Optimising molecular dynamic potentials to dynamical structure factor data
Anders J. Markvardsen, RL McGreevy · Science and Technology Facilities Council · 2017
This paper reports an algorithm and software for producing dynamical models of materials at the atomic scale using classical molecular dynamics simulation, but simultaneously refining the potential energy parameters based on experimentally determined dynamical and structural information such as the dynamical structure factor S(Q, w). The software is called MDMC (Molecular Dynamics - Monte Carlo) (https://github.com/MDMCproject) and the first pre-release version used in this work is https://doi.org/10.5281/zenodo.1068365. The algorithm follows from two previous very successful algorithms Reverse Monte Carlo (RMC) (McGreevy & Pusztai, 1988)(McGreevy, 2001) and Empirical Potential Structure Refinement (EPSR) (Soper, 1996) (Soper, 2007), though these are only used to produce structural models based on structural data. In this paper the algorithm is presented together with initial results for a model system, liquid argon, based on the data of (van Well, Verkerk, de Graaf, Suck, & Copley, 1985).