Algoritmo de ponto interior para programação linear baseado no FDIPA

Abstract

[EN] In this work we use an algorithm of interior points and feasible directions called FDIPA, ”Feasible Direction Interior Point Algorithm”, for solving linear optimization problems. In each iteration FDIPA calculates a feasible descent direction of the problem by a Newton-type iteration to solve the Karush-Kuhn-Tucker (KKT) conditions; generating two linear systems of equations. Numerical techniques are proposed to solve them efficiently, in particular by means of a preconditioned conjugate gradient method, in which we obtain a criterion to truncate it. Finally, several test problems will be solved and compared with results from the literature.