Liu & Yau hit the arXiv (again)

Today, Chien-Hao Liu and Shing-Tung Yau (yep, the one of Calabi-Yau fame) posted a paper titled Nontrivial Azumaya noncommutative schemes, morphisms therefrom, and their extension by the sheaf of algebras of differential operators: D-branes in a B-field background à la Polchinski-Grothendieck Ansatz. (Btw. is there a limit to the number of characters one can put in an arXiv-title? Surely, it should be less than a text-message?)

To get my attention, they might as well have stopped after the first comma : ‘Nontrivial Azumaya noncommutative schemes’. If there is one noncommutative scheme which isn’t noncommutative at all, it should be an Azumaya scheme surely … nontrivial, or not.

But hey, I downloaded the pdf-file and surely enough their Key-words included all of the current arty-farty-fanshi-wanshi terms : Azumaya scheme, Azumaya structure, B-field, D-brane, D-string, D-module, deformation quantization, gerbe, Higgs/spectral pair, moduli stack, morphism, Polchinski-Grothendieck Ansatz, quantum spectral curve, sheaf of algebras of differential operators, twisted sheaf, valuation criterion.

Wow! Perhaps, I should at least try to understand some of this…

Woops … I must have been living in another galaxy or have been preoccupied with entirely different things lately, for this is but the 5th (!) in a series of papers with equally catchy titles. Here’s the list:

Yeah well, my bad. So, I’ve printed it all out, made a pot of coffee and started reading…

I’m finished for now, but don’t quite get it. So, has anyone worked through these papers and got something out of them?

In part 1 they define ‘Azumaya-type noncommutative space’ which appear to be just an \mathcal{O}_X-sheaf of orders on a (commutative) scheme X. But, for their ‘physical’ applications they seem to require that this order is a split Azumaya algebra (in parts 1-4) and a general sheaf of Azumaya algebras (in part 5).

Here’s what I believe to understand of their approach. I’ll just sketch the affine case (that is, working with rings rather than schemes) as they use only tame (that is, central) localizations to glue things together.

They want to have a ‘geometric’ (ie. space and stuff) interpretation of the natural extension of Grothendieck’s ‘functor of points’ approach. That is, they consider for a given algebra A the covariant functor

h_A~:~\C-\wis{algebras} \rightarrow \wis{sets} \qquad B \mapsto Alg_{\C}(A,B)

and want to construct their ‘geometric’ object containing as much information as this functor. In the cmmutative case, Grothendieck showed how one could do this : take the prime spectrum of the algebra and adorn it with a structure sheaf, that is, look at the affine scheme ~(spec(A),\mathcal{O}_A) of your algebra. Then he showed that their is a duality between the two approaches, that is, to any algebra morphism A \rightarrow B, there is a corresponding scheme-morphism ~(spec(B),\mathcal{O}_B) \rightarrow (spec(A),\mathcal{O}_A).

In the noncommutative case we face the immediate problem that there isn’t even a map from prime ideals of B to prime ideals of A!

Lie & Yau solve this problem as follows : their ‘geometric’ object associated to a complex algebra A does not just consists of twosided prime ideals of A itself, but also of all prime ideals of all (!) Z(A)-subrings of A. Might sound like a crazy idea at first, but it ties in nicely with old results by Artin and Procesi.

They showed that there is(!) a natural map between prime ideals whenever the algebra map f~:~A \rightarrow B is a ‘central extension’, meaning that B = f(A)Z(B) with Z(B) being the center of B. So, what Liu&Yau propose it to take all prime ideals of all possible central extensions to resolve the problem.

For example, an Azumaya-type noncommutative point of size n, corresponds to M_n(\C) should really be thought of as all prime-ideals of all \C-subalgebras of M_n(\C). They call it the M_n(\C) ‘noncommutative cloud’. As a consequence, their ‘generalized morphisms’ from this noncommutative point to one of their “Azumaya-type noncommutative spaces” associated to an algebra A is nothing but the collection of all n-dimensional semi-simple representations of A.

Right, now that we connect their approach with the little we algebraists know, would any of you be interested in running through these papers together in a couple of posts, trying to connect the NCAG-content to ’stringy’ stuff like D-branes, Chan-Paton modules and Higgsing/un-Higgsing, whathever all this means?