Nested Sampling of Materials' Potential Energy Surfaces: Case Study of Zirconium

The nested sampling (NS) method was originally proposed by John Skilling
to calculate the evidence in Bayesian inference. The method has since been utilised in various research fields, and here we focus on how NS has been adapted to sample the Potential Energy Surface (PES) of atomistic systems, enabling the straightforward estimation of the partition function.
Using two interatomic potential models of zirconium, we demonstrate the workflow and advantages of using nested sampling to calculate pressure-temperature phase diagrams.
Without any prior knowledge of the stable phases or the phase transitions, we are able to identify the melting line, as well as the transition between the body-centred-cubic and hexagonal-close-packed structures.
Supporting data for Proceedings paper of MaxEnt Conference 2022.