hPINN: Physics-informed neural networks with hard constraints

The source code for the paper L. Lu, R. Pestourie, W. Yao, Z. Wang, F. Verdugo, & S. G. Johnson. Physics-informed neural networks with hard constraints for inverse design. SIAM Journal on Scientific Computing, 43(6), B1105-B1132, 2021.

Code

The code depends on the deep learning package DeepXDE v0.9.1. If you want to use the latest DeepXDE, you need to modify the code.

Holography

• hPINN
• FDFD: Jupyter Notebook
• FEM: Jupyter Notebook

Fluids in Stokes flow

• hPINN

Cite this work

If you use this code for academic research, you are encouraged to cite the following paper:

@article{lu2021physics,
  author  = {Lu, Lu and Pestourie, Raphael and Yao, Wenjie and Wang, Zhicheng and Verdugo, Francesc and Johnson, Steven G},
  title   = {Physics-informed neural networks with hard constraints for inverse design},
  journal = {SIAM Journal on Scientific Computing},
  volume  = {43},
  number  = {6},
  pages   = {B1105-B1132},
  year    = {2021},
  doi     = {10.1137/21M1397908}
}

Questions

To get help on how to use the code, simply open an issue in the GitHub "Issues" section.