OAI/Analysis/Brenier/CellGeometry.lean
23 declarations, 0 contain sorry, 0 declared as axioms
- def
OAI.Problem358.CellGeometry.closedCellline 16Source documentation
The closed cell of one affine branch, including ties.
def closedCell (K : Set (E d)) (y : ι → E d) (c : ι → ℝ) (i : ι) : Set (E d) := …
- theorem
OAI.Problem358.CellGeometry.closedCell_subsetline 19 - theorem
OAI.Problem358.CellGeometry.isClosed_closedCellline 22 - theorem
OAI.Problem358.CellGeometry.isCompact_closedCellline 33 - theorem
OAI.Problem358.CellGeometry.convex_closedCellline 38 - theorem
OAI.Problem358.CellGeometry.exists_mem_closedCellline 49 - theorem
OAI.Problem358.CellGeometry.iUnion_closedCellline 56 - theorem
OAI.Problem358.CellGeometry.exists_strict_active_interiorline 70Source documentation
Positive-volume cells have strictly active points in the interior, once affine ties have been removed as a null set.
theorem exists_strict_active_interior {K : Set (E d)} (hconv : Convex ℝ K) (y : ι → E d) (c : ι → ℝ) (i : ι) (hpos : volume (closedCell K y c i) ≠ 0) (hties : ∀ᵐ x ∂volume, ∀ j, j ≠ i → inner ℝ x (y j) + c j ≠ inner ℝ x (y i) + c i) : ∃ x ∈ interior K, ∀ j, j ≠ i → inner ℝ x (y j) + c j < inner ℝ x (y i) + c i := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.combo_mem_closedCellline 89Source documentation
Convex interpolation of the intercepts and source points remains in the same cell.
theorem combo_mem_closedCell {K : Set (E d)} (hK : Convex ℝ K) (y : ι → E d) (c e : ι → ℝ) (i : ι) {x z : E d} (hx : x ∈ closedCell K y c i) (hz : z ∈ closedCell K y e i) {a b : ℝ} (ha : 0 ≤ a) (hb : 0 ≤ b) (hab : a + b = 1) : a • x + b • z ∈ closedCell K y (fun j => a * c j + b * e j) i := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.midpoint_mem_closedCellline 102Source documentation
The opposite-intercept midpoint inclusion used in the moment estimate.
theorem midpoint_mem_closedCell {K : Set (E d)} (hK : Convex ℝ K) (y : ι → E d) (c q : ι → ℝ) (i : ι) (s : ℝ) {x z : E d} (hx : x ∈ closedCell K y (fun j => c j - s * q j) i) (hz : z ∈ closedCell K y (fun j => c j + s * q j) i) : (1 / 2 : ℝ) • x + (1 / 2 : ℝ) • z ∈ closedCell K y c i := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.exists_affine_equality_interiorline 118Source documentation
An affine equality plane separating two strictly active interior points meets the interior of the convex source.
theorem exists_affine_equality_interior {K : Set (E d)} (hK : Convex ℝ K) {y z : E d} {c e : ℝ} {p q : E d} (hp : p ∈ interior K) (hq : q ∈ interior K) (hp_lt : inner ℝ p y + c < inner ℝ p z + e) (hq_lt : inner ℝ q z + e < inner ℝ q y + c) : ∃ x ∈ interior K, inner ℝ x y + c = inner ℝ x z + e := by … - theorem
OAI.Problem358.CellGeometry.pairwise_aedisjoint_closedCellline 149Source documentation
Distinct-site closed cells overlap only along null affine tie planes.
theorem pairwise_aedisjoint_closedCell {ρ : Measure (E d)} (hρ : ρ ≪ volume) (K : Set (E d)) (y : ι → E d) (c : ι → ℝ) (hy : Function.Injective y) : Pairwise (fun i j => AEDisjoint ρ (closedCell K y c i) (closedCell K y c j)) := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.sum_measure_closedCellline 162Source documentation
The closed cells form a finite partition for all measures absolutely continuous with respect to volume.
theorem sum_measure_closedCell [Fintype ι] [Nonempty ι] {ρ : Measure (E d)} (hρ : ρ ≪ volume) {K : Set (E d)} (hK : IsClosed K) (y : ι → E d) (c : ι → ℝ) (hy : Function.Injective y) : (∑ i, ρ (closedCell K y c i)) = ρ K := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.sum_integral_closedCellline 171Source documentation
Integrable functions can be summed cellwise despite the shared boundary faces.
theorem sum_integral_closedCell [Fintype ι] [Nonempty ι] {V : Type*} [NormedAddCommGroup V] [NormedSpace ℝ V] {ρ : Measure (E d)} (hρ : ρ ≪ volume) {K : Set (E d)} (hK : IsClosed K) (y : ι → E d) (c : ι → ℝ) (hy : Function.Injective y) {f : E d → V} (hf : IntegrableOn f K ρ) : (∑ i, ∫ x in closedCell K y c i, f x ∂ρ) = ∫ x in K, f x ∂ρ := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.sum_real_measure_closedCellline 185Source documentation
Real-valued cell masses also add to the total source mass.
theorem sum_real_measure_closedCell [Fintype ι] [Nonempty ι] {ρ : Measure (E d)} [IsFiniteMeasure ρ] (hρ : ρ ≪ volume) {K : Set (E d)} (hK : IsClosed K) (y : ι → E d) (c : ι → ℝ) (hy : Function.Injective y) : (∑ i, ρ.real (closedCell K y c i)) = ρ.real K := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.exists_strict_active_interior_of_injectiveline 196Source documentation
Injective sites discharge the null-tie premise in strict activation.
