A Genetic Algorithm for Spline Least Squares Calculations

In this paper we describe a method for least squares calculations with a spline function using a genetic algorithm (GA). Spline functions, which are piecewise polynomials, are the most successful approximating functions when their knots (the joining points) are considered as variables. This treatment results in a nonlinear least squares problem with many local minima. We apply the GA to search for the global minimum of the fitting criterion AIC (Akaike's Information Criterion). A Numerical example is given.