Efficient Algorithms for Geometric Partial Matching
Abstract
Let and be two point sets in the plane of sizes and respectively (assume ), and let be a parameter. A matching between and is a family of pairs in so that any point of appears in at most one pair. Given two positive integers and , we define the cost of matching to be where is the -norm. The geometric partial matching problem asks to find the minimum-cost size- matching between and . We present efficient algorithms for geometric partial matching problem that work for any powers of -norm matching objective: An exact algorithm that runs in time, and a -approximation algorithm that runs in time. Both algorithms are based on the primal-dual flow augmentation scheme; the main improvements involve using dynamic data structures to achieve efficient flow augmentations. With similar techniques, we give an exact algorithm for the planar transportation problem running in time.
Cite
@article{arxiv.1903.09358,
title = {Efficient Algorithms for Geometric Partial Matching},
author = {Pankaj K. Agarwal and Hsien-Chih Chang and Allen Xiao},
journal= {arXiv preprint arXiv:1903.09358},
year = {2019}
}