Homework 1

David Sirajuddin
Department of Nuclear Engineering and Engineering Physics
Physics 715 - Statistical Mechanics
February 8, 2010
1. Consider the Schrodinger equation:
i
d
dt
=

H
where

H is time-independent.
(a) What happens to this equation under the time-reversal transformation: t t
and all quantities are complex conjugated? What are the physical implications of
your result?
Under the time-reversal transformation and complex conjugation, Schrodingers
equation takes the form:
i
d
d(t)

(r, t) =

H

(r, t)
+i
d
dt

(r, t) =

H

(r, t)
where the superscripts and denote the complex conjugate and the Hermitian
adjoint respectively. Given that the Hamiltonian is independent of time, the
Hermitian adjoint operation returns the same operator since the Hamiltonian is
then self-adjoint (

H

=

H). Thus,
+i
d
dt

(r, t) =

H

(r, t)
This is of the same form as the original Schrodingers equation. The equation can
be recasted identically if there exists a time-reveral operator L that is invertible,
such that,
+Li
d
dt

(r, t) = L

H

(r, t)
i
d
dt
(L

(r, t)) =

H[L

(r, t)]
i
d
dt

(r, t) =

H

(r, t)
1
Provided that the Hamiltonian and the time-reversal operator commute. Thus,
the procedure explicitly shows that the complex conjugation and time-reversal of
Schrodingers equation produces another solution in a special case. The physical
implications of taking t t suggest there exists no preferred direction in time.
That is to say, the procedure informs that Schrodingers equation is reversible.
(b) How does your result to part (a) appear to contradict the Second Law of Thermody-
namics?
The conclusion of part (a) admits the consequence of reversibility, i.e. a deter-
ministic evolution is predictable in both directions of time. The contradiction of
the Second Law involves the tenet that the entropy S of a system must always
increase. Solutions involving time propogating in the opposite direction provides
implies it is possible for the entropy to decrease, or that extrapolation backwards
in time could admit instances of higher entropy than at later times.
(c) How can you resolve the apparent contradiction of part (b)?
The contradiction is resolved by noting the criterion for its applicability, i.e the
commutation between the Hamiltonian and a time-reversal operator. It is sug-
gested that this special case is never realized in reality for macroscopic systems.
The Schrodinger equation is an isolated system with its outside interactions mod-
eled through the Hamiltonian (in a potential eld term). The averaging procedure
of the equation to compute tractable quantities ruins the reversability.
2. (Huang 6.3) Consider a system of N free particles in which the energy of each particle
can assume two and only two distinct values, 0 and E(E > 0). Denote by n
0
and n
1
the occupation numbers of the energy level 0 and E, respectively. The total energy of
the system is U.
(a) Find the entropy of such a system.
The entropy S is calculated as per the recipe provided by Huang, using the denition
S = k ln , where k is Boltzmanns constant, and is the total number of microstates
the system may achieve. The quantity is calculated by noting the N free particles may
only take on two distinct values. The number of states may be modeled as a function
of the ratio U/E (i.e. = (U/E)). This allows a direct computation of the number of
microstates by way of the binomial coecient,
_
N
r
_
=
N!
r!(Nr)!
, for the ratio r = U/E,
without repeat counting.
The choice of counter r = U/E is furnished by use of the total energy U. For example,
suppose N = 4 providing for the total number of possible states to be 16. If n
0
= 3
particles are at zero energy, and n
1
= 1 particles are at energy E, then r = U/E =
(0+0+0+E)/E = 1, such that
_
N
1
_
= 4 possible states are calculated. This computation
predicts the correct number of states. If N = 4, n
0
= n
1
= 2, then r = U/E =
(0 + 0 +E +E)/E = 2E/E = 2, so that
_
N
2
_
= 6 microstates.
Thus, given this formulation, the entropy S may be expressed as,
S = k ln
_
N!
r!(N r)!
_
= k [ln N! ln r! ln(N r)!] (1)
2
Each term is of the form ln x!, where x is an integer. An approximation for such terms
may be obtained by approximating the sum as a continuous function,
ln x! = ln [x(x 1)(x 1) . . . 2 1]
= ln x + ln(x 1) +. . . + ln 2 + ln 1
=
x

