Induced Modules for Affine Lie Algebras
We study induced modules of nonzero central charge with arbitrary multiplicities over affine Lie algebras. For a given pseudo parabolic subalgebra P of an affine Lie algebra G, our main result establishes the equivalence between a certain category of P-induced G-modules and the category of weight P-...
Збережено в:
| Дата: | 2009 |
|---|---|
| Автори: | , |
| Формат: | Стаття |
| Мова: | English |
| Опубліковано: |
Інститут математики НАН України
2009
|
| Назва видання: | Symmetry, Integrability and Geometry: Methods and Applications |
| Онлайн доступ: | http://dspace.nbuv.gov.ua/handle/123456789/149179 |
| Теги: |
Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
|
| Назва журналу: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Цитувати: | Induced Modules for Affine Lie Algebras / V. Futorny, I. Kashuba // Symmetry, Integrability and Geometry: Methods and Applications. — 2009. — Т. 5. — Бібліогр.: 22 назв. — англ. |
Репозитарії
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
irk-123456789-149179 |
|---|---|
| record_format |
dspace |
| spelling |
irk-123456789-1491792019-02-20T01:27:22Z Induced Modules for Affine Lie Algebras Futorny, V. Kashuba, I. We study induced modules of nonzero central charge with arbitrary multiplicities over affine Lie algebras. For a given pseudo parabolic subalgebra P of an affine Lie algebra G, our main result establishes the equivalence between a certain category of P-induced G-modules and the category of weight P-modules with injective action of the central element of G. In particular, the induction functor preserves irreducible modules. If P is a parabolic subalgebra with a finite-dimensional Levi factor then it defines a unique pseudo parabolic subalgebra Pps, P Ì Pps. The structure of P-induced modules in this case is fully determined by the structure of Pps-induced modules. These results generalize similar reductions in particular cases previously considered by V. Futorny, S. König, V. Mazorchuk [Forum Math. 13 (2001), 641-661], B. Cox [Pacific J. Math. 165 (1994), 269-294] and I. Dimitrov, V. Futorny, I. Penkov [Comm. Math. Phys. 250 (2004), 47-63]. 2009 Article Induced Modules for Affine Lie Algebras / V. Futorny, I. Kashuba // Symmetry, Integrability and Geometry: Methods and Applications. — 2009. — Т. 5. — Бібліогр.: 22 назв. — англ. 1815-0659 2000 Mathematics Subject Classification: 17B65; 17B67 http://dspace.nbuv.gov.ua/handle/123456789/149179 en Symmetry, Integrability and Geometry: Methods and Applications Інститут математики НАН України |
| institution |
Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| collection |
DSpace DC |
| language |
English |
| description |
We study induced modules of nonzero central charge with arbitrary multiplicities over affine Lie algebras. For a given pseudo parabolic subalgebra P of an affine Lie algebra G, our main result establishes the equivalence between a certain category of P-induced G-modules and the category of weight P-modules with injective action of the central element of G. In particular, the induction functor preserves irreducible modules. If P is a parabolic subalgebra with a finite-dimensional Levi factor then it defines a unique pseudo parabolic subalgebra Pps, P Ì Pps. The structure of P-induced modules in this case is fully determined by the structure of Pps-induced modules. These results generalize similar reductions in particular cases previously considered by V. Futorny, S. König, V. Mazorchuk [Forum Math. 13 (2001), 641-661], B. Cox [Pacific J. Math. 165 (1994), 269-294] and I. Dimitrov, V. Futorny, I. Penkov [Comm. Math. Phys. 250 (2004), 47-63]. |
| format |
Article |
| author |
Futorny, V. Kashuba, I. |
| spellingShingle |
Futorny, V. Kashuba, I. Induced Modules for Affine Lie Algebras Symmetry, Integrability and Geometry: Methods and Applications |
| author_facet |
Futorny, V. Kashuba, I. |
| author_sort |
Futorny, V. |
| title |
Induced Modules for Affine Lie Algebras |
| title_short |
Induced Modules for Affine Lie Algebras |
| title_full |
Induced Modules for Affine Lie Algebras |
| title_fullStr |
Induced Modules for Affine Lie Algebras |
| title_full_unstemmed |
Induced Modules for Affine Lie Algebras |
| title_sort |
induced modules for affine lie algebras |
| publisher |
Інститут математики НАН України |
| publishDate |
2009 |
| url |
http://dspace.nbuv.gov.ua/handle/123456789/149179 |
| citation_txt |
Induced Modules for Affine Lie Algebras / V. Futorny, I. Kashuba // Symmetry, Integrability and Geometry: Methods and Applications. — 2009. — Т. 5. — Бібліогр.: 22 назв. — англ. |
| series |
Symmetry, Integrability and Geometry: Methods and Applications |
| work_keys_str_mv |
AT futornyv inducedmodulesforaffineliealgebras AT kashubai inducedmodulesforaffineliealgebras |
| first_indexed |
2025-07-12T21:35:28Z |
| last_indexed |
2025-07-12T21:35:28Z |
| _version_ |
1837478589958717440 |
| fulltext |
Symmetry, Integrability and Geometry: Methods and Applications SIGMA 5 (2009), 026, 14 pages
Induced Modules for Affine Lie Algebras?
Vyacheslav FUTORNY and Iryna KASHUBA
Institute of Mathematics, University of São Paulo,
Caixa Postal 66281 CEP 05314-970, São Paulo, Brazil
E-mail: futorny@ime.usp.br, kashuba@ime.usp.br
Received October 20, 2008, in final form March 01, 2009; Published online March 04, 2009
doi:10.3842/SIGMA.2009.026
Abstract. We study induced modules of nonzero central charge with arbitrary multiplicities
over affine Lie algebras. For a given pseudo parabolic subalgebra P of an affine Lie algeb-
ra G, our main result establishes the equivalence between a certain category of P-induced
G-modules and the category of weight P-modules with injective action of the central element
of G. In particular, the induction functor preserves irreducible modules. If P is a parabolic
subalgebra with a finite-dimensional Levi factor then it defines a unique pseudo parabolic
subalgebra Pps, P ⊂ Pps. The structure of P-induced modules in this case is fully deter-
mined by the structure of Pps-induced modules. These results generalize similar reductions
in particular cases previously considered by V. Futorny, S. König, V. Mazorchuk [Forum
Math. 13 (2001), 641–661], B. Cox [Pacific J. Math. 165 (1994), 269–294] and I. Dimitrov,
V. Futorny, I. Penkov [Comm. Math. Phys. 250 (2004), 47–63].
Key words: affine Kac–Moody algebras; induced modules; parabolic subalgebras; Borel
subalgebras
2000 Mathematics Subject Classification: 17B65; 17B67
1 Introduction
It is difficult to over-estimate the importance of Kac–Moody algebras for modern mathe-
matics and physics. These algebras were introduced in 1968 by V. Kac and R. Moody as
a generalization of simple finite-dimensional Lie algebras, by relaxing the condition of Cartan
matrix to be positive definite. We address to [20] for the basics of the Kac–Moody theory.
Affine Lie algebras are the most studied among infinite-dimensional Kac–Moody algebras,
and have very wide applications. They correspond to the case of positive semidefinite matrix
(det(aij) = 0, with positive principal minors).
All results in the paper hold for both untwisted and twisted affine Lie algebras of rank
greater than 1. Let G be an affine Kac–Moody algebra with a 1-dimensional center Z = Cc.
A natural way to construct representations of affine Lie algebras is via induction from
parabolic subalgebras. Induced modules play an important role in the classification problem
of irreducible modules. For example, in the finite-dimensional setting any irreducible weight
module is a quotient of the module induced from an irreducible module over a parabolic
subalgebra, and this module is dense (that is, it has the largest possible set of weights) as
a module over the Levi subalgebra of the parabolic [9, 10, 7]. In particular, dense irreducible
module is always torsion free if all weight spaces are finite-dimensional. In the affine case,
?This paper is a contribution to the Special Issue on Kac–Moody Algebras and Applications. The full
collection is available at http://www.emis.de/journals/SIGMA/Kac-Moody algebras.html
mailto:futorny@ime.usp.br
mailto:kashuba@ime.usp.br
http://dx.doi.org/10.3842/SIGMA.2009.026
http://www.emis.de/journals/SIGMA/Kac-Moody_algebras.html
2 V. Futorny and I. Kashuba
a similar conjecture [11, Conjecture 8.1] singles out induced modules as construction devices
for irreducible weight modules. This conjecture has been shown for A
(1)
1 [12, Proposition 6.3]
and for all affine Lie algebras in the case of modules with finite-dimensional weight spaces
and nonzero central charge [18]. In the latter case, a phenomenon of reduction to modules
over a proper subalgebra (finite-dimensional reductive) provides a classification of irreducible
modules. Moreover, recent results of I. Dimitrov and D. Grantcharov [6] show the validity of
the conjecture also for modules with finite-dimensional weight spaces and zero central charge.
For highest weight modules (with respect to nonstandard Borel subalgebras) with nonzero
central charge such reduction was shown in [4, 17]. The case of induced modules from
a parabolic subalgebra with a finite-dimensional Levi factor was considered in [15]. In par-
ticular, it was shown that such categories of modules are related to projectively stratified
algebras [16]. A more general setting of toroidal Lie algebras was considered in [5] for in-
duced modules from general Borel subalgebras.
The main purpose of the present paper is to show, that in the affine setting all known
cases of the reduction are just particular cases of a general reduction phenomenon for modules
with nonzero central charge.
We assign to each parabolic subset P of the root system ∆ of G, the parabolic subalgebra
GP = G0
P ⊕ G+
P of G with the Levi subalgebra G0
P (cf. Section 3.1). The subalgebra G0
P is
infinite-dimensional if and only if P ∩ −P contains imaginary roots. If, in the same time,
P ∩−P contains some real roots then we define our key subalgebra G
ps
P ⊂ GP , which will be
called pseudo parabolic subalgebra.
Denote by U(G) the universal enveloping algebra of G.
Let P = P0 ⊕ N be a (pseudo) parabolic subalgebra of G with the Levi subalgebra P0.
If N is a weight P0-module then it can be viewed as a P-module with a trivial action of N.
Then one can construct the induced G-module ind(P,G;N) = U(G)⊗P N . Hence
ind(P,G) : N 7−→ ind(P,G;N)
defines a functor from the category of weight P0-modules to the category of weight G-modules.
Denote by W (P0) the full subcategory of weight P0-modules on which the central element
c acts injectively. Our main result is the following reduction theorem.
Theorem 1.1. Let G be affine Lie algebra of rank greater than 1, P = P0 ⊕ N a pseudo
parabolic subalgebra of G, P0 infinite-dimensional and ind0(P,G) the restriction of the induc-
tion functor ind(P,G) onto W (P0). Then the functor ind(P,G) preserves the irreducibles.
This result allows to construct new irreducible modules for affine algebras using parabolic
induction from affine subalgebras.
Theorem 1.1 follows from a more general result (see Theorems 3.1 and 4.1) which estab-
lishes an equivalence of certain categories of modules.
In the case when the Levi factor G0
P is finite-dimensional, we define a certain subalgebra
mP ⊂ G, which leads to a parabolic decomposition G = N−
P ⊕mP ⊕N+
P with N+
P ⊂ G+
P and
G0
P ⊂ mP (cf. Section 2). If mP is finite-dimensional then G0
P = mP and G+
P = N+
P . In this
case there is no reduction. If mP is infinite-dimensional then mP ⊕ N+
P is pseudo parabolic
subalgebra and the reduction theorem above implies Theorem 8 in [15].
The structure of the paper is as follows. In Section 2 we recall the classification of Borel
subalgebras and parabolic subalgebras. Section 3 is devoted to the study of parabolic and
pseudo parabolic induction. In particular, we prove Theorem 3.1 that describes the structure
Induced Modules for Affine Lie Algebras 3
of induced modules. In the last section we introduce certain categories of G-modules and of
P-modules and establish their equivalence (Theorem 4.1).
2 Preliminaries on Borel subalgebras
and parabolic subalgebras
Let H be a Cartan subalgebra of G with the following decomposition
G = H⊕ (⊕α∈H∗\{0}Gα),
where Gα := {x ∈ G | [h, x] = α(h)x for every h ∈ H}. Denote by ∆ = {α ∈ H∗\{0} |Gα 6= 0}
the root system of G. Let π be a basis of the root system ∆. For a subset S ⊂ π denote
by ∆+(S) the subset of ∆, consisting of all linear combinations of elements of S with nonnega-
tive coefficients. Thus, ∆+(π) is the set of positive roots with respect to π. Let δ ∈ ∆+(π) be
the indivisible imaginary root. Then the set of all imaginary roots is ∆im = {kδ|k ∈ Z\{0}}.
Denote by G a Heisenberg subalgebra of G generated by the root spaces Gkδ, k ∈ Z \ {0}.
Denote by Sπ a root subsystem generated by S and δ. Let S+
π = Sπ ∩∆+(π). For a subset
T ⊂ ∆ denote by G(T ) the subalgebra of G generated by the root subspaces Gα, α ∈ T , and
let H(T ) = H ∩G(T ). The subalgebra G(−T ) will be called the opposite subalgebra to G(T ).
Let S ⊂ π, S = ∪iSi where all Si’s are connected and Si ∩ Sj = ∅ if i 6= j.
Proposition 2.1 ([15, Proposition 2]). G(Sπ) = GS + G(S) + H, where GS =
∑
i G
i,
[Gi,Gj ] = 0, i 6= j, Gi is the derived algebra of an affine Lie algebra of rank |Si| + 1,
[GS , G(S)] = 0, G(S) ⊂ G, G(S) + (G ∩GS) = G, GS ∩G(S) = ∩iG
i = Z.
Let V be a weight G-module, that is V = ⊕µ∈H∗Vµ, Vµ = {v ∈ V |hv = µ(h)v,∀h ∈ H}. If
V is irreducible then c acts as a scalar on V , it is called the central charge of V . A classification
of irreducible modules in the category of all weight modules is still an open question even in
the finite-dimensional case. In the affine case, a classification is only known in the subcat-
egory of modules with finite-dimensional weight spaces ([18] for nonzero charge and [6] for
zero charge), and in certain subcategories of induced modules with some infinite-dimensional
weight spaces [11, 15]. If V is a weight module (with respect to a fixed Cartan subalgebra)
then we denote by w(V ) the set of weight, that is w(V ) = {λ ∈ H∗|Vλ 6= 0}.
The affine Lie algebra G has the associated simple finite-dimensional Lie algebra g (for
details see [20]). Of course, in the untwisted case G is just the affinization of g:
G = g⊗ C
[
t, t−1
]
⊕ Cc⊕ Cd,
where d is the degree derivation: d(x⊗ tn) = n(x⊗ tn), d(c) = 0, for all x ∈ G, n ∈ Z.
A closed subset P ⊂ ∆ is called a partition if P ∩(−P ) = ∅ and P ∪(−P ) = ∆. In the case
of finite-dimensional simple Lie algebras, every partition corresponds to a choice of positive
roots in ∆, and all partitions are conjugate by the Weyl group. The situation is different in
the infinite-dimensional case. In the case of affine Lie algebras the partitions are divided into
a finite number of Weyl group orbits (cf. [19, 11]).
Given a partition P of ∆, we define a Borel subalgebra BP ⊂ G generated by H and the
root spaces Gα with α ∈ P . Hence, in the affine case not all of the Borel subalgebras are
conjugate but there exists a finite number of conjugacy classes.
4 V. Futorny and I. Kashuba
A parabolic subalgebra of G corresponds to a parabolic subset P ⊂ ∆, which is a closed
subset in ∆ such that P ∪ (−P ) = ∆. Given such a parabolic subset P , the corresponding
parabolic subalgebra GP of G is generated by H and all the root spaces Gα, α ∈ P .
The conjugacy classes of Borel subalgebras of G are parameterized by the parabolic subal-
gebras of the associated finite-dimensional simple Lie algebra g. We just recall the construc-
tion in the untwisted case. Parabolic subalgebras of g are defined as above, they correspond
to parabolic subsets of the roots system of g. Let p = p0 ⊕ p+ be a parabolic subalgebra of g
containing a fixed Borel subalgebra b of g. Define
B(p) = p+ ⊗ C[t, t−1]⊕ p0 ⊗ tC[t]⊕ b⊕ Cc⊕ Cd.
For any Borel subalgebra B of G there exists a parabolic subalgebra p of g such that B is
conjugate to B(p) [19, 11].
Any Borel subalgebra conjugated to B(g) is called standard. It is determined by a choice of
positive roots in G. Another extreme case p0 = H corresponds to the natural Borel subalgebra
of G.
We will also use the geometric description of Borel subalgebras in G following [8]. Let
W = SpanR ∆, n = dim W . Let
F = {{0} = Fn ⊂ Fn−1 ⊂ · · · ⊂ F1 ⊂ F0 = W}
be a flag of maximal length in W . The flag F of maximal length is called oriented if, for
each i, one of the connected components of Fi\Fi+1 is labelled by + and the other one is
labelled by −. Such an oriented flag F determines the partition
P := ∆ ∩
(
∪i (Fi\Fi+1)+
)
of ∆. Denote P 0
i = ∆ ∩ Fi. The subsets P 0
i are important invariants of the partition P .
Next statement follows immediately from the description of partitions of root systems [13].
Proposition 2.2. Let P ⊂ ∆ be a partition. There exists an oriented flag of maximal length
F = {{0} = Fn ⊂ Fn−1 ⊂ · · · ⊂ F1 ⊂ F0 = W} which determines the partition P .
Given a partition P of ∆ we will denote by F (P ) the corresponding oriented flag of
maximal length.
Let λ : BP → C be a 1-dimensional representation of BP . Then one defines an induced
Verma type G-module
MP (λ) = U(G)⊗U(BP ) C.
The module M∆+(π)(λ) is a classical Verma module with the highest weight λ [20]. In the
case of natural Borel subalgebra we obtain imaginary Verma modules studied in [14].
Let BP = H⊕NP , where NP is generated by Gα, α ∈ P . Note that the module MP (λ) is
U(N−
P )-free, where N−
P is the opposite subalgebra to NP . The theory of Verma type modules
was developed in [11]. It follows immediately from the definition that Verma type module
with highest weight λ has a unique maximal submodule. Also, unless it is a classical Verma
module, it has both finite and infinite-dimensional weight spaces and it can be obtained
using the parabolic induction from a classical Verma module M with highest weight λ over
a certain infinite-dimensional Lie subalgebra. Moreover, if the central charge of such Verma
Induced Modules for Affine Lie Algebras 5
type module is nonzero, then the structure of this module is completely determined by the
structure of module M , which is well-known [11, 4, 17].
