Prepared for submission to JHEP

On the zoology of 2d N = (0, 2) dualities

gauge theories with antisymmetric matter
arXiv:2504.16544v1 [hep-th] 23 Apr 2025

Antonio Amariti,a Pietro Glorioso,a,b Chiara Mascherpa,a,b and Andrea Zanettia,b

a INFN, Sezione di Milano, Via Celoria 16, I-20133 Milano, Italy
b Dipartimento di Fisica, Università degli studi di Milano, Via Celoria 16, I-20133
E-mail: [email protected], [email protected],
[email protected], [email protected]

Abstract: In this paper we investigate and propose new dualities involving 2d gauge
theories with N = (0, 2) supersymmetry. In the first part of the paper we focus on
SU(n) gauge theories with two antisymmetric chirals. The gauge theories are non-
anomalous if we consider, in addition to such matter content, nf fundamental and na
antifundamental chirals, provided the constraint nf + na = 4. By exploring the five
possibile scenarios arising from this constraint we provide in each case evidences of
a dual LG description, by matching the ’t Hooft anomalies and deriving the relation
between the elliptic genera in terms of other more fundamental dualities. In the second
part of the paper we provide a 4d origin for a gauge/LG duality already stated in the
literature, that does not descend from any known s-confining duality. In the last part
of the paper we focus on dualities for SU(n) and USp(2n) models with antisymmetric
Fermi multiplets, obtained from dimensional reduction of 4d parent dualities.
Contents
1 Introduction 1

2 Basic dualities 4
2.1 USp(2n) with 2n + 3 □ chirals and 1 □ Fermi 5
2.2 USp(2n) with 2n + 2 fundamentals 7
2.3 SU(N ) with N + x □ chirals, N − x + y □ chirals, y □ Fermi 7

3 SU(N ) with two antisymmetric chirals 9

3.1 SU(2n) with four fundamentals 10
3.2 SU(2n + 1) with four fundamentals 17
3.3 SU(2n) with three fundamentals and one antifundamental 21
3.4 SU(2n + 1) with three fundamentals and one antifundamental 25
3.5 SU(2n) with two fundamentals and two antifundamentals 27
3.6 SU(2n + 1) with two fundamentals and two antifundamentals 29
3.7 SU(2n) with one fundamentals and three antifundamentals 31
3.8 SU(2n + 1) with one fundamentals and three antifundamentals 34
3.9 SU(2n) with four antifundamentals 36
3.10 SU(2n + 1) with four antifundamentals 39

4 A 4d parent for SU(2n) with (A, Ã) and 4 □ 41

5 Dualities involving antisymmetric Fermi fields 44

5.1 A gauge/LG duality with an antisymmetric Fermi 44
5.2 A gauge/gauge duality with an antisymmetric Fermi 48

6 Conclusions 50

A Conventions on the elliptic genus 52

1 Introduction
The existence of IR dualities is a fascinating aspect of quantum field theories, because it
opens up the possibility of alternative field theoretical descriptions of the same physical
setup. Ubiquitous examples have been obtained in the supersymmetric setup, where

–1–
dualities played a prominent role in different dimensions and degree of supersymmetry.
An ultimate goal would be achieving a connection with the real world by uncovering
dualities beyond the supersymmetric case. For this reason it is useful to take an in-
termediate step by finding dualities with a low amount of supercharges. While in 4d
the lowest possible cases require four supercharges, in lower dimensional physics the
number of supersymmetry generators can be lowered. An interesting setup involves
2d dualities with N = (0, 2) supersymmetry, as in this case a whole class of dualities
can be achieved by dimensional reduction of 4d Seiberg-like dualities from a twisted
compactification on a two sphere.
Typically, such compactification of 4d dual pairs results in relations between collec-
tion of theories in lower dimension, obscuring the possible claims on pure 2d dualities.
However, in [1], elaborating on the results of localization, the authors discuss a pre-
scription that allows to infer the existence of 2d dualities starting from 4d ones. The
prescription requires to fix the R charges of the 4d fields to be non-negative integers.
In this way the whole contribution to the 4d S 2 × T 2 topologically twisted index [2]
comes from the zero flux sector only, discarding the sum over the remaining ones. By
translating the same constraints imposed on the electric side to the magnetic one via
the 4d duality map, one obtains a candidate 2d N = (0, 2) duality. The ’t Hooft
anomaly matching are automatically satisfied and the equivalence of the elliptic genera
is a consequence of the equivalence of the 4d topologically twisted indices.
In principle one cannot trust such dualities without any hesitation and further
checks are necessary. For example the theories obtained in this way usually have non-
compact direction in the target space. For this reason the c-extremization procedure
of [3, 4] requires some care. Furthermore, the matching of the elliptic genera in this
case is reliable only in the massive case.
Remarkable checks of such dualities have been performed in the literature when one
restricts to the s-confining limit of Seiberg duality [5] and Intriligator-Pouliot duality
[6]. In such cases the dual theories correspond to Landau-Ginzburg (LG) models.
Furthermore, as observed in [1], in the USp(2n) case the prescription described so
far does not apply to the reduction of the Intriligator-Pouliot beyond the s-confining
regime, whereas this is allowed for Seiberg duality.
Restricting to the reduction of s-confining gauge theories, it has been shown in
[7–9] that the other 4d s-confining dualities reduce to 2d gauge/LG dualities that can
be derived in a pure 2d picture by generalizing the tensor deconfinement technique of
[10, 11]. This approach of deriving dualities from dualities was used recently in various
dimensions [12–24] in order to show that several dualities in presence of tensor matter
are consequence of more basic ones with only fundamental matter. A relevant result
for our discussion was obtained in [25], where it was shown that the classification of

–2–
[26] for SU(N ) and USp(2n) dualities can be obtained using only two basic s-confining
dualities, i.e. the s-confining limits of Seiberg and Intriligator-Pouliot dualities. The
same phenomenon has been generalized in [9] for 2d dualities obtained by applying the
prescription of [1] to such s-confining cases1 .
Furthermore, the idea of deriving 2d dualities with tensorial matter in terms of
basic gauge/LG dualities can be applied to other cases, where a 4d parent s-confining
description is absent. Some examples have already been studied in [8, 9], but more
cases are expected. A reason behind such expectations is that the constraints imposed
by gauge anomalies are milder in 2d than in 4d. In addition, the similarities between
the dualities found in [8, 9] and the 3d ones with a similar field content obtained in
[15, 19, 22, 27] suggest that 2d gauge/LG dualities should exist also for SU(N ) gauge
theories with two antisymmetric chirals.
Motivated by this last observation, in this paper we continue our investigations of 2d
N = (0, 2) SUSY QFTs dual to LG models, providing a generalization of the previous
results found in [9]. We mostly focus on SU(N ) gauge theories with antisymmetric and
fundamental matter, considering both chiral and Fermi fields.
The first class of models under investigation consists of gauge/LG dualities without
a known 4d origin, i.e. no 4d confining dualities available in the literature give rise
to the 2d dualities argued here through a topological twist. We discuss SU(N ) gauge
theories with two antisymmetric, nf fundamental and na antifundamental chirals, such
that nf +na = 4. Notice that, even if such dualities do not descend from any 4d parent,
they share a similarity with analogous 3d dualities proposed in [27], and further checked
in [22, 28]. Such relation agrees with the results of [9], where the various 2d dualities
were shown to be in line, even in absence of a 4d parent, with analogous 3d dualities.
The main checks of the dualities proposed here consist of ’t Hooft anomaly matching
and of the study of the elliptic genus. While the first check is straightforward, the
matching of the elliptic genera in this case cannot be inferred from the one of any 4d
partition function, and its analysis is a crucial test of the duality. Here we show that
the matching of the elliptic genera reduces to the one of simpler dualities, that hold in
absence of any tensorial matter field and that have a 4d origin. The analysis is very
similar to the one performed in [7–9] and it requires a duality that is not completely
under control in 2d. The auxiliary duality involves an USp(2n) gauge theory with
2N + 3 fundamental chirals and 1 fundamental Fermi. This model is dual to an LG,
provided that a symmetry of the electric theory is obstructed2 . Such obstruction is due
1
The sporadic cases have not been studied in detail in [9], but the same result is expected to hold
for them as well.
2
We will often refer, with a slight abuse of notation, to the electric and the magnetic phase for the
2d models under investigation, in analogy, when it exists, with the 4d counterpart.

–3–
to the choice of R charges used to perform the twist on the two-sphere. As observed in
[1, 7] this is reminiscent of the obstruction to the generation of axial (and topological)
symmetries in 3d N = 2 gauge theories. Here in the 2d picture we do not have an
analogous candidate playing the role of the monopole superpotential3 , nevertheless we
claim that the identity derived from the reduction of the zero flux sector of the S 2 × T 2
topologically twisted index to the elliptic genus is reliable, and can be used to provide
a derivation of the dualities under investigation. Assuming the matching of the elliptic
genera for such duality, we will provide the relative identity for the duality involving
SU(N ) with two antisymmetric chirals and four fundamentals. Such proof requires a
recursive analysis that leads to the towers of operators, expected from the field theory
duality, that one can conjecture from the 3d analogous found in [27]. Once this duality
with nf = 4 is established, we will show that all the other cases with na ̸= 0 can be
derived from the one with nf = 4, i.e. in the recursive derivation we always find a
deformed version of the model with nf = 4.
In the second part of the paper we revisit the problem of 2d dualities without
a known 4d parent, by considering the case of SU(2n) with an antisymmetric flavor
and four fundamentals discussed in [9]. Here we present a 4d parent duality, with a
non-vanishing superpotential, that reduces to the 2d expected one by twisting along an
assignation of non-negative integer R charges.
The last part of the paper focuses on SU(N ) and USp(2n) dualities in presence
of a charged antisymmetric Fermi field. These models are obtained by reducing 4d
dualities that have been guessed in the literature from the analysis of the superconfor-
mal index [30]. A physical derivation of such 4d parent dualities was then provided in
[20, 21]. We study the twist of these dualities on the two sphere by choosing integer
non-vanishing R symmetries and in this way we argue the existence of new 2d dualities.
In the first case the duality involves an SU(N ) theory with N fundamentals and one
antisymmetric Fermi whose dual model is identified with an LG model. The second
case regards a duality between an USp(2n) theory and an USp(2N − 2) theory, both
with an antisymmetric Fermi, 4N fundamentals and non-trivial J-terms.

2 Basic dualities

In this section we review some of the relevant basic dualities that we will use in the
paper in order to derive the matching of the elliptic general for the dualities proposed
below. These dualities are already known, but they have been less discussed in the
3
The notion of boundary monopoles in [29] seems promising for understanding a dynamical origin
for such obstructions.

–4–
literature, for this reason we provide a detailed discussion and some further comments
in order to employ them below.
In general we consider the 4d/2d reduction of dualities by compactifying on S 2 and
by applying the prescription introduced in [1]: i.e. given a 4d chiral multiplet with R
charge R, this reduces to
• R − 1 Fermi multiplets if R > 1 ,

• 1 − R chiral multiplets if R < 1 ,

• no multiplets if R = 1 ,
while the 4d vector multiplet reduces to a 2d vector multiplet. As discussed in [1] this
set of rules, provided that we have integer and non-negative R charges, allows to obtain
a 2d N = (0, 2) theory from a 4d N = 1 one. Here we further restrict our attention
to cases without any field with R > 2 in order to avoid the generation of possible
non-abelian symmetries in 2d.

2.1 USp(2n) with 2n + 3 □ chirals and 1 □ Fermi

We start our survey by discussing a duality with a 4d origin that will play a relevant
role in our search of new 2d gauge/LG dualities. It corresponds to USp(2n) with 2n + 3
fundamental chirals and one Fermi. The duality can be derived from 4d using USp(2n)
with 2n + 4 fundamentals by assigning an R charge equal to 2 to one fundamental and
R charges equal to zero to the remaining 2n + 3 fundamentals. The 2d field content
is anomaly free and the dual meson splits into a chiral antisymmetric meson with
(2n + 3)(n + 2) components ΦMab and 2n + 3 Fermi Ψa .
The superpotential of the 2d theory is
 
W2d = ϵa1 ,...,a2n+3 ΦMa1 a2 . . . ΦMa2n+1 a2n+2 Ψa2n+3 ≡ Φn+1
M Ψ (2.1)
and it descends from the 4d superpotential of the confining theory, corresponding to
the limiting case of the Intriligator-Pouliot duality [6]. In the following, we will adopt
a 2d notation, where we will refer to the J-term for the Fermi Ψa instead of the 2d
superpotential. Moreover, in all the examples discussed in this paper the E-terms are
vanishing.
The global charges of the fields are
U(1)Q SU(2n + 3)
Q 1 □
ψ −2n − 3 · (2.2)
ΦM = Q2
2 □
□
Ψ = Qψ −2n − 2 □

–5–
 ψ ΦM

Q Ψ
2n 2n + 3 2n+3

Figure 1: In this figure we represents the quivers associated to the duality between USp(2n)
with 2n + 3 fundamental chirals and one Fermi, and the LG model. Dashed lines are for
Fermi fields while continuous lines are for chiral multiplets. The dots refer to the singlets.

As a first check one can see that the anomalies

κQQ = −4n(n + 1)(2n + 3) , κSU(2n+3)2 = n , (2.3)

are equal on both sides of the duality.

Here is a crucial aspect of the duality proposed above: on the gauge theory side of
the duality we have not included an axial symmetry, which gives opposite charge to the
2n + 3 chirals and to the Fermi. However, such symmetry is not forbidden by the field
content and by the requirements from anomaly cancellation. At the level of localization
such a symmetry is not generated in the reduction of the S 2 ×T 2 index to the 2d elliptic
genus along the lines of the prescription of [1], i.e. by assigning vanishing R charge to
2n + 2 fields and R = 2 to the (2n + 4)-th 4d N = 1 chiral. As discussed originally in [1]
and elaborated in [7], in this case the absence of such a symmetry is very similar to the
obstruction to the generation of an axial symmetry in the 4d/3d reduction of dualities
due to finite size effects [31]. In the 4d/2d reduction of [1], restricting our attention to
the cases with R = 0, 1 and 2, such analogy is ubiquitous for all the anomaly free R
charge assignations where there are no fields with R = 1, i.e. each 4d field gives origin
either to a 2d N = (0, 2) chiral or to a 2d N = (0, 2) Fermi.
We expect that analogous finite size effects have to be considered in such 2d cases
as well. Even if we postpone a physical discussion about the origin of such effect to
future investigation, it is worth here to comment more on such issue. At the level of
localization the obstruction imposed by the anomalies translates into a constraint on
the fugacities. Borrowing the mathematical terminology, such constraints are denoted
as balancing conditions in the literature of the 4d superconformal index (see for example
[30]). Similar constraints arise in the analysis of the 3d three sphere partition function
when monopole superpotential (for example KK monopole superpotentials signalling
the presence of finite size effects) are considered. Similarly, in the 4d/2d reduction of
the S 2 × T 2 index to the elliptic genus, the 4d constraints on the fugacities survive
when only fields with R charges R = 0 and R = 2 are considered in 4d. For this reason
we expect that such balancing conditions signal the fact that finite size effects have to

–6–
be considered in such cases. We will return on such issue below in the paper when we
will apply the duality discussed here in order to deconfine the antisymmetric tensors4 .
At the level of the elliptic genus the duality discussed in this subsection translates
into the identity
Q2n+3
(2n+3; ·) −2n−3 θ (qua /x2n+2 )
IUSp(2n) (x⃗u; x ; ·) = Q a=1 2
, (2.4)
1≤a<b≤2n+3 θ(x ua ub )

with 2n+3
Q
a=1 ua = 1. We refer the reader to Appendix A for the conventions adopted
here on the elliptic genus.

