A Sparse Multicover Bifiltration of Linear Size

TL;DR

This paper introduces a linear-size approximation method for the multicover bifiltration of point clouds, making it computationally feasible while maintaining accuracy, applicable to various metric spaces.

Contribution

It presents a $(1+ ext{epsilon})$-approximation of the multicover bifiltration with linear size, significantly reducing complexity compared to previous methods.

Findings

Achieves linear size $O(|X|)$ for the approximation

Applicable to subdivision Rips bifiltration on bounded doubling spaces

Maintains $(1+ ext{epsilon})$ accuracy in approximation

Abstract

The $k$-cover of a point cloud $X$ in $\mathbb{R}^{d}$ at radius $r$ is the set of all points within distance $r$ of at least $k$ points of $X$. By varying $r$ and $k$ we obtain a two-parameter filtration known as the multicover bifiltration. This bifiltration has received attention recently due to being choice-free and robust to outliers. However, it is hard to compute: the smallest known equivalent simplicial bifiltration has $O(|X|^{d+1})$ simplices. We introduce a $(1+\epsilon)$-approximation of the multicover bifiltration of linear size $O(|X|)$, for fixed $d$ and $\epsilon$. The methods also apply to the subdivision Rips bifiltration on metric spaces of bounded doubling dimension, yielding analogous results.