Computational Experience of an Augmented Lagrangian for Nonlinear Programming as a Merit Function in a Primal-Dual Interior-Point Method
In this paper we extend the L² augmented Lagrangian function from equality constrained problems to nonlinear programming problems. The augmentation is strongly influenced by interior-point philosophy. Only information about the equality constraints and nonnegativity of the variables involved in the first order necessary conditions are considered in the augmentation term. The parameter choice in the augmentation is taken in order to make the Newton direction of a perturbed first order necessary condition a descent direction of the corresponding augmented Lagrangian. We present numerical results of a line-search interior-point method where the merit function is theaugmented Lagrangian.
Citable link to this pagehttps://hdl.handle.net/1911/101872
MetadataShow full item record
- CAAM Technical Reports