关于对数积分的Stieltjes近似误差
On the Stieltjes Approximation Error to the Logarithmic Integral
AI总结:
本文研究了Stieltjes近似误差对logarithmic integral的误差,推导了无条件分析界,并提供了首个完全明确的全局界。
摘要:
我们研究了经典Stieltjes渐近展开式对logarithmic integral(对数积分)的近似误差ε(x)=li*(x)-li(x)。我们的分析基于离散值ε_k=ε(e^k)及其增量Δ_k=ε_{k+1}-ε_k,我们推导了新的无条件分析界。利用每个增量Δ_k的精确积分表示,结合与相关核积分相关的精确上界和下界估计,我们得到了对于所有k≥1的ε_k可计算且一致的界,因此对于所有x≥e的ε(x)也得到了界。我们证明了以下无条件界:对于所有e≤x≤e^{1000},有1/3√(2π/ln(x)) + o(1/√(ln(x))) ≤ ε(x) ≤ 1/3√(2π/ln(x)) + o(1/√(ln(x)));对于所有x>e^{1000},有1/3√(2π/ln(x)) + o(1/√(ln(x))) - C_l ≤ ε(x) ≤ 1/3√(2π/ln(x)) + o(1/√(ln(x))) + C_r,其中C_l=0.0000035462和C_r=0.0000021511。这些结果建立了Stieltjes近似误差的第一个完全明确的全局界。最后,我们的发现强烈支持猜想行为:ε(x)=1/3√(2π/ln(x)) + o(1/√(ln(x))),对于x≥e。
英文摘要:
We study the approximation errors $\varepsilon(x)=\operatorname{li}_{*}(x)-\operatorname{li}(x)$ and $\varepsilon_{S}(x)=\operatorname{li}_{S}(x)-\operatorname{li}(x)$ arising from the classical Stieltjes--van Boven and pure Stieltjes asymptotic expansions of the logarithmic integral, respectively. Our analysis establishes a top-down reduction based on the discrete error sequence $\varepsilon_n=\varepsilon(e^n)$ and its localized partial variations, for which we derive new, unconditional analytic bounds. By developing precise integral representations for these error increments and establishing sharp, fully explicit upper and lower estima-tes for the underlying kernel integrals, we obtain uniform bounds valid for every integer $n\geq 1$. These discrete results are subsequently lifted to the real domain to provide comprehensive estimates for all real $x\geq e$. In particular, we prove that $\varepsilon(x) = \frac{1}{3} \sqrt{2π/\log x} + O(1/\log x)$ as $x \to \infty$. Furthermore, we deploy these continuous bounds to uncover the fine structural behavior of the pure Stieltjes expansion, demonstrating that $|\varepsilon_{S}(x)| = \sqrt{2π}\,|1/3-α|/\sqrt{\log x} + O(1/\log x)$ as $x\to\infty$, where $n=\lfloor\log x\rfloor$ and $α=\log x-n$. This establishes a novel and explicit asymptotic description of the absolute pure Stieltjes error, fully capturing its non-trivial oscillatory dependence on the fractional part of the logarithm.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。