Dirichlet's theorem on Diophantine approximation and homogeneous flows (Q928467)

The authors consider generalisations of Dirichlet's theorem in the setting of systems of linear forms. If \(Y\) denotes a real \(m \times n\) matrix and \({\mathcal T}\) is an unbounded set of vectors \((t_1, \dots, t_{m+n}) \in \mathbb R_+^{m+n}\) with \(\sum_{i=1}^m t_i = \sum_{j=1}^{n} t_{m+j}\), Dirichlet's theorem can be \(\varepsilon\)-improved for \(Y\) along \({\mathcal T}\) if the system of inequalities \[ | Y_i {\mathbf q} - p_i| < \varepsilon e^{-t_i} \quad (i=1, \dots, m), \qquad | q_j | < \varepsilon e^{t_j} \quad (j=1, \dots, n) \] has a solution \({\mathbf q} \in \mathbb Z^n \setminus \{0\}\), \({\mathbf p} \in \mathbb Z^m\) for every sufficiently large \({\mathbf t} \in {\mathcal T}\). The problem considered is to prove that the set of points for which this is the case is small. The usual form of the problem, which goes back to \textit{H. Davenport} and \textit{W. M. Schmidt} [Sympos. Math., Roma 4, 113--132 (1970; Zbl 0226.10032)] comes about when \({\mathcal T}\) consists of all vectors \((t/m, \dots, t/m, t/n, \dots, t/n)\) with \(t \in \mathbb R_+\). It is shown that for \({\mathcal T}\) satisfying a certain technical condition and for any \(\varepsilon < 1\), the set of matrices \(Y\) for which Dirichlet's Theorem can be \(\varepsilon\)-improved for \(Y\) along \({\mathcal T}\) is of Lebesgue measure zero. Additionally it is shown that in the case of a single linear form (\(m=1\)), this is also the case when we replace Lebesgue measure by a member of a larger class of measures and \(\varepsilon\) is sufficiently small. In the latter case, no technical condition is needed on \({\mathcal T}\). The class of measures includes the natural measures on suitably regular smooth manifolds as well as certain fractal sets. The proofs of the two results rely on a translation of the problem into a problem of studying equidistribution and quantitative non-divergence for an action on the homogeneous space \(\text{SL}_{m+n}(\mathbb R)/\text{SL}_{m+n}(\mathbb Z)\). For the proof of the first theorem, the series of ideas date back to \textit{G. A. Margulis} [On some aspects of the theory of Anosov systems. Berlin: Springer (2004; Zbl 1140.37010)]. The proof of the second theorem stems from the developments following the seminal paper of \textit{D. Y. Kleinbock} and \textit{G. A. Margulis} [Ann. Math. (2) 148, No. 1, 339--360 (1998; Zbl 0922.11061)]. The paper is concluded with a nice section on applications of the main results.