Machine Learning Forcefield for Silicate Glasses

TL;DR

This paper introduces a novel, efficient methodology for developing accurate interatomic forcefields for silicate glasses using ab initio data, Gaussian process regression, and Bayesian optimization, demonstrated on silica.

Contribution

A new parameterization approach combining ab initio simulations, Gaussian process regression, and Bayesian optimization for silicate glasses.

Findings

Produces a highly accurate forcefield for silica

Reduces need for intuition in parameterization

Offers a transferable method for silicate glasses

Abstract

Developing accurate, transferable, and computationally-efficient interatomic forcefields is key to facilitate the modeling of silicate glasses. However, the high number of forcefield parameters that need to be optimized render traditional parameterization methods poorly efficient or potentially subject to bias. Here, we present a new forcefield parameterization methodology based on ab initio molecular dynamics simulations, Gaussian process regression, and Bayesian optimization. By taking the example of glassy silica, we show that our methodology yields a new interatomic forcefield that offers an unprecedented description of the atomic structure of silica. This methodology offers a new route to efficiently parameterize new empirical interatomic forcefields for silicate glasses with very limited need for intuition.