5
$\begingroup$

In a Jordan algebra elements $a$ and $b$ are said to operator-commute, whenever $a \circ (b \circ x) = b \circ (a \circ x)$ for every other element $x$. (That is: $T_aT_b = T_bT_a$, writing $T_x(y) = x \circ y$.) In a JB-algebra elements $a$ and $b$ operator-commute if and only if they generate an associative subalgebra. (See e.g. p.44 "Jordan operator algebras" by Hanche-Olsen and Størmer.) Does this generalise from pairs to arbitrary subsets?

Question: Assume $A$ is a JB-algebra. $S \subseteq A$ is a subset of pairwise operator-commuting elements. Is the algebra generated by $S$ in $A$ associative?

$\endgroup$

1 Answer 1

2
$\begingroup$

Yes, this is true. I couldn't find any proof of the statement you quote in the article, and even after emailing the authors I didn't get any wiser, so I decided to work out the details myself, see my paper Commutativity in Jordan Operator Algebras. My main result is that if $a$ and $b$ operator commute then they indeed generate an associative JB-algebra of mutually operator commuting elements. In particular, $a^2$, $b^2$ and $a*b$ also commute with each other and with $a$ and $b$.

Now suppose $S$ contains more than two elements. Consider $J(S)$ the Jordan algebra generated by $S$ which consists of "polynomials" in the elements of $S$. For such an element $p(s_1,s_2,\ldots,s_n)$ we can always decompose the operator $T_p$ into simple terms $T_{s_i}$, $T_{s_i*s_j}$ and $T_{s_i^2}$ by repeatedly applying the "normalisation equation" (Eq. 1 in my paper). By the two-element case we know that these all operator commute, hence any two elements of $J(S)$ operator commute. Now we just need to close $J(S)$ in the norm to get the JB-algebra generated by $S$. As the Jordan product is continuous, this algebra also consists of operator commuting elements.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.