Algoritmo de Karmarkar y matrices ralas

Abstract

This is the second of a series of two articles en which we study the Karmarkar’s method. In this article we are going to show how can we use sparse matrix theory to get an efficient implementation of the Karmarkar’s process presented in the first article. In phase I of the Karmarkar’s process, it was evident how the size of the technological matrix increased. However, the new matrix has a special structure in which we observed the presence of zero’s blocks that make it a sparse matrix. We will discuss here some techniques to be used with this kind of matrix. Finally we propose a Kamarkar’s variant that takes advantage of this situation.