, , , February 23, 2005

The algebraic deformation theory of Giaquinto and Zhang [GZ] laid out a framework for certain types of deformations of associative algebras, namely those arising from actions of bialgebras on algebras. This work generalized known examples of deformations coming from the action of a Lie algebra (equivalently of its universal enveloping algebra). Giaquinto and Zhang illustrated their ideas by constructing a new deformation formula from the action of a noncommutative bialgebra, namely, the universal enveloping algebra of a certain nonabelian Lie algebra. This Lie algebra is spanned by the

n \times n

matrix units

E_{1, p}, E_{p, n}

for

p = 1, . . ., n

, along with the

n \times n

diagonal matrices of trace 0, and so it is an abelian extension of a Heisenberg Lie algebra when

n \geq 3

, and it is a two-dimensional nonabelian Lie algebra when

n = 2

. In recent work, Grunspan [G] applied the deformation formula for the

n = 2

case to solve the open problem of giving an explicit deformation of the Witt algebra.

In their study of deformations and orbifolds, Căldăraru, Giaquinto, and the second author [CGW] constructed a deformation from a noncocommutative (and noncommutative) bialgebra – the Drinfel'd double of a Taft algebra. This was related to unpublished work of Giaquinto and Zhang and was the first known example of an explicit formula for a formal deformation of an algebra arising from a noncocommutative bialgebra.

The present paper began as an attempt to put the example of [CGW] into a general context. The algebra in [CGW] to undergo deformation was a crossed product of a polynomial algebra and the group algebra of a finite group. We expected that generalizations of the Drinfel'd double of a Taft algebra, such as the two-parameter restricted quantum groups

u_{r, s} (s l_{n})

of our paper [BW3] , would act on similar crossed products, potentially leading to deformations. Indeed,

u_{r, s} (s l_{n})

does act on such a crossed product, but we were unable to generalize the deformation formula of [CGW] directly to deform the multiplication in the crossed product via the

u_{r, s} (s l_{n})

-action. Instead, a generalization in a different direction proved more successful in deforming these particular types of algebras (see [Wi] ). However, as we discuss in Sections 3 through 5 in this paper, certain related noncommutative algebras and their crossed products with group algebras also carry a

u_{r, s} (s l_{n})

-action. Since

u_{r, s} (s l_{n})

is a quotient of the infinite-dimensional two-parameter quantum group

U_{r, s} (s l_{n})

, such algebras also have an action of

U_{r, s} (s l_{n})

, and in some cases they admit formal deformations arising from this action.

In order to obtain these examples, we found it necessary to generalize the definitions of twisting elements and universal deformation formulas given in [GZ] , from elements in bialgebras to operators on particular categories of modules. This we do in Section 2. There are known connections between twisting elements and

R

-matrices of quasitriangular Hopf algebras (see Section 2 for the details). In Section 3, we factor the

R

-matrix of the quasitriangular Hopf algebra

u_{r, s} (s l_{n})

, where

r

and

s

are roots of unity (under a mild numerical constraint), and obtain a twisting element from one of the factors. In the special case of

u_{1, - 1} (s l_{2})

, we show that this twisting element leads to a universal deformation formula, and via this alternate approach, essentially recover the example of [CGW] .

In Section 4, we continue under the assumption that

r

and

s

are roots of unity, but consider the infinite-dimensional quantum group

U_{r, s} (s l_{n})

. We show that an analogue of an

R

-matrix for

U_{r, s} (s l_{n})

gives rise to a universal deformation formula for certain types of algebras. In such formulas, it appears to be necessary to work with the quantum group

U_{r, s} (s l_{n})

, or more particularly with

U_{q, q^{- 1}} (s l_{n})

, rather than with its one-parameter quotient

U_{q} (s l_{n})

, whose relations do not permit the types of actions we describe in Section 5.

For the definitions and general theory in Section 2, we assume only that our algebras are defined over a commutative ring

K

. For the quantum groups in Sections 3 through 5, we assume

K

is a field of characteristic 0 containing appropriate roots of unity.

2 Twisting elements and deformation formulas

Let

B

be an associative bialgebra over a commutative ring

K

with coproduct

Δ : B \to B \otimes B

and counit

ε : B \to K

. An associative

K

-algebra

A

is called a left

B

-module algebra if it is a left

B

-module such that

\begin{matrix} b . 1 = ε (b) 1, and \end{matrix}

(2.1)

\begin{matrix} b . (a a^{'}) = \sum_{(b)} (b_{(1)} . a) (b_{(2)} . a^{'}) \end{matrix}

(2.2)

for all

b \in B

and

a, a^{'} \in A

. Here we have adopted the Heyneman-Sweedler convention for the coproduct,

Δ (b) = \sum_{(b)} b_{(1)} \otimes b_{(2)}

Let

C

be a category of

B

-module algebras. An element

F \in B \otimes B

is a twisting element for

C

(based on

B

) if

\begin{matrix} (ε \otimes i d) (F) = i d \otimes i d = (i d \otimes ε) (F), and \end{matrix}

(2.3)

\begin{matrix} [(Δ \otimes i d) (F)] (F \otimes i d) = [(i d \otimes Δ) (F)] (i d \otimes F) \end{matrix}

(2.4)

as operators on the

K

-modules

A \otimes A

and

A \otimes A \otimes A

, respectively, for all objects

A

C

, where

i d

denotes the identity operator. More generally, we allow

F

to be a formal infinite sum of elements of

B \otimes B

, provided that

F

and the operators on each side of equations ( 2.3 ) and ( 2.4 ) are well-defined operators on the appropriate objects.

The following theorem is essentially the same as [GZ,Thm. 1.3] , but stated in terms of the category

C

. We include a proof for completeness.

Theorem 2.5. Let

A

be an associative algebra in a category

C

B

-module algebras having multiplication map

μ = μ_{A}

, and assume

F

is a twisting element for

C

. Let

A_{F}

denote the

K

-module

A

with multiplication map

μ \circ F

. Then

A_{F}

is an associative algebra with multiplicative identity

1 = 1_{A}

Proof. Equations ( 2.1 ) and ( 2.3 ) imply that

1_{A}

is the multiplicative identity in

A_{F}

. Associativity in

A_{F}

is equivalent to the identity

μ \circ F \circ (μ \otimes i d) \circ (F \otimes i d) = μ \circ F \circ (i d \otimes μ) \circ (i d \otimes F)

of functions from

A \otimes A \otimes A

A

. We prove this identity by applying the associativity of

μ

, ( 2.4 ), and ( 2.2 ) twice:

\begin{matrix} μ \circ F \circ (μ \otimes i d) \circ (F \otimes i d) & = & μ \circ (μ \otimes i d) \circ [(Δ \otimes i d) (F)] \circ (F \otimes i d) \end{matrix}

\begin{matrix} = & μ \circ (i d \otimes μ) \circ [(i d \otimes Δ) (F)] \circ (i d \otimes F) \end{matrix}

\begin{matrix} = & μ \circ F \circ (i d \otimes μ) \circ (i d \otimes F) . \end{matrix}

(This identity may be viewed as a commutative diagram as in [GZ,Thm. 1.3] .) □

Known examples of twisting elements include those arising from quasitriangular Hopf algebras, and we recall these ideas next, with a few details for clarity. A Hopf algebra

H

is quasitriangular if the antipode

S

is bijective, and there is an invertible element

R \in H \otimes H

such that

\begin{matrix} τ (Δ (h)) = R Δ (h) R^{- 1} \end{matrix}

(2.6)

for all

h \in H

, where

τ (a \otimes b) = b \otimes a

\begin{matrix} (Δ \otimes i d) (R) & = & R^{13} R^{23}, and \end{matrix}

(2.7)

\begin{matrix} (i d \otimes Δ) (R) & = & R^{13} R^{12} . \end{matrix}

(2.8)

The notation is standard; for example, if

R = \sum_{i} R_{i}^{1} \otimes R_{i}^{2}

, then

R^{13} = \sum_{i} R_{i}^{1} \otimes 1 \otimes R_{i}^{2}

It can be shown that

R^{12} R^{13} R^{23} = R^{23} R^{13} R^{12}

(see for example [M,Prop. 10.1.8] ). Thus left multiplication of ( 2.7 ) by

R^{12} = R \otimes 1

and left multiplication of ( 2.8 ) by

R^{23} = 1 \otimes R

yields

(R \otimes 1) [(Δ \otimes i d) (R)] = (1 \otimes R) [(i d \otimes Δ) (R)],

which is very similar to ( 2.4 ), but applies to right

H

-module algebras. For left

H

-module algebras,

R^{- 1}

(which equals

(S \otimes i d) (R)

[M,(10.1.10)] ) satisfies ( 2.4 ) (as noted in [GZ,pp. 139–40] ). The element

R_{21} : = τ (R)

can also be shown to satisfy ( 2.4 ) by the following argument:

Interchange the factors in ( 2.7 ) and ( 2.8 ) to obtain

\begin{matrix} (i d \otimes Δ) (R_{21}) & = & R^{21} R^{31}, and \end{matrix}

(2.9)

\begin{matrix} (Δ \otimes i d) (R_{21}) & = & R^{32} R^{31} . \end{matrix}

(2.10)

Using the fact that

R_{21}

is itself an

R

-matrix for

H^{coop}

(that is, for

H

with the opposite coproduct), we see that

R^{32} R^{31} R^{21} = R^{21} R^{31} R^{32} .

(This may also be shown directly.) Thus, multiplying ( 2.9 ) on the right by

R^{32}

and multiplying ( 2.10 ) on the right by

R^{21}

yields

\begin{matrix} [(Δ \otimes i d) (R_{21})] (R_{21} \otimes 1) = [(i d \otimes Δ) (R_{21})] (1 \otimes R_{21}) . \end{matrix}

(2.11)

Again, as

R_{21}

is an

R

-matrix for

H^{coop}

, it satisfies ( 2.3 ) [M,(10.1.11)] . Therefore we have shown the following.

Proposition 2.12. Let

H

be a quasitriangular Hopf algebra with R -matrix

R

. Then

R_{21}

is a twisting element for left

H

-module algebras.

Now assume

A

is an associative

K

-algebra with multiplication map

μ = μ_{A}

. Let

t

be an indeterminate, and extend the scalars for

A

by setting

A [[t]] : = K [[t]] \otimes_{K} A

Here

K [[t]]

is the algebra of formal power series in

t

with coefficients in

K

. The algebra

A [[t]]

is graded with homogeneous components

A [[t]]_{i} = t^{i} A

for each

i \in Z^{\geq 0}

. Tensor powers of

A [[t]]

are also graded in the usual way. A map

f : A [[t]] \to A [[t]]

is said to have

deg (f) \geq j

f

takes

A [[t]]_{i}

\oplus_{k \geq j} A [[t]]_{i + k}

for all

i

A formal deformation of

A

is an associative

K [[t]]

-algebra structure on the

K [[t]]

-module

A [[t]]

for which multiplication takes the form

μ_{A} + t μ_{1} + t^{2} μ_{2} + \dots .

Each

μ_{i} : A \otimes_{K} A \to A

is assumed to be a

K

-linear map, extended to be

K [[t]]

-linear. Associativity imposes certain constraints on the maps

μ_{i}

; for example,

μ_{1}

must be a Hochschild two-cocycle:

μ_{1} (a \otimes b) c + μ_{1} (a b \otimes c) = a μ_{1} (b \otimes c) + μ_{1} (a \otimes b c)

for all

a, b, c \in A

. For further discussion of the conditions on the maps

μ_{i}

, see [GZ,p.141] . Let

B

be a

K

-bialgebra,

C

be a category of

B [[t]]

-module algebras of the form

A [[t]]

for

K

-algebras

A

, and

F

be a twisting element for

C

. We say that

F

is a universal deformation formula for

C

(based on

B

) if

\begin{matrix} F = 1 \otimes 1 + \sum_{j \geq 1} F_{j}^{1} \otimes F_{j}^{2} \end{matrix}

(2.13)

for some elements

F_{j}^{1}, F_{j}^{2}

B [[t]]

, which when viewed as transformations on

A [[t]] \otimes_{K [[t]]} A [[t]]

(for any

A [[t]]

C

) satisfy

deg (F_{j}^{1} \otimes F_{j}^{2}) \geq 1

for each

j \geq 1

In this case,

A [[t]]

has a (new) associative algebra structure with multiplication

μ_{A [[t]]} \circ F

and multiplicative identity

1_{A [[t]]} = 1_{A}

by Theorem 2.5 . In addition, this multiplication has the following property: for all

a, b \in A

μ_{A [[t]]} \circ F (a \otimes b) \equiv μ_{A} (a \otimes b) m o d (t) .

Note that our definition of a universal deformation formula generalizes that of [GZ,Defn.1.13] , where

F

is required to have an expression

1 \otimes 1 + t F_{1} + t^{2} F_{2} + \dots

with each

F_{i} \in B \otimes B

, and where

C

is the category of

B [[t]]

-module algebras arising from

B

-module algebras by extension of scalars. The differences are that here

F

is not required to satisfy ( 2.3 ) and ( 2.4 ) on module algebras outside of category

C

, and

B

is not required to take each algebra

A

C

to itself, but only to

A [[t]]

We illustrate such actions in Section 5.

The next example is well-known.

Example 2.14. Let

K

be a field of characteristic 0,

B

be a commutative

K

-bialgebra, and

C

be any category of

B [[t]]

-module algebras of the form

A [[t]]

where

A

is a

B

-module algebra. Let

P

be the space of primitive elements of

B

and

p \in P \otimes P

. Then

exp (t p) = \sum_{i = 0}^{\infty} \frac{t^{i}}{i!} p^{i}

is a universal deformation formula for

C

(see [GZ,Thm.2.1] for a proof ).

The following theorem on universal deformation formulas generalizes a consequence of [GZ,Thm.1.3andDefn.1.13] .

Theorem 2.15. Let

B

be a

K

-bialgebra and

C

be a category of

B [[t]]

-module algebras of the form

A [[t]]

, where

A

is a

K

-algebra. Let

F

be a universal deformation formula for

C

based on

B

. Then for each

A [[t]]

C

μ_{A [[t]]} \circ F

defines a formal deformation of

A

Proof. Let $A [[t]] \in C$ . By Theorem 2.5 , $μ_{A [[t]]} \circ F$ provides an associative algebra structure on $A [[t]]$ . For $a, a^{'} \in A$ , we may write $F_{j}^{1} . a = \sum_{k \geq 0} φ_{j}^{k} (a) t^{k} and F_{j}^{2} . a^{'} = \sum_{l \geq 0} ψ_{j}^{l} (a^{'}) t^{l},$ where $φ_{j}^{k} : A \to A$ , $ψ_{j}^{l} : A \to A$ are $K$ -linear functions. The coefficient of $t^{i}$ in $μ_{A [[t]]} \circ (F_{j}^{1} \otimes F_{j}^{2}) (a \otimes a^{'})$ is $\sum_{k + l = i} φ_{j}^{k} (a) ψ_{j}^{l} (a^{'})$ . As $deg (F_{j}^{1} \otimes F_{j}^{2}) \geq 1$ , the coefficient of $t^{0}$ in $μ_{A [[t]]} \circ (F_{j}^{1} \otimes F_{j}^{2}) (a \otimes a^{'})$ is 0 for each $j \geq 1$ . Therefore we have $μ_{A [[t]]} \circ F (a \otimes a^{'}) = μ_{A} (a \otimes a^{'}) + t μ_{1} (a \otimes a^{'}) + t^{2} μ_{2} (a \otimes a^{'}) + \dots$ where $μ_{i} (a \otimes a^{'}) = \sum_{j \geq 1} \sum_{k + l = i} φ_{j}^{k} (a) ψ_{j}^{l} (a^{'}) .$ This is necessarily well-defined for each $i$ , since $F$ is assumed to be a well-defined operator on $A [[t]] \otimes_{K [[t]]} A [[t]]$ . As $F_{j}^{1}$ , $F_{j}^{2}$ act linearly on $A$ for each $j$ , the same is true of the functions $φ_{j}^{k}$ , $ψ_{j}^{l}$ , and thus $μ_{i}$ is bilinear. □

3 Twisting elements from finite quantum groups

Drinfel'd doubles of finite-dimensional Hopf algebras provide a wealth of examples of quasitriangular Hopf algebras. In particular, the two-parameter restricted quantum groups

u_{r, s} (s l_{n})

of [BW3] are Drinfel'd doubles under some mild assumptions on

r

and

s

. We discuss implications of this for twisting and deforming algebras later on. The restricted quantum group

u_{r, s} (s l_{n})

is a quotient of the unital associative

K

-algebra

U_{r, s} (s l_{n})

, which we introduce next.

Let

ε_{1}, . . ., ε_{n}

denote an orthonormal basis of a Euclidean space

E = R^{n}

with an inner product

〈, 〉

. Set

Π = {α_{j} = ε_{j} - ε_{j + 1} | j = 1, . . ., n - 1}

and

Φ = {ε_{i} - ε_{j} | 1 \leq i \neq j \leq n}

. Then

Φ

is a finite root system of type A

_{n - 1}

with

Π

a base of simple roots. Let

K

be a field. Choose

r, s \in K^{\times}

with

r \neq s

. The algebra

U = U_{r, s} (s l_{n})

is the unital associative

K

-algebra generated by

e_{j}, f_{j} (1 \leq j < n)

, and

ω_{i}^{\pm 1}, (ω_{i}^{'})^{\pm 1}

(

1 \leq i < n

), subject to the following relations.

(R1) The $ω_{i}^{\pm 1}, (ω_{j}^{'})^{\pm 1}$ all commute with one another and $ω_{i} ω_{i}^{- 1} = ω_{j}^{'} (ω_{j}^{'})^{- 1} = 1,$
(R2) $ω_{i} e_{j} = r^{〈 ε_{i}, α_{j} 〉} s^{〈 ε_{i + 1}, α_{j} 〉} e_{j} ω_{i}$ and $ω_{i} f_{j} = r^{- 〈 ε_{i}, α_{j} 〉} s^{- 〈 ε_{i + 1}, α_{j} 〉} f_{j} ω_{i},$
(R3) $ω_{i}^{'} e_{j} = r^{〈 ε_{i + 1}, α_{j} 〉} s^{〈 ε_{i}, α_{j} 〉} e_{j} ω_{i}^{'}$ and $ω_{i}^{'} f_{j} = r^{- 〈 ε_{i + 1}, α_{j} 〉} s^{- 〈 ε_{i}, α_{j} 〉} f_{j} ω_{i}^{'}$ ,
(R4) $[e_{i}, f_{j}] = \frac{δ_{i, j}}{r - s} (ω_{i} - ω_{i}^{'}) .$
(R5) $[e_{i}, e_{j}] = [f_{i}, f_{j}] = 0 if | i - j | > 1,$
(R6) $e_{i}^{2} e_{i + 1} - (r + s) e_{i} e_{i + 1} e_{i} + r s e_{i + 1} e_{i}^{2} = 0,$ $e_{i} e_{i + 1}^{2} - (r + s) e_{i + 1} e_{i} e_{i + 1} + r s e_{i + 1}^{2} e_{i} = 0,$
(R7) $f_{i}^{2} f_{i + 1} - (r^{- 1} + s^{- 1}) f_{i} f_{i + 1} f_{i} + r^{- 1} s^{- 1} f_{i + 1} f_{i}^{2} = 0,$ $f_{i} f_{i + 1}^{2} - (r^{- 1} + s^{- 1}) f_{i + 1} f_{i} f_{i + 1} + r^{- 1} s^{- 1} f_{i + 1}^{2} f_{i} = 0 .$

The algebra

U

becomes a Hopf algebra over

K

with

ω_{i}

ω_{i}^{'}

group-like and

\begin{matrix} Δ (e_{i}) = e_{i} \otimes 1 + ω_{i} \otimes e_{i}, & Δ (f_{i}) = 1 \otimes f_{i} + f_{i} \otimes ω_{i}^{'}, \end{matrix}

\begin{matrix} ε (e_{i}) = 0, & ε (f_{i}) = 0, \end{matrix}

\begin{matrix} S (e_{i}) = - ω_{i}^{- 1} e_{i}, & S (f_{i}) = - f_{i} (ω_{i}^{'})^{- 1} . \end{matrix}

Let

U^{0}

be the group algebra generated by all

ω_{i}^{\pm 1}

(ω_{i}^{'})^{\pm 1}

, and let

U^{+}

(respectively,

U^{-}

) be the subalgebra of

U

generated by all

e_{i}

(respectively,

f_{i}

We define

ℰ_{j, j} = e_{j} and ℰ_{i, j} = e_{i} ℰ_{i - 1, j} - r^{- 1} ℰ_{i - 1, j} e_{i} (i > j),

ℱ_{j, j} = f_{j} and ℱ_{i, j} = f_{i} ℱ_{i - 1, j} - s ℱ_{i - 1, j} f_{i} (i > j) .

The algebra

U

has a triangular decomposition

U \sim = U^{-} \otimes U^{0} \otimes U^{+}

, and as discussed in [BKL] , the subalgebras

U^{+}

U^{-}

respectively have monomial PBW (Poincaré-Birkhoff-Witt) bases

\begin{matrix} ℰ : = {ℰ_{i_{1}, j_{1}} ℰ_{i_{2}, j_{2}} \dots ℰ_{i_{p}, j_{p}} | (i_{1}, j_{1}) \leq (i_{2}, j_{2}) \leq \dots \leq (i_{p}, j_{p}) lexicographically}, \end{matrix}

(3.1)

\begin{matrix} ℱ : = {ℱ_{i_{1}, j_{1}} ℱ_{i_{2}, j_{2}} \dots ℱ_{i_{p}, j_{p}} | (i_{1}, j_{1}) \leq (i_{2}, j_{2}) \leq \dots \leq (i_{p}, j_{p}) lexicographically} . \end{matrix}

(3.2)

The spaces

U^{\leq 0} = U^{-} \otimes U^{0}

and

U^{\geq 0} = U^{0} \otimes U^{+}

are Hopf subalgebras of

U

We view

U^{\leq 0}

as a Hopf algebra with the opposite coproduct

Δ^{op}

. Then there is a Hopf pairing

(|) : U^{\leq 0} \times U^{\geq 0} \to K

given by

\begin{matrix} \begin{matrix} (f_{i} | e_{j}) & = & \frac{δ_{i, j}}{s - r} \\ (ω_{i}^{'} | ω_{j}) & = & r^{〈 ε_{j}, α_{i} 〉} s^{〈 ε_{j + 1}, α_{i} 〉} \end{matrix} \end{matrix}

(3.3)

for

1 \leq i, j < n

, and all other pairings between generators are 0. Pairings between more complicated expressions can be computed from those values by applying the coproducts

Δ

U^{\geq 0}

and

Δ^{op}

U^{\leq 0}

according to the following rules: (3.4)

(i) $(1 | b) = ε (b), (b^{'} | 1) = ε (b^{'})$
(ii) $(b^{'} | b c) = (Δ^{op} (b^{'}) | b \otimes c) = \sum (b_{(1)}^{'} | b) (b_{(2)}^{'} | c)$
(iii) $(b^{'} c^{'} | b) = (b^{'} \otimes c^{'} | Δ (b)) = \sum (b^{'} | b_{(1)}) (c^{'} | b_{(2)})$

for all

b, c \in U^{\geq 0}

and

b^{'}, c^{'} \in U^{\leq 0}

For the rest of this section, we assume that

r

and

s

are roots of unity. Let

r

be a primitive

d

th root of unity,

s

be a primitive

d^{'}

th root of unity, and

ℓ

be the least common multiple of

d

and

d^{'}

. We also assume that

K

contains a primitive

ℓ

th root of unity

θ

and that

r = θ^{y}

s = θ^{z}

, where

y, z

are nonnegative integers.

It is shown in [BW3] that all

ℰ_{i, j}^{ℓ}

ℱ_{i, j}^{ℓ}

ω_{i}^{ℓ} - 1

, and

(ω_{i}^{'})^{ℓ} - 1

(

1 \leq j \leq i < n

) are central in

U_{r, s} (s l_{n})

. The ideal

I_{n}

generated by these elements is in fact a Hopf ideal [BW3,Thm.2.17] . This is evident from the following expressions for the coproduct given in [BW3,(2.24)andtheensuingtext] :

\begin{matrix} Δ (ℰ_{i, j}^{ℓ}) = ℰ_{i, j}^{ℓ} \otimes 1 + ω_{i, j}^{ℓ} \otimes ℰ_{i, j}^{ℓ} + s^{ℓ (ℓ - 1) / 2} (1 - r^{- 1} s)^{ℓ} \sum_{p = j}^{i - 1} ℰ_{i, p + 1}^{ℓ} ω_{p, j}^{ℓ} \otimes ℰ_{p, j}^{ℓ} \end{matrix}

(3.5)

\begin{matrix} Δ (ℱ_{i, j}^{ℓ}) = 1 \otimes ℱ_{i, j}^{ℓ} + ℱ_{i, j}^{ℓ} \otimes (ω_{i, j}^{'})^{ℓ} + r^{- ℓ (ℓ - 1) / 2} (1 - r^{- 1} s)^{ℓ} \sum_{p = j}^{i - 1} ℱ_{p, j}^{ℓ} \otimes ℱ_{i, p + 1}^{ℓ} (ω_{p, j}^{'})^{ℓ} \end{matrix}

(3.6)

where

ω_{p, j} = ω_{p} ω_{p - 1} \dots ω_{j}

and

ω_{p, j}^{'} = ω_{p}^{'} ω_{p - 1}^{'} \dots ω_{j}^{'}

. As

I_{n}

is a Hopf ideal, the quotient

\begin{matrix} u : = u_{r, s} (s l_{n}) : = U_{r, s} (s l_{n}) / I_{n} \end{matrix}

(3.7)

is a Hopf algebra, called the restricted two-parameter quantum group. Furthermore,

u

is finite-dimensional, as can be readily seen from the PBW-bases ( 3.1 ) and ( 3.2 ).

The algebra

U

is graded by the root lattice

Q

s l_{n}

by assigning

deg (e_{i}) = α_{i}, deg (f_{i}) = - α_{i}, and deg (ω_{i}) = 0 = deg (ω_{i}^{'}) .

Since the generators of

I_{n}

are homogeneous,

u

inherits the grading. Note that by ( 3.3 ) and ( 3.4 ), if

X \in ℰ

and

Y \in ℱ

with

(Y | X) \neq 0

, then

deg (X Y) = 0

Let

ℰ_{ℓ}

denote the set of monomials in

ℰ

having each

ℰ_{i, j}

appear with exponent at most

ℓ - 1

. Identifying cosets in

u

with their representatives, we may assume

ℰ_{ℓ}

is a basis for the subalgebra of

u

generated by the elements

e_{i}

The following proposition will allow us to use the pairing ( 3.3 ) on the quotient algebra

u

Proposition 3.8. The ideal

I_{n}

is contained in the radical of the pairing

(|)

U_{r, s} (s l_{n})

. Thus there is an induced pairing on the quotient

u_{r, s} (s l_{n})

Proof. For the group-like elements, note that for each pair

i, j

\begin{matrix} (ω_{i}^{'} | ω_{j}^{ℓ} - 1) & = & (ω_{i}^{'} | ω_{j}^{ℓ}) - (ω_{i}^{'} | 1) \end{matrix}

\begin{matrix} = & (r^{〈 ε_{j}, α_{i} 〉} s^{〈 ε_{j + 1}, α_{i} 〉})^{ℓ} - 1 = 0, \end{matrix}

r

and

s

are

ℓ

th roots of 1. Thus

ω_{j}^{ℓ} - 1

is in the radical of

(|)

, and similarly for

(ω_{i}^{'})^{ℓ} - 1

Now consider

ℰ_{i, j}^{ℓ}

and let

X

be any monomial in

f_{1}, \dots, f_{n - 1}

of degree

- ℓ α_{i} - \dots - ℓ α_{j}

. Such a monomial is the only type that potentially has a nonzero pairing with

ℰ_{i, j}^{ℓ}

. As

i \geq j

and

ℓ \geq 2

, we may write

X = Y f_{k}

for some

k

. By ( 3.4 )(iii) and ( 3.5 ),

(X | ℰ_{i, j}^{ℓ}) = (Y \otimes f_{k} | ℰ_{i, j}^{ℓ} \otimes 1 + ω_{i, j}^{ℓ} \otimes ℰ_{i, j}^{ℓ} + s^{ℓ (ℓ - 1) / 2} (1 - r^{- 1} s)^{ℓ} \sum_{p = j}^{i - 1} ℰ_{i, p + 1}^{ℓ} ω_{p, j}^{ℓ} \otimes ℰ_{p, j}^{ℓ}) .

Again as

ℓ \geq 2

and

deg f_{k} = - α_{k}

, each term above is 0. □

We will need to apply the following results from [BW3] , where the corresponding pairing differs from ( 3.3 ) by nonzero scalar multiples. This difference does not affect the results. (See the proof of [BW3,Thm.4.8] where a relevant adjustment is made.)

Proposition 3.9. [BW3,(5.8)andLem. 4.1] Let

θ

be a primitive

ℓ

th root of unity, and suppose

r = θ^{y}

and

s = θ^{z}

. Let

b

be the subalgebra of

u

generated by

ω_{i}, e_{i}

(1 \leq i < n)

and

b^{'}

be the subalgebra of

u

generated by

ω_{i}^{'}, f_{i}

(1 \leq i < n)

. Then the Hopf pairing

(|)

b^{'} \times b

satisfying ( 3.3 ) and ( 3.4 ) is nondegenerate if

\begin{matrix} (y^{n - 1} - y^{n - 2} z + - \dots + (- 1)^{n - 1} z^{n - 1}, ℓ) = 1, \end{matrix}

(3.10)

(that is, the first expression in the parentheses is relatively prime to

ℓ

, the least common multiple of the orders of

r

and

s

as roots of 1).

We will use the definition of the Drinfel'd double

D (b)

given in [BW3] :

D (b) = b \otimes (b^{*})^{coop}

as a coalgebra, where

(b^{*})^{coop}

denotes the dual Hopf algebra with the opposite coproduct. As an algebra,

b

and

(b^{*})^{coop}

become subalgebras of

D (b)

under identifications with

b \otimes 1

and

1 \otimes b^{*}

, respectively, and

(1 \otimes b) (a \otimes 1) = \sum_{(a), (b)} (b_{(1)} | S^{- 1} (a_{(1)})) (b_{(3)} | a_{(3)}) a_{(2)} \otimes b_{(2)} .

Proposition 3.11. [BW3,Thm.4.8] Assume

r = θ^{y}

and

s = θ^{z}

, where

θ

is a primitive

ℓ

th root of unity, and suppose that ( 3.10 ) holds. Then there is an isomorphism of Hopf algebras

u_{r, s} (s l_{n}) \sim = D (b)

, where

D (b)

is the Drinfel'd double of the Hopf subalgebra

b

u_{r, s} (s l_{n})

generated by

ω_{i}, e_{i} (1 \leq i < n)

In general, whenever

u \sim = u_{r, s} (s l_{n}) \sim = D (b)

, then

u

is quasitriangular with

R

-matrix,

R = \sum b \otimes b^{*},

where

b

runs over a basis of

b

and

b^{*}

runs over the dual basis of the dual space

b^{*}

, which can be identified with the Hopf subalgebra

b^{'}

u

generated by

ω_{i}^{'}, f_{i}

(

1 \leq i < n

) but with the opposite coproduct

Δ^{op}

[BW3,Lem.4.1] .

To illustrate this result in a very special case, take

r = q

, a primitive

ℓ

th root of unity, and

s = q^{- 1}

. Then

u = u_{q, q^{- 1}} (s l_{n})

is isomorphic to the Drinfel'd double of

b

when

n

and

ℓ

are relatively prime. The quotient of

u

by the ideal generated by the elements

ω_{i}^{'} - ω_{i}^{- 1}

for

1 \leq i < n

is a one-parameter restricted quantum group related to those which have played a significant role in the study of algebraic groups in the case

ℓ = p

, a prime (see for example, [AJS] ).

Lemma 3.12. Assume

u_{r, s} (s l_{n})

is a Drinfel'd double as in Proposition 3.11 . Then its

R

-matrix factors,

\begin{matrix} R = R_{e, f} R_{ω, ω^{'}}, \end{matrix}

(3.13)

with

R_{e, f} = \sum ɛ \otimes ɛ^{*}

, where

ɛ

runs over the basis

ℰ_{ℓ}

ℓ

-power truncated monomials, and

R_{ω, ω^{'}} = \sum w \otimes w^{*}

, where

w

runs over the basis

Ω : = {ω^{\underset{̲}{c}} = ω_{1}^{c_{1}} \dots ω_{n - 1}^{c_{n - 1}} | 0 \leq c_{i} < ℓ

for all

i}

of the group algebra generated by the

ω_{i} (1 \leq i < n)

Proof. Let

{ɛ^{*} | ɛ \in ℰ_{ℓ}}

denote elements of

b^{'}

dual to the vectors of

ℰ_{ℓ}

with respect to this pairing so that

((ɛ^{'})^{*} | ɛ) = δ_{ɛ^{'}, ɛ}

for all

ɛ^{'}, ɛ \in ℰ_{ℓ}

. For example, if

ɛ = ℰ_{1, 1} ℰ_{2, 2} = e_{1} e_{2}

, then

ɛ^{*}

is a linear combination of

ℱ_{1, 1} ℱ_{2, 2} = f_{1} f_{2}

and

ℱ_{2, 1} = f_{2} f_{1} - s f_{1} f_{2}

. Similarly, let

{w^{*} | w \in Ω}

denote the basis of

K G^{'}

dual to

Ω

, where

G^{'}

is the group generated by the

ω_{i}^{'}

. By the triangular decomposition of

u_{r, s} (s l_{n})

, the elements

{ɛ w | ɛ \in ℰ_{ℓ}, w \in Ω}

form a basis of

b

(see for example, [BW3,(2.16)] ). Moreover using

Δ (e_{i}) = e_{i} \otimes 1 + ω_{i} \otimes e_{i}

and

Δ^{op} (f_{i}) = f_{i} \otimes 1 + ω_{i}^{'} \otimes f_{i}

, the properties of the pairing in ( 3.4 ), and degree considerations, we have

\begin{matrix} ((ɛ^{'})^{*} (w^{'})^{*} | ɛ w) & = & ((ɛ^{'})^{*} \otimes (w^{'})^{*} | Δ (ɛ w)) \end{matrix}

\begin{matrix} = & ((ɛ^{'})^{*} \otimes (w^{'})^{*} | Δ (ɛ) (w \otimes w)) \end{matrix}

\begin{matrix} = & ((ɛ^{'})^{*} | ɛ w) ((w^{'})^{*} | w) \end{matrix}

\begin{matrix} = & (Δ^{op} ((ɛ^{'})^{*}) | ɛ \otimes w) δ_{w^{'}, w} \end{matrix}

\begin{matrix} = & ((ɛ^{'})^{*} | ɛ) (1 | w) δ_{w^{'}, w} \end{matrix}

\begin{matrix} = & δ_{ɛ^{'}, ɛ} δ_{w^{'}, w} . \end{matrix}

This calculation shows that

ɛ^{*} w^{*}

is the dual basis element of

ɛ w

relative to the pairing. Thus,

\begin{matrix} R & = & \sum_{ɛ \in ℰ_{ℓ}, w \in Ω} ɛ w \otimes ɛ^{*} w^{*} \end{matrix}

\begin{matrix} = & (\sum_{ɛ \in ℰ_{ℓ}} ɛ \otimes ɛ^{*}) (\sum_{w \in Ω} w \otimes w^{*}) . \end{matrix}

□

It follows from Lemma 3.12 that

R_{21} = R_{f, e} R_{ω^{'}, ω}

, where

R_{f, e} = \sum_{ɛ \in ℰ_{ℓ}} ɛ^{*} \otimes ɛ

(R_{e, f})_{21}

and

R_{ω^{'}, ω} = \sum_{w \in Ω} w^{*} \otimes w = (R_{ω, ω^{'}})_{21}

. Equation ( 2.11 ) holds for

R_{21}

, that is

\begin{matrix} [(Δ \otimes i d) (R_{f, e} R_{ω^{'}, ω})] (R_{f, e} R_{ω^{'}, ω} \otimes 1) = [(i d \otimes Δ) (R_{f, e} R_{ω^{'}, ω})] (1 \otimes R_{f, e} R_{ω^{'}, ω}) . \end{matrix}

(3.14)

We will show that the factor

R_{f, e}

alone satisfies this equation and so is a twisting element.

Let

C

be the category of left

u

-module algebras. Let

G

be the group generated by the

ω_{i}

and

G^{'}

the group generated by the

ω_{i}^{'}

(

1 \leq i < n

). Since

ω_{i}, ω_{i}^{'} (1 \leq i < n)

generate the finite abelian group

G \times G^{'}

of order a power of

ℓ

, and

K

contains a primitive

ℓ

th root of 1, each

A \in C

decomposes into a direct sum of common eigenspaces (or weight spaces) for the

ω_{i}, ω_{i}^{'}

. That is,

A = \oplus_{χ} A_{χ}

for some group homomorphisms

χ : G \times G^{'} \to K^{\times}

, where

ω_{i} . a = χ (ω_{i}) a

and

ω_{i}^{'} . a = χ (ω_{i}^{'}) a

for all

a \in A_{χ}

i = 1, \dots, n - 1

. We will use this decomposition to show that

R_{f, e}

is itself a twisting element for

C

Theorem 3.15. Let

u : = u_{r, s} (s l_{n})

. The factor

R_{f, e}

of the opposite

R_{21}

of the

R

-matrix for

u

is a twisting element for the category of left

u

-module algebras.

