A class of generalized positive linear maps on matrix algebras (Q2435494)

The authors investigate properties of a class of positive linear maps on matrix algebras over the complex numbers. The parameters of the construction of this class are positive real numbers and a permutation. The authors find conditions on these parameters which ensure that the corresponding positive linear map is atomic, decomposable, or completely positive. These notions refer to certain properties when a linear map on a matrix algebra is extended to a tensor product by application to every component. A map is called completely positive if this extension is again positive for every tensor product with any other matrix algebra over the complex numbers. The other notions refer to the possibility to either decompose the linear map into sums of maps fulfilling certain positivity requirements or not. Under certain conditions on the parameters, complete positivity of a map of this class is characterized. The authors give applications in quantum information theory. For a non-zero positive linear map its structural physical approximation is considered and if it is not completely positive also its entanglement witness and when it is optimal. For some conditions it is shown when the structural physical approximation of a map of this class is separable, i.e., can be decomposed into a sum of tensor products of matrices. For certain other conditions it is shown that the entanglement witness of the transpose of a map from this class is optimal.