Influence of nonlinear node operating states on the performance of photonic reservoir computing
Tong Zhao, Baofeng Feng, 崔天培, Zhimin Bai, Pengfa Chang, Lijun Qiao, Xiaopeng Fan · Optics Express · 2026
Photonic reservoir computing (PRC) has attracted significant attention in artificial intelligence and information processing owing to its advantages of high speed, low power consumption, and inherent parallelism. The operating state of the nonlinear node strongly influences the computational performance of the PRC, yet most existing studies rely almost exclusively on the traditional operating point located in the steady-state region before the Hopf bifurcation. In this work, we adopt a dual-delay optical feedback structure to create another bifurcation pathway in which the system can transition directly from the steady state toward a quasi-periodic regime, bypassing the periodic state. We compare the steady state located before this quasi-periodic bifurcation with the conventional pre-Hopf steady state, and analyze their respective impacts on reservoir performance. The results highlight the importance of selecting an appropriate operating state for specific tasks and provide new insights for nonlinear optimization in PRC systems.