jackknify
is a Python-based package that jackknifes ALMA visibilities to create noise realizations from the observations.
Jackknifing is a simple but effective tool to characterize the underlying noise distribution of any type of data set. This tool specifically is implemented for interferometric data. jackknify
splits half the visibilities randomly in two subsets, then multiplies one half with -1 so that when the data is binned, any signal present in the data is averaged out. This creates observation-specific noise realization of the data, which can be used to for instance, sample the likelihood a false detection.
The full methodology can be found here and in an upcoming paper, which is still in preparation.
jackknify
itself can be installed through
pip install jackknify
or alternatively
python -m pip install git+https://github.com/Joshiwavm/jackknify
or from the source
git clone https://github.com/Joshiwavm/jackknify
cd jackknify
pip install -e .
jackknify
uses casatask
and casatools
to interface with CASA measurements. casatask
and casatools
requires casadata
to load. Sadly, this is a ~350 MB sized file making the installment a bit slow. Further, when performing line searches, we make use of the package interferopy
, which is a Python-based package for common tasks used in the observational radio/mm interferometry data analysis.
If you want to run jackknify
on a Mac with an Apple Silicon chip, run it in a Rosetta terminal. To open a Rosetta session in your terminal, run:
/usr/bin/arch -x86_64 /bin/zsh --login
Further, casadata
will download and store examples sets into the folder: ~/.casa/data. However, it might not have permission from the local machine to do so. If such an error comes up. Just run:
mkdir ~/.casa/data
To make the folder. That should solve most problems.
For your convenience, there are notebooks on how to run and use jackknify
for line inference. You can find them in the docs/notebooks folder. Also, check out the documentation here.