Relative convexity and its applications (Q499650): Difference between revisions

If \(f: K\to \mathbb{R}\) is a function defined on a compact closed subset of \(\mathbb{R}^n\), and \(V\) is an infinite convex subset of \(K\), then \(a\in K\) is called a point of convexity (p.o.c.) of \(f\) relative to \(V\) if the inequality \[ f(a)\leq \sum_{k=1}^n\lambda_kf(x_k) \] holds for all \(x_k\in V\) and every positive weights \(\lambda_k\) such that \(\sum_{k=1}^n\lambda_k=1\) and \(\sum_{k=1}^n\lambda_kx_k=a\). It follows immediately that if \(f\) is convex, then every point in its its domain is a point of convexity relative to \(K\). But the class of functions possessing such points is much broader (e.g. every local minimum is a point of convexity relative to some neighboorhood). The authors discuss the existence of p.o.c and give numerous examples of classical theorems for convex functions that remain valid if convexity gets replaced with the new concept. Some examples are: the extension of Hardy-Littlewood-Pólya theorem of majorization and Popoviciu's inequality. An application to mathematical finance is given. A nice reading.