Proof. Let

A

be a left

u

-module algebra, and apply both sides of ( 3.14 ) to

a \otimes a^{'} \otimes a^{''} \in A_{χ} \otimes A_{ψ} \otimes A_{φ}

, for weights

χ, ψ, φ

A

. Note that

R_{ω^{'}, ω}

acts as the scalar

ξ (χ, ψ) : = \sum_{w \in Ω} χ (w^{*}) ψ (w)

a \otimes a^{'}

. Applying the left side of ( 3.14 ) to

a \otimes a^{'} \otimes a^{''}

, we have

\begin{matrix} [(Δ \otimes i d) (R_{f, e} R_{ω^{'}, ω})] (R_{f, e} \otimes 1) ξ (χ, ψ) a \otimes a^{'} \otimes a^{''} \end{matrix}

\begin{matrix} = [(Δ \otimes i d) (R_{f, e})] (R_{f, e} \otimes 1) ξ (χ \cdot ψ, φ) ξ (χ, ψ) a \otimes a^{'} \otimes a^{''}, \end{matrix}

since each term of

R_{f, e}

when applied to

a \otimes a^{'}

changes the weight of

a

and the weight of

a^{'}

by multiplication by functions which are inverses of one another, with a net effect of no change at all in the weight. Similarly, applying the right side of ( 3.14 ) to

a \otimes a^{'} \otimes a^{''}

, we have

\begin{matrix} [(i d \otimes Δ) (R_{f, e} R_{ω^{'}, ω})] (1 \otimes R_{f, e}) ξ (ψ, φ) a \otimes a^{'} \otimes a^{''} \end{matrix}

\begin{matrix} = [(i d \otimes Δ) (R_{f, e})] (1 \otimes R_{f, e}) ξ (χ, ψ \cdot φ) ξ (ψ, φ) a \otimes a^{'} \otimes a^{''} . \end{matrix}

Now

R_{ω^{'}, ω}

is itself an

R

-matrix for the group algebra

K [G \times G^{'}]

, because our constructions show that this group algebra is the Drinfel'd double of

K [G]

. Similar calculations to those above prove that

ξ (χ \cdot ψ, φ) ξ (χ, ψ) = ξ (χ, ψ \cdot φ) ξ (ψ, φ)

for all weights

χ, ψ, φ

. (Note this implies that

ξ

is a 2-cocycle for the dual group to

G \times G^{'}

.) Moreover, as

R_{ω^{'}, ω}

is invertible, none of these values is zero. Thus we may cancel the factors

ξ (χ \cdot ψ, φ) ξ (χ, ψ)

and

ξ (χ, ψ \cdot φ) ξ (ψ, φ)

from the above expressions and apply ( 3.14 ) to obtain

[(Δ \otimes i d) (R_{f, e})] (R_{f, e} \otimes 1) (a \otimes a^{'} \otimes a^{''}) = [(i d \otimes Δ) (R_{f, e})] (1 \otimes R_{f, e}) (a \otimes a^{'} \otimes a^{''}),

as desired. We also have

(ε \otimes 1) (R_{f, e}) = 1 \otimes 1 = (1 \otimes ε) (R_{f, e})

by the definition of

R_{f, e}

. □

Examples 3.16. We will give some examples of

u

-module algebras, which may be twisted by

R_{f, e}

according to Theorems 2.5 and 3.15 . First let

V

be the natural

n

-dimensional module for

U = U_{r, s} (s l_{n})

; that is,

V

has a basis

v_{1}, \dots, v_{n}

and

\begin{matrix} e_{i} . v_{j} & = & δ_{i, j - 1} v_{j - 1} \end{matrix}

\begin{matrix} f_{i} . v_{j} & = & δ_{i, j} v_{j + 1} \end{matrix}

\begin{matrix} ω_{i} . v_{j} & = & r^{δ_{i, j}} s^{δ_{i, j - 1}} v_{j}, \end{matrix}

\begin{matrix} ω_{i}^{'} . v_{j} & = & r^{δ_{i, j - 1}} s^{δ_{i, j}} v_{j}, \end{matrix}

\begin{matrix}  \end{matrix}

where

v_{0} = 0 = v_{n + 1}

. Because each

ℰ_{i, j}^{2}

acts as 0,

V

becomes a

u

-module via the induced action. This action extends to a representation of

u

on the tensor algebra

T (V) = \oplus_{k = 0}^{\infty} V^{\otimes k}

, where the action on

V^{\otimes k}

is by

Δ^{k - 1}

for

k \geq 1

. When

k = 0

, then

V^{\otimes k} = K

, and the action is given by the counit. By definition then,

T (V)

is a

u

-module algebra.

Let

J

be the (two-sided) ideal of

T (V)

generated by all elements of the form

v_{i} \otimes v_{j} - r v_{j} \otimes v_{i} (j > i) .

Then

J

is homogeneous,

J = \oplus_{k = 2}^{\infty} J_{k}

, where

J_{k} = J \cap V^{\otimes k}

. In fact,

J

is a

u

-submodule of

T (V)

: A computation such as in [BW2,Prop.5.3] shows that

J_{2} : = J \cap V^{\otimes 2}

is a

u

-submodule of

V^{\otimes 2}

, and as

J = T (V) \otimes J_{2} \otimes T (V)

, the result follows. Thus, the quantum plane

K_{r} [x_{1}, \dots, x_{n}]

, with

x_{i} x_{j} = r x_{j} x_{i}

j > i

, is isomorphic to the

u

-module algebra

T (V) / J

, under the identification

x_{i} = v_{i} + J

Let

x (\underset{̲}{d}) = x (d_{1}, \dots, d_{n}) = x_{1}^{d_{1}} \dots x_{n}^{d_{n}} .

Then the

u

-module action is given by

\begin{matrix} e_{i} . x (\underset{̲}{d}) & = & r^{d_{i} - d_{i + 1} + 1} [d_{i + 1}] x (d_{1}, \dots, d_{i} + 1, d_{i + 1} - 1, \dots, d_{n}), \end{matrix}

\begin{matrix} f_{i} . x (\underset{̲}{d}) & = & r^{d_{i + 1} - d_{i} + 1} [d_{i}] x (d_{1}, \dots d_{i} - 1, d_{i + 1} + 1, \dots, d_{n}), \end{matrix}

\begin{matrix} ω_{i} . x (\underset{̲}{d}) & = & r^{d_{i}} s^{d_{i + 1}} x (\underset{̲}{d}), \end{matrix}

\begin{matrix} ω_{i}^{'} . x (\underset{̲}{d}) & = & r^{d_{i + 1}} s^{d_{i}} x (\underset{̲}{d}), \end{matrix}

where

[d] : = \frac{r^{d} - s^{d}}{r - s} .

Other quotients of

T (V)

provide examples as well: Choose a positive integer

p

, and let

J

be the ideal of

T (V)

generated by the

u

-submodule of

T (V)

having as generators all

v_{i}^{\otimes p}

. As an abbreviation, we will write

v_{i}^{p} = v_{i}^{\otimes p}

, and similarly leave out the tensor product symbol in the notation for words in

v_{1}, \dots, v_{n}

. For example, if

n = 2

and

p = 2

, the

u

-submodule of

V^{\otimes 2}

generated by

v_{1}^{2}

and

v_{2}^{2}

has basis

v_{1}^{2}, v_{2}^{2}, v_{1} v_{2} + s v_{2} v_{1}

. If

n = 2

and

p = 3

, the

u

-submodule of

V^{\otimes 3}

generated by

v_{1}^{3}

and

v_{2}^{3}

has basis

v_{1}^{3}, v_{2}^{3}, v_{1} v_{2}^{2} + s v_{2} v_{1} v_{2} + s^{2} v_{2}^{2} v_{1}, v_{1}^{2} v_{2} + s v_{1} v_{2} v_{1} + s^{2} v_{2} v_{1}^{2}

q

is a primitive third root of 1, and

r = q

s = q^{- 1}

, then

u = u_{q, q^{- 1}} (s l_{2})

acts on the quotient of

T (V)

by the ideal generated by

v_{1}^{2} v_{2} + s v_{1} v_{2} v_{1} + s^{2} v_{2} v_{1}^{2}

v_{1} v_{2}^{2} + s v_{2} v_{1} v_{2} + s^{2} v_{2}^{2} v_{1}

, which is the down-up algebra

A (1 + s^{2}, - s^{2}, 0) = A (1 + r, - r, 0)

in the notation of [BR] . This example is related to one appearing in the work of Montgomery and Schneider [MS] on skew derivations and actions of the double of the Taft algebra.

The family of down-up algebras

A (1 + r, - r, 0)

(where

r

is an arbitrary primitive

ℓ

th root of unity) is especially interesting, as the finite-dimensional modules for these algebras are completely reducible [CM] . This family includes the universal enveloping algebra

U (s l_{2})

for

r = 1

, the universal enveloping algebra

U (o s p_{1, 2})

of the Lie superalgebra

o s p_{1, 2}

for

r = - 1

, and the algebras appearing in this example for

r

a primitive third root of 1.

Example 3.17. Here we generalize the last example by allowing

n

to be arbitrary. Again we suppose

r = q

s = q^{- 1}

for

q

a primitive third root of 1, but take

u = u_{r, s} (s l_{n})

for any

n \geq 3

. Consider the subspace

Y

V^{\otimes 3}

spanned by the following elements:

\begin{matrix} y (i, j, k) & = & v_{i} v_{j} v_{k} + s v_{j} v_{i} v_{k} + s v_{i} v_{k} v_{j} + s^{2} v_{j} v_{k} v_{i} + s^{2} v_{k} v_{i} v_{j} + v_{k} v_{j} v_{i} \end{matrix}

\begin{matrix} (1 \leq i < j < k \leq n), \end{matrix}

\begin{matrix} y (i, i, k) & = & v_{i}^{2} v_{k} + s v_{i} v_{k} v_{i} + s^{2} v_{k} v_{i}^{2} (1 \leq i < k \leq n), \end{matrix}

\begin{matrix} y (i, k, k) & = & v_{i} v_{k}^{2} + s v_{k} v_{i} v_{k} + s^{2} v_{k}^{2} v_{i} (1 \leq i < k \leq n), \end{matrix}

\begin{matrix} y (i, i, i) & = & 0 (1 \leq i \leq n) . \end{matrix}

From the formulas below it is easy to see that

Y

is a

u

-submodule of

T (V)

\begin{matrix} ω_{t} . y (i, j, k) & = & r^{〈 ε_{t}, ε_{i} + ε_{j} + ε_{k} 〉} s^{〈 ε_{t + 1}, ε_{i} + ε_{j} + ε_{k} 〉} y (i, j, k), \end{matrix}

\begin{matrix} ω_{t}^{'} . y (i, j, k) & = & r^{〈 ε_{t + 1}, ε_{i} + ε_{j} + ε_{k} 〉} s^{〈 ε_{t}, ε_{i} + ε_{j} + ε_{k} 〉} y (i, j, k), \end{matrix}

\begin{matrix}  \end{matrix}

for all

1 \leq i \leq j \leq k \leq n

; and using the fact that

r + s = - 1

, we have

\begin{matrix} e_{t} . y (t + 1, j, k) & = & y (t, j, k) (t + 1 \leq j < k) o r (t + 1 < j = k), \end{matrix}

\begin{matrix} e_{t} . y (i, t + 1, k) & = & (- 1)^{δ_{i, t}} y (i, t, k) (i < t + 1 \leq k), \end{matrix}

\begin{matrix} e_{t} . y (i, j, t + 1) & = & (- 1)^{δ_{j, t}} y (i, j, t) (i \leq j \leq t), \end{matrix}

\begin{matrix} e_{t} . y (i, j, k) & = & 0 o t h e r w i s e; \end{matrix}

\begin{matrix} f_{t} . y (t, j, k) & = & (- 1)^{δ_{j, t + 1}} y (t + 1, j, k) (t < j \leq k), \end{matrix}

\begin{matrix} f_{t} . y (i, t, k) & = & (- 1)^{δ_{k, t + 1}} y (i, t + 1, k) (i \leq t < k), \end{matrix}

\begin{matrix} f_{t} . y (i, j, t) & = & y (i, j, t + 1) (i \leq j < t) o r (i < j = t), \end{matrix}

\begin{matrix} f_{t} . y (i, j, k) & = & 0 o t h e r w i s e . \end{matrix}

The restricted quantum group

u

acts on the quotient algebra

A = T (V) / 〈 Y 〉

obtained by factoring out the ideal generated by

Y

. In the quotient, each pair

v_{i}, v_{k}

(

1 \leq i < k \leq n

) generates a down-up algebra

A (1 + s^{2}, - s^{2}, 0) = A (1 + r, - r, 0)

Example 3.18. As a special instance of Theorem 3.15 , consider the algebra

u_{1, - 1} (s l_{2})

for which

R_{f, e} = 1 \otimes 1 - 2 f \otimes e .

