Alphabeta Math
Pipeline-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.

✓ 8 results · all verified · 1 also independently AI-judged
Every result on this page is machine-checked by a proof checker and read in full and owner-audited; the judge is an additional, independent cross-model AI review of the proofs. The 7 not AI-judged were verified by owner audit (typically over a confirmed judge false positive), not failures.

Singular and Parabolic Kazhdan–Lusztig Theory

1 · Prerequisites

2 · Summary

This page develops the two parabolic Hecke modules attached to a standard parabolic subgroup and the two categories in which their canonical bases appear. The spherical and antispherical modules are introduced together with their standard bases, right actions and bar involutions; the parabolic Kazhdan–Lusztig bases are then constructed by the same triangular induction as the ordinary basis, and their coefficients are identified with the ordinary Kazhdan–Lusztig polynomials: an alternating sum over the Levi Weyl group in the antispherical case, and the longest-element reduction my,x=pw0Jy,w0Jx in the spherical case.

The second half separates two constructions that are often conflated. The singular integral block of category O is treated through the facet tensor-weight exclusion lemma and exact translation functors: standards and costandards translate to standards and costandards, a simple survives exactly when its label is a minimal coset representative, and singular multiplicities are spherical parabolic values at one, with the big antidominant projective carrying every distinct singular standard once. Separately, the genuine locally finite parabolic category is built from a standard parabolic and its Levi subalgebra, proved to be a highest weight category with generalized Verma standards and generalized-standard flags, and its multiplicities are the antispherical values ny,x(1), with the weighted inverse character formula. The finite computations are collected on singular-and-parabolic-kazhdan-lusztig-theory-examples.

3 · Logical flowchart

4 · Definitions, theorems and proofs

DefinitionDefinition: Literature-sourcedProof: Not applicableOpen item page →

The parabolic Hecke module and the antispherical standard basis

Definition

Work with the normalized type-A Hecke algebra Hv(n) over A=Z[v±1] of The normalized type-A Hecke algebra and its bar involution: generators Hs with Hs2=1+(v−1−v)Hs, standard basis {Hw:w∈Sn}, bar involution Hs‾=Hs−1=Hs−(v−1−v) and Kazhdan--Lusztig basis {H‾w} of Existence and uniqueness of the Kazhdan–Lusztig basis, so that H‾s=Hs+v. The underlying Coxeter data are those of The symmetric group Sym⁡(X): the bijections of a set X under composition, Permutation Weyl group and inversion length and The generic type-A Hecke algebra.

Fix J⊆S={s1,…,sn−1}, let WJ=⟨s:s∈J⟩⊆Sn be the standard parabolic subgroup and let ΦJ=Φ∩span⁡R{αs:s∈J} be its standard type-A root subsystem, all in the conventions of Standard parabolic subgroups, descent-free one- and two-sided representatives, parabolic and reflection subgroups.

The two characters. Let HJ⊆Hv(n) be the subalgebra generated by {Hs:s∈J}. For s∈J the quadratic relation makes v−1 and −v the two roots of x2−(v−1−v)x−1=0, so there are well-defined algebra homomorphisms φ+,φ−:HJ→A with

φ+(Hs)=v−1,φ−(Hs)=−v(s∈J).

Viewing A through φ+ (resp. φ−) as a right HJ-module gives the spherical parabolic module MJ:=A⊗HJHv(n) and the antispherical parabolic module NJ:=A⊗HJHv(n), right Hv(n)-modules under right multiplication on the second factor.

Coset representatives. Throughout, Φ+ is the positive system of Φ and ΦJ+=ΦJ∩Φ+, and

JW:={w∈Sn:ℓ(sw)>ℓ(w) for all s∈J},WJ:={w∈Sn:ℓ(ws)>ℓ(w) for all s∈J}

are the left-minimal and right-minimal representatives of the cosets WJw and wWJ, respectively. By Support, intrinsic parabolic presentations, minimal coset representatives and length additivity, with the type-A identification and Intersections of standard parabolics, the parabolic root subsystem, and global minimality of coset representatives every w∈Sn factors uniquely as w=ly with l∈WJ, y∈JW and ℓ(w)=ℓ(l)+ℓ(y), and inversion interchanges JW and WJ.

Standard bases. For x∈JW define mx:=1⊗Hx∈MJ,nx:=1⊗Hx∈NJ. The families {mx:x∈JW} and {nx:x∈JW} are the spherical standard basis and antispherical standard basis, respectively. Their basis property is proved in Standard bases, right actions and bar involutions on the parabolic Hecke modules ↗ (1), by the free left HJ-module decomposition Hv(n)=⨁x∈JWHJHx and tensoring with the character modules.

Remarks

  • Well-definedness of φ±. Assign every generator Hs of the full presented algebra Hv(n) the same scalar v−1, respectively −v. Direct expansion verifies the quadratic relation; both sides of each braid or commutation relation have the same scalar value. Thus the assignments define algebra maps Hv(n)→A. Restricting these maps to HJ gives φ±, without assuming a presentation for the embedded subalgebra.
  • Module structure. For a right HJ-module B, the tensor product B⊗HJHv(n) is a right Hv(n)-module by right multiplication on the second factor; the two modules MJ and NJ differ only in the HJ-action on A, which is trivial (v−1) for MJ and sign (−v) for NJ. The standard basis vectors are defined above; their basis property, right actions and compatible bars are proved in Standard bases, right actions and bar involutions on the parabolic Hecke modules ↗.
  • Normalization. Spherical is φ+(Hs)=v−1 (the trivial character at the parameter of the anti-spherical reference) and antispherical is φ−(Hs)=−v (the sign character). The two are not exchanged by any unit, and the small parabolic S3/S2 exhibits different canonical coefficients; this is why both modules are recorded. The singular-block multiplicities of this page are read in the spherical normalization.
  • Root subsystem. ΦJ is a closed subsystem of the type-A root system and ΦJ+=ΦJ∩Φ+ is its positive system (Standard parabolic subgroups, descent-free one- and two-sided representatives, parabolic and reflection subgroups, Intersections of standard parabolics, the parabolic root subsystem, and global minimality of coset representatives (2)).
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

Standard bases, right actions and bar involutions on the parabolic Hecke modules

Statement

Retain the notation of The parabolic Hecke module and the antispherical standard basis and put mx:=1⊗Hx, nx:=1⊗Hx for x∈JW, and Cs:=H‾s=Hs+v for s∈S. Then:

(1) (Bases.) The sets {mx:x∈JW} and {nx:x∈JW} are A-bases of MJ and NJ respectively. Indeed every w∈Sn factors uniquely as w=ly with l∈WJ, y∈JW, length-additively, and φ+(Hl)=v−ℓ(l) and φ−(Hl)=(−v)ℓ(l) are units of A, so 1⊗Hw=φ±(Hl) (1⊗Hy) in both modules.

(2) (Right action.) For s∈S and x∈JW, mxCs={mxs+v mx,xs∈JW, xs>x,mxs+v−1mx,xs∈JW, xs<x,(v+v−1)mx,xs∉JW,nxCs={nxs+v nx,xs∈JW, xs>x,nxs+v−1nx,xs∈JW, xs<x,0,xs∉JW.

(3) (Bar.) The assignment a⊗h↦aˉ⊗hˉ (bar on A and the bar involution of Hv(n)) descends to A-antilinear involutions of MJ and NJ that satisfy X⋅h‾=X‾⋅hˉ and mx‾=mx+∑y<xay,xmy,nx‾=nx+∑y<xby,xny with ay,x,by,x∈A and top coefficient ax,x=bx,x=1.

(4) (Parabolic property.) If x∈JW and s∈S with xs∉JW, then xs=rx for a unique simple r∈J and xs>x.

Facts & Assumptions

Given: The data of The parabolic Hecke module and the antispherical standard basis: J⊆S={s1,…,sn−1}, WJ, JW, the modules MJ,NJ with φ+(Hs)=v−1, φ−(Hs)=−v, the standard basis {Hw} and the elements Cs=Hs+v; Sn=Sym⁡({1,…,n}), si=(i i+1), and ℓ is inversion length.

[F1]

Parabolic factorization: every w∈Sn is uniquely w=ly with l∈WJ, y∈JW, and ℓ(w)=ℓ(l)+ℓ(y); consequently ℓ(uy)=ℓ(u)+ℓ(y) for u∈WJ, y∈JW (Support, intrinsic parabolic presentations, minimal coset representatives and length additivity, with the type-A identification (3), Intersections of standard parabolics, the parabolic root subsystem, and global minimality of coset representatives (3), Standard parabolic subgroups, descent-free one- and two-sided representatives, parabolic and reflection subgroups (2)). For a reduced expression w=si1⋯sik one has Hw=Hsi1⋯Hsik, and if ℓ(w)=ℓ(u)+ℓ(y) then Hw=HuHy (The normalized type-A Hecke algebra and its bar involution, The standard basis of the generic type-A Hecke algebra).

[F2]

Right multiplication by a generator (The normalized type-A Hecke algebra and its bar involution): HwHs=Hws if ℓ(ws)=ℓ(w)+1, and HwHs=Hws+(v−1−v)Hw if ℓ(ws)=ℓ(w)−1.

[F3]

Bar (The Hecke bar involution is well defined): bar is a multiplicative A-semilinear involution with Hs‾=Hs−1=Hs−(v−1−v), Hw‾=Hw−1−1, and φ±(Hs‾)=φ±(Hs)‾, the conjugate under v↦v−1.

[F4]

Bruhat order (Basic properties of the Bruhat order on Sn (a)-(c)): length-increasing simple multiplication is a Bruhat cover; y≤w implies ℓ(y)≤ℓ(w) with equality only for y=w; a reduced subword of a reduced expression for w represents an element ≤w. An arbitrary subword need not be reduced. Deletion of pairs of letters reduces any word to a reduced word for the same permutation (Support, intrinsic parabolic presentations, minimal coset representatives and length additivity, with the type-A identification (1)).

[F5]

Length and left descents (Support, intrinsic parabolic presentations, minimal coset representatives and length additivity, with the type-A identification (4), Inversions, inversion number, the sign sgn⁡(σ)=(−1)inv⁡(σ), and even and odd permutations): ℓ(x)=inv⁡(x), and for a simple reflection si the left multiple six swaps the values i and i+1 in the one-line notation; the total inversion contribution from pairs involving any third value is unchanged, while the contribution of {i,i+1} changes by one, so ℓ(six)<ℓ(x) if and only if i+1 occurs to the left of i in x, equivalently ℓ(six)>ℓ(x) if and only if i occurs to the left of i+1.

[F6]

Blocks (Support, intrinsic parabolic presentations, minimal coset representatives and length additivity, with the type-A identification (2), Standard parabolic subgroups, descent-free one- and two-sided representatives, parabolic and reflection subgroups (1)): if J decomposes into maximal runs of consecutive indices, then WJ is the direct product of the symmetric groups on the corresponding consecutive value blocks, and by [F5] an element x lies in JW if and only if inside every such block the values occur in increasing order in the one-line notation of x.

Proof

technique · direct: exhibit the free left $H_J$-basis, compute the right action from the standard multiplication rules, prove the parabolic property from the type-$A$ block criterion, and expand bar along subwords
1.1F1F2givenalgebra

The bases. The A-span of {Hu:u∈WJ} is exactly HJ: it contains the generators and is closed under their left and right multiplication by the quadratic multiplication rule, while every Hu is a reduced product of generators in J. By [F1], the standard basis partitions as {HuHy:u∈WJ, y∈JW}. Hence multiplication is an isomorphism of left HJ-modules ⨁y∈JWHJ→Hv(n), (hy)y↦∑yhyHy: expanding each hy in the independent Hu proves injectivity and the unique factorization proves surjectivity. Tensoring this isomorphism with either character module A gives A⊗HJHv(n)≅⨁y∈JWA. The images of the coordinate generators are my, respectively ny, proving both bases.

2.1F2F4step 1.1algebra

The two minimal branches. Let x∈JW, s∈S and suppose xs∈JW. Since xs=(xsx−1)x is the left translate of x by a transposition and ℓ(xs)=ℓ(x)±1, [F4] shows that xs>x holds exactly when ℓ(xs)=ℓ(x)+1, and xs<x exactly when ℓ(xs)=ℓ(x)−1. If xs>x then [F2] gives HxHs=Hxs, so by step 1.1 and the module structure mxCs=1⊗(Hxs+vHx)=mxs+v mx, and likewise nxCs=nxs+v nx. If xs<x then HxHs=Hxs+(v−1−v)Hx, so mxCs=mxs+(v−1−v)mx+v mx=mxs+v−1mx, and likewise nxCs=nxs+v−1nx.

2.2F1F2F4F5F6step 1.1algebra

The parabolic property. Let x∈JW and s=sj, and suppose xs∉JW; by [F5] xs has a left descent in J: some r∈J has r+1 to the left of r in xs. Since x∈JW has no such descent, while xs differs from x only by the swap of the entries at positions j and j+1, the only changed relative order is that of the values a:=x(j), b:=x(j+1), so {r,r+1}={a,b}. The order a=r+1, b=r is excluded because it would already be a left descent of x. Hence a=r and b=r+1 with r∈J; the two permutations xs and srx coincide because both are obtained from x by interchanging the values r and r+1. Moreover ℓ(srx)>ℓ(x) because r∈J and x∈JW, so ℓ(xs)=ℓ(x)+1 and xs>x by F4,(b); thus HxHs=Hxs=Hsrx=HrHx by [F1] and [F2]. Therefore mxCs=1⊗(Hxs+vHx)=φ+(Hr)mx+v mx=(v−1+v)mx and nxCs=φ−(Hr)nx+v nx=(−v+v)nx=0. Uniqueness of r holds because r is the value whose swap produces xs; this proves (4) and the third branch of (2).

2.3F3step 1.1algebra

The bar descent. The map β(a⊗h):=aˉ⊗hˉ is well defined on A⊗HJHv(n): it suffices to check aφ±(t)‾⊗hˉ=aˉ⊗th‾ for t=Hs, s∈J, where the left side is aˉ φ±(Hs)‾⊗hˉ and the right side is aˉ⊗Hs‾hˉ=aˉ⊗(Hs−(v−1−v))hˉ=aˉ φ±(Hs−(v−1−v))⊗hˉ, and φ+(Hs−(v−1−v))=v−1+(v−v−1)=v=v−1‾, φ−(Hs−(v−1−v))=−v+(v−v−1)=−v−1=−v‾ by [F3]; multiplicativity extends this to all t. The map is A-antilinear, involutive (β2=id because bar is an involution on A and on Hv(n)), and satisfies β(X⋅h′)=β(X)⋅h′‾ because (a⊗h)⋅h′=a⊗hh′ maps to aˉ⊗hh′‾=aˉ⊗hˉhˉ′=(aˉ⊗hˉ)⋅hˉ′.

3.1F1F2F3F4step 1.1step 2.3algebra

Triangular expansion and top coefficient. For x=si1⋯sik reduced, [F3] gives Hx‾=∏m=1k(Hsim−(v−1−v)). The term choosing every generator is Hx. Every other term is a scalar times a product of fewer than k generators selected from this word. Successive use of [F2] expands such a product into Hz whose group labels are products of subwords of the selected word: each multiplication retains the old label or appends the current generator. Deletion of pairs in [F4] reduces each such group word to a reduced subword of the original reduced expression for x. Thus z≤x, and ℓ(z)<k because at most k−1 letters were retained, so z<x. Consequently Hx‾=Hx+∑z<xczHz. For each z<x factor z=uy length-additively with u∈WJ, y∈JW. A reduced expression of z contains a reduced suffix for y, so y≤z<x. In either module 1⊗Hz=φ±(Hu)(1⊗Hy). Applying the bar from step 2.3 therefore gives exactly the triangular expansions in (3), with top coefficient one and all lower coefficients in A.

4.1step 1.1step 2.1step 2.2step 2.3step 3.1∎

Parts (1), (2), (3) and (4) are proved in steps 1.1, 2.1 together with 2.2, 2.3 together with 3.1, and 2.2 respectively.

Remarks

The independence of {mx} uses the free left HJ-basis {Hy} of Hv(n) and not merely the spanning property; a unit coefficient alone would not prove independence. The top coefficient one in (3) is the coefficient of Hx in the expansion of the inverse generators; it is not a claim that all lower coefficients lie in vZ[v]. The bar is multiplicative; the balancing check in step 2.3 is exactly where the multiplicative convention (rather than an anti-automorphism convention) is used.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

Existence and uniqueness of the parabolic Kazhdan–Lusztig bases

Statement

Retain the notation of The parabolic Hecke module and the antispherical standard basis and Standard bases, right actions and bar involutions on the parabolic Hecke modules. For every x∈JW there is a unique Cx∈MJ and a unique Dx∈NJ with (i) Cx‾=Cx and Dx‾=Dx, and (ii) Cx∈mx+∑y<xvZ[v] my and Dx∈nx+∑y<xvZ[v] ny, the sums over the lower Bruhat ideal of x inside JW. The sets {Cx}x∈JW and {Dx}x∈JW are A-bases of MJ and NJ. No coefficientwise positivity is asserted.

Facts & Assumptions

Given: The modules MJ,NJ with their standard bases {mx},{nx} indexed by JW, and the bar involutions of Standard bases, right actions and bar involutions on the parabolic Hecke modules; A=Z[v±1].

[F1]

{mx:x∈JW} and {nx:x∈JW} are A-bases of MJ and NJ (Standard bases, right actions and bar involutions on the parabolic Hecke modules (1)).

[F2]

The bar is an A-antilinear involution with my‾=my+∑x<yax,ymx and ny‾=ny+∑x<ybx,ynx, where the sums range over the lower Bruhat ideal of y inside JW and ax,y,bx,y∈A (Standard bases, right actions and bar involutions on the parabolic Hecke modules (3)).

[F3]

Bruhat order on Sn is a finite graded partial order: y<x implies ℓ(y)<ℓ(x), and lower ideals are finite, so the sets {y∈JW:y≤x} are finite and admit induction on length (The Bruhat order on Sn by rank inequalities, Basic properties of the Bruhat order on Sn (b),(e)).

[F4]

A=Z[v±1] and vZ[v]∩v−1Z[v−1]={0} (The normalized type-A Hecke algebra and its bar involution).

[F5]

The ordinary Kazhdan--Lusztig triangular construction over a finite Bruhat ideal, with bar-duality matrix R satisfying RˉR=I, diagonal one and lower-triangular support, produces the unique bar-fixed elements with the prescribed triangular support and shows they form a basis (Existence and uniqueness of the Kazhdan–Lusztig basis, step 1.1 of its proof).

Proof

technique · direct: run the triangular bar-fixed induction of [Existence and uniqueness of the Kazhdan–Lusztig basis](/item/thm-existence-and-uniqueness-of-the-kazhdan-lusztig-basis) over the finite lower Bruhat ideal inside ${}^JW$, for the two modules simultaneously
1.1F1F2algebra

The coefficient matrices. Define rx,y∈A for x,y∈JW by my‾=∑x≤yrx,ymx, so rx,y=ax,y for x<y, ry,y=1, and rx,y=0 when x≰y; define sx,y analogously from ny‾=∑x≤ysx,ynx. Applying bar to my‾=∑x≤yrx,ymx and using my‾‾=my gives my=∑x≤yrx,y‾ mx‾=∑x′≤y(∑x:x′≤x≤yrx,y‾rx′,x)mx′; by [F1] the coefficient of mx′ is δx′,y, that is, RRˉ=I for the unitriangular matrix R=(rx,y). A finite unitriangular matrix is invertible, so also RˉR=I; the same identities hold for S=(sx,y).

2.1F1F2F4F5step 1.1algebra

Construction of the coefficients. Fix w∈JW and increase d=ℓ(w)−ℓ(x) over the finite lower ideal Iw={x∈JW:x≤w} of [F3], starting with cw,w=1. Suppose x<w and cy,w has been constructed for every y∈Iw with x<y≤w, satisfying cy,w=∑y≤z≤wry,zcz,w‾; note that all cited labels lie in Iw. Put ax:=∑x<y≤wrx,ycy,w‾. Then ax‾=∑x<y≤wrx,y‾cy,w=∑x<y≤z≤wrx,y‾ry,zcz,w‾=−∑x<z≤wrx,zcz,w‾=−ax, where the middle equality substitutes the induction equations and the third uses RˉR=I from step 1.1: the sum over x≤y≤z is zero for x<z, and its omitted diagonal term is rx,x‾rx,z=rx,z. Write ax=∑m∈Zγmvm; anti-invariance under v↦v−1 gives γ−m=−γm, hence γ0=0. Define cx,w:=∑m>0γmvm, so cx,w∈vZ[v] and cx,w−cx,w‾=ax. The induction is finite and uses no choice principle.

3.1F1F2step 2.1algebra

Bar invariance and the coefficient equations. Set Cw:=∑y∈Iwcy,wmy. The coefficient of mx in Cw‾ is ∑x≤y≤wrx,ycy,w‾. For x=w this is cw,w‾=1=cw,w; for x<w it is cx,w‾+ax=cx,w by step 2.1, and for x∉Iw the support of rx,y makes the coefficient zero (x≤y≤w would put x∈Iw). Hence Cw‾=Cw and the coefficient equation cx,w=∑x≤y≤wrx,ycy,w‾ holds for every x≤w∈JW; this proves the existence, the bar invariance and the triangular support Cw∈mw+∑y<wvZ[v]my of (i),(ii). Repeating the identical construction with the matrix S in place of R produces Dw∈nw+∑y<wvZ[v]ny with Dw‾=Dw.

4.1F1F4step 3.1algebra

Uniqueness. Suppose h=∑y<wcymy∈MJ is bar-fixed with every cy∈vZ[v]. If h≠0, choose a Bruhat-maximal x in its finite support Iw∖{w}. The coefficient of mx in h‾ is cx‾: no y>x is supported, and rx,y≠0 requires x≤y. Since h=h‾ and the my are independent, cx=cx‾; but bar maps vZ[v] into v−1Z[v−1], so the common value lies in vZ[v]∩v−1Z[v−1]={0} by [F4], a contradiction. Hence h=0: the bar-fixed element with the prescribed triangular support is unique, and likewise in NJ.

5.1F1step 3.1step 4.1algebra∎

Bases. The transition from (mx)x∈JW to (Cx)x∈JW is unitriangular over the finite Bruhat order (each Cx=mx+∑y<xcy,xmy), so it is invertible over A; hence {Cx} is an A-basis of MJ. The same holds for {Dx} in NJ. This proves the theorem for both modules.

Remarks

The construction is the ordinary Kazhdan--Lusztig triangular induction run over the finite poset (JW,≤) with the induced bar-duality matrix; the only inputs are the bar involution, the top coefficient one of the induced bar, and the lower-triangular support. Only the listed facts, hence no coefficientwise positivity, degree or parity statement, enter; those are not asserted here.

DefinitionDefinition: Literature-sourcedProof: Not applicableOpen item page →

The parabolic Kazhdan–Lusztig bases and polynomials

Definition

Retain the notation of The parabolic Hecke module and the antispherical standard basis and write the parabolic Kazhdan--Lusztig bases of Existence and uniqueness of the parabolic Kazhdan–Lusztig bases as

Cx=∑y≤x, y∈JWmy,x my,Dx=∑y≤x, y∈JWny,x ny,x∈JW,

with coefficients my,x,ny,x∈Z[v] supported on the lower Bruhat ideal, mx,x=nx,x=1 and my,x,ny,x∈vZ[v] for y<x.

The spherical parabolic Kazhdan--Lusztig polynomials are My,xJ(q):=v−ℓ(x)+ℓ(y)my,x,q=v−2, and the antispherical (sign) parabolic Kazhdan--Lusztig polynomials are Ny,xJ(q):=v−ℓ(x)+ℓ(y)ny,x. The parity and degree bounds established in Parabolic Kazhdan–Lusztig polynomials are alternating sums of ordinary ones ↗ make these polynomials in q; that reduction theorem discharges the well-definedness asserted here and is recorded in justified_by. Values at v=1 are written my,x(1), ny,x(1).

Normalization dictionary. With Py,xJ(u) the parabolic polynomials of Deodhar (1987) attached to the subset J, one has My,xJ=Py−1,x−1J for u=−1 and Ny,xJ=Py−1,x−1J for u=q, in the index convention of Deodhar and of the anti-spherical category. Thus the spherical family is the u=−1 normalization and the antispherical family the u=q normalization; the two are not related by a unit.

Spherical reduction to ordinary coefficients. If WJ is finite with longest element w0J, then my,x=pw0Jy, w0Jxfor all y,x∈JW, where p is the coefficient of the ordinary Kazhdan--Lusztig basis of Kazhdan–Lusztig polynomials in the classical q-normalization and H‾w=∑y≤wpy,wHy in the normalization of The normalized type-A Hecke algebra and its bar involution.

Remarks

  • Well-definedness. The coefficients my,x,ny,x exist and are unique by the parabolic basis theorem; the triangular support and the vanishing of my,x,ny,x for y≰x are part of that theorem. The assertions that the shifted coefficients are polynomials in q=v−2 (parity and degree bound) are not part of the existence theorem: they are proved in the alternating-sum reduction theorem Parabolic Kazhdan–Lusztig polynomials are alternating sums of ordinary ones ↗, which is the recorded justification of this definition. No parity clause is read back into the existence theorem.
  • Both families are canonical-basis coefficients, not chosen representatives. Cx and Dx are the unique bar-fixed elements with triangular support of the parabolic basis theorem; the symbols my,x and ny,x are their standard-basis coefficients in MJ and NJ respectively.
TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)Open item page →

Parabolic Kazhdan–Lusztig polynomials are alternating sums of ordinary ones

Statement

Retain the notation of The parabolic Kazhdan–Lusztig bases and polynomials.

(1) (Antispherical alternating sum.) For all x,y∈JW, ny,x=∑z∈WJ(−1)ℓ(z) vℓ(z) pzy, x, the sum over all z∈WJ (the product zy taken in Sn, and p⋅,x the ordinary coefficient of the Kazhdan--Lusztig basis of Existence and uniqueness of the Kazhdan–Lusztig basis, so the inner coefficients vanish unless zy≤x).

(2) (Spherical reduction.) If WJ is finite with longest element w0J then for all x,y∈JW, my,x=pw0Jy, w0Jx.

In particular both families are computable from the ordinary Kazhdan--Lusztig coefficients of Sn, with no separate recursion.

Facts & Assumptions

Given: The modules MJ,NJ, the elements Cx=∑y≤xmy,xmy and Dx=∑y≤xny,xny of Existence and uniqueness of the parabolic Kazhdan–Lusztig bases, the ordinary basis H‾w=∑y≤wpy,wHy of Hv(n) in the normalization of The normalized type-A Hecke algebra and its bar involution, and, in part (2), a finite WJ with longest element w0J.

[F1]

Ordinary Kazhdan--Lusztig data: existence and uniqueness of the bar-fixed triangular basis, pw,w=1, py,w=0 unless y≤w, and for y<w the degree bound py,w=vℓ(w)−ℓ(y)+lower and the parity py,w∈vℓ(w)−ℓ(y)Z[v−2] (Existence and uniqueness of the Kazhdan–Lusztig basis); The standard bar expansion is triangular: invert the finite unitriangular transition from the standard basis to the bar-fixed ordinary basis, express Hw in that basis, and apply bar. This gives Hw‾=Hw+∑z<wczHz.

[F2]

Ordinary multiplication (Multiplication by a generator in the Kazhdan–Lusztig basis): if sw<w, then (Hs+v)H‾w=(v+v−1)H‾w, hence HsH‾w=v−1H‾w. If sw>w, then (Hs+v)H‾w=H‾sw+∑sz<z<wμ(z,w)H‾z.

[F3]

Parabolic Hecke module data: {my},{ny} are A-bases, the right actions of Cs are as in Standard bases, right actions and bar involutions on the parabolic Hecke modules (2), (3): the bar is the balanced multiplicative one with triangular expansion.

[F4]

Parabolic basis theorem: the elements Cx,Dx exist uniquely as the bar-fixed triangular elements, and their coefficients satisfy mx,x=nx,x=1 and my,x,ny,x∈vZ[v] for y<x (Existence and uniqueness of the parabolic Kazhdan–Lusztig bases, The parabolic Kazhdan–Lusztig bases and polynomials).

[F5]

Parabolic factorization and Bruhat order: w=uy with u∈WJ, y∈JW is unique and length-additive; if y≤x then y is a reduced subword of every reduced expression of x, and y≤z holds whenever z=uy is length-additive (Support, intrinsic parabolic presentations, minimal coset representatives and length additivity, with the type-A identification, Basic properties of the Bruhat order on Sn, Standard parabolic subgroups, descent-free one- and two-sided representatives, parabolic and reflection subgroups).

[F6]

For finite WJ with longest element wJ=w0J, one has ℓ(swJ)=ℓ(wJ)−1 for s∈J. In type A, wJ reverses each consecutive value block, so these descents follow directly from inversion length. For x∈JW, length additivity gives ℓ(swJx)=ℓ(wJ)−1+ℓ(x)<ℓ(wJx), and [F2] therefore gives HsH‾wJx=v−1H‾wJx.

Proof

technique · compare the antispherical quotient of an ordinary canonical element and the spherical integral left-eigenmodule with their parabolic canonical bases
1.1F3F4F5algebra

The antispherical image. Let Φ:Hv(n)→NJ=A⊗HJHv(n), Φ(h):=1⊗h. It is A-linear and satisfies Φ(hˉ)=Φ(h)‾ by [F3]. Since H‾x is bar-fixed, so is Φ(H‾x); expanding H‾x=∑w≤xpw,xHw and using the parabolic factorization w=zy with z∈WJ, y∈JW and 1⊗Hw=φ−(Hz)ny=(−v)ℓ(z)ny gives Φ(H‾x)=∑y∈JW(∑z∈WJ(−v)ℓ(z)pzy,x)ny. The top coefficient at y=x is px,x=1 (only z=1 contributes, since zx≤x with z≠1 would give ℓ(zx)>ℓ(x)); and for a coefficient at y≠x, every contributing term has zy≤x, hence y≤zy by [F5] and y<x. For y<x, every contributing ordinary coefficient has zy<x: equality zy=x would contradict the unique parabolic factorization of the minimal representative x. Thus pzy,x∈vZ[v], and multiplication by (−v)ℓ(z) preserves this property. Hence Φ(H‾x) is bar-fixed in nx+∑y<xvZ[v]ny, so by uniqueness [F4] it equals Dx and matching coefficients gives (1).

1.2F1F5F6algebra

The integral parabolic eigenline. Put L=ℓ(wJ) and CJ=∑u∈WJvL−ℓ(u)Hu. Pairing u with su where ℓ(su)=ℓ(u)+1, the left multiplication rule gives Hs(auHu+asuHsu)=asuHu+(au+(v−1−v)asu)Hsu. This equals v−1(auHu+asuHsu) iff asu=v−1au. Iterating along reduced words proves that the simultaneous left eigenvectors in HJ are exactly ACJ: their coefficients satisfy au=v−ℓ(u)ae=vL−ℓ(u)awJ. The same pairing on the right gives CJHs=v−1CJ. Bar preserves the left eigen-equations, since conjugating HsX=v−1X gives Hs−1X‾=vX‾ and hence HsX‾=v−1X‾. By [F1] its triangular standard expansion leaves the coefficient of HwJ in CJ‾ equal to one. The integral eigenline calculation therefore forces CJ‾=CJ. Every u∈WJ is a reduced subword of a reduced expression for the block-reversing longest permutation wJ (sort a block permutation up to the reverse order by adjacent ascents); thus u≤wJ. Consequently CJ has the ordinary canonical triangular form, and [F1] identifies it with H‾wJ.

2.1F1step 1.1algebra

Well-definedness and degree bounds for the antispherical family. In the formula of step 1.1, a term with pzy,x≠0 has zy≤x and ℓ(zy)=ℓ(z)+ℓ(y) by [F5], so pzy,x∈vℓ(x)−ℓ(zy)Z[v−2] by [F1]; multiplying by (−v)ℓ(z) gives a contribution in vℓ(x)−ℓ(y)Z[v−2]. Hence ny,x∈vℓ(x)−ℓ(y)Z[v−2], so Ny,xJ(q)=vℓ(y)−ℓ(x)ny,x lies in Z[v−2]=Z[q] with q=v−2, and ny,x has parity ℓ(x)−ℓ(y) modulo two. This discharges the well-definedness of the antispherical polynomials.

2.2F2F3F5F6step 1.2algebra

The integral spherical image. Define Θ:MJ→Hv(n) by Θ(1⊗h)=CJh. Step 1.2 proves balancing and bar compatibility. Decompose Hv(n)=⨁y∈JWHJHy as in [F3]. An element X=∑yhyHy satisfies HsX=v−1X for all s∈J iff each hy does. The integral eigenline calculation of step 1.2 then gives hy=cyCJ with cy∈A. Thus the image of Θ is exactly this integral eigenmodule, and it is injective: the coefficient of HwJy in CJHy=∑uvL−ℓ(u)Huy is one, and no other coset contributes to that standard label. By [F2] in the descent case [F6], H‾wJx lies in this image, with unique preimage Xx=∑ycymy.

3.1F1F4F5step 2.2algebra

Triangularity and coefficient comparison. The disjoint coset expansions in step 2.2 give cy=pwJy,wJx. The coefficient cx is one. If cy≠0, then wJy≤wJx, so y≤wJy≤wJx. Choose a reduced expression for wJx consisting of a word in J followed by a reduced expression for x. By [F5] it contains a reduced subword for y. Such a subword cannot select any letter from the initial J-word: its first selected letter would be in J, making it a left descent of y, contrary to y∈JW. Thus the subword lies entirely in x, proving y≤x. For y≠x, the ordinary coefficient pwJy,wJx lies in vZ[v] by [F1]. Bar compatibility and injectivity give Xx‾=Xx. Therefore Xx has exactly the defining canonical form in [F4], so Xx=Cx and my,x=pwJy,wJx. This proves (2), including zero coefficients outside the lower ideal.

4.1F1step 2.1step 3.1algebra∎

Degree bounds and conclusion. The ordinary degree bound of [F1] gives pw0Jy,w0Jx∈vℓ(x)−ℓ(y)Z[v−2], so my,x has parity ℓ(x)−ℓ(y) and My,xJ(q)=vℓ(y)−ℓ(x)my,x∈Z[q]. Both families are therefore computable from the ordinary coefficients by (1) and (2), and both polynomial families are well defined.

Remarks

The antispherical formula is Deodhar's alternating sum; the spherical formula is the reduction my,x=pw0Jy,w0Jx for finite WJ. The proof uses only the ordinary basis theorem and multiplication formula, the induced module structure, and the comparison of bar-fixed triangular elements; it does not read a parity clause back into the existence theorem of the parabolic basis, and it uses no choice principle.

DefinitionDefinition: Literature-sourcedProof: Not applicableOpen item page →

The singular integral block and its dot stabilizer

Definition

Assume the Axiom of Choice (The Axiom of Choice), inherited from the published block theory. Work in category O of a finite-dimensional complex semisimple Lie algebra g with Cartan h, positive system Φ+, simple roots α1,…,αr, Weyl group W, Weyl vector ρ and dot action w⋅λ=w(λ+ρ)−ρ, as in The classical BGG category O and Dot-Weyl facets and single-wall translation data.

(1) Dot-antidominant integral weights. A weight μ is dot-antidominant if ⟨μ+ρ,α∨⟩≤0 for every α∈Φ+, and integral if ⟨μ,αi∨⟩∈Z for all i (The integral Weyl group of a weight, Integral, dominant, and strictly dominant weights).

(2) The dot stabilizer. Fix such a weight μ and put J=J(μ):={i:⟨μ+ρ,αi∨⟩=0},Wμ:=WJ=⟨si:i∈J⟩. Then the dot stabilizer {v∈W:v⋅μ=μ} equals WJ. Indeed, for i∈J the reflection si fixes μ+ρ, so WJ lies in the stabilizer; conversely the stabilizer of an integral dot-antidominant weight is generated by the reflections it contains and is a standard parabolic (Finite Weyl closed chambers and stabilizers, applied to the dominant vector −μ−ρ), and by Standard parabolic subgroups, descent-free one- and two-sided representatives, parabolic and reflection subgroups a standard parabolic subgroup is WI for the set I of simple reflections it contains, which is exactly J. In particular w′⋅μ=w⋅μ if and only if w−1w′∈WJ, that is, w′∈wWJ.

(3) The singular integral block. The singular integral block Oμ is the block of Central-character summands refine into linkage blocks of the central character χμ of μ; by Central characters are dot-Weyl orbits two central characters agree exactly when the highest weights lie in one dot orbit, and by Generalized central-character decomposition of O the block is a summand of O. Its simple objects L(w⋅μ), standard objects Δ(w⋅μ)=M(w⋅μ), costandard objects ∇(w⋅μ) and projective covers P(w⋅μ) are indexed by w∈W (Standard and costandard objects, Verma modules); by (2) the label w⋅μ depends only on the coset wWJ, so the labels can be normalized by the minimal right coset representatives

WJ={w∈W:ℓ(ws)>ℓ(w) for all s∈J},

which are exactly the elements of minimal length in the cosets wWJ (Support, intrinsic parabolic presentations, minimal coset representatives and length additivity, with the type-A identification (3)); inversion carries WJ onto the minimal left coset representatives JW.

(4) Highest-weight structure and multiplicities. The block is a highest-weight category: it has enough projectives (Category O has enough projectives), projective covers have finite Verma flags (Projectives in category O have finite Verma flags), the standard classes and simple classes are bases of the Grothendieck group (The Grothendieck group and character of O, Simple and standard bases of K0(O)), and it satisfies BGG reciprocity (P(w⋅μ):Δ(y⋅μ))=[M(y⋅μ):L(w⋅μ)] (BGG reciprocity, Finite Verma flags and their multiplicities). The multiplicity matrix is unitriangular for the refinement of the dominance order on the dot orbit: (P(w⋅μ):Δ(w⋅μ))=1, and (P(w⋅μ):Δ(y⋅μ))≠0 only if y⋅μ≥w⋅μ (The triangular restriction on projective Verma flags); the ordinary standard/costandard Hom and Ext1 vanishing used on this page is Standard-costandard Hom and Ext-one orthogonality.

(5) Regular versus singular. The block is regular when J=∅ (the weight lies on no wall) and singular when J≠∅; the singular translation statements of this page use this notation, and J=∅ recovers the regular case. The genuine regular-parabolic multiplicity statements below have their own separate explicit dominant regular base and are not consequences of the notation here.

Remarks

  • Type-A comparison. When W=Sn with its type-A simple reflections, JW is the common standard-basis label set of the two parabolic Hecke modules of The parabolic Hecke module and the antispherical standard basis; the singular labels WJ correspond to that set by inversion. This comparison is not a prerequisite of the general block definition.

  • Choice. The Axiom of Choice is inherited with the published block theory: restricted duality, enough projectives, Verma flags of projectives and the Grothendieck-basis theorem are quoted from items that assume it (The Axiom of Choice, Standard and costandard objects, Category O has enough projectives, Projectives in category O have finite Verma flags, Simple and standard bases of K0(O)), and no new choice principle is introduced here.

  • Highest-weight projective kernels. In an extension 0→M(β)→X→M(α)→0 with α≮β, lift the highest vector of weight α. Every positive-root image lands in the submodule at a weight strictly above α, impossible because that would imply α<β. Thus the lift is highest, and The universal property of Verma modules splits the extension. The projective flag in (4) has its labelled standard once and all other labels strictly higher. The preceding splitting swaps a higher section followed immediately by the labelled submodule section, moving the labelled section to the final quotient. Its kernel therefore has a flag of strictly higher standards, giving the projective-kernel axiom.

  • Stabilizer notation. Here Wμ denotes the dot stabilizer WJ. The integral-reflection group called Wμ in The integral Weyl group of a weight is instead all of W for integral μ. In particular its linkage class is the entire dot orbit, so the integral central-character summand in (3) is one linkage block.

  • No new multiplicity assertion. Everything in (4) is quoted structure of the published block theory; the first genuinely new multiplicity statements are the translation, singular and parabolic multiplicities proved later on this page. In particular no claim is made here that the singular matrix equals an ordinary Kazhdan--Lusztig matrix or a parabolic one.

  • Label conventions. WJ (right minima) is used for the singular block, JW (left minima) for the genuine regular parabolic category later; the two families are exchanged by inversion (Standard parabolic subgroups, descent-free one- and two-sided representatives, parabolic and reflection subgroups (2)), and the small S3/S2 matrices of the companion examples exhibit both label orders.

LemmaStatement: AI-adaptedProof: AI-adaptedprecheck passOpen item page →

The facet tensor-weight exclusion lemma

Statement

Assume the Axiom of Choice (The Axiom of Choice). Let λ be an integral dot-antidominant dot-regular weight and let μ be an integral dot-antidominant weight such that the facet Fμ lies in the closure of the chamber Fλ, i.e. ⟨μ+ρ,α∨⟩=0 for the positive roots α∈ΦJ+ of the parabolic subsystem ΦJ of J=J(μ) defined in The singular integral block and its dot stabilizer, and ⟨μ+ρ,α∨⟩ has the same sign as ⟨λ+ρ,α∨⟩ for every remaining positive root (Dot-Weyl facets and single-wall translation data). Let ν be the unique dominant weight of the linear orbit W(μ−λ) (Finite Weyl closed chambers and stabilizers) and let E=L(ν) (Finite-dimensional simple modules are classified by dominant highest weights). Then for all w,w′∈W and every weight γ of E: if w′⋅μ=w⋅λ+γ, then w′⋅μ=w⋅μ and γ=w(μ−λ).

Consequently, for every w the tensor E⊗Δ(w⋅λ) has exactly one Verma factor whose central character is that of μ, namely Δ(w⋅μ), and its multiplicity in the Verma flag computed by Finite-dimensional tensoring preserves Verma flags is one.

Facts & Assumptions

Given: The Axiom of Choice; integral dot-antidominant weights λ and μ with λ dot-regular and Fμ in the closure of Fλ in the sense of the statement, J=J(μ), the dot stabilizer WJ, the unique dominant weight ν of W(μ−λ), and E=L(ν); write λ∙=λ+ρ and μ∙=μ+ρ.

[F1]

The set J=J(μ)={i:⟨μ+ρ,αi∨⟩=0} is the set of simple indices vanishing on μ∙; the dot stabilizer {v∈W:v⋅μ=μ} equals WJ=⟨si:i∈J⟩, the positive roots vanishing on μ∙ are exactly ΦJ+, and Stab⁡W(μ∙)=Stab⁡W(−μ∙)=WJ (The singular integral block and its dot stabilizer, Finite Weyl closed chambers and stabilizers, Dot-Weyl facets and single-wall translation data). Also λ∙ is regular, −λ∙ is dominant regular and −μ∙ is dominant.

[F2]

Every weight γ of L(ν) satisfies ∣γ∣≤∣ν∣, with equality exactly when γ∈Wν, and every weight of Wν occurs in L(ν) with multiplicity one (Weights of a finite-dimensional simple module lie in the norm ball); moreover ∣ν∣=∣μ∙−λ∙∣ because ν∈W(μ−λ)=W(μ∙−λ∙) and the form is W-invariant (Finite Weyl closed chambers and stabilizers).

