On the dimension growth of groups. (Q2428016)

In his seminal work on the large-scale approach to groups, \textit{M. Gromov} [Geometric group theory. Volume 2: Asymptotic invariants of infinite groups. Cambridge: Cambridge University Press (1993; Zbl 0841.20039)] introduced the notion of asymptotic dimension. It turns out that finitely generated groups with finite asymptotic dimension satisfy many conjectures (such as the Novikov higher signature conjecture). Many large classes of groups have finite asymptotic dimension, but it is fairly easy to give examples of groups with infinite asymptotic dimension. The paper under review takes a dimension-theoretic approach to groups with infinite asymptotic dimension. In particular, the authors define (for each \(\lambda>0\)) the \(\lambda\)-\(\dim X\) of a metric space \(X\). (Here \(\lambda\)-\(\dim X\leq n\) if there is a uniformly bounded cover of \(X\) that can be decomposed into \(n+1\) \(\lambda\)-disjoint families.) The growth of this function \(d_X(\lambda)\) is a quasi-isometry invariant, so it makes sense to determine the dimension growth of a finitely generated group with infinite asymptotic dimension. One famous example of a finitely generated group with infinite asymptotic dimension is the R.~Thompson group \(F\). It is known that \(F\) contains a copy of \(\mathbb Z^m\) for each \(m\), and so has infinite asymptotic dimension. On the other hand, it was shown by \textit{D. S. Farley}, [Int. Math. Res. Not. 2003, No. 45, 2409-2414 (2003; Zbl 1113.22005)], that \(F\) coarsely embeds into Hilbert space. Since all groups with finite asymptotic dimension have G.~Yu's Property A and all Property A groups embed coarsely in Hilbert space, a natural question is whether \(F\) has Property A. This turns out to be very difficult. The paper under review seeks to understand \(F\) by computing its dimension growth. Indeed \textit{N. Ozawa} [Int. J. Algebra Comput. 22, No. 2, Article ID 1250011 (2012; Zbl 1238.54015)] has shown that a metric space with subexponential dimension growth has Property A. Thus, if \(F\) were to have subexponential dimension growth, this would settle the question of \(F\) having property A in the affirmative. Unfortunately, the growth of \(d_F(\lambda)\) turns out to be exponential, but the paper contains very interesting results about growth of \(d_G(\lambda)\) and the amenability of group \(G\). In particular, the largest dimension of an amenable group that the authors are able to prove is \(e^{\sqrt\lambda}\), so there is hope that the dimension growth of \(F\) actually means it is not amenable.