OAI/Analysis/Brenier/FixedMass.lean

GitHub source file

24 declarations, 0 contain sorry, 0 declared as axioms

  1. def OAI.Problem358.FixedMassIntercepts.masses line 15

    Source documentation

    The actual closed-cell mass vector, without a choice of maximizing selector.

    def masses (K : Set (E d)) (y : ι → E d) (c : ι → ℝ) : ι → ℝ :=
    …
  2. theorem OAI.Problem358.FixedMassIntercepts.closedCell_ae_eq_selector line 19

    Source documentation

    Null ties identify every maximizing selector's fibers with the closed cells.

    theorem closedCell_ae_eq_selector {K : Set (E d)} (hK : MeasurableSet K)
        (y : ι → E d) (hy : Function.Injective y) (c : ι → ℝ)
        (k : E d → ι)
        (hk : ∀ x j, ⟪x, y j⟫ + c j ≤ ⟪x, y (k x)⟫ + c (k x)) (i : ι) :
        CellGeometry.closedCell K y c i =ᵐ[uniformMeasure K] {x | k x = i} := by
    …

    Mentions: OAI.Problem358.CellGeometry.closedCell

  3. theorem OAI.Problem358.FixedMassIntercepts.masses_eq_selector line 36

    Source documentation

    The selector law and geometric closed-cell masses have exactly the same real masses.

    theorem masses_eq_selector {K : Set (E d)} (hK : MeasurableSet K)
        (y : ι → E d) (hy : Function.Injective y) (c : ι → ℝ)
        (k : E d → ι)
        (hk : ∀ x j, ⟪x, y j⟫ + c j ≤ ⟪x, y (k x)⟫ + c (k x)) (i : ι) :
        masses K y c i = (uniformMeasure K).real {x | k x = i} := by
    …

    Mentions: OAI.Problem358.FixedMassIntercepts.masses

  4. theorem OAI.Problem358.FixedMassIntercepts.existsUnique_normalized line 49

    Source documentation

    Prescribed positive masses uniquely determine a normalized intercept vector. This is an actual existence and uniqueness result for the geometric cells; it has no cell-differentiability hypothesis.

    theorem existsUnique_normalized {K : Set (E d)}
        (hK : IsCompact K) (hconv : Convex ℝ K) (hne : (interior K).Nonempty)
        (m : ι → ℝ) (hm : ∀ i, 0 < m i) (hsum : ∑ i, m i = 1)
        (y : ι → E d) (hy : Function.Injective y) :
        ∃! c : CellLaplacian.normalizedSpace m, masses K y c = m := by
    …

    Mentions: OAI.Problem358.CellLaplacian.normalizedSpace, OAI.Problem358.FixedMassIntercepts.masses

  5. def OAI.Problem358.FixedMassIntercepts.intercepts line 78

    Source documentation

    The canonical normalized intercept vector. It is zero on configurations with repeated sites; all uses of the prescribed-mass specification have distinct sites.

    def intercepts {K : Set (E d)}
        (hK : IsCompact K) (hconv : Convex ℝ K) (hne : (interior K).Nonempty)
        (m : ι → ℝ) (hm : ∀ i, 0 < m i) (hsum : ∑ i, m i = 1)
        (y : ι → E d) : CellLaplacian.normalizedSpace m :=
    …

    Mentions: OAI.Problem358.CellLaplacian.normalizedSpace

  6. theorem OAI.Problem358.FixedMassIntercepts.intercepts_spec line 86
  7. theorem OAI.Problem358.FixedMassIntercepts.eq_intercepts line 94
  8. def OAI.Problem358.CellGeometry.normalizedMassMap line 116

    Source documentation

    The actual closed-cell masses, restricted to normalized intercepts.

    def normalizedMassMap (K : Set (E d)) (m : ι → ℝ) :
        ((ι → E d) × CellLaplacian.normalizedSpace m) → (ι → ℝ) :=
    …

    Mentions: OAI.Problem358.CellLaplacian.normalizedSpace

  9. theorem OAI.Problem358.CellGeometry.eventually_injective_sites line 121

    Source documentation

    Distinctness of finitely many sites persists on a neighborhood.

