Professional Documents
Culture Documents
This is page i
Printer: Opaque this
Silvio Salinas
19 August 2011
ii
Preface
Silvio Salinas
Institute of Physics, University of São Paulo
srasalinas@gmail.com
iv Preface
This is page 1
Printer: Opaque this
1
Introduction to Statistical Physics
for large N .
0.16
A)
0.14
0.12
0.1
PN(N1)
0.08
0.06
0.04
0.02
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
N1/N
0.06
B)
0.05
0.04
P(N1)
0.03
0.02
0.01
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
N1/N
Graphs of P N(N1) (dots) versus N1/N, and the corresponding Gaussian distribution (solid
line), for N=36 (top figure) and N=216 (at bottom).
and
P (AB) = P (B) P (A j B) = P (A) P (B j A) :
(viii) Considering a third event C, show that
P (B j A) P (B) P (A j B)
= ;
P (C j A) P (C) P (A j C)
which is an expression of Bayes’ theorem.
and that
Z Z
p (y) = dx1 dx2 p (x1 ) p (x2 ) (y x1 x2 ) :
p (y) = y exp ( y) :
Additional exercises
2
hN1 it+ t = 1 hN1 it + 1;
N
100
90
80
N1 (t)
70
60
50
40
0 50 100 150 200 250 300 350 400 450 500
t (passos)
Figura 1: Simulation for the Ehrenfest urn model. Graph of N1 versus the discrete time
t, for N=100 and the initial condition N10=N. The solid line represents the theoretical
result for the time evolution of the average value of N1.
This is page 11
Printer: Opaque this
2
Statistical Description of a Physical
System
Vsph EN :
where D > 0 and the spin variable Sj may assume the values 1
or 0, for all j = 1; 2; 3:::. This spin Hamiltonian describes the
e¤ects of the electrostatic environment on spin-1 ions. An ion
in states 1 has energy D > 0, and an ion in state 0 has zero
N! X 1
(U; N ) = U U
;
N D
! N D
N !N !
1 u u u u u
ln ! ln 2 1 ln 1 ln ;
N D D D D D
for N; U ! 1, with U=N = u …xed. This last expression is the
entropy per particle in units of Boltzmann’s constant, kB .
*** Let´s write the number of microscopic con…gurations with
N0 ions with spin S = 0, N+ ions with spin +1, and N ions
with spin 1,
N!
(N0 ; N+ ; N ) = :
N !N0 !N+ !
N 0 + N+ + N = N
and
D (N+ + N ) = U:
U=D
N! X U
!
D
= U U U
:
N D
! D
!N =0 D
N !N !
1 (N0 + N+ + N N) + 2 (DN+ + DN U) :
Using Stirling´s expansion, we take derivatives with respect to
all of the arguments. It is easy to eliminate the Lagrange mul-
tipliers. The maximum is given by
N e+ = U ;
e =N e0 = N
N
U
;
2D D
from which we have the same asymptotic expression
1 u u u u u
ln ln 2 1 ln 1 ln ;
N D D D D D
Additional exercises
Z
+1 " N
!#
1 X
dk exp ik R x2i =
2 i=1
1
Z
+1
d=2
R 2
= dk exp ikR :
d ik
1
ln N ! = N ln N N + O (ln N ) ;
Z xo 1
"
2
exp [ n ] dx = " exp [ n ] :
1
xo 2
"
Therefore,
" exp [ n ] In C;
which leads to
Z b
" exp [ n ] exp [nf (x0 )] exp [nf (x)] dx C exp [nf (x0 )] :
a
This is page 21
Printer: Opaque this
3
Overview of Classical
Thermodynamics
Use this result to show that the speci…c heat of an ideal gas does
not depend on volume. Show that
@ @ T
= :
@p T;N @T p;N
@s @p
=) = ;
@v T @T v
which leads to the …rst identity.
The proof of the second identity requires similar tricks.
= V fo (T ) exp ;
kB T
p= = fo (T ) exp :
V kB T
Thus, we can write
p
= kB T ln ;
fo (T )
which is identical to the expression for the chemical potential
in the …rst exercise, if we make o = 0 and po (T ) = fo (T ).
