Re: Testing for product of Gaussians

Herman - thanks for the idea on the stronger test, I'll definitely use
that in the K-S framework. As to what will be actually be tested by my
stats - I think I see what you are getting at, but I have no reason to
suspect my random number generator and can make my bins as large or
small as I please, although do realise that both of these factors could
skew my results. Perhaps I could ask a related question?

Using the computer to repeatedly generate random numbers and then jump
around based upon them to simulate diffusion is very inefficient but
for one reason or another is what I am doing. As well having the
desirable side effect of testing that something hasn't gone grossly
wrong with my computer program/mapping between diffusion coefficients
and random walk step lengths, what I really want to do is be able to
use a statistical test to give a lower bound on the minimum number
particles moving around I can use with in my final program for the
complicated problem I am actually going to tackle, while still being to
claim that some number of particles somewhat smaller "gives
statistically acceptible results for the simpler pure diffusion
problem". Am I approaching this correctly do you think, or is there
something else I should be doing (possibly asking in
sci.math.num-analysis?)

Thanks to all again,

.