OAI/Analysis/DiskMaximal/Envelopes.lean
48 declarations, 0 contain sorry, 0 declared as axioms
- def
OAI.GeometricFilter.filtline 57Source documentation
Causal dyadic averaging, with both the present and the complete past.
def filt (c : ℕ → ℝ) : ℕ → ℝ | 0 => c 0 | i+1 => c (i+1)+filt c i/2
- lemma
OAI.GeometricFilter.filt_nonnegline 60 - lemma
OAI.GeometricFilter.le_filtline 64 - lemma
OAI.GeometricFilter.filt_leline 68 - lemma
OAI.GeometricFilter.filt_monoline 72 - lemma
OAI.GeometricFilter.filt_mulline 76 - lemma
OAI.GeometricFilter.filt_sumline 80 - lemma
OAI.GeometricFilter.filt_tail_strongline 97 - theorem
OAI.GeometricFilter.filt_tailline 115Source documentation
The complete pre-birth history costs a fixed amount, not the birth index.
theorem filt_tail {c : ℕ → ℝ} {A : ℝ} (hA : 0 ≤ A) (hc0 : ∀ i, 0 ≤ c i) (hcA : ∀ i, c i ≤ A) {b m : ℕ} (hbm : b ≤ m) : (∑ i ∈ Ico b (m+1), filt c i) ≤ 2*∑ i ∈ Ico b (m+1), c i+4*A := by …Mentions: OAI.GeometricFilter.filt
- theorem
OAI.GeometricFilter.double_tailline 123Source documentation
Two geometric convolutions are summable on every occupied lifetime.
theorem double_tail {c : ℕ → ℝ} {A : ℝ} (hA : 0 ≤ A) (hc0 : ∀ i, 0 ≤ c i) (hcA : ∀ i, c i ≤ A) {b m : ℕ} (hbm : b ≤ m) : (∑ i ∈ Ico b (m+1), filt (filt c) i) ≤ 4*∑ i ∈ Ico b (m+1), c i+16*A := by …Mentions: OAI.GeometricFilter.filt
- lemma
OAI.GeometricFilter.radius_sumline 133 - def
OAI.ActualLabels.spatialConstantline 151 - lemma
OAI.ActualLabels.spatialConstant_posline 152 - lemma
OAI.ActualLabels.spatialConstant_geline 156 - lemma
OAI.ActualLabels.scaled_marginal_distanceline 160 - theorem
OAI.ActualLabels.filtered_spatial_costline 175Source documentation
Actual spatial comparison at every point of the neighborhood. The gradient versions are pointwise bounded; differentiability is only needed later for IBP.
theorem filtered_spatial_cost (h : ℂ → ℂ) (r : ℕ → ℝ) (q : ℕ → ℤ × ℤ) (x y : ℂ) (j : ℕ) (hr : ∀ i ≤ j, 0 < r i) (hstep : ∀ i < j, r i = 2*r (i+1)) (hx : ∀ i ≤ j, boxNorm (x-point (r i) (q i)) < 199*r i) (hy : ∀ i ≤ j, boxNorm (y-point (r i) (q i)) < 199*r i) (hdist : ‖x-y‖ ≤ 104*r j) : switchCost h r x j+derivativeCost h r x j ≤ spatialConstant*filt (energyRate h r q y) j := by …Mentions: OAI.ActualLabels.switchCost, OAI.ActualLabels.derivativeCost, OAI.ActualLabels.spatialConstant, OAI.ActualLabels.energyRate
- lemma
OAI.DyadicHats.scale_monoline 229 - lemma
OAI.DyadicHats.descending_stepline 231 - lemma
OAI.DyadicHats.descending_compareline 234 - def
OAI.DyadicHats.neighborhoodline 239 - lemma
OAI.DyadicHats.point_mem_neighborhoodline 240 - def
OAI.ActualMarkTrees.pathVertexline 249 - lemma
OAI.ActualMarkTrees.terminal_path_localityline 251 - lemma
OAI.ActualMarkTrees.path_neighborhood_geometryline 264Source documentation
The neighborhoods used by assembly remain within the actual 200-neighbor energy window at every earlier row, including the complete pre-birth history.
