Documentation

LeanPool.Superdiffusion.Algsuperdiff.Section3.Provider.Whitney.BoundaryDistance

BoundaryDistance #

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

The two-sided boundary distance of a Whitney layer #

This module proves the assertion ABK26 records without proof β€” "the cubes in 𝒲(β–‘_m, n) lie at distance roughly 3^{m-n} from the boundary βˆ‚β–‘_m" β€” and the two consequences its proof of l.bad.clusters.geometry draws from it:

Why the lower half is stated at the centre only #

The consumer two_thirds_zpow_lt_norm_cubeCenter_sub_of_layers pairs the lower half at the centre of the shallower cube against the upper half at the centre of the deeper one, so the centre-only lower bound suffices; it is also the form for which the index arithmetic is exact.

Provider hypotheses #

Both halves are proved for an abstract depth profile hn : β„• β†’ β„•; no monotonicity, no gap bound and no dimension hypothesis is used. The adjacency clause adds only 1 ≀ hn n for the two layers involved (consequence 3: kβ‚€ β‰₯ 1). Layer indices are automatically β‰₯ 1 because whitneyLayer m hn 0 = βˆ….

References #

Elementary cube arithmetic #

Structure of a layer cube #

The interior cube of a layer-n Whitney cube: the depth-n non-boundary-touching descendant of β–‘_m produced by the recursion.

The boundary ancestor of a layer-n Whitney cube: the depth-(n-1) descendant of β–‘_m that still touches βˆ‚β–‘_m.

theorem Algsuperdiff.Section3.Provider.Whitney.whitneyLayer_index_unique {d : β„•} {m : β„€} {hn : β„• β†’ β„•} (hmono : Monotone hn) {n n' : β„•} {Q : HCPolySupport.TriadicCube d} (hQ : Q ∈ whitneyLayer m hn n) (hQ' : Q ∈ whitneyLayer m hn n') :
n = n'

Distinct layers of a nondecreasing profile sit at distinct scales, so a Whitney cube determines its layer index.

Additivity of the extreme index #

E_{t+s} = 3^s E_t + E_s: subdividing s further generations rescales the extreme index and adds the extreme index of the new depth.

The two-sided boundary distance #

, the index bound. A layer-n Whitney cube has every index coordinate at most E_{n+h_n} - 3^{h_n}: it is a depth-h_n descendant of a non-touching depth-n cube, whose index is at most E_n - 1.

, the lower half, at the centre: a layer-n Whitney cube has its centre at β„“^∞ distance strictly more than 3^{m-n} from βˆ‚β–‘_m.

, the upper half: every point of a layer-n Whitney cube is within 3Β·3^{m-n} of βˆ‚β–‘_m in the β„“^∞ sense, because the cube sits inside a boundary-touching cube of size 3^{m-n+1}.

Layer separation and adjacency #

The layer-separation constant c = 2/3: Whitney cubes two or more layers apart have centres more than (2/3)Β·3^{m-n} apart in the β„“^∞ sense. The constant is dimension-free.

theorem Algsuperdiff.Section3.Provider.Whitney.whitneyLayer_index_close_of_cubeTouch {d : β„•} {m : β„€} {hn : β„• β†’ β„•} (hn1 : βˆ€ (j : β„•), 1 ≀ hn j) {n n' : β„•} {Q R : HCPolySupport.TriadicCube d} (hQ : Q ∈ whitneyLayer m hn n) (hR : R ∈ whitneyLayer m hn n') (htouch : CubeTouch Q R) :
n ≀ n' + 1 ∧ n' ≀ n + 1

The adjacency clause: two touching Whitney cubes lie in equal or neighbouring layers. Per consequence 2 this uses no percolation input and no largeness of kβ‚€; only h_n β‰₯ 1 enters.