3
$\begingroup$

This may be a standard question answered in a book, or article. I don't know. I know that there exist related results with $\lim^1$-sequences (Rosenberg and Schochet).

What is

$KK(C_0(X),\mathbb{C})$

for $X =$ Cantor set?

What is it more generally when $C_0(X)$ is an abelian AF-algebra?

My only question is, are these homology groups mostly UNCOUNTABLE?

Note that, for example,

$KK(\oplus_{n=1}^\infty \mathbb{C} , \mathbb{C}) = \prod_{n=1}^\infty \mathbb{Z}$

is uncountable.

$\endgroup$
5
  • $\begingroup$ Have you got access to Higson and Roe's book? I haven't got my copy to hand, but that is the first place I would look. $\endgroup$
    – Yemon Choi
    Commented Feb 2, 2015 at 22:17
  • $\begingroup$ I know almost nothing about KK-theory, but it seems to me that uncountability would follow from the fact that every infinite metrizable Stone space has $\mathbb{N}\cup\{\infty\}$ as a retract. $\endgroup$ Commented Feb 2, 2015 at 23:59
  • $\begingroup$ Isn't it just the dual group of K$_0 (A)$ (here $A = C(X)$, but could be any C*-algebra)? That is, Hom$(K_0(A),Z)$ (since K$_0 C = Z$). For $X$ a Cantor set, this is uncountable. $\endgroup$ Commented Feb 3, 2015 at 3:43
  • $\begingroup$ And because $K_0(A)$ is free on countably many generators (if $A = C(X)$ where $X$ is Cantor), its dual is the product. $\endgroup$ Commented Feb 3, 2015 at 3:46
  • $\begingroup$ @David: Maybe you could explain this with dual = $Hom(K_0(A),Z)$. When is it so and why (ok one explanation is the main answer). $\endgroup$
    – hans
    Commented Feb 3, 2015 at 21:04

1 Answer 1

2
$\begingroup$

As David Handelman already pointed out, by continuity of $K$-theory we obtain that $K_0(C(X)) \cong \bigoplus_{\mathbb{N}} \mathbb{Z}$ and $K_1(C(X)) = 0$. Since $C(X)$ is commutative, it lies in the Bootstrap class of $KK$-theory (see Blackadar, "Operator algebras", section V.1.5.2, p. 414). In particular, the universal coefficient theorem holds for $C(X)$ (see V.1.5.8, p. 416 in Blackadar), i.e. $$ 0 \to Ext^1_{\mathbb{Z}}(K_*(C(X)), K_*(\mathbb{C})) \to KK_*(C(X),\mathbb{C}) \to Hom(K_*(C(X)),K_*(\mathbb{C})) \to 0 $$ is exact and the map $\delta$ from Ext to KK has degree $1$, the map $\gamma$ from KK to Hom has degree $0$. Since $K_*(C(X))$ is free, we obtain that $\gamma$ is an isomorphism. This yields $$ KK(C(X),\mathbb{C}) = KK_0(C(X),\mathbb{C}) \cong Hom(K_0(C(X)), \mathbb{Z}) = \prod_{\mathbb{N}} \mathbb{Z} , $$ which is uncountable.

If $A$ is a commutative AF-algebra, then we have $K_1(A) = 0$ and $A$ also lies in the Bootstrap class (because it is commutative), so we still obtain an isomorphism $$ KK_0(A,\mathbb{C}) \cong Hom(K_0(A), \mathbb{Z}) $$ in this case. Since $K_0(A)$ is the direct limit over countable abelian groups, it should itself be countable again (feel free to correct me if I am wrong here). Therefore $KK(A,\mathbb{C})$ should be at most uncountable.

$\endgroup$
4
  • $\begingroup$ Bitte, nur ein N am hinten. $\endgroup$ Commented Feb 3, 2015 at 14:36
  • 2
    $\begingroup$ @DavidHandelman: Whoops. I am sorry. Unfortunately, writing "man" with double N is a reflex you easily develop in Germany. $\endgroup$ Commented Feb 3, 2015 at 14:55
  • $\begingroup$ @Ulrich: Very good answer, thank you very much! Thanks also to the others who commented on this. $\endgroup$
    – hans
    Commented Feb 3, 2015 at 21:02
  • $\begingroup$ PS: my question was: are most cases of $C_0(X)$ uncountable? (not: at most uncountable). $\endgroup$
    – hans
    Commented Feb 3, 2015 at 21:08

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .