Weyl modules for the twisted loop algebras
Vyjayanthi Chari, Ghislain Fourier, Prasad Senesi

TL;DR
This paper extends the concept of Weyl modules from untwisted to twisted affine algebras, establishing an identification that enables computation of their dimensions and characters using known untwisted results.
Contribution
It introduces a new framework linking Weyl modules of twisted affine algebras to those of untwisted algebras, facilitating their analysis.
Findings
Identification of twisted Weyl modules with untwisted ones
Explicit computation of dimensions and characters
Application of known untwisted results to twisted cases
Abstract
The notion of a Weyl module, previously defined for the untwisted affine algebras, is extended here to the twisted affine algebras. We describe an identification of the Weyl modules for the twisted affine algebras with suitably chosen Weyl modules for the untwisted affine algebras. This identification allows us to use known results in the untwisted case to compute the dimensions and characters of the Weyl modules for the twisted algebras.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAlgebraic structures and combinatorial models · Advanced Topics in Algebra · Nonlinear Waves and Solitons
Weyl modules for the twisted loop algebras
Vyjayanthi Chari, Ghislain Fourier and Prasad Senesi
VC: Department of Mathematics, University of California, Riverside, CA 92521
GF: Mathematisches Institut, Universität zu Köln, 50931 Köln
PS: Department of Mathematics, University of California, Riverside, CA 92521
Abstract.
The notion of a Weyl module, previously defined for the untwisted affine algebras, is extended here to the twisted affine algebras. We describe an identification of the Weyl modules for the twisted affine algebras with suitably chosen Weyl modules for the untwisted affine algebras. This identification allows us to use known results in the untwisted case to compute the dimensions and characters of the Weyl modules for the twisted algebras.
VC was partially supported by the NSF grant DMS-0500751
1. Introduction
The notion of Weyl modules for the untwisted affine Lie algebras was introduced in [6] and was motivated by an attempt to understand the category of finite dimensional representations of the untwisted quantum affine algebra. Specifically, the Weyl modules were conjectured to be the limit of certain irreducible representations of the quantum affine algebras. It was proved that the conjecture was true for and that this conjecture would follow if the dimensions of the Weyl modules were known. H. Nakajima has pointed out recently that the dimension formula follows by using results of [2] and [12].
Another approach to proving the dimension formula for the Weyl modules can be found in [4] for and in [10] for the general simply laced case. These papers also make the connection between Weyl modules and the Demazure modules for affine Lie algebras and also with the fusion product defined by [7]. The approach in these papers is rather simple and show that one can study the Weyl modules from a purely classical viewpoint. Other points of interest and generalizations of these can be found in [8].
We now turn our attention to the case of the twisted affine algebras. None of the quantum machinery is available and in fact there are rather few results on the category of finite dimensional representations of the twisted quantum affine algebras [1], [5]. These results do show however that one can make a similar conjecture; i.e that one can define a notion of the Weyl module for the twisted affine Lie algebras such that they are the specializations of irreducible modules in the quantum case. To do this, one requires the Weyl modules to be universal in a suitable sense. One of the difficulties is in the case of the algebras of type , which are not built up entirely of algebras isomorphic to ; and indeed one needs to understand on its own. Thus, we use results of [9], [13] to arrive at the correct definition of the Weyl modules.
The next question clearly is to determine the dimensions of the Weyl modules and also their decomposition as modules for the underlying finite–dimensional simple Lie algebra. In the untwisted case these questions can be answered either by using the fusion product of [7] or the fact that the modules are specializations of modules for the quantum affine algebra. Both these techniques are unavailable to us in the twisted case, as far as we know the notion of fusion product does not admit a generalization to the twisted algebras. We get around these difficulties by identifying the Weyl modules for the twisted algebras , with suitably chosen Weyl modules for the untwisted algebra . We then use all the known results in the untwisted case to complete our analysis of the twisted algebras. In conclusion, we note that some of the methods we use in this paper give simpler proofs of some of the results in [6].
2. The untwisted loop algebras and the modules W({\mbox{\boldmath\pi}}).
2.1.
Throughout the paper (resp. ) denotes the set of complex (resp. non–zero complex) numbers, and (resp. ) the set of integers (resp. non–negative) integers. Given a Lie algebra we denote by the universal enveloping algebra of and by denotes the loop algebra of . Specifically, we have
[TABLE]
with commutator given by
[TABLE]
for , . We identify with the subalgebra of . Given , we let be the automorphism defined by extending for all , .
Given and let be the quotient of by the ideal .
Lemma**.**
Let be such that has distinct coordinates. For all , we have
[TABLE]
Proof.
