1
$\begingroup$

Consider the topology $\tau_1$, the standard topology on $\mathbb R$, and $\tau_2$, the cocountable topology (here we will consider a set open with respect to $\tau_2$ if its complement is either finite or countable). We define $\tau=\tau_1\lor \tau_2$ as the collection of subsets $W\subseteq \mathbb R$ such that for every $x \in W$, there exist subsets $U,V \subseteq \mathbb R$ with $x \in U$, $x \in V$, $U \in \tau_1$, $V \in \tau_2$, and $U \cap V \subseteq W$. Show that $(\mathbb R, \tau)$ is not regular.

I know that a regular topological space is defined as a $T_1$ topological space such that for every $x \in \mathbb R$ and closed set $E \subseteq \mathbb R$ with $x \notin E$, there are disjoint open sets $U,V \subseteq \mathbb R$ such that $x \in U$ and $E \subseteq V$. So, I want to show that there exists some $p \in \mathbb R$ and a closed set $E \subseteq \mathbb R$ such that $p \notin E$ and $U$ and $V$ are not disjoint for every $U,V \subseteq \mathbb R$ with $p \in U$ and $E \subseteq V$.

I'm not really sure how to set this proof up. I have a proposition that says a regular topological space is Hausdorff, so one idea I had was to show that $(\mathbb R, \tau)$ is not Hausdorff, but I'm not sure if it would be better to attempt a direct proof or not. For a direct proof, I want to consider a set $E$ which is countable (or finite). Then, $E$ will be closed with respect to $\tau$. I'm not sure how to find a $p$ to fit my needs, though.

$\endgroup$
  • 1
    $\begingroup$ $(X,\tau)$ is Hausdorff as $\tau_1 \subseteq \tau$, so that route is out. $\endgroup$ – Henno Brandsma Apr 15 '18 at 21:17
1
$\begingroup$

Note that $\mathbb{Q}$ is closed in $\tau$, as $\mathbb{Q}$ is countable and so $\mathbb{R} \setminus \mathbb{Q} \in \tau_2 \subseteq \tau$ and so its complement is closed.

Also, $\sqrt{2} \notin \mathbb{Q}$. Suppose we have $U$ and $V$ in $\tau$ such that $\sqrt{2} \in U$, $\mathbb{Q} \subseteq V$.

By definition of $\tau$ we can find a Euclidean open set $O_1$ and an at most countable set $N_1 \subseteq \mathbb{R}$ such that $$\sqrt{2} \in O_1 \setminus N_1 \subseteq U$$

We can find a rational number $q \in O_1$, as $\mathbb{Q}$ is dense in the usual topology on the reals.

As $V$ is open and $q \in V$ we again by definition of $\tau$ find a Euclidean open set $O_2$ and an at most countable set $N_2 \subseteq \mathbb{R}$ such that $$q \in O_2 \setminus N_2 \subseteq V$$

But then note that $O_1 \cap O_2$ is a non-empty Euclidean subset of $\mathbb{R}$ (it contains $q$ e.g.) and so it contains uncountably many irrational numbers. We can thus find an irrational $p$ such that

$$p \in O_1 \cap O_2 , p \notin N_1 \cup N_2$$

And we can immediately check that $p \in U \cap V \neq \emptyset$. This implies that $(X,\tau)$ is not regular.

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.