English

Near-Optimal Algorithms for Point-Line Covering Problems

Computational Geometry 2021-02-16 v3 Data Structures and Algorithms

Abstract

We study fundamental point-line covering problems in computational geometry, in which the input is a set SS of points in the plane. The first is the Rich Lines problem, which asks for the set of all lines that each covers at least λ\lambda points from SS, for a given integer parameter λ2\lambda \geq 2; this problem subsumes the 3-Points-on-Line problem and the Exact Fitting problem, which -- the latter -- asks for a line containing the maximum number of points. The second is the NP-hard problem Line Cover, which asks for a set of kk lines that cover the points of SS, for a given parameter kNk \in \mathbb{N}. Both problems have been extensively studied. In particular, the Rich Lines problem is a fundamental problem whose solution serves as a building block for several algorithms in computational geometry. For Rich Lines and Exact Fitting, we present a randomized Monte Carlo algorithm that achieves a lower running time than that of Guibas et al.'s algorithm [Computational Geometry 1996], for a wide range of the parameter λ\lambda. We derive lower-bound results showing that, for λ=Ω(nlogn)\lambda =\Omega(\sqrt{n \log n}), the upper bound on the running time of this randomized algorithm matches the lower bound that we derive on the time complexity of Rich Lines in the algebraic computation trees model. For Line Cover, we present two kernelization algorithms: a randomized Monte Carlo algorithm and a deterministic algorithm. Both algorithms improve the running time of existing kernelization algorithms for Line Cover. We derive lower-bound results showing that the running time of the randomized algorithm we present comes close to the lower bound we derive on the time complexity of kernelization algorithms for Line Cover in the algebraic computation trees model.

Keywords

Cite

@article{arxiv.2012.02363,
  title  = {Near-Optimal Algorithms for Point-Line Covering Problems},
  author = {Jianer Chen and Qin Huang and Iyad Kanj and Ge Xia},
  journal= {arXiv preprint arXiv:2012.02363},
  year   = {2021}
}
R2 v1 2026-06-23T20:43:25.936Z