A Gauss-Newton approach for solving constrained optimization problems using differentiable exact penalties (Q1949585): Difference between revisions

The authors construct an exact penalty function for equality and inequality constrained optimization problems, by extending the one considered by \textit{T. A. de André} and \textit{P. J. S. Silva} [Comput. Optim. Appl. 47, No. 3, 401--429 (2010; Zbl 1208.90173)] for variational inequalities, and use a generalized semismooth Newton method to solve a reformulation of the KKT conditions as a system of equations. The method is proved to be locally superlinearly convergent under rather weak assumptions. To globalize it, an appropriate merit function is proposed. Some numerical experiments are reported.