Efficient evolutionary algorithms for the clustering problem in directed graphs

We present improvements in the performance of standard genetic algorithms (GAs) as regards the solution of highly complex combinatorial optimization problems. These improvements are related to some modifications in the GA, including local search and/or diversification procedures. The performance of each proposed version is evaluated through a graph partitioning problem. Extensive computational experiments show that our evolutionary algorithms outperform a genetic algorithm proposed in the literature, by significantly improving the quality of the final solutions with similar computational times.