I am trying to understand better direct image sheaves. To do so, I want to start working in a particular and easy example.

Let $\pi:X\rightarrow \mathbb{P}^2$ be the blow up of $\mathbb{P}^2$ in a point, and $E\subset X$ the exceptional curve. What can we say about $\pi_*\mathcal{O}_X(E)$ or $\pi_*\mathcal{O}_X$? Are they invertible sheaves? If so, which is the integer $n$ such that they coincide with $\mathcal{O}_{\mathbb{P}^2}(n)$?

More in general, we can ask the same questions about the sheaves $\pi_*\mathcal{O}_X(aL+bE)$, where $a,b\in \mathbb{Z}$ and $L$ is the pullback of a line in $\mathbb{P}^2$.

  1. $\pi_*(\mathcal{O}_X) \cong \mathcal{O}_{\mathbb{P}^2}$. In general given $f:X\rightarrow Y$ a birational projective morphism of noetherian integral schemes, with $Y$ normal, then $f_*\mathcal{O}_X = \mathcal{O}_Y$. The proof is in Hartshorne III.11.4.
  2. $\pi_*(\mathcal{O}_X(-E)) \cong \mathcal{I}_p$, where $\mathcal{I}_p$ is the ideal sheaf of $p$ (where here $p$ is the point you're blowing up). To see this, consider the exact sequence

$$0\rightarrow \mathcal{O}_X(-E) \rightarrow \mathcal{O}_X \rightarrow \mathcal{O}_p \rightarrow 0$$ Applying $\pi_*$, we have that

$$0\rightarrow \pi_*\mathcal{O}_X(-E) \rightarrow \pi_*\mathcal{O}_X \rightarrow \pi_* \mathcal{O}_E = \mathcal{O}_p \rightarrow 0$$

where here we note that even though $\pi_*$ is not right exact in general, it is here because you can check by hand that the map is surjective, hence the claim. You can repeat this argument for $\pi_*(\mathcal{O}_X(aE))$ for any $a<0$.

  1. $\pi_*(\mathcal{O}_X(E)) \cong \pi_*(\mathcal{O}_X) \cong \mathcal{O}_{\mathbb{P}^2}$.

Tensor the above short exact sequence by $\mathcal{O}_X(E)$ and apply $\pi_*$. We have that the following sequence is exact:

$$0\rightarrow \pi_*\mathcal{O}_X \rightarrow \pi_*\mathcal{O}_X(E) \rightarrow \pi_* \mathcal{O}_E(E)$$

Now we claim that $\pi_*\mathcal{O}_E(E) = 0$. Since this is supported on $p$, this is the same thing as the vector space $H^0(E,\mathcal{O}_E(E))$. Now using the fact that the self intersection of $E$ is $-1$, $\mathcal{O}_E(E)$ is a line bundle of degree $-1$ on $E\cong \mathbb{P}^1$ which has no non-zero global sections. Hence we deduce that $\pi_*(\mathcal{O}_X(E)) \cong \pi_*(\mathcal{O}_X)$. You can repeat this argument for $\pi_*(\mathcal{O}_X(aE))$ for any $a>0$.

  1. Finally for $\pi_*\mathcal{O}_X(aL+bE)$, you can compute this using the projection formula. Namely, you have $\pi_*(\mathcal{O}_X(aL) \otimes \mathcal{O}_X(bE)) = \pi_*(\pi^*\mathcal{O}_{\mathbb{P}^2}(aL) \otimes \mathcal{O}_X(bE)) = \mathcal{O}_{\mathbb{P}^2}(aL) \otimes \pi_*\mathcal{O}_X(bE)$.

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.