PDE discovery using UBIC

Using the uncertainty-penalized Bayesian information criterion (UBIC) originally proposed in Adaptive Uncertainty-Penalized Model Selection for Data-Driven PDE Discovery to discover governing PDEs.

• Please first install our modified pysindy package archived at this OneDrive link (Password: UBIC). The more updated version is avaiable at this repository.
• Please visit the Example directory to run experiments.
• To use the L0BnB best-subset solver, please install the package.

Citing UBIC

@article{thanasutives2024adaptive,
  author={Thanasutives, Pongpisit and Morita, Takashi and Numao, Masayuki and Fukui, Ken-ichi},
  journal={IEEE Access},
  title={Adaptive Uncertainty-Penalized Model Selection for Data-Driven PDE Discovery},
  year={2024},
  volume={12},
  pages={13165-13182},
  doi={10.1109/ACCESS.2024.3354819}
}

Note

• UBIC-play and PDE-FIND are the development versions used to produce recent results in the paper.
• This repository with the PDE datasets included can be dowloaded at this OneDrive link (Password: UBIC).
• Please refer to the ``3D_exp'' folder for the GS experiment