Table of Contents
Fetching ...

On the structure of (dart, odd hole)-free graphs

Chính T. Hoàng

TL;DR

This work analyzes (dart, odd hole)-free graphs and establishes a three-way structural dichotomy: such a graph is either perfect, has $\alpha(G) \le 2$, or admits a join/co-join partition into two smaller graphs. Leveraging this structure, the authors design a polynomial-time algorithm for optimal graph coloring by combining perfect-graph coloring, maximum-matchings in the complement, and recursive composition in the partition case, and show that these graphs are perfectly divisible with $\chi(G) \le \omega(G)^2$. They also prove that while coloring and large-stable-set computation are tractable, computing the maximum clique and minimum clique cover remain NP-hard for this class. The results advance understanding of the complexity landscape for odd-hole-free and related graph classes and provide practical coloring algorithms for this family, along with open directions such as identifying odd-hole-free graphs that are not perfectly divisible.

Abstract

A hole is a chordless cycle with at least four vertices. A hole is odd if it has an odd number of vertices. A dart is a graph which vertices $a, b, c, d, e$ and edges $ab, bc, bd, be, cd, de$. Dart-free graphs have been actively studied in the literature. We prove that a (dart, odd hole)-free graph is perfect, or does not contain a stable set on three vertices, or is the join or co-join of two smaller graphs. Using this structure result, we design a polynomial-time algorithm for finding an optimal colouring of (dart, odd hole)-free graphs. A graph $G$ is perfectly divisible if every induced subgraph $H$ of $G$ contains a set $X$ of vertices such that $X$ meets all largest cliques of $H$, and $X$ induces a perfect graph. The chromatic number of a perfectly divisible graph $G$ is bounded by $ω^2$ where $ω$ denotes the number of vertices in a largest clique of $G$. We prove that (dart, odd hole)-free graphs are perfectly divisible.

On the structure of (dart, odd hole)-free graphs

TL;DR

This work analyzes (dart, odd hole)-free graphs and establishes a three-way structural dichotomy: such a graph is either perfect, has , or admits a join/co-join partition into two smaller graphs. Leveraging this structure, the authors design a polynomial-time algorithm for optimal graph coloring by combining perfect-graph coloring, maximum-matchings in the complement, and recursive composition in the partition case, and show that these graphs are perfectly divisible with . They also prove that while coloring and large-stable-set computation are tractable, computing the maximum clique and minimum clique cover remain NP-hard for this class. The results advance understanding of the complexity landscape for odd-hole-free and related graph classes and provide practical coloring algorithms for this family, along with open directions such as identifying odd-hole-free graphs that are not perfectly divisible.

Abstract

A hole is a chordless cycle with at least four vertices. A hole is odd if it has an odd number of vertices. A dart is a graph which vertices and edges . Dart-free graphs have been actively studied in the literature. We prove that a (dart, odd hole)-free graph is perfect, or does not contain a stable set on three vertices, or is the join or co-join of two smaller graphs. Using this structure result, we design a polynomial-time algorithm for finding an optimal colouring of (dart, odd hole)-free graphs. A graph is perfectly divisible if every induced subgraph of contains a set of vertices such that meets all largest cliques of , and induces a perfect graph. The chromatic number of a perfectly divisible graph is bounded by where denotes the number of vertices in a largest clique of . We prove that (dart, odd hole)-free graphs are perfectly divisible.
Paper Structure (9 sections, 8 theorems, 1 figure)

This paper contains 9 sections, 8 theorems, 1 figure.

Key Result

Theorem 3.1

Let $G$ be a (dart, odd hole)-free graph. Then one of the following holds for $G$.

Figures (1)

  • Figure 1: The claw, dart and banner

Theorems & Definitions (8)

  • Theorem 3.1
  • Lemma 3.2: Lemma 2.2 in DonXu2022
  • Lemma 3.3: Lemma 2.4 in DonXu2022
  • Theorem 3.4
  • Lemma 3.5: Hoàng, Lemma 7.3 in Hoa2018
  • Theorem 3.6: Dong, Song and Xu DonXu2022
  • Theorem 3.7: Hoàng and McDiarmid HoaMcd2002
  • Lemma 3.8: Hoàng and McDiarmid HoaMcd2002