• Aucun résultat trouvé

Modular representations of cyclotomic Hecke algebras of type G(r,p,n)

N/A
N/A
Protected

Academic year: 2021

Partager "Modular representations of cyclotomic Hecke algebras of type G(r,p,n)"

Copied!
16
0
0

Texte intégral

(1)

HAL Id: hal-00002867

https://hal.archives-ouvertes.fr/hal-00002867

Submitted on 17 Sep 2004

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Modular representations of cyclotomic Hecke algebras of type G(r,p,n)

Gwenaelle Genet, Nicolas Jacon

To cite this version:

Gwenaelle Genet, Nicolas Jacon. Modular representations of cyclotomic Hecke algebras of type G(r,p,n). International Mathematics Research Notices, Oxford University Press (OUP), 2006, pp.Art.

ID 93049. �hal-00002867�

(2)

ccsd-00002867, version 1 - 17 Sep 2004

Modular representations of cyclotomic Hecke algebras of type G ( r, p, n )

Gwena¨ elle Genet

and Nicolas Jacon

Abstract

We give a classification of the simple modules for the cyclotomic Hecke algebras overCin the modular case. We use the unitriangular shape of the decomposition matrices of Ariki-Koike algebras and Clifford theory.

1 Introduction

Letr, pand nbe integers such thatpdividesr. The complex reflection group of typeG(r, p, n) is defined to be the groups ofn×npermutation matrices such that the entries are either 0 orrth roots of unity and the product of the nonzero entries is a rp-root of unity.

In [2] and [5], Ariki and Brou´e-Malle have defined a Hecke algebra associated to each complex reflection groupG(r, p, n). It can be seen as a deformation of the algebra of the group G(r, p, n). According to a conjecture of Brou´e and Malle, such algebras, called cyclotomic Hecke algebras, should occur as endo- morphism algebras of the Lusztig induced character. The representation theory of cyclotomic Hecke algebras of typeG(r,1, n) (also known as Ariki-Koike alge- bras) is beginning to be well understood. They are cellular algebras, the simple modules have been classified in both semi-simple and modular cases and the decomposition matrices are known in characteristic 0 (see [21] for a survey of these results).

In [2], Ariki has shown that a Hecke algebra of typeG(r, p, n) can be consid- ered as the 0-component of a graded system for a Hecke algebra of typeG(r,1, n) with a special choice of parameters. As a consequence, in the semi-simple case, he has given a complete set of non isomorphic simple modules by using Clifford theory. In the modular case, partial results have been obtained by Hu in [17]

(see also [16] and [18] where the case r =p = 2 which corresponds to Hecke algebras of type Dn is studied). The main problem is that the restrictions of the simple modules for non semi-simple Ariki-Koike algebras are much more complicated to describe than for semi-simple ones.

The purpose of this paper is precisely to give a parametrization of the simple modules for the non semi-simple cyclotomic Hecke algebras of type G(r, p, n)

Address: UFR Math´ematiques - Case 7012 - Universit´e Denis Diderot - Paris 7 - 75251 PARIS Cedex 05, France. E-mail address: genet@math.jussieu.fr

Address: Universit´e de Caen, UFR des Sciences D´epartement de Math´ematiques Laboratoire LMNO, Campus II 14032 Caen cedex, France. E-mail address: ja- con@math.unicaen.fr

(3)

over C. It comes from the “canonical basic set” introduced by Geck and Rouquier in [13] and extended by the second author in [20]. This set induces a parametrization of the simple modules of non semi-simple Ariki-Koike alge- bras by some FLOTW multipartitions (this kind of multipartitions has been defined by Foda, Leclerc, Okado, Thibon and Welsh in [9]). We proceed in an analogous way than in [14, Theorem 2.1], using the unitriangular shape of the decomposition matrices of Ariki-Koike algebras and Clifford theory.

2 Cyclotomic Hecke algebras, Clifford theory

2.A Cyclotomic Hecke algebras

Letr, pandnbe integers such that pdividesr. We denote d:= r

p. LetRbe an integral ring,qa unit ofRand a sequencex= (x1, . . . , xd) of elements inR.

Denote Hq,xr,p,n(R), the Hecke R-algebra associated to the complex reflection group G(r, p, n), for the parameters q, x, defined by the following presenta- tion:

generators: a0, . . . , an,

relations: (a0−x1). . .(a0−xd) = 0 (ai−q)(ai+ 1) = 0, 1≤i≤n, a1a3a1=a3a1a3,

aiai+1ai=ai+1aiai+1, 2≤i≤n−1, (a1a2a3)2= (a3a1a2)2,

a1ai=aia1, 4≤i≤n,

aiaj=ajai, 2≤i < j ≤n, j≥i+ 2, a0a1a2= (q−1a1a2)2−ra2a0a1+ (q−1)

r−2

X

i=1

(q−1a1a2)1−ka0a1,

=a1a2a0, a0ai=aia0, 3≤i≤n.