    theorem eventually_injective_sites {y : ι → E d} (hy : Function.Injective y) :
        ∀ᶠ z in 𝓝 y, Function.Injective z := by
    …
  10. theorem OAI.Problem358.CellGeometry.sum_normalizedMassMap line 134

    Source documentation

    The coordinate mass map takes values in probability vectors whenever the sites are distinct. This is derived from the actual closed-cell partition.

    theorem sum_normalizedMassMap [Nonempty ι] {K : Set (E d)} (hK : IsCompact K)
        (hne : (interior K).Nonempty) (m : ι → ℝ)
        (p : (ι → E d) × CellLaplacian.normalizedSpace m)
        (hy : Function.Injective p.1) :
        ∑ i, normalizedMassMap K m p i = 1 := by
    …

    Mentions: OAI.Problem358.CellLaplacian.normalizedSpace, OAI.Problem358.CellGeometry.normalizedMassMap

  11. theorem OAI.Problem358.CellGeometry.normalizedMassResidual_eq_zero_iff line 154

    Source documentation

    The normalized mass residual vanishes exactly at the prescribed physical closed-cell masses, without a selector or residual-range assumption.

    theorem normalizedMassResidual_eq_zero_iff [Nonempty ι] {K : Set (E d)} (hK : IsCompact K)
        (hne : (interior K).Nonempty) (m : ι → ℝ) (hm : ∑ i, m i = 1)
        (p : (ι → E d) × CellLaplacian.normalizedSpace m)
        (hy : Function.Injective p.1) :
        CellLaplacian.massResidual m hm (normalizedMassMap K m) p = 0 ↔
          ∀ i, (uniformMeasure K).real (closedCell K p.1 p.2 i) = m i := by
    …

    Mentions: OAI.Problem358.CellLaplacian.normalizedSpace, OAI.Problem358.CellLaplacian.massResidual, OAI.Problem358.CellGeometry.normalizedMassMap, OAI.Problem358.CellGeometry.closedCell

  12. def OAI.Problem358.CellGeometry.normalizedConfigInclusion line 165

    Source documentation

    Include the normalization hyperplane into the unrestricted configuration space.

    def normalizedConfigInclusion (m : ι → ℝ) :
        ((ι → E d) × CellLaplacian.normalizedSpace m) →L[ℝ] ((ι → E d) × (ι → ℝ)) :=
    …

    Mentions: OAI.Problem358.CellLaplacian.normalizedSpace

  13. theorem OAI.Problem358.CellGeometry.normalizedConfigInclusion_apply line 169
  14. theorem OAI.Problem358.CellGeometry.contDiffAt_normalizedMassMap_of_raw line 175

    Source documentation

    Restrict geometric C¹ regularity on ordinary site/intercept coordinates to the weighted normalization hyperplane used by the implicit function theorem.

    theorem contDiffAt_normalizedMassMap_of_raw {K : Set (E d)} (m : ι → ℝ)
        (p : (ι → E d) × CellLaplacian.normalizedSpace m)
        (hM : ContDiffAt ℝ 1
          (fun z : (ι → E d) × (ι → ℝ) => fun i =>
            (uniformMeasure K).real (closedCell K z.1 z.2 i))
          (p.1, (p.2 : ι → ℝ))) :
        ContDiffAt ℝ 1 (normalizedMassMap K m) p :=
    …

    Mentions: OAI.Problem358.CellLaplacian.normalizedSpace, OAI.Problem358.CellGeometry.closedCell, OAI.Problem358.CellGeometry.normalizedMassMap

  15. theorem OAI.Problem358.CellGeometry.hasFDerivAt_normalizedMassMap_of_raw line 186

    Source documentation

    The normalized mass derivative is the unrestricted geometric derivative composed with the actual hyperplane inclusion.

    theorem hasFDerivAt_normalizedMassMap_of_raw {K : Set (E d)} (m : ι → ℝ)
        (p : (ι → E d) × CellLaplacian.normalizedSpace m)
        (A : ((ι → E d) × (ι → ℝ)) →L[ℝ] (ι → ℝ))
        (hD : HasFDerivAt
          (fun z : (ι → E d) × (ι → ℝ) => fun i =>
            (uniformMeasure K).real (closedCell K z.1 z.2 i)) A
          (p.1, (p.2 : ι → ℝ))) :
        HasFDerivAt (normalizedMassMap K m) (A ∘L normalizedConfigInclusion m) p := by
    …

    Mentions: OAI.Problem358.CellLaplacian.normalizedSpace, OAI.Problem358.CellGeometry.closedCell, OAI.Problem358.CellGeometry.normalizedMassMap, OAI.Problem358.CellGeometry.normalizedConfigInclusion

  16. theorem OAI.Problem358.CellLaplacian.normalizedSpace_eq_zero_of_subsingleton line 214

    Source documentation

    For one site, the weighted normalization hyperplane is the zero space. This uses only the total mass, and does not need positivity or a graph lemma.

    theorem normalizedSpace_eq_zero_of_subsingleton (m : ι → ℝ)
        (hsum : ∑ i, m i = 1) (q : normalizedSpace m) : q = 0 := by
    …

    Mentions: OAI.Problem358.CellLaplacian.normalizedSpace

  17. theorem OAI.Problem358.CellLaplacian.subsingleton_normalizedSpace line 225
  18. theorem OAI.Problem358.CellGeometry.closedCell_eq_of_subsingleton line 236

    Source documentation

    A one-site closed affine cell is the entire source set.

    theorem closedCell_eq_of_subsingleton {d : ℕ} {ι : Type*} [Subsingleton ι]
        (K : Set (E d)) (y : ι → E d) (c : ι → ℝ) (i : ι) :
        closedCell K y c i = K := by
    …

    Mentions: OAI.Problem358.CellGeometry.closedCell

  19. theorem OAI.Problem358.FixedMassIntercepts.intercepts_eq_zero_of_subsingleton line 254

    Source documentation

    The actual globally selected normalized intercept is zero for a one-site target.

    @[simp] theorem intercepts_eq_zero_of_subsingleton {K : Set (E d)}
        (hK : IsCompact K) (hconv : Convex ℝ K) (hne : (interior K).Nonempty)
        (m : ι → ℝ) (hm : ∀ i, 0 < m i) (hsum : ∑ i, m i = 1)
        (y : ι → E d) : intercepts hK hconv hne m hm hsum y = 0 :=
    …

    Mentions: OAI.Problem358.FixedMassIntercepts.intercepts

  20. theorem OAI.Problem358.FixedMassIntercepts.contDiff_intercepts_of_subsingleton line 261

    Source documentation

    The one-site intercept selection is smooth independently of any cell calculus.

    theorem contDiff_intercepts_of_subsingleton {K : Set (E d)}
        (hK : IsCompact K) (hconv : Convex ℝ K) (hne : (interior K).Nonempty)
        (m : ι → ℝ) (hm : ∀ i, 0 < m i) (hsum : ∑ i, m i = 1)
        (n : ℕ∞ω) : ContDiff ℝ n (intercepts hK hconv hne m hm hsum) := by
    …

    Mentions: OAI.Problem358.FixedMassIntercepts.intercepts

  21. theorem OAI.Problem358.FixedMassIntercepts.hasDerivAt_intercepts_path_of_subsingleton line 274

    Source documentation

    Even an arbitrary site path gives a constant normalized intercept path if there is only one site. In particular its intercept energy is exactly zero.

    theorem hasDerivAt_intercepts_path_of_subsingleton {K : Set (E d)}
        (hK : IsCompact K) (hconv : Convex ℝ K) (hne : (interior K).Nonempty)
        (m : ι → ℝ) (hm : ∀ i, 0 < m i) (hsum : ∑ i, m i = 1)
        (y : ℝ → ι → E d) (t : ℝ) :
        HasDerivAt (fun s => intercepts hK hconv hne m hm hsum (y s)) 0 t := by
    …

    Mentions: OAI.Problem358.FixedMassIntercepts.intercepts

  22. theorem OAI.Problem358.FixedMassIntercepts.contDiffAt_intercepts_of_massDifferential line 301

    Source documentation

    The actual canonical prescribed-mass intercepts are C¹ at each distinct-site configuration where the geometric mass map is C¹ with its connected cell Laplacian as intercept derivative. Neither continuity nor uniqueness of a chosen branch is assumed: both are supplied by the geometric closed-cell construction.

