Calculate the log-likelihood for an ACT foreground model.
This is a quick (and somewhat dirty) recode of most of the chi^2 functionality of the original DR4 Baseline Multi-Frequency Likelihood from ACTPol, available here.
You need numpy, scipy and python3 installed to run this. For additional functionality, you also need camb and/or sacc installed (see links below).
import loglike in your python script. You must create a Likelihood instance for access to the functionality, e.g. like = loglike.Likelihood().
You can load in a plaintext dataset by calling like.load_plaintext(). It has a number of parameters, referring to the three files (model spectra, covariance matrix, window functions) and an optional path to where these files are loaded in. You need to give two arrays to provide the number of bins expected from the files and which cross-spectrum each bin set refers to (indexed using like.freqs). You can load in a dataset from a SACC file by calling like.load_sacc(). Note that loading from a plaintext assumes you set all parameters such as number of spectra and bins beforehand, whereas loading from a SACC file will make the Likelihood class overwrite values with what is found in the SACC file. You can load in a (plaintext) leakage dataset by calling like.load_leakage().
You can load in a LCDM model by calling either like.load_cells() (to load it from a plaintext file) or by calling like.load_cells_camb() to have a background cosmology generated via camb.
When all is prepared and done, you can calculate the log-likelihood for a foreground model by calling like.loglike(). You have to provide TT/TE/EE foreground models (you can disable individual models by setting like.use_ee = False for example, but note not every combination of TT/TE/EE is allowed).
This code is presented as-is with no warranty for its functionality. Please refer to the original FORTRAN code if you wonder about the underlying functionality.
This code came to be by stitching together a great variety of different codesets, and thus may not be entirely stable, optimal or clean. The original author may not update this code in the (nearby) future, and they also cannot guarantee bug-free functionality.
If you find any bugs and you desperately want to push a commit to patch it, feel free to contact the original author.
This code was made by Hidde Jense and is available for usage and modification free of charge. Please contact Hidde Jense if you wish to incorporate this code into your own project(s). You are free to use and modify this code for personal or academical use, so long as you refer to the original author when you do so.
Based on original code presented in Choi et al. 2020 and Aiola et al. 2020. Original code available in FORTRAN here. Please reference them if you use their code at a later stage.
SACC is created by Joe Zuntz and is available here.
CAMB is created by Antony Lewis et al., and is available here.