Diffusion-Based Electromagnetic Inverse Design of Scattering Structured Media

We present a conditional diffusion model for electromagnetic inverse design thatgenerates structured media geometries directly from target differential scatteringcross-section profiles, bypassing expensive iterative optimization. Our 1D U-Netarchitecture with feature-wise linear modulation learns to map desired angularscattering patterns to $2\times2$ dielectric sphere structure, naturally handling the non-uniqueness of inverse problems by sampling diverse valid designs. Trained on11,000 simulated metasurfaces, the model achieves median MPE below 19% onunseen targets (best: 1.39%), outperforming CMA-ES evolutionary optimizationwhile reducing design time from hours to seconds. These results demonstrate thatemploying diffusion models is promising for advancing electromagnetic inverse design research, potentially enabling rapid exploration of complex metasurface architectures and accelerating the development of next-generation photonic and wirelesscommunication systems. The code is publicly available at https://anonymous.4open.science/r/inverse_design_scattering_metasurface-BDE8.