Abstract:
Most models of agents and multi-agent systems include information about possible states of the system (that defines relations between states and their external characteristics), and information about relationships between states. Qualitative models of this kind assign no numerical measures to these relationships. At the same time, quantitative models assume that the relationships are measurable, and provide numerical information about the degrees of relations. In this paper, we explore the analogies between some qualitative and quantitative models of agents/processes, especially those between transition systems and Markovian models.
Typical analysis of Markovian models of processes refers only to the expected utility that can be obtained by the process. On the other hand, modal logic offers a systematic method of describing phenomena by combining various modal operators. Here, we try to exploit linguistic features, offered by propositional modal logic, for analysis of Markov chains and Markov decision processes. To this end, we propose Markov temporal logic – a multi-valued logic that extends the branching time logic CTL*.

Keywords: Temporal logic, Markov chains, Markov decision processes.

Computational Intelligence Group @ Technical University of Clausthal
Human Media Interaction Group @ University of Twente
Computer Science Group @ University of Gdansk	Last modified 2008-03-10