Alphabeta Math
TheoremStatement: Literature-sourcedProof: AI-adaptedPipeline-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.

Soergel's Struktursatz: the functor V is fully faithful on projectives

Statement

Assume the Axiom of Choice. Let Oλ be a regular integral block with antidominant projective P=P(w0⋅λ) and Soergel functor V=Hom⁡O(P,−), and let C be the coinvariant algebra, identified with End⁡O(P) by The endomorphisms of the antidominant projective form the coinvariant algebra, so that V takes values in C-modules. Then for every M∈Oλ and every projective Q∈Oλ the canonical map V ⁣:Hom⁡Oλ(M,Q)⟶Hom⁡C-mod(VM,VQ) is an isomorphism. In particular V is fully faithful on the full subcategory of projective objects of the block, and it is faithful on all of Oλ after passing to the quotient by the Serre subcategory of objects killed by V.

Facts & Assumptions

Given: AC (The Axiom of Choice), the regular integral block, its projective-injective P=P(a) for a=w0⋅λ, and V=Hom⁡(P,−).

[F1]

The central action identifies End⁡(P) with the local coinvariant algebra C, with maximal ideal C+ and residue field C (The endomorphisms of the antidominant projective form the coinvariant algebra, The coinvariant algebra of a finite Weyl group). The top multiplication pairing is perfect and the socle of C is one-dimensional (The coinvariant algebra is a complete intersection with one-dimensional socle).

[F2]

P is projective-injective with a once-each Verma flag; V is exact, dim⁡VX=[X:L(a)], every Verma has socle L(a), and VΔ(x⋅λ) is one-dimensional (The antidominant projective and the Soergel functor).

[F3]

Restricted duality fixes simples and interchanges projective covers with injective hulls; enough projectives therefore gives enough injectives (Restricted duality is exact and involutive on O, Category O has enough projectives, Projective covers in O are indecomposable and unique, Fitting decomposition in a finite-length abelian category).

[F4]

Projectives are Verma-filtered, and finite tensor-block functors and wall crossings are exact and biadjoint; every indecomposable projective is a summand of a wall word applied to Δ(λ) (Projectives in category O have finite Verma flags, Translation functors are exact and biadjoint, The wall-crossing functor and its unit and counit, Wall-crossing compositions produce the projective with the top standard label, Indecomposable projective functors are indexed by the Weyl group).

Proof

technique · direct; the annihilator argument follows Soergel, §2.3, Lemma 6 and Struktursatz 9 (preprint pp. 15–16)
1.1F2F3F4algebra

An indecomposable injective has simple socle: embed its finite-length socle into the direct sum of the injective hulls from [F3], extend that embedding to the whole object by injectivity of the target, and note that the kernel is zero because any nonzero finite-length submodule meets the socle. Injectivity of the source splits the embedding; indecomposability and uniqueness of injective hulls then give one hull. Every simple submodule of a Verma-filtered object embeds into some flag factor (take the first flag position it meets), so [F2] forces its label to be a. Hence every projective-injective indecomposable is the injective hull of L(a), and is isomorphic to P. In particular the socle of P is L(a), once.

1.2F1F2algebra

For a finite C-module N, define L(N)=P⊗CN by a finite presentation: the cokernel of the corresponding map between finite sums of P exists in O. Tensor-Hom adjunction gives L⊣V; exactness of V and VP=C give VL(N)≅N by applying V to that presentation. The counit LVM→M becomes an isomorphism under V, so its kernel and cokernel lie in ker⁡V. This supplies the equivalence Oλ/ker⁡V≃C-mod: in the quotient every object is isomorphic to one of the L(N), and adjunction identifies their quotient Hom maps with module maps.

2.1F1F2step 1.1algebra

Choose a linear functional ω:C→C nonzero on the top degree and zero on the other degrees. By [F1], c↦(d↦ω(cd)) identifies C with its vector-space dual as a C-module. For a finite C-module N, Hom⁡C(N,C∗)≅N∗ by evaluation at 1, with inverse ℓ↦(n↦[c↦ℓ(cn)]). Thus C is injective among finite C-modules. The two contravariant functors Hom⁡(−,P) and Hom⁡C(V−,C) are exact. Their canonical comparison is an isomorphism on simples by step 1.1 and [F2]; it is nonzero at L(a) because P↠L(a)↪P is nonzero. Induction on finite length, using the two short exact Hom sequences, gives the comparison for all sources and target P.

2.2F1F2F4F5step 1.1algebra

The unique maximal-weight Verma factor of P lifts to an embedded Δ(λ) with Verma-filtered quotient. To see this, choose its position in a Verma flag. Earlier factors have no weights above λ or at λ, so a lift of its highest vector is killed by n+. The universal Verma map is injective because its composite to that flag factor is the identity. Its image splits that stage, and quotienting the full flag removes just this factor. Every Verma is annihilated by C+ under the central action [F1], so Δ(λ)⊆K:={v∈P:C+v=0}. If K/Δ(λ)≠0, it is a submodule of the Verma-filtered quotient; its socle contains L(a) by the flag argument in step 1.1. Exactness of V would give dim⁡VK≥2, since dim⁡VΔ(λ)=1. But inclusion into P identifies VK with a subspace of the one-dimensional socle of VP=C: every map P→K has image killed by C+. This is impossible. Thus K=Δ(λ).

3.1F1F2step 2.1step 2.2

The comparison for target P in step 2.1 intertwines the central action. Its subspace of maps annihilated by C+ is exactly Hom⁡(M,K) on the category-O side and Hom⁡C(VM,soc⁡C) on the module side. Step 2.2 and exactness identify these subspaces with Hom⁡(M,Δ(λ)) and Hom⁡C(VM,VΔ(λ)). This proves the comparison for the dominant Verma without an invalid induction for merely left-exact functors.

4.1F4step 1.1step 3.1step 1.2algebra

A wall word F and its adjoint word G send P to projective-injective objects; by step 1.1 these are finite sums of P. Therefore VX=0 implies VFX=Hom⁡(GP,X)=0, and likewise for G. The exact adjunction descends to the Serre quotient: its unit and counit induce maps there, and the triangle identities descend as equalities. Under step 1.2 it is an adjunction between the induced C-module functors. Combining the original adjunction, step 3.1 and this descended adjunction gives the canonical chain Hom⁡(M,FΔ(λ))≅Hom⁡(GM,Δ(λ))≅Hom⁡C(VGM,VΔ(λ))≅Hom⁡C(VM,VFΔ(λ)).

5.1F4step 1.2step 4.1∎

Every projective is a summand of a finite sum of the wall words in [F4]. The natural comparison in step 4.1 respects the summand idempotents, so restriction gives the required isomorphism for every projective target Q and every source M. Step 1.2 also proves faithfulness on the quotient. No claim of full faithfulness on all original objects is made.

Depends on

Used by

Dependency tree · two levels

115 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