Principal congruence link complements (Q2339513)

This paper is concerned with the exceptional topological properties enjoyed by the hyperbolic three-manifolds which are constructed by purely arithmetic means. The authors take the latter to mean that they are obtained as a quotient of hyperbolic-three space \(\mathbb{H}^3\) by a subgroup \(\Gamma\) of one of the so-called Bianchi groups \(\mathrm{PSL}(2,\mathcal{O}_d)\) (where \(\mathcal{O}_d\) the ring of integers in an imaginary quadratic field). Moreover \(\Gamma\) has to be defined by a system of congruences modulo an ideal \(I\) in \(\mathcal{O}_d\). This last condition is equivalent to the condition that \(\Gamma\supset\Gamma_d(I)\) where \(\Gamma_d(I)\) is the principal congruence subgroup of level \(I\), that is \(\Gamma = \ker\left(\mathrm{PSL}(2,\mathcal{O}_d) \to \mathrm{PSL}(2,\mathcal{O}_d/I)\right)\) Such a \(\Gamma\) is called a congruence subgroup (of \(\mathrm{PSL}(2,\mathcal{O}_d)\)), and the quotient \(\Gamma\backslash\mathbb{H}^3\) is called a congruence manifold, with the added qualificative of principal whenever \(\Gamma\) is such (this is the terminology in the paper under review). A theorem of Zograf [\textit{P. Zograf}, J. Reine Angew. Math. 414, 113--116 (1991; Zbl 0709.11031)] states that for a given closed surface \(S\) there are only finitely many congruence subgroups \(\Gamma\) of \(\mathrm{PSL}(2,\mathbb{Z})\) such that the quotient \(\Gamma\backslash\mathbb{H}^2\) is diffeomorphic to \(S\) minus a set of points (this is false if one allows \(\Gamma\) to be only a finite-index subgroup of \(\mathrm{PSL}(2,\mathbb{Z})\)). In three dimensions we can ask whether, given a closed three-manifold \(N\), there are only finitely many links \(L\subset N\) such that \(N\setminus L\) is a congruence manifold (this is also known to be false if one asks merely for subgroups of finite index instead of congruence subgroups). This is true for principal congruence manifolds and can be seen as follows. It is easy to see that the dimension of the first cuspidal cohomology of a link complement in \(N\) is bounded by the first Betti number of \(N\), hence the result that the cuspidal cohomology of Bianchi groups goes to infinity implies that there are only finitely many Bianchi groups that can contain fundamental groups of link complements. On the other hand the result that the systole of a link complement is bounded (a theorem of \textit{C. C. Adams} and \textit{A. W. Reid} [Math. Proc. Camb. Philos. Soc. 128, No. 1, 103--110 (2000; Zbl 0958.57015)] when \(N\) is not hyperbolic, and of \textit{G. S. Lakeland} and \textit{C. J. Leininger} [Algebr. Geom. Topol. 14, No. 3, 1441--1460 (2014; Zbl 1293.57010)] in general) implies that only finitely many principal congruence subgroups in a given group can belong to a link complement. The question for general congruence manifolds remains open and no new advance towards it is given in this paper. What it does is giving quantitative results for the principal case when \(N = \mathbb{S}^3\): the authors give a list of the principal manifolds that are known to be diffeomorphic to link complements (the proof is done by explicitely giving the links). In cases \(d=1,3\) all principal congruence subgroups of \(\mathrm{PSL}(2,\mathcal{O}_d)\) giving link complements are known by the work of M. Goerner.