9
$\begingroup$

The real rank for C*-algebras was defined by Brown-Pedersen in [1] as a noncommutative analog of covering dimension. Given a unital C*-algebra $A$, its real rank $\mathrm{rr}(A)$ is the smallest natural number $n$ (possibly $n=0$) such that for every $n+1$-tuple $(x_0,\ldots,x_n)$ of selfadjoint elements in $A$ and every $\varepsilon>0$ there exists an $n+1$-tuple $(y_0,\ldots,y_n)$ of selfadjoint elements in $A$ such that $\|x_k-y_k\|<\varepsilon$ for $k=0,\ldots,n$ and such that $\sum_k y_k^*y_k$ is invertible. If no such $n$ exists, one sets $\mathrm{rr}(A)=\infty$. If $A$ is nonunital, one sets $\mathrm{rr}(A)=\mathrm{rr}(\widetilde{A})$ for the minimal unitization $\widetilde{A}$.

Question: Do we have $\mathrm{rr}(M_n(A))\leq\mathrm{rr}(A)$?

It seems that this ought to be true, but I couldn't find such a statement anywhere. For the stable rank (which is defined analogous to the real rank, but using not necessarily selfadjoint tuples) Rieffel proved the formula $\mathrm{sr}(M_n(A))=\lceil \frac{\mathrm{sr}(A)-1}{n}\rceil+1$. Rieffel's method does not seem to generalize to the real rank. In fact, I can show that for the real rank there cannot be a formula computing the real rank of $M_n(A)$ in terms of the real rank of $A$. But one would at least hope for the possiblity to bound $\mathrm{rr}(M_n(A))$ in terms of $\mathrm{rr}(A)$.

[1] Brown, Pedersen. C*-algebras of real rank zero, J. Funct. Anal. 99 (1991)

[2] Rieffel. Dimension and stable rank in the K-theory of C*-algebras, Proc. Lond. Math. Soc. 46 (1983)

$\endgroup$
2
  • $\begingroup$ You are almost surely aware of this, but the answer is known to be affirmative in case $\mathrm{rr}(A) = 0$ by Theorem 2.10 in Brown & Pedersen. $\endgroup$ Commented Jan 30, 2020 at 16:21
  • 1
    $\begingroup$ Yes, one can handle specific cases of low rank. If rr(A)=0, then rr(M_n(A))=0. If rr(A)=1 and sr(A)=1, then sr(M_n(A))=1 and thus rr(M_n(A))\leq 1 (and actually =1), But even the general case of rr(A)=1 is not clear to me. $\endgroup$ Commented Jan 31, 2020 at 10:28

0

You must log in to answer this question.