Fast Bayesian force fields from active learning: study of inter-dimensional transformation of stanene
JSON Export
{
"updated": "2021-03-21T16:48:39.854852+00:00",
"id": "645",
"metadata": {
"id": "645",
"status": "published",
"_files": [
{
"description": "README",
"size": 1083,
"key": "README.txt",
"checksum": "md5:c78306356a58379e40453bc94fa0cfb4"
},
{
"description": "Python code of this method, including building spline interpolations and Bayesian active learning",
"size": 96386,
"key": "flare.zip",
"checksum": "md5:62cbb48b16b2703d0eb31dfc7cc4341a"
},
{
"description": "LAMMPS source code of MGP pair style",
"size": 3556897,
"key": "LMP.zip",
"checksum": "md5:3177c52a24b94c6bc2edfeda19e1e217"
},
{
"description": "Training and testing data of stanene and bulk tin systems",
"size": 471254,
"key": "Data.zip",
"checksum": "md5:4dd1cbfc0b81b93cb5a4877e94f5a59a"
},
{
"description": "GP models and MGP coefficient files of LAMMPS for stanene and bulk systems",
"size": 404507479,
"key": "Models.zip",
"checksum": "md5:9d36b692a81ba8cd4338f6ccf1390440"
},
{
"description": "LAMMPS molecular dynamics of stanene phase transition and ab-initio molecular dynamics of liquid tin",
"size": 239747413,
"key": "MD.zip",
"checksum": "md5:8ab5b2285543e668ade1bd07b6e77fb7"
}
],
"contributors": [
{
"givennames": "Yu",
"familyname": "Xie",
"affiliations": [
"John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, USA"
],
"email": "xiey@g.harvard.edu"
},
{
"givennames": "Jonathan",
"familyname": "Vandermause",
"affiliations": [
"John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, USA"
],
"email": "jonathan_vandermause@g.harvard.edu"
},
{
"givennames": "Lixin",
"familyname": "Sun",
"affiliations": [
"John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, USA"
],
"email": "lixinsun@g.harvard.edu"
},
{
"givennames": "Andrea",
"familyname": "Cepellotti",
"affiliations": [
"John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, USA"
],
"email": "acepellotti@seas.harvard.edu"
},
{
"givennames": "Boris",
"familyname": "Kozinsky",
"affiliations": [
"John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, USA",
"Robert Bosch LLC, Research and Technology Center, Cambridge, Massachusetts 02142, USA"
],
"email": "bkoz@g.harvard.edu"
}
],
"conceptrecid": "470",
"doi": "10.24435/materialscloud:qg-99",
"references": [
{
"url": "https://www.nature.com/articles/s41524-021-00510-y",
"citation": "Xie, Y., Vandermause, J., Sun, L. et al. Bayesian force fields from active learning for simulation of inter-dimensional transformation of stanene. npj Comput Mater 7, 40 (2021).",
"comment": "Journal article in which the method is described and the data is discussed",
"type": "Journal reference",
"doi": "10.1038/s41524-021-00510-y"
},
{
"url": "https://www.nature.com/articles/s41524-020-0283-z",
"citation": "J. Vandermause, S.B. Torrisi, S. Batzner, Y. Xie, L. Sun, A.M. Kolpak, B. Kozinsky, npj Comput Mater 6, 20 (2020).",
"comment": "Paper and code on which this method is based",
"type": "Journal reference",
"doi": "10.1038/s41524-020-0283-z"
},
{
"url": "https://github.com/mir-group/flare",
"citation": "J. Vandermause, S.B. Torrisi, S. Batzner, Y. Xie, L. Sun, A.M. Kolpak, B. Kozinsky, npj Comput Mater 6, 20 (2020).",
"comment": "Code in which the method is implemented",
"type": "Software"
}
],
"title": "Fast Bayesian force fields from active learning: study of inter-dimensional transformation of stanene",
"publication_date": "Dec 01, 2020, 20:20:31",
"description": "Gaussian process (GP) regression is one promising technique of constructing machine learning force fields with built-in uncertainty quantification, which can be used to monitor the quality of model predictions. A current limitation of existing GP force fields is that the prediction cost grows linearly with the size of the training data set, making accurate GP predictions slow. \nIn this work, we exploit the special structure of the kernel function to construct a mapping of the trained Gaussian process model, including both forces and their uncertainty predictions, onto spline functions of low-dimensional structural features. This method is incorporated in the Bayesian active learning workflow for training of Bayesian force fields. \nTo demonstrate the capabilities of this method, we construct a force field for stanene and perform large scale dynamics simulation of its structural evolution. \nWe provide a fully open-source implementation of our method, as well as the training and testing examples with the stanene dataset.",
"mcid": "2020.155",
"edited_by": 162,
"version": 3,
"is_last": true,
"owner": 162,
"license_addendum": "",
"keywords": [
"machine learning",
"molecular dynamics",
"Bayesian inference",
"stanene",
"phase transition"
],
"_oai": {
"id": "oai:materialscloud.org:645"
},
"license": "Creative Commons Attribution 4.0 International"
},
"revision": 10,
"created": "2020-11-20T03:04:58.057822+00:00"
}