Applying local search to the feedback vertex set problem

Authors:
Philippe Galinier;Eunice Lemamou;Mohamed Wassim Bouzidi
Affiliations:
Ecole Polytechnique de Montréal, Montreal, Canada;Ecole Polytechnique de Montréal, Montreal, Canada;Ecole Polytechnique de Montréal, Montreal, Canada
Venue:
Journal of Heuristics
Year:
2013

Citing 7
Cited 0

A contraction algorithm for finding small cycle cutsets

Journal of Algorithms
Approximation algorithms for the vertex feedback set problem with applications to constraint satisfaction and Bayesian inference

SODA '94 Proceedings of the fifth annual ACM-SIAM symposium on Discrete algorithms
Algorithm 815: FORTRAN subroutines for computing approximate solutions of feedback set problems using GRASP

ACM Transactions on Mathematical Software (TOMS)
Computers and Intractability: A Guide to the Theory of NP-Completeness

Computers and Intractability: A Guide to the Theory of NP-Completeness
Node-and edge-deletion NP-complete problems

STOC '78 Proceedings of the tenth annual ACM symposium on Theory of computing
Computing Minimum Feedback Vertex Sets by Contraction Operations and its Applications on CAD

ICCD '99 Proceedings of the 1999 IEEE International Conference on Computer Design
Approximation algorithms for the loop cutset problem

UAI'94 Proceedings of the Tenth international conference on Uncertainty in artificial intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

The feedback vertex set problem (FVSP) consists in making a given directed graph acyclic by removing as few vertices as possible. In spite of the importance of this NP-hard problem, no local search approach had been proposed so far for tackling it. Building on a property of acyclic graphs, we suggest in this paper a new representation of the solutions of the FVSP (feedback sets). Thanks to this solution representation, we are able to design a local transformation (equivalent to a neighborhood) that changes a feedback set into a new one. Based on this neighborhood, we have developed a simulated annealing algorithm for the FVSP. Our experiments show that our algorithm outperforms the best existing heuristic, namely the greedy adaptive search procedure by Pardalos et al.