NRR.PowerDiagram.BodyCellPartition — partition properties of restricted cells #
The restricted power (Laguerre) cells bodyCellSet K s w i = (K : Set Plane) ∩ cell s w i
form a finite almost‑disjoint cover of a convex body K:
bodyCellSet_subset— each restricted cell is contained inK.iUnion_bodyCellSet— the restricted cells coverK(needs[NeZero n]: with no sites the union is empty while a convex body is nonempty, so the covering claim is false forn = 0).bodyCellSet_inter_null— for an injective site family, two distinct restricted cells overlap on a Lebesgue‑null set.
The nondegeneracy needed for the null‑overlap statement is sepNormal s i j ≠ 0. This is
not implied by i ≠ j alone, but it is implied by Function.Injective s together with
i ≠ j; the bridging lemma sepNormal_ne_zero_of_injective records that implication.
No nonemptiness of restricted cells is assumed, no ConvexPartition is bundled, and no
equal‑area properties are proved here.
A restricted power cell is contained in the body K.
The restricted power cells cover the body K. Requires at least one site ([NeZero n]):
the full cells cover the plane by iUnion_cell, so intersecting with K recovers K.
For an injective family of sites, distinct indices give a nonzero separating normal:
sepNormal s i j = 2 • (sⱼ - sᵢ) is nonzero because sᵢ ≠ sⱼ.
Null overlap. For an injective site family, the overlap of two distinct restricted power
cells is Lebesgue‑null: it embeds into the overlap of the full (nondegenerate) cells, which is
null by cell_inter_null.