# stacks/stacks-project

Switch branches/tags
Nothing to show
Fetching contributors…
Cannot retrieve contributors at this time
4864 lines (4317 sloc) 162 KB
 \input{preamble} % OK, start here. % \begin{document} \title{Sheaves of Modules} \maketitle \phantomsection \label{section-phantom} \tableofcontents \section{Introduction} \label{section-introduction} \noindent In this chapter we work out basic notions of sheaves of modules. This in particular includes the case of abelian sheaves, since these may be viewed as sheaves of $\underline{\mathbf{Z}}$-modules. Basic references are \cite{FAC}, \cite{EGA} and \cite{SGA4}. \medskip\noindent We work out what happens for sheaves of modules on ringed topoi in another chapter (see Modules on Sites, Section \ref{sites-modules-section-introduction}), although there we will mostly just duplicate the discussion from this chapter. \section{Pathology} \label{section-pathology} \noindent A ringed space is a pair consisting of a topological space $X$ and a sheaf of rings $\mathcal{O}$. We allow $\mathcal{O} = 0$ in the definition. In this case the category of modules has a single object (namely $0$). It is still an abelian category etc, but it is a little degenerate. Similarly the sheaf $\mathcal{O}$ may be zero over open subsets of $X$, etc. \medskip\noindent This doesn't happen when considering locally ringed spaces (as we will do later). \section{The abelian category of sheaves of modules} \label{section-kernels} \noindent Let $(X, \mathcal{O}_X)$ be a ringed space, see Sheaves, Definition \ref{sheaves-definition-ringed-space}. Let $\mathcal{F}$, $\mathcal{G}$ be sheaves of $\mathcal{O}_X$-modules, see Sheaves, Definition \ref{sheaves-definition-sheaf-modules}. Let $\varphi, \psi : \mathcal{F} \to \mathcal{G}$ be morphisms of sheaves of $\mathcal{O}_X$-modules. We define $\varphi + \psi : \mathcal{F} \to \mathcal{G}$ to be the map which on each open $U \subset X$ is the sum of the maps induced by $\varphi$, $\psi$. This is clearly again a map of sheaves of $\mathcal{O}_X$-modules. It is also clear that composition of maps of $\mathcal{O}_X$-modules is bilinear with respect to this addition. Thus $\textit{Mod}(\mathcal{O}_X)$ is a pre-additive category, see Homology, Definition \ref{homology-definition-preadditive}. \medskip\noindent We will denote $0$ the sheaf of $\mathcal{O}_X$-modules which has constant value $\{0\}$ for all open $U \subset X$. Clearly this is both a final and an initial object of $\textit{Mod}(\mathcal{O}_X)$. Given a morphism of $\mathcal{O}_X$-modules $\varphi : \mathcal{F} \to \mathcal{G}$ the following are equivalent: (a) $\varphi$ is zero, (b) $\varphi$ factors through $0$, (c) $\varphi$ is zero on sections over each open $U$, and (d) $\varphi_x = 0$ for all $x \in X$. See Sheaves, Lemma \ref{sheaves-lemma-points-exactness}. \medskip\noindent Moreover, given a pair $\mathcal{F}$, $\mathcal{G}$ of sheaves of $\mathcal{O}_X$-modules we may define the direct sum as $$\mathcal{F} \oplus \mathcal{G} = \mathcal{F} \times \mathcal{G}$$ with obvious maps $(i, j, p, q)$ as in Homology, Definition \ref{homology-definition-direct-sum}. Thus $\textit{Mod}(\mathcal{O}_X)$ is an additive category, see Homology, Definition \ref{homology-definition-additive-category}. \medskip\noindent Let $\varphi : \mathcal{F} \to \mathcal{G}$ be a morphism of $\mathcal{O}_X$-modules. We may define $\Ker(\varphi)$ to be the subsheaf of $\mathcal{F}$ with sections $$\Ker(\varphi)(U) = \{ s \in \mathcal{F}(U) \mid \varphi(s) = 0 \text{ in } \mathcal{G}(U)\}$$ for all open $U \subset X$. It is easy to see that this is indeed a kernel in the category of $\mathcal{O}_X$-modules. In other words, a morphism $\alpha : \mathcal{H} \to \mathcal{F}$ factors through $\Ker(\varphi)$ if and only if $\varphi \circ \alpha = 0$. Moreover, on the level of stalks we have $\Ker(\varphi)_x = \Ker(\varphi_x)$. \medskip\noindent On the other hand, we define $\Coker(\varphi)$ as the sheaf of $\mathcal{O}_X$-modules associated to the presheaf of $\mathcal{O}_X$-modules defined by the rule $$U \longmapsto \Coker(\mathcal{G}(U)\to \mathcal{F}(U)) = \mathcal{F}(U)/\varphi(\mathcal{G}(U)).$$ Since taking stalks commutes with taking sheafification, see Sheaves, Lemma \ref{sheaves-lemma-stalk-sheafification} we see that $\Coker(\varphi)_x = \Coker(\varphi_x)$. Thus the map $\mathcal{G} \to \Coker(\varphi)$ is surjective (as a map of sheaves of sets), see Sheaves, Section \ref{sheaves-section-exactness-points}. To show that this is a cokernel, note that if $\beta : \mathcal{G} \to \mathcal{H}$ is a morphism of $\mathcal{O}_X$-modules such that $\beta \circ \varphi$ is zero, then you get for every open $U \subset X$ a map induced by $\beta$ from $\mathcal{G}(U)/\varphi(\mathcal{F}(U))$ into $\mathcal{H}(U)$. By the universal property of sheafification (see Sheaves, Lemma \ref{sheaves-lemma-sheafification-presheaf-modules}) we obtain a canonical map $\Coker(\varphi) \to \mathcal{H}$ such that the original $\beta$ is equal to the composition $\mathcal{G} \to \Coker(\varphi) \to \mathcal{H}$. The morphism $\Coker(\varphi) \to \mathcal{H}$ is unique because of the surjectivity mentioned above. \begin{lemma} \label{lemma-abelian} Let $(X, \mathcal{O}_X)$ be a ringed space. The category $\textit{Mod}(\mathcal{O}_X)$ is an abelian category. Moreover a complex $$\mathcal{F} \to \mathcal{G} \to \mathcal{H}$$ is exact at $\mathcal{G}$ if and only if for all $x \in X$ the complex $$\mathcal{F}_x \to \mathcal{G}_x \to \mathcal{H}_x$$ is exact at $\mathcal{G}_x$. \end{lemma} \begin{proof} By Homology, Definition \ref{homology-definition-abelian-category} we have to show that image and coimage agree. By Sheaves, Lemma \ref{sheaves-lemma-points-exactness} it is enough to show that image and coimage have the same stalk at every $x \in X$. By the constructions of kernels and cokernels above these stalks are the coimage and image in the categories of $\mathcal{O}_{X, x}$-modules. Thus we get the result from the fact that the category of modules over a ring is abelian. \end{proof} \noindent Actually the category $\textit{Mod}(\mathcal{O}_X)$ has many more properties. Here are two constructions we can do. \begin{enumerate} \item Given any set $I$ and for each $i \in I$ a $\mathcal{O}_X$-module we can form the product $$\prod\nolimits_{i \in I} \mathcal{F}_i$$ which is the sheaf that associates to each open $U$ the product of the modules $\mathcal{F}_i(U)$. This is also the categorical product, as in Categories, Definition \ref{categories-definition-product}. \item Given any set $I$ and for each $i \in I$ a $\mathcal{O}_X$-module we can form the direct sum $$\bigoplus\nolimits_{i \in I} \mathcal{F}_i$$ which is the {\it sheafification} of the presheaf that associates to each open $U$ the direct sum of the modules $\mathcal{F}_i(U)$. This is also the categorical coproduct, as in Categories, Definition \ref{categories-definition-coproduct}. To see this you use the universal property of sheafification. \end{enumerate} Using these we conclude that all limits and colimits exist in $\textit{Mod}(\mathcal{O}_X)$. \begin{lemma} \label{lemma-limits-colimits} Let $(X, \mathcal{O}_X)$ be a ringed space. \begin{enumerate} \item All limits exist in $\textit{Mod}(\mathcal{O}_X)$. Limits are the same as the corresponding limits of presheaves of $\mathcal{O}_X$-modules (i.e., commute with taking sections over opens). \item All colimits exist in $\textit{Mod}(\mathcal{O}_X)$. Colimits are the sheafification of the corresponding colimit in the category of presheaves. Taking colimits commutes with taking stalks. \item Filtered colimits are exact. \item Finite direct sums are the same as the corresponding finite direct sums of presheaves of $\mathcal{O}_X$-modules. \end{enumerate} \end{lemma} \begin{proof} As $\textit{Mod}(\mathcal{O}_X)$ is abelian (Lemma \ref{lemma-abelian}) it has all finite limits and colimits (Homology, Lemma \ref{homology-lemma-colimit-abelian-category}). Thus the existence of limits and colimits and their description follows from the existence of products and coproducts and their description (see discussion above) and Categories, Lemmas \ref{categories-lemma-limits-products-equalizers} and \ref{categories-lemma-colimits-coproducts-coequalizers}. Since sheafification commutes with taking stalks we see that colimits commute with taking stalks. Part (3) signifies that given a system $0 \to \mathcal{F}_i \to \mathcal{G}_i \to \mathcal{H}_i \to 0$ of exact sequences of $\mathcal{O}_X$-modules over a directed set $I$ the sequence $0 \to \colim \mathcal{F}_i \to \colim \mathcal{G}_i \to \colim \mathcal{H}_i \to 0$ is exact as well. Since we can check exactness on stalks (Lemma \ref{lemma-abelian}) this follows from the case of modules which is Algebra, Lemma \ref{algebra-lemma-directed-colimit-exact}. We omit the proof of (4). \end{proof} \noindent The existence of limits and colimits allows us to consider exactness properties of functors defined on the category of $\mathcal{O}$-modules in terms of limits and colimits, as in Categories, Section \ref{categories-section-exact-functor}. See Homology, Lemma \ref{homology-lemma-exact-functor} for a description of exactness properties in terms of short exact sequences. \begin{lemma} \label{lemma-exactness-pushforward-pullback} Let $f : (X, \mathcal{O}_X) \to (Y, \mathcal{O}_Y)$ be a morphism of ringed spaces. \begin{enumerate} \item The functor $f_* : \textit{Mod}(\mathcal{O}_X) \to \textit{Mod}(\mathcal{O}_Y)$ is left exact. In fact it commutes with all limits. \item The functor $f^* : \textit{Mod}(\mathcal{O}_Y) \to \textit{Mod}(\mathcal{O}_X)$ is right exact. In fact it commutes with all colimits. \item Pullback $f^{-1} : \textit{Ab}(Y) \to \textit{Ab}(X)$ on abelian sheaves is exact. \end{enumerate} \end{lemma} \begin{proof} Parts (1) and (2) hold because $(f^*, f_*)$ is an adjoint pair of functors, see Sheaves, Lemma \ref{sheaves-lemma-adjoint-pullback-pushforward-modules} and Categories, Section \ref{categories-section-adjoint}. Part (3) holds because exactness can be checked on stalks (Lemma \ref{lemma-abelian}) and the description of stalks of the pullback, see Sheaves, Lemma \ref{sheaves-lemma-pullback-abelian-stalk}. \end{proof} \begin{lemma} \label{lemma-j-shriek-exact} Let $j : U \to X$ be an open immersion of topological spaces. The functor $j_! : \textit{Ab}(U) \to \textit{Ab}(X)$ is exact. \end{lemma} \begin{proof} Follows from the description of stalks given in Sheaves, Lemma \ref{sheaves-lemma-j-shriek-abelian}. \end{proof} \begin{lemma} \label{lemma-section-direct-sum-quasi-compact} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $I$ be a set. For $i \in I$, let $\mathcal{F}_i$ be a sheaf of $\mathcal{O}_X$-modules. For $U \subset X$ quasi-compact open the map $$\bigoplus\nolimits_{i \in I} \mathcal{F}_i(U) \longrightarrow \left(\bigoplus\nolimits_{i \in I} \mathcal{F}_i\right)(U)$$ is bijective. \end{lemma} \begin{proof} If $s$ is an element of the right hand side, then there exists an open covering $U = \bigcup_{j \in J} U_j$ such that $s|_{U_j}$ is a finite sum $\sum_{i \in I_j} s_{ji}$ with $s_{ji} \in \mathcal{F}_i(U_j)$. Because $U$ is quasi-compact we may assume that the covering is finite, i.e., that $J$ is finite. Then $I' = \bigcup_{j \in J} I_j$ is a finite subset of $I$. Clearly, $s$ is a section of the subsheaf $\bigoplus_{i \in I'} \mathcal{F}_i$. The result follows from the fact that for a finite direct sum sheafification is not needed, see Lemma \ref{lemma-limits-colimits} above. \end{proof} \section{Sections of sheaves of modules} \label{section-sections} \noindent Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. Let $s \in \Gamma(X, \mathcal{F}) = \mathcal{F}(X)$ be a global section. There is a unique {\it map of $\mathcal{O}_X$-modules $$\mathcal{O}_X \longrightarrow \mathcal{F}, \ f \longmapsto fs$$ associated to $s$}. The notation above signifies that a local section $f$ of $\mathcal{O}_X$, i.e., a section $f$ over some open $U$, is mapped to the multiplication of $f$ with the restriction of $s$ to $U$. Conversely, any map $\varphi : \mathcal{O}_X \to \mathcal{F}$ gives rise to a section $s = \varphi(1)$ such that $\varphi$ is the morphism associated to $s$. \begin{definition} \label{definition-globally-generated} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. We say that $\mathcal{F}$ is {\it generated by global sections} if there exist a set $I$, and global sections $s_i \in \Gamma(X, \mathcal{F})$, $i \in I$ such that the map $$\bigoplus\nolimits_{i \in I} \mathcal{O}_X \longrightarrow \mathcal{F}$$ which is the map associated to $s_i$ on the summand corresponding to $i$, is surjective. In this case we say that the sections $s_i$ {\it generate} $\mathcal{F}$. \end{definition} \noindent We often use the abuse of notation introduced in Sheaves, Section \ref{sheaves-section-stalks} where, given a local section $s$ of $\mathcal{F}$ defined in an open neighbourhood of a point $x \in X$, we denote $s_x$, or even $s$ the image of $s$ in the stalk $\mathcal{F}_x$. \begin{lemma} \label{lemma-globally-generated} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. Let $I$ be a set. Let $s_i \in \Gamma(X, \mathcal{F})$, $i \in I$ be global sections. The sections $s_i$ generate $\mathcal{F}$ if and only if for all $x\in X$ the elements $s_{i, x} \in \mathcal{F}_x$ generate the $\mathcal{O}_{X, x}$-module $\mathcal{F}_x$. \end{lemma} \begin{proof} Omitted. \end{proof} \begin{lemma} \label{lemma-tensor-product-globally-generated} \begin{slogan} The tensor product of globally generated sheaves of modules is globally generated. \end{slogan} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$, $\mathcal{G}$ be sheaves of $\mathcal{O}_X$-modules. If $\mathcal{F}$ and $\mathcal{G}$ are generated by global sections then so is $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}$. \end{lemma} \begin{proof} Omitted. \end{proof} \begin{lemma} \label{lemma-generated-by-local-sections} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. Let $I$ be a set. Let $s_i$, $i \in I$ be a collection of local sections of $\mathcal{F}$, i.e., $s_i \in \mathcal{F}(U_i)$ for some opens $U_i \subset X$. There exists a unique smallest subsheaf of $\mathcal{O}_X$-modules $\mathcal{G}$ such that each $s_i$ corresponds to a local section of $\mathcal{G}$. \end{lemma} \begin{proof} Consider the subpresheaf of $\mathcal{O}_X$-modules defined by the rule $$U \longmapsto \{ \text{sums } \sum\nolimits_{i \in J} f_i (s_i|_U) \text{ where } J \text{ is finite, } U \subset U_i \text{ for } i\in J, \text{ and } f_i \in \mathcal{O}_X(U) \}$$ Let $\mathcal{G}$ be the sheafification of this subpresheaf. This is a subsheaf of $\mathcal{F}$ by Sheaves, Lemma \ref{sheaves-lemma-characterize-epi-mono}. Since all the finite sums clearly have to be in $\mathcal{G}$ this is the smallest subsheaf as desired. \end{proof} \begin{definition} \label{definition-generated-by-local-sections} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. Given a set $I$, and local sections $s_i$, $i \in I$ of $\mathcal{F}$ we say that the subsheaf $\mathcal{G}$ of Lemma \ref{lemma-generated-by-local-sections} above is the {\it subsheaf generated by the $s_i$}. \end{definition} \begin{lemma} \label{lemma-generated-by-local-sections-stalk} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. Given a set $I$, and local sections $s_i$, $i \in I$ of $\mathcal{F}$. Let $\mathcal{G}$ be the subsheaf generated by the $s_i$ and let $x\in X$. Then $\mathcal{G}_x$ is the $\mathcal{O}_{X, x}$-submodule of $\mathcal{F}_x$ generated by the elements $s_{i, x}$ for those $i$ such that $s_i$ is defined at $x$. \end{lemma} \begin{proof} This is clear from the construction of $\mathcal{G}$ in the proof of Lemma \ref{lemma-generated-by-local-sections}. \end{proof} \section{Supports of modules and sections} \label{section-support} \begin{definition} \label{definition-support} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. \begin{enumerate} \item The {\it support of $\mathcal{F}$} is the set of points $x \in X$ such that $\mathcal{F}_x \not = 0$. \item We denote $\text{Supp}(\mathcal{F})$ the support of $\mathcal{F}$. \item Let $s \in \Gamma(X, \mathcal{F})$ be a global section. The {\it support of $s$} is the set of points $x \in X$ such that the image $s_x \in \mathcal{F}_x$ of $s$ is not zero. \end{enumerate} \end{definition} \noindent Of course the support of a local section is then defined also since a local section is a global section of the restriction of $\mathcal{F}$. \begin{lemma} \label{lemma-support-section-closed} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. Let $U \subset X$ open. \begin{enumerate} \item The support of $s \in \mathcal{F}(U)$ is closed in $U$. \item The support of $fs$ is contained in the intersections of the supports of $f \in \mathcal{O}_X(U)$ and $s \in \mathcal{F}(U)$. \item The support of $s + s'$ is contained in the union of the supports of $s, s' \in \mathcal{F}(U)$. \item The support of $\mathcal{F}$ is the union of the supports of all local sections of $\mathcal{F}$. \item If $\varphi : \mathcal{F} \to \mathcal{G}$ is a morphism of $\mathcal{O}_X$-modules, then the support of $\varphi(s)$ is contained in the support of $s \in \mathcal{F}(U)$. \end{enumerate} \end{lemma} \begin{proof} This is true because if $s_x = 0$, then $s$ is zero in an open neighbourhood of $x$ by definition of stalks. Similarly for $f$. Details omitted. \end{proof} \noindent In general the support of a sheaf of modules is not closed. Namely, the sheaf could be an abelian sheaf on $\mathbf{R}$ (with the usual archimedean topology) which is the direct sum of infinitely many nonzero skyscraper sheaves each supported at a single point $p_i$ of $\mathbf{R}$. Then the support would be the set of points $p_i$ which may not be closed. \medskip\noindent Another example is to consider the open immersion $j : U = (0 , \infty) \to \mathbf{R} = X$, and the abelian sheaf $j_!\underline{\mathbf{Z}}_U$. By Sheaves, Section \ref{sheaves-section-open-immersions} the support of this sheaf is exactly $U$. \begin{lemma} \label{lemma-support-sheaf-rings-closed} Let $X$ be a topological space. The support of a sheaf of rings is closed. \end{lemma} \begin{proof} This is true because (according to our conventions) a ring is $0$ if and only if $1 = 0$, and hence the support of a sheaf of rings is the support of the unit section. \end{proof} \section{Closed immersions and abelian sheaves} \label{section-closed-immersions} \noindent Recall that we think of an abelian sheaf on a topological space $X$ as a sheaf of $\underline{\mathbf{Z}}_X$-modules. Thus we may apply any results, definitions for sheaves of modules to abelian sheaves. \begin{lemma} \label{lemma-i-star-exact} Let $X$ be a topological space. Let $Z \subset X$ be a closed subset. Denote $i : Z \to X$ the inclusion map. The functor $$i_* : \textit{Ab}(Z) \longrightarrow \textit{Ab}(X)$$ is exact, fully faithful, with essential image exactly those abelian sheaves whose support is contained in $Z$. The functor $i^{-1}$ is a left inverse to $i_*$. \end{lemma} \begin{proof} Exactness follows from the description of stalks in Sheaves, Lemma \ref{sheaves-lemma-stalks-closed-pushforward} and Lemma \ref{lemma-abelian}. The rest was shown in Sheaves, Lemma \ref{sheaves-lemma-equivalence-categories-closed-abelian}. \end{proof} \noindent Let $\mathcal{F}$ be a sheaf on $X$. There is a canonical subsheaf of $\mathcal{F}$ which consists of exactly those sections whose support is contained in $Z$. Here is the exact statement. \begin{lemma} \label{lemma-sections-support-in-closed} Let $X$ be a topological space. Let $Z \subset X$ be a closed subset. Let $\mathcal{F}$ be a sheaf on $X$. For $U \subset X$ open set $$\Gamma(U, \mathcal{H}_Z(\mathcal{F})) = \{s \in \mathcal{F}(U) \mid \text{ the support of }s\text{ is contained in }Z \cap U\}$$ Then $\mathcal{H}_Z(\mathcal{F})$ is an abelian subsheaf of $\mathcal{F}$. It is the largest abelian subsheaf of $\mathcal{F}$ whose support is contained in $Z$. The construction $\mathcal{F} \mapsto \mathcal{H}_Z(\mathcal{F})$ is functorial in the abelian sheaf $\mathcal{F}$. \end{lemma} \begin{proof} This follows from Lemma \ref{lemma-support-section-closed}. \end{proof} \noindent This seems like a good opportunity to show that the functor $i_*$ has a right adjoint on abelian sheaves. \begin{lemma} \label{lemma-i-star-right-adjoint} Let $i : Z \to X$ be the inclusion of a closed subset into the topological space $X$. Denote\footnote{This is likely nonstandard notation.} $i^! : \textit{Ab}(X) \to \textit{Ab}(Z)$ the functor $\mathcal{F} \mapsto i^{-1}\mathcal{H}_Z(\mathcal{F})$. Then $i^!$ is a right adjoint to $i_*$, in a formula $$\Mor_{\textit{Ab}(X)}(i_*\mathcal{G}, \mathcal{F}) = \Mor_{\textit{Ab}(Z)}(\mathcal{G}, i^!\mathcal{F}).$$ In particular $i_*$ commutes with arbitrary colimits. \end{lemma} \begin{proof} Note that $i_*i^!\mathcal{F} = \mathcal{H}_Z(\mathcal{F})$. Since $i_*$ is fully faithful we are reduced to showing that $$\Mor_{\textit{Ab}(X)}(i_*\mathcal{G}, \mathcal{F}) = \Mor_{\textit{Ab}(X)}(i_*\mathcal{G}, \mathcal{H}_Z(\mathcal{F})).$$ This follows since the support of the image via any homomorphism of a section of $i_*\mathcal{G}$ is contained in $Z$, see Lemma \ref{lemma-support-section-closed}. \end{proof} \begin{remark} \label{remark-i-star-right-adjoint} In Sheaves, Remark \ref{sheaves-remark-i-star-not-exact} we showed that $i_*$ as a functor on the categories of sheaves of sets does not have a right adjoint simply because it is not exact. However, it is very close to being true, in fact, the functor $i_*$ is exact on sheaves of pointed sets, sections with support in $Z$ can be defined for sheaves of pointed sets, and $i^!$ makes sense and is a right adjoint to $i_*$. \end{remark} \section{A canonical exact sequence} \label{section-canonical-exact-sequence} \noindent We give this exact sequence its own section. \begin{lemma} \label{lemma-canonical-exact-sequence} Let $X$ be a topological space. Let $U \subset X$ be an open subset with complement $Z \subset X$. Denote $j : U \to X$ the open immersion and $i : Z \to X$ the closed immersion. For any sheaf of abelian groups $\mathcal{F}$ on $X$ the adjunction mappings $j_{!}j^*\mathcal{F} \to \mathcal{F}$ and $\mathcal{F} \to i_*i^*\mathcal{F}$ give a short exact sequence $$0 \to j_{!}j^*\mathcal{F} \to \mathcal{F} \to i_*i^*\mathcal{F} \to 0$$ of sheaves of abelian groups. For any morphism $\varphi : \mathcal{F} \to \mathcal{G}$ of abelian sheaves on $X$ we obtain a morphism of short exact sequences $$\xymatrix{ 0 \ar[r] & j_{!}j^*\mathcal{F} \ar[r] \ar[d] & \mathcal{F} \ar[r] \ar[d] & i_*i^*\mathcal{F} \ar[r] \ar[d] & 0 \\ 0 \ar[r] & j_{!}j^*\mathcal{G} \ar[r] & \mathcal{G} \ar[r] & i_*i^*\mathcal{G} \ar[r] & 0 }$$ \end{lemma} \begin{proof} The functoriality of the short exact sequence is immediate from the naturality of the adjunction mappings. We may check exactness on stalks (Lemma \ref{lemma-abelian}). For a description of the stalks in question see Sheaves, Lemmas \ref{sheaves-lemma-j-shriek-abelian} and \ref{sheaves-lemma-stalks-closed-pushforward}. \end{proof} \section{Modules locally generated by sections} \label{section-locally-generated} \noindent Let $(X, \mathcal{O}_X)$ be a ringed space. In this and the following section we will often restrict sheaves to open subspaces $U \subset X$, see Sheaves, Section \ref{sheaves-section-open-immersions}. In particular, we will often denote the open subspace by $(U, \mathcal{O}_U)$ instead of the more correct notation $(U, \mathcal{O}_X|_U)$, see Sheaves, Definition \ref{sheaves-definition-restriction}. \medskip\noindent Consider the open immersion $j : U = (0 , \infty) \to \mathbf{R} = X$, and the abelian sheaf $j_!\underline{\mathbf{Z}}_U$. By Sheaves, Section \ref{sheaves-section-open-immersions} the stalk of $j_!\underline{\mathbf{Z}}_U$ at $x = 0$ is $0$. In fact the sections of this sheaf over any open interval containing $0$ are $0$. Thus there is no open neighbourhood of the point $0$ over which the sheaf can be generated by sections. \begin{definition} \label{definition-locally-generated} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. We say that $\mathcal{F}$ is {\it locally generated by sections} if for every $x \in X$ there exists an open neighbourhood $U$ such that $\mathcal{F}|_U$ is globally generated as a sheaf of $\mathcal{O}_U$-modules. \end{definition} \noindent In other words there exists a set $I$ and for each $i$ a section $s_i \in \mathcal{F}(U)$ such that the associated map $$\bigoplus\nolimits_{i \in I} \mathcal{O}_U \longrightarrow \mathcal{F}|_U$$ is surjective. \begin{lemma} \label{lemma-pullback-locally-generated} Let $f : (X, \mathcal{O}_X) \to (Y, \mathcal{O}_Y)$ be a morphism of ringed spaces. The pullback $f^*\mathcal{G}$ is locally generated by sections if $\mathcal{G}$ is locally generated by sections. \end{lemma} \begin{proof} Given an open subspace $V$ of $Y$ we may consider the commutative diagram of ringed spaces $$\xymatrix{ (f^{-1}V, \mathcal{O}_{f^{-1}V}) \ar[r]_{j'} \ar[d]_{f'} & (X, \mathcal{O}_X) \ar[d]^f \\ (V, \mathcal{O}_V) \ar[r]^j & (Y, \mathcal{O}_Y) }$$ We know that $f^*\mathcal{G}|_{f^{-1}V} \cong (f')^*(\mathcal{G}|_V)$, see Sheaves, Lemma \ref{sheaves-lemma-push-pull-composition-modules}. Thus we may assume that $\mathcal{G}$ is globally generated. \medskip\noindent We have seen that $f^*$ commutes with all colimits, and is right exact, see Lemma \ref{lemma-exactness-pushforward-pullback}. Thus if we have a surjection $$\bigoplus\nolimits_{i \in I} \mathcal{O}_Y \to \mathcal{G} \to 0$$ then upon applying $f^*$ we obtain the surjection $$\bigoplus\nolimits_{i \in I} \mathcal{O}_X \to f^*\mathcal{G} \to 0.$$ This implies the lemma. \end{proof} \section{Modules of finite type} \label{section-finite-type} \begin{definition} \label{definition-finite-type} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. We say that $\mathcal{F}$ is of {\it finite type} if for every $x \in X$ there exists an open neighbourhood $U$ such that $\mathcal{F}|_U$ is generated by finitely many sections. \end{definition} \begin{lemma} \label{lemma-pullback-finite-type} Let $f : (X, \mathcal{O}_X) \to (Y, \mathcal{O}_Y)$ be a morphism of ringed spaces. The pullback $f^*\mathcal{G}$ of a finite type $\mathcal{O}_Y$-module is a finite type $\mathcal{O}_X$-module. \end{lemma} \begin{proof} Arguing as in the proof of Lemma \ref{lemma-pullback-locally-generated} we may assume $\mathcal{G}$ is globally generated by finitely many sections. We have seen that $f^*$ commutes with all colimits, and is right exact, see Lemma \ref{lemma-exactness-pushforward-pullback}. Thus if we have a surjection $$\bigoplus\nolimits_{i = 1, \ldots, n} \mathcal{O}_Y \to \mathcal{G} \to 0$$ then upon applying $f^*$ we obtain the surjection $$\bigoplus\nolimits_{i = 1, \ldots, n} \mathcal{O}_X \to f^*\mathcal{G} \to 0.$$ This implies the lemma. \end{proof} \begin{lemma} \label{lemma-extension-finite-type} Let $X$ be a ringed space. The image of a morphism of $\mathcal{O}_X$-modules of finite type is of finite type. Let $0 \to \mathcal{F}_1 \to \mathcal{F}_2 \to \mathcal{F}_3 \to 0$ be a short exact sequence of $\mathcal{O}_X$-modules. If $\mathcal{F}_1$ and $\mathcal{F}_3$ are of finite type, so is $\mathcal{F}_2$. \end{lemma} \begin{proof} The statement on images is trivial. The statement on short exact sequences comes from the fact that sections of $\mathcal{F}_3$ locally lift to sections of $\mathcal{F}_2$ and the corresponding result in the category of modules over a ring (applied to the stalks for example). \end{proof} \begin{lemma} \label{lemma-finite-type-surjective-on-stalk} Let $X$ be a ringed space. Let $\varphi : \mathcal{G} \to \mathcal{F}$ be a homomorphism of $\mathcal{O}_X$-modules. Let $x \in X$. Assume $\mathcal{F}$ of finite type and the map on stalks $\varphi_x : \mathcal{G}_x \to \mathcal{F}_x$ surjective. Then there exists an open neighbourhood $x \in U \subset X$ such that $\varphi|_U$ is surjective. \end{lemma} \begin{proof} Choose an open neighbourhood $U \subset X$ of $x$ such that $\mathcal{F}$ is generated by $s_1, \ldots, s_n \in \mathcal{F}(U)$ over $U$. By assumption of surjectivity of $\varphi_x$, after shrinking $U$ we may assume that $s_i = \varphi(t_i)$ for some $t_i \in \mathcal{G}(U)$. Then $U$ works. \end{proof} \begin{lemma} \label{lemma-finite-type-stalk-zero} Let $X$ be a ringed space. Let $\mathcal{F}$ be an $\mathcal{O}_X$-module. Let $x \in X$. Assume $\mathcal{F}$ of finite type and $\mathcal{F}_x = 0$. Then there exists an open neighbourhood $x \in U \subset X$ such that $\mathcal{F}|_U$ is zero. \end{lemma} \begin{proof} This is a special case of Lemma \ref{lemma-finite-type-surjective-on-stalk} applied to the morphism $0 \to \mathcal{F}$. \end{proof} \begin{lemma} \label{lemma-support-finite-type-closed} \begin{slogan} Over any ringed space, sheaves of modules of finite type have closed support. \end{slogan} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. If $\mathcal{F}$ is of finite type then support of $\mathcal{F}$ is closed. \end{lemma} \begin{proof} This is a reformulation of Lemma \ref{lemma-finite-type-stalk-zero}. \end{proof} \begin{lemma} \label{lemma-finite-type-quasi-compact-colimit} Let $X$ be a ringed space. Let $I$ be a preordered set and let $(\mathcal{F}_i, f_{ii'})$ be a system over $I$ consisting of sheaves of $\mathcal{O}_X$-modules (see Categories, Section \ref{categories-section-posets-limits}). Let $\mathcal{F} = \colim \mathcal{F}_i$ be the colimit. Assume (a) $I$ is directed, (b) $\mathcal{F}$ is a finite type $\mathcal{O}_X$-module, and (c) $X$ is quasi-compact. Then there exists an $i$ such that $\mathcal{F}_i \to \mathcal{F}$ is surjective. If the transition maps $f_{ii'}$ are injective then we conclude that $\mathcal{F} = \mathcal{F}_i$ for some $i \in I$. \end{lemma} \begin{proof} Let $x \in X$. There exists an open neighbourhood $U \subset X$ of $x$ and finitely many sections $s_j \in \mathcal{F}(U)$, $j = 1, \ldots, m$ such that $s_1, \ldots, s_m$ generate $\mathcal{F}$ as $\mathcal{O}_U$-module. After possibly shrinking $U$ to a smaller open neighbourhood of $x$ we may assume that each $s_j$ comes from a section of $\mathcal{F}_i$ for some $i \in I$. Hence, since $X$ is quasi-compact we can find a finite open covering $X = \bigcup_{j = 1, \ldots, m} U_j$, and for each $j$ an index $i_j$ and finitely many sections $s_{jl} \in \mathcal{F}_{i_j}(U_j)$ whose images generate the restriction of $\mathcal{F}$ to $U_j$. Clearly, the lemma holds for any index $i \in I$ which is $\geq$ all $i_j$. \end{proof} \begin{lemma} \label{lemma-set-isomorphism-classes-finite-type-modules} Let $X$ be a ringed space. There exists a set of $\mathcal{O}_X$-modules $\{\mathcal{F}_i\}_{i \in I}$ of finite type such that each finite type $\mathcal{O}_X$-module on $X$ is isomorphic to exactly one of the $\mathcal{F}_i$. \end{lemma} \begin{proof} For each open covering $\mathcal{U} : X = \bigcup U_j$ consider the sheaves of $\mathcal{O}_X$-modules $\mathcal{F}$ such that each restriction $\mathcal{F}|_{U_j}$ is a quotient of $\mathcal{O}_{U_j}^{\oplus r}$ for some $r_j \geq 0$. These are parametrized by subsheaves $\mathcal{K}_i \subset \mathcal{O}_{U_j}^{\oplus r_j}$ and glueing data $$\varphi_{jj'} : \mathcal{O}_{U_j \cap U_{j'}}^{\oplus r_j}/ (\mathcal{K}_j|_{U_j \cap U_{j'}}) \longrightarrow \mathcal{O}_{U_j \cap U_{j'}}^{\oplus r_{j'}}/ (\mathcal{K}_{j'}|_{U_j \cap U_{j'}})$$ see Sheaves, Section \ref{sheaves-section-glueing-sheaves}. Note that the collection of all glueing data forms a set. The collection of all coverings $\mathcal{U} : X = \bigcup_{j \in J} U_i$ where $J \to \mathcal{P}(X)$, $j \mapsto U_j$ is injective forms a set as well. Hence the collection of all sheaves of $\mathcal{O}_X$-modules gotten from glueing quotients as above forms a set $\mathcal{I}$. By definition every finite type $\mathcal{O}_X$-module is isomorphic to an element of $\mathcal{I}$. Choosing an element out of each isomorphism class inside $\mathcal{I}$ gives the desired set of sheaves (uses axiom of choice). \end{proof} \section{Quasi-coherent modules} \label{section-quasi-coherent} \noindent In this section we introduce an abstract notion of quasi-coherent $\mathcal{O}_X$-module. This notion is very useful in algebraic geometry, since quasi-coherent modules on a scheme have a good description on any affine open. However, we warn the reader that in the general setting of (locally) ringed spaces this notion is not well behaved at all. The category of quasi-coherent sheaves is not abelian in general, infinite direct sums of quasi-coherent sheaves aren't quasi-coherent, etc, etc. \begin{definition} \label{definition-quasi-coherent} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. We say that $\mathcal{F}$ is a {\it quasi-coherent sheaf of $\mathcal{O}_X$-modules} if for every point $x \in X$ there exists an open neighbourhood $x\in U \subset X$ such that $\mathcal{F}|_U$ is isomorphic to the cokernel of a map $$\bigoplus\nolimits_{j \in J} \mathcal{O}_U \longrightarrow \bigoplus\nolimits_{i \in I} \mathcal{O}_U$$ The category of quasi-coherent $\mathcal{O}_X$-modules is denoted $\QCoh(\mathcal{O}_X)$. \end{definition} \noindent The definition means that $X$ is covered by open sets $U$ such that $\mathcal{F}|_U$ has a {\it presentation} of the form $$\bigoplus\nolimits_{j \in J} \mathcal{O}_U \longrightarrow \bigoplus\nolimits_{i \in I} \mathcal{O}_U \longrightarrow \mathcal{F}|_U \longrightarrow 0.$$ Here presentation signifies that the displayed sequence is exact. In other words \begin{enumerate} \item for every point $x$ of $X$ there exists an open neighbourhood such that $\mathcal{F}|_U$ is generated by global sections, and \item for a suitable choice of these sections the kernel of the associated surjection is also generated by global sections. \end{enumerate} \begin{lemma} \label{lemma-direct-sum-quasi-coherent} Let $(X, \mathcal{O}_X)$ be a ringed space. The direct sum of two quasi-coherent $\mathcal{O}_X$-modules is a quasi-coherent $\mathcal{O}_X$-module. \end{lemma} \begin{proof} Omitted. \end{proof} \begin{remark} \label{remark-infinite-direct-sum-quasi-coherent-not} Warning: It is not true in general that an infinite direct sum of quasi-coherent $\mathcal{O}_X$-modules is quasi-coherent. For more esoteric behaviour of quasi-coherent modules see Example \ref{example-quasi-coherent}. \end{remark} \begin{lemma} \label{lemma-pullback-quasi-coherent} Let $f : (X, \mathcal{O}_X) \to (Y, \mathcal{O}_Y)$ be a morphism of ringed spaces. The pullback $f^*\mathcal{G}$ of a quasi-coherent $\mathcal{O}_Y$-module is quasi-coherent. \end{lemma} \begin{proof} Arguing as in the proof of Lemma \ref{lemma-pullback-locally-generated} we may assume $\mathcal{G}$ has a global presentation by direct sums of copies of $\mathcal{O}_Y$. We have seen that $f^*$ commutes with all colimits, and is right exact, see Lemma \ref{lemma-exactness-pushforward-pullback}. Thus if we have an exact sequence $$\bigoplus\nolimits_{j \in J} \mathcal{O}_Y \longrightarrow \bigoplus\nolimits_{i \in I} \mathcal{O}_Y \longrightarrow \mathcal{G} \longrightarrow 0$$ then upon applying $f^*$ we obtain the exact sequence $$\bigoplus\nolimits_{j \in J} \mathcal{O}_X \longrightarrow \bigoplus\nolimits_{i \in I} \mathcal{O}_X \longrightarrow f^*\mathcal{G} \longrightarrow 0.$$ This implies the lemma. \end{proof} \noindent This gives plenty of examples of quasi-coherent sheaves. \begin{lemma} \label{lemma-construct-quasi-coherent-sheaves} Let $(X, \mathcal{O}_X)$ be ringed space. Let $\alpha : R \to \Gamma(X, \mathcal{O}_X)$ be a ring homomorphism from a ring $R$ into the ring of global sections on $X$. Let $M$ be an $R$-module. The following three constructions give canonically isomorphic sheaves of $\mathcal{O}_X$-modules: \begin{enumerate} \item Let $\pi : (X, \mathcal{O}_X) \longrightarrow (\{*\}, R)$ be the morphism of ringed spaces with $\pi : X \to \{*\}$ the unique map and with $\pi$-map $\pi^\sharp$ the given map $\alpha : R \to \Gamma(X, \mathcal{O}_X)$. Set $\mathcal{F}_1 = \pi^*M$. \item Choose a presentation $\bigoplus_{j \in J} R \to \bigoplus_{i \in I} R \to M \to 0$. Set $$\mathcal{F}_2 = \Coker\left( \bigoplus\nolimits_{j \in J} \mathcal{O}_X \to \bigoplus\nolimits_{i \in I} \mathcal{O}_X \right).$$ Here the map on the component $\mathcal{O}_X$ corresponding to $j \in J$ given by the section $\sum_i \alpha(r_{ij})$ where the $r_{ij}$ are the matrix coefficients of the map in the presentation of $M$. \item Set $\mathcal{F}_3$ equal to the sheaf associated to the presheaf $U \mapsto \mathcal{O}_X(U) \otimes_R M$, where the map $R \to \mathcal{O}_X(U)$ is the composition of $\alpha$ and the restriction map $\mathcal{O}_X(X) \to \mathcal{O}_X(U)$. \end{enumerate} This construction has the following properties: \begin{enumerate} \item The resulting sheaf of $\mathcal{O}_X$-modules $\mathcal{F}_M = \mathcal{F}_1 = \mathcal{F}_2 = \mathcal{F}_3$ is quasi-coherent. \item The construction gives a functor from the category of $R$-modules to the category of quasi-coherent sheaves on $X$ which commutes with arbitrary colimits. \item For any $x \in X$ we have $\mathcal{F}_{M, x} = \mathcal{O}_{X, x} \otimes_R M$ functorial in $M$. \item Given any $\mathcal{O}_X$-module $\mathcal{G}$ we have $$\Mor_{\mathcal{O}_X}(\mathcal{F}_M, \mathcal{G}) = \Hom_R(M, \Gamma(X, \mathcal{G}))$$ where the $R$-module structure on $\Gamma(X, \mathcal{G})$ comes from the $\Gamma(X, \mathcal{O}_X)$-module structure via $\alpha$. \end{enumerate} \end{lemma} \begin{proof} The isomorphism between $\mathcal{F}_1$ and $\mathcal{F}_3$ comes from the fact that $\pi^*$ is defined as the sheafification of the presheaf in (3), see Sheaves, Section \ref{sheaves-section-ringed-spaces-functoriality-modules}. The isomorphism between the constructions in (2) and (1) comes from the fact that the functor $\pi^*$ is right exact, so $\pi^*(\bigoplus_{j \in J} R) \to \pi^*(\bigoplus_{i \in I} R) \to \pi^*M \to 0$ is exact, $\pi^*$ commutes with arbitrary direct sums, see Lemma \ref{lemma-exactness-pushforward-pullback}, and finally the fact that $\pi^*(R) = \mathcal{O}_X$. \medskip\noindent Assertion (1) is clear from construction (2). Assertion (2) is clear since $\pi^*$ has these properties. Assertion (3) follows from the description of stalks of pullback sheaves, see Sheaves, Lemma \ref{sheaves-lemma-stalk-pullback-modules}. Assertion (4) follows from adjointness of $\pi_*$ and $\pi^*$. \end{proof} \begin{definition} \label{definition-sheaf-associated} In the situation of Lemma \ref{lemma-construct-quasi-coherent-sheaves} we say $\mathcal{F}_M$ is the {\it sheaf associated to the module $M$ and the ring map $\alpha$}. If $R = \Gamma(X, \mathcal{O}_X)$ and $\alpha = \text{id}_R$ we simply say $\mathcal{F}_M$ is the {\it sheaf associated to the module $M$}. \end{definition} \begin{lemma} \label{lemma-restrict-quasi-coherent} Let $(X, \mathcal{O}_X)$ be a ringed space. Set $R = \Gamma(X, \mathcal{O}_X)$. Let $M$ be an $R$-module. Let $\mathcal{F}_M$ be the quasi-coherent sheaf of $\mathcal{O}_X$-modules associated to $M$. If $g : (Y, \mathcal{O}_Y) \to (X, \mathcal{O}_X)$ is a morphism of ringed spaces, then $g^*\mathcal{F}_M$ is the sheaf associated to the $\Gamma(Y, \mathcal{O}_Y)$-module $\Gamma(Y, \mathcal{O}_Y) \otimes_R M$. \end{lemma} \begin{proof} The assertion follows from the first description of $\mathcal{F}_M$ in Lemma \ref{lemma-construct-quasi-coherent-sheaves} as $\pi^*M$, and the following commutative diagram of ringed spaces $$\xymatrix{ (Y, \mathcal{O}_Y) \ar[r]_-\pi \ar[d]_g & (\{*\}, \Gamma(Y, \mathcal{O}_Y)) \ar[d]^{\text{induced by }g^\sharp} \\ (X, \mathcal{O}_X) \ar[r]^-\pi & (\{*\}, \Gamma(X, \mathcal{O}_X)) }$$ (Also use Sheaves, Lemma \ref{sheaves-lemma-push-pull-composition-modules}.) \end{proof} \begin{lemma} \label{lemma-quasi-coherent-module} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $x \in X$ be a point. Assume that $x$ has a fundamental system of quasi-compact neighbourhoods. Consider any quasi-coherent $\mathcal{O}_X$-module $\mathcal{F}$. Then there exists an open neighbourhood $U$ of $x$ such that $\mathcal{F}|_U$ is isomorphic to the sheaf of modules $\mathcal{F}_M$ on $(U, \mathcal{O}_U)$ associated to some $\Gamma(U, \mathcal{O}_U)$-module $M$. \end{lemma} \begin{proof} First we may replace $X$ by an open neighbourhood of $x$ and assume that $\mathcal{F}$ is isomorphic to the cokernel of a map $$\Psi : \bigoplus\nolimits_{j \in J} \mathcal{O}_X \longrightarrow \bigoplus\nolimits_{i \in I} \mathcal{O}_X.$$ The problem is that this map may not be given by a matrix'', because the module of global sections of a direct sum is in general different from the direct sum of the modules of global sections. \medskip\noindent Let $x \in E \subset X$ be a quasi-compact neighbourhood of $x$ (note: $E$ may not be open). Let $x \in U \subset E$ be an open neighbourhood of $x$ contained in $E$. Next, we proceed as in the proof of Lemma \ref{lemma-section-direct-sum-quasi-compact}. For each $j \in J$ denote $s_j \in \Gamma(X, \bigoplus\nolimits_{i \in I} \mathcal{O}_X)$ the image of the section $1$ in the summand $\mathcal{O}_X$ corresponding to $j$. There exists a finite collection of opens $U_{jk}$, $k \in K_j$ such that $E \subset \bigcup_{k \in K_j} U_{jk}$ and such that each restriction $s_j|_{U_{jk}}$ is a finite sum $\sum_{i \in I_{jk}} f_{jki}$ with $I_{jk} \subset I$, and $f_{jki}$ in the summand $\mathcal{O}_X$ corresponding to $i \in I$. Set $I_j = \bigcup_{k \in K_j} I_{jk}$. This is a finite set. Since $U \subset E \subset \bigcup_{k \in K_j} U_{jk}$ the section $s_j|_U$ is a section of the finite direct sum $\bigoplus_{i \in I_j} \mathcal{O}_X$. By Lemma \ref{lemma-limits-colimits} we see that actually $s_j|_U$ is a sum $\sum_{i \in I_j} f_{ij}$ and $f_{ij} \in \mathcal{O}_X(U) = \Gamma(U, \mathcal{O}_U)$. \medskip\noindent At this point we can define a module $M$ as the cokernel of the map $$\bigoplus\nolimits_{j \in J} \Gamma(U, \mathcal{O}_U) \longrightarrow \bigoplus\nolimits_{i \in I} \Gamma(U, \mathcal{O}_U)$$ with matrix given by the $(f_{ij})$. By construction (2) of Lemma \ref{lemma-construct-quasi-coherent-sheaves} we see that $\mathcal{F}_M$ has the same presentation as $\mathcal{F}|_U$ and therefore $\mathcal{F}_M \cong \mathcal{F}|_U$. \end{proof} \begin{example} \label{example-quasi-coherent} Let $X$ be countably many copies $L_1, L_2, L_3, \ldots$ of the real line all glued together at $0$; a fundamental system of neighbourhoods of $0$ being the collection $\{U_n\}_{n \in \mathbf{N}}$, with $U_n \cap L_i = (-1/n, 1/n)$. Let $\mathcal{O}_X$ be the sheaf of continuous real valued functions. Let $f : \mathbf{R} \to \mathbf{R}$ be a continuous function which is identically zero on $(-1, 1)$ and identically $1$ on $(-\infty, -2) \cup (2, \infty)$. Denote $f_n$ the continuous function on $X$ which is equal to $x \mapsto f(nx)$ on each $L_j = \mathbf{R}$. Let $1_{L_j}$ be the characteristic function of $L_j$. We consider the map $$\bigoplus\nolimits_{j \in \mathbf{N}} \mathcal{O}_X \longrightarrow \bigoplus\nolimits_{j, i \in \mathbf{N}} \mathcal{O}_X, \quad e_j \longmapsto \sum\nolimits_{i \in \mathbf{N}} f_j 1_{L_i} e_{ij}$$ with obvious notation. This makes sense because this sum is locally finite as $f_j$ is zero in a neighbourhood of $0$. Over $U_n$ the image of $e_j$, for $j > 2n$ is not a finite linear combination $\sum g_{ij} e_{ij}$ with $g_{ij}$ continuous. Thus there is no neighbourhood of $0 \in X$ such that the displayed map is given by a matrix'' as in the proof of Lemma \ref{lemma-quasi-coherent-module} above. \medskip\noindent Note that $\bigoplus\nolimits_{j \in \mathbf{N}} \mathcal{O}_X$ is the sheaf associated to the free module with basis $e_j$ and similarly for the other direct sum. Thus we see that a morphism of sheaves associated to modules in general even locally on $X$ does not come from a morphism of modules. Similarly there should be an example of a ringed space $X$ and a quasi-coherent $\mathcal{O}_X$-module $\mathcal{F}$ such that $\mathcal{F}$ is not locally of the form $\mathcal{F}_M$. (Please email if you find one.) Moreover, there should be examples of locally compact spaces $X$ and maps $\mathcal{F}_M \to \mathcal{F}_N$ which also do not locally come from maps of modules (the proof of Lemma \ref{lemma-quasi-coherent-module} shows this cannot happen if $N$ is free). \end{example} \section{Modules of finite presentation} \label{section-finite-presentation} \begin{definition} \label{definition-finite-presentation} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. We say that $\mathcal{F}$ is of {\it finite presentation} if for every point $x \in X$ there exists an open neighbourhood $x\in U \subset X$, and $n, m \in \mathbf{N}$ such that $\mathcal{F}|_U$ is isomorphic to the cokernel of a map $$\bigoplus\nolimits_{j = 1, \ldots, m} \mathcal{O}_U \longrightarrow \bigoplus\nolimits_{i = 1, \ldots, n} \mathcal{O}_U$$ \end{definition} \noindent This means that $X$ is covered by open sets $U$ such that $\mathcal{F}|_U$ has a {\it presentation} of the form $$\bigoplus\nolimits_{j = 1, \ldots, m} \mathcal{O}_U \longrightarrow \bigoplus\nolimits_{i = 1, \ldots, n} \mathcal{O}_U \to \mathcal{F}|_U \to 0.$$ Here presentation signifies that the displayed sequence is exact. In other words \begin{enumerate} \item for every point $x$ of $X$ there exists an open neighbourhood such that $\mathcal{F}|_U$ is generated by finitely many global sections, and \item for a suitable choice of these sections the kernel of the associated surjection is also generated by finitely many global sections. \end{enumerate} \begin{lemma} \label{lemma-finite-presentation-quasi-coherent} Let $(X, \mathcal{O}_X)$ be a ringed space. Any $\mathcal{O}_X$-module of finite presentation is quasi-coherent. \end{lemma} \begin{proof} Immediate from definitions. \end{proof} \begin{lemma} \label{lemma-kernel-surjection-finite-free-onto-finite-presentation} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a $\mathcal{O}_X$-module of finite presentation. \begin{enumerate} \item If $\psi : \mathcal{O}_X^{\oplus r} \to \mathcal{F}$ is a surjection, then $\Ker(\psi)$ is of finite type. \item If $\theta : \mathcal{G} \to \mathcal{F}$ is surjective with $\mathcal{G}$ of finite type, then $\Ker(\theta)$ is of finite type. \end{enumerate} \end{lemma} \begin{proof} Proof of (1). Let $x \in X$. Choose an open neighbourhood $U \subset X$ of $x$ such that there exists a presentation $$\mathcal{O}_U^{\oplus m} \xrightarrow{\chi} \mathcal{O}_U^{\oplus n} \xrightarrow{\varphi} \mathcal{F}|_U \to 0.$$ Let $e_k$ be the section generating the $k$th factor of $\mathcal{O}_X^{\oplus r}$. For every $k = 1, \ldots, r$ we can, after shrinking $U$ to a small neighbourhood of $x$, lift $\psi(e_k)$ to a section $\tilde e_k$ of $\mathcal{O}_U^{\oplus n}$ over $U$. This gives a morphism of sheaves $\alpha : \mathcal{O}_U^{\oplus r} \to \mathcal{O}_U^{\oplus n}$ such that $\varphi \circ \alpha = \psi$. Similarly, after shrinking $U$, we can find a morphism $\beta : \mathcal{O}_U^{\oplus n} \to \mathcal{O}_U^{\oplus r}$ such that $\psi \circ \beta = \varphi$. Then the map $$\mathcal{O}_U^{\oplus m} \oplus \mathcal{O}_U^{\oplus r} \xrightarrow{\beta \circ \chi, 1 - \beta \circ \alpha} \mathcal{O}_U^{\oplus r}$$ is a surjection onto the kernel of $\psi$. \medskip\noindent To prove (2) we may locally choose a surjection $\eta : \mathcal{O}_X^{\oplus r} \to \mathcal{G}$. By part (1) we see $\Ker(\theta \circ \eta)$ is of finite type. Since $\Ker(\theta) = \eta(\Ker(\theta \circ \eta))$ we win. \end{proof} \begin{lemma} \label{lemma-pullback-finite-presentation} Let $f : (X, \mathcal{O}_X) \to (Y, \mathcal{O}_Y)$ be a morphism of ringed spaces. The pullback $f^*\mathcal{G}$ of a module of finite presentation is of finite presentation. \end{lemma} \begin{proof} Exactly the same as the proof of Lemma \ref{lemma-pullback-quasi-coherent} but with finite index sets. \end{proof} \begin{lemma} \label{lemma-quasi-coherent-limit-finite-presentation} Let $(X, \mathcal{O}_X)$ be a ringed space. Set $R = \Gamma(X, \mathcal{O}_X)$. Let $M$ be an $R$-module. The $\mathcal{O}_X$-module $\mathcal{F}_M$ associated to $M$ is a directed colimit of finitely presented $\mathcal{O}_X$-modules. \end{lemma} \begin{proof} This follows immediately from Lemma \ref{lemma-construct-quasi-coherent-sheaves} and the fact that any module is a directed colimit of finitely presented modules, see Algebra, Lemma \ref{algebra-lemma-module-colimit-fp}. \end{proof} \begin{lemma} \label{lemma-finite-presentation-quasi-compact-colimit} Let $X$ be a ringed space. Let $I$ be a preordered set and let $(\mathcal{F}_i, \varphi_{ii'})$ be a system over $I$ consisting of sheaves of $\mathcal{O}_X$-modules (see Categories, Section \ref{categories-section-posets-limits}). Assume \begin{enumerate} \item $I$ is directed, \item $\mathcal{G}$ is an $\mathcal{O}_X$-module of finite presentation, and \item $X$ has a cofinal system of open coverings $\mathcal{U} : X = \bigcup_{j\in J} U_j$ with $J$ finite and $U_j \cap U_{j'}$ quasi-compact for all $j, j' \in J$. \end{enumerate} Then we have $$\colim_i \Hom_X(\mathcal{G}, \mathcal{F}_i) = \Hom_X(\mathcal{G}, \colim_i \mathcal{F}_i).$$ \end{lemma} \begin{proof} Let $\alpha$ be an element of the right hand side. For every point $x \in X$ we may choose an open neighbourhood $U \subset X$ and finitely many sections $s_j \in \mathcal{G}(U)$ which generate $\mathcal{G}$ over $U$ and finitely many relations $\sum f_{kj} s_j = 0$, $k = 1, \ldots, n$ with $f_{kj} \in \mathcal{O}_X(U)$ which generate the kernel of $\bigoplus_{j = 1, \ldots, m} \mathcal{O}_U \to \mathcal{G}$. After possibly shrinking $U$ to a smaller open neighbourhood of $x$ we may assume there exists an index $i \in I$ such that the sections $\alpha(s_j)$ all come from sections $s_j' \in \mathcal{F}_i(U)$. After possibly shrinking $U$ to a smaller open neighbourhood of $x$ and increasing $i$ we may assume the relations $\sum f_{kj} s'_j = 0$ hold in $\mathcal{F}_i(U)$. Hence we see that $\alpha|_U$ lifts to a morphism $\mathcal{G}|_U \to \mathcal{F}_i|_U$ for some index $i \in I$. \medskip\noindent By condition (3) and the preceding arguments, we may choose a finite open covering $X = \bigcup_{j = 1, \ldots, m} U_j$ such that (a) $\mathcal{G}|_{U_j}$ is generated by finitely many sections $s_{jk} \in \mathcal{G}(U_j)$, (b) the restriction $\alpha|_{U_j}$ comes from a morphism $\alpha_j : \mathcal{G} \to \mathcal{F}_{i_j}$ for some $i_j \in I$, and (c) the intersections $U_j \cap U_{j'}$ are all quasi-compact. For every pair $(j, j') \in \{1, \ldots, m\}^2$ and any $k$ we can find we can find an index $i \geq \max(i_j, i_{j'})$ such that $$\varphi_{i_ji}(\alpha_j(s_{jk}|_{U_j \cap U_{j'}})) = \varphi_{i_{j'}i}(\alpha_{j'}(s_{jk}|_{U_j \cap U_{j'}}))$$ see Sheaves, Lemma \ref{sheaves-lemma-directed-colimits-sections} (2). Since there are finitely many of these pairs $(j, j')$ and finitely many $s_{jk}$ we see that we can find a single $i$ which works for all of them. For this index $i$ all of the maps $\varphi_{i_ji} \circ \alpha_j$ agree on the overlaps $U_j \cap U_{j'}$ as the sections $s_{jk}$ generate $\mathcal{G}$ over this overlap. Hence we get a morphism $\mathcal{G} \to \mathcal{F}_i$ as desired. \end{proof} \begin{remark} \label{remark-condition-necessary} In the lemma above some condition beyond the condition that $X$ is quasi-compact is necessary. See Sheaves, Example \ref{sheaves-example-conditions-needed-colimit}. \end{remark} \begin{lemma} \label{lemma-finite-presentation-stalk-free} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a finitely presented $\mathcal{O}_X$-module. Let $x \in X$ such that $\mathcal{F}_x \cong \mathcal{O}_{X, x}^{\oplus r}$. Then there exists an open neighbourhood $U$ of $x$ such that $\mathcal{F}|_U \cong \mathcal{O}_U^{\oplus r}$. \end{lemma} \begin{proof} Choose $s_1, \ldots, s_r \in \mathcal{F}_x$ mapping to a basis of $\mathcal{O}_{X, x}^{\oplus r}$ by the isomorphism. Choose an open neighbourhood $U$ of $x$ such that $s_i$ lifts to $s_i \in \mathcal{F}(U)$. After shrinking $U$ we see that the induced map $\psi : \mathcal{O}_U^{\oplus r} \to \mathcal{F}|_U$ is surjective (Lemma \ref{lemma-finite-type-surjective-on-stalk}). By Lemma \ref{lemma-kernel-surjection-finite-free-onto-finite-presentation} we see that $\Ker(\psi)$ is of finite type. Then $\Ker(\psi)_x = 0$ implies that $\Ker(\psi)$ becomes zero after shrinking $U$ once more (Lemma \ref{lemma-finite-type-stalk-zero}). \end{proof} \section{Coherent modules} \label{section-coherent} \noindent The category of coherent sheaves on a ringed space $X$ is a more reasonable object than the category of quasi-coherent sheaves, in the sense that it is at least an abelian subcategory of $\textit{Mod}(\mathcal{O}_X)$ no matter what $X$ is. On the other hand, the pullback of a coherent module is almost never'' coherent in the general setting of ringed spaces. \begin{definition} \label{definition-coherent} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. We say that $\mathcal{F}$ is a {\it coherent $\mathcal{O}_X$-module} if the following two conditions hold: \begin{enumerate} \item $\mathcal{F}$ is of finite type, and \item for every open $U \subset X$ and every finite collection $s_i \in \mathcal{F}(U)$, $i = 1, \ldots, n$ the kernel of the associated map $\bigoplus_{i = 1, \ldots, n} \mathcal{O}_U \to \mathcal{F}|_U$ is of finite type. \end{enumerate} The category of coherent $\mathcal{O}_X$-modules is denoted $\textit{Coh}(\mathcal{O}_X)$. \end{definition} \begin{lemma} \label{lemma-coherent-finite-presentation} Let $(X, \mathcal{O}_X)$ be a ringed space. Any coherent $\mathcal{O}_X$-module is of finite presentation and hence quasi-coherent. \end{lemma} \begin{proof} Let $\mathcal{F}$ be a coherent sheaf on $X$. Pick a point $x \in X$. By (1) of the definition of coherent, we may find an open neighbourhood $U$ and sections $s_i$, $i = 1, \ldots, n$ of $\mathcal{F}$ over $U$ such that $\Psi : \bigoplus_{i = 1, \ldots, n} \mathcal{O}_U \to \mathcal{F}$ is surjective. By (2) of the definition of coherent, we may find an open neighbourhood $V$, $x \in V \subset U$ and sections $t_1, \ldots, t_m$ of $\bigoplus_{i = 1, \ldots, n} \mathcal{O}_V$ which generate the kernel of $\Psi|_V$. Then over $V$ we get the presentation $$\bigoplus\nolimits_{j = 1, \ldots, m} \mathcal{O}_V \longrightarrow \bigoplus\nolimits_{i = 1, \ldots, n} \mathcal{O}_V \to \mathcal{F}|_V \to 0$$ as desired. \end{proof} \begin{example} \label{example-coherent-not-Noetherian} Suppose that $X$ is a point. In this case the definition above gives a notion for modules over rings. What does the definition of coherent mean? It is closely related to the notion of Noetherian, but it is not the same: Namely, the ring $R = \mathbf{C}[x_1, x_2, x_3, \ldots]$ is coherent as a module over itself but not Noetherian as a module over itself. See Algebra, Section \ref{algebra-section-coherent} for more discussion. \end{example} \begin{lemma} \label{lemma-coherent-abelian} Let $(X, \mathcal{O}_X)$ be a ringed space. \begin{enumerate} \item Any finite type subsheaf of a coherent sheaf is coherent. \item Let $\varphi : \mathcal{F} \to \mathcal{G}$ be a morphism from a finite type sheaf $\mathcal{F}$ to a coherent sheaf $\mathcal{G}$. Then $\Ker(\varphi)$ is finite type. \item Let $\varphi : \mathcal{F} \to \mathcal{G}$ be a morphism of coherent $\mathcal{O}_X$-modules. Then $\Ker(\varphi)$ and $\Coker(\varphi)$ are coherent. \item Given a short exact sequence of $\mathcal{O}_X$-modules $0 \to \mathcal{F}_1 \to \mathcal{F}_2 \to \mathcal{F}_3 \to 0$ if two out of three are coherent so is the third. \item The category $\textit{Coh}(\mathcal{O}_X)$ is a weak Serre subcategory of $\textit{Mod}(\mathcal{O}_X)$. In particular, the category of coherent modules is abelian and the inclusion functor $\textit{Coh}(\mathcal{O}_X) \to \textit{Mod}(\mathcal{O}_X)$ is exact. \end{enumerate} \end{lemma} \begin{proof} Condition (2) of Definition \ref{definition-coherent} holds for any subsheaf of a coherent sheaf. Thus we get (1). \medskip\noindent Assume the hypotheses of (2). Let us show that $\Ker(\varphi)$ is of finite type. Pick $x \in X$. Choose an open neighbourhood $U$ of $x$ in $X$ such that $\mathcal{F}|_U$ is generated by $s_1, \ldots, s_n$. By Definition \ref{definition-coherent} the kernel $\mathcal{K}$ of the induced map $\bigoplus_{i = 1}^n \mathcal{O}_U \to \mathcal{G}$, $e_i \mapsto \varphi(s_i)$ is of finite type. Hence $\Ker(\varphi)$ which is the image of the composition $\mathcal{K} \to \bigoplus_{i = 1}^n \mathcal{O}_U \to \mathcal{F}$ is of finite type. \medskip\noindent Assume the hypotheses of (3). By (2) the kernel of $\varphi$ is of finite type and hence by (1) it is coherent. \medskip\noindent With the same hypotheses let us show that $\Coker(\varphi)$ is coherent. Since $\mathcal{G}$ is of finite type so is $\Coker(\varphi)$. Let $U \subset X$ be open and let $\overline{s}_i \in \Coker(\varphi)(U)$, $i = 1, \ldots, n$ be sections. We have to show that the kernel of the associated morphism $\overline{\Psi} : \bigoplus_{i = 1}^n \mathcal{O}_U \to \Coker(\varphi)$ has finite type. There exists an open covering of $U$ such that on each open all the sections $\overline{s}_i$ lift to sections $s_i$ of $\mathcal{G}$. Hence we may assume this is the case over $U$. We may in addition assume there are sections $t_j$, $j = 1, \ldots, m$ of $\Im(\varphi)$ over $U$ which generate $\Im(\varphi)$ over $U$. Let $\Phi : \bigoplus_{j = 1}^m \mathcal{O}_U \to \Im(\varphi)$ be defined using $t_j$ and $\Psi : \bigoplus_{j = 1}^m \mathcal{O}_U \oplus \bigoplus_{i = 1}^n \mathcal{O}_U \to \mathcal{G}$ using $t_j$ and $s_i$. Consider the following commutative diagram $$\xymatrix{ 0 \ar[r] & \bigoplus_{j = 1}^m \mathcal{O}_U \ar[d]_\Phi \ar[r] & \bigoplus_{j = 1}^m \mathcal{O}_U \oplus \bigoplus_{i = 1}^n \mathcal{O}_U \ar[d]_\Psi \ar[r] & \bigoplus_{i = 1}^n \mathcal{O}_U \ar[d]_{\overline{\Psi}} \ar[r] & 0 \\ 0 \ar[r] & \Im(\varphi) \ar[r] & \mathcal{G} \ar[r] & \Coker(\varphi) \ar[r] & 0 }$$ By the snake lemma we get an exact sequence $\Ker(\Psi) \to \Ker(\overline{\Psi}) \to 0$. Since $\Ker(\Psi)$ is a finite type module, we see that $\Ker(\overline{\Psi})$ has finite type. \medskip\noindent Proof of part (4). Let $0 \to \mathcal{F}_1 \to \mathcal{F}_2 \to \mathcal{F}_3 \to 0$ be a short exact sequence of $\mathcal{O}_X$-modules. By part (3) it suffices to prove that if $\mathcal{F}_1$ and $\mathcal{F}_3$ are coherent so is $\mathcal{F}_2$. By Lemma \ref{lemma-extension-finite-type} we see that $\mathcal{F}_2$ has finite type. Let $s_1, \ldots, s_n$ be finitely many local sections of $\mathcal{F}_2$ defined over a common open $U$ of $X$. We have to show that the module of relations $\mathcal{K}$ between them is of finite type. Consider the following commutative diagram $$\xymatrix{ 0 \ar[r] & 0 \ar[r] \ar[d] & \bigoplus_{i = 1}^{n} \mathcal{O}_U \ar[r] \ar[d] & \bigoplus_{i = 1}^{n} \mathcal{O}_U \ar[r] \ar[d] & 0 \\ 0 \ar[r] & \mathcal{F}_1 \ar[r] & \mathcal{F}_2 \ar[r] & \mathcal{F}_3 \ar[r] & 0 }$$ with obvious notation. By the snake lemma we get a short exact sequence $0 \to \mathcal{K} \to \mathcal{K}_3 \to \mathcal{F}_1$ where $\mathcal{K}_3$ is the module of relations among the images of the sections $s_i$ in $\mathcal{F}_3$. Since $\mathcal{F}_3$ is coherent we see that $\mathcal{K}_3$ is finite type. Since $\mathcal{F}_1$ is coherent we see that the image $\mathcal{I}$ of $\mathcal{K}_3 \to \mathcal{F}_1$ is coherent. Hence $\mathcal{K}$ is the kernel of the map $\mathcal{K}_3 \to \mathcal{I}$ between a finite type sheaf and a coherent sheaves and hence finite type by (2). \medskip\noindent Proof of (5). This follows because (3) and (4) show that Homology, Lemma \ref{homology-lemma-characterize-weak-serre-subcategory} applies. \end{proof} \begin{lemma} \label{lemma-coherent-structure-sheaf} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be an $\mathcal{O}_X$-module. Assume $\mathcal{O}_X$ is a coherent $\mathcal{O}_X$-module. Then $\mathcal{F}$ is coherent if and only if it is of finite presentation. \end{lemma} \begin{proof} Omitted. \end{proof} \begin{lemma} \label{lemma-finite-type-to-coherent-injective-on-stalk} Let $X$ be a ringed space. Let $\varphi : \mathcal{G} \to \mathcal{F}$ be a homomorphism of $\mathcal{O}_X$-modules. Let $x \in X$. Assume $\mathcal{G}$ of finite type, $\mathcal{F}$ coherent and the map on stalks $\varphi_x : \mathcal{G}_x \to \mathcal{F}_x$ injective. Then there exists an open neighbourhood $x \in U \subset X$ such that $\varphi|_U$ is injective. \end{lemma} \begin{proof} Denote $\mathcal{K} \subset \mathcal{G}$ the kernel of $\varphi$. By Lemma \ref{lemma-coherent-abelian} we see that $\mathcal{K}$ is a finite type $\mathcal{O}_X$-module. Our assumption is that $\mathcal{K}_x = 0$. By Lemma \ref{lemma-finite-type-stalk-zero} there exists an open neighbourhood $U$ of $x$ such that $\mathcal{K}|_U = 0$. Then $U$ works. \end{proof} \section{Closed immersions of ringed spaces} \label{section-closed-immersion} \noindent When do we declare a morphism of ringed spaces $i : (Z, \mathcal{O}_Z) \to (X, \mathcal{O}_X)$ to be a closed immersion? \medskip\noindent Motivated by the example of a closed immersion of normal topological spaces (ringed with the sheaf of continuous functors), or differential manifolds (ringed with the sheaf of differentiable functions), it seems natural to assume at least: \begin{enumerate} \item The map $i$ is a closed immersion of topological spaces. \item The associated map $\mathcal{O}_X \to i_*\mathcal{O}_Z$ is surjective. Denote the kernel by $\mathcal{I}$. \end{enumerate} Already these conditions imply a number of pleasing results: For example we prove that the category of $\mathcal{O}_Z$-modules is equivalent to the category of $\mathcal{O}_X$-modules annihilated by $\mathcal{I}$ generalizing the result on abelian sheaves of Section \ref{section-closed-immersions} \medskip\noindent However, in the Stacks project we choose the definition that guarantees that if $i$ is a closed immersion and $(X, \mathcal{O}_X)$ is a scheme, then also $(Z, \mathcal{O}_Z)$ is a scheme. Moreover, in this situation we want $i_*$ and $i^*$ to provide an equivalence between the category of quasi-coherent $\mathcal{O}_Z$-modules and the category of quasi-coherent $\mathcal{O}_X$-modules annihilated by $\mathcal{I}$. A minimal condition is that $i_*\mathcal{O}_Z$ is a quasi-coherent sheaf of $\mathcal{O}_X$-modules. A good way to guarantee that $i_*\mathcal{O}_Z$ is a quasi-coherent $\mathcal{O}_X$-module is to assume that $\mathcal{I}$ is locally generated by sections. We can interpret this condition as saying $(Z, \mathcal{O}_Z)$ is locally on $(X, \mathcal{O}_X)$ defined by setting some regular functions $f_i$, i.e., local sections of $\mathcal{O}_X$, equal to zero''. This leads to the following definition. \begin{definition} \label{definition-closed-immersion} A {\it closed immersion of ringed spaces}\footnote{This is nonstandard notation; see discussion above.} is a morphism $i : (Z, \mathcal{O}_Z) \to (X, \mathcal{O}_X)$ with the following properties: \begin{enumerate} \item The map $i$ is a closed immersion of topological spaces. \item The associated map $\mathcal{O}_X \to i_*\mathcal{O}_Z$ is surjective. Denote the kernel by $\mathcal{I}$. \item The $\mathcal{O}_X$-module $\mathcal{I}$ is locally generated by sections. \end{enumerate} \end{definition} \noindent Actually, this definition still does not guarantee that $i_*$ of a quasi-coherent $\mathcal{O}_Z$-module is a quasi-coherent $\mathcal{O}_X$-module. The problem is that it is not clear how to convert a local presentation of a quasi-coherent $\mathcal{O}_Z$-module into a local presentation for the pushforward. However, the following is trivial. \begin{lemma} \label{lemma-i-star-quasi-coherent} Let $i : (Z, \mathcal{O}_Z) \to (X, \mathcal{O}_X)$ be a closed immersion of ringed spaces. Let $\mathcal{F}$ be a quasi-coherent $\mathcal{O}_Z$-module. Then $i_*\mathcal{F}$ is locally on $X$ the cokernel of a map of quasi-coherent $\mathcal{O}_X$-modules. \end{lemma} \begin{proof} This is true because $i_*\mathcal{O}_Z$ is quasi-coherent by definition. And locally on $Z$ the sheaf $\mathcal{F}$ is a cokernel of a map between direct sums of copies of $\mathcal{O}_Z$. Moreover, any direct sum of copies of the {\it the same} quasi-coherent sheaf is quasi-coherent. And finally, $i_*$ commutes with arbitrary colimits, see Lemma \ref{lemma-i-star-right-adjoint}. Some details omitted. \end{proof} \begin{lemma} \label{lemma-i-star-reflects-finite-type} Let $i : (Z, \mathcal{O}_Z) \to (X, \mathcal{O}_X)$ be a morphism of ringed spaces. Assume $i$ is a homeomorphism onto a closed subset of $X$ and that $\mathcal{O}_X \to i_*\mathcal{O}_Z$ is surjective. Let $\mathcal{F}$ be an $\mathcal{O}_Z$-module. Then $i_*\mathcal{F}$ is of finite type if and only if $\mathcal{F}$ is of finite type. \end{lemma} \begin{proof} Suppose that $\mathcal{F}$ is of finite type. Pick $x \in X$. If $x \not \in Z$, then $i_*\mathcal{F}$ is zero in a neighbourhood of $x$ and hence finitely generated in a neighbourhood of $x$. If $x = i(z)$, then choose an open neighbourhood $z \in V \subset Z$ and sections $s_1, \ldots, s_n \in \mathcal{F}(V)$ which generate $\mathcal{F}$ over $V$. Write $V = Z \cap U$ for some open $U \subset X$. Note that $U$ is a neighbourhood of $x$. Clearly the sections $s_i$ give sections $s_i$ of $i_*\mathcal{F}$ over $U$. The resulting map $$\bigoplus\nolimits_{i = 1, \ldots, n} \mathcal{O}_U \longrightarrow i_*\mathcal{F}|_U$$ is surjective by inspection of what it does on stalks (here we use that $\mathcal{O}_X \to i_*\mathcal{O}_Z$ is surjective). Hence $i_*\mathcal{F}$ is of finite type. \medskip\noindent Conversely, suppose that $i_*\mathcal{F}$ is of finite type. Choose $z \in Z$. Set $x = i(z)$. By assumption there exists an open neighbourhood $U \subset X$ of $x$, and sections $s_1, \ldots, s_n \in (i_*\mathcal{F})(U)$ which generate $i_*\mathcal{F}$ over $U$. Set $V = Z \cap U$. By definition of $i_*$ the sections $s_i$ correspond to sections $s_i$ of $\mathcal{F}$ over $V$. The resulting map $$\bigoplus\nolimits_{i = 1, \ldots, n} \mathcal{O}_V \longrightarrow \mathcal{F}|_V$$ is surjective by inspection of what it does on stalks. Hence $\mathcal{F}$ is of finite type. \end{proof} \begin{lemma} \label{lemma-i-star-equivalence} Let $i : (Z, \mathcal{O}_Z) \to (X, \mathcal{O}_X)$ be a morphism of ringed spaces. Assume $i$ is a homeomorphism onto a closed subset of $X$ and $i^\sharp : \mathcal{O}_X \to i_*\mathcal{O}_Z$ is surjective. Denote $\mathcal{I} \subset \mathcal{O}_X$ the kernel of $i^\sharp$. The functor $$i_* : \textit{Mod}(\mathcal{O}_Z) \longrightarrow \textit{Mod}(\mathcal{O}_X)$$ is exact, fully faithful, with essential image those $\mathcal{O}_X$-modules $\mathcal{G}$ such that $\mathcal{I}\mathcal{G} = 0$. \end{lemma} \begin{proof} We claim that for a $\mathcal{O}_Z$-module $\mathcal{F}$ the canonical map $$i^*i_*\mathcal{F} \longrightarrow \mathcal{F}$$ is an isomorphism. We check this on stalks. Say $z \in Z$ and $x = i(z)$. We have $$(i^*i_*\mathcal{F})_z = (i_*\mathcal{F})_x \otimes_{\mathcal{O}_{X, x}} \mathcal{O}_{Z, z} = \mathcal{F}_z \otimes_{\mathcal{O}_{X, x}} \mathcal{O}_{Z, z} = \mathcal{F}_z$$ by Sheaves, Lemma \ref{sheaves-lemma-stalk-pullback-modules}, the fact that $\mathcal{O}_{Z, z}$ is a quotient of $\mathcal{O}_{X, x}$, and Sheaves, Lemma \ref{sheaves-lemma-stalks-closed-pushforward}. It follows that $i_*$ is fully faithful. \medskip\noindent Let $\mathcal{G}$ be a $\mathcal{O}_X$-module with $\mathcal{I}\mathcal{G} = 0$. We will prove the canonical map $$\mathcal{G} \longrightarrow i_*i^*\mathcal{G}$$ is an isomorphism. This proves that $\mathcal{G} = i_*\mathcal{F}$ with $\mathcal{F} = i^*\mathcal{G}$ which finishes the proof. We check the displayed map induces an isomorphism on stalks. If $x \in X$, $x \not \in i(Z)$, then $\mathcal{G}_x = 0$ because $\mathcal{I}_x = \mathcal{O}_{X, x}$ in this case. As above $(i_*i^*\mathcal{G})_x = 0$ by Sheaves, Lemma \ref{sheaves-lemma-stalks-closed-pushforward}. On the other hand, if $x \in Z$, then we obtain the map $$\mathcal{G}_x \longrightarrow \mathcal{G}_x \otimes_{\mathcal{O}_{X, x}} \mathcal{O}_{Z, x}$$ by Sheaves, Lemmas \ref{sheaves-lemma-stalk-pullback-modules} and \ref{sheaves-lemma-stalks-closed-pushforward}. This map is an isomorphism because $\mathcal{O}_{Z, x} = \mathcal{O}_{X, x}/\mathcal{I}_x$ and because $\mathcal{G}_x$ is annihilated by $\mathcal{I}_x$ by assumption. \end{proof} \section{Locally free sheaves} \label{section-locally-free} \noindent Let $(X, \mathcal{O}_X)$ be a ringed space. Our conventions allow (some of) the stalks $\mathcal{O}_{X, x}$ to be the zero ring. This means we have to be a little careful when defining the rank of a locally free sheaf. \begin{definition} \label{definition-locally-free} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. \begin{enumerate} \item We say $\mathcal{F}$ is {\it locally free} if for