Quadratic convex reformulation for graph partitionning problems

Abstract : Many graph partitionning problems can be formulated by quadratic programs (QP) with binary variables and linear and quadratic constraints. We apply the general approach that consists in first reformulating the initial (QP) into an equivalent program (QP ). Problem (QP ) has the additional property that its continuous relaxation is a convex quadratic problem. It can then be solved by branch-and-bound based on quadratic convex relaxation. We show how variants of this general approach perform on many graph partitionning problems.
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-01126262
Contributor : Laboratoire Cedric <>
Submitted on : Friday, March 6, 2015 - 11:48:04 AM
Last modification on : Thursday, February 6, 2020 - 5:31:08 PM

Identifiers

  • HAL Id : hal-01126262, version 1

Collections

Citation

Sourour Elloumi, Amélie Lambert. Quadratic convex reformulation for graph partitionning problems. IFIP TC 7 / 2013 System Modelling and Optimization, Sep 2013, Klagenfurt, Austria. pp.1. ⟨hal-01126262⟩

Share

Metrics

Record views

19