2.2 USp(2n) with 2n + 2 fundamentals

This duality has been obtained in [1] and further studied in [32] for n = 1. It can
be obtained from the reduction of the 4d s-confining limit of the USp(2n) Intriligator-
Pouliot duality, i.e. in presence of 2n + 4 fundamentals and vanishing superpotential.
In this case the twist needs to be performed by fixing the R charge of two fundamentals
to R = 1, and by assigning vanishing R charge to the remaining ones.
The 2d duality maps an USp(2n) gauge theory with 2n + 2 fundamental chirals Q
to an LG theory with an antisymmetric chiral A = Q2 in addition to a Fermi Ψ with
J-term JΨ = PfA. At the level of the elliptic genus the relative identity is
(2n+2,·) θ (q(u1 . . . u2n+2 )−1 )
IUSp(2n) (⃗u; ·; ·) = Q , (2.5)
1≤a<b≤2n+2 θ(ua ub )

where we stress that the numerator on the RHS is written in this way in order to make
apparent the structure of the J-term for the Fermi Ψ. Using the fact that θ(q/x) = θ(x)
Q
we could also write the numerator as θ( ua ), which would correspond to exchanging
the J term with an E term. We refer the reader to Appendix A for the conventions
adopted here on the elliptic genus.

2.3 SU(N ) with N + x □ chirals, N − x + y □ chirals, y □ Fermi

Here we present the duality introduced in [1]. We first start from a 4d SU(N ) gauge
theory with N + x flavors. The Seiberg dual of this theory is a 4d SU(x) gauge theory
with N + x flavors, a chiral meson and a superpotential given by W = q̃φq. An
anomaly-free R symmetry for the electric theory requires that
N +x
X 1
NX+x
1

R(Qi ) − 1 + R(Q̃i ) − 1 + N = 0 . (2.6)
i=1
2 i=1
2
4
Observe that this 2d confining duality has been recently used in [8] in order to prove 2d confining
dualities with antisymmetric matter.

–7–
 φ
ψ N −x+y
Q̃ N −x+y
q̃
Q N +x x
N +x N
q y
λ
Ψ y

Figure 2: Quiver diagrams of the electric theory (left) and the magnetic theory (right).

Any R charge assignation must respect this constraint. Using the 4d duality dictionary,
in the magnetic theory we have
N +x N +x
1X 1X
R(qi ) = R(Q̃j ) − R(Q̃i ) , R(q̃i ) = R(Qj ) − R(Qi ) ,
x j=1 x j=1 (2.7)
R(φij ) = R(Qi ) + R(Q̃j ) .

To perform the reduction to 2d we assign R = 0 to N + x fundamental chirals and

to N − x + y antifundamental chirals, R = 1 to some other 2(x − y) antifundamental
chirals and R = 2 to the last y antifundamental chirals. One can easily check that such
assignation respects the constraint (2.6). Therefore we obtain a 2d theory with N + x
fundamental chiral multiplets Q, N − x + y antifundamental chiral multiplets Q̃ and y
antifundamental Fermi multiplets Ψ, as shown on the left side of the Figure 2. There
is a restriction on the allowed values of x and y, indeed Seiberg duality requires x ≥ 2
and from the reduction rules we see that x > y 5 .
On the other hand, from (2.7) the field content of the 4d magnetic theory reduces
as follows: the 4d antifundamental chirals become N + x antifundamental chirals q̃
in 2d, the 4d fundamental chirals split into y fundamental chirals q and N − x + y
fundamental Fermi fields6 ψ in 2d. The 4d chiral meson splits into a 2d chiral meson
φ and a 2d Fermi meson λ. The magnetic theory is represented on the right side of
Figure 2. The original 4d superpotential splits into two J-terms

Jλ = q̃q, Jψ = q̃φ . (2.8)

In order to check this new 2d duality we verify that the ’t Hooft anomalies coincide.
The global symmetries of both electric and magnetic theories are listed below.
5
The case x = y is similar to the one studied in subsection 2.1, where a 4d anomalous axial
symmetry is not generated in 2d. We will not need such duality in the present discussion and for this
reason we omit to review this case.
6
Or equivalently antifundamentals, trading the relative J terms with opportune E-terms.

–8–
 SU(N + x) SU(N − x + y) SU(y) U(1)Q U(1)Q̃ U(1)Ψ U(1)R
Q □ · · 1 0 0 0
Q̃ · □ · 0 1 0 0
Ψ · · □ 0 0 −1 1
N
q · · □ −1 − x 0 1 0 (2.9)
N
q̃ □ · · x
0 0 0
N
ψ · □ · −1 − x −1 0 1
φ □ □ · 1 1 0 0
λ □ · □ 1 0 −1 1

Notice that we have an explicit duality dictionary to write q̃, φ and λ in terms of
the electric variables, but not for q and ψ. However, we can fix their U(1) charges from
the superpotential (2.8). From the table above one can read the ’t Hooft anomalies
and find out that they match:
N
κSU(n)2 = 2
, κQQ = −κQR = N (N + x) ,
κRR = N 2 + 1 , κQ̃Q̃ = −κQ̃R = N (N − x + y) , (2.10)
κQQ̃ = κQΨ = κQ̃Ψ = 0 , κΨΨ = −κΨR = −N y .

At the level of the elliptic genus the identity that follows from the reduction of the
index on S 2 × T 2 is
QN +x Qy
(N +x;N −x+y;y;·;·) ⃗ a=1 θ(qui ha ) (y;N +x;N −x+y;·;·) ⃗ ⃗ ⃗
ISU(N ) (⃗u; ⃗v ; h; ·; ·) = QNi=1
+x QN −x+y
ISU(x) (h̃, ũ, ṽ; ·; ·)
i=1 j=1 θ(ui vj )
(2.11)
−1
QN +x x1 −1
QN +x − x1 N +x − 1
and h̃a = h−1
Q
with ũi = ui ℓ=1 uℓ , ṽj = vj ℓ=1 uℓ a ℓ=1 uℓ . Again, we refer
x

the reader to Appendix A for the conventions adopted here on the elliptic genus.

3 SU(N ) with two antisymmetric chirals

In this section we discuss the existence of LG duals for 2d N = (0, 2) SU(N ) gauge the-
ories with two antisymmetric tensors A1,2 , nf fundamentals and na antifundamentals,
with nf + na = 4. The cases with even N = 2n rank and with odd N = 2n + 1 rank are
different for any allowed choice of nf and na , therefore we need to treat each case sep-
arately. In each one we propose the duality and check the ’t Hooft anomaly matching.
Furthermore, we provide a derivation of the matching of the elliptic genera by employ-
ing other identities obtained by the reduction of 4d Seiberg and Intriligator-Pouliot
duality.

–9–
3.1 SU(2n) with four fundamentals
In this case the dual LG model has three gauge invariant operators corresponding to
chiral fields interacting through a J-term with a Fermi multiplet. The field content of
the gauge theory and of the dual LG model is represented in the table below.

SU(2n) SU(2) SU(4) U(1)A U(1)Q U(1)R0

A □ □ · 1 0 0
□
Q □ · □ 0 1 0
n
Tn · ⊗sym □ · n 0 0 (3.1)
Tn−1 · ⊗n−1 □ n−1 2 0
sym □ □
n−2
Tn−2 · ⊗sym □ · n−2 4 0
2n−2
Ψ · ⊗sym □ · −2n + 2 −4 1

The three chirals Tj correspond to the following gauge invariant combinations of

the charged fields

Tn = An , Tn−1 = An−1 Q2 , Tn−2 = An−2 Q4 , (3.2)

while the J-term is

2
JΨ = Tn Tn−2 + Tn−1 , (3.3)
We have computed the ’t Hooft anomalies in the two phases and we have observed
that they match. Explicitly, we have
n(2n−1)
κSU(2)2 = 2
, κAA = −κAR0 = 2n(2n − 1) ,
κSU(4)2 = n , κQQ = −κQR0 = 8n , (3.4)
κR0 R0 = 6n + 1 , κAQ = 0 .

The further evidence that we provide to corroborate the existence of this duality
consists of showing that the matching of the elliptic genera of the two phases is a
consequence of other identities. These can be derived from the matching of the S 2 × T 2
topologically twisted index of well stated 4d s-confining parent dualities for special
unitary and symplectic SQCD.
The identity that we want to prove is
(4,·,·,2,·)
ISU(2n) (⃗u; ·; ·; ⃗t; ·) = (3.5)
Q2n−2   2(2n−2−j) 2j Q4 
j=0 θ q/ t1 t2 a=1 ua
= Qn 2(n−j) 2j
Q 4 Q n−1 2(n−1−j) 2j
Qn−2 2(n−2−j) 2j Q4
,
j=0 θ t1 t2 a<b j=0 θ t1 t2 ua ub j=0 θ t1 t2 a=1 ua

– 10 –
 1 1
A1 A2
W1 W2

P1 P2
2n 2n−2 2n 2n−2

Qa Ψ̂1 R̃1 R̃2 Ψ̂2

M1 M2
4 1 2 Q3,4 1

1 1 1 1
Ã1 Ã2
W1 W2
Q̃1 Q̃2
P̃1 P̃2
2n−2 2n−2 2n−2 C2 2n−2 C1

Λ2 Λ1
N2 N1
L2 L1 L2 L1
1 2 1 1 2 1
M1 Q̃3,4 M2 M1 Q̃3,4 M2

Figure 3: Deconfinement steps of SU(2n) with two antisymmetrics and four fundamentals.

where we refer to the appendix A for the details on the conventions on the elliptic genus
used here. The collection of θ-functions on the RHS of the identity (3.5) represents the
Fermi Ψ (numerator) and the chirals Tj (denominator) in the dual LG theory.
In the following, in order to prove the identity, we will break some of the global
symmetries by moving all the theta functions from the denominator of the RHS to the
numerator of LHS of (3.5). We will further employ the relation

θ(x) = θ(q/x) . (3.6)

At the field theory level this operation corresponds to adding a J-term on the gauge
theory side by setting a gauge invariant combination of chiral fields (the one that we
have moved from the RHS to the LHS) to zero using the equations of motion. The
net effect of this operation on the elliptic genus consists of adding a Fermi on the
gauge theory side and of removing a chiral on the LG side. We denote the new Fermi
that appears on the LHS as Fermi flipper, borrowing the terminology used in higher
dimensional theories with four supercharges for chiral multiplets.

– 11 –
 In this case we also redefine the singlets Tn , Tn−1 and Tn−2 as Tn−j,j , for j =
0, . . . , n−1, Tab
n−j−1,j for j = 0, . . . , n and 1 ≤ a < b ≤ 4 and Tn−j−2,j for j = 0, . . . , n−2
respectively. This redefinition is quite useful in the following, because we will explicit
break the SU(2) flavor symmetry and expressing the fields as T allows us to identify
the U(1) ⊂ SU(2) charges. Furthermore, the Fermi Ψ is redefined as Ψ2n−2−j,j with
j = 0, . . . , 2n − 2 accordingly. The electric theory is then modified by flippling all the
combinations T, which corresponds to adding the following J-terms

ΨTn−j,j = An−j j
1 A2 , j = 0, . . . , n ,
ΨTab
n−j−1,j
= An−j−1
1 Aj2 Qa Qb , j = 0, . . . , n − 1, 1 ≤ a < b ≤ 4 , (3.7)
ΨTn−j−2,j = An−j−2
1 Aj2 Q1 Q2 Q3 Q4 , j = 0, . . . , n − 2 .

In the dual theory, we are then left with only the 2n − 1 free Fermi Ψ2n−2−j,j .
Then the crucial aspect of the analysis consists of finding an auxiliary dual quiver
gauge theory with new gauge groups and new bifundamentals, where the original an-
tisymmetric fields A1,2 are absent. In the case at hand, such an auxiliary quiver corre-
sponds to the second one in Figure 3.
There are two symplectic gauge groups USp(2n−2)1,2 and each one is characterized
by one USp/SU bifundamental P1,2 , one fundamental W1,2 and one Fermi Ψ̂1,2 . There
are also two new SU(2n) fundamentals R̂1,2 and two J-terms

JΨ̂1,2 = R̃1,2 P1,2 + W1,2 M1,2 . (3.8)

Observe that the procedure manifestly breaks the SU(4) × SU(2) global symmetry.
2
The original fields A1,2 and Q1,2 correspond here to the combinations P1,2 and P1,2 W1,2
respectively. In addition, the J-terms (3.7) are modified as

ΨTn−j,j = P12n−2j P22j , j = 0, . . . , n ,

2(n−j−1)
ΨT1,2 = P1 P22j P1 W1 P2 W2 , j = 0, . . . , n − 1 , (3.9)
n−j−1,j
2(n−j−1)
ΨTℓ,a = P1 P22j Pℓ Wℓ Qa , j = 0, . . . , n − 1, ℓ = 1, 2, a = 3, 4 ,
n−j−1,j
2(n−j−1)
ΨT3,4 = P1 P22j Q3 Q4 , j = 0, . . . , n − 1 ,
n−j−1,j
2(n−j−2)
ΨTn−j−2,j = P1 P22j P1 W1 P2 W2 Q3 Q4 , j = 0, . . . , n − 2 .

The original quiver in Figure 3 can be recovered if we use the duality reviewed
in subsection 2.1, provided we remove an extra non-anomalous axial symmetry that is
allowed by the field content.
As we stressed above, we do not have a dynamical mechanism that removes such
a symmetry. A similar problem arises in 3d, where the extra symmetry is removed by

– 12 –
the presence of a (linear) monopole superpotential. The procedure carried out here
was indeed applied in [22] to the analysis of a 3d duality with the same electric matter
content7 found in [27], where in the auxiliary quiver two linear monopole superpotentials
terms were added to the theory in order to remove the axial symmetries. This analogy
is helpful in order to justify the existence of a dynamical constraint also in 2d, because
the models discussed here can be obtained from the 3d ones as boundary dualities along
the lines of the construction in [29]. If we apply such construction to the auxiliary quiver
we then have to consider the contribution of the boundary monopoles, that forbids the
generation of the axial symmetry in the half index. Here a similar constraint exists
also at the level of the elliptic genus. This last observation descends from the reduction
of the S 2 × T 2 partition function to the elliptic genus, where the choice of R charges
discussed in subsection 2.1, prevents the extra symmetry to be generated.
Having this caveat in mind we can proceed by dualizing the SU(2n) gauge node.
This model has 4n − 2 fundamentals and 2 antifundamentals and the necessary duality
has been reviewed in subsection 2.3. Here we consider the case with y = 0.
The model obtained in this case is represented in the third quiver of figure 3. The
non-trivial aspect of this model is that the constraints that we imposed on the axial
symmetries for the USp(2n−2) nodes above are not required here anymore. The reason
is that there are linear J-term in the dual phase that remove some of the Fermi and
chiral fields, and non-anomalous charges in the leftover field content are exactly the
ones visible in the matching of the elliptic genera.
In this case the J-terms for the Fermi Λ1,2 obtained in the duality are
X
JΛ1 = Q̃a L1,a + P̃1 M1 W1 + N1 P̃2 ,
a=3,4
X
JΛ2 = Q̃a L2,a + P̃2 M2 W2 + N2 P̃1 . (3.10)
a=3,4

The last step in Figure 3 corresponds to confining the two USp(2n − 2) gauge
nodes. In this case, they both have 2n fundamentals and the dual models are LG
2
theories, where the two combinations P̃1,2 are conjugated antisymmetric chirals. There
are also four SU(2n − 2) antifundamentals, and the J-terms obtained after solving the
equation of motion are
X 
JΨ(0) = Q̃1 Ãn−2
1 Q̃ a L2,a + M2 Q̃2 + C1 PfÃ1 , (3.11)
1
a=3,4
X 
JΨ(0) = Q̃2 Ãn−2
2 Q̃a L1,a + M1 Q̃1 + C2 PfÃ2 . (3.12)
2
a=3,4

7
More precisely the duality that was proved in [22] through tensor deconfinement had 4 fundamental
and two antisymmetric chirals 3d N = 2.

– 13 –
Notice that we have denoted the Fermi flippers, arising from the gauge/LG duality of
(0) (0)
the symplectic gauge nodes discussed in subsection 2.2, as Ψ1 and Ψ2 , as they will
constitute the extremal fields in the Fermi tower denoted as Ψ in equation (3.1).
In this way we have obtained a duality between the original SU(2n) model with
two antisymmetric and four fundamental chirals, and an SU(2n − 2) model with two
conjugate antisymmetric and four antifundamental chirals, in addition to a set of Fermi
and chiral singlets.
We now iterate the procedure until we reach an SU(2n − 4) gauge theory with
the same charged field content of the original one, i.e. two antisymmetric and four
fundamental chirals. In this way we have a simpler relation between the elliptic genera,
that can be recursively applied to obtain the final relation between the original model
and an LG model. Such relation is
 1 1

(4,·,·,2,·) ⃗ (4,·,·,2,·) ⃗
ISU(2n) (⃗u; ·; ·; t; ·) = ISU(2n−4) (t1 t2 ) ⃗u; ·; ·; (t1 t2 ) t; ·
n−2 n−2

4n−4
Q4
2 4n−6
Q4

θ q/(t2,1 a=1 ua ) θ q/(t1,2 t2,1 a=1 ua )
×
2n−4
Q4
Q 2n−2

. (3.13)
θ t2n 1,2 θ t1,2 a=1 ua a<b θ t1,2 ua ub

The relation is derived by applying the various steps of deconfinement, duality and
reconfinement discussed above and summarized graphically in Figure 3 twice. The
relations used to derive (3.13) have been summarized in Section 2 and they correspond
to (2.4), (2.11) and (2.5) respectively. Observe that we kept the chiral fields unflipped
in the relation (3.13), while they appeared as Fermi fields in the field theory discussion
in the SU(2n) side.
Each term in the second line of the (3.13) corresponds to one of the expected
singlets of the duality, in which the procedure has apparently broken the global SU(2)
non-abelian global symmetry8 . More concretely:

• θ(t2n n
1,2 ) corresponds to the combinations A1,2

• θ(t2n−4
Q4 n−2
1,2 a=1 ua ) corresponds to A1,2 Q1 Q2 Q3 Q4

• θ(t2n−2 n−1
Q
a<b 1,2 ua ub ) corresponds to A1,2 Qa Qb , antisymmetric in the SU(4) indices

(0) (0)
• θ q/(t4n−4
Q4

2,1 a=1 ua ) correspond to the two Fermi Ψ2n−2,0 and Ψ0,2n−2

(0) (0)
• θ q/(t21,2 t4n−6
Q4

2,1 a=1 ua ) correspond to the two Fermi Ψ2n−1,1 and Ψ1,2n−1 .
8
Also the SU(4) global symmetry is broken by our procedure, but it is actually recovered in the
SU(2n − 4) theory.

– 14 –
The (0) superscript in the Fermi fields indicates that they have been created at the
zero-th iteration of the procedure.
A recursive application of this relation leads to an SU(4) gauge theory model for
even n or to an SU(2) gauge theory for odd n. In general, at the j-th step we obtain
an SU(2n − 4j) gauge theory, and the leftover integral is
 j j

(4,·,·,2,·) ⃗
ISU(2n−4j) (t1 t2 ) n−2j ⃗u; ·; ·; (t1 t2 ) n−2j t; · . (3.14)

The Fermi and chiral singlets generated at this step contribute as

2(n−j+1) 2(j−1)
• θ(t1,2 t2,1 ) corresponding to the combinations An−j+1
1,2 Aj−1
2,1

2(n−j) 2(j−1)
• n−j j−1
Q
θ(t1,2 t2,1 ua ub ) corresponding to A1,2
a<b A2,1 Qa Qb , antisymmetric in the
SU(4) indices
2(n−j−1) 2(j−1) Q4
• θ(t1,2 t2,1 n−j−1 j−1
a=1 ua ) corresponding to A1,2 A2,1 Q1 Q2 Q3 Q4
4(n−j) 4(j−1) Q4 (j) (j)
• θ(q/(t1,2 t2,1 a=1 ua )) associated to the two Fermi Ψ2n−2j,2j−2 and Ψ2j−2,2n−2j
2(2j−1) 2(2n−2j−1) (j)
• θ(q/(t1,2
Q4
t2,1 a=1 ua )) corresponding to the two Fermi Ψ2n−2j−1,2j−1
(j)
and Ψ2j−1,2n−2j−1 .

The Fermi generated at this stage constitute another bit of the tower of states Ψ.
In order to complete the proof of (3.5) we must distinguish the cases n = 2m and
n = 2m + 1 as discussed above.

The case of SU(4m)

In this case we can use the recurrence relations until we get to an SU(4) gauge the-
ory with two antisymmetric A1,2 and four fundamentals Q1,2,3,4 . Their fugacities are
m−1 m−1
(t1 t2 ) 2 t1,2 and (t1 t2 ) 2 ua respectively.
This model is dual (see [9]) to an LG theory with chirals φi that contribute to the
elliptic genus as

• φ1 = A2 → a<b θ t2m−2


t2m
Q
1 2 ua ub

• φ2 = PfA1 → θ t2m+2 t2m−2

1 2

• φ3 = A1 Q2 → a<b θ t2m 2m−2

Q

1 t2 ua ub

• φ4 = Q4 → θ (t1 t2 )2(m−1) 4a=1 ua

Q

• φ5 = PfA2 → θ t2m−2 t2m+2

1 2

– 15 –
 • φ6 = A1 A2 → θ ((t1 t2 )2m ).

There are in addition three Fermi singlets, whose charges (and contribution to the
elliptic genus) can be read from the J-terms9 . We have

4(m−1)
• JΨ11 ⊃ φ21 → θ(q/(t4m
Q4
1 t2 a=1 ua ))

4(m−1) 4m
• JΨ33 ⊃ φ23 → θ(q/(t1
Q4
t2 a=1 ua ))

2(2m−1) 2(2m−1)
• JΨ13 ⊃ φ1 φ3 → θ(q/(t1
Q4
t2 a=1 ua )) .

These last Fermi singlets are the remaining ones needed for the full Fermi tower Ψ.
Once the contributions arising from the last duality are added to the ones obtained
from the recursive relation above, we restore the original SU(2) symmetry and obtain
the relation (3.5) for n = 2m.

The case of SU(4m + 2)

In order to complete the proof of (3.5) we need to discuss the case SU(4m + 2) as well.
In this case we arrive at an SU(2) gauge theory with four fundamentals Q1,2,3,4 with
fugacities (t1 t2 )2m ua after m iterations. The two antisymmetric are singlets and they
2(m+1) 2(m−1)

contribute to the elliptic genus as θ t1,2 t2,1 . This gauge theory is dual to an LG
model, where the singlets are given by the SU(4) two-index antisymmetric chiral Qa Qb


and a Fermi. The contribution of the chiral to the elliptic genus is a<b θ (t1 t2 )2m ua ub ,
Q

while the Fermi contributes as θ q/((t1 t2 )4m 4a=1 ua ) , compatibly with the J-terms
Q

that would be generated in the duality in absence of flippers in the original SU(2n)
model. Again, once the contributions of the last duality are added to the ones obtained
from the recursive relation above, we restore the original SU(2) symmetry and obtain
the relation (3.5) for n = 2m + 1 as well.
This concludes the proof of the identity (3.5). Observe that at the level of the
interaction we end up with a model with only Fermi multiplets, as we have flipped all
the chirals Tj in the original gauge theory. In the following, we will actually use the
un-flipped (or partially flipped) duality in order to prove that the other cases of SU(N )
with two antisymmetric chirals, nf < 4 fundamental chirals and na > 1 antifundamental
chirals are dual to LG models.
9
Here we have flipped all the chiral combinations Tn,n−1,n−2 . Then, the LG model will only have
free Fermi multiplets and vanishing J-terms. Nevertheless we can read their charges from the un-
flipped duality discussed in [9].

– 16 –
3.2 SU(2n + 1) with four fundamentals
In this case the dual LG model has two chiral gauge invariant operators corresponding
to chiral fields interacting through a J-term with a Fermi multiplet. The field content
of the gauge theory and of the dual LG model is represented in the following table

SU(2n + 1) SU(2) SU(4) U(1)A U(1)Q U(1)R0

A □ □ · 1 0 0
□
Q □ · □ 0 1 0 (3.15)
Tn · ⊗nsym □ □ n 1 0
n−1
Tn−1 · ⊗sym □ □ n−1 3 0
2n−1
Ψ · ⊗sym □ · −2n + 1 −4 2

The chirals Tj correspond to the following gauge invariant combinations of the charged
fields
Tn = An Q , Tn−1 = An−1 Q3 , (3.16)
while the J-term is
JΨ = Tn Tn−1 . (3.17)
We have computed the ’t Hooft anomalies in the two phases and we have observed that
they match. Explicitly we have
n(2n+1)
κSU(2)2 = 2
, κAA = −κAR0 = 2n(2n + 1) ,
2n+1
κSU(4)2 = 2
, κQQ = −κQR0 = 8n + 4 , (3.18)
κR0 R0 = 6n + 4 , κAQ = 0 .

In the following we will provide a further check of this duality by showing that the
elliptic genera of the two models match, provided the validity of other identities that
already appeared in the literature, and that descend from the reduction of 4d dualities
to 2d. The derivation is very similar to the one spelled out above, and for this reason
we will be more sketchy and refer the reader to the derivation in subsection 3.1 for
further details. The identity that we need to prove in order to check the validity of the
duality is
Q2n−1   2(2n−1−j) 2j Q4 
(4,·,·,2,·) j=0 θ q/ t1 t2 a=1 ua
ISU(2n+1) = Q Q  Q Qn−1  2(n−1−j) 2j  . (3.19)
4 n 2(n−j) 2j 4
a=1 j=0 θ t1 t2 u a a<b<c j=0 θ t1 t2 u a u b u c

Again, we flip all the terms in the denominator of the RHS of (3.19) by moving the
theta functions to the numerator of LHS and by using the relation (3.6). These theta

– 17 –
functions are associated to Fermi flippers for the operators Tn and Tn−1 on the gauge
theory side. Then we re-define Tn and Tn−1 as Tan−j,j , for j = 0, . . . , n and a = 1, . . . , 4
and Tabc
n−j−1,j for j = 0, . . . , n − 1 and 1 ≤ a < b < c ≤ 4 and the Fermi Ψ is redefined
as Ψ2n−1−j,j with j = 0, . . . , 2n − 1 accordingly.

A1 A2
P1 P2
2n−2 2n+1 2n−2

2n+1
Ψ̂1 R̃1 R̃2 Ψ̂2
Q Q
1 4 1
4

Ã1 Ã2
P̃1 P̃2
2n−2 2n−1 2n−2

2n−1
N2 Λ2 Λ1 N1

1 4 1 Q̃
L2 L1
Q̃ 1 4 1
L1 L2

Figure 4: Deconfinement steps of SU(2n+1) with two antisymmetrics and four fundamentals.

The electric theory is modified by the addition of the following J-terms

ΨaTn−j,j = An−j j
1 A2 Qa , j = 0, . . . , n, a = 1, . . . , 4 ,
ΨTabc
n−j−1,j
= An−j−1
1 Aj2 Qa Qb Qc , j = 0, . . . , n − 1, 1 ≤ a < b < c ≤ 4 . (3.20)

In the dual theory we are then left only with the free Fermi fields Ψ2n−1−j,j .
We proceed by finding an auxiliary quiver, shown in Figure 4, where the two
antisymmetric chirals A1,2 are traded with two USp(2n − 2) auxiliary gauge groups. In
this case the process does not break the SU(4) global symmetry, differently from the
case studied in subsection 3.1.
There are two USp/SU bifundamental P1,2 , two SU(2n + 1) fundamentals R̃1,2 and
two Fermi fields Ψ̂1,2 . There are also two J-terms JΨ̂1,2 = R̃1,2 P1,2 . The original fields
2
A1,2 correspond here to the combinations P1,2 . In addition, the J-terms (3.20) are
modified as
2(n−j−1)
ΨTan−j,j = P12n−2j P22j Qa , ΨTabc
n−j−1,j
= P1 P22j Qa Qb Qc . (3.21)

– 18 –
The original quiver in Figure 4 can be recovered if we exploit the duality reviewed in
subsection 2.1 provided that we remove an extra, non-anomalous, axial symmetry that
is allowed by the field content.
This requirement is translated into a constraint imposed on the elliptic genus, which
prevents fugacities for the axial symmetries to be turned on both for the chirals and the
Fermi charged under the two USp(2n − 2) factors. Again, we interpret the constraint
as in the discussion of subsection (3.1).
We proceed by dualizing the SU(2n+1) gauge node. This model has 4n fundamen-
tals and 2 antifundamentals and the duality we need has been reviewed in subsection
2.3. Here we consider the case with y = 0. The model obtained in this case is rep-
resented in the third quiver of Figure 4. Again, the constraints imposed on the axial
symmetries for the USp(2n − 2) nodes above are not required here anymore. In this
case the J-terms for the Fermi Λ1,2 obtained in the duality are

JΛ1 = L1 Q̃ + N1 P̃2 , JΛ2 = L2 Q̃ + N2 P̃1 . (3.22)

The last step in Figure 4 corresponds to confine the two USp(2n − 2) gauge nodes.
In this case they are both characterized by 2n fundamentals, and the dual models are
2
LG theories, where the two combinations P̃1,2 define conjugated antisymmetric chirals.
There are also four SU(2n−2) antifundamentals, and the J-terms obtained after solving
the equation of motion are

JΨ(0) = Ãn−1
1 L2 Q̃ , JΨ(0) = Ã2n−1 L1 Q̃ . (3.23)
1 2

Again we have denoted the Fermi flippers, arising from the gauge/LG duality of
(0) (0)
the symplectic gauge nodes discussed in subsection 2.2, as Ψ1 and Ψ2 , as they will
constitute the extremal fields in the Fermi tower denoted as Ψ in equation (3.15).
In this way we have obtained a duality between the original SU(2n + 1) model with
two antisymmetric and four fundamental chirals and an SU(2n − 1) model with two
conjugate antisymmetric and four antifundamental chirals, in addition to a set of Fermi
and chiral singlets.
At this point of the discussion we iterate the procedure until we reach an SU(2n −
3) gauge theory with the same charged field content of the original one, i.e. two
antisymmetric and four fundamental chirals. In this way we get a simpler relation
between the elliptic genera, that can be recursively applied in order to obtain the final
relation between the original model and an LG model. Such relation is obtained by
applying the relations (2.4), (2.11) and (2.5), and it is
 2 2

(4,·,·,2,·) (4,·,·,2,·)
ISU(2n+1) (⃗u; ·; ·; ⃗t; ·) = ISU(2n−3) (t1 t2 ) 2n−3 ⃗u; ·; ·; (t1 t2 ) 2n−3 ⃗t; ·

– 19 –
    
2(2n−1) Q4 2(2n−2) 2 Q4
θ q/(t1,2 a=1 ua ) θ q/(t1,2 t2,1 a=1 ua )
× Q4 2n

Q 2n−2

.(3.24)
a=1 θ t1,2 ua a<b<c θ t1,2 ua ub uc

Analogously to the case studied in subsection 3.1 we keep the chiral fields unflipped
at the level of the elliptic genus, while they appear as Fermi fields in the field theory
discussion in the SU(2n + 1) side. Each term in the second line of the (3.24) corre-
sponds to one of the expected singlets of the duality. These are consistent with the
deconfinement procedure which apparently broke the global SU(2) non-abelian global
symmetry. We leave to the interested reader the details of the matching in this case.
A recursive application of this relation leads to an SU(5) gauge theory model for
even n or to an SU(3) gauge theory for odd n. In general, at the j-th step we obtain
an SU(2n + 1 − 4j) gauge theory, and the leftover integral is
 2j 2j

(4,·,·,2,·) ⃗
ISU(2n−4j+1) (t1 t2 ) 2n+1−4j ⃗u; ·; ·; (t1 t2 ) 2n+1−4j t; · . (3.25)

In addition, the Fermi and chiral singlets generated at this step can be worked out
from the relation (3.24) in complete analogy to what was done in subsection 3.1. The
details of the derivation are straightforward, thus we omit them.
To conclude the proof of (3.19), we must distinguish the cases n = 2m and n =
2m + 1.

The case of SU(4m + 1)

In this case we can use the recurrence relations until we obtain an SU(5) gauge the-
ory with two antisymmetric A1,2 and four fundamentals Q1,2,3,4 . Their fugacities are
2(m−1) 2(m−1)
(t1 t2 ) 5 t1,2 and (t1 t2 ) 5 ua respectively.
Now, we can further deconfine the two antisymmetric tensors A1,2 , obtaining two
USp gauge groups. By repeating the various steps in Figure 4 we get an SU(3) gauge
theory, where the two conjugate antisymmetric correspond to two extra fundamentals.
The SU(3) gauge theory has then two fundamentals and four antifundamentals and it
is dual to an LG model (see [9]). Translating the discussion above to the corresponding
identities among elliptic genera, we find the expected relation (3.19) in the case of
n = 2m.

The case of SU(4m + 3)

In order to complete the proof of (3.19) we need to discuss the case SU(4m + 3) as well.
In this case we get, after m iterations, an SU(3) gauge theory with four fundamentals
2m
Q1,2,3,4 with fugacities (t1 t2 ) 3 ua and two antisymmetric chirals, corresponding to two
2m
extra antifundamentals, with fugacities (t1 t2 ) 3 t1,2 . Again, the SU(3) gauge is dual to

– 20 –
an LG theory (see [9]) and we find the expected identity (3.19) in the case of n = 2m+1
as well.
This concludes the proof of the identity (3.19). In the end we obtained a model
with only Fermi multiplets, as all the chirals Tj in the original gauge theory have been
flipped. In the following we will actually use the un-flipped (or partially flipped) duality
in order to prove the cases of SU(2n + 1) with two antisymemtric chirals and (4 − j, j)
pairs of fundamentals and antifundamental.
In the next subsections we study the other cases, in which we decrease the number of
fundamentals and increase the number of antifundamentals. As a general comment we
will see that the derivation in such cases is simpler than above, because the proof does
not require an iterative construction. Indeed after deconfining the two antisymmetric
tensors and dualizing the unitary and the symplectic gauge nodes we will get SU(N )
models with two conjugate antisymmetric and four antifundamental chirals, possibly in
addition to a non-trivial set of J-terms. Such models are (modulo charge conjugation)
exactly the ones studied so far. It follows that we can use the results of subsection
3.1 and 3.2 in order to prove that the other SU(N ) models with two antisymmetric
chirals, nf < 4 fundamental chirals and na > 1 antifundamental chirals are dual to LG
theories.

3.3 SU(2n) with three fundamentals and one antifundamental

In this case the dual LG model has five chiral gauge invariant operators corresponding
to chiral fields interacting through a J-term with a Fermi multiplet. The field content
of the gauge theory and of the dual LG model is represented in the table below.

SU(2n) SU(2) SU(3) U(1)A U(1)Q U(1)Q̃ U(1)R0

A □ □ · 1 0 0 0
□
Q □ · □ 0 1 0 0
Q̃ □ · · 0 0 1 0
M · · □ 0 1 1 0 (3.26)
Tn · ⊗nsym □ · n 0 0 0
Tn−1 · ⊗n−1
sym □ □ n−1 2 0 0
Pn · ⊗n−2
sym □ □ n 1 1 0
Pn−1 · ⊗n−3
sym □ · n−1 3 1 0
2n−3
Ψ · ⊗sym □ · −2n + 1 −3 −1 2

The chirals M , Tj , Pj correspond to the following gauge invariant combinations of

– 21 –
the charged fields

M = QQ̃, Tn = An , Tn−1 = An−1 Q2 , Pn = An−1 (AQ̃)Q, Pn−1 = An−2 (AQ̃)Q3 ,

(3.27)
while the J-term is
JΨ = M Tn−1 Tn + Tn Pn−1 + Tn−1 Pn . (3.28)

We have computed the ’t Hooft anomalies in the two phases and they match. Explicitly
we have
κAA = −κAR0 = 2n(2n − 1) ,
κSU(2)2 = n(2n−1)
2
,
κQQ = −κQR0 = 6n ,
κSU(3)2 = n , (3.29)
κQ̃Q̃ = −κQ̃R0 = 2n ,
κR0 R0 = 6n + 1 ,
κAQ = κAQ̃ = 0 .
Again, we aim to corroborate the validity of the duality by matching the elliptic
genera using other relations already derived in the literature. In this case, differently
from the cases above, we will not flip all the chirals in the dual LG description, but
only the meson M . We will see that the expected J-term JΨ can be reconstructed
with our procedure starting from the J-term (3.17) that we expect from the duality of
subsection 3.2.

A1 A2

P1 P2
2n 2n−2 2n 2n−2

Q Q̃ Q Q̃
ΨM ΨM
3 1 3 1

P̃1 P̃2 Ã1 Ã2

2n−2 2n−1 2n−2

q̃ ψ 2n−1
N2
3 1 q̃ p̃
N1
3 1

Figure 5: Deconfinement steps of SU(2n) with two antisymmetrics, three fundamentals and
one antifundamental.

– 22 –
 The identity that we want to prove is
Q2n−3
θ q/(tj+1 2n−j−2


(3,1,·,2,·) j=0 1 t2 u1 u2 u3 v)
ISU(2n) (⃗u; v; ·; ⃗t; ·)
= Q3 Qn−2 n−1−j j+1

(3.30)
a=1 θ (ua v) j=0 θ t1 t2 ua v
1
× Qn
Qn−1 Q
.
j n−j j n−j−1
ua ub · n−3 n−2−j j+1

Q
j=0 θ t1 t2 · j=0 1≤a<b≤3 θ t1 t2 j=0 θ t1 t2 u 1 u2 u3 v

In this case, in order to prove this identity through the techniques used above, we
will flip the chirals combinations QQ̃ and PfA1,2 . The first one corresponds to the
singlet M in the J-term (3.28), while the other descends from the singlet Tn , and such
flip breaks the SU(2) flavor symmetry.
In the following we will redefine the singlets Tn , Tn−1 , Pn and Pn−1 as Tn−j,j with
j = 0, . . . , n, Tn−j−1,j with j = 0, . . . , n − 1, Pn−j−1, j+1 with j = 0, . . . , n − 2 and
Pn−j−2, j+1 with j = 0, . . . , n − 3. Accordingly the Fermi Ψ becomes Ψ−j−1,−2n+j+2
with j = 0, . . . , 2n − 3.
The Fermi flippers of the combinations QQ̃ and PfA1,2 give origin to the following
J-terms in the original SU(2n) gauge theory

JΨM = QQ̃ , JΨA1,2 = PfA1,2 . (3.31)

We now proceed by trading the two antisymmetric with two USp(2n − 2)1,2 gauge
groups, each one connected to the original SU(2n) gauge group through a bifundamental
P1,2 . In this case the flippers ΨA1,2 disappear and we are left with only JΨM . Observe
that in this case the original theory is recovered by dualizing the two symplectic gauge
theory using the duality between USp(2n) with 2n + 2 fundamentals Q and the LG
model characterized by the singlets M = Q2 and ψ0 and J-term Jψ0 = Pf M . This
duality was originally proposed in [1] and further studied in [7, 32]. We have reviewed
the basic aspects of the duality in subsection 2.2.
The SU(2n) gauge theory has 4n − 1 fundamentals and one antifundamental. It
can be dualized according to the rules explained in subsection 2.3, and it gives origin
to the third quiver in Figure 5. In this case the J-term is given by Jψ = N2 P̃2 + N1 P̃1 .
The last step consists of dualizing the two USp(2n − 2) groups, each one with 2n chiral
fundamentals to an LG model. The final quiver is the fourth one in figure 5. In this case
(0) n−1
there are two J-terms associated to two Fermi singlets Ψ1,2 . They read JΨ(0) = Ã1,2 p̃.
1,2
We ended up with an SU(2n − 1) gauge theory with four antifundamentals and
two conjugate antisymmetric. This is, up to a conjugation, the same field content of
the confining duality studied in subsection 3.2. In addition we have the non-vanshing
J-terms JΨ(0) .
1,2

– 23 –
 The theory is then dual to an LG model where the J-term (3.17) is modified by
symmetry breaking pattern implied by the presence of JΨ(0) .
1,2
In order to take such breaking into account let us consider an SU(2N + 1) gauge
theory with four antifundamentals Q̃ = {q̃1 , q̃2 , q̃3 , p̃} and two conjugate antisymmetric
tensors à = {Ã1 , Ã2 }. The singlets of the unbroken phase are T̃N and T̃N −1 and the 2d
superpotential10
2N
X −1 X
N N
X −1
(a) (bcd)
W2d = |ϵabcd |ψj−2N +1,−j T̃k,N −k T̃ℓ,N −1−ℓ δj+k+ℓ,2N −1 , (3.32)
j=0 k=0 ℓ=0

where we used the notations of 3.2 for the SU(2) and SU(4) indices. Then we fix
N = n − 1 and consider the effect of JΨ(0) . The singlets T̃n−1 and T̃n−2 are split as
1,2

(a) (0)
T̃j,n−1−j = Ãj1 Ãn−1−j
2 q̃a , T̃j,n−1−j = Ãj1 Ãn−1−j
2 R̃− ,
(0) (ab)
(3.33)
T̃j,n−2−j = Ãj1 Ãn−2−j
2 q̃1 q̃2 q̃3 , T̃j,n−2−j = Ãj1 Ãn−2−j
2 q̃a q̃b R̃− .
In this way the superpotential of the LG for the deformed theory is
2n−3
XX n−2
n−1 X
(a) (bc) (0) (0)
W2d = ψj−2n+3,−j (|ϵabc |T̃k,n−1−k T̃ℓ,n−2−ℓ + T̃k,n−1−k T̃ℓ,n−2−ℓ )δj+k+ℓ,2n−3
j=0 k=0 ℓ=0
(0) (0) (0) (0)
+ ψ1 T̃n−1,0 + ψ2 T̃0,n−1 , (3.34)
where the deformation in the second line can be integrated out using the equations of
motion.
The superpotential (3.34) is equivalent to the one that can be read from the J-term
(3.28) once the dictionary between the singlets T̃ and the singlets (3.27) is specified
and after setting the combinations QQ̃ and PfA1,2 in (3.28) to zero.

(0) (ab) (c)

Tn−1−j, j+1 = T̃j, n−2−j , Tn−1−j, j = |ϵabc |T̃j, n−1−j ,
(a) (bc) (0)
Pn−1−j, j+1 = |ϵabc |T̃j, n−1−j , Pn−1−j, j = T̃j ,n−1−j , (3.35)
Ψ−j−1, j−2n+2 = ψj−2n+3, −j
The same strategy explained in the field theory discussion can be applied at the
level of the elliptic genus. In this case the four steps in Figure 5 can be reproduced at
the level of the elliptic genus as in the previous cases of SU(N ) with two antisymmetric
and four fundamentals. The last step corresponds to plugging the identity (3.19) inside
the elliptic genus of the SU(2n − 1) gauge theory. In this way we obtain the relation
(3.30) as expected.
10
Here we use the notion of 2d superpotential instead of the one of J-terms because the structure
of the former is more intuitive

– 24 –
3.4 SU(2n + 1) with three fundamentals and one antifundamental
In this case the dual LG model has five chiral gauge invariant operators corresponding
to chiral fields interacting through a J-term with a Fermi multiplet. The field content
of the gauge theory and of the dual LG model is represented in the following table

SU(2n + 1) SU(2) SU(3) U(1)A U(1)Q U(1)Q̃ U(1)R0

A □ □ · 1 0 0 0
□
Q □ · □ 0 1 0 0
Q̃ □ · · 0 0 1 0
M · · □ 0 1 1 0 (3.36)
Tn · ⊗nsym □ □ n 1 0 0
Tn−1 · ⊗n−1
sym □ · n−1 3 0 0
n−1
Pn+1 · ⊗sym □ · n+1 0 1 0
Pn · ⊗n−2
sym □ □ n 2 1 0
2n−2
Ψ · ⊗sym □ · −2n −3 −1 2

We have computed the ’t Hooft anomalies in the two phases and they match. Explicitly:

n(2n+1)
κAA = −κAR0 = 2n(2n + 1) ,
κSU(2)2 = 2
,
2n+1
κQQ = −κQR0 = 6n + 3 ,
κSU(3)2 = 2
, (3.37)
κQ̃Q̃ = κQ̃R0 = 2n + 1 ,
κR0 R0 = 6n + 4 ,
κAQ = κAQ̃ = 0 .

