Documentation

LeanPool.HighContrastHomogenization.Support.Probability.RescaledLaw

Coarse-graining support: Support.Probability.RescaledLaw #

Imported from the Apache-2.0 CoarseGraining development at commit c7ddd76c08ade64fed1b8d2ca51be14dfee8deb4.

Triadically rescaled coefficient laws and random scale parameters.

The Chapter 4 probability layer is mostly law-centric: coefficient fields are sampled from a measure on CoeffField d. This module adds the basic transport API for passing from a law to its triadically rescaled law and records the lightweight random-scale package used by later quenched-scale arguments.

noncomputable def HCPolySupport.triadicDilateVec {d : ℕ} (n : ℕ) (x : Vec d) :
Vec d

Dilate a vector by the triadic factor 3^n.

Equations
Instances For
    noncomputable def HCPolySupport.triadicDilateSet {d : ℕ} (n : ℕ) (U : Set (Vec d)) :
    Set (Vec d)

    The image of a set under triadic dilation by 3^n.

    Equations
    Instances For
      def HCPolySupport.triadicScaleIntShift {d : ℕ} (n : ℕ) (z : Fin d → ℤ) :
      Fin d → ℤ

      The integer shift in the original variables corresponding to an integer shift after triadic rescaling.

      Equations
      Instances For
        noncomputable def HCPolySupport.rescaleCoeffField {d : ℕ} (n : ℕ) (a : CoeffField d) :

        Rescale a coefficient field by the triadic factor 3^n.

        The rescaled field is x ↦ a(3^n x). Thus integer translations of the rescaled field correspond to translations of the original field by integer vectors multiplied by 3^n.

        Equations
        Instances For
          @[simp]

          Triadic dilation sends bounded subsets of the ambient space to bounded subsets.

          theorem HCPolySupport.localTestObservable_rescaleCoeffField_eq_const_mul {d : ℕ} (k : ℕ) (e e' : Vec d) (φ : Vec d → ℝ) (a : CoeffField d) :
          localTestObservable e e' φ (rescaleCoeffField k a) = ((3 ^ k) ^ d)⁻¹ * localTestObservable e e' (fun (y : Vec d) => φ ((3 ^ k)⁻¹ • y)) a
          theorem HCPolySupport.localFiniteTestObservable_rescaleCoeffField_eq {d : ℕ} {ι : Type} (n : ℕ) (I : Finset ι) (e e' : ι → Vec d) (φ : ι → Vec d → ℝ) (a : CoeffField d) :
          localFiniteTestObservable I e e' φ (rescaleCoeffField n a) = localFiniteTestObservable I e e' (fun (k : ι) (y : Vec d) => ((3 ^ n) ^ d)⁻¹ * φ k ((3 ^ n)⁻¹ • y)) a