Does "regular" implies collectionwise hausdorff?

A topological space is said to be collectionwise Hausdorff if given any closed discrete collection of points in the topological space, there are pairwise disjoint open sets containing the points.

I believe it and try to proof it, however I'm not sure. Thanks in advance:)


No, it doesn't. The Sorgenfrey plane is regular, as the Sorgenfrey line is normal, but it isn't collectionwise Hausdorff as $\{(-x,x)\mid x \in \mathbb R\}$ is discrete and closed, but there aren't pairwise disjoint open neighbourhoods.

To prove this, let $U_x = [-x,-x+\epsilon_x) \times [x,x+\epsilon_x)$ be a neighbourhood of $(-x,x)$ in the Sorgenfrey plane, then there is some $\epsilon > 0$ such that $E_\epsilon= \{x \in \mathbb R\mid \epsilon_x > \epsilon \}$ is uncountable (as $\mathbb R = \bigcup_n E_{1/n}$). Then $E_\epsilon$ has an accumulation point in $\mathbb R$, choose $x_1, x_2 \in E_\epsilon$ with $|x_1 - x_2| < \frac \epsilon 4$, then $U_{x_1} \cap U_{x_2}\ne \emptyset$.

To answer your addional question in the comment below: Let $X$ be regular and $A =\{a_n \mid n \in \mathbb N\}$ closed, discrete and countable. For each $n \in \mathbb N$ there are by regularity and closedness of all subsets of $A$ in $X$ disjoint open sets $U_n \ni x_n$, $V_n \supseteq A \setminus \{x_n\}$. Now let $W_n = U_n \cap \bigcap_{i < n} V_i$, then $W_n$ is an open set containing $x_n$ (as finite intersection of such). Now let $n < m$, then $W_n \subseteq U_n$, $W_m \subseteq V_n$, so $W_n \cap W_m = \emptyset$. Hence $X$ is (don't know if this is a common term, but IMHO it describes the property) countable collectionwise Hausdorff.

| cite | improve this answer | |
  • $\begingroup$ I see; to prove it by the baire property of the $R$. However, if for countable closed discrete subspace, it is true? $\endgroup$ – Paul Jul 4 '12 at 12:59
  • $\begingroup$ Hm ... have to think about it ... $\endgroup$ – martini Jul 4 '12 at 13:10
  • $\begingroup$ @John: Added sth. above, does this work? $\endgroup$ – martini Jul 4 '12 at 14:59
  • $\begingroup$ @martini: For the added bit you can also construct the $W_n$ recursively. Choose a nbhd $W_0$ of $a_0$ whose closure is disjoint from $\{a_n:n>0\}$. Given $W_k$ for $k<n$, choose a nbhd $W_n$ of $a_n$ whose closure is disjoint from $\bigcup_{k<n}\operatorname{cl}W_k\cup\{a_k:k>n\}$. $\endgroup$ – Brian M. Scott Jul 4 '12 at 17:43
  • $\begingroup$ @martini, it really help me:) $\endgroup$ – Paul Jul 5 '12 at 0:31

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.