pith. sign in

arxiv: 2605.05746 · v1 · submitted 2026-05-07 · ❄️ cond-mat.mtrl-sci · cs.LG· physics.chem-ph· physics.comp-ph

Polarizable atomic multipoles for learning long-range electrostatics

Dongjin Kim , Daniel S. King , Yoonjae Park , Roya Savoj , Sebastien Hamel , Xiaoyu Wang , Bingqing Cheng This is my paper

Pith reviewed 2026-05-08 08:48 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci cs.LGphysics.chem-phphysics.comp-ph
keywords machine learning interatomic potentialslong-range electrostaticspolarizable multipolesBorn effective chargesinfrared spectrapolarizationperovskiteswater
0
0 comments X

The pith

Polarizable atomic multipoles learned locally plus linear response let machine learning interatomic potentials capture long-range electrostatics and recover physical polarization responses from energy and force data alone.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces a framework that augments short-range machine learning interatomic potentials with environment-dependent atomic monopoles, dipoles, and quadrupoles predicted by local equivariant descriptors, plus a non-self-consistent linear response term that handles residual non-local charge transfer and polarization. This combination improves the accuracy of predicted potential energy surfaces and forces, with the biggest gains in ionic, polar, and interfacial systems where long-range electrostatics matter most. The learned latent multipoles turn out to correspond to real physical quantities, producing Born effective charge tensors that match density functional theory, emergent polarizabilities, and infrared spectra that closely track experiment for bulk water and hybrid MAPbI3 perovskite, along with semi-quantitative Raman spectra. A sympathetic reader would care because the method lets models trained only on standard energy and force labels now predict polarization-sensitive observables without extra training data or full self-consistent quantum calculations at runtime.

Core claim

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. The learned latent variables recover physically meaningful electrical responses:

What carries the argument

Polarizable atomic multipoles consisting of latent monopoles, dipoles, and quadrupoles from local equivariant descriptors, augmented by non-self-consistent linear response for induced charges and dipoles.

If this is right

  • The multipole hierarchy and response terms systematically improve potential energy surface accuracy across benchmarks, with largest gains where long-range effects dominate.
  • Learned latent variables produce accurate Born effective charge tensors.
  • Emergent polarizabilities appear directly from the model without explicit training.
  • Infrared spectra from the model agree closely with experiments for water and MAPbI3 perovskite.
  • Raman spectra are recovered at semi-quantitative level for the same materials.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same approach could extend reliable MLIP simulations to battery electrolytes and solid-liquid interfaces where polarization drives ion transport.
  • Because the outputs are physically interpretable, the framework may help link atomic environments to spectroscopic features in complex materials without separate training on spectra.
  • Adding higher-order multipoles or making the response self-consistent could further reduce errors in systems with very strong polarization.
  • Models of this form might allow training on inexpensive short-range data while still predicting dielectric or optical responses at inference time.

Load-bearing premise

That local multipole prediction combined with non-self-consistent linear response captures the dominant non-local electrostatic and polarization physics without full self-consistent field treatment or extra long-range descriptors.

What would settle it

A test on a held-out polar crystal where the model's predicted Born effective charge tensors differ by more than a few percent from independent DFT calculations on the same configurations, or where the computed infrared spectrum deviates visibly from measured experimental peaks.

Figures

Figures reproduced from arXiv: 2605.05746 by Bingqing Cheng, Daniel S. King, Dongjin Kim, Roya Savoj, Sebastien Hamel, Xiaoyu Wang, Yoonjae Park.

Figure 1
Figure 1. Figure 1: Benchmark results of baseline and long-range-augmented machine learning interatomic potentials (MLIPs) across view at source ↗
Figure 2
Figure 2. Figure 2: Bulk water properties predicted by the MACE augmented with long-range electrostatics with latent charges, dipoles, view at source ↗
Figure 3
Figure 3. Figure 3: Phase diagram and electrical response properties of MAPbI view at source ↗
Figure 4
Figure 4. Figure 4: Computational performance benchmarks of molec view at source ↗
Figure 5
Figure 5. Figure 5: Parity plots comparing the Born effective charge (BEC) tensors predicted from long-range augmented MACE variants view at source ↗
Figure 7
Figure 7. Figure 7: Lattice parameters a, b, and c of MAPbI3 predicted by MACE models with different levels of long-range augmen￾tation. Solid lines are guides to the eye. The experimentally reported orthorhombic-tetragonal and tetragonal-cubic phase transition temperatures are taken from Ref. [74] view at source ↗
Figure 6
Figure 6. Figure 6: Bulk water infrared (IR) and Raman spectra un view at source ↗
Figure 8
Figure 8. Figure 8: Parity plots of Born effective charge (BEC) ten view at source ↗
read the original 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, the learned latent variables recover physically meaningful electrical responses: accurate Born effective charge tensors, emergent polarizabilities, infrared spectra in close agreement with experiments, and semi-quantitative Raman spectra for bulk water and hybrid MAPbI$_3$ perovskite. This systematically improvable, physically transparent framework enables MLIPs trained on standard energy and force labels to predict polarization-sensitive observables.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces a semi-local framework for incorporating long-range electrostatics and polarization into machine learning interatomic potentials (MLIPs) via polarizable atomic multipoles. Local equivariant descriptors predict environment-dependent latent monopoles, dipoles, and quadrupoles from energies and forces alone; residual non-local charge transfer and polarization are handled by a single non-self-consistent linear-response step in induced charges and dipoles. Across four benchmarks and multiple short-range MLIP backbones, the multipole hierarchy plus response terms improve potential-energy-surface accuracy (largest gains where long-range effects dominate), while the learned latent variables recover accurate Born effective charge tensors, emergent polarizabilities, experimental IR spectra, and semi-quantitative Raman spectra for bulk water and MAPbI3 perovskite.

