Sheaf of Modules¶

A sheaf whose local sections are modules over a sheaf of rings, with scalar multiplication compatible with restriction and gluing.

Version: v1 · 2026-10-03 · History
Domain-specific #: 13608
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomains: Algebraic Geometry, Sheaf Theory → Mathematics

Core Idea¶

A sheaf of modules does two things at once. It assigns compatible local sections to open subsets of a space, as a sheaf does, and it permits those sections to be added and multiplied by local scalars. The scalars themselves vary over the space: they are supplied by a sheaf of rings \(\mathcal O\) on \(X\). For each open \(U\), \(\mathcal F(U)\) is an \(\mathcal O(U)\)-module, and restriction from \(U\) to \(V\subseteq U\) must obey \((fs)|_V=f|_V\,s|_V\). Compatible sections over an open cover also glue uniquely. Stacks Project Definition 6.10.1 packages this as a presheaf of \(\mathcal O\)-modules whose underlying presheaf of abelian groups is a sheaf.^[1]

The module action is the residual that a general sheaf lacks. A ringed space \((X,\mathcal O)\) supplies the variable scalar system, but does not by itself specify the additional object \(\mathcal F\). Conversely, a bare module over one global ring does not yet say how its elements restrict and glue over a chosen space. The abstraction is the coupled, local-to-global linear structure, not simply the word “module” appearing in a geometric setting.

Smooth vector fields on a manifold and module-associated sheaves on an affine scheme exhibit the same construction with quite different local sections. In the first case the scalars are smooth functions and the sections are vector fields. In the second, an \(R\)-module \(M\) produces a sheaf \(\widetilde M\) on \(\operatorname{Spec}R\), with \(M_f\) on standard opens \(D(f)\). The affine correspondence is special: it classifies quasi-coherent sheaves on an affine scheme, not every sheaf of modules on every ringed space.^[2]^[3]

Structural Signature¶

Sig role-phrases: ringed space and scalar sheaf → local module sections → linear restriction → unique gluing → stalkwise algebraic tests → optional geometric subclasses.

Ringed space and scalar sheaf. \(X\) has a sheaf of rings \(\mathcal O\), not merely a single unlabeled coefficient ring. Its restriction maps tell which scalar on \(V\) corresponds to a scalar on \(U\). A scheme and a smooth manifold are different possible bases; neither is obligatory for the generic definition.^[1]
Local module sections. Every \(\mathcal F(U)\) is an \(\mathcal O(U)\)-module. Addition and scalar multiplication are therefore meaningful on the same open set. An arbitrary sheaf of sets does not meet this type requirement.^[1]
Linear restriction. The restriction maps for \(\mathcal F\) are additive and commute with the corresponding restrictions of \(\mathcal O\). Thus localizing a calculation and then multiplying gives the same section as multiplying before restriction.^[1]
Unique gluing. If sections on a cover agree after restricting to pairwise overlaps, they assemble uniquely into a section on the union. A presheaf can satisfy all the module laws and still fail this sheaf axiom.^[1]
Stalkwise algebraic tests. A morphism between module sheaves is \(\mathcal O\)-linear locally. Kernels are obtained sectionwise, but the sheaf cokernel is the sheafification of the sectionwise quotient; exactness is tested on stalks. These are consequences of the identity, not further admission criteria.^[4]
Optional geometric subclasses. Quasi-coherent, coherent, locally free and invertible sheaves impose extra local-presentation, finiteness or freeness requirements. An arbitrary \(\mathcal O\)-module sheaf need not belong to any of them.^[5]

What It Is Not¶

It is not just a sheaf. The live Sheaf node already accounts for restriction, locality and gluing. An \(\mathcal O\)-module sheaf adds modules over a specified ring sheaf and linearity of restriction. Two otherwise identical underlying abelian sheaves can carry different module actions, so forgetting the action can erase a meaningful distinction.^[1]

It is not the ringed space itself. The pair \((X,\mathcal O)\) is the ambient base. \(\mathcal F\) is a further sheaf acted on by \(\mathcal O\); the structure sheaf \(\mathcal O\) is an important self-module example, but an example must not replace the general definition.

It is not automatically quasi-coherent, coherent, locally free, or a sheaf of algebras. Quasi-coherence requires a local presentation by free module sheaves; on general ringed spaces its behavior is less well controlled than on schemes. Coherence and local freeness restrict further. A sheaf of algebras additionally has compatible multiplication among its own sections, not merely scalar action from \(\mathcal O\).^[5]

Nor is it the category of all such sheaves or the property of being generated by global sections. Those are related, separately typed constructions, not synonyms of one module sheaf.

Scope of Application¶

The formal definition applies to sheaves of modules over a sheaf of rings on a topological space. One can use it in algebraic geometry, differential geometry, and other ringed-space contexts without first assuming that the base is a scheme or that the module has finite rank. In differential geometry, smooth functions act pointwise on smooth sections of vector bundles. In algebraic geometry, the structure sheaf acts on many geometric sheaves, including \(\widetilde M\) on affine schemes.^[2]^[3]

Results about particular subclasses must carry their hypotheses. On \(X=\operatorname{Spec}R\), the functors \(M\mapsto\widetilde M\) and \(\mathcal F\mapsto\Gamma(X,\mathcal F)\) identify \(R\)-modules with quasi-coherent \(\mathcal O_X\)-modules. That does not establish such an equivalence for all \(\mathcal O_X\)-modules or for every ringed space. Even the pleasant closure properties of quasi-coherent sheaves require care outside schemes.^[3]^[5]

Clarity¶

To recognize the object, ask four typed questions in order: What is \(X\)? What sheaf of rings \(\mathcal O\) supplies the scalars? What modules \(\mathcal F(U)\) are assigned to opens? Do the restrictions preserve scalar multiplication and do compatible sections glue uniquely? This separates the sheaf-of-modules assertion from subsequent questions about rank, presentation, support or cohomology.^[1]

The order matters. One can verify module arithmetic separately on every open and still have no sheaf because compatible local sections fail to glue. One can have a sheaf of abelian groups with perfect gluing but no chosen \(\mathcal O\)-action. And one can know that \(\mathcal F\) is an \(\mathcal O\)-module sheaf without knowing whether it is generated by global sections. Each missing or added condition changes the claim.

Manages Complexity¶

A global geometric object can have locally different coordinate descriptions. The sheaf-of-modules structure allows linear calculations to be performed on the opens where data are manageable, while restriction and gluing explain how those calculations relate. The smooth tangent sheaf, for instance, has coordinate bases locally even when no single global basis exists; the module sheaf records their common invariant object rather than privileging one chart. An affine quasi-coherent sheaf can similarly be studied via localizations \(M_f\) instead of a separate unrelated module for every open.^[2]^[3]

This compression is useful only when it preserves the exact type of operation. The sectionwise kernel of a sheaf map is already a sheaf, but the sectionwise quotient used to form a cokernel may require sheafification. Declaring that every construction is simply “open by open” would hide a precisely relevant local-to-global failure.^[4]

Abstract Reasoning¶

The constitutive inference begins with an \(\mathcal O\)-module presheaf. On each inclusion \(V\subseteq U\), restriction must be a homomorphism relative to \(\mathcal O(U)\to\mathcal O(V)\). The sheaf axiom then asks whether local sections that agree on overlaps exist as one section, and whether that section is unique. Only after both tests pass may we invoke theorems about module sheaves.^[1]

Consider a map \(\varphi:\mathcal F\to\mathcal G\). Its kernel can be tested by asking, on each open, which sections map to zero. For the cokernel, however, the presheaf \(U\mapsto\mathcal G(U)/\varphi(\mathcal F(U))\) is only the starting object; sheafification gives the actual sheaf cokernel. Stalks then recover ordinary module kernels and cokernels, and exactness can be checked pointwise at all stalks. This explains why the category is abelian while naive openwise quotients can fail to be the correct object.^[4]

