Classical Mechanics Problem 1: Central Potential

Solution
a)

Integrals of motion for a central potential V (r):

Angular Momentum
L = rvt = r2
Energy per unit mass E = 12 r 2 + vt2 + V (r) = 21 r 2 + Veff (r)
where vt is the tangential velocity and Veff is defined as
Veff (r) = V (r) +

L2
2r2

If the orbit is circular, the distance of the test body from the origin is invariant: r = 0,
which implies that the body is always at the equilibrium-distance:
dVeff
=0
dr

dV
L2
v2
= 3 = t = r 2
dr
r
r

then
L
= = 2 =
r

1 dV
r dr

1/2

so for the period we get

T =
b)

2
= 2

1 dV
r dr

1/2

Write the orbit as in the statement of the problem:

r(t) = r0 + (t)

with

dVeff
(r0 ) = 0
dr

and

2 r02 .

The energy per unit mass is now E = 12 2 + Veff (r0 + ), and since is small we may
Taylor-expand the potential as
Veff (r0 + ) = Veff (r0 ) +

dVeff
1 d2 Veff
(r0 ) +
(r0 )2 + O(3 )
2
dr
2
dr
| {z }
=0

so then

1 2 1 d2 Veff
+
(r0 )2 + O(3 ) = const.
2
2 dr2
In the above equation we readily recognize the equation of the simple harmonic oscillator with
1/2
2
d Veff
r =
dr2 r=r0
E Veff (r0 ) =

and its general solution is

p

E Veff (r0 )
cos [r (t t0 )]
r

(t) =
where t0 is an arbitrary constant.

Now return to writing r in terms of V (r) instead of Veff (r).

d2 Veff
d2 V
3L2
d2 V
d2 V
3 dV
2
=
+
=
+
3
=
+

dr2
dr2
r4
dr2
dr2
r dr
1/2

1/2
2
3 dV
1 d
d V
3 dV
r =
+
=
r
dr2
r dr r=r0
r3 dr
dr
r=r0

r2 =

And the radial period is

Tr =
c)

2
r

Stability is determined by the sign of r2 . For stability: r2 > 0, so

1 d
r3 dr

3 dV
r
>0
dr

for the Yukawa-potential

V (r) =

GM kr
e
r

so the condition is

GM
1 d
3 dV
r
= 3 ekr 1 + kr (kr)2 > 0
1 + kr (kr)2 > 0
3
r dr
dr
r

!
!

51
5+1
2
1 + kr (kr) =
+ kr
kr > 0
2
2
which is satisfied only if

kr <

Therefore circular orbits are unstable for

kr >

5+1
2

5+1
2

d)

The outermost stable circular orbit is at

r0 =

5+1
2k

its energy per unit mass is

1
1
E = V (r0 ) + (r0 )2 = V (r0 ) +
2
2
GM kr0
1 GM kr0
e
(kr0 1) =
e
2 r0
r0

dV
r
dr r=r0

51
4

!
>0

If r0 is decreased only slightly, E > 0 still and the orbit is absolutely stable
The effective potential for the Yukawa-potential has the form shown in Figure 1.
Veff

Figure 1: Effective potential against distance from the origin

Classical Mechanics Problem 2: Planar Double Pendulum

Solution

q1

q2

a)

L=T V
The moment of inertia for a uniform rod of length l and mass m is
I=

1 2
ml
3

about one of the ends

and

1
ml2 about the rods center
12
The kinetic energy term we can decompose into three parts:
Ic =

T = T1 + T2,rot + T2,trans
where T1 is the kinetic energy of the first rod, T2,trans is the translational energy of
the center of mass of the second rod and T2,rot is its rotational energy about its center
of mass. Then
1
T1 = ml2 12
6
1
T2,rot =
ml2 22
24
and

1
T2,trans = m x 2c + y c2
2
where xc and yc are the coordinates of the second rods center of mass, so
l
sin 2
2
l
yc = l cos 1 cos 2
2

xc = l sin 1 +

from which

1
x 2c + y c2 = l2 12 + 22 + 1 2 (sin 1 sin 2 + cos 1 cos 2 )
4
4

The potential energies are simply Vi = mgyc,i , where yc,i are the vertical coordinates of
the rods centers of mass. Since both rods are uniform, yc,i are simply the coordinates
of the centers. Thus,

l
l
V1 = mg cos 1 ; V2 = mg l cos 1 + cos 2
2
2
The full Lagrangian is then
L = T1 + T2,rot + T2,trans V1 V2

1
3
2
1
= ml2 12 + 22 + 1 2 cos (1 2 ) + mgl
cos 1 + cos 2
3
6
2
2
b)

Expand the Langrangian from part a) for small angles. The only function we have to
deal with is
1
cos = 1 2 + O(4 )
2
Since we are going to look for normal modes with j = j exp (it), where the j 1,
we immediately see that in the term 1 2 cos (1 2 ), the -dependence in the cosine
can be dropped, because even the first -dependent term gives a fourth order correction.
Then the approximate Lagrangian is

2
1
3
1
L = ml2 12 + 22 + 1 2 mgl 12 + 22 + const.
3
6
4
4
The Euler-Lagrange equations are
L
d L
=

dt j
j
so in the specific case:
4
1 +
3
1
1 +
2

1
2 +
2
1
2 +
3

g 3
1 = 0
l 2

g 1
2 = 0
l 2

if we now look for normal modes, as mentioned, the above set of equations takes the
form
#"
#
"

4 2
3g
1 2
1
3 2 l
2
1 2 1 g
=0
1 2
2
2
3 2 l
Non-trivial solutions exist if the determinant of the matrix on the left is zero. Denoting
2 = g/l, we can write this condition as

3
1
1
2
4

= 0,
3
2
3
2
4
5

that is
whose solutions are

7 2 7
3
+ =0
36
6
4
6
= 3 ,
7

so finally

c)

6
3
7

1/2
g
l

To sketch the eigenmodes, find eigenvectors of the matrix in part b).

2 = g/l (low-frequency mode)

3
8
1

2 =

1 =
2 7 1 1
3
3

(2 7 1)/3 > 0 and real, therefore the two pendula are in phase;
2 = + g/l (high-frequency mode)

3
8
1
2 =

1 =
2 7 1 1
3
3

(2 7 1)/3 < 0 and real, therefore the two pendula are perfectly out of phase.

Figure 2: The low- (a) and high-frequency (b) normal modes of the planar double pendulum.

Electromagnetism Problem 1
Solution
a)

Normal modes are products of harmonic standing waves in the x, y and z directions.
For their frequencies, we have
1/2

q
nx 2 ny 2 nz 2
= c kx2 + ky2 + kz2 = c
+
+
;
a
b
b

nx , ny , nz Z+

Since a > b, the lowest frequency has nx = 1 and either ny = 1, nz = 0 or ny = 0, nz = 1

(note that ny = 0, nz = 0 does not satisfy the boundary condition Ek,at wall = 0). Since
~ k y, the boundary conditions require
we are told to pick the mode with E
~ t) = E0 y sin x sin z cos t
E(r,
a
b
The magnetic induction we can get from Faradays Law:

~
B
Ey
Ey
~ =x
= c E
c
zc
t
z
x

x
z

x
z
= x
cE0 sin
cos
zcE0 cos
sin
cos t
b
a
b
a
a
b

x
z
z
x
z
~ t) = cE0 x
B(r,
sin
cos
cos
sin
sin t

b
a
b
a
a
b
where the frequency is (by the argument above)

= c
b)

1
1
+ 2
a2
b

1/2

At a boundary of media, the discontinuity in the normal component of the electric field
is 4 times the surface charge density , so
Ey (x, 0, z) = 4
E0
x
z
sin
sin
cos t
4
a
b
(x, b, z) = (x, 0, z)

(x, 0, z) =

and
(0, y, z) = (a, y, z) = (x, y, 0) = (x, y, b) 0
Similarly, at the boundary of media the discontinuity of the tangential component of
the magnetic field is given by the surface current ~
~ =
n
B
7

4
~
c

where n
is a unit vector normal to the surface, so

c2 E0 z
x
z
x

x
z
~(x, 0, z) =
sin
cos
+ cos
sin
sin t
4
b
a
b
a
a
b
~(x, b, z) = ~(x, 0, z)
z
c2 E0 y
sin
sin t
4 a
b
~(a, y, z) = ~(0, y, z)
~(0, y, z) =

c2 E0 y
x
sin
sin t
4 b
a
~(x, y, b) = ~(x, y, 0)

~(x, y, 0) =

c)

Since there is no charge on the b b sides, the force there is purely magnetic and is
given by
Z

1
~
~ d2 x
F (t) =
~ B
2c
bb

E 2 c2
F~ (x = 0, t) = 0 2 2 x

8 a

x
=

Zb

Zb

dz sin2

dy
0

c b
E0 sin t
4 a

z
sin2 t
b

F~ (x = a, t) = F~ (x = 0, t)
The forces point outwards from the box on both sides (as is indicated by the sign in
the equation above).
d)

Start with the sides where y = const. The magnetic component of the force can be
written as above
E 2 c2
F~mag (y = 0, t) = 0 2 y
8

Za
dx
0

Zb
dz
0

1
x
z
1
x
z
cos2
sin2
+ 2 sin2
cos2
2
a
a
b
b
a
b

2 1 b
1 c
a
E0 sin t
+
=
y
2 2
4 a
b
To simplify this result further, use from part a)

1
1
1
b
a
2
2 = (c)2
+
=
ab(c)
+
a2
b2
a
b
Then

ab
E0
~
sin t
Fmag (y = 0, t) =
y
4
2 3
8

sin2 t

The electric component of the force can be written as

Z
F~el (y = 0, t) =

= y

1 ~ 2
E2
E d x = y 0 cos2 t
2
8
1
2

E0
cos t
4

Za

Zb
dz sin2

dx
0

x
z
sin2
a
b

ab
3

F~tot (y = 0, t) = F~el (y = 0, t) + F~mag (y = 0, t)

2

E0
ab 2
cos t sin2 t
= y
4
2 3
2
E0
ab
= y
cos 2t
4
2 3
and
F~tot (y = b, t) = F~tot (y = 0, t)
There net force on the top and bottom sides oscillates between the inward and outward direction with half the period of the lowest frequency mode. In a time average,
therefore, this force cancels.
Next, calculate the force on the sides where z = const. Again, there is no charge,
therefore no electric component; the force is purely magnetic
E 2 c2
F~ (z = 0, t) = 0 2 2 z
8 b

Zb

Za
dx sin2

dy
0

x
sin2 t
a

2
a c
E0 sin t
=
z
b 4
F~ (z = b, t) = F~ (z = 0, t)
The magnetic force is pushing the a b walls outwards, too (sign!).
e)

From the Maxwell stress tensor, the force per unit surface area is
1 ~ ~
E2
B2
1 ~ ~
f~ =
E(E n
)
n
+
B(B n
)
n

4
8
4
8
~ = 0 and B
~ x
On the x = const. walls n
=
x, E
= 0, so
E 2 c2
z
B2
= 02 2 x
sin2
sin2 t
f~(x = {0, a}, t) = x
8
8 a
b
which is exactly the integrand from part c).

~ E
~ y) = E 2 y and B
~ y = 0, so we get
On the y = const. walls n
=
y , E(
1
f~(y = {0, b}, t) = (E 2 B 2 )
y
8
2
E0
x
z
=
cos2 t sin2
sin2
8
a
b

z
1
E02 c2 1
x
2
2 x
2
2 z
2
sin
+ 2 sin
cos

cos
sin t y
8 2
a2
a
b
b
a
b
the sum of the first two integrands from part d).
~ z) = 0 and B
~ z = 0, so we get
On the z = const. walls n
=
z , (E
B2
E 2 c2
x
f~(z = {0, b}, t) = z = 0 2 2 z sin2
sin2 t
8
8 b
a
the last integrand from part d).

y
b
F(x)
z

s
b
F(z)

k(x)

k(y)

k(z)
a

k(z)
s

F(z)
k(x)

k(y)

F(x)
x

Figure 3: Average total forces, surface charges and surface currents on the cavity.

10

Electromagnetism Problem 2: Waves in a Dilute Gas

Solution
(see Feynman Lectures on Physics, vol. II, chapter 32)
a)

The EM wave is travelling in the x

direction; it has a transverse electric field, so assume
Ey
= 0. Then the electron in the atom behaves classically as a damped, driven
harmonic oscillator

me y + y + 02 y = qE0 eit
with the solution
y(t) =

1
qE(t)
.
2
0 + i me

For the dipole moment per unit volume:

P = na (q)y =

02

1
na q 2 E
2 i me

Therefore the volume polarizability is, according to the definition given,

() =

P
1
na q 2
= 2
0 E
0 2 i 0 me

(A quantum mechanical derivation would give this same expression multiplied by the
oscillator strength f for the transition.)
b)

With no free charges or currents, Maxwells equations read

B
t
D
H=
t

D = 0;

E=

B = 0;

and B = 0 H, D = 0 E + P = 0 (1 + )E for a single frequency . This gives us the

following wave-equation
2D
1

2 D = 0.
2
t
0 0 (1 + )
Now let D be that of a plane wave: D ei(kxt) . Then
k 2 = 0 0 (1 + ) 2 = (1 + )
n() =

p
1 + ()

2
c2

One can also get this result by using the microscopic E, B and P fields:

1
P
2
E = P; c B =
+E
0
t 0
11

2E
1 P
c2 2 E =
t2
0 t

also

2P
P
na q 2
2
E.
+

P
=

0
t2
t
me

Together these give us k 2 = (1 + ) 2 /c2 for a plane wave, as before. (Note that we
are neglecting dipole-dipole interactions in the dilute gas.)
c)

We start by noting that according to Fourier-analysis

1
E(x, t) =
2

dk ei(kx(k)t) E(k)

E(k)
=

dx e

ikx

E(x, 0) =

=e

2 2

dx ei(kkc )xx

/(2 2 )

i(kkc )2 2 /2

Now Taylor-expand (k) about k = kc :

d
(k) = (kc ) +
(k kc ) + O{(k kc )2 }
dk kc
kc vph + vg (k kc ) + O{(k kc )2 }
where, by definition, vph and vg are the phase- and group-velocities, respectively. Now
let K = k kc . Then
1
E(x, t) =
2
E(x, t) = e
d)

ikc (xvph t) e

For the dilute gas, n =

is complex)

dK eikc (xvph t)+iK(xvg t)K

d
=
dk

2 /2

(xvg t)2 /(2 2 )

From part c)
vg =

2 2

dk
d

1
=

= eikc (xvph t) N (x vg t, )

c
n

1
d log n
1+
.
d log

1 + 1 + /2, which we will write as n = nr + ini (for

nr 1 +
ni

02 2
na q 2
20 me (02 2 )2 + 2 2

na q 2

2
20 me (0 2 )2 + 2 2
12

Here the real part nr of the index of refraction determines the dispersion, and the
imaginary part ni determines the absorption/gain coefficient. At = 0 :
nr = 1

vg =

d log nr
na q 2
=
d log
20 me 2

and

na q 2
20 me 2

1
c

1+

na q 2
20 me 2

Note that vg > c at = 0 . This is called anomalous dispersion. It does not violate causality because signals (information) cannot travel faster than the minimum of
(vph , vg ), and now vph = c (since nr = 1). Also, the waves are damped by the electronic
resonance maximally at = 0 .

13

Quantum Mechanics Problem 1

Solution
1. The ground state will have no nodes, so we can pick the even part of the general solution
of the free Schr
odinger equation inside the well. Outside the well, square-integrability
demands the solutions to vanish at infinity. The wave-function for the ground state is
then
|x| < w

|x| > w
(x) = Ae|x|

(x) = cos kx

Both the wave-function and its derivative has to be continuous at the boundaries of
the well:
cos kw = Aew

:
d
:
dx

k sin kw = Aew

k tan kw =

Directly from Schrodingers equation:

|x| < w
E=

|x| > w

~2 k 2
2m

E =

~2 2
+ V0
2m

~2 k 2
~2 2
=
+ V0
2m
2m
From which we get the transcendental equation:

2mV0
k tan kw =
k2
~2

1/2

Let k denote the positive root of the equation above, and introduce the following
notation:

2mV0

kc =
and
kmax =
.
~
2w
Clearly, the LHS of the equation diverges at kmax ; and the RHS describes a circle with
radius kc , as shown in Fig. 4.
For the energy we have
E=

14

~2 k 2
2m

k tankw
kc

H L

*
k* * tan k* w

(kc2- k2)1/2

k* *

kc

kmax

Figure 4: Graphical representation of the solution of the transcendental equation

2. Write the result of part 1 in the non-dimensional form:

1/2
2mw2 V0
2
kw tan kw =
(kw)
~2
According to the condition given in the statement of the problem, the radius of the
circle on the RHS (that in Fig. 4) goes to infinity, therefore
k kmax
and
E

2
~2 kmax
~2 2
=
2m
8mw2

3. The potential barrier on the low-potential side of the well, denoted a in the figure, will
be finite (for any E), so the particle will eventually escape by the tunnel-effect.
4. E = 0, because the perturbation is odd (and therefore its integral with the square of
the ground-state wave-function vanishes).
5.
1
F =
~

Za
dx

p
2m(V (x) B)

V (x) = V0 eEx
B = V0 eEa

15

a=

V0 B
eE

V(x)

V0
a

F =

2m
~

Za
dx

(V0 B) eEx

2m 2
1
3/2
=

(V0 B eEx)
~ 3
eE
w

2m 2 1
3/2
(V0 B eEw)
=
~ 3 eE
6. Write the energy of the particle as
B=

1
mv 2 .
2

Then

2B
.
m
The time it takes for the particle to bounce back and forth once is
v2 =

T =

4w
,
v

so it hits the right wall with frequency

=

v
4w

v 2F
Probability to escape
=
e
unit time
4w

Lifetime

16

4w 2F
e
v

Quantum Mechanics Problem 2

Solution
1. Drop the t-label for simplicity. Then we have

cos
sin eit
H=B
,
sin eit
cos
and for the eigenvectors solve

cos
sin eit

sin eit
cos

x
y

x
y

with the normalization condition |x|2 + |y|2 = 1. From the vector-equation

y = eit

sin
x
cos 1

and with the normalization, we end up with

cos (/2)
|+i =
sin (/2)eit

|i =

sin (/2)
cos (/2)eit

2. Decompose the state-vector as

|i = c+ |+i + c |i
and write Schrodingers equation in terms of these vectors:
d
|i = H|i
dt

d
d
i~ c+ |+i + c+ |+i + c |i + c |i = B [c+ |+i c |i]
dt
dt
i~

or in the (|+i, |i) basis

"
# "
#"
#
d
d
c+
|+i
i~h+| dt
|i
c+
B i~h+| dt
d
i~
=
d
d
dt c
c
i~h| dt
|+i
B i~h| dt
|i
which, with the given concrete form of the vectors, is
"
# "
#"
#
c+
c+
B + ~ sin2(/2)
~ cos (/2) sin (/2)
d
i~
=
.
dt c
c
~ sin (/2) cos (/2)
B + ~ cos2(/2)
Now use the identities
1
(1 cos )
2
1
cos2 (/2) = (1 + cos )
2
1
sin (/2) cos (/2) = sin
2
sin2 (/2) =

17

to get
d
i~
dt

"

c+
c

1
=
2

"

#"

2B ~ cos

~ sin

~ sin

2B + ~ cos

c+
c

#
.

Note that in the last equation we dropped the part of the Hamiltonian that was proportional to the identity, since that gives only a time dependent phase that is identical
for the coefficients c , c+ . This we can rewrite in the form:
"
# "
#"
#
c+
Dz Dx
c+
d
i~
=
.
dt c
Dx Dz
c
with the solution
"

c+

(
= exp

= cos

i
~

"

~
|D|t
~

Dz

Dx

Dx

Dz

#) "

1
0

~ sin
iD

~
|D|t
~

!
.

And so

c+ = cos

|c+ | = cos

3. For B ~, Dz D, so

~
|D|t
~

~
|D|t
~

Dz
i
sin
~
|D|

18

~
|D|t
~

~
Dz2
2 |D|t
i 2 sin
D
~

|c+ |2 1

(Adiabatic theorem)

Statistical Mechanics and Thermodynamics Problem 1

Thermodynamics of a Non-Interacting Bose Gas
Solution
a)
np =

Ep =

e(Ep ) 1

p2
2m

At and below TBEC = 0. At exactly TBEC , there are no atoms in the condensate and
V
N=
2~3

d3 p
ep2 /(2m)

= (2~)

2m

3/2

Z
4
0

x2 dx
1
{z }

ex 2

=I1

n=

1
2 2

kTBEC =

2mkT
~2
2 2 n
I1

3/2

2/3

I1
~2
2m
(2 points)

b)

The above integral with = 0 also applies below TBEC , but it then gives the number
of non-condensed atoms. So on an isotherm below Vcritical
Nnon-condensed is constant
T is constant
p is constant
(think of the kinetic origin of pressure)

p
Classical Gas p 1/V
Bose Gas
Phase Transition Line
5/3
pBEC V

V
(2 points)

19

c)
V
U=
2~3

I2
= Nc
I1
cv =

p2
d3 p
3
= (2~) V
2
2m ep /(2m) 1

2m

2m

5/2

4
2m

x4 dx
ex2 1
0
| {z }
=I2

1
I2
= Nc kT T 5/2
2m
I1

5 I2
V
Nc k =
2 I1
2 2

2mkT
~2

3/2

5
k
I2 T 3/2
2
(2 points)

d)

From the reversibility of the Carnot-cycle:

dS1 = dS2

for 1 cycle

S1 = S2

for the entire process

S
U
3/2
dU = T dS pdV T
=
= cv = aT
| {z }
|{z}
T V
T V

from c)

=0

cv
dS
=
dT
T
Therefore the entropy transfer in the entire process is

ZT0
Si =

cv
dT = a
T

Ti

S1 + S2 = 0

ZT0
T 1/2 dT =

2 3/2
3/2
a T0 Ti
3

Ti

3/2

T0

1 3/2
3/2
T1 + T2
2

ZT0
Heat transferred to F1 :

Q1 =

T dS =

2 5/2
5/2
a T0 T1
.
5

T dS =

2 5/2
5/2
a T2 T0
.
5

T1

ZT2
Heat transferred from F2 :

Q2 =
T0

Therefore the total work done by the Carnot-machine is

W = Q2 Q1 =

2 5/2
5/2
5/2
a T1 + T2 2T0
5
(4 points)

20

Statistical Mechanics and Thermodynamics Problem 2

Phase Transition in a Superconductor
Solution
a)

S
Q
=
T
T H
T H

S
S
dS =
dT
+
dM
T M
M T

S
S M
S
=
+
T H
T M
M T T H
| {z }

cH

=0

where the last term is zero because M is independent of T . Then

S
Q
cH = T
=
cM
T M
T M
(2 points)

b)

The transition takes place at constant T and H. The thermodynamic function whose
variables are T and H is the Gibbs-potential:
dG = SdT M dH
Gsuper = Gnormal at every point on HC (T ), so dGS = dGN which we then write as
SS dT MS dH = SN dT MN dH
|{z}
=0

4
SN SS
dH
dHC
=
=
(SN SS )
=

dT trans.
dT
MS
V HC (T )
line

(3 points)

c)

By the third law S 0 as T 0. But the figure shows HC (T = 0) is finite. Therefore

dHC
0 as
dT

T 0.

The transition is second order where SN SS = 0, that is, the latent heat equals zero.
SN SS =

21

V
dHC
HC (T )
4
dT

 At T = 0 the transition is second order because both entropies go to zero.

At T = TC (H = 0) the transition is second order since HC (T ) = 0 and dHC /dT
is finite.
At all other temperatures the transition is first order since both HC (T ) and
dHC /dT are finite.
(2 points)

d)

Use H and T as variables

dS(H, T ) =

S
S
dH
+
dT
H T
T H

S
cH
=

H T
T

S
T H

Maxwell
relation

=0
T H

Z
S=

cH
a
dT = T 3 V
T
3
b
= T 3 V + T V
3

ba
SN SS =
T 3 V + T V
3

T < TC
T > TC
T = TC (H = 0)

ba
TC2
3

1/2
3
TC (H = 0) =
ba
=

(3 points)

22