# Self-Consistency Error Correction for Accurate Machine Learning Potentials from Variational Monte Carlo

https://mdr.nims.go.jp/datasets/b0a6afa5-0e89-450a-8778-95f82a194887

## File

- [2504.20481v2.pdf](https://mdr.nims.go.jp/filesets/6ffe5718-f584-4211-8462-cce18f00a380/download) ([Detail](https://mdr.nims.go.jp/filesets/6ffe5718-f584-4211-8462-cce18f00a380.md))

## Id

b0a6afa5-0e89-450a-8778-95f82a194887

## Local identifier



## Visibility

open_to_public

## State

published

## Created at

2025-12-01T07:40:28.172152Z

## Updated at

2025-12-01T23:30:12.929767Z

## Published at

2025-12-01T23:23:29.532034Z

## Doi



## First published url

https://doi.org/10.1021/acs.jctc.5c00715

## Date published

2025-10-14

## Recorded date published

2025-10-14

## Resource type

journal_article

## Manuscript type

accepted_manuscript

## Collection



## Title

- title: Self-Consistency Error Correction for Accurate Machine Learning Potentials
    from Variational Monte Carlo
  title_type: original
  lang: en

## Description

- description: Variational Monte Carlo (VMC) can be used to train accurate machine
    learning interatomic potentials (MLIPs), enabling molecular dynamics (MD) simulations
    of complex materials on time scales and for system sizes previously unattainable.
    VMC training sets are often based on partially optimized wave functions (WFs)
    to circumvent expensive energy optimizations of the whole set of WF parameters.
    However, frozen variational parameters lead to VMC forces and pressures not consistent
    with the underlying potential energy surface, a bias called the self-consistency
    error (SCE). Here, we demonstrate how the SCE can spoil the accuracy of MLIPs
    trained on these data, taking high-pressure hydrogen as test case. We then apply
    a recently introduced SCE correction [ Phys. Rev. B 109, 205151 (2024)] to generate
    unbiased VMC training sets based on a Jastrow-correlated single determinant WF
    with frozen Kohn-Sham orbitals. The MLIPs generated within this framework are
    significantly improved and can approach in quality those trained on datasets built
    with fully optimized WFs. Our conclusions are further supported by MD simulations,
    which show how MLIPs trained on SCE-corrected datasets systematically yield more
    reliable physical observables. Our framework opens the possibility of constructing
    extended high-quality training sets with VMC.
  description_type: abstract
  lang: und

## Creator

- name: Giacomo Tenti
  role: author
- name: Kousuke Nakano
  role: author
  orcid: https://orcid.org/0000-0001-7756-4355
  organization: National Institute for Materials Science
- name: Michele Casula
  role: author

## Contact agent



## Publisher

organization: American Chemical Society (ACS)

## Managing organization



## Keyword

- subject: Quantum Monte Carlo
  schema: not_defined
- subject: Variational Monte Carlo
  schema: not_defined
- subject: Atomic forces
  schema: not_defined

## Rights

- identifier: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

## Other identifier(s)



## Data origin

- data_origin_type: other

## Embargo



## Journal

- title: Journal of Chemical Theory and Computation
  issn: '15499618'
  volume: '21'
  issue: '19'
  start_page: 9335
  end_page: 9346

## Conference



## Related item



## Funding

- identifier: EHPC-EXT-2024E01-064
  funder_name: European High Performance Computing Joint Undertaking
- identifier: JPMXS0320220025
  funder_name: Ministry of Education, Culture, Sports, Science and Technology
- identifier: M24AN006
  funder_name: Murata Science and Education Foundation
- identifier: JPMJPR24J9
  funder_name: Japan Science and Technology Agency (JST)
- identifier: Hpc AlliaNce for Applications and supercoMputing Innovation
  funder_name: European High Performance Computing Joint Undertaking

## Instrument



## Instrument operator



## Instrument managing organization



## Measurement method



## Specimen



## Chemical composition



## Structure for specimen



## Structural feature for specimen



## Specific property for specimen



## Process for specimen treatment



## Computational method



## Energy level/transition state



## Software



## Custom property



## Fileset

- id: 6ffe5718-f584-4211-8462-cce18f00a380
  filename: 2504.20481v2.pdf
  content_type: application/pdf
  size: 1189066
  md5: '028ca840adcdbe116ef741b225eae745'

## Thumbnail

fileset_id: 6ffe5718-f584-4211-8462-cce18f00a380
filename: 2504.20481v2.pdf