Since if , it is standard that
[TABLE]
and the lemma now follows trivially. ∎
2.2. The simple Lie algebras and their representations
Let be any finite-dimensional complex simple Lie algebra and a Cartan subalgebra of and the corresponding Weyl group. Let be the set of roots of with respect to , an index set for a set of simple roots (and hence also for the fundamental weights), the set of positive roots, (resp. ) the span of the simple roots (resp. fundamental weights) and be the highest root in . Given let be the corresponding root space, we have
[TABLE]
Fix a Chevalley basis , for and set
[TABLE]
In particular for ,
[TABLE]
Given a finite–dimensional representation of on a complex vector space , we can write
[TABLE]
Set . It is well–known that
[TABLE]
and that is isomorphic to a direct sum of irreducible representations. The set of isomorphism classes of irreducible finite–dimensional –modules is in bijective correspondence with and for any let be an element of the corresponding isomorphism class. Then is generated by an element satisfying the relations:
[TABLE]
2.3. Identities in
For it is easy to see that the elements span a subalgebra of which is isomorphic to . We shall need the following formal power series in with coefficients in . For , set
[TABLE]
Given a power series in with coefficients in an algebra , let be the coefficient of (). The following result was proved in [11, Lemma 7.5], (see [6, Lemma 1.3] for the formulation in this notation).
Lemma**.**
Let .
[TABLE]
∎
2.4. The monoid
Let be the monoid of –tuples of polynomials {\mbox{\boldmath\pi}}=(\pi_{1},\cdots,\pi_{n}) in an indeterminate with constant term one, with multiplication being defined component wise. For and , set
[TABLE]
and for , set
[TABLE]
Clearly any {\mbox{\boldmath\pi}}^{+}\in\cal P^{+} can be written uniquely as a product
[TABLE]
for some and distinct elements and in this case we set {\mbox{\boldmath\pi}}^{-}=\prod_{k=1}^{\ell}{\mbox{\boldmath\pi}}_{\lambda_{i},{}\ {}a_{i}^{-1}}. Define a map by {\mbox{\boldmath\pi}}\to\lambda_{\mbox{\boldmath\pi}}=\sum_{i\in I}\deg(\pi_{i})\omega_{i}.
2.5. The modules W({\mbox{\boldmath\pi}}), V({\mbox{\boldmath\pi}}).
Given {\mbox{\boldmath\pi}}=(\pi_{i})_{i\in I}\in\cal P^{+}, let W({\mbox{\boldmath\pi}}) be the –module generated by an element w_{\mbox{\boldmath\pi}} with relations:
[TABLE]
where \lambda_{\mbox{\boldmath\pi}}=\sum_{i\in I}(\deg\pi_{i})\omega_{i}, {\mbox{\boldmath\pi}}^{+}={\mbox{\boldmath\pi}}, and . It is not hard to see that if we write {\mbox{\boldmath\pi}}=\prod_{k=1}^{\ell}{\mbox{\boldmath\pi}}_{\lambda_{\ell},a_{\ell}} where are all distinct, then for
[TABLE]
Let and let \tau_{b}W({\mbox{\boldmath\pi}}) be the –module obtained by pulling back W({\mbox{\boldmath\pi}}) through the automorphism of . The next result is standard.
Lemma**.**
- (i)
Let {\mbox{\boldmath\pi}}\in\cal P^{+}. Then W({\mbox{\boldmath\pi}})=\mathbf{U}(L(\mathfrak{n}^{-}))w_{\mbox{\boldmath\pi}}, and hence we have,
[TABLE]
In particular, the module W({\mbox{\boldmath\pi}}) has a unique irreducible quotient V({\mbox{\boldmath\pi}}).
- (ii)
For , we have \tau_{b}W({\mbox{\boldmath\pi}})\cong W({\mbox{\boldmath\pi}}_{b}), where {\mbox{\boldmath\pi}}=(\pi_{i}(u))_{i\in I} and {\mbox{\boldmath\pi}}_{b}=(\pi_{i}(b^{-1}u))_{i\in I}. In particular we have
[TABLE]
2.6.
The modules W({\mbox{\boldmath\pi}}) were initially defined and studied in [6] and a formula was conjectured for their dimension. Parts (i) and (ii) of the next theorem were proved in [6]. Part (iii) was proved in [6] in the case of , for it was proved in [4] and for the general simply laced case in [10]. Part (iii) can be deduced for the general case by using results of [2],[12],[14] for quantum affine algebras.
Theorem 1**.**
- (i)
Given {\mbox{\boldmath\pi}}=(\pi_{i})_{i\in I} with unique decomposition {\mbox{\boldmath\pi}}=\prod_{k=1}^{\ell}{\mbox{\boldmath\pi}}_{\lambda_{\ell},a_{\ell}}, we have an isomorphism of –modules
[TABLE]
- (ii)
Let be any finite–dimensional –module generated by an element such that
[TABLE]
Then there exists {\mbox{\boldmath\pi}}\in\cal P^{+} such that the assignment w_{\mbox{\boldmath\pi}}\to v extends to a surjective homomorphism W({\mbox{\boldmath\pi}})\to V of –modules.
- (iii)
Let and . Suppose that . Then
[TABLE]
∎
2.7. Annihilating ideals for W({\mbox{\boldmath\pi}}).
The next proposition is implicit in [6] but since it plays a big role in this paper we make it explicit and give a proof.
Proposition**.**
Let {\mbox{\boldmath\pi}}=\prod_{r=1}^{\ell}{\mbox{\boldmath\pi}}_{\lambda_{r},a_{r}}\in\cal P^{+}. There exists an integer N=N({\mbox{\boldmath\pi}}) such that
[TABLE]
Proof.
We begin by proving that for all
[TABLE]
Set N_{i}=\lambda_{\mbox{\boldmath\pi}}(h_{i}). Using the defining relations of W({\mbox{\boldmath\pi}}) and Lemma Lemma,
[TABLE]
We also have
[TABLE]
Combining these we get
[TABLE]
But it is elementary to see that
[TABLE]
which proves (2.3). Since is generated by the elements , , it is immediate that there exists such that
[TABLE]
Since and W({\mbox{\boldmath\pi}})\cong\mathbf{U}(L(\mathfrak{n}^{-}))w_{\mbox{\boldmath\pi}} as vector spaces, we get
[TABLE]
Since any element in is in the span of elements of the form , we now get
[TABLE]
∎
Corollary**.**
Given {\mbox{\boldmath\pi}}\in\cal P^{+} with unique decomposition {\mbox{\boldmath\pi}}=\prod_{r=1}^{\ell}{\mbox{\boldmath\pi}}_{\lambda_{r},a_{r}}\in\cal P^{+}, there exists such that the action of on W({\mbox{\boldmath\pi}}) factors through to an action of on W({\mbox{\boldmath\pi}}) and W({\mbox{\boldmath\pi}})=\mathbf{U}(L(\mathfrak{n}^{-}_{\mathbf{a},N}))w_{\mbox{\boldmath\pi}}.
3. The twisted algebras and the modules W({\mbox{\boldmath\pi}}^{\sigma})
3.1.
Assume from now on that is simply–laced and that is a non–trivial diagram automorphism of of order . In particular induces a permutation of and and we have
[TABLE]
Let be a primitive root of unity, we have
[TABLE]
Given any subalgebra of which is preserved by , set . It is known that is a simple Lie algebra, is a Cartan subalgebra and that is an irreducible representation of for all . Moreover,
[TABLE]
The following table describes the various possibilities for , and the structure of as a –module, here is the highest short root of and .
[TABLE]
*From now we set , , the sets , etc. are defined similarly. The set of – orbits of has the same cardinality as and we identify with a subset of . In the case when is of type we assume that corresponds to the unique short simple root of . We shall also fix a primitive root of unity. *
Suppose that is one of the sets , or and assume that and that if is of type . Define subsets of by
[TABLE]
If is of type , then we set,
[TABLE]
Finally if is of type and , set,
[TABLE]
In the rest of this paper in the case when is of type , we shall only be interested in elements such that and we let denote this subset of . Moreover we regard as an element of as follows:
[TABLE]
3.2.
Let be the automorphism defined by extending,
[TABLE]
for , . Then is or order and we let be the subalgebra of fixed points of . Clearly,
[TABLE]
Lemma**.**
Let and assume that if is of type . The subalgebra of spanned by the elements is canonically isomorphic to . If is of type the subalgebra of spanned by the elements is canonically isomorphic to . ∎
3.3. Identities in
Suppose that either is not of type and or that is of type and . Define power series with coefficients in by,
[TABLE]
If is not of type and , then we set
[TABLE]
Finally, if is of type and we have,
[TABLE]
Lemma**.**
Let .
- (i)
If is not of type and or is of type and , we have
[TABLE]
- (ii)
If is not of type and ,
[TABLE]
- (iii)
If is of type , we have
- (a)
2. (b)
where
[TABLE]
Proof.
