EUVIMEDEUVIMED entdecken
Crossref

Δ-learning for transferable machine learning interatomic potentials

Nguyen Thien Phuc Tu iD, Christopher N Rowley iD

The Journal of Chemical Physics · 07.08.2026 · 165 (5)

DOI: 10.1063/5.0342595 ↗

Kurz gefasst

Fachartikel in der Zeitschrift The Journal of Chemical Physics, online erschienen am 06.08.2026. Verfasst von Nguyen Thien Phuc Tu und 1 weiteren Autor:innen. Beteiligte Einrichtung: Department of Chemistry, Carleton University, Ottawa, Ontario K1S 5B6, Canada. Die Arbeit nennt 56 Literaturangaben. Gefördert von: NSERC of Canada.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Machine-learning interatomic potentials (MLIPs) trained by directly learning the total interatomic interaction energies can suffer from limited transferability, unphysical behavior beyond a finite cutoff, and large errors for out-of-distribution geometries such as transition states and uncommon conformers. We evaluate Δ-learning (delta-learning) as a remedy by training an ANI-style high-dimensional neural network (HDNNP) potential as a correction to predict PBE0/aug-cc-pVTZ energies from a third-order tight-binding density functional theory (DFTB3) baseline model. On a held-out test set derived from the modified ANI-1x training set, the Δ-learning model (named ANIDFTB-Δ) achieves a mean absolute error (MAE) of 0.82 kcal/mol, while the HDNNP model (ANIPBE0-Direct) has an MAE of 2.01 kcal/mol. On selected GMTKN55 benchmarks, the Δ model systematically improves relative energies for conformers and tautomers and avoids catastrophic outliers on challenging structures. The good physical description of DFTB3 significantly reduces the error in proton-transfer transition states in the PX13 benchmark and intermolecular interactions in the DES370K dataset in comparison with reference target PBE0. The long-range electrostatics in DFTB3 also partially correct the long-range behavior of local descriptor MLIPs outside of their predetermined cutoff, reducing the MAE from 0.098 to 0.019 kcal/mol. The computational cost of this method is incrementally more expensive than DFTB3 alone, making it practical for extensive simulations of moderately sized systems, although the DFTB3 step makes the Δ model much more costly than a simple HDNNP alone.

Autor:innen und Institutionen

  • Department of Chemistry, Carleton University, Ottawa, Ontario K1S 5B6, Canada.

ORCID-Profile: Nguyen Thien Phuc Tu, Christopher N Rowley

Förderung

  • NSERC of Canada

Zitieren

APA

Phuc Tu, N. T., & Rowley, C. N. (2026). Δ-learning for transferable machine learning interatomic potentials. The Journal of Chemical Physics, 165(5). https://doi.org/10.1063/5.0342595

Vancouver

Phuc Tu NT, Rowley CN. Δ-learning for transferable machine learning interatomic potentials. The Journal of Chemical Physics. 2026;165(5). doi:10.1063/5.0342595

BibTeX
@article{PhucTu2026,
  author = {Phuc Tu, Nguyen Thien and Rowley, Christopher N},
  title = {Δ-learning for transferable machine learning interatomic potentials},
  journal = {The Journal of Chemical Physics},
  year = {2026},
  volume = {165},
  number = {5},
  doi = {10.1063/5.0342595}
}
Kostenlos starten →