Minimal Denominators Lying in Subsets of the Ring of Polynomials over a Finite Field

Basic Information

• Paper ID: 2510.07787
• Title: Minimal Denominators Lying in Subsets of the Ring of Polynomials over a Finite Field
• Author: Noy Soffer Aranov (Graz University of Technology)
• Classification: math.NT (Number Theory), cs.NA (Numerical Analysis), math.NA, math.PR (Probability)
• Publication Date: November 12, 2025 (arXiv v2)
• Paper Link: https://arxiv.org/abs/2510.07787

Abstract

This paper investigates the minimal denominator problem in subsets of the polynomial ring over finite fields. Given a subset $\mathcal{S}\subseteq \mathbb{F}_q[x]$ and fixed $n,m\in \mathbb{N}$, the author studies two related problems: (1) Continuous case: the distribution of minimal denominators $Q\in \mathcal{S}$ such that $\frac{P}{Q}$ lies in a ball $B(\alpha, q^{-n})$ for some $\mathbf{P}\in \mathbb{F}_q[x]^m$; (2) Discrete case: the corresponding discrete probability distribution when $N\in \mathbb{F}_q[x]$ satisfies $\deg(N)=n$. The paper proves that for any infinite subset $\mathcal{S}$, any $n$, and any dimension $m$, the probability distributions of these two random variables are exactly equal. This is significantly stronger than results in the real case (Balazard and Martin only proved asymptotic equality of means).

Research Background and Motivation

Problem Background

1. Classical Problem Origin: Kruyswijk and Meijer KM77 studied the function defined on integers: $q_j(N) = \min\left\{q \in \mathbb{N} : \exists p \in \mathbb{Z} : \frac{p}{q} \in \left[\frac{j}{N}, \frac{j+1}{N}\right)\right\}$ which returns the minimal denominator of a rational number falling in a given interval. They proved $S(N) = \sum_{j=0}^{N-1} q_j(N) \asymp N^{3/2}$ and conjectured $S(N) \sim \frac{16}{\pi^2}N^{3/2}$.
2. Continuous Analogue: Meiss and Sanders SM20 proposed a continuous version: $q_{\min}(x, \delta) = \min\left\{q \in \mathbb{N} : \exists \frac{p}{q} \in \left(x-\frac{\delta}{2}, x+\frac{\delta}{2}\right), \gcd(p,q)=1\right\}$
3. Important Progress: Balazard and Martin BM23 proved the Kruyswijk-Meijer conjecture and established a connection between discrete and continuous cases: $S(N) - N\int_0^1 q_{\min}\left(x, \frac{1}{N}\right)dx \ll N^{4/3}\ln^2 N$

Research Motivation

1. Special Properties of Function Fields: Function fields possess the ultrametric property, which makes ball intersection behavior more regular (either disjoint or equal), potentially leading to more precise results.
2. Restricted Denominator Sets: Existing research focuses mainly on all denominators; this paper considers denominators restricted to specific subsets $\mathcal{S}$, which has important applications in number theory.
3. High-Dimensional Generalization: Extending one-dimensional results to arbitrary dimensions $m$, which is significant in Diophantine approximation theory.
4. Exact Equality vs. Asymptotic Relations: The real case only yields asymptotic relations, while function field structure may allow exact distributional equalities.

Core Contributions

1. Main Theorem (Theorem 1.3): Proves that for any infinite set $\mathcal{S} \subseteq \mathbb{F}_q[x]_{\text{monic}}$, any dimension $m$, any $n\in\mathbb{N}$, and any $N\in R_{=n}$, the probability distributions in discrete and continuous cases are exactly equal: $\mathbb{P}(a \in R^m_{<n} : d_{N,\mathcal{S}}(a) = k) = \nu_m(\alpha \in \mathfrak{m}^m : \deg_{\min,\mathcal{S}}(\alpha, q^{-n}) = k)$
2. Denominator Distribution Theorem (Theorem 1.7): Not only are the degree distributions equal, but the distributions of actual minimal denominator polynomials are also exactly equal: $\mathbb{P}(a \in R^m_{<n} : Q_{N,\mathcal{S}}(a) = Q) = \nu(\alpha \in \mathfrak{m}^m : Q_{\min,\mathcal{S}}(\alpha, q^{-n}) = Q)$
3. Exact Distribution Formula (Corollary 1.5): Provides exact probability formulas for the unrestricted denominator case. For example, when $n\geq 2$: $\mathbb{P}(a \in R_{<n} : d_{N,R_{\text{monic}}}(a) = k) = \begin{cases} q^{-n} & k=0 \\ \frac{q-1}{q^{n-2k+1}} & k \leq \lceil\frac{n}{2}\rceil \\ 0 & \text{else} \end{cases}$
4. Hadamard Gap Sequences (Theorem 1.10): For the multiplicative semigroup $\mathcal{S} = \{P^d : d\geq 0\}$, provides complete distribution formulas, which relate to base-$P$ expansions of Laurent series.
5. Zero Error Result: Unlike the $O(N^{4/3}\ln^2 N)$ error in the real case, the function field case achieves zero error, representing a fundamental improvement.

Detailed Methodology

Problem Definition

Continuous Case: Given $\alpha \in K_\infty^m$ (the $m$-dimensional Laurent series space) and $n\in\mathbb{N}$, define $\deg_{\min,\mathcal{S}}(\alpha, q^{-n}) = \min\left\{d : \exists \frac{P}{Q} \in K^m, (P,Q)\in\hat{R}^{m+1}, Q\in\mathcal{S}\cap R_{=d}, \left\|\alpha - \frac{P}{Q}\right\| < \frac{1}{q^n}\right\}$

Discrete Case: Given $N\in R_{=n}$ and $a\in R^m_{<n}$, define $d_{N,\mathcal{S}}(a) = \deg_{\min,\mathcal{S}}\left(\frac{a}{N}, \frac{1}{|N|}\right)$

where:

• $R = \mathbb{F}_q[x]$ is the polynomial ring over a finite field
• $\hat{R}^{m+1}$ denotes primitive vectors (i.e., vectors with gcd equal to 1)
• $\|\cdot\|$ is the maximum norm: $\|v\| = \max_i |v_i|$
• $|f| = q^{\deg(f)}$ is the absolute value

Core Technical Framework

1. Farey Fractions and Ball Coverings

Define $\mathcal{S}$-Farey fractions: $\mathcal{F}^m_{k,\mathcal{S}} = \left\{\frac{P}{Q} \in K : (P,Q)\in\hat{R}^{m+1}, \|P\| < |Q| \leq q^k, Q\in\mathcal{S}\right\}$

Key quantity $f_{m,n,\mathcal{S}}(k)$: represents the number of disjoint balls in the union $\bigcup_{P/Q\in\mathcal{F}^m_{k,\mathcal{S}}} B(P/Q, q^{-(n+1)})$.

2. Application of Ultrametric Property

Lemma 1.1 (Key Lemma): For $\alpha_1, \alpha_2 \in K_\infty^m$ and $r>0$:

• If $\|\alpha_1 - \alpha_2\| > r$, then $B(\alpha_1, r) \cap B(\alpha_2, r) = \emptyset$
• If $\|\alpha_1 - \alpha_2\| \leq r$, then $B(\alpha_1, r) = B(\alpha_2, r)$

This property is the essential difference between function fields and real fields, making ball intersection behavior completely determined.

3. Separation Condition

Definition 3.2: A Farey fraction $\frac{P}{Q} \in \mathcal{F}^m_{k,\mathcal{S}}$ is called $(\mathcal{S}, n)$-separated if for all $\frac{A}{B} \in \mathcal{F}^m_{\deg(Q),\mathcal{S}} \setminus \{\frac{P}{Q}\}$, $\left\|\frac{P}{Q} - \frac{A}{B}\right\| \geq q^{-n}$

This concept is analogous to Marklof's Mar13, Mar24 study of "gaps" in Farey fractions.

Proof Strategy

Core of Theorem 1.3 Proof

Step 1: Establish volume formula for continuous case $\nu_m(\alpha \in \mathfrak{m}^m : \deg_{\min,\mathcal{S}}(\alpha, q^{-n}) > k) = 1 - \frac{f_{m,n,\mathcal{S}}(k)}{q^{mn}}$

Step 2: Establish counting formula for discrete case $\#\{a\in R^m_{<n} : d_{N,\mathcal{S}}(a) > k\} = q^{mn} - f_{m,n,\mathcal{S}}(k)$

Step 3: Use ultrametric property to prove ball correspondence

• $B(P/Q, q^{-(n+1)}) \cap B(A/B, q^{-(n+1)}) = \emptyset$ if and only if
• $B(NP/Q, q^{-1}) \cap B(NA/B, q^{-1}) = \emptyset$

This ensures that the number of disjoint balls is the same in both cases.

Step 4: Obtain probability equality by differencing $\mathbb{P}(d_{N,\mathcal{S}}(a) = k) = \frac{f_{m,n,\mathcal{S}}(k) - f_{m,n,\mathcal{S}}(k-1)}{q^{mn}} = \nu_m(\deg_{\min,\mathcal{S}}(\alpha, q^{-n}) = k)$

Core of Theorem 1.7 Proof

Use Lemma 3.1 to establish: $Q_{\min,\mathcal{S}}(\alpha, q^{-n}) = Q$ is equivalent to the existence of a $(\mathcal{S}, n)$-separated $P/Q$.

Define $f_{m,n,\mathcal{S},Q}(k)$ as the number of disjoint balls corresponding to all $(\mathcal{S}, n)$-separated Farey fractions with denominator $Q$, then complete the proof similarly to Theorem 1.3.

Experimental Setup

This is a pure mathematics theory paper with no numerical experiments or datasets. All results are rigorous mathematical proofs.

Theoretical Verification

The paper verifies theoretical correctness through:

1. Special Case Verification: Corollary 1.5 provides exact formulas for $\mathcal{S} = R_{\text{monic}}$, which can be cross-checked with results from Ara26.
2. Consistency Checks: Verifies that probabilities sum to 1, for example: $\sum_{k=0}^{n} \mathbb{P}(d_{N,\mathcal{S}}(a) = k) = 1$
3. Boundary Cases:
   • When $k \leq n/2$, Corollary 2.1 gives $f_{m,n,\mathcal{S}}(k) = \#\mathcal{F}^m_{k,\mathcal{S}}$
   • When $1\in\mathcal{S}$, $\mathbb{P}(d_{N,\mathcal{S}}(a)=0) = q^{-mn}$

Experimental Results

Main Theoretical Results

1. Complete Equivalence (Theorem 1.3)

For all $k = 0, 1, \ldots, m_{\mathcal{S}}(n)$: $\mathbb{P}(a \in R^m_{<n} : d_{N,\mathcal{S}}(a) = k) = \nu_m(\alpha \in \mathfrak{m}^m : \deg_{\min,\mathcal{S}}(\alpha, q^{-n}) = k)$

Significance: This is an exact equality rather than an asymptotic relation, with zero error.

2. Expected Value Equality (Corollary 1.4)

$\mathbb{E}[d_{N,\mathcal{S}}(a)] = \frac{1}{|N|}\sum_{a\in R^m} d_{N,\mathcal{S}}(a) = \int_{\mathfrak{m}^n} \deg_{\min,\mathcal{S}}(\alpha, |N|^{-1})d\nu_m(\alpha)$

This generalizes the Balazard-Martin result and reduces the error term from $O(N^{4/3}\ln^2 N)$ to zero.

3. Exact Distribution for Unrestricted Denominators (Corollary 1.5)

When $\mathcal{S} = R_{\text{monic}}$, $n\geq 2$, $m=1$: