[ what's new? ] -High efficiency improvement making the algorithm O(n) instead of O(n^2) using a space partitioning hash table -Zoom correction (the higher the zoom the bigger things appear on the screen) -Replaced "ture" by "true" in "resolve collision" ; ) This optimization adds a parameter to the simulation called "hashing factor" which should be around the inverse of the average object radius To get a significant improvement go to TurboWarp because the higher is n the higher is the difference between O(n^2) and O(n) /!\ doesn't handle object deletion You can use it without credit