Mathematical Methods of Physics

Collection of problems and solutions

Maxim Zabzine, Joel Ekstrand
Department of Physics and Astronomy, Uppsala University
2010
1 Vector elds, curved coordinates, operators, integrals
1.1 (Yngve, vektoranalys, 1.8.2) Show that u v is divergence-free if u and v are rotation-
free.
1.2 (Yngve, vektoranalys, 1.8.4) Show that v = r is divergence-free if is a constant
vector eld (r is the position vector).
1.3 (Yngve, vektoranalys, 2.4.7) Use cylindrical coordinates and determine
1. v = r, where v is the velocity, is a constant angle velocity, andr is the position
vector.
2. the rotation of v, i.e. v.
1.4 (Yngve, vektoranalys,1.8.16) The potential from an electric dipol p is given by
(r) =
pr
4
0
r
3
.
Determine

E = at the point r
0
.
1.5 (Exam FFM NV 060322, 1) Show that the vector eld

F = 20x
9
yz x +(2x
10
z +10y
9
) y +(2x
10
y +25z
24
) z
is conserved and there exists a scalar potential. Calculate the line integral
_
C

Fdr for the curve

starting at the point (0, 0, 0) and nishing at (1, 1, 1).
1.6 (Yngve, vektoranalys, 2.4.13) The vector eld

F has the cartesian components
F
x
=
y
x
2
+y
2
, F
y
=
x
x
2
+y
2
, F
y
= 0.
1. Determine the components of

F in cylindrical coordinates.
2. What is the rotation of

F, i.e.

F ?
3. Calculate W =
_

F dr, where the integration is along the unit circle.
1
4. Is the results of 2 and 3 in agreement with Stokes theorem?
1.7 (Yngve, vektoranalys, 1.10.1) Calculate the contour integral
_
(4,4)
(1,1)

F dr where the co-

ordinates are cartesian (x, y)-coordinates.

F has the cartesian components
F
x
=kx, F
y
=ky, F
y
= 0,
where k is a constant. Calculate the integral along different contours:
1. Along (1, 1) (4, 1) (4, 4) in the xy-plane.
2. Along x = y.
1.8 (Homework 1 FFM NV 2006, 2) Calculate
_
mr
r
3
_
, where m is constant vector.
1.9 Use Gauss law to show that
_
S
u d

S = 0 if Laplace equation is fullled.

1.10 (Homework 1 FFM QX 2008, 1) Calculate

_
mr
r
7
+3r
_
,
where m is constant vector andr = x x +y y +z z.
1.11 (Homework 1 FFM QX 2008, 2) The vector eld

F is dened by

F = (yz +2xz
2
) x +xz y +(xy +2x
2
z) z.
Calculate the line integral
_
C

F dr where C is any curve you wish starting at point (0, 0, 0) and

nishes at point (1, 2, 1).
1.12 Calculate
_

B d

S, where

B =

A.
1.13 (Homework 2 FFM NV 2006, 1) The vector eld is dened by

F = 2xz x +2yz
2
y +(x
2
+2y
2
z 1) z.
Show that for any curve C the integral
_
C

F dr = 0
1.14 (Homework 2 FFM NV 2006, 2) Evaluate the surface integral
_
S
r d

S over the sphere

x
2
+y
2
+z
2
= R
2
.
1.15 (Exam 011012, 1) Calculate the contour integral
_
C
r dr, where C is the contour de-
scribed byr() = 2a(1cos)( xcos + ysin), a is a constant and [0, 2]
1.16 Show that
_
C
r dr = 2A z, where A is the area bounded by the curve C, and C is in the
xy-plane.
2
1.17 (Exam FMM QX 061023, 1) Consider the vector eld

F =
r
r
3
+5r,
where
r = x x +y y +z z, r =|r|.
Show that

F is conserved and there exists a scalar potential. Calculate the line integral
_
C

Fdr
for the curve starting at the point with the cartesian coordinate (1, 0, 0) and nishing at the point
with the cartesian coordinate (1, 1, 1).
2 Heat and wave equation. One dimensional problems.
2.1 (Homework 3 FFM NV 2006, 1) Determine the steady-state concentration distribution
outside the sphere of radius R, if the concentration on the sphere is c
0
.
Hint: use the diffusion equation and think about the correct boundary condition at innity.
2.2 (Exam FFM NV 060322, 2) Assume that the temperature is spherically symmetric, u(r, t).
Consider the heat ow (without sources) between two concentric spheres of radii R
1
and R
2
,
R
2
> R
1
. Determine the steady-state temperature distribution, if the temperature of outer sphere
is u
2
and the inner sphere is u
1
. Assuming the constant specic heat c and the constant mass
density , calculate the total heat energy for this state-state solution.
2.3 (Homework 2 FFM QX 2008) Consider the diffusion of a chemical pollutant described
by the diffusion equation without sources:
1. Determine the steady-state concentration distribution between two concentric spheres
with radii R
1
and R
2
, R
1
< R
2
. The concentration on the outer sphere is c
2
and on the
inner sphere is c
1
.
2. Calculate the total mass of the pollutant between two spheres for this steady-state con-
centration distribution.
3. Calculate the total ow of pollutant through the outer sphere.
2.4 (Homework 3 FFM NV 2006, 2) Consider a steady-state temperature distribution in a
volume V with the boundary given by surface S. Please calculate the total heat ow through
surface S and give possible physical explanation for the answer.
2.5 (Exam FFM QX 061023, 2) Assume that the temperature is spherically symmetric, u(r, t).
Consider the heat ow (without sources) between two concentric spheres of radii R
1
and R
2
,
R
2
> R
1
. Determine the steady-state temperature distribution, if the temperature of outer sphere
is u
2
and the inner sphere is perfectly insulated. Assuming the constant specic heat c and the
constant mass density , calculate the total heat energy for this steady-state solution.
3
3 Two and three dimensional problems. Separation of variables.
3.1 (Homework 3 FFM QX 2007) Solve the following PDE
u
t
=

2
u
x
2
+2
u
x
+u ,
with the boundary conditions u(0, t) = 0 , u(L, t) = 0 . The initial condition is u(x, 0) = f (x).
3.2 (Exam FFM NV 060322, 3) Consider the vibration u(r, , t) of a circular membrane of
radius a which is described by the two-dimensional wave equation

2
u
t
2
= c
2
u
Find a solution which satises u(a, , t) = 0, u(r, , 0) = 0,
u
t
(r, , 0) = f (r)sin3.
3.3 (Exam FFM NV 060322, 4) Solve the Laplace equation insider the sphere of the radius
a with the following boundary condition
u(a, , ) = f ()cos
Hint: use the following conventions for the Laplace operator in the spherical coordinates:
u =
1
r
2

r
_
r
2
u
r
_
+
1
r
2
sin

_
sin
u

_
+
1
r
2
sin
2

2
u

2
.
3.4 (Exam FFM NV 060322, 5) On the interval [0, ] solve the heat equation with the source
u
t
=

2
u
x
2
+10sin(6x)e
t
with the boundary conditions
u(0, t) = 1, u(, t) = 3
and the initial condition
u(x, 0) = 1+
2

x +6sin6x +3sinx
3.5 (Exam FFM QX 070109, 4) Consider the two-dimensional heat equation on the disk of
radius b
u
t
= 2u
Find a solution u(, , t) which satises
u(b, , t) = 0, u(, , 0) = f ()sin3.
4
3.6 (Exam FFM QX 070109, 5) Solve the Laplace equation outside of sphere of the radius
a with the following boundary condition
u(a, , ) = P
1
3
(cos)sin
3.7 (Exam FFM QX 070109, 3) Using the separation of variables write the general solution
for the PDE
u
t
=

2
u
x
2
+5u
where u(x, t) satises the following boundary conditions
u(0, t) = 0, u(1, t) = 0
3.8 (Homework 4 FFM NV 2006) Solve the PDE
u
t
= k

2
u
x
2
u, where k and are con-
stants. The boundary conditions are u(0, t) =0, u(L, t) =0 and the initial condition is u(x, 0) =
f (x).
What happens with the solution when t ? [Hint: the limit depends from k and .]
3.9 (Homework 5 FFM NV 2006) Solve the heat equation
u
t
= ku
inside a cylinder (of radius a and height H) subject to the initial condition,
u(r, , z, 0) = f (r, z)
independent of , if the boundary conditions are
u(r, , 0, t) = 0
u(r, , H, t) = 0
u
r
(a, , z, t) = 0
[Hint: try to use the independence of f from as soon as possible, this will simplify some of
the calculations]
3.10 (Homework 4 FFM QX 2007) Consider the two-dimensional heat equation on the disk
of radius b
u
t
= 4u .
Find a solution u(, , t) which satises
u(b, , t) = 0 ,
u(, , 0) = f ()sin(7) +g()sin(3) .
5
3.11 (Homework 5 FFM QX 2007) Solve the Laplace equation
u = 0
between two concentric spheres of the radius a and b, a < b. The boundary conditions are
u(a, , ) = f ()cos ,
u(b, , ) = g()cos .
3.12 (Homework 6 FFM NV 2006) Solve the wave equation

2
u
t
2
= c
2
u
inside a sphere of radius a subject to boundary conditions u(a, , , t) = 0 and the initial condi-
tions u(r, , , 0) = F(r)sin2,
u
t
(r, , , 0) = 0.
3.13 (Homework 3 FFM QX 2008) Solve the following PDE
u
t
= k

2
u
x
2
u,
where k and are constants. The boundary conditions are
u
x
(0, t) = 0, u(L, t) = 0
and the initial condition is
u(x, 0) = f (x).
What happens with the solution when t ? [Hint: the limit depends from k and .]
3.14 (Homework 4 FFM QX 2008) Consider the two-dimensional heat equation on the disk
of radius 1
u
t
=u .
Find a solution u(, , t) which satises the following boundary and initial conditions
u(1, , t) = 0 , u(, , 0) = f ()sin2 +g()sin .
3.15 (Homework 5 FFM QX 2008) Solve the wave equation

2
u
t
2
= c
2
u
outside of a sphere of a radius a subject to the following boundary conditions
u(a, , , t) = 0 ,
and the initial conditions
u(r, , , 0) = 0 ,
u
t
(r, , , 0) = f (r)sin2 .
6
4 Inhomogeneous problems.
4.1 (Homework 7 FFM NV 2008,1) Solve the heat equation with the source
u
t
=

2
u
x
2
+sin7xe
3t
on the interval [0, ] subject to
u(0, t) = 1, u(, t) = 0, u(x, 0) = 0
7
Solutions
S 1.1 Use (a

b) =

b (a) a (

b)
S 1.2 We are free to choose a coordinate system so =
z
z
(
z
z r) =
z
[ z (x x +y y +z z)] =
z
(y x +x y) =

z
(
y
x
x +
x
y
y) = 0.
S 1.5 The vector eld has a potential f if the following is satised

