Table of Contents
Fetching ...

Separable Drawings: Extendability and Crossing-Free Hamiltonian Cycles

Oswin Aichholzer, Joachim Orthaber, Birgit Vogtenhuber

TL;DR

The paper introduces separable drawings, a broad relaxation of pseudospherical drawings where each edge has a witness curve that meets every other edge at most once, and uses this framework to study two classic questions: extendability to simple drawings of $K_n$ and the existence of crossing-free Hamiltonian substructures. It proves that any separable drawing of a graph on $n$ vertices extends to a simple drawing of $K_n$, and that separable drawings of $K_n$ contain a crossing-free Hamiltonian cycle and are plane Hamiltonian connected. It also shows that generalized convex drawings and 2-page book drawings are separable, placing separable drawings strictly above these classes and strengthening recent results on plane Hamiltonicity. Additionally, the work develops a rotation-system perspective, with flips and triangle mutations linking separability to realizability, and analyzes recognition complexity (polynomial-time for simple drawings of $K_n$, NP-complete for general graphs).

Abstract

Generalizing pseudospherical drawings, we introduce a new class of simple drawings, which we call separable drawings. In a separable drawing, every edge can be closed to a simple curve that intersects each other edge at most once. Curves of different edges might interact arbitrarily. Most notably, we show that (1) every separable drawing of any graph on $n$ vertices in the plane can be extended to a simple drawing of the complete graph $K_{n}$, (2) every separable drawing of $K_{n}$ contains a crossing-free Hamiltonian cycle and is plane Hamiltonian connected, and (3) every generalized convex drawing and every 2-page book drawing is separable. Further, the class of separable drawings is a proper superclass of the union of generalized convex and 2-page book drawings. Hence, our results on plane Hamiltonicity extend recent work on generalized convex drawings by Bergold et al. (SoCG 2024).

Separable Drawings: Extendability and Crossing-Free Hamiltonian Cycles

TL;DR

The paper introduces separable drawings, a broad relaxation of pseudospherical drawings where each edge has a witness curve that meets every other edge at most once, and uses this framework to study two classic questions: extendability to simple drawings of and the existence of crossing-free Hamiltonian substructures. It proves that any separable drawing of a graph on vertices extends to a simple drawing of , and that separable drawings of contain a crossing-free Hamiltonian cycle and are plane Hamiltonian connected. It also shows that generalized convex drawings and 2-page book drawings are separable, placing separable drawings strictly above these classes and strengthening recent results on plane Hamiltonicity. Additionally, the work develops a rotation-system perspective, with flips and triangle mutations linking separability to realizability, and analyzes recognition complexity (polynomial-time for simple drawings of , NP-complete for general graphs).

Abstract

Generalizing pseudospherical drawings, we introduce a new class of simple drawings, which we call separable drawings. In a separable drawing, every edge can be closed to a simple curve that intersects each other edge at most once. Curves of different edges might interact arbitrarily. Most notably, we show that (1) every separable drawing of any graph on vertices in the plane can be extended to a simple drawing of the complete graph , (2) every separable drawing of contains a crossing-free Hamiltonian cycle and is plane Hamiltonian connected, and (3) every generalized convex drawing and every 2-page book drawing is separable. Further, the class of separable drawings is a proper superclass of the union of generalized convex and 2-page book drawings. Hence, our results on plane Hamiltonicity extend recent work on generalized convex drawings by Bergold et al. (SoCG 2024).

Paper Structure

This paper contains 2 sections, 2 theorems, 1 figure.

Table of Contents

  1. Introduction
  2. Preliminaries

Key Result

Lemma 3

Let $\gamma_{e}$ be a witness of a separator edge $e$ in a simple drawing $\mathcal{D}$. Then every edge $f$ of $\mathcal{D}$ that connects two vertices on the same side of $\gamma_{e}$ is fully contained in that side.

Figures (1)

  • Figure 1: \ref{['fig:flip-rs']} A rotation system corresponding to a convex straight-line drawing of $K_7$. The only possible flip of the edge $e = \{ 2, 6 \}$ is marked. \ref{['fig:flip-drawing']} As we implicitly show in the proof of \ref{['lem:separator-equivalence']}, \ref{['lem:sep-equiv-item-3']}$\Rightarrow$\ref{['lem:sep-equiv-item-1']}, a flip of $e$ in the rotation system corresponds to redrawing $e$ (the dashed version is after the flip, the solid version before) in any simple drawing realizing the rotation system.

Theorems & Definitions (4)

  • Conjecture 1
  • Conjecture 2
  • Lemma 3
  • Lemma 5