Alphabeta Math
LemmaStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedprecheck passaudited 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.

The adele quotient V_X/(K + A_X) computes H^1 of the structure sheaf

Statement

Assume the Axiom of Choice as inherited from the coherent-cohomology suppliers. Let X be a smooth curve over the field k that is proper over k; thus X is a geometrically integral, connected, separated k-scheme of finite type and dimension one. Write K=k(X) for its function field and let p range over the closed points of X. For each closed point let Ap:=OX,p^ be the completion of the local ring OX,p, with fraction field Kp, and put VX:={(fp)∈∏pKp  :  fp∈Ap for all but finitely many p},AX:=∏pAp, embedding K diagonally into VX. Then:

  1. K is a k-subspace of VX, and VX is a K-module under componentwise multiplication;
  2. for every f∈K the quotient (fAX+AX)/AX is a finite-dimensional k-vector space; in the notation B<C for "(B+C)/C is finite-dimensional over k", this says fAX<AX for every f∈K, so K acts on VX through endomorphisms θ with θAX<AX; moreover, for all f,g∈K and every finite set S of closed points outside which f, g and g−1 are all integral (the condition on g−1 being void when g=0), the tail subspace AT:=∏p∈TAp over T:=X∖S satisfies fAT+fgAT+fg−1AT⊆AT;
  3. K∩AX is the space of global sections H0(X,OX);
  4. the cokernel VX/(K+AX) is canonically isomorphic, as a k-vector space, to H1(X,OX), and is finite-dimensional;
  5. if X is geometrically integral (as every smooth curve over k is) then K∩AX=k.

The finite-support condition in the definition of VX is exactly what makes the diagonal image of K lie in VX (part 1); parts 2, 3 and 4 are the hypotheses and the conclusion of the adelic presentation of H1 used for the residue theorem.

Facts & Assumptions

Given: a field k, a smooth curve X over k that is proper over k, its function field K=k(X), its closed points p, the completions Ap=OX,p^ with fraction fields Kp, and the Axiom of Choice.

[F1]

X is nonempty, geometrically integral, separated and of finite type over k, with chain dimension one; it has a generic point η with OX,η=K=k(X), and for every nonempty affine open U=Spec⁡A⊆X the ring A is a Noetherian domain with fraction field K and OX,p=Ap for the closed points p∈U; consequently X is quasi-compact and locally Noetherian, hence its underlying space is Noetherian and every open subset of X is quasi-compact (Curves over a field, The function field of an irreducible classical affine variety, Function field of an integral finite-type scheme, The field of fractions Frac⁡(D)=(D∖{0})−1D of an integral domain, Zero divisor, and integral domain: a commutative ring with 1≠0 and no zero divisors, Frac⁡(D) is a field and d↦d/1 embeds the integral domain D, The stalk of the affine structure sheaf at a prime is A_p).

[F2]

For a closed point p the local ring OX,p is a discrete valuation ring with fraction field K and maximal ideal generated by a uniformizer tp; every nonzero f∈K has a well-defined order ord⁡p(f)∈Z, and f∈OX,p exactly when ord⁡p(f)≥0 (Local rings at closed points of smooth curves are discrete valuation rings, Order codimension one rational function, Localisation at a prime ideal: Rp=(R∖p)−1R).

[F3]

For f∈K× the set of closed points with ord⁡p(f)<0 is finite: on an affine chart U=Spec⁡A write f=a/b with a,b∈A, b≠0, so that every pole of f in U is a maximal ideal of A containing b; since A is a one-dimensional Noetherian domain, A/(b) is zero-dimensional and Noetherian, hence has only finitely many maximal ideals, and a finite affine cover of X bounds all poles (Curves over a field, The closed points of the prime spectrum are exactly the maximal ideals, In a nonzero commutative ring, every proper ideal is contained in a maximal ideal, Localisation at a prime ideal: Rp=(R∖p)−1R).

[F4]

Ap is a local ring whose maximal ideal is generated by the image of tp; the completion map OX,p→Ap is injective, Kp=Frac⁡(Ap) contains K=Frac⁡(OX,p) as a subfield with K∩Ap=OX,p, the equality Kp=Ap[tp−1] holds, and fAp=tpord⁡p(f)Ap for f∈K×; for f=0 one has fAp=0 (The I-adic completion of a module, Local rings at closed points of smooth curves are discrete valuation rings).

[F5]

The residue field κ(p)=OX,p/mp of a closed point p of a finite-type k-scheme is a finite extension of k, so dim⁡k(Ap/tpmAp)=m [κ(p):k] for every m≥0; in particular tp−mAp/Ap is a finite-dimensional k-vector space for every m (Curves over a field, Local rings at closed points of smooth curves are discrete valuation rings, The I-adic completion of a module).

[F6]

Sheaf cohomology Hq(X,−) is the right derived functor of global sections on sheaves of abelian groups, a short exact sequence of sheaves yields a natural long exact sequence of cohomology groups, and for a nonempty affine open U=Spec⁡A one has A=Γ(U,OX) (Sheaf cohomology as right derived global sections, A sheaf on a topological space, The derived long exact sequence, Quasi-coherent module on a scheme).

[F7]

On the irreducible space X the constant sheaf with value a k-vector space is flasque, a direct sum of skyscraper sheaves at closed points is flasque, every flasque sheaf on X has vanishing cohomology in positive degrees, and the global sections of a direct sum of skyscraper sheaves at the closed points are the direct sum of their values (Constant sheaves on irreducible spaces are flasque and acyclic, The constant sheaf is the sheaf of locally constant functions, Flasque abelian sheaves are Γ-acyclic).

[F8]

For every coherent OX-module F and every q≥0 the k-vector space Hq(X,F) is finite-dimensional, because X is proper over k; in particular H0(X,OX) and H1(X,OX) are finite-dimensional, H0(X,OX) is a k-subalgebra and a domain, and if X is geometrically integral then for an algebraic closure kˉ/k the ring K⊗kkˉ is a domain (Finite-dimensional coherent cohomology over a field, Function field of an integral finite-type scheme, Curves over a field).

[F9]

The Axiom of Choice is The Axiom of Choice.

Proof

technique · direct
1.1F1F2F3F4F9given

Set up the two ambient objects. By [F1] the generic point η has function field K=k(X), and by [F2] and [F4] each closed point p carries the valuation vp=ord⁡p whose valuation ring OX,p lies in K, with K⊆Kp. By [F3] every f∈K lies in OX,p⊆Ap for all but finitely many p, so the diagonal map K→∏pKp lands in VX; it is injective because each map K→Kp is an inclusion of fields by [F4]. If f∈K and (gp)∈VX, then gp∈Ap for all but finitely many p and f∈Ap for all but finitely many p by [F3], so fgp∈Ap for all but finitely many p; hence componentwise multiplication makes VX a K-module containing K as a k-subspace. This is part 1; the Axiom of Choice [F9] enters only through the cited suppliers.

1.2F2F3F4F5

Compute the quotient attached to a single f∈K. If f=0, then (fAX+AX)/AX=0. Now suppose f≠0 and let S={p:ord⁡p(f)<0}, finite by [F3], and put np=−ord⁡p(f)≥0 for p∈S. Since fAp=tp−npAp by [F4], inside Kp one has fAp+Ap=tp−npAp, and (fAp+Ap)/Ap=tp−npAp/Ap is a k-vector space of dimension np[κ(p):k] by [F5] for p∈S, while fAp⊆Ap by [F4] for p∉S. Hence fAX+AX⊆(∏p∈Stp−npAp)×∏p∉SAp, and the quotient (fAX+AX)/AX embeds into ⨁p∈Stp−npAp/Ap, a finite-dimensional k-vector space. Therefore fAX<AX for every f∈K, and for each such f the multiplication-by-f endomorphism θf of VX satisfies θf(AX)=fAX<AX. This is the first assertion of part 2.

1.3F2F3F4

Verify the tail condition. Let f,g∈K and let S be a finite set of closed points outside which f, g and, when g≠0, g−1 are all integral; put T=X∖S. If f=0, all three tail images are zero, interpreting the g−1 term as zero when g=0. If f≠0 and g=0, only fAT remains, and it lies in AT because f is integral at every p∈T. If f,g≠0, integrality gives f,g,g−1∈Ap at each p∈T, so their products give fAp+fgAp+fg−1Ap⊆Ap. Taking products proves the tail inclusion. Such finite sets S exist by [F3] applied only to the nonzero functions among f,g,g−1. This proves the second assertion of part 2 without taking the order of zero.

1.4F1F3F4F6

Identify K∩AX with the global sections. Let f∈K. By [F4], f∈Ap if and only if f∈OX,p, so f∈K∩AX exactly when f is regular at every closed point of X. Choose a finite affine cover of X by opens U=Spec⁡A (possible since X is quasi-compact and of finite type over k by [F1]) and let f∈K=Frac⁡(A). If f∉A, the ideal of denominators I={a∈A:af∈A} is a nonzero proper ideal of the Noetherian domain A, so it is contained in a maximal ideal m of A by [F3]; the closed points of U are exactly the maximal ideals of A [F3], with OX,p=Am for the corresponding closed point p by [F1] and [F6], and f∉Am, because an expression f=b/s with b∈A and s∈A∖m would give sf=b∈A, hence s∈I⊆m, a contradiction. Passing to the contrapositive, f regular at every closed point of U forces f∈A; applying this on the finitely many charts shows that f∈K∩AX if and only if f∈Γ(X,OX)=H0(X,OX). This is part 3.