every point $x \in X$ there exists a set $I$ and an open neighbourhood $x \in U \subset X$ such that $\mathcal{F}|_U$ is isomorphic to $\bigoplus_{i \in I} \mathcal{O}_X|_U$ as an $\mathcal{O}_X|_U$-module. \item We say $\mathcal{F}$ is {\it finite locally free} if we may choose the index sets $I$ to be finite. \item We say $\mathcal{F}$ is {\it finite locally free of rank $r$} if we may choose the index sets $I$ to have cardinality $r$. \end{enumerate} \end{definition} \noindent A finite direct sum of (finite) locally free sheaves is (finite) locally free. However, it may not be the case that an infinite direct sum of locally free sheaves is locally free. \begin{lemma} \label{lemma-locally-free-quasi-coherent} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. If $\mathcal{F}$ is locally free then it is quasi-coherent. \end{lemma} \begin{proof} Omitted. \end{proof} \begin{lemma} \label{lemma-pullback-locally-free} Let $f : (X, \mathcal{O}_X) \to (Y, \mathcal{O}_Y)$ be a morphism of ringed spaces. If $\mathcal{G}$ is a locally free $\mathcal{O}_Y$-module, then $f^*\mathcal{G}$ is a locally free $\mathcal{O}_X$-module. \end{lemma} \begin{proof} Omitted. \end{proof} \begin{lemma} \label{lemma-rank} Let $(X, \mathcal{O}_X)$ be a ringed space. Suppose that the support of $\mathcal{O}_X$ is $X$, i.e., all stalk of $\mathcal{O}_X$ are nonzero rings. Let $\mathcal{F}$ be a locally free sheaf of $\mathcal{O}_X$-modules. There exists a locally constant function $$\text{rank}_\mathcal{F} : X \longrightarrow \{0, 1, 2, \ldots\}\cup\{\infty\}$$ such that for any point $x \in X$ the cardinality of any set $I$ such that $\mathcal{F}$ is isomorphic to $\bigoplus_{i\in I} \mathcal{O}_X$ in a neighbourhood of $x$ is $\text{rank}_\mathcal{F}(x)$. \end{lemma} \begin{proof} Under the assumption of the lemma the cardinality of $I$ can be read off from the rank of the free module $\mathcal{F}_x$ over the nonzero ring $\mathcal{O}_{X, x}$, and it is constant in a neighbourhood of $x$. \end{proof} \begin{lemma} \label{lemma-map-finite-locally-free} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $r \geq 0$. Let $\varphi : \mathcal{F} \to \mathcal{G}$ be a map of finite locally free $\mathcal{O}_X$-modules of rank $r$. Then $\varphi$ is an isomorphism if and only if $\varphi$ is surjective. \end{lemma} \begin{proof} Assume $\varphi$ is surjective. Pick $x \in X$. There exists an open neighbourhood $U$ of $x$ such that both $\mathcal{F}|_U$ and $\mathcal{G}|_U$ are isomorphic to $\mathcal{O}_U^{\oplus r}$. Pick lifts of the free generators of $\mathcal{G}|_U$ to obtain a map $\psi : \mathcal{G}|_U \to \mathcal{F}|_U$ such that $\varphi|_U \circ \psi = \text{id}$. Hence we conclude that the map $\Gamma(U, \mathcal{F}) \to \Gamma(U, \mathcal{G})$ induced by $\varphi$ is surjective. Since both $\Gamma(U, \mathcal{F})$ and $\Gamma(U, \mathcal{G})$ are isomorphic to $\Gamma(U, \mathcal{O}_U)^{\oplus r}$ as an $\Gamma(U, \mathcal{O}_U)$-module we may apply Algebra, Lemma \ref{algebra-lemma-fun} to see that $\Gamma(U, \mathcal{F}) \to \Gamma(U, \mathcal{G})$ is injective. This finishes the proof. \end{proof} \begin{lemma} \label{lemma-direct-summand-of-locally-free-is-locally-free} Let $(X, \mathcal{O}_X)$ be a ringed space. If all stalks $\mathcal{O}_{X, x}$ are local rings, then any direct summand of a finite locally free $\mathcal{O}_X$-module is finite locally free. \end{lemma} \begin{proof} Assume $\mathcal{F}$ is a direct summand of the finite locally free $\mathcal{O}_X$-module $\mathcal{H}$. Let $x \in X$ be a point. Then $\mathcal{H}_x$ is a finite free $\mathcal{O}_{X, x}$-module. Because $\mathcal{O}_{X, x}$ is local, we see that $\mathcal{F}_x \cong \mathcal{O}_{X, x}^{\oplus r}$ for some $r$, see Algebra, Lemma \ref{algebra-lemma-finite-projective}. By Lemma \ref{lemma-finite-presentation-stalk-free} we see that $\mathcal{F}$ is free of rank $r$ in an open neighbourhood of $x$. (Note that $\mathcal{F}$ is of finite presentation as a summand of $\mathcal{H}$.) \end{proof} \section{Tensor product} \label{section-tensor-product} \noindent Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$, $\mathcal{G}$ be $\mathcal{O}_X$-modules. We have briefly discussed the tensor product in the setting of change of rings in Sheaves, Sections \ref{sheaves-section-presheaves-modules} and \ref{sheaves-section-sheafification-presheaves-modules}. In exactly the same way we define first the {\it tensor product presheaf} $$\mathcal{F} \otimes_{p, \mathcal{O}_X} \mathcal{G}$$ as the rule which assigns to $U \subset X$ open the $\mathcal{O}_X(U)$-module $\mathcal{F}(U) \otimes_{\mathcal{O}_X(U)} \mathcal{G}(U)$. Having defined this we define the {\it tensor product sheaf} as the sheafification of the above: $$\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G} = (\mathcal{F} \otimes_{p, \mathcal{O}_X} \mathcal{G})^\#$$ This can be characterized as the sheaf of $\mathcal{O}_X$-modules such that for any third sheaf of $\mathcal{O}_X$-modules $\mathcal{H}$ we have $$\Hom_{\mathcal{O}_X} (\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}, \mathcal{H}) = \text{Bilin}_{\mathcal{O}_X}(\mathcal{F} \times \mathcal{G}, \mathcal{H}).$$ Here the right hand side indicates the set of bilinear maps of sheaves of $\mathcal{O}_X$-modules (definition omitted). \medskip\noindent The tensor product of modules $M, N$ over a ring $R$ satisfies symmetry, namely $M \otimes_R N = N \otimes_R M$, hence the same holds for tensor products of sheaves of modules, i.e., we have $$\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G} = \mathcal{G} \otimes_{\mathcal{O}_X} \mathcal{F}$$ functorial in $\mathcal{F}$, $\mathcal{G}$. And since tensor product of modules satisfies associativity we also get canonical functorial isomorphisms $$(\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}) \otimes_{\mathcal{O}_X} \mathcal{H} = \mathcal{F} \otimes_{\mathcal{O}_X} (\mathcal{G} \otimes_{\mathcal{O}_X} \mathcal{H})$$ functorial in $\mathcal{F}$, $\mathcal{G}$, and $\mathcal{H}$. \begin{lemma} \label{lemma-stalk-tensor-product} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$, $\mathcal{G}$ be $\mathcal{O}_X$-modules. Let $x \in X$. There is a canonical isomorphism of $\mathcal{O}_{X, x}$-modules $$(\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G})_x = \mathcal{F}_x \otimes_{\mathcal{O}_{X, x}} \mathcal{G}_x$$ functorial in $\mathcal{F}$ and $\mathcal{G}$. \end{lemma} \begin{proof} Omitted. \end{proof} \begin{lemma} \label{lemma-tensor-product-sheafification} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}'$, $\mathcal{G}'$ be presheaves of $\mathcal{O}_X$-modules with sheafifications $\mathcal{F}$, $\mathcal{G}$. Then $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G} = (\mathcal{F}' \otimes_{p, \mathcal{O}_X} \mathcal{G}')^\#$. \end{lemma} \begin{proof} Omitted. \end{proof} \begin{lemma} \label{lemma-tensor-product-exact} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{G}$ be an $\mathcal{O}_X$-module. If $\mathcal{F}_1 \to \mathcal{F}_2 \to \mathcal{F}_3 \to 0$ is an exact sequence of $\mathcal{O}_X$-modules then the induced sequence $$\mathcal{F}_1 \otimes_{\mathcal{O}_X} \mathcal{G} \to \mathcal{F}_2 \otimes_{\mathcal{O}_X} \mathcal{G} \to \mathcal{F}_3 \otimes_{\mathcal{O}_X} \mathcal{G} \to 0$$ is exact. \end{lemma} \begin{proof} This follows from the fact that exactness may be checked at stalks (Lemma \ref{lemma-abelian}), the description of stalks (Lemma \ref{lemma-stalk-tensor-product}) and the corresponding result for tensor products of modules (Algebra, Lemma \ref{algebra-lemma-tensor-product-exact}). \end{proof} \begin{lemma} \label{lemma-tensor-product-pullback} Let $f : (X, \mathcal{O}_X) \to (Y, \mathcal{O}_Y)$ be a morphism of ringed spaces. Let $\mathcal{F}$, $\mathcal{G}$ be $\mathcal{O}_Y$-modules. Then $f^*(\mathcal{F} \otimes_{\mathcal{O}_Y} \mathcal{G}) = f^*\mathcal{F} \otimes_{\mathcal{O}_X} f^*\mathcal{G}$ functorially in $\mathcal{F}$, $\mathcal{G}$. \end{lemma} \begin{proof} Omitted. \end{proof} \begin{lemma} \label{lemma-tensor-product-permanence} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$, $\mathcal{G}$ be $\mathcal{O}_X$-modules. \begin{enumerate} \item If $\mathcal{F}$, $\mathcal{G}$ are locally generated by sections, so is $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}$. \item If $\mathcal{F}$, $\mathcal{G}$ are of finite type, so is $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}$. \item If $\mathcal{F}$, $\mathcal{G}$ are quasi-coherent, so is $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}$. \item If $\mathcal{F}$, $\mathcal{G}$ are of finite presentation, so is $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}$. \item If $\mathcal{F}$ is of finite presentation and $\mathcal{G}$ is coherent, then $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}$ is coherent. \item If $\mathcal{F}$, $\mathcal{G}$ are coherent, so is $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}$. \item If $\mathcal{F}$, $\mathcal{G}$ are locally free, so is $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}$. \end{enumerate} \end{lemma} \begin{proof} We first prove that the tensor product of locally free $\mathcal{O}_X$-modules is locally free. This follows if we show that $(\bigoplus_{i \in I} \mathcal{O}_X) \otimes_{\mathcal{O}_X} (\bigoplus_{j \in J} \mathcal{O}_X) \cong \bigoplus_{(i, j) \in I \times J} \mathcal{O}_X$. The sheaf $\bigoplus_{i \in I} \mathcal{O}_X$ is the sheaf associated to the presheaf $U \mapsto \bigoplus_{i \in I} \mathcal{O}_X(U)$. Hence the tensor product is the sheaf associated to the presheaf $$U \longmapsto (\bigoplus\nolimits_{i \in I} \mathcal{O}_X(U)) \otimes_{\mathcal{O}_X(U)} (\bigoplus\nolimits_{j \in J} \mathcal{O}_X(U)).$$ We deduce what we want since for any ring $R$ we have $(\bigoplus_{i \in I} R) \otimes_R (\bigoplus_{j \in J} R) = \bigoplus_{(i, j) \in I \times J} R$. \medskip\noindent If $\mathcal{F}_2 \to \mathcal{F}_1 \to \mathcal{F} \to 0$ is exact, then by Lemma \ref{lemma-tensor-product-exact} the complex $\mathcal{F}_2 \otimes \mathcal{G} \to \mathcal{F}_1 \otimes \mathcal{G} \to \mathcal{F} \otimes \mathcal{G} \to 0$ is exact. Using this we can prove (5). Namely, in this case there exists locally such an exact sequence with $\mathcal{F}_i$, $i = 1, 2$ finite free. Hence the two terms $\mathcal{F}_2 \otimes \mathcal{G}$ are isomorphic to finite direct sums of $\mathcal{G}$. Since finite direct sums are coherent sheaves, these are coherent and so is the cokernel of the map, see Lemma \ref{lemma-coherent-abelian}. \medskip\noindent And if also $\mathcal{G}_2 \to \mathcal{G}_1 \to \mathcal{G} \to 0$ is exact, then we see that $$\mathcal{F}_2 \otimes_{\mathcal{O}_X} \mathcal{G}_1 \oplus \mathcal{F}_1 \otimes_{\mathcal{O}_X} \mathcal{G}_2 \to \mathcal{F}_1 \otimes_{\mathcal{O}_X} \mathcal{G}_1 \to \mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G} \to 0$$ is exact. Using this we can for example prove (3). Namely, the assumption means that we can locally find presentations as above with $\mathcal{F}_i$ and $\mathcal{G}_i$ free $\mathcal{O}_X$-modules. Hence the displayed presentation is a presentation of the tensor product by free sheaves as well. \medskip\noindent The proof of the other statements is omitted. \end{proof} \begin{lemma} \label{lemma-tensor-commute-colimits} Let $(X, \mathcal{O}_X)$ be a ringed space. For any $\mathcal{O}_X$-module $\mathcal{F}$ the functor $$\textit{Mod}(\mathcal{O}_X) \longrightarrow \textit{Mod}(\mathcal{O}_X) , \quad \mathcal{G} \longmapsto \mathcal{F} \otimes_\mathcal{O} \mathcal{G}$$ commutes with arbitrary colimits. \end{lemma} \begin{proof} Let $I$ be a preordered set and let $\{\mathcal{G}_i\}$ be a system over $I$. Set $\mathcal{G} = \colim_i \mathcal{G}_i$. Recall that $\mathcal{G}$ is the sheaf associated to the presheaf $\mathcal{G}' : U \mapsto \colim_i \mathcal{G}_i(U)$, see Sheaves, Section \ref{sheaves-section-limits-sheaves}. By Lemma \ref{lemma-tensor-product-sheafification} the tensor product $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}$ is the sheafification of the presheaf $$U \longmapsto \mathcal{F}(U) \otimes_{\mathcal{O}_X(U)} \colim_i \mathcal{G}_i(U) = \colim_i \mathcal{F}(U) \otimes_{\mathcal{O}_X(U)} \mathcal{G}_i(U)$$ where the equality sign is Algebra, Lemma \ref{algebra-lemma-tensor-products-commute-with-limits}. Hence the lemma follows from the description of colimits in $\textit{Mod}(\mathcal{O}_X)$. \end{proof} \section{Flat modules} \label{section-flat} \noindent We can define flat modules exactly as in the case of modules over rings. \begin{definition} \label{definition-flat} Let $(X, \mathcal{O}_X)$ be a ringed space. An $\mathcal{O}_X$-module $\mathcal{F}$ is {\it flat} if the functor $$\textit{Mod}(\mathcal{O}_X) \longrightarrow \textit{Mod}(\mathcal{O}_X), \quad \mathcal{G} \mapsto \mathcal{G} \otimes_\mathcal{O} \mathcal{F}$$ is exact. \end{definition} \noindent We can characterize flatness by looking at the stalks. \begin{lemma} \label{lemma-flat-stalks-flat} Let $(X, \mathcal{O}_X)$ be a ringed space. An $\mathcal{O}_X$-module $\mathcal{F}$ is flat if and only if the stalk $\mathcal{F}_x$ is a flat $\mathcal{O}_{X, x}$-module for all $x \in X$. \end{lemma} \begin{proof} Assume $\mathcal{F}_x$ is a flat $\mathcal{O}_{X, x}$-module for all $x \in X$. In this case, if $\mathcal{G} \to \mathcal{H} \to \mathcal{K}$ is exact, then also $\mathcal{G} \otimes_{\mathcal{O}_X} \mathcal{F} \to \mathcal{H} \otimes_{\mathcal{O}_X} \mathcal{F} \to \mathcal{K} \otimes_{\mathcal{O}_X} \mathcal{F}$ is exact because we can check exactness at stalks and because tensor product commutes with taking stalks, see Lemma \ref{lemma-stalk-tensor-product}. Conversely, suppose that $\mathcal{F}$ is flat, and let $x \in X$. Consider the skyscraper sheaves $i_{x, *} M$ where $M$ is a $\mathcal{O}_{X, x}$-module. Note that $$M \otimes_{\mathcal{O}_{X, x}} \mathcal{F}_x = \left(i_{x, *} M \otimes_{\mathcal{O}_X} \mathcal{F}\right)_x$$ again by Lemma \ref{lemma-stalk-tensor-product}. Since $i_{x, *}$ is exact, we see that the fact that $\mathcal{F}$ is flat implies that $M \mapsto M \otimes_{\mathcal{O}_{X, x}} \mathcal{F}_x$ is exact. Hence $\mathcal{F}_x$ is a flat $\mathcal{O}_{X, x}$-module. \end{proof} \noindent Thus the following definition makes sense. \begin{definition} \label{definition-flat-at-point} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $x \in X$. An $\mathcal{O}_X$-module $\mathcal{F}$ is {\it flat at $x$} if $\mathcal{F}_x$ is a flat $\mathcal{O}_{X, x}$-module. \end{definition} \noindent Hence we see that $\mathcal{F}$ is a flat $\mathcal{O}_X$-module if and only if it is flat at every point. \begin{lemma} \label{lemma-colimits-flat} Let $(X, \mathcal{O}_X)$ be a ringed space. A filtered colimit of flat $\mathcal{O}_X$-modules is flat. A direct sum of flat $\mathcal{O}_X$-modules is flat. \end{lemma} \begin{proof} This follows from Lemma \ref{lemma-tensor-commute-colimits}, Lemma \ref{lemma-stalk-tensor-product}, Algebra, Lemma \ref{algebra-lemma-directed-colimit-exact}, and the fact that we can check exactness at stalks. \end{proof} \begin{lemma} \label{lemma-j-shriek-flat} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $U \subset X$ be open. The sheaf $j_{U!}\mathcal{O}_U$ is a flat sheaf of $\mathcal{O}_X$-modules. \end{lemma} \begin{proof} The stalks of $j_{U!}\mathcal{O}_U$ are either zero or equal to $\mathcal{O}_{X, x}$. Apply Lemma \ref{lemma-flat-stalks-flat}. \end{proof} \begin{lemma} \label{lemma-module-quotient-flat} Let $(X, \mathcal{O}_X)$ be a ringed space. \begin{enumerate} \item Any sheaf of $\mathcal{O}_X$-modules is a quotient of a direct sum $\bigoplus j_{U_i!}\mathcal{O}_{U_i}$. \item Any $\mathcal{O}_X$-module is a quotient of a flat $\mathcal{O}_X$-module. \end{enumerate} \end{lemma} \begin{proof} Let $\mathcal{F}$ be an $\mathcal{O}_X$-module. For every open $U \subset X$ and every $s \in \mathcal{F}(U)$ we get a morphism $j_{U!}\mathcal{O}_U \to \mathcal{F}$, namely the adjoint to the morphism $\mathcal{O}_U \to \mathcal{F}|_U$, $1 \mapsto s$. Clearly the map $$\bigoplus\nolimits_{(U, s)} j_{U!}\mathcal{O}_U \longrightarrow \mathcal{F}$$ is surjective, and the source is flat by combining Lemmas \ref{lemma-colimits-flat} and \ref{lemma-j-shriek-flat}. \end{proof} \begin{lemma} \label{lemma-flat-tor-zero} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $$0 \to \mathcal{F}'' \to \mathcal{F}' \to \mathcal{F} \to 0$$ be a short exact sequence of $\mathcal{O}_X$-modules. Assume $\mathcal{F}$ is flat. Then for any $\mathcal{O}_X$-module $\mathcal{G}$ the sequence $$0 \to \mathcal{F}'' \otimes_\mathcal{O} \mathcal{G} \to \mathcal{F}' \otimes_\mathcal{O} \mathcal{G} \to \mathcal{F} \otimes_\mathcal{O} \mathcal{G} \to 0$$ is exact. \end{lemma} \begin{proof} Using that $\mathcal{F}_x$ is a flat $\mathcal{O}_{X, x}$-module for every $x \in X$ and that exactness can be checked on stalks, this follows from Algebra, Lemma \ref{algebra-lemma-flat-tor-zero}. \end{proof} \begin{lemma} \label{lemma-flat-ses} \begin{slogan} Kernels of epimorphisms and extensions of flat sheaves of modules over a ringed space are again flat. \end{slogan} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $$0 \to \mathcal{F}_2 \to \mathcal{F}_1 \to \mathcal{F}_0 \to 0$$ be a short exact sequence of $\mathcal{O}_X$-modules. \begin{enumerate} \item If $\mathcal{F}_2$ and $\mathcal{F}_0$ are flat so is $\mathcal{F}_1$. \item If $\mathcal{F}_1$ and $\mathcal{F}_0$ are flat so is $\mathcal{F}_2$. \end{enumerate} \end{lemma} \begin{proof} Since exactness and flatness may be checked at the level of stalks this follows from Algebra, Lemma \ref{algebra-lemma-flat-ses}. \end{proof} \begin{lemma} \label{lemma-flat-resolution-of-flat} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $$\ldots \to \mathcal{F}_2 \to \mathcal{F}_1 \to \mathcal{F}_0 \to \mathcal{Q} \to 0$$ be an exact complex of $\mathcal{O}_X$-modules. If $\mathcal{Q}$ and all $\mathcal{F}_i$ are flat $\mathcal{O}_X$-modules, then for any $\mathcal{O}_X$-module $\mathcal{G}$ the complex $$\ldots \to \mathcal{F}_2 \otimes_{\mathcal{O}_X} \mathcal{G} \to \mathcal{F}_1 \otimes_{\mathcal{O}_X} \mathcal{G} \to \mathcal{F}_0 \otimes_{\mathcal{O}_X} \mathcal{G} \to \mathcal{Q} \otimes_{\mathcal{O}_X} \mathcal{G} \to 0$$ is exact also. \end{lemma} \begin{proof} Follows from Lemma \ref{lemma-flat-tor-zero} by splitting the complex into short exact sequences and using Lemma \ref{lemma-flat-ses} to prove inductively that $\Im(\mathcal{F}_{i + 1} \to \mathcal{F}_i)$ is flat. \end{proof} \noindent The following lemma gives one direction of the equational criterion of flatness (Algebra, Lemma \ref{algebra-lemma-flat-eq}). \begin{lemma} \label{lemma-flat-eq} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a flat $\mathcal{O}_X$-module. Let $U \subset X$ be open and let $$\mathcal{O}_U \xrightarrow{(f_1, \ldots, f_n)} \mathcal{O}_U^{\oplus n} \xrightarrow{(s_1, \ldots, s_n)} \mathcal{F}|_U$$ be a complex of $\mathcal{O}_U$-modules. For every $x \in U$ there exists an open neighbourhood $V \subset U$ of $x$ and a factorization $$\mathcal{O}_V^{\oplus n} \xrightarrow{A} \mathcal{O}_V^{\oplus m} \xrightarrow{(t_1, \ldots, t_m)} \mathcal{F}|_V$$ of $(s_1, \ldots, s_n)|_V$ such that $A \circ (f_1, \ldots, f_n)|_V = 0$. \end{lemma} \begin{proof} Let $\mathcal{I} \subset \mathcal{O}_U$ be the sheaf of ideals generated by $f_1, \ldots, f_n$. Then $\sum f_i \otimes s_i$ is a section of $\mathcal{I} \otimes_{\mathcal{O}_U} \mathcal{F}|_U$ which maps to zero in $\mathcal{F}|_U$. As $\mathcal{F}|_U$ is flat the map $\mathcal{I} \otimes_{\mathcal{O}_U} \mathcal{F}|_U \to \mathcal{F}|_U$ is injective. Since $\mathcal{I} \otimes_{\mathcal{O}_U} \mathcal{F}|_U$ is the sheaf associated to the presheaf tensor product, we see there exists an open neighbourhood $V \subset U$ of $x$ such that $\sum f_i|_V \otimes s_i|_V$ is zero in $\mathcal{I}(V) \otimes_{\mathcal{O}(V)} \mathcal{F}(V)$. Unwinding the definitions using Algebra, Lemma \ref{algebra-lemma-relations} we find $t_1, \ldots, t_m \in \mathcal{F}(V)$ and $a_{ij} \in \mathcal{O}(V)$ such that $\sum a_{ij}f_i|_V = 0$ and $s_i|_V = \sum a_{ij}t_j$. \end{proof} \begin{lemma} \label{lemma-flat-locally-finite-presentation} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be locally of finite presentation and flat. Then $\mathcal{F}$ is locally a direct summand of a finite free $\mathcal{O}_X$-module. \end{lemma} \begin{proof} After replacing $X$ by the members of an open covering, we may assume there exists a presentation $$\mathcal{O}_X^{\oplus r} \to \mathcal{O}_X^{\oplus n} \to \mathcal{F} \to 0$$ Let $x \in X$. By Lemma \ref{lemma-flat-eq} we can, after shrinking $X$ to an open neighbourhood of $x$, assume there exists a factorization $$\mathcal{O}_X^{\oplus n} \to \mathcal{O}_X^{\oplus n_1} \to \mathcal{F}$$ such that the composition $\mathcal{O}_X^{\oplus r} \to \mathcal{O}_X^{\oplus n} \to \mathcal{O}_X^{\oplus n_1}$ annihilates the first summand of $\mathcal{O}_X^{\oplus r}$. Repeating this argument $r - 1$ more times we obtain a factorization $$\mathcal{O}_X^{\oplus n} \to \mathcal{O}_X^{\oplus n_r} \to \mathcal{F}$$ such that the composition $\mathcal{O}_X^{\oplus r} \to \mathcal{O}_X^{\oplus n} \to \mathcal{O}_X^{\oplus n_r}$ is zero. This means that the surjection $\mathcal{O}_X^{\oplus n_r} \to \mathcal{F}$ has a section and we win. \end{proof} \section{Constructible sheaves of sets} \label{section-constructible} \noindent Let $X$ be a topological space. Given a set $S$ recall that $\underline{S}$ or $\underline{S}_X$ denotes the constant sheaf with value $S$, see Sheaves, Definition \ref{sheaves-definition-constant-sheaf}. Let $U \subset X$ be an open of a topological space $X$. We will denote $j_U$ the inclusion morphism and we will denote $j_{U!} : \Sh(U) \to \Sh(X)$ the extension by the empty set described in Sheaves, Section \ref{sheaves-section-open-immersions}. \begin{lemma} \label{lemma-surjection} Let $X$ be a topological space. Let $\mathcal{B}$ be a basis for the topology on $X$. Let $\mathcal{F}$ be a sheaf of sets on $X$. There exists a set $I$ and for each $i \in I$ an element $U_i \in \mathcal{B}$ and a finite set $S_i$ such that there exists a surjection $\coprod_{i \in I} j_{U_i!}\underline{S_i} \to \mathcal{F}$. \end{lemma} \begin{proof} Let $S$ be a singleton set. We will prove the result with $S_i = S$. For every $x \in X$ and element $s \in \mathcal{F}_x$ we can choose a $U(x, s) \in \mathcal{B}$ and $s(x, s) \in \mathcal{F}(U(x, s))$ which maps to $s$ in $\mathcal{F}_x$. By Sheaves, Lemma \ref{sheaves-lemma-j-shriek} the section $s(x, s)$ corresponds to a map of sheaves $j_{U(x, s)!}\underline{S} \to \mathcal{F}$. Then $$\coprod\nolimits_{(x, s)} j_{U(x, s)!}\underline{S} \to \mathcal{F}$$ is surjective on stalks and hence surjective. \end{proof} \begin{lemma} \label{lemma-filtered-colimit-constructibles} Let $X$ be a topological space. Let $\mathcal{B}$ be a basis for the topology of $X$ and assume that each $U \in \mathcal{B}$ is quasi-compact. Then every sheaf of sets on $X$ is a filtered colimit of sheaves of the form \begin{equation} \label{equation-towards-constructible-sets} \text{Coequalizer}\left( \xymatrix{ \coprod\nolimits_{b = 1, \ldots, m} j_{V_b!}\underline{S_b} \ar@<1ex>[r] \ar@<-1ex>[r] & \coprod\nolimits_{a = 1, \ldots, n} j_{U_a!}\underline{S_a} } \right) \end{equation} with $U_a$ and $V_b$ in $\mathcal{B}$ and $S_a$ and $S_b$ finite sets. \end{lemma} \begin{proof} By Lemma \ref{lemma-surjection} every sheaf of sets $\mathcal{F}$ is the target of a surjection whose source is a coprod $\mathcal{F}_0$ of sheaves the form $j_{U!}\underline{S}$ with $U \in \mathcal{B}$ and $S$ finite. Applying this to $\mathcal{F}_0 \times_\mathcal{F} \mathcal{F}_0$ we find that $\mathcal{F}$ is a coequalizer of a pair of maps $$\xymatrix{ \coprod\nolimits_{b \in B} j_{V_b!}\underline{S_b} \ar@<1ex>[r] \ar@<-1ex>[r] & \coprod\nolimits_{a \in A} j_{U_a!}\underline{S_a} }$$ for some index sets $A$, $B$ and $V_b$ and $U_a$ in $\mathcal{B}$ and $S_a$ and $S_b$ finite. For every finite subset $B' \subset B$ there is a finite subset $A' \subset A$ such that the coproduct over $b \in B'$ maps into the coprod over $a \in A'$ via both maps. Namely, we can view the right hand side as a filtered colimit with injective transition maps. Hence taking sections over the quasi-compact opens $U_b$, $b \in B'$ commutes with this coproduct, see Sheaves, Lemma \ref{sheaves-lemma-directed-colimits-sections}. Thus our sheaf is the colimit of the cokernels of these maps between finite coproducts. \end{proof} \begin{lemma} \label{lemma-constructible-comes-from-finite} Let $X$ be a spectral topological space. Let $\mathcal{B}$ be the set of quasi-compact open subsets of $X$. Let $\mathcal{F}$ be a sheaf of sets as in Equation (\ref{equation-towards-constructible-sets}). Then there exists a continuous spectral map $f : X \to Y$ to a finite sober topological space $Y$ and a sheaf of sets $\mathcal{G}$ on $Y$ with finite stalks such that $f^{-1}\mathcal{G} \cong \mathcal{F}$. \end{lemma} \begin{proof} We can write $X = \lim X_i$ as a directed limit of finite sober spaces, see Topology, Lemma \ref{topology-lemma-spectral-inverse-limit-finite-sober-spaces}. Of course the transition maps $X_{i'} \to X_i$ are spectral and hence by Topology, Lemma \ref{topology-lemma-directed-inverse-limit-spectral-spaces} the maps $p_i : X \to X_i$ are spectral. For some $i$ we can find opens $U_{a, i}$ and $V_{b, i}$ of $X_i$ whose inverse images are $U_a$ and $V_b$, see Topology, Lemma \ref{topology-lemma-descend-opens}. The two maps map $$\beta, \gamma : \coprod\nolimits_{b \in B} j_{V_b!}\underline{S_b} \longrightarrow \coprod\nolimits_{a \in A} j_{U_a!}\underline{S_a}$$ whose coequalizer is $\mathcal{F}$ correspond by adjunction to two families $$\beta_b, \gamma_b : S_b \longrightarrow \Gamma(V_b, \coprod\nolimits_{a \in A} j_{U_a!}\underline{S_a}), \quad b \in B$$ of maps of sets. Observe that $p_i^{-1}(j_{U_{a, i}!}\underline{S_a}) = j_{U_a!}\underline{S_a}$ and $(X_{i'} \to X_i)^{-1}(j_{U_{a, i}!}\underline{S_a}) = j_{U_{a, i'}!}\underline{S_a}$. It follows from Sheaves, Lemma \ref{sheaves-lemma-compute-pullback-to-limit} (and using that $S_b$ and $B$ are finite sets) that after increasing $i$ we find maps $$\beta_{b, i}, \gamma_{b, i} : S_b \longrightarrow \Gamma(V_{b, i}, \coprod\nolimits_{a \in A} j_{U_{a, i}!}\underline{S_a}) , \quad b \in B$$ which give rise to the maps $\beta_b$ and $\gamma_b$ after pulling back by $p_i$. These maps correspond in turn to maps of sheaves $$\beta_i, \gamma_i : \coprod\nolimits_{b \in B} j_{V_{b, i}!}\underline{S_b} \longrightarrow \coprod\nolimits_{a \in A} j_{U_{a, i}!}\underline{S_a}$$ on $X_i$. Then we can take $Y = X_i$ and $$\mathcal{G} = \text{Coequalizer}\left( \xymatrix{ \coprod\nolimits_{b = 1, \ldots, m} j_{V_{b, i}!}\underline{S_b} \ar@<1ex>[r] \ar@<-1ex>[r] & \coprod\nolimits_{a = 1, \ldots, n} j_{U_{a, i}!}\underline{S_a} } \right)$$ We omit some details. \end{proof} \begin{lemma} \label{lemma-constructible-in-constant} Let $X$ be a spectral topological space. Let $\mathcal{B}$ be the set of quasi-compact open subsets of $X$. Let $\mathcal{F}$ be a sheaf of sets as in Equation (\ref{equation-towards-constructible-sets}). Then there exist finitely many constructible closed subsets $Z_1, \ldots, Z_n \subset X$ and finite sets $S_i$ such that $\mathcal{F}$ is isomorphic to a subsheaf of $\prod (Z_i \to X)_*\underline{S_i}$. \end{lemma} \begin{proof} By Lemma \ref{lemma-constructible-comes-from-finite} we reduce to the case of a finite sober topological space and a sheaf with finite stalks. In this case $\mathcal{F} \subset \prod_{x \in X} i_{x, *}\mathcal{F}_x$ where $i_x : \{x\} \to X$ is the embedding. We omit the proof that $i_{x, *}\mathcal{F}_x$ is a constant sheaf on $\overline{\{x\}}$. \end{proof} \section{Flat morphisms of ringed spaces} \label{section-flat-morphisms} \noindent The pointwise definition is motivated by Lemma \ref{lemma-flat-stalks-flat} and Definition \ref{definition-flat-at-point} above. \begin{definition} \label{definition-flat-morphism} Let $f : X \to Y$ be a morphism of ringed spaces. Let $x \in X$. We say $f$ is said to be {\it flat at $x$} if the map of rings $\mathcal{O}_{Y, f(x)} \to \mathcal{O}_{X, x}$ is flat. We say $f$ is {\it flat} if $f$ is flat at every $x \in X$. \end{definition} \noindent Consider the map of sheaves of rings $f^\sharp : f^{-1}\mathcal{O}_Y \to \mathcal{O}_X$. We see that the stalk at $x$ is the ring map $f^\sharp_x : \mathcal{O}_{Y, f(x)} \to \mathcal{O}_{X, x}$. Hence $f$ is flat at $x$ if and only if $\mathcal{O}_X$ is flat at $x$ as an $f^{-1}\mathcal{O}_Y$-module. And $f$ is flat if and only if $\mathcal{O}_X$ is flat as an $f^{-1}\mathcal{O}_Y$-module. A very special case of a flat morphism is an open immersion. \begin{lemma} \label{lemma-pullback-flat} Let $f : X \to Y$ be a flat morphism of ringed spaces. Then the pullback functor $f^* : \textit{Mod}(\mathcal{O}_Y) \to \textit{Mod}(\mathcal{O}_X)$ is exact. \end{lemma} \begin{proof} The functor $f^*$ is the composition of the exact functor $f^{-1} : \textit{Mod}(\mathcal{O}_Y) \to \textit{Mod}(f^{-1}\mathcal{O}_Y)$ and the change of rings functor $$\textit{Mod}(f^{-1}\mathcal{O}_Y) \to \textit{Mod}(\mathcal{O}_X), \quad \mathcal{F} \longmapsto \mathcal{F} \otimes_{f^{-1}\mathcal{O}_Y} \mathcal{O}_X.$$ Thus the result follows from the discussion following Definition \ref{definition-flat-morphism}. \end{proof} \begin{definition} \label{definition-flat-module} Let $f : (X, \mathcal{O}_X) \to (Y, \mathcal{O}_Y)$ be a morphism of ringed spaces. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. \begin{enumerate} \item We say that $\mathcal{F}$ is {\it flat over $Y$ at a point $x \in X$} if the stalk $\mathcal{F}_x$ is a flat $\mathcal{O}_{Y, f(x)}$-module. \item We say that $\mathcal{F}$ is {\it flat over $Y$} if $\mathcal{F}$ is flat over $Y$ at every point $x$ of $X$. \end{enumerate} \end{definition} \noindent With this definition we see that $\mathcal{F}$ is flat over $Y$ at $x$ if and only if $\mathcal{F}$ is flat at $x$ as an $f^{-1}\mathcal{O}_Y$-module because $(f^{-1}\mathcal{O}_Y)_x = \mathcal{O}_{Y, f(x)}$ by Sheaves, Lemma \ref{sheaves-lemma-stalk-pullback}. \section{Symmetric and exterior powers} \label{section-symmetric-exterior} \noindent Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be an $\mathcal{O}_X$-module. We define the {\it tensor algebra} of $\mathcal{F}$ to be the sheaf of noncommutative $\mathcal{O}_X$-algebras $$\text{T}(\mathcal{F}) = \text{T}_{\mathcal{O}_X}(\mathcal{F}) = \bigoplus\nolimits_{n \geq 0} \text{T}^n(\mathcal{F}).$$ Here $\text{T}^0(\mathcal{F}) = \mathcal{O}_X$, $\text{T}^1(\mathcal{F}) = \mathcal{F}$ and for $n \geq 2$ we have $$\text{T}^n(\mathcal{F}) = \mathcal{F} \otimes_{\mathcal{O}_X} \ldots \otimes_{\mathcal{O}_X} \mathcal{F} \ \ (n\text{ factors})$$ We define $\wedge(\mathcal{F})$ to be the quotient of $\text{T}(\mathcal{F})$ by the two sided ideal generated by local sections $s \otimes s$ of $\text{T}^2(\mathcal{F})$ where $s$ is a local section of $\mathcal{F}$. This is called the {\it exterior algebra} of $\mathcal{F}$. Similarly, we define $\text{Sym}(\mathcal{F})$ to be the quotient of $\text{T}(\mathcal{F})$ by the two sided ideal generated by local sections of the form $s \otimes t - t \otimes s$ of $\text{T}^2(\mathcal{F})$. \medskip\noindent Both $\wedge(\mathcal{F})$ and $\text{Sym}(\mathcal{F})$ are graded $\mathcal{O}_X$-algebras, with grading inherited from $\text{T}(\mathcal{F})$. Moreover $\text{Sym}(\mathcal{F})$ is commutative, and $\wedge(\mathcal{F})$ is graded commutative. \begin{lemma} \label{lemma-local-tensor-algebra} In the situation described above. The sheaf $\wedge^n\mathcal{F}$ is the sheafification of the presheaf $$U \longmapsto \wedge^n_{\mathcal{O}_X(U)}(\mathcal{F}(U)).$$ See Algebra, Section \ref{algebra-section-tensor-algebra}. Similarly, the sheaf $\text{Sym}^n\mathcal{F}$ is the sheafification of the presheaf $$U \longmapsto \text{Sym}^n_{\mathcal{O}_X(U)}(\mathcal{F}(U)).$$ \end{lemma} \begin{proof} Omitted. It may be more efficient to define $\text{Sym}(\mathcal{F})$ and $\wedge(\mathcal{F})$ in this way instead of the method given above. \end{proof} \begin{lemma} \label{lemma-stalk-tensor-algebra} In the situation described above. Let $x \in X$. There are canonical isomorphisms of $\mathcal{O}_{X, x}$-modules $\text{T}(\mathcal{F})_x = \text{T}(\mathcal{F}_x)$, $\text{Sym}(\mathcal{F})_x = \text{Sym}(\mathcal{F}_x)$, and $\wedge(\mathcal{F})_x = \wedge(\mathcal{F}_x)$. \end{lemma} \begin{proof} Clear from Lemma \ref{lemma-local-tensor-algebra} above, and Algebra, Lemma \ref{algebra-lemma-colimit-tensor-algebra}. \end{proof} \begin{lemma} \label{lemma-pullback-tensor-algebra} Let $f : (X, \mathcal{O}_X) \to (Y, \mathcal{O}_Y)$ be a morphism of ringed spaces. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_Y$-modules. Then $f^*\text{T}(\mathcal{F}) = \text{T}(f^*\mathcal{F})$, and similarly for the exterior and symmetric algebras associated to $\mathcal{F}$. \end{lemma} \begin{proof} Omitted. \end{proof} \begin{lemma} \label{lemma-presentation-sym-exterior} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}_2 \to \mathcal{F}_1 \to \mathcal{F} \to 0$ be an exact sequence of sheaves of $\mathcal{O}_X$-modules. For each $n \geq 1$ there is an exact sequence $$\mathcal{F}_2 \otimes_{\mathcal{O}_X} \text{Sym}^{n - 1}(\mathcal{F}_1) \to \text{Sym}^n(\mathcal{F}_1) \to \text{Sym}^n(\mathcal{F}) \to 0$$ and similarly an exact sequence $$\mathcal{F}_2 \otimes_{\mathcal{O}_X} \wedge^{n - 1}(\mathcal{F}_1) \to \wedge^n(\mathcal{F}_1) \to \wedge^n(\mathcal{F}) \to 0$$ \end{lemma} \begin{proof} See Algebra, Lemma \ref{algebra-lemma-presentation-sym-exterior}. \end{proof} \begin{lemma} \label{lemma-tensor-algebra-permanence} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. \begin{enumerate} \item If $\mathcal{F}$ is locally generated by sections, then so is each $\text{T}^n(\mathcal{F})$, $\wedge^n(\mathcal{F})$, and $\text{Sym}^n(\mathcal{F})$. \item If $\mathcal{F}$ is of finite type, then so is each $\text{T}^n(\mathcal{F})$, $\wedge^n(\mathcal{F})$, and $\text{Sym}^n(\mathcal{F})$. \item If $\mathcal{F}$ is of finite presentation, then so is each $\text{T}^n(\mathcal{F})$, $\wedge^n(\mathcal{F})$, and $\text{Sym}^n(\mathcal{F})$. \item If $\mathcal{F}$ is coherent, then for $n > 0$ each $\text{T}^n(\mathcal{F})$, $\wedge^n(\mathcal{F})$, and $\text{Sym}^n(\mathcal{F})$ is coherent. \item If $\mathcal{F}$ is quasi-coherent, then so is each $\text{T}^n(\mathcal{F})$, $\wedge^n(\mathcal{F})$, and $\text{Sym}^n(\mathcal{F})$. \item If $\mathcal{F}$ is locally free, then so is each $\text{T}^n(\mathcal{F})$, $\wedge^n(\mathcal{F})$, and $\text{Sym}^n(\mathcal{F})$. \end{enumerate} \end{lemma} \begin{proof} These statements for $\text{T}^n(\mathcal{F})$ follow from Lemma \ref{lemma-tensor-product-permanence}. \medskip\noindent Statements (1) and (2) follow from the fact that $\wedge^n(\mathcal{F})$ and $\text{Sym}^n(\mathcal{F})$ are quotients of $\text{T}^n(\mathcal{F})$. \medskip\noindent Statement (6) follows from Algebra, Lemma \ref{algebra-lemma-free-tensor-algebra}. \medskip\noindent For (3) and (5) we will use Lemma \ref{lemma-presentation-sym-exterior} above. By locally choosing a presentation $\mathcal{F}_2 \to \mathcal{F}_1 \to \mathcal{F} \to 0$ with $\mathcal{F}_i$ free, or finite free and applying the lemma we see that $\text{Sym}^n(\mathcal{F})$, $\wedge^n(\mathcal{F})$ has a similar presentation; here we use (6) and Lemma \ref{lemma-tensor-product-permanence}. \medskip\noindent To prove (4) we will use Algebra, Lemma \ref{algebra-lemma-present-sym-wedge}. We may localize on $X$ and assume that $\mathcal{F}$ is generated by a finite set $(s_i)_{i \in I}$ of global sections. The lemma mentioned above combined with Lemma \ref{lemma-local-tensor-algebra} above implies that for $n \geq 2$ there exists an exact sequence $$\bigoplus\nolimits_{j \in J} \text{T}^{n - 2}(\mathcal{F}) \to \text{T}^n(\mathcal{F}) \to \text{Sym}^n(\mathcal{F}) \to 0$$ where the index set $J$ is finite. Now we know that $\text{T}^{n - 2}(\mathcal{F})$ is finitely generated and hence the image of the first arrow is a coherent subsheaf of $\text{T}^n(\mathcal{F})$, see Lemma \ref{lemma-coherent-abelian}. By that same lemma we conclude that $\text{Sym}^n(\mathcal{F})$ is coherent. \end{proof} \begin{lemma} \label{lemma-whole-tensor-algebra-permanence} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$ be a sheaf of $\mathcal{O}_X$-modules. \begin{enumerate} \item If $\mathcal{F}$ is quasi-coherent, then so is each $\text{T}(\mathcal{F})$, $\wedge(\mathcal{F})$, and $\text{Sym}(\mathcal{F})$. \item If $\mathcal{F}$ is locally free, then so is each $\text{T}(\mathcal{F})$, $\wedge(\mathcal{F})$, and $\text{Sym}(\mathcal{F})$. \end{enumerate} \end{lemma} \begin{proof} It is not true that an infinite direct sum $\bigoplus \mathcal{G}_i$ of locally free modules is locally free, or that an infinite direct sum of quasi-coherent modules is quasi-coherent. The problem is that given a point $x \in X$ the open neighbourhoods $U_i$ of $x$ on which $\mathcal{G}_i$ becomes free (resp.\ has a suitable presentation) may have an intersection which is not an open neighbourhood of $x$. However, in the proof of Lemma \ref{lemma-tensor-algebra-permanence} we saw that once a suitable open neighbourhood for $\mathcal{F}$ has been chosen, then this open neighbourhood works for each of the sheaves $\text{T}^n(\mathcal{F})$, $\wedge^n(\mathcal{F})$ and $\text{Sym}^n(\mathcal{F})$. The lemma follows. \end{proof} \section{Internal Hom} \label{section-internal-hom} \noindent Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$, $\mathcal{G}$ be $\mathcal{O}_X$-modules. Consider the rule $$U \longmapsto \Hom_{\mathcal{O}_X|_U}(\mathcal{F}|_U, \mathcal{G}|_U).$$ It follows from the discussion in Sheaves, Section \ref{sheaves-section-glueing-sheaves} that this is a sheaf of abelian groups. In addition, given an element $\varphi \in \Hom_{\mathcal{O}_X|_U}(\mathcal{F}|_U, \mathcal{G}|_U)$ and a section $f \in \mathcal{O}_X(U)$ then we can define $f\varphi \in \Hom_{\mathcal{O}_X|_U}(\mathcal{F}|_U, \mathcal{G}|_U)$ by either precomposing with multiplication by $f$ on $\mathcal{F}|_U$ or postcomposing with multiplication by $f$ on $\mathcal{G}|_U$ (it gives the same result). Hence we in fact get a sheaf of $\mathcal{O}_X$-modules. We will denote this sheaf $\SheafHom_{\mathcal{O}_X}(\mathcal{F}, \mathcal{G})$. There is a canonical evaluation'' morphism $$\mathcal{F} \otimes_{\mathcal{O}_X} \SheafHom_{\mathcal{O}_X}(\mathcal{F}, \mathcal{G}) \longrightarrow \mathcal{G}.$$ For every $x \in X$ there is also a canonical morphism $$\SheafHom_{\mathcal{O}_X}(\mathcal{F}, \mathcal{G})_x \to \Hom_{\mathcal{O}_{X, x}}(\mathcal{F}_x, \mathcal{G}_x)$$ which is rarely an isomorphism. \begin{lemma} \label{lemma-internal-hom} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$, $\mathcal{G}$, $\mathcal{H}$ be $\mathcal{O}_X$-modules. There is a canonical isomorphism $$\SheafHom_{\mathcal{O}_X} (\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G}, \mathcal{H}) \longrightarrow \SheafHom_{\mathcal{O}_X} (\mathcal{F}, \SheafHom_{\mathcal{O}_X}(\mathcal{G}, \mathcal{H}))$$ which is functorial in all three entries (sheaf Hom in all three spots). In particular, to give a morphism $\mathcal{F} \otimes_{\mathcal{O}_X} \mathcal{G} \to \mathcal{H}$ is the same as giving a morphism $\mathcal{F} \to \SheafHom_{\mathcal{O}_X}(\mathcal{G}, \mathcal{H})$. \end{lemma} \begin{proof} This is the analogue of Algebra, Lemma \ref{algebra-lemma-hom-from-tensor-product}. The proof is the same, and is omitted. \end{proof} \begin{lemma} \label{lemma-internal-hom-exact} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$, $\mathcal{G}$ be $\mathcal{O}_X$-modules. \begin{enumerate} \item If $\mathcal{F}_2 \to \mathcal{F}_1 \to \mathcal{F} \to 0$ is an exact sequence of $\mathcal{O}_X$-modules, then $$0 \to \SheafHom_{\mathcal{O}_X}(\mathcal{F}, \mathcal{G}) \to \SheafHom_{\mathcal{O}_X}(\mathcal{F}_1, \mathcal{G}) \to \SheafHom_{\mathcal{O}_X}(\mathcal{F}_2, \mathcal{G})$$ is exact. \item If $0 \to \mathcal{G} \to \mathcal{G}_1 \to \mathcal{G}_2$ is an exact sequence of $\mathcal{O}_X$-modules, then $$0 \to \SheafHom_{\mathcal{O}_X}(\mathcal{F}, \mathcal{G}) \to \SheafHom_{\mathcal{O}_X}(\mathcal{F}, \mathcal{G}_1) \to \SheafHom_{\mathcal{O}_X}(\mathcal{F}, \mathcal{G}_2)$$ is exact. \end{enumerate} \end{lemma} \begin{proof} Omitted. \end{proof} \begin{lemma} \label{lemma-stalk-internal-hom} Let $(X, \mathcal{O}_X)$ be a ringed space. Let $\mathcal{F}$, $\mathcal{G}$ be $\mathcal{O}_X$-modules. If $\mathcal{F}$ is finitely presented then the canonical map $$\SheafHom_{\mathcal{O}_X}(\mathcal{F}, \mathcal{G})_x \to \Hom_{\mathcal{O}_{X, x}}(\mathcal{F}_x, \mathcal{G}_x)$$ is an isomorphism. \end{lemma} \begin{proof} By localizing on $X$ we may assume that $\mathcal{F}$ has a presentation $$\bigoplus\nolimits_{j = 1, \ldots, m} \mathcal{O}_X \longrightarrow \bigoplus\nolimits_{i = 1, \ldots, n} \mathcal{O}_X \to \mathcal{F} \to 0.$$ By Lemma \ref{lemma-internal-hom-exact} this gives an exact sequence $0 \to \SheafHom_{\mathcal{O}_X}(\mathcal{F}, \mathcal{G}) \to \bigoplus\nolimits_{i = 1, \ldots, n} \mathcal{G} \longrightarrow \bigoplus\nolimits_{j = 1, \ldots, m} \mathcal{G}.$ Taking stalks we get an exact sequence