Counting intrinsic Diophantine approximations in simple algebraic groups (Q2677810): Difference between revisions

Let \(G\subset\mathrm{GL}\) be a simply-connected \(\mathbb{Q}\)-almost simple linear algebraic group defined over \(\mathbb{Q}\) and let \(S\) be a (finite or infinite) set of primes. Suppose that for at least one \(p \in S\), \(G(\mathbb{Q}_p)\) is non-compact. This implies that \(G(\mathbb{Z}[S^{-1}])\) is dense in \(G(\mathbb{R})\). To quantify this denseness, define the height function of \(r \in\mathrm{Mat}_N(\mathbb{Q})\), \[ H_f(r) = \prod_{p \text{ a prime }} \max(1, \Vert r \Vert_p), \] where \(\Vert r \Vert_p\) is the maximal \(p\)-adic norm among the entries of \(r\), and for \(x \in G(\mathbb{R})\) and \(\delta > 0\) let \[ \omega_S(x,\delta) = \min\{H_f(r) : \rho(x,r) \le \delta \text{ and } r \in G(\mathbb{Z}[S^{-1}])\}, \] where \(\rho\) is the left-invariant Riemannian metric on \(G(\mathbb{R})\). In previous work [Int. Math. Res. Not. 2013, No. 21, 5002--5058 (2013; Zbl 1370.11077)], the authors established the existence of a constant \(\kappa > 0\), such that for any \(x \in G(\mathbb{R})\) and any \(\delta\) small enough, \(\omega_S(x,\delta) \le \delta^{-\kappa}\). In the present manuscript, the authors obtain a quantitative generalisation of the above estimate. Concretely, let \(G_S\) denote the restricted direct product of the groups \(G(\mathbb{Q}_p)\) over \(p \in S\), let \(m_p\) denote the normalised Haar measure on \(G(\mathbb{Q}_p)\), and let \(m_S\) denote the product measure on \(G_S\). Extend the height \(H_f\) to \(G_S\) and let \(B_S(h) = \{g \in G_S : H_f(g) \le h\}\). For the infinite place, let \(G_\infty =G(\mathbb{R})\) and let \(m_\infty\) denote the Haar measure, normalised so that \(G(\mathbb{Z}[S^{-1}])\) has co-volume one in \(G_\infty \times G_S\). Finally, for \(h > \delta^{-\kappa}\), let \[ N_S(x,\delta, h) = \vert \{r \in G(\mathbb{Z}[S^{-1}]) : \rho(x,r) \le \delta\text{ and }H_f(r) \le h \}\vert. \] It is shown that for every \(x \in G_\infty\), there exist \(\theta, c_1, c_2(x) > 0\) such that \[ N_S(x,\delta, h) = m_\infty(B(x,\delta)) m_S(B_S(h))(1+ O_S(\delta^{-d/(d+1)}m_S(B_S(h))^{-\theta d/(d+1)})), \] where \(d=\dim(G)\), \(\delta \in [c_1 m_S(B_S(h))^{-\theta}, c_2(x))\) and \(B(x,\delta)\) denotes the Riemannian ball with centre \(x\) and radius \(\delta\). Furthermore, \(c_2(x)\) is uniform for \(x\) in bounded sets.