1

Let $X$ be a Locally compact Hausdorff space and $Y$ is a compact subspace of $X$. Let $\phi$ be a state on $C(Y)$. Then can we extend to $\phi$ to $C_0(X)$? Suppose if $T:f \mapsto f|_X$ is the homomorphism from $C_0(X)$ to $C(Y)$, then will the map $\tilde{\phi}=\phi\circ T$ be a state on $C_0(X)$?

budi
  • 1,780
  • I think you mean $T: f \mapsto f_Y$. Also, clearly $\phi \circ T$ will be positive (its a composition of positive maps), so it suffices to consider whether or not this $\tilde{\phi}$ functional has norm 1. – PStheman Aug 24 '20 at 20:41
  • Yes. That is a typo. It is $f|_Y$ – budi Aug 25 '20 at 04:53

1 Answers1

1

Okay, so I think this is true. As per my comment, we just need to check that this functional has norm 1. Positivity is in turn equivalent to $\lim_\lambda \tilde{\phi}(e_\lambda) = \|\tilde{\phi}\|$ for some (or any) approximate unit $(e_\lambda)$ for $C_0(X)$. So to solve this, lets find an approximate unit $(e_\lambda)$ which satisfies $\lim_\lambda \tilde{\phi}(e_\lambda) = 1$.

Just as in the wikipedia article for C*-algebras (https://en.wikipedia.org/wiki/C*-algebra#Commutative_C*-algebras), there is an approximate unit $(f_K)$, indexed by compact subsets $K \subseteq X$ for which $f_K|K = 1$ (Tietze extension/link in the comments). With this idea in mind, its not hard to construct a net $(f_K)$, indexed by compact subsets $K \subseteq X$ which contain $Y$, such that $f_K|_K = 1$. This is our desired approximate unit: $$ \lim_K \tilde{\phi}(f_K) = \lim_K \phi \circ T(f_K) = \lim_K \phi(f_K|_Y) = \lim_K 1 = 1. $$

PStheman
  • 1,581
  • It'd also be interesting to consider this question in further generality: given $T: A \to B$ a surjective $$-homomorphism between C$$-algebras and state $\phi$ on $B$, when can we say that $\phi \circ T$ is a state on $A$? Would love to have an answer to that. I suppose I'll ask it if its not answered here. Edit: an answer where $B$ is unital and $A$ is not would be interesting (a unital *-hom will obviously give that $\phi \circ T$ is a state). – PStheman Aug 24 '20 at 21:22
  • If I'm not wrong, to use Tietze you need regular, which is not true for all locally compact spaces. – Martin Argerami Aug 24 '20 at 23:02
  • @MartinArgerami it seems normality of the space is whats necessary in Tietze's theorem (according to Munkres' "Topology"). Also it seems that LCH spaces are regular (Exercise 3 of chapter 32 in Munkres). Nonetheless I've done a search and found this link (https://math.stackexchange.com/questions/1188694/tietze-extension-theorem-in-lch-spaces) for extending functions from compact subsets to the whole space. – PStheman Aug 24 '20 at 23:26
  • Yes, the proof I know of Tietze works fine in general for compact subset of locally compact Hausdorff. – Martin Argerami Aug 25 '20 at 22:31