Hi Jacalyn,

It's difficult to know exactly what the problem is, but it might have to do with the interaction of PBS with the parallel package (which Amelia uses). You might try to see if you can run the example scripts on PBS:

library(parallel)

example(mclapply)

If those don't work, there might be an issue between parallel and PBS.

Cheers,

Matt

~~~~~~~~~~~

Matthew Blackwell

Assistant Professor of Political Science

University of Rochester

url: http://www.mattblackwell.org

On Fri, Jul 12, 2013 at 2:11 PM, Jacalyn Huband <jmh5ad@virginia.edu> wrote:

I have been able to run the "multicore" option of amelia on a single (Linux) computer with 8 cores, but I am having problems porting my program to a (Linux) cluster.

Specifically, I have requested (through the PBS resource management software) a single node with 16 cores. My program appears to run correctly, but then gets killed by PBS (error message: ncpus 4.01 exceeded limit 1). In the amelia function, I'm setting m = 5, parallel = "multicore", and ncpus = 1. I've tried other values for ncpus (including 5, 10, and 16), but the job still gets killed. (Note: It does not get killed if m = 1, but that defeats the purpose of using multiple cores.)

Is the amelia function trying to spawn threads of which I am unaware?

J. Huband

--
Amelia mailing list served by HUIT
[Un]Subscribe/View Archive: http://lists.gking.harvard.edu/?info=amelia
More info about Amelia: http://gking.harvard.edu/amelia
Amelia mailing list
Amelia@lists.gking.harvard.edu

To unsubscribe from this list or get other information:

https://lists.gking.harvard.edu/mailman/listinfo/amelia