Journal of Algebra 252 (2002) 1–19

www.academicpress.com

Generalized Virasoro and super-Virasoro

algebras and modules of the intermediate series
Yucai Su a and Kaiming Zhao b,∗
a Department of Mathematics, Shanghai Jiaotong University, Shanghai 200030, PR China
b Institute of Mathematics, Academy of Mathematics and System Sciences,
Chinese Academy of Sciences, Beijing 100080, China
Received 14 January 2000
Communicated by Georgia Benkart

1. Introduction

Recently, a number of new classes of infinite-dimensional simple Lie algebras

over a field of characteristic 0 were discovered by several authors (see the
references at the end of this paper [7–17]). Among those algebras, are the
generalized Witt algebras. The higher-rank Virasoro algebras was introduced by
Patera and Zassenhaus [1]; they are 1-dimensional universal central extensions of
some generalized Witt algebras [3]; and the higher-rank super-Virasoro algebras
was introduced by Su [2].
In this paper, we will introduce and study generalized Virasoro and super-
Virasoro algebras which are slightly more general than higher-rank Virasoro
and super-Virasoro algebras. In Section 2, we give the definition and all the
automorphisms of generalized Virasoro algebras, which are different from those
given in [1] for higher-rank Virasoro algebras. In Section 3, we exhibit many two-
dimensional subalgebras of generalized Virasoro algebras, which should occur
but were not collected in [1] for higher-rank Virasoro algebras. We believe our
results are correct although they are different from those given in [1]. Section 4
is devoted to determination of all modules of the intermediate series over the
generalized Virasoro algebras; see Theorem 4.6. Then in Section 5 we introduce
the notion of generalized super-Virasoro algebras and determine the modules

* Corresponding author.
E-mail address: [email protected] (K. Zhao).

0021-8693/02/$ – see front matter  2002 Elsevier Science (USA). All rights reserved.
PII: S 0 0 2 1 - 8 6 9 3 ( 0 2 ) 0 0 0 2 1 - 2
2 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19

of the intermediate series over the generalized super-Virasoro algebras; see

Theorem 5.4. Our work here generalizes the results of [1,2] in the following
two ways: first, the ground field is any field of characteristic 0, not necessarily
algebraically closed; second, the subgroup M of F is arbitrary, not necessarily of
finite rank.

2. Generalized Virasoro algebras

Let A be a torsion-free abelian additive group, F any field of characteristic 0,

T a vector space over F. Denote by FA the group algebra of A over F. The
elements t x , x ∈ A, form a basis of this algebra, and the multiplication is defined
by t x · t y = t x+y . The tensor product W = FA ⊗F T is a free left FA-module.
We denote an arbitrary element of T by ∂. We shall write t x ∂ instead of t x ⊗ ∂.
Choose a pairing φ : T × A → F which is F-linear in the first variable and additive
in the second one. For convenience we shall also use the following notations:
φ(∂, x) = ∂, x
= ∂(x) ∀∂ ∈ T , x ∈ A. (2.1)
The following bracket

x   
t ∂1 , t y ∂2 = t x+y ∂1 (y)∂2 − ∂2 (x)∂1 ∀x, y ∈ A, ∂1 , ∂2 ∈ T (2.2)
makes W = W (A, T , φ) into a Lie algebra, which was referred to as a generalized
Witt algebra and studied in [3], The following result was proved there.

Theorem 2.1. (1) W = W (A, T , φ) is simple if and only if A = 0 and φ is

nondegenerate.
(2) If θ : W → W  is a Lie algebra isomorphism between two simple algebras
W = W (A, T , φ) and W  = W (A , T  , φ  ), then there exists χ ∈ Hom(AF∗ )
where F∗ is the multiplicative group F \ {0}, isomorphisms σ : A → A and
τ : T → T  satisfying τ (∂), σ (x)
= ∂, x
such that θ (t x ∂) = χ(x)t σ (x)τ (∂).
(3) For a simple generalized Witt algebra W = W (A, T , φ), the second
cohomology group H 2 (W, F) = 0 if dim T
2, and if T = F∂ is 1-dimensional,
then H 2 (W, F) is 1-dimensional, spanned by the cohomology class [ψ] where
ψ : W × W → F is the 2-cocycle defined by ψ(t x ∂, t y ∂) = δx+y,0∂(x)3 , x, y ∈ A.

In this paper we are interested in the 1-dimensional universal central extension

of the simple algebras W (A, T , φ) in the case when T = F∂ is 1-dimensional. In
fact, this leads us to the following definition.

Definition 2.2. Let M be an abelian additive subgroup of F, the generalized

Virasoro algebra Vir[M] is defined to be the Lie algebra with F-basis {Lµ , c |
µ ∈ M}, subject to the following commutation relations:
 3 
[Lµ , Lν ] = (ν − µ)Lµ+ν + 121
µ − µ δµ+ν,0 c,
for µ, ν ∈ M. (2.3)
[c, Lµ ] = [c, c] = 0,
 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19 3

It is straightforward to verify that Vir[M] is M-graded:

 
FLµ , if µ = 0,
Vir[M] = Vir[M]µ , Vir[M]µ =
FL0 ⊕ Fc, if µ = 0,
µ∈M


Vir[M]µ , Vir[M]ν ⊆ Vir[M]µ+ν . (2.4)
From the above theorem we see that Vir[M]  Vir[M  ] if and only if there exists
a ∈ F∗ such that M  = aM. It is clear that Vir[Z] is the ordinary Virasoro algebra,
and if M is of rank n, Vir[M] is a higher-rank Virasoro algebra defined and studied
in [1]. We can easily deduce the following theorem from Theorem 2.1.

Theorem 2.3. (i) For any χ ∈ Hom(M, F∗ ), the mapping

ϕχ : Vir[M] → Vir[M], Lx → χ(x)Lx ∀x ∈ M; c → c
is an automorphism of Vir[M].
(ii) For any a ∈ S(M) := {α ∈ F | αM = M}, the mapping
ϕa : Vir[M] → Vir[M], Lx → a −1 Lax ∀x ∈ M; c → a −1 c
is an automorphism of Vir[M].
(iii) Aut(Vir[M])  Hom(M, F∗ )  S(M).

In Theorem 3 of [1], mappings in (a) are not automorphisms of Vir[M],

mappings in (c) should be modified to (ii) of the above theorem. The reason of
those errors are the inaccuracy of Lemma 1 in [1] which we shall discuss in the
next section.

