Satisfying solutions to inverse design problems in photonics
Résumé
We have used global optimization algorithms (PSO, Differential Evolution, CMA-ES, Nelder Mead,...) to solve simple well posed photonics problems (how to reflect a given wavelength, reflect a broad range of wavelength, diffract a given wavelength, how to favor the absorption by a photovoltaic device, excite a long range surface mode, separate polarizations, excite a waveguide). Algorithms are run a large number of times allowing to assess the reliability of each solution.