Index of Hadamard multiplication by positive matrices. II (Q5946197)

[For part I see \textit{D. Stojanoff}, ibid. 290, No. 1-3, 95-108 (1999; Zbl 0937.15016).] The authors study some problems of the index theory of conditional expectations [see \textit{S. Popa}, Classification of subfactors and their endomorphisms, AMS-CBMS, No. 86, Providence (1995; Zbl 0865.46044) and its references]. For each \(n{\times}n\) positive semidefinite matrix \(A\) the authors define the minimal index \(I(A)= \max\{\lambda \geq 0: A\circ B \succeq \lambda B\) for all \(B \succeq 0\}\) and, for each norm \(N\), the \(N\)-index \(I_{N}(A) = \min\{N(A\circ B): B \succeq 0\) and \(N(B)=1\}\), where \(A{\circ}B=[a_{ij}b_{ij}]\) is the Hadamard or Schur product of \(A=[a_{ij}]\) and \(B=[b_{ij}]\) and \(B \succeq 0\) means that \(B\) is a positive semidefinite matrix. They investigate a comparison between these indices for different choices of the norm \(N\). As an application they find, for each bounded invertible selfadjoint operator \(S\) on a Hilbert space, the best constant \(M(S)\) such that \(\|STS+S^{-1}TS^{-1}\|\geq M(S)\|T \|\) for all \(T \succeq 0\).