Convergence of the dyadic reconstruction grid with geometric error
ProvedHairer.dyadic_grid_limit_existsanalysisregularity-structures
Let be a modelled distribution with positive regularity for a regularity structure and model, and write . Use the tensor-product smooth-transition partition of unity at scales , with weights and centres . For every smooth compactly supported test function , set
Only finitely many summands are nonzero. There exists a single linear distribution such that, for every such test,
for some independent of .
This constructs the limit of the dyadic grid approximations. The constant may depend on the fixed test function; the uniform local estimate over rescaled tests needed for the full reconstruction theorem is a separate statement.
Preamble
import Definitions.Def_Hairer_Model open BigOperators Filter Hairer open scoped Topology noncomputable section
Formal statement
theorem Hairer.dyadic_grid_limit_exists
{d : ℕ} {s : Fin d → ℕ} (hs : IsScaling s) {γ : ℝ} (hγ : 0 < γ)
{A : Set ℝ} {E : A → Type} [∀ a : A, NormedAddCommGroup (E a)]
[∀ a : A, NormedSpace ℝ (E a)]
{G : Subgroup (ModelSpace A E ≃ₗ[ℝ] ModelSpace A E)} {one : ModelSpace A E}
(hT : IsRegularityStructure A E G one)
{r : ℕ} {Pi : Pt d → ModelSpace A E →ₗ[ℝ] Distrib d}
{Gam : Pt d → Pt d → ModelSpace A E ≃ₗ[ℝ] ModelSpace A E}
(hmod : IsModel s r G Pi Gam)
{f : Pt d → ModelSpace A E} (hf : IsModelled s γ Gam f) :
let W : Pt d → ℝ := fun z ↦
∏ i, (Real.smoothTransition (z i + 1) - Real.smoothTransition (z i))
let δ : ℕ → ℝ := fun n ↦ (1 / 2 : ℝ) ^ (n + 2)
let X := fun (n : ℕ) (j : Fin d → ℤ) (i : Fin d) ↦ δ n ^ s i * (j i : ℝ)
let Rn := fun (n : ℕ) (φ : testFunctions d) ↦ ∑ᶠ j : Fin d → ℤ,
(Pi (X n j) (f (X n j))).eval
(fun y ↦ W (fun i ↦ y i / δ n ^ s i - (j i : ℝ)) * φ.val y)
∃ R : Distrib d,
(∀ φ : testFunctions d, Tendsto
(fun n ↦ Rn n φ) atTop (𝓝 (R φ))) ∧
∀ φ : testFunctions d, ∃ C : ℝ, 0 ≤ C ∧ ∀ n : ℕ,
|R φ - Rn n φ| ≤
C * ((1 / 2 : ℝ) ^ γ) ^ n := by sorry
Source
Auxiliary construction for M. Hairer, A theory of regularity structures (2014), Theorem 3.10. This formalization uses a tensor-product smooth-transition grid and establishes its pointwise distributional limit with a geometric tail estimate.