Documentation

LeanPool.Superdiffusion.Algsuperdiff.Section3.Provider.Diffusivity.Corrector.OscillationCubeFamily

OscillationCubeFamily #

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

The concentric triadic cube family of e.nablaw.oscillations #

e.nablaw.oscillations is stated for a mesh point z in 3^n Zd ∩ cu_K with `z

  U_j := z + cu_{n+j} = openCubeAtScale z (n + j) ,   0 ≤ j ≤ m - h - n ,

all centred at the same z. Only U_0 is a triadic cube in the lattice sense (z ∈ 3^n Zd); the coarser members U_j, j ≥ 1, are cubes of side 3^{n+j} centred at a point of the finer lattice, so they are not members of Homogenization.TriadicCube at their own scale. This is why the whole argument must be run on Homogenization.openCubeAtScale, and why this module exists.

CoarseGraining supplies openCubeAtScale together with openness, measurability, the product description, the translation identity and the dilation identity, but no nesting, no volume and no convexity. Those three facts are what the nested-recentring route consumes from the cube family itself, and they are proved here.

Contents #

References #

The half-width appearing in openCubeAtScale is the ordinary integer power 3 ^ m; this removes Real.rpow from every downstream computation.

Membership in the concentric cube z + cu_m, with the half-width written as an integer power of three.

The containment criterion.

If every coordinate of the offset z - w is small enough to fit the cube of scale m inside the cube of scale n, then it does. Taking z = w gives the concentric nesting z + cu_m ⊆ z + cu_n for m ≤ n; taking w = 0, n = K gives the interior-mesh hypothesis z + cu_{m-h} ⊆ cu_K of e.nablaw.oscillations.

Concentric nesting: the cube family U_j = z + cu_{n+j} is increasing.

Convexity of the concentric cube, the hypothesis under which the mean-value inequality turns a bound on sup |grad h| into the birth-scale forcing size.