I'm trying to solve the following problem:

Let $(X,T)$ be a second-countable, normal topological space, and let $U \subset X$ be an open set. Show that there exists a continuous function $f:X\rightarrow[0,1]$ such that $f(x) \neq 0$ iff $x \in U$.

Now, I know that in a metric space, every open set is $F_\sigma$. Does this holds also for a general second-countable space? I've got a hunch that the answer is positive, but I couldn't prove it.

If so, solving the problem would be easy, since $U=\bigcup_{i=1}^\infty F_i$, where each $F_i$ is closed, and so there exists a closed set $F$ such that $F \subset U \subset X$, and by applying Urysohn's lemma to $U^c$ and $F$ we are done.


Yes, this is true. Suppose $B_n, n \in \mathbb{N}$ is a countable base for $X$, and $U$ is open. Then for each $x \in U$ we pick by regularity some open $V_x$ such that $x \in V_x \subseteq \overline{V_x} \subseteq U$. Then pick some $n(x)$ with $x \in B_{n(x)} \subseteq V_x$. Then all $\cup \{\overline{B_{n(x)}}: x \in U \}$ is equal to $U$ (all $x$ in $U$ are in their own $B_{n(x)}$; all of these have closures inside $U$ by construction, so their union is exaclty $U$) and shows $U$ is an $F_\sigma$ (as we have countably many $n(x)$ (many $x$ will give the same $n(x)$, but that does not matter, we have countably many (at most) different closed sets).


By the Urysohn metrization theorem, a second-countable, normal space is metrizable, which solves your exercise.

  • $\begingroup$ I'm not familiar yet with this theorem, is there any way to circumvent it? $\endgroup$ – Brassican Jan 5 '16 at 17:52
  • $\begingroup$ Let, each open subset of a topological space X is normal. Does this necessarily mean the space is hereditary normal? $\endgroup$ – math is love Jan 26 '17 at 20:54

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.