[pymvpa] Antwort: Re: searchlight analysis fails

Mon Jun 12 10:11:00 UTC 2017

Dear Nick,

thanks for your answer. Unfortunately, I still end up in an error no matter
what computer I use (but both are VMs). On the one, I just have one CPU, on
the other I tried with setting the nproc = 1 and =4, but the result is the
same (please see below). And no matter what I set, it's always just one CPU
busy with ~100% (RAM is always some free left), so I guess it's one thread,
right? the pprocess package is installed  (version: 0.5-1+nd0~nd13.10+1
+nd14.04+1).

I'd be happy for any advice on this.
Thanks
Lydia

############ Traceback ###############
Running searchlight with radius: 0 ...
[SLC] DBG:     Starting computing block for 338688 elements
Traceback (most recent call last):
  File "load.py2", line 81, in <module>
    sl_map = sl(ds_copy)
  File "/usr/lib/python2.7/dist-packages/mvpa2/base/learner.py", line 258,
in __call__
    return super(Learner, self).__call__(ds)
  File "/usr/lib/python2.7/dist-packages/mvpa2/base/node.py", line 137, in
__call__
    result = self._call(ds, **(_call_kwargs or self._get_call_kwargs(ds)))
  File "/usr/lib/python2.7/dist-packages/mvpa2/measures/searchlight.py",
line 154, in _call
    results = self._sl_call(dataset, roi_ids, nproc)
  File "/usr/lib/python2.7/dist-packages/mvpa2/measures/searchlight.py",
line 375, in _sl_call
    self._proc_block(roi_ids, dataset, self.__datameasure)]
  File "/usr/lib/python2.7/dist-packages/mvpa2/measures/searchlight.py",
line 471, in _proc_block
    res = measure(roi)
  File "/usr/lib/python2.7/dist-packages/mvpa2/base/learner.py", line 258,
in __call__
    return super(Learner, self).__call__(ds)
  File "/usr/lib/python2.7/dist-packages/mvpa2/base/node.py", line 137, in
__call__
    result = self._call(ds, **(_call_kwargs or self._get_call_kwargs(ds)))
  File "/usr/lib/python2.7/dist-packages/mvpa2/measures/base.py", line 514,
in _call
    return super(CrossValidation, self)._call(ds)
  File "/usr/lib/python2.7/dist-packages/mvpa2/measures/base.py", line 337,
in _call
    result = node(sds)
  File "/usr/lib/python2.7/dist-packages/mvpa2/base/learner.py", line 258,
in __call__
    return super(Learner, self).__call__(ds)
  File "/usr/lib/python2.7/dist-packages/mvpa2/base/node.py", line 137, in
__call__
    result = self._call(ds, **(_call_kwargs or self._get_call_kwargs(ds)))
  File "/usr/lib/python2.7/dist-packages/mvpa2/measures/base.py", line 619,
in _call
    measure.train(dstrain)
  File "/usr/lib/python2.7/dist-packages/mvpa2/base/learner.py", line 137,
in train
    self._train(ds)
  File "/usr/lib/python2.7/dist-packages/mvpa2/clfs/libsvmc/svm.py", line
147, in _train
    svmprob = _svm.SVMProblem(labels, src )
  File "/usr/lib/python2.7/dist-packages/mvpa2/clfs/libsvmc/_svm.py", line
221, in __init__
    svmc.double_setitem(y_array, i, y[i])
TypeError: in method 'double_setitem', argument 3 of type 'double'
Exception AttributeError: "SVMProblem instance has no attribute 'data'" in
<bound method SVMProblem.__del__ of <SVMProblem: size = 90>> ignored

Von:	Nick Oosterhof <n.n.oosterhof at googlemail.com>
An:	Development and support of PyMVPA
            <pkg-exppsy-pymvpa at lists.alioth.debian.org>
Datum:	10/06/2017 11:11
Betreff:	Re: [pymvpa] searchlight analysis fails
Gesendet von:	"Pkg-ExpPsy-PyMVPA" <pkg-exppsy-pymvpa-bounces
            +lydia.hellrung=econ.uzh.ch at lists.alioth.debian.org>

On 9 June 2017 at 18:03, <lydia.hellrung at econ.uzh.ch> wrote:
  I'm trying to run a searchlight analysis, but I just end up in error
  messages.
  I followed your searchlight example from the tutorial (detrend, z-score
  and kick out non-used conditions) and then calling the sl function
  exactly as in your tutorial (sl_map = sl(ds_copy)).
  I'm running it on a computer with lots of memory and several CPUs now.
  When I load the full dataset (size 180,338688), I end up with traceback
  #1 (please see below).
  When I load a mask I have less data (size180, 213), I end up with
  traceback #2 (also below).

  I just found such an issue in your mail archive with 3D nifti, but here
  it's 4D nifti.
  What am I missing here?

Not sure what's causing this, but I wonder if it may be related to a high
number of threads that may not play well with pprocess. Could you try to
run the searchlight with as additional parameter in the call: nproc=1
(disabling multiple processes), and maybe also some other values (e.g.
nproc=4 or 8)? Do you know how many threads the searchlight runs with?
_______________________________________________
Pkg-ExpPsy-PyMVPA mailing list
Pkg-ExpPsy-PyMVPA at lists.alioth.debian.org
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20170612/733f5751/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: graycol.gif
Type: image/gif
Size: 105 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20170612/733f5751/attachment.gif>