Alphabeta Math
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.

Schemes Subschemes and Morphisms Locally of Finite Type — Examples

1 · Prerequisites

2 · Summary

These examples show how gluing produces familiar and nonseparated schemes, why nilpotent structure survives beyond topology, where finite type and finite presentation diverge, and how a dense open immersion has full scheme-theoretic image in an integral target.

3 · Logical flowchart

4 · Definitions, theorems and proofs

None yet.

5 · Examples, counterexamples and false statements

ExampleConstruction: Literature-sourcedVerification: Literature-sourcedaudited 2026-09-07Open item page →

The projective line from two affine charts

Example

Let k be a field. Let U=Spec⁡k[t] and V=Spec⁡k[u]. Their principal opens D(t) and D(u) have coordinate rings k[t,t−1] and k[u,u−1]. The isomorphism u↦t−1 glues U and V along these opens. With two charts there is no nontrivial triple-overlap cocycle; the resulting scheme is, by definition in this construction, the projective line Pk1; equivalently, the notation here names precisely this two-chart gluing.

ExampleConstruction: Literature-sourcedVerification: Literature-sourcedjudge pass (gpt-5.6-terra)audited 2026-09-07Open item page →

The affine line with doubled origin

Example

Let k be a field. Glue two copies of Spec⁡k[t] by the identity on D(t). The two copies of every nonzero point are identified, but the two closed points defined by the maximal ideal (t) remain distinct. This is the affine line with doubled origin; it is retained as the standard later test case for separatedness.

ExampleConstruction: Literature-sourcedVerification: Literature-sourcedjudge pass (gpt-5.6-terra)audited 2026-09-07Open item page →

Two infinitesimal structures at the origin

Example

Inside Spec⁡k[t], the ideals (t) and (t2) define Spec⁡k and Spec⁡k[t]/(t2). Both have the single-point support V(t), but the latter has a nonzero nilpotent class of t, so these are distinct closed subschemes.

ExampleConstruction: Literature-sourcedVerification: Literature-sourcedjudge pass (gpt-5.6-terra)audited 2026-09-07Open item page →

Reduction of the dual-number point

Example

Let k be a field. For A=k[ϵ]/(ϵ2), the nilradical is (ϵ). Hence Spec⁡Ared=Spec⁡(A/(ϵ))=Spec⁡k. The underlying space has one point before and after reduction, but only the former has a nonzero nilpotent.

CounterexampleConstruction: Literature-sourcedVerification: Literature-sourcedjudge pass (gpt-5.6-terra)audited 2026-09-07Open item page →

A closed subset has many scheme structures

Statement refuted

A closed subset of an affine scheme determines its closed subscheme structure.

Facts & Assumptions

Given: A field k.

[F1]

Closed immersions into Spec⁡A are classified, up to unique isomorphism over the target, by quotient rings A/I Closed immersions into affine schemes are quotient spectra.

Counterexample

technique · direct
1.1given

The ideals (t) and (t2) in k[t] have the same radical (t), so their quotient-spectrum closed immersions have the same underlying closed subset V(t).

2.1F1step 1.1∎

The ring k[t]/(t) is reduced, whereas the nonzero class of t in k[t]/(t2) is nilpotent. Their quotient rings are therefore not isomorphic, so [F1] gives distinct closed subscheme structures with the same support.

CounterexampleConstruction: Literature-sourcedVerification: Literature-sourcedjudge pass (gpt-5.6-terra)audited 2026-09-07Open item page →

An infinite disjoint union is locally but not globally finite type

Statement refuted

Every locally finite-type morphism is finite type.

Facts & Assumptions

Given: A field k.

[F1]

A finite-type morphism is locally of finite type and quasi-compact Locally finite type and finite type morphisms.

Counterexample

technique · direct
1.1given

Let X=∐m≥1Ak1 and map it to Spec⁡k. Each component is an affine finite-type chart, so the map is locally of finite type.

2.1F1step 1.1∎

The inverse image of the one-point base is covered by the open components, and no finite subfamily covers X. Thus the map is not quasi-compact; by [F1] it is not finite type.

CounterexampleConstruction: Literature-sourcedVerification: Literature-sourcedjudge pass (gpt-5.6-terra)audited 2026-09-07Open item page →

Finite type need not mean finite presentation

Statement refuted

Every finite-type morphism is of finite presentation.

Facts & Assumptions

Given: Let k be a field, A=k[x1,x2,…], and I=(x1,x2,…). The ideal I is not finitely generated: a finite list of its elements involves only finitely many variables and cannot generate a later variable.

[F1]

A finite-type morphism is locally defined by finite-type ring maps Locally finite type and finite type morphisms.

[F2]

A locally finite-presentation morphism is locally defined by finitely presented ring maps Locally finite presentation morphisms.

Counterexample

technique · direct
1.1F1given

The quotient map A→A/I is generated as an A-algebra by the empty set, so it is of finite type and its affine scheme morphism is finite type.

2.1F2step 1.1algebra∎

The source is the single point corresponding to I. Every affine target neighbourhood contains some D(f) with f∉I. Modulo I, such an f is a nonzero scalar. The vector space I/I2 has basis given by the classes of the xi, and localization at f leaves it infinite-dimensional because f acts on it by that nonzero scalar. Hence If/If2≅(I/I2)f is not finitely generated over Af/If≅k, so If itself is not finitely generated. Thus Af→Af/If is not finitely presented on any target neighbourhood of the source point. By [F2], the affine morphism is not locally of finite presentation and therefore not of finite presentation.

ExampleConstruction: Literature-sourcedVerification: AI-adaptedjudge pass (gpt-5.6-terra)audited 2026-09-07Open item page →

The scheme-theoretic image of a dense open immersion

Example

Let X be an integral scheme and let j:U↪X be a dense open immersion. Then the scheme-theoretic image of j is X itself. This does not assume that j is quasi-compact.

Facts & Assumptions

Given: An integral scheme X and a dense open immersion j:U↪X.

[F1]

Every nonempty affine open of an integral scheme is the spectrum of a domain Integral schemes.

[F2]

The scheme-theoretic image, when it exists, is the smallest closed subscheme through which the morphism factors Scheme-theoretic image.

Verification

technique · direct
1.1given

Let Z↪X be a closed subscheme through which j factors, and let I be its ideal sheaf. On a nonempty affine open V=Spec⁡A, every a∈I(V) restricts to zero on the dense open U∩V.

2.1F1step 1.1choose

Choose a nonempty principal open D(f)⊆U∩V. By [F1], A is a domain and f≠0; the equality a/1=0 in Af gives fna=0 for some n, hence a=0. Thus I∣V=0.

3.1F2step 2.1∎

The same holds on every nonempty affine open, while the empty opens carry only the zero ideal. Hence I=0, so every closed factorization of j contains X itself. By [F2], the smallest such factorization is X.

Sources