Therefore, we have Boyle’s law and the usual expressions for T
and .
and
p @s 1
= = a1=4 v 3=4 1=2
u :
T @v u 4
The Gibbs free energy per particle is given by the Legendre
transformation
g 1 p
=s u v;
T T T
where u and v come from the equations of state. Note that g
has to be given in terms of T and p.
dU = T dS + f dL + dN:
du = T ds + Hdm;
CH
T = H;
cH T
h = u + pv;
@2h @T T
= = > 0:
@s2 p @s p cp
Also, we have
@2h @v
= = v s < 0:
@p2 s @p s
1 @2s @2s
2 @u2 @u@v
du
d s= du dv @2s @2s :
2 @u@v @v 2
dv
and
@2s @2s
+ > 0:
@u2 @v 2
Now it is straightforward to relate these derivatives of the en-
tropy with positive physical quantities (as the compressibilities
and the speci…c heats).
This is page 29
Printer: Opaque this
4
Microcanonical Ensemble
30 4. Microcanonical Ensemble
1 1 2 (1 u=D)
= ln :
kB T D u=D
2D exp ( D)
u= ;
1 + 2 exp ( D)
v
! = ! (v) = ! o A ln ;
vo
4. Microcanonical Ensemble 31
Except for an additive constant, show that the entropy per par-
ticle is given by
! !
X N ~j N~j
s = kB ln ;
j
N N
n o
where N ~j is the set of occupation numbers at equilibrium.
Using the continuum limit of the Boltzmann gas, show that the
entropy depends on temperature according to a term of the form
kB ln T (note the correction!).
32 4. Microcanonical Ensemble
! !
X X
+ 1 N Nj + 2 E j Nj :
j j
ej = exp (
N 1 2 j) ;
so that
ej
N exp ( 2 j)
=X :
N exp ( 2 j)
j
X ej
N E
j = = u;
j
N N
4. Microcanonical Ensemble 33
34 4. Microcanonical Ensemble
s = kB [v ln v (v 1) ln (v 1)] ;
4. Microcanonical Ensemble 35
36 4. Microcanonical Ensemble
This is page 37
Printer: Opaque this
5
Canonical Ensemble
38 5. Canonical Ensemble
0), sketch graphs of the internal energy, the entropy, and the
speci…c heat versus temperature. Indicate the behavior of these
quantities in the limits T ! 0 and T ! 1. Calculate the
expected value of the “quadrupole moment,”
* N +
1 X 2
Q= S ;
N i=1 i
" N N
#
X X X
= exp D Si2 + oH Si =
S1 ;S2 ;:::;SN i=1 i=1
( )N
X
= exp DS 2 + o HS =
S
1 @
u= ln Z:
N@
1 @
m= ln Z;
N @H
1 @
Q= ln Z:
D @D
5. Canonical Ensemble 39
X N
X
H= J i i+1 oH i;
i=1;3;5;:::;N 1 i=1
@m
= (T; H) = :
@H T
( )N=2
X
= exp [ J 1 2 + oH ( 1 + 2 )] =
1; 2
40 5. Canonical Ensemble
sinh (2 o H)
= o ;
cosh (2 o H) + exp ( 2 J)
whose derivative with respect to H yields the susceptibility.
From the free energy,
1
f= ln Z;
N
we calculate the entropy,
@f
s = s (T; H) = :
@T H
s (T; H = 0) ! kB ln 2;
5. Canonical Ensemble 41
1
n = ~! o n + ; with n = 1; 3; 5; 7; ::::
2
42 5. Canonical Ensemble
5. Canonical Ensemble 43
8+1 +1 9N
<Z Z m! 2 2 = 2
N
= dpdq exp p2 q = :
: 2m 2 ; !
1 1
*
7. Consider again the preceding problem. The canonical par-
tition function can be written as an integral form,
Z1
Z( )= (E) exp ( E) dE;
0
Z1 Z
+i1
1
= dE (E) exp [ (E 0 E)] d = (E 0 ) :
2
0 i1
Inserting the result of the previous exercise, we have
Z
+i1
N
1 2
(E) = exp ( E) d ;
2 !
i1
44 5. Canonical Ensemble
1=2 2
(E) 2 u2 exp N ln + ln u + 1 :
!
