I made the following experiment:

http://lzkiss.netfirms.com/cgi-bin/igperl/igp.pl?dir=random&name=funcrand

1./ Get an arbitrary function f(x)  which is continuous in the [$from, 
$to] interval.

2./ Approximate the integral of f(x) + min(f(x)). This is necessary to 
get non decreasing integral function. Also normalize the integral 
function, for easier calculations and graphics, so the last value of 
that will equal 1.

3./ Get the values of the inverse of this integral for given number of 
computer generated uniform  random numbers, and save it as the 
transformed random set.

4. Group the set and compare visually and using the good-fit chi square 
test with the original function. The experiment says that the density of 
the new set follows the original function.

Question: could be this proven or rejected mathematically?

laszlo

.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to