2
$\begingroup$

Let $C\subset\mathbb{P}^2$ be a plane nodal cubic curve with a unique singular point $O$ at the origin. Then I consider its normalization, denoted by $\widetilde{C}$ and let $\pi:\widetilde{C}\rightarrow C$ be the normalization map. Then I consider the associated functor $\pi^*:\mathrm{Perf}(C)\rightarrow\mathrm{Perf}(\widetilde{C})$, where $\mathrm{Perf}$ is the derived category of perfect complexes. I was wondering how I can describe the category $\mathrm{Ker}\pi^*$, I think this category is nonzero and the objects in $\mathrm{Ker}\pi^*$ should support on the nodal singularity $O$.

$\endgroup$
2
  • $\begingroup$ Can you not use in this particular case Orlov's SOD of the blowup to give a description of this Ker \pi category? $\endgroup$
    – AT0
    Commented Jan 26 at 23:44
  • $\begingroup$ I think I know what it looks like, maybe I will answer my question myself and write an answer here. $\endgroup$
    – user41650
    Commented Jan 27 at 2:06

1 Answer 1

5
$\begingroup$

The kernel is zero.

Indeed, on the one hand, since $\pi$ is an isomorphism over the complement of $O$, any object in $\operatorname{Ker}(\pi^*)$ is supported at $O$.

On the other hand, if $\tilde{O}$ is the fiber of $\pi$ over the node, $i \colon O \to C$ and $\tilde{i} \colon \tilde{O} \to \tilde{C}$ are the embeddings, and $\pi_O \colon \tilde{O} \to O$ is the restriction of $\pi$ then $$ \tilde{i}^* \circ \pi^* \cong \pi_O^* \circ i^*. $$ The functor $\pi_O^*$ is obviously fully faithful, hence any object in the kernel of $\pi^*$ must lie in the kernel of $i^*$, but then it must be supported away from $O$.

Thus, the kernel of $\pi^*$ is zero.

$\endgroup$
6
  • $\begingroup$ The pullback map on K-groups has a copy of the multiplicative group in its kernel. $\endgroup$ Commented Jan 28 at 14:43
  • $\begingroup$ Right, but this does not contradict to the vanishing of the kernel. $\endgroup$
    – Sasha
    Commented Jan 28 at 15:17
  • $\begingroup$ I know, I just thought I would mention it. $\endgroup$ Commented Jan 28 at 16:18
  • $\begingroup$ @Jason Starr, could you please elaborate on what you commented? It seems that what you said is related to the original motivation for my post--trying to categorify the Jacobian of such nodal curve, which is the $\mathbb{C}^*$-extension of Jacobian of normalization. We try to find an "exact sequence" of categories which categorifies the exact sequence $0\rightarrow\mathbb{C}^*\rightarrow Jac(C)\rightarrow Jac(\widetilde{C})\rightarrow 0$. $\endgroup$
    – user41650
    Commented Jan 30 at 22:12
  • 1
    $\begingroup$ @user41650 There is a categorical resolution of singularities of the curve $C$ obtained by gluing the derived category of $\tilde{C}$ and the derived category of a point, such that the resolution functor from the perfect derived category of $C$ to this categorical resolution is fully faithful (in particular, its kernel is zero). Perhaps, this is the categorification you want. $\endgroup$
    – Sasha
    Commented Feb 1 at 5:42

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .