7

It is a standard result that the open ball $$B^2=\{(x,y)\in\mathbb{R}^2:x^2+y^2<1\}$$ is homeomorphic to $\mathbb{R}^2$ itself. Also, distorting $B^2$ by any continuous bijective transformation will also give a path-connected open set homeomorphic to $\mathbb{R^2}$. So, I was wondering the following:

Question: Are all path-connected open subsets of $\mathbb{R}^2$ homeomorphic to $\mathbb{R}^2$?

I state the question in $\mathbb{R}^2$ for simplicity, but of course we could ask the same question in $\mathbb{R}^n$ which I suspect will have the same answer.

Chern
  • 73

2 Answers2

9

Take $\Bbb R^2-0$. A rather weird result is that every nonempty proper simply connected subset of $\Bbb R^2$ is homeomorphic, actually biholomorphic, to $B(0,1)$. This is the Riemann mapping theorem. It is rather weird because a biholomorphic mapping preserves angles and orientation, and one can produce rather erratic simply connected subspaces of $\Bbb R^2$ which can be mapped "nicely" into $B(0,1)$. As an example, take $(0,1)\times (0,1)$, and from the point $(0,0)$ draw a line of length say $1/n$ and angle of $2\pi /n$ radians for each $n$. This looks like a square with a "duster deleted in the corner" and is simply connected, but it is hard to imagine a conformal mapping of this onto the unit disk.

Pedro
  • 122,002
  • Right, indeed. Thanks for the quick answer. – Chern Dec 11 '14 at 22:45
  • Are all simply-connected open subsets of $\mathbb{R}^2$ homeomorphic to $\mathbb{R}^2$? – Chern Dec 11 '14 at 22:46
  • 1
    @Chern Yes! The Riemann mapping theorem states that there is a holomorphic map from any such region to the unit disk. – D Wiggles Dec 11 '14 at 22:47
  • @Chern You read my mind. =) – Pedro Dec 11 '14 at 22:48
  • @PedroTamaroff This is awesome. Is that particular to $\mathbb{R}^2$, or does it hold in any dimension? – Spenser Dec 11 '14 at 22:56
  • @Spenser It is particular to $\Bbb C$. I do not know if it generalizes. You can google "Riemann mapping theorem" for more information. – Pedro Dec 11 '14 at 23:09
  • 1
    @Spenser It does not generalize to $\mathbb{R}^n$ for $n\geq 3$ because, for example, $\mathbb{R}^n-{0}$ is simply connected and open. But there is maybe a condition on the homology of the set that would be enough though. I don't know. – Chern Dec 11 '14 at 23:14
  • @Chern +1, yet I never liked people using famous mathematicians' names for users. It feels slightly disrespectful. Unless you're actually a descendant of Chern. =/ – Pedro Dec 11 '14 at 23:18
  • @PedroTamaroff Thanks for the advice on how to choose a name. I sort of agree with you now that I think about it. – Chern Dec 11 '14 at 23:30
  • 3
    @Chern The Whitehead manifold is an open contractible subset of $\Bbb R^3$ that's not homeomorphic to $\Bbb R^3$. So if you want to generalize it, you need to use things that aren't homotopy invariant. I wouldn't hold out much hope. –  Dec 11 '14 at 23:39
1

Take any small enough open neighborhood of the $n$-sphere $S^n$, with $n\ge 1$, in $\mathbb R^{n+1}$. That is a path-connected open subset of the ambient space $\mathbb R^{n+1}$ which is not homeomorphic to $\mathbb R^{n+1}$.

Ittay Weiss
  • 79,840
  • 7
  • 141
  • 236