A package of LAMMPS software enabling simulations using polynomial machine learning potentials

A polynomial MLP represents the potential energy as a polynomial function of linearly independent polynomial invariants of the O(3) group. Developed polynomial MLPs are available in Polynomial Machine Learning Potential Repository.

“Tutorial: Systematic development of polynomial machine learning potentials for elemental and alloy systems”, A. Seko, J. Appl. Phys. 133, 011101 (2023)

@article{pypolymlp,
    author = {Seko, Atsuto},
    title = "{"Tutorial: Systematic development of polynomial machine learning potentials for elemental and alloy systems"}",
    journal = {J. Appl. Phys.},
    volume = {133},
    number = {1},
    pages = {011101},
    year = {2023},
    month = {01},
}

(lammps-polymlp-package is tested using LAMMPS_VERSION "lammps-23Jun2022”)

1. Copy all the components in the lammps-polymlp-package to the latest lammps source code directory as

   > cp -r lammps-polymlp-package/lib/polymlp $(lammps_src)/lib
   > cp -r lammps-polymlp-package/src/POLYMLP $(lammps_src)/src
   

2. Add "polymlp" to variable PACKAGE defined in $(lammps_src)/src/Makefile and activate polymlp package as

   > cat $(lammps_src)/src/Makefile
       PACKAGE = \
       adios \
       amoeba \
       ...
       poems \
       polymlp \
       ptm \
       ...
       ml-iap \
       phonon
       ...
   > ulimit -s unlimited
   > cd $(lammps_src)/src
   > make yes-polymlp
   

3. If necessary, modify compile options and flags in $(lammps_src)/src/MAKE/Makefile.serial and $(lammps_src)/lib/polymlp/Makefile.lammps

4. Change directory to $(lammps_src)/src and build lammps binary files. (It requires approximately ten minutes to one hour for compiling polymlp_gtinv_data.cpp.)

   > make serial -j 36
   

Note that MPI is not currently available for the lammps-polymlp package, but OpenMP is supported. Even when using the OpenMP-supported lammps-polymlp package, you should compile LAMMPS using the command make serial. Parallel calculations can be performed using OpenMP with lmp_serial.

Molecular dynamics and other calculations available in LAMMPS, including those using user packages, can be performed by following the standard LAMMPS documentation, except for the specification of pair_style and pair_coeff. The following LAMMPS input commands specify a machine learning potential (e.g., for a binary Ti–Al system):

pair_style  polymlp
pair_coeff * * polymlp.lammps Ti Al

If you are using a polynomial MLP generated by pypolymlp (version ≥ 0.9.0), the potential file can be specified as follows:

pair_style  polymlp
pair_coeff * * polymlp.yaml Ti Al

In hybrid polynomial MLPs, the potential energy is represented as a sum of multiple polynomial MLPs, where the regression coefficients of all models are determined simultaneously. When performing simulations with hybrid polynomial MLPs in LAMMPS, pair_style hybrid command can be used:

pair_style hybrid polymlp polymlp polymlp
pair_coeff * * polymlp.yaml.1 Ti Al
pair_coeff * * polymlp.yaml.2 Ti Al
pair_coeff * * polymlp.yaml.3 Ti Al