delensalot

Curved-sky iterative CMB lensing reconstruction and bias calculation.

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

Installation

Download the project, navigate to the root folder and execute the command,

python setup.py install

Make sure that you have the latest plancklens and that you are using the plancklensdev branch

cd plancklens
git checkout plancklensdev

This is needed to give delensalot control over plancklens's mpi implementation, which we conveniently set up at this branch.

You will need to install jupyter for the tutorials found in first_steps/notebooks/, and possibly an ipykernel to create a jupyter-kernel out of the environment in which you install delensalot.

Set up a jupyter-kernel with delensalot

To run the tutorials with a jupyter kernel, you will have to install delensalot in it. Assuming you are using conda for your package management,

conda create --name delensalot
conda activate delensalot

Then, install your favourite packages,

conda install pip, numpy

Now, go to the delensalot directory, and install, including its requirements,

cd </path/to/delensalot>
python3 -m pip install -r requirements .
python3 setup.py develop

Eventually, create your kernel using the environment at which you just installed all packages,

python3 -m ipykernel install --user --name=delensalot

Installation troubles

Frequent problems are

• attrs. If there are errors related to the metamodel, make sure you have attrs (not attr, which is a different package) installed and updated/upgraded (version 23.1.0 should do)
• astropy. If there are errors related to ducc0, chances are it's because of an old astropy installation. Make sure you have the latest astropy.

Usage

The quickest way: anafast(), map2delblm() or map2tempblm()

delensalot comes with two handy functions to get you started very easily. To get a delensed power spectrum, simply import delensalot and run anafast():

import delensalot
delensedmap = delensalot.anafast(obsmaps, lmax_cmb=lmax_cmb, beam=beam, itmax=itmax, noise=noise, verbose=True)

To get a delensed B map, simply import delensalot and run map2delblm():

import delensalot
delensedmap = delensalot.map2delblm(obsmaps, lmax_cmb=lmax_cmb, beam=beam, itmax=itmax, noise=noise, verbose=True)

here obsmaps is the observed Q and U map you may have... gotten from somewhere. Then all what is left to do is to tell delensalot about,

• the maximum \ell (lmax_cmb) of your CMB map you'd like to use,
• the beam (sims_beam) of the transfer function of the observed maps,
• how many iterations (itmax) you'd like to perform,
• and the noise level (noise) of the observation.

If you are interested in the B-lensing template, instead use map2tempblm(),

import delensalot
Blenstemplate = delensalot.map2tempblm(obsmaps, lmax_cmb=lmax_cmb, beam=beam, itmax=itmax, noise=noise, verbose=True)

Run a configuration file

To run a configuration file <path-to-config/conf.py>, type in your favorite bash,

python3 run.py -r <path-to-config/conf.py>

delensalot supports MPI,

srun --nodes <nnodes> -n <taskspernode> python3 run.py -r <path-to-config/conf.py>

If you'd like to know the status of the analysis done with <path-to-config/conf.py>, run,

python3 run.py -s <path-to-config/conf.py>

interactive mode

delensalot supports interactive mode. See first_steps/notebooks/ for our tutorials.

help

Type python3 run.py [-h] for quickhelp,

usage: run.py [-h] [-p NEW] [-r RESUME] [-s STATUS] [-purgehashs PURGEHASHS]

delensalot entry point.

optional arguments:
  -h, --help            show this help message and exit
  -r RESUME             Absolute path to config file to run/resume.
  -s STATUS             Absolute path for the analysis to write a report.

Dependencies

uses

• Plancklens
• lenspyx
• DUCC

Doc

Documentation may be found [HERE]

Use with HPC

delensalot can be computationally demanding. We have parallelized the computations across the simulation index in most cases. Assuming you have MPI set up and srun is available, you can simply run MPI-supported delensalot via,

srun -MPI_paramX X -MPI_paramY Y python3 <path-to-delensalot>/run.py -r <path-to-config-file>

If you have troubles, your HPC-center can help.