scientific article; zbMATH DE number 1071775

From MaRDI portal
Publication:4359090





It is well known that the notion of convexity and the subdifferential of convex functions are important tools for the derivation of optimality and duality assertions. In this connection, we have the basic result that a continuous function \(f: X\to\mathbb{R}\) (where \(X\) is a normal vector space) is convex if and only if it is the supremum of affine-linear functions, i.e., if \[ f(x)= \sup\{l(x)+ c\mid l\in X^*, c\in\mathbb{R}, l(y)+ c\leq f(y)\;\forall y\in X\}. \] This duality assertion can be expressed by the notion of Fenchel conjugation since the term on the right-hand side turns out to be the second Fenchel dual of \(f\). Moreover, by weak conditions on the topological structure of \(X\), Mazure and Asplund showed many years ago that continuous convex functions are differentiable ``almost everywhere in the sense of Gâteaux and Fréchet, respectively, i.e., the subdifferential of \(f\) is a singleton ``almost everywhere. In the present paper, the author gives an extension of the mentioned results for spaces \(X\) without linear structure. Replacing the dual space \(X^*\) by an arbitrary family \(\Phi\) of functions, he obtains the \(\Phi\)-convexification of a function \(f\) according to \[ f^\Phi(x)= \sup\{\phi(x)+ c\mid \phi\in\Phi, c\in\mathbb{R}, \phi(y)+c\leq f(y)\;\forall y\in X\}. \] In this sense a function \(f\) is called to be \(\Phi\)-convex if it coincides with \(f^\Phi\). In the same manner the \(\Phi\)-subdifferential \(\partial_\Phi f\) of a function is introduced and important properties of this set in connection with optimality assertions for nonlinear optimization problems are pointed out. Regarding spaces \(X\) with a metric structure and using additional algebraic and topological assumptions on the set \(\Phi\) it is shown that also the mentioned theorems of Mazure and Asplund can be extended to \(\Phi\)-convex functions. In the last sections, the author gives some results regarding the connections between local uniform convexity and local uniform smoothness of \(\Phi\)-convex functions but also to special forms of quadratic regularization of functions in Hilbert spaces.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4359090)