Bug in rnorm. (PR#1664)

Duncan Murdoch dmurdoch@pair.com
Thu, 13 Jun 2002 21:03:16 -0400


On Fri, 14 Jun 2002 00:39:58 +0200 (MET DST), p.dalgaard@biostat.ku.dk
wrote:

>With your script, I get a nice horizontal line both with 1.4.0 and
>1.5.0-patched.
>
>(RedHat Linux 7.1 on a PC) 

I get what Rolf described in 1.5.0-patched for Windows with the
default generators: 

RNGkind()
[1] "Marsaglia-Multicarry" "Kinderman-Ramage"  

I switched to the Ahrens-Dieter normal RNG (using
RNGkind(,'Ahrens-Dieter')), and things were fine.  

Keeping the default normal generator but switching to the
Mersenne-Twister uniform also fixed it:

> RNGkind()
[1] "Mersenne-Twister" "Kinderman-Ramage"

Is it possible you're not using the default generators?  If you are,
this is quite weird, because we're on the same hardware.  Why would
the OS or compiler matter in a calculation like this??

Duncan Murdoch
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._