First order rigidity of non-uniform higher rank arithmetic groups (Q2419718): Difference between revisions

Rigidity properties of lattices in semisimple Lie groups are the lifeblood of the subject of Lie groups and discrete subgroups. The paper under review here is an important work that proves a totally new type of rigidity result. In model theory, the notion of quasi-axiomatizability is a form of rigidity -- it says that a finitely generated group \(\Gamma\) is QA if every finitely generated group that satisfies the same first-order sentences in the language of groups as \(\Gamma\), is necessarily isomorphic to \(\Gamma\). For this reason, the authors call this property first-order rigidity. The authors of the paper prove: Any abstract group that is commensurable to an irreducible non-uniform lattice in a higher rank Lie group is first-order rigid. Apart from lattices like \(\mathrm{SL}_n(\mathbb{Z})\) for \(n \geq 3\), this includes \(\mathrm{SL}_2(\mathbb{Z}[1/p])\) too, which is a lattice in \(\mathrm{SL}_2(\mathbb{R}) \times \mathrm{SL}_2(\mathbb{Q}_p)\). This type of result is far from true in rank \(1\). Two crucial notions are that of a group being prime and an element of a group having the Brenner property. A group \(G\) is said to be prime if it is an elementary submodel of every group \(H\) for which the sentences that are true are exactly those that are true for \(G\). The second property is interesting -- it is crystallized from a property proved for certain elements in \(\mathrm{SL}(n,\mathbb{Z})\) (\(n \geq 3\)) in a classical 1960 paper of \textit{J. L. Brenner} [Ann. Math. (2) 71, 210--223 (1960; Zbl 0103.26302)]. The main result of the present paper asserts: Theorem. Let \(\Gamma\) be any finitely generated, linear group satisfying: \begin{itemize} \item[(i)] the FC-center of \(\Gamma\) is finite, \item[(ii)] \(\Gamma\) is prime, \item[(iii)] there exists \(b \in \Gamma\) which has the Brenner property and there is a constant \(k\) so that every finitely generated subgroup of \(Z(C_{\Gamma}(b))\) is \(k\)-generated. \end{itemize} Then, \(\Gamma\) is first-order rigid. The authors easily verify that \(\mathrm{SL}(n,\mathbb{Z})\) for \(n \geq 3\) satisfies the three hypotheses above (\(E_{1n}\) satisfies the Brenner property in \(\mathrm{SL}(n,\mathbb{Z})\) as seen in Brenner's paper) and hence, as a corollary, it is first-order rigid. For proving the general theorem, the authors first show that super-rigid lattices are prime. Incidentally, the converse is false. An outcome of this result is that an abstract group commensurable with an irreducible non-uniform lattice in a higher rank group is co-Hopfian (every injective endomorphism is an automorphism). For producing an element in the general case that satisfies the Brenner property, the authors use \textit{G. A. Margulis}'s [Discrete subgroups of semisimple Lie groups. Berlin etc.: Springer-Verlag (1991; Zbl 0732.22008)] arithmeticity theorem and the method of proof of an old result of \textit{M. S. Raghunathan} [Publ. Math., Inst. Hautes Étud. Sci. 46, 107--161 (1976; Zbl 0347.20027), Theorem 2.1] in his affirmative solution in 1976 of the congruence subgroup problem for all groups of rank \(\geq 2\). Following the proof of the main theorem, the authors point out that the concept of first-order rigidity is not preserved when we go to a finite index subgroup or to a finite extension. In the final section, the authors make several interesting remarks and raise intriguing questions. These include the question of first-order rigidity for: (a) uniform higher tank arithmetic lattices, and for: (b) arithmetic groups satisfying the congruence subgroup property. In particular, an affirmative answer to the latter question would imply that uniform lattices in \(\mathrm{Sp}(n,1)\) fail to have the congruence subgroup property (as they are known not to be first-order rigid).