Normalization of SMT-LIB scripts
Satisfiability modulo theories (SMT) is about determining the satisfiability of logical formulas over a range of one or more theories. SMT solvers are programs that are designed to determine the satisfiability of logical formulas and to find a satisfying model. Since the SMT problem is known to be NP-complete, the SMT initiative holds annual competitions to further advance and stimulate the techniques and tools used by the SMT community. To prevent unfair entries to the competition, a benchmark scrambler with a random seed is used to rename variables, randomly permutes arguments and hides benchmark names for each of the formulas. This paper is about researching the possibility of proving satisfiability of expressions that are semantically equal but not necessarily syntactical. The purpose of this project was to develop a computer program that is able to normalize benchmark scripts to a standard defined format such that by scrambling syntactically equal scripts would result in the same result when normalized. The implementation that was conceived is incomplete and inefficient for larger SMT scripts in terms of computational complexity due to time constraints however, the research of that was conceived during this thesis provides an interesting insight in the complexity of normalizing benchmark scripts and its relation to graph isomorphism.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)