Convex hyperspaces of probability measures and extensors in the asymptotic category (Q554416): Difference between revisions

For metric spaces \((X, d_{X}), \; (Y, d_{Y})\), a mapping \(f: X \to Y\) is called asymptotically \((\lambda, \varepsilon)\)-Lipschitz for \(\lambda >0, \; \varepsilon >0\) if \(d_{Y}(f(x), f(x')) \leq \lambda d_{X}(x, x') + \varepsilon\) for all \(x,x'\in X\). If every closed ball of a metric space is compact, then the metric space is called proper; a map between metric spaces is called proper if preimages of bounded sets are bounded. The category \(\mathcal{A}\) has as objects proper metric spaces, and its morphisms are the proper asymptotically Lipschitz maps. A metric space \(Y\) (not necessarily an object of \(\mathcal{A}\)) is an absolute extensor, in asymptotic dimension \(0\), AE(0) for category \(\mathcal{A}\) if for every proper asymptotically Lipschitz map \(f: A \to Y \), defined on a closed subset \(A\) of a proper metric space \(X\) of asymptotic dimension \(0\), there exists a proper asymptotically Lipschitz extension \( \bar{f}: X \to Y\). For a metric space \(X\), \(P(X)\) denotes the spaces of probability measures with compact supports and \( P(X)\) is given the Kantorovich metric; \(ccP(X)\) denotes the set of all nonempty compact convex subsets of \(P(X)\) and is given the Hausdorff metric. The main result is the following example: The authors give an example of two proper metric spaces \(X, Y\), each of asymptotic dimension \(0\), such that \(X\) a closed subset of \(Y\). They consider the mapping \(f: X\to ccP(X), \; x \mapsto \delta_{x}, \) and prove that this does not have a Lipschitz extension \(\bar{f}: Y \to ccP(X)\). Thus \(ccP(X)\) is not an absolute extensor.