Extension of Simons’ inequality (Q5313329)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 2199861
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Extension of Simons’ inequality |
scientific article; zbMATH DE number 2199861 |
Statements
Extension of Simons’ inequality (English)
0 references
29 August 2005
0 references
Simons' inequality
0 references
convex sets in topological vector spaces
0 references
convex functions
0 references
uniformly convergent convex combinations
0 references
Banach space geometry
0 references
0 references
0.9650189
0 references
0.8426108
0 references
0 references
\textit{S. Simons} [Pac. J. Math. 40, 703--708 (1972; Zbl 0237.46012)] obtained the following: Let \(S\) be a set and \(T\) a subset of \(S\). Suppose \((x_ n)\) be a sequence in \(l_\infty(S) \) and assume that for every \[ x \in C_ 1 = \left\{\sum^{\infty}_ {n=1}\lambda_ n x_ n : \lambda_ n \geq 0, ~~ \sum^{\infty}_{n=1} = 1\right\}, \] there exists a \(t \in T \) such that \[ x(t) = \sup_{s\in S} x(s). \] Then \[ \inf_{x \in C_ 0}\sup_{s \in S} x(s) \leq \sup_{t \in T}\limsup_ n x_ n (t), \] where \(C_ 0\) is the convex hull of the sequence \((x_ n)\). The above result is today known as Simons' inequality and has applications in real analysis and the geometry of Banach spaces. In this paper, the authors obtain an extension of the above result by replacing the continuity assumption on \(f(.,s),~ s \in S\), of \textit{R. Deville} and \textit{C. Finet} [Rev. Mat. Univ. Complut. 14, 95--104 (2001; Zbl 0993.46003)] with purely algebraic properties. Specifically, the authors establish the following results: Let \(S\) be a set and \(T\) a subset of \(S\). Let \(C\) be a subset of a Hausdorff topological vector space which is invariant under infinite convex combinations. Let \(f: C\times S \to \mathbb R\) be a bounded function such that the functions \(f(.,t): C \to \mathbb{R}\) are convex for all \(t \in T\) and \(f(\lambda x, s) = \lambda f(x,s)\) whenever \(\lambda > 0,~ x, \lambda x \in C\) and \(s \in S.\) Let \((x_ n)\) be a sequence in \(C\). Assume that for every \[ x \in C_ 1 = \left\{\sum^{\infty}_{n=1} \lambda_ n x_ n : \lambda_ n \geq 0,~~ \sum^{\infty}_{n=1} \lambda_ n = 1\right\}, \] there exists \(t \in T\) satisfying \[ f(x,t) = \sup_{s \in S} f(x,s). \] Then \[ \inf_{x \in C_ 1}\sup_{s \in S} f(x,s) \leq \sup_{t \in T}\limsup_ n f( x_ n,t). \] If \(-C_ 1 \subset C,\) then the set \(C_ 1\) in the above inequality can be replaced by \(\text{conv}\left\{x_ 1, x_ 2, \dots\right\}.\)
0 references