(Note we omit subscripts when discussing

s l_{2}

, as there is only one of each type of generator.) We know from Theorem 3.15 that this is a twisting element, but this also may be verified directly quite easily. This quantum group has as a quotient

u_{- 1} (s l_{2}) : = u_{1, - 1} (s l_{2}) / (ω - ω^{'}),

and in

u_{- 1} (s l_{2})

, the images of

e

and

f

commute by relation (R4). Thus the defining relations of

u_{- 1} (s l_{2})

are all homogeneous with respect to powers of

e

Consequently, the following is a twisting element based on

u_{- 1} (s l_{2}) [[t]]

, as may be checked directly:

F = 1 \otimes 1 - 2 t f \otimes e .

Thus

F

is a universal deformation formula for the category of

u_{- 1} (s l_{2}) [[t]]

-module algebras arising from

u_{- 1} (s l_{2})

-module algebras by extension of scalars. In fact, there is a one-parameter family of such deformation formulas, given by

F_{c} = 1 \otimes 1 + c t f \otimes e

, for any

c \in K

. The choice

c = 1

yields precisely the universal deformation formula of [CGW,Lem. 6.2] that was applied to a certain

u_{- 1} (s l_{2})

-module algebra

A

(given by a crossed product of a polynomial ring with a finite group) to obtain a formal deformation of

A

However, it is an accident due to the choices

ℓ = 2

n = 2

for

u_{1, - 1} (s l_{2})

that this technique produces a universal deformation formula directly, as relation (R4) of

u_{r, s} (s l_{n})

is not homogeneous with respect to the powers of the

e_{i}

(nor of the

f_{i}

), and in general, there is no reasonable quotient in which it becomes so. In the next section we will remedy this situation by returning to the infinite-dimensional Hopf algebra

U = U_{r, s} (s l_{n})

and choosing an action of

U

that itself incorporates the indeterminate

t

. Using this infinite-dimensional Hopf algebra necessitates a more complicated (but related) construction of a twisting element.

4 Twisting elements from infinite quantum groups

We consider the infinite-dimensional quantum group

U = U_{r, s} (s l_{n})

, defined in the previous section, where

r

and

s

are roots of unity. We adopt the same notation as before, so that

ℓ

is the least common multiple of the orders of

r

and

s

as roots of

1

, and

Q = \oplus_{i = 1}^{n - 1} Z α_{i}

is the root lattice of

s l_{n}

. We will also assume ( 3.10 ) holds, so that we may use Proposition 3.9 . Let

K

be a commutative

K

-algebra, and extend coefficients for

U

K

U_{K} = K \otimes_{K} U

. (For the applications in the next section,

K

will be an extension of the field

K

by powers of an indeterminate.) As in [BW1] , we will be interested in weight modules, this time for

U_{K}

. We will consider only

U_{K}

-modules that are free as

K

-modules. For every

λ \in Q

, define the corresponding algebra homomorphism

\hat{λ} : U^{0} \to K

\begin{matrix} \hat{λ} (ω_{j}) = r^{〈 ε_{j}, λ 〉} s^{〈 ε_{j + 1}, λ 〉} and \hat{λ} (ω_{j}^{'}) = r^{〈 ε_{j + 1}, λ 〉} s^{〈 ε_{j}, λ 〉}, \end{matrix}

(4.1)

which may be extended to yield an algebra homomorphism from

U_{K}^{0}

K

. As

r

and

s

are roots of unity, the set

\hat{Q} = {\hat{λ} | λ \in Q}

is finite.

Let

M

be a

U_{K}

-module, assumed to be free as a

K

-module. If as a

U^{0} K

-module,

M

decomposes into the direct sum of eigenspaces

M_{χ} = {m \in M | (ω_{i} - χ (ω_{i})) m = 0 = (ω_{i}^{'} - χ (ω_{i}^{'})) m for all i}

for algebra homomorphisms

χ : U^{0} K \to K

, we say

U^{0} K

acts semisimply on

M

. The homomorphisms

χ

such that

M_{χ} \neq 0

are called the weights of

M

. Note that

\begin{matrix} e_{j} . M_{χ} \subseteq M_{χ \cdot \hat{α_{j}}} and f_{j} . M_{χ} \subseteq M_{χ \cdot (\hat{- α_{j}})} \end{matrix}

(4.2)

by (R2), (R3), and ( 4.1 ). It follows that if

M

is a simple

U_{K}

-module with a nonzero weight space

M_{χ}

, then

U^{0} K

acts semisimply on

M

, since the (necessarily direct) sum of the weight spaces is a submodule. In fact, if

M = U K . m

, any cyclic

U K

-module generated by a weight vector

m \in M_{χ}

, then

U^{0} K

acts semisimply on

M

, and all the weights of

M

are of the form

χ \cdot \hat{ζ}

(

ζ \in Q

Let

N

be the category of unital

U K

-modules

M

that are free as

K

-modules and satisfy the following conditions:

( $N 1$ ) $U^{0} K$ acts semisimply on $M$ ;
( $N 2$ ) For each $i, j$ ( $1 \leq j \leq i < n$ ), both $ℰ_{i, j}^{ℓ}$ and $ℱ_{i, j}^{ℓ}$ annihilate $M$ .

Note that

N

is closed under direct sums and quotients, and it follows from ( 3.5 ) and ( 3.6 ) that

N

is closed under tensor products.

Examples of modules

M

satisfying (

N

2) are the

u_{r, s} (s l_{n})

-modules (see the previous section) with scalars extended to

K

, which can be viewed as

U K

-modules.

However we will also be interested in some modules in category

N

that do not correspond to

u_{r, s} (s l_{n})

-modules, where neither

ω_{i}^{ℓ}

nor

(ω_{i}^{'})^{ℓ}

acts as the identity. We remark that (

N

2) has been included so that certain operators will be well-defined. It is not necessarily true that

e_{i}, f_{i}

act nilpotently on all finite-dimensional modules as happens in the non-root of unity case [BW1,Cor.3.14] . The argument used there fails, as

\hat{Q}

is finite.

For any two modules

M

and

M^{'}

N

, we will construct a

U K

-module homomorphism

R_{M^{'}, M} : M^{'} \otimes M \to M \otimes M^{'}

by Jantzen's method (see [J,Ch. 7] ). The main difference between the two-parameter version of Jantzen's method (see [BW1,Sec. 4] ) and what we are about to do here is that now we are assuming both

r

and

s

are roots of unity, a case excluded from consideration in [J, BW1] . As a consequence, we will not know whether

R_{M^{'}, M}

is invertible. However we will make the necessary adjustments to show that most of the arguments of [BW1,Secs. 4,5] , in particular those needed to obtain a twisting element, apply in this context.

The desired function

R_{M^{'}, M}

will be the composition of three

K

-linear functions

τ

\tilde{f}

, and

F

, where

τ

is the map that interchanges tensor factors as before, and

\tilde{f}

and

F

are as follows. (Ultimately we will show that

F

is a twisting element, accounting for the choice of notation.) As

M

satisfies (

N

1), by ( 4.2 ) there are algebra homomorphisms

χ : U_{K}^{0} \to K

such that

M = \oplus_{χ} M (χ)

, where

M (χ)

is a

U_{K}

-submodule having weights contained in

χ \cdot \hat{Q}

. (For example, we may take the sum over a set of representatives

χ

of cosets of

\hat{Q}

A l g_{K} (U_{K}^{0}, K)

.) For the purposes of this section, it will suffice to deal with each summand

M (χ)

separately. Assume

M = M (χ)

and

M^{'} = M^{'} (ψ)

are such modules. Let

λ, μ \in Q

and

m \in M_{χ \cdot \hat{λ}}

m^{'} \in M_{ψ \cdot \hat{μ}}

. Set

\tilde{f} (m \otimes m^{'}) = f_{χ, ψ} (λ, μ) (m \otimes m^{'})

where

f_{χ, ψ} : Q \times Q \to K^{\times}

is given by

f_{χ, ψ} (λ, μ) = ψ (ω_{λ}^{- 1}) χ (ω_{μ}^{'}) (ω_{μ}^{'} | ω_{λ}^{- 1})

with

ω_{λ} = ω_{1}^{λ_{1}} \dots ω_{n - 1}^{λ_{n - 1}}

ω_{μ}^{'} = (ω_{1}^{'})^{μ_{1}} \dots (ω_{n - 1}^{'})^{μ_{n - 1}}

λ = \sum_{i = 1}^{n - 1} λ_{i} α_{i}

μ = \sum_{i = 1}^{n - 1} μ_{i} α_{i}

, and the Hopf pairing is as in ( 3.3 ), ( 3.4 ). Thus

f_{1, 1}

is the function

f

of [BW1,(4.2)] as restricted to the root lattice

Q

, and it can be shown that

\tilde{f}

generalizes the function

\tilde{f}

there by looking at cosets of the weight lattice modulo

Q

. In particular, the following hold:

\begin{matrix} \begin{matrix} f_{χ, ψ} (λ + μ, ν) & = & f_{χ, ψ} (λ, ν) ψ (ω_{μ}^{- 1}) (ω_{ν}^{'} | ω_{μ}^{- 1}) \\ f_{χ, ψ} (λ, μ + ν) & = & f_{χ, ψ} (λ, μ) χ (ω_{ν}^{'}) (ω_{ν}^{'} | ω_{λ}^{- 1}) \\ (ω_{μ}^{'} | ω_{λ}^{- 1}) & = & \hat{μ} (ω_{λ}^{- 1}) = \hat{λ} (ω_{μ}^{'}) . \end{matrix} \end{matrix}

(4.3)

The definition of

F

is similar to that of

Θ

in [BW1,Sec.4] . We will construct

F

as a sum of elements of

U \otimes U

, which then may also be considered elements of

U_{K} \otimes_{K} U_{K}

. The subalgebra

U^{+}

U

generated by

1

and

e_{i}

(

1 \leq i < n

) may be decomposed as

U^{+} = \oplus_{ζ \in Q^{+}} U_{ζ}^{+}

where

U_{ζ}^{+} = {x \in U^{+} | x is homogeneous of degree ζ}

, and

Q^{+} = \oplus_{i = 1}^{n - 1} Z^{\geq 0} α_{i}

For each

ζ \in Q^{+}

, let

{\bar{U}}_{ζ}^{+}

be the linear span of all PBW basis elements in

U_{ζ}^{+}

(see ( 3.1 )) in which the power of each

ℰ_{i, j}

is less than

ℓ

. Similarly define

{\bar{U}}_{- ζ}^{-}

. By Propositions 3.8 and 3.9 , the spaces

{\bar{U}}_{ζ}^{+}

{\bar{U}}_{- ζ}^{-}

are nondegenerately paired under the assumption that ( 3.10 ) holds. We define

\begin{matrix} F = \sum_{ζ \in Q^{+}} F_{ζ} \end{matrix}

(4.4)

where

F_{ζ} = \sum_{k = 1}^{{\bar{d}}_{ζ}} v_{k}^{ζ} \otimes u_{k}^{ζ}

{\bar{d}}_{ζ} = {dim}_{K} {\bar{U}}_{ζ}^{+}

{u_{k}^{ζ}}_{k = 1}^{{\bar{d}}_{ζ}}

is a basis for

{\bar{U}}_{ζ}^{+}

, and

{v_{k}^{ζ}}_{k = 1}^{{\bar{d}}_{ζ}}

the dual basis for

{\bar{U}}_{- ζ}^{-}

. Note that if

{\bar{U}}_{ζ}^{+} = 0

, then

F_{ζ} = 0

, and if

ζ \in Q \ Q^{+}

, we will also set

F_{ζ} = 0

for convenience.

Δ (e_{i}) = e_{i} \otimes 1 + ω_{i} \otimes e_{i}

, for all

x \in U_{ζ}^{+}

we have

Δ (x) \in \sum_{0 \leq ν \leq ζ} U_{ζ - ν}^{+} ω_{ν} \otimes U_{ν}^{+},

where

ν \leq ζ

means

ζ - ν \in Q^{+}

. For each

i

, there are elements

p_{i} (x)

and

p_{i}^{'} (x) \in U_{ζ - α_{i}}^{+}

such that

\begin{matrix} Δ (x) & = & x \otimes 1 + \sum_{i = 1}^{n - 1} p_{i} (x) ω_{i} \otimes e_{i} + π (x), \end{matrix}

\begin{matrix} Δ (x) & = & ω_{ζ} \otimes x + \sum_{i = 1}^{n - 1} e_{i} ω_{ζ - α_{i}} \otimes p_{i}^{'} (x) + π^{'} (x), \end{matrix}

where

π (x)

(respectively,

π^{'} (x)

) is a sum of terms involving products of more than one

e_{j}

in the second factor (respectively, in the first factor). Similarly, if

y \in U_{- ζ}^{-}

, we define

p_{i} (y)

and

p_{i}^{'} (y)

\begin{matrix} Δ (y) & = & y \otimes ω_{ζ}^{'} + \sum_{i = 1}^{n - 1} p_{i} (y) \otimes f_{i} ω_{ζ - α_{i}}^{'} + π (y), \end{matrix}

\begin{matrix} Δ (y) & = & 1 \otimes y + \sum_{i = 1}^{n - 1} f_{i} \otimes p_{i}^{'} (y) ω_{i}^{'} + π^{'} (y) . \end{matrix}

The following identities from [BW1] hold in our context, as the proof consists of calculations that do not use properties of the parameters

r

and

s

Lemma 4.5. [BW1,Lem.4.6,Lem.4.8] For all

x \in U_{ζ}^{+}

x^{'} \in U_{ζ^{'}}^{+}

, and

y \in U^{-}

, the following hold:

(i) $(f_{i} y | x) = (f_{i} | e_{i}) (y | p_{i}^{'} (x)) = (s - r)^{- 1} (y | p_{i}^{'} (x))$ .
(ii) $(y f_{i} | x) = (f_{i} | e_{i}) (y, p_{i} (x)) = (s - r)^{- 1} (y | p_{i} (x))$ .
(iii) $f_{i} x - x f_{i} = (s - r)^{- 1} (p_{i} (x) ω_{i} - ω_{i}^{'} p_{i}^{'} (x))$ .

