A memetic algorithm for graph coloring

All a algorithm proposed ;EÞwith G corresponds independen coloring as a canonica in [7] [17] as greedy constructive heuristics are often used to generate initial solutions for advanced metaheuristic algorithms. On the other hand, local search based metaheuristic algorithms have been widely used to tackle the graph coloring problem in col algorithm which is the first application of Tabu Search to graph This paper presents MACOL, a hybrid metaheuristic algorithm integrating a tabu search procedure with an evolutionary algorithm for solving the graph coloring problem. In this algorithm, we highlight the importance of the diversity of individuals and the balance between intensification and diversification.

To achieve this goal, we devise an adaptive multi-parent crossover operator, which is an extension of the greedy partition crossover (GPX) presented in [14]. Furthermore, we highlight the diversity of the population by defining a new replacement criterion for the * Corresponding author. Tel.: +33 2 41 73 52 94. E-mail addresses: zhipeng.lui@gmail.com, lu@info.univ-angers.fr (Z. Lü), hao@ European Journal of Operational Research 203 (2010) 241–250 Contents lists available European Journal of Operational info.univ-angers.fr (J.-K. Hao). [20]. Due to its NP-Hardness, heuristics and metaheuristics are pre- ferred to handle large and hard coloring problems. We briefly re- view below some of the most representative algorithms. The first heuristic approaches to solving the graph coloring problem were based on greedy construction, which color the verti- ces of the graph one by one guided by a predefined greedy func- tion. These algorithms are very fast by nature but their quality is generally unsatisfactory. The best known algorithms in this class are the largest saturation degree heuristic (DSATUR) [3] and the recursive largest first heuristic (RLF) [28]. In...

Website: smart.hust.edu.cn | Filesize: -
No of Page(s): 10
Download A memetic algorithm for graph coloring.pdf