Performing optical logic operations by a diffractive neural network

This paper introduces a novel design strategy for performing optical logic operations using a diffractive neural network. The approach utilizes plane waves as input signals, eliminating the need for precise control of input light properties (phase difference, polarization, intensity, beam size). A compound Huygens' metasurface acts as the diffractive neural network, spatially encoding the input plane wave and decoding it through hidden layers to direct the light to designated output areas representing logic states '1' and '0'. The same metasurface, after training, can perform all seven basic optical logic operations. The concept is experimentally demonstrated at microwave frequencies for NOT, OR, and AND gates.

Chao Qian, Xiao Lin, Xiaobin Lin, Jian Xu, Yang Sun, Erping Li, Baile Zhang, Hongsheng Chen