An efficient local search heuristic with row weighting for the unicost set covering problem

Research output: Contribution to journalArticlepeer-review

Authors

  • Chao Gao
  • Xin Yao
  • Thomas Weise
  • Jinlong Li

Colleges, School and Institutes

External organisations

  • University of Science and Technology of China
  • USTC-Birmingham Joint Research Institute in Intelligent Computation and Its Applications (UBRI)

Abstract

The Set Covering Problem (SCP) is NP-hard. We propose a new Row Weighting Local Search (RWLS) algorithm for solving the unicost variant of the SCP, i.e., USCPs where the costs of all sets are identical. RWLS is a heuristic algorithm that has three major components united in its local search framework: (1) a weighting scheme, which updates the weights of uncovered elements to prevent convergence to local optima, (2) tabu strategies to avoid possible cycles during the search, and (3) a timestamp method to break ties when prioritizing sets. RWLS has been evaluated on a large number of problem instances from the OR-Library and compared with other approaches. It is able to find all the best known solutions (BKS) and improve 14 of them, although requiring a higher computational effort on several instances. RWLS is especially effective on the combinatorial OR-Library instances and can improve the best known solution to the hardest instance CYC11 considerably. RWLS is conceptually simple and has no instance-dependent parameters, which makes it a practical and easy-to-use USCP solver.

Details

Original languageEnglish
Pages (from-to)750-761
Number of pages12
JournalEuropean Journal of Operational Research
Volume246
Issue number3
Early online date22 May 2015
Publication statusPublished - 1 Nov 2015

Keywords

  • Combinatorial optimization, Row weighting local search, Unicost set covering problem