sin
n
L
x , s
n
(t) =
_
L
0
S(x, t) sin
n
L
xdx
_
L
0
sin
2 n
L
xdx
n=1
( v
n
+ c
2
_
n
L
_
2
v
n
) sin
n
L
x =
n=1
s
n
sin
n
L
x
266
v
n
+
_
c n
L
_
2
v
n
= s
n
v
n
(0) coecient of expanding F(x)
v
n
(0) coecient of expanding G(x)
v
n
= c
1
..
cos
c n
L
t + c
2
..
sin
c n
L
t +
_
t
0
s
n
()
sin
c n
L
(t )
c n
L
d
v
n
(0)
v
n
(0)
c n
L
(see 8.3.12-13)
u = v + w
267
b. u(0, t) = 0
u
x
(L, t) = 0 Homogeneous. b.c.
n
=
__
n
1
2
_

L
_
2
n
= sin
(n
1
2
)
L
x
n = 1, 2,
u =
n=1
u
n
(t) sin
(n
1
2
)
L
x
S =
n=1
s
n
(t) sin
(n
1
2
)
L
x
u
n
+ c
2
_
(n
1
2
)
L
_
2
u
n
= s
n
u
n
= c
1
cos
_
n
1
2
_
c
L
t +c
2
sin
_
n
1
2
_
c
L
t +
_
t
0
s
n
()
sin (n 1/2)

L
c (t )
__
n
1
2
_
c
L
_
d
u
n
(0) coecients of f(x)
c
2
= 0 (since u
t
= 0)
268
c. u
x
(0, t) = A(t)
u(L, t) = 0
w = ax + b
w
x
(0, t) = A(t) = a a = A(t)
w(L, t) = 0 = aL + b b = aL b = A(t)L
w = A(t)x A(t)L = A(t) (x L)
w
xx
= 0 w
tt
=

A(x L)
v = u w
u = v + w
v
tt
c
2
v
xx
= S(x, t)

A(t) (x L)

S (x, t)
v (x, 0) = f(x) A(0)(x L) F(x)
v
t
(x, 0) = 0

A(0) (x L) G(x)
continue as in b.
269
4a. u
tt
c
2
u
xx
= xt
u(x, 0) = sin x
u
t
(x, 0) = 0
u(0, t) = 1
u(L, t) = t
w(x, t) =
t 1
L
x + 1 ; w
t
=
x
L
w
tt
= 0
v
tt
c
2
v
xx
= xt
v(0, t) = v(L, t) = 0
n
=
_
n
L
_
2
n
= sin
n
L
x n = 1, 2,
v(x, 0) = sin x
_
x
L
+ 1
_
v
t
(x, 0) = 0
x
L
v(x, t) =
n=1
v
n
(t) sin
n
L
x
xt =
n=1
s
n
(t) sin
n
L
x s
n
(t) =
_
L
0
xt sin
n
L
xdx
_
L
0
sin
2 n
L
xdx
sin x +
x
L
1 =
n=1
v
n
(0) sin
n
L
x v
n
(0) =
_
L
0
(sin x +
x
L
1) sin
n
L
xdx
_
L
0
sin
2 n
L
xdx
x
L
=
n=1
v
n
(0) sin
n
L
x v
n
(0) =

_
L
0
x
L
sin
n
L
xdx
_
L
0
sin
2 n
L
xdx
v
n
+
_
n
L
_
2
c
2
v
n
= s
n
(t)
v
n
= c
1
cos c
_
n
t + c
2
sin c
_
n
t +
_
t
0
s
n
()
sin c
n
(t )
c
n
d
v
n
(0) = c
1
v
n
(0) = c
2
c
n
continue as in 3b.
270
b. u
tt
c
2
u
xx
= xt
u(x, 0) = sin x
u
t
(x, 0) = 0
u
x
(0, t) = t
u
x
(L, t) = t
2
w(x, t) =
t
2
t
2L
x
2
+ tx w
t
=
2t 1
2L
x
2
+ x
w
tt
=
x
2
L
w
xx
=
t
2
t
L
Thus w(x, 0) = 0, w
t
(x, 0) = x
x
2
2L
Let v = u w
then
v
tt
c
2
v
xx
= xt
x
2
L
+ c
2
t
2
t
L
. .
s(x,t)
v
x
(0, t) = v
x
(L, t) = 0
n
=
_
n
L
_
2
n
= cos
n
L
x n = 0, 1, 2,
v(x, 0) = sin x since w(x, 0) = 0
v
t
(x, 0) = 0 x +
x
2
2L
v(x, t) =
1
2
v
0
(t) +
n=1
v
n
(t) cos
n
L
x
s(x, t) =
1
2
s
0
(t) +
n=1
s
n
(t) cos
n
L
x
s
n
(t) =
1
L
_
L
0
_
xt
x
2
L
+ c
2
t
2
t
L
_
cos
n
L
xdx n = 0, 1, 2,
sin x =
1
2
v
0
(0) +
n=1
v
n
(0) cos
n
L
x
v
n
(0) =
_
L
0
sin x cos
n
L
xdx
_
L
0
cos
2
n
L
xdx
n = 0, 1, 2,
x +
x
2
2L
=
1
2
v
0
(0) +
n=1
v
n
(0) cos
n
L
x
271
v
n
(0) =
_
L
0
_
x +
x
2
2L
_
cos
n
L
xdx
_
L
0
cos
2
n
L
xdx
n = 0, 1, 2,
v
n
+
_
n
L
_
2
c
2
v
n
= s
n
(t)
v
0
= s
0
(t)
The solution of the ODE for n = 0 is obtained by integration twice and using the initial
conditions
v
0
(t) =
_
t
0
_
_

0
s
0
()d
_
d + v
0
(0) + t v
0
(0)
v
n
= C
n
cos c
n
L
t + D
n
sin c
n
L
t +
_
t
0
s
n
()
sin c
n
L
(t )
c
n
L
d
v
n
(0) = C
n
v
n
(0) = D
n
c
n
L
v
n
(t) = v
n
(0) cos
cn
L
t +
L v
n
(0)
cn
sin
cn
L
t +
_
t
0
s
n
()
sin
cn
L
(t )
cn
L
d
Now that we have all the coecients in the expansion of v, recall that u = v + w.
272
c. u
tt
c
2
u
xx
= xt
u(x, 0) = sin x
u
t
(x, 0) = 0
u(0, t) = 0
u
x
(L, t) = t
w(x, t) = xt ; w
t
= x
w
tt
= 0 w
xx
= 0
w(x, 0) = 0 w
t
(x, 0) = x
v
tt
c
2
v
xx
= xt
v(0, t) = v
t
(L, t) = 0

n
=
_
(n1/2)
L
_
2
n
= sin
(n1/2)
L
x n = 1, 2,
v(x, 0) = sin x
v
t
(x, 0) = x
v(x, t) =
n=1
v
n
(t) sin
(n 1/2)
L
x
xt =
n=1
s
n
(t) sin
(n 1/2)
L
x s
n
(t) =
_
L
0
xt sin
(n1/2)
L
xdx
_
L
0
sin
2 (n1/2)
L
xdx
v(x, 0) = sin x =
n=1
v
n
(0) sin
(n 1/2)
L
x v
n
(0) =
_
L
0
sin x sin
(n1/2)
L
xdx
_
L
0
sin
2 (n1/2)
L
xdx
v
t
(x, 0) = x =
n=1
v
n
(0) sin
(n 1/2)
L
x v
n
(0) =

_
L
0
xsin
(n1/2)
L
xdx
_
L
0
sin
2 (n1/2)
L
xdx
v
n
+
_
(n1/2)
L
_
2
c
2
v
n
= s
n
(t)
v
n
= c
1
cos c
_
n
t + c
2
sin c
_
n
t +
_
t
0
s
n
()
sin c
n
(t )
c
n
d
v
n
(0) = c
1
v
n
(0) = c
2
c
n
continue as in 3b.
273
d. u
tt
c
2
u
xx
= xt
u(x, 0) = sin x
u
t
(x, 0) = 0
u
x
(0, t) = 0
u
x
(L, t) = 1
w(x, t) =
x
2
2L
; w
t
= 0
w
tt
= 0 w
xx
=
1
L
Thus w(x, 0) =
x
2
2L
, w
t
(x, 0) = 0
v
tt
c
2
v
xx
= xt +
c
2
L
. .
s(x,t)
v
x
(0, t) = v
x
(L, t) = 0
n
=
_
n
L
_
2
n
= cos
n
L
x n = 0, 1, 2,
v(x, 0) = sin x
x
2
2L
v
t
(x, 0) = 0
v(x, t) =
1
2
v
0
(t) +
n=1
v
n
(t) cos
n
L
x
s(x, t) =
1
2
s
0
(t) +
n=1
s
n
(t) cos
n
L
x
s
n
(t) =
1
L
_
L
0
_
xt +
c
2
L
_
cos
n
L
xdx n = 0, 1, 2,
v(x, 0) = sin x
x
2
2L
=
1
2
v
0
(0) +
n=1
v
n
(0) cos
n
L
x
v
n
(0) =
1
L
_
L
0
_
sin x
x
2
2L
_
cos
n
L
xdx n = 0, 1, 2,
v
t
(x, 0) = 0 =
1
2
v
0
(0) +
n=1
v
n
(0) cos
n
L
x v
n
(0) = 0 n = 0, 1, 2,
v
n
+
_
n
L
_
2
c
2
v
n
= s
n
(t)
v
0
= s
0
(t)
274
The solution of the ODE for n = 0 is obtained by integration twice and using the initial
conditions
v
0
(t) =
_
t
0
_
_

0
s
0
()d
_
d + v
0
(0)
v
n
= C
n
cos c
n
L
t + D
n
sin c
n
L
t +
_
t
0
s
n
()
sin c
n
L
(t )
c
n
L
d
v
n
(0) = C
n
v
n
(0) = 0 = D
n
c
n
L
D
n
= 0
v
n
(t) = v
n
(0) cos
cn
L
t +
_
t
0
s
n
()
sin
cn
L
(t )
cn
L
d
Now that we have all the coecients in the expansion of v, recall that u = v + w.
275
5. u
tt
u
xx
= 1
u(x, 0) = f(x)
u
t
(x, 0) = g(x)
u(0, t) = 1
u
x
(L, t) = B(t)
w(x, t) = xB(t) + 1; w
t
= x

B(t)
w
tt
= x

B(t) w
xx
= 0
w(x, 0) = xB(0) + 1 w
t
(x, 0) = x

B(0)
v
tt
v
xx
= 1 x

B(t) + 0 S(x, t)
v(x, 0) = f(x) xB(0) 1 F(x)
v
t
(x, 0) = g(x) x

B(0) G(x)
v(0, t) = v
t
(L, t) = 0

n
=
_
(n 1/2)
L
_
2
n
= sin
(n 1/2)
L
x n = 1, 2,
v(x, t) =
n=1
v
n
(t) sin
(n 1/2)
L
x
S(x, t) =
n=1
s
n
(t) sin
(n 1/2)
L
x
s
n
(t) =
_
L
0
S(x, t) sin
(n1/2)
L
xdx
_
L
0
sin
2 (n1/2)
L
xdx
F(x) =
n=1
v
n
(0) sin
(n 1/2)
L
x v
n
(0) =
_
L
0
F(x) sin
(n1/2)
L
xdx
_
L
0
sin
2 (n1/2)
L
xdx
G(x) =
n=1
v
n
(0) sin
(n 1/2)
L
x v
n
(0) =
_
L
0
G(x) sin
(n1/2)
L
xdx
_
L
0
sin
2 (n1/2)
L
xdx
v
n
+
_
(n1/2)
L
_
2
v
n
= s
n
(t)
v
n
= c
1
cos
_
n
t + c
2
sin
_
n
t +
_
t
0
s
n
()
sin
n
(t )
n
d
v
n
(0) = c
1
v
n
(0) = c
2
n
continue as in 3b.
276
8.4 Poissons Equation
Problems
1. Solve
2
u = S(x, y), 0 < x < L, 0 < y < H,
a.
u(0, y) = u(L, y) = 0
u(x, 0) = u(x, H) = 0
Use a Fourier sine series in y.
b.
u(0, y) = 0 u(L, y) = 1
u(x, 0) = u(x, H) = 0
Hint: Do NOT reduce to homogeneous boundary conditions.
c.
u
x
(0, y) = u
x
(L, y) = 0
u
y
(x, 0) = u
y
(x, H) = 0
In what situations are there solutions?
2. Solve the following Poissons equation
2
u = e
2y
sin x, 0 < x < , 0 < y < L,
u(0, y) = u(, y) = 0,
u(x, 0) = 0,
u(x, L) = f(x).
277
1. a.
2
u = s(x, y)
u(0, y) = u(L, y) = 0
u(x, 0) = u(x, H) = 0 sin
n
H
y
Use a Fourier sine series in y (we can also use a Fourier sine series in x or a double Fourier
sine series, because of the boundary conditions)
u(x, y) =
n=1
u
n
(x) sin
n
H
y
S(x, y) =
n=1
s
n
(x) sin
n
H
y
s
n
(x) =
_
H
0
s(x, y) sin
n
H
y dy
_
H
0
sin
2 n
H
y dy
n=1
_
n
H
_
2
u
n
sin
n
H
y
. .
uyy
+ u
n
sin
n
H
y
. .
uxx
n=1
s
n
(x) sin
n
H
y
u
n
(x)
_
n
H
_
2
u
n
(x) = s
n
(x)
Boundary conditions are coming from u(0, y) = u(L, y) = 0
n=1
u
n
(0) sin
n
H
y = 0 u
n
(0) = 0
n=1
u
n
(L) sin
n
H
y = 0 u
n
(L) = 0
(*) u
n
(x) =
sinh
n
H
(L x)
n
H
sinh
n
H
L
_
x
0
s
n
() sinh
n
H
d
+
sinh
n
H
x
n
H
sinh
n
H
L
_
L
x
s
n
() sinh
n
H
(L ) d
Lets check by using (*)
u
n
(0) = 1
st
term the integral is zero since limits are same
2
nd
term the numerator is zero = sinh
n
H
0
u
n
(L) = 1
st
term the numerator sinh
n
H
(L L) = 0
2
nd
term the integral is zero since limits of integration are the same.
278
u
n
=

n
H
cosh
n
H
(L x)
n
H
sinh
n
H
L
_
x
0
s
n
() sinh
n
H
d +
sinh
n
H
(L x)
n
H
sinh
n
H
L
s
n
(x) sinh
n
H
x
. .
integrand at upper limit
+
n
H
cosh
n
H
x
n
H
sinh
n
H
L
_
L
x
s
n
() sinh
n
H
(L) d +
sinh
n
H
x
n
H
sinh
n
H
L
_
s
n
(x) sinh
n
H
(L x)
_
. .
integrand at lower limit
Lets add the second and fourth terms up
1
n
H
sinh
n
H
L
s
n
(x)
sinh
n
H
(L x) sinh
n
H
x sinh
n
H
x sinh
n
H
(L x)
. .
=0
u
n
=
_
n
H
_
2
sinh
n
H
(L x)
n
H
sinh
n
H
L
_
x
0
s
n
() sinh
n
H
d +
n
H
cosh
n
H
(L x)
n
H
sinh
n
H
L
s
n
(x) sinh
n
H
x
+
_
n
H
_
2
sinh
n
H
x
n
H
sinh
n
H
L
_
L
x
s
n
() sinh
n
H
(L) d +
n
H
cosh
n
H
x
n
H
sinh
n
H
L
_
s
n
(x) sinh
n
H
(L x)
_
Lets add the second and fourth terms up
sn (x)
sinh
n
H
L
sinh
n
H
x cosh
n
H
(L x) + cosh
n
H
x sinh
n
H
(L x)
. .
=sinh
n
H
(x (Lx)) =sinh
n
H
L
= s
n
(x)
The integral terms in u
n
are exactly
_
n
H
_
2
u
n
and thus the ODE is satised.
279
b.
2
u = S(x, y)
u(0, y) = 0 u(L, y) = 1
u(x, 0) = u(x, H) = 0 sin
n
H
y
Use a Fourier sine series in y
u(x, y) =
n=1
u
n
(x) sin
n
H
y
S(x, y) =
n=1
s
n
(x) sin
n
H
y
s
n
(x) =
_
H
0
S(x, y) sin
n
H
y dy
_
H
0
sin
2 n
H
y dy
n=1
_
n
H
_
2
u
n
sin
n
H
y
. .
uyy
+ u
n
sin
n
H
y
. .
uxx
n=1
s
n
(x) sin
n
H
y
u
n
(x)
_
n
H
_
2
u
n
(x) = s
n
(x)
Boundary conditions are coming from u(0, y) = 0 u(L, y) = 1
n=1
u
n
(0) sin
n
H
y = 0 u
n
(0) = 0
n=1
u
n
(L) sin
n
H
y = 1 u
n
(L) =
2
H
_
H
0
1 sin
n
H
y dy
u
n
(L) =
4
n
for n odd and 0 for n even. (see (5.8.1)
For n even the solution is as in 1a (since u
n
(L) = 0)
For n odd, how would the solution change?
Let
n
=
4
nL
x, then
n
= 0
Let v
n
= u
n

n
then v
n
(0) = 0 and v
n
(L) = u
n
(L)
n
(L) = 0
and
v
n

_
n
H
_
2
v
n
=
_
n
H
_
2
4x
nL
+ s
n
. .
This is the sn to be used in (*) in 1a
280
c.
2
u = S(x, y)
u
x
(0, y) = 0 u
x
(L, y) = 0 cos
n
L
x
u
y
(x, 0) = u
y
(x, H) = 0 cos
m
H
y
Use a double Fourier cosine series
u(x, y) =
n=0
m=0
u
nm
cos
n
L
x cos
m
H
y
S(x, y) =
n=0
m=0
s
nm
cos
n
L
x cos
m
H
y
s
nm
=
_
H
0
_
L
0
S(x, y) cos
m
H
y cos
n
L
xdxdy
_
H
0
_
L
0
cos
2
m
H
y cos
2
n
L
xdxdy
n=0
m=0
(u
nm
)
_
_
n
L
_
2
+
_
m
H
_
2
_
cos
n
L
x cos
m
H
y = S(x, y)
Thus
u
nm
_
_
n
L
_
2
+
_
m
H
_
2
_
= s
nm
Substituing for s
nm
, we get the unknowns u
nm
u
nm
=
_
H
0
_
L
0
S(x, y) cos
m
H
y cos
n
L
xdxdy
_
_
n
L
_
2
+
_
m
H
_
2
_
_
H
0
_
L
0
cos
2
m
H
y cos
2
n
L
xdxdy
What if
nm
= 0? (i.e. n = m = 0)
Then we cannot divide by
nm
but in this case we have zero on the left
_
H
0
_
L
0
S(x, y) dxdy = 0
This is typical of Neumann boundary conditions.
281
2.
2
u = e
2y
sin x
u(0, y) = 0 u(, y) = 0 sin nx
u(x, 0) = 0 u(x, L) = f(x)
Use a Fourier sine series in x
u(x, y) =
n=1
u
n
(y) sin nx
S(x, y) is already in a Fourier sine series in x with the coecients s
1
(y) = e
2y
and all
the other coecients are zero.
n=1
n
2
u
n
sin nx
. .
uxx
+ u
n
sin nx
. .
uyy
n=1
s
n
(x) sin nx
u
n
(y) n
2
u
n
(y) = 0 for n ,= 1
u
1
(y) u
1
(y) = e
2y
Boundary conditions are coming from u(x, 0) = 0 u(x, L) = f(x)
n=1
u
n
(0) sin nx = 0 u
n
(0) = 0
n=1
u
n
(L) sin nx = f(x) u
n
(L) =
2
L
_
L
0
f(x) sin nxdx
The solution of the ODEs is
u
1
(y) =
1
3
e
2y
. .
particular solution
+
1
sinh y +
1
cosh y
and
u
n
(y) =
n
sinh ny +
n
cosh ny n ,= 1
Since u
1
(0) = 0 we have
1
3
+
1
= 0
1
=
1
3
Since u
n
(0) = 0 we have
n
= 0 n ,= 1
Using u
1
(L) we have
1
3
e
2L
+
1
sinh L
1
3
cosh L = u
1
(L). This gives a value for
1
1
=
u
1
(L) +
1
3
cosh L
1
3
e
2L
sinh L
Using u
n
(L) we get a vlaue for
n
282
n
=
u
n
(L)
sinh nL
n ,= 1
Now we can write the solution
u(x, y) =
_
1
sinh y
1
3
cosh y +
1
3
e
2L
_
sin x +
n=2
_
u
n
(L)
sinh nL
sinh ny
_
sin nx
with
1
as above.
283
9 Fourier Transform Solutions of PDEs
9.1 Motivation
9.2 Fourier Transform pair
Problems
1. Show that the Fourier transform is a linear operator, i. e.
T (c
1
f(x) + c
2
g(x)) = c
1
T (f(x)) + c
2
T (g(x)) .
2. If F() is the Fourier tranform of f(x), show that the inverse Fourier transform of
e
i
F() is f(x ) . This is known as the shift theorem.
3. Determine the Fourier transform of
f(x) =
_
0 [x[ > a
1 [x[ < a .
4. Determine the Fourier transform of
f(x) =
_
x
0
(t)dt.
5. Prove the scaling theorem
T (f(ax)) =
1
[a[
T (f(x)) .
6. If F() is the Fourier tranform of f(x), prove the translation theorem
T
_
e
iax
f(x)
_
= F( a).
284
1.
T (c
1
f(x) + c
2
g(x)) =
1
2
_

(c
1
f(x) + c
2
g(x)) e
ix
dx
The integral of a sum is the sum of the integrals:
= c
1
1
2
_

f(x)e
ix
dx
. .
F(f(x))
+c
2
1
2
_

g(x)e
ix
dx
. .
F(g(x))
= c
1
T (f(x)) + c
2
T (g(x))
2.
T
1
_
e
i
F()
_
=
_

F()e
i
e
ix
d
=
_

F()e
i(x)
d
= f(x )
3.
T (f(x)) =
1
2
_

f(x)e
ix
dx
=
1
2
_
a
a
1 e
ix
dx
=
1
2
1
i
e
ix
a
a
=
1
2i
_
e
ia
e
ia
_
. .
2i sin a
=
2i sin a
2i
=
sin a
285
4. Dierentiating
f(x) =
_
x
(t)dt
we get
df
dx
= (x).
Now we can either use the denition (9.2.1) or we can use (9.4.2) saying
T
_
df
dx
_
= iT(f)
to get
T() = iT
__
x
(t)dt
_
.
Therefore dividing by i
T
__
x
(t)dt
_
=
1
i
T().
Remember the assumption is required because we need the function to be transformed to
tend to zero at innity.
5. Say a > 0, then
T (f(ax)) =
1
2
_

e
ix
f(ax)dx
can be transformed by the substitution y = ax to (remember dy = a dx)
T (f(ax)) =
1
2a
_

e
i/ay
f(y)dy =
1
a
F
_
a
_
If a < 0 then the transformation reverses the limits of integration and that will pull a negative
sign in front. So we have
1
a
multiplying the integral and thats
1
[a[
in this case.
286
6.
T
_
e
iax
f(x)
_
=
1
2
_

e
iax
f(x)e
ix
dx
=
1
2
_

f(x)e
i(a)x
dx
= F( a)
287
9.3 Heat Equation
Problems
1. Use Fourier transform to solve the heat equation
u
t
= u
xx
+ u, < x < , t > 0,
u(x, 0) = f(x) .
288
1.
The Fourier transform of the equation is
U
t
(, t) =
2
U + U = (1
2
)U
subject to
U(, 0) = F()
Thus
U(, t) = e
t
F()e
2
t
The inverse Fourier transform is
u(x, t) = e
t
_

F()e
2
t
e
ix
d
since e
t
is independent of . The integral is the same as was done in class and the solution
is (see (9.3.8))
u(x, t) = e
t
1
4t
_

f()e
(x)
2
4t
d
289
9.4 Fourier Transform of Derivatives
Problems
1. Solve the diusion-convection equation
u
t
= ku
xx
+ cu
x
, < x < ,
u(x, 0) = f(x) .
2. Solve the linearized Korteweg-de Vries equation
u
t
= ku
xxx
, < x < ,
u(x, 0) = f(x) .
u
xx
+ u
yy
= 0 , 0 < x < L, < y < ,
subject to
u(0, y) = g
1
(y) ,
u(L, y) = g
2
(y) .
u
tt
= u
xx
, < x < ,
u(x, 0) = 0 ,
u
t
(x, 0) = g(x) .
u
xx
+ u
yy
= 0 , < x < , y > 0
u
y
(x, 0) = g(x) .
290
1. Use Fourier transform to solve the heat equation
u
t
= ku
xx
+ cu
x
, < x < , t > 0,
u(x, 0) = f(x) .
Taking the Fourier transform, we have
U
t
= k
2
U + icU, t > 0,
U(, 0) = F() .
The solution of this initial value problem in t is
U(, t) = F()e
(k
2
ic)t
Now we nd the inverse Fourier transform
u(x, t) =
_

F()e
(k
2
ic)t
e
ix
d
u(x, t) =
_

e
ict
F()
. .
H()
e
k
2
t
. .
G()
e
ix
d
Therefore
u(x, t) = h g, using the convolution theorem
where
h = T
1
(H()) = f(x + ct) see problem 2 in section 9.2
g = T
1
(G()) =
_

kt
e
x
2
4kt
Therefore
u(x, t) =
1
2
_

f( + ct)
_

kt
e
(x)
2
4kt
d
291
2.
The Fourier transform is
U
t
= (i)
3
kU = ik
3
U
The solution of this dierential equation with the initial condition U(, 0) = F() is given
by
U = F() e
ik
3
t
Let
G() = e
ik
3
t
then
U(, t) = F()G()
We can now use the convolution theorem. If
g(x) =
_

e
ik
3
t
e
ix
d =
_

e
i(k
3
tx)
d
then
u(x, t) =
1
2
_

f()g(x )d
The question is how to nd g(x)
Let k
3
t = s
3
/3 after using symmetry we get
g(x) =
_

e
i(k
3
tx)
d = 2
_

0
cos(k
3
tx)d =
2
(3kt)
1/3
_

0
cos
_
s
3
3

sx
(3kt)
1/3
_
ds
g(x) =
2
(3kt)
1/3
A
i
_
x
(3kt)
1/3
_
where A
i
(x) is the Airy function (the solution of y
xy = 0 satisfying lim
x
y = 0 and
y(0) =
3
2/3
(2/3)
.)
The plot of Airy function Ai(x) is given as gure 60.
292
10 8 6 4 2 0 2 4 6 8 10
0.8
0.6
0.4
0.2
0
0.2
0.4
0.6
Airy function
Figure 60: Airy function
3.
u
xx
+ u
yy
= 0 , 0 < x < L, < y < ,
subject to
u(0, y) = g
1
(y) ,
u(L, y) = g
2
(y) .
Use Fourier transform in y
U
xx
(x, )
2
U(x, ) = 0
U(0, ) = G
1
()
U(L, ) = G
2
()
The solution is
U(x, ) = G
2
()
sinh x
sinh L
+ G
1
()
sinh (L x)
sinh L
Now take the inverse transform.
293
4. The Fourier transform of the PDE and the initial conditions
U
tt
=
2
U
U(, 0) = 0
U
t
(, 0) = G()
The solution is
U = A() cos t + B() sint
where A() = 0 and B() = G()/
U(, t) =
G()
sin t
Using the inverse transform formula
u(x, t) =
_

G()
sin t
e
ix
d
This is a convolution of g(x) with the function
f(x) =
_
0 [x[ > t
[x[ < t .
since (see table of transforms)
T(h(x)) =
1
sin a
for
h(x) =
_
0 [x[ > a
1 [x[ < a .
Note that h(x) = f(x)
u(x, t) =
1
2
_

g(x )
. .
initial condition
f()
. .
if ||<t
d
u(x, t) =
1
2
_
t
t
g(x ) d
294
5. Let v = u
y
, then v satises Laplaces equation subject to v(x, 0) = g(x). Based on the
results of the example in this section
v(x, y) =
1
2
_

g()
2y
(x )
2
+ y
2
d .
Therefore
u(x, y) =
1
_
y
_

g()

(x )
2
+
2
dd
Now reverse the order of integration and execute the integration with respect to , we have
u(x, y) =
1
log
_
(x )
2
+ y
2
_
g()d .
295
9.5 Fourier Sine and Cosine Transforms
Problems
1. Derive the Fourier cosine transform of e
x
2
.
2. Determine the inverse cosine transform of e
(Hint: use dierentiation with respect

to a parameter)
3. Solve by Fourier sine transform:
u
t
= ku
xx
, x > 0 , t > 0
u(0, t) = 1 ,
u(x, 0) = f(x) .
u
t
= ku
xx
, x > 0 , t > 0
u
x
(0, t) = 0 ,
u(x, 0) = f(x) .
5. Prove the convolution theorem for the Fourier sine transforms, i.e. (9.5.14) and (9.5.15).
6. Prove the convolution theorem for the Fourier cosine transforms, i.e. (9.5.16).
7.
a. Derive the Fourier sine transform of f(x) = 1.
b. Derive the Fourier cosine transform of f(x) =
_
x
0
(t)dt.
c. Derive the Fourier sine transform of f(x) =
_
x
0
(t)dt.
8. Determine the inverse sine transform of
1
(Hint: use integration with respect to a

parameter)
296
1.
C
_
e
x
2
_
=
2
_

0
e
x
2
cos xdx by denition
Using the symmetry (since the integrand is an even function) we have
C
_
e
x
2
_
=
1
e
x
2
e
ix
dx
Recall the relationship between the cosine and complex exponentials.
C
_
e
x
2
_
= 2T
_
e
x
2
_
=
1
2
4
297
2.
C
1
_
e
_
=
_

0
e
. .
(e
)
cos xd
=

_

0
e
cos xd
=

C
_
e
_
. .
=

x
2
+
2
from table
=

x
2
+
2
=
1 (x
2
+
2
) 2
(x
2
+
2
)
2
=
x
2
+
2
2
2
(x
2
+
2
)
2
=
x
2
2
(x
2
+
2
)
2
298
3. Since u is given on the boundary, we use the Fourier sine transform:
U
t
= k
_
2
u(0, t)
2
U
_
Substitute the boundary condition, we get
U
t
=
2k

2
kU
The solution of this ODE is
U(, t) = c() e
k
2
t
+
2
Use the initial condition

F() = U(, 0) = c() +
2
Therefore
c() = F()
2
Plug this c in the solution

U(, t) =
_
F()
2
_
e
k
2
t
+
2
This can be written as

S[u] = U(, t) = F() e
k
2
t
. .
S[f]C[g]
+
2
e
k
2
t
where
G() = e
k
2
t
We now use convolution
S[f] C[g] =
1
_

0
f() [g(x ) g(x + )] d
where g is the inverse cosine transform of G.
Since
C[e
x
2
] =
2
4
e
2
4
we need
1
4
= kt or =
1
4kt
, so
C[e
x
2
4kt
] =
2
_
4
1
4kt
e
kt
2
299
or
1
2
_

kt
C[e
x
2
4kt
] = e
kt
2
Therefore
g =
1
2
_

kt
e
x
2
4kt
Now the rst term is
1
2
_

kt
1
_

0
f()
_
e
(x)
2
4kt
e
(x+)
2
4kt
_
d
The second term is
S
1
[
2
] = 1
The last term is again by convoultion of 1 with the same function g, that is
1
2
_

kt
1
_

0
1
_
e
(x)
2
4kt
e
(x+)
2
4kt
_
d
If we decide to use (9.5.15) then remember that the inverse sine transform is giving the
constant 1 for x 0. Combining all these terms
u(x, t) =
1
2
1
kt
_

0
f()
_
e
(x)
2
4kt
e
(x+)
2
4kt
_
d +1
1
2
1
kt
_

0
1
_
e
(x)
2
4kt
e
(x+)
2
4kt
_
d
u(x, t) =
1
4kt
_

0
[f() 1]
_
e
(x)
2
4kt
e
(x+)
2
4kt
_
d + 1
300
4. Since the boundary condition is on u
x
, we have to use Fourier cosine transform:
U
t
= k
2
U since u
x
(0, t) = 0
The solution is
U = F()
. .
C[f]
e
k
2
t
. .
C[g]
This g is exactly the same as in the previous problem.
301
5. To prove (9.5.14), we start with the denition of inverse Fourier sine transform
S
1
[H] = h(x) =
_

0
S(f)C(g) sinxd
Substitute for the Fourier sine transform of f
S
1
[H] = h(x) =
_

0
_
2
_

0
f() sin d
_
C(g) sinxd
We now rearrange and put the integral over inside
S
1
[H] = h(x) =
2
_

0
f()
_

0
C(g) sin sin xd d
We can use the trigonometric identity
sin sin x =
1
2
cos (x )
1
2
cos (x + )
and get two integrals
S
1
[H] = h(x) =
1
_

0
f()
__

0
C(g) cos (x ) d
_

0
C(g) cos (x + ) d
_
d
Now each of the inner integrals is inverse Fourier cosine transform of g at x and x + .
Thus
S
1
[H] = h(x) =
1
_

0
f() [g(x ) g(x + )] d
To prove (9.5.15), we substitute for the inverse Fourier cosine transform of g and go through
similar arguments
S
1
[H] = h(x) =
_

0
_
2
_

0
g() cos d
_
S(f) sinxd
S
1
[H] = h(x) =
2
_

0
g()
_

0
S(f) cos sin xd d
cos sin x =
1
2
sin ( + x)
1
2
sin ( x)
S
1
[H] = h(x) =
1
_

0
g()
__

0
S(f) sin( + x) d
_

0
S(f) sin ( x) d
_
d
Now each of the inner integrals is inverse Fourier sine transform of f at + x and x.
Thus
S
1
[H] = h(x) =
1
_

0
g() [f( + x) f( x)] d
302
6. To prove (9.5.16), we start with the denition of inverse Fourier cosine transform
C
1
[H] = h(x) =
_

0
C(f)C(g) cos xd
Substitute for the Fourier cosine transform of g
C
1
[H] = h(x) =
_

0
_
2
_

0
g() cos d
_
C(f) cos xd
C
1
[H] = h(x) =
2
_

0
g()
_

0
C(f) cos cos xd d
cos cos x =
1
2
cos (x ) +
1
2
cos (x + )
C
1
[H] = h(x) =
1
_

0
g()
__

0
C(f) cos (x ) d +
_

0
C(f) cos (x + ) d
_
d
Now each of the inner integrals is inverse Fourier cosine transform of f at x and x + .
Thus
C
1
[H] = h(x) =
1
_

0
g() [f(x ) f(x + )] d
303
7.
a. By denition
S(1) =
2
_

0
sin xdx
=
2
cos x =
2
.
b. The Fourier cosine transform C
__
x
0
(t)dt
_
can be derived by using
(x) =
df
dx
and (9.5.6)
S
_
df
dx
_
= C(f).
Divide by we get
C
__
x
0
(t)dt
_
=
1
S()
c. As in part c. the Fourier sine transform S
__
x
0
(t)dt
_
can be derived by using
(x) =
df
dx
and (9.5.5)
C
_
df
dx
_
=
2
f(0) + S(f)
for
f(x) =
_
x
0
(t)dt
which satises f(0) = 0. Divide by we get
S
__
x
0
(t)dt
_
=
1
C()
304
8. Note that
1
=
_

e
t
dt.
Therefore
S
1
_
1
_
= S
1
__

e
t
dt
_
=
_

x
x
2
+ t
2
dt
=
_

1
x(1 + (t/x)
2
dt
=
1
x
arctan(t/x)
t
t=
=

2x

1
x
arctan
_
x
_
.
305
9.6 Fourier Transform in 2 Dimensions
Problems
u
tt
= c
2
2
u, < x < , < y < ,
u(x, y, 0) = f(x, y),
u
t
(x, y, 0) = 0.
306
1. Fourier transform in two dimension of the given wave equation yields:
U
tt
= c
2
_
[[
2
_
U = c
2
[[
2
U(
1
,
2
, t)
The solution is
U(
1
,
2
, t) = A() cos c[[t + B() sin c[[t
Using the Initial conditions in the transform domain, we get
U(
1
,
2
, t) = F() cos c[[t
. .
G()
By the convolution theorem
u(x, t) =
1
(2)
2
_

f( r
0
) g(r r
0
) d r
0
We only need to nd
g(r) =
_

cos c[[t e
i r
d
307
10 Greens Functions
10.1 Introduction
10.2 One Dimensional Heat Equation
Problems
1. Consider the heat equation in one dimension
u
t
=

2
u
x
2
+ Q(x, t), 0 < x < 1, t > 0,
u(x, 0) = f(x),
u(0, t) = A(t),
u(1, t) = B(t).
Obtain a solution in the form (10.2.16).
2. Consider the same problem subject to the homogeneous boundary conditions
u
x
(0, t) = u
x
(1, t) = 0
a. Obtain a solution by any method.
b. Obtain a solution in the form (10.2.16).
3. Solve the wave equation in one dimension
2
u
t
2
=

2
u
x
2
+ Q(x, t), 0 < x < 1, t > 0,
u(x, 0) = f(x),
u
t
(x, 0) = g(x),
u(0, t) = 0,
u(1, t) = 0.
Dene functions such that a solution in a similar form to (10.2.16) exists.
4. Solve the above wave equation subject to
a.
u
x
(0, t) = u
x
(1, t) = 0.
b.
u
x
(0, t) = 0, u
x
(1, t) = B(t).
c.
u(0, t) = A(t), u
x
(1, t) = 0.
308
1. u
t
= u
xx
+ Q(x, t) 0 x 1 t > 0
u(x, 0) = f(x)
u(0, t) = A(t)
u(1, t) = B(t)
Solution: Let w(x, t) = A(t) + x[B(t) A(t)] and let
v(x, t) = u(x, t) w(x, t)
Then v
t
= v
xx
+ y(x, t) where y(x, t) = Q(x, t) w
t
+ w
xx
v(x, 0) = g(x) f(x) A(0) x[B(0) A(0)]
v(0, t) = v(1, t) = 0
The Homogeneous solution has eigenfunctions and eigenvalues
n
(x) = sin(n x),
n
= (n)
2
, n = 1, 2, . . .
y(x, t) =
n=1
y
n
(t)
n
(x)
where
y
n
(t) =
_
1
0
y(x, t) sin(n x) dx
_
1
0
sin
2
(n x) dx
= 2
_
1
0
y(x, t) sin(n x) dx
Let v(x, t) =
n=1
v
n
(t)
n
(x) then v(x, 0) = g(x) =
n=1
v
n
(0)
n
(x)
So v
n
(0) =
_
1
0
g(x) sin(n x) dx
_
1
0
sin
2
(n x) dx
= 2
_
1
0
g(x) sin(n x) dx
Substitute in the equation:
n=1
v
n
(t) sin(n x) =
n=1
((n)
2
)v
n
(t) sin(n x) +
n=1
y
n
(t) sin(n x)
or
n=1
_
v
n
(t) + (n)
2
v
n
(t) y
n
(t)
_
sin(n x) = 0
so v
n
(t) + (n)
2
v
n
(t) = y
n
(t)
where v
n
(0) = 2
_
1
0
g(x) sinn x dx
This has a solution v
n
(t) = v
n
(0)e
(n)
2
t
+
_
t
0
y
n
()e
(n )
2
(t)
d
(using variation of parameters). Thus
309
v(x, t) =
n=1
sin(n x)
_
e
(n)
2
t
2
_
1
0
g(s) sin(n s)ds +
_
t
0
2
_
1
0
y(s, ) sin(n s)ds e
(n)
2
(t)
_
d
=
_
1
0
g(s)
_
2
n=1
sin(n x)e
(n)
2
t
sin(n s)
_
ds
+
_
1
0
_
t
0
y(s, )
_
2
n=1
sin(n x)e
(n )
2
(t)
sin(n s)
_
dds
so v(x, t) =
_
1
0
[f(x) A(0) x[B(0) A(0)] ] G(x; s, t)ds
+
_
1
0
_
t
0
[Q(x, t) A
(t) + x[B
(t) A
(t)] ] G(x; s, t )dds

where G(x; s, t) = 2
n=1
sin(n x)e
(n)
2
t
sin(n s)
310
2. u
t
= u
xx
+ Q(x, t) 0 x 1 t > 0
u(x, 0) = f(x)
u
x
(0, t) = u
x
(1, t) = 0
(a). The homogeneous solution has eigenfunctions and eigenvalues
n
(x) = cos(n x)
n
= (n)
2
n = 0, 1, 2, . . .
so u(x, t) =
n=0
a
n
(t) cos(n x)
a
n
(0) =
_
1
0
f(x) cos(n x) dx
_
1
0
cos
2
(n x) dv
= 2
_
1
0
f(x) cos(n x) dx
Expanding Q(x, t) =
n=0
q
n
(t) cos(n x)
where the coecients
q
n
(t) = 2
_
1
0
Q(x, t) cos(n x) dx
so a
n
(t) = a
n
(0)e
(n)
2
t
+ e
(n)
2
t
_
t
0
q
n
()e
(n)
2
d
Thus
u(x, t) =
n=0
cos(n x)
_
2
_
1
0
f(s) cos(n s)ds e
(n)
2
t
+
e
(n)
2
t
_
t
0
_
1
0
2Q(s, ) cos(n s)ds e
(n)
2
d
_
=
_
1
0
f(s)
_

n=0
2 cos(n x) cos(n s)e
(n)
2
t
_
ds+
_
1
0
_
t
0
Q(s, )
_

n=0
(n)
2
(t)
_
dds
(b). u(x, t) =
_
1
0
f(s)G(x; s, t)ds +
_
1
0
_
t
0
Q(s, )G(x; s, t )dds
where G(x; s, t) =
n=0
(n)
2
t
311
3. u
tt
= u
xx
+ Q(x, t) 0 x 1 t > 0
u(x, 0) = f(x)
u
t
(x, 0) = g(x)
u(0, t) = u(1, t) = 0
The homgeonous solution has eigenfunctions and eigenvalues
Q
n
(x) = sin(n x),
n
= (n)
2
n = 1, 2, . . .
Q(x, t) =
n=1
q
n
(t) sin(nx) so q
n
(t) = 2
_
1
0
Q(x, t) sin(n x) dx
Let u(x, t) =
n=1
A
n
(t) sin(nx)
Then f(x) =
n=1
A
n
(0) sin(nx) and g(x) =
n=1
A
n
(0) sin(nx)
So A
n
(t) + (n)
2
A
n
(t) = q
n
(t) where
A
n
(0) = 2
_
1
0
f(x) sin(nx) dx and A
n
(0) = 2
_
1
0
g(x) sin(nx) dx
The solution is then
A
n
(t) = K
1n
cos(nt) + K
2n
sin(nt) + [A
n
(t)]
p
where the particular solution is
[A
n
(t)]
p
=
cos(nt)
n
_
t
0
sin(n)q
n
()d +
sin(nt)
n
_
t
0
cos(n)q
n
()d
K
1n
= 2
_
1
0
f(s) sin(ns)ds and K
2n
=
2
n
_
1
0
g(s) sin(ns)ds
So u(x, t) =
n=1
sin(nx)
_
2
_
1
0
f(s) sin(ns)ds cos(nt) +
2
n
_
1
0
g(s) sin(ns)ds sin(nt)
cos(nt)
n
_
t
0
sin(n)2
_
1
0
Q(s, ) sin(ns)dsd
+
sin(nt)
n
_
t
0
cos(n)2
_
1
0
Q(s, ) sin(ns)dsd
_
Rearrange:
u(x, t) =
_
1
0
f(s)
_

n=1
2 sin(nx) sin(ns) cos(nt)
_
ds
312
+
_
1
0
g(s)
_

n=1
2
n
sin(nx) sin(ns) sin(nt)
_
ds
+
_
1
0
_
t
0
Q(s, )
_

n=1
2
n
sin(nx) sin(ns)(cos(nt)) sin(n)
_
dds
+
_
1
0
_
t
0
Q(s, )
_

n=1
2
n
sin(nx) sin(ns) sin(nt) cos(n)
_
dds
u(x, t) =
_
1
0
f(s)
_

n=1
2 sin(nx) sin(ns) cos(nt)
_
ds
+
_
1
0
g(s)
_

n=1
2
n
_
ds
+
_
1
0
_
t
0
Q(s, )
_

n=1
2
n
sin(nx) sin(ns) sin(nt) cos(n) cos(nt) sin(n)
_
dds
But sin(nt) cos(nt) cos(nt) sin(n) = sin(n(t ))
So Let G(x; s, t) =
n=1
2
n
Then u(x, t) =
_
1
0
f(s)G
t
(x; s, t)ds +
_
1
0
g(s)G(x; s, t)ds
+
_
1
0
_
t
0
Q(s, )G(x; s, t )dds
313
4 a. u
tt
= u
xx
+ Q(x, t) 0 x 1 t > 0
u(x, 0) = f(t)
u
t
(x, 0) = g(x)
u
x
(0, t) = u
x
(1, t) = 0
For the homogenous problem, the eigenfunctions and eigenvalues are
n
(x) = cos(n x),
n
= (n)
2
n = 0, 1, 2, . . .
Expand in terms of the eigenfunctions:
Q(x, t) =
n=1
q
n
(t) cos(n x)
where q
n
(t) = 2
_
1
0
Q(x, t) cos(n x) dx n = 1, 2, . . .
Let u(x, t) =
n=0
A
n
(t) cos(n x).
For n = 0:
A
0
(0) =
_
1
0
f(x) dx
A
0
(0) =
_
1
0
g(x) dx
q
0
(t) =
_
1
0
Q(x, t) dx
A
0
(t) = q
0
(t) =
_
1
0
Q(x, t) dv
A
0
(t) =
t
2
2
_
1
0
Q(x, t) dx + c
1
t + c
2
A
0
(0) =
_
1
0
f(x) dx = c
2
A
0
(0) =
_
1
0
g(x) dx = c
1
A
0
(t) =
t
2
2
_
1
0
Q(s, t)ds + t
_
1
0
f(s) ds +
_
1
0
g(s) ds
For n > 0:
u(x, 0) = f(x) =
n=1
A
n
(0) cos(n x)
where A
n
(0) = 2
_
1
0
f(x) cos(nx) dx
314
u
t
(x, 0) = g(x) =
n=1
A
n
(0) cos(n x)
where A
n
(0) = 2
_
1
0
g(x) cos(nx) dx
A
n
(t) + (n)
2
A
n
(t) = q
n
(t)
A
n
(t) = k
1n
cos(nt) + k
2n
sin(nt) + [A
n
(t)]
p
[A
n
(t)]
p
=
cos(nt)
n
_
t
0
sin(n)q
n
()d +
sin(nt)
n
_
t
0
cos(n)q
n
()d
k
1n
= 2
_
1
0
f(s) cos(ns)ds
k
2n
=
2
n
_
1
0
g(s) sin(ns)ds
So u(x, t) =
n=1
cos(n x)
_
2
_
1
0
f(s) cos(ns)ds cos(nt)
+
2
n
_
1
0
g(s) cos(ns)ds sin(nt)
cos(nt)
n
_
t
0
sin(n)2
_
1
0
Q(s, ) cos(ns)dsd
+
sin(nt)
n
_
t
0
cos(n)2
_
t
0
Q(s, ) cos(ns)dsd
_
+ A
0
(t)
u(x, t) =
_
1
0
f(s)
_

n=1
2 cos(n x) cos(ns) cos(nt)
_
ds
+
_
1
0
g(s)
_

n=1
2
n
cos(n x) cos(ns) sin(nt)
_
ds
+
_
1
0
_
t
0
Q(s, )
_

n=1
2
n
cos(n x) cos(ns) sin(n(t ))
_
dds
+A
0
(t)
where we have used the identity
[sin(nt) cos(n) cos(nt) sin(n)] = sin(n(t ))
Let G(x; s, t) =
n=1
2
n
cos(n x) cos(ns) sin(nt)
315
Then u(x, t) = +
_
1
0
f(s)G
t
(x; s, t) ds +
_
1
0
g(s)G(x; s, t) ds
_
1
0
_
t
0
Q(s, )G(x; s, t )dds +
t
2
2
_
1
0
Q(s, t)ds +t
_
1
0
f(s)ds +
_
1
0
g(s)ds
316
4b. Same as 4a, but u
x
(0, t) = 0 u
x
(1, t) = (t)
Let w(x, t) =
x
2
2
(t), so w
x
(0, t) = 0 w
x
(1, t) = (t)
v(x, t) = u(x, t) w(x, t)
v(x, 0) = f(x)
x
2
2
(0)
v
t
(x, 0) = g(x)
x
2
2

(0)
v
x
(0, t) = v
x
(1, t) = 0
v
tt
= v
xx
+ Q(x, t) w
tt
(x, t) + w
xx
(x, t)
w
tt
(x, t) =
x
2
2

