Abstract:We introduce the notion of quantum Markov decision process (qMDP) as a semantic model of nondeterministic and concurrent quantum programs. It is shown by examples that qMDPs can be used in analysis of quantum algorithms and protocols. We study various reachability problems of qMDPs both for the finite-horizon and for the infinite-horizon. The (un)decidability and complexity of these problems are settled, and the relationship between one of them and the joint spectral radius problem, a long-standing open problem in matrix analysis and control theory, is clarified. Some of these results show a certain separation between the MDP and qMDP models. We also develop an algorithm for finding an optimal scheduler for a large class of qMDPs. Finally, the results of reachability problems are applied in the analysis of the safety problem for qMDPs.

Reachability Analysis of Quantum Markov Decision Processes.