Re: Random number generator without duplicates
Re: Random number generator without duplicates
- Subject: Re: Random number generator without duplicates
- From: Emmanuel <email@hidden>
- Date: Fri, 20 Apr 2001 18:14:51 +0200
At 13:59 +0200 20/04/01, Bill Briggs wrote:
>
>
At 12:23 AM -0700 20/04/01, Bryan Harris wrote:
>
> > From that result, you can see that Applescript (v 1.6) produces
>
>no appreciable
>
>> bias.
>
>
>
>Aside from your (single-run) experiment results, I'd love to see
>
>your proof of this statement.
>
>
I ran it several times, increasing the number of iterations. Went to
>
bed and let it run 3,000,000 times, with collections after a
>
million. The results were as follows:
>
>
{99903, 99363, 100703, 100181, 100183, 99828, 100606, 99779, 99933, 99521}
>
{99941, 100070, 100043, 99955, 99908, 100184, 100269, 99777, 100073, 99780}
>
{99932, 99889, 99527, 100427, 100141, 99808, 100453, 99570, 99929, 100324}
>
>
{299776, 299322, 300273, 300563, 300232, 299820, 301328,
>
299126, 299935, 299625}
>
>
Of course I couldn't do an infinite number events, but I don't see
>
anything that looks like a bias in the results. The largest variation
>
shows a difference of about 0.3%.
If they were ideal random variables, typical variation should be square
root of (10 / 3,000,000), i.e. 0.0018, very consistent with what you found
as a largest variation.
Emmanuel