graph-pes 0.0.2

Potential Energy Surfaces on Graphs

graph-pes is a framework built to accelerate the development of machine-learned potential energy surface (PES) models that act on graph representations of atomic structures.

Features

• Experiment with new model architectures by inheriting from our GraphPESModel base class.
• Train your own or existing models (e.g., SchNet, NequIP, PaiNN, MACE, etc.).
• Easily configure distributed training, learning rate scheduling, weights and biases logging, and other features using our graph-pes-train command line interface.
• Use our data-loading pipeline within your own training loop.
• Run molecular dynamics simulations via LAMMPS (or ASE) using any GraphPESModel and the pair_style graph_pes LAMMPS command.

Quickstart

For a 0-install quickstart experience, please see this Google colab, which you can also find in our documentation.

Want to try this out locally? Run the following commands:

# optionally create a new environment
conda create -n graph-pes python=3.10
conda activate graph-pes

# install graph-pes
pip install graph-pes

# download a config file
wget https://tinyurl.com/graph-pes-qm7-quickstart

# train a model
graph-pes-train qm7-quickstart.yaml

Contributing

Contributions are welcome! If you find any issues or have suggestions for new features, please open an issue or submit a pull request on the GitHub repository.

License

graph-pes is licensed under the MIT License.

Acknowledgments

graph-pes builds upon the following open-source projects:

• PyTorch
• LAMMPS
• ASE

We are grateful for the contributions of the developers and maintainers of these projects.

---