Efficient iterated greedy for the two-dimensional bandwidth minimization problem

• Published in: European journal of operational research Vol. 306; no. 3; pp. 1126 - 1139
• Main Authors: Cavero, Sergio, Pardo, Eduardo G., Duarte, Abraham
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.05.2023
• Subjects: Bandwidth; Combinatorial optimization; Graph layout problem; Heuristics; Iterated greedy; Iterated greedy; Combinatorial optimization; Heuristics; Graph layout problem; Bandwidth
• ISSN: 0377-2217; 1872-6860
• DOI: 10.1016/j.ejor.2022.09.004

•Heuristic algorithms for the Two-Dimensional Bandwidth Minimization Problem (2DBMP).•New greedy strategies for the construction of efficient solutions.•Iterated greedy for the 2DBMP.•Favorably empirical comparison with previous algorithms. Graph layout problems are a family of combinatorial optimization problems that consist of finding an embedding of the vertices of an input graph into a host graph such that an objective function is optimized. Within this family of problems falls the so-called Two-Dimensional Bandwidth Minimization Problem (2DBMP). The 2DBMP aims to minimize the maximum distance between each pair of adjacent vertices of the input graph when it is embedded into a grid host graph. In this paper, we present an efficient heuristic algorithm based on the Iterated Greedy (IG) framework hybridized with a new local search strategy to tackle the 2DBMP. Particularly, we propose different designs for the main IG procedures (i.e., construction, destruction, and reconstruction) based on the trade-off between intensification and diversification. Additionally, the improvement method incorporates three advanced strategies: an efficient way to evaluate the objective function of neighbor solutions, a tiebreak criterion to deal with “flat landscapes”, and a neighborhood reduction technique. Extensive experimentation was carried out to assess the IG performance over state-of-the-art methods, emerging our approach as the most competitive algorithm. Specifically, IG finds the best solutions for all instances considered in considerably less execution time. Statistical tests corroborate the merit of our proposal.