PRAiSE (for Probabilistic Reasoning As Symbolic Evaluation) is a library developed at SRI International's Artificial Intelligence Center, which implements the ideas presented in the following paper:
"Probabilistic Inference Modulo Theories",
de Salvo Braz, R., O'Reilly, C., Gogate, V., Dechter, R.,
Proc. of the 25th International Joint Conference on Artificial Intelligence (IJCAI-16).
(revised version, original version, bibtex)
- propositional variables
- equality on categorical types
- equality and inequalities over bounded integers (more specifically, difference arithmetic).
- linear real arithmetic
- algebraic data types
- random functions (also described as relational variables, or uninterpreted functions - the type of inference common in the lifted first-order probabilistic inference literature).
Your possible next steps:
- Play with the demo (requires a browser that supports Java Web Start)
- Read the User Guide
- Use the command-line interface to use the library as an external solver
- Use the library through Maven in your own JVM projects (see "Latest Maven information" at "Getting Started")
- Install the project to use in your own JVM projects or become a contributor