Theoretical Statistical Physics: Prof. Dr. Christof Wetterich Institute For Theoretical Physics Heidelberg University

Theoretical Statistical Physics
Prof. Dr. Christof Wetterich

Institute for Theoretical Physics
Heidelberg University
Last update: March 25, 2014

Script prepared by Robert Lilow,
using earlier student's script by Birgir sgeirsson, Denitsa Dragneva and Emil Pavlov.
For corrections and improvement suggestions please send a mail to
lilow@stud.uni-heidelberg.de.
Contents
I.
Foundations of Statistical Physics
1. Introduction
1.1. From Microphysics to Macrophysics . . . . . . . . . . . . . . . . . . . . . . . .
1.2. Macroscopic laws and simplicity . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3. Fundamental laws are of probabilistic nature . . . . . . . . . . . . . . . . . . .
11
11
12
12
2. Basic concepts of statistical physics

2.1. Probability distribution and expectation values . . . . . . . . . . .
2.1.1. States (Microstates) . . . . . . . . . . . . . . . . . . . . . .
2.1.2. Probability distribution . . . . . . . . . . . . . . . . . . . .
2.1.3. Observables . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.4. Expectation value . . . . . . . . . . . . . . . . . . . . . . .
2.1.5. Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.6. Expectation values and probability distribution . . . . . .
2.2. Simple examples . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.1. Two-state system . . . . . . . . . . . . . . . . . . . . . . .
2.2.2. Four-state system . . . . . . . . . . . . . . . . . . . . . . .
2.3. Reduced system or subsystem . . . . . . . . . . . . . . . . . . . .
2.4. Probabilistic observables . . . . . . . . . . . . . . . . . . . . . . .
2.5. Quantum statistics and density matrix . . . . . . . . . . . . . . . .
2.5.1. Expectation value . . . . . . . . . . . . . . . . . . . . . . .
2.5.2. Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.3. Diagonal operators . . . . . . . . . . . . . . . . . . . . . .
2.5.4. Properties of the density matrix . . . . . . . . . . . . . . .
2.5.5. Diagonal elements of dene the probability distribution
2.5.6. Positivity constraints for the elements of . . . . . . . . .
2.5.7. Non-diagonal operators . . . . . . . . . . . . . . . . . . .
2.5.8. Pure quantum states . . . . . . . . . . . . . . . . . . . . .
2.5.9. Change of the basis by unitary transformation . . . . . . .
2.5.10. Pure and mixed states . . . . . . . . . . . . . . . . . . . .
2.6. Microstates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.7. Partition function . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.8. Continuous distribution and continuous variables . . . . . . . . .
15
15
15
15
15
16
16
16
17
17
18
18
20
23
23
24
24
25
25
26
26
29
29
30
31
33
33
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2.9. Probability density of energy . . . . . . . . . . . . . . . . . . .

2.9.1. Equal probability for all values . . . . . . . . . . . .
2.9.2. Boltzmann distribution for microstates . . . . . . . .
2.9.3. Number of one particle states in a momentum interval
2.10. Correlation functions . . . . . . . . . . . . . . . . . . . . . . .
2.10.1. Conditional probability p( B | A ) . . . . . . . . . . . .
2.10.2. Product observable B A . . . . . . . . . . . . . . . . .
2.10.3. Correlation function . . . . . . . . . . . . . . . . . . .
2.10.4. Independent observables . . . . . . . . . . . . . . . . .
2.10.5. Classical correlation function . . . . . . . . . . . . . .
2.10.6. Quantum correlations . . . . . . . . . . . . . . . . . .
3. Statistical description of many-particles systems
3.1. Chain with N lattice sites . . . . . . . . . . . . .
3.1.1. Probability distribution . . . . . . . . . .
3.1.2. Macroscopic probability . . . . . . . . .
3.1.3. Expectation value . . . . . . . . . . . . .
3.1.4. Dispersion . . . . . . . . . . . . . . . . .
3.2. Uncorrelated probability distribution . . . . . .
3.2.1. N spins on a chain . . . . . . . . . . . .
3.2.2. Correlation functions . . . . . . . . . . .
3.2.3. Independent measurement . . . . . . . .
3.3. Statistical Ensembles . . . . . . . . . . . . . . .
3.4. Gaussian distribution . . . . . . . . . . . . . . .
3.5. Thermodynamic limit . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
35
35
36
36
38
38
38
39
39
40
41
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
43
43
43
43
45
45
46
46
47
49
50
51
53
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
55
55
55
56
56
57
61
61
63
63
65
65
66
68
68
70
70
71
4. Equilibrium Ensembles
4.1. Microcanonical ensemble . . . . . . . . . . . . . . . . . . . . . .
4.1.1. Fundamental statistical postulate . . . . . . . . . . . . .
4.1.2. Microcanonical partition function . . . . . . . . . . . . .
4.1.3. Entropy S . . . . . . . . . . . . . . . . . . . . . . . . . .
4.1.4. Partition function for ideal classical gas . . . . . . . . .
4.1.5. Entropy of an ideal classical gas . . . . . . . . . . . . . .
4.1.6. Thermodynamic limit . . . . . . . . . . . . . . . . . . . .
4.1.7. The limit E E 0 . . . . . . . . . . . . . . . . . . . . . .
4.1.8. Derivatives of the entropy . . . . . . . . . . . . . . . . .
4.2. Canonical ensemble . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.1. System in a heat bath . . . . . . . . . . . . . . . . . . . .
4.2.2. Canonical partition function and Boltzmann factor . . .
4.2.3. Relation between temperature and energy . . . . . . . .
4.2.4. Energy distribution . . . . . . . . . . . . . . . . . . . . .
4.2.5. Equivalence of microcanonical and canonical ensemble .
4.2.6. Temperature . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.7. Temperature and energy distribution in subsystems . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4.2.8. Thermal equilibrium . . . . . . . . .

4.3. Grand canonical ensemble . . . . . . . . . .
4.4. Relations for entropy . . . . . . . . . . . . .
4.4.1. Entropy in canonical ensemble . . .
4.4.2. Entropy in grand canonical ensemble
4.4.3. General expression for entropy . . .
4.4.4. Entropy and density matrix . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5. Thermodynamics
5.1. Thermodynamic potentials . . . . . . . . . . . . . . . . . .
5.1.1. State variables, equations of state . . . . . . . . . .
5.1.2. Thermodynamic potentials and partition functions
5.1.3. Dierential relations . . . . . . . . . . . . . . . . .
5.1.4. Legendre transform . . . . . . . . . . . . . . . . . .
5.1.5. Properties of entropy . . . . . . . . . . . . . . . . .
5.1.6. Zero of absolute temperature . . . . . . . . . . . .
5.1.7. 3rd law of thermodynamics . . . . . . . . . . . . .
5.2. Reversible and irreversible processes . . . . . . . . . . . .
5.2.1. Subsystems and constraints . . . . . . . . . . . . .
5.2.2. Removing the constraint . . . . . . . . . . . . . . .
5.2.3. Increase of entropy by removing the constrains . .
5.2.4. 2nd law of thermodynamics . . . . . . . . . . . . .
5.2.5. Reversible processes for isolated systems . . . . . .
5.2.6. Reversible processes for open systems . . . . . . .
5.2.7. Thermodynamics and non-equilibrium states . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
71
72
73
73
74
74
75
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
77
77
77
79
80
81
83
84
85
85
85
86
87
88
89
89
90
II. Statistical Systems
91
6. Ideal quantum gas

6.1. Occupation numbers and functional integral . . . .
6.2. Partition function for the grand canonical ensemble
6.3. Mean occupation numbers . . . . . . . . . . . . . .
6.3.1. Bosons . . . . . . . . . . . . . . . . . . . . .
6.3.2. Fermions . . . . . . . . . . . . . . . . . . .
6.3.3. Mean energy . . . . . . . . . . . . . . . . .
6.3.4. Chemical potential . . . . . . . . . . . . . .
6.3.5. Boltzmann statistics . . . . . . . . . . . . .
6.4. Photon gas . . . . . . . . . . . . . . . . . . . . . . .
6.4.1. Black body radiation . . . . . . . . . . . . .
6.4.2. Spectrum of black body radiation . . . . . .
6.4.3. Energy density . . . . . . . . . . . . . . . .
6.4.4. Canonical partition function . . . . . . . . .
6.4.5. Free energy and equation of state . . . . . .
93
93
94
95
95
96
97
97
98
99
99
101
103
104
105
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
6.5. Degenerate Fermi gas . . . . . . . . . . . . . . . . . . . . . . . . . . .

6.5.1. Quasiparticles . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.5.2. The limit T 0 . . . . . . . . . . . . . . . . . . . . . . . . . .
6.5.3. Computation of the Fermi energy . . . . . . . . . . . . . . . .
6.5.4. Thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . .
6.5.5. Ground state energy . . . . . . . . . . . . . . . . . . . . . . .
6.5.6. One-particle density of states . . . . . . . . . . . . . . . . . .
6.5.7. 1 () for free non-relativistic fermions . . . . . . . . . . . . .
6.5.8. E(T , N ) for arbitrary temperature . . . . . . . . . . . . . . . .
6.5.9. Smoothing of Fermi surface . . . . . . . . . . . . . . . . . . .
6.5.10. Computation of (T ) from the normalisation condition . . . .
6.5.11. Occupation numbers for particles and holes . . . . . . . . . .
6.5.12. Solve for (T ) . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.6. Non-relativistic bosons . . . . . . . . . . . . . . . . . . . . . . . . . .
6.6.1. Chemical potential . . . . . . . . . . . . . . . . . . . . . . . .
6.6.2. Number of bosons in the ground state . . . . . . . . . . . . .
6.6.3. Qualitative behaviour of n(E
) for T 0 . . . . . . . . . . .
6.6.4. Gibbs potential . . . . . . . . . . . . . . . . . . . . . . . . . .
6.7. Classical ideal gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.7.1. Classical approximation . . . . . . . . . . . . . . . . . . . . .
6.7.2. Gibbs potential . . . . . . . . . . . . . . . . . . . . . . . . . .
6.7.3. Thermal de Broglie wavelength . . . . . . . . . . . . . . . . .
6.7.4. Thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . .
6.7.5. Validity of the approximation for a diluted gas (classical gas) .
6.8. Bose-Einstein condensation . . . . . . . . . . . . . . . . . . . . . . . .
6.8.1. n() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.8.2. Critical temperature . . . . . . . . . . . . . . . . . . . . . . .
6.8.3. Ground state of bosonic gas . . . . . . . . . . . . . . . . . . .
6.8.4. Bose-Einstein condensation . . . . . . . . . . . . . . . . . . .
7. Gas of interacting particles
7.1. Interactions and complexity . . . . . . . . . . . .
7.2. Real gases . . . . . . . . . . . . . . . . . . . . . .
7.3. Mean eld approximation . . . . . . . . . . . . . .
7.3.1. Classical statistics in phase space . . . . .
7.3.2. Average potential mean eld . . . . . .
7.3.3. Expansion of canonical partition function
(1)
7.3.4. Computation of Z ww
. . . . . . . . . . . .
7.3.5. Excluded volume . . . . . . . . . . . . . .
7.3.6. Mean potential . . . . . . . . . . . . . . .
7.3.7. Free energy . . . . . . . . . . . . . . . . .
7.4. Equation of state . . . . . . . . . . . . . . . . . . .
7.5. Specic heat of real gases . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
105
105
106
106
107
107
108
109
110
110
111
112
113
114
114
115
115
115
116
116
116
117
118
118
119
119
119
120
121
.
.
.
.
.
.
.
.
.
.
.
.
123
123
123
124
124
125
125
126
127
127
128
128
129
7.6. Equipartition theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
129
8. Magnetic systems
8.1. Magnetization . . . . . . . . . . . . . . . . . . . . . . .
8.1.1. Degrees of freedom . . . . . . . . . . . . . . . .
8.1.2. Hamiltonian . . . . . . . . . . . . . . . . . . . .
8.1.3. Magnetization . . . . . . . . . . . . . . . . . . .
8.1.4. Thermodynamic potential . . . . . . . . . . . .
8.1.5. Eective potential . . . . . . . . . . . . . . . .
8.1.6. Minimal principle . . . . . . . . . . . . . . . . .
8.2. Spontaneous symmetry breaking . . . . . . . . . . . .
8.2.1. Symmetries . . . . . . . . . . . . . . . . . . . .
8.2.2. Vanishing magnetic eld . . . . . . . . . . . . .
8.2.3. Domains . . . . . . . . . . . . . . . . . . . . . .
8.2.4. General form of U . . . . . . . . . . . . . . . .
8.3. Landau theory and phase transition . . . . . . . . . . .
8.3.1. Quartic expansion of U . . . . . . . . . . . . . .
8.3.2. Phase transition for j = 0 . . . . . . . . . . . .
8.3.3. Discontinuity in magnetization for non-zero B
8.4. Magnetic susceptibility . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
133
133
133
134
134
135
136
137
137
137
138
138
139
139
139
140
142
143
9. Phase transitions
9.1. Order of phase transition . . . . . . .
9.1.1. First order phase transition .
9.1.2. Second order phase transition
9.1.3. Probabilistic interpretation . .
9.2. Liquid-gas phase transition . . . . . .
9.2.1. Eective potential . . . . . .
9.2.2. First order transition . . . . .
9.2.3. Compressibility . . . . . . . .
9.2.4. Chemical potential . . . . . .
9.3. Phase diagram . . . . . . . . . . . . .
9.3.1. Vapor pressure curve . . . . .
9.3.2. p T diagram . . . . . . . . .
9.3.3. Clausius-Clapeyron relation .
9.3.4. Latent heat . . . . . . . . . .
9.3.5. Superheating, supercooling .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
147
147
147
147
149
150
150
151
152
152
153
153
153
154
155
155
10. Critical Phenomena

10.1. Critical exponents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.1.1. Specic heat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.1.2. Order parameter as function of temperature . . . . . . . . . . . . .
10.1.3. Susceptibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.1.4. Order parameter at critical temperature as function of the source
.
.
.
.
.
.
.
.
.
.
157
157
157
157
157
158
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
10.1.5. Mean eld exponents . . . . . . . . . . . . . . . . . . . . . . . . .

10.1.6. Universal critical exponents . . . . . . . . . . . . . . . . . . . . .
10.2. Correlation functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.2.1. Correlation length . . . . . . . . . . . . . . . . . . . . . . . . . .
10.2.2. Correlation function and generating functional . . . . . . . . . .
10.2.3. Correlation function and susceptibility . . . . . . . . . . . . . . .
10.2.4. Critical exponent of correlation length and anomalous dimension
10.2.5. Critical opalescence . . . . . . . . . . . . . . . . . . . . . . . . .
10.3. Universality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.3.1. Macroscopic laws for second order phase transition . . . . . . . .
10.3.2. Ising model in one dimension . . . . . . . . . . . . . . . . . . . .
10.3.3. Ising model in d > 1 dimensions . . . . . . . . . . . . . . . . . .
10.3.4. Renormalization ow . . . . . . . . . . . . . . . . . . . . . . . . .
10.3.5. Fixed points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
158
160
161
161
162
163
164
164
164
164
165
167
170
173
Part I.
Foundations of Statistical Physics
1. Introduction
1.1. From Microphysics to Macrophysics
Fundamental laws for basic interactions do not explain everything, e. g. the description of
macroscopic objects, consisting of many particles (solids, uids, gases).
A typical number of particles is Avogadro's number: N a ' 6 1023 . It denotes the number of
molecules per mol, e. g. the number of water molecules in 18 g water.
Typical questions are:
connections between pressure (p), volume (V ), and temperature (T ) of gases (equation of
state)
specic heat capacity of a material
conductivity of a solid, colour of a uid
connection between magnetization and magnetic eld
melting of ice
How does a refrigerator work?
How does a supernova explode?
energy budget of the Earth's atmosphere
diusion through a membrane in a plant
particle beam in a high energy collider
Typical area of application:
systems with many degrees of freedom macroscopic objects
For example a water droplet:
Consider an isolated system a droplet of water with volume V and number of particles N .
A classical mechanical description needs 6N initial conditions 6N degrees of freedom.
The distance between two trajectories, which initially are close to each other, grows
exponentially with time. Initial state can never be prepared (known) with arbitrary
precision: Limit for predictability (weather forecast)
11
1. Introduction
Try deterministic classical mechanics as a description of a water droplet for one second.
One needs more information than can be saved in the entire universe, even if one stores
one bit in every volume with Planck size (1034 m) 3
The goal of the statistical physics is to develop laws for the macrostates by deriving them from
the microphysics with the help of statistical methods.
1.2. Macroscopic laws and simplicity

qualitatively new laws
new concepts, e. g. entropy
statistic laws are very exact, relative uctuations
1
N
predictive power only for systems with many degrees of freedom; uncertainty about the
behaviour of single atoms (quantum physics), but precise predictions about the whole
system
thermodynamics
material properties
non-equilibrium processes
Scales for macroscopic systems:
L mean distance between particles in uids and solids ' 107 cm, N 1
t timescale of the microscopic motions:
motion of an atom: 1013 s
motion of an electron (in solids): 1015 s
1.3. Fundamental laws are of probabilistic nature

Fundamental laws are of probabilistic nature
The role of the statistical physics in the building of physics is huge: it is the foundation of the
modern physics! There are two two major strategies in physics, as illustrated in Figure 1.1:
1. Fundamental theory: systems with many (innite) degrees of freedom; QFT, String theory
(for scattering of two particles) are formulated as statistical theories.
2. The role of statistics to understand complex systems is obvious.
12
1.3. Fundamental laws are of probabilistic nature
Abstraction
Understanding
of the complexity
Fundamental laws
(1)
(2)
Figure 1.1.: The two major strategies in physics
To calculate the predictions, which follow from the fundamental theory, one uses methods of
statistical physics just like in this lecture.
Basic laws of physics are statistical in nature!
They make statements about:
the probability that, when measuring an observable A once, one gets the value Ai
the conditional probability for two measurements: if one measures a value Ai for an observable A, what is the probability that one measures the value B j for a second observable
B?
Remarks
1. From microphysics to microphysics and from Planck's scale (1019 GeV) to the energy scale
of LHC (1000 GeV) 16 orders of magnitude.
2. Universality: one does not consider all the details of the microscopic laws predictive
power of the QFT!
3. Conceptional revolution due to the statistical physics.
Modern physics: Plank, Boltzmann, Einstein.
(((( probabilistic theory predictive power
4. (
Determinism
(
13

2.1. Probability distribution and expectation values
2.1.1. States (Microstates)
We will denote states with and the number of states with . In the beginning, we will examine
only a nite number of states < .
2.1.2. Probability distribution

To every state we associate a probability p , the probability that this state is realized:
p .
(2.1)
The probabilities p have to obey two conditions

X
p 0
positivity,
(2.2)
p = 1
normalization.
(2.3)
The range of p is therefore
0 p 1 .
(2.4)
The probability distribution is {p } = {p1 ,p2 ,p3 , . . . }, which is an ordered sequence of numbers.
2.1.3. Observables
Capital roman letters are used for observables, e. g. A. A classical observable A has a xed value
A in every state :
A Spec A ,
(2.5)
where the spectrum Spec A of an observable is the set of all possible measurement values { A },
A R.
The probability to measure a certain value A is given by:
X

pA =
p ,
(2.6)

A = A
where one sums over all states for which A = A . A linear combination of two classical
observables A and B is again a classical observable:
C = f AA + f B B
C = f A A + f B B .
with
f A,B R ,
(2.7)
(2.8)
15
A(2)
A(3)
0
0
1
0
0
1
0
0
Table 2.1.: A system with A( ) =
And a function of A is an observable, too:

f (A)
= f (A ) .
(2.9)
2.1.4. Expectation value

The expectation value of an observable is dened as
hAi =
p A .
(2.10)
p (A hAi) 2
(2.11)
(A2 2 A hAi + hAi2 )
(2.12)
p A2 hAi2
(2.13)
2.1.5. Dispersion
The dispersion is given by
A2 =
= hA2 i hAi2 .
(2.14)
2.1.6. Expectation values and probability distribution

The expectation values of a complete set of observables determine the probability distribution
{p } uniquely. Consider observables A( ) with A( ) = , as illustrated in Table 2.1. Then, it
is
X
hA( ) i =
p A( ) = p .
(2.15)
The observables A( ) are called projectors. They can be considered as a basis for observables.
All classical observables can be expressed as linear combinations of projector observables. An
association of A( ) with possible events (i. e. that state is realized) leads to the Kolmogorov
axioms for probabilities.
16
2.2. Simple examples
2.2. Simple examples

2.2.1. Two-state system
Typical two-state systems would a coin, a spin, occupied / unoccupied states or computer bits.
States, possibilities and observables in these are usually denoted by:
: , or h1i, h2i or 1, 0
p : p , p or p1 , p2 ,
p1 + p2 = 1
A: number up number down, spin, particle number
As an example, let us dene the observable A, for which holds that A1 = 1 for state and
A2 = 1 for state . Then, we can express the spin in z-direction in terms of A:
Sz =
~
A.
2
(2.16)
The number of particles is given by N = 12 (A + 1), which corresponds to N 1 = 1 and N 2 = 0.

The expectation values of A and N are
hAi = p1A1 + p2A2
(2.17)
= p1 (1 p1 )
(2.18)
= 2p1 1 ,
(2.19)
hN i = h 12 (A + 1)i
(2.20)
= p1 N 1 + p2 N 2
(2.21)
= p1 .
(2.22)
Their dispersions are easily calculated. For A we rst need

hA2 i = p1A21 + p2A22
(2.23)
= p1 + (1 p1 )
(2.24)
= 1,
(2.25)
where we used A21 = A22 = 1. Then,

A2 = hA2 i hAi2
= 1 (2p1 1)
(2.26)
2
= 4p1 (1 p1 ) ,
(2.27)
(2.28)
and therefore
A2 = 0
for p1 = 0
or p1 = 1 .
(2.29)
And for N accordingly:

hN 2 i = 1 p1 + 0 p2 = p1 ,
(2.30)
N = p1 (1 p1 ) .
(2.31)
17
11
10
01
00
|1i
|2i
|3i
|4i
Table 2.2.: States in a four-state system
|1i
A
B
C
2
1
1
|2i
0
1
1
|3i
0
1
1
|4i
2
1
1
total spin
spin 1
spin 2
Table 2.3.: Observables in a four-state system
2.2.2. Four-state system

An example of a four-state system are two particles, each with a spin in two possible directions.
Dierent possibilities to denote these states are listed in Table 2.2. We can now introduce a
variable called the total spin A. Other possible observables are the spin of the rst particle B
and the spin of the second particle C. Values of these observables are given in Table 2.3.
For the particular case of equal probabilities for all states, p1 = p2 = p3 = p4 = 14 , we can
compute the expectation values:
hAi = hBi = hCi = 0 ,
(2.32)
and the dispersions:

hB 2 i = hC 2 i = 1 ,
hA2 i = 2 ,
(2.33)
B = C = 1 ,
A = 2 .
(2.34)
2.3. Reduced system or subsystem

For most parts of this lecture the basic concepts of section 2.1, classical statistics, are sucient. So,
we want to describe quantum systems, but often use classical statistics. This needs justication!
The following parts of chapter 2 motivate why this is possible for many purposes.
The basic observation is that classical statistics and quantum statistics have much in common!
Many quantum features appear when one considers subsystems of classical statistical systems.
When only part of the information in {p } is needed or available, we can integrate out the
redundant information, which results in a reduced probability distribution.
Examples
1. Consider the previous example of a four-state system again. Suppose we only have
information about spin 1. Then, we can combine the probabilities for states with the
18
2.3. Reduced system or subsystem
same B and arbitrary C . As a consequence, the system turns into an eective two-state
system.
Here, = | 1 i groups together these microscopic states = |1i, = |2i, in which B
has the same values B = 1. Similarly, = | 2 i groups together = |3i and = |4i with
B = 1. The states of this reduced ensemble are illustrated in Table 2.4. The probabilities
of the subsystem states are
p1 = p1 + p2 ,
p2 = p3 + p4 .
(2.35)
(2.36)
Using B1 = 1 and B2 = 1, which are the classical observalbles in the subsystem, the
expectation value of B can be computed from the subsystem:
hBi = p1 B1 + p2 B2
= p1 p2
(2.37)
(2.38)
= p1 + p2 p3 p4 .
(2.39)
As it should be, this is the same as computed from the underlying microscopic system:
hBi =
4
X
B p .
(2.40)
=1
In this case, B is a system observable, which means that the information in the system
is enough to compute the distribution of the measurement values. In contrast, A and C
are environment observables.
2. Consider the case when A is the system observable of the subsystem. Thus, the system
becomes an eective three-state system with
A1 = 2 ,
A2 = 0 ,
A3 = 2 .
(2.41)
The probabilities for these states are

p1 = p1 ,
p2 = p2 + p3 ,
p3 = p4 .
(2.42)
State | 2 i of the subsystem groups together the states |2i and |3i of the microsystem that
both have A = 0. The states | 1 i and | 3 i of the subsystem are identical to the microstates
|1i and |4i with A = 2 and A = 2. We will continue examining this example in
section 2.4.
19
| 1 i
| 2 i
Table 2.4.: States of the reduced ensemble in example 1
2.4. Probabilistic observables

In every state a classical observable has a xed value A . Is this the most general case? No,
more general is the concept of probabilistic observables. In every state a probabilistic observable
takes its possible measurement values B with a certain probability w ( B ).
Examples
1. Water on earth:
Divide all probes of H2 O-molecules into three states (using some criterion):
= 1: solid (ice)
= 2: liquid (water)
= 3: gas (vapor)
So, p are the probabilities to nd water in the form of solid, liquid or gas. One would
like to determine the average density of water on earth, and the dispersion. Therefore,
one needs the probability distribution of density in each of the three states w 1 (n), w 2 (n),
w 3 (n). (the distributions depend on the state, but in each state the density has no xed
value). Then, the average density is given by

hni = p1 dn n w 1 (n) + p2 dn n w 2 (n) + p3 dn n w 3 (n)
(2.43)
= p1n1 + p2n2 + p3n3 ,
where
(2.44)

n =
dn n w (n)
is the mean value in state . We use normalized probability distributions:
(2.45)
dn w (n) = 1.
The Motivation of these concepts is, to characterize microstates by density and state
of water. So, we have microstates (n, ) with probability distribution P (n, ), obeying
X
dn P (n, ) = 1 .
(2.46)
In this setting, n is a classical observable with xed value n in each state (n, ). It is

hni = dn P (n, ) n ,
(2.47)

hn2 i = dn P (n, ) n 2 .
(2.48)
20
2.4. Probabilistic observables
Dene probabilities for states of reduced system:

p =
dn P (n, ) ,
(2.49)
P (n, )
.
p
(2.50)
dn w (n) = 1 .
(2.51)
and probabilities to nd n in a state :

w (n) =
Their normalization is given by
Then,
P (n, ) = p w (n) ,
and we can rewrite (2.47) and (2.48) as
X
hni =
dn w (n) n
hn i =
2
p n ,
(2.53)
p (n2 ) .
(2.54)

p
(2.52)
dn w (n) n2 =
One nds (n2 ) , (n ) 2 due to uctuations of n in state . Therefore,

X
hn2 i ,
(n ) 2 .
(2.55)
For hni the specication of n for every state is sucient (at this point no distinction from
classical observable), but hn2 i needs additional information about distribution of values n
in each state .
n is a probabilistic observable on the level of states .
2. Subsystem of states with total spin:
Probabilistic observables typically occur for subsystems. For example we can identify
the two-spin system with the microstates (n, ) of water. Then, the three states
characterized by total spin A in subsection 2.2.2:
A1 = 2 ,
A2 = 0 ,
A3 = 2 ,
(2.56)
are analogous to the states ice, water, vapor in the rst example.
21
For this subsystem the rst spin B is a probabilistic observable. As B1 = 1 and B3 = 1, B

behaves as a classical observable in the states = 1, 3. But what about B in state = 2?
In that case, there is no xed value for B. Instead, we have the probabilities
p2
p2 + p3
p3
w 2 (1) =
p2 + p3
w 2 (1) =
to nd B = 1 ,
(2.57)
to nd B = 1 .
(2.58)
The mean value of B in state = 2 is

p2 p3
B2 = w 2 (1) 1 + w 2 (1) (1) =
.
p2 + p3
(2.59)
hBi = p1 B2 + p2 B2 + p3 B3
p2 p3
= p1 + (p2 + p3 )
p4
p +p
| {z } |2{z }3
(2.60)
Let us check this:
p2
(2.61)
B 2
= p1 + p2 p3 p4
q. e. d.
(2.62)
However,
(B2 )2 = 1 ,
(2.63)
which diers from (B2 ) 2 ! Thus,

hB 2 i = 1 ,
3
X
p B2 .
(2.64)
=1
This is a clear sign of a probabilistic observable. And probabilistic observables need

additional information about the probability distribution of measurement values in a
given state, which we have sketched in Figure 2.1. So, we conclude:
B environment observable
additional information
B system observable.
(2.65)
We have listed all the observables that we just have computed in Table 2.5.
3. Quantum mechanics:
Consider two-state quantum mechanics:
=1:
=2:
where Sk = ~2 Sk .
22
Sz = 1 ,
Sz = 1 ,
(2.66)
(2.67)
2.5. Quantum statistics and density matrix
Figure 2.1.: Probability distribution of the probabilistic variable B in given state
| 1 i
p
A
N
B
B2
p1
2
2
1
1
| 2 i
| 3 i
p2 + p3 p4
0
2
1
0
p2 p3
1
p2 +p3
1
1
Table 2.5.: Observables in the reduced system in example 2
The additional observable Sx has no xed value in the states , only probabilities to nd
Sx = 1:
w 1 (1) = a ,
w 1 (1) = 1 a ,
(2.68)
w 2 (1) = b ,
(Sx )1 = 2a 1 ,
w 2 (1) = 1 b ,
(Sx )2 = 2b 1 .
(2.69)
(2.70)
Its expectation value is given by

hSx i = p1 (2a 1) + p2 (2b 1)
= 2p1a + 2p2b 1 .
(2.71)
(2.72)
So,
1 hSx i 1 ,
(2.73)
where hSx i = 1 for a = b = 1 and hSx i = 1 for a = b = 0.

