delensalot

Curved-sky optimal CMB lensing reconstruction and bias calculation. Delensalot takes an observed CMB map and returns an optimal estimate of the underlying lensing field.

If you use delensalot for your work, please consider citing the ApJ publication CMB-S4: Iterative internal delensing and r constraints.

This figure shows the full sky optimal lensing potential reconstruction for various CMB observations in a 5 times 5 degree patch. The input lensing potential is shown in the leftmost figure.

Features

• Curved-sky analysis
• Anisotropic noise model support
• Masked-sky support
• Quadratic estimator (QE) implementation via Plancklens
• Mock data generation using lenspyx
• Supports various estimators (TT, EE, BB, MV, EE+EB, ..)

Installation

Requires Python 3.9–3.12. First install the two dependencies that are not on PyPI:

pip install git+https://github.com/carronj/plancklens
pip install git+https://github.com/carronj/lenspyx

Note: building plancklens requires a Fortran compiler (gfortran). On macOS: brew install gcc. On Linux: sudo apt-get install gfortran.

Then install delensalot:

git clone https://github.com/NextGenCMB/delensalot.git
cd delensalot
pip install -e .

Setting up a conda environment

conda create --name delensalot python=3.11
conda activate delensalot
conda install pip numpy

pip install git+https://github.com/carronj/plancklens
pip install git+https://github.com/carronj/lenspyx

cd </path/to/delensalot>
pip install -e .

To use delensalot in Jupyter notebooks, add it as a kernel:

pip install ipykernel
python -m ipykernel install --user --name=delensalot

Installation troubleshooting

• attrs errors — make sure you have the attrs package (not attr, which is different): pip install --upgrade attrs. Version 23.1.0 or newer is required.
• astropy / ducc0 errors — usually caused by an outdated astropy. Run pip install --upgrade astropy.
• plancklens build fails — make sure gfortran is installed (see above).
• Still stuck? — run python check_install.py for a full dependency report with fix hints.

Usage

Interactive mode

See first_steps/notebooks/ for tutorials. The minimal working example notebook interactive_mwe.ipynb is a good starting point.

Parameter files

python3 <parfile>.py

See first_steps/parameter_files/ for examples.

MPI / HPC

delensalot supports MPI for parallelisation across simulation indices:

srun -n <ntasks> python3 run.py -r <path-to-config-file>

Dependencies

• Plancklens
• lenspyx
• DUCC
• numpy, scipy, healpy, astropy, attrs, psutil, logdecorator

Documentation

Documentation can be found at delensalot.readthedocs.io.