CIGNN

CIGNN (Charge Integrated Graph Neural Network) is a machine learning potential model designed to predict atomic charges, energies, and forces.

This model enables the generation of customized potentials tailored for specific materials, offering high flexibility and accuracy for advanced materials simulations.

---

Installation and Requirements

Install from PyPI

This is the recommended method for installing CIGNN:

git clone https://github.com/CNMD-POSTECH/CIGNN
cd CIGNN

conda env create -f setup.yaml
conda activate cignn_env
pip install --upgrade pip
pip install -e .

Ensure you update the setup.yaml file to match your system's configuration. Specify the appropriate versions of torch and torch-scatter based on your system's environment and hardware.

If you need to install a specific version of PyTorch, refer to the official PyTorch installation guide and choose the options that are compatible with your system.

---

Usage

Before running the commands, make sure to set the PYTHONPATH environment variable:

export PYTHONPATH=$(pwd)

Train Charge

To train a CIGNN (CPM) model, use the following command:

cignn-train-q
    --config=./script/run_train_q.yaml

Train Energy & Force

To train a CIGNN (EFPM) model, use the following command:

cignn-train
    --config=./script/run_train.yaml

Prediction

To run prediction with a trained model, use the following command:

cignn-predict
    --config=./script/run_predict.yaml

Make sure to adjust the YAML file to fit your specific dataset and requirements.

The example datasets are located in the script/example_dataset folder, and the example models are located in the example-1 folder.

Pretrained Models

Pretrained model files are available for immediate use without additional training:

• Charge model (CPM): ./modelset/q.pth.tar
• Energy & Force model (EFPM): ./modelset/ef.pth.tar

---

Molecular Dynamics

To run MD with a trained model, follow these steps:

1. Navigate to the cnmp folder:

2. Read the README.md file inside the cnmp folder for detailed instructions:

---

Error

If you encounter issues during installation or usage, please refer to the error.md file for troubleshooting.
If the issue persists, feel free to open an issue in this GitHub repository.

---

References

If you use CIGNN in your research, please consider citing the following work:

• Hyo Gyeong Shin, Seong Hun Kim, Eun Ho Kim, Jun Hyeong Gu, Jaeseon Kim, Seon-Gyu Kim, Shin Hyun Kim, Youngjun Park, Hyo Kim, Sunghyun Kim, Duk-Hyun Choe, Donghwa Lee.
  Charge Integrated Graph Neural Network-based Machine Learning Potential for Amorphous and Non-stoichiometric Hafnium Oxide

---

Contact

For inquiries, feel free to reach out to:

• Hyo Gyeong Shin (hyogyeong@postech.ac.kr)

For bug reports or feature requests, please open an issue on GitHub Issues.

---

License

CIGNN is distributed under the MIT License.

---

Contributors

This repository includes contributions from Hyo Gyeong Shin, Seong Hun Kim, Donghwa Lee, and other collaborators.