Re: largely OT: a random-number generator and bzip2


"megagurka" <megagurka@xxxxxxxxx> wrote in message
news:1127386708.874672.191270@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
> What's wrong with an existing PRNG's like the Mersenne twister or some
> cipher based algorithm? Anyway, you can use Diehard for measuring the
> quality of a set of random numbers.
>
well, it would cost me the experience of doing it myself.

I had underestimated initially how much I would end up messing with it
(original intent had more been to do something comprable to 'rand').

now the mystery is what to do with it exactly, well, I have a use, but,
since it exists I can put it in a few places (eg: both in my physics lib,
and probably the core part of my "general" lib).

it did give a few vaguely interesting thoughts related to entropy.

what else? dunno...

diehard:

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:: This is the BIRTHDAY SPACINGS TEST ::

....
:::::::::::::::::::::::::::::::::::::::::
The 9 p-values were
.299080 .260979 .319984 .527126 .754357
.730711 .661000 .565234 .950486
A KSTEST for the 9 p-values yields .369946

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:: THE OVERLAPPING 5-PERMUTATION TEST ::

chisquare for 99 degrees of freedom=143.443; p-value= .997632
chisquare= 5.626 for 3 d. of f.; p-value= .877569
chisquare= 4.033 for 3 d. of f.; p-value= .764404

....

TEST SUMMARY, 25 tests on 100,000 random 6x8 matrices
These should be 25 uniform [0,1] random variables:
.259442 .023760 .642690 .902768 .340336
.415662 .868232 .998924 .308241 .189915
.262428 .936387 .847822 .520848 .703802
.102346 .182456 .560653 .848073 .806561
.695445 .248770 .619447 .107232 .760210
brank test summary for rand0.bin
The KS test for those 25 supposed UNI's yields
KS p-value= .120352

....


lots of tests. all in all, I *think* it did ok (it doesn't help I don't
understand much of what is going on or how to interpret the results).

> /JN
>


.