    theorem contDiffAt_intercepts_of_massDifferential {K : Set (E d)}
        (hK : IsCompact K) (hconv : Convex ℝ K) (hne : (interior K).Nonempty)
        (m : ι → ℝ) (hm : ∀ i, 0 < m i) (hsum : ∑ i, m i = 1)
        (y : ι → E d) (hy : Function.Injective y)
        (g : ι → ι → ℝ) (hsym : ∀ i j, g i j = g j i)
        (hnonneg : ∀ i j, 0 ≤ g i j) (G : SimpleGraph ι)
        (hG : G.Connected) (hpos : ∀ i j, G.Adj i j → 0 < g i j)
        (A : ((ι → E d) × CellLaplacian.normalizedSpace m) →L[ℝ] (ι → ℝ))
        (hM : ContDiffAt ℝ 1 (CellGeometry.normalizedMassMap K m)
          (y, intercepts hK hconv hne m hm hsum y))
        (hD : HasFDerivAt (CellGeometry.normalizedMassMap K m) A
          (y, intercepts hK hconv hne m hm hsum y))
    …

    Mentions: OAI.Problem358.CellLaplacian.normalizedSpace, OAI.Problem358.CellGeometry.normalizedMassMap, OAI.Problem358.FixedMassIntercepts.intercepts

  23. theorem OAI.Problem358.FixedMassIntercepts.hasDerivAt_intercepts_comp_and_mass_constraint line 339

    Source documentation

    Along a differentiable site path, the actual canonical intercept derivative satisfies the differentiated fixed-mass constraint. Its weighted mean vanishes because it belongs to the normalization subspace.

    theorem hasDerivAt_intercepts_comp_and_mass_constraint {K : Set (E d)}
        (hK : IsCompact K) (hconv : Convex ℝ K) (hne : (interior K).Nonempty)
        (m : ι → ℝ) (hm : ∀ i, 0 < m i) (hsum : ∑ i, m i = 1)
        (Y : ℝ → (ι → E d)) (v : ι → E d) (t : ℝ)
        (hY : HasDerivAt Y v t) (hy : Function.Injective (Y t))
        (hs : DifferentiableAt ℝ (intercepts hK hconv hne m hm hsum) (Y t))
        (A : ((ι → E d) × CellLaplacian.normalizedSpace m) →L[ℝ] (ι → ℝ))
        (hD : HasFDerivAt (CellGeometry.normalizedMassMap K m) A
          (Y t, intercepts hK hconv hne m hm hsum (Y t))) :
        let c := fun q => intercepts hK hconv hne m hm hsum (Y q)
    …

    Mentions: OAI.Problem358.FixedMassIntercepts.intercepts, OAI.Problem358.CellLaplacian.normalizedSpace, OAI.Problem358.CellGeometry.normalizedMassMap

  24. theorem OAI.Problem358.FixedMassIntercepts.laplacian_intercepts_comp_deriv line 364

    Source documentation

    Coordinate form of the fixed-mass derivative equation. This is the actual constraint consumed by the symmetric flux / intercept energy calculation.

    theorem laplacian_intercepts_comp_deriv {K : Set (E d)}
        (hK : IsCompact K) (hconv : Convex ℝ K) (hne : (interior K).Nonempty)
        (m : ι → ℝ) (hm : ∀ i, 0 < m i) (hsum : ∑ i, m i = 1)
        (Y : ℝ → (ι → E d)) (v : ι → E d) (t : ℝ)
        (hY : HasDerivAt Y v t) (hy : Function.Injective (Y t))
        (hs : DifferentiableAt ℝ (intercepts hK hconv hne m hm hsum) (Y t))
        (A : ((ι → E d) × CellLaplacian.normalizedSpace m) →L[ℝ] (ι → ℝ))
        (hD : HasFDerivAt (CellGeometry.normalizedMassMap K m) A
          (Y t, intercepts hK hconv hne m hm hsum (Y t)))
        (g : ι → ι → ℝ)
        (hA : ∀ q : CellLaplacian.normalizedSpace m, A (0, q) = CellLaplacian.apply g q) :
        CellLaplacian.apply g
    …

    Mentions: OAI.Problem358.FixedMassIntercepts.intercepts, OAI.Problem358.CellLaplacian.normalizedSpace, OAI.Problem358.CellGeometry.normalizedMassMap, OAI.Problem358.CellLaplacian.apply