The chirals M , Tj and Pj correspond to the following gauge invariant combinations of

the charged fields

M = QQ̃ , Tn = An Q , Tn−1 = An−1 Q3 , Pn+1 = An (AQ̃) , Pn = An−1 (AQ̃)Q2

(3.38)
while the J-term is
JΨ = M Tn2 + Tn−1 Pn+1 + Tn Pn . (3.39)

Again we aim to corroborate the validity of the duality by matching the elliptic
genera using other relations already derived in the literature. In this case we will flip
the meson M and the combinations An1 Q1 and An2 Q2 in the dual LG and we show that
the expected J-term in (3.39) can be reconstructed by our procedure using the J-term
expected from the duality of subsection 3.1. The derivation is very similar to the one
discussed in subsection 3.3, and for this reason we will omit many technical details in
this case.

– 25 –
 A1 A2

P1 P2
2n+1 2n 2n+1 2n

Q1,2 Q3 Q̃ R1 Q3 Q̃ R2
ΨM3 ΨM3
2 1 1 1 1 1 1

1 1 Ã1 Ã2

R1 R2
P̃1 P̃2 q̃1 q̃2
2n 2n 2n 1 2n 1

q̃3 ψ q̃3 q̃
N2 M2
1 1 1 1
N1 M1

Figure 6: Deconfinement steps of SU(2n + 1) with two antisymmetrics, three fundamentals

and one antifundamental.

The identity we want is

Q2n−2 j+1 2n−j−1
(3,1,·,2,·) j=0 θ(q/(t1 t2 u1 u2 u3 v))
ISU(2n+1) (⃗u; v; ·; ⃗t; ·) = Q3 Qn Q j n−j (3.40)
a=1 θ(ua v) j=0 a θ(t1 t2 ua )
1
× Qn−1 j n−j−1 Qn−2 .
u1 u2 u3 )θ(tn−j j+1
θ(tn−j−1 tj+1
Q
j=0 (θ(t1 t2 1 t2 v)) j=0 a<b 1 2 ua ub v)

In order to prove this relation we deconfine the two antisymmetric as in Figure 6,

obtaining two USp(2n) gauge nodes, two bifundamentals P1,2 and two new fundamen-
2
tals R1,2 . The original antisymmetric chirals A1,2 correspond to the combinations P1,2 .
Furthermore, two out of the three original fundamentals Q1,2 correspond in this quiver
to the composites P1,2 R1,2 . The J-term flipping the operator M splits here as

JΨM1 = Q̃P1 R1 , JΨM2 = Q̃P2 R2 , JΨM3 = Q̃Q3 , (3.41)

signalling the explicit breaking of the SU(3) flavor symmetry in the process.
The SU(2n + 1) gauge theory has 4n + 1 fundamentals and one antifundamental.
It can be dualized according to the rules explained in subsection 2.3, and it gives
origin to the third quiver in Figure 6. In this case the new J-term generated by the

– 26 –
duality is given by Jψ = N2 P̃2 + N1 P̃1 . Furthermore, the two J-terms for ΨM1,2 become
JΨM1,2 = Q̃N1,2 . The last step consists of dualizing the two USp(2n) groups, each one
with 2n + 2 chiral fundamentals to an LG. The final quiver is the fourth one in Figure
(0)
6. In this case there are two J-terms associated to two Fermi singlets Ψ1,2 . They read
JΨ(0) = Ãn−1
1 q̃1 q̃ and JΨ(0) = Ãn−1
2 q̃2 q̃.
1 2
Again we are in the situation described in subsection 3.3, where after deconfining
the tensors, dualizing the unitary gauge group and confining the tensors back, we have,
up to an overall charge conjugation, a model already described in 3.1. We can the use
the duality of such model with an LG one in order to prove the validity of (3.40) and to
reconstruct (up to flippers) the expected J-term (3.39). The details of the derivation
are straightforward and we leave them to the interested reader.

3.5 SU(2n) with two fundamentals and two antifundamentals

In this case the dual LG model has seven chiral gauge invariant operators corresponding
to chiral fields interacting through a J-term with a Fermi multiplet. The field content
of the gauge theory and of the dual LG model is represented in the table below.

SU(2n) SU(2) SU(2) SU(2) U(1)A U(1)Q U(1)Q̃ U(1)R0

A □ □ · · 1 0 0 0
□
Q □ · □ · 0 1 0 0
Q̃ □ · · □ 0 0 1 0
M · · □ □ 0 1 1 0
B · □ · · 1 0 2 0 (3.42)
Tn · ⊗nsym □ · · n 0 0 0
Tn−1 · ⊗n−1
sym □ · · n−1 2 0 0
n−2
Pn · ⊗sym □ □ □ n 1 1 0
Pn+1 · ⊗n−3
sym □ · · n+1 0 2 0
n−4
R · ⊗sym □ · · n 2 2 0
Ψ · ⊗2n−4
sym □ · · −2n −2 −2 2

The chirals B, M , Tj , Pj and R correspond to the following gauge invariant com-

binations of the charged field

Tn = An , Pn = An−1 (AQ̃)Q , M = QQ̃ , R = An−2 (AQ̃)2 Q2 ,

(3.43)
Tn−1 = An−1 Q2 , Pn+1 = An−1 (AQ̃)2 , B = AQ̃2 ,

while the J-term is

JΨ = M 2 Tn2 + M Pn Tn + Tn R + Tn−1 Pn+1 + Pn2 + BTn−1 Tn . (3.44)

– 27 –
The ’t Hooft anomalies match in the two phases and are explicitly
n(2n−1)
κSU(2)2A = 2
, κAA = −κAR0 = 2n(2n − 1) ,
κSU(2)2Q = κSU(2)Q̃2 = n , κQQ = −κQR0 = κQ̃Q̃ = −κQ̃R0 = 4n , (3.45)
κR0 R0 = 6n + 1 , κAQ = κAQ̃ = 0 .
In the following we show the matching of the elliptic genera along the same lines of
the discussion performed in the other examples. In this case we will flip the meson M
and the baryons B, PfA1,2 in the dual LG, and we will show that the expected J-term
in (3.44) can be reconstructed by our procedure using the J-term expected from the
duality of subsection 3.1.
At the level of the elliptic genus the identity associated to this duality is
Q2n−4 j+2 2n−j−2
(2,2,·,2,·) j=0 θ(q/(t1 t2 u1 u2 v1 v2 ))
ISU(2n) (⃗u; ⃗v ; ·; ⃗t; ·) = Q2 Qn j n−j Qn−1 n−1−j j
a,b=1 θ(ua vb )θ(t1,2 v1 v2 ) j=0 θ(t1 t2 ) j=0 θ(t1 t2 u1 u2 )
1
× Qn−4 n−j−2 j+2 Qn−3 n−j−1 j+2 Qn−2 Q2 n−j−1 j+1 .
j=0 θ(t1 t2 u1 u2 v1 v2 ) j=0 θ(t1 t2 v1 v2 ) j=0 a,b=1 θ(t1 t2 ua vb )
(3.46)
In order to prove this relation we deconfine the two antisymmetric as in Figure 7,
obtaining two USp(2n − 2) gauge nodes, with two bifundamentals P1,2 . The original
2
antisymmetric chirals A1,2 correspond to the combinations P1,2 . The J-terms at this
stage are the ones that flip the operators M and B
JΨM = Q̃Q , JΨB1 = Q̃2 P12 , JΨB2 = Q̃2 P22 . (3.47)
The SU(2n) gauge theory has 4n − 2 fundamentals and two antifundamentals. It can
be dualized according to the rules explained in subsection 2.3, giving origin to the
third quiver in Figure 7. In this case the new J-term generated by the duality is
Jψ = N2 P̃2 + N1 P̃1 . In addition, the two J-terms for ΨB1 and ΨB2 become JΨB1 = N12 .
and JΨB2 = N22 respectively.
The last step consists in dualizing the two USp(2n − 2) groups, each one with 2n
chiral fundamentals, to an LG model. The final quiver is the fourth one in Figure 7.
(0)
In this case there are two J-terms associated to two Fermi singlets Ψ1,2 . They read
JΨ(0) = Ãn−2 2
1,2 p̃ .
1,2
Similarly to the cases discussed above we have, up to an overall charge conjugation,
a model already described in 3.1. We can the use the duality of such model with an
LG one in order to prove the validity of (3.46) and to reconstruct (up to flippers) the
expected J-term (3.44). The details of the derivation are straightforward and we leave
them to the interested reader.

– 28 –
 A1 A2
2n−2 2n−2
P1 P2
2n 2n

Q Q̃ Q Q̃
ΨM ΨM
2 2 2 2

2n−2 2n−2
P̃1 P̃2 Ã1 Ã2
2n−2

q̃ ψ N2 2n−2

N1
2 2
q̃ p̃

2 2

Figure 7: Deconfinement steps of SU(2n) with two antisymmetrics, two fundamentals and
two antifundamentals.

3.6 SU(2n + 1) with two fundamentals and two antifundamentals

In this case the dual LG model has six chiral gauge invariant operators corresponding
to chiral fields interacting through a J-term with a Fermi multiplet. The field content
of the gauge theory and of the dual LG model is represented in the table below.

SU(2n + 1) SU(2) SU(2) SU(2) U(1)A U(1)Q U(1)Q̃ U(1)R0

A □ □ · · 1 0 0 0
□
Q □ · □ · 0 1 0 0
Q̃ □ · · □ 0 0 1 0
M · · □ □ 0 1 1 0
B · □ · · 1 0 2 0 (3.48)
Tn · ⊗nsym □ □ · n 1 0 0
Tn−1 · ⊗n−3
sym □ □ · n+1 1 2 0
Pn · ⊗n−1
sym □ · □ n+1 0 1 0
Pn−1 · ⊗n−2
sym □ · □ n 2 1 0
2n−3
Ψ · ⊗sym □ · · −2n − 1 −2 −2 2

– 29 –
 The chirals Tj , Pj correspond to the following gauge invariant combinations of the
charged fields

Tn = An Q , Tn−1 = An−1 (AQ̃)2 Q , Pn = An (AQ̃) , Pn−1 = An−1 (AQ̃)Q2 . (3.49)

In addition there are a meson M = QQ̃ and the baryons B1,2 = A1,2 Q̃2 , while the
expected J-term compatible with the global symmetry is

JΨ = M Tn Pn + BTn2 + Tn Tn−1 + Pn Pn−1 . (3.50)

We have matched the ’t Hooft anomalies in the two phases. Explicitly we have
n(2n+1)
κSU(2)2A = 2
, κAA = −κAR0 = 2n(2n + 1) ,
2n+1
κSU(2)2Q = κSU(2)Q̃2 = 2
, κQQ = −κQR0 = κQ̃Q̃ = −κQ̃R0 = 4n + 2 , (3.51)
κR0 R0 = 6n + 4 , κAQ = κAQ̃ = 0 .

In the following we show the matching of the elliptic genera along the same lines of
the discussion performed in the other examples. In this case we will flip the combina-
tions QQ̃, A1,2 Q̃2 and An1,2 Q and we will show that the expected J-term in (3.50) can
be reconstructed by our procedure starting from the J-term expected from the duality
of subsection 3.2.
At the level of the elliptic genus the identity associated to this duality is
Q2n−3 j+2 2n−j−1
(2,2,·,2,·) j=0 θ(q/(t1 t2 u1 u2 v1 v2 ))
⃗
ISU(2n+1) (⃗u; ⃗v ; ·; t; ·) = Q2 Qn Qn−1 n−j j+1
θ(t1,2 v1 v2 ) a=1 ( j=0 θ(tj1 t2n−j ua ) · j=0 θ(t1 t2 va ))
1
× Q2 Q2 Qn−3 n−1−j j+2 Qn−2 Q2 n−1−j j+1 .
a,b=1 θ(u a v b ) a=1 ( j=0 θ(t 1 t2 ua v 1 v2 ) · j=0 a=1 θ(t 1 t 2 u 1 u 2 va ))
(3.52)

In order to prove this relation we deconfine the two antisymmetric as in Figure

8, obtaining two USp(2n) gauge nodes, with two bifundamentals P1,2 and two funda-
mentals R1,2 . The original antisymmetric chirals A1,2 correspond to the combinations
2
P1,2 while the original fundamentals Q1,2 correspond to the combinations P1,2 R1,2 . The
J-terms at this stage are the ones flipping the operators M and B and they are

JΨMa = Q̃Pa Ra , JΨBa = Q̃2 Pa2 , (3.53)

for a = 1, 2. The SU(2n + 1) gauge theory has 4n fundamentals and two antifunda-
mentals. It can be dualized according to the rules explained in subsection 2.3, and it
gives origin to the third quiver in Figure 8. In this case the new J-term generated by

– 30 –
 A1 A2

P1 P2
2n+1 2n 2n+1 2n

Q Q̃ R1 Q̃ R2
ΨM ΨM1 ΨM2
2 2 1 2 1

Ã1 Ã2
P̃1 P̃2
2n 2n−1 2n

N1 ψ N2 2n−1
R1 R2
2
q̃1 p̃ q̃2
ΨM1 ΨM2
1 1 1 2 1

Figure 8: Deconfinement steps of SU(2n + 1) with two antisymmetrics, two fundamentals

and two antifundamentals.

the duality is Jψ = N2 P̃2 + N1 P̃1 . In addition the J-terms for ΨMa and ΨBa become
JΨMa = Ra Na and JΨBa = Na2 .
The last step consists of dualizing the two USp(2n) groups, each one with 2n + 2
chiral fundamentals, to an LG. The final quiver is the fourth one in Figure 8. In
(0)
this case there are two J-terms associated to two Fermi singlets Ψ1,2 . They read
JΨ(0) = Ãn−2 2
1,2 q̃1,2 p̃ .
1,2

Similarly to the cases discussed above we have, up to an overall charge conjugation,

a model already described in 3.2. We can the use the duality of such model with an LG
in order to prove the validity of (3.52) and to reconstruct (up to flippers) the expected
J-term (3.50). The details of the derivation are straightforward and we leave them to
the interested reader.

3.7 SU(2n) with one fundamentals and three antifundamentals

In this case the dual LG model has six chiral gauge invariant operators corresponding
to chiral fields interacting through a J-term with a Fermi multiplet. The field content
of the gauge theory and of the dual LG model is represented in the table below.

