Implements gaussian expectation propagation for any tree-like probabilistic graphical model.

Documentation website: https://sphinxteam.github.io/tramp.docs

• python>=3.6
• numpy/pandas/scipy/matplotlib
• networkx==1.11
• daft

Warning Currently the package does not support networkx 2.xx and will throw unexpected errors. We plan to upgrade to networkx 2.xx at some point.

To install the package, go to the folder where setup.py is located and run:

pip install .

or if you want to install in development mode (changes to the repository will immediately affect the installed package without needing to re-install):

pip install --editable .

To install the package on a remote machine directly from the github repo:

pip install git+https://github.com/sphinxteam/tramp.git

See installing from sources for more details. In both cases, the necessary requirements should be automatically installed.

The package is presented in more details in the corresponding paper on arXiv

Illustrating notebooks and scripts are gathered in the tramp_notebooks repo.

Codes corresponding to the examples presented in the above mentioned paper can be found in the tramp_examples repo.

Both the SPHINX team and the SMILE team acknowledge funding from: