Documentation

MathlibNt.SieveTheory.LinearSieve.Suzuki.SuzukiClaim145CaseALowS

Suzuki Claim 14.5, Case A: the low-s source branch #

For κ = κ̂ = 1 and β = 2, Suzuki applies Lemma 14.3 and the factorial estimate used in (14.7), first obtaining

L^M / M! * exp L ≤ exp (-s log s + s log log (3K) + O (log K + s)).

The uniform target below is exactly that first big-O step. It precedes the subsequent comparison with

exp (-s log s - s log log (3s) + sqrt K / 2)

and therefore must be discharged before the latter comparison can be used. No Claim-14.5 conclusion is assumed here.

The first missing scalar inequality in Claim 14.5, Case A, low-s range.

A is the absolute big-O coefficient and K₀ is the source choice that K is sufficiently large. The quantifiers are uniform in K, D, the natural ceiling cutoff, and s; in particular this is not a finite-D maximum or an eventual-D statement.

Equations
Instances For

    Suzuki (14.7), with the factorial estimated by the elementary Stirling lower bound, gives the first scalar big-O target in the low-s branch.

    theorem MathlibNt.SieveTheory.claim145_caseA_lowS_lemma14_3_upper (S : BoundingSieve) {N D z : } {K s : } (hlocal : SwitchingPrinciple.HasDimensionOneLocalProductBound S K) (hD : 2 D) (hz : z = D ^ (1 / s)⌉₊) (hs : 2 s) :
    suzukiActualT S N D z suzukiSourceL (↑z) K ^ (s - 2⌋₊ + 1) / (s - 2⌋₊ + 1).factorial * Real.exp (suzukiSourceL (↑z) K)

    Lemma 14.3 reaches the left side of the frozen low-s scalar target with no dependence on the depth N.

    Proposition 13.1(ii) supplies the lower side of (14.6), uniformly in the parity/depth. This is the target-scale edge used after the two scalar exponential comparisons in Suzuki's low-s branch.

    theorem MathlibNt.SieveTheory.claim145_caseA_highS_final_of_equation14_6_lower (S : BoundingSieve) (H : SwitchingPrinciple.SuzukiLemma144KappaOne.Section13HatLayers) {N D z : } {d Δ K s C1 Θ A C145 C : } (hlocal : SwitchingPrinciple.HasDimensionOneLocalProductBound S K) (hD : 2 D) (hz : z = D ^ (1 / s)⌉₊) (hs : 2 s) (hK : 1 < K) (hC1 : 0 < C1) ( : 0 Θ) (hhigh : K / Real.log K s) (hlogrel : Real.log K 4 * Real.log s) (hsmall : Real.log D C1 * K ^ Θ) (hsourceLarge : Real.exp 1 * suzukiSourceL (↑z) K s - 2) (hexponent : suzukiSourceL (↑z) K + (s - 2) * (1 + Real.log (suzukiSourceL (↑z) K) - Real.log (s - 2)) -s * Real.log s + s * Real.log (Real.log (3 * K)) + A * (Real.log K + s)) (hgrowth : C1 * 16 ^ Θ * Real.log s ^ (2 * Θ) * (Real.log (3 * K) * Real.log (3 * s) ^ 2) s ^ (d - 2 * Θ)) (hC145 : 0 C145) (habsorb : Real.exp (A * (Real.log K + s) + C * s - s * Real.log (Real.log (3 * s))) C145 * (SwitchingPrinciple.SuzukiLemma144KappaOne.claim14_5VProduct S D * (Real.exp K / (Real.log D * SwitchingPrinciple.SuzukiLemma144KappaOne.sourceSigma (↑D) d))) * s * Real.log D ^ (-Δ)) (hlower : SwitchingPrinciple.SuzukiLemma144KappaOne.claim14_5VProduct S D * (Real.exp K / (Real.log D * SwitchingPrinciple.SuzukiLemma144KappaOne.sourceSigma (↑D) d)) * ((1 + s ^ d / Real.log D) ^ s * s * SwitchingPrinciple.SuzukiLemma144KappaOne.proposition131iiLowerProfile C s) * Real.log D ^ (-Δ) SwitchingPrinciple.SuzukiLemma144KappaOne.claim14_5Scale S H N (↑D) d Δ (SwitchingPrinciple.SuzukiLemma144KappaOne.sourceSigma (↑D) d) K s) :

    Final pointwise assembly of the high-s half of Claim 14.5, Case A.

    The two hypotheses hsourceLarge and hexponent are the explicit large-K threshold hidden in the first O(log K + s) on pp. 82--83. The hypothesis habsorb is the second, final O(s) absorption (including all harmless Euler and logarithmic prefactors). They are deliberately exposed: the source only asserts the result after choosing K sufficiently large, and replacing them by a finite-D maximum or compactness would not formalize that assertion.

    Everything between these threshold inequalities is closed here: Lemma 14.3, the already proved high-coordinate logarithmic gain, the production lower form of (14.6), and the literal Claim14_5Bound.

    theorem MathlibNt.SieveTheory.claim145_caseA_highS_final (S : BoundingSieve) (H : SwitchingPrinciple.SuzukiLemma144KappaOne.Section13HatLayers) (hH : SwitchingPrinciple.SuzukiLemma144KappaOne.Section13HatSourceContract H) {N D z : } {d Δ K s C1 Θ A C145 : } (hlocal : SwitchingPrinciple.HasDimensionOneLocalProductBound S K) (hD : 2 D) (hz : z = D ^ (1 / s)⌉₊) (hs : 2 s) (hK : 1 < K) (hC1 : 0 < C1) ( : 0 Θ) (hhigh : K / Real.log K s) (hlogrel : Real.log K 4 * Real.log s) (hsmall : Real.log D C1 * K ^ Θ) (hsourceLarge : Real.exp 1 * suzukiSourceL (↑z) K s - 2) (hexponent : suzukiSourceL (↑z) K + (s - 2) * (1 + Real.log (suzukiSourceL (↑z) K) - Real.log (s - 2)) -s * Real.log s + s * Real.log (Real.log (3 * K)) + A * (Real.log K + s)) (hgrowth : C1 * 16 ^ Θ * Real.log s ^ (2 * Θ) * (Real.log (3 * K) * Real.log (3 * s) ^ 2) s ^ (d - 2 * Θ)) (hC145 : 0 C145) :

    Source-contract form of the preceding assembly. It chooses the uniform Proposition-13.1(ii) constants supplied by (14.6); once the displayed source large-K thresholds hold for those constants, the conclusion is the final Claim-14.5 bound, not an intermediate scalar comparison.

    Suzuki's second displayed comparison in Case A, uniformly throughout 2 ≤ s ≤ √K / log K. Above one source threshold the multiplicative constant can in fact be chosen to be one.

    theorem MathlibNt.SieveTheory.claim145_caseA_lowS_146_prefactor_eventually {C1 Θ d Δ C : } (hC1 : 0 C1) ( : 0 < Θ) (hd : 0 < d) (hsource : 2 / d < 1 / Θ) (hΔ0 : 0 Δ) (hΔ1 : Δ 1) (hC : 0 C) :
    ∀ᶠ (K : ) in Filter.atTop, 2 K ∀ (D s : ), 2 D2 ss K / Real.log KReal.log D C1 * K ^ ΘReal.log D / Real.log 2 * (1 + K / Real.log 2) * Real.log D ^ (1 + Δ) * SwitchingPrinciple.SuzukiLemma144KappaOne.sourceSigma D d * Real.exp (C * s) Real.exp (K / 2)

    Strengthened (14.6) prefactor absorption, including the moving sourceSigma D d denominator factor.