For all

y \in U_{- ζ}^{-}

y^{'} \in U_{- ζ^{'}}^{-}

, and

x \in U^{+}

, the following hold:

(iv) $(y | e_{i} x) = (f_{i} | e_{i}) (p_{i} (y) | x) = (s - r)^{- 1} (p_{i} (y) | x)$ .
(v) $(y, x e_{i}) = (f_{i} | e_{i}) (p_{i}^{'} (y) | x) = (s - r)^{- 1} (p_{i}^{'} (y) | x)$ .
(vi) $e_{i} y - y e_{i} = (s - r)^{- 1} (p_{i}^{'} (y) ω_{i}^{'} - ω_{i} p_{i} (y))$ .

Since the spaces

{\bar{U}}_{ζ}^{+}

and

{\bar{U}}_{- ζ}^{-}

are nondegenerately paired with dual bases

{u_{k}^{ζ}}_{k = 1}^{{\bar{d}}_{ζ}}

and

{v_{k}^{ζ}}_{k = 1}^{{\bar{d}}_{ζ}}

, for each

x \in {\bar{U}}_{ζ}^{+}

and

y \in {\bar{U}}_{- ζ}^{-}

, we have

\begin{matrix} x = \sum_{k = 1}^{{\bar{d}}_{ζ}} (v_{k}^{ζ} | x) u_{k}^{ζ} and y = \sum_{k = 1}^{{\bar{d}}_{ζ}} (y | u_{k}^{ζ}) v_{k}^{ζ} . \end{matrix}

(4.6)

The next lemma is a modified version of [BW1,Lem.4.10] or [J,Lem.7.1] . The main distinction is that here the identities are as operators on modules rather than as elements of

U \otimes U

Lemma 4.7. Let

ζ \in Q^{+}

and

1 \leq i < n

. Then the following relations hold for operators on tensor products of pairs of modules in category

N

(i) $(ω_{i} \otimes ω_{i}) F_{ζ} = F_{ζ} (ω_{i} \otimes ω_{i})$ and $(ω_{i}^{'} \otimes ω_{i}^{'}) F_{ζ} = F_{ζ} (ω_{i}^{'} \otimes ω_{i}^{'})$ .
(ii) $(e_{i} \otimes 1) F_{ζ} + (ω_{i} \otimes e_{i}) F_{ζ - α_{i}} = F_{ζ} (e_{i} \otimes 1) + F_{ζ - α_{i}} (ω_{i}^{'} \otimes e_{i})$ .
(iii) $(1 \otimes f_{i}) F_{ζ} + (f_{i} \otimes ω_{i}^{'}) F_{ζ - α_{i}} = F_{ζ} (1 \otimes f_{i}) + F_{ζ - α_{i}} (f_{i} \otimes ω_{i})$ .

Proof. We will verify that (ii) holds. Identity (iii) is similar to (ii), and (i) is immediate from (R2) and (R3). In the calculations below, we use Lemma 4.5 (iv)–(vi), ( 4.6 ), and the fact that on a module in category

N

any element of

U_{- ζ}^{-}

acts as its projection in

{\bar{U}}_{- ζ}^{-}

\begin{matrix} (e_{i} \otimes 1) F_{ζ} - F_{ζ} (e_{i} \otimes 1) & = & \sum_{k = 1}^{{\bar{d}}_{ζ}} (e_{i} v_{k}^{ζ} - v_{k}^{ζ} e_{i}) \otimes u_{k}^{ζ} \end{matrix}

\begin{matrix} = & \frac{1}{s - r} \sum_{k} (p_{i}^{'} (v_{k}^{ζ}) ω_{i}^{'} - ω_{i} p_{i} (v_{k}^{ζ})) \otimes u_{k}^{ζ} \end{matrix}

\begin{matrix} = & \frac{1}{s - r} \sum_{k} (\sum_{j = 1}^{{\bar{d}}_{ζ - α_{i}}} (p_{i}^{'} (v_{k}^{ζ}) | u_{j}^{ζ - α_{i}}) v_{j}^{ζ - α_{i}}) ω_{i}^{'} \otimes u_{k}^{ζ} \end{matrix}

\begin{matrix} - \frac{1}{s - r} \sum_{k} ω_{i} (\sum_{j = 1}^{{\bar{d}}_{ζ - α_{i}}} (p_{i} (v_{k}^{ζ}) | u_{j}^{ζ - α_{i}}) v_{j}^{ζ - α_{i}}) \otimes u_{k}^{ζ} \end{matrix}

\begin{matrix} = & \sum_{k} (\sum_{j} (v_{k}^{ζ} | u_{j}^{ζ - α_{i}} e_{i}) v_{j}^{ζ - α_{i}}) ω_{i}^{'} \otimes u_{k}^{ζ} \end{matrix}

\begin{matrix} - \sum_{k} ω_{i} (\sum_{j} (v_{k}^{ζ} | e_{i} u_{j}^{ζ - α_{i}}) v_{j}^{ζ - α_{i}}) \otimes u_{k}^{ζ} \end{matrix}

\begin{matrix} = & \sum_{j} v_{j}^{ζ - α_{i}} ω_{i}^{'} \otimes (\sum_{k} (v_{k}^{ζ} | u_{j}^{ζ - α_{i}} e_{i}) u_{k}^{ζ}) \end{matrix}

\begin{matrix} - \sum_{j} ω_{i} v_{j}^{ζ - α_{i}} \otimes (\sum_{k} (v_{k}^{ζ} | e_{i} u_{j}^{ζ - α_{i}}) u_{k}^{ζ}) \end{matrix}

\begin{matrix} = & \sum_{j = 1}^{{\bar{d}}_{ζ - α_{i}}} (v_{j}^{ζ - α_{i}} ω_{i}^{'} \otimes u_{j}^{ζ - α_{i}} e_{i} - ω_{i} v_{j}^{ζ - α_{i}} \otimes e_{i} u_{j}^{ζ - α_{i}}) \end{matrix}

\begin{matrix} = & F_{ζ - α_{i}} (ω_{i}^{'} \otimes e_{i}) - (ω_{i} \otimes e_{i}) F_{ζ - α_{i}} . \end{matrix}

□

Notice that (ii) and (iii) of the above lemma hold even in the cases where

{\bar{U}}_{ζ}^{+} = 0

but

{\bar{U}}_{ζ - α_{i}}^{+} \neq 0

, because in these cases both sides of each equation annihilate modules in category

N

The following is a modification of [BW1,Thm.4.11] or [J,Thm.7.3] .

Theorem 4.8. Let

M

and

M^{'}

be modules in category

N

. Then the map

R_{M^{'}, M} = F \circ \tilde{f} \circ τ : M^{'} \otimes M \to M \otimes M^{'}

is a homomorphism of

U

-modules.

Proof. We must prove that the action of each generator of

U

commutes with the map

R

. By Lemma 4.7 ,

ω_{i}

and

ω_{i}^{'}

commute with

R

. We will check this for

e_{i}

and leave the similar calculation for

f_{i}

as an exercise. We may assume

M = M (χ)

and

M^{'} = M^{'} (ψ)

for some algebra homomorphisms

χ, ψ : U_{K}^{0} \to K

. Let

m \in M_{χ \cdot \hat{λ}}

m^{'} \in M_{ψ \cdot \hat{μ}}^{'}

. By ( 4.3 ),

\begin{matrix} (F \circ \tilde{f} \circ τ) Δ (e_{i}) (m^{'} \otimes m) \end{matrix}

\begin{matrix} = & (F \circ \tilde{f}) (m \otimes e_{i} m^{'} + e_{i} m \otimes ω_{i} m^{'}) \end{matrix}

\begin{matrix} = & f_{χ, ψ} (λ, μ + α_{i}) F (m \otimes e_{i} m^{'}) + f_{χ, ψ} (λ + α_{i}, μ) F (e_{i} m \otimes ω_{i} m^{'}) \end{matrix}

\begin{matrix} = & f_{χ, ψ} (λ, μ) χ (ω_{i}^{'}) \hat{λ} (ω_{i}^{'}) F (1 \otimes e_{i}) (m \otimes m^{'}) \end{matrix}

\begin{matrix} + f_{χ, ψ} (λ, μ) ψ (ω_{i}^{- 1}) \hat{μ} (ω_{i}^{- 1}) F (e_{i} \otimes ω_{i}) (m \otimes m^{'}) . \end{matrix}

Now we may replace

F

\sum_{ζ \in Q} F_{ζ}

\sum_{ζ \in Q} F_{ζ - α_{i}}

χ (ω_{i}^{'}) \hat{λ} (ω_{i}^{'}) m

ω_{i}^{'} m

, and

ψ (ω_{i}^{- 1}) \hat{μ} (ω_{i}^{- 1}) ω_{i} m^{'}

m^{'}

. Thus we obtain the following expression to which we apply Lemma 4.7 (ii):

\begin{matrix} f_{χ, ψ} (λ, μ) ((\sum_{ζ} F_{ζ - α_{i}}) (ω_{i}^{'} \otimes e_{i}) + (\sum_{ζ} F_{ζ}) (e_{i} \otimes 1)) (m \otimes m^{'}) \end{matrix}

\begin{matrix} = & f_{χ, ψ} (λ, μ) ((e_{i} \otimes 1) (\sum_{ζ} F_{ζ}) + (ω_{i} \otimes e_{i}) (\sum_{ζ} F_{ζ - α_{i}})) (m \otimes m^{'}) \end{matrix}

\begin{matrix} = & Δ (e_{i}) F \tilde{f} τ (m^{'} \otimes m) . \end{matrix}

□

We will need the following relation [BW1,(5.1)] :

\begin{matrix} (y ω_{ζ}^{'} | x ω_{η}) = (y | x) (ω_{ζ}^{'} | ω_{η}) \end{matrix}

(4.9)

for all

x \in U_{γ}^{+}

and

y \in U_{- γ}^{-}

. A similar identity was proven and used in our Lemma 3.12 .

Lemma 4.10. If

x \in {\bar{U}}_{γ}^{+}

and

y \in {\bar{U}}_{- γ}^{-}

, then the following hold as relations of operators on a tensor product of two modules from category

N

(i) $Δ (x) = \sum_{0 \leq ζ \leq γ} \sum_{i, j} (v_{i}^{γ - ζ} v_{j}^{ζ} | x) u_{i}^{γ - ζ} ω_{ζ} \otimes u_{j}^{ζ}$ ,
(ii) $Δ (y) = \sum_{0 \leq ζ \leq γ} \sum_{i, j} (y | u_{i}^{γ - ζ} u_{j}^{ζ}) v_{j}^{ζ} \otimes v_{i}^{γ - ζ} ω_{ζ}^{'}$ .

Proof. As in [BW1,Lem.5.2] , we may write $Δ (x) = \sum_{ζ, i, j} c_{i, j}^{ζ} u_{i}^{γ - ζ} ω_{ζ} \otimes u_{j}^{ζ}$ for some scalars $c_{i, j}^{ζ}$ , this time as an identity of operators on a tensor product of modules from category $N$ . For all $k, l$ , and $ν$ , by ( 3.4 ) and ( 4.9 ), we have $(v_{k}^{γ - ν} v_{l}^{ν} | x) = \sum_{ζ, i, j} c_{i, j}^{ζ} (v_{k}^{γ - ν} | u_{i}^{γ - ζ} ω_{ζ}) (v_{l}^{ν} | u_{j}^{ζ}) = c_{k, l}^{ν} .$ The proof of (ii) is similar. □

Now let

F^{op} = \sum_{γ \in Q^{+}} \sum_{i = 1}^{{\bar{d}}_{γ}} u_{i}^{γ} \otimes v_{i}^{γ}

F^{12} = \sum_{γ \in Q^{+}} \sum_{i} v_{i}^{γ} \otimes u_{i}^{γ} \otimes 1

, similarly

F^{i j}

, and

F_{f}^{i j} = F^{i j} \circ {\tilde{f}}_{i j}

, where

{\tilde{f}}_{i j}

denotes

\tilde{f}

applied to the

i, j

tensor slots. Recall the notation

R^{i j}

for

R

introduced in Section 2, where the notation

R_{21}

was reserved for

τ (R)

Lemma 4.11. The following holds as an identity of operators on a tensor product of three modules from category

N

(Δ \otimes i d) (F^{o p}) \circ {\tilde{f}}_{31} \circ {\tilde{f}}_{32} = F_{f}^{31} \circ F_{f}^{32}

Proof. Let

M, M^{'}, M^{''} \in N

and assume

M = M (χ)

M^{'} = M^{'} (ψ)

, and

M^{''} = M^{''} (φ)

. Let

m \in M_{χ \cdot \hat{λ}}

m^{'} \in M_{ψ \cdot \hat{μ}}

m^{''} \in M_{φ \cdot \hat{ν}}^{''}

. The left side of (i) applied to

m \otimes m^{'} \otimes m^{''}

\begin{matrix} (Δ \otimes i d) (F^{op}) \circ {\tilde{f}}_{31} \circ {\tilde{f}}_{32} \end{matrix}

\begin{matrix} = & f_{φ, ψ} (ν, μ) f_{φ, χ} (ν, λ) (Δ \otimes i d) (\sum_{γ, k} u_{k}^{γ} \otimes v_{k}^{γ}) \end{matrix}

\begin{matrix} = & f_{φ, ψ} (ν, μ) f_{φ, χ} (ν, λ) \sum_{γ, k} \sum_{ζ, i, j} (v_{i}^{γ - ζ} v_{j}^{ζ} | u_{k}^{γ}) u_{i}^{γ - ζ} ω_{ζ} \otimes u_{j}^{ζ} \otimes v_{k}^{γ} \end{matrix}

\begin{matrix} = & f_{φ, ψ} (ν, μ) f_{φ, χ} (ν, λ) \sum_{γ, ζ, i, j} u_{i}^{γ - ζ} ω_{ζ} \otimes u_{j}^{ζ} \otimes (\sum_{k} (v_{i}^{γ - ζ} v_{j}^{ζ} | u_{k}^{γ}) v_{k}^{γ}) \end{matrix}

\begin{matrix} = & f_{φ, ψ} (ν, μ) f_{φ, χ} (ν, λ) \sum_{γ, ζ, i, j} u_{i}^{γ - ζ} ω_{ζ} \otimes u_{j}^{ζ} \otimes v_{i}^{γ - ζ} v_{j}^{ζ} . \end{matrix}

On the other hand, by ( 4.3 ),

\begin{matrix} F_{f}^{31} \circ F_{f}^{32} (m \otimes m^{'} \otimes m^{''}) \end{matrix}

\begin{matrix} = & f_{φ, ψ} (ν, μ) \sum_{η, ζ, i, j} f_{φ, χ} (ν - ζ, λ) u_{i}^{η} m \otimes u_{j}^{ζ} m^{'} \otimes v_{i}^{η} v_{j}^{ζ} m^{''} \end{matrix}

\begin{matrix} = & f_{φ, ψ} (ν, μ) f_{φ, χ} (ν, λ) \sum_{η, ζ, i, j} χ (ω_{ζ}) \hat{λ} (ω_{ζ}) u_{i}^{η} m \otimes u_{j}^{ζ} m^{'} \otimes v_{i}^{η} v_{j}^{ζ} m^{''} . \end{matrix}

Re-summing over

γ = ζ + η

, and replacing

χ (ω_{ζ}) \hat{λ} (ω_{ζ}) m

ω_{ζ} m

, we see that (i) holds. □

The next result is the quantum Yang-Baxter equation for

R = F \circ \tilde{f} \circ τ

Theorem 4.12. Let

M, M^{'}, M^{''} \in N

. Then

R^{12} \circ R^{23} \circ R^{12} = R^{23} \circ R^{12} \circ R^{23}

as maps from

M \otimes M^{'} \otimes M^{''}

M^{''} \otimes M^{'} \otimes M

Proof. If

σ

is a permutation of

{1, 2, 3}

, let

τ_{σ}

denote the corresponding permutation of three tensor factors, that is

τ_{σ} (m_{1} \otimes m_{2} \otimes m_{3}) = m_{σ^{- 1} (1)} \otimes m_{σ^{- 1} (2)} \otimes m_{σ^{- 1} (3)}

In particular, if

σ

equals the transposition

(i j)

, we write simply

τ_{i j}

, as

τ_{i j}

is just the same as

τ

applied to tensor slots

i

and

j

in that case.

Note that

τ_{σ} \circ F_{f}^{i j} = F_{f}^{σ (i) σ (j)} \circ τ_{σ}

for all

σ

, and that

{\tilde{f}}_{31} \circ {\tilde{f}}_{32} \circ F^{12} = F^{12} \circ {\tilde{f}}_{31} \circ {\tilde{f}}_{32}

by two applications of ( 4.3 ). We apply these identities, Theorem 4.8 , and Lemma 4.11 to obtain the following:

\begin{matrix} R^{12} R^{23} R^{12} & = & τ_{12} τ_{23} F_{f}^{31} F_{f}^{32} R^{12} \end{matrix}

\begin{matrix} = & τ_{12} τ_{23} (Δ \otimes i d) (F^{op}) {\tilde{f}}_{31} {\tilde{f}}_{32} F^{12} {\tilde{f}}_{12} τ_{12} \end{matrix}

\begin{matrix} = & τ_{12} τ_{23} (Δ \otimes i d) (F^{op}) F^{12} {\tilde{f}}_{31} {\tilde{f}}_{32} {\tilde{f}}_{12} τ_{12} \end{matrix}

\begin{matrix} = & τ_{12} τ_{23} (Δ \otimes i d) (F^{op}) F^{12} {\tilde{f}}_{12} τ_{12} {\tilde{f}}_{32} {\tilde{f}}_{31} \end{matrix}

\begin{matrix} = & τ_{12} τ_{23} (Δ \otimes i d) (F^{op}) R^{12} {\tilde{f}}_{32} {\tilde{f}}_{31} \end{matrix}

\begin{matrix} = & τ_{12} τ_{23} R^{12} (Δ \otimes i d) (F^{op}) {\tilde{f}}_{32} {\tilde{f}}_{31} \end{matrix}

\begin{matrix} = & τ_{12} τ_{23} F_{f}^{12} τ_{12} F_{f}^{31} F_{f}^{32} \end{matrix}

\begin{matrix} = & F_{f}^{23} τ_{12} τ_{23} τ_{12} F_{f}^{31} F_{f}^{32} \end{matrix}

\begin{matrix} = & F_{f}^{23} τ_{23} τ_{12} τ_{23} F_{f}^{31} F_{f}^{32} \end{matrix}

\begin{matrix} = & R^{23} R^{12} R^{23} . \end{matrix}

□

We will need one more lemma in order to obtain the hexagon identities, from which it will follow that

F

is a twisting element for

N

Lemma 4.13. The following are identities of operators on a tensor product of three modules from

N

(i) $(Δ \otimes i d) (F_{γ}) = \sum_{0 \leq ζ \leq γ} (F_{γ - ζ})^{23} (F_{ζ})^{13} (1 \otimes ω_{ζ}^{'} \otimes 1)$ .
(ii) $(i d \otimes Δ) (F_{γ}) = \sum_{0 \leq ζ \leq γ} (F_{γ - ζ})^{12} (F_{ζ})^{13} (1 \otimes ω_{ζ} \otimes 1)$ .
(iii) ${\tilde{f}}_{12} \circ (F_{ζ})^{13} = (F_{ζ})^{13} \circ (1 \otimes ω_{ζ} \otimes 1) \circ {\tilde{f}}_{12} .$
(iv) ${\tilde{f}}_{23} \circ (F_{ζ})^{13} = (F_{ζ})^{13} \circ (1 \otimes ω_{ζ}^{'} \otimes 1) \circ {\tilde{f}}_{23} .$

Proof. We will check (ii) and (iv); the proofs of (i) and (iii) are similar. By Lemma 4.10 and ( 4.6 ), considering operators on modules we have

\begin{matrix} (i d \otimes Δ) (F_{γ}) & = & \sum_{k} \sum_{ζ, i, j} v_{k}^{γ} \otimes (v_{i}^{γ - ζ} v_{j}^{ζ} | u_{k}^{γ}) u_{i}^{γ - ζ} ω_{ζ} \otimes u_{j}^{ζ} \end{matrix}

\begin{matrix} = & \sum_{ζ, i, j} (\sum_{k} (v_{i}^{γ - ζ} v_{j}^{ζ} | u_{k}^{γ}) v_{k}^{γ}) \otimes u_{i}^{γ - ζ} ω_{ζ} \otimes u_{j}^{ζ} \end{matrix}

\begin{matrix} = & \sum_{ζ, i, j} v_{i}^{γ - ζ} v_{j}^{ζ} \otimes u_{i}^{γ - ζ} ω_{ζ} \otimes u_{j}^{ζ} \end{matrix}

\begin{matrix} = & \sum_{ζ} (F_{γ - ζ})^{12} (F_{ζ})^{13} (1 \otimes ω_{ζ} \otimes 1), \end{matrix}

which proves (ii).

Let

M, M^{'}, M^{''} \in N

and

m \in M_{χ \cdot \hat{λ}}

m \in M_{ψ \cdot \hat{μ}}^{'}

m^{''} \in M_{φ \cdot \hat{ν}}^{''}

. By ( 4.3 ),

\begin{matrix} {\tilde{f}}_{23} (F_{ζ})^{13} (m \otimes m^{'} \otimes m^{''}) & = & f_{ψ, φ} (μ, ν + ζ) \sum_{i} v_{i}^{ζ} m \otimes m^{'} \otimes u_{i}^{ζ} m^{''} \end{matrix}

\begin{matrix} = & f_{ψ, φ} (μ, ν) ψ (ω_{ζ}^{'}) (ω_{ζ}^{'} | ω_{μ}^{- 1}) \sum_{i} v_{i}^{ζ} m \otimes m^{'} \otimes u_{i}^{ζ} m^{''} \end{matrix}

\begin{matrix} = & f_{ψ, φ} (μ, ν) \sum_{i} v_{i}^{ζ} m \otimes ω_{ζ}^{'} m^{'} \otimes u_{i}^{ζ} m^{''} \end{matrix}

\begin{matrix} = & (F_{ζ})^{13} (1 \otimes ω_{ζ}^{'} \otimes 1) {\tilde{f}}_{23} (m \otimes m^{'} \otimes m^{''}), \end{matrix}

which proves (iv). □

Next we will prove the hexagon identities.

Theorem 4.14. Let

M, M^{'}, M^{''} \in N

. Then the following are identities of maps from

M \otimes M^{'} \otimes M^{''}

M^{''} \otimes M \otimes M^{'}

(respectively,

M^{'} \otimes M^{''} \otimes M

(i) $R^{12} \circ R^{23} = (i d \otimes Δ) (F) \circ {\tilde{f}}_{12} \circ {\tilde{f}}_{13} \circ τ_{12} \circ τ_{23}$ .
(ii) $R^{23} \circ R^{12} = (Δ \otimes i d) (F) \circ {\tilde{f}}_{23} \circ {\tilde{f}}_{13} \circ τ_{23} \circ τ_{12}$ .

Proof. We will prove (ii). The proof of (i) is similar. Let

m \otimes m^{'} \otimes m^{''} \in M_{χ \cdot \hat{λ}} \otimes M_{ψ \cdot \hat{μ}} \otimes M_{φ \cdot \hat{ν}}

. By Lemma 4.13 , we have

\begin{matrix} R^{23} R^{12} (m \otimes m^{'} \otimes m^{''}) & = & F^{23} {\tilde{f}}_{23} τ_{23} F^{12} {\tilde{f}}_{12} τ_{12} (m \otimes m^{'} \otimes m^{''}) \end{matrix}

\begin{matrix} = & F^{23} {\tilde{f}}_{23} F^{13} {\tilde{f}}_{13} (m^{'} \otimes m^{''} \otimes m) \end{matrix}

\begin{matrix} = & \sum_{ζ} F^{23} (F_{ζ})^{13} (1 \otimes ω_{ζ}^{'} \otimes 1) {\tilde{f}}_{23} {\tilde{f}}_{13} (m^{'} \otimes m^{''} \otimes m) \end{matrix}

\begin{matrix} = & \sum_{γ} \sum_{0 \leq ζ \leq γ} (F_{γ - ζ})^{23} (F_{ζ})^{13} (1 \otimes ω_{ζ}^{'} \otimes 1) {\tilde{f}}_{23} {\tilde{f}}_{13} (m^{'} \otimes m^{''} \otimes m) \end{matrix}

\begin{matrix} = & \sum_{γ} (Δ \otimes i d) (F_{γ}) {\tilde{f}}_{23} {\tilde{f}}_{13} (m^{'} \otimes m^{''} \otimes m) \end{matrix}

\begin{matrix} = & (Δ \otimes i d) (F) {\tilde{f}}_{23} {\tilde{f}}_{13} τ_{23} τ_{12} (m \otimes m^{'} \otimes m^{''}) . \end{matrix}

□

Finally we show that

F

is a twisting element for

N

Theorem 4.15. Let

M, M^{'}, M^{''}

be modules in category

N

. Then for

F

as defined in ( 4.4 ) we have

[(Δ \otimes i d) (F)] (F \otimes i d) = [(i d \otimes Δ) (F)] (i d \otimes F)

as operators on

M \otimes M^{'} \otimes M^{''}

. Thus

F

is a twisting element for any subcategory of

N

consisting of

U

-module algebras.

Proof. In Theorem 4.14 , multiply (i) by $R^{12}$ on the right, multiply (ii) by $R^{23}$ on the right, and apply Theorem 4.12 to obtain the identity $[(Δ \otimes i d) (F)] \circ {\tilde{f}}_{23} \circ {\tilde{f}}_{13} \circ τ_{23} \circ τ_{12} \circ R^{23} = [(i d \otimes Δ) (F)] \circ {\tilde{f}}_{12} \circ {\tilde{f}}_{13} \circ τ_{12} \circ τ_{23} \circ R^{12}$ as functions from $M \otimes M^{'} \otimes M^{''}$ to $M^{''} \otimes M^{'} \otimes M$ , for any $M, M^{'}, M^{''} \in N$ . It may be checked that this is equivalent to the identity stated in the theorem, using ( 4.3 ). Thus ( 2.4 ) holds, and ( 2.3 ) is immediate from the definition of $F$ . □

We remark that all of the above arguments apply equally well to

U = U_{r, s} (s l_{n})

, if

r s^{- 1}

is not a root of unity and our

U

-module algebras are (possibly infinite) direct sums of modules arising from the category

O

modules of [BW1,Sec. 4] by shifting weights by some functions

χ : U_{K}^{0} \to K

. The function

\tilde{f}

there merely needs to be replaced by our more general function

\tilde{f}

here. The element

Θ

from that paper provides another example of a twisting element similar to the well-known examples arising from

R

-matrices of one-parameter quantum groups.

5 Deformation formulas from infinite quantum groups

Let

U = U_{r, s} (s l_{n})

, and assume throughout this section that

r = θ^{y}

and

s = θ^{z}

are both roots of unity with

y, z

satisfying condition ( 3.10 ). Let

K = K [[t, t^{- 1}]

, Laurent polynomials in

t

with finitely many negative powers of

t

. Let

X

be any finite-dimensional

U

-module such that the extension

X_{K} = K \otimes_{K} X

is a

U_{K}

-module in category

N

. Let

T (X_{K}) = T_{K} (X_{K})

be the tensor algebra of

X_{K}

over

K

, considered as a

U_{K}

-module algebra. As

N

is closed under tensor products and infinite direct sums,

T (X_{K})

is also in category

N

. Therefore Theorem 4.15 may be applied to twist the multiplication of

T (X_{K})

We wish instead to go further and produce a deformation formula that may be applied to quotients of

T (X_{K})

to obtain formal deformations. We define a new action of

U

T (X_{K})

denoted

*

, that is

K

-linear, by specifying

\begin{matrix} e_{i} * x & = & (e_{i} . x) t \end{matrix}

\begin{matrix} f_{i} * x & = & f_{i} . x \end{matrix}

\begin{matrix} ω_{i} * x & = & (ω_{i} . x) t, ω_{i}^{- 1} * x = (ω_{i}^{- 1} . x) t^{- 1} \end{matrix}

\begin{matrix} ω_{i}^{'} * x & = & (ω_{i}^{'} . x) t, (ω_{i}^{'})^{- 1} * x = ((ω_{i}^{'})^{- 1} . x) t^{- 1} \end{matrix}

for all

x \in X

, and extending to

T (X_{K})

via the

U K

-module algebra conditions ( 2.1 ), ( 2.2 ). (We are just forcing the generators of

U K

to act as scalar multiples of their original actions, the scalars being powers of

t

.) The powers of

t

arising in the new action

*

are chosen so that the defining relations of

U

are preserved.

One way to obtain this module structure is to let

U

act on the tensor product

K [[t, t^{- 1}] \otimes_{K} X

via coproducts, making

K [[t, t^{- 1}]

a weight module with the single weight

χ (ω_{i}) = χ (ω_{i}^{'}) = t

so that

e_{i}

f_{i}

annihilate

K [[t, t^{- 1}]

(resulting in the above action on the tensor product

K [[t, t^{- 1}] \otimes_{K} X

via their coproducts). Observe that this new

U

-module structure on

T (X_{K})

yields a module in category

N

, as the only change is in the weights

χ

Theorem 5.1. The operator

F

defined in ( 4.4 ) is a universal deformation formula for any subcategory of

N

(over

K = K [[t, t^{- 1}]

) consisting of

U_{K}

-module algebras for which ( 2.13 ) holds.

Proof. This follows immediately from Theorem 4.15 and the definition of a universal deformation formula for a category. □

Of course, the

U_{K}

-module algebras

T (X_{K})

constructed above satisfy the hypotheses of the theorem, so by Theorem 2.15 ,

μ_{T (X_{K})} \circ F

defines a formal deformation of

T (X)

X

is any finite-dimensional vector space, the Hochschild cohomology groups

H H^{i} (T (X))

are

0

for

i \geq 2

(e.g. see [W,Prop.9.1.6] ). Thus all formal deformations of

T (X)

are equivalent to the trivial one, and so we are interested in applying Theorem 5.1 to proper quotients. For example, we may take the truncated tensor algebra obtained by letting

p

be a fixed positive integer and taking the quotient by the ideal

W_{p} : = \oplus_{k \geq p} X^{\otimes k}

. The second Hochschild cohomology group of a truncated tensor algebra is nontrivial. (See for example [C,Lem.4.1] , where we need to identify our

T (X) / W_{p}

with Cibils'

K Q / ℱ^{p}

Q

the quiver with one vertex and

dim (X)

loops, and

ℱ

the ideal of the path algebra

K Q

generated by the loops.)

Example 5.2. Next we consider some examples that are group crossed products.

Let

V

be the natural

n

-dimensional module for

U = U_{r, s} (s l_{n})

, defined in Section 3. We will specify an action of

U

on a crossed product

T (V) # A

, where

A

is the abelian group

(Z / ℓ Z)^{n - 1}

on generators

a_{1}, \dots, a_{n - 1}

, written multiplicatively.

Choose arbitrary

ℓ

th roots of unity

β_{i}

(

1 \leq i \leq n - 1

), and let an action of

A

T (V)

as automorphisms be defined by

a_{i} . v_{j} = β_{i} v_{j}

for

i = 1, \dots, n - 1

and

j = 1, \dots, n

. The smash product

T (V) # A

is the vector space

T (V) \otimes_{K} A

with the multiplication

(x \otimes a) (y \otimes b) = x (a . y) \otimes a b,

for

x, y \in T (V)

and

a, b \in A

. Define the following action of

U

on the smash product

(T (V) # A) [[t, t^{- 1}]

\begin{matrix} e_{i} . v_{j} & = & δ_{i, j - 1} v_{i} a_{i} t, \end{matrix}

\begin{matrix} f_{i} . v_{j} & = & δ_{i, j} v_{i + 1}, \end{matrix}

\begin{matrix} ω_{i} . v_{j} & = & r^{δ_{i, j}} s^{δ_{i, j - 1}} v_{j} a_{i} t, ω_{i}^{- 1} . v_{j} = r^{- δ_{i, j}} s^{- δ_{i, j - 1}} v_{j} a_{i}^{- 1} t^{- 1}, \end{matrix}

\begin{matrix} ω_{i}^{'} . v_{j} & = & r^{δ_{i, j - 1}} s^{δ_{i, j}} v_{j} a_{i} t, (ω_{i}^{'})^{- 1} . v_{j} = r^{- δ_{i, j - 1}} s^{- δ_{i, j}} v_{j} a_{i}^{- 1} t^{- 1}, \end{matrix}

where

U

acts trivially on

A

. (It is possible to modify this example to involve nontrivial actions of the

ω_{i}

ω_{i}^{'}

A

.) It may be checked that

(T (V) # A) [[t, t^{- 1}]

is a

U

-module algebra. Note that

T (V) # A

is in category

N

, which may be seen by decomposing

K A

as an algebra into a direct sum of copies of

K

(as

A

is a finite abelian group), and partitioning the elements of each

V^{\otimes i} # A

accordingly, (or simply by letting

K = K [[t, t^{- 1}] A

be the ring of coefficients). If elements of

A

are assigned degree 0, it may be verified that

F

satisfies equation ( 2.13 ), and so yields a formal deformation of

T (V) # A

. This example is similar in some respects to that given in [CGW] , although in that example it was possible to take a further quotient of

T (V)

, namely a polynomial algebra, because of the special nature of the parameters.

Just as for

T (V)

, formal deformations of

T (V) # A

are necessarily infinitesimally trivial, as the second Hochschild cohomology group is again trivial. However, we may again truncate the tensor algebra and consider the resulting crossed product with

A

as a

U

-module algebra and corresponding deformations.

References

H.H. Andersen, J.C. Jantzen, and W. Soergel, Representations of quantum groups at a $p$ th root of unity and of semisimple groups in characteristic $p$ : Independence of $p$ , Astérisque 220 Société Math. de France 1994.
G. Benkart, S.-J. Kang, and K.-H. Lee, On the center of two-parameter quantum groups, submitted.
G. Benkart and T. Roby, Down-up algebras, J. Algebra, 209 (1998), 305–344; Addendum 213 (1999), 378.
G. Benkart and S. Witherspoon, Two-parameter quantum groups and Drinfel'd doubles, Algebr. Represent. Theory, 7 (2004), 261–286.
G. Benkart and S. Witherspoon, Representations of two-parameter quantum groups and Schur-Weyl duality, “Hopf Algebras: Proceedings from an International Conference held at DePaul University”, Bergen, Catoiu, and Chin, eds., 65–92, Lecture Notes in Pure and Appl. Math. 237 Dekker, New York, 2004.
G. Benkart and S. Witherspoon, Restricted two-parameter quantum groups, “Representations of Finite Dimensional Algebras and Related Topics in Lie Theory and Geometry”, 293–318. Fields Inst. Commun. 40, Amer. Math. Soc., Providence, RI, 2004.
A. Căldăraru, A. Giaquinto, and S. Witherspoon, Algebraic deformations arising from orbifolds with discrete torsion, J. Pure Appl. Algebra 187 (2004), 51–70.
P.A.A.B. Carvalho and I.M. Musson, Down-up algebras and their representation theory, J. Algebra 228 (2000), 286–310.
C. Cibils, Rigidity of truncated quiver algebras, Adv. Math. 79 (1990), no. 1, 18–42.
A. Giaquinto and J. Zhang, Bialgebra actions, twists, and universal deformation formulas, J. Pure Appl. Algebra 128 (1998), 133–151.
C. Grunspan, Quantizations of the Witt algebra and of simple Lie algebras in characteristic $p$ , J. Algebra 280 (2004), 145–161.
J. C. Jantzen, Lectures on Quantum Groups, vol. 6, Graduate Studies in Math., Amer. Math. Soc., Providence, 1996.
S. Montgomery, Hopf Algebras and Their Actions on Rings, CBMS Conf. Math. Publ., vol. 82, Amer. Math. Soc., Providence, 1993.
S. Montgomery and H.J. Schneider, Skew derivations of finite-dimensional algebras and actions of the double of the Taft Hopf algebra, Tsukuba J. Math. 25 (2001), no. 2, 337–358.
C. A. Weibel, An Introduction to Homological Algebra, Cambridge University Press, 1994.
S. Witherspoon, Skew derivations, Hopf algebras and deformations of group crossed products, submitted (http://www.math.tamu.edu/~sjw/pub/skew.pdf).

Department of Mathematics, University of Wisconsin, Madison,Wisconsin 53706, USA E-mail address : benkart@math.wisc.edu Department of Mathematics, Texas A&M University, College Station,Texas 77843, USA E-mail address : sjw@math.tamu.edu

February 23, 2005

Georgia Benkart

Sarah Witherspoon