MEANS: python package for Moment Expansion Approximation, iNference and Simulation

We present a free, user-friendly tool implementing an efficient moment expansion approximation with parametric closures that integrates well with the IPython interactive environment. Our package enables the analysis of complex stochastic systems without any constraints on the number of species and moments studied and the type of rate laws in the system. In addition to the approximation method our package provides numerous tools to help non-expert users in stochastic analysis.

Documentation

Tutorial

A tutorial on getting started with MEANS can be found in the directory tutorial in this repository.

API Reference

API reference can be found on Read the Docs. Read the Docs also autogenerates a PDF file for every build, which can be downloaded from here.

Installation

Dependancy: Sundials

Prior to installation of this package ensure sundials libraries are installed on your system.

The exact steps to install them will depend on your system and configuration.

Generally, the easiest way to install these packages is via the system's package manager. For instance, on Mac OS X these libraries can be installed via HomeBrew:

brew install sundials

If you are using ubuntu, you can use apt-get to install the packages:

apt-get install libsundials-cvode1 libsundials-cvodes2 libsundials-ida2 libsundials-idas0 libsundials-kinsol1 libsundials-nvecserial0 libsundials-serial libsundials-serial-dev

Users of anaconda python distribution, could use the recipe provided by chemreac to install it:

conda install -c https://conda.anaconda.org/chemreac sundials

At this point, make sure to note the path where the package manager installs the libraries to. For instance, apt-get is likely to put the libraries to /usr/lib whereas anaconda is likely to put it in /home/username/anaconda/lib. Ensure this lib directory is in LD_LIBRARY_PATH variable.

export LD_LIBRARY_PATH="/your/sundials/home/lib:$LD_LIBRARY_PATH"

Also note the directory prefix without the lib as you might need to specify this path as --sundials-homewhen istallingassimulo` (see the troubleshooting section below).

On other linux systems one might need to compile the sundials libraries from source. Visit sundials homepage for instructions on how to obtain the latest release of the library.

For the newest release of sundials, follow instructions in the pdf distributed with the sundials source. Ensure that -DCMAKE_C_FLAGS=-fPIC is set when installing.

If you are installing an older version of the library, then make sure that

--enable-shared, --enable-static, --with-gnu-ld and --with-cflags=-fPIC flags are set.

Similarly, ensure that the --prefix is set to /usr/local, or alternatively see troubleshooting section below.

MEANS: Stable Release

Once sundialslibraries are installed correctly, MEANS can be installed through pip, just like any other python package:

pip install means

Note that due to the way pip handles dependencies, you might need to install cython, numpy prior to means:

pip install cython pip install numpy

Verifying Installation

Since a lot of things can go wrong during the installation, it is important to verify the installation before proceeding with the package.

Particularly, please run the tests and ensure they pass. To run the tests, install nose package first:

pip install nose

Then run the following command: nosetests means.tests If no tests failed, proceed straight to the tutorial. Otherwise, consult the troubleshooting section.

MEANS: Development Version

The development version of this package can be installed by downloading the code, navigating to src directory and typing pip install -e .

You may similarly need to install numpy and cython for this command to work, as in the section above.

Troubleshooting

Symptom: Tests fail with No module named sundials

This error indicates that assimulo installation failed to find the sundials in the system path. Please ensure that you set appropriate flags in sundials installation, as described in previous section and attempt to reinstall assimulo as follows:

pip uninstall assimulo pip install assimulo

If you installed the sundials libraries to prefix other than /usr/local (for instance, ArchLinux installs the library to /usr by default), install assimulo as follows:

pip install assimulo --install-option="--sundials-home=/your/sundials/installation/prefix" Replacing the /your/sundials/installation/prefix with the appropriate prefix.

Once this is done try running tests again.

Symptom: Tests fail with ImportError: /usr/lib/libsundials_idas.so.0: undefined symbol: dscal_

The symptom above indicates incorrect linkage between libsundials_idas.so and BLAS/LAPACK libraries. A workaround for this issue is to add libblas.so liblapack.so to LD_PRELOAD environment variable. To do this, make sure to set this environment variable first:

export LD_PRELOAD='libblas.so liblapack.so'

And then run everything else as usual.