1
$\begingroup$

Is a topological space $(\mathbb{R}, \tau)$ normal, irrespective of the nature of $\tau$?

For convenience I post the definition of normality:

Definition A topological space $(X, \tau)$ is said to be normal, if for closed sets $C$, $D \subseteq X$, if $C \cap D = \emptyset$, then there exist open sets $U$ and $V$ such that $C \subseteq U$, $D \subseteq V$ and $U \cap V = \emptyset$

Using the definition above it seems to be the case for me that if disjoint subsets $C$ and $D$ of $\mathbb{R}$ can be found, a proof similar to this one can be given. I quote from the chosen answer of that post:

Every metric space is normal, in particular $\mathbb R^n$. The proof goes as follows:

For each $a\in A$, let $r_a=\frac{1}{3}d(a,B)$, and for each $b\in B$, let $s_b=\frac{1}{3}d(b,A)$. Now define $U=\bigcup_{a\in A}B(a,r_a)$ and $V=\bigcup_{b\in B}B(b,s_b)$. It is not hard to show that $A\subseteq U,$ $B\subseteq V,$ and $U\cap V=\varnothing$.

If no two such sets can be found, the statement would be vacuously true like in this case.

Am I missing anything crucial?

$\endgroup$
6
  • 3
    $\begingroup$ The topology has to matter, or it would be a property of sets not spaces. Give $\mathbb{R}$ the topology with open sets $\varnothing$ and any set that contains $1$. $\endgroup$
    – Randall
    Commented Jun 21, 2021 at 20:46
  • $\begingroup$ The fact that $\mathbb{R}$ is a metric space already answers your question, right? Indeed, $\mathbb{R}$ is normal, irrespective of $\tau$. $\endgroup$
    – jasnee
    Commented Jun 21, 2021 at 20:50
  • 1
    $\begingroup$ @jasnee but it is only metric under a metrizable topology. There are others. $\endgroup$
    – Randall
    Commented Jun 21, 2021 at 20:51
  • $\begingroup$ Yes, I can see, how the quoted proof works with the usual topology on $\mathbb{R}$, @Randall. I just don't see how any other fact than the sets $C$ and $D$ being closed and disjoint are used in that proof. And if no such sets can be found, the premise of the definition of normality is not satisfied and so the statement of normality as a whole would be true. Wouldn't it? $\endgroup$ Commented Jun 21, 2021 at 20:51
  • $\begingroup$ @greg. No. The example I gave has plenty of disjoint closed sets, but it still isn't normal. $\endgroup$
    – Randall
    Commented Jun 21, 2021 at 20:52

2 Answers 2

4
$\begingroup$

Normality is a topological property because its very definition relies on open and closed sets. It is not a property of bare sets. The topology of choice must matter.

Give $\mathbb{R}$ the topology with open sets $\varnothing$ and any subset $A$ of $\mathbb{R}$ with $1 \in A$. Check that this is indeed a topology on $\mathbb{R}$ and that a set $C$ is closed in this topology if and only if $1 \notin C$ or $C=\mathbb{R}$.

Now, consider the disjoint closed sets $C=\{4\}$ and $D=\{5\}$. There cannot exist open sets $U$ and $V$ surrounding $C$ and $D$ respectively with $U \cap V = \varnothing$ because both $U$ and $V$ must contain $1$ by definition.

$\endgroup$
2
  • $\begingroup$ Ah, great! This makes sense, thank you so much! $\endgroup$ Commented Jun 21, 2021 at 21:01
  • $\begingroup$ Happy to help. This kind of example and the cofinite topology together make really good sources for checking definitions and intuition, etc. I rely on them all the time. $\endgroup$
    – Randall
    Commented Jun 21, 2021 at 21:01
3
$\begingroup$

There are lots of non-normal topologies on $\mathbb{R}$, or indeed (almost) any set whatsoever. For example, the cofinite topology on an infinite set is never normal.

It's a good exercise to exactly determine which sets cannot be equipped with a non-normal topology. You can check by hand that all topologies on a set with at most two elements are normal; what if there are more than two elements in the set?

Suppose $X$ is a set with more than two elements. Let $a,b$ be distinct elements of $X$ and consider the topology $\tau$ generated by $\{X\setminus\{a\},X\setminus\{b\}\}$. Do you see why this is not normal?

$\endgroup$
1
  • $\begingroup$ Thanks. I didn't even think of the cofinite topology, but now that you mention it, it definitely makes sense :) $\endgroup$ Commented Jun 21, 2021 at 21:03

You must log in to answer this question.

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