[FieldTrip] Problems using fieldtrip with Matlab Distributed Computing Server
Haris Tzagarakis
haristz at umn.edu
Fri May 1 21:58:02 CEST 2015
Hi There,
I have been trying to use fieldrtip in conjunction with Matlab Distributed
Computing Server on a supercomputing resource and have been running into a
problem that seems to come from paths/global variables?
in detail: I use fieldtrip in conjunction with my own code for an analysis.
The analysis works fine on a "normal"/non-parallel call. I then set out to
parallelize the code and split the analysis to a different node for every
subject I have. I have access to a supercomputing cluster (not uniquely for
matlab/biomed use), which runs Matlab Distributed Computing Server, so job
scheduling etc can happen from within matlab. I believe this is different
from the 'peer distributed' approach described in the faq. I prepared a
script which is then run with the matlab 'batch' command : job =
batch('parscript3b', 'Profile', 'Itasca_MSI', 'Pool', 10); I attach the
script in its current form here. This call should pass the script to a top
level node that then distributes the loop in 10 nodes, one for each
subject. What always happens however is that I get an error at the top
level node saying: Error using ft_hastoolbox (line 469) the FREESURFER
toolbox is not installed, see http://surfer.nmr.mgh.harvard.edu/fswiki
I have read the help/comments text and some of the code in ft_hastoolbox
and ft_defaults (I in fact always call ft_defaults at startup - my startup
file is also attached here). This in conjunction with what is mentioned in
MATLAB help makes me think that there is something about the way paths and
global variables are handled by MDCS that does not agree with Fieldtrip. I
have been trying to go around the problem by liberally using ft_defaults in
my script code (as can be seen) and I also added some lines to the
ft_hastoolbox function (in all its 4 instances in the private folders) to
check explicitly for the fieldtrip locations in my system but none of this
has worked.
I would be grateful for any insight you might have.
With Thanks and Best Wishes,
Haris
--
Charidimos [Haris] Tzagarakis MD, PhD, MRCPsych
Senior Research Associate
University of Minnesota Dept of Neuroscience
office:
Brain Sciences Center
Minneapolis VA Medical Center
Tel:612-467-1363
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20150501/5a92944b/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: parscript3b.m
Type: text/x-csrc
Size: 679 bytes
Desc: not available
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20150501/5a92944b/attachment-0002.c>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: startup_current.m
Type: text/x-csrc
Size: 552 bytes
Desc: not available
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20150501/5a92944b/attachment-0003.c>
More information about the fieldtrip
mailing list