Origins of Universal Machine Learning Force-Field Errors in Multicomponent Materials

2026-10-07

AI Summary

This study addresses the insufficient generalization evaluation of universal machine learning force fields for multicomponent materials by constructing a large-scale multicomponent configuration benchmark dataset based on density functional theory and spherical harmonics representations. It systematically reveals how training coverage, geometric heterogeneity, and elemental response disparities influence force prediction errors, identifying for the first time that compression-side force errors significantly exceed those on the tension side. Furthermore, this work establishes an intrinsic correlation between electronic band responses and model fitting difficulty. By optimizing the spherical harmonic degree, prediction errors for most elements are effectively reduced, providing critical guidance for force field architecture selection and data sampling strategies.

Abstract

Universal machine learning force-field generalization to multicomponent environments generated by compositional design remains insufficiently assessed. We construct a benchmark of 7,599 multicomponent configurations inspired by high-entropy design, elemental substitution and anion mixing. Eleven pretrained models are evaluated against density functional theory for energies, forces and stresses, with assessment extended to elastic, vibrational and adsorption-related properties. Force errors are analysed through training-reference coverage, local geometric heterogeneity, distance directionality and elemental response. Distances to training-reference environments reveal a qualitative association between coverage differences and increasing errors, while substantial variation remains at similar distances. Higher-error groups show greater local geometric heterogeneity, although OMat24 provides broad coverage of these environments. Relative to training-reference pair medians, errors remain low near the median, rise steeply on the compression side and increase more weakly on the extension side. After matching element pairs and absolute distance deviations, compression-side force errors are 1.81-1.95 times extension-side errors. Model-predicted pairwise interaction curves show greater curvature under compression. Fitting difficulty in independent elemental systems correlates with electronic band-energy responses to atomic displacements and Fermi-level shifts, and a similar pattern is observed in multicomponent systems. In parameter-matched comparisons, spherical-harmonic representations with maximum degrees of 2 and 4 lower test force errors for 38 and 40 of 43 elements, respectively, while differences in elemental difficulty remain. These findings inform force-field selection for experimental compositional design and identify targets for training-data sampling and model representations.

Problem

• universal machine learning force-field
• multicomponent materials
• generalization error
• compositional design

Innovation

• Universal Machine Learning Force-Fields
• Multicomponent Materials
• Error Analysis
• Local Geometric Heterogeneity
• Spherical Harmonic Representations