{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,13]],"date-time":"2026-02-13T23:16:51Z","timestamp":1771024611227,"version":"3.50.1"},"reference-count":49,"publisher":"IOP Publishing","issue":"3","license":[{"start":{"date-parts":[[2024,8,14]],"date-time":"2024-08-14T00:00:00Z","timestamp":1723593600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"},{"start":{"date-parts":[[2024,8,14]],"date-time":"2024-08-14T00:00:00Z","timestamp":1723593600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"funder":[{"DOI":"10.13039\/100016186","name":"Idaho Operations Office, U.S. Department of Energy","doi-asserted-by":"crossref","award":["DE-AC07-05ID14517"],"award-info":[{"award-number":["DE-AC07-05ID14517"]}],"id":[{"id":"10.13039\/100016186","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/100006132","name":"Office of Science","doi-asserted-by":"crossref","award":["DE-SC0016507"],"award-info":[{"award-number":["DE-SC0016507"]}],"id":[{"id":"10.13039\/100006132","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Mach. Learn.: Sci. Technol."],"published-print":{"date-parts":[[2024,9,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Machine learning approaches have recently emerged as powerful tools to probe structure-property relationships in crystals and molecules. Specifically, machine learning interatomic potentials (MLIPs) can accurately reproduce first-principles data at a cost similar to that of conventional interatomic potential approaches. While MLIPs have been extensively tested across various classes of materials and molecules, a clear characterization of the anharmonic terms encoded in the MLIPs is lacking. Here, we benchmark popular MLIPs using the anharmonic vibrational Hamiltonian of ThO<jats:sub>2<\/jats:sub> in the fluorite crystal structure, which was constructed from density functional theory (DFT) using our highly accurate and efficient irreducible derivative methods. The anharmonic Hamiltonian was used to generate molecular dynamics (MD) trajectories, which were used to train three classes of MLIPs: Gaussian approximation potentials, artificial neural networks (ANN), and graph neural networks (GNN). The results were assessed by directly comparing phonons and their interactions, as well as phonon linewidths, phonon lineshifts, and thermal conductivity. The models were also trained on a DFT MD dataset, demonstrating good agreement up to fifth-order for the ANN and GNN. Our analysis demonstrates that MLIPs have great potential for accurately characterizing anharmonicity in materials systems at a fraction of the cost of conventional first principles-based approaches.<\/jats:p>","DOI":"10.1088\/2632-2153\/ad674a","type":"journal-article","created":{"date-parts":[[2024,7,25]],"date-time":"2024-07-25T00:47:50Z","timestamp":1721868470000},"page":"030502","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":7,"title":["Benchmarking machine learning interatomic potentials via phonon anharmonicity"],"prefix":"10.1088","volume":"5","author":[{"ORCID":"https:\/\/orcid.org\/0009-0002-8854-4215","authenticated-orcid":true,"given":"Sasaank","family":"Bandi","sequence":"first","affiliation":[]},{"given":"Chao","family":"Jiang","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3812-2751","authenticated-orcid":true,"given":"Chris A","family":"Marianetti","sequence":"additional","affiliation":[]}],"member":"266","published-online":{"date-parts":[[2024,8,14]]},"reference":[{"key":"mlstad674abib1","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.100.014303","article-title":"Group theoretical approach to computing phonons and their interactions","volume":"100","author":"Fu","year":"2019","journal-title":"Phys. Rev. B"},{"key":"mlstad674abib2","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.106.014314","article-title":"Generalized quasiharmonic approximation via space group irreducible derivatives","volume":"106","author":"Mathis","year":"2022","journal-title":"Phys. Rev. B"},{"key":"mlstad674abib3","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.107.174302","article-title":"Precisely computing phonons via irreducible derivatives","volume":"107","author":"Bandi","year":"2023","journal-title":"Phys. Rev. B"},{"key":"mlstad674abib4","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.107.094303","article-title":"Anharmonic phonon behavior via irreducible derivatives: self-consistent perturbation theory and molecular dynamics","volume":"107","author":"Xiao","year":"2023","journal-title":"Phys. Rev. B"},{"key":"mlstad674abib5","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-022-29939-5","article-title":"E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials","volume":"13","author":"Batzner","year":"2022","journal-title":"Nat. Commun."},{"key":"mlstad674abib6","doi-asserted-by":"publisher","first-page":"41","DOI":"10.1038\/s41524-022-00721-x","article-title":"Representations of molecules and materials for interpolation of quantum-mechanical simulations via machine learning","volume":"8","author":"Langer","year":"2022","journal-title":"npj Comput. Mater."},{"key":"mlstad674abib7","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.108.058301","article-title":"Fast and accurate modeling of molecular atomization energies with machine learning","volume":"108","author":"Rupp","year":"2012","journal-title":"Phys. Rev. Lett."},{"key":"mlstad674abib8","doi-asserted-by":"publisher","DOI":"10.1063\/1.3553717","article-title":"Atom-centered symmetry functions for constructing high-dimensional neural network potentials","volume":"134","author":"Behler","year":"2011","journal-title":"J. Chem. Phys."},{"key":"mlstad674abib9","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.87.184115","article-title":"On representing chemical environments","volume":"87","author":"Bart\u00f3k","year":"2013","journal-title":"Phys. Rev. B"},{"key":"mlstad674abib10","doi-asserted-by":"publisher","first-page":"5255","DOI":"10.1021\/acs.jctc.7b00577","article-title":"Prediction errors of molecular machine learning models lower than hybrid dft error","volume":"13","author":"Faber","year":"2017","journal-title":"J. Chem. Theory Comput."},{"key":"mlstad674abib11","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.104.136403","article-title":"Gaussian approximation potentials: the accuracy of quantum mechanics, without the electrons","volume":"104","author":"Bart\u00f3k","year":"2010","journal-title":"Phys. Rev. Lett."},{"key":"mlstad674abib12","doi-asserted-by":"publisher","first-page":"8772","DOI":"10.1039\/D3NR00399J","article-title":"Gaussian approximation potentials for accurate thermal properties of two-dimensional materials","volume":"15","author":"Kocabas","year":"2023","journal-title":"Nanoscale"},{"key":"mlstad674abib13","doi-asserted-by":"publisher","DOI":"10.1088\/0256-307X\/37\/12\/126301","article-title":"Accuracy of machine learning potential for predictions of multiple-target physical properties","volume":"37","author":"Ouyang","year":"2020","journal-title":"Chin. Phys. Lett."},{"key":"mlstad674abib14","doi-asserted-by":"publisher","DOI":"10.1088\/1361-648X\/abcf61","article-title":"Thermal conductivity of h-bn monolayers using machine learning interatomic potential","volume":"33","author":"Zhang","year":"2020","journal-title":"J. Phys.: Condens. Matter"},{"key":"mlstad674abib15","doi-asserted-by":"publisher","DOI":"10.1016\/j.commatsci.2021.111012","article-title":"Developing machine learning potential for classical molecular dynamics simulation with superior phonon properties","volume":"202","author":"Wei","year":"2022","journal-title":"Comput. Mater. Sci."},{"key":"mlstad674abib16","doi-asserted-by":"publisher","DOI":"10.1063\/5.0069443","article-title":"Recent advances in lattice thermal conductivity calculation using machine-learning interatomic potentials","volume":"130","author":"Arabha","year":"2021","journal-title":"J. Appl. Phys."},{"key":"mlstad674abib17","doi-asserted-by":"publisher","first-page":"15","DOI":"10.1016\/0927-0256(96)00008-0","article-title":"Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set","volume":"6","author":"Kresse","year":"1996","journal-title":"Comput. Mater. Sci."},{"key":"mlstad674abib18","doi-asserted-by":"publisher","first-page":"558","DOI":"10.1103\/PhysRevB.47.558","article-title":"Ab initiomolecular dynamics for liquid metals","volume":"47","author":"Kresse","year":"1993","journal-title":"Phys. Rev. B"},{"key":"mlstad674abib19","doi-asserted-by":"publisher","DOI":"10.1088\/1361-648X\/ac066b","article-title":"The stochastic self-consistent harmonic approximation: calculating vibrational properties of materials with full quantum and anharmonic effects","volume":"33","author":"Monacelli","year":"2021","journal-title":"J. Phys.: Condens. Matter"},{"key":"mlstad674abib20","doi-asserted-by":"publisher","DOI":"10.1002\/qute.202200131","article-title":"Temperature dependent anharmonic phonons in quantum paraelectric ktao3 by first principles and machine learned force fields","volume":"6","author":"Ranalli","year":"2023","journal-title":"Adv. Quantum Technol."},{"key":"mlstad674abib21","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevMaterials.7.L030801","article-title":"Quantum paraelectricity and structural phase transitions in strontium titanate beyond density functional theory","volume":"7","author":"Verdi","year":"2023","journal-title":"Phys. Rev. Mater."},{"key":"mlstad674abib22","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.113.185501","article-title":"Lattice anharmonicity and thermal conductivity from compressive sensing of first-principles calculations","volume":"113","author":"Zhou","year":"2014","journal-title":"Phys. Rev. Lett."},{"key":"mlstad674abib23","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.100.184308","article-title":"Compressive sensing lattice dynamics. I. general formalism","volume":"100","author":"Zhou","year":"2019","journal-title":"Phys. Rev. B"},{"key":"mlstad674abib24","doi-asserted-by":"publisher","first-page":"273","DOI":"10.1111\/j.1467-9868.2011.00771.x","article-title":"Regression shrinkage and selection via the lasso: a retrospective","volume":"73","author":"Tibshirani","year":"1996","journal-title":"J. R. Stat. Soc. B"},{"key":"mlstad674abib25","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.92.054301","article-title":"Self-consistent phonon calculations of lattice dynamical properties in cubic SrTio3 with first-principles anharmonic force constants","volume":"92","author":"Tadano","year":"2015","journal-title":"Phys. Rev. B"},{"key":"mlstad674abib26","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.98.146401","article-title":"Generalized neural-network representation of high-dimensional potential-energy surfaces","volume":"98","author":"Behler","year":"2007","journal-title":"Phys. Rev. Lett."},{"key":"mlstad674abib27","article-title":"e3nn: euclidean neural networks","author":"Geiger","year":"2022"},{"key":"mlstad674abib28","doi-asserted-by":"publisher","DOI":"10.1016\/j.mtphys.2020.100181","article-title":"A unified deep neural network potential capable of predicting thermal conductivity of silicon in different phases","volume":"12","author":"Li","year":"2020a","journal-title":"Mater. Today Phys."},{"key":"mlstad674abib29","doi-asserted-by":"publisher","DOI":"10.1063\/5.0025051","article-title":"A deep neural network interatomic potential for studying thermal conductivity of \u03b2 -Ga2O3","volume":"117","author":"Li","year":"2020","journal-title":"Appl. Phys. Lett."},{"key":"mlstad674abib30","doi-asserted-by":"publisher","DOI":"10.1063\/5.0009550","article-title":"Transferability of neural network potentials for varying stoichiometry: phonons and thermal conductivity of MnxGey compounds","volume":"127","author":"Mangold","year":"2020","journal-title":"J. Appl. Phys."},{"key":"mlstad674abib31","doi-asserted-by":"publisher","DOI":"10.7567\/1882-0786\/ab36bc","article-title":"Simulating lattice thermal conductivity in semiconducting materials using high-dimensional neural network potential","volume":"12","author":"Minamitani","year":"2019","journal-title":"Appl. Phys. Express"},{"key":"mlstad674abib32","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevMaterials.3.074603","article-title":"Machine-learning-based interatomic potential for phonon transport in perfect crystalline si and crystalline si with vacancies","volume":"3","author":"Babaei","year":"2019","journal-title":"Phys. Rev. Mater."},{"key":"mlstad674abib33","doi-asserted-by":"publisher","DOI":"10.1016\/j.mtphys.2019.100140","article-title":"Thermal conductivity modeling using machine learning potentials: application to crystalline and amorphous silicon","volume":"10","author":"Qian","year":"2019","journal-title":"Mater. Today Phys."},{"key":"mlstad674abib34","doi-asserted-by":"publisher","DOI":"10.1016\/j.commatsci.2022.111472","article-title":"Accelerated computation of lattice thermal conductivity using neural network interatomic potentials","volume":"211","author":"Choi","year":"2022","journal-title":"Comput. Mater. Sci."},{"key":"mlstad674abib35","doi-asserted-by":"publisher","first-page":"4475","DOI":"10.1007\/s11664-023-10403-z","article-title":"A deep neural network potential to study the thermal conductivity of MnBi2Te4 and Bi2Te3\/MnBi2Te4 superlattice","volume":"52","author":"Qu","year":"2023","journal-title":"J. Electron. Mater."},{"key":"mlstad674abib36","doi-asserted-by":"publisher","DOI":"10.1016\/j.commatsci.2022.111280","article-title":"Teanet: universal neural network interatomic potential inspired by iterative electronic relaxations","volume":"207","author":"Takamoto","year":"2022","journal-title":"Comput. Mater. Sci."},{"key":"mlstad674abib37","article-title":"Mace: higher order equivariant message passing neural networks for fast and accurate force fields","author":"Batatia","year":"2022"},{"key":"mlstad674abib38","article-title":"Directional message passing for molecular graphs","author":"Gasteiger","year":"2020"},{"key":"mlstad674abib39","doi-asserted-by":"publisher","first-page":"718","DOI":"10.1038\/s43588-022-00349-3","article-title":"A universal graph deep learning interatomic potential for the periodic table","volume":"2","author":"Chen","year":"2022","journal-title":"Nat. Comput. Sci."