full period

51 views
Skip to first unread message

Euriece Jan

unread,
Dec 12, 2021, 4:49:23 AM12/12/21
to prng
I have read "Scrambled Linear Pseudorandom Number Generators", but I could not understand logically how to find the parameter of the full period. 

Can anyone help me understand it?

Pierre L'Ecuyer

unread,
Dec 12, 2021, 9:11:24 AM12/12/21
to pr...@googlegroups.com, euri...@gmail.com

When a generator is scrambled at the output only, its period is normally the same as the original linear recurrence (not scrambled).  The following papers (and the references given there) discuss the period length of these linear generators, modulo 2 and modulo m > 2:

  1. f2lin.pdf P. L'Ecuyer and F. Panneton, ``F_2-Linear Random Number Generators'', published in "Advancing the Frontiers of Simulation: A Festschrift in Honor of George Samuel Fishman," C. Alexopoulos, D. Goldsman, and J. R. Wilson Eds., Springer-Verlag, 2009, 169-193.
  2. combmrg2.pdf P. L'Ecuyer, ``Good Parameter Sets for Combined Multiple Recursive Random Number Generators'', Shorter version in Operations Research, 47, 1 (1999), 159--164.

On 12/12/2021 4:49 AM, Euriece Jan wrote:
I have read "Scrambled Linear Pseudorandom Number Generators", but I could not understand logically how to find the parameter of the full period. 

Can anyone help me understand it?
--
You received this message because you are subscribed to the Google Groups "prng" group.
To unsubscribe from this group and stop receiving emails from it, send an email to prng+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/prng/ff93dee9-fc32-4fed-94de-2bdff2478770n%40googlegroups.com.
-- 
Pierre L'Ecuyer, Professeur Titulaire
CIRRELT, GERAD, and DIRO, Université de Montréal, Canada
http://www.iro.umontreal.ca/~lecuyer
Message has been deleted

Euriece Jan

unread,
Dec 12, 2021, 12:32:53 PM12/12/21
to prng
Thank you very much. Do these two papers show how to find the period of xoroshiro?

2021年12月12日日曜日 23:11:24 UTC+9 lec...@iro.umontreal.ca:

Pierre L'Ecuyer

unread,
Dec 12, 2021, 1:26:58 PM12/12/21
to pr...@googlegroups.com

Xoroshiro uses an F_2-linear recurrence, so the theory and methods in the first paper apply.  For more specific discussions, see the xoroshiro paper by Blackman and Vigna, in ACM Trans. on Math Software,  http://arxiv.org/abs/1805.01407

Reply all
Reply to author
Forward
0 new messages