Sharp phase transition theorems for hyperbolicity of random groups. (Q1889938)

The study of random groups was initiated by \textit{M. Gromov} [in his paper in Essays in group theory, Publ., Math. Sci. Res. Inst. 8, 75-263 (1987; Zbl 0634.20015)] in which he stated that almost every group is hyperbolic. Later on, in his paper [Geometric group theory, Vol. 2: Asymptotic invariants of infinite groups, Lond. Math. Soc. Lect. Note Ser. 182 (1993; Zbl 0841.20039)], \textit{M. Gromov} introduced a precise model for random groups which depends upon the choice of a density \(d\) between \(0\) and \(1\), and where one chooses a length \(\ell\) and picks at random a set \(R\) of \((2m-d)^{2\ell}\) uniformly chosen cyclically reduced words of length \(\ell\) in the letters \(a_1^{\pm 1},\dots,a_m^{\pm 1}\). With these notations, Gromov states the following theorem: if \(d<1/2\) then the probability that the presentation defines an infinite hyperbolic group tends to \(1\) as \(\ell\to\infty\); if \(d>1/2\) then the probability that the presentation \(\langle a_1,\dots,a_m;R\rangle\) defines a trivial group or a cyclic group of order \(2\) tends to \(1\) as \(\ell\to\infty\). In the paper under review, the author writes a proof of that theorem of Gromov's and he obtains similar results in more general situations. Thus, while Gromov's theorem gives conditions for a random quotient of a free group to be hyperbolic, the author obtains results concerning the following question: when is a random quotient of a hyperbolic group hyperbolic? The results obtained say that, as Gromov's result, there is a critical density \(d\) under which with high probability the quotient of a hyperbolic group (with ``harmless torsion'') is hyperbolic and above which it is trivial. The critical density is characterized explicitly. The hypothesis on harmless torsion is discussed in the paper and shown to be necessary. Hyperbolic groups with harmless torsion include torsion-free groups and free products of torsion-free groups and/or finite groups.