The reduced dynamics of the system S, interacting with the environment E, is not given by a linear map, in general. However, if it is given by a linear map, then this map is also Hermitian. In order that the reduced dynamics of the system is given by a linear Hermitian map, there must be some restrictions on the set of possible initial states of the system-environment or on the possible unitary
... [Show full abstract] evolutions of the whole SE.
In this paper, adding an ancillary reference space R, we assign to each convex set of possible initial states of the system-environment $\mathcal{S}$, for which the reduced dynamics is Hermitian, a tripartite state $\omega_{RSE}$, which we call it the reference state, such that the set $\mathcal{S}$ is given as the steered states from the reference state $\omega_{RSE}$,. The set of possible initial states of the system is also given as the steered set from a bipartite reference state $\omega_{RS}$. The relation between these two reference states is as $\omega_{RSE}=id_{R}\otimes \Lambda_{S}(\omega_{RS})$, where $id_{R}$ is the identity map on R and $\Lambda_{S}$ is a Hermitian assignment map, from S to SE. As an important consequence of introducing the reference state $\omega_{RSE}$, we generalize the result of
[F. Buscemi, Phys. Rev. Lett. 113, 140502 (2014)}]: We show that, for a U-consistent subspace, the reduced dynamics of the system is completely positive, for arbitrary unitary evolution of the whole system-environment U, if and only if the reference state $\omega_{RSE}$ is a Markov state. In addition, we show that the evolution of the set of system-environment (system) states is determined by the evolution of the reference state $\omega_{RSE}$ ($\omega_{RS}$).
