Work

Windowed Weil forms under RH alone: an e^{−μ^{6/5+ε}} lower bound and a barrier from distinct zeros

Year: 2026
Type: preprint
Abstract: Let λ_min(√μ) be the smallest eigenvalue of Weil's quadratic form on test functions supported in the multiplicative window [μ^{−1/2}, μ^{1/2}]. Under the Riemann Hypothesis (RH) and a hypothesis on close pairs of zeros it is known that −log λ_min(√μ) = (4π + o(1))μ, while the earlier bound under RH alone is −log λ_min(√μ) ≤ exp(μ^ϖ), ϖ > 1/2. We prove, under RH alone, that −log λ_min(√μ) ≤ C μ^{6/5} log μ (1 + log log μ). The proof rests on an upper bound for the number of close pairs of zeros in an interval, uniform in the position of the interval, which is obtained from Montgomery's explicit formula and a mean value estimate of Montgomery and Vaughan. In the other direction we show, under RH, that a deficit of distinct zeros up to height T against the number (T/2π) log μ forces λ_min(√μ) to be small. Consequently, if −log λ_min(√μ) ≤ μ^A for all large μ, then the proportion of distinct zeros up to height T is at least 1/A − o(1); so a bound of this kind from RH alone with A < 1.1796 would improve the best proportion of distinct zeros known to us under RH, while the bound proved here has A = 6/5 + ε. We also show that the exponent 1/2 of the earlier bound cannot be lowered by arguments which use the zeros only through their symmetry and the classical bound for S(t), and that for every B > 2 a separation condition of the form |γ − γ'| ≥ exp(−(log|γ|)^B), even for simple points, does not give −log λ_* = O(μ) for point sets of this kind. No statement about the truth of RH is made. All results are conditional on RH or concern model point sets, as stated in the paper. The paper does not claim a proof of RH. It builds on the author's preprint "The smallest eigenvalue of windowed Weil forms decays like e^{−4πμ} under RH and hypotheses on close pairs of zeros" (doi:10.5281/zenodo.24246078). Not yet reviewed by independent human experts. Generative AI systems were used extensively in this work under the author's direction; see the section "Use of generative AI" in the paper.
Author: Mori, Ryota
Language: English
Open Access status: green
OpenAlex ID: W7234018101
Cites: 3
Cited by: 2
Related to: 10