PHY 402

STATISTICAL PHYSICS
Statistical Physics:Introduction

I Statistical physics is the study of the universal statistical and

probabilistic law which govern the microscopic behaviour and
properties of systems composed of a large number of particles,
N of the order of 1023 .

I It is a microscopic theory of thermal Physics that describes

macroscopic quantities from taking mean values of microscopic
quantities e.g temperature (macroscopic quantity) is given by
the mean kinetic energy of particles (microscopic quantity).

I It has applications in different fields including biology, chem-

istry, neurology and even in social sciences.
Statistical Physics: Basic Concepts

I Degree of freedom is the number of spatial coordinate that

could be used to describe the position of the constituent of the
system. N-particle system has N degree of freedom of position
coordinates qi (i =1, 2, ..., N).
I State of a system - The classical state of a system is given
by specifying the state ψi = (qi , pi ) of each particle i in the
system; where pi is the momentum of the ith particle.
I The Microscopic ψ = {ψi : i = 1, 2, ..., N} of statistical physics
is called a microstate while the macroscopic state Ψ = (S, V,
n) of thermal physics is called a macrostate.
Statistical Physics: Basic Concepts

I For a N-particle system, phase space is a 6N dimensional

space. Each particle is described in 6 dimensions (3 spatial
coordinates and 3 momenta). A point in this space is specified
by giving a particular set of values for the 6N coordinates and
momenta. This point can be denoted as:

rp = (r1 , ..., rN , p1 , ..., PN )

Each point represent the state of the system

I A phase trajectory is a curve, in phase space, which consists
of the states through through which a system evolves in a given
time interval.
Statistical Physics: Basic Concepts
I The probability density ρ(p, q) of a system in phase space
is the probability, dP, of finding the system in an infinitesimal
volume, dV, of phase space, where

ρ(p, q) = ρ(p1 , p2 , ..., pN , q1 , q2 , ..., qN )

dV = dpd = dp1 dp2 ...dpN dq1 dq2 ...dqN
dP = ρdpdq (1)
dP is a probability distribution. The total probability of finding
the system somewhere in the phase space is unity.
Z Z
dP = ρdpdq = 1 (2)
phase space phase space
I ρ(p, q) which is also known as the statistical distribution func-
tion, is the fundamental problem to be determined in statistical
physics. Once ρ(p, q) is known for any system, then, its micro-
scopic properties are known.
Statistical Physics: Average values of physical quantities
I Knowledge of ρ allows us to calculate the average, that is the
macroscopic (or thermodynamic) value of any physical quantity
f through
Z
¯
hf i = f = f (p, q)ρ(p, q)dpdq (3)
phase space

This average is known as statistical or ensemble average.

The time average, however, is expressed as:
Z T
hf i = f¯ = lim f (t)dt (4)
T →∞ 0

where f (t) is the actual value of the physical quantity mea-

sured at time t. T is the total time during which the system
is monitored. Calculation of the average quantities is over a
continuum state.
Statistical Physics: Average values of physical quantities
I For a system which evolves only through a discrete set of
states equation (3) becomes:
X
hf i = f¯ = fi Pi (5)
allstatesi

where fi is the value of the quantity when the system is in

state i and Pi is the probability of the state i.
The normalizing Pi :
X
Pi = 1 (6)
all states i
Equation (4) becomes:
1 X
hf i = f¯ = fi Pi (7)
T
all states i

where T is the number of values fi .

Statistical Physics: Average values of physical quantities

I ρ12 for composite of two subsystems is given by

ρ12 = ρ1 ρ2 (8)
that is,
hf1 f2 i = hf1 ihf2 i (9)
Statistical Physics: Thermodynamic limit and fluctuation
in calculated values

I A fundamental result of statistical physics is that the

fluctuation or variation (∆f = f − hf i) in the values of f
relative to the mean value hf i is proportional to N1 .
The fundamental statistical physics result is expressible as:
p
h(∆f )2 i 1
∝
hf i N

This means that in thermodynamic limit(N → ∞), these

fluctuations tend to zero. In fact, if fluctuations lower than
0.01hf i are considered irrelevant, then statistical physics
theory will give accurate predictions of the properties of
system sizes of about 10,000 particles and above.
Statistical Physics: Thermodynamic limit and fluctuation
in calculated values

I Note that the root mean square fluctuation or deviation

h(∆f )2 i is used instead of actual deviation h∆f i
I Try this: Show that

h(∆f )2 i = hf 2 i − (hf i)2

Statistical Physics: Liouville’s Theorem

Liouville’s theorem states that statistical distribution function

ρ(p, q) is constant along the phase trajectories of the system. In
other words, the probability density of the states of a system in
phase space is constant.
Statistical Physics: Statistical definition of Entropy

I In thermal physics, the entropy of a system may change

without a change in its heat content. Such a change in
entropy can be accounted for in statistical physics by a
change in the number of accessible microstates of the system.

S ∝ ∆Ω (10)

I The statistical physics definition of entropy (in dimensionless

units) is
S = ln∆Ω (11)
The number of microstates ∆Ω ≥ 1, meaning that S ≥ 0,
that is entropy cannot be negative.
Statistical Physics: Statistical definition of Entropy
I The entropy S = 0 corresponds to a system with only one
state ∆Ω = 1. According to the third law of thermal physics,
we can only have the occurrence ∆Ω = 1 at absolute zero.
An example of such a system is an ideal (perfect) crystal at
T=0K.
I To express entropy in energy units, we recall from
thermodynamics that

dE = TdS − pdV + µdn (12)

It implies S must be in units of JK −1 which is also the unit

of the Boltzmann’s constant kB . Thus,

S = kB ln∆Ω (13)

This equation in the unit of energy/Kelvin is known as the

Boltzmann’s equation.
Statistical Physics: Statistical definition of Entropy

I Given two subsystem 1 and 2 with number of microstates

∆Ω1 and ∆Ω2 , respectively, the total number of microstates

∆Ω = ∆Ω1 ∆Ω2 (14)

This implies

S = kB ln(∆Ω1 ∆Ω2 )
S = kB ln∆Ω1 + kB ln∆Ω2

S = S1 + S2 (15)
That is entropy is additive and is extensive variable.