1.5F1F3F4F7

Build the two auxiliary sheaves. Let K0 be the constant sheaf with value K on X, and let K1 be the sheaf U↦⨁p∈UKp/Ap, the direct sum over the closed points of the skyscraper sheaves with values Kp/Ap; this is a sheaf because every open subset of X is quasi-compact by [F1], so a compatible family over a cover is already determined by its members over a finite subcover, and it is flasque because a section over an open U extends by zero to X by [F7]. Define the map of sheaves δ:K0→K1 on sections over a nonempty open U by sending f∈K=K0(U) to the class (f+Ap)p∈U; this is well defined because f∈Ap for all but finitely many p by [F3], so the family has finite support. Then ker⁡δ has sections over a nonempty open U equal to {f∈K:f∈Ap for every closed point p∈U}, and the cokernel of δ on global sections is VX/(K+AX).

2.1F1F2F3F4step 1.4step 1.5

Prove that 0→OX→K0→δK1→0 is exact. The map OX→K0 sends a regular function to itself viewed in K; it is injective on every nonempty open because Γ(U,OX)⊆K, and its image is killed by δ. By construction ker⁡δ has sections over a nonempty open U equal to {f∈K:f∈Ap for all closed points p∈U}, which by the chart computation of step 1.4 (applied to the members of a finite affine cover of the quasi-compact open U) is exactly Γ(U,OX); hence ker⁡δ=OX. For the surjectivity of δ it suffices to check stalks: at a closed point p the stalk of K0 is K and the stalk of K1 is Kp/Ap, and the map K→Kp/Ap is surjective because K+Ap=Kp: since Kp=Ap[tp−1] by [F4], every x∈Kp equals ctp−m with c∈Ap and m≥0, and the image of the completion map OX,p→Ap being dense supplies b∈OX,p with b≡c mod tpmAp, so that btp−m∈K and x−btp−m∈Ap. At the generic point η the stalk of K1 is 0 because the class of a local section of the direct sum of skyscrapers is killed after removing its finitely many support points, so δ is surjective there as well. Thus the sequence of sheaves is exact.

2.2F1F7step 1.5

Compute the cohomology of K0 and K1. Since K0 is the constant sheaf with value K on the irreducible space X [F1], [F7] gives H0(X,K0)=K and Hq(X,K0)=0 for every q>0. Since K1 is flasque, [F7] gives Hq(X,K1)=0 for every q>0, and its global sections are H0(X,K1)=⨁pKp/Ap=VX/AX, the direct sum over all closed points, again by [F7].

3.1F6step 1.4step 2.1step 2.2

Apply the long exact sequence. The short exact sequence of step 2.1 yields, by [F6], the exact sequence 0→H0(X,OX)→K→VX/AX→H1(X,OX)→0, where the middle map K→VX/AX sends f to (f+Ap)p and its cokernel is VX/(K+AX); here the cohomology of K0 and K1 is that computed in step 2.2. Hence H1(X,OX)≅VX/(K+AX) canonically as k-vector spaces, and the injective map H0(X,OX)→K identifies H0(X,OX) with the kernel of K→VX/AX, which is K∩AX in agreement with step 1.4. This is part 4, except for the finiteness of VX/(K+AX).

4.1F1F8step 1.4step 3.1∎

Finiteness and the constant-field case. Since X is proper over k and OX is coherent, [F8] gives dim⁡kH1(X,OX)<∞, so step 3.1 makes VX/(K+AX) finite-dimensional, which completes part 4. For part 5 let L:=K∩AX=H0(X,OX) by step 1.4; by [F8] the ring L is a finite-dimensional k-algebra and a domain, and L≠0 because it contains the unit of K. If X is geometrically integral and kˉ/k is an algebraic closure, then K⊗kkˉ is a domain by [F8], and the inclusion L↪K remains injective after tensoring with the flat k-module kˉ, so L⊗kkˉ is a domain of dimension [L:k] over the algebraically closed field kˉ. A finite-dimensional commutative domain over a field is a field, since a nonzero element has injective and hence bijective multiplication; thus L⊗kkˉ is a field of degree [L:k] over kˉ, forcing [L:k]=1 and L=k, which is part 5.

Depends on

Used by

Dependency tree · two levels

125 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