Comparing \(n\)-fold and \(m\)-fold hyperspaces. (Q1408723): Difference between revisions

By a continuum we understand a nondegenerate compact connected metric space. The hyperspace of all closed nonempty subsets of \(X\) which have at most \(n\) components is denoted by \({\mathcal C}_{n}(X)\), and it is always endowed with the Hausdorff metric. A monotone map is a continuous function with connected fibers. A continuum \(X\) which is irreducible between two points \(p\) and \(q\) is said to be of type \(\lambda\), provided that there exists a monotone map \(\gamma\) from \(X\) to the unit interval \(I\) such that \(\gamma (p) =0\), \(\gamma (q)=1\) and \(int_{X}(\gamma^{-1}(t))=\emptyset\) for each \(t\in I\). In addition, if \(\gamma^{-1}(t)\subset cl_{X}(\gamma^{-1}(]t,1]))\) for each \(t<1\), and \(\gamma^{-1}(t)\subset cl_{X}(\gamma^{-1}([0,t[))\) for each \(t>0\), then \(X\) is said to be of type \(\lambda^{\star}\). The paper under review deals with the following question: suppose that dim\([{\mathcal C}_{n}(X)]\) is finite and \({\mathcal C}_{n}(X)\) is homeomorphic to \({\mathcal C}_{m}(X)\), is it true that \(n=m\) ? Notice that the hypothesis ``dim\([{\mathcal C}_{n}(X)]\) finite'' in the previous question is a natural one. In fact, if \(X\) is locally connected and each arc in \(X\) has empty interior, then \({\mathcal C}_{n}(X)\) is homeomorphic to the Hilbert cube [\textit{S. Macías}, Topology Appl. 109, 237--256 (2001; Zbl 0979.54013)]. The author supposes that dim\([{\mathcal C}_{n}(X)]\) is finite, \({\mathcal C}_{n}(X)\) is homeomorphic to \({\mathcal C}_{m}(X)\) and \(n\leq m\) and, under these conditions, he presents the following results: (i) if \(X\) is almost hereditarily indecomposable, then \(n=m\), (ii) \(m\leq 2n\), and (iii) if \(X\) contains a subcontinuum \(Z\) of type \(\lambda^{\star}\), then \(n=m\). The previous results imply that the equality \(n=m\) holds whenever \(X\) is hereditarily indecomposable or it contains an arc.