On the Classification of Finite Quasi-Quantum Groups over Abelian Groups
Hua-Lin Huang, Gongxiang Liu, Yuping Yang*, and Yu Ye
School of Mathematical Sciences, Huaqiao University, Quanzhou 362021, China
hualin.huang@hqu.edu.cnDepartment of Mathematics, Nanjing University, Nanjing 210093, China
gxliu@nju.edu.cnSchool of Mathematics and statistics, Southwest University, Chongqing 400715, China
yupingyang@swu.edu.cnSchool of Mathematical Sciences, Wu Wen-Tsun Key Laboratory of Mathematics, University of Science and Technology of China, Hefei, Anhui 230026, CHINAandHefei National Laboratory, University of Science and Technology of China, Hefei 230088, CHINAyeyu@ustc.edu.cn
Abstract.
Using a variety of methods developed in the theory of finite-dimensional quasi-Hopf algebras, we classify all finite-dimensional coradically graded pointed coquasi-Hopf algebras over abelian groups. As a consequence, we partially confirm the generation conjecture of pointed finite tensor categories due to Etingof, Gelaki, Nikshych and Ostrik.
As a continuation to a series of previous works [17, 18, 20], this paper completes the classification problem of finite-dimensional coradically graded pointed coquasi-Hopf algebras over abelian groups. Throughout, we work over an algebraically closed field of characteristic zero. Unless stated otherwise, in this paper all spaces, maps, (co)algebras, (co)modules, and categories, etc., are over
The classification of finite-dimensional pointed Hopf algebras over finite abelian groups was completed over the last two decades, and a systematic approach (in particular Weyl groupoids and arithmetic root systems) was established, see [3, 5, 6, 14, 15]. Meanwhile, Etingof and Gelaki proposed to classify pointed finite tensor categories. By the Tannakian formalism [11], this amounts to a classification of certain finite quasi-quantum groups, namely finite-dimensional elementary quasi-Hopf algebras, or dually finite-dimensional pointed coquasi-Hopf algebras. In the pioneering works [7, 8, 9, 13, 4], a few examples and classification results of such algebras, and consequently the associated pointed finite tensor categories, are thus obtained.
In [17, 18, 20], the authors of the present paper continue the study of the classification problem of finite quasi-quantum groups and several interesting classification results are also obtained. To explain this and our main result of this paper, we need some concrete notations.
Once and for all, let be a finite abelian group and be a -cocycle on Based on an analog of the lifting method in the theory of finite-dimensional pointed Hopf algebras, a complete understanding of the Nichols algebras in the twisted Yetter-Drinfeld module category is the crux for the classification of finite-dimensional pointed coquasi-Hopf algebras. A twisted Yetter-Drinfeld module is said to be of diagonal type, if it is a direct sum of -dimensional twisted Yetter-Drinfeld modules. The associated Nichols algebra is called diagonal if is so. Let be a pointed coquasi-Hopf algebra over , and the coradically graded coquasi-Hopf algebra associated to . The coinvariant subalgebra of will be a twisted Yetter-Drinfeld module in for certain , and is called diagonal if is diagonal as a twisted Yetter-Drinfeld module.
In [17, 18], we classified all finite-dimensional Nichols algebras of diagonal type in and proved that every finite-dimensional pointed coquasi-Hopf algebra of diagonal type must be the form of .
The aim of this paper is to study Nichols algebras of nondiagonal type and the main result is the following (see Theorem 3.1 for an equivalent form) .
Theorem 0.1.
Let be a Nichols algebra of nondiagonal type with . Then is infinite dimensional.
Here is the support group of (see the paragraph after Definition 2.3). Under assumption that , which is natural for us since the braided Hopf algebra structure of is determined by rather than (see the paragraph after Proposition 2.6), our result tells us that every finite-dimensional Nichols algebra must be of diagonal type finally, which already was classified in our previous works. As consequences, we can get the structure of general finite-dimensional pointed coquasi-Hopf algebras now (see Theorems 5.1 and 5.2).
Theorem 0.2. If is a finite-dimensional pointed coquasi-Hopf algebra over finite abelian group , then for a Yetter-Drinfeld module of finite type .
In [11], Etingof, Gelaki, Nikshych and Ostrik conjecture that every pointed finite tensor category over a field of characteristic zero is tensor generated by objects of length . Let be the set of isomorphism classes of simple objects in a pointed tensor category . Then is naturally a group under tensor product. Using our classification result and some useful result in [18], we can partially prove the conjecture (see Theorem 5.5).
Theorem 0.3. Let be a pointed finite tensor category over a field of characteristic zero. If the group is abelian, then is tensor generated by objects of length .
The paper is organized as follows. In Section 2, we recall some necessary notions and particularly introduce a method to study Nichols algebras in called change of based groups. Sections 3 and 4 are designed to give a proof of the above Theorem 0.1. The last section is devoted to the classification of finite-dimensional pointed coquasi-Hopf algebras and the generation problem of pointed finite tensor categories.
2. Preliminaries
In this section, we recall some necessary notions and basic facts about pointed coquasi-Hopf algebras, twisted Yetter-Drinfeld modules and Nichols algebras. The reader is referred to [11, 17, 18] for related concepts and notations.
2.1. Pointed coquasi-Hopf algebras
A coquasi-Hopf algebra is a coalgebra equipped with a
compatible quasi-algebra structure and a quasi-antipode. Namely,
there exist two coalgebra homomorphisms
a convolution-invertible map
called the associator, a coalgebra antimorphism and two
functions such that for all the following equalities hold:
The triple is called a quasi-antipode. is called a pointed coquasi-Hopf algebra if is a pointed coalgebra, i.e., every simple comodule of is -dimensional.
Let be a coalgebra, the coradical of is the sum of all simple subcoalgebras of . Fix a coalgebra with coradical , define inductively as follows: for each , define
Then we get a filtration which is called the coradical filtration of . A coquasi-Hopf algebra has a coradical filtration since it is a coalgebra.
Given a coquasi-Hopf algebra let be its coradical filtration, and let
the corresponding coradically graded coalgebra. Then naturally
inherits from a graded coquasi-Hopf algebra structure. The
corresponding graded associator satisfies
for all homogeneous elements
unless they all lie in
Similar conditions hold for and A coquasi-Hopf algebra is called coradically graded if as coquasi-Hopf algebras.
If is a pointed coquasi-Hopf algebra, then is a pointed cosemisimple coquasi-Hopf algebra, which is determined by a group together with a -cocycle on as follows.
Example 2.1.
Let be a group. Clearly the group algebra is a Hopf algebra with and
for any . Let be a normalized -cocycle on , i.e.
(2.1)
(2.2)
for all . By linearly extending, becomes a
convolution-invertible map. Define two linear functions
by
for any . Then together with these and
makes a coquasi-Hopf algebra, which will be written as in the following. The comodule category of forms a tensor category, which is called a Gr-category and denoted by .
Let us now consider the construction of Gr-categories which will be especially important in this paper. The crux to determine all the Gr-categories is to give a complete list of the representatives of the -cohomology classes in for all groups However, when is a finite abelian group, the problem was solved in [18], and a list of the representatives of can be given as follows.
Let denote the set of nonnegative integers, the ring of integers, and the cyclic group of order Any finite abelian group is of the form with
for Denote by the set of all -sequences
(2.3)
such that for , where and are ordered in the lexicographic order of their indices. We denote by the sequence (2.3) in the following. Let be a generator of . For any , define
(2.4)
Here and below stands for an -th primitive root of unity.
According to [18, Proposition 3.8], forms a complete set of representatives of the normalized -cocycles on up to -cohomology.
2.2. Nichols algebras of Yetter-Drinfeld modules
Nichols algebras are very important for the construction of pointed coquasi-Hopf algebras. For our purpose, we are mainly concerned with the Nichols algebras in the Yetter-Drinfeld module category of the coquasi-Hopf algebra , where is a finite abelian group and is a normalized 3-cocycle on . To emphasize we denote the Yetter-Drinfeld category of as . For convenience, we call an object in a twisted Yetter-Drinfeld module. Define
(2.5)
for all . By direct computation one can show that is a 2-cocycle on . The construction of category of twisted Yetter-Drinfeld modules can be summarized as follows, the detailed computations can be found in [18, 19].
Definition 2.2.
An object in is a -graded vector space ( as a -comodule) with each a projective
-representation with respect to the 2-cocycle namely for any we have
(2.6)
The module structure on the tensor product is determined by
(2.7)
The associativity and the braiding constraints of are given respectively by
(2.8)
(2.9)
for all
Let be a -cocycle on as given in (2.1).
One can verify directly that
(2.10)
Suppose is -representation, is a -representation, then is a -representation.
In particular, the dual object of is a -representation and , see [20, Proposition 2.5] for details.
