Strong Matching of Points with Geometric Shapes

TL;DR

This paper introduces algorithms for finding large strong matchings in geometric graphs defined by points and convex shapes, providing improved bounds for various shapes such as disks, triangles, and squares.

Contribution

It presents new algorithms with provable lower bounds for strong matchings in geometric graphs based on different convex shapes, improving previous bounds.

Findings

For diametral-disks, strong matchings of size at least (n-1)/17.

For equilateral-triangles, strong matchings of size at least (n-1)/9.

For squares, strong matchings of size at least (n-1)/4, improving previous bounds.

Abstract

Let $P$ be a set of $n$ points in general position in the plane. Given a convex geometric shape $S$, a geometric graph $G_S(P)$ on $P$ is defined to have an edge between two points if and only if there exists an empty homothet of $S$ having the two points on its boundary. A matching in $G_S(P)$ is said to be $strong$, if the homothests of $S$ representing the edges of the matching, are pairwise disjoint, i.e., do not share any point in the plane. We consider the problem of computing a strong matching in $G_S(P)$, where $S$ is a diametral-disk, an equilateral-triangle, or a square. We present an algorithm which computes a strong matching in $G_S(P)$; if $S$ is a diametral-disk, then it computes a strong matching of size at least $\lceil \frac{n-1}{17} \rceil$, and if $S$ is an equilateral-triangle, then it computes a strong matching of size at least $\lceil \frac{n-1}{9} \rceil$. If $S$ can be a downward or an upward equilateral-triangle, we compute a strong matching of size at least $\lceil \frac{n-1}{4} \rceil$ in $G_S(P)$. When $S$ is an axis-aligned square we compute a strong matching of size $\lceil \frac{n-1}{4} \rceil$ in $G_S(P)$, which improves the previous lower bound of $\lceil \frac{n}{5} \rceil$.