A (partial) differential equation is an equation that relates one or more functions and their (partial) derivatives. Such equations play a prominent role in many disciplines including engineering, physics, economics, and biology.
This project was built based on Physics Informed Deep Learning (Part I): Data-driven Solutions of Nonlinear Partial Differential Equations.
The method is completely data-driven, since no analytical solution is required. The needed data for the ANN training are created during the process. During the training, the ANN minimizes the boundary condition loss function, applying a regularisation term related to the differential equation.
In the context of this project, PyTorch library was used.