[ase-users] Failing Tests Related to Scientific_netcdf

[Eric Hermes]

Robert,

The first error is related to a bug in your installed version of Numpy, and the remaining are due to a bug in your installed version of ScientificPython (as you have identified). Specifically, according to the aptly-named "scientificpython_bug.py" test, ScientificPython version 2.8 or greater is required for NetCDF support.

I am not a Mac user, so I am not intimately familiar with MacPorts, but it would appear that when you installed NetCDF it did not build/install the static library version (netcdf.a). This appears to be a requested enhancement in the MacPorts bug tracker: https://trac.macports.org/ticket/51303. I expect your best bet is to manually install NetCDF and switch BUILD_SHARED_LIBS on in cmake if you wish to continue using the ScientificLinux package. Alternatively, you can use system Python and manually install ASE's dependencies following the guide on the ASE wiki here: https://wiki.fysik.dtu.dk/ase/install.html.

Eric

From: ase-users-bounces at listserv.fysik.dtu.dk [mailto:ase-users-bounces at listserv.fysik.dtu.dk] On Behalf Of Robert Wexler via ase-users
Sent: Monday, August 22, 2016 2:17 PM
To: ase-users at listserv.fysik.dtu.dk
Subject: [ase-users] Failing Tests Related to Scientific_netcdf

Dear ASE Users,

I tested my installation and received 4 errors.

My machine specific information is as follows:
platform        Darwin-15.5.0-x86_64-i386-64bit
python-2.7.10   /usr/bin/python
ase-3.11.0      /Users/<username>/ase/ase/
numpy-1.8.0rc1  /System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python/numpy/
scipy-0.13.0b1  /System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python/scipy/
test-dir        /var/folders/vb/858_vzmn18gc7hd524j0gmf80000gn/T/ase-test-fEK8kI

The 4 errors I received were:

======================================================================
ERROR: niggli.py
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/<username>/ase/ase/test/__init__.py", line 53, in testfile
    {'display': self.display})
  File "/Users/<username>/ase/ase/test/niggli.py", line 123, in <module>
    niggli_reduce(conf)
  File "/Users/<username>/ase/ase/build/tools.py", line 631, in niggli_reduce
    scpos = np.dot(atoms.get_scaled_positions(), np.linalg.inv(C).T)
  File "/Users/<username>/ase/ase/atoms.py", line 1459, in get_scaled_positions
    fractional = np.linalg.solve(self.cell.T, self.positions.T).T
  File "/System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python/numpy/linalg/linalg.py", line 380, in solve
    r = gufunc(a, b, signature=signature, extobj=extobj)
ValueError: cannot remove a zero-sized axis from an iterator

======================================================================
ERROR: scientificpython_bug.py
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/<username>/ase/ase/test/__init__.py", line 53, in testfile
    {'display': self.display})
  File "/Users/<username>/ase/ase/test/scientificpython_bug.py", line 10, in <module>
    import Scientific.IO.NetCDF as netcdf
  File "/Library/Python/2.7/site-packages/Scientific/IO/NetCDF.py", line 165, in <module>
    from Scientific_netcdf import *
ImportError: No module named Scientific_netcdf

======================================================================
ERROR: fio/oi.py
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/<username>/ase/ase/test/__init__.py", line 53, in testfile
    {'display': self.display})
  File "/Users/<username>/ase/ase/test/fio/oi.py", line 90, in <module>
    write(fname1, atoms, format=format)
  File "/Users/<username>/ase/ase/io/formats.py", line 259, in write
    io.write(filename, images, **kwargs)
  File "/Users/<username>/ase/ase/io/etsf.py", line 12, in write_etsf
    ETSFWriter(filename).write_atoms(atoms)
  File "/Users/<username>/ase/ase/io/etsf.py", line 35, in __init__
    from Scientific.IO.NetCDF import NetCDFFile
  File "/Library/Python/2.7/site-packages/Scientific/IO/NetCDF.py", line 165, in <module>
    from Scientific_netcdf import *
ImportError: No module named Scientific_netcdf

======================================================================
ERROR: jacapo/jacapo.py
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/<username>/ase/ase/test/__init__.py", line 53, in testfile
    {'display': self.display})
  File "/Users/<username>/ase/ase/test/jacapo/jacapo.py", line 5, in <module>
    from ase.calculators.jacapo import Jacapo
  File "/Users/<username>/ase/ase/calculators/jacapo/__init__.py", line 3, in <module>
    from ase.calculators.jacapo.jacapo import *
  File "/Users/<username>/ase/ase/calculators/jacapo/jacapo.py", line 21, in <module>
    from Scientific.IO.NetCDF import NetCDFFile as netCDF
  File "/Library/Python/2.7/site-packages/Scientific/IO/NetCDF.py", line 165, in <module>
    from Scientific_netcdf import *
ImportError: No module named Scientific_netcdf

I'm not sure what is causing the first error but the second error, i.e. "ImportError: No module named Scientific_netcdf", seems to be related to my installation of netCDF and ScientificPython2.8. I used MacPorts to install netcdf 4.4.1 and received the following output:

<computername>:~ <username>$ sudo port install netcdf
Password:
--->  Computing dependencies for netcdf
--->  Cleaning netcdf
--->  Scanning binaries for linking errors
--->  No broken files found.

Then I tried, however unsuccessfully, to install ScientificPython2.8 and received the following output:

<computername>:ScientificPython-2.8.1 <username>$ python setup.py build
netCDF not found, the netCDF module will not be built!
If netCDF is installed somewhere on this computer,
please set NETCDF_PREFIX to the path where
include/netcdf.h and lib/netcdf.a are located
and re-run the build procedure.
running build
running build_py
running build_ext
running build_scripts

I located netcdf.h in /opt/local/include/ but was unable to find netcdf.a. I then modified the variable NETCDF_PREFIX in setup.py with /opt/local but received the same output upon trying to build.

As a side comment, my main motivation for using ASE is to be able to perform LAMMPS total energy calculations as part of a Python script. So perhaps, these errors are insignificant and I can move on without worry. I would, however, like to get all functionalities of ASE running on my computer so that I can explore them later.

Does anybody have experience with the problem I am facing and can suggest ways to correct the errors that show up during testing?

Thank you all so much for you help. I am looking forward to your responses.

Best,
Robert Wexler

---------------------------------------------
Robert B. Wexler
PhD Candidate in Chemistry
University of Pennsylvania, 2018
rwexler at sas.upenn.edu<mailto:rwexler at sas.upenn.edu>
(215) 801-8741<tel:%28215%29%20801-8741>
---------------------------------------------
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20160822/a6db2e30/attachment-0001.html>