We describe the necessary steps required to succesfully install pysmc. The most important prior step is to satisfy the dependencies of the code.
There are two different categories of packages on which we rely. The Required ones have to be there no matter what. The Optional ones can be skipped but without them you will loose some of the functionality of pysmc or/and you will not be able to run all the examples in the tutorial.
The following packages are required:
- Numpy for linear algebra.
- SciPy for some root finding methods.
- PyMC for general probabilistic model definition and MCMC sampling. We suggest that you install the latest PyMC version from GitHub .
The following packages are optional but highly recommended:
- MPI4PY to enable the parallel capabilities of pysmc.
- matplotlib for plotting.
- FiPy in order to run the diffusion_example.
As soon as you are done installing the packages above, you can fetch pysmc from GitHub by:
git clone https://github.com/ebilionis/pysmc.git
Then, all you have to do is enter the pysmc directory that was created and run:
python setup.py install
If you want to put the code in an non default location, simply do:
python setup.py install --prefix=/path/to/your/directory
If you do the latter, make sure you update your PYTHONPATH variable:
export PYTHONPATH=/path/to/your/directory/lib/python2.7/site-packages:$PYTHONPATH