Physics 116C

Solutions to Homework Set #6

Fall 2011

Boas, p. 643, problem 13.5-3(b)

Find the steady-state temperature distribution in a solid cylinder of height H and radius a if the top and
curved surfaces are held at 0 and the base at 100.
We have to solve the equation
2 u(r, , z) = 0

(1)

subject to the boundary conditions

u(r, , H) = u(a, , z) = 0 ,

u(r, , 0) = 100

(2)

After the usual separation of variables in cylindrical coordinates u(r, , z) = R(r)()Z(z) we are left
with
1
Z
1
(rR ) + 2 =
= K 2
(3)
rR
r
Z
The R equation can also be separated and gives

r

() = ein ,
(rR ) + K 2 r 2 =
= n2
=
(4)
R(r) = Jn (Kr)
R

n has to be an integer so that ( + 2) = (). The Z equation gives Z(z) = A sinh Kz + B cosh Kz =
A sinh K(H z), were we have parametrized the solution so that Z(H)=0. The full solution is
u(r, , z) = Jn (Kr) sinh K(H z)ein

(5)

We now have to impose the boundary conditions (2), but first we notice that they have a rotational
symmetry around the z axis (that is, there is no -dependence): then there will be no -dependence in
the answer, that is, we have n = 0. For the other boundary conditions,
u(a, , z) = 0 = J0 (Ka) = 0 = Ka = xm is a zero of the Bessel function J0 (x)
u=

X
m

cm J0 (xm r/a) sinh[xm (H z)/a]

u(r, , 0) = 100 =

(6)
(7)

cm J0 (xm r/a) sinh(xm H/a)

(8)

Because the set {J0 (xm r/a) , m = 1, 2, . . .} is an orthogonal set in (0, a), we can find the coefficient cn by
multiplying expression (8) by J0 (xn r/a) and integrating between 0 and a. Then we get (following Boas,
p.641)
200
sinh(xn H/a)cn =
(9)
xn J1 (xn )
or

X
200
J0 (xm r/a) sinh[xm (H z)/a].
(10)
u(r, , z) =
xm J1 (xm ) sinh(xm H/a)
m=1

Boas, p. 643, problem 13.5-5

A flat circular plate of radius a has a temperature distribution u(r, ) = 100r sin . At t = 0, the
circumference of the plate is held at 0. Find the time dependent temperature distribution.
We have to solve the problem


1
u = 2 u,
t
2

u(r, , 0) = 100r sin ,

u(a, , t) = 0, t > 0

(11)

We solve the differential equation in the usual way:

u = F (x, y)T (t) =
The solution for T is T (t) = ek

2 2 t

1 2
1 T
F = 2
= k2
F
T

(12)

, while for F we have

r
(rR ) + k2 r 2 =
= n2 =
F (x, y) = R(r)() =
R

() = ein ,
R(r) = Jn (kr)

(13)

The boundary condition R(a) = 0 tells us that ka is a zero of the n-th Bessel function; if we let xmn
be the m-th zero of the Bessel function Jn , then ka = xmn . Unlike in the last problem, we now have a
-dependence in the initial condition:
X
u(r, , 0) = 100r sin =
Jn (xmn r/a)(Amn cos n + Bmn sin n)
(14)
m,n

By inspection, we already see there will be no cos n dependence on the right hand side (Amn = 0). If we
multiply the previous equation by sin l and integrate over (0, 2), we have
100r

sin l sin d = 100J1 (r)l1 =

0

Jn (xmn r/a)Bml ln ,

(15)

n,m

= Bml = 0 for l 6= 1,

100r =

J1 (xm1 r/a)Bm1 .

If we now multiply by rJ1 (xk1 r/a) and integrate over r (0, a), we get
Z a
Z a
ar d
a3
2
r J1 (xk1 r/a)dr = 100
100
rJ2 (xk1 r/a)dr = 100
J2 (xk1 ) =
xk1
0 xk1 dr
0
Z a
X
a2
rJ1 (xm1 r/a)J1 (xk1 r/a)dr = Bk1 J22 (xk1 )
=
Bm1
2
0
m
Hence,
Bk1 =

(16)

200a
.
xk1 J2 (xk1 )

(17)
(18)

(19)

So that the final solution is

u=

X
k=1

200a
2
2
2
J1 (xk1 r/a)exk1 /a t
xk1 J2 (xk1 )

with xk1 =zeros of J1 (x).

(20)

Boas, p. 643, problem 13.5-7

Find the steady state temperature distribution in a solid cylinder of height 20 and radius 3 if the flat ends
are held at 0 and the curved surface is at 100.
We have to solve the problem


u(r, , z) = 0 ,

u(r, , 0) = u(r, , 20) = 0,

u(a, , z) = 100.

(21)

Following the derivation on pp. 639640 of Boas, we write

u(r, , z) = R(r)()Z(z) ,
and obtain

11 d
R r dr

dR
r
dr

dR
r
dr

1 d2 Z
1 1 d2
+
= 0.
r 2 d 2
Z dz 2

(22)

(23)

Choosing the opposite sign for the separation constant as compared to eq. (5.4) on p. 639 of Boas, we
have
(
sin kz ,
1 d2 Z
2
(24)
= k ,
Z=
Z dz 2
cos kz .
Inserting this result back into (23),
11 d
R r dr

1 1 d2
k2 = 0 .
r 2 d 2

We can separate the variables by multiplying through by r 2 ,



dR
1 d2
r d
k2 r2 = 0 .
r
+
R dr
dr
d 2

(25)

(26)

Thus we have,
(
sin n ,
=
cos n .

1 d2
= n2 ,
d 2

Inserting this back into (26), we obtain the radial equation,



dR
r d
r
n2 k 2 r 2 = 0 ,
R dr
dr

(27)

(28)

or

d2 R
dR
(k2 r 2 + n2 )R = 0 .
(29)
+r
2
dr
dr
Comparing this differential equation with eq. (17.2) on p. 595 of Boas, we identify the possible solutions
as:
(
In (kr) ,
(30)
R(r) =
Kn (kr) .
r2

However, since the present problem includes the origin, we demand that the radial solution should be
finite as r 0. In light of the small r behavior of In (kr) and Kn (kr) as specified on p. 604 of Boas, we
must exclude Kn (kr) as a possible solution. Hence,
R(r) = In (kR) .
3

(31)

If we now impose the requirement that the temperature is held at 0 at z = 0 and 100 on the surface of
the solid cylinder, then it follows that n = 0 since the boundary value temperature has no -dependence.
Hence, the allowed solutions for the temperature are given by:
u(r, , z) = I0 (kr) sin kz .

(32)

Next, we impose the boundary condition that the temperature is held at 0 at z = 20. This implies
that 20k = n or k = m/20, for m = 1, 2, 3, . . .. Hence, the solution must be of the form,

u(r, , z) =

cm I0 (mr/20) sin(mz/20) .

(33)

m=1

Finally, we impose the boundary condition that u(3, , z) = 100. This yields
100 =

cm I0 (3m/20) sin(mz/20) ,

(34)

m=1

which we recognize as a Fourier sine series. Thus, we can solve for the Fourier coefficients, cm I0 (3m/20),
20
Z 20

1
200

cm I0 (3m/20) =
cos(mz/20)
100 sin(mz/20)dz =

10 0
m
0

0 ,

200
[1 (1)m ] =
=

m
400 ,
m

for even m ,

(35)
for odd m .

This equation determines the cm , which we can insert back into (33) to obtain the final solution,
 mz 
 mr 
400 X
1
u(r, , z) =
sin
.
I0

mI0 (3m/20)
20
20

(36)

odd m

Boas, p. 644, problem 13.5-12

Solve the Laplaces equation in 2 dimensions in polar coordinates and solve the r and equation. Solve
the problem of the steady-state temperature in a circular plate if the upper semicircular boundary is held
at 100 and the lower at 0.
The Laplacian in polar coordinates is
1 d
=
r dr
2



d
1 d2
r
+ 2 2
dr
r d

(37)

so that Laplaces equation 2 u = 0 becomes (with u = R(r)())

r

(rR ) =
= n2
R

(38)

The angular dependence is = ein and n must be an integer for the function to be single-valued,
( + 2) = (). The radial equation is
r 2 R + rR n2 R = 0
4

(39)

To solve this we try a solution R = r and substituting we find = n. The solution to Laplaces
equation is then
X

u=
r n (An sin n + Bn cos n) + r n (An sin n + Bn cos n)
(40)
n=0

To solve the requested problem, first we put A = B = 0 because the temperature would diverge in
the center of the plate. Then we apply the boundary condition u(a, ) = 100H( + ) where H is the
Heaviside step function:
X
an (An sin n + Bn cos n)
(41)
u(a, ) = 100H( + ) =
n=0

