This preview shows page 1. Sign up to view the full content.
Unformatted text preview: ms is that the feasible region has no
interior. If the interior is non-empty, then the KKT conditions are necessary for a
convex programming problem.
Question: Are the KKT conditions sufficient for local optimality? That is, does ̅ a KKT
̅ a local min? Recall the KKT conditions.
∑ ̅ ̅ Consider any point ̅ with for
̅ ̅ ̅ need not be a local minimum (could , does ̅ a KKT point f (x) g1 (x) g 2 ( x) 0 ̅
. Then ̅ is a KKT point by selecting However, we know that a point with
be maximum or a sad...
View Full Document
- Fall '13