Efficient Parallelization of Message Passing Neural Network Potentials for Large-Scale Molecular Dynamics
JACS Au · 22.06.2026 · 6 (6) · 3239-3248
DOI: 10.1021/jacsau.6c00210 ↗Kurz gefasst
Fachartikel in der Zeitschrift JACS Au, online erschienen am 25.05.2026. Verfasst von Junfan Xia und 1 weiteren Autor:innen. Beteiligte Einrichtung: State Key Laboratory of Precision and Intelligent Chemistry, University of Science and Technology of China, Hefei, Anhui 230026, China. Die Arbeit nennt 82 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: National Natural Science Foundation of China, Strategic Priority Research Program of the Chinese Academy of Sciences, Quantum Science and Technology-National Science and Technology Major Project.
Abstract (Originalsprache)
High Resolution Image Download MS PowerPoint Slide Machine learning potentials have achieved great success in accelerating atomistic simulations, among which message passing neural networks (MPNNs) have become increasingly prevalent thanks to their superior accuracy. However, MPNN potentials are difficult to be efficiently parallelized due to their semilocal architecture. Here, we propose an efficient scheme for parallelizing MPNN potentials, in which additional data communication is minimized among local atoms only in each layer without redundant computation, thus scaling linearly with the message-passing depth. This scheme is first tested on several bulk systems─including silver, liquid water, and high-entropy alloys─demonstrating excellent strong- and weak-scaling performance up to over 100 million atoms, extending MPNN applicability to an unprecedented scale. Moreover, we develop a universal potential trained on a comprehensive data set of C, H, O, and N, and leverage this parallel algorithm to perform efficient reactive MD simulations of graphene formation via acetylene–oxygen detonation, a central process in carbon nanomaterial synthesis. With atomic resolution, our simulations reveal how oxygen actively mediates the reaction network and identify the O 2 /C 2 H 2 ratio that optimally favors graphene-precursor formation, offering mechanistic insights into the experimentally observed necessity of oxygen for controlling graphene quality and yield. The proposed parallelization framework can be readily extended to other MPNN potentials.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- State Key Laboratory of Precision and Intelligent Chemistry, University of Science and Technology of China, Hefei, Anhui 230026, China.
- School of Chemistry and Materials Science, Department of Chemical Physics, University of Science and Technology of China, Hefei, Anhui 230026, China.
- Hefei National Laboratory, University of Science and Technology of China, Hefei 230088, China.
ORCID-Profile: Junfan Xia, Bin Jiang
Förderung
- National Natural Science Foundation of China
- Strategic Priority Research Program of the Chinese Academy of Sciences
- Quantum Science and Technology-National Science and Technology Major Project
Zitieren
APA
Xia, J., & Jiang, B. (2026). Efficient Parallelization of Message Passing Neural Network Potentials for Large-Scale Molecular Dynamics. JACS Au, 6(6), 3239-3248. https://doi.org/10.1021/jacsau.6c00210
Vancouver
Xia J, Jiang B. Efficient Parallelization of Message Passing Neural Network Potentials for Large-Scale Molecular Dynamics. JACS Au. 2026;6(6):3239-3248. doi:10.1021/jacsau.6c00210
BibTeX
@article{Xia2026,
author = {Xia, Junfan and Jiang, Bin},
title = {Efficient Parallelization of Message Passing Neural Network Potentials for Large-Scale Molecular Dynamics},
journal = {JACS Au},
year = {2026},
volume = {6},
number = {6},
pages = {3239--3248},
doi = {10.1021/jacsau.6c00210}
}
EUVIMED