OpenEuphoria

Re: 32-bit random numbers

• Posted by Al Getz <Xaxo at aol.com> Jul 03, 2004
• 580 views

Hello again,

I have to agree with Rob, that the probability distribution *might*
not be acceptable for some things, but i dont think Derek's 
solution was that far off.  I'd make a slight change to improve
it a little and i'd use if myself for everything except applications
that require encryption of data:

constant n=#10000

function Rand32()
  --no reason why you cant make this more inline...
  atom a,b,c

  a=rand(n)-1
  b=rand(n)-1
  c=a*n+b
  return c
end function

P.S. We havent heard from Wolf yet???

Take care,
Al

And, good luck with your Euphoria programming!

My bumper sticker: "I brake for LED's"


Robert Craig wrote:
> 
> Tommy Carlier wrote:
> > How to create a 32-bit random integer between 1 and n (n = 32-bit integer
> > atom):
> > 
> > function rand32(atom n)
> >     if n <= #3FFFFFFF then return rand(n)
> >     else return rand(n - #3FFFFFFF) + rand(#3FFFFFFF) - 1
> > end function
> > 
> > PS: if you leave out the -1 at the end, the number 1 will never be chosen.
> 
> That will work, but keep in mind that the probability distribution
> of the sum of two random numbers is not flat. For example,
> if you roll two dice (1-6), the sum is far more likely to 
> be 7 than 12. So what would I do? I'm not sure, but I guess you
> could take two random 16-bit numbers and concatenate them into
> a 32-bit number. I'd still have some slight worries though,
> since the two 16-bit numbers would be consecutive numbers
> from the same random number generator. Are they truly independent?
> Of course not, but then rand() is really just giving you 
> "pseudo-random" numbers anyway.
> 
> Regards,
>    Rob Craig
>    Rapid Deployment Software
>    <a href="http://www.RapidEuphoria.com">http://www.RapidEuphoria.com</a>
>