An Estimator for Matching Size in Low Arboricity Graphs with Two Applications

TL;DR

This paper introduces a simple degree-based estimator for maximum matching size in bounded arboricity graphs, achieving better approximation factors for planar graphs and enabling efficient streaming and distributed algorithms.

Contribution

The paper presents a new estimator that does not require knowing the arboricity and improves approximation for planar graphs, with applications in streaming and distributed models.

Findings

Provides a $ ext{α}+2$ approximation for bounded arboricity graphs.

Achieves a 3.5-approximation for planar graphs.

Enables space-efficient streaming and distributed algorithms for matching size approximation.

Abstract

In this paper, we present a new simple degree-based estimator for the size of maximum matching in bounded arboricity graphs. When the arboricity of the graph is bounded by $\alpha$, the estimator gives a $\alpha+2$ factor approximation of the matching size. For planar graphs, we show the estimator does better and returns a $3.5$ approximation of the matching size. Using this estimator, we get new results for approximating the matching size of planar graphs in the streaming and distributed models of computation. In particular, in the vertex-arrival streams, we get a randomized $O(\frac{\sqrt{n}}{\epsilon^2}\log n)$ space algorithm for approximating the matching size within $(3.5+\epsilon)$ factor in a planar graph on $n$ vertices. Similarly, we get a simultaneous protocol in the vertex-partition model for approximating the matching size within $(3.5+\epsilon)$ factor using $O(\frac{n^{2/3}}{\epsilon^2}\log n)$ communication from each player. In comparison with the previous estimators, the estimator in this paper does not need to know the arboricity of the input graph and improves the approximation factor for the case of planar graphs.