Some convex programs without a duality gap (Q959958): Difference between revisions

The author studies the convex programming problem: \[ v_{P} = \min f_o(x) \text{ s.t. } f_{i}(x) \leq 0,\;i=1,\ldots,m\tag{P} \] where each \(f_i:\mathbb{R}^{n}\rightarrow (-\infty,\infty)\) is a proper convex lower semicontinuous function and its associated dual: \[ v_{D} = \max q(\mu) \text{ where } q(\mu) = \inf_{x}\{f_o(x)+ \sum _{i=1}^m f_{i}(x)\}, \mu \geq 0.\tag{D} \] The main result concerns the non-existence of the duality gap i.e. \(v_{P} = v_{D}\). It is shown that for separable functions \(f_o, f_1,\ldots,f_m\) the relation \(v_{P} = v_{D}\) holds under the assumption of \(\text{dom } f_o \subseteq \bigcap_{1=1}^{m} \text{ dom } f_{i}\) where \(\text{dom } f = \{x|f(x) < \infty\}\). The author gives also a sufficient condition involving weakly analytic functions.