0
$\begingroup$

Let $M$ be a topological space. I want to show the equivalence of the following statements.

  1. There is an open subset $U\subset M$ and a homeomorphism $\Psi:U\rightarrow\mathbb{R}^n$.
  2. There are open subsets $U^\prime\subset M$, $V\subset\mathbb{R}^n$ and a homeomorphism $\varphi:U^\prime\rightarrow V$.
  3. There is an open subset $U^{\prime\prime}\subset M$, an open ball $B\subset \mathbb{R}^n$ and a homeomorphism $\phi:U^{\prime\prime}\rightarrow B$.

  • Let (1) be given. Then for any open subset $V\subset\mathbb{R}^n$ let $U^\prime:=\Psi^{-1}(V)$. Since $\Psi$ is a homeomorphism, the map $\varphi:=\Psi|_{U^\prime}$ is also a homeomorphism.

  • Let (2) be given. Since $V$ is open, it contains an open ball $B$. Now let $U^{\prime\prime}:=\varphi^{-1}(B)$. Then again, $\phi:=\varphi|_{U^{\prime\prime}}$ is a homeomorphism.


Now, how do I show the last implication from (3) to (1)? I thought of showing that any open ball is homeomorphic to $\mathbb{R}^n$ itself. But that's false. Then I thought of composing a homeomorphism onto $\mathbb{R}^n$ out of homeomorphisms $\phi_i$ onto disjoint open balls $B_i$, such that $\cup B_i=\mathbb{R}^n$. But my intuition tells me that that's impossible.

Any thoughts?

$\endgroup$
6
  • 1
    $\begingroup$ "I thought of showing that any open ball is homeomorphic to $ℝ^n$ itself. But that's false." Are you sure about that? $\endgroup$ Commented Jun 14, 2018 at 10:36
  • $\begingroup$ Incidentally, you can simplify the proof for (1) $\implies$ (2), by just taking $V = \mathbb R^n$ (and $U' = U$, $\varphi = \Psi$). $\endgroup$ Commented Jun 14, 2018 at 10:43
  • $\begingroup$ Oh, you're right - it is indeed right. $\endgroup$ Commented Jun 14, 2018 at 10:57
  • $\begingroup$ Manifold not space, correct? $\endgroup$
    – Randall
    Commented Jun 14, 2018 at 11:00
  • $\begingroup$ Yes, indeed. In the original assignment it is a topological manifold. It is to show that there are equivalent definitions of the latter. But actually those structures aren't even necessary for the proposition itself to hold. Therefore, I will amend my original post by simply requiring $M$ to be a topological space. $\endgroup$ Commented Jun 14, 2018 at 11:06

2 Answers 2

1
$\begingroup$

You had a good idea because there is always a homeomorphism between $R^n$ and the open ball of $ R^n$. Think of the one-dimensional case in which you would like to map $R$ to $(-1,1)$: In this case the good map can be $f(x)=\frac{2}{\pi}arctan(x)$ . Oviously you can generalize this concept to a higher size.

$\endgroup$
1
$\begingroup$

Since any open subset $B\subset \mathbb{R} ^n$ is homeomorphic to $\mathbb{R} ^n$ by some $\phi^\prime$ we can define the map $\Psi:=\phi^\prime\circ\phi$. As a composition of homeomorphisms it is itself a homeomorphism.

The proposition is therefore valid, q.e.d..

$\endgroup$

You must log in to answer this question.

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