Putnam-Fuglede theorem and the range-kernel orthogonality of derivations (Q5957245)

For a complex Hilbert space \(H\) and \(A,B\in{\mathcal B}(H)\), denote by \(d_{AB}:{\mathcal B}(H)\to{\mathcal B}(H)\) either of the operators \(E\mapsto AT-TB\) and \(T\mapsto ATB-T\). The classical Putnam-Fuglede theorem says that \(\text{Ker} d_{AB}=\text{Ker} d_{A^*B^*}\), provided both \(A\) and \(B\) are normal. Motivated by this fact, one says that the pair \((A,B)\) has the property \(\text{PF}(d)\) if \(\text{Ker} d_{AB}\subseteq\text{Ker} d_{A^*B^*}\). Moreover, given \(S\in{\mathcal B}(H)\), one says that the pair \((A,B)\) has the property \(\text{PF}(d(S))\) if \(S\in\text{Ker} d_{AB}\cap\text{Ker} d_{A^*B^*}\). In the paper under review, one investigates the relationship between the range-kernel orthogonality of the operator \(d_{AB}\) (in the usual sense in the setting of normed spaces) and the properties \(\text{PF}(d)\) and \(\text{PF}(d(S))\). To this end, one first proves that the pair \((A,B)\) has the property \(\text{PF}(d(S))\) if and only if \(A|S^*|=|S^*|A\), \(B|S|=|S|B\) and \(d_{AB}(U)=0\) (Theorem~2.1), where \(S=U|S|\) is the polar decomposition of \(S\). Some already known results of \textit{T.~Okuyama} and \textit{K.~Watanabe} [Proc. Am. Math. Soc. 126, 2631-2634 (1998; Zbl 0901.47011)] and \textit{B. P.~Duggal} [ibid. 129, 83-87 (2001; Zbl 0958.47015)] are recovered as consequences of this characterization. One then makes use of that characterization to prove the following fact (Theorem 3.6): Assume that \(1<p<\infty\) and \(S=U|S|\) (as above) belongs to the Schatten class \({\mathcal C}_p\). Then the following assertions are equivalent: (i) We have \[ \min\{\|d_{AB}(X)+S\|_0\|d_{A^*B^*}(X)+S\|_p\} \geq\|S\|_p \] for all \(X\in{\mathcal C}_p\). (ii) The pair \((A,B)\) has the property \(\text{PF}(d(S))\). (iii) The pair \((A,B)\) has the property \(\text{PF}(d(U))\) and we have \[ \min\{\|\delta_A(X)+|S^*|\|_0 \|\delta_{A^*}(X)+|S^*|\|_0 \|\delta_B(X)+|S|\|_0 \|\delta_{B^*}(X)+|S|\|_p \}\geq\|S\|_p \] for all \(XX\in{\mathcal C}_p\), where \(\delta_Y(X)=YX-XY\) whenever \(Y\in{\mathcal B}(H)\). We conclude by mentioning that the paper under review is a well-written one, and adds important pieces of information to the circle of ideas related to various extensions of the Putnam-Fuglede theorem started 25 years ago by \textit{G.~Weiss} [Trans. Am. Math. Soc. 246, 193-209 (1978; Zbl 0403.47009)].