Skip to content

Latest commit

 

History

History
44 lines (26 loc) · 970 Bytes

README.md

File metadata and controls

44 lines (26 loc) · 970 Bytes
Raw

OpenFOAM Workshop 17 Training - Towards physics-based deep learning in OpenFOAM: Combining OpenFOAM with the PyTorch C++ API

Author

Tomislav Maric, MMA, Mathematics Department, TU Darmstadt, [email protected]

Installation

Dependencies

  • OpenFOAM-v2312
  • python Pandas, matplotlib

Installation

   pinnfoam> ./Allwmake

Usage

In pinnfoam/run/unit_box_domain

    unit_box_domain > blockMesh && pinnFoamSetSphere && pinnFoam

Visualization

Run

    unit_box_domain > paraview --state=visualize.pvsm

To view training loss diagrams jupyter notebook and open pinnFoam-visualize-single-training.ipynb', then execute Re-run and Clear Data`.

Grid Search

A primitive Grid Search using python.subprocess is implemented in run/pinnFoam-grid-search.ipynb, just re-run this Jupyter notebook. Visualization of the Grid Search is done by run/pinnFoam-visualize.ipynb.

License

GPL v3.0