On a class of languages recognizable by probabilistic reversible decide-and-halt automata

We analyze the properties of probabilistic reversible decide-and-halt automata (DH-PRA) and show that there is a strong relationship between DH-PRA and 1-way quantum automata. We show that a general class of regular languages is not recognizable by DH-PRA by proving that two “forbidden” constructions in minimal deterministic automata correspond to languages not recognizable by DH-PRA. The shown class is identical to a class known to be not recognizable by 1-way quantum automata. We also prove that the class of languages recognizable by DH-PRA is not closed under union and other non-trivial Boolean operations.     

Algebraic Results on Quantum Automata

We use tools from the algebraic theory of automata to investigate the class of languages recognized by two models of Quantum Finite Automata (QFA): Brodsky and Pippenger's end-decisive model (which we call BPQFA) and a new QFA model (which we call LQFA) whose definition is motivated by implementations of quantum computers using nucleo-magnetic resonance (NMR). In particular, we are interested in LQFA since NMR was used to construct the most powerful physical quantum machine to date. We give a complete characterization of the languages recognized by LQFA and by Boolean combinations of BPQFA. It is a surprising consequence of our results that LQFA and Boolean combinations of BPQFA are equivalent in language recognition power.