Uniform grid
Spatial partitioning that divides space into equal cells and bins objects into them, so collision tests only run against neighbours in the same cell.
Each object is placed in the cell or cells it overlaps, and a collision or neighbour query only needs to look in its own cell and the adjacent ones rather than across the whole scene. When objects are similar in size and evenly spread, both insertion and lookup are O(1) on average, the structure is cache-friendly and easily parallelised — which is why it is a common broad phase in 2D physics and in boids.
Its weakness is the "teapot in a stadium": few objects or wildly varying sizes either waste cells or overflow them. Then spatial hashing, hierarchical grids or adaptive trees like quadtrees and octrees do better.