Δ-HGPR: A Hierarchical Gaussian Process Framework for Active-Learning Δ-Potentials
Journal of Chemical Theory and Computation · 10.11.2026 · 22 (21) · 11153-11167
DOI: 10.1021/acs.jctc.6c01498 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Chemical Theory and Computation, online erschienen am 24.10.2026. Verfasst von Takahito Nakajima. Beteiligte Einrichtung: RIKEN Center for Computational Science , 7-1-26 Minatojima-minami-machi, Chuo-ku, Kobe, Hyogo650-0047, Japan. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 43 Literaturangaben.
Abstract (Originalsprache)
Δ-learning, which uses machine learning to represent only the difference between a high-accuracy reference and a lower-accuracy base calculator, is a promising route to high-fidelity molecular dynamics (MD) with limited training data, but it has lacked a unified protocol combining hierarchical descriptors with well-defined equilibrium sampling. We propose Δ-HGPR (hierarchical Gaussian process regression for Δ-learning), a framework for learning difference potentials, and its MD instantiation, Δ-HGPR-MD. The framework combines (i) a difference potential written as an additive radial (inverse-distance) term plus a many-body machine-learning correction, determined by a single regularized least-squares fit; (ii) the ensemble force standard deviation as an active-learning indicator with an automatically scaled threshold; and (iii) a difference potential held fixed within each round so that equilibrium NVT sampling is well-defined in both exploration and production. The many-body term of Δ-HGPR-MD uses smooth overlap of atomic positions (SOAP) descriptors with random Fourier features (Δ-HGPR-RFF). On a toy system whose difference potential contains a 1/r contribution beyond the cutoff of the local descriptor, the many-body term alone reproduces none of this force, however many features it is given, whereas the hierarchy recovers 99% of it, with a reach set by the cutoff of its radial term. In MD, the hierarchy reproduces about 80% of the shift of the ion-pair distribution toward the reference, while the many-body term alone leaves it where the base places it. On malonaldehyde and the Zundel cation H5O2+ with three base calculators, Δ-HGPR-RFF removes barrier errors of up to 0.1 eV under one protocol, giving free-energy barriers statistically indistinguishable from DFT; free-energy perturbation shows that the error contributed by the learned correction is below 0.01 eV on average in every case. For H5O2+with ANI-2x, whose dynamics collapse without correction, it recovers the Zundel geometry and a barrier and transition count consistent with DFT.
Autor:innen und Institutionen
- RIKEN Center for Computational Science , 7-1-26 Minatojima-minami-machi, Chuo-ku, Kobe, Hyogo650-0047, Japan.
ORCID-Profile: Takahito Nakajima
Zitieren
APA
Nakajima, T. (2026). Δ-HGPR: A Hierarchical Gaussian Process Framework for Active-Learning Δ-Potentials. Journal of Chemical Theory and Computation, 22(21), 11153-11167. https://doi.org/10.1021/acs.jctc.6c01498
Vancouver
Nakajima T. Δ-HGPR: A Hierarchical Gaussian Process Framework for Active-Learning Δ-Potentials. Journal of Chemical Theory and Computation. 2026;22(21):11153-11167. doi:10.1021/acs.jctc.6c01498
BibTeX
@article{Nakajima2026,
author = {Nakajima, Takahito},
title = {Δ-HGPR: A Hierarchical Gaussian Process Framework for Active-Learning Δ-Potentials},
journal = {Journal of Chemical Theory and Computation},
year = {2026},
volume = {22},
number = {21},
pages = {11153--11167},
doi = {10.1021/acs.jctc.6c01498}
}
EUVIMED