Towards a practical parallelisation of the revised simplex method

Contributed talk the IMA Conference on Numerical Linear Algebra and Optimisation: 14th September 2010

J. A. J. Hall, E. Smith and Q. Huangfu

Abstract

The revised simplex method is frequently the most efficient method of solving linear programming (LP) problems. Current architecture trends dictate that exploiting parallelism must be considered as a means of improving computational performance. This talk will outline the challenge of developing a parallel revised simplex solver of practical value and introduce two approaches that are the subject of current research.