fff (Fast Functions For fMRI)

============================================================================
 OVERVIEW
============================================================================

Quickie. From the trunk directory, type:

 python setup.py install

Yet some conditions are required for this to work.  


============================================================================
 PERMISSIONS
============================================================================

fff depends on NumPy and SciPy, the scientific python packages, as
well as on the LAPACK library (written in fortran), although the fff
distribution uncludes a lite version of LAPACK that is built when
running setup.py if LAPACK is not found on your system.

If LAPACK is installed but not found, it is probably because it is in
a non-standard path. You may then create a site.cfg file in the root
fff directory (similarly to numpy) in order to tell the setup where
LAPACK is installed on your system. An example site.cfg is versioned
in this directory.


============================================================================
 PERMISSIONS
============================================================================

Running python setup.py install will result in an attempt to write in
the Python site-packages directory and will fail if you don't have the
appropriate permissions. You may install fff in an alternative
directory, using:

 python setup.py install --home=/my/directory

This will actually install fff under /my/directory/lib/python/. 

Alternatively, you can build fff but skip installation by just
running:

  python setup.py build 

but you will then need to adjust your PYTHONPATH environment variable
similarly to the example below (using bash):

 export PYTHONPATH=$PWD:/build/lib.linux-i686-2.5/fff/:$PYTHONPATH


============================================================================
 NUMPY / SCIPY 
============================================================================
 
To download and install both numpy and scipy, please visit:

	http://numpy.scipy.org

To test whether numpy is properly installed on your system, launch a
python terminal (for instance, ipython), and type:
 
  import numpy
  print numpy.__version__
  numpy.test()
 
The same procedure applies to scipy.

Note that only numpy is required for building fff. Some fff modules
actually work without scipy.


============================================================================
 LAPACK
============================================================================
 
fff optionnaly links with a complete LAPACK library, whose path can be
specified in the [lapack_opt] section of the configuration file if
needed (see site.cfg.example).
 
The standard LAPACK distribution is found at:

    http://www.netlib.org/lapack/
	
To build it, download the complete package (file: lapack.tgz) and
uncompress it. You should see a file called make.inc.example in the
directory that has just been created. Rename that file make.inc and
optionally edit it to modify some variables). Then type:

	   make blaslib
	   make lib
	   make (optionally for tests)

This should create at least two static libraries: blas$PLAT.a and
lapack$PLAT.a (where PLAT is set in make.in). Rename them libblas.a
and liblapack.a, respectively, and move them to an appropriate path
such as /usr/local/lib/ or $YOURHOME/lib.

For more details on building LAPACK, as well as using optimized LAPACK
implementations depending on your system (Mac OS X, Linux or 
Microsoft Windows), please visit:

       http://www.scipy.org/Installing_SciPy
  

============================================================================
 TESTING fff
============================================================================

You should now be able to launch python and type:

  import fff
  fff.test()

This will test only those modules that are loaded automatically when
initializing fff. To test other modules or packages, you need to
import them explicitely before running fff.test().

For a sub-package <package>, the command fff.<package>.test() is
available and results in testing that package only.