An effective direct-search algorithm for inverse design of fiber grating filter

In this paper, a new approach to the solution of fiber grating synthesis problems is developed. The proposed method is an optimization approach based on the evolutionary programming (EP) algorithm. When compared to the genetic algorithms for fiber grating synthesis, EP algorithm only uses the mutation process of continuous variables and without using the binary coding and crossover processes. We find that such a simplified evolutionary algorithm may have resulted in a higher convergence velocity as well as higher reliability. To verify the feasibility and the effectiveness of the algorithm, several practical filters have been designed and analyzed.