[pymvpa] errors when running sphere_gnbsearchlight+balancer

[Tren Huang]

Interesting... I don't get this error under Python 2.7.3+PyMVPA 2.2.0.

Tren

On Sun, Aug 4, 2013 at 3:05 PM, Tren Huang <tren.huang at gmail.com> wrote:

> Dear PyMVPA experts,
>
> I successfully ran through the following code using Python 2.6.2+PyMVPA
> 2.2.0:
>
> from mvpa2.tutorial_suite import *
> ds=get_raw_haxby2001_data(roi='vt')
> part=NFoldPartitioner()
> sl=sphere_gnbsearchlight(GNB(),part,postproc=mean_sample())
> map=sl(ds)
>
> However, the same code with a new partitioner led to errors:
>
> part=ChainNode([NFoldPartitioner(),
>                 Balancer(limit='partitions',apply_selection=True)],
>                 space='partitions')
>
> The error messages were:
>
> Traceback (most recent call last):
>   File "debug.py", line 9, in <module>
>     map=sl(ds)
>   File "/usr/local/lib/python2.6/dist-packages/mvpa2/base/learner.py",
> line 239, in __call__
>     return super(Learner, self).__call__(ds)
>   File "/usr/local/lib/python2.6/dist-packages/mvpa2/base/node.py", line
> 96, in __call__
>     result = self._call(ds)
>   File
> "/usr/local/lib/python2.6/dist-packages/mvpa2/measures/searchlight.py",
> line 141, in _call
>     results = self._sl_call(dataset, roi_ids, nproc)
>   File
> "/usr/local/lib/python2.6/dist-packages/mvpa2/measures/adhocsearchlightbase.py",
> line 354, in _sl_call
>     if not (np.all(p.sa[targets_sa_name].value == labels)):
> ValueError: shape mismatch: objects cannot be broadcast to a single shape
>
> Any ideas about what's wrong here?
>
> Thanks!
> Tren
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20130804/89e8465a/attachment.html>