Alphabeta Math
TheoremStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedaudited 2026-10-02
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

Rational sections of line bundles are Cartier divisors

Statement

Let X be an integral scheme (Integral schemes), let L be an invertible OX-module (Invertible sheaves) and let s∈Γ(X,KX(L)) be a rational section, i.e. a nonzero meromorphic section of L (Rational section line bundle). Then:

  1. (the divisor) the coefficients of s in any trivialization glue to a well-defined Cartier divisor div⁡C(s) on X (Cartier divisor);
  2. (the pair) there is a canonical isomorphism φ:OX(div⁡C(s))→L of OX-modules carrying the canonical rational section 1D of OX(D), D=div⁡C(s), to s; hence (L,s)≅(OX(D),1D) (Invertible sheaf of cartier divisor, The sheaf of a Cartier divisor is invertible);
  3. (surjectivity) conversely, for every Cartier divisor D on X the canonical section 1D is a rational section of the invertible sheaf OX(D) and div⁡C(1D)=D;
  4. (the equivalence relation) for two rational sections s,s′ one has div⁡C(s)=div⁡C(s′) if and only if there is an isomorphism ψ:L→L′ of invertible sheaves with ψ(s)=s′; thus div⁡C induces a bijection from isomorphism classes of pairs (L,s) to Cartier divisors on X.

Facts & Assumptions

Given: An integral scheme X with generic point η and function field K(X)=OX,η (Integral schemes), an invertible OX-module L, and a nonzero meromorphic section s∈Γ(X,KX(L)), KX(L)=L⊗OXKX (Rational section line bundle, Tensor product of sheaves of modules).

[F1]

On the integral scheme X the sheaf KX is the constant sheaf with value K(X), and KX(L) is the constant sheaf with value the stalk Lη, a one-dimensional K(X)-vector space; every nonempty open subset of X contains η, is irreducible and connected, and a rational section is a nonzero element of that one-dimensional space, equal to a K(X)-multiple of the germ of any chosen local generator (Rational section line bundle, Sheaf total quotient rings, Integral schemes).

[F2]

L is invertible, i.e. locally free of rank one: for every x∈X there is an open neighbourhood U and a generator e∈L(U) such that OU→L∣U, a↦ae, is an isomorphism; if two sections generate L on a common nonempty open W, then each is a unit multiple of the other, and the unit is a unit of OX(W) (Invertible sheaves).

[F3]

A Cartier divisor on X is a global section of KX×/OX×; a family of meromorphic units fi∈KX(Ui)× with fi/fj∈OX(Ui∩Uj)× glues along the cover to such a global section, and refining the cover or replacing the fi by unit multiples does not change it (Cartier divisor).

[F4]

For a Cartier divisor D with local datum (Ui,fi) the sheaf OX(D)⊆KX satisfies OX(D)∣Ui=fi−1OUi, is invertible and freely generated by fi−1; the constant meromorphic function 1 is a global section of KX(OX(D)), the canonical rational section 1D, with 1D∣Ui=fi⋅fi−1 (Invertible sheaf of cartier divisor, The sheaf of a Cartier divisor is invertible).

[F5]

Sheaf sections and morphisms that agree on an open cover glue uniquely (A sheaf on a topological space).

[F6]

A Cartier divisor is effective exactly when its local equations lie in OX and their germs act injectively by multiplication; this condition is unchanged by multiplying equations by regular units (Effective cartier divisor).

Proof

1.1F1F2

Coefficients of a rational section. Let A be the set of all pairs (U,e) with U⊆X a nonempty open subset and e∈L(U) a generator of L over U; by [F2] every point of X lies in the first component of a member of A, so these pairs cover X, and no choice is used because A is determined by a formula. By [F1] the sheaf KX(L) is the constant sheaf with value the one-dimensional K(X)-vector space Lη and the sections over the connected open U are exactly Lη, so for a pair (U,e) the germ eη is a nonzero vector and there is a unique f∈K(X) with s∣U=fe; since s≠0, this f lies in K(X)×. Write f=f(U,e) for this coefficient.

2.1F1F2step 1.1

Unit ratios. Let (U,e),(V,e′) be pairs of A with nonempty intersection W. Both e∣W and e′∣W generate L∣W, so e∣W=u e′∣W for a unit u∈OX(W)× by [F2]. Restricting the identities s∣U=f(U,e)e and s∣V=f(V,e′)e′ to W and substituting gives f(U,e)u e′∣W=f(V,e′)e′∣W in the one-dimensional K(X)-vector space Lη of [F1]; since the germ eη′ is nonzero, f(U,e)u=f(V,e′) and hence f(U,e)/f(V,e′)=u−1∈OX(W)×.

2.2F3F4step 1.1

Every Cartier divisor arises. Let D be a Cartier divisor on X with local datum (Ui,fi) consisting of meromorphic units with unit ratios on overlaps [F3]. Since X is integral each fi lies in K(X)× and, by [F4], OX(D) is invertible with generator fi−1 on Ui and canonical rational section 1D satisfying 1D∣Ui=fi⋅fi−1. The coefficient of 1D in the trivialization fi−1 is therefore fi, so div⁡C(1D) is represented by the same datum (Ui,fi) and equals D by [F3]; in particular 1D≠0 is a rational section and the construction is surjective onto CaDiv⁡(X).

