Dear FSL listers,
Following up on an unanswered query, has anyone else had problems running the output of setup_masks through randomise or randomise_parallel?
Randomise seems to throw an exception when the number of columns in the matrix is greater than the number of rows.
Under normal circumstances there would be more subjects (i.e. rows) than EVs (i.e. columns), but as I understand it the whole point of the setup_masks script is to add an EV of a single 1 and multiple 0s for each subject/row. This gives a square like this:
/Matrix
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
However, with the addition my two group EVs and 2 covariate EVs, the resulting final matrix in the .mat file is always going to have more columns than rows, e.g.:
/Martix
1 0 1.3 12 1 0 0 0
1 0 1.2 19 0 1 0 0
0 1 1.6 14 0 0 1 0
0 1 1.1 13 0 0 0 1
The error I get is this:
ERROR: Program failed
An exception has been thrown
Logic error:- detected by Newmat: Want no. Rows >= no. Cols
MatrixType = Rect # Rows = 4; # Cols = 8
Trace: SVD.
Exiting
___________
ALONGSIDE this problem, randomise seems to become EXTREMELY memory hungry when trying to run on the output of setup_masks. I often get the 'killed: 9' error when randomise tries to load all the voxelwise EVs (--vxf option), so can't even get to the stage of the error reported above (I'm trying to run this on >100 subjects).
Any thoughts much appreciated, or even reports of people suffering similar problems,
Best wishes,
James
|