act_dr6_lenslike 1.0.0

Likelihood for the Atacama Cosmology Telescope DR6 CMB lensing data.

ACT DR6 Lensing Likelihood

This repository contains likelihood software for the ACT DR6 CMB lensing analysis. If you use this software and/or the associated data, please cite both of the following papers:

• Madhavacheril, Qu, Sherwin, MacCrann, Li et al ACT Collaboration (2023), arxiv:2304.05203
• Qu, Sherwin, Madhavacheril, Han, Crowley et al ACT Collaboration (2023), arxiv:2304.05202

In addition, if you use the ACT+Planck lensing combination variant from the likelihood, please also cite:

• Carron, Mirmelstein, Lewis (2022), arxiv:2206.07773, JCAP09(2022)039

Chains

A pre-release version of the chains from Madhavacheril et al are available here. Please make sure to read the README file.

Step 1: clone this repository

Step 2: download and unpack data

Download the likelihood data tarball for ACT DR6 lensing from NASA's LAMBDA archive.

Extract the tarball into the act_dr6_lenslike/data/ directory in the cloned repository such the directory v1.1 is directly inside it. Only then should you proceed with the next steps.

Step 3: install Python package

Install with symbolic links

pip install -e . --user

Step 4: Run tests

python setup.py test

Step 5: use in Python codes

Generic Python likelihood

import act_dr6_lenslike as alike

variant = 'act_baseline'
lens_only = False # use True if not combining with any primary CMB data
like_corrections = True # should be False if lens_only is True

# Do this once
data_dict = alike.load_data(variant,lens_only=lens_only,like_corrections=like_corrections)
# This dict will now have entries like `data_binned_clkk` (binned data vector), `cov`
# (covariance matrix) and `binmat_act` (binning matrix to be applied to a theory
# curve starting at ell=0).

# Get cl_kk, cl_tt, cl_ee, cl_te, cl_bb predictions from your Boltzmann code.
# These are the CMB lensing convergence spectra (not potential or deflection)
# as well as the TT, EE, TE, BB CMB spectra (needed for likelihood corrections)
# in uK^2 units. All of these are C_ell (not D_ell), no ell or 2pi factors.
# Then call
lnlike=alike.generic_lnlike(data_dict,ell_kk,cl_kk,ell_cmb,cl_tt,cl_ee,cl_te,cl_bb)

Cobaya likelihood

Your Cobaya YAML or dictionary should have an entry of this form

likelihood:
    act_dr6_lenslike.ACTDR6LensLike:
        lens_only: False
        stop_at_error: True
        lmax: 4000
        variant: act_baseline

No other parameters need to be set. (e.g. do not manually set like_corrections or no_like_corrections here).

Important parameters

• variant should be
  • act_baseline for the ACT-only lensing power spectrum with the baseline multipole range
  • act_extended for the ACT-only lensing power spectrum with the extended multipole range (L<1250)
  • actplanck_baseline for the ACT+Planck lensing power spectrum with the baseline multipole range
  • actplanck_extended for the ACT+Planck lensing power spectrum with the extended multipole range (L<1250)
• lens_only should be
  • False when combining with any primary CMB measurement
  • True when not combining with any primary CMB measurement

Recommended theory accuracy

For CAMB calls, we recommend the following (or higher accuracy):

• lmax: 4000
• lens_margin:1250
• lens_potential_accuracy: 4
• AccuracyBoost:1
• lSampleBoost:1
• lAccuracyBoost:1
• halofit_version:mead2016

Planned updates

• Cobaya YAML files
• CosmoSIS wrapper