Multi-dimensional metric approximation by primitive points (Q2339682): Difference between revisions

The goal of this article is to refine some classical results on \textit{metrical Diophantine approximation} for systems of linear or affine inequalities, with additional constraints on the solutions involving coprimality by blocks. Definitions: For \(\mathbf v\in \mathbb R^d\), let us denote by \(|\mathbf v|\) the \textit{supremum norm} in \(\mathbb R^d\). When \(\mathbf q \in \mathbb Z^m \subset \mathbb R^m\) and \(\mathbf p \in \mathbb Z^n \subset \mathbb R^n\), the vector \(\left(\begin{smallmatrix} \mathbf q\\ \mathbf p\end{smallmatrix} \right)\) in \(\mathbb R^{m+n}\) is conventionally written as \((\mathbf q,\mathbf p)^t\) (with \(t\) for transpose vector notation). For \(m, n \in \mathbb N\) let us denote by Mat\(_{n,m}(\mathbb R)\) the vector space of \(n \times m\) matrices with real entries. A matrix \(X \in\) Mat\(_{n,m+n}(\mathbb R)\) will be expressed in the form \(X = (\Theta,\Phi)\) with \(\Theta\in \) Mat\(_{n,m}(\mathbb R)\) and \(\Phi \in \) Mat\(_{n,n}(\mathbb R)\). For any subset \(\sigma = \{i_1,\dots i_\nu\} \) of \(\{1,\dots, d\}\) having \(\nu\geq 2\) elements, let us denote by \(P(\sigma)\) the set of integer points \(\mathbf v =(v_1, \dots , v_d )^t\) such that \(\gcd(v_{i_1}, \dots, v_{i_\nu} ) = 1\). Let \(\pi\) be a partition of \(\{1,\dots, d\}\) formed by a family of subsets \(\pi_j\), each one having at least two elements. Then, let us denote by \(P(\pi)\) the set of integer points \(\mathbf v \in \mathbb Z^d\) such that \(\mathbf v \in P(\pi_ j )\) for all components \(\pi_j\) . {Results:} The authors prove the following refinement of \textit{J. W. S. Cassels} [An introduction to Diophantine approximation. Cambridge: At the University Press (1957; Zbl 0077.04801)] and \textit{V. G. Sprindzhuk} [Metric theory of Diophantine approximations. New York etc.: John Wiley \& Sons (1979; Zbl 0482.10047)]: Theorem. Let \(n,m \in \mathbb N\) and \(\pi\) be a partition of \(\{1, \ldots ,m+n\}\) such that every component of \(\pi\) has at least \(n + 1\) elements. Let \(\psi : \mathbb N \rightarrow (0,\infty)\) be a function such that the mapping \(x \rightarrow x^{m-1}\psi(x)^n\) is non-increasing. If \(\sum_{j\geq 1}j^{m-1}\psi ( j )^n =\infty\), then for almost every pair \((\Theta,\mathbf y) \in\) Mat\(_{n,m}(\mathbb R) \times \mathbb R^n\), there exist infinitely many points \((\mathbf q,\mathbf p)^t \in P(\pi)\) such that \[ |\Theta\mathbf q +\mathbf p -\mathbf y| \leq \psi(|\mathbf q|). \tag{1} \] Conversely, if the series \(\sum_{j\geq 1}j^{m-1}\psi( j )^n\) converges, then for almost every pair \((\Theta,\mathbf y) \in\) Mat\(_{n,m}(\mathbb R)\times\mathbb R^n\), there exist only finitely many integer points \((\mathbf q,\mathbf p)^t \in \mathbb Z^{m+n}\) for which (1) holds. The authors prove a refinement of the Khintchine-Groshev theorem (see for instance in \textit{V. Beresnevich} and \textit{S. Velani} [Int. Math. Res. Not. 2010, No. 1, 69--86 (2010; Zbl 1241.11086)]) for the homogeneous case \(\mathbf y=0\): Theorem. Let \(n,m, \pi\) and \(\psi\) as before. If \(\sum_{j\geq 1} j^{m-1}\psi( j )^n =\infty \) then for almost every \(\Theta\in \) Mat\(_{n,m}(\mathbb R)\), there exist infinitely many points \((\mathbf q,\mathbf p)^t \in P(\pi)\) such that \[ |\Theta\mathbf q +\mathbf p| \leq \psi (|q|). \tag{2} \] Conversely, if \(\sum_{j\geq 1} j^{m-1}\psi( j )^n < \infty\) then for almost every \(\Theta \in \) Mat\(_{n,m}(\mathbb R)\), there exist only finitely many \((\mathbf q,\mathbf p)^t\in \mathbb Z^{m+n}\) for which (2) holds. For general systems with \textit{free coefficients}, they obtain: Theorem. Let \(n,m, p\) and \(\psi\) as before. If \(\sum_{j\geq 1} j^{m-1}\psi( j )^n = \infty\) then for every \(\mathbf y \in \mathbb R^n\) and for almost every matrix \(X = (\Theta, \Phi) \in\) Mat\(_{n,m+n}(\mathbb R)\), there exist infinitely many points \((\mathbf q,\mathbf p)^t \in P(\pi)\) such that \[ |\Theta\mathbf q +\Phi\mathbf p -\mathbf y| \leq \psi(|\mathbf q|).\tag{3} \] Conversely, if \(\sum_{j\geq 1} j^{m-1}\psi( j )^n < \infty\) then for almost every \(X = (\Theta,\Phi) \in\) Mat\(_{n,m+n}(\mathbb R)\), there exist only finitely many \((\mathbf q,\mathbf p)^t \in \mathbb Z^{m+n}\) for which (3) holds.