MetaSMT: a unified interface to SMT-LIB2

Various problems from artificial intelligence and formal methods are solved utilizing Satisfiability Modulo Theories (SMT) solvers. Selecting the best SMT solver for a specific application, however, is a daunting task. In this paper, we present the novel metaSMT TCP server and client architecture which can be used to solve SMT instances expressed in SMT-LIB2 by multiple solver processes in parallel. The metaSMT TCP server provides a unified interface for SMT-LIB2 instances with the capability to either use the API or the file interface of a solver process and thus serves as a highly customizable portfolio solver. We show that the run-time overhead required by the metaSMT TCP server and client architecture is marginal using selected benchmarks from SMT-LIB.