Print

Print


Hi Christian,

Running Melodic on a large data set of 6 subjects with about 1100 volumes per subject, I got a "segmentation fault" error while the final report was being created. Do you have any idea what may be causing this? It seems to me that it happened while the report was creating the "bp_Smodes" figure. I expected a likely RAM issue in the beginning and so I was checking the RAM usage that remained about 36% throughout the one day of running melodic (it's a 16G RM machine).

Thanks,
Danial


On Mar 16, 2010, at 5:54 PM, Christian F. Beckmann wrote:

Hi,

It's easiest to grab the time courses from the tXX.txt files within the report direcotry. These contain (in your case) 10 columns, where the first one is the first principal eigenvector of the other 9 time courses (which themselves are the subject-specific tcs).
hth
Christian

On 15 Mar 2010, at 17:03, Danial Lashkari wrote:

Hi,

I'm applying Tensor ICA to my data from a long visual experiment with 9
subjects. First, I was wondering where exactly the normalized time course
for each component is save. Let's say I have 50 time points and 20
components, the melodic_Tmodes includes a 50 by 200 matrix. I'm assuming
that the second dimension includes the average overall time course of each
component along with its contribution to the response of each of the 9
subjects, making the full (1+9)*20=200 dimensions in this matrix (or the
first column in each of t*.txt files for * between 1 to 20). Is that true?
Otherwise, where can I find the time course that is presented in the
index000.html?

Thanks,
Danial