Boundary modulus of a finite canonical product
ProvedZeroFreeBridge.norm_finprod_canonicalFactor_zpow_eq_onearda-rh65k-transfernumber-theoryriemann-zeta
Let have finite support , and let be the defined canonical factor. For every ,
The product has unit modulus on the boundary circle.
Preamble
import Mathlib
/- PHASE 4 (dVP frontier, BLASCHKE correction): the finite Blaschke product has modulus 1 on the
sphere — the fact that removes the lossy AP term.
The monomial factorization `ζ = ∏(·-ρ)^m · g` fails to give `M' = O(L)`: near a boundary zero the
decomposition `log‖g‖ = log‖ζ‖ - log‖P‖` throws away a cancellation, so `AP` blows up. The
classical fix uses BLASCHKE factors, and Mathlib already has them:
`canonicalFactor R w z = (R² - conj w · z)/(R(z-w))` (`Analysis/Complex/CanonicalDecomposition`),
with `‖canonicalFactor R w z‖ = 1` for `z ∈ sphere 0 R`, `w ∈ ball 0 R`
(`norm_canonicalFactor_eval_circle_eq_one`), and the full decomposition `MeromorphicOn.exists_
canonicalDecomp`: `f =ᶠ (∏ᶠ_u canonicalFactor^{-divisor u}) • g` with `g` zero-free on the ball.
This file lifts the single-factor modulus-1 fact to the whole finite product: for a finitely
supported exponent `n` with support in `ball 0 R`,
`‖(∏ᶠ_u (canonicalFactor R u)^{n u}) z‖ = 1` for `z ∈ sphere 0 R`.
Consequently `‖f‖ = ‖g‖` on the sphere (up to the codiscrete exceptional set) — so the boundary
`M'` is `sup log‖f‖ - log‖g c‖ = O(L)` with NO zero-factor term. conjecture1_proved = False.
-/
open Complex Metric
section
namespace ZeroFreeBridge
end ZeroFreeBridge
open ZeroFreeBridge
Formal statement
theorem ZeroFreeBridge.norm_finprod_canonicalFactor_zpow_eq_one {R : ℝ} (n : ℂ → ℤ)
(hn : (Function.support n).Finite)
(hsupp : ∀ u ∈ hn.toFinset, u ∈ ball (0 : ℂ) R)
{z : ℂ} (hz : z ∈ sphere (0 : ℂ) R) :
‖(∏ᶠ u, (canonicalFactor R u) ^ (n u)) z‖ = 1 := by sorry
Source