Phys.

3341 Homework Solution Set #5

Yoav Kallus

1 Reif 3.3
(a) The number of states of energy E, as we worked out in last week’s problem
set is

N − E/µH N − E/µH N − E/µH
Ω(E) ≃ exp N log N − N − log + (1)
2 2 2

N + E/µH N + EµH N + E/µH
− log + . (2)
2 2 2

And in the limit that E ≪ µN H, this becomes

 2 !
N E
Ω(E) ≃ exp N log 2 − . (3)
2 N µH

Therefore, for the two weakly coupled systems described in the problem, the
combined number of states for a given partition of the energy is
 2  2 !
′ ′ N E N′ E′
Ω(E, E ) ≃ exp N log 2 + N log 2 − − . (4)
2 N µH 2 N ′ µ′ H

If E ′ = E0 − E, then this is maximized when the derivative of log Ω(E, E0 − E)

with respect to E is zero:
! !
1 Ẽ 1 Ẽ ′
− + ′ =0 (5)
µH N µH µ H N ′ µ′ H
Ẽ/N µ2
= ′2 . (6)
Ẽ ′ /N ′ µ

(b) Using (6) and the fact that E0 = Ẽ + Ẽ ′ = bN µH + b′ N ′ µ′ H, we obtain

N µ2
Ẽ = (bN µH + b′ N ′ µ′ H) (7)
N µ2 + N ′ µ′2
N ′ µ′2
Ẽ ′ = (bN µH + b′ N ′ µ′ H). (8)
N µ2 + N ′ µ′2

1
 (c) Since there is no work done (no change in external parameters), the heat
absorbed is just
N µ2
Q = Ef − Ei = (bN µH + b′ N ′ µ′ H) − bN µH (9)
N µ2 + N ′ µ′2
N µ2 ′ ′ ′ N ′ µ′2
= (b N µ H) − (bN µH) (10)
N µ2 + N ′ µ′2 N µ2 + N ′ µ′2
µb′ − µ′ b
= N N ′ µµ′ H . (11)
N µ2 + N ′ µ′2
(d) By the fundamental postulates of statistical mechanics, we know that
P (E, E0 − E) ∼ Ω(E, E0 − E) (12)
 2  2 !
N E N ′ E0 − E
∼ exp − . (13)
2 N µH 2 N ′ µ′ H
The argument in the exponent is quadratic (degree-2 polynomial) in E, so it
can be expressed as a(E − c)2 + d. By inspection we have that the coefficient of
E 2 is a = (1/2N µ2 H 2 ) + (1/2N ′ µ′2 H 2 ). The minimum of the quadratic, as we
figured out in parts (a+b) occurs at c = Ẽ. The constant term can be thrown
away, since the (12) is only a proportionality any way. So
!
(E − Ẽ)2
P (E) ∼ exp , (14)
2σ 2
where σ 2 = 1/2a. Alternatively, you could work out the algebra, complete the
square, and get the same answer. To normalize the distribution, we just recall
the form of the normalized Gaussian:
!
1 (E − Ẽ)2
P (E) = √ exp . (15)
2πσ 2 2σ 2
(e) The dispersion of a Gaussian is just given by σ 2 :
1
(∆∗ E)2 = σ 2 = (16)
(1/N µ H ) + (1/N ′ µ′2 H 2 )
2 2
′ 2 ′2 2
NN µ µ H
= . (17)
N µ2 + N ′ µ′2
(f) For the case when N ′ ≫ N , A′ acts as a thermal bath. From (17) we get
(∆∗ E)2 = N µ2 H 2 . (18)
From (7) we get
N µ2 b′ H
Ẽ = . (19)
µ′
When we combine (18) and (19), we get
|∆∗ E| µ′
= √ (20)
Ẽ N µb′

2
2 Reif 3.6
The partial pressure of Helium in the bulb will equilibrate with that in the
room, while the partial pressure of all other components will remain unchanged.
Therefore, the helium in the bulb will contribute a partial pressure of 1 atmo-
sphere, as will the total partial pressure of all the components of the air in the
bulb (helium is negligible as a component of air). The total pressure will be 2
atmospheres.

3 Elser 5.1
(a) For a random walk in three dimensions starting at the origin and ending at
r = (x, y, z),
N
X
r= Si . (21)
i=1

The x-component, on which we’ll focus is just

N
X
x= Sx,i . (22)
i=1

Since the individual steps are i.i.d. (identically and independently distributed),
the mean and standard deviations of x are just

x = N Sx (23)
(x − x)2 = N (Sx − Sx )2 . (24)

Each step S is uniformly distributed over a sphere of radius L. By symmetry of

reflection about yz-plane, we have that Sx = 0 and from symmetry of permuting
the three axes, we have that (Sx − Sx )2 = Sx2 = Sy2 = Sz2 = |S|2 /3 = L2 /3.
Therefore, (x − x)2 = x2 = N L2 /3.
We compare this value of to the value of x2 we get from the probability
distribution given: P (x, y, z) ∝ exp[−(x2 + y 2 + z 2 )/(2σ 2 )]. First, note that the
marginal probability density over x is just P (x) ∝ exp[−x2 /2σ 2 ], as x, y, and
z are independent (only true in the large-N limit). As this is a Gaussian, we
simply have x2 = σ 2 , and so σ 2 = N L2 /3.
(b) From Reif (3.12.5) we have that the force in the x-direction is given by
∂ log Ω
Fx = kB T (25)
∂x 
∂ log Ω −x2 − y 2 − z 2

= kB T (26)
∂x 2σ 2
kB T x
=− 2 . (27)
σ

3
Note that this is a tensive force (acting to contact the system), not an expan-
sive force, like we get for gases. This is of course what we expect since we
know rubber bands to be elastic, resisting expansion. We find that the force is
proportional to the extension of the rubber band, and the Hooke’s law constant
is temperature depndent: kB T /σ 2 . Counterintuitively, the hotter the rubber
band is, the stiffer it is.
(c) We forgot to include the kinetic contribution to the entropy, which is
due to the fact that the polymer strands are actually continually wriggling
about (see movie on course website). Therefore, if we expand the rubber band
adiabatically, the kinetic contribution to the entropy must grow to compensate
for the decrease in structural entropy. The result is that the polymer strands
will wriggle about more vigorously (exploring a greater region of momentum
space), as is clearly seen in the movie on the course website.