Re: How S chooses the .Random.seed

[Patrick Burns <pburns@pburns.seanet.com>]

Not updating the random seed when an error occurs is
done on purpose -- and is a good thing.  That way if an
error only happens with some random seeds, you have
an opportunity to fix your function.

If you want to change the seed, you can use "set.seed".

S Poetry has some other cautions on random seeds.

Patrick Burns

Burns Statistics
patrick@burns-stat.com
+44 (0)20 8525 0696
http://www.burns-stat.com
(home of S Poetry and "A Guide for the Unwilling S User")

roula tsonaka wrote:

> Dear S-PLUS users,
>
> I have made a function that is based on an iterative
> algorithm which uses randomly chosen initial values
> from a distribution. After many trials I realized that
> when an error occurs and the procedures stops, the
> re-running of the function starts from the same
> initial values. In order to check out if this is a
> defect of my function I tried the following function:
>
> ff <- function(n){
>  x <- rnorm(n)
>  print(x)
>  stop("problem")  
> }
> ff(10)
>
> which shows that every time the same values are
> simulated after the procedure has stopped. Thus, I
> would like to ask whether such a problem has something
> to do with how S chooses the .Random.seed or is due to
> something else
> and how it can be overcome. I am using S-PLUS 6.1
> Release 1
> in Windows XP Home.
>
> Thanks in advance for your suggestions. 
>
> Roula
>
> =====
> -----------------------------------------------------------
> Spyridoula I. Tsonaka
> Graduate Student
> Department of Statistics
> Athens University of Economics and Business 
> (http://www.stat-athens.aueb.gr)
> ____________________________________________________________
> Do You Yahoo!?
> Αποκτήστε τη δωρεάν @yahoo.gr διεύθυνση σας στο http://www.otenet.gr
> --------------------------------------------------------------------
> This message was distributed by s-news@lists.biostat.wustl.edu.  To
> unsubscribe send e-mail to s-news-request@lists.biostat.wustl.edu with
> the BODY of the message:  unsubscribe s-news
>
>
>