2
$\begingroup$

This question is about (a special case of) the varieties discussed here: Does the preimage of the Slodowy slice in $T^*G/P$ have a name?.

Let $G = SL_n(\mathbb{C}), \mathfrak{g} = \mathfrak{sl}_n(\mathbb{C})$; $\mathcal{N}$ its nilpotent cone and let $z \in \mathfrak{g}$. The Springer resolution is $\widetilde{\mathcal{N}}=T^*G/B = \{ (0 \subset V_1 \subset \cdots \subset V_n), x | x V_{i} \subset V_{i-1} \}$; the natural map $\pi: \widetilde{\mathcal{N}} \rightarrow \mathcal{N}$ is a resolution of singularities.

Let $e \in \mathcal{N}$ be a nilpotent (I'm in the case where $z$ is a $2$-block nilpotent, but I'm not sure if this simplifies matters much). Using Jacobson-Morozov, pick an $\mathfrak{sl}_2$-triple $\{ e, h, f \}$; then the Slodowy slice is defined to be $S = e + Z_{\mathfrak{g}}(f)$ (where $Z_{\mathfrak{g}}(f)$ is the centralizer of $f$ in $\mathfrak{g}$).

Let $S' = S \cap \mathcal{N}$, and $\widetilde{S'} = \pi^{-1}(S') = \{ (V_{\bullet}), x | x \in S, x V_{i} \subset V_{i-1} \}$. It is known that $\widetilde{S'}$ is smooth.

My question: given a point $t = \{ (V_{\bullet}), x\} \in \widetilde{S'}$, what is $T_{t}(\widetilde{S'})$ (the tangent space to $\widetilde{S'}$ at $t$)? Since $\widetilde{S'}$ is a subvariety of $\widetilde{\mathcal{N}}$, the answer should be as a subspace of $T_t(\widetilde{\mathcal{N}}) = \mathfrak{n}_t^* \oplus \mathfrak{n}_t$, where $\mathfrak{n}_t$ is the nilpotent radical of the Borel $\mathfrak{b}_t$ corresponding to $(V_{\bullet}) \in G/B$. Here the tangent space $T_t(\widetilde{\mathcal{N}}) = \mathfrak{n}_t^* \oplus \mathfrak{n}_t$, because $\widetilde{\mathcal{N}} \rightarrow G/B$ is a vector bundle; the tangent space $T_{(V_{\bullet})}(G/B) = \mathfrak{g}/\mathfrak{b}_t \simeq \mathfrak{n}_t^*$ and the fiber above $(V_{\bullet})$ is $\mathfrak{n}_t$.

$\endgroup$
2
$\begingroup$

Here's an answer of sorts: by assumption we have that a point $((V_\cdot),x)$ of $\widetilde{\mathcal N}$ is in $\widetilde{S}$ if and only if $[f,x]=h$. So we get the obvious linearization of this condition on ${\mathfrak n}_t^*\oplus{\mathfrak n}_t$, namely that a point $(u,y)\in({\mathfrak g}/{\mathfrak b}_t)\oplus{\mathfrak n}_t$ is in the tangent space if and only if $[f,[u,x]+y]=0$.

EDIT: This isn't quite right because of course this isn't well-defined for $u\in{\mathfrak g}/{\mathfrak b}_t$. I have essentially identified ${\mathfrak g}/{\mathfrak b}_t$ and ${\mathfrak n}_t^*$ with the nilradical of the negative Borel to ${\mathfrak b}_t$, but that isn't right.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.