# Global sections of projective schemes

Let $Y$ be a closed subscheme of $X = \operatorname{Proj} S$, where $S = k[x_0,\dots,x_n]$, $k$ algebraically closed. Then $Y = \operatorname{Proj} (S/I)$ for a homegenous ideal $I$ of $S$. How can we understand the global sections $\Gamma(Y,\mathcal{O}_Y(d))$ of the twisted sheaf $\mathcal{O}_Y(d)$ on $Y$? By Exercise II.5.9(b) in Hartshorne, we know that for large $d$ these sections coincide with the degree $d$ component of the ring $S/I$. How large does $d$ need to be for this to be true? Also, what about for smaller $d$?

• I believe that those numbers depend on the vanishing of the cohomology module $H^1(X, I_Y)$, and the Castelnuovo-Mumford regularity of Y may answer your questions. – Youngsu Mar 23 '18 at 8:39
• The issue is that i am trying to understand the relationship of $\Gamma(Y,\mathcal{O}_Y(d))$ with $S/I$. I can see that these two agree for large $d$ but i am not sure what $\Gamma(Y,\mathcal{O}_Y(d))$ is for small $d$. – Manos Mar 28 '18 at 5:39
• By the way, i am very familiar with Castelnuovo-Mumford regularity of modules (but not of schemes). – Manos Apr 1 '18 at 9:56

Since $(S/I)^\sim = \mathcal O_Y$, there is a natural map $S/I\to \Gamma_*(\mathcal O_Y) = \bigoplus_{d\in\mathbb Z}H^0(Y,\mathcal O_Y(d))$ which is an isomorphism in large degrees and whose kernel and co-kernel are given by the local cohomology of $S/I$ with respect to the irrelevant ideal. More precisely, the above map sits in an exact sequence $$0\to H^0_{\mathfrak m}(S/I)\to S/I\to \Gamma_*(\mathcal O_Y)\to H_{\mathfrak m}^1(S/I)\to 0.$$ This can be found in Eisenbud's Geometry of syzygies, Appendix 1 section A1B. In there, you can find more information on (the vanishing of) these local cohomology groups.