A data-driven approach to systematically and reproduciably optimize force field parameters

Features:

• Efficient adoption of parallism to perform sampling with a parallel simulator
• Diverse choices of force field functional forms supported by the chosen simulator
• Flexible inclusions of distinct physical properties as reference data
• Modular design to facilitate the exstensions with user-defined:
  • objective functional forms
  • sampling methods/force field potential functional forms
  • optimization algorithms

Software required:

• A compiled MD/MC packages exectuable (LAMMPS is already supported)

• Slurm Workload Manager (or equivalent)

• Python 3.7

• Other Python libraries: pytest==5.4.2, numpy==1.18.1

• Intel Fortran compiler (> version 18.0.3). gfortran (4.8.5) also compiles successfully, but the program may not work with a long absolute file path.

Installation on Linux:

conda is the recommended package manager. If 'conda' is not not found or root previliage is required, you can download anaconda: https://www.anaconda.com/products/individual to your home directory. Then, install the package: https://docs.anaconda.com/anaconda/install/

• create a conda environment with specific version of numpy and python:

conda create -n "env name" python=3.7 numpy=1.18.1 

• copy this GitHub repo to your local directory:

git clone https://github.com/jingxiangguo/Python-force-field-parameterization-workflow.git 

cd Python-force-field-parameterization-workflow

• install the package to your conda environment

pip install -e .

• A Fortran library "fortranAPI" directory comes with the package. Inside Fortran routines are C-interoperable, and thus can be callable through Python using ctypes modules.

• Note that this Fortran library can be used independently. It is used to accelerate some numerical intensive calculations in Python

• To compile it, run GNU "make" command.

make

• Test your installation.

optimize -h

usage: optimize [-h] -c CORES -i INPUT -j JOB [-m MODE] [-Ref REFERENCEDATA]
                [-prep PREPSYSTEM]

This is a Python software package implementing a force field parameters
optimization workflow

• Run some unit-testings to further check the installation (You may need to install pytest)

conda install pytest

pytest optimizer/

pytest fortranAPI/

Examples:

• Check out the "tutorial" directory for more details

References:

[1]: Chan, H., Cherukara, M. J., Narayanan, B., Loeffler, T. D., Benmore, C., Gray, S. K., & Sankaranarayanan, S. K. R. S. (2019). Machine learning coarse grained models for water. Nature Communications, 10(1), 379. https://doi.org/10.1038/s41467-018-08222-6

[2]: Gao, F., & Han, L. (2012). Implementing the Nelder-Mead simplex algorithm with adaptive parameters. Computational Optimization and Applications, 51(1), 259–277. https://doi.org/10.1007/s10589-010-9329-3

[3]: Wang, L.P., Chen, J., & Van Voorhis, T. (2013). Systematic Parametrization of Polarizable Force Fields from Quantum Chemistry Data. Journal of Chemical Theory and Computation, 9(1), 452–460. https://doi.org/10.1021/ct300826t

[4]: Ercolessi, F., & Adams, J. B. (1994). Interatomic Potentials from First-Principles Calculations: The Force-Matching Method. Europhysics Letters ({EPL}), 26(8), 583–588. https://doi.org/10.1209/0295-5075/26/8/005

[5]: Sundararaman, S., Huang, L., Ispas, S., & Kob, W. (2018). New optimization scheme to obtain interaction potentials for oxide glasses. Journal of Chemical Physics, 148(19). https://doi.org/10.1063/1.5023707