In 1934, Jordan, von Neumann and Wigner gave a nice classification of finitedimensional simple Jordan algebras that are 'formally real', meaning that a sum of squares is zero only if each term in the sum is zero. In 1983 Zelmanov generalized this to all simple Jordan algebras, but unlike the original result, Zelmanov's 'classification' is not a neat list. What about formally real Jordan algebras, not necessarily finitedimensional? What does the classification of these look like?

13$\begingroup$ Welcome. If you put your last name in the search box in the upper right, you'll find a number of questions you may find of interest, as people have invoked your writings. Meanwhile, if you register you can avoid having multiple ID's on MO, that happened to me. Also click on "Users," I think you will recognize some names. Not mine, can't be helped. $\endgroup$– Will JagyNov 9, 2010 at 6:36

$\begingroup$ The work by Zelmanov that you refer to appears to be this one: ams.org/mathscinetgetitem?mr=688595 (MR0688595, for future reference). $\endgroup$– Harald HancheOlsenNov 11, 2010 at 14:27
1 Answer
The formally real Jordan algebras include the class of JBalgebras, a class of normed Jordan algebra which is the Jordan algebra equivalent of C*algebras. From this you might imagine that obtaining a complete classification is a rather nontrivial task. In this class you also find JWalgebras, which are JBalgebras with a predual, thus corresponding to von Neumann algebras. Some of the concepts from C* and von Neumann algebra theory carry over to the Jordan algebra setting, but this margin is too narrow to summarize what is known. With apologies for tooting my own horn here, in 1984 I coauthored a book “Jordan Operator Algebras” with Erling Størmer which pretty much summarized the state of the art at the time. (Since then I have left that field, so I don't know if a lot has happened since.)

8$\begingroup$ JB, were you just fishing for someone to mention JBalgebras? $\endgroup$ Nov 9, 2010 at 13:42

1$\begingroup$ Heh. Well, JB in this context stands for Jordan–Banach. $\endgroup$ Nov 9, 2010 at 15:09

$\begingroup$ Thanks, Harald. No thanks, Allen. :) I would really like to see a wild multitude of simple formally real Jordan algebras. I'm not sure I see them yet. Harald, I guess you're hinting that I can get such a thing from any C*algebra that lacks nontrivial starideals. But this reminds me that my question said nothing about a norm, or topology. I know lots of infinitedimensional C*algebras that lack nontrivial closed starideals, but how about C*algebras that don't have any nontrivial starideals? $\endgroup$ Nov 11, 2010 at 5:58

$\begingroup$ John, I was painfully aware at the time that I did not really answer your question. Still, I hoped (and still hope) that at least it did shed some light on some corner of it. Regarding ideals, I am sure you are aware that in a unital C*algebra (or unital Banach algebra more generally), the closure of a nontrivial ideal is again nontrivial, since a neighbourhood of the unit consists of invertible elements. Nonunital algebras is presumably a whole different kettle of fish. $\endgroup$ Nov 11, 2010 at 14:13