A double-contour representation of the saddle integral
ProvedRSInt.rsJ_fubiniarda-rh65k-transfernumber-theoryriemann-zeta
For , set and define and . Then
This representation supports the bound on the saddle integral.
Preamble
import Definitions.Def_ArdaRH65K_RS_B2
import Definitions.Def_ArdaRH65K_RS_C0
import Definitions.Def_ArdaRH65K_RS_Entire
import Definitions.Def_ArdaRH65K_RS_Fold
import Definitions.Def_ArdaRH65K_RS_Kernel
import Definitions.Def_ArdaRH65K_RS_Mellin
import Definitions.Def_ArdaRH65K_RS_Strip
import Mathlib
import Theorems.Thm_RSInt_gauss_fourier_dn
import Theorems.Thm_RSInt_integrable_of_continuous_of_tail
/- RS_C0.lean -- lane RS, brick B3 (the Riemann-Siegel remainder), part 1: RIEMANN'S Psi.
Mathlib-only (imports `RS_B2`).
The main term of the Riemann-Siegel remainder is the Mordell integral with Gaussian e^{2 pi i w^2}
(tau = 2):
rsJ p = lineUp 0 (w |-> e^{2 pi i w^2} / (e^{i pi (p+w)} - e^{-i pi (p+w)})), 0 < p < 1.
Its value is not elementary, but the combination that enters Z(t) is:
2 Re(e^{-i pi/8} rsJ p) = cos(2 pi (p^2 - p - 1/16)) / cos(2 pi p) = Psi(p) (Gabcke's C0).
Proof (no residue theorem): write e^{i pi w^2} as a Gaussian Fourier integral along the (1-i)
line, swap the two line integrals (Fubini), evaluate the inner one by the tau = 1 Mordell closed
form (RS_Kernel), split the outer one into a tau = -1 Mordell integral (closed form by conjugation
and one pole crossing) and a tau = -2 integral that is the conjugate of rsJ itself (symmetry
rsJ(1-p) = rsJ(p)). This gives rsJ + e^{i pi/4} conj(rsJ) = elementary, whose real part after the
phase e^{-i pi/8} is Psi.
conjecture1_proved = False.
-/
open Complex MeasureTheory Filter Topology Set
open scoped Real ComplexConjugate
noncomputable section
section
namespace RSInt
end RSInt
open RSInt
/-! ## 1. Linearity for `lineDn`, translations -/
/-! ## 2. The Gaussian as a Fourier integral along the (1-i) line -/
/-! ## 3. The tau = 2 Mordell integral and its symmetry -/
/-! ## 4. Fubini: rsJ as an iterated integral -/
Formal statement
theorem RSInt.rsJ_fubini {p : ℝ} (hp : 0 < p) (hp1 : p < 1) :
rsJ p = cexp (↑π / 4 * I) * lineDn 0 (fun u => cexp (-(↑π * I * u ^ 2)) *
lineUp 0 (fun w => cexp (↑π * I * w ^ 2) / rsD (p + w) * cexp (2 * ↑π * I * u * w))) := by sorry
Source