existence of charaterization of amenable groups by complementation? - MathOverflow most recent 30 from http://mathoverflow.net2013-05-25T04:08:48Zhttp://mathoverflow.net/feeds/question/40795http://www.creativecommons.org/licenses/by-nc/2.5/rdfhttp://mathoverflow.net/questions/40795/existence-of-charaterization-of-amenable-groups-by-complementationexistence of charaterization of amenable groups by complementation?BigBill2010-10-01T21:50:06Z2010-10-02T08:12:45Z
<p>Recall that we say that a closed space $F$ of a Banach space $E$ is complemented if there exists a contractive projection $P$ from $E$ onto $F$.</p>
<blockquote>
Do you know a charaterization of discrete amenable groups by the existence of a complementation of a closed space $F$ of a Banach space $E$?
</blockquote>
<p>More precisely, the required charaterization is<br>
For all discrete group $G$, there exists a Banach space $E_G$ and a closed space $F_G$ of $E_G$ such that
$G$ is amenable if and only if $F_G$ is complemented in $E_G$.</p>
http://mathoverflow.net/questions/40795/existence-of-charaterization-of-amenable-groups-by-complementation/40820#40820Answer by Matthew Daws for existence of charaterization of amenable groups by complementation?Matthew Daws2010-10-02T08:12:45Z2010-10-02T08:12:45Z<p>Yes: a <em>discrete</em> group $G$ is amenable if and only if the reduced group C*-algebra $C^*_r(G)$ is nuclear, see E.C. Lance, On nuclear $C^{\ast} $-algebras.
J. Functional Analysis 12 (1973), 157--176. This is then equivalent to <code>$W^*(G) = C^*_r(G)^{**}$</code> being an <em>injective</em> von Neumann algebra: which by definition means that if $W^*(G) \subseteq B(H)$ then there is a <em>contractive</em> projection from $W^*(G)$ to $B(H)$.</p>
<p>I'm pretty sure you could look at the group von Neumann algebra $VN(G)$ instead, but I cannot recall the correct reference (but it's all in Runde's book "Lectures on Amenability"). Note that all this only works because $G$ is discrete.</p>
<p>Now, the problem is that you do need ``contractive'' projection here: it's still a conjecture if just having a bounded projection is enough.</p>
<p>Also, I'm sure there are other answers (and perhaps some that are easier: even a streamlined approach to all this uses a lot of operator algebra theory)...</p>