Performance Study of Different Evolutionary Algorithms for Coding Metasurface Design

Numeric encoding of metasurfaces can be seamlessly integrated with evolutionary algorithms to achieve intelligent design of metasurfaces. In this paper, we present a fully automated design approach for the design of reflective coding metasurfaces. In addition, we conduct a comparative performance study for three distinct evolutionary algorithms (i.e, BPSO, DE, and GA) in the optimization of two pixelized patterns corresponding to two coding units for a passive coding metasurface. Our investigations show that the BPSO algorithm achieves the most favorable performance for the design optimization task of coding metasurfaces under consideration.