Polarizable atomic multipoles for learning long-range electrostatics
Abstract
Long-range electrostatics and polarization remain central obstacles to extending machine learning interatomic potentials (MLIPs) to ionic, polar, and interfacial systems. Here we introduce a semi-local framework for learning electrostatics from energies and forces using polarizable atomic multipoles. Local equivariant descriptors predict environment-dependent latent monopoles, dipoles, and quadrupoles, while residual non-local charge transfer and polarization are captured by non-self-consistent linear response in induced charges and dipoles. Across four diverse benchmarks and four short-range MLIP architectures, the multipole hierarchy and response terms systematically improve potential energy surface accuracy, with the largest gains in systems where long-range effects are essential. More importantly, physically meaningful electrical responses emerge without direct supervision. The learned latent multipoles yield accurate Born effective charge tensors and infrared spectra in close agreement with experiments. The induced-dipole extension introduces new capabilities: it predicts polarizabilities and thereby enables semi-quantitative Raman spectra for bulk water and hybrid MAPbI3 perovskite, as well as the essential features of the surface-specific vibrational sum-frequency generation spectrum at the water-air interface. In ferroelectric HfO2, the predicted electrical response also captures LO-TO splitting and polarization switching. This systematically improvable, physically transparent framework enables MLIPs trained on standard energy and force labels to predict polarization-sensitive observables.
Related
- Determination of fundamental properties of nitrogen from first principles. I. Atomic polarizabilities and long-range dispersion coefficients
- Exact periodic electrostatics for latent-charge machine-learning potentials
- Radial Kernel Attention for Long-Range Modeling in Polymer Machine Learning Force Fields
- WISP: Learning spin-resolved electron localization from atomic environments
- MPID Polarizable Force Field: GPU Implementation and Biomolecular Simulations
BibTeX
@misc{park2026polarizable,
title = {{Polarizable atomic multipoles for learning long-range electrostatics}},
author = {Yoonjae Park and Dongjin Kim and Daniel S. King and Nam H. Đào and Roya Savoj and Sebastien Hamel and Xiaoyu Wang and Bingqing Cheng},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2605.05746},
eprint = {2605.05746},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.05746},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Polarizable atomic multipoles for learning long-range electrostatics
AU - Yoonjae Park
AU - Dongjin Kim
AU - Daniel S. King
AU - Nam H. Đào
AU - Roya Savoj
AU - Sebastien Hamel
AU - Xiaoyu Wang
AU - Bingqing Cheng
PY - 2026
DA - 2026/09/01
PB - arXiv
DO - 10.48550/arxiv.2605.05746
UR - https://arxiv.org/abs/2605.05746
AB - Long-range electrostatics and polarization remain central obstacles to extending machine learning interatomic potentials (MLIPs) to ionic, polar, and interfacial systems. Here we introduce a semi-local framework for learning electrostatics from energies and forces using polarizable atomic multipoles. Local equivariant descriptors predict environment-dependent latent monopoles, dipoles, and quadrupoles, while residual non-local charge transfer and polarization are captured by non-self-consistent linear response in induced charges and dipoles. Across four diverse benchmarks and four short-range MLIP architectures, the multipole hierarchy and response terms systematically improve potential energy surface accuracy, with the largest gains in systems where long-range effects are essential. More importantly, physically meaningful electrical responses emerge without direct supervision. The learned latent multipoles yield accurate Born effective charge tensors and infrared spectra in close agreement with experiments. The induced-dipole extension introduces new capabilities: it predicts polarizabilities and thereby enables semi-quantitative Raman spectra for bulk water and hybrid MAPbI_3 perovskite, as well as the essential features of the surface-specific vibrational sum-frequency generation spectrum at the water-air interface. In ferroelectric HfO_2, the predicted electrical response also captures LO-TO splitting and polarization switching. This systematically improvable, physically transparent framework enables MLIPs trained on standard energy and force labels to predict polarization-sensitive observables.
ER -