I'm trying to understand the definition of homotopy between Kasparov modules as presented in Blackadar's book on K-theory for operator algebras. $A,B$ will be C*-algebras, while $E$ will denote a Hilbert $B$-module. We define homotopy:

Let $i=0,1$ and let $f_{i}$ denote the respective evaluation morphism from $IB:=C([0,1],B)$ to $B$. Two Kasparov $A$-$B$ modules $(E_{i},\phi_{i},F_{i})$ are homotopic if there exists an $A$-$IB$ module $(E,\phi,F)$ such that $(E \otimes_{f_{i}}B,f_{i}\circ\phi,f_{i,*}F)\simeq^{unitary}(E,\phi_{i},F_{i})$.

Here $E\otimes_{f_{i}}B$ denotes the graded tensor product with respect to $f_{i}$. My troubles are with the definition of the third Kasparov module. In the definition of a Kasparov module we need $f_{i}\circ\phi$ to be a graded *-homomorphism from $A$ to $\mathbb{B}(E\otimes_{f_{i}}B)$. However, $\phi:A\rightarrow\mathbb{B}(E)$ while $f_{i}:IB\rightarrow B$, so obvious composition doesn't work. One could try to simply demand $f_{i}\circ\phi:=\phi\otimes 1$, but this is independent of $f_{i}$ and hence most likely the false definition. Similarly, I don't see where we naturally apply the evaluation morphism with an element of $\mathbb{B}(E)$, for example $F$, to turn it into an element of $\mathbb{B}(E\otimes_{f_{i}}B)$.

Thank you for your help!


1 Answer 1


The notation is a little confusing, yes. In fact, you are right that you could just take $f_i \circ \phi := \phi \otimes 1$. This looks like it is independent of $f_i$, but it's not: $f_i$ is hidden in the definition of the tensor product. If you want to clarify this, write it as $\phi\otimes_{f_i}1$.

Effectively, you should see $E$ as a family of $A$-$B$-bimodules indexed by $i \in [0,1]$. The tensor product $E\otimes_{f_i}B$ is the fibre at $i$. The representation $\phi\otimes_{f_i}1$ is the restriction of the $A$ action to this fibre.

  • 1
    $\begingroup$ You are of course right that $f_{i}$ plays a role in the definition of the tensor product. I guess we then write $f_{i,*}F:=F\otimes_{f_{i}} 1$ as well? $\endgroup$ Commented Sep 25, 2014 at 10:32

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.