S.T.E.P.: the easiest way to optimize a function

Most of the algorithms for global optimization making use of the concept of population exploit very little of the information provided by agents in the population in order to choose the next point to evaluate. We develop a new method called S.T.E.P. (Select The Easiest Point) which determines the next point to evaluate by analysing the usefulness of evaluating the function at a certain position. Moreover, we will see that this method permits one to precisely define the heuristic of the search. We also prove its convergence under certain conditions