reported ?RNG bug

[Thomas Lumley]

On Mon, 9 Nov 1998, Mai Zhou wrote:

> RNG in R and Splus 3.4
> 
> Prof. Ripley asked the details of the example. 
> We were doing parametric bootstrap, so it is similar to simulation.
> Anyway here is the details.

The following code replicates this without needing any specific
hand-coding of numbers each time it is run (except for the slight chance
that y2 will be shorter than 1900 elements)

  x1 <- rexp(100, rate=0.3)
  theta<-4
  x2 <- x1[x1<theta] 
  x3 <- x2[1:19]
  y1<-rexp(4000, rate=0.3)
  y2<-y1[y1< max(x3)]
  y3<-y2[1:1900]   
  y4 <- matrix(y3, ncol=19, byrow=T)
  maxboot <- apply(y4, 1, max)
  biasboot <- mean(maxboot) - max(x3)  
  biasboot

I ran this 1000 times on S-PLUS 3.4 for Solaris, Rpre0.63 for Solaris,
Guido's R 0.61.3 for Windows (NT)  and R0.61.3 for Linux (the oldest R I
have lying around) and got

S-PLUS 		-0.3040023
Rpre0.63	-0.3083820
R 0.61.3	-0.3054932
WinNT	 	-0.307758

I don't have easy access to RJune, but it seems that on the other current
and recent Rs the example works consistently.

Anyone else tried this on RJune?

	-thomas


Thomas Lumley
Assistant Professor, Biostatistics
University of Washington, Seattle.

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._