Hello Vasilisa,
If you upload your design.con/mat to
http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi
and let me know the 6-digit reference number, we will check what is going wrong.
Many Regards
Matthew
> Dear FSL users,
>
> I try to set up a randomise 2-group design with two covariates (one continous one categorical) but I'm keeping to receive the same error.
> My design matrix :
> /NumWaves 4
> /NumPoints 33
> /PPheights 1 1 1 1
> /Matrix
> EV1 EV2 EV3 (gender) EV4
> 0 1 1 demeaned cov4
> 0 1 -1 demeaned cov4
> 0 1 1 demeaned cov4
> 1 0 -1demeaned cov4
> 1 0 -1 demeaned cov4
> 1 0 1 demeaned cov4
> .
> .
> .
> my design.con are:
> /NumWaves 4
> /NumContrasts 4
> /PPheights 1 1 1 1
> /Matrix
> 1 -1 0 0
> -1 1 0 0
> 0 0 0 1
> 0 0 0 -1
>
> the randomise command is : randomise options: -i all_FA_skeletonised -o tbss -m mean_FA_skeleton_mask -d design.mat -t design.con -n 500 --T2 -V
>
> The error:
> Loading Data:
> Data loaded
> ERROR: Program failed
>
> An exception has been thrown
> Runtime error:- detected by Newmat: process fails to converge
>
> MatrixType = Rect # Rows = 33; # Cols = 3
> Trace: SVD; pinv.
>
>
> Exiting
>
> It seems that it does not process all the 4 columns of the design...
> Does it look like some text:syntax error or something is conceptually wrong?
> I tried to set up the matrix via GLM gui but the error persists.
> I also tried to run it with the non-demeaned covariate since Jeannette's recent guideline explains that I probably do not need mean centering as I'm not interested in the individual group means.
> I also tried to run it with the demeaned data but it is the same error.
> I recently ran a similar design on Mac with the same version of FSL 4.1.7 and it seemed to work. Now I'm working on Linux machine.
> would greatly appreciate any comments and prompts.
> Vasilisa.
>
|