Centered Classes Lie In The Diffusion Image Closure
ProvedFunctionalInequalities.KLS.LogConcave.HeatSemigroup.centered_mem_closure_compactSmoothDiffusionImage_readableThis formalizes the mean-zero density statement recorded as Lemma A.6 of Letwin (arXiv:2607.24164v1), with the reviewed relation recorded as partial.
Let be finite-dimensional, let be an additive Haar measure, let be log-concave with
where . If has zero mean,
then belongs to the closure in of the literal images of smooth compactly supported functions under the diffusion operator compactSmoothDiffusionImage:
The conclusion is a closure statement for every mean-zero class; it does not state the Brascamp–Lieb inequality itself.
Original Lean declaration header
Within namespace FunctionalInequalities.KLS.LogConcave.HeatSemigroup. Shown in its original module context. Proof omitted.
theorem centered_mem_closure_compactSmoothDiffusionImage
{ν : ProbabilityMeasure V} (m : Measure V) [m.IsAddHaarMeasure]
(hν : FunctionalInequalities.KLS.LogConcave ν) {U : V → ℝ}
(hνU : (ν : Measure V) = m.tilted (-U)) (hU : ContDiff ℝ 1 U)
(g : Lp ℝ 2 (ν : Measure V)) (hg : L2.mean ν g = 0) :
g ∈ closure (compactSmoothDiffusionImage hU)
Module navigation
Module: KLS/LogConcave/HeatSemigroup/Range.
Definition modules: L2/MeanVariance, Probability/LogConcave.
Supporting theorems: mem_closure_range_of_inner_eq_zero, eq_const_mean_of_closedWeightedOperator_eq_zero, exists_seq_smooth_tendsto_adjointComp_of_localRegularity, interior_support_nonempty_of_eq_tilted, toLp_mul_mem_closedWeightedOperator_domain_of_orthogonal.
Source excerpt, lines 45–50 · Frozen source 408915f.
import Definitions.Def_KLSModular_FunctionalInequalities_KLS_LogConcave_HeatSemigroup_Range
open MeasureTheory Filter Topology Set InnerProductSpace
variable {V : Type*} [NormedAddCommGroup V] [InnerProductSpace ℝ V] [CompleteSpace V]
[MeasurableSpace V] [BorelSpace V] [SecondCountableTopology V] [FiniteDimensional ℝ V]
theorem FunctionalInequalities.KLS.LogConcave.HeatSemigroup.centered_mem_closure_compactSmoothDiffusionImage_readable
{ν : ProbabilityMeasure V} (m : Measure V) [m.IsAddHaarMeasure]
(hν : FunctionalInequalities.KLS.LogConcave ν) {U : V → ℝ}
(hνU : (ν : Measure V) = m.tilted (-U)) (hU : ContDiff ℝ 1 U)
(g : Lp ℝ 2 (ν : Measure V)) (hg : L2.mean ν g = 0) :
g ∈ closure (compactSmoothDiffusionImage hU) := by sorry