Model sum two-term asymptotic
ProvedAnalyticCombinatorics.Ch8.Partitions.Erdos.model_sum_two_term_asympanalytic-combinatoricslean4partitions-remaindersmooth-driftsupporting-lemma
Model sum two-term asymptotic: |model_sum − (μ̄ + μA/√n)| ≤ K/n for large n.
The identity model_sum = c₁·M₁ + c₂·M₂ + c₃·M₃ + c₄·M₄ (Lemma model_sum_eq_moments)
is combined with the banked moment asymptotics.
The full literal source contract is:
theorem model_sum_two_term_asymp :
∃ K : ℝ, 0 < K ∧ ∀ᶠ n : ℕ in atTop,
|(∑' m : ℕ, if m = 0 then (0:ℝ) else modelSummand n m * rhoDropModel n m)
- (3 / massLam + (3 / (2 * massLam ^ 2)) / Real.sqrt (n : ℝ))|
≤ K / (n : ℝ)
Preamble
import Definitions.Def_AnalyticCombinatorics_Ch8_Partitions_MuNumModel import Definitions.Def_AnalyticCombinatorics_Ch8_Partitions_SigmaRecurrence import Definitions.Def_AnalyticCombinatorics_Ch8_Partitions_UpperBound import Definitions.Def_AnalyticCombinatorics_Ch8_Partitions_ErdosKernel import Definitions.Def_AnalyticCombinatorics_Ch8_Partitions_MassRateMoments import Definitions.Def_AnalyticCombinatorics_Ch8_Partitions_MassRateApprox2 import Mathlib set_option autoImplicit true set_option relaxedAutoImplicit false set_option maxSynthPendingDepth 3 /-! # MuNum two-term expansion `|muNum n − (μ̄ + μA/√n)| ≤ K/n` for large `n`, where `μ̄ = 3/massLam` and `μA = 3/(2·massLam²)`. -/ set_option maxHeartbeats 4000000 open Filter Finset open scoped Topology BigOperators open AnalyticCombinatorics AnalyticCombinatorics.Ch8 AnalyticCombinatorics.Ch8.Partitions AnalyticCombinatorics.Ch8.Partitions.Erdos /-! ### Algebraic expansion of the model product -/ /-! ### Model sum expressed via sigma moments (infinite-sum identity) -/ /-! ### Small lemmas -/ /-! ### Two-term asymptotic from the moment asymptotics -/
Formal statement
theorem AnalyticCombinatorics.Ch8.Partitions.Erdos.model_sum_two_term_asymp :
∃ K : ℝ, 0 < K ∧ ∀ᶠ n : ℕ in atTop,
|(∑' m : ℕ, if m = 0 then (0:ℝ) else modelSummand n m * rhoDropModel n m)
- (3 / massLam + (3 / (2 * massLam ^ 2)) / Real.sqrt (n : ℝ))|
≤ K / (n : ℝ) := by sorrySource