0

I am a beginner in studying functional analysis, so please pardon me if my argument seems absurd. I am having hard time, coming in terms with the proof of "Compact subsets of $\mathbb{R}$, are closed". Now lets say that I have a set $ A = [1,2) $, and follow the proof. If I take the fixed point $x$ as $ {2} \in \mathbb{R} \backslash A $, then for each point $ y \in A$, I $\mathbf{can}$ find small disjoint neighborhoods for $x$ and $y$. I am not sure if these small neighborhoods of $y$ cover $A$, but it is given that $A$ is compact. Or is it that I have to prove that these neighborhoods of $y$ doesn't cover $A$? Thanks

Or Shahar
  • 1,766
Prashanth
  • 103

1 Answers1

5

$A$ is not compact:

Let $\alpha=\{(-1,1.1),(1,2-\frac{1}{n}) \mid \forall n \in \mathbb{N}\}$ be an open cover of $A$.

There is not any finite subcover of $\alpha$.

Moreover, in metric spaces, any compact set is closed and bounded, and Heine–Borel theorem says that in $\mathbb{R}^n$, any bounded and closed set is compact.

Edit: I tried to comment this but it is too long:

@Prashanth If I'm right, the proof's claim you sent is: A compact set $K$ in a Hausdorff topological space $X$ is closed. Which means that the complement of any compact set $K$ in a $T_2$ topological space $X$ is open. To see that, you have to show that $\forall x \in K^c$, there is an open neighbourhood $V$ such that $x \in V \subseteq K^c$ , (The definition of open set). and to show that, try to use the definition of compactness; the fact that any finite intersection of open sets is an open set; and the fact that $X \in T_2$ ($X$ is Hausdorff).

Hint: given a point $x \in K^c$ , because $X$ is Hausdorff, try to look at the (maybe infinite) union of the open sets $U_k\subseteq K$ such that $k\in U_k \subseteq K$ such that $U_k \cap V =\emptyset$, when $x\in V\subseteq K^c$, and $V$ is an open set. Do this for all $x\in K^c$, for all $x\in K^c$ use the fact that $K$ is compact and see what you got.

Or Shahar
  • 1,766
  • 1
    Minor quibble on terminology. One does not "Let" a specific set "be" an open cover of $A$. Since you've specified exactly what $\alpha$ is, it either is an open cover or it is not. This $\alpha$ is easily proven to be an open cover. But that is a matter of proof, not a hypothesis. What you should have said is "Let $\alpha = \ldots$. Then $\alpha$ is an open cover of $A$." Or you could just say "$\alpha := \ldots$ is an open cover of $A$." – Paul Sinclair Dec 06 '20 at 17:45
  • @PaulSinclair Thanks for correcting me! I appreciate it since English is not my native language. – Or Shahar Dec 06 '20 at 18:29
  • @OrShahar Thanks for answering. Actually I know open sets are not compact, because some open covers do not have finite subcover. However I have a problem with this specific proof. I think this was not clear in the question. All the standard text books have this proof. In this, an exterior point is picked and it goes on to show that a compact set is closed,( or for me it can be a closed set, ok). But how does this proof, prove that compact set cannot be an open set (in Hausdorff topology)? I cannot see this, and this is bothering me. Thanks – Prashanth Dec 07 '20 at 02:48
  • 1
    @Prashanth If I'm right, the proof's claim you sent is: A compact set $K$ in a Hausdorff topological space $X$ is closed. Which means that the complement of any compact set $K$ in a $T_2$ topological space $X$ is open. To see that, you have to show that $\forall x \in K^c$, there is an open neighbourhood $V$ such that $x \in V \subseteq K^c$ , (The definition of open set). and to show that, try to use the definition of compactness; the fact that any finite intersection of open sets is an open set; and the fact that $X \in T_2$ ($X$ is Hausdorff). – Or Shahar Dec 07 '20 at 09:38
  • @OrShahar I really thank you for replying. So, as you have mentioned in the hint, if I want to do the same but with proof by contradiction. So I take $K = (1,2)$, then I choose x = 2, (i.e a boundary), and use the fact that K is compact, then where will the argument fail? For any point $x \in K^{C}$, all I get is $K^{C}$ having both interior and boundary points. – Prashanth Dec 08 '20 at 01:58
  • 1
    @Prashanth If I'm right' you ask what will fail if we let $K=(1,2)$ be a compact set. The proof fails when you take an open neighbourhood $V$ of $2\in V \subseteq K^c$, which does not exist because $2$ is a boundary point of $K=(1,2)$, which means $\forall U$ such that $U$ is open and $2\in U$, $U\cap K \neq \phi$ (Definition). Let me know if its answer your question. – Or Shahar Dec 08 '20 at 11:04
  • 1
    @OrShahar oh ok, now I see it, I was trying to take open neighborhoods deliberately avoiding 2 in it. Thanks a lot, yes this answered my question. – Prashanth Dec 09 '20 at 00:52