Saving local searches in global optimization

This thesis concerns the use of local searches within global optimization algorithms. In particular, we focus our attention on the strategies to decide whether to start or not a local search from a starting point. More specifically, our aim is to avoid the waste of computational effort due to local searches which lead to already detected local minima or to local minimizers with a poor function value. Our clustering-based strategies can be easily used to solve large scale global optimization problems and to enhance the performance of any memetic algorithm in general.