Densely related groups (Q2292304)

From MaRDI portal





scientific article
Language Label Description Also known as
default for all languages
No label defined
    English
    Densely related groups
    scientific article

      Statements

      Densely related groups (English)
      0 references
      0 references
      0 references
      3 February 2020
      0 references
      The authors develop a systematic study of a quasi-isometry invariant among finitely generated groups named the \textit{relation range}. This invariant was first introduced in [\textit{B. H. Bowditch}, Comment. Math. Helv. 73, No. 2, 232--236 (1998; Zbl 0924.20032)] to give a short proof that there are continuously many quasi-isometry classes of \(2\)-generator groups with small cancellation. Given a group \(G\) and a finite generating set \(S\) of \(G\), let \(F_S\) be the free group on \(S\). A \textit{relation} in \(G\), with respect to \(S\), is an element in the kernel of the natural map \(F_S \twoheadrightarrow G\). Define \(\mathcal{R}_S(G) \subseteq \mathbb{N}\) as follows: \(n \in \mathcal{R}_S(G)\) if there is a relation in \(G\) of length \(n\), with respect to the word length associated to \(S\), which cannot be written as a product of conjugates of relations of lesser length. Define the \textit{relation range} of \(G\) as the equivalence class of \(\mathcal{R}_S(G)\) for the relation \(\sim\) identifying two subsets of \(\mathbb{N}\) if their images by \(x \mapsto \log_2(1 + x)\) are at finite Hausdorff distance from each other. The authors give a new proof that \(\mathcal{R}(G)\) does not dependent on the choice of \(S\) but only on the quasi-isometry class of \(G\). They actually obtain a stronger statement (Theorem 1.5) from which they infer that if \(H\) is a group retract of \(G\), then there are subsets \(I \subseteq J \subseteq \mathbb{N}\) such that \(\mathcal{R}(H) \sim I\) and \(\mathcal{R}(G) \sim I\). They demonstrate further that, for every \(I \subseteq \mathbb{N}\), there is a lacunary hyperbolic group \(G\) such that \(\mathcal{R}(G) \sim I\). The same holds true if we instead consider the extensions \(G\) of a 3-nilpotent locally finite normal subgroup by an abelian group, or the groups \(G\) obtained by removing any subset of defining relations from the presentation \(\langle x, t \mid [t^nxt^{-n}, x]=1, \, n \ge 1 \rangle\) of \(\mathbb{Z} \wr \mathbb{Z}\). Within each class, a continuum of pairwise non-quasi-isometric groups is thus constructed. The paper focuses subsequently on the finitely generated groups \(G\) satisfying \(\mathcal{R}(G) \sim \mathbb{N}\). Such groups, which are necessarily infinitely related, are called \textit{densely related}. A remarkable result of the paper is that a densely related group has no simply connected asymptotic cone. Several criteria implying that a group is densely related are given. The authors show that a finitely generated restricted wreath product \(H \wr G\), with \(G\) infinite and \(H \neq \{1\}\), is densely related. They also provide a device to build a densely related group out of a finitely generated non-Hopfian group (Corollary 4.20) and they prove, using a similar device (Theorem 4.15), that the first Grigorchuk group is densely related. Eventually, the authors consider two classes of soluble groups: the class of finitely generated center-by-metabelian groups and the class of finitely generated nilpotent-by-cyclic groups. For \(G\) in any of these classes, they prove that \(G\) is either finitely presented or densely related. The paper culminates with Corollary 4.32, establishing the same dichotomy for the groups which satisfy a law and which are of the form \(M \rtimes \mathbb{Z}\) with \(M\) coherent. Numerous results are phrased in the more general setting of the compactly generated locally compact groups. The paper is written with great care and is richly illustrated. This makes it a recommended read for researchers interested in geometric group invariants tied to algorithmic properties. Note that the proof of Theorem 4.17 has a flaw: Theorem 4.15 cannot apply because condition (3) doesn't hold in the considered setting. Nonetheless, the theorem is true. It can be proven by means of the algorithm solving the word problem in the first Grigorchuk group and by resorting to Lysenok's presentation [\textit{I. G. Lysenok}, Math. Notes 38, 784--792 (1985; Zbl 0595.20030); translation from Mat. Zametki 38, No. 4, 503--516 (1985)].
      0 references
      0 references
      group presentation
      0 references
      quasi-isometry
      0 references
      asymptotic cone
      0 references
      relation range
      0 references
      densely related group
      0 references
      lacunary hyperbolic group
      0 references
      Grigorchuk group
      0 references
      wreath product
      0 references
      graph product
      0 references
      center-by-metabelian group
      0 references
      small cancellation
      0 references
      compactly generated locally compact group
      0 references
      0 references
      0 references
      0 references
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references