# Spectral sequence associated to elliptic fibration degenerates?

Let $\phi:S\rightarrow \mathbb{CP}^1$ be an elliptic fibration of a K3 surface. When is the Leray spectral sequence associated to the fibration $E_2$-degenerate? Are there any good criteria for the $E_2$-degeneration?

-

It's always true with $\mathbb{Q}$ coefficients. It follows from a general result of Zucker, who proved Leray degenerates whenever you have a projective map to curve. But in this case, it's simpler to check it by hand. There is only one (two !) differentials to worry about $$d_2:H^0(\mathbb{P}^1, R^1\phi_*\mathbb{Q})\to H^2(\mathbb{P}^1,\mathbb{Q})$$ But right side injects into $H^2(S)$, so this has to vanish, we also have $$d_2:H^0(\mathbb{P}^1, R^2\phi_*\mathbb{Q})\to H^2(\mathbb{P}^1,R^1\phi_*\mathbb{Q})$$ You get vanishing of the second map if you observe that the edge map from $H^2(S)$ to the left side is also surjective (you can see this if you identify it with the sum of $H^2$ of the fibres).
There should be two differentials to worry about, no? $H^0(\mathbf P^1,R^2 \phi_\ast \mathbf Q) \to H^2(\mathbf P^1,R^1\phi_\ast \mathbf Q)$ and $H^0(\mathbf P^1, R^1 \phi_\ast \mathbf Q) \to H^2(\mathbf P^1,R^0\phi_\ast \mathbf Q)$. –  Dan Petersen May 18 '14 at 17:31