The complexity of CTL* + linear past

  • Authors:
  • Laura Bozzelli

  • Affiliations:
  • Università di Napoli Federico II, Napoli, Italy

  • Venue:
  • FOSSACS'08/ETAPS'08 Proceedings of the Theory and practice of software, 11th international conference on Foundations of software science and computational structures
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

We investigate the complexity of satisfiability and finite-state model-checking problems for the branching-time logic CTL*lp, an extension of CTL* with past-time operators, where past is linear, finite, and cumulative. It is well-known that CTL*lp has the same expressiveness as standard CTL*, but the translation of CTL*lp into CTL* is of nonelementary complexity, and no elementary upper bounds are known for its satisfiability and finite-state model checking problems. In this paper, we provide an elegant and uniform framework to solve these problems, which non-trivially extends the standard automata-theoretic approach to CTL* model-checking. In particular, we show that the satisfiability problem for CTL*lp is 2Exptime-complete, which is the same complexity as that of CTL*, but for the existential fragment of CTL*lp, the problem is Expspace-complete, hence exponentially harder than that of the existential fragment of CTL*. For the model-checking, the problem is already Expspace-complete for the existential and universal fragments of CTL*lp. For full CTL*lp, the proposed algorithm runs in time polynomial in the size of the Kripke structure and doubly exponential in the size of the formula. Thus, the exact complexity of model-checking full CTL*lp remains open: it lies somewhere between Expspace and 2Exptime.