[R] Parallel Programming

[Bert Gunter]

Inline below.

On Fri, Sep 21, 2012 at 9:05 AM, R. Michael Weylandt
<michael.weylandt at gmail.com> wrote:
> On Fri, Sep 21, 2012 at 5:43 AM, Tjun Kiat Teo <teotjunk at gmail.com> wrote:
>> I am trying to do parallel programming and I tried this
>>
>> library(doSNOW)
>> library(foreach)
>>
>> testfunc<-function(x){
>> x<-x+1
>> x
>> }
>>
>> noc<-2
>>
>> cl <- makeCluster(do.call(rbind,rep(list("localhost"),noc)), type = "SOCK")
>> registerDoSNOW(cl)
>> clusterExport(cl=cl,c("testfunc.r"))
>>
>>
>> testl<-foreach(pp=1:2) %dopar% {
>> testfunc(pp)
>> }
>>
>>
>> And this works but if I try to enclose my commands inside a text file
>> to be sourced it doesn't work
>>
>> noc<-2
>>
>> testfunc<-function(x){
>> x<-x+1
>> x
>> }
>>
>> cl <- makeCluster(do.call(rbind,rep(list("localhost"),noc)), type = "SOCK")
>>
>> registerDoSNOW(cl)
>>
>> clusterExport(cl=cl,c("a","testfunc.r"))
>>
>> testl<-foreach(pp=1:2)) %dopar% {
>> source("test.r")
>> }
>
> I'm not sure this is a parallelization issue: when you source() a
> file, it doesn't return the last value calculated in quite the way
> that running a function directly does.

I believe you are getting confused here.

1) The return of the source() function is a list, not a value, as
Michael's example showed. This appears to be undocumented, however.

2) source() parses and evals the file's contained code in the
specified environment, so, e.g. all assignments are carried out in
that environment, **unlike** functions in which the assignments are
local to the function environment.

e.g. Suppose test.R has the code:
x <- 2

Then:
> x <- 5
> source("test.R")
> x
[1] 2

...  but this last value is **Not** autoprinted, however. As the docs say:

"Note that running code via source differs in a few respects from
entering it at the R command line. Since expressions are not executed
at the top level, auto-printing is not done. "

For these reasons, it seems generally wiser to encapsulate desired
behavior in a function rather than sourcing in a script from a file,
though there are of course issues here, too (e.g. scoping, etc.).

Cheers,
Bert


>
> E.g.,
>
> system("echo 2+2 > test.R")
> x <- source("test.R")
>
> identical(x,4) # FALSE
>
> str(x) # Look at what actually returned
>
> source("test.R")[[1]] # Actual result
>
> Perhaps that's tripping you up?
>
> Cheers,
> Michael
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.



-- 

Bert Gunter
Genentech Nonclinical Biostatistics

Internal Contact Info:
Phone: 467-7374
Website:
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm