Physics-informed neural network for simulation of guided waves in materials
Solution of PDEs in complex media is challenging.
Wave propagation in complex media is challenging because of their high frequency nature.
In literature FEM is used, FEM demands that the mesh size should be smaller than the wavelength of the propagation. Smaller mesh size means higher computational time.
PINNs utilize the universal nature of polynomial functions as surrogate to solve the partial differential equations.
Parallel processing nature of neural nets on GPUs make it faster than Serial processing nature of Newmark beta on CPUs.