An implementation of the Komatsu-Spergel-Wandelt bispectrum estimator for modern CMB data.

• Python>=3.7
• a C compiler (tested with gcc and icc)
• fftw
• intel MKL library (see instructions below)
• Python>=3.6
• numpy
• scipy
• cython
• pytest
• mpi4py
• h5py
• healpy
• camb
• optweight

Start by making sure the MKL library is loaded in your environment. On most clusters this can be achieved by loading a predefined module. On the Princeton della and tiger clusters you can use module load intel-mkl (see here for more information). On NERSC you can use load intel (see here). Once you have loaded the module, check if the MKLROOT environment variable has been set (echo $MKLROOT).

Once the MKL environment has been set, git clone this repository, go into the directory and run:

$ make

$ pip install .

Run tests:

$ make check

Consider adding the -e flag to the pip install command to enable automatic updating of code changes when developing.

As an alternative to the pip install . command, you can run:

$ make python

to simply build the python package inside this directory.

• libpshtlight extracted from libpsht. Copyright Martin Reinecke. Licensed under the GNU GPL v.2.
• hyperspherical extracted from CLASS (see directory for more information)
• libpsht python wrappers adapted from wavemoth. Copyright Dag Sverre Seljebotn. Licensed under the GNU GPL v.2.