Author: Sune Fischer
Date: 02:12:38 12/31/01
Go up one level in this thread
On December 30, 2001 at 21:42:21, Eelco de Groot wrote: >A very simple test doing without any statistics you often read about would be to >use two consecutive generated numbers as coordinates of a point. Plot the points >on the screen and the screen should get filled up at random with many points. If >any two numbers used as a coordinate pair are not independant this becomes >evident as patterns in the output. The points may cluster along a line for >instance. It does not say much about the degree of "unrandomness" and the human >eye is quick to see patterns in such output when there really isn't any (just >think about all the star constellations that we imagined to be all sorts of >things, mythological heroes, animals, cooking utensils etc., from prehistory >onwards). I believe the minimum distance test above does something like this and >then uses statistics to get a measure of the degree of interdependancy of >points. Yes, that is a good test also. BTW. Numerical Recipes has a chapter 7 about PRNGs: http://www.ulib.org/webRoot/Books/Numerical_Recipes/bookc.html >As we are being off topic already I hope I can take advantage of the occasion to >say: "Happy New Year Everyone!" and that goes for me 2. > Eelco Sune
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.