Cobaya likelihood for the DR6 foreground-marginalized data.
Author: Hidde T. Jense
See the ACT DR6 power spectra and likelihoods paper for a full description.
This is a cobaya likelihood for the foreground-marginalized DR6 data.
The easiest way to install the code is to first clone the repository somewhere with
git clone [email protected]:ACTCollaboration/DR6-ACT-lite.git .
You can install the package then as
pip install -e .
This will locally pip-install the dataset.
At this point, you can install the likelihood data with
cobaya-install act_dr6_cmbonly
Which will download the correct data file and install it at your cobaya packages path.
If you want to use the Planck dataset cut at the ACT multipole, you need to install the data with
cobaya-install act_dr6_cmbonly.PlanckActCut
You can now run the tests with
pytest -v --pyargs act_dr6_cmbonly
If the tests return without any error (i.e. with only warnings), then the code is probably correctly installed. You may get some tests which get skipped if you do not install the differentiable likelihood (see below) - you do not need to worry about this. You can then attempt to run chains with
cobaya-run yamls/act-lcdm.yaml
(and other data combinations if you want, see the files yamls/).
If you cannot access the data, there is a simulated test dataset stored inside and loaded from the pip dist (the directory where pip installed your ACT DR6 lite likelihood). It may be necessary to obey the cobaya convention and put everything inside your packages path (this is the directory where cobaya will look for data). For this, you will have to manually move the dataset over there with
mkdir -p $COBAYA_PACKAGES_PATH/data/ACTDR6CMBonly
cp act_dr6_cmbonly/data/dr6_data_cmbonly.fits $COBAYA_PACKAGES_PATH/data/ACTDR6CMBonly/dr6_data_cmbonly.fits
If you ever get an error that the likelihood cannot locate the data, then you can add a --debug flag to your cobaya call, which will make the likelihood output a message for the exact path where it is trying to locate the data from.
By default, the likelihood will look for the data in either
<pip directory>/act_dr6_cmbonly/data/if no cobaya packages path is given, or<cobaya packages path>/data/ACTDR6CMBonly/if a cobaya packages path is given.
We provide a set of example files in the yamls/ folder. In addition, we have a separate repository where we keep our most up-to-date parameter and settings files.
The example files include a series of single-parameter extension models that can be run with ACT-lite. For example, to recover the ACT constraints on LCDM, you can run
cobaya-run yamls/act-lcdm.yaml
There are a variety of single-parameter extensions provided in the yamls/parameters subfolder. To include for example Neff into your run, you can make a file where you simply change the parameters to include the neff.yaml file. You can do this by changing the params block to:
params: !defaults [params/cosmology, params/neff]
(Make sure to rename your output target so you don't overwrite your LCDM chain!)