Alphabeta Math
TheoremStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedprecheck passjudge pass (gpt-6.1-sol)audited 2026-10-08
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.

The Riemann-Roch theorem on a compact Riemann surface

Statement

Assume the full Axiom of Choice (The Axiom of Choice). Let X be a compact Riemann surface of topological genus g, let D be a divisor on X, and put ED:=OX(D) and FD:=KX⊗ED∗, where KX:=Λ1,0T∗X is the canonical holomorphic line bundle (Divisors, principal divisors and canonical divisors on a Riemann surface, The holomorphic line bundle associated to a divisor, Serre duality on a compact Riemann surface). Write hq(X,G):=dim⁡CHq(X,G) when finite, ℓ(D):=dim⁡H0(X,ED), i(D):=dim⁡H1(X,ED), and χ(ED):=ℓ(D)−i(D) (Finite-dimensionality of the cohomology of a divisor on a compact Riemann surface). Then:

  1. The spaces H0(X,ED), H1(X,ED), and H0(X,FD) are finite-dimensional, and i(D)=h0(X,FD) (Finite-dimensionality of the cohomology of a divisor on a compact Riemann surface, Serre duality on a compact Riemann surface).

  2. The intrinsic Riemann–Roch formula is ℓ(D)−h0(X,FD)=χ(ED)=deg⁡D+1−g.

  3. There exists a nonzero meromorphic differential on X. For any such differential η, let Kη:=(η) be its canonical divisor. The isomorphism OX(Kη)≅KX identifies H0(X,FD) with L(Kη−D), so i(D)=ℓ(Kη−D) and the formula becomes ℓ(D)−ℓ(Kη−D)=deg⁡D+1−g. This expression is independent of the differential, and the identity is invariant under replacing D by a linearly equivalent divisor. For D=0, it gives ℓ(0)=1, i(0)=g, and χ(OX)=1−g.

Facts & Assumptions

Given: Full AC, a compact Riemann surface X of topological genus g, and a divisor D.

[F1]

Full AC implies countable choice by restricting a choice function to any countable family of nonempty sets (The Axiom of Choice, The Axiom of Countable Choice (ACω)).

[F2]

A divisor on compact X has finite support and degree the sum of its finitely many coefficients. Every principal divisor has degree zero, and any two canonical divisors are linearly equivalent (Divisors, principal divisors and canonical divisors on a Riemann surface).

[F3]

For a canonical divisor Kη=(η), the divisor bundle satisfies OX(Kη)≅KX, and tensoring with OX(−D)≅OX(D)∗ identifies OX(Kη−D) with FD (The holomorphic line bundle associated to a divisor).

[F4]

Under countable choice, a compact Riemann surface admits a compatible Riemannian metric (Hermitian metric and L2 pairing on a compact Riemann surface).

[F5]

Under countable choice, every holomorphic line bundle admits a Hermitian metric (Hermitian metric and L2 pairing on a compact Riemann surface).

[F6]

With compatible metrics supplied, the groups H0(X,OX(D)) and H1(X,OX(D)) are finite-dimensional (Finite-dimensionality of the cohomology of a divisor on a compact Riemann surface).

[F7]

The notation is ℓ(D)=dim⁡H0(X,OX(D))=dim⁡L(D), i(D)=dim⁡H1(X,OX(D)), and χ(OX(D))=ℓ(D)−i(D) (Finite-dimensionality of the cohomology of a divisor on a compact Riemann surface).

[F8]

For every divisor A and point p, χ(OX(A+[p]))=χ(OX(A))+1 (The point-divisor exact sequence and the Euler-characteristic step).

[F9]

The adjacent cohomology spaces in that point-divisor sequence are finite-dimensional (The point-divisor exact sequence and the Euler-characteristic step).

[F10]

For topological genus g, ℓ(0)=1, i(0)=h1(X,OX)=g, and χ(OX)=1−g (The Euler characteristic of the structure sheaf is one minus the genus).

[F11]

For ED=OX(D), Serre duality identifies H1(X,ED) with the complex-linear dual of H0(X,FD), and both spaces are finite-dimensional with equal dimensions (Serre duality on a compact Riemann surface).

[F12]
[F13]

Any two canonical divisors are linearly equivalent (Divisors, principal divisors and canonical divisors on a Riemann surface).

[F14]

If D′∼D, multiplication by the inverse of a meromorphic function with divisor D′−D gives the line-bundle isomorphism OX(D)≅OX(D′) (The holomorphic line bundle associated to a divisor).

[F15]

The zero-divisor bundle is canonically trivial, OX(0)≅X×C (The holomorphic line bundle associated to a divisor).

