Renormings concerning the lineability of the norm-attaining functionals (Q313517): Difference between revisions

Norm attaining functions have received high levels of attention for over 50 years, dating at least from the remarkable work of \textit{E. Bishop} and \textit{R. R. Phelps} [Bull. Am. Math. Soc. 67, 97--98 (1961; Zbl 0098.07905)]. In the last 15 years, interest in lineability has increased markedly (see, e.g., [\textit{R. M. Aron} et al., Lineability. The search for linearity in mathematics. Boca Raton, FL: CRC Press (2016; Zbl 1348.46001)]). It was somehow natural that attention should be directed at the ``intersection'' of these two areas. Among the early contributors to this area is \textit{G. Godefroy}, who posed in [Extr. Math. 16, No. 1, 1--25 (2001; Zbl 0986.46009)] the following question: Let \(X\) be a Banach space and let \(NA(X) = \{ \varphi \in X^* \mid \varphi \text{ attains its norm}\}\). Is \(NA(X)\) lineable; that is, does \(NA(X)\) contain an infinite dimensional vector space? Basing his work on a construction of \textit{C. J. Read} [``Banach spaces with no proximinal subspaces of codimension 2'', Preprint, \url{arXiv:1307.7958}, Isr. J. Math. 223, 493--504 (2018; Zbl 1397.46010)], \textit{M. Rmoutil} [``Norm-attaining functionals and proximinal subspaces'', Preprint, \url{arXiv:1503.06112}, J. Funct. Anal. 272, No.~3, 918--928 (2017; Zbl 1359.46008)] has constructed a Banach space \(X\) such that \(NA(X)\) does not even contain a \(2\)-dimensional vector space. With this in mind, it is natural to look for situations in which the answer to Godefroy's question [loc. cit.] is affirmative, and this is the main contribution of this paper. To explain it, one needs the notion of filling subspaces: For \(V \subset \ell_\infty(\Lambda)\), let \(\mathrm{supp}(V) :\equiv \bigcup \{\mathrm{supp}(v) \mid v \in V\}.\) The subspace \(V\) is said to be filling if for every infinite subset \(A \subset \mathrm{supp}(V)\) there exists a unit vector \(x \in V\) such that \(\mathrm{supp}(x) \subset A\) and \(x\) attains its (sup-)norm. The authors note that every filling subspace \(V \subset \ell_\infty(\Gamma)\) is such that the set of its norm-attaining functionals contains a subspace of cardinality equal to that of \(\mathrm{supp}(V)\). One main result is the following theorem: For any Banach space \(X,\) there exists a biorthogonal system \((x_i, x_i^*)_{i \in I}\) such that \(\Lambda = \{x_i^* \mid i \in I\}\) is norming if and only if \(X\) is linearly isometric to a filling subspace of \(\ell_\infty(V)\). If this occurs, then \(NA(X)\) contains a vector space of dimension \(\mathrm{card}(\Lambda)\).