1 2
s = ln + ln u + 1;
kB !
where 8 1
< 2 m! 2 q 2 ; for q > 0;
V (q) =
: 1
2
m! 2 q 2 + ; for q < 0;
with > 0.
(a) Obtain the canonical partition function of this classical
system. Calculate the internal energy per oscillator, u = u (T ).
What is the form of u (T ) in the limits ! 0 and ! 1?
(b) Consider now the quantum analog of this model in the
limit ! 1. Obtain an expression for the canonical partition
function. What is the internal energy per oscillator of this quan-
tum analog?
*** The classical partition function is given by Z = Z1N , where
Z1 = [exp ( ) + 1] :
!
5. Canonical Ensemble 45
46 5. Canonical Ensemble
This is page 47
Printer: Opaque this
6
The Classical Gas in the Canonical
Formalism
Thus,
@ n+2
u= ln Z1 = kB T:
@ 2n
from which we obtain the Helmholtz free energy (and the value
of D). It interesting to check the constant value of the molecular
speci…c heat, c = (9=2) kB .
The second Hamiltonian represents a more complicated model
of a diatomic molecule. The partition function is given by
3Z Z
2 m
Z1 = d r1 d3 r2 exp [
3
j(j!
r1 ! r2 j ro )j]
3
2 m
V I;
where
Zro Z1
I=4 r2 exp [ (ro r)] dr+4 r2 exp [ (r ro )] dr =
0 ro
2ro2 4 2
= + 3 exp ( ro ) :
( ) ( )4
Now it is interesting to obtain the speci…c heat as a function of
temperature.
Z Z2 Z
+1 Z
+1
p2 p2'
Z1 = d d' dp dp' exp [ +
ma2 ma2 sin2
0 0 1 1
4 2 ma2 2 sinh ( E)
+ E cos ] = :
E
Thus, we have
1
h cos i = coth ( E) = L( E) ;
E
Za Zb
2
=2 r dr 2 r2 [exp ( Vo ) 1] dr:
0 a
This is page 53
Printer: Opaque this
7
The Grand Canonical and Pressure
Ensembles
Show that this same form of entropy still holds in the pressure
ensemble (with a suitable probability distribution).
1 (h c)3
ln Z f (zo ) = 1 ln :
N 8 v
It is interesting to note that the same expression comes from
the asymptotic form
( )
N
1 1 1 8 V
ln ZN = ln :
N N N ! (h c)3
@ @
( N )2 = Nj2 hNj i2 = z z ln ( ; z) :
@z @z
q
Obtain an expression for the relative deviation ( N )2 = hNj i
of an ideal gas of classical monatomic particles.
*** For a classical ideal gas of monatomic particles, we have
3=2
@ zV 2 m
hN i = z ln ( ; z) = 3
@z h
and
3=2
@ @ zV 2 m
( N )2 = z z ln ( ; z) = 3 = hN i :
@z @z h
p
The relative deviation is thus of order 1= N .
@U
( E)2 = Ej2 hEj i2 = ;
@ z;V
where
( E)2 can
= N kB T 2 cV
is the average quadratic deviation of energy in the canonical
ensemble. Finally, show that
" #
@U @N @U @N
+
@N T;V @T V; T @N T;V @ T;V
" #2
1 @N @U
= > 0;
T @ T;V @N T;V
since
@N
( N ) 2 = kB T > 0:
@ T;V
(z; V ) = (1 + z)V 1 + z V
;
for n = 0; 1; :::.
Also, note the thermodynamic limit,
1
ln ! ln (1 + z) + ln z; for z > 1
V
and
1
ln ! ln (1 + z) ; for z < 1:
V
For z > 1, we have the equations of state
p 1 z 2 + (1 + z)
= ln (1 + z) + ln z; and = :
kB T v z (1 + z)
This is page 59
Printer: Opaque this
8
The Ideal Quantum Gas
1. Obtain the explicit forms of the ground state and of the …rst
excited state for a system of two free bosons, with zero spin,
con…ned to a one-dimensional region of length L. Repeat this
problem for two fermions of spin 1=2.
*** The psingle-particle states are given by ' (x) = A sin (kx),
where k = 2mE=~ and kL = n , with n = 1; 2; 3:::. Note that
' (x) vanishes at x = 0 and x = L.
The ground state is given by C sin ( x1 =L) sin ( x2 =L), where
C is a normalization constant.
The …rst excited state is given by
Z1
V ~2 k 2
4 k 2 dk ln 1 + z exp =
(2 )3 2m
0
1
V 4 3 ~2 k 2
= 3f k ln 1 + z exp +
(2 ) 3 2m 0
Z1 1
2 ~2 k 2 ~2 k 2
+ 4 k 2 dk z 1
exp +1 g
3 2m 2m
0
2 X j 2
1
= U;
3 j z exp ( j) +1 3
1 2
A = p~ o;
2m
X Z1
A ~2 k 2
ln = z exp ( j) = z 2 k exp + o ;
j
(2 )2 2m
0
~2 ~2
r = = ;
2kB I 2kB M ro2
This is page 65
Printer: Opaque this
9
The Ideal Fermi Gas
we have
Z 1 d=2
Ld 1 2m d=2 1
N= Cd f ( )d =
(2 )d 0 2 ~2
Z 1
d
= L D( )f ( )d ;
0
3=2
mkB T
= V :
2 n~2
The classical limit is given by
3=2
mkB T
ln cl =z V ;
2 ~2
from which we have
3=2
@ mkB T
N = z ln cl =z V :
@z 2 ~2
Thus,
3=2
N 2 ~2
z = exp ( )= ;
V mkB T
Therefore,
" 2
#
3 3 3 3
1
ln z = ln +A + = ln + ;
v v v 23=2 v
Also, we have
Z1 Z 1
1 2 exp (x)
f ( )d ! f +
2 1 [exp (x) + 1]2
0
Z 1
1 x2 exp (x) dx 1 2 1 2
+ 2g = 1+ :
( )2 1 [exp (x) + 1] 2 ( )2 3
From these results, we write a low-temperature expression for
the internal energy (except for vanishingly exponential correc-
tions), " #
2 2
1 T
U = N F 1+ ;
2 3 TF
where
2 ~2 N
F = ;
m A
which leads to a linear dependence of the speci…c heat with
temperature as in the three-dimensional case.
and
3
U = N F:
4
The asymptotic form of the speci…c heat at low tempera-
tures comes from a straightforward application of Sommerfeld´s
method. We should note that
Z 1 Z 3
2 1 3 1 x exp (x) dx
f ( )d = 1+ dx !
0 3 [exp (x) + 1]2
1 3 2 1
! 1+ ;
3 ( )2
and
Z 1 Z 1 4
3 1 4 x exp (x) dx
f ( )d = 1+ dx !
0 4 [exp (x) + 1]2
2
1 4 1 7 4 1
! 1+6 + ;
4 3 ( )2 15 ( )4
where Z 1
exp (x) dx 7 4
dx = :
1 [exp (x) + 1]2 15
as can be checked in a good integral table (in Gradshteyn and
Ryzhik, for example).
We then write
3 3 2 1
F = 1+ ;
( )2
and
2
3 1 4 1 7 4 1
U= N 3 1+6 + :
4 F 3 ( )2 15 ( )4
and Z 1
3=2
U = VC f ( )d ;
0
where
1
f( )= :
exp [ ( )] + 1
At low temperatures, f ( ) is almost a step function, so the
derivative df =d has a pronounced peak at = , which provides
Z 1 Z +1 3=2
1=2 2 3=2 x exp (x)
f ( )d ! 1+ dx;
0 3 1 [exp (x) + 1]2
which is the main idea of Sommerfeld´s scheme. Except for these
exponential corrections, we have the series expansion
Z 1
1=2 2 3=2 3 1
f ( )d = fI0 + I2 +
0 3 8 ( )2
3 1 1
+ I4 +O g;
128 ( )4 ( )6
where Z +1
exp (x)
I0 = dx = 1;
1 [exp (x) + 1]2
Z +1
x2 exp (x) 2
I2 = 2 dx = ;
1 [exp (x) + 1] 3
Z +1
x4 exp (x) 7 2
I4 = d x = ;
[exp (x) + 1]2 1 15
and so on (as you can check in the integral table of Gradshteyn
and Ryzhik, for example). We then have the expansion
3N 3=2 3=2 3 2 1 7 2 1 1
= F = 1+ 2 + 4 + O( ) ;
2 VC 24 ( ) 640 ( ) ( )6
We left for the reader the task of obtaining the numerical coef-
…cients (M2 and M4 ). It is then easy to write an expansion of
the internal energy in terms of (even) powers of (T =TF ).
Thus,
a
pV = U:
d
In order to derive the Fermi energy, we write
Z 1
@ Ld kd 1
N = z ln ( ; z; V ) = C d dk;
@z (2 )d 0 z 1 exp ( ck a ) + 1
p
~; = cp BH ;
M = CH + O H 3 :
ln = ln + + ln ;
with X
ln = ln f1 + z exp [ (c~k B H)]g :
!
k
We then write
@
N =z ln = N+ + N ;
@z
with
3 Z1
V 1 2
N = 2 f ( )d ;
2 ~c
0
where
1
f ( )= :
exp [ ( BH ) + 1]
As the Fermi energy is the chemical potential at zero tempera-
ture, we have
3 Z BH
Z F + BH
V 1 F
2 2
N= 2 d + d ;
2 ~c 0 0
2 N
3
F +3 F ( B H) =3 2
(~c)3 :
V
For BH << F (H = 0), we have
1=3 2
2 1=3 N B
F = 3 ~c H2 + O H4 :
V 2 )5=3 N 1=3
3 (3 ~c V
M= B (N+ N );
which leads to
2
3 BN
M= H + O H3 ;
2 )1=3 N 1=3
(3 ~c V
M = N h cos i = N L ( H) ;
where
1
L (x) = coth (x)
x
is the Langevin function.
(c) Show that the susceptibility in zero …eld is given by the
Curie law,
N 2
o = :
3kT
where
Z Z
Z1 = d exp ( H cos ) = 2 sin exp ( H cos ) d =
0
4
= sinh ( H) :
H
Thus,
1
M =N + coth ( H) = N L ( H) ;
H
*** The grand partition function for this system was written
as 1 Z
eHL2 X L +1
ln = 2 dkz
hc n=0 2 1
~2 kz2 He 1
ln 1 + z exp n+ :
2m mc 2
Since H=mc << 1, we can use Euler´s formula to rewrite this
expression,
Z Z 1
eHL2 L +1 ~2 kz2 He
ln = 2 dkz dx ln 1 + z exp x
hc 2 1 0 2m mc
Z
eHL2 L +1 He 1
2 dkz :
hc 2 1 mc z 1 exp ~2 kz2 + 1
2m
~2 kz2 He
= + x;
2m mc
This is page 81
Printer: Opaque this
10
Free Bosons: Bose-Einstein
Condensation; Photon Gas
15 g5=2 (1) 3
cV (T = T0 ) = kB > kB :
4 g3=2 (1) 2
d=2
where Cd = 2 = (d=2). With a suitable change of variables,
we have
2=d
(2 )1=2 ~2 N 2 1
kB T0 = ;
2m Ld Cd I
where
Z1
xd=2 dx
I= :
exp (x) 1
0
N (2 mkB T0 )3=2 1
= 3
g3=2 (1) + g3=2 exp ;
V h kB T0
!~k;j = ck;
U = V T n:
2 3
X X 1 5
= exp 4 ~!~k;j a~yk;j a~k;j + =
2
~k;j
n o
n!
k ;j
(1 )
Y X 1 Y exp 1
~ck
2
= exp ~!~k;j n+ = :
! n=0
2 ! 1 exp ( ~ck)
k ;j k ;j
1X X
ln Z = ~ck ln [1 exp ( ~ck)] ;
2! !
k ;j k ;j
1 @ 1 X ~!~k;j
u= ln Z = :
N@ N
~k;j exp ~!~k;j 1
U = V T n;