3
$\begingroup$

Let $X$ be a compact Hausdorff space and $C(X)$ the Banach space of continuous $\mathbb K$-valued functions equipped with the supremum norm. We denote the dual space of $C(X)$ by $C(X)^*$.

A well-known theorem, due to Arens & Kelley (1947), states that the extreme points of the unit ball of $C(X)^*$ are precisely the functionals of the form$$\alpha= \theta\cdot\delta_x,$$ where $\theta$ is an unitary scalar and $\delta_x$ is the functional "evaluation at $x$", that is, $ \delta_x(f)=f(x), $ for every $f\in C(X)$.

I wonder if there is a version of this theorem for $C_0(X)$ spaces, that is, the Banach space of continuous $\mathbb K$-valued functions which vanish at infinity for some locally compact (not necessarily compact) Hausdorff space $X$.

Actually, I'm pretty sure I have seen people using this theorem for the $C_0(X)$ spaces, however I still fail to find any proof of this. I was able to prove, by myself, a version of this theorem for $C_0(X)$ spaces in the case $\mathbb K=\mathbb R$. I still have no answer for the case $\mathbb K=\mathbb C$.

$\endgroup$
  • $\begingroup$ Well, your hint seems fine to me. This way is so much better. The onliest proof I knew (the one at Dunford's book) didn't use the representation theorem. Instead it strongly depends on the compactness of $X$. $\endgroup$ – André Porto Dec 6 '17 at 18:59
1
$\begingroup$

Using a form of Riesz representation theorem, one can argue as follows. A unit-norm functional on $C_0(X)$ is represented by a complex measure $\mu$ of total variation $1$. If the support of $\mu$ consists of one point, it is a unimodular multiple of Dirac mass. Otherwise, $\operatorname{supp}\mu$ contains two distinct points $a\ne b$. Let $U,V$ be disjoint neighborhoods of $a$ and $b$. By the definition of support, $|\mu|(U)>0$ and $|\mu|(V)>0$. Let $t = |\mu|(U)$; by the above $0<t<1$. Define $$ \lambda = t^{-1} \mu_{|U}\quad \text{and} \quad \nu = (1-t)^{-1} \mu_{|U^c} $$ Then both $\lambda$ and $\nu$ represent unit-norm functionals, and $$ \mu = t \lambda + (1-t)\nu $$ proving that $\mu$ is not extreme.

$\endgroup$

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.