Arithmetic distributions of convergents arising from Jacobi-Perron algorithm (Q2502995): Difference between revisions

For a real number \(x \in (0,1)\), let \(p_n/q_n\) denote the \(n\)th convergent of \(x\) arising from the regular continued fraction of \(x\). \textit{H. Jager} and \textit{P. Liardet} [Indag. Math. 50, No. 2, 181--197 (1988; Zbl 0655.10045)] showed that for almost every \(x\), for every integer \(m \geq 2\), the sequence of pairs \((p_n, q_n)\) is equidistributed over the set \(\{(p,q) \in ({\mathbb Z}/m{\mathbb Z})^2 : \langle p,q \rangle = {\mathbb Z}/m{\mathbb Z}\}\). Here, \(\langle p, q \rangle\) denotes the subgroup generated by \(p\) and \(q\). The result of \textit{H. Jager} and \textit{P. Liardet} [loc.cit.] was deduced by proving the ergodicity of a certain skew product of the dynamical system arising from iterations of the Gauss map with \(\text{SL}(2, {\mathbb Z}/m{\mathbb Z})\). In the present paper, the authors extend this result to higher dimensions by proving the corresponding result for the Jacobi-Perron algorithm. As with the one-dimensional case, the numerators and denominators of the convergents arising from this \(d\)-dimensional continued fraction algorithm equidistribute over the subset of points in \(({\mathbb Z}/m{\mathbb Z})^{d+1}\) for which the coordinates generate the group \({\mathbb Z}/m{\mathbb Z}\). As in the one-dimensional case, the proof depends on proving the ergodicity of a skew product of the Jacobi-Perron system with a finite matrix group.