griff - 2012-07-23

Hey, So I'm trying to create a parallel processed tool to run the simulations with. I can do it using python's subprocess module, but had to do some work around. I create the model and run the sampling in a separate process. Which works well, but I can only save the database from the sampling. The model MCMC object will be lost (as far as I know…wasn't able to find a way to pickle that as well). So in my main process I can load the db, but in order to do any more resampling, I will need to recreate the model MCMC dictionary again. Because I am just a software engineer, and not an analyst, I'm unsure of the repercussions of doing this. Will recreating the model MCMC dictionary, in exactly the same fashion that I had before the sampling, affect the results?

Also, is there a way to retrieve the Models last state from the DB. I read online that the DB can in fact hold this information, but was unsure as how to retrieve it.

In the mean time I'm forced to run this tool using the Threading module, and I'm losing about 3 to 5 times the speed. :\

Thanks for any help in advance!