Parts (i) and (ii) are immediate consequences of Lemma Lemma and Lemma Lemma. Part (iii) is deduced from [13], [9, Lemma 5.36], exactly as (i) and (ii) were deduced from Garland in [6]. ∎
3.4. The monoid
Let be the form on induced by the Killing form of normalized so that . For and , and not of type let
[TABLE]
while if is of type we set for , , ,
[TABLE]
Let be the monoid generated by the elements {\mbox{\boldmath\pi}}^{\sigma}_{\lambda,a}. Define a map by
[TABLE]
if is not of type and
[TABLE]
if is of type . It is clear that any {\mbox{\boldmath\pi}}^{\sigma}\in\cal P^{+}_{\sigma} can be written (non–uniquely) as product
[TABLE]
where and have distinct coordinates. We call any such expression a standard decomposition of {\mbox{\boldmath\pi}}^{\sigma}.
3.5. The set \mathbf{i}({\mbox{\boldmath\pi}}^{\sigma})
Given and , define elements by,
[TABLE]
Define a map as follows. Given {\mbox{\boldmath\pi}}\in\cal P^{+} write
[TABLE]
and set
[TABLE]
Note that is well defined since the choice of is unique and set
[TABLE]
We now give an explicit description of the set \mathbf{i}({\mbox{\boldmath\pi}}^{\sigma}). Recall that given , we also regard as in Section 3.1. In addition, define for .
Lemma**.**
- (i)
Let and . We have,
[TABLE]
and for and ,
[TABLE]
- (ii)
Let {\mbox{\boldmath\pi}}^{\sigma}=\prod_{k=1}^{\ell}\prod_{\epsilon=0}^{m-1}\prod_{i\in I_{0}}({\mbox{\boldmath\pi}}^{\sigma}_{\omega_{i},\zeta^{\epsilon}a_{k}})^{m_{k,\epsilon,i}} be a decomposition of {\mbox{\boldmath\pi}}^{\sigma} into linear factors for not of type . Then
[TABLE]
where the product of the sets is understood to be the set of products of elements of the sets.
In the case of , let {\mbox{\boldmath\pi}}^{\sigma}=\prod_{k=1}^{\ell}\prod_{\epsilon=0}^{1}\prod_{i\in I_{0}}({\mbox{\boldmath\pi}}^{\sigma}_{(1+\delta_{i,n})\omega_{i},\zeta^{\epsilon}a_{k}})^{m_{k,\epsilon,i}} be a decomposition of {\mbox{\boldmath\pi}}^{\sigma} into linear factors. Then
[TABLE]
- (iii)
In particular, we have
[TABLE]
where and .
Proof.
The first statement is trivially checked, noting that if is a fixed point of , then {\mbox{\boldmath\pi}}^{\sigma}_{\omega_{i},a}={\mbox{\boldmath\pi}}^{\sigma}_{\omega_{i},\zeta^{r}a} for . The other statements follow immediately from the first one. ∎
From here on we shall assume that, unless otherwise noted, the element {\mbox{\boldmath\pi}}\in\mathbf{i}({\mbox{\boldmath\pi}}^{\sigma}) chosen is of the form given in (iii) of the lemma.
3.6. The modules W({\mbox{\boldmath\pi}}^{\sigma}), V({\mbox{\boldmath\pi}}^{\sigma})
Given {\mbox{\boldmath\pi}}^{\sigma}=(\pi_{i,\sigma})_{i\in I_{0}}\in\cal P^{+}_{\sigma}, the Weyl module W({\mbox{\boldmath\pi}}^{\sigma}) is the -module generated by an element w_{{\mbox{\boldmath\pi}}^{\sigma}} with relations:
[TABLE]
for all and . If {\mbox{\boldmath\pi}}^{\sigma}=\prod_{k=1}^{\ell}{\mbox{\boldmath\pi}}_{\lambda_{k},a_{k}}^{\sigma}\in\cal P_{\sigma}^{+}, it is not hard to see that for , we have if not of type ,
[TABLE]
and for of type ,
[TABLE]
3.7.
For we have and we let \tau_{b}W({\mbox{\boldmath\pi}}^{\sigma}) be the –module obtained by pulling back W({\mbox{\boldmath\pi}}^{\sigma}) through . The next result is proved by standard methods.
Lemma**.**
- (i)
Let {\mbox{\boldmath\pi}}^{\sigma}\in\cal P^{+}_{\sigma}. Then W({\mbox{\boldmath\pi}}^{\sigma})=\mathbf{U}(L^{\sigma}(\mathfrak{n}^{-}))w_{\mbox{\boldmath\pi}}^{\sigma}, and hence we have,
[TABLE]
In particular, the module W({\mbox{\boldmath\pi}}^{\sigma}) has a unique irreducible quotient V({\mbox{\boldmath\pi}}^{\sigma}).
