Deep Convolutional Neural Networks to Predict Mutual Coupling Effects in Metasurfaces

Metasurfaces have provided a novel and promising platform for realizing compact high-performance optical devices. The conventional metasurface design approach assumes periodic boundary conditions each element, which is inaccurate in most cases since near-field coupling effects between elements will change when the element surrounded by nonidentical structures. In this paper, deep learning proposed to predict actual electromagnetic (EM) responses of target meta-atom placed large array with taken into account. predicting neural network takes physical specifications its neighbors as input, calculates phase amplitude milliseconds. This can be used optimize metasurfaces’ efficiencies combined optimization algorithms. To demonstrate efficacy methodology, improvements efficiency beam deflector metalens over are obtained. Moreover, it shown that correlations metasurface's performance errors caused mutual not bound certain (materials, shapes, etc.). As such, envisioned readily applied explore improve various designs.