Let $E$ be a compact seperable space, for example $E=[0,1]\subset\mathbb{R}$. Denote by $$\ell_1(E):=\{u:=(u_x\in\mathbb{C}:x\in F):F\text{ is a countable subset of }E \text{ and } \|u\|_{\ell_1(E)}:=\sum_{x\in F}|u_x|<\infty\}$$ the Banach space of functions on $E$ that is integrable with respect to the counting measure on $E$. For any $u=(u_1,u_2,\ldots)\in \ell_1(E)$, suppose the corresponding countable subset $F$ is $\{x_1,x_2,\ldots\}\subset E$, then $u$ corresponds to a finite signed Baire measure $$\mu_u(S)=\sum\limits_{x_i\in S\cap F}u_i$$ for any subset $S$ of $E$. With the definition of total variation, we have $$\|\mu_u\|_{TV}=\sum\limits_{i=1}^\infty |u_i|\quad (=\|u\|_{\ell_1(E)})$$ And for any measurable function $f$ on $E$, the integration will be $$\int_E fd\mu_u=\sum\limits_{i=1}^\infty f(x_i)\cdot u_i.$$ By Riesz-Markov-Kakutani Theorem, we know $\mu_u$ is an element of dual space $M(E)$ of continous function space $C(E)$. It seems that we can embed $\ell_1(E)$ into $M(E)$. Is this correct? Is the space $\ell_1(E)$ separable? And do we have some reference of the space $\ell_1(E)$? Thanks a lot!
$\begingroup$
$\endgroup$
4
-
5$\begingroup$ Yes, all this is correct. You don't need Riesz-Markov-Kakutani, you can easily check from first principles that $f \mapsto \sum_i f(x_i) u_i$ is a bounded linear functional on $C(E)$. $\ell^1(E)$ is definitely not separable unless $E$ is countable; this is also easy to check (the point masses are an uncountable set at pairwise distance 1). I don't think there are many references on this space per se as there isn't a lot to be said about it. $\endgroup$– Nate EldredgeCommented Jan 29, 2021 at 18:00
-
2$\begingroup$ Just to follow up on @NateEldredge's answer: it is also possible to show directly that the mapping he defines from $\ell^1(E)\to C(E)^*$ is not just contractive but isometric. (Urysohn's lemma ought to do the job, but I have not thought through all the details) $\endgroup$– Yemon ChoiCommented Jan 29, 2021 at 18:11
-
3$\begingroup$ Just an aside to @Nate's answer: the point masses are at pairwise distance 2. $\endgroup$– Dirk WernerCommented Jan 29, 2021 at 18:15
-
$\begingroup$ A term to look for is maybe "discrete measure", see en.m.wikipedia.org/wiki/Atom_(measure_theory)#Discrete_measures $\endgroup$– Matthew DawsCommented Jan 29, 2021 at 22:31
Add a comment
|