Model exploration and analysis for quantitative safety refinement in probabilistic B

The role played by counterexamples in standard system analysis is well known; but less common is a notion of counterexample in probabilistic systems refinement. In this paper we extend previous work using counterexamples to inductive invariant properties of probabilistic systems, demonstrating how they can be used to extend the technique of bounded model checking-style analysis for the refinement of quantitative safety specifications in the probabilistic B language. In particular, we show how the method can be adapted to cope with refinements incorporating probabilistic loops. Finally, we demonstrate the technique on pB models summarising a one-step refinement of a randomised algorithm for finding the minimum cut of undirected graphs, and that for the dependability analysis of a controller design.