We have the following special cases:

• ifr=p= 1,Hq,xr,p,n(R) is the Hecke algebra of typeAn−1,

• ifr= 2 and p= 1,Hq,xr,p,n(R) is the Hecke algebra of typeBn,

• ifr=p= 2,Hq,xr,p,n(R) is the Hecke algebra of typeDn,

• ifp= 1,Hq,xr,p,n(R) is the Ariki-Koike algebra defined in [4].

Suppose the ringR contains apth root of unity ηp and apth root yi ofxi, for each i ∈ [1, d]. We define a sequence Q = (Q1, . . . , Qr) of elements in R from the sequence x: for j ∈ [1, r], j = sp+t with s ∈ [0, d−1] and t ∈ [1, p], let Qj := ys+1ηpt−1. Then, Hq,Qr,1,n(R) is the R-algebra defined by

generators: T1, . . . , Tn,

relations: (T1p−x1). . .(T1p−xd) = 0, (Ti−q)(Ti+ 1) = 0, 2≤i≤n, T1T2T1T2=T2T1T2T1,

TiTi+1Ti=Ti+1TiTi+1, 2≤i≤n−1, TiTj=TjTi, 1≤i < j≤n, j≥i+ 2.

(4)

We use [2, Proposition 1.6] and identifya0 with T1p, a1 withT1−1T2T1 and ai with Ti, for i ∈ [2, n], such that Hq,xr,p,n(R) is considered as a subalgebra of Hq,Qr,1,n(R). Then by [15, §4.1], the Ariki-Koike algebraHq,Qr,1,n(R) is graded over the cyclotomic Hecke algebraHq,xr,p,n(R) by a cyclic group of orderp:

Hq,Qr,1,n(R) =

p−1

M

i=0

T1iHq,xr,p,n(R).

Put H:=Hq,Qr,1,n(R) andH :=Hq,xr,p,n(R). We can define an automorphism gHH ofH by

gHH(h) :=T1−1hT1, for allh∈H.

We also have an automorphismfHH ofHwhich is defined on the grading by fHH(T1jh) =ηjpT1jh, forj∈[0, p−1], h∈H.

The next theorem of Dipper and Mathas will be very useful in the following of this paper.

Theorem 2.1 ([8]) With the previous notations, assume that we have a parti- tion of Q:

Q=Q1a Q2a

. . .a Qs such that

fΓ(q, Q) = Y

1≤α<β≤s

Y

Qi∈Qα Qj∈Qβ

Y

−n<a<n

(qaQi−Qj)

is a unit of R. Then, Hq,Qr,1,n(R)is Morita-equivalent to the following algebra:

M

n1,...,ns>0 n1+...+ns=n

Hq,Qr 1

1,1,n1(R)⊗. . .⊗Hq,Qr s

s,1,ns(R) where for i∈[1, s],ri=|Qi|.

2.B Clifford theory

Lety1,y2,...,yd andv be indeterminates overC. Fori∈[1, d], put xi:=ypi

and fori∈[1, r], i=sp+t withs∈[0, d−1],t∈[1, p], ηp:= exp(2iπp ), Qi:=ηt−1p ys+1.

Form the sequencesx:= (x1, . . . ,xd) andQ:= (Q1, . . . ,Qr).

(5)

Let A := C[x1,x−11 , . . . ,xd,x−1d ,v,v−1], K := C(x1, . . . ,xd,v) its field of fractions.

Letθ:A→Cbe a ring homomorphism such thatCis the field of fractions ofθ(A).

We put fori∈[1, d],

xi:=θ(xi), fori∈[1, r],

Qi:=θ(Qi) and

v:=θ(v).

Note that for i ∈ [1, r], i = sp+t with s ∈ [0, d−1], t ∈ [1, p], then Qi:=ηpt−1ys+1, for a complex number ys+1 such thatys+1p =xs+1.

We also form the sequencesx:= (x1, . . . , xd) andQ:= (Q1, . . . , Qr).

We will note for shortH, resp. H, for the Ariki-Koike algebrasHv,Qr,1,n(K), resp. Hv,Qr,1,n(C) andH, resp. H, for the cyclotomic Hecke algebrasHv,xr,p,n(K), resp. Hv,xr,p,n(C).

Since H, resp. H, is graded overH, resp. H, we will give some results which comes from Clifford theory.

Let H be the Ariki-Koike algebra H, resp. H, and H be the cyclotomic Hecke algebra H, resp. H. In both cases, we will denote by Res the functor of restriction from the category of the H-modules to the category of the H- modules and by Ind the functor H⊗H−from the category of theH-modules to the category of theH-modules.

LetV be aH-simple module andU be a simple submodule of ResV. Put of,V := min{k∈N>0|fkV ≃V}

where f := fHH and for i ∈ [0, of,V −1], fiV is the H-module with the same underlying space as V and the structure of module is given by composing the original action ofHwithfi.

In the same way, let

og,U:= min{k∈N>0| gkU ≃U}

where g :=gHH and for i ∈ [0, og,U −1], giU is the H-module with the same underlying space asU with the action ofH twisted bygi.

Denote [X :Y] the multiplicity of a simple moduleY in the semi-simple mod- uleX.

Lemma 2.2 Denote for shortg:=gHH andf :=fHH. Every simpleH-moduleU appears as a direct summand of the restriction ResV of a simple H-module V with multiplicity-free. With such modules U and V, a conjugate giU of U, i ∈ [0, og,U −1], occurs in the restriction of the conjugate fjV of V, for all

(6)

j ∈[0, of,V −1]and the conjugates of V are the only simple H-modules whose restrictions contain such a giU. Besides

ResV ≃

of,V−1

M

i=0

fiV, (1)

IndU ≃

og,U−1

M

i=0

giU, (2)

p = og,Uof,V. (3)

Proof: It is easy to see that

ResV ≃[ResV :U]

og,U−1

M

i=0

giU, (4)

(see [6, Theorem 11.1]). Using [14, Proposition 2.2], we get

IndU ≃[IndU :V]

of,V−1

M

i=0 fiV

as well as p

of,Vog,U

= [ResV :U][IndU :V]. (5) As ResV is g-stable, [6, Proposition 11.14] implies that EndH(Ind ResV) is graded over EndH(ResV) by a cyclic group of orderp. By (4) and (5), the di- mension of EndH(Ind ResV) overCis opf,V2 and by the grading over EndH(ResV), this dimension is equal top[ResV :U]2og,U. By comparison with (5), it follows that

[ResV :U] = [IndU :V]. (6)

Now, by definition of of,V, for shorto, there exists anH-automorphism t:V →foV.

So, tpo is anH-isomorphism. By Schur’s lemma,tpo is a scalar. There exists a complex numberc such that (ct)po = IdV. Then

ResV ≃

p o−1

M

i=0

Ker(t

c−ηopiIdV).

It is easy to see that for each i ∈[0,po−1], Ker(ct−ηpoiIdV) is an H-module isomorphic to giKer(tc −IdV). So the semi-simple module ResV decomposes into more than po simple modules and by (5) by less than po simple modules.

Therefore ResV decomposes into exactly po simple modules and [IndU :V] = 1 so [ResV :U] = 1 by (6).

As the module induced from a simpleH-module toHis semi-simple, the asser- tions of the Lemma are proved.

(7)

2.C The semi-simple case

We keep the notations of the previous subsection.

We study the representation theory of the algebras H := Hv,Qr,1,n(K) and H:=Hv,xr,p,n(K). By [1, Main Theorem],His a split semi-simple algebra. This implies thatH is also split semi-simple.

A complete set of non isomorphic modules forHhas been given in [4] and [7].

Let Πrn be the set of r-partitions of n. We say that λ = λ(i)

i∈[1,r] is a r- partition of sizenif for everyi,λ(i) is a partition and Σri=1|λ(i)|=n.

For each r-partition λ ∈ Πrn, we can associate an Hv,Qr,1,n(A)-module Sv,Q(λ) called a “Specht module” that is free over A. Here, we use the definition of

“classical” Specht module contrary to [7] where the dual Specht modules are defined. We have the following theorem.

Theorem 2.3 ([4],[7]) The following set is a complete set of non isomorphic absolutely irreducible Hv,Qr,1,n(K)-modules:

nSKv,Q(λ) :=K⊗ASv,Q(λ)| λ∈Πrno .

For eachλ∈ Πrn, the set of standard tableaux of shape λis a basis of the underlyingC-vector space ofSKv,Q(λ) (see [4]). Using [4, Propositions 3.16, 3.17]

and some combinatorial properties, it is easy to get the

Proposition 2.4 Letλ∈Πrn. The map

hv,Qλ :fHHSKv,Q(λ)→SKv,Q ̟(λ)

(7) that sends a standard λ-tableau T = (Ti)i∈[1,r] to the standard ̟(λ)-tableau

̟(T) := (T̟−1(i))i∈[1,r]is anH-isomorphism where̟is the permutation ofSr defined byQ̟(i):=ηpQi,for alli∈[1, r]and̟(λ) = (λ(̟−1(1)), . . . , λ(̟−1(r))).

Define

oλ:= min{k∈N>0k(λ) =λ}.