In quantum statistics the expectation value is computed as
,
hAi = tr (A)
(2.74)
23
where A is the operator associated with A and is the density matrix.

Let us look at the particular case of a pure state. For pure states the density matrix can be
expressed as
= ,
(2.75)
where is a complex vector eld. Then, we nd
i = A
h |A|
(2.76)
= A
(2.77)
= A
.
= tr (A)
(2.78)
(2.79)
2.5.2. Dispersion
Accordingly, the dispersion is given by
hA2 i = tr (A2 ) ,
(2.80)
A = hA i hAi .
2
(2.81)
2.5.3. Diagonal operators

When A is diagonal:
A1 0 0
*.
.. +/
.
.. 0 A2 . .
. //
A = . . .
/,
.
.. ..
. . . . 0 //
, 0 0 AN A = An ,
(2.82)
(2.83)
then the expectation value is just
hAi = tr (A)
X
=
A
(2.84)
(2.85)
A .
(2.86)
So, one only needs the diagonal elements of the density matrix. They can be associated with
probabilities p = , where 0. This way, we recover classical statistics:
hf (A)i =
24
p f (A ) .
(2.87)
Example
A two-state system:

11
Let A = A01 A02 and = 12
12
22 ,
then:
!
!
!
12
A1 0
A1 11 A2 12
=
,
22
0 A2
A1 12
A2 22
A = 11
12
= A1 11 + A2 22 ,
tr (A)
X
X
hAi =
A =
p A .
n
(2.88)
(2.89)
(2.90)
2.5.4. Properties of the density matrix

The density matrix is a N N matrix with the following properties:
1. Hermicity: = , so all eigenvalues n are real
2. Positivity: n : n 0
3. Normalization: tr = 1, so
n = 1 and n : 0 n 1
2.5.5. Diagonal elements of define the probability distribution

From the properties in subsection 2.5.4 it follows that 0 and
the properties of a probability distribution.
= 1. So, { } has all
Example
Quantum mechanical two-state system:
In this case, the density matrix is a complex 2 2 matrix of the form
!
11
12
=
,
12 1 11
with 11 R, 0 11 1. Consider the observable
!
1 0
A =
.
0 1
(2.91)
(2.92)
We can nd its expectation value:

"
!
!#
~
1 0
11
12
hAi = tr

(2.93)
0 1
12 1 11
2
!
~
11
12
= tr
(2.94)
12
1 + 11
2
~
= (2 11 1) .
(2.95)
2
From the classical statistics we have hAi = p1A1 + p2A2 . If we set p1 = 11 , p2 = 1 11 ,
A1 = ~2 , A2 = ~2 , then we will get the same result (2.95).
25
11 (1 11 )
1
4
11
1
2
Figure 2.2.: Positivity condition for two state density matrix
2.5.6. Positivity constraints for the elements of

To determine the eigenvalues n of the density matrix (2.91), we have to solve
0 = ( 11 )(1 11 ) f
(2.96)
= + 11 (1 11 ) f ,
(2.97)
where
f = | 12 | 2 0 .
(2.98)
(2.99)
We nd
1,2 =
1
2

1 4[ 11 (1 11 ) f ] .
Using the positivity conditions for the eigenvalues, 1,2 0, we arrive at

11 (1 11 ) f 0 ,
(2.100)
11 (1 11 ) 0 .
(2.101)
and therefore,
The full positivity condition for the two-state matrix is
f = | 12 | 2 11 (1 11 ) .
(2.102)
In Figure 2.2 we have plotted 11 (1 11 ). From the positivity condition it follows that | 12 | 2
only can take values on or below this curve.
2.5.7. Non-diagonal operators

One may ask why we need the information in 12 , as for the expectation value it holds hAi =
P
A , provided that A is diagonal. The answer is that 12 carries additional information

for the probabilistic observables.
26
hSz i
hSx i
2
hS x i, hSz2 i
|1i
|2i
12 + 12
1
12 + 12
1
Table 2.6.: Mean values of spin observables in the states |1i and |2i
Example
Spin in arbitrary direction Si = ~2 Si :
The spin components Sz and Sx are dierent types of observables:
Sz : Classical observable

Sz = 1 ,
Sz = 1 ,
1
2
Sz2 = 1 ,
Sz2 = 1 .
1
(2.103)
(2.104)
Sx : Probabilistic observable
There is a probability distribution for nding the values +1 or 1 in the states |1i
and |2i:
hSx i1 = hSx i2 = 12 + 12
.
(2.105)
But in both states the square has a xed value:

Sx2 = Sx2 = 1 .
1
(2.106)
For a better overview, we have listed these observables in Table 2.6.

The probabilities to nd Sx = 1 in state |1i are given by
w 1 (1) = 12 (1 + 12 + 12
),
w 1 (1) =
1
2 (1
12
12
),
(2.107)
(2.108)
such that
w 1 (1) + w 1 (1) = 1 .
(2.109)
With these probabilities we arrive at
hSx i1 = w 1 (1) w 1 (1) = 12 + 12

.
(2.110)
Both, w 1 (1) and w 1 (1), have to be positive, which can be inferred from the condition
| 12 + 12
| 1,
(2.111)
or equivalently
| Re 12 |
1
2
(2.112)
27
This follows from (2.102):

| Re 12 | =
| 12 | 2 | Im 12 | 2
(2.113)
11 (1 11 ) | Im 12 | 2
(2.114)
11 (1 11 )
(2.115)
1
2
(2.116)
By similar calculations we nd the same results for w 2 (1) and w 2 (1).

We shall check if classical statistics (with probabilistic observables) and quantum statistics
produce discrepancies:
hSx i = p1 hSx i1 + p2 hSx i2
Classical:
= 11 ( 12 +
= 12 +
12
)
(2.117)
+ (1 11 )( 12 +
12
12
)
(2.118)
(2.119)
= 2 Re 12
(2.120)
"
Quantum:
!
!#
0
1
11
12
hSx i = tr
12
22
1 0
!
11
= tr 12
22 12
(2.121)
(2.122)
= 12 + 12
(2.123)
= 2 Re 12 .
(2.124)
One can see that (2.120) and (2.124) give exactly the same result.
Actually, this could be generalized for any hermitian quantum operator
!
A11 A12
A=A =
.
A12 A22
(2.125)
The probabilistic observable has mean values in the states |1i and |2i, given by hAi1 and hAi2 :
hAi1 = A11 + 12
A12 + 12 A12
hAi2 = A22 +
12
A12
12 A12
(2.126)
.
(2.127)
It turns out that quantum mechanics can be interpreted as classical statistics, but with additional
information for probabilistic observables. As long as we are dealing with diagonal observables,
there is no dierence between both approaches. In this case, non-diagonal elements of are of
no importance. Actually, the last statement is only true for a given point in time or for a static .
Non-diagonal elements of are important for the time evolution of the probability distribution:
{p (t )} { (t )} ,
where the sum over repeated indices is implied.
28
(2.128)
2.5.8. Pure quantum states

Consider a special form of the density matrix = , where is a complex vector with
P
N components and = 1. In other words, is a wave function. Then,
hAi = tr (A)
= A
= A
(2.129)
(2.130)
(2.131)
i,
= h |A|
(2.132)
where, again, the sum over repeated indices is implied, and

p = = | | 2 0 .
(2.133)
As already mentioned, only is needed for diagonal observables. So, for diagonal observables
there is no dierence between pure and mixed states.
We can state the condition for pure states as
2 = ,
(2.134)
(2.135)
because
(2.136)
= .
(2.137)
2.5.9. Change of the basis by unitary transformation

A unitary transformation is of the kind 0 = U U , where U U = 1. 0 has still all
the properties of the density matrix: ( 0 ) = 0, tr = 1, positivity. Operators also can be
. Nevertheless, the expectation values are invariant
transferred unitarily: A A0 = U AU
under this transformation:
hAi = tr ( 0A0 )
(2.138)
)
= tr (U U U AU
)
= tr (U AU
(2.139)
= tr (U U A )
= tr (A ) .
(2.141)
(2.140)
(2.142)
Let us list several important properties:

can be non-diagonal in the general case.
Even if A is diagonal, U AU
If can be diagonalized, then 0 = diag ( 1 , 2 , . . . , N )
29
lz
pure states
1
*.
.0
,0
density matrix
probability
-1
1
*. +/
.0/
,00
0
0
0
*. +/
.1/
,00
1
0
0
*. +/
.0/
,10
0
0
0
+
0//
0-
0
*.
.0
,0
p1
0
+
0//
0-
p2
0
*.
.0
,0
0
+
0//
1-
p3
Table 2.7.: z-component of the angular momentum
Pure states are invariant:

2 =
02 = 0 .
(2.143)
One can see this via

02 = U U U U
= U U
2
= UU
(2.144)
(2.145)
(2.146)
= .
0
(2.147)
From this it follows that n2 = n , leaving only the possible values n = 0, 1. This means that
0 = and therefore,
0 = ei ,
(2.148)
= ei ,
(2.149)
which is a pure state.
2.5.10. Pure and mixed states

Consider an observable l = 1, which could be associated with the angular momentum. Then, lz
could be -1, 0 or 1. The system is described in Table 2.7.
Of course, p1 + p2 + p3 = 1 must be satised. The density matrix of the system is
p
0 0
* 1
+
= .. 0 p2 0 // .
, 0 0 p3 -
(2.150)
This is a mixed state as long as there are pi , 0, 1. One could check and see that in this case
p2 0 0
*. 1 2
+
= . 0 p2 0 // , 0 .
2
, 0 0 p3 2
30
(2.151)
2.6. Microstates
Example
Let us examine a system characterized by the following density matrix:
1
0 0
*. 3 1
+
= . 0 3 0 // .
(2.152)
1
,0 0 3 Here, all the states lz = 1, 0, 1 have equal probability, so hlz i = 0. If we compare this
with a pure state:
1
1 * +
= ..1//
(2.153)
3 1
, leading to the density matrix
1 1 1
+
1 *.
= .1 1 1// .
(2.154)
3
1
1
1
,
-
we can see that both, (2.152) and (2.154), have the same diagonal elements. Consequently,
hlz i = 0, again. They dier, though, in the mean values of the probabilistic observables,
because of the dierent o-diagonal elements.
2.6. Microstates
Back to the quantum-mechanical point of view: provided that | i is a complete orthonormal
basis, then are microstates.
Examples
1. Hydrogen atom:
Here, = (n,l,m,s)
is a multiple index. We assume that the microstates are countable,

although it is commonly otherwise. Often, they are also limited, e. g. only microstates
with E < E50 . The energy of a state with quantum number n is given by
E=
E
.
n 2
(2.155)
Let us consider the following states:

1,2 :
3,...,10 :
n = 1, l = 0, m = 0, s = 1; E = E 0 ,
E0
n = 2, l = 0, 1, m, s; E = ,
4
(2.156)
(2.157)

where N = 2 + 8 = 10. The energy operator is given by E = E 0 diag 1, 1, 14 , . . . , 14 . Let

1
1
the distribution be uniform: = diag 10
, . . . , 10
. Then, the expectation value of the
energy is

1
hEi = 10
2 + 84 (E 0 ) = 0.4 E0 .
(2.158)
31
cos
Lx

mx =
1
2
sin
2
Lx
mx = 1
x
cos
3
Lx

mx =
3
2
Figure 2.3.: Wave functions of a particle in a box
2. A particle in a box:
The particle is described by the wave equation
~
= E .
2M
(2.159)
Consider a cubic box with an edge length L. Then, the solution is found to be
=e
=e
ipx x
~
ip x
~
ipy y
~
ipz z
~
0 ,
(2.160)
(2.161)
where
1 2
p .
(2.162)
2M

The periodical boundary conditions give L2 ,y,z = L2 ,y,z (and the same for y and
z). Then px,y,z have discrete values:
E=
p=~
2
m,
L
(2.163)
with
m = (m x ,my ,mz ) ,
m x,y,z Z .
Three examples of these periodical wave functions are illustrated in Figure 2.3.
32
(2.164)
2.7. Partition function
2.7. Partition function

What is the number of states for a given Energy E of a single particle (N = 1) in an onedimensional (d = 1) cubic box with edge length L? We will denote the number of states in
~2 2
[E,E + E] by (E,V ). For our purposes, assume that E 2M
. It is
L2
p2
2 ~2 n 2
=
,
2M
2M L2
(2.165)
2 ~2 n 2
< E + E .
2M L2
(2.166)
so,
E<
If E E, we can approximate
leading to
1 E
E + E E 1 +
2 E
1 E
= E+ ,
2 E
2ME
L<n<
~
2ME
1 2M
L+
~
2~ E
(2.167)
(2.168)
! 1/2
LE .
(2.169)
Therefore, the number of states (E,V ) is given by

V 2M
(E,V ) =
2~ E
! 1/2
E ,
(2.170)
which means that V !
2.8. Continuous distribution and continuous variables

We will examine the continuous description of closely situated discrete States. We need to
dene the probability density p(x ) with:

dx p(x ) = 1
(2.171)
dx p(x ) A(x ) = hAi .
We will use the following denitions:

f
I (x ): interval of states which belong to the interval x
dx
2 ,x
dx
2
): mean probability in I (x ),
p(x
(x ): number of states belonging to I (x ),
33
A(x ): hAix , i. e, mean value of A in interval I (x ) (probabilistic variable).

We can express the probability of the reduced system with states x as
p(x ) dx =
) (x ) ,
p = p(x
(2.172)
I (x )
and A(x ) is given by

P
I (x ) p A
A(x ) = hAix = P
.
I (x ) p
(2.173)
Let us check if the continuous case is equivalent to the discrete one:

dx p(x ) A(x ) =
dx p(x ) A(x )
(2.174)
X
x
! P
I (x ) p A
p
P
I (x ) p
I (x )
| {z }
X
(2.175)
dx p (x )
X X
(2.176)
p A
x I (x )
p A
(2.177)
q. e. d.

P
Note that hA2 i = dx p(x ) A2 (x ) = p A holds only, if the dispersion of A vanishes in the
continuum limit dx 0.
We shall see that p(x ) depends on the choice of the variables x. For that purpose, consider
the explicit coordinate transformation
x = f (x 0 ) ,
df
dx = 0 dx 0 .
dx
(2.178)
(2.179)
The expectation value of A in the new coordinates becomes:

hAi =

dx p(x ) A(x ) =
dx 0
df
p[f (x 0 )] A[f (x 0 )] .
0
dx
| {z } | {z }
p 0 (x 0 )
(2.180)
A0 (x 0 )
This can be generalized for many variables:

hAi =
34
dx 1 dx n p(x 1 , . . . ,x n ) A(x 1 , . . . ,x n )
(2.181)
2.9. Probability density of energy
where p(x ) is transformed by the Jacobi determinant:

df
p 0 (x 0 ) = 0 p[f (x 0 )] ,
dx
x i = fi (x j0 ) ,
df 1
* dx 10
..
.
d f
.
0 = det .. ...
.
dx
df 1
, dx n0
(2.182)
(2.183)
df n
dx 10
..
.
+/
// .
/
df n
dx n0
(2.184)

The probability density of energy can be inferred from
p(E) dE = p(E)
via
(E)
dE ,
E
(2.185)
(E) change of symbols (E) continuum limit (E)
,
E
E
E
(2.186)
leading us to
p(E) dE = p(E)
dE .
E
(2.187)
An example would be one particle in a linear box (d = 1, N = 1).
2.9.1. Equal probability for all values

We know that
V 2M
(E) (E) =
2~ E
So, we can conclude that
Then,
V
2~
p j (E) =
0
V 2M
2~ E
1
2M /2 1
Z
E
!1/2
E .
(2.188)
(2.189)
!1/2
for E < E max ,

else.
(2.190)
Here,
Z is a constant factor, which could be determined by the normalization condition

dE p(E) = 1:
!1/2
Emax
V
2M
Z=
dE
.
(2.191)
2~ 0
E
35
(a) Equipotential distribution
(b) Boltzmann distribution
Figure 2.4.: Comparison of equipotential and Boltzmann distribution
2.9.2. Boltzmann distribution for microstates

The Boltzmann distribution is given by the formula:
E
p = Z 1 e kbT = Z 1 e T .
(2.192)
We have set k b = 1, which means that we measure the temperature in energy units. E is the
energy of state . In Figure 2.4 the Boltzmann distribution is compared to an equipotential one.
We nd
!1/2
E
V 2M
pb (E) =
e T Z 1 ,
(2.193)
2~ E
Emax
hEi =
dE p(E) E ,
(2.194)
0
Emax
2
hE i =
dE p(E) E 2 .
(2.195)
0
It should be marked that p(E) is the macroscopic probability distribution and p is the microscopic one.
2.9.3. Number of one particle states in a momentum interval

Number of one particle states in a momentum interval p x , py , p z
First, the one-dimensional case:
(p) =
L
px .
2~
(2.196)
Periodical boundary conditions would give

px <
2~
n < px + px .
L
(2.197)
V
For an arbitrary dimension d we have (p) = (2~)
d p 1 pd , which in the most usual case
of d = 3 is
3
V
=
p1 p2 p3 .
(2.198)
px py pz
(2~) 3
36
Number of one particle states with energy E for arbitrary d

Using the usual energy-momentum relation, we nd
=
E
dd p
!
p2
d
E ,
p1 pd
2M
V
=
E
(2~)d
(2.199)
!
p2
d p
E .
2M
d
(2.200)
For d = 1 it is
!

p2
1
E = 2M
d /2 ( E)
2M
0
!1/2
2M
=
,
E
dp
with =
p2
2M .
We have used that |p| =

(2.202)
2ME and
dp = 2
(2.203)
d|p|
=2
1 1
2M /2 .
2
So, we arrive at
L 2M
=
E
2~ E
For d = 3 we have
(2.201)
(2.204)
!1/2
.
(2.205)
d p = 4
d|p| |p| 2 ,
(2.206)
with =
|p | 2
2M .
So, we nd
r
!

2
p
M 1/2
dd p
E = 4
d ( E)
2M
2M
2
0
| {z
} |{z}
|p |
= 4 2 M /2 E /2 ,
3
(2.207)
|p | 2
(2.208)
and the number of states per energy is given by
V
3
1
=
4 2 M /2 E /2 .
3
E
(2~)
(2.209)
37
Boltzmann probability distribution for one particle

This Boltzmann probability distribution for one particle is given by
E 1
e TZ ,
E
where Z is the normalization factor. To nd it, we use

dE p(E) = 1 .
p(E) =
So, we get
Z=
dE
0
where
= F E 1/2 . Therefore,

Z =F
and
E
e T ,
E
1
dE E /2 eE/T = F Z
p(E) = Z 1 E /2 e T
1
(2.210)
(2.211)
(2.212)
(2.213)
(2.214)
So, it is possible to nd an expression for this probability, where ~ does not longer appear.
2.10. Correlation functions

If there are two observables A and B, what is the probability that a measurement of A and
B yields specic values A and B ? Imagine that B is measured after A, then the probability
becomes p( B , A ).
2.10.1. Conditional probability p(B |A )

The probability to nd B , if A is measured for A, is called conditional probability and is
written as p( B | A ). It is
p( B , A ) = p( B | A ) p( A ) .
(2.215)
2.10.2. Product observable B A

The product observable is dened as
hB Ai =
p( B , A ) B A
(2.216)
( B , A )
XX
p( B | A ) p( A ) B A .
(2.217)
B A
Its spectrum is given by

Spec (B A) = { B A } .
hB Ai depends on conditional probabilities, not only on p( B ) and p( A ).
38
(2.218)
2.10.3. Correlation function

The (connected) correlation function is dened as
hB Aic = hB Ai hBi hAi .
(2.219)
If hB Aic , 0, then B and A are correlated.

Example
A falling marble:
Imagine a marble that is falling out of pipe on a prism, as illustrated in Figure 2.5a.
Moreover, it can roll down on any of the sides of the prism with equal probability. However,
there are two detectors A and B on the left side. The corresponding measurement values
are A = 1 and B = 1 if the marble goes through the detectors and 0 otherwise.
The probability to measure A = 1 is 12 . It is easy to see that p( B | A ) = 1. Therefore,
p( B , A ) = 12 . With this we calculate
hB Ai =
hAi =
1
2,
X
(2.220)
pA A =
hBi =
A
X
hBi hAi =
hB Aic =
1
4
1
2
1
2
pB B =
|{z}
(2.221)
1
2
(2.222)
1
2
(2.223)
1
4
1
4
(2.224)
Is hB Ai = hA Bi true? Although it is true in this case, in general it is not. A simple

example is shown in Figure 2.5b.
2.10.4. Independent observables

If the measurement of A does not inuence B, then the condtional probability is independent of
A :
p( B | A ) = p( B ) .
(2.225)
So, only for independent observables it is

p( B , A ) = p( B ) p( A ) .
(2.226)
39
1
2
1
2
1
2
1
2
A
1
2
1
2
B
(a) Two possibilities
(b) Three possibilities
Figure 2.5.: A marble falling down on a prism
But we must keep in mind that this is not true in general! In this case, the correlation function
can be written as
XX
hB Ai =
p( B ) p( A ) B A
(2.227)
B A
p( B ) B
! X
!
p( A ) A
(2.228)
= hBi hAi .
(2.229)
Then,
hB Aic = 0 .
(2.230)
It is important to note that the reverse statement is not true!
2.10.5. Classical correlation function

It is
(B A) = B A .
(2.231)
The classical correlation function is

! X
!
X
X
hBAic =
p B A
p B
p A .
{z
hBAi
} |
(2.232)
{z
hBi
}|
{z
hAi
In general, hBAic , 0. Setting B = A, we nd

hAAic = A2 .
There is an easy recipe for calculating conditional probabilities p( B | A ):
40
(2.233)
1. We eliminate all for which A , A .

2. Now we have a new system with states 0 and
p 0
p 0 = P
.
0 p 0
(2.234)
It is A 0 = A and B 0 = B .
3. The result is:
p( B , A ) =
A = A
B = B
p .
One should note that after the measurement of A the relative probabilities
p
p
(2.235)
p 0
p0
are the same as
before the measurement of A:

p 0 p
=
.
p0 p
(2.236)
In this lecture we will be mostly concerned with classical products of two observables, i. e.
BA = AB, and with classical correlation functions, which are special correlation functions.
However, if the relative probabilities change after the measurement of A, then we have dierent
conditional probabilities: B A , BA. Note that every consistent choice of conditional probabilities p( B | A ) leads to the denition of a product of two observables B A. Many dierent
products and correlation functions are possible. In general, the correct choice depends on the
way how measurements are performed. This choice is important for the statistical description of
systems with few states like atoms (with bounded energy), where the measurement of a Fermi
observable has a strong inuence on the possible outcome of the measurement of a second
observable.
2.10.6. Quantum correlations

We now have
hB Ai = tr
!
AB + BA
,
2
B being diagonal: A = diag (A ). Therefore,

with A,
X
hB Ai =
A B .
(2.237)
(2.238)
The quantum product coincides with the classical one, in this case. However, if A and B do
not commute, this is no longer correct, because in that case A and B cannot simultaneously be
diagonal.
41
3. Statistical description of many-particles

systems
N is the particle number. Typically N N Avogadro 6 1023 .
3.1. Chain with N lattice sites

3.1.1. Probability distribution
We consider N fermions being distributed over the chain. If site j is occupied by a fermion, we
denote this by s j = +1 . If it is not, we denote it by s j = 1 . The possible states of such a
system is given by = {s j } where s j = 1 for every j. So = 2 N . A possible conguration is
shown in Figure 3.1. Later, in subsection 3.2.1, we will interpret this as a chain of spins. In the
following, for notational convenience, we will denote any quantity f , which depends on the
complete microscopical state {s j }, as f {s j } f ({s j }).
Let us dene the number of fermions N . For every collection {s j }, the corresponding value of
the fermion number is
X
1
N {s j } =
(3.1)
2 (s j + 1) .
j
Lattice sites are independent and we say that the probability of a site being occupied is q. So,
1 q is the probability of it being unoccupied. Then
p{s j } =
N f
Y
j=1
1
2

g
+ q 12 s j
= q N (1 q) N N .
(3.2)
(3.3)
3.1.2. Macroscopic probability

We want to compute the macroscopic probability to nd N fermions (particles), that is p(N ):
p(N ) = q N (1 q) N N (N )
where
(N ) =
N!
N ! (N N )!
(3.4)
(3.5)
is the number of microscopic states with N particles.
43
Figure 3.1.: Fermions on a chain
It is given by the number of possible ways to distribute N indistinguishable particles on N

dierent sites:
Particle 1
N sites
Particle 2
(N 1) sites
Particle 3
(N 2) sites
(1) = N
N (N 1)
(2) =
2
N (N 1)(N 2)
(3) =
23
..
.
All in all, we nd the probability to be given by the binomial distribution

p(N ) = q N (1 q) N N
N!
.
N ! (N N )!
(3.6)
Using the general relation

(q + r )
N
X
N!
q N r N N ,
N
!
(N
N
)!
N =0
(3.7)
we can prove that p(N ) has the correct normalization

N
X
p(N ) = 1
(3.8)
N =0
by setting r = 1 q.
We conclude that there are two dierent levels at which one can characterize this system.
One can consider the 2 N microstates micro = {s j } or the N + 1 macrostates macro = N .
44
3.1. Chain with N lattice sites

The expectation value is given by
hN i =
=
=
N
X
p(N ) N
(3.9)
N =0
N
X
N!
N q N (1 q) N N
N
!
(N
N
)!
N =0
N
X
N!
q q N r N N
N
!
(N
N
)!
q
N =0
N
X
N!
=q
q N r N N
q N =0 N ! (N N )!

