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

Pushforward and vanishing for an affine point blowup

Statement

Assume the Axiom of Choice, inherited from the Proj construction and from the cohomology suppliers (The Axiom of Choice). Let A be a commutative ring with 1, let I=(x,y)⊆A be an ideal generated by a regular sequence (x,y) (for example A a regular local ring of dimension two with parameters x,y), let π ⁣:Bl⁡ISpec⁡A→Spec⁡A be the blowup of Blowup of a scheme along an ideal sheaf and let E be its exceptional subscheme. Then the two standard charts X0=D+(xt) and X1=D+(yt) cover Bl⁡ISpec⁡A, the ordered Čech complex of this two-element affine cover computes the cohomology of OBl⁡ISpec⁡A (Cech cohomology computes quasi-coherent cohomology on a separated scheme), and H0(Bl⁡ISpec⁡A,O)=A,H1(Bl⁡ISpec⁡A,O)=0,Hq(Bl⁡ISpec⁡A,O)=0  (q≥2). Consequently π∗OBl⁡ISpec⁡A=OSpec⁡A and Rqπ∗OBl⁡ISpec⁡A=0 for all q>0.

Facts & Assumptions

Given: A commutative ring A, an ideal I=(x,y)⊆A such that x is a nonzerodivisor of A and y is a nonzerodivisor of A/(x), the Rees algebra R(I)=⨁n≥0Intn (Rees algebra sheaf of a finite type ideal), the blowup π ⁣:Bl⁡ISpec⁡A=Proj⁡R(I)→Spec⁡A (Blowup of a scheme along an ideal sheaf), and the relative projective line PA1=Proj⁡A[u,v] with twisting sheaves O(d) (Twisting sheaf on Proj).

[F1]

Affine blowup standard charts and overlaps: For I=(f0,…,fr) and Bi=A[I/fi] the standard opens D+(fit)=Spec⁡Bi cover Bl⁡ISpec⁡A, with overlap identifications given by the ratios (fjt)/(fit).

[F2]

Closed subschemes of projective space and saturated ideals: A homogeneous ideal a⊆A[u,v] determines a closed subscheme V+(a)↪PA1, equal to Proj⁡(A[u,v]/a) under the canonical closed immersion, and V+(a)∩D+(u)=Spec⁡(A[u,v](u)/a(u)).

[F3]

Hypersurface cohomology sequence: For f∈A[u,v] homogeneous of degree d>0 with every dehomogenisation f/ud, f/vd a nonzerodivisor of the corresponding chart ring, the multiplication map OPA1(−d)→⋅fOPA1 and the structure map of the closed immersion i ⁣:V+(f)↪PA1 form a short exact sequence 0→OPA1(−d)→⋅fOPA1→i♯i∗OV+(f)→0, and the induced long exact sequence of sheaf cohomology computes the cohomology of V+(f) from that of the twists OPA1(d).

[F4]

Cohomology of O(d) on projective space: On PA1 one has H0(O)=A, H1(O)=0, Hq(O)=0 for q≥2, and H0(O(−1))=0, H1(O(−1))=0; indeed Hq(PA1,O(d))=0 unless q=0 or q=1, with H0 the degree-d part of A[u,v] for d≥0 and H1 the free A-module on the Laurent monomials ue0ve1 with e0,e1<0 and e0+e1=d, which is zero for d=−1.

[F5]

Cech cohomology computes quasi-coherent cohomology on a separated scheme: For a quasi-compact separated scheme X, a finite affine open cover and a quasi-coherent OX-module F, the canonical comparison map Hˇq(U,F)→Hq(X,F) is an isomorphism for every q≥0.

[F6]

Čech complex for a two-open cover: For a cover by two open sets U0,U1, the ordered Čech complex has C0=F(U0)⊕F(U1), C1=F(U0∩U1) and Cp=0 for p≥2, with δ0(s0,s1)=s1∣U0∩U1−s0∣U0∩U1; hence Hˇ0=ker⁡δ0, Hˇ1=coker⁡δ0 and Hˇp=0 for p≥2.

[F7]

Higher direct images localize over an affine base: For a quasi-compact separated morphism f ⁣:X→S and a quasi-coherent OX-module F, each Rqf∗F is quasi-coherent (Higher direct image of a sheaf), and for every affine open V=Spec⁡A⊆S there is a canonical isomorphism (Rqf∗F)∣V≅Hq(f−1V,F)~.

[F8]

Closed immersions of schemes, Quasi-compact and quasi-separated schemes, Separated morphism of schemes: A closed subscheme of a quasi-compact scheme is quasi-compact, and a closed subscheme of a separated scheme is separated; the relative projective line PA1 is quasi-compact and separated over Spec⁡A.

Proof

1.1F1

The graded A-algebra homomorphism A[u,v]→R(I) with u↦xt, v↦yt is surjective because In=(x,y)n is generated by the monomials xn−iyi, and its kernel is (xv−yu): for a homogeneous F=∑i=0nciun−ivi with F(x,y)=0, reduction modulo x gives cnyn≡0(modx), so cˉn=0 because y is a nonzerodivisor modulo x, say cn=xd; then F−dvn−1(xv−yu) has zero vn-coefficient, hence equals uG for a homogeneous G of degree n−1, and evaluating at (x,y) gives xG(x,y)=0, so G(x,y)=0 because x is a nonzerodivisor, and induction on n down to degree 0 yields F∈(xv−yu), while conversely xv−yu↦x⋅yt−y⋅xt=0; hence R(I)≅A[u,v]/(xv−yu) as graded A-algebras.

1.2F1F2

The element f:=xv−yu∈A[u,v] is a nonzerodivisor: as a polynomial in v over A[u] its leading coefficient is the nonzerodivisor x, so fg=0 forces the top v-coefficient of g to vanish and descending induction gives g=0; the dehomogenisation f/u=x(v/u)−y is a nonzerodivisor of A[v/u] for the same reason, and yU−x is a nonzerodivisor of A[U], since comparing coefficients of Ui in (yU−x)g=0 gives −xg0=0 and ygi−1=xgi for i≥1, so g0=0 and induction gives all gi=0; hence by [F2] the blowup is identified with the closed subscheme Bl⁡ISpec⁡A=Proj⁡(A[u,v]/(xv−yu))=V+(xv−yu)⊆PA1, whose standard charts D+(u)∩V+(f)=Spec⁡(A[v/u]/(x(v/u)−y))=X0 and D+(v)∩V+(f)=Spec⁡(A[u/v]/(y(u/v)−x))=X1 are affine and cover it, matching the standard blowup charts of [F1] under the identification of u,v with xt,yt.

1.3F4

On PA1 the groups H0(O)=A, H1(O)=0, Hq(O)=0 for q≥2 and H0(O(−1))=H1(O(−1))=0 hold by [F4] with n=1.

2.1F3step 1.2

Applying [F3] to the homogeneous degree-one element f=xv−yu, whose chart dehomogenisations are nonzerodivisors by step 1.2, gives a short exact sequence of OPA1-modules 0→OPA1(−1)→⋅fOPA1→i∗OBl⁡ISpec⁡A→0, where i is the closed immersion of step 1.2.

3.1F4step 2.1step 1.3

The long exact cohomology sequence of step 2.1, with Hq(PA1,i∗OBl⁡)≅Hq(Bl⁡ISpec⁡A,O) for the closed immersion i and the groups of step 1.3, gives H0(Bl⁡ISpec⁡A,O)≅coker⁡(H0(O(−1))→H0(O))=A, then H1(Bl⁡ISpec⁡A,O)=0 because it is squeezed between H1(O)=0 and H2(O(−1))=0, and for q≥2 the group Hq(Bl⁡ISpec⁡A,O) is squeezed between Hq(O)=0 and Hq+1(O(−1))=0.

4.1F5F6F8step 1.2step 3.1

By [F8] the blowup Bl⁡ISpec⁡A is a closed subscheme of the quasi-compact separated PA1, hence quasi-compact and separated, so the two-element affine cover of step 1.2 has Čech cohomology computing the sheaf cohomology of O by [F5] and, by [F6], a Čech complex concentrated in degrees 0 and 1 with Hˇ0=ker⁡δ0 and Hˇ1=coker⁡δ0, so its Čech groups are A, 0 and 0 in degrees 0, 1 and ≥2, consistently with step 3.1.

5.1F7F8step 3.1∎

The structural morphism π is quasi-compact and separated, being the composite of the closed immersion of step 1.2 with the quasi-compact separated structure morphism of PA1 from [F8], and OBl⁡ is quasi-coherent, so [F7] with V=Spec⁡A identifies Rqπ∗O with the sheaf associated to the A-module Hq(Bl⁡ISpec⁡A,O), which is A in degree 0 and 0 for q>0 by step 3.1; hence π∗O=A~=OSpec⁡A and Rqπ∗O=0 for every q>0.

Remarks

  • Regularity of the sequence (x,y) enters twice: to identify the Rees algebra with the incidence algebra A[u,v]/(xv−yu) (step 1.1) and to make the dehomogenisations of xv−yu nonzerodivisors (step 1.2). For a general pair of generators the map A[u,v]→R(I) has a nontrivial kernel in general, and the blowup need not be a hypersurface in PA1.
  • The theorem is stated for an affine base precisely so that the direct image computation reduces to the two cohomology modules A and 0; over a nonaffine base the same proof applies over each affine open, and the sheaf statement is the affine-local one of [F7].
  • The two-chart computation never inverts x or y in A: the overlap identification inverts the ratio v/u only, consistent with [F1].

Depends on

Used by

Dependency tree · two levels

93 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources