How to validate machine-learned interatomic potentials

TL;DR

This paper reviews the principles and best practices for validating machine-learned interatomic potentials, emphasizing both numerical accuracy and physical consistency to ensure reliable atomistic simulations.

Contribution

It provides a comprehensive overview of validation methods and offers practical recommendations for assessing ML potentials in materials modeling.

Findings

Guidelines for error metric selection

Importance of physical validation methods

Recommendations for community best practices

Abstract

Machine learning (ML) approaches enable large-scale atomistic simulations with near-quantum-mechanical accuracy. With the growing availability of these methods there arises a need for careful validation, particularly for physically agnostic models - that is, for potentials which extract the nature of atomic interactions from reference data. Here, we review the basic principles behind ML potentials and their validation for atomic-scale materials modeling. We discuss best practice in defining error metrics based on numerical performance as well as physically guided validation. We give specific recommendations that we hope will be useful for the wider community, including those researchers who intend to use ML potentials for materials "off the shelf".