Covering versus partitioning with the Cantor space (Q2105041)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Covering versus partitioning with the Cantor space
scientific article

    Statements

    Covering versus partitioning with the Cantor space (English)
    0 references
    0 references
    8 December 2022
    0 references
    In this extremely pleasant paper, the author investigates the relationship between ``being coverable'' and ``being partitionable'', for various topological spaces. If \(X\) and \(Y\) are topological spaces, we say that \(X\) is \textit{\(Y\)-coverable} if \(X\) can be covered by homeomorphic copies of \(Y\); and we say that \(X\) is \textit{\(Y\)-partitionable} if it can be partitioned into homeomorphic copies of \(Y\). Thus, clearly, being \(Y\)-partitionable implies being \(Y\)-coverable, and the paper under review explores various circumstances under which the converse implication holds. After the first section containing an introduction, Section 2 establishes as its main result that every first countable space \(X\) with \(|X|\leq\mathfrak c\) is \(2^\omega\)-partitionable iff it is \(2^\omega\)-coverable. There is also a nice example showing that the condition \(|X|\leq\mathfrak c\) is necessary, and the bound of \(\mathfrak c\) is sharp. The next section's main result is that, for a metrizable space \(X\) satisfying \(|X|\leq\mathfrak c^{+\omega}\), being \(2^\omega\)-partitionable is equivalent to being \(2^\omega\)-coverable. The section also contains several lemmas leading to the construction of an example showing that the condition \(|X|\leq\mathfrak c^{+\omega}\) is necessary, and the bound of \(\mathfrak c^{+\omega}\) is sharp. Finally, in Section 4, the main result states that, for a completely metrizable space \(X\), four conditions are equivalent: being \(2^\omega\)-partitionable, being \(2^\omega\)-coverable, having no nonempty open \(\sigma\)-discrete subspace, and having no isolated points. There is, at the end, a short final section showing that the previous results (from all previous sections) remain true when one replaces \(2^\omega\) with any zero-dimensional Polish space without isolated points. A few final comments establish that the hypothesis that the incumbent space contains no isolated points is necessary, even though, on the other hand, for every zero-dimensional Polish space \(Y\) (regardless of whether it has isolated points), every \(2^\omega\)-partitionable space must be \(Y\)-partitionable as well.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    partitions
    0 references
    coverings
    0 references
    Cantor space
    0 references
    Baire space
    0 references
    0 references