[SciPy-Dev] ANN: SciPy 0.8.0 beta 1

[josef.pktd at gmail.com]

On Fri, Jun 11, 2010 at 7:54 PM, Derek Homeier
<derek at astro.physik.uni-goettingen.de> wrote:
> Hi Josef,
>
>>> FAIL: test_stats.test_kstest
>>> ----------------------------------------------------------------------
>>> Traceback (most recent call last):
>>>  File "/sw/lib/python2.6/site-packages/nose/case.py", line 186, in runTest
>>>    self.test(*self.arg)
>>>  File "/sw/lib/python2.6/site-packages/scipy/stats/tests/test_stats.py", line 1078, in test_kstest
>>>    np.array((0.0072115233216310994, 0.98531158590396228)), 14)
>>>  File "/sw/lib/python2.6/site-packages/numpy/testing/utils.py", line 441, in assert_almost_equal
>>>    return assert_array_almost_equal(actual, desired, decimal, err_msg)
>>>  File "/sw/lib/python2.6/site-packages/numpy/testing/utils.py", line 765, in assert_array_almost_equal
>>>    header='Arrays are not almost equal')
>>>  File "/sw/lib/python2.6/site-packages/numpy/testing/utils.py", line 609, in assert_array_compare
>>>    raise AssertionError(msg)
>>> AssertionError:
>>> Arrays are not almost equal
>>>
>>> (mismatch 100.0%)
>>>  x: array([ 0.007,  0.985])
>>>  y: array([ 0.007,  0.985])
>>
>> maybe the precision (decimal 14) is too high for this test across platforms
>>
>> Could you check how large the difference is ?
>>
>> np.random.seed(987654321)
>> x = stats.norm.rvs(loc=0.2, size=100)
>> np.array(stats.kstest(x,'norm', alternative = 'greater')) -
>>                np.array((0.0072115233216310994, 0.98531158590396228))
>>
>> (my line numbers differ, but this should be the right test given your numbers)
>
>
> yes, just a decimal or two too high, if I got the numbers right:
> # OS X 10.5 i386 / 10.6 x86_64:
> array([  8.67361738e-18,   1.66533454e-15])
>
> # OS X 10.5 ppc:
> array([  2.05955045e-13,  -7.16759985e-13])

interesting that there are differences in the calculations, but for
the test we can just reduce the precision to decimal=12 to avoid the
test failure.

Thanks,
Josef

>
> Cheers,
>                                                Derek
>
> _______________________________________________
> SciPy-Dev mailing list
> SciPy-Dev at scipy.org
> http://mail.scipy.org/mailman/listinfo/scipy-dev
>