In this repository, we provide all necessary files to reproduce our work in "Machine Learning Diffusion Monte Carlo Forces".

Necessary software:

• QMCPACK
• QuantumESPRESSO
• PySCF
• AMPtorch

For python environment, we follow the setup tutorial here.

Due to the file size limit by Github, an extra compressed .zip file including all training and test dataset are provided in below link.

https://drive.google.com/file/d/1NJuZSZlur-7os5HtXxOKgpmXuZxWtqjF/view?usp=sharing

The repository is organized based on the examples we studied in our work. You can find the example-related python scripts in each example subfolder.

For example,

- C2
  - data: scripts to generate training, validation, and test data
  - train: scripts to train different AMPtorch models
  - md_start_eq: scripts for MD simulations using different trained models
  - geom_opt: scripts for geometry optimization using different trained models
  - notebooks_analysis: scripts for data analysis and all figures used in our work