Table of Contents
Fetching ...

Arbitrary orientations of cycles in oriented graphs

Guanghui Wang, Yun Wang, Zhiwei Zhang

TL;DR

The paper resolves Häggkvist and Thomason's long-standing question by showing that for all large $n$ an oriented graph with minimum semidegree at least $\frac{3n-1}{8}$ contains every possible orientation of a Hamilton cycle, with a tightened bound $\frac{3n-4}{8}$ for nearly directed cycles. The authors leverage a robust-outexpander framework, a delta-extremal partition $(W,X,Y,Z)$, and a combination of probabilistic embedding, the Diregularity Lemma, and the Blow-up Lemma to realize embeddings in both expansion and near-extremal regimes. They split the analysis into two main cases based on sinks in the Hamilton cycle: when $C$ has few sinks, and when $C$ has many sinks, developing two tailored embedding strategies that balance clusters and exploit special edges, respectively. A pancyclicity extension shows that the same threshold yields cycles of every orientation and every length, except for a small class of exceptional graphs. Overall, the work advances exact threshold results for Hamiltonian orientations in oriented graphs and strengthens the toolkit for embedding complex structures via regularity-based methods, with potential transversal generalizations to families of digraphs.

Abstract

We show that every sufficiently large oriented graph $G$ with minimum indegree and outdegree both at least $(3|V(G)|-1)/8$ contains every orientation of a Hamilton cycle. This result improves the approximate bound established by Kelly and resolves a long-standing problem posed by Häggkvist and Thomason in 1995. The degree condition is tight and it can be improved to $(3|V(G)|-4)/8$ for Hamilton cycles that are nearly directed, generalizing a classic result by Keevash, Kühn and Osthus. Additionally, we derive a pancyclicity result for arbitrary orientations. More precisely, the above degree condition suffices to guarantee the existence of cycles of every possible orientation and every possible length unless $G$ is isomorphic to one of the exceptional oriented graphs.

Arbitrary orientations of cycles in oriented graphs

TL;DR

The paper resolves Häggkvist and Thomason's long-standing question by showing that for all large an oriented graph with minimum semidegree at least contains every possible orientation of a Hamilton cycle, with a tightened bound for nearly directed cycles. The authors leverage a robust-outexpander framework, a delta-extremal partition , and a combination of probabilistic embedding, the Diregularity Lemma, and the Blow-up Lemma to realize embeddings in both expansion and near-extremal regimes. They split the analysis into two main cases based on sinks in the Hamilton cycle: when has few sinks, and when has many sinks, developing two tailored embedding strategies that balance clusters and exploit special edges, respectively. A pancyclicity extension shows that the same threshold yields cycles of every orientation and every length, except for a small class of exceptional graphs. Overall, the work advances exact threshold results for Hamiltonian orientations in oriented graphs and strengthens the toolkit for embedding complex structures via regularity-based methods, with potential transversal generalizations to families of digraphs.

Abstract

We show that every sufficiently large oriented graph with minimum indegree and outdegree both at least contains every orientation of a Hamilton cycle. This result improves the approximate bound established by Kelly and resolves a long-standing problem posed by Häggkvist and Thomason in 1995. The degree condition is tight and it can be improved to for Hamilton cycles that are nearly directed, generalizing a classic result by Keevash, Kühn and Osthus. Additionally, we derive a pancyclicity result for arbitrary orientations. More precisely, the above degree condition suffices to guarantee the existence of cycles of every possible orientation and every possible length unless is isomorphic to one of the exceptional oriented graphs.

Paper Structure

This paper contains 22 sections, 37 theorems, 34 equations, 4 figures, 1 table.

Key Result

Theorem 1.1

There exists an integer $n_0$ such that every oriented graph $G$ on $n\geqslant n_0$ vertices with minimum semidegree $\delta^0(G)\geqslant (3n-1)/8$ contains every possible orientation of a Hamilton cycle.

Figures (4)

  • Figure 1: The oriented graph $G$ in Proposition \ref{['PROP-degreesharp']}. The bold edges indicate that all possible edges are present and have the directed shown. The set $W$ has size $\lfloor n/4\rfloor$ and $|X|=|Z|=\lceil n/4\rceil$. Each of $W$ and $Y$ spans an almost regular tournament, that is, the indegree and outdegree of every vertex differ by at most one. The oriented graph induced by $X$ and $Z$ is an almost regular bipartite tournament. Table \ref{['TAB-degreesharp']} in Appendix \ref{['APPSEC-table1']} will help readers to better check every vertex of $G$ has the correct indegree and outdegree.
  • Figure 2: An illustration of how to embed $C$ when it contains few sinks. The black diamonds, white squares and black circles indicate the sources, sinks and normal vertices of $C$, respectively. The white circle on the path $L_2$ in (b) indicates that it is a bad vertex of $G$. In the proof we use directed 13-paths to cover bad vertices but here we use a 5-path $L_2$ for an illustration. Note that the blue fat directed path $R_1$ has the same direction as the cycle $C$ but the red dashed directed path $R_2$ has the opposite direction.
  • Figure 3: An illustration of how randomised embedding method may be applied to $G$.
  • Figure 4: An illustration of how to cover all vertices in $W^\prime\backslash(B\cup U)$ by embedding $P_W$. The path in $G[W^\prime\backslash(B\cup U)]$ is the auxiliary path $P_W^{\ast}$ obtained from $P_W$ by replacing several short subpaths with edges.

Theorems & Definitions (73)

  • Theorem 1.1
  • Proposition 1.2
  • Theorem 1.3: kellyJCTB100
  • Theorem 1.4: kellyEJC18
  • Theorem 1.5
  • Theorem 4.1: Chernoff Bound 1
  • Lemma 4.2: debiasioEJC22
  • Theorem 4.3: Chernoff Bound 2
  • Lemma 4.4
  • proof
  • ...and 63 more