Linear Information Forwarding in Highly Nonlinear Fibers
Glitta R. Cheeran, Sobhi Saeed, Bennet Fischer, Mario Chemnitz · 2025
Optical neural networks are an emerging field at the intersection of neuroscience, artificial intelligence, and optics, aiming to replicate the human brain's functionality in optical systems. Highly nonlinear fibers (HNLFs) have recently been demonstrated as a powerful optical system that mimics multi-layer neuromorphic computing [1]. However, in such nonlinear systems, the information encoded in the input is dissipated and cannot be retrieved at the output. Here, we explore a new phenomena called periodic spectral peaking [2] for its capability to preserve phase-encoded information during nonlinear transformations in fibers. In this phenomenon a narrow spectral dip on a pulse turns into a sharp, intense peak through nonlinear interference. We investigated this process by introducing narrow-band phase shifts on the input spectra of femtosecond pulses normal-dispersion and their propagation through HNLF in simulation and experiment.