lemma path_neighborhood_geometry {N : ℕ} (v : Grid N → ℂ) {r : ℝ} (hr : 0 < r) {m : ℕ} {q : Grid N} {p : List (Grid N)} (hp : activePath (tree v m 1 q) p) {l : ℕ} (hl : l ≤ m) {x : ℂ} (hx : x ∈ neighborhood (scale r (m-l)) (pathVertex q p l)) {i : ℕ} (hi : i ≤ m) (hil : i ≤ l+1) : ‖x-point r (lastVertex q p)‖ ≤ 104*scale r (m-i) ∧ boxNorm (x-point (scale r (m-i)) (pathVertex q p i)) < 199*scale r (m-i) ∧ boxNorm (point r (lastVertex q p)-point (scale r (m-i)) (pathVertex q p i)) < 199*scale r (m-i) := by …Mentions: OAI.ActualMarkTrees.tree, OAI.ActualMarkTrees.pathVertex, OAI.ActualMarkTrees.lastVertex
- lemma
OAI.GeometricFilter.filt_mono_online 291 - lemma
OAI.GeometricFilter.shifted_tailline 295 - def
OAI.DyadicHats.localSupline 305 - lemma
OAI.DyadicHats.localSup_leline 307 - lemma
OAI.DyadicHats.le_localSupline 312 - lemma
OAI.DyadicHats.localSup_nonnegline 315 - lemma
OAI.ActualLabels.derivativeCost_nonnegline 322 - def
OAI.ActualLabels.cellCostline 328 - lemma
OAI.ActualLabels.cellCost_nonnegline 333 - lemma
OAI.ActualMarkTrees.path_filtered_costline 347 - lemma
OAI.ActualMarkTrees.path_cellCostline 361 - lemma
OAI.ActualMarkTrees.path_cellCost_sumline 399 - def
OAI.ActualMarkTrees.packingChargeline 425 - lemma
OAI.ActualMarkTrees.packingCharge_pathline 428 - lemma
OAI.ActualMarkTrees.tree_packing_boundline 458 - def
OAI.ActualMarkTrees.packingConstantline 470 - lemma
OAI.ActualMarkTrees.packingConstant_posline 473 - theorem
OAI.ActualMarkTrees.spatial_packingline 478 - def
OAI.CommonPrices.rootMassline 514 - lemma
OAI.CommonPrices.rootMass_nonnegline 516 - def
OAI.CommonPrices.sizeErrorsline 519 - lemma
OAI.CommonPrices.sizeErrors_priceline 525 - lemma
OAI.CommonPrices.sizeErrors_gradientline 544 - theorem
OAI.CommonPrices.size_cell_priceline 561Source documentation
The endpoint-bin gradient price, with all root terms retained rather than silently using the cone estimate at one-sided maximizing radii.
theorem size_cell_price : ∃ C : ℝ, 0 < C ∧ ∀ {N : ℕ} {r R : ℝ} (hr : 0 < r) (hR : 0 ≤ R) (hN : R/r+1 ≤ (N:ℝ)) {g : ℂ → ℝ}, ContDiff ℝ 1 g → Integrable (gradient g) → (∀ y, gradient g y ≠ 0 → ‖y‖ ≤ R) → ∀ (k : ℕ) (p : ℤ × ℤ) {n : ℂ}, n ∈ fixedPalette.labels → ∀ Q : Set ℂ, volume Q ≤ ENNReal.ofReal 1 → ∃ F : ℂ → ℝ≥0∞, Measurable F ∧ (∫⁻ X in Q, F X) ≤ ENNReal.ofReal (C*(scale r k)⁻¹* (rootMass (gradient g) r k (cellRoots N p)+cellPrice N r (gradient g) k (cellRoots N p))) ∧ ∀ X, ‖X‖ ≤ 2 → ∀ t, EndpointRadius (fun Y => g (point (scale r k) p+scale r k • Y)) X t → ENNReal.ofReal ‖∫ Y in Metric.ball X t, gradient (fun Z => g (point (scale r k) p+scale r k • Z)) Y‖ ≤ F X := by …Mentions: OAI.CommonPrices.rootMass, OAI.CommonPrices.cellRoots, OAI.CommonPrices.cellPrice