2019/12/18 by Flávio José Mendes Coelho, Coelho, Flávio José Mendes
Decision Sciences · #Artificial Intelligence (cs.AI) #Business and Management Studies #FOS: Computer and information sciences #Other Computer Science (cs.OH)
paper · pdf · doi:10.48550/arxiv.1912.11533
openalex publication_date 2019/12/18 · openalex created_date 2024/04/10 · openalex updated_date 2026/07/28
A classic graph coloring problem is to assign colors to vertices of any graph so that distinct colors are assigned to adjacent vertices. Optimal graph coloring colors a graph with a minimum number of colors, which is its chromatic number. Finding out the chromatic number is a combinatorial optimization problem proven to be computationally intractable, which implies that no algorithm that computes large instances of the problem in a reasonable time is known. For this reason, approximate methods and metaheuristics form a set of techniques that do not guarantee optimality but obtain good solutions in a reasonable time. This paper reports a comparative study of the Hill-Climbing, Simulated Annealing, Tabu Search, and Iterated Local Search metaheuristics for the classic graph coloring problem considering its time efficiency for processing the DSJC125 and DSJC250 instances of the DIMACS benchmark.