A linear kernel for co-path/cycle packing

  • Authors:

  • Zhi-Zhong Chen;Michael Fellows;Bin Fu;Haitao Jiang;Yang Liu;Lusheng Wang;Binhai Zhu

  • Affiliations:

  • Department of Mathematical Sciences, Tokyo Denki University, Hatoyama, Saitama, Japan;The University of New Castle, Callaghan, NSW, Australia;Department of Computer Science, University of Texas-American, Edinburg, TX;Department of Computer Science, Montana State University, Bozeman, MT;Department of Computer Science, University of Texas-American, Edinburg, TX;Department of Computer Science, City University of Hong Kong, Kowloon, Hong Kong;Department of Computer Science, Montana State University, Bozeman, MT

  • Venue:
  • AAIM'10 Proceedings of the 6th international conference on Algorithmic aspects in information and management
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Bounded-Degree Vertex Deletion is a fundamental problem in graph theory that has new applications in computational biology. In this paper, we address a special case of Bounded-Degree Vertex Deletion, the Co-Path/Cycle Packing problem, which asks to delete as few vertices as possible such that the graph of the remaining (residual) vertices is composed of disjoint paths and simple cycles. The problem falls into the well-known class of 'node-deletion problems with hereditary properties', is hence NP-complete and unlikely to admit a polynomial time approximation algorithm with approximation factor smaller than 2. In the framework of parameterized complexity, we present a kernelization algorithm that produces a kernel with at most 37k vertices, improving on the super-linear kernel of Fellows et al.'s general theorem for Bounded-Degree Vertex Deletion. Using this kernel, and the method of bounded search trees, we devise an FPT algorithm that runs in time O*(3.24k). On the negative side, we show that the problem is APX-hard and unlikely to have a kernel smaller than 2k by a reduction from Vertex Cover.