1: \begin{abstract}

2: A preconditioning strategy for the Powell-Hestenes-Rockafellar 

3: Augmented Lagrangian method (ALM) is presented. The scheme exploits the

4: structure of the Augmented Lagrangian Hessian. It is a modular

5: preconditioner consisting of two blocks. The first one is associated with the

6: Lagrangian of the objective while the second administers the Jacobian of the

7: constraints and possible low-rank corrections to the Hessian.

8: The proposed updating strategies 

9: take advantage of

10: ALM convergence results and avoid frequent refreshing. Constraint 

11: administration takes into account complementarity over the Lagrange multipliers 

12: and admits 

13: relaxation.

14: The preconditioner is designed for problems where constraint quantity is 

15: small compared to the search space. A virtue of the scheme is that it is  

16: agnostic to the preconditioning technique used for the Hessian of the 

17: Lagrangian function.

18: The strategy described can be used for linear and nonlinear preconditioning.

19: Numerical experiments report on spectral properties of  

20: preconditioned matrices from Matrix Market while some optimization 

21: problems where taken from the CUTEst collection. Preliminary results indicate 

22: that the 

23: proposed scheme could be attractive and further experimentation is encouraged.

24: \end{abstract}

25: