{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,7,8]],"date-time":"2025-07-08T07:47:08Z","timestamp":1751960828749,"version":"3.37.3"},"reference-count":33,"publisher":"IOP Publishing","issue":"4","license":[{"start":{"date-parts":[[2024,11,12]],"date-time":"2024-11-12T00:00:00Z","timestamp":1731369600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"},{"start":{"date-parts":[[2024,11,12]],"date-time":"2024-11-12T00:00:00Z","timestamp":1731369600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"funder":[{"name":"Strategic Priority Research Program of Chinese Academy of Sciences","award":["XDB0500201"],"award-info":[{"award-number":["XDB0500201"]}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"crossref","award":["12134012"],"award-info":[{"award-number":["12134012"]}],"id":[{"id":"10.13039\/501100001809","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":[[2024,12,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Deep learning for predicting the electronic-structure Hamiltonian of quantum systems necessitates satisfying the covariance laws, among which achieving SO(3)-equivariance without sacrificing the non-linear expressive capability of networks remains unsolved. To navigate the harmonization between SO(3)-equivariance and expressiveness, we propose HarmoSE, a deep learning method synergizing two distinct categories of neural mechanisms as a two-stage encoding and regression framework. The first stage corresponds to group theory-based neural mechanisms with inherent SO(3)-equivariant properties prior to the parameter learning process, while the second stage is characterized by a non-linear 3D graph Transformer network we propose, featuring high capability on non-linear expressiveness. Their combination lies in the point that, the first stage predicts baseline Hamiltonians with abundant SO(3)-equivariant features extracted, assisting the second stage in empirical learning of equivariance; and in turn, the second stage refines the first stage\u2019s output as a fine-grained prediction of Hamiltonians using powerful non-linear neural mappings, compensating for the intrinsic weakness on non-linear expressiveness capability of mechanisms in the first stage. Our method enables precise, generalizable predictions while capturing SO(3)-equivariance under rotational transformations, and achieves state-of-the-art performance in Hamiltonian prediction tasks under multiple mean absolute error (MAE) metrics, such as the average MAE across all samples and matrix elements, the MAE for challenging samples, the MAE for different Hamiltonian blocks, and the MAE for the challenging blocks. It also demonstrates significant improvements in accuracy for downstream quantities, such as occupied orbital energy and the electronic wavefunction, as measured by MAE and cosine similarity, respectively.<\/jats:p>","DOI":"10.1088\/2632-2153\/ad8d30","type":"journal-article","created":{"date-parts":[[2024,10,30]],"date-time":"2024-10-30T22:58:00Z","timestamp":1730329080000},"page":"045038","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":2,"title":["Towards harmonization of SO(3)-equivariance and expressiveness: a hybrid deep learning framework for electronic-structure Hamiltonian prediction"],"prefix":"10.1088","volume":"5","author":[{"ORCID":"https:\/\/orcid.org\/0009-0006-5459-5259","authenticated-orcid":true,"given":"Shi","family":"Yin","sequence":"first","affiliation":[]},{"given":"Xinyang","family":"Pan","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2006-7109","authenticated-orcid":true,"given":"Xudong","family":"Zhu","sequence":"additional","affiliation":[]},{"given":"Tianyu","family":"Gao","sequence":"additional","affiliation":[]},{"given":"Haochong","family":"Zhang","sequence":"additional","affiliation":[]},{"given":"Feng","family":"Wu","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2050-134X","authenticated-orcid":true,"given":"Lixin","family":"He","sequence":"additional","affiliation":[]}],"member":"266","published-online":{"date-parts":[[2024,11,12]]},"reference":[{"key":"mlstad8d30bib1","doi-asserted-by":"publisher","first-page":"5024","DOI":"10.1038\/s41467-019-12875-2","article-title":"Unifying machine learning and quantum chemistry with a deep neural network for molecular wavefunctions","volume":"10","author":"Sch\u00fctt","year":"2019","journal-title":"Nat. Commun."},{"key":"mlstad8d30bib2","first-page":"pp 14434","article-title":"Se(3)-equivariant prediction of molecular wavefunctions and electronic densities","volume":"vol 34","author":"Unke","year":"2021"},{"key":"mlstad8d30bib3","doi-asserted-by":"publisher","first-page":"29","DOI":"10.1016\/j.scib.2021.09.010","article-title":"Neural network representation of electronic structure from ab initio molecular dynamics","volume":"67","author":"Gu","year":"2022","journal-title":"Sci. Bull."},{"key":"mlstad8d30bib4","doi-asserted-by":"publisher","first-page":"367","DOI":"10.1038\/s43588-022-00265-6","article-title":"Deep-learning density functional theory Hamiltonian for efficient ab initio electronic-structure calculation","volume":"2","author":"Li","year":"2022","journal-title":"Nat. Comput. Sci."},{"key":"mlstad8d30bib5","doi-asserted-by":"publisher","first-page":"182","DOI":"10.1038\/s41524-023-01130-4","article-title":"Transferable equivariant graph neural networks for the hamiltonians of molecules and solids","volume":"9","author":"Zhong","year":"2023","journal-title":"npj Comput. Mater."},{"key":"mlstad8d30bib6","doi-asserted-by":"publisher","first-page":"2848","DOI":"10.1038\/s41467-023-38468-8","article-title":"General framework for E(3)-equivariant neural network representation of density functional theory hamiltonian","volume":"14","author":"Gong","year":"2023","journal-title":"Nat. Commun."},{"key":"mlstad8d30bib7","doi-asserted-by":"publisher","first-page":"B864","DOI":"10.1103\/PhysRev.136.B864","article-title":"Inhomogeneous electron gas","volume":"136","author":"Hohenberg","year":"1964","journal-title":"Phys. Rev."},{"key":"mlstad8d30bib8","doi-asserted-by":"publisher","first-page":"A1133","DOI":"10.1103\/PhysRev.140.A1133","article-title":"Self-consistent equations including exchange and correlation effects","volume":"140","author":"Kohn","year":"1965","journal-title":"Phys. Rev."},{"article-title":"Artificial intelligence for science in quantum, atomistic, and continuum systems","year":"2023","author":"Zhang","key":"mlstad8d30bib9"},{"article-title":"e3nn: Euclidean neural networks","year":"2022","author":"Geiger","key":"mlstad8d30bib10"},{"article-title":"Spherical channels for modeling atomic interactions","year":"2022","author":"Zitnick","key":"mlstad8d30bib11"},{"key":"mlstad8d30bib12","first-page":"pp 40412","article-title":"Efficient and equivariant graph networks for predicting quantum hamiltonian","author":"Yu","year":"2023"},{"key":"mlstad8d30bib13","first-page":"pp 1889","article-title":"Exploiting cyclic symmetry in convolutional neural networks","volume":"vol 48","author":"Dieleman","year":"2016"},{"key":"mlstad8d30bib14","first-page":"pp 2892","article-title":"Equivariance through parameter-sharing","author":"Ravanbakhsh","year":"2017"},{"article-title":"Covariant compositional networks for learning graphs","year":"2018","author":"Kondor","key":"mlstad8d30bib15"},{"key":"mlstad8d30bib16","first-page":"pp 27420","article-title":"Reducing SO(3) convolutions to SO(2) for efficient equivariant gnns","author":"Passaro","year":"2023"},{"article-title":"Equiformerv2: improved equivariant transformer for scaling to higher-degree representations","year":"2024","author":"Liao","key":"mlstad8d30bib17"},{"key":"mlstad8d30bib18","doi-asserted-by":"publisher","first-page":"2514","DOI":"10.1016\/j.scib.2024.06.011","article-title":"Universal materials model of deep-learning density functional theory hamiltonian","volume":"69","author":"Wang","year":"2024","journal-title":"Sci. Bull."