Scipy Superpack Errors and Failures

[MDenno]

Hello:

I have been trying to get numpy/scipy installed on my Mac (OSX 10.6.4) for a few weeks now and just can't seem to make it all work.  I have tried 2 general approaches, building from source and using Chris Fonnesbeck's Scipy Superpack, both of which have been giving me problems.  After much installing and uninstalling of various programs/libraries I just did a fresh install of OSX and gave the Superpack  (10-22-2010 version) another try, but I am still having problems.  I don't think it is working correctly based on running the tests.

First, some information about my setup:

$ gcc --version
i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5659)

$ gfortran --version
GNU Fortran (GCC) 4.2.1 (Apple Inc. build 5659) + GF 4.2.4

$ python
Python 2.6.1 (r261:67515, Feb 11 2010, 00:51:29)
[GCC 4.2.1 (Apple Inc. build 5646)]

FFTW 3.2.2 built as follows:
        ./configure CC="gcc -arch x86_64" CXX="g++ -arch x86_64" CPP="gcc -E" CXXCPP="g++ -E"
        make
        sudo make install
       
XCode installed from xcode322_2148_developerdvd.dmg
       
Are there any other dependencies? UMFPACK?  ARPACK?

After installing and running the tests is it normal to have lots of warnings and some errors and failures for scipy?  I ran the following:

>>> import scipy
>>> scipy.test()
Running unit tests for scipy
NumPy version 2.0.0.dev
NumPy is installed in /Library/Python/2.6/site-packages/numpy-2.0.0.dev_20101022-py2.6-macosx-10.6-universal.egg/numpy
SciPy version 0.9.0.dev
SciPy is installed in /Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy
Python version 2.6.1 (r261:67515, Feb 11 2010, 00:51:29) [GCC 4.2.1 (Apple Inc. build 5646)]
nose version 0.11.0

[lots of warnings omitted - I would be happy to provide them if they are needed, but they are quite long]

======================================================================
ERROR: test_complex_nonsymmetric_modes (test_arpack.TestEigenComplexNonSymmetric)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/tests/test_arpack.py", line 269, in test_complex_nonsymmetric_modes
    self.eval_evec(m,typ,k,which)
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/tests/test_arpack.py", line 250, in eval_evec
    eval,evec=eigen(a,k,which=which)
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/arpack.py", line 397, in eigen
    params.iterate()
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/arpack.py", line 222, in iterate
    raise RuntimeError("Error info=%d in arpack" % self.info)
RuntimeError: Error info=-8 in arpack

======================================================================
ERROR: test_nonsymmetric_modes (test_arpack.TestEigenNonSymmetric)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/tests/test_arpack.py", line 206, in test_nonsymmetric_modes
    self.eval_evec(m,typ,k,which)
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/tests/test_arpack.py", line 188, in eval_evec
    eval,evec=eigen(a,k,which=which,**kwds)
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/arpack.py", line 397, in eigen
    params.iterate()
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/arpack.py", line 222, in iterate
    raise RuntimeError("Error info=%d in arpack" % self.info)
RuntimeError: Error info=-8 in arpack

======================================================================
ERROR: test_starting_vector (test_arpack.TestEigenNonSymmetric)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/tests/test_arpack.py", line 216, in test_starting_vector
    self.eval_evec(self.symmetric[0],typ,k,which='LM',v0=v0)
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/tests/test_arpack.py", line 188, in eval_evec
    eval,evec=eigen(a,k,which=which,**kwds)
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/arpack.py", line 397, in eigen
    params.iterate()
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/arpack.py", line 222, in iterate
    raise RuntimeError("Error info=%d in arpack" % self.info)
RuntimeError: Error info=-8 in arpack

======================================================================
FAIL: gaussian filter 3
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Library/Python/2.6/site-packages/nose-0.11.0-py2.6.egg/nose/case.py", line 183, in runTest
    self.test(*self.arg)
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/ndimage/tests/test_ndimage.py", line 468, in test_gauss03
    assert_almost_equal(output.sum(), input.sum())
  File "/Library/Python/2.6/site-packages/numpy-2.0.0.dev_20101022-py2.6-macosx-10.6-universal.egg/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

======================================================================
FAIL: line-search Newton conjugate gradient optimization routine
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/optimize/tests/test_optimize.py", line 177, in test_ncg
    assert_(self.gradcalls == 18, self.gradcalls) # 0.8.0
  File "/Library/Python/2.6/site-packages/numpy-2.0.0.dev_20101022-py2.6-macosx-10.6-universal.egg/numpy/testing/utils.py", line 34, in assert_
    raise AssertionError(msg)
AssertionError: 16

======================================================================
FAIL: test_complex_symmetric_modes (test_arpack.TestEigenComplexSymmetric)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/tests/test_arpack.py", line 159, in test_complex_symmetric_modes
    self.eval_evec(self.symmetric[0],typ,k,which)
  File "/Library/Python/2.6/site-packages/scipy-0.9.0.dev_20101022-py2.6-macosx-10.6-universal.egg/scipy/sparse/linalg/eigen/arpack/tests/test_arpack.py", line 148, in eval_evec
    assert_array_almost_equal(eval,exact_eval,decimal=_ndigits[typ])
  File "/Library/Python/2.6/site-packages/numpy-2.0.0.dev_20101022-py2.6-macosx-10.6-universal.egg/numpy/testing/utils.py", line 774, in assert_array_almost_equal
    header='Arrays are not almost equal')
  File "/Library/Python/2.6/site-packages/numpy-2.0.0.dev_20101022-py2.6-macosx-10.6-universal.egg/numpy/testing/utils.py", line 618, in assert_array_compare
    raise AssertionError(msg)
AssertionError:
Arrays are not almost equal

(mismatch 100.0%)
 x: array([ 1.07188725 +6.23436023e-08j,  4.91291142 -3.25412906e-08j], dtype=complex64)
 y: array([ 5.+0.j,  6.+0.j], dtype=complex64)

----------------------------------------------------------------------
Ran 4771 tests in 78.922s

FAILED (KNOWNFAIL=13, SKIP=41, errors=3, failures=3)
<nose.result.TextTestResult run=4771 errors=3 failures=3>

Thank you in advance for your help.

Matt

[Tommy Grav]

On Oct 28, 2010, at 4:13 PM, MDenno wrote:

>
> Hello:
>
> I have been trying to get numpy/scipy installed on my Mac (OSX 10.6.4) for a
> few weeks now and just can't seem to make it all work.  I have tried 2
> general approaches, building from source and using Chris Fonnesbeck's Scipy
> Superpack, both of which have been giving me problems.  After much
> installing and uninstalling of various programs/libraries I just did a fresh
> install of OSX and gave the Superpack  (10-22-2010 version) another try, but
> I am still having problems.  I don't think it is working correctly based on
> running the tests.
>
> First, some information about my setup:
>
> $ gcc --version
> i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5659)
>
> $ gfortran --version
> GNU Fortran (GCC) 4.2.1 (Apple Inc. build 5659) + GF 4.2.4
>
> $ python
> Python 2.6.1 (r261:67515, Feb 11 2010, 00:51:29)
> [GCC 4.2.1 (Apple Inc. build 5646)]
>
> FFTW 3.2.2 built as follows:
> ./configure CC="gcc -arch x86_64" CXX="g++ -arch x86_64" CPP="gcc -E"
> CXXCPP="g++ -E"
> make
> sudo make install

As far as I can tell the Superpack is 32bit only installation. Since you have installed
FFTW as a 64bit library, I assume that python is also a 64bit version and then you
are going to have to install numpy/scipy from scratch (something that is not a
very easy thing to get right on the mac in 64bit).

Cheers
  Tommy

_______________________________________________
SciPy-User mailing list
SciPy-User@...
http://mail.scipy.org/mailman/listinfo/scipy-user

[David Cournapeau]

On Fri, Oct 29, 2010 at 5:17 AM, Tommy Grav <tgrav@...> wrote:

> As far as I can tell the Superpack is 32bit only installation. Since you have installed
> FFTW as a 64bit library, I assume that python is also a 64bit version and then you
> are going to have to install numpy/scipy from scratch (something that is not a
> very easy thing to get right on the mac in 64bit).

FFTW is not used by scipy since 0.7.0, also,

David
_______________________________________________
SciPy-User mailing list
SciPy-User@...
http://mail.scipy.org/mailman/listinfo/scipy-user

[MDenno]

Tommy Grav wrote:

As far as I can tell the Superpack is 32bit only installation. Since you have installed
FFTW as a 64bit library, I assume that python is also a 64bit version and then you
are going to have to install numpy/scipy from scratch (something that is not a
very easy thing to get right on the mac in 64bit).

Cheers
  Tommy

Tommy,

Thanks for your reply.  I am not sure how to tell if the Superpack is build for 64 bits for sure but the website says it is.  

http://stronginference.com/scipy-superpack/

I also tried building FFTW with 32 and 64 bit flags and it didn't help.  However, I guess based on Dave's reply below FFTW is not used anymore.

Maybe I will try from source again or macports or something.

Matt

[MDenno]

> FFTW is not used by scipy since 0.7.0, also,

David,

Thanks for the clarification.  I think I will wait and see if anyone else has any suggestions, etc regarding the superpack.  If not maybe I will try from source again.  I will start a separate thread for that if/when I have problems.

Thanks again,

Matt

[Tommy Grav]

Also I think you can do

file python

(or other programs and libraries) to find out wether it is 32bit, 64bit
or both

[balder: tgrav ~]> file /usr/local/bin/python
/usr/local/bin/python: Mach-O universal binary with 2 architectures
/usr/local/bin/python (for architecture i386): Mach-O executable i386
/usr/local/bin/python (for architecture x86_64): Mach-O 64-bit executable x86_64

[balder: tgrav ~/Downloads/matplotlib-1.0.0/build/lib.macosx-10.6-intel-2.7/matplotlib]> file _delaunay.so
_delaunay.so: Mach-O universal binary with 2 architectures
_delaunay.so (for architecture i386): Mach-O bundle i386
_delaunay.so (for architecture x86_64): Mach-O 64-bit bundle x86_64

Cheers
 Tommy


On Oct 28, 2010, at 9:08 PM, MDenno wrote:

>
>
>
>> FFTW is not used by scipy since 0.7.0, also,
>
> David,
>
> Thanks for the clarification.  I think I will wait and see if anyone else
> has any suggestions, etc regarding the superpack.  If not maybe I will try
> from source again.  I will start a separate thread for that if/when I have
> problems.
>
> Thanks again,
>
> Matt
> --
> View this message in context: http://old.nabble.com/Scipy-Superpack-Errors-and-Failures-tp30079638p30081981.html
> Sent from the Scipy-User mailing list archive at Nabble.com.
>
> _______________________________________________
> SciPy-User mailing list
> SciPy-User@...
> http://mail.scipy.org/mailman/listinfo/scipy-user

_______________________________________________
SciPy-User mailing list
SciPy-User@...
http://mail.scipy.org/mailman/listinfo/scipy-user

[MDenno]

> Also I think you can do

> file python

Tommy,

Thanks for the tip.  I have checked a number of the libraries in the matplotlib, numpy and scipy directories and they all appear to be x86_64 at a minimum. Many of them are built for multiple arch.

Can I ask you what method you used to get scipy/numpy installed?

Matt