On the strategy synthesis problem in MDPs: probabilistic CTL and rolling windows.

14 février 2022

Damien Busatto-Gaston (Université libre de Bruxelles)

In this talk I will present ongoing work on Markov decision processes.

Given an MDP and a formula phi, the strategy synthesis problem asks if there exists a strategy for the MDP such that the resulting Markov chain satisfies phi.
This challenging problem is known to be undecidable for the probabilistic temporal logic PCTL.

We study a class of formulae that can be seen as a fragment of PCTL where a local, bounded horizon property is enforced globally. Moreover, we allow for linear expressions
in the probabilistic inequalities.

We show that this class of formulae is at the frontier of decidability, depending on the type of strategies considered. In particular, strategy synthesis becomes decidable when strategies are either deterministic or memoryless, while the general problem remains undecidable.

We compare with results on the entire PCTL logic and consider applications to the PCTL satisfiability problem.