What distribution must I sample from so that the distribution of x*y is uniform between [a,b], where x and y are independently sampled from the distribution?
Update: I've been informed (from -c help -i probability) that the answer is 1/sqrt(-pi log x) 0<x<1.
Triggers a Euler-identity-like surprise: where did "pi" and "e" (in "log") come from?!
No comments :
Post a Comment