Approximating the Number of Attraction Basins of a Function by Means of Clustering and Evolutionary Algorithms