Equipartition Theorem

Motivation

Consider a system in thermal equilibrium with the constraint

ipiEi=C,

the energy distribution is given by the Boltzmann distribution

pi=eβEij=1neβEj

Defining

Z(β)=j=1neβEj,

then the expected energy is

E(β)=i=1nEieβEiZ(β)=βlnZ(β).

You then determine β by solving

βlnZ(β)=C.

In most cases this equation

C=iEieβEiieβEi

has a unique solution β=β(C) (because E(β) is a strictly decreasing function of β), so the temperature T=1/(kβ) is completely fixed by the desired E=C. Thus there is indeed a one‐to‐one correspondence between the coolness β (hence the temperature) and the imposed mean energy C.

A particularly simple correspondence happens for special cases:

Equipartition theorem

If a system at temperature T has a Hamiltonian in which each independent degree of freedom appears only quadratically,

H=i=1N12aixi2,

then at thermal equilibrium the average energy per quadratic degree of freedom is

12aixi2=12kT,

and hence

E=12NkT.

Clarification on degrees of freedom

H(q,p)=p22m+12kq2

is a quadratic form on R2, so equipartition counts two quadratic terms and gives

H=2×12kT=kT.

Applicability

Examples

Limitations