This package implements the NequIP-LES shown in A universal augmentation framework for long-range electrostatics in machine learning interatomic potentials.

In particular, NequIP-LES implements the LES library as an extension package for the NequIP framework.

• Installation
• Usage
• License
• LAMMPS Integration
• Citation
• Contact and questions

Nequip-LES requires the nequip and allegro packages. Details on nequip and allegroand their required PyTorch versions can be found in the nequip docs.

Nequip-LES can be installed using pip

git clone https://github.com/ChengUCB/NequIP-LES.git
pip install -e . 

Installing Nequip-LES in this way will also install the nequip package from PyPI and les package from GitHub.

The Nequip-LES package provides the Nequip-LES model for use within the NequIP framework. The framework's documentation describes how to train, test, and use models.

Nequip-LES now supports both the NequIP and Allegro.

A minimal example of a config file for training a Nequip-LES model is provided at configs/tutorial_les.yaml.

You can use the Allegro model by changing base_model: nequip to base_model: allegro in model details.

This project is licensed under the CC BY-NC 4.0 License.

LAMMPS Integration has not been tested yet.

If you use this code in your academic work, please cite:

@article{Kim2025universal,
  title = {A Universal Augmentation Framework for Long-Range Electrostatics in Machine Learning Interatomic Potentials},
  author = {Kim, Dongjin and Wang, Xiaoyu and Zhong, Peichen and King, Daniel S. and Inizan, Theo Jaffrelot and Cheng, Bingqing},
  journal={arXiv preprint arXiv:2507.14302},
  year = {2025}

And also consider citing:

1. Latent Ewald summation for machine learning of long-range interactions

2. Machine learning of charges and long-range interactions from energies and forces

3. Machine learning interatomic potential can infer electrical response

4. The original NequIP paper

5. The Allegro paper

6. The e3nn equivariant neural network package used by NequIP, through its preprint and/or code

If you find a bug or have a proposal for a feature, please post it in the Issues or reach out to [email protected]