Algebraic Geometry – Universal Property of Sheafification

algebraic-geometrysheaf-theory

Given a presheaf $\mathcal{F}$ there is a sheaf $\mathcal{F}^+$ and a morphism $\theta: \mathcal{F}\to\mathcal{F}^+$ with the property that for any sheaf $\mathcal{G}$ and any morphism $\varphi: \mathcal{F}\to \mathcal{G}$ there is a unique morphism $\overline{\varphi}:\mathcal{F}^+\to\mathcal{G}$ such that $\varphi=\overline{\varphi}\circ \theta$.

If $U$ is open I know how to construct $\mathcal{F}^+(U)$ as a set of functions $$f:U\to\bigcup_{p\in U}\mathcal{F}_p$$

such that $f(p)\in\mathcal{F}_p$ and for every $p\in U$ there is an open set $p\in V_p\subset U$ and $g\in \mathcal{F}(V_p)$ such that for all $q\in V_p$ the germ $g_q=f(q)$.

But I don't know how to construct the $\overline{\varphi}$ map.

Best Answer

If you have a map of presheaves $\varphi : \mathcal{F} \to \mathcal{G}$, this induces a map on stalks

$$ \varphi_p : \mathcal{F}_p \to \mathcal{G}_p $$

by the universal property of direct limits. Thus if we have an $f \in \mathcal{F}^+(U)$, we can compose $f$ with the map

$$ \bigsqcup_{p \in U} \varphi_p : \bigsqcup_{p \in U} \mathcal{F}_p \to \bigsqcup_{p \in U}\mathcal{G}_p $$

to get a map $U \to \bigsqcup_{p \in U} \mathcal{G}_p$. Call this $\tilde{\varphi}(f)$.Then we get a morphism $\tilde{\varphi}\colon\mathcal{F}^+\to \mathcal{G}^+$.

In fact, for every $p\in U$ there exists an open neighborhood $p \in V_p \subset U$ and $h \in \mathcal{F}(V_p)$ such that $h_q =f(q)$ for all $q \in V_p$ by definition of $\mathcal{F}^+$, and $\tilde{\varphi}(f)(q)=\varphi_q(f(q))=\varphi_q(h_q)=(\varphi(V_p)(h))_q$. So we can conclude $\tilde{\varphi}(f)\in \mathcal{G}^+(U)$.

Since $\mathcal{G}$ is assumed to be sheaf, then canonical morphism $\theta'\colon \mathcal{G}\to\mathcal{G}^+$ is isomorphic and so $\overline{\varphi}\colon={\theta'}^{-1}\circ\tilde{\varphi}$ gives a well defined morphism $\mathcal{F}^+ \to \mathcal{G}$.

Then we want $\overline{\varphi}\circ\theta=\varphi$.For this, take arbitary $x\in X$, then it is enough to see stalk at $x$; $\varphi_x={\theta'_x}^{-1}\circ\tilde{\varphi}_x\circ\theta_x$.This is easy to check.Uniqueness follows from the fact that $\theta_x$ is isomorphic for any $x\in X$.