Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sun-barr!newstop!texsun!convex!convex.COM
From: dodson@convex.COM (Dave Dodson)
Newsgroups: comp.lang.fortran
Subject: Re: Random Number Functions--Generate versus Input
Message-ID: <108690@convex.convex.com>
Date: 13 Nov 90 22:11:04 GMT
References: <1990Nov8.170334.24579@athena.mit.edu> <1990Nov8.215752.7075@athena.mit.edu> <108474@convex.convex.com> <1990Nov13.053812.19914@athena.mit.edu>
Sender: news@convex.com
Reply-To: dodson@convex.COM (Dave Dodson)
Organization: Convex Computer Corporation; Richardson, TX
Lines: 17

In article <1990Nov13.053812.19914@athena.mit.edu> oliver@athena.mit.edu (James D. Oliver III) writes:
>Your algorithm is the same as the original scheme proposed by von Neumann
>for generating random vectors.  It's efficiency is given by the area of a
>unit cube divided by a unit sphere, so 18/pi=5.73 random numbers plus a
>square root must be generated for each vector, as opposed to the 2.55 plus
>a square root for the Marsaglia method I posted in a recent follow-up

The intended contribution in my posting was how to vectorize the generation
of random unit vectors.  I suggest you combine Marsaglia's rejection algorithm
with my use of whenfle to produce a bunch of unit vectors in vector mode.
With the entire process vectorized, I would expect it to be much faster to
generate them than to input them.

----------------------------------------------------------------------

Dave Dodson		                             dodson@convex.COM
Convex Computer Corporation      Richardson, Texas      (214) 497-4234