OAI/Analysis/DiskMaximal/Telescoping.lean
40 declarations, 0 contain sorry, 0 declared as axioms
- lemma
OAI.DiskScales.width_addline 22 - lemma
OAI.DiskScales.sum_width_tailline 26 - lemma
OAI.DiskScales.dist_width_tailline 35 - lemma
OAI.LocalCancellation.moment_kernel_pairing_subline 50 - lemma
OAI.LocalCancellation.total_correctionline 61 - lemma
OAI.LocalCancellation.balanced_kernel_gap_identityline 72 - lemma
OAI.LocalCancellation.kernel_error_boundline 114 - lemma
OAI.LocalCancellation.coarse_bound_scaledline 127 - lemma
OAI.LocalCancellation.scaled_ordinary_inactiveline 133 - lemma
OAI.LocalCancellation.scaled_corrected_inactiveline 148 - lemma
OAI.LocalCancellation.scaled_corrected_activeline 163 - lemma
OAI.DiskScales.successor_discountline 181 - lemma
OAI.DiskScales.geometric_den_posline 189 - lemma
OAI.DiskScales.coarse_error_scaleline 192 - lemma
OAI.LocalCancellation.inactive_kernel_distanceline 224 - lemma
OAI.LocalCancellation.kernel_gap_errorsline 242 - lemma
OAI.LocalCancellation.corrected_kernel_gapline 289 - def
OAI.DiskScales.errorGapConstantline 338 - def
OAI.DiskScales.directionGapConstantline 341 - lemma
OAI.DiskScales.errorGapConstant_posline 344 - lemma
OAI.DiskScales.directionGapConstant_geline 349 - lemma
OAI.DiskScales.directionGapConstant_posline 357 - lemma
OAI.DiskScales.direction_cap_scaleline 362 - lemma
OAI.LocalCancellation.direction_sequence_pointline 396Source documentation
The full finite/infinite active-width argument at a fixed center. The cap premise is exactly the already proved shell concentration output, not a radius selection or a projection estimate.
lemma direction_sequence_point {Cs κ d t r : ℝ} {e : ℂ} (a : ConeDensity e κ) (b : ι → ErrorPiece Cs) (he : ‖e‖ = 1) (hκ : 0 < κ) (hCs : 0 ≤ Cs) (hd : 0 < d) (hdq : d ≤ 1/4) (hr : r ∈ Icc (1 : ℝ) 2) (hAt : errorPrice b (1/100) ≤ t) (k : ∀ n : ℕ, RadiusDensity (r-width d n) (r+width d n) ⟨216/width d n, div_nonneg (by norm_num) (width_pos hd n).le⟩) (x : ℂ) (hM : ∀ n, (∫ y, star (y-x)*(kernel (k n) x y : ℂ)*a.density y)+ (∑ v, ∫ y, star (y-x)*(kernel (k n) x y : ℂ)*(b v).density y) = 0) (hcap : ∀ j, t*(width d j)^(1/10 : ℝ) ≤ coarseMass b (width d j) (thinShell x r (width d j)) → (∫ y in Metric.ball x (r-width d j), ‖a.density y‖) ≤ (errorGapConstant Cs/κ)*t*(width d j)^(-(1/50 : ℝ)) → ∃ z ∈ thinShell x r (width d j), …Mentions: OAI.LocalCancellation.ConeDensity, OAI.LocalCancellation.ErrorPiece, OAI.LocalCancellation.errorPrice, OAI.LocalCancellation.coarseMass
- def
OAI.LocalCancellation.directionShellBoundline 531 - theorem
OAI.LocalCancellation.direction_sequenceline 543Source documentation
Radius-independent exceptional union for the full directional sequence. The cap property is produced by shell concentration for the actual densities.
theorem direction_sequence {e : ℂ} {κ Cs d t : ℝ} (a : ConeDensity e κ) (b : ι → ErrorPiece Cs) (he : ‖e‖ = 1) (hκ : 0 < κ) (hCs : 0 ≤ Cs) (hd : 0 < d) (hdq : d ≤ 1/4) (hgeom : ∀ δ : ℝ, 0 < δ → δ ≤ d → δ ≤ (δ^(1/5 : ℝ))^2/1600) (ht : 0 < t) (hAt : errorPrice b (1/100) ≤ t) : ∃ E : Set ℂ, MeasurableSet E ∧ volume E ≤ ∑' n : ℕ, directionShellBound b κ t (width d n) ∧ ∀ x ∉ E, ∀ r ∈ Icc (1 : ℝ) 2, ∀ k : ∀ n : ℕ, RadiusDensity (r-width d n) (r+width d n) ⟨216/width d n, div_nonneg (by norm_num) (width_pos hd n).le⟩, (∀ n, (∫ y, star (y-x)*(kernel (k n) x y : ℂ)*a.density y)+ (∑ v, ∫ y, star (y-x)*(kernel (k n) x y : ℂ)*(b v).density y) = 0) → …Mentions: OAI.LocalCancellation.ConeDensity, OAI.LocalCancellation.ErrorPiece, OAI.LocalCancellation.errorPrice, OAI.LocalCancellation.directionShellBound
- def
OAI.DiskMoments.NeighborhoodRadiusline 597 - lemma
OAI.DiskMoments.neighborhood_kernelline 601 - lemma
OAI.LocalCancellation.mixed_moment_pairing_decompositionline 619 - lemma
OAI.LocalCancellation.mixed_ball_pairing_decompositionline 638 - theorem
OAI.LocalCancellation.direction_rawline 651Source documentation
Directional estimate at all genuine neighborhood-maximizing radii, with only the original local density equality.
theorem direction_raw {e : ℂ} {κ Cs d t : ℝ} (a : ConeDensity e κ) (b : ι → ErrorPiece Cs) (he : ‖e‖ = 1) (hκ : 0 < κ) (hCs : 0 ≤ Cs) (hd : 0 < d) (hdq : d ≤ 1/4) (hgeom : ∀ δ : ℝ, 0 < δ → δ ≤ d → δ ≤ (δ^(1/5 : ℝ))^2/1600) (ht : 0 < t) (hAt : errorPrice b (1/100) ≤ t) {g : ℂ → ℝ} (hg : ContDiff ℝ 1 g) : ∃ E : Set ℂ, MeasurableSet E ∧ volume E ≤ ∑' n : ℕ, directionShellBound b κ t (width d n) ∧ ∀ x ∉ E, (∀ᵐ y, y ∈ Metric.ball x 3 → a.density y+∑ j, (b j).density y = gradient g y) → ∀ r, NeighborhoodRadius g x r → max 0 (-inner ℝ e (∫ y in Metric.ball x r, gradient g y)) ≤ …Mentions: OAI.LocalCancellation.ConeDensity, OAI.LocalCancellation.ErrorPiece, OAI.LocalCancellation.errorPrice, OAI.LocalCancellation.directionShellBound
- lemma
OAI.DiskScales.third_power_identityline 680 - lemma
OAI.DiskScales.direction_shell_tail_scaledline 686 - lemma
OAI.DiskScales.direction_start_identityline 738 - lemma
OAI.DiskScales.exists_direction_parametersline 760 - def
OAI.DiskScales.directionTailConstantline 786 - lemma
OAI.DiskScales.directionTailConstant_posline 788 - theorem
OAI.LocalCancellation.direction_tailline 801Source documentation
A genuine superlevel bound with exponent 34/25 > 1 for all neighborhood maxima, independent of the number of positive or cancelling pieces.
theorem direction_tail {Cs κ : ℝ} (hCs : 0 ≤ Cs) (hκ : 0 < κ) : ∃ L : ℝ, 1 ≤ L ∧ ∀ {ι : Type*} [Fintype ι] {e : ℂ} (a : ConeDensity e κ) (b : ι → ErrorPiece Cs), ‖e‖ = 1 → 0 < errorPrice b (1/100) → ∀ t : ℝ, L*errorPrice b (1/100) ≤ t → ∀ g : ℂ → ℝ, ContDiff ℝ 1 g → ∃ E : Set ℂ, MeasurableSet E ∧ volume E ≤ ENNReal.ofReal (directionTailConstant Cs κ*(errorPrice b (1/100)/t)^(34/25 : ℝ)) ∧ ∀ x ∉ E, (∀ᵐ y, y ∈ Metric.ball x 3 → a.density y+∑ j, (b j).density y = gradient g y) → ∀ r, NeighborhoodRadius g x r → max 0 (-inner ℝ e (∫ y in Metric.ball x r, gradient g y)) ≤ t := by …Mentions: OAI.LocalCancellation.ConeDensity, OAI.LocalCancellation.ErrorPiece, OAI.LocalCancellation.errorPrice
- lemma
OAI.LocalCancellation.cone_ball_nonnegline 861 - theorem
OAI.LocalCancellation.direction_majorantline 870Source documentation
Full Borel integrable directional price. This is the hard local cancellation estimate, not an assumed maximal-operator endpoint.
theorem direction_majorant {Cs κ : ℝ} (hCs : 0 ≤ Cs) (hκ : 0 < κ) : ∃ C : ℝ, 0 < C ∧ ∀ {ι : Type*} [Fintype ι] {e : ℂ} (a : ConeDensity e κ) (b : ι → ErrorPiece Cs), ‖e‖ = 1 → ∀ g : ℂ → ℝ, ContDiff ℝ 1 g → ∀ Q : Set ℂ, volume Q ≤ ENNReal.ofReal 1 → ∃ F : ℂ → ℝ≥0∞, Measurable F ∧ (∫⁻ x in Q, F x) ≤ ENNReal.ofReal (C*errorPrice b (1/100)) ∧ ∀ x, (∀ᵐ y, y ∈ Metric.ball x 3 → a.density y+∑ j, (b j).density y = gradient g y) → ∀ r, NeighborhoodRadius g x r → ENNReal.ofReal (max 0 (-inner ℝ e (∫ y in Metric.ball x r, gradient g y))) ≤ F x := by …Mentions: OAI.LocalCancellation.ConeDensity, OAI.LocalCancellation.ErrorPiece, OAI.LocalCancellation.errorPrice