Author: Janosch Zwerensky
Date: 07:21:34 10/29/03
Go up one level in this thread
On October 29, 2003 at 06:56:06, Dan Andersson wrote: > It would be OK to concatenate if you used a cryptologically safe PRNG like x^2 >mod M. I doubt strongly that the method you propose is cryptographically safe. Off the top of my head, I'd say that the cycle length will be at most M-1 (which case will not happen for M>2) and equal to one at worst, that most of the time it will be *much* worse than M-1 and that the behaviour of the least significant bit will be quite nonrandom even if one disregards the abovementioned facts. > But it is a tad unsound when using linear, fibonacci and generalized >shift feedback PRNGs. I would assume that getting random numbers of 2n binary digits by concatenating consecutively generated numbers of n digits length is always unsound when the numbers generated by the pseudo random number generator give complete information about the state of the prng (which is the case for example with your method above or with linear congruential generators), because among the 2^2n pairs of n-bit numbers, only 2^n will have any chance at all to be chosen by the prng (because the prng has period 2^n at most in this case). regards, Janosch
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.