English

Coresets for Farthest Point Problems in Hyperbolic Space

Computational Geometry 2025-11-03 v1

Abstract

We show how to construct in linear time coresets of constant size for farthest point problems in fixed-dimensional hyperbolic space. Our coresets provide both an arbitrarily small relative error and additive error ε\varepsilon. More precisely, we are given a set PP of nn points in the hyperbolic space HD\mathbb{H}^D, where D=O(1)D=O(1), and an error tolerance ε(0,1)\varepsilon\in (0,1). Then we can construct in O(n/εD)O(n/\varepsilon^D) time a subset PεPP_\varepsilon \subset P of size O(1/εD)O(1/\varepsilon^D) such that for any query point qHDq \in \mathbb{H}^D, there is a point pεPεp_\varepsilon \in P_\varepsilon that satisfies dH(q,pε)(1ε)dH(q,fP(q))d_H(q,p_\varepsilon) \geq (1-\varepsilon)d_H(q,f_P(q)) and dH(q,pε)dH(q,fP(q))εd_H(q,p_\varepsilon) \geq d_H(q,f_P(q))-\varepsilon, where dHd_H denotes the hyperbolic metric and fP(q)f_P(q) is the point in PP that is farthest from qq according to this metric. This coreset allows us to answer approximate farthest-point queries in time O(1/εD)O(1/\varepsilon^D) after O(n/εD)O(n/\varepsilon^D) preprocessing time. It yields efficient approximation algorithms for the diameter, the center, and the maximum spanning tree problems in hyperbolic space.

Keywords

Cite

@article{arxiv.2510.27491,
  title  = {Coresets for Farthest Point Problems in Hyperbolic Space},
  author = {Eunku Park and Antoine Vigneron},
  journal= {arXiv preprint arXiv:2510.27491},
  year   = {2025}
}
R2 v1 2026-07-01T07:15:39.640Z