QBF Programming with the Modeling Language Bule

We introduce Bule, a modeling language for problems from the complexity class PSPACE via quantified Boolean formulas (QBF) - that is, propositional formulas in which the variables are existentially or universally quantified. Bule allows the user to write a high-level representation of the problem in a natural, rule-based language, that is inspired by stratified Datalog. We implemented a tool of the same name that converts the high-level representation into DIMACS format and thus provides an interface to aribtrary QBF solvers, so that the modeled problems can also be solved. We analyze the complexity-theoretic properties of our modeling language, provide a library for common modeling patterns, and evaluate our language and tool on several examples.

Authors

• Jean Christoph Jung
• Valentin Mayer-Eichberger
• Abdallah Saffidine

Keywords

• Modeling
• QBF Programming
• CNF Encodings