Robust and Scalable Flat‐Optics on Flexible Substrates via Evolutionary Neural Networks

Abstract

Flexible Flat-Optics This work, presented in article 2100105 by Andrea Fratalocchi and co-workers, implements an artificial intelligent, end-to-end pipeline for the experimental realization of flexible, broadband ultra-flat optics for vectorial light control. The system takes advantage of “physical” neural networks, engineered into suitably designed optical nanoresonators. The design framework also considers fabrication robustness, furnishing devices that maintain their optical response even under strong stress perturbations and deformations.