Skip to main navigation Skip to search Skip to main content

NC algorithms for partitioning planar graphs into induced forests and approximating NP-hard problems

  • Tokyo Denki University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

2 Scopus citations

Abstract

It is well known that the vertex set of every planar graph can be partitioned into three subsets each of which induces a forest. Previously, there has been no NC algorithm for computing such a partition. In this paper, we design an optimal NC algorithm for computing such a partition for a given planar graph. It runs in O(log n log* n) time using O(n/(log n log* n)) processors on an EREW PRAM. This algorithm implies optimal NC approximation algorithms for many NP-hard maximum induced subgraph problems on planar graphs with a performance ratio of 3. We also present optimal NC algorithms for partitioning the vertex set of a given K4-free or K2,3-free graph into two subsets each of which induces a forest. As consequences, we obtain optimal NC algorithms for 4-coloring K4-free or K2,3-free graphs which are previously unknown to our knowledge.

Original languageEnglish
Title of host publicationGraph-Theoretic Concepts in Computer Science - 21st International Workshop, WG 1995, Proceedings
EditorsManfred Nagl
PublisherSpringer Verlag
Pages275-289
Number of pages15
ISBN (Print)3540606181, 9783540606185
DOIs
StatePublished - 1995
Event21st International Workshop on Graph-Theoretic Concepts in Computer Science, WG 1995 - Aachen, Germany
Duration: Jun 20 1995 → Jun 22 1995

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1017
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference21st International Workshop on Graph-Theoretic Concepts in Computer Science, WG 1995
Country/TerritoryGermany
CityAachen
Period06/20/95 → 06/22/95

Fingerprint

Dive into the research topics of 'NC algorithms for partitioning planar graphs into induced forests and approximating NP-hard problems'. Together they form a unique fingerprint.

Cite this