Algorithme de points intérieurs pour la programmation quadratique convexe

Abstract

Abstract: In this work, we have studied a primal-dual interior-point algorithm with weights for solving a convex quadratic problem (CQP) (see [6]). The strategy of this algorithm is to replace the centrality equation by the weight, with at each iteration, the full Newton step is used to find an approximate solution. This is followed by comparative numerical tests based on the choice of θ to see its influence on the numerical behavior of this algorithm.