m=1
ln m
Multiplying by a form of unity: 1 = (m + 1) - m,
ln x! =
x

m=1
ln m[(m+ 1) m]
. .
m
=
x

m=1
ln mm

_
x
1
ln mdm
ln x! xln x x (2)
This is, of course, Stirlings approximation. Applying Eqn. (2) with each term in Eqn.
(1),
ln N! N ln N N
ln r! r ln r r
ln(N r)! (N r) ln(N r) (N r)
ln
_
N!
r!(N r)!
_
= N ln N N r ln r +r (N r) ln(N r) + (N r)
= N ln N r ln r
$
$
$
$
(N r) N ln(N r) +r ln(N r) +
$
$
$
$
(N r)
= N [ln N ln(N r)] +r [ln(N r) ln r]
= N ln
_
N
N r
_
+r ln
_
N r
r
_
= N ln
_
N
N r
_
+r ln
_
N r
r
_
ln
_
N!
r!(N r)!
_
= N ln
_
1
r
N
_
+r ln
_
N
r
1
_
Inputting this result into Eqn. (1):
S = k ln (r)
kr ln
_
N
r
1
_
kN ln
_
1
r
N
_
3
Recalling that r = U/E gives a nal expression for the entropy S,
S = k
_
U
E
_
ln
_
N
_
U
E
_
1
1
_
kN ln
_
1
1
N
_
U
E
__
(3)
(b) Find the most probable values of n
0
and n
1
, and nd the mean square uctuations of
these quantities.
Given that the particles may only take on an energy of 0 or E, the occupation numbers
n
0
and n
1
follow a binomial distribution. The most probable values of n
0
and n
1
correspond to the modes of the distribution. The proceeding work derives the mode n
for n = n
0
, n
1
follows that of Ross [2].
Given the discrete binomial probability (P) mass function for N particles
P(i) =
_
N
i
_
p
i
(1 p)
ni
,
where i = 0, 1, . . . , N, and p = 1/2 is the probability of taking on an energy 0 or E, the
mode is proposed to be the largest integer less than or equal to n = (N +1)p = (N +
1)/2|. This is to say, the cumulative probability PN = i increases monotonically to
a peak value (mode) and decreases for a values larger than i corresponding to n. The
proof proceeds by determining the ratio Pn = i/Pn = i 1.
Pn = i
Pn = i 1
=
N!
i!(Ni)!
p
i
(1 p)
Ni
N!
(Ni+1)!(i1)!
p
i1
(1 p)
Ni+1
=
&
& N!
i
$
$
$
(i1)!
$
$
$
(Ni)!

p
i1
p
$
$
$
$
$$
(1 p)
Ni
&
& N!
(Ni+1)
$
$
$
(Ni)!
$
$
$
(i1)!

p
i1
$
$
$
$
$$
(1 p)
Ni
(1 p)
(Sorry for the mess above, I will clean it up if I have enough time)
=
(N i + 1)p
i(1 p)
Thus, an index i corresponds to greater frequency if Pn = i Pn = i 1, i.e., if
and only if
(N i + 1)p i(1 p)
(N + 1)p

ip i

ip
i (N + 1)p|
Identifying the particular value i as the mode n, and the probability of taking on an
energy as p = 1/2 completes the proof,
Most probable occupation number n = (N + 1)p| (4)
4
where n = n
0
, n
1
is constrained by the criterion n
0
+n
1
= N
The mean square uctuation (n)
2
) is given by,
(n)
2
) =
n
2
) n)
2
n)
2
(5)
Again, owing to both occupation numbers having the same distribution, n = n
0
, n
1
.
Ross indicates the expectation value of the rst and second moments of the occupation
numbers are given by [2]:
n) = Np
n
2
) = Np[(N 1)p + 1]
Proof of these propositions are not provided, but may be located in Ross [2], or any
standard probability text. For the probability p = 1/2, this admits n)
2
= N
2
/4, and
n
2
) = Np[(N 1)p + 1]
=
N
2
_
N 1
2
+ 1
_
=
N
2
_
N
2

