Repo contains all the files for optimizing the Second Harmonic Generation of Gallium Nitride with circular, triangular, and rectangular metasurfaces. ".fsp" sample files are also included to run simulations in Lumerical. Here are some sample images of the simulations:
The goal of this project is to optimize Second Harmonic Generation (SHG). SHG is a process by which an input source of frequency omega gets doubled when it interacts with an non-centrosymmetric material (asymmetric about the origin). GaN is useful because it's a dielectric material with a relatively high index of refraction, making it ideal to perform SHG. The project uses Bayesian Optimization, naive reinforcement learning, and random forest regression to learn and train different parameters.