Documentation

LeanPool.Superdiffusion.Support.HighContrast.Coupled.Stampacchia.LevelRecursion

LevelRecursion #

Ported from scottnarmstrong/CoarseGraining at c7ddd76c08ade64fed1b8d2ca51be14dfee8deb4. Module imports, support namespace and local option directives adapted for Lean Pool.

Chebyshev level bound and the recursion assembly #

Two purely analytic helpers used by the generic De Giorgi core (Stampacchia/DeGiorgiCore.lean):

Both are proved at default heartbeats, no sorry.

theorem HCPolySupport.real_chebyshev_level {α : Type u_1} {m0 : MeasurableSpace α} {μ : MeasureTheory.Measure α} {p : ENNReal} (hp0 : p ≠ 0) (hptop : p ≠ ⊤) {h : α → ℝ} (hfin : MeasureTheory.eLpNorm h p μ ≠ ⊤) {ε : ℝ} (hε : 0 ≤ ε) {S : Set α} (hSsub : ∀ x ∈ S, ε ≤ h x) :
ε * (μ S).toReal ^ (1 / p.toReal) ≤ (MeasureTheory.eLpNorm h p μ).toReal

Real Chebyshev level bound. If S ⊆ {x | ε ≤ h x} with ε ≥ 0 and ‖h‖_{L^p(μ)} < ∞, then ε · (μ S)^{1/p} ≤ ‖h‖_{L^p(μ)} in real numbers.

theorem HCPolySupport.sq_level_recursion_of_le {ε r a b N₁ N₂ R : ℝ} (hε : 0 ≤ ε) (hr0 : 0 ≤ r) (hr1 : r ≤ 1) (ha : 0 ≤ a) (hb : 0 ≤ b) (hN₁ : ε * a ^ r ≤ N₁) (hN₂ : ε * b ^ r ≤ N₂) (hsum : N₁ + N₂ ≤ R) (hR : 0 ≤ R) :
ε ^ 2 * (a + b) ^ (2 * r) ≤ R ^ 2

Squaring step of the level recursion. From the two per-copy Chebyshev bounds ε · (vol S₁)^{1/p} ≤ N₁, ε · (vol S₂)^{1/p} ≤ N₂ and a bound N₁ + N₂ ≤ R, together with 0 ≤ ε and 0 < 1/p ≤ 1, we obtain ε² · (vol S₁ + vol S₂)^{2/p} ≤ R².