Bound for the canonical-factor correction
ProvedZeroFreeBridge.norm_correction_sum_learda-rh65k-transfernumber-theoryriemann-zeta
Let , be finite, , and . Then
This controls the correction in the logarithmic-derivative decomposition.
Preamble
import Mathlib
/- PHASE 4 (dVP frontier, BLASCHKE item (d3)): the Blaschke correction sum is O(L).
Expanding `DlvpBlaschkeSplit.logDeriv_split_off_zeros` via `DlvpCanonicalLogDeriv.logDeriv_
canonicalFactor`, each term splits into the monomial Herglotz contribution and a CORRECTION
`(n u)·conj u/(R² − conj u·z)`. The correction is bounded because its pole `R²/conj u` lies
OUTSIDE the disk: `|R² − conj u·z| ≥ R² − ‖u‖‖z‖ ≥ R(R − ‖z‖) > 0`, so per term
`‖(n u)·conj u/(R² − conj u·z)‖ ≤ |n u|/(R − ‖z‖)` ,
and summing gives `‖Σ correction‖ ≤ (Σ |n u|)/(R − ‖z‖)` — a constant `O(1)` times the zero count
`Σ|n u| = O(L)`. Function-agnostic in `n`. conjecture1_proved = False (NOT a proof of RH).
-/
open Complex Metric
section
namespace ZeroFreeBridge
end ZeroFreeBridge
open ZeroFreeBridge
Formal statement
theorem ZeroFreeBridge.norm_correction_sum_le {R : ℝ} (hR : 0 < R) (n : ℂ → ℤ) (s : Finset ℂ)
(hsupp : ∀ u ∈ s, u ∈ ball (0 : ℂ) R) {z : ℂ} (hz : ‖z‖ < R) :
‖∑ u ∈ s, (n u : ℂ) * (starRingEnd ℂ) u / ((R : ℂ) ^ 2 - (starRingEnd ℂ) u * z)‖
≤ (∑ u ∈ s, |(n u : ℝ)|) / (R - ‖z‖) := by sorry
Source