},{"key":"mlstad674abib40","doi-asserted-by":"publisher","first-page":"1031","DOI":"10.1038\/s42256-023-00716-3","article-title":"Chgnet as a pretrained universal neural network potential for charge-informed atomistic modelling","volume":"5","author":"Deng","year":"2023","journal-title":"Nat. Mach. Intell."},{"key":"mlstad674abib41","doi-asserted-by":"publisher","DOI":"10.1063\/1.4812323","article-title":"Commentary: the materials project: a materials genome approach to accelerating materials innovation","volume":"1","author":"Jain","year":"2013","journal-title":"APL Mater."},{"key":"mlstad674abib42","doi-asserted-by":"publisher","DOI":"10.1016\/j.egyai.2023.100286","article-title":"Screening outstanding mechanical properties and low lattice thermal conductivity using global attention graph neural network","volume":"14","author":"Ojih","year":"2023","journal-title":"Energy AI"},{"key":"mlstad674abib43","doi-asserted-by":"publisher","first-page":"474","DOI":"10.1002\/smm2.1074","article-title":"Novel machine learning framework for thermal conductivity prediction by crystal graph convolution embedded ensemble","volume":"3","author":"Sun","year":"2021","journal-title":"SmartMat"},{"key":"mlstad674abib44","doi-asserted-by":"publisher","first-page":"21","DOI":"10.1038\/s41524-024-01207-8","article-title":"Machine learning potential assisted exploration of complex defect potential energy surfaces","volume":"10","author":"Jiang","year":"2024","journal-title":"npj Comput. Mater."},{"key":"mlstad674abib45","article-title":"Expressive programming for computational physics in fortran 95+","author":"Cs\u00e1nyi","year":"2007"},{"key":"mlstad674abib46","doi-asserted-by":"publisher","DOI":"10.1088\/1361-648X\/ab82d2","article-title":"f90wrap: an automated tool for constructing deep python interfaces to modern fortran codes","volume":"32","author":"Kermode","year":"2020","journal-title":"J. Phys.: Condens. Matter"},{"key":"mlstad674abib47","doi-asserted-by":"publisher","first-page":"1827","DOI":"10.1021\/acs.jctc.8b00770","article-title":"Library-based lammps implementation of high-dimensional neural network potentials","volume":"15","author":"Singraber","year":"2019","journal-title":"J. Chem. Theory Comput."},{"key":"mlstad674abib48","article-title":"See Supplemental Material at [URL will be inserted by publisher] for machine learning model details, more computational details for the calculation of observables, and additional linewidth, lineshift, and thermal conductivity plots."},{"key":"mlstad674abib49","article-title":"Supplemental data for benchmarking machine learning interatomic potentials via phonon anharmonicity","author":"Bandi","year":"2024"}],"container-title":["Machine Learning: Science and Technology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad674a","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad674a\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad674a","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad674a\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad674a\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad674a\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad674a\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad674a\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,8,14]],"date-time":"2024-08-14T08:46:55Z","timestamp":1723625215000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad674a"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,8,14]]},"references-count":49,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2024,8,14]]},"published-print":{"date-parts":[[2024,9,1]]}},"URL":"https:\/\/doi.org\/10.1088\/2632-2153\/ad674a","relation":{},"ISSN":["2632-2153"],"issn-type":[{"value":"2632-2153","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,8,14]]},"assertion":[{"value":"Benchmarking machine learning interatomic potentials via phonon anharmonicity","name":"article_title","label":"Article Title"},{"value":"Machine Learning: Science and Technology","name":"journal_title","label":"Journal Title"},{"value":"paper","name":"article_type","label":"Article Type"},{"value":"\u00a9 2024 The Author(s). Published by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2024-04-04","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2024-07-18","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2024-08-14","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}