A practical optimality condition without constraint qualifications for nonlinear programming
Source file as
Postscript Document (.ps)
,
Portable Document Format (.pdf)
José Mario Martínez ,
Benar Fux Svaiter
Keywords:
Optimality conditions, Karush-Kuhn-Tucker, minimization algorithms, constrained optimization
Abstract:
A new optimality condition for minimization with general constraints is introduced. Unlike the KKT conditions, this condition is satisfied by local minimizers of nonlinear programming problems, independently of constraint qualifications. The new condition implies, and is strictly stronger than, Fritz-John optimality conditions. Sufficiency for convex programming is proved.
MSC 2000:
90C30 Nonlinear programming
