On the minimal denominator problem in function fields (Q6908137)

Let \(\mathcal{R}=\mathbb{F}_q[x]\) the ring of polynomials over the finite field \(\mathbb{F}_q\) with \(q\) elements where \(q=p^r\) with \(p\) a prime and \(r\in \mathbb{N}\). Let \(\mathcal{K}\) be the field of fractions of \(\mathcal{R}\). Define an absolute value on \(\mathcal{K}\) by \(\bigg|\frac fg\bigg|=q^{\deg(f)-\deg(g)}\) where \(f,g\in \mathcal{R}\) and \(g\neq 0\). Then the completion of \(\mathcal{K}\) with respect to \(\mid \cdot \mid\) is \(\mathcal{K}_{\infty}=\{\underset{n=-N}{\overset{\infty}{\sum}}a_nx^{-n}: a_n\in \mathbb{F}_q\}\). On \(\mathcal{K}_{\infty}^n\) define the supremum norm as \(\parallel (v_1,v_2,\cdots,v_n)\parallel=\max_{i=1,2,\cdots,n}\mid v_i\mid\). The author studies the problem on minimal denominator in function fields, that is, the following quantity.\N\[\N\deg_{\min}(\boldsymbol{\alpha},\delta)=\min\bigg\{d: \text{ exists } \frac{\boldsymbol{P}}{Q}, \deg(Q)=d,\bigg|\bigg|\boldsymbol{\alpha}-\frac{\boldsymbol{P}}{Q}\bigg|\bigg|<\delta\bigg\},\N\]\Nwhere \(\boldsymbol{\alpha}=(\alpha_1,\cdots,\alpha_n)\in \mathcal{K}_{\infty}^n\) and \(\boldsymbol{P}=(P_1,\cdots,P_n)\in \mathcal{R}^n\). We say that \(Q\) is a minimal denominator for \(\boldsymbol{\alpha}\) if \(\deg(Q)=\deg_{\min}(\boldsymbol{\alpha},\delta)\) and \(\bigg|\bigg| \boldsymbol{\alpha}-\frac{\boldsymbol{P}}{Q}\bigg|\bigg|<\delta\). Let \(\boldsymbol{\mathfrak{m}}=\{\alpha\in \mathcal{K}_{\infty}: \mid \alpha\mid\leq q^{-1}\}\). One can write any \(\alpha\in \mathcal{K}_{\infty}\) as \(\alpha=[\alpha]+\{\alpha\}\) where \([\alpha]\in \mathcal{R}\) and \(\{\alpha\}\in \boldsymbol{\mathfrak{m}}\). Similarly for \(\boldsymbol{\alpha}=(\alpha_1,\cdots,\alpha_n)\in \mathcal{K}_{\infty}^n\) one can write \(\boldsymbol{\alpha}=[\boldsymbol{\alpha}]+\{\boldsymbol{\alpha}\}\) where \([\boldsymbol{\alpha}]=(\alpha_1,\cdots,\alpha_n)\) and \(\{\boldsymbol{\alpha}\}=(\{\alpha_1\},\cdots,\{\alpha_n\})\). If \(Q\) is a minimal denominator for \(\boldsymbol{\alpha}\) then it is also a minimal denominator for \(\{\boldsymbol{\alpha}\}\). So it suffices to assume that \(\boldsymbol{\alpha}\in \boldsymbol{\mathfrak{m}}^n\).\N\NThe author proves the theorem that for \(n=1\) the probability\N\[\N\mathbb{P}(\deg_{\min}(\alpha,q^{-1})=d)=\begin{cases} \frac 1q & \text{ if }d=0,\\\N\frac{q-1}{q} & \text{ if }d=1, \end{cases}\N\]\Nand for every \(k\geq2\)\N\[\N\mathbb{P}(\deg_{\min}(\alpha,q^{-k})=d)=\begin{cases} q^{-k} & \text{ if }d=0,\\\N\frac{q-1}{q^{k-2d+1}} & \text{ if } d\leq \lceil \frac k2\rceil, d\in \mathbb{N},\\\N0 & \text{ else.} \end{cases}\N\]\NThe author also gives a precise value for \(\mathbb{E}[\deg_{\min}(\bullet,q^{-k})]\).\N\NThe author shows that for every \(\boldsymbol{\alpha}\in \boldsymbol{\mathfrak{m}}^n\) and for every \(k\geq 1\), there exists a unique monic polynomial \(Q\in\mathcal{R}\), such that \(\deg(Q)=\deg_{\min}(\boldsymbol{\alpha},q^{-k})\) and \(\parallel Q\boldsymbol{\alpha}\parallel <q^{-k}\). This unique monic polynomial is denoted by \(Q_{\min}(\boldsymbol{\alpha},q^{-k})\). For \(n=1\), then the author calculates for a monic polynomial \(Q\) with \(\deg(Q)\leq \lceil \frac k2\rceil\) the probability of \(Q_{\min}(\alpha,q^{-k})=Q\) for every \(k\geq 1\). In particular, the author shows that if \(Q\) is an irreducible monic polynomial of degree \(d\), then \(\mathbb{P}(Q_{\min}(\alpha,q^{-k})=Q)=\frac{q^d-1}{q^k}\). In higher dimensions for \(n\in \mathbb{N}\), for \(Q\in \mathcal{R}\), the author also shows that \(Q_{\min}(\boldsymbol{\alpha},q^{-k})=Q\) if and only if for every \(1\leq i\leq n,\ Q_{\min}(\alpha_i,q^{-k})\) divides \(Q\) and there exists \(1\leq i\leq n\) such that \(Q_{\min}(\alpha_i,q^{-k})=Q\). Further the author also shows that for \(n,k\in \mathbb{N}, \boldsymbol{\alpha}\in \boldsymbol{\mathfrak{m}}^n\), \(\deg_{\min}(\boldsymbol{\alpha},q^{-k})\leq \frac{nk}{n+1}\). Moreover for every \(d\leq \frac{nk}{n+1}\), the author gives an upper bound for the probability:\N\[\N\mathbb{P}(\deg_{\min}(\boldsymbol{\alpha},q^{-k})\leq d)\leq q^{-kn-(n+1)d}.\N\]\NThen the author discusses a \(P\)-adic invariant of the minimal denominator problem, which is motivated by the \(P\)-adic Littlewood conjecture. Let \(P\in \mathcal{R}\) be an irreducible polynomial and \(\alpha\in \boldsymbol{\mathfrak{m}}\) and let \(k\geq 1\). Define \(\deg_{\min, P}(\alpha,q^{-k})=\inf_{m\geq 0}\deg_{\min}(P^m\alpha,q^{-k})\). The author shows the following result: For every irreducible \(P\in \mathcal{R}\) and for \(1\leq d\leq \lceil\frac k2\rceil\),\N\[\N\dim_{H}\{\alpha\in \mathfrak{m}: \deg_{\min,P}(\alpha,q^{-k})\geq d\}=\frac{\log_q(\mid P\mid^k-\mid P\mid^{2(d-1)})}{k\deg(P)}\N\]\Nwhere \(\dim_H(E)\) is the Hausdorff dimension of \(E\subseteq \mathcal{K}_{\infty}^n\).