Dear all, I have tried to install scipy on ournew university-wide cluster, but I am unsure as to whether the install was completely successful. I checked out scipy from svn, and because that alone gave me "scipy_base not found", I needed to look through the mail on this list to find out that there exists a separate "scipy_core" module, which also has to be checked out and installed separately. Now both of these are installed. However, I only get 246 tests on level=10. I remember that this was much more in the past. Did I forget to install something? I also get one failure, which is listed below. I must say that I am not exactly running on a standard system... We run on a cluster of Opterons, use pathscale as compiler, and the BLAS and LAPACK are from the AMD Core Math Library instead of ATLAS. Once this system is working and validated, I will post detailed installation instructions for this setup. Best, Giovanni ====================================================================== FAIL: check_dot (scipy.lib.blas.test_blas.test_fblas1_simple) ---------------------------------------------------------------------- Traceback (most recent call last): File "/data/home/u0038151/lib/python2.4/site-packages/scipy/lib/blas/tests/test_blas.py", line 75, in check_dot assert_almost_equal(f([3j,-4,3-4j],[2,3,1]),-9+2j) File "/data/home/u0038151/lib/python2.4/site-packages/scipy_test/testing.py", line 662, in assert_almost_equal assert round(abs(desired - actual),decimal) == 0, msg AssertionError: Items are not equal: DESIRED: (-9+2j) ACTUAL: (-9+2.1839129395899578e-317j) ---------------------------------------------------------------------- Ran 246 tests in 1.658s
Giovanni Samaey wrote:
I have tried to install scipy on ournew university-wide cluster, but I am unsure as to whether the install was completely successful.
Coming back to this after an absence to do some other things: I believe that, at present, it is still not possible to build scipy from the SVN archive on a machine on which scipy was never before installed. If I do 'rm -rf scipy* weave gui_thread' in /usr/lib/python2.4, and then try: svn co http://svn.scipy.org/svn/scipy/trunk scipy cd scipy setup.py build it immediately complains that there is no scipy_distutils module. The best thing to do, still, seems to be to download Fernando Perez's snapshot of the last CVS archive at http://ipython.scipy.org/tmp/scipy_cvs_2005-07-29.tgz and build that. Even using that archive, I'm finding that "python setup.py bdist_rpm" in scipy/scipy_core doesn't work any more; I think it used to.
Stephen Walton wrote:
Giovanni Samaey wrote:
I have tried to install scipy on ournew university-wide cluster, but I am unsure as to whether the install was completely successful.
Coming back to this after an absence to do some other things:
and now I see Nils Wagner's previous post on this issue, which is to check out scipy_core and scipy both from SVN. My apologies.
and now I see Nils Wagner's previous post on this issue, which is to check out scipy_core and scipy both from SVN. My apologies.
Yes this works -- my question was if it is normal to have so many tests missing, and this one failure that I mentioned.
_______________________________________________ SciPy-user mailing list SciPy-user@scipy.net http://www.scipy.net/mailman/listinfo/scipy-user
participants (2)
-
Giovanni Samaey -
Stephen Walton