The approximation properties determined by operator ideals (Q526792)

Let \({\mathcal{F}}\) denote the operator ideal of finite-rank bounded linear operators. Recall that a Banach space \(X\) has the approximation property (AP) if \(\text{id}_X\in \overline{{\mathcal{F}}(X,X)}^{\tau_c}\), where \(\text{id}_X\) is the identity map on \(X\) and \(\tau_c\) is the topology of uniform convergence on compact subsets of \(X\). Let \(\mathcal{A}\) be an operator ideal. Following \textit{J. M. Delgado} and \textit{C. Piñeiro} [Stud. Math. 214, No. 1, 67--75 (2013; Zbl 1432.46008)], \(X\) is said to have the AP with respect to \({\mathcal{A}}\) (\(\text{AP}_{\mathcal{A}}\)) if \(\text{id}_X\in \overline{{\mathcal{F}}(X,X)}^{\tau_c({\mathcal{A}})}\), where \(\tau_c({\mathcal{A}})\) is the topology of uniform convergence on \({\mathcal{A}}\)-compact subsets of \(X\). Recall that a closed subset \(K\) of \(X\) is \({\mathcal{A}}\)-compact if there exist a Banach space \(Z\), an operator \(T\in {\mathcal{A}}(Z,X)\), and a compact subset \(M\) of \(Z\) such that \(K\subset T(M)\). This concept is due to \textit{B. Carl} and \textit{I. Stephani} [Math. Nachr. 119, 77--95 (1984; Zbl 0575.47015)]. Delgado and Piñeiro proved in [loc. cit.] that \(X\) has the \(\text{AP}_{\mathcal{A}}\) whenever \(X^{**}\) the \(\text{AP}_{\mathcal{A}}\). The authors introduce the notion of the right AP with respect to \({\mathcal{A}}\) (right \(\text{AP}_{\mathcal{A}}\)). Note that by [\textit{A. Lissitsin}, Stud. Math. 211, No. 3, 199--214 (2012; Zbl 1275.46009), Theorem 2.4] this notion coincides with the AP whenever \(\mathcal A \supset \mathcal K\). Relying on the Lima-Nygaard-Oja isometric version of the Davis-Figiel-Johnson-Pełczyński factorization lemma, the authors refine this result: if \(X^*\) has the \(\text{AP}_{\mathcal{A}}\), then \(X\) has the right \(\text{AP}_{{\mathcal{A}}^{\text{dual}}}\), and if \(X^*\) has the right \(\text{AP}_{\mathcal{A}}\), then \(X\) has the \(\text{AP}_{{\mathcal{A}}^{\text{dual}}}\). Now, let \({\mathcal{A}}\) be a Banach operator ideal and let \(\lambda\geq 1\). Following Lassalle, Oja, and Turco [\textit{S. Lassalle} et al., J. Approx. Theory 205, 25--42 (2016; Zbl 1343.46013)], \(X\) is said to have the weak \(\lambda\)-bounded AP for \({\mathcal{A}}\) (weak \(\lambda\)-BAP for \({\mathcal{A}}\)) if for every Banach space \(Y\) and for each operator \(T\in {\mathcal{A}}(X,Y)\), there exists a net \((S_\alpha)_\alpha\) in \({\mathcal{F}}(X,X)\) such that \(\sup_\alpha\| TS_\alpha\|_{{\mathcal{A}}}\leq\lambda\| T\|_{{\mathcal{A}}}\) and \(S_\alpha \longrightarrow \text{id}_X\) in the strong operator topology. The authors prove some results which are similar or symmetric to those from Lassalle et al. [loc. cit.], using similar methods of proof. For the symmetric versions, they introduce the notion of the left weak \(\lambda\)-BAP for \({\mathcal{A}}\). \textit{S. Lassalle} and \textit{P. Turco} [J. Funct. Anal. 265, No. 10, 2452--2464 (2013; Zbl 1298.47031)] studied systematically \({\mathcal{A}}\)-compact sets and related concepts basing on a clever way to measure the size of \({\mathcal{A}}\)-compact sets and to make the \({\mathcal{A}}\)-compact operators a Banach operator ideal. The authors extend the notion of the \({\mathcal{A}}\)-compactness to the \({\mathcal{A}}\)-\(p\)-compactness, \(1\leq p\leq\infty\), by replacing the compact subsets with the \(p\)-compact subsets. Then the \({\mathcal{A}}\)-compactness is the same as the \({\mathcal{A}}\)-\(\infty\)-compactness. Correspondingly, they prove some results which generalize results of Lassalle and Turco [loc. cit.], using similar methods of proof. The reviewer would also suggest to look at [\textit{A. Lissitsin}, Arch. Math. 105, No. 2, 163--171 (2015; Zbl 1333.46016)] for a more general approach that uses systems of seminorms on spaces of operators. In particular: {\parindent=6mm \begin{itemize}\item[--] Lemma 2.2 follows from the well-known equalities \[ \mathcal A \circ \mathcal K = \mathcal A \circ \mathcal K \circ \mathcal K = \mathcal A \circ \text{Op}(\mathsf{SepRefl}) \circ \mathcal K \] because \(\tau_c\) is the topology induced by seminorms of right composition with operators from \(\mathcal K\), cf. [loc. cit., Proposition 4] (here \(\mathsf{SepRefl}\) denotes the space ideal of separable reflexive spaces). \item[--] The necessity of Theorem 2.5 follows from [loc. cit., Proposition 5] applied for \(\mathcal A \circ \mathcal K\) (and the well-known fact that \(\mathcal F(X^\ast) \subset \overline{\mathcal F(X)^a}^{\tau_c}\)). \end{itemize}} The reviewer thanks Aleksei Lissitsin for the above details.