DNNsolver: Accurate and Efficient Deep Learning Modeling of Polarization-Sensitive Diffractive Neural Networks

DNNsolver proposes a deep learning-based method for accurately and efficiently modeling polarization-sensitive diffractive neural networks composed of cascaded diffractive layers. The approach addresses the limitation of conventional thin element approximation (TEA) which fails to capture electromagnetic coupling within layers, while avoiding the computational expense of full-wave FDTD simulations. This research contributes to the field of optical computing by providing an efficient modeling solution for high-density, polarization-sensitive diffractive optical systems.