Encoding The Lexicographic Ordering Constraint in SAT Modulo Theories

Hani A. Elgabou,Alan M. Frisch
Abstract:This paper presents eight different SMT encodings of the lexicographic ordering constraint. These constraints are helpful in breaking some kinds of symmetries in combinatorial decision and optimisation problems. The encodings are obtained from the literature then translated into an SMT suitable form. We have done this using two methods, the first is by directly translating the encodings into SMT. The second starts by rewriting the encodings in MiniZinc language, flattening them into FlatZinc instances then using a tool called fzn2smt to translate them to SMT. We evaluate the encodings on a suite of instances of the Social Golfer problem, which is well known for its highly symmetric models. This shows that different encodings of the lexicographic ordering constraint perform best on different problem instances and that no one encoding is dominant.
What problem does this paper attempt to address?