[F3]

For dominant ξ,η in the real span of the roots one has ∣ξ−wη∣≥∣ξ−η∣ for every w∈W, with equality exactly when wη∈Stab⁡W(ξ)η (A dominant vector minimises its distance to a dominant weight).

[F4]

The tensor E⊗Δ(λ′) has a finite Verma flag with multiplicities dim⁡Eη−λ′ (Finite-dimensional tensoring preserves Verma flags, Finite semisimple PBW and highest-weight construction).

[F5]

Two weights have the same central character exactly when they lie in one dot-Weyl orbit (Central characters are dot-Weyl orbits).

Proof

Proof technique: transport the tensor-weight identity into a norm comparison, squeeze it to equality, and read off the surviving factor; this is the single-wall proof of The single-wall tensor-weight exclusion lemma with the stabilizer {1,s} replaced by the full dot stabilizer WJ.

1.1givenalgebra

Let w,w′∈W and let γ be a weight of E with w′⋅μ=w⋅λ+γ. Since w′⋅μ−[w⋅λ]=w′(μ∙)−w(λ∙), setting x=(w′)−1w gives xλ∙=μ∙−(w′)−1γ, that is, (w′)−1γ=μ∙−xλ∙.

2.1F1F2F3step 1.1

By [F2] one has ∣γ∣≤∣ν∣=∣μ∙−λ∙∣, and W-invariance of the form gives ∣(w′)−1γ∣=∣γ∣, so the identity of step 1.1 yields ∣μ∙−xλ∙∣≤∣μ∙−λ∙∣. On the other hand [F3] applied to the dominant vectors ξ=−μ∙ and η=−λ∙ (dominant and regular by [F1]) gives ∣μ∙−λ∙∣=∣ξ−η∣≤∣ξ−xη∣=∣μ∙−xλ∙∣.

3.1F1F2F3step 1.1step 2.1algebra

The two inequalities of step 2.1 are equalities, so the equality case of [F3] applies: xη∈Stab⁡W(ξ)η, and Stab⁡W(ξ)=WJ by [F1], that is, x(−λ∙)∈WJ(−λ∙). Choose u∈WJ with xλ∙=uλ∙; then u−1x fixes the regular vector λ∙, so u−1x=1 and x=u∈WJ. Since WJ is the dot stabilizer of μ by [F1], x⋅μ=μ, and therefore w′⋅μ=w⋅(x⋅μ)=w⋅μ. Moreover γ=w′⋅μ−w⋅λ=w⋅μ−w⋅λ=w(μ−λ). Finally ∣γ∣=∣μ∙−λ∙∣=∣ν∣ and γ=w(μ−λ)∈W(μ−λ)=Wν, so by [F2] the weight γ occurs in E with multiplicity one.

4.1F4F5step 3.1algebra

For the reformulation, fix w and let η be a weight in the dot orbit of μ, so η=w′⋅μ for some w′ and η has the central character of μ by [F5]. If dim⁡Eη−w⋅λ≠0, then γ:=η−w⋅λ is a weight of E with w′⋅μ=w⋅λ+γ, so step 3.1 gives η=w′⋅μ=w⋅μ and γ=w(μ−λ); conversely dim⁡Ew(μ−λ)=1 by step 3.1. Hence among labels with the central character of μ only Δ(w⋅μ) occurs, with multiplicity one, in the Verma flag of E⊗Δ(w⋅λ) supplied by [F4].

5.1step 3.1step 4.1∎

Steps 3.1 and 4.1 prove both formulations: the tensor-weight identity forces w′⋅μ=w⋅μ and γ=w(μ−λ) with multiplicity one, and the only Verma factor of E⊗Δ(w⋅λ) with central character χμ is Δ(w⋅μ), once.

Remarks

The Choice assumption is inherited from the published translation and norm items; no new choice principle is used. The lemma is the facet form of the single-wall exclusion: for a parabolic stabilizer WJ the equality case of the dominant norm comparison selects the full stabilizer orbit WJ(−λ∙) instead of the two-element wall orbit, and regularity of λ∙ then pins x inside WJ.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

Translation to the wall on Verma, simple and projective objects

Statement

Assume the Axiom of Choice (The Axiom of Choice). Let (λ,μ) be as in The facet tensor-weight exclusion lemma: λ integral dot-antidominant and dot-regular, μ integral dot-antidominant with Fμ in the closure of the chamber Fλ, J=J(μ), Wμ=WJ. Let ν be the unique dominant weight of W(μ−λ) and let T:=Tλμ=pr⁡χμ∘(L(ν)⊗−)∘incl⁡χλ ⁣:Oλ⟶Oμ,T′:=Tμλ, be the translation functors of Translation functors by tensoring and projection, which are exact and biadjoint (Translation functors are exact and biadjoint). Then:

(1) TΔ(w⋅λ)≅Δ(w⋅μ) and T∇(w⋅λ)≅∇(w⋅μ) for every w∈W.

(2) TL(w⋅λ)≅L(w⋅μ) if w∈WJ, and TL(w⋅λ)=0 otherwise; equivalently TL(w⋅λ)≠0 if and only if w lies in the set WJ of minimal representatives of the cosets wWJ.

(3) (Big projective on the wall.) Let Ξμ:=T−ρμL(−ρ) be the translate of the simple object of the most singular block. Then Ξμ is projective and injective in Oμ, it is the projective cover P(1⋅μ) of L(μ), and (Ξμ:Δ(w⋅μ))=1 for every distinct orbit label (equivalently for its representatives w∈WJ); consequently (P(μ):Δ(y⋅μ))=1 for every distinct label y∈WJ and [Δ(w⋅μ):L(μ)]=1 for all w∈W.

Facts & Assumptions

Given: The facet data (λ,μ) of The facet tensor-weight exclusion lemma with J=J(μ), Wμ=WJ, the translating weight ν and E=L(ν); the functors T=Tλμ, T′=Tμλ of Translation functors by tensoring and projection with E∗=L(ν)∗; write λ∙=λ+ρ, μ∙=μ+ρ.

[F1]

T=pr⁡χμ∘(E⊗−)∘incl⁡χλ and T′=pr⁡χλ∘(E∗⊗−)∘incl⁡χμ; inclusions and projections are exact, and O is the direct sum of its central-character blocks (Translation functors by tensoring and projection, Generalized central-character decomposition of O, Central characters are dot-Weyl orbits).

[F2]

T and T′ are exact and each is both a left and a right adjoint of the other; they send projectives to projectives and injectives to injectives (Translation functors are exact and biadjoint).

[F3]

The unique-factor clause: for every w∈W, the tensor E⊗Δ(w⋅λ) has exactly one Verma factor with central character χμ, namely Δ(w⋅μ), of multiplicity one (The facet tensor-weight exclusion lemma).

[F4]

Tensoring with E preserves finite Verma flags: E⊗Δ(η) has a finite Verma flag with factors Δ(η+γ), γ running over the weights of E with multiplicity dim⁡Eγ; the flag factors of E⊗X for Verma-filtered X are computed by convolution (Finite-dimensional tensoring preserves Verma flags).

[F5]

The block Oμ: J(μ) indexes the vanishing simple coroot pairings of μ∙, WJ is the dot stabilizer of μ, the labels w⋅μ depend only on wWJ, and WJ is the set of minimal right coset representatives (The singular integral block and its dot stabilizer, Dot-Weyl facets and single-wall translation data).

[F6]

Restricted Chevalley duality: D(M)η=Mη∗ with the τ-twisted action; D is an exact contravariant involution of O, preserves weight-space dimensions, simple composition multiplicities and central-character blocks, and D(L(η))≅L(η), D(Δ(η))=∇(η) (Restricted duality is exact and involutive on O, Standard and costandard objects). For finite-dimensional E, the natural tensor-dual identification is D(E)⊗D(M)≅D(E⊗M), with the Chevalley-twisted dual D(E), not the ordinary dual E∗.

[F8]

For a dominant integral weight ν, every weight γ of L(ν) satisfies ∣γ∣≤∣ν∣, with equality exactly when γ∈Wν, and every weight of the orbit Wν occurs in L(ν) with multiplicity one (Weights of a finite-dimensional simple module lie in the norm ball).

[F7]

Each Verma has the unique simple head L(η) (A Verma module has a unique simple quotient), whose dual is the simple socle of its costandard by [F6]. BGG data: a nonzero g-homomorphism M(η)→M(ζ) exists if and only if η↑ζ in the strong linkage order (The BGG criterion for homomorphisms between Verma modules), such a map is injective (A nonzero homomorphism between Verma modules is injective), BGG reciprocity (P(η):Δ(ζ))=[Δ(ζ):L(η)] holds with finitely many nonzero multiplicities (BGG reciprocity, Finite Verma flags and their multiplicities, Projectives in category O have finite Verma flags), and Hom⁡(Δ(η),∇(ζ)) is 0 or one-dimensional according as η≠ζ or η=ζ (Standard-costandard Hom and Ext-one orthogonality). Every object of O has finite length and finite-length objects decompose into indecomposables (The simple objects of O, Fitting decomposition in a finite-length abelian category, Projective covers in O are indecomposable and unique, Category O has enough projectives).

[F9]

Orbit and weight bounds: for dot-antidominant integral η, applying Dominant integral weights are maxima of their Weyl orbits to −η−ρ gives w⋅η≥η for every w. Verma weights lie in η−Q+ and the highest-weight space is one dimensional (Weights of a Verma module lie below lambda). By the central-character and simple-label facts in [F1], [F7], every composition factor of M(η) has its highest weight in W⋅η; the two bounds force that weight to equal η. Exactness of weight spaces and the one-dimensional top then force length one. Thus M(η)=L(η), including singular η and η=−ρ.

[F10]

Finite-Weyl root/length criterion (Finite Weyl strong exchange and deletion): ℓ(wsj)<ℓ(w) iff wαj<0, and ℓ(siv)>ℓ(v) iff v−1αi>0. Positive roots of a parabolic subsystem are nonnegative combinations of its simple roots, so wαj>0 for all j∈J implies wΦJ+⊆Φ+.

Proof

technique · direct: compute $T$ on standards by the unique surviving factor, transport to costandards by restricted duality, decide simple survival by the nonminimal Verma embedding, and manufacture the big projective from the most singular block
1.1F1F3F4algebra

Standards (first clause of (1)). By [F3], the tensor E⊗Δ(w⋅λ) has a finite Verma flag by [F4] in which the unique factor with central character χμ is Δ(w⋅μ), once. The exact projection pr⁡χμ applied to this flag kills every factor whose central character differs from χμ and carries the remaining Δ(w⋅μ) isomorphically, so TΔ(w⋅λ)≅Δ(w⋅μ) for every w.

1.2F1F2F4F5F6F7F8F9F10algebra

The big projective (3). By [F9], M(−ρ)=L(−ρ). Its dot orbit is a singleton, so BGG reciprocity [F7] gives a one-factor Verma flag for P(−ρ), proving P(−ρ)=L(−ρ). Simple-preserving duality makes it injective too. Put E0=L(w0μ∙) and S=T−ρμ, whose adjoint is S′=Tμ−ρ; the ordinary dual E0∗ has highest weight −μ∙. By [F2], Ξμ=SL(−ρ) is projective and injective. Tensoring M(−ρ) with E0 and projecting retains exactly the factors Δ(−ρ+γ) with γ∈Wμ∙, each once by [F4], [F8]. These are precisely the distinct standards Δ(y⋅μ). Likewise, since L(μ)=M(μ) by [F9], tensoring it with E0∗ and projecting to χ−ρ retains only Δ(−ρ), once, from the extremal weight −μ∙. Thus S′L(μ)=L(−ρ), and adjunction gives dim⁡Hom⁡(Ξμ,L(μ))=1. The resulting epimorphism splits off the projective cover P(μ): lift it through the cover, use essentiality to make the lift surjective, and split by projectivity. For every y, there is an injection M(μ)↪M(y⋅μ). Indeed induct on a reduced expression y=siv: [F10] gives v−1αi>0, so ⟨v(μ∙),αi∨⟩≤0. If it is zero the weights coincide; otherwise the positive integral pairing at siv⋅μ supplies the one-step strong linkage and the injection M(v⋅μ)↪M(siv⋅μ) by [F7]. Compose these injections. BGG then gives (P(μ):Δ(y⋅μ))≥1 for every distinct label. The direct-summand complement of P(μ) in Ξμ is projective and has a Verma flag by [F7]. Additivity and the already computed one-per-label flag of Ξμ force every flag multiplicity of the complement to vanish, so the complement is zero. Hence Ξμ=P(μ) and all the displayed standard and composition multiplicities are one.

2.1F5F7F10step 1.1algebra

Nonminimal vanishing (second half of (2)). Suppose w∉WJ; then there is a simple s∈J with ℓ(ws)<ℓ(w), equivalently wαs<0 by [F10]. Then w⋅λ−ws⋅λ=w(λ−s⋅λ)=⟨λ∙,αs∨⟩ wαs with ⟨λ∙,αs∨⟩<0 and wαs<0, so w⋅λ−ws⋅λ∈Q+∖{0}: ws⋅λ is strictly below w⋅λ. Put β=−wαs∈Φ+; then sβw=ws and ⟨wλ∙,β∨⟩=−⟨λ∙,αs∨⟩∈Z>0. This is a one-step strong linkage, so [F7] gives a nonzero (hence injective) map Δ(ws⋅λ)→Δ(w⋅λ) by [F7]. Applying the exact functor T and step 1.1 gives a map Δ(ws⋅μ)→Δ(w⋅μ); now ws⋅μ=w⋅(s⋅μ)=w⋅μ because s∈J fixes μ [F5], so this is an injective endomorphism of the finite-length module Δ(w⋅μ), hence an isomorphism. If N is the cokernel of Δ(ws⋅λ)→Δ(w⋅λ), exactness of T gives TN=0; since L(w⋅λ) is a quotient of N (the kernel of Δ(w⋅λ)↠L(w⋅λ) contains Δ(ws⋅λ) because the latter has no composition factor L(w⋅λ), its weights being strictly below), TL(w⋅λ)=0.

2.2F1F6step 1.1algebra

Costandards (second clause of (1)). The evaluation pairing gives D(E⊗M)≅D(E)⊗D(M) naturally, and D(E)≅E for the finite-dimensional simple E by [F6]. Hence E⊗∇(w⋅λ)=E⊗DΔ(w⋅λ)≅D(E⊗Δ(w⋅λ)). Since D preserves central-character blocks and is exact, it commutes with the inclusion and projection defining the translation functor, so T∇(w⋅λ)=pr⁡χμ(E⊗∇(w⋅λ))≅D(pr⁡χμ(E⊗Δ(w⋅λ)))≅DΔ(w⋅μ)=∇(w⋅μ) by step 1.1.

