Study on the efficiency in the numerical integration of size-structured population models: Error and computational cost

Abstract

We describe a procedure which is useful to select an appropriate numerical method in a size-structured population model. We consider four different numerical methods based on finite difference schemes or characteristics curves integration. We compute an analytical approximation in terms of the discretization parameters for the theoretical error principal terms and the computational cost. Thus, we show the efficiency curve that allows to select the best relationship between the discretization parameters for each numerical method. Finally, we obtain the most efficient numerical method for each test.