CELL MODULES FOR TYPE AWEBS STUART MARTIN AND ROBERT A. SPENCER ABSTRACT .We examine the cell modules for the category of type Anwebs and their

2025-04-30 0 0 366.04KB 27 页 10玖币
侵权投诉
CELL MODULES FOR TYPE AWEBS
STUART MARTIN AND ROBERT A. SPENCER
ABSTRACT.
We examine the cell modules for the category of type
An
webs and their
natural cellular forms. We modify the bases of these modules, as described by Elias, to
obtain an orthogonal basis of each cell module. Hence, we calculate the determinant of the
Gram matrix with respect to such bases.
These Gram determinants are given in terms of intersection forms, computed from
certain traces of clasps — higher order Jones-Wenzl morphisms. Additionally, the modified
basis is constructed using these clasps, and each clasp is constructed using traces of smaller
clasps.
In [Eli15], Elias conjectures a value for these intersection forms and verifies it in types
A1,A2and A3. This paper concludes with a proof of the conjecture in type An.
1. INTRODUCTION
The category of (tensor products of) fundamental representations of
Uq(sln)
comes
equipped with many structural features. Indeed, it is monoidal, cellular and diagrammatic.
These three structures intersect and interact in a manner which permits us to use one to
deduce facts about the other.
The monoidal structure comes from the ability to take tensor product of representations
of
Uq(sln)
. This gives us the structure of a monoid on the representations. It is associative
and has an identity: the trivial representation
C
. However, for a general choice of
q
, it is
not symmetric: XYis not isomorphic to YX.
The category is cellular because it satisfies the conditions of being an OACC [EMTW20]
or, more simply, that each algebra
End(X)
is cellular in the sense of Graham and Lehrer
[GL96].
We say it is diagrammatic, due to the existence of a presentation by planar diagram
generators and relations [CKM14], known as webs or spiders.
Of the three, the cellular structure is perhaps the most interesting, as it contains the
most information. However, as we shall see, its cellular structure can be built up out of
monoidal generators for the objects and diagrammatic generators for the morphisms. This
is the interplay we shall exploit.
Cellular categories are equipped with cell modules. These can be thought of as the
cell modules of each of the cellular endomorphism algebras,
End(X)
. If the category is
semisimple, these cell modules are indecomposable. If it is not, they provide us with a
complete list of indecomposable objects as their heads.
To be exact, each module inherits a bilinear pairing from the homomorphism space
structure and the quotients of the cell modules by this radical (where the quotient is not
zero) are a complete set of indecomposable modules.
These categories also admit an integral lattice, and through this, their modular represen-
tation theory can be studied. For example, we may take the indecomposable cell modules
1
arXiv:2210.09639v2 [math.RT] 29 Jan 2023
2 STUART MARTIN AND R. A. SPENCER
over characteristic zero
1
and study their decompositions in positive characteristic. This
decomposition is controlled, to the first order, by the bilinear form on the cell modules.
For example, it gives a Jantzen-like filtration.
Since the integral lattice forms a basis of these modules, even under specialisation, we
can study certain properties of the bilinear form over semisimple characteristic in order to
deduce properties over positive characteristic. One such property is the determinant of
the form, which this paper investigates.
In a semisimple cellular category, each object with a weight has an associated idempotent
known as a clasp. Indeed, if
X
is an object of weight
λ
, there is a morphism
JWXEnd X
which is idempotent and which is killed by left- or right-composition by any morphism
factoring through an object of weight less than λ.
In this paper we will frequently refer to “simple” examples as both inspiration and
touchstones. The simplest example will be that of the Temperley-Lieb category,
TL
. In this
category (which manifests when studying
Uq(sl2)
) the clasps are known as Jones-Wenzl
elements. They all exist, which is to say the category is semisimple iff
q
is not a root of
unity.
The diagrammatic category equivalent to the category of fundamental representations of
Uq(sln)
is known as the category of webs or spiders. The concepts were introduced by
Kuperberg for the rank 2 Cartan types [Kup96]. Type Awebs were developed by Cautis,
Kamnitzer and Morrison using a variation on Howe duality [CKM14]. Thereafter Bodish
et. al. extended the construction to type Cn[BERT21].
These presentations are in terms of generators and relations — a quotient of a certain
category by a tensor ideal. This doesn’t describe the category completely though, and
the problem of determining a basis for the morphism spaces was started by Elias (in type
An
) [Eli15], and continued by Bodish for type
C2
[Bod20] and Bodish and Wu in type
G2[BW21].
Due to a very similar construction, it turns out that calculating clasps in the web
categories can be done nearly concurrently to finding bases. Indeed, we now have
constructions for clasps in type An[Eli15], C2[Bod21] and G2[BW21].
The familiar construction of Jones-Wenzl idempotents forms an important archetype
for these higher order clasps. Indeed, the Jones-Wenzl idempotents satisfy
(1.1) JWn+1
.
.
..
.
.=JWnJWn
.
.
..
.
..
.
.[n]
[n+1]JWn
.
.
..
.
..
Two features of this construction generalise. Firstly, clasps are inductively defined in terms
of smaller clasps. This is sometimes known as the “triple-clasp formula” (each of the two
morphisms in the right hand side of eq. (1.1) is hiding a further
JWn1
) and comes from
plethysm rules.
Secondly, the factor [n]/[n+1]comes from a tracing rule:
(1.2)
.
.
..
.
.
JWn=[n+1]
[n]
.
.
..
.
.
JWn1.
1In fact, the characteristic should be a ”semisimple mixed characteristic”.
CELL MODULES FOR TYPE AWEBS 3
Indeed, it is the reciprocal of the coefficient of the identity diagram. In the more general
case, this trace is called an intersection form and calculating their numerical values can be
difficuly. Though these have been calculated for types
An
for
n
4,
C2
and
G2
, in general
we must revert to a conjecture due to Elias for these constants.
In this paper we elucidate the connection between these intersection forms and the Gram
determinants of cell modules in the category. In particular we give expressions for all the
cell Gram determinants in terms of these intersection forms. We then go on to show that
Elias’ conjecture holds in type
A
, by finding pre-images of two distinguished bases across
the quantum skew Howe duality
While this paper concerns type
A
webs, it should be read with a more general as-
pect. Some results, such as lemma 3.1 hold in more general “object adapted cellular
categories” [Spe22]. These include web categories for other Coxeter types, and diagram-
matic Soergel bimodule categories. If such a category has a monoidal structure which
suitably interfaces with the category structure, then further constructions (such as those of
the ladder basis and triple-clasps) should generalise. However, a discussion of monoidal,
cellular categories is beyond the scope of this paper.
The remainder of this paper is laid out as follows. In section 2 we review the definition the
web category and its intersection forms, and introduce Elias’ conjecture for these forms.
Then, in section 3 we examine the cell module of this category and recount the ladder
basis. In section 4, new bases are found which give recurrant solutions to the determinant
of the cell module. We also determine an orthogonal basis. Finally, section 5 shows that a
previously-known basis of
Uq(glm)
is a pre-image of this orthogonal basis. Using this we
are able to confirm Elias’ conjectured formula for the intersection forms.
2. THE CATEGORY OF TYPE AWEBS
Let us recall the definition of type
A
webs. This work was initiated by Kuperberg [Kup96]
who derived the diagrammatic form for all rank two Lie type, and completed by Cautis,
Kamnitzer and Morrison who gave diagrammatics for general type An[CKM14].
The aim is to study the representation theory of
Uq(sln)
over
Q(q)
. To do this, it will
be sufficient to study the category of (tensor products of)
fundamental representations
of
Uq(sln)
, denoted
Fundn
. This category encodes the full category of finite dimensional
representations through its Karoubian envelope. The reader is directed to [Eli15,
§
3] for a
full description of the appropriate idempotents, known as clasps.
Remark 2.1.The reader may wish to overlook the quantum deformation on a first reading.
The appropriate mental notational substitutions are
Uq(sln)7sln,q7→ 1, [n]7n.
The results that hold for undeformed
sln
often also hold for
Uq(sln)
under the above
notational substitution (see for example the definition of a Jones-Wenzl idempotent using
quantum numbers). However, this is not always straightforward. For example, the
representation theory of
sln
occurs as a limit of the theory of
Uq(sln)
as
q
1. In this
paper, we make a point of keeping our results valid for the general case.
Recall that
Uq(sln)
has
n
1 fundamental representations, namely
{Vi:=ViCn}n1
i=1
.
Here
Cn
is given the standard action of
Uq(sln)
. In this way we can also consider
V0
=
4 STUART MARTIN AND R. A. SPENCER
Vn
=C
. If
i+j=k
then there are natural
Uq(sln)
-maps
ViVjVk
and
VkViVj
.
Note that these maps are not inverse.
The objects in the category
Fundn
are of the form
Vi1Vi2⊗ · · · ⊗ Vik
and the morphisms
are generated (through composition and tensor product) by the aforementioned maps
between
ViVj
and
Vk
. If
i=i1i2· · · ik
is a sequence of elements of
{
1,
. . .
,
n
1
}
we will
write Vifor Vi1Vi2⊗ · · · ⊗ Vik.
Remark 2.2.Strictly, we should define the objects to also permit tensor factors of the duals
of the
Vi
, but since
V
i
=Vni
what we have constructed is sufficient. In [Eli15] our
category is denoted by Fund+.
The composition of the map from
Vk
to
ViVj
and the map from
ViVj
to
Vk
again is
a multiple of the identity map on
Vk
. Indeed, let
[n]=(qnq(n1)/(qq1)
be the
n
-th
quantum number. Then if
[n]!= [n][n1]· · · [1]
and
[n
r]= [n]!/([r]![nr]!)
, the multiple
is [k
j]=[k
i].
Remark 2.3.Our analysis will actually hold over any pointed ring wherein
q
is not a root of
unity. In these characteristics, all the quantum numbers are invertable and so the necessary
objects will be well defined. When
q
is specialised to a root of unity, certain quantum
numbers and quantum binomial coefficients will vanish, and the required clasps will not
be well defined. See [MS21] for the correct idempotents in the A1case.
To provide an easier-to-work with category, we introduce the category of
webs
or
spiders
. This category denoted
Spn
is diagrammatic and is equivalent as a category to
Fundn. Thus we can think of these interchangeably.
The objects are taken from the set of finite words in letters
{
1,
. . .
,
n
1
}
and the object
i
in
Spn
corresponds to the object
Vi
in
Fundn
. As such, tensor product on words is simply
concatenation.
Morphisms are linear combinations of
diagrams2
. A diagram is an oriented, decorated,
planar graph with edges labelled by the letters
{
1,
. . .
,
n
1
}
. These graphs are taken up
to isotopy. Each nontrivial vertex is trivalent and if, when read left-to-right, one edge
splits into two, the sum of the outgoing labels equal that of the incoming. A similar result
holds for the reverse.
For example, the two classes of generating morphisms are represented as
(2.1) k
i
j
k
i
j
We take the precaution of removing all edges that are labelled with
n
or 0. Indeed, since
tensoring with
C
is a null operation, we lose no information on the object level. On the
morphism level, vertices with an nor 0 are turned into bivalent vertices (simply edges).
2
Technically we are working with the
ladder
incarnation of diagrams from [CKM14], however no generality
is lost
CELL MODULES FOR TYPE AWEBS 5
Example 2.1.
When
n=
2so we are examining the representation theory of
Uq(sl2)
, the
well known category to consider is the Temperley-Lieb category. Here, the object set is simply
{1}' {}'N.
Since
n=
2, there is only a single label for edges and the only allowed vertex is bivalent.
However, the statement that graphs are taken up to isotopy means that we recover the usual
definition of the Temperley-Lieb category as planar matchings of points.
Example 2.2.
When considering
Uq(sl3)
there are two objects which are dual to each other. We
will often consider the set
{+
,
}
instead of
{
1, 2
}
. Instead of labeling the edges with
{
1, 2
}
we will orient them with the understanding that bivalent vertices have one incident edge of each
orientation (and hence can be taken up to isotopy or ignored) and trivalent vertices are all sources
or sinks.
An example diagram from (+ + −−−+) to (+ + +) might be
+
+
+
+
+
+
There are relations on the morphisms in this category, as suggested by the comment
after remark 2.3 but their exact forms are not important for our study. They can be found
in [CKM14].
2.1.
Weights.
Each fundamental representation
Vi
has highest weight a fundamental
dominant weight
vi
. Viewed as a partition,
vi= (
1
i
, 0
ni)
. For any object
x
in
Fund
, let
wt(x)
be
ivxi
. If this is viewed as a
n+
1-part partition
(λ1
,
· · ·
,
λn
, 0
)
of
ixi
then
x
contains
λiλi+1
copies of
i
. The set of all weights, denoted
Λ
, is equipped with the
dominance partial order.
2.2.
Compatible Families.
Let
λΛ
and pick
ϕλ={ϕx,y:xy}
be a set of morphisms
between objects of weight
λ
. Let
J<λ
be the ideal spanned by morphisms that factor
through objects of weight less than λ. We say that ϕλis a compatible family if
for each pair of objects, xand y, of weight λ, there is a unique ϕx,y:xyin ϕλ,
ϕx,yϕz,x=ϕz,ymodulo J<λfor any three composable elements of ϕλ, and
ϕx,x=idxmodulo J<λ.
The two examples of compatible families we will meet are clasps and neutral ladders.
Lemma 2.3. There is a compatible family of morphisms such that ϕx,x=idxexactly.
These morphisms are known as (families of) neutral ladders.
Proof.
We present the construction here and signpost the reader to the proof that it suffices.
Set
ϕx,x=idx
. Now, suppose
x
and
y
differ in only two adjacent values, which is to say
x=x(0)· · · x(i1)x(i)x(i+1)x(i+2)· · · x(k)
摘要:

CELLMODULESFORTYPEAWEBSSTUARTMARTINANDROBERTA.SPENCERABSTRACT.WeexaminethecellmodulesforthecategoryoftypeAnwebsandtheirnaturalcellularforms.Wemodifythebasesofthesemodules,asdescribedbyElias,toobtainanorthogonalbasisofeachcellmodule.Hence,wecalculatethedeterminantoftheGrammatrixwithrespecttosuchbases...

收起<<
CELL MODULES FOR TYPE AWEBS STUART MARTIN AND ROBERT A. SPENCER ABSTRACT .We examine the cell modules for the category of type Anwebs and their.pdf

共27页,预览5页

还剩页未读, 继续阅读

声明:本站为文档C2C交易模式,即用户上传的文档直接被用户下载,本站只是中间服务平台,本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私,请立即通知玖贝云文库,我们立即给予删除!
分类:图书资源 价格:10玖币 属性:27 页 大小:366.04KB 格式:PDF 时间:2025-04-30

开通VIP享超值会员特权

  • 多端同步记录
  • 高速下载文档
  • 免费文档工具
  • 分享文档赚钱
  • 每日登录抽奖
  • 优质衍生服务
/ 27
客服
关注