A critical review of machine learning interatomic potentials and Hamiltonian

Abstract

Machine learning interatomic potentials (ML-IAPs) and machine learning Hamiltonian (ML-Ham) have revolutionized atomistic and electronic structure simulations by offering near ab initio accuracy across extended time and length scales. In this Review, we summarize recent progress in these two fields, with emphasis on algorithmic and architectural innovations, geometric equivariance, data efficiency strategies, model-data co-design, and interpretable AI techniques. In addition, we discuss key challenges, including data fidelity, model generalizability, computational scalability, and explainability. Finally, we outline promising future directions, such as active learning, multi-fidelity frameworks, scalable message-passing architectures, and methods for enhancing interpretability, which is particularly crucial for the field of AI for Science (AI4S). The integration of these advances is expected to accelerate materials discovery and provide deeper mechanistic insights into complex material and physical systems.

Keywords

Machine learning interatomic potentials / machine learning Hamiltonian / ab initio molecular dynamics / density functional theory / AI for science

References

[1] Krishtal A,Genova A.Subsystem density-functional theory as an effective tool for modeling ground and excited states, their dynamics and many-body interactions.J Phys Condens Matter2015;27:183202

[2] Kohn W.Self-consistent equations including exchange and correlation effects.Phys Rev1965;140:A1133-8

[3] Brockherde F,Li L,Burke K.Bypassing the Kohn-Sham equations with machine learning.Nat Commun2017;8:872 PMCID:PMC5636838

[4] Li H,Zou N.Deep-learning density functional theory Hamiltonian for efficient ab initio electronic-structure calculation.Nat Comput Sci2022;2:367-77 PMCID:PMC11499279

[5] Kochkov D,Sanchez-Gonzalez A,Clark BK. Learning ground states of quantum Hamiltonians with graph networks. arXiv 2021, arXiv:2110.16390. https://doi.org/10.48550/arXiv.2110.06390. (accessed 30 Jun 2025)

[6] Behler J.Generalized neural-network representation of high-dimensional potential-energy surfaces.Phys Rev Lett2007;98:146401

[7] Bartók AP,Kondor R.Gaussian approximation potentials: the accuracy of quantum mechanics, without the electrons.Phys Rev Lett2010;104:136403

[8] Zhang L,Wang H,E W.Deep potential molecular dynamics: a scalable model with the accuracy of quantum mechanics.Phys Rev Lett2018;120:143001

[9] Chan H,Cherukara MJ.Machine learning classical interatomic potentials for molecular dynamics from first-principles training data.J Phys Chem C2019;123:6941-57

[10] Ko TW.Recent advances and outstanding challenges for machine learning interatomic potentials.Nat Comput Sci2023;3:998-1000

[11] Yang Z,Li Y,Chen CY.Efficient equivariant model for machine learning interatomic potentials.npj Comput Mater2025;11:1535

[12] Ahmad W,Chithrananda S,Ramsundar B. ChemBERTa-2: towards chemical foundation models. arXiv 2022, arXiv:2209.01712. https://doi.org/10.48550/arXiv.2209.01712. (accessed 30 Jun 2025)

[13] Li J,Wang Y.Mol‐BERT: an effective molecular representation with BERT for molecular property prediction.Wirel Commun Mob Comput2021;2021:7181815

[14] Batzner S,Sun L.E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials.Nat Commun2022;13:2453 PMCID:PMC9068614

[15] Duignan TT.The potential of neural network potentials.ACS Phys Chem Au2024;4:232-41 PMCID:PMC11117678

[16] Dong L,Yang Z,Lu Y.Accurate piezoelectric tensor prediction with equivariant attention tensor graph neural network.npj Comput Mater2025;11:1546

[17] Zitnick CL,Kolluru A. Spherical channels for modeling atomic interactions. arXiv 2022, arXiv:2206.14331. https://doi.org/10.48550/arXiv.2206.14331. (accessed 30 Jun 2025)

[18] Frank JT,Müller KR.A Euclidean transformer for fast and stable machine learned force fields.Nat Commun2024;15:6539 PMCID:PMC11303804

[19] Yuan Z,Li H.Equivariant neural network force fields for magnetic materials.Quantum Front2024;3:55

[20] Yu H,Hong L.Spin-dependent graph neural network potential for magnetic materials.Phys Rev B2024;109:144426

[21] Wang H,Han J.DeePMD-kit: a deep learning package for many-body potential energy representation and molecular dynamics.Comput Phys Commun2018;228:178-84

[22] Sokolovskiy V,Miroshkina O.Meta-GGA SCAN functional in the prediction of ground state properties of magnetic materials: review of the current state.Metals2023;13:728

[23] Kirklin S,Meredig B.The Open Quantum Materials Database (OQMD): assessing the accuracy of DFT formation energies.npj Comput Mater2015;1:BFnpjcompumats201510

[24] Ramakrishnan R,Rupp M.Quantum chemistry structures and properties of 134 kilo molecules.Sci Data2014;1:140022

[25] Chmiela S,Sauceda HE,Schütt KT.Machine learning of accurate energy-conserving molecular force fields.Sci Adv2017;3:e1603015 PMCID:PMC5419702

[26] Chmiela S,Unke OT. Accurate global machine learning force fields for molecules with hundreds of atoms. arXiv 2022, arXiv 2209.14865. https://doi.org/10.48550/arXiv.2209.14865. (accessed 30 Jun 2025)

[27] Smith JS,Roitberg AE.ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost.Chem Sci2017;8:3192-203 PMCID:PMC5414547

[28] Smith JS,Lubbers N,Roitberg AE.Less is more: sampling chemical space with active learning.J Chem Phys2018;148:241733

[29] Smith JS,Zubatyuk R.Approaching coupled cluster accuracy with a general-purpose neural network potential through transfer learning.Nat Commun2019;10:2903 PMCID:PMC6602931

[30] Devereux C,Huddleston KK.Extending the applicability of the ANI Deep learning molecular potential to sulfur and halogens.J Chem Theory Comput2020;16:4192-202

[31] Schütt KT,Kindermans PJ,Müller KR.SchNet - a deep learning architecture for molecules and materials.J Chem Phys2018;148:241722

[32] Eastman P,Dotson DL.SPICE, a dataset of drug-like molecules and peptides for training machine learning potentials.Sci Data2023;10:11 PMCID:PMC9813265

[33] Chanussot L,Goyal S.Open Catalyst 2020 (OC20) dataset and community challenges.ACS Catal2021;11:6059-72

[34] Tran R,Shuaibi M.The Open Catalyst 2022 (OC22) Dataset and Challenges for Oxide Electrocatalysts.ACS Catal2023;13:3066-84

[35] Jain A,Hautier G.Commentary: The Materials Project: a materials genome approach to accelerating materials innovation.APL Mater2013;1:011002

[36] Dunn A,Ganose A,Jain A.Benchmarking materials property prediction methods: the Matbench test set and Automatminer reference algorithm.npj Comput Mater2020;6:138

[37] Bursch M,Hansen A.Best-practice DFT protocols for basic molecular computational chemistry.Angew Chem Int Ed Engl2022;61:e202205735 PMCID:PMC9826355

[38] Ko TW.Data-efficient construction of high-fidelity graph deep learning interatomic potentials.npj Comput Mater2025;11:1550

[39] Batatia I,Kovács DP.The design space of E(3)-equivariant atom-centred interatomic potentials.Nat Mach Intell2025;7:56-67 PMCID:PMC11769842

[40] Fu X,Barroso-Luque L. Learning smooth and expressive interatomic potentials for physical property prediction. arXiv 2025, arXiv 2502,12147. https://doi.org/10.48550/arXiv.2502.12147. (accessed 30 Jun 2025)