3. Finite-dimensional subalgebras of Vir[M]

In this section we shall discuss finite-dimensional subalgebras of Vir[M].


It suffices to consider this problem for only the centerless algebra Vir[M] =
Vir[M]/Fc.


Lemma 3.1. Let T be any finite-dimensional subalgebras of Vir[M]. Then
dim T  3. If dim T = 3, then there exists a nonzero n ∈ M such that T =
FLn + FL0 + FL−n .

Proof. On the contrary, suppose dim T
4 and Xi (i = 1, 2, 3, 4) are four

linearly independent elements in T . We choose a total ordering “” on M
compatible with the group structure. For an element X = µ∈M aµ Lµ ∈ Vir[M], 
we define supp(X) = {µ ∈ M | aµ = 0}. Let max{supp(Xi )} = xi . We may
assume that x1 > 0. If x2 > 0 also, and x1 = x2 , then X1 , X2 generate an infinite-

dimensional subalgebra of Vir[M], a contradiction. So x2 = x1 . By subtracting
4 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19

a suitable multiple of X1 from X2 , we get X2 with max{supp(X2 )}  0. In

this way we may assume that max{supp(Xi )}  0 for i = 2, 3, 4. We consider
min{supp(Xi )} = xi  0 for i = 2, 3, 4. Assume x2 < 0. By subtracting suitable
multiples of X2 from X3 , X4 we get respectively two linearly independent
elements X3 , X4 such that X3 , X4 ∈ FL0 , a contradiction. Thus dim T  3.
If dim T = 3, from the above discussion we can choose a basis {Y1 , Y2 , Y3 }
such that supp(Y1 )
0, supp(Y2 ) = 0, supp(Y3 )  0. Then there exists a nonzero
integer n such that T = FLn + FL0 + FL−n . ✷

Lemma 3.2. (i) Let x ∈ M \ {0}. Then for any positive integer n and any α ∈ F,

the following two elements span a two-dimensional subalgebra of Vir[M]:
X = L0 + αL−x , Y = exp(α ad L−x )Lnx .

(ii) If T is a two-dimensional subalgebra of Vir[M], then T is not abelian.
If further X = L0 + αL−x ∈ T , then there is an element Y ∈ T such that
Y = exp(α ad L−x )Lnx for some integer n > 0.

Proof. Part (i) is easy to verify. We omit the details.

(ii) It is clear that T is not abelian. Since dim T = 2, there exists another
Y ∈ T \ FX such that [X, Y ] = aX + bY for some a, b ∈ F. Further we can change
Y so that [X, Y ] = bY = 0. Applying exp(ad(−aL−x )) to [X, Y ] = bY , we get

    
L0 , exp ad(−bL−x ) Y = exp ad(−bL−x ) Y,
where exp(ad(−bL−x ))Y can be an infinite sum. Thus there exists a positive in-
teger n such that exp(ad(−bL−x ))Y = b Lnx , i.e., Y = b exp(b ad L−x )Lnx . ✷


The above lemma does not determine all 2-dimensional subalgebras of Vir[M].
Let X = (3/16)L0 + L1 + L2 , Y = (3/16)L0 + (1/16)L−1 + (1/16 )L−2 , then
2

FX + FY is a subalgebra. This subalgebra and the ones in Lemma 3.2 are

not included in [1, Lemma 1 and Theorem 4]. There are certainly some other

subalgebras. It is not easy to determine all 2-dimensional subalgebras of Vir[M].
Here we are not able to solve this problem, but we can reduce this problem to

determining all 2-dimensional subalgebras of Vir[Z].


Lemma 3.3. Let {X, Y } be a basis of a two-dimensional subalgebra T of Vir[M].
Then span{supp(X), supp(Y )} is a group of rank 1, i.e., isomorphic to Z.

Proof. It is clear that span{supp X, span Y } is a free group of finite rank. We may
assume that [X, Y ] = bY = 0. Choose a total ordering “” on M compatible
with the group structure. If max{supp(Y )} > 0 and max{supp(X)} > 0, they
must be equal. By subtracting a suitable multiple of Y from X, we get X with
max{supp(X )} = 0. Then [X , Y ] = bY = 0. If supp(X ) = {0}, we obtain that
supp(Y ) is a singleton, the lemma is true in this case. Suppose min{supp(X )} < 0.
 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19 5

If min{supp(Y )} < 0, they must be equal since otherwise dim T > 2. By

subtracting a suitable multiple of X from Y , we get Y  with min{supp(Y  )}
0.
Since span{supp(X), supp(Y )} = span{supp(X ), supp(Y  )} if it is a group of
rank > 1, then we can choose another ordering  of M such that one of the
following is true:

(1) max{supp(Y  )} > 0 and max{supp(X )} > 0;

(2) min{supp(X )} < 0 and min{supp(Y  )} < 0.

In either cases we deduce that dim T > 2, a contradiction. Thus our lemma
follows. ✷

4. Modules of the intermediate series

Let Vir[M] be the generalized Virasoro algebra defined in Definition 2.2.

A Harish–Chandra module over Vir[M] is a module V such that

V= Vλ , Vλ = {v ∈ V | L0 v = λv}, dim Vλ < ∞, ∀λ ∈ F. (4.1)
λ∈F

Definition 4.1. A module of the intermediate series over Vir[M] is an indecom-

posable Harish–Chandra module V such that dim Vλ  1 for all λ ∈ F.

For any a, b ∈ F, as in the Virasoro algebra case, one can define the three series
of Vir[M]-modules Aa,b (M), Aa (M), Ba (M); they all have basis {vµ | µ ∈ M}
with actions cvµ = 0 and
Aa,b (M): Lµ vν = (a + ν + µb)vµ+ν ; (4.2a)
Aa (M): Lµ vν = (ν + µ)vµ+ν , if ν = 0,
Lµ v0 = µ(µ + a)vµ ; (4.2b)
Ba (M): Lµ vν = νvµ+ν , if ν = −µ,
Lµ v−µ = −µ(µ + a)v0 ; (4.2c)

for all µ, ν ∈ M. We use Aa,b (M), Aa (M), Ba (M) to denote the nontrivial sub-
quotient of Aa,b (M), Aa (M), Ba (M), respectively. As results in [4], Aa,b (M) =
Aa,b (M) if and only if a ∈ M and b = 0, or a ∈ M and b = 1. Note that we made
a slight change in the modules Aa (M), Ba (M). But the notation here is obviously
neater and simpler than the old ones. You can see this from the following theorem
which is similar to [2, Proposition 2.2].

