Equation (21): genuine horizontal boundary decay #
The sole quantitative analytic input is a pointwise bound for the actual
L'/L quotient throughout the closed strip. This file does not prove that
input. The smoothing-order condition is explicit and may be paid eventually
by a caller. Constants may depend on this fixed cell.
Writing a = PerronScale x, N = PerronOrder x + 1, σ = Eq21Sigma x,
α = Alpha x, and y = x/(p₁p₂), the proof constructs
K = 2 * (σ⁻¹ + a^N) * (2*d)^r, B = (M*y^α)*K, and
C = 2*B*(α-σ). The coarse inequality 1 + log z ≤ z suffices
because no conductor-uniform constant is requested.
Horizontal integrability is proved separately using measurability and the pointwise majorant on a finite interval. Thus no nonintegrability/zero-integral convention is used, and no extra zero-free hypothesis is necessary here. The existing contour deformation still requires its own genuine zero-free input; this result does not discharge it.
Full true kernel power, reduced to a square tail only after paying the polynomial weight. The small-height estimate uses the positive real part.
Actual integrand pointwise decay on the whole strip. The character coefficient is bounded by one, never silently deleted.
Each actual horizontal section is measurable. This uses the entire nonprincipal L-function and its measurable derivative, not zero integral conventions. Nonvanishing is not needed for this measurability fact.
The two genuine horizontal integrals are integrable, and their oriented bottom-minus-top difference has square decay. The conclusion carries the factor two for the two edges and their actual length. No horizontal bound or integrability premise is assumed.
The exact horizontal_decay output consumed by the contour bridge.