Download data/lean/OAI/Analysis/Fourier/Restriction/DensityLayers.lean from introvoyz045/math: direct link, hf CLI and curl.
- Browser
- Download file 1.82 kB
-
https://huggingface.co/datasets/introvoyz045/math/resolve/main/data/lean/OAI/Analysis/Fourier/Restriction/DensityLayers.lean
- Command line
-
hf download hf://datasets/introvoyz045/math/data/lean/OAI/Analysis/Fourier/Restriction/DensityLayers.lean
-
curl -L -o DensityLayers.lean https://huggingface.co/datasets/introvoyz045/math/resolve/main/data/lean/OAI/Analysis/Fourier/Restriction/DensityLayers.lean
1.82 kB
| import OAI.Analysis.Fourier.Restriction.Sparse.Counting | |
| import Mathlib.Data.Nat.Find | |
| noncomputable section | |
| namespace OAI.Analysis.Fourier.Restriction | |
| /-- Every point has a first density scale. This elementary stopping-time | |
| step is the source of the small number of sparse families in epsilon removal. -/ | |
| theorem exists_density_layer {X : Type*} [PseudoMetricSpace X] [DecidableEq X] | |
| (E : Finset X) (K N : ℕ) (hN : 0 < N) (hcard : E.card ≤ K ^ N) | |
| (r : ℕ → ℝ) (hr0 : 0 < r 0) (x : X) (hx : x ∈ E) : | |
| ∃ j < N, | |
| K ^ j ≤ (E.filter (fun y ↦ dist x y < 2 * r j)).card ∧ | |
| (E.filter (fun y ↦ dist x y < 2 * r (j + 1))).card ≤ K ^ (j + 1) := by | |
| classical | |
| let P (k : ℕ) : Prop := 0 < k ∧ k ≤ N ∧ | |
| (E.filter (fun y ↦ dist x y < 2 * r k)).card ≤ K ^ k | |
| have hex : ∃ k, P k := ⟨N, hN, le_rfl, (Finset.card_filter_le _ _).trans hcard⟩ | |
| let k := Nat.find hex | |
| have hk : P k := Nat.find_spec hex | |
| have hkpos : 0 < k := hk.1 | |
| have hkle : k ≤ N := hk.2.1 | |
| have hnext : k - 1 + 1 = k := by omega | |
| refine ⟨k - 1, by omega, ?_, ?_⟩ | |
| · by_cases hkone : k = 1 | |
| · have hcount : 0 < (E.filter (fun y ↦ dist x y < 2 * r 0)).card := by | |
| apply Finset.card_pos.mpr | |
| exact ⟨x, Finset.mem_filter.mpr ⟨hx, by simpa using (mul_pos (by norm_num : (0 : ℝ) < 2) hr0)⟩⟩ | |
| convert Nat.succ_le_of_lt hcount using 1 <;> simp [hkone] | |
| · have hnot : ¬ P (k - 1) := Nat.find_min hex (by omega) | |
| have hkminus : 0 < k - 1 := by omega | |
| have hkminusN : k - 1 ≤ N := by omega | |
| have hlarge : K ^ (k - 1) < (E.filter (fun y ↦ dist x y < 2 * r (k - 1))).card := by | |
| by_contra! h | |
| exact hnot ⟨hkminus, hkminusN, h⟩ | |
| exact hlarge.le | |
| · simpa only [hnext] using hk.2.2 | |
| end OAI.Analysis.Fourier.Restriction | |