Sparse Gaussian Process Potentials: Application to Lithium Diffusivity in Superionic Conducting Solid Electrolytes

TL;DR

This paper introduces a scalable sparse Gaussian process regression method for machine learning interatomic potentials, demonstrating its effectiveness in simulating lithium diffusivity and discovering new phases in solid electrolytes.

Contribution

A novel sparse Gaussian process approach with adaptive sampling for efficient and accurate interatomic potential modeling in materials science.

Findings

Successfully reproduces melting and glass-crystallization temperatures of Li7P3S11

Simulates lithium diffusivity accurately in solid electrolytes

Identifies a new phase with lower lithium diffusivity

Abstract

For machine learning of interatomic potentials a scalable sparse Gaussian process regression formalism is introduced with a data-efficient on-the-fly adaptive sampling algorithm. With this approach, the computational cost is effectively reduced to those of the Bayesian linear regression methods whilst maintaining the appealing characteristics of the exact Gaussian process regression. As a showcase, experimental melting and glass-crystallization temperatures are reproduced for Li7P3S11, Li diffusivity is simulated, and an unchartered phase is revealed with much lower Li diffusivity which should be circumvented.