On positive linear maps between matrix algebras (Q1087962): Difference between revisions

Let \(M_ n\) be the algebra of \(n\times n\) complex matrices. For \(A\in M_ n\), A* denotes the transpose of the complex conjugate of A. Then \(A\in M_ n\) is said to be positive if \(A=A*\) and A has non-negative eigenvalues (more commonly called positive semi-definite). A linear map \(\Phi\) : \(M_ n\to M_ m\) is unital if \(\Phi (I)=I\). \(\Phi\) is called positive if \(\Phi\) (A) is positive for all positive A. Furthermore \(\Phi\) is called completely positive if the block matrix \([\Phi (A_{ij})]^ k_{i,j=1}\) is positive for all positive \([A_{ij}]^ k_{i,j=1}\in M_{nk}\) and for all \(k\geq 1\) and \(\Phi\) is called completely copositive if the same condition holds with \(A_{ij}\) replaced by \(A_{ji}\). A positive linear map is decomposable if it is the sum of a completely positive linear map and a completely copositive linear map. Equivalent by a result of \textit{M. Choi} (ibid. 10, 285-290 (1975; Zbl 0327.15018)) \(\Phi\) is decomposable if and only if there exist \(n\times m\) matrices \(V_ i\) and \(U_ i\) such that \(\Phi (A)=\sum V^*_ iAV_ i+\sum U^*_ jA_ j^{tr}U_ j\) for all \(A\in M_ n\), where \(A^{tr}\) denotes the transpose of A. Choi also showed that not every positive linear map: \(M_ 3\to M_ 3\) is decomposable and \textit{Woronowicz} [Rep. Math. Phys. 10, 165-183 (1976)] showed that there exist positive linear maps: \(M_ 2\to M_ 4\) which are not decomposable but gave no examples. In this paper explicit examples of such maps are constructed. Also Woronowicz has shown that if \(\Phi\) : \(M_ n\to M_ m\) is a unital decomposable positive linear map, then for every \(A,T\in M_ n\) with \(T=T^*\) and \(T-A^*A\) and \(T-A^*A\) positive, \(\Phi (T)-\Phi (A^*)\Phi (A)\) and \(\Phi (T)-\Phi (A)\Phi (A^*)\) are positive and conjectured that every unital positive map has this property. \textit{Kirchberg} [Rep. Math. Phys. 18, 113-116 (1980)] showed that there exist counterexamples to Woronowicz's conjecture but did not give specific counterexamples. In this paper a map \(\Phi\) : \(M_ 4\to M_ 2\) is constructed which is an explicit counterexample.