Documentation

LeanPool.Superdiffusion.Algsuperdiff.Section24.Sensitivity.Provider.ResponseUnconditional.CubeSensitivity

CubeSensitivity #

Ported from scottnarmstrong/Superdiffusion at 7f20476385d4af418f110303085ad0740c9eba46. Module imports, support namespace and local option directives adapted for Lean Pool.

The conditional J-sensitivity at a mesoscopic cube, unconditionally #

Source: ABK26.

At the mesoscopic depth H = mesoscaleDepth s (1 + C₀ ‖∇h‖ λ_{s,2}^{-1}(□₀;a)) the dimensionless quantity ‖∇h‖ |R| λ_{3/8,2}^{-1}(R;a) is at most 1/C₀ at every triadic descendant R of the unit cube of depth H, with no hypothesis whatsoever on h. With C₀ chosen large enough this gate

so the affine differential inequality of Provider.BigLambda.PathODE integrates at δ = 1 to the per-cube estimate

J(R, a+h, p, q) ≤ 3 J(R, a, p, q)
                + K₁ |p|² λ_{3/8,2}^{-1}(R) (V(R) + 3^{-2H} ‖∇h‖²)
                + K₁ 3^{-2H} λ_{3/8,2}^{-2}(R) ‖∇h‖² |p·q| ,

with V(R) the cube-local L² value budget of Provider.ResponseUnconditional.ValueLipschitz.

Every declaration in this module is an internal helper for the Section 2.4 sensitivity providers.

Two pieces of abstract path arithmetic #

theorem Algsuperdiff.Section24.Sensitivity.Provider.ResponseUnconditional.dhBound_to_path_step {CA CB P G0 Gt Y T : ℝ} (hCAP : 0 ≤ CA * P) (hCB : 0 ≤ CB) (hGt : 0 ≤ Gt) (hGle : Gt ≤ 2 * G0) (hY : 0 ≤ Y) (_hT : 0 ≤ T) :
1 / 2 * (CA * P * (√Gt * √Y) + CB * (√Gt * √Y * (√Gt * √Y + √Gt * √T))) ≤ 1 / 2 * CA * P * (√2 * √G0) * √Y + 1 / 2 * CB * (2 * G0) * Y + 1 / 2 * CB * (2 * G0) * √T * √Y

The pointwise conversion of the general-loading D_h bound into the affine differential inequality, with the path gauge replaced by twice its base value.

theorem Algsuperdiff.Section24.Sensitivity.Provider.ResponseUnconditional.path_ode_le {y y' : ℝ → ℝ} {A₁ B₁ T : ℝ} (hderiv : ∀ (τ : ℝ), HasDerivAt y (y' τ) τ) (hynn : ∀ (τ : ℝ), 0 ≤ y τ) (_hA₁ : 0 ≤ A₁) (hB₁ : 0 ≤ B₁) (hB₁le : B₁ ≤ 1 / 4) (hT : 0 ≤ T) (hstep : ∀ (τ : ℝ), 0 ≤ τ → τ ≤ 1 → y' τ ≤ A₁ * √(y τ) + B₁ * y τ + B₁ * √T * √(y τ)) :
y 1 ≤ 2 * y 0 + 4 * A₁ ^ 2 + 4 * B₁ ^ 2 * T

Exact integration of the affine differential inequality at δ = 1.

It is large enough both for the mesoscopic pure-flux spine (factor 4 C_flux) and for the Term BC coefficient of the general-loading D_h bound (factor 4 C_BC).

Equations
  • One or more equations did not get rendered due to their size.
Instances For

    The mesoscopic depth attached to the base coefficient and the perturbation.

    Equations
    • One or more equations did not get rendered due to their size.
    Instances For

      The unconditional deep-cube gate #

      The unconditional deep-cube gate. At the mesoscopic depth, the dimensionless quantity governing both the mesoscopic gauge bootstrap and the Term BC coefficient is at most 1/C₀, with no hypothesis on h.

      The mesoscopic gauge along the path #

      The two ODE coefficients #

      The Term A coefficient of the path ODE at a mesoscopic cube.

      Equations
      • One or more equations did not get rendered due to their size.
      Instances For

        The Term BC coefficient of the path ODE at a mesoscopic cube.

        Equations
        • One or more equations did not get rendered due to their size.
        Instances For

          The error constant #

          The error constant of the per-mesoscopic-cube J-sensitivity.

          Equations
          • One or more equations did not get rendered due to their size.
          Instances For

            The per-mesoscopic-cube J-sensitivity #

            The per-mesoscopic-cube J-sensitivity at a general loading.