Skip to content

Latest commit

 

History

History
137 lines (92 loc) · 4.37 KB

INSTALL.rst

File metadata and controls

137 lines (92 loc) · 4.37 KB
Raw

Installation procedure for Healpy

Requirements

Healpy depends on the HEALPix C++ and cfitsio C libraries. Source code for both is included with Healpy and is built automatically, so you do not need to install them yourself. Only Linux and MAC OS X are supported, not Windows.

Binary installation with conda (RECOMMENDED)

Conda forge provides a conda channel with a pre-compiled version of healpy for linux 64bit and MAC OS X platforms, you can install it in Anaconda with:

conda config --add channels conda-forge
conda install healpy

Source installation with Pip

It is possible to build the latest healpy with pip

pip install --user healpy

If you have installed with pip, you can keep your installation up to date by upgrading from time to time:

pip install --user --upgrade healpy

On Linux with newer compilers many users reported compilation errors like configure: error: cannot run C compiled programs, the solution is to specifiy the flags for the C and CXX compiler:

CC=gcc CXX=g++ CFLAGS='-fPIC' CXXFLAGS='-fPIC' pip install --user healpy

Installation on Mac OS with MacPorts

If you are using a Mac and have the MacPorts package manager, it's even easer to install Healpy with:

sudo port install py27-healpy

Binary apt-get style packages are also available in the development versions of Debian (sid) and Ubuntu (utopic).

Almost-as-quick installation from official source release

Healpy is also available in the Python Package Index (PyPI). You can download it with:

curl -O https://pypi.python.org/packages/source/h/healpy/healpy-1.7.4.tar.gz

and build it with:

tar -xzf healpy-1.7.4.tar.gz
pushd healpy-1.7.4
python setup.py install --user
popd

If everything goes fine, you can test it:

python

>>> import matplotlib.pyplot as plt
>>> import numpy as np
>>> import healpy as hp
>>> hp.mollview(np.arange(12))
>>> plt.show()

or run the test suite with nose:

cd healpy-1.7.4 && python setup.py test

Building against external Healpix and cfitsio

Healpy uses pkg-config to detect the presence of the Healpix and cfitsio libraries. pkg-config is available on most systems. If you do not have pkg-config installed, then Healpy will download and use (but not install) a Python clone called pykg-config.

If you want to provide your own external builds of Healpix and cfitsio, then download the following packages:

If you are going to install the packages in a nonstandard location (say, --prefix=/path/to/local), then you should set the environment variable PKG_CONFIG_PATH=/path/to/local/lib/pkgconfig when building. No other environment variable settings are necessary, and you do not need to set PKG_CONFIG_PATH to use Healpy after you have built it.

Then, unpack each of the above packages and build them with the usual configure; make; make install recipe.

Development install

Developers building from a snapshot of the github repository need:

  • autoconf and libtool (in Debian or Ubuntu: sudo apt-get install autoconf automake libtool pkg-config)
  • cython > 0.16
  • run git submodule init and git submodule update to get the bundled HEALPix sources

the best way to install healpy if you plan to develop is to build the C++ extensions in place with:

python setup.py build_ext --inplace

then add the healpy/healpy folder to your PYTHONPATH.

In case of compilation errors, see the note above in the pip section.

Clean

When you run "python setup.py", temporary build products are placed in the "build" directory. If you want to clean out and remove the build directory, then run:

python setup.py clean --all