pinn

How to run this example

This is an example of showing how PINN (physical informed neural network) works. The PINN is a data-driven solutions of partial differential equations proposed by Raissi et al., 2018.

1. Run generate_model_geometry.py for generating the velocity model and geometry files.
2. In this step, the script forward.py will help you model snapshots of pressure wavefields. The data will be stored in a folder named wavefields. These wavefield are training data for PINN.
3. Run the cells in pinn.ipynb and enjoy your PINN trip.