Random number generators discussion

Random number generators | www.agner.org

thread New PRNG - Ribeiro Alvo - 2009-05-18
last reply New PRNG - Agner - 2009-05-18
Author:  Date: 2009-05-18 00:27
Agner, see the prng algorithm on


What do you think about this


Ribeiro Alvo

Author: Agner Date: 2009-05-18 06:58
You are writing:
this algorithm is non periodic. Is only limited by the physical capacity of the computer's memory.
This is not true. There are 253 possible double precision numbers between 0 and 1 so the period cannot possibly be longer than 253. You have not proven that the period is 253 for all values of s. It is possible that the period could be as low as 1 for certain values of s. It is also possible that it can have a rho-shaped trajectory. This means for example a, b, c, d, e, f, g, h, f, g, h, f, g, h, f, g, h, ... where the letters represent different x-values.

The TESTU01 battery of tests is better than DIEHARD. You can find it here

Try to test your generator with TESTU01 with different values of s.