1
2
+ 1
_
=
N
2
_
N
2
+
1
2
_
n
2
) =
N
2
4

N
4
Inserting these expressions into Eqn. (5) provides,
(n)
2
) =

N
2
4

N
4

N
2
4
N
2
/4
=
N/4
N
2
/4
=

N/4
N
$
$
$$
(N/4)
=
1
N
Taking the absolute value, the nal result may be quoted,
Mean square uctuation of occupancy number (n)
2
) =
1
N
(6)
Which, reassuringly, tends to zero as N , i.e. the result obeys the strong/weak
laws of large numbers.
5
(c) Find the temperatue as a function of U, and show that it can be negative.
The temperature T may be obtained from the entropy S calculated in part (a),
T =
_
U
S
_
V

1
T
=
_
S
U
_
V
Where U is the total internal energy of the system. Noting that U = U(S), the partial
derivative may translated to an ordinary derivative:
1
T
=
_
S
U
_
V

1
T
=
dS
dU
and the constant volume condition is understood. The dierentiation is more conve-
niently facilitated by using the form of entropy given above Eqn. (3), where S = S(r),
r = U/E. Translating dieretial operators, it is written,
d
dU
=
d
dr
dr
dU
=
1
E
d
dr
so that
dS
dU
=
1
E
dS
dr
. Calculating the derivative then admits T = T(r),
1
T
=
dS
dU
=
1
E
dS
dr
=
1
E
d
dr
_
kr ln
_
N
r
1
_
kN ln
_
1
r
N
_
_
=
k
E
_
ln
_
N
r
1
_

r
r

2
N
r
1
+&
& N

1
N
1
r
N
_
=
k
E
_
ln
_
N
r
1
_

&
&
&
& 1
N r
+
&
&
&
& 1
N r
_
=
k
E
_
ln
_
N
r
1
__
T(r) =
E/k
ln
_
N
r
1
_
Recalling r = U/E yields the temperature T as a function of total energy U
T(U) =
E/k
ln
_
N
U/E
1
_ (7)
To show that the system may take on negative temperatures, the logarithm argument
must be between 0 and 1. Explicitly, a condition can be found,
6
ln
_
N
U/E
1
_
< 0
N
U/E
1 < 1
N
U/E
< 2
U/E > N/2
A similar calculation for the lower bound, ln[N(U/E) 1] > 0 gives an upper limit for
U/E < N. Thus, the system may have negative temperatures if the ratio N/2 < U/E < N.
This statement equivalently implies that more than half the free particles must take on
an energy E. For if the total energy U =

i
n
i
= n
0
(0) +n
1
(E) = n
1
(E), this indicates
that n
1
> N/2 so that U/E > (N/2)(E/E) > N/2.
(d) What happens when a system of negative temperature is allowed to exchange heat with
a system of positive temperature?
So as to not contradict the Clausius statement, the hotter temperature (T > 0) must
transfer heat to the lower temperature (T < 0) until an equilibrium temperature is
reached. Curiously, the result of part (c) shows T(U) to admit a trend that binds only
the occupancy number n
1
(population of particles with energy E). The function T(U)
shows that in order to gain higher temperature, n
1
must decrease so that n
0
increases
(concentration of particles with zero energy). The process is counterintuitive in the
respect that to increase the temperature of the system, the total energy U must be
lowered.
3. Find the equations governing an adiabatic transformation of an ideal gas.
A set of equations relating the pressure (P), temperature (T), and the volume (V )
may be obtained in terms of the ratio of specic heats = C
p
/C
V
.
(a) TV
1
= constant
Beginning with the rst law in dierential form,
dU = dQ
..
=0, adiabatic
dW
dU = dW
= PdV
dU = PdV (8)
To obtain an expression for the internal energy dU, the denition of specic heat
at constant volume, C
V
is used [1],
C
V
=
_
U
T
_
V
7
For an ideal gas, U = U(T) C
V
=
_
U
T
_
V

dU
dT
. Thus, the dierential is
separable and the internal energy may be written as dU = C
v
dT. Consolidating
this result with Eqn. (8) gives,
C
v
dT = PdV
C
v
dT = NkT
dV
V
(9)
where in the last step, the ideal gas law (PV = NkT) was used. A substitution
for the quantity Nk may be extracted as per Huang, by examining the denition
of the specic heats at constant pressure and volume if the enforcement is made
that C
V
,= C
V
(T) [1].
dU = C
V
dT
U(T) = C
V
T + @
@
@
@@
constant
. .
may choose to be zero
U(T) = C
V
T
Inserting this result into U in the denition of enthalthy H = U + PV and
replacing P with the ideal gas law (P = NkT) yields,
H = (C
V
+Nk)T
_
H
T
_
V
=
dH
dT
= C
V
+Nk = C
p
as according to the denition of specic heat at constant pressure. This implies
that
Nk = C
p
C
V
Inputting this result into Eqn. (9):
8
C
v
dT = (C
p
C
V
)T
dV
V
C
V
dT
T
= (C
p
C
V
)
dV
V
dT
T
=
C
p
C
V
C
V
dV
V
dT
T
=
_
C
p
C
V
1
_
. .
=(1)
dV
V
dT
T
= ( 1)
dV
V
dT
T
+ ( 1)
dV
V
= 0
ln T + ( 1) ln V = constant
ln(TV
1
) = constant
And, the proposition is proved,
TV
1
= constant (10)
(b) PV

= constant
Beginning with the First Law (dQ = 0), the denition of specic heat at constant
volume, and the ideal gas law result from above,
dU = PdV First Law
dU = C
V
dT Denition of specic heat at constant volume
PV = (C
p
C
V
)T Ideal gas law withNk = C
p
C
V
Dierentiating the ideal gas law gives,
(C
p
C
V
)dT = PdV +V dP
dT =
1
C
p
C
V
(PdV +V dP)
Inserting this result in for dT in the second equation above,
9
dU =
C
V
C
p
C
V
(PdV +V dP)
=
1
C
p
/C
v
. .
=
1
(PdV +V dP)
dU =
1
1
(PdV +V dP)
Finally, substituting this form into the First law yields,
PdV = V dP = ( 1)PdV
$
$
$
PdV +V dP = PdV +$
$
$
PdV
V dP = PdV
dP
P
=
dV
V
dP
P
+
dV
V
=
ln P + ln V = constant
ln(PV

) = constant
Which implies,
PV

= constant (11)
(c) P
1
T

= constant
The nal equation may be obtained from a proper combination of the previous
results (Eqns. (10) and (11)) to eliminate the volume dependence.
Considering two states (1 and 2), Eqns (10) and (11) may be written as
T
1
V
1
1
= T
2
V
1
2

T
1
T
2
=
_
V
2
V
1
_

_
V
2
V
1
_
1
and
P
1
V

1
= P
2
V

2

P
1
P
2
=
_
V
2
V
1
_

Substituting the right hand equation of the second into the right hand side of the
rst gives,
10
T
1
T
2
=
_
P
1
P
2
__
P
1
P
2
_
1/
T
1
T
2
=
_
P
1
P
2
_
11/
T
1
P
1+1/
1
= T
2
P
1+1/
2
T
1
P

