Spatial Hash Broadphase

Naive collision is quadratic and dies at a few hundred entities. The usual fix is a uniform grid, and the usual grid is subtly wrong: entities spanning a cell boundary get inserted into ONE cell, so two objects touching across the seam never test against each other. Collisions are missed only near boundaries, which makes it look like a rare physics glitch rather than a systematic bug.

The second classic error is returning both (A,B) and (B,A), so collision responses apply twice and objects jitter.

VERIFIED: 36 tests, 6/6 mutations caught. Tested differentially against brute force over randomised scenes -- the pair set must be identical, which is the only test that actually proves a broadphase. It caught a bug class beyond the two documented: trusting bucket co-membership without the exact overlap filter produces false-positive pairs.

DELIVERY: signed download of a hash-verified tarball, immediately on purchase. Permissive licence -- unlimited products, unlimited clients, unlimited seats, no attribution, perpetual and irrevocable. One restriction: do not republish the source as source.

Interface

What you call, and what comes back. Types and signatures only — the implementation ships with the source.

  export function aabbOverlaps(a: AABB, b: AABB): boolean;