LSSTDESC / NaMaster

A unified pseudo-Cl framework
BSD 3-Clause "New" or "Revised" License
56 stars 26 forks source link

NaMaster

Build Status Docs Status Coverage Status

NaMaster is a C library, Python module and standalone program to compute full-sky angular cross-power spectra of masked fields with arbitrary spin and an arbitrary number of known contaminants using a pseudo-Cl (aka MASTER) approach. The code also implements E/B-mode purification and is available in both full-sky and flat-sky modes.

Installation

There are different ways to install NaMaster. In rough order of complexity, they are:

Conda forge

Unless you care about optimizing the code, it's worth giving this one a go. The conda recipe for NaMaster is currently hosted on conda-forge (infinite kudos to Mat Becker for this). In this case, installing NaMaster means simply running:

conda install -c conda-forge namaster

If that works for you and you don't care about optimizing the code too much, skip the rest of this section. If you don't have admin permissions, you can give virtual environments a try (or else follow the instructions below).

PyPI

NaMaster is also hosted on PyPI. Installing it should be as simple as running:

python -m pip install pymaster [--user]

(add --user if you don't have admin permissions). Note that this will compile the code on your machine, so you'll need to have installed its dependencies.

From source

If all the above fail, try to install NaMaster from its source. You should first clone this github repository. Then follow these steps:

1. Install dependencies.

Install the dependencies listed here. Note that some of them (HEALPix) may not be necessary, as pymaster will attempt to install them automatically.

2. Install the python module

Installing the python module pymaster should be as simple as running

python setup.py install [--user]

or, even better, if you can use pip:

pip install . [--user]

where the optional --user flag can be used if you don't have admin privileges.

You can check that the python installation works by running the unit tests:

pytest -vv pymaster

Note that the test directory, containing all unit tests, also contains all the sample python scripts described in the documentation.

If you installed pymaster via pip, you can uninstall everything by running

pip uninstall pymaster

Note that the C library is automatically compiled when installing the python module. If you care about the C library at all, or you have trouble compiling it, see the next section.

3. Install the C code (optional)

The script scripts/install_libnmt.sh contains the instructions run by setup.py to compile the C library (libnmt.a). You may have to edit this file or make sure to include any missing compilation flags if setup.py encounters issues compiling the library.

If you need the C library for your own code, scripts/install_libnmt.sh installs it in _deps/lib and _deps/include. Note that the script process will also generate an executable namaster, residing in _deps/bin that can be used to compute power spectra. The use of this program is discouraged over using the python module.

You can check that the C code works by running

make check

If all the checks pass, you're good to go.

Installing on Mac

NaMaster can be installed on Mac using any of the methods above as long as you have either the clang compiler with OpenMP capabilities or the gcc compiler. Both can be accessed via homebrew. If you don't have either, you can still try the conda installation above.

Note: NaMaster is not supported on Windows machines yet.

Documentation

The following sources of documentation are available for users:

Dependencies

NaMaster has the following dependencies, which should be present in your system before you can install the code from source:

Besides these, NaMaster will attempt to install the following additional dependency. If this fails, or if you'd like to use your own preinstalled versions, follow these instructions:

Licensing, credits and feedback

You are welcome to re-use the code, which is open source and freely available under terms consistent with BSD 3-Clause licensing (see LICENSE).

If you use NaMaster for any scientific publication, we kindly ask you to cite this github repository and the companion paper https://arxiv.org/abs/1809.09603. Special kudos should go to the following heroes for their contributions to the code:

For feedback, please contact the author via github issues or email (david.alonso@physics.ox.ac.uk).