When replicating this paper Hybrid Evolutionary Algorithms for Graph Coloring,P. Galinier and J.K. Hao. Hybrid evolutionary algorithms for graph coloring. Journal of Combinatorial Optimization, 3(4):379–397, 1999. ,
I found that when α=0.6(which is considered∈(0,1)) as the paper said, it didn't work as I expected. That value seems to make the tabu tenure too small to avoid being stuck at a plateau.
It couldn't even fix dsjc0125.1(k=5) or dsjc0125.5(k=17), which were fixed when I increased α to 3. For dsjc0250.5, I tested different values of α and found that α=5 may perform best. Besides, none of the value perform as good as the paper said.
I am just a sophomore so did I make a mistake in my program? Or α really needs to be bigger?May someone who once replicated this paper provide me with a copy?