If $Y$ is a curve of bidegree $(a,b)$ on a smooth quadric surface $Q\subset \mathbb{P}^3$, how do we see that it is arithmetically Cohen-Macaulay (ACM, for short) iff $|a-b|\leq 1$?

If (like me) you aren't familiar with what ACM means, two equivalent conditions in this case are:

1) $H^1(\mathcal{I}_Y(\ell))=0$ for all $\ell\in \mathbb{Z}$, where $\mathcal{I}_Y$ is the ideal sheaf for $Y$ in $\mathbb{P}^3$.

2) $k[x_0,\ldots,x_3]=\bigoplus_{n\in\mathbb{Z}}{H^{0}(\mathscr{O}_{P^3}(n))}\rightarrow \bigoplus_{n\in\mathbb{Z}} {H^0(\mathscr{O}_Y(n))}$ is surjective.

(This is exercise 8.1(a) in Hartshorne's Deformation Theory, and the equivalent conditions are from Proposition 8.6.)


We will use your condition (2). Since $Q$ is a hypersurface, it is ACM (by Hartshorne's Algebraic Geometry, III, Exc. 5.5(a)), so it suffices to show $$\bigoplus_{n \in \mathbf{Z}} H^0(Q,\mathcal{O}_Q(n)) \longrightarrow \bigoplus_{n \in \mathbf{Z}} H^0(Y,\mathcal{O}_Y(n))$$ is surjective if and only if $\lvert a - b \rvert \le 1$.

Remark. If you don't like the usage of the Künneth formula, you can also prove your statement using a lot of long exact sequences. An outline is in Hartshorne's Algebraic Geometry, III, Exc. 5.6, especially part (b), (3).

Consider the exact sequence $$ 0 \longrightarrow \mathcal{O}_Q(n-a,n-b) \longrightarrow \mathcal{O}_Q(n) \longrightarrow \mathcal{O}_Y(n) \longrightarrow 0 $$ The long exact sequence on cohomology is $$\cdots \longrightarrow H^0(Q,\mathcal{O}_Q(n)) \longrightarrow H^0(Y,\mathcal{O}_Y(n)) \longrightarrow H^1(Q,\mathcal{O}_Q(n-a,n-b)) \longrightarrow H^1(Q,\mathcal{O}_Q(n)) \longrightarrow \cdots$$

Now if $\lvert a - b \rvert \le 1$, then since $Q \cong \mathbf{P}^1 \times \mathbf{P}^1$, we have, by the Künneth formula, \begin{align*} H^1(Q,\mathcal{O}_Q(n-a,n-b)) &= \left( H^0(\mathbf{P}^1,\mathcal{O}_{\mathbf{P}^1}(n-a)) \otimes H^1(\mathbf{P}^1,\mathcal{O}_{\mathbf{P}^1}(n-b)) \right)\\ &\quad\quad\oplus \left( H^1(\mathbf{P}^1,\mathcal{O}_{\mathbf{P}^1}(n-a)) \otimes H^0(\mathbf{P}^1,\mathcal{O}_{\mathbf{P}^1}(n-b)) \right) \end{align*} is zero, since if $\lvert a - b \rvert = \lvert (n-b) - (n-a) \rvert \le 1$, then in each summand, one of $H^0$ or $H^1$ must vanish by the cohomology of $\mathbf{P}^1$.

On the other hand, if $\lvert a - b \rvert \ge 2$, then suppose without loss of generality that $a > b$. Letting $n = b$, the long exact sequence from above becomes $$\cdots \longrightarrow H^0(Q,\mathcal{O}_Q(b)) \longrightarrow H^0(Y,\mathcal{O}_Y(b)) \longrightarrow H^1(Q,\mathcal{O}_Q(b-a,0)) \longrightarrow H^1(Q,\mathcal{O}_Q(b)) \longrightarrow \cdots$$ But, again by the Künneth formula, $$H^1(Q,\mathcal{O}_Q(b)) = \left( H^0(\mathbf{P}^1,\mathcal{O}_{\mathbf{P}^1}(b)) \otimes H^1(\mathbf{P}^1,\mathcal{O}_{\mathbf{P}^1}(b))\right)^{\oplus 2} = 0$$ while $$H^1(Q,\mathcal{O}_Q(b-a,0)) = H^1(\mathbf{P}^1,\mathcal{O}_{\mathbf{P}^1}(b-a)) \ne 0$$ and so $H^0(Q,\mathcal{O}_Q(b)) \to H^0(Y,\mathcal{O}_Y(b))$ is not surjective.

  • 1
    $\begingroup$ Maybe it is worthwhile to emphasize that $\mathcal O_Q(b)=\mathcal O_Q(b,b)$ in the notation of Takumi's nice answer above. Anyway, +1. $\endgroup$ – Georges Elencwajg Apr 22 '16 at 9:27

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.