📌 Overview

This project provides a fast and accurate script for structural relaxation of atomic systems using the ORB model—an attention-augmented Graph Network-based Simulator (GNS) . ORB is a type of Message Passing Neural Network (MPNN) that uses orbital features to predict forces and energies with high fidelity.

By leveraging ASE (Atomic Simulation Environment) and the pre-trained ORB force field, this script enables reliable relaxation of atomic structures, suitable for rapid screening or high-throughput simulations.

ORB Structure Relaxation

Relax atomic structures using pre-trained neural network models from ORB-models, integrated with the Atomic Simulation Environment (ASE).

This project uses a pre-trained ORB model to perform structural relaxation on atomic systems. The relaxation process is carried out with the BFGS optimizer in ASE. It supports both CPU and CUDA devices.

Orbital-Based Relaxation with ORB Model and ASE

This project performs a structural relaxation using the pre-trained ORB (Orbital Regression Based) force field model in combination with the Atomic Simulation Environment (ASE).

💪 Requirements

• Python 3.8+
• ASE
• NumPy
• Matplotlib
• orb_models (from ORB GitHub repository)

📦 Installation

Create a virtual environment and install the required packages:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

📂 Files and Folders

• data/structure.vasp – Input structure for relaxation.
• src/relax_orb.py – Main Python script to perform structure relaxation.
• outputs/ – Contains the relaxed structure and plots.

🚀 How to Run

Run the relaxation script:

python src/relax_orb.py

📜 License

This project is licensed under the MIT License.

---

📬 Citation

If you use the ORB model in your research or projects, please cite the following work:

Benjamin Rhodes, Sander Vandenhaute, Vaidotas Šimkus, James Gin, Jonathan Godwin, Tim Duignan, and Mark Neumann. Orb-v3: atomistic simulation at scale. arXiv:2504.06231, 2025. https://arxiv.org/abs/2504.06231

BibTeX:

@misc{rhodes2025orbv3atomisticsimulationscale,
  title={Orb-v3: atomistic simulation at scale},
  author={Benjamin Rhodes and Sander Vandenhaute and Vaidotas {\v{S}}imkus and James Gin and Jonathan Godwin and Tim Duignan and Mark Neumann},
  year={2025},
  eprint={2504.06231},
  archivePrefix={arXiv},
  primaryClass={cond-mat.mtrl-sci},
  url={https://arxiv.org/abs/2504.06231},
}