Idempotent probability measures on ultrametric spaces (Q924245)

From MaRDI portal
Revision as of 10:02, 28 June 2024 by ReferenceBot (talk | contribs) (‎Changed an Item)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
scientific article
Language Label Description Also known as
English
Idempotent probability measures on ultrametric spaces
scientific article

    Statements

    Idempotent probability measures on ultrametric spaces (English)
    0 references
    0 references
    0 references
    15 May 2008
    0 references
    The authors introduce a metric on the set of idempotent probability measures defined on an ultrametric space. Their construction determines a functor on the category of ultrametric spaces and nonexpanding maps, which turns out to be the functorial part of a monad on this category. Let \(X\) be a compact Hausdorff space and let \(C(X)\) denote the space of continuous functions on \(X\). Let \({\mathbf 1}\in C(X)\) denote the constant \(1\) function, define \(\odot : \mathbb R \times C(X) \to C(X)\), \(c \odot \varphi = c \cdot {\mathbf 1}+\varphi\) and \(\oplus : C(X) \times C(X) \to C(X)\), \( \varphi \oplus \psi = \max\{\varphi, \psi\}\). A functional \(\mu : C(X) \to \mathbb R\) is an \textit{idempotent probability measure}, or Maslov measure, if for every \(c \in \mathbb R\) and \(\varphi, \psi \in C(X)\) we have \(\mu(c \cdot{\mathbf 1})=c\), \(\mu (c \odot \varphi)=c \cdot \mu(\varphi)\) and \(\mu( \varphi \oplus \psi) = \max\{\mu( \varphi), \mu( \psi)\}\). The set of idempotent probability measures on \(X\) is denoted by \(I(X)\). If \(X\) and \(Y\) are compact Hausdorff spaces and \(f : X \to Y\) is a continuous function, \(I(f) : I(X) \to I(Y)\) is defined by \([I(f)(\mu)](\varphi) = \mu(\varphi \circ f)\) \((\varphi \in C(Y))\). If \(X\) and \(Y\) are arbitrary Tychonov spaces, \(I(X)\) is defined as the direct limit of the directed system \[ \{I(A), I(i_{AB}) : A,B \in \exp(X),\;A \subseteq B\} \] where \(\exp(X)\) stands for the hyperspace of nonempty compact subsets of \(X\) and for \(A,B \in \exp(X)\) with \(A \subseteq B\), \(i_{AB} : A \to B\) is the inclusion map. For a continuous function \(f : X \to Y\), \(I(f) : I(X) \to I(Y)\) is defined by \(I(f)(\mu)= I(f|_{A})(\mu)\) where \(A \subseteq X\) is any compact set satisfying \(\mu \in I(A)\). Now let \((X,d)\) be an ultrametric space. For every \(\varepsilon > 0\), let \(\mathcal F _{\varepsilon }(X)\) be the set of those functions \(\varphi \in C(X)\) which satisfy that for every \(y \in \varphi[X]\), \(\varphi^{-1}(y)\) is the union of open balls of radii \(\varepsilon\). A central result of the paper is that the function \(\hat d : I(X) \times I(X) \to \mathbb R\), \[ \hat d(\mu, \nu) = \inf\{\varepsilon > 0 : \mu(\varphi) = \nu(\varphi)\;(\varphi \in \mathcal F_{\varepsilon}(X))\} \] is an ultrametric on \(I(X)\). Moreover, if \((X,d)\) is complete then \((I(X), \hat d)\) is complete, as well. With this metric on \(I(X)\), \(I\) turns out to be a functor on the category of ultrametric spaces and nonexpanding maps. It is also obtained that \(I\) is the functorial part of a monad on this category which contains the hyperspace monad.
    0 references
    idempotent probability measure
    0 references
    ultrametric space
    0 references
    Maslov measure
    0 references
    nonexpanding map
    0 references
    hyperspace monad
    0 references
    functor
    0 references

    Identifiers

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