Skip to main content

DPTools: CLI toolkit and python library for working with deepmd-kit.

Project description

DPTools

Deep Potential Tools (DPTools) provides a command-line interface and python library to simplify training and deploying DeePMD-kit machine learning potentials (MLPs), also known as ML force fields. The primary goal of DPTools is to condense workflows for training DP MLPs and running atomistic simulations with LAMMPS on HPC systems into a handful of intuitive CLI commands. It is intended for scientists with knowledge of quantum mechanics-based ab-initio simulation methods who are interested in effortlessly transitioning to ML-based approaches to greatly increase computational throughput. It requires no prior experience with DeePMD-kit or LAMMPS software, only familiarity with the popular Atomic Simulation Environment (ASE) python package is needed.

Main Features

  • Setup deepmd-kit training sets from VASP output or common ASE formats
  • Train ensemble of DP models
  • Generate parity plots to assess accuracy of MLP energy and force predictions
  • Intelligently sample and select new training configurations from DPMD trajectories
  • Easily setup and run different atomistic simulations in LAMMPS:
    • Single point energy calculations
    • Structure geometry optimizations
    • Structure unit cell optimizations
    • Molecular dynamics (NVT and NPT ensembles)
    • Equations of State and bulk moduli calculations
    • Vibratrional/phonon modes using the finite differences approach
    • Other common simulation methods available upon request
  • Supports Slurm job submission on HPC systems
  • Setup and run simulations on thousands of structures with a single command

Documentation

For detailed descriptions on setting up and using DPTools, visit the official documentation.

Quick Install

The current stable version (1.0.0) of DPTools can be installed using pip with the following command:

pip install dpmdtools

To verify that the installation was completed successfully, run the command:

dptools --version

Support

If you are having issues with DPTools, create an issue here. For more assistance, new feature requests, or general inquiries, feel free to contact Ty at tsours@ucdavis.edu.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

dpmdtools-1.0.1.tar.gz (44.5 kB view hashes)

Uploaded Source

Built Distribution

dpmdtools-1.0.1-py3-none-any.whl (56.6 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page