Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
To: Kbwms <Kbwms AT aol DOT com>, djgpp AT delorie DOT com
From: Nate Eldredge <nate AT cartsys DOT com>
Subject: Re: Problem with Function rand()
Date: Sun, 26 Apr 1998 20:23:06 -0700
Message-ID: <19980427032245.AAH5484@ppp100.cartsys.com>
Precedence: bulk

At 12:23  4/26/1998 EDT, Kbwms wrote:
>While completing an article entitled _Quick and Portable Random
>Number Generators_ with Gerald P. Dwyer (Clemson University), I
>ran tests on random number generators in several C libraries,
>including Microsoft, Borland, and GNU.  The 16-bit versions in
>the Microsoft and Borland libraries seem to behave well enough
>but their cycle-length is far too short for serious work.
>
>The availability of 32-bit output from a Lehmer generator led
>me to test the version of rand() in the GNU C library.  When
>possible, the first test to be conducted is the spectral test.
>(See Knuth, The Art of Computer Programming, Vol 2, 1981, pp.
>89-110.)

I doubt you mean the *GNU* C library, else this post is off topic here.

[snip discussion of using different constants in RNG]

Okay, that is worth considering.

Incidentally, did you try `random' in its various modes? See its source for
info on how to use them.

Nate Eldredge
nate AT cartsys DOT com