3.1F3step 2.1

The Cartier divisor. The coefficients f(U,e)∈KX(U)×=K(X)× have unit ratios on overlaps by step 2.1, so by [F3] they glue to a global section div⁡C(s)∈CaDiv⁡(X), and step 2.1 shows moreover that any two pairs give local equations differing by a unit, so the class is independent of all choices.

3.2F4step 2.1

The local isomorphisms. Fix a pair (U,e)∈A and put f=f(U,e). By [F4] the sheaf OX(D), D=div⁡C(s), is freely generated by f−1 over U, so there is a unique OU-linear map φ(U,e):OX(D)∣U→L∣U with φ(U,e)(f−1)=e; because it carries a generator to a generator, it is an isomorphism. Let (V,e′) be a second pair with coefficient h=f(V,e′) and put W=U∩V; by step 2.1 we have h=fu and e=ue′ there for a unit u, so f−1=uh−1 and φ(U,e)(h−1)=φ(U,e)(u−1f−1)=u−1e=e′=φ(V,e′)(h−1); the maps agree on the generator h−1 of OX(D)∣W, hence agree on all sections.

4.1F5step 3.2

Gluing. By step 3.2 the isomorphisms φ(U,e) agree on all intersections, so by [F5] they glue to an OX-linear morphism φ:OX(D)→L, which is an isomorphism because it restricts to an isomorphism on each member of the cover.

4.2F1F2F5F6step 1.1step 3.1

Global sections and effectivity. The inclusion OX↪KX induces an inclusion L↪KX(L), as seen in any frame. Thus s belongs to Γ(X,L) exactly when every frame coefficient f(U,e) belongs to OX(U): these local sections then glue by [F5]. Each such coefficient has a nonzero germ at every point of U, since a zero germ would make it vanish on a nonempty open containing η, contrary to f(U,e)≠0 in K(X). The local rings are domains, so multiplication by these germs is injective. By [F6], this is equivalent to D=div⁡C(s) being effective. Conversely, any effective local equation differs from a frame coefficient by a regular unit, so all frame coefficients lie in OX and s is a global section of L.

5.1F4F5step 4.1

The canonical section maps to s. The constant meromorphic function 1 is the rational section 1D of OX(D) with 1D∣U=f⋅f−1 for every pair (U,e) by [F4]. Under the induced map KX(φ) its restriction to U maps to KX(φ(U,e))(ff−1)=fφ(U,e)(f−1)=fe=s∣U by step 3.2, so KX(φ)(1D)=s by the uniqueness in [F5]. Hence (L,s)≅(OX(D),1D), which proves 2.

6.1F3step 1.1step 2.2step 3.2step 4.1step 5.1

Pair isomorphism and equality of divisors. Let s∈Γ(X,KX(L)) and s′∈Γ(X,KX(L′)) be rational sections with div⁡C(s)=D and div⁡C(s′)=D′. If D=D′, steps 3.2, 4.1 and 5.1 produce isomorphisms φ:OX(D)→L with φ(1D)=s and φ′:OX(D)→L′ with φ′(1D)=s′, so ψ=φ′∘φ−1:L→L′ is an isomorphism with ψ(s)=s′. Conversely, if ψ:L→L′ is an isomorphism with ψ(s)=s′ and (U,e) is a pair for L with coefficient f, then (U,ψ(e)) is a pair for L′ and ψ(s)∣U=ψ(fe)=fψ(e) shows that its coefficient is again f; hence the two families of local equations define the same Cartier divisor, that is, div⁡C(s)=div⁡C(s′). Therefore div⁡C is a bijection from isomorphism classes of pairs to Cartier divisors.

7.1step 3.1step 5.1step 2.2step 6.1∎

Conclusion. Assertion 1 is step 3.1, assertion 2 is steps 3.2, 4.1 and 5.1, assertion 3 is step 2.2, and assertion 4 is step 6.1; this proves the theorem.

No choice principle is used: the family of pairs (U,e) is determined by a formula, the coefficients are uniquely determined by s and e, and the gluing maps are the unique maps on free generators. For L=OX and s=1 the coefficient is 1 on every chart, so div⁡C(1)=0 and φ is the identity isomorphism OX(0)=OX. More generally for a principal rational function s=g∈K(X)× on L=OX the divisor div⁡C(g) is the principal Cartier divisor of g. For example, on X=Spec⁡k[t] and D=−[0], OX(D)=tOX and 1D=1 is rational but is not a global section of that sheaf. By step 4.2, s belongs to Γ(X,L) exactly when D is effective. This global-section condition is stronger than being a regular meromorphic section in the terminology of Rational section line bundle, where “regular” means nonzero; the present theorem allows arbitrary poles. The scheme X is integral, hence nonempty, so there is no empty case; and no Noetherian, normal or separatedness hypothesis is needed, since only the constant-sheaf description of KX(L) and the local description of OX(D) are used.

Depends on

Used by

Dependency tree · two levels

34 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources