Design of Metasurface with Near Couplings Based on Deep Learning Driven Space Mapping

Near couplings will lead to low efficiency electromagnetic waves manipulation of the metasurface. In this paper, a space mapping design framework driven by deep learning is proposed for the optimization of metasurafce with near couplings meta-atoms. The deep learning neural network is training to predict the phase of the atom accurately, in which process the near couplings of the designed meta-atoms are considered. The deep learning and the full wave simulation are employed as the coarse and fine models of the spacing mapping algorithm. Numerical results verify the effectiveness of the optimization framework.