\(L\)-orthogonality in Daugavet centers and narrow operators (Q2235875)

Let \(X, Y\) be Banach spaces, \(L(X, Y)\) be the space of all bounded linear operators acting from \(X\) to \(Y\). A nonzero operator \(G \in L(X, Y)\) is a \textit{Daugavet center}, if \(\|G + T\| = \|G\| + \|T\|\) for every \(T \in L(X, Y)\) of \(\operatorname{rank} T = 1\). A Banach space \(X\) has the \textit{Daugavet property} if the identity operator \(\mathrm{Id} \in L(X, Y)\) is a Daugavet center. An element \(u \in S_{X^{**}}\) is called \textit{\(L\)-orthogonal} to \(X\) if \(\|x + u\| = 1 + \|x\|\) for every \(x \in X\). In [\textit{A. Rueda Zoca}, Banach J. Math. Anal. 12, No. 1, 68--84 (2018; Zbl 1391.46017)] the following connection between these concepts was established: a separable space \(X\) has the Daugavet property if and only if the set of elements \(L\)-orthogonal to \(X\) is \(w^*\)-dense in \(S_{X^{**}}\). This result on abundance of $L$-orthogonal elements extends to spaces \(X\) with the Daugavet property of \(\operatorname{dens}(X) =\omega_1\), but does not extend to spaces of large density characters [\textit{G.~López-Pérez} and \textit{A.~Rueda Zoca}, Adv. Math. 383, Article ID 107719, 17~p. (2021; Zbl 1473.46013)]. The paper under review extends the above results to Daugavet centers and elucidates the role of \(L\)-orthogonal elements for the theory of narrow operators (a class of operators closely related to the Daugavet property). It is shown that if \(\operatorname{dens}(Y) \leqslant \omega_1\) and \(G \in S_{L(X, Y)}\) is a Daugavet center with separable range then, for every non-empty reatively \(w^\ast\)-open subset \(W \subset B_{X^{\ast \ast}}\), its image \(G^{\ast \ast}(W)\) contains some element \(L\)-orthogonal to \(Y\). For a separable \(X\) with the Daugavet property and a narrow operator \(T \in L(X, Y)\), it is demonstrated that for every \(y \in B_X\) and non-empty \(w^\ast\)-open subset \(W \subset B_{X^{\ast \ast}}\) there is an element \(u \in W\) which is \(L\)-orthogonal to \(X\) such that \(T^{\ast \ast}(u) = T(y)\). In the particular case of separable \(T^\ast( Y^\ast)\) the previous result extends to \(\operatorname{dens}(X) = \omega_1\), but none of the previous results hold true in large density characters, in particular, there is a counterexample for \(\omega_2\) under the set-theoretic assumption \(2^c = \omega_2\).