[BubRoss]

Poisson disc sampling is a brute force technique and isn't an analytic method for packing. Any sample pattern can be relaxed if you are finding neighbor points and using multiple iterations. Poisson sampling is just trying and rejecting many samples based on their neighbors' distances, there aren't any deep principles being used.

It is a lot like a bubble sort in that it is O(n^2) and is useful in practice on tiny sets and as a benchmark.

(Also that link is very cool, but it looks closer to n-rooks / latin hypercube sampling in that it used a grid, which means it won't work in higher dimensions)