BrainSTAT installers

  • Windows, built for python2.4.
  • Windows, built for python2.3 -- assuming you have installed python2.3 packages of all that follow below.
  • Unix/Linux, source distribution.

Windows instructions

  • Install python, version 2.4. BrainSTAT works with python versions 2.3 and 2.4. A direct link to the installer can be found here.
  • Install numarray. A direct link to the installer can be found here.
  • Install GSL, the GNU Scientific Library. This is needed for some special functions. A direct link to the installer is available here. Expand this directory into C:\gsl. The path "C:\gsl\bin" will eventually need to be added to your path, this can be done using the (rough) sequence of mouseclicks "My Computer->Control Panel->System->Advanced->Environment Variables->Path." and adding ";C:\gsl\bin" to the list of paths. If the path "C:\Python24" is not listed, this will be have to be added as well.
  • Note: GSL support is disabled in the current installer because of a path/DLL problem in Windows -- this has very little effect on the capabilities, only the non-parametric module is affected.
  • Install the Windows installer of BrainSTAT (see above).
  • Unless you want to change the numerix entry of your matplotlibrc file, found by default in C:\Python24\share\matplotlib, you may also have to install Numeric. This duplicity will eventually disappear as NumPy takes over. A direct link to the installer can be found here. Using both numarray and Numeric obviously costs some overhead, and will be fixed in a future version.

Visualization for windows

  • Install matplotlib for 2d plots. A direct link the installer can be found here.
  • Install GTK for Win32. A direct link to the installer can be found here.
  • Install PyGTK for Win32. A direct link to the installer can be found here.
  • Install the python bindings of wxWindows, a direct link to the installer can be found here.
  • Install a Traits.UI modified version of traits, the installer is here.
  • Install VTK. A direct link to the installer can be found here.
  • Note: VTK/mayavi support for windows is not quite ready yet -- problems running mayavi as a script from python with the above release of VTK.
  • Install mayavi. A direct link to the installer can be found here.

Using IPython for Windows

The standard IDLE environment that ships with Python for windows is a bit awkward to work with. Fernando Perez's IPython (interactive python) has some very nice features for Unix and Windows. A direct link to a directory containing the Windows installer can be found here, though before installing IPython the following three packages should be installed.

  • The pywin32 extensions. A direct link to the installer can be found here.
  • The ctypes package. A direct link to the installer can be found here.
  • The readlines package. A direct link to the installer can be found here.

Linux instructions

  • You will need all the above packages mentioned in the Windows installation directory (most are available on debian/unstable). However, you will need the MINC source code, specifically the header file "minc_def.h" -- this is now included in the Include directory of the source code. The binary libraries and include available from debian do NOT include this file, which is needed for compiling the extension.

Debian packages

  • For simplicity, I will list the debian packages needed -- any *nix distribution with these packages should be enough to compile BrainSTAT. Packges:
    • Python: python2.4, python2.4-dev (nothing is strongly tied to 2.4, and python2.3 compiles fine - BrainSTAT does assume at least 2.3 as it uses Sets in a few places)
    • numarray: python2.4-numarray, python2.4-numarray-ext
    • GSL: gsl-bin, libgsl0, libgsl0-dev
    • MINC: minc-tools, libminc0, libminc0-dev
    • NetCDF: netcdf-bin, netcdfg-def, netcdfg3
    • VTK: libvtk4, libvtk4-dev, python2.4-vtk
    • GTK: python-gtk2, python-gtk2-dev
    • Ipython (not necessary, but useful): python2.4-ipython
    • Mayavi: mayavi

Matplotlib

  • There is no debian package for matplotlib. Download the latest source from here.
  • Untar/gz the source distribution and run the usual 
    python setup.py install

Build/install instructions

  • Untar the source distribution (see above), and modify the setup.py file to reflect your directory structure, i.e. you will need to make sure that the minc, gsl and netcdf libraries and include files are accessible from python. Then, run the usual 
    python setup.py install