From Energy-Force Weighting to Primal-Dual Optimization of Machine-Learned Interatomic Potentials
Abstract
Machine-learned interatomic potentials are commonly fitted by weighted-sum scalarization, which combines energy and force errors in a single loss. A nominal weight, however, identifies a potential only relative to the complete fitting protocol. We therefore treat energy--force balancing as a protocol-dependent problem of physical model selection. For fixed-basis atomic cluster expansion models of molten LiCl, liquid H2O, and Si, the resulting scalarization paths contain dominated states at extreme weights. Their nondominated subsets depend on the solver, and the out-of-distribution Si path is nonmonotone. We replace direct weight selection by minimizing the regularized energy objective subject to an upper bound on the normalized force loss. Projected dual ascent adjusts the Lagrange multiplier from the force-constraint residual. A frozen-multiplier limited-memory quasi-Newton refinement then returns the final model. Across the three systems, the constrained procedure reaches the solver-matched low-error region and gives measured speedups of $6.7−−8.9$ over completed scalarization scans under the stated timing convention. Physical-property calculations show that first-shell geometry is comparatively insensitive to the selected balance, whereas transport and solid-state observables vary more strongly. These results identify the fitted model--protocol pair as the relevant object of energy--force model selection and support force-constrained fitting as an explicit selection rule.
Related
- Accurate and Rapid Vibrational Spectra of Molecular Crystals from Machine-Learned Interatomic Potentials
- Neural Constrained Charge Equilibration for Extracting Charge Transport Parameters from Machine Learned Interatomic Potentials
- FACTO: fast atom-centered tensor operator for accurate and efficient foundational machine-learned interatomic potentials
- Towards Out-of-Distribution Uncertainty Quantification for Machine-Learned Interatomic Potentials via Pre-Training
- SoLiD26: A First Principles Solid-Liquid Interface Dataset for Machine-learned Interatomic Potentials
BibTeX
@misc{wang2026from,
title = {{From Energy-Force Weighting to Primal-Dual Optimization of Machine-Learned Interatomic Potentials}},
author = {Chenyu Wang and Yangshuai Wang and Lei Zhang},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.00876},
eprint = {2610.00876},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.00876},
note = {CC0},
}
RIS
TY - UNPB
TI - From Energy-Force Weighting to Primal-Dual Optimization of Machine-Learned Interatomic Potentials
AU - Chenyu Wang
AU - Yangshuai Wang
AU - Lei Zhang
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2610.00876
UR - https://arxiv.org/abs/2610.00876
AB - Machine-learned interatomic potentials are commonly fitted by weighted-sum scalarization, which combines energy and force errors in a single loss. A nominal weight, however, identifies a potential only relative to the complete fitting protocol. We therefore treat energy--force balancing as a protocol-dependent problem of physical model selection. For fixed-basis atomic cluster expansion models of molten LiCl, liquid H_2O, and Si, the resulting scalarization paths contain dominated states at extreme weights. Their nondominated subsets depend on the solver, and the out-of-distribution Si path is nonmonotone. We replace direct weight selection by minimizing the regularized energy objective subject to an upper bound on the normalized force loss. Projected dual ascent adjusts the Lagrange multiplier from the force-constraint residual. A frozen-multiplier limited-memory quasi-Newton refinement then returns the final model. Across the three systems, the constrained procedure reaches the solver-matched low-error region and gives measured speedups of $6.7−−8.9$ over completed scalarization scans under the stated timing convention. Physical-property calculations show that first-shell geometry is comparatively insensitive to the selected balance, whereas transport and solid-state observables vary more strongly. These results identify the fitted model--protocol pair as the relevant object of energy--force model selection and support force-constrained fitting as an explicit selection rule.
ER -