Global optimization of complex optical structures using Baysian optimization based on Gaussian processes

TL;DR

This paper introduces a Bayesian optimization method using Gaussian processes to efficiently optimize complex optical structures, demonstrated through shaping a reflective metasurface to achieve desired diffraction properties.

Contribution

It applies Bayesian optimization with Gaussian processes to optical structure design, improving efficiency over traditional parameter scans and comparing multiple approaches.

Findings

Bayesian optimization accelerates optical structure design.

Different acquisition functions and kernels impact optimization performance.

Method reduces computational cost of optical simulations.

Abstract

Numerical simulation of complex optical structures enables their optimization with respect to specific objectives. Often, optimization is done by multiple successive parameter scans, which are time consuming and computationally expensive. We employ here Bayesian optimization with Gaussian processes in order to automatize and speed up the optimization process. As a toy example, we demonstrate optimization of the shape of a free-form reflective meta surface such that it diffracts light into a specific diffraction order. For this example, we compare the performance of six different Bayesian optimization approaches with various acquisition functions and various kernels of the Gaussian process.