Theorem 4.2. Among the Vir[M]-modules Aa,b (M), Aa (M), Ba (M) for a, b ∈ F,
and their nontrivial subquotients, we have only the following module isomor-
phisms:
6 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19

(i) Aa,b (M)  Aa  ,b (M) if a − a  ∈ M;

(ii) Aa,0(M)  Aa  ,1 (M) for a ∈ / M with a − a  ∈ M;
(iii)  
Aa,b (M)  Aa  ,b (M) if a − a  ∈ M;
(iv) Aa,0(M)  Aa  ,1 (M) for a ∈ F with a − a  ∈ M;
(v) Aa (M)  Bb (M)  A0,0 (M) for a, b ∈ F.

The following lemma is quite clear.

Lemma 4.3. Suppose M0 ⊆ M is a subgroup of M, and V is a Harish–Chandra

module over Vir[M] with a weight α. If

 Aa,b (M), α = a,


 a
B (M), α = 0,
V  Aa (M), α = 0,


 A0,0 (M) ⊕ Fv0 , α = 0,

A0,0 (M), α = 0,
for some a, b ∈ F, then there exists x0 ∈ M such that

 Aa,b (M0 ),


  Ba (M0 ),
V (α + x0 , M0 ) := Vα+x0 +z  Aa (M0 ),

 
z∈M0  A0,0(M0 ) ⊕ Fv0 ,

A0,0(M0 ),
respectively, and in the last four cases α + x0 ∈ M0 .

Su [2] proved the following theorem.

Theorem 4.4. Suppose M ⊆ F is an additive subgroup of rank n, and F is

algebraically closed of characteristic 0. A module of the intermediate series
over Vir[M] is isomorphic to one of the following: Aa,b (M), Aa (M), Ba (M)
for a, b ∈ F, and their nonzero subquotients.

It is natural to ask: are these three series of modules the only modules of the
intermediate series over the generalized Virasoro algebra Vir[M]? You will see the
affirmative answer. Before we proceed the proof, we need an auxiliary theorem.
Similar to Theorem 2.3 of [5] (where the result is the same as in the following
theorem but only for Vir[Z]) we have following result.

Theorem 4.5. Suppose M ⊆ F is an additive subgroup of rank n. Let V =

j ∈M Vj be a M-graded Vir[M]-module with dim V > 1 and dim Vj  1 for
all j ∈ M. Suppose there exists a ∈ F such that L0 acts on Vj as the scalar a + j .
Then V is isomorphic to one of the following for appropriate α, β ∈ F:

(i) Aα,β (M),

 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19 7

(ii) A0,0 ⊕ Fv0 as direct sum of Vir[M]-modules,

(iii) Aα (M),
(iv) Bα (M),
(v) A0,0 (M).

Proof. We may assume that Z ⊂ M. Denote the algebraically closed extension

F. Write sl2 = 
of F by  Fd1 + 
Fd0 + Fd−1 which is the 3-dimensional simple
Lie algebra. Consider the  F extensions V = 
F ⊗F V , Vir[M] = F ⊕F Vir[M],
Vi =F ⊕F Vi . It is well known that any submodule of a M-graded module is still
M-graded.

Claim 1. V has no finite-dimensional Vir[M]-submodule of dimension > 1.

Suppose, to the contrary, that U is a finite-dimensional Vir[M]-submodule

of V of dimension > 1. From [2, Theorem 2.1], we see that each irreducible
subquotient of U must be a 1-dimensional trivial Vir[M]-module. Contrary to the
fact that each weight space of V with respect to L0 is 1-dimensional. So Claim 1
holds.
This claim implies that if U is a subquotient of V then dim U = 1 or ∞.

Claim 2. If U is a Vir[M]-subquotient of V and dim U = ∞, then U ∩ Va+x = 0

for all x ∈ M with x + a = 0.

Suppose, to the contrary, that U ∩ Va+x0 = 0 for x0 ∈ M with x0 + a = 0.

Choose x1 ∈ M such that U ∩ Va+x1 = 0. Consider the Vir[Z(x1 − x0 )]-module
U  = k∈Z Va+x0 +k(x1 −x0 ) . By [5, Theorem 2.3] we must have U   A0,0 (Z(x1 −
x0 )). This forces a + x0 = 0, contrary to the choice of x0 . Thus Claim 2 follows.
We shall determine V case by case.

Case 1. V is irreducible.
It follows from [2, Theorem 2.1] that V  Aα,β (M) for appropriate α, β ∈  F.

Note that α ∈ F. If V  A0,0(M), we can choose a basis {vi | i ∈ M with i = 0} for
V such that Li vj = j vi+j with v0 = 0. There exists γ ∈ 
F such that γ v1 ∈ V \ {0}.
 
Then γ vi ∈ V . Thus V  A0,0 (M). If V  A0,0(M), we can choose a basis
{vi | i ∈ M} for V such that
Li vj = (j + α + iβ)vi+j . (4.3)
There exists an integer i0 such that L±1 vi = 0 for all integer i
i0 . We may
assume that γ vi0 ∈ V where γ ∈ F \ {0}. Define

i−1
wi+i0 = γ Li1 vi0 = γ (α + β + j + i0 )vi+i0 ∀i
0.
j =0
8 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19

Then wi ∈ V for all i
i0 , and Lj wi ∈ V for all j
−2, i > i0 + 2. Using (4.3)
we see that
L−1 wi+i0 = (α + β + i0 + i − 1)(α − β + i0 + i)wi+i0 −1 ,
L−2 wi+i0 = (α + β + i0 + i − 1)(α + β + i0 + i − 2)
× (α − 2β + i0 + i)wi+i0 −1 .
It follows that
(α + β + i0 + i − 1)(α − β + i0 + i) ∈ F ∀i
2,
(α + β + i0 + i − 1)(α + β + i0 + i − 2)(α − 2β + i0 + i) ∈ F ∀i
2.
We can deduce that β(β − 1), −2β(β − 1)(β − 2) ∈ F. Thus we get β ∈ F.
Therefore if we replace vi with γ vi for all i ∈ Z, then vi ∈ V and (4.3) still holds.
Thus V  Aα,β (M) with α, β ∈ F.

Case 2. V is reducible.

Let U be a proper submodule of V . From Claims 1 and 2 we know that U is

irreducible and that dim U = 1 or ∞.
Subcase 1. dim U = 1.
Setting U = V i0 , we see that L0 V i0 = 0. Then V /U is irreducible. From
Case 1 we must have V /U  A0,0 (M) :=  F ⊗F A0,0 (M). If V is decomposable,
then V = A0,0 (M)⊕V 0 as Vir-modules, which implies that V = A0,0(M)⊕V0 as
Vir-modules. If V is indecomposable, by [2, Theorem 2.1] we have V  A0,0 (M),
or V  B α (M) for appropriate α ∈ 
F. In the first case we have V  A0,0(M).
In the second case, by changing the basis we can see that, in fact, α ∈ F and
V  Bα (M).
Subcase 2. dim U = ∞.
Claim 1 tells us that dim(V /U ) = 1, and V /U = V 0 . This implies that
L0 V0 = 0 and V = U ⊕ V 0 as subspaces. We deduce that U is irreducible. By
Case 1 we have U  A0,0 (M). If V is decomposable, then V = A0,0 (M) ⊕ V i0
as submodules. And it follows that V = A0,0 (M) ⊕ V0 as Vir-submodules. If V
is indecomposable, then [2, Theorem 2.1] ensures V  A0,1 (M), or V  Aα (M)
for appropriate α ∈ 
F. In the first case we have V  A0,1 (M). In the second case,
by changing the basis we can see that, in fact, α ∈ F and V  Aα (M). ✷

Now we are ready to classify modules of the intermediate series over Vir[M]
for any M and any field F of characteristic 0.

Theorem 4.6. A module of the intermediate series over Vir[M] is isomorphic to

one of the following: Aa,b (M), Aa (M), Ba (M) for a, b ∈ F, and their nonzero
subquotients.
 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19 9

Proof. Suppose V is a module of the intermediate series. As it is indecomposable,

there exists some a ∈ F such that

V= Vµ+a . (4.4)
µ∈M

If a ∈ M, we choose a = 0.
For x ∈ M and any subgroup N of M, we define

V (a + x, N) = Va+x+z . (4.5)
z∈N

It clear that V (a + x, N) is a Vir[N]-module.

If dim V = 1, then V is the 1-dimensional trivial Vir[M]-module. Next we
shall always assume that dim V > 1.
If M  Z, [5, Theorem 2.3] ensures that Theorem 4.6 holds. Next we shall
always assume that M  Z.

Claim 1. Va+x = 0 for all x ∈ M with a + x = 0.

Suppose, to the contrary, that Va+x0 = 0 for x0 ∈ M with x0 + a = 0.

Choose
x1 ∈ M such that Va+x1 = 0. Consider the Vir[Z(x1 − x0 )]-module U =

k∈Z Va+x0 +k(x1 −x0 ) . By [5, Theorem 2.3] we must have U  A0,0 (Z(x1 − x0 )).
This forces a + x0 = 0, contrary to the choice of x0 . Thus Claim 1 follows.
From Claim 1 and Theorem 4.5 we know that c acts trivially on V .
Now we fix an arbitrary nonzero z0 ∈ M, let {xi | i ∈ I } be the set of all
of cosets of Zz0 in M. If a ∈ M, we choose a0 = 0. It is clear
representatives
that V = i∈I V (a + xi , Zz0 ), see (4.5) for the definition. From Claim 1 and
Theorem 4.5 we know that, for any i ∈ I \ {0}, there exists bi ∈ F such that
V (a + xi , Zz0 )  Aa+xi ,bi (Zz0 ), (4.6)
and



Aa+x0 ,b0 (Zz0 ),

 Bα (Zz0 ),

V (a + x0 , Zz0 )  Aα (Zz0 ), for some b0 , α ∈ F.

 A (Zz0 ) ⊕ Fv0 ,


 A0,0
 (Zz ),
0,0 0

Suppose M0 is a maximal subgroup of M satisfying the following two

conditions:

(C1) Let {xi | i ∈ I  } be the set of all representatives of cosets of M0 in M with

0 ∈ I  and x0 = 0. Then for all i ∈ I  \ {0},
V (a + xi , M0 )  Aa+xi ,bi (M0 ), for some bi ∈ F;
10 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19




Aa+x0 ,b0 (M0 ),


 α (M0 ),
B
(C2) V (a + x0 , M0 )  Aα (M0 ), for some b0 , α ∈ F.

 
 A0,0 (M0 ) ⊕ Fv0 ,

 A (M ),
0,0 0

From the above discussion and Zorn’s Lemma we know that such an M0 exists.
It suffices to show that M0 = M.
Otherwise we suppose M0 = M. So |I  | > 1, and x1 ∈ M \ M0 for 1 ∈ I  .
Denote M1 = M0 + Zx1 . Let {yi | i ∈ J } be the set of all representatives of
cosets of M1 in M with 0 ∈ J and y0 = 0, and {ix1 | i ∈ K} be the set of all
representatives of cosets of M0 in M1 , where K ⊂ Z.

Claim 2. For any fixed j ∈ J we have


 A
 a+yj ,α 1
(M ),


 α 1
B (M ),
V (a + yj , M1 )  Aα (M1 ), for some α ∈ F.

 A0,0 (M1 ) ⊕ Fv0 ,


 A (M ),
0,0 1

We shall show this claim in two cases.

Case 1. For all i ∈ K,

V (a + ix1 + yj , M0 )  Aa+ix1 +yj ,bi (M0 ), for some bi ∈ F. (4.7)
From Theorem 4.2, we have two choices for some bi . From (4.7) we deduce
that
V (a + ix1 + yj , Zz0 )  Aa+yj +ix1 ,bi (Zz0 ). (4.8)
For any z0 ∈ M0 \ {0}, rank(Zz0 + Zxi ) = 1 or 2. Then for all i ∈ K, [2,
Theorem 2.1] and Lemma 4.3 ensure that

V (a + ix1 + yj , Zz0 + Zx1 ) = V (a + yj , Zz0 + Zx1 )


= Va+yj +z  Aa+yj ,b (Zz0 + Zx1 ) for some b ∈ F.
z∈Zz0 +Zx1

Thus for all i ∈ K,

V (a + ix1 + yj , Zz0 )  Aa+yj +ix1 b (Zz0 ), (4.9)

V (a + yj , Zx1 )  Aa+yj ,b (Zx1 ). (4.10)
Combining (4.8) with (4.9), we can choose bi such that b = bi , for all i ∈ Z; i.e.,
V (a + ix1 + yj , M0 )  Aa+ix1 +yj ,b (M0 ). (4.11)
 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19 11

Using (4.10) and (4.11), we choose a basis vyj +z ∈ Va+yj +z for all z ∈ M1 and
all i ∈ K as follows:

(N1) Choose vkx1 +yj ∈ Va+kx1 +yj for all k ∈ Z such that
Lix1 vyj +kx1 = (a + yj + kx1 + ix1 b)vyj +kx1 +ix1 ∀i, k ∈ Z;
(N2) For any i ∈ K, choose vix1 +yj +z ∈ Va+ix1 +yj +z for all z ∈ M0 \ {0} such
that
Ly vyj +ix1 +z = (a + yj + ix1 + z + by)vyj +ix1 +z+y ∀y, z ∈ M0 .

Applying Lemma 4.3 to (4.10) gives

V (a + yj , Zz0 + Zx1 )  Aa+yj ,b (Zz0 + Zx1 ). (4.12)
Then we can choose a basis vy j +z ∈ Va+yj +z for all z ∈ Zz0 + Zx1 with vy j = vyj
such that
Ly vy j +ix1 +z = (a + yj + ix1 + z + by)vy j +ix1 +z+y .
Combining this with (N1) and (N2), we deduce that vy j +z = vyj +z for all z ∈
Zz0 + Zx1 ; i.e.,
Ly vy j +z = (a + yj + z + by)vy j +z+y ∀y, z ∈ Zz0 + Zx1 .
Since z0 ∈ M0 is arbitrary, the above equation holds for all y, z ∈ M1 . This yields
that V (a + yj , M1 )  Aa+yj ,b (M1 ). So Claim 2 is true for this case.

Case 2. a = 0 and



Bα (M0 ),
  A (M ),
α 0
V (0, M0 ) = Vz  A (M ) ⊕ Fv , for some α ∈ F. (4.13)

 0,0 0 0
z∈M0  A (M ),
0,0 0

In this case, j = 0, yj = 0. It is clear that, for any i ∈ K \ {0}, since ix1 ∈

/ M0
we have
V (ix1 , M0 ) = Aix1 ,bi (M0 ). (4.14)
For any z ∈ M0 \ Zx1 , rank(Zz + Zxi ) = 1 or 2. If M0 ⊂ Zx1 , then M1  Z;
Claim 2 is automatically true. Now suppose z0 ∈ M0 \ Zx1 is a fixed arbitrary
element. Then for all i ∈ K, [2, Theorem 2.1], Lemma 4.3 and (4.13) ensure that

 Bα (Zz0 + Zx1 ),

 A (Zz + Zx ),
α 0 1
V (ix1 , Zz0 + Zx1 ) = V (0, Zz0 + Zx1 )  A (Zz + Zx ) ⊕ Fv ,

 0 1 0
 A0,0
 (Zz + Zx ),
0,0 0 1

respectively. Thus
12 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19




Aix1 ,0 (Zz0 ),
 Aix ,1 (Zz0 ),
V (ix1, Zz0 )  ∀i ∈ K \ {0}
1
 Aix ,0 (Zz0 ), (4.15)

 A 1 (Zz ),
ix ,0 0
 1



Bα (Zz 0 ),
 A (Zz ),
α 0
V (0, Zz0 )  A (Zz0 ) ⊕ Fv0 , (4.16)


 A0,0
 (Zz ),
0,0 0

 A (Zx 1 ),
 z0 ,0
 Az0 ,1 (Zx1 ),
V (z0 , Zx1 )  A (Zz ), (4.17)

 z ,0 0
 A 0 (Zx ),
z0 ,0 1

respectively. From (4.14) and (4.15) we can choose bi (i ∈ K \ {0}) such that
bi = b = 0, 1, 0, or 0 (4.18)
corresponding to the four cases, respectively. Using (4.15)–(4.18) and (4.14), we
choose a basis vix1 +z ∈ Vix1 +z for all z ∈ M0 and all z ∈ M0 and all i ∈ K as
follows:

(N1 ) Choose vz0 +kx1 ∈ Vz0 +kx1 for all k ∈ Z such that
L v
 ix z +j x = (z0 + j x1 )vz0 +(i+j )x1 ,
 L 1 v 0 1 = (z + (i + j )x
ix1 z0 +j x1 0 1 )vz0 +(i+j )x1 ,
∀i, j ∈ Z,

 L ix 1 vz0 +j x 1 = (z0 + j x 1 )vz0 +(i+j )x1 ,
Lix1 vz0 +j x1 = (z0 + j x1 )vz0 +(i+j )x1 ,
respectively;
(N2 ) For any j ∈ K, choose vj x1 +z ∈ Vj x1 +z for all z ∈ M0 \ {z0 } such that, if
j ∈ K \ {0},
L v
 z z +j x +y = (z0 + j x1 + y)vz0 +j x1 +z+y ,
 L v 0 1 = (z + z + j x + y)v
z z0 +j x1 +y 0 1 z0 +j x1 +z+y ,
∀y, z ∈ M0 ,

 L z vz0 +j x 1 +y = (z0 + j x 1 + y)vz0 +j x1 +z+y ,
Lz vz0 +j x1 +y = (z0 + j x1 + y)vz0 +j x1 +z+y ,
respectively; if j = 0, ∀y, z ∈ M0 ,
 L v = yv for y + z = 0, Lz v−z = z(z + α)v0 ,
 z y z+y
 L v = (z + y)v
z y z+y for y = 0, Lz v0 = z(z + α)vz ,
 Lz vy = yvz+y
 for y(y + z) = 0,
Lz vy = yvz+y for y(y + z) = 0,
respectively.

Note that this choice of basis can always be done although we have sometimes
M0 ∩ Zx1 = 0.
 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19 13

For any z1 ∈ M0 , (4.13) yields (4.13) with M0 replaced into Zz1 . Applying
Lemma 4.3 gives (4.13) with M0 replaced into Zz1 + Zx1 . Then we can choose a
basis vz 0 +z ∈ Vz0 +z for all z ∈ M2 := Zz1 + Zx1 with vz 0 = vz0 such that, for all
y, z ∈ M2 ,


 Lz v  = (z0 + y)v  0 +z+y  = z(z + α)v  ,
if z0 + y + z = 0, Lz v−z
 L vz 0 +y = (z + z +zy)v   
0
z z0 +y 0 z0 +z+y if z0 + y = 0, Lz v0 = z(z + α)vz ,

 Lz v  = (z0 + y)v  ,
 L vz 0 +y = (z + y)vz 0 +z+y ,
z z0 +y 0 z0 +z+y
respectively. Combining this with (N1 ) and (N2 ), we deduce that vz 0 +z = vz0 +z
for all z ∈ Zz1 + Zx1 . Since z1 ∈ M1 is arbitrary, we can deduce (4.13) with M0
replaced by M1 . So Claim 2 is true for this subcase.

Claim 2 is contrary to the choice of M0 . Therefore we must have M0 = M.

This completes the proof of this theorem. ✷

5. Generalized super-Virasoro algebras

In this section, we shall first introduce the notion of the generalized super-
Virasoro algebras which generalizes the notion of the high-rank super-Virasoro
algebras introduced in [2], and then determine the modules of the intermediate
series over the generalized super-Virasoro algebras.
Roughly speaking, a generalized super-Virasoro algebra is a Lie superalgebra
which is a nontrivial Z/2Z-graded extension of a generalized Virasoro algebra.
Thus, suppose SVir[M] = SVir0̄ [M] ⊕ SVir1̄ [M] is a Z/2Z-graded extension
of a generalized Virasoro algebra Vir[M] such that SVir0̄ [M] = Vir[M] and
SVir1̄ [M] is a nontrivial irreducible Vir[M]-module of the intermediate series.
By Theorems 4.2 and 4.6, SVir1̄ [M] is a subquotient module of Aα,b (M) for
some α, b ∈ F, so by rewriting (4.2a), there is a subset M  of M, where M  = M
/ M, b = 0, 1 or M  = M \ {0} if α = 0, b = 0, 1, such that there exists a
if α ∈
basis {Gµ | µ ∈ α + M  } and
[Lµ , Gν ] = (ν + µb)Gµ+ν ,
∀µ ∈ M, ν ∈ α + M  . (5.1)
[c, Gν ] = 0,
Since we are considering a nontrivial extension, we have 0 = [Gµ , Gν ] ∈ Vir[M]
for some µ, ν ∈ M  . As it has the weight µ + ν, thus we have µ + ν ∈ M and so
2α ∈ M. (5.2)
In general, since [SVir1̄ [M], SVir1̄ [M]] ⊂ SVir0̄ [M] = Vir[M], we can write
[Gµ , Gν ] = xµ,ν Lµ+ν + δµ+ν,0 yµ c ∀µ, ν ∈ α + M  , (5.3)
for some xµ,ν , yµ ∈ F. Applying ad Lλ , λ ∈ M, to (5.3), using (5.1) and definition
(2.3), we obtain that
14 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19

(µ + λb)xµ+λ,ν + (ν + λb)xµ,ν+λ = (µ + ν − λ)xµ,ν , (5.4a)

δµ+λ+ν,0 (µ + λb)yµ+λ + δµ+ν+λ,0 (ν + λb)yµ
 3 
= 12
1
λ − λ δλ+µ+ν,0 xµ,ν , (5.4b)

holds for all λ ∈ M, µ, ν ∈ α + M  .

We shall consider (5.4) in two cases.

Case 1. Suppose [SVir1̄ [M], SVir1̄ [M]] ⊆ Fc, i.e., xµ,ν = 0 for all µ, ν ∈ α + M  .
By taking ν = µ ∈ α + M  , λ = −2µ ∈ M in (5.4b), we obtain
µ(1 − 2b)(y−µ + yµ ) = 0 ∀µ ∈ α + M  . (5.5)
First suppose b = 12 . Then (5.5) gives y−µ = −yµ . On the other hand, from the
definition of Lie superalgebras we have
y−µ = yµ ∀µ ∈ α + M  . (5.6)
This forces yµ = 0 ∀µ ∈ α + M  and so we obtain the trivial extension. Therefore
b = 12 and M  = M. Then in (5.4b), by taking λ = −(ν + µ) ∈ M, we obtain

2 (µ − ν)(y−ν − yµ ) = 0 ∀µ, ν ∈ α + M.
1
(5.7)
In particular, by setting ν = α, we have yµ = y−α for all α = µ ∈ α + M, and by
(5.6) we have yα = y−α . This shows that yµ = y ∈ F must be a nonzero scalar.
By rescaling the basis {Gµ | µ ∈ α + M} if necessary, we can suppose yµ = 1 for
all µ ∈ α + M; and so from (5.1) and (5.3), we have
 
[Lµ , Gν ] = ν + 12 µ Gµ+ν , [c, Gν ] = 0,
∀µ ∈ M, ν, λ ∈ α + M. (5.8)
[Gν , Gλ ] = δν+λ,0 c,
It is immediate to check that 
SVir[M, α] = span{Lµ , Gν , c | µ ∈ M, ν ∈
α + M} with commutation relations (2.3) and (5.8) defines a Lie superalgebra
for any subgroup M of F and α ∈ F such that 2α ∈ M.

Case 2. Suppose [SVir1̄ [M], SVir1̄ [M]]  Fc; i.e., xµ,ν = 0 for some µ, ν ∈
α + M .
Using the super-Jacobian identity,




Gλ , [Gµ , Gν ] = [Gλ , Gµ ], Gν − Gµ , [Gλ , Gν ] ∀µ, ν, λ ∈ α + M  ,
(5.9)
by applying Gλ to (5.3), and using (5.1), we have
     
− λ + (µ + ν)b xµ,ν = ν + (λ + µ)b xλ,µ + µ + (λ + ν)b xλ,ν
∀λ, µ, ν ∈ α + M  . (5.10)
 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19 15

Setting λ = ν = µ ∈ α + M  in (5.10), we obtain

3µ(1 + 2b)xµ,µ = 0 ∀µ ∈ α + M  . (5.11)
If b = − 12 , then
xµ,µ = 0 ∀0 = µ ∈ α + M  . (5.12)
By setting ν = µ in (5.4a), and then substituting µ + λ by ν, using (5.12) and
noting that xµ,ν = xν,µ , we obtain
 
2 µ(1 − b) + νb xµ,ν = 0 ∀µ, ν ∈ α + M  . (5.13)
Suppose xµ0 ,ν0 = 0 for some µ0 , ν0 ∈ α + M  ; then (5.13) and xµ0 ,ν0 = xν0 ,µ0
give µ0 (1 − b) + ν0 b = 0 = ν0 (1 − b) + µ0 b. From this we have µ0 = −ν0 ; thus
we obtain
xµ,ν = 0 ∀µ, ν ∈ α + M  , µ = −ν. (5.14)
But when µ = −ν, by taking λ = 0 in (5.4a), we again obtain xµ,−µ = 0. Thus
xµ,ν = 0 for all µ, ν ∈ α + M  , a contradiction. Therefore we obtain b = − 12 .
So, M  = M. Then in (5.10), by letting λ = µ, we obtain xµ,ν = xµ,µ for all
µ, ν ∈ α + M. Therefore we have xµ,ν = xν,ν = xα,ν = xα,α is a nonzero scalar.
By rescaling basis {Gµ | µ ∈ α + M} if necessary, we can suppose xµ,ν = 2 for
all µ, ν ∈ α + M.
By letting ν = µ ∈ α + M, λ = −2µ ∈ M in (5.4b), we obtain
 
µyµ = − 121
µ 4µ2 − 1 ∀µ ∈ α + M. (5.15)
Thus if α ∈
/ M, we obtain that
 
yµ = − 13 µ2 − 14 , (5.16)
holds for all µ ∈ α + M. If α ∈ M, then (5.16) holds for all µ = 0. To prove that
(5.16) also holds for µ = 0 when α ∈ M, after setting ν = 0, λ = −µ ∈ M in
(5.4b), we see this immediately. Thus (5.1) and (5.3) have the following forms:
 
[Lµ , Gν ] = ν − 12 µ Gµ+ν , [c, Gν ] = 0,
  ∀µ ∈ M, ν, λ ∈ α + M. (5.17)
[Gν , Gλ ] = 2Lν+λ − 13 ν 2 − 14 δν+λ,0 c,
It is immediate to check that SVir[M, α] = span{Lµ , Gν , c | µ ∈ M, ν ∈
α + M} with commutation relations (2.3) and (5.17) defines a Lie superalgebra
for any subgroup M of F and α ∈ F such that 2α ∈ M.

Thus we have in fact proved the next lemma.

Lemma 5.1. Suppose W = W0̄ ⊕W1̄ is a Lie superalgebra such that W0̄  Vir[M]
and W1̄ is an irreducible Vir[M]-module of the intermediate series, where M is
an additive subgroup of F. Then W is 
SVir[M, α] or SVir[M, α] for a suitable
α ∈ F with 2α ∈ M.

This result leads us the following definition.

16 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19

Definition 5.2. A generalized super-Virasoro algebra is a Lie superalgebra


SVir[M, α] defined by (2.3) and (5.8), or SVir[M, α] defined by (2.3) and (5.17),
where M is a subgroup of F, α ∈ F with 2α ∈ M.

Now we are in a position to consider modules of the intermediate series

over the generalized super-Virasoro algebras. Since SVir[M, α] is just a trivial
extension of Vir[M] modulo the center Fc, the modules of the intermediate series
over 
SVir[M, α] are simply those modules over the generalized Virasoro algebra
Vir[M]. Thus we shall be only interested in considering SVir[M, α].
First we give the precise definition.

Definition 5.3. A Z/2Z-graded vector space V = V0̄ ⊕ V1̄ is called a module of

the intermediate series over SVir[M, α] if it is an indecomposable SVir[M, α]-
module such that

SVirσ [M, α]Vτ ⊂ Vσ +τ ∀σ, τ ∈ Z/2Z,

dimF (Vσ )λ  1, where (Vα )λ = {v ∈ Vσ | L0 v = λv},
∀λ ∈ F, σ ∈ Z/2Z. (5.18)

As in [2,6], there exist three series of modules SAa,b (M, α), SAa (M, α),
SBa (M, α) for a, b ∈ F comparable with those of the generalized Virasoro
algebra defined in (4.2). Precisely, SAa,b (M, α), SAa (M, α) have basis {vµ | µ ∈
M} ∪ {wν | ν ∈ α + M} and SBa [M, α] has basis {vν | ν ∈ α + M} ∪ {wµ | µ ∈ M}
such that the central element c acts trivially and

SAa,b (M, α): Lλ vµ = (a + µ + λb)vλ+µ ,

  
Lλ wν = a + ν + λ b − 12 wλ+ν ,
  
Gη vµ = wη+µ , Gη wν = a + ν + 2η b − 12 vν+ν , (5.19a)
SAa (M, α): Lλ vµ = (µ + λ)vλ+µ , µ = 0, Lλ v0 = λ(λ + a)vλ ,
 
λ
Lλ wλ = ν + wλ+ν ,
2
Gη vµ = wη+µ , µ = 0, Gη w0 = (2η + a)w0,
Gη wν = (ν + η)vη+ν , (5.19b)
 
λ
SBa (M, α): Lλ vν = ν + vλ+µ , Lλ wµ = µwλ+µ , µ = −λ,
2
Lλ w−λ = −µ(µ + a)w0 ,
Gη vν = wη+ν , ν = −λ, Gη v−η = (2η + a)w0 ,
Gη wµ = µvη+µ , (5.19c)
for all λ, µ ∈ M, ν, η ∈ α + M.
 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19 17

Now we are ready to prove following theorem.

Theorem 5.4. A module of the intermediate series over SVir[M, α] is isomorphic

to one of the following: SAa,b (M, α), SAa (M, α), SBa (M, α) for a, b ∈ F, and
their nonzero subquotients.

Proof. Let V = V0̄ ⊕ V1̄ be a module of the intermediate series over

SVir[M, α].
Since V is indecomposable, there exists a ∈ F such that V0̄ = λ∈M (V0̄ )a+λ
and V1̄ = λ∈M (V1̄ )a+α+λ . We always take a = 0 if a ∈ M. Since V0̄ , V1̄ are
Vir[M]-modules such that dim(Vσ )λ  1, ∀σ ∈ Z/2Z, λ ∈ F, they have the form
Aa,b (M), Aa (M), Ba (M) or their subquotients or the direct sum of the two
composition factors of A0,0 (M). Since we can interchange V0̄ and V1̄ if necessary,
it suffices to consider the following cases.

Case 1. Suppose V0̄ = Fv0 .

