I am familiar with the Morita theory of rings, and the hermitian Morita theory of rings with involution, and I am trying to understand some parallels and differences with the Morita theory of C*-algebras.

In the algebraic version, we are interested in the monoid structure of the Morita equivalence classes of $R$-algebras (where R is a commutative ring), given by the tensor product over $R$. In particular, the invertible elements of this monoid are given by the Azumaya algebras over $R$, and they form the Brauer group of $R$. Do similar phenomena occur for C*-algebras?

For von Neumann algebras, I asked a question on Math.SE, and someone commented that the theory is basically empty: since factors of type I are Morita-trivial, and a tensor product of a factor of type II or III with another factor is again of type II or III, in the end the only way to be Morita-invertible is to be Morita-trivial.

Do C*-algebras offer more theory? I understand that in that context one has to be a little more careful: the Morita-equivalence I care about is the strong Morita equivalence (as defined by imprimitivity bimodules). Also, talking about tensor products can be awkward, so maybe I should restrict to nuclear C*-algebras (but if there are results of Morita-invertibility for well-chosen tensor products on non-nuclear algebras, I am also interested in hearing about it).

Clearly, one has to restrict to unital and central algebras, so in the end my question is the following:

If $A$ is a central unital C*-algebra (maybe nuclear), and there exists $B$ such that $A\otimes B$ is strongly Morita-equivalent to $\mathbb{C}$ (for some tensor product if $A$ is not nuclear), does it follow that $A$ itself is strongly Morita-equivalent to $\mathbb{C}$?

I am also interested in similar results for real C*-algebras (maybe even more so).

  • 1
    $\begingroup$ There has definitely been stuff on the Cstar version of Azumaya algebras, with reference to the Brauer group, but I don't recall a precise reference right now (maybe Raeburn, or Raeburn and Taylor, or D. Williams?) - my impression was that this was more actively pursued in the 1970s and 1980s. $\endgroup$
    – Yemon Choi
    Dec 31 '19 at 18:11
  • $\begingroup$ @YemonChoi Indeed, thanks to your comment I have come across the book "Morita equivalence and continuous-trace C*-algebras" by Raeburn and Williams, which seems to address my question rather well. $\endgroup$ Jan 2 '20 at 11:39

I know my answer is coming a bit late, but the answer to your question is: yes. If $A$ is a $C^\ast$-algebra, and there exists a $C^\ast$-algebra $B$ such that $A\otimes_\alpha B$ is strongly Morita equivalent to $\mathbb C$ for some $C^\ast$-tensor product $\otimes_\alpha$, then $A\cong \mathcal K(H)$ is the compact operators on some Hilbert space $H$, and in particular $A$ is strongly Morita equivalent to $\mathbb C$. If $A$ is unital (as in the question), then $H$ is finite dimensional, so $A$ is a matrix algebra.

It is well-known that strong Morita equivalence preserves: (1) being a Type I $C^\ast$-algebra (also called GCR); and (2) the spectrum. Hence $A\otimes_\alpha B$ is of Type I and its spectrum is a point. In particular $A\otimes_\alpha B$ is simple so $\otimes_\alpha$ is the minimal tensor product (since the $A\otimes_{\min{}} B$ is a quotient of $A\otimes_{\alpha} B$). By Theorem 2 in Tomiyama's paper "Applications of Fubini type theorem to the tensor products of C∗-algebras. Tohoku Math. J. (2) 19 (1967), 213–226." it follows that $A$ and $B$ are Type I. By Theorem B.45 in Raeburn and William's book on Morita equivalence, it follows that the spectrum of $A$ is a singleton. As Naimark's problem* is true amongst Type I $C^\ast$-algebras it follows that $A \cong \mathcal K(H)$ for some Hilbert space $H$.

*Naimark's problem: if $A$ is a $C^\ast$-algebra for which the spectrum is a singleton, is $A$ isomorphic to the compact operators on some Hilbert space?

  • $\begingroup$ ... in the last sentence, perhaps "is $A$ isomorphic" instead of "$A$ is isomorphic"? $\endgroup$
    – Nik Weaver
    Jan 15 '20 at 5:41
  • 1
    $\begingroup$ Thanks, Nik, it has been changed. $\endgroup$
    – Jamie Gabe
    Jan 16 '20 at 0:20

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.