Commutativity in Jordan operator algebras

From MaRDI portal
Publication:2184792



Abstract: While Jordan algebras are commutative, their non-associativity makes it so that the Jordan product operators do not necessarily commute. When the product operators of two elements commute, the elements are said to operator commute. In some Jordan algebras operator commutation can be badly behaved, for instance having elements a and b operator commute, while a2 and b do not operator commute. In this paper we study JB-algebras, real Jordan algebras which are also Banach spaces in a compatible manner, of which C*-algebras are examples. We show that elements a and b in a JB-algebra operator commute if and only if they span an associative sub-algebra of mutually operator commuting elements, and hence operator commutativity in JB-algebras is as well-behaved as it can be. Letting Qa denote the quadratic operator of a, we also show that positive a and b operator commute if and only if Qab2=Qba2. We use this result to conclude that the unit interval of a JB-algebra is a sequential effect algebra as defined by Gudder and Greechie.


Let \((E, \cdot)\) be a linear Jordan algebra. Two elements \(a,b\in E\) are said to \textit{operator commute} if \(T_aT_b=T_bT_a\), where for any \(x\in E\), \(T_x: E\to E\) is defined by \(T_x(y)=x\cdot y\) for all \(y\in E\). In the paper the author prove (Theorem 3.13) that for two elements \(a, b\) in a JB-algebra \(A\) the following conditions are equivalent: \begin{itemize} \item[(1)] \(a, b\) operator commute, \item[(2)] the JB-subalgebra \(\overline{\langle a, b\rangle}\) generated by \(a, b\) is associative, \item[(3)] \(\overline{\langle a, b\rangle}\) is associative and their elements operator commute mutually, \item[(4)] \(a\) and \(a^2\) operator commute with \(b\) and \(b^2\). \end{itemize} It is already known that for a general Jordan algebra \(E\) the implications (1) \(\Rightarrow\) (4) and (2) \(\Rightarrow\) (1) don't hold. The proof of Theorem 3.13 is reduced to JBW-algebras, via the embedding of \(A\) into its bidual \(A^{**}\), and divided into two cases, via the structure theory of JBW-algebras, i.e. when \(A^{**}\) is purely exceptional or a JW-algebra. \textbf{Reviewer's remark}. A subalgebra \(C\) of a Jordan algebra \(A\) is said to be a \textit{strongly associative subalgebra} of \(A\) whenever any two elements of \(C\) operator commute (see Section 8 in Chapter I of [\textit{N. Jacobson}, Structure and representations of Jordan algebras. Providence, RI: American Mathematical Society (AMS) (1968; Zbl 0218.17010)]). With this terminology, condition (3) in the theorem can be read as follows: \(\overline{\langle a, b \rangle}\) is a strongly associative subalgebra of \(A\). Therefore the implication (2)\(\Rightarrow\)(3) (which is the one with a more involved proof) can be read as follows: Every associative subalgebra of a JB-algebra (generated by two elements) is a strongly associative subalgebra. A forerunner of this implication for JC-algebras is due to Topping (see Proposition 1 in [\textit{D. M. Topping}, Jordan algebras of self-adjoint operators. Providence, RI: American Mathematical Society (AMS) (1965; Zbl 0137.10203)]), and the general version of this implication is due to \textit{H. N. Boyadjiev} and \textit{M. A. Youngson} [C. R. Acad. Bulg. Sci. 33, 1589--1590 (1980; Zbl 0457.46036)], and was rediscovered by \textit{L. J. Bunce} [Proc. Edinb. Math. Soc., II. Ser. 26, 353--360 (1983; Zbl 0531.46038)]. A proof of this result can be found in Theorem 6.1.118 in the book [\textit{M. Cabrera García} and \textit{ Rodríguez Palacios}, Non-associative normed algebras. Volume 2. Representation theory and the Zel'manov approach. Cambridge: Cambridge University Press (2018; Zbl 1390.17001)].











This page was built for publication: Commutativity in Jordan operator algebras

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2184792)