Abstract:This paper studies the logarithmic moments of the smallest denominator of all rationals in a shrinking interval with random center. Convergence follows from the more general results in [
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about the asymptotic behavior of the logarithmic moments of the "least denominator". Specifically, the author studies the logarithmic moments of the least denominator among all rational numbers in intervals with shrinking random centers. This problem stems from a numerical research problem proposed by Meiss and Sander when they were studying the breakdown of invariant tori in integrable systems. They focused on the distribution of the least resonance order and hoped to prove its limit law and the convergence of its expected value.
### Problem Background
1. **Definition of the Least Denominator**:
The paper considers the smallest denominator \(q\) among all rational numbers \(\frac{p}{q}\) in an interval \([x - \frac{\delta}{2}, x + \frac{\delta}{2}]\). Formally, it is defined as:
\[
q_{\text{min}}(x, \delta) = \min \left\{ q \in \mathbb{N} : \exists p \in \mathbb{Z}, \frac{p}{q} \in \left( x - \frac{\delta}{2}, x + \frac{\delta}{2} \right) \right\}
\]
2. **Research Motivation**:
When studying the breakdown of invariant tori in volume - preserving mappings, Meiss and Sander proposed a numerical research problem: for randomly selected \(x\), calculate the distribution of \(\log q_{\text{min}}(x, \delta)\) and seek the convergence of its limit law and its expected value.
3. **Existing Results**:
Chen and Haynes have already studied the asymptotic behavior of the expected value (without taking the logarithm) of \(q_{\text{min}}(x, \delta)\).
### Main Contributions of the Paper
- **Limit Theorem**:
The author proves that as \(\delta \to 0\), the distribution of \(\log q_{\text{min}}(x, \delta)\) converges to a limit distribution and gives a specific formula for this limit distribution.
- **Explicit Formula**:
The paper derives an explicit formula for the logarithmic moments of this limit distribution, especially providing a specific expression for the moment \(\mu_n\) in the one - dimensional case.
- **High - Dimensional Generalization**:
The author also discusses the convergence of the logarithmic moments of the least denominator in higher dimensions and applies it to the study of the least resonance order proposed by Meiss and Sander.
### Formula Summary
- **Limit Density Function**:
\[
\eta_{\log}(s) = \frac{6}{\pi^2} \begin{cases}
e^{2s} & \text{if } s \leq 0 \\
-e^{2s} + 2 + 4s & \text{if } 0 \leq s \leq \log 2 \\
-e^{2s} + 2 + 2e^{2s} \sqrt{\frac{1}{4} - e^{-2s}} - 4 \log \left( \frac{1}{2} + \sqrt{\frac{1}{4} - e^{-2s}} \right) & \text{if } s \geq \log 2
\end{cases}
\]
- **Explicit Formula for Moments**:
\[
\mu_n = \frac{n!}{2^n} \sum_{k = 0}^n \sum_{j = 0}^k \frac{(\log 2)^{n - k}}{(n - k)!} \frac{(-1)^j}{2^j (k - j)!} \rho_{k - j}
\]
where,
\[
\rho_n = \frac{24}{\pi^2} \left( \frac{n!}{2^{n+1}} \sum_{j}