English

Interpolation and Model Checking for Nonlinear Arithmetic

Logic in Computer Science 2021-06-09 v1 Programming Languages Symbolic Computation

Abstract

We present a new model-based interpolation procedure for satisfiability modulo theories (SMT). The procedure uses a new mode of interaction with the SMT solver that we call solving modulo a model. This either extends a given partial model into a full model for a set of assertions or returns an explanation (a model interpolant) when no solution exists. This mode of interaction fits well into the model-constructing satisfiability (MCSAT) framework of SMT. We use it to develop an interpolation procedure for any MCSAT-supported theory. In particular, this method leads to an effective interpolation procedure for nonlinear real arithmetic. We evaluate the new procedure by integrating it into a model checker and comparing it with state-of-art model-checking tools for nonlinear arithmetic.

Keywords

Cite

@article{arxiv.2106.04340,
  title  = {Interpolation and Model Checking for Nonlinear Arithmetic},
  author = {Dejan Jovanović and Bruno Dutertre},
  journal= {arXiv preprint arXiv:2106.04340},
  year   = {2021}
}

Comments

To be published in CAV 2021