MathOverflow will be down for maintenance for approximately 3 hours, starting Monday evening (06/24/2013) at approximately 9:00 PM Eastern time (UTC-4).

## Actions on Sⁿ with quotient Sⁿ

What is known about isometric actions on $\mathbb S^n$ such that the quotient space is homeomorphic to $\mathbb S^n$?

• I am mostly interested in (maybe trivial) properties of such actions for large $n$. Say, it is true that any such action is generated by rotations around $\mathbb S^{n-2}$'s; what else?
• I see that the orientation preserving part of Coxeter's group has this property.
• Now I see that there are other examples for $\mathbb S^3$, thanks to Lee Mosher. It seems that taking joints you get such examples in higher dimensions.
-
Your $S^n$'s didn't come out well; I put them into TeX. – Lee Mosher Jul 25 at 13:24
I prefer this way, it visible on the question list. – Anton Petrunin Jul 25 at 13:43
It's probably browser dependent. On my browser it looks the same in the question list as on the actual page, which is to say, indecipherable, like a little box with six binary digits. – Lee Mosher Jul 25 at 13:55
That's why, as you'll see, I reverted it briefly, I thought it might look better in the question list. But it didn't. – Lee Mosher Jul 25 at 13:56
I just learned that this question is very close to conjecture on p.9 of Lectures on orbifolds and reflection groups by Michael W. Davis, math.osu.edu/~davis.12/papers/… – Anton Petrunin Jul 30 at 12:20

In the following article of M.A.Mikhailova (М.А. Михайлова)

Изв. АН СССР. Сер. матем., 48:1 (1984)

О ФАКТОРПРОСТРАНСТВЕ ПО ДЕЙСТВИЮ КОНЕЧНОЙ ГРУППЫ, ПОРОЖДЕННОЙ ПСЕВДООТРАЖЕНИЯМИ.

it is proven:

Theorem. The quotient $\mathbb R^n/G$ by a linear action of a finite group $G$ is homeomorphic to $\mathbb R^n$ if and only if $G$ is generated by pseudo-reflections (i.e, rotations of $\mathbb R^n$ that fix a subspace of codimension 2).

The proof relies on a complete classification of finite groups generated by pseudo-reflections (there is a reference to this classification at the end of the article)

(there should be of course an English translation of this article, but I can not find it now).

-
She probably wanted to say "PL-homeomorphic" otherwise cone over double suspension over Poincare sphere would be a counterexample. – Anton Petrunin Jul 29 at 14:45
The article is available at iopscience.iop.org/0025-5726/24/1 for real money. I was able to get a copy through the proxy server of my university library. – Lee Mosher Jul 29 at 19:26

Your question translates into the language of orbifolds as saying: what is known about spherical $n$-orbifolds with underlying space homeomorphic to $S^n$?

In $S^2$, the examples you give are all there are.

Orbifolds with the geometry of $S^3$ were enumerated by William Dunbar in his thesis. His published paper MR1118824 contains the enumeration of the 21 oriented $S^3$-orbifolds which do not have a circle fibration over a 2-orbifold. The equivalence relation here is up to orientation preserving isometry; if you allow orientation reversing isometry then the list is cut down somewhat. Each of the 21 has underlying space homeomorphic to $S^3$. At the end of Dunbar's paper you will see that exactly 8 of the 21 are orientable double covers of Coxeter group quotients, with the corresponding Dynkin diagrams listed out explicitly. That leaves 13 examples as you ask for in $S^3$.

-