Article ID: | iaor20126635 |
Volume: | 54 |
Issue: | 4 |
Start Page Number: | 669 |
End Page Number: | 687 |
Publication Date: | Dec 2012 |
Journal: | Journal of Global Optimization |
Authors: | Wang Changyu, Xiu Naihua, Zhou Jinchuan |
Keywords: | programming: nonlinear |
In this paper, we introduce a generalized proximal Lagrangian function for the constrained nonlinear programming problem and discuss existence of its saddle points. In particular, the local saddle point is obtained by using the second‐order sufficient conditions, and the global saddle point is given without requiring compactness of constraint set and uniqueness of the optimal solution. Finally, we establish equivalent relationship between global saddle points and exact penalty representations.