English

Computing Optimal Kernels in Two Dimensions

Computational Geometry 2023-03-15 v2

Abstract

Let PP be a set of nn points in 2\Re^2. For a parameter ε(0,1)\varepsilon\in (0,1), a subset CPC\subseteq P is an \emph{ε\varepsilon-kernel} of PP if the projection of the convex hull of CC approximates that of PP within (1ε)(1-\varepsilon)-factor in every direction. The set CC is a \emph{weak ε\varepsilon-kernel} of PP if its directional width approximates that of PP in every direction. Let kε(P)\mathsf{k}_{\varepsilon}(P) (resp.\ kεw(P)\mathsf{k}^{\mathsf{w}}_{\varepsilon}(P)) denote the minimum-size of an ε\varepsilon-kernel (resp. weak ε\varepsilon-kernel) of PP. We present an O(nkε(P)logn)O(n\mathsf{k}_{\varepsilon}(P)\log n)-time algorithm for computing an ε\varepsilon-kernel of PP of size kε(P)\mathsf{k}_{\varepsilon}(P), and an O(n2logn)O(n^2\log n)-time algorithm for computing a weak ε\varepsilon-kernel of PP of size kεw(P){\mathsf{k}}^{\mathsf{w}}_{\varepsilon}(P). We also present a fast algorithm for the Hausdorff variant of this problem. In addition, we introduce the notion of \emph{ε\varepsilon-core}, a convex polygon lying inside ch(P)\mathsf{ch}(P), prove that it is a good approximation of the optimal ε\varepsilon-kernel, present an efficient algorithm for computing it, and use it to compute an ε\varepsilon-kernel of small size.

Keywords

Cite

@article{arxiv.2207.07211,
  title  = {Computing Optimal Kernels in Two Dimensions},
  author = {Pankaj K. Agarwal and Sariel Har-Peled},
  journal= {arXiv preprint arXiv:2207.07211},
  year   = {2023}
}

Comments

To appear in SoCG 2023