Interaction embedding via factorized linear models for multi-element machine learning interatomic potentials
The Journal of Chemical Physics · 07.11.2026 · 165 (17)
DOI: 10.1063/5.0345011 ↗Kurz gefasst
Fachartikel in der Zeitschrift The Journal of Chemical Physics, online erschienen am 05.11.2026. Verfasst von Sung Hoon Jung und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Chemistry, University of Texas at Austin, Austin, Texas 78712, USA. Die Arbeit nennt 59 Literaturangaben. Gefördert von: Welch Foundation.
Abstract (Originalsprache)
Linear machine-learning interatomic potentials, such as the Ultra-Fast Force Field (UF3), deliver near-density functional theory accuracy at low computational cost. Their number of parameters, however, grows combinatorially with the number of chemical species, limiting their use on multi-element systems. We address this scaling problem with pseudo-interaction UF3 (PI-UF3), a factorized linear model in which the Z(k) real k-body potentials are expressed as linear combinations of a small number P(k) ≪ Z(k) of shared pseudo-interactions. Learned element embeddings are standard in nonlinear potentials; inside a linear model, the embedding becomes a rank constraint on the coefficient matrix of each body order, and the resulting bilinear problem is trained with alternating least squares, where each block update is one closed-form linear solve. We benchmark PI-UF3 on a five-element refractory high-entropy alloy (HEA) and a subset of MD22, training on forces, with restarts and cross-validation. On the chemically homogeneous HEA, compression acts as implicit regularization: with three-body terms included, PI-UF3’s test force root-mean-squared error drops below the unfactorized UF3 baseline with 3% of the baseline’s parameters on the predefined split and 7% on every cross-validation fold, and the maximally compressed model recovers a chemically meaningful ordering of homonuclear interaction strengths consistent with experimental bulk melting temperatures. On the chemically heterogeneous MD22 subset, where UF3 is not expressive enough to overfit, PI-UF3 approaches the baseline from above and reaches parity with it at about two thirds of the parameter count. Compression thus regularizes only where the unfactorized model overfits, while a parameter reduction is available in both regimes.
Autor:innen und Institutionen
- Department of Chemistry, University of Texas at Austin, Austin, Texas 78712, USA.
ORCID-Profile: Sung Hoon Jung, Ilgar Baghishov, Graeme Henkelman
Förderung
- Welch Foundation
Zitieren
APA
Jung, S. H., Baghishov, I., & Henkelman, G. (2026). Interaction embedding via factorized linear models for multi-element machine learning interatomic potentials. The Journal of Chemical Physics, 165(17). https://doi.org/10.1063/5.0345011
Vancouver
Jung SH, Baghishov I, Henkelman G. Interaction embedding via factorized linear models for multi-element machine learning interatomic potentials. The Journal of Chemical Physics. 2026;165(17). doi:10.1063/5.0345011
BibTeX
@article{Jung2026,
author = {Jung, Sung Hoon and Baghishov, Ilgar and Henkelman, Graeme},
title = {Interaction embedding via factorized linear models for multi-element machine learning interatomic potentials},
journal = {The Journal of Chemical Physics},
year = {2026},
volume = {165},
number = {17},
doi = {10.1063/5.0345011}
}
EUVIMED