1D Burgers equation simulated by PINN
This is an implementation of PINN(s) on TensorFlow 2 to solve Burgers equation (1D Navier-Stokes eq. with no pressure gradient / external force) under Dirichlet boundary condition w/o training data (data to fit initial & boundary conditions need to be provided). This is keras-utilized version unlike other two of my repos (PINN_wave / PINN_von_Karman).
Usage
Simply type
python main.py
to run the entire code. Basic parameters (e.g., network architecture, batch size, initializer, etc.) are found in
params.py
and could be modified depending on the problem setup.
Environment
Tested on
python 3.8.10
with the following:
| Package | Version |
|---|---|
| numpy | 1.22.1 |
| scipy | 1.7.3 |
| tensorflow | 2.8.0 |
Reference
[1] Raissi, M., Perdikaris, P., Karniadakis, G.E.: Physics-informed neural networks: