Hilbertian matrix cross normed spaces arising from normed ideals (Q2267685)

Motivated by the description of the matrix norms in the interpolating spaces between the row and the column operator spaces \(R\) and \(C\) by the complex interpolation method given in [\textit{G.\,Pisier}, ``The operator Hilbert space \(OH\), complex interpolation and tensor norms'' (Mem.\ Am.\ Math.\ Soc.\ 585) (1996; Zbl 0932.46046)], the author introduces a Hilbertian matrix cross normed space \(H(\Phi,\Psi)\) associated with a pair of symmetric norming functions \((\Phi,\Psi)\) with \(\Phi \geq \Psi\), where for a separable Hilbert space \(H\) with an orthonormal basis \((\xi_i)\), the matrix norm of an operator \(T = \sum \xi_i \otimes T_i \in H \otimes M_n\) is defined as \(\| T \|_{H(\Phi,\Psi)}= \sup_x \frac{\|T_i x T_i^\ast \|_\Psi}{\| x \|_\Phi}\). This construction indeed gives a matrix cross normed space (that is, \(\| x \otimes A\|_n=\| x \|\, \|A \|_{M_n}\) for all \(x \in H(\Phi,\Psi)\) and \(A \in M_n\)), but not necessarily an operator space (an example using the KyFan \(2\)-norm is given). The author proves a necessary condition for this to happen. For the case when \(\Phi=\Psi\), this condition implies that \(\Phi\) is a cross norm (that is, \(\| x \otimes y\|=\|x\|\,\|y\|\) for all \(x,y \in H(\Phi,\Phi)\)), and thus, as the author shows in the final section without any further use of operator space techniques, \(\Phi\) has to coincide with some Schatten \(p\)-norm (\(1 \leq p \leq \infty\)). The author also provides a sufficent condition: If \(\Phi\) is the \(2\)-convexification of a symmetric norming function and \(\Psi\) the adjoint of such a \(2\)-convexification, then \(H(\Phi,\Psi)\) is an operator space. In general, a pair of symmetric normed ideals that are not necessarily the Schatten class or satisfy the above sufficient condition may still give rise to an operator space. The author studies the space of completely bounded mappings between the matrix cross normed spaces obtained in this way and shows that the multiplicator norm naturally appears as the completely bounded norm.