thecomplexityoftemporallogicmodelchecking-Checking.PDF

1 The Complexity of Temporal Logic Model Checking Ph. Schnoebelen1 1 Introduction Temporal logic. Logical formalisms for reasoning about time and the timing of events appear in several ?elds: physics, philosophy, linguistics, etc. Not surprisingly, they also appear in computer science, a ?eld where logic is ubiquitous. Here temporal logics are used in automated reasoning, in planning, in semantics of programming languages, in arti?cial intelligence, etc. There is one area of computer science where temporal logic has been unusually successful: the speci?cation and veri?cation of programs and sys- tems, an area we shall just call “programming” for simplicity. In today’s curricula, thousands of programmers ?rst learn about temporal logic in a course on model checking! Temporal logic and programming. Twenty ?ve years ago, Pnueli iden- ti?ed temporal logic as a very convenient formal language in which to state, and reason about, the behavioral properties of parallel programs and more generally reactive systems [Pnu77, Pnu81]. Indeed, correctness for these sys- tems typically involves reasoning upon related events at di?erent moments of a system execution [OL82]. Furthermore, when it comes to liveness prop- erties, the expected behavior of reactive systems cannot be stated as a static property, or as an invariant one. Finally, temporal logic is well suited to ex- pressing the whole variety of fairness properties that play such a prominent role in distributed systems [Fra86]. For these applications, one usually restricts oneself to propositional tem- poral logic: on the one hand, this does not appear to be a severe limi- tation in practice, and on the other hand, this restriction allows decision procedures for validity and entailment, so that, at least in principle, the above-mentioned reasoning can be automated. Model checking. Generally speaking, model checking is the algorithmic veri?cation that a given logic formula holds