Existence of contractive projections on preduals of \(JBW^*\)-triples (Q532614)

The so-called ``contractive projection principle'' is a milestone result in the theory of JB\(^*\)-triples. This remarkable principle, independently established by \textit{W. Kaup} [Math. Scand. 54, 95--100 (1984; Zbl 0578.46066)], \textit{Y. Friedman} and \textit{B. Russo} [J. Funct. Anal. 60, 56--79 (1985; Zbl 0558.46035)], and \textit{L. L. Stacho} [Acta Sci. Math. 44, 99--124 (1982; Zbl 0505.58008)], affirms that the range of a contractive projection on a JB\(^*\)-triple is linearly isometric (equivalently, JB\(^*\)-triple isomorphic) to another JB\(^*\)-triple. That is, the category of JB\(^*\)-triples is stable under contractive projections, stability which is not always preserved in the setting of C\(^*\)-algebras. Additional examples of classes of Banach spaces enjoying a good stability property under certain types of projections have also been described, for example: \(\ell_p\), \(L^p\), \(\ell_p\)-direct sums of \(S^p\) (Schatten classes with \(p\neq 2\)), and \(\ell_p\)-direct sums of \(L^p(\Omega,H)\) (\(H\) Hilbert space) (\(1\leq p<\infty\)) are ``stable'' under contractive projections (cf. \textit{A. Grothendieck} [Can. J. Math. 7, 552--561 (1955; Zbl 0065.34503)], \textit{R. G. Douglas} [Pac. J. Math. 15, 443--462 (1965; Zbl 0148.12203)],\textit{T. Ando} [Pac. J. Math. 17, 391--405 (1966; Zbl 0192.23304)], \textit{S. J. Bernau, H. E. Lacey} [Pac. J. Math. 53, 21--41 (1974; Zbl 0287.46038)], \textit{L. Tzafriri} [Isr. J. Math. 7, 9--15 (1969; Zbl 0184.15103)], [\textit{J. Lindenstrauss, L. Tzafriri}, Classical Banach spaces. 1: Sequence spaces. 2. Function spaces. Classics in Mathematics. Berlin: Springer-Verlag. (1996; Zbl 0852.46015)], \textit{J. Arazy, Y. Friedman} [Mem. Am. Math. Soc. 459 (1992; Zbl 0748.47040)], and \textit{Y. Raynaud} [Rev. Mat. Complut. 17, No. 2, 485--512 (2004; Zbl 1066.46025)]), C\(^*\)-algebras and JC\(^*\)-algebras are ``stable'' under completely positive unital projections and under positive unital projections, respectively [\textit{E. G. Effros, E. Størmer} [Math. Scand. 45, 127--138 (1979; Zbl 0455.46059)], TRO's behave stable under completely contractive projections [\textit{M. A. Youngson}, Q. J. Math., Oxf. II. Ser. 34, 507--511 (1983; Zbl 0542.46029)]. It should be noted that TRO's, C\(^*\)- and JC\(^*\)-algebras are not stable under contractive projections, while JB\(^*\)-triples are not stable under bounded projections. It is very natural to consider the converse of the above stability property in the following sense: Let \(\mathfrak{B}\) be the category of Banach spaces and contractions. A sub-category \(S\) of \(\mathfrak{B}\) is \textit{projectively rigid} if it has the property that whenever \(A\) is an object of \(S\) and \(X\) is a subspace of \(A\) which is isometric to an object in \(S\), then \(X\) is the range of a morphism of \(S\) on \(A\) which is a projection. Examples of projectively rigid categories include the following: \(\ell_p\) (\(1< p<\infty\)), \(L^p\), \(S^p\) (Schatten classes) (\(1\leq p<\infty\)) and preduals of von Neumann algebras (cf. \textit{A. Pelczynski} [Stud. Math. 19, 209--228 (1960; Zbl 0104.08503)], \textit{T. Ando} [loc. cit.], \textit{S. J. Bernau, H. E. Lacey} [loc. cit.], \textit{J. Arazy, Y. Friedman} [Isr. J. Math. 26, 151--165 (1977; Zbl 0345.47035)], \textit{E. Kirchberg} [Invent. Math. 112, No. 3, 449--489 (1993; Zbl 0803.46071)]). Preduals of TRO's are ``completely projectively rigid'', that is, a subspace \(X\) of the predual of a TRO \(T\) is completely 1-complemented in \(T\) if, and only if, \(X\) is completely isometrically isomorphic to the predual of some TRO (see \textit{P. W. Ng, N. Ozawa} [Pac. J. Math. 205, No. 1, 171--195 (2002; Zbl 1064.46044)]). A similar statement holds for \(S^p\) (\(1\leq p<\infty\), \(p\neq 2\)) (cf. \textit{C. Le Merdy, E. Ricard, J. Roydor} [Trans. Am. Math. Soc. 361, No. 2, 849--887 (2009; Zbl 1190.46043)]). The above forerunners lead us to the question whether the predual of a general JBW\(^*\)-triple is projectively rigid. In this line, Ng and Ozawa conjectured that the predual of every JC\(^*\)-triple is projectively rigid. The paper under review studies the property of being projectively rigid for preduals of JBW\(^*\)-triples. The authors show that the conjecture of Ng and Ozawa is not, in general, true. It is further shown that the counterexample to this conjecture is, essentially, the unique setting where this property fails. The main result of the paper is the following remarkable theorem which characterises projective rigidity for preduals of JBW\(^*\)-triples: Theorem. Let \(X\) be a subspace of the predual \(A_*\) of a JBW\(^*\)-triple \(A\). If \(X\) is isometric to the predual of another JBW\(^*\)-triple, then there is a contractive projection \(P\) on \(A_*\) such that \(X = P(A_*)\oplus^{\ell_1} Z\), where \(Z\) is isometric to a direct sum of spaces of the form \(L^1(\Omega,H)\) where \(H\) is a Hilbert space of dimension at least two, \(P(A_*)\) is isometric to the predual of some JBW\(^*\)-triple with no such \(L^1(\Omega,H)\)-summand, and \(P(Z) = 0\). The above theorem culminates the study of projective rigidity for preduals of JBW\(^*\)-triples. The proofs are new and the authors develop a series of new technical results which are interesting by themselves. The paper is extremely well detailed with many references and connections with previously commented forerunners.