home *** CD-ROM | disk | FTP | other *** search
- Path: hubcap.clemson.edu!hubcap!mjs
- From: mjs@hubcap.clemson.edu (M. J. Saltzman)
- Newsgroups: comp.lang.c,comp.lang.c++
- Subject: Re: RANDOM NUMBER GENERATOR
- Date: 3 Jan 96 23:44:35 GMT
- Organization: Clemson University
- Message-ID: <mjs.820712675@hubcap>
- References: <ABg4lwmmK4@lh.ioffe.rssi.ru> <4cf1d2$1ts@oldfart.ecl.wustl.edu>
- NNTP-Posting-Host: hubcap.clemson.edu
-
- "Klaus L. Nordstrom" <kln@news.hpn.org> writes:
-
- >"Eugene I Levin" <Levin@lh.ioffe.rssi.ru> wrote:
- >>
- >>Tom, you are RIGHT! The congruent generator which is employed in
- >>rand() is the bad one! And everybody know that.
-
- This should read "commonly employed in rand(). The implementor is
- under no obligation to use that same old random number generator,
- and indeed many implementations don't. It would be nice if more
- implementors paid more careful attention to this issue, though.
-
- >If one uses <stdlib.h>, does one not automatically get access to
- >srand48() and drand48()? These routines generates good quality random
- >numbers with long periodicities.
-
- One does not in ANSI/ISO Standard C, although many Unix implementations
- provide them as an extension.
-
- >Also, you may wish to check "numerical Recipies in C", authored by
- >Teukolsky and Press.
-
- >-kln.
-
- --
- Matthew Saltzman
- Clemson University Math Sciences
- mjs@clemson.edu
-