EXACT STECHKIN WEIGHTS AND IMPROVED ZERO-FREE REGIONS FOR THE RIEMANN ZETA-FUNCTION AND DIRICHLET L-FUNCTIONS
In Stechkin's differencing step the logarithmic derivative is read at a point σ = ½ + a and at shifted points σ + δₘ with weights κₘ, and the resulting kernel must be nonnegative on every pair of zeros ρ, 1 − ρ̄. For a single shift we show, by an exact factorisation of the pair kernel, that for 0 < a < ½ the largest admissible weight is a/(a + δ) whenever δ ≥ (√5 − 1)a, and that this persists under every smoothing whose even extension has a nonnegative Fourier transform. For several shifts the pair condition is linear in the weights, so the best weights are found by linear programming; their sum, which multiplies the conductor term, can be brought close to ½ but always exceeds ½. With a ten-shift weight whose shifts follow the reading point, Kadiri's argument gives the classical zero-free region for Dirichlet L-functions of modulus 3 ≤ q ≤ 400 000 with the constant 4.9774 in place of 5.60, with consequences for the explicit bounds of Bennett, Martin, O'Bryant and Rechnitzer for primes in arithmetic progressions. With the same weight in the argument of Bellotti, Trudgian and Yang, together with sharper error terms, a prime sum evaluated at each prime power, a new trigonometric polynomial, a sharpened form of Ford's zero-free region and a new bound for |ζ(½ + it)| for t ≥ 2.9·10¹², proved by weighted Weyl differencing and an explicit B-process, we obtain ζ(σ + it) ≠ 0 for σ ≥ 1 − 1/(4.6892 log t), t ≥ 3, in place of 4.896. All numerical inequalities in the proofs are verified in interval arithmetic.
Authors
- Eric Fodge
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-17
- DOI
- https://doi.org/10.5281/zenodo.24325328
- Primary Topic
- Analytic Number Theory Research
- Type
- preprint