# Simulator-based training of generative models for the inverse design of metasurfaces

## TL;DR

This paper introduces a novel population-based global optimization method for metasurfaces, leveraging a generative neural network trained with electromagnetic simulations to efficiently discover high-performance designs.

## Contribution

It presents a new optimization algorithm that combines neural network training with electromagnetic adjoint methods for inverse metasurface design, outperforming traditional topology optimization.

## Key findings

- Generated devices achieve efficiencies comparable or superior to standard methods.
- The neural network distribution shifts towards high-performance regions during training.
- The method is applicable to other gradient-based optimization problems.

## Abstract

Metasurfaces are subwavelength-structured artificial media that can shape and localize electromagnetic waves in unique ways. The inverse design of these devices is a non-convex optimization problem in a high dimensional space, making global optimization a major challenge. We present a new type of population-based global optimization algorithm for metasurfaces that is enabled by the training of a generative neural network. The loss function used for backpropagation depends on the generated pattern layouts, their efficiencies, and efficiency gradients, which are calculated by the adjoint variables method using forward and adjoint electromagnetic simulations. We observe that the distribution of devices generated by the network continuously shifts towards high performance design space regions over the course of optimization. Upon training completion, the best generated devices have efficiencies comparable to or exceeding the best devices designed using standard topology optimization. Our proposed global optimization algorithm can generally apply to other gradient-based optimization problems in optics, mechanics and electronics.