openAI-math-dataset / lean /OAI /Probability /SKGap /Gaussian /GaussianFiniteIBPError.lean
roseking's picture
Add lean source and binary files (part 128)
7d9f027 verified
Raw History Blame Contribute Delete
1.88 kB
import Mathlib
import OAI.Probability.SKGap.Localization.CoordinateDeriv
namespace OAI
noncomputable section
open MeasureTheory ProbabilityTheory InformationTheory Real Set
open scoped NNReal ENNReal
open Filter
open scoped Topology
namespace SKGap
variable {κ : Type*} [Fintype κ] [DecidableEq κ]
theorem gaussianFinite_IBP_error (i : κ)
{f d : (κ → ℝ) → ℝ} {L : ℝ≥0} {C D : ℝ} {s : Set (κ → ℝ)}
(hf : LipschitzWith L f) (hC : ∀ x, ‖f x‖ ≤ C)
(hd : Measurable d) (hD : ∀ x, ‖d x‖ ≤ D) (hs : MeasurableSet s)
(hder : ∀ x ∈ s, HasDerivAt (fun z => f (Function.update x i z)) (d x) (x i)) :
|(∫ x, x i*f x ∂Measure.pi (fun _ : κ => gaussianReal 0 1))-
∫ x, d x ∂Measure.pi (fun _ : κ => gaussianReal 0 1)| ≤
((L:ℝ)+D)*(Measure.pi (fun _ : κ => gaussianReal 0 1)).real sᶜ := by
let μ := Measure.pi (fun _ : κ => gaussianReal 0 1)
have hi : Integrable (finiteCoordinateDeriv i f) μ :=
Integrable.of_bound (measurable_finiteCoordinateDeriv i hf.continuous).aestronglyMeasurable
(L:ℝ) (ae_of_all _ (norm_finiteCoordinateDeriv_le i hf))
have hid : Integrable d μ := Integrable.of_bound hd.aestronglyMeasurable D (ae_of_all _ hD)
rw [gaussianFinite_lipschitz_IBP i hf hC,← integral_sub hi hid]
have he : (fun x => finiteCoordinateDeriv i f x-d x) =
sᶜ.indicator (fun x => finiteCoordinateDeriv i f x-d x) := by
funext x
by_cases hx : x ∈ s
· simp only [Set.indicator_of_notMem (show x ∉ sᶜ by simpa using hx),finiteCoordinateDeriv]
rw [(hder x hx).deriv,sub_self]
· rw [Set.indicator_of_mem hx]
rw [he,integral_indicator hs.compl]
exact norm_setIntegral_le_of_norm_le_const (measure_lt_top μ sᶜ) (fun x _ =>
(norm_sub_le _ _).trans (add_le_add (norm_finiteCoordinateDeriv_le i hf x) (hD x)))
end SKGap
end
end OAI