Suppose that $A$ is a separable, simple, non-unital C*-algebra. Let $\varphi$ be an approximately inner automorphism on $A\otimes\cal K$, meaning that there exists a sequence of unitaries $v_n$ in the multiplier algebra $\mathcal{M}(A\otimes\mathcal{K})$ with $v_n x v_n^* \stackrel{n\to\infty}{\longrightarrow} \varphi(x)$ for all $x\in A\otimes\cal K$. We also denote by $\varphi$ the induced automorphism on the multiplier algebra.

Let $e_{11}\in\cal K$ denote the obvious rank one projection.

My question: Is $1\otimes e_{11}$ Murray-von-Neumann equivalent to $\varphi(1\otimes e_{11})$ in the multiplier algebra $\mathcal{M}(A\otimes\mathcal{K})$?

It is obvious that this is true if $A$ is unital, but unclear (to me) otherwise. I don't know whether simplicity of $A$ is a red herring here, but that is the case that I am interested in for now.

  • What is $1\otimes e_{11} $ if $A $ is not unital? – vap Apr 19 '17 at 18:11
  • It is the multiplier given by $(1\otimes e_{11})(a\otimes e_{kl}) = \delta_{1,k}\cdot a\otimes e_{1l}$ for all $a\in A$ and $k,l\in\mathbb{N}$. (Analogous formula from the right) – Gabor Szabo Apr 20 '17 at 9:16
  • Thanks. So, for $n$ large and $x $ a projection, $v_nxv_n^*$ is close to $\varphi (x) $ and therefore unitarily equivalent to it. Then $v_nxv_n^*$ is equivalent to $x $. Is there a problem with this argument? – vap Apr 20 '17 at 14:26
  • Yes. The problem is that, as $A$ is non-unital, the element $1\otimes e_{11}$ is a multiplier not in $A\otimes\cal K$, so the limit $\varphi(1\otimes e_{11}) = \lim_{n\to\infty} v_n (1\otimes e_{11}) v_n^*$ is in the strict topology, but a priori not in norm. – Gabor Szabo Apr 21 '17 at 10:36
  • This is true if $A$ has stable rank one (no simplicity). Is that interesting already? Also true if $A$ is simple purely infinite (since then it would be stable by Zhang's dichotomy). – Leonel Robert Apr 25 '17 at 14:46
up vote 3 down vote accepted

If $A$ has stable rank one then this can be deduced from the following result: If $B$ has stable rank one and $a,b\in B_+$ are Cuntz equivalent then the right ideals $\overline{aB}$ and $\overline{bB}$ are isomorphic as Hilbert modules. This is Theorem 3 of

Coward, Kristofer T.; Elliott, George A.; Ivanescu, Cristian. The Cuntz semigroup as an invariant for $C^*$-algebras. J. Reine Angew. Math. 623 (2008), 161--193.

See also Proposition 1 of

Ciuperca, Alin; Elliott, George A.; Santiago, Luis. On inductive limits of type-I $C^*$-algebras with one-dimensional spectrum. Int. Math. Res. Not. IMRN 2011, no. 11, 2577--2615.

It's a little awkward to translate this into an answer to the question. Let me try.

First, this: Let $p,q\in M(B)$ be multiplier projections such that $pB\cong qB$ as Hilbert modules. Then $p$ is Murray-von Neumann equivalent to $q$. Proof: Let $v\colon pB\to qB$ be an isomorphism. Extend $v$ to $B$ setting it to 0 on $(1-p)B$. Then $v\in M(B)$ (regarded as the adjointable operators on $B$) and $v^*v=p$, $vv^*=q$.

Now let $B$ be separable and of stable rank one. Let $\phi$ be an approximately inner automorphism extended to $M(B)$. Let us show that any projection $p\in M(B)$ is Murray-von Neumann equivalent to $\phi(p)$. It suffices to show that $pB\cong \phi(p)B$. Let $c\in B$ be a strictly positive element of $pBp$ (exists since $B$ is separable). Then $pB=\overline{cB}$. Since $c^{1/n}\uparrow p$ strictly and $\phi$ is strictly continuous, we also have $\phi(p)B=\overline{\phi(c)B}$. But $c$ and $\phi(c)$ are approximately unitarily equivalent, so they are Cuntz equivalent. We can use the theorem recalled above.

To answer the question, we take $B=A\otimes \mathcal K$ and $p=1\otimes e_{11}$. Notice that to answer the question we only need $A$ $\sigma$-unital rather than separable.

Remark: There exist examples of positive elements $a,b$ in an $A$ (of sr=2) that are approximately unitarily equivalent and yet $\overline{aA}$ and $\overline{bA}$ are not isomorphic as Hilbert modules. However, (1) $A$ is non-simple and they way the examples work they can't be tweaked to get this, (2) The unitary conjugations moving $a$ to $b$ don't seem to converge pointwise to an automorphism. So there is a chance that the question has a positive answer without sr1.

  • Very nice, thanks! Since this is already very useful to me, I accept this as an answer. – Gabor Szabo Apr 27 '17 at 14:57

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.