15
$\begingroup$

Let $X$ be an affine scheme and $\mathcal{E}$ a finitely generated locally free sheaf on $X$. It is obvious that $\mathcal{E}$ is a projective object in the category Qcoh$(X)$ since we can pass to rings and modules.

My question is: is $\mathcal{E}$ still projective when we consider it in the larger category $\mathcal{O}_X$-mod which consists of all $\mathcal{O}_X$-modules? If it is true, how to prove it in a "cohomological way"?

I'm not sure whether this question is suitable for mathoverflow. Please feel free to move it if necessary.

$\endgroup$

1 Answer 1

21
$\begingroup$

This answer is inspired by the discussion at this question. Let $X$ be an integral affine scheme admitting an open cover $X=U\cup V$ with $U$, $V$ and $X$ all distinct. I claim that $\mathscr O_X$ is not projective in $\operatorname{Mod}(\mathscr O_X)$. Let $i\colon U\hookrightarrow X$ and $j\colon V\hookrightarrow X$ be the inclusion morphisms. There is an obvious surjection $p\colon i_! \mathscr O_U\oplus j_! \mathscr O_V\twoheadrightarrow \mathscr O_X$ (check surjectivity on stalks). If $\mathscr O_X$ were projective, $p$ would split, so we would have $\Gamma(\mathscr O_X)\hookrightarrow \Gamma(i_!\mathscr O_U)\oplus \Gamma(j_! \mathscr O_V)=0$, a contradiction.

$\endgroup$
2
  • $\begingroup$ Sorry for obvious question. Why is $\Gamma(i_! \mathcal{O}_U) \oplus \Gamma(i_! \mathcal{O}_V) = 0$? $\endgroup$
    – user113988
    Dec 1, 2016 at 13:49
  • 1
    $\begingroup$ The lower shriek is this case is just extension by zero, so $\Gamma(W, i_! \mathcal{O}_U) = \Gamma(W,\mathcal{O}_U)$ if $W\subset U$, and $0$ otherwise. So in this case $\Gamma(i_! \mathcal{O}_U) = \Gamma(i_! \mathcal{O}_V) = 0$, since $U$ and $V$ are strict subsets of $X$. $\endgroup$ Dec 1, 2016 at 16:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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