A partition function estimator

Abstract

We propose an estimator that allows us to calculate the value of a simple system’s partition function using finite sampling. The core idea is to neglect the contribution from high energy microstates, which are difficult to be sampled properly, and then calculate a volume correction term to compensate for this. As a proof of concept, the estimator is applied to calculate the partition function for several model systems, ranging from a simple harmonic oscillator to a Lennard-Jones fluid with hundreds of particles. Our results agree well with the numerically exact solutions or reference data, demonstrating that efficiently estimating partition functions for the studied example cases is possible and computationally affordable.