Alphabeta Math
Pipeline-generated
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.

✓ 10 results · all verified · 3 also independently AI-judged
Every result on this page is machine-checked by a proof checker and read in full and owner-audited; the judge is an additional, independent cross-model AI review of the proofs. The 7 not AI-judged were verified by owner audit (typically over a confirmed judge false positive), not failures.

Riemann Roch for Curves via Euler Characteristics — Examples

1 · Prerequisites

2 · Summary

These examples and counterexamples exercise the Euler-characteristic form of Riemann–Roch on explicit curves and divisors, and record the failure modes that the page's deliberately one-sided statements leave open.

On the projective line the theorem is an identity between computed numbers. For D=d[∞] the divisor spaces are L(d[∞]) with dimensions ℓ(D)=max⁡(d+1,0) and i(D)=max⁡(−d−1,0), so ℓ(D)−i(D)=d+1=deg⁡k(D)+1−g for every integer d, including the negative range where the left-hand side is the negative of the index of speciality. A negative right-hand side is therefore not a contradiction: for D=−d[∞] with d≥2 the identity reads 0−(d−1)=−d+1, and the negative lower bound cannot count or ensure nonzero sections. At d=1 both cohomology dimensions are zero: the equality and the count of zero independent sections are consistent, while the bound ℓ(D)≥0 still ensures no nonzero section. On the same curve, principal divisors of degree zero are computed from rational functions: the divisor of (t−a)/(t−b) is [a]−[b], its class is trivial by the classification of divisors on P1, and ℓ=1, i=0 on both sides of Riemann–Roch.

The genus-zero boundary cases separate the geometric and arithmetic behaviour. A smooth plane conic with a rational point has arithmetic genus zero and a divisor of degree one, so it is isomorphic to the projective line and the projection from the rational point computes its divisor spaces; the conic x2+y2+z2=0 over R has the same arithmetic genus and no rational point, so it carries no divisor of degree one and is not isomorphic to PR1, although it becomes a projective line after base change to C. The empty divisor exhibits the normalization at the heart of the definitions: ℓ(0)=h0(OC)=1 and i(0)=h1(OC)=g, so χ(OC)=1−g, the system ∣0∣ is a single point in every genus, and the zero divisor is nonspecial exactly in genus zero.

The remaining examples test the point-addition and positivity statements. The jump satisfies 0≤ℓ(D+p)−ℓ(D)≤[κ(p):k]; both endpoints occur on the projective line, and intermediate values can occur: over R, the point p=V(t2+1) has residue degree two, while D=−2[∞] gives a jump of one; the zero divisor of a curve of positive genus, for instance a smooth plane quartic, shows that the Riemann inequality is strict for special divisors by exactly i(D); a pencil spanned by 1 and a function with poles only at one point realizes the finite morphism to the projective line constructed on the main page, with the fibre over infinity equal to the pole divisor; and a sufficiently positive divisor in a fixed ample direction is nonspecial, so Riemann–Roch counts its sections exactly. The example keeps the fixed-direction threshold of the vanishing theorem and does not assert the universal bound above degree 2g−2, which requires the duality pair that follows.

3 · Logical flowchart

4 · Definitions, theorems and proofs

None yet.

5 · Examples, counterexamples and false statements

ExampleConstruction: Literature-sourcedVerification: AI-adaptedprecheck passaudited 2026-10-02Open item page →

Riemann-Roch on the projective line for every degree

Example

Assume the Axiom of Choice inherited from the current projective-line cohomology and divisor/Picard suppliers.

Let k be a field, let Pk1 have coordinate t with point at infinity ∞, and let D=d[∞] for an integer d. Since OPk1(d) is isomorphic to OPk1(d[∞]) — the class of the point at infinity generates Pic⁡(Pk1)=Z with O(1) corresponding to 1 (The Picard group of the projective line) — the explicit cohomology of the twists gives l(D)=h0(O(d))=max⁡(d+1,0),i(D)=h1(O(d))=max⁡(−d−1,0), the values being read off from Global sections of projective twists and Top cohomology of projective twists. Hence l(D)−i(D)=d+1=deg⁡k(D)+1−g with g=0 for every integer d: on the projective line Riemann-Roch is an identity between explicit numbers, including the negative-degree range where l(D)=0 and i(D)=−d−1 compensates. The divisors of degree at least −1 are exactly the nonspecial ones here, and the complete linear system ∣D∣ of Complete linear system is nonempty exactly for d≥0.

Scaffold repair, recorded for the owner. The frozen scaffold statement cited the examples-page item ex-cohomology-o-d-projective-line-all-d for the cohomology of the twists. That item is homed on the examples page cohomology-of-quasi-coherent-sheaves-on-affine-and-projective-schemes-examples, and an examples-page item may not depend on another examples-page item, so the citation is replaced here by the published A-page corollaries Global sections of projective twists and Top cohomology of projective twists, which contain the same values for n=1; every promised claim is preserved. The example's negative-degree compensation clause is also corrected to l(D)=0 and i(D)=−d−1 with l−i=d+1, so that the displayed identity is true at every d.

The current Divisors on the projective line are classified by degree and The Picard group of the projective line give the divisor-to-twist identification used in the calculation. The dimension notation is supplied by The Riemann-Roch dimension l(D), The index of speciality i(D) and Special and nonspecial divisors, and the complete linear system by Complete linear system.

Facts & Assumptions

Given: the Axiom of Choice inherited from the current projective-line cohomology and divisor/Picard suppliers; a field k, the projective line Pk1 with coordinate t and point at infinity ∞, and the divisor D=d[∞] for an integer d.

[F1]

Projective-line data: Pk1 is a smooth proper geometrically integral curve over k of genus 0; the coordinate section x0 of O(1) vanishes exactly at infinity with multiplicity one, so div⁡(x0)=[∞] and O(1)≅O(∞) with deg⁡kO(1)=1; every divisor D′ is linearly equivalent to deg⁡k(D′)[∞], and the degree homomorphism on divisor classes is an isomorphism, so linearly equivalent divisors have equal degree and [κ(∞):k]=1 (Divisors on the projective line are classified by degree, The Picard group of the projective line).

[F2]

Global sections: for n=1 and A=k one has H0(Pk1,O(d))≅k[x0,x1]d for d≥0 and H0(Pk1,O(d))=0 for d<0; the monomials x0ax1d−a for 0≤a≤d form a k-basis of k[x0,x1]d, so h0(O(d))=d+1 for d≥0 and h0(O(d))=0 for d<0 (Global sections of projective twists).

[F3]

Top cohomology: for n=1 the group H1(Pk1,O(d)) vanishes for d>−2 and for d≤−2 it is free of rank (−d−11)=−d−1 over k≠0; so h1(O(d))=0 for d≥−1 and h1(O(d))=−d−1 for d≤−2 (Top cohomology of projective twists).

[F4]

Degree: for a divisor D=∑xnx[x] the k-degree is deg⁡k(D)=∑xnx[κ(x):k], sum over the finite support, and it is a group homomorphism on the divisor group (Degree divisor proper curve, Divisors on a smooth proper curve).

[F5]

Riemann-Roch dimensions: l(D)=dim⁡kL(D)=h0(D)=dim⁡kH0(C,OC(D)) and i(D)=h1(D)=dim⁡kH1(C,OC(D)), so l(D)=h0(O(D)) and i(D)=h1(O(D)) for the attached sheaf (The Riemann-Roch dimension l(D), The index of speciality i(D)).

[F6]

Riemann-Roch: l(D)−i(D)=deg⁡k(D)+1−g with i(D)≥0, and l(D)=deg⁡k(D)+1−g exactly for the nonspecial divisors (Riemann-Roch as l minus i, Special and nonspecial divisors).

[F7]

Nonspeciality: D is nonspecial exactly when i(D)=0, and special exactly when i(D)≥1; equivalently nonspeciality is the equality case l(D)=deg⁡k(D)+1−g (Special and nonspecial divisors).

[F8]

The complete linear system ∣D∣={ D′ effective:D′ is linearly equivalent to D } is in bijection with the set of k-lines in L(D) and is empty exactly when L(D)=0 (Complete linear system).

[F10]

The Axiom of Choice is available and is inherited only through the suppliers named above; the computations below evaluate explicit formulas and select nothing (The Axiom of Choice).

Verification

technique · direct computation of $l(D)$ and $i(D)$ from the explicit cohomology of the twists, followed by a case check on the sign of $d$ for the Riemann-Roch identity, the nonspecialty threshold and the nonemptiness of the complete linear system
1.1F1F4F5F9

Degree and attached sheaf. By [F1] the residue degree of the point at infinity is [κ(∞):k]=1 and the degree homomorphism is defined on linear-equivalence classes, so for D=d[∞] the degree is deg⁡k(D)=d⋅[κ(∞):k]=d by [F4]. By [F1] the class of ∞ generates Pic⁡(Pk1)≅Z with O(1)↦1, so by the current divisor/Picard route [F9] one has O(D)=O(d[∞])≅O(1)⊗d≅O(d); hence l(D)=l(0+d[∞])=h0(O(d)) and i(D)=h1(O(d)) by [F5], and g=g(Pk1)=0 by [F1].

2.1F2F3step 1.1

The explicit values. By [F2] applied with n=1 and A=k, the space H0(Pk1,O(d)) has dimension d+1 for d≥0 and vanishes for d<0, because k[x0,x1]d has dimension d+1 as the space of homogeneous polynomials of degree d in two variables; hence h0(O(d))=max⁡(d+1,0). By [F3] applied with n=1, H1(Pk1,O(d))=0 for d>−2, that is for d≥−1, while for d≤−2 its rank is (−d−11)=−d−1; hence h1(O(d))=max⁡(−d−1,0). Combining with step 1.1, l(D)=max⁡(d+1,0),i(D)=max⁡(−d−1,0) for every integer d.

3.1F4F6step 1.1step 2.1

Riemann-Roch as an identity between explicit numbers. If d≥−1 then −d−1≤0, so i(D)=max⁡(−d−1,0)=0, while d+1≥0 gives l(D)=max⁡(d+1,0)=d+1; then l(D)−i(D)=d+1. If instead d≤−2 then d+1<0, so l(D)=max⁡(d+1,0)=0, while −d−1≥1 gives i(D)=−d−1; then l(D)−i(D)=0−(−d−1)=d+1. In both cases, by step 1.1, l(D)−i(D)=d+1=deg⁡k(D)+1−g, which is [F6] with g=0 and deg⁡k(D)=d; the two cases d≥−1 and d≤−2 exhaust Z and meet no other, and for d≤−2 the section space l(D)=0 is compensated by i(D)=−d−1>0, while at d=−1 both dimensions and the right-hand side are zero, so the right-hand side stays correct even where it is negative.

3.2F1F4F6F7F8step 2.1

Nonspecialty and the complete linear system. By [F7] the divisor D is nonspecial exactly when i(D)=max⁡(−d−1,0)=0, that is exactly when −d−1≤0, i.e. d≥−1: the divisors d[∞] of degree at least −1 are exactly the nonspecial ones, and by [F6] the same threshold is the equality case l(D)=d+1 of the Riemann inequality, while for d≤−2 one has l(D)=0>d+1 and D is special. For the complete linear system, if d≥0 then D=d[∞] is an effective divisor and D is linearly equivalent to itself, so D∈∣D∣ and ∣D∣ is nonempty; if d<0 and D′=∑xnx[x] is effective with D′ linearly equivalent to D, then deg⁡k(D′)=deg⁡k(D)=d<0 by the well-definedness of the degree homomorphism on classes [F1], while effectivity gives nx≥0 and hence deg⁡k(D′)=∑xnx[κ(x):k]≥0 by [F4] and [κ(x):k]≥1, a contradiction; so ∣D∣ is empty for d<0. This agrees with [F8], under which ∣D∣ is in bijection with the k-lines in L(D): by step 2.1, l(D)=max⁡(d+1,0) is at least 1 exactly for d≥0.

4.1F2F3F9F10step 1.1step 2.1step 3.1step 3.2∎

Assembly and choice accounting. Step 2.1 computes l(D)=max⁡(d+1,0) and i(D)=max⁡(−d−1,0); step 3.1 verifies l(D)−i(D)=d+1=deg⁡k(D)+1−g for every integer d by an exhaustive case check on d≥−1 versus d≤−2; step 3.2 identifies the nonspecial divisors with the degrees d≥−1 and shows ∣D∣ nonempty exactly for d≥0, in agreement with the section dimension. The sheaf identification O(D)≅O(d) and the identity l(D)=h0(O(D)) of step 1.1 use the current interfaces [F9]; the numerical values of steps 2.1 and 3.1 depend on them only through that identification, and the cohomology values themselves are the published A-page corollaries [F2] and [F3]. The Axiom of Choice enters only through the suppliers recorded in [F10]; every value above is computed from explicit formulas, and no family of objects is selected.

ExampleConstruction: Literature-sourcedVerification: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)audited 2026-10-02Open item page →

A smooth conic with a rational point is a projective line

Example

Assume the Axiom of Choice inherited from the current plane-genus, rational-point and cohomology suppliers.

Let k be a field of characteristic not two, let C=V+(F)⊆Pk2 be a smooth plane conic that is a curve (integral of dimension one — the hypothesis under which Arithmetic genus of a plane curve applies), and let p∈C be a k-rational point. Then:

  1. deg⁡k[p]=[κ(p):k]=1, so D=[p] is a divisor of degree one (Degree divisor proper curve);
  2. A genus-zero curve with a degree-one divisor is the projective line applies once g(C)=0: Arithmetic genus of a plane curve gives arithmetic genus pa(C)=(2−1)(2−2)2=0, which for a smooth curve is the genus, and a degree-one divisor is present, so C≅Pk1;
  3. under such an isomorphism the degree-one divisor [p] corresponds to a degree-one divisor [q] of Pk1, and the projective-line computation gives l([q])=h0(O(1))=2 and i([q])=0 (Global sections of projective twists); dimensions of cohomology are invariant under isomorphism, so l([p])=2,i([p])=0: Riemann-Roch on C reads 2−i([p])=1+1−0, forcing i([p])=0, and the two-dimensional space L([p]) is spanned by 1 and a coordinate function with a single simple pole at p, the coordinate of the isomorphism C≅Pk1 supplied by the rational-point theorem.

The classical form of this computation is the projection parametrisation: for each line ℓ through p the residual intersection C∩ℓ pairs the second point of ℓ∩C with p, giving the pencil ∣[p]∣ and the coordinate above; the reverse direction — that the quadratic Veronese image of Pk1 is such a conic — is the batch-6 examples-page item ex-quadratic-veronese-conic, which is not consumable here because examples-page items are leaves.

Scaffold repair, recorded for the owner. The frozen scaffold cited the examples-page items ex-quadratic-veronese-conic and ex-rational-parametrization-circle-conic. Both are leaves and cannot carry a load; the citations are replaced by the A-page rational-point theorem A genus-zero curve with a degree-one divisor is the projective line, the published A-page computation of h0(O(1))=2 Global sections of projective twists, and the local argument of items 1–3. Every promised numerical claim (deg⁡k[p]=1, pa=0, g=0, C≅Pk1, l([p])=2, i([p])=0, and the reading 2−i=1+1−0) is preserved. The explicit line-pencil description of ∣[p]∣ is recorded as the classical geometric picture rather than as a consumed claim, with its would-be supplier named above.

The current Arithmetic genus of a plane curve gives the arithmetic genus; smoothness identifies it with the curve genus. The current A genus-zero curve with a degree-one divisor is the projective line supplies the isomorphism, and the published projective-space cohomology result Global sections of projective twists supplies the section dimension. The proof transports cohomology through the isomorphism using the current cohomology and Riemann-Roch interfaces cited below.

Facts & Assumptions

Given: the Axiom of Choice inherited from the current plane-genus, rational-point and cohomology suppliers; a field k of characteristic not two, a smooth plane conic curve C=V+(F)⊆Pk2 with arithmetic genus computed by the plane-curve theorem, and a k-rational point p∈C.

[F1]

Divisors and degree: deg⁡k(D)=∑xnx[κ(x):k] is a group homomorphism, and a k-rational point has residue degree one, so deg⁡k[p]=1 (Degree divisor proper curve).

[F2]

Plane conic arithmetic genus: a curve X=V+(G) cut out by a nonzero homogeneous form of degree d≥1 has H0(X,OX)=k and pa(X)=1−χ(OX)=(d−1)(d−2)2; for d=2 this is 0, and for a smooth curve the arithmetic genus is the genus (Arithmetic genus of a plane curve, Curves over a field; the genus g(C)=1−χ(C,OC) is the one of Genus via the Euler characteristic, where the agreement with the arithmetic genus in the smooth case is recorded).

[F3]

Rational-point theorem: a smooth proper geometrically integral curve of genus 0 over k that admits a divisor of degree one is isomorphic to Pk1 (A genus-zero curve with a degree-one divisor is the projective line).

[F4]

Cohomology of the twists on the projective line: Pk1 is a smooth proper geometrically integral curve of genus 0; for a k-rational point q the degree-one divisor [q] is linearly equivalent to [∞], so O([q])≅O(1) with deg⁡kO(1)=1, while H0(Pk1,O(1))≅k[x0,x1]1 has dimension 2; hence l([q])=h0(O(1))=2 (Global sections of projective twists, Divisors on the projective line are classified by degree).

[F5]

Riemann-Roch and the index of speciality: l(D)−i(D)=deg⁡k(D)+1−g with i(D)=h1(D)≥0; i(D)=0 if and only if D is nonspecial, equivalently if and only if the identity l(D)=deg⁡k(D)+1−g is an equality (Riemann-Roch as l minus i, The index of speciality i(D), Special and nonspecial divisors, The Riemann-Roch dimension l(D)).

[F6]

Invariance under isomorphism: l(D)=dim⁡kH0(C,OC(D)) and i(D)=dim⁡kH1(C,OC(D)) are dimensions of cohomology groups of the attached invertible sheaf, so an isomorphism of curves carrying D to a divisor D′ carries OC(D) to OC′(D′) and preserves l and i (The Riemann-Roch dimension l(D), Sheaf cohomology as right derived global sections, The index of speciality i(D)).

[F7]

The Axiom of Choice is available and is inherited only through the rational-point and Riemann-Roch suppliers above; the computation evaluates the given conic, point and isomorphism and selects nothing beyond them (The Axiom of Choice).

Proof

technique · compute the degree of $[p]$, derive genus zero from the plane-conic arithmetic genus, apply the rational-point theorem, and transport $l$ and $i$ from the projective line to obtain $l([p])=2$, $i([p])=0$
1.1F1F2

Degree one and genus zero. By [F1] the divisor D=[p] has degree [κ(p):k]=1 because p is k-rational. By [F2] the plane conic has pa(C)=(2−1)(2−2)2=0, and since C is smooth, g(C)=pa(C)=0.

2.1F3step 1.1

The conic is a projective line. The curve C is smooth proper and geometrically integral by hypothesis and has genus 0 by step 1.1, and it carries the degree-one divisor [p] of step 1.1; [F3] therefore gives a k-isomorphism φ:C→Pk1.

3.1F4F5F6step 2.1

The two-dimensional space of the point. Under the isomorphism φ of step 2.1 the degree-one divisor [p] corresponds to a degree-one divisor [q] of Pk1, and by [F4] l([q])=2 with O([q])≅O(1) and g(Pk1)=0; Riemann-Roch [F5] on Pk1 at [q] therefore reads 2−i([q])=1+1−0, so i([q])=0. By the invariance [F6] of l and i under isomorphism, l([p])=2 and i([p])=i([q])=0. Equivalently, L([p]) contains the constants and a coordinate function with a single simple pole at p, so its dimension is at least two, while [F5] gives l([p])=2+i([p]) and the value l([p])=2 forces i([p])=0.

4.1F1F5F7step 1.1step 3.1∎

Riemann-Roch on the conic and conclusion. Riemann-Roch on C at the degree-one divisor reads 2−i([p])=1+1−0, which with i([p])=0 of step 3.1 is the identity 2=2; by [F5] the divisor [p] is nonspecial, and the equality case of the Riemann inequality holds at a rational point of a genus-zero conic. The classical projection parametrisation of C from p realizes the pencil ∣[p]∣ and the coordinate of the isomorphism; the example therefore exhibits explicitly the rational-point hypothesis of the genus-zero theorem in the conic case. The Axiom of Choice is inherited only through the suppliers of [F7]; nothing is selected beyond the given conic, point and isomorphism.

CounterexampleConstruction: Literature-sourcedVerification: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)audited 2026-10-02Open item page →

A genus-zero curve need not be the projective line

Counterexample

Assume the Axiom of Choice inherited from the current smoothness, properness, plane-genus and rational-point suppliers.

Let k=R and let C=V+(x2+y2+z2)⊆PR2 be the real projective conic. Then:

  1. C is a smooth proper geometrically integral curve over R. On each of the three standard affine charts, after permuting coordinates, the equation is 1+u2+v2=0. In its chart ring the Jacobian entries 2u,2v generate the unit ideal, since 1=(−u/2)(2u)+(−v/2)(2v); each prime therefore has a neighborhood on which one Jacobian minor is invertible, giving a standard smooth presentation. Thus all three charts are smooth over R. The conic is a closed subscheme of the proper R-scheme PR2, hence proper. Over C the form F is irreducible: a factorisation x2+y2+z2=L1L2 into linear forms would, after setting z=0, make x2+y2=(x+iy)(x−iy) the product of the restrictions of L1,L2, so by unique factorisation in C[x,y] the restrictions are units times x+iy and x−iy. After rescaling and, if necessary, interchanging the factors, write L1=x+iy+αz and L2=x−iy+γz. Expanding gives L1L2=x2+y2+(α+γ)xz+i(γ−α)yz+αγz2. Comparing with F gives α+γ=0, i(γ−α)=0, and αγ=1. The first two equations force γ=α and 2α=0, hence α=γ=0 in characteristic zero, contradicting αγ=1. Thus F is irreducible over C. The scheme CC=V+(F) is reduced and irreducible; it is nonempty since it contains [i:0:1]. The field C is an algebraic closure of R, so this is the geometric fibre used to establish geometric integrality over R; since C is algebraically closed, the same scheme is geometrically integral over C. Its chain dimension, and that of C over R, are one by the direct affine-chart calculation in [F4].
  2. g(C)=0. By Arithmetic genus of a plane curve applied to the curve C cut out by the degree-two form, the arithmetic genus is pa(C)=(2−1)(2−2)2=0, and for a smooth curve the arithmetic genus is the genus (Genus via the Euler characteristic), so g(C)=0.
  3. C has no R-rational point. If x,y,z∈R with x2+y2+z2=0, then x=y=z=0; since [0:0:0] is not a point of PR2, the conic has no R-point, C(R)=∅.
  4. The hypothesis of A genus-zero curve with a degree-one divisor is the projective line is not satisfied. For every closed point p∈C, its residue field is a finite extension of R; it is separable and simple, so an irreducible real minimal polynomial for a generator has degree 1 or 2 (A maximal ideal of an affine algebra has finite residue field over the base field, Fields of characteristic zero, finite fields, and algebraically closed fields are perfect, Every algebraic extension of a perfect field is separable, A finite extension generated by elements all but possibly one of which are separable is simple, The evaluation kernel and the unique monic irreducible minimal polynomial of an algebraic element, A simple algebraic extension is its minimal-polynomial quotient and has power basis 1,a,…,an−1 and degree n, An irreducible polynomial in R[x] has degree 1 or 2). Degree 1 would make κ(p)=R and give an R-point by Field-valued points and local-ring points, contradicting item 3. Thus every closed point has degree 2. Every divisor, including a signed divisor D=∑pnp[p] with arbitrary integers np, has degree deg⁡RD=2∑pnp, which is even (Degree divisor proper curve). In particular, no divisor has degree one, so the theorem's degree-one-divisor hypothesis fails.
  5. Geometrically the conic is a projective line. Over C the point [i:0:1] satisfies i2+02+1=0, so CC has a C-rational point; CC is a smooth proper geometrically integral genus-zero curve over C by items 1 and 2 applied over C, and A genus-zero curve with a degree-one divisor is the projective line therefore gives CC≅PC1. Thus the same curve becomes a projective line after base change to C.
  6. C≇PR1. The line PR1 has the R-rational point [1:0], and an R-isomorphism would induce a bijection on R-rational points; since C(R)=∅ by item 3, no such isomorphism exists.

Over a non-algebraically-closed field, genus zero therefore does not determine the curve: the real conic is a projective line geometrically and a form of P1 with no rational point arithmetically.

Scaffold repair, recorded for the owner. The frozen scaffold cited the examples-page items ex-base-change-real-conic-to-complex and ex-projective-conic-standard-charts; examples-page items are leaves and cannot carry a load, and both uses are replaced here by the explicit computations in items 1, 3, 5 and 6 (F and its partials, the sign of a sum of squares over R, the evaluation at [i:0:1], and the transport of rational points along an isomorphism).

The current Arithmetic genus of a plane curve supplies the genus computation, and the current A genus-zero curve with a degree-one divisor is the projective line is used only to state the missing rational-point hypothesis. The explicit smoothness, geometric-integrality, residue-degree, and real-point arguments below establish the counterexample directly.

Facts & Assumptions

Given: the Axiom of Choice inherited from the current smoothness, properness, plane-genus and rational-point suppliers; the field k=R, the form F=x2+y2+z2, the closed subscheme C=V+(F)⊆PR2, and its base change CC to C.

[F1]

Smoothness: for either K=R or K=C, each of the three standard affine charts of CK is presented as AK=K[u,v]/(1+u2+v2). Its Jacobian row is (2u,2v), and in AK one has 1=(−u/2)(2u)+(−v/2)(2v). Thus no prime contains both entries; around every prime one of them is invertible, so the one-equation presentation is standard smooth there by Standard smooth presentations and locally standard smooth maps. By [F2] the structure map CK→Spec⁡K is proper, hence of finite type by Proper morphisms, so the finite-type hypothesis in Smooth morphisms via local standard smooth presentations holds. Therefore every chart, and hence CK, is smooth over K.

[F2]

Properness: for either K=R or K=C, PK2 is proper over K; the closed immersion CK↪PK2 is proper, and its composition with the structure morphism is proper (Finite-dimensional projective space is proper over every base, Closed immersions are proper, Properness survives composition, Proper morphisms).

[F3]

Geometric integrality: over C the form F has no factorisation into linear forms. Any proper factorisation of a homogeneous quadratic over a field has two degree-one factors; writing each as its degree-one homogeneous part plus a constant, the degree-zero and degree-one parts of their product force both constants to vanish. Thus it suffices to test homogeneous linear factors. If F=L1L2, then restricting to z=0 and using unique factorisation in C[x,y] lets us rescale and order the factors as L1=x+iy+αz and L2=x−iy+γz. Their product has xz, yz, and z2 coefficients α+γ, i(γ−α), and αγ, respectively. Equality with F requires α+γ=0, i(γ−α)=0, and αγ=1; the first two force α=γ=0 in characteristic zero, contradicting the third. Thus F is irreducible in C[x,y,z]; its principal ideal is prime by the finite-variable UFD lemma, so CC is reduced and irreducible. It is nonempty since [i:0:1]∈CC. The extension C/R is algebraic (C/R has power basis 1,i and degree 2) and C is algebraically closed (The complex numbers are algebraically closed), so C is an algebraic closure of R (An algebraic closure of a field). Hence CC is the geometric fibre defining geometric integrality of C over R; for CC over C, take the algebraic closure to be C itself. The fibres are integral in the sense of Geometric properties of fibres, giving the geometric-integrality assertions in Curves over a field (Geometric fibres and geometric points, Every finite-variable polynomial ring over a field is a UFD, with prime irreducibles and principal height-one primes).

[F4]

Scheme chain dimension: for K∈{R,C}, the three standard affine charts of CK have coordinate ring AK=K[u,v]/(1+u2+v2) (Relative projective space from standard charts). By [F3], F is irreducible over C and therefore over R; for either field K its homogeneous ideal is prime by the UFD property of K[x,y,z] (Every finite-variable polynomial ring over a field is a UFD, with prime irreducibles and principal height-one primes). Each chart ring is the degree-zero subring of the localization of this homogeneous domain at one coordinate, hence is a nonzero domain. The map K[u]→AK is injective: if a nonzero q(u) lay in (1+u2+v2), then in K[u][v] it would equal (1+u2+v2)h for nonzero h, contradicting additivity of degree in v (Over an integral domain, degrees add under multiplication of nonzero polynomials). Thus u is transcendental over K, while v is algebraic over K(u) by v2+u2+1=0; the transcendence-degree tower formula gives trdeg⁡KFrac⁡(AK)=1 (Transcendence degree is additive in finite towers). The affine-domain dimension theorem gives Krull dimension dim⁡AK=1 (Affine-domain dimension equals transcendence degree). The ring AK is Noetherian as a quotient of a finite-variable polynomial ring, so its spectrum is Noetherian (Finite-variable polynomial algebras over fields are Noetherian by finite generators, The spectrum of a Noetherian ring is a Noetherian topological space). Every nonempty irreducible closed subset of this spectrum has a prime ideal as its unique generic point, and distinct primes have distinct closures; thus chains of nonempty irreducible closed subsets have exactly the lengths of chains of prime ideals. The chart's chain dimension is therefore its Krull dimension, one (A Zariski-closed subset is irreducible exactly when its radical defining ideal is prime, and then it has a unique generic point, Krull dimension of a nonzero ring, Chain dimension and the empty-space convention). Each of the three charts is Noetherian. A descending chain of closed subsets of CK stabilizes after restriction to each chart; since the cover is finite, the maximum of those three stabilization indices works on all of CK. Thus CK is Noetherian, and the open-cover dimension lemma gives its chain dimension as the supremum of the three chart dimensions, namely one (Dimension can be computed on an open cover).

[F5]

Arithmetic genus: a curve X=V+(G) cut out by a nonzero homogeneous form of degree d≥1 has pa(X)=(d−1)(d−2)2; for d=2 this is 0, and for a smooth curve pa=g (Arithmetic genus of a plane curve, Genus via the Euler characteristic).

[F6]

Residue-field degrees: if p is a closed point, an affine neighborhood Spec⁡A of p is of finite type over R, and p remains closed there, so κ(p)=A/mp is finite over R (Curves over a field, A maximal ideal of an affine algebra has finite residue field over the base field). The field R is perfect because it has characteristic zero (Fields of characteristic zero, finite fields, and algebraically closed fields are perfect), so the finite extension κ(p)/R is separable (Every algebraic extension of a perfect field is separable) and simple (A finite extension generated by elements all but possibly one of which are separable is simple). If κ(p)=R(α), the minimal polynomial of α is irreducible and has degree [κ(p):R] (The evaluation kernel and the unique monic irreducible minimal polynomial of an algebraic element, A simple algebraic extension is its minimal-polynomial quotient and has power basis 1,a,…,an−1 and degree n); every irreducible real polynomial has degree 1 or 2 (An irreducible polynomial in R[x] has degree 1 or 2). If this degree were 1, then κ(p)=R and the residue-field point would give an R-morphism Spec⁡R→C (Field-valued points and local-ring points).

[F7]

No real points and even divisor degrees: every real sum of squares x2+y2+z2 vanishes only at the origin, so C(R)=∅. Therefore [F6] excludes residue degree 1 for every closed point, and every closed point has residue degree 2. A divisor is a finite signed combination D=∑pnp[p] of closed points and has degree deg⁡RD=∑pnp[κ(p):R] (Degree divisor proper curve, The residue field at a point of an affine scheme); hence deg⁡RD=2∑pnp is even.

[F8]

The rational-point theorem: a smooth proper geometrically integral curve of genus 0 over a field K that admits a divisor of degree one (equivalently a K-rational closed point) is isomorphic to PK1 (A genus-zero curve with a degree-one divisor is the projective line); and an isomorphism of K-schemes induces a bijection of K-rational points, while PK1 has the K-point [1:0].

[F9]

The Axiom of Choice is assumed in the local-standard-smooth definition used in [F1], in the three properness results used in [F2], in the Noetherian-spectrum and irreducible-closed-subset correspondences used in [F4], and in the genus-zero rational-point theorem [F8]; the arithmetic-genus route in [F5] also inherits the properness suppliers. The factorisation, chart-dimension computations, and residue-field degree argument require no further choice principle, and the algebraic closure used here is the explicitly given C (The Axiom of Choice).

Proof

technique · verify by explicit computation that the real conic is a smooth proper geometrically integral genus-zero curve with no real point, acquire a complex point after base change, and compare $\mathbb R$-points to separate it from $\mathbb P^1_{\mathbb R}$
1.1F1F2F3F4

The conic is a smooth proper geometrically integral curve. By [F1], for each K=R,C, the Jacobian row on each of the three charts has a unit entry in a neighborhood of every prime; the standard smooth presentation criterion gives smoothness at every point. In particular C is smooth over R. By [F2], C is proper over R; by [F3], the algebraic-closure fibre CC is integral, so C is geometrically integral over R; and [F4] directly computes chain dimension one for both C and CC. Therefore C is a smooth proper geometrically integral curve over R.

1.2F6F7

No rational point, hence no degree-one divisor. Every real solution of x2+y2+z2=0 is (0,0,0), which is not a point of PR2, so C(R)=∅. By [F6] every closed point has residue degree either 1 or 2, and degree 1 would give a real point; hence every closed point has degree 2. For any signed divisor D=∑pnp[p], [F7] gives deg⁡RD=2∑pnp, an even integer. Thus no divisor has degree one and the hypothesis of [F8] fails.

2.1F5step 1.1

Genus zero. Applying [F5] to the curve C cut out by the degree-two form gives pa(C)=(2−1)(2−2)/2=0, and since C is smooth the arithmetic genus equals the genus, so g(C)=0.

3.1F1F2F3F4F5F8step 1.1step 2.1

The complex picture. Over C the point [i:0:1] satisfies i2+02+1=0, so CC has a C-rational point; by steps 1.1 and 2.1 applied over C (with [F1]–[F5] read over the algebraically closed field C of characteristic 0), CC is a smooth proper geometrically integral curve of genus 0, and [F8] gives CC≅PC1.

4.1F8F9step 1.2step 3.1∎

The two curves are not isomorphic over R. The line PR1 has the R-rational point [1:0], while C has none by step 1.2; an R-isomorphism would induce a bijection on R-rational points (an isomorphism of functors of points), so C≇PR1. Hence a genus-zero curve over a non-algebraically-closed field need not be a projective line, and the rational-point hypothesis of [F8] cannot be dropped; geometrically the conic is a projective line, so genus zero does not determine the curve arithmetically. The Axiom of Choice is inherited only through the suppliers of [F9]; nothing is selected.

ExampleConstruction: AI-generatedVerification: AI-adaptedprecheck passaudited 2026-10-02Open item page →

The jump l(D+p) - l(D) ranges from zero to the residue degree

Example

Assume the Axiom of Choice inherited from the current divisor, residue-field and projective-line suppliers.

Let k be a field, let C be a smooth proper geometrically integral curve over k (Curves over a field), let D be a divisor and let p be a closed point. By Monotonicity of L(D) in the divisor the quotient L(D+p)/L(D) embeds k-linearly into the residue field κ(p), so the jump is bounded: 0≤l(D+p)−l(D)≤[κ(p):k], and both extremes occur. The intermediate values occur as well: the jump need not be 0 or [κ(p):k], as case (iv) below shows in residue degree two.

On Pk1, with coordinate t=x1/x0 and point at infinity ∞=[0:1]=V(x0) (Divisors on the projective line are classified by degree):

  1. for D=−2[q] and p=[a] with q=[0] and a≠0 rational, both D and D+p=−2[q]+[a] have negative degree, so L(D)=L(D+p)=0 and the jump is 0;
  2. for D=−[q] and p=[a] with a≠q rational, L(D)=0 while L(D+p) is the one-dimensional space spanned by (t−q)/(t−a), whose divisor is [q]−[a]=−(D+p), so the jump is 1=[κ(p):k];
  3. over k=R, for D=0 and p=V(t2+1) of residue degree two, L(0) is the constant field with l(0)=1, while p is linearly equivalent to 2[∞] because div⁡(t2+1)=[p]−2[∞], so O(p)≅O(2) and L(p)={A/(t2+1):deg⁡A≤2} is three-dimensional: the jump is 2=[κ(p):k];
  4. over k=R, for D=−2[∞] and p=V(t2+1), one has L(D)=0 and L(D+p) is the one-dimensional space spanned by 1/(t2+1), so the jump is 1 inside residue degree 2.

The residue-degree case therefore really occurs, the bound is sharp in both extremes, and the jump is in general an intermediate integer of the interval [0,[κ(p):k]].

Scaffold repair, recorded for the owner. The frozen scaffold statement claimed that "the jump l(D+p)−l(D) is either 0 or [κ(p):k]". That strengthening is false: case (iv) exhibits a jump of 1 with residue degree 2. The statement above keeps every promised instance (i)-(iii) with their computations, corrects the general claim to the true bound 0≤l(D+p)−l(D)≤[κ(p):k] of Monotonicity of L(D) in the divisor, and adds case (iv) as the disproof of the false reading. The scaffold's parenthetical in (ii), "a function with divisor [a]−[q]", is also corrected: the spanning function (t−q)/(t−a) has divisor [q]−[a]=−(D+p).

The current Monotonicity of L(D) in the divisor supplies the general residue-field bound. The current Riemann-Roch-space, principal-divisor, Cartier-sheaf, and Cartier-to-Weil interfaces used for the displayed projective-line calculations are The space L(D), Principal weil divisor and class group, Invertible sheaf of cartier divisor, Cartier and Weil divisors agree on a smooth curve, On an integral scheme, Cartier divisors modulo principal divisors compute the Picard group and The Picard group of the projective line.

Facts & Assumptions

Given: the Axiom of Choice inherited from the current divisor, residue-field and projective-line suppliers; a field k, a smooth proper geometrically integral curve C, a divisor D, a closed point p, and the computations on Pk1 listed in the statement.

[F1]

On Pk1 with coordinate t=x1/x0: Pk1 is a smooth proper geometrically integral curve of genus 0; for a monic irreducible g∈k[t] of degree d the closed point pg=V(g)⊆Spec⁡k[t] has [κ(pg):k]=d and div⁡(g)=[pg]−d[∞]; div⁡(x0)=[∞] and O(1)≅O(∞) with deg⁡kO(1)=1; every divisor on Pk1 is linearly equivalent to deg⁡k(D)[∞] (Divisors on the projective line are classified by degree, Twisting sheaf on Proj).

[F2]

Every nonzero A∈k[t] factors as a unit times a product of monic irreducibles (For every field F, F[x] is a unique factorisation domain); for a monic irreducible g the point pg is a closed point of U0=Spec⁡k[t] with residue degree d=deg⁡g and ord⁡pg(g)=1, ord⁡x(g)=0 for every other closed point x of U0, and ord⁡∞(g)=−d (Divisors on the projective line are classified by degree, Order codimension one rational function).

[F3]

The order ord⁡x at a closed point is a group homomorphism k(C)×→Z with ord⁡x(fg)=ord⁡x(f)+ord⁡x(g), ord⁡x(f−1)=−ord⁡x(f); ord⁡x(f)≥0 exactly for f∈OC,x, and ord⁡x(f)=0 exactly for units. The local ring OC,p is a discrete valuation ring with uniformizer tp and residue field κ(p)=OC,p/(tp), and [κ(p):k]=dim⁡kκ(p) (Order codimension one rational function, Local rings at closed points of smooth curves are discrete valuation rings, The residue field at a point of an affine scheme, Degree divisor proper curve).

[F4]

The current The space L(D) gives the order description of L(D) and its global-section identification; Principal weil divisor and class group gives the additive divisor convention. The current Invertible sheaf of cartier divisor, Cartier and Weil divisors agree on a smooth curve, On an integral scheme, Cartier divisors modulo principal divisors compute the Picard group and The Picard group of the projective line identify the attached sheaves, including O(p)≅O(2) in case (3).

[F5]

The general bound is Monotonicity of L(D) in the divisor: L(D)⊆L(D+p), the quotient L(D+p)/L(D) embeds k-linearly into κ(p), whence 0≤l(D+p)−l(D)≤[κ(p):k], and the spaces are finite-dimensional with l(D)=dim⁡kL(D)=h0(D) (The Riemann-Roch dimension l(D), Finite-dimensionality of the Riemann-Roch space, Finite-dimensional vector space, and its dimension dim⁡FV; infinite-dimensional means having no finite basis).

[F6]

The Axiom of Choice enters only through the suppliers of [F1]-[F5], exactly those recorded there; the explicit computations below select only the normalized spanning functions exhibited (The Axiom of Choice).

Verification

technique · direct computation of the order conditions on $\mathbb P^1_k$ in each of the four cases, with the bound of the one-point lemma supplying the general inequality
1.1F2F3F4F5

Set-up. By [F5] one has L(D)⊆L(D+p) and 0≤l(D+p)−l(D)≤[κ(p):k] for every divisor D and closed point p on C; by [F4] a rational function f∈k(C)× lies in L(D) if and only if ord⁡x(f)+nx≥0 for every closed point x, where nx is the coefficient of D. On Pk1 every nonzero rational function is a quotient A/B of nonzero polynomials (Divisors on the projective line are classified by degree, For every field F, F[x] is a unique factorisation domain), and by [F2] and [F3] its orders are ord⁡x(A/B)=ord⁡x(A)−ord⁡x(B) at every closed point, with ord⁡∞(A/B)=deg⁡B−deg⁡A for A/B in lowest terms.

1.2F1F2F3F4

Case (1): D=−2[q], p=[a], q=[0], a≠0. A rational function f=A/B in lowest terms lies in L(D) exactly when ord⁡q(f)≥2 and ord⁡x(f)≥0 for every x≠q: the second condition says f has no poles at all, so B is constant and f=A is a polynomial, while the first says the polynomial A has a zero of order at least 2 at q; the condition at infinity ord⁡∞(A)=−deg⁡A≥0 forces deg⁡A=0, and then ord⁡q(A)=0<2. Hence L(D)=0. The same argument with the single change ord⁡a(f)≥−1, applied to L(D+p)=L(−2[q]+[a]), says f=A/B is in L(D+p) only if f=A/(t−a) with deg⁡A≤1 and A divisible by t2 at q, which is impossible for a nonzero polynomial of degree at most one; hence L(D+p)=0 and the jump is 0.

1.3F1F2F3F4

Case (2): D=−[q], p=[a], a≠q. Here f∈L(D) requires ord⁡q(f)≥1 and ord⁡x(f)≥0 for x≠q: again f=A is a polynomial with a zero at q, and ord⁡∞(A)=−deg⁡A≥0 forces deg⁡A=0, contradicting the zero at q; so L(D)=0. For D+p=[a]−[q] the conditions are ord⁡q(f)≥1, ord⁡a(f)≥−1 and ord⁡x(f)≥0 for all other x: writing f=A/(t−a) with A∈k[t], the condition at infinity is ord⁡∞(f)=1−deg⁡A≥0, so deg⁡A≤1, and the condition at q is A(q)=0; hence A=c(t−q) and L(D+p) is the one-dimensional span of (t−q)/(t−a), a nonzero function with ord⁡q=1, ord⁡a=−1, all other orders zero, so div⁡((t−q)/(t−a))=[q]−[a]=−(D+p). The jump is 1, and for the rational point p=[a] one has [κ(p):k]=1.

1.4F1F2F3F4

Case (3): k=R, D=0, p=V(t2+1). By [F1] the polynomial t2+1 is monic irreducible of degree two with closed point p of residue degree [κ(p):k]=2 and div⁡(t2+1)=[p]−2[∞], so [p] is linearly equivalent to 2[∞]. The space L(0) consists of the rational functions with ord⁡x(f)≥0 at every closed point: these are the polynomials (since the denominator of a reduced fraction would give a pole) with ord⁡∞(A)=−deg⁡A≥0, i.e. the constants, so l(0)=1. For L(p): a reduced fraction f=A/B with all orders at least 0 except possibly ord⁡p(f)≥−1 has B=1 or B=t2+1, so f=A/(t2+1) with A∈R[t], and the condition at infinity is ord⁡∞(f)=2−deg⁡A≥0, i.e. deg⁡A≤2. Hence L(p)={A/(t2+1):deg⁡A≤2}, spanned by 1/(t2+1), t/(t2+1), t2/(t2+1): these three are linearly independent because clearing the denominator turns a relation into a polynomial identity of degree at most two. Therefore l(p)=3 and the jump is 2=[κ(p):k], realized over the non-algebraically-closed field R.

2.1F2F3F4step 1.1

Case (4): k=R, D=−2[∞], p=V(t2+1). Here f∈L(D) requires ord⁡∞(f)≥2 and ord⁡x(f)≥0 for every x≠∞: as in step 1.2 this forces f=A to be a polynomial with −deg⁡A≥2, impossible for a nonzero polynomial, so L(D)=0. In L(D+p) the conditions are ord⁡∞(f)≥2, ord⁡p(f)≥−1 and ord⁡x(f)≥0 for all other x: a reduced fraction with no pole outside p is of the form f=A/(t2+1) with A∈R[t], and 2−deg⁡A=ord⁡∞(f)≥2 forces deg⁡A=0; hence L(D+p) is the one-dimensional span of 1/(t2+1), whose orders are +2 at ∞ and −1 at p, so the jump is 1 while [κ(p):k]=2. This is the intermediate value: the jump is neither 0 nor the full residue degree.

3.1F1F4F5F6step 1.2step 1.3step 1.4step 2.1∎

Assembly and choice accounting. Steps 1.2-2.1 exhibit jumps 0, 1 and 2 on Pk1, and in particular the extreme value [κ(p):k] occurs for [κ(p):k]=1 in case (2) and for [κ(p):k]=2 in case (3), while case (4) gives the intermediate value 1 in residue degree 2; the general inequality 0≤l(D+p)−l(D)≤[κ(p):k] is [F5]. The sheaf restatement of case (3), O(p)≅O(2), follows from the current Cartier/Picard route [F4]. The Axiom of Choice is inherited only through the suppliers of [F1]-[F5], as recorded in [F6]; no infinite selection is made above, since the spanning functions are exhibited by formulas.

CounterexampleConstruction: Literature-sourcedVerification: AI-adaptedprecheck passaudited 2026-10-02Open item page →

The Riemann inequality is not an equality for special divisors

Counterexample

Assume the Axiom of Choice as inherited from the Riemann--Roch, curve-divisor, Jacobian, weighted-Bezout, and projective-properness suppliers. Let k be any field and let C be a smooth proper geometrically integral curve over k of genus g≥1 (Genus via the Euler characteristic). The zero divisor D=0 satisfies l(0)=1,i(0)=h1(C,OC)=g≥1, whereas the Riemann inequality gives only l(0)≥deg⁡k(0)+1−g=1−g. Thus the inequality is strict, with excess l(0)−(deg⁡k(0)+1−g)=g=i(0). More generally, Riemann--Roch gives l(D)=deg⁡k(D)+1−g+i(D) for every divisor D, so equality with the lower bound holds exactly when i(D)=0, that is, exactly when D is nonspecial (Special and nonspecial divisors). In particular, every special divisor gives a strict inequality.

A concrete instance is the Fermat quartic X=V+(x04+x14+x24)⊆Pk2 over an algebraically closed field k of characteristic zero. Steps 1.2, 1.3, 2.3, and 3.1 verify scheme-theoretically that it is smooth, pure of dimension one, geometrically integral, and of genus three. Its zero divisor therefore has l(0)=1, i(0)=3, and the strict inequality 1>−2.

Supplier status. The Fermat geometry is proved below without using the generic complete-intersection existence claim. The genus formula and the Riemann--Roch, degree, and divisor interfaces cited below are draft suppliers in this run; this repair does not certify their separate proofs.

Facts & Assumptions

Given: the Axiom of Choice inherited from the Riemann--Roch, curve-divisor, Jacobian, weighted-Bezout, and projective-properness suppliers; a field k, a smooth proper geometrically integral curve C over k of genus g≥1, and the zero divisor D=0 on C.

[F1]

Riemann--Roch as l minus the index of speciality gives, for every divisor D, l(D)−i(D)=deg⁡k(D)+1−g,i(D)=h1(C,OC(D))≥0. Thus equality in the Riemann inequality holds exactly when i(D)=0, which is the definition of nonspeciality. (Riemann-Roch as l minus i, Special and nonspecial divisors)

[F2]

The zero divisor has l(0)=dim⁡kH0(C,OC)=1, since the global sections of the structure sheaf on a proper integral curve are canonically k. (The Riemann-Roch dimension l(D), Functions on a proper curve)

[F3]

The index of speciality of the zero divisor is i(0)=h1(C,OC)=g(C). (The index of speciality i(D), Genus via the Euler characteristic)

[F4]

The zero divisor has degree zero, and divisor degree is the additive weighted sum of closed-point coefficients. (Degree divisor proper curve)

[F5]

In an affine plane chart, the smoothness criterion for a scheme presented by the actual equation f is given by an invertible 1×1 Jacobian minor. At a rational closed point, regularity of its local ring is equivalent to Jacobian rank 2−dim⁡Am, even if the actual ideal (f) is not radical. (Relative Jacobian criterion with its presentation hypothesis, Jacobian rank detects regularity at closed points)

[F6]

In the Fermat calculation, k is algebraically closed. For a nonzero nonunit equation f in a chart ring k[u,v], all irreducible components of the hypersurface have dimension one: each minimal prime over (f) has height one by the principal ideal theorem, and the affine-domain dimension formula gives quotient dimension one. At a closed point with maximal ideal m, its residue field is finite over k and hence equals k; the dimension formula gives dim⁡k[u,v]m=2. A minimal prime over f in this local ring is nonzero and has height one by the principal ideal theorem. No prime can lie strictly between it and m, since that would give a chain of length at least three in a ring of dimension two. Therefore the hypersurface local ring has dimension one. The polynomial ring is Noetherian. The same minimal-prime calculation gives dimension one for every nonempty affine chart component. Every projective irreducible component meets a standard chart, and its intersection is a chart component, so every projective component has dimension one; the open-cover dimension lemma gives scheme dimension one as well. (Finite-variable polynomial algebras over fields are Noetherian by finite generators, The dimension formula for affine domains, Affine-domain dimension equals transcendence degree, Krull's principal ideal theorem, Dimension can be computed on an open cover, A maximal ideal of an affine algebra has finite residue field over the base field)

[F7]

The scheme-theoretic projective Bezout formula gives a nonempty finite intersection for coprime positive-degree forms and computes its local lengths; over an algebraically closed field the residue-degree weights are all one. (Algebraic Bezout formula as a sum of local scheme lengths)

[F8]

A smooth finite-type scheme over an algebraically closed field has regular local rings. (Classical and scheme smoothness over a perfect field)

[F9]

The published arithmetic-genus theorem gives pa(X)=(d−1)(d−2)/2 for an integral plane curve cut out by a homogeneous form of degree d (Arithmetic genus of a plane curve). For the smooth proper geometrically integral curve established in steps 1.2, 1.3, and 2.3, the current genus definition identifies g(X)=pa(X)=1−χ(OX) (Genus via the Euler characteristic).

[F10]

The Axiom of Choice is inherited from the Riemann--Roch, curve and divisor, dimension and Jacobian, weighted-Bezout, and projective-properness suppliers used here. (The Axiom of Choice)

[F11]

A finite-variable polynomial ring over a field is a UFD, so every irreducible polynomial in it is prime. (Every finite-variable polynomial ring over a field is a UFD, with prime irreducibles and principal height-one primes)

[F12]

Projective space over a field is proper; closed immersions and compositions of proper morphisms are proper. Hence a closed subscheme of Pk2 is proper over k. (Finite-dimensional projective space is proper over every base, Closed immersions are proper, Properness survives composition)

Proof

Proof technique: compute the zero-divisor terms, prove the general strictness statement by rearranging Riemann--Roch, and realize the genus-three case by an explicit Fermat quartic.

1.1F1F2F3F4F5F6F7F8F9F10step 2.2step 3.1∎

The zero divisor is special. By [F3], i(0)=g≥1, so it is nonzero; hence [F1] says that D=0 is special. By [F2], l(0)=1. [F1, F2, F3] 1.2 (Smoothness of the Fermat quartic.) Let k be algebraically closed of characteristic zero, set F=x04+x14+x24, and let X=V+(F)⊆Pk2. In the chart xi≠0, set xi=1 and write the other coordinates as u,v; the equation is f=1+u4+v4. The opens D(u) and D(v) cover this affine hypersurface, since no prime containing both u and v can contain f=1+u4+v4. On D(u), the derivative ∂f/∂u=4u3 is a unit; on D(v), ∂f/∂v=4v3 is a unit. Each is therefore a standard smooth presentation by [F5], so the three projective charts show that X is smooth over k. The scheme is nonempty: choose a∈k with a4=−1; then [1:a:0]∈X. [F5, given] 1.3 (Scheme dimension.) In each of the three standard charts, identified by projective hypersurface affine pieces, the coordinate ring is k[u,v]/(f) with f=1+u4+v4, a nonzero nonunit. The polynomial ring k[u,v] is Noetherian. Every minimal prime q over (f) is nonzero and has height one by [F6] and the principal ideal theorem. The dimension formula gives trdeg⁡kFrac⁡(k[u,v]/q)=1, and affine-domain dimension equals this transcendence degree. Thus every irreducible component in every nonempty chart has dimension one. Since the standard charts cover X, it is pure of dimension one; the same calculation at a closed point gives local dimension one. The scheme X is proper because it is the closed subscheme V+(F)↪Pk2: projective space is proper over k, and the closed immersion and composite are proper by [F12]. [F6, F12, given] 2.1 The inequality at D=0 is strict. By [F1] and [F4], l(0)−i(0)=deg⁡k(0)+1−g=1−g. Using step 1.1 gives l(0)=1>1−g=deg⁡k(0)+1−g because g≥1, and the excess is 1−(1−g)=g=i(0). [F1, F4, step 1.1] 2.2 For any divisor D, rearranging [F1] gives l(D)=deg⁡k(D)+1−g+i(D). Since i(D)≥0, the Riemann inequality is strict exactly when i(D)>0, which is exactly when D is special; equality holds exactly for nonspecial divisors. This proves the general claim independently of the concrete example. [F1, step 1.1] 2.3 (Integrality.) We show that F is square-free and irreducible. Suppose an irreducible homogeneous factor G occurs at least twice. Choose a line ℓ=0 not containing G; [F7] gives a closed point p∈V+(G,ℓ). On a chart through p, the actual equation f of X lies in the square of the maximal ideal, so its Jacobian row is zero. The local ring has dimension one by step 1.3, and [F5] says it is not regular, contradicting smoothness from step 1.2 and [F8]. Hence F is square-free. If the square-free F were reducible, choose a nonconstant irreducible factor G and let H be the product of the remaining factors. Then G,H are coprime and have positive degree. By [F7], they meet at a closed point p. There f=gh lies in the square of the maximal ideal, so the Jacobian row again vanishes. The local ring has dimension one, contradicting regularity exactly as above. Thus F is irreducible. Since k[x0,x1,x2] is a UFD by [F11], (F) is prime, and X is integral. As k is algebraically closed, this proves geometric integrality. [F5, F6, F7, F8, step 1.2, step 1.3] 3.1 (Genus three.) Steps 1.2, 1.3, and 2.3 show that X is a smooth proper geometrically integral plane curve cut out by a homogeneous quartic. The arithmetic-genus theorem in [F9] gives pa(X)=(4−1)(4−2)2=3, and the genus definition in [F9] identifies g(X)=pa(X) for this smooth curve. Hence g(X)=3, as needed for the concrete instance in the Counterexample section. [F9, step 1.2, step 1.3, step 2.3] 4.1 (Conclusion and choice accounting.) Steps 1.1 and 2.1 show that the zero divisor on any curve of genus at least one gives a strict Riemann inequality with excess exactly i(0). Step 2.2 proves the stated criterion for all divisors. Steps 1.2, 1.3, 2.3, and 3.1 give the promised characteristic-zero genus-three example, where the inequality is 1>−2. The Axiom of Choice [F10] is inherited through the Riemann--Roch, affine-dimension, Jacobian, and Bezout suppliers; no additional choice is made.

ExampleConstruction: Literature-sourcedVerification: AI-adaptedprecheck passaudited 2026-10-02Open item page →

A principal divisor of degree zero on the projective line

Example

Assume the Axiom of Choice inherited from the current divisor and projective-line cohomology suppliers.

Let k be a field and let Pk1 have coordinate t=x1/x0 on the standard chart U0=Spec⁡k[t], with point at infinity ∞=[0:1]=V(x0) (Divisors on the projective line are classified by degree). Let a≠b be distinct k-rational points of U0, so a,b∈k and the associated closed points are [a]=V(t−a) and [b]=V(t−b). The rational function f=t−at−b ∈ k(t)× has divisor div⁡(f)=[a]−[b], which is a principal divisor of degree deg⁡k([a]−[b])=1−1=0. Its divisor class satisfies the degree shift and Riemann-Roch numerically: the attached invertible sheaf O(div⁡f)=O([a])⊗O([b])−1 is isomorphic to OPk1, since the class of a principal divisor is trivial and since deg⁡kdiv⁡(f)=0 on Pk1 forces the class to be trivial under the isomorphism Pic⁡(Pk1)→Z (The Picard group of the projective line). Hence l(div⁡f)=h0(O)=1,i(div⁡f)=h1(O)=0, by the explicit cohomology of the structure sheaf (Global sections of projective twists, Top cohomology of projective twists). Alternatively L(div⁡f) is the one-dimensional k-space spanned by 1/f=t−bt−a, because div⁡(g)+div⁡(f)≥0 for a nonzero g∈k(t) is equivalent to div⁡(gf)≥0, and a rational function on Pk1 with no poles is constant. In particular f∉L(div⁡f): the divisor div⁡(f)+div⁡(f)=2[a]−2[b] is not effective, so the nonzero elements of L(div⁡f) are the scalar multiples of 1/f and not those of f. Hence χ(O(div⁡f))=1=χ(O) with deg⁡kdiv⁡(f)=0, and Riemann-Roch reads 1−0=0+1−0 on both sides, the right-hand side being deg⁡kdiv⁡(f)+1−g with g=g(Pk1)=0 (Riemann-Roch as l minus i). The same computation for f=g(t) a monic polynomial of degree d gives div⁡(g)=Z(g)−d[∞],Z(g)=∑imi[pi], where g=∏igimi is the factorisation of g into monic irreducibles gi of degree di and pi=V(gi); this divisor has degree ∑imidi−d=0, showing that the individual zero and pole parts need not be trivial even though the class is principal.

Scaffold repair, recorded for the owner. The frozen scaffold statement claimed that "alternatively L(div⁡f) consists of the scalar multiples of f because div⁡(g)+div⁡(f)≥0 forces g/f to have no poles". That is false as written: the condition is equivalent to div⁡(gf)≥0, hence to gf∈k, so L(div⁡f) consists of the scalar multiples of 1/f, and f itself is not in L(div⁡f) because 2[a]−2[b] is not effective. The statement above keeps every other promised claim and records the corrected spanning function; the general degree-zero claim is deg⁡k([a]−[b])=0, computed directly below.

The current principal-divisor, Cartier/Picard, and line-bundle interfaces used below are Principal weil divisor and class group, Invertible sheaf of cartier divisor, Addition of Cartier divisors is tensor product of their sheaves, On an integral scheme, Cartier divisors modulo principal divisors compute the Picard group and Cartier and Weil divisors agree on a smooth curve. The degree zero asserted for this example is computed explicitly from [a]−[b]; no general principal-divisor degree theorem is needed.

Facts & Assumptions

Given: the Axiom of Choice inherited from the current divisor and projective-line cohomology suppliers; a field k, the projective line Pk1 with chart U0=Spec⁡k[t], coordinate t and point at infinity ∞=[0:1]; distinct k-rational points a≠b, with closed points [a]=V(t−a) and [b]=V(t−b); and the rational function f=(t−a)/(t−b)∈k(t)×.

[F1]

Projective-line data: Pk1 is a smooth proper geometrically integral curve over k of genus 0; for every monic irreducible g∈k[t] of degree d the closed point p=V(g) of U0 has [κ(p):k]=d and div⁡(g)=[p]−d[∞]; every divisor D on Pk1 is linearly equivalent to deg⁡k(D)[∞], and the degree homomorphism deg⁡k from CaDiv⁡(Pk1)/Prin⁡(Pk1) to Z is an isomorphism (Divisors on the projective line are classified by degree).

[F2]

Divisors and orders: a divisor on a curve is a finite formal Z-linear combination of closed points, with effectiveness read coefficientwise; for a closed point x of a smooth curve the order ord⁡x at x is a homomorphism on k(C)×, so ord⁡x(gh)=ord⁡x(g)+ord⁡x(h), and ord⁡x(f)≥0 exactly when f is regular at x (Divisors on a smooth proper curve, Order codimension one rational function).

[F3]

Degree: the k-degree of a divisor is deg⁡k(D)=∑xnx[κ(x):k], a group homomorphism Div⁡(C)→Z; for a rational point x one has [κ(x):k]=1 (Degree divisor proper curve, Divisors on the projective line are classified by degree).

[F4]

The Riemann-Roch space: for a divisor D on a smooth proper geometrically integral curve, L(D)={g∈k(C)×:div⁡(g)+D≥0}∪{0} is the k-subspace of functions whose poles are no worse than −D, membership being read coefficientwise as ord⁡x(g)+nx≥0 at every closed point x, and the divisor of a rational function is div⁡(g)=∑xord⁡x(g)[x] (The space L(D)).

[F5]

The integer l(D)=dim⁡kL(D)=h0(D)=dim⁡kH0(C,OC(D)) is the dimension of the Riemann-Roch space, with hi(D)=dim⁡kHi(C,OC(D)); in particular l(0)=1 for the zero divisor (The Riemann-Roch dimension l(D), The space L(D)).

[F6]

The polynomial ring F[x] over a field is a unique factorisation domain (For every field F, F[x] is a unique factorisation domain).

[F7]

Sections of the structure sheaf: OX(0)=OX for the twisting sheaf of Proj⁡, and H0(Pk1,OPk1(0))≅k[x0,x1]0, so h0(OPk1)=1 (Twisting sheaf on Proj, Global sections of projective twists).

[F8]

Top cohomology vanishes: H1(Pk1,OPk1(0))=0, so h1(OPk1)=0 (Top cohomology of projective twists).

[F9]

Riemann-Roch: for every divisor D on a smooth proper geometrically integral curve of genus g one has l(D)−i(D)=deg⁡k(D)+1−g with i(D)=h1(D)≥0 (Riemann-Roch as l minus i).

[F10]

Degree shift of the Euler characteristic: χ(C,OC(D))−χ(C,OC)=deg⁡k(D) for every divisor D, where χ=h0−h1 (Riemann-Roch in Euler-characteristic form: the degree shift).

[F11]

The Picard group of the projective line is Z: the degree homomorphism induces an isomorphism Pic⁡(Pk1)→Z, so an invertible sheaf of degree zero is isomorphic to OPk1 (The Picard group of the projective line).

[F12]

The current Invertible sheaf of cartier divisor attaches the sheaf to a Cartier divisor; Addition of Cartier divisors is tensor product of their sheaves gives its tensor and dual identities; On an integral scheme, Cartier divisors modulo principal divisors compute the Picard group identifies principal divisors with the trivial line-bundle class; Cartier and Weil divisors agree on a smooth curve identifies Cartier and Weil divisors on this smooth curve; and Principal weil divisor and class group supplies the principal-divisor convention. These interfaces give O(div⁡f)≅O and the displayed tensor expression. The degree zero in this example is computed directly from [a]−[b].

[F13]

The Axiom of Choice is available and is inherited only through the suppliers named above; the computations below select nothing beyond the given points and functions (The Axiom of Choice).

Verification

technique · direct computation of $\operatorname{div}(f)$ and of $L(\operatorname{div}f)$ on $\mathbb P^1_k$ from the order of vanishing at the closed points, compared with the explicit cohomology of $\mathcal O$ and the Riemann-Roch identity
1.1F1F3F6

The two rational points and the divisors of t−a, t−b. The polynomials t−a and t−b are monic of degree 1, hence irreducible in k[t]: a factorisation into non-units would give two factors of degree at least one, whose degrees add to 1. So both are monic irreducibles of degree d=1, and by [F1] applied to g=t−a and to g=t−b, div⁡(t−a)=[a]−[∞],div⁡(t−b)=[b]−[∞], with [κ(a):k]=1=[κ(b):k]. Since a≠b, the maximal ideals (t−a)≠(t−b) are distinct, so [a]≠[b] and [a]−[b]≠0.

2.1F2F3F4step 1.1

The divisor of f. The order at a closed point is additive in products and quotients [F2], so for every closed point x, ord⁡x(f)=ord⁡x(t−a)−ord⁡x(t−b). Summing against [x] as in [F4] and using step 1.1, div⁡(f)=div⁡(t−a)−div⁡(t−b)=[a]−[∞]−[b]+[∞]=[a]−[b]. This is a principal divisor by construction — it is the divisor of the nonzero rational function f — and its degree is deg⁡k([a]−[b])=[κ(a):k]−[κ(b):k]=1−1=0 by [F3] and step 1.1. In particular the zero part [a] and the pole part [b] are individually nontrivial while the total degree vanishes.

2.2F1F2F3F6step 1.1

A general monic polynomial. Let g∈k[t] be monic of degree d≥0. By [F6] it factors as g=∏igimi with pairwise distinct monic irreducibles gi of degree di and exponents mi≥1, where ∑imidi=d. Setting pi=V(gi) and Z(g)=∑imi[pi], additivity [F2] and the divisor formula of [F1] give div⁡(g)=∑imidiv⁡(gi)=∑imi([pi]−di[∞])=Z(g)−d[∞], a principal divisor; its degree is deg⁡kZ(g)−deg⁡k(d[∞])=∑imidi−d=0 by [F3] and the residue-degree formula [κ(pi):k]=di of [F1]. For d≥1 the zero part Z(g) and the pole part d[∞] are both nonzero, while Z(g)∼d[∞] because their difference is the principal divisor div⁡(g); for d=0 the polynomial is g=1 and div⁡(g)=0, which is the case Z(g)=0 and 0[∞]=0 of the formula. Taking g=t−a recovers div⁡(t−a)=[a]−[∞] of step 1.1.

3.1F1F11F12step 2.1

The triviality of the class and the sheaf. Since div⁡(f) is principal, its class in CaDiv⁡(Pk1)/Prin⁡(Pk1) is zero, and the degree isomorphism of [F1] is injective, so the class of a degree-zero divisor is trivial: div⁡(f)∼0 by step 2.1. By the current Cartier/Picard dictionary [F12], the principal-divisor case of the sheaf attachment gives O(div⁡f)≅OPk1(0)≅OPk1, and the tensor identities give O(div⁡f)≅O([a])⊗O([b])−1; the same conclusion via degrees uses [F11], under which a degree-zero invertible sheaf on Pk1 is isomorphic to OPk1.

3.2F1F2F4F5F6step 2.1

The sections of div⁡(f), directly. Let g∈k(t)×. By [F4], g∈L(div⁡f) if and only if ord⁡x(g)+ord⁡x(f)≥0 at every closed point x, which by the additivity of [F2] is the same as ord⁡x(gf)≥0 at every closed point x, that is div⁡(gf)≥0 by [F4]. So g∈L(div⁡f) if and only if gf∈L(0), and the space L(0) is the constant field k: a nonzero rational function h∈k(t)× factors by [F6] as h=c∏igini with c∈k×, pairwise distinct monic irreducibles gi of degree di and exponents ni∈Z, so that div⁡(h)=∑ini([pi]−di[∞]) by [F1] and the additivity of the orders [F2]; effectivity forces ni≥0 for every i (from the coefficient at pi) and −∑inidi≥0 (from the coefficient at ∞), and since di≥1 while ni≥0 this gives ni=0 for all i and h=c. Hence g∈L(div⁡f) if and only if gf=c for some c∈k, i.e. L(div⁡f)=k⋅(1/f), a one-dimensional space spanned by 1/f=(t−b)/(t−a), and l(div⁡f)=1 by [F5]. In particular f∉L(div⁡f): the function f corresponds to c=gf=f2∉k, and directly div⁡(f)+div⁡(f)=2[a]−2[b] has coefficient −2<0 at [b], so it is not effective.

