Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

We know the Stone-Weierstrass theorem for locally compact Hausdorff spaces (LCH) which states the following:

Theorem: Suppose $X$ is LCH. A subalgebra $\mathcal{A}$ of $C_0(X)$ is dense if and only if it separates points ($\forall x,y \in X : x\neq y \implies \exists f \in \mathcal{A}: f(x) \neq f(y)$) and vanishes nowhere $(\forall x \in X \exists f \in \mathcal{A} : f(x) \neq 0$) and is closed under complex conjugation1.

It's also easy to show that $C_b(X)$, the continuous and bounded functions on a topological space $X$, is a Banach space2. It would be obvious to try and show a variant of Stone-Weierstrass for $C_b(X)$ where $X$ is merely Hausdorff, and since it's obviously dense in itself, for such a theorem to exist it would be required that $C_b(X)$ separates points (that it vanishes nowhere is clear: it contains the constant functions).

I've tried to come up with an example of a Hausdorff space where $C_b(X)$ fails to separate points (this could for example be a space where every non-constant continuous function is unbounded), but to no avail. So does anyone happen to have an instructive example lying around?

PS: It's an interesting exercise to prove that given a LCH space $X$ the continuous functions with compact support, $C_{00}(X)$, separates points and vanishes nowhere. You get to use many theorems from topology in the process of constructing a continuous function such that $f(x)=1$ and $f(y)=0$ given distinct points $x,y \in X$ [Hint: Find a good compact subset and use Urysohn's lemma].

[1]: In the case of real-valued functions this is essentially a no-op, so including it in the theorem statement doesn't hurt.

[2]: [Car00] shows in Lemma 10.8 that $B(X)$, the set of bounded functions on a set $X$ is a Banach space, and it's an immediate corollary from Thm 10.4 that $C_b(X)$ is closed in $B(X)$ for $X$ a metric space. Of course this generalises readily to the case of $X$ a topological space.

share|cite|improve this question
It's always a good idea to consult Steen and Seebach's Counterexamples in topology for such questions. I haven't checked the details, but should provide an example (see the last paragraph). – t.b. Feb 27 '11 at 13:53
up vote 10 down vote accepted

Looking at the table at the back of Counterexamples in Topology, the first space which is Hausdorff but not Urysohn (which in that book means $C(X)$ doesn't separate points) is the relatively prime integer topology. That is, the topology on $\mathbb{N}\setminus \{0\}$ with basis all sets of the form $\{b+na|n \in \mathbb{N}\}$ for $a$, $b$ coprime.

Hausdorff is easy: if $k, l \in \mathbb{N}\setminus \{0\}$, take a prime $p$ larger than both and consider the open neighbourhoods $\{k+np \}, \{l+np \}$.

$C(X)$ fails to separate points because any two basic open sets have nondisjoint closures, since the closure of $\{b+na \}$ contains all multiplies of $a$.

share|cite|improve this answer

I'll just point out that boundedness is not the obstruction: any space that admits a nonconstant real-valued continuous function admits one which is bounded, and if $C(X,\mathbb{R})$ separates points then so does $C_b(X,\mathbb{R})$. This is easy to see: if $f$ is continuous and $f(x_1)=a < b = f(x_2)$, then $g(x) = (f(x) \vee a) \wedge b$ is continuous and bounded and also has $g(x_1)=a<b=g(x_2)$.

However, continuous functions with compact support may not be enough if $X$ is not locally compact, even if it is metrizable. For instance, it's a nice exercise to check that if $X$ is an infinite-dimensional Banach space, the only real-valued continuous function on $X$ with compact support is the zero function, i.e. in your notation $C_{00}(X) = \{0\}$. (I think most authors call this space either $C_0(X)$ or $C_c(X)$.)

share|cite|improve this answer
I've only ever seen $C_c(X)$ (most textbooks around here) and $C_{00}(X)$ (I think T. Tao uses this on his mathematical blog - I used it because I figured it would be familiar for most readers) for the compactly supported functions on $X$. What really puzzles me is writing $C_0(X)$ for it… what would one then use for the functions that vanish at infinity? – kahen Feb 28 '11 at 1:41
Many papers don't need to refer to the functions that vanish at infinity. Alternatively, some use $C_\infty(X)$ for this. – Nate Eldredge May 7 '11 at 12:56

I recommend this short 1971 note of T.E. Gantner. He first proves:

Theorem: Each regular space $Z$ can be embedded as a subspace of a regular space $Q(Z)$ such that every continuous, real-valued function on $Q(Z)$ is constant on $Z$.

He applies the theorem as follows: take $X_0$ to be a one-point space, and inductively, for all $n \in \mathbb{N}$, define $X_{n+1} = Q(X_n)$. Let $X = \lim_{n \rightarrow \infty} X_n$, endowed with the direct limit (or final) topology. Then $X$ is an infinite regular space on which every continuous real-valued function is constant.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.