(t)
w
xx
(x, t) = (t)
So u(x, t) = v(x, t) + w(x, t) where v(x, t) is as in 4a.
with

f(x) = f(x)
x
2
2
(0), g(x) = g(x)
x
2
2

(0)
and

Q(x, t) = Q(x, t)
x
2
2

(t) + (t)
u(x, t) =
x
2
2
(t) +
t
2
2
_
1
0
_
Q(s, t)
s
2
2

(t) + (t)
_
ds
+ t
_
1
0
(f(s)
s
2
2
(0))ds +
_
1
0
(g(s)
s
2
2

(0))ds
+
_
1
0
(f(s)
s
2
2
(0))G
t
(x; s, t)ds +
_
1
0
(g(s)
s
2
2

(0))G(x; s, t)ds
+
_
1
0
_
t
0
_
Q(s, ))
s
2
2

() + ()
_
G(x; s, t )dds
where G(x; s, t ) =
n=1
2
n
cos(nx) cos(ns) sin(n(t ))
317
4c. u(0, t) = A(t) u
x
(1, t) = 0
Let w(x, t) = (1 x)
2
A(t)
Then w(0, t) = A(t) w
x
(1, t) = 2(1 x)A(t)[
x=1
= 0
As before letting v(x, t) = v(x, t) w(x, t)
v
tt
= v
xx
+

Q(x, t)
where

Q(x, t) = Q(x, t) + (1 x)
2
A
+ 2A(t)
v(x, 0) = f(x) (1 x)
2
A(0)

f(x)
v
t
(x, 0) = g(x) (1 x)
2
A
(0) g(x)
v(0, t) = v
x
(1, t) = 0
The homogenous equation has eigenfunctions and eigenvalues
n
(x) = cos
_
n
1
2
_
x
n
=
__
n
1
2
_
_
2
n = 1, 2, . . .
Q(x, t) =
n=1
q
n
(t) cos
__
n
1
2
_
x
_
where q
n
(t) = 2
_
1
0
Q(x, t) cos
__
n
1
2
_
x
_
dx
v(x, t) =
n=1
A
n
(t) cos
__
n
1
2
_
x
_
v(x, 0) =
n=1
A
n
(0) cos
__
n
1
2
_
x
_
=

f(x)
where A
n
(0) =
_
1
0
f(x) cos(n
1
2
)xdx
_
1
0
cos
2
(n
1
2
)xdx
A
n
(0) = 2
_
1
0
f(x) cos
__
n
1
2
_
x
_
dx
v
t
(x, 0) =
n=1
A
n
(0) cos
__
n
1
2
_
x
_
= g(x)
where A
n
(0) = 2
_
1
0
g(x) cos
_
n
1
2
_
xdx
A
n
(t) +
__
n
1
2
_
_
2
A
n
(t) = q
n
(t)
318
The solution is
A
n
(t) = K
1n
cos
__
n
1
2
_
t
_
+ K
2n
sin
__
n
1
2
_
t
_
+ [A
n
(t)]
p
where the particular solution is
[A
n
(t)]
p
=
cos
__
n
1
2
_
t
_
(n
1
2
)
_
t
0
sin
__
n
1
2
_
_
q
n
()d
+
sin
__
n
1
2
_
t
_
(n
1
2
)
_
t
0
cos
__
n
1
2
_
_
q
n
()d
K
1n
= 2
_
1
0
f(x) cos
__
n
1
2
_
x
_
dx
K
2n
=
2
(n
1
2
)
_
1
0
g(x) cos
__
n
1
2
_
x
_
dx
u(x, t) = v(x, t) + w(x, t)
=
n=1
A
n
(t) cos
__
n
1
2
_
x
_
+ (1 x)
2
A(t)
= (1x)
2
A(t)+
n=1
cos
__
n
1
2
_
x
_ _
2
_
1
0
f(s) cos
__
n
1
2
_
s
_
ds cos
_
(n
1
2
)t
_
+
2
(n
1
2
)
_
1
0
g(s) cos
__
n
1
2
_
s
_
ds sin
__
n
1
2
_
t
_
+2
_
1
0
_
t
0
Q(s, ) cos
__
n
1
2
_
s
_
sin
_
(n
1
2
)t
_
(n
1
2
)
cos
__
n
1
2
_
cos
__
n
1
2
_
t
_
(n
1
2
)
sin
__
n
1
2
_
dds
Let G(x; s, t) =
n=1
2
(n
1
2
)
cos
__
n
1
2
_
x
_
cos
__
n
1
2
_
s
_
Then u(x, t) = (1 x)
2
A(t) +
_
1
0
_
f(s) (1 s)
2
A(0)
_
G
t
(x; s, t)ds
+
_
1
0
_
g(s) (1 s)
2
A
(0)
_
G(x; s, t)ds
+
_
1
0
_
t
0
_
Q(s, ) + (1 s)
2
A
() + 2A()
_
G(x; s, t )dds
319
10.3 Greens Function for Sturm-Liouville Problems
Problems
1. Show that Greens function is unique if it exists.
Hint: Show that if there are 2 Greens functions G(x; s) and H(x; s) then
_
1
0
[G(x; s) H(x; s)] f(s)ds = 0.
2. Find Greens function for each
a.
ku
xx
= f(x), 0 < x < L,
u
(0) = 0,
u(L) = 0.
b.
u
xx
= F(x), 0 < x < L,
u
(0) = 0,
u
(L) = 0.
c.
u
xx
= f(x), 0 < x < L,
u(0) u
(0) = 0,
u(L) = 0.
3. Find Greens function for
ky
+ y = 0, 0 < x < 1,
y(0) y
(0) = 0,
y(1) = 0.
4. Find Greens function for the initial value problem
Ly = f(x),
y(0) = y
(0) = 0.
Show that the solution is
y(x) =
_
x
0
G(x; s)f(s)ds.
5. Prove (10.3.22).
320
1. Assume there exist 2 Greens functions G(x, s), H(x, s)
Then
_
1
0
G(x, s)f(s)ds =
_
1
0
H(x, s)f(s)ds for all f.
So
_
1
0
[G(x, s) H(x, s)] f(s)ds = 0 for all f.
G(x, s) H(x, s) = 0 G(x, s) = H(x, s)
G if it exists, is unique.
321
2a. ku
xx
= f(x) 0 < x < L p(x) = k
u
(0) = 0 u(L) = 0
Solve: ku
= 0 u
(0) = 0
kv
= 0 v(L) = 0
Choose c = 1 so kW = 1
u = ax + b u
(0) = a = 0 u = b
v = x + v(L) = 0 = L + = L v = (x L)
W =
b (x L)
0
= b
kb = 1
Let b = 1, then =
1
k
Then u(x) = 1
and v(x) =
1
k
x +
L
k
=
1
k
(x L)
G(x, s) =
u(s) v(x) 0 s x 1
u(x) v(s) 0 x s 1
G(x, s) =
1
k
(x L) 0 s x 1
1
k
(s L) 0 x s 1
322
2b. v
xx
= f(x) 0 < x < L
u
(0) = 0 u
(L) = 0 p(x) = 1
u
(x) = 0 u
(0) = 0 u = a
v
(x) = 0 v
(L) = 0 v =
W =
a
0 0
= 0 = c
So
G(x, s) =
a 0 s x 1
a 0 x s 1
Therefore Greens function is a constant
(For a Newman problem, the solution is not unique).
323
2c. u
xx
= f(x) 0 < x < L
u(0) u
(0) = 0 u(L) = 0
u
= 0 u = ax + b (a(0) + b) a = 0 b = a
u = ax + a
v
= 0 v = x + v(L) = 0 L = = L
v = (x L)
Let c = 1
W =
ax + a (x L)
a
= 1 (ax + a) a(x L) = 1
Let a = 1 x + x + L = 1 or (1 + L) = 1 =
1
1 + L
u = x + 1
v =
1
1 + L
(x L)
G(x, s) =
(s 1)(x L) / (1 + L) 0 s x 1
(x 1)(s L) / (1 + L) 0 x s 1
324
3. ky
+ y = 0 0 < x < 1
y(0) y
(0) = 0 y(1) = 0
ku
+ u = 0
u

k
u = 0
Let
2
= /k then u = ae
x
+ be
x
u(0) u
(0) = 0 ae
x
+ be
x
a
e
x
+ b
e
x
[
x=0
= 0
a + b a + b = 0 a(1 ) + b(1 + ) = 0
b = a
(1 + )
(1 + )
u = a
_
e
x
+
1
+ 1
e
x
_
kv
+ v = 0 v(1) = 0
v
2
v = 0 v = e
x
+ e
x
v(1) = 0 = e
+ e
=
e
= e
2
v =
_
e
x
e
2
e
x
_
W = k =
a
_
e
x
+
( 1)
( + 1)
e
x
_

_
e
x
e
2
e
x
_
a
_
e
x
+
_
2
+
+ 1
_
e
x
_

_
e
x
+ e
2
e
x
_
= a
_
e
x
+
( 1)
( + 1)
e
x
_
_
e
x
+ e
2
e
x
_
a
_
e
x
e
2
e
x
_
_
e
x
e
x
_
+ 1
__
= a
_
e
2
+

2
+ 1
+ e
2
+
_
+ 1
__
= 2a
_
e
2
+

2
+ 1
_
325
= k/2a
_
e
2
+

2
+ 1
_
Let a = 1
G(x; s) =
_
e
x
+
1
+ 1
e
x
_

k
2(e
2
+

2
+1
_
e
s
e
2
e
s
_
0 s x 1
(by symetry) 0 x s 1
326
4. Lets take Ly =
d
dx
_
p
dy
dx
_
+ q y
y(x) =
_
x
0
G(x; s)f(s)ds clearly satises y(0) = 0, since, in this case,
both limits of integration are zero.
y
(x) = G(x; x)f(x) +

_
x
0
G(x; s)
x
f(s)ds
y
(0) = 0 implies G(x; x) = 0

Dierentiate again after multiplying by p(x)
d
dx
_
p
d
dx
_
+ q y =
d
dx
_
p(x)
_
x
0
G(x; s)
x
f(s)ds
_
+q(x)
_
x
0
G(x; s)f(s)ds
Note that p(x) and q(x) can be put inside the integral on s !
=
G(x; s)
x
s=x
p(x)f(x)
+
_
x
0
_

x
_
p(x)
G(x; s)
x
__
f(s)ds
+
_
x
0
q(x)G(x; s) f(s)ds
Thus Ly =
_
x
0
LGf(s)ds +
_
p(x)
G(x
j
s)
x
s=x
_
f(x)
In order to get f(x) on the right hand side we must have
LG = 0 (will anihilate the integral)
and
x
G(x; s)
s=x
=
1
p(x)
we also need
G(x; x) = 0 (seen earlier)
327
Another way to solve problem 4
Let Ly (py
+ qy = f
y(0) = y
(0) = 0
Then Lv = 0 has 2 linearly independent solutions v
1
and v
2
.
Dene w(x) = v
1
(x)
_
x
0
v
2
(s)f(s)ds v
2
(x)
_
x
0
v
1
(s)f(s)ds
Then w
(x) = v
1
(x)
_
x
0
v
2
(s)f(s)ds v
2
(x)
_
x
0
v
1
(s)f(s)ds
+v
1
(x)v
2
(x)f(x) v
2
(x)v
1
(x)f(x)
So: w
(x) = v
1
(x)
_
x
0
v
2
(s)f(s)ds v
2
(x)
_
x
0
v
1
(s)f(s)ds
So (pw
(x))
=
d
dx
[p(x)v
1
(x)]
_
x
0
v
2
(s)f(s)ds
d
dx
[p(x)v
2
(x)]
_
x
0
v
1
(s)f(s)ds
+p(x) [v
1
(x)v
2
(x) v
2
(x)v
1
(x)] f(x)
= q(x)w(x) + p(x) [v
1
(x)v
2
(v) v
2
(v)v
1
(x)]
. .
=c(see problem 5 next)
f(x)
Hence (pw
+ qw = cf
where w(0) = w
(0) = 0
So y =
w
c
is a solution to
(py
+ qy = f
with y(0) = y
(0) = 0
Thus y(x) =
_
x
0
f(s)
v
1
(x)v
2
(s) v
2
(x)v
1
(s)
p(x) [v
1
(x)v
2
(x) v
2
(x)v
1
]
ds
So y(x) =
_
x
0
f(s)G(x; s)ds
where
G(x; s) =
0 x < s
v
1
(x)v
2
(s) v
2
(x)v
1
(s)
p(x) [v
1
(x)v
2
(x) v
2
(x)v
1
(x)]
x > s
328
And we can see that
LG = 0 For x > s
G(x; x) = 0
G
x
(x; x) =
1
p(x)
329
5. Prove p(x)W(x) = c
u, v both satisfythe same ODE i.e.
(pu
+ qu = 0
(pv
+ qv = 0
W(x) = uv
vu
To compute p(u
v v
u) we dierentiate the Wronskian W(x)

dW
dx
= u
+ uv
vu
uv
vu
=
dW
dx
(pu
+ qu = 0 p
pu
+ qu = 0
also p
pv
+ qv = 0
Divide by p
p
p
u
+
qv
p
= 0 OR u
=
p
p
u
+
qu
p
p
p
v
+
qv
p
= 0 OR v
=
p
p
v
+
qv
p
Thus uv
vu
= u(
p
p
v
+
qv
p
) v(
p
p
u
+
qu
p
) =
p
p
[uv
vu
dW
dx
=
p
p
W so W
p = Wp
(Wp)
= W
p + Wp
so using the above (Wp)
= W
p W
p = 0
(Wp)
= 0 and Wp = c
330
10.4 Dirac Delta Function
Problems
1. Derive (10.4.3) from (10.4.2).
2. Show that (10.4.8) satises (10.4.7).
3. Derive (10.4.9)
Hint: use a change of variables = c(x x
i
).
331
1. f(x) =
_
b
a
f(x
i
)(x x
i
) dx (10.4.2)
Let f(x) 1 x(, ). Then f(x
i
) = 1 for all x
i
So f(x) = 1 =
_

1 (x x
i
) dx
332
2. Show (10.4.8) satises (10.4.7)
H(x x
i
) =
_
x
( x
i
)d (10.4.8)
H(x x
i
) =
0 x < x
i
1 x > x
i
(10.4.7)
Let F(, x) be dened by
F(, x) =
1 < x
0 > x
Then
_
x
( x
i
)d =
_

F(, x)( x
i
)d
since on the other interval F 0.
But using (10.4.5), the integral on the right is F(x
i
, x)
Notice that
F(x
i
, x) =
1 x
i
< x
0 x
i
> x
which is the denition of the Heaviside function
H(x x
i
)
Thus we get (10.4.8)
333
3. Prove: [c(x x
i
)] =
1
[ c [
(x x
i
)
Compute the intergral:
_

f(x) [c(x x
0
)] dx =
make a transformation: y = c(x x
0
)
then dx =
1
c
dy
for c > 0 the limits:
_

f
_
y
c
+ x
0
_
(y)
1
c
dy =
1
c
f(x
0
)
for c < 0 the limits:
_

f
_
y
c
+ x
0
_
(y)
1
c
dy =
1
c
f(x
0
)
Combining the two:
_

f(x) [c (x x
0
)] dx =
1
[c[
f(x
0
)
But f(x
0
) =
_

f(x)(x x
0
) dx
therefore
_

f(x) [c(x x
0
)] dx =
1
[c[
_

f(x)(x x
0
) dx
and we have the required relationship for functions.
334
10.5 Nonhomogeneous Boundary Conditions
Problems
1. Consider
u
t
= u
xx
+ Q(x, t), 0 < x < 1, t > 0,
subject to
u(0, t) = u
x
(1, t) = 0,
u(x, 0) = f(x).
a. Solve by the method of eigenfunction expansion.
b. Determine the Greens function.
c. If Q(x, t) = Q(x), independent of t, take the limit as t of part (b) in order to
determine the Greens function for the steady state.
2. Consider
u
xx
+ u = f(x), 0 < x < 1,
u(0) = u(1) = 0.
Determine the Greens function.
3. Give the solution of the following problems in terms of the Greens function
a. u
xx
= f(x), subject to u(0) = A, u
x
(1) = B.
b. u
xx
+ u = f(x), subject to u(0) = A, u(1) = B.
c. u
xx
= f(x), subject to u(0) = A, u
x
(1) + u(1) = 0.
4. Solve
dG
dx
= (x s),
G(0; s) = 0.
Show that G(x; s) is not symmetric.
5. Solve
u
xxxx
= f(x),
u(0) = u(1) = u
x
(0) = u
xx
(1) = 0,
by obtaining Greens function.
335
1. u
t
= u
xx
+ Q(x, t) 0 x 1 t 0
subject to: u(0, t) = u
x
(1, t) = 0 u(x, 0) = f(x)
a.
n
(x) = sin
_
n
1
2
_
x, n = 1, 2, . . .
n
=
__
n
1
2
_
_
2
, n = 1, 2, . . .
Expand: Q(x, t) =
n=1
q
n
(t) sin
_
n
1
2
_
x
The coecients are:
q
n
(t) = 2
_
1
0
Q(x, t) sin
_
n
1
2
_
xdx
Expand: u(x, t) =
n=1
u
n
(t) sin
_
n
1
2
_
x
It was shown in Chapter 8 that the coecients are:
u
n
(t) = e
(n
1
2
)
2
2
t
_
2
_
1
0
f(x) sin
_
n
1
2
_
xdx
_
+
_
t
0
q
n
()e
(n
1
2
)
2
2
(t)
d
So: u(x, t) =
_
1
0
f(s)
_

n=1
2 sin
__
n
1
2
_
s
_
sin
__
n
1
2
_
x
_
e
(n
1
2
)
2
2
t
_
ds
+
_
1
0
_
t
0
Q(s, )
_

n=1
2 sin
__
n
1
2
_
s
_
sin
__
n
1
2
_
x
_
e
(n
1
2
)
2
2
(t)
_
dds
b. G(x; s, t) =
n
n=1
2 sin
__
n
1
2
_
s
_
sin
__
n
1
2
_
x
_
e
(n
1
2
)
2
2
t
336
c. If Q(x, t) = Q(x), nd the steady state solution
lim
t
u(x, t) =
_
1
0
f(s)
_

n=1
2 sin
_
n
1
2
_
s sin
_
n
1
2
_
x
_
ds
_
lim
t
e
(n
1
2
)
2
2
t
_
. .
=0
+
_
1
0
Q(s)
n=1
2 sin
_
n
1
2
_
s sin
_
n
1
2
_
xds lim
t
e
(n
1
2
)
2
2
t
_
t
0
e
(n
1
2
)
2
d
. .
e
(n
1
2
)
2
2
t
1
_
n
1
2
_
2
2
. .
1 e
(n
1
2
)
2
2
t
_
n
1
2
_
2
2
. .
1
_
n
1
2
_
2
2
=
_
1
0
Q(s)
n=1
2 sin
_
n
1
2
_
s
sin
_
n
1
2
_
x
_
n
1
2
_
2
2
ds
Therefore
lim
t
u(x, t) =
_
1
0
Q(s)
n=1
2
_
n
1
2
_
2
2
sin
_
n
1
2
_
s sin
_
n
1
2
_
x
ds
337
2. u
xx
+ u = f(x) 0 < x < 1
subject to u(0) = u(1) = 0
Solve the 2 ODEs to get G:
u
= u
u(0) = 0
u(x) = a sin x + b cos x u(0) = 0 b = 0

v
= v
v(1) = 0
v(x) = sin x + cos x v(1) = 0 sin 1 + cos 1 = 0 = tan1

u = a sin x
v = sin x tan1 cos x
W =
a sin x (sin x tan1 cos x)

a cos x (cos x + tan 1 sinx)
= a
_
sin xcos x + tan 1 sin
2
x sin xcos x + tan 1 cos
2
x
_
W = atan 1
So atan 1 = c let c = tan1 a =
1
u =
1
sin x v = (sin x tan1 cos x)

G(x; s) =
sin s (sinx tan1 cos x) s x

1
sin x(sin s tan1 cos s) x s

Therefore:
G(x; s) =
sin s [sin x tan 1 cos x] 0 s x 1

sin x[sin s tan 1 cos s] 0 x s 1
338
3 a. u
xx
= f(x)
u(0) = A u
x
(1) = B
To get G(x; s),
solve the homogeneous equation with homogeneous boundary conditions
u
xx
= 0 u(0) = 0 u
x
(1) = 0
u
= 0 u(0) = 0 u = ax + b b = 0 u = ax
v
= 0 v
(0) = 0 v = x + v
(1) = = 0 v =
W =
ax
a 0
= +1
..
coecent of uxx
a = 1 a = 1/
Let = 1 then a = 1
So u = x v = 1
and
G(x; s) =
s s x
x x s
u(s) =
_
1
0
G(x; s)f(x) dx G(1; s)B A
dG(x; s)
dx
[
x=0
G(1; s) = s
dG
dx
=
0 s x
1 x s
so
dG
dx
[
x=0
= 1
u(s) =
_
1
0
G(x; s)f(x) dx + A+ Bs
339
where G(x; s) =
s s x
x x s
Check the answer by substituting: u(s) =
_
1
0
G(x; s)f(x) dx + A + Bs
Write this as 2 integrals substituting for G:
u(s) =
_
s
0
xf(x) dx +
_
1
s
sf(x) dx + A+ Bs
u(0) =
_
0
0
xf(x) dx
. .
=0
+
_
1
0
0 f(x) dx + A+ B 0 = +A checks
Recall how to dierentiate an integral whose limits depend on the variable of integration
u
(s) = sf(s) + sf(s)

. .
=0
+
_
1
s
f(x) dx + B
Deerentiate again: u
(s) = f(s) thus the equation checks

u
(1) = B the second boundary condition checks

340
3. b. u
xx
+ u = f(x)
subject to u(0) = A u(1) = B
From problem 2,
G(x; s) =
sin s [sin x tan1 cos x] s x

sin x[sin s tan1 cos s] x s
If x s :
dG(x; s)
dx
= sin s cos x + sin s tan1 sin x
So at x = 1,
dG
dx
[
x=1
= sin s cos 1 + sin s tan1 sin 1
If x s;
dG(x; s)
dx
= cos x[sin s tan 1 cos s]
So at x = 0,
dG
dx
[
x=0
= sin s tan1 cos s
u(s) =
_
1
0
G(x; s)f(x)dx + B[sin s(cos 1 + tan1 sin 1)] A[sin s tan 1 cos s]
341
3. c. u
xx
= f(x)
u(0) = A u
x
(1) + u(1) = 0
The homogeneous equation with homogeneous boundary conditions:
u
xx
= 0 u(0) = 0 u
x
(1) + u(1) = 0
u
= 0 u(0) = 0 u = ax + b; b = 0 u = ax
v
= 0 v
(1) + v(1) = 0 v = x + ; + + = 0
2 + = 0; = 2 v = x 2
W =
ax x 2
a
= 2a = c = 1 a =
1
2
Let a = 1 =
1
2
Therefore
u(x) = x; v(x) =
1
2
(x 2)
G(x; s) =
1
2
s(x 2) s x
1
2
x(s 2) x s
If x s
dG(x, s)
dx
=
1
2
s
dG
dx
x=1
=
1
2
s
dG
dx
x=0
=
s 2
2
u(s) =
_
1
0
G(x; s)f(x) dx uG
x
1
0
+ Gu
x
1
0
uG
x
1
0
= u(1)
s
2
u(0)
. .
A
(s 2)
2
Gu
x
1
0
= G[
x=1
u
x
(1) G[
x=0
. .
=0
u
x
(0) =
s
2
u
x
(1)
u(s) =
_
1
0
G(x; s)f(x) dx
s
2
u(1) + A
(s 2)
2

s
2
u
x
(1)
342
=
_
1
0
G(x; s)f(x) dx + A
(s 2)
2
u(1)
s
2
(u(1) + u
x
(1))
. .
=0
u(s) =
_
1
0
G(x; s)f(x) dx +
1
2
A(s 2)
343
4. Solve
dG
dx
= (x s) subject to G(0; s) = 0
Since
d
dx
(H(x s)) = (x s) G = H(x s)
G(0; s) = H(s)
G(x; s) =
0 x s
1 x s
Therefore G is NOT symmetric.
344
5. Solve u
xxxx
= f(x)
subject to
u(0) = u(1) = u
(0) = u
(1) = 0
Since Lu = u
xxxx
with the stated boundary conditions is self-adjoint, we have
_
1
0
(uLGGLu) dx = 0
for u, G satisfying the boundary conditions.
In particular, if
Lu = f, LG(x; s) = (x s)
then
_
1
0
uLGdx
_
1
0
GLu dx =
_
1
0
u(x s) dx
_
1
0
G(s; x)f(x) dx
= u(s)
_
1
0
G(s; x)f(x) dx.
Therefore u(x) =
_
1
0
G(x; s)f(s) ds.
Hence we seek a solution to LG(x; s) = (x s)
G(0; s) = G(1; s) = G
x
(0; s) = G
xx
(1; s) = 0
Also
_
s
+
s
G
xxxx
(x; s) dx =
_
s
+
s
(x s) dx = 1
G
xxx
s
+
s
= 1.
Furthermore, G
xx
, G
x
, and G are continuous at x = s
Using LG = 0 if x ,= s,
we can dene G(x; s) on [0, s) and (s, 1], and use the above conditions to determine the
parameters as follows:
G(x; s) =
ax
3
+ bx
2
+ cx + d x < s
x
3
+ x
2
+ x + x > s
345
G(0; s) = 0 d = 0
G
x
(0; s) = 0 c = 0
G(1; s) = 0 + + + = 0
G
xx
(1; s) = 0 6 + 2 = 0 = 3
3 + + = 0
= 2
G(x; s) =
ax
3
+ bx
2
x
3
x
2
+ x + 2
G
xx
s
+
s
= 1 6 6a = 1 a =
1
6
G(x; s) =
_

1
6
_
x
3
+ bx
2
x
3
3x
2
+ x + 2
G
xx
[
s
+ = G
xx
[
s
6
_

1
6
_
s + 2b = 6s 6
s + 2b = 6 b =
s
2
3
G(x; s) =
_

1
6
_
x
3
+
_
s
2
3
_
x
2
x
3
3x
2
+ x + 2
G
x
[
s
+ = G
x
[
s
3(
1
6
)s
2
+ (s 6)s = 3s
2
6s +
346
1
2
s
2
+ s
2
6s = 6s +
=
1
2
s
2
G(x; s) =
(
1
6
)x
3
+ (
s
2
3)x
2
x
3
3x
2
+
s
2
2
x + 2
s
2
2
G[
s
+ = G[
s
(
1
6
)s
3
+ (
s
2
3)s
2
= s
3
3s
2
+
s
3
2
+ 2
s
2
2
s
3
s
3
6
+
s
3
2
3s
2
= s
3
3s
2
+
s
3
2
+ 2
s
2
2
2 =
s
2
2

s
3
6
=
s
2
4

s
3
12
G(x; s) =
x
3
_
s
3
12
+
s
2
4

1
6
_
+ x
2
_
s
3
4

3
4
s
2
+
s
2
_
x < s
x
3
_
s
3
12
+
s
2
4
_
+ x
2
_
s
3
4

3
4
s
2
_
+ x
_
s
2
2
_
+
_
s
2
2

s
3
6
_
s
2
2
x > s
G(x, s) =
x
3
_
s
3
12
+
s
2
4

1
6
_
+ x
2
_
s
3
4

3
4
s
2
+
s
2
_
x < s
s
3
_
x
3
12
+
x
2
4

1
6
_
+ s
2
_
x
3
4

3
4
x
2
+
x
2
_
x > s
347
10.6 Fredholm Alternative And Modied Greens Functions
Problems
1. Use Fredholm alternative to nd out if
u
xx
+ u = + x, 0 < x < ,
subject to
u(0) = u() = 0,
has a solution for all or only for certain values of .
2. Without determining u(x), how many solutions are there of
u
xx
+ u = cos x
a. = 1 and u(0) = u() = 0.
b. = 1 and u
x
(0) = u
x
() = 0.
c. = 1 and u(0) = u() = 0.
d. = 2 and u(0) = u() = 0.
3. Are there any values of for which there are solutions of
u
xx
+ u = + x, < x <
u() = u(),
u
x
() = u
x
()?
4. Consider
u
xx
+ u = 1
a. Find the general solution.
b. Obtain the solution satisfying
u(0) = u() = 0.
Is your answer consistent with Fredholm alternative?
c. Obtain the solution satisfying
u
x
(0) = u
x
() = 0.
Is your answer consistent with Fredholm alternative?
5. Obtain the solution for
u
xx
u = e
x
,
348
u(0) = 0, u
(1) = 0.
6. Determine the modied Greens function required for
u
xx
+ u = F(x),
u(0) = A, u() = B.
Assume that F satises the solvability condition. Obtain the solution in terms of the modied
Greens function.
349
1. u
xx
+ u = + x 0 x
u(0) = u() = 0
Homogenous: u
xx
+ u = 0 u(0) = u() = 0
u
H
(x) = sin x ,= 0
So
_

0
( + x) sin xdx = cos x[
0
+ (sin x xcos x)[
0
= (1 1) cos = +2 + = 0 =
2
350
2 a. u
xx
+ u = 0 = 1 u(0) = u() = 0
u
H
= sin x
_

0
sin xcos xdx =
1
2
sin
2
x[
0
= 0
innite number of solutions
2 b. u
x
(0) = u
x
() = 0
Has u
H
= cos x
_

0
cos
2
xdx =

2
,= 0
no solution
2 c. u
xx
= u u(0) = u() = 0
u
H
= e
x
No nontrivial homogenous solution
unique solution
2 d. u
xx
= 2u
u
H
= sin( x)
2
= +2 =
2
u
H
(0) = sin(
20) = 0 but u
H
() = sin(
2) ,= 0 no nontrivial solution
unique solution
351
3. u
xx
+ u = + x x [, ]
u() = u()
u
x
() = u
x
()
v
xx
= v v
H
= cos x + sin x
_

( + x)(cos x + sin x) dx = 0
Note that integrating an odd function on the symmetric interval gives zero. So we have left:
_

cos xdx +
_

xsin xdx = sin x
+ (sin x xcos x)
= (0 0) + (0 0) + (0 0) cos cos() = +2 ,= 0
Regardless of value of < f, cos x >,= 0 no solution.
352
4 a. u
xx
+ u = 1
Homogenous: u
xx
+ u = 0 u(x) = c
1
cos x + c
1
sin x
Get a particular solution of the inhomogeneous
u
p
= A u
p
+ u
p
= 1 0 + A = 1 A = 1
So u
g
= u
H
+ u
p
= c
1
cos x + c
2
sin x + 1
4 b. u(0) = 0 c
1
+ 1 = 0
u() = 0 c
1
+ 1 = 0
No solution for the system No solution.
_

0
(1) sin xdx = cos x[
0
= (1) + (1) = 2 ,= 0
No solution by Fredholm either
4 c. u
x
(0) = u
x
() = 0
u
x
(0) = c
2
cos 0 = 0 u
x
() = c
2
cos = 0
c
2
= 0 innite number of solutions
u = c
1
cos x + 1
_

0
1cos x = sin x[
0
= 0 innite number of solutions by Fredholm.
353
5. This problem is similar to the example (10.6.2)-(10.6.3). In this case A = 1 which
is NOT n
2
for any integer. Thus the nonhomogeneous has a unique solution (Fredholm
alternative). The solution can be found by the method of eigenfunction expansion. Note
that the eigenfunctions are sin
_
n
1
2
_
x, n = 1, 2, . . .. Expandind the right hand side
in terms of the eigenfunctions we have
e
x
=
n=1
n
sin
_
n
1
2
_
x,
with coecients given by
n
=
_
0
e
x
sin
_
n
1
2
_
xdx
_
0
sin
2
_
n
1
2
_
xdx
Using the same expansion for u, we get the coecients u
n
by substitution in the dierential
equation
u
n
=

n
1 +
_
n
1
2
_
2
2
354
6. u
xx
+ u = F(x) u(0) = A u() = B
Solve the homogenous problem rst to nd

G(x, s)
u
xx
+ u = 0 u(0) = 0 = u()
The homogenous problem has u
H
= sin x as a nontrivial solution
(Assume
_

0
F(x) sin xdx = 0)
_

0
sin x[(x s) + c sin x] dx = 0 = sin s + c
_

0
sin
2
xdx
. .
2
c =
sin s
/2
L
G =
d
2

G
dx
2
+

G = (x s) +
2 sin xsin s
G(0; s) =

G(; s) = 0
Assume

G is continuous at x = s
_
s
+
s
d
2

G
dx
2
dx +
_
s
+
s
Gdx =
_
s
+
s
(x s) dx +
_
s
+
s
2 sin xsin s
dx
d

G
dx
s
+
s
+ 0 = 1 + 0 (by continuity)
For x ,= s
d
2

G
dx
2
+

G =
2 sin xsin s
Try variation of parameters

G
+

G =
_
2 sin s
_
sin x
G
H
= c
1
cos x + c
2
sin x
G
p
(x) = u
1
(x) cos x + u
2
(x) sin x
p
(x) = u
1
(x) cos x u
1
(x) sin x + u
2
(x) sin x + u
2
(x) cos x
= [u
1
cos x + u
2
sin x]
. .
set this to zero
+[u
1
sin x + u
2
cos x]
p
(x) = u
1
sin x + u
2
cos x u
1
cos x u
2
sin x
355
p
+

G
p
= u
1
sin x + u
2
cos x =
2 sins
sin x
Therefore u
1
and u
2
satisfy the system
u
1
cos x + u
2
sin x = 0
u
1
sin x + u
2
cos x = sin x
_
2 sin s
_
The Wronskian is:
W = cos
2
x + sin
2
x = 1
The solution is:
u
1
= sin xsin x
_
2 sin s
_
u
1
=
2 sin s
__
sin
2
xdx
_
=
2 sin s
1
2
cos xsin x +
x
2
_
u
1
=
1
sin s sin xcos x

x
sin s
u
2
= cos xsin x
_
2 sin s
_
u
2
=
2 sin s
_
cos xsin xdx =
2 sin s
_
1
2
sin
2
x
_
=
1
sin s sin
2
x
The solution is
G(x; s) = c
1
cos x+c
2
sin x+
1
cos xsin s [cos xsin x x]+

1
sin s sin
2
xsin x
= c
1
cos x + c
2
sin x
x
cos xsin s +
sin s
sin x
_
cos
2
x + sin
2
x
_
= c
1
cos x + c
2
sin x
x
cos xsin s +
sin s
sin x
So
G(x; s) =
sin s sin x
x
cos xsin s + c
1
cos x + c
2
sin x x < s
1
sin xsin s
x
cos xsin s + c
3
cos x + c
4
sin x x > s
From the endpoints: G(0; s) = 0 c
1
= 0
356
G(; s) = 0 c
3
(1) + sin s = 0 c
3
= +sin s
G(x; s) =
sin s sin x
x
cos xsin s + c
2
sin x x < s
1
sin xsin s
x
cos xsin s + sin s cos x + c

4
sin x x > s
Continuity at x = s; c
2
sin s = +sin s cos s + c
4
sin s
c
2
= c
4
+ cos s
G(x; s) =
sin s sin x
x
cos xsin s + cos s sin x + c sin x x < s

1
sin xsin s
x
cos xsin s + cos xsin s + c sin x x > s

Checking jump in derivative
d

G
dx
s
+
s
= 1 cos s cos x + sin xsin s
xs
= sin
2
(s) + cos
2
(s) = 1
Symmetry: Letting c =
s
cos s and reordering terms
G(x; s) =
sin s sin x + cos s sin x

x
cos xsin s
s
cos s sin x x < s

1
sin xsin s + cos xsin s

s
cos s sinx
x
cos xsin s x > s

To obtain a solution using

G(x, s), notice
_

0
_
uL
G

GLu
_
dx =
_

0
_
u(

G
+

G)

G(u
+ u)
_
dx
=
_

0
_
u

G

Gu
+ (u

Gu

G)
_
dx =
_
u

G

Gu
0
= u
0
= B

G
0
357
d

G
dx
[
=
_
1
sin s cos x sin xsin s

s
cos s cos x
1
cos xsin s +
x
sin xsin s
_
[
x=
=
1
sin s +
s
cos s +
1
sin s =
s
cos s
d

G
dx
[
0
=
_
1
sin s cos x + cos s cos x

1
cos xsin s +
x
sin xsin s
s
cos s cos x
_
[
x=0
=
1
sin s + cos s
1
sin s
s
cos s =
_
1
s
_
cos s
Also notice:
_

