Difference between revisions of "Python Install libraries"

Line 5: Line 5:
 
This installation path should work for all versions of Mac OS X, because every library involved and the python interpreter is self-compiled. Make sure to use the system compiler to be compatible to system libraries.
 
This installation path should work for all versions of Mac OS X, because every library involved and the python interpreter is self-compiled. Make sure to use the system compiler to be compatible to system libraries.
  
 
+
# Install [http://mxcl.github.com/homebrew/ homebrew] using the default settings.
# Install [http://mxcl.github.com/homebrew/|homebrew] using the default settings.
 
 
#: <code>ruby -e "$(curl -fsSkL raw.github.com/mxcl/homebrew/go)"</code>
 
#: <code>ruby -e "$(curl -fsSkL raw.github.com/mxcl/homebrew/go)"</code>
 
# Install necessary libraries and python.
 
# Install necessary libraries and python.
Line 20: Line 19:
 
# Install the python-imaging library, which is required for Basemap.
 
# Install the python-imaging library, which is required for Basemap.
 
#: <code>pip install PIL</code>
 
#: <code>pip install PIL</code>
# Basemap itself is unfortunately not available via pip, so you have to compile and install yourself. Download and extract the [http://sourceforge.net/projects/matplotlib/files/matplotlib-toolkits/|source code from the sourceforge repository]. Then in the extracted directory run
+
# Basemap itself is unfortunately not available via pip, so you have to compile and install yourself. Download and extract the [http://sourceforge.net/projects/matplotlib/files/matplotlib-toolkits/ source code from the sourceforge repository]. Then in the extracted directory run
 
#: <code>python setup.py build</code>
 
#: <code>python setup.py build</code>
 
#: <code>python setup.py install</code>
 
#: <code>python setup.py install</code>
 +
 +
== Testing your installation ==
 +
 +
Open an python or ipython shell and type
 +
# to import and test numpy:
 +
#:<code>import numpy as np</code>
 +
#:<code>np.test()</code>
 +
# to import and test scipy:
 +
#:<code>import scipy as sp</code>
 +
#:<code>sp.test()</code>
 +
# to import and test matplotlib
 +
#:<code>import matplotlib as mpl</code>
 +
#:<code>mpl.test()</code>
 +
# to import Basemap
 +
#:<code>from mpl_toolkits.basemap import Basemap</code>
 +
There should be no errors during the imports and during the tests.
  
 
== De-Installation ==
 
== De-Installation ==

Revision as of 22:14, 24 January 2013

Installing numpy, scipy, matplotlib and Basemap on Mac OS X can be challenging.

Installation using homebrew

This installation path should work for all versions of Mac OS X, because every library involved and the python interpreter is self-compiled. Make sure to use the system compiler to be compatible to system libraries.

  1. Install homebrew using the default settings.
    ruby -e "$(curl -fsSkL raw.github.com/mxcl/homebrew/go)"
  2. Install necessary libraries and python.
    brew install gfortran python geos
  3. Install numpy via the python package manager pip which homebrew installed automatically with python.
    pip install numpy
  4. Install scipy
    pip install scipy
  5. Install matplotlib
    pip install matplotlib
  6. Install ipython (not necessary, but recommended replacement for the standard python shell)
    pip install ipython
  7. Install the python-imaging library, which is required for Basemap.
    pip install PIL
  8. Basemap itself is unfortunately not available via pip, so you have to compile and install yourself. Download and extract the source code from the sourceforge repository. Then in the extracted directory run
    python setup.py build
    python setup.py install

Testing your installation

Open an python or ipython shell and type

  1. to import and test numpy:
    import numpy as np
    np.test()
  2. to import and test scipy:
    import scipy as sp
    sp.test()
  3. to import and test matplotlib
    import matplotlib as mpl
    mpl.test()
  4. to import Basemap
    from mpl_toolkits.basemap import Basemap

There should be no errors during the imports and during the tests.

De-Installation

Follow the above commands in the opposite order and replace install by uninstall.

Common pitfalls

  • Make sure you have no other version of gcc installed. Check the output of
    which gcc
    to make sure to use the system compiler.
  • Do not mix pre-compiled and self-compiled libraries.