Channel-state duality refers to the correspondence between quantum channels and bipartite states, and it is extremely useful and fruitful in quantum information theory. It is often called the Jamio\l{}kowski isomorphism, the Jamio\l{}kowski-Choi isomorphism, or the Choi-Jamio\l{}kowski isomorphism. We trace the original roots of this duality from a historic perspective, clarify the somewhat misleading term ``isomorphism,'' reveal the underlying subtle nature, and highlight the seemingly similar but actually different features of the original correspondences in the manner of Pillis, Jamio\l{}kowski, and Choi that lead to the duality. We further illustrate some fundamental properties and diverse applications.
We consider the problem of separability of quantum channels via the Choi matrix representation given by the Choi-Jamio\l{}kowski isomorphism. We explore three classes of separability across different cuts between systems and ancillae, and we provide a solution based on the mapping of the coordinates of the Choi state (in a fixed basis) to a truncated moment sequence (tms)
y. This results in an algorithm which gives a separability certificate using semidefinite programming. The computational complexity and the performance of it depend on the number of variables
n in the tms and on the size of the moment matrix
Mt(y) of order
t. We exploit the algorithm to numerically investigate separability of families of two-qubit and single-qutrit channels; in the latter case we can provide an answer for examples explored earlier through the criterion based on the negativity
N, a criterion which remains inconclusive for Choi matrices with
N=0.
Provided a maximally entangled state vector in
H⊗H, the Choi-Jamiolkowski isomorphism of quantum information theory (or channel/state duality) produces a mapping of
CP(H,K) (completely positive maps of density operators on
H tothose on
K) onto the density operators on
K⊗H, which extends to a unitary map. Here,
H and
K are {finite-dimensional} Hilbert spaces.The nature of this construction, the form of the associated formulas, and the possibility of versions which do not require an entangled vector seed, are illuminated from the perspective of {naturality}, in the sense of Eilenberg and MacLane.There is a unique such correspondence (``Choi map'') natural in both
H and
K which produces density operators on either
L(H,K) (Hilbert-Schmidt inner product) or
K⊗H∗ instead of
K⊗H. In particular, no entangled state input is required. There are no such binatural maps carried on
K⊗H, but those natural in only
K are canonically associated with unitary maps
H→H∗ and derive from the binatural maps. The construction recovers the standard Choi-Jamiolkowski formulae and shows how they express naturality.
In chapter 7 we will study quantum channels. We will see how the so-called Choi–Jamiołkowski isomorphism—which links quantum channels with quantum states—allows for semidefinite programming techniques to be applied to quantum channels. We will then see how we can estimate an uncharacterised channel, calculate the diamond norm (which characterises the operational distinguishability between channels) and finally how to use duality to uncover a link between a channel optimisation problem and quantum entropies.
Channel--state duality identifies completely positive maps with bipartite states. Its generalisation to von Neumann algebras is known to produce states not on
M⊗ˉN but on the tensor product of
M with the \emph{opposite} algebra of
N, or equivalently with a commutant; returning to
N itself requires a transposition, and it has been observed that this step typically fails. We determine exactly when it does not fail. Our main result is that a Choi--Jamiołkowski type isomorphism which is natural in the first algebra exists if and only if the second algebra is
∗-anti-isomorphic to itself. Naturality cannot be dropped: whenever
M is anti-isomorphic to itself such an isomorphism exists for trivial reasons, since the order structure of a predual is a Jordan invariant and cannot distinguish an algebra from its opposite. Consequently no such correspondence exists for the type III factors constructed by Connes. We show in addition that the domain admits no description in terms of properties of individual maps: the space of all normal maps with the operator norm is too large, and neither complete boundedness nor complete positivity cuts it down to the right size.