ParaXpress: an experimental extension of the FICO Xpress-Optimizer to solve hard MIPs on supercomputers. (4th May 2018)

• Record Type: Journal Article
• Authors: Shinano, Yuji; Berthold, Timo; Heinz, Stefan
• Is Part Of: Optimization methods and software. Volume 33:Number 3(2018)
• Page Start: 530
• Page End: 539
• Publication Date: 2018-05-04
• Subjects: mixed-integer programming -- distributed memory parallelization; Mathematical optimization -- Periodicals; Algorithms -- Periodicals; 519.7
• DOI: 10.1080/10556788.2018.1428602
• Languages: English
• ISSNs: 1055-6788

Abstract

Abstract : The Ubiquity Generator (UG ) is a general framework for the external parallelization of mixed integer programming (MIP) solvers. In this paper, we presentParaXpress, a distributed memory parallelization of the powerful commercial MIP solverFICO Xpress . Besides sheer performance, an important feature ofXpress is that it provides an internal parallelization for shared memory systems. When aiming for a best possible performance ofParaXpress on a supercomputer, the question arises how to balance the internalXpress parallelization and the external parallelization byUG against each other. We provide computational experiments to address this question and we show computational results for runningParaXpress on a Top500 supercomputer, using up to 43, 344 cores in parallel.