Performances's study on crossover operators keeping good schemata for some scheduling problems

In this paper, we deal with genetic algorithms (GAs) solving some permutation scheduling problems. Considering only the permutation codes and only the crossover-phase of a GA on a very large population, a comparison study of di erent crossover operators was done earlier according to some performance indicators. Here, we develop a tool generating a large scale of scheduling instances and of permutation codes (i.e. the initial population). We again run only the crossover-phase of a GA on this very large population with some crossover operators, and they are considered independently. This new comparison is made on permutation scheduling problems. Statistics are then computed in order to validate or not the results obtained previously with the performance indicators.