PAL2 (PTA Algorithm Library)

PAL2 is a Bayesian inference package for pulsar timing data. PAL2 is a re-write of the original PAL in which we use a very similar signal dictionary formalism introduced in [piccard] (https://github.com/vhaasteren/piccard). This code is meant to perform a large number of analyses on pulsar timing data including:

Noise characterization
Detection and characterization of GWB (isotropic and anisotropic) and continuous GWs
Non-linear pulsar timing
Dispersion measure variation modeling

Dependencies

The code has a large number of Python and non-Python dependencies. The python dependencies should should be installed automatically when following the directions below. You will have to manually install the following non-Python packages (preferably with a package manager):

[tempo2] (https://bitbucket.org/psrsoft/tempo2.git)
hdf5
healpix
openmpi

PAL2 also supports MultiNest but it must be installed separately.

[MultiNest] (http://ccpforge.cse.rl.ac.uk/gf/project/multinest/) (version 2.18 or later])
[PyMultiNest] (https://github.com/JohannesBuchner/PyMultiNest)

Installation

You will need to have tempo2 and a working C, C++ and Fortran compiler.

Anaconda Install

The easiest way for installation is to use Anaconda or Miniconda. Anaconda comes with many of the packages that we will need whereas Miniconda only comes with a small base Python installation. Once you have Anaconda (or Miniconda) installed you should get a academic license and install the MKL optimizations for fast linear algebra. With either installation it is best to set up an python environment with all of the needed dependencies. This can easily be done in Anaconda with

conda env create -f environment.yml
source activate pal2_conda

Be sure to follow the instructions and activate the Anaconda environment. This will create an Anaconda environment with nearly all of the required dependencies installed except libstempo. The libstempo package can be installed with

pip install git+https://github.com/vallis/libstempo@master

The installation will look for a TEMPO2 installation. To use a specific install you can use the --install-option="--with-tempo2=..." option. To finalize the installation do

python setup.py install

Standard Install

You can also install PAL2 with a python virutal environment. However for some clusters this may not be appropriate and you can follow the instructions below but append a --user flag on all of the pip and setup.py commands. Once you have the virualenv activated first do:

pip install numpy
pip install cython
pip install -r requirements.txt

This will install most of the dependencies except libstempo. The libstempo package can be installed with

pip install libstempo --install-option="--with-tempo2=$TEMPO2"

if you have your TEMPO2 environment variable set correctly. To finalize the installation do

python setup.py install

Known Issues

There are a few known issues with the PAL2 code that are being addressed:

During installation pip may have problems with the healpy package on Mac OSX. To overcome any potential problems you may need to include export CC=/path/to/gcc and export CXX=/path/to/g++.
the mark6 likelihood function is currently not compatible with cross correlations in the GWB. You will need to use the --noCorrelations flag with PAL2_run.py or use the incCorrelations=False flag in the likelihood if using your own interface.
You may have problems installing basemap with pip. You will first need to install geo with your package manager. Then you can install basemap via pip install basemap --allow-external basemap --allow-unverified basemap.

Example Data

We have included two datasets for testing purposes.

Open dataset 1 from the IPTA MDC. Good for testing GWB detection and characterization.
The 5-year data release from NANOGrav. Good for testing realistic noise models.

Manual

Example ipython notebook can be found in here.

Contact

[Justin Ellis] (mailto:justin.ellis18@gmail.com)

paulthebaker/PAL2