Abstract |
: |
Integer Programming-based Local Search (IPbLS) is a kind of local search. IPbLS is based on the first-choice hill-climbing search and uses integer programming to generate a neighbor solution. IPbLS has been applied to solve various NP-hard combinatorial optimization problems like knapsack problem, set covering problem, set partitioning problem, and so on. In this paper, we investigate the effect of problem reduction in the IPbLS experimentally using the n-queens maximization problem. The characteristics of IPbLS are examined by comparing IPbLS using strong problem reduction with IPbLS using weak problem reduction, and also IPbLS is compared with other local search strategies like simulated annealing. Experimental results show the importance of problem reduction in IPbLS. |