Hi Holly,
My models were all with just two nodes but 5 hours per step sounds too
much indeed. Eight nodes is indeed quite heavy. Something like 4 is
more reasonable. Look what's the most CC had in any of her papers. One
thing you can do is reduce the number of frequency modes. The default
is 8 and it's definitely too much. Try reducing it to something like 4
and if there is still anything interesting in the 4th mode, increase
it by 1 etc. If not perhaps reduce it even to 3.
Best,
Vladimir
> From: Holly Rossiter [mailto:[log in to unmask]]
> Sent: 16 March 2011 16:43
> I meant to ask you a question about DCM while I was there. I’m running DCM
> for induced on my data and it is taking forever. Literally about 5 hours for
> one line of the estimation and it keeps crashing. I know it’s slightly
> different to SSR so might not be comparable but roughly how long does a
> normal DCM-SSR take for one subject-one model? I’ve got 8 nodes and 120
> trials of 3 seconds to analyse but i’m wondering if I need to change my
> parameters or something to simplify things.
>
>
>
> Thanks,
>
>
>
> Holly
|