L2C compiles lifted inference or weighted first-order model counting (WFOMC) into C++ programs.
Make sure you have Ruby and g++ installed on your machine and the default Ruby version is 2.3.0 or higher.
Running the L2C compiler for an input theory:
$ ruby L2C.rb -f <filename>
For example, to find the WFOMC of example1 in the examples folder, use:
$ ruby L2C.rb -f examples/example1.wmc
Specify the elimination ordering heuristic using -h. The possible values are MNL for MinNestedLoops and MTS for MinTableSize. The default heuristic is MNL.
If MNL is selected, specify the number of stochastic local search iterations using -k. The default value is 25.
Specify if the generated C++ programs must be readable or not using -r. The possible values are true (for readable) and false (for not readable). The default value is false. Generating readable C++ codes might be slightly slower.
As an example, the following command runs L2C for example1 using MNL with 20 stochastic local search iterations and outputs readable C++ code:
$ ruby L2C.rb -f examples/example1.wmc -h MNL -k 20 -r true
See manual.pdf for the documentation.
Refer to the following publications for the details of the algorithms used in this compiler:
-
Seyed Mehran Kazemi and David Poole
Knowledge Compilation for Lifted Probabilistic Inference: Compiling to a Low-Level Program
In Proceedings of the 15th International Conference on Principles of Knowledge Representation and Reasoning, 2016
-
Seyed Mehran Kazemi and David Poole
Why is Compiling Lifted Inference into a Low-Level Program so Efficient?
In Proceedings of the Sixth International Workshop on Statistical Relational AI, 2016
-
Seyed Mehran Kazemi and David Poole
Elimination Ordering in Lifted First-Order Probabilistic Inference
In Proceedings of the 28th AAAI Conference, 2014
Seyed Mehran Kazemi
Computer Science Department
The University of British Columbia
201-2366 Main Mall, Vancouver, BC, Canada (V6T 1Z4)
http://www.cs.ubc.ca/~smkazemi/
Licensed under the GNU General Public License Version 3.0. https://www.gnu.org/licenses/gpl-3.0.en.html
Copyright (C) 2016 Seyed Mehran Kazemi