xgarrido/LAT_MFLike

Multifrequency Likelihood

LAT Multifrequency Likelihood

An external likelihood using cobaya.

Installing the code

The easiest way to install and to use mflike likelihood is via pip

pip install mflike

If you want to dig into the code, you'd better clone this repository to some location

git clone https://github.com/simonsobs/LAT_MFLike.git /where/to/clone

Then you can install the mflike likelihood and its dependencies via

pip install -e /where/to/clone

The -e option allow the developer to make changes within the mflike directory without having to reinstall at every changes. If you plan to just use the likelihood and do not develop it, you can remove the -e option.

Installing LAT likelihood data

Preliminary simulated data can be found at NERSC. You can download them by yourself but you can also use the cobaya-install binary and let it do the installation job. For instance, if you do the next command

cobaya-install /where/to/clone/examples/mflike_example.yaml -p /where/to/put/packages

data and code such as CAMB will be downloaded and installed within the /where/to/put/packages directory. For more details, you can have a look to cobaya documentation.

Running/testing the code

You can test the mflike likelihood by doing

cobaya-run /where/to/clone/examples/mflike_example.yaml -p /where/to/put/packages

which should run a MCMC sampler for the simulation n°44 (i.e. data_sacc_00044.fits in the mflike_example.yaml file) using the combination of TT, TE and EE spectra (i.e. polarizations: ['TT', 'TE', 'ET', 'EE']). The results will be stored in the chains/mcmc directory.

Simulation releases

• v0.8 release has spectra produced with high accuracy camb parameters up to ell = 9000, the foreground components are integrated in frequency using top-hat bandpasses, stored in the sacc files
• v0.7.1 is a bugfix release of v0.7 release where the mono-frequency and dirac bandpasses have been correctly set with the sacc files
• v0.7 release includes the ACT like foregrounds. Simulation parameters are also stored within sacc metadata and the associated dict file can be viewed here.
• v0.6 release make use of CMB map based simulations (see https://github.com/simonsobs/map_based_simulations/blob/master/201911_lensed_cmb/README.md). Only temperature foregrounds were considered.

Releases prior to v0.6 version were development simulations and should not be used for simulation analysis.