This is a collection of tools for using Planck spacecraft data and simulations with the LGMCA component separation algorithm.

It includes likelihoods to run MCMC chains with the Cobaya monte carlo sampling software.

First, clone this repository & enter the directory:

$ git clone git@github.com:jhod0/lgmca_planck_tools.git
$ cd lgmca_planck_tools

Then, install via pip:

$ pip install .

This will use the setup.py in this repository. If you wish to edit the code in this repository without having to reinstall it every time, add the --editable flag to pip install.

Runnable via python -m lgmca_planck_tools.invert. Requires the lgmca_inv program to be accessible on the PATH.

You will need to install extra data, such as LGMCA mixing weights, planck simulations, masks, and maps.

This package includes likelihoods to be run with the Cobaya cosmological MCMC sampler. They are tested to work with Cobaya version 2.0.5+.

Once this package is installed you can add a likelihood to any cobaya init file, e.g.:

likelihood:
  lgmca_planck_tools.like.FFP8Like:
    data_vector_file: /path/to/data/vector.fits
    cov_file: /path/to/data/covariance.txt
    do_rayleigh: false
    lmin: 70
    lmax: 2000
    dl: 30

This will load a CMB spectrum (D_\ell = \ell (\ell + 1) C_\ell / (2 \pi), in units of \mu K^2, stored via healpy.write_cl()), and a D_\ell covariance, and run an MCMC chain sampling cosmological parameters to fit to the spectrum.

It will bin the input vector from \ell = 30 to \ell = 2000 in bins of 30, and not attempt to account for Rayleigh scattering.

TODO: other likelihoods