Alphabeta Math
LemmaStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedprecheck pass
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.

Rouquier complexes satisfy the three-term braid relation

Statement

For 1≤i≤n−2, with s=si and t=si+1, Fi⊗RFi+1⊗RFi ≃ Fi+1⊗RFi⊗RFi+1in Kb(Re-grmod), with no grading shift. More precisely, expanding the two total complexes gives 8-term complexes; using Bi⊗RBi≅Bi(1)⊕Bi(−1) and the rank-two decompositions BiBi+1Bi≅Bi,i+1,i⊕Bi and Bi+1BiBi+1≅Bi,i+1,i⊕Bi+1 with no shift on any summand, each complex is a direct sum of a contractible summand whose extra bimodule is Bi, respectively Bi+1, and which has an invertible differential block, and a surviving complex built from Bi,i+1,i; cancelling the contractible summands is Gaussian elimination, and the two survivors have the same terms and shifts, with differentials identified by the degreewise sign isomorphism (1,1,−1,1) written below. Thus the displayed homotopy equivalence holds. The chain maps and contractions are written explicitly.

Facts & Assumptions

Given: Adjacent indices i,i+1 with 1≤i≤n−2, the complexes Fi,Fi+1 of The positive and negative Rouquier generator complexes and the Bott–Samelson products BiBi+1Bi=Bi⊗RBi+1⊗RBi, Bi+1BiBi+1.

[F1]

Rank one. Bi⊗RBi≅Bi(1)⊕Bi(−1) and Bi+1⊗RBi+1≅Bi+1(1)⊕Bi+1(−1), with the summands the middle-slot idempotent images of the decomposition R=Rs⊕αsRs, respectively R=Rt⊕αtRt. (The rank-one Soergel bimodule square splits).

[F2]

Rank two. BiBi+1Bi≅Bi,i+1,i⊕Bi and Bi+1BiBi+1≅Bi,i+1,i⊕Bi+1 with no additional shift, where Bi,i+1,i=R⊗RWi,i+1R(3) is the rank-two longest bimodule (Rank-two type-A Soergel bimodule decompositions, The rank-two longest type-A Soergel bimodule).

[F3]

Gaussian elimination. An invertible differential block φ:U→V in a fixed biproduct decomposition of two adjacent terms of a cochain complex can be cancelled: the complex is homotopy equivalent to the reduction obtained by deleting U,V and replacing the differential by its Schur complement, and the deleted part is the contractible two-term complex [U→φV] (Gaussian elimination splits a contractible two-term complex, An invertible cochain differential block and its candidate reduction).

[F4]

Totalization. The signed tensor totalization of bounded complexes is associative up to the canonical degree-zero reassociation and has Koszul differential d(x⊗y)=d(x)⊗y+(−1)px⊗d(y); a shift on a factor is a shift on the tensor product with the same totalization differential (Bounded graded bimodule complexes and signed tensor totalization).

Proof

technique · two explicit Gaussian eliminations to a symmetric four-term complex
1.1F4algebra

Write S=Bs, T=Bt, L=Bi,i+1,i, and let ms,mt be multiplication. The total complex K=FsFtFs has terms STS, (TS)(1)⊕(SS)(1)⊕(ST)(1), S(2)⊕T(2)⊕S(2) and R(3) in degrees 0,1,2,3. Call the degree-one terms U,V,W and degree-two terms A,B,C, in this order. The tensor signs give d0=(ms⊗1⊗1,1⊗mt⊗1,1⊗1⊗ms); d1 has blocks U→A=−mt⊗1, U→B=−1⊗ms, V→A=ms⊗1, V→C=−1⊗ms, W→B=ms⊗1, W→C=1⊗mt, and the other blocks zero; d2=(ms,−mt,ms).

1.2F1givenalgebra

Put x=xi, y=xi+1, z=xi+2 and use the coordinate roots βs=x−y, βt=y−z, distinct from the balanced roots in the generator definition. Set Ds(f)=(f−s(f))/(2βs), the coefficient of βs in R=Rs⊕βsRs. Define J(r⊗r′)=−r⊗βt⊗1⊗r′−r⊗1⊗βt⊗r′,p(r⊗f⊗g⊗r′)=rDs(fg)⊗r′. The map p:STS→S is balanced because Ds is Rs-linear and the middle multiplication is Rt-balanced. For J:S→STS, unit insertion into SS is Rs-balanced, and insertion of βt⊗1+1⊗βt into the middle T is a bimodule map: this element commutes with R, by R=Rt⊕βtRt and βt2∈Rt. Both J and p have internal degree zero. Since s(βt)=βt+βs, one has Ds(βt)=−1/2 and hence pJ(r⊗r′)=−2rDs(βt)⊗r′=r⊗r′.

2.1F1F2step 1.2algebra

Define j:L→STS by j(r⊗r′)=r⊗1⊗1⊗r′; invariants in R⟨s,t⟩ slide across all three dividers, so j is balanced and degree zero, and pj=0. The bimodule STS is generated by g0=1⊗1⊗1⊗1 and gx=1⊗x⊗1⊗1: first expand the second middle slot in the Rt-basis {1,z} and slide its invariant coefficients to the first middle slot, then expand that slot in the Rs-basis {1,x} and slide its invariant coefficients left; the remaining z in the second middle slot slides right because z∈Rs. Let e=Jp. With u=x+y−z, balancing gives p(g0)=0, p(gx)=12(1⊗1) and (1−e)g0=g0,(1−e)gx=12(ug0+g0u), since the two inserted βt tensors sum to ug0+g0u−2gx. Thus ker⁡p=im⁡(1−e)=im⁡j. By F2 and pJ=1, ker⁡p and L have equal dimensions in each graded degree; these dimensions are finite because R is a polynomial ring with positive-degree variables. The graded surjection j:L→ker⁡p is therefore an isomorphism. This establishes the specific decomposition STS=j(L)⊕J(S) without assuming splitting maps from the abstract decomposition.

3.1F1F3step 1.1step 1.2step 2.1

In V=(SS)(1) use the coordinate middle decomposition V+⊕V−=S(2)⊕S; replacing the balanced root by its unit multiple βs changes neither summand. Projection to V− is the coefficient map r⊗f⊗r′↦rDs(f)⊗r′. Its composite with the V component of d0 is exactly p, so the block J(S)→V− is pJ=1 and the block j(L)→V− is zero. Cancel this identity pivot by F3. The surviving degree-zero term is L, and its components into U,V+,W are the outer-unit inclusions with signs +,+,+, since d0j(r⊗r′) inserts a middle 1 in each of those terms.

4.1F1F3F4step 1.1step 3.1algebra

The next pivot is V+→A: multiplication on r⊗1⊗r′ sends it to r⊗r′, so it is the identity S(2)→S(2). Its component into C is minus the identity. Cancel this pivot by F3; the Schur complement replaces the C row by the sum of the old C and A rows. The resulting complex H is L⟶(TS)(1)⊕(ST)(1)⟶S(2)⊕T(2)⟶R(3), with dH0=(jts,jst), where jts(r⊗r′)=r⊗1⊗r′ and likewise for jst, and dH1=(−mt⊗11⊗mt−1⊗msms⊗1),dH2=(ms,−mt). The preceding differential retains its U,W components under the elimination, and the following differential retains its C,B components; these are the displayed formulas. Each map has internal degree zero with the written shifts.

5.1step 1.2step 2.1step 3.1step 4.1algebra

For K′=FtFsFt, repeat steps 1.2–3.1 with s,t exchanged; Dt(βs)=−1/2 gives the same identity pivot. The complement calculation follows by interchanging x,z, which negates both coordinate roots and therefore leaves Jp unchanged. Put its survivors into the same order L, (TS)(1)⊕(ST)(1), S(2)⊕T(2), R(3). The resulting H′ has dH′0=dH0, dH′1=−dH1 and dH′2=−dH2, as follows by exchanging s,t in the matrix of step 4.1 and reordering its two rows and columns. Hence the degreewise maps (1,1,−1,1) form an explicit chain isomorphism q:H→H′.

6.1F3step 3.1step 4.1step 5.1∎

For either identity pivot write the differential block as (abc1). The Gaussian chain isomorphism T to the reduced complex plus the identity pair has components Tn=(10c1) and Tn+1=(1−b01), and identity elsewhere. Its retraction is π=pr⁡T, inclusion ι=T−1in⁡ and homotopy h=T−1kT, where k is the identity from the pivot target back to its source and zero elsewhere. For the two eliminations set i=ι1ι2, p=π2π1, h=h1+ι1h2π1, and similarly i′,p′,h′ for K′. F3 gives pi=1H, 1K−ip=dh+hd and the primed identities. Thus Φ=i′qp and Ψ=iq−1p′ are explicit homotopy-inverse chain maps. All entries are the displayed neighboring blocks and identity pivots, with internal degree zero, so no grading shift is introduced.

Remarks

The local splitting calculation uses coordinate roots, as in Libedinsky §§4.3–4.4. These roots differ by unit signs from the balanced roots of the generator definition; the positive differentials are multiplication and do not change. The abstract rank-two decomposition is used only for the graded dimension comparison in step 2.1. The specific splitting maps and both identity pivots are verified here.

Depends on

Used by

Dependency tree · two levels

24 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