[R-sig-hpc] run parallel R sessions

[Norm Matloff]

I'm not sure what you mean by not sharing data on the a,b values, but
the rest of your remarks suggest some sharing.  If so, my packge Rdsm on
CRAN might be helpful.  It enables separate R invocations to share data
with each other, including lists.

Norm Matloff

On Thu, Nov 04, 2010 at 12:06:32PM -0400, Prasenjit Kapat wrote:
> Hi,
> 
> Looking for suggestions to run parallel R sessions. Consider the
> following setup:
> 
> params.R: this file sets up a few simulation parameters which I
> generally modify manually
> funs.R: collection of functions needed for the simulation
> script.R: runs the main simulation using the parameters defined in
> params.R and the functions in funs.R
> 
> There are two parameters, say, a, b (defined in params.R) for which I
> want to run (in parallel) the simulation over a large number of
> possible values. Suppose the simulation code (script.R) produces a
> list. I would like to collect all these lists and then do some more
> investigation of some of the simulation output over the grid of a x b
> values.
> 
> How, do you suggest, should I go about implementing this? I would
> prefer that the R thread/process for (a_i, b_j) NOT interact (ie not
> share any data) with that for (a_m, b_n).
> 
> Thanks for any help,
> -- 
> Prasenjit
> 
> PS:
> The compute servers are 64bit RedHat EL 5.5 running R 2.11.1 (8 cores)
> 
> _______________________________________________
> R-sig-hpc mailing list
> R-sig-hpc at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-hpc