Improving the accuracy of the neuroevolution machine learning potential for multi-component systems

TL;DR

This paper enhances the neuroevolution potential framework by introducing an optimized descriptor for multi-component systems, significantly improving regression accuracy without added computational cost.

Contribution

It proposes a new atom-environment descriptor with optimized factors for multi-component systems, improving accuracy in machine-learning potentials.

Findings

Improved regression accuracy for multi-component systems.

No increase in computational cost during molecular dynamics simulations.

Enhanced neural network training via optimized descriptors.

Abstract

In a previous paper [Fan Z \textit{et al}. 2021 Phys. Rev. B, \textbf{104}, 104309], we developed the neuroevolution potential (NEP), a framework of training neural network based machine-learning potentials using a natural evolution strategy and performing molecular dynamics (MD) simulations using the trained potentials. The atom-environment descriptor in NEP was constructed based on a set of radial and angular functions. For multi-component systems, all the radial functions between two atoms are multiplied by some fixed factors that depend on the types of the two atoms only. In this paper, we introduce an improved descriptor for multi-component systems, in which different radial functions are multiplied by different factors that are also optimized during the training process, and show that it can significantly improve the regression accuracy without increasing the computational cost in MD simulations.