Heuristics for matrix bandwidth reduction

In this work, we provide two heuristic algorithms for the matrix bandwidth reduction problem. The first is a genetic algorithm and the second uses node label adjustments. Experiments show these heuristics improve solution quality when compared with the well-known GPS algorithm and recently-developed methods using tabu search and GRASP with Path Relinking. Further, the node adjustment approach obtains solutions at speeds comparable to the fast GPS algorithm.

• Type: Journal
• Year: 2006
• Where: EOR
• Authors: Andrew Lim, Brian Rodrigues, Fei Xiao
• Keywords: Algorithm, EOR 2006, Matrix Bandwidth Reduction, Node Label Adjustments