It is clear that oλ=of,Sv,Q

K (λ)and ResSKv,Q(λ)≃

p −1

M

i=0

Ker(hoλλ −ηopλiIdSv,Q

K (λ)), wherehoλλ:fSKv,Q(λ)→SKv,Q λ

sends a standardλ-tableauTto the standard λ-tableau̟oλ(T). Fori∈[0,op

λ −1], put

SKv,Q(λ, i) := Ker(hoλλ−ηopλiIdSv,Q

K (λ)). (8)

(8)

LetL be a set of representatives ofr-partitions for the action of the cyclic group generated by̟ over Πrn. It follows from Lemma 2.2, the

Proposition 2.5 The set

{SKv,Q(λ, i) ; i∈[0, p oλ

−1], λ∈ L}

is a complete set of non isomorphic simple H-modules.

IfTis a standardλ-tableau, then forj∈[0,op

λ −1], the element (T, j) :=

p −1

X

i=0

ηp−oλij̟oλi(T) belongs to the eigenspace Ker(hoλλ−ηopλjIdSv,Q

K (λ)). It comes from the fact that T = oλ

p

p −1

X

j=0

(T, j), that for every j ∈ [0,op

λ −1], the eigenspace Ker(hoλλ − ηopλjIdSv,Q

K (λ)) is generated by the (T, j), Ta standard λ-tableau. So we have just described the simpleH-modules, see also [2].

The following part is now concerned with the non semi-simple case: simple Hq,Qr,1,n(C)-modules andHq,xr,p,n(C)-modules.

2.D The non semi-simple case

We keep the notations of§2.B.

2.D.1 Decomposition maps

The aim of this part is to define the decomposition maps for H := Hq,Qr,1,n(C) and H := Hq,xr,p,n(C). Recall that A := C[x1,x−11 , . . . ,xd,x−1d ,v,v−1], K :=

C(x1, . . . ,xd,v) and thatθ:A→Cis a ring homomorphism.

Using [10], there exists a discrete valuation ringOwith maximal idealI(O) such that A ⊂ O and I(O)∩A = ker(θ). Let k0 := O/I(O) be the residue field of O. Denote by : O →k0 the canonical map. We obtain the algebras Hq,Qr,1,n(k0) andHq,xr,p,n(k0). The fieldk0 can be considered as an extension of C that is the quotient field of θ(A). By [12, Lemma 7.3.4], there is an isomor- phism between the Grothendieck groups of finitely generated H-modules and Hq,Qr,1,n(k0)-modules (resp. H-modules andHq,xr,p,n(k0)-modules):

R0(H)≃R0(Hq,Qr,1,n(k0)) and R0(H)≃R0(Hq,xr,p,n(k0)).

As a consequence, we obtain well-defined decomposition maps by choosingO- forms for the simpleHq,Qr,1,n(K)-modules (resp. Hq,xr,p,n(K)-modules) and reducing them moduloI(O):

dr,1,n:R0(H)→R0(H),

(9)

dr,p,n:R0(H)→R0(H).

For a simpleH-moduleV andW a simpleH-module, there exist non neg- ative integers d(1)V,M with M a simple H-module and d(p)W,N with N a simple H-module such that:

dr,1,n([V]) = X

M∈Irr(H)

d(1)V,M[M]

and dr,p,n([W]) = X

N∈Irr(H)

d(p)W,N[N].

By using the same argument as in [11, Lemma 5.2] (see also [12, Theorem 7.4.3.c]), we obtain the following result.

Proposition 2.6 The following diagram commutes:

R0(H) −−−−→Res R0(H)

dr,1,n

 y

 yd

r,p,n

R0(H) −−−−→Res R0(H)

Moreover, for any simple H-moduleV and any simpleH-moduleM, we have d(1)V,M =d(1)fH

HV,f HHM

,

and for any simpleH-moduleW and any simpleH-moduleN, we have:

d(p)W,N =d(p)gH HW,gHHN

.

In the following, we will see that under some additionnal hypotheses, the above property leads to some interesting results about the decomposition map of cyclotomic Hecke algebras of typeG(r, p, n) and about the simpleH-modules.

2.D.2 Simple modules of non semi-simple Ariki-Koike algebras We will work under the following hypothesis. We assume thatv is a primitive root of unity,

v:=ηe= exp(2iπ e ), for a positive integere >1. Let

f := gcd(e, p), e=f e, p=f p, thenηfppee.

By the definition of Q and Theorem 2.1, without lost of generality, we can suppose that there exist integers v1 ≤. . . ≤ vδ that belong to [0, e−1] such that Qconsists of the complex numbersηeviηjp, whereiranges over [1, δ] andj ranges over [0, p−1]. Thenr=δf p and we can splitQintop sets:

Q=Q1a . . .a

Qp,

(10)

withQj formed by theηveiηplp+j,i∈[1, δ] andl∈[0, f−1]. We considerQj as an ordered sequence

Qj = (ηj−1p ηev1, . . . , ηpj−1ηevδ, ηpp+j−1ηve1, . . . , ηpp+j−1ηevδ, . . . η(f−1)pp +j−1ηev1, . . . , η(f−1)pp +j−1ηevδ). (9)

Remark: First note that, forj∈[1, p],Qjj−1p Q1and

Q1= (ηve1, . . . , ηveδ, ηve1+e, . . . , ηevδ+e, . . . , ηev1+(f−1)e, . . . , ηevδ+(f−1)e), withv1≤. . .≤vδ ≤v1+e ≤. . .≤vδ+ (f−1)e.

Note also that the quotient of two elements that belong toQj, forj∈[1, p], is a power ofv=ηewhile the quotient of an element ofQj by an element ofQl, l∈[1, p],l6=j is not such a power, by definition off.

If we order the elements of Q by the ordered elements of Q1, then the ordered elements ofQ2, etc, Theorem 2.1 implies that the Ariki-Koike algebra H is Morita equivalent to the algebra

M

n1,...,np′ ≥0 n1+...+np=n

Hq,Qf δ,1,n1

1(C)⊗. . .⊗Hq,Qp

f δ,1,np(C).

So it is clear thatH is also Morita equivalent to the algebra M

n1,...,np′ ≥0 n1+...+np=n

Hq,Qf δ,1,n1

1(C)⊗. . .⊗Hq,Qf δ,1,n1

p(C).

We orderQand we split it intop sets following what we have done forQ:

Q=Q1a . . .a

Qp.

2.D.3 Parametrization of simple H-modules

We can now give a parametrization for the simple H-modules. By [3, Theo- rem 2.5], the simple modules of the Ariki-Koike algebraHq,Qf δ,1,n1

i(C),i∈[1, p], are the quotient modules Dv,QC 1(λ) :=SCv,Q1(λ)/J(SCv,Q1(λ)) of the specialisa- tions of the Specht modules SCv,Q1(λ) := C⊗Sv,Q1(λ) by their radical of Ja- cobson, with λa Kleshchev f δ-partition (for the parametersQ1) of sizeni. So the simple modules ofHare labelled by thep-tuples of Kleshchevf δ-partitions (associated toQ1) (λ1, . . . , λp) with

p

X

i=1

i|=n. We will denote this set by Λ0 and forλ∈Λ0,Dv,QC (λ) the corresponding simpleH-module.

(11)

In [20], another parametrization for the simpleHq,Qf δ,1,n1

i(C)-modules has been found by using Lusztig’s a-function (see Theorem 2.7 below). Following this paper, we will associate to each r-partition an a-value. To do this, we first describe the a-value of af δ-partitionµ= (µ(1), µ(2), ..., µ(f δ)).

Fork= 1, ..., δ,s= 1, ..., f andj= (s−1)δ+kwe put wj :=vk+ (s−1)e so thatwj is the power of thejth-component of Q1.

Then, forj:= 1, ..., f δ, we define:

m(j):=wj−je f δ +e and fors= 1, ..., n, we put:

Bs′(j):=µ(j)s−s+n+m(j)

where we use the convention that µ(j)p := 0 ifpis greater than the height of µ(j). Forj = 1, ..., f δ, letB′(j)= (B′(j)1 , ..., B′(j)n ). Then, we define:

a(1)f δ(µ) := X

0≤i≤j<f δ (a,b)∈B′(i)×B′(j)

a>bif i=j

min{a, b} − X

1≤i,j≤f δ a∈B′(i)

1≤k≤a

min{k, m(j)}+g(n).

whereg(n) is a rational number which only depends on them(j) and onn(the expression ofg is given in [20]).

Let λ ∈ Πrn, write λ = (λ(1), . . . , λ(r)) as λ = (λ1, . . . , λp) where, for i∈[1, p],λi = λ(f δ(i−1) + 1), . . . , λ(f δ(i−1) +f δ)

. Finally, we define

a(1)r (λ) :=

p

X

i=1

a(1)f δi).

To each Specht module SKv,Q1(λ) of Hq,Qf δ,1,n1 (K), n ∈ [0, n] andλ an f δ- partition, we also associate ana-value

a(1)f δ SKv,Q1(λ)

:=a(1)f δ(λ).

To each simpleHq,Qf δ,1,n1 (C)-module DCv,Q1(λ), n ∈[0, n] andλa Kleshchev f δ-partition of sizen, we again associate ana-value

a(1)f δ(Dv,QC 1(λ)) := min{a(1)f δ(µ)| µ∈Πf δn, d(1)

SKv,Q1(µ),Dv,QC 1(λ)6= 0}.

Theorem 2.7 ([19, Theorem 2.3.8]) Let λ be a Kleshchev f δ-partition of n for a non-negative integer n. There exists a unique µ := κ(λ) such that a(1)f δ(DCv,Q1(λ)) =a(1)f δ(µ) andd(1)

SKv,Q1(µ),Dv,QC 1(λ)6= 0. The function κis a bijec- tion between the set of the Kleshchev f δ-partitions of size n and the set of the FLOTW f δ-partitions of sizen.

(12)

Recall what is a FLOTWf δ-partitionλ= λ(1), . . . , λ(f δ)

of sizenassoci- ated to the parametersv andQ1, see [9]. Denote each partitionλ(i), i∈[1, f δ], as (λ(i)1, λ(i)2, . . .). The multipartitionλsatisfies

1. fori∈[1, f],j ∈[1, δ−1],kpositive integer, λ (i−1)δ+j

k ≥λ (i−1)m+j+ 1

k+vj+1−vj, 2. fori∈[1, f−1],kpositive integer,

λ(iδ)k≥λ(iδ+ 1)k+v1+e−vδ, 3. for kpositive integer,

λ(f δ)k ≥λ(1)k+v1+e−vδ, as well as

4. to each node of the diagram ofλwhich is located in the ath row and the bth column of thecth partition ofλ(for two poditive integersaandband 1 ≤ c ≤ f δ), we associate its residue ηb−a+ve c. Then, for any positive integerk, the cardinality of the set of the residues associated to the nodes in both thekth columns and the right rims of the Young diagrams ofλis note.

In the same way, to each Specht module SKv,Q(λ) of H, λ ∈ Πrn, λ = (λ1, . . . , λp), we can definea(1)r (SKv,Q(λ)) :=a(1)r (λ) that is the sum

p

X

i=1

a(1)f δi).

To each simpleH-moduleDv,QC (λ),λ∈Λ0, we associate thea-value a(1)r (Dv,QC (λ)) := min{a(1)r (µ)|µ∈Πrn, d(1)

Sv,QK (µ),Dv,QC (λ)6= 0}. (10) With Theorem 2.1, it is clear that there exists a unique µ:= κ(λ) such that a(1)r (DCv,Q(λ)) =a(1)r (µ) andd(1)

Sv,QK (µ),Dv,QC (λ)6= 0. In fact, κ(λ) = (κ(λ1), . . . , κ(λp)) ifλ= (λ1, . . . , λp).

The function κ is a bijection between Λ0 and Λ1 the set of the p-tuples of FLOTWf δ-partitions (λ1, . . . , λp) of size

p

X

i=1

i|=n.

2.D.4 Preliminary results

In order to prove the main Theorem, we will need some preliminary results.

First, recall that we have ordered Q by the ordered elements of Q1, then by the ordered elements ofQ2 etc as in §2.D.2. Let ̟ be the permutation of Sr defined in Proposition 2.4 and let

λ:= (λ1[1], ..., λ1[f δ], λ2[1], ..., λ2[f δ], ..., λp[1], ..., λp[f δ])

be anr-partition. Hence, following the notations of the previous paragraph, we haveλj[i] =λ(f δ(j−1) +i). Then, it is easy to verify that

(13)

• fori∈[1,(p−1)f δ], then ̟(i) =i+f δ,

• fori∈[r−f δ+ 1, r−δ], then̟(i) =i−r+ (f+ 1)δ,

• fori∈[r−δ+ 1, r], then ̟(i) =i−r+δ.

Then, by Proposition 2.4, we have:

SKv,Q(λ)≃SKv,Q ̟(λ)

where ̟(λ) = (λp[f δ−δ+ 1], . . . , λp[f δ],λp[1], . . . , λp[f δ−δ], λ1[1], . . . , λ1[f δ], . . . , λp−1[1], . . . , λp−1[f δ]).

Proposition 2.8 Letλ∈Πrn, thena(1)r (SKv,Q(λ)) =a(1)r (fHHSKv,Q(λ)). Besides, if λ∈Λ0, thena(1)r (Dv,QC (λ)) =a(1)r (fHHDv,QC (λ)).

Proof: By the definition of thea-value (see§2.D.3), we have:

a(1)r (Sv,QK (λ)) :=

p

X

i=1

a(1)f δi[1], ..., λi[f δ])

Thus, by using the above properties, it is sufficient to show thata(1)f δp[1], . . . , λp[f δ])=a(1)f δp[f δ−δ+ 1], . . . , λp[f δ], λp[1], . . . , λp[f δ−δ]).

Using the notations of §2.D.3, for i = 1, ..., δ and j = 1, ..., f we write m(i)[j] :=m((j−1)δ+i). Then, all we have to do is to prove thatm(i)[j] doesn’t depend onj. We havem(i)[j] =vi−((j−1)δ+i)e

f δ + (j−1)e=vi−ie

f because ef =e. Hence, we obtain the desired result.

With the definition (10), it is trivial that the previous result about Specht modules implies that, for anyλ∈Λ0,a(1)r (Dv,QC (λ)) =a(1)r (fHHDv,QC (λ)).

With Proposition 2.8 and Lemma 2.2, we can associate ana-value to each simpleH-moduleU. If U appears in the restriction ResSKv,Q(λ), for λ∈Πrn, put

a(p)r (U) :=a(1)r SKv,Q(λ) .

In the same way, we can associate ana-value to each simpleH-moduleW. IfW appears in the restriction ResDv,QC (λ), forλ∈Λ0, then

a(p)r (W) :=a(1)r (Dv,QC (λ)).

Another useful proposition is the following one.

(14)

Proposition 2.9 Let λ ∈ Λ1, then ̟(λ) ∈ Λ1, where ̟ is defined in the Proposition 2.4.

Proof: Writeλ∈Λ1 as λ= (λ1, . . . , λp) where λi = (λi[1], . . . , λi[f δ]) is a FLOTWf δ-partition forQ1. We will see that̟(λ)∈Λ1so the result will be proved. The r-partition ̟(λ) is equals to (λp[f δ−δ+ 1] , . . . , λp[f δ],λp[1]

, . . . , λp[f δ−δ], λ1[1], . . . , λ1[f δ], . . . λp−1[1], . . . , λp−1[f δ]). It belongs to Λ1 if and only if (λp[f δ−δ+ 1] , . . . , λp[f δ], λp[1] , . . . , λp[f δ−δ]) is a FLOTW f δ-partition. It is easy to verify that the three first conditions hold.

Now, the set of the residues associated to the nodes in both thekthcolumns and the right rims of the Young diagrams of this last multipartition consists of the residues of λp but multiplied by ηee. So the fourth condition about FLOTW multipartitions hold as well as the Proposition.

3 A parametrization of the simple H

-modules

We keep the above notations. The aim of this section is to describe a parametriza- tion of the simpleH-modules, withH=Hv,Qr,p,n(C). The proof is highly inspired by [14, Theorem 2.1].

Theorem 3.1 For all simple H-module N, there exists a simple H-module WN such that

dr,p,n([WN]) = [N] + X

a(p)r (L)<a(p)r (N)

dWN,L[L],

anda(p)r (N) =a(p)r (WN).

Let Λ1 := {WN | N ∈ Irr(H)}. Then Λ1 consists in the SKv,Q(λ, i), with λ∈Λ1∩ L,i∈[0,op

λ−1]with the notations of the Proposition 2.5 and the map N ∈Irr(H)7→WN ∈Λ1 is bijective. SoΛ1 is a parametrization of the simple H-modules.

Proof : Fix a simpleH-moduleN. By Lemma 2.2, there existsλ∈Λ0such that N appears in the restriction ResDCv,Q(λ). We have

dr,1,n([SCv,Q κ(λ)

]) = [DCv,Q(λ)] + X

a(p) r (Dv,Q

C (µ))<

a(p) r (Dv,Q

C (λ))

d(1)

Sv,QK (κ(λ)),DCv,Q(µ)[DCv,Q(µ)],

(11) anda(1)r SCv,Q(κ(λ))

=a(1)r (Dv,QC (λ)). With the notations of (8) and Proposi- tion 2.6, this equality implies that

p oκ(λ)−1

X

i=0

dr,p,n([SCv,Q(κ(λ), i)]) =

p of H

H,Dv,Q C (λ)

−1

X

i=0

[(gHH)iN] + [N], (12)

(15)

whereNis a sum ofH-modules,D, witha-valuea(p)r (D)< a(p)r (Dv,QC (λ)) and, for alli,j,a(p)r (SCv,Q(κ(λ), i)) =a(p)r ((gHH)iN).

Denote by τ(λ)∈Λ0, the multipartition defined byDCv,Q(τ(λ)) :=fHHDv,QC (λ).

If we twist the action ofHbyfHH, the equality (11) gives thatκ(τ(λ)) =̟(κ(λ)) by Proposition 2.9 and by definition ofκ. As this map is bijective, it is clear that

ofH

H,Sv,QK (κ(λ)) =ofH

H,DCv,Q(λ)

and therefore for alli∈[0,op

κ(λ) −1], ogH

H,Sv,QK (κ(λ),i)=ogH

H,N

with the last equality of Lemma 2.2.

Recall that for i∈[0,oκ(λ)p −1],SKv,Q(κ(λ), i) is conjugate to SKv,Q(κ(λ),0) by (gHH)i. The equality (12) implies that there exists an unique i∈ [0,oκ(λ)p −1]

such that

dr,p,n([SCv,Q(κ(λ), i)]) = [N] + [N′′], (13) where N′′ is a sum of H-modules, D, with a-value a(p)r (D) < a(p)r (N). This construction makes clear all the statements of the Theorem.