4.1F5F7F8F12step 3.1step 3.2

The cohomological reading. By step 3.1, O(div⁡f)≅OPk1; by [F7] the structure sheaf has h0(OPk1)=1, and by [F8] it has h1(OPk1)=0. Therefore l(div⁡f)=h0(div⁡f)=1 and i(div⁡f)=h1(div⁡f)=0, in agreement with the direct computation of step 3.2; here l(D)=h0(D) and i(D)=h1(D) are the dimensions attached to D and its sheaf [F5], the sheaf-theoretic equality being the current dictionary [F12].

5.1F1F9F10step 2.1step 4.1

Riemann-Roch and the Euler characteristic. The Euler characteristics are χ(OPk1)=h0(O)−h1(O)=1−0=1 and, by step 4.1, χ(O(div⁡f))=h0(div⁡f)−h1(div⁡f)=1−0=1, so χ(O(div⁡f))=χ(O) as the degree shift [F10] requires for the degree-zero divisor div⁡(f): χ(O(div⁡f))−χ(O)=0=deg⁡kdiv⁡(f) by step 2.1. Riemann-Roch [F9] reads l(div⁡f)−i(div⁡f)=1−0=1=deg⁡kdiv⁡(f)+1−g=0+1−0, the genus of the projective line being g=g(Pk1)=0 by [F1] and the degree being computed in step 2.1.

6.1F10F12F13step 2.1step 2.2step 3.1step 3.2step 4.1step 5.1∎

Assembly and the current supplier route. Step 2.1 computes div⁡(f)=[a]−[b] of degree zero, step 3.1 identifies the divisor class as trivial and the attached sheaf as OPk1, steps 3.2 and 4.1 compute L(div⁡f)=k⋅(1/f) and (l,i)=(1,0), step 5.1 reads the degree shift and Riemann-Roch as 1−0=0+1−0, and step 2.2 gives div⁡(g)=Z(g)−d[∞] for every monic polynomial of degree d. The sheaf identifications of steps 3.1 and 4.1 use the current interfaces [F12]; the degree-zero claim is computed explicitly in step 2.1, and the direct divisor calculations of steps 2.1, 2.2 and 3.2 use the current order route. The Axiom of Choice enters only through the suppliers recorded in [F13]: the functions f, 1/f and the factorisations are exhibited by formulas, and no family of objects is selected.

ExampleConstruction: Literature-sourcedVerification: AI-adaptedprecheck passaudited 2026-10-02Open item page →

A pencil of functions with poles at one point defines a finite map to the projective line

Example

Assume the Axiom of Choice inherited from the current bounded-pole, linear-system and finite-map suppliers.

Let k be a field, let C be a smooth proper geometrically integral curve over k, and let f∈k(C)× be a nonconstant rational function whose poles all lie at a single closed point p, of order m≥1; thus (f)∞=m[p] is the pole divisor of f, and f∈L(np) for every n≥m — the case "all poles at p, of order at most n" of the scaffold. Write D0:=(f)∞=m[p]. Then:

  1. 1 and f are linearly independent elements of L(np) for every n≥m, hence also of L(D0).
  2. In the space L(D0) the subspace V0=k⋅1+k⋅f is two-dimensional and base-point-free: the section sf of OC(D0) has unit coefficient in a local trivialization at p, and s1 has unit coefficient at every point away from p. The morphism φV0:C→Pk1 attached to V0 by A base-point-free linear system defines a morphism to projective space is exactly the finite morphism φf of A nonconstant rational function defines a finite map to the projective line, of degree [k(C):k(f)] and with fibre over infinity equal to the pole divisor (f)∞.
  3. For n>m the same two elements 1,f in the larger space L(np) are not base-point-free: since div⁡(1)+np=np and div⁡(f)+np=(f)0+(n−m)p, the point p lies in both divisors, so p is a base point. Among the divisors np≥(f)∞ the base-point-free hypothesis therefore holds exactly at the pole divisor D0 (n=m).
  4. On the projective line, with coordinate t and D=[∞]=(t)∞, the pair 1,t inside L([∞]) is base-point-free and the attached morphism is [1:t], the identity of Pk1: it is finite of degree one with fibre over infinity the single point [∞]. For n≥2 the same pair inside L(n[∞]) has base point ∞ (the two divisors n[∞] and [0]+(n−1)[∞] both contain ∞), so no morphism is attached there; the identity is attached to the pole divisor [∞].
  5. The morphism attached to a base-point-free subspace of L(D) depends on the subspace and not only on D: on Pk1 with D=2[∞] the pencils V1=k⋅1+k⋅t2 and V2=k⋅1+k⋅(t2+t) are both base-point-free subspaces of the same L(2[∞]), and the attached morphisms satisfy φV1♯(t)=t2 and φV2♯(t)=t2+t; no fractional linear transformation M(u)=au+bcu+d satisfies M(t2)=t2+t, so the two morphisms are not related by the projective-linear action of PGL2(k) on the target and are genuinely different.
  6. By Rational functions with poles bounded at one point, for every closed point p of residue degree d and every n≥1 with nd+1−g≥2 a nonconstant f∈L(np) of exactly this kind exists, so the construction is nonempty. The simplest instance is C=Pk1 with f=t, whose only pole is at infinity and for which φf is the identity; this realizes the construction of Finite morphisms from a curve to the projective line in the case where the only pole is at infinity.

Scaffold repair, recorded for the owner. The frozen scaffold claimed that 1 and f span a base-point-free subspace of L(np) for a pole order "at most n", and that "the same two sections, viewed inside the larger space L(n[∞]), define the same morphism". Both clauses are false when the pole order m at p is strictly smaller than n, and false on Pk1 for n≥2: in L(np) both div⁡(1)+np=np and div⁡(f)+np=(f)0+(n−m)p contain p, so p is a base point and the hypothesis of the base-point-free morphism theorem fails. The repair keeps every promised object — the two sections, the two-dimensional subspace, the identification of its morphism with φf, the projective-line identity at D=[∞], and the dependence of the morphism on the chosen subspace rather than on D alone — and states base-point-freeness at the correct divisor, the pole divisor (f)∞; the enlarged divisors are handled as the base-point case in item 3.

The current Rational functions with poles bounded at one point supplies the nonconstant function in the existence clause. The Riemann-Roch-space and base-point-free interfaces are The space L(D), Base points and base-point-free linear systems and A base-point-free linear system defines a morphism to projective space. The finite map and its pole fibre are supplied by A nonconstant rational function defines a finite map to the projective line and Finite morphisms from a curve to the projective line.

Facts & Assumptions

Given: the Axiom of Choice inherited from the current bounded-pole, linear-system and finite-map suppliers; a field k, a smooth proper geometrically integral curve C over k, a nonconstant f∈k(C)× whose poles all lie at a single closed point p, of order m≥1, an integer n≥m, and D0=(f)∞=m[p].

[F1]

Curve, orders and pole divisors: closed points of C have order functions ord⁡x on k(C)×; a divisor is a finite formal integral combination of closed points; the positive and negative parts of div⁡(f) are (f)0 and (f)∞, and div⁡(f)=(f)0−(f)∞ (Curves over a field, Divisors on a smooth proper curve, Divisor support positive negative parts).

[F2]

The Riemann-Roch space: L(D)={g∈k(C)×:div⁡(g)+D≥0}∪{0} is a k-subspace of k(C), characterized coefficientwise by ord⁡x(g)+nx≥0; the promised section dictionary identifies L(D)=H0(C,OC(D)) and attaches to g∈L(D) the section sg with div⁡(sg)=div⁡(g)+D, so that the ratio sg1/sg0 of two sections of OC(D) is the rational function g1/g0 (The space L(D)).

[F3]

Base points: a closed point x is a base point of a subspace V⊆L(D) when every nonzero g∈V has x in the support of div⁡(g)+D, and V is base-point-free when it has no base point; the base-point-free condition is exactly the surjectivity of the evaluation morphism of any basis (Base points and base-point-free linear systems).

[F4]

The base-point-free morphism: a base-point-free subspace V⊆L(D) of dimension r+1≥1 carries a k-morphism φV:C→Pkr, well defined up to the projective-linear action of PGLr+1(k) on the target, with OC(D)≅φV∗O(1) under which the coordinate sections pull back to the sections of V (in particular φV♯(x1/x0)=s1/s0 for a chosen basis s0,s1 when r=1); conversely, a morphism φ:C→Pkr together with an isomorphism α:φ∗O(1)→OC(D) has base-point-free span Vφ∋α(φ∗xi), and the morphism attached to the data is φ (A base-point-free linear system defines a morphism to projective space).

[F5]

The morphism of a nonconstant function: a nonconstant f∈k(C)× defines a finite locally free k-morphism φf:C→Pk1 with φf♯(t)=f, of degree [k(C):k(f)]≥1, whose fibre over infinity is the pole divisor (f)∞; and with A=(f)∞ one has OC(A)≅φf∗O(1) (A nonconstant rational function defines a finite map to the projective line, Finite morphisms from a curve to the projective line).

[F6]

The projective line: Pk1 is a smooth proper geometrically integral curve of genus 0 with affine coordinate t=x1/x0, origin [1:0] and point at infinity ∞=[0:1]=V(x0); the coordinate section x0 of O(1) satisfies div⁡(x0)=[∞] and O(1)≅O([∞]) with deg⁡kO(1)=1, while div⁡(x1)=[ [1:0] ] (Divisors on the projective line are classified by degree, Relative projective space from standard charts).

[F7]

Existence of functions with a bounded single pole: for every closed point p of residue degree d=[κ(p):k]≥1 and every n≥1 with nd+1−g≥2 there is a nonconstant f∈L(np), every pole of which lies at p with order at most n (Rational functions with poles bounded at one point).

[F8]

The Axiom of Choice is available and is inherited only through the suppliers of [F2], [F4], [F5] and [F7]; the example selects nothing beyond the given curve, point and function (The Axiom of Choice).

Proof

technique · compute the two divisors $\operatorname{div}(1)+D$ and $\operatorname{div}(f)+D$ for $D=(f)_\infty$ and for the larger $np$, apply the base-point-free morphism theorem and its converse, and carry out the two explicit projective-line computations
1.1F1F2

The two sections and their divisors. Since f is nonconstant, 1 and f are linearly independent in k(C); since f∈L(np) and 1∈L(np) for n≥m by [F2] (as div⁡(f)+np≥0 and np≥0), they span a two-dimensional subspace of L(np) and of L(D0). By [F1] the pole divisor of f is (f)∞=m[p] with m≥1, and the divisor attached to 1 and f in L(np) is div⁡(1)+np=np and div⁡(f)+np=(f)0−m[p]+n[p]=(f)0+(n−m)[p].

1.2F2F3F4F6

The same divisor with two different subspaces. On Pk1 let D=2[∞]. The two subspaces V1=k⋅1+k⋅t2 and V2=k⋅1+k⋅(t2+t) of the single space L(2[∞]) are two-dimensional. Both are base-point-free: by [F2] and [F6], div⁡(1)+2[∞]=2[∞] and div⁡(t2)+2[∞]=2[0], and div⁡(t2+t)+2[∞]=[0]+[−1] (as div⁡(t2+t)=[0]+[−1]−2[∞]), so in each pair the constant 1 is a unit away from infinity and the second section is a unit at infinity; by [F3] there is no base point. By [F4] each subspace attaches a morphism C→Pk1, and the pullback of the coordinate is the ratio of the two basis sections: φV1♯(t)=t2 and φV2♯(t)=t2+t. If the two morphisms were related by the projective-linear action of PGL2(k) on the target, some M(u)=au+bcu+d would satisfy M(t2)=t2+t; clearing denominators, at2+b=(t2+t)(ct2+d)=ct4+ct3+dt2+dt, so comparing coefficients gives c=0, then d=0, then a=b=0, contradicting that M is invertible. Hence φV1 and φV2 are not related by the target action: the morphism attached to a base-point-free subspace of L(D) depends on the subspace, not only on D.

2.1F2F3step 1.1

Base-point-freeness at the pole divisor. Take n=m and D0=m[p]. By step 1.1, div⁡(f)+D0=(f)0, which does not contain p, so the section f does not vanish at p; and div⁡(1)+D0=D0=m[p] is supported at p, so the constant section 1 does not vanish at any other point. By [F3] no point of C lies in both divisors, so V0=k⋅1+k⋅f⊆L(D0) is base-point-free of dimension two.

3.1F2F4F5step 2.1

The attached morphism is φf. By [F5] there is a finite locally free morphism φf:C→Pk1 with φf♯(t)=f, of degree [k(C):k(f)], whose fibre over infinity is (f)∞, and an isomorphism OC(D0)≅φf∗O(1). Under the section dictionary [F2] the pullbacks φf∗x0 and φf∗x1 correspond to the rational functions 1 and f, whose span is V0; the converse clause of [F4], applied with r=1, D=D0 and that isomorphism, gives that V0 is base-point-free — recovering step 2.1 — and that the morphism attached to the data (OC(D0);1,f) is φf. Hence φV0=φf, so φV0 is finite of degree [k(C):k(f)] with fibre over infinity equal to (f)∞.

3.2F2F3F4step 1.1

The enlarged divisors have the base point p. Let n>m. By step 1.1 both div⁡(1)+np=np and div⁡(f)+np=(f)0+(n−m)p contain p, since n≥1 and n−m≥1; by [F3] the point p is a base point of k⋅1+k⋅f⊆L(np), so the hypothesis of [F4] fails and no morphism is attached. Together with step 2.1 this shows that among the divisors np≥(f)∞ the pair 1,f is base-point-free exactly at the pole divisor D0=(f)∞ (n=m).

4.1F2F3F4F6step 3.1

The projective-line identity. Let C=Pk1 and f=t, so that (t)∞=[∞]=D0 by [F6] (the coordinate section x0 has divisor [∞]). By step 3.1 the attached morphism φV0 is φt, finite of degree [k(Pk1):k(t)]=1 with fibre over infinity the single point [∞]; and the converse clause of [F4], applied with φ=idP1, D=[∞] and the isomorphism O(1)≅O([∞]) of [F6] carrying x0↦s1 and x1↦st (matching divisors [∞] and [0] by [F2]), shows that the morphism attached to the data (O([∞]);1,t) is the identity [1:t]. For n≥2, the same two elements of L(n[∞]) have div⁡(1)+n[∞]=n[∞] and div⁡(t)+n[∞]=[0]+(n−1)[∞], both containing ∞, so by [F3] ∞ is a base point of k⋅1+k⋅t⊆L(n[∞]) and no morphism is attached to the larger system.

5.1F5F7F8step 3.1step 4.1∎

Realization and conclusion. By [F7] the example is nonempty: for every closed point p of residue degree d and every n≥1 with nd+1−g≥2 there is a nonconstant f∈L(np) with all poles at p of order at most n, and the construction above attaches to the pole divisor (f)∞ the base-point-free pencil k⋅1+k⋅f with morphism exactly φf; on Pk1 the case f=t is the simplest instance, in which the only pole is at infinity and φf is the identity. This realizes the construction of Finite morphisms from a curve to the projective line in the single-pole case. The Axiom of Choice declared in [F8] is inherited only through the suppliers of [F2], [F4], [F5] and [F7], and nothing is selected beyond the given curve, point and function.

ExampleConstruction: Literature-sourcedVerification: AI-adaptedprecheck passaudited 2026-10-02Open item page →

A sufficiently positive divisor is nonspecial and Riemann-Roch counts its sections

Example

Assume the Axiom of Choice inherited from the current fixed-direction vanishing, divisor and Riemann-Roch suppliers.

Let k be a field, let C be a smooth proper geometrically integral curve over k of genus g=g(C) (Curves over a field, Genus via the Euler characteristic), let φ:C→Pk1 be a finite k-morphism, and let A be an effective divisor on C with OC(A)≅φ∗OPk1(1). Let D0 be any divisor on C and let n0=n0(D0,φ) be an integer supplied for D0 and φ by the fixed-direction vanishing theorem. Then for every n≥n0 and every effective divisor E the divisor D=D0+nA+E is nonspecial, H1(C,OC(D))=0 (Sufficiently positive divisors in a fixed direction are nonspecial), so Riemann-Roch computes its dimension exactly: l(D)=deg⁡k(D)+1−g (Riemann-Roch as l minus i).

The concrete instance is the projective line. Take C=Pk1, φ=id and A=[∞], so that φ∗O(1)=O(1)≅O([∞]); then g=0 and, for every d∈Z, l(d[∞])=h0(Pk1,O(d))={d+1,d≥0,0,d<0,i(d[∞])=h1(Pk1,O(d))={0,d≥−1,−d−1,d≤−2, while deg⁡k(d[∞])=d⋅[κ(∞):k]=d. Consequently:

  1. for d≥−1 the divisor d[∞] is nonspecial, i(d[∞])=0, and Riemann-Roch reads l(d[∞])=d+1=deg⁡k(d[∞])+1−g — including the boundary case d=−1, where l=i=0 (the sheaf O(−1) has no sections);
  2. for d≤−2 the divisor is special, with l(d[∞])=0 and i(d[∞])=−d−1≥1, and Riemann-Roch reads 0−(−d−1)=d+1;
  3. so in this family nonspeciality holds exactly for d≥−1, and the exact threshold is the boundary value d=−1; the general statement only provides some threshold n0(D0,φ) depending on D0 and on the fixed morphism, which this instance computes exactly.

The "sufficiently positive" threshold of the general statement is the fixed-direction one of the vanishing theorem; the example does not assert the universal bound "deg⁡k(D)>2g−2 implies i(D)=0", which requires Serre duality and belongs to the next pair.

Scaffold repair, recorded for the owner. The frozen scaffold cited the examples-page item ex-cohomology-o-d-projective-line-all-d (on another page) and ex-riemann-roch-projective-line-divisor (on this page) for the projective-line values l(d[∞])=d+1 and the nonspecial range. Both are examples and cannot carry a load; the citations are replaced by the published A-page suppliers Global sections of projective twists, Top cohomology of projective twists and Divisors on the projective line are classified by degree, together with The Picard group of the projective line for the identification O(d[∞])≅O(d) of the attached sheaves. Every promised numerical claim is preserved: l(d[∞])=d+1 for d≥0, nonspeciality exactly for d≥−1, the boundary case d=−1 with l=i=0, and the fixed-direction restriction.

The current Sufficiently positive divisors in a fixed direction are nonspecial supplies the fixed-direction vanishing, and Riemann-Roch as l minus i supplies the dimension identity. The current projective-line divisor and Picard interfaces Divisors on the projective line are classified by degree and The Picard group of the projective line identify the attached twists; the published cohomology corollaries supply their dimensions.

Facts & Assumptions

Given: the Axiom of Choice inherited from the current fixed-direction vanishing, divisor and Riemann-Roch suppliers; a field k, a smooth proper geometrically integral curve C over k of genus g, a finite k-morphism φ:C→Pk1, an effective divisor A with OC(A)≅φ∗O(1), a divisor D0 and an integer n0=n0(D0,φ) supplied by the fixed-direction vanishing theorem for D0; and the instance C=Pk1, φ=id, A=[∞], d∈Z.

[F1]

Fixed-direction nonspeciality: for every n≥n0 and every effective E, the divisor D0+nA+E is nonspecial, that is, H1(C,OC(D0+nA+E))=0; the threshold depends on D0 and on φ and is not a bound in the degree (Sufficiently positive divisors in a fixed direction are nonspecial, Special and nonspecial divisors).

[F2]

Riemann-Roch and the index of speciality: l(D)−i(D)=deg⁡k(D)+1−g with i(D)=h1(D)≥0, and D is nonspecial exactly when i(D)=0, equivalently when l(D)=deg⁡k(D)+1−g; here l(D)=h0(D) and i(D)=h1(D) (Riemann-Roch as l minus i, Genus via the Euler characteristic, The index of speciality i(D), The Riemann-Roch dimension l(D)).

[F3]

Projective-line data: Pk1 is a smooth proper geometrically integral curve of genus 0; the point at infinity has residue degree [κ(∞):k]=1; the coordinate section x0 vanishes exactly at infinity with multiplicity one, so div⁡(x0)=[∞] and O(1)≅O([∞]) with deg⁡kO(1)=1 (Divisors on the projective line are classified by degree).

[F4]

The current The Picard group of the projective line identifies the degree class of every invertible sheaf on Pk1 with a unique twist. Together with the divisor-to-line-bundle interface in Divisors on the projective line are classified by degree, this gives O(d[∞])≅O(d) for every d∈Z.

[F5]

Degree of a multiple: deg⁡k is a group homomorphism on divisors with deg⁡k[x]=[κ(x):k] for a closed point, so with [F3] one has deg⁡k(d[∞])=d (Degree divisor proper curve, Divisors on a smooth proper curve).

[F6]

Cohomology of the twists of the projective line: for n=1, H0(Pk1,O(d))≅k[x0,x1]d for d≥0 and 0 for d<0, while H1(Pk1,O(d))=0 for d≥−1 and has dimension (−d−11)=−d−1 for d≤−2; with [F4] these are l(d[∞]) and i(d[∞]) (Global sections of projective twists, Top cohomology of projective twists, The Riemann-Roch dimension l(D), The index of speciality i(D)).

[F7]

Fixed direction only: the nonspeciality theorem is stated for the fixed ample direction A; it is explicitly not claimed there that every divisor of degree greater than 2g−2 is nonspecial, which requires the duality pair following this page (Sufficiently positive divisors in a fixed direction are nonspecial).

[F8]

The Axiom of Choice is available and is inherited only through the suppliers of [F1], [F2], [F3], [F4] and [F6]; the computation evaluates the given data and selects nothing (The Axiom of Choice).

Proof

technique · apply the fixed-direction nonspeciality theorem and Riemann-Roch to the general divisor $D_0+nA+E$, then compute both sides explicitly on the projective line with the coordinate divisor $d[\infty]$
1.1F1

The general statement. By [F1], for n≥n0 and E≥0 the divisor D=D0+nA+E satisfies H1(C,OC(D))=0.

1.2F3F4

The projective-line instance and its sheaves. Let C=Pk1, φ=id and A=[∞]. By [F3] one has O(1)≅O([∞])=O(A)=φ∗O(1), so this is an instance of the general data. By [F4] the attached invertible sheaves satisfy O(d[∞])≅O(d) for every d∈Z; the identifications use the current divisor/Picard interfaces of [F3] and [F4].

1.3F3F5

The degree. By [F5] and [F3] the residue degree of infinity is 1, so deg⁡k(d[∞])=d⋅[κ(∞):k]=d.

2.1F1F2step 1.1

Riemann-Roch for D. By [F2] the identity l(D)−i(D)=deg⁡k(D)+1−g holds with i(D)=h1(D); by step 1.1 i(D)=0 and D is nonspecial, so l(D)=deg⁡k(D)+1−g.

2.2F6step 1.2

The dimensions of the twists. By [F6] and step 1.2, l(d[∞])=h0(O(d)) is d+1 for d≥0 and 0 for d<0, while i(d[∞])=h1(O(d)) is 0 for d≥−1 and −d−1 for d≤−2. In particular i(d[∞])=0 exactly when d≥−1.

3.1F2F3F6step 2.2step 1.3

Nonspeciality and Riemann-Roch on the projective line. Combine steps 2.2 and 1.3. If d≥−1, then i(d[∞])=0, so d[∞] is nonspecial by [F2], and the identity of [F2] reads l(d[∞])=d+1=deg⁡k(d[∞])+1−g since g=0 by [F3]; at d=−1 this is l=i=0. If d≤−2, then i(d[∞])=−d−1≥1, so d[∞] is special by [F2], while l(d[∞])=0; the identity of [F2] reads 0−(−d−1)=d+1=deg⁡k(d[∞])+1−g. Hence in this family nonspeciality holds exactly for d≥−1, with d=−1 the exact threshold, whereas the general theorem supplies only the threshold n0(D0,φ) along the fixed direction.

4.1F1F7F8step 2.1step 3.1∎

Fixed-direction restriction and choice accounting. The general statement is the fixed-direction form of [F1]: it is a bound along the one ample direction A, depending on D0 and φ, and [F7] records that no universal bound deg⁡k(D)>2g−2 is asserted here; that bound belongs to the duality pair following this page. The Axiom of Choice of [F8] is inherited only through the suppliers of [F1], [F2], [F3], [F4] and [F6] (the fixed-direction vanishing theorem, the divisor–tensor dictionary and the cohomology of projective space), and the computation selects nothing beyond the given curve, morphism, divisor and integer d.

CounterexampleConstruction: Literature-sourcedVerification: AI-adaptedprecheck passaudited 2026-10-02Open item page →

A negative right-hand side does not contradict Riemann-Roch

Counterexample

Assume the Axiom of Choice inherited from the current divisor, cohomology and Riemann-Roch suppliers.

Let k be a field, let m≥1 be an integer, let Pk1 be the projective line over k with point at infinity ∞ and coordinate t, and put D:=−m[∞]. Then deg⁡k(D)=−m<0,l(D)=0,i(D)=m−1, so Riemann-Roch on Pk1 (g=0) reads 0−(m−1)=1−m=deg⁡k(D)+1−g; for m=2 this is the identity 0−1=−1=−2+1. The explicitly refuted readings are:

  1. "l(D)=deg⁡k(D)+1−g for every divisor D": at D=−2[∞] one has l(D)=0 while deg⁡k(D)+1−g=−1, and a negative integer is not the dimension of any k-vector space;
  2. "there exist deg⁡k(D)+1−g independent sections": for every m≥2 the number 1−m is negative, so it cannot count sections, and indeed l(D)=0;
  3. "the Riemann inequality produces sections": the inequality l(D)≥1−m is vacuous for m≥2, since its right-hand side is negative.

There is no contradiction with the vanishing l(D)=0 for deg⁡k(D)<0 (No sections in negative degree): that corollary asserts exactly the value l(D)=0 computed here and is proved without the Riemann inequality, because its nonpositive lower bound cannot ensure a nonzero section. The Euler characteristic l(D)−i(D)=1−m is an integer that may be negative, and the compensation is supplied by H1, whose dimension i(D)=m−1 is not zero as soon as m≥2; the right-hand side of Riemann-Roch is therefore not itself the dimension of a space of sections. At the boundary m=1, where deg⁡k(D)+1−g=0 and l(D)=i(D)=0, the first two numerical readings are consistent. The third reading still fails: the bound l(D)≥0 does not ensure a nonzero section. Thus m=2 is the first failure of the first two readings and the first negative right-hand side; the third reading fails already at m=1.

Scaffold repair, recorded for the owner. The frozen scaffold statement cited the examples-page item ex-cohomology-o-d-projective-line-all-d for the cohomology of the twists, and it also cited the same-page example ex-riemann-roch-projective-line-divisor. Both are examples-page items: the first is homed on the examples page cohomology-of-quasi-coherent-sheaves-on-affine-and-projective-schemes-examples, and an examples-page item may not be consumed by another item, while the second is an example on this very page and likewise cannot carry the load. The two citations are replaced here by the published A-page corollaries Global sections of projective twists and Top cohomology of projective twists (the same values at n=1, with the monomial count d+1 for d≥0 and the rank (−d−11)=m−1 for d=−m≤−2) together with the A-page suppliers Divisors on the projective line are classified by degree and The Picard group of the projective line; every promised claim is preserved.

The current divisor-to-line-bundle route uses Divisors on the projective line are classified by degree and The Picard group of the projective line, while The Riemann-Roch dimension l(D) and The index of speciality i(D) identify the two cohomology dimensions. The published projective-line cohomology corollaries give their explicit values.

Facts & Assumptions

Given: the Axiom of Choice inherited from the current divisor, cohomology and Riemann-Roch suppliers; a field k, an integer m≥1, the projective line Pk1 with point at infinity ∞ and coordinate t, and the divisor D=−m[∞].

