Equivalent-graded-index photonic-crystal fiber (EGI-PCF) is a modified photonic-crystal fiber (PCF) where the cladding is not required to be periodic. A radial gradient of the equivalent refractive index distribution provides extra freedom to efficiently tailor the dispersion of EGI-PCF. However, such additional complexity makes the manual fiber design much more challenging where the time consumption becomes too much to be afforded eventually. In this paper, we for the first time propose the method of computer-aided inverse fiber design (CAFD) based on empirical dispersion formula to circumvent this hurdle. We apply the differential evolution (DE) genetic algorithm to seek for the optimized EGI-PCF design with a target of specific dispersion profile. In the dispersion oriented EGI-PCF design, the use of empirical formula instead of numerical simulation (e.g., finite-element method) is proved to greatly reduce the run-time of CAFD and simplifies the fitness function evaluation as much as possible. We adopt our method and demonstrate a design of EGI-PCF with a near-zero ultra-flattened dispersion in the whole S+C+L band, which merely costs a few seconds of a laptop to return an optimized fiber design. Thus improvement of efficiency is about three orders of magnitude faster than any previous CAFD attempts as far as we learn. In the end, we use our method to explore an EGI-PCF design with a unique dispersion profile of all-normal dispersion-decreasing along the fiber length centered around 1 μm that plays a key role for the self-similar pulse evolution in a fiber laser/amplifier.
Read full abstract