Tailoring 3D non-Rayleigh speckles via physics-informed neural networks

Optical speckles, a disordered photonic phenomenon that originates from the interference of multiple scattered coherent waves, typically obey Rayleigh statistics. Beyond this classical regime, featuring customizable statistical properties and high-order correlations, non-Rayleigh speckles present potential in both fundamental and applied research. Here, we introduce a physics-informed neural network for designing three-dimensional non-Rayleigh speckle fields. As a proof of concept, we experimentally realize volumetric sub-Rayleigh and propagation-persistent super-Rayleigh speckles on this versatile framework. Our work paves a novel, to the best of our knowledge, way for statistical light manipulation and engineering implementation.