Strong subdifferentiability and local Bishop-Phelps-Bollobás properties (Q2293207)

The paper under review deals with some Bishop-Phelps-Bollobás properties of local nature in the context of multilinear mappings. Recall the following definitions: Let \(X_1,\dots, X_N, Y\) be Banach spaces and denote by \(L(X_1,\dots, X_N;Y)\) the space of \(N\)-linear mappings \(A:X_1\times\cdots \times X_N\longrightarrow Y\). According to [\textit{S. Dantas} et al., J. Math. Anal. Appl. 468, No. 1, 304--323 (2018; Zbl 1412.46022), Definition 2.1]: (1) It is said that \((X_1,\dots, X_N; Y)\) has the local Bishop-Phelps-Bollobás point property for \(N\)-linear mappings (denoted in short \(L_{p,p}\)) if, given \(\varepsilon>0\) and \((x_1,\dots, x_N)\in S_{X_1}\times\ldots \times S_{X_N}\), there exists \(\eta(\varepsilon, (x_1,\dots, x_N))>0\) such that, whenever \(A\in L(X_1,\dots, X_N; Y)\) with \(\Vert A\Vert=1\) satisfies that \[\Vert A(x_1,\dots, x_N)\Vert>1-\eta(\varepsilon, (x_1,\dots, x_N)),\] then there exists a norm-one element \(B\in L(X_1,\ldots, X_N;Y)\) such that \[\Vert B(x_1,\ldots, x_N)\Vert=1\mbox{ and }\Vert B-A\Vert<\varepsilon.\] (2) It is said that \((X_1,\dots, X_N; Y)\) has the local Bishop-Phelps-Bollobás operator property for \(N\)-linear mappings (denoted in short \(L_{o,o}\)) if, given \(\varepsilon>0\) and a norm-one element \(A\in L(X_1,\dots, X_N; Y)\), there exists \(\eta(\varepsilon,A)>0\) such that, whenever \((x_1,\dots, x_N)\in S_{X_1}\times\ldots\times S_{X_N}\) satisfies \[\Vert A(x_1,\dots, x_N)\Vert>1-\eta(\varepsilon, A),\] then there exists an element \((z_1,\dots, z_N)\in S_{X_1}\times\ldots\times S_{X_N}\) such that \[\Vert A(z_1,\dots, z_N)\Vert=1\mbox{ and }\max\limits_{1\leq j\leq N}\Vert x_j-z_j\Vert<\varepsilon.\] On the one hand, the above mentioned properties have a uniform counterpart, namely, the ``Bishop-Phelps-Bollobás point property for \(N\)-linear mappings'' (BPBpp) and the ``Bishop-Phelps-Bollobás operator property for \(N\)-linear mappings'' (BPBop), which are defined in their respective case taking the function \(\eta\) only depending on the positive \(\varepsilon\). On the other hand, notice that all the above mentioned properties are motivated by the classical Bishop-Phelps-Bollobás theorem, its Banach space property version for bounded operators or for multilinear mappings. Finally, notice that both \(L_{o,o}\) and \(L_{p,p}\) are very restrictive properties. As a matter of fact, notice that, if \((X_1,\dots, X_N;Y)\) satisfies the \(L_{o,o}\)-property, then every element of \(L(X_1,\dots, X_N;Y)\) attains its norm and, if one considers multilinear mappings of the form \(A(x_1,\dots, x_N)=x_1^*(x_1)x_2^*(x_2)\dots x_N^*(x_N)\) for \(x_i^*\in X_i^*\), then it follows that every element of \(X_i^*\) attains its norm for every \(i\), which implies that \(X_i\) is reflexive by the classical James theorem. For this reason, the appearance of quite restrictive Banach spaces properties implying reflexivity, such as uniform convexity, is rather natural in this context. Coming back to the content of the paper under review, after some preliminary results connecting the properties \(L_{o,o}, L_{p,p}\), BPBpp and BPBop of \((X_1,\dots, X_N,Y)\) and \((X_1,\dots, X_N,\mathbb K)\), where \(\mathbb K\) stands for the scalar field, one of the main results of the paper (Theorem 2.7) establishes conditions under which the pair \((\ell_p,\ell_q,\mathbb K)\) has or does not have the properties \(L_{p,p}\) and/or \(L_{o,o}\), namely: \begin{itemize} \item[(a)] If \(2<p,q<\infty\), then \((\ell_p,\ell_q,\mathbb K)\) satisfies \(L_{p,p}\). \item[(b)] If \(1<p,q<\infty\), then \((\ell_p,\ell_q,\mathbb K)\) satisfies \(L_{o,o}\) if, and only if, \(pq>p+q\). \end{itemize} As a curious consequence of the above mentioned Theorem 2.7, the authors obtain in Corollary 2.8 that, if \(2<p,q<\infty\), then the norm of \(\ell_p\widehat{\otimes}_\pi \ell_q\) is strongly subdifferentiable at every point of \(S_{\ell_p\widehat{\otimes}_\pi \ell_q}\). In Section 3, the authors study the properties \(L_{p,p}\) and \(L_{o,o}\) when dealing with the space \(\ell_1^N\). To be more precise, the main results in this section are: \begin{itemize} \item[(1)] (Theorem 3.3) \((\ell_1^N,Y;\mathbb K)\) has the \(L_{p,p}\)-property if, and only if, the norm of \(Y\) is strongly subdifferentiable. \item[(2)] (Theorem 3.6) If \(Y\) has the approximate hyperplane series property (in particular, if \(Y\) is finite dimensional or uniformly convex), then \((\ell_1^N,Y)\) has the \(L_{p,p}\)-property. \end{itemize}