Prolate ellipsoids don't work; Mercury does not detect overlaps
Blockiness of 0.1 causes contact point detection to fail
The paper by Alexander P. on which the Mercury algorithm is based says convergence is guaranteed for blockiness >0.25
Dariel will take a look whether this can be fixed
Dariel corrected the distribution of angles (relative to the velocity vector) by sin(theta)/2, but the data for spheres (AR=1) is not perfectly uniform (it should be); also, the data has a large standard deviation
I told him to use the data files instead of the restart files because they contain more timesteps; and look for errors