2

I thought I knew that but it seems I don't.

Let $\alpha$ be a countable, limit ordinal $\alpha>\omega$. Give $\alpha$ its order topology. What is the Stone-Čech compactification of $\alpha$? Is there any reason why it should be $\beta \omega$?

Stefan Hamcke
  • 27,733
Jorg
  • 691

2 Answers2

2

If $\alpha>\omega$, then the subspace $\omega+1$ of $\alpha$ is compact. It is therefore a compact subset of $\beta\alpha$ and hence a closed subset. But every infinite closed subset of $\beta\omega$ contains a copy of $\beta\omega$, so $\beta\omega$ contains no set homeomorphic to $\omega+1$. Thus, $\beta\alpha$ cannot be homeomorphic to $\beta\omega$.

If $\alpha=\beta+\omega$ for some limit ordinal $\beta$, then $\alpha$ is homeomorphic to the disjoint union of $\beta+1$ and $\omega$, and since $\beta+1$ is compact, $\beta\alpha$ is homeomorphic to $(\beta+1)\sqcup\beta\omega$.

It’s not clear to me just what happens when $\alpha$ is more complicated, even for $\omega^2$.

Brian M. Scott
  • 616,228
  • Could you please explain where does the $\beta\omega$-bit come from? – Jorg Jun 09 '13 at 20:32
  • @usao: The $\beta\omega$ part of $(\beta+1)\sqcup\beta\omega$? From the $\omega$ part of $(\beta+1)\sqcup\omega$, i.e., the points $\beta+n$ for $n\ge 1$ in $\beta+\omega$. Or did you mean the fact about $\beta\omega$ that I used in the first paragraph? – Brian M. Scott Jun 09 '13 at 20:35
  • Ah, I see, sure! – Jorg Jun 09 '13 at 20:36
  • In the case where $\alpha$ is another ordinal plus $\omega$, the world would be a better place if the other ordinal were not named $\beta$ when Stone-Cech compactifications are present. – Andreas Blass Jun 09 '13 at 21:48
  • @Andreas: :-) I noticed that as soon as I posted. I did think about changing it, but then I decided to stick with the OP's notation. – Brian M. Scott Jun 10 '13 at 04:55
1

If $\alpha > \omega$, then $\beta \alpha$ is not homeomorphic to $\beta \omega$. For $\omega + 1$ is a countably infinite subset of $\beta \alpha$ which is open and closed, while $\beta \omega$ contains no such subset.

Suppose $U \subset \beta \omega$ is infinite, open and closed. Set $V = U \cap \omega$. If $V$ is finite then it is closed, so $U \setminus V$ is a nonempty open set containing no point of the dense set $\omega$, which is absurd. Thus $V$ must be infinite. By the universal property of the Stone-Čech compactification, the closure $\bar{V}$ is homeomorphic to $\beta V$ which is uncountable. Hence as $\bar{V} \subset U$, $U$ is uncountable.

Stefan Hamcke
  • 27,733
Nate Eldredge
  • 97,710