Embedding theorems for classes of convex sets (Q1078806)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Embedding theorems for classes of convex sets
scientific article

    Statements

    Embedding theorems for classes of convex sets (English)
    0 references
    1986
    0 references
    The author improves the well known embedding theorems of \textit{H. Rådström} [Proc. Am. Math. Soc. 3, 165-169 (1952; Zbl 0046.333)] and \textit{L. Hörmander} [Ark. Mat. 3, 181-186 (1955; Zbl 0064.105)], using the original constructive method of Rådström and its refinement proposed by \textit{C. Godet-Thobie} and \textit{Pham the Lai} [C.R. Acad. Sci., Paris, Ser. A 271, 84-87 (1970; Zbl 0197.094)]. For a Hausdorff locally convex real vector space E the embedding space will be constructed in two steps. Step 1. Let \({\mathbb{F}}\) be the family of all non-empty closed bounded convex subsets of E ordered by inclusion and with respect to the following operations: (A,B)\(\to \overline{A+B}\) and \((\rho,A)=\rho A\) (A,B\(\in {\mathbb{F}}\), \(\rho\in [0,\infty))\). Let \(\{p_{\gamma}| \gamma \in \Gamma \}\) be the family of seminorms which determines the topology of E, and \(d_{\gamma}(A,B):=\inf \{\epsilon \in (0,\infty)| \quad A\subset B+U_{\gamma}(\epsilon),\quad B\subset A+U_{\gamma}(\epsilon)\}\) where \(U_{\gamma}(\epsilon)=\{x\in E| \quad p_{\gamma}(x)<\epsilon \}.\) The first main result asserts that \(\{d_{\gamma}| \gamma \in \Gamma \}\) is a separating family of monoton, positively homogeneous, translation invariant semimetrics on \({\mathbb{F}}\), and it turns \({\mathbb{F}}\) into a Hausdorff locally convex-solid semilattice. Step 2. In \({\mathbb{F}}\times {\mathbb{F}}\) the pair (A,B) and (C,D) by definition are equivalent if \(\overline{A+D}=\overline{B+C}\) holds. Let \({\mathbb{G}}\) be the set of all equivalence classes \(<A,B>\). On \({\mathbb{G}}\) define \[ <A,B>+<C,D> := <\overline{A+C},\overline{B+D}>;\quad \rho<A,B> := \begin{cases} <\rho A,\rho B>\;&\text{if \(\rho\in\mathbb{R}_+\)} \\ <(-\rho)A,(-\rho)B>\;&\text{if \(\rho\in\mathbb{R}\setminus\mathbb{R}_+\)} \end{cases} \] and the order relation \(<A,B>\leq <C,D>\Leftrightarrow \overline{A+D}\subset \overline{B+C}.\) The second main theorem states: \({\mathbb{G}}\) is a Riesz space and \(\{r_{\gamma}| \gamma \in \Gamma \}\) where \(r_{\gamma}(<A,B>):=d_{\gamma}(A,B)\) is a separating family of Riesz seminorms of type M. The topological completion of \({\mathbb{G}}\), denoted by \({\mathbb{H}}\), will be the embedding space of \({\mathbb{F}}\), and the mapping \(j: {\mathbb{F}}\to {\mathbb{H}}\) defined by \(j(A)=<A,\{0\}>\) the embedding map. The main embedding theorem (Theorem 4.5) asserts that: \({\mathbb{H}}\) is a complete Hausdorff locally convex-solid and j is additive, positively homogeneous, isotone, preserves finite suprema, continuous; \({\mathbb{F}}\) and j(\({\mathbb{F}})\) are homeomorphic and the linear hull of j(\({\mathbb{F}})\) is dense in \({\mathbb{H}}.\) In the metrizable case, if E is a Fréchet space, then \({\mathbb{F}}\) is complete (Theorem 5.2) and if E is normed, then \({\mathbb{H}}\) is an AM-space with unit and j is an isometry. A similar embedding theorem construction, \({\mathbb{H}}_ 0\), for the class \({\mathbb{F}}_ 0\) of compact sets from \({\mathbb{F}}\) is given in Section 7. In the normable case \({\mathbb{H}}\) and \({\mathbb{H}}_ 0\) can be represented as a space of continuous functions on some compact space. The paper includes a brief discussion of some actual and potential applications of these results to the Debreu integral, limit theorem for random sets, martingales, fixed point theorems, mathematical economics and interval mathematics.
    0 references
    integration of random sets
    0 references
    fixed-point theorems for set-valued
    0 references
    maps
    0 references
    embedding theorem of Rådström
    0 references
    embedding theorem of
    0 references
    Hörmander
    0 references
    classes of convex sets
    0 references
    topological vector lattices
    0 references
    embedding space
    0 references
    limit theorem for random sets
    0 references
    mathematical economics
    0 references
    interval mathematics
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references