2
$\begingroup$

Let $A$ and $B$ be $C^*$-algebras. Let $A_0$ be a dense *-subalgebra of $A$, and let $B_0$ a dense *-subalgebra of $B$. Assume that $A_0$ is isomorphic to $B_0$.

Finally, assume that $K_0(A)$ is finitely generated. Must $K_0(B)$ also be finitely generated?

The isomorphism between $A_0$ and $B_0$ is an algebraic isomorphism between algebras over the field of complex numbers; it need not be a $*$-isomorphism.

$\endgroup$
5
  • 3
    $\begingroup$ If $A_0$ and $B_0$ are not necessarily isomorphic as $*$-algebras, then in what category are they assumed isomorphic? $\endgroup$
    – LSpice
    Mar 3, 2018 at 18:28
  • 3
    $\begingroup$ How about the following. Take $A_0 = B_0 = {\bf C}G$ (the group algebra on the group $G$), and suppose that $G$ is not amenable. Then the maximum and minimum norms yield non-isomorphic C*-algebras $A$ and $B$ respectively. I am not that familiar with the horde of results in this area, but there presumably are examples where one of them has merely ${\bf Z}$ as its $K_0$, and the other one has lots of projection-equivalence classes, enough to guarantee non-finite generation. $\endgroup$ Mar 3, 2018 at 21:28
  • $\begingroup$ Of course what I meant by maximum and minimum norms were the full and regular representation norms, but that occurred to me just more than five minutes after ... $\endgroup$ Mar 3, 2018 at 21:34
  • $\begingroup$ It's certainly your right to rollback edits to the question, but why insist on $C*$ in place of $C^*$, and why the '---'s? $\endgroup$
    – LSpice
    Mar 3, 2018 at 21:57
  • 3
    $\begingroup$ When a dense subalgebra is invariant under analytic functional calculus, its K-theory is isomorphic to that of the ambient algebra, so I suggest you check whether that property holds on the examples you are interested. $\endgroup$
    – Ruy
    Mar 8, 2018 at 0:56

2 Answers 2

3
$\begingroup$

No. To obtain a counterexample, you just need a C*-algebra $A$ with finitely generated K-theory and a quotient $A/I$ of $A$ which does not have finitely generated K-theory and the quotient algebraically preserves some dense subalgebra of $A$. Here is the easiest example of this situation that I could think of.

Let $A=C[0,1]$ and let $B$ be the C*-algebra of convergent sequences. Then $K_0(A)$ is finitely generated while $K_0(B)$ is not. Define the surjective map $\pi:A\rightarrow B$ by $\pi(f)(n)=f(1/n).$ Let $A_0\subseteq A$ be polynomials and let $B_0=\pi(A_0),$ which is a dense subalgebra of $B.$ Since $\pi$ is injective on $A_0$ it defines a *-isomorphism between $A_0$ and $B_0.$

$\endgroup$
0
$\begingroup$

We know that $K_0(A) \neq K_0(B)$ in general. Take for example, what Handelman pointed out, the full and reduced group $C^*$-algebras of a group $G$ which is not $K$-amenable, i.e. $K_0(C^* G) \neq K_0(C^*_r G)$.

It is fair to assume that we may find an example where $K_0(A)=0$ but $K_0(B) \neq 0$.

If you accept this, then you immediately have a counterexample.

Take $a_0:=\bigoplus_{\mathbb{N}} A_0$ with two distinct closures $a:=\bigoplus A$ ($C^*$-direct inifinite sum of $A$) and similarly $b:=\bigoplus B$.

Then $a_0$ is dense in $a$ and $b$.

But $K_0(a) = \bigoplus K_0(A) =0$ is finitely generated, and $K_0(b)= \bigoplus K_0(B)$ is infinitely generated.

That is why I am confident that your question is answered to the negative.

$\endgroup$

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.