We analyze the Namioka-Phelps minimal and maximal tensor products of compact convex sets which arise as the state spaces of unital C
∗-algebras. Relatedly, we study entanglement in (infinite dimensional) C
∗-algebras. While the minimal Namioka-Phelps tensor product of the state spaces of two C
∗-algebras is the well-known set of separable (= un-entangled) states on the (minimal) tensor product of the C
∗-algebras, we also describe the more elusive maximal Namioka-Phelps tensor product of state spaces of C
∗-algebras. We show that the minimal and maximal tensor products of state spaces of C
∗-algebras agree precisely when one of the two C
∗-algebras is commutative, which confirms Barker's conjecture in the case where the compact convex sets are state paces of C
∗-algebras.
Further, the Namioka-Phelps tensor product of the trace simplexes of two or more unital C
∗-algebras is shown to be the trace simplex of the (minimal or maximal) tensor product of the C
∗-algebras. This enables a systematic way of determining the trace simplex of a tensor product of C
∗-algebras.
Using the natural duality between linear functionals on tensor products of
C∗-algebras with the trace class operators on a Hilbert space
H and linear maps of the
C∗-algebra into
B(H), we give two characterizations of separability, one relating it to abelianness of the definite set of the map, and one on tensor products of nuclear and UHF
C∗-algebras.
A pair of quantum channels is said to be incompatible if they cannot be realized as marginals of a single channel. This paper addresses the general structure of the incompatibility of completely positive channels with a fixed quantum input space and with general outcome operator algebras. We define a compatibility relation for such channels by identifying the composite outcome space as the maximal (projective) C*-tensor product of outcome algebras. We show theorems that characterize this compatibility relation in terms of the concatenation and conjugation of channels, generalizing the recent result for channels with quantum outcome spaces. These results are applied to the positive operator valued measures (POVMs) by identifying each of them with the corresponding quantum-classical (QC) channel. We also give a characterization of the maximality of a POVM with respect to the post-processing preorder in terms of the conjugate channel of the QC channel. We consider another definition of compatibility of normal channels by identifying the composite outcome space with the normal tensor product of the outcome von Neumann algebras. We prove that for a given normal channel, the class of normally compatible channels is upper bounded by a special class of channels called tensor conjugate channels. We show the inequivalence of the C*- and normal compatibility relations for QC channels, which originates from the possibility and impossibility of copying operations for commutative von Neumann algebras in C*- and normal compatibility relations, respectively.
The operational structure of quantum couplings and entanglements is studied and classified for semi-finite von Neumann algebras. We show that the classical–quantum correspondences, such as quantum encodings, can be treated as diagonal semi-classical (d-) couplings, and the entanglements, characterized by truly quantum (q-) couplings, can be regarded as truly quantum encodings. The relative entropy of the d-compound and entangled states leads to two different types of entropy for a given quantum state: the von Neumann entropy, which is achieved as the maximum of mutual information over all d-entanglements, and the dimensional entropy, which is achieved at the standard entanglement (true quantum entanglement) coinciding with a d-entanglement only in the case of pure marginal states. The d-and q-information of a quantum noisy channel are, respectively, defined via the input d- and q-encodings, and the q-capacity of a quantum noiseless channel is found to be the logarithm of the dimensionality of the input algebra. The quantum capacity may double the classical capacity, achieved as the supremum over all d-couplings (or encodings) bounded by the logarithm of the dimensionality of a maximal Abelian subalgebra.
The operational structure of quantum couplings and entanglements is studied and classified for semi–finite von Neumann algebras. We show that the classical–quantum correspondences, such as quantum encodings, can be treated as diagonal semi–classical (d–) couplings, and the entanglements, characterized by truly quantum (q–) couplings, can be regarded as truly quantum encodings. The relative entropy of the d–compound and entangled states leads to two different types of entropy for a given quantum state: the von Neumann entropy, which is achieved as the maximum of mutual information over all d–entanglements, and the dimensional entropy, which is achieved at the standard entanglement (true quantum entanglement) coinciding with a d–entanglement only in the case of pure marginal states. The d– and q–information of a quantum noisy channel are, respectively, defined via the input d– and q–encodings, and the q–capacity of a quantum noiseless channel is found to be the logarithm of the dimensionality of the input algebra. The quantum capacity may double the classical capacity, achieved as the supremum over all d–couplings (or encodings) bounded by the logarithm of the dimensionality of a maximal Abelian subalgebra.
We consider four norms on tensor product spaces that have appeared in quantum information theory and demonstrate duality relationships between them. We show that the product numerical radius is dual to the robustness of entanglement, and we similarly show that the S(k)-norm is dual to the projective tensor norm. We show that, analogous to how the product numerical radius and the S(k)-norm characterize k-block positivity of operators, there is a natural version of the projective tensor norm that characterizes Schmidt number. In this way we obtain an elementary new proof of the cross norm criterion for separability, and we also generalize both the cross norm and realignment criteria to the case of arbitrary Schmidt number.