6
$\begingroup$

Let's say we have a $C^{*}$-algebra $A$, a map $\alpha \in Aut(A)$ with $\alpha ^{n}=id$ and $n\in \mathbb{N}$. Of what structure is the crossed product $A \rtimes_{\alpha}\mathbb{Z}_{n}$ then?

The construction I know begins with the space $C_c(\mathbb{Z}_n, A)$ of formal polynomials $\sum_{k\in\mathbb{Z}_{n}}t^kx_k$ where $x_0, ..., x_{n-1} \in A$ on which one can define a suitable addition, multiplication, involution and a $C^*$-norm. The crossed product $A \rtimes_{\alpha}\mathbb{Z}_{n}$ is the completion of $C_c(\mathbb{Z}_n, A)$ by this norm, then.

My question: Since the definition of the norm for crossed products looks rather complicated, what do the elements of $C_c(\mathbb{Z}_n, A)$ look like? Isn't $C_c(\mathbb{Z}_n, A)$ already complete since $\mathbb{Z}_n$ is finite or do I miss something?

$\endgroup$
4
$\begingroup$

In the construction you mention, I don't know what the "suitable" norm is.

Remember that the goal of the cross product is to implement the automorphisms by unitary conjugation.

The way I know the construction is, you embed $A\subset B(H)$ and $C_c(\mathbb Z_n,A)\subset B(H^n)$ in the following way. We actually consider $$C_c(\mathbb Z_n,A)\subset M_n(A),$$ by embedding $A\hookrightarrow M_n(A)$ via $$a\longmapsto \pi(a)=\sum_{k=1}^n\alpha^{k}(a)\otimes E_{kk},$$ and $\mathbb Z_n\hookrightarrow M_n(A)$ by $$m\longmapsto U_m=\sum_{k=1}^nE_{k,k+m-1},$$ where all the sums in the coordinates are interpreted in $\mathbb Z_n$.

One can show that $$\tag{1}U_m\pi(a)U^*_m=\pi(\alpha^m(a)),$$ and that $$\tag{2} C_c(\mathbb Z_n,A)=C^*(\{\pi(A),U_1,\ldots,U_n\})=\{X\in M_n(A):\ X_{kj}=\alpha^{k-1}(X_{1,j-k+1})\}. $$ Now we can assign to $C_c(\mathbb Z_n, A)$ its natural operator norm in $B(H^n)$, which agrees with norm pointwise convergence of the coordinates. With this, it is easy to check that $C_c(\mathbb Z_n, A)$ is closed. So $(1)$ gives a concrete expression for $A\rtimes_\alpha\mathbb Z_n$.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Hello Martin, thanks for your answer. I guess you mean $C_c(\mathbb{Z}_n, A)$ and $A \rtimes_\alpha \mathbb{Z}_n$ instead of $C_c(\mathbb{Z}, A)$ and $A \rtimes_\alpha \mathbb{Z}$ at the end? And if I understand you correctly we have $A \rtimes_\alpha \mathbb{Z}_n = C_c(\mathbb{Z}, A)$ since the right hand side is closed, then? $\endgroup$ – worldreporter14 Aug 19 '17 at 19:44
  • $\begingroup$ Yes, absolutely. With $\mathbb Z$ the ideas are the same, but you don't get a closed algebra. $\endgroup$ – Martin Argerami Aug 19 '17 at 20:20

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.