Extending Thomassen's conjecture to directed graphs

TL;DR

This work introduces and analyzes a directed analogue of Thomassen's conjecture by defining F-avoidability in digraphs via minimum out-degree. It proves that all orientations of the odd cycles $C_3$ and $C_5$ are avoidable, but that certain directed bipartite and tree orientations are not, establishing that a straightforward extension of Thomassen's conjecture to digraphs is false. A complete classification is given for regular-avoidable digraphs: exactly those that are not grounded forests, with a robust probabilistic toolkit (LLL, Chernoff bounds) and structured reductions (3-partite typing, layering, and $d$-out-arborescence gadgets) driving the results. The paper also lays out a set of open problems, including the fate of all odd-cycle orientations and the broader characterization of avoidable digraphs, inviting further development of the theory and techniques in directed graph settings.

Abstract

A famous conjecture by Thomassen from 1983 asserts that for any given $k,g\in \mathbb{N}$ there exists some $d=d(k,g)\in \mathbb{N}$ such that every graph of minimum degree at least $d$ contains a subgraph of minimum degree at least $k$ and girth at least $g$. In this paper, we initiate the systematic study of the directed analogs of Thomassen's conjecture one obtains when replacing minimum degree by minimum out-degree. Concretely, we study which digraphs $F$ are avoidable in the sense that there exists $d_F:\mathbb{N}\rightarrow \mathbb{N}$ such that every digraph of minimum out-degree at least $d_F(k)$ contains an $F$-free subdigraph of minimum out-degree at least $k$. Among our main results, we show that all orientations of $C_3$ and $C_5$ are avoidable, while one-directed orientations of complete bipartite graphs and all oriented trees are not avoidable. This, in particular, shows that the most direct extension of Thomassen's conjecture to digraphs is false. We also fully characterize which digraphs are avoidable when restricting the setting to regular host digraphs. Finally, we raise numerous attractive open problems in the hope of sparking further progress.

Figures (1)

• Figure 1: The four different orientations of $C_5$

Theorems & Definitions (30)

• Theorem 1.1
• Theorem 1.2: Dellamonica, Koubek, Martin and Rödl directed_cycles
• Theorem 1.3
• Theorem 1.4
• Theorem 1.5
• Lemma 2.1: The Lovász Local Lemma LLLshearer1985problem
• Theorem 2.2: noga
• Lemma 2.3: chernoff
• proof : Proof of Theorem \ref{['thm:complete directed pair']}
• Lemma 3.1