Running

Running the simulations

Now, for the moment you’ve been waiting for (I assume?): running the simulation study. To run all the simulations, run:

$ glurmo -r 5 .

This will submit all five simulations. Note that you can replace “5” with any positive number, and glurmo will run that many (or at least the maximum number in the study). Note that glurmo will keep track of what simulations have and have not been run; if you run the same command above again, glurmo will submit no jobs, since all jobs will either be running or completed.

These scripts are quite simple, and should run basically instantaneously (though of course this depends on the slurm load). Once the scripts have completed, we can look at what has changed in the sub-directories.

results

The results folder should now have 5 files: results___0.RData through results___4.RData. These are the results we saved from each of the respective scripts.

slurm_out

This directory contains the output generated by slurm for each simulation. Per our slurm file, there will be two files per simulation: output___{index} and output___{index}.Rout. The output___{index} files should be empty, but the .Rout files show what code was run and how long it took. They will also show errors when they occur, which is extremely useful for debugging simulations.

slurm_errors

slurm_errors contains the error output for each simulation’s slurm script. Each output file will be of the form error___{index}. However, since nothing went wrong in these simulations, all files should be empty.