Applications of Siegel's lemma to a system of linear forms and its minimal points (Q2165896)

Consider a real matrix \(\Theta\) consisting of rows \((\theta_{i,1},\ldots , \theta_{i,n})\) for \(1 \le i \le m\). The problem of making the system of linear forms \(x_1 \theta_{i,1} + \ldots + x_n \theta_{i,n} - y_i\) small for integers \(x_j\), \(y_i\) classically leads to consider an ordinary and a uniform exponent of approximation, denoted by \(w(\Theta)\) and \(\widehat{w}(\Theta)\) respectively; both depend implicitly on \(m\) and \(n\). They satisfy \(\frac{n}{m}\le \widehat{w}(\Theta)\le w(\Theta)\) by Minkowski's convex body theorem. For \(m =1\) or \(n=1\), a sharp lower bound for the ratio \(w(\Theta)/\widehat{w}(\Theta)\) was recently established by \textit{A. Marnat} and \textit{N. G. Moshchevitin} [Mathematika 66, No. 3, 818--854 (2020; Zbl 1503.11100)] (Theorem 2.1 of this paper): in the case \(m=1\) and \(n\ge 2\), we have \(w(\Theta)/\widehat{w}(\Theta)\ge G_{1,n}\) where \(G_{1,n}\) is the unique positive real root of \(1-\widehat{w}(\Theta)+\sum_{j=1}^{n-1} x^j\). The author gives a short proof (Theorem 3.3) of this result when \(m=1\) under an additional assumption on the best approximation integer vectors associated to \(\Theta\), which is too technical to be written here (Definition 3.1). His bound applies to any \(m \ge 2\) but, as the author mentions, it is probably not optimal in this case. This result complements a similar conditional result of Moshchevitin, who imposed a different assumption on the best approximations. The author's assumption is satisfied in particular for \(m = 1\), \(n = 2\) and this enables him to recover a previous theorem of Jarnik (Theorem 3.9). He also proves a criteria concerning the number of consecutive best approximation vectors which are linearly independent (Theorem 4.1). The main tool in the proofs is a variant due to Davenport and Schmidt of Siegel's classical lemma on the size of integral solutions of linear systems with integer coefficients.