11

This is a question about the proofs of Kazhdan-Lusztig's conjectures for category $\mathcal{O}$ using higher representation theory (avoiding Beilinson-Bernstein's geometric localization theory).

Using Bernstein-Frenkel-Khovanov and generalizations (Sussan, Stroppel-Mazorchuk, etc), the Kazhdan-Lusztig conjectures in type A are equivalent to the following statement: the classes of the simple modules in their categorification correspond to a "dual canonical basis" in a tensor product representation of $\mathfrak{sl}_k$ (for appropriately chosen $k$).

Using Losev-Webster and Webster, that statement can be deduced. They construct a theory of tensor product categorifications in type A (i.e. existence + uniqueness).

Can [Losev-Webster] be simplified, so that the connection to Soergel's J.AMS paper (www.ams.org/jams/1990-03-02/S0894-0347-1990-1029692-5/) becomes clear? It would be interesting to construct a theory of tensor product categorifications, with the input data being a Dynkin diagram (i.e. a simple Lie algebra).

Puraṭci Vinnani
  • 2,186
  • 4
    MO is NOT a polymath-type platform. If you want to do this as polymath, get in touch with the polymath people. If you want to ask spin-off questions from this one, why not do so with separate questions? – Yemon Choi Apr 02 '18 at 23:00
  • How do I get in touch with them? I've been reading the website, https://terrytao.wordpress.com/2018/01/24/polymath-proposal-upper-bounding-the-de-bruijn-newman-constant/ But I'm not sure.. – Puraṭci Vinnani Apr 03 '18 at 00:25
  • https://mathoverflow.net/questions/219638/proposals-for-polymath-projects – Francois Ziegler Apr 03 '18 at 00:28
  • Thanks. I still don't see how to propose a polymath project (I've read the comments there carefully). If you can that answer that question precisely, please let me know - and I will move on to the next step. – Puraṭci Vinnani Apr 03 '18 at 02:45
  • 5
    I'm voting to close this question given the direction it's now taken, because I believe this site is not the right platform for this kind of endeavour, although the individual problems/questions might be appropriate as individual questions – Yemon Choi Apr 04 '18 at 20:47
  • Ok I've changed the wording slightly. Let me know what you think. – Puraṭci Vinnani Apr 04 '18 at 20:53
  • I find it slightly confusing what you mean in (B). For example, it does not really make sense to try to show that the combinatorics of Lusztig's conjecture matches the setup, since we know that it will not unless the prime is very large. The nice feature of the work by Riche-Williamson (and more recently also together with Achar and Makisumi) is that it gives the correct answer as long as the prime exceeds the Coxeter number. (cont.) – Tobias Kildetoft Apr 05 '18 at 08:04
  • This does still leave some primes where we don't quite know what is going on, since passing to characters of simple modules requires that $p\geq 2h-2$ (or that we assume Donkin's tilting conjecture, but even then, the picture is not nearly as nice as when $p\geq 2h-2$ since there is no longer a direct analogue of BGG reciprocity between tilting modules and simple modules). – Tobias Kildetoft Apr 05 '18 at 08:06
  • 1
    I recommend editing the question back to its original version, so that the posted answers make sense. I agree with Yemon Choi's comment above that some of the questions currently posed could be asked separately. – j.c. Apr 05 '18 at 09:28

2 Answers2

20

This is not a truly independent proof. Both of the papers of mine above use the decomposition theorem for various collections of algebraic varieties, which essentially include the proof of the original KL conjecture as special cases. You should think of the techniques in those papers as a generalization of Soergel's JAMS paper, rather than as an independent approach to the same material.

Ben Webster
  • 43,949
4

Concerning your other question (4), it's probably too early to say what will eventually happen with Lusztig's various conjectures in prime characteristic. By now there is a lot of literature, following Williamson's cautionary observations about Lusztig's Santa Cruz conjecture in 1979 modeled on the statement of the Kazhdan-Lusztig conjecture. The modular situation turns out to be even more complicated than expected. See for example the papers and preprints listed on Achar's and Williamson's homepages, such as here or here.

While the earlier work by Andersen-Jantzen-Soergel proved Lusztig's first conjecture for all "sufficiently large" primes $p$, the explicit bound on $p$ found afterward by Fiebig is extremely large. There remains a serious problem about "intermediate" primes, and no firm conjecture about what happens when $p$ is smaller than the Coxeter number. While the current work encourages hopes for a solution, the end result is likely to be far more complicated than the Kazhdan-Lusztig conjecture. It's definitely worthwhile to seek alternative approaches but probably premature to expect a simple answer involving just the Kazhdan-Lusztig polnomials for an affine Weyl group.

Jim Humphreys
  • 52,369