TensorInference is an open source   Julia   package for probabilistic inference over discrete graphical models. It leverages tensor-based technology for efficiently solving various inference tasks.

Features

TensorInference supports finding solutions to the most common probability inference tasks of the UAI inference competitions, which include:

• PR: The partition function or probability of evidence
• MAR: The marginal probability distribution over all variables given evidence
• MAP: The most likely assignment to all variables given evidence
• MMAP: The most likely assignment to the query variables after marginalizing out the remaining variables

Installation

Install TensorInference through the Julia package manager:

pkg> add TensorInference

Examples

Usage examples can be found in the examples folder, and for a comprehensive introduction to the package read the documentation .

Citing

If you use TensorInference as part of your research, teaching, or other activities, please consider citing our work.

Questions and Contributions

Please open an issue if you encounter any problems, or have any feature requests.