A compactly supported rational adelic Schwartz–Bruhat test that violates the printed same-sign Poisson formula
ProvedMasulloRH.printed_adelic_poisson_counterexampleadelic-analysispoisson-summationriemann-hypothesis
Let u=(1,−1) on the rational adeles and let χ(x)=ψ_Q(ux), the Fourier character with the same local signs printed in Examples A.3.1.5–A.3.1.8. There is a compactly supported adelic Schwartz–Bruhat test f with nonnegative real part that vanishes at every diagonal rational point and equals 1 at uΔ(1/3). For every additive Haar measure the real part of the covolume-normalized sum of its χ-Fourier transform over the ordinary diagonal rational lattice is at least 1. Thus the geometric lattice sum is 0 and the printed unshifted Fourier sum cannot equal it. The witness works for every Haar normalization. The separately proved finite-sign-even Poisson result describes a sector where the same-sign convention does work.
Preamble
import Mathlib import Definitions.Def_NumberField_StandardGlobalAddCharRat import Definitions.Def_NumberField_AdelicBox import Definitions.Def_NumberField_AdelicFourier set_option autoImplicit false set_option backward.isDefEq.respectTransparency.types false
Formal statement
open NumberField NumberField.StandardAddChar NumberField.AdelicFourier MeasureTheory
open scoped NumberField.AdeleRing
namespace MasulloRH
theorem printed_adelic_poisson_counterexample [MeasurableSpace 𝔸[ℚ]] [BorelSpace 𝔸[ℚ]] :
let u : 𝔸[ℚ] := (1,-1)
let χ := psiQ.mulShift u
∃ f ∈ schwartzBruhat ℚ, HasCompactSupport f ∧
f (u * algebraMap ℚ 𝔸[ℚ] (1/3)) = 1 ∧ (∀ x, 0 ≤ (f x).re) ∧
(∀ r : ℚ, f (algebraMap ℚ 𝔸[ℚ] r) = 0) ∧
∀ (μ : Measure 𝔸[ℚ]) [μ.IsAddHaarMeasure],
(1:ℝ) ≤ (((μ (AdelicBox.adelicBox ℚ)).toReal : ℂ)⁻¹ *
∑' r : ℚ, fourierIntegral χ μ f (algebraMap ℚ 𝔸[ℚ] r)).re := by sorry
end MasulloRHSource
Masullo, The Music of the Primes, Theorem 2.2.3.2 at the identity idele, Remark 2.2.3.3, and the local convention in Examples A.3.1.5–A.3.1.8. The independent mccorvie/riemann-masullo project identified the sign issue in erratum F13, https://github.com/mccorvie/riemann-masullo/blob/main/ERRATA.md. This proof constructs a compactly supported adelic test separating the two rational lattices and quantifies the failure in the same normalization as 8b98b7eb.