Infusing Self-Consistency into Density Functional Theory Hamiltonian Prediction via Deep Equilibrium Models

📌 Introduction

Infusing Self-Consistency into Density Functional Theory Hamiltonian Prediction via Deep Equilibrium Models.

🚀 Quickstart

Install dependencies

# clone project
git clone https://github.com/Zun-Wang/DEQHNet.git
cd DEQHNet

# [OPTIONAL] create conda environment
[Optional] conda create -n DEQHNet python=3.10
[Optional] conda activate DEQHNet

# Recommed to install part of dependencies in advance
# Take `cuda121` version as an example
pip install torch==2.1.2 --index-url https://download.pytorch.org/whl/cu121
pip install pyg_lib torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.1.0+cu121.html
pip install torch_geometric==2.3.0

pip install pytorch-lightning==1.8.3

pip install pyscf==2.2.1
conda install psi4 python=3.9 -c conda-forge

pip install requirements.txt

pip install -e .

Train DEQHNet, e.g.,

cp auxiliary.gbs src/QHNet/
cd src/QHNet/
set basis AUXILIARY
python src/QHNet/train_wH.py dataset=uracil model=QHNet model.version=DEQHNet

Citation

@inproceedings{wang2024infusing,
  title={Infusing Self-Consistency into Density Functional Theory Hamiltonian Prediction via Deep Equilibrium Models},
  author={Wang, Zun and Liu, Chang and Zou, Nianlong and Zhang, He and Wei, Xinran and Huang, Lin and Wu, Lijun and Shao, Bin},
  booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems},
  year={2024},
  url={https://openreview.net/forum?id=PSVkinBs4u}
}

Acknowledgements

This project is based on the repo AIRS.