Hi Bastian,
that's fine, you just add the results. What makes you skeptical? Remember that the intensity value for a voxel in fdt_paths represents the number of curves that have been launched from the seed and go through. This won't depend (assuming some sort of convergence) on the way you launch the curves, serially or in parallel.
Cheers,
Stam
On 21 Oct 2011, at 09:00, Bastian Cheng wrote:
> Thank you very much! At least, I can stop searching for missed options on my SGE.
>
> If I may add another question:
>
> I can of course split one probtrackx-job in 4 parallel jobs using the --rseed option (thereby choosing --rseed=1, -=2, =3 and =4 and 1/4 of streamlines each, i.e. 4 times 1250 to get 5000).
>
> This successfully reduces processing time, but I'm ending up with 4 different outputs: How do I get them together to form the desired result (as if I had tracked 5000)?
>
> Is this method solid anyway? I am a bit skeptical...
>
> Thanks again for your help!
>
> Bastian
>
|