Hybrid ant models with a transition policy for solving a complex problem

Combinatorial problems arising in diverse application domains and the growing complexity of many real-world situations emphasize the need for efficient solving methods. One such problem is the bandwidth minimization problem having broad applications in engineering, science, logistics or information recovery. This well-knownNP-complete problem refers to the minimization of the bandwidth for non-zero entries of a sparse symmetric matrix by permuting its rows and columns. A new ant-based approach to matrix bandwidth minimization is proposed. The introduced model is based on the hybridization of the Ant Colony System technique with new problem-tailored local search mechanisms. Computational experiments show a good behaviour of the proposed method for the considered set of bandwidth minimization problem instances.