Improved approximation for Fr\'echet distance on c-packed curves matching conditional lower bounds

TL;DR

This paper presents an improved algorithm for approximating the Fréchet distance on c-packed curves, achieving near-optimal runtime in high dimensions by leveraging properties that avoid certain lower bounds.

Contribution

The authors develop a faster approximation algorithm for c-packed curves that matches conditional lower bounds, using novel projections and extensions of greedy algorithms.

Findings

Achieves near-linear time approximation in high dimensions.

Matches conditional lower bounds assuming SETH.

Introduces a projection technique for short subcurves.

Abstract

The Fr\'echet distance is a well-studied and very popular measure of similarity of two curves. The best known algorithms have quadratic time complexity, which has recently been shown to be optimal assuming the Strong Exponential Time Hypothesis (SETH) [Bringmann FOCS'14]. To overcome the worst-case quadratic time barrier, restricted classes of curves have been studied that attempt to capture realistic input curves. The most popular such class are c-packed curves, for which the Fr\'echet distance has a $(1+\epsilon)$-approximation in time $\tilde{O}(c n /\epsilon)$ [Driemel et al. DCG'12]. In dimension $d \ge 5$ this cannot be improved to $O((cn/\sqrt{\epsilon})^{1-\delta})$ for any $\delta > 0$ unless SETH fails [Bringmann FOCS'14]. In this paper, exploiting properties that prevent stronger lower bounds, we present an improved algorithm with runtime $\tilde{O}(cn/\sqrt{\epsilon})$. This is optimal in high dimensions apart from lower order factors unless SETH fails. Our main new ingredients are as follows: For filling the classical free-space diagram we project short subcurves onto a line, which yields one-dimensional separated curves with roughly the same pairwise distances between vertices. Then we tackle this special case in near-linear time by carefully extending a greedy algorithm for the Fr\'echet distance of one-dimensional separated curves.