Documentation

File

1. Datset. The Dataset for Mg, Cu, and MgCu.
2. lammps. The LAMMPS plugin for using MLFF model.
3. calss_potential. The Class Potentials used as a comparison.
4. model.pt. The MLFF model with completed training.

Install LAMMPS plugin and Use MLFF

# 1. Download LAMMPS and Libtorch
# 2. Modify the path of Libtorch in lammps/Makefile.mpi, then copy the following files to the specified directory
cp -rf lammps/MLFF LAMMPS/src
cp -rf lammps/Makefile.mpi LAMMPS/src/MAKE
# 3. Install LAMMPS
module load mkl mpi compiler gcc
cd LAMMPS/src
make yes-mlff
make mpi -j 8
# 4. Installation complete, generate lmp_mpi executable. Check that lammps is installed correctly by entering the following command in the directory LAMMPS/src.
lmp_mpi
# 5. Call MLFF with the following command. 
pari_style mlff model.pt
pair_coffe * * 12 29   # Use Mg Cu
pair_coffe * * 12      # Only Use Mg 

Note: software version. LAMMPS: lammps-23Jun2022; Libtorch: 2.2.0 gcc: >=9.0

Cite

1. If you use the Dataset or MLFF model, , please cite our papers:

@article{GENG2025113486, 
title = {Exploring the strength-concentration relationship of MgCu bioalloys at low copper content using machine learning force fields},
journal = {Computational Materials Science},
volume = {246},
pages = {113486},
year = {2025},
issn = {0927-0256},
doi = {https://doi.org/10.1016/j.commatsci.2024.113486},
url = {https://www.sciencedirect.com/science/article/pii/S0927025624007079},
author = {Xingze Geng and Lin-Wang Wang and Xiangying Meng}
}