Numerical Simulations for Fitting Parameters of Linear and Logistic-Type Fractional-, Variable-Order Equations - Comparision of Methods.

In the work variable-, fractional-order backward difference of the Grunwald-Letnikov type is presented. The backward difference is used to generate simulated experimental data to which additional noise signal is added. Using prepared data four different algorithms of finding the parameter of the order function (assuming that the general family of the function is known) and constant \(\lambda \) coefficient are compared. The algorithms are: trust region algorithm, particle swarm algorithm, simulated annealing algorithm and genetic algorithm.