On quantization dimensions of idempotent probability measures (Q2065108)

Let \(\mathcal{F}\) be a seminormal metrizable functor in the category \textit{Comp} of compact Hausdorff spaces, \((X,\rho)\) be a metric compactum, \(\rho_{\mathcal{F}}\) a functorial extension of the metric \(\rho\) onto \(\mathcal{F}(X)\) and \(\mathcal{F}_{n}(X)\) the subspace of \(\mathcal{F}(X)\), consisting of points \(\xi\), such that support \(\mbox{supp}(\xi)\) contains at most \(n\) elements (\(n\in\mathbb{N}\)). The set \(\underset{n\in\mathbb{N}}\bigcup\mathcal{F}_{n}(X)\) is dense in \(\mathcal{F}(X)\). For each \(\xi\in\mathcal{F}(X)\) and \(\varepsilon>0\) put \(N(\xi,\varepsilon)=\min\{n:\,\,\rho_{\mathcal{F}}(\xi,\mathcal{F}_{n}(X))\leq\varepsilon\}\). For any point \(\xi\notin\underset{n\in\mathbb{N}}\bigcup\mathcal{F}_{n}(X)\), the number \(N(\xi,\varepsilon)\) increases indefinitely when \(\varepsilon\to 0\). The rate of this increase is characterized by the value \[ \dim_{\mathcal{F}}(\xi)=\underset{\varepsilon\to 0}\lim \frac{\log N(\xi,\varepsilon)}{-\log\varepsilon}, \] which we call the dimension of finite approximation of the point \(\xi\) (if the specified limit does not exist, the upper and lower limits are considered, and we get the upper \(\overline{\dim}_{\mathcal{F}}(\xi)\) or the lower \(\underline{\dim}_{\mathcal{F}}(\xi)\) dimension of finite approximation). \textbf{Definition 2.1.} A functional \(\mu:C(X)\to\mathbb{R}\) is called an idempotent probability measure if for any \(f,g\in C(X)\) and \(c\in\mathbb{R}\) 1) \(\mu(c_{X})=c\); 2) \(\mu(c_{X}+f)=c+\mu(f)\); 3) \(\mu(\max\{f,g\})=\max\{\mu(f),\mu(g)\}\). The set of idempotent probability measures is denoted by \(I(X)\). Any functional \(\mu\in I(X)\) is \textit{order-preserving}. In this paper the author proves the following results. \textbf{Theorem 3.2.} The metric \(\rho_{I}\) determines a metrization of the functor \(I\). \textbf{Theorem 5.1.} For any metric compactum \((X,\rho)\) of dimension \(\overline{\dim}_{B}X=a\) and any real number \(b\) with \(0\leq b\leq a\) there exists a measure \(\mu_{b}\in I(X)\) such that \(\overline{D}_{I}(\mu_{b})=b\) and \(\mbox{supp}(\mu_{b})=X\). \textbf{Theorem 5.5.} If in a metric compactum \((X,\rho)\) the cardinality of \(\varepsilon\)-separated sets is locally bounded and \(l\underline{\dim}_{B}X=a>0\), then for any nonnegative number \(b<a\) there exists a closed subset \(Z_{b}\subset X\) such that \(\underline{\dim}_{B}Z_{b}=b\). \textbf{Theorem 5.7.} If in a metric compactum \((X,\rho)\) the cardinality of \(\varepsilon\)-separated sets is locally bounded and \(l\underline{\dim}_{B}X=a>0\), then for of any nonnegative number \(b<a\) there exists a measure \(\mu_{b}\in I(X)\) for which \(\underline{D}_{I}(\mu_{b})=b\) and \(\mbox{supp}(\mu)=X\).