System specification formalisms should come with suitable property specification languages and effective verification tools. We sketch a framework for the verification of quantified temporal properties of systems with dynamically evolving structure. We consider visual specification formalisms like graph transformation systems (GTS) where program states are modelled as graphs, and the program behaviour is specified by graph transformation rules. The state space of a GTS can be represented as a graph transition system (GTrS), i.e. a transition system with states and transitions labelled, respectively, with a graph, and with a partial morphism representing the evolution of state components. Unfortunately, GTrSs are prohibitively large or infinite even for simple systems, making verification intractable and hence calling for appropriate abstraction techniques.
State-of-the-art in GTS LogicsAfter the pioneering works on monadic second-order logic (MSO) [7], various graph logics have been proposed and their connection with topological properties of graphs investigated [8]. The need to reason about the evolution of graph topologies has then led to combining temporal and graph logics in propositional temporal logics using graph formulae as state observations (e.g. [4]). However, due to the impossibility to interleave the graphical and temporal dimensions it was not possible to reason on the evolution of single graph components. To overcome this limitation, predicate temporal logics were proposed (e.g. [2,16]), where edge and node quantifiers can be interleaved with temporal operators.More recent approaches [2] propose quantified µ-calculi combining the fixpoint and modal operators with MSO. These logics fit at the right level of abstraction for GTSs, allowing to reason on the topological structure of a state, and on the evolution of its components. We refer to § 8 of [11] for a more complete discussion. Unfortunately, the semantical models for such logics are less clearly cut. Current solutions are not perfectly suited to model systems with dynamic structure, where components might get merged [2,16], or (re)allocated [2]. These problems are often solved by restricting the class of admissible models or by reformulating the state transition relation, hampering the meaning of the logic.