Let P ⊂ ∆ be a parabolic subset, P ∪ (−P ) = ∆, GP the corresponding parabolic subal-
gebra of G. Set P 0 = P ∩ −P . Then
G = G−
P ⊕G0
P ⊕G+
P ,
where G±
P =
∑
α∈P\(−P ) G±α and G0
P is generated by H and the subspaces Gα with α ∈ P 0.
The subalgebra G0
P is the Levi factor of G. A classification of parabolic subsets of ∆ and
parabolic subalgebras of G was obtained in [11, 13].
Every parabolic subalgebra P ⊂ G containing a Borel subalgebra B has a Levi decom-
position P = P0 ⊕ N, with the Levi factor P0 and N ⊂ B. Following [11] we say that P
has type I if P0 is finite-dimensional reductive Lie algebra, and P has type II if P0 contains
the Heisenberg subalgebra G, generated by the imaginary root spaces of G. In the latter
case, P0 is an extension of a sum of some affine Lie subalgebras by a central subalgebra and
by a certain subalgebra of G [10]. Note that the radical N is solvable only for the type II
parabolic subalgebras. Type I parabolic subalgebras are divided also into two essentially
different types depending on whether N belongs to some standard Borel subalgebra (type Ia)
or not (type Ib).
It is easy to see that there are parabolic subalgebras P which do not correspond to any
triangular decomposition of G [11]. In fact, if P contains Gα+kδ for some α and infinitely
many both positive and negative integers k then this parabolic subalgebra does not correspond
to any triangular decomposition of G. In particular, this is always the case for parabolic
subalgebras of type II.
If P = P0 ⊕N is a parabolic subalgebra of type II then as soon as N contains Gα+kδ for
some real root α and some k ∈ Z, it also contains Gα+rδ for all r ∈ Z.
Geometrically parabolic subsets correspond to partial oriented flags of maximal length.
Let P be a parabolic subset which contains a partition P̃ and F (P̃ ) the corresponding full
oriented flag: F (P̃ ) = {{0} = Fn ⊂ Fn−1 ⊂ · · · ⊂ F1 ⊂ F0}. If P 0 ⊂ Fk for some k and P 0 is
not in Fk+1 then P is completely determined by the partial oriented flag
F (P ) = {{0} = Fk ⊂ · · · ⊂ F1 ⊂ F0 = W}.
Here P 0 = Fk ∩ ∆, P0 = G0
P . The corresponding parabolic subalgebra P has type II if
∆Im ⊂ Fk and it has type Ib if ∆Im ⊂ Fs for some s, 1 ≤ s < k.
Let P = P0⊕N be a non-solvable parabolic subalgebra of type II, P0 = [P0,P0]⊕G(P)+H,
where G(P) ⊂ G is the orthogonal completion (with respect to the Killing form) of the
Heisenberg subalgebra of [P0,P0], that is G(P)+([P0,P0]∩G) = G and G(P)∩ [P0,P0] = Cc.
Note that by Proposition 2.1, [P0,P0] is a sum of affine subalgebras of G. Let
G(P) = G(P)− ⊕ Cc⊕G(P)+
be a triangular decomposition of G(P). Define a pseudo parabolic subalgebra Pps = Pps
0 ⊕Nps,
where Pps
0 is generated by the root spaces Gα, α ∈ P ∩−P ∩∆re and H, while Nps is generated
by the root spaces Gα, α ∈ P \ (−P ) and G(P)+. Then Pps is a proper subalgebra of P.
Suppose P = P0 ⊕N is a parabolic subalgebra of type Ib, P the corresponding parabolic
subset and F (P ) the partial flag. Then P0 is a finite-dimensional reductive Lie algebra.
Assume Gδ ⊂ N and ∆Im ⊂ Fs, with the largest such s, 1 ≤ s < k. Note that for any α ∈ P 0,
6 V. Futorny and I. Kashuba
P \ P 0 contains the roots of the form α + kδ and −α + kδ for all k > 0. Denote by mP
a subalgebra of G generated by Fs ∩ ∆ and H. This is an infinite-dimensional Lie algebra
which contains P0 and m̃P = mP ∩ P is a parabolic subalgebra of mP of type Ia.
Let NP be the span of all root subspaces Gβ , β ∈ P which are not in mP , NP ⊂ N. Then
P = m̃P ⊕NP . It follows immediately that
Proposition 2.3 ([15, Proposition 3]). m̃P = mP ⊕NP is a parabolic subalgebra of G of
type II and G = N−
P ⊕mP ⊕NP , where N−
P is the opposite algebra to NP .
Hence, any parabolic subalgebra of type Ib can be extended canonically to the parabolic
subalgebra of type II. Moreover, it can be extended canonically to the pseudo parabolic
subalgebra m̃
ps
P :
P ⊂ m̃
ps
P ⊂ m̃P .
3 Parabolic induction
Let P be a parabolic subset of ∆. Let N be a weight (with respect to H) module over the
parabolic subalgebra P = GP (respectively pseudo parabolic subalgebra Pps), with a trivial
action of G+
P (respectively (Gps
P )+), and let
MP (N) = ind(GP ,G;N), Mps
P (N) = ind(Gps
P ,G;N)
be the induced G-modules. If N is irreducible then MP (N) (respectively Mps
P (N)) has
a unique irreducible quotient LP (N) (respectively Lps
P (N)). If N is irreducible P-module
such that G(P)+ acts trivially on N then MP (N) ' Mps
P (N) and LP (N) ' Lps
P (N).
If G0
P 6= G then LP (N) is said to be parabolically induced. Following [7] we will say that
irreducible G-module V is cuspidal if it is not of type LP (N) for any proper parabolic subset
P ⊂ ∆ and any N .
We see right away that a classification of irreducible G-modules reduces to the classification
of all irreducible cuspidal modules over Levi subalgebras of G. Namely we have
Proposition 3.1. Let V be an irreducible weight G-module. Then there exists a parabolic
subalgebra P = P0⊕N of G (possibly equal G) and an irreducible weight cuspidal P0-module N
such that V ' LP (N).
Example 3.1.
