Re: Prng Alvo

On Fri, 17 Apr 2009 15:35:07 -0700 (PDT), Phoenix
<ribeiroalvo@xxxxxxxxx> wrote:


I put the algorithm in this forum because I was suggested by a person
of NIST.But, also put in sci.math and almost no discussion.
About the seed 1.000000000000001, Is no entropy.
Any number has exactly the same entropy as any other number. You need
to either find a PRNG that accepts any seed or introduce a seeding
mechanism that avoids all bad seeds.

See this to put on the algorithm program:

IF s=INT(s) THEN s=67583.87256347297

I use this on my BASIC routine to avoid that kind of problems.
The probality of the seed to be bad, is litle.
You need a probability of zero. There are already PRNGs which can
accept any seed value.

But we all now the seeds on the PRNG are like the biologic seeds:
Goods and bads.
Generate implies this: Whith a good seed and a generator, we have good
results, otherwise not.
Marsagia tell us about the bad and good seeds: The famous 6969, if I
am not wrong.

If you allow me, I ask to you a small favor. Can you colaborate white
me, making a java script to generate random integer numbers to put on
my web pag?
Javascript is not the same as Java. I can write Java, I cannot write

I do not no to write in java.
This is the Java that I used:

static public void main(String[] args) {
// See
double seed = 0.06;
System.out.println("seed = " + seed);
double x = seed;
int count = 0;
for (int i = 0; i < 500; ++i) {
final double temp = x * seed;
x = temp - Math.floor(temp);
if (count == 50) {
count = 0;
} // end if
} // end for


} // end main()

At this point I was looking at how fast small seeds (< 1.0) converged
on zero.