FARM

Foreground All-scale Radio Modeller.

Requirements

• python 3.8+ (required for most up-to-date numpy library)

Python modules

• ARatmospy (no version information available)
• astropy (developed with v5.0)
• h5py (developed with v3.6.0)
• jupyter (developed with v1.0.0)
• matplotlib (developed with v3.5.0)
• numpy (developed with v1.22.1)
• oskarpy (no version information available)
• pandas (developed with v1.4.1)
• powerbox (developed with v0.6.1)
• py21cmfast (no version information available)
• pygdsm (no version information available)
• reproject (developed with v0.8)
• scipy (developed with v1.7.3)
• toml (developed with v0.10.2)
• tools21cm (developed with custom fork)

System

• casa (developed with v6.4.0.16)
• gsl (developed with v2.7)
• miriad (developed with v20220120)
• oskar (developed with v2.8.2)

Installation help

For the python modules detailed above (and their dependencies (e.g. gsl), installation within a virtual environment is a possible method. Instructions for virtual environment creation and installing those dependencies listed above is shown below (requires a conda insallation).

NOTE: Installation of py21cmFAST on MacOSX leads to issues during compilation with the error, clang: error: unsupported option '-fopenmp' being thrown when installing via pip install . To properly resolve this and install py21cmFAST is detailed here.

conda create -n farm python>3.8
conda activate farm

# General package requirements
conda install pip git astropy gsl h5py jupyter matplotlib numpy pandas scipy toml reproject

# Installation of reproject can fail, in which case:
conda install -c conda-forge reproject

# py21cmFAST
# OpenMP enabled libraries of fftw are available through conda-forge, using "conda install fftw" does not include openmp threads
conda install -c conda-forge fftw  # Only needed for MacOS
conda install -c conda-forge/label/cf201901 gcc  # Only needed for MacOS
conda install -c conda-forge 21cmFAST

# powerbox
conda install -c conda-forge pyfftw
pip install git+git://github.com/steven-murray/powerbox.git

# Global diffuse sky model
pip install git+https://github.com/telegraphic/pygdsm

# OSKAR
pip install 'git+https://github.com/OxfordSKA/OSKAR.git@master#egg=oskarpy&subdirectory=python'

# Our fork of the ARatmospy repository for simulating the TEC screen
pip install git+https://github.com/SimonP2207/ARatmospy

# Our fork of tools21cm
pip install git+https://github.com/SimonP2207/tools21cm.git

A note on miriad and casa

Both miriad and casa limit the possible number of antennae in an array.

Basic Use

As a CLI application

python3 /path/to/farm/main.py /path/to/config.toml

As an imported library

from pathlib import Path
import astropy.units as u
from astropy.coordinates import SkyCoord
import farm

dims  = 512, 512  # n_pixels in x/R.A., n_pixels in y/declination
field_of_view = 8.  # deg
cell_size = field_of_view / dims[0]  # deg

coord = SkyCoord("01:02:03.4", "05:06:07.89",
                 frame='fk5', unit=(u.hourangle, u.degree))

# Use MHD model for Galactic small-scale structure SkyComponent. This is loaded 
# from data/Gsynch_SKAs.fits directly and thus its frequency information is 
# fixed.
gssm = farm.SkyComponent.load_from_fits(fitsfile=farm.DATA_FILES['MHD'],
                                        name='GSSM', cdelt=cell_size,
                                        coord0=coord)

# Use GSM2016 model (Zheng et al, 2016) for Galactic diffuse-scale structure 
# SkyComponent. Add the same frequencies as the GSSM sky component to enable 
# combination.
gdsm = farm.SkyComponent(dims, cell_size, coord,
                         tb_func=farm.tb_functions.gdsm2016_t_b)
gdsm.add_frequency(gssm.frequencies)

# Create SkyModel instance which will handle the combination of the various 
# SkyComponent instances. Add frequency information to it
skymodel = farm.SkyModel(dims, cell_size, coord, gssm.frequencies)

# Add individual SkyComponent instances to the SkyModel
skymodel += (gssm, gdsm)  # Equivalent to skymodel.add_component((gssm, gdsm))

# Write the sky model to a .fits file
skymodel.write_fits(Path("test_skymodel_Inu.fits"), unit='JY/SR')

For co-developers

Testing

For the unit-testing, Python's standard library package, unittest is used. Command line use is:

python -m unittest /path/to/farm/tests/test_*.py

Otherwise, good IDEs (such as PyCharm) have inbuilt testing which can be set up for unittest. This is the preferred approach due to the inhuilt debugger which can be run in conjunction with the testing package.