PINN(s): Physics-Informed Neural Network(s) for von Karman vortex street
This is an implementation of PINN(s) on TensorFlow 2 to learn the flow field of von Karman vortex street, and estimate the fluid density and kinemetic viscosity.
Usage
Simply type
python main.py
to run the entire code in
src
directory. This will load data in
input
and starts training to estimate the fluid density (rho), and kinematic viscosity (nu). 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: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, Journal of Computational Physics, Vol. 378, pp. 686-707, 2019. (paper)
