Structure oriented search algorithm to solve knapsack problems

In this paper, we analyze the landscape of knapsack problems and propose a new search method called structure oriented search algorithm (SOSA). The heuristic search methods such as a simulated annealing and a genetic algorithm can be applied to complex optimization problems such as combinatorial ones. However, these methods are often redundant and inefficient. The key idea of the SOSA is approximating the landscape structure of the search space, so that the sampling domain can be directly moved to the promising region. The numerical experiments to solve knapsack problems have been conducted and the results are compared with that of genetic algorithms. The experimental results show the effectiveness of our method from the viewpoint of accuracy and calculation cost.