},{"key":"mlstad8d30bib19","first-page":"pp 2017","article-title":"Spatial transformer networks","author":"Jaderberg","year":"2015"},{"article-title":"Steerable CNNS","year":"2017","author":"Cohen","key":"mlstad8d30bib20"},{"article-title":"Tensor field networks: rotation-and translation-equivariant neural networks for 3D point clouds","year":"2018","author":"Thomas","key":"mlstad8d30bib21"},{"article-title":"Se(3)-transformers: 3D roto-translation equivariant attention networks","year":"2020","author":"Fuchs","key":"mlstad8d30bib22"},{"article-title":"Equiformer: Equivariant graph attention transformer for 3d atomistic graphs","year":"2023","author":"Liao","key":"mlstad8d30bib23"},{"article-title":"QH9: a quantum Hamiltonian prediction benchmark for QM9 molecules","year":"2023","author":"Yu","key":"mlstad8d30bib24"},{"article-title":"Infusing self-consistency into density functional theory hamiltonian prediction via deep equilibrium models","year":"2024","author":"Wang","key":"mlstad8d30bib25"},{"article-title":"Self-consistency training for density-functional-theory hamiltonian prediction","year":"2024","author":"Zhang","key":"mlstad8d30bib26"},{"key":"mlstad8d30bib27","doi-asserted-by":"publisher","first-page":"361","DOI":"10.1002\/andp.19263840404","article-title":"Quantisierung als eigenwertproblem","volume":"384","author":"Schr\u00f6dinger","year":"1926","journal-title":"Ann. Phys., Lpz."},{"key":"mlstad8d30bib28","doi-asserted-by":"publisher","first-page":"178","DOI":"10.1016\/j.cpc.2018.03.016","article-title":"Deepmd-kit: a deep learning package for many-body potential energy representation and molecular dynamics","volume":"228","author":"Wang","year":"2018","journal-title":"Comput. Phys. Commun."},{"key":"mlstad8d30bib29","doi-asserted-by":"publisher","first-page":"4962","DOI":"10.1021\/acs.jpclett.9b02037","article-title":"Embedded atom neural network potentials: efficient and accurate machine learning with a physically inspired representation","volume":"10","author":"Zhang","year":"2019","journal-title":"J. Phys. Chem. Lett."},{"key":"mlstad8d30bib30","doi-asserted-by":"publisher","first-page":"6424","DOI":"10.1038\/s41467-023-42148-y","article-title":"universal machine learning for the response of atomistic systems to external fields","volume":"14","author":"Zhang","year":"2023","journal-title":"Nat. Commun."},{"key":"mlstad8d30bib31","doi-asserted-by":"publisher","first-page":"43","DOI":"10.1038\/nature26160","article-title":"Unconventional superconductivity in magic-angle graphene superlattices","volume":"556","author":"Cao","year":"2018","journal-title":"Nature"},{"key":"mlstad8d30bib32","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.132.036501","article-title":"Fractional chern insulator in twisted bilayer MoTe2","volume":"132","author":"Wang","year":"2024","journal-title":"Phys. Rev. Lett."},{"key":"mlstad8d30bib33","doi-asserted-by":"publisher","first-page":"224","DOI":"10.1038\/s41563-023-01713-y","article-title":"Dynamically tunable moir\u00e9 exciton rydberg states in a monolayer semiconductor on twisted bilayer graphene","volume":"23","author":"He","year":"2024","journal-title":"Nat. Mater."}],"container-title":["Machine Learning: Science and Technology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad8d30","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad8d30\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad8d30","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad8d30\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad8d30\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad8d30\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad8d30\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad8d30\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,11,12]],"date-time":"2024-11-12T13:45:48Z","timestamp":1731419148000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad8d30"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,11,12]]},"references-count":33,"journal-issue":{"issue":"4","published-online":{"date-parts":[[2024,11,12]]},"published-print":{"date-parts":[[2024,12,1]]}},"URL":"https:\/\/doi.org\/10.1088\/2632-2153\/ad8d30","relation":{},"ISSN":["2632-2153"],"issn-type":[{"type":"electronic","value":"2632-2153"}],"subject":[],"published":{"date-parts":[[2024,11,12]]},"assertion":[{"value":"Towards harmonization of SO(3)-equivariance and expressiveness: a hybrid deep learning framework for electronic-structure Hamiltonian prediction","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":"2024-06-21","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2024-10-30","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2024-11-12","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}