• Let V be an irreducible weight cuspidal g-module then V ⊗ C[t, t−1] is an irreducible
cuspidal G-module with zero central charge.
• Modules obtained by the parabolic induction from cuspidal modules over the Heisenberg
subalgebra are called loop modules [3].
• Pointed (that is, all weight spaces are 1-dimensional) cuspidal modules were studied
in [22].
A Levi subalgebra of G is cuspidal if it admits a weight cuspidal module. All cuspidal Levi
subalgebras of type Ia and Ib parabolics were classified in [9]. They are the subalgebras with
simple components of type A and C. All cuspidal Levi factors of type II parabolic subalgebras
were described in [13, 11]. For any affine Lie algebra the simplest Levi subalgebra of type II
is a Heisenberg subalgebra. Below we provide a list of all other Levi subalgebras of type II
with the connected root system.
Induced Modules for Affine Lie Algebras 7
G P0
A
(1)
n A
(1)
k , 1 ≤ k ≤ n− 1
B
(1)
n A
(1)
k , 1 ≤ k ≤ n− 1, C
(1)
2 , B
(1)
k , 3 ≤ k ≤ n− 1
C
(1)
n A
(1)
k , 1 ≤ k ≤ n− 1, C
(1)
k , 2 ≤ k ≤ n− 1
D
(1)
n A
(1)
k , 1 ≤ k ≤ n− 1, D
(1)
k , 4 ≤ k ≤ n− 1
G
(1)
2 , D
(3)
4 A
(1)
1
F
(1)
4 A
(1)
1 , A
(1)
2 , C
(1)
2 , C
(1)
3 , B
(1)
3
E
(1)
l , l = 6, 7, 8 A
(1)
k , 1 ≤ k ≤ l − 1, D
(1)
k , 4 ≤ k ≤ l − 1, E
(1)
k , 6 ≤ k ≤ l − 1
A
(2)
2n A
(1)
k , 1 ≤ k ≤ n− 1, A
(2)
2k , 1 ≤ k ≤ n− 1, E
(1)
k , 6 ≤ k ≤ l − 1
D
(2)
n A
(1)
k , 1 ≤ k ≤ n− 2, D
(2)
k , 3 ≤ k ≤ n− 1, E
(1)
k , 6 ≤ k ≤ l − 1
A
(2)
2n−1 A
(1)
k , 1 ≤ k ≤ n− 2, A
(2)
2k−1, 3 ≤ k ≤ n− 1, D
(2)
3
E
(2)
6 A
(1)
1 , A
(1)
2 , D
(2)
3 , D
(2)
4 , D
(2)
5
A nonzero element v of a G-module V is called P-primitive if G+
P v = 0. Let QP be the
free Abelian group generated by P 0. The following statement is standard.
Proposition 3.2. Let V be an irreducible weight G-module with a P-primitive element of
weight λ, P 0 6= ∆, N =
∑
ν∈QP
Vλ+ν . Then N is an irreducible GP -module and V is
isomorphic to LP (N).
If V is generated by a B-primitive element v ∈ Vλ then V is a highest weight module with
highest weight λ. If P is of type Ia then MP (N) is a generalized Verma module [15, Section 2].
A classification of all irreducible N with finite-dimensional weight spaces (and hence of LP (N)
if P is of type I) is known due to Proposition 3.2, [21] and [9]. Also a classification is known
when P is of type II, N has finite-dimensional weight spaces and a nonzero charge [18]. In
this case, N is the irreducible quotient of ind(P0,P ′;N ′), where P ′ = P ′
0 ⊕N′ is a parabolic
subalgebra of P0 of type Ia and N′ is an irreducible P ′-module with a trivial action of N′.
3.1 Reduction theorem for type II
Let π be a basis of the root system ∆, α0 ∈ π such that −α + δ ∈
∑
β∈π\{α0} Zβ and either
−α + δ ∈ ∆ or 1/2(−α + δ) ∈ ∆. Let π̇ = π \ {α0} and ∆̇+ the free semigroup generated
by π̇. Choose a proper subset S ∈ π̇ and the root subsystem Sπ generated by S and δ. Set
P+ = {α + nδ|α ∈ ∆̇+ \ Sπ, n ∈ Z} ∩∆.
Then PS = Sπ ∪P+ is a parabolic subset with Sπ = PS ∩−PS . Let GPS
be the corresponding
parabolic subalgebra. Then it is of type II with
G0
PS
= G(Sπ) =
∑
α∈Sπ
Gα ⊕ H.
Proposition 3.3 ([11]). If P is a parabolic subalgebra of G of type II then there exist π, α0
and S as above such that P is conjugate to GPS
.
8 V. Futorny and I. Kashuba
Hence, it suffices to consider the parabolic subalgebras of type II in the form GPS
.
Let Sπ = ∪iSi be the decomposition of Sπ into connected components and let G0
PS
=∑
i Gi ⊕G(PS) be the corresponding decomposition of G0
PS
(see Proposition 2.1).
Theorem 3.1. Let G be of rank > 1, P a parabolic subset of ∆ such that P ∩−P contains real
and imaginary roots simultaneously. Consider a weight G
ps
P -module V which is annihilated
by (Gps
P )+ and on which the central element c acts injectively (object of W (Gps
P )). Then for
any submodule U of ind(Gps
P ,G;V ) there exists a submodule VU of V such that
U ' ind(Gps
P ,G;VU ).
In particular, ind(Gps
P ,G;V ) is irreducible if and only V is irreducible.
Proof. The proof follows general lines of the proof of Lemma 5.4 in [11]. Denote Mps(V ) =
ind(Gps
P ,G;V ) and M̂ps(V ) =
∑
ν∈QP ,λ∈w(V ) Mps(V )λ+ν = 1 ⊗ V. Then M̂ps(V ) is a GP -
submodule of Mps(V ) isomorphic to V , which consists of P-primitive elements.
Let G
ps
P = P0 ⊕N, N− is the opposite subalgebra to N.
Let U be a nonzero submodule of Mps(V ) and v ∈ U a nonzero homogeneous element.
Then
v =
∑
i∈I
uivi,
where ui ∈ U(N−) are linearly independent homogeneous, vi ∈ Mps(V ).
