Minkowski difference
The set of all vector differences between two shapes; two convex shapes overlap exactly when their Minkowski difference contains the origin.
For shapes A and B the Minkowski difference is A ⊖ B = { a − b : a ∈ A, b ∈ B }. Its key property drives collision detection: A and B intersect if and only if the origin lies inside A ⊖ B, and the distance from the origin to the difference's surface is the separation (or penetration) distance.
The GJK algorithm never builds the whole difference — it samples it lazily with support functions, walking a simplex toward the origin, which is why GJK handles arbitrary convex shapes with no per-pair special-casing. Contrast the Minkowski sum (A ⊕ B), used for configuration-space planning and for inflating obstacles.