Re: [xsl] help with random numbers

["Michael Kay michaelkay90@xxxxxxxxx" <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx>]

> That's a good question.  The main symptom so far is that the first
> hundred times I ran the simulation, 80 of the runs produced the same
> trivial result: a birth-and-death process in which the initial
> individual died before reproducing - the runs differed on how long that
> individual lived, so they were not completely identical, but the family
> trees they produced were isomorphic: one node with birth and death
> dates.
>

I suspect that result is fairly realistic. Certainly until modern times, most
people didn't live long enough to reproduce, and I've heard the claim that of
all the people who lived in England 1000 years ago, only a tiny handful have
descendants who are living today.

Beware of the natural tendency to distrust your analysis because it doesn't
produce the answers you were expecting. I learned that on my first paid
programming assignment, which was doing some data analysis for an archeologist
-- I eventually had to tell him that if he would only tell me what answers he
wanted, I could easily write a program that delivered them. I also knew a
zoology PhD student whose first assignment was to test the statistical methods
her supervisor was using, by using the algorithms to estimate the size of the
population of London buses - because the answer was known, it could be used to
test the reliability of the method.

Having said that, it does seem that your method of using small integers as
seeds and then only using two or three numbers from the resulting sequence is
flawed.

Michael Kay
Saxonica