Nonlinear centralizers in homology. II: The Schatten classes (Q1982552)

``Most decent Banach (and quasi-Banach) spaces carry natural module structures over a selected group of algebras.'' This is the point of view the author adopts, following the pioneering work of Nigel Kalton, in his studies developed through the papers [\textit{F.~Cabello Sánchez}, Proc. Am. Math. Soc. 142, No.~3, 949--955 (2014; Zbl 1287.47021); \textit{F.~Cabello Sánchez}, Math. Ann. 358, No.~3--4, 779--798 (2014; Zbl 1306.46027), which is Part~I; \textit{F.~Cabello Sánchez}, J. Math. Anal. Appl. 418, No.~1, 317--335 (2014; Zbl 1346.46065); \textit{F.~Cabello Sánchez} et al., Adv. Math. 294, 454--488 (2016; Zbl 1356.46051); \textit{F.~Cabello Sánchez}, J. Noncommut. Geom. 11, No.~4, 1395--1412 (2017; Zbl 1409.46036); \textit{F.~Cabello Sánchez} et al., J. Funct. Anal. 280, No.~4, Article ID 108863, 37 p. (2021; Zbl 1466.46070)], to which we must add the paper under review. Our plan is to describe the contents of the paper and explain why the Kalton-Cabello approach is important. Everything is contained in the motto: Most decent Banach (and quasi-Banach) spaces carry natural module structures over a selected group of algebras. Thus, a study of the existence and behaviour of exact sequences $0\longrightarrow Y \longrightarrow Z\longrightarrow X\longrightarrow 0$ of (quasi) Banach spaces, which is what is usually called (quasi) Banach space homology, will be enriched, completed or possible only after involving those additional structures. The simplest structure is the standard $\mathbb C$-module structure that stems from the fact of being a complex (quasi) Banach space. Beyond that, spaces with unconditional basis (say, $\ell_p$) have an additional $\ell_\infty$-module structure, Köthe spaces over a measure space $S$ (say $L_p$) enjoy an additional $L_\infty(S)$-module structure, most function spaces on a compact $K$, say $C[0,1]$, have an additional $C(K)$-module structure, and standard operator ideals on a Hilbert space $\mathcal H$, say the Schatten classes $\mathfrak S_p$, have an additional $\mathfrak L(\mathcal H)$-module structure. The behaviour of exact sequences is (or can be) different depending on the category one is considering. For instance, an exact sequence $0\longrightarrow \ell_2 \longrightarrow Z\longrightarrow \ell_2\longrightarrow 0$ in the category of Banach $\mathbb C$-modules is not necessarily an exact sequence of $\ell_\infty$-modules. And, even if it is, it could happen (although it does not) that the sequence is trivial as (quasi) Banach spaces but not as $\ell_\infty$-modules. Let us present the classical example, a combination of [\textit{N. J. Kalton} and \textit{N. T. Peck}, Trans. Am. Math. Soc. 255, 1--30 (1979; Zbl 0424.46004); \textit{N. J. Kalton}, Nonlinear commutators in interpolation theory. Providence, RI: American Mathematical Society (AMS) (1988; Zbl 0658.46059); \textit{N. J. Kalton}, Trans. Am. Math. Soc. 333, No. 2, 479--529 (1992; Zbl 0776.46033); \textit{F. Cabello Sánchez}, Math. Ann. 358, No. 3--4, 779--798 (2014; Zbl 1306.46027)]: nontrivial sequences $0\longrightarrow \ell_q \longrightarrow Z\longrightarrow \ell_p\longrightarrow 0$ in the category of quasi Banach spaces exist for all values $0<p,q\leq \infty$ and for all values $1<p$; and for $q<\infty$ in the category of Banach spaces. However, exact sequences of $\ell_\infty$-modules are only possible when $p=q$. Let us call $\mathrm{Ext}_A(X,Y)$ the space of exact sequences $0\longrightarrow Y \longrightarrow Z\longrightarrow X\longrightarrow 0$ in the category of $A$-quasi Banach modules. The richer structure is reflected here in a surprising fact: $\mathrm{Ext}_{\ell_\infty}(\ell_p, \ell_p)= \mathrm{Ext}_\mathbb C(\ell_1, \mathbb C)$, independently of $p$! The paper under review extends the study to exact sequences of Schatten classes $\mathfrak S_p$ as $\mathfrak L(\mathcal H)$-modules by showing that \[ \mathrm{Ext}_{\mathfrak L(\mathcal H)} (\mathfrak S_p, \mathfrak S_q) = \begin{cases} 0, &\text{ if }0 < q < p \leq \infty \text{ or }p = q = \infty; \\ \mathrm{Ext}_{\mathbb C} (\mathfrak S_1, \mathbb C), &\text{ if } q = p <\infty; \\ \mathrm{Ext}_{\mathbb C}(\mathcal H, \mathcal H), &\text{ if } 0 <p<q \leq \infty. \end{cases} \] All this is startling since one could have easily believed that the more rigid the structure, the more rigid becomes the structure of exact sequences. However, things are not so: nontrivial sequences $0\longrightarrow \mathfrak S_3 \longrightarrow \diamondsuit \longrightarrow \mathfrak S_2\longrightarrow 0$ of $\mathfrak L(\mathcal H)$-modules exist, say, while no nontrivial sequences $0\longrightarrow L_3 \longrightarrow \diamondsuit \longrightarrow L_2\longrightarrow 0$ of $L_\infty$-modules exist. Passing to techniques, the reader expecting categorical mumbo jumbo will be surprised. The paper is a lesson on hard analysis about vector valued functions, just as [\textit{F. Cabello Sánchez}, Math. Ann. 358, No. 3--4, 779--798 (2014; Zbl 1306.46027)]. Why? Because, and this is the beautiful discovery of [\textit{N. J. Kalton} and \textit{N. T. Peck}, Trans. Am. Math. Soc. 255, 1--30 (1979; Zbl 0424.46004); \textit{N. J. Kalton}, Nonlinear commutators in interpolation theory. Providence, RI: American Mathematical Society (AMS) (1988; Zbl 0658.46059); \textit{N. J. Kalton}, Trans. Am. Math. Soc. 333, No. 2, 479--529 (1992; Zbl 0776.46033)], then continued in [\textit{F. Cabello Sánchez} et al., Adv. Math. 294, 454--488 (2016; Zbl 1356.46051)], the space $\mathrm{Ext}_A(X,Y)$ corresponds to a space of homogeneous functions, called $A$-centralizers. And that is the true object of study in this paper.