Dye's theorem for tripotents in von Neumann algebras and JBW\(^*\)-triples (Q2029187)

This paper is a perfect example of the ongoing research to find invariants to classify and characterize von Neumann algebras, JBW\(^*\)-algebras and JBW\(^*\)-triples. The seeking of certain subsets which are, in general, non-stable for linear combinations nor for other algebraic or geometric properties, but contain the complete information of the whole space, is a current problem on preservers. To understand the new advances it is perhaps worth to look at classical results. A celebrated result due to Dye states if \(\varphi: P(M) \to P(N)\) is an orthoisomorphism (i.e., a bijection preserving orthogonality in both directions) between the projection lattices of two von Neumann algebras \(M\) and \(N\), where \(M\) does not have a type \(I_2\) direct summand, then there is a Jordan \(^*\)-isomorphism \(J: M \to N\) extending \(\varphi\) [\textit{H. A. Dye}, Ann. Math. (2) 61, 73--89 (1955; Zbl 0064.11002)]. Dye's theorem can be regarded as a generalization of the famous Wigner and Ulhorn theorems about symmetries of quantum systems and the logic theoretic version of this principle, respectively. A~version of Dye's theorem in the setting of JW\(^*\)-algebras was established by \textit{L. J. Bunce} and \textit{J. D. M. Wright} [Expo. Math. 11, No.~1, 91--95 (1993; Zbl 0772.46037)]. An orthomorphism \(\varphi\) between JW\(^*\)-algebras \(M\) and \(N\) is a map from the set of projections of \(M\) into the set of projections of \(N\) such that: \begin{itemize} \item[1.] \(\varphi(0)=0\); \item[2.] \(\varphi(1-p)=1-\varphi(p)\); \item[3.] If \(p\perp q\), then \(\varphi(p)\perp \varphi(q)\) and \(\varphi(p+q)=\varphi(p)+\varphi(q)\). \end{itemize} Relying on the Bunce-Wright-Mackey-Gleason theorem, \textit{L. J. Bunce} and \textit{J. D. M. Wright} [Bull. Am. Math. Soc., New Ser. 26, No. 2, 288--293 (1992; Zbl 0759.46054)] proved that an orthomorphism \(\varphi: M\to N\) can be extended to a Jordan \(^*\)-homomorphism of the corresponding JW\(^*\)-algebras provided that \(M\) has no direct summand of type \(I_2\). In the particular case of von Neumann algebras, this is just Dye's theorem. By replacing the lattice of projections in \(B(H)\) by the set of all partial isometries with the natural partial order provided by orthogonality, \textit{L. Molnár} proved in [Arch. Math. 78, No. 1, 43--50 (2002; Zbl 1037.47025)] that the partial order set of all partial isometries (\(\equiv\) tripotents) in \(B(H)\) is a real Jordan invariant for \(B(H)\) when dim\((H)\geq 3\). More concretely, let \(PI(H)\) be the set of all partial isometries acting on a Hilbert space \(H\) with dim\((H)\geq 3\). Let \(\varphi: PI(H)\to PI(H)\), be a bijection preserving orthogonality and order in both directions. If \(\varphi\) is continuous at some nonzero element \(u\) of \(PI(H)\), then \(\varphi\) extends to a real linear triple automorphism of \(B(H)\) for the triple product given by \(\{a,b,c\} = \frac12 (a b^* c + c b^* a)\). In the note under review, the author explores the information contained in the partial order set of all tripotents in a JBW\(^*\)-algebra \(M\), or more generally in a JBW\(^*\)-triple \(E\), with the aim of determining if the structure of \(M\) or \(E\) is determined by this partial order set. Let us recall that a mapping \(\varphi\) between two posets \(P\) and \(Q\) endowed with a relation of orthogonality is called orthomorphism (respectively, an orthoisomorphism) if it preserves orthogonality relation in one direction (respectively, if it is a bijection preserving orthogonality relation in both