GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE

SCHUR–WEYL DUALITY

FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

Abstract. We explicate relations among the Gelfand–Graev modules for central cov-
ers, the Euler–Poincaré polynomial of the Arnold–Brieskorn manifold, and the quantum
affine Schur–Weyl duality. These three objects and their relations are dictated by a
arXiv:2210.16138v1 [math.NT] 28 Oct 2022

permutation representation of the Weyl group.

Contents
1. Introduction 1
2. The Whittaker functor FWh 4
3. An Euler–Poincaré polynomial and stability 6
4. Quantum affine Schur–Weyl duality and FWh 9
5. Some remarks 19
References 19

1. Introduction
It might be appropriate to retitle the paper as “A short tale of a permutation repre-
sentation of the Weyl group”, since the main content concerns objects and their relations
as depicted in the following diagram:

f12 (indG
µn ×U − ι × ψ)
I

f02

(1.1) EP(MAB , X) W y Y /YQ,n f23

f01
f03

f13 FSW ,
where
(i) W y Y /YQ,n denotes the permutation representation
σX : W −→ Perm(XQ,n )
of the Weyl group W of a root datum Φ acting on X := XQ,n := Y /YQ,n with Y
the cocharacter lattice and YQ,n ⊂ Y a certain sublattice (see §2 for details);
(ii) EP(MAB , X) denotes the Euler–Poincaré polynomial of the Arnold–Brieskorn
manifold MAB (see [Arn69, Bri73]), the complement of the full Coxeter hyper-
plane arrangement of the root datum Φ;
2010 Mathematics Subject Classification. Primary 11F70; Secondary 22E50, 20G42.
Key words and phrases. covering groups, Iwahori–Hecke algebras, Gelfand–Graev representations,
Coxeter hyperplane arrangement, quantum affine Schur–Weyl duality.
1
2 FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

(iii) (indG I
µn ×U − ι × ψ) =: V is the Iwahori-component of the Gelfand–Graev represen-
tation of an n-fold central cover G of a connected reductive group G with root
datum Φ, see [GGK];
ˆ α )) is the quantum affine Schur–Weyl duality
(iv) FSW : ML (H(Sraff )) → ML (Uq (sl(n
functor of left modules, defined for the affine Hecke algebra H(Sraff ) of GLr and
ˆ α )), see [CP96, DDF12, Ant].
the quantum affine group Uq (sl(n
Now we elaborate on our main results relating the above objects. Note, some notation
is taken from [GGK]. We provide precise references when needed.
Henceforth, we fix a p-adic field F with prime ideal pF and ring of integers OF . We
have G := G(F ) for a smooth connected split reductive group scheme G over OF with
root datum Φ. Assume that F × contains the full group µn of n-th roots of unity and
gcd(p, n) = 1. For a maximal torus T ⊂ G let
Y := Hom(Gm , T)
denote the cocharacter lattice. Let W be the Weyl group of (G, T). Partially depending
on a W -invariant quadratic form
Q : Y −→ Z,
one has a central extension
(n) pr
µn G G.
(n)
For any H ⊂ G, let H = pr−1 (H). In particular, G = G . For every root α, we define
n
nα := ∈ N,
gcd(n, Q(α∨ ))
which appears in (iv) and only depends on the length of α.
Let U − ⊂ B − = T U − be the opposite unipotent radical of a fixed Borel subgroup
B = T U. Let
ψ : U − −→ C
be a non-degenerate character of conductor pF , i.e., ψ ◦ α∨ : F → C is of conductor pF
for every simple root α. The cover G splits canonically over U and U − , which are thus
viewed as subgroups of G.
We consider only ι-genuine representations of G, where µn acts via a faithful character
ι : µn ֒→ C× . Define K := G(OF ), and let I := B ∩ K be the Iwahori subgroup. We fix
a splitting sK : K ֒→ G and identify K and I with their image in G. This gives
∞
HI := Cc,ι (I\G/I),
the ι-genuine Iwahori–Hecke algebra. By Borel and Casselman, the map
π 7→ π I
gives a bijection between irreducible Iwahori-spherical representations of G and simple
modules over HI .
In [GGK], we determined the structure of V = (indG I
µn ×U − ι × ψ) as a right HI -module
for X -splitting covers (see Definition 2.1). In particular, the permutation representation
σX in (i) underlies this structure. We explain this in §2, which gives the relation f02 in
(1.1). Note, in this paper we use V, while we used V I in [GGK] for the same module.
For any algebra A, we denote by ML (A), MR (A) the category of left and right modules
over A respectively. Arising from the Gelfand–Graev module V is the Whittaker functor
FWh : MR (HI ) −→ MR (EndHI (V))
 GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE SCHUR–WEYL DUALITY 3

given by π 7→ HomHI (V, π ∨ ), where π ∨ denotes the contragredient of π. Naturally,

Whψ (π) := HomHI (V, π ∨ )
is called the Whittaker space of π. We show in Proposition 2.3 that if n is stable (see
Definition 2.2), then the functor FWh gives an equivalence of categories, and in particular
HI and EndHI (V) are Morita equivalent. For Kazhdan–Patterson covers and Savin covers
of GLr , stability is equivalent to the inequality nα > r, see Example 2.4. Such results
provide a partial understanding of the affine q-Schur algebra EndHI (V). In the linear case,
Bushnell–Henniart determine the structure of EndHI (V) for general Bernstein classes
[BH03, Theorem 4.3].
(n)
In §3, we determine the stability of n for covers G of semisimple simply-connected
G. To this end, we must compute the Whittaker dimension of the theta representation
(n)
dim Whψ (Θ(G )).
Using Sommers’ computation of the character of the permutation representation σX
[Som97], we show that this dimension is equal to
r
Y
−1 r −1 −1
|W | · n · EP(MAB , −n ) = |W | · (n − mj ),
j=1

where the mj ’s are the exponents of the Weyl group W . This dimension is also related to
the characteristic polynomial of the Coxeter hyperplane arrangment by results of Orlik–
Solomon [OS83]. We also give an analogous formula for the Steinberg representation
(n)
St(G ) in terms of EP(MAB , X). For details, see Theorem 3.1. This yields the link f01 .
Moreover, by an alternative characterization of stability, Theorem 3.1 enables us to
determine when HI ⊆ V as HI -modules. This provides the connection f12 in (1.1).
Recall that the space Whψ (π) is naturally a Z(G)-module, where Z(G) denotes the
center of G. Part of the goal of our paper is to show that for type A groups it is naturally
ˆ α ))-module via relating FWh to FSW . In §4, we recall the quantum affine Schur–
a Uq (sl(n
Weyl functor FSW for GLr which arises from the vector space VSW ≃ (C[Z])⊗r endowed
with commuting actions
ˆ
Uq (sl(m)) y VSW x H(Sraff )
for general m, r ∈ N. We highlight that the above action of H(Sraff ) on VSW depends on
ˆ
m sensitively, while the action of Uq (sl(m)) is a “standard” one (see §4.1). The presence
of σX in describing this action of H(Sraff ) on VSW is clear by a simple translation of
terminology from that in [Gre99, DDF12, Ant] for example, see the proof of Theorem 4.2.
This gives the link f03 .
(n)
For covers of GLr of type (C1) (see §2), one has
H(Sraff ) ≃ HI
and we make such an identification. Fix a certain character εHI : HI → C such that
εHI |HW equals the sign character εHW of HW (see §4.3). Our first result in §4 is Theorem
4.2, which contains two related statements. First,
V ≃ VSW ⊗C εHI
ˆ α )), H(S aff ))-bimodule, V inherits the left
as right HI -modules. Since VSW is a (Uq (sl(n r
ˆ
Uq (sl(nα ))-action from VSW , which might be referred to as a process of quantization.
Second, this gives us a precise relation between FSW and FWh .
4 FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

Following this, we use FSW to identify the local scattering matrix associated with
(n)
intertwining operators on genuine principal series I(χs ), s = (s1 , ..., sr ) ∈ Cr of GLr
with an R-matrix, up to a flipping map. See Theorem 4.6 and the subsequent discussion.
This result was already proved in [BBB19] by showing that the R-matrix (suitably
twisted) gives a solution to the Yang–Baxter equation of metaplectic ice model, and
it coincides with the local scattering matrix; thus, a natural map was defined from
Whψ (I(χs )I ) to the quantum space
V (nα s1 ) ⊗ ... ⊗ V (nα sr ),
where V (u) is the shifted standard evaluation Uq (sl(n ˆ α ))-module associated with any
u ∈ C, see Definition 4.1. The map is equivariant with respect to intertwiners and
homomorphisms associated with R-matrices.
Here we give a reproof of [BBB19, Theorem 1]. Indeed, the functor FSW provides a
natural substitute for the above map. Also, the functorial and monoidal properties of
FSW , coupled with its relation with FWh (Theorem 4.2), imply that the local scattering
ˆ α ))-homomorphism. This gives Theorem 4.6. The above
matrix represents a Uq (sl(n
results, especially the aforementioned “quantization”, give the connection f23 in (1.1).
In view of the relation between FSW and FWh , the polynomial EP(MAB , X) then
determines the faithfulness of FSW , see Remark 4.3. This gives the connection f13 , which
is in effect the composite of f12 and f23 .
In the last section §5, we add several remarks regarding existing generalizations of FSW
to general Cartan types in the literature. Inevitably, we are not able to give exhaustive
references on those deep topics, but can only content ourselves with a brief mention which
pertains to our discussions above.
1.1. Acknowledgement. We would like to thank Valentin Buciumas and Dennis Gaits-
gory for some communications on an earlier version of the paper. We are especially
grateful to Buciumas for very helpful comments and clarifications on the earlier work
relevant to the topics discussed here. The work of F. G. is partially supported by NSFC–
12171422.

2. The Whittaker functor FWh

(n)
In this section, let G = G be a Brylinski–Deligne cover of G. Recall the bilinear
form BQ (y, z) := Q(y + z) − Q(y) − Q(z) associated with Q and the W -stable sublattice
YQ,n = {y ∈ Y : BQ (y, z) ∈ nZ for all z ∈ Y } .
The Weyl group W acts naturally on XQ,n = Y /YQ,n.
(n)
Definition 2.1. A covering group G is X -splitting if the W -equivariant quotient map
f : Y ։ XQ,n
has a W -equivariant section.
Consider GLr with lattice Y given with the standard Z-basis {ei : 1 6 i 6 r}. A
(n)
Brylinski–Deligne cover GLr of GLr is associated with p, q ∈ Z such that
(
2p if i = j,
B(ei , ej ) =
q if i 6= j.
We have Q(α∨ ) = 2p − q. The Kazhdan–Patterson covers are those with Q(α∨ ) = −1,
where p corresponds to the parameter c in [KP84]. Meanwhile, Savin’s “nice” cover is
the one with p = −1, q = 0. These covers are X -splitting by [GGK, Example 6.5].
 GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE SCHUR–WEYL DUALITY 5

The following two classes of Brylinski–Deligne covers, which are also X -splitting cov-
ers, are our main interest in this paper.
(n)
(C1) A Brylinski–Deligne cover GLr satisfying
n|q
is called of type (C1). Such covers satisfy the block commutativity, i.e., for
partition r = (r1 , ..., rk ) of r with associated standard Levi subgroup Mr =
GLr1 × ... × GLrk , one has an isomorphism of groups
Y (n)

GLri /H → Mr ,
i
Q
where H = {(ζi )i : i ζi = 1}. Moreover, for such covers
YQ,n = nα Y with nα = n/ gcd(2p, n).
These covers are X -splitting by [GGK, Example 6.5].
As concrete examples, recall that the Savin covers are Brylinski–Deligne covers
with (p, q) = (−1, 0), and thus are of type (C1). On the other hand, for Kazhdan–
Patterson covers of type (C1) n must be odd.
(C2) Let G be semisimple and simply-connected. In [GGK, Corollary 6.4] we show that
all oasitic covers (see Defintion 6.1 of loc. cit.) of G are X -splitting. And in the
tables of [GGK, §6.1], we explicitly list the n’s such that an n-fold cover of such
G is oasitic.
Now assume G is an X -splitting cover. For every W -orbit Oy ⊂ XQ,n , one may view
Oy ⊂ Y such that
Wy := StabW (y, Y ) = StabW (y, X )
is a parabolic Weyl subgroup. Let
HWy ⊂ HW ⊂ HI
be the subalgebra associated with Wy and εWy be the sign character of HWy . In [GGK,
Theorem 5.18] we show that
M
(2.1) V= εWy ⊗HWy HI
Oy ⊂XQ,n

as a right HI -module. We write

VOy := εWy ⊗HWy HI .
Consider the Whittaker functor
FWh : MR (HI ) MR (EndHI (V)),
of right modules defined by
FWh (π) := Whψ (π) := HomHI (V, π ∨ ).
The action of a ∈ EndHI (V) on f ∈ HomHI (V, π ∨ ) is given by
(f · a) = f ◦ a.
Definition 2.2. Let G be an X -splitting n-fold cover. The degree n and also G are
called stable if the following equivalent conditions are satisfied:
(i) there is a free W -orbit in XQ,n ;
(ii) HI is a direct summand of V as HI -modules;
(iii) every theta representation Θ(G, χ) is generic, i.e., FWh (Θ(G, χ)) 6= 0.
6 FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

Here a theta representation Θ(G, χ) is the Langlands quotient of the genuine principal
series I(χ) associated with an “exceptional” genuine central character χ of Z(T ), see
[KP84, Gao17]. By (2.1), we have Oy ⊂ XQ,n is a free W -orbit if and only if VOy ≃ HI ,
and this immediately gives the equivalence between (i) and (ii). The equivalence of (i)
and (iii) is discussed in the proof of [GGK, Theorem 8.8]. It is easy to see that if n is big
enough, then it is stable since Definition 2.2 (i) is satisfied then.
(n)
Proposition 2.3. Let G be an X -splitting cover. If n is stable, then V is a progen-
erator of HI -mod. In this case, the functor FWh gives an equivalence of categories, and
thus HI and EndHI (V) are Morita-equivalent.
Proof. The HI -module V is a progenerator if V is projective, finitely generated, and is a
generator (see [Roc09, §1.5.1]). The module V is projective by [GGK, Lemma 8.6]. Since
V is projective, V is a generator if and only if for any nonzero σ ∈ MR (HI ), the space
HomHI (V, σ) 6= 0.
Since n is stable, this follows from Definition 2.2 (ii). Clearly, V is finitely-generated in
view of its structure given in (2.1).
Since V is a progenerator, it follows from general category theory (see [Roc09, §1.5.1])
that FWh gives an equivalence of categories. This completes the proof. 
Example 2.4. Let GLr be any X -splitting cover, for example of Kazhdan–Patterson or
Savin type, or of type (C1). It then follows from Lemma 3.1 and Theorem 3.7 of [GT22]
that n is stable if and only if nα > r. In fact, for general G, the results in [GT22, §3.1–3.5,
Theorem 3.7] already give a lower bound for stable n.
(n)
Remark 2.5. If G is a stable X -splitting cover such that HI ≃ H(G, I), then the
Morita equivalence between HI and EndHI (V) also follows from the work of Vigneras
[Vig03]. For every subset J ⊂ ∆ of the simple roots ∆, let vJε ∈ H(G, I) (resp. vJ1 )
denote the idempotent corresponding to εHW (J ) (resp. 1). This gives
M 
S ε := EndH(G,I) vJε H(G, I) .
J⊂∆

Its analogue S 1 was studied in [Vig03, MS19]. For stable n, the algebra EndHI (V) and
S ε are Morita equivalent and thus might be both called affine q-Schur algebras, as is the
case for S 1 in [Vig03, MS19]. However, we note that they are not identical since VOy ,
which corresponds to certain vJε y H(G, I), has multiplicities in V in general; that is, it
is possible that Jy = Jy′ for Oy 6= Oy′ . Such multiplicities are given in Sommers’ work
[Som97] for oasitic covers of semisimple simply-connected G, which we will use in §3.
In type A, such multiplicities are also important in relating FWh to the quantum affine
Schur–Weyl functor, see §4 below.

