DeePMD-kit v3: A Multiple-Backend Framework for Machine Learning Potentials
In recent years, machine learning potentials (MLPs) have become indispensable tools in physics, chemistry, and materials science, driving the development of software packages for molecular dynamics (MD) simulations and related applications. These packages, typically built on specific machine learning frameworks, such as TensorFlow, PyTorch, or JAX, face integration challenges when advanced applications demand communication across different frameworks. The previous TensorFlow-based implementation of the DeePMD-kit exemplified these limitations. In this work, we introduce DeePMD-kit version 3, a significant update featuring a multibackend framework that supports TensorFlow, PyTorch, JAX, and PaddlePaddle backends, and demonstrate the versatility of this architecture through the integration of other MLP packages and of differentiable molecular force fields. This architecture allows seamless back-end switching with minimal modifications, enabling users and developers to integrate DeePMD-kit with other packages using different machine learning frameworks. This innovation facilitates the development of more complex and interoperable workflows, paving the way for broader applications of MLPs in scientific research.
Authors
- Junhan Chang (ORCID: https://orcid.org/0000-0003-3195-0773)
- Zhaoxi Yu
- Denghui Lu (ORCID: https://orcid.org/0000-0003-0977-3635)
- Timothy J. Giese (ORCID: https://orcid.org/0000-0002-0653-9168)
- Sigbjørn Løland Bore (ORCID: https://orcid.org/0000-0002-8620-4885)
- Anyang Peng (ORCID: https://orcid.org/0000-0002-0630-2187)
- Koki Muraoka (ORCID: https://orcid.org/0000-0003-1830-7978)
- Siyuan Han
- Chenxing Luo (ORCID: https://orcid.org/0000-0003-4116-6851)
- Wanrun Jiang (ORCID: https://orcid.org/0000-0001-7833-9862)
- Yun‐Pei Liu
- Ruosong Shi
- Hananeh Oliaei (ORCID: https://orcid.org/0000-0002-0828-6897)
- Jinzhe Zeng (ORCID: https://orcid.org/0000-0002-1515-8172)
- Anurag Kumar Singh (ORCID: https://orcid.org/0000-0001-9696-9016)
- Jiaxin Zhu (ORCID: https://orcid.org/0000-0002-3471-4728)
- Pinghui Mo (ORCID: https://orcid.org/0009-0009-4591-8644)
- Chengqian Zhang (ORCID: https://orcid.org/0000-0001-7829-0587)
- Haohui Que
- Zhaonan Ding (ORCID: https://orcid.org/0000-0003-2026-8795)
- Guolin Ke (ORCID: https://orcid.org/0000-0002-1227-7221)
- Xinzijian Liu (ORCID: https://orcid.org/0000-0003-2242-6881)
- Shaochen Shi (ORCID: https://orcid.org/0000-0003-3457-403X)
- Yong‐Bin Zhuang (ORCID: https://orcid.org/0000-0001-5182-8084)
- Zhangmancang Xu
- Darrin M. York (ORCID: https://orcid.org/0000-0002-9193-7055)
- Ben Bin Xu (ORCID: https://orcid.org/0000-0002-6747-2016)
- Zhengtao Huang (ORCID: https://orcid.org/0009-0003-9371-4458)
- Jiayu Dai (ORCID: https://orcid.org/0000-0001-7423-7500)
- Qiyu Zeng (ORCID: https://orcid.org/0009-0004-2304-0213)
- Zhenyu Wang (ORCID: https://orcid.org/0009-0007-7927-6753)
- Yifan Li (ORCID: https://orcid.org/0000-0003-3217-0592)
- Handong Wang (ORCID: https://orcid.org/0000-0001-5989-9338)
- Duo Zhang (ORCID: https://orcid.org/0000-0001-9591-2659)
- Ye Ding (ORCID: https://orcid.org/0009-0005-1986-2879)
- Linfeng Zhang (ORCID: https://orcid.org/0000-0002-8470-5846)
- Weihong Xu
- Yiming Du
- Zhaoqing Liu (ORCID: https://orcid.org/0009-0007-8286-9090)
- Weile Jia (ORCID: https://orcid.org/0000-0001-8539-8326)
- Chun Cai (ORCID: https://orcid.org/0000-0001-6242-0439)
- Xuejian Qin (ORCID: https://orcid.org/0009-0003-0386-4226)
- Xiangyu Zhang (ORCID: https://orcid.org/0009-0003-0940-5565)
- Zhe Deng (ORCID: https://orcid.org/0000-0002-0990-6031)
- S Z He (ORCID: https://orcid.org/0009-0007-9793-4510)
- Hangrui Bi
- Yan Wang (ORCID: https://orcid.org/0009-0005-2485-0899)
Institutions
- University of Science and Technology of China (CN)
- University of Illinois Urbana-Champaign (US)
- Hunan University (CN)
- National University of Defense Technology (CN)
- University of Oslo (NO)
- University of Toronto (CA)
- Xiamen University (CN)
- Jilin University (CN)
- Wuhan University of Technology (CN)
- Chinese Academy of Sciences (CN)
- Princeton University (US)
- Peking University (CN)
- Beijing Normal University (CN)
- Shanghai Astronomical Observatory (CN)
- Art Institute of Portland (US)
- China Academy of Engineering Physics (CN)
- Chinese Academy of Engineering (CN)
- Institute of Computing Technology (CN)
- Rutgers Sexual and Reproductive Health and Rights (NL)
- Quantitative BioSciences (US)
- Integrated Optical Circuit Consultants (United States) (US)
- Smart Material (Germany) (DE)
- Division of Materials Science and Engineering (AU)
- Jilin Medical University (CN)
- Center for Interdisciplinary Studies (IN)
- SP Technology (South Korea) (KR)
- Institute of Applied Physics and Computational Mathematics (CN)
- Collaborative Innovation Center of Chemistry for Energy Materials (CN)
- University of Chinese Academy of Sciences (CN)
- Ningbo Institute of Industrial Technology (CN)
- State Key Laboratory of Advanced Technology For Materials Synthesis and Processing
- École Polytechnique Fédérale de Lausanne (CH)
- The University of Tokyo (JP)
- Columbia University (US)
- Baidu (China) (CN)
Publication Details
- Journal
- Journal of Chemical Theory and Computation
- Published
- 2025-05-02
- DOI
- https://doi.org/10.1021/acs.jctc.5c00340
- Citations
- 50
- Primary Topic
- Machine Learning in Materials Science
- Type
- article
- Field-Weighted Citation Impact
- 25.89
Funders
- National Science Foundation
- National Natural Science Foundation of China
- China Scholarship Council
- Norges Forskningsråd
- Xiamen University
- Science and Technology Program of Hunan Province
- National Key Research and Development Program of China
- National Institute of General Medical Sciences