Given a root ϕ ∈ ∆ denote by ht(ϕ) the number of simple roots of N− in the decomposition
of ϕ and by ht1(ϕ) the number of all simple roots in the decomposition of ϕ. Suppose
ui ∈ U(N−)−ϕi . We can assume that all ϕi’s have the same ht. Let ht(ϕi) = 1 for all i.
Choose i0 such that ht1(ϕi0) is the least possible. Then there exists a nonzero x ∈ N such
that 0 6= xv ∈ Mps(V ) and [x, ui0 ] ∈ U(G∩N−) since [N,N−]∩G = G(Gps
P ). But U(G∩N−)
is irreducible G(Gps
P )-module. Hence, there exists y ∈ U(N) such that y[x, ui0 ]vi0 = vi0 . In
the same time yxuivi = 0 if i 6= i0. Thus, we obtain vi0 ∈ U and v−ui0vi0 ∈ U . Applying the
induction on |I| we conclude that vi ∈ U for all i ∈ I. This completes the proof in the case
ht(ϕ) = 1. The induction step is considered similarly. Hence, U is generated by U ∩Mps(V )
which implies the statements. �
Corollary 3.1. For each i, let Vi be an irreducible Gi-module with a nonzero action of the
central element. Then ind(Gps
P ,G;⊗iVi) is irreducible.
Proof. Since ⊗iVi is irreducible G
ps
P -module, the statement follows immediately from Theo-
rem 3.1. �
Corollary 3.2. Let V be an irreducible weight non-cuspidal G-module with an injective action
of the central element c. Then there exists a parabolic subalgebra P = P0 ⊕ N of G and an
irreducible weight cuspidal P0-module N such that V ' LP (N), where P is the corresponding
parabolic subset of ∆. Moreover, V ' MP (N) if P0 is infinite-dimensional and NpsN = 0.
Proof. First statement is obvious. If NpsN = 0 then MP (N) ' Mps
P (N) which is irreducible
by Theorem 3.1. �
If P0 = H then Corollary 3.2 implies reduction theorem for Verma type modules [4, 17].
Note that in general ind(GP ,G;N) need not be irreducible if N is irreducible. On the
other hand we have
Induced Modules for Affine Lie Algebras 9
Corollary 3.3. For each i, let Vi be an irreducible Gi-module with the action of the central
element by a nonzero scalar a, V an irreducible highest weight G(GP )-module with highest
weight a. Then
ind(GP ,G;⊗iVi ⊗ V )
is irreducible.
Proof. Note that V is isomorphic to the Verma module with highest weight a. Then
ind(GP ,G;⊗iVi ⊗ V ) ' ind(Gps
P ,G;⊗iVi)
which is irreducible by Corollary 3.1. �
Corollary 3.4. Let λ ∈ H∗, λ(c) 6= 0, B a non-standard Borel subalgebra of G, P corre-
sponding partition of ∆ and
F = {{0} = Fn ⊂ · · · ⊂ F1 ⊂ F0 = W}
the corresponding full flag of maximal length. Suppose δ ∈ Fs−1 \ Fs for some s, 1 ≤ s < n.
Denote ms the Lie subalgebra of G generated by the root subspaces with roots in ∆∩Fs−1 and
by H. Then ms is infinite-dimensional and MP (λ)s = U(ms)vλ is a highest weight module
over ms. Moreover, MP (λ) is irreducible if and only if MP (λ)s is irreducible.
Proof. Indeed, the flag F defines a parabolic subalgebra P of G whose Levi subalgebra
is ms. Hence,
MP (λ) ' ind(P,G;MP (λ)s).
The statement follows immediately from Corollary 3.2. �
3.2 Reduction theorem for type Ib
Let P = P0 ⊕ N be a parabolic subalgebra of G of type Ib, dimP0 < ∞, P corresponding
parabolic subset of ∆ and
F = {{0} = Fk ⊂ · · · ⊂ F1 ⊂ F0 = W}
the corresponding partial flag of maximal length. Then P0 has the root system Fk ∩ ∆.
Since P is of type Ib, there exists s, 1 ≤ s < k, such that δ ∈ Fs and δ /∈ Fs+1. Then,
a Lie subalgebra mP of G generated by H and the root spaces corresponding to the roots
from Fs ∩∆, is infinite-dimensional. Obviously, it can be extended to a parabolic subalgebra
mP ⊕NP of G of type II, where mP is the Levi subalgebra.
Corollary 3.5. Let P be a parabolic subset of ∆ such that G0
P is finite-dimensional and mP
is infinite-dimensional. Consider a weight mP -module V which is annihilated by N
ps
P and on
which the central element c acts injectively. Then for any submodule U of ind(mP ⊕NP ,G;V )
there exists a submodule VU of V such that
U ' ind(mP ⊕NP ,G;VU ).
In particular, ind(mP ⊕NP ,G;V ) is irreducible if and only if V is irreducible.
10 V. Futorny and I. Kashuba
Proof. Consider the pseudo parabolic subalgebra m
ps
P ⊕N
ps
P ⊂ mP ⊕NP , GP ⊂ m
ps
P ⊕N
ps
P .
Then
ind(mP ⊕NP ,G;U(mP )V ) ' ind(mps
P ⊕N
ps
P ,G;U(mps
P )V ),
since U(mps
P )V ' ind(G0
P ,mps
P ;V ) and U(mP )V ' U(mps
P )V ⊗ M, where M is the hig-
hest weight G(GP )-module of highest weight a. Hence, the statement follows from Theo-
rem 3.1. �
Consider now a weight P-module V such that N (and hence Nps) acts trivially on V and c
acts by a multiplication by a nonzero scalar. Then
MP (V ) = ind(P,G;V ) ' ind
(
m
ps
P ⊕N
ps
P ,G;U(mps
P )V
)
and we obtain immediately
Corollary 3.6. Let P be a parabolic subset of ∆ such that G0
P is finite-dimensional and mP
is infinite-dimensional. Consider an irreducible weight G0
P -module V which is annihilated
by G+
P and on which the central element c acts injectively. Then for any submodule U of
MP (V ) there exists a submodule VU of U(mP )V such that
U ' ind(mP ⊕N,G;VU ).
Moreover, MP (V ) is irreducible if and only if U(mP )V is is irreducible.
Corollary 3.6 is essentially Theorem 8 of [15]. In particular it reduces the case of parabolic
subalgebras of type Ib to the case of parabolic subalgebras of type II.
