Piecewise \(^\ast\)-homomorphisms and Jordan maps on \(C^\ast\)-algebras and factor von Neumann algebras (Q1754634)

The celebrated Mackey-Gleason theorem proved by \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)] has proved useful to attack many different problems on von Neumann algebras. The Mackey-Gleason theorem was immediately applied to explore the ``quasi-linearity problem''. Let us briefly discuss this problem. Let \(A_{sa}\) denote the self-adjoint part of a \(C^*\)-algebra \(A\). A Banach space-valued mapping \(\rho: A_{sa}\to X\) is called piecewise linear if it satisfies the following:{\parindent=0.7cm\begin{itemize}\item[(i)] \(\rho (a+b) = \rho(a) +\rho(b)\), whenever \(a\) and \(b\) commute (this is called piecewise additivity); \item[(ii)] \(\rho(\lambda a) = \lambda \rho(a)\) for all \(\lambda\in \mathbb{R}\), \(a\in A_{sa}\). \end{itemize}} The quasi-linearity problem asks whether every Banach space-valued piecewise linear map \(\rho : A_{sa}\to X\), where \(A\) is a \(C^*\)-algebra containing no quotient isomorphic to the \(C^*\)-algebra of two by two matrices, and \(\rho\) is bounded on the unit ball, is linear on \(A_{sa}\). Bunce and Wright [loc. cit.] gave a complete solution to this problem in the setting of von Neumann algebras. However, little or nothing is known when \(A\) is a \(C^*\)-algebra. The paper under review is a significant contribution towards a solution to the quasi-linearity problem and to the Mackey-Gleason problem for general \(C^*\)-algebras. One of the main conclusions is presented in Theorem 2.6 and shows that every continuous bijection between self-adjoint parts of \(C^*\)-algebras that preserves the triple product \((a, b) \mapsto aba\), and is linear on commutative subspaces, is already linear. Given a \(C^*\)-algebra \(A\), the set of all positive invertible elements in \(A\) is denoted by \(A_+^{-1}\). A mapping \(\varphi : A^{-1}_+\to B^{-1}_+\) is said to be a piecewise Jordan triple homomorphism if the identity \(\varphi(aba) = \varphi(a) \varphi(b) \varphi(a)\) holds for all \(a, b \in A_{+}^{-1}\) such that \(ab = ba\). In the setting of von Neumann algebras, the author proves in Theorem 3.3 that, if \(A\) is a von Neumann algebra without type I\(_2\) direct summand, and \(\varphi :A^{-1}_+\to B^{-1}_+ \) is a weak\(^*\) to weak\(^*\) continuous piecewise homomorphism, then there is a bounded linear map \(f:A_{sa}\to B_{sa}\) preserving commutativity such that \(\varphi (a) = e^{f(\log(a))}\) for all \(a\in A_+^{-1}\). If we also assume that \(M\) and \(N\) are von Neumann factors not of type I\(_1\) nor I\(_2\), then each weak\(^*\) to weak\(^*\) continuous piecewise isomorphism \(\varphi: M_+^{-1} \to N_+^{-1}\) is of the form \(\varphi (a) = e^{\psi(\log(a))} \Theta (a^c)\) for all \(a\in A_+^{-1}\), where \(\Theta : M_{sa}\to N_{sa}\) is a linear Jordan isomorphism, \(c\) is a nonzero real number, and \(\psi\) is a bounded real-valued linear functional on \(M_{sa}\). The author also describes the same type of bicontinuous maps between unitary groups of von Neumann factors.