3. An Euler–Poincaré polynomial and stability

In this section, we consider exclusively a semisimple simply-connected G and an oasitic
(n)
cover G of G, such that for any short coroot α∨ ,
Q(α∨ ) = −1.
∨
The unramified genuine exceptional characters χ form a torsor over Z(G ), which
(n)
is trivial in this case; thus, there is a unique theta representation Θ(G ), the unique
(n)
irreducible quotient of I(χ). The unique irreducible subrepresentation St(G ) of I(χ) is
the analogue of the Steinberg representation.
 GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE SCHUR–WEYL DUALITY 7

(n)
Using work of Sommers [Som97], we determine dim Whψ (Θ(G )) explicitly in terms
of the Euler–Poincaré polynomial of the Coxeter hyperplane arrangement of G. This
gives the stability condition for n as well.
For semisimple simply-connected G, one has Y = Y sc , which is Z-spanned by the
simple coroots. For each root α ∈ Φ, consider the hyperplane
Hα := {v ∈ Y ⊗ R : hv, αi = 0} .
We have the set
A := {Hα : α ∈ Φ}
of the full Coxeter hyperplane arrangements in Y ⊗ R. Let L = L(A) be the set of
intersections of hyperplanes in A. We consider
Y ⊗R∈L
by taking the empty intersection of elements in A. Elements in L are ordered, and we
write x < y if y ⊆ x. We also use x 6 y to represent x < y or x = y. The Mobius
function
µ : L × L −→ Z
is uniquely defined by requiring µ(x, x) = 1,
X
µ(z, y) = 0 if x < y,
z: x6z6y

and µ(x, y) = 0 otherwise. The characteristic polynomial of L is thus given by

X
ω(L, X) := µ(Y ⊗ R, x) · X dim x .
x∈L

On the other hand, consider the Arnold–Brieskorn complex manifold

[
MAB := Y ⊗ C − Hα ⊗ C
α∈Φ

and its Euler–Poincaré polynomial

X
EP(MAB , X) := dim H i (MAB , C) · X i .
i>0

(n)
Theorem 3.1. Let G be an oasitic cover of a semisimple simply-connected G such that
Q(α∨ ) = −1 for any long root α. Then one has
r
(n) ω(L, n) nr · EP(MAB , −n−1 ) −1
Y
(3.1) dim Whψ (Θ(G )) = = = |W | · (n − mj ),
|W | |W | j=1

where mj , 1 6 j 6 r are the exponents of the Weyl group W . Similarly,

(3.2)
r
(n) (−1)r · ω(L, −n) nr · EP(MAB , n−1 ) −1
Y
dim Whψ (St(G )) = = = |W | · (n + mj ).
|W | |W | j=1

Proof. Before we begin, we note that we cite results from [GGK] that include the as-
sumption Q(α∨ ) = 1, for short coroots. When G is semisimple, these results also hold
when Q(α∨ ) = −1, for short coroots.
First, note that we have
Y r
r −1
ω(L, X) = X · EP(MAB , −X ) = (X − mj ) ∈ C[X],
j=1
8 FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

where the first equality was a classical result of Orlik–Solomon [OS83] and the second
equality already follows from work of Brieskorn [Bri73, Théorèm 6].
It follows from [GGK, Theorem 8.8] that
(n)
dim Whψ (Θ(G )) = hεW , σX iW = # {free W -orbits in XQ,n } .
Let P1 , P2 , ..., Pk ⊂ W be the conjugacy classes of parabolic subgroups of W such that
P1 = W, Pk = {1} .
(n)
Note that XQ,n = Y sc /nY sc . Since G is an oasitic cover (see [GGK, §6.1]), we see that
n is “very good” in the terminology of [Som97, Definition 3.5]. Following this, we have
from [Som97, Lemma 4.2] that
k
M
σX = Pi (1Pi )
mi (n) · IndW
i=1

for well-defined mi (n) ∈ Z>0 . Furthermore, it is shown in [Som97, Proposition 5.1] that
ω(L, n)
mk (n) = .
|W |
This gives
k
X ω(L, n)
hεW , σX iW = mi (n) · hεPi , 1Pi i = mk (n) = .
i=1
|W |
(n)
Now we consider the Steinberg representation St(G ). Again by [GGK, Theorem 8.8],
(n)
dim Whψ (St(G )) = h1W , σX iW = # {all W -orbits in XQ,n } .
For every w ∈ W , let s(w) denote the least number of reflections whose product is w and
consider
d(w) := dim(Y ⊗ R)w ,
the dimension of the set of fixed points of w in Y ⊗ R. One has d(w) = r − s(w). Let
χσX denote the character of σX . It was shown in [Som97, Proposition 3.9] that
χσX (w) = nd(w)
for every w ∈ W . We also note that
εW (w) = (−1)s(w) .
Thus,
(n) 1 X d(w)
dim Whψ (St(G )) = n
|W | w∈W
(−1)r X
= εW (w) · (−n)d(w)
|W | w∈W
(−1)r
= · ω(L, −n)
|W |
Yr
= |W |−1 · (n + mj ).
j=1

This completes the proof. 

 GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE SCHUR–WEYL DUALITY 9

The above recovers those explicit formulas for Sp2r and G2 computed in [Gao17, §8.2–
8.3]. Applying n = 1 in (3.2), since St(G) Q
is generic with unique Whittaker model, we
get the classical Weyl order formula |W | = rj=1 (1 + mj ), see [Sol66, Corollary 2.3].
Now assume that m1 ≤ m2 ≤ . . . ≤ mr .
(n)
Corollary 3.2. Let G be as in Theorem 3.1. Then n is stable if and only if n > mr ,
in which case
FWh : MR (HI ) −→ MR (EndHI (V))
gives an equivalence of categories.
Proof. If n > mr , then n is stable by Theorem 3.1. Conversely, the result follows by
comparing [GGK, Tables 1, 2], with the exponents of the associated Weyl group. We
include the details for a group of type E8 . The other cases are similar.
By [GGK, Table 2], if G is an oasitic cover of type E8 , then n is not divisible by 2, 3,
or 5. The exponents of the E8 Weyl group are 1, 7, 11, 13, 17, 19, 23, 29 (e.g. see [Bou02]).
These are exactly the positive integers less than 30 that are not divisible by 2, 3, or 5.
Since n is stable it follows that n > 29, as desired. 
Remark 3.3. We note that Theorem 3.1 is compatible with Example 2.4, since mr = r
(n)
for type Ar groups. For GLr , it is possible to compute χσX similarly as in [Som97] and
thus obtain a formula for dim Whψ (Θ(GLr , χ)) and dim Whψ (St(GLr , χ)), where χ is an
exceptional character. Alternatively, one can argue as follows in a special case. For this
(n)
purpose, we consider a Kazhdan–Patterson cover GLr satisfying gcd(n, r) = 1. The
(n) (n) (n)
covering subgroup SLr is thus an oasitic cover. In this case, the pair (GLr , SLr ) is
an isotypic pair in the sense of [GSSb, Definition 2.23]. In particular, one has
(n) (n)
(3.3) dim Whψ (St(GLr , χ)) = |Y /(Y sc + YQ,n )| · dim Whψ (St(SLr , χ0 ))
(n)
for the unique exceptional character χ0 for SLr obtained from the restriction of χ. In
fact, (3.3) holds for any irreducible constituent (in particular, Θ(GLr , χ) here as well)
of a regular unramified principal series of GLr , see [GSSb, §4.3.1]. Combining (3.3) and
(3.2), we get
r−1  
n 1 Y 1 r+n−1
dim Whψ (St(GLr , χ)) = · · (n + j) = · ,
|YQ,n/nY | r! j=1 |YQ,n /nY | r

where |YQ,n/nY | = gcd(n, 2rp + r − 1). We note that this formula can also be obtained
from [Zou, Theorem 4.7], which gives the Whittaker dimensions of general square inte-
grable representations of the Kazhdan–Patterson covers.