An additional predicate can narrow the class. For example, quasi-coherence asks for local presentations in terms of sums of copies of \(\mathcal O\). It is not a hidden clause in the sheaf-of-modules definition. Keeping the predicate separate prevents a theorem about affine schemes from being smuggled into an arbitrary ringed space.^[5]^[3]

Knowledge Transfer¶

The same recognition test transfers from smooth geometry to affine algebraic geometry: choose the base scalar sheaf, construct local modules, verify restriction compatibility, and verify gluing. What changes is the meaning of the sections and the available supplementary theorems. Vector fields are smooth sections with pointwise smooth-function action; \(\widetilde M\) has algebraic sections governed by localization on \(D(f)\). Neither setting is a metaphor for the other: both satisfy the literal module-sheaf axioms.^[2]^[3]

The affine equivalence is not transferable merely because a different ringed space also has a structure sheaf. The Stacks Project explicitly warns that quasi-coherent sheaves can have poor category behavior on general ringed spaces. Knowledge transfer therefore requires a second, setting-specific check of the subclass and theorem hypotheses after the shared object identity has been established.^[5]

Examples¶

Smooth vector fields on a manifold¶

Let \(X\) be a smooth manifold, \(\mathcal O(U)=C^\infty(U)\), and \(\mathcal F(U)=\mathfrak X(U)\), the smooth vector fields on \(U\). A smooth function \(f\) multiplies a vector field \(V\) pointwise to give \(fV\). Restricting both to \(V_0\subseteq U\) gives \((fV)|_{V_0}=f|_{V_0}V|_{V_0}\). Smooth vector fields defined on covering coordinate patches and agreeing on overlaps glue into one field. A local coordinate frame exhibits local freeness, but this is an added property of this example, not the definition of every sheaf of modules.^[2]^[1]

Mapped back: smooth manifold and smooth-function sheaf → modules of local vector fields → scalar-compatible restriction → gluing of agreeing fields → a locally free example of the general module-sheaf identity.

A module on an affine scheme¶

Let \(R=k[t]\), \(M=R/(t)\), and \(X=\operatorname{Spec}R\). The associated sheaf \(\widetilde M\) has \(\widetilde M(D(f))=M_f\); further restriction corresponds to further localization. Its stalk at the closed point \((t)\) is nonzero, whereas away from that point the stalk is zero. Thus the local modules can encode a point-supported algebraic object rather than a smooth vector field spread across a manifold. The sheaf is quasi-coherent by construction but need not be locally free of a fixed positive rank. The general affine equivalence concerns quasi-coherent \(\mathcal O_X\)-modules.^[3]

Mapped back: affine scheme and structure sheaf → localized modules \(M_f\) → compatible localization/restriction → associated-sheaf gluing → quasi-coherent, point-supported instance.

Near miss: local modules without gluing¶

Suppose each open receives an \(\mathcal O(U)\)-module and each inclusion a compatible linear map, but two agreeing sections on overlapping opens have no section on their union. These data form an \(\mathcal O\)-module presheaf, not a sheaf of modules. Sheafification can repair the missing gluing, but doing so constructs a new object rather than proving the original assignment already had the property.^[1]

Structural Tensions¶

Sectionwise calculation vs. sheaf-level result. Computing modules over a chosen open makes algebra concrete, yet sheaf-level cokernels and colimits may require new glued local sections. Neither the sectionwise calculation nor the gluing condition can be discarded. Diagnostic: Does the proposed output satisfy the sheaf axiom as stated, or must the sectionwise construction be sheafified?^[4]
General module sheaf vs. stronger geometric subclass. The broad definition accommodates smooth sections and many algebraic sheaves. Stronger claims—affine module equivalence, finite presentation, local freeness—need extra hypotheses. Narrowing all sheaves to these subclasses loses genuine examples, while applying their theorems to every sheaf overstates what the definition proves. Diagnostic: Which particular local-presentation or finiteness condition supports this downstream claim?^[5]^[3]

Structural–Framed Character¶

The entry is structural within a mathematical frame. Evaluative weight: satisfying module and gluing laws is a mathematical fact, not a judgment that a sheaf is useful or elegant. Human-practice dependence: mathematicians choose a base space, scalar sheaf and notation, but the resulting restriction and gluing identities can be proved or refuted independently of preference. Institutional origin: the terminology arose in mathematical practice; no institutional declaration makes incompatible local sections glue. Vocabulary travel: the exact \(\mathcal O\)-module laws travel from smooth manifolds to affine schemes, whereas loose talk of “local modules” without a ringed space does not. Import versus recognition: one recognizes the structure by checking the local module action and sheaf axiom, not by importing the term into any spatially indexed collection.^[1]^[3]

Its character: a structurally reusable, domain-specific mathematical object. The pattern “local linear data plus restriction and gluing” is not itself a Prime abstraction because \(\mathcal O(U)\)-modules and sheaf axioms are indispensable, not optional examples of a context-free mechanism.

Structural Core vs. Domain Accent¶

The core is the paired assignment \(U\mapsto(\mathcal O(U),\mathcal F(U))\), the scalar-compatible restriction law, and the existence and uniqueness of gluing. The domain accent determines what the sections mean: vector fields, functions, algebraic modules or another mathematical object. Those accents can change without changing the formal identity, provided the same axioms are proved.

The core should not be abstracted so far that “local things combine” becomes enough. An arbitrary collection of local records might glue, but without an acting sheaf of rings and module-compatible restrictions it is not a sheaf of modules. Conversely, a specific affine \(\widetilde M\) should not be made the universal template; it is one notably tractable subclass instance.

This entry is a kind of Sheaf.

The proposed upward edge is to the live domain-specific Sheaf node, by strict subsumption. The module action is the added identity-bearing condition. The live Ringed Space node supplies ambient structure but is not a genus of \(\mathcal F\); a module sheaf is not itself the pair \((X,\mathcal O)\). No Prime parent is asserted merely because local-to-global reasoning sounds portable.

Relationships to Other Abstractions¶

Current abstraction Sheaf of Modules Domain-specific

Parents (1) — more general patterns this builds on

Sheaf of Modules is a kind of Sheaf Domain-specific

Every sheaf of modules is an underlying sheaf with additional local module structure.

Hierarchy path (1) — routes to 1 parentless root

Sheaf of Modules → Sheaf

Neighborhood in Abstraction Space¶

Sheaf of Modules sits in a moderately populated region (50^th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Sheaves & Birational Geometry (22 abstractions)

Nearest neighbors

Ringed Space — 0.89
Picard Group — 0.88
Ideal sheaf — 0.86
Associative algebra — 0.86
Algebraic Variety — 0.85

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With¶

Sheaf. More general local-to-global assignment; module action is not required.
Ringed space. The space and scalar sheaf on which an \(\mathcal O\)-module sheaf is defined.
Coherent or quasi-coherent sheaf. Extra local-presentation or finiteness conditions, not synonyms for all module sheaves.
Sheaf of algebras. Adds compatible internal multiplication and unit beyond an \(\mathcal O\)-module action.
Category of sheaves of abelian groups. A category of objects, not one instance of a sheaf of modules.
Sheaf generated by global sections. A property enjoyed by some module sheaves, not required in general.

References¶

[1] The Stacks Project, “Sheaves of modules,” Section 6.10, Definition 6.10.1, original mathematical text. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k

[2] MIT Mathematics, “Automorphic Forms and Vector Bundles,” original lecture note, opening definition of smooth sections and section-space construction, HTML text. The ordinary smooth-vector-field module example is derived from pointwise scalar multiplication and the Stacks definition. registry ↩a ↩b ↩c ↩d ↩e

[3] The Stacks Project, “Quasi-coherent sheaves on affines,” Section 26.7, Lemmas 26.7.1–26.7.5, original mathematical text. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i

[4] The Stacks Project, “The abelian category of sheaves of modules,” Section 17.3, construction preceding and including Lemma 17.3.1, original mathematical text. registry ↩a ↩b ↩c ↩d

[5] The Stacks Project, “Quasi-coherent modules,” Section 17.10, opening warning and Definition 17.10.1, original mathematical text. registry ↩a ↩b ↩c ↩d ↩e ↩f