- (ii)
For , we have \tau_{b}W({\mbox{\boldmath\pi}}^{\sigma})\cong W({\mbox{\boldmath\pi}}_{b}^{\sigma}), where {\mbox{\boldmath\pi}}^{\sigma}=(\pi_{i}(u))_{i\in I} and {\mbox{\boldmath\pi}}^{\sigma}_{b}=(\pi_{i}(b^{-1}u))_{i\in I}. In particular we have
[TABLE]
∎
3.8. The main theorem
In the rest of this paper we shall prove the following result.
Theorem 2**.**
- (i)
Let {\mbox{\boldmath\pi}}^{\sigma}\in\cal P^{+}_{\sigma}. For all {\mbox{\boldmath\pi}}\in\mathbf{i}({\mbox{\boldmath\pi}}^{\sigma}), we have
[TABLE]
- (ii)
Let {\mbox{\boldmath\pi}}^{\sigma}\in\cal P^{+}_{\sigma} and assume that \prod_{k=1}^{\ell}\prod_{\epsilon=0}^{m-1}{\mbox{\boldmath\pi}}^{\sigma}_{\lambda_{k,\epsilon},\zeta^{\epsilon}a_{k}}\in\cal P^{+}_{\sigma} is a standard decomposition of . As –modules, we have
[TABLE]
- (iii)
Suppose that \prod_{\epsilon=0}^{m-1}{\mbox{\boldmath\pi}}^{\sigma}_{\lambda_{\epsilon},\zeta^{\epsilon}a}\in\cal P^{+}_{\sigma}. Then
[TABLE]
- (iv)
Let and . We have for not of type
[TABLE]
and for of type
[TABLE]
- (v)
Let be any finite–dimensional –module generated by an element such that
[TABLE]
Then there exists {\mbox{\boldmath\pi}}^{\sigma}\in\cal P^{+}_{\sigma} such that the assignment w_{{\mbox{\boldmath\pi}}^{\sigma}}\to v extends to a surjective homomorphism W({\mbox{\boldmath\pi}}^{\sigma})\to V of –modules.
∎
4. Proof of Theorem 2
4.1. Annihilating ideals for W({\mbox{\boldmath\pi}}^{\sigma}).
Proposition**.**
Let {\mbox{\boldmath\pi}}^{\sigma}=\prod_{r=1}^{\ell}{\mbox{\boldmath\pi}}^{\sigma}_{\lambda_{r},a_{r}}\in\cal P^{+}_{\sigma}. There exists an integer N=N({\mbox{\boldmath\pi}}) such that
[TABLE]
Proof.
The subalgebra is canonically isomorphic to . It follows from the defining relations that
[TABLE]
and hence, \mathbf{U}(L^{m}(\mathfrak{g}_{0}))w_{{\mbox{\boldmath\pi}}^{\sigma}} is a quotient of the –module W_{\mathfrak{g_{0}}}({\mbox{\boldmath\pi}}_{m}) where
[TABLE]
It follows from (2.4) that
[TABLE]
for some , where is the highest root in .
Assume first that is not of type , then
[TABLE]
The first equality in (4.2) gives \left(x^{-}_{\theta_{0}}\otimes\prod_{r=1}^{\ell}(t^{m}-a_{r}^{m})^{N}\right)W({\mbox{\boldmath\pi}}^{\sigma})=0,. One deduces now as in the untwisted case that
[TABLE]
Applying to the preceding equation and using the second equality in (4.2) gives
[TABLE]
for all and the result is proved.
Assume now that is of type . This time, we use the fact that
[TABLE]
together with (4.1) to conclude that
[TABLE]
Hence
[TABLE]
for some . Since the element generates as a –module and , we can now prove by similar arguments that for some ,
[TABLE]
Next, using the fact that , we get
[TABLE]
which finally gives
[TABLE]
and completes the proof.
∎
Given positive integers , and a subalgebra of such that , let
[TABLE]
Corollary**.**
Let {\mbox{\boldmath\pi}}^{\sigma}=\prod_{r=1}^{\ell}{\mbox{\boldmath\pi}}^{\sigma}_{\lambda_{r},a_{r}}\in\cal P^{+}_{\sigma} be a standard decomposition of {\mbox{\boldmath\pi}}^{\sigma} and set There exists such that
[TABLE]
4.2.
Proposition**.**
For all {\mbox{\boldmath\pi}}^{\sigma}\in\cal P^{+}_{\sigma}, the –module W({\mbox{\boldmath\pi}}^{\sigma}) is finite–dimensional.
Proof.
Let u\in W({\mbox{\boldmath\pi}}^{\sigma}) and write u=yw_{{\mbox{\boldmath\pi}}^{\sigma}} for some . The adjoint action of the subalgebras on and hence on is nilpotent. Using the defining relations we get immediately that for some , we have
[TABLE]
This implies that is a finite–dimensional –submodule of W({\mbox{\boldmath\pi}}^{\sigma}), and hence W({\mbox{\boldmath\pi}}^{\sigma}) is isomorphic to a direct sum of –modules. Write,
[TABLE]
where W({\mbox{\boldmath\pi}}^{\sigma})_{\mu}=\{u\in W({\mbox{\boldmath\pi}}^{\sigma}):hu=\mu(h)u,\ \ \forall\ \ h\in\mathfrak{h}_{0}\}. The representation theory of now implies that
[TABLE]
Since W({\mbox{\boldmath\pi}}^{\sigma})_{\mu}=0 unless and the number of elements in with this property is finite we get that W({\mbox{\boldmath\pi}}^{\sigma})_{\nu}=0, for all but finitely many . The proposition follows if we prove that \dim(W({\mbox{\boldmath\pi}}^{\sigma})_{\nu})<\infty for all .
Choose and as in Corollary Proposition. Then
[TABLE]
where
[TABLE]
Since this subspace is finite–dimensional it follows that \dim(W({\mbox{\boldmath\pi}}^{\sigma})_{\nu})<\infty as required. ∎
4.3.
Let and . The inclusion obviously induces a Lie algebra map , where . The following proposition will play a crucial role in the proof of Theorem 2.
Proposition**.**
Let be such that and have distinct coordinates. For all we have an isomorphism of Lie algebras,
[TABLE]
for all . In particular, the composite map is surjective.
Proof.
The proof that
[TABLE]
is an obvious modification of the one given in Lemma Lemma which also shows now that it is sufficient to prove the proposition when . For this, let and where . Then,
[TABLE]
which proves that is injective. The proposition follows by noting that
[TABLE]
∎
4.4.
We note some elementary observations which we use without further comment. Any –module (resp. ) is obviously a –module (resp. –module). Moreover if is such that and have distinct coordinates then for all , any –module is also a –module and we write it as . Similarly if we start with a –module we get a –module which we write as . Note also that if is an –module, then
[TABLE]
4.5.
Lemma**.**
Let {\mbox{\boldmath\pi}}^{\sigma}\in\cal P^{+}_{\sigma}, and assume that {\mbox{\boldmath\pi}}\in\mathbf{i}({\mbox{\boldmath\pi}}^{\sigma}).
- (i)
There exists and with and having distinct coordinates such that W({\mbox{\boldmath\pi}}) and W({\mbox{\boldmath\pi}}^{\sigma}) are modules for both and .
- (ii)
In particular,
[TABLE]
and V({\mbox{\boldmath\pi}})_{\mathfrak{g}^{\sigma}_{\mathbf{a}^{m},N}} is an irreducible –module.
Proof.
Let {\mbox{\boldmath\pi}}=\prod_{k=1}^{\ell}{\mbox{\boldmath\pi}}_{\lambda_{k},a_{k}}, where and have distinct coordinates. Proposition Proposition implies that W({\mbox{\boldmath\pi}})=\mathbf{U}(\mathfrak{g}_{\mathbf{a},N})w_{\mbox{\boldmath\pi}}. Using proposition Proposition we see that W({\mbox{\boldmath\pi}}) is also module for and so we get
[TABLE]
Similarly Proposition Proposition implies that W({\mbox{\boldmath\pi}}^{\sigma}) is a module for and hence for . Since V({\mbox{\boldmath\pi}}) is an irreducible module for , it follows that it is also irreducible as a –module and the proposition is proved. ∎
The following proposition proves (i) of Theorem 2.
Proposition**.**
Let {\mbox{\boldmath\pi}}^{\sigma}\in\cal P^{+}_{\sigma}, {\mbox{\boldmath\pi}}\in\mathbf{i}({\mbox{\boldmath\pi}}^{\sigma}).
- (i)
Regarded as –module W({\mbox{\boldmath\pi}}) is a quotient of W({\mbox{\boldmath\pi}}^{\sigma}) and hence
[TABLE]
- (ii)
For , the –module structure on W({\mbox{\boldmath\pi}}^{\sigma}) (resp. V({\mbox{\boldmath\pi}}^{\sigma})) extends to an –module action on W({\mbox{\boldmath\pi}}^{\sigma}) (resp. V({\mbox{\boldmath\pi}}^{\sigma})).
- (iii)
The module W({\mbox{\boldmath\pi}}^{\sigma})_{\mathfrak{g}_{\mathbf{a},N}} is a –module quotient of W({\mbox{\boldmath\pi}}).
Proof.
Using (3.1), (3.2) and the fact that \mathbf{r}({\mbox{\boldmath\pi}})={\mbox{\boldmath\pi}}^{\sigma}, we see that w_{\mbox{\boldmath\pi}} satisfies the defining relations of W({\mbox{\boldmath\pi}}^{\sigma}). Part (i) follows if we prove that W({\mbox{\boldmath\pi}})=\mathbf{U}(L^{\sigma}(\mathfrak{g}))w_{\mbox{\boldmath\pi}}. But this is true because proposition Proposition and proposition Lemma prove that there exists such that
[TABLE]
It now follows that V({\mbox{\boldmath\pi}})_{\mathfrak{g}_{\mathbf{a}^{m},N}} is the irreducible quotient of W({\mbox{\boldmath\pi}}^{\sigma}) and hence is isomorphic to V({\mbox{\boldmath\pi}}^{\sigma}) as –modules.
To prove (ii), note that that we have a surjective homomorphism of Lie algebras
[TABLE]
such that the restriction of to is just the canonical surjection. Moreover
[TABLE]
and hence
[TABLE]
Since \dim(W({\mbox{\boldmath\pi}}^{\sigma}))<\infty, it follows from Theorem 1(i) that W({\mbox{\boldmath\pi}}^{\sigma})_{\mathfrak{g}_{\mathbf{a},N}} is a quotient of W(\tilde{\mbox{\boldmath\pi}}) for some \tilde{\mbox{\boldmath\pi}}\in\cal P^{+}. Since the module W(\tilde{\mbox{\boldmath\pi}}) has an unique irreducible quotient V(\tilde{\mbox{\boldmath\pi}}), part (iii) follows if we prove that
[TABLE]
But this follows from part (i) and (4.4) and part (iii) is now proved. ∎
4.6.
The next proposition proves part (ii) of Theorem 2.
Proposition**.**
Let {\mbox{\boldmath\pi}}^{\sigma}=\prod_{k=1}^{\ell}\prod_{\epsilon=0}^{m-1}{\mbox{\boldmath\pi}}^{\sigma}_{\lambda_{k,\epsilon},\zeta^{\epsilon}a_{k}}\in\cal P^{+}_{\sigma} and assume that and have distinct coordinates. As –modules, we have
[TABLE]
Proof.
For , set
[TABLE]
It is checked easily that the element \otimes_{k=1}^{\ell}{w_{{\mbox{\boldmath\pi}}^{\sigma}_{k}}} satisfies the defining relations of w_{{\mbox{\boldmath\pi}}^{\sigma}} and hence we have an –module map,
[TABLE]
The proposition follows if we prove that this map is surjective. For then, taking {\mbox{\boldmath\pi}}\in\mathbf{i}({\mbox{\boldmath\pi}}^{\sigma}) and {\mbox{\boldmath\pi}}_{k}\in\mathbf{i}({\mbox{\boldmath\pi}}_{k}^{\sigma}), we have
[TABLE]
where the first equality uses part (i) of Theorem 2 and the last equality follows from Theorem 1(iii). To prove that is surjective choose so that W({\mbox{\boldmath\pi}}^{\sigma}) is a module for and also so that for all the algebra acts on W({\mbox{\boldmath\pi}}_{k}^{\sigma}) where and we have
[TABLE]
On the other hand by Proposition Proposition we have
[TABLE]
and hence \otimes_{k=1}^{\ell}W({\mbox{\boldmath\pi}}_{k}^{\sigma}) is cyclic for , i.e
[TABLE]
This proves that is a surjective map of –modules and the proof of the proposition is complete.
∎
4.7.
We now prove (iii) of Theorem 2. Recall that in Section 2.1, we have identified elements of with elements of and hence for each and we have elements {\mbox{\boldmath\pi}}_{\lambda,a}\in\cal P^{+} and {\mbox{\boldmath\pi}}_{\lambda,a}^{\sigma}\in\cal P^{+}_{\sigma}. Moreover, {\mbox{\boldmath\pi}}_{\lambda,a}\in\mathbf{i}({\mbox{\boldmath\pi}}_{\lambda,a}^{\sigma}).
Proof.
Choose , such that
[TABLE]
Using Lemma Lemma, Theorem 2(ii) and Theorem 2(i) in that order gives,
[TABLE]
Since , we have \prod{\mbox{\boldmath\pi}}_{\sigma^{\epsilon}(\lambda_{\epsilon}),\zeta^{\epsilon}b_{\epsilon}}\in\mathbf{i}(\prod{\mbox{\boldmath\pi}}^{\sigma}_{\lambda_{\epsilon},b_{\epsilon}}) and so by Theorem 2(i) we get,
[TABLE]
Theorem 1 gives,
[TABLE]
And since {\mbox{\boldmath\pi}}_{\sum_{\epsilon=0}^{m-1}\sigma^{\epsilon}(\lambda_{\epsilon}),a}\in\mathbf{i}(\prod_{\epsilon=0}^{m-1}{\mbox{\boldmath\pi}}^{\sigma}_{\lambda_{\epsilon},\zeta^{\epsilon}a}), we get
[TABLE]
which completes the proof.
∎
4.8.
We now prove Theorem 2 (iv). By Theorem 2(i), we have
[TABLE]
Theorem 1 gives if not of type that
[TABLE]
and for of type gives
[TABLE]
which completes the proof.
4.9.
We now prove Theorem 2(v). This part of the proof is very similar to the one given in [6] in the untwisted case and we shall only give a sketch of the proof. Thus, let be an –module, assume that is finite–dimensional and that it is generated by an element such that
[TABLE]
Let be such that for all . Since is finite–dimensional it follows from the representation theory of the subalgebras , that and also that
[TABLE]
Moreover if is of type , we find by working with the subalgebra that
[TABLE]
Applying to both sides of (4.5), ( if of type ) we find by using Lemma 3.3 (i), (ii), that
[TABLE]
while if is of type , we apply to both sides of (4.6) and using Lemma 3.3(iii), we find
[TABLE]
Set
[TABLE]
and let {\mbox{\boldmath\pi}}^{\sigma}=(\pi_{i}^{\sigma})_{i\in I_{0}}. The preceding arguments show that {\mbox{\boldmath\pi}}^{\sigma} is an –tuple of polynomials. We claim that
[TABLE]
which now shows that is a quotient of W({\mbox{\boldmath\pi}}^{\sigma}). To prove that \lambda=\lambda_{\mbox{\boldmath\pi}} is equivalent to proving that
[TABLE]
for all , if if is not of type and for all if is of type and if is of type
[TABLE]
It is now easy to see (keeping in mind that ) that the following Lemma implies (4.7).
Lemma**.**
*Let be a finite–dimensional –module and let be such that . For all ( for of type ), we have
[TABLE]
and for of type , we have
[TABLE]
Proof.
The proof of the first statement is given in [6, Proposition 1.1] and the key ingredient in that proof is Lemma 3.4 (i). The proof when and of type is entirely similar and one uses Lemma 3.4 (iii)(a) with .
∎
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] T. Akasaka An integral PBW basis of the quantum affine algebra of type A 2 ( 2 ) subscript superscript 𝐴 2 2 A^{(2)}_{2} Publ. Res. Inst. Math. Sci. 38, no. 4 (2002), 803 894
- 2[2] J. Beck and H. Nakajima Crystal bases and two-sided cells of quantum affine algebras , Duke Math. J. 123, no. 2 (2004), 335 402
- 3[3] V. Chari On the fermionic formula and the Kirillov-Reshetikhin conjecture , Internat. Math. Res. Notices 2001, no. 12, 629–654.
- 4[4] V. Chari and S. Loktev Weyl Demazure and fusion modules for the current algebra of 𝔰 𝔩 r + 1 𝔰 subscript 𝔩 𝑟 1 \mathfrak{sl}_{r+1} , Adv. Math. 207 (2006), no. 2, 928–960.
- 5[5] V. Chari and A. Pressley Twisted quantum affine algebras Commun. Math. Phys. 196, no. 2, pp. 461-476 (1998).
- 6[6] V. Chari and A. Pressley Weyl modules for classical and quantum affine algebras , Represent. Theory 5 (2001), 191–223.
- 7[7] B. Feigin and S. Loktev On Generalized Kostka Polynomials and the Quantum Verlinde Rule , Differential topology, infinite-dimensional Lie algebras, and applications, Amer. Math. Soc. Transl. Ser. 2 Vol. 194 (1999), p.61-79.
- 8[8] B. Feigin, A. N. Kirillov and S. Loktev Combinatorics and geometry of higher level Weyl modules , math.QA/0503315.
