Unified theory of atom-centered representations and message-passing machine-learning schemes
JSON Export
{
"metadata": {
"edited_by": 576,
"owner": 150,
"description": "Data-driven schemes that associate molecular and crystal structures with their microscopic properties share the need for a concise, effective description of the arrangement of their atomic constituents. Many types of models rely on descriptions of atom-centered environments, that are associated with an atomic property or with an atomic contribution to an extensive macroscopic quantity. Frameworks in this class can be understood in terms of atom-centered density correlations (ACDC), that are used as a basis for a body-ordered, symmetry-adapted expansion of the targets. Several other schemes, that gather information on the relationship between neighboring atoms using \"message-passing\" ideas, cannot be directly mapped to correlations centered around a single atom. We generalize the ACDC framework to include multi-centered information, generating representations that provide a complete linear basis to regress symmetric functions of atomic coordinates, and provides a coherent foundation to systematize our understanding of both atom-centered and message-passing, invariant and equivariant machine-learning schemes.\n\nThis record contains the data and code required to reproduce the results from the corresponding paper, computing message-passing inspired machine learning features built on top of density correlation. The data used in this article is a subset of other existing datasets, which can be found online:\n\n- methane dataset: https://archive.materialscloud.org/record/2020.105\n- NaCl dataset: https://github.com/dilkins/TENSOAP/tree/ea671154b3642b4ec879a4292a4dd4399ddbdea6/example/random_nacl\n- QM7 and QM9 with dipole moments: https://archive.materialscloud.org/record/2020.56",
"keywords": [
"machine learning",
"message passing",
"reproducibility",
"MARVEL/DD2",
"PASC"
],
"is_last": true,
"title": "Unified theory of atom-centered representations and message-passing machine-learning schemes",
"status": "published",
"license_addendum": null,
"doi": "10.24435/materialscloud:3f-g3",
"conceptrecid": "1293",
"_files": [
{
"checksum": "md5:70d7f45c49b725a8f95a3e7c763a291c",
"description": "Extended XYZ file containing the methane structures and energies",
"key": "methane-40k.xyz",
"size": 13648472
},
{
"checksum": "md5:f8a03c12c20114ef88237f61c725560d",
"description": "Scripts used to compute message passing features and use them to train models for methane",
"key": "methane.zip",
"size": 6652
},
{
"checksum": "md5:9e76b4a8972a2e829335163fe31c7d7c",
"description": "Extended XYZ file containing the random NaCl structures and energies",
"key": "random-nacl.xyz",
"size": 7154931
},
{
"checksum": "md5:e03a06c9b6c4657590280d78797cc14d",
"description": "Scripts used to compute message passing features and use them to train models for NaCl",
"key": "nacl.zip",
"size": 7703
},
{
"checksum": "md5:6b437f8c3ad640ac3676c8e43d98e965",
"description": "Extended XYZ file containing the subset of QM7 used in this study",
"key": "qm7_shuffled_chno.xyz",
"size": 8697956
},
{
"checksum": "md5:a44c0931c24788625daa60ce1afea85b",
"description": "Dipole moments of QM7 structures in numpy format",
"key": "qm7_chno_dipole.npy",
"size": 165104
},
{
"checksum": "md5:97406dd69fd0ec945fcadf432a340051",
"description": "Extended XYZ file containing the subset of QM9 used in this study",
"key": "qm9_1000_test.xyz",
"size": 1101053
},
{
"checksum": "md5:c90dc2576831316f43df392bfcf91af7",
"description": "Dipole moments of QM9 structures in numpy format",
"key": "qm9_dipole.npy",
"size": 24104
},
{
"checksum": "md5:2966a6260def1b073e2a3c90b6e354f1",
"description": "Scripts used to compute message passing features and use them to train models for QM7/QM9",
"key": "qm7_dipole.zip",
"size": 6573
}
],
"references": [
{
"citation": "J. Nigam, S. Pozdnyakov, G. Fraux, and M. Ceriotti, arXiv:2202.01566 [stat.ML]",
"comment": "Preprint where this data is used as examples for the message passing density correlation features",
"url": "https://arxiv.org/abs/2202.01566",
"doi": "10.48550/arXiv.2202.01566",
"type": "Preprint"
}
],
"contributors": [
{
"givennames": "Jigyasa",
"affiliations": [
"Laboratory of Computational Science and Modelling, Institute of Materials, \u00c9cole Polytechnique F\u00e9d\u00e9rale de Lausanne, 1015 Lausanne, Switzerland",
"National Centre for Computational Design and Discovery of Novel Materials (MARVEL), \u00c9cole Polytechnique F\u00e9d\u00e9rale de Lausanne, 1015 Lausanne, Switzerland"
],
"familyname": "Nigam",
"email": "jigyasa.nigam@epfl.ch"
},
{
"givennames": "Sergey",
"affiliations": [
"Laboratory of Computational Science and Modelling, Institute of Materials, \u00c9cole Polytechnique F\u00e9d\u00e9rale de Lausanne, 1015 Lausanne, Switzerland"
],
"familyname": "Pozdnyakov",
"email": "sergey.pozdnyakov@epfl.ch"
},
{
"givennames": "Guillaume",
"affiliations": [
"Laboratory of Computational Science and Modelling, Institute of Materials, \u00c9cole Polytechnique F\u00e9d\u00e9rale de Lausanne, 1015 Lausanne, Switzerland"
],
"familyname": "Fraux",
"email": "guillaume.fraux@epfl.ch"
},
{
"givennames": "Michele",
"affiliations": [
"Laboratory of Computational Science and Modelling, Institute of Materials, \u00c9cole Polytechnique F\u00e9d\u00e9rale de Lausanne, 1015 Lausanne, Switzerland",
"National Centre for Computational Design and Discovery of Novel Materials (MARVEL), \u00c9cole Polytechnique F\u00e9d\u00e9rale de Lausanne, 1015 Lausanne, Switzerland"
],
"familyname": "Ceriotti",
"email": "michele.ceriotti@epfl.ch"
}
],
"_oai": {
"id": "oai:materialscloud.org:1294"
},
"publication_date": "Mar 24, 2022, 15:00:55",
"mcid": "2022.44",
"version": 1,
"id": "1294",
"license": "Creative Commons Attribution 4.0 International"
},
"revision": 8,
"created": "2022-03-21T10:14:13.582721+00:00",
"id": "1294",
"updated": "2022-03-24T14:00:55.332005+00:00"
}