Documentation

LeanPool.Superdiffusion.Algsuperdiff.Section4.Provider.ExcessDecay.BoundaryCoveringGeometry

BoundaryCoveringGeometry #

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

The well-placed cube of the boundary covering step #

ABK26's Step 2 applies the boundary coarse-grained Caccioppoli inequality on a truncated window (x+□_{n+1}) ∩ □_m and concludes an energy bound on (x+□_n) ∩ □_m, "after a simple covering argument". CoarseGraining's theorem, however, lives on a cube: an outer triadic cube Q, a Dirichlet patch openCubeAtScale x (Q.scale-1), and the core caccioppoliCoreSet Q x = □_Q ∩ openCubeAtScale x (Q.scale-2).

This module constructs that cube and proves the three inclusions the application needs.

The construction #

With hm := 3^m/2, hk := 3^k/2 and the gap A := hm - hk ≥ 0 (this is where k ≤ m enters), the centre is the coordinatewise clamp

  c_i := max (-A) (min A (x_i)) ,        c := wellPlacedCentre x m k .

Clamping is exactly "anchor on the met faces": in a coordinate where the window stays well inside □_m the cube is centred at x_i, and in a coordinate where the window reaches a face the cube slides until its own face coincides with that face of □_m. Consequently

What is not done here #

No measure theory (that is BoundaryCoveringVolume) and no analysis: this module contains only inclusions of sets.

References #

1. Membership in a translated set #

A point lies in the translate c + S exactly when its back-translate lies in S.

2. The well-placed centre #

The clamping gap 3^m/2 - 3^k/2 of the well-placed cube.

Equations
Instances For

    The centre of the well-placed cube: the coordinatewise clamp of x into the set of centres whose cube of side 3^k still fits inside □_m.

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

      3. The cube stays inside the domain #

      The well-placed cube never leaves □_m. This is what makes the printed "the equation holds on the covering cube" legitimate, and it forces the Dirichlet locus of the Caccioppoli application into ∂□_m.

      4. The cube covers the truncated window #

      The well-placed cube covers the truncated window.

      For every scale j ≤ k the anchor's window (x+□_j) ∩ □_m lies inside the translated cube c + □_k. At j = k-1 this is the covering of the Caccioppoli Dirichlet patch; at j = k-2 it is the covering of the energy window.

      5. The window sits in CoarseGraining's Caccioppoli core #

      The covering inclusion in CoarseGraining's own window shape.

      In the translated frame the anchor's window (x+□_j) ∩ □_m with j ≤ k-2 is contained in caccioppoliCoreSet (□_k) (x-c), the core of CoarseGraining's boundary coarse Caccioppoli inequality at the outer cube □_k and centre x-c. Together with image_add_wellPlacedCentre_subset_openCubeSet this is the geometric content of the printed "simple covering argument".