Installation

Prerequisites

The Trees module uses the MPI-based openbtmixing Python package (See [1] and [2]). Before installing Taweret, please prepare your system as indicated in the OpenBT mixing installation documentation. Note that for some package managers, developer library packages such as libopenmpi-dev or libmpich-dev might need to be installed in addition to the base MPI packages such as openmpi-bin or mpich. openbtmixing also requires the package ninja to be preinstalled, so this may need to be added to the user's system prior to configuring openbtmixing.

Otherwise, the list of dependences is as short as possible to keep the installation process streamlined and allow for minimal, clean installations; however, if a user would like to run the Jupyter notebooks in the associated Jupyter Book, dependences for the notebooks will need to be installed in the relevant environment. These dependences are located in the Jupyter notebooks, and hence can be quickly installed by running the import cell at the top of each notebook.

The bilby sampler comes with the ability to use a suite of samplers---in the case of Taweret, we also have not listed all samplers as dependences. However, the user can (and should) install any samplers that they wish to use and bilby will be able to use them through its wrapper in the Taweret package.

OpenBT References

  1. OpenBT Repository (https://bitbucket.org/mpratola/openbt/src/master/).

  2. OpenBT Repository with Model Mixing (https://github.com/jcyannotty/OpenBT).

Windows Users

While in the past Taweret was known to work with Windows 10 using an appropriately configured Ubuntu 20.04 Windows Subsystem for Linux (WSL) build, it is presently tested only using macOS and Ubuntu installations. Windows users that would like to see if Taweret will work for them might find these instructions useful for helping to keep their Ubuntu WSL up to date, or to install additional features like X support.

Standard Installation

Taweret is available via pip install

pip install Taweret

If you prefer to use conda to setup your Python environment, you can still pip install Taweret, but be sure to conda install pip first, so you conda environment knows where to look.

Alternative Installation

Alternatively, you can clone the repository and install Taweret into your Python environment in developer or editable mode from the clone by running

pip install -e .

Testing

A Taweret installation can be tested directly by executing

>>> import Taweret
>>> Taweret.__version__
>>> Taweret.test()

An Important Note

pip maintains a cache of previously built wheels. Therefore a new Taweret installation may be faulty if openbtmixing was installed using a previously cached wheel that was built with a different or incompatible MPI installation. If some of the trees tests are failing, users can first determine if a clean reinstall addresses the failures by:

  1. uninstalling Taweret,

  2. removing the openbtmixing wheel from cache (e.g., python -m pip cache remove openbtmixing),

  3. and reinstalling Taweret.