PolynomialTime Approximation Schemes for Geometric几何多项式的多项式时间近似方案.pptVIP

Polynomial-Time Approximation Schemes for Geometric Intersection Graphs Authors: T. Erlebach, L. Jansen, and E. Seidel Presented by: Ping Luo 10/17/2005 Agenda Preliminaries Disk partition and plane subdivision Dynamic programming Basic Concepts Intersection graph: for a set V of geometric objects, the corresponding intersection graph is the undirected graph with vertex set V and an edge between two vertices if the corresponding objects intersect. Disk graph: a disk graph is the intersection graph of a set of disks Disk graph recognition is NP-hard Disk graph – an example Problem definitions Maximum Weight Independent Set (MWIS) For a given set of geographic objects, find a subset of disjoint objects such that the total weight is maximum. This paper deals with disks and is the topic of this talk. Minimum Weight Vertex Cover (MWVC) Find a subset of a given objects with minimum total weight such that, for any two intersection objects, at least one is in the subset. Basic idea of the PTAS algorithm for MWIS The plane is partitioned into squares on each level. Some disks in each level are removed so that different squares on the same level yield independent subproblems with respect to all disks that are on this level or smaller levels. Each square has at most a constant number of disks so that enumeration is possible Agenda Preliminaries Disk partition and plane subdivision Dynamic programming Partition disks into levels Zoom in or out to make the largest diameter to be 1. The smallest is denoted as dmin Compute disk levels Determine disk level j of each disk with diameter d: Partition disks – an example Plane subdivision Definitions I Observation about j-squares in different levels For any j, , every (j+1)-square is completely contained in some j-square. Every j-square is the union of (j+1)-squares. Definition II - D(r,s) Definitions III Useful upper bounds The number of disks (levelj) that intersect a j-square is bounded by The numb