🔬 SCDM：Discovery of 2D Materials via Symmetry-Constrained Diffusion Model

A symmetry-constrained diffusion model (SCDM) that integrates space group symmetry into the generative process for 2D material discovery.

✨ Key Features

• Symmetry-aware Generation: Incorporates space group symmetry constraints in diffusion process
• Wyckoff Position Processing: Automated data preprocessing for crystallographic data
• 2D Material Focus: Specialized for two-dimensional material discovery

🚀 Getting Started

Prerequisites

• Python 3.9+
• PyTorch 2.0+
• CUDA-enabled GPU (recommended cu121)

Installation

# Clone repository
git clone https://github.com/yourusername/SCDM.git

# Install dependencies
pip install -r requirements.txt

📖 Usage

Step 1: Data Preprocessing

0.Prepare your crystallographic data in cif format under ..SCDM/poscar/

1.Use python csv_create.py for encoding all the data under ..SCDM/poscar/

2.Modify file paths in npz.py

# your data csv here
csv_file = '../SCDM/POSCAR/xxx.csv'
# your poscar data here
directory_path = '../POSCAR'

3.Generate processed dataset: python npz.py --output materials_dataset.npz

Step 2: Model training

python TRAIN.py\
    --dataset materials_dataset.npz 
    --batch_size 24 
    --num_epochs 100 
    --save_dir models

Step 3: Generation

0. Load Your Trained Model before generating new materials, load your trained model:

best_model_path = saved_path + '/' + 'BestModel.pth'
#your trained model here
best_model_path ='../saved_train_models/BestModel.pth'
DDPM.load_state_dict(torch.load(best_model_path)) 

1.Run the generation script to create new 2D materials:

python Generate.py

🤝 Citation

@article{xu2025discovery,
  title={Discovery of 2D Materials via Symmetry-Constrained Diffusion Model},
  author={Xu, Shihang and Chu, Shibing and Mrad, Rami and Zhang, Zhejun and Li, Zhelin and Jiao, Runxian and Chen, Yuanping},
  journal={The Journal of Physical Chemistry C},
  volume={129},
  number={14},
  pages={6794--6802},
  year={2025},
  publisher={ACS Publications}
}

📜 License

This project is licensed under the MIT License.

📞 Contact

If there are any issues and questions, please leave us issues on Github or send me email (Shibing Chu c◎ujs dot edu dot cn).