Faster combinatorialapproximation algorithms for multicommodity flow problems and its applocation

Abstract

A multi-commodity flow problem is to find a deasible flow that satisfies the demand of each source to the sink in a directed network while the flow on each edge is within its capacity. In this paper we implement polynomial-time combinatorial approximation algorithms for ε - optimal multi-commodity flow problems. A new algorithm Optima is proposed and good test results are obtained. Application of Optima to a system-optimized multi-user network flow problem is given where substantially good test results have shown the capacity of Optima in dealing with problems of multi-commodity flow associated.