3
$\begingroup$

Suppose $X$ is a normal (Hausdorff) zero dimensional topological space and $C\subset X$ is a closed set. Let $f:C\rightarrow \mathbb N$ be continuous. By the Tietze extension theorem, there exists a continuous $g:X\rightarrow \mathbb R$ such that $f\subset g$. I want to know if we can guarantee that there exists a continuous $g:X\rightarrow \mathbb N$ such that $f\subset g$ (even if f is not bounded).

I know that in general it's not true since that would imply that every normal topological space that has at least two points is not connected, therefore I added the zero dimensional hypothesis.

Edit: If it's not true, can I fix it by adding an extra hypothesis?

$\endgroup$
3
$\begingroup$

According to this paper, theorem 6, it is true for so called ultranormal spaces: spaces where any two disjoint closed sets are separated by a clopen set. Then we can even take any complete metric space as the codomain, including $\mathbb{N}$. This condition is clearly necessary (for any two disjoint closed non-empty sets extend the map that sends one closed set to $0$ and the other to $1$, which is a continuous map into $\mathbb{N}$ and take the inverse image of $\{0\}$). Not all zero-dimensional normal spaces are ultranormal (Roy has constructed a classical counterexample, to which the paper also refers.)

$\endgroup$

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.