Combinatorial optimization is a branch of mathematical optimization that has important applications in many fields. The submodularity function property frequently appears in many combinatorial systems and enables efficiently finding solutions to these problems that would otherwise be intractable. This project is the result of my ETH Master Thesis and offers the fastest to date algorithm implementation for general Submodular Function Minimization and an extensible High-performance foundation. Furthermore, the following four applications are included: Semi-supervised Clustering, Text Corpus Selection, Quadratic Potentials and Minimum Graph Cut.
By downloading the source code you are agreeing to the terms and conditions. Please check the dual license scheme we offer: the GNU Affero General Public License Version 3 and a separate license suitable for commercial use.
The project Wiki (under construction) provides the installation guide and a minimal users manual to get you started reusing the library.
The best documentation available is the Thesis script that provides detailed coverage not only of the algorithm but also the High-performance foundation and the software design.