Skip to content
Snippets Groups Projects
user avatar
authored

Python module for binary_c

Docstring coverage: docstring coverage Test coverage: test coverage

Binary population synthesis code that interfaces with binary_c. Based on a original work by Jeff Andrews (can be found in old_solution/ directory). Updated and extended for Python3 by David Hendriks, Robert Izzard.

The current release is version version, make sure to use that version number when installing!

Requirements

To run this code you need to at least have installations of:

  • Python 3.6 or higher
  • binary_c version 2.1.7 or higher

And the following python packages (which will get installed automatically when installing with pip):

  • numpy
  • pytest
  • h5py
  • pathos
  • pandas
  • astropy
  • matplotlib
  • py_rinterpolate

Environment variables

Before compilation you need to have certain environment variables:

Required:

  • BINARY_C should point to the root directory of your binary_c installation
  • LD_LIBRARY_PATH should include $BINARY_C/src and whatever directories are required to run binary_c (e.g. locations of libgsl, libmemoize, librinterpolate, etc.)
  • LIBRARY_PATH should include whatever directories are required to build binary_c (e.g. locations of libgsl, libmemoize, librinterpolate, etc.)
  • GSL_DIR should point to the root location where you installed GSL to. This root dir should contain bin/, lib/ etc

Build instructions

First, make sure you have built binary_c (See $BINARY_C/doc/binary_c2.pdf section: installation for all the installation instructions for binary_c)) and that it functions correctly.

Installation via PIP:

To install this package via pip:

pip install binarycpython

This will install the latest stable installation that is available on pip.

Installation from source:

We can also install the package from source, which is useful for development versions and when you want to modify the code. It is recommended that you install this into a virtual environment. From within the root directory, run

./install.sh

This will install the package, along with all the dependencies.

If this is not the first time you install the package, but rather rebuild it because you made changes in either binary_c or binarycpython, you can run

./install_without_dependencies.sh

to reinstall just binarycpython.

After installation

After installing the code via source it is useful to run the test suite before doing any programming with it. The test suite is stored in binarycpython/tests and running python main.py in there will run all the tests.

Examples

See the examples/ directory for example scripts and notebooks. The documentation contains example pages as well.

Usage notes

Make sure that with every change/recompilation you make in binary_c, you also rebuild this package. Whenever you change the sourcecode of this package, you need to reinstall it into your virtualenvironment as well

Documentation

Look in the docs/ directory. Within the build/html/ there is the html version of the documentation. The

Development:

If you want to contribute to the code, then it is recommended that you install the packages in development_requirements.txt:

pip install -r development_requirements.txt

FAQ/Issues:

Building issues with binary_c itself:

  • see the documentation of binary_c (in doc/).
  • If you have MESA installed, make sure that the $MESASDK_ROOT/bin/mesasdk_init.sh is not sourced. It comes with its own version of some programs, and those can interfere with installing.

When Pip install fails:

  • Run the installation with -v and/or --log <logfile> to get some more info
  • If gcc throws errors like gcc: error: unrecognized command line option ‘-ftz’; did you mean ‘-flto’?, this might be due to that the python on that system was built with a different compiler. It then passes the python3.6-config --cflags to the binarycpython installation, which, if done with gcc, will not work. Try a different python3.6. I suggest using pyenv to manage python versions. If installing a version of python with pyenv is not possible, then try to use a python version that is avaible to the machine that is built with the same compiler as binary_c was built with.
  • if pip installation results in No files/directories in /tmp/pip-1ckzg0p9-build/pip-egg-info (from PKG-INFO), try running it verbose (-v) to see what is actually going wrong.
  • If pip terminates with the error FileNotFoundError: [Errno 2] No such file or directory: '<...>/binary_c-config' Then make sure that the path to your main $BINARY_C directory is set correctly.

Other:

  • When running jupyter notebooks, make sure you are running the jupyter installation from the same virtual environment.
  • When the output of binary_c seems to be different than expected, you might need to rebuild this python package. Everytime binary_c is compiled, this package needs to be rebuilt too.