I have a 3d data cube. For every point I measure the property A which is a gaussian variable of mean m and variance s and it's also a function of the density d at every point. [itex] A(x,y,z)=f(d(x,y,z))e^(-(X-m)^2/(2\sigma^2)) [/itex] X is a random number. Now let's say I want to sample the distribution. I mean, I divide my 3d cube in boxes of size delta X. What happen to the function f?