1: \begin{abstract}

2: \noindent 

3: When an iterative method is applied to solve the linear equation system 

4: in interior point methods (IPMs), the attention is usually placed on accelerating their convergence 

5: by designing appropriate preconditioners, but the linear solver is applied 

6: as a black box with a standard termination criterion which asks 

7: for a sufficient reduction of the residual in the linear system. 

8: Such an approach often leads to an unnecessary ''oversolving'' of linear 

9: equations. 

10: %In this paper, it is demonstrated how an IPM can preserve the 

11: %polynomial worst-case complexity when relying on an inner termination 

12: %criterion that is not based on the residual of the linear system. 

13: In this paper, an IPM that relies on an inner termination criterion not based on the residual of the linear system is introduced and analyzed. Moreover, 

14: new indicators for the early termination of the inner iterations are derived 

15: from a deep understanding of IPM needs.

16: The new technique has been adapted to the Conjugate Gradient (CG) 

17: and to the Minimum Residual method (MINRES) applied in the IPM context. 

18: The new criterion has been tested on a set of quadratic optimization 

19: problems including compressed sensing, image processing and instances 

20: with partial differential equation constraints, and it has been

21: compared to standard residual tests with variable tolerance.

22: Evidence gathered from these computational experiments shows that the new 

23: technique delivers significant improvements in terms of inner (linear) 

24: iterations and those translate into significant savings of the IPM 

25: solution time.

26: \end{abstract}