I see that the partition function is associated to the way particles are "partitioned" among energy levels. The equipartition theorem divides (partitions) average energy among degrees of freedom allowed to the particles.

Equating the average energy based on the partition function, given by $$\langle E\rangle = -\frac{\partial \ln Z}{\partial \beta}\,,$$ with the average energy of the system given by the equipartition theorem, given by $\langle E\rangle = (f/2)kT$, where $f$ is the total number of degrees of freedom in the system, $\beta = 1/kT$, and $\langle E\rangle$ is the average energy of the system, we get $$ -\frac{\partial \ln Z}{\partial \beta} = \frac{f}{2}kT\,. $$ Simplifying, we get $$ \ln Z = -\frac{f}{2}\ln(kT) + const.\,, $$ or, $$ Z = (kT)^{-f/2} \times const. $$ So, is it correct to say that:

  1. the partition function $Z$ shows how the average energy is partitioned among the degrees of freedom at a particular temperature OR
  2. the "partitions" of energy associated with the partition function $Z$ are actually the degrees of freedom allowed to the particles at a particular temperature.

1 Answer 1


The partition function is related to the equipartition theorem (the theorem is derived from that function) but the word "partition" refers to a different thing in the two cases. Here is a general way to think of the partition function in any ensemble.

If we assign a "multiplicity" $A_i$ to microstate $i$, then the probability of microstate is $$ p_i = \frac{A_i}{\sum_i A_i} = \frac{A_i}{\text{partition function}} $$ The denominator is the partition function of the ensemble and the summation is over all microstates that are compatible with the macroscopic state, whether this is $E,V,N$ (microcanonical), $T$, $V$, $N$ (canonical) or other.

In the canonical ensemble we have $A_i = e^{-\beta E_i}$, where $E_i$ is the energy of the microstate. It is the special form of the energy and its dependence on the degrees of freedom that leads to equipartition of energy. More specifically, equipartition is achieved in the limit $1/\beta = k T \to \infty$

So, while the partition function and the equipartition theorem are related to each other, the partition function is a more general tool that describes the probability of microstate, while the equipartition theorem is a result that is based on the partition function in the limit of high temperature.

  • $\begingroup$ This makes sense, thank you. On the other hand, do the derivation of the partition function Z in terms of degrees of freedom (shown in the description) and the inferences derived from it seem valid, in the equipartition limit 1/β=kT→∞ like you mentioned? $\endgroup$
    – Walrus
    Apr 6, 2023 at 16:53
  • $\begingroup$ @Walrus Technically you are correct but only in the limit $T\to\infty$. The whole point of the partition function is to express it as a function of the microscopic state: $Z = \frac{1}{h^{3N}N!}\int e^{-\beta E} d r_i^{3N} d p_i^{3N}$, with $E=\sum_i \frac{p_i^2}{2m} + V(r_1,\cdots r_N)$. This function conveys everything we can know about a system composed of $N$ interacting particles. $\endgroup$
    – Themis
    Apr 6, 2023 at 19:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.