F =

f .
In components it can be written as the following set of PDEs
f
x
= 20x
9
yz
f
y
= 2x
10
z +10y
9
f
z
= 2x
10
y +25z
24
These equations have the solution
f (x, y, z) = 2x
10
yz +y
10
+z
25
+constant
The line integral is calculates as
_
C

Fdr =
_
C
d f = f (1, 1, 1) f (0, 0, 0) = 4
S 1.8 The answer is (
mr
r
3
) = 3
mr
r
5
r
m
r
3
.
S 1.10 The answer is

_
mr
r
7
+3r
_
= 9 .
S 1.11 Let us show that

F is conserved vector eld, namely

F =

f =
f
x
x +
f
y
y +
f
z
z.
8
This implies the following set of PDEs
f
x
= yz +2xz
2
f
y
= xzy
f
z
= xy +2x
2
z
which have the following solution
f (x, y, z) = xyz +x
2
z
2
+constant
Since the vector eld is conserved the line integral is independent from the concrete curve C
and it is given by the following expression
_
C

F dr =
_
C

f dr =
_
C
d f = f (1, 2, 1) f (0, 0, 0) = 3 .
S 1.13 Let us show that

F is conserved vector eld, namely

F =

f =
f
x
x +
f
y
y +
f
z
z.
This implies the following set of PDEs
f
x
= 2xz
f
y
= 2yz
2
f
z
= x
2
+2y
2
z 1
which have the following solution
f (x, y, z) = x
2
z +y
2
z
2
z +constant
Using the Stockes theorem we have
_
C

f dr =
_
S

f ) ds, where we have used that

f ) = 0.
S 1.14 Using the Gauss theorem we get
_
S
r ds =
_
V
(

r)dV = 3
_
V
dV
9
Next we have to calculate the volume of the ball of radius R (x
2
+y
2
+z
2
R
2
). It is easy to do
in the spherical coordinates
_
V
dV =
R
_
0

_
0
2
_
0
drddr
2
sin =
4
3
R
3
,
where we use the following formula
dV = h
1
h
2
h
3
du
1
du
2
du
3
= r
2
sindrdd.
The nal answer is
_
S
r ds = 4R
3
S 1.17 The vector eld has a potential f if the following is satised

F =

f .
In components it can be written as the following set of PDEs
f
x
=
x
(x
2
+y
2
+z
2
)
3/2
+5x
f
y
=
y
(x
2
+y
2
+z
2
)
3/2
+5y
f
z
=
z
(x
2
+y
2
+z
2
)
3/2
+5z
These equations have the solution
f (x, y, z) =
1
_
x
2
+y
2
+z
2
+
5
2
(x
2
+y
2
+z
2
) +constant
Alternatively one can nd the function f in spherical coordinates. In spherical coordninates

F
is written as

F =
_
r
2
+5r
_
r
and the gradient of f in spherical coordinates has a form

f =
f
r
r +
1
r
f

+
1
r sin
f

.
Thus we arrive to the following equation
f
r
= r
2
+5r
and thus f is
f (r) =
1
r
+
5
2
r
2
,
10
which is the same as above when we rewrite it in the cartesian coordinates.
The line integral is calculates as
_
C

Fdr =
_
C
d f = f (1, 1, 1) f (1, 0, 0) = 6
1

3
S 2.1 The diffusion equation is
c
t
= kc,
where c(r, t) is concentration and k is the diffusion coefcient. We have to look for a steady-
state solution,
c
t
= 0 and thus c satises the Laplace equation,
c = 0.
Moreover the problem has spherical symmetry and therefore c depends only on the radial coor-
dinate . Using the form of in spherical coordinates
c =
1

2
c

_
+
1

2
sin

_
sin
c

_
+
1

2
sin
2

2
c

2
we arrive at the following equation
1

2
d
d
_

2
dc
d
_
= 0.
This equation has the following solution
c() = B
A

,
where B and A are constants which can be xed by imposing the correct boundary conditions.
We look for the solution outside the sphere, R . On the sphere we have
c(R) = c
0
.
At innite distance from the sphere we would expect that there will be no any chemical and
thus
, c() 0.
We can conclude that the solution of the problem is
c() =
c
0
R

.
11
S 2.2 To nd a spherically symmetric steady-state we have to solve
1
r
2
d
dr
(r
2
du
dr
) = 0,
which is the Laplace equation for u(r) written in the spherical coordinates. The equation has a
solution
u(r) =
A
r
+B,
where A and B are constants. The constants can be xed by imposing the boundary conditions
u(R
1
) = u
1
, u(R
2
) = u
2
.
Thus we get
A =
(u
2
u
1
)R
1
R
2
R
2
R
1
, B =
u
2
R
2
u
1
R
1
R
2
R
1
and
u(r) =
(u
1
u
2
)R
1
R
2
(R
2
R
1
)r
+
u
2
R
2
u
1
R
1
R
2
R
1
The total heat energy is
c
_
V
udV = c
R
2
_
R
1
dr

_
0
d
2
_
0
d sin r
2
(
A
r
+B) = 2cA(R
2
1
R
2
2
) +
4
3
cB(R
3
2
R
3
1
)
S 2.3 1. The diffusion equation is
c
t
= kc,
where c(r, t) is concentration and k is the diffusion coefcient. We have to look for a steady-
state solution,
c
t
= 0 and thus c satises the Laplace equation,
c = 0.
Moreover the problem has spherical symmetry and therefore c depends only on the radial coor-
dinate r. Using the form of in spherical coordinates
c =
1
r
2

r
_
r
2
c
r
_
+
1
r
2
sin

_
sin
c

_
+
1
r
2
sin
2

2
c

2
we arrive at the following equation
1
r
2
d
dr
_
r
2
dc
dr
_
= 0.
This equation has the following solution
c(r) = B
A
r
,
12
where B and A are constants which can be xed by imposing the correct boundary conditions.
We look for the solution between two concentric spheres, R
1
r R
2
. On the sphere we have
c(R
1
) = c
1
, c(R
2
) = c
2
.
We can conclude that the solution of the problem is
c(r) =
(c
2
c
1
)R
1
R
2
(R
1
R
2
)r
+
c
2
R
2
c
1
R
1
R
2
R
1
.
A =
(c
1
c
2
)R
1
R
2
R
1
R
2
, B =
c
2
R
2
c
1
R
1
R
2
R
1
.
2. The mass of pollutant is given by the volume integral over the space between two concentric
spheres
_
V
c dV =
R
2
_
R
1
dr

_
0
d
2
_
0
d r
2
sin
_
B
A
r
_
= 4
B
3
(R
3
2
R
3
1
) 4
A
2
(R
2
2
R
2
1
) .
3. The ux vector is proportional to the gradient of the concentration c

J =K
0

c .
In the spherical coordinates we have

J =K
0
A
r
2
r .
The total heat ow through the outer sphere S is given by surface integral
_
S

J ds =K
0

_
0
d
2
_
0
d
A
R
2
2
R
2
2
sin =4AK
0
.
S 2.4 The steady-state temperature distribution satises the Laplace equation
u = 0.
The heat ow vector is proportional to the gradient of the temperature

=K
0

u.
The total heat ow through the surface S is given by surface integral
_
S

ds =K
0
_
S
u ds =K
0
_
V

2
u dV = 0,
where we have applied the Gauss theorem and used the fact that u satises the Laplace equation
(remember that =
2
=

).
Thus for the steady state temperature distribution the total ow through any closed surface is
zero. This is natural since the total heat energy in V does not change in time (there are no
sources).
13
S 2.5 To nd a spherically symmetric steady-state we have to solve
1
r
2
d
dr
(r
2
du
dr
) = 0,
which is the Laplace equation for u(r) written in the spherical coordinates. The equation has a
solution
u(r) =
A
r
+B,
where A and B are constants. The constants can be xed by imposing the boundary conditions
du
dr
(R
1
) = 0, u(R
2
) = u
2
,
where the rst condition corresponds to the perfectly insulated boundary. Thus we get
B = u
2
and
u(r) = u
2
The total heat energy is
c
_
V
udV = c
R
2
_
R
1
dr

_
0
d
2
_
0
d sin r
2
u
2
=
4
3
cu
2
(R
3
2
R
3
1
)
S 3.1 The equation should be solved by the method of separation of variables. Let us look for
the solutions of the form
u(x, t) = (x)G(t) .
Plugging this into the equation we get

dG
dt
= G
_
d
2

dx
2
+2
d
dx
+
_
,
and
1
G
dG
dt
=
1

_
d
2

dx
2
+2
d
dx
+
_
= .
Thus we get two ODEs
dG
dt
+G = 0 ,
d
2

dx
2
+2
d
dx
+ = ,
which should be supplemented by the boundary conditions
(0) = 0 , (L) = 0 .
The equation for G is solved very easily
G(t) = Ae
t
.
14
For we can look for the solutions of the form = e
x
, where
( +1)
2
=
and we get
=i

1 .
If > 0 then the solution can be written as
(x) = e
x
_
Acos(

x) +Bsin(

x)
_
.
Imposing the boundary conditions we get
(x) = Ae
x
sin
_
nx
L
_
, =
_
n
L
_
2
, n = 1, 2, 3, ...
There will be no non-trivial solutions for 0 (see the lectures notes for the derivation).
The general solution of the problem will be
u(x, t) =

n=1
A
n
e
(
n
L
)
2
t
e
x
sin
_
nx
L
_
.
Using the orthogonality of sins we get
A
n
=
2
L
L
_
0
dx f (x)e
x
sin
_
nx
L
_
.
S 3.2 We have to solve the following equation

2
u
t
2
= c
2
_
1
r

r
_
r
u
r
_
+
1
r
2

2
u

2
_
Assuming that
u(t, r, ) = G(t)F(r)g()
we get the following set of ODEs
d
2
g
d
2
=g, g( +2) = g()
r
2
d
2
F
dr
2
+r
dF
dr
+(r
2
)F = 0, F(a) = 0, |F(0)| <
d
2
G
dt
2
=c
2
G
The rst equation has a solution
g() = Acos(m) +Bsin(m), = m
2
, m = 0, 1, 2, ...
15
and the second
F(r) =CJ
m
(
_

n
r),
n
=
_
z
mn
a
_
2
, J
m
(z
mn
) = 0, n = 1, 2, ...
The solution for G is
G(t) = Dcos(c
_

n
t) +E sin(c
_

n
t),
where we can drop cos in order to satisfy the rst initial condition. Thus we get
u(t, r, ) =

m=0

n=1
[B
mn
sin(m) +A
mn
cos(m)] sin(c
_

n
t)J
m
(
_

n
r)
Imposing the remaining initial condition we arrive at the solution of the problem
u(t, r, ) =

n=1
B
n
sin(c
_

n
t)sin(3)J
3
(
_

n
r)
with
B
n
=
a
_
0
dr f (r)rJ
3
(

n
)
c

n
a
_
0
drrJ
2
3
(

n
r)
S 3.3 We have to solve
1
r
2

r
_
r
2
u
r
_
+
1
r
2
sin

_
sin
u

_
+
1
r
2
sin
2

2
u

2
= 0
The problem is solved by the separation of the variables
u(r, , ) = f (r)h()g()
We get the following set of ODEs
d
2
g
d
2
=g, g( +2) = g()
d
d
_
sin
dh
d
_
+
_
sin
m
2
sin
_
h = 0
r
2
d
2
f
dr
2
+2r
d f
dr
f = 0
The rst equation has a solution
g() = Asin(m) +Bcos(m), = m
2
, m = 0, 1, 2, ...
For the second equation in order to have a nite solution everywhere we have to require =
l(l +1) with l m
h() =CP
m
l
(cos)
16
For f we have
f (r) = Dr
l
+Er
l1
and we keep only the solution nite at the origin. Thus the solution of Laplace equation inside
a sphere is
u(r, , ) =

m=0

l=m
r
l
[A
ml
cos(m) +B
ml
sin(m)]P
m
l
(cos)
Imposing the boundary conditions we arrive at the nal solution of the problem
u(r, , ) =

l=1
a
l
A
l
cosP
1
l
(cos)
with
A
l
=

_
0
f ()P
1
l
(cos)sin d
a
l

_
0
(P
1
l
(cos))
2
sin d
S 3.4 First we would like to switch to the homogeneous BC
u(x, t) = 1+
2

x +v(x, t)
where v(0, t) = 0, v(, t) = 0 and v satises the same equation as u. We can look for v in the
form
v(x, t) =

n=1
a
n
(t)sin(nx)
Substituting this into PDE we get
da
n
dt
=n
2
a
n
, n = 6
da
6
dt
=36a
6
+10e
t
with the initial conditions
a
6
(0) = 6, a
1
(0) = 3, a
n
(0) = 0, n = 1, 6
So the nal solution is
u(x, t) = 1+
2

x +3e
t
sinx +
_
(6
10
35
)e
36t
+
10
35
e
t
_
sin6x
S 3.5 We have to solve the following equation
u
t
= 2
_
1

_
+
1

2
u

2
_
17
Assuming that
u(, , t) = G(t)F()g()
we get the following set of ODEs
d
2
g
d
2
=g, g( +2) = g()

2
d
2
F
d
2
+
dF
d
+(
2
)F = 0, F(b) = 0, |F(0)| <
dG
dt
=2G
The rst equation has a solution
g() = Acos(m) +Bsin(m), = m
2
, m = 0, 1, 2, ...
and the second
F() =CJ
m
(

), =
_
z
mn
b
_
2
, J
m
(z
mn
) = 0, n = 1, 2, ...
The solution for G is
G(t) = De
2t
.
Combing all we get the following solution of heat equation together with boundary conditions
u(, , t) =

m=0

n=1
[B
mn
sin(m) +A
mn
cos(m)]e
(
zmn
b
)
2
2t
J
m
(
z
mn
b
)
Imposing the initial condition we arrive at the solution of the problem
u(, , t) =

n=1
B
n
e
(
z
3n
b
)
2
2t
sin(3)J
3
(
z
3n
b
)
with
B
n
=
b
_
0
d f ()J
3
(
z
3n
b
)
b
_
0
d (J
3
(
z
3n
b
))
2
S 3.6 We have to solve
1
r
2

r
_
r
2
u
r
_
+
1
r
2
sin

_
sin
u

_
+
1
r
2
sin
2

2
u

2
= 0
The problem is solved by the separation of the variables
u(r, , ) = f (r)h()g()
18
We get the following set of ODEs
d
2
g
d
2
=g, g( +2) = g()
d
d
_
sin
dh
d
_
+
_
sin
m
2
sin
_
h = 0
r
2
d
2
f
dr
2
+2r
d f
dr
f = 0
The rst equation has a solution
g() = Asin(m) +Bcos(m), = m
2
, m = 0, 1, 2, ...
For the second equation in order to have a nite solution everywhere we have to require =
l(l +1) with l m
h() =CP
m
l
(cos)
For f we have
f (r) = Dr
l
+Er
l1
and we keep only the solution nite at the innity. Thus the solution of Laplace equation inside
a sphere is
u(r, , ) =

m=0

l=m
r
l1
[A
ml
cos(m) +B
ml
sin(m)]P
m
l
(cos)
Imposing the boundary conditions and using the orthogonality of associated Legendre functions
(together with the orthogonality of sin and cos) we arrive at the nal solution of the problem
u(r, , ) =
_
a
r
_
4
P
1
3
(cos)sin
S 3.7 We have to solve
u
t
=

2
u
x
2
+5u.
Assuming that u(x, t) = G(t)(x) we get

dG
dt
= G
_
d
2

x
2
+5
_
which implies the following ODEs
dG
t
=G
d
2

dx
2
+( +5) = 0
with the last equation supplemented by the boundary conditions
(0) = 0, (1) = 0.
19
The equation for is solved by
(x) = Asin(nx), = (n)
2
5, n = 1, 2, ...
It is important to notice that > 0 for all n = 1, 2, .. since for
2
> 5. The equation for G is
solved by
G(t) = Ae
t
Combing all together we get the following solution
u(x, t) =

n=1
A
n
e
(5(n)
2
)t
sin(nx)
S 3.8 We have a linear homogeneous PDE with the linear homogeneous boundary conditions
and thus we can solve the problem by the separation of variables.
Let us solve PDE
u
t
= k

2
u
x
2
u
assuming the following form of the solution
u(x, t) = h(t)(x).
PDE reduces to the following equation
1
h
dh
dt
= k
1

d
2

dx
2

Using the standard argument for the separation of variables we get
d
2

dx
2
=
with being a constant and
dh
dt
=(k +)h.
The last equation can be easily solved
h(t) = Ae
(k+)t
.
For we have to specify the boundary conditions
(0) = 0, (L) = 0.
The system for has been considered many times during the lectures and it has the following
solution
(x) = Asin
_
nx
L
_
, =
_
n
L
_
2
, n = 1, 2, 3, ...
The solution of the original PDE can be written as
u(x, t) =

n=1
A
n
e
((
n
L
)
2
k+)t
sin
_
nx
L
_
20
Finally we have to impose the initial conditions
u(x, 0) =

n=1
A
n
sin
_
nx
L
_
= f (x)
Using the orthogonality of sines we get the following expressions for As
A
n
=
2
L
L
_
0
f (x)sin
_
nx
L
_
dx
Now about the limit t . If k + > 0 then
u(x, t) 0
Otherwise u(x, t) will go to innity.
S 3.9 We have a linear homogeneous PDE with the linear homogeneous boundary conditions.
Therefore we can apply the method of separation of variables. Assume our solution of the form
u(r, , z, t) = G(t)(r, , z)
then we arrive to the followin problems
dG
dt
=kG
+ = 0, (r, , 0) = 0, (r, , H) = 0,

r
(a, , z) = 0.
First equation can be easily solved by
G(t) = Ae
kt
and second problem should be solved by the separation of variables. Assuming that
(r, , z) = f (r)g()h(z)
we arrive to the following systems of equations
d
2
h
dz
2
=h, h(0) = 0, h(H) = 0
d
2
g
d
2
=g, g( +2) = g(), ,
dg
d
( +2) =
dg
d
()
r
2
d
2
f
dr
2
+r
d f
dr
+
_
( )r
2

_
f = 0, | f (0)| <,
d f
dr
(a) = 0
First two equations can be solved easily as follows
h(z) = Asin
_
nz
H
_
, =
_
n
H
_
2
, n = 1, 2, 3, ...
21
g() = Asin(m) +Bcos(m), = m
2
, m = 0, 1, 2, ....
However we can observe at this stage that the initial conditions do not have any dependence on
and therefore the solution is independent from . Thus we can conclude that = 0. Thus the
equation for f becomes
r
2
d
2
f
dr
2
+r
d f
dr
+

r
2
f = 0, | f (0)| <,
d f
dr
(a) = 0
where =

+
_
n
H
_
2
. Assuming that

> 0 the solution (which is nite at r = 0) is

f (r) =CJ
0
(

r).
Next we have to impose the boundary conditions at r = a. We use the following property of
Bessel functions (see the lecture notes)
J
1
(x) =
dJ
0
dx
.
Thus we get that
J
1
(

a) = 0,

l
=
_
z
1l
a
_
2
, l = 1, 2, 3, ...
where z
1l
are zeros of J
1
. For

0 no solution exists due to the properties of modied Bessel

functions (no zeros).
Thus the solution of the PDE with the boundary conditions is
u(r, , z, t) =

n=1

l=1
A
nl
e

nl
kt
sin
_
nz
H
_
J
0
(
_

l
r),
where

nl
=

l
+
_
n
H
_
2
Finally we have to impose the initial conditions
u(r, , z, 0) =

n=1

l=1
A
nl
sin
_
nz
H
_
J
0
(
_

l
r) = f (r, z)
Using the orthogonality of sines and J
0
we get
A
nl
=
2
H
_
0
dz
a
_
0
dr r f (r, z)sin
_
nz
H
_
J
0
(
_

l
r)
H
H
_
0
dr rJ
2
0
(
_

l
r)
S 3.10 (Some details of the solution which have been presented on the lectures is not written
out). We have to solve the heat equation in cylindrical coordinates ignoring the dependence
from z. We solve the following equation
u
t
= 4
_
1

_
+
1

2
u

2
_
22
Assuming that
u(, , t) = G(t)F()g()
we get the following set of ODEs
d
2
g
d
2
=g, g( +2) = g()

2
d
2
F
d
2
+
dF
d
+(
2
)F = 0, F(b) = 0, |F(0)| <
dG
dt
=4G
The rst equation has a solution
g() = Acos(m) +Bsin(m), = m
2
, m = 0, 1, 2, ...
and the second
F() =CJ
m
(
_

mn
),
mn
=
_
z
mn
a
_
2
, J
m
(z
mn
) = 0, n = 1, 2, ... ,
where we dropped the Bessel function of second kind since it goes to innity at = 0. The
solution for G is
G(t) = De
4t
.
Thus we get
u(, , t) =

m=0

n=1
[B
mn
sin(m) +A
mn
cos(m)]e
4
mn
t
J
m
(
_

mn
)
Imposing the remaining initial condition we arrive at the solution of the problem
u(t, r, ) =

n=1
_
B
7n
e
4
7n
t
sin(7)J
7
(
_

7n
) +B
3n
e
4
3n
t
sin(3)J
3
(
_

3n
)
_
with
B
7n
=
a
_
0
d f ()J
7
(

7n
)
a
_
0
d (J
7
(

7n
))
2
and
B
3n
=
a
_
0
d g()J
3
(

3n
)
a
_
0
d (J
3
(

3n
))
2
23
S 3.11 We have to solve
1
r
2

r
_
r
2
u
r
_
+
1
r
2
sin

_
sin
u

_
+
1
r
2
sin
2

2
u

2
= 0
The problem is solved by the separation of the variables
u(r, , ) = f (r)h()g()
We get the following set of ODEs
d
2
g
d
2
=g, g( +2) = g()
d
d
_
sin
dh
d
_
+
_
sin
m
2
sin
_
h = 0
r
2
d
2
f
dr
2
+2r
d f
dr
f = 0
The rst equation has a solution
g() = Asin(m) +Bcos(m), = m
2
, m = 0, 1, 2, ...
For the second equation in order to have a nite solution everywhere we have to require =
l(l +1) with l being a positive integer such that l m
h() =CP
m
l
(cos)
For f we have
f (r) = Dr
l
+Er
l1
and we keep both the solutions, since we solve the problem between two spheres. Thus the
solution of Laplace equation between two spheres is
u(r, , ) =

m=0

l=m
r
l1
[A
ml
cos(m) +B
ml
sin(m)]P
m
l
(cos)+
+

m=0

l=m
r
l
[D
ml
cos(m) +C
ml
sin(m)]P
m
l
(cos)
Imposing the boundary conditions we arrive at the nal solution of the problem
u(r, , ) =

l=1
_
r
l1
A
l
+r
l
B
l
_
P
1
l
(cos)cos
with
a
l1
A
l
+a
l
B
l
=

_
0
f ()P
1
l
(cos)sin d

_
0
(P
1
l
(cos))
2
sin d
24
b
l1
A
l
+b
l
B
l
=

_
0
g()P
1
l
(cos)sin d

_
0
(P
1
l
(cos))
2
sin d
.
Finally the coefcients are
A
l
=

_
0
_
b
l
f () a
l
g()
_
P
1
l
(cos)sin d
(b
l
a
l1
a
l
b
l1
)

_
0
(P
1
l
(cos))
2
sin d
B
l
=

_
0
_
b
l1
f () a
l1
g()
_
P
1
l
(cos)sin d
(b
l1
a
l
a
l1
b
l
)

_
0
(P
1
l
(cos))
2
sin d
S 3.12 We have a linear homogeneous PDE with linear homogeneous BC and thus we can
apply the method of separation of variables. Our equation is
1
c
2

2
u
t
2
=
1
r
2

r
_
r
2
u
r
_
+
1
r
2
sin

_
sin
u

_
+
1
r
2
sin
2

2
u

2
We could observe that initial conditions require that u is independent of . Thus we can simplify
the equation
1
c
2

2
u
t
2
=
1
r
2

r
_
r
2
u
r
_
+
1
r
2
sin
2

2
u

2
.
However although we assumed the independence of the solution from the right hand side
depends on . Thus we can proceed only if

2
u

2
= 0
and this will contradict the IC unless F(r) = 0.
The problem has a trivial solution only if F(r) = 0 otherwise there is no solution.
Alternatively you could separate the variables
u(r, , , t) = G(t) f (r)h()g()
(see the lecture notes) and nd among other things
h() = P
m
l
(cos), g() = Acos(m) +Bsin(m)
with m l. The initial conditions require that m = 2 and at the same time there is no de-
pendence. So we will arrive at the same conclusion, that there exists only trivial solution when
F(r) = 0.
25
S 3.13 We have a linear homogeneous PDE with the linear homogeneous boundary conditions
and thus we can solve the problem by the separation of variables.
Let us solve PDE
u
t
= k

2
u
x
2
u ,
assuming the following form of the solution
u(x, t) = h(t)(x) .
PDE reduces to the following equation
1
h
dh
dt
= k
1

d
2

dx
2
.
Using the standard argument for the separation of variables we get
d
2

dx
2
=
with being a constant and
dh
dt
=(k +)h .
The last equation can be easily solved
h(t) = Ae
(k+)t
.
For we have to specify the boundary conditions
d
dx
(0) = 0 , (L) = 0 .
Lets consider the different cases. For the case > 0 the general solution is
(x) = Asin(

x) +Bcos(

x) .
Imposing the boundary conditions we arrive at the solution
(x) = Acos
_
(n+
1
2
)x
L
_
, =
_
(n+
1
2
)
L
_
2
, n = 0, 1, 2, 3, ...
For the case 0 there are no solutions satisfying the boundary conditions. Thus the solution
of the original PDE can be written as
u(x, t) =

n=0
A
n
e
((
(n+1/2)
L
)
2
k+)t
cos
_
(n+
1
2
)x
L
_
.
Finally we have to impose the initial conditions
u(x, 0) =

n=0
A
n
cos
_
(n+
1
2
)x
L
_
= f (x)
26
Using the orthogonality of coss we get the following expressions for As
A
n
=
2
L
L
_
0
f (x)cos
_
(n+
1
2
)x
L
_
dx
Now about the limit t . If
k
_
(n+
1
2
)
L
_
2
+ > 0
(for all ns) then
u(x, t) 0
Otherwise u(x, t) will go to innity.
S 3.14 We have to solve the heat equation in cylindrical coordinates ignoring the dependence
from z. We solve the following equation
u
t
=
_
1

