0
$\begingroup$

Let $\Phi$ be an irreducible root system in a Euclidean vector space $V$. Let $W$ denote its Weyl group. Choose a base $\Delta=\{\alpha_1,...,\alpha_r\}$ for $\Phi$. Then $\Delta$ is a basis for $V$. Let $$F_\Delta=\{\omega_1^\vee, ...,\omega_r^\vee\}\subset V^*$$ denote the dual basis. This is the set of fundamental coweights associated to $\Delta$. Now observe that for every $w\in W$, the set $w.\Delta$ is also a base for $\Phi$ with corresponding fundamental coweights $w.F_\Delta$. Moreover, $$ \bigcup_{w\in W} w.\Delta = \Phi. $$

Question: What does the set $\displaystyle X:=\bigcup_{w\in W} w.F_\Delta$ look like? E.g. how many elements does it have? Is there a direct way to obtain $X$ from $\Phi$ without choosing a base?

Added in edit: Note that each element $x\in X\subset V^*$ defines a hyperplane in $V$; namely, $\mathrm{ker}(x)$. I am also interested in the hyperplane arrangement defined by $X$. Is there anything known about this?

$\endgroup$
5
  • $\begingroup$ If $\Phi$ is of Type $A_n$ then $\# X = 2^{n+1}-2$. $\endgroup$ Commented Nov 3, 2023 at 2:21
  • $\begingroup$ In general, the size of the $W$-orbit of $\omega_i^\vee$ is $\#W/\#W_i$ where $W_i$ is the maximal parabolic subgroup corresponding to node $i$ (i.e., the Weyl group of the Dynkin diagram we get by deleting node $i$). It’s not clear you can get a better formula for $\#X$ than just summing this over all $i$. $\endgroup$ Commented Nov 3, 2023 at 2:32
  • $\begingroup$ In Type A, something special happens: every fundamental weight is minuscule. So you can characterize $X$ as the set of those nonzero coweights $\omega^{\vee}$ which have $(\omega^{\vee}, \alpha) \in \{0,\pm1\}$ for all roots $\alpha \in \Phi$. But this characterization does not extend to other types. $\endgroup$ Commented Nov 3, 2023 at 13:01
  • $\begingroup$ Thanks a lot Sam. $\endgroup$
    – Dr. Evil
    Commented Nov 4, 2023 at 7:22
  • $\begingroup$ Again in Type $A_n$, the hyperplane arrangement you mention at the end is (basically) the so-called "resonance arrangement." This is a classic example of a hyperplane arrangement about which it is very hard to say anything. See e.g. the prior MO questions mathoverflow.net/questions/62764 and mathoverflow.net/questions/372427. $\endgroup$ Commented Nov 4, 2023 at 13:41

0

You must log in to answer this question.

Browse other questions tagged .