Hello,
I'm trying to use Multiblock PLS with Parallel Computing Toolbox, in order to speed up the lengthy computations. However, when I try to run the analysis using either Regular or Non-Rotated Multiblock, I get the following error:
Starting parallel pool (parpool) using the 'local' profile ... connected to 4 workers.
Error using splithalf_perm_par>(parfor supply) (line 42)
Index exceeds matrix dimensions.
Error in splithalf_perm_par (line 42)
parfor op = 1:num_perm
Error in pls_analysis (line 1440)
[sop, ucorr_distrib, vcorr_distrib] = splithalf_perm_par(sop, ...
Error in fmri_pls_analysis (line 295)
result = pls_analysis(grp_datamat, num_subj_lst, num_conditions, opt);
Error in bfm_analysis_ui>ExecutePLS (line 2346)
fmri_pls_analysis(PLSoptions.profiles, 'MULTIBLOCK', ...
Error in bfm_analysis_ui (line 118)
ExecutePLS;
Caused by:
Index exceeds matrix dimensions.
Error while evaluating uicontrol Callback
When I switch off the parallel pool, the same analysis runs smoothly:
Parallel pool using the 'local' profile is shutting down.
op = 1
I do not get the error for other variants of PLS - Behavioral and Task run smoothly on multiple cores. I'm using MATLAB R2013b on Windows machine, with the latest version of your software. Is this a bug? Is there an easy workaround?
Cheers,
Marcel
Hi Marcel,
With the holidays there was some delay in getting back to you. I am not 100% sure what the source of the error is and will need to run a few tests. Can you give me a better idea of the nature of the dataset you are using so I can create a toy example to test?
Hi Marcel,
With the holidays there was some delay in getting back to you. I am not 100% sure what the source of the error is and will need to run a few tests. Can you give me a better idea of the nature of the dataset you are using so I can create a toy example to test?
Hello,
I'm analyzing a simple blocked paradigm, with 2 task and 1 rest conditions. I got 30 subjects. The behavioral table consists of 9 variables, so it's size is 90x3. I hope this helps.
Best regards,
Marcel
Baycrest is an academic health sciences centre fully affiliated with the University of Toronto
Privacy Statement - Disclaimer - © 1989-2024 BAYCREST HEALTH SCIENCE. ALL RIGHTS RESERVED