= q (q + r ) N
q

(3.10)
(3.11)
(3.12)
r =1q
r =1q
(3.13)
r =1q

= N (q + r ) N 1q
r =1q
= Nq .
(3.14)
(3.15)
To arrive at this result, we used the following trick: We expressed N via a q-derivative acting
on q N . Thereto we considered r to be independent of q. Only in the end, after using (3.7) and
performing the q-derivative, we substituted r = 1 q again.
3.1.4. Dispersion
The dispersion can be computed using the same trick as above:
hN 2 i =
N
X
p(N ) N 2
(3.16)
N =0
!2

= q
(q + r ) N
q
r =1q

=q
Nq (q + r ) N 1
q
r =1q
f
g
= qN (q + r ) N 1 + (N 1) q (q + r ) N 2
r =1q
2
= qN + q N (N 1)
= q 2 N 2 + N (q q 2 ) .
(3.17)
(3.18)
(3.19)
(3.20)
(3.21)
45
From the expectation value and the dispersion we can calculate the variance
N 2 = hN 2 i hN i2
(3.22)
= Nq (1 q) .
(3.23)
and the relative standard deviation

Nq (1 q)
Nq
s
1q
1
=
q
N
p
1
=
1q.
hN i
N
=
hN i
Notice that the relative standard deviation is proportional to

sharper with increasing N . For example:
N = 1020
(3.24)
(3.25)
(3.26)
1
N
! So, the distribution gets
N
1010 !
hN i
(3.27)
This is one of the main reasons why statistical physics has such a predictive power. The
statistical uncertainty is so small that one virtually can calculate deterministically
by using mean values.
3.2. Uncorrelated probability distribution

Here, uncorrelated means that the probabilities q do not depend on other lattice sites. It does
not mean that correlation functions between dierent sites have to vanish.
3.2.1. N spins on a chain

We consider a chain of N spins. Therefore the total spin S, which is a macroscopic observable,
is given by
X
S=
sj .
(3.28)
j
In quantum mechanics the spin is ~2 S. The number of -spins (compare to (3.1)) is

N =
=
46
N
X
1
2 (s j
+ 1)
j=1
1
1
2S + 2 N
(3.29)
(3.30)
and therefore
S = 2N N
(3.31)
= N (2q 1) .
(3.32)
Spin expectation value, dispersion, variance and relative standard deviation are then found to
be
hSi = 2 hN i N
(3.33)
= N (2q 1) ,
(3.34)
hS 2 i = 4 hN 2 i 4 N hN i + N 2 ,
(3.35)

2
S 2 = 4 hN 2 i hN i
(3.36)
= 4 N 2
(3.37)
= 4 Nq (1 q) ,
(3.38)
S
1
,
hSi
N
(3.39)
where the last equation only holds if hSi , 0, i. e. q , 12 .

Now, using equation (3.6), the macroscopic probability is given by

p(S ) = p N = N2+S
=q
N +S
2
(1 q)
N S
2
N!

N +S
N S
!
!
2
2
(3.40)
(3.41)
and the microscopic probability by

p{s j } =
N f
Y
j=1
N
Y
1
2

g
+ q 12 s j
P (s j ) .
(3.42)
(3.43)
j=1
The latter is a product of independent probabilities for each spin s j . That is because we assumed
uncorrelated spins, i. e. the spin s j does not know anything about its neighbours. In the
following this property is used to compute correlation functions.
3.2.2. Correlation functions

Let us compute the correlation function for l , k in a microstate = {s j }:
X
hsl sk i =
sl sk p{s j }
(3.44)
{s j }
47
Using (3.43) and the fact that the sum over all micro-states can be written as
X Y X
=
j
{s j }
X X
(3.45)
s j =1
s 1 =1 s 2 =1
(3.46)
s N =1
we nd
hsl sk i =
Y X Y

P (s j ) sl sk
j
Y X
j
(3.47)
s j =1
P (s j ) sl sk
(3.48)
s j =1
Y X
P (s j )
X X
j,l,k s j =1

P (sl )P (sk ) sl sk .
Moreover, we can use the normalisation of p{s j }

Y X
P (s j ) = 1
j
j,l,k s j =1
(3.50)
s j =1
to compute
Y X
(3.49)
sl =1 s k =1
P (s j ) = P
sl =1
s k =1 P (sl )P (s k )
(3.51)
With this, we nally arrive at

P
hsl sk i =
sl =1
s k =1 P (sl )P (s k ) sl s k
sl =1
s k =1 P (sl )P (s k )
(3.52)
Notice, that this is the same expression as for a system consisting of only two spins! We conclude
that in the case of uncorrelated spins the spins s j , j , l,k, do not matter for hsl sk i.
By a similar calculation we also get an expression for the expectation value:
P
s =1 P (s k ) s k
hsk i = Pk
.
(3.53)
s k =1 P (s k )
So, obviously
hsl sk i = hsl i hsk i ,
(3.54)
which leads to the observation that the connected correlation function, dened in (2.219) as
hsl sk ic = hsl sk i hsl i hsk i ,
(3.55)
vanishes for the system we are looking at:

hsl sk ic = 0 .
48
(3.56)
This resembles the fact that we are dealing with uncorrelated spins here.
In contrast to that, an example for a system with correlated spins is the Ising model. Its
microscopic probability is given by
P
p{s j } = e
j (s j s j+1 )
(3.57)
In this case there are next neighbour interactions, so the spins know about each other. We will
come back to this later.
3.2.3. Independent measurement

Consider a sequence of independent measurements of one spin. We denote the individual
measurements by j = 1, . . . , N . Then such a sequence is denoted as
{s j } = {s 1 ,s 2 , . . . ,s N } .
(3.58)
If the probability for one spin is P (sk ), the probability for a given sequence is
Y
p{s j } =
P (s j ) ,
(3.59)
just like throwing dice N times.

The average of s for a sequence of N measurements is
hSi
N
= 2q 1 .
hsi =
(3.60)
(3.61)
Of course, there always is some statistical error for a nite sequence, i. e. the value of the total
spin S uctuates between dierent sequences. The dispersion is computed as
S 2 = hS 2 i hSi2 ,
X
hS 2 i =
p{s j } S 2 ,
where
(3.62)
(3.63)
{s j }
S2
and
X
2
sm
(3.64)
From (3.38) we know that

S = 2
q
N
q (1 q) ,
(3.65)
which can be related to s via

S
N
p
2 q (1 q)
=
N
1
.
N
s =
(3.66)
(3.67)
(3.68)
49
Therefore, large sequences give accurate results.

But beware of the fact that the statistical measurement error s has nothing to do with the
uctuations s j of the actual spin:
s j2 = hs j2 i hs j i2
= 1 (2q 1)
(3.69)
2
= 4 q (1 q) .
(3.70)
(3.71)
So,
s j = 2
q (1 q) ,
q (1 q)
= s .
(3.72)
s is an eect of measuring, whereas s j is a quantum mechanical eect as every single

measurement either yields s j = +1 or s j = 1 (if q = 12 , s j = 1). Or in other words, s is a
macroscopic, whereas s j is a microscopic variable.
Summary
1. Dierent physical situations are described by the same probability distribution p{s j }:
a) Fermions on a lattice of N sites,
b) N independent spins,
c) Random walk (exercise),
d) Sequence of measurements.
2. One major theme in statistical physics is to compute macroscopic probability distributions
from microscopic probability distributions.
3.3. Statistical Ensembles

A statistical ensemble is constituted by specifying the states and the probability
distribution p or the density matrix , respectively.
In such an ensemble dierent operators / variables can be dened.
We can imagine an ensemble as a collection of innitely many similar, uncorrelated experimental set-ups in which we can measure the values of the state variables. From this their
probability distributions can be inferred (sequences of measurements with N ).
For example, consider an ensemble of 1-particle systems with spin, i. e. an ensemble of 2-state
systems ( = {,}). This is a sequence of measurements on a single spin and should not be
confused with systems with many degrees of freedom (many spins). These would be described
by ensembles with many lattice sites or particles.
Of course, real sequences of measurements can only be nite.
50
3.4. Gaussian distribution
3.4. Gaussian distribution

Systems with many uncorrelated degrees of freedom are approximated by a Gaussian
distribution. This approximation gets exact in the limit of innite number of particles
N .
The continuous description is given by
p(x ) = c e
(x x ) 2
2 2
(3.73)
where x = hxi is the mean value of the macroscopic variable x, = x = x 2 is the variance
and c = c ( ) is a normalization constant which is determined by demanding

1=
dx p(x )
= c ( )
(3.74)
2 2 .
(3.75)
1 1
c ( ) =
.
2
The generalization to several variables x i is given by:
(3.76)
We nd
p = c e 2 Ai j
1
(x i xi )(x j x j )
(3.77)
where Ai j are the components of a positive denite symmetric Matrix A, i. e. a matrix with only
positive eigenvalues. As an example we can think about fermions on a 1-dimensional chain.
We will show that it is the correct large N limit of the binomial distribution
p(N ) =
N!
q N r N N ,
N ! (N N )!
r = 1q.
(3.78)
Using Stirling's formula

lim ln(N !) = N + 12 ln N N + 12 ln (2) + O(N 1 ) ,
(3.79)
we get
(N ) = ln p(N )

= N + 12 ln N N + 12 ln (2)

N + 12 ln N + N 12 ln (2)

N N + 12 ln (N N ) + N N 12 ln (2)
(3.80)
(3.81)
+ N ln q + (N N ) ln (1 q)

= N + 12 ln N N + 12 ln N N N + 12 ln (N N )
+ N ln q + (N N ) ln (1 q) 12 ln (2) .
(3.82)
51
It can be shown that

(N )
(N N 2 )
,
2 N 2
(3.83)
where
N = Nq ,
(3.84)
N = Nq (1 q) ,
2
(3.85)
for N close to the maximum N .

The fact that the maximum is given by N = Nq, can be proven by computing the rst
derivative of :

0=
(3.86)
N
N =N
N + 12
N N +
+ ln (N N ) +
N
N N
1q
N N
= ln
ln
+ O( N 1 ) .
q
N
= ln N
So, up to N 1 corrections we nd
1
2
+ ln q ln (1 q)
(3.87)
(3.88)
1q
N N
=
,
q
N
(3.89)
N = Nq .
(3.90)
and therefore
Moreover, from the second derivative we can infer the variance:
1
2

=
N 2 N 2
(3.91)
N =N
1
1 N
1
1
=

2
+
2
N N N
N (N N ) (N N ) 2 N
N
=
+ O( N 2 ) .
(N N ) N
!
(3.92)
(3.93)
So, using N = Nq, up to N 2 corrections it is
1
N
=
,
N 2
(N Nq) Nq
(3.94)
and therefore
N = Nq (1 q) .
(3.95)
Furthermore, the third derivative yields
3
N2

=
N 3 N =N
(N N ) 2 N 2
(3.96)
All in all, this leads to the central limit theorem of statistics which states that any uncorrelated
ensemble can be described by a Gaussian probability distribution in the limit N .
52
3.5. Thermodynamic limit
3.5. Thermodynamic limit

The thermodynamic limit is given by V or N , respectively. In this limit the Gaussian
distribution becomes extremely narrow. For example
N = 1020
N 2 = 2N = 2 1020 ,
(3.97)
when q = 1/2. If we then consider a small deviation

N = N (1 + 105 ) ,
(3.98)
(N N ) 2
p(N )
= e 2 N 2
p( N )
(3.99)
1010 1040
4 1020
(3.100)
the relative probability is given by
= e
=e
14 1010
10
109
(3.101)
.
(3.102)
And for a deviation

it is given by
N = N (1 + 1010 )
(3.103)
1
p(N )
= e 4 = O(1) .
p( N )
(3.104)
N 5 N . N . N + 5 N
(3.105)
Therefore, only values

play a role. And in this regime the deviations of the actual probability distribution from a
Gaussian distribution are extremely small!
53
Equilibrium states are described by statistical ensembles with a static probability
distribution.
If the probability distribution {p } is time independent, then all expectation values of the observables are time independent, too. One example would be an isolated system after suciently
long time or, respectively, just parts of the isolated system. Another one would be an open
system being immersed in a heat bath.
Basic postulate for equilibrium states
An isolated system approaches an equilibrium state. After suciently long time an
isolated system can be described by an equilibrium ensemble.
This conjecture was made by Boltzmann, but was never proven. Actually, it most probably is not
true, as there are two problems with it. First, the necessary time scale could be extremely large.
For example think of systems like water, honey or glass. And second, the microscopic equations
are invariant under time reversal, but the time direction always points towards equilibrium.
But even though the basic postulate may not be strictly true, and despite the existence of
important practical systems which do not equilibrate, the basic postulate is useful for many
applications and it often provides very good approximations.
4.1. Microcanonical ensemble

In the following we will be examining an isolated system in a volume V with xed energy E
and xed number of particles N .
4.1.1. Fundamental statistical postulate

Microstates
These are the states of a statistical ensemble, given by all possible quantum states with specied
V , E, N . Note that in quantum mechanics one has to choose a complete basis of eigenstates to
E, N in a given volume V . Then one can identify
.
(4.1)
Or equivalently: An isolated system in equilibrium can be found with equal probability in each
of its allowed states.
55
Microcanonical ensemble
The number of microstates with given E and N is denoted by (E, N ). Then the
probability of the system being in one these is given by
p =
1
.
(E, N )
(4.2)
This raises several arguments in favour of the basic postulate for equilibrium states. To begin
with, no state is preferred to another state 0. In addition to that, this so-called equipartition
of states (i. e. equal p ) is constant in time. But one has to be careful. The existence of an
asymptotic state for t is only valid for large enough N . Moreover there is an obstruction
due to conserved correlations, i. e. a system whose initial correlations dier from the ones of
the microcanonical ensemble cannot reach the latter asymptotically.
Note that for a microcanonical ensemble all expectation values and classical correlations for
classical observables are uniquely xed.
4.1.2. Microcanonical partition function

The microcanonical partition function is denoted by Z mic , where
Z mic = (E) ,
p =
1
Z mic
(4.3)
(4.4)
Z mic (E, N ,V ) depends on the thermodynamical variables E, N ,V . Recall that (E) (E) is
the number of states with energy in between E and E + E.
4.1.3. Entropy S
For a microcanonical ensemble the entropy is given by:
S = k b ln ,
(4.5)
where k b is the Boltzmann constant. Setting it to 1 simply means to measure the temperature in
energy units, e. g. eV. Then we have
S (E, N ,V ) = ln (E, N ,V ) .
(4.6)
This equation is valid for isolated systems in equilibrium. The entropy S is an extensive
thermodynamic potential, i. e. it is additive when uniting subsystems. One can conclude that,
in the limit of large N , the entropy is proportional to N :
SN
56
for
N .
(4.7)
Examples
1. Consider N uncorrelated particles where each of them can be in F dierent states. Furthermore, say the energy is given by E = cN , independent of the particles states. Then
there are F N possible microstates.
= FN
(4.8)
ln = N ln F
(4.9)
S = k b N ln F
(4.10)
2. Consider N uncorrelated particles on N lattice sites with E = cN and N = bV .

N!
N ! (N N )!
S = ln N ! ln N ! ln (N N )!
(4.11)
(4.12)
Applying Stirling's formula on ln N ! for large N :

ln N ! N ln N N
(4.13)
one nds
S = [N ln N N ] [N ln N N ] [(N N ) ln (N N ) (N N )]
N N
N
= N ln
+ N ln
N !
N N !
N
N
= N ln
1 N ln 1
N
N
"
!
!#
bV
bV
N
= N ln
1
ln 1
.
N
N
bV
(4.14)
(4.15)
(4.16)
(4.17)
Now it is obvious that S is extensive, since one can write it as N times a function of the
particle density n = VN :
S = N f (n) .
(4.18)
For small n we can approximate
"
!
#
b
S = N ln
1 +1 .
n
(4.19)
4.1.4. Partition function for ideal classical gas

To calculate the partition function for an ideal classical gas, we have to count the number of
quantum states of N particles in a box with volume V (e. g. with periodic boundary conditions)
for a xed total energy E. We will approach this step by step.
57
1) Quantum states for one particle (N = 1)

p = (px ,py ,pz )
Assigning an occupation number n(p) to each of these states, we nd
X
1=
n(p) ,
(4.20)
(4.21)
E=
n(p)E(p) ,
(4.22)
with E(p) =
p2
.
2M
(4.23)
The partition sum can then be expressed as

X

=
p 2
p
2M
!
p2
3
= dp 3
E E
p
2M
!

p2
V
3
=
d
p
E
E
2M
(2~) 3
V
3
1
=
4
2 M /2 E /2 E .
3
(2~)
3
=E
(4.24)
(4.25)
(4.26)
2) Quantum states for two particles (N = 2)

Using the same denitions as in the one particle case, we now get
X
2=
n(p) ,
(4.27)
E=
n(p)E(p) .
(4.28)
We should discriminate between fermions, for which n(p) {0, 1}, and bosons, for which n(p)
can take arbitrary values. In any case, we can split up the partition sum into two parts:
= 1 + 2 .
(4.29)
Here, 1 is the number of states with n(p 1 ) = 1, n(p 2 ) = 1 for some p 1 , p 2 , and 2 is the
number of states with n(p) = 2 for some p. These two contributions are given by
1 =
1X X
2 p p ,p
1
E=
p 21 +p 22
2M
2 = (b) P
2 = p p 2
E= M
58
(4.30)
for fermions,
for bosons.
(4.31)
The factor 12 in (4.30) is due to the fact that we are dealing with indistinguishable quantum
mechanical particles. So, swapping both particles does not count as a dierent state.
The full partition sum then takes the form
!
1 X X
1 (b)
bosons

=

.
(4.32)
fermions
2 p p E= p 21 +p 22 2 2
2M
1
2
Note that in (4.32) the p 2 sum runs over all possible momentum values, whereas in (4.30) it
excludes p 2 = p 1 .
If we neglect the 12 2(b) term in (4.32) we arrive at the approximate description of a classical
gas. Then there is no distinction between fermions and bosons any more. Performing the two
momentum sums analogously to the one particle case leads to
!2
!

p 21 + p 22
1
V
3
3
=
d p1 d p2
E E .
(4.33)
2
2M
(2~) 3
3) Quantum states for N particles
The extension to the general N particle case is straight-forward:
X
N =
n(p) ,
(4.34)
E=
n(p)E(p) ,
(4.35)
1
=
N!

3
d p 1 d pn
V
(2~) 3
!N
N p2
X
j
j=1
2M
!
E E .
(4.36)
As in the two particle case, equation (4.36) neglects the particular behaviour when two or more
momenta commute. Again, this is the classical approximation where no distinction between
fermions and bosons is made.
4) Partition function
We dene
=
.
(4.37)
E
E
To nd , we have to evaluate a 3N -dimensional integral. For this purpose, we will perform a
change of variables:
p 2j
p j,k = 2ME x j,k ,

= x 2j E .
(4.38)
2M
where the x j are chosen such that
X
x 2j E = E .
(4.39)
=
59
Then
d p1 d p N = (2ME)
3
3N/2
d3 x 1 d3 x N ,
(4.40)
X p 2j
2M
X
j
x 2j E .
(4.41)
For our purposes, we also introduce the 3N -component vector y = (x 1,1 ,x 1,2 ,x 1,3 ,x 2,1 , . . . ).
Now, nally, we can calculate :
1
V
=
N ! (2~) 3
!N
(2ME)
1
E
dy1 dy3N
1 N ME
=
V
N!
22 ~2
The factor
3N/2
! 3N/2

E
(4.42)
(4.43)
F=
yi2 E
1
F .
E
comes from applying the relation (E X ) =

X
dy1 dy3N 1
(X ) and
1
E
3N
X
i=1
!
yi2
(4.44)
is the volume of the (3N 1)-dimensional unit sphere S 3N , which is equal to the surface of a
3N -dimensional unit ball. It can be calculated from the general formula
F=
=
3N
2
3N/2
2
3N
2

1 !
(4.45)
3N/2
(4.46)
As a consistency check we compute (4.46) in the case N = 1:

F=
2
1
2!
/2
3
(4.47)
4 3
= /2
= 4
q. e. d.
Here we have used 12 ! =
partition sum is given by

60
3
2
1
2
2
N!
3N
2
(4.48)
(4.49)
in the second line. Using the denition (4.37), the
M

2
1 ! 2~
! 3N/2
E
3N/2
VN
E
.
E
(4.50)
Finally, using
E
E
1
3N
, we arrive at
Z mic = =
N!
1

3N
2
M
2
2~
!
! 3N/2
E
3N/2
VN .
(4.51)
N is typically of the order 6 1026 . Therefore, Z mic increases extremely rapidly with E and V !
4.1.5. Entropy of an ideal classical gas

With the help of our favorite approximated Stirling's formula (4.13) we can nd the entropy of
an ideal classical gas S:
!
3N
M
3N
3N
S=
ln
+
ln
E
+
N
ln
V
ln
N
!
ln
!
(4.52)
2
2~2
2
2
(
)
1
M
1
1
1
1 1 3N 1
= k b 3N
ln
+ ln V + ln E ln N + ln
+
.
(4.53)
2 2~2 3
2
3
3 2
2
2
So,
(
S = k b 3N
)
1 V
1 2E 1
M
5
ln + ln
+ ln
+
.
3 N 2 3 N 2 2~2 6
(4.54)
3N = f is the number of degrees of freedom. In a compact form we can write:
and therefore
S = c kb f
(4.55)
(E) E cf ,
(4.56)
where c depends on the system. So, the partition sum increases rapidly with increasing energy.
This form of the entropy and the partition sum applies too many other systems, too.
It is important to note, that the factor N1 ! in (E) was necessary for the entropy to be an
extensive quantity. This factor is absent in the classical description! It was rst introduced by
Max Planck in 1900 and was heavily discussed back then.
4.1.6. Thermodynamic limit

The thermodynamic limit is described by
V ,
N ,
E .
(4.57)
In this limit the useful quantities to describe the system are:

N
=n
V
E
=
V
(particle density) ,
(4.58)
(energy density) ,
(4.59)
61
Extensive
Intensive
p (pressure)
n
s = VS (entropy density)
V
N
E
S
Table 4.1.: Examples for extensive and intensive quantities
as these stay constant in the limiting process. So, mathematically more rigorously we dene
the limiting process via an auxiliary parameter a:
V aV ,
N aN ,
E aE
(4.60)
and consider the limit

a .
(4.61)
We make the distinction between extensive quantities (A), where

hAi
= nite , 0 ,
V V
and the intensive quantities (B), where
lim
lim hAi = nite.
(4.62)
(4.63)
Some extensive and intensive quantities are listed in table Table 4.1.
Again note, that the entropy is an extensive quantity:
S = 3k bnV c (n, ) ,
(4.64)
where
1 2 1
1
M
5
ln
ln n + ln
+ .
(4.65)
2
2
3n
3
2 2~
6
On the other hand, ratios of extensive quantities, such as n, and s, are intensive quantities.
By setting k b = 1 and ~ = 1 we nd the entropy density to be given by
c (n, ) =
s=
S
V
(4.66)

1/2

1
1
= 3n ln n /3 + ln
+ ln M /2 + const. .
n
(4.67)
So
r
s = 3n ln
!
M
+ const. .
n5/3
(4.68)
In general we have to be careful with units, as we can only take logarithms of dimensionless
numbers. But a quick dimensional analysis conrms this here:
M M1 ,
(4.69)
M ,
(4.70)
nM .
(4.71)
4
3
62
4.1.7. The limit E E 0

We now denote by E 0 the quantum mechanical energy of the ground state and choose E to be
an energy dierence smaller than the dierence between E 0 and the lowest excited state E 1 :
E < E 1 E 0 .
(4.72)
So, if the ground state is non-degenerate, then

lim (E) = 1
(4.73)
lim S (E) = 0 .
(4.74)
EE 0
and as a consequence
EE 0
On the other hand, if we have a q-fold degeneracy, then

lim S = k b ln q k b f .
EE 0
(4.75)
So, often the assumption is made that S vanishes in this limit, as ln q is tiny compared to the
usually huge f . Then
S
0
(4.76)
V
in the thermodynamic limit.
4.1.8. Derivatives of the entropy

Temperature
First recall that up to now, the partition sum and the entropy have been functions of the state
variables E, N , and V :
Z mic = (E, N ,V ) ,
(4.77)
S (E, N ,V ) = k b ln (E, N ,V ) .
(4.78)
Now we take the derivative with respect to E at xed particle number and volume:
1
S

=
.
T
E N ,V
(4.79)
This denes the temperature T = T (E, N ,V ). It can be used as a new thermodynamic state
variable. So, we can make a variable substitution in order to express the entropy not as a
function of E but of T: S = S (T , N ,V ). Moreover, we see that the temperature is an intensive
quantity, as S and E are extensive and so their ratio must be intensive.
63
Pressure
Now we take the derivative with respect to V at xed particle number and energy:
p
S

=
.
T
V E,N
(4.80)
This denes the pressure p. And it can immediately be concluded that the pressure is an intensive
quantity.
Let us know compute this for our ideal gas of particles:
!
S
V
= kb
N ln + f (E, N )
(4.81)
V
V
N
N
= kb
(4.82)
V
p
= .
(4.83)
T
From this we can infer the equation of state for an ideal classical gas:
pV = k b NT .
(4.84)
In thermodynamics T and p are determined by the change of the number of available states
when changing E and V .
Chemical potential
Now we take the derivative with respect to N at xed volume and energy:
S

= T
N
(4.85)
E,V
This denes the chemical potential .

