Cornell University, Physics Department Fall 2013

PHYS-3341 Statistical Physics Prof. Itai Cohen

Solutions to Problem Set 2
David C. Tsang, Woosong Choi, Phil Kidd
2.1: Classical Particle in a 1-D Box
Reif 2.1: A particle of mass m is free to move in one dimension. Denote its
coordinate by x and its momentum by p. Suppose that this particle is conned
within a box so as to be located between x = 0 and x = L, and suppose that its
energy is known to lie between E and E + E. Draw the classical phase space
of this particle, indicating the regions of this space which are accessible to the
particle.
L
x
p
p(E)
p(E) + dP
-p(E)
-p(E) - dP
0
Figure 1: Solution to problem 2.1
Assuming the particle is conned classically to the box 0 x L, we have
E <
p
2
2m
< E +E
solving for p we dene

2mE < p <

_
2m(E +D) (1)
linearizing we see
p(E) =

2mE p =
_
m
2E
E (2)
2
1
2.2: Two Particles in a Box
Reif 2.2: Consider a system consisting of two weakly interacting particles,
each of mass m and free to move in one dimension. Denote the respective position
coordinates of the two particles by x
1
and x
2
, their respective momenta by p
1
and
p
2
. The particles are conned within a box with end walls located at x = 0 and
x = L. The total energy of the system is known to lie between E and E + E.
Since it is dicult to draw in four-dimensional phase space, draw seperately the
part of phase space involving x
1
and x
2
, and that involving p
1
and p
2
. Indicate
on these diagrams the regions of phase space accessible to the system.
L
x
1
p
1
p
2
p(E)
p(E)+dP
Figure 2: Solution for problem 2.2
Again, the classical particles are bound to the region 0 < x < L. Due to the weak coupling
between the particles, the momenta must obey
E <
1
2m
(p
2
1
+p
2
2
) < E +E (3)
which bounds the solution between two circles in momentum space,
2mE < p
2
1
+p
2
2
< 2m(E +E) (4)
with radii

2mE and
_
2m(E +E) =

2mE +
_
m/2EE. 2
2
2.3: Ensemble of Harmonic Oscillators
Reif 2.3: Consider an ensemble of classical one-dimensional harmonic oscilla-
tors.
(a) Let the displacement x of an oscillator as a function of time t be given by
x = Acos(t+). Assume that the phase angel is equally likely to assume
any value in its range 0 < < 2. The probability w()d that lies in
the range between and +d is then simply w()d = (2)
1
d. For any
xed time t, nd the probability P(x)dx that x lies between x and x + dx
by summing w()d over all angles for which x lies in this range. Express
P(x) in terms of A and x.
(b) Consider the classical phase space for such an ensemble of oscillators, their
energy being known to lie in the small range between E and E+E. Calcu-
late P(x)dx by taking the ratio of that volume of phase space lying in this
energy range and in the range between x and x + dx to the total volume
of phase space lying in the energy range between E and E + E. Express
P(x) in terms of E and x. by relating E to the amplitude A, show that the
result is the same as that obtained in part (a).
(a) We have
p(x)dx =

w()
|dx/d|
d
= 2
dx
2Asin(t +)
=
dx

A
2
x
2
(b) We have the energy as a function of the amplitude:
E =
p
2
2m
+
kx
2
2
=
kA
2
2
The equal energy (equal amplitude) contour, in phase space is an ellipse (see Fig 2.3.1
in Reif). If we make the transformation p
2
= p
2
/(mk), we get a circle as the equal
energy (amplitude) contour. A
2
= x
2
+p
2
. Now, the phase space volume lying between
E and E + E is represented by the area of a shell between A and A + A where A
is a function of E.
W(A)A = 2AA.
In order to calculate which portion of this shell lies between x and dx, we need to move
polar coordinates
cos =
x
A
, d =
dx
Asin
=
dx

A
2
x
2
.
3
Therefore, the area of the two parts of the shell that lies between x and x +dx is
W(x, A)dxA = 2AdA =
2AdxA

A
2
x
2
.
and the probability of being in this interval is
p(x)dx =
W(x, A)dxA
W(A)A
=
dx

A
2
x
2
4
2.4: Magnetization of Spins
Reif 2.4: Consider an isolated system consisting of a large number N of very
weakly interacting localized particles of spin
1
2
. Each particle has a magnetic
moment which can point either parallel of antiparallel to an applied eld H.
The energy E of the system is then E = (n
1
n
2
)H, where n
1
is the number
of spins aligned parallel to H and n
2
the number of spins aligned antiparallel to
H.
(a) Consider the energy range between E and E + E where E is very small
compared to E but is microscopically large so that E H. What is the
total number of states (E) lying in this energy range?
(b) Write down an expression for ln (E) as a function of E. Simplify this
expression by applying Stirlings formula in its simplest form
ln n! nln n n.
(c) Assume that the energy E is in a region where (E) is appreciable, i.e., that
it is not close to the extreme possible values NH which it can assume.
In this case apply a Gaussian approximation to part (a) to obtain a simple
expression for (E) as a function of E.
(a) This problem is like counting the number of states in a binomial distribution. Note
this is only counting the number of states. There are two possible states for each
of the spins, i.e. parallel and antiparallel. E = (n
1
n
2
)H can be written as
E = (2n
1
N)H using N = n
1
+n
2
. Thus, by counting the number of states for a
specic value of n
1
, we can directly relate it to the number of states within the energy
range. The number of states for n
1
is
(n
1
) =
N!
n
1
!(N n
1
)!
Now, (E, E+E) can be found by counting the number of n
1
within the energy range
E, E + E. Since E 2H, we can approximate the number to be |
E
dE/dn
|. Since
E = (2n
1
N)H,
dE
dn
= 2H. Thus,
(E, E +E) = (n
1
)|
1
dE/dn
|E =
N!
n
1
!(N n
1
)!
E
2H
where we can substitute n
1
=
1
2
(N
E
H
) to get
(E, E +E) =
N!
(N/2 E/2H)!(N/2 +E/2H)!
E
2H
.
5
(b) Using the result of (a),
ln (E) = ln N! ln(N/2 E/2H)! ln(N/2 + E/2H)! ln 2H
apply Stirlings formula to this we get
ln (E) = N ln N N ln 2H
N E/H
2
ln
N E/H
2
+
N E/H
2

N +E/H
2
ln
N +E/H
2
+
N +E/H
2
.
= N ln N ln 2H
N E/H
2
ln
N E/H
2

N +E/H
2
ln
N +E/H
2
(c) Although this is not a random walk, the formula of the number of states is proportional
to the random walk probability with p = q =
1
2
. In particular, if we put a factor of
= 2
N
in front of the expression and cancel it by adding a factor of p
n
q
Nn
=
1
2
N
,
then our distribution is identical to the binomial distrbution with the additional factor
of . Thus, using the Gaussian approximation (1.6.4) on the result of (a),
(E)dE =

2
e
(E

E)
2
/2
2
dE
=
2
N

2NH
e

E
2
2N(H)
2
dE

E = (p q)Nl = 0, = 2

Npql = 2
_
N
1
4
H =

NH, and = 2
N
has been used.
We can also show this starting from the result of (b). Assuming E is not close to
NH, i.e.
E
H
N, we can use ln(1 x) x for x 1. Then,
ln
N E/H
2
= ln
N
2
(1
E
HN
)
= ln
N
2
+ ln(1
E
HN
)
ln
N
2

E
HN
Using this in the equation for ln (E),
ln (E) = N ln N ln 2H
N E/H
2
(ln
N
2

E
HN
)
N +E/H
2
(ln
N
2
+
E
HN
)
= N ln 2 ln 2H
1
2N
(
E
H
)
2
(5)
This yields the Gaussian approximation
(E) =
2
N
2H
e

E
2
2N(H)
2
.
6
(Note: Two results have constant factor dierence because we used a simple form of
Sterlings approximation, thereby dropping some terms in the logarithmic scale.) 2
7
2.5: Exact and Inexact dierentials
Reif 2.6: Consider the innitesimal quantity
(x
2
y)dx +xdy dF (6)
(a) Is this an exact dierential?
(b) Evaluate the integrals
_
dF between the points (1, 1) and (2, 2) along the
straight line paths connecting the folllowing points:
(1, 1) (1, 2) (2, 2)
(1, 1) (2, 1) (2, 2)
(1, 1) (2, 2)
(c) Suppose that both sides of the dF equation are divided by x
2
. This yields
the quantity dG = dF/x
2
Is dG and exact dierential?
(d) Evaluate the integral
_
dG along the three pahts of part (b).
(a) For something to be an exact dierential it must be expressed as
dF = A(x, y)dx +B(x, y)dy =
F
x
dx +
F
y
dy
If this was the case then we would have

y
A
F
(x, y) =

2
F
xy
=

x
B
F
(x, y),
however we see instead that
A
F
y
=

y
(x
2
y)
= 1
B
F
x
=

x
x
= 1

A
F
y
=
B
F
x
hence dF cannot be an exact dierential.
(b) Evaluating
_
dF we see
F[(1, 1) (1, 2) (2, 2)] = 0 +
_
2
1
1dy +
_
2
1
(x
2
2)dx + 0 =
4
3
F[(1, 1) (2, 1) (2, 2)] =
_
2
1
(x
2
1)dx + 0 +
_
2
1
2dy =
10
3
F[(1, 1) (2, 2)] =
_
2
1
(x
2
x)dx +
_
2
1
ydy =
7
3
8
(c) We have
dG = (1
y
x
2
)dx +
1
x
dy
giving
A
G
y
=

y
(1
y
x
2
)
=
1
x
2
B
G
x
=

x
1
x
=
1
x
2

A
G
y
=
B
G
x
Hence dG is an exact dierential.
(d) Using the denition of dG,
G[(1, 1) (1, 2) (2, 2)] = 0 +
_
2
1
dy +
_
2
1
(1 2/x
2
)dx + 0 = 1
G[(1, 1) (2, 1) (2, 2)] =
_
2
1
(1 1/x
2
)dx + 0 +
_
2
1
dy/2 = 1
G[(1, 1) (2, 2)] =
_
2
1
(1 x/x
2
)dx +
_
2
1
dy/y = 1
The integral
_
dG does not vary between the three paths.
9
2.6: Quantum Particle in a Box
Reif 2.7: Consider a particle conned within a box in the shape of a cube of
edges L
x
= L
y
= L
z
. The possible energy levels of this particle are then given by
E =

2
2m

2
_
n
2
x
L
2
x
+
n
2
y
L
2
y
+
n
2
z
L
2
z
_
.
(a) Suppose that the particle is in a given state specied by particular values
of the three integers n
x
, n
y
, n
z
. By considering how the energy of this state
must change when the length L
x
of the box is changed quasistaticallly by a
small amount dL
x
, show that the force exerted by the particle in this state
on a wall perpendicular to the x-axis is given by F
x
= E/L
x
.
(b) Calculate explicitly the force per unit area (or pressure on this wall. By
averaging over all possible states, nd an expression for the mean pressure
on this wall. (Exploit the property that the average values n
2
x
= n
2
y
= n
2
z
must be equal by symmetry). Show that this mean pressure can be very
simply expressed in terms of the mean energy

E of the particle and the
volume V = L
x
L
y
L
z
of the box.
(a) We have
E =

2

2
2m
_
n
2
x
L
2
x
+
n
2
y
L
2
y
+
n
2
z
L
2
z
_
For a small change in distance L
x
, we have the change in energy
E =
E
L
x
L
x
= 2

m
n
2
x
L
3
x
L
x
. (7)
Which corresponds to work F
x
L
x
= W = E. Hence we have
F
x
=
E
L
x
= 2

m
n
2
x
L
3
x
. (8)
(b) To calculate the mean value of the pressure we rst see pressure is dened as
P =
F
x
A
x
=
F
x
L
y
L
z
the mean value is then

P =

F
x
L
y
L
z
=

2

2
m
n
2
x
L
3
x
1
L
y
L
z
To nd an expression for n
2
x
we notice

E =

2

2
2m
_
n
2
x
L
2
+
n
2
y
L
2
+
n
2
z
L
2
_
10
with L = L
x
= L
y
= L
z
. Noting that by symmetry n
2
x
= n
2
y
= n
2
z
,
n
2
x
=
2mL
2

E
3
2

2
(9)
Hence we have

P =
2
3

E
V
. (10)
where V = L
3
. 2
11
2.7: Wire Under Tension
Reif 2.9: The tension in a wire is increased quasi-statically from F
1
to F
2
. If
the wire has length L, cross-sectional area A, and Youngs modulus Y, calculate
the work done.
The force F required to stretch by L a wire with initial length L, cross-sectional area
A, and Youngs modulus Y is given by
F
A
= Y
L
L
(11)
The work done in increasing the force from F
1
to F
2
is
W =
_
pdV =
_
Fdx (12)
So, changing variables so that the integral runs from L
1
to L
2
, with L
i
=
L
Y A
F
i
, we
have
_
L
2
L
1
Y A
L
ldl =
Y A
2L
(L
2
2
L
2
1
) =
L
2Y A
(F
2
2
F
2
1
) (13)
12