theorem exists_strict_active_interior_of_injective [Countable ι] {K : Set (E d)} (hK : Convex ℝ K) (y : ι → E d) (c : ι → ℝ) (hy : Function.Injective y) (i : ι) (hpos : volume (closedCell K y c i) ≠ 0) : ∃ x ∈ interior K, ∀ j, j ≠ i → inner ℝ x (y j) + c j < inner ℝ x (y i) + c i := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.exists_strict_active_interior_of_massline 208Source documentation
Positive cell mass for an absolutely continuous source already guarantees a strictly active interior point.
theorem exists_strict_active_interior_of_mass [Countable ι] {ρ : Measure (E d)} (hρ : ρ ≪ volume) {K : Set (E d)} (hK : Convex ℝ K) (y : ι → E d) (c : ι → ℝ) (hy : Function.Injective y) (i : ι) (hpos : ρ (closedCell K y c i) ≠ 0) : ∃ x ∈ interior K, ∀ j, j ≠ i → inner ℝ x (y j) + c j < inner ℝ x (y i) + c i := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.exists_equality_interior_of_positive_cellsline 218Source documentation
Every equality plane between positive-mass cells cuts the interior of the source.
theorem exists_equality_interior_of_positive_cells [Countable ι] {ρ : Measure (E d)} (hρ : ρ ≪ volume) {K : Set (E d)} (hK : Convex ℝ K) (y : ι → E d) (c : ι → ℝ) (hy : Function.Injective y) {i j : ι} (hij : i ≠ j) (hi : ρ (closedCell K y c i) ≠ 0) (hj : ρ (closedCell K y c j) ≠ 0) : ∃ x ∈ interior K, inner ℝ x (y i) + c i = inner ℝ x (y j) + c j := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.ae_fiber_eq_closedCellline 244Source documentation
With distinct slopes, any maximizing index selects the closed affine cell almost everywhere. No measurability of the supplied selector is required.
theorem ae_fiber_eq_closedCell [Countable ι] {K : Set (E d)} (hK : MeasurableSet K) (y : ι → E d) (c : ι → ℝ) (hy : Function.Injective y) (k : E d → ι) (hk : ∀ᵐ x ∂uniformMeasure K, ∀ j, ⟪x, y j⟫ + c j ≤ ⟪x, y (k x)⟫ + c (k x)) (i : ι) : {x | k x = i} =ᵐ[uniformMeasure K] closedCell K y c i := by …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.measure_fiber_eq_closedCellline 262Source documentation
Closed-cell masses are exactly the masses of any maximizing selector.
theorem measure_fiber_eq_closedCell [Countable ι] {K : Set (E d)} (hK : MeasurableSet K) (y : ι → E d) (c : ι → ℝ) (hy : Function.Injective y) (k : E d → ι) (hk : ∀ᵐ x ∂uniformMeasure K, ∀ j, ⟪x, y j⟫ + c j ≤ ⟪x, y (k x)⟫ + c (k x)) (i : ι) : uniformMeasure K {x | k x = i} = uniformMeasure K (closedCell K y c i) := …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.real_fiber_eq_closedCellline 273Source documentation
The real-valued mass form used in the moving-potential energy estimate.
theorem real_fiber_eq_closedCell [Countable ι] {K : Set (E d)} (hK : MeasurableSet K) (y : ι → E d) (c : ι → ℝ) (hy : Function.Injective y) (k : E d → ι) (hk : ∀ᵐ x ∂uniformMeasure K, ∀ j, ⟪x, y j⟫ + c j ≤ ⟪x, y (k x)⟫ + c (k x)) (i : ι) : (uniformMeasure K).real {x | k x = i} = (uniformMeasure K).real (closedCell K y c i) := …Mentions: OAI.Problem358.CellGeometry.closedCell
- theorem
OAI.Problem358.CellGeometry.exists_joint_measurable_selector_closedCell_massesline 286Source documentation
A single globally measurable selector works for all parameters. Its masses agree with closed-cell masses at every parameter with distinct slopes, so no injectivity at endpoint parameters or outside the time interval is imposed.
theorem exists_joint_measurable_selector_closedCell_masses {P : Type*} [MeasurableSpace P] [MeasurableSpace ι] [Fintype ι] [Nonempty ι] {K : Set (E d)} (hK : MeasurableSet K) (y : P → ι → E d) (c : P → ι → ℝ) (hy : ∀ i, Measurable (fun p => y p i)) (hc : ∀ i, Measurable (fun p => c p i)) : ∃ k : P × E d → ι, Measurable k ∧ (∀ p x j, ⟪x, y p j⟫ + c p j ≤ ⟪x, y p (k (p, x))⟫ + c p (k (p, x))) ∧ ∀ p, Function.Injective (y p) → ∀ i, (uniformMeasure K).real {x | k (p, x) = i} = … - theorem
OAI.Problem358.CellGeometry.exists_joint_measurable_selector_prescribed_massesline 312Source documentation
Prescribed closed-cell masses on a chosen set of parameters are realized by one jointly measurable selector, even when sites collide elsewhere.
theorem exists_joint_measurable_selector_prescribed_masses {P : Type*} [MeasurableSpace P] [MeasurableSpace ι] [Fintype ι] [Nonempty ι] {K : Set (E d)} (hK : MeasurableSet K) (y : P → ι → E d) (c m : P → ι → ℝ) (s : Set P) (hy : ∀ i, Measurable (fun p => y p i)) (hc : ∀ i, Measurable (fun p => c p i)) (hinj : ∀ p ∈ s, Function.Injective (y p)) (hm : ∀ p ∈ s, ∀ i, (uniformMeasure K).real (closedCell K (y p) (c p) i) = m p i) : ∃ k : P × E d → ι, Measurable k ∧ (∀ p x j, ⟪x, y p j⟫ + c p j ≤ …Mentions: OAI.Problem358.CellGeometry.closedCell