Development of a Software to Optimize Parameters of Interatomic Potentials for Solid Systems

Abstract

Generally, it is difficult to develop practical interatomic potentials which can be used in classical molecular dynamics calculations, since a method for development had not been established. In order to solve such problems, we had proposed a framework to develop interatomic potentials. However, the framework had not been widely used due to the difficulty of coding a process to optimize potential-parameters, which was involved in the framework. In this work, a practical software to optimize potential-parameters for solid systems, was developed based on real-coded genetic algorithm (GA). Fitness (=optimization-function) in GA for potential-parameter optimization could be calculated in practical time, and a concept of the corresponded module in GA-code was proposed. The developed software has an extensible structure: a user can define a new crystal-structure, or a new potential function to use in potential-parameter optimization by defining a corresponded class for that in separated header files. Convenient interfacial class to calculate equilibrium material properties of crystals and that to calculate material properties of user-inputted atomic-structures were prepared to define optimization-function. Several sample input files and header files were available for easy starting. As a result of distribution, several effective interatomic potentials were developed by using the software. As a sample of the development of interatomic potential, Tersoff-type potential for SiO2 systems were shown.