Approximation algorithms for solving the line-capacitated minimum Steiner tree problem

In this paper, we address the line-capacitated minimum Steiner tree problem (the Lc-MStT problem, for short), which is a variant of the (Euclidean) capacitated minimum Steiner tree problem and defined as follows. Given a set X = { r 1 , r 2 , … , r n } of n terminals in R 2 , a demand function d : X...

Full description

Saved in:
Bibliographic Details
Published in:Journal of global optimization Vol. 84; no. 3; pp. 687 - 714
Main Authors: Li, Jianping, Wang, Wencheng, Lichen, Junran, Liu, Suding, Pan, Pengxiang
Format: Journal Article
Language:English
Published: New York Springer US 01.11.2022
Springer
Springer Nature B.V
Subjects:
ISSN:0925-5001, 1573-2916
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, we address the line-capacitated minimum Steiner tree problem (the Lc-MStT problem, for short), which is a variant of the (Euclidean) capacitated minimum Steiner tree problem and defined as follows. Given a set X = { r 1 , r 2 , … , r n } of n terminals in R 2 , a demand function d : X → N and a positive integer C , we are asked to determine the location of a line l and a Steiner tree T l to interconnect these n terminals in X and at least one point located on this line  l such that the total demand of terminals in each maximal subtree (of T l ) connected to the line l , where the terminals in such maximal subtree are all located at the same side of this line l , does not exceed the bound C . The objective is to minimize total weight ∑ e ∈ T l w ( e ) of such a Steiner tree T l among all line-capacitated Steiner trees mentioned-above, where weight w ( e ) = 0 if two endpoints of that edge e ∈ T l are located on the line l and otherwise weight w ( e ) is the Euclidean distance between two endpoints of that edge e ∈ T l . In addition, when this line l is as an input in R 2 and ∑ r ∈ X d ( r ) ≤ C holds, we refer to this version as the 1-line-fixed minimum Steiner tree problem (the 1Lf-MStT problem, for short). We obtain three main results. (1) Given a ρ st -approximation algorithm to solve the Euclidean minimum Steiner tree problem and a ρ 1 L f -approximation algorithm to solve the 1Lf-MStT problem, respectively, we design a ( ρ st ρ 1 L f + 2 ) -approximation algorithm to solve the Lc-MStT problem. (2) Whenever demand of each terminal r ∈ X is less than  C 2 , we provide a ( ρ 1 L f + 2 ) -approximation algorithm to resolve the Lc-MStT problem. (3) Whenever demand of each terminal r ∈ X is at least C 2 , using the Edmonds’ algorithm to solve the minimum weight perfect matching as a subroutine, we present an exact algorithm in polynomial time to resolve the Lc-MStT problem.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0925-5001
1573-2916
DOI:10.1007/s10898-022-01163-x