A Kadec-Pelczyński dichotomy-type theorem for preduals of JBW\(^\ast\)-algebras (Q891073)

A bounded sequence in the Banach space \(L_1[0;1]\) admits a subsequence that can be written as the sum of a disjointly supported sequence and a uniformly integrable (or, equivalently, weakly convergent) sequence. This useful splitting into two sequences of opposite nature is known as the Ka\-dec-{Peł\-czyńs\-ki}-Ro\-sen\-thal splitting lemma. (It holds in \(L_p[0;1]\), \(p<\infty\), too, but the paper under review naturally has to restrict to the case \(p=1\).) \textit{N. Randrianantoanina} [Math. Proc. Camb. Philos. Soc. 132, No. 1, 137--154 (2002; Zbl 1037.46055)] and, with a different proof, \textit{Y. Raynaud} and \textit{Q.-H. Xu} [J. Funct. Anal. 203, No. 1, 149--196 (2003; Zbl 1056.46056)] generalized the splitting lemma to preduals of von Neumann algebras and the authors of the paper under review succeed in generalizing it to preduals of JBW\(^*\)-algebras. The result reads: If \((\varphi_n)\) is a bounded sequence in the predual \(M_*\) of a JBW\(^*\)-algebra \(M\), then there exist a subsequence \((\varphi_{n_k})\), a weakly convergent sequence \((\xi_k)\) and a sequence \((\psi_k)\) of pairwise orthogonal elements in \(M_*\) such that \(\varphi_{n_k}=\xi_k+\psi_k\); moreover, the \(\psi_k\) can be written \(\psi_k=\varphi_{n_k}P_2(p_k)\), where the \(p_k\) are pairwise orthogonal projections in \(M\) with their corresponding 2-Peirce projections \(P_2(p_k)\). The authors rather follow Randrianantoanina's approach [loc.\,cit.], but had to develop tricky auxiliary results for their proof. One of it concerns the definition of uniform integrability of a bounded set \(K\) in \(M_*\) which coincides with the one of Randrianantoanina and of Raynaud-Xu (in the case of von Neumann algebras) because it is shown to be equivalent to relative weak compactness of \(K\). Remark: Meanwhile, the splitting lemma has been generalized to JBW\(^*\)-triples by the second named author and the reviewer [Stud. Math. 227, No. 1, 77--95 (2015; Zbl 1360.46062)]; although they rather follow the approach of Raynaud and Xu [loc. cit.], their proof benefits from several technical results of the paper under review.