From: Robert Klemme <bob.news@...>
Date: 2004-09-16T08:05:01+09:00
Subject: Re: Bignum Troubles


"James Edward Gray II" <james@grayproductions.net> schrieb im Newsbeitrag 
news:DDD9DF2E-0761-11D9-9076-000A95BA45F8@grayproductions.net...
> On Sep 15, 2004, at 2:32 PM, Jim Weirich wrote:
>
>> Several things look wrong here.
>
> That's because your math is a lot better than mine.  <laughs>  I 
> recognized the division as >> 2**16, but I didn't get the 2**14 modulo.
>
> You're right about the array, it was dumb.
>
> You also optimized my routine incredibly with the % to the seed.  Very 
> nice.

The algorithm looks to me, as if random numbers were not equally distributed 
because of the "% limit".  Is that so and if so is that a desired effect?

> Ironically, as it turns out, my number generator was fine, if super slow. 
> I did fine the problem elsewhere.  Guess it wasn't such a good guess after 
> all.

Ha! -)

> Thanks to all who helped!

You're welcome.

Just one other remark, there is a lib that changes the semantics of int 
division.  I don't remember the name though.  So 5 / 3 is not always a 
Fixnum.

Kind regards

    robert