Type, cotype and twisted sums induced by complex interpolation (Q1686035)

The outcome of a series of paper of \textit{N. J. Kalton} (mainly, [Mem. Am. Math. Soc. 385, 85 p. (1988; Zbl 0658.46059)] and [Trans. Am. Math. Soc. 333, No. 2, 479--529 (1992; Zbl 0776.46033)]) is the existence of an intimate connection between complex interpolation and twisted sums. Reduced to its bare bones, and cheating just a bit, this relation can be formulated as: complex interpolation pairs \((X_0, X_1)\) of Köthe spaces and twisted sums \(Z(X_\theta)\) of the Banach spaces \(X_\theta= (X_0, X_1)_\theta\) are the same thing. Let us recall that, when we say that \(Z(X_\theta)\) is a twisted sum space, we mean that it contains a subspace isomorphic to \(X_\theta\) and such that \(Z(X_\theta)/X_\theta\) is isomorphic to \(X_\theta\). Even if there still remain many aspects of the correspondence to be clarified, the natural next step is to connect properties of the interpolation scale \((X_\theta)\) and properties of the twisted sums \(Z(X_\theta)\). Two central properties of twisted sum spaces are: triviality (namely, \(X_\theta\) is complemented in \(Z(X_\theta))\) and singularity (namely, \(X_\theta\) is not complemented in any subspace \(E\) of \(Z(X_\theta)\) containing it and such that \(E/X_\theta\) is infinite dimensional). The paper [\textit{J. M. F. Castillo} et al., Trans. Am. Math. Soc. 369, No. 7, 4671--4708 (2017; Zbl 1427.46050)] introduced several tools to perform such a study, always working inside the context of Köthe spaces. The two core techniques are the introduction of different parameters \(M_n(\cdot)\) valid for Köthe spaces and the inequality \[ \left\|\Omega_\theta\left(\sum_{i=1}^n x_i\right)-\sum_{i=1}^n \Omega_\theta(x_i)-\log\frac{M_n(X_0)}{M_n(X_1)}\sum_{i=1}^n x_i \right\|\leq C_\theta M_n(X_0)^{1-\theta}M_n(X_1)^\theta \tag{1} \] connecting the behaviour of the induced twisted sum \(Z(X_\theta)\), represented here by the map \(\Omega_\theta\), and the parameters \(M_n\). The paper under review transplants the tools and results of the paper above to the context of arbitrary interpolation scales. To that end, the author has to do two things: {\parindent=6mm \begin{itemize}\item[(i)] to introduce working parameters making sense for arbitrary Banach spaces and then \item[(ii)] obtain for them inequalities of type (1). \end{itemize}} All this the author does, and moreover makes some of the results work in the context of interpolation families and not only of interpolation pairs. The clever idea behind (i) is to use the finite type/cotype parameters and the clever idea behind (ii) is to state and prove an average form of (1). The last section of the paper contains four applications: while the first two applications, to \(\ell_p\) and \(L_p\) spaces, are standard -- in fact, they are there just to show that the author's methods also work for Köthe spaces -- the second two are to Schatten classes (see also [\textit{J. Suárez de la Fuente}, Rocky Mt. J. Math. 44, No. 6, 2093--2102 (2014; Zbl 1328.46015)]) and to twisted sums of spaces without the compact approximation property.