Abstract:We introduce approximation functions of $li(x)$ for all $x\ge e$: (1) $\displaystyle li_{\underline{\omega},\alpha}(x) = \frac{x}{\log(x)}\left( \alpha\frac{\underline{m}!}{\log^{\underline{m}}(x)} + \sum_{k=0}^{\underline{m}-1}\frac{k!}{\log^{k}(x)} \right)$, and (2) $\displaystyle li_{\overline{\omega},\beta}=\frac{x}{\log(x)}\left( \beta\frac{\overline{m}!}{\log^{\overline{m}}(x)} + \sum_{k=0}^{\overline{m}-1}\frac{k!}{\log^{k}(x)} \right)$ with $0 < \omega < 1$ a real number, $\alpha \in \{ 0, \underline{\kappa}\log(x) \}$, $\underline{m} = \lfloor \underline{\kappa}\log(x) \rfloor$, $\beta \in \{ \overline{\kappa}\log(x), 1 \}$, $\overline{m} = \lfloor \overline{\kappa}\log(x) \rfloor$, and $\underline{\kappa} < \overline{\kappa}$ the solutions of $\kappa(1-\log(\kappa)) = \omega$. Since the error of approximating $li(x)$ using Stieltjes asymptotic series $\displaystyle li_{*}(x) = \frac{x}{\log(x)}\sum_{k=0}^{n-1}\frac{k!}{\log^{k}(x)} + (\log(x)-n)\frac{xn!}{\log^{n+1}(x)}$, with $\displaystyle n = \lfloor \log(x) \rfloor$ for all $x\ge e$, satisfies $\displaystyle |\varepsilon(x)| = |li(x)-li_{*}(x)| \le 1.265692883422\ldots$, by using Stirling's approximation and some facts about $\log(x)$ and floor functions, we show that $\displaystyle \varepsilon_{0}(x) = li(x) - li_{\underline{1/2},0}(x)$, $\displaystyle \underline{\varepsilon}(x) = li(x) - li_{\underline{1/2},\underline{\kappa}\log(x)}(x)$, $\displaystyle \overline{\varepsilon}(x) = li(x) - li_{\overline{1/2},\overline{\kappa}\log(x)}(x)$, and $\varepsilon_{1}(x) = li_{\overline{1/2},1}(x) - li(x)$ belong to $O\left(\sqrt{\frac{x}{\log(x)}}\right)$. Moreover, we conjecture that $li_{0}(x) \le \pi(x) \le li_{1}(x)$ and $\underline{li}(x) \le \pi(x) \le \overline{li}(x)$ for all $x \ge e$, here $\pi(x)$ is the prime counting function and we show that if one of those conjectures is true then the Riemann Hypothesis is true.

A Robust Approximation to a Lambert-Type Function

New approximations to the principal real-valued branch of the Lambert W-function

Computing the Lambert W function in arbitrary-precision complex interval arithmetic

Solar Cells, Lambert W and the LogWright Functions

Complex branches of a generalised Lambert $W$ function arising from $p,q$--binomial coefficients

On the Stieltjes Approximation Error to Logarithmic Integral

Analytical approximations as close as desired to special functions

Additional applications of the Lambert W function to solid state physics

Strong Approximation Theorems for Geometrically Weighted Random Series and Their Applications

Approximation for the Theis Well Function Using Ramanujan’s Series and Bounds for the Exponential Integral

Acknowledgment of priority: Usage of the Lambert W function in statistics

A Fast Algorithm to Calculate Exponential Function E~x

Element-Free Approximation of Generalized Regularized Long Wave Equation

A rational approximation for efficient computation of the Voigt function in quantitative spectroscopy

Computation of the Wright function from its integral representation

$L^p$ Error Estimates for Approximation by Sobolev Splines and Wendland Functions on $\mathbb{R}^d$

Generalized Karagiannidis–Lioumpas Approximations and Bounds to the Gaussian Q-Function With Optimized Coefficients

Lambert W random variables - a new family of generalized skewed distributions with applications to risk estimation

New error bounds for Legendre approximations of differentiable functions

Defining Upper and Lower Bounding Functions of $li(x)$ with $\displaystyle O\left(\sqrt{\frac{x}{\log(x)}}\right)$ Error Using Truncated Asymptotic Series

The Wright function -- hypergeometric representation and symbolical evaluation