Inverse Optical Waveguide Analysis: Leveraging Intensity Distributions to Infer Waveguide Structures

We present an innovative approach for optical waveguide analysis using inverse design, focusing on achieving desired mode or intensity distributions. This method overcomes the limitations of conventional mode solvers used for designing structures like optical fibers, which haven’t been harmoniously integrated with gradient-based optimization due to numerous challenges, such as risks of local minima and extensive computational time. Our approach incorporates a uniquely derived cost function focused on a singular variable: the refractive index. It leverages the Adam optimization algorithm to avoid local minima and employs Gaussian mixture models to efficiently control the refractive index range, allowing the optimization of various waveguide structures based solely on mode intensity distributions. Our primary goal is to craft refractive index distributions that match predefined target intensity distributions precisely. We showcase two waveguide designs optimized for Sun-like and Ultra-Flat intensity profiles to illustrate the effectiveness and versatility of our methodology, highlighting its potential to develop high-efficiency optical devices in a compact form.