Gaussian Product Source Stein Identity
ProvedFunctionalInequalities.KLS.LogConcave.GaussianSmoothing.gaussian_product_source_stein_identity_readableThis statement combines the affine source identity with the translated Gaussian identity on a product coupling.
Let , let be , with and globally Lipschitz, and assume . Let be globally Lipschitz and bounded by , let be continuous linear, let , and let . Define
Then the two displayed functions are integrable under , and
This is the exact product-coupling identity and includes all stated integrability conclusions.
Original Lean declaration header
Within namespace FunctionalInequalities.KLS.LogConcave.GaussianSmoothing. Shown in its original module context. Proof omitted.
set_option maxHeartbeats 800000 in
theorem gaussian_product_source_stein_identity {n : ℕ}
{u h : EuclideanSpace ℝ (Fin n) → ℝ} {L G Lh : NNReal}
(hu : ContDiff ℝ 1 u) (hL : LipschitzWith L u)
(hG : LipschitzWith G (gradient u))
(hnorm : (∫ y, Real.exp (-u y) ∂volume) = 1)
(hh : ContDiff ℝ 1 h) (hhLip : LipschitzWith Lh h)
(C : ℝ) (hbound : ∀ x, ‖h x‖ ≤ C)
(S : EuclideanSpace ℝ (Fin n) →L[ℝ] EuclideanSpace ℝ (Fin n))
(v : EuclideanSpace ℝ (Fin n)) {s : ℝ} (hs : 0 < s) :
let P := ((volume : Measure (EuclideanSpace ℝ (Fin n))).tilted (-u)).prod
(FunctionalInequalities.KLS.LogConcave.GaussianSmoothing.gaussianLaw (V := EuclideanSpace ℝ (Fin n)) s)
Integrable (fun p : EuclideanSpace ℝ (Fin n) × EuclideanSpace ℝ (Fin n) =>
⟪S (gradient u p.1) + p.2, v⟫ * h (S (gradient u p.1) + p.2)) P ∧
Integrable (fun p : EuclideanSpace ℝ (Fin n) × EuclideanSpace ℝ (Fin n) =>
⟪S.adjoint v, action (hessianMatrix u p.1)
(S.adjoint (gradient h (S (gradient u p.1) + p.2)))⟫ +
(2 * s)⁻¹ * fderiv ℝ h (S (gradient u p.1) + p.2) v) P ∧
(∫ p : EuclideanSpace ℝ (Fin n) × EuclideanSpace ℝ (Fin n),
⟪S (gradient u p.1) + p.2, v⟫ * h (S (gradient u p.1) + p.2) ∂P) =
∫ p : EuclideanSpace ℝ (Fin n) × EuclideanSpace ℝ (Fin n),
⟪S.adjoint v, action (hessianMatrix u p.1)
(S.adjoint (gradient h (S (gradient u p.1) + p.2)))⟫ +
(2 * s)⁻¹ * fderiv ℝ h (S (gradient u p.1) + p.2) v ∂P
Module navigation
Module: KLS/LogConcave/Density/GaussianSmoothing/ProductCoupling.
Definition modules: Analysis/BakryEmery/HessianOperator, KLS/LogConcave/Density/GaussianSmoothing/Basic, Probability/Stein/Kernel.
Supporting theorems: gaussianLaw_translated_directional_identity_of_bounded_lipschitz_test, gaussian_product_source_stein_pairings_integrable, source_affine_stein_identity_of_bounded_lipschitz_test.
Previous: gaussian_product_source_stein_pairings_integrable · Next: gaussian_product_compact_dual_bound.
Source excerpt, lines 124–147 · Frozen source 408915f.
import Definitions.Def_KLSModular_FunctionalInequalities_KLS_LogConcave_Density_GaussianSmoothing_ProductCoupling open MeasureTheory open FunctionalInequalities.Probability.Stein open scoped ContDiff RealInnerProductSpace Topology set_option maxHeartbeats 800000
theorem FunctionalInequalities.KLS.LogConcave.GaussianSmoothing.gaussian_product_source_stein_identity_readable {n : ℕ}
{u h : EuclideanSpace ℝ (Fin n) → ℝ} {L G Lh : NNReal}
(hu : ContDiff ℝ 1 u) (hL : LipschitzWith L u)
(hG : LipschitzWith G (gradient u))
(hnorm : (∫ y, Real.exp (-u y) ∂volume) = 1)
(hh : ContDiff ℝ 1 h) (hhLip : LipschitzWith Lh h)
(C : ℝ) (hbound : ∀ x, ‖h x‖ ≤ C)
(S : EuclideanSpace ℝ (Fin n) →L[ℝ] EuclideanSpace ℝ (Fin n))
(v : EuclideanSpace ℝ (Fin n)) {s : ℝ} (hs : 0 < s) :
let P := ((volume : Measure (EuclideanSpace ℝ (Fin n))).tilted (-u)).prod
(FunctionalInequalities.KLS.LogConcave.GaussianSmoothing.gaussianLaw (V := EuclideanSpace ℝ (Fin n)) s)
Integrable (fun p : EuclideanSpace ℝ (Fin n) × EuclideanSpace ℝ (Fin n) =>
⟪S (gradient u p.1) + p.2, v⟫ * h (S (gradient u p.1) + p.2)) P ∧
Integrable (fun p : EuclideanSpace ℝ (Fin n) × EuclideanSpace ℝ (Fin n) =>
⟪S.adjoint v, action (hessianMatrix u p.1)
(S.adjoint (gradient h (S (gradient u p.1) + p.2)))⟫ +
(2 * s)⁻¹ * fderiv ℝ h (S (gradient u p.1) + p.2) v) P ∧
(∫ p : EuclideanSpace ℝ (Fin n) × EuclideanSpace ℝ (Fin n),
⟪S (gradient u p.1) + p.2, v⟫ * h (S (gradient u p.1) + p.2) ∂P) =
∫ p : EuclideanSpace ℝ (Fin n) × EuclideanSpace ℝ (Fin n),
⟪S.adjoint v, action (hessianMatrix u p.1)
(S.adjoint (gradient h (S (gradient u p.1) + p.2)))⟫ +
(2 * s)⁻¹ * fderiv ℝ h (S (gradient u p.1) + p.2) v ∂P := by sorry