Long-range electrostatics for machine learning interatomic potentials is easier than we thought

2025-12-19

AI Summary

Existing machine learning interatomic potentials (MLIPs) commonly lack explicit long-range electrostatic interactions, severely limiting their reliability for interfaces, polar/ionic materials, charge-transfer reactions, and biomolecules. To address this, we propose two physics-guided design principles: (i) adopting a Coulombic functional form with environment-dependent atomic charges, and (ii) avoiding explicit fitting to ambiguous DFT partial charges—thereby reducing model complexity and improving generalizability. Building upon the Latent Ewald Summation (LES) framework, our method integrates environment-dependent charges, short-range MLIP enhancement, charge relaxation, and tensor-targeted learning, requiring only standard energy and force labels. It achieves high-fidelity modeling of charge distributions, electrostatic responses, and long-range electrostatics. We validate the approach across diverse challenging systems, demonstrating high accuracy, strong transferability, and broad applicability—without auxiliary charge or dipole labels.

Technology Category

• Machine Learning: Kernel Methods
• Natural Language Processing: (Large) Language Models
• Multiagent Systems: Multiagent Learning

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: ML for personalized search and recommendations

Abstract

The lack of long-range electrostatics is a key limitation of modern machine learning interatomic potentials (MLIPs), hindering reliable applications to interfaces, charge-transfer reactions, polar and ionic materials, and biomolecules. In this Perspective, we distill two design principles behind the Latent Ewald Summation (LES) framework, which can capture long-range interactions, charges, and electrical response just by learning from standard energy and force training data: (i) use a Coulomb functional form with environment-dependent charges to capture electrostatic interactions, and (ii) avoid explicit training on ambiguous density functional theory (DFT) partial charges. When both principles are satisfied, substantial flexibility remains: essentially any short-range MLIP can be augmented; charge equilibration schemes can be added when desired; dipoles and Born effective charges can be inferred or finetuned; and charge/spin-state embeddings or tensorial targets can be further incorporated. We also discuss current limitations and open challenges. Together, these minimal, physics-guided design rules suggest that incorporating long-range electrostatics into MLIPs is simpler and perhaps more broadly applicable than is commonly assumed.

Problem

• Addresses lack of long-range electrostatics in machine learning interatomic potentials
• Proposes a framework to capture long-range interactions using standard training data
• Enables application to interfaces, ionic materials, and charge-transfer reactions

Innovation

• Latent Ewald Summation framework captures long-range electrostatics
• Uses Coulomb functional form with environment-dependent charges
• Avoids explicit training on ambiguous DFT partial charges