1: \begin{abstract}

2:   Interior Point Methods (IPM) rely on the Newton method for solving

3:   systems of nonlinear equations. Solving the linear systems which

4:   arise from this approach is the most computationally expensive task

5:   of an interior point iteration. If, due to problem's inner

6:   structure, there are special techniques for efficiently solving

7:   linear systems, IPMs enjoy fast convergence and are able to solve

8:   large scale optimization problems. It is tempting to try to replace

9:   the Newton method by quasi-Newton methods. Quasi-Newton approaches

10:   to IPMs either are built to approximate the Lagrangian function for

11:   nonlinear programming problems or provide an inexpensive

12:   preconditioner. In this work we study the impact of using

13:   quasi-Newton methods applied directly to the nonlinear system of

14:   equations for general quadratic programming problems. The cost of

15:   each iteration can be compared to the cost of computing correctors

16:   in a usual interior point iteration. Numerical experiments show that

17:   the new approach is able to reduce the overall number of matrix

18:   factorizations and is suitable for a matrix-free implementation.\\

19: 

20:   \noindent \textbf{Keywords}: Broyden Method, Quasi-Newton, Interior

21:   Point Methods, Matrix-free, Quadratic Programming Problems

22: \end{abstract}

23: