1: \begin{abstract}

2: In the context of augmented Lagrangian approaches for solving semidefinite 

3: programming problems, we investigate the possibility of eliminating the positive semidefinite constraint on the dual matrix by employing a factorization. 

4: Hints on how to deal with the resulting unconstrained maximization of the augmented Lagrangian are given. We further use the approximate maximum of 

5: the augmented Lagrangian with the aim of improving the convergence rate of alternating direction augmented Lagrangian frameworks. 

6: Numerical results are reported, showing the benefits of the approach.\end{abstract}