Real-variable estimates for [I71, Theorem 5] #
With L = log y, λ = log log 3y (TheoremFive.lam) and ξ = L / λ^{11/5} (xi), this file
collects the "for y ≥ y₀" facts used in the proof of Theorem 5:
eventually_scalar:λ,log λ,λ^{1/5}are large,λ - 1 ≤ log L ≤ λ,24 λ ≤ ξ,16 λ² ≤ L,λ/2 ≤ log (ξ/2),4 log λ + c ≤ log ξ ≤ λ;crude_exponent_le: the comparison of the crude count ofrosserSet lower y y^{1/(ξ-1)}withy exp(-ξ log ξ + (3/2) ξ log λ) / (8 L⁴)([I71, p. 27]);exists_slack: the lossesδ(e^{32}, y) (1 + C ξ² pntE(y^{1/ξ})) - 1of Corollary 3 are at most(log log y)² / (8 L²), reusingTheoremFour.eventually_delta_mul_le.
theorem
IwaniecLinearSieve.TheoremFive.eventually_scalar
(c : ℝ)
(hc : 100 ≤ c)
:
∀ᶠ (y : ℝ) in Filter.atTop, 3 ≤ y ∧ c ≤ Real.log (lam y) ∧ c ≤ lam y ^ (1 / 5) ∧ c ≤ lam y ∧ Real.log (Real.log y) ≤ lam y ∧ lam y - 1 ≤ Real.log (Real.log y) ∧ 24 * lam y ≤ xi y ∧ 16 * lam y ^ 2 ≤ Real.log y ∧ lam y / 2 ≤ Real.log (xi y / 2) ∧ 4 * Real.log (lam y) + c ≤ Real.log (xi y) ∧ Real.log (xi y) ≤ lam y
The scalar facts about L = log y, λ = lam y and ξ = xi y used for Theorem 5, valid for
large y; c ≥ 100 is a free threshold.
The crude part ([I71, p. 27]) #
theorem
IwaniecLinearSieve.TheoremFive.crude_exponent_le
{L ℓ lam ξ a : ℝ}
(hL : 0 < L)
(hℓ : ℓ = Real.log L)
(hℓlam : ℓ ≤ lam)
(hlam : 100 ≤ lam)
(ha : 12 ≤ a)
(hxilam : 4 * (ξ * lam ^ 2) ≤ L)
(h24 : 24 * lam ≤ ξ)
(h16 : 16 * lam ^ 2 ≤ L)
(hlogξ : Real.log ξ ≤ lam)
:
The comparison at the bottom of [I71, p. 27], in exponential form: with ℓ = log L,
8 L⁴ (1 + e^{L + 2ξ - L/λ} + 2 e^{L - ξ log ξ + ξ a + 5ξ}) ≤ e^{L - ξ log ξ + (3/2) ξ a}
(here a = log λ), under the scalar facts of eventually_scalar.