On homogeneous and inhomogeneous Diophantine approximation over the fields of formal power series (Q2305389)

The authors prove analogues of several classical and recent results in Diophantine approximation in the field of formal Laurent series with coefficients from a finite field. In this setting, one considers the polynomial ring \(\mathbb{F}[X]\) over a finite field \(\mathbb{F}[X]\) as an analogue of the integers, the field of rational functions \(\mathbb{F}(X)\) as an analogue of the rational numbers and obtain the field of formal Laurent series \(\mathbb{F}((X^{-1}))\) by completing \(\mathbb{F}(X)\) with respect to the absolute value induced by \(\Vert P \Vert = \vert \mathbb{F} \vert^{\deg P}\) for \(P \in \mathbb{F}[X]\). Thus, one obtains a locally compact and complete field of positive characteristic, in which it is natural to study the approximation properties of general elements by rational functions. In higher dimensional vector spaces over \(\mathbb{F}((X^{-1}))\), define the norm \(\Vert y \Vert\) to be the maximum of the absolute values of the coordinates of \(y\). Within this setting, the authors first obtain an analogue of Kronecker's Theorem in higher dimensions. Subsequently, for \(A\) an \(n \times m\) matrix over \(\mathbb{F}((X^{-1}))\) and \(\theta \in \mathbb{F}((X^{-1}))^n\), two Diophantine exponents are defined. For \(H > 1\), consider the inequalities \[ \vert \langle Ax - \theta \rangle \vert \le H^{-\omega}, \quad \Vert x \Vert \le H, \] where \(\vert \langle y \rangle \vert\) denotes the distance from the vector \(y\) to the nearest vector in \(\mathbb{F}[X]^n\) in the metric induced by the norm \(\Vert \cdot \Vert\). The exponent \(\omega(A, \theta)\) is the \textit{supremum} of all \(\omega > 0\) such that this system of inequalities has solutions \(x \in \mathbb{F}[X]^m\) for arbitrarily large values of \(H\). Similarly, the exponent \(\hat{\omega}(A, \theta)\) is the \textit{supremum} of all \(\omega > 0\) such that this system of inequalities has solutions \(x \in \mathbb{F}[X]^m\) for all sufficiently large \(H\). In analogy with the results of \textit{Y. Bugeaud} and \textit{M. Laurent} [Mosc. Math. J. 5, No. 4, 747--766 (2005; Zbl 1119.11039)], it is shown that for a fixed matrix \(A\), \[ \omega(A, \theta) \ge\frac{1}{\hat{\omega}(A, 0)}, \quad \hat{\omega}(A, \theta) \ge\frac{1}{\omega(A, 0)}, \] with equality for almost all \(\theta\) with respect to the Haar measure. Furthermore, for \(\theta \notin A \mathbb{F}[X]^m + \mathbb{F}[X]^n\), \(\hat{\omega}(A, \theta) \le \omega(A, 0)\). Finally, for a fixed matrix \(A\) and \(\varepsilon > 0\), the set \[ \mathrm{Bad}^{\varepsilon} (A) = \bigg \{\theta \in \mathbb{F}((X^{-1}))^n : \Vert \theta \Vert < 1, \liminf_{x \in \mathbb{F}[X]^m , \Vert x \Vert \rightarrow \infty} \Vert x \Vert^{m/n} \vert \langle Ax -\theta \rangle \vert \ge \varepsilon \bigg\} \] is studied. Lower bounds for the Hausdorff dimension of these sets are obtained, and in analogy with recent results of the first author et al. [``Hausdorff dimension in inhomogeneous Diophantine approximation'', Int. Math. Res. Not. (to appear)], it is shown that under certain special conditions on \(A\), these sets have maximal dimension for some strictly positive values of \(\varepsilon\). In the case \(m=n=1\), these conditions are necessary and sufficient.