Documentation

LeanPool.AnomalousDiffusion.AVenhance.Infra.Section5.RelativeError.RelativeInitialDefectPositiveTransport

Backward transport on positive windows. No continuity at zero is needed when the source integral is uniformly bounded on every such window.

theorem AVenhance.Infra.Section5.RelativeError.relative_initial_backward_transport_integrated_pos {F source : ℝ → Homogenization.Vec 2 → ℝ} (X : ℝ → Ergodic.PeriodicVolumePreservingDiffeomorphism 2) (hFper : ∀ (t : ℝ), IsZ2Periodic (F t)) (hSourcePer : ∀ (t : ℝ), IsZ2Periodic (source t)) (hComp : ContinuousOn (fun (p : ℝ × Homogenization.Vec 2) => F p.1 ((X p.1).toFun p.2)) (Set.Ioi 0 ×ˢ Set.univ)) (hSourceComp : ContinuousOn (fun (p : ℝ × Homogenization.Vec 2) => source p.1 ((X p.1).toFun p.2)) (Set.Ioi 0 ×ˢ Set.univ)) (hMaterial : ∀ (t : ℝ), 0 < t → ∀ (x : Homogenization.Vec 2), HasDerivAt (fun (s : ℝ) => F s ((X s).toFun x)) (source t ((X t).toFun x)) t) {b D : ℝ} (hb : 0 < b) (hD : 0 ≤ D) (hzero : F b = 0) (hSourceIntegral : ∀ t ∈ Set.Ioc 0 b, ∫ (s : ℝ) in t..b, √(l2NormSq (source s)) ≤ D) (t : ℝ) :
t ∈ Set.Ioc 0 b → √(l2NormSq (F t)) ≤ 2 * D

Backward FTC and compact maximum absorption control every positive slice. The source primitive is bounded on [t,b], so the argument never uses time zero.

theorem AVenhance.Infra.Section5.RelativeError.relative_initial_source_L1_of_positive_L2 {t b D : ℝ} (ht : 0 < t) (htb : t ≤ b) (hD : 0 ≤ D) {g : ℝ → ℝ} (hg : ContinuousOn g (Set.Icc t b)) (hquad : ∫ (s : ℝ) in t..b, g s ^ 2 ≤ D ^ 2) :
∫ (s : ℝ) in t..b, g s ≤ √b * D

Time Cauchy--Schwarz on a positive subwindow, with an abstract source amplitude. No source trace at zero is assumed.