Abstract

Accurately forecasting the high-dimensional chaotic dynamics of semiconductor laser (SL) networks is essential in photonics research. In this study, we propose a spatiotemporal multiplexed photonic reservoir computing (STM-PRC) architecture, specifically designed for parallel prediction of the high-dimensional chaotic dynamics in complex SL networks.

This is accomplished by decomposing the prediction task into multiple simplified reservoirs, leveraging the intrinsic topological characteristics of the network. Additionally, we introduce a dimensionality reduction technique for high-dimensional chaotic datasets, which exploits the symmetrical properties of the network topology and cluster synchronization patterns derived from complex network theory.

This approach further simplifies the prediction process and enhances the computational efficiency of the parallel STM-PRC system. The feasibility and effectiveness of the proposed framework are demonstrated through numerical simulations and corroborated by experimental validation. Our results expand the application potential of SL networks in all-optical communication systems and suggest new directions for optical information processing.