2
$\begingroup$

Let $\mathcal D$ be an upper semicontinuous decomposition of $\mathbb S^n$ and let $\mathcal D'\subset\mathcal D$ be the set of non-singletons. The decomposition space $^{\mathbb S^n}/_{\mathcal D}$ is homeomorphic to $\mathbb S^n$ if $\mathcal D$ is "shrinkable", and there are many conditions on $\mathcal D$ that ensure this. (For example, there are numerous such conditions in Daverman's book "Decompositions of Manifolds".)

However, the statements I've been able to find assume, in one way or another, that $\mathcal D'$ is countable.

Is there some set of conditions on an upper semicontinuous decomposition $\mathcal D$ of $\mathbb S^n$ that ensures that the decomposition space $^{\mathbb S^n}/_{\mathcal D}$ is homeomorphic to $\mathbb S^n$ (e.g. by ensuring that $\mathcal D$ is shrinkable), but that does not require the set $\mathcal D'$ of non-singletons in $\mathcal D$ to be countable?

(In the situation I have in mind, $\cup_{D\in\mathcal D'}D$ has dense, full-measure complement, each $D\in\mathcal D'$ is homeormorphic to a closed ball in $\mathbb E^k$ for some $k<n$, and $^{\mathbb S^n}/_{\mathcal D}$ is compact, Hausdorff, and second-countable. Moreover, the quotient $\mathbb S^n\to^{\mathbb S^n}/_{\mathcal D}$ has a very nice description. However, my $\mathcal D'$ is definitely uncountable.)

$\endgroup$
1
  • $\begingroup$ For the sake of completeness I have to mention the Bing shrinking criterion (although you are certainly aware of it). Apart from this I am aware of a recent [paper][1] of Kasprowski and Powell, where they consider so called toroidal decompositions (the decomposition elements arise as a intersection of embedded solid tori). Usually these have uncountably many nontrivial decomposition elements). You might want to have a look at it. [1]: arxiv.org/pdf/1307.0154v3.pdf $\endgroup$ Commented Aug 7, 2015 at 19:22

1 Answer 1

4
$\begingroup$

Bing's Dogbone Space stems from a decomposition of $S^3$ unto points and an uncountable collection of tame arcs. The decomposition space is not a manifold, so the decomposition is non-shrinkable. W. T. Eaton (Proc. Amer. Math. Soc. 39 (1973), 379--387) presented a higher dimensional analog of the Dogbone Space, involving a decomposition of $S^n$, $n>3$, into points and tame arcs that yields a nonmanifold.

Positive results in the general setting are nowhere near as rich as when the set of nondegenrate elements is countable. The best of them is Edwards's Cell-kike Approximation Theorem, which promises that for $n>4$ a cell-like decomposition of an $n$-manifold $M$ is shrinkable if the quotient space is finite dimensional and satisfies the Disjoint Disks Property.

$\endgroup$
2
  • $\begingroup$ Welcome to mathoverflow professor Daverman! $\endgroup$
    – Paul Fabel
    Commented Aug 8, 2015 at 4:56
  • $\begingroup$ Thank you very much, Prof. Daverman. Both the dogbone warning and the Disjoint Discs suggestion will be quite helpful. $\endgroup$
    – Mark Hagen
    Commented Aug 10, 2015 at 14:21

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .