Operational Characterization of Divisibility of Dynamical Maps

Divisibility of dynamical maps turns out to be a fundamental notion in characterising Markovianity of quantum evolution, although the decision problem for divisibility itself is computationally intractable. In this work, we propose the operational characterisation of divisibility of dynamical maps by exploiting distinguishability of quantum channels. We prove that distinguishability for any pair of quantum channels does not increase under divisible maps, and then, in terms of channel distinguishability with entanglement between system and $k$-dimensional ancillas, provide the operational characterization for the full hierarchy of the so-called $k$-divisiblity $(k=1,2,\ldots)$. Finally, from the fact that min-entropy corresponds to the information-theoretic measure of distinguishability, the entropic characterisation to divisible maps is also provided.