3
$\begingroup$

I am going to describe a "degeneracy functor" $$\delta_\mathfrak{q} \ : \ \text{Par}(\text{SL}_n,B_n) \ \to \ \text{Par}(\text{SL}_{n-1},B_{n-1})$$ from parabolics of a big group to a smaller group (both containing given Borels), and my question is: what is the morally correct definition of this functor which extends at least to arbitrary reductive groups $G$?

The functor is attached to something like $$\mathfrak{q} \ = \ \left( \begin{array}{cccccc} \color{white}{\blacksquare} & \color{white}{\blacksquare} & \color{white}{\blacksquare} & \color{white}{\blacksquare} & \color{white}{\blacksquare} & \color{white}{\blacksquare} \\ & \color{yellow}{\blacksquare} & \color{yellow}{\blacksquare} & \color{yellow}{\blacksquare} & \color{white}{\blacksquare} & \color{white}{\blacksquare} \\ & \color{yellow}{\blacksquare} & \color{yellow}{\blacksquare} & \color{yellow}{\blacksquare} & \color{white}{\blacksquare} & \color{white}{\blacksquare} \\ &\color{yellow}{\blacksquare} & \color{yellow}{\blacksquare} & \color{yellow}{\blacksquare} & \color{white}{\blacksquare} & \color{white}{\blacksquare} \\ & & & & \color{white}{\blacksquare} & \color{white}{\blacksquare} \\ & & & & & \color{white}{\blacksquare} \end{array} \right),$$ where I suspect $\mathfrak{q}B$ should be viewed as a standard parabolic in $\text{SL}_n$.

The functor $\delta_\mathfrak{q}$ takes a parabolic $B_n\subseteq \color{blue}{P}\subseteq \text{SL}_n$ and "$\mathfrak{q}$-blockifies" it, e.g.

$$\left( \begin{array}{cccccc} \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ \color{blue}{\blacksquare}& \color{green}{\blacksquare} & \color{green}{\blacksquare} & \color{green}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ &\color{yellow}{\blacksquare}& \color{green}{\blacksquare} & \color{green}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & \color{yellow}{\blacksquare}& \color{yellow}{\blacksquare} & \color{green}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & & & & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & & & & & \color{blue}{\blacksquare} \end{array} \right) \ \mapsto \ \left( \begin{array}{ccccc} \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ \color{blue}{\blacksquare}& \color{green}{\blacksquare} & \color{green}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & \color{yellow}{\blacksquare}& \color{green}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & & & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & & & & \color{blue}{\blacksquare} \end{array} \right)$$

$$\left( \begin{array}{cccccc} \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ \color{blue}{\blacksquare}& \color{green}{\blacksquare} & \color{green}{\blacksquare} & \color{green}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ &\color{green}{\blacksquare}& \color{green}{\blacksquare} & \color{green}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & \color{yellow}{\blacksquare}& \color{yellow}{\blacksquare} & \color{green}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & & & & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & & & & & \color{blue}{\blacksquare} \end{array} \right) \ \mapsto \ \left( \begin{array}{ccccc} \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ \color{blue}{\blacksquare}& \color{green}{\blacksquare} & \color{green}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & \color{green}{\blacksquare}& \color{green}{\blacksquare} & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & & & \color{blue}{\blacksquare} & \color{blue}{\blacksquare} \\ & & & & \color{blue}{\blacksquare} \end{array} \right)$$ In the above, the parabolic $P$ on the left is drawn in $\color{blue}{\text{blue}}/\color{green}{\text{green}}$, and its image $\delta_\mathfrak{q}(P)$ on the right is also drawn in $\color{blue}{\text{blue}}/\color{green}{\text{green}}$. We draw $\mathfrak{q}$ on the left in $\color{yellow}{\text{yellow}}/\color{green}{\text{green}}$.

More precisely, the functor takes $P$, "adds in" the rest of $\mathfrak{q}$ if $P$ contains a root of $\mathfrak{q}$, then "contracts the $\text{GL}_3$ in $\mathfrak{q}$ to a copy of $\text{GL}_2$", i.e. $\delta_\mathfrak{q}(P_I)= P_{d(I)}$ where $d:[n]\to [n-1]$ is the degeneracy map induced by contracting the two simple roots in $\mathfrak{q}$, and $P_I$ is the standard parabolic attached to subset $I$ of roots.

To reiterate my question: is this an example of something that works for arbitrary (B,N)-pairs/Tits systems/reductive groups/Coxeter groups/[any class of such groups/root data bigger than just ADE]?

$\endgroup$
10
  • 1
    $\begingroup$ Could you say more about what "contracts the $\operatorname{GL}_3$ in $\mathfrak q$ to a copy of $\operatorname{GL}_2$" means? I can't quite follow why dots are green or yellow after contraction. Or does it matter? $\endgroup$ Commented Aug 16 at 23:18
  • $\begingroup$ @LSpice Is the edit clearer? r.e. colour choices - I coloured in the $3\times 3$ and $2\times 2$ blocks in $\text{SL}_n,\text{SL}_{n-1}$ yellow and the parabolics $P,\delta_{\mathfrak{q}}(P)$ blue (so that their intersection is green). $\endgroup$ Commented Aug 17 at 11:42
  • $\begingroup$ I am still not sure what controls whether squares in the "contracted" parabolic are green or yellow, but I think it doesn't matter. The operation of "adding in" $\mathfrak q$ is easy enough: we have that $\mathfrak q$ determines a facet of the root-hyperplane arrangement, and then I think that we write $P = P_G(\lambda)$ and project $\lambda$ onto that facet. The "contraction" operation seems to be taking the centraliser of $\operatorname{diag}(1, \dotsc, 1, \lambda, 1, \dotsc, 1)$ and then projecting onto the $\operatorname{GL}_{n - 1}$ factor. $\endgroup$ Commented Aug 17 at 12:45
  • 1
    $\begingroup$ If that description is correct, then the first step always makes sense, and the second relies on the fact that Levi subgroups of $\operatorname{GL}_n$ are of the form $\operatorname{GL}_r \times \operatorname{GL}_s$, which is not true for general groups (not even general groups of type $\mathsf A_n$, but here you are using the bijection between parabolics of $\operatorname{SL}_n$ and $\operatorname{GL}_n$). Is something like this, adapted to the shape of different Levis for different root data, in the spirit of what you want? $\endgroup$ Commented Aug 17 at 12:47
  • 1
    $\begingroup$ Oh, on further reflection, I think I might better understand something. In terms of colour, the yellow block on the right-hand side of the first $\delta_{\mathfrak q}(P)$ example does not belong to $\delta_{\mathfrak q}(P)$? $\endgroup$ Commented Aug 17 at 13:19

0

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.