Physics-informed deep-learning applications to experimental fluid mechanics

Introduction

The code in this repository features a Python implemention of Physics-informed neural networks (PINNs) for applications to experimental fluid mechanics.

More details about the implementation and results from the training are available in "Physics-informed deep-learning applications to experimental fluid mechanics", Hamidreza Eivazi, Yuning Wang, Ricardo Vinuesa(2024,Measurement Science and Technology)

Pre-requisites

The code was run successfully using Tensorflow>=2.6.0, using 1 GPU for training. In addition, scipy is necessary for implementing optimization algorithm

Data

The dataset used for training and testing are available in order to ensure the reproducibility of the results.

Please, get in touch using the email address for correspondance in the paper to arrange the transfer.

Training and inference

The PINNs training and prediction can be performed after cloning the repository.

Take cylinder_wake as example:

git clone  https://github.com/KTH-FlowAI/Physics-informed-deep-learning-applications-to-experimental-fluid-mechanics.git
cd cylinder_wake/training
python training.py

For postprocessing the reults, it can be performed as follows:

cd ../post-processing
python plot_cylinder_pod.py

Note that for running the 2D_Burgers, the data is required to be generated by running the code. To do this:

cd 2D_Burgers
python Burgers_2d_solve.py