Atom-centered machine-learning force field package

Journal article

Lei Li, Ryan A Ciufo, Jiyoung Lee, Chuan Zhou, Bo Lin, Jaeyoung Cho, Naman Katyal, Graeme Henkelman
Computer Physics Communications, 2023

DOI: 10.1016/j.cpc.2023.108883

Cite

APA Click to copy
Li, L., Ciufo, R. A., Lee, J., Zhou, C., Lin, B., Cho, J., … Henkelman, G. (2023). Atom-centered machine-learning force field package. Computer Physics Communications. https://doi.org/10.1016/j.cpc.2023.108883

Chicago/Turabian Click to copy
Li, Lei, Ryan A Ciufo, Jiyoung Lee, Chuan Zhou, Bo Lin, Jaeyoung Cho, Naman Katyal, and Graeme Henkelman. “Atom-Centered Machine-Learning Force Field Package.” Computer Physics Communications (2023).

MLA Click to copy
Li, Lei, et al. “Atom-Centered Machine-Learning Force Field Package.” Computer Physics Communications, 2023, doi:10.1016/j.cpc.2023.108883.

BibTeX Click to copy

@article{lei2023a,
  title = {Atom-centered machine-learning force field package},
  year = {2023},
  journal = {Computer Physics Communications},
  doi = {10.1016/j.cpc.2023.108883},
  author = {Li, Lei and Ciufo, Ryan A and Lee, Jiyoung and Zhou, Chuan and Lin, Bo and Cho, Jaeyoung and Katyal, Naman and Henkelman, Graeme}
}

In recent years, machine learning algorithms have been widely used for constructing force fields with an accuracy of ab initio methods and the efficiency of classical force fields. Here, we developed a python-based atom-centered machine-learning force field (PyAMFF) package to provide a simple and efficient platform for fitting and using machine learning force fields by implementing an atom-centered neural-network algorithm with Behler-Parrinello symmetry functions as structural fingerprints. The following three features are included in PyAMFF: (1) integrated Fortran modules for fast fingerprint calculations and Python modules for user-friendly integration through scripts and facile extension of future algorithms; (2) a pure Fortran backend to interface with the software, including the long-timescale dynamic simulation package EON, enabling both molecular dynamic simulations and adaptive kinetic Monte Carlo simulations with machine-learning force fields; and (3) integration with the Atomic Simulation Environment package for active learning and ML-based algorithm development. Here, we demonstrate an efficient parallelization of PyAMFF in terms of CPU and memory usage and show that the Fortran-based PyAMFF calculator exhibits a linear scaling relationship with the number of symmetry functions and the system size.