[F1]

Projective-line data: Pk1 is a smooth proper geometrically integral curve over k (Curves over a field) of genus 0; the point at infinity has residue degree [κ(∞):k]=1; the coordinate section x0 of O(1) vanishes exactly at infinity with multiplicity one, so div⁡(x0)=[∞] and O(1)≅O(∞) with deg⁡kO(1)=1; and every divisor D′ on Pk1 is linearly equivalent to deg⁡k(D′)[∞], the degree homomorphism on divisor classes being an isomorphism (Divisors on the projective line are classified by degree).

[F2]

Picard group: the degree homomorphism induces an isomorphism Pic⁡(Pk1)→Z under which the class of OPk1(d) corresponds to d, so every invertible sheaf on Pk1 is isomorphic to O(d) for a unique integer d (The Picard group of the projective line).

[F3]

Divisors and degree: a divisor on a smooth proper curve is a finite formal Z-linear combination of closed points, and deg⁡k(D)=∑xnx[κ(x):k] is a group homomorphism on the divisor group (Divisors on a smooth proper curve, Degree divisor proper curve).

[F4]

Dimensions: l(D)=dim⁡kL(D)=h0(C,OC(D)) and i(D)=h1(C,OC(D))=dim⁡kH1(C,OC(D)) are nonnegative integers, and a dimension over a field is nonnegative by definition (The Riemann-Roch dimension l(D), The index of speciality i(D), Finite-dimensional vector space, and its dimension dim⁡FV; infinite-dimensional means having no finite basis).

[F5]

Riemann-Roch: for every divisor D on the curve one has l(D)−i(D)=deg⁡k(D)+1−g with g=g(C) and i(D)≥0, with equality l(D)=deg⁡k(D)+1−g exactly when i(D)=0 (Riemann-Roch as l minus i).

[F6]

Negative degree: if deg⁡k(D)<0 then L(D)=0 and l(D)=0, by the effective-divisor argument; the Riemann inequality is not used, since its right-hand side is nonpositive and cannot ensure a nonzero section (No sections in negative degree).

[F7]

Explicit cohomology of the twists at n=1: H0(Pk1,O(d))≅k[x0,x1]d for d≥0, with monomial basis x0ax1d−a (0≤a≤d) of size d+1, and H0=0 for d<0; H1(Pk1,O(d))=0 for d>−2 and is free of rank (−d−11)=−d−1 for d≤−2 when k≠0 (Global sections of projective twists, Top cohomology of projective twists).

[F8]

Riemann inequality: l(D)≥deg⁡k(D)+1−g for every divisor D (The Riemann inequality).

[F9]

Riemann inequality vacuity and speciality: the inequality is an equality exactly when i(D)=0, so for i(D)≥1 the divisor is special and the inequality is strict (The index of speciality i(D)).

[F10]

The current Divisors on the projective line are classified by degree and The Picard group of the projective line identify the divisor class and attached twist, while The Riemann-Roch dimension l(D) and The index of speciality i(D) identify the cohomology dimensions. These interfaces give OPk1(−m[∞])≅OPk1(−m) and the readings l(D)=h0(O(D)), i(D)=h1(O(D)) used at steps 1.1 and 2.1.

[F11]

The Axiom of Choice is available and is inherited only through the cohomology, divisor and Riemann-Roch suppliers recorded above; the computation below evaluates explicit formulas at the given divisor and selects nothing (The Axiom of Choice).

Proof

technique · compute both sides of Riemann-Roch at $D=-m[\infty]$ from the explicit cohomology of the twists on $\mathbb P^1_k$, then read off the failure of the equality and existence readings for $m\ge2$
1.1F1F2F3F10

The divisor and its degree; the attached sheaf. By [F1] the residue degree of infinity is [κ(∞):k]=1, so by [F3] the degree of D=−m[∞] is deg⁡k(D)=−m⋅[κ(∞):k]=−m<0. Still by [F1], O(1)≅O(∞), and by the current divisor/Picard route [F10] dualizing and tensoring give O(−m)≅O(1)⊗(−m)≅O(−m[∞])=O(D); equivalently, both classes equal −m under the isomorphism Pic⁡(Pk1)≅Z of [F2] with O(1)↦1.

2.1F4F6F7step 1.1

Sections and index of speciality of D. By [F7] with d=−m: the group H0(Pk1,O(−m)) vanishes because −m<0, and H1(Pk1,O(−m)) vanishes for m=1 (as −1>−2) while for m≥2 it is free of rank (m−11)=m−1; thus h0(O(−m))=0 and h1(O(−m))=m−1 for every m≥1. By the isomorphism O(D)≅O(−m) of step 1.1 and the definitions [F4] of l and i, l(D)=h0(O(D))=h0(O(−m))=0,i(D)=h1(O(D))=h1(O(−m))=m−1. Independently [F6] gives l(D)=0 directly from deg⁡k(D)<0 of step 1.1, so the two computations of l(D) agree.

3.1F1F5step 1.1step 2.1

Riemann-Roch at D. By [F5] applied to the divisor D of step 1.1, with g=g(Pk1)=0 from [F1] and the values of step 2.1, l(D)−i(D)=0−(m−1)=1−m=−m+1=deg⁡k(D)+1−g, an identity for every m≥1; at m=2 it reads 0−1=−1=−2+1. The right-hand side 1−m is negative exactly when m≥2, and equals 0 at the boundary m=1.

3.2F4F6F8F9step 1.1step 2.1

The refuted readings and the role of H1. For every m≥2 step 2.1 gives l(D)=0 while deg⁡k(D)+1−g=1−m<0; since a dimension over k is nonnegative ([F4]), the equality l(D)=deg⁡k(D)+1−g fails, and the negative number 1−m cannot be the number of independent sections in L(D). The Riemann inequality [F8] reads 0≥1−m, which is true for m≥2 but vacuous: its right-hand side is negative, so it guarantees no nonzero section, in agreement with l(D)=0. The discrepancy is exactly the index of speciality: by step 2.1, i(D)=m−1≥1 for m≥2, so by [F9] D is special, the Riemann inequality is strict, and the h1 term 0−(m−1)=1−m supplies the negative compensation. For m=1 one has i(D)=0, l(D)=0 and 1−m=0, so the first two numerical readings hold, but l(D)≥0 still ensures no nonzero section. The third reading therefore fails already at m=1, while the first two fail first at m=2, explicitly by D=−2[∞] with l(D)=0, i(D)=1 and 0−1=−1.

4.1F4F6F11step 1.1step 2.1step 3.1step 3.2∎

Conclusion and choice accounting. Step 1.1 computes deg⁡k(−m[∞])=−m and identifies the attached sheaf with O(−m); step 2.1 computes l(D)=0 and i(D)=m−1; step 3.1 verifies the Riemann-Roch identity 0−(m−1)=−m+1; and step 3.2 exhibits the failure of the readings "l(D)=deg⁡k(D)+1−g" and "there exist deg⁡k(D)+1−g sections" for every m≥2, first at D=−2[∞] where the right-hand side is −1, while the negative-degree vanishing l(D)=0 of [F6] remains consistent because the H1 term compensates. No contradiction with the Riemann inequality arises: the inequality gives no positive lower bound in this family: it is 0≥0 at m=1 and 0≥1−m with negative right-hand side for m≥2. The Axiom of Choice is inherited only through the suppliers of [F11]; the computation selects nothing beyond the given field and integer m.

ExampleConstruction: Literature-sourcedVerification: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)audited 2026-10-02Open item page →

The empty divisor, its Euler characteristic and the genus boundary cases

Example

Assume the Axiom of Choice inherited from the current cohomology, dimension and Riemann-Roch suppliers.

Let k be a field, let C be a smooth proper geometrically integral curve over k of genus g=g(C) (Genus via the Euler characteristic), and let D=0 be the empty divisor. Then l(0)=h0(C,OC)=1,i(0)=h1(C,OC)=g, so the Euler characteristic of the structure sheaf is χ(C,OC)=h0(C,OC)−h1(C,OC)=1−g, and Riemann-Roch on the empty divisor reads 1−g=0+1−g, the normalization that makes χ(C,OC)=1−g the definition of the genus. The complete linear system ∣0∣ consists of the single divisor 0 in every genus, so dim⁡k∣0∣=0=l(0)−1. The two boundary genera: for g=0 one has χ=1 with i(0)=0, so the empty divisor is nonspecial — for Pk1 this reads h0(O)=1, h1(O)=0 — while for g=1 one has χ=0 with l(0)=1=i(0), the first and simplest case of a divisor of degree zero that is special, where the Riemann inequality for D=0 is the strict bound 1≥0 with gap i(0)=1. In genus zero the equality case l(D)=deg⁡k(D)+1−g holds at D=0, and it fails at D=0 as soon as g≥1.

Scaffold repair, recorded for the owner. The frozen scaffold cited the examples-page item ex-cohomology-o-d-projective-line-all-d for the projective-line values h0(O)=1, h1(O)=0. Examples-page items are leaves and cannot carry a load; the citation is replaced by the published A-page corollaries Global sections of projective twists and Top cohomology of projective twists together with Divisors on the projective line are classified by degree, which give the same values at d=0 and g(Pk1)=0. Every promised claim is preserved.

The current Riemann-Roch as l minus i supplies the Riemann-Roch identity, The dimension of a complete linear system supplies the dimension formula for ∣0∣, and Complete linear system identifies the linear system. The current structure-sheaf and cohomology suppliers give the empty divisor and genus boundary values.

Facts & Assumptions

Given: the Axiom of Choice inherited from the current cohomology, dimension and Riemann-Roch suppliers; a field k, a smooth proper geometrically integral curve C over k of genus g, and the empty divisor D=0 on C.

[F1]

The structure sheaf: the canonical map k→H0(C,OC) is an isomorphism, so h0(C,OC)=l(0)=1; in particular the zero divisor has a one-dimensional space of sections, spanned by the constant function 1 (Functions on a proper curve, The Riemann-Roch dimension l(D)).

[F2]

The index of speciality and the genus: i(0)=h1(C,OC)=g(C), the genus, which is the dimension of H1 of the structure sheaf (The index of speciality i(D), Genus via the Euler characteristic, The Riemann-Roch dimension l(D)).

[F3]

Euler characteristic: χ(C,OC)=h0(C,OC)−h1(C,OC)=1−g, the Euler characteristic of the coherent sheaf OC (Euler characteristic of a coherent sheaf, Genus via the Euler characteristic).

[F4]

Riemann-Roch: for every divisor D one has l(D)−i(D)=deg⁡k(D)+1−g, with i(D)≥0 and equality l(D)=deg⁡k(D)+1−g exactly when i(D)=0; a divisor is nonspecial exactly when i(D)=0, and special exactly when i(D)≥1 (Riemann-Roch as l minus i, Special and nonspecial divisors).

[F5]

The complete linear system of the empty divisor: ∣0∣={0} is a single point, and dim⁡k∣0∣=0=l(0)−1 for every genus; generally ∣D∣ is nonempty exactly when l(D)≥1 (The dimension of a complete linear system, Complete linear system, Divisors on a smooth proper curve).

[F6]

The projective line: Pk1 is a smooth proper geometrically integral curve of genus 0; for d=0 its twists satisfy H0(Pk1,O)≅k[x0,x1]0≅k and H1(Pk1,O)=0 (Divisors on the projective line are classified by degree, Global sections of projective twists, Top cohomology of projective twists).

[F7]

The Axiom of Choice is available and is inherited only through the cohomology, degree and Riemann-Roch suppliers recorded above; the computation below evaluates the fixed divisor D=0 and selects nothing (The Axiom of Choice).

Proof

technique · evaluate $h^0$ and $h^1$ at the empty divisor, read off the Euler characteristic and Riemann-Roch identity, identify $|0|$, and check the genus-zero and genus-one boundary cases
1.1F1F2

The dimensions at the empty divisor. By [F1] the space L(0) is one-dimensional, spanned by the constants, so l(0)=h0(C,OC)=1; by [F2] the index of speciality is i(0)=h1(C,OC)=g.

2.1F3F4step 1.1

Euler characteristic and Riemann-Roch at the empty divisor. By [F3] the Euler characteristic is χ(C,OC)=h0−h1=1−g, and the Riemann-Roch identity [F4] at D=0, with deg⁡k(0)=0 because the empty divisor has empty support, reads l(0)−i(0)=1−g=0+1−g, that is, 1−g=0+1−g: the genus is exactly the normalization constant that makes this identity hold.

2.2F5step 1.1

The complete linear system of the empty divisor. By [F5] the complete linear system ∣0∣ consists of the single divisor 0, so dim⁡k∣0∣=0=l(0)−1 by step 1.1, in every genus.

3.1F4F6step 1.1step 2.1

The genus boundary cases. If g=0, then step 1.1 gives i(0)=0, so by [F4] the empty divisor is nonspecial, χ=1 by step 2.1, and the identity is an equality; the projective line realizes this with h0(O)=1, h1(O)=0 by [F6]. If g=1, then step 1.1 gives l(0)=1=i(0), so the empty divisor is special by [F4], χ=0 by step 2.1, and the Riemann inequality for D=0 reads l(0)=1≥0=deg⁡k(0)+1−g; it is strict with gap exactly i(0)=1. In genus zero the equality case l(D)=deg⁡k(D)+1−g holds at D=0, and for every g≥1 it fails at D=0 by step 1.1 and [F4].

4.1F2F3F5F7step 1.1step 2.1step 2.2step 3.1∎

Conclusion and choice accounting. The empty divisor has l(0)=1, i(0)=g and Euler characteristic 1−g, Riemann-Roch at D=0 is the identity 1−g=0+1−g, and ∣0∣={0} with dim⁡k∣0∣=0; the empty divisor is nonspecial exactly in genus zero and is the simplest special divisor in genus one. The Axiom of Choice is inherited only through the suppliers of [F7]; nothing is selected.

Sources