Self-learning path integral hybrid Monte Carlo with mixed ab initio and machine learning potentials for modeling nuclear quantum effects in water

Thomsen, Bo; Nagai, Yuki; Kobayashi, Keita; Hamada, Ikutaro; Shiga, Motoyuki

doi:https://doi.org/10.1063/5.0230464

Self-learning path integral hybrid Monte Carlo with mixed ab initio and machine learning potentials for modeling nuclear quantum effects in water

Thomsen, Bo; Nagai, Yuki; Kobayashi, Keita et al.

Journal of Chemical Physics, 2024, 161(20), 204109

Number of Access:530（2025-08-14 05:02 Counts）

Identifier to cite or link to this item: https://hdl.handle.net/11094/99630

Link to primary information

File	Format	Terms of use	Size	Views	Date.Available	Description	information
JChemPhys_161_20_204109	pdf	None	933 KB	39	2025.01.17

Item Information

Output File Export EndNote Basic Export Mendeley

Title	Self-learning path integral hybrid Monte Carlo with mixed ab initio and machine learning potentials for modeling nuclear quantum effects in water
Creator	NRID 1000000834178 1000000834178 CreatorName Thomsen, Bo
	NRID 1000020587026 1000020587026 CreatorName Nagai, Yuki
	NRID 1000000468860 1000000468860 CreatorName Kobayashi, Keita
	NRID 1000080419465 1000080419465 ORCID 0000-0001-5112-2452 0000-0001-5112-2452 researchmap permalink ihamada ihamada Scopus Author ID 7005614511 7005614511 CreatorName Hamada, Ikutaro
	NRID 1000040370407 1000040370407 CreatorName Shiga, Motoyuki
Description	Thomsen Bo et al., J. Chem. Phys. 161, 204109 (2024) https://doi.org/10.1063/5.0230464
Abstract	The introduction of machine learned potentials (MLPs) has greatly expanded the space available for studying Nuclear Quantum Effects computationally with ab initio path integral (PI) accuracy, with the MLPs’ promise of an accuracy comparable to that of ab initio at a fraction of the cost. One of the challenges in development of MLPs is the need for a large and diverse training set calculated by ab initio methods. This dataset should ideally cover the entire phase space, while not searching this space using ab initio methods, as this would be counterproductive and generally intractable with respect to computational time. In this paper, we present the self-learning PI hybrid Monte Carlo Method using a mixed ab initio and ML potential (SL-PIHMC-MIX), where the mixed potential allows for the study of larger systems and the extension of the original SL-HMC method [Nagai et al., Phys. Rev. B 102, 041124 (2020)] to PI methods and larger systems. While the MLPs generated by this method can be directly applied to run long-time ML-PIMD simulations, we demonstrate that using PIHMC-MIX with the trained MLPs allows for an exact reproduction of the structure obtained from ab initio PIMD. Specifically, we find that the PIHMC-MIX simulations require only 5000 evaluations of the 32-bead structure, compared to the 100 000 evaluations needed for the ab initio PIMD result.
Publisher	American Institute of Physics
Source Title	Journal of Chemical Physics
Volume (Issue)	161(20)
Page	204109
Date of Issued	2024-11-28
Language	English
Handle URL	https://hdl.handle.net/11094/99630
PISSN	00219606
eISSN	10897690
Relation.isIdenticalTo	DOI (Publisher Version) https://doi.org/10.1063/5.0230464
Access Rights	open access
CopyRight	This article may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This article appeared in Thomsen Bo et al., J. Chem. Phys. 28 November 2024; 161(20): 204109 and may be found at https://doi.org/10.1103/PhysRevMaterials.8.114004.
oaire:version	Accepted Manuscript
Category	学術雑誌論文 Journal Article

Resource Type	journal article
Local Type	Journal Article
DCTERMS.bibliographicCitation	Journal of Chemical Physics. 2024, 161(20), p. 204109
DC.title	Self-learning path integral hybrid Monte Carlo with mixed ab initio and machine learning potentials for modeling nuclear quantum effects in water
DC.creator	Thomsen, Bo
	Nagai, Yuki
	Kobayashi, Keita
	Hamada, Ikutaro
	Shiga, Motoyuki
DC.publisher	American Institute of Physics
DC.language' scheme='DCTERMS.RFC1766	English
DCTERMS.issued' scheme='DCTERMS.W3CDTF	2024-11-28
DC.identifier	https://doi.org/10.1063/5.0230464
DC.description	Thomsen Bo et al., J. Chem. Phys. 161, 204109 (2024) https://doi.org/10.1063/5.0230464
DCTERMS.abstract	The introduction of machine learned potentials (MLPs) has greatly expanded the space available for studying Nuclear Quantum Effects computationally with ab initio path integral (PI) accuracy, with the MLPs’ promise of an accuracy comparable to that of ab initio at a fraction of the cost. One of the challenges in development of MLPs is the need for a large and diverse training set calculated by ab initio methods. This dataset should ideally cover the entire phase space, while not searching this space using ab initio methods, as this would be counterproductive and generally intractable with respect to computational time. In this paper, we present the self-learning PI hybrid Monte Carlo Method using a mixed ab initio and ML potential (SL-PIHMC-MIX), where the mixed potential allows for the study of larger systems and the extension of the original SL-HMC method [Nagai et al., Phys. Rev. B 102, 041124 (2020)] to PI methods and larger systems. While the MLPs generated by this method can be directly applied to run long-time ML-PIMD simulations, we demonstrate that using PIHMC-MIX with the trained MLPs allows for an exact reproduction of the structure obtained from ab initio PIMD. Specifically, we find that the PIHMC-MIX simulations require only 5000 evaluations of the 32-bead structure, compared to the 100 000 evaluations needed for the ab initio PIMD result.
DC.rights	This article may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This article appeared in Thomsen Bo et al., J. Chem. Phys. 28 November 2024; 161(20): 204109 and may be found at https://doi.org/10.1103/PhysRevMaterials.8.114004.
DC.format	99630
citation_title	Self-learning path integral hybrid Monte Carlo with mixed ab initio and machine learning potentials for modeling nuclear quantum effects in water
citation_author	Thomsen, Bo
	Nagai, Yuki
	Kobayashi, Keita
	Hamada, Ikutaro
	Shiga, Motoyuki
citation_publisher	American Institute of Physics
citation_language	English
citation_date	2024-11-28
citation_journal_title	Journal of Chemical Physics
citation_volume	161
citation_issue	20
citation_firstpage	204109
citation_doi	https://doi.org/10.1063/5.0230464
citation_public_url	https://hdl.handle.net/11094/99630

Self-learning path integral hybrid Monte Carlo with mixed ab initio and machine learning potentials for modeling nuclear quantum effects in water

Link to primary information

Item Information

Show details