Accelerating Parameter Synthesis Using Semi-algebraic Constraints

Abstract

We propose a novel approach to parameter synthesis for parametrised Kripke structures and CTL specifications. In our method, we suppose the parametrisations form a semi-algebraic set and we utilise a symbolic representation using the so-called cylindrical algebraic decomposition of corresponding multivariate polynomials. Specifically, we propose a new data structure allowing to compute and efficiently manipulate such representations. The new method is significantly faster than our previous method based on SMT. We apply the method to a set of rational dynamical systems representing complex biological mechanisms with non-linear behaviour.