1
$\begingroup$

Suppose that $X$ is a locally compact topological space. Let $M(X)$ denote the Banach space of regular Borel measures on $X$. It is known that the bidual of $C_0(X)$ is a commutative $C^*-$algebra. Denote the spectrum of $C_0(X)^{**}$ by $\tilde{X}$.

Since $C_0(X)^{**}$ is a unital $C^*-$algebra it can be shown that $C(X^{\infty})$ is a subspace of $C_0(X)^{**}$, where $X^{\infty}$ is the one-point compactification for $X$. Now for each $f\in C_0(X)$ we have that $f(\infty)=0$.

Let $x\in\tilde{X}$ and consider the point mass $\delta_x$. Then $\delta_x|_{C_0(X)}$ is a multiplicative linear functional on $C_0(X)$ and therefore it belongs to $X^{\infty}$. Let $\pi(x)$ be the corresponding element in $X^{\infty}$. Therefore we can define a mapping $\pi:\tilde{X}\to X^{\infty}$.

If $x\in X$ is an isolated point then it can be shown that $\pi^{-1}(x)=\{x\}$. What about the converse, if $\pi^{-1}(x)=\{x\}$ can we conclude that $x$ is an isolated point of $X$?

$\endgroup$

1 Answer 1

3
$\begingroup$

Yes, you can conclude this. Let ${\rm Bor}(X)$ be the set of bounded Borel measurable scalar functions on $X$, equipped with sup norm. Then ${\rm Bor}(X)$ embeds isometrically as a subspace of $C_0(X)^{**}$. Now let $x$ be any non-isolated point of $X$ and find a net $(x_\lambda)$ in $X$ which converges to $x$, such that $x_\lambda \neq x$ for all $\lambda$. Now each $\delta_{x_\lambda}$ belongs to $M(X)$, so by the natural embedding of $M(X)$ in $M(X)^{**}$ it gives rise to an element $\tilde{x}_\lambda$ of $\tilde{X}$. Finally, since $\tilde{X}$ is compact we can find a cluster point $y$ of the net $(\tilde{x}_\lambda)$ in $\tilde{X}$. Then $\pi(y) = \pi(\tilde{x}) = x$, but $y \neq \tilde{x}$; we know this because the function $\chi_{\{x\}} \in {\rm Bor}(X)$ which is $1$ at $x$ and $0$ elsewhere, when realized as an element of $C_0(X)^{**} \cong C(\tilde{X})$, evaluates to $1$ at $\tilde{x}$ and to $0$ at $y$.

$\endgroup$
2
  • $\begingroup$ Very nice! Thank you very much Nik! $\endgroup$
    – Bob
    Feb 26, 2014 at 18:07
  • $\begingroup$ Sure, no problem. $\endgroup$
    – Nik Weaver
    Feb 26, 2014 at 19:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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