An automated framework for exploring and learning potential-energy surfaces

TL;DR

This paper presents autoplex, an automated framework that streamlines the exploration and fitting of potential-energy surfaces to facilitate machine learning in atomistic simulations, reducing manual effort and increasing efficiency.

Contribution

The introduction of autoplex, an open-source automated software for exploring and fitting potential-energy surfaces, enhancing the efficiency of machine learning in materials modeling.

Findings

Demonstrated capability on titanium-oxygen, SiO2, water, and phase-change materials.

Showed automation accelerates the development of machine-learned interatomic potentials.

Illustrated interoperability with existing computational software.

Abstract

Machine learning has become ubiquitous in materials modelling and now routinely enables large-scale atomistic simulations with quantum-mechanical accuracy. However, developing machine-learned interatomic potentials requires high-quality training data, and the manual generation and curation of such data can be a major bottleneck. Here, we introduce an automated framework for the exploration and fitting of potential-energy surfaces, implemented in an openly available software package that we call autoplex (`automatic potential-landscape explorer'). We discuss design choices, particularly the interoperability with existing software architectures, and the ability for the end user to easily use the computational workflows provided. We show wide-ranging capability demonstrations: for the titanium-oxygen system, SiO2, crystalline and liquid water, as well as phase-change memory materials. More generally, our study illustrates how automation can speed up atomistic machine learning -- with a long-term vision of making it a genuine mainstream tool in physics, chemistry, and materials science.