OAI/Geometry/RCD/Harmonic/AffineProjection.lean

GitHub source file

16 declarations, 0 contain sorry, 0 declared as axioms

The actual L² projection onto the affine coordinate span

  1. def OAI.RegularPointCoordinates.restrictedOne line 18
  2. def OAI.RegularPointCoordinates.affineCoordinateMap line 23

    Source documentation

    Constants and linear combinations of the given coordinates, restricted as actual L² functions on the testing set.

    def affineCoordinateMap : (ℝ × Euclidean n) →L[ℝ] Lp ℝ 2 (μ.restrict s) :=
    …

    Mentions: OAI.RegularPointCoordinates.Euclidean

  3. theorem OAI.RegularPointCoordinates.affineCoordinateMap_apply line 28
  4. def OAI.RegularPointCoordinates.affineCoordinateSpace line 33
  5. instance OAI.RegularPointCoordinates.affineCoordinateSpace_finiteDimensional line 36
  6. def OAI.RegularPointCoordinates.affineProjection line 42

    Source documentation

    The least-squares affine fit exists without assuming independence of its generators. Coefficient uniqueness can be added after normalization.

    def affineProjection : Lp ℝ 2 (μ.restrict s) →L[ℝ] Lp ℝ 2 (μ.restrict s) :=
    …
  7. theorem OAI.RegularPointCoordinates.affineProjection_has_coefficients line 45
  8. theorem OAI.RegularPointCoordinates.affineProjection_orthogonal line 53
  9. theorem OAI.RegularPointCoordinates.affineProjection_best line 61

    Source documentation

    The affine fit minimizes the L² error among all coordinate-affine functions, with the unnormalized integral convention.

    theorem affineProjection_best (f : Lp ℝ 2 (μ.restrict s))
        (c : ℝ) (v : Euclidean n) :
        ‖f - affineProjection μ hQ u s hfin f‖ ≤
          ‖f - affineCoordinateMap μ hQ u s hfin (c, v)‖ := by
    …

    Mentions: OAI.RegularPointCoordinates.Euclidean, OAI.RegularPointCoordinates.affineProjection, OAI.RegularPointCoordinates.affineCoordinateMap

  10. def OAI.RegularPointCoordinates.affineResidual line 78

    Source documentation

    The residual is a bounded linear map, so all scalar directional fits can be assembled linearly before taking their matrix coefficients.

    def affineResidual : Lp ℝ 2 (μ.restrict s) →L[ℝ] Lp ℝ 2 (μ.restrict s) :=
    …
  11. theorem OAI.RegularPointCoordinates.affineResidual_norm_le line 81
  12. theorem OAI.RegularPointCoordinates.affineProjection_norm_le line 89
  13. theorem OAI.RegularPointCoordinates.affineResidual_integral line 95

    Source documentation

    Orthogonality to constants is the vanishing of the actual integral, including when the original ambient measure is infinite.

    theorem affineResidual_integral (f : Lp ℝ 2 (μ.restrict s)) :
        ∫ x, affineResidual μ hQ u s hfin f x ∂μ.restrict s = 0 := by
    …

    Mentions: OAI.RegularPointCoordinates.affineResidual

  14. theorem OAI.RegularPointCoordinates.affineResidual_coordinate_inner line 111

    Source documentation

    The residual is orthogonal to each original coordinate in L².

    theorem affineResidual_coordinate_inner (f : Lp ℝ 2 (μ.restrict s)) (i : Fin n) :
        ⟪affineResidual μ hQ u s hfin f,
          SobolevSpace.restrictLp μ hQ s (u i)⟫_ℝ = 0 := by
    …

    Mentions: OAI.RegularPointCoordinates.affineResidual, OAI.RegularPointCoordinates.SobolevSpace.restrictLp

  15. theorem OAI.RegularPointCoordinates.affineProjection_controlled_coefficients line 122

    Source documentation

    Uniform nondegeneracy of the value coordinates gives uniform bounds for the coefficients of their least-squares fit.

    theorem affineProjection_controlled_coefficients (f : Lp ℝ 2 (μ.restrict s))
        {C : ℝ} (hC : 0 ≤ C)
        (hbound : ∀ c v, |c| + ‖v‖ ≤ C * ‖affineCoordinateMap μ hQ u s hfin (c, v)‖) :
        ∃ c : ℝ, ∃ v : Euclidean n,
          affineProjection μ hQ u s hfin f = affineCoordinateMap μ hQ u s hfin (c, v) ∧
          |c| + ‖v‖ ≤ C * ‖f‖ := by
    …

    Mentions: OAI.RegularPointCoordinates.affineCoordinateMap, OAI.RegularPointCoordinates.Euclidean, OAI.RegularPointCoordinates.affineProjection

  16. theorem OAI.RegularPointCoordinates.affineResidual_eq_zero_of_mem line 139

    Source documentation

    Vanishing of an affine residual follows from affinity itself. This is the final contradiction after Euclidean Euler rigidity in sharp projection.

    theorem affineResidual_eq_zero_of_mem (f : Lp ℝ 2 (μ.restrict s))
        (hf : f ∈ affineCoordinateSpace μ hQ u s hfin) :
        affineResidual μ hQ u s hfin f = 0 := by
    …

    Mentions: OAI.RegularPointCoordinates.affineCoordinateSpace, OAI.RegularPointCoordinates.affineResidual