Lifting convex approximation properties from Banach spaces to their dual spaces and the related local reflexivity (Q2634150)

A Banach space \(X\) is \(\lambda\)-extendably locally reflexive (\(\lambda\)-ELR) if for all finite-dimensional subspaces \(E\) of \(X^{**}\) and \(F\) of \(X^*\), and all \(\varepsilon > 0\), there exists a bounded linear operator \(T\) on \(X^{**}\) such that \(T(E) \subset X\), \(\|T\| \leq \lambda + \varepsilon\), and \(x^*(Tx^{**}) = x^{**}(x^*)\) for all \(x^* \in F\) and \(x^{**} \in E\). \textit{W. B. Johnson} and \textit{T. Oikhberg} [Ill. J. Math. 45, No. 1, 123--137 (2001; Zbl 1004.46008)] proved that, if \(X\) is \(\lambda\)-ELR and has the \(\mu\)-bounded approximation property, then \(X^*\) has the \(\mu \lambda\)-bounded approximation property. The proof used the principle of local reflexivity. The question arises of how to lift other approximation properties. Let \(A\) be a convex subset of the bounded linear operators on \(X\) containing \(0\). Then \(X\) has the \(\lambda\)-bounded \(A\)-approximation property if the identity can be approximated uniformly on compact subsets by \(\lambda\)-bounded operators from \(A\). Suitable versions of the \(\lambda\)-ELR and the principle of local reflexivity are introduced and the authors give a unified approach to the lifting of these convex approximation properties from a Banach space to its dual.