Diffractive interconnects: all-optical permutation operation using diffractive networks

Abstract Permutation matrices form an important computational building block frequently used in various fields including, e.g., communications, information security, and data processing. Optical implementation of permutation operators with relatively large number input–output interconnections based on power-efficient, fast, compact platforms is highly desirable. Here, we present diffractive optical networks engineered through deep learning to all-optically perform operations that can scale hundreds thousands between input output field-of-view using passive transmissive layers are individually structured at the wavelength scale. Our findings indicate capacity network approximating a given operation increases proportional trainable transmission elements system. Such deeper designs pose practical challenges terms physical alignment diffraction efficiency We addressed these by designing misalignment tolerant arbitrarily selected operations, experimentally demonstrated, for first time, operates THz part spectrum. Diffractive might find applications in, image encryption, processing, along telecommunications; especially carrier frequencies wireless communications approaching THz-bands, presented potentially serve as channel routing interconnection panels networks.