DMCA
Experiments With a Network Design Algorithm Using Epsilon-Approximate Linear Programs (1998)
| Citations: | 8 - 3 self |
BibTeX
@MISC{Bienstock98experimentswith,
author = {Daniel Bienstock},
title = {Experiments With a Network Design Algorithm Using Epsilon-Approximate Linear Programs},
year = {1998}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
We describe an upper-bound algorithm for multicommodity network design problems that relies on new results for approximately solving certain linear programs, and on the greedy heuristic for set-covering problems. 1 Introduction. Network design problems are mixed-integer programs that have the following broad structure. Given a graph, and a set of "demands" -- positive amounts to be routed between pairs of vertices -- capacity must be added to the edges and/or vertices of the graph, in discrete amounts, and at minimum cost, so that a feasible routing is possible. Problem of this form are increasingly important in telecommunications applications, because of the great expense inherent in maintaining and upgrading metropolitan networks. A wide variety of special cases have been studied. For example, one may be constrained to using a fixed family of paths to carry out the routing, or to using a single path for each demand, or to using integral flows. The precise manner in which capacit...
Keyphrases
feasible routing network design problem upgrading metropolitan network positive amount single path special case telecommunication application integral flow following broad structure certain linear program new result fixed family great expense inherent mixed-integer program set-covering problem vertex capacity discrete amount precise manner multicommodity network design problem upper-bound algorithm wide variety minimum cost
