A complete characterization of strong duality in nonconvex optimization with a single constraint (Q454276)

Let \(X\) be a real locally convex topological vector space, \(Y\) be a normed space. \(P\) has a closed convex cone in \(Y\), let \(C\) be a subset of \(X\). The authors consider the following nonconvex minimization problem: \[ \inf f(x)\text{ subject to }g(x)\in -P,\;x\in C, \] where \(f: C\to g: C\to Y\). The Lagrangian dual problem to this problem is formulated and the strong duality of the two problems is proved under a generalized Slater-type condition. A new complete characterization of the strong duality with a single constraints is presented. It is shown that the strong duality holds in this case without the standard Slater condition. Theoretical results are demonstrated on appropriately chosen examples.