0
_
uL
G

GLu
_
dx =
_

0
u
_
(x s) +
2
sin s sin x
_
dx
_

0
GF dx
= u(s) +
_

0
u(x)
2 sins sin x
dx
. .
a multiple of the homogenous solution
_

0
GF(x) dx
so disregard to get particular solution
So u(s)
_

0
GF(x) dx = B
d

G
dx
x=
A
d

G
dx
x=0
Interchanging x and s, we get
u(x) =
_

0
G(x; s)F(s)ds +
B
xcos x A
_
1
x
_
cos x
358
10.7 Greens Function For Poissons Equation
Problems
1. Derive Greens function for Poissons equation on innite three dimensional space. What
is the condition at innity required to ensure vanishing contribution from the boundary
integral?
2. Show that Greens function (10.7.37) satises the boundary condition (10.7.35).
3. Use (10.7.39) to obtain the solution of Laplaces equation on the upper half plane subject
to
u(x, 0) = h(x)
4. Use the method of eigenfunction expansion to determine G(r; r
0
) if
2
G = (r r
0
), 0 < x < 1, 0 < y < 1
subject to the following boundary conditions
G(0, y; r
0
) = G
x
(1, y; r
0
) = G
y
(x, 0; r
0
) = G
y
(x, 1; r
0
) = 0
5. Solve the above problem inside a unit cube with zero Dirichlet boundary condition on
all sides.
6. Derive Greens function for Poissons equation on a circle by using the method of images.
7. Use the above Greens function to show that Laplaces equation inside a circle of radius
with
u(r, ) = h() for r =
is given by Poissons formula
u(r, ) =
1
2
_
2
0
h(
0
)

2
r
2
r
2
+
2
2r cos(
0
)
d
0
.
8. Determine Greens function for the right half plane.
9. Determine Greens function for the upper half plane subject to
G
y
= 0 on y = 0.
Use it to solve Poissons equation
2
u = f
u
y
= h(x), on y = 0.
Ignore the contributions at innity.
10. Use the method of images to solve
2
G = (r r
0
)
in the rst quadrant with G = 0 on the boundary.
359
1. Solve
2
u = f(r), r = (x, y, z) 1
3
Greens function G(r; r
0
) satises
2
G = (x x
0
) (y y
0
) (z z
0
)
Because of symmetry we have
1
2
d
d
_
2
dG
d
_
= 0 for ,= 0
where = [x x
0
[
The solution is done by integration
2
dG
d
= C
Divide by
2
and integrate again
G() =
C
+ D
To obtain the constants, we integrate over a sphere of radius containing the point (x
0
, y
0
, z
0
),
thus
_ _ _

2
Gdx dy dz =
_ _ _
(x x
0
) (y y
0
) (z z
0
)dxdy dz = 1.
But by Greens formula, the left hand side is
_ _
S
G ndS =
_ _
S
G
dS =
G
4
2
where S is the surface of the sphere.
Thus
G
4
2
= 1
G
=
1
4
2
Since G =
C
+ D
then
G
=
C
2
. Comparing
G
we have
360
C
2
=
1
4
2
=C =
1
4
We can take D = 0, thus
G() =
1
4
The condition at innity can be obtained by using Greens formula as in the text.
lim
_ _
S
(uGGu) ndS = 0
lim

2
_
u
G
G
u
_
= 0
or by using G from above:
lim
_
u +
u
_
= 0
361
2. Show that
G =
1
2
ln[r r
0
[
1
2
ln[r r
0
[
satises
G(x, 0; x
0
, y
0
) = 0
Recall
r = (x, y)
r
0
= (x
0
, y
0
)
r
0
= (x
0
, y
0
)
G(x, 0; x
0
, y
0
) =
1
2
ln
_
(x x
0
)
2
+ (0 y
0
)
2
1
2
ln
_
(x x
0
)
2
+ (0 + y
0
)
2
Since the terms under square root signs are identical we get zero for G.
362
3. Use
u(r) =
_ _
f(r
0
) G(r; r
0
)dr
0
+
_

h(x
0
)
y/
(x x
0
)
2
+ y
2
dx
0
To obtain the solution of
2
u = 0 y > 0
u(x, 0) = h(x).
Since this is Laplaces equation f(r) 0 and the rst integral is zero, thus
u(r) =
_

h(x
0
)
y
(x x
0
)
2
+ y
2
dx
0
363
4. Use the method of eigenfunction expansion to determine G(r; r
0
) on [0, 1] [0, 1]
Because of the boundary conditions, the eigenfunctions are: cos ny n = 0, 1, . . .
G(r; r
0
) =
n=0
g
n
(x) cos ny
or
G(r; r
0
) =
n=1
g
n
(y) cos
_
n +
1
2
_
x
We take the rst expansion and substitute in the equation
n=0
g
n
(x) cos ny +
n=1
(n
2
2
)g
n
(x) cos ny = 0, r ,= r
0
.
g
n
(x) n
2
2
g
n
(x) = 0 n = 1, 2, . . .
g
0
= 0
Subject to:
g
n
(0) = 0
g
n
(1) = 0
Solve, apply the jump condition for the derivative and so on.
364
5.
2
G = (r r
0
) 0 x, y, z 1
G(0, y, z) = G(1, y, z) = 0
G(x, 0, z) = G(x, 1, z) = 0
G(x, y, 0) = G(x, y, 1) = 0
Because of the boundary conditions, the eigenfunctions are: sin ny sin mz, n = 1, 2, . . . ,
m = 1, 2, . . .
Thus
G(r; r
0
) =
n=1
m=1
g
nm
(x) sin ny sin mz
(other possibilities exist, depending on the two variables we use. We can even take expansion
in all 3 variables !)
Substitute in the equation
n=1
m=1
g
nm
(x) sin ny sin mz+
n=1
m=1
(n
2
2
m
2
2
)g
nm
(x) sin ny sin mz = 0, r ,= r
0
.
g
nm
(x) (n
2
2
+ m
2
2
)g
nm
(x) = 0, n, m = 1, 2, . . .
g
nm
(0) = g
nm
(1) = 0,
The boundary conditions are the result of using the two boundary conditions we didnt use
in getting the eigenfunction, in this case G(0, y, z) = G(1, y, z) = 0.
Solve the boundary value problem to get g
nm
(x)
365
6. Use the method of images to derive Greens function for Poissons equation on a circle.
2
G(r; r
0
) = (r r
0
) [r[ a
G(r; r
0
) = 0 for [r[= a
Let r
0
be the reected image of a point r
0
(see gure). Then if G is the Greens function for
the whole plane, we have
2
G(r; r
0
) = (r r
0
) (r r
0
)
* * *
x
0 x
x
0
*
Figure 61: The case = 0
So
G(r; r
0
) =
1
2
ln[r r
0
[
1
2
ln[r r
0
[+C
=
1
4
ln
[r r
0
[
2
[r r
0
[
2
+ C
To nd C, we note that when [r[= a G(r; r
0
) = 0 and with = 0 all 3 points coincide
=C =
1
4
ln
[a r
0
[
2
[a r
0
[
2
For ,= 0 (see gure) G(r; r
0
) = 0
yields
1
4
_
ln
[r r
0
[
2
[r r
0
[
2
ln
[a r
0
[
2
[a r
0
[
2
_
= 0
x
0
x
x
0
*
Figure 62: The case ,= 0

366
So for all [r[ = a,
[r r
0
[
2
[r r
0
[
2
=
[a r
0
[
2
[a r
0
[
2
Note
[r r
0
[
2
= [r[
2
+[r
0
[
2
2[r[ [r
0
[ cos
[r r
0
[
2
= [r[
2
+[r
0
[
2
2[r[ [r
0
[ cos
Therefore
[r
0
[ =
a
2
[r
0
[
2
Note that when r
0
is the center of the circle that r
0
is at innity.
Hence
G(r; r
0
) =
1
4
ln
_
[r r
0
[
2
a
2
[r r
0
[
2
[r
0
[
2
_
367
7. The solution in general is given by
u(r) =
_ _
f(r
0
)
. .
right hand side of equation=0
G(r; r
0
)dr
0
+
_
h(r
0
)
. .
u on the boundary.
r
0
G(r; r
0
) ndS
The normal to the circle is in the direction of radius.
=
_
h(r
0
)

r
0
G(r; r
0
)[
r
0
=a
dS
Convert G obtained in problem 6 to polar coordinates, dierentiate and substitute in the
integral.
368
8. Let
r = (x, y)
r
0
= (x
0
, y
0
)
Then
r
0
= (x
0
, y
0
)
is the image for the right half plane.
Therefore
G(r; r
0
) =
1
2
_
ln [r r
0
[ ln[r r
0
[
_
in a similar fashion to problem 6.
The solution of Poissons equation in general is
u(r) =
_ _
f(r
0
)G(r; r
0
)dr
0
+
_
u(r
0
)
r
0
G(r; r
0
) ndS
369
9. For the upper half plane
G(r; r
0
) =
1
4
ln
(x x
0
)
2
+ (y y
0
)
2
(x x
0
)
2
+ (y + y
0
)
2
Since r
0
= (x
0
, y
0
)
It is straightforward to check that
G
y
= 0, when y = 0.
The solution is given by
u(r) =
_ _
f(r
0
)G(r; r
0
)dr
0
+
_

G(x, y; x
0
, 0)h(x
0
)dx
0
370
10. 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.
( x
0
, y
0
) ( x
0
**
, y
0
**
) = ( x
0
, y
0
)
( x
0
*
, y
0
*
) = ( x
0
, y
0
) ( x
0
***
, y
0
***
) = ( x
0
, y
0
)
2
G = (r r
0
) (r r
0
) (r r
0
)
+(r r
0
)
G =
1
2
ln
[r r
0
[ [r r
0
[
[r r
0
[ [r r
0
[
=
1
4
ln
[(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
]
It is easy to see that on the axes G = 0.
371

PDE - Most Important

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PDE - Most Important

Uploaded by

Copyright:

Available Formats

PARTIAL DIFFERENTIAL EQUATIONS

z (using the transformation)

, therefore when using the atarred equations

This equation can be solved.

= 2(1)(4)(1) + 5((4)(1) + 1(1)) + 2(4)(1)(1) = 8 25 + 8 = 9

Rewrite this as a system of two rst order ODEs,

Using the initial condition

Now substitute back:

f(x ct) for x ct > 0

dierentiation with respect to argument

F(ct) + G(ct) = h(t)

sin xcos ct x + ct < 0

f(x + ct) + f(x ct)

We now consider three cases depending on the sign of

Thus the solution is a combination of sine and cosine

We now consider three cases depending on the sign of

L = 0, thus the coecient A

Thus the solution is a combination of sine and cosine

We now consider three cases depending on the sign of

Thus the solution is a combination of sine and cosine

We now consider three cases depending on the sign of

Thus the solution is a combination of sine and cosine

We now consider three cases depending on the sign of

Thus the solution is a combination of sine and cosine

L = 0 then we are left with sin

L = 0 which is not possible (the cosine and

L ,= 0 and upon dividing by it we get

f(x) =zero on the rest

This agrees with previous result.

(r, 0) = u(r, /2) = 0, u(1, ) = f().

only positive exponent

Since we are solving < 0 trivial solution

(x) + aX(x) + X(x) = 0, a > 0,

(x) + 2X(x) + X(x) = 0, > 1,

(b) p(b) v(b) u

Figure 59: Domain for problem 2 of 7.4

r) to satisfy [ R(0) [ <

This is Bessels equation with = 0

This is zero for m = 0

(H) = 0 [ R(0) [ <

This solution satises the boundedness at the origin. The eigenvalues

= 0, 0 r < a, 0 < < , 0 < < 2,

= 0, 0 r < a, 0 < < , 0 < < ,

= 0, 0 < < , 0 < < 2

s in the equation, we get (7.7.21) with

Coecients of expansion of initial condition for v

Solution for homogeneous.

(Hint: use dierentiation with respect

(Hint: use integration with respect to a

Use the initial condition

Plug this c in the solution

This can be written as

(t)] ] G(x; s, t )dds

(x) = G(x; x)f(x) +

(0) = 0 implies G(x; x) = 0

u) we dierentiate the Wronskian W(x)

so using the above (Wp)

u(x) = a sin x + b cos x u(0) = 0 b = 0

v(x) = sin x + cos x v(1) = 0 sin 1 + cos 1 = 0 = tan1

a sin x (sin x tan1 cos x)

sin x v = (sin x tan1 cos x)

sin s (sinx tan1 cos x) s x

sin x(sin s tan1 cos s) x s