Alphabeta Math
LemmaStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedprecheck passjudge pass (gpt-6-luna)verified 2026-11-05 (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.

Finite partial coinvariants on a fixed pointed compact curve

Statement

Assume full AC as in compact-curve cohomology. Let C be a finite disjoint union of compact Riemann surfaces with distinct incoming marks x1,...,xN and outgoing marks y1,...,yM. Every component contains an incoming mark. Fix local coordinates at every mark and a tuple a=(a1,...,aM) of nonnegative integers. Let V be C2-cofinite and of CFT type, and W an admitted grading-restricted generalized joint module at the incoming marks, with no tensor-factorization assumption.

Write VC for the state bundle and D=∑xi. Define its outgoing lattice V(C,a) by retaining VC away from the outgoing marks and taking the O-submodule spanned by θj(aj+r)v for homogeneous v in Vr at yj. This is a filtered union of finite-rank bundles. Let Ja be the span of the incoming residue actions on W of global sections of V(C,a)⊗KC with finite poles only at D. Then W/Ja is finite dimensional. In particular ordinary coinvariants, obtained with no outgoing marks, are finite dimensional. This is a fixed-curve assertion; no relative coherence, analytic sewing or representability is inferred.

Facts & Assumptions

Given: Exactly the curve, marks, coordinates, module and pole lattices in the Statement.

[F1]

Finite generation of C2-cofinite generalized joint modules gives finite generation and a compatible nonnegative multigrading with finite pieces on W. Modes of a weight-r state have component degree r-n-1.

[F2]

Difference-one spanning and finite degree spans for cyclic weak modules gives strict-index spanning with a finite positive-weight representative set A of V/C2(V), and a generator cutoff. Component actions commute.

[F3]

Virasoro modes give an Aut O action with field covariance gives the finite state bundles, with positive Virasoro modes lowering state weight and diagonal scaling by the rth power in state weight r. Formal field and residue covariance for generalized VOA modules makes the residue action coordinate covariant.

[F4]

The Riemann-Roch theorem on a compact Riemann surface supplies a canonical divisor representing KC. Serre duality on a compact Riemann surface says H1(O(B)) is dual to H0(K(-B)) on each compact component, and these spaces are finite dimensional. A negative-degree line bundle has no nonzero holomorphic section. Long exact sequence of sheaf cohomology supplies the exact lifting sequence. The same theorem applies to finite extensions by filtering into line-bundle quotients.

Proof

1.1F1F2algebra

Choose finitely many incoming generators homogeneous for F1's auxiliary multigrading. For each component action and generator choose the truncation cutoff on the finite set A. Commute distinct component blocks and apply F2 successively to them. This spans W by products of strict-index A-words in the separate components on those generators, with a uniform upper index cutoff in each component. This reduction is valid even when W does not split: commute all other blocks past the block being reduced, apply F2 to that component's cyclic submodule of the generator, and commute back. If A is empty, only the empty words remain, so the finite incoming generator set spans W; hence W and its quotient W/Ja are finite dimensional directly.

1.2F3givenconstruct

Let E be the maximum of {0}∪{wt(a):a∈A}, so E=0 if A is empty and the bundle and vanishing calculations below remain defined. Set FE=V(C,a),≤E⊗KC. This is a finite-rank holomorphic bundle. To see its filtration explicitly, the state-weight filtration of F3 has graded weight-r quotient Vr⊗KC(−r). At the outgoing points the lattice definition twists this quotient by O(−∑j(aj+r)yj). Thus FE has successive quotients finite direct sums of KC1−r(−∑j(aj+r)yj),0≤r≤E. The outgoing elementary modifications are locally free because their displayed lattice frames are powers of a local parameter times a vector-space basis; the filtration quotients have the stated frames.

2.1F1F2step 1.1algebra

Fix K. The spanning strings all of whose mode indices are at least -K form a finite set: their indices lie in the finite interval from -K to the generator cutoff, are strict within each component, and their states lie in A. Thus their total auxiliary degrees have a maximum nu(K), enlarged to include all generator degrees and zero. A vector homogeneous of degree n>ν(K) is therefore a sum of homogeneous terms a(−l)(i)w° with a in A and l>K. Indeed in some component the first, smallest strict index is less than -K; move its mode past the other components to the far left. Its remaining tail has degree n-wt(a)-l+1.

2.2F4step 1.2algebra

On each component deg D>0. After tensoring every quotient in step 1.2 with O(kD), its degree tends to positive infinity as k tends to infinity. Choose k0 so that all these finitely many line quotients have degree greater than the degree of KC for every k≥k0, on every component. F4 then gives zero H1 for every quotient, hence H1(C,FE(kD))=0 for k≥k0 by induction through the finite filtration. This establishes the exact vanishing needed below from line duality; no unnamed vector-bundle vanishing theorem is invoked.

3.1F3F4step 2.2construct

Choose K≥E+k0 and n>ν(K). It suffices by step 2.1 to reduce a term w=a(−l)(i)w°, with l>K and r=wt⁡(a)≥1. The finite quotient sheaf FE(lD)/FE(k0D) is supported on the incoming marks. In it prescribe the principal part ati(−l)dti at xi and zero at the other marks, setting all intermediate coefficients below order -k0 to zero. The long exact sequence and H1(C,FE(k0D))=0 lift this prescription to a global sigma in H0(C,FE(lD)). Consequently its expansion at xi is ati(−l)dti plus a state-valued one-form of pole order at most k0, while at every other incoming point it has pole order at most k0. All its state coefficients have weight at most E.

4.1F1F3step 3.1algebra

By definition sigma acting on w° belongs to Ja. Its incoming residue action equals w plus a remainder wδ. Each remainder term increases total auxiliary degree by at most E+k0-1: a coefficient vtpdt with p≥−k0 gives mode v(p), of degree wt(v)-p-1. Although a regular Taylor tail is infinite, only finitely many of its terms act on the fixed w°, by truncation on the finite state basis of V≤E. Since w° has degree n-r-l+1, the remainder has degree at most n+E+k0−r−l<n. Hence w is congruent to a sum of strictly smaller-degree vectors modulo Ja.

5.1F1step 4.1algebra∎

The auxiliary grading is nonnegative. Induction on n now reduces every vector to W≤ν(K) modulo Ja. This is a finite sum of finite-dimensional grading pieces by F1, so it is finite dimensional. Its image surjects onto W/Ja, proving the Statement. The argument did not require Ja to be graded: every reduction used an explicit finite remainder of smaller input degree. Taking M=0 gives the ordinary fixed-surface case.

Depends on

Dependency tree · 2 levels

Sources