PolySCIP

  • PolySCIP is a new solver for multi-criteria integer and multi-criteria linear programs handling an arbitrary number of objectives. It is available as an official part of the non-commercial constraint integer programming framework SCIP. It utilizes a lifted weight space approach to compute the set of supported extreme non-dominated points and unbounded non-dominated rays, respectively. The algorithmic approach can be summarized as follows: At the beginning an arbitrary non-dominated point is computed (or it is determined that there is none) and a weight space polyhedron created. In every next iteration a vertex of the weight space polyhedron is selected whose entries give rise to a single-objective optimization problem via a combination of the original objectives. If the ptimization of this single-objective problem yields a new non-dominated point, the weight space polyhedron is updated. Otherwise another vertex of the weight space polyhedron is investigated. The algorithm finishes when all vertices of the weight space polyhedron have been investigated. The file format of PolySCIP is based on the widely used MPS format and allows a simple generation of multi-criteria models via an algebraic modelling language.

Export metadata

  • Export Bibtex
Metadaten
Author:Ralf Borndörfer, Sebastian Schenker, Martin Skutella, Timo Strunk
Editor:G.-M. Greuel, Thorsten Koch, Peter Paule, Andrew Sommese
Document Type:In Proceedings
Parent Title (English):Mathematical Software – ICMS 2016, 5th International Conference, Berlin, Germany, July 11-14, 2016, Proceedings
Volume:9725
First Page:259
Last Page:264
Series:Lecture Notes in Computer Science
Publisher:Springer International Publishing
Year of first publication:2016
ISBN:978-3-319-42431-6
DOI:http://dx.doi.org/10.1007/978-3-319-42432-3_32

$Rev: 13581 $