GPUMD: A package for constructing accurate machine-learned potentials and performing highly efficient atomistic simulations
Journal article, 2022

We present our latest advancements of machine-learned potentials (MLPs) based on the neuroevolution potential (NEP) framework introduced in Fan et al. [Phys. Rev. B 104, 104309 (2021)] and their implementation in the open-source package gpumd. We increase the accuracy of NEP models both by improving the radial functions in the atomic-environment descriptor using a linear combination of Chebyshev basis functions and by extending the angular descriptor with some four-body and five-body contributions as in the atomic cluster expansion approach. We also detail our efficient implementation of the NEP approach in graphics processing units as well as our workflow for the construction of NEP models and demonstrate their application in large-scale atomistic simulations. By comparing to state-of-the-art MLPs, we show that the NEP approach not only achieves above-average accuracy but also is far more computationally efficient. These results demonstrate that the gpumd package is a promising tool for solving challenging problems requiring highly accurate, large-scale atomistic simulations. To enable the construction of MLPs using a minimal training set, we propose an active-learning scheme based on the latent space of a pre-trained NEP model. Finally, we introduce three separate Python packages, viz., gpyumd, calorine, and pynep, that enable the integration of gpumd into Python workflows.

Author

Zheyong Fan

Bohai University

Yanzhou Wang

Aalto University

Penghua Ying

Harbin Institute of Technology

Keke Song

University of Science and Technology Beijing

Junjie Wang

Nanjing University

Yong Wang

Nanjing University

Zezhu Zeng

The University of Hong Kong

Ke Xu

Xiamen University

Eric Lindgren

Chalmers, Physics, Condensed Matter and Materials Theory

Magnus Rahm

Chalmers, Physics, Condensed Matter and Materials Theory

Alexander J. Gabourie

Stanford University

Jiahui Liu

University of Science and Technology Beijing

Haikuan Dong

Bohai University

Jianyang Wu

Xiamen University

Yue Chen

The University of Hong Kong

Zheng Zhong

Harbin Institute of Technology

Jian Sun

Nanjing University

Paul Erhart

Chalmers, Physics, Condensed Matter and Materials Theory

Yanjing Su

University of Science and Technology Beijing

Tapio Ala-Nissila

Aalto University

Journal of Chemical Physics

0021-9606 (ISSN) 1089-7690 (eISSN)

Vol. 157 11 114801

Phase behavior and electronic properties of mixed halide perovskites from atomic scale simulations

Swedish Research Council (VR) (2020-04935), 2020-12-01 -- 2024-11-30.

Hydrogen trapping by carbides in steel

Swedish Research Council (VR) (2021-05072), 2021-12-01 -- 2025-11-30.

Analysis and Modelling Service for Engineering Materials Studied with Neutrons

Swedish Research Council (VR) (2018-06482), 2018-11-01 -- 2020-12-31.

SwedNESS

Swedish Foundation for Strategic Research (SSF) (GSn15-0008), 2016-07-01 -- 2021-06-30.

Swedish Foundation for Strategic Research (SSF) (GSn15-0008), 2017-01-01 -- 2020-12-31.

Subject Categories

Computational Mathematics

Bioinformatics (Computational Biology)

Computer Science

Infrastructure

C3SE (Chalmers Centre for Computational Science and Engineering)

Areas of Advance

Materials Science

DOI

10.1063/5.0106617

PubMed

36137808

More information

Latest update

12/12/2023