Second-order local and tail accuracy records
DefinitionAnalyticCombinatorics_Ch8_SaddlePoint_SecondOrderHAdmissibleanalytic-combinatoricslean4saddle-point
The record stores b3,b4 and an eventually positive vanishing correction scale c2. Its proof fields require Q2=O(c2), weighted central L1 accuracy o(c2), and uniform angular tail accuracy at the same scale. These are supplied accuracy data, not automatic consequences of analyticity.
Definition code
import Definitions.Def_AnalyticCombinatorics_Ch8_SaddlePoint_HAdmissible
import Definitions.Def_AnalyticCombinatorics_Ch8_SaddlePoint_SecondOrderCore
import Mathlib.Analysis.Real.Pi.Bounds
import Mathlib
import Mathlib.Probability.Distributions.Gaussian.Real
import Mathlib.Analysis.Analytic.Basic
import Mathlib.Analysis.Analytic.OfScalars
import Mathlib.Analysis.Analytic.RadiusLiminf
import Mathlib.RingTheory.PowerSeries.Basic
import Mathlib.Algebra.Algebra.Rat
import Mathlib.Data.Complex.Basic
set_option autoImplicit true
set_option relaxedAutoImplicit false
set_option maxSynthPendingDepth 3
/-!
# Second-order H-admissible wrapper
This is a clean add-on to the first-order `HAdmissible` interface. It records
the third and fourth logarithmic saddle derivatives, the robust correction
scale, and the local/tail hypotheses used by the second-order saddle layer.
-/
open Complex Filter Asymptotics MeasureTheory
open scoped Topology Real NNReal ENNReal Interval
noncomputable section
/-- Second-order Hayman data over an existing first-order H-admissible instance. -/
structure SecondOrderHAdmissible {f : ℂ → ℂ}
{p : FormalMultilinearSeries ℂ ℂ ℂ} (hf : HAdmissible f p) where
b3 : ℝ → ℝ
b4 : ℝ → ℝ
corrScale : ℝ → ℝ
corrScale_pos : ∀ᶠ r : ℝ in hf.radFilter, 0 < corrScale r
corrScale_tendsto_zero : Tendsto corrScale hf.radFilter (𝓝 0)
corrScale_dominates_correction :
∃ K : ℝ, 0 ≤ K ∧
∀ᶠ r : ℝ in hf.radFilter,
saddleCorrectionScale hf.b b3 b4 r ≤ K * corrScale r
local_second_L1 :
Tendsto
(fun r : ℝ =>
(∫ x in -(hf.δ r * Real.sqrt (hf.b r))..(hf.δ r * Real.sqrt (hf.b r)),
‖Complex.exp (-(x ^ 2) / 2) *
(saddleLocalRatio f hf.a hf.b r (x / Real.sqrt (hf.b r)) -
saddleSecondPoly hf.b b3 b4 r x)‖) / corrScale r)
hf.radFilter (𝓝 0)
tail_second_uniform :
∃ E : ℝ → ℝ,
(∀ᶠ r : ℝ in hf.radFilter, 0 ≤ E r) ∧
Tendsto
(fun r : ℝ => Real.sqrt (2 * Real.pi * hf.b r) * E r / corrScale r)
hf.radFilter (𝓝 0) ∧
(∀ᶠ r : ℝ in hf.radFilter, ∀ n : ℕ, ∀ θ : ℝ,
hf.δ r ≤ |θ| → |θ| ≤ Real.pi → ‖saddleGAt f r n θ‖ ≤ E r)
Source