Alphabeta Math
LemmaStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedjudge pass (gpt-6.1-sol)audited 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.

Addition of Cartier divisors is tensor product of their sheaves

Statement

Let X be a scheme and let D,E be Cartier divisors on X (Cartier divisor), with associated invertible sheaves OX(D) and OX(E) (Invertible sheaf of cartier divisor). Then there are canonical isomorphisms of OX-modules OX(D+E)≅OX(D)⊗OXOX(E),OX(−D)≅OX(D)∨, where OX(D)∨=HomOX(OX(D),OX) is the dual sheaf (The internal Hom sheaf of two module sheaves). If D and E are represented on a common open cover {Ui} by meromorphic units fi and gi with regular-unit ratios, then the first isomorphism carries the local generator (figi)−1 of OX(D+E)∣Ui to fi−1⊗gi−1, and the second carries fi∈OX(−D)(Ui) to the functional λi on OX(D)∣Ui=fi−1OUi with λi(fi−1)=1.

Facts & Assumptions

Given: A scheme X and Cartier divisors D,E on X, represented on a common open cover {Ui}i∈I by meromorphic units fi∈KX(Ui)× for D and gi∈KX(Ui)× for E, with fi/fj,gi/gj∈OX(Ui∩Uj)×.

[F1]

A Cartier divisor is a global section of KX×/OX×; the group law is induced by multiplication of local equations, so that if D is represented by (Ui,fi) and E by (Ui,gi) on a common cover then D+E is represented by (Ui,figi), the zero divisor 0 is represented by the constant equation 1, and −D is represented by (Ui,fi−1); passing to a common refinement or replacing equations by regular-unit multiples does not change the divisor (Cartier divisor).

[F2]

For a Cartier divisor D with equations fi one has OX(D)∣Ui=fi−1OUi⊆KX, the sheaf OX(D) is well defined independently of the datum, and OX(0)=OX; the section fi−1 generates OX(D) on Ui, so OX(D) is invertible (Invertible sheaf of cartier divisor, The sheaf of a Cartier divisor is invertible).

[F3]

The tensor product of OX-modules is the sheafification of the presheaf tensor product; if L∣U=OU ⁣⋅s and M∣U=OU ⁣⋅t are free of rank one on an open set U, then L⊗M∣U=OU ⁣⋅(s⊗t) is free of rank one on U. The dual HomOX(L,OX) of a free rank-one module OU is free of rank one with dual basis λ characterised by λ(1)=1, and formation of duals and tensor products is compatible with restriction to open subsets (Tensor product of sheaves of modules, The internal Hom sheaf of two module sheaves, Invertible sheaves).

[F4]

For an invertible sheaf L the evaluation pairing L∨⊗L→OX, φ⊗s↦φ(s), is an isomorphism, and the transition units of L∨ are the inverses of those of L (Dual of a line bundle is its tensor inverse).

[F5]

Sections of a sheaf on an open cover glue uniquely when they agree on the pairwise overlaps (A sheaf on a topological space).

Proof

1.1F1F2

Sum and inverse equations. On the common cover, D+E is represented by the equations figi and −D by fi−1, and the associated sheaves satisfy OX(D+E)∣Ui=(figi)−1OUi, OX(D)∣Ui=fi−1OUi, OX(E)∣Ui=gi−1OUi and OX(−D)∣Ui=fiOUi.

1.2F2F3

Dual of a local generator. For each i let λi∈HomOX(OX(D),OX)(Ui) be the functional determined by λi(fi−1)=1; it is a basis of the free rank-one OUi-module OX(D)∨∣Ui by [F3]. Hence OX(D)∨∣Ui=OUi ⁣⋅λi.

2.1F2F3F5step 1.1

The addition isomorphism. For each i there is a unique OUi-linear isomorphism φi:OX(D)⊗OX(E)∣Ui→OX(D+E)∣Ui sending a (fi−1⊗gi−1) to a (figi)−1, and the φi agree on overlaps and glue to a global isomorphism φ:OX(D)⊗OX(E)→OX(D+E) by [F5]. Indeed, both sides are free of rank one on Ui, with the displayed generators. On an overlap Ui∩Uj write u=fi/fj and v=gi/gj, units of OX(Ui∩Uj); then fi−1⊗gi−1=(uv)−1 (fj−1⊗gj−1) and (figi)−1=(uv)−1 (fjgj)−1, so the transition units of source and target coincide in the displayed trivialisations and φi, φj agree on the overlap. Hence the φi glue, and the glued map is an isomorphism because it is one on every chart.

2.2F2F3F5step 1.2

The inverse isomorphism. There are unique OUi-linear isomorphisms τi:OX(−D)∣Ui→OX(D)∨∣Ui sending a fi to a λi, where λi(fi−1)=1, and these τi agree on overlaps and glue by [F5] to an isomorphism τ:OX(−D)→OX(D)∨. Indeed, on Ui∩Uj write fi=ufj with u=fi/fj a unit; then the dual bases satisfy λi=uλj, because λi(fi−1)=λi(u−1fj−1)=1 forces λi=uλj. Hence afi=aufj↦auλj=aλi, so τi and τj agree on the overlap, and τ is an isomorphism because each τi carries the basis fi of OX(−D)∣Ui to the basis λi of OX(D)∨∣Ui.

3.1F4step 2.1step 2.2∎

Conclusion. There are canonical isomorphisms OX(D+E)≅OX(D)⊗OX(E) and OX(−D)≅OX(D)∨; the first is characterised by (figi)−1↦fi−1⊗gi−1 and the second by fi↦λi with λi(fi−1)=1. The second is the canonical inverse of OX(D) described by [F4]: combining it with the first for the pair (D,−D) gives OX(−D)⊗OX(D)≅OX(0)=OX, the evaluation pairing. The construction uses only the given equations; no trivialisations are chosen and no choice principle is used.

On the empty scheme all three sheaves are the zero module sheaf, which is the unique O∅-module, and both canonical isomorphisms are the identity of that module. For E=0, whose equations are gi=1, the addition isomorphism reads OX(D)≅OX(D)⊗OX, the canonical unit isomorphism; for D=0 it reads OX(E)≅OX⊗OX(E). Taking E=−D gives OX(D)⊗OX(−D)≅OX, recovering the evaluation isomorphism of [F4] from the divisor side. If the divisors are represented on two different covers, one first passes to a common refinement, which changes neither the divisors nor their associated sheaves by [F1] and [F2].

Depends on

Used by

Dependency tree · two levels

19 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