Example 3.2.
• Examples of irreducible dense modules with non-zero central charge were constructed
in [2] as tensor products oh highest and lowest weight modules. Applying functor of
pseudo parabolic induction to these modules one obtains new examples of irreducible
modules over G with infinite-dimensional weight spaces.
• Series of irreducible cuspidal modules over the Heisenberg subalgebra with a non-zero
central charge were constructed in [1]. These modules have infinite-dimensional weight
spaces. We can not apply functor of pseudo parabolic induction to these modules since
the action of the Heisenberg subalgebra is torsion free. On the other hand, in the case
of A
(1)
1 , the functor of parabolic induction applied to such modules gives again new
irreducible modules.
Parabolic induction can be easily generalized to the non-weight case as follows (cf. [15]).
Let P = P0⊕N be a parabolic subalgebra of type II, P corresponding parabolic subset of ∆,
P 0 = P ∩ −P . Let H′ be the linear span of [Gα,G−α], α ∈ P 0 and HP a complement of H′
in H such that [P0,HP ] = 0.
Let Λ be an arbitrary Abelian category of P0-modules (note that Λ may have a different
Abelian structure than the category of modules over P0). Given V ∈ Λ and λ ∈ H∗
P one
makes V into a P-module with h|V = λ(h)Id for any h ∈ HP and NV = 0. Then one can
construct a G-module MP (V, λ) by parabolic induction. It follows from the construction that
MP (V, λ) is HP -diagonalizable. Analogously to Theorem 3.1 one can show
Induced Modules for Affine Lie Algebras 11
Theorem 3.2. If the central element c acts injectively on V ∈ Λ then for any submodule U
of MP (V, λ) there exists a submodule VU of V such that
U ' MP (VU , λ).
In particular, MP (V, λ) is irreducible if and only V is irreducible.
4 Categories of induced modules
Let G be an affine Lie algebra of rank > 1, P = P0⊕N a pseudo parabolic subalgebra of G of
type II, P corresponding parabolic subset of ∆, W (P0) the category of weight (with respect
to H) P0-modules V with an injective action of c.
Denote by O(G,P) the category of weight G-modules M such that the action of the central
element c on M is injective and M contains a nonzero P-primitive element. Modules MP (V )
and LP (V ) are typical objects of O(G,P).
For M ∈ O(G,P) we denote by MN the subspace of M consisting of P-primitive elements,
that is the subspace of N-invariants. Clearly, MN is a P0-module and hence MN ∈ W (P0).
Let Õ(G,P) be the full subcategory of O(G,P) whose objects M are generated by MN.
Again MP (V ) and LP (V ) are objects of Õ(G,P).
Both categories O(G,P) and Õ(G,P) are closed under the operations of taking submo-
dules, quotients and countable direct sums.
The parabolic induction provides a functor
I : W (P0) → O(G,P), V 7→ MP (V ) = ind(P,G;V ).
Note that MP (N) ' Mps
P (N). The canonical image of V in MP (V ) is annihilated by N and,
hence, I(V ) is generated by its P-primitive elements. Thus I(V ) ∈ Õ(G,P).
In the opposite direction we have a well defined functor
R : O(G,P) → W (P0), M 7→ MN.
Denote by R̃ the restriction of R onto the category Õ(G,P).
We need the following lemma.
Lemma 4.1.
• Let V ∈ W (P0), M a subquotient of MP (V ). Then w(MN) ⊂ w(V ).
• MP (V ⊕ V ′) ' MP (V )⊕MP (V ′).
• Let M ∈ Õ(G,P) be generated by P-primitive elements then M is a direct sum of
modules of type MP (V ).
Proof. Without loss of generality we will assume that M is a quotient of MP (V ). Suppose
M = MP (V )/M ′. Then MN ' V/(M ′ ∩ V ) (identifying V with 1⊗ V ) by Theorem 3.1, and
the first statement follows. Second statement follows immediately from the definition. Let
M ∈ Õ(G,P) and MN = M ′ ⊕M ′′ is a direct sum of P0-modules. If M is generated by MN
then M ' MP (MN) by Theorem 3.1, and hence M ' MP (M ′)⊕MP (M ′′). �
12 V. Futorny and I. Kashuba
Theorem 4.1.
• The functor I : W (P0) → O(G,P) is a left adjoint to the functor R : O(G,P) → W (P0),
that is R ◦ I is naturally isomorphic to the identity functor on W (P0).
• The functors R̃ and I are mutually inverse equivalences of W (P0) and the subcategory
Õ(G,P).
Proof. Let V ∈ W (P0). Clearly, V is naturally embedded into MP (V )N. On the other hand,
w(MP (V )N) ⊂ w(V ) by Lemma 4.1. Since U(P0)V ' V then MP (V )N ' V . If M = MP (V )
then R(M) ' V and (I ◦R)(MP (V )) ' MP (V ) by Theorem 3.1. If M is an arbitrary object
in O(G,P) then M is generated by MN = R(M) and MN = ⊕iMi, where Mi are P0-modules
and w(Mi)∩w(Mj) = ∅ if i 6= j. Then M ' ⊕iMP (Mi) by Theorem 3.1. On the other hand,
I(M (N)) ' ⊕iMP (Mi) by Lemma 4.1. Hence, (I ◦R)(M) ' M , implying the statement. �
Theorem 1.1 is an immediate consequence of Theorem 3.1 or Theorem 4.1.
Let F (P ) be a partial flag of the parabolic subset P of ∆:
F (P ) = {{0} = Fk ⊂ · · · ⊂ F1 ⊂ F0 = W},
P 0 = Fk∩∆, P0 = G0
P , ∆Im ⊂ Fk. Then P0 is a subalgebra of G generated by Gα, α ∈ Fk∩∆
and H.
Fix s, k ≤ s < n. Then δ ∈ Fs and the Lie subalgebra ms
P ⊂ G, generated by H and
the root spaces corresponding to the roots from Fs ∩∆, is infinite-dimensional. Obviously, it
can be extended to a parabolic subalgebra of G of type II, where ms
P is the Levi subalgebra:
Ps = ms
P ⊕Ns, Ns ⊂ N. If V is a P-module with NN = 0 then V ′ = U(ms
P )V is a Ps-module,
NsV
′ = 0. If V ′ is a Ps-module with a trivial action of Ns and injective action of the central
element c then the structure of the induced module ind(Ps,G;V ′) is completely determined
by the structure of V ′ by Theorem 3.1. In particular, this module is irreducible if and only
if V ′ is irreducible. Moreover, since
MP (V ) ' ind(Ps,G;V ′),
we have the following interesting observation.
