vibespatial.constructive.tiled_union

Exact device collective union with bounded topology tiles.

The logical operation is one polygonal unary union. Its physical work is segment, tile, face, and boundary-atom shaped: small arrangements use one collective topology plan, while large arrangements are clipped into disjoint tiles, solved locally, and stitched by noded coverage assembly.

ADR-0002: constructive coordinates and topology remain fp64. ADR-0044: inputs and outputs remain device-resident OwnedGeometryArray. ADR-0046: dispatch uses segment-peer pressure rather than public row count.

Attributes

cp

Functions

single_group_polygon_collective_coverage_gpu(...)	Return exact interior-disjoint tile coverage for a large polygon union.
single_group_polygon_collective_union_gpu(...)	Union one all-valid polygon group through bounded collective topology.

Module Contents

vibespatial.constructive.tiled_union.cp = None

vibespatial.constructive.tiled_union.single_group_polygon_collective_coverage_gpu(owned: vibespatial.geometry.owned.OwnedGeometryArray, *, force_tile_count: int | None = None) → vibespatial.geometry.owned.OwnedGeometryArray | None

Return exact interior-disjoint tile coverage for a large polygon union.

The carrier is intended for downstream relation/grouped consumers. Its row union is the exact logical union, but shared tile seams remain until the consumer’s grouped constructive reduction or final seam stitch.

vibespatial.constructive.tiled_union.single_group_polygon_collective_union_gpu(owned: vibespatial.geometry.owned.OwnedGeometryArray, *, force_tile_count: int | None = None) → vibespatial.geometry.owned.OwnedGeometryArray

Union one all-valid polygon group through bounded collective topology.

The tiled variant clips source geometry into mutually interior-disjoint cells. Local unions therefore never feed overlapping complex aggregate polygons into another binary overlay. The final operation only nodes and removes shared tile seams from a proven coverage.