MASSACHUSETTS INSTITUTE OF TECHNOLOGY

DEPARTMENT OF PHYSICS
Academic Programs Phone: (617) 253-4851
Room 4-315 Fax: (617) 258-8319

DOCTORAL GENERAL EXAMINATION

WRITTEN EXAM — WITH SOLUTIONS
September 4, 2015

DURATION: 75 MINUTES PER SECTION

 MASSACHUSETTS INSTITUTE OF TECHNOLOGY
DEPARTMENT OF PHYSICS
Academic Programs Phone: (617) 253-4851
Room 4-315 Fax: (617) 258-8319

DOCTORAL GENERAL EXAMINATION

WRITTEN EXAM - CLASSICAL MECHANICS — WITH SOLUTIONS
September 4, 2015

DURATION: 75 MINUTES

1. This examination has two problems. Read both problems carefully be-
fore making your choice. Submit ONLY one problem. IF YOU SUBMIT
MORE THAN ONE PROBLEM FROM A SECTION, BOTH WILL BE
GRADED, AND THE PROBLEM WITH THE LOWER SCORE WILL
BE COUNTED.
2. If you decide at this time not to take this section of the exam, please
inform the faculty proctor. ONCE YOU BEGIN THE EXAM, IT WILL
BE COUNTED.
3. Calculators may not be used.
4. No books or reference materials may be used.
Classical Mechanics 1: Striking the sphere
A hard sphere of radius R is fixed at the center of a spherically symmetric potential U (r).
The potential U (r) declines monotonically to zero as r −→ ∞.
A beam of test particles is aimed at the sphere from a great distance. All the particles
are moving parallel to one another with speed v  c. The number density of particles is n
particles/cm3 , and each particle has a mass m.

(a) (4 pts) Calculate the cross-section for striking the sphere, in terms of U (R), v, and m.

(b) (2 pts) Is it possible for the cross-section to vanish? If so, under what conditions?

(c) (4 pts) Now take U (r) = −GM m/r, the gravitational potential between the sphere of
mass M and the test particle of mass m.
Whenever a particle strikes the sphere, the particle sticks to the sphere and increases
the sphere’s mass but without appreciably changing the radius. Calculate the time
required for the sphere’s mass to increase from an initial value of Mi to a final value
of Mf .

1
• Solutions by J. Winn (August 2015). An earlier version of this problem also appeared on
the Fall 1992 exam.

(a) The figure below illustrates the case of a repulsive potential, although the result is the
same for an attractive potential. Consider the particles with impact parameter b just
large enough for the point of closest approach to equal R. All particles with smaller
impact parameters will collide with the sphere.

To figure out b in terms of given quantities, use the conservation of energy,

1 2 1 2
mv = mv0 + U (R), (1)
2 2
as well as the conservation of angular momentum,

mvb = mv0 R, (2)

giving v0 = v(b/R). Combining these results,

 2
1 2 1 2 b
mv = mv + U (R) (3)
2 2 R
 2
b 2U (R)
1 = + (4)
R mv 2
 
2 2 2U (R)
b = R 1− . (5)
mv 2

The cross-section for hitting the sphere is therefore

 
2 2 2U (R)
σ = πb = πR 1 − . (6)
mv 2

2
(b) From our previous answer we see that for the cross-section to vanish, U (R) must be
positive. Specifically, σ vanishes when
1 2
mv = U (R), (7)
2
and the particles with b = 0 have just enough energy to touch the sphere.

(c) In time dt, the number of particles hitting the sphere is nvσdt. Therefore

dM
= mnvσ (8)
dt  
dM 2 2GM m
= mnvπR 1 + (9)
dt mv 2 R
 
dM M
= K 1+ , (10)
dt M0

where we have defined K ≡ mnvπR2 and M0 ≡ v 2 R/2G. Now we may integrate:

dM
= Kdt (11)
1 + M/M0
Z Mf
dM
= Kt (12)
Mi 1 + M/M0
M0 ln (M0 + Mf ) − M0 ln (M0 + Mi ) = Kt (13)
 
M0 M0 + Mf
t = ln (14)
K M0 + Mi
 2 
v v R/2G + Mf
t = ln . (15)
2πGmnR v 2 R/2G + Mi

3
Classical Mechanics 2: Particle sliding on a rotating circular wire
A pointlike particle of mass m is constrained to move on a circular wire of radius R. The
particle can slide without friction. The circular wire spins with constant angular speed ω
about the vertical diameter. The force of gravity mg is acting on the particle.

(a) (2 pts) Write the Lagrangian for the system using θ as the generalized coordinate.
Identify an effective potential V (θ).

(b) (2 pts) Write down the Euler Lagrange equation and derive the equation of motion in
terms of m, g, R, ω, and θ. (Do not solve it.)

(c) (2 pts) Find constant values θi , i = 1, 2, ..., for which θ(t) = θi is a stationary solution
of the equation of motion. Express your answer in terms of ω, R, and g. Do all solutions
exist for all values of ω?

(d) (4 pts) Consider now small displacements from each of the equilibrium values θi iden-
tified in part (c). Determine, as a function of ω whether such displacements lead to
stable or unstable oscillations. If the small-amplitude oscillation is stable, determine
the corresponding oscillation frequency Ωi . Summarize your results in a graph where
you plot the θi as functions of ω and label the various parts of the curves as ”stable”
or ”unstable”.

4
• Solutions by Markus Klute (August 2015) An earlier version of this problem also appeared
on the Spring 2009 exam.

(a) The velocity ~v of the particle is

~v = R θ̇ ~eθ + ωR sin θ ~eφ . (1)

The Lagrangian L is then given by

1  2
L = T − V = m R θ̇ ~eθ + ωR sin θ ~eφ − mgR (1 − cos θ) . (2)
2
Simplifying and dropping constants
1 1 1
L = m R2 θ̇2 + m ω 2 R2 sin2 θ + mgR cos θ = m R2 θ̇2 − V (θ) , (3)
2 2 2
with
1
V (θ) = − m ω 2 R2 sin2 θ − mgR cos θ . (4)
2
(b) The Euler Lagrange equation  
d ∂L ∂L
= (5)
dt ∂ θ̇ ∂θ
gives
1
m R2 2θ̈ = −mgR sin θ + mω 2 R2 sin θ cos θ . (6)
2
Canceling out constants results in
g
θ̈ = − sin θ + ω 2 sin θ cos θ . (7)
R

(c) If θ is constant then θ̈ = 0, so from the part (b) of this problem we get
 g
sin θ ω 2 cos θ − = 0. (8)
R
There are three solutions

θ1 = 0 , (9)
 g 
θ2 = cos−1 , (10)
Rω 2
θ3 = π. (11)

The solutions θ1 and θ3 exist for all values of ω. The solution θ2 exists for
g
ω2 > ≡ ω02 . (12)
R

5
(d) From the equation of motion we find

∂L ∂V
m R2 θ̈ = =− . (13)
∂θ ∂θ
∂V
Around the equilibrium points θi we write θ = θi +  and from |
∂θ θi
we find

∂ 2V
m R2 ¨ = − . (14)
∂θ2 θi

The oscillation frequency Ωi is therefore

1 ∂ 2V
Ω2i = . (15)
mR2 ∂θ2 θi

Using the potential from part (a) gives

∂ 2V
= −m ω 2 R2 (cos2 θ − sin2 θ) + m g R cos θ . (16)
∂θ2
We therefore find
Ω2i = −ω 2 (2 cos2 θi − 1) + ω02 cos θi . (17)

For θ1 = 0 we find
Ω2i = ω02 − ω 2 . (18)
This point is stable for ω < ω0 and unstable for ω > ω0 .

ω02
For θ2 satisfying cos θ2 = ,
we find
ω2
 4
ω04 ω04

2 2 2ω0 2
Ωi = −ω − 1 + = ω − . (19)
ω4 ω2 ω2

This is stable for ω > ω0 .

For θ3 = π we find
Ω2i = −ω02 − ω 2 < 0 , (20)
which is always unstable.

6
7
 MASSACHUSETTS INSTITUTE OF TECHNOLOGY
DEPARTMENT OF PHYSICS
Academic Programs Phone: (617) 253-4851
Room 4-315 Fax: (617) 258-8319

DOCTORAL GENERAL EXAMINATION

WRITTEN EXAM - ELECTRICITY AND MAGNETISM — WITH SOLUTIONS
September 4, 2015

DURATION: 75 MINUTES

1. This examination has two problems. Read both problems carefully be-
fore making your choice. Submit ONLY one problem. IF YOU SUBMIT
MORE THAN ONE PROBLEM FROM A SECTION, BOTH WILL BE
GRADED, AND THE PROBLEM WITH THE LOWER SCORE WILL
BE COUNTED.
2. If you decide at this time not to take this section of the exam, please
inform the faculty proctor. ONCE YOU BEGIN THE EXAM, IT WILL
BE COUNTED.
3. Calculators may not be used.
4. No books or reference materials may be used.
Electromagnetism 1: Magnetic mirror
When a charged particle orbits around magnetic field lines while also drifting along the
field into a region of higher field strength, then the particle experiences a force that reduces
the component of velocity parallel to the field. This force slows the motion along the field
lines and may reverse it. This is the basis of a “magnetic mirror,” illustrated below.

In this problem we will investigate this phenomenon using a cylindrical coordinate system
in which the z-axis is the symmetry axis, r denotes the cylindrical radius (the distance from
the z-axis), and φ is the azimuthal angle measured from the x-axis, as illustrated below.

You may find it useful to remember

∇ ~ = 1 ∂ (rAr ) + 1 ∂Aφ + ∂Az .

~ ·A (1)
r ∂r r ∂φ ∂z

1
 ~ which is axially symmetric around the z-axis. The z-
Consider a magnetic field B
component is
Bz (r, z) = B0 + B00 z, (2)
where B00 is a constant.
We inject a particle with mass m and charge q > 0 at point A, located at x = r0 and
y = z = 0. The particle’s initial velocity has both horizontal and vertical components:

~v = −vh0 ŷ + vz0 ẑ, (3)

q
where vh0 = Br
mc 0 0
and vz0  vh0 . Both components of the velocity are non-relativistic.

(a) (1 pt) Calculate the radial component of the magnetic field, Br (r, z).

(b) (1 pt) Show that throughout the subsequent motion,

v 2 (t) = vh0
2 2
+ vz0 . (4)

Now assume that during each orbit around the z-axis, the horizontal velocity is nearly
in the −φ̂ direction and the change in radius ∆r of the orbit is negligible compared to the
instantaneous radius r.

(c) (1 pt) What is the radius r of the orbit as a function of vφ , Bz and physical constants?

(d) (2 pts) Find an equation for the z-dependence of vh (the horizontal speed) as the
particle drifts in the z-direction. One way to do so is to write the equation of motion
for the φ̂ component of the velocity, and then use vz dt = dz.

(e) (3 pts) Using your result from part (c), integrate your equation from part (d) to show
that the horizontal speed varies with z as
s
vh (z) Bz (z)
= . (5)
vh0 B0

(f) (2 pts) Find the value of z for which vz = 0. This is the reflection point of the
magnetic mirror, where the spiraling particle stops its upward motion and starts moving
downward.
Express your answer entirely in terms of B0 , B00 , vh0 and vz0 .

2
• Solutions by J. Winn (August 2015), based on a version of this problem that appeared in
the Spring 1986 exam.

~ to vanish,
(a) Requiring the divergence of B
1 ∂ ∂Bz
(rBr ) + = 0 (6)
r ∂r ∂z
∂
(rBr ) = −B00 r (7)
∂r
r2
rBr = − B00 + constant. (8)
2
Note that the constant of integration must be zero, for Br to remain finite at r = 0.
Therefore
rB 0
Br = − 0 . (9)
2
(b) The only force on the particle is the Lorentz force, which cannot change the kinetic
energy of the particle:
q
F~ = ~v × B ~ (10)
c
dEK
= F~ · ~v = 0 (11)
dt
1 1
EK = mv 2 = m(vh2 + vz2 ) = constant (12)
2 2
from which the desired result follows.