[41] Chmiela S,Müller KR.Towards exact molecular dynamics simulations with machine-learned force fields.Nat Commun2018;9:3887 PMCID:PMC6155327

[42] Choudhary K,Liang T,Hennig RG.Evaluation and comparison of classical interatomic potentials through a user-friendly interactive web-interface.Sci Data2017;4:160125 PMCID:PMC5283064

[43] Wilkinson MD,Aalbersberg IJ.The FAIR Guiding Principles for scientific data management and stewardship.Sci Data2016;3:160018 PMCID:PMC4792175

[44] Blaiszik B,Pruyne J,Tuecke S.The materials data facility: data services to advance materials science research.JOM2016;68:2045-52

[45] Varughese B,Loeffler TD,Cherukara MJ.Active and transfer learning of high-dimensional neural network potentials for transition metals.ACS Appl Mater Interfaces2024;16:20681-92

[46] Zhang L,Wang H,E W.Active learning of uniformly accurate interatomic potentials for materials simulation.Phys Rev Mater2019;3:023804

[47] Zhang L,Su J.Data mining new energy materials from structure databases.Renew Sust Energy Rev2019;107:554-67

[48] Focassio B,Schleder GR.Performance assessment of universal machine learning interatomic potentials: challenges and directions for materials’ surfaces.ACS Appl Mater Interfaces2025;17:13111-21

[49] Kim Y,Yang C,Gu GX.Deep learning framework for material design space exploration using active transfer learning and data augmentation.npj Comput Mater2021;7:609

[50] Mosquera-Lois I,Ganose AM.Machine-learning structural reconstructions for accelerated point defect calculations.npj Comput Mater2024;10:1303

[51] Omee SS,Dong R,Hu J.Structure-based out-of-distribution (OOD) materials property prediction: a benchmark study.npj Comput Mater2024;10:1316

[52] Musaelian A,Johansson A.Learning local equivariant representations for large-scale atomistic dynamics.Nat Commun2023;14:579 PMCID:PMC9898554

[53] Ma X,He R.Active learning of effective Hamiltonian for super-large-scale atomic structures.npj Comput Mater2025;11:1563

[54] Deng B,Jun K.CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling.Nat Mach Intell2023;5:1031-41

[55] Liao YL,Das A. EquiformerV2: improved equivariant transformer for scaling to higher-degree representations. arXiv 2023, arXiv:2306.12059. https://doi.org/10.48550/arXiv.2306.12059. (accessed 30 Jun 2025)

[56] Riebesell J,Benner P. Matbench discovery - a framework to evaluate machine learning crystal stability predictions. arXiv 2023, arXiv.2308.14920. https://doi.org/10.48550/arXiv.2308.14920. (accessed 30 Jun 2025)

[57] Merchant A,Schoenholz SS,Cheon G.Scaling deep learning for materials discovery.Nature2023;624:80-5 PMCID:PMC10700131

[58] McClean JR,Lee J.What the foundations of quantum computer science teach us about chemistry.J Chem Phys2021;155:150901

[59] Castelvecchi D.The AI-quantum computing mash-up: will it revolutionize science? Nature. 2024.

[60] Allen AEA,Matin S.Learning together: towards foundation models for machine learning interatomic potentials with meta-learning.npj Comput Mater2024;10:1339

[61] Zhong Y,Yang J,Xiang H.Universal machine learning Kohn–Sham Hamiltonian for materials.Chinese Phys Lett2024;41:077103

[62] Curchod BFE.Ab initio nonadiabatic quantum molecular dynamics.Chem Rev2018;118:3305-36

[63] Babadi M,Martin I,Demler E.Theory of parametrically amplified electron-phonon superconductivity.Phys Rev B2017;96:014512

[64] Zou J,Lin D,Hou S. Deep learning accelerated quantum transport simulations in nanoelectronics: from break junctions to field-effect transistors. arXiv 2024, arXiv:2411.08800. https://doi.org/10.48550/arXiv.2411.08800. (accessed 30 Jun 2025)

[65] Yan B. Spin-orbit coupling: a relativistic effect. 2016. https://tms16.sciencesconf.org/data/pages/SOC_lecture1.pdf. (accessed 30 Jun 2025)

[66] Inorganic Chemistry II review: 6.1 solid state structures. https://library.fiveable.me/inorganic-chemistry-ii/unit-6/solid-state-structures/study-guide/qnl67GdXxyZ74VJP. (accessed 30 Jun 2025)

[67] Gong X,Zou N,Duan W.General framework for E(3)-equivariant neural network representation of density functional theory Hamiltonian.Nat Commun2023;14:2848 PMCID:PMC10199065

[68] Schütt KT,Tkatchenko A,Maurer RJ.Unifying machine learning and quantum chemistry with a deep neural network for molecular wavefunctions.Nat Commun2019;10:5024 PMCID:PMC6858523

[69] Unke OT,Gastegger M,Smidt T. SE(3)-equivariant prediction of molecular wavefunctions and electronic densities. In Advances in Neural Information Processing Systems 34 (NeurIPS 2021). 2021. https://proceedings.neurips.cc/paper/2021/hash/78f1893678afbeaa90b1fa01b9cfb860-Abstract.html. (accessed 30 Jun 2025)

[70] Thomas N,Kearnes S. Tensor field networks: rotation- and translation-equivariant neural networks for 3D point clouds. arXiv 2018, arXiv:1802.08219. https://doi.org/10.48550/arXiv.1802.08219. (accessed 30 Jun 2025)

[71] Passaro S. Reducing SO(3) convolutions to SO(2) for efficient equivariant GNNs. arXiv 2023, arXiv:2302.03655. https://doi.org/10.48550/arXiv.2302.03655. (accessed 30 Jun 2025)

[72] Wang Y,Tang Z. DeepH-2: enhancing deep-learning electronic structure via an equivariant local-coordinate transformer. arXiv 2024, arXiv:2401.17015. https://doi.org/10.48550/arXiv.2401.17015. (accessed 30 Jun 2025)

[73] Zhouyin Z,Liu M,Zhang L. Learning local equivariant representations for quantum operators. arXiv 2024, arXiv:2407.06053. https://doi.org/10.48550/arXiv.2407.06053. (accessed 30 Jun 2025)

[74] Li Y,Huang L. Enhancing the scalability and applicability of Kohn-Sham Hamiltonians for molecular systems. arXiv 2025, arXiv:2502.19227. https://doi.org/10.48550/arXiv.2502.19227. (accessed 30 Jun 2025)

[75] Yin S,Wang F. TraceGrad: a framework learning expressive SO(3)-equivariant non-linear representations for electronic-structure Hamiltonian prediction. arXiv 2024, arXiv:2405.05722. https://doi.org/10.48550/arXiv.2405.05722. (accessed 30 Jun 2025)

[76] Yu H,Qian X,Ji S. Efficient and equivariant graph networks for predicting quantum Hamiltonian. arXiv 2023, arXiv:2306.04922. https://doi.org/10.48550/arXiv.2306.04922. (accessed 30 Jun 2025)

[77] Zhong Y,Su M,Xiang H.Transferable equivariant graph neural networks for the Hamiltonians of molecules and solids.npj Comput Mater2023;9:1130

[78] Tang Z,Lin P.A deep equivariant neural network approach for efficient hybrid density functional calculations.Nat Commun2024;15:8815 PMCID:PMC11470148

[79] Li H,Fu J.Deep-learning density functional perturbation theory.Phys Rev Lett2024;132:096401

[80] Yin S,Zhu X.Towards harmonization of SO(3)-equivariance and expressiveness: a hybrid deep learning framework for electronic-structure Hamiltonian prediction.Mach Learn Sci Technol2024;5:045038

[81] Li H,Gong X,Duan W.Deep-learning electronic-structure calculation of magnetic superstructures.Nat Comput Sci2023;3:321-7 PMCID:PMC10766521

[82] Gong X,Duan W.Generalizing deep learning electronic structure calculation to the plane-wave basis.Nat Comput Sci2024;4:752-60 PMCID:PMC11499277

[83] Wang Y,Tang Z.Universal materials model of deep-learning density functional theory Hamiltonian.Sci Bull2024;69:2514-21

[84] Tang H,He W.Approaching coupled-cluster accuracy for molecular electronic structures with multi-task learning.Nat Comput Sci2025;5:144-54

[85] Zhong Y,Zhang B.Accelerating the calculation of electron-phonon coupling strength with machine learning.Nat Comput Sci2024;4:615-25

[86] Ma Y,Zhong Y,Gong X.Transferable machine learning approach for predicting electronic structures of charged defects.Appl Phys Lett2025;126:044103

[87] Gu Q,Pandey SK,Zhang L.Deep learning tight-binding approach for large-scale electronic simulations at finite temperatures with ab initio accuracy.Nat Commun2024;15:6772 PMCID:PMC11310461

[88] Zheng H,Christensen R.The ab initio non-crystalline structure database: empowering machine learning to decode diffusivity.npj Comput Mater2024;10:1469

[89] Huang P,Faleev M.Unveiling the complex structure-property correlation of defects in 2D materials based on high throughput datasets.npj 2D Mater Appl2023;7:369

[90] Zhou Z,He Q,Li F.Machine learning guided appraisal and exploration of phase design for high entropy alloys.npj Comput Mater2019;5:265

[91] Chen Y,Wang H.DeePKS: a comprehensive data-driven approach toward chemically accurate density functional theory.J Chem Theory Comput2021;17:170-81

[92] Ou Q,Li W,Chen Y.DeePKS model for halide perovskites with the accuracy of a hybrid functional.J Phys Chem C2023;127:18755-64

[93] Kakkad J,Sharma K,Medya S. A survey on explainability of graph neural networks. arXiv 2023, arXiv:2306.01958. https://doi.org/10.48550/arXiv.2306.01958. (accessed 30 Jun 2025)

[94] Chen Y,Han B. Interpretable and generalizable graph neural networks via subgraph multilinear extension. 2024. https://openreview.net/forum?id=dVq2StlcnY. (accessed 30 Jun 2025)

[95] Miao S,Li P. Interpretable and generalizable graph learning via stochastic attention mechanism. arXiv 2022, arXiv 2201.12987. https://doi.org/10.48550/arXiv.2201.12987. (accessed 30 Jun 2025)

[96] Hou B,Qiu DY.Unsupervised representation learning of Kohn-Sham states and consequences for downstream predictions of many-body effects.Nat Commun2024;15:9481

[97] Miao S,Liu M. Interpretable geometric deep learning via learnable randomness injection. arXiv 2022, arXiv 2210.16966. https://doi.org/10.48550/arXiv.2210.16966. (accessed 30 Jun 2025)

[98] Zhang C,Zhong Y,Tao ZG.Advancing nonadiabatic molecular dynamics simulations in solids with E(3) equivariant deep neural hamiltonians.Nat Commun2025;16:2033

[99] Zhang X,Lu J,Shen L. Physics-integrated neural network for quantum transport prediction of field-effect transistor. arXiv 2024, arXiv 2408.17023. https://doi.org/10.48550/arXiv.2408.17023. (accessed 30 Jun 2025)

[100] Chen X,Liu P.Unconventional magnons in collinear magnets dictated by spin space groups.Nature2025;640:349-54 PMCID:PMC11981943

[101] Törmä P,Bernevig BA.Superconductivity, superfluidity and quantum geometry in twisted multilayer systems.Nat Rev Phys2022;4:528-42

[102] Zhao Y,Zhou J,Shin S.Effects of loosely bound electrons and electron–phonon interaction on the thermoelectric properties of electrenes.J Mater Chem C2024;12:14496-504