(All spaces are Hausdorff.)

This question is a variant of my previous question.

Let $X$ be a completely regular space, that is, for every closed set $F\subseteq X$ and $x\not\in F,$ there exists a continuous function $g:X\to [0,1]$ such that $g(F) = \{0\}$ and $g(x) =1.$

Question: For every closed $G_\delta$ set $F\subseteq X,$ does there exist a continuous function $f:X\to [0,1]$ such that $f=0$ on $F$ and $f\neq 0$ outside $F?$

A subset $U\subseteq X$ is a zero set if there exists a continuous function $g:X\to [0,1]$ such that $g^{-1}(\{0\}) = U.$

It is well-known that if $X$ is normal then all closed $G_\delta$ sets are zero sets.

However, I am not sure whether the same holds for completely regular space.

  • $\begingroup$ I know that $X$ normal implies that all closed $G_\delta$'s are zero-sets, but I believe the reverse does not hold. If it did your question would be useless. Do you have a reference ? "well-known" it is not, it's not mentioned in Engelking, a standard reference for such matters. $\endgroup$ – Henno Brandsma Jan 5 '19 at 14:26
  • 2
    $\begingroup$ Note that this can only be true when $F$ is closed and $G_\delta$. You cannot ask this for all closed sets. $\endgroup$ – Henno Brandsma Jan 5 '19 at 14:27
  • $\begingroup$ @HennoBrandsma yes, you are right. I have edited my post. $\endgroup$ – Idonknow Jan 5 '19 at 14:59
  • $\begingroup$ @HennoBrandsma by the way, if the converse holds, why would by question be useless? $\endgroup$ – Idonknow Jan 5 '19 at 15:00
  • 1
    $\begingroup$ @PaulFrost that is the direction I agreed with, yes. The OP originally claimed that if all closed $G_\delta$ sets are zero-sets then $X$ is normal. This I doubt the truth of, as said. $\endgroup$ – Henno Brandsma Jan 5 '19 at 18:46

There is a well-known counterexample. Let $S$ be the Sorgenfrey line, that is the real line endowed with the Sorgenfrey topology (generated by the base consisting of half-intervals $[a,b)$, $a<b$). It is well-known that a product $X=S\times S$ is a Tychonoff but non-normal space (see, for instance, Examples 1.4.4 and 2.3.12 in [Eng]). Let $D=\{(x,-x)\in S\times S: x\in S\}$. Then any subset $Y$ of $D$ is closed in $X$, and we have $2^{|D|}=2^{\frak c}$ such subsets. Moreover, since $Y=\bigcap_{n\in\Bbb N} Y+[0,1/n)\times [0,1/n)$, $Y$ is a $G_\delta$-subset of $X$. On the other hand, let $C$ be a countable dense subset of $X$. By Theorem 2.1.9 in [Eng], each continuous real-valued function $f$ on $X$ is uniquely determined by its restriction on $C$, so there are at most $\frak c^\omega=\frak c$ such functions. Thus most subsets of $D$ are not zero-sets.


[Eng] Ryszard Engelking, General Topology, 2nd ed., Heldermann, Berlin, 1989.


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.