I am reading the presentation of Cuntz' proof of Bott periodicity for $C^*$-algebras in Wegge-Olsen (Thm. 11.2.1). Here one considers the short exact sequence of $C^*$-algebras $$0 \longrightarrow \mathcal{T}_0 \longrightarrow \mathcal{T} \stackrel{q}{\longrightarrow} \mathbb{C} \longrightarrow 0,$$ where $\mathcal{T}$ is the Toeplitz algebra and $q$ is the homomorphism determined by sending the shift operator $S \in \mathcal{T}$ to $1$. Clearly, we also have the map $j: \mathbb{C} \rightarrow \mathcal{T}$, and $q \circ j = \mathrm{id}_{\mathbb{C}}$.

Now, the main step in the proof is to show that the induced map in $K$-theory $q_*: K_0(\mathcal{T}) \rightarrow K_0(\mathbb{C})$ is an isomorphism. But then, it is claimed that it was also clear that also for any $C^*$-algebra $A$, one has that also $$(\mathrm{id}_A \otimes q)_*: K_0(A \otimes \mathcal{T}) \rightarrow K_0(A)$$ is an isomorphism.

I do understand that since all $C^*$-algebras in the short exact sequence are nuclear, the sequence $$0 \longrightarrow A \otimes \mathcal{T}_0 \longrightarrow A \otimes \mathcal{T} \stackrel{\mathrm{id}_A \otimes q}{\longrightarrow} A \longrightarrow 0$$ is exact for any $C^*$-algebra. But I do not see why knowing that $q_*$ is an isomorphism implies the same thing for $(\mathrm{id}_A \otimes q)_*$. Any suggestions?


1 Answer 1


The conclusion drawn in the book of Wegge-Olsen is wrong (explained below), but can, however, easily be tweaked to a correct proof. What is shown is that $j\circ q$ is homotopic to the identity on $\mathcal T$ and hence the same is true after tensoring with $A$. It follows that $K_\ast(\mathcal T_0 \otimes A) = 0$ for every $C^\ast$-algebra $A$.

A quotient map $q$ of $C^\ast$-algebras is an isomorphism in $K$-theory if and only if $K_\ast(\ker q) =0$ (by six-term exactness). So the statement that a quotient map $q$ (with splitting $j$) induces an isomorphism in $K$-theory implies that $(q\otimes_\alpha id_A)_\ast$ is an isomorphism is equivalent to the statement that $K_\ast(B)=0$ implies $K_\ast(B \otimes_\alpha A) = 0$ (for $\otimes_\alpha$ being either the maximal or minimal tensor product). And this is in general wrong. A $C^\ast$-algebra $A$ satisfies the maximal (respectively minimal) Künneth theorem if and only if $K_\ast(B \otimes_{\max} A) = 0$ (resp. $K_\ast(B \otimes_{\min{}} A) = 0$) for every $C^\ast$-algebra $B$ with $K_\ast(B) = 0$, see https://arxiv.org/abs/1111.7228 Theorems 3.1 and 4.1. This paper also contains examples of $C^\ast$-algebras that don't satisfy the maximal/minimal Künneth theorem. These examples are modelled after Skandalis' counterexamples to the Universal Coefficient Theorem.

  • $\begingroup$ Thanks, I thought so. $\endgroup$ May 11, 2020 at 4:17

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.