The Topic

The constraint-based product configurator CAS Merlin is at its core an optimizing SAT (Boolean Satisfiability) solver. The rules of a product are represented as a set of propositional clauses. During the configuration process we need to find a satisfying assignment to these clauses under various assumptions. This is an NP-complete task, which must be accomplished at every single configuration click. One way to improve the performance of this calculation is to preprocess the clause set in certain ways. There already exist many algorithms developed for improving clause sets for classical SAT solving, however, they are not all applicable for our product configurator. The topic of the thesis is to identify which clause set optimization methods are applicable (and invent novel methods) and then efficiently implement and evaluate them in our product configurator Merlin.

Your Tasks

• Do a literature research to assemble a list of algorithms for (propositional logic) clause set optimization, also known as knowledge compilation.
• Evaluate these methods and identify those that are applicable in a constraint based product configurator like Merlin.
• Possibly invent and develop your own novel methods.
• Implement and experimentally evaluate the applicable methods in Merlin using real-world customer data.
• Summarize the results in your thesis and/or scientific publication(s).

Your Skills

• You can find, read and understand scientific papers relevant for a given topic.
• Creativity in approaching hard problems.
• Sick JAVA coding skillz (so that the programs you write run fast).
• Not being scared off by hard (NP-Hard) problems.
• Communicating your ideas clearly and precisely in English.
• Working together with others in German is also easy for you.