Optimized de la Vallée Poussin region estimate
ProvedZeroFreeBridge.dlvp_region_boundarda-rh65k-transfernumber-theoryriemann-zeta
Let , , and , and write . Suppose
Then
This is the real-variable optimization used in the zero-free-region proof.
Preamble
import Mathlib
/- PHASE 4 (dVP frontier, the σ-OPTIMIZATION — the headline region bound): turn the cleared region GAP
into the de la Vallée Poussin bound `β ≤ 1 - c/(A·L)` (`= 1 - c/log|t|`).
`ZeroFreeRegion.dlvp_region_gap` produces, at any `σ > 1`, the gap
`(σ-1)·(1 - (σ-1)·B) ≤ (1-β)·(3 + (σ-1)·B)`, B = 3A + 5AL.
The classical dVP optimization CHOOSES `σ - 1 = δ = 1/(2B)`, so `δ·B = 1/2` and the gap collapses to
`δ/2 ≤ (7/2)(1-β)`, i.e. `1 ≤ 14 B (1-β)`. Since `B = A(3+5L) ≤ 8AL` for `L ≥ 1`, this gives
`112 A L (1-β) ≥ 1`, hence
`β ≤ 1 - 1/(112 · A · L)`,
the zero-free region with an EXPLICIT constant `c = 1/112` (per unit `A`). This is the final step of
the reduction skeleton: composed with `dlvp_region_gap` at `σ = 1 + 1/(2B)`, it yields the region from
the three Borel–Carathéodory inputs. Self-contained real-arithmetic. conjecture1_proved = False.
-/
section
namespace ZeroFreeBridge
end ZeroFreeBridge
open ZeroFreeBridge
Formal statement
theorem ZeroFreeBridge.dlvp_region_bound (A L β δ : ℝ) (hA : 0 < A) (hL : 1 ≤ L) (hδpos : 0 < δ)
(hδB : δ * (3 * A + 5 * (A * L)) = 1 / 2)
(hgap : δ * (1 - δ * (3 * A + 5 * (A * L)))
≤ (1 - β) * (3 + δ * (3 * A + 5 * (A * L)))) :
β ≤ 1 - 1 / (112 * (A * L)) := by sorry
Source