4
$\begingroup$

Is there an example of simple and non-nuclear(non-amenable) $C^*$-algebra?

$\endgroup$
  • 2
    $\begingroup$ Look for C*-simple groups. For example, Powers has shown that the reduced group C*-algebra of a nonabelian free group is simple. $\endgroup$ – Ulrich Pennig Oct 20 '15 at 8:19
  • $\begingroup$ @UlrichPennig Why not leave this as an answer? Although this is standard knowledge for specialists it strikes me that it would not be universally known $\endgroup$ – Yemon Choi Oct 20 '15 at 13:27
9
$\begingroup$

Following Yemon Choi's suggestion I turn my comment into an answer:

Lance gave a characterization of amenability in terms of the reduced group $C^*$-algebra: A discrete group $G$ is amenable if and only if $C^*_r(G)$ is nuclear. Therefore one approach to finding examples of non-nuclear simple $C^*$-algebras might be to look for non-amenable groups, such that their reduced group $C^*$-algebra is simple.

Powers showed in the paper "Simplicity of the $C^*$-algebra associated with the free group on two generators" that this is in fact true for $G = \mathbb{F}_2$. But the story does not end here. A lot of interesting research has been done concerning the question of $C^*$-simplicity of discrete groups. You might want to look for example at the fairly recent paper of Breuillard, Kalantar, Kennedy and Ozawa (arXiv).

$\endgroup$
  • $\begingroup$ The only link I can find for Powers' paper is behind a paywall, sorry. $\endgroup$ – Ulrich Pennig Oct 20 '15 at 15:33
  • 3
    $\begingroup$ Just because it is something I always go on about, I shall do so here :) ... Lance proved that $G$ is amenable iff ${\rm C}_r^*(G)$ is nuclear, using one of the original definitions of nuclearity. Then Bunce proved, without requiring any of Lance's results, that if ${\rm C}_r^*(G)$ is amenable as a Banach algebra and $G$ is discrete, then $\ell^\infty(G)$ has a (left or right or two-sided) invariant mean, so $G$ is amenable. I mention these in order to emphasize that non-nuclearity & non-amenability of ${\rm C}_r^*(F_2)$ can be discussed without mentioning the names of Connes or Haagerup :) $\endgroup$ – Yemon Choi Oct 20 '15 at 15:59
  • $\begingroup$ @YemonChoi: and yet you mentioned them ... $\endgroup$ – Nik Weaver Oct 20 '15 at 18:51
  • $\begingroup$ @NikWeaver Who says I have to mention pink elephants? Oh, damn :) $\endgroup$ – Yemon Choi Oct 20 '15 at 18:59
  • $\begingroup$ BTW I hope my comment wasn't perceived as denigrating the amenable iff nuclear results, which I hold in very high regard. It's just that sometimes they are used when there are easier, and more informative, approaches for particular examples one is considering $\endgroup$ – Yemon Choi Oct 20 '15 at 19:02

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.