Coresets for Continuous $k$-Center in Hyperbolic Space

Abstract

We study the continuous $k$-center problem in fixed-dimensional hyperbolic space. The input is a finite point set $P \subset \mathbb H^D$, while the centers may be placed anywhere in the ambient space. A nonempty subset $P_\varepsilon\subseteq P$ is an $\varepsilon$-coreset if every exact optimal continuous $k$-center solution for $P_\varepsilon$ is a $(1+\varepsilon)$-approximate solution for $P$. Our construction uses Gonzalez's farthest-first traversal, whose radius $R_G$ gives a constant-factor estimate of the optimum. When $R_G$ is bounded, we map each cluster to the origin of the Poincar\'e ball model and apply a local Euclidean grid. For larger $R_G$, we partition each cluster into radial shells and project them onto representative hyperbolic spheres, avoiding a discretization of the ambient ball. By the hyperbolic law of cosines, the intersection of a hyperbolic ball with each sphere corresponds to a spherical cap. A recursive $k$-cap coverage certificate then replaces each shell with a radius-independent set of input witnesses that preserves its farthest-point behavior for every set of at most $k$ centers. The resulting coreset $P_\varepsilon$ has size $|P_\varepsilon| = \left({k}/{\varepsilon}\right)^{O(kD)}$ and satisfies $\Phi_P(C_{P_\varepsilon}) \le (1+\varepsilon)\operatorname{OPT}_k(P)$ for every optimal continuous $k$-center solution $C_{P_\varepsilon}$ of $P_\varepsilon$. It can be constructed in $O\left( nkD\left({k}/{\varepsilon}\right)^{O(kD)} \right)$ time. Consequently, for fixed $D$, $k$, and $\varepsilon$, the coreset has constant size and can be computed in linear time. We complement this upper bound by showing that, for every fixed $D \ge 1$, every $k \ge 2$, and all sufficiently small $\varepsilon$, there are instances for which every $\varepsilon$-coreset has size $\Omega\left(k\varepsilon^{-(D-1)/2}\right)$.