Efficiently Enumerating Scaled Copies of Point Set Patterns

TL;DR

This paper presents optimal algorithms for enumerating scaled copies of point set patterns, including axis-parallel squares and general fixed sets, matching known lower bounds and solving longstanding open problems.

Contribution

It introduces the first worst-case optimal algorithm for enumerating scaled axis-parallel squares and extends this to general fixed patterns in higher dimensions.

Findings

Algorithm for axis-parallel squares runs in O(n√n) time

Extended algorithm works for any fixed pattern in d-dimensional space

Results match known lower bounds, confirming optimality

Abstract

Problems on repeated geometric patterns in finite point sets in Euclidean space are extensively studied in the literature of combinatorial and computational geometry. Such problems trace their inspiration to Erd\H{o}s' original work on that topic. In this paper, we investigate the particular case of finding scaled copies of any pattern within a set of $n$ points, that is, the algorithmic task of efficiently enumerating all such copies. We initially focus on one particularly simple pattern of axis-parallel squares, and present an algorithm with an $O(n\sqrt{n})$ running time and $O(n)$ space for this task, involving various bucket-based and sweep-line techniques. Our algorithm is worst-case optimal, as it matches the known lower bound of $\Omega(n\sqrt{n})$ on the maximum number of axis-parallel squares determined by $n$ points in the plane, thereby solving an open question for more than three decades of realizing that bound for this pattern. We extend our result to an algorithm that enumerates all copies, up to scaling, of any full-dimensional fixed set of points in $d$-dimensional Euclidean space, that works in time $O(n^{1+1/d})$ and space $O(n)$, also matching the corresponding lower bound due to Elekes and Erd\H{o}s.