Now we must multiply this equation

R by sin m or cos m and integrate between (0, 2) in order to find
respectively An and Bn . Because 0 cos m d = m0 , only the constant term in the cosine series is
non-zero. It is given by
Z
Z
2

d = 2 = B0 = 50
d = B0
100
0
0
Z
200
1
for odd n, and 0 otherwise
100 sin n d =
an An =
0
n
We can rewrite the solution as
200 X  r n sin n
u = 50 +

a
n

(42)

(43)

(44)

odd n

Boas, p. 650, problem 13.7-7

Find the steady state temperature distribution inside a sphere of radius 1 with the following surface
temperature distribution:

cos ,
0 < < /2,
(45)
u(1, , ) =
0,
/2 < < .
In spherical coordinates, the Laplacian is




d
d2
1 d
1
d
1
2
2 d
= 2
r
+ 2
sin
+ 2 2
r dr
dr
r sin d
d
r sin d2
After separating the variables, u(r, , ) = R(r)()(), we have


d
m2
1 d
2
= 0,
sin
+ k
(r R ) = kR ,
sin d
d
sin2

= m2

(46)

(47)

The R equation is solved by setting k = l(l + 1) and the solutions are r l , r l1 . The second solution
diverges at the origin so we do not consider it. The equation gives the associated Legendre polynomials
in cos , while the equation is solved by the sine and the cosine. The full solution to Laplaces equation
is
u(r, , ) = r l Plm (cos )eim
(48)
m must be integer for to be single valued, while l must be integer for Plm to not diverge at = 0, . As
the initial condition (45) has no -dependence, the only allowed value will be m = 0, which gives us the
Legendre polynomials Pl (cos ). We can write our solution as
u(r, , ) =

X
l=0

cl r l Pl (cos ) .

(49)

We now impose the boundary condition at r = 1. Defining x cos ,

(
)

X
x,
0 < x < q,
u(r = 1) =
=
cl Pl (x) .
0,
1 < x < 0

(50)

l=1

We can now use the orthogonality of the Legendre polynomials,

cl =

2l + 1
2

u(r = 1)Pl (x)dx =

R1

1 dx Pl Pm

2l + 1
2

2lm
2l+1

and find

xPl (x)dx .

(51)

To evaluate this integral, we first employ the recursion relation given in eq (5.8) on p. 570 of Boas,
lPl (x) = (2l 1)xPl1 (x) (l 1)Pl2 (x) .

(52)

1
[(l + 1)Pl+1 (x) + lPl1 (x)] .
2l + 1

(53)

Shifting l l + 1, we obtain
xPl (x) =
Hence,

xPl (x)dx =



Z 1
Z 1
1
Pl1 (x)dx .
Pl+1 (x)dx + l
(l + 1)
2l + 1
0
0

We first consider the cases of l = 0 and l = 1.

Z 1
Z 1
xdx =
xP0 (x)dx =
0

1
2

xP1 (x)dx =

x2 dx =

1
3

(54)

For l 2, we make use of the result of problem 1223.3 on p. 615 of Boas,

Z 1
Z 1
(1)n (2n 1)!!
.
P2n+1 (x)dx =
P2n (x)dx = 0 , for n > 0 , and
2n+1 (n + 1)!
0
0

(55)

(56)

Thus (54) and (56) yield

Z

xP2n+1 (x)dx = 0 ,

for n = 1, 2, 3, . . . ,

(57)



1
(1)n (2n + 1)!! (1)n (2n 3)!!
xP2n (x)dx =

4n + 1
2n+1 (n + 1)!
2n1 (n 1)!
=

(1)n (2n 3)!!

[(2n + 1)(2n 1) 4n(n + 1)]
(4n + 1)2n+1 (n + 1)!

(1)n+1 (2n 3)!!

,
2n+1 (n + 1)!

for n = 1, 2, 3, . . .

(58)

Hence, (51) yields

c0 =

1
4,

c1 =

1
2

c2n

(1)n+1 (4n + 1( 2n 3)!!

,
=
2n+2 (n + 1)!

c2n+1 = 0 ,

for n = 1, 2, 3, . . .

(59)

We can read the first few terms as

c0 =

1
,
4

c1 =

1
,
2

c2 =

5
,
16

c3 = 0 ,

c4 =

3
...
32

(60)

Inserting these results into (49), we obtain

u(r, , ) =

1
4

+ 12 r cos +

X
(1)n+1 (4n + 1( 2n 3)!!

2n+2 (n + 1)!

n=1

r 2n P2n (cos ) .

