Kenan Sulayman wrote:
> Hello PHP internals,
>
> I've been running several million tests regarding the
> arithmetically-random-functions.
>
> The returning value at the end:
> Avg: mt_r: 1073848211.4106 r: 16384.155746685 ~8016909
> Avg: mt_r: 1073904131.0286 r: 16384.164383921 ~8503575
> Avg: mt_r: 1074010520.4456 r: 16384.091213092 ~13136327
>
> Shape:
> Avg: mt_r: { avg of mt_rand() } r: { avg of rand() } ~ { value of $i (
> iterations ) }
>
> Means:
> Average of mersenne twister in 8016909 it's is 1073848211.4106 (
> corresponding to that in 8503575 it's the avg. is 1073904131.0286).
> That's acceptable.
>
> The rand(x) function alters the avg in 5M iterations really by just 0.1 ?
> Wow, that's creepy.
>
You're exhausting the period. It's not creepy, it's expected, for a
random number generator with only 16 bits of state.
> Is there any way - to make it better ( or a bit as good as the m_twister is
> ? ) ?
>
No, apparently it's conventional to expose whatever crap PRNG the system
has lying around instead of forcing programmers to use one of the two
perfectly good ones that PHP bundles.
-- Tim Starling
--
PHP Internals - PHP Runtime Development Mailing List
To unsubscribe, visit: http://www.php.net/unsub.php