3.1F6F7step 1.1step 2.2algebra

A translated simple is simple or zero. For every x∈W, apply exact T to Δ(x⋅λ)↠L(x⋅λ)↪∇(x⋅λ). Steps 1.1 and 2.2 give a surjection Δ(x⋅μ)↠TL(x⋅λ) followed by an injection into ∇(x⋅μ). If the intermediate object is nonzero, the composite is nonzero; the one-dimensional standard-costandard Hom of [F7] makes its image the canonical simple L(x⋅μ). The injection identifies the intermediate object with that image. Thus TL(x⋅λ) is either zero or L(x⋅μ), without any unstated Bruhat support assertion.

4.1F5step 1.1step 2.1step 3.1algebra

Minimal survival. Let w∈WJ and translate a composition series of Δ(w⋅λ). Its image is Δ(w⋅μ), whose head L(w⋅μ) occurs as a composition factor. By step 3.1 that factor is the image of some L(x⋅λ) with x⋅μ=w⋅μ. Step 2.1 kills every nonminimal x, and [F5] says that the coset wWJ has exactly one minimal representative, namely w. Thus TL(w⋅λ) is nonzero and equals L(w⋅μ) by step 3.1. Together with step 2.1 this proves (2).

5.1step 1.1step 2.1step 1.2step 2.2step 3.1step 4.1∎

Part (1) is steps 1.1 and 2.2; part (2) is steps 2.1 and 3.1 together with the simplicity statement of step 4.1; part (3) is step 1.2.

Remarks

Upper closure. The upper closure in Dot-Weyl facets and single-wall translation data preserves strictly positive pairings and allows a negative pairing to become zero. Consequently Fw⋅μ⊆Fw⋅λ+ holds exactly when wΦJ+⊆Φ+: zeros at the translated walls are allowed precisely when the corresponding regular pairings were negative. This is equivalent to wαj>0 for all j∈J, hence to w∈WJ by [F10]. The criterion therefore agrees with (2). In sl2, F0+=(−1,∞) excludes the wall −1, whereas F−2+=(−∞,−1] includes it, in agreement with TL(0)=0 and TL(−2)=L(−1). This is Chen Definition 3.9, Theorem 3.12 and Example 3.16; ordinary closure must not be substituted for upper closure.

Only the big antidominant projective Ξμ=P(μ) carries every distinct singular standard exactly once; other singular projectives need not contain all standards, so no universal lower bound (P:Δ)≥1 is asserted for arbitrary projectives.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

Singular category O multiplicities are parabolic Kazhdan–Lusztig values

Statement

Assume the Axiom of Choice (The Axiom of Choice) and work in the type-A root datum W=Sn of the page's Hecke module and ordinary multiplicity theorem. Let (λ,μ) be as in The facet tensor-weight exclusion lemma with J=J(μ), so Oμ is the singular integral block of The singular integral block and its dot stabilizer with labels normalized by the minimal right coset representatives WJ, and let w0 be the longest element of W. Then for all y,w∈WJ, [M(y⋅μ):L(w⋅μ)]=(P(w⋅μ):Δ(y⋅μ))=Pyw0, ww0(1), the value at q=1 of the ordinary Kazhdan--Lusztig polynomial attached to the pair (yw0,ww0) in the classical normalization of Kazhdan–Lusztig polynomials in the classical q-normalization. Equivalently, put k(w)=w0Jw−1w0∈JW. Then [M(y⋅μ):L(w⋅μ)]=mk(y),k(w)(1). This is the spherical family. The antispherical family instead governs the genuine regular parabolic locally finite category with dominant labels; changing to an opposite positive system does not identify the two families. In particular all singular multiplicities are computable from the ordinary Kazhdan--Lusztig coefficients of Sn, and no individual polynomial coefficients beyond the value at one occur.

Facts & Assumptions

Given: The facet data (λ,μ) with J=J(μ), Wμ=WJ, the regular block Oλ, the singular block Oμ with labels WJ (The singular integral block and its dot stabilizer, Translation to the wall on Verma, simple and projective objects), and the translation functor T=Tλμ.

[F1]

The block conventions: labels w⋅μ depend only on the coset wWJ and WJ is a set of minimal representatives; BGG reciprocity (P(w⋅μ):Δ(y⋅μ))=[M(y⋅μ):L(w⋅μ)] holds; Δ(w⋅μ)=M(w⋅μ); restricted duality fixes label pairs; all objects have finite length (The singular integral block and its dot stabilizer, BGG reciprocity, Standard and costandard objects).

[F2]

Wall translation: T is exact, TΔ(z⋅λ)≅Δ(z⋅μ) for all z, and TL(z⋅λ)≅L(z⋅μ) for z∈WJ while TL(z⋅λ)=0 for z∉WJ (Translation to the wall on Verma, simple and projective objects (1),(2)); Translation functors are exact and biadjoint supplies exactness and the biadjunction.

[F3]

Ordinary regular multiplicities: in a regular integral block of type A with dominant regular base λdom, [M(y⋅λdom):L(w⋅λdom)]=Py,w(1) for y,w∈W, with vanishing unless y≤w (Simple Verma composition multiplicities are Kazhdan–Lusztig values, The regular integral dot-orbit labelling convention).

[F4]

Ordinary Kazhdan--Lusztig polynomial data: Pa,b(1) is defined by the classical normalization, and simultaneous inversion gives Pa,b=Pa−1,b−1 (Kazhdan–Lusztig polynomials in the classical q-normalization, Existence and uniqueness of the Kazhdan–Lusztig basis). The latter supplier also gives uniqueness of the bar-fixed triangular basis; the quadratic/braid presentation and normalized standard elements are those of The normalized type-A Hecke algebra and its bar involution, and the conjugation-by-w0 symmetry needed below is proved from that uniqueness in step 1.3.

[F5]

Spherical reduction: for y,x∈JW, my,x=pw0Jy,w0Jx and pa,b=vℓ(b)−ℓ(a)Pa,b(v−2) with Pa,b(1) the value at q=1 (Parabolic Kazhdan–Lusztig polynomials are alternating sums of ordinary ones (2), The parabolic Kazhdan–Lusztig bases and polynomials); the labelling WJ=(JW)−1 is Support, intrinsic parabolic presentations, minimal coset representatives and length additivity, with the type-A identification.

Proof

technique · direct: push a composition series of the regular standard through the exact translation functor, count only the surviving minimal-representative simple, and translate the labels to the dominant base
1.1F1F2algebra

Composition multiplicities on the wall. Fix y,w∈WJ and a composition series of Δ(y⋅λ) with factors L(z⋅λ), z∈W. Exactness of T [F2] turns it into a series of TΔ(y⋅λ)≅Δ(y⋅μ) with factors TL(z⋅λ); by [F2] the only nonzero factors occur for z∈WJ, and then the factor is L(z⋅μ), which can equal the simple L(w⋅μ) only for z=w since distinct WJ-labels have distinct dot translates [F1]. Hence [Δ(y⋅μ):L(w⋅μ)]=[Δ(y⋅λ):L(w⋅λ)].

1.2F3algebra

Translate to the dominant base. Let λdom:=w0⋅λ; since λ is dot-antidominant regular, λdom is dominant regular and λ=w0⋅λdom, so for every z∈W, z⋅λ=(zw0)⋅λdom. Applying [F3] in the regular block of λdom gives [Δ(y⋅λ):L(w⋅λ)]=[M(yw0⋅λdom):L(ww0⋅λdom)]=Pyw0,ww0(1), the ordinary value supported on yw0≤ww0.

1.3F4F5algebra

The spherical dictionary. Conjugation by w0 permutes the type-A simple reflections, so the assignment Hz↦Hw0zw0, v↦v is an algebra automorphism: it preserves the quadratic and braid presentations, the bar, length and Bruhat order. It sends a bar-fixed triangular element with top Hb to one with top Hw0bw0; uniqueness in [F4] therefore gives pa,b=pw0aw0,w0bw0. Now set k(w)=w0Jw−1w0. Inversion sends WJ to JW [F5]. In type A, left-minimality means that the values in each consecutive J-block occur in increasing order: the left descent si occurs exactly when i+1 precedes i in one-line notation. Right multiplication by w0 reverses all positions, making those values decrease; left multiplication by w0J reverses the values inside each block, restoring increasing order. Thus k(w)∈JW. The spherical reduction [F5] gives mk(y),k(w)=pw0Jk(y),w0Jk(w)=py−1w0,w−1w0=pw0y,w0w=pyw0,ww0, using simultaneous inversion from [F4] and the conjugation symmetry just proved. Evaluating at v=1 and applying the classical normalization yields mk(y),k(w)(1)=Pyw0,ww0(1).

2.1F1step 1.2algebra

The projective version. By BGG reciprocity [F1], (P(w⋅μ):Δ(y⋅μ))=[M(y⋅μ):L(w⋅μ)]=Pyw0,ww0(1); this is the second displayed identity.

3.1F5step 1.2step 2.1step 1.3∎

Conclusion. Combining the preceding steps, [M(y⋅μ):L(w⋅μ)]=(P(w⋅μ):Δ(y⋅μ))=Pyw0,ww0(1)=mk(y),k(w)(1) for all y,w∈WJ, so the singular multiplicities are spherical parabolic values at one. The antispherical family is a different family (it is the sign-module normalization, used for the genuine regular parabolic category with dominant labels), and no change of positive system identifies the two; this is exactly the normalization distinction recorded in The parabolic Kazhdan–Lusztig bases and polynomials.

Remarks

Open obligation. The proof uses the batch-14 interfaces thm-simple-verma-composition-multiplicities-are-kazhdan-lusztig-values and def-regular-integral-dot-orbit-labelling-convention through fact [F3], cited in step 1.2 for the single-term reduction [M(yw0⋅λdom):L(ww0⋅λdom)]=Pyw0,ww0(1); both were scaffolded but unwritten when this item was authored, and both are now written in items/ with statements matching this use. The Step-3 decision remains escalated because only the owner may refresh an escalation receipt; the exact supplier IDs and the consuming step are recorded in the pair report. The labelling convention is used only through its explicitly quoted clauses (regular block data, Bruhat reversal and the Kazhdan--Lusztig normalization), not through its Soergel dictionary.

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6.1-sol)Open item page →

Standard parabolic and Levi subalgebras from a subset of simple roots

Definition

Assume the Axiom of Choice (The Axiom of Choice), inherited from the published root-space and sl2 suppliers. Let g be a finite-dimensional complex semisimple Lie algebra with Cartan subalgebra h, root system Φ (The root set is a reduced crystallographic root system, Root-space decomposition, Root spaces of a complex semisimple Lie algebra are one-dimensional), positive system Φ+ with base of simple roots Δ={α1,…,αr} (Positive systems and simple roots, Reduced crystallographic Euclidean root system), and let J be a subset of the simple indices {1,…,r}. Write ΦJ for the root subsystem spanned by the simple roots indexed by J, ΦJ=Φ∩span⁡R{αj:j∈J}, and let Hj=hαj∈h be the coroot of αj (Coroot of a Lie-algebra root via Killing-dual vector of a root). Put hJ=span⁡C{Hj:j∈J},zJ={h∈h:αj(h)=0 for all j∈J}, and define the following subspaces of g, where gα is the root space of α (Root-space decomposition):

sJ=hJ⊕⨁α∈ΦJgα,lJ=sJ⊕zJ=h⊕⨁α∈ΦJgα, uJ=⨁β∈Φ+∖ΦJ+gβ,vJ=⨁β∈Φ−∖ΦJ−gβ,pJ=lJ⊕uJ.

Here ΦJ±=ΦJ∩Φ± and Φ± are the positive and negative roots with n±, b as in Positive and negative nilpotent subalgebras and the Borel. The construction depends only on the positive system and on J. The following structural facts are part of the definition (proofs and references in the Remarks; rad⁡ and nilrad⁡ are those of Solvable radical and Nilradical).

(1) Decomposition of the Cartan. h=hJ⊕zJ and dim⁡hJ=∣J∣, so dim⁡zJ=r−∣J∣.

(2) The semisimple part. sJ is a Lie subalgebra of g; it is semisimple, and its root system with respect to the Cartan subalgebra hJ is ΦJ, with positive system ΦJ+ and Weyl group WJ=⟨sj:j∈J⟩≤W (Weyl group, Standard parabolic subgroups, descent-free one- and two-sided representatives, parabolic and reflection subgroups).

(3) The reductive Levi algebra. lJ is a Lie subalgebra with root system ΦJ relative to h; it is reductive as in Simple, semisimple, and reductive Lie algebras, with center zJ and derived algebra [lJ,lJ]=sJ; h is a Cartan subalgebra of lJ and the Weyl group of (lJ,h) is WJ. The algebra lJ need not be semisimple: its center zJ is nonzero as soon as J≠{1,…,r}.

(4) Parabolic and radicals. lJ, uJ, vJ and pJ are Lie subalgebras (Brackets of root spaces); uJ is an ideal of pJ, lJ stabilizes uJ and vJ, and pJ=lJ⋉uJ is a semidirect product (Levi subalgebras and Levi decompositions). Moreover nilrad⁡(pJ)=uJ and rad⁡(pJ)=zJ⊕uJ. Consequently [pJ,pJ]=sJ⊕uJ and the quotient pJ/rad⁡(pJ)≅sJ is semisimple: in the sense of Levi subalgebras and Levi decompositions, sJ is a semisimple Levi factor of pJ and lJ is the reductive complement to the nilradical. The parabolic pJ contains b and pJ=b⊕⨁α∈ΦJ−gα, the span of the negative root spaces of ΦJ; for J=∅ one has pJ=b, and for J={1,…,r} one has lJ=sJ=g and uJ=vJ=0.

(5) The positive grading. There exists ζ∈zJ with αj(ζ)=0 for j∈J and αi(ζ)=1 for i∉J. With respect to ad⁡(ζ) every root space gβ has weight β(ζ), so:

  • lJ has degree 0 (all of ΦJ vanishes on ζ);
  • uJ has strictly positive integral degrees β(ζ)∈Z>0;
  • vJ has strictly negative integral degrees.

In particular uJ is nilpotent by this positive grading, and both uJ and vJ have finite-dimensional fixed-degree slices of their PBW algebras: dim⁡U(uJ)d<∞ and dim⁡U(vJ)d<∞ for every d∈Z (PBW gives an ordered monomial basis for the enveloping algebra).

Remarks

  • The Cartan splitting (1). The pairing matrix A=(αi(Hj))i,j∈J is the principal submatrix of the Cartan matrix of (Φ,Δ) with respect to the base {αj:j∈J} of ΦJ (Simple roots form a signed integral basis). By Properties of finite-type Cartan matrices (4) the symmetrization DAD−1 is positive definite, so A is nonsingular. Hence if h=∑jcjHj satisfies αi(h)=0 for all i∈J, then Ac=0 and c=0; so hJ∩zJ=0 and dim⁡hJ=∣J∣. Counting dimensions gives h=hJ⊕zJ and dim⁡zJ=r−∣J∣.
  • Semisimplicity of sJ (2). The Killing-dual map is linear, so α=∑j∈Jmjαj gives Hα=∑j∈JmjHαj∈hJ (Killing-dual vector of a root, Coroot of a Lie-algebra root). Opposite-root brackets therefore lie in hJ, and root brackets give subalgebra closure. Use its intrinsic Killing form κ, which need not equal the ambient Killing restriction. Its root-space decomposition gives κ(h,h′)=∑α∈ΦJα(h)α(h′) on hJ. On the real coroot span this is positive definite: the simple-root pairing matrix is nonsingular by (1), and the sum of squares vanishes only when every simple-root value is zero. Thus its complexification is nondegenerate there. Trace-form invariance (Trace forms are symmetric and invariant) gives orthogonality of weights whose sum is nonzero: applying a Cartan element gives (α+β)(h)κ(xα,xβ)=0. For a normalized sl2 triple eα,fα,hα (The root sl_2 triple), invariance gives κ(hα,hα)=2κ(eα,fα)≠0. One-dimensionality of the root spaces makes each opposite pair nondegenerate. Hence κ is nondegenerate on sJ, and Cartan's semisimplicity criterion proves semisimplicity. Every root restricts nontrivially to hJ, since it vanishes on zJ and h=hJ⊕zJ. The weight decomposition then shows that the centralizer and normalizer of hJ in sJ are both hJ; thus it is a Cartan subalgebra. Restriction identifies its roots with ΦJ and its simple reflections with sj, j∈J, extended by the identity on zJ.
  • Reductive lJ (3). Since zJ centralizes hJ and each root space of ΦJ (αj(h)=0 for h∈zJ), it commutes with sJ, so lJ=zJ⊕sJ is a Lie subalgebra with central summand; [lJ,lJ]=sJ because every root vector is a bracket with a Cartan element, and the simple coroots spanning hJ are the brackets [ej,fj]. The center contributes nothing. Its center is zJ together with the center of sJ, which is zero, and the displayed sum is direct. Hence lJ is reductive in the sense of Simple, semisimple, and reductive Lie algebras, with root system ΦJ relative to h and Weyl group WJ. For J≠ all indices, zJ≠0 by (1), so lJ is not semisimple.
  • Brackets, ideals and grading (4)-(5). For roots α,β∈ΦJ (i.e. sums of simple roots in J) one has α+β∈ΦJ whenever the sum is a root; if the sum is neither a root nor zero, the bracket is zero. This gives the bracket closure for lJ; the mixed brackets with uJ likewise stay in uJ∪lJ∪{0} by Brackets of root spaces and the signs of the simple-root coefficients (Simple roots form a signed integral basis), so uJ is an ideal of pJ and lJ preserves uJ and vJ; this gives the semidirect product pJ=lJ⋉uJ, and pJ=b⊕⨁α∈ΦJ−gα is a restatement of lJ=h⊕⨁α∈ΦJgα. The simple roots form a basis of h∗, so there is a unique ζ∈h with the prescribed simple-root values. Its zero values on J put it in zJ. Every positive root outside ΦJ has a positive sum of coefficients outside J; negative roots have the opposite sign. This proves (5), and brackets add degrees, proving nilpotency of uJ. For maximality, let I be a nilpotent ideal of pJ. Its image in lJ is nilpotent and has zero projection to the semisimple sJ, since nilpotent algebras are solvable. If its central projection contained h≠0, then some i∉J would have αi(h)≠0. Choose x=h+a∈I with a∈uJ and 0≠ei∈gαi. The element [x,ei]∈I has degree-one component αi(h)ei; the contributions from a have degree at least two. For every k≥0, (ad⁡x)k[x,ei] has degree-one component αi(h)k+1ei≠0. This contradicts nilpotency of I, whose lower central series makes ad⁡x nilpotent on I. Thus I⊆uJ, proving the nilradical assertion. Finally zJ⊕uJ is a solvable ideal with semisimple quotient sJ, so it is exactly rad⁡(pJ); and by Levi subalgebras and Levi decompositions the semisimple complement sJ is a Levi factor of pJ. The fixed-degree PBW finiteness in (5) uses finite-dimensionality of each root space, the finite height bound on positive roots occurring in a fixed degree, and the PBW basis theorem (PBW gives an ordered monomial basis for the enveloping algebra).
  • The cases J=∅ and J all. For J=∅ one has hJ=0, zJ=h, ΦJ=∅, lJ=h and uJ=n+, so pJ=b. For J={1,…,r} one has zJ=0, ΦJ=Φ, uJ=vJ=0 and pJ=g. These are the two extreme cases of the standard parabolic construction.
  • Choice. The Axiom of Choice is inherited from the published root-space and sl2 suppliers used in the justifications; no further choice principle enters this construction.
DefinitionDefinition: Literature-sourcedProof: Not applicableOpen item page →

Parabolic category O for a standard parabolic

Definition

Assume the Axiom of Choice (The Axiom of Choice), inherited from the published category-O theory. Let p=pJ=lJ⊕uJ be the standard parabolic of Standard parabolic and Levi subalgebras from a subset of simple roots, with reductive parabolic Levi lJ, nilradical uJ and Weyl group WJ; the ambient category is O of The classical BGG category O with standard objects the Verma modules Verma modules.

(1) Dominance for the Levi. A weight λ is p-dominant if ⟨λ,α∨⟩∈Z≥0 for all α∈ΦJ+ (this is dominance for the Levi root system; in particular λ is integral for ΦJ; see Integral, dominant, and strictly dominant weights for the ambient notion). No integrality condition is imposed on the simple coroots outside J, or on the central part zJ of the Levi.

(2) Generalized Verma modules. For a p-dominant λ let E(λ) be the finite-dimensional simple p-module obtained from the finite-dimensional simple lJ-module of highest weight λ (Finite-dimensional simple modules are classified by dominant highest weights) by letting uJ act by zero. Define the generalized Verma module

Mp(λ):=U(g)⊗U(p)E(λ).

(3) The parabolic category. The parabolic category Op is the full subcategory of O whose objects are the finitely generated h-semisimple g-modules on which U(p) acts locally finitely, that is, on which every vector lies in a finite-dimensional p-submodule; equivalently, every vector lies in a finite-dimensional lJ-submodule and is annihilated by a power of uJ (local finiteness for the nilpotent uJ).

The following structural facts are part of the definition; the arguments are given in the Remarks.

(4) Generalized Vermas lie in Op. For each p-dominant λ, Mp(λ) is an object of Op with highest weight λ, and the natural surjection M(λ)↠Mp(λ) exhibits Mp(λ) as the largest quotient of the Verma module M(λ) that lies in Op.

(5) Simples. The simple objects of Op are exactly the simple highest-weight modules L(λ) with λ p-dominant (The simple objects of O).

(6) Serre property. Op is closed under subobjects, quotients and extensions in O; in particular it is an abelian subcategory of O stable under finite direct sums.

(7) Costandard objects. For p-dominant λ put ∇p(λ):=DMp(λ), the restricted dual of Restricted duality is exact and involutive on O. These are the costandard objects of Op; each is naturally a submodule of the ordinary dual Verma DM(λ). They need not be quotients of ordinary dual Vermas.

Remarks

  • Finite Levi modules. A finite-dimensional lJ-module with h-semisimple action splits as the direct sum of its center-weight spaces, and on each such space, which is a finite-dimensional module for the semisimple algebra sJ, Weyl's complete reducibility theorem (Weyl's complete reducibility theorem) applies. Hence such a module is a direct sum of finite-dimensional simple lJ-modules, and those are parametrized by weights integral and dominant for ΦJ, with center zJ acting through the weight (Finite-dimensional simple modules are classified by dominant highest weights, Standard parabolic and Levi subalgebras from a subset of simple roots (3)).
  • Why Mp(λ)∈Op (4). Inflating E(λ) to p by uJ⋅E(λ)=0 is a module structure because uJ is an ideal of p and uJ acts on E(λ) by zero. Poincar'e--Birkhoff--Witt (PBW gives an ordered monomial basis for the enveloping algebra) applied to the triangular decomposition g=vJ⊕p (with vJ the sum of the negative outside root spaces) identifies Mp(λ) with U(vJ)⊗E(λ) as a vector space, and more precisely each bounded negative ζ-degree piece of U(vJ) is finite dimensional (Remark (5) of Standard parabolic and Levi subalgebras from a subset of simple roots) and lJ-stable. Positive uJ raises the ζ-degree and hence lowers this degree after PBW reordering; since it annihilates the degree-zero Levi part, each such finite piece is p-stable. This proves p-local finiteness of Mp(λ) directly, with finite-dimensional p-stable pieces. The natural surjection M(λ)↠Mp(λ) (induced by vλ↦1⊗eλ) shows that Mp(λ) is a quotient of the Verma module, hence an object of O (Verma and finite-dimensional weight modules belong to O); it has the same highest weight λ and a unique simple head L(λ), since every simple quotient is also a simple quotient of M(λ) (A Verma module has a unique simple quotient), and all its composition factors lie below λ because it is a quotient of M(λ) (Every object of O has finite length, The simple objects of O).
  • Largest quotient (4). For a parabolic quotient Q of M(λ), let v‾ be the image of the highest vector. Every positive root vector kills it. Its finite Levi span is, by complete reducibility, a cyclic highest-weight copy of E(λ) if v‾≠0, and zero otherwise. The radical kills this entire span because it kills v‾ and [uJ,lJ]⊆uJ. Tensor induction therefore factors the quotient map through Mp(λ).
  • Local finiteness and Serre closure (6). Inside O, p-local finiteness is equivalent to local nilpotence of the negative simple root vectors fj, j∈J. Necessity follows because fj lowers weights and a finite-dimensional weight module has finite weight support. Conversely, the cyclic Levi module of a weight vector is finitely generated, Cartan-semisimple and locally nJ+-finite, hence is an object of the semisimple Levi's category O and has finite length. Local nilpotence of fj descends to its simple highest-weight factors. If v is the top of such a factor and k is least with fjkv=0, the relation ejfjkv=k(λ(hj)−k+1)fjk−1v forces λ(hj)=k−1∈Z≥0 (The root sl_2 triple). The finite-dimensional highest-weight classification makes each factor finite dimensional, so the cyclic Levi module is finite dimensional. A vector with several weights is handled by their finite sum. Moreover U(uJ)m is finite dimensional because uJ⊆n+. PBW gives U(p)m=U(lJ)U(uJ)m; the finitely many basis vectors of that radical span each have a finite Levi span. This proves local finiteness. For extensions, kill a vector's image by one power of fj, then its resulting vector in the submodule by another. Local nilpotence also passes to subobjects and quotients, so this criterion proves the Serre property in the ambient abelian category.
  • Simples (5). A simple g-module L(λ) lies in Op if and only if λ is p-dominant: necessity holds because the highest weight vector generates a finite-dimensional p-module, whose finite-dimensional sJ-part has λ as a highest weight of a finite Levi simple, forcing ⟨λ,α∨⟩∈Z≥0 for α∈ΦJ+; sufficiency holds because L(λ) is a quotient of Mp(λ)∈Op and Op is closed under quotients by the previous paragraph.
  • Duality and costandards (7). Restricted duality (Restricted duality is exact and involutive on O) is exact and contravariant on O. A locally finite semisimple Levi representation with finite h-weight spaces is an algebraic direct sum of finite-dimensional Levi simples: taking an increasing sequence of finite-dimensional Levi submodules exhausting the space, complete reducibility splits each inclusion, the successive spans exhaust the space, and every weight meets only finitely many summands (the weight space is finite dimensional); the restricted dual is then the direct sum of the finite duals, again a finite-dimensional-module direct sum. Hence D preserves Op; the natural map DMp(λ)→DM(λ) is injective (dual of the Verma quotient), so ∇p(λ) is naturally a submodule of the ordinary dual Verma. Its simple socle is L(λ) by duality from the simple head of Mp(λ). It is not generally a quotient of DM(λ): dualizing, that would say M(λ) embeds in Mp(λ), which fails as soon as the Verma surjection has a nonzero kernel (for instance g=sl2, J={s} and λ=0, when Mp(0)=L(0) is one dimensional).
  • Choice. The Axiom of Choice is inherited from the published category-O theory quoted above (restricted duality, finite length, enough structures); no new choice principle is introduced.
TheoremStatement: Literature-sourcedProof: AI-adaptedOpen item page →

Parabolic category O is a highest weight category

Statement

Assume the Axiom of Choice (The Axiom of Choice). Let p=lJ⊕uJ and Op be as in Parabolic category O for a standard parabolic. Then:

(1) (Simples.) L(λ)∈Op if and only if λ is p-dominant (integral and dominant for the Levi coroots); every simple object of Op is of this form. No integrality condition is imposed on the coroots outside J.

(2) (Standards.) For every p-dominant λ the generalized Verma Mp(λ) lies in Op, has finite length, has L(λ) as its unique simple quotient (head), and [Mp(λ):L(μ)]≠0 forces μ≤λ in the dominance order; Mp(λ) is the largest quotient of M(λ) lying in Op.

(3) (Enough projectives.) The inclusion Op⊆O is exact and has a left adjoint Lp (the largest p-locally-finite quotient); Lp is right exact and sends projective objects of O to projective objects of Op. Consequently, for every p-dominant λ the object Pp(λ):=Lp(P(O)(λ)) is a projective cover of L(λ) in Op, and Op has enough projectives and injectives.

(4) (Generalized-standard flags, BGG reciprocity and duality.) Every projective in any fixed generalized central-character summand has a finite generalized Verma flag. Each projective cover surjects to its labelled generalized standard with a kernel filtered by higher-labelled generalized standards. The simple and generalized-standard classes form bases of the finite summand Grothendieck group, and characters are injective on this group. For p-dominant λ,μ, (Pp(λ):Mp(μ))=[Mp(μ):L(λ)], the multiplicities being finite; moreover restricted duality restricts to an exact contravariant involution of Op exchanging Mp(λ) and ∇p(λ), and Ext⁡Op1(Mp(λ),∇p(μ))=0 for all p-dominant λ,μ. Thus each nonzero central-character summand is a finite highest-weight category for the root order μ≤λ  ⟺  λ−μ∈Q+, and Op is their categorical direct sum, with each object supported on finitely many summands. A central-character summand need not be connected. Ambient integrality is not required for any of these assertions.

Facts & Assumptions

Given: A finite-dimensional complex semisimple Lie algebra with the standard parabolic p=l⊕u, negative outside radical v, grading element ζ∈z(l), and the full locally p-finite subcategory Op of ordinary O. Write N=U(v) and B=U(p). The grading has αj(ζ)=0 for j∈J and αi(ζ)=1 for i∉J; its integer degrees are distinct from ambient integrality of highest weights.

[F1]

Parabolic objects and Serre closure (Parabolic category O for a standard parabolic, Standard parabolic and Levi subalgebras from a subset of simple roots): finite Levi simples E(λ) are inflated with radical acting by zero; Δp(λ)=U(g)⊗BE(λ) has simple head L(λ) and lies in the Serre subcategory Op; the ordinary Verma surjects onto it. Duality preserves the subcategory and ∇p(λ) embeds in the ordinary dual Verma.

[F3]

Ordinary weights and orthogonality (Weights of a Verma module lie below lambda, The universal property of Verma modules, Standard-costandard Hom and Ext-one orthogonality): ordinary Verma weights lie in λ−Q+; a highest-weight vector generates a quotient of that Verma; ordinary standard-costandard Hom is δλμC. Here Q+ is the nonnegative integral simple-root cone; its strict root order is transitive.

[F4]

Levi splitting and PBW (Weyl's complete reducibility theorem, Finite-dimensional simple modules are classified by dominant highest weights, PBW gives an ordered monomial basis for the enveloping algebra): finite-dimensional h-semisimple Levi modules split into simples, since the center splits into weight spaces and the derived Levi is semisimple; U(g)=N⊗B by PBW; N has nonpositive ζ-degrees, N0=C and finite-dimensional homogeneous pieces.

[F5]

Central summands (Generalized central-character decomposition of O, Central characters are dot-Weyl orbits): every object has finite central-character support, different summands have no morphisms, and projection to each summand is exact. For any central character χ, its parabolic simple labels form the finite set Σχ of p-dominant weights in its dot-Weyl orbit; no ambient integrality or connectedness is assumed.

[F6]

Restricted duality (Restricted duality is exact and involutive on O) is exact, simple-preserving and involutive, exchanges ordinary standards and costandards, and preserves the parabolic subcategory. Projectivity and injectivity have their represented-Hom meaning (Projective object, Injective object). The Axiom of Choice (The Axiom of Choice) is inherited and used for bases, maximal kernels, splittings and the indicated covers and lifts.

Proof

technique · direct
1.1F1F2F3F4

Simples and standards. By [F1], p-dominant highest weights are exactly the simple labels in Op. Their generalized standards are quotients of the ordinary Vermas with the same one-dimensional highest-weight space and simple head. They have finite length by [F2], and their weights, and hence highest weights of composition factors, lie in λ−Q+ by [F3]. For any parabolic quotient of the ordinary Verma, the image of its highest vector is killed by the full positive nilpotent algebra. Its finite Levi span is the cyclic highest-weight simple E(λ) by complete reducibility, and the radical kills that span since it kills the highest vector and is Levi-stable. The induction universal property therefore factors the quotient map through Δp(λ). This proves the largest-quotient assertion locally. This proves parts (1) and (2).

1.2F1F2F6algebra

Largest locally finite quotient. For M∈O, let K be the set of its subobjects K such that M/K∈Op. It contains M, and finite length bounds ℓ(M/K), so choose K0 for which this quotient length is maximal. For every K∈K, the diagonal map embeds M/(K∩K0) in M/K⊕M/K0. Serre closure gives K∩K0∈K. Maximality and the epimorphism M/(K∩K0)→M/K0 force equality of lengths; its kernel is therefore zero and K0⊆K. Thus K0 is the least kernel among all parabolic quotients. For a map M→V with V parabolic, its image is parabolic and its kernel lies in K, so the map factors uniquely through M/K0. This proves the functorial universal property of LpM=M/K0 and its left adjunction to the exact inclusion. It is right exact as a left adjoint.

1.3F1F3F6algebra

The correct Hom arrows. The ordinary Verma surjection is M(η)↠Δp(η), and the costandard inclusion is ∇p(ξ)↪DM(ξ). Composing with these maps injects parabolic Hom⁡(Δp(η),∇p(ξ)) into the ordinary standard-costandard Hom of [F3]. Thus it is zero for different labels and at most one-dimensional for equal labels. In the equal case the nonzero composite through L(η) is a homomorphism from the standard head to the costandard socle, so its dimension is one.

1.4F1F2F3F4F6algebra

Every parabolic standard-costandard extension splits. Consider 0→∇p(ξ)→X→Δp(η)→0. First suppose ξ−η∉Q+∖{0}, including equality and incomparability. Choose a weight-η lift v of the source highest vector. If a positive root vector sent v nontrivially to weight η+α, its image in the standard quotient would be zero, so this weight would occur in the costandard. But all costandard weights are ≤ξ; hence η<η+α≤ξ would imply η<ξ, contrary to the case assumption. Thus v is killed by every positive root vector. Its Levi span is finite dimensional because X is locally parabolic-finite. By [F4] it is the cyclic finite Levi highest-weight simple E(η): complete reducibility decomposes the span, and the nonzero components of a highest-weight vector of weight η belong to copies of that same simple, with cyclic diagonal span one copy. The radical kills this whole span since it kills v and [l,u]⊆u. Induction therefore gives a map Δp(η)→X whose composite onto the quotient is the identity on its generating top, hence the identity. The extension splits. In the remaining case η<ξ, duality gives an extension with reversed pair (ξ,η), to which the first case applies; dualizing the splitting back finishes the proof. Consequently parabolic Ext⁡1(Δp(η),∇p(ξ))=0 for every pair, proved inside this category. The argument uses the absence of a weight obtained by raising the chosen top, not a total ordering of incomparable highest weights.

1.5F1F2F3F4F5algebra

Integer bounds in every central summand. Fix any central character χ with a parabolic label η∈Σχ. Put d=max⁡{(ν−η)(ζ):ν∈Σχ, (ν−η)(ζ)∈Z}. This is a nonnegative integer, since the set is finite and contains 0 from ν=η. Every weight γ of an object W in this summand belongs to ν−Q+ for some composition-factor label ν∈Σχ. Because Q+(ζ)⊆Z≥0, if γ(ζ)−η(ζ)∈Z then (ν−η)(ζ)∈Z and γ(ζ)−η(ζ)≤d. Thus the ζ-eigenvalues in the coset η(ζ)+Z have a uniform upper bound. This compares integer differences only, without ordering complex eigenvalues in different cosets or requiring a dot-dominant maximum.

2.1F1F2F4F5F6step 1.5algebra

Finite bounded induced projectives. Let E=E(η) and let B≤d(E) be U(p)⊗U(l)E modulo the positive-radical PBW terms of degree greater than d. The excluded terms form a submodule since the Levi preserves degree and the radical increases it. The quotient is finite dimensional by [F4]. A Levi map E→W extends to the induced p-module, and its extension kills those terms by step 1.5: ζ acts on E by η(ζ) and a degree-k term would map to eigenvalue η(ζ)+k. The descending degree filtration of B≤d(E) has semisimple Levi slices with radical acting by zero. Refining the slices into Levi simples and inducing exactly by PBW gives a finite generalized-standard flag; Serre closure places the induced module in Op. By central projection and induction adjunction, Q(E)=pr⁡χ(U(g)⊗BB≤d(E)) represents Hom⁡l(E,W) on this summand. This functor is exact: lift a weight basis of the finite Levi image through an epimorphism, take the sum of the finite Levi spans of the lifts, and split its epimorphism onto that image by [F4]. The inclusion E(η)↪L(η) induces a nonzero map Q(E)→L(η), hence an epimorphism. Thus Q(E) is projective and maps onto this simple.

2.2F1F2F6step 1.2algebra

Projective covers and enough projectives. The adjunction gives Hom⁡Op(LpP,−)=Hom⁡O(P,−) on the parabolic subcategory, so an ordinary projective has parabolic projective image. For a parabolic simple L(λ), apply Lp to the ordinary projective cover to obtain LpP(O)(λ)↠L(λ). If a subobject of this image surjects onto the simple, its inverse image in the ordinary cover does also and is the whole cover by essentiality. Hence the image map is essential and is the required parabolic projective cover. An ordinary projective epimorphism onto any parabolic object becomes a parabolic projective epimorphism by right exactness. Dualizing gives enough injectives. This proves part (3) on all the objects in the stated parabolic category.

3.1F1F2F4F5step 2.2step 2.1algebra

These projectives begin with actual PBW flags. Filter B≤d(E) by descending radical degree. Each degree slice is finite-dimensional and semisimple for the Levi, with radical acting by zero on the slice. PBW makes induction from B exact, so inducing this filtration gives a finite generalized-standard flag on U(g)⊗BB≤d(E). A generalized standard has scalar central character, because its ordinary Verma precursor does; exact central projection therefore either retains or kills each whole standard slice. Hence Q(E) has an actual finite generalized-standard flag. As an N-module the unprojected induction is graded free of finite rank, with basis the finite graded space B≤d(E). Central projection is an N-linear grading-preserving direct-summand projection. Every parabolic projective cover in the chosen summand is a grading-preserving summand of the corresponding Q(E): lift the epimorphism Q(E)→L(η) to its cover, use essentiality for surjectivity, then split by projectivity.

4.1F4F6step 3.1algebra

Graded direct summands stay free. For a graded N-module bounded above, the equality V=N<0V forces V=0: a nonzero homogeneous component of highest occupied degree cannot be obtained from strictly higher-degree components. Let R be a grading-preserving direct summand of a finite graded free N-module. Then R/N<0R is finite dimensional. Choose a homogeneous basis of this quotient and lifts to R, giving a map from a finite graded free module G onto R; its cokernel is bounded above and equals its own N<0 multiple, so the map is surjective by the preceding argument. It splits as a graded N-map since R is a graded projective summand. The kernel K then has K/N<0K=0, because the induced map on generator quotients is an isomorphism and reduction of the split sequence is split exact. The same bounded-above argument gives K=0. Thus R is graded free of finite rank. This proves the summand claim without importing the flag criterion or assuming finite-dimensional degree-zero pieces of an enveloping Levi algebra.

5.1F1F2F4F5step 3.1step 4.1induction

Free finite rank gives generalized-standard flags. Let R be one of these parabolic g-modules, graded free of finite N-rank. Within its fixed ζ-eigenvalue coset its degrees are bounded above and its components finite dimensional. Let a be its highest occupied degree. The finite Levi module Ra is semisimple and the positive radical kills it, as it raises degree. A homogeneous free basis shows that all its degree-a basis vectors are part of an N-basis, so the induction map U(g)⊗BRa=N⊗Ra→R is injective and its quotient is again graded free, with those finitely many generators removed. Decomposing Ra into finite Levi simples gives an actual submodule that is a finite sum of generalized standards. Repeating on the quotient terminates since each step reduces the finite N-rank. All subobjects and quotients lie in the selected central summand; hence their simple Levi labels are the allowed labels there. This proves that every projective cover, and therefore every finite sum and summand of them, has a finite generalized-standard flag. The grading is only the actual semisimple ζ-action, shifted on its coset; it is not a Koszul assumption. This proves the flag part of (4).

6.1F1F2F6step 2.2step 1.3step 1.4step 5.1algebra

Reciprocity. For a generalized-standard flag of Pp(λ), steps 1.3 and 1.4 make Hom⁡(−,∇p(μ)) exact along its flag sequences and show that its dimension counts precisely the Δp(μ) sections. On the other hand projectivity makes Hom⁡(Pp(λ),−) exact along a composition series of ∇p(μ). Its value on a simple has dimension one for L(λ) and zero for every other label, by the essential simple head of the projective cover. Consequently (Pp(λ):Δp(μ))=[∇p(μ):L(λ)]=[Δp(μ):L(λ)], the last equality by simple-preserving duality. This also proves independence of flag multiplicities and their finiteness. By standard support, only μ≥λ occurs, and the diagonal multiplicity is one.

7.1F1F2F3F4step 1.4step 5.1step 6.1algebra

Labelled standard quotient and its higher-standard kernel. The same lifting argument as in step 1.4, with Δp(β) in place of the costandard, shows Ext⁡1(Δp(α),Δp(β))=0 whenever α≮β: its submodule has no weight strictly above the lifted top α, and finite Levi splitting supplies the section. In the flag of Pp(λ) from step 5.1, the labelled standard occurs once and every other standard has strictly higher label by step 6.1. Whenever the labelled standard precedes a higher standard in the flag, the corresponding two-section subquotient splits by this Ext vanishing. Swapping those two sections moves the labelled standard upwards in the flag; finitely many swaps place it last. Thus there is an actual epimorphism Pp(λ)→Δp(λ) with a kernel having a finite flag of strictly higher standards. This is the highest-weight projective-kernel axiom, obtained without the tilting or homological-criterion suppliers.

8.1F1F2F5F6step 1.1step 1.2step 2.2step 1.3step 1.4step 5.1step 6.1step 7.1∎

Bases, characters and conclusion. In any fixed central-character summand there are finitely many allowed labels by [F5]. Their simple classes form its finite-length Grothendieck basis; the standard-to-simple matrix has diagonal one and entries only below the diagonal after a linear extension of the root order. It is therefore invertible over Z, so standard classes give another basis. A finite relation between the simple characters has a maximal highest-weight label with a nonzero coefficient; its highest-weight monomial occurs in that simple with coefficient one and in no remaining incomparable or lower simple, so the relation cannot vanish. Inducting proves that characters are injective on this Grothendieck group. The constructions above establish the simples and standard support, the reflector and enough projectives, the local Hom/Ext orthogonality, actual generalized-standard flags, reciprocity and the labelled higher-standard kernel. Duality has remained the ordinary restricted duality restricted to the actual parabolic subcategory. Thus every nonzero summand has the finite highest-weight structure asserted in (4), including nonintegral and disconnected summands. The exact central projections preserve the parabolic subcategory by Serre closure, giving its categorical direct-sum decomposition with finite support. These are all four parts of the Statement.

