Descriptive compact spaces and renorming (Q4821083)

A compact topological space \((K,\tau)\) is \textit{fragmentable} by a certain metric \(d\) on \(K\) if for every \(\varepsilon>0\) and for every non-empty \(A\subset K\), there exists \(U\in\tau\) such that \(A\cap U\not=\emptyset\) and diam\((A\cap U)<\varepsilon\). This concept is due to \textit{J.~E.\ Jayne} and \textit{C.~A.\ Rogers} [Acta Math.\ 155, 41--79 (1985; Zbl 0588.54020 )] and has been extensively studied. A subclass \(\mathcal D\) of the former consists of the so-called \textit{descriptive compacta}, formed by the compact spaces \(K\) with a \(\sigma\)-isolated network, i.e., a family \(\mathcal H:=\{ H_i:i\in I \}\) of subsets of \(K\) such that \(I=\bigcup_{n\in\mathbb{N}} I_n\), where \(\{ H_i,i\in I_n \}\) is isolated, and such that every \(U\in\tau\) is an union of members of \(\mathcal H\). This class, in a more general context, has been introduced by [\textit{R.~W.\ Hansell}, Serdica Math.\ J.\ 27, No.~1, 1--66 (2001; Zbl 0982.46012)]. In this very interesting paper, the authors study the Banach spaces \(X\) having a linearly dense descriptive \(\sigma(X,F)\)-compact subset \(K\), where \(F\subset X^*\) is some total subspace. The main theorem says that \(X\) admits a coarser \(\sigma(X,F)\)-locally uniformly rotund norm (LUR) and the topology \(\sigma(X,F)\) admits a \(\sigma\)-isolated network. If, moreover, \(K\) is fragmented by the metric \(p_F\) of uniform convergence on \(B_{X^*}\cap F\), then \(X\) admits a coarser norm which is \(p_F\)-LUR and \(\sigma(X,F)\)-lsc. The norming character of \(F\) is equivalent to the fact that the previous norms can be taken equivalent to the original one. It is worth observing that this result contains the celebrated LUR renorming theorem of WCG Banach spaces due to Troyanski and covers also the case of dual spaces where \((B_{X^*},w^*)\) is descriptive. As a corollary, they obtain that in the setting of the main theorem, \(X\) admits an equivalent rotund norm. Byproducts of the techniques used give, for example, that a compact space is descriptive iff it embeds as a \(w^*\)-compact subset of a \(w^*\)-LUR dual Banach space. The authors give a systematic study of properties (some new and some already known) of descriptive compacta and notice that a compactum built by \textit{S.~Argyros} and \textit{S.~Mercourakis} [Rocky Mt.\ J.\ Math.\ 23, No.~2, 395--446 (1993; Zbl 0797.46009)] is a Corson and Rosenthal descriptive compactum which is not Gul'ko. Still open is the question of knowing if a Corson fragmentable compactum should be descriptive, something true if the fragmentic metric is lsc.