SYMMETRIES OF GROTHENDIECK RINGS IN REPRESENTATION

THEORY

DAVID HERNANDEZ
arXiv:2501.03024v1 [math.RT] 6 Jan 2025

Abstract. This is a written version of the invited lecture at the 9th European Congress
of Mathematics in July 2024 in Sevilla. We review certain new symmetries of Grothendieck
rings that have emerged in representation theory.

Contents
1. Grothendieck group 2
2. Monoidal categories 3
3. Grothendieck ring 4
4. Cluster algebra symmetries 6
5. Shifted quantum affine algebras 10
6. Symmetry of q-characters 13
References 15

Categories of representations of groups, or more generally, bialgebras and quantum

groups, are endowed with monoidal structures. The Grothendieck ring of such a cate-
gory has a basis consisting of classes of simple objects, with structure constants defined
as multiplicities of simple representations in the decomposition of tensor products. The
resulting rings together with their bases can have a very intricate structure, and many of
them are fundamental with important applications (for example, leading to algorithms to
compute the dimension of simple representations).
In recent years, new symmetries of Grothendieck rings have emerged, which can be used
to better understand these rings. We will discuss these symmetries and illustrate some of
their applications.
First, cluster symmetries give powerful tools to investigate the structure of the Grothendieck
rings [HL1]. As an application, we have obtained the proof of long-standing conjectures on
character formulas for simple finite-dimensional representations of quantum affine algebras
(which are infinite-dimensional quantum groups) [FHOO].
Secondly, motivated by symmetries of quantum integrable models arising in physics, we
have established remarkable relations in associated Grothendieck rings [FH1], generalizing
the celebrated Baxter relations; these relations led us to the discovery of a surprising new
action of the Weyl group on these Grothendieck rings [FH2]. These Weyl group symmetries
are also closely related to the cluster symmetries [GHL].

1
2 DAVID HERNANDEZ

1. Grothendieck group
To start with, let us remind the general construction Grothendieck groups.
1.1. Grothendieck group of an additive category. Let C be an additive category with
distinguished short exact sequences
(1.1) A → B → C.
For example, one may consider the case of an abelian category with all ordinary exact
sequences.
Definition 1.1. The Grothendieck group K(C) of C is the group generated by the elements
[M ] for each isomorphism class of objects M in C, with the relations
[A] + [C] = [B]
for each exact sequence (1.1).
We obtain the important algebraic structure K(C) attached to the category C. We will
discuss many examples in this lecture.
The main original motivation appeared in algebraic geometry. Consider the C category
of coherent sheaves on an algebraic variety X. The Grothendieck group
K(X) = K(C)
is one of the starting points of K-theory in algebraic geometry. Historically, such construc-
tions were first studied by Grothendieck for categories of coherent and locally free sheaves
on schemes, in proving the Riemann–Roch theorem.
Another basic but fundamental example is obtained for a field k : the category C = V of
finite-dimensional k-vector spaces. The isomorphism classes are parametrized by the dimen-
sion. Then V is semi-simple and the direct sum descends to the sum in the Grothendieck
group, that is for V and W in V we have
[V ⊕ W ] = [V ] + [W ].
This implies that K(V) is freely generated by the class [k] of the one-dimensional vector
space
K(V) = Z[k] ≃ Z
as we have
[V ] = dim(V ).[k]
for V in V.
1.2. Grothendieck groups in representation theory. We now discuss the Grothendieck
group of finite-dimensional representations of an algebra.
Let A be an algebra over a field k and F be the category of finite-dimensional represen-
tations of A.
We have the following classical and fundamental result.
Theorem 1.2. [Jordan-Hölder series] Each object M in F admits a series of subobjects
M = M0 ⊃ M1 ⊃ · · · ⊃ MN = {0}
with Mi /Mi+1 simple object.
The number nV (M ) of occurrence of a simple object V (as a quotient Mi /Mi+1 ) depends
only on M , and not on the series of subobjects.
 SYMMETRIES OF GROTHENDIECK RINGS IN REPRESENTATION THEORY 3

Such a series is called a composition series or a Jordan-Hölder series of the representation.

A simple quotient Mi /Mi+1 is called a composition factor of M , and its number of occurrence
is called its multiplicity.
For completeness, let us briefly remind the arguments of the proof.
Proof. The existence of the series is obtained by induction on the dimension of the repre-
sentation. The uniqueness of nV (M ) is also proved by induction on the dimension : it is
obvious for a simple representation. In general, assume that we have two series Mi and Mi′ as
above. If M1 = M1′ , we can conclude using the induction hypothesis for this representation.
Otherwise
M ⊃ M1 ⊃ M1 ∩ M1′
and
M ⊃ M1′ ⊃ M1 ∩ M1′
are the first few terms of composition series with the same composition factors. Using the
induction hypothesis, they have the same composition factors. 
One can then define the length of a module M as the sum
X
l(M ) = nV (M ).
V simple class in F

As a consequence of Theorem 1.2, we obtain that

M
K(F) = Z[V ]
V simple class

has a canonical basis parametrized by simple classes. The Grothendieck group is a group
with a distinguished basis.
The coefficient on the basis are the multiplies of simple representations, that is for M
object in F, one has X
[M ] = nV (M )[V ].
V simple class
In general, an element X
nV [V ]
V simple class
with all nV ≥ 0 is called the class of an actual representation.
If one of the nV < 0, this is the class of a virtual representation. The Grothendieck group
allows to treat in a uniform way actual and virtual representations.

2. Monoidal categories
We now assume that an abelian category C has a monoidal structure, that is we have
tensor products M ⊗ N for M, N objects in C with isomorphisms
(M ⊗ N ) ⊗ P ≃ M ⊗ (N ⊗ P )
as well as a unit object 1 in C so that for M object in C
M ⊗ 1 ≃ 1 ⊗ M ≃ M.
We have additional compatibility properties with the abelian structure (see [ML]).
Remark 2.1. We do not assume that the category is necessarily a tensor category, that is
we do not assume the category admits dualities compatible with the monoidal structure.
4 DAVID HERNANDEZ

Let us discuss important examples of monoidal categories.

1) The category V of finite-dimensional k-vector spaces has a natural monoidal structure.
2) For a group G, the category of finite-dimensional representations of G (that is of
its group-algebra k[G]) has a monoidal structure. The action of g ∈ G on a pure tensor
v ⊗ w ∈ V ⊗ W in a tensor product of representations is defined by
g.(v ⊗ w) = (g.v) ⊗ (g.w).

3) For a Lie algebra g, the category of finite-dimensional representations of g (that is of

its universal enveloping algebra U(g)) has a monoidal structure. The action of g ∈ g on a
pure tensor v ⊗ w ∈ V ⊗ W in a tensor product of representations is defined by
g.(v ⊗ w) = (g.v) ⊗ w + v ⊗ (g.w).
This is an infinitesimal version of the group action above.
4) More generally : for a Hopf algebra H, that is an algebra with a coproduct and a
counit
∆ : H → H ⊗ H and ǫ : H → k.
The category of finite-dimensional representations of H has a monoidal structure. This
includes all examples above, and new examples, in particular the Drinfeld-Jimbo quantum
groups
Uq (g)
associated to q ∈ C∗ \ {−1, 1} quantum parameter and g simple Lie algebra. It is a q-
deformation of U(g), originated from mathematical physics. For example, Uq (sl2 ) is the
algebra by generators E, F , K ±1 and relations
K − K −1
KE = q 2 EK , KF = q −2 F K and [E, F ] = .
q − q −1

5) Hecke algebras, of their variations such as quiver Hecke algebras (as defined by
Khovanov-Lauda and Rouquier) quiver rise categories of representations which have a
monoidal structures via convolutions.
6) In various geometric contexts, convolution diagrams
p q
X1 ← X2 → X3
of regular morphisms p, q between remarkable varieties allow to consider convolution prod-
ucts and then monoidal structures on certain categories of sheaves (see for instance [CG, BR]
and references therein).

3. Grothendieck ring
3.1. Grothendieck ring of a monoidal category. Let us define the Grothendieck ring
of C monoidal category.
Proposition 3.1. The Grothendieck group K(C) of C inherits a ring structure such that
[M ].[N ] = [M ⊗ N ]
for M , N objects in C and the class [1] ∈ K(C) is the neutral element.
 SYMMETRIES OF GROTHENDIECK RINGS IN REPRESENTATION THEORY 5

Example 3.2. C = V the category of finite-dimensional vector spaces on a field k, then the
group isomorphism
K(C) ≃ Z
is a ring isomorphism. This follows for example from the formula dim(V ⊗W ) = dim(V )dim(W ).

Remark 3.3. If the category is braided, that is

M ⊗N ≃ N ⊗M

for all objects M, N in C, then the Grothendieck ring is commutative. The converse state-
ment is false : there are non braided monoidal categories whose Grothendieck ring is com-
mutative. We will see examples below.

3.2. Grothendieck rings in representation theory. Let C be the category of finite-

dimensional representations of a bialgebra algebra H. The Grothendieck ring
M
K(C) = Z[V ]
V simple

has a canonical basis ([V ])V simple , and the constant structures on this basis are positive con-
stant structures : they are given by the multiplicities of simple modules in tensor products
(sometimes called Clebsch-Gordan coefficients or fusion coefficients).

Remark 3.4. If H is a Hopf algebra, C is a tensor category, and dualities induce ring
automorphisms of K(C), which preserve the canonical basis.

Example 3.5. let C be the category of finite-dimensional representations of the Lie algebra
sl2 (C). Then we have a ring isomorphism

K(C) ≃ Z[X]

with the polynomial ring in one variable X. Here

X = [V ] where V = C2 is the natural representation.

It is well-known that sl2 (C) has a simple complex representation of dimension n + 1 for any
n ≥ 0, and that this representation is unique up to isomorphism. Let us denote by Qn the
isomorphism class of this representation in K(C). Then the basis (Qn )n≥0 of simple classes
satisfies the induction relation

(3.2) Q2n = Qn+1 Qn−1 + 1.

One can compute the canonical basis of simple classes

(Q0 , Q1 , Q2 , · · · ) = (1, X, X 2 − 1, X 3 − 2X, · · · ),

which is different that the standard monomial basis of Z[X].

Remark 3.6. Note that, independently on Grothendieck rings, it is not a priori completely
that the inductive sequence (3.2) with initial conditions Q0 = 1 and Q1 = X, a priori in
Q(X), defines a sequence in the ring Z[X].
6 DAVID HERNANDEZ

3.3. Applications of Grothendieck rings in representation theory. The study of

study of Grothendieck rings of categories of representations lead to interesting developments
in various directions. Let us list some of them.

Representation Theory : there is close relation between the understanding of the global
structure of Grothendieck rings and the description of the structure of simple representa-
tions. For example, if one knows the structure of certain simple representations, one may
extract informations on the simple constituents of their tensor products.

Example 3.7. This approach will be discussed below to compute the characters of simple
representations of quantum affine algebras from the characters of fundamental representa-
tions, following [N1] and then [H1, FHOO].

Categorification : one can realize sometimes an algebra A as a Grothendieck ring:

A ≃ K(C).

The categorified structure A inherits a canonical basis from the canonical basis of the
Grothendieck ring.

Example 3.8. The coordinate ring C[N ], with N unipotent subgroup of a Lie group G,
can be realized from representations of quiver Hecke algebras [KL, R] or of quantum affine
algebras [HL2], leading to a categorical interpretation of its dual canonical basis. A basic
example is also explained in Example 3.5.

Quantum integrable models : certain Grothendieck rings can be identified with the ring of
commuting operators of a quantum integrable model, via the transfer-matrix construction.
Indeed, a quantum integrable model involves a space W (the quantum space) with a family
of commuting operators on this space (including quantum Hamiltonians). The transfer-
matrix construction defines a ring morphism

T : K0 (C) → End(W )

where C is the category of representation of an algebra of symmetry of the system. Hence,

any relation in K0 (C) (a ”universal relation”) gives relations between the eigenvalues of the
operators on V , for any related quantum integrable model.

Example 3.9. The famous XXZ-model can be realized using the representation theory of
the quantum affine algebra associated to sl2 (see below).

The aim of this lecture is to discuss certain symmetries of Grothendieck rings, and their
consequences. The main examples of symmetries we present are Cluster symmetries and
Weyl group symmetries.

4. Cluster algebra symmetries

We discuss the first symmetries of Grothendieck rings in this lecture, in the context of
cluster algebras.
 SYMMETRIES OF GROTHENDIECK RINGS IN REPRESENTATION THEORY 7

4.1. A quick review on cluster algebras. The theory of cluster algebras was introduced
by Fomin-Zelevinsky [FZ]. A cluster algebra is a commutative algebra with a distinguished
set of generators grouped into overlapping subsets (the clusters, obtained by an inductive
process called mutation). Each element of the cluster algebra can be expressed as a rational
fraction into the elements of a cluster. From a geometric point of view, one might think
about a variety with various sets of local coordinates, and each element of the algebra of
functions over this variety can be expressed locally in terms of the local coordinates. Then
the mutations can be understood as a regular change of coordinates.
More precisely, the cluster algebra AQ is a commutative algebra associated with a quiver
Q (without loops or 2-cycles). It has a distinguished set of generators (the cluster variables)
defined by a combinatorial process. Cluster algebras have many incarnations in various
fields, and in particular in representation theory as we will discuss below.
The cluster algebra AQ attached to the quiver Q (with set of vertices Q0 ) is a subring of
the field
F = Q(Xi )i∈Q0 .
with free variables Xi which are called the initial cluster variables (together with the initial
quiver Q, they form the initial cluster). The cluster algebra AQ is defined as the subalgebra
of F generated by the cluster variables, obtained inductively from the initial variables by
an inductive process called mutations. For example, the first step mutated variables Xi∗ are
defined by the formula : Y Y
Xi Xi∗ = Xj + Xj ,
j→i j←i
were the arrows → and ← are arrows in the initial quiver Q (the quiver gets also mutated
in the process). The number of cluster variables is not necessarily finite. In addition,
the cluster variables are grouped into overlapping subsets called clusters, which are all in
bijection with Q0 . The cluster monomials are defined as the monomials into the cluster
variables of the same cluster. In some situations, one may have additional non-mutable
cluster variables : they are called frozen variables and they belong to all clusters.
One of the fundamental properties of cluster algebras is the Laurent phenomenon : any
cluster variable can be expressed as a Laurent polynomial in the cluster variables of a given
seed. In the initial seed, this can be written as
AQ ⊂ Z[Xi±1 ]i∈Q0 .
Example 4.1. Consider a Q quiver of type A2 :
• −→ •
We have the initial cluster variables (X1 , X2 ), and five cluster variables :
1 + X2 1 + X1 1 + X1 + X2
X1 , X2 , , , .
X1 X2 X1 X2
There are also five clusters :
   
1 + X2 1 + X2 1 + X1 + X2
(X1 , X2 ), , X2 , , ,
X1 X1 X1 X2
   
1 + X1 1 + X1 + X2 1 + X1
, , , X1 .
X2 X1 X2 X2
Remark 4.2. This example should not be misleading : in general, the number of clusters
is not equal to the number of cluster variables.
8 DAVID HERNANDEZ

4.2. Monoidal categorification of cluster algebras. In this lecture we focus on the

relation between cluster algebras and monoidal categories (there are also important relations
with additive categories, see [K] and references therein).
Definition 4.3. [HL1] A monoidal category M is a monoidal categorification of a cluster
algebra A is there exists a ring isomorphism
φ : A → K(M)
so that the cluster monomials are sent to simple classes.
In [HL1], a stronger version is also discussed, where it is required that φ induces a
bijection between the cluster monomials and the real simple classes (whose tensor square is
also simple). In the definition above, the simple modules corresponding to cluster monomials
are called reachable modules. They have the property to be real, as the square of a cluster
monomial is a cluster monomial.
A monoidal categorification can be seen as a cluster symmetry of the Grothendieck ring,
or as a categorical realization of the cluster algebra. It gives useful informations on the
monoidal category : for example, it points out remarkable simple representations (corre-
sponding to cluster variables) and it gives a factorization of simple representations corre-
sponding to cluster monomials (in the same way as cluster monomials factorize as a product
of cluster variables).
The original examples [HL1] were obtained from the finite-dimensional representations
of quantum affine algebras. Many developments about monomial categorifications in this
context have followed, one can cite for instance [BC, KKKO, N2, Q], and different contexts
[CW] for instance.
In all known examples, the cluster symmetry is related to the defect of symmetry of the
category which produces relations in the Grothendieck ring : we have objects V , W so that
V ⊗ W not isomorphic to W ⊗ V
but
[V ⊗ W ] = [W ⊗ V ]
in the Grothendieck ring (examples will be given below).
4.3. Quantum affine algebras. Consider g a complex simple Lie algebra (for example,
g = sl2 ) of rank n (that is n = 1 for g = sl2 ). Then we can construct the affine Kac-Moody
algebra ĝ, which is a central extension of the loop algebra1
g ⊗ C[t, t−1 ].
For q ∈ C∗ a quantum parameter (which is assumed to not root of unity), we have the
corresponding quantum affine algebra : Uq (ĝ). It is a Hopf algebra and a q-deformation of
the universal enveloping algebra U(ĝ).
Let C be the category of finite-dimensional representations of Uq (ĝ) : it has a very
rich structure. This monoidal category if not semi-simple and not braided. The simple
objects of C have been parametrized in terms of Drinfeld polynomials by Chari-Pressley,
that is of n-tuples (Pi (z))1≤i≤n of monic polynomials. In particular, for 1 ≤ i ≤ n and
a ∈ C∗ , we have the fundamental representation Vi (a) which corresponds to the n-tuple
(1, · · · , 1, 1 − za, 1, · · · , 1) with a degree 1 polynomial in position i.
1We do not consider the derivation element in this lecture.
 SYMMETRIES OF GROTHENDIECK RINGS IN REPRESENTATION THEORY 9

Theorem 4.4. [FR] The Grothendieck ring K(C) is commutative and polynomial
K(C) ≃ Z[Xi,a ]1≤i≤n,a∈C∗ ,
with Xi,a = [Vi,a ] class of a fundamental representation.
4.4. Monoidal categorification - examples. As a first example, consider the complex
Lie algebra g = sl2 . We define M as the monoidal Serre subcategory of the category C of
ˆ 2 )-representations generated by two fundamental
finite-dimensional representations of Uq (sl
representations
V1 (1), V1 (q 2 ).
Then we obtain a monoidal categorification
K(M) ≃ AQ
with Q of type A2 (as above) with one frozen variable. The total number of cluster variables
is 3. Indeed, there is a simple representation W in M of dimension 3 so that we have exact
sequences
0 → 1 → V1 (q 2 ) ⊗ V1 (1) → W → 0,

0 → W → V1 (1) ⊗ V1 (q 2 ) → 1 → 0,
which come from degenerated braidings (which are not isomorphisms as the two tensor
products are not isomorphic) :
V1 (q 2 ) ⊗ V1 (1) → V1 (1) ⊗ V1 (q 2 ) and V1 (1) ⊗ V1 (q 2 ) → V1 (q 2 ) ⊗ V1 (1).
These exact sequences correspond to the unique mutation relation in the cluster algebra
AQ :
[V1 (q 2 )][V1 (1)] = 1 + [W ].
The two clusters are (V1 (q 2 ), W ) and (V1 (1), W ). As an application, we obtain that every
simple representations in M can be factorized into these 3 representations. Note that there
are infinitely many simple classes in this category M, which have the following form with
a, b ≥ 0 :
(V1 (1))⊗a ⊗ W b and (V1 (q 2 ))⊗a ⊗ W b .
As another example, consider the complex Lie algebra g = sl3 . We define M as the
ˆ 3 )-representations
monoidal Serre subcategory of the category C of finite-dimensional Uq (sl
generated by the four fundamental representations
V1 (1), V1 (q 2 ), V2 (q), V2 (q 3 ).
Then we obtain a monoidal categorification
K(M) ≃ AQ
with Q of type A2 (as above) with two additional frozen variables. The total number of
cluster variables is 7.
As an application, we obtain that every simple representation in M can be factorized into
these 7 representations. Note that there are infinitely many simple classes in this category
M.
10 DAVID HERNANDEZ

4.5. Global structures. For a general simple Lie algebra g, a cluster algebra structure
has been obtained on K(C− ) for C− a large subcategory2 of C [HL3] . Then Kashiwara-
Kim-Oh-Park established the following conjectured in [HL3].
Theorem 4.5. [KKOP] The category C− of finite-dimensional representations of a quantum
affine algebra Uq (ĝ) is a monoidal categorification of a cluster algebra.
As an application, we could prove for simple representations corresponding to cluster
monomials, a Conjecture that I formulated in 2004 [H1] for non simply-laced types. Using
cluster algebra methods, the result extends results obtained by Nakajima [N1] in simply
laced-types.
Theorem 4.6. [FHOO] The dimension (and character) of simple representations corre-
sponding to cluster monomials can be obtained from an algorithm (à la Kazhdan-Lusztig).
Remark 4.7. The algorithm is partly based on the existence of a quantization of the
Grothendieck rings, introduced in [N1, VV] for simply-laced types using quiver varieties
(and generalized in [H1] using a different method). The deformation is then identified with
a natural quantization of the cluster algebra (the quantum cluster algebra). Recently, with
R. Fujita [FuH], we propose a construction of monoidal Jantzen filtrations for monoidal
categories, with certain degenerated braidings, in order to give a purely categorical con-
struction of such quantum Grothendieck rings. This is based on distinguished filtrations by
submodules :
M = M0 ⊃ M1 ⊃ · · · ⊃ MN = {0}
which lead to deformed classes :
X
[M ]t = tr [Mr /Mr−1 ],
r≥0
and to a deformation ∗ of the product of the Grothendieck ring, by considering filtrations
of tensor products.
As an illustration, in the first example of Section 4.4, we have the filtrations
V1 (q 2 ) ⊗ V1 (1) ⊃ 1 ⊃ {0} and V1 (1) ⊗ V1 (q 2 ) ⊃ W ⊃ {0}
which lead to the deformed products
[V1 (q 2 )] ∗ [V1 (1)] = t[W ] + 1 and [V1 (1)] ∗ [V1 (q 2 )] = t−1 [W ] + 1.

5. Shifted quantum affine algebras

The next step is to handle shifted quantum groups. The shifted quantum affine algebras
form a new class of quantum groups closely related to (quantized K-theoretical) Coulomb
branches(in the sense of Braverman-Finkelberg-Nakajima [BFN]). They were introduced
by Finkelberg-Tsymbaliuk [FT] in the study of these K-theoretical Coulomb branches.
Consider g and q as above. The algebra Uµq (ĝ) can be seen as a variation of the quantum
affine algebra Uq (ĝ) depending on a shift parameter : a coweight µ of the Lie algebra g (for
example in the sl2 -case, we can see µ as an integer). For µ = 0, U0q (ĝ) is essentially Uq (ĝ).
These algebras have a very interesting representation theory.
Theorem 5.1. [H2] Uµq (ĝ) has a non-zero finite-dimensional representation if and only if
µ is codominant.
2This large subcategory is sufficient to describe all simple modules by using spectral parameter shifts.
 SYMMETRIES OF GROTHENDIECK RINGS IN REPRESENTATION THEORY 11

In general, Uµq (ĝ) has an abelian category Oµ of representations (which might be finite-
dimensional or infinite-dimensional). Recall n is the rank of g.
Theorem 5.2. [H2] The simple objects in Oµ are parametrized by n-tuples of rational
fractions Ψ = (ψi (z))1≤i≤n regular at 0, with the degree condition :
deg(ψi (z)) = αi (µ)
where αi is the simple root of g attached to the vertex i of the Dynkin diagram.
Example 5.3. For 1 ≤ i ≤ n, a ∈ C∗ , consider Ψ defined by ψj (z) = 1 if j 6= i and
1 − zaq −ri
ψi (z) = q ri ,
1 − zaq ri
with ri the length of the simple root αi . Then we recover the finite-dimensional fundamental
representation Vi (a) of the ordinary quantum affine algebra (the rational fractions ψj (z)
should not be confused with the Drinfeld polynomials).
Example 5.4. For 1 ≤ i ≤ n, a ∈ C∗ , consider
Ψ = ((1 − za)δi,j )1≤j≤n .
ω∨
The associated simple representation of Uq i (ĝ) is called the positive prefundamental repre-
sentation L+ ∨
i,a (here ωi is a fundamental coweight). It is of dimension 1 !

Example 5.5. For 1 ≤ i ≤ n, a ∈ C∗ , consider

Ψ = ((1 − za)−δi,j )1≤j≤n .
−ωi∨
The associated simple representation of Uq (ĝ) is called the negative prefundamental rep-
−ω ∨
resentation L−
i,a .
It is infinite dimensional. In addition, Uq i (ĝ) contains a copy of the
ordinary quantum affine Borel algebra Uq (b̂). Then, restricted to this subalgebra, L−
i,a is
the simple prefundamental representation constructed in [HJ] as a limit of simple finite-
dimensional representations.
Without loss of generality, we may assume that all roots and poles of the rational fractions
are integral powers of q.
One obtains an abelian category
M
Osh = Oµ
µ coweight

The (completed) sum of Grothendieck groups

M
K0 (Osh ) = K0 (Oµ )
µ
3
has a (topological) ring structure induced by a the fusion product construction (which
defines a representation V ∗ W in Oµ+λ from simple representations V , W in Oµ and Oλ
respectively). This ring can be considered as an analogue of a Grothendieck ring. In
addition, it is commutative.
3It will be proved in [HZ] that the subcategory of finite length representations is stable by fusion product,
and so it gives rise to a (non-topological) ring structure on K0 (Osh ).
12 DAVID HERNANDEZ

Consider the subcategory Csh ⊂ Osh of finite-dimensional representations of shifted quan-

tum affine algebras. Its Grothendieck ring K(Csh ) contains as a subring the Grothendieck
ring K(C) of finite-dimensional representations of the ordinary quantum affine algebra.
Theorem 5.6. [HL3, KKOP, H2] K0 (Csh ) is isomorphic to a cluster algebra AΓ∞ , with
an explicit quiver Γ∞ . The initial cluster variables are classes of positive prefundamental
representations. The cluster monomials correspond to certain classes of simple modules.
Example 5.7. For g = sl2 , the quiver Γ∞ is the infinite linear quiver :
· · · −→ • −→ • −→ · · ·
The initial seed is formed of classes of 1-dimensional positive prefundamental representa-
tions
· · · −→ L+
1,q −2
−→ L+ +
1,1 −→ L1,q 2 −→ · · ·
(up to factors by invertible representations, that we omit in these notes for clarity). The
first step mutations are given by certain remarkable relations called Baxter T Q-relations
[L+ + +
1,a ][V1 (a)] = [L1,aq 2 ] + [L1,aq −2 ]

where V1 (a) : 2-dimensional fundamental representation. In fact, this relation is also rele-
vant from the point of view of quantum integrable models.
Theorem 5.8. [GHL] The Grothendieck ring K0 (Osh ) is isomorphic to (a completion of )
a cluster algebra AΓ′∞ , with an explicit quiver Γ′∞ .
The proof is partly based on additional symmetries of the Grothendieck ring that will be
discussed in the next section.
Example 5.9. For g = sl2 , the new quiver Γ′∞ is obtained from the older quiver Γ∞ by just
inverting the direction of one arrow :
· · · −→ • −→ •←−• −→ • −→ · · ·
Then the initial seed is made of positive and negative prefundamental representations :
· · · −→ L+
1,q 2
−→ L+ − −
1,1 ←− L1,q −2 −→ L1,q −4 −→ · · ·

The mutation at the vertex corresponding to L+

1,1

· · · −→ L+
1,q 2
←− L− − −
1,1 −→ L1,q −2 −→ L1,q −4 −→ · · ·

as we have the remarkable QQ-relation

(5.3) [L+ − + +
1,1 ][L1,1 ] = 1 + [L1,q 2 ][L1,q −2 ].

Similarly, the mutation at the vertex corresponding to L−

1,q −2
translated the seed in the other
direction :
· · · −→ L+
1,q 2
−→ L+ + −
1,1 −→ L1,q −2 ←− L1,q −4 −→ · · ·

In general, we conjecture the following.

Conjecture 5.10. [GHL] All cluster monomials in AΓ′∞ correspond to classes of simple
objects in O through our isomorphism.
This general Conjecture is, for moment, only known in the following case.
Theorem 5.11. [GHL] The conjecture is true for g = sl2 .
 SYMMETRIES OF GROTHENDIECK RINGS IN REPRESENTATION THEORY 13

6. Symmetry of q-characters
We now discuss certain group symmetries of Grothendieck rings.
6.1. Classical Theory. Let us briefly the well-known Weyl group symmetry of classical
characters. Let g be a complex finite-dimensional simple Lie algebra of rank n, and K0 (g)
be the Grothendieck ring of its finite-dimensional representations. The character morphism,
which encodes the dimension of weight spaces of finite-dimensional representations, defines
an injective ring morphism
χ : K0 (g) → Z[yi±1 ]1≤i≤n .
The image of χ can be characterized as an invariant subring
Im(χ) = (Z[yi±1 ]1≤i≤n )W
where the Weyl group W is generated by the simple reflexions si , 1 ≤ i ≤ n which act by
the explicit formula
−δ
Y Cji
si (yj ) = yj ai ij where ai = yk .
1≤k≤n
Here ai corresponds to a simple root, and the Cji are coefficients of the Cartan matrix of g.
Example 6.1. For g = sl2 , the Cartan matrix C = (2) and we have
a1 = y12
s1 (y1 ) = y1 a−1 −1
1 = y1 ,
s21 (y1 ) = y1
s1 (y1 + y1−1 ) = y1 + y1−1 .
Im(χ) = (Z[y1±1 ])W = Z[y1 + y1−1 ].
6.2. q-characters. The analogue of characters for finite-dimensional representations of
quantum affine algebras Uq (ĝ) are q-characters (as defined by Frenkel-Reshetikhin [FR]).
It defines an injective ring morphism on the Grothendieck ring K(C) of the category C of
finite-dimensional representations (here we assume that the quantization parameter q ∈ C∗
is not a root of unity) :
±1
χq : K(C) → Y = Z[Yi,a ]1≤i≤n,a∈C∗ .
It is a refinement of characters for the action of the underlying finite-type quantum group
Uq (g) (as we can recover the ordinary by removing the spectral parameters a).
Example 6.2. For a ∈ C∗ , the fundamental representation V1 (a) of Uq (slˆ2 ) satisfies :
−1
χq (V1 (a)) = Y1,a + Y1,aq 2.

If we forget the spectral characters, we recover the ordinary character Y1 + Y1−1 .

There are monomial analogs of simple roots (we give here the formula for simply-laced
types) :
Y
−1
Ai,a = Yi,aq−1 Yi,aq Yj,a .
j|Cj,i =−1
For example for g = sl2 , we have A1,a = Y1,aq−1 Y1,aq .
We may wonder what is the analogue of the Weyl group symmetry in this context.
14 DAVID HERNANDEZ

6.3. Symmetry of q-characters. We introduce in [FH2] the following operators :

δi,j
−δi,j Σi,aq −3
Θi (Yj,a ) = Yj,a Ai,aq−1 δ ,
i,j
Σi,aq −1

for 1 ≤ i, j ≤ n and a ∈ C∗ . Here Σi,a is the solution of the q-difference equation

Σi,a = 1 + A−1
i,a Σi,aq −2
in a sum M
Π= Ỹw
w∈W

of completions Ỹw of Y, parametrized by the Weyl group.

Example 6.3. For g = sl2 , we have
−1 Σ1,aq−3
Θ1 (Y1,a ) = Y1,aq −2 .
Σ1,aq−1
Here Σ1,a is the couple
(1 + A−1 −1
1,a (1 + A1,aq −2 (1 + · · · ),
−A1,aq2 (1 + A1,aq4 (1 + · · · ))).
It belongs to
Π = Ye ⊕ Ys1 .
Remark 6.4. One component of these operators is related to operators defined in [I] on a
different space.
Now Y embeds in Π diagonally, and we have the following analogue of the classical result.
Theorem 6.5. [FH2] The Θi are involutions and define a Weyl group action (with the
simple reflexions acting as Θi ). Moreover, we have the ring of invariants
YW = Im(χq ).
6.4. Interpretation of the Weyl group action. Let us explain an interpretation of the
Weyl group action above, which connected to the cluster algebra symmetry.
Theorem 6.6. [FH1, H2] The replacement of the variables :
[L+
i,aq −1
]
Yi,a =
[L+
i,aq ]

in the q-character of finite-dimensional representation gives a well-defined relation in (the

fraction field) of K(Osh ).
Note that originally, this Theorem was established in [FH1] in the context of representa-
tions of Borel subalgebra of quantum affine algebras (as in Example 5.5, in order to prove
Frenkel-Reshetikhin conjecture on spectra of quantum integrable models), but it can also
be stated for representations of shifted quantum affine algebras.
Motivated by quantum integrable models and Bethe Ansatz equations, we expect that the
class [L+
i,a ] can by replaced in the Theorem by the class of other simple representations, that
we denote by L̃+ i,a (see [FH3]). This is exactly how one can write the defining formula of the
 SYMMETRIES OF GROTHENDIECK RINGS IN REPRESENTATION THEORY 15

operator Θi , which is now interpreted as a substitution of simple classes in the Grothendieck

ring.
ω∨
We can now relate this picture to cluster algebra symmetry. Indeed, starting from Qa i =
[L+i,a ], and using the Weyl group action, one can introduce for a general w ∈ W , i ∈ I :
w(ωi∨ )
Qa ∈ K(Osh ).
w(ωi∨ )
Theorem 6.7. [FH4][Generalized QQ-systems] The series Qa satisfy the QQ-system
(ws )(ω ∨ ) w(ω ∨ ) (ws )(ω ∨ ) w(ω ∨ )
Y w(ωj∨ )
Qaq i i Qaq−1i − Qaq−1i i Qaq i = Qa .
j|Ci,j =−1

Hence the same relation is satisfied by elements in an orbit under the Weyl group action.
Example 6.8. For g = sl2 , this QQ-system is the quantum Wronskian relation :
(6.4) Q̃aq Qaq−1 − Q̃aq−1 Qaq = 1,
ω∨ −ω1∨
where we have denoted Qa = Qa 1 and Q̃a = Qa .
These QQ-system relations are crucial in the construction of the cluster algebra structure
on K(Osh ) in [GHL], as they are appear in mutations of the cluster algebra (for example,
compare Equation (5.3) and Equation (6.4)). Hence, cluster symmetries are related to group
symmetry of Grothendieck rings in these examples, which one could find quite intriguing.

References
[BC] M. Brito and V. Chari, Tensor products and q-characters of HL-modules and monoidal categori-
fications, J. Éc. polytech. Math. 6 (2019), 581–619
[BFN] A. Braverman, M. Finkelberg and H. Nakajima, Towards a mathematical definition of
Coulomb branches of 3-dimensional N=4 gauge theories, II, Adv. Theor. Math. Phys. 22 (2018),
no. 5, 1071–1147.
[BR] P. Baumann and S. Riche, Notes on the geometric Satake equivalence, Relative aspects in repre-
sentation theory, Langlands functoriality and automorphic forms, 1–134, Lecture Notes in Math.,
2221, CIRM Jean-Morlet Ser., Springer, Cham, 2018.
[CG] N. Chriss and V. Ginzburg, Representation theory and complex geometry, Birkhäuser Boston,
Inc., Boston, MA, 1997.
[CW] S. Cautis and H. Williams, Cluster theory of the coherent Satake category, J. Amer. Math. Soc.
32 (2019), no. 3, 709–778.
[FH1] E. Frenkel and D. Hernandez, Baxter’s Relations and Spectra of Quantum Integrable Models,
Duke Math. J. 164 (2015) 2407–2460.
[FH2] E. Frenkel and D. Hernandez, Spectra of quantum KdV Hamiltonians, Langlands duality, and
affine opers, Comm. Math. Phys. 362 (2018) 361–414.
[FH3] E. Frenkel and D. Hernandez, Weyl group symmetry of q-characters, Preprint
arXiv:2211.09779.
[FH4] E. Frenkel and D. Hernandez, Extended Baxter relations and QQ-systems for quantum affine
algebras, Comm. Math. Phys. 405 (2024), 190.
[FR] E. Frenkel and N. Reshetikhin, The q-characters of representations of quantum affine algebras
and deformations of W -Algebras, in Recent Developments in Quantum Affine Algebras and related
topics, Contemp. Math. 248 (1999), 163–205 (corrected version available at arXiv:math/9810055).
[FuH] R. Fujita and D. Hernandez, Monoidal Jantzen filtrations, Preprint arXiv:2402.13544.
[FHOO] R. Fujita, D. Hernandez, S-J. Oh and H. Oya, Isomorphisms among quantum Grothendieck
rings and cluster algebras, Preprint arXiv:2304.02562.
[FT] M. Finkelberg and A. Tsymbaliuk, Multiplicative slices, relativistic Toda and shifted quantum
affine algebras, in Progr. Math. 330(2019), 133–304.
16 DAVID HERNANDEZ

[FZ] S. Fomin and A. Zelevinsky, Cluster algebras I: Foundations, J. Amer. Math. Soc. 15 (2002),
497–529.
[GHL] C. Geiss, D. Hernandez and B. Leclerc, Representations of shifted quantum affine algebras
and cluster algebras I: The simply laced case, Proc. Lond. Math. Soc. (3) 129 (2024), no. 3, No.
e12630.
[H1] D. Hernandez, Algebraic approach to q, t-characters, Adv. Math. 187, 1–52 (2004).
[H2] D. Hernandez, Representations of shifted quantum affine algebras, Int. Math. Res. Not. IMRN
2023, no. 13, 11035–11126.
[HJ] D. Hernandez and M. Jimbo, Asymptotic representations and Drinfeld rational fractions, Com-
pos. Math. 148 (2012), 1593–1623.
[HL1] D. Hernandez and B. Leclerc, Cluster algebras and quantum affine algebras, Duke Math. J.
154 (2010), no. 2, 265–341.
[HL2] D. Hernandez and B. Leclerc, Quantum Grothendieck rings and derived Hall algebras, J. Reine
Angew. Math. 701 (2015), 77–126.
[HL3] D. Hernandez and B. Leclerc, A cluster algebra approach to q-characters of Kirillov-Reshetikhin
modules, J. Eur. Math. Soc. (JEMS) 18 (2016), no. 5, 1113–1159.
[HZ] D. Hernandez and H. Zhang, in progress.
[I] R. Inoue, Cluster realizations of Weyl groups and q-characters of quantum affine algebras, Lett.
Math. Phys. 111 (2021) 4 (arXiv:2003.04491).
[K] B. Keller, Cluster algebras and cluster monomials, European Congress of Mathematics, 399–413,
Eur. Math. Soc., Zürich, 2013.
[KKKO] S-J. Kang, M. Kashiwara, M. Kim and S-J. Oh, Monoidal categorification of cluster algebras,
J. Amer. Math. Soc. 31 (2018), no. 2, 349–426.
[KKOP] M. Kashiwara, M. Kim, S-J. Oh, and E. Park, Monoidal categorification and quantum affine
algebras II, Invent. Math. 236 (2024), no. 2, 837–924.
[KL] M. Khovanov and A. Lauda, A diagrammatic approach to categorification of quantum groups I,
Represent. Theory 13 (2009), 309–347.
[ML] S. Mac Lane, Categories for the working mathematician, Second edition. Graduate Texts in Math-
ematics, 5. Springer-Verlag, New York, 1998
[N1] H. Nakajima, Quiver varieties and t-analogs of q-characters of quantum affine algebras, Ann. of
Math. (2) 160, no. 3, 1057–1097 (2004).
[N2] H. Nakajima, Quiver varieties and cluster algebras, Kyoto J. Math. 51 (2011), no. 1, 71–126.
[Q] F. Qin, Triangular bases in quantum cluster algebras and monoidal categorification conjectures,
Duke Math. J. 166 (2017), no. 12, 2337–2442.
[R] R. Rouquier, 2-Kac-Moody algebras, arXiv:0812.5023.
[VV] M. Varagnolo and E. Vasserot, Perverse sheaves and quantum Grothendieck rings, Studies in
memory of Issai Schur (Chevaleret/Rehovot, 2000), Progr. Math. 210, Birkhäuser Boston, Boston,
MA, 345–365 (2003).

IMJ-PRG, Université Paris Cité and Sorbonne Université, CNRS, F-75006, Paris, France