4. Quantum affine Schur–Weyl duality and FWh

In this section, we consider exclusively a cover G = GLr of type (C1). The goal is to
show that
FWh : MR (HI ) −→ MR (EndHI (V))
gives rise to the quantum affine Schur–Weyl functor
ˆ α ))
FSW : ML (H(Sraff )) → ML (Uq (sl(n
ˆ α ))-modules. Here
valued in the category of Uq (sl(n
Sraff = Y ⋊ Sr = Zr ⋊ Sr
10 FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

is the extended affine Weyl group of GLr . A first link between FWh and FSW is that one
has naturally
(4.1) HI ≃ H(Sraff )
for a type (C1) cover of GLr , see [Sav88, McN12, GG18].
Let C[Z] be the C-vector space with a basis given by {vi : i ∈ Z}. Let
VSW := (C[Z])⊗r .
The functor FSW arises from commuting (Uq (sl(nˆ α )), H(S aff ))-actions on the left and
r
right hand side of VSW . We briefly describe the actions below, where some details are
referred to [CP96, Ant].
ˆ
4.1. Left Uq (sl(m))-action ˆ
on VSW . We first recall the quantum affine group Uq (sl(m))
for any m ∈ N given as follows. First, let α be the highest root of the root system of
†

type A. Write
α0 = −α† .
Let 

CˆA := αi , αj∨ 06i,j6m−1
be the generalized Cartan matrix of type Ar . For convenience, we write


ai,j = αi , αj∨ .
ˆ
Now, Uq (sl(m)) is the quantized affine Lie algebra associated with CˆA . More precisely, it
is the C-algebra with generators

Ei , Fi , Ki±1 : 0 6 i 6 m − 1
and relations (see [CP96, §2.4] and [Ant, §2.5])
(R1) Ki Ki−1 = 1 = Ki−1 Ki and Ki Kj = Kj Ki .
(R2) Ki Ej = q ai,j · Ej Ki and Ki Fj = q −ai,j · Fj Ki .
(R3) [Ei , Fj ] = δi,j · (Ki − Ki−1 )/(q − q −1 ).
(R4)
Ei2 Ei±1 + Ei±1 Ei2 = (q + q −1 )Ei Ei±1 Ei
Fi2 Fi±1 + Fi±1 Fi2 = (q + q −1 )Fi Fi±1 Ei
Ei Ej = Ej Ei if i − j 6= 0, 1, m − 1
Fi Fj = Fj Fi if i − j 6= 0, 1, m − 1.
The computation of i ± 1 and i − j above are all valued in {0, 1, ..., m − 1}, the
fixed set of representatives of Z/mZ.
In terms of the basis {vi : i ∈ Z} of C[Z], one has an action of Uq (sl(m))
ˆ on C[Z] given
by the following:
Ei · vj = δi+1,j · vj−1
Fi · vj = δi,j · vj+1
Ki · vj = q δi,j −δi+1,j · vj .
Again, the computation of i + 1 and j ± 1 in the subscripts of the δ-functions above are
valued in the set {0, 1, ..., m − 1} of representatives for Z/mZ.
ˆ
To have an action of Uq (sl(m)) on VSW = (C[Z])⊗r , we note that the quantum group
ˆ
Uq (sl(m)) has a comultiplication map
ˆ
∆ : Uq (sl(m)) ˆ
Uq (sl(m)) ˆ
⊗C Uq (sl(m))
 GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE SCHUR–WEYL DUALITY 11

given by
Ei 7→ Ei ⊗ Ki−1 + 1 ⊗ Ei , Fi 7→ Fi ⊗ 1 + Ki ⊗ Fi , and Ki 7→ Ki ⊗ Ki .
This gives a well-defined representation
ˆ
∆r−1 : Uq (sl(m)) −→ EndC ((C[Z])⊗r ),
ˆ
equivalently, a left action Uq (sl(m)) y VSW .
Definition 4.1. For every s ∈ C, the subspace
Sm,s := SpanC {vi − q s vi+m : i ∈ Z} ⊂ C[Z]
ˆ
is Uq (sl(m))-stable, and we call V (s) := C[Z]/Sm,s the shifted evaluation representation
ˆ
of Uq (sl(m)) associated with s.
ˆ
Here V (s) is exactly the Uq (sl(m))-module V (q s ) in [CP96, §2.4], for a = q s there.
Recall the standard evaluation representation Vev (b) := (Vq♮ )qb , b ∈ C, where the latter
is in the notation of [CP95, Example 12.3.17]. Then it follows from loc. cit. that
V (s) = Vev (2/(m + 1) + s), whence called the shifted evaluation representation.
4.2. Right H(Sraff )-action on VSW and FSW . There is a canonical map
v : Y ֒→ C[Y ], y 7→ vy ,
and similarly
ṽ : Y /mY ֒→ C[Y /mY ] = (Cm )⊗r , y 7→ ṽy .
In terms of the standard Z-basis {ei : 1 6 i 6 r} of Y , one has
C[Y ] = C[e±1 ±1 ±1
1 , e2 , ..., er ].

Clearly,
( r
)
X
(4.2) Rm := ij · ej : 0 6 ij 6 m − 1
j=1

is a set of representatives of Y /mY . For every integer ij ∈ [0, m − 1] we write

ṽij := ṽ(ij · ej ).
Thus, a basis of C[Y /mY ] = (Cm )⊗r is given by
Bm = {ṽi1 ⊗ ṽi2 ⊗ ... ⊗ ṽir : 0 6 ij 6 m − 1 for every j} .
One has the C-vector space isomorphism
τm : (Cm )⊗r ⊗C C[Y ] −→ VSW
given by
(ṽi1 ⊗ ... ⊗ ṽir ) ⊗ (ek11 · ... · ekr r ) 7→ vi1 −k1 m ⊗ ... ⊗ vir −kr m .
Let (γ, C[Y /mY ]) be any HW -module afforded on the space C[Y /mY ]. From the
Bernstein presentation,
H(Sraff ) = HW ⊗C C[Y ].
Thus as C-vector spaces
VSW ≃ (Cm )⊗r ⊗C C[Y ] ≃ (Cm )⊗r ⊗γ,HW H(Sraff ).
This gives a right action of H(Sraff ) on VSW depending on the choice of γ.
ˆ
If γ is chosen arbitrarily, then the left Uq (sl(m))-action and right H(Sraff )-action on
VSW may not commute. To obtain a commuting action, γ should be “quantized” from
12 FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

the permutation action of Sr on (Cm )⊗r . We describe this special action denoted by γm
♮

as follows.
Following notations of [Gre99, DDF12, Ant], let
Λ(m, r)
denote the set of all λ := (λ1 , λ2 , ..., λm ) ∈ Zm>0 satisfying
X
λi = r.
i=1

For each λ ∈ Λ(m, r), consider

yλ = (1, 1, ..., 1, 2, 2, ..., 2, ...., m, m, ..., m) ∈ (Z/mZ)r .
| {z } | {z } | {z }
λ1 λ2 λm

It is clear that the stabilizer subgroup of yλ in W = Sr is

Wλ := Wyλ = Sλ1 × ... × Sλm .
Then one has a C-vector isomorphism
M
(4.3) 1 ⊗HWλ HW −→ (Cm )⊗r .
λ∈Λ(m,r)

This gives the sought after HW -module

♮
(γm , (Cm )⊗r )
with HW -action transported via (4.3). We also obtain the right action
VSW x H(Sraff )
♮
arising from γm .
It is known (see [Ant, Corollary 2.55]) that the above left and right actions on VSW
commute, which we write as
(4.4) ˆ
Uq (sl(m)) y VSW x H(Sraff ).
This commuting action gives rise to the so-called quantum affine Schur–Weyl functor
ˆ
FSW : ML (H(Sraff )) −→ ML (Uq (sl(m)))
given by
π 7→ VSW ⊗H(Sraff ) π =: FSW (π).
The functor FSW can be viewed as the composite Fϕ ◦ F1 , where
F1 : ML (H(Sraff )) −→ ML (EndH(Sraff ) (VSW )), π 7→ VSW ⊗H(Sraff ) π.
On the other hand, (4.4) also gives an algebra homomorphism
ˆ
ϕ : Uq (sl(m)) −→ EndH(Sraff ) (VSW ),
which is surjective if m > r, see [DG07, Theorem 3.2.1] or [DDF12, Theorem 3.8.3]. Here
ϕ induces
ˆ
Fϕ : ML (EndH(S aff ) (VSW )) −→ ML (Uq (sl(m))).
r

It is clear that FSW = Fϕ ◦ F1 .

 GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE SCHUR–WEYL DUALITY 13

4.3. Comparison of FWh and FSW . We first revert to the discussion of the Gelfand–
(n)
Graev module V for a cover GLr of type (C1). We write
M
VX := εWy ⊗HWy HW ,
Oy ⊂XQ,n

which is a right HW -module, then it is clear from (2.1) that

V ≃ VX ⊗HW HI .
(n)
For these GLr , one has YQ,n = nα Y . We set HI ≃ HW ⊗C C[YQ,n ] as C-vector spaces
and define the modified extended affine Weyl group to be
W̃ex := YQ,n ⋊ W.
Recall that there is the modified coroot lattice
sc
YQ,n = nα · Y sc ⊂ YQ,n
which gives the modified affine Weyl group
sc
W̃aff := YQ,n ⋊ W ⊂ W̃ex .
Associated with W̃aff is a subalgebra HW̃aff ⊂ HI . In fact, there is a subgroup Ω ⊂ W̃ex
such that
HI ≃ HW̃aff ⊗C C[Ω],
where the twisted algebra multiplication of the right hand side is given in [IM65, Propo-
sition 3.8] or [Sol21, §1.4]. Let
εHI : HI −→ C
be the unique character which extends the sign character εW̃aff of HW̃aff and also
εHI |C[YQ,n] = Ct− , with t− = (q (1−r)/2 , q (3−r)/2 , ..., q (r−1)/2 ),
see [Sol21, §2.3].
For any rings A, B and a functor
F : ML (A) −→ ML (B)
of left modules over these rings, we denote by
F o : MR (Ao ) −→ MR (B o )
the naturally associated functor of right modules over the opposite rings Ao , B o of A and
B. That is, for any σ ∈ ML (A) one has
F o (σ o ) := F (σ)o ,
where (−)o denotes the corresponding module over the opposite ring.
Let H be one of the Hecke algebras H(Sraff ), HW̃aff , or HI . There is a lattice Z such that
H = HW ⊗ C[Z] as vector spaces. The algebra H is isomorphic to its opposite algebra
in view of the star involution
(−)⋆ : H −→ H, Tw⋆ = Tw−1 , θz⋆ = θz
for any Tw ∈ HW and θz ∈ C[Z] (see [BC, §2.3]). This gives an isomorphism of algebras
I ⋆ : Ho −→ H
given by I ⋆ (x) = x⋆ . Thus, if A = H is an affine Hecke algebra, then using I ⋆ we will
view F and F o both to be defined from the category of (left and right) modules over H.
14 FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

(n)
Theorem 4.2. Let G = GLr be a cover of GLr of type (C1). Then one has an isomor-
phism
VX ≃ γn♮ α ⊗C εW
of HW -modules and thus
V ≃ VSW ⊗C εHI
as HI -modules, where we make the identification HI ≃ H(Sraff ) as in (4.1). Moreover,
o
(4.5) FSW (π ⊗C εHI ) = Fϕo ◦ FWh (π ∨ ) and FSW
o
(π) = Fϕo ◦ FWh (π ∨ ⊗C εHI )
for every π ∈ MR (HI ).
Proof. We have
M M
γn♮ α ⊗C εW = εWλ ⊗HWλ HW and VX = εWy ⊗HWy HW .
λ∈Λ(nα ,r) Oy ⊂XQ,n

Since XQ,n = Y /nα Y , it is easy to see that the following map is a well-defined bijection.
Λ(nα , r) −→ {W -orbits in XQ,n } , λ 7→ W (yλ )
This immediately implies VX ≃ γn♮ α ⊗C εW and also V ≃ VSW ⊗ εHI .
Recall that we view F1o as defined on MR (HI ) via I ⋆ . For any π ∈ MR (HI ) we have
F1o (π) = (VSW ⊗HI π o )o = π ⊗HI VSW
o
,
o
where VSW is the (HI , (EndHI VSW )o )-module opposite to VSW . For every σ ∈ MR (HI )
we set
σ ∗ := HomHI (σ, HI ),
which is also an (HI , EndHI (σ))-bimodule. Now
FWh (π) = HomHI (V, π ∨ ) = π ∨ ⊗HI V ∗
where the last equality follows from the fact that V is finite-generated projective HI -
module, see [Bou98, Page 271]. Note
M M
V∗ = HomHI (εWy ⊗HWy HI , HI ) = HomHWy (εWy , HI ).
Oy ⊂XQ,n Oy ⊂XQ,n

Again, since εWy is finitely-generated projective HWy -module, we have

HomHWy (εWy , HI ) = HI ⊗HWy ε∗Wy ,
where ε∗Wy = HomHWy (εWy , HWy ) as left HWy -module. Since the algebra HWy is semisim-
ple and contains εWy with multiplicity one, we have dimC ε∗Wy = 1. Moreover,
εWy ≃ ε∨Wy ≃ ε∗Wy ,
as HWy -modules, where the middle denotes the dual representation. Hence,
M M
(4.6) V∗ ≃ Vo ≃ HI ⊗HWy εWy , VSW ∗
≃ VSWo
≃ HI ⊗HWy 1Wy
Oy ⊂XQ,n Oy ⊂XQ,n

as left HI -modules.
The above also gives that
(EndHI VSW )o ≃ (EndHI V)o ≃ (V ⊗HI V ∗ )o ≃ (V ∗ )o ⊗HI V o ≃ EndHI V = EndHI VSW ,
where the fourth isomorphism follows from (4.6) and the rest are canonical. With this
∗ o
identification, one can also check easily that VSW ≃ VSW as right EndHI (VSW )-modules.
Thus, for every π ∈ MR (HI ) we have
FWh (π) = (π ∨ ⊗ εHI ) ⊗HI VSW
∗
= F1o (π ∨ ⊗ εHI )
 GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE SCHUR–WEYL DUALITY 15

as right EndHI (VSW )-modules. Since FSW = Fϕ ◦ F1 , the first equality (4.5) follows,
which also gives the second equality. 
It is immediate from Theorem 4.2 that for every Iwahori-spherical representation π one
has
o
dim Whψ (π ∨ ) = dim FSW (π I ⊗ εHI ).
Moreover, if nα > r, then FSW is a faithful functor (see [Ant, Theorem 2.60]) and this
gives that dim Whψ (π) > 0 for every π in this case. We mention that the above equality
and the faithfulness for nα > r were first observed by V. Buciumas. In fact, he already
noticed that one can relate Whittaker models of such π to quantum affine representations
by using the work of Chari–Pressley [CP96].
ˆ α )). In this case, one can
Remark 4.3. We may consider Uq (ĝl(nα )) instead of Uq (sl(n
define the natural left Uq (ĝl(nα )) action on VSW which still commutes with the right
HI -action. It then gives rise to the quantum affine Schur–Weyl functor
′
FSW : ML (HI ) −→ ML (Uq (ĝl(nα ))),
which is faithful if nα > r. This faithfulness can be obtained as follows. First, ϕ′ :
Uq (ĝl(nα )) ։ EndHI (VSW ) is surjective if nα > 2, see [DDF12, Theorem 3.8.1]. On the
other hand, if nα > r, then FWh gives an equivalence of categories from Proposition 2.3
′,o
and Example 2.4. Since FSW (π ⊗ εHI ) = Fϕo′ ◦ FWh (π ∨ ) as in (4.5) still holds, we see that
′
FSW is faithful for nα > r.
4.4. Local scattering matrices and R-matrices. Henceforth, we write
ˆ α )))
F ϕ = F o ◦ FWh : MR (HI ) −→ MR (Uq (sl(n
Wh ϕ

for simplicity of notations.

(n)
First, we consider the principal series representation of GLr of type (C1). We have
T ≃ (T 1 × ... × T r )/H,
n o
ζj = 1 and T j ⊂ T is the covering torus associated to Yj := Zej .
Q
where H = (ζj )j : j

One has Z(T j ) = (Tj )nα , which splits naturally over (Tj )nα by the section s in the notation
of [GGK, §2.3]. Thus,
Z(T j ) ≃s µn × (Tj )nα
and this gives the Weyl-invariant unramified genuine character ωs,j of Z(T j ) trivial on
(Tj )nα . Consider the character χsj : Tj → C× given by χsj (aej ) = |a|sj . This gives
i(χsj ) := i(ωs,j ) ⊗ χsj ∈ Irr(T j ),
where i(ωs,j ) is uniquely determined by ωs,j by the Stone-von Neumann theorem. For
s = (s1 , ..., s2 ) ∈ Cr , we have
i(χs ) := ⊗j i(χsj ) ∈ Irr(T )
and thus the principal series representation I(χs ) := IndG
B (i(χs )). We can identify natu-
rally I(χs ) ≃ I(χ−s ).
∨

Since
o
FSW ˆ α )))
: MR (HI ) −→ MR (Uq (sl(n
is a monoidal functor (see [CP96, Proposition 4.8]), one has
(4.7) o
FSW (I(χs )I ) ≃ FSW
o
(i(χs1 )I1 ) ⊗ ... ⊗ FSW
o
(i(χsr )Ir ),
where
Ij = I ∩ Tj = Tj (OF ).
16 FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

In fact, we can directly verify (4.7) as follows.

Proposition 4.4. For every 1 6 j 6 r one has
o
(4.8) FSW (i(χsj )Ij ) ≃ V (nα sj ),
where V (nα sj ) is the shifted evaluation representation in Definition 4.1. Moreover,
(4.9) o
FSW (I(χs )I ) ≃ V (nα s1 ) ⊗ ... ⊗ V (nα sj ) ⊗ ... ⊗ V (nα sr ).
Proof. Write Yj = Zej , YQ,n,j = Znα ej and XQ,n,j = Yj /YQ,n,j . Note that
o
FSW (i(χsj )Ij ) = i(χsj )Ij ⊗HIj VSW
o
,
where HIj = C[YQ,n,j ] = C[Znα ej ] and VSW
o
= HIj ⊗C C[XQ,n,j ]. Also, in this case
i(χsj )Ij = Csj ,
the one-dimensional unramified subspace of i(χsj ) affording the character
χsj : C[YQ,n,j ] → C.
o
It is then easy to check that FSW (i(χsj )Ij ) ≃ V (nα sj ).
For the second isomorphism, we note I(χs )I = i(χs )I∩T ⊗C[YQ,n ] HI and thus
o
FSW (I(χs )I ) = I(χs )I ⊗HI VSW
∗

≃ i(χs )I∩T ⊗C[YQ,n ] (VSW

∗
|C[YQ,n] )
O r
≃ i(χsj )Tj ⊗HIj (HIj ⊗ C[XQ,n,j ])
j=1
Or
o
≃ FSW (i(χsj )Ij ).
j=1

This gives (4.7). Coupled with (4.8), we obtain (4.9). 

Corollary 4.5. The unramified principal series I(χs ) is reducible if and only if q =
q nα (si −sj ) for some i 6= j.
Proof. This follows immediately from combining Proposition 4.4 and [CP96, §4.8 Corol-
lary]. In fact, more directly, since HI ≃ H(Sraff ) for GLr of type (C1), the points of
reducibility of I(χs ) are exactly those of its Shimura lifted principal series I(χs ) of GLr
modulo an nα -scaling, where χs is a linear character of T ⊂ GLr corresponding to χs . 
In view of the Theorem 4.2, the isomorphism in (4.7) or (4.9) is an incarnation of
Rodier’s heredity of Whittaker models [Rod73]. We make this precise as follows. First,
one has a natural C-isomorphism
η(χs ) : Whψ (I(χ−s )I ⊗ εHI ) −→ Whψ (I(χ−s )I ).
Indeed, one has
Whψ (I(χ−s )I ⊗ εHI ) = HomHI (V, I(χs )I ⊗ ε∨HI ) = HomHW (VX , I(χs )I ⊗ εW ).
As HW -modules, I(χs )I ≃ HW for any principal series. Thus, one has a natural identifi-
cation I(χs )I ⊗ εW ≃ I(χs )I as HW -modules and
Whψ (I(χ−s )I ⊗ εHI ) = HomHW (VX , I(χs )I ) ≃ Whψ (I(χ−s )I ).
This gives the above map η(χs ). For every j one has the natural C-isomorphisms
(4.10) C[Zej /Znα ej ] ≃ FWh
ϕ
(i(χ−sj )Ij ) ≃ FSW
0
(i(χsj )Ij ) ≃ V (nα sj ).
 GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE SCHUR–WEYL DUALITY 17

Also, the same argument as in the proof of Proposition 4.4 gives the C-isomorphisms
Whψ (I(χ−s )I ) ≃ I(χs )I ⊗HI V ∗
Or
≃ i(χsj )Tj ⊗HIj (HIj ⊗ C[XQ,n,j ])
(4.11) j=1
Or r
O
ϕ
≃ FWh (i(χ−sj )Ij ) ≃ V (nα sj ),
j=1 j=1

which can be referred to as the Rodier’s heredity. It is easy to see that the diagram
Nr
j=1 V (nα sj )
(4.12)
η(χs )
o
FSW (I(χs )I ) ϕ
FWh (I(χ−s )I )

of vector spaces commutes. Here the left and right slanted arrows are given by (4.9) and
(4.11) respectively.
For every simple root β = βk with 1 6 k 6 r − 1, we have the simple reflection
wβ = (k, k + 1) ∈ Sr associated with β. Consider the intertwining map T (wβ ) : I(χs ) −→
I(wβ χs ) as in McNamara [McN12, Equation (7.2)], where they are written Twβ . Passing
to I-fixed vectors gives an HI -module map I(χs )I −→ I(wβ χs )I , which we still write as
T (wβ ). Note wβ χs = χwβ (s) and further wβ acts on s = (s1 , ..., sj , ..., sr ) via that on the
index j. Consider the map
T ♯ (wβ ) : I(wβ χ−s )I −→ I(χ−s )I .
We then have the following commutative diagram arising from (4.12):

Nr
j=1 V (nα sj )

F(wβ )∗
o
FSW (I(χs )I ) η(χs )
ϕ
FWh (I(χ−s )I )
(4.13)
Nr
j=1 V (nα swβ (j) )
o (T (w ))
FSW β T ♯ (wβ )∗

o
FSW (I(wβ χs )I ) wβ
ϕ
FWh (I(wβ χ−s )I ).
η( χ s)

o
Here, the front square diagram, and in particular the two maps FSW (T (wβ )) and T ♯ (wβ )∗
♯ ∗
are induced from T (wβ ) and T (wβ ). The map F(wβ ) naturally arises from the front
o
square in view of (4.12). In fact, by the functorial and monoidal property of FSW , one
has (noting that wβ = (k, k + 1) ∈ Sr )
F(wβ )∗ = id ⊗ ... ⊗ F(wβ ) ⊗ ... ⊗ id
where
F(wβ ) : V (nα sk ) ⊗ V (nα sk+1 ) −→ V (nα sk+1 ) ⊗ V (nα sk )
ˆ α ))-homomorphism.
is a Uq (sl(n
18 FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