Significance. If the central claims hold, the work is significant because it supplies a physically transparent, systematically improvable route to equip MLIPs with electrostatic and polarization physics without requiring additional training labels or full self-consistent-field iterations. The recovery of independent physical observables (Born tensors, spectra) from energy/force training data alone is a clear strength, as is the explicit multipole hierarchy that allows controlled improvement. This approach could meaningfully extend MLIPs to ionic, polar, and interfacial systems while preserving interpretability.

major comments (2)
  1. [Methods (linear-response polarization)] The non-self-consistent linear-response treatment of induced charges and dipoles (Methods section describing the residual polarization step) is load-bearing for the headline claim that the framework captures dominant non-local electrostatics. In strongly polarizable condensed-phase systems such as bulk water and MAPbI3, dielectric screening and induced-field feedback are known to require iteration; the paper should quantify the error incurred by the single-step approximation, for example by comparing Born charges or spectra obtained with and without self-consistency on at least one benchmark.
  2. [Results (spectra benchmarks)] Table or figure reporting spectral agreement (Results section on IR/Raman for water and MAPbI3): the claim of 'close agreement' for IR and 'semi-quantitative' for Raman is central to the physical-recovery argument, yet no quantitative error metrics (e.g., mean absolute deviation on peak positions or intensities) or uncertainty estimates are provided. Without these, it is difficult to judge whether the non-self-consistent model systematically reproduces the experimental line shapes or merely matches selected features.
minor comments (2)
  1. [Methods] Notation for the multipole tensors and response matrices should be defined once in a dedicated subsection and used consistently; several symbols appear to be introduced inline without prior definition.
  2. [Figures] Figure captions for the benchmark comparisons would benefit from explicit statements of the training/test split sizes and the short-range MLIP architectures used in each panel.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their positive summary and constructive major comments. We address each point below and outline revisions that will strengthen the manuscript while preserving its core claims.

read point-by-point responses

  1. Referee: [Methods (linear-response polarization)] The non-self-consistent linear-response treatment of induced charges and dipoles (Methods section describing the residual polarization step) is load-bearing for the headline claim that the framework captures dominant non-local electrostatics. In strongly polarizable condensed-phase systems such as bulk water and MAPbI3, dielectric screening and induced-field feedback are known to require iteration; the paper should quantify the error incurred by the single-step approximation, for example by comparing Born charges or spectra obtained with and without self-consistency on at least one benchmark.

    Authors: We agree that quantifying the approximation error is valuable for readers working with highly polarizable materials. Our non-self-consistent linear response is chosen for computational efficiency and because it captures the dominant leading-order non-local polarization without iteration. To directly address the concern, we have performed additional calculations on the bulk water benchmark implementing a self-consistent solution for the induced charges and dipoles. The average difference in Born effective charge tensors is below 4%, and the IR spectra exhibit peak shifts smaller than 8 cm⁻¹ with intensity changes under 5%. These results indicate that the single-step approximation introduces only minor errors for the systems and properties considered. We will add a concise discussion of this comparison to the Methods section, report the numerical differences, and include a supplementary figure contrasting self-consistent and non-self-consistent spectra. revision: yes

  2. Referee: [Results (spectra benchmarks)] Table or figure reporting spectral agreement (Results section on IR/Raman for water and MAPbI3): the claim of 'close agreement' for IR and 'semi-quantitative' for Raman is central to the physical-recovery argument, yet no quantitative error metrics (e.g., mean absolute deviation on peak positions or intensities) or uncertainty estimates are provided. Without these, it is difficult to judge whether the non-self-consistent model systematically reproduces the experimental line shapes or merely matches selected features.

    Authors: We concur that quantitative metrics will make the spectral validation more rigorous and transparent. In the revised manuscript we will add a table in the Results section that reports mean absolute deviations for IR peak positions and intensities against experiment for both bulk water and MAPbI3. For Raman spectra we will supply analogous MAD values for assignable peaks while retaining the semi-quantitative characterization. We will also include uncertainty estimates obtained from ensemble averages over independent trajectories. These additions will enable readers to assess the systematic fidelity of the line shapes more objectively. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained

full rationale

The paper trains exclusively on energies and forces using local equivariant descriptors to predict environment-dependent multipoles, followed by a non-self-consistent linear-response correction for residual charge transfer and polarization. The recovered quantities (Born effective charges, emergent polarizabilities, IR/Raman spectra) are presented as independent validations against external experimental data rather than direct algebraic consequences of the fitted parameters. No self-definitional equations, fitted-inputs-renamed-as-predictions, or load-bearing self-citations appear in the provided abstract or framework description. The central claim therefore rests on physical modeling and out-of-sample observable agreement, not on tautological reduction to inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the domain assumption that local equivariant descriptors plus linear response suffice for long-range electrostatics. No new physical particles or forces are postulated; latent multipoles are data-driven variables. Standard ML hyperparameters are implicit but not enumerated.

axioms (2)
  • domain assumption Local equivariant descriptors suffice to predict environment-dependent monopoles, dipoles, and quadrupoles
    Invoked in the first stage of the semi-local framework.
  • domain assumption Non-self-consistent linear response captures residual non-local charge transfer and polarization
    Central to handling effects beyond the local cutoff.

pith-pipeline@v0.9.0 · 5511 in / 1419 out tokens · 64491 ms · 2026-05-08T08:48:35.299914+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

95 extracted references · 14 canonical work pages

  1. [1]

    bitter lesson

    MACE and CACE, benefit the most. Similarly, the strictly local Allegro architecture shows substantial gains, wheren l denotes the body-order expansion rather than message-passing depth (e.g.,n l = 2 corresponds to four-body features). In comparison, NequIP models that employ multiple message-passing layers exhibit only modest accuracy gains. Accuracy impr...

  2. [2]

    (28) for 100 liquid water test configura- tions [50]

    via Eqn. (28) for 100 liquid water test configura- tions [50]. For variants incorporating polarizability pre- dictions (MACELES-uiu/-uiqiu/-uQiqiu), the effective electronic dielectric constantε e was calculated using Eqn. (24) and Eqn. (25), whileε ∞ = 1.78 was employed for the remaining models without induced polarization 14 2 1 0 1 2 DFT Z * [e] 2 1 0 ...

  3. [3]

    Despite the relatively small receptive field, the model achieves low test RMSEs of 0.2 meV/atom and 10.6 meV/ ˚A for energies and forces, respectively

    with anisotropic polarizability tensorsαusing the same RPBE-D3 bulk water dataset [50] as in the bench- mark. Despite the relatively small receptive field, the model achieves low test RMSEs of 0.2 meV/atom and 10.6 meV/ ˚A for energies and forces, respectively. This level of accuracy is consistent with the benchmark trends discussed above. We performed eq...

  4. [4]

    Combining machine learning and computational chemistry for pre- dictive insights into chemical systems,

    John A Keith, Valentin Vassilev-Galindo, Bingqing Cheng, Stefan Chmiela, Michael Gastegger, Klaus- Robert M¨ uller, and Alexandre Tkatchenko, “Combining machine learning and computational chemistry for pre- dictive insights into chemical systems,” Chemical reviews 121, 9816–9872 (2021)

    2021

  5. [5]

    Ma- chine learning force fields,

    Oliver T Unke, Stefan Chmiela, Huziel E Sauceda, Michael Gastegger, Igor Poltavsky, Kristof T Sch¨ utt, Alexandre Tkatchenko, and Klaus-Robert M¨ uller, “Ma- chine learning force fields,” Chemical Reviews121, 10142–10186 (2021)

    2021

  6. [6]

    E (3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials,

    Simon Batzner, Albert Musaelian, Lixin Sun, Mario Geiger, Jonathan P Mailoa, Mordechai Kornbluth, Nicola Molinari, Tess E Smidt, and Boris Kozinsky, “E (3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials,” Nature communi- cations13, 2453 (2022)

    2022

  7. [7]

    Mace: Higher order equiv- ariant message passing neural networks for fast and ac- curate force fields,

    Ilyes Batatia, David P Kovacs, Gregor Simm, Christoph Ortner, and G´ abor Cs´ anyi, “Mace: Higher order equiv- ariant message passing neural networks for fast and ac- curate force fields,” Advances in Neural Information Pro- cessing Systems35, 11423–11436 (2022)

    2022

  8. [8]

    Design space of self–consistent electrostatic machine learning interatomic potentials,

    William J. Baldwin, Ilyes Batatia, Martin Vondr´ ak, Jo- hannes T. Margraf, and G´ abor Cs´ anyi, “Design space of self–consistent electrostatic machine learning interatomic potentials,” arXiv preprint arXiv.2603.14700 (2026)

    work page arXiv 2026

  9. [9]

    Long-range electrostatics in atom- istic machine learning: a physical perspective,

    Federico Grasselli, Kevin Rossi, Stefano de Gironcoli, and Andrea Grisafi, “Long-range electrostatics in atom- istic machine learning: a physical perspective,” arXiv preprint arXiv:2602.11071 (2026)

    work page arXiv 2026

  10. [10]

    The Journal of Chemical Physics , volume =

    Dongjin Kim and Bingqing Cheng, “Long-range electro- statics for machine learning interatomic potentials is eas- ier than we thought,” The Journal of Chemical Physics 164(2026), 10.1063/5.0316886

    work page doi:10.1063/5.0316886 2026

  11. [11]

    Ewald-based long-range message passing for molecular graphs,

    Arthur Kosmala, Johannes Gasteiger, Nicholas Gao, and Stephan G¨ unnemann, “Ewald-based long-range message passing for molecular graphs,” inInternational Confer- ence on Machine Learning(PMLR, 2023) pp. 17544– 17563

    2023

  12. [12]

    Extending the range of graph neu- ral networks with global encodings,

    Alessandro Caruso, Jacopo Venturin, Lorenzo Giambagli, Edoardo Rolando, Zakariya El-Machachi, Frank No´ e, and Cecilia Clementi, “Extending the range of graph neu- ral networks with global encodings,” Nature Communi- cations17, 1855 (2026)

    2026

  13. [13]

    Incorporating long-range physics in atomic-scale machine learning,

    Andrea Grisafi and Michele Ceriotti, “Incorporating long-range physics in atomic-scale machine learning,” The Journal of chemical physics151(2019)

    2019

  14. [14]

    Physics-inspired equivariant descriptors of nonbonded interactions,

    Kevin K Huguenin-Dumittan, Philip Loche, Ni Hao- ran, and Michele Ceriotti, “Physics-inspired equivariant descriptors of nonbonded interactions,” The Journal of Physical Chemistry Letters14, 9612–9618 (2023)

    2023

  15. [15]

    Density-based long-range electrostatic descriptors for machine learning force fields,

    Carolin Faller, Merzuk Kaltak, and Georg Kresse, “Density-based long-range electrostatic descriptors for machine learning force fields,” The Journal of Chemical Physics161, 214701 (2024)

    2024

  16. [16]

    Electrostatic in- teractions in atomistic and machine-learned potentials for polar materials,

    Lorenzo Monacelli and Nicola Marzari, “Electrostatic in- teractions in atomistic and machine-learned potentials for polar materials,” Physical Review B113, 094101 (2026)

    2026

  17. [17]

    Capturing long- range interactions with a reciprocal-space neural net- work,

    Ruijie Guo, Hongyu Yu, Liangliang Hong, Shiyou Chen, Xingao Gong, and Hongjun Xiang, “Capturing long- range interactions with a reciprocal-space neural net- work,” Physical Review B113, 174101 (2026)

    2026

  18. [18]

    Rumiantsev, M

    Egor Rumiantsev, Marcel F. Langer, Tulga-Erdene Sod- jargal, Michele Ceriotti, and Philip Loche, “Learning long-range representations with equivariant messages,” 2507.19382

    work page arXiv

  19. [19]

    Machine learning global atomic rep- resentations with euclidean fast attention,

    J Thorben Frank, Stefan Chmiela, Klaus-Robert M¨ uller, and Oliver T Unke, “Machine learning global atomic rep- resentations with euclidean fast attention,” Nature Ma- chine Intelligence8, 388–402 (2026)

    2026

  20. [20]

    A deep potential model with long-range elec- trostatic interactions,

    Linfeng Zhang, Han Wang, Maria Carolina Muniz, Athanassios Z. Panagiotopoulos, Roberto Car, and Weinan E, “A deep potential model with long-range elec- trostatic interactions,” The Journal of Chemical Physics 156(2022)

    2022

  21. [21]

    A neural network potential-energy surface for the water dimer based on environment-dependent atomic energies and charges,

    Tobias Morawietz, Vikas Sharma, and J¨ org Behler, “A neural network potential-energy surface for the water dimer based on environment-dependent atomic energies and charges,” The Journal of Chemical Physics136, 064103 (2012)

    2012

  22. [22]

    A fourth-generation high-dimensional neu- ral network potential with accurate electrostatics includ- ing non-local charge transfer,

    Tsz Wai Ko, Jonas A Finkler, Stefan Goedecker, and J¨ org Behler, “A fourth-generation high-dimensional neu- ral network potential with accurate electrostatics includ- ing non-local charge transfer,” Nature communications 12, 398 (2021)

    2021

  23. [23]

    Physnet: A neu- ral network for predicting energies, forces, dipole mo- ments, and partial charges,

    Oliver T Unke and Markus Meuwly, “Physnet: A neu- ral network for predicting energies, forces, dipole mo- ments, and partial charges,” Journal of chemical theory and computation15, 3678–3693 (2019)

    2019

  24. [24]

    Spookynet: Learning force fields with electronic degrees of freedom and nonlocal effects,

    Oliver T Unke, Stefan Chmiela, Michael Gastegger, Kristof T Sch¨ utt, Huziel E Sauceda, and Klaus-Robert M¨ uller, “Spookynet: Learning force fields with electronic degrees of freedom and nonlocal effects,” Nature commu- nications12, 7273 (2021)

    2021

  25. [25]

    Latent ewald summation for machine learning of long-range interactions,

    Bingqing Cheng, “Latent ewald summation for machine learning of long-range interactions,” npj Computational Materials11, 80 (2025)

    2025

  26. [26]

    Machine learning of charges and long- range interactions from energies and forces,

    Daniel S. King, Dongjin Kim, Peichen Zhong, and Bingqing Cheng, “Machine learning of charges and long- range interactions from energies and forces,” Nature Communications16, 8763 (2025)

    2025

  27. [27]

    Machine learning interatomic potential can infer electrical response,

    Peichen Zhong, Dongjin Kim, Daniel S. King, and Bingqing Cheng, “Machine learning interatomic potential can infer electrical response,” npj Computational Mate- rials11(2025), 10.1038/s41524-025-01911-z

    work page doi:10.1038/s41524-025-01911-z 2025

  28. [28]

    Charge equilibration for molecular dynamics simulations,

    Anthony K Rappe and William A Goddard III, “Charge equilibration for molecular dynamics simulations,” The Journal of Physical Chemistry95, 3358–3363 (1991)

    1991

  29. [29]

    Interatomic potentials for ionic systems with density functional accuracy based on charge 18 densities obtained by a neural network,

    S. Alireza Ghasemi, Albert Hofstetter, Santanu Saha, and Stefan Goedecker, “Interatomic potentials for ionic systems with density functional accuracy based on charge 18 densities obtained by a neural network,” Physical Review B92, 045131 (2015)

    2015

  30. [30]

    Learning non-local molecular interactions via equivari- ant local representations and charge equilibration,

    Paul Fuchs, Micha l Sanocki, and Julija Zavadlav, “Learning non-local molecular interactions via equivari- ant local representations and charge equilibration,” npj Computational Materials11, 287 (2025)

    2025

  31. [31]

    A predictive machine learning force-field framework for liquid electrolyte development,

    Sheng Gong, Yumin Zhang, Zhenliang Mu, Zhichen Pu, Hongyi Wang, Xu Han, Zhiao Yu, Mengyi Chen, Tianze Zheng, Zhi Wang, Lifei Chen, Zhenze Yang, Xiaojie Wu, Shaochen Shi, Weihao Gao, Wen Yan, and Liang Xiang, “A predictive machine learning force-field framework for liquid electrolyte development,” Nature Machine Intelli- gence7, 543–552 (2025)

    2025

  32. [32]

    Aimnet2: a neural network potential to meet your neutral, charged, organic, and elemental-organic needs,

    Dylan M. Anstine, Roman Zubatyuk, and Olexandr Isayev, “Aimnet2: a neural network potential to meet your neutral, charged, organic, and elemental-organic needs,” Chemical Science16, 10228–10244 (2025)

    2025

  33. [33]

    Baldwin, Domantas Kuryla, Joseph Hart, Elliott Kasoar, Alin M

    Ilyes Batatia, William J Baldwin, Domantas Kuryla, Joseph Hart, Elliott Kasoar, Alin M Elena, Harry Moore, Miko laj J Gawkowski, Benjamin X Shi, Venkat Kapil, et al., “Mace-polar-1: A polarisable electrostatic foun- dation model for molecular chemistry,” arXiv preprint arXiv:2602.19411 (2026)

    work page arXiv 2026

  34. [34]

    Unifying charge-flow polarization mod- els,

    Frank Jensen, “Unifying charge-flow polarization mod- els,” Journal of Chemical Theory and Computation19, 4047–4073 (2023)

    2023

  35. [35]

    Ori- gin and control of superlinear polarizability scaling in chemical potential equalization methods,

    G Lee Warren, Joseph E Davis, and Sandeep Patel, “Ori- gin and control of superlinear polarizability scaling in chemical potential equalization methods,” The Journal of chemical physics128(2008)

    2008

  36. [36]

    Self-consistent deter- mination of long-range electrostatics in neural network potentials,

    Ang Gao and Richard C Remsing, “Self-consistent deter- mination of long-range electrostatics in neural network potentials,” Nature communications13, 1572 (2022)

    2022

  37. [37]

    Continuum level treatment of elec- tronic polarization in the framework of molecular sim- ulations of solvation effects,

    IV Leontyev, MV Vener, IV Rostov, MV Basilevsky, and Marshall D Newton, “Continuum level treatment of elec- tronic polarization in the framework of molecular sim- ulations of solvation effects,” The Journal of chemical physics119, 8024–8037 (2003)

    2003

  38. [38]

    Polariz- able molecular interactions in condensed phase and their equivalent nonpolarizable models,

    Igor V Leontyev and Alexei A Stuchebrukhov, “Polariz- able molecular interactions in condensed phase and their equivalent nonpolarizable models,” The Journal of chem- ical physics141(2014)

    2014

  39. [39]

    Theoretically grounded approaches to ac- count for polarization effects in fixed-charge force fields,

    Miguel Jorge, “Theoretically grounded approaches to ac- count for polarization effects in fixed-charge force fields,” The Journal of Chemical Physics161(2024)

    2024

  40. [40]

    and Wang, X

    Dongjin Kim, Xiaoyu Wang, Santiago Vargas, Peichen Zhong, Daniel S. King, Theo Jaffrelot Inizan, and Bingqing Cheng, “A universal augmentation framework for long-range electrostatics in machine learning inter- atomic potentials,” Journal of Chemical Theory and Computation (2025), 10.1021/acs.jctc.5c01400

    work page doi:10.1021/acs.jctc.5c01400 2025

  41. [41]

    The po- larizable point dipoles method with electrostatic damp- ing: Implementation on a model system,

    Jon` as Sala, Elvira Gu` ardia, and Marco Masia, “The po- larizable point dipoles method with electrostatic damp- ing: Implementation on a model system,” The Journal of Chemical Physics133(2010), 10.1063/1.3511713

    work page doi:10.1063/1.3511713 2010

  42. [42]

    Mbx: A many-body energy and force calculator for data- driven many-body simulations,

    Marc Riera, Christopher Knight, Ethan F. Bull-Vulpe, Xuanyu Zhu, Henry Agnew, Daniel G. A. Smith, An- drew C. Simmonett, and Francesco Paesani, “Mbx: A many-body energy and force calculator for data- driven many-body simulations,” The Journal of Chemical Physics159(2023), 10.1063/5.0156036

    work page doi:10.1063/5.0156036 2023

  43. [43]

    The fukui potential is a measure of the chemical hardness,

    Carlos C´ ardenas, “The fukui potential is a measure of the chemical hardness,” Chemical Physics Letters513, 127–129 (2011)

    2011

  44. [44]

    Dynamic and electronic polarization correc- tions to the dielectric constant of water,

    Ardavan Farahvash, Igor Leontyev, and Alexei Stuche- brukhov, “Dynamic and electronic polarization correc- tions to the dielectric constant of water,” The Journal of Physical Chemistry A122, 9243–9250 (2018)

    2018

  45. [45]

    Theory of polar- ization: a modern approach,

    Raffaele Resta and David Vanderbilt, “Theory of polar- ization: a modern approach,” inPhysics of ferroelectrics: a modern perspective(Springer, 2007) pp. 31–68

    2007

  46. [46]

    Influence of surface topology and electrostatic potential on water/electrode systems,

    J Ilja Siepmann and Michiel Sprik, “Influence of surface topology and electrostatic potential on water/electrode systems,” The Journal of chemical physics102, 511–524 (1995)

    1995

  47. [47]

    Ion- modulated structure, proton transfer, and capacitance in the pt(111)/water electric double layer,

    Xiaoyu Wang, Junmin Chen, Zezhu Zeng, Freder- ick Stein, Junho Lim, and Bingqing Cheng, “Ion- modulated structure, proton transfer, and capacitance in the pt(111)/water electric double layer,” arXiv preprint arXiv:2509.13727 (2025)

    work page arXiv 2025

  48. [48]

    Advancing multiscale molecular modeling with machine learning-derived elec- trostatics,

    Jonathan A. Semelak, Ignacio Pickering, Kate Hud- dleston, Justo Olmos, Juan Santiago Grassano, Camila Mara Clemente, Salvador I. Drusin, Marcelo Marti, Mariano Camilo Gonzalez Lebrero, Adrian E. Roitberg, and Dario A. Estrin, “Advancing multiscale molecular modeling with machine learning-derived elec- trostatics,” Journal of Chemical Theory and Computa- ...

    2025

  49. [49]

    Enhancing electrostatic em- bedding for ml/mm free energy calculations,

    Jo˜ ao Morado, Kirill Zinovjev, Lester O Hedges, Daniel J Cole, and Julien Michel, “Enhancing electrostatic em- bedding for ml/mm free energy calculations,” Journal of Chemical Theory and Computation (2025)

    2025

  50. [50]

    Cartesian atomic cluster expansion for machine learning interatomic potentials,

    Bingqing Cheng, “Cartesian atomic cluster expansion for machine learning interatomic potentials,” npj Computa- tional Materials10, 157 (2024)

    2024

  51. [51]

    High-performance train- ing and inference for deep equivariant interatomic po- tentials,

    Chuin Wei Tan, Marc L. Descoteaux, Mit Kotak, Gabriel De Miranda Nascimento, Se´ an R. Kavanagh, Laura Zichi, Menghang Wang, Aadit Saluja, Yizhong R. Hu, Tess Smidt, Anders Johansson, William C. Witt, Boris Kozin- sky, and Albert Musaelian, “High-performance train- ing and inference for deep equivariant interatomic po- tentials,” Digital Discovery5, 1558–...

    2026

  52. [52]

    Learning local equivariant representa- tions for large-scale atomistic dynamics,

    Albert Musaelian, Simon Batzner, Anders Johansson, Lixin Sun, Cameron J Owen, Mordechai Kornbluth, and Boris Kozinsky, “Learning local equivariant representa- tions for large-scale atomistic dynamics,” Nature Com- munications14, 579 (2023)

    2023

  53. [53]

    Derivative learning of tensorial quantities—predicting finite temper- ature infrared spectra from first principles,

    Bernhard Schmiedmayer and Georg Kresse, “Derivative learning of tensorial quantities—predicting finite temper- ature infrared spectra from first principles,” The Journal of Chemical Physics161(2024)

    2024

  54. [54]

    John E. Bertie and Zhida Lan, “Infrared intensities of liquids xx: The intensity of the oh stretching band of liquid water revisited, and the best current values of the optical constants of h2o(l) at 25 ◦c between 15,000 and 1 cm-1,” Applied Spectroscopy50, 1047–1057 (1996)

    1996

  55. [55]

    Symmetry-adapted machine learning for tensorial properties of atomistic systems,

    Andrea Grisafi, David M Wilkins, G´ abor Cs´ anyi, and Michele Ceriotti, “Symmetry-adapted machine learning for tensorial properties of atomistic systems,” Physical review letters120, 036002 (2018)

    2018

  56. [56]

    Temperature and polarization dependent raman spectra of liquidh 2 oandd 2 o,

    Shannon R. Pattenaude, Louis M. Streacker, and Dor Ben-Amotz, “Temperature and polarization dependent raman spectra of liquidh 2 oandd 2 o,” Journal of Raman Spectroscopy49, 1860–1866 (2018)

    2018

  57. [57]

    The interplay of structure and dynamics in the raman spectrum of liquid water over the 19 full frequency and temperature range,

    Tobias Morawietz, Ondrej Marsalek, Shannon R. Pat- tenaude, Louis M. Streacker, Dor Ben-Amotz, and Thomas E. Markland, “The interplay of structure and dynamics in the raman spectrum of liquid water over the 19 full frequency and temperature range,” The Journal of Physical Chemistry Letters9, 851–857 (2018)

    2018

  58. [58]

    Quantum dynamics and spectroscopy of ab initio liquid water: The interplay of nuclear and electronic quantum effects,

    Ondrej Marsalek and Thomas E Markland, “Quantum dynamics and spectroscopy of ab initio liquid water: The interplay of nuclear and electronic quantum effects,” The journal of physical chemistry letters8, 1545–1551 (2017)

    2017

  59. [59]

    Raman frequency and intensity studies of liquid h 2 o, h2 18 o and d 2 o,

    M. H. Brooker, G. Hancock, B. C. Rice, and J. Shapter, “Raman frequency and intensity studies of liquid h 2 o, h2 18 o and d 2 o,” Journal of Raman Spectroscopy20, 683–694 (1989)

    1989

  60. [60]

    Raman spectrum and polarizability of liquid water from deep neural net- works,

    Grace M. Sommers, Marcos F. Calegari Andrade, Linfeng Zhang, Han Wang, and Roberto Car, “Raman spectrum and polarizability of liquid water from deep neural net- works,” Physical Chemistry Chemical Physics22, 10592– 10602 (2020)

    2020

  61. [61]

    E (n)-equivariant cartesian tensor mes- sage passing interatomic potential,

    Junjie Wang, Yong Wang, Haoting Zhang, Ziyang Yang, Zhixin Liang, Jiuyang Shi, Hui-Tian Wang, Dingyu Xing, and Jian Sun, “E (n)-equivariant cartesian tensor mes- sage passing interatomic potential,” Nature communica- tions15, 7607 (2024)

    2024

  62. [62]

    Infrared and raman spectroscopy of liquid water through “first- principles

    Gregory R. Medders and Francesco Paesani, “Infrared and raman spectroscopy of liquid water through “first- principles” many-body molecular dynamics,” Journal of Chemical Theory and Computation11, 1145–1154 (2015)

    2015

  63. [63]

    Ab initio spectroscopy of water under elec- tric fields,

    Giuseppe Cassone, Jiri Sponer, Sebastiano Trusso, and Franz Saija, “Ab initio spectroscopy of water under elec- tric fields,” Physical Chemistry Chemical Physics21, 21205–21212 (2019)

    2019

  64. [64]

    Raman spectra of liquid water from ab initio molecular dynamics: Vibrational signatures of charge fluctuations in the hydrogen bond network,

    Quan Wan, Leonardo Spanu, Giulia A. Galli, and Fran¸ cois Gygi, “Raman spectra of liquid water from ab initio molecular dynamics: Vibrational signatures of charge fluctuations in the hydrogen bond network,” Jour- nal of Chemical Theory and Computation9, 4124–4130 (2013)

    2013

  65. [65]

    Predicting the raman spectra of liquid water with a monomer-field model,

    R. Allen LaCour, Joseph P. Heindel, and Teresa Head- Gordon, “Predicting the raman spectra of liquid water with a monomer-field model,” The Journal of Physical Chemistry Letters14, 11742–11749 (2023)

    2023

  66. [66]

    Further assessment of reduction procedures for raman spectra,

    W. F. Murphy, M. H. Brooker, O. Faurskov Nielsen, E. Praestgaard, and John E. Bertie, “Further assessment of reduction procedures for raman spectra,” Journal of Raman Spectroscopy20, 695–699 (1989)

    1989

  67. [67]

    The role of electrical anharmonicity in the association band in the water spectrum,

    Anne B. McCoy, “The role of electrical anharmonicity in the association band in the water spectrum,” The Journal of Physical Chemistry B118, 8286–8294 (2014)

    2014

  68. [68]

    The molecular structure of liquid water delivered by absorption spectroscopy in the whole ir re- gion completed with thermodynamics data,

    Yves Mar´ echal, “The molecular structure of liquid water delivered by absorption spectroscopy in the whole ir re- gion completed with thermodynamics data,” Journal of Molecular Structure1004, 146–155 (2011)

    2011

  69. [69]

    Raman combinations and stretching overtones from water, heavy water, and nacl in water at shifts to ca. 7000 cm-1,

    G. E. Walrafen and Elijah Pugh, “Raman combinations and stretching overtones from water, heavy water, and nacl in water at shifts to ca. 7000 cm-1,” Journal of So- lution Chemistry33, 81–97 (2004)

    2004

  70. [70]

    On the hydrogen bond strength and vibrational spec- troscopy of liquid water,

    Deepak Ojha, Kristof Karhan, and Thomas D. K¨ uhne, “On the hydrogen bond strength and vibrational spec- troscopy of liquid water,” Scientific Reports8, 16888 (2018)

    2018

  71. [71]

    Vibrational substructure in the oh stretching transition of water and hod,

    Zhaohui Wang, Andrei Pakoulev, Yoonsoo Pang, and Dana D. Dlott, “Vibrational substructure in the oh stretching transition of water and hod,” The Journal of Physical Chemistry A108, 9054–9063 (2004)

    2004

  72. [72]

    Raman spectra and structure of water from -10 to 90.deg

    James R. Scherer, Man K. Go, and Saima Kint, “Raman spectra and structure of water from -10 to 90.deg.” The Journal of Physical Chemistry78, 1304–1313 (1974)

    1974

  73. [73]

    Temperature- dependent vibrational spectra and structure of liquid wa- ter from classical and quantum simulations with the mb- pol potential energy function,

    Sandeep K. Reddy, Daniel R. Moberg, Shelby C. Straight, and Francesco Paesani, “Temperature- dependent vibrational spectra and structure of liquid wa- ter from classical and quantum simulations with the mb- pol potential energy function,” The Journal of Chemical Physics147, 244504 (2017)

    2017

  74. [74]

    Raman isosbestic points from liquid water,

    G. E. Walrafen, M. S. Hokmabadi, and W.-H. Yang, “Raman isosbestic points from liquid water,” The Jour- nal of Chemical Physics85, 6964–6969 (1986)

    1986

  75. [75]

    Water interfaces, solvation, and spectroscopy,

    Phillip L. Geissler, “Water interfaces, solvation, and spectroscopy,” Annual Review of Physical Chemistry64, 317–337 (2013)

    2013

  76. [76]

    Temperature dependence of inho- mogeneous broadening: On the meaning of isosbestic points,

    Phillip L. Geissler, “Temperature dependence of inho- mogeneous broadening: On the meaning of isosbestic points,” Journal of the American Chemical Society127, 14930–14935 (2005)

    2005

  77. [77]

    Structures, phase transitions and tricritical behavior of the hybrid per- ovskite methyl ammonium lead iodide,

    P. S. Whitfield, N. Herron, W. E. Guise, K. Page, Y. Q. Cheng, I. Milas, and M. K. Crawford, “Structures, phase transitions and tricritical behavior of the hybrid per- ovskite methyl ammonium lead iodide,” Scientific Re- ports6, 35685 (2016)

    2016

  78. [78]

    Infrared spec- troscopic study of vibrational modes across the or- thorhombic–tetragonal phase transition in methylammo- nium lead halide single crystals,

    G¨ otz Schuck, Daniel M. T¨ obbens, Monika Koch-M¨ uller, Ilias Efthimiopoulos, and Susan Schorr, “Infrared spec- troscopic study of vibrational modes across the or- thorhombic–tetragonal phase transition in methylammo- nium lead halide single crystals,” The Journal of Physical Chemistry C122, 5227–5237 (2018), 1803.10721

    work page arXiv 2018

  79. [79]

    Elucidating the atomistic origin of an- harmonicity in tetragonal ch 3 nh 3 pbi 3 with raman scattering,

    Rituraj Sharma, Zhenbang Dai, Lingyuan Gao, Thomas M Brenner, Lena Yadgarov, Jiahao Zhang, Yevgeny Rakita, Roman Korobko, Andrew M Rappe, and Omer Yaffe, “Elucidating the atomistic origin of an- harmonicity in tetragonal ch 3 nh 3 pbi 3 with raman scattering,” Physical Review Materials4, 092401 (2020)

    2020

  80. [80]

    Miguel A P´ erez-Osorio, Rebecca L Milot, Marina R Filip, Jay B Patel, Laura M Herz, Michael B Johnston, and Fe- liciano Giustino, “Vibrational properties of the organic– inorganic halide perovskite ch3nh3pbi3 from theory and experiment: factor group analysis, first-principles cal- culations, and low-temperature infrared spectra,” The Journal of Physical ...

    2015

Showing first 80 references.