Suppose, on a smooth projective complex variety $X$ that we are given an effective divisor $D$ and $A\in\mathrm{Pic}(D)$ a globally generated line bundle on $D$ with $r$ independent sections. Then we have a surjection $\mathcal{O}_X^{\oplus r}\to A$ (here we view $A$ as a torsion sheaf on $X$ with support on $D$) and thus an exact sequence $$0\to F\to\mathcal{O}_X^{\oplus r}\to A \to 0$$ where $F$ is the kernel sheaf. Is $F$ locally free?

What about the same situation but with a general vector bundle $V$ in place of $\mathcal{O}_X^{\oplus r}$ ?


The question is local, so let us look at everything over the local ring at a point. We then have a regular local ring $R$, a nonzero element $f\in R$ (the local equation of $D$), and an $R$-module $A$ which is annihilated by $f$ and is invertible as a module over $R/(f)$. We want to know whether $A$ has projective dimension $\leq 1$ over $R$ (because then the kernel of any surjection from a projective module to $A$ is projective). Since $A$ is invertible over $R/(f)$, it is a direct summand of a free module over $R/(f)$, so it suffices to show that $R/(f)$ has projective dimension $\leq 1$ over $R$. But this is easy, since $0\to R\stackrel{f}{\to} R\to R/(f)\to 0$ is a projective resolution of $R/(f)$ (since $f$ is nonzero and $R$ is a domain). Thus $R/(f)$ has projective dimension $\leq 1$, and hence so does $A$.

More generally, this argument shows that the kernel of any surjection from a locally free sheaf on $X$ to a locally free sheaf on $D$ is locally free.

  • $\begingroup$ Dear Eric, thanks for your answer. A small question: why is the kernel of any surjection from a projective module to $A$ projective when $A$ has projective dimension 1? $\endgroup$ – Heitor Fontana Dec 3 '15 at 13:04
  • 1
    $\begingroup$ If $0\to F\to V\to A\to 0$ is exact and $V$ is projective, then by the long exact sequence in Ext, $\operatorname{Ext}^i(F,M)=\operatorname{Ext}^{i+1}(A,M)$ for all $i>0$ and all modules $M$. So if $\operatorname{Ext}^j(A,M)=0$ for $j>1$, then $\operatorname{Ext}^i(F,M)=0$ for $i>0$. $\endgroup$ – Eric Wofsey Dec 3 '15 at 14:36

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.