Could you define “hanging out” and “prefer”?
I’d guess you’ve (locally) minimized some energy defined by those two phrases, but I can imagine other models, often easier.
Comment by Douglas Knight — February 11, 2007 @ 12:08 pm
The prior probability distribution for the location of each number is a Gaussian. The likelihood of one number being a divisor of another is a (Gaussian) function of their distance (in 2-space). This is a (local) maximum-posterior probability model.