(c) Since the motion is nearly circular, we have

mvh2 qvh Bz
= , (13)
r c
which gives
mvh c
r= . (14)
qBz
(d) The φ component of the Lorentz equation of motion is
dvφ q
Fφ = m = (vz Br − vr Bz ) , (15)
dt c
and since we are assuming a nearly circular orbit, vr ≈ 0. Thus
dvφ q q rB 0
= vz Br = − vz 0 . (16)
dt mc mc 2
Dividing both sides by vz , and using vh = −vφ (the particle circulates in the −φ
direction) we obtain
dvh dvh q rB00
= = . (17)
vz dt dz mc 2

3
(e) Replace r by mvφ c/qBz in the preceding equation, giving

dvh q mvh cB00

= (18)
dz mc 2qBz
1 dvh 1 B00
= . (19)
vh dz 2 Bz

We integrate to find vh (z):

vh
1 0 z
Z Z
dvh dz
= B0 (20)
vh0 vh 2 B0 + B00 z
  0 
vh 1 Bz (z)
ln = ln (21)
vh0 2 B0
s
vh (z) Bz (z)
= . (22)
vh0 B0

(f) Denote by zr the height of the reflection point, at which vz (zr ) = 0. From part (b),

[vh (zr )]2 = vh0

2 2
+ vz0 . (23)

Now we make use of our result from part (e), by writing

 2  2
vh (zr ) vz0
= 1+ (24)
vh0 vh0
2
B0 + B00 zr

vz0
= 1+ (25)
B0 vh0

allowing us to solve for zr :   2

B0 vz0
zr = . (26)
B00 vh0

4
Electromagnetism 2: Electromagnetic waves in a plasma
A plane electromagnetic wave of angular frequency ω propagates in a uniform plasma
with electron density Ne . The plasma is locally neutral, ρ = 0. The electromagnetic wave
generates periodic currents within the plasma that, as the problem will show, modify the
index of refraction of the medium compared to that of the vacuum.
Assume the plasma has no resistivity and neglect radiation pressure effects as well as
currents due to the ions.

~ r, t) in the plasma to the wave’s electric field E(~

(a) (3 pts) Relate the current J(~ ~ r, t) or
derivatives thereof. Assume magnetic forces can be neglected.

(b) (3 pts) Write down the appropriate Maxwell equations and derive the wave equation.
Find the dispersion relation ω(k) and the lowest frequency electromagnetic wave that
can propagate the plasma.

(c) (2 pts) Find the phase and group velocity for electromagnetic waves in the plasma.
Compare those velocities with c, the speed of light in vacuum.

(d) (1 pt) Find the index of refraction n of the plasma as a function of frequency.

(e) (1 pt) If a plane electromagnetic wave is incident on a plane interface between a vacuum
and the plasma, what is the critical angle for total reflection, measured from the normal
to the interface?

5
• Solutions by Markus Klute (August 2015) An earlier version of this problem also appeared
on the Fall 2010 exam.

(a) As stated in the problem, we ignore the ions and focus on the electrons. The motion
of an electron at location ~r is
m ~s¨ = −eE(~
~ r, t) , (1)
where ~s is the small displacement vector of the electron, lying along the direction of
the electric field. Because of the harmonic nature of the wave, the solution for ~s will
also be harmonic with angular frequency ω so that ~s¨ = −ω 2~s. We thus get
e ~
~s(~r, t) = E(~r, t) . (2)
mω 2
It follows that the velocity of the electron is given by
~ r, t)
e ∂ E(~
~v (~r, t) = . (3)
mω 2 ∂t
The overall current density in the plasma at ~r is given by
~ r, t)
e2 Ne ∂ E(~
J~ = −eNe v(~r, t) = − . (4)
mω 2 ∂t

(b) In Gaussian units the appropriate Maxwell equations are

~ ·B
∇ ~ = 0 (5)
~ ·E
∇ ~ = 4πρ = 0 (6)
~
∇ ~ = 4π J~ + 1 ∂ E
~ ×B (7)
c c ∂t
~
∇ ~ = − 1 ∂B
~ ×E (8)
c ∂t
Note that ρ is taken to be zero due to the condition of local charge neutrality. We now
use the results for J~ to simplify Maxwell’s equations for the plasma.
~
∇ ~ = − 1 ∂B
~ ×E (9)
c ∂t
2 ~ ~
∇ ~ = − 4πe Ne ∂ E + 1 ∂ E
~ ×B (10)
m c ω 2 ∂t c ∂t
 2
 ~
1 4πe Ne ∂ E
= 1− (11)
c m ω2 ∂t

We can combine these two equations, while making use of ∇~ ·E

~ = 0, to find the wave
equation in the plasma

4πe2 Ne 1 ∂ 2 E
 ~
2~
∇ E = 1− . (12)
m ω2 c2 ∂t2

6
 We can write this as   ~
ωp2 1 ∂ 2 E
2~
∇ E = 1− 2 , (13)
ω c2 ∂t2
where ωp is the plasma frequency
4πe2 Ne
   2 
2 e Ne
ωp ≡ or . (14)
m cgs 0 m mks
The wave equation admits to plane wave solutions with dispersion relation
ωp2 ω 2
 
2
k = 1− 2 , (15)
ω c2
or equivalently, q
ω= ωp2 + k 2 c2 . (16)
Since k has no real solution for ω < ωp , we conclude that the propagating waves must
have a frequency above the plasma frequency.
(c) The phase and group velocities are given by
ω(k) dω
vp = and vg = . (17)
k dk
We thus find r
c kc2 ωp2
vp = q and vg = =c 1− 2 . (18)
ωp2 ω ω
1− ω2
The group velocity is less than c and the phase velocity larger than c.
(d) The index of refraction n is defined from the phase velocity as vp = nc . We thus find
r
ωp2
n= 1− 2 . (19)
ω
Note that n < 1.
(e) A plane wave incident on a planar boundary with a medium having an effective index
of refraction n is subject to Snell’s law
nvac sin θinc = nplasma sin θref r , (20)
where the angles are measured with respect to the direction normal to the interface.
For our problem, this results in
sin θinc = n sin θref r . (21)
Since the index of refraction n of the medium is less than unity the refracted angle
reaches 90◦ before the incident angle reaches 90◦ . Thus, the critical incident angle,
after which the incident radiation is totally reflected, is
sin θcritical = n . (22)

7
 MASSACHUSETTS INSTITUTE OF TECHNOLOGY
DEPARTMENT OF PHYSICS
Academic Programs Phone: (617) 253-4851
Room 4-315 Fax: (617) 258-8319

DOCTORAL GENERAL EXAMINATION

WRITTEN EXAM - STATISTICAL MECHANICS — WITH SOLUTIONS
September 4, 2015

DURATION: 75 MINUTES

1. This examination has two problems. Read both problems carefully be-
fore making your choice. Submit ONLY one problem. IF YOU SUBMIT
MORE THAN ONE PROBLEM FROM A SECTION, BOTH WILL BE
GRADED, AND THE PROBLEM WITH THE LOWER SCORE WILL
BE COUNTED.
2. If you decide at this time not to take this section of the exam, please
inform the faculty proctor. ONCE YOU BEGIN THE EXAM, IT WILL
BE COUNTED.
3. Calculators may not be used.
4. No books or reference materials may be used.
Statistical Mechanics 1: Potts Model
The q-state Potts model generalizes the Ising model. There is a variable σi ∈ {1, 2, . . . , q}
at each lattice site. The Hamiltonian is given by a sum over nearest neighbors:
3J X
HPotts = − δσi ,σj . (1)
2
hi,ji

There are N lattice sites. Assume J > 0. For parts (a) and (b), q ≥ 2 is arbitrary, and for
(c) and (d), we assume q = 3.

(a) (1 pt) What is the entropy of the system at T = 0?

(b) (3 pts) For this part only, suppose the N sites are arranged on a line with open bound-
ary conditions. Write down the (exact) free energy.

(c) (1 pt) For the q = 3 case, show that the model is equivalent to
X
H = −J ~si · ~sj (2)
hi,ji

with each ~si restricted to take values in the set

     
1 −1/2
√ −1/2
√
~si ∈ , , (3)
0 3/2 − 3/2
P
(d) (5 pts) Let m~ denote the mean-field magnetization i ~si /N . Suppose that each site
in the lattice interacts with z other sites. Use a mean-field approximation to derive
an expression for the free energy. (Your answer should be left in terms of a solution
to a transcendental equation.) In other words, replace the true nearest-neighbor inter-
actions with an approximation in which every site interacts with every other site with
an interaction strength rescaled appropriately. Is there a first-order (i.e. discontinuous
in m) phase transition?
[Hint: it may be helpful to consider m
~ of the form (m, 0).]

1
• Solutions by Aram Harrow (August 2015)

(a) kB ln(q)

(b) X
Z= Mσ1 ,σ2 Mσ2 ,σ3 · · · MσN −1 ,σN , (4)
σ1 ,...,σN

where Mi,i = e−β3J/2 and Mi,j = 1 for i 6= j. Let I denote the q × q identity matrix
√
and u = 1/ q, where 1 is the all-ones vector. Then

M = quuT + (e−β3J/2 − 1)I,

and so M N = (e−β3J/2 + q − 1)N uuT + (e−β3J/2 − 1)N (I − uuT ). We compute

X
Z= (M N )i,j = quT M N u = q(e−β3J/2 + q − 1)N ,
i,j

and so
F = −kB T (ln(q) + N ln(e−β3J/2 + q − 1)). (5)

(c) The inner product ~si · ~sj is either 1 (when i = j) or −1/2 (when i 6= j). This yields
the same Hamiltonian up to an overall additive constant.

(d) We can replace the sum over hi, ji with Nz times the sum over all pairs i, j. Then we
get
z
E = −J N 2 m ~ ·m~ = −JzN |m|~ 2.
N
~ = (m, 0) then the energy is E = −JzN m2 .
If m
The entropy in this case can be determined from the populations N1 , N2 , N3 . Let
Ni = N pi . We can determine p1 , p2 , p3 by solving the linear system of equations
p2 + p3
p1 − =m (6)
2
p2 − p3 = 0 (7)
p1 + p 2 + p3 = 1 (8)
(9)
1+2m 1−m
which have solution p1 = 3
, p2 = p3 = 3
. The entropy is
3  
X N kB 3 2(1 − m) 3
S = N kB pi ln(1/pi ) = (1 + 2m) ln + ln . (10)
i=1
3 1 + 2m 3 1−m

Thus the free energy is

  
2 kB T 3 2(1 − m) 3
F = E − T S = N −Jzm − (1 + 2m) ln + ln .
3 1 + 2m 3 1−m
(11)

2
To calculate the derivative, observe that
 
3 2
~
X ∂~p 1
∇S(p 1 , p2 , p3 ) = (− ln(pi ) − 1)ei and = −1 , (12)
∂m 3
i=1 −1
dS
so that dm
= N kB 32 ln 1+2m
1−m
.
The equilibrium value of m then satisfies
1 dF 2 1−m
0= = −2Jzm − kB T ln , (13)
N dm 3 1 + 2m
or equivalently
1−m
= exp(−3βJzm), (14)
1 + 2m
where β ≡ 1/kB T .
There is always a solution at m = 0 corresponding to the disordered state. Whether
there is a solution at m > 0 depends on the value of βJz. To see this, we evaluate the
second derivative of F , which is proportional to
1 2
−3βJz − + .
1 − m 1 + 2m
At m = 0 this is 1 − 3βJz which is negative for βJz > 1/3. When this occurs there is
spontaneous order. This means that there is a first-order phase transition.

3
Statistical Mechanics 2: Magnetic susceptibility of a spinful Boltzmann gas
Consider a classical ideal gas of N spin-1/2 atoms moving in a container of volume V .
In the presence of a weak external magnetic field H the energy of the nth such atom may
be taken to be
p~2
E(~pn , σn ) = n − γHσn (1)
2m
Here σn = ±1 describes the two possible spin orientations of the atom and p~n is the momen-
tum of the atom. γ is a positive constant.

(a) (2 pts) Calculate the change in free energy due to the magnetic field.

(b) (3 pts) Calculate the average magnetization per atom hM i = N1 γh n σn i.

P

(c) (3 pts) Calculate the variance in the magnetization hM 2 i − hM i2 .

dhM i
(d) (2 pts) Use the above to calculate the magnetic susceptibility χ = dH
. How is the
susceptibility related to the variance?

4
• Solutions by Senthil Todadri (August 2015)

(a) Partition function Z is given by

  !N
~2
d3 p X −β
Z p
1 N 2m
−γσH
Z= V e (2)
N! 2π~ σ=±1

The translational and spin degrees of freedom are decoupled. Therefore the total
partition function factorizes as
Z = Ztrans Zs (3)
with Ztrans independent of H (the partition function of an ideal gas of spinless atoms),
and
Zs = (2 cosh(βγH))N (4)
The free energy
N
F = Ftrans − ln(2 cosh(βγH)) (5)
β
(with Ftrans = − β1 ln Ztrans ).

(b) The average magnetization per atom is obtained as

1 ∂F
hM i = − (6)
N ∂H
This gives
hM i = γtanh(βγH) (7)

(c) The variance is obtained from the partition function through

1 ∂ 2Z
hM 2 i − hM i2 = (8)
N 2 β 2 ∂H 2
1 ∂F
= − 2 (9)
N β ∂H
2
γ
= Sech2 (βγH) (10)
N

(d) Differentiatiing hM i with respect to field we get the magnetic susceptibility

χ = βγ 2 Sech2 (βγH) (11)

We also have
kB T χ
hM 2 i − hM i2 = (12)
N

5
 MASSACHUSETTS INSTITUTE OF TECHNOLOGY
DEPARTMENT OF PHYSICS
Academic Programs Phone: (617) 253-4851
Room 4-315 Fax: (617) 258-8319

DOCTORAL GENERAL EXAMINATION

WRITTEN EXAM - QUANTUM MECHANICS — WITH SOLUTIONS
September 4, 2015

DURATION: 75 MINUTES

1. This examination has two problems. Read both problems carefully be-
fore making your choice. Submit ONLY one problem. IF YOU SUBMIT
MORE THAN ONE PROBLEM FROM A SECTION, BOTH WILL BE
GRADED, AND THE PROBLEM WITH THE LOWER SCORE WILL
BE COUNTED.
2. If you decide at this time not to take this section of the exam, please
inform the faculty proctor. ONCE YOU BEGIN THE EXAM, IT WILL
BE COUNTED.
3. Calculators may not be used.
4. No books or reference materials may be used.
Quantum Mechanics 1: A spin-1/2 particle in a magnetic field
A spin-1/2 particle interacts with a magnetic field via the Hamiltonian
     
~ 0 1 0 −i 1 0
H = ~σ · B, σx = , σy = , σz = . (1)
1 0 i 0 0 −1

~
(We absorb the other relevant constants into the definition of B.)

(a) (5 pts) Find the eigenvalues and eigenvectors of H. You may find it convenient to
~ as
write B  
B sin(θ) cos(φ)
~ =  B sin(θ) sin(φ)  .
B
B cos(θ)

~ = B0 ẑ. Suppose at time t = 0 the spin is initially pointing in the +x̂

(b) (5 pts) Let B
direction. After time t = T , the spin is measured along the ŷ direction. What are the
possible outcomes and what are their probabilities?

1
• Solutions by Aram Harrow (August 2015)

(a) Observe that the Pauli matrices satisfy the multiplication rule {σi , σj } = 2δi,j I, where
{X, Y } = XY + Y X is the anticommutator and I is the 2 × 2 identity matrix. From
this we calculate
3
1
2 1X X
H = {H, H} = Bi Bj {σi , σj } = Bi Bj δi,j = B 2 .
2 2 i,j=1 i,j

Thus H has eigenvalues in the set {±B}. Since trH = 0, H must have one eigenvalue
equal to B and one equal to −B. To calculate these we need to solve
cos(θ) sin(θ)e−iφ
    
x x
B = ±B ,
sin(θ)eiφ − cos(θ) y y
for x, y. Assume without loss of generality that x is real (since we can always add an
overall phase to make this true) and that sin(θ) 6= 0 (by continuity). Then from
±x = cos(θ)x + sin(θ)e−iφ y,
we find that e−iφ y is also real. We can thus write x = cos(α) and y = sin(α)eiφ . The
eigenvalue equation then becomes
     
cos(α) cos(θ) cos(α) + sin(θ) sin(α) cos(θ − α)
± = = .
sin(α)eiφ (sin(θ) cos(α) − cos(θ) sin(α))eiφ sin(θ − α)eiφ
The +B eigenstate then has α = θ/2 while the −B eigenstate has α = π + θ/2, and
so the corresponding eigenvectors are
   
cos(θ/2) − sin(θ/2)
|+Bi = and |−Bi = (2)
sin(θ/2)eiφ cos(θ/2)eiφ

(b) The state after time T is

1 e−iBT /~ 1 e−iω0 T
     
−iHT /~ 1 1
e √ =√ iBT /~ =√ iω T ,
2 1 2 e 2 e 0
where we have defined ω0 = B/~.  The
 possible outcomes after a ŷ measurement are
1
±ŷ, corresponding to vectors √12 . The corresponding probabilities are
±i
2
e−iω0 T ± ieiω0 T
Pr[±ŷ] = .
2
These cases can be separately calculated as
π π 2
e−iω0 T +i 4 + eiω0 T −i 4
Pr[+ŷ] = = sin2 (ω0 T − π/4) (3)
2
π π 2
e−iω0 T −i 4 + eiω0 T +i 4
Pr[−ŷ] = = sin2 (ω0 T + π/4) (4)
2

2
Quantum Mechanics 2: Two interacting fermions
Consider two identical fermions of mass m interacting with each other through an at-
tractive harmonic potential. The Hamiltonian is

p21 p2 k
H= + 2 + (x1 − x2 )2 (1)
2m 2m 2
x1,2 are the coordinates of the two fermions, and p1,2 are the conjugate momenta. For
simplicity assume that the spins of both fermions are polarized in the same direction (e.g,
in the ↑ direction) so that we may ignore the spin degree of freedom.

(a) (2 pts) State the restriction imposed by Fermi statistics on acceptable wave functions
ψ(x1 , x2 ) of this system.

(b) (2 pts) Rewrite H in terms of center-of-mass and relative coordinates.

(c) (3 pts) Ignoring the restriction imposed by Fermi statistics what is the ground state
energy? What is the energy of the first excited bound state?

(d) (3 pts) Including the effects of Fermi statistics what is the ground state energy? What
is the degeneracy of the ground state?

3
• Solutions by Senthil Todadri (August 2015)

(a) The wave function must be antisymmetric under exchange: ψ(x1 , x2 ) = −ψ(x2 , x1 ).

(b) Introduce center-of-mass R = x1 +x

2
2
, and relative coordinates x = x1 − x2 . The
corresponding conjugate momenta are P = p1 + p2 and p = p1 −p 2
2
. In terms of these
the Hamiltonian is
P2 p2 kx2
H= + + (2)
4m m 2
(c) As P commutes with the rest of the Hamiltonian we can set fix it to determine the
eigen-energies. The ground state and the lowest energy excited bound states will have
P = 0. The relative
q motion is described by a 3d simple harmonic oscillator at a
2k
frequency ω = m
.
Therefore ground state energy is 23 ~ω and first excited bound state is at 52 ~ω. The
ground state is singly degenerate and the excited bound state is triply degenerate.

(d) The antisymmetry condition means that the orbital angular momentum L of the rela-
tive coordinate must be odd. In the ground state of the 3d oscillator, L = 0 and hence
this is not acceptable once we impose Fermi statistics. The first excited bound states
have L = 1, and hence they are acceptable states. Thus for fermions, the ground state
energy is 52 ~ω, and the ground state is three fold degenerate.