Research on influence of solving quality based on different initializing solution algorithm in tabu search

Many research results show that TS (tabu search or taboo search) is depended on the selection of initial solutions. For good initial solutions, TS can find better results with quicker speed in the solution space, but poor initial solutions may decrease the TS convergence speed. This paper investigates and compares three common initializing algorithms (greedy, insertion and randomization) to solve the TSP (traveling salesman problem). The experiment shows that if we pursue different solving qualities, and face different problem sizes, we should select different initializing algorithms to generate the initial solutions.