Advanced Search

A new derivation of the classic asymptotic expansion of the $n$-th prime is presented. A fast algorithm for the computation of its terms is also given, which will be an improvement of that by Salvy (1994). Realistic bounds for the error with ${li}^{-1}\left(n\right)$, after having retained the first $m$ terms, for $1\le m\le 11$, are given. Finally, assuming the Riemann Hypothesis, we give estimations of the best possible $r_3$ such that, for $n\ge r_3$, we have $p_n\>s_3\left(n\right)$ where $s_3\left(n\right)$ is the sum of the first four terms of the asymptotic expansion.

We introduce the real valued real analytic function κ(t) implicitly defined by $e^{2\pi i\kappa \left(t\right)}=-e^{-2i\vartheta \left(t\right)}\left({\zeta }^{\text{'}}(1/2-it)\right)/\left({\zeta }^{\text{'}}(1/2+it)\right)$ (κ(0) = -1/2). By studying the equation κ(t) = n (without making any unproved hypotheses), we show that (and how) this function is closely related to the (exact) position of the zeros of Riemann’s ζ(s) and ζ’(s). Assuming the Riemann hypothesis and the simplicity of the zeros of ζ(s), it follows that the ordinate of the zero 1/2 + iγₙ of ζ(s) is the unique solution to the equation κ(t) = n.