4
$\begingroup$

Let $A$ be a $C^*$-algebra, we denote with $V(A)$ the semigroup of Murray-von Neumann equivalence classes of projections in matrices over $A$ (as usual). In https://arxiv.org/pdf/math/0310340.pdf, above definition 5.1. in section 5 (on page 20), there is mentioned the following fact:

If $A$ is a $C^*$-algebra and $p$ a projection in $A$, and $n\in\mathbb{N}_{\ge 1}$, then $nx=[p]$ has a solution $x\in V(A)$ if and only if there is a unital embedding $M_n\to pAp$. $(\star)$

I need help to prove the direction $\Rightarrow$. (The reason why I take a closer look to this statement is that I need $\star$ for an other proof.)

To prove that there is such a unital embedding, one strategy could be (if I'm not mistaken): Find mutually orthogonal projections $f_1,\cdots f_n$ in pAp, with $f_1\sim f_2 \sim \cdots \sim f_n$ (where $\sim$ is Murray-von Neumann equivalence) and such that $1=f_1+ \cdots +f_n$. Then you can extend the set $\{f_1,\cdots,f_n\}$ to matrix units of $M_n$.

Now, if $nx=[p]$ for an $x\in V(A)$. Write $x=[x']$ for a projection $x'\in M_m(A)$ for some natural number $m$. Then $\operatorname{diag}(x',x',\cdots,x')\sim_0 p.$ I.e. if $p$ is a projection in $M_k(A)$, then there exists a partial isometry $v\in M_{k,nm}$ such that $\operatorname{diag}(x',x',\cdots,x')=vv^*$ and $p=v^*v$ (it's Murray-von Neumann equivalence in different sized matrices). But I don't know how to get these projections $f_i$ (from here)...

How to get these projections $f_i$/ or can you provide an alternative approach to prove it/ or can you provide a reference for a proof of $\star$? Thank you.

$\endgroup$
4
$\begingroup$

Using your notation, we have diag$(x',...,x')\in M_n\otimes M_m(A).$ Then $M_n$ is isomorphic to the algebra (let's call it $B$) generated by $e_{ij}\otimes x'$ where $e_{ij}$ are matrix units for $M_n.$ Since $e_{ij}\otimes x'$ (and hence everything in $B$) commutes with $vv^*$, the map from $B$ to $pAp$ defined by $b\mapsto v^*bv$ is a *-homomorphism that takes the unit of $B$ (which is diag$(x',...,x')$) to $p.$

$\endgroup$

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.