(61)

The first few terms of the series are:

u(r, , ) =

1
4

+ 21 r cos +

5 2
3 4
16 r P2 (cos ) 32 r P4 (cos ) +

(62)

Boas, p. 650, problem 13.7-11

Find the steady state temperature distribution inside a hemisphere if the spherical surface is held at 100
and the equatorial plane at 0.
We already have the solution for a full sphere with the upper half held at 100 and the lower half at 0
(Boas, p.649, eq. (7.15)):

Z
X
2l + 1 1
0, 1 < x < 0
u0 =
dx f (x)Pl (x) ,
f (x) =
(63)
100cl r l Pl (cos ),
cl =
1,
0<x<1
2
1
l
1
Z 1


1
1
cl = 2
dx f (x)[Pl+1 Pl1 ] = 2 [Pl+1 Pl1 ] = 12 (Pl1 (0) Pl+1 (0)) =
(64)
0
( 1
l
l+1 (2l+1)!!
l
2l+1
12 (1)
= 21 (1) 2(2l1)!!
(1 + 2(l+1)
),
c2l+1 = 12 (1) 2(2l1)!!
l l!
l+1 (l+1)!
l l!
2
(65)
=
1
c2l = 0 ,
c0 = 2

u0 = 100 21 P0 (cos ) + 34 rP1 (cos )

7 3
11 5
16 r P3 (cos ) + 32 r P5 (cos ) + . . .

(66)

Let u1 be a solution corresponding to the upper half held at 0 and the lower half at 100:

Z
X
2l + 1 1
1, 1 < x < 0
l
u1 =
100 dl r Pl (cos ),
dl =
dx g(x)Pl (x) ,
g(x) =
0,
0<x<1
2
1
l
Z
Z
Z
2l + 1 0
2l + 1 1
2l + 1 1
dx g(x)Pl (x) =
dx Pl (x) =
dxPl (x) = (1)l cl
(67)
dl =
2
2
2
1
1
0
This tells us that d0 = 12 , d2l = 0, d2l+1 = c2l+1 : then u1 is

7 3
u1 = 100 12 P0 (cos ) + 34 rP1 (cos ) 16
r P3 (cos ) +

11 5
32 r P5 (cos ) +

...

(68)

Finally, u0 + u1 will still be a solution of Laplaces equation, and satisfy the boundary conditions with the
two hemispheres held at 100o and 100 . This also corresponds to have the central plane at 0o , so that
the if we consider u = u0 + u1 </2 we got the solution to our original problem:

u = 200 34 rP1 (cos )

7 3
11 5
16 r P3 (cos ) + 32 r P5 (cos ) +

...

(69)

Boas, p. 651, problem 13.7-16

Separate the wave equation in spherical coordinates and find that the , solutions are the spherical
harmonics Ylm (, ) = Plm (cos )eim and the r solutions are the spherical Bessel functions jl (kr) and
yl (kr)
The wave equation is

1 2u
,
v 2 t2
where v is the speed of the wave. In spherical coordinates, the Laplacian is




d2
d
1
d
1
1 d
2 d
2
r
+ 2
sin
+ 2 2
= 2
r dr
dr
r sin d
d
r sin d2
2 u =

(70)

(71)

After separating the variables, u(r, , , t) = F (r, , )T (t) = R(r)()()T (t), we have
1 T
2 F
= 2
= k2 = T (t) = eikvt ,
2 F + k2 F = 0,
F
c T


1 2
d2
d
1
d
1
2 2
(r R ) + k r =
= l(l + 1)
sin
+
2
R
sin d
d
sin d2

(72)
(73)

which implies that



d
1 d2
sin d
= m2 ,
(r 2 R ) + (k2 r 2 l(l + 1))R ,
sin
=
d
d
d2


1 d
d
m2
= 0,
= m2
sin
+ l(l + 1)
sin d
d
sin2

(74)

The R equation is precisely the spherical Bessel equation in the variable kr, x2 y +2xy +[x2 l(l+1)]y = 0
so that its solutions are jl (kr) and yl (kr). The equation gives the associated Legendre polynomials in
cos , Plm (cos ), while the equation is solved by eim .

Boas, p. 658, problem 13.8-1

Show that the gravitational potential V = Gm

r satisfy Laplaces equation.
We have to show that 2 V 2 1r = 0. For that we express the Laplacian in spherical coordinates and
note that the only contribution comes from the radial part:


1
1 d
1 d
21
2 d
= 2
= 2 (1) = 0 , for r 6= 0 .
(75)
r
r
r dr
dr r
r dr
With this we have proven that the gravitational potential satisfies Laplaces equation at all points in space
away from the origin.

Boas, p. 658, problem 13.8-3

Solve Poissons equation for a charge q inside a grounded sphere to obtain the potential V inside the
sphere. Sum the series solution and state the image method of solving the problem.
Let the charge q be at (0, 0, a) inside the sphere, a < R. Poissons equation is
2 V = ,

= q(x)(y)(z a)

(76)

This is solved by
1
u(x, y, z) =
4

(x , y , z )dx dy dz
p

(x x )2 + (y y )2 + (z z )2

=p

q
x2 + y 2 + (z a)2

(77)

We still have to satisfy the boundary condition V (r = R) = 0. For this, we search for another solution
satisfying it and add it to our particular solution (77); the solutions to Laplaces equation in spherical
coordinates is (as we found earlier solving eq. (47))
(Alm r l + Blm r l1 )Plm (cos )eim

(78)

As we are searching for a solution in the inside of the sphere, we take B = 0 so that our result does not
diverge at r = 0. Because the problem is symmetric around the z axis, there will be no dependence,
that is m = 0. The sum of our two solutions is
X
q
+
Al r l Pl (cos )
(79)
V =
r 2 2ar cos + a2
l
where we have expressed the particular solution (77) in spherical coordinates.
The coefficients are chosen in order to satisfy boundary condition
V (r = R) = 0 =
= Al =

qal
R2l+1

X
X
X al
q
P
(cos
)
+
Al Rl Pl (cos ),
+
Al Rl Pl (cos ) = q
l
Rl+1
R2 2aR cos + a2
l
l
l
(80)

where in the first line we have expanded the potential in terms of Legendre polynomials. The solution
satisfying the boundary condition is
V =

X al r l
q
Pl (cos )
q
R2l+1
r 2 2ar cos + a2
l

We can sum the second term (using

q

(81)

l P (x) = 1/ 1 2xh + h2 ):
h
l
l

X al r l
qR/a
q X  ar l
q
=p
P
(cos
)
=

Pl (cos ) = p
l
2l+1
2
2
2
2
4
2
R
R
R
R 1 2(ar/R ) cos + a r /R
r 2(rR2 /a) cos + R4 /a
l
l
2

R
This has the same form of (77), but for a charge qR
a situated at (0, 0, a ). Then the problem of the
grounded sphere with a charge inside it is equivalent to the electrostatic problem of having two charges
2
of different magnitude located at (0, 0, a) and (0, 0, Ra ). The second charge is called image charge.

10

Boas, p. 664, problem 13.10-19

A long conducting cylinder is placed parallel to the z axis in an originally uniform electric field in the
negative x direction. The cylinder is held at zero potential. Find the potential in the region outside the
cylinder.
First, we recall the solution to Laplaces equation in cylindrical coordinates that we found in Problem 4,
eq. (40)
X
 X n
V =
r n (An sin n + Bn cos n) + r n (An sin n + Bn cos n) =
r (An sin n + Bn cos n) (82)
n

n=0

As we will be looking for the potential outside the cylinder, the coefficients multiplying r n must be zero.
The electric field E is in the x direction, that is, E = E0 i = V so that V = E0 x = E0 r cos . We
want a solution to Laplaces equation (not Poissons, as there are no charges) such that V (r = a) = 0 and
V = E0 r cos for large r (far away from the cylinder). Then, we pick
X
V = E0 r cos +
r n (An cos n + Bn sin n),
satisfying V (r = a) = 0
(83)
n

The coefficients are quickly picked by inspection: Bn = 0, An = 0 except for A1 , which satisfies
1
0 = E0 a cos + A1 cos = A1 = E0 a2
a
and finally our solution is
V = E0

11

a2
r
r

cos

(84)

(85)

Boas, p. 664, problem 13.10-20

Use Problem 7 to find the characteristic vibration frequencies of a sound in a spherical cavity.
The vibration modes of a sound in a spherical cavity are solutions to the wave equation (70) subject to
the boundary condition u(r = a) = 0. The general solution was given by




jl (kr)
cos kvt
im
m
u(r, , , t) =
Pl (cos )e
(86)
yl (kr)
sin kvt
where v is the speed of sound. Because we are looking at the solution inside the sphere, the yl does not
contribute. The boundary condition will be satisfied if
jl (ka) = 0, that is, if ka = l is a zero of the spherical Bessel function

(87)

The time dependence of the solution is given by sines and cosines, that is, oscillatory modes with an
angular frequency = kv so that the frequencies of the normal modes are
=

l v

=
2
2a

where l is a zero of the spherical Bessel jl .

10

(88)