Squared Gradient Bound for a Scaled Cutoff Product
ProvedFunctionalInequalities.Poincare.SmoothTest.norm_gradient_scaled_mul_sq_le_modularThis is a pointwise estimate for the gradient of a function multiplied by the designated scaled cutoff.
Let be a finite-dimensional complete real inner product space. Denote the fixed bump Energy.Euclidean.LipschitzApproximation.bump by , and its scaled cutoff Energy.Euclidean.LipschitzApproximation.scaled R by . Assume:
- and .
- for every , where the norm is the operator norm of the Fréchet derivative.
- is differentiable at .
Then
This technical estimate controls the energy contribution introduced by cutoff multiplication. It requires no measure or global differentiability assumption on .
Superseded upload. This snapshot was replaced by FunctionalInequalities.Poincare.SmoothTest.norm_gradient_scaled_mul_sq_le_readable: open the readable replacement. The replacement preserves the frozen source and mathematical statement, with a compact declaration context. This historical node remains available for existing imports.
import Definitions.Def_KLSModular_FunctionalInequalities_Poincare_SmoothTest_WeakLimit
noncomputable section
open MeasureTheory ProbabilityTheory Filter Topology
open scoped ContDiff
section
namespace FunctionalInequalities
end FunctionalInequalities
open FunctionalInequalities
section
namespace FunctionalInequalities.Poincare
end FunctionalInequalities.Poincare
open FunctionalInequalities FunctionalInequalities.Poincare
section
namespace FunctionalInequalities.Poincare.SmoothTest
end FunctionalInequalities.Poincare.SmoothTest
open FunctionalInequalities FunctionalInequalities.Poincare FunctionalInequalities.Poincare.SmoothTest
variable {E : Type*} [NormedAddCommGroup E] [InnerProductSpace ℝ E] [CompleteSpace E]
[MeasurableSpace E] [BorelSpace E]
section Cutoff
variable [FiniteDimensional ℝ E]
omit [MeasurableSpace E] [BorelSpace E]
theorem FunctionalInequalities.Poincare.SmoothTest.norm_gradient_scaled_mul_sq_le_modular {R M : ℝ} (hR : 0 < R)
(hM : ∀ x : E, ‖fderiv ℝ (⇑(Energy.Euclidean.LipschitzApproximation.bump (E := E))) x‖ ≤ M)
{f : E → ℝ} {x : E} (hf : DifferentiableAt ℝ f x) :
‖gradient (fun y => Energy.Euclidean.LipschitzApproximation.scaled R y * f y) x‖ ^ 2 ≤
2 * ‖gradient f x‖ ^ 2 + 2 * (M / R) ^ 2 * f x ^ 2 := by sorry