A series of absolutely indecomposable positive maps in matrix algebras (Q1801446): Difference between revisions

Consider the set \(P(M_ n)\) of all positive (semidefinite) linear maps of the matrix algebra \(M_ n(\mathbb{C})\) into itself with the usual inner product. A linar map \(\tau\) of \(M_ n(\mathbb{C})\) is called ``\(k\)- positive'', if it is positive, and the induced mapping on \(M_ k\bigl(M_ n(\mathbb{C})\bigr)\) defined by \(\bigl[a_{ij}]\mapsto[\tau(a_{ij})\bigr]\) is also positive. Similarly, \(\tau\) is called ``\(k\)-copositive'', if \(\tau\) is positive and the induced mapping \([a_{ij}]\mapsto\bigl[\tau(a_{ji})\bigr]\) is positive. It has been shown that for \(n=2\), each positive map \(\tau\) can be written as a sum of two maps, one of which is \(k\)-positive for all \(k\) and the other \(k\)-copositive for all \(k\); but this is not true for \(n=3\) [see \textit{S. L. Woronowicz}, Rep. Math. Phys. 10, 165-183 (1976; Zbl 0347.46063) and \textit{M. D. Choi}, Linear Algebra Appl. 12, 95-100 (1975; Zbl 0336.15014)]. In the present paper the author constructs, for each \(n\geq 4\), an example of a positive map \(\tau\) which is not the sum of a 2-positive map and a 2- copositive map.