It is clear from (4.13) that F(wβ )∗ can be represented by any matrix form of T ♯ (wβ )∗ .
We write
s∗ = −wβ (s) = (−s1 , −s2 , ..., −sk+1 , −sk , ..., −sr ) ∈ Cr .
The operator T ♯ (wβ )∗ is represented by a so-called local scattering square matrix
[S(y ′ , y; wβ , χs∗ )]y′ ,y∈Rnα ,
of size |XQ,n |, where Rnα ⊂ Y is the set of representatives of XQ,n in (4.2). This
matrix was first studied by Kazhdan–Patterson [KP84] for GLr andPthen generalized by
McNamara [McN16] for general G. More precisely, for every y = j yj ej ∈ Rnα with
yj ∈ [0, nα − 1], one has
y ∈ Whψ (I(χ−s ) ),
I
λ−s
the naturally associated element via (4.10) and (4.11). The matrix [S(y ′ , y; wβ , χs∗ )] is
then by definition the one satisfying
X ∗
(4.14) T ♯ (wβ )∗ (λ−s
y ′ ) = S(y ′ , y; wβ , χs∗ ) · λsy .
y∈Rnα

j s
Note
P that we also have uyj ∈ V (nα sj ) for every yj as above, using (4.10). Hence, for
y = j yj ej ∈ Rnα , one has

usy := usy11 ⊗ ... ⊗ usyjj ⊗ ... ⊗ usyrr ∈ V (nα s1 ) ⊗ ... ⊗ V (nα sr ),

which corresponds to λ−s

y in the right top slanted arrow in (4.13). Similar correspondence
s∗ −s∗
holds for λy and uy . The commutativity of (4.13) and the discussion above immediately
give the following.

Theorem 4.6. Let the domain and codomain ˆ α ))-homomorphism F(wβ )∗ be

of the Uq (sl(n
 s  ∗
endowed with the bases uy′ : y ′ ∈ Rnα and uy−s : y ∈ Rnα respectively. Then F(wβ )∗
is represented by the local scattering matrix [S(y ′ , y; wβ , χs∗ )]y′ ,y∈Rnα .
Consider the matrix
[M(y ′, y; wβ , χs∗ )] := [S(wβ (y ′), y; wβ , χs∗ )]y,y′ ∈Rnα
and let
R(wβ , χs∗ ) ∈ EndC (V (nα s1 ) ⊗ ... ⊗ V (nα sr ))
be the C-endomorphism
represented by [M(y ′, y; wβ , χs∗ )]y,y′ ∈Rnα with respect to the
basis usy : y ∈ Rnα . It is then easy to see that

F(wβ )∗ = τ (wβ ) ◦ R(wβ , χs∗ )

where τ (wβ ) is the obvious C-homomorphism induced from the flipping map
V (nα sk ) ⊗ V (nα sk+1 ) −→ V (nα sk+1 ) ⊗ V (nα sk ), v ⊗ w 7→ w ⊗ v.
As a consequence of (4.13), the F(wβ )∗ ’s satisfy the braid relation. Thus, the opera-
tors R(wβ , χs∗ ) solve the quantum Yang–Baxter equation, and the representing matrices
[M(y ′, y; wβ , χs∗ )] can be properly called R-matrices. One can check easily that it agrees
with the one given in [BBB19, Page 105]. We also note that the matrix [M(y ′ , y; wβ , χs∗ )]
is exactly a “local coefficients matrix” associated with I(χs∗ ) and wβ as in [GSSa, §3.2].
 GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE SCHUR–WEYL DUALITY 19

5. Some remarks
The functor FWh is defined for any root system type, while
ˆ α )))
FSW : ML (H(Sraff )) −→ ML (Uq (sl(n
is for type A only. It is a natural question to ask for a generalization of FSW , if we replace
the domain or codomain of FSW by more general root system types.
First, if one considers MR (Uq (ĝ)) for a general semisimple Lie algebra g, then it was
shown by [KKK18, Fuj20] that there exists a natural functor
g
M
FSW : ML (RJ (β)) −→ ML (Uq (ĝ)),
β

where RJ (β) is certain quiver algebra given by Khovanov–Lauda [KL09] and Rouquier
g
[Rou]. Many results are established regarding FSW , especially those pertaining to the
g
categorification of the two sides of FSW .
On the other hand, instead of H(Sraff ) in FSW above, if one considers affine Hecke
algebras of general Cartan type, for example of type B or C, then the recent work of
W.-Q. Wang, H.-C. Bao and their collaborators [BKLW18, BWW18, FLL+ 20] have made
advances in the framework of quantum symmetric pairs. In particular, a Schur–Weyl type
duality between affine Hecke algebras of type Br or Cr and certain coideal subalgebra of
Uq (gl(k)) is established. We refer the reader to loc. cit. and references therein for details.
In the geometric setting, the work of Gaitsgory and Lysenko [Gai08, GL18, Lys17]
represents some significant progress towards understanding certain Whittaker category
in the metaplectic setting and also its relation to the quantum groups in the framework
of the quantum Langlands program (see [Gai]). It will be very interesting to see what
would be the analogue of such results in the classical p-adic context, and conversely what
role the quantum affine Schur–Weyl functor plays in the formulation of some geometric
results.

References
[Ant] Jonas Antor, Affine versions of Schur–Weyl duality, Master’s Thesis, 2020, available at
https://www.math.uni-bonn.de/ag/stroppel/Masterarbeit_Antor-4-1.pdf.
[Arn69] V. I. Arnol′d, The cohomology ring of the group of dyed braids, Mat. Zametki 5 (1969), 227–231
(Russian). MR242196
[BKLW18] Huanchen Bao, Jonathan Kujawa, Yiqiang Li, and Weiqiang Wang, Geometric Schur duality
of classical type, Transform. Groups 23 (2018), no. 2, 329–389, DOI 10.1007/s00031-017-9447-
4. MR3805209
[BWW18] Huanchen Bao, Weiqiang Wang, and Hideya Watanabe, Multiparameter quantum Schur dual-
ity of type B, Proc. Amer. Math. Soc. 146 (2018), no. 8, 3203–3216, DOI 10.1090/proc/13749.
MR3803649
[BC] Dan Barbasch and Dan Ciubotaru, Star operations for affine Hecke algebras, preprint, available
at https://arxiv.org/abs/1504.04361v1.
[Bou98] Nicolas Bourbaki, Algebra I. Chapters 1–3, Elements of Mathematics (Berlin), Springer-Verlag,
Berlin, 1998. Translated from the French; Reprint of the 1989 English translation [ MR0979982
(90d:00002)]. MR1727844
[Bou02] , Lie groups and Lie algebras. Chapters 4–6, Elements of Mathematics (Berlin),
Springer-Verlag, Berlin, 2002. Translated from the 1968 French original by Andrew Pressley.
[Bri73] Egbert Brieskorn, Sur les groupes de tresses [d’après V. I. Arnol′d], Séminaire Bourbaki, 24ème
année (1971/1972), Exp. No. 401, Springer, Berlin, 1973, pp. 21–44. Lecture Notes in Math.,
Vol. 317 (French). MR0422674
[BBB19] Ben Brubaker, Valentin Buciumas, and Daniel Bump, A Yang-Baxter equation for
metaplectic ice, Comm. Number Theory and Physics 13 (2019), no. 1, 101–148, DOI
10.4310/CNTP.2019.v13.n1.a4.
20 FAN GAO, NADYA GUREVICH, AND EDMUND KARASIEWICZ

[BBBF18] Ben Brubaker, Valentin Buciumas, Daniel Bump, and Solomon Friedberg, Hecke modules from
metaplectic ice, Selecta Math. (N.S.) 24 (2018), no. 3, 2523–2570, DOI 10.1007/s00029-017-
0372-0. MR3816510
[BH03] Colin J. Bushnell and Guy Henniart, Generalized Whittaker models and the Bernstein center,
Amer. J. Math. 125 (2003), no. 3, 513–547. MR1981032
[CP95] Vyjayanthi Chari and Andrew Pressley, A guide to quantum groups, Cambridge University
Press, Cambridge, 1995. Corrected reprint of the 1994 original. MR1358358
[CP96] , Quantum affine algebras and affine Hecke algebras, Pacific J. Math. 174 (1996), no. 2,
295–326. MR1405590
[DDF12] Bangming Deng, Jie Du, and Qiang Fu, A double Hall algebra approach to affine quantum
Schur-Weyl theory, London Mathematical Society Lecture Note Series, vol. 401, Cambridge
University Press, Cambridge, 2012. MR3113018
[DG07] S. R. Doty and R. M. Green, Presenting affine q-Schur algebras, Math. Z. 256 (2007), no. 2,
311–345, DOI 10.1007/s00209-006-0076-1. MR2289877
[FLL+ 20] Zhaobing Fan, Chun-Ju Lai, Yiqiang Li, Li Luo, Weiqiang Wang, and Hideya Watanabe,
Quantum Schur duality of affine type C with three parameters, Math. Res. Lett. 27 (2020),
no. 1, 79–114. MR4088809
[Fuj20] Ryo Fujita, Geometric realization of Dynkin quiver type quantum affine Schur-Weyl duality,
Int. Math. Res. Not. IMRN 22 (2020), 8353–8386, DOI 10.1093/imrn/rny226. MR4216691
[Gai08] Dennis Gaitsgory, Twisted Whittaker model and factorizable sheaves, Selecta Math. (N.S.) 13
(2008), no. 4, 617–659, DOI 10.1007/s00029-008-0053-0. MR2403306
[Gai] , Quantum Langlands Correspondence, available at https://arXiv.1601.05279.
[GL18] Dennis Gaitsgory and Sergey Lysenko, Parameters and duality for the metaplectic geometric
Langlands theory, Selecta Math. (N.S.) 24 (2018), no. 1, 227–301, DOI 10.1007/s00029-017-
0360-4. MR3769731
[GG18] Wee Teck Gan and Fan Gao, The Langlands-Weissman program for Brylinski-Deligne exten-
sions, Astérisque 398 (2018), 187–275 (English, with English and French summaries). L-groups
and the Langlands program for covering groups. MR3802419
[Gao17] Fan Gao, Distinguished theta representations for certain covering groups, Pacific J. Math. 290
(2017), no. 2, 333–379, DOI 10.2140/pjm.2017.290.333.
[GGK] Fan Gao, Nadya Gurevich, and Edmund Karasiewicz, Genuine pro-p Iwahori–Hecke algebras,
Gelfand–Graev representations, and some applications, preprint (2022, 63 pages), available at
https://arxiv.org/abs/2204.13053.
[GSSa] Fan Gao, Freydoon Shahidi, and Dani Szpruch, Local coefficients and gamma factors for
principal series of covering groups, Memoirs of the AMS (2019, accepted), available at
https://arxiv.org/abs/1902.02686.
[GSSb] , Restrictions, L-parameters, and local coefficients for genuine representations, preprint
(2021, submitted), available at https://arxiv.org/abs/2102.08859.
[GT22] Fan Gao and Wan-Yu Tsai, On the wavefront sets associated with theta representations, Math.
Z. 301 (2022), no. 1, 1–40, DOI 10.1007/s00209-021-02894-5. MR4405642
[Gre99] R. M. Green, The affine q-Schur algebra, J. Algebra 215 (1999), no. 2, 379–411, DOI
10.1006/jabr.1998.7753. MR1686197
[IM65] N. Iwahori and H. Matsumoto, On some Bruhat decomposition and the structure of the Hecke
rings of p-adic Chevalley groups, Inst. Hautes Études Sci. Publ. Math. 25 (1965), 5–48.
MR185016
[KKK18] Seok-Jin Kang, Masaki Kashiwara, and Myungho Kim, Symmetric quiver Hecke algebras
and R-matrices of quantum affine algebras, Invent. Math. 211 (2018), no. 2, 591–685, DOI
10.1007/s00222-017-0754-0. MR3748315
[KP84] D. A. Kazhdan and S. J. Patterson, Metaplectic forms, Inst. Hautes Études Sci. Publ. Math.
59 (1984), 35–142. MR743816
[KL09] Mikhail Khovanov and Aaron D. Lauda, A diagrammatic approach to categorification of quan-
tum groups. I, Represent. Theory 13 (2009), 309–347, DOI 10.1090/S1088-4165-09-00346-X.
MR2525917
[Lys17] Sergey Lysenko, Twisted Whittaker models for metaplectic groups, Geom. Funct. Anal. 27
(2017), no. 2, 289–372, DOI 10.1007/s00039-017-0403-1. MR3626614
[McN12] Peter J. McNamara, Principal series representations of metaplectic groups over local fields,
Multiple Dirichlet series, L-functions and automorphic forms, Progr. Math., vol. 300,
 GENUINE WHITTAKER FUNCTOR AND THE QUANTUM AFFINE SCHUR–WEYL DUALITY 21

Birkhäuser/Springer, New York, 2012, pp. 299–327, DOI 10.1007/978-0-8176-8334-413.

MR2963537
[McN16] , The metaplectic Casselman-Shalika formula, Trans. Amer. Math. Soc. 368 (2016),
no. 4, 2913–2937, DOI 10.1090/tran/6597. MR3449262
[MS19] Vanessa Miemietz and Catharina Stroppel, Affine quiver Schur algebras and p-adic GLn ,
Selecta Math. (N.S.) 25 (2019), no. 2, Paper No. 32, 66, DOI 10.1007/s00029-019-0474-y.
MR3948934
[OS83] Peter Orlik and Louis Solomon, Coxeter arrangements, Singularities, Part 2 (Arcata, Calif.,
1981), Proc. Sympos. Pure Math., vol. 40, Amer. Math. Soc., Providence, RI, 1983, pp. 269–
291. MR713255
[Roc09] Alan Roche, The Bernstein decomposition and the Bernstein centre, Ottawa lectures on ad-
missible representations of reductive p-adic groups, Fields Inst. Monogr., vol. 26, Amer. Math.
Soc., Providence, RI, 2009, pp. 3–52, DOI 10.1090/fim/026/01. MR2508719
[Rod73] François Rodier, Whittaker models for admissible representations of reductive p-adic split
groups, Harmonic analysis on homogeneous spaces (Proc. Sympos. Pure Math., Vol. XXVI,
Williams Coll., Williamstown, Mass., 1972), Amer. Math. Soc., Providence, R.I., 1973, pp. 425–
430. MR0354942
[Rou] Raphael Rouquier, 2-Kac-Moddy algebras, available at https://arxiv.org/abs/0812.5023.
[Sav88] Gordan Savin, Local Shimura correspondence, Math. Ann. 280 (1988), no. 2, 185–190, DOI
10.1007/BF01456050. MR929534
[Sol21] Maarten Solleveld, Affine Hecke algebras and their representations, Indag. Math. (N.S.) 32
(2021), no. 5, 1005–1082, DOI 10.1016/j.indag.2021.01.005. MR4310011
[Sol66] Louis Solomon, The orders of the finite Chevalley groups, J. Algebra 3 (1966), 376–393, DOI
10.1016/0021-8693(66)90007-X. MR199275
[Som97] Eric Sommers, A family of affine Weyl group representations, Transform. Groups 2 (1997),
no. 4, 375–390, DOI 10.1007/BF01234541. MR1486037
[Vig03] Marie-France Vignéras, Schur algebras of reductive p-adic groups. I, Duke Math. J. 116 (2003),
no. 1, 35–75, DOI 10.1215/S0012-7094-03-11612-9. MR1950479
[Zou] Jiandi Zou, Metaplectic correspondence and applications, preprint, available at
https://arxiv.org/abs/2209.05852v1.

F. Gao: School of Mathematical Sciences, Zhejiang University, 866 Yuhangtang Road,

Hangzhou, China 310058
Email address: gaofan@zju.edu.cn

N. Gurevich: Department of Mathematics, Ben Gurion University of the Negev, Be’er

Sheva, Israel 8410501
Email address: ngur@math.bgu.ac.il

E. Karasiewicz: Department of Mathematics, University of Utah, Salt Lake City,

USA 84112
Email address: karasiewicz@math.utah.edu