theorem
MathlibNt.SieveTheory.LiuWeight.liuPanPrimitiveCofactorLedger_eq_source
(N A₁ A₂ D m : ℕ)
(f : ℕ → ℝ)
:
liuPanPrimitiveCofactorLedger N A₁ A₂ D m f = AnalyticNumberTheory.LargeSieve.PanLow.nonprincipalLow
(AnalyticNumberTheory.LargeSieve.panSourceG (fun (a : ℕ) => ↑(f a)) m)
(AnalyticNumberTheory.LargeSieve.panSourceD m) N A₁ A₂ D
Both cofactor screens are exactly the canonical high-source screens.
theorem
MathlibNt.SieveTheory.LiuWeight.eventually_pan_conductor_bounds
(B : ℝ)
(hB : 0 ≤ B)
:
∀ᶠ (N : ℕ) in Filter.atTop, 1 ≤ Real.log ↑N ∧ 1 ≤ ⌊AnalyticNumberTheory.LargeSieve.ChenLiuCoprimeProducer.lowConductor N B⌋₊ ∧ ⌊AnalyticNumberTheory.LargeSieve.ChenLiuCoprimeProducer.lowConductor N B⌋₊ ≤ panModulusCutoff N B ∧ ↑(panModulusCutoff N B) ≤ √↑N ∧ panModulusCutoff N B ≤ N
theorem
MathlibNt.SieveTheory.LiuWeight.liuPanPrimitiveCofactorLedger_log_saving
(U : ℝ)
(hU : 0 < U)
:
∃ (C : ℝ),
0 < C ∧ ∃ (B : ℝ),
0 ≤ B ∧ ∃ (N₀ : ℕ),
∀ N ≥ N₀,
∀ m ∈ Finset.Icc 1 (panModulusCutoff N B),
liuPanPrimitiveCofactorLedger N (liuPanSourceIntervalLower N B) (liuPanSourceIntervalUpper N)
(panModulusCutoff N B) m (liuWeight N (liuSourceZ10 N) (liuSourceY3 N)) ≤ C * ↑N / Real.log ↑N ^ U
Full primitive ledger, uniformly in the cofactor. No high or low estimate is a theorem parameter. The exponent B is chosen before N and m.