SciELO - Scientific Electronic Library Online

 
vol.48 número3Non-homogeneous N-Koszul AlgebrasClassification of split TTF-triples in module categories índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados

Revista

Articulo

Indicadores

  • No hay articulos citadosCitado por SciELO

Links relacionados

  • No hay articulos similaresSimilares en SciELO

Compartir


Revista de la Unión Matemática Argentina

versión impresa ISSN 0041-6932versión On-line ISSN 1669-9637

Rev. Unión Mat. Argent. v.48 n.3 Bahía Blanca  2007

 

On pointed Hopf algebras associated with alternating and dihedral groups

Nicolás Andruskiewitsch and Fernando Fantino

Abstract. We classify finite-dimensional complex pointed Hopf algebras with group of group-like elements isomorphic to 𝔸5 . We show that any pointed Hopf algebra with infinitesimal braiding associated with the conjugacy class of π ∈ 𝔸n is infinite-dimensional if the order of π is odd except for π = (12 3) in 𝔸4 . We also study pointed Hopf algebras over the dihedral groups.

2000 Mathematics Subject Classification. 16W30; 17B37

This work was partially supported by Agencia Córdoba Ciencia, ANPCyT-Foncyt, CONICET and Secyt (UNC)

Dedicado a María Inés Platzeck en sus # 𝔸5 años

 

Introduction

In this article, we continue the work of [AZAF] on the classification of finite-dimensional complex pointed Hopf algebras H with G = G (H ) non-abelian. We follow the Lifting Method - see [AS2] for a general reference; in particular, we focus on the problem of determining when the dimension of the Nichols algebra associated with conjugacy classes of G is infinite. The paper is organized as follows. In Section 1, we review some general facts on Nichols algebras corresponding to finite groups. We discuss the notion of absolutely real element of a finite group in subsection 1.2. We then provide generalizations of [AZ, Lemma 1.3], a basic tool in [AZAF], see Lemmata 1.8 and 1.9. Section 2 is devoted to pointed Hopf algebras with coradical ℂ 𝔸 n . We prove that any finite-dimensional complex pointed Hopf algebra H with G (H ) ≃ 𝔸5 is isomorphic to the group algebra of 𝔸5 ; see Theorem 2.6. This is the first finite non-abelian group G such that all pointed Hopf algebras H with G (H ) = G are known. We also prove that dim 𝔅 (Oπ, ρ) = ∞ , for any π in 𝔸n of odd order, except for π = (1 23 ) or π = (13 2) in 𝔸 4 - see Theorem 2.3. This last case is particularly interesting. It corresponds to a "tetrahedron" rack with constant cocycle ω ∈ 𝔾3 - 1 . The technique in the present paper does not provide information on the corresponding Nichols algebra. We also give partial results on pointed Hopf algebras with groups 𝔸4 and 𝔸6 , and on Nichols algebras 𝔅 (O π,ρ) , with |π| even. In Section 3, we apply the technique to conjugacy classes in dihedral groups. It turns out that it is possible to decide when the associated Nichols algebra is finite-dimensional in all cases except for M (Ox,sgn ) (if n is odd), or M (Ox, sgn ⊗ sgn) or M (Ox, sgn ⊗ɛ) or M (Oxy,sgn ⊗ sgn) or M (Oxy, sgn ⊗ɛ) (if n is even). See below for undefined notations. We finally observe in Section 4 that there is no finite-dimensional Hopf algebra with coradical isomorphic to the Hopf algebra  J (ℂ 𝔸5 ) discovered in [Ni], except for  J (ℂ𝔸5 ) itself.

1. Generalities

For s ∈ G we denote by  s G the centralizer of s in G . If H is a subgroup of G and s ∈ H we will denote  H O s the conjugacy class of s in H . Sometimes we will write in rack notations x ⊳ y = xyx -1 , x , y ∈ G . Also, if (V, c) is a braided vector space, that is c ∈ GL (V ⊗ V ) is a solution of the braid equation, then 𝔅 (V ) denotes its Nichols algebra.

We denote by 𝔾n the group of n -th roots of 1 in ℂ .

1.1. Preliminaries. Let G be a finite group, O a conjugacy class of G , s ∈ O fixed, ρ an irreducible representation of Gs , M (O,ρ ) the corresponding irreducible Yetter-Drinfeld module. Let t = s 1 , …, t M be a numeration of O and let gi ∈ G such that gi ⊳ s = ti for all 1 ≤ i ≤ M . Then M (O,ρ ) = ⊕1 ≤i≤M gi ⊗ V . Let giv := gi ⊗ v ∈ M (O, ρ) , 1 ≤ i ≤ M , v ∈ V . If v ∈ V and 1 ≤ i ≤ M , then the coaction and the action of g ∈ G are given by

δ(giv) = ti ⊗ giv, g ⋅ (giv) = gj(γ ⋅ v),

where ggi = gjγ , for some 1 ≤ j ≤ M and γ ∈ Gs . The Yetter-Drinfeld module M (O, ρ ) is a braided vector space with braiding given by

c(giv ⊗ gjw) = ti ⋅ (gjw ) ⊗ giv = gh(γ ⋅ v) ⊗ giv
(1)

for any 1 ≤ i,j ≤ M , v,w ∈ V , where tigj = gh γ for unique h , 1 ≤ h ≤ M and γ ∈ Gs . Since s ∈ Z (Gs) , the Schur Lemma says that

s acts by a scalar qss on V.
(2)

Let G be a finite non-abelian group. Let O be a conjugacy class of G and let ρ be an irreducible representation of the centralizer  s G of a fixed s ∈ O . Let M (O, ρ) be the irreducible Yetter-Drinfeld module corresponding to (O, ρ) and let 𝔅 (O, ρ) be its Nichols algebra. As explained in [AZAF], we look for a braided subspace U of M (O,ρ ) of diagonal type such that the dimension of the Nichols algebra 𝔅 (U ) is infinite. This implies that the dimension of 𝔅 (O, ρ) is infinite too.

Lemma 1.1. If W is a subspace of V such that c(W ⊗ W ) = W ⊗ W and dim 𝔅 (W ) = ∞ , then dim 𝔅 (V) = ∞ .

Recall that a braided vector space (V,c) is of diagonal type if there exists a basis v1,...,vθ of V and non-zero scalars qij , 1 ≤ i,j ≤ θ , such that c(v ⊗ v ) = q v ⊗ v i j ij j i , for all 1 ≤ i,j ≤ θ . A braided vector space (V, c) is of Cartan type if it is of diagonal type and there exists aij ∈ ℤ , - |qii| < aij ≤ 0 such that  aij qijqji = qii for all 1 ≤ i ⁄= j ≤ θ ; by |qii| we mean ∞ if qii is not a root of 1, otherwise it means the order of qii in the multiplicative group of the units in ℂ . Set aii = 2 for all 1 ≤ i ≤ θ . Then (aij)1≤i,j≤ θ is a generalized Cartan matrix.

Theorem 1.2. ([H, Th. 4], see also [AS1, Th. 1.1]). Let (V, c) be a braided vector space of Cartan type. Then dim 𝔅 (V ) < ∞ if and only if the Cartan matrix is of finite type.

We say that s ∈ G is real if it is conjugate to s-1 ; if s is real, then the conjugacy class of s is also said to be real. We say that G is real if any s ∈ G is real.

The next application of Theorem 1.2 was given in [AZ]. Let G be a finite group, s ∈ G , O the conjugacy class of s ,  s ρ : G → GL (V ) irreducible;  × qss ∈ ℂ was defined in (2).

Lemma 1.3. Assume that s is real. If dim 𝔅 (O,ρ ) < ∞ then q = - 1 ss and s has even order.

If  -1 s ⁄= s , this is [AZ, Lemma 2.2]; if  2 s = id then qss = ±1 but qss = 1 is excluded by Lemma 1.1.

The class of real groups includes finite Coxeter groups. Indeed, all the characters of a finite Coxeter group are real valued, see subsection 1.2 below, and [BG] for H 4 . Therefore, we have:

Theorem 1.4. Let G be a finite Coxeter group. If s ∈ G has odd order, then dim 𝔅 (Os, ρ) = ∞ , for any ρ ∈ ^Gs .

1.2. Absolutely real groups. Let G be a finite group. We say that s ∈ G is absolutely real if there exists an involution σ in G such that σsσ = s-1 . If this happens, any element in the conjugacy class of s is absolutely real and we will say that the conjugacy class of s is absolutely real. We say that G is absolutely real if any s ∈ G is so. The finite Coxeter groups are absolutely real. Indeed,

  1. the dihedral groups are absolutely real, by straightforward computations.
  2. the Weyl groups of semisimple finite dimensional Lie algebras are absolutely real, by [C, Th. C (iii), p. 45].
  3. H3 is absolutely real, by Proposition 1.7 below.
  4. H4 is absolutely real, we have checked it using GAP3, [S].

Remark 1.5. Let G , H be finite groups. We note:

  • (s,t) ∈ G × H is absolutely real iff both s ∈ G and t ∈ H are absolutely real.
  • G × H is absolutely real iff both G and H are absolutely real.
  • Assume H abelian. Then H is absolutely real iff H has exponent 2, i. e. H ≃ ℤn 2 for some integer n .
  • If G is absolutely real and H is abelian of exponent 2 then G × H is absolutely real.

We first discuss when an element of 𝔸n is absolutely real. Assume that π ∈ 𝕊n is of type (1m1,2m2, ...,nmn) . Then π ∈ 𝔸n iff  ∑ mj j even is even.

Lemma 1.6. (a). If m1 ≥ 2 , then π is absolutely real in 𝔸n .

(b). If ∑ (m4h + m4h+3 ) h∈ℕ is even then π is absolutely real in 𝔸n .

Proof. Let τj := (1 2 ...j) for some j and take

 { (1 j - 1)(2 j - 2 )⋅⋅⋅(k - 1 k + 1), if j = 2k is even, gj = (1 j - 1)(2 j - 2 )⋅⋅⋅(k k + 1), if j = 2k + 1 is odd.

It is easy to see that gjτjgj = τ -1 j , g2= id j and

 { k-1 sgn(g ) = (- 1) , if j = 2k is even, j (- 1)k, if j = 2k + 1 is odd.

To prove (b), observe that there exists an involution σ ∈ 𝕊n such that σπ σ = π- 1 , which is a product of "translations" of the g j 's. Since the sign of σ is (- 1)∑h∈ℕ(m4h+m4h+3) , σ ∈ 𝔸 n iff ∑ h∈ℕ(m4h + m4h+3 ) is even; (b) follows. We prove (a). By assumption there are at least two points fixed by π , say n - 1 , n . By the preceding there exists an involution σ ∈ 𝕊n-2 such that σπ σ = π -1 . If σ ∈ 𝔸n-2 ⊂ 𝔸n we are done, otherwise take ^σ = σ(n - 1 n) ∈ 𝔸n ; ^σ is an involution and ^σπ ^σ = π- 1 . □

Proposition 1.7. The groups 𝔸 5 and H 3 are absolutely real.

Proof. The type of π ∈ 𝔸 5 is either (15) , (12,31) , (1,22) or (51) ; in the first two cases π is absolutely real by Lemma 1.6 part (a), in the last two by part (b). Since H3 ≃ 𝔸5 × ℤ2 (see [Hu, Section 2.13]), then the Coxeter group is absolutely real by Remark 1.5. □

1.3. Generalizations of Lemma 1.3. The next two Lemmata are variations of [AZ, Lemma 2.2]. A result in the same spirit appears in [FGV]. We deal with elements s having a power in O , the conjugacy class of s . Clearly, if sj = σs σ-1 is in O , then sjl = σlsσ-l is in O , for every l . So, sj|σ| = s ; this implies that |s| divides  |σ| j - 1 . Hence

pict

with N := |qss| , recall (2).

Lemma 1.8. Let G be a finite group, s ∈ G , O the conjugacy class of s and ρ ∈ ^Gs . Assume that there exists an integer j such that s , sj and sj2 are distinct elements and sj is in O . If dim 𝔅 (O, ρ) < ∞ , then s has even order and qss = - 1 .

Proof. We assume that dim 𝔅 (O, ρ) < ∞ , thus N > 1 . It is easy to see that

pict

for every l , h . We will call  l tl := sj , gl := σl , l = 0 , 1 , 2 ; so tl = glsgl-1 , for l = 0 , 1 , 2 . The other relations between tl 's and gh 's are obtained from (4). For v ∈ V - 0 and l = 1 or 2 , we define Wl := ℂ - span of {g0v,glv} . Hence, Wl is a braided vector subspace of M (O,ρ ) of Cartan type with

 ( j|σ|-l) ( ) Ql = qssl qss , Al = 2 a12(l) , qjss qss a21(l) 2

where a (l) = a (l) ≡ j |σ|- l + jl mod (N ) 12 21 . Since dim 𝔅 (O, ρ) < ∞ , we have that a12(l) = a21(l) = 0 or - 1 . We consider now two cases.
(i) Let us suppose that a12(1) = a21(1) = 0 . This implies that j|σ|-1 + j ≡ 0 mod (N ) . Since N divides j|σ| - 1 , we have that N divides j2 + 1 . We consider now two possibilities.
  • Assume that a12(2) = a21(2) = 0 . Then j |σ|- 2 + j2 ≡ 0 mod (N ) . Since N divides j|σ| - 1 , we have that N divides  4 j + 1 . So, - 1 ≡ 1 mod (N ) ; hence the result follows.
  • Assume that a12(2) = a21(2) = - 1 . Then j|σ|-2 + j2 ≡ - 1 mod (N ) . We can see that N divides j4 + j2 + 1 . This implies that N divides 1 , a contradiction.

(ii) Let us suppose that a12(1) = a21(1) = - 1 . This implies that j|σ|-1 + j ≡ - 1 mod (N ) . Since N divides j|σ| - 1 , we have that N divides j2 + j + 1 . We consider now two possibilities.

  • Assume that a12(2) = a21(2) = 0 . Then  |σ|-2 2 j + j ≡ 0 mod (N ) . So, N divides  4 j + 1 . It is easy to see that N divides  2 j . Since j and |s| are relatively prime, N must be 1 , a contradiction.
  • Assume that a12(2) = a21(2) = - 1 . This means that the subspace ^ W := ℂ - span of {g0v, g1v,g2v} of M (O, ρ) is of Cartan type with

     ( j|σ|-1 j|σ|-2) ( ) qss qss qss|σ|-1 2 - 1 - 1 Q = ( qjss qss qjss ) , A = ( - 1 2 - 1) . qjs2s qjss qss - 1 - 1 2

    By Theorem 1.2, we have that dim 𝔅 (O, ρ) = ∞ , a contradiction.

This concludes the proof. □

Lemma 1.9. Let G be a finite group, s ∈ G , O the conjugacy class of s and  ^ s ρ = (ρ,V ) ∈ G such that dim 𝔅 (O, ρ) < ∞ . Assume that there exists an integer j such that  j s ⁄= s and  j s is in O .

  1. If deg ρ > 1 , then s has even order and q = - 1 ss .
  2. If deg ρ = 1 , then either s has even order and qss = - 1 , or qss ∈ 𝔾3 - 1 .

Proof. We will proceed and use the notation as in the proof of Lemma 1.8. If  j2 s ⁄= s , then the result follows by Lemma 1.8. Assume that  2 sj = s . This implies that |s| divides j2 - 1 , so N divides j2 - 1 .

(a) Let v1 and v2 in V linearly independent and let W = ℂ - span of {g0v1, g0v2, g1v1, g1v2} , with g0 := id and g1 := σ . Thus W is a braided vector subspace of M (O, ρ) of Cartan type with

 ( |σ|-1 |σ|-1) ( ) qss qss qjss qjss 2 a12 a13 a14 || qss qss qj|σ|-1 qj|σ|-1|| || a21 2 a23 a24|| Q = ( qj qj ssq ssq ) , A = ( a31 a32 2 a34) , sjs ssj ss ss a a a 2 qss qss qss qss 41 42 43

where aij = aji , i ⁄= j , a12 ≡ 2 ≡ a34 mod (N ) , a13 = a14 = a23 = a24 and

a ≡ j|σ|-1 + j mod (N ). 13

If a12 = 0 or a34 = 0 , then N divides 2 and the result follows. Besides, if a13 = 0 , then j|σ|-1 + j ≡ 0 mod (N ) ; this implies that N divides j2 + 1 , so N divides 2 and the result follows. On the other hand, if aij = - 1 , for all i,j , we have that the matrix A is not of finite type; hence dim 𝔅 (O, ρ) = ∞ , from Theorem 1.2. This is a contradiction by hypothesis. Therefore, (a) is proved.

(b) For v ∈ V - 0 we define W := ℂ - span of {g0v,g1v} , with g0 := id and g1 := σ . Hence, W is a braided vector subspace of M (O,ρ ) of Cartan type with

 ( |σ|-1) ( ) qss qjss 2 a12 Q = qjss qss , A = a21 2 ,

where  |σ|-1 a12 ≡ j + j mod (N ) . Since dim 𝔅 (O, ρ) < ∞ , we have that a12 = 0 or - 1 . We consider now two possibilities.
(i) Assume that a12 = 0 . This implies that j|σ|-1 + j ≡ 0 mod (N ) . Since N divides j|σ| - 1 , we have that N divides j2 + 1 . Thus, N divides 2 ; hence N = 2 , and the result follows.
(ii) Assume that a12 = - 1 . This implies that  |σ|-1 j + j ≡ - 1 mod (N ) . Since N divides  |σ| j - 1 , we have that N divides  2 j + j + 1 . Hence, N divides j + 2 . If p is a prime divisor of N , then p divides j - 1 or j + 1 , because N divides j2 - 1 . If p divides j + 1 , then p divides 1 , a contradiction. So, N divides j - 1 . Hence, N divides 3 , i.e. N = 3 and the result follows. □

2. On Nichols algebras over 𝔸n

We recall that we will denote O or O π the conjugacy class of an element π in 𝔸n , and ρ in  ^π 𝔸 n , a representative of an isomorphism class of irreducible representations of  π 𝔸 n . We want to determine pairs (O, ρ ) , for which dim 𝔅 (O, ρ) = ∞ , following the strategy given in [AZAF]; see also [].

