Neural Networks for wave propagation
Neural-Networks, Wave-Propagation, PINN, Neural-Operators
The study of wave propagation is at the core of applications in several areas of physics, ranging from quantum theory
to applications for reservoir monitoring in geophysics. Despite being widely investigated and applied over the last few decades,
some points related to the cost associated with the computational simulation of wave propagation in non-homogeneous media,
the application in irregular domains, and the solution in the presence of corrupted data still prove to be difficult today and are
studied in a number of scientific fields. In light of this, recent advances in machine learning and, more specifically, in neural
networks have been shown to be potential tools to assist in complex tasks such as these. By means of physics-informed neural
networks and neural operators, we study here distinct applications for wave propagation where we include already known
information about the properties of these solutions in order to improve the networks' learning.