[SciPy-user] Test failure

[Arnd Baecker]

On Thu, 19 Jan 2006, Pawel Wielgus wrote:

> Dear SciPy Users,
>
> I've just installed SciPy for the first time in my life (with a lot of
> problems when using svn core and scipy,

have you tried a recent numpy/scipy svn, i.e.:
  svn co http://svn.scipy.org/svn/numpy/trunk numpy
  svn co http://svn.scipy.org/svn/scipy/trunk scipy
(see also http://new.scipy.org/Wiki for more information
- under-development)
Concerning the naming: scipy core etc.: have a look
at  http://numeric.scipy.org/
which explains this.
To summarize: it is really numpy+scipy what you want. ;-)

It would be great if you could post the
details of your problems.
I am sure that these can be solved!
(Actually, with the most recent numpy/scipy it should
work fine...)

> but smoothly and succesfuly when
> using Scipy_core-0.3.2.tar.gz and SciPy-0.3.2.tar.gz +FFTW +ATLAS
> binaries -DJBFTT [for DJBFFT I had some compilation problems, but
> honestly I did not dig it deeper]). Anyway, finally I run some tests.
> After t=scipy.test() I have:
> Ran 972 tests in 2.042s
> OK
>
> But after scipy.test(level=10,verbosity=2) I receive an error:
> ======================================================================
> FAIL: check_normal (scipy.stats.morestats.test_morestats.test_anderson)
> ----------------------------------------------------------------------

[...]

> Is there any particular reason for that? What does it mean?

This is a statistical test, so it may fail from time to time.
If you rerun `scipy.test(10)` usually this test will pass.
It has been discussed on scipy-dev to run statistical
tests a couple of times and only if it fails all
the time to raise an error (not yet implemented).

Best, Arnd