Diophantine FLINT-HILLS series
Nikos Mantzakouras, Carlos López Zapata
TL;DR
This work studies Diophantine Dirichlet-type series with trig terms, focusing on the unsolved Flint-Hills series $\sum_{n \ge 1} \frac{\csc^2(n)}{n^3}$ and its connections to irrationality measures. It develops a convergence framework via $\text{Riemann–Stieltjes}$ integrals, Hölder continuity, and Abel summation, tying the analysis to Fermi-Dirac/Bose-Einstein integrals and polylogarithms. The main results include a convergent representation for the Flint-Hills series, an explicit constant $c_1 \approx 78.1160806386$, and a new upper bound $\mu(\pi) \le 2.5$ for the irrationality measure of $\pi$, enriching the link between diophantine approximation and series convergence. The paper also sketches a broader program using Weierstrass elliptic curves and polygamma expansions to analyze generalized Diophantine Dirichlet series and their connections to modular forms and elliptic curves, suggesting substantial avenues for future research.
Abstract
We prove the convergence of the FLINT-HILLS series and establish new criteria for a similar type of diophantine or lacunary series, which faces issues due to spaced long terms coming from the trigonometric nature of functions, e.g., cosecant in the FLINT-HILLS series. We connect the FLINT-HILLS series to the Fermi-Dirac integral via the Riemann-Stieltjes integral and YOUNG'S inequality criteria but also proved that the upper bound of the irrationality measure of pi is equal or lower than 2.5 expected if the FLINT-HILLS series converged.