The following is a helpful criterion to decide when a conjugacy class of an even permutation π in 𝕊n splits in 𝔸n .

Proposition 2.1. [JL, Proposition 12.17] Let π ∈ 𝔸n , with n > 1 .

  1. If π commutes with some odd permutation in 𝕊 n , then  𝔸n 𝕊n O π = Oπ and  π π [𝕊n : 𝔸 n] = 2 .
  2. If π does not commute with any odd permutation in 𝕊n , then O 𝕊πn splits into two conjugacy classes in 𝔸n of equal size, with representatives π and (1 2)π(1 2) , and 𝕊πn = 𝔸 πn .

Remarks 2.2. (i) Notice that if π satisfies (1) of Proposition 2.1, then π is real. The reciprocal is not true, e.g. consider τ5 = (12 34 5) in 𝔸5 .

(ii) One can see that if π in 𝔸n is of type (1m1, 2m2,...,nmn ) , then π satisfies (2) of Proposition 2.1 if and only if m1 = 0 or 1 , m = 0 2h and m ≤ 1 2h+1 , for all h ≥ 1 . Thus, if π ∈ 𝔸 n has even order, then π is real.

We state the main Theorem of the section.

Theorem 2.3. Let π ∈ 𝔸n and ρ ∈ ^𝔸πn . Assume that π is neither (1 23) nor (1 32) in 𝔸4 . If dim 𝔅 (O π,ρ) < ∞ , then π has even order and qππ = - 1 .

Proof. If |π| is even the result follows by Lemma 1.3 and Remark 2.2 (ii). Let us suppose that |π| ≥ 5 and odd . If  -1 π is in Oπ , then the result follows by Lemma 1.3. Assume that π -1 ⁄∈ O π . We consider two cases.

(i) If π2 ∈ O π , then π4 is in O π , and π4 ⁄= π2 because |π| ≥ 5 . Hence, the result follows from Lemma 1.8.

(ii) Assume that  2 π ⁄∈ O π . We know that there exist σ and  ′ σ in 𝕊n , necessarily odd permutations, such that  - 1 -1 π = σ πσ and π2 = σ′π σ′-1 . Then σ′′ = σ σ′ ∈ 𝔸n and π- 2 = σ′′πσ ′′-1 ; so π-2 is in Oπ . This implies that π4 is in Oπ , and π4 ⁄= π -2 because 5 ≤ |π| is odd. Now, the result follows from Lemma 1.8.

Finally, let us suppose that |π| = 3 , with type (1a,3b) . If a ≥ 2 or b ≥ 2 , then π is real, by Lemma 1.6 (a) and Remark 2.2, respectively. Hence, the result follows by Lemma 1.3. This concludes the proof. □

2.1. Case 𝔸3 . Obviously, 𝔸3 ≃ ℤ3 ; thus 𝔸3 is not real. This case was considered in [AS1, Theorem 1.3].

2.2. Case 𝔸4 . It is straightforward to check that 𝔸4 is not real, since (12 3) is not real in 𝔸 4 . Let π in 𝔸 4 ; then the type of π may be (14) , (22) or (1,3) . If the type of π is  4 (1 ) , then dim 𝔅 (O π,ρ) = ∞ , for any ρ in ^𝔸4 , by Lemma 1.1. If the type of π is (1,3) , then π is not real; moreover we have

O (1 2 3) = {(1 2 3),(1 3 4),(1 4 2),(2 4 3 )},

O (1 3 2) = {(1 3 2),(1 2 4),(1 4 3),(2 3 4 )},

and  π 𝔸4 = ⟨π ⟩ ≃ ℤ3 . If  ^ π ρ ∈ 𝔸 4 is trivial, then dim 𝔅 (O π,ρ) = ∞ ; otherwise it is not known.

The following result is a variation of [AZ, Theorem 2.7].

Proposition 2.4. Let π in 𝔸4 of type  2 (2 ) . Then dim 𝔅 (O π,ρ) = ∞ , for every ρ in ^𝔸 π4 .

Proof. We can assume that π = (1 2)(3 4) . If we call t1 := π , t := (1 3)(24) 2 and t := (1 4)(23) 3 , then O = {t ,t ,t} π 1 2 3 and  π 𝔸 4 = ⟨t1⟩ × ⟨t2⟩ ≃ ℤ2 × ℤ2 . If g1 = id , g2 = (1 32) and g3 = (1 23 ) , then  - 1 tj = gjπgj , j = 1,2,3 , and pict

Let ρ in ^𝔸π4 and M (O π,ρ) := g1v ⊕ g2v ⊕ g3v , where ⟨v⟩ is the vector space affording ρ . Thus M (O π,ρ) is a braided vector space with braiding given by - see (1)- c(gjv ⊗ gjv) = gjt1 ⋅ v ⊗ gjv and c(g v ⊗ g v) = g t ⋅ v ⊗ g v j 1 1 j j , j = 1,2,3 and pict

Clearly, dim 𝔅 (O π,ɛ ⊗ ɛ) = dim 𝔅 (O π,ɛ ⊗ sgn ) = ∞ , by Lemma 1.1. If we consider ρ = sgn ⊗ ɛ (resp. sgn⊗ sgn ), then M (O π,ρ) is of Cartan type with matrix of coefficients (qij)ij given by

 ( ) ( ) - 1 - 1 1 - 1 1 - 1 Q = ( 1 - 1 - 1) , ( resp. Q = ( - 1 - 1 1 ) ). - 1 1 - 1 1 - 1 - 1

In both cases the Cartan matrix is  ( ) 2 - 1 - 1 A = ( - 1 2 - 1 ) - 1 - 1 2 . Therefore, dim 𝔅 (Oπ, ρ) = ∞ , by Theorem 1.2. □

2.3. Case 𝔸5 . Here is the key step in the consideration of this case.

Lemma 2.5. Let π ∈ 𝔸5 . Then dim 𝔅 (O π,ρ) = ∞ , for every ρ in ^𝔸 π 5 .

Proof. Let π ∈ 𝔸5 . If the type of π is either  5 (1 ) ,  2 (1 ,3) or (5) , we have that dim 𝔅 (O π,ρ) = ∞ , by Lemma 1.3 and Proposition 1.7. Let us assume that the type of π is (22) . For j = 1,2,3 , let tj and gj be as in the proof of Proposition 2.4. By Proposition 2.1 and straightforward computations, we have that O 𝔸5 = O 𝕊5 π π and  π 𝔸 5 = ⟨t1⟩ × ⟨t2⟩ ≃ ℤ2 × ℤ2 . Notice that  𝔸5 tj ∈ O π , j = 1,2,3 . Let ρ ∈ ^𝔸 π5 and W := g1v ⊕ g2v ⊕ g3v , where ⟨v⟩ is the vector space affording ρ ; then W is a braided vector subspace of M (O π,ρ) . Therefore, dim 𝔅 (O π,ρ) = ∞ , by the same argument as in the proof of Proposition 2.4. □

As an immediate consequence of Lemma 2.5 we have the following result.

Theorem 2.6. Any finite-dimensional complex pointed Hopf algebra H with G (H ) ≃ 𝔸5 is necessarily isomorphic to the group algebra of 𝔸5 .

Proof. Let H be a complex pointed Hopf algebra with G(H ) ≃ 𝔸5 . Let  ℂ𝔸5 M ∈ ℂ𝔸5YD be the infinitesimal braiding of H -see [AS2]. Assume that H ⁄= ℂ 𝔸5 ; thus M ⁄= 0 . Let N ⊂ M be an irreducible submodule. Then dim 𝔅 (N ) = ∞ , by Lemma 2.5. Hence, dim 𝔅 (M ) = ∞ and dim H = ∞ . □

2.4. Case 𝔸6 . Let π be in 𝔸6 . If the type of π is (16) , (12,22) , (13,3 ) , (32) or (1,5) , then π is absolutely real by Lemma 1.6, and if the type of π is (2,4) , then π is real because it has even order - see Remark 2.2 (ii). Hence, 𝔸6 is a real group. We summarize our results in the following statement.

Theorem 2.7. Let M (O, ρ) be an irreducible Yetter-Drinfeld module over ℂ 𝔸6 , corresponding to a pair (O, ρ) . If dim 𝔅 (O, ρ) < ∞ , then O = O π , with π = (1 2)(34 56) , and ρ = sgn ∈ ^ℤ4 .

Remark 2.8. In this Theorem we do not claim that the condition is sufficient.

Proof. Let π be in 𝔸 6 . If the type of π is

  • (16) , then dim 𝔅 (O π,ρ) = ∞ , for any ρ in ^𝔸 π6 , by Lemma 1.1.
  •  3 (1 ,3) ,  2 (3 ) or (1,5) , then dim 𝔅 (O π,ρ) = ∞ , for any ρ in ^𝔸 π6 , by Lemma 1.3.

Let us suppose that the type of π is  2 2 (1 ,2 ) ; we can assume that π = (1 2)(34) . It is easy to check that

 π 𝔸6 = ⟨a := (34 )(5 6), b := (13 24)(5 6)⟩ ≃ 𝔻4.

Notice that π = b2 . It is known that ^𝔻4 = {ρ1, ρ2, ρ3, ρ4, ρ5} , where ρj , j = 1 , 2 , 3 and 4 , are the following characters

pict

and ρ5 is the 2 -dimensional representation given by

pict

It is clear that ρj(π) = 1 , j = 1 , 2 , 3 and 4 . Then dim 𝔅 (Oπ, ρ) = ∞ , by Lemma 1.1. Let us consider now that ρ = ρ5 . We define t1 := (12 )(3 4) , t2 := (13 )(2 4) , t3 := (14)(2 3) , g1 := id , g2 := (13 2) and g3 := (12 3) . It is clear that

pict

If  ( ) i v1 := 1 we have that ρ(t1)(v1) = - v1 and ρ(t2)(v1) = v1 = - ρ(t3)(v1) . We define W := ℂ - span of {g v ,g v ,g v } 1 1 2 1 3 1 . Then W is a braiding subspace of M (O π,ρ) of Cartan type with

pict

Since A is not of finite type we have that dim 𝔅 (Oπ, ρ) = ∞ , by Theorem 1.2.

Finally, let us assume that the type of π is (2,4) . Then O π has 90 elements and 𝔸π = ⟨π ⟩ ≃ ℤ 6 4 . We call ^ℤ = { χ ,χ ,χ ,χ } 4 0 1 2 3 , where  l χl(π) = i , l = 0 , 1 , 2 , 3 . It is clear that if ρ = χl , with l = 0 , 1 or 3 , then ρ(π) ⁄= - 1 . This implies that dim 𝔅 (O π,ρ) = ∞ , by Lemma 1.3. □

Remark 2.9. We can see that every maximal abelian subrack of O (12)(3456) has two elements. Hence, M (O (12)(3456),ρ) is a negative braided space in the sense of [AF].

2.5. Case 𝔸m , m ≥ 7 . Let π ∈ 𝔸m , with |π | even. We now investigate the Nichols algebras associated with π by reduction to the analogous study for the orbit of π in 𝕊 m , [AF]. By Remark 2.2 (ii), O 𝔸m = O 𝕊m π π and [𝕊π : 𝔸 π] = 2 m m . So, we can determinate the irreducible representations of  π 𝔸m from those of  π 𝕊 m . We know that if the type of π is  b1 b2 bm (1 ,2 ,...,m ) , then 𝕊 πm = T1 ⋅⋅⋅Tm, with  b Ti ≃ ℤii⋊ 𝕊bi , 1 ≤ i ≤ m .

Some generalities and notation. Let G be a finite group, H a subgroup of G of index two, and η a representation of G . It is easy to see that

 { ′ η(g) ,if g ∈ H, η (g) := - η(g) ,if g ∈ G \ H,

defines a new representation of G . Notice that ResGH η = ResGH η′ . On the other hand, any representation ρ of H defines a representation ρ- of H , call it the conjugate representation of ρ , given by ρ(h ) := ρ(ghg -1) , for every h ∈ H , where g is an arbitrary fixed element in G \ H . Since g is unique up to multiplication by an element of H , the conjugate representation is unique up to isomorphism.

Let s ∈ H such that OHs = OGs ; thus [Gs : Hs ] = 2 . Let η in G^s . Then we have two cases:

  1. η ⁄≃ η′ . If  s ρ := ResGHs η , then ρ ∈ ^Hs ,  -- ρ ≃ ρ and IndGss ρ ≃ η ⊕ η′ H .
  2.  ′ η ≃ η . We have that  Gs -- Res Hs η ≃ ρ ⊕ ρ and  Gs Gs-- IndHs ρ ≃ η ≃ Ind Hsρ .

Moreover, if ρ is an irreducible representation of Hs , then ρ is a restriction of some  ^ s η ∈ G or is a direct summand of  Gs Res Hs η as in (ii), see [FH, Ch. 5].

Remark 2.10. If η ∈ ^Gs and ρ := ResGss η H , it is easy to check that

pict

as braided vector spaces.

We apply these observations to the case G = 𝕊m and H = 𝔸m . We use some notations given in [AF, Section II.D].

Lemma 2.11. Assume that the type of π is ((2r)n) , with r ≥ 1 and n even. Let ρ in  ^π 𝔸 m , with m = 2rn .

  • If qππ ⁄= - 1 , then dim 𝔅 (O π,ρ) = ∞ .
  • If  -- ρ ≃ ρ and qππ = - 1 , then
    • if r = 1 , then dim 𝔅 (O, ρ) = ∞ .
    • Assume that r > 1 . If deg ρ > 1 , then dim 𝔅 (O, ρ) = ∞ . Assume that deg ρ = 1 . If ρ = χr,...,r ⊗ μ , with r even or odd, or if ρ = χc,...,c ⊗ μ , with r even and  r c = 2 or 3r 2- , where μ = ɛ or sgn , then the braiding is negative; otherwise, dim 𝔅 (O, ρ) = ∞ .

Proof. (a) follows by Remark 2.2 (ii) and Lemma 1.3. (b). Since ρ ≃ ρ- ,  𝕊πm ρ = Res 𝔸πm(η ) , with η ∈ ^𝕊πm ,  ′ η ⁄≃ η and  𝕊πm ′ Ind 𝔸πm ρ ≃ η ⊕ η . Notice that η (π ) = - Id because ρ(π) = - Id . Now, as the racks are the same, i.e. O 𝔸πm = O𝕊πm , we can apply [AF, Theorem 1]. □

Remark 2.12. Keep the notation of the Lemma. If ρ is not isomorphic to its conjugate representation ρ- , then there exists η ∈ ^𝕊π m such that  𝕊πm -- Res 𝔸πm(η) = ρ ⊕ ρ and  𝕊πm ′ 𝕊πm -- Ind𝔸πm ρ ≃ η ≃ η ≃ Ind 𝔸πm ρ . Clearly, η(π) and -- ρ(π) act by scalar - 1 , and we have that M (O𝕊πm ,η) ≃ M (O 𝔸πm ,ρ) ⊕ M (O𝔸πm ,ρ) as braided vector spaces. We do not get new information with the techniques available today.

3. On Nichols algebras over 𝔻n

We fix the notation: the dihedral group 𝔻n of order 2n is generated by x and y with defining relations x2 = e = yn and xyx = y-1 . Let ω be a primitive n -th root of 1 and let χ be the character of ⟨y⟩ , χ (y) = ω . If s ∈ 𝔻 n then we denote the conjugacy class by On s or simply Os .

Theorem 3.1. Let M (O, ρ) be the irreducible Yetter-Drinfeld module over ℂ 𝔻n corresponding to a pair (O, ρ) . Assume that its Nichols algebra 𝔅 (O, ρ) is finite-dimensional.

  1. If n is odd, then (O, ρ) = (Ox, sgn ) , where sgn ∈ ^𝔻xn , 𝔻xn = ⟨x⟩ ≃ ℤ2 .
  2. If n = 2m is even, then (O, ρ) is one of the following:
    1. (Oym, ρ) where  ^ ρ ∈ 𝔻n satisfies  m ρ(y ) = - 1 .
    2. (Oyh, χj) where 1 ≤ h ≤ n - 1 , h ⁄= m and ωhj = - 1 .
    3. (Ox, sgn⊗ sgn ) or (Ox, sgn ⊗ ɛ) , where sgn ⊗ sgn , sgn ⊗ ɛ ∈ ^𝔻xn , 𝔻xn = ⟨x⟩ ⊕ ⟨ym ⟩ ≃ ℤ2 × ℤ2 .
    4. (Oxy, sgn ⊗ sgn ) or (Oxy,sgn ⊗ ɛ) , where sgn ⊗ sgn ,  ^xy sgn ⊗ ɛ ∈ 𝔻n ,  xy m 𝔻 n = ⟨xy⟩ ⊕ ⟨y ⟩ ≃ ℤ2 × ℤ2 .

In the cases (i) and (ii) the dimension is finite. In the cases (iii) and (iv), the braiding is negative in the sense of [AF].

Remark 3.2. There are isomorphisms of braided vector spaces

pict

Remark 3.3. Assume for simplicity that n is odd and that n = de , where d , e are integers ≥ 2 . Then the (indecomposable) rack On x is a disjoint union of e racks isomorphic to Od x ; in other words, On x is an extension of Oe x by Od x (and vice versa), see [AG, Section 2]. Thus, there is an epimorphism of braided vector spaces  n e M (O x,sgn ) → M (O x,sgn) , as well as an inclusion M (Odx,sgn ) → M (Onx,sgn ) . The techniques available today do not allow to compute the Nichols algebra 𝔅 (Onx ,sgn ) from the knowledge of the Nichols algebra 𝔅 (Oe ,sgn) x .

Remark 3.4. In Theorem 3.1 we do not claim that the conditions are sufficient. See Tables 1, 2. For instance, it is known that  n dim 𝔅 (Ox ,sgn ) < ∞ when n = 3 - see [MS]; for other odd n , this is open.

|----------------------------|------------|---------|----------| |Orbit |Isotropy |Rep. |dim 픅 (V) | | |group | | | |e---------------------------|픻-----------|any------|∞---------| |---------±h-----------------|-n----------|---------|----------| |Oyh = {y }, h ⁄= 0, |ℤn ≃ ⟨y ⟩ |any |∞ | || Oyh-|=-2------------------|------------|---------|----------| |Ox = {xyh : 0 ≤ h ≤ n- 1}, |ℤ2 ≃ ⟨x⟩ |ɛ |∞ | || Ox |= n | | | | | | |sgn-------|negative--| | | | | | |----------------------------|-----------------------braiding---| | |
Table 1. Nichols algebras of irreducible Yetter-Drinfeld modules over 𝔻n , n odd.

Let us now proceed with the proof of Theorem 3.1.

Proof. If s = id , then q = 1 ss and dim 𝔅 (O, ρ) = ∞ , from Lemma 1.1.

We consider now two cases.

CASE 1: n odd.

(I) If s = yh , with 1 ≤ h ≤ n , it is easy to see that O h = {yh,y- h} y and 𝔻yh = ⟨y ⟩ ≃ ℤ n n . Then ℤ^ = {χ }n n ll=1 , where  l χl(y) = ω , with  i2π- ω = exp (n ) ∈ 𝔾n a primitive n -th root of 1 . Let us consider M (Oyh,χl) ; it is a braided vector space of diagonal type. If qss ⁄= - 1 , then dim 𝔅 (Oyh, χl) = ∞ , from Lemma 1.3. Assume qss = - 1 ; so we have - 1 = χl(s) = χl(yh) = ωlh. This is a contradiction because n is odd.

(II) If s = x , then  n- 1 Ox = {x, xy,...,xy } and  x 𝔻 n = ⟨x⟩ ≃ ℤ2 . Clearly, dim 𝔅 (Ox, ɛ) = ∞ . On the other hand, M (Ox,sgn ) is a negative braided vector space, since every abelian subrack of Ox has one element; indeed xyjxyk = xykxyj , 0 ≤ j,k ≤ n - 1 , if and only if j = k .

Therefore, the part (a) of the Theorem is proved.

|------------------------------|------------|------------|----------| |Orbit |Isotropy |Rep. |dim 픅 (V )| |------------------------------|group-------|------------|----------| |e-----------------------------|픻n----------|any---------|∞---------| |Oym = {ym }, | Oym |= 1 |픻n | |∞ | | | |(V, ρ) ∈ | | | | |픻^n, | | | | |ρ(ym ) = 1 | | | | |------------|----------| | | |(V, ρ) ∈ |2dim V | | | | ^ | | | | |픻n, | | | | |ρ(ym ) = | | |------------------------------|------------|--1---------|----------| |Oyh = {y±h}, h ⁄= 0,m, |ℤn ≃ ⟨y⟩ |χj, |4 | || O h |= 2 | |ωhj = - 1 | | | y | |--j---------|----------| | | |χ h,j |∞ | |------------------------------|------------|ω---⁄=--- 1--|----------| |Ox = {xy2h : 0 ≤ h ≤ m - 1} |ℤ2 × ℤ2 ≃ |ɛ ⊗ ɛ, |∞ | || Ox |= m |⟨x⟩ ⊕ ⟨ym⟩ |ɛ-⊗-sgn-----|----------| | | |sgn ⊗ sgn, |negative | | | |sgn ⊗ ɛ |braiding | |----------2h+1----------------|------------|------------|----------| |Oxy = {xy : 0 ≤ h ≤ m - |ℤ2 × ℤ2 ≃m |ɛ ⊗ ɛ, |∞ | |1} |⟨xy⟩ ⊕ ⟨y ⟩ |ɛ ⊗ sgn | | || Ox |= m | |------------|----------| | | |sgn ⊗ sgn, |negative | | | |sgn ⊗ ɛ |braiding | |------------------------------|------------------------------------- | |
Table 2
. Nichols algebras of irreducible Yetter-Drinfeld modules over 𝔻n , n = 2m even.

CASE 2: n even. Let us say n = 2m .

(I) If s = ym , then O m = {ym } y and 𝔻ym = 𝔻 n n . Clearly, dim 𝔅 (Oym, ρ) = ∞ , for every  ^ ρ ∈ 𝔻n with ρ(s) = Id . On the other hand, if (ρ,V ) ∈ ^𝔻n is such that ρ(s) = - Id , then it is straightforward to prove that 𝔅 (Oym, ρ) = ∧ (V ) , the exterior algebra of V ; hence dim 𝔅 (O m, ρ) = 2dim V y .

(II) If  h s = y , h ⁄= 0,m ; then  h - h Oyh = {y ,y } and  yh 𝔻 n = ⟨y⟩ ≃ ℤn . From Lemma 1.3, it is clear that dim 𝔅 (Oyh, χl) = ∞ , for every l such that χl(yh) ⁄= - 1 , i.e. ωhl ⁄= - 1 . On the other hand, it is easy to see that 𝔅 (O h,χ ) = ∧ (M (O h,χ )) y l y l , hence dim 𝔅 (O h,χ ) = 4 y l , for every χ l with  h χl(y ) = - 1 .

(III) If s = x , then  2h Ox = {xy : 0 ≤ h ≤ m - 1} and  x m 𝔻n = ⟨x ⟩ ⊕ ⟨y ⟩ ≃ ℤ2 × ℤ2 . From Lemma 1.1, dim 𝔅 (Ox,ɛ ⊗ ɛ) = dim 𝔅 (Ox, ɛ ⊗ sgn) = ∞ .

For the cases ρ = sgn ⊗ ɛ or sgn ⊗ sgn , we note the following fact.

  1. If m is odd and 0 ≤ j,k ≤ m - 1 , we have that

    xy2jxy2k = xy2kxy2j if and only if j = k.

  2. If m is even and 0 ≤ j ≤ k ≤ m - 1 , we have that

     2j 2k 2k 2j m- xy xy = xy xy if and only if k = j or j + 2 .

The cases (i) and (ii) say that every maximal abelian subrack of Ox has one and two elements, respectively. Hence, in both cases the braiding is negative. Indeed, the result is obvious for the case (i), while in the case (ii) we have that if  2j j j -1 tj := xy = xy x(xy ) and  2k k k -1 tk := xy = xy x (xy ) commute in Ox , then qjj = - 1 = qkk and qjkqkj = 1 ; thus the braiding is negative.

(IV) If s = xy , then Oxy = {xy2h+1 : 0 ≤ h ≤ m - 1} and 𝔻xy = ⟨xy⟩ ⊕ ⟨ym⟩ ≃ ℤ2 × ℤ2 n . The result follows as in (III) using the isomorphism 𝔻 → 𝔻 n n , x ↦→ xy , y ↦→ y . □

4. On Nichols algebras over semisimple Hopf algebras

Let A be a Hopf algebra. Let J ∈ A ⊗ A be a twist and let  J A be the corresponding twisted Hopf algebra. If A is a Hopf subalgebra of a Hopf algebra H , then J is a twist for H and AJ is a Hopf subalgebra of HJ . Now, if A is semisimple, then this induces a bijection

{isoclasses of Hopf algebras with coradical ≃ A } -∽→ {isoclasses of Hopf algebras with coradical ≃ AJ },
(7)

that preserves standard invariants like dimension, Gelfand-Kirillov dimension, etc. Let now A = ℂ 𝔸5 and let J ∈ A ⊗ A be the non-trivial twist defined in [Ni]. By (7), we conclude immediately from Theorem 2.6.

Theorem 4.1. Let H be a finite-dimensional Hopf algebra with coradical isomorphic to (ℂ𝔸5 )J . Then H ≃ (ℂ𝔸5 )J .

Again, this is the first classification result we are aware of, for finite-dimensional Hopf algebras with coradical isomorphic to a fixed non-trivial semisimple Hopf algebra. Recently, a semisimple Hopf algebra B ≃ (ℂ 𝔻 × 𝔻 )J′ 3 3 was discovered in [GN]. This Hopf algebra B is simple, that is it has no non-trivial normal Hopf subalgebra. Since there are finite-dimensional non-semisimple pointed Hopf algebras with group 𝔻3 ≃ 𝕊3 , there is a finite-dimensional non-semisimple Hopf algebra with coradical isomorphic to B .

Acknowledgement. We are grateful to Professor John Stembridge for information on Coxeter groups, in particular reference [BG]. We thank Matías Graña, Sebastián Freyre and Leandro Vendramín for interesting discussions.

References

[AF]    N. Andruskiewitsch and F. Fantino, On pointed Hopf algebras associated with unmixed conjugacy classes in 𝕊m , J. Math. Phys. 48 (2007), 033502-1 - 033502-26.        [ Links ]

[AG]    N. Andruskiewitsch and M. Graña, From racks to pointed Hopf algebras, Adv. Math. 178 (2003), 177 - 243.        [ Links ]

[AS1]    N. Andruskiewitsch and H.-J. Schneider, Finite quantum groups and Cartan matrices, Adv. Math. 154 (2000), 1-45.        [ Links ]

[AS2]    N. Andruskiewitsch, Pointed Hopf Algebras, in "New directions in Hopf algebras", 1-68, Math. Sci. Res. Inst. Publ. 43, Cambridge Univ. Press, Cambridge, 2002.        [ Links ]

[AZ]    N. Andruskiewitsch and S. Zhang, On pointed Hopf algebras associated to some conjugacy classes in 𝕊n , Proc. Amer. Math. Soc. 135 (2007), 2723-2731.        [ Links ]

[BG]    C. T. Benson and L. C. Grove, The Schur indices of the reflection group H4 , J. Algebra, 27 (1973), 574-578.        [ Links ]

[C]    R. W. Carter, Conjugacy classes in the Weyl group, Compos. Math. 25 (1972), 1-59.        [ Links ]

[FGV]    S. Freyre, M. Graña and L. Vendramin, On Nichols algebras over SL (2,𝔽q) and GL(2,𝔽q) , preprint math.QA/0703498.        [ Links ]

[FH]    W. Fulton and J. Harris, Representation theory, Springer-Verlag, New York 1991.        [ Links ]

[GN]    C. Galindo and S. Natale, Simple Hopf algebras and deformations of finite groups, Math. Res. Lett., to appear.        [ Links ]

[Gñ]    M. Graña, On Nichols algebras of low dimension, Contemp. Math. 267 (2000), 111-134.        [ Links ]

[H]    I. Heckenberger, The Weyl groupoid of a Nichols algebra of diagonal type, Invent. Math. 164 (2006), 175-188.        [ Links ]

[Hu]    J. E. Humphreys, Reflection groups and Coxeter groups, Cambridge University Press, Cambridge 1990.        [ Links ]

[JL]    A. James and M. Liebeck, Representations and characters of groups, Cambridge University Press, Cambridge 2001.        [ Links ]

[MS]    A. Milinski and H-J. Schneider, Pointed Indecomposable Hopf Algebras over Coxeter Groups, Contemp. Math. 267 (2000), 215-236.        [ Links ]

[Ni]    D. Nikshych, K0 -rings and twisting of finite dimensional semisimple Hopf algebras, Commun. Algebra 26 (1998), 321-342. (Corrigendum: Commun. Algebra 26 (1998), 2019.)        [ Links ]

[S]    Martin Schönert et al. "GAP - Groups, Algorithms, and Programming - version 3 release 4 patchlevel 4". Lehrstuhl D für Mathematik, Rheinisch Westfälische Technische Hochschule, Aachen, Germany, 1997.        [ Links ]

Nicolás Andruskiewitsch
Facultad de Matemática, Astronomía y Física,
Universidad Nacional de Córdoba. CIEM - CONICET.
Medina Allende s/n
(5000) Ciudad Universitaria, Córdoba, Argentina
andrus@mate.uncor.edu

Fernando Fantino
Facultad de Matemática, Astronomía y Física,
Universidad Nacional de Córdoba. CIEM - CONICET.
Medina Allende s/n
(5000) Ciudad Universitaria, Córdoba, Argentina
fantino@mate.uncor.edu

Recibido: 16 de febrero de 2007
Aceptado: 2 de agosto de 2007

Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons