| Popis: |
Evolutionary algorithms are very popular and are frequently applied to many different optimization problems. Reasons for this success include that methods of this kind are of global nature, very robust, and only require minimal assumptions on the optimization problem. It is also known that such methods need quite a few resources to generate accurate approximations of the solution sets. As a remedy, researchers have used hybrid (or memetic) algorithms, i.e., evolutionary algorithms coupled with local search for which mainly techniques from mathematical programming are utilized. Such hybrids typically yield satisfying results, the problem, however, remains that the algorithms are relatively expensive since the gradients have to be computed or approximated at each given candidate solution that is designated for local search. |