_
+
1

2
u

2
_
Assuming that
u(, , t) = G(t)F()g()
we get the following set of ODEs
d
2
g
d
2
=g, g( +2) = g()

2
d
2
F
d
2
+
dF
d
+(
2
)F = 0, F(1) = 0, |F(0)| <
dG
dt
=G
The rst equation has a solution
g() = Acos(m) +Bsin(m), = m
2
, m = 0, 1, 2, ...
and the second
F() =CJ
m
(z
mn
),
mn
= (z
mn
)
2
, J
m
(z
mn
) = 0, n = 1, 2, ... ,
where we dropped the Bessel function of second kind since it goes to innity at = 0. The
solution for G is
G(t) = De
t
.
Thus we get
u(, , t) =

m=0

n=1
[B
mn
sin(m) +A
mn
cos(m)]e
(z
mn
)
2
t
J
m
(z
mn
)
27
Imposing the remaining initial condition we arrive at the solution of the problem
u(t, r, ) =

n=1
_
B
1n
e
(z
1n
)
2
t
sin()J
1
(z
1n
) +B
2n
e
(z
2n
)
2
t
sin(2)J
2
(z
2n
)
_
with
B
1n
=
a
_
0
d g()J
1
(z
1n
)
a
_
0
d (J
1
(z
1n
))
2
and
B
2n
=
a
_
0
d f ()J
2
(z
2n
)
a
_
0
d (J
2
(z
2n
))
2
S 3.15 Solve the problem inside of a sphere rst.
We have a linear homogeneous PDE with linear homogeneous BC and thus we can apply the
method of separation of variables. Our equation is
1
c
2

2
u
t
2
=
1
r
2

r
_
r
2
u
r
_
+
1
r
2
sin

_
sin
u

_
+
1
r
2
sin
2

2
u

2
We could observe that initial conditions require that u is independent of . Thus we can simplify
the equation
1
c
2

2
u
t
2
=
1
r
2

r
_
r
2
u
r
_
+
1
r
2
sin

_
sin
u

_
.
We separate the variables
u(r, , , t) = G(t) f (r)h()
and arrive at the following equations and BC
d
d
_
sin
dh
d
_
+ sinh = 0 , |h(0)| < , |h()| < ,
d
dr
_
r
2
d f
dr
_
+(r
2
) f (r) = 0 , f (a) = 0 , f <
d
2
G
dt
2
+c
2
G = 0 ,
The solutions of the rst equation are
h() = P
l
(cos) , l = 0, 1, 2, ...
and the second equation is solved by
f (r) = r
1/2
J
l+
1
2
(
_

lk
r) ,
lk
=
_
z
(l+1/2)k
a
_
2
, k = 1, 2, 3, ...
28
where z
(l+1/2)k
are the zeros of J
l+
1
2
. The last equation is solved as follows
G(t) = Acos(

ct) +Bsin(

ct) .
Imposing the one of IC
u
t
(r, , , t) = 0
we get
u(r, , t) =

l=0

k=1
A
lk
r
1/2
J
l+
1
2
(
_

lk
r)P
l
(cos)cos(
_

lk
ct) .
Imposing the second IC we get
A
lk
=
a
_
0
dr

_
0
dF(r)sin2 r
3/2
sinJ
l+
1
2
(
_

lk
r)P
l
(cos)
a
_
0
dr

_
0
d r sin(J
l+
1
2
(
_

lk
r))
2
(P
l
(cos))
2
Now lets comment on the solution outside of a sphere. In this case the solution for f will be
f
lk
(r) = r
1/2
J
l+
1
2
(
_

lk
r) +r
1/2
Y
l+
1
2
(
_

lk
r) ,
where
lk
is now dened as
J
l+
1
2
(
_

lk
a) +Y
l+
1
2
(
_

lk
a) = 0 .
Due to the general properties of the SL problem there is the following orthogonality condition

_
a
dr r f
lk
(r) f
ln
(r) = 0 , k = n .
S 4.1 First we have to switch to the homogeneous BC. We can do it by the following change
v(x, t) = u(x, t) 1+
x

.
v satises the following PDE
v
t
=

2
v
x
2
+sin(7x)e
3t
on the interval [0, ] subject to
v(0, t) = 0, v(, t) = 0, v(x, 0) =
x

1.
We look for the solution in the form
v(x, t) =

n=1
a
n
(t)sin(nx)
29
which satises BC. Substituting this solution into the PDE we get the following set of ODEs
da
n
dt
=n
2
a
n
, n = 7,
da
7
dt
=49a
7
+e
3t
.
These ODEs have the solutions in the form
a
n
(t) = e
n
2
t
a
n
(0), n = 7
a
7
(t) = e
49t
_
a
7
(0)
1
46
_
+
1
46
e
3t
.
Imposing the IC
v(x, 0) =

n=1
a
n
(0)sin(nx) =
x

1
we get
a
n
(0) =
2

_
0
sin(nx)
_
x

1
_
dx.
Thus the nal solution is
u(x, t) = 1
x

+
1
46
_
e
3t
e
49t
_
sin(7x) +

n=1
a
n
(0)e
n
2
t
sin(nx)
30