A Yetter-Drinfeld module is called diagonal if is direct sum of -dimensional twisted Yetter-Drinfeld modules. For a simple Yetter-Drinfeld module in , there exists some such that and we define in this case. Recall that a -cocycle on is called symmetric if for all . By (2.6), it is not hard to show that a simple Yetter-Drinfeld module with is -dimensional if and only if is symmetric.
Let be a nonzero object in By we denote the tensor algebra in generated freely by It is clear that is isomorphic to as a linear space, where means
This induces a natural -graded structure on Define a comultiplication on by a counit by and an antipode by These provide a graded Hopf algebra structure on in the braided tensor category
Definition 2.3.
The Nichols algebra of is the quotient Hopf algebra in where is the unique maximal graded Hopf ideal contained in .
A Nichols algebra is called of diagonal type if is diagonal. Suppose is direct sum of simple objects, then we will say that the rank of is . According to [20, Proposition 3.1], is a -graded algebra with , where is a set of free generators of .
For , we will call the based group of and . Let be direct sum of simple Yetter-Drinfeld modules in , the degree of for , the subgroup will be called the support group of .
Next we will recall the definition of the twisting of a Nichols algebra through a 2-cochain of . Let , and let be a -cochain of . There is a new action of on determined by
(2.11)
for homogenous element and Here is the -degree of . We denote by , and one can verify that Moreover there is a tensor equivalence
which takes to and
for
Let be a Nichols algebra in Then is a Hopf algebra in with multiplication determined by
(2.12)
for all homogenous elements . Using the same terminology as for coquasi-Hopf algebras, we say that and are twist equivalent, or is a twisting of .
Lemma 2.4.
[18, Lemma 2.12]
The twisting of is a Nichols algebra in and .
2.3. Reduction
The study of Nichols algebras in is deeply related on the -cocycle on . Recall that a -cocycle on is called an abelian -cocycle if is pointed, i.e. each simple object of is -dimensional.
A key observation in [18] is that every Nichols algebra in is twist equivalent to a Nichols algebra in a normal Yetter-Drinfeld category when is an abelian -cocycle.
Suppose , is a generator of for all , and is an abelian -cocycle on Then up to cohomology must be of the form
(2.13)
The folowing lemma follows immediately.
Lemma 2.5.
(1).
Each -cocycle of a finite cyclic group or direct sum of two finite cyclic groups is abelian;
(2).
Let be a Nichols algebra of rank or rank , then is an abelian -cocycle on .
Suppose . Associated to there is a finite group defined by
(2.14)
Let
(2.15)
be the canonical epimorphism. The following proposition is very important for the study of Nichol algebras of diagonal type.
Proposition 2.6.
[18, Proposition 3.15]
Suppose that is an abelian -cocycle on Then is a -coboundary on , namely, there is a -cochain of such that .
Since the Yetter-Drinfeld module structure of is determined by the based group , while the braided
Hopf algebra structure of and the braiding are determined by the support group . So if the braided Hopf algebra structure of is the only concern, we can omit some Yetter-Drinfeld module information of and realize it in a new Yetter-Drinfeld category. Let and , we will say that is isomorphic to if there is a linear isomorphism which preserves the multiplication and comultiplication.
For a Yetter-Drinfeld module , we use to denote the comodule structure map of .
Suppose and , where is a finite abelian group. Let and be the support groups of and respectively. If there is a linear isomorphism and
a group epimorphism such that:
for any . Then is isomorphic to .
Let and be two finite groups and a group epimorphism, be a section of , that is . With these notations, we have following lemma.
Lemma 2.8.
Let . Then there is an object such that as linear spaces and the Yetter-Drinfeld module structure is determined by
(2.16)
(2.17)
for any . Moreover, we have .
Proof.
We need to show that the space with action and coaction of defined by (2.16) and (2.17) is a Yetter-Drinfeld module in . Let . Then it is obvious that we have such that as vector spaces for all . We only need to prove that is a projective -representation associated to
-cocycle . Let and , we have
So we have proved that is a projective -representation associated to the
-cocycle , and hence .
The isomorphism follows from Lemma 2.7 immediately.
∎
Let be a Nichols algebra in . By lemma 2.8, is isomorphic to a Nichols algebra in . If is an abelian -cocycle, then there is a -cochain of such that . According to Lemma 2.4, is a Nichols algebra in , which is twist equivalent to . So we obtain the following proposition.
Proposition 2.9.
Let be an abelian -cocycle and be a Nichols algebra in . Then is twist equivalent to a Nichols algebra in .
Let , if the action of the support group on is diagonal, then has a basis such that
(2.18)
where for . Such a basis is called a standard basis of .
The following lemma follows from [18, Lemma 4.1] immediately.
Lemma 2.10.
Let , the following three conditions are equivalent:
(1).
has a standard basis.
(2).
The action of support group on is diagonal.
(3).
is an abelian -cocycle on .
Now suppose has a standard basis , then we can define a nondirected graph associated to as follows:
1)
There is a bijecton from to the set of vertices of .
2)
For all the vertex is labelled by
3)
For all the number of edges between and is either or If or then otherwise and the edge is labelled by for all
The diagram is called the generalized Dynkin diagram of . Note that a Nichols algebra of diagonal type always has a generalized Dynkin diagram. It is also helpful to point out that if the generalized Dynkin diagram exists, it does not depend on the choice of the standard basis of . It is not hard to see that if has a generalized Dynkin diagram, then also have the same generalized Dynkin diagram with for any -cochain of . So combining this with Proposition 2.9, we have the following important proposition.
Proposition 2.11.
Let be a Nichols algebra with a standard basis. Then is twist equivalent to a Nichols algebra in , and the two Nichols algebras have the same generalized Dynkin diagrams.
According to this proposition, all finite-dimensional Nichols algebras with a standard basis can be determined by Heckenberger’s classification result of arithmetic root systems [15]. Note that if is rank or rank , then must be a finite cyclic group or direct product of two finite cyclic groups. According to (2.1), all the -cocycles on finite cyclic group or direct product of two finite cyclic groups must be abelian. So a Nichols algebra of rank or rank always has a standard basis. One of the main result in [20] is as follows.
Proposition 2.12.
[20, Proposition 3.18]
Suppose is a simple Yetter-Drinfeld module of nondiagonal type with . Then is finite dimensional if and only if is one of the following two types:
(I).
for all ;
(II).
and for all , here is a -rd primitive root of unity.
3. Nichols algebras
In this section, we will study Nichols algebras without a standard basis in , where is a finite abelian group and is a -cocycle on . The main result is as follows.
Theorem 3.1.
Suppose that has no standard basis, then is infinite dimensional.
See Remark 3.14 for the proof. Since a Nichols algebra of rank or always has a standard basis, so our start point will be Nichols algebras of rank .
3.1. Nichols algebras of rank
Suppose is a Nichols algebra of rank . If is an abelian -cocycle on , then the dimension of can be determined by Proposition 2.11. So in this subsection, we will mainly consider the case that is nonabelian.
Definition 3.2.
Let be a -cocycle on . An element is called an -element if for all .
Lemma 3.3.
Suppose is a -cocycle on , , then is a -element.
Proof.
For each element , we have
∎
Lemma 3.4.
Let be a finite abelian group and a -cocycle on . Then for any , we have
Take such that for some . Let be the minimal positive integer such that
(3.9)
for some . Note that such integer exits and since
(3.10)
Since and is an irreducible projective -representation with respect to , must be spanned by
In fact, let . Then we have , and for all we have
(3.11)
So we get
(3.12)
We claim that . Let be a primitive -th roots of unit and be an -th root of . Then are all -th roots of . For all , we set
Then for all we have
(3.13)
Then ’s are clearly linearly independent since they correspond to different eigenvalues. Then forms a basis of , and are all eigenvalues of when viewed as a linear transformation on .
Next we prove that .
Notice that
(3.14)
Inductively we have
(3.15)
So we have , and hence
for some scaler . This implies that span a sub-Yetter-Drinfeld module of , and hence since is simple. So we obtain that and hence .
The last statement is obvious and the proof is completed.
∎
By this lemma, we have the following important proposition.
Proposition 3.6.
Let be a direct sum of simple twisted Yetter-Drinfeld modules with . Then we have
(3.16)
Proof.
Let , . Then since . By Lemma 3.5, we have ,
, , and (3.16) follows from Lemma 3.4.
∎
Now we can consider nondiagonal Nichols algebras of rank . Firstly, we have the following propositions.
Proposition 3.7.
Let be a direct sum of simple twisted Yetter-Drinfeld modules with . If , then is infinite dimensional.
Proof.
Let , . Since , we have . Let , and . Then by Proposition 3.6. According to Proposition 2.12, if is not a simple Yetter-Drinfeld module of type (I) for some , then must be infinite dimensional, hence is infinite dimensional.
In the following, we assume that are simple Yetter-Drinfeld modules of type (I). By Lemma 3.5, has a basis such that
(3.17)
(3.18)
where . Here (3.17) follows from the fact that is a simple Yetter-Drinfeld module of type (I).
Similarly, also has a basis such that
(3.19)
(3.20)
where .
Let and . Since is direct sum of two cyclic groups, must be an abelian -cocycle on by Lemma 2.5. This implies is a Yetter-Drinfeld module of diagonal type in .
In the following, let be a submodule of We will consider the generalized Dynkin diagram .
If for , then the generalized Dynkin diagram (with unlabeled edges) associated to is
By Propsition 2.11, is twist equivalent to a Nichols algebra in , and has the same generalized Dynkin diagram with . Comparing the classification of generalized Dynkin diagrams of finite-dimensional Nichols algebras in [15], and hence must be infinite dimensional. This implies that is infinite dimensional.
If for some , then the generalized Dynkin diagram (with unlabeled edges) has a subdiagram of the form
Comparing the classification result in [15], again we have is infinite dimensional, and hence is infinite dimensional.
In either case is infinite dimensional, hence so is since is a subalgebra of .
∎
Proposition 3.8.
Let be a direct sum of simple twisted Yetter-Drinfeld modules and . Assume that , and at least one of ’s is of
type (II). Then is infinite dimensional.
Proof.
Without loss of generality, we can assume that is a simple twisted Yetter-Drinfeld module of type (II). In what follows we will prove that (similarly for ) is infinite dimensional, which forces is infinite dimensional.
Let and for , and . Then is an abelian -cocycle on by Lemma 2.5, and hence is a Nichols algebra of diagonal type in . By Lemma 3.5, has a basis such that
(3.21)
(3.22)
Here is a root of unit satisfying . In the following, we need to consider two cases: (a) has type (I), (b) has type (II).
Here .
If , the generalized Dynkin diagram (with unlabeled edges) of is
If , the generalized Dynkin diagram of is
By Propsition 2.11, is twist equivalent to some Nichols algebra in , and have the same generalized Dynkin diagram with . By checking up the classification of generalized Dynkin diagrams of finite-dimensional Nichols algebras of diagonal type in [15], we can see that and hence is infinite dimensional.
Here and is a fixed number.
If , then the generalized Dynkin diagram (with unlabeled edges) of is
If , then the generalized Dynkin diagram (with unlabeled edges) of is
In both cases is infinite dimensional by the classification of generalized Dynkin diagrams of finite-dimensional Nichols algebras of diagonal type in [15].
∎
It remains to consider where are simple twisted Yetter-Drinfeld modules of type (I) and , . We have the following theorem.
Theorem 3.9.
Let be simple Yetter-Drinfeld modules of type (I) such that , and . Then the Nichols algebra is infinite dimensional.
The proof of the theorem is quite technical and lengthy. To avoid digressing from the present main theme, we postpone the proof to the next section. With the help of this theorem, we can prove the following proposition.
Proposition 3.10.
Let be a direct sum of simple Yetter-Drinfeld modules and . If and are all simple Yetter-Drinfeld modules of
type (I), then is infinite dimensional.
Proof.
Let and for . Let be the abelian group with free generators such that for . Then it is obvious that there is a group epimorphism such that , . Let be a section of ( that is ) such that for all . For each , let be the Yetter-Drinfeld module associated to define by (2.16)-(2.17). Let . By Lemma 2.8, we have . On the other hand, by Theorem 3.9, is infinite dimensional, thus is also infinite dimensional.
∎
Combining Proposition 3.7, 3.8 and 3.10, the following theorem is clear.
Theorem 3.11.
Let be a nondiagonal Nichols algebra of rank with . Then is infinite dimensional.
Corollary 3.12.
Let be a Nichols algebra of rank , and a nonabelian -cocycle on . Then is infinite dimensional.
Proof.
Let and . Then the Nichols algebra can be realized in , and the rank of must be greater than or equal to .
For each , let be a nonzero simple Yetter-Drinfeld submodule of and . It is clear . By Lemma 2.10, is nondiagonal since is nonabelian on , thus are all nondiagonal in . Now is infinite dimensional by Theorem 3.11, and so is since .
∎
3.2. The general case
In this subsection, we will give a classification of finite-dimensional Nichols algebras in , where is a finite abelian group and is a -cocycle on . Firstly, we have the following theorem.
Theorem 3.13.
Let such that is nonabelian. Then is infinite dimensional.
Proof.
By Lemma 2.7, can be realized in . Since each -cocycle of cyclic group or direct sum of two cyclic groups is abelian, so the rank of must be greater than or equal to because is nonabelian. Thus is infinite dimensional by Corollary 3.12.
∎
Remark 3.14.
Since the conditions “ is abelian” and “ has a standard basis” are equivalent by Lemma 2.10, Theorem 3.13 is equivalent to Theorem 3.1.
We draw the following immediate consequences.
Corollary 3.15.
Suppose is a finite-dimensional Nichols algebra and . Then must be of diagonal type.
Corollary 3.16.
Suppose is a finite-dimensional Nichols algebra. Then we have
(1).
is abelian and has a standard basis;
(2).
is isomorphic to a Nichols algebra of diagonal type in .
Proof.
(1) follows from Theorem 3.13, and (2) follows from Proposition 3.10.
∎
3.3. Classification
Next we will present a classification of finite-dimensional Nichols algebras in .
Let be an arithmetic root system, where is a set of free generator of and is a bicharacter of . For each positive root , define
.
Then the height of is defined by
(3.27)
Let be a Yetter-Drinfeld module with a standard basis , . Then there is a pair associated to , where is a set of free generator of and is a bicharacter of given by
(3.28)
Remark 3.17.
If has another standard basis , and a pair associated to with respect to the basis . Then there is an isomorphism such that
for all .
Definition 3.18.
A Yetter-Drinfeld module is said to be of finite type if
(1)
has a standard basis;
(2)
is an arithmetic root system and for all , where is a pair associated to .
Theorem 3.19.
Let be a Yetter-Drinfeld module. Then is finite dimensional if and only if is of finite type.
Proof.
Firstly, suppose that is finite dimensional. By Corollary 3.16, is an abelian -cocycle on , thus has a standard basis. Let , , and a pair associated to . By Proposition 2.11, is isomorphic to a Nichols algebra in . Note that can be obtained from by change of based groups and twisting, which do not change the pair associted to , thus is also a pair associated to . On the other hand, since is finite dimensional, is an arithmetic root system. Moreover, is finite dimensional implies that the nilpotent index of a root vector of each root is finite, i.e., for all .
Conversely, Suppose that is finite type. By Proposition 2.11, is twist equivalent to a Nichols algebra in , and have the same generalized Dynkin diagram with . So is the arithmetic root system of . Furthermore, is finite dimensional since for all . This implies is infinite dimensional since is twist equivalent to .
∎
The main task of this section is to prove Theorem 3.9. Firstly, we will consider a special case.
4.1. A special case
In this subsection, we will prove the following proposition.
Proposition 4.1.
Suppose is a finite abelian group, is a -cocycle on given by
(4.1)
for all , where .
Let be simple Yetter-Drinfeld modules of type (I) such that and . Then is infinite dimensional.
In what follows, the Yetter-Drinfeld category and modules in it are assumed satisfying the conditions of this proposition.
Lemma 4.2.
The simple module has a basis satisfying
(4.2)
(4.3)
(4.4)
Here such that .
Proof.
Since is of type (I) and , by Lemma 3.5, has a basis such that
Here
Let such that . It is clear that . Let , we get (4.2)-(4.4).
∎
Similar to Lemma 4.2, we have the following two lemmas.
Lemma 4.3.
The simple module has a basis satisfying
(4.5)
(4.6)
(4.7)
Here are numbers such that .
Lemma 4.4.
The simple module has a basis such that
(4.8)
(4.9)
(4.10)
where are numbers
In the following, will be called structure constants of respectively. It is clear that the structure constants depend on the choice of the bases of .
Remark 4.5.
The structure constants can be changed to be and if we transform the bases of and respectively.
For example, let the constants will changed to be . Let the constants will be . Let
the constants will changed to be .
Proposition 4.6.
Keep the previous notations, we have
(1).
If , then is infinite dimensional.
(2).
If , then is infinite dimensional.
(3).
If , then is infinite dimensional.
Proof.
(1). Let and . Then it is obvious that and is a Nichols algebra of diagonal type.
The generalized Dynkin diagram is
Comparing the classification result of finite-dimensional Nichols algebras in , we obtain that is infinite dimensional.
(2). Let . Then it is clear that is a basis of which satisfies
The rest of the proof is similar to (1).
(3). Let . Then it is clear that is a basis of which satisfies
Similarly, let . Then is a basis of and we have
The same as (1), if , then is infinite dimensional.
∎
If the structure constants of satisfy , and , then one can show that are all finite dimensional, the proof is the same as that of [20, Proposition 5.1]. In what follows, we will prove that is infinite dimensional. For -graded homogenous elements , we denote .
Proposition 4.7.
Suppose , and , then is infinite dimensional.
Proof.
Let be the tensor algebra of , the maximal -graded Hopf ideal contained in . Thus by definition. According to Proposition 2.9, is -graded with , where are free generators of .
Next we will prove that is infinite dimensional. Without loss of generality, we can assume that , and by
Remark 4.5.
The remaining proof will be divided into four steps.
Step 1. We will consider the comultiplications of some elements in the spaces . In we have
Step 2. We will prove that and are linear independent in . Firstly we have
(4.27)
Here the third identity follows from (4.11) and (4.14). Similarly we have
(4.28)
(4.29)
and
(4.30)
Now let such that
By (4.27)-(4.30), the homogenous term in the comultiplication of the left hand side of the above identity with degree is
On the other hand, this element must be zero since comultiplication keep -degrees. This implies because and are linear independent by (4.21)-(4.22) and
is a basis of .
Step 3. Let
We will prove that and can not be linear spanned by .
Firstly we have
(4.31)
Similarly, one can show that By (4.27)-(4.30), the comultiplications of and can be written as the forms of
(4.32)
(4.33)
(4.34)
(4.35)
Where . In the following, let be the subset of defined by
(4.36)
Here means the smaller number of . It is clear that , , , . Moreover, for all homogenous elements such that , , we have .
In this subsection, let , be a nonabelian -cocycle on ,
be simple twisted Yetter-Drinfeld modules of type (C1) such that , for . In what follows, we denote , . By (2.1), we have
(4.39)
Where for , for , . Furthermore , we have the following lemma.
Proof of Theorem 3.9.
Let and be the -cocycles of given by
(4.42)
(4.43)
By (4.39) and Lemma 4.8, we have . By (2.13), is an abelian -cocycle of . Let such that , and be the epimorphism determined by
(4.44)
Let be the section of given by
(4.45)
Then we have an object defined by (2.16)-(2.17), and by Lemma 2.8.
By Proposition 2.6, is a -coboundary of . Let be the -cochain of such that . So we have
for all . By Proposition 4.1, is infinite dimensional. So and are also infinite dimensional.
5. Finite quasi-quantum groups over abelian groups
In this section, we will give a classification of finite-dimensional coradically graded pointed coquasi-Hopf algebras over finite abelian groups.
Let be a coradically graded pointed coquasi-Hopf algebra over a finite abelian group . Then for a -cocycle on . Let be the coinvariant subalgebra of .
With these notations, we have
Theorem 5.1.
Assume that is finite dimensional. Then coinvariant subalgebra of is a Nichols algebra in .
Proof.
First we have for is coradically graded. Since is finite dimensional, we have is also finite dimensional since is a subquotient of . This implies that is an abelian -cocycle on
by Corollary 3.16. So we have by [18, Proposition 5.1].
∎
Now we can give a classification of finite-dimensional coradically graded pointed coquasi-Hopf algebras over finite abelian groups.
Theorem 5.2.
(1).
Let be a Yetter-Drinfeld module of finite type. Then is a finite-dimensional pointed coquasi-Hopf algebra.
(2).
Let be a finite-dimensional coradically graded pointed coquasi-Hopf algebra over a finite abelian group, . Then we have for a Yetter-Drinfeld module of finite type
.
Proof.
(1). It follows from Theorem 3.19 that is finite dimensional, thus is finite dimensional.
(2). Let be the coinvariant subalgebra of . Then . Let . Then by Theorem 5.1, is a Nichols algebra in . So for the Yetter-Drinfeld module , and is of finite type since and hence is finite dimensional.
∎
Finally, we will consider the generation problem of pointed finite tensor categories. We partially prove the following conjecture due to Etingof, Gelaki, Nikshych and Ostrik.
Conjeture 5.3.
Every pointed finite tensor categories over a field of characteristic zero is tensor generated by objects of length two.
In fact, this conjecture can be viewed as a generalization of Andruskiewitsch-Schneider conjecture. Let be a pointed finite tensor categories. Then it is well known that for a finite-dimensional pointed coquasi-Hopf algebra , see [12] for details. In [20], we prove the following proposition.
Proposition 5.4.
[20, Proposition 4.10]
Let be a pointed finite tensor category, and a finite-dimensional pointed coquasi-Hopf algebra such that . Then is tensor generated by objects of length two if and only if is generated by
group-like elements and skew-primitive elements.
With the help of Theorem 5.2 and Proposition 5.4, we can prove the following theorem.
Theorem 5.5.
Let be a pointed finite tensor category over a field of characteristic zero such that is an abelian group. Then is tensor generated by objects of length two.
Proof.
Let be a finite-dimensional pointed coquasi-Hopf algebra such that . By Proposition 5.4, we only need to show that is generated by group-like elements and skew-primitive elements. Let be the based field of ,
the algebraically closure of . Let . Then is also a pointed coquasi-Hopf algebra with the structure induced from that of , and it is obvious that is generated by group-like elements and skew-primitive elements if and only if is generated by group-like elements and skew-primitive elements. On the other hand, we have by Theorem 5.2, where and is a Yetter-Drinfeld module of finite type in for some -cocyce on . Thus , and hence are generated by group-like elements and skew-primitive elements .
Therefore is generated by group-like elements and skew-primitive elements.
∎
acknowledgments
This work is supported by the National Natural Science Foundation of China (Nos. 12271243, 12371037, 12131015, 12161141001 and 12371042), the Fundamental Research Funds for the Central Universities (No. SWU-XDJH202305), the Innovation Program for Quantum Science and Technology (No. 2021ZD0302902), and the Natural Science Foundation of Chongqing (No. cstc2021 jcyj-msxmX0714). The first author is also supported by Fuzhou-Xiamen-Quanzhou National Independent Innovation Demonstration Zone Collaborative Innovation Platform under Grant 2022FX5.
References
[1]
Andruskiewitsch, Nicolás; Heckenberger, István; Schneider, Hans-Jürgen: The Nichols algebra of a semisimple Yetter-Drinfeld module. Amer. J. Math. 132 (2010), no. 6, 1493-1547.
[2]
Andruskiewitsch, Nicolás; Schneider, Hans-Jürgen: finite quantum groups and Cartan matrices. Adv. Math 154 (2000), 1-45.
[3]
Andruskiewitsch, Nicolás; Schneider, Hans-Jürgen: On the classification of finite-dimensional pointed Hopf algebras. Ann. of Math. (2) 171 (2010), no. 1, 375-417.
[5]
Angiono, Iván Ezequiel: On Nichols algebras of diagonal type. J. Reine Angew. Math. 683 (2013), 189?251,
[6]
Angiono, Iván Ezequiel; Iglesias, Agustín García: Liftings of Nichols algebras of diagonal type II: all liftings are cocycle deformations. Sel. Math. New Series. 25 (2019), no. 1, Paper No. 5, 95.
[7]
Etingof, Pavel; Gelaki, Shlomo: Finite-dimensional quasi-Hopf algebras with radical of codimension 2. Math. Res. Lett. 11 (2004), no. 5-6, 685-696.
[8]
Etingof, Pavel; Gelaki, Shlomo: On radically graded finite-dimensional quasi-Hopf algebras. Mosc. Math. J. 5 (2005), no. 2, 371-378.
[9]
Etingof, Pavel; Gelaki, Shlomo: Liftings of graded quasi-Hopf algebras with radical of prime codimension. J. Pure Appl. Algebra 205 (2006), no. 2, 310-322.
[10]
Etingof, Pavel; Gelaki, Shlomo: The small quantum groups as a quantum double. J. Algebra 322 (2009), no. 7, 2580-2585.
[16]
Heckenberger, István; Schneider, Hans-Jürgen: Root systems and Weyl groupoids for Nichols algebras. Proc. Lond. Math. Soc. (3) 101 (2010), no. 3, 623-654.
[17]
Huang, Hua-Lin; Liu, Gongxiang; Yang, Yuping; Ye, Yu: Finite quasi-quntum groups of rank two. Trans. Amer. Math. Soc. Ser. B 8 (2021), 635-678.
[18]
Huang, Hua-Lin; Liu, Gongxiang; Yang, Yuping; Ye, Yu: Finite quasi-quntum groups of diagonal type. J. Reine. Angew. Math. 759 (2020), 201-243.
[19]
Huang, Hua-Lin; Yang, Yuping: Quasi-quantum linear spaces. J. Noncommut. Goem. 9 (2015), 1227-1259
[20]
Huang, Hua-Lin; Yang, Yuping; Zhang, Yinhuo: On nondiagonal finite quasi-quantum groups over finite abelian groups. Sel. Math. New Series. 24 (2018), 4197-4221.
[21]
Majid, Shahn: Algebras and Hopf algebras in braided categories. Advances in Hopf algebras (Chicago, IL, 1992), 55-105, Lecture Notes in Pure and Appl. Math., 158, Dekker, New York, 1994.
[22]
Majid, Shahn: Quantum double for quasi-Hopf algebras. Lett. Math. Phys. 45 (1998), no. 1, 1-9.