Corollary 4.1. If V ∈ W (P0), s is such that k ≤ s < n and V ′ = U(ms
P )V then the
submodule structure of the induced module M = ind(Ps,G;V ′) is determined by the submodule
structure of V (as in Theorem 3.1). In particular, M is irreducible if and only if V is
irreducible.
Denote by W (ms
P ) the category of weight ms
P -modules with injective action of the central
element c and by O(G,Ps) the category of weight G-modules M such that the action of c
on M is injective and M contains a nonzero Ps-primitive element. Modules MPs(V ) and
LPs(V ) are the objects of O(G,Ps), V ∈ W (Ps). For M ∈ O(G,Ps) we denote by MNs the
subspace of M consisting of Ps-primitive elements. Denote by Õ(G,Ps) the full subcategory
of O(G,Ps) whose objects M are generated by MNs .
Then we have the following functors
Is : W (ms
P ) → O(G,Ps), V 7→ MPs(V ) = ind(Ps,G;V ),
Rs : O(G,Ps) → W (ms
P ), M 7→ MNs
and R̃s the restriction of Rs onto Õ(G,Ps).
Induced Modules for Affine Lie Algebras 13
Theorem 4.2. For any s, k ≤ s < n,
• the functor Is is a left adjoint to Rs;
• the functors R̃s and Is are mutually inverse equivalences of W (ms
P ) and the subcategory
Õ(G,Ps).
Note that for k ≤ s < r we have
W (ms
P )
Is //
� _
��
Õ(G,Ps)
R̃s
oo
� _
��
W (mr
P )
Ir // Õ(G,Pr)
R̃r
oo
Hence, Is is just the restriction of Ir onto W (ms
P ), while R̃s is the restriction of R̃r onto
Õ(G,Ps).
Remark 4.1. One can establish similar category equivalences for non-weight modules (cf.
Section 5 in [15]).
Acknowledgements
The first author is supported in part by the CNPq grant (301743/2007-0) and by the Fapesp
grant (2005/60337-2). The second author is supported by the Fapesp grant (2007/025861).
References
[1] Bekkert V., Benkart G., Futorny V., Weyl algebra modules, Kac–Moody Lie Algebras and Related Topics,
Contemp. Math. 343 (2004), 17–42.
[2] Chari V., Pressley A., New unitary representations of loop groups, Math. Ann. 275 (1986), 87–104.
[3] Chari V., Integrable representations of affine Lie algebras, Invent. Math. 85 (1986), 317–335.
[4] Cox B., Verma modules induced from nonstandard Borel subalgebras, Pacific J. Math. 165 (1994),
269–294.
[5] Dimitrov I., Futorny V., Penkov I., A reduction theorem for highest weight modules over toroidal Lie
algebras, Comm. Math. Phys. 250 (2004), 47–63.
[6] Dimitrov I., Grantcharov D., Private communication.
[7] Dimitrov I., Mathieu O., Penkov I., On the structure of weight modules, Trans. Amer. Math. Soc. 352
(2000), 2857–2869.
[8] Dimitrov I., Penkov I., Partially integrable highest weight modules, Transform. Groups 3 (1998), 241–253.
[9] Fernando S., Lie algebra modules with finite-dimensional weight spaces. I, Trans. Amer. Math. Soc. 322
(1990), 757–781.
[10] Futorny V., The weight representations of semisimple finite-dimensional Lie algebras, in Algebraic Struc-
tures and Applications, Kiev University, 1988, 142–155.
[11] Futorny V., Representations of affine Lie algebras, Queen’s Papers in Pure and Applied Mathematics,
Vol. 106, Queen’s University, Kingston, ON, 1997.
[12] Futorny V., Irreducible non-dense A
(1)
1 -modules, Pacific J. Math. 172 (1996), 83–99.
14 V. Futorny and I. Kashuba
[13] Futorny V., The parabolic subsets of root systems and corresponding representations of affine Lie al-
gebras, in Proceedings of the International Conference on Algebra (Novosibirsk, 1989), Contemp. Math.
131 (1992), part 2, 45–52.
[14] Futorny V., Imaginary Verma modules for affine Lie algebras, Canad. Math. Bull. 37 (1994), 213–218.
[15] Futorny V., König S., Mazorchuk V., Categories of induced modules for Lie algebras with triangular
decomposition, Forum Math. 13 (2001), 641–661.
[16] Futorny V., König S., Mazorchuk V., Categories of induced modules and projectively stratified algebras,
Algebr. Represent. Theory 5 (2002), 259–276.
[17] Futorny V., Saifi H., Modules of Verma type and new irreducible representations for affine Lie algebras,
in Representations of Algebras (Ottawa, ON, 1992), CMS Conf. Proc., Vol. 14, Amer. Math. Soc.,
Providence, RI, 1993, 185–191.
[18] Futorny V., Tsylke A., Classification of irreducible nonzero level modules with finite-dimensional weight
spaces for affine Lie algebras, J. Algebra 238 (2001), 426–441.
[19] Jakobsen H.P., Kac V.G., A new class of unitarizable highest weight representations of infinite-
dimensional Lie algebras, in Nonlinear Equations in Classical and Quantum Field Theory (Meudon/Paris,
1983/1984), Lecture Notes in Phys., Vol. 226, Springer, Berlin, 1985, 1–20.
[20] Kac V.G., Infinite-dimensional Lie algebras, 3rd ed., Cambridge University Press, Cambridge, 1990.
[21] Mathieu O., Classification of irreducible weight modules, Ann. Inst. Fourier (Grenoble) 50 (2000), 537–
592.
[22] Spirin S., Z2-graded modules with one-dimensional components over the Lie algebra A
(1)
1 , Funktsional.
Anal. i Prilozhen. 21 (1987), 84–85.
1 Introduction
2 Preliminaries on Borel subalgebras and parabolic subalgebras
3 Parabolic induction
3.1 Reduction theorem for type II
3.2 Reduction theorem for type Ib
4 Categories of induced modules
References
|