Modified genetic algorithm for inverse design of anti-resonant hollow core fiber with low confinement loss

Low-loss anti-resonant hollow core fibers (AR-HCF) are important for optical communication systems, photonics-enabled sensors, and exploring nonlinear dynamics with gas-light interaction. However, the design of AR-HCF that facilitates optical pulse propagation with desirable properties remains highly intricate and time-consuming. In this work, a modified genetic algorithm (MGA) with a small dataset is presented to reduce the confinement loss (CL) of AR-HCF at the telecom wavelength. The algorithm has been modified through a non-uniform mutation process and a simulated binary crossover method to enhance its searching capability and ensure a stable convergence. The inverse design method for AR-HCF, which combines the algorithm with a full vector finite-element method based on a modal solver, can be implemented to search for approximately optimal designs. In our study, MGA is utilized to optimize the tube diameter and thickness of single-ring tubular AR-HCF and nested anti-resonant nodeless fiber (NANF) at 1550 nm. As a result, optimized ratios of 25.4% and 79.8% are achieved for a core diameter of 40 µm, respectively. In addition, MGA also provides the corresponding optimal range of fiber parameters, which is helpful for actual fabrication.