Planarity of Streamed Graphs

TL;DR

This paper introduces a new concept of planarity for streaming graphs, analyzes its computational complexity, and provides efficient algorithms for specific cases, advancing understanding of dynamic graph visualization.

Contribution

It defines the notion of stream planarity with a window size, proves NP-completeness for general cases, and offers faster algorithms for certain restricted scenarios.

Findings

NP-complete for constant window size

NP-complete with backbone graph for all ω ≥ 2

O(n+ωm)-time algorithms for specific cases

Abstract

In this paper we introduce a notion of planarity for graphs that are presented in a streaming fashion. A $\textit{streamed graph}$ is a stream of edges $e_1,e_2,...,e_m$ on a vertex set $V$. A streamed graph is $\omega$-$\textit{stream planar}$ with respect to a positive integer window size $\omega$ if there exists a sequence of planar topological drawings $\Gamma_i$ of the graphs $G_i=(V,\{e_j \mid i\leq j < i+\omega\})$ such that the common graph $G^{i}_\cap=G_i\cap G_{i+1}$ is drawn the same in $\Gamma_i$ and in $\Gamma_{i+1}$, for $1\leq i < m-\omega$. The $\textit{Stream Planarity}$ Problem with window size $\omega$ asks whether a given streamed graph is $\omega$-stream planar. We also consider a generalization, where there is an additional $\textit{backbone graph}$ whose edges have to be present during each time step. These problems are related to several well-studied planarity problems. We show that the $\textit{Stream Planarity}$ Problem is NP-complete even when the window size is a constant and that the variant with a backbone graph is NP-complete for all $\omega \ge 2$. On the positive side, we provide $O(n+\omega{}m)$-time algorithms for (i) the case $\omega = 1$ and (ii) all values of $\omega$ provided the backbone graph consists of one $2$-connected component plus isolated vertices and no stream edge connects two isolated vertices. Our results improve on the Hanani-Tutte-style $O((nm)^3)$-time algorithm proposed by Schaefer [GD'14] for $\omega=1$.