An Efficient Lazy SMT Solver for Nonlinear Numerical Constraints

Verification and testing for cyber-physical system may involve constraints that have both Boolean and nonlinear numerical expressions. Such constraints are undecidable, so efficiently giving out a solution whose error is within a given bound when the constraints are satisfiable or the answer that there are some contradicts in the constraints when the constraints are unsatisfiable is important. This paper presents an efficient method to solve Boolean combinations of nonlinear numerical constraints. Under the lazy SMT framework, this method combines an iterative optimization method and interval analysis together to solve numerical constraints. A prototype tool is implemented and some experiments were made. The experimental results show that our method is efficient and can solve many problems that other tools cannot.