Deciding Probabilistic Simulation between Probabilistic Pushdown Automata and Finite-State Systems

Abstract

This paper studies the decidability and computational complexity of checking probabilistic simulation pre-order between probabilistic pushdown automata (pPDA) and (probabilistic)finite-state systems. We show that checking classical and combined probabilistic similarity are EXPTIME-complete in both directions and become polynomial if both the number of control states of the pPDA and the size of the finite-state system are fixed. These results show that checking probabilistic similarity is as hard as checking similarity in the standard, i.e., non-probabilistic setting.