Vector analog computing via on-demand metasurface dispersive polarization transformation
Hui Yang, Jie Xu, Meiyu Peng, Hairong He, Yuting Jiang, Dian Yu, Rui‐Bo Jin, Yingjie Gu, Yueqiang Hu, Huigao Duan, Hui Jing
Abstract
Optical analog computing can potentially feature high-throughput parallel processing with ultralow power and high speed and is promising for efficient signal processing. Previous platforms have mainly focused on scalar computing with optical intensities, which is highly sensitive to environmental disturbance and has been primarily restricted to single or basic computations because of intrinsic fixed correlation between the input and output signals. To our knowledge, for the first time, we use polarization vectors for optical analog computing with a single-layered metasurface to overcome these restrictions. The underlying mechanism is on-demand polarization transformation on the dispersive Poincaré spheres, constructing intrinsic variable correlations between the incident polarization vectors and output signals. We choose the universal logical gates and mathematical function operations as two specific examples. Experimental results of our vector computing metadevices exhibit minimal errors relative to the target values. Our work opens up an avenue for advanced optical signal processing across both classical and quantum domains.