Machine-Learning-Enhanced Polarization Splitter in Silicon-Integrated Dual-Core Photonic Crystal Fiber

In this research, we propose a novel design for a compact polarization fiber using a dual-core hexagonal Photonic Crystal Fiber (PCF) approach. The primary goal is to optimize the structural parameters of the PCF to attain a coupling length ratio of 2 at a wavelength of 1.55 μm, thereby enhancing the device’s efficiency. In a pioneering approach, we incorporate machine-learning techniques to optimize the structural parameters. The finite element method (FEM) is employed in synergy with machine-learning algorithms, enabling automated and efficient fine-tuning of the structural parameters to achieve the desired coupling length ratio. The envisioned advantages of this innovative design encompass a compact device size, reduced fiber length, and superior performance, underscoring the potential for substantial advancements in photonic fiber technology.