[SciPy-User] FAIL: gaussian filter 3

[Warren Weckesser]

On Sat, Nov 20, 2010 at 5:15 AM, Nils Wagner
<nwagner at iam.uni-stuttgart.de>wrote:

> Hi all,
>
> is this a known issue ?
>
> ======================================================================
> FAIL: gaussian filter 3
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File
>
> "/home/nwagner/local/lib64/python2.6/site-packages/nose-0.11.2.dev-py2.6.egg/nose/case.py",
> line 183, in runTest
>     self.test(*self.arg)
>   File
>
> "/home/nwagner/local/lib64/python2.6/site-packages/scipy/ndimage/tests/test_ndimage.py",
> line 468, in test_gauss03
>     assert_almost_equal(output.sum(), input.sum())
>   File
> "/home/nwagner/local/lib64/python2.6/site-packages/numpy/testing/utils.py",
> line 463, in assert_almost_equal
>     raise AssertionError(msg)
> AssertionError:
> Arrays are not almost equal
>  ACTUAL: 49993304.0
>  DESIRED: 49992896.0
>
>

Hi Nils,

Thanks for reporting the problem.  It looks like this failure was introduced
in r6837, when the ndimage tests were cleaned up.  The old version of
test_gauss03() computed the sums of the 32 bit floats using 64 bit
accumulators, but the new version did not.  However, simply changing the
size of the accumulators did not fix the test, because r6837 actually fixed
a bug in the old test: the old test was testing the difference of the actual
and expected sums, rather than the *absolute value* of the difference.  It
turns out that the default precision requested by the test was too high.

The test has been fixed in r6927.

Warren
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20101120/43839cbf/attachment.html>