No abstract

Uniform coloured Petri nets can be abstracted to their skeleton, the place/transition net that simply turns the coloured tokens into black tokens. A coloured net and its skeleton are related by a net morphism [Des91, PGE98]. For the application of the skeleton as an abstraction method in the model checking process, we need to establish a simulation relation [Mil89] between the state spaces of the two nets. Then, universal temporal properties (properties of the ACT L * logic) are preserved. The abstraction relation induced by a net morphism is not necessarily a simulation relation, due to a subtle issue related to deadlocks [Fin92]. We discuss several situations where the abstraction relation induced by a net morphism is as well a simulation relation, thus preserving ACT L * properties. We further propose a partition refinement algorithm for folding a place/transition net into a coloured net. This way, skeleton abstraction becomes available for models given as place/transition nets. Experiments demonstrate the capabilities of the proposed technology. Using skeleton abstraction, we are capable of solving problems that have not been solved before in the Model Checking Contest [KGH + 19].

Uniform coloured Petri nets can be abstracted to their skeleton, the place/transition net that simply turns the coloured tokens into black tokens. A coloured net and its skeleton are related by a net morphism [1, 2]. For the application of the skeleton as an abstraction method in the model checking process, we need to establish a simulation relation [3] between the state spaces of the two nets. Then, universal temporal properties (properties of the ACTL* logic) are preserved. The abstraction relation induced by a net morphism is not necessarily a simulation relation, due to a subtle issue related to deadlocks [4]. We discuss several situations where the abstraction relation induced by a net morphism is as well a simulation relation, thus preserving ACTL* properties. We further propose a partition refinement algorithm for folding a place/transition net into a coloured net. This way, skeleton abstraction becomes available for models given as place/transition nets. Experiments demonstrate the capabilities of the proposed technology. Using skeleton abstraction, we are capable of solving problems that have not been solved before in the Model Checking Contest [5].