A support theorem for delta \((s,t)\)-convex mappings (Q2350289): Difference between revisions

Let \(X, Y\) be real Banach spaces, let \(D\subset X\) be a convex set and let \(s,t\in (0,1)\) be fixed real numbers. The author says that a map \(F:D\rightarrow Y\) is delta \((s,t)\)-convex with a control function \(f:D\rightarrow \mathbb{R}\) if the inequality \(\|tF(x)+(1-t)F(y) - F(sx+(1-s)y)\|\leq tf(x) +(1-t)f(y) -f(sx+(1-s)y)\) holds for all \(x,y\in D\). He establishes necessary and sufficient conditions for a given map \(F\) to be delta \((s,t)\)-convex controlled by \(f\) in terms of the dual space \(Y^*\), and proves that in the case \(Y=\mathbb{R}\) endowed with the usual metric, a map \(F:D\rightarrow \mathbb{R}\) is delta \((s,t)\)-convex if and only if \(F\) is the difference of two \((s,t)\)-convex functions, that is, the difference of two maps \(f_1,f_2:D\rightarrow [-\infty,\infty)\) satisfying \(\bigwedge_{x,y\in D} f_j(sx+(1-s)y)\leq tf(x)+(1-t)f(y)\), \(j=1,2.\) His main result states that the existence of support mappings at an arbitrary point characterizes the delta \((s,t)\)-convexity, to be more precise, \(F\) is delta \((s,t)\)-convex with control \(f\) if and only if for every point \(y\in D\) there exist \((s,t)\)-affine maps (that is, \((s,t)\)-convex functions for which the equality holds in the definition of \((s,t)\)-convexity) \(A_y:D\rightarrow Y\) and \(a_y:D\rightarrow \mathbb{R}\) such that \(\|F(x)-A_y(x)\|\leq f(x)-a_y(x)\) for all \(x\in D\) and, moreover, \(A_y(y)=F(y)\), \(a_y(y)=f(y)\). It is interesting to emphasize that the present definition of delta \((s,t)\)-convexity is inspired by the definition of delta-convexity and its characterization by a functional inequality (with \(f,F\) continuous) given by \textit{L. Veselý} and \textit{L. Zajíček} [Diss. Math. 289, 48 p. (1989; Zbl 0685.46027)]. Notice that now the introduced definition of delta \((s,t)\)-convexity is given without any regularity assumption on \(F\) and \(f\).