The Topic

The constraint-based product configurator CAS Merlin is at its core an optimizing SAT (Boolean Satisfiability) solver. Merlin's core algorithm, similarly to academic SAT solvers, is based on heuristic search of (partial) truth assignments. The performance of the search highly depends on the guidance of a branching (also called decision) heuristic. There already exist many decision heuristics developed for classical SAT and MaxSAT solving, however, they may not all be efficient in our product configurator, since there are certain differences between Merlin and a standard SAT/MaxSAT solver. The topic of the thesis is to implement and experimentally evaluate various existing (and newly invented by the author) decision heuristics in the product configurator Merlin.

Your Tasks

• Do a literature research to assemble a list of decision heuristics for SAT (Boolean Satisfiability) and MaxSAT (Maximum Satisfiability) solvers.
• Evaluate these heuristics and identify those, which may be suitable for our product configurator Merlin.
• Invent and develop your own novel decision heuristic(s).
• Implement and experimentally evaluate the collected heuristics in Merlin using real-world customer data.
• Summarize the results in your thesis and/or scientific publication(s).

Your Skills

• Reading and understanding scientific papers.
• 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.