DiffPy-CMI installation from sources

Downloaded the most recent DiffPy-CMI tarball and follow the steps below.

1 Install system software

DiffPy-CMI requires the system software dependencies which can be installed from command line using a suitable package manager. Here are installation commands for several supported systems.

Ubuntu

For Ubuntu or other Debian-based Linux distributions use apt-get:

sudo apt-get install \
    libgsl0-dev libboost-all-dev python-dev \
    python-setuptools python-numpy python-scipy \
    python-matplotlib python-lxml ipython \
    build-essential scons git zsh

Fedora

Use yum for Fedora or RedHat Linux:

sudo yum install \
    gsl-devel boost-devel python-devel \
    python-setuptools numpy scipy  \
    python-matplotlib python-lxml \
    python-ipython-notebook \
    gcc-c++ scons git zsh

Mac OS X

For Mac OS X the system dependencies can be installed using the MacPorts software manager. A similar package system Homebrew works as well, but has been considerably less tested with DiffPy-CMI.

For best results with MacPorts follow these tips:

  • We recommend to upgrade to the latest version of OS X.

  • Install Xcode. If already present, we recommend to upgrade to the latest version.

  • Be patient, it may take several hours to install all the dependencies using MacPorts.

  • Installation command may fail on the first run, but usually works when repeated. See MacPorts FAQ for more help.

To install system dependencies with MacPorts, use:

sudo port install \
    python27 py27-setuptools py27-ipython py27-lxml \
    gsl boost py27-numpy py27-scipy py27-matplotlib scons git

Make sure the MacPorts versions of IPython and Python are active by running the following commands:

sudo port select --set ipython ipython27
sudo port select --set python python27

Important: When finished installing the Mac OS X dependencies, adjust the shell environment so that MacPorts Python is the first in the PATH. This can be accomplished by adding the following line to either of .profile or .zshenv file in your HOME directory

export PATH="/opt/local/bin:$PATH"

2 Install DiffPy-CMI

Unzip the DiffPy-CMI tarball into a directory of your choice. Execute the included install script and follow the prompts.

# replace VERSION to match the actual filename
tar xzf diffpy_cmi-VERSION.tar.gz
cd diffpy_cmi-VERSION
./install

The install process may take a while and produce a plentiful output, but should work if all required software is in place.

3 Test

Execute the included test script, which should report no warnings nor errors:

./runtests.sh

If there are failures they are most likely due to missing software or incorrect versions of Python or other libraries being used together. After addressing these issues it may be necessary to recompile the sources by running ./install --build.

Remarks

The expanded diffpy_cmi directory can be renamed or moved to a different location and the software should still work. The only requirement is to update the diffpy_cmi.pth symbolic link so it points to the new location. This can be done by running the ./install script again or by following the steps below.

The Python interpreter must have a symbolic link pointing to the diffpy_cmi.pth file in one of the directories where it looks for .pth files. This is normally set up by the install script. If that process somehow fails, the preferred pth directory for a single-user installation can be found using

python -c 'import site; print site.USER_SITE'

For a system-wide installation the standard pth locations are

python -c 'import site; print site.getsitepackages()'

Once the pth directory has been established, navigate to the base diffpy_cmi directory and create the symbolic link with

ln -si $PWD/diffpy_cmi.pth /path/to/the/pth/directory/

Note

It is essential to use a symbolic link. Making a copy of the diffpy_cmi.pth file will not work.

Note

The installation of DiffPy-CMI is entirely contained under the expanded diffpy_cmi directory. The software can be completely uninstalled by deleting that directory and removing the symbolic link.

Contacts

If you need help with installing this software, please check discussions or post your question to the diffpy-dev group.

List of software dependencies

  • gsl - GNU Scientific Library is collection of routines for numerical analysis.

  • boost - a set of useful C++ libraries.

  • python-dev - development tools for Python modules.

  • python-setuptools - enhancements to the Python distutils.

  • numpy - general-purpose array-processing for large multi-dimensional arrays in Python.

  • scipy - the fundamental library for scientific computing with Python.

  • matplotlib - a Python 2D plotting library.

  • python-lxml - a Python library for processing XML.

  • ipython - an enhanced interactive Python shell.

  • scons - a software build tool.

  • git - a version control system.

  • zsh - an interactive shell and powerful scripting language.