[F16]

For every divisor A, the global holomorphic sections of OX(A) identify with L(A) (The holomorphic line bundle associated to a divisor).

[F17]

A Riemann surface is nonempty and connected and has holomorphic coordinate charts (Riemann surfaces and holomorphic atlases).

[F18]

If X is compact and deg⁡A<0, then L(A)=0 (Divisors, principal divisors and canonical divisors on a Riemann surface).

[F19]

The divisor bundle OX(A) has a canonical meromorphic section sA=fiei with (sA)=A, where ej=(fi/fj)ei; also OX(−A)≅OX(A)∗ (The holomorphic line bundle associated to a divisor).

[F20]

Holomorphic sections have holomorphic coefficients in holomorphic frames, and meromorphic sections of KX are exactly meromorphic differentials. A nonzero meromorphic differential has no identically zero germ on connected X (Holomorphic line bundles and meromorphic sections on a Riemann surface, Meromorphic differentials, orders and residues).

Proof

The proof computes the Euler characteristic by finite point-divisor increments, then applies Serre duality to the canonical line bundle. Applying the intrinsic formula to a sufficiently negative point divisor produces a nonzero meromorphic differential, and hence the unconditional divisor form.

1.1F1F4F5F6F7given

Full AC gives countable choice by [F1]. Choose compatible metrics on X and ED using [F4, F5], solely to invoke the finiteness result [F6]; the formula below does not depend on these auxiliary metrics. Thus ℓ(D) and i(D) are finite and χ(ED)=ℓ(D)−i(D) by [F7].

2.1F1F2F4F5F6F8F9F10step 1.1algebra

Write D=∑p∈Snp[p], where S is finite by [F2]. The finite sequence from 0 to D uses only finitely many intermediate divisors. By [F1], [F4], and [F5], choose compatible metrics on X and each associated line bundle; [F6] then makes every intermediate Euler characteristic finite and defined. Starting at the zero divisor, apply [F8] np times to add [p] when np>0. When np<0, apply [F8] to A−[p] to obtain χ(OX(A−[p]))=χ(OX(A))−1, and repeat −np times; [F9] supplies finiteness for each adjacent pair. This finite sequence reaches D and changes χ by ∑p∈Snp=deg⁡D. By [F10], its initial value is χ(OX)=1−g, so χ(ED)=1−g+deg⁡D.

3.1F7F11step 1.1step 2.1algebra

Set FD=KX⊗ED∗. By [F11], i(D)=h0(X,FD). Substituting this equality into the definition of χ(ED) and using step 2.1 gives ℓ(D)−h0(X,FD)=1−g+deg⁡D. Finiteness of H0(X,FD) follows from the same perfect duality and finiteness of H1(X,ED).

4.1F2F17F18F19F20step 3.1chooseconstructalgebra

By [F17], choose p∈X and put n=g+2>0 and A=−n[p]. Since deg⁡A=−n<0, [F18] gives ℓ(A)=0. Applying step 3.1 to A and using the dual-bundle identity in [F19] gives h0(X,KX⊗OX(n[p]))=n+g−1=2g+1>0. Choose a nonzero holomorphic section σ of this bundle. In local coordinates and holomorphic divisor-bundle frames, write σ=ai dzi⊗ei and sn[p]=fiei as in [F19], and define η∣Ui=(ai/fi) dzi. Each quotient is meromorphic. If ej=gijei and dzj=kijdzi on an overlap, then ai=kijgijaj and fi=gijfj, so ai/fi=kij(aj/fj); the differentials therefore glue by [F20]. Near p, the local equation fi has order n, so η has pole order at most n; away from p, fi is a holomorphic unit, so η is holomorphic. Since σ≠0, division by sn[p] gives η≠0. By [F20], its germs are not identically zero, and [F2] gives the finite-support canonical divisor (η).

5.1F3F12F13F14F16step 3.1step 4.1algebra

A nonzero meromorphic differential exists by step 4.1. For any such differential η, [F3] identifies FD≅OX(Kη−D), and [F16] identifies its global sections with L(Kη−D). Together with step 3.1, this gives i(D)=ℓ(Kη−D) and the divisor form. If η′ is another nonzero meromorphic differential, [F13] gives Kη′∼Kη, so [F3] yields the same dimension. If D′∼D, then [F14] identifies OX(D′) with OX(D) and their dual twists with the same KX; [F12] gives deg⁡D′=deg⁡D. Hence both sides of the formula are invariant under this replacement.

6.1F10F15step 2.1algebra∎

For D=0, [F15] identifies E0 with OX. By [F10], ℓ(0)=1 and i(0)=g, so the formula reads 1−g=1−g.

Depends on

Used by

Dependency tree · two levels

117 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