The following question arose while thinking about a step in the proof of Huybrechts-Lehn, Theorem 1.3.1 (the Harder-Narasimhan filtration for the projective line $\mathbb{P}^{1}$):

Setup: Let $k$ be a field, let $X$ be a projective $k$-scheme, let $\mathcal{O}_{X}(1)$ be a fixed very ample line bundle on $X$, let $E$ be a vector bundle on $X$. For an integer $b \in \mathbb{Z}$, let $\mathrm{ev}_{E(b)} : \mathrm{H}^{0}(X,E(b)) \otimes_{k} \mathcal{O}_{X} \to E(b)$ be the evaluation map and let $E_{b} \subseteq E$ be the image of the twist $\mathrm{ev}_{E(b)}(-b) : \mathrm{H}^{0}(X,E(b)) \otimes_{k} \mathcal{O}_{X}(-b) \to E$.

Question: Why is $E_{b} \subset E_{b+1}$?

(And does this inclusion depend on a choice of global section $s \in \Gamma(X,\mathcal{O}_{X}(1))$?)

Remark: Note that $\mathrm{Hom}_{\mathcal{O}_{X}}(\mathcal{O}_{X}(-b),\mathcal{O}_{X}(-b-1)) = 0$ so there is no map $\varphi : \mathrm{H}^{0}(X,E(b)) \otimes_{k} \mathcal{O}_{X}(-b) \to \mathrm{H}^{0}(X,E(b+1)) \otimes_{k} \mathcal{O}_{X}(-b-1)$ satisfying $\mathrm{ev}_{E(b+1)}(-b-1) \circ \varphi = \mathrm{ev}_{E(b)}(-b)$.


There is a commutative diagram $$\require{AMScd} \begin{CD} H^0(E(b)) \otimes H^0(\mathcal{O}(1)) \otimes \mathcal{O}(-b-1) @>{\mathrm{ev}_{\mathcal{O}(1)}}>> H^0(E(b)) \otimes \mathcal{O}(-b) \\ @VVV @V{\mathrm{ev}_{E(b)}}VV \\ H^0(E(b+1)) \otimes \mathcal{O}(-b-1) @>{\mathrm{ev}_{E(b+1)}}>> E \end{CD} $$ where the left vertical arrow is the multiplication of global sections. The top arrow is surjective (because $\mathcal{O}(1)$ is very ample), hence the composition of the top and right arrows has image $E_b$. By commutativitity of the diagram it is contained (canonically) in the image $E_{b+1}$ of the bottom arrow.

| cite | improve this answer | |
  • $\begingroup$ Thanks very much. I think your argument shows that we only really need $\mathcal{O}(1)$ to be globally generated: Let $X$ be a scheme, let $E$ be a quasicoherent $\mathcal{O}_{X}$-module, let $\mathcal{L}$ be a globally generated line bundle. Then $\operatorname{im} \mathrm{ev}_{E} \subseteq \operatorname{im} (\mathrm{ev}_{E \otimes \mathcal{L}} \otimes \mathcal{L}^{-1})$. $\endgroup$ – user2831784 Sep 23 at 18:29
  • $\begingroup$ Yes, that's correct. $\endgroup$ – Sasha Sep 23 at 18:32

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.