The framework of the model is shown in the figure below:
This code was used to predict phonon DOS using our developed CATGNN model which was used to produce the following work:
- Al-Fahdi, M., Lin, C., Shen, C., Zhang, H., & Hu, M. Rapid prediction of phonon density
of states by crystal attention graph neural network and High-Throughput screening of candidate
substrates for wide bandgap electronic cooling. Materials Today Physics, 2025, 101632.
Note: that the correct figures 1-3 in the above work are in the corrigendum in the following link. - please cite the above work if you use the code
the following packages are required to run the code:
torch=2.5.1
torch-geometric=2.6.1
torch-scatter=2.1.2
e3nn=0.5.1
Jarvis-tools=2024.10.30
scikit-learn=1.2.2
other versions might work, but those versions were successful in running the code
1- untar the data directory by running:
tar -xvzf data.tar
2- you can edit the model parameters from the file "model_params.yaml" and the parameters should be straightforward to edit.
3- you can simply run the following line to run the code:
python main.py
- Please consider reading my published work in Google Scholar using this link thank you :)
- also please let me know if more features are needed to be added and/or improved