Best uniform convex approximation on a compact convex set (Q1329335)

Let \(K\) be a compact convex subset of \(\mathbb{R}^ n\), \(C(K)\) the space of continuous functions on \(K\) with the uniform norm \(\| f \| = \sup \{| f(x) | : x \in K\}\), and \(P(K)\) the closed cone of continuous convex functions on \(K\). A function \(g^* \in P(K)\) is said to be a best uniform convex approximation to \(f \in C(K)\) if \(\| f - g^* \| = \text{dist} (f,P(K)) : = \inf \{\| f - g \| : g \in P(K)\}\). When \(K\) is one dimensional, it is known that best uniform convex approximation always exist. A characterization of best uniform convex approximation of alternate type and a duality theorem that expresses the arise of the best uniform convex approximation in forms of the supremum of some linear functionals on \(C(K)\) are also known when \(K\) is one- dimensional. In this note these problems are discussed when \(K\) is not necessarily one-dimensional. It is shown that when \(K\) is two- dimensional, a best uniform convex approximation always exists. It is remarked that the author is not able to find a compact convex subset \(K\) of \(\mathbb{R}^ n\) and a continuous function \(f\) on \(K\) such that \(f\) does not admit any uniform convex continuous approximation.