6

Let $X$ be a compact Hausdorff space without any isolated point. Show that $X$ is uncountable.

As $X$ is compact Hausdorff, it is normal. then for any two distinct points $x$ and $y$, we have a continuous map $f$ from $X$ to $[0,1]$ such that $f(x)=0$ and $f(y)=1.$
As $X$ is compact, $f(X)$ is closed and bounded in $[0,1]$. Now I want to show that $f(X)$ contains an interval by using the fact that $X$ has no isolated point. Some one Help me.

Stefan Hamcke
  • 27,733
Jana
  • 475
  • 2
    $f(X)$ need not include an interval, so you should use an alternative approach. Also, you (or whoever posed the question originally) should add the hypothesis that $X$ is nonempty. – Andreas Blass Jun 28 '13 at 00:48
  • May be relevant: http://math.stackexchange.com/questions/281332/cardinality-of-a-locally-compact-space-without-isolated-point – Seirios Jun 28 '13 at 11:35

2 Answers2

6

One way to do this is via a version of the Baire Category Theorem, which states that every locally compact Hausdorff space is a Baire space, i.e. it is not the union of a countable set of nowhere dense subsets. Recall that compact implies locally compact, so $X$ is not the union of a countable set of nowhere dense subsets. But $$X=\bigcup_{x\in X}\{x\}$$ and each $\{x\}$ is nowhere dense since $x$ is not an isolated point, thus $X$ must be uncountable.

Alex Becker
  • 60,569
3

The idea of using normality to show that $X$ can be mapped onto the interval $[0,1]$ is a good one, but it doesn’t work: if $X=A\cup B$, where $A$ and $B$ are disjoint clopen (= closed and open) sets, your function will be

$$f:X\to[0,1]:x\mapsto\begin{cases}0,&\text{if }x\in A\\1,&\text{if }x\in B\end{cases}$$

or the opposite one that takes $A$ to $1$ and $B$ to $0$: it will assume only the two values $0$ and $1$ and therefore will show only that $|X|\ge 2$, which doesn’t help.

This answer to a slightly different question uses an idea that can be modified to work here. In fact, it’s simpler in this setting. The basic idea is to start with non-empty open sets $U_0$ and $U_1$ whose closures are disjoint. Inside $U_0$ find non-empty open sets $U_{00}$ and $U_{01}$ whose closures are disjoint, and inside $U_1$ find non-empty open sets $U_{10}$ and $U_{11}$ whose closures are disjoint. In general, if $\sigma$ is any finite sequence of zeroes and ones, inside $U_\sigma$ find non-empty open sets $U_{\sigma^\frown 0}$ and $U_{\sigma^\frown 1}$ whose closures are disjoint. (Here $\sigma^\frown i$ is the finite sequence obtained by appending $i$ to $\sigma$.) The hypotheses on $X$ ensure that this is possible; why?

Now show that if $\sigma=\langle i_k:k\in\Bbb N\rangle$ is any infinite sequence of zeroes and ones, then

$$\bigcap_{k\in\Bbb N}\operatorname{cl}U_{i_0i_1\dots i_k}\ne\varnothing\;,$$

and use this to show that $|X|\ge 2^\omega=\mathfrak{c}$.

You may find that it helps your intuition to think of the sets $\operatorname{cl}U_\sigma$ as being like the closed intervals in the construction of the middle-thirds Cantor set: $\operatorname{cl}U_0$ is like $\left[0,\frac13\right]$, $\operatorname{cl}U_1$ is like $\left[\frac23,1\right]$, $\operatorname{cl}U_{01}$ is like $\left[\frac29,\frac13\right]$, and so on.

Brian M. Scott
  • 616,228
  • Going further, does a compact Hausdorff space without isolated point contain a subspace homeomorphic to the Cantor space? – PatrickR Apr 25 '23 at 21:02
  • @PatrickR: No. For $E\subseteq(0,1)\subseteq\Bbb R$ let $\Bbb D(E)=\big([0,1]\times{0}\big)\cup\big(E\times{1}\big)$ with the lexicographic order topology. (E.g., $\Bbb D\big((0,1)\big)$ is the version of the double arrow space without isolated points.) Then $\Bbb D(E)$ is a compact, separable LOTS with no isolated points, and $\Bbb D(E)$ contains a copy of the Cantor set iff $(0,1)\setminus E\subseteq\Bbb R$ contains a copy of the Cantor set. In particular, $\Bbb D\big((0,1)\big)$ does not contain a copy of the Cantor set. – Brian M. Scott Apr 27 '23 at 21:41
  • Thank you, this seems like a wonderful example. The part I don't see is the assertion about containing a copy of the Cantor set. Should I ask a question about the double arrow space not containing such a copy? (or maybe you have a reference?) – PatrickR Apr 27 '23 at 22:20
  • 1
    @PatrickR: I don’t have a pointer to the result for general $\Bbb D(E)$, but Proposition $\mathbf{2.1}$ of this paper gives you a proof for the double arrow space in particular. – Brian M. Scott Apr 27 '23 at 22:49
  • 1
    Thanks for the reference! – PatrickR Apr 28 '23 at 00:13
  • @PatrickR: My pleasure! – Brian M. Scott Apr 28 '23 at 04:15