We see that we can compute all these variables, if we just have the entropy at hand. So,
basically the entropy already contains the whole thermodynamics! From here on, one can
develop the complete formal machinery of thermodynamics, resting upon the thermodynamic
potentials, such as S, E, F and G, and their derivatives (Legendre transformations).
Note that up to now T is only a denition, but it will be motivated in the following.
We conclude this section by giving a little summary about entropy in the case of a microcanonical
ensemble description of a system in equilibrium:
S = k b ln
S is extensive, i. e. S V , S f N
64
S
V
0 for E E 0 (3rd law of thermodynamics)
1
T
S (E,N ,V )

E

N ,V
(important part of the 2nd law of thermodynamics)
4.2. Canonical ensemble
Reservoir
System
R
S
E
Figure 4.1.: System connected to a reservoir with energy exchange

4.2.1. System in a heat bath
A canonical ensemble describes a system in a heat bath, intuitively speaking. We consider a
closed system with a volume V . So, energy can be exchanged with the environment, but the
particle number N remains constant. Additionally, there is a heat bath characterized by the
temperature T . In the following we will derive all properties of this compound system.
The system S is connected to the reservoir (heat bath) R with which it can exchange energy,
but no particles. This is illustrated in Figure 4.1. But it is important that the total system is
isolated. Moreover we demand that
Vr Vs ,
(4.86)
Nr Ns 1 ,
(4.87)
i. e. the reservoir R is much larger than the system S concerning volume as well as particle
number.
The total system can be described by a microcanonical ensemble with total energy
E g = E s + E r = const.
(4.88)
However, E r and E s are not xed individually. More precisely, the total system is described by a
Hamiltonian
H = H s + H r + H int ,
(4.89)
where H int describes the interaction between S and R. We assume this coupling to be weak:
| hH int i | hH s i , hH r i .
(4.90)
Now we denote by |n i the quantum states of S and by | i the quantum states of R. They form
a complete basis, i. e. a quantum state (microstate) of the full system can always be expressed as
= n ,
(4.91)
65
where = (n, ). Now the energies are given by

E sn = n H s n ,

E r = H r ,
(4.92)
(4.93)
E gn = E sn + E r = E g .
(4.94)
And from the statistical postulate it follows that

pn = g1 (E g ) .
(4.95)
We now consider operators which only depend on n, i. e. they are independent of the properties
of the reservoir:
XX
hAi =
pn An
(4.96)
=
n
X
p n An ,
(4.97)
where
pn =
pn .
(4.98)
In other words, we look at the reduced system S. This restriction to leads to the concept of a
canonical ensemble. Our task is now to compute the reduced probability distribution pn . These
En
considerations will lead us to the famous Boltzmann factor e kbT .
4.2.2. Canonical partition function and Boltzmann factor

From (4.94), (4.95) and (4.98) we conclude that
X
pn =
g1 (E g ) (E sn + E r E g ) E g .
(4.99)
This means that the sum only runs over such which full the constraint (4.94). So, the sum
is just the number of states in R. Therefore pn will depend on E sn and the properties of the
reservoir (Tr ) and we conclude that
pn (E s ) =
r (E r )

.
g (E g ) Er =Eg Es
(4.100)
As we demanded E s E r , E r is close to E g and we can make a Taylor expansion of ln r (E r )

around E g :
ln r (E r ) ln r (E g ) + (E r E g )
ln r (E g ) .
(4.101)
E g
| {z } |
{z
}
E s
For
66
Er
Es
this gets exact (ln r f r ln E r ).
We have dened

ln r (E)
,
E
E=Er
(4.102)
where the evaluation at E r instead of E g is ne, as this dierence is a higher order eect in the
Taylor expansion. Note that 0 for monotonously increasing r (E).
So, just depends on the density of states of the reservoir:
=
1 S r

k b E E=Er
(4.103)
1
,
k bTr
(4.104)
and we infer that

=
with Tr being the temperature of the reservoir. r is now given by

r (E r ) = r (E g ) e (Er Eg ) ,
(4.105)
and inserting this into (4.100) leads to

1 E s
pn (E s ) = Z can
e
.
Here,
Z can =
g (E g )
r (E g )
(4.106)
(4.107)
is the canonical partition function. It only depends on properties of the reservoir. It can be
computed from the normalisation of pn :
X
pn = 1 .
(4.108)
n
So,
e En .
(4.109)
(E) e E
(4.110)
Z can =
Or, summing over energy states instead:

Z can =
X
E

=
dE (E) e E .
(4.111)
So all in all,
Z can = Z can (, N ,V ) ,
(4.112)
where N and V are the particle number and the volume of S. So, Z can only depends on R via .
67
4.2.3. Relation between temperature and energy

Let us next compute the mean energy of our statistical ensemble. First, we recall the probability
to be given by
1 E n
pn = Z can
e
,
(4.113)
where
Z can =
e En
(4.114)
and, using the convention k b = 1,

=
1
1
=
.
T Tr
(4.115)
Then we nd
E = hEi =
(4.116)
pn E n
P En
En
ne
= P
E n
e
n
X
= ln
e En .
(4.117)
(4.118)
So, the mean energy is given by
E = ln Z can .
(4.119)
4.2.4. Energy distribution

The probability density to nd an energy E is
w (E) = (E) e E
E
c s f s E
(4.120)
,
(4.121)
where f s , the number of degrees of freedom, is usually a huge number. Then, this distribution
describes an extremely sharp peak, as sketched in Figure 4.2.
Say, that w (E) is given by a Gaussian distribution:
E
( E ) 2
w (E) = const. e 2 E 2 .
(4.122)
Then we can determine E from the maximum of w or, respecitvely ln w:

2
(E E)
+ const.
2 E 2
= c s f s ln E E ,
ln w =
68
(4.123)
(4.124)
E
Figure 4.2.: Energy distribution
where we inserted (4.121) in the second line. We get

0=
cs fs

=
.
E E
E
(4.125)
This means that

cs fs
E =
= cs fs T ,
(4.126)
and therefore
T =
E
,
cs fs
(4.127)
so the temperature can be interpreted as something like the energy per degree of freedom.
The energy dispersion can be inferred from the second derivative:
cs fs
1
2
= 2 .
=
2
2
E
E E
E
(4.128)
So,
E = p
(4.129)
E
= p
.
E
cs fs
(4.130)
cs fs
1
In the thermodynamic limit the relative energy uctuations are negligible. So, we can assume
E = E and associate statistical quantities with actual numbers, this way. This means that we
can actually speak of the energy in a canonical ensemble. Then, microcanonical and canonical
ensemble are equivalent!
69
Microcanonical
Canonical
E mic
Z mic (E) = (E)
Z can ( )
Table 4.2.: Related quantities in microcanonical and canonical ensemble
4.2.5. Equivalence of microcanonical and canonical ensemble

Dierent quantities, characterizing the microcanonical or canonical ensemble, can be related,
as shown in Table 4.2. Note that the energy E is xed in this case.
They are related via
The question is, if we identify E mic
E = ln Z can ( ) ,
(4.131)
mic =
ln Z mic (E) .
(4.132)
E
= E can , can we also con conclude mic = can , i. e. Ts = Tr ?
4.2.6. Temperature
Microcanonical ensemble (T = Ts )
mic =
ln
1

=
E N ,V k bT
(4.133)
Here, T is an absolute temperature. So, its units are Kelvin or eV, if we set k b = 1. For example
1
300 K 40
eV. Recall that T is an intensive quantity and that it can be interpreted as an energy
per degree of freedom:
Es
T = 1 =
.
(4.134)
cs fs
In the case of an ideal gas we can compute T from the entropy
S=
via
3N
2E
ln
+ f (V , N )
2
3N
(4.135)
1
S
3N

=
=
.
T
E N ,V
2E
(4.136)
E
E 1
= = T,
3N
f
2
(4.137)
This means that
and therefore the average (kinetic) energy per particle is

E
3
= T,
N
2
as each particle has 3 degrees of freedom.
70
(4.138)
Canonical ensemble (T = Tr )
T is the characteristic quantity for the canonical ensemble, i. e. it is a property of the total
system. So to speak, we can forget the concept of the reservoir. T can be measured by
measuring the mean energy hEi in the system.
As a next step, we will show that Tr = Ts .
4.2.7. Temperature and energy distribution in subsystems

Consider a system that is divided into two arbitrary sized subsystems s1 and s2 with energies
E 1 and E 2 , respectively. Using the notation E = E 1 and E g = E 1 + E 2 , the probability to nd the
system in a state with a certain energy E is
w (E) = 1 (E) 2 (E g E) g1 (E g ) .
(4.139)
It is
As said before, w (E) is a sharply peaked distribution, dominated by E max = E.

ln w
E
E=E

=
ln 1 (E)
+
ln 2 (E g E)
E
E=E E
E=E
0=
= 1 2
1
1
=
,
T1 T2
(4.140)
(4.141)
(4.142)
(4.143)
where we used the denition of in the third equation. The minus sign in front of 2 arises
from the derivative of the argument E g E of 2 with respect to E. So, we nd
T1 = T2 .
(4.144)
We conclude, that we can dene the temperature of any ensemble with small uctuations of
E around E via

1
=
ln (E)
.
(4.145)
T
E
E=E
But note, that in general the temperature is determined by the energy dependency of the
partition sum.
4.2.8. Thermal equilibrium

0th law of thermodynamics
If two systems are in thermal equilibrium with a third system, they are also in
mutual thermal equilibrium. There exists a variable characterising the equilibrium
state, the temperature, that has the same value for the two systems.
Of course this is also valid for several systems in thermal equilibrium (T1 = T2 = . . . = Tn ).
71
Reservoir
System
R
S
E, N
Figure 4.3.: System connected to a reservoir with energy and particle exchange
4.3. Grand canonical ensemble

The three dierent types of statistical ensembles are, next to the volume V , characterized by
dierent quantities:
Microcanonical:
N ,E
Canonical:
N ,T
Grand canonical:
,T
The grand canonical ensemble describes an open system S in equilibrium that is able to exchange
energy and particles with a reservoir R, as illustrated in Figure 4.3. For example one can consider
a part of a gas volume.
We will treat it in the same manner as we treated the canonical ensemble. The probability for
a certain state is now given by
1 ( E n + N n )
pn (En , Nn ) = Z gc
e
,
(4.146)
where
ln r (E g , N g ) ,
E
=
ln r (E g , N g ) .
N
We dene the chemical potential as
= .
(4.147)
(4.148)
(4.149)
Then, the probability takes the form

1 (E n N n )
pn (En , Nn ) = Z gc
e
,
the grand canonical partition function is given by

XX
Z gc =
(E, N ,V ) e (E N )
E

=
72
dE
dN (E, N ,V ) e (E N ) ,
(4.150)
(4.151)
(4.152)
4.4. Relations for entropy
and the partition sum is

(E, N ,V ) = (E, N ,V ) E N .
(4.153)
The averaged quantities can again be extracted via derivatives of the logarithm of the partition
function:
E N = ln Z gc (, ,V ) ,
N =
ln Z gc (, ,V ) .

(4.154)
(4.155)

4.4.1. Entropy in canonical ensemble
We can approximate the probability density (E) in the partition function

Z can = dE (E) e E
by means of a saddle point expansion:
Z can
e E
= (E)
dE e 2 (EE ) + . . .
|
{z
}
q
(4.156)
(4.157)
2
a
where
cf
.
E2
(4.158)
E
2
e E
= (E)
E
cf
| {z }
(4.159)
a=
Then the partition function takes the form
Z can
O(1)
and its logarithm is, using E = E,

ln Z can = ln (E) E + . . .
| {z } |{z}
f
(4.160)
Then, altogether, we nd the entropy to be given by

!
S = ln Z can + E = 1
ln Z can .
(4.161)
Note that, as said before, the dierence between the xed energy E for the microcanonical
and E = hEi = E for the canonical ensemble is negligible. It vanishes in the thermodynamic
limit.
73
4.4.2. Entropy in grand canonical ensemble

By the same reasoning as before we get
ln Z gc = ln (E) (E N )
(4.162)
S = ln Z gc + (E N ) = 1
ln Z gc .
(4.163)
and therefore the entropy is
4.4.3. General expression for entropy

In general, the entropy can be expressed in the following form:
X
S=
p ln p ,
(4.164)
where we had n in the previous cases. The validity of this expression can be checked easily:
In the microcanonical case it is
1
p = ,
(4.165)
so we have
X 1
1
S=
ln
(4.166)
!
1 X
= ln
1
(4.167)

|{z}
= ln
q. e. d.
(4.168)
And in the canonical case it is

1
e E ,
Z can
ln p = ln Z can E ,
p =
(4.169)
(4.170)
so we have
S=
(4.171)
p ln p

p ln Z can + E
(4.172)
= ln Z can + E
q. e. d.
Similarly, it can be checked in the grand canonical case.
74
(4.173)
4.4.4. Entropy and density matrix

The entropy is related to the density matrix via
S = tr ln .
(4.174)
This can also be considered as a general denition of the entropy, as it coincides with S = ln (E)
for microcanonical, canonical and grand canonical ensembles.
Most importantly, this expression is valid in any basis. Therefore it also holds when diagonalizing , which proves (4.164) more generally:
X
S=
ln
(4.175)
=
p ln p
q. e. d.
(4.176)
75
5. Thermodynamics
5.1. Thermodynamic potentials
5.1.1. State variables, equations of state
Classical thermodynamics describes equilibrium states and transitions from one equilibrium
state to another. Historically it was developed before statistical physics, but it can be derived as
an eective macroscopic theory from the microscopic statistical theory. Without any doubt it
has an immense importance regarding technical aspects. It is mostly axiomatic, based on the
laws of thermodynamics and their consequences. Classical thermodynamics only uses a few
macroscopic variables (both extensive and intensive) such as E, T , S, N , , V , p. We distinguish
state variables or independent thermodynamic variables, which characterise the ensemble, and
dependent thermodynamic variables. For the dierent types of ensembles (systems) this is
illustrated in Table 5.1
The thermodynamic limit provides a functional relationship between the state variables.
Usually, no average values are used, because hEi E, etc.
There are several practical question, which we will try to answer in this chapter:
How much energy should be deposited in a system, so that its temperature rises by a
certain amount, provided the volume stays constant?
How much will the pressure increase during this process?
How is heat transformed into mechanical energy? What is the eciency of machines?
When is a process reversible and when not?
The dependence of the dependent variables on the state variables is called equation of state.
Examples
1. Equation of state for an ideal gas:
pV = RT n M = k bT N
(5.1)
n M : number of moles
R: gas constant (R = 8.31 molJ K )
This follows from
p
S

= .
V E,N T
(5.2)
77
5. Thermodynamics
microcanonical ensemble (isolated system)

canonical ensemble (closed system)
grand canonical ensemble (open system)
State variables
Dependent variables
ln Z mic (E, N ,V )
ln Z can (T , N ,V )
ln Z gc (T , ,V )
S,T , ,p
S,E, ,p
S,E, N ,p
Table 5.1.: Dierent kinds of ensembles (systems) and their corresponding state and dependent variables
It can be concluded that

p = Tn .
(5.3)
2. Van der Waals equation of state for a real gas:

(p + an 2 ) =
nT
1 bn
(5.4)
a,b: constants specic to the gas

The task of the statistical physics is to explain the equations of state and to derive their
functional form. On the other hand, classical thermodynamics examines the consequences of
their existence. There are seven variables and only 4 equations, which automatically makes
three of the variables independent (state variables), but there also are additional thermodynamic
conditions. As already mentioned, variables split into two types: extensive (E, N ,V ,S) and
intensive (T , ,p). However, the choice of the state variables cannot be arbitrary at least one
of them has to be extensive.
Example
Grand canonical ensemble with state variables T , ,V :
S = V f S (T , )
(5.5)
E = V f E (T , )
(5.6)
N = V f N (T , )
(5.7)
p = fp (T , )
(5.8)
Therefore, (p,T , ) cannot be used as three independent variables at the same time!
If the particle number is not conserved (e. g. photon), there are only two state variables, as
= 0. Then,
p = fp (T ) .
(5.9)
So, there is a relation between pressure and temperature. Dening =
E
V,
we nd
p = fp ( ) ,
(5.10)
p = 13 .
(5.11)
and in the special case of photons it is
78
5.1.2. Thermodynamic potentials and partition functions

S = S (E, N ,V ) =
k b ln Z mic (E, N ,V )
(entropy)
(5.12)
F = F (T , N ,V ) =k bT ln Z can (T , N ,V )
(free energy)
(5.13)
J = J (T , ,V ) =k bT ln Z gc (T , ,V )
(Gibbs potential)
(5.14)
S
E
S
N
S
V
N ,V
E,V
E,N
1
T
(5.15)
(5.16)
p
T
(5.17)
At this point, these are just denitions. Their physical meaning will be explained later. Note that,
traditionally, in thermodynamics one denotes the xed quantities explicitly. These 3 relations
together with S (E, N ,V ) lead to 4 relations for the 7 variables in total.
We now use
S = ln Z can + E
and =
1
T
(5.18)
to nd
F
E
= ln Z can = S ,
T
T
(5.19)
and therefore
F = E TS .
(5.20)
S = ln Z gc + (E N ) ,
(5.21)
E N
J
= ln Z gc = S
,
T
T
(5.22)
On the other hand, using
we have
and therefore
J = E N T S
(5.23)
J = F N .
(5.24)
or
79
5. Thermodynamics
5.1.3. Dierential relations

Recall the following denitions:
1
S

=
,
T
E N ,V
S
,
=
T
N E,V
p
S

=
.
T
V
(5.25)
(5.26)
(5.27)
E,N
With these, one arrives at the following expressions for the dierentials of S:
dS =
p
1
dE dN + dV ,
T
T
T
(5.28)
and of E:
dE = T dS +
|{z}
heat
Q
dN
|{z}
added particles
EN
pdV .
|{z}
(5.29)
work
A
1st law of thermodynamics: conservation of energy (Helmholtz, Mayer)

A thermodynamic system can be characterized by its energy. The energy can be
changed by adding heat, work or matter. For an isolated system the energy is
conserved.
It is very important to note that heat, work and added particles are not properties of an
equilibrium system! Statements like Heat is added / lost make no sense! But these quantities
may describe dierences between two equilibrium systems.
Heat
For example, the heat is dened as
Q = E 2 E 1
for xed V , N .
(5.30)
Therefore it accounts for a change in temperature:

Q = E(T2 ,V , N ) E(T1 ,V , N ) .
(5.31)
Again, note that this no property of a system. Neither is it a thermodynamic variable. But it
can be used as a thermodynamic denition of the entropy:
Q = T dS .
80
(5.32)
ds
Fp
Figure 5.1.: Mechanical work
Mechanical work
A denition of mechanical work can be given via a slow change of volume at xed S, N (compare
Figure 5.1):
A = Fp ds
(5.33)
= |Fp | ds
(5.34)
= p a ds
(5.35)
= p dV ,
(5.36)
where p is the pressure and a the area.

We can identify dierent signs of A with dierent physical situations:
A < 0 :
system does work,
(5.37)
A > 0 :
systems absorbs mechanical energy.
(5.38)
In general, A describes the change in energy by work at xed S, N . For example, it also
describes the change of an external magnetic eld instead of the change of volume.
We note that the statistical denition of pressure coincides with the mechanical denition
force per area, but the statistical one is more general!
5.1.4. Legendre transform

The change from S to F at xed V , N is one example of a Legendre transform. In this case,
the conjugate variables are E and = T1 . To be formally more precise, its actually a transform
between the following quantities:
S (E)
F ( ) = F ( ) .
(5.39)
Let us look at this in more detail. The new variable is :
S
=.
E
(5.40)
Then, the Legendre transform of S is given as

F = E S ,
(5.41)
81
5. Thermodynamics
and therefore
F = E TS .
(5.42)
The derivative of F with respect to is
F
E S
=E+

E
S E
=E+
|{z}
E
(5.43)
(5.44)
=E,
the original variable, as it should be.
On the other hand, we nd
(5.45)
F

= S ,
T N ,V
(5.46)
dF = SdT + dN pdV .
(5.47)
dJ = SdT N d pdV ,
(5.48)
using (5.20). And therefore

Similarly, it is
which can be concluded, using
ln Z gc = N ,
J = ln Z gc ,
so,
J
= N ,
(5.49)
(5.50)
(5.51)
as well as
!
S = 1
ln Z gc
= 1
( J )
J
= J + J + 2
J
=
.
T
Now we have everything at hand to construct thermodynamics:
82
(5.52)
(5.53)
(5.54)
(5.55)
1. As the basis: S = ln , the central probabilistic quantity

P
(equivalent to S = p ln p = tr ( ln ))
2. Denitions of T , ,p as derivatives of S:
dS =
p
1
dE dN + dV
T
T
T
(5.56)
3. Computation of the partition function for the given system S (E, N ,V ), or equivalently
F (T , N ,V ) or J (T , ,V )
(aim of statistical physics, learn counting)
Note that no other than these 3 axioms are needed for thermodynamics! And all fundamental
laws of thermodynamics can be derived from these! We already have shown the latter for the
0th and the 1st law.
5.1.5. Properties of entropy

1) Additivity
Consider two systems with energies E 1 and E 2 (neglect again H int ), then
g = 1 2 ,
ln g = ln 1 + ln 2 ,
Sg = S1 + S2 .
(5.57)
(5.58)
(5.59)
2) Maximum principle for entropy

We permit energy exchange between these two systems, with the constraint
E1 + E2 = Eg .
(5.60)
We dene E 1 = E and want to nd the condition for the maximum of the probability distribution
W (E), which describes the equilibrium state:
w (E) =
1 (E 1 ) 2 (E 2 )

.
g (E g )
E1 +E2 =Eg
(5.61)
From the considerations in 1) we can conclude that a maximum of w (E) corresponds to a

maximum of ln 1 + ln 2 , which corresponds to a maximum of S 1 + S 2 and therefore to a
maximum of S g . This means that in equilibrium the system chooses E 1 such that the total
entropy is maximal, with the constraint E 1 + E 2 = E g .
In equilibrium entropy is maximal, subject to constraints.
Of course, the individual energies E 1 , E 2 can be computed from the maximum of S 1 + S 2 with
the condition E 1 + E 2 = E g .
83
5. Thermodynamics
Another property of the entropy can be inferred from this. Let us start with two isolated
systems in equilibrium, with arbitrary energies E 1(0) , E 2(0) and E g = E 1(0) + E 2(0) . They have
entropies S 1(0) , S 2(0) . Then we bring them into thermal contact (i. e. we allow energy exchange).
This leads to an energy ow E 1(0) E 1 , E 2(0) E 2 until S 1 + S 2 is maximal for a given E g . This
means that
S 1 + S 2 S 1(0) + S 2(0) .
(5.62)
Entropy can only increase!
This is a key ingredient of the 2nd law of thermodynamics!
5.1.6. Zero of absolute temperature

Approaching its ground state, the partition sum of a system will approach a value of the order
of 1:
EE 0
(E 0 ) = 0 = O(1) ,
(5.63)
and therefore the entropy will approach 0:
EE 0
(or S S 0 ,
Using
S0
V
S 0
(5.64)
ln
1
=
,
E E0 0 E
(5.65)
0).
and therefore, by the denition of the absolute temperature,

! 1
ln
E
T =
0
,
E
we nd
(5.66)
EE0
,
E
(5.67)
E E f ,
(5.68)
(5.69)
as, choosing E 0 0,
e f .
This means that in the thermodynamic limit the ground state temperature approaches 0:
f
T f ef 0 ,
(5.70)
T (E = E 0 ) = 0 .
(5.71)
or, in other words,

The properties (5.63) and (5.67) of (E) are illustrated in Figure 5.2.
84
5.2. Reversible and irreversible processes
ln
E
E0
Figure 5.2.: Behaviour of the partition sum at zero temperature
S0
0
0+
Figure 5.3.: Entropy for T 0+
5.1.7. 3rd law of thermodynamics

3rd law of thermodynamics (empirical by Planck (Nernst))
ForT 0+ the entropy obeys S S 0 , where S 0 is independent of the thermodynamic
variables of the system.
So, the entropy for T 0+ will behave as sketched in Figure 5.3. An example for this would be
the hyper-ne splitting in atoms.

5.2.1. Subsystems and constraints
We consider an isolated system in equilibrium with xed energy E (or interval [E,E + E],
respectively). Furthermore we recall that = Z mic is the number of (micro-)states with energy
E, which the system can occupy, and that the entropy is dened as S = k b ln .
In general, the system is determined by some constraints. These correspond to particular
parameters:
= (E,y1 , . . . ,yk )
(5.72)
(or intervals [y1 ,y1 + y1 ], . . . , [yk ,yk + yk ]).
85
5. Thermodynamics
Gas 1
(E 1 ,V1 , N 1 )
Vacuum
Gas
(V1 )
Gas 2
Thermally isolating wall
Wall
(a) Example 1 (xed volume V1 )
(b) Example 2 (xed energy E 1 )
Figure 5.4.: Examples of constraints
Examples
1. Fixed volume V1 (Figure 5.4a):
V1 : volume occupied by the gas
2. Fixed energy of a subsystem E 1 (Figure 5.4b):
E 1 : energy of gas 1
V1 : volume of gas 1
N 1 : particle number of gas 1
This example can be considered as two isolated subsystems:
g (E,E 1 ,V ,V1 , N , N 1 ) = (E 1 ,V1 , N 1 ) (E E 1 ,V V1 , N N 1 ) .
(5.73)
5.2.2. Removing the constraint

After sucient time the system will reach a new equilibrium state:
i f .
(5.74)
f i .
(5.75)
Thereby, one always has

All the states compatible with the constraint are still possible. But additionally, new states with
xed energy E become possible, which previously have been excluded by the constraint.
In general, removing the constraint in an isolated system leads to a nal state whose entropy
is larger than or equal to the entropy of the initial state:
Sf Si .
(5.76)
(The nal and initial states are equilibrium states. It is not important what happens in between.)
86
Examples
1. Removing the wall:
i V1N
(5.77)
f V N
(5.78)
f i
for V > V1
(5.79)
2. Wall becomes thermally conductive (subsystem is a canonical ensemble):

i = i1 i2
(5.80)
m 3N1/2
1
E 1
3N
3N
3N
i1
V1 1/2 E 1 1/2
(diluted gas at high energy)
2
E1
N 1 ! 2 1 ! 2~
f i
(i w (E 1 ), f w (E1 ))
E i Ei
f i for E 1 , E 1max = E1 (recall Figure 4.2)
(5.83)
f i
(5.84)
for T1 = T2
(5.81)
(5.82)
5.2.3. Increase of entropy by removing the constrains

The probability that the system without constraints, after suciently long time, still stays in its
initial state is
i
wi =
.
(5.85)
f
That is because 1f is the probability that the system is in one particular macrostate (of the
system without constraints) and i is the number of microstates satisfying the initial conditions.
Of course,
w i 1 if i f .
(5.86)
Examples
1. Probability that the gas still occupies a volume V1 after removing the wall for V = 2V1 :
1 N
wi =
(5.87)
2
(factor
1
2
for each molecule)
2. Probability that the energy of gas 1 remains E 1 after the wall becomes thermally conductive:
w i = w (E 1 ) E
(5.88)
For the interval [E max 12 E,E max + 12 E] with
E
E
1
N
E 1 .
w (E max ) E = w (E)
it is
(5.89)
For E 1 , E we nd w (E 1 ) to be exponentially suppressed.
87
5. Thermodynamics
Let us look at the eect of removing the constraint on a general variable y (e. g. y = E 1 , the
energy in the subvolume V1 ). Beforehand, it has a xed initial value yi . Afterwards, it satises a
probability distribution characterized by
w (y)y (y) ,
where, in general, y , yi , with y being the position of the maximum of w (y),
generalization to multiple variables y1 , . . . ,yk is straightforward.)
Now we consider the limit
N .
(5.90)
w
y y
= 0. (The
(5.91)
This will lead to a sharp value yf = y = ymax (again, generally yf , yi ). yf is determined by the
entropy becoming maximal ((ymax )):
S (yf ) maximal.
(5.92)
Without constraints the parameters will take such values that the entropy becomes maximal (at
given boundary conditions like V , N , E of the system).
Example
y is the energy in a subvolume:
Beforehand, S (E, N ,V ; y) is calculated with given constraints. Afterwards, we have to
satisfy yS
= 0 to nd ymax .
E,N ,V
The entropy of a system grows when approaching the equilibrium state.

Note that this is a direct consequence of our two postulates! (It is not necessarily valid if one of
the basic postulates is not obeyed!)
Initial states with the probability distribution not being a uniform distribution over all possible
microstates (equilibrium distribution) are very unlikely for large N :
i
1,
f
Si < Sf .
wi =
(5.93)
(5.94)
So, from the equilibrium state point of view approaching the equilibrium state corresponds to
the transition from an unlikely to a likely state (content of the rst postulate).
5.2.4. 2nd law of thermodynamics

The entropy of an isolated system cannot decrease.
Historical formulations:
Clausius:
Heat cannot ow from lower to higher temperature.
88
Kelvin:
It is impossible to produce work continuously by lowering the temperature of a
system, without other changes in the environment.
In other words, the latter formulation states the impossibility of perpetuum motion machines.
Short summary of the laws of thermodynamics
0th law: T , T1 = T2
1st law: E, conservation of E
2nd law: S, S cannot decrease
3rd law: S min = 0
5.2.5. Reversible processes for isolated systems

A process (i) (f ) in an isolated system is reversible, if the process (f ) (i) is
possible, too. Otherwise, the process is irreversible.
In an isolated system an increase of entropy indicates an irreversible process. On the other
hand, a reversible process is, concerning the isolated total system, adiabatic (the entropy of the
total system stays constant).
5.2.6. Reversible processes for open systems

It must be possible, after performing some process, to perform its reversed process, so that the
system gets back into its initial state, without changing the environment. The Carnot cycle is an
example of such a reversible process (if it is quasi-static). The temperature dierence between
the reservoirs is lowered and work is performed. Its reversed process describes a fridge: work
is performed in order to enlarge the temperature dierence between the reservoirs. The p V
diagram of the Carnot cycle is shown in Figure 5.5.
Irreversible processes, on the other hand, are dissipative processes. Typically, in such a process
the energy is being distributed over several degrees of freedom (conversion of mechanical
energy into heat). Examples are mechanical friction, eddy currents, viscose friction and
balancing the temperature of two heat reservoirs.
89
5. Thermodynamics
(a)
(d)
(b)
(c)
V
Figure 5.5.: Carnot cycle
5.2.7. Thermodynamics and non-equilibrium states

The course of an irreversible process cannot be described by thermodynamic potentials. But its
nal and initial states can be described, if they are equilibrium states.
90
Part II.
Statistical Systems
91

In an ideal gas the interactions between its particles can be neglected. This can be used as an
approximation for several kinds of physical systems like:
molecules
photons (radiation)
electrons
quasi-particles in solids (phonons, magnetons )
It provides a good approximation especially at small densities, but also for many other problems,
e. g. electrons in solids.
6.1. Occupation numbers and functional integral

We consider some one-particle states . In the case of momentum states, we have = p =
(p1 ,p2 ,p3 ). To each of these states we can assign an energy E = E(p) and an occupation number
n = n(p). A given sequence of occupation numbers {n } = {n(p)} corresponds to a given
function n(p). Depending on the type of particles we consider, the occupation numbers can
attain dierent value ranges:
fermions:
n = 0, 1
(6.1)
bosons:
n = 0, 1, 2, . . . ,
(6.2)
A microstate of the full system is given by the sequence of occupation numbers:

{n } .
(6.3)
This can be used to express the sum over all microstates via the one-particle states:
X X

(6.4)
{n }
X
X
n1
n2
YX
(6.5)
nk
(6.6)
93
This way, every possible sequence appears exactly once! For the dierent types of particles this
yields:
1 !
X Y X
fermions:
=
(6.7)
{n }
bosons:
{n }
n =0
!
X
n =0
(6.8)
As said before, we can relate a sequence of occupation numbers {n } to a function n(p). This
means that we also can identify the sum over all possible sequences with a sum over all possible
functions n(p):
X

D n(p) .
(6.9)
{n }
This is a functional integral!
6.2. Partition function for the grand canonical ensemble

Using the results from the previous section, we can express the grand canonical partition
function as
X
Z gc =
e (E N )
(6.10)
=
e (
E n
n )
(6.11)
{n }
n (E )
(6.12)
e n (E )
(6.13)
{n }
XY
{n }
Y X! Y
0
!
(6.14)
n0
Y X
n (E )
n (E )
!
.
(6.15)
The last step may seem confusing, but by explicitly writing it out, it is easy to see that every
summand in (6.14) appears exactly once and with the same weight in (6.15), too.
The logarithm of the partition function is given by

X X
ln Z gc =
ln
en (E ) .
(6.16)
For fermions this yields

(f)
ln Z gc
=
94

ln 1 + e (E ) ,
(6.17)
6.3. Mean occupation numbers
whereas for bosons we nd (j = n )

(b)
ln Z gc
ln
"X

(E )
j#
(6.18)
j=0
Applying the formula for the geometric series:

1 + x + x2 + x3 + =
1
,
1x
(6.19)
(6.20)
we arrive at
(b)
ln Z gc
=
ln
e (E )
Now we only have to carry out sums over one-particle states! This is an enormous simplicaP
P
P
tion as it is much easier to perform rather than = {n } ! But note that this just works
for free particles without any interactions.

6.3.1. Bosons
In general it is
N =
(6.21)
N p
X N e (E N )
Z gc
1
ln Z gc .

(6.22)
(6.23)
So, using (6.20), for bosons we nd

N =

X" 1
#
(E )
ln 1 e

X 1 e (E )
=
1 e (E )
X e (E )
=
,
(E )
1
e
|
{z
}
(6.24)
(6.25)
(6.26)
(b)
and therefore
N =
n(b) ,
(6.27)
95
with n(b) being the mean occupation number of the state :

n(b) =
1
e (E )
(Bose-Einstein statistics)
If we also allow for state dependent , this generalizes to

X X
P
Z (; 1 , . . . , ) =
e (E n n )
n1
and
n =
(6.28)
(6.29)

1
ln Z
.

=
(6.30)
For bosons this yields

ln Z (; ) =
ln
1
1
e (E )
(6.31)
and n(b) as before.
6.3.2. Fermions
As for bosons, the mean occupation number of fermions is given by the sum of the mean
one-particle occupation numbers:
X
N =
n(f) ,
(6.32)
where in this case

n(f)

1
(E )
=
ln 1 + e

e (E )
=
.
1 + e (E )
(6.33)
(6.34)
And therefore
n(f) =
1
e (E )
+1
(Femi-Dirac-statistics)
(6.35)
From this it is easy to see that n(f) only can take values in the following range:
0 n(f) 1 .
(6.36)
Moreover, we can look at some special cases:
96
n(f) 0
for E T ,
n(f) 1
for E < 0
(6.37)
and
|E | T .
(6.38)
6.3.3. Mean energy

In general, the mean energy is computed as
!

E=
+
ln Z gc

= ln Z gc + N .
(6.39)
(6.40)
For bosons it is
X (E )e (E )
+ N
1 e (E )
X
X
=
(E )n(b) +
n(b)
E =
(6.41)
(6.42)
E n(b) .
(6.43)
Performing the same calculation for fermions, we nd that both are given by
X
E =
E n .
(6.44)
Similar formulas can be found for other macroscopic quantities, too! Note that n itself already
is a macroscopic quantity!
6.3.4. Chemical potential

One question we have not discussed yet is: Which value should we choose for ?
1) Conserved number of particles
Examples for conserved numbers of particles are:
N e : electric charge and lepton number
Ni : number of molecules of a certain type, if chemical transitions are negligible
N b : baryon number (in the early universe)
In general,
1
N =
ln Z gc (, = N (, ) ,

(6.45)
= ( N ,T ) .
(6.46)
so we conclude
In the thermodynamic limit the grand canonical and the canonical ensemble are equivalent,
which leads to
= (N ,T ) .
(6.47)
Note that, if there are several conserved quantum numbers Ni , there also are several independent
i .
97
2) No conserved number of particles

Examples for cases where the number of particles is not conserved are photons and protons. In
these cases the canonical ensemble does not depend on N , i. e.
0.
(6.48)
P
But there is no restriction of the summands in {n } , anyway, as N is no conserved quantum
number. Here, the mean occupation numbers characterize the canonical ensemble.
3) Approximatively conserved quantum numbers

If one considers time scales, which are much shorter than the decay time of the particles, then
one can consider the corresponding particle number as approximatively conserved and assume
, 0.
(6.49)
6.3.5. Boltzmann statistics

In the regime of large energies (compared to T ):
E T
(6.50)
the mean occupation number of a state is approximately

n e (E ) ,
(Maxwell-Boltzmann statistics for classical particles)
(6.51)
which holds for bosons and fermions, equally. This recovery of the classical description in
the large energy limit is an example of the correspondence principle between QM and classical
mechanics.
We also can write this as
n = C e
E
T
ET
= C e
E
(6.52)
(6.53)
to emphasize the famous Boltzmann factor e T .

Note that E is the energy of the one-particle state it is not to be mistaken for the energy
1 e E .
E of a microstate like, for example, in p = Z can
As a comparison recall the computation of (E, N ) earlier in this lecture, with the approximation that is dominated by microstates where each is occupied at most once, which
corresponds to
n 1 .
(6.54)
98
6.4. Photon gas
Hole
Figure 6.1.: Hole in a box
6.4. Photon gas

Think of a box with a small hole in it, like in Figure 6.1. The electromagnetic radiation within
the box will be in equilibrium, with the temperature T given by the temperature of the walls.
The hole is needed for observation. If there is no radiation from the outside entering through
the hole, this system is an example of a black body.
The concept of a black body was crucial in the development of modern physics, as classical
eld theory yields wrong results (large deviations from the measured frequency distribution of
the radiation). Historically, two of the most important milestones in understanding this as a
photon gas in thermal equilibrium have been the quantisation of electromagnetic energy by
Planck in 1900 (~) and Einstein's idea of quantising the the electromagnetic radiation itself as
photons in 1905 (explanation of the photoelectric eect).
From a modern statistical physics point of view, a photon gas is a gas of non-interacting
particles whose particle number is not conserved ( = 0). So it is described by a canonical
ensemble. Note that in the case of free photons n(p) = n is conserved for every p (). Such
an ensemble with , 0 can be very dierent from black body radiation ( = 0), e. g. radio
waves.
6.4.1. Black body radiation

Consider a gas of photons in thermal equilibrium with a temperature T . As = 0, we immediately can conclude
1
n = E
.
(6.55)
e 1
So, to proceed, one-photon states and their energies E are necessary.
To get them, we start with the Maxwell equation (wave equation for the electric eld):
1 2 E
= 2 E .
c 2 t 2
(6.56)
E = E (r ) eit ,
(6.57)
Assuming
it takes the form
2 E (r ) +
2
E = 0,
c2
(6.58)
99
which is solved by
E(r ) = A eik r .
(6.59)
The exact form of boundary conditions is not important, so we choose periodic ones:
k=
2
n
L
with n x,y,z Z .
(6.60)
From this we can conclude the dispersion law

k2 =
2
,
c2
= c |k | .
(6.61)
Now, using the fact that the diversion of the electric eld vanishes (no sources):
E (r ) = 0 ,
(6.62)
we can directly infer the transversality of electromagnetic waves:

k A = 0.
(6.63)
This means that there are 2 degrees of freedom for every k (massless particle with spin 1 and
helicity H = 1).
Quantisation
For an electromagnetic wave, energy and momentum are given by
E = ~ ,
(6.64)
p = ~k .
(6.65)
Therefore, the one-photon state energy is, using (6.60),

E = ~c |k | =
2~c
,
L
(6.66)
where the states are given by

= (n x ,ny ,nz ,H )
with n x,y,z Z , H = 1 .
(6.67)
We also can split up the states as

= ( t , i ) ,
(6.68)
t (n x ,ny ,nz )
translational degrees of freedom,
(6.69)
i H
inner degrees of freedom.
(6.70)
into
Note that the translational degrees of freedom are the same for all kinds of particles, e. g. atoms,
molecules.
100
6.4. Photon gas
Density of the translational states

For large volumes all translational states lie very densely, as E L1 . Therefore, we can replace
the sums over the states by integrals:
X
2V
d3p
,
(2~) 3
(6.71)
d3 k
,
(2~) 3
(6.72)
where we dened the volume V = L3 . To arrive at (6.72), we used p = ~k and the fact that E
is independent of H . So, the factor of 2 arises due to the two possible values of the helicity
H = 1.
6.4.2. Spectrum of black body radiation

The mean number of photons per volume (photon density n)
in the momentum interval [k,k+dk]
is, using (6.55), (6.72) and E = ~,
n(k
) d3 k =
2
e ~
d3 k
.
1 (2) 3
(6.73)
With the dispersion law = c |k | we can deduce

d3k = 4|k | 2 d|k |
4
= 3 2 d ,
c
(6.74)
(6.75)
so the mean number of photons per volume in the frequency interval [, + d] is

n()
d =
8
2
d .
(2c) 3 e ~ 1
(6.76)
Note that there is no divergence for 0!

Using E = ~, we can easily infer the mean photon energy per volume in the frequency
interval [, + d] as
~
3
u (,T ) d = 2 3 ~
d .
(6.77)
c e
1
This is the frequency spectrum of black body radiation. It is plotted in Figure 6.2. Dening
= ~ = k~
, we recover the original form
bT
u (,T ) d =
(k bT ) 4 3
d ,
2 (~c) 3 e 1
(6.78)
postulated by Planck in 1900. Its maximum is roughly at ~ 3k bT or 3, respectively.
101
10
~
k bT
Figure 6.2.: Frequency spectrum of black body radiation
Of course, we also can deduce the wavelength spectrum, using =

1
u (,T ) d = 162 ~c
e
2~c
kbT
d
.
5
1
2c
:
(6.79)
Let us look at an example: the spectrum of sunlight. Its maximum is at max 500 nm
which is just what one would expect from a black body radiation spectrum at the sun's surface
temperature T 5800 K. This maximum lies in the regime of green visible light which, as an
interesting side note, is also the maximum of the sensitivity of the human eye!
Furthermore, we can consider the low frequency limit ~ k bT , where the frequency
spectrum can be approximated by
~ 3
d
2c 3 ~
k bT
= 2 3 2 d ,
c
u (,T ) d =
(6.80)
(6.81)
which recovers classical electrodynamics (Einstein 1905). In this limit ~ does not appear, any
more.
The extrapolation to large frequencies ~ k bT , on the other hand, leads to a divergence:
d u (,T ) .
(6.82)
This demonstrates the incompleteness of the theory! So, classical electrodynamics has to
be modied. A solution is given by quantum statistics of photons, which accounts for the
particle-wave duality (for large occupation number we are in the regime of classical elds,
whereas for small occupation numbers we are in the regime of classical particles).
102
6.4. Photon gas
6.4.3. Energy density

The energy density is given by
E
= =
V
d u (,T )
(6.83)
d u (,T )

3
(k bT ) 4
= 2 3 3
d
.
c ~ 0
e 1
|
{z
}
(6.84)
0
(6.85)
4
15
So, we conclude that

=
2 (k bT ) 4
.
15 (c~) 3
(Stefan-Boltzmann law)
(6.86)
This law contains several fundamental constants:

c: unication of space and time: special relativity
~: quantum physics
k b : associates T to energy, microphysical understanding of thermodynamics
Natural units
A frequently used system of units are the natural units. They are dened via
~ = c = kb = 1 .
(6.87)
This has some consequences:

k b = 1: Temperature is measured in units of energy, e. g.
1 K = 8.617 105 eV .
(6.88)
c =: Mass is measured in units of energy, e. g.

m e = 511 keV ,
(6.89)
1 eV = 1.783 10
33
g,
(6.90)
and distances are measured in units of time, e. g, light seconds.

~ = 1: Time and distances are measured in inverse units of energy, e. g.
1
1 fm = 1015 m =
,
197 MeV
1 m = 5.068 106 (eV)1 .
(6.91)
(6.92)
In Table 6.1 some physical quantities and constants and their respective mass (energy) dimensions in natural units are listed.
103
Quantity
Mass dimension
E
T
L,t
V
N
M
M
M 1
M 3
M0
M
M4
M0
M
M0
M0
M0
M0
Z mic,can,gc
F
S
e
Q
Table 6.1.: Mass dimension of some physical quantities
6.4.4. Canonical partition function

The logarithm of the canonical partition function for a photon gas is computed as
ln Z can (,V ) =
Together with (6.72) and E =
1
.
1 e E
(6.93)
1
d 2 ln
1 e
{z
}
(6.94)
ln
this yields
ln Z can
V k bT
= 2
~c
!3
|0
2 3
V ,
45
4
45
(6.95)
where we used natural units in the last step again. Let us check, if this gives the correct result
for the energy density:
E=
=
ln Z can
2 4
V
15
(6.96)
(6.97)
and therefore
E
2
= T4
V
15
which agrees with (6.86), using natural units.
=
104
(6.98)
6.5. Degenerate Fermi gas
Remarks
1. Not only E itself is a macroscopic observable, but the energy distribution as a function of
the frequency is one, too.
2. From Z can other quantities, e. g. the pressure of the photon gas, easily can be attained.
thermodynamic potentials classical thermodynamics
6.4.5. Free energy and equation of state

Using (6.95), the free energy is found to take the form
F = k bT ln Z can
= VT 4 ,
3
where we dened
=
or, using natural units,
k b4 2
15c 3 ~3
2
,
15
respectively. Then, the entropy, energy and pressure are given by
!
F
4
S=
= VT 3 ,
T V 3
=
E = F + T S = VT 4 ,
!
F
p=
= T4 .
T V 3
(6.99)
(6.100)
(6.101)
(6.102)
(6.103)
(6.104)
(6.105)
Combining the expressions for E and p and using = VE , we arrive at the equation of state for a
photon gas:
1
p = ,
(6.106)
3
where, as a reminder, the energy density is given by
=
2 4
T .
15
(6.107)

6.5.1. Quasiparticles
At low temperature and high pressure quantum statistics become important! A typical example
of this are electrons in metals. They can be considered as quasi free and are, due to their spin of
s = 12 , fermions. Recall that the mean fermionic occupation number is given by
n(f) =
1
.
e (E ) + 1
(6.108)
105
n
1
Figure 6.3.: Mean occupation number in zero temperature limit
6.5.2. The limit T 0

We consider the low temperature regime:
k bT |E | .
(6.109)
Thereby, the zero temperature limit of the mean occupation number depends on the relative
size of E and :
k bT 0
n 0
k bT 0
n 1
for E > ,
(6.110)
for E < .
(6.111)
This describes a step function as plotted in Figure 6.3.

We dene the Fermi energy f via
f = lim (N ,T ) ,
T 0
(6.112)
and the Fermi momentum pf or k f , respectively, via the usual energy momentum relation:
f =
pf2
~2k f2
=
,
2M
2M
(6.113)
with M being the eective mass of the quasiparticle.
6.5.3. Computation of the Fermi energy

We consider electrons or atoms with 2 internal degrees of freedom (i = 2). Then, the number
of particles is computed as

V
N =2
d3 k
(6.114)
(2) 3
k 2 <k f2
kf
V
d|k | |k | 2
2 0
V
= 2 k f3 .
3
106
(6.115)
(6.116)
So, using n =
N
V
, the Fermi momentum and energy are given by

k f = (32n) /3 ,
1
f =
(6.117)
1
2
(32n) /3 .
2M
(6.118)
This way, we also know (n) at T = 0.

For free non-relativistic particles one also can dene the Fermi temperature:
k bTf = f .
(6.119)
For example the Fermi temperature of electrons in copper is approximately 80 000 K. So, we see
that in this case (and also in general) k bT f is a very good approximation.
6.5.4. Thermodynamics
for our purposes the energy E(T ,V , N ) is the most suitable thermodynamic potential. To
calculate it, we perform an expansion in TTf . In the lowest order approximation the energy is
given by the ground state energy:
E = E0 .
(6.120)
6.5.5. Ground state energy

The ground state energy is computed as
E 0 = lim E
T 0
=2
V
(2) 3
(6.121)

d3 k
k2
2M
(6.122)
k 2 <k f2
kf
V 1
= 2
d|k | |k | 4
2M 0
V
=
k5
102 M f
3 k f2 N
=
,
10 M
(6.123)
(6.124)
(6.125)
where we used (6.116) in the last step. Also using (6.113), leads us to
3
E0 = N f .
5
(6.126)
So, in the low temperature regime we have

E
3
= +...
N
5
(6.127)
107
One might wonder if
E
N
is constant? The answer is no! It depends on the density due to

~2
2
(32n) /3 .
2M
= f =
(6.128)
An additional particle always has the energy f ! (not the mean energy of the already present
particles)
So far, we have found the zero temperature equation of state for our Fermi gas:
E(T = 0, N ,V ) =
3
2
(32n) /3 N .
10M
(6.129)
6.5.6. One-particle density of states

We would like to nd a more general type of dispersion relation
E (k ) ~ (k ) ,
(6.130)
which also considers the interaction with the lattice. Note that denotes the one-particle energy
here, not the energy density VE . It is given by
=
~2
|k | 2
2M
(6.131)
for a single atomic non-relativistic gas and by

= ~c |k |
(6.132)
for a relativistic Fermi gas. In principle, this does not change anything regarding the computation
of k f , now we just allow for a general function f (k f ). But nevertheless,
N = i
still holds, as it is a general result.
So, the energy is now computed as
E = i
= i
V
(2) 3
V
(2) 3
V 3
k
62 f
(6.133)

d3k (k ) n(k
)
k 2 <k f2

(6.134)

d
d3k [ (k )] n()
,
(6.135)
and therefore
E=N
d 1 ( ) n(
) ,
(6.136)
d3 k
[ (k )] ,
(2) 3
(6.137)
where we have dened

V
1 ( ) = i
N
108
Figure 6.4.: Energy states
the number of states per particle for given or rather , as sketched in Figure 6.4. (It corresponds
to the quantity (E) for statistics of one particle in the rst part of this lecture, as it has mass
dimension M 1 .) It is an important microscopic property of a specic Fermi gas, as it inuences
the conductivity of metals, for example.
6.5.7. 1 () for free non-relativistic fermions

Inserting the non-relativistic dispersion relation (6.131) into the expression (6.137) for 1 leads
to
!

iV
~2 |k | 2
1 () = 2
d|k | |k | 2
.
(6.138)
2 N 0
| 2M
{z }
(k )
The result is
1 ( ) =
3 3/2 1/2
( ) ,
2 f
which is only non-vanishing if > 0.

Let us check if this reproduces the correct ground state energy:
f
E0
=
d 1 ()
N
0
3 3 2 5
= f /2 f/2
2
5
3
= f
q. e. d.
5
(6.139)
(6.140)
(6.141)
(6.142)
Can we determine 1 ( ) from thermodynamics, too? Often it is dicult to calculate it from

rst principles! But while the whole function 1 () may be dicult to compute, one can infer
some important properties. For example, one would like to know how thermodynamic variables,
as specic heat, depend on 1 ( ), i. e. how they depend on the microphysics.
One uses thermodynamic observables for the measurement of microphysical properties.
109
The method, used for this, is to compute how observables depend on 1 () by leaving 1 () a
free function.
6.5.8. E(T , N ) for arbitrary temperature

We already know that
E(T , ) = N
d 1 ( )
e ( )
+1
(6.143)
which we obtain by inserting (6.108) into (6.136). It holds for arbitrary T . So, in principle, this
yields a general equation of state for NE , but only if we know the form of (T ). To compute it,
we start with the general expression for the number of particles:
X
N =
n
(6.144)

=
d n()
i V
d3 k
[ (k )]
(2) 3
(6.145)
=N
d 1 () n()
.
(6.146)
From this we can read o the normalisation condition

1
d 1 () ( )
= 1,
e
+1
(6.147)
which, for known 1 (), determines (T ).

As a cross check, we look at the T = 0 case again. In this case it is
n()
= ( f ) ,
leading to

E0 = N
d 1 ()
(6.148)
(6.149)
which agrees with (6.140).
6.5.9. Smoothing of Fermi surface

If we look at temperatures T > 0, the Fermi surface is not a sharp step function, any more.
But as long as we are still considering low temperatures, it just deviates slightly from this, as
illustrated in Figure 6.5, i. e. it becomes a smooth step function, as only electrons close to the
Fermi edge play a role at these small temperatures. This can be seen by looking at the deviation
of the energy at T , 0 from the ground state energy (T = 0) for xed N , V :
"
#

1
E E0 = N
d 1 ( ) ( )
( f ) .
(6.150)
e
+1
For T Tf the main contribution to this deviation comes from values close to f . Note that
f = f (n), which follows from the calculation at T = 0, whereas = (T ), as , f for T , 0.
A remark towards nomenclature: If electrons have an energy larger than f , we call them
(excited) particles. If they have an energy less than f , we can call them holes.
110
n
1
f
Figure 6.5.: Smooth Fermi surface
6.5.10. Computation of (T ) from the normalisation condition

We know that
"
d 1 ()
1
e ( ) + 1
( f ) = 0 .
(6.151)
To calculate it, we will perform the so-called Sommerfeld approximation: We perform a Taylor
expansion of 1 () around f :
1 () = 1 ( f ) + 10 ( f ) ( f ) + . . .
(6.152)
= f + ,
(6.153)
x = ( f ) ,
(6.154)
( ) = ( f )
(6.155)
and dene
as well as as the new variable
such that
= x .
(6.156)
This leads to
0 ( f ) x
0=
dx 1 ( f ) + 1
"
#"
1
ex e + 1
(x ) + . . .
(6.157)
Expanding it for small :

1
ex
1
(1 ) + 1
+1
1

(ex + 1) 1
(6.158)
ex
ex
ex +1
1
ex
+
,
(ex + 1)
(ex + 1) 2
(6.159)
(6.160)
111
Figure 6.6.: The dierence in between the occupation numbers at T , 0 and T = 0
yields
0 ( f ) x
0=
dx 1 ( f ) + 1
"
#"
#
ex
f (x ) + x
,
(e + 1) 2
(6.161)
where we dened
f (x ) =
ex
1
(x ) ,
+1
(6.162)
Now, (6.161) is just a linear expression in , but the x-integrations still have to be performed.
6.5.11. Occupation numbers for particles and holes

Note that the function f (x ), which characterizes the dierence in between the occupation
numbers at T , 0 and T = 0, is odd:
1
(x )
ex + 1
x
e
= x
+ (x ) 1
e +1
ex (1 + ex )
=
+ (x )
ex + 1
1
= x
+ (x )
e +1
= f (x ) .
f (x ) =
(6.163)
(6.164)
(6.165)
(6.166)
(6.167)
This means that at T , 0 the occupation number for the particles is enhanced in the same extent
as it is lowered for the holes, when compared to the occupation number at T = 0. It is easy to
this by looking at the plot of f (x ) in Figure 6.6.
We also can immediately conclude that
dx f (x ) = 0 .
112
(6.168)
Moreover, some straightforward calculations yield

2
dx x f (x )
=
,
6
ex
dx x
= 1,
(e + 1) 2
ex
dx x x
= 0,
(e + 1) 2
where the last one follows directly from
ex
(ex +1) 2
(6.169)
(6.170)
(6.171)
being even.
6.5.12. Solve for (T )

Inserting (6.168) to (6.171) into (6.161) yields
1 ( f ) =
10 ( f ) 2
.
(6.172)
By reinserting = f + , we arrive at an expression for :

= f
2 10 ( f )
(k bT ) 2 .
6 1 ( f )
(6.173)
Let us compute this explicitly for the example of free non-relativistic particles:
3 3/2 1/2

2 f
3 3 1
10 = f /2 /2
4
1 =
Therefore it is
3 1
,
2 f
3
10 ( f ) = f2 .
4
1 ( f ) =
10 ( f ) 1 1
= ,
1 ( f ) 2 f
(6.174)
(6.175)
(6.176)
and we end up with

2 k bT
= f 1
12 f
!2
!
+... .
A similar calculation can be performed for the energy:

"
#

E E0
1
=
d 1 () ( )
( f )
f
N
e
+1
2
f 1 ( f ) + (k bT ) 2 [1 ( f ) + f 10 ( f )] .
6
(6.177)
(6.178)
(6.179)
Using = f and inserting (6.173) leads to

E = E0 +
2
(k bT ) 2 1 ( f ) N .
6
(6.180)
113
3 He
gas (single atomic)

liquid
electrons in Na
white dwarfs (electrons)
neutron stars
3 He
Tf [K]
T [K]
0.006
12
37 000
1011
4 1012
300
3
300
104
Table 6.2.: Fermi temperature Tf vs. characteristic temperature T of dierent materials
By dening the Fermi temperature in a more general context via

k bTf =
3
,
21 ( f )
(6.181)
which also reproduces the result k bTf = f for free non-relativistic particles, we can bring this
into the form
2 (k bT ) 2
E = E0 +
N.
(6.182)
4 k bTf
This way, one also can consider Tf as a measure for 1 ( f ).
With an expression for the energy at hand we now can calculate the specic heat
E

.
T V ,N
(6.183)
2
T
kb N
,
2
Tf
(6.184)
cv =
It is given by
cv =
which is linear in T ! Indeed, this is in good agreement with experimental results for metals at
low temperatures! (the lattice contribution is c v T 3 ) Moreover, the measurement of Tf allows
us to determine the density of states at the Fermi edge 1 ( f ). This as an example of getting
microphysical information out of the macroscopic thermodynamic behaviour of a system! In
Table 6.2 we have listed the Fermi temperature in comparison to the characteristic temperature
of some materials.
6.6. Non-relativistic bosons

We are considering single atomic bosonic gases without any internal degrees of freedom excited
(i = 1), e. g. hydrogen. So, the mean occupation number is
n =
114
1
.
e (E ) 1
(6.185)
6.6. Non-relativistic bosons
We demand n 0. It immediately follows that < E 0 , if E 0 is the lowest energy. Without loss
of generality we set E 0 = 0, i. e. < 0 (otherwise rescale E and by some additive shift, as
p2
only their dierence E matters, which is shift invariant), for example E = 2M .
6.6.2. Number of bosons in the ground state

The mean occupation of the ground state is given by
n0 =
(6.186)
1
1 1
1
= ,
(6.187)
For small | | this becomes

n0 =
(6.188)
which can become very large for 0 . Macroscopically, n0 corresponds to Bose-Einstein

condensation, i. e. nN 0 denotes the fraction of condensed bosons (for superuids nN 0 is the
superuid density). In the thermodynamic limit V we distinguish the following cases:
n0
0
N
n0
ns > 0
N
no condensate,
(6.189)
Bose-Einstein condensate,
(6.190)
) for T 0
6.6.3. Qualitative behaviour of n(E
For smaller T at xed N there are fewer particles at high E due to stronger Boltzmann suppression. On the other hand, of course, this has to be compensated by more particles at small E .
We have plotted this in Figure 6.7. It follows that the low temperature limit of is
T 0
(T , N ) 0 .
(6.191)
J = T ln Z gc = pV .
(6.192)
(6.193)
6.6.4. Gibbs potential

The Gibbs potential was dened as
In this case, it is given by

J = T
= T i V
ln
e (E )

d3 k
T1
ln
1
e
(2) 3
~2 k 2
2M

(6.194)
115
n
T2 > T1
T2
T1
E
E0
Figure 6.7.: Occupation number for dierent temperatures at xed N
6.7. Classical ideal gas

6.7.1. Classical approximation
The classical approximation is given by
E T ,
(6.195)
~2 k 2
T .
2M
(6.196)
or, inserting E ,
Regarding the occupation number, this leads to

n =
1
e (E )
e (E )
(6.197)
for bosons and fermions, equally. So, the classical approximation is only valid for n 1.
6.7.2. Gibbs potential

Using this approximation, the Gibbs potential is given by

J = T i V
T i V
=
22
d3k T1
e
(2) 3
~2 k 2
2M
1
2 T
d|k | |k | e
~2 |k | 2
2M
(6.198)

,
(6.199)
which, of course, is again the same for bosons and fermions. So,
J = T i
116
V
e T I3 ,
22
(6.200)
6.7. Classical ideal gas
where
I3 =
0
dx x 2 ebx

2
dx ebx
0
r !
1
2 b
=
b
1 1 3
= /2b /2 ,
4
=
with
b=
(6.201)
(6.202)
(6.203)
(6.204)
~2
.
2Mk bT
(6.205)
6.7.3. Thermal de Broglie wavelength

We dene the thermal de Broglie wavelength as
2~2
(T ) =
Mk bT
! 1/2
,
(6.206)
or, using natural units, as

2
(T ) =
MT
! 1/2
.
(6.207)
Due to the relation between temperature and kinetic energy:

p2
T
2M
(6.208)
we also can associate it to the momentum via

MT p 2 .
(6.209)
Recalling the denition (6.205) of b, we nd

2 = 4b ,
(6.210)
and therefore
b /2 = 3 (4) /2 .
3
(6.211)
This way, we can express the factor I 3 in the Gibbs potential (6.200) in terms of the thermal de
Broglie wavelength:
I 3 = 22 3 .
(6.212)
117
6.7.4. Thermodynamics
The Gibbs potential is now given by
J = T i V 3 e T ,
(6.213)
J = pV .
(6.214)
as well as
So, the number of particles is computed as
J
N =
1
= J
T
pV
=
,
T
!
(6.215)
T ,V
(6.216)
(6.217)
leading to the equation of state:

pv = NT .
(6.218)
6.7.5. Validity of the approximation for a diluted gas (classical gas)

We want to check the validity of the classical approximation
p2
T .
2M
For < 0 it is sucient to check if
(6.219)
1.
T
(6.220)
pv
T
J
=
T
= i V 3 e T ,
(6.221)
We know that
N =
(6.222)
(6.223)
leading to
eT =
N 3
,
iV
(6.224)
and therefore
V
= ln i + ln
.
T
N 3
From this we can read o that the approximation is valid if
V
3 (T ) .
N
118
(6.225)
(6.226)
6.8. Bose-Einstein condensation
Therefore, recalling that T 1/2 , the classical approximation is valid for a diluted gas at
suciently high temperature:
n 3 (T ) ,
(6.227)
i. e. with far fewer than 1 particle per volume 3 (T ).

We want to examine what happens, if we keep n xed, while lowering T .
6.8.1. n()
The density is given by
N
n=
=
V
Using
d3 k
(2) 3
1
e
k2
2M
(6.228)

d3 k
1
=
d|k | |k | 2
(2) 3 22

p
1
= 2
d|k | 2 |k | 2
4
(6.229)
(6.230)
and dening
|k | 2
2MT
(T ) 2 |k | 2
=
,
4
z=
this becomes
1 4
n= 2
4 2
! 3/2
and therefore
2
n3 =
dz
0

dz
(6.231)
(6.232)
1
ez
1
.
ez 1
(6.233)
(6.234)
As is restricted to 0, we are interested in the limit 0 , i. e. 0+ . One can read

o that n3 increases in this limit.
6.8.2. Critical temperature

Let us compute n3 for = 0 (its maximal value):
2
1
3
(n )c =
dz z z
.
e 1
|
{z
}

3
2
(6.235)
3
2
119
Using

3
2
1
2
, we arrive at
(n3 )c =

3
2
(6.236)
So, the critical wavelength is given by

c =
1/3
3
2
nc /3 ,
1
(6.237)
which, using
2
,
MTc
leads us to an expression for the critical temperature:
Tc =
Using

3
2
2c =
(6.238)
2 3 2/3 2/3
2
nc .
M
(6.239)
2.61, gives us the approximate result
1 2/3
n .
(6.240)
M c
So, if we lower T until it reaches Tc for a given n, reaches 0. But what happens for T < Tc ? Is
there no solution?
Tc 3.31
6.8.3. Ground state of bosonic gas

At T = 0 the gas is in its ground state and has minimal energy. Therefore, all bosons have to be
in the state with the lowest energy E 0 . In a box with periodic boundary conditions E 0 is given
by
~2k 0 2
E0 =
,
(6.241)
2M
with ~k0 being the lowest possible discrete momentum. And in a cavity it is given by
E 0 = ~ 0 ,
with 0 being the lowest resonance frequency.
But if we now demand
n =0 (T = 0) = N ,
(6.242)
(6.243)
we have a problem, as this diverges in the thermodynamic limit. The macroscopic number of
particles in the ground state is
n =0 N
=
= n0 .
(6.244)
V
V
So, the atoms in the ground
d3 k state are not properly counted in the transition from the discrete
P
sum to the integral (2)
3 . The total number number of atoms is given by
n = n 0 (T ) + n(T
)
(6.245)
where n(T
) is the number of non-condensed thermal atoms. If n 0 , 0 we speak of Bose-Einstein
condensation (BEC).
120
n0
T
Tc
Figure 6.8.: Number of condensed atoms
Figure 6.9.: Bose-Einstein condensation
6.8.4. Bose-Einstein condensation

At T < Tc it is = 0 and the number of non-condensed thermal atoms is given by

n(T
) = 32 (T ) 3 < n ,
(6.246)
and therefore the remaining atoms have to form a Bose-Einstein condensate:

n 0 (T ) = n n(T
) > 0.
(6.247)
The qualitative behaviour of this function is plotted in Figure 6.8. At the critical temperature it
is
n 0 (Tc ) = 0
(6.248)
and it occurs a second order phase transition! For n 0 , 0 we also observe superuidity!
Experimental images of a BEC are shown in Figure 6.9.
We can express the number of condensed atoms in terms of the critical temperature. To do
so, we use
!3/2
n(T
)
3 (Tc )
T
= 3
=
,
(6.249)
n(T
c)
(T )
Tc
121
as (T ) T 1/2 . So, as n(T

c ) = n, it follows
!3/2
T
n(T
)=
Tc
n,
and therefore
(6.250)
!3/2 #
"
T
n 0 (T ) = 1
Tc
n.
(6.251)
This has consequences for the total energy, too, as only particles with E , E 0 = 0 contribute
if T < Tc :

i M 3/2 T 5/2 V
z 3/2
E=
dz z
.
(6.252)
1
2 2 ~3 |0 {ze }

5
5
2 2
So, using

5
5
2 2 =
3
4

5
2
we arrive at
E
T
0.77
NT
Tc
For comparison, for a classical gas we had
122
E
NT
= 32 .
1.341 ,
(6.253)
!3/2
.
(6.254)

7.1. Interactions and complexity
So far we only considered free particles. Adding interactions between these particles is a major
change that will lead to a considerable increase in complexity! For example we will encounter
new concepts like phase transitions and long range correlations.
Interactions are dealt with dierently, depending on the considered temperature regime. At
low temperatures only few states are excited. Collective modes of these states can be described
as interactions between quasi-particles. Examples are acoustic waves and phonons in solids or
magnons and spin waves in magnetic systems. At high temperatures, on the other hand, one
introduces interactions as perturbations of the free theory.
We will nd that many systems show some kind of microscopical order at low temperatures,
whereas they are widely disordered at high temperatures. We are going to examine transitions
between these two phases.
The aim of statistical physics is to describe all of these phenomena. Starting from known
microphysics (atomic physics), it serves as a bridge to macrophysics. The key elements to
achieve this, are statistical uctuations and correlations.
7.2. Real gases

Contrary to ideal gases, one has to consider the interactions between atoms or molecules,
respectively, to describe real gases. A typical form of the potential V (R) between these particles
is the Lenard-Jones potential:
"
V (R) = V0
R0
R
!12
R0
2
R
!6 #
,
(7.1)
with R being the distance between two particles. This potential is used to describe Van-der-Waals
forces at long distances, as well as the eect of Pauli exclusion at short distances. We have
plotted it in Figure 7.1a. A simpler approximation would be

V (R) =
V R0 6
0
R
for R < R 0 ,
for R < R 0 .
(7.2)
This describes hard spheres with an attractive interaction. It is sketched in Figure 7.1b.
123
R0
R0
(a) Lenard-Jones potential
(b) Hard spheres with attractive interaction
Figure 7.1.: Typical interaction potentials
Virial expansion
An important technique, used to describe real gases, is the virial expansion. It is an expansion
of the equation of states in terms of the density n = VN :
pV
= 1 + B 2 (T ) n + B 3 (T ) n 2 + . . . ,
N k bT
(7.3)
where the Bi are the so-called virial coecients. Using the grand canonical partition function,
we nd
p
1
ln Z gc =
V
k bT
= f (n,T )
(7.4)
(7.5)
= n (1 + B 2 (T ) n + B 3 (T ) n + . . . ) .
2
(7.6)
7.3. Mean field approximation

Another important approach is the mean eld approximation. In this approximation one
considers a molecule to see an eective potential, generated by all the other molecules. This
way, the complicated interacting system is reduced to an one-particle description with an
eective Hamiltonian. Note that this is a self-consistent description and that it does not depend
on a small density as expansion parameter! Moreover, we will work in classical statistics.
7.3.1. Classical statistics in phase space

The general Hamiltonian of a system of interacting particles is given by
H=
X p12
i
124
2M
X
ij
V (r i r j ,
(7.7)
7.3. Mean eld approximation
P
where i j means the sum over pairs of particles. The problem here is that in quantum mechanics
pi and r i do not commute. So, it is not easy to diagonalize H ! Therefore, we will consider the
classical limit for xed N . In this limit, the sum over microstates becomes
X
1 Y
N ! i=1
d3pi
.
(2~) 3
d3 r i
(7.8)
The factor N1 ! takes care of the fact that we are dealing with identical particles. It can be shown
that this is the correct limit by using the Wigner representation of the quantum density matrix.
If we had free particles, then the spatial integration would just lead to a volume factor for
each particle:
N
X
d3pi
1 NY
=
V
.
(7.9)
N!
(2~) 3
i=1
This is an exact result!
7.3.2. Average potential mean field

With the interaction potential in the Hamiltonian (7.7) one can not perform the spatial integration
that easy, any more. At this point, the mean eld approximation comes into play by replacing
X
X
V (r i r j )
Ueff (r i ) .
(7.10)
ij
Here, Ueff is an average potential seen by one particle, generated by the interaction with all
other particles.
With this replacement the canonical partition function factorizes:
Z can
f
g

2
1 Y d3pi d3r i Pj p2Mj +Pjk V (r j r k )
=
e
N! i
(2~) 3
f 2
g

i +U (r )
1 Y d3pi d3r i p2M
eff i
=
e
.
N! i
(2~) 3
(7.11)
(7.12)
7.3.3. Expansion of canonical partition function

The canonical partition function can now be expressed as
Z can
f 2
g#N
p
d3p d3r 2M
+Ueff (r )
e
(2~) 3
"
#N
N
3 Ueff (r )]
= Z ideal V
dre
1
N!
"
(1) N
= Z ideal Z ww
,
(7.13)
(7.14)
(7.15)
125
with Z ideal being the canonical partition function for an ideal gas. Using this, the free energy
becomes
F = k bT ln Z can
(7.16)
= k bT ln Z ideal N k bT
= F ideal N k bT
(1)
ln Z ww
(7.17)
(1)
ln Z ww
,
(7.18)
where F ideal is the free energy of an ideal gas:

N 2~2
= N k bT 1 ln
V mk bT
"
F ideal
!3/2 # !
,
(7.19)
or, splitting up the logarithm,

F ideal = N k bT + N k bT ln
N 3
mk bT
N k bT ln
.
V
2
2~2
(7.20)
(1)
7.3.4. Computation of Z ww
(1)
To compute this deviation Z ww
from the ideal gas partition function, we have to consider two
eects:
1. excluded volume,
2. attractive interaction.
It is
(1)
Z ww
1
=
V
d3r e Ueff (r ) .
(7.21)
This expression will be modied due to the two mentioned eects:

1. Vex is the region, where Ueff .
2. In the remaining space we exchange Ueff by a mean potential W :
e W =
1
V Vex
d3r e Ueff (r ) .
(7.22)
V Vex
Together, this leads to

(1)
Z ww
=
126
V Vex W
e
.
V
(7.23)
7.3. Mean eld approximation
7.3.5. Excluded volume

We have to determine Vex , the mean excluded volume for an atom. To begin with, the total
excluded volume is N Vex , which can be expressed as
N Vex =
1
N (N 1)
2 {z }
|
4 3
a ,
3
|{z}
(7.24)
number of pairs excl. volume

per pair
12 N
with a being the minimal distance between two particles, i. e. a = R 0 in Figure 7.1b. So, we nd
1 4 3
N
a
2
3
2 3
=
a N
3
= b0 N ,
Vex
(7.25)
(7.26)
(7.27)
where we dened the virial coecient

b0 =
2 3
a .
3
(7.28)
7.3.6. Mean potential

Moreover, we have to compute the W , the mean potential seen by a given atom. We will do
this in a similar fashion. The potential energy of all atoms is given by
N W =
1
N (N 1) u ,
2
with u being the mean potential energy per pair. It is

u = 4
dR R 2 W (R) (V Vex )1 ,
a
(7.29)
(7.30)
and therefore
1
W = N u
2

2N
=
dR R 2 W (R) .
V Vex a
(7.31)
(7.32)
Introducing another virial coecient b1 , we can write this as

N
W = b1 .
V
(7.33)
Note that W n!
127
Altogether, we arrive at
V b0 N kWT
e b ,
V
V b0 N
W
= ln
V
k bT
V b0 N
N
= ln
+
b1 .
V
V k bT
(1)
Z ww
=
(1)
ln Z ww
(7.34)
(7.35)
(7.36)
7.3.7. Free energy

Using the result (7.36), the free energy becomes
(1)
F = F ideal N k bT ln Z ww
N 3
mk bT
V b0 N
Nb1
= N k bT + N k bT ln N k bT ln
N k bT ln
N k bT
.
2
V
2
2~
V
k bTV
(7.37)
(7.38)
So, we arrive at
F = N k bT + N k bT ln
N
3
mk bT
N2
N k bT ln
b
.
1
V b0 N 2
2~2
V
(7.39)
7.4. Equation of state

Using the general relation
p=
F

,
N N ,T
(7.40)
we arrive at the equation of state:

N k bT
N
p=
b1
V b0 N
V
!2
.
(7.41)
Note that here only the structural form is important. The computation of b0 and b1 can be
improved by a self-consistent computation of Ueff (r .
Often, it is more useful to express these equations just in terms of the density n = VN and T .
So, instead of F we look at VF :
F
n
3
mk bT
= n k bT + n k bT ln
n k bT ln
b 1n 2 ,
V
1 b 0n 2
2~2
(7.42)
and the equation of state becomes

p=
nk bT
b 1n 2 .
1 b 0n
(7.43)
This is the van der Waals equation. One nds it to describe the phase transition from gas to
liquid qualitatively correct!
128
7.5. Specic heat of real gases
7.5. Specific heat of real gases

The specic heat was dened as
S

T T ,N
2 F
= T 2
T
cV = T
(7.44)
(7.45)
V ,N
S
,
=V
lnT V n
with
S
V
(7.46)
the entropy density:

!
S
F

=
V
T V n
= k bn k bn ln
(7.47)
n
3
mk bT 3
+ k bn ln
+ k bn .
1 b 0n 2
2~2 2
(7.48)
So,
!
S
5
n
3 mk bT
= k bn
ln
+ ln
,
V
2
1 b 0n 2
2~2
which is independent of b1 .
This leads to
cV
s
=
V
lnT V
!
= 3 k n ,
b

2
n
(7.49)
(7.50)
and we nally arrive at

cV =
3
kb N .
2
(7.51)
So, regarding the specic heat there is no dierence between a van der Waals gas and an ideal
gas!
7.6. Equipartition theorem

At high temperatures one nds the specic heat to be proportional to N with a system dependent
prefactor, for example:
Monatomic (real) gas:
Phonons:
3
kb N ,
2
cV = 3 k b N .
cV =
(7.52)
(7.53)
In general, it is
cV =
1
kb N f1 ,
2
(7.54)
129
with f 1 being the number of excited degrees of freedom per particle. The total number f excited
degrees of freedom is f = N f 1 . So, for example one nds:
Monatomic gas:
f1 = 3
(translations),
(7.55)
Phonons:
f1 = 6
(position and velocity),
(7.56)
Diatomic molecule:
f 1 = 3 + 2 + (2)
(translations, rotations and vibrations). (7.57)
Regarding the diatomic molecule: note that one (longitudinal) vibration corresponds to f 1 = 2.
Alternatively, one can split up its total f 1 = 7 into 6 velocity components and the distance
between the atoms.
The relation (7.54) is always true, if the quadratic approximation for the energy holds:
X
E = E0 +
b q 2 .
(7.58)
Here, the q are the excited degrees of freedom of the system and the b are suitable prefactors.
P
labels the excited degrees of freedom and therefore = N f 1 . We denote the non-excited
degrees of freedom by q.
So, for a general degree of freedom qr one nds qr {q , q}.
The average energy of such a system is computed as

hEi =
P
2
P
Dqr (E 0 + 0 b 0q 2 0 ) e E0 b q
,
P

2
Dqr e E0 b q
(7.59)
where

Dqr =
dq 1 dqr dq N
(7.60)
Dq Dq .
(7.61)
It is easy to see that the Dq integration and the factor e E0 in enumerator and denominator
cancel. Moreover, we can pull the E 0 out of the remaining Dq integration. We end up with
b q 2
dq e
hEi =
Q
130
!
P
2
0 b 0q 0
b q 2
dq e
(7.62)
7.6. Equipartition theorem
For each summand only one integration is non-trivial:

b 0 q 2
dq 0 b 0q 2 0 e 0
hEi E 0 =

b 0 q 2
dq 0 e 0
0
!
X
b 0 q 2
=
ln dq 0 e 0
0
!
r
X
ln
b 0
0
| {z }
X
(7.63)
(7.64)
(7.65)
12 ln + const.
X 1
=
.
2
0
(7.66)
And therefore,
1
k bT N f 1 .
(7.67)
2
Note that we implicitly assumed a classical treatment, i. e. we neglected the discreteness
of energy levels. This is only valid if the dierence between the ground state energy and the
energy levels in the -channel is small compared to k bT :
hEi = E 0 +
k bT E 1, E 0 .
(7.68)
131
8. Magnetic systems
8.1. Magnetization
8.1.1. Degrees of freedom
We consider a lattice with points x. In this case it is a cubic lattice:
x = na ,
(8.1)
n = (n 1 ,n 2 ,n 3 )
with nk Z ,
(8.2)
where a is the lattice spacing. There are classical spins a (x ) at each lattice site x which satisfy
3
X
a (x ) a (x ) = 1 .
(8.3)
a=1
The microstates of this system are given by the possible spin congurations a (x ), the
so-called lattice elds. Moreover, we have a Hamiltonian
H = H [a (x )] ,
(8.4)
where we identify H E . Of course, this can be generalized to N -component elds:

a (x )
N
X
with a = 1, . . . , N ,
a (x ) a (x ) = 1 .
(8.5)
(8.6)
a=1
One example of such a system is the Ising model, which is characterized by N = 1 and
(x ) = 1 (discrete number of microstates). So, its microstates are given by
: { (x )} {s j }
with s j = 1
(8.7)
and the corresponding probabilities take the form

p =
e E
e H
=
,
Z
Z
(8.8)
as we are describing it as a canonical ensemble. Note that we are not dealing with uncorrelated
probabilities as in chapter 3, any more!
133
8. Magnetic systems
e2
e1
Figure 8.1.: Unit lattice vectors
8.1.2. Hamiltonian
We are considering the following Hamiltonian:
XXX
XX
H =
a (x ) a (x + e ) c B
a (x ) Ba .
x
{z
(8.9)
H int
Here, c B is the magnetic moment, for which we assume

cB > 0 .
(8.10)
e is the unit lattice vector in -direction, as shown in Figure 8.1. Of course, the sum over
runs from 1 to the dimension d of the lattice (d = 3 for a cubic lattice):
X
d
X
(8.11)
=1
So, the properties of H int are that it only considers next-neighbour interactions and that it tries
to align the spins. On the other hand, B is a magnetic eld which tries to align the spins in the
same direction as B.
The ground state of the system is characterized by the limit . In this limit the spin
expectation values will be proportional to the magnetic eld:
ha (x )i Ba .
(8.12)
Note that the coupling multiplying H int can be absorbed into 0.
8.1.3. Magnetization
The magnetization is dened as
Ma =
ha (x )i .
(8.13)
Since Ba is an intensive quantity, Ma is an extensive one. Now, the question is: How does the
magnetization depend on the magnetic eld?
M (B) = ?
134
(8.14)
8.1. Magnetization
8.1.4. Thermodynamic potential

As always, the canonical partition function is given by
X
X
Z=
e H =
e E .
(8.15)
From (8.9) we can read o that it includes a source term, which is linear in the eld (x ).
Compenentwise, the corresponding source eld is given by
ja = c B Ba .
(8.16)
So, instead of considering the partition function as B-dependent, we also can consider it as
j-dependent:
Z (T ,B) = Z (T , j) ,
(8.17)
which is a mere change of notation. But we will stick to the notation using j, as it is the same
notation as in QFT. Note that the volume V and the particle number N do not play a role here
they are no thermodynamic variables!
Again, we can perform the transition from a sum over states to a functional integral:
! X
#
X Y" Y
da (x )
a (x ) a (x ) 1 .
(8.18)
On the other hand, for the discrete Ising model (N = 1) we just have
X Y X
=
.
(8.19)
(x )=1
We dene
W (T , j) = ln Z (T , j) ,
(8.20)
which is related to the free energy in the following way:

F (T ,B) = TW (T , j) .
(8.21)
This way, the basic relation for the magnetization takes the form
W
= Ma
ja
(8.22)
which can be seen by
W
1 Z
=
ja
Z ja
1 X Hint +Px
=
e
Z ja
*X
+
=
a (x )
(8.23)
a (x ) j a (x )
(8.24)
(8.25)
= Ma .
(8.26)
135
8. Magnetic systems
Therefore, one also can compute the magnetization from the free energy:
M
F
,
B
(8.27)
i. e. we can identify some analogies to our description of gases:

B ,
(8.28)
MN,
(8.29)
(Gibbs potential).
(8.30)
W (T , j)
as thermodynamic potential,
(8.31)
T,j
as thermodynamic variables.
(8.32)
F J
So, in the following we will use
8.1.5. Eective potential

We denote the number of lattice sites by N . Then Ma N . So, we introduce the magnetization
per lattice site a via
Ma
a =
.
(8.33)
N
Moreover, we dene the eective potential U ( a ,T ):
U =
g
1 f
W (j,T ) + Ma ja ,
N
(8.34)
where the sum over repeated indices is implied. Obviously,

U =
W (j,T )
+ a ja ,
N
(8.35)
which is a Legendre transformation.

Note that the sources ja depend on the b . The exact dependence can be inferred by inverting
the relation
1 W
a (jb ) =
,
(8.36)
N jb
leading to
U
= ja .
a
(8.37)
This is the basic relation between magnetization and magnetic eld, which can be proven by
direct computation:
!
U
jb
W (j,T )
= ja +
b
.
(8.38)
a
a
jb N
| {z }
b
In the following we want to compute U ()!
136
8.2. Spontaneous symmetry breaking
8.1.6. Minimal principle

We dene
U (, j) = U () j .
(8.39)
The magnetization is determined by the minimum of U (, j) with respect to at xed j:
U
= 0.
j
(8.40)
Ma = N 0,a .
(8.41)
N U = U jM
(8.42)
= W
F
= ,
T
(8.43)
We denote this minimum by 0 . Then it is
We can relate U and F via
(8.44)
where F (T , j; M ) is the free energy with the constraint that M takes the given value (by
inverting M (j), one gets the free energy F (T , j) in standard form). This is similar to the entropy
of a system with constraint y (S (y) maximal F (y) minimal).
What is the probability to nd a given M? To answer that, we need the probability distribution
w (M ):
w (M ) =
X

P
e H [a (x )]
x
(M ) e E (M )
S (M ) T1
=e
)
F (M
T
=e
(8.45)
a (x )=M a
E (M )
(8.46)
(8.47)
(8.48)
where we used F = E T S in the last step. From this we conclude that the maximum of w (M )
corresponds to the minimum of F (M ) and therefore to the minimum of U (). For j = 0 this
then corresponds to a minimum of U :
U
= 0.
a
(8.49)
8.2. Spontaneous symmetry breaking

8.2.1. Symmetries
Obviously, H is invariant under lattice translations. Moreover, H int is invariant under lattice
rotations and reections. But these symmetries are (at least partially) broken by B , 0 (j , 0).
137
8. Magnetic systems
O (N )-symmetry
a (x ) and ja (x ) are N -component vectors. So, H int is
invariant under N -dimensional rotations, i. e. SO (N )-invariant,
invariant under a (x ) a (x ) (internal reection).
From a group theoretical point of view the symmetry group SO (N ) together with reections
P
yields the symmetry group O (N ), which describes all linear transformations that leave a (a a )
invariant. Therefore, H int is O (N )-invariant.
But the source term breaks this O (N )-symmetry for xed j , 0. So, only simultaneous rotations
of a (x ) and ja (x ) leave H invariant. From this we can conclude that W (j,T ) is invariant under
O (N ) rotations of ja and that, as a also is a vector under rotations, U ( a ) is invariant under
O (N ) rotations of a .
This symmetry of U (N ) leads to a simplication of the computations. As already said, U is
invariant with respect to O (N )-symmetry and the explicit breaking by a magnetic eld appears
only in the source term. Therefore, the solution of
U
= ja
a
(8.50)
is not O (N )-symmetric for a given ja , 0. This means, that the magnetic eld singles out a
particular direction.
8.2.2. Vanishing magnetic field

What happens for B = 0 ? Or more precisely, what happens in the limit ja 0 ? There
are two possibilities. Either there is a non-vanishing magnetization a , 0 in an arbitrary
direction, as no direction is preferred (ground-state degeneracy''). If this happens, we speak of
spontaneous symmetry breaking (SSB). Or there is no magnetization ( a = 0). In this case the
O (N )-symmetry is preserved.
Example
In the Ising model, = 1 and = 1 are equivalent. So the magnetisation has to choose
one of the two possible values. In the ground state, the minimum of H , all spins are
aligned, so
either
(x ) = 1
for all x, = 1 ,
(8.51)
or
(x ) = 1
for all x, = 1 .
(8.52)
8.2.3. Domains
A domain is a continuous region of spins pointing in the same direction. In Figure 8.2 we
sketched three neighbouring domains. At the boundary between two domains there is a positive
138
8.3. Landau theory and phase transition
domain
+ domain
Figure 8.2.: Spin domains
interaction energy. Therefore, the probability p of a domain-microstate is smaller than the

probability p of the ground state. But on the other hand, there are more domain-states than
ground states (only two in the Ising model).
The magnetization is given by
X
M=
p M .
(8.53)
But what wins? p or the number of contributions? We expect, that for low T the system will
have near ground state order with , 0 (SSB), and that for high T the system will be completely
disordered with = 0 (SYM). There will be a phase transition between these two cases at the
critical temperature Tc .
Note that in practice domains are visible as Weiss domains in ferromagnets.
8.2.4. General form of U

As we have already pointed out, the eective potential U ( a ) must be invariant under rotations.
Let us dene
= 12 a a .
(8.54)
This is the simplest expression, which is invariant under rotations (O (N ) includes all transformations that leave a a invariant). Therefore, U has to be a function of . This simplies the
problem, as now U only depends on the one variable and not explicitly on N variables a .

8.3.1. Quartic expansion of U
We expand U up to order 4 ( 2 ):
U = U0 + 2 + 12 2 ,
(8.55)
where U0 is irrelevant for the magnetization, as it drops out when performing the derivative:
U
= ja .
a
(8.56)
We assume > 0, as otherwise U has no absolute minimum.

Derivatives with respect to a can be expressed in terms of derivatives with respect to :
U
U
U
=
=
a .
a
a
(8.57)
139
8. Magnetic systems
8.3.2. Phase transition for j = 0

We look for the minimum of U (as U = U for j = 0):
0=
U
= 2 a + a
a
= ( 2 + 12 b b ) a .
(8.58)
(8.59)
There are two cases:

2 > 0: Only solution is a = 0 (no magnetisation, SYM, Figure 8.3).
2 < 0: Minimum of U for = 0 (SSB, Figure 8.4), where
0 =
2
,
0 0 =
2 2
.
(8.60)
This is the so-called mexican hat potential.

In between, at 2 = 0 there is a phase transition (Figure 8.5) At this point there is no quadratic
term and the potential is just given by
U =
( a a ) 2 .
8
(8.61)
Because | 0 |( 2 ) is a continuous function, as one can see in Figure 8.6, we speak of a second
order phase transition.
Critical temperature
2 = 2 (T ) ,
(8.62)
= (T )
(8.63)
are continuous functions of the temperature. Thereby, 2 (T ) is monotonically increasing and

we know that
2 (T = 0)
< 0,
(8.64)
(T large) > 0 .
(8.65)
2 (T = Tc ) = 0 ,
(8.66)
The critical temperature is dened via
i. e. it is the temperature where the phase transition occurs. The resulting phase diagram is
sketched in Figure 8.7.
140
U
2 > 0
Figure 8.3.: Eective potential in the symmetric phase (SYM)
U
2 < 0
Figure 8.4.: Eective potential in the spontaneously broken phase (SSB)
U
2 = 0
Figure 8.5.: Eective potential at the phase transition
141
8. Magnetic systems
| 0 |
SSB
SYM
Figure 8.6.: | 0 | as a function of 2
T
Tc
Figure 8.7.: Phase diagram of Landau theory
8.3.3. Discontinuity in magnetization for non-zero B

Now, we are considering the general case j , 0. So, up to quartic order in , U is given by
U = 12 2 2 + 18 4 j ,
(8.67)
which we have plotted in Figure 8.8 for dierent cases of j So, we are looking for the solutions
of
U
0=
= 2 + 12 3 j .
(8.68)
This is a cubic equation, i. e. there are either one or three extrema.

The position of the absolute minimum is called the order parameter. In this case, when
varying B (i. e. j), there is a jump in this order parameter, a discontinuity, as you can see from
Figure 8.9. What is the origin of this? The physical phenomenon behind this is hysteresis.
142
8.4. Magnetic susceptibility
U = U j
j>0
j=0
j<0
Figure 8.8.: Modied eective potential U for dierent cases of j
0
j
Figure 8.9.: Jump of the order parameter

The magnetic susceptibility is the isothermal change of magnetization with magnetic eld.
N =1
It is given by
T =
or equivalently,
=

,
B T
(8.69)
.
j
(8.70)
(8.71)
By inverting it:
1 =
143
8. Magnetic systems
2
Figure 8.10.: Magnetic susceptibility for j = 0
and using j =
, we nd (j = 0)
1 =
2U

.
2 =0
(8.72)
In the two dierent phases this is computed as

SYM-phase ( 2 > 0):
1 = 2
SSB-phase ( 2 < 0):

1 = 2 + 32 2
=0
2
= + 3 0 .
1
,
2
(8.73)
(8.74)
(8.75)
For j = 0 the latter can be evaluated, using 0 = (for 2 < 0):

1 = 2 0 = 2 2
1
1
= 2.
2 0
2
(8.76)
So, by combining (8.73) and (8.76), we can plot the magnetic susceptibility for j = 0 over the
complete 2 range in Figure 8.10. We see that it diverges at the phase transition ( 2 0, T Tc ).
For j , 0 we would nd a nite susceptibility at Tc .
N >1
In this case, we have a susceptibility tensor:
a
.
jb
(8.77)
2U
,
a b
(8.78)
ab =
Dening the mass matrix
2
mab
=
144
we nd the matrix identity

= (m2 ) 1 .
(8.79)
This is easily proven by

2
ab mbc
a U
=
jb c b
a jb
=
jb c
a
=
c
= ac .
!
(8.80)
(8.81)
(8.82)
(8.83)
For j = 0 the mass matrix is given by

2
mab
=
2U
,
a b 0
(8.84)
with
U = 2 + 12 2 .
(8.85)
2
mab
= 2 ab ,
(8.86)
In the SYM-phase this leads to

which has N equal eigenvalues 2 . So, the susceptibility tensor then has N equal eigenvalues
2 . In the SSB-phase on the other hand, it is
2
mab
= ( 2 + 0 ) ab + 0,a 0,b .
(8.87)
Take
0,1 = 0 ,
0,a,1 = 0
0 = 12 02 ,
(8.88)
for a , 1, b , 1,
(8.89)
which leads to
m211 = 2 0 ,
1
mab
=0
and therefore
2 0 0 0
*.
+
0
0
0//
.
2
m = .. .
. . .. // .
. ./
. ..
0 0, 0
(8.90)
So, there are

N 1
massless modes (Goldstone modes) and
(8.91)
massive mode (radial mode).
(8.92)
This means that in the SSB- phase the susceptibility tensor has N 1 diverging eigenvalues for
j 0, 0 const. , 0 (N diverging eigenvalues for j 0, 0 0).
145
8. Magnetic systems
In the special case of N = 2 this corresponds to a complex eld

=
( 1 + i 2 ) ,
(8.93)
with
= = = 12 ( 12 + 22 ) .
(8.94)
In this case the system experiences superuidity in the SSB-phase, which is related to the
Goldstone mode.
The case of N = 4 corresponds to two complex elds. Its relativistic version is the Higgs
mechanism (Goldstone bosons become longitudinal components of massive vector bosons in
presence of local gauge symmetry).
146
9.1. Order of phase transition
We speak of phase transitions when we observe non-analytic behaviour of thermodynamic
potentials or their derivatives, while varying thermodynamic variables.
9.1.1. First order phase transition

A rst order phase transition is characterized by a discontinuous jump of the expectation value of
a thermodynamic potential or its derivative. In chapter 9 we already encountered an example of
this: the discontinuity of the magnetization if the magnetic eld changes direction, as sketched
in Figure 8.9 (recall that M and B j). In this example, the thermodynamic potential U ,
that needs to be minimized, has two separated local extrema. The absolute minimum changes
from one to the other at B = 0, as illustrated in Figure 8.8. At B = 0 both minima have equal U .
Of course, this is only the special case of a magnetic system. In general, rst order phase
transitions may occur when varying some other quantity of a system, like temperature or
pressure. For example, this is the case for the liquid-gas phase transition. There, the absolute
minimum of the corresponding thermodynamic potential changes from liquid to gas phase at
T = Tc , as we will work out later in this chapter. It is illustrated in Figure 9.2.
9.1.2. Second order phase transition

Contrary to a rst order phase transition, a second order phase transition is characterized by a
continuous behaviour of the expectation value. But susceptibilities or other derivatives diverge.
Again, we already encountered an example of this: the magnetization for j = 0 (B = 0). In that
case, the eective potential was given by
U = 2 (T ) + 12 2 .
(9.1)
Thereby, 2 (T ) increases monotonically and 2 (Tc ) = 0. For temperatures T near the critical
temperature Tc it behaves like
2 = a (T Tc ) + . . . ,
(9.2)
with some constant a. From this the behaviours of 0 and 0 follow directly:
2 (T )
a (Tc T )
0 (T ) =
=
,
q
p
0 (T ) = 2 0 (T ) = c Tc T ,
(9.3)
(9.4)
147
where c =
2a
.
We already had sketched this behaviour in Figure 8.7.
In general, one nds 0 to behave like a power law near the critical temperature:
0 (Tc T ) ,
(9.5)
with being a so-called critical exponent. In this case = 12 (mean eld value).
For our example the derivative of 0 with respect to T is given by
0
1

.
T
Tc T
(9.6)
Indeed, this diverges for T Tc .

Assuming j , 0 again, the eective potential at Tc is computed as
U = 12 2 ,
(9.7)
U = 18 ( a a ) 2 ja a .
(9.8)
and therefore
What is the form of 0 (j) in this case? To compute it, we take ja = j a1 , leading to 0,a = 0 a1 .
Then, we nd
U
j=
= 03 ,
(9.9)
1 2
and therefore
0 j /3 ,
1
M B /3 .
1
(9.10)
In general, one nds M to behave like

M B / ,
1
(9.11)
with being another critical exponent. In this case = 3 (mean eld value).
The derivative of M with respect to B is given by
M
2
B /3 ,
(9.12)
which diverges for B 0.

The origin of all of this is that the minimum of the thermodynamic potential U splits up
into two minima at T = Tc , as shown in Figures 8.3 to 8.5 (recall that 2 < 0, 2 = 0 and
2 > 0 correspond to T < Tc , T = Tc and T > Tc , respectively). This is typical for spontaneous
symmetry breaking and can occur in other situations, too.
148
9.1. Order of phase transition
w
T < Tc
w
T = Tc
w
T > Tc
Figure 9.1.: Probability distribution for dierent temperatures
9.1.3. Probabilistic interpretation

The probability to nd some value is given by
w () ec U ( ) ,
(9.13)
where c in the thermodynamic limit, as c V or c N for lattice sites. The two

non-degenerate local minima of U correspond to two local maxima of the probability distribution,
as sketched in Figure 9.1. Thereby, the relative probability of these two maxima depends on the
temperature.
A dierent state becomes most probable at a rst order phase transition.
There is a striking phenomenon: The microscopic properties of water molecules are the same
for T < Tc and T > Tc , but a very small change in the average kinetic energy (temperature)
produces qualitatively dierent properties of water and vapor!
149
Magnetic systems
Real gas
j
U
U
v
p
f
Table 9.1.: Magnetic systems vs. real gas
9.2. Liquid-gas phase transition

9.2.1. Eective potential
The Van der Waals equation of state for real gases is given by
k b NT
N
p=
b1
V b0 N
V
!2
,
(9.14)
with b0 , b1 being the so-called virial coecients. The free energy F is related to the pressure via
F
= p ,
V
(9.15)
and has the explicit form

"
F = k bT N ln
N
3 mk bT
1 ln
V b0 N
2
2~2
In the following we will use the intensive variable v =

spondingly, we dene
F
f = ,
N
such that
V
N
b1
1
n
N2
.
V
(9.16)
(volume per particle). Corre(9.17)
f
= p ,
v
(9.18)
(v,T
,p) = f (v,T ) + pv .
(9.19)
and
plays the role of U for magnetic systems, i. e. the maximal probability w (v) corresponds to
the minimum of (v):

= 0.
(9.20)
v T ,p
In Table 9.1 we have listed the corresponding quantities of magnetic systems and real gases.
But note that there is one important dierence: there is no O (N )-symmetry for the real gas.
150
9.2. Liquid-gas phase transition
Liquid
Gas
Liquid
Gas
v
(a) T < Tc
(b) T > Tc
Figure 9.2.: Eective potential 1 above and below the critical temperature
n
Liquid
Gas
T
Tc
Figure 9.3.: Jump in density during liquid-gas transition
9.2.2. First order transition

We look for local minima of (v)
for xed T , p. Therefore, we split up (v)
into a v-dependent
and a non-v-dependent part:
(v)
= 0 (T ) + 1 (v,T ,p) ,
(9.21)
with (k b = ~ = 1)
b1
+ pv ,
v
(9.22)
1
T
b1
=
+ 2 +p .
v
v b0 v
(9.23)
1 = T ln (v b0 )
leading to
0=
We nd that the equation of state has three solutions for given T , p (2 minima, 1 maximum),
as sketched in Figure 9.2. For T = Tc both local minima have equal 1 . Moreover, there is a jump
in density at T = Tc , as shown in Figure 9.3.
151
9.2.3. Compressibility
The isothermal compressibility T is dened as
p
1

= V
.
T
V T ,N
(9.24)
Note that V < 0, as larger pressure corresponds to smaller volume. Therefore T > 0 (if T < 0,
this would be an unstable situation).
So, in terms of intensive variables,
p
1
= v
,
T
v
and therefore
T
b1
=
+ 2
v T v v b0 v
(9.25)
!
.

T
(9.26)
!
2
T
b1
=
,
+
v 2 T ,p v v b0 v 2 T
(9.27)
2
1
= v 2 .
T
v v0
(9.28)
By comparing this to
we nd
As T > 0, this corresponds to a minimum of .

Moreover, we conclude that T v corresponds
to the susceptibility in magnetic systems.

The chemical potential is given by
= (v
0) ,
(9.29)
= f + pv 0
1
= (F + pV )
N
1
= (F JGibbs )
N
1
= (N )
N
=.
(9.30)
JGibbs = F N = pV
(9.35)
as one can see as follows:
(9.31)
(9.32)
(9.33)
(9.34)
Recall that
152
9.3. Phase diagram
and note that N is the free enthalpy

G (N ,T ,p) = F + pV .
(9.36)
We are considering chemical reactions where N atoms of a given species can be bound either
in molecule A or molecule B. The corresponding variable is the concentration, e. g.
cb =
#molecules B
.
#molecules A + #molecules B
It replaces v. Then, the probability to nd some value c b is proportional to e
(9.37)
N (c b )
T
9.3. Phase diagram

9.3.1. Vapor pressure curve
The vapor pressure curve describes the coexistence of uid and gas at the critical value of (p,T ).
There are two conditions:
1.
v
v2
= 0, from which we get the equation of state and the solution for the two minima v 1 ,
as a function of p,
2. (v
1 ) = (v
2 ), which is equivalent to l = g (liquid-gas coexistence condition) and from
which we get the vapor pressure curve pc (T ).
So, there is only one free parameter (for xed N , V ), e. g. T or p:
given p
Tc ,
(9.38)
given T
pc .
(9.39)
9.3.2. p T diagram
The free enthalpies of the two phases are
G l = N l l (p,T ) ,
G g = N g g (p,T ) .
(9.40)
We distinguish the following cases:

l < g :
all particles in liquid phase (N l = N , N g = 0),
(9.41)
l > g :
all particles in gas phase (N l = 0, N g = N ),
(9.42)
l = g :
transition line (N l , N g , 0, N l + N g = N ).
(9.43)
The transition line is described by p = pc (T ), which we have sketched in Figure 9.4. So,
coexistence in equilibrium is possible. Note that this is a universal curve, as it is independent of
Nl , Ng .
153
p
l = g
Liquid
l < g
Gas
l > g
T
Figure 9.4.: p T transition line between liquid and gas phase
9.3.3. Clausius-Clapeyron relation

We can set up a dierential equation for the transition line by combining the coexistence
conditions of two innitesimally distant points on the transition line (changing the index
notation from l,g to 1,2):
1 (T ,p) = 2 (T ,p) ,
(9.44)
1 (T + dT ,p + dp) = 2 (T + dT ,p + dp) ,
(9.45)
d 1 = d 2 ,
(9.46)
to arrive at
with
d i =
i
i
dT +
dp .
T
p
|{z}
|{z}
S
N
(9.47)
V
N
The latter is the so-called Duhem-Gibbs relation:

V
S
d = dp
dT .
N
N
|{z}
(9.48)
When expressing it in terms of intensive quantities and combining it with (9.46) we arrive at
v 1 dp s1 dT = v 2 dp s2 dT ,
(9.49)
(s s ) dT = (v 2 v 1 ) dpc ,
| 2{z 1}
| {z }
(9.50)
which we can rewrite as

s
to nd the Clausius-Clapeyron equation:

dpc
s
=
.
dT
v
(9.51)
It tells us that the slope of pc (T ) is given by the ratio of entropy dierence and volume dierence.
154
9.3. Phase diagram
9.3.4. Latent heat

There is a jump in entropy at the phase transition. So, heat has to be absorbed or released (the
phase transition can also be adiabatic if the change in T is slow). It is called latent heat and is
given by
L 12 = T S = T (S 2 S 1 ) .
(9.52)
Combining it with the Clausius-Clapeyron equation gives us
L 12 =
dT
T V .
dpc
(9.53)
The latent heat is necessary to separate the water molecules during the transition from the
liquid to the gas phase or to break up the crystal structure during the transition from the ice
(solid) to the liquid phase, respectively.
9.3.5. Superheating, supercooling

Let us introduce the following time scales:
t a : typical time scale of establishing the equilibrium in the liquid (relaxation time),
t b : typical time scale of establishing the equilibrium in the gas,
t t : time scale of the liquid-gas transition (forming of droplets / bubbles).
For T < Tc the probability to be in the gas phase is much smaller than the probability to be in
the liquid phase:
wb wa ,
(9.54)
as sketched in Figure 9.5. But for the states in between these two phases it is even smaller:
w (v) w b
for v a < v < v b .
(9.55)
Therefore, small uctuations near B, e. g. at x, tend to return to B (stability of B regarding small

uctuations). An equivalent explanation would be to say that B is a local minimum of G, even
if T < Tc , as illustrated in Figure 9.6.
The typical time scale of a uctuation returning to B is given by t b , which is usually much
shorter than t t (overcoming a barrier; similar to tunneling in QM: exponentially suppressed
probabilities). Let us now consider an experimental time scale t e in between these:
tb te tt .
(9.56)
In this case, the system has not enough time for the transition B A, but instead stays in the
local equilibrium B even beyond the critical point. This phenomenon is called supercooling
(one speaks of superheating if during the transition A B the systems stays in A beyond the
critical point). Eventually, the local instability will become extremal when B becomes a saddle
point and nally a maximum. In practice one can observe the coexistence of water and vapor
not only at Tc .
For magnets the same eect appears in the form of hysteresis. Indeed, this is a characteristic
eect for rst order transitions in general.
155
w
T > Tc
T < Tc
A
Liquid
B
Gas
Figure 9.5.: Probability distributions for liquid and gas phase
G
T < Tc
Figure 9.6.: Free enthalpy in liquid phase
156

10.1. Critical exponents
For second order phase transitions the eective potential describes correctly the qualitative
behaviour of the system in the regimeT Tc . But a closer examination of the singularities yields
values for the critical exponents, which deviate from the simple mean eld approximation we
used in subsection 9.1.2.
In the following we assume B = 0 for magnetic systems or p = p , the pressure at the endpoint
of the critical line, for liquid-gas transitions. These are just the conditions to observe second
order phase transitions. Moreover, we parametrize the behaviour for T Tc by power laws,
with critical exponents.
10.1.1. Specific heat

The power law for the specic heat reads
c (T ) = c reg + A |T Tc | ,
(10.1)
with c reg being the simplest approximation expression for the specic heat, being a critical
exponent, and A being some constant. We have plotted (10.1) in Figure 10.1 for dierent cases
of .
10.1.2. Order parameter as function of temperature

The order parameter of magnetic systems is the magnetization per lattice site = M
N . Its
temperature dependence was illustrated in Figure 8.7. Close to the critical temperature one has
M
(Tc T )
N
(10.2)
for T < Tc (compare (9.5)) and M = 0 for T > Tc . Here, is another critical exponent.
10.1.3. Susceptibility
The susceptibility in magnetic systems corresponds to the compressibility T in uids. At
B = 0 / p = p both behave the same close to the critical temperature:
|T Tc | ,
(10.3)
(10.4)
T |T Tc |
This hold for T > Tc as well as for T < Tc . They both share the same critical exponent .
157
(a) < 0
(b) > 0
Figure 10.1.: Specic heat for dierent values of
10.1.4. Order parameter at critical temperature as function of the source

We can take a look at the source-dependence of the order parameters, too. For magnetic
systems this source is the external magnetic eld B, for uids it is the pressure p (or rather its
deviation from the critical value p ). Again, both types of systems show the same behaviour:
|n(p)| |p p | / ,
M

1
(B) |B| / ,
N

1
(10.5)
(10.6)
with some critical exponent . For magnetic systems we already found this relation in (9.11).
10.1.5. Mean field exponents

Let us rst calculate / collect the mean eld values of the critical exponents. In general, they are
extracted from an eective potential of the form
U =
A (T Tc ) 2 + 4 + U0 ,
2
8
(10.7)
as already used in (9.1). We will need its rst two derivatives:
= A (T Tc ) + 3 ,
2
2
U
3
= A (T Tc ) + 2 .
2
(10.8)
(10.9)
The rst derivative also directly gives us the position of the minimum for SSB:
1 2 A (Tc T )
=
,
2 0
as we already had found in (9.4). For SYM it is 0 = 0.
158
(10.10)
10.1. Critical exponents
1)
The specic heat is dened as
c = T
2 F
,
T 2
(10.11)
where
F = T N U ( 0 ) ,
(10.12)
as we had found in (8.44). So, by inserting 0 into U , we nd
T U 0
F T U ( 0 ) =
T U 0
1 A 2T
2
(Tc T ) 2
for SYM,
for SSB.
(10.13)
And by applying the denition of c, we arrive at
c reg
c
c reg
A 2T
with
c reg T 2
(3T 2Tc )
for SYM,
for SSB,
2U0
U0
+T
.
T 2
T
(10.14)
(10.15)
There appears no T Tc in the expression for c, so = 0. Or, in other words, c has no pole at
T = Tc . But there is a jump in specic heat at the phase transition:

A2Tc
c sym c ssb =
.
Tc
(10.16)
0 (Tc T ) /2
(10.17)
2)
From
1
we can read o = 12 .
3)
By combining (8.72) and (10.9) we nd
2U
!1

(T Tc ) 1 .

phi 0
(10.18)
So, it follows = 1.
159
4)
At T = Tc the eective potential is given by
U =
Moreover, we know that
4
.
8
U
( 0 ) = j .
(10.19)
(10.20)
So, we can immediately conclude that 0 j 1/3 and therefore = 3.

We see that these mean eld exponents are independent of the dimension d or the number of
components N (in case of O (N )-symmetry). Experimental results show that the actual values
of the critical exponents dier from the mean eld values and that they depend on d and N . But
otherwise they are universal! For example, one nds, even experimentally, the same exponents
for liquid-gas transitions and uniaxial magnets. Note that both are described by an Ising model
with N = 1.
So, we conclude that the critical exponents are independent of the precise details of a system
(e. g. the exact form of the equation of state for uids, i. e. b0 , b1 and so on). Furthermore, they
are the same for dierent types of systems, if these share the same dimension and symmetries
(d, N ). We say, they are in the same Ising universality class. So, for example, liquid-gas
transitions, uniaxial magnets and many other systems belong to the same class.
As these universality classes only depend on d and N , there are only few of them. But why?
Why are 2 parameters sucient to x four critical exponents? And why do they only depend on
d and N ? The understanding of this needs the profound concept of renormalization. Part of the
answer arises from the existence of additional scaling laws that relate the critical exponents,
e. g.
=
,
(10.21)
1
2 = + 2 + .
(10.22)
The concepts which are necessary to understand where the scaling relations come from and to
understand this universal behaviour of systems with equal symmetries, in general, have been
developed by Kadano, Wilson and Wegner. We will introduce these concepts in the following
sections.
10.1.6. Universal critical exponents

In Table 10.1 we have summarized the mean eld results of the previous subsection and compare
them to more sophisticated theoretical calculations as well as to experimental values. Note that
the results of the two-dimensional Ising model are exact. It is the only interacting model that
has been solved exactly, so far. This solution was found by Onsager in 1944.
160
Theoretical results:
Mean eld
0.5
Ising model in d = 2
0.125
Renormalization group in d = 3
n=1
n=3
0.11
0.1159
1
1.75
3
15
0.3249
0.3647
1.2371(4)
1.3866
4.817
4.802
Experimental results:
Liquid-gas / uniaxial magnet (n = 1)
0.1
0.34
1.2
4.4
Isotropic magnet (n = 3)
0.15
0.36
1.25
4.3
Table 10.1.: Universal critical exponents

10.2.1. Correlation length
The question we would now like to answer is: What does the density at one point r 1 know
about the density at another point r 2 , as the local density uctuates even in equilibrium? And
from which distance on are n(r 1 ) and n(r 2 ) uncorrelated, i. e. what is the correlation length of
the system? For magnetic systems we would ask the same questions for the magnetisation, but
we will stick to the uid notation in the following.
To answer these questions, let us look at the correlation function (2-point function):
Df
g f
gE
G (r 1 ,r 2 ) = n(r 1 ) n n(r 2 ) n
(10.23)
D
E
= n(r 1 ) n(r 2 ) n2 .
(10.24)
Note that, due to the translational and rotational symmetry of our system, G can only depend
on the distance between two points:
G (r 1 ,r 2 ) = G (|r 1 r 2 |) = G (|r |) = G (r ) ,
(10.25)
with r = r 1 r 2 . Often, the correlation function shows a very typical behaviour for large r :
r
G (r ) e
(10.26)
where is the correlation length.

G (r ) is the central quantity to describe the local distribution of uctuations regardless of
whether one deals with the amorphous structure of polymers or the galaxy distribution in the
universe, for example. Moreover, it has to play an important role in phase transitions. As we
have already seen, second order phase transitions correspond to a order-disorder transition, e. g.
T > Tc :
T < Tc :
M = 0,
|M | > 0
(10.27)
(10.28)
161
in magnetic systems at B = 0. To make this transition possible, the spins have to know of
each other, i. e. they have to be correlated. But not only that, the correlation length has to be a
function of temperature, too, to explain these dierent behaviours above and below the critical
temperature.
10.2.2. Correlation function and generating functional

Let us assume a very general form of a partition function for a magnetic system:
X
P
Z [j (r )] =
e H [ (r )]+ r (r ) j (r ) ,
(10.29)
where j (r ) are articial local sources. They resemble an arbitrary inhomogeneous magnetic
eld. Moreover, H [(r )] and Z [j (r )] are functionals, i. e. functions of functions. Note that they
are not just functions of some constant or j.
Now, the so-called generating functional is dened as
W = ln Z .
(10.30)
Its rst derivative with respect to j (r ) is the expectation value of :

P
W
1 X
=
(r ) e H + j
j (r ) Z
D
E
= (r ) ,
and its second derivative is exactly the 2-point correlation function:

"
# "X
#
P
P
2W
1 X
H + j
0 H + j
= 2
(r ) e
(r ) e
j (r ) j (r 0 )
Z
P
1 X
+
(r ) (r 0 ) e H + j
Z
D
E D
ED
E
= (r ) (r 0 ) (r ) (r )
Df
D
Eg f
D
Eg E
= (r ) (r )
(r 0 ) (r 0 )
=G (r ,r 0 ) .
(10.31)
(10.32)
(10.33)
(10.34)
(10.35)
(10.36)
If we now demand translational symmetry, it is

D E
(r = ,
(10.37)
G (r ,r ) = G (r r ) .
0
(10.38)
And if we further demand symmetry under spatial rotations, it is again

G (r ,r 0 ) = G (|r r 0 |) .
162
(10.39)
10.2.3. Correlation function and susceptibility

Now, let us look at what happens, if we replace j (r ) by j:
j (r )
j,
(10.40)
with j being independent of r . Then, the rst two derivatives of W with respect to j yield
E
W X D
=
(r )
j
r
= N ,
(10.41)
(10.42)
g f
gE
2W X Df
(r 0 )
=
(r
)
j 2
r ,r 0
X
=
G (r ,r 0 )
(10.43)
(10.44)
r ,r 0
=N
G (r , 0)
(10.45)
G (r ) ,
(10.46)
=N
X
r
where we used translational and rotational symmetry in the last two steps.
On the other hand, the susceptibility is dened as
j
1 W
=
j N j
1 2W
=
.
N j 2
(10.47)
(10.48)
(10.49)
So, the susceptibility and the correlation function are related via
=
G (r ) .
(10.50)
While we have considered a magnetic system, so far, these calculations can be performed
completely analogously for uids. In that case, one ends up with a relation between the
compressibility and the correlation function:
1
T = 2
nT

d3r G (r ) .
(10.51)
163
10.2.4. Critical exponent of correlation length and anomalous dimension

As we have already mentioned, at large r the correlation function behaves like a exponential
cuto:
r
lim G (r ) e .
(10.52)
r
So, using the relations between susceptibility / compressibility and the correlation function,
we can conclude that and T are nite as long as remains nite. But as we have seen in
section 8.4, the susceptibility diverges at the critical temperature (e. g. recall Figure 8.10). The
same holds for the compressibility. We can immediately infer that the correlation length has to
diverge at T = Tc , too:
T Tc
,T ,
T Tc
(10.53)
(10.54)
To describe this behaviour, we introduce another critical exponent :

|T Tc | .
(10.55)
In the mean eld approximation one nds = 12 . Moreover, the large r limit of G at T = Tc can,
in general, be expressed in the following way:
lim G (r ) r (d 2+) ,
(10.56)
with the so-called anomalous dimension being yet another critical exponent. Its mean eld
value is = 0.
10.2.5. Critical opalescence

Considering a uid, density uctuations will lead to uctuations in the refractive index. It
follows that light with a wavelength close to the correlation length will experience stronger
scattering. As we have seen that the correlation length becomes large in the vicinity of a second
order phase transition, it will eventually reach the wavelength of visible light, causing the
usually transparent uid to appear cloudy. This phenomenon is known as critical opalescence.
10.3. Universality
10.3.1. Macroscopic laws for second order phase transition
At second order phase transitions the correlation length diverges, leading to long-range phenomena. This is the reason for universality: the long-range behaviour does not depend on details of
the microscopic interactions, any more, except for very few properties, like the dimension, the
symmetries and the number of relevant degrees of freedom. These properties determine the
universality class of the system.
164
10.3. Universality
We have seen that the mean led theory is not capable of describing this long-range behaviour
properly. But nevertheless, exact statements are possible! One nds laws which are only loosely
connected to the microphysics and mainly arise from the statistics of many body systems. These
eective laws strongly depend on the relevant distance scale. For example, particle and atomic
physics are suitable to describe systems at very small scales, whereas these systems are better
described by the laws of chemistry and biology at way larger, macroscopic scales.
So, what we are looking for, is a framework to explain this qualitatively dierent behaviour on
dierent scales to describe the transition from microscopic laws on small scales to macroscopic
laws on large scales. This framework is the so-called renormalization group. It describes the
distance scale dependence of the eective laws and makes it possible to cast the qualitative
description of these central ideas into a mathematically rigorous form.
In the following subsections, we will introduce this concept via the block spin description,
developed by Kadano.
10.3.2. Ising model in one dimension

Let us rst look at the Ising model in one dimension. Using the abbreviation
J = K ,
(10.57)
its partition function takes the form

Z=
eK
i =1 s i s i +1
(10.58)
eKsi si +1 .
(10.59)
PN
{s i }
N
XY
{s i } i=1
Now, we introduce the concept of a block lattice. Therefore, we relabel the spins in the following
way:
i +1
for i odd (spins on the block lattice, block-spins),
(10.60)
2
i
s j0 : j =
for i even (intermediate spins).
(10.61)
2
It is illustrated in Figure 10.2.
The idea now is to integrate out the intermediate spins, to obtain a new Ising model for the
block spins sj . We achieve this by always grouping two subsequent factors of the product in
(10.59) into one block factor. This way, we arrive at a product of blocks:
sj :
j=
Z=
N/2
XY
eK sj s j eK sj+1 s j
0
(10.62)
{s i } j=1
XY
{s i }
(10.63)
eKs j (sj +s j+1 ) .
(10.64)
XY
{s i }
eK (sj s j +s j+1 s j )
0
165
2
1
j for s
j for s 0
Figure 10.2.: One-dimensional block lattice
Now, we can split the conguration sum, as s j0 only appears in block j:

!
XY X
0
Z=
eKs j (sj +s j+1 ) .
{sj }
Next, we perform the inner sum (integrate out the intermediate spins):

X Y
Z=
eK (sj +s j+1 ) + eK (sj +s j+1 ) .
{sj }
(10.65)
s j0 =1
(10.66)
This way, we arrive at a new Ising model for the block spins:
XY 0
Z=
eK sj sj +1 +C ,
{sj }
(10.67)
with a suitably dened K 0 and some constant C.

To determine K 0, we use the condition that for all possible spin congurations the two
expressions (10.66) and (10.67) for Z have to be identical:
eK (sj +s j+1 ) + eK (sj +s j+1 ) = eK sj sj+1 +C ,
0
(s j , sj+1 ) :
++ :
e2K + e2K
+ :
=e
e2K + e2K
Inverting (10.70) leads to

1
2
(10.69)
,
(10.70)
K 0 +C
(10.71)
=e
(10.68)
K 0 +C
=e
+ :
K 0 +C
=e
K 0 +C
= eK C ,
0
(10.72)
(10.73)
and multiplying this with (10.69) gives us

0
2K
2K
1
e
+
e
= e2K .
2
(10.74)
So, our new coupling K 0 is given by

K0 =
166
1
2
ln
f
1
2
e2K + e2K
g
1
2
ln cosh 2K .
(10.75)
10.3. Universality
The additive constant C is irrelevant and can be neglected. With that, (10.67) has exactly the
same form as (10.59). So, if the original lattice had lattice spacing a, the new partition function
is describing a lattice with lattice spacing 2a and a coupling K 0. But, of course, both expressions
of the partition function are the same we are still describing the same model! In other words,
(10.67) is describing the original model, just on a larger length scale. And on this length scale it
behaves like a model on a lattice with doubled lattice spacing and eective coupling K 0:
Renormalization: K K 0 ,
Lattice spacing:
(10.76)
a 2a .
Repeating the renormalization step (10.76) leads to a length scale dependent K (l ):

K
K (a)
K0
K (2a)
K 00
K (4a)
K 000
(10.77)
K (8a)
(10.78)
This procedure describes the renormalization group. It is a group, whose elements are all
possible sequences of renormalization steps. Actually, if you want to be mathematically more
rigorous, it is just a semigroup, as it does not contain the inverses of its elements.
Now, let us look at the dierent cases for K. If K > 0, then
so,
e2K =
1
2
e2K < e2K ,
(10.79)
(10.80)

e2K + e2K < e2K ,
and therefore,
0 < K0 < K ,
(10.81)
i. e. the coupling decreases as the length scale increases. So, the interaction goes to zero, if one
is going to larger and larger length scales:
l
K (l ) 0 .
(10.82)
This means that it is possible to treat this system perturbatively, as an expansion in K.

In the other case, K < 0, we speak of an antiferromagnet. But then (10.75) tells us that K 0 > 0,
nevertheless. So, from there on it just behaves the same as the K > 0 case.
10.3.3. Ising model in d > 1 dimensions

New interactions
If we now consider a lattice in d > 1 dimensions, integrating out the microscopic modes
(the intermediate spins) leads to way more complicated interactions, as we have sketched in
Figure 10.3 for d = 2. And performing further renormalization steps will lead to even more and
more complicated interactions. So, in the following we will have to nd a systematic way to do
so.
167
Figure 10.3.: Complicated interactions in two-dimensional block lattice. The symbol denotes block
spins in the rst renormalization step. Links denote eective interactions. In addition to
next-neighbour interactions one now also nds diagonal interactions. A second renormalization step leads to the block spins . This induces even more complicated interactions, as
demonstrated by the dashed link.
2a
Figure 10.4.: Block spins in two-dimensional lattice
Block spins
Let us generalize the concept of block spins, that we have introduced in the one-dimensional
case. A block will now consist out of 2d spins, as illustrated in Figure 10.4 for d = 2. We dene
the spin of the k-th block as
X
si
sk =
(10.83)
,
2d
i block(k )
where we have chosen an arbitrary normalisation. Now, these block spins represent our new
degrees of freedom, i. e. we have reduced the number of degrees of freedom.
The partition function is given by
Z=
e J
<i j > s i s j
(10.84)
{s i =1}
X
{s i =1}
168
e H0 [si ] .
(10.85)
10.3. Universality
Due to (10.83) we can write

X

d
dsk sk 2 /2
si = 1 .
(10.86)
block
For each block we can now insert such a 1 into the partition function:
Z=
dsk
k block lattice
X

d
sk 2 /2
si e H0 [si ]
{s i =1}
original lattice
|
=
Y
(10.87)
block
{z
e H1 [sk ]
dsk e H1 [sk ] .
(10.88)
As in the one-dimensional case, we end up with an equivalent partition function, formulated on

a block lattice.
Renormalization flow of the coupling
In the same way, as we have done it in the one-dimensional Ising model, we can repeat this
blocking procedure, i. e. this renormalization step:
H0
H1
H2
H3
(10.89)
This way, we end up with a so-called renormalization ow J (L) of the coupling, where L = 2n a.
Moreover, new couplings are generated by this procedure.
New variables
P
The sum of the spins in a block k, i block(k ) si , can take values from 2d to 2d . For our
normalisation (10.83) the spacing between the values sk amounts to 2d/2 , instead of 1 for the
original spins. The range of sk is given by
sk [2 /2 , 2 /2 ] ,
d
(10.90)
instead of si [1, 1] for the original spins. Thus, both range and spacing of the block variable
is dierent from the original variable.
After many renormalization steps we can replace our block spin by a continuous function :
sk
(x ) ,
(10.91)
where < (x ) < .
169
Potential
The interactions between the spins of one given block spin are described by the potential U (sk ).
They are called ultra local or on-site interactions. Due to the symmetry under sk sk it
can only depend on the square of the spin:
U (sk ) = U (sk2 ) .
(10.92)
For the microscopic Ising spins there is only one possible value si2 = 1, so there are no on-site
interactions; i. e. microscopic spins si do not interact with themselves. For block spins, on the
other hand, U (sk2 ) is not trivial any more, as sk2 can take values between 0 and 2d . And in the
continuum limit, U [(x )] is a function of a continuous variable .
Let us denote the potential at step n of the blocking procedure, i. e. at eective lattice spacing
l = 2n a, by U [; l]. So, the potential changes as a function of length scale. Now, the Hamiltonian
at length scale l can be expressed as a potential for the on-site interactions plus additional terms
H :
X
H [(x ); l] =
U [(x ); l] + H .
(10.93)
x
Here, H vanishes if all (x ) are equal, i. e. if we are considering a homogeneous eld (x ) = .

Let us assume
H 0 .
(10.94)
Then, H [(x ); l] is minimal for a homogeneous eld 0 , with 0 being the minimum of U [(x ); l].
In general, this minimum changes due to uctuations. So it is length-scale dependent: 0 (l ).
10.3.4. Renormalization flow

Principle of the renormalization flow
H [(x ); l] describes a family of lattice models with dierent l. But they all correspond to the
same microscopic model. So, we can conclude that the physical laws take dierent forms on
dierent length scales l. For example, the form of the on-site interaction potential U might
depend on l in the way sketched in Figure 10.5. Its interpretation would be that small domains
favour magnetisation (non-zero position of the minimum of U ), whereas they get averaged out
for large domains.
The general principle behind this is, that for H [; l] the uctuations with wavelength smaller
than l are averaged / integrated out. On the other hand, uctuations with wavelength larger
than l remain to be done.
Momentum space
All of this can be formulated more elegantly in momentum space. The momentum scale,
belonging to the length scale l, is
k .
(10.95)
l
So, uctuations with momenta q 2 > k 2 are integrated out.
170
10.3. Universality
(a) l = 100 a
(b) l = 109 a
Figure 10.5.: Example of an potential on dierent length scales
In momentum space, Hamiltonian and potential are denoted by H [; k] and U [; k]. The
microscopic limit of the Hamiltonian is given by
H [; k = /a ] = H 0 [] ,
(10.96)
and its k 0 limit can be written as

H [; k = 0] = NU [] + H .
(10.97)
In the latter limit, all uctuations are integrated out. The potential in this limit is called the
eective potential:
U [] = U [; k 0] .
(10.98)
Flow equation
The central equation of the renormalization group is the ow equation:
U [; k]
= ... ,
ln k
(10.99)
where the r. h. s. still remains to be computed. This equation describes the change of the potential
U when varying the momentum scale. To extract the free energy for constant (magnetisation),
it is sucient to solve this equation in the k 0 limit. Equivalently, we can consider this
equation in real space:
U [; l]
= ... ,
(10.100)
l
and solve it in the l limit.
As a simplication, we restrict the potential to a quartic polynomial at every scale l:
U [; l] = 12 l ( 0,l ) 2 ,
(10.101)
= 12 2 .
(10.102)
with
171
Figure 10.6.: Flow of the potential towards larger l
Here, l and 0,l depend on l and 0,l is the minimum of U [; l].

Now we want to set up ow equations for 0,l and l . Assuming a 2 l2 (l 0,l ) 1 as an
approximation, one nds in three dimensions
3c 1 1
0 = 2 2 ,
l
2 l
9c 2
= 2 2 ,
l
4
(10.103)
(10.104)
with some constants c 1 , c 2 of order O(1). (This result requires a computation not covered by this
lecture.) We can directly infer that 0 and decrease for increasing l. The qualitative change of
U due to that has been sketched in Figure 10.6.
Solution of flow equation
Let us rst consider the running minimum. The equation we have to solve, is
3c 1 dl
22 l 2
3c 1
= 2 d(l 1 ) .
2
d 0 =
(10.105)
(10.106)
The solution to this dierential equation is given by

!
3c 1 1 1
0 = 0 + 2
,
2 l l 0
(10.107)
with initial value 0 = 0 (l 0 ), which depends on T and J .

Let us examine its large length scale behaviour:
l
0 (l ) 0
3c 1
= 0 0,c .
22l 0
So, we can distinguish three dierent cases:

0 > 0,c : In this case, 0 (l ) > 0. So, it describes the ordered phase with SSB.
172
(10.108)
10.3. Universality
0 < 0,c : In this case, 0 (l ) reaches 0 before l . So, it describes the disordered phase,
where the symmetry is unbroken (SYM).
0 = 0,c : At this point, the phase transition occurs! The corresponding critical temperature
can be deduced from 0 (Tc ) = 0,c .
Now, let us consider the running coupling. Its dierential equation is given by
d
9c 2
= d(1 ) = 2 dl .
2
(10.109)
Integrating this, yields
1
1
9c 2
= 2 (l l 0 ) ,
0 4
where 0 = (l 0 ). Solving this for gives us
!1
1
9c 2
=
+
(l l 0 )
0 42
!1
9c 2 0
= 0 1 +
(l l 0 ) .
42
(10.110)
(10.111)
(10.112)
In the l limit, this becomes

=
42 1
.
9c 2 l
(10.113)
Note that this is independent of the initial value! This is an example of how the memory of the
microscopic physics is lost at macroscopic scales!
10.3.5. Fixed points

Now, we want to look for xed points of the system, i. e. points, where the ow of and 0
vanishes. Actually, we want to look for xed points of the dimensionless quantities
= l ,
(10.114)
= 0l ,
(10.115)
with respect to the dimensionless scale

t = ln
l
.
l0
(10.116)
This is necessary to nd truly scale-independent xed points of the system!

For we nd
d
d
= l ( 0l )
dt
dl
d 0
= l 2
.
dl
(10.117)
(10.118)
173
(b) Flow of
(a) Flow of
Figure 10.7.: Flow of the dimensionless quantities towards larger l
So, using (10.107), we arrive at

= t = +
3c 1
.
22
(10.119)
Similarly, we compute the t-derivative of :

d
d
= l 2
,
dt
dl
(10.120)
and, using (10.112), nd

= t = +
9c 2 2
.
42
(10.121)
Here, we adopted the usual QFT notation of -functions. Note that no mass or length scale
appears in these ow equations!
We have sketched the ow of in Figure 10.7a. Thereby, the arrows indicate the ow towards
larger l, i. e. smaller t. We see that, independent of the starting point, ows to the xed point
for l . From
( ) = 0
(10.122)
it follows
4
=
.
9c 2
(10.123)
So, we can conrm the large length scale behaviour found in (10.113):
l

4
=
.
l
9c 2l
(10.124)
This ow of the eective couplings towards an infrared xed point is related to the observed
loss of memory of the microscopic physics. The details of the microphysics determine the starting
point of the ow of the coupling. But as we have seen, for larger (macroscopic) lenght scales the
174
10.3. Universality
coupling ows towards this xed point independent of the specic starting point! This is also
the deep origin of the universality of the critical exponents. They are related to the behaviour
of the -functions around the xed point. This behaviour is determined by the form of the
potential U . And this, in turn, is determined by the dimension and the symmetries of our system.
Universality follows from the ow!
Contrary to the coupling, the ow of , as sketched in Figure 10.7b, shows the existence of an
ultraviolet xed point . Every other starting point than = leads to a divergent ow for
l . So, the critical point is described by = :
=
T = Tc .
(10.125)
That is because the critical point is characterized by 0 = l 0 for l . And this is only
possible for = .
So, we see that at the critical point all quantities are on a xed point. Therefore, our system
is scale-invariant at the critical point! This behaviour is characteristic for second order phase
transitions. But note that this needs some tuning for , in contrast to the infrared xed point
for .
175

Theoretical Statistical Physics: Prof. Dr. Christof Wetterich Institute For Theoretical Physics Heidelberg University

Uploaded by

Copyright:

Available Formats

Theoretical Statistical Physics: Prof. Dr. Christof Wetterich Institute For Theoretical Physics Heidelberg University

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Theoretical Statistical Physics: Prof. Dr. Christof Wetterich Institute For Theoretical Physics Heidelberg University

Uploaded by

Copyright:

Available Formats

Theoretical Statistical Physics

Prof. Dr. Christof Wetterich

Last update: March 25, 2014

Foundations of Statistical Physics

2. Basic concepts of statistical physics

2.9. Probability density of energy . . . . . . . . . . . . . . . . . . .

4.2.8. Thermal equilibrium . . . . . . . . .

II. Statistical Systems

6. Ideal quantum gas

6.5. Degenerate Fermi gas . . . . . . . . . . . . . . . . . . . . . . . . . . .

7.6. Equipartition theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10. Critical Phenomena

10.1.5. Mean eld exponents . . . . . . . . . . . . . . . . . . . . . . . . .

Foundations of Statistical Physics

1.2. Macroscopic laws and simplicity

1.3. Fundamental laws are of probabilistic nature

1.3. Fundamental laws are of probabilistic nature

Figure 1.1.: The two major strategies in physics

2. Basic concepts of statistical physics

2.1.2. Probability distribution

The probabilities p have to obey two conditions

The range of p is therefore

2. Basic concepts of statistical physics

Table 2.1.: A system with A( ) =

And a function of A is an observable, too:

2.1.4. Expectation value

(A2 2 A hAi + hAi2 )

2.1.6. Expectation values and probability distribution

2.2. Simple examples

2.2. Simple examples

The number of particles is given by N = 12 (A + 1), which corresponds to N 1 = 1 and N 2 = 0.

Their dispersions are easily calculated. For A we rst need

where we used A21 = A22 = 1. Then,

And for N accordingly:

2. Basic concepts of statistical physics

Table 2.2.: States in a four-state system

Table 2.3.: Observables in a four-state system

2.2.2. Four-state system

and the dispersions:

2.3. Reduced system or subsystem

2.3. Reduced system or subsystem

The probabilities for these states are

2. Basic concepts of statistical physics

Table 2.4.: States of the reduced ensemble in example 1

2.4. Probabilistic observables

is the mean value in state . We use normalized probability distributions:

2.4. Probabilistic observables

Dene probabilities for states of reduced system:

and probabilities to nd n in a state :

One nds (n2 ) , (n ) 2 due to uctuations of n in state . Therefore,

2. Basic concepts of statistical physics

For this subsystem the rst spin B is a probabilistic observable. As B1 = 1 and B3 = 1, B

The mean value of B in state = 2 is

Let us check this:

which diers from (B2 ) 2 ! Thus,

This is a clear sign of a probabilistic observable. And probabilistic observables need

2.5. Quantum statistics and density matrix

Figure 2.1.: Probability distribution of the probabilistic variable B in given state

Table 2.5.: Observables in the reduced system in example 2