– 31 –
 SU(2n) SU(2) SU(3) U(1)A U(1)Q U(1)Q̃ U(1)R0
A □ □ · 1 0 0 0
□
Q □ · · 0 1 0 0
Q̃ □ · □ 0 0 1 0
M · · □ 0 1 1 0
B · □ □ 1 0 2 0 (3.54)
Tn · ⊗nsym □ · n 0 0 0
Pn · ⊗n−2
sym □ □ n 1 1 0
Pn+1 · ⊗n−3
sym □ □ n+1 0 2 0
Πn+1 · ⊗n−5
sym □ · n+1 1 3 0
2n−5
Ψ · ⊗sym □ · −2n − 1 −1 −3 2
The chirals Tj , Pj correspond to the following gauge invariant combinations of the
charged fields

Tn = An , Pn = An−1 (AQ̃)Q , Pn+1 = An−1 (AQ̃)2 ,

Πn+1 = An−2 (AQ̃)3 Q ,
(3.55)
2
In addition there are a meson M = Q̃Q and the baryons B1,2 = A1,2 Q̃ , while the
expected J-term compatible with the global symmetry is

JΨ = BTn Pn + Pn Pn+1 + Tn Πn+1 + M Tn Pn+1 . (3.56)

We checked the matching of the ’t Hooft anomalies in the two theories. Explicitly we
have
κAA = −κAR0 = 2n(2n − 1) ,
κSU(2)2 = n(2n−1)
2
,
κQQ = −κQR0 = 2n ,
κSU(3) = n , (3.57)
κQ̃Q̃ = −κQ̃R0 = 6n ,
κR0 R0 = 6n + 1 ,
κAQ = κAQ̃ = 0 .
In the following we show the matching of the eblliptic genera along the same lines of
the discussion performed in the other examples. In this case we will flip the meson M
and the baryons B1,2 and PfA1,2 in the dual LG and we will show that the expected
J-term in (3.56) can be reconstructed by our procedure using the J-term expected from
the duality of subsection 3.2. At the level of the elliptic genus the identity associated
to this duality is
Q2n−5
(1,3,·,2,·) θ(q/(tj+3
j=0 1 t2
2n−j−1
uv1 v2 v3 ))
ISU(2n) (u; ⃗v ; ·; ⃗t; ·) =Q Q Qn−3 n−1−j j+2
a θ(uva ) a<b (θ(t1,2 va vb )) j=0 θ(t1 t2 va vb )

– 32 –
 A1 A2
2n−2 2n−2
P1 P2
2n ΨB1 ΨB2 2n ΨB1 ΨB2

Q Q̃ Q Q̃
ΨM ΨM
1 3 1 3

2n−2 2n−2
P̃1 P̃2
Ã1 Ã2
2n−3
ΨB1 ΨB2

q̃ ψ N2 2n−3

N1 1 3
q̃ p̃

1 3

Figure 9: Deconfinement steps of SU(2n) with two antisymmetrics, one fundamental and
three antifundamentals.

1
× Qn−5 . (3.58)
t2 uv1 v2 v3 ) nj=0
n−j−2 j+3 Qn−2 Q
θ(t1n−j tj2 ) j=0 n−1−j j+1
Q
j=0 θ(t1 a θ(t1 t2 uva )

In order to prove this relation we deconfine the two antisymmetric as in Figure 9,

obtaining two USp(2n − 2) gauge nodes, with two bifundamentals P1,2 . The original
2
antisymmetric chirals A1,2 correspond to the combinations P1,2 . The J-term at this
stage are the ones flipping the operators M and B and they read

JΨM = Q̃Q , JΨB1,2 = Q2 P1,2

2
. (3.59)

The SU(2n) gauge theory has 4n − 1 fundamentals and one antifundamental. It can be
dualized according to the rules explained in subsection 2.3, and it gives origin to the
third quiver in Figure 9.
In this case the new J-term generated by the duality is Jψ = N2 P̃2 + N1 P̃1 . In
addition the J-terms for ΨB1 and ΨB2 become JΨB1 = N12 and JΨB2 = N22 respectively.
The last step consists of dualizing the two USp(2n − 2) groups, each one with 2n
chiral fundamentals, to an LG. The final quiver is the fourth one in Figure 9. In this case
(0) n−3 3
there are two J-terms associated to two Fermi singlets Ψ1,2 . They read JΨ(0) = Ã1,2 p̃ .
1,2

– 33 –
 Again we have, up to an overall charge conjugation, a model already described in
3.2. We can the use the duality of such model with an LG in order to prove the validity
of (3.58) and to reconstruct (up to flippers) the expected J-term (3.56). The details of
the derivation are straightforward and we leave them to the interested reader.

3.8 SU(2n + 1) with one fundamentals and three antifundamentals

In this case the dual LG model has six chiral gauge invariant operators corresponding
to chiral fields interacting through a J-term with a Fermi multiplet. The field content
of the gauge theory and of the dual LG model is represented in the table below.

SU(2n + 1) SU(2) SU(3) U(1)A U(1)Q U(1)Q̃ U(1)R0

A □ □ · 1 0 0 0
□
Q □ · · 0 1 0 0
Q̃ □ · □ 0 0 1 0
M · · □ 0 1 1 0
B · □ □ 1 0 2 0 (3.60)
Tn · ⊗nsym □ · n 1 0 0
Pn+1 · ⊗n−1
sym □ □ n+1 0 1 0
Pn+2 · ⊗n−4
sym □ · n+2 0 3 0
R · ⊗n−3
sym □ □ n+1 1 2 0
2n−4
Ψ · ⊗sym □ · −2n − 2 −1 −3 2

The chirals Tj , Pj and R correspond to the following gauge invariant combinations

of the charged fields

Tn = An Q , Pn+1 = An (AQ̃) , Pn+2 = An−1 (AQ̃)3 , R = An−1 (AQ̃)2 Q , (3.61)

In additions there are other singlets identified with the meson M = QQ̃ and the baryons
B1,2 = A1,2 Q̃2 . The J-term compatible with the global symmetry is
2
JΨ = M Pn+1 + BPn+1 Tn + Tn Pn+2 + Pn+1 R . (3.62)

We checked the matching of the ’t Hooft anomalies in the two theories. Explicitly we
have
κAA = −κAR0 = 2n(2n + 1) ,
κSU(2)2 = n(2n+1)
2
,
κQQ = −κQR0 = 2n + 1 ,
κSU(3)2 = 2n+1
2
(3.63)
κQ̃Q̃ = −κQ̃R0 = 6n + 3 ,
κR0 R0 = 6n + 4 ,
κAQ = κAQ̃ = 0 .

– 34 –
 1 R2
A1 A2 R1 ψ

P1 P2
ΨM 2n 2n+1 2n
2n+1
ΨB1 ΨB2
ΨB1 ΨB2
Q Q̃ Q̃
ΨM
1 3 3

1
q

P̃1 P̃2 Ã1 Ã2

ΨM 2n 2n−1 2n

N1 λ
2n−1
3 N2
q q̃
ΨM
ΨB1 ΨB2 1 3

Figure 10: Deconfinement steps of SU(2n + 1) with two antisymmetrics, one fundamental
and three antifundamentals.

In the following we show the matching of the elliptic genera along the same lines of
the discussion performed in the other examples. In this case we found more convenient11
to flip the meson M and the baryons B1,2 , An1,2 Q and An1,2 A2,1 Q̃2,1 in the dual LG and
we will show that the expected J-term in (3.62) can be reconstructed by our procedure
using the J-term expected from the duality of subsection 3.2.
At the level of the elliptic genus the identity associated to this duality is
Q2n−4 j+3 2n−j−1
(1,3,·,2,·) j=0 θ(q/(t1 t2 uv1 v2 v3 ))
⃗
ISU(2n+1) (u; ⃗v ; ·; t; ·) = Q Q Qn−3 n−1−j j+2
a θ(uva ) a<b (θ(t1,2 va vb )) j=0 θ(t1 t2 uva vb )
1
× Qn−4 n−j−1 j+3 Qn n−j j Qn−1 Q n−j j+1 . (3.64)
j=0 θ(t 1 t 2 v1 v2 v 3 ) j=0 θ(t1 t2 u) j=0 a θ(t 1 t 2 v a )
In order to prove this relation we deconfine the two antisymmetric as in Figure 10,
obtaining two USp(2n) gauge nodes, with two bifundamentals P1,2 .
11
There are other possible patterns and flippers that can be used to prove the duality through tensor
deconfinement, but we found that the choice made here allows for a simpler derivation.

– 35 –
 Actually the procedure adopted here is slighlty different from the cases studied
above and it deserves some further comment. In the various cases studied so far the
two symplectic gauge nodes could be reconfined simultaneously to give the original
model. Here such reconfinement must be done in two steps. Indeed there is a field,
corresponding to the Fermi ψ in the second quiver of Figure 10 that gives a non-
vanishing J-term and which requires, after reconfining one of the two USp(2n) gauge
nodes, to integrate out some massive combinations of chirals and Fermi.
The J-term for the fields ψ is

Jψ = P1 R1 + P2 R2 . (3.65)

In addition in this phase there are other J-terms, originating from the flippers for M
and B1,2 discussed above

2
JΨM = Q̃(P1 R1 − P2 R2 ), JΨB1,2 = P1,2 Q̃2 . (3.66)

The SU(2n + 1) gauge node has now 4n fundamentals, three antifundamentals and
one Fermi. It can be dualized according to the rules explained in subsection 2.3, but
in this case with y = 1. The dual phase corresponds to the third quiver in Figure 10.
In this case the new J-term generated by the duality is Jλ = N2 P̃2 + N1 P̃1 . In addition
the other J-terms in this phase are

JΨB1 = N12 , JΨB2 = N22 , JΨM = q(N1 P̃1 − N2 P̃2 ) (3.67)

where the last term is obtained after integrating out the massive fields.
The last step consists of dualizing the two USp(2n) groups, each one with 2n + 2
chiral fundamentals, to an LG. The final quiver is the fourth one in Figure 10. In
(0)
this case there are two J-terms associated to two Fermi singlets Ψ1,2 . They read
JΨ(0) = Ãn−2 3
1,2 q̃ . There is a further J term for the field ΨM that reads JΨM = q q̃.
1,2
Again we have, up to an overall charge conjugation, a model already described in
3.2. We can the use the duality of such model with an LG in order to prove the validity
of (3.64) and to reconstruct (up to flippers) the expected J-term (3.62). The details of
the derivation are straightforward and we leave them to the interested reader.

3.9 SU(2n) with four antifundamentals

In this case the dual LG model has four chiral gauge invariant operators corresponding
to chiral fields interacting through a J-term with a Fermi multiplet. The field content

– 36 –
of the gauge theory and of the dual LG model is represented in the table below.

SU(2n) SU(2) SU(4) U(1)A U(1)Q̃ U(1)R0

A □ □ · 1 0 0
□
Q̃ □ · □ 0 1 0
B · □ □ 1 2 0
□ (3.68)
Tn · ⊗nsym □ · n 0 0
Tn−1 · ⊗n−3 □ n+1 2 0
sym □ □
Tn−2 · ⊗n−6
sym □ · n+2 4 0
2n−6
Ψ · ⊗sym □ · −2n − 2 −4 2

The chirals Tj and B correspond to the following gauge invariant combinations of

the charged fields

Tn = An , Tn−1 = An−1 (AQ̃)2 , Tn−2 = An−2 (AQ̃)4 , B = AQ̃2 , (3.69)

while the J-term compatible with the global symmetry is

2
JΨ = Tn Tn−2 + Tn−1 + B 2 Tn2 + BTn Tn−1 . (3.70)

We computed and matched the ’t Hooft anomalies in the two dual theories. Explicitly
we have
n(2n−1)
κSU(2)2 = 2
, κAA = −κAR0 = 2n(2n − 1) ,
κSU(4)2 = n , κQ̃Q̃ = −κQ̃R0 = 8n , (3.71)
κR0 R0 = 6n + 1 , κAQ̃ = 0 .

In the following we show the matching of the elliptic genera along the same lines of
the discussion performed in the other examples. In this case we will flip the baryons B
and PfA1,2 in the dual LG and we show that the expected J-term in (3.70) can be re-
constructed by our procedure using the J-term expected from the duality of subsection
3.2.
At the level of the elliptic genus the identity associated to this duality is
Q2n−6 j+4 2n−j−2 Q4
(·,4,·,2,·) j=0 θ(q/(t1 t2 a=1 va ))
ISU(2n) (·; ⃗v ; ·; ⃗t; ·) = Q Qn−3 n−1−j j+2
a<b (θ(t1,2 va vb ) j=0 θ(t1 t2 va vb ))
1
× Qn−6 n−j−2 j+4 Q4 Qn n−j j . (3.72)
j=0 θ(t1 t2 a=1 va ) j=0 θ(t1 t2 )

– 37 –
 A1 A2
2n−2 2n−2
P1 P2
2n ΨB1 ΨB2
2n ΨB1 ΨB2

Q̃

4 Q̃ 4

Ã1 Ã2
2n−2 2n−2
P̃1 P̃2
2n−4
ΨB1 ΨB2 2n−1
N1 N2
λ
q̃
4 4

Figure 11: Deconfinement steps of SU(2n) with two antisymmetrics and four antifundamen-
tals.

In order to prove this relation we deconfine the two antisymmetric as in Figure 11,
obtaining two USp(2n − 2) gauge nodes, with two bifundamentals P1,2 . The original
2
antisymmetric chirals A1,2 correspond to the combinations P1,2 . The J-term at this
stage is the one flipping the operator B, corresponding to
JΨB1,2 = Q̃2 P1,2
2
. (3.73)
The SU(2n) gauge theory has 4n − 4 fundamentals and four antifundamentals. It can
be dualized according to the rules explained in subsection 2.3, and it gives origin to the
third quiver in Figure 11.
In this case the new J-term generated by the duality is Jλ = N2 P̃2 + N1 P̃1 . In
addition the J-terms for ΨB1 and ΨB2 become JΨB1 = N12 and JΨB2 = N22 (where the
flavor indices are left implicit in these relations).
The last step consists of dualizing the two USp(2n − 2) groups, each one with 2n
chiral fundamentals, to an LG. The final quiver is the fourth one in Figure 9. In this case
(0) n−4 4
there are two J-terms associated to two Fermi singlets Ψ1,2 . They read JΨ(0) = Ã1,2 q̃ .
1,2
Again we have, up to an overall charge conjugation, a model already described in
3.2. We can the use the duality of such model with an LG in order to prove the validity
of (3.72) and to reconstruct (up to flippers) the expected J-term (3.70). The details of
the derivation are straightforward and we leave them to the interested reader.

– 38 –
3.10 SU(2n + 1) with four antifundamentals
Here we conclude this section with the last case of our classification of 2d N = (0, 2)
gauge/LG dualities for SU(N ) gauge theories with two antisymmetric chirals. In this
case the dual LG model has three chiral gauge invariant operators corresponding to
chiral fields interacting through a J-term with a Fermi multiplet. The field content of
the gauge theory and of the dual LG model is represented in the table below.
SU(2n + 1) SU(2) SU(4) U(1)A U(1)Q U(1)R0
A □ □ · 1 0 0
□
Q̃ □ · □ 0 1 0
B · □ □ 1 2 0 (3.74)
□
n−1
Pn+1 · ⊗sym □ □ n+1 1 0
n−4
Pn+2 · ⊗sym □ □ n+2 3 0
2n−5
Ψ · ⊗sym □ · −2n − 3 −4 2
The chirals Pj and B correspond to the following gauge invariant combinations of
the charged fields

Pn+1 = An (AQ̃) , Pn+2 = An−1 (AQ̃)3 , B = ÃQ2 , (3.75)

while the most general J-term compatible with the global symmetry is
2
JΨ = Pn+1 Pn+2 + BPn+1 . (3.76)

We have computed the ’t Hooft anomalies in the two phases and they match. Explicitly
we have
n(2n+1)
κSU(2)2 = 2
, κAA = −κAR0 = 2n(2n + 1) ,
κSU(4)2 = n , κQ̃Q̃ = −κQ̃R0 = 8n , (3.77)
κR0 R0 = 6n + 4 , κAQ̃ = 0 .

In the following we show the matching of the elliptic genera along the same lines
of the discussion performed in the other examples.
Actually in this case we need to adopt the same strategy used in the cases analyzed
in subsections 3.1 and 3.2 in order to trade the two antisymmetric with two symplectic
gauge groups. Concretely, we flip them by using the confining duality reviewed in
subsection 2.1.
In this case we also flip the baryon B in the dual LG and we show that the expected
J-term in (3.76) can be reconstructed by our procedure using the J-term expected from
the duality of subsection 3.1.

– 39 –
 A1 A2 P1 P2
2n−2 2n+1 2n−2

2n+1 R̃1 R̃2

ΨB1 ΨB2 ψ1 ψ2
4 Q̃
Q̃
4 1 ΨB1 ΨB2 1

1 1

L1 η2 η1 L2 Ã1 Ã2

2n−2 2n−5 2n−2

P̃1 P̃2 2n−1
λ
q̃
N1 4 N2
4

ΨB1 ΨB2

Figure 12: Deconfinement steps of SU(2n + 1) with two antisymmetrics and four antifunda-
mentals.

At the level of the elliptic genus the identity associated to this duality is
Q2n−5 j+4 2n−j−1 Q4
(·,4,·,2,·) j=0 θ(q/(t1 t2 a=1 va ))
ISU(2n+1) (·; ⃗v ; ·; ⃗t; ·) = Q
a<b θ(t1,2 va vb )
1
×Q Qn−4 n−j−1 j+3 Q Qn−1 n−j j+1 . (3.78)
a<b<c j=0 θ(t1 t2 va vb vc ) a j=0 θ(t1 t2 va )

In order to prove this relation we deconfine the two antisymmetrics as in Figure

12, obtaining two USp(2n − 2) gauge nodes, with two bifundamentals P1,2 and two
fundamental Fermi ψ1,2 . The original antisymmetric chirals A1,2 correspond to the
2
combinations P1,2 . There are also two J-terms Jψ1,2 = P1,2 R̃1,2 and in addition there
are J-terms flipping the operators B1,2 , corresponding to

JΨB1,2 = Q̃2 P1,2

2
. (3.79)

As discussed in subsection 3.1 in this phase we must impose a further constraint on the
charges in order to prevent the generation of an axial symmetry for each USp(2n − 2)

– 40 –
gauge symmetry. We refer the reader to the discussion there for further comments on
this problem.
The SU(2n + 1) gauge theory has 4n − 4 fundamentals and four antifundamentals.
It can be dualized according to the rules explained in subsection 2.3, and it gives origin
to the third quiver in Figure 12.
In this case the new J-terms generated by the duality are Jλ = N2 P̃2 + N1 P̃1 and
Jη1,2 = L1,2 P̃1,2 . In addition the J-terms for ΨB1 and ΨB2 become JΨB1 = N12 and
JΨB2 = N22 respectively (again omitting the antisymmetric flavor indices).
The last step consists of dualizing the two USp(2n − 2) groups, each one with
2n chiral fundamentals, to an LG. The final quiver is the fourth one in Figure 10.
(0)
In this case there are two J-terms associated to two Fermi singlets Ψ1,2 . They read
JΨ(0) = C1,2 Ãn−4 3
1,2 q̃ .
1,2
Again we have, up to an overall charge conjugation, a model already described in
3.2. We can the use the duality of such model with an LG in order to prove the validity
of (3.78) and to reconstruct (up to flippers) the expected J-term (3.76). The details of
the derivation are straightforward and we leave them to the interested reader.

4 A 4d parent for SU(2n) with (A, Ã) and 4 □

In this section we discuss a 4d duality that, once topologically twisted on a S 2 using the
prescription of [1], gives origin to a 2d gauge/LG duality. The 2d model corresponds
to a 2d duality studied in [9] using tensor deconfinement. The claim made in [9] was
that the model did not have a 4d origin in terms of any known 4d s-confining duality
in the classification of [26]. Such a classification is complete for models with vanishing
superpotential.
Despite this fact, here we obtain a parent 4d duality between two gauge theories
in which the dual phase reduces in 2d to the expected LG model.
Here, in order to prove the 4d duality we will use tensor deconfinement. Further-
more, the presence of the superpotential will trigger non-trivial RG flows in the dual
phases. The 4d original model corresponds to SU(2n) SQCD with four fundamental
flavors and one antisymmetric flavor. We turn on the following superpotential
n−1
X
n−1
W =A Q3 Q4 + βi Tr(AÃ)i + sA PfA + sà Pf à + Ma Qa Q̃ + γ Ãn−2 Q̃4 . (4.1)
i=1

with a = 1, 2. We then deconfine the antisymmetric A obtaining a new USp(2n − 2)

gauge group with an SU/USp bifundamental P . In addition we consider two USp(2n −
2) fundamentals R3,4 as in Figure 13. The original fields A and Q3,4 correspond in

– 41 –
 Ã

Q Q̃
2 2n 4 Σ2,A
A, Ã 4
Σ3,F
P
R3,4
Q, Q̃ 2n−2 2 2 2n−2 2
2n 4 Σ2,F R3,4

Figure 13: In this Figure we provide a quiver description of the 4d confining gauge theory
studied in this section. The original model corresponds to SU(2n) with four fundamental
flavors and one antisymmetric flavor. The theory has a non-trivial superpotential given in
formula (4.1). The second quiver corresponds to the model where we have deconfined the
SU(2n) antisymmetric tensor. The third model is obtained after using the s-confining duality
of [26] on the SU(2n) gauge node.

this phase to the combinations P 2 and P R3,4 respectively. The superpotential for this
phase is
n−1
X
W = βi Tr(P 2 Ã)i + sà Pf à + Ma Qa Q̃ + γ Ãn−2 Q̃4 . (4.2)
i=1

To proceed we notice that the SU(2n) gauge node has one conjugate antisymmetric,
2n fundamentals and 4 antifundamentals. The model is then s-confining in terms of
the following SU(2n) gauge invariant combinations

Σ1 = Ãn−1 Q̃2 , Σ2,A = ÃP 2 , Σ2,F = ÃP Q, Σ2,s = ÃQ2

(4.3)
Σ3,F = Q̃P, Σ3,s = Q̃Q, Σ4 = Q2 P 2n−2 , Σ5 = Pf Ã, Σ6 = Ãn−2 Q̃4

where Σ2,A is in the (tracless) antisymmetric representation of the leftover USp(2n − 2)

gauge group, while Σ3,F is in the fundamental of such USp(2n − 2). The fields Σ5,6
and Σ3,s are set to zero in the chiral ring by the superpotential (4.2) and we can ignore
them in the following. The other Σ-fields are USp(2n − 2) singlets. The superpotential
for this phase, after integrating out the massive fields, is
n−1
X
W = Σ1 Σ23,F (Σ2,s Σn−2
2,A + n−3 2
Σ2,A Σ2,F ) + βi Tr(Σ2,A )i + Σ4 Σ21 . (4.4)
i=2

In this way we have obtained a duality between SU(2n) with an antisymmetric flavor,
four fundamental flavors and superpotential (4.1), and an USp(2n − 2) gauge theory
with eight fundamental, one antisymmetric and superpotential (4.4). This last theory

– 42 –
is not confining but it has 72 many (self 12 –) dual phases (with a different structure of
flippers. Indeed the global symmetry is known to enhance to E7 × U(1) [33]).
This duality gives origin to the 2d gauge/LG duality discussed in [9] once we
compactify the models using the prescription of [1] by fixing the R charges of the fields
RQ̃,Ã,A = 0 and RQ = 1. The former become chirals in 2d N = (0, 2) while the latter do
not give rise to any 2d field. Then, the singlets βi , sA , s̃Ã and γ have R charge R = 2
and become Fermi fields in 2d, that we denote as Ψβi , ΨsA ,Ψsà and Ψγ respectively.
The singlets Ma , on the other hand, have R charge 1 and they do not give rise to any
2d field.
The 4d superpotential (4.1) of the electric theory gives origin to the 2d J-terms

Jψβi = Tr(AÃ)i , JψsA = PfA, Jψs = Pf Ã, Jψγ = Ãn−2 Q̃4 , (4.5)
Ã

with i = 2, . . . , n − 1, where we kept the same names for the 2d N = (0, 2) chirals of
the corresponding 4d N = 1 chiral multiplets.
By following the duality map, in the dual USp(2n − 2) gauge theory the antisym-
metric Σ2,A , the fundamentals Σ3,F and the singlet Σ1 have R charge R = 0 and they
survive as 2d N = (0, 2) chirals while the fields Σ2,F and R3,4 have R charge R = 1
and do not survive in 2d.
On the other hand, the singlets Σ2,s and Σ4 have R charge R = 2 and they give
origin to the Fermi fields ψΣ2,s and ψΣ4 respectively. The 4d superpotential (4.4) of the
dual theory gives origin to the 2d J-terms

Jψβi = Tr (Σ2,A )i , JψΣ2,s = Σ1 Σ23,F Σn−2

2,A , JψΣ4 = Σ21 , (4.6)

with i = 2, . . . , n − 2, where again we kept the same names for the 2d N = (0, 2) chirals
of the corresponding 4d N = 1 chiral multiplets.
The J-terms in (4.5) and (4.6) are equivalent to the 2d superpotentials discussed
in formulas (4.72) and (4.76) of [9], where the dual phase was further dualized to an
LG.
Summarizing, in this section we have obtained a parent 4d duality between two
gauge theories that, once compactified to 2d, gives origin to the 2d gauge/LG duality
studied in [9]. The 4d duality in this case does not correspond to an s-confining theory,
even if it is still possible that there exist other 4d s-confining models that can be reduced
to the same 2d dualities. We leave such possibility to future investigations.
12
In the sense that the dual theories have an USp(2n − 2) gauge group with an antisymmetric and
eight fundamentals but a different structure of singlets and superpotential.

– 43 –
5 Dualities involving antisymmetric Fermi fields
In this section we focus on SU(n) and USp(2n) gauge theories with (anti)-fundamental
chirals and with a Fermi field in the antisymmetric representation. The situation is
different from the cases treated above and in [7–9], because in these cases the technique
of tensor deconfinement does not apply. This is due to the fact that, to the best of
our knowledge, the fundamental gauge/LG dualities proposed in the literature never
include antisymmetric Fermi fields charged under a non-abelian flavor symmetry. This
constitutes an obstruction to proving dualities using the deconfining technique. Despite
this fact, here we observe that some dualities with charged antisymmetric Fermi fields
can be obtained by using the prescription of [1] from 4d parent dualities proposed in the
literature starting from the matching of the superconformal index. We further check
the 2d dualities obtained by matching the ’t Hooft anomalies.

5.1 A gauge/LG duality with an antisymmetric Fermi

The first 4d duality under investigation has been proposed in [30, 34] and further
investigated in [35], where it was referred to as SWV duality. A proof of the dualities
in terms of other fundamental dualities was then provided in [20].
The electric theory in this duality is an SU(n + 1) gauge theory with an antisym-
metric A, 2n antifundamentals Q̃ and n + 3 fundamentals Q with 4d superpotential
ele
WSWV = AQ̃2 . (5.1)

The dual theory is a WZ model, described by the electric baryon B = Qn+1 and
meson M = QQ̃ with superpotential
mag
WSWV = BM 2 . (5.2)

We are interested in a flipped version of this duality, where we break the global
SU(n + 3) symmetry by partially flipping the baryonic operator B. On the electric side,
we consider the split Q → (Q1 , Q2 ), corresponding to the decomposition of SU(n +
3) → SU(n + 1) × SU(2). Such symmetry breaking pattern is enforced in the electric
superpotential by adding a singlet b and considering the flipped superpotential
ele
WSWV = AQ̃2 + bQn+1
1 . (5.3)

On the magnetic side, under the symmetry breaking pattern above, the baryon B
is decomposed into three fields: an SU(n + 1) antisymmetric BA , an SU(n + 1) × SU(2)
bifundamental BV and a singlet BS . Similarly, the meson M splits into M1 and M2

– 44 –
respectively in the fundamental representation of SU(n + 1) and SU(2). The dual
superpotential becomes

mag
WSWV = BA M12 + BV M1 M2 + BS M22 + bBS , (5.4)

and after integrating out the massive fields it reduces to

mag
WSWV = BA M12 + BV M1 M2 , (5.5)

as expected from the addition of the baryonic flipper in (5.3).

The superconformal index still matches across the dual phases, as the addition of
the flipper just moves the relative elliptic gamma function from one side of the duality to
the other without modifying the balancing condition, or equivalently, the requirements
from the anomaly freedom.
In the next step we fix the non-negative integer R charges of the flipped version
of the duality. We fix the R charge of the antisymmetric to RA = 2 and, consistently
with the anomaly freedom condition, we fix the R charges of the fields Q2 to RQ2 = 1.
The charges of Q1 and Q̃ are then vanishing, consistently with the global symmetries.
The singlet b in (5.3) on the other hand has Rb = 2.
The 2d theory obtained using such assignation of charges is an SU(n + 1) electric
gauge theory with an antisymmetric Fermi ΨA , 2n antifundamental chirals Q̃ and n + 1
fundamental chirals Q with 2d J-terms

JΨA = Q̃2 , JΨB = Qn+1 , (5.6)

where the Fermi ψb descend from the 4d singlet b.

In the 4d dual theory the R-charges that can be read from the duality map are
RBA = 2, RM1 = 0 and RM2 = RBV = 1. The field BA becomes a Fermi field, denoted
as ψ, in the antisymmetric representation of the SU(n + 1) flavor symmetry. The
meson M1 becomes a chiral, denoted by ϕM , in the antifundamental representation of
the SU(n + 1) flavor symmetry (and in the fundamental of the other USp(2n) flavor
symmetry). On the other hand, the fields BS and M2 do not give rise to any 2d field.
The magnetic description is an LG model with J-term JΨ = Φ2M where we can
further associate ΦM to the combination QQ̃. This relation and the structure of the
J-term fix the charges of the fields in the dual picture. This results into the table of
global charges below.

– 45 –
 USp(2n) SU(n + 1) U(1)A U(1)Q U(1)R
ΨA · · 1 0 1
1
Q̃ □ · −2 0 0
Q · □ 0 1 0 (5.7)
ΨB · · 0 −n − 1 1
ΦM □ □ 1 − 12 0
Ψ · □ −2 1 1
□

ΨB ΨA Ψ

Q Q̃ ϕM
n+1 n+1 2n n+1 2n

Figure 14: In this figure we represent the quivers associated to the duality between SU(n+1)
with an antisymmetric Fermi and the LG model. Dashed lines are for Fermi fields while
continuous lines are for chiral multiplets. The dots refer to the singlets.

We now provide some checks of the duality proposed here obtained from the re-
duction of the parent 4d SWV duality. First of all, we have computed the ’t Hooft
anomalies and checked their matching. Explicitly, we have

n+1
κSU(n+1)2 = , κUSp(2n)2 = n + 1 , κQQ = κAA = κQA = 0 ,
2
3n(n + 1) n(n + 1)
κRR = , κAR = , κRQ = −n(n + 1) . (5.8)
2 2
The validity of the duality also implies the relation between the ellitpic genera
n+1 Q ±1
(q; q)2n+2 1≤i<j≤n+1 θ(qT zi zj )θ((zi /zj ) )
I
∞ dzi
Y
Iele = Qn+1 Qn+1 Qn ±1 −1
√
(n + 1)! JK i=1 2πizi i=1 ( m=1 θ(zi s m ) k=1 θ(tk zi / T ))
n+1
!
−1 −1
Q
1≤m<ℓ≤n+1 θ(qT sm sℓ )
Y
×θ q s−1
ℓ = Q n+1 Q n ±1
√ = Imag , (5.9)
ℓ=1 m=1 k=1 θ(s m tk / T)

with n+1
Q
i=1 zi = 1. We explitly checked the validity of this identity for the cases n = 1, 2,
where the duality reduces to a flipped version of the ones studied in subsection 2.3.
Indeed, in the first case the antisymmetric Fermi disappears, while in the second case
it is equivalent to an antifundamental Fermi field.

– 46 –
 A further check can be obtained through a non-anomalous gauging of a subgroup
of the USp(2n) flavor symmetry group. For example, let us start by considering the
case of n = 2m − 1 such that the flavor symmetry is USp(4m − 2). Then, we gauge an
USp(2m − 2) subgroup of this symmetry. The final USp(2m) × USp(2m − 2) is then
non-anomalous and the superpotential for this model becomes

W = ΨA Q̃21 + ΨA Q̃22 + ΨB Q2m , (5.10)

where Q1 is the SU(2m) × USp(2m − 2) bifundamentals while Q2 represents the other

2m SU(2m) fundamentals whose flavor symmetry is not gauged.
The USp(2m − 2) gauge theory has 2m fundamentals and it can be dualized into
an LG model. The superpotential after this duality becomes

W = ΨA A + ΨA Q̃22 + ψσ PfA + ΨB Q2m . (5.11)

The Fermi ΨA and the chiral antisymmetric A = Q21 are massive and can be integrated
out, leaving us with

W = ΨA A + ΨA Q̃22 + ψσ PfA + ΨB Q2m . (5.12)

By integrating out the massive fields the superpotential for the remaining SU(2m)
gauge theory becomes
W = ψσ Q̃2m2 + ΨB Q
2m
, (5.13)
that is dual to an LG with a meson ΦM = QQ̃2 with superpotential

W = Ψ̂ det ΦM . (5.14)

The same gauging can be performed in the dual phase in which the model becomes

W = Ψ(Φ2M1 + Φ2M2 ) , (5.15)

where ΦM1 is charged under the USp(2m − 2) gauge symmetry, while ΦM2 is charged
under the USp(2m) flavor symmetry. The USp(2m − 2) gauge group is dual to an LG
model and we get
W = Ψ(A + Φ2M2 ) + Ψ0 PfA , (5.16)
with A = Φ2M1 . By integrating out the massive fields, such superpotential becomes

W = Ψ0 Pf(ΦM2 )2 = Ψ0 det ΦM2 , (5.17)

which is equivalent to (5.14).

– 47 –
5.2 A gauge/gauge duality with an antisymmetric Fermi
The second duality in presence of charged antisymmetric Fermi fields that we consider
relates two symplectic gauge theories with USp(2n) and USp(2n − 2) gauge group
respectively. Again, the duality is obtained from a 4d parent, that was originally
proposed in [30] from the matching of the associated superconformal indices obtained
in [36]. The duality was then derived from a physical perspective in [21].
The 4d duality relates models characterized by symplectic gauge groups, an anti-
symmetric field together with fundamental flavors which interact with the antisymmet-
ric, breaking the flavor symmetry with various possible patterns. Here we focus on a
specific pattern, suitable to give origin to a 2d N = (0, 2) duality with a charged anti-
symmetric Fermi field. In this case the flavor symmetry is SU(4) × USp(4n − 2). There
are four fundamentals P and 4n − 2 fundamentals Q, which interact with the USp(2n)
antisymmetric A through a superpotential W = AQ2 . The dual picture corresponds to
an USp(2n − 2) gauge theory with four fundamentals p, 4n − 2 fundamentals q and an
antisymmetric, denoted as a. The superpotential of this dual phase is

W = N M 2 + M pq + aq 2 , (5.18)

where the singlets N and M correspond in the electric phase to the gauge invariant
combinations P 2 and P Q.
Again, we consider a flipped version of this duality by breaking the SU(4) flavor
symmetry to SU(2)1 ×SU(2)2 , denoting the two fundamentals as P1 and P2 and flipping
the combination P22 by adding a singlet B in the electric phase. The electric theory
has then superpotential
WU Sp(2n) = AQ2 + BP22 . (5.19)
On the magnetic side the superpotential becomes

WU Sp(2n−2) = M1 p1 q + M2 p2 q + aq 2 + M12 N1 + M1 M2 L, (5.20)

where M1 = QP1 , M2 = QP2 , L = P1 P2 and N1 = P12 .

We then compactify the theory on S 2 by fixing RP1 = RQ = 0, RA = RS = 2 and
RP2 = 1. In this way we obtain a 2d N = (0, 2) USp(2n) electric gauge theory with an
antisymmetric Fermi ΨA , 4n − 2 fundamental chirals Q and 2 fundamental chirals P1
with J-terms YΨA = Q2 and JΨB = P12 . In the dual theory the R charges, read from the
duality map, are Rq = RM1 = 0, Ra = RN1 = Rp1 = 2 and RM2 = RL = Rp2 = 1. The
magnetic description is an USp(2n − 2) gauge theory with an antisymmetric Fermi Ψa ,
4n − 2 fundamental chirals q and 4 fundamentals Fermi multiplets Ψp1 , with J-terms
JΨp1 = qΦM1 , JΨa = q 2 and JΨN1 = Φ2M1 .

– 48 –
 ΨN1 Ψa
ΨB ΨA
Ψp 1 q
2 2n−2 4n−2
P1 Q
2 2n 4n−2
ΦM1

Figure 15: In this figure we represent the quivers associated to the duality between USp(2n)
and USp(2n − 2) with an antisymmetric Fermi. Dashed lines are for Fermi fields while
continuous lines are for chiral multiplets. The dots refer to the singlets.

The duality map in 2d can be read once we associate the chiral ΦM1 to the electric
mesonic combination P1 Q. The global charges for this duality are then summarized in
the table below.
USp(2n) USp(4n − 2) SU(2) U(1)A U(1)P U(1)R
ΨA □ · · 1 0 1
□
Q □ □ · − 12 0 0
P1 □ · □ 0 1 0
ΨB · · · 0 −2 1
USp(2n − 2) USp(4n − 2) SU(2) U(1)A U(1)P U(1)R (5.21)
q □ □ · − 12 0 0
1
ΦM1 · □ □ −2 1 0
Ψp1 □ · □ 1 1 1
Ψa □ · · 1 0 1
□
ΨN1 · · · 1 −2 1

As a check we computed the ’t Hooft anomalies for this duality and we have showed
that they match. We have

κP P = κAA = κAP = 0 , κRP = 2(1 − 2n) , κRA = n(2n − 1) ,

κSU(2)2 = n , κUSp(4n−2)2 = 2n , κRR = 4n2 − 1 . (5.22)

For completeness we also provide the expression for the elliptic genus in the electric
and in the magnetic case. On the electric side we have
 
q
(q; q)2n n−1 Qn
θ(qT ) θ n Q ±1 ±1 ±1 ±1 ±2
i<j θ(zi zj )θ(qT zi zj ) i=1 θ(zi )
I Y
∞ t1 t2
Iele = Qn Q2 Q2n−1 ±1 − 21 ±1
(5.23)
n! 2n JK i=1 i=1
±1
a=1 θ(zi ta ) ℓ=1 θ(zi T sℓ )

– 49 –
while on the dual side we have
 
qT
(q; q)2n−2
∞ θ(qT )n−2
θ t1 t2
Imag = Q2n−1 − 12
(n − 1)! 2n−1 ℓ=1 θ(s±1ℓ t1,2 T )
  
Q ±1 ±1 ±1 ±1
Q n−1 ±2
Q2 qT ±1
I n−1
Y i<j θ(z i zj )θ(qT zi zj ) i=1 θ(z i ) a=1 θ z
ta i
Qn−1 Q2n−1 1
±1 − 2 ±1
. (5.24)
JK i=1 i=1 ℓ=1 θ(z i T s ℓ )

6 Conclusions
In this paper we have studied 2d N = (0, 2) dualities with antisymmetric matter,
extending the program started in [9]. The first class of dualities under investigation
correspond to SU(N ) gauge theories with 2 antisymmetric, nf fundamental and na
antifundamental chirals, with nf + na = 4. Our main claim is that such models have a
dual description in terms of LG models, with chiral and Fermi multiplets and non-trivial
J-terms. Such gauge/LG dualities cannot in general be predicted from the prescription
of [1] by a 4d/2d reduction of s-confining dualities and a check of their validity requires,
in principle, a pure 2d analysis. We have checked such claims by verifying the ’t
Hooft anomaly matching and by studying the elliptic genus. Our analysis has shown
that the matching of the elliptic genera follows from other basic identities that can
be obtained from the dimensional reduction of 4d s-confining dualities. By assuming
the matching of the 4d index on S 2 × T 2 defined in [2, 37, 38], the matching of the
elliptic genera for such basic dualities follows from the prescription of [1]. The basic
identity employed here have been reviewed in section 2. A crucial relation is the one
discussed in subsection 2.1. Despite the fact that the field content in this case admits an
additional axial symmetry, the identity 2.4 holds only if this extra symmetry is absent.
This suggests that, on the field theory side, some non-perturbative effect must be
invoked on the gauge side of the duality to lift such dangerous symmetry. As originally
discussed in [1, 7], this expectation is reminiscent of similar results obtained in the circle
reduction of 4d dualities, keeping the effects of the KK monopole when considering the
3d effective dualities at finite size of the circle [31]. Here we have used the basic identity
for this 2d duality under the assumption that such a non-perturbative effect lifts the
axial symmetry from the spectrum. As we commented in the introduction one can
think of such a 2d duality as arising from a 2d boundary one [29] of the 3d effective
theory on the circle. In such a case the matching of the half-index does not depend
on the axial symmetry, which is expected to be lifted by the presence of the monopole
superpotential in the bulk theory. We refer the reader to [39] for examples of this kind.
In this paper we have left to future investigation an appropriate discussion on such non-
perturbative effect, and looked at the consequences of the duality of subsection 2.1. We

– 50 –
have observed that, by deconfining the antisymmetric chirals using this duality and then
dualizing the original SU(N ) gauge node, there is no need to call for non-perturbative
effects anymore, and it is then possible to re–confine the antisymmetric chirals using
the ordinary gauge/LG for USp(2n) with 2n + 2 fundamental chirals. By iterating the
procedure, we have been able to find the expected identities for the proposed dualities
with nf = 4 and na = 0. Furthermore, we have shown that all the other dualities with
na > 0 arise as a consequence of the ones with na = 0.
In the second part of the paper we studied a 4d duality between a SU(2n) theory
with an antisymmetric and four fundamental flavors, and a USp(2n) one with an anti-
symmetric and eight fundamentals. In absence of a superpotential the former has a nine
dimensional Cartan for the non-anomalous non–R global symmetry group, compatible
with the enhancement to D6 × U(1)3 in presence of an opportune set of flippers [40].
Analogously, the latter is compatible with an E7 ×U(1) enhancement [33]. Here we have
broken the nine dimensional Cartan of the SU(2n) gauge theory by introducing a 4d
superpotential term between an antisymmetric and two fundamentals. Such a breaking
has allowed us to state a 4d duality using the tensor deconfinement technique. Despite
the fact that such a duality can be interesting per se, here we have studied its fate upon
compactification on S 2 with non-negative integer R-charges through the prescription
of [1]. We have found that SU(2n) with an antisymmetric flavor and four fundamental
chirals is dual symplectic gauge theory is related to the LG theory expected from the
analysis of [9]. In this sense the 4d duality obtained here is a parent of the 2d duality
proposed in [9]. In the last part of the paper we obtained 2d dualities in presence of
charged antisymmetric Fermi fields. Such dualities are especially interesting because
it is not possible to deduce them from a pure 2d perspective by using the deconfin-
ing technique, or in other words, because we are not aware of any candidate basic 2d
duality with Fermi fields in the antisymmetric representation of the flavor symmetry.
We conclude with some general comments on the IR dynamics. We expect that the
gauge/LG dualities discussed in the paper give rise to interacting CFT in the IR. The
reason behind such expectation is that in presence of non-compact target spaces the
global symmetries associated to such non-compact directions get removed from the c-
extremization procedure. In particular, the trial R-symmetry discussed in the various
example is also the exact one. The central charge is then positive and there is no
violation of the unitarity bounds. The presence of a non-compact target space implies
also that we cannot turn off the fugacities in the matching of the elliptic genera. This
signals that the dualities discussed here have to be interpreted as dualities in the mass
deformed case, where we are only in presence of isolated vacua. Similar observations
have been made in the literature for dualities with charged antisymmetric chiral fields
in [7–9].

– 51 –
Acknowledgments

The work of the authors has been supported in part by the Italian Ministero dell’Istruzione,
Università e Ricerca (MIUR), in part by Istituto Nazionale di Fisica Nucleare (INFN)
through the “Gauge Theories, Strings, Supergravity” (GSS) research project.

A Conventions on the elliptic genus

In this appendix we fix the conventions used in the paper for the elliptic genus. We
refer here to the elliptic genus in the NSNS sector, defined originally in [41, 42] (see
[43, 44] for the definition in the RR sector [41, 42]). The index is given by
Y
I(⃗u; q) ≡ I(⃗u) ≡ TrNSNS (−1)F q L0 ucaa , (A.1)
a

where q = e2πiτ and τ refers to the complex structure of T 2 . Formula (A.1) can be
interpreted as a Witten index refined by the flavor fugacities ua . Flat directions render
the index divergent in presence of non-compact target space, therefore, it is crucial to
keep the flavor fugacities turned on. This can be interpreted as a restriction of the
analysis to the massive theory.
Denoting the gauge group as G, the elliptic genus can be computed as an integral
over the Cartan torus of G, expressed in terms of the gauge fugacities z
I rk G
1 Y dzi
I(u) = IV (⃗z)Iχ (⃗z, ⃗u)Iψ (⃗z, ⃗u), (A.2)
|W | JK i=1 2πizi

where JK refers to the fact that the integral is computed by using the Jeffrey-Kirwan
prescription.
The vector, the chiral and the Fermi multiplets contribute respectively as
Y
IV (⃗z) = (q; q)2rk
∞
G
θ (z αG ) ,
αG
Y 1
Iχ (⃗z, ⃗u) