Problems for the course FYS4130

Yuri Galperin
February 19, 2013
Introduction to statistical methods. Simple
random walks
0.1. Simple calculations with probabilities
Three people shoot the same target, partial probabilities to hit the target being p
1
, p
2
, and p
3
,
respectively.
1. Find the probability that no one will hit the target.
2. Find the probability to nd at least one bullet in the target.
3. Find the distribution function P
n
to nd n bullets in the target and check that
n
P
n
= 1.
4. Find average value of the number of bullets in the target, n, and mean square deviation,
(n)
2
(n n)
2
.
5. Find numerical values of P
n
, n, and (n)
2
for p
1
= 0.8, p
2
= 0.9, p
3
= 0.7.
Solution 0.1: Independent event are those in which a shooter misses the target. Then the solu-
tions are:
1. P
0
= (1p
1
)(1p
2
)(1p
3
).
2. P
n1
= 1P
0
= 1(1p
1
)(1p
2
)(1p
3
).
3. The distribution has the form:
P
0
= (1p
1
)(1p
2
)(1p
3
),
P
1
= p
1
(1p
2
)(1p
3
) + p
2
(1p
3
)(1p
1
) + p
3
(1p
1
)(1p
2
),
P
2
= p
1
p
2
(1p
3
) + p
2
p
3
(1p
1
) + p
3
p
1
(1p
2
),
P
3
= p
1
p
2
p
3
.
4. n =
3
n=1
p
n
, (n)
2
=
3
n=1
p
n
(1p
n
).
2
5. Numerical values are: P
0
= 0.006, P
1
= 0.092, P
2
= 0.398, P
3
= 0.504, n = 2.4, (n)
2
=
0.46.
The numerical results are obtained with the program Maple 6, see the le.
0.2. One-dimensional random walk. Binomial distribution
Consider a asymmetric random walk with the probability p for a hop to the right and q =1p
for a hop to the left. The probability W
N
(m) for for m hops to the right from total number of hops
N is given by the binomial distribution (??)
W
N
(m) =
N!
m!(Nm)!
p
m
(1p)
Nm
.
1. Find the probability P
N
(M) for a total displacement M after N hops.
2. Find this probability for a symmetric case, p = 1/2.
3. Calculate average displacement

M after N hops. Hint: it is easier to use the equality

M = 2 mN and calculate m.
4. For the same situation calculate the dispersion (M)
2
= (M

M)
2
).
5. Compare

M
_
(M)
2
and

M. What happens for symmetric random walk?
6. Two drunks start out together at the origin, each having equal probability of making a step
to the left or to the right along the x axis. Find the probability that they meet again after
each one makes N steps. It is understood that they make steps simultaneously. Hint: It is
practical to consider their relative motion.
Solution 0.2:
1. Since the total displacement M is the difference of number m of the hops to the right and
Nm hops to the left, we have
m = (N+M)/2, Nm = (NM)/2.
As a result,
P
N
(M) =
N!
[(N+M)/2]![(NM)/2]!
p
(N+M)/2
(1p)
(NM)/2
. (1)
2. For p = 1/2,
P
N
(M) =
N!
2
N
[(N+M)/2]![(NM)/2]!
. (2)
3
3. Let us start with calculation of m. We have
m =
N

m=0
mW
N
(m) =
N

m=0
m
N!
m!(Nm)!
p
m
q
Nm
=
N

m=0
N!
m!(Nm)!
q
Nm
_
p
d
dp
p
m
_
= p
d
dp
_
N

m=0
N!
m!(Nm)!
p
m
q
Nm
_
= p
d
dp
(p+q)
N
= pN(p+q)
N1
= pN.
Here we have used that p+q = 1. Then

M = 2 mN = (2p1)N = (pq)N.
4. Again,
M
2
= (2mN)
2
= 4m
2
4N m+N
2
= 4m
2
+(14p)N
2
,
M
2

M
2
= 4m
2
+(14p)N
2
(2p1)
2
N
2
= 4m
2
4p
2
N
2
.
Now we can use the same trick as in the previous problem to calculate m
2
.
m
2
=
N

m=0
m
2
W
N
(m) =
N

m=0
m
2
N!
m!(Nm)!
=
N

m=0
N!
m!(Nm)
!
_
p
d
dp
_
2
p
m
q
Nm
=
_
p
d
dp
_
2
(p+q)
N
= p
_
N(p+q)
N1
+ p(N(N1)(p+q)
N2

= (Np)
2
+Np(1p).
As a result, we obtain
M
2
= 4(Np)
2
+4Np(1p) 4(pN)
2
= 4Np(1p). (3)
5. We have

M
=

4Npq
N
2
(pq)
2
=
1

2pq
|pq|
.
6. The relative motion corresponds to 2N steps, and the total displacement m=0. Thus, from
Eq. (2) we get
P
2N
(0) =
(2N)!
(2
N
N!)
2
.
4
Basic principles of thermostatics
0.3. Van der Waals gas
The fundamental relation for the Van der Waals gas is given by
s = s
0
+R ln[(v b)/(v
0
b)] +(3/2)R lnsinh[c(u+a/v)] , (4)
(all notations in this section are according the the book [?].
(i) Show that the corresponding equation of state is given by
(P+a/v
2
)(v b) =R T . (5)
(i) Expand the Van der Walls equation of state (5) in the form of virial expansion and deter-
mine 3 rst virial coefcients.
Solution 0.3:
(i) By denition T
1
=
_
s
u
_
V,N
. Consequently,
T
1
=
3R
2
cosh(u+a/v)
sinh(u+a/v)
(6)
Following the Maxwell relation
_
S
V
_
T
=
_
P
T
_
V
we get
P
T
=
_
s
v
_
u,N
=
R
v b

3R
2
cosh(u+a/v)
sinh(u+a/v)
a
v
2
=
R
v b

1
T
a
v
2
. (7)
As a result, we obtaine the VdW equation (5).
5
(ii) Expanding Eq. (5) in powers of v
1
we get
P =
TR
v
_
1+
ba/R T
v
+
b
2
v
2
+. . .
_
.
Thus
B(T) = ba/R T , C(T) = b
2
.
0.4. Which of the following, offered as fundamental relations, are acceptable and which violate
one or more of fundamental principles of thermostatics? (R, N
0
, V
0
and S
0
are constants)
(i) S = NR
_
1+
_
(E/E
0
)
3/2
(V/V
0
)(N/N
0
)
5/2
__
(ii) S = R(EV/E
0
V
0
)
2
(N/N
0
)
3
(iii) S = tan
_
(E/E
0
)(V/V
0
)
2
(N/N
0
)
3

(iv) S = R(EVN/E
0
V
0
N
0
)
1/3
(v) S = NRexp
_
(E/E
0
)
2
(V/V
0
)
2

(vi) S = NRcoth
_
(E/E
0
)
2
(V/V
0
)(N/N
0
)

(vii) S = R(EN/E
0
N
0
)
1/2
exp
_
(VN
0
/V
0
N)
2

(viii) S = R(NE/N
0
E
0
)
2
(V/V
0
)
1/5
.
Solution 0.4: Main principle: the quantities S and V are additive in the particle number. Thus,
the energy can be expressed as E = N f (S/N,V/N), or S/N = f
1
(E/N,V/N) = f
1
(u, v). Let us
put S = Ns, E = Nu, V = Nv and rewrite the equations as
(i) s = R
_
1+
_
(u/E
0
)
3/2
(v/V
0
)N
5/2
0
__
(ii) s = N
3
0
R(uv/E
0
V
0
)
2
(iii) s = N
1
tan
_
N
3
0
(u/E
0
)(v/V
0
)
2

(iv) s = N
1/3
0
R(uv/E
0
V
0
)
1/3
(v) s = Rexp
_
(u/E
0
)
2
(v/V
0
)
2

(vi) s = Rcoth
_
(u/E
0
)
2
(v/V
0
)(N
4
/N
0
)

(vii) s = R(u/E
0
N
0
)
1/2
exp
_
(vN
0
/V
0
)
2

6
(viii) s = N
16/5
R(u/N
0
E
0
)
2
(v/V
0
)
1/5
.
We observe that only Eqs. (i), (ii), (iv) (v) and (vii) meet the requirements. Thus, we are left with
(i) s = R
_
1+
_
(u/E
0
)
3/2
(v/V
0
)N
5/2
0
__
(ii) s = N
3
0
R(uv/E
0
V
0
)
2
(iv) s = N
1/3
0
R(uv/E
0
V
0
)
1/3
(v) s = Rexp
_
(u/E
0
)
2
(v/V
0
)
2

(vii) s = R(u/E
0
N
0
)
1/2
exp
_
(vN
0
/V
0
)
2

For all these expression the entropy is an increasing function of the energy that is requires for
the temperature to be positive, remember that T
1
= (S/E)
V,N
.
Then, the pressure must also be positive, and the pressure is given by the relation P =
(u/v)
s,N
. Thus, for a constant entropy the energy u must be decreasing function of vol-
ume. This requirement is met only for Eqs. (i), (ii) and (iv). Thus we are left with equations
(i) s = R
_
1+
_
(u/u
0
)
3/2
(v/v
0
)
__
(ii) s = N
1
0
R(uv/u
0
v
0
)
2
(iv) s = N
1
0
R(uv/u
0
v
0
)
1/3
where u
0
= E
0
/N
0
, v
0
=V
0
/N
0
. These relations do not contradict general principles.
0.5. Consider an isolated composite system of constant total volume, which consists of two
subsystems separated by a movable wall that permits ow of heat across it but is non-permeable
by the ow of matter. Find out the condition of equilibrium between the two systems using
(a) the energy minimum principle
(b) the entropy maximum principle
Solution 0.5: Denote partial energies as E
i
, partial entropies as S
i
, and partial volumes as V
i
for
i = 1, 2. We have
E = E
1
+E
2
= const , S = S
1
+S
2
= const , V =V
1
+V
2
= const .
We have also
T =
_
E
S
_
V
=
__
S
E
_
V
_
1
, P =
_
E
V
_
S
= T
_
S
V
_
E
7
(a) Minimum energy
E
S
1
=
E
1
S
1
+
E
2
S
2
S
2
S
1
=
E
1
S
1

E
2
S
2
T
1
= T
2
,
E
V
1
=
E
1
V
1
+
E
2
V
2
V
2
V
1
=
E
1
V
1

E
2
V
2
P
1
= P
2
.
(b) Maximum entropy
S
E
1
=
S
1
E
1
+
S
2
E
2
E
2
E
1
=
S
1
E
1

S
2
E
2
T
1
= T
2
,
S
V
1
=
S
1
V
1
+
S
2
V
2
V
2
V
1
=
S
1
V
1

S
2
V
2
P
1
/T
1
= P
2
/T
2
.
0.6. Now assume that the wall in the problem 0.5 is permeable to the ow of molecules. Find
out the condition of equilibrium between the two systems using
(a) the energy minimum principle
(b) the entropy maximum principle
Solution 0.6:
(a) Remembering the expression for thermodynamic potential dE = T dS PdV dN and
using previous considerations we obtain the condition for the equilibrium as

1
=
2
.
(b) Following the previous considerations we get (S
1
/N
1
) = (S
2
/N
2
). Since
dS =
dE
T


T
dN
we obtain the condition

1
/T
1
=
2
/T
2

1
=
2
.
0.7. Prove the following consequences of the third law of thermodynamics
(i) c
P
and c
V
of simple uids must vanish at T = 0.
(ii) Coefcient of thermal expansion vanishes at T = 0
Solution 0.7:
(i) For i =V, P, according to the denition, c
i
= T (s/T)
i
= (S/lnT)
i
. Since S tends to a
nite limit (zero) and lnT .
(ii) The thermal expansion coefcient = (V/T)
P
can be determined by the Maxwell rela-
tion (V/T)
P
=(S/P)
T
. Since S 0 at T 0 we observe that 0.
8
Basic principles of statistical mechanics
0.8. The Hamiltonian of a collection of N non-interacting one-dimensional classic harmonic
oscillators is given by
H =
1
2M
N

i=1
p
2
i
+
k
2
N

i=1
x
2
i
.
Calculate the internal energy of the system using the rules of calculation for isolated systems.
Hint: Apply the law of equipartition of the energy.
Solution 0.8 According to the equipartition law, the average energy for each degree of freedom
is the same. Here we have 2N degrees of freedom. The average energy for one degree of freedom
is
E
1
=

dp(p
2
/2M)e
p
2
/2MkT

dpe
p
2
/2MkT
=
kT
2
.
The total energy is
E = 2NE
1
= NkT .
0.9. Model for rubber elasticity.
The rubber is assumed to consist of a polymer chain of N rod-like monomers, each of length
a and is sibjected to a force F in the +X direction. Each monomer can point independently along
any of X,Y, Z axes, in either the + or direction. The energy is only X-dependent; = aF for
the monomer pointimg along X, = aF for the monomer pojnting along +X, = 0 for the
monomer along Y and Z.
(a) Calculate the partition function for the N-monomer chain.
(b) Show that the linear thermal expansivity is negative, as for the real rubber. Interpret the
result physically.
9
Solution 0.9 .
(a) Each monomer has the partition function
Z
1
= 2+2+e
Fa
+e
Fa
= 4+2cosh(Fa)
the total partition function being
Z = Z
N
1
= 2
N
[2+cosh(Fa)]
N
.
(b) If N
+
states point on the average to +X, N

states point to X, etc., then

L
X
= Na(
+

), L
Y
= Na(
+

), L
Z
= Na(
+

),
We immedately obtain that L
Y
=L
Z
= 0. Furthermore,

= e
aF
/Z
1
, L
X
= Na
sinhaF
2+coshaF
.
We immediately get L
X
/ > 0. Consequently, L
X
/T < 0. Rubber contracts upon
heating.
0.10. The partition function for a system of some kind of particles is
Z
N
=
_
(V Nb)/
3

N
exp(aN
2
/V),
where
=
_
2 h
2
/mk
B
T
and a and b are constants, V is the volume and N is the number of particles; all other symbols
have their usual meaning.
(a) Find the internal energy E(N, T,V).
(b) Find the pressure P(N, T,V).
(c) Find the entropy S(N, T,V).
(d) Is this expression for S a valid fundamental relation, except perhaps at T = O? If not, what
is wrong, and how ca Z
N
be appropriately corrected?
Hint: Recall Gibbs paradox.
10
Solution 0.10
(a) Since E =lnZ/ with (k
B
T)
1
we rewrite the partition function as
lnZ
N
= Nln(V Nb) 3Nln+aN
2
/V
with =
_
h/m. Having in mind that / = 1/2 = k
B
T/2 we get:
E = (3/2)Nk
B
T aN
2
/V .
(b) Let us dene the Helmholtz free energy
F =k
B
T lnZ
N
=k
B
T
_
Nln(V Nb) 3Nln+aN
2
/V

.
We have
P =
_
F
V
_
T
=
Nk
B
T
V Nb
+
aN
2
V
2
.
(c)
S =
_
F
T
_
V,N
= Nk
B
_
3
2
+ln
V Nb

3
_
.
(d) Entropy is not an additive quantity. The states created by permutation of the particles are
actually the same, so the partition function Z
N
should be divided by N!. In the main
approximation it will result in the expression
S = Nk
B
_
3
2
+ln
v b

3
_
, v V/N.
Another point is that the entropy does not vanish as T 0. One cannot correct this prop-
erty within classical statistics.
0.11. Calculate the partition function and the free energy for a ideal classical gas consisting
of N molecules at temperature T contained in a vertical column of height H and cross-sectional
area A in the presence of the of non-vanishing constant acceleration g due to gravity.
11
Solution 0.11 The total energy of the molecule in the gravitational eld is
E =
p
2
2m
+mgz
where z-axis is directed perpendicular to the Earth surface. Then,
Z
1
=

d
3
r

d
3
p
(2 h)
3
e
(p
2
/2m+mgz)
= A

H
0
dze
mgz

d
3
p
(2 h)
3
e
p
2
/2m
.
The rst integral is equal to (mg)
1
_
1e
mgH
_
. The second integral can be calculated in
the spherical coordinates. Since d
3
p = p
2
dpdd and kinetic energy depends only on p we can
write:
d
3
p
(2 h)
3
=
p
2
dp
2
2
h
3
=
m

2md
2
2
h
3
g()d, .
Here p
2
/2m, while
g() = g
0

1/2
, g
0
=
(2m)
3/2
4
2
h
3
is the density of states. Thus the second integral is
g
0

0
d
1/2
e

= g
0

3/2

0
x
1/2
e
x
dx =

2
g
0

3/2
.
Collecting all the factors we obtain
Z
1
=Z
1e
mgH
mgH
, Z =V

2
(2m)
3/2
4
2
h
3

3/2
=
V

3
.
Finally,
Z
N
=
Z
N
1
N!
=
V
N

3N
N!
_
1e
mgH
mgH
_
N
.
0.12. Calculate the partition function and the free energy for an ideal classical gas consist-
ing of N molecules at temperature T contained in a vessel and subjected to a centrifugal force
M
2
z
2
/2, where z is the distance of the particle from the axis of rotation and is the angular
velocity of rotation of the centrifuge.
12
Solution 0.12: When the external eld is present, the integrand in the partition function con-
tains an extra factor e
U
where U M
2
z
2
/2. Then one has to replace volume in the usual
expression for the partition function by

d
3
r e
U
. This procedure yields an extra factor
1
V

d
3
r e
U
=
2L
R
2
L

R
0
zdze
M
2
z
2
/2
=
2
M
2
R
2

M
2
R
2
/2
0
de

=
2
M
2
R
2
_
e
M
2
R
2
/2
1
_
.
Thus,
Z =
2Z
0
M
2
R
2
_
e
M
2
R
2
/2
1
_
, F = F
0
Nk
B
T ln
2k
B
T
M
2
R
2
_
e
M
2
R
2
/2k
B
T
1
_
.
0.13. Consider an ideal monoatomic gas of N molecules in the presence of an external magnetic
eld H, where each molecule behaves as an Ising spin. Calculate the free energy, energy, and
entropy and interpret the result physically. Find the limit of S at T 0.
Solution 0.13: The energy of the Ising spin Sin magnetic eld can be written as U = S
H
H
where S
H
acquires the values S. Consequently, the partition function can be written as
Z
1
= Z
0

e
SH
= 2cosh(SH).
Here Z
0
allows for non-magnetic degrees of freedom. Consequently,
Z = Z
N
1
/N! = (Z
N
0
/N!)[2cosh(SH)] ,
F F
0
= (N/)ln[2ecosh(SH)] ,
E E
0
= Z/ =NSHtanh(SH),
(SS
0
)/k
B
= (E F) = Nln[2ecosh(SH)] NSHtanh(SH).
0.14. Evaluate the contribution of a one-dimensional anharmonic oscillator having a potential
V(x) = cx
2
gx
3
f x
4
to the heat capacity. Discuss the the dependence of the mean value of
the position x of the oscillator on the temperature T. Here c, g, f are positive constants. Usually,
g c
3/2
(k
B
T)
1/2
and f c
2
/k
B
T.
Solution 0.14. Since g and f are small let us try to apply perturbation theory. Since the typical
value of the displacement x = (k
B
T/c)
1/2
we obtain
g x
3
/k
B
T = (k
B
T)
1/2
c
3/2
1, f x
4
/k
B
T = f k
B
T/c
2
1.
13
Thus one can expand the exponential to obtain
e
V(x)
e
cx
2
_
1gx
3
f x
4
_
.
As a result,
Z = Z
0

dxe
V(x)

_

c
_
1+
3 f
4c
2
_
.
Here Z
0
is the contribution of kinetic energy. Consequently,
lnZ = lnZ
0
+(1/2)ln(/c) (1/2)ln+ln(1+3 f /4c
2
)
= lnZ
0
+(1/2)ln(/c) (1/2)ln+3 f /4c
2
,
E = lnZ
0
/lnZ/
= 1/2+1/2+3 f /4
2
c
2
= k
B
T +3 f (k
B
T)
2
/4c
2
,
C = k
B
_
1+3 f k
B
T/2c
2
_
.
To estimate x we calculate
x =

dxxe
V(x)

dxe
V(x)
g

0
x
4
dxe
cx
2

0
dxe
cx
2
=
3
4
g
c
2
=
3
4
g x

1/2
c
3/2
x.
We have x T. 0.15. The energy of anharmonic oscillator is given by
H = p
2
/2m+bx
2n
where n is a positive integer and n > 1. Consider a thermodynamic system consisting of a large
number of these identical noninteracting oscillators.
(a) Derive the single oscillator partition function.
(b) Calculate an average kinetic energy of an oscillator.
(c) Calculate an average potential energy of an oscillator.
(d) Show that the heat capacity is
C = (Nk
B
/2)(1+1/n).
14
Solution 0.15.
(a)
Z
1
=

dp
2 h
e
p
2
/2m

dxe
bx
2n
Z
k
Z
p
,
Z
k
=
m
1/2
h(2)
1/2
,
Z
p
=
(1/2n)
n(b)
1/2n
.
(b)
E
k
=lnZ
k
/ = k
B
T/2.
(c)
E
p
=lnZ
p
/ = k
B
T/2n.
(d) Straightforward.
0.16. Suppose the expression S = k
B
r
P
r
lnP
r
is accepted as a denition of the entropy.
Imagine that a system A
1
has probability P
(1)
r
of being in a state r and a system A
2
has probability
P
(2)
s
of being in a state s. Then
S
1
=k
B
r
P
(1)
r
lnP
(1)
r
, S
2
=k
B
s
P
(2)
s
lnP
(2)
s
.
Each state of a composite system A consisting of A
1
and A
2
can then be labeled by the pair of
numbers, r, s. Let the probability of A being found in the state r, s be denoted by P
rs
, and the
corresponding entropy is k
B
r,s
P
rs
lnP
rs
.
(a) If A
1
and A
2
are weakly interacting so they are statistically independent, then P
rs
= P
(1)
r
P
(2)
s
Show that under such circumstances the entropy is simply additive, i. e. S = S
1
+S
2
.
(b) Suppose that A
1
and A
2
are not weakly so that P
rs
=P
(1)
r
P
(2)
s
. One has, of course, the general,
relations
s
P
rs
= P
(1)
r
,
r
P
rs
= P
(2)
s
, and
r,s
P
rs
= 1. Show that
S(S
1
+S
2
) = k
B
rs
P
rs
ln
P
(1)
r
P
(2)
s
P
rs
.
Moreover, by using the inequality
lnx 1x,
show that S S
1
+S
2
, where the equality holds only if P
rs
= P
(1)
r
P
(2)
s
for all r and s. This
means that the existence of correlation between the systems leads to a situation less random
that where the systems are completely independent of each other.
15
Solution 0.16: Let us start from the part (b), and the 1st part will be a limiting case. We have
(S
1
+S
2
)k
B
=

r
P
(1)
r
lnP
(1)
r
+

r
P
(2)
s
lnP
(2)
s
=

r,s
P
rs
ln(P
(1)
r
+

s,r
P
rs
lnP
(2)
s
=

r,s
P
rs
ln
_
P
(1)
r
P
(2)
s
_
.
If P
rs
= P
(1)
r
P
(2)
s
then S = S
1
+S
2
. Now
S(S
1
+S
2
) = k
B
r,s
P
rs
ln
P
(1)
r
P
(2)
s
P
rs
k
B
r,s
P
rs
_
P
(1)
r
P
(2)
s
P
rs
1
_
= k
B
r,s
_
P
(1)
r
P
(2)
s
P
rs
_
= 0.
0.17. Consider a system distributed over its accessible states r in accordance with a probability
distribution P
r
, and let its entropy be dened by the relations
S =k
B
r
P
r
lnP
r
,

r
P
r
= 1.
Compare this distribution with the canonical distribution
P
(0)
r
=
e
E
r

r
e
E
r
corresponding to the same mean energy E, i. e.

r
P
r
E
r
=

r
P
(0)
r
E
r
=E.
The entropy of the canonical distribution is
S
0
=k
B
r
P
(0)
r
lnP
(0)
r
.
(a) Show that
SS
0
= k
B
r
P
r
ln
P
(0)
r
P
r
.
(b) Using the inequality lnx x 1 show that S
0
S; the equality sign holds only if P
r
= P
(0)
r
for all states r. This shows that, for a specic value of mean energy, the entropy S is a
maximum for the canonical distribution.
16
Solution 0.17:
(a)
(SS
0
)/k
B
=

r
_
P
r
lnP
r
P
(0)
r
lnP
(0)
r
_
=

r
_
P
r
lnP
r
P
(0)
r
(E
r
lnN )
_
= lnN E

r
P
r
lnP
r
=

r
P
r
ln
P
(0)
r
P
r
.
(b) We have

r
P
r
ln
P
(0)
r
P
r

r
P
r
_
P
(0)
r
P
r
1
_
=

r
_
P
(0)
r
P
r
_
= 0.
0.18. A one-dimensional normal distribution of zero mean and standard deviation is given by
P(x) =
1

2
e
x
2
/2
2
.
(a) Show that it entropy is (k
B
/2)ln(2e
2
).
(b) show that for given

x
2
P(x)dx =
2
, the normalized probability distribution having the largest entropy in the one-dimensional
normal distribution.
Solution 0.18:
(a) By denition,
S/k
B
=lnP =

x
2
2
2

1
2
ln(2
2
)
_
P(x)dx =
1
2
ln(2e
2
).
(b) Let us construct the functional
F [P(x),
1
,
2
] =

dxP(x)lnP(x) +
1
(

dxx
2
P(x)
2
) +
2
(

dxP(x) 1).
17
Maximizing entropy under these two constraints implies that the variational derivation
F /[P(x)] = 0 or equivalently:
lnP(x) 1+
1
x
2
+
2
= 0,
where the constants
1
and
2
are determined from the constraints:
1
=
1
2
2
and
2
=
1ln

2
2
. Hence, the Gaussian distribution:
P(x) =
1

2
2
e
x
2
/2
2
.
0.19. Using the method of canonical ensemble. calculate the partition function, average energy
and specic heat of a system consisting of N noninteracting quantum harmonic oscillators and
show that these expressions do reduce to the corresponding classical results in the appropriate
classical limit.
Solution 0.19: Since
n
= h(n+1/2) we obtain
Z
1
= e
h/2

n=0
e
hn
= [2sinh( h/2)]
1
, Z = Z
N
1
,
E = N
h
2
+N
h
e
h
1
,
C/k
B
=
E
T
= N( h)
2
e
h
_
e
h
1
_
2
.
Classical limiting case at h 1 is obvious.
0.20. Let us denote the average occupation of the n-th single-particle level by the symbol f
n
.
For fermions, consider the variational entropy
S
var
=

n
[ f
n
ln f
n
+(1 f
n
)ln(1 f
n
)]
and for bosons, consider the variational entropy
S
var
=

n
[ f
n
ln f
n
+(1+ f
n
)ln(1+ f
n
)] .
Show that if S
var
is maximized, subject to the conditions

n
f
n
= N,

n
f
n

n
= E
then the resulting distribution is the FD distribution in the 1st case and BE distribution in the
2nd case. Moreover, at the maximum S
var
is equal to the equilibrium entropy of the FD and DE
systems, respectively.
18
Solution 0.20: Let us construct the functional

S = S
var
N+E =

n
[f
n
ln f
n
(1 f
n
)ln(1 f
n
) f
n
+
n
f
n
] .
We have

S
f
n
= ln
1 f
n
f
n
+
n
= 0
1 f
n
f
n
= e
(
n
)
.
Consequently,
f
n
=
1
e
(
n
)
1
.
Substituting these solutions to the expressions for the variational entropy we nd the equilibrium
values of the entropy.
19
Chapter 1
Fluctuations, correlations and response
1.1. Consider a closed system consisting of N noninteracting classical one-dimensional har-
monic oscillators at a temperature T. Using the canonical ensemble, show that the most probable
energy of the system is identical to its average energy.
Solution 1.1: Let us start with the density of states for one-dimensional classical oscillator with
the Hamiltonian H = p
2
/2m+m
2
x
2
/2. We have
D() =

dpdx
2 h

_

p
2
2m

m
2
x
2
2
_
.
To calculate this integral let us rst recall the relation

b
a
dx[ f (x)] =

s
1
| f

(x
s
)|
where x
s
are roots of equation f (x) = 0 belonging to the interval (a, b). In our case,
x
s
=

