Computing the sequence of $k$-cardinality assignments

Research output: Contribution to journalArticle

Abstract

The k-cardinality assignment problem asks for finding a maximal (minimal) weight of a matching of cardinality k in a weighted bipartite graph Kn,n, k≤n. The algorithm of Gassner and Klinz from 2010 for the parametric assignment problem computes in time O(n3) the set of k-cardinality assignments for those integers k≤n which refer to "essential" terms of a corresponding maxpolynomial. We show here that one can extend this algorithm and compute in a second stage the other "semi-essential" terms in time O(n2), which results in a time complexity of O(n3) for the whole sequence of k=1,...,n-cardinality assignments. The more there are assignments left to be computed at the second stage the faster the two-stage algorithm runs. In general, however, there is no benefit for this two-stage algorithm on the existing algorithms, e.g. the simpler network flow algorithm based on the successive shortest path algorithm which also computes all the k-cardinality assignments in time O(n3)
Original languageEnglish
Number of pages19
JournalarXiv.org e-Print archive
Publication statusPublished - 8 Apr 2021

Keywords

  • math.OC
  • cs.DS

Cite this