The Fan Lab at Stanford University has developed an ultra-fast, physics-augmented, deep learning enhanced surrogate field solver for high-speed electromagnetic simulation and optimization. Denoising WaveY-Net uses a two-stage approach to target different field error sources.