1
= T
2
P

2
T

1
P
1
1
= T

2
P
1
2
Which shows,
T

P
1
= constant (12)
4. An automobile tire containing N molecules of air at pressure times atmospheric
pressure, p
tire
= p
air
, and at an atmospheric temperature Tair is punctured so that
all the air settles down to a uniform temperature and pressure. What is the resulting
change in entropy? Use the ideal gas law, p = nkT where n is the number density.
Explain why one might have worried about how to model the atmosphere. Should one
hold the atmospheric pressure constant? or the volume? Explain why, for a tire of
realistic size, such fears are groundless.
The process is irreversible, but the change in entropy may notwithstanding be obtained
by a dierence between state values when obtained from a reversible path. The air in
the tire begins at a pressure p
tire
= p
air
and at a temperature T
air
. After expanding
into the atmosphere (reservoir), the gas must settle at an equilibrium temperature and
pressure to that of the atmosphere (T
air
and p
air
)). Labeling the two states: before
puncture (state 1) and after equilibrium has been reached (state 2), it is evident the
process is isothermal (T
1
= T
2
= T
air
). Invoking the First Law (dU = 0, U = U(T)),
dQ = dW = pdV
The work pdV may be expressed in terms of other parameters by dierentiating the
ideal gas law,
PV = NkT PdV = NkdT V dP = NkdT NkT
dP
P
where the ideal gas law was used again in the nal step. The First Law above, then
admits
11
dQ = NkdT NkT
dP
P
dQ
T
= Nk
dT
T
. .
=0,T=constant
Nk
dP
P
Integrating both sides with the identication of the entropy change S =
_
dQ
T
,
_
dQ
T
S = Nk ln
_
P
2
P
1
_
= Nk ln
_
p
air
p
tire
_
= Nk ln
_

p
air

p
air
_
Implying,
S = Nk ln (13)
5. (Huang 1.2)
(a) An engine is represented by the cyclic transformation shown in the accompanying
T S diagram, where A denotes the area of the shaded region and B the area of
the region below it. Show that this engine is not as ecient sa a Carnot engine
operating between the highest and the lowest available temperatures.
According to the gure, the eciency of the engine may be immediately written
down,
=
W
Q
in
12
Sirajuddin, David Homework 1 Physics 715, Spring 2010
where Q
i
n is the heat input, and the word W = Q
in
Q
out
as usual. Inspecting
the gure, these heats correspond to according to Q =
_
TdS. Thus, with the
identication Q
in
= A + B, and Q
out
= B, then W = A + B B = A. The
eciency is then seen to be,
=
A
A +B
=
1
1 +B/A
For a Carnot engine, the engine operates at the highest and lowest temperature
reservoirs available, and is the shape of a rectangle. The only change is that the
area A
carnot
> A. The eciency of this engine is given as,

carnot
=
1
1 +B/A
carnot
Comparing the two engines directly may be done with a ratio,

carnot

=
1 +B/A
1 +B/A
carnot
It is immediately seen that since A
carnot
> A, the ratio is greater than unity
(
carnot
/ > 1) revealing that the eciency of the Carnot engine is greater.
(b) Show that an arbitrary reversible engine cannot be more ecient than a Carnot
engine operating between the highest and the lowest available temperatures.
This part is an extension of the previous. Using the same gure, suppose an
engine is operating with any shaped cycle between the two temperature limits
given. A Carnot cycle is a rectangle, while an arbitrary engine is of any shape.
Again, the only quantity that changes is the area A on the T-S diagram. This
area changes with the particular shape of the cycle on the diagram. Given the
relation from before,
=
1
1 +B/A
With B always constant. The ecieny increases with A. However, it is noted that
the maximum area possible is obtained from a rectangular cycle, which is the same
as a Carnot engine. Thus, Carnots theorem is proven: No reversible engine can be
more ecient than a Carnot engine operating between the highest and the lowest
available temperature.
References
[1] Huang, Kerson Statistical Mechanics 1987. John Wiley & Sons, Inc. Canada.
[2] Ross, Sheldon A First Course in Probability, 6th Ed., Prentice Hall, Inc. , 2002.
13