Remarks

This proof constructs the generalized-standard projective flags by finite graded PBW and Levi splitting, including the Levi center. It uses neither the homological flag criterion nor tilting classification, and does not apply the merely right-exact reflector to an ordinary Verma flag.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

Parabolic category O multiplicities are parabolic Kazhdan–Lusztig values

Statement

Assume the Axiom of Choice (The Axiom of Choice). Fix a dominant regular integral weight λ of the complex type-A Lie algebra and the standard parabolic pJ, with reductive Levi lJ and longest Levi Weyl element wJ (Standard parabolic and Levi subalgebras from a subset of simple roots). The pJ-dominant weights in this regular block are exactly y⋅λ for y∈JW. Write ΔJ(y⋅λ)=Mp(y⋅λ), L(x⋅λ) and PJ(x⋅λ) for its generalized standards, simples and projective covers (Parabolic category O for a standard parabolic). Then for x,y∈JW, [ΔJ(y⋅λ):L(x⋅λ)]=(PJ(x⋅λ):ΔJ(y⋅λ))=∑u∈WJ(−1)ℓ(u)Puy,x(1)=ny,x(1).

These are the antispherical sign coefficients, with support y≤x and diagonal 1. The precise character/Grothendieck identity supplying the formula is ch⁡ΔJ(y⋅λ)=∑u∈WJ(−1)ℓ(u)ch⁡Δ(uy⋅λ).

For the same highest weights in the antidominant base λanti=w0⋅λ, the actual longest-left label is z=yw0, since y⋅λ=z⋅λanti. Its shortest-left encoding is a=lf(y):=wJyw0∈JW, so the weight in that encoding is (wJa)⋅λanti, not a⋅λanti. With b=lf(x), the reindexed matrix is N~a,b:=nlf−1(a),lf−1(b)(1)=nwJaw0, wJbw0(1). It is the same multiplicity matrix with both indices permuted and has reversed support in the transported order. It is not in general na,b(1). Every label retains its explicit dot-baseweight. The highest-weight category theorem applies also to singular integral parabolic blocks, but this ordinary regular KL matrix formula assumes the ambient central character is regular. No singular/parabolic category equivalence, grading or Koszul duality is asserted.

Facts & Assumptions

Given: A dominant regular integral weight λ of type A, the standard parabolic pJ=lJ⊕uJ with Levi Weyl group WJ and longest element wJ, the parabolic category Op with generalized Vermas ΔJ(ξ)=Mp(ξ) (Parabolic category O for a standard parabolic, Parabolic category O is a highest weight category), and the completed character ring with its geometric-series inverses (The completed formal character ring, Geometric series are invertible in the completed character ring).

[F1]

Parabolic highest-weight structure: Mp(ξ) lies in Op with head L(ξ) and finite length, its composition factors are parabolic simples, and in a fixed integral block every projective has a finite generalized-standard flag with BGG reciprocity (Pp(ξ):Mp(η))=[Mp(η):L(ξ)]; the parabolic simple/standard classes form Grothendieck bases and characters are injective on the block (Parabolic category O is a highest weight category (1)-(4), Parabolic category O for a standard parabolic).

[F2]

The local PBW model: ch⁡Mp(ξ)=ch⁡E(ξ)⋅∏β∈Φ+∖ΦJ+(1−e−β)−1 and, by the PBW theorem, the identity holds in the completed character ring; the outside denominator is the PBW denominator of the negative outside radical vJ (The PBW model of a Verma module, PBW gives an ordered monomial basis for the enveloping algebra, Standard parabolic and Levi subalgebras from a subset of simple roots (5)).

[F3]

Weyl character formula for sJ: for a p-dominant integral ξ, ch⁡E(ξ)=∑u∈WJ(−1)ℓ(u)eu(ξ+ρJ)∏α∈ΦJ+(eα/2−e−α/2) with ρJ half the sum of the positive Levi roots (The Weyl character formula); here WJ fixes ρ−ρJ because (ρ−ρJ)(Hj)=0 for every j∈J, and ΦJ+=ΦJ∩Φ+.

[F4]

Ordinary regular multiplicities: for a dominant regular integral base, [M(z⋅λ):L(x⋅λ)]=Pz,x(1) with vanishing unless z≤x (Simple Verma composition multiplicities are Kazhdan–Lusztig values, The regular integral dot-orbit labelling convention), and the alternating sum identity ny,x(1)=∑u∈WJ(−1)ℓ(u)Puy,x(1) (Parabolic Kazhdan–Lusztig polynomials are alternating sums of ordinary ones (1) at v=1); the labelling dictionary WJ=(JW)−1 and the coset minima are Standard parabolic subgroups, descent-free one- and two-sided representatives, parabolic and reflection subgroups, Support, intrinsic parabolic presentations, minimal coset representatives and length additivity, with the type-A identification.

[F5]

The regular-to-regular translations of Translation identifies the regular integral blocks identify labelled ordinary standards, simples and projective covers between dominant regular integral blocks of the same type-A root datum, so [F4] holds in every such block, not only the principal one.

Proof

technique · direct: prove the finite-Levi character identity, compare ordinary simple coefficients using the regular Kazhdan-Lusztig multiplicity theorem, and read off the antispherical matrix and its projective version
1.1F2F3algebra

Levi character identity. Multiply the Weyl character formula [F3] for E(ξ) by the outside PBW denominator of [F2]. For each u∈WJ, u(ρ−ρJ)=ρ−ρJ by [F3], so eu(ξ+ρJ)+(ρ−ρJ)=eu(ξ+ρ); dividing by the full PBW denominator ∏α∈Φ+(eα/2−e−α/2)=∏α∈ΦJ+(…)∏β∉ΦJ+(…) turns each summand into ch⁡Δ(u⋅ξ) (The PBW model of a Verma module). Hence, for p-dominant integral ξ, ch⁡ΔJ(ξ)=∑u∈WJ(−1)ℓ(u)ch⁡Δ(u⋅ξ) in the completed character ring; this is the displayed identity at ξ=y⋅λ.

1.2F4algebra

The dominant labels. Put A=λ+ρ, strictly dominant and integral. The weight y⋅λ is pJ-dominant iff ⟨yA,αj∨⟩−1≥0 for every j∈J. The integer ⟨A,y−1αj∨⟩ is positive exactly when y−1αj is a positive root: strict dominance makes positive-root coroot pairings positive integers and negative-root pairings negative. Thus the condition is y−1αj>0 for all j∈J, equivalently no LEFT descent sjy<y: in type A, y−1αj>0 means that the position of value j precedes that of j+1, exactly the inversion-length ascent condition for swapping those values. These are precisely the shortest-left representatives JW in the coset-minimum convention of [F4]; no inversion of the weight label is performed.

2.1F1F4F5step 1.1step 1.2algebra

Comparing simple coefficients. By [F1] the parabolic simples are the L(x⋅λ), x∈JW. Taking the coefficient of L(x⋅λ) in the identity of step 1.1 (with ξ=y⋅λ, y∈JW) and using [F4] and [F5] for the ordinary Vermas in the given dominant regular integral block gives [ΔJ(y⋅λ):L(x⋅λ)]=∑u∈WJ(−1)ℓ(u)[M(uy⋅λ):L(x⋅λ)]=∑u∈WJ(−1)ℓ(u)Puy,x(1)=ny,x(1). The sum is finite and the coefficients with uy≰x vanish; in particular the support is y≤x with diagonal nx,x(1)=1.

3.1F1step 2.1algebra

Projective multiplicities. By the parabolic BGG reciprocity of [F1], (PJ(x⋅λ):ΔJ(y⋅λ))=[ΔJ(y⋅λ):L(x⋅λ)]=ny,x(1); the multiplicities are finite by the finite generalized-standard flag of [F1]. This proves the second identity.

3.2F1F4step 2.1algebra

Native longest-left labels. Let λanti=w0⋅λ. Dot-action composition gives y⋅λ=(yw0)⋅λanti and x⋅λ=(xw0)⋅λanti. If y has no left descent in J, each y−1αj is positive, so (yw0)−1αj=w0y−1αj is negative; hence z=yw0 is the longest representative in its left WJ-coset. Its shortest-left encoding is a=wJz=wJyw0, with inverse y=wJaw0; similarly b=wJxw0 has inverse x=wJbw0. Consequently the same multiplicity of step 2.1 is [ΔJ((wJa)⋅λanti):L((wJb)⋅λanti)]=nwJaw0,wJbw0(1)=N~a,b. This is an honest permutation of rows and columns, with support transported from wJaw0≤wJbw0, not a claim that N~a,b=na,b(1). Choosing the transported reversed order reads the same matrix as lower triangular.

4.1step 1.1step 2.1step 3.1step 3.2∎

The preceding steps prove the multiplicity formula, its projective form, the character identity and the reindexed version; the statement that only the regular ambient central character is assumed, and that no singular/parabolic equivalence or grading is used, is exactly the scope of the quoted suppliers [F1]--[F5].

Remarks

Open obligation. The proof uses the batch-14 interfaces thm-simple-verma-composition-multiplicities-are-kazhdan-lusztig-values and def-regular-integral-dot-orbit-labelling-convention through fact [F4], cited in steps 1.2, 2.1 and 3.2 for the ordinary multiplicity value Puy,x(1), and prop-translation-identifies-regular-integral-blocks through fact [F5], cited in step 2.1 for the labelled transport of the ordinary multiplicities across dominant regular integral blocks. All three were scaffolded but unwritten when this item was authored and are now written in items/ with statements matching these uses. The Step-3 decision remains escalated because only the owner may refresh an escalation receipt; the exact supplier IDs and consuming steps are recorded in the pair report. The ordinary KL formula is asserted only for the regular ambient central character; the parabolic highest-weight structure itself retains arbitrary integral blocks. No singular/parabolic equivalence, Koszul duality or grading is imported.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

The parabolic Kazhdan–Lusztig character formula

Statement

Assume the Axiom of Choice (The Axiom of Choice) and the dominant regular integral conventions of Parabolic category O multiplicities are parabolic Kazhdan–Lusztig values. Put N=(ny,x(1))y,x∈JW, with rows indexed by generalized standards and columns by simples in a Bruhat-compatible order (The parabolic Kazhdan–Lusztig bases and polynomials). Then N is upper unitriangular and, for each x, ch⁡L(x⋅λ)=∑y≥x(N−1)x,y ch⁡ΔJ(y⋅λ). Its inverse entries are integers computed by the finite signed chain sum of products of na,b(1); they are weighted incidence inverses, not generally ordinary Bruhat Mobius values. In the antidominant base λanti=w0⋅λ, set a=lf(y)=wJyw0 and encode the same weight as (wJa)⋅λanti. The reindexed matrix is N~a,b=Nlf−1(a),lf−1(b) and its inverse is reindexed by the same permutation; thus the same formula has reversed support in the transported order. This does not replace the entries by na,b(1). This is the ungraded character formula for the regular parabolic block, and constructs no Koszul or Ext grading.

Facts & Assumptions

Given: The regular parabolic block with generalized standards ΔJ(y⋅λ), simples L(x⋅λ) and projective covers PJ(x⋅λ) indexed by JW, the multiplicity matrix N=(ny,x(1)) of Parabolic category O multiplicities are parabolic Kazhdan–Lusztig values, and the Grothendieck group with its standard and simple bases (The Grothendieck group and character of O, Simple and standard bases of K0(O)).

[F1]

Multiplicity theorem: [Mp(y⋅λ):L(x⋅λ)]=ny,x(1) with support y≤x and diagonal one, and (PJ(x⋅λ):Mp(y⋅λ))=ny,x(1) (Parabolic category O multiplicities are parabolic Kazhdan–Lusztig values).

[F2]

The simple and generalized-standard classes are bases of the finite block Grothendieck group, characters are injective on it, and the standard and simple matrices are unitriangular; the blocks are finite (Parabolic category O is a highest weight category (4), Simple and standard bases of K0(O), BGG reciprocity).

[F3]

The Bruhat order on the finite set JW is a partial order, and the index range is finite, so the strictly upper triangular matrix U=N−I relative to such an order is nilpotent (Parabolic category O multiplicities are parabolic Kazhdan–Lusztig values, The parabolic Kazhdan–Lusztig bases and polynomials); the inverse entries are the finite alternating sums of products of the na,b(1) along strictly increasing chains (Inverse Kazhdan–Lusztig polynomials, The Kazhdan–Lusztig inversion formula provide the ordinary inversion-formula comparison).

Proof

technique · direct finite linear algebra on the unitriangular multiplicity matrix
1.1F1F2algebra

The matrix equation. In the Grothendieck group write My:=[ΔJ(y⋅λ)] and Lx:=[L(x⋅λ)]; by [F1] and the Jordan-Holder theorem in the finite-length parabolic block, My=∑x[ ΔJ(y⋅λ):L(x⋅λ) ]Lx=∑x≥yny,x(1)Lx. Since the simple classes form a basis by [F2], this is the matrix equation M=NL with N=(ny,x(1)).

1.2F1F3algebra

Unitriangularity and inversion. By [F1], nx,x(1)=1 and ny,x(1)=0 unless y≤x; the diagonal of N is one and U:=N−I is strictly upper triangular for the finite Bruhat-compatible order on JW. Hence U is nilpotent (Um=0 for m>#JW) and N−1=I−U+U2−⋯ is a finite sum of integer matrices; the (x,y)-entry of N−1 is supported on y≥x and equals the finite signed chain sum ∑m≥0(−1)m∑x=z0<z1<⋯<zm=y∏inzi,zi+1(1) over strictly increasing chains. These are weighted incidence inverses of the finite poset, not generally ordinary Mobius values.

2.1F1F2step 1.1step 1.2algebra

The character formula. Solving step 1.1 by step 1.2 gives L=N−1M with N−1 read as a matrix whose rows are indexed by the simples and whose columns are indexed by the generalized standards, that is, ch⁡L(x⋅λ)=∑y(N−1)x,ych⁡ΔJ(y⋅λ); since (N−1)x,y is supported on y≥x, this is exactly the displayed formula.

3.1F1step 1.2step 2.1algebra∎

Reindexing and conclusion. Use a=lf(y)=wJyw0 as the shortest-left encoding of the actual longest-left label yw0 in the antidominant base, so the same weight is (wJa)⋅λanti. The corrected dictionary in Parabolic category O multiplicities are parabolic Kazhdan–Lusztig values gives N~a,b=Nlf−1(a),lf−1(b). Reindexing both coordinates is conjugation by a permutation matrix, hence N~a,b−1=Nlf−1(a),lf−1(b)−1; therefore ch⁡L((wJa)⋅λanti)=∑bN~a,b−1ch⁡ΔJ((wJb)⋅λanti), with the original support transported by lf. Choosing the transported reversed order makes this matrix lower triangular. This preserves every character formula and constructs no Koszul or Ext grading.

5 · Examples, counterexamples and false statements

None yet.

Sources