The limit set of subgroups of arithmetic groups in \(\mathrm{PSL}(2,\mathbb C)^q\times\mathrm{PSL}(2,\mathbb R)^r\). (Q2264078)

The main goal of this article is to connect the arithmetic and the geometric properties of subgroups of irreducible lattices in \(\mathrm{PSL}(2,\mathbb C)^q\times\mathrm{PSL}(2,\mathbb R)^r\). The author is in particular interested in those groups whose projection in one of the factors is a subgroup of an arithmetic Fuchsian or Kleinian group and investigates their limit sets. In Section 1, the author addresses the question of finding a criterion for a nonelementary subgroup \(\Gamma\) of \(\mathrm{PSL}(2,\mathbb C)\) to be Zariski dense over \(\mathbb R\). It is known that a nonelementary subgroup of \(\mathrm{PSL}(2,\mathbb K)\) is Zariski dense over \(\mathbb K\) for \(\mathbb K=\mathbb R\) and \(\mathbb C\) [\textit{G. Cornelissen} and \textit{M. Marcolli}, J. Geom. Phys. 58, No. 5, 619-632 (2008; Zbl 1174.58004)]. The author proves that \(\Gamma\) is Zariski dense over \(\mathbb R\) if and only if the set of traces of all elements of \(\Gamma\) is not contained in \(\mathbb R\) (Corollary 1.5). This result immediately follows after the author proves the same result for Schottky groups (Lemma 1.4). Section 2 provides definitions of several limit sets of a subgroup of \(\mathrm{PSL}(2,\mathbb C)^q\times\mathrm{PSL}(2,\mathbb R)^r\). Let \(\mathbb H^n\) denote the upper half \(n\)-space equipped with the hyperbolic distance. Then \(\mathrm{PSL}(2,\mathbb C)^q\times\mathrm{PSL}(2,\mathbb R)^r\) acts on \((\mathbb H^3)^q\times(\mathbb H^2)^r\) by isometries and also on its \textit{geodesic boundary} \(\partial((\mathbb H^3)^q\times(\mathbb H^2)^r)\), the set of equivalence classes of geodesic rays: Two (unit speed) geodesic rays are equivalent to each other if they are asymptotic. The geodesic boundary contains the \textit{regular boundary} \(\partial((\mathbb H^3)^q\times(\mathbb H^2)^r)^{\mathrm{reg}}\) which consists of equivalence classes of geodesic rays such that their projection to each factor space is a non-constant geodesic ray in \(\mathbb H^3\) or in \(\mathbb H^2\). There is a natural identification of \(\partial((\mathbb H^3)^q\times(\mathbb H^2)^r)^{\mathrm{reg}}\) with \((\partial\mathbb H^3)^q\times(\partial\mathbb H^2)^r\times\mathbb{RP}_+^{q+r-1}\). Let \(\Gamma\) be a subgroup of \(\mathrm{PSL}(2,\mathbb C)^q\times\mathrm{PSL}(2,\mathbb R)^r\). The \textit{limit set} \(\mathcal L_\Gamma\) of \(\Gamma\) is \(\overline{\Gamma(x)}\cap\partial((\mathbb H^3)^q\times(\mathbb H^2)^r)\), where \(\Gamma(x)\) is the \(\Gamma\)-orbit of \(x\in(\mathbb H^3)^q\times(\mathbb H^2)^r\), and it is independent of \(x\). The \textit{regular limit set} of \(\Gamma\) is \(\mathcal L_\Gamma^{\mathrm{reg}}=\mathcal L\cap\partial((\mathbb H^3)^q\times(\mathbb H^2)^r)^{\mathrm{reg}}\). The projections \(F_\Gamma\) and \(P_\Gamma\) of \(\mathcal L_\Gamma^{\mathrm{reg}}\) to \((\partial\mathbb H^3)^q\times(\partial\mathbb H^2)^r\) and to \(\mathbb{RP}_+^{q+r-1}\) are called the \textit{Furstenberg limit set} and the \textit{projective limit set} of \(\Gamma\), respectively. Let \(p_j\) denote the projection of \(\mathrm{PSL}(2,\mathbb C)^q\times\mathrm{PSL}(2,\mathbb R)^r\) to the \(j\)-th factor for \(j=1,\ldots,q+r\). In this article \(\Gamma\) is called \textit{nonelementary} if \(p_j(\Gamma)\) is nonelementary for all \(j\) and if for all \(g\in\Gamma\) and \(j\), \(p_j(g)\) is either loxodromic or elliptic of infinite order. Section 3 treats nonelementary groups. Let \(\Gamma\) be nonelementary. Then \(\mathcal L_\Gamma^{\mathrm{reg}}\) is non-empty (Lemma 3.3). For loxodromic or elliptic \(g\) in \(\mathrm{PSL}(2,\mathbb C)^q\times\mathrm{PSL}(2,\mathbb R)^r\), let \(2x_j\in\mathbb R_{\geq 0}\) be the translation length of \(p_j(g)\). Then the \textit{translation direction} of \(g\) is \(L(g)=(2x_1:\ldots:2x_{q+r})\) in \(\mathbb{RP}^{q+r-1}_+\). The \textit{limit cone} of a nonelementary group \(\Gamma\) is the closure of \(L(\Gamma)=\{L(g):g\in\Gamma\}\) in \(\mathbb{RP}^{q+r-1}_+\). If \(q+r\geq 2\), then its projective limit set \(P_\Gamma\) is convex and its closure in \(\mathbb{RP}^{q+r-1}_+\) is equal to the limit cone of \(\Gamma\) and hence the limit cone is convex (Proposition 3.8). This is a partial extension of a theorem by \textit{Y. Bénoist} [Geom. Funct. Anal. 7, No. 1, 1-47 (1997; Zbl 0947.22003)]. After a preparatory section (Section 4) which reviews characterizations of arithmetic groups in \(\mathrm{PSL}(2,\mathbb R)\) by Takeuchi and those in \(\mathrm{PSL}(2,\mathbb C)\) by Maclachlan and Reid, the author considers in Section 5 subgroups of arithmetic groups in \(\mathrm{PSL}(2,\mathbb C)^q\times\mathrm{PSL}(2,\mathbb R)^r\). The author determines the group with the smallest possible projective limit set. If \(\Gamma\) is a finitely generated subgroup of an irreducible arithmetic group in \(\mathrm{PSL}(2,\mathbb C)^q\times\mathrm{PSL}(2,\mathbb R)^r\) with \(q+r\geq 2\), then \(P_\Gamma\) is a one point set if and only if \(\Gamma\) is a conjugate by an element in \(\mathrm{GL}(2,\mathbb C)^q\times\mathrm{ GL}(2,\mathbb R)^r\) of a subgroup of \(\{(\sigma_1(s),\ldots,\sigma_{q+r}(s)):s\in S\}\), where \(S\) is an arithmetic Fuchsian or Kleinian group and \(\sigma_j\) is either the identity or complex conjugate (Theorem 5.10 and Corollary 5.17). If \(r>0\) and \(p_j(\Gamma)\) is nonelementary for some \(j\), then under the same assumption, \(\mathcal L_\Gamma\) is embedded homeomorphically in a circle if and only if \(p_j(\Gamma)\) is contained in an arithmetic Fuchsian group. Moreover, \(\mathcal L_\Gamma\) is homeomorphic to a circle if and only if \(p_j(\Gamma)\) is a cofinite arithmetic Fuchsian group (Theorem 5.18 and Corollary 5.19).