Installation¶

Dependencies¶

In addition to an MPI implementation (e.g. MPICH or Open MPI), the pyCTQW.MPI module depends on the following components:

• Python >= 2.7
• NumPy >= 1.6.0
• PETSc >= 3.4.2
• SLEPc >= 3.4.1
• petsc4py 3.4 or petsc4py-dev
• mpi4py (recommended, used for some plotting)
• matplotlib (recommended, for node plotting and graph visualisation)
• SciPy (recommended, for some I/O operations)
• NetworkX (recommended, graph visualisation)

Installation using pip¶

After ensuring NumPy and petsc4py are installed (and all PETSc, SLEPc and MPI environment variables are properly set), pyctqw can be installed using pip:

$ pip install pyCTQW

$ pip install --allow-unverified pyCTQW pyCTQW==dev

Installing pyCTQW.MPI from source code¶

Alternatively, the source code can be downloaded and compiled manually:

1. Download the latest version of pyctqw, extract the pyCTQW archive, and cd into the extracted directory:

   $ wget https://github.com/josh146/pyctqw/archive/1.1.0.tar.gz -O pyctqw-1.1.0.tar.gz
   $ tar xzvf pyctqw-1.1.0.tar.gz
   $ cd pyctqw-1.1.0
   

2. Ensure that your PETSc and SLEPc environment variables are correctly set; for example,

   $ export PETSC_DIR=/path/to/petsc
   $ export PETSC_ARCH=linux-gnu
   $ export SLEPC_DIR=/path/to/slepc
   

   If you are unsure what your PETSc or SLEPc variables should be, please refer to their documentation.

   Important

   If you plan to install pyCTQW.MPI using root to a system directory, the PETSc and SLEPc environment variables must be available to the root user.

3. Compile the Python module pyCTQW.MPI by running

   $ python setup.py build
   

4. System-wide install:

   $ sudo -E python setup.py install
   

   where the command -E ensures that the environment variables set in step 3 are passed to the root.

   Note

   If you do not have root access, or the above command does not appear to work, you can install the package locally by running

   $ python setup.py install --user
   

   Now, have a go running some of the Examples!

Optional: build documentation¶

If Sphinx is installed, the documentation can be compiled by running

$ pip install -r docs/requirements.txt
$ make docs

Known Issues¶

• Non-mpi fallback modes not present yet