Limit sets of cyclic quaternionic Kleinian groups (Q6163398)

Kleinian groups are discrete subgroups of \(\mathrm{SL}(2,\mathbb{C})\) acting on the Riemann sphere which is the boundary of the three-dimensional hyperbolic space. A Kleinian group acts properly discontinuously on the complement of its limit set. A complex Kleinian group means a discrete subgroup of \(\mathrm{PSL}(3,\mathbb{C})\) with a non-empty domain of discontinuity. The aim of the paper under review is to begin an investigation of quaternionic Kleinian groups, that is to say, discrete subgroups of \(\mathrm{PSL}(3,\mathbb{H})\) with a non-empty domain of discontinuity, where \(\mathbb{H}\) denotes the ring of Hamilton's quaternions. A first step for this study is to obtain a good understanding of the limit sets of cyclic subgroups of \(\mathrm{PSL}(3,\mathbb{H})\). In the paper, the Kulkarni limit sets of these groups are calculated. Let \(G\) be a subgroup of \(\mathrm{SL}(3, \mathbb{H})\) and let us consider its natural action on the two-dimensional quaternionic projective space \(X=\mathbb{P}_{\mathbb{H}}^2\). Consider the following three sets \(L_0(G):=\) the closure of the set of points of \(X\) which have an infinite isotropy group, \(L_1(G):=\) the closure of the cluster points of orbits of points in \(X \backslash L_0(G)\), and \(L_2(G):=\) the closure of the cluster points of \(\{g(K)\}_{g \in G}\), where \(K\) runs over all the compact subsets of \(X \backslash\left\{L_0(G) \cup L_1(G)\right\}\). The Kulkarni limit set of \(G\) is defined as \( \Lambda(G):=L_0(G) \cup L_1(G) \cup L_2(G) \) and the Kulkarni domain of discontinuity of \(G\) is \(\Omega(G)=X \backslash \Lambda(G)\). Let \(G\) be a cyclic quaternion Klein group with generator \(\tilde{g}\). Let \(g\) be its lift in \(\mathrm{SL}(3,\mathbb{H})\). The authors present their results in Tables 1, 2 and 3, according to whether the element \(\tilde{g}\) is elliptic, loxodromic or parabolic, respectively. These classes are divided in several subclasses according to the eigenvalues of the matrix \(g\). Each table gives \(L_0(G)\), \(L_1(G)\), \(L_2(G)\) and \(\Lambda(G)\). As an example, we have here Table 3: \bigskip \( \begin{array}{lllll} \hline \text { Kulkarni Set } & \mathrm{L}_0(\mathrm{G}) & \mathrm{L}_1(\mathrm{G}) & \mathrm{L}_2(\mathrm{G}) & \Lambda(\mathrm{G}) \\ \text { Vertical translation } & \mathbb{L}\left\{e_1, e_3\right\} & \left\{e_1\right\} & \left\{e_1\right\} & \mathbb{L}\left\{e_1, e_3\right\} \\ \text { Non-vertical translation } & \left\{e_1\right\} & \left\{e_1\right\} & \mathbb{L}\left\{e_1, e_2\right\} & \mathbb{L}\left\{e_1, e_2\right\} \\ \text { Rational ellipto-parabolic } & \mathbb{L}\left\{e_1, e_3\right\} & \left\{e_1\right\} & \left\{e_1\right\} & \mathbb{L}\left\{e_1, e_3\right\} \\ \text { Irrational ellipto-parabolic } & \left\{e_1, e_3\right\} & \mathbb{L}\left\{e_1, e_3\right\} & \left\{e_1\right\} & \mathbb{L}\left\{e_1, e_3\right\} \\ \text { Ellipto-translation } & \left\{e_1\right\} & \left\{e_1\right\} & \mathbb{L}\left\{e_1, e_2\right\} & \mathbb{L}\left\{e_1, e_2\right\} \\ \hline \\ \end{array} \) where \(\{e_1, e_2, e_3\}\subset \mathbb{P}_\mathbb{H}^2\) is the projection of the standard basis \(\{\mathbf{e_1}, \mathbf{e_2}, \mathbf{e_3}\}\) de \(\mathbb{H}^3\) in \(\mathbb{P}_\mathbb{H}^2\), and \(\mathbb{L}\{p,q\}\) is the quaternionic projective line passing through \(p, q \in \mathbb{P}_\mathbb{H}^2\). The results in Tables 1--3 are obtained in Sections 3--5 in Theorems 3.1, 4.1 and 5.1, respectively. Section 1 provides a very clear introduction. Section 2 deals with Jordan forms in \(M(n, \mathbb{H})\), the quaternion projective space, projective and pseudo-projective transformations, and some other preliminaries. In the last Section 6, the authors study the extended Conze-Guivarc'h limit set. Let \(G \subset \mathrm{PSL}(3, \mathbb{H})\) act on the dual space \((\mathbb{P}_{\mathbb{H}}^2)^*\). A point \(q\in (\mathbb{P}_{\mathbb{H}}^2)^*\) is a limit point of \(G\) if there exists an open subset \(U\subset (\mathbb{P}_{\mathbb{H}}^2)^*\) and a sequence \(\{g_n\} \subset G\) of distinct elements such that for every \(p\in U\), \(\lim_{n\rightarrow \infty} g_n\cdot p=q\). The set of limit points is called the extended Conze-Guivarc'h limit set and is denoted \(\hat{L}(G)\). For the cyclic subgroups of \(\mathrm{PSL}(3,\mathbb{H})\), the limit sets \(\hat{L}(G)\) are calculated and the results are presented in Table 4.