Fast Simulation of Large-Scale Metasurfaces Based on Intelligent Local Periodic Approximation

Metasurfaces have proved powerful in manipulating electromagnetic waves and realizing various functionalities. Better performance and capabilities require larger metasurfaces and more complex meta-elements, thus lifting the computational burden in the design process. Rigorous simulations are intractable for a large metasurface composed of tens of thousands of meta-elements. Fast algorithms with approximations have been developed for efficient simulations and designs of metasurfaces. For example, simply summing up the scattering field caused by each individual meta-element would result in an approximated scattered field without considering the mutual coupling between meta-elements. To partially account for the mutual coupling effects, the idea of local periodic approximation was developed by adding adjacent regions of each meta-element when simulating them individually. Although it effectively reduces the simulation time, the rigorous simulation of local meta-elements with adjacent regions is required at runtime. The forward simulation time remains unaffordable in the design process that needs thousands of iterations.