directions). If \(P\) and \(Q\) are generalized orthomodular posets, a mapping \(\varphi: P \to Q\) is called a quantum logic morphism if \(\varphi(0) =0\) and for every \(a, b \in P\) with \(a\perp b\) we have \(\varphi(a) \perp \varphi(b)\) and \(\varphi(a\vee b) = \varphi(a)\vee\varphi(b)\). A bijection \(\varphi: P \to Q\) such that both \(\varphi\) and \(\varphi^{-1}\) are quantum logic morphisms is called a quantum logic isomorphism. Let \(E\) and \(F\) be JB\(^*\)-triples, and let \(\varphi : U(E) \to U(F)\) be a map between the poset of all tripotents in \(E\) and \(F\), respectively. A system of maps \(\Phi(\varphi) = (\Phi_u: E_2(u) \to F_2 (\varphi(u)))_{u\in U(E)}\) is called a consistent system of Jordan \(^*\)-homomorphisms corresponding to \(\varphi\) if the following conditions are satisfied: \begin{itemize} \item[1.] For each \(u\in U(E),\) the mapping \(\Phi_u: E_2(u) \to F_2 (\varphi(u))\) is a unital Jordan \(^*\)-homomorphism between the unital JB\(^*\)-algebras \(E_2(u)\) and \(F_2(\varphi(u))\). \item[2.] If \(u \leq v\) in \(U(E)\), then \(\Phi_u\) and \(\Phi_v\) coincide on \(E_2(u)\). \end{itemize} If each \(\Phi_u\) is a Jordan \(^*\)-isomorphism and \(\varphi\) is a bijection, we say that the consistent system \(\Phi(\varphi)\) is a consistent system of Jordan \(^*\)-isomorphisms. It is proved in this note that if \(\varphi : U(E) \to U(F)\) is a mapping between the posets of all tripotents in two JB\(^*\)-triples and \(\Phi(\varphi)\) is a consistent system of Jordan \(^*\)-isomorphisms, then \(\varphi\) is a quantum logic isomorphism. Furthermore, assuming that \(E\) and \(F\) are JBW\(^*\)-triples, for each quantum logic morphism \(\varphi : U(E) \to U(F)\) there is a unique consistent system of Jordan \(^*\)-homomorphisms \(\Phi(\varphi)\) corresponding to \(\varphi\) (see Theorem~4.6). Assuming that \(E\) is regular and \(\varphi\) is a quantum logic isomorphism, then there is a unique consistent system of Jordan *-isomorphisms \(\Phi(\varphi)\) corresponding to \(\varphi\) (Theorem~4.7). A set \(S\) in a JBW\(^*\)-triple \(E\) is called triple bounded if the set of all range tripotents of elements in \(S\) admits an upper bound in \(U(E)\). A mapping between JB\(^*\)-triples \(J: E\to F\) is locally additive if \(J(x + y) = J(x) + J(y)\) whenever the set \(\{x, y\}\) is triple bounded. The mapping \(J\) is called a local triple Jordan homomorphism if it is real homogenous, locally additive and preserves tripotents. We say that \(J: E\to F\) is a local triple isomorphism if it is a bijection such that both \(J\) and \(J^{-1}\) are local triple homomorphisms. Every local triple homomorphism between JBW\(^*\)-triples restricts to a quantum logic morphism between the posets of tripotents and satisfies the additional property of preserving antipodal points (Proposition~5.3). If \(E\) and \(F\) are regular JBW\(^*\)-triples, each quantum logic isomorphism \(\varphi : U(E) \to U(F)\) preserving antipodal points extends to a local triple isomorphism. It is finally proved that, in the setting of von Neumann algebras, the posets of partial isometries are a Jordan invariant. More concretely, let \(M\) and \(N\) be von Neumann algebras such that \(U(M)\) and \(U(N)\) are quantum logic isomorphic, then \(M\) and \(N\) are Jordan \(^*\)-isomorphic (Theorem~6.3). This is a pioneering paper, well written, which motivates many potential problems and questions on the poset of all tripotents in a JB\(^*\)-algebra or a JB\(^*\)-triple, and will have an undoubted influence on future research. Its reading is pleasant and recommended for readers who approach this topic.