Re #1. Ah, maybe then it's because I included --nomask? I assumed there were no zero entries, so it wouldn't be a problem.
Re #2. Thanks, that makes sense. So it sounds like increasing those values might speed up the process but likely not by much since each incremental step is working with a larger chunk of data?
And, assuming there's no way to run MELODIC multithreaded, would my best option for speeding things up be to run migp prior to MELODIC with the Matlab code referenced in the migp paper?
Thanks for your help,
Andrew
|