Solving the graph coloring problem via hybrid genetic algorithms

Let G = (V,E) an undirected graph, V corresponds to the set of vertices and E corresponds to the set of edges, we focus on the graph coloring problem (GCP), which consist to associate a color to each vertex so that two vertices connected do not possess the same color. In this paper we propose a new hybrid genetic algorithm based on a local search heuristic called DBG to give approximate values of χ(G) for GCP. The proposed algorithm is evaluated on the DIMACS benchmarks and numerical results show that the proposed approach achieves highly competitive results, compared with best existing algorithms.