openAI-math-dataset / lean /OAI /NumberTheory /TwoPoint /Walks /CanonicalComplexPrefix.lean
roseking's picture
Add lean source and binary files (part 118)
180bc03 verified
Raw History Blame Contribute Delete
4.1 kB
import OAI.NumberTheory.TwoPoint.Walks.CanonicalComplexTesting
import OAI.NumberTheory.TwoPoint.Bounds.ComplexPrefixBound
namespace OAI
/-! Canonical directed complex prefix estimate with all boundary terms explicit. -/
namespace TwoPointCorrelations
open Finset Filter
open scoped Classical
theorem ModFiveThetaInput.eventually_canonical_complex_prefix
(hprime : ModFiveThetaInput) (hBr : BravermanDepth22Input) :
∃ A : ℕ, 1000 ≤ A ∧
∀ (h : ℕ) (_hh : 0 < h) (E : Finset ℕ)
(hE : ∀ p, p.Prime → p ∣ h → p ∈ E) (W : ℝ) (hW : 1 ≤ W),
∀ᶠ L : ℝ in atTop,
∀ (hL : 1 ≤ L) (η : ℝ), 0 < η → η ≤ 1 →
∀ eligible : ℕ → ℕ → Prop,
(∀ d q, eligible d q → PaddingPairEligible L η d q) →
∀ (D : ℤ), 0 < D →
(∀ d q, eligible d q → D ≤ (h * q * d : ℕ) ∧ (h * q * d : ℕ) < 2 * D) →
let J := primeSupplyCount W L
let P := centeredPrimeBands E (L ^ (199 / 200 : ℝ)) W J
let Qp := paddingPrimeSupply E L
let Q := boundedPaddingDivisors Qp ⌊100 * Real.log L⌋₊
let data := canonicalTraceFamily h E W L eligible hL hW hE
let active := fun d q => (d, q) ∈ data.pairs
let keep := fun z => ¬ProhibitedSite h ⌊L ^ (1 / 10 : ℝ)⌋₊ active z
let M := ⌈Real.exp (103 * L)⌉₊
let K := Real.exp (4 * J)
let R := Real.exp 1 * (2 * (K * (2 * Real.exp 150 * Real.sqrt W) ^ J))
∀ (F G : ℤ → ℂ), (∀ n, ‖F n‖ ≤ 1) → (∀ n, ‖G n‖ ≤ 1) →
∀ N : ℕ, Real.exp (L ^ A / 2) ≤ (N : ℝ) →
‖retainedComplexPrefix P Q Qp actualPaddingCoefficient active L K W
(fun _ => actualPaddingDegreeCut Qp L) h (fun _ _ _ => True) keep F G N‖ ≤
(40 * ((3 * R) * (2 * M * paddingTiltNormalizer Qp) +
((M : ℝ) * (2 * K * (8 * W) ^ J) * (5 : ℝ) ^ (400 * Real.log L)) *
Real.exp (-(2 * ⌊L⌋₊ : ℕ)))) / (L * M) +
((h * ⌊Real.exp (100 * L + 1)⌋₊ : ℕ) : ℝ) / M *
(K / L * (5 : ℝ) ^ (400 * Real.log L) * (8 * W) ^ J) +
2 * (M : ℝ) / N * (K / L * (5 : ℝ) ^ (400 * Real.log L) * (8 * W) ^ J) := by
obtain ⟨A, hA, hb⟩ := hprime.eventually_canonical_complex_blocks hBr
refine ⟨A, hA, ?_⟩
intro h hh E hE W hW
filter_upwards [hb h hh E hE W hW, hprime.eventually_actual_pool_masses E W hW]
with L hb hmass
intro hL η hη hηone eligible he D hD hs
dsimp only
let J := primeSupplyCount W L
let P := centeredPrimeBands E (L ^ (199 / 200 : ℝ)) W J
let Qp := paddingPrimeSupply E L
let Q := boundedPaddingDivisors Qp ⌊100 * Real.log L⌋₊
let data := canonicalTraceFamily h E W L eligible hL hW hE
let active := fun d q => (d, q) ∈ data.pairs
let keep := fun z => ¬ProhibitedSite h ⌊L ^ (1 / 10 : ℝ)⌋₊ active z
let M := ⌈Real.exp (103 * L)⌉₊
let K := Real.exp (4 * J)
let R := Real.exp 1 * (2 * (K * (2 * Real.exp 150 * Real.sqrt W) ^ J))
intro F G hF hG N hN
have hNp : 0 < N := by exact_mod_cast (Real.exp_pos _).trans_le hN
have hMp : 0 < M := Nat.one_le_ceil_iff.mpr (Real.exp_pos _)
have hp := centeredPrimeBands_prime E (L ^ (199 / 200 : ℝ)) W J
have hd := centeredPrimeBands_disjoint E (L ^ (199 / 200 : ℝ)) W J
(Real.rpow_nonneg (by linarith) _) (by linarith)
have hstep (e : ((j : Fin J) → P j) × Q) :
retainedPrimeStep active h e ≤ h * ⌊Real.exp (100 * L + 1)⌋₊ := by
apply retainedPrimeStep_le h hη hηone
intro d q hmem
exact he d q (canonicalTraceFamily_pair_eligible h E W L eligible hL hW hE d q hmem)
have ht := hb hL η hη hηone eligible he D hD hs F G hF hG N hN
exact retainedComplexPrefix_bound P hp hd M (h * ⌊Real.exp (100 * L + 1)⌋₊) N
Q Qp actualPaddingCoefficient active L K W (fun _ => actualPaddingDegreeCut Qp L)
h (fun _ _ _ => True) keep F G hF hG hMp hNp (by linarith) (Real.exp_pos _).le
(by linarith) (fun q _ => actualPaddingCoefficient_nonneg q)
(fun j => (hmass.2.1 j).2.1) hstep _ ht
end TwoPointCorrelations
end OAI