PRISim (Precision Radio Interferometer Simulator)

A modular radio interferometer array simulator, including the radio sky and instrumental effects.

Installation

Note that currently this package only supports Python 2.6+, and not Python 3.

Non-Python Dependencies

The only non-python dependencies are openmpi and xterm. These are not required, but do provide extra functionality. These can usually be installed via a distro package manager (eg. for Arch Linux, the package names are exactly openmpi and xterm).

Using Anaconda

If using the Anaconda python distribution, many of the packages may be installed using conda. While these dependencies will be installed automatically with the installation procedure below (i.e. with pip), usually conda users will want to install these with conda explicitly before installing prisim.

The conda-appropriate packages can be installed with

conda install mpi4py progressbar psutil pyyaml h5py astropy matplotlib numpy scipy scikit-image

NOTE: at this time, you must install scikit-image via conda, or else it will try to install packages that are incompatible with python 2. Full python 3 support is coming soon.

Finally, either install PRISim directly:

pip install git+https://github.com/nithyanandan/PRISim

or clone it into a directory and from inside that directory issue the command:

pip install .

Getting Package Data

First try using the following (from anywhere on your computer, but inside your env):

setup_prisim_data.py

If this does not work, try a manual download, as follows:

Find the data/ directory under PRISim installation folder which is usually in

/path/to/Anaconda/envs/YOURENV/lib/python-2.7/site-packages/prisim/

Download the contents of PRISim Data from either Google Drive

or the zipped version from Zenodo (.zip or tar.gz) or Google Drive (.tar.gz)

Extract the contents of the zipped file and place it under

/path/to/Anaconda/envs/YOURENV/lib/python-2.7/site-packages/prisim/data/

Testing MPI for PRISim

On terminal, run:

mpirun -n 2 test_mpi4py_for_prisim.py

and you must see a message that the test was successful. Otherwise you will have to ask your system administrator to install openmpi

It has also been noted that it is preferable to install mpi4py using

conda install mpi4py

rather than

pip install mpi4py

because the pip installation seems to get the paths to the MPI libraries mixed up.

Basic Usage

Run on terminal:

mpirun -n nproc run_prisim.py -i parameterfile.yaml

or

mpirun -n nproc xterm -e run_prisim.py -i parameterfile.yaml

where, nproc is the number of processors (say, 4), and use of option xterm -e opens an xterm window where you can view the progress of each of the processes.

Data Size

Data size is proportional to n_bl x nchan x n_acc

Citing and acknowledging use of PRISim:

If you use PRISim for your research, please acknowledge and cite it using the bibtex entry from https://doi.org/10.5281/zenodo.2548116