Abstracts
Abstract
The sphere method for solving linear programs operates with only a subset of constraints in the model in each iteration, and thus has the advantage of handling instances which may not be very sparse. In this paper we discuss enhancements, and improved versions Sphere Methods 2, 2.1 which improve performance by 20 to 80%. Additional steps that can improve performance even more are also presented.
Keywords:
- Linear Programming (LP),
- Interior point methods (IPM.s),
- ball centers of a polytope,
- solving LPs using matrix inversions sparingly,
- Sphere methods-1, 2, 2.1 for large scale LPs
Download the article in PDF to read it.
Download