Solutions to Final Exam

Physics 604
Dec. 15, 2003
This CLOSED BOOK exam is to be completed in 2 hours. Choose THREE of the four problems if you attempt all four,
be sure to indicate clearly which three should be graded. You must explain your reasoning to receive full credit.
Initialization
ClearAll@"Global`"D;
Off@General::spell, General::spell1D

$DefaultFont = 8"Times", 12<;

$TextStyle = 8FontFamily "Times", FontSize 12, FontSlant "Italic"<;
Needs@"Graphics`Master`"D;
Needs@"Utilities`Notation`"D;

Problem 1 (points per part: 4, 4, 2)

Sound waves in a pipe of slowly varying cross section A@xD satisfy the equation
A@xD 2 y@x, tD
i
y@x, tD y
2
- jjA@xD zz 0
2
c
x k
t
x {

Consider an exponential horn with A@xD a x .

a) Show that there exist travelling waves of the form
y@x, tD = -b x Cos@k x - w tD
where b is independent of w and determine the dispersion relation k@wD.
b) Describe this solution. What happens for low frequencies? Develop and describe an alternative solution, if
needed, for low frequencies.
c) Evaluate and sketch the phase and group velocities for propagating waves as functions of either k or w (your
choice).
a)
We simply substitute the proposed solution into the wave equation to obtain

w2
- 2 Exp@Ha - bL xD Cos@k x - w tD + H Exp@Ha - bL xD Hb Cos@k x - w tD + k Sin@k x - w tDL L 0
c
x

SolutionsToFinal.nb

or
w2
- 2 Cos@k x - w tD +
c
Ha - bL Hb Cos@k x - w tD + k Sin@k x - w tDL + H-k b Sin@k x - w tD + k 2 Cos@k x - w tDL 0
Collecting the coefficients of the sine and cosine functions provides the two equations
w2
- 2 + Ha - bL b + k 2 0
c
Ha - bL k - k b 0

such that
a
b =
2
a2
w2
a2
w2
2 - %%%%
%
- 2 + + k 2 0 k = $%%%%%%%%%%%%%%%%
c
4
c
4
This solution is an attenuated travelling wave where the attenuation length, b-1 , is independent of frequency. Hence, the
balance between high and moderately low frequencies is preserved as the sound propagates.
b)

However, there is a low-frequency cut-off, wc = a c 2 = b c, below which waves do not propagate because k becomes
complex. For very low frequencies, we substitute a trial solution of the form
w2
y@x, tD = -b x Cos@w tD - 2 + bHa - bL 0
c
to obtain

w2
a
a2
- %%%%
%
b = $%%%%%%%%%%%%%%%%
4
c2
2

for standing waves. Both damping coefficients are positive and must be retained, but after some distance the smaller one
will dominate and determines the maximum penetration of low-frequencies vibrations into the horn. Notice that at w 0
there is a static solution with b 0 that is not attenuated.
c)
The phase and group velocities are given by
a2
w
%%%%%%%
1 + %
%%
v p = = c $%%%%%%%%
k
4 k2

w
c2
w w c2 k k vg = =
vp
k

such that
v p vg = c2

These velocities are sketched below as functions of k a. Notice that v p diverges and vg vanishes at k = 0.

SolutionsToFinal.nb

vc

1 y12 i
1 y12
i
z
z
PlotAEvaluateA9j
=E,
j1 +
z ,j
z
j1 +
4 k2 {
4 k2 {
k
k
8k, 0.6, 5<, PlotRange 8Automatic, 80.8, 1.2<<, Frame True,
FrameLabel 8"ka", "vc"<, PlotStyle 88<, Dashing@80.02, 0.02<D<,
PlotLegend 8"vp c", "vg c"<, LegendPosition 80.3, 0.3<,
LegendSize 80.6, 0.25<, LegendShadow NoneE;

v p c
vg c

1.15
1.1
1.05
1
0.95
0.9
0.85
0

ka
It may also be of interest to sketch these velocities as functions of w wc instead. Notice that v p diverges and vg vanishes
at wc , and are complex for smaller frequencies.
12
1 y12
i 2 1y
i
2
j
z
z
PlotAEvaluateA9 j
,

=E,

j
j
z
z
4{
4{
k
k
8, 0.5, 5<, PlotRange 8Automatic, 80.4, 1.5<<, Frame True,
FrameLabel 8"c ", "vc"<, PlotStyle 88<, Dashing@80.02, 0.02<D<,
PlotLegend 8"vp c", "vg c"<, LegendPosition 80.3, 0.3<,
LegendSize 80.6, 0.25<, LegendShadow NoneE;

v p c
vg c

1.4

vc

1.2
1
0.8
0.6
0

3
wwc

SolutionsToFinal.nb

Problem 2 (10 points)

Use the Laplace transform to solve the initial-value problem
y @tD + y@tD t -t , y@0D 1
t

solution
The Laplace transform of this integro-differential equation reads
s Hs + 2L
s y - 1 + s-1 y Hs + 1L-1 y =

Hs + 1L Hs2 + 1L
The solution transform is resolved into partial fractions according to
s Hs + 2L
C
A
B
A Hs2 + 1L + B Hs + 1L Hs - L + C Hs + 1L Hs + L s2 + 2 s
+ +
Hs + 1L Hs2 + 1L
s-
s+1
s+
which yields the system of equations
A+B+C 1
B H1 - L + C H1 + L 2
A-B+C 0
The second gives
2 - H1 + L C
B =
1-
and substitution into the third gives

2
H1 - L A - 2 + H1 + L C + H1 - L C 0 A = H1 - CL
1-

Finally, substitution into the first equation gives

3-
1+3
2 H1 - CL + 2 - H1 - L C + H1 - L C 0 C = =
4
4

such that

2 1+
1
A = J N = H1 + L2 = -
4
4
1-
2
8 - H1 + L H3 - L
8 - H4 + 2 L
H4 - 2 L H1 + L
3+
B =
= =
=
4 H1 - L
4 H1 - L
8
4

Thus,

s Hs + 2L
3- 1
1 1
3+ 1
y =
= - + +
2
Hs + 1L Hs + 1L
4 s-
2 s+1
4 s+

can be inverted using

1
1
3+
3-
-a t y@tD = - -t + - t + t
4
4
s+a
2

SolutionsToFinal.nb

Therefore, we obtain
1
y@tD = H3 Cos@tD + Sin@tD - -t L
2

Problem 3 (10 points)

Consider an integral of the form

g@xD x

where g@xD is well-behaved on the positive real axis but is not symmetric with respect to the sign of x, such that the
integration interval cannot be extended for use with a great semicircle. If the corresponding g@zD is a meromorphic
function that decreases sufficiently rapidly for z and is sufficiently small at the origin, one can often use
f @zD = g@zD Log@zD with a branch cut on the positive real axis and a PacMan contour of the form shown below.
Apply this method for the following integral.

Hx + 1L Hx2 + 2 x + 2L

For this integral one could resolve the integrand into partial fractions instead, but to obtain full credit you must
employ the method described above, which can be useful for more complicated problems that are not suitable for
partial-fraction decomposition. Be sure to justify neglect of any portions of the contour that do not contribute.

solution
Let

Log@zD
Log@zD
f @zD =
=

2
Hz + 1L Hz + 2 z + 2L
Hz + 1L Hz + 1 + L Hz + 1 - L

with a branch cut on the positive real axis, such that

SolutionsToFinal.nb
Log@xD
z = x + f @zD =

Hx + 1L Hx2 + 2 x + 2L
Log@xD + 2 p
z = x - f @zD =

Hx + 1L Hx2 + 2 x + 2L
z
Log@RD + q
z = R q f @zD >
0
R2
q

For large R we use

while near the origin

z
z = q f @zD HLog@D + qL 0
q

Thus, we can neglect both circular portions and obtain

Log@zD
x
1

z
= -

2
2 p C Hz + 1L Hz + 1 + L Hz + 1 - L
0 Hx + 1L Hx + 2 x + 2L
The right-hand side can now be evaluated as the sum of the residues for z = -1, -1 - , -1 + , such that

x
Log@-1D
Log@-1 - D
Log@-1 + D
= -J + + N

2 + 2 x + 2L
Hx
+
1L
Hx
HL
H-L
H-L
H-2
L
HL H2 L
0
1
5p
1
3p
!!!
!!!
= -J p - JLogA 2 E + N - JLogA 2 E + NN
2
4
2
4
Therefore,

x
!!!

= LogA 2 E
Hx + 1L Hx2 + 2 x + 2L

Problem 4 (5 points per part)

The curved surface of a cylinder of radius a is grounded while the endcaps at z = L 2 are maintained at opposite
potentials y@x, f, L 2D = V @x, fD.
a) Develop an expansion for the electrostatic potential y@x, f, zD within the cylinder and express the coefficients in
terms of the appropriate integral over V @x, fD.
b) Determine the coefficients for the simple case V @x, fD = V0 where V0 is constant.

Possibly useful information:

1 2
1 i y
2
2 = jjx zz + 2 2 + 2
x f
x x k x {
z

1 R i
m2 y
2 R
2 + + jjk 2 - 2 zz R 0 R = A Jm @k xD + B Nm @k xD
x
x x
x {
k

a
a2
Jm @km,n aD = 0 Jm @km,n xD2 x x = Jm @km,n aD2
2
0

m 1 Hxm Jm @xDL = xm Jm-1 @xD

x

SolutionsToFinal.nb

a)
The Laplacian takes the form

1 i y
1 2
2
2 = jjx zz + 2 2 + 2
x x k x {
x f
z

in cylindrical coordinates. Laplace's equation

2 y 0

y@x, f, zD = R@xD F@fD Z@zD

with boundary conditions

R@0D finite, R@aD 0
F@f + 2 pD F@fD
L
L
ZB- F -ZB F
2
2
then separates into
1 2 F
2 -m2
F f
1 2 Z
2 k 2
Z z

F = m f , m = 0, 1, 2,
Z = Sinh@k zD

1 R i
m2 y
2 R
2 + + jjk 2 - 2 zz R 0 R = Jm @k xD
x
x x
x {
k

where Jm is the Bessel function of the first kind, the one that is finite at the origin. Note that although the Bessel equation
is insensitive to the sign of m, the conventional definition J-m @xD = H-Lm Jm @xD simplifies some of the relationships among
Bessel functions. The boundary condition at the curved surface requires k km,n where the index n enumerates the roots
of
Jm @km,n aD 0

Thus, the general solution takes the form

y@x, f, zD = ym,n Jm @km,n xD Sinh@km,n zD m f

m=- n=1

where the yn,m are numerical coefficients obtained by matching the boundary conditions at z = L 2 according to
ym,n = Ip a2 Jm @km,n aD2 Sinh@km,n L 2DM

-1

2p

f x x - m f Jm @km,n xD V @x, fD
a

b)

The expansion coefficients for V @x, fD = V0 require the integrals

2p

f - m f = 2 p dm,0

a
k0,n a
k0,n a
1
a
1

J0 @xD x x = 2
Hx J1 @xDL x = J1 @k0,n aD
J0 @k0,n xD x x = 2
k0,n
x
k0,n
k0,n
0
0
0

SolutionsToFinal.nb

Thus, the electrostatic potential becomes

J0 @k0,n xD
Sinh@k0,n zD
2 V0
y@x, f, zD =
a n=1 km,n J1 @k0,n aD Sinh@k0,n L 2D