Hybrid Boundary Element – Physics Informed Neural Network Formulation for Electromagnetics Problems

In this contribution the authors propose a hybrid BEM-PINN method for the resolution of the partial differential equations arising in electromagnetism. The method retains the advantages of both integral methods (compact representation and no need to mesh large domains) as well as differential methods, where the term “differential” is here referred to the Automatic Differentiation implemented during the PINN training. The method is easy to implement and adds an additional flexibility to PINN applications.