2
m
2
_

p
2
2m
_
, | f

(x
s
)| = m
2
|x
s
| =

2m
2
_

p
2
2m
_
Consequently,
D() =
2
2 h

2m
2

2m
0
dp
_
p
2
/2m
=
2
2 h

2m
2

2m =
1
h
.
Consequently, the number of states per particle is / h. The thermodynamical weight is then
(/ h)
N
= (E/N h)
N
. The entropy to be maximized is then
S = k
B
Nln
E
N

E
T
20
that yields for the most probable E the equation
S
E
= k
B
N
E

1
T
= 0 E = Nk
B
T .
1.2. Calculate the mean-square uctuation in energy, and the mean-square fractional uctuation
in the energy of
(a) a quantum harmonic oscillator,
(b) a collection of N identical quantum harmonic oscillators.
Examine the results in the high-temperature limit.
Solution 1.2:
(a) Let us count the energy form the ground state energy E
0
= h/2. Then
Z
1
=

n=1
e
n h
=
1
1e

,
E =
lnZ
1

=
h
e
h
1
= hN (),
E
2
=
1
Z
1

2
Z
1

2
= h
2

2
N ()[2N () +1] ,
(E)
2
= ( h)
2
N ()[N () +1] ,
(E)
2
/E
2
= [N () +1] /N ().
(b) In the case of M oscillators one has to multiply both E and (E)
2
by M. Thus, (E)
2
=
[N () +1] /MN (). At high temperature, this ratio is approximately 1/M.
1.3. Show that for an ideal classical Ising magnet
(a) the thermal uctuation in the magnetization (M)
2
is related to the thermal energy k
B
T
and the corresponding response function, namely susceptibility through
(M)
2
= k
B
TN,
(b) and
(M)
2

1/2
/M 1/

N.
21
Solution 1.3:
(a) The energy is =H. Then,
Z
1
= 2cosh(H),
M =
1
H
lnZ
1

= tanh(H),
=
M
H
=

2
cosh
2
(H)
,
M
2
=
1
H
2
Z
1

2
Z
1

2
=
2
,
(M)
2
=
2
_
1tanh
2
(H)

=

2
cosh
2
(H)
.
Thus
(M)
2
= k
B
TN.
(b) Follows from statistical independence of different spins.
1.4. V
1
is the volume of a small subsystem of a very large container of volume V, which is lled
with N atoms of an ideal classical monoatomic gas. Denote the probability of nding a particular
gas particle in the subsystem volume by p =V
1
/V. Find an expression for the Probability P(N
1
)
that there are N
1
particles in the subsystem. From this expression , nd the relative variance
(N
1
N
1
)
2
. Show that the relative variance vanishes in the thermodynamic limit, i. e. when
V , N with N/V = const and simultaneously V
1
with p = constant.
Solution 1.4: Let us use the binomial distribution
P(N
1
) =
N!
N
1
!(NN
1
)!
_
V
1
V
_
N
1
_
1
V
1
V
_
NN
1
=
N!
N
1
!(NN
1
)!
p
N
1
(1p)
NN
1
.
Assuming N
1
N we write N! (NN
1
)!N
N
1
to get the Poisson distribution.
P(N
1
) =
(Np)
N
1
N
1
!
(1p)
N

(Np)
N
1
N
1
!
e
Np
.
We have
N
1
= e
Np

N
1
=0
N
1
(Np)
N
1
N
1
!
= Npe
Np

N
1
=1
(Np)
N
1
1
(N
1
1)!
= Np,
N
2
1
= e
Np

N
1
=0
N
2
1
(Np)
N
1
N
1
!
= Npe
Np

N
1
=1
N
1
(Np)
N
1
1
(N
1
1)!
= Npe
Np

N
1
=1
[1+(N
1
1)] (Np)
N
1
1
(N
1
1)!
= Np+(Np)
2
,
(N
1
)
2
= Np, (N
1
)
2
/N
1

2
= (Np)
1
.
22