Let wν = Gν v0 , ν ∈ α + M, then applying Lµ , Gη to Gν v0 , by (5.17), we have
 
µ
Lµ wν = Lµ Gν v0 = ν − wµ+ν ∀µ ∈ M, ν ∈ α + M,
2
Gη wν = Gη Gν v0 = −Gν wη ∀η, ν ∈ α + M. (5.20)
Since Gη wν = 0 for all η + ν = 0, we have
(ν − η)w2η+ν = L2η wν = G2η wν = 0. (5.21)
Thus wν = 0 for all ν ∈ α + M and V is a trivial module.

Case 2. Suppose V0̄ = A0,0 (M).

Then we can choose a basis {vµ | µ ∈ M  }, where M  = M \ {0} such that
Lλ vµ = (λ + µ)vλ+µ ∀λ ∈ M, µ ∈ M  . (5.22)
First fix 0 = µ0 ∈ M. For any ν ∈ α + M, then ν − µ0 ∈ α + M and we set
wν = Gν−µ0 vµ0 . If 2ν − µ0 = 0, then we have Gν−µ0 wν = L2(ν−µ0 ) vµ0 =
(2ν − µ0 )v2ν−µ0 = 0; and so wν = 0. Also if ν0 ∈ α + M such that 2ν0 − µ0 = 0,
then Gν0 w2ν = G2ν0 vµ0 = L2ν0 vµ0 = 3µ0 v2µ0 = 0. This shows that wν = 0 for
0
all ν ∈ α + M. Therefore we can choose a basis {wν | ν ∈ α + M} of V1̄ such that
there exists b ∈ F and
Lµ wν = (ν + bµ)wµ+ν (5.23)
holds for µ ∈ M, ν ∈ α + M, ν = 0, µ + ν = 0. For any 0 = ν ∈ α + M, applying
L2ν to w−ν , we have ν(2b − 1)wν = L2ν w−ν = Gν · Gν w−ν ∈ Gν (V0̄ )0 = {0}.
Thus b = 12 and so (5.23) holds for all µ ∈ M, ν ∈ α + M. Now suppose
Gν vµ = aν,µ wν+µ , Gν wη = bν,η xν+η ,
0 = µ ∈ M, ν, η ∈ α + M, (5.24)
18 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19

for some aν,µ , bν,η ∈ F. We can suppose aµ,ν = 1 for some µ, ν. Applying
Gν , L2ν to (5.24), by (5.22) and (5.23) we obtain
aν,µ bν,ν+µ = µ + 2ν, bν,η aν,ν+η = η + ν,
(2ν + µ)aν,µ = (2ν + µ)aν,µ+2ν ,
(3ν + η)bν,η = (η + ν)bν,η+2ν . (5.25)
Now one can proceed exactly as in [6, Section 2] to prove that aν,µ = 1 and
bν,η = η + ν; so that V is the quotient module of SA0 (M).

Case 3. Suppose V0̄ = Aa (M), Ba (M), A0,0 (M) or direct sum of two submodules
A0,0 (M) ⊕ Fv0 .

Similar to the proof of Cases 2, it is not difficult to show V = SAa (M),

SBa (M), SA0,0 (M) or direct sum of two submodules SA0,0 (M) ⊕ F0 (this cannot
occur since V is indecomposable).

Case 4. Suppose both V0̄ and V1̄ are simple Vir[M]-module of type Aa,b (M).
As in Case 2, we can deduce similar equation as (5.25) and the proof is exactly
analogous to that of [2, Theorem 3.1]. We obtain that V is of type SAa,b (M). ✷

References

[1] J. Patera, H. Zassenhaus, The higher rank Virasoro algebras, Comm. Math. Phys. 136 (1991)
1–14.
[2] Y.C. Su, Harish–Chandra modules of the intermediate series over the high rank Virasoro algebras
and high rank super-Virasoro algebras, J. Math. Phys. 35 (1994) 2013–2023.
[3] D.Z. Djokovic, K. Zhao, Derivations, isomorphisms, and second cohomology of generalized Witt
algebras, Trans. Amer. Math. Soc. 350 (2) (1998) 643–664.
[4] V.G. Kac, V. Raina, Bombay Lectures on Highest Weight Representations of Infinite Dimensional
Lie Algebras, World Scientific, Singapore, 1987.
[5] J.M. Osborn, K. Zhao, Doubly Z-graded Lie algebras containing a Virasoro algebra, J. Alge-
bra 219 (1999) 266–298.
[6] Y.C. Su, Classification of Harish–Chandra modules over the super-Virasoro algebras, Comm.
Algebra 23 (1995) 3653–3675.
[7] D.Z. Djokovic, K. Zhao, Some simple subalgebras of generalized Block algebras, J. Algebra 192
(1997) 74–101.
[8] D.Z. Djokovic, K. Zhao, Some infinite-dimensional simple Lie algebras in characteristic 0 related
to those of Block, J. Pure Appl. Algebra 127 (2) (1998) 153–165.
[9] D.Z. Djokovic, K. Zhao, Generalized Cartan type W Lie algebras in characteristic zero,
J. Algebra 195 (1997) 170–210.
[10] D.Z. Djokovic, K. Zhao, Generalized Cartan type S Lie algebras in characteristic zero,
J. Algebra 193 (1997) 144–179.
[11] D.Z. Djokovic, K. Zhao, Second cohomology of generalized Cartan type S Lie algebras in
characteristic zero, J. Pure Appl. Algebra 136 (1999) 101–126.
[12] J.M. Osborn, K. Zhao, Infinite dimensional Lie algebras of type L, Comm. Algebra, accepted.
 Y. Su, K. Zhao / Journal of Algebra 252 (2002) 1–19 19

[13] J.M. Osborn, K. Zhao, Infinite dimensional Lie algebras of generalized Block type, Proc. Amer.
Math. Soc. 127 (6) (1999) 1641–1650.
[14] J.M. Osborn, K. Zhao, Generalized Poisson bracket and Lie algebras of type H in characteristic 0,
Math. Z. 230 (1999) 107–143.
[15] J.M. Osborn, K. Zhao, Generalized Cartan type K Lie algebras in characteristic 0, Comm.
Algebra 25 (1997) 3325–3360.
[16] D.P. Passman, Simple Lie algebras of Witt type, J. Algebra 206 (1998) 682–692.
[17] X. Xu, New generalized simple Lie algebras of Cartan type over a field with characteristic 0,
J. Algebra 224 (2000) 23–58.