Denoising diffusion probabilistic models for replica exchange MD simulation

Implementation of Denoising Diffusion Probabilistic Model for data from replica exchange MD. This implementation was transcribed from the Tensorflow version Ho, J., Jain, A. and Abbeel, P., 2020. arXiv:2006.11239. and a modified Pytorch version here.

pip3 install einops
pip3 install pillow
pip3 install torchvision
pip3 install tqdm

To train the model, you can run the run_training.py after modifying the parameters insde the script.

python run_training.py

To use the trained model to generate figures, you can run the gen_sample.py after modifying the parameters insde the script.

python gen_sample.py

A demo of using this model can be found in DPPM_REMD_SciML.ipynb.

@article{wang2021denoising,
  title={Denoising diffusion probabilistic models for replica exchange},
  author={Wang, Yihang and Tiwary, Pratyush},
  journal={arXiv preprint arXiv:2107.07369},
  year={2021}
}

@misc{ho2020denoising,
    title={Denoising Diffusion Probabilistic Models},
    author={Jonathan Ho and Ajay Jain and Pieter Abbeel},
    year={2020},
    eprint={2006.11239},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}