---
title: "A New Node Centroid Algorithm for Bandwidth Minimization"
authors:
  - "Andrew Lim"
  - "Brian Rodrigues"
  - "Fei Xiao"
doi: ""
url: "https://ijcai.org/proceedings/2003/266"
pdf: "https://ijcai.org/proceedings/2003/0266.pdf"
venue: "Proceedings of the Eighteenth International Joint Conference on Artificial Intelligence"
year: 2003
pages: "1544-1544"
track: ""
keywords:
---

# A New Node Centroid Algorithm for Bandwidth Minimization

Andrew Lim, Brian Rodrigues, Fei Xiao

*Proceedings of the Eighteenth International Joint Conference on Artificial Intelligence*, pages 1544-1544.

## Abstract

We propose a Node Centroid method with HillClimbing to solve the well-known matrix bandwidth minimization problem, which is to permute rows and columns of the matrix to minimize its bandwidth. Many heuristics have been developed for this NP-complete problem including the Cuthill-McKee (CM) and the Gibbs, Poole and Stockmeyer (GPS) algorithms. Recently, heuristics such as Simulated Annealing, Tabu Search and GRASP have been used, where Tabu Search and the GRASP with Path Relinking have achieved significantly better solution quality than the CM and GPS algorithms. Experimentation shows that the Node Centroid method achieves the best solution quality when compared with these while being much faster than the newly-developed algorithms.