Hydra groups. (Q373477)

Some lines of historical nature seem to be appropriate, even if they are related more to the philological aspect than the mathematical context of the paper under review. Following ``The Theogony'' of Hesiod (probably written around 700 B.C.) on the genealogies of the Greek gods, we read about Hydra at 313-318, where it is noted that it was the son of Ekhidna and Typhon. The mother, Ekhidna, was half snake and half woman, while the father, Typhon, was the last son of Gaea and Tartarus, born to destroy Zeus. He was in fact a monster, but Zeus was able to trap him underneath Mount Etna. Ekhidna, on the other hand, was the son of Ceto and Phorcys, two primordial sea gods, who generated other monsters, too. The genealogy of Hydra shows that the ancient Greek philosophers had no idea about the regeneration of some rectiles, starfishes or octopuses. As usual in Mythology, the reasons are justified by means of other non-reasons in the genealogy of the gods. Hercules' second labour was to fight against Hydra. We read about his battle in ``Bibliotheca'' [2. 77-80] attributed to Apollodorus of Athens (born probably around 180 B.C.) but also in ``Bibliotheca Historica'' [4. 11. 5] of Diodorus Siculus (active around 60 and 30 B.C.). One of the main problems was to understand how to stop the regeneration of the heads of Hydra, since they regrow each time. With the help of Iolaus, Hercules understood that it was necessary to burn each head, after cut, so that the regeneration would have been stopped. This reflects the victory of the human knowledge on the forces of the nature, which is a ``leitmotif'' in most of the classical mythology. The problem of the regeneration is very interesting in several contexts of the science. The paper under review shows a wonderful idea in geometric group theory. Inspired by the battle of Hercules against Hydra, the authors construct some groups whose geometric properties are surprising. First of all, the algorithm of regeneration is formalized via the ``Hydra functions'' \(\mathcal H_k\), defined at p.508. These are very fast and related to some known functions in number theory, the ``Ackermann functions'' \(A_k\). Proposition 1.2 illustrates that they are actually equivalent for all \(k\geq 1\). The main result deals with the description of the asymptotic properties of the groups \[ \Gamma_k=\langle a_1,\ldots,a_k,t, p\mid t^{-1}a_1t=a_1,\;t^{-1}a_it=a_ia_{i-1}\;(i>1),\;[p,a_it]=1\;(i>0)\rangle \] for \(k\geq 2\). Indeed Theorem 1.4 shows that the Dehn function of \(\Gamma_k\) is equivalent to \(A_k\). This property is very interesting and the authors show a series of deep consequences in recent studies of geometric group theory. On the other hand, if we omit the generator \(p\) and the relations \([p,a_it]=1\) from the above \(\Gamma_k\), we obtain \(G_k\) and the authors look at the distortion function \(\mathrm{Dist}^{G_k}_{H_k}\) of a rank-\(k\) free subgroup \(H_k\) of \(G_k\). They show that \(\mathrm{Dist}^{G_k}_{H_k}\) is also equivalent to \(A_k\). Moreover, \(G_k\) turns out to be free-by-cyclic, one-relator, biautomatic and \(\mathrm{CAT}(0)\). The details are illustrated in Theorem 1.3. I don't enter into the technical aspects of the proofs, which are a wise combination of methods of geometric group theory, number theory and topology.