On the category of quotient Banach spaces after Wegner (Q2161113): Difference between revisions

The research in this paper stems from Section 9 of [\textit{J. M. F. Castillo}, Extr. Math. 37, No.~1, 1--56 (2022; Zbl 1501.46060)], entitled ``The heart of Banach spaces'', namesake of \textit{S.-A. Wegner}'s paper [J. Pure Appl. Algebra 221, No.~11, 2880--2909 (2017; Zbl 1377.46053)]. The core problem is that the category of Banach space operators (linear continuous maps) is not abelian, which, in a slightly simplified formulation, means that an injective (monic) and surjective (epic) map is not necessarily an isomorphism. This sets an obstacle when one attempts to make diagrammatic reasonings and introduces a disruptive element in every categorical reasoning one intends to perform in Banach spaces. On the other hand, categorical reasonings can indeed be performed in Banach spaces, one way or another. And this is the conundrum: why? A way of tackling the problem is making a full embedding of Banach spaces in a bigger abelian category (``full'' means that no new morphisms appear in the new category while working with standard Banach spaces). Such a bigger category is called a heart; and there are, in principle, many possible constructions for hearts of Banach spaces, from the first try in \textit{L. Waelbroeck}'s work [Banach Cent. Publ. 8, 553--562 (1982; Zbl 0492.46012); Bornological quotients. With the collaboration of Guy Noël. Bruxelles: Académie Royale de Belgique (2005; Zbl 1084.46001)] to the Clausen-Scholze try [\textit{P. Scholze}, Lectures on analytic geometry, \url{www.math.uni-bonn.de/people/scholze/Analytic.pdf}]. The authors focus on the Waelbroeck construction after Wegner, with a twist of their own: the objects in the new category are injective operators between Banach spaces and a morphism from \(f: X'\to X\) to \(g: Y'\to Y\) is a commutative diagram \[ \begin{tikzcd} X'\ar[r, "{u'}"] \ar[d, "f" '] & Y' \ar[d, "g"] \\ X\ar[r, "{u}" '] & Y \end{tikzcd}\tag{1} \] The equality notion for these morphisms is as follows: a morphism is trivial if some operator \(r:X\to Y'\) exists making the diagram \[ \begin{tikzcd} X'\ar[r, "{u'}"] \ar[d, "f" '] & Y' \ar[d, "g"] \\ X\ar[r, "{u}" '] \ar[ur, "r"] & Y \end{tikzcd} \] commutative. Now comes the localization with respect to a type of morphisms that Wegner called \textit{pulations}: those for which the diagram (1) is simultaneously a pullback and pushout diagram. The resulting category is the Waelbroeck-Wegner category \(\mathbf{W}\), a heart of Banach spaces. In this setting, the algebraical half of the team of authors shows that: \begin{itemize} \item Every object of \(\mathbf{W}\) is isomorphic to one of the form \(L\to \ell_1(I)\) for some set \(I\). \item Given two Banach spaces \(X,Y\), \(\mathrm{Ext}_{\mathbf {Ban}}^n(X,Y) = \mathrm{Ext}_{\mathbf W}^n(X,Y)\) for all \(n\geq 1\). \item \(\mathbf W\) is a colimit; i.e., the obvious Abelianization functor \(\mathbf{Ban}\to \mathbf W\) has the required universal property. \end{itemize} On the other hand, the Banach space half of the team provides a plethora of extremely nice applications and transpositions of results to and from Banach spaces to \(\mathbf{W}\), such as: \begin{itemize} \item An object of \(\mathbf{W}\) is projective if and only if isomorphic to some \(\ell_1(I)\) for some set \(I\). \item \(\mathbf{W}\) has enough projectives and no injectives (except \(0\)). \item A \(\mathbf{W}\)-version of the Lindenstrauss-Rosenthal theorem. \item Rambling considerations about Hilbert spaces or strictly singular objects in \(\mathbf{W}\). \end{itemize} As a whole, the paper is a must for people who feel the frame of Banach spaces too narrow when attacking some of the difficult problems in the theory and need to unearth new ideas.