A deterministic linear time algorithm for geometric separators and its applications

  • Authors:
  • David Eppstein;Gary L. Miller;Shang-Hua Teng

  • Affiliations:
  • Univ. of California, Irvine;Carnegie Mellon Univ., Pittsburgh, PA;Massachusetts Institute of Technology, Cambridge

  • Venue:
  • SCG '93 Proceedings of the ninth annual symposium on Computational geometry
  • Year:
  • 1993

Quantified Score

Hi-index 0.00

Visualization

Abstract

We give a deterministic linear time algorithm for finding a smallcost sphere separator of a k-plyneighborhood system &Fgr; in any fixed dimension, where ak-ply neighborhood system inRd is a collection ofn balls such that no points in thespace is covered by more than kballs. The sphere separator intersects at most Ok1dnd-1d balls of &Fgr; and it divides the remaining of &Fgr;into two parts: those in the interior and those in the exterior of thesphere, respectively, so that the larger part contains at most&dgr;n balls d+1d+2d. This result improves the 0(n2) timedeterministic algorithm of Miller and Teng [29] and answers a majoralgorithmic open question posed by Miller, Teng, Thurston and Vavasis[23,25].The deterministic algorithm hinges on the use of a new method forderiving the separator property of neighborhood systems. Using thisalgorithm, we devise an O(kn + n logn) time deterministic algorithm for computing theintersection graph of a k-plyneighborhood system. We give an O(n logn) time algorithm for constructing a linear space,O(log n) query time search structurefor a geometric query problem associated withk-ply neighborhood systems, and weuse this data structure in an algorithm for approximating the value ofk within a constant factor in timeO(n log n). We also develop adeterministic linear time algorithm for finding an Ok1dnd-1d-separator for ak-nearest neighborhood graph ind dimensions.