Application of real-type tabu search in function optimization problems

An optimization solution performance of tabu search is influenced by initial solution, selection of neighbor solution, and size of tabu list etc. In this paper, we proposed a real-type tabu search (RTS) for function optimization, which uses belief space to create a neighbor solution. Belief space is made of upper 60% neighbors to effectively restrict searching limit, so it can improve searching time and local or global searching capability of RTS. Also short-term and long-term memory based tabu lists adequate to RTS are implemented to search a different region. All of theses procedures are independently applied to each determinant value for quick convergance and effective searching process. In order to show the usefulness of the proposed method, the RTS is applied to the minimization problems such as, De Jong functions, Ackley function, and Griewank functions etc., the results are compared with those of genetic algorithm (GA) or evolutionary programming (EP)