A deep neural network for general scattering matrix

Abstract The scattering matrix is the mathematical representation of characteristics any scatterer. Nevertheless, except for scatterers with high symmetry like spheres or cylinders, does not have analytical forms and thus can only be calculated numerically, which requires heavy computation. Here, we developed a well-trained deep neural network (DNN) that calculate without at speed thousands times faster than finite element solvers. Interestingly, obtained from DNN inherently satisfies fundamental physical principles, including energy conservation, time reversal reciprocity. Moreover, inverse design based on made possible by applying gradient descent algorithm. Finally, demonstrate an application DNN, to desired properties under special conditions. Our work proposes convenient solution learning problems.