Remark: The above theorem is only concerns with the case wherevis a primitive eth-root of unity. Ifv isn’t a root of unity, it is readily checked that an analogue of this theorem holds by replacing Λ1by the set of Kleshchev multipartitions Λ0 at e=∞. The proof of this result can be easily obtained following the outline of [14].

References

[1] S.Ariki, On the Semi-simplicity of the Hecke Algebra of (Z/rZ)≀Sn. J.

Algebra, 169 : 216-225, 1994.

[2] S.Ariki,Representation theory of a Hecke algebra ofG(r, p, n). J. Algebra, 177 : 164-185, 1995.

[3] S. Ariki, On the classification of simple modules for cyclotomic Hecke algebras of type G(m,1,n) and Kleshchev multipartitions. Osaka J. Math., 38, no. 4, 827-837, 2001.

[4] S.Ariki, K.Koike, A Hecke algebra of (Z/rZ)≀Sn and construction of its irreducible representations. Adv. Math., 106, no.2 : 216-243, 1994.

[5] M.Brou´e, G.Malle,Zyklotomische Heckealgebren. Ast´erisque, 212 : 119- 189, 1993.

[6] C. Curtis, I. Reiner, Methods of Representation Theory vol.1,2. Wiley, New York, 1990.

[7] R. Dipper, G. James, A.Mathas, Cyclotomic q-Schur algebras. Math.

Z., 229, no. 3 : 385-416., 1998.

(16)

[8] R. Dipper, A. Mathas, Morita equivalences of Ariki-Koike algebras.

Math.Z., 240, no. 3 : 579-610, 2002.

[9] O. Foda, B.Leclerc, M. Okado, J-Y Thibon, T. Welsh, Branching functions ofA(1)n−1and Jantzen-Seitz problem for Ariki-Koike algebras. Adv.

Math., 141, no. 2 : 322-365, 1999.

[10] M.Geck, Representations of Hecke algebras at roots of unity. S´eminaire Bourbaki. Vol. 1997/98. Ast´erisque No. 252 : 33-55 : 1993.

[11] M. Geck, On the representation theory of Iwahori-Hecke algebras of ex- tended finite Weyl groups. Representation theory (electronic journal), 4 : 370-397, 2000.

[12] M.Geck, G.Pfeiffer,Characters of Finite Coxeter Groups and Iwahori- Hecke Algebras. Oxford Science Publications, Oxford University Press, 2000.

[13] M. Geck, R.Rouquier, Filtrations on projective modules for Iwahori- Hecke algebras. Modular representation theory of finite groups (Char- lottesville, VA, 1998), de Gruyter, Berlin : 211-221, 2001.

[14] G.Genet,On the decomposition matrices for graded algebras. J. Algebra, 274 : 523-542, 2004.

[15] G. Genet, Inclusion d’alg`ebres de Hecke et nombres de d´ecomposition.

Ph.D thesis, Paris VII, 2004.

[16] J.Hu,Crystal bases and simple modules for Hecke algebra of type Dn. J.

Algebra, 267 : 7-26, 2003.

[17] J.Hu, Modular representations of Hecke algebras of typeG(p, p, n). J. Al- gebra, 274, 446-490, 2004.

[18] N.Jacon,Sur les repr´esentations modulaires des alg`ebres de Hecke de type Dn. J. Algebra, 274 : 607-628, 2004.

[19] N.Jacon,Repr´esentations modulaires des alg`ebres de Hecke et des alg`ebres de Ariki-Koike . Ph.D thesis, Lyon I, 2004.

[20] N. Jacon, On the parametrization of the simple modules for Ariki-Koike algebras at roots of unity. preprint.

[21] A. Mathas,The representation theory of the Ariki-Koike and cyclotomic q-Schur algebras. Adv. Studies Pure Math. to appear.

Références

Documents relatifs

The main purpose of this note is to give explicit semi-simple decompositions of D R -modules with a finite support, where D R is the ring of differential operators associated with

We note that Boys and Mathas [Bo, BoMa] already studied a restriction of the isomorphism of [BrKl] to the fixed point subalgebra of an automorphism, in order to study

We prove that cyclotomic Yokonuma–Hecke algebras of type A are cyclotomic quiver Hecke algebras and we give an explicit isomorphism with its inverse, using a similar result of

The main point of this paper is to prove a similar decomposition theorem for some generalisations of quiver Hecke algebras and hence obtain an analogue of the Dipper–Mathas

The existence of basic sets has been shown for all Hecke algebras of finite Weyl groups in the case where the associated “weight function” is positive ([10]), for special cases

The idea is to use results of Ariki which give an interpretation of the decom- position map using the theory of canonical basis of Fock spaces. This theory can.. not be applied to

Rapha¨ el Rouquier, Gunter Malle, Michel B., Sungsoon Kim,. &amp;

In particular, we conjecture that decomposition matrices of cyclotomic v-Schur algebras are obtained by specializing at q = 1 some transition matrices between the standard basis and