/TSM-PDE

Code of ICML paper arxiv.org/abs/2302.08105

Primary LanguagePython

TSM-PDE

See "A Neural PDE Solver with Temporal Stencil Modeling" for the paper associated with this codebase.

Parts of the codebase is adapted from google/jax-cfd.

TSM Illustration

Setup

conda env create -f environment.yml
conda activate cfd
cd jax-cfd
pip install jaxlib
pip install -e ".[complete]"
cd ..

Data

Both the training and evaluation data can be deterministically generated. Please see the data_generation for more details.

Reproduction

Please check the reproducing_scripts for more details.

Pretrained Checkpoints

Please download the pretrained model checkpoints from here.

Reference

If you found this codebase useful, please consider citing the following papers:

Temporal Stencil Modeling:

@article{sun2023tsm,
  title={A Neural PDE Solver with Temporal Stencil Modeling},
  author={Sun, Zhiqing and Yang, Yiming and Yoo, Shinjae},
  journal={arXiv preprint arXiv:2302.08105},
  year={2023}
}

Learned Interpolation:

@article{Kochkov2021-ML-CFD,
  author = {Kochkov, Dmitrii and Smith, Jamie A. and Alieva, Ayya and Wang, Qing and Brenner, Michael P. and Hoyer, Stephan},
  title = {Machine learning{\textendash}accelerated computational fluid dynamics},
  volume = {118},
  number = {21},
  elocation-id = {e2101784118},
  year = {2021},
  doi = {10.1073/pnas.2101784118},
  publisher = {National Academy of Sciences},
  issn = {0027-8424},
  URL = {https://www.pnas.org/content/118/21/e2101784118},
  eprint = {https://www.pnas.org/content/118/21/e2101784118.full.pdf},
  journal = {Proceedings of the National Academy of Sciences}
}