1
$\begingroup$

Given a nondiscrete compact Hausdorff space $K$, does there always exist a real-valued function $f$ on $K$ that is not locally constant? Why/why not?

In http://arxiv.org/abs/math/9505204 the authors show that there are compact Hausdorff spaces $K$ such that all $f \in C(K)$ are locally constant on a dense subset (example: $\beta \mathbb{N} \setminus \mathbb{N}$). However, they do not mention anything about functions that are locally constant everywhere.

$\endgroup$
2
  • $\begingroup$ Formally, not: if $K$ is discrete, any function is locally constant. $\endgroup$ Commented Feb 3, 2014 at 16:50
  • $\begingroup$ Added nondiscrete to the question, thanks. $\endgroup$ Commented Feb 3, 2014 at 16:58

2 Answers 2

5
$\begingroup$

A $P$-space is a completely regular space where the countable intersection of open sets is open. It is well known and easy to prove that that a completely regular space is a $P$-space if and only if every continuous function is locally constant. In fact, there are several other necessary and sufficient conditions for whether a space is a $P$-space and these conditions can be found in the book Rings of Continuous Functions by Gillman and Jerison. The only compact $P$-spaces are the finite spaces: First take note that every $P$-space is zero-dimensional(i.e. it has a basis of clopen sets). If $X$ is an infinite $P$-space, and $A\subseteq X$ is a countable subset, then for each $a,b\in A$ with $a\neq b$, let $\{C_{a,b,1},C_{a,b,2}\}$ be a partition of $X$ into two clopen sets such that $a\in C_{a,b,1},b\in C_{a,b,2}$. Let $P=\bigcap_{a,b\in P}\{C_{a,b,f(a,b)}|f:A^{2}\setminus 1_{A}\rightarrow\{1,2\}\}.$ Then $P$ is a partition of $X$ into infinitely many clopen sets. Therefore $X$ cannot be compact.

$\endgroup$
1
-1
$\begingroup$

Since $K$ is infinite, you can find a sequence $(x_n)$ in $K$ converging to a point $x$. There exists a function $f_n: K\rightarrow [0,1]$ with $f(x)=0$, $f(x_n)\neq 0$ (compactness is not necessary, you just need that functions separate points: uniformizable is enough). Take $f=\sum 2^{-n}f_n$ .

$\endgroup$
7
  • 1
    $\begingroup$ How can you find a converging sequence? Don't you need some countability axiom for that? $\endgroup$ Commented Feb 3, 2014 at 17:29
  • $\begingroup$ Compactness is a necessary condition. $\endgroup$ Commented Feb 3, 2014 at 17:36
  • $\begingroup$ What I meant is that you don't need compactness to separate points. But you need it to find a convergent sequence -- in a compact (Hausdorff) space any infinite set contains a convergent subsequence. $\endgroup$
    – abx
    Commented Feb 3, 2014 at 17:48
  • $\begingroup$ @abx: No, it doesn't. Any net has a convergent subnet, but that subnet need not necessarily be a sequence. $\endgroup$ Commented Feb 3, 2014 at 17:58
  • 1
    $\begingroup$ A slight variant of this argument does work: take $(x_n)$ to be any countable discrete subset, and instead of a single point $x$ take the set of all accumulation points. $\endgroup$ Commented Feb 3, 2014 at 18:51

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .