R help - Jun 2004 - how to initialize random seed properly ?

I want to start R processes on multiple processors from single shell 
script
and I want all of them to have different random seeds.
One way of doing this is

        sleep 2 # (with 'sleep 1' I am often getting the same number)
               ...
        set.seed(unclass(Sys.time()))

Is there a simpler way without a need to sleep between invoking
different R processes ?

Ryszard

For reproducibility, you probably want to specify the starting seeds.
I'd worry about using systime.   (for selecting the seeds, you might
consider random draws from a uniform).

best,
-tony

ryszard.czerminski at pharma.novartis.com writes:
> I want to start R processes on multiple processors from single shell 
> script
> and I want all of them to have different random seeds.
> One way of doing this is
>
>         sleep 2 # (with 'sleep 1' I am often getting the same
number)
>                ...
>         set.seed(unclass(Sys.time()))
>
> Is there a simpler way without a need to sleep between invoking
> different R processes ?
>
> Ryszard
>
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html
>
-- 
rossini at u.washington.edu            http://www.analytics.washington.edu/ 
Biomedical and Health Informatics   University of Washington
Biostatistics, SCHARP/HVTN          Fred Hutchinson Cancer Research Center
UW (Tu/Th/F): 206-616-7630 FAX=206-543-3461 | Voicemail is unreliable
FHCRC  (M/W): 206-667-7025 FAX=206-667-4812 | use Email

CONFIDENTIALITY NOTICE: This e-mail message and any attachme...{{dropped}}

> I want to start R processes on multiple processors from single shell 
> script
> and I want all of them to have different random seeds.
The usual way would be using the process ID for the initialization.
Or the shell script could just hand out consecutive numbers 1, 2, 3, ...

  Lutz

Here is my best solution so far using $RANDOM in bash
or maybe somebody has "pure R"  solution ?

#!/bin/bash
for i in 1 2; do
p=tmp$i
cat > $p.R << EOF
set.seed($RANDOM)
cat('rnorm(3) =', rnorm(3), '\n')
EOF
nice R CMD BATCH --no-save --no-restore $p.R $p.log &
done
sleep 3
grep rnorm tmp[12].log

Best regards,
Ryszard





Ryszard Czerminski/PH/Novartis at PH
Sent by: r-help-bounces at stat.math.ethz.ch
06/09/2004 03:24 PM

 
        To:     r-help at stat.math.ethz.ch
        cc: 
        Subject:        [R] how to initialize random seed properly ?


I want to start R processes on multiple processors from single shell 
script
and I want all of them to have different random seeds.
One way of doing this is

        sleep 2 # (with 'sleep 1' I am often getting the same number)
               ...
        set.seed(unclass(Sys.time()))

Is there a simpler way without a need to sleep between invoking
different R processes ?

Ryszard

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html

perldoc -f srand
has a quick and good advice on initializing a random sequence.
I second the recommendation to prepare the list of random numbers 
in advance for sake of reproducibility.

	Itay
--------------------------------------------------------------
itayf at fhcrc.org		Fred Hutchinson Cancer Research Center