Hybrid Approaches based on Simulated Annealing and an Exact Algorithm for Mixed Integer Programming Problems

摘要

This paper proposes two hybrid approaches based on the exact algorithm and metaheuristic, simulated annealing, for mixed integer programming problems. The first algorithm tries to control the tradeoff between the computation time and the solution quality by reducing the number of variables of the instances to be solved by the exact algorithm. The variable reduction is performed on the basis of searching by simulated annealing in the hybrid algorithm. The second hybrid algorithm considers not only the problem size reduction but also the computation time constraint of searching for the exact algorithm. The experimental evaluation shows that the exact algorithm and SA complement each other and the second hybrid algorithm performs better than the others.