Size and Quality of Quantum Mechanical Data Set for Training Neural Network Force Fields for Liquid Water
Autor(a) principal: | |
---|---|
Data de Publicação: | 2023 |
Outros Autores: | , , |
Tipo de documento: | Artigo |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UNESP |
Texto Completo: | http://dx.doi.org/10.1021/acs.jpcb.2c09059 http://hdl.handle.net/11449/248312 |
Resumo: | Molecular dynamics simulations have been used in different scientific fields to investigate a broad range of physical systems. However, the accuracy of calculation is based on the model considered to describe the atomic interactions. In particular, ab initio molecular dynamics (AIMD) has the accuracy of density functional theory (DFT) and thus is limited to small systems and a relatively short simulation time. In this scenario, Neural Network Force Fields (NNFFs) have an important role, since they provide a way to circumvent these caveats. In this work, we investigate NNFFs designed at the level of DFT to describe liquid water, focusing on the size and quality of the training data set considered. We show that structural properties are less dependent on the size of the training data set compared to dynamical ones (such as the diffusion coefficient), and a good sampling (selecting data reference for the training process) can lead to a small sample with good precision. |
id |
UNSP_ae1ca8c32d53c20251babcad52296013 |
---|---|
oai_identifier_str |
oai:repositorio.unesp.br:11449/248312 |
network_acronym_str |
UNSP |
network_name_str |
Repositório Institucional da UNESP |
repository_id_str |
2946 |
spelling |
Size and Quality of Quantum Mechanical Data Set for Training Neural Network Force Fields for Liquid WaterMolecular dynamics simulations have been used in different scientific fields to investigate a broad range of physical systems. However, the accuracy of calculation is based on the model considered to describe the atomic interactions. In particular, ab initio molecular dynamics (AIMD) has the accuracy of density functional theory (DFT) and thus is limited to small systems and a relatively short simulation time. In this scenario, Neural Network Force Fields (NNFFs) have an important role, since they provide a way to circumvent these caveats. In this work, we investigate NNFFs designed at the level of DFT to describe liquid water, focusing on the size and quality of the training data set considered. We show that structural properties are less dependent on the size of the training data set compared to dynamical ones (such as the diffusion coefficient), and a good sampling (selecting data reference for the training process) can lead to a small sample with good precision.Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)Centro de Ciências Naturais e Humanas Universidade Federal do ABC, São PauloInstituto de Física Universidade de São PauloInstitute of Theoretical Physics São Paulo State UniversityInstitute of Theoretical Physics São Paulo State UniversityUniversidade Federal do ABC (UFABC)Universidade de São Paulo (USP)Universidade Estadual Paulista (UNESP)Gomes-Filho, Márcio S.Torres, AlbertoReily Rocha, Alexandre [UNESP]Pedroza, Luana S.2023-07-29T13:40:22Z2023-07-29T13:40:22Z2023-02-16info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/article1422-1428http://dx.doi.org/10.1021/acs.jpcb.2c09059Journal of Physical Chemistry B, v. 127, n. 6, p. 1422-1428, 2023.1520-52071520-6106http://hdl.handle.net/11449/24831210.1021/acs.jpcb.2c090592-s2.0-85147507003Scopusreponame:Repositório Institucional da UNESPinstname:Universidade Estadual Paulista (UNESP)instacron:UNESPengJournal of Physical Chemistry Binfo:eu-repo/semantics/openAccess2023-07-29T13:40:22Zoai:repositorio.unesp.br:11449/248312Repositório InstitucionalPUBhttp://repositorio.unesp.br/oai/requestopendoar:29462024-08-05T19:18:29.556503Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)false |
dc.title.none.fl_str_mv |
Size and Quality of Quantum Mechanical Data Set for Training Neural Network Force Fields for Liquid Water |
title |
Size and Quality of Quantum Mechanical Data Set for Training Neural Network Force Fields for Liquid Water |
spellingShingle |
Size and Quality of Quantum Mechanical Data Set for Training Neural Network Force Fields for Liquid Water Gomes-Filho, Márcio S. |
title_short |
Size and Quality of Quantum Mechanical Data Set for Training Neural Network Force Fields for Liquid Water |
title_full |
Size and Quality of Quantum Mechanical Data Set for Training Neural Network Force Fields for Liquid Water |
title_fullStr |
Size and Quality of Quantum Mechanical Data Set for Training Neural Network Force Fields for Liquid Water |
title_full_unstemmed |
Size and Quality of Quantum Mechanical Data Set for Training Neural Network Force Fields for Liquid Water |
title_sort |
Size and Quality of Quantum Mechanical Data Set for Training Neural Network Force Fields for Liquid Water |
author |
Gomes-Filho, Márcio S. |
author_facet |
Gomes-Filho, Márcio S. Torres, Alberto Reily Rocha, Alexandre [UNESP] Pedroza, Luana S. |
author_role |
author |
author2 |
Torres, Alberto Reily Rocha, Alexandre [UNESP] Pedroza, Luana S. |
author2_role |
author author author |
dc.contributor.none.fl_str_mv |
Universidade Federal do ABC (UFABC) Universidade de São Paulo (USP) Universidade Estadual Paulista (UNESP) |
dc.contributor.author.fl_str_mv |
Gomes-Filho, Márcio S. Torres, Alberto Reily Rocha, Alexandre [UNESP] Pedroza, Luana S. |
description |
Molecular dynamics simulations have been used in different scientific fields to investigate a broad range of physical systems. However, the accuracy of calculation is based on the model considered to describe the atomic interactions. In particular, ab initio molecular dynamics (AIMD) has the accuracy of density functional theory (DFT) and thus is limited to small systems and a relatively short simulation time. In this scenario, Neural Network Force Fields (NNFFs) have an important role, since they provide a way to circumvent these caveats. In this work, we investigate NNFFs designed at the level of DFT to describe liquid water, focusing on the size and quality of the training data set considered. We show that structural properties are less dependent on the size of the training data set compared to dynamical ones (such as the diffusion coefficient), and a good sampling (selecting data reference for the training process) can lead to a small sample with good precision. |
publishDate |
2023 |
dc.date.none.fl_str_mv |
2023-07-29T13:40:22Z 2023-07-29T13:40:22Z 2023-02-16 |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/article |
format |
article |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
http://dx.doi.org/10.1021/acs.jpcb.2c09059 Journal of Physical Chemistry B, v. 127, n. 6, p. 1422-1428, 2023. 1520-5207 1520-6106 http://hdl.handle.net/11449/248312 10.1021/acs.jpcb.2c09059 2-s2.0-85147507003 |
url |
http://dx.doi.org/10.1021/acs.jpcb.2c09059 http://hdl.handle.net/11449/248312 |
identifier_str_mv |
Journal of Physical Chemistry B, v. 127, n. 6, p. 1422-1428, 2023. 1520-5207 1520-6106 10.1021/acs.jpcb.2c09059 2-s2.0-85147507003 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.none.fl_str_mv |
Journal of Physical Chemistry B |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
1422-1428 |
dc.source.none.fl_str_mv |
Scopus reponame:Repositório Institucional da UNESP instname:Universidade Estadual Paulista (UNESP) instacron:UNESP |
instname_str |
Universidade Estadual Paulista (UNESP) |
instacron_str |
UNESP |
institution |
UNESP |
reponame_str |
Repositório Institucional da UNESP |
collection |
Repositório Institucional da UNESP |
repository.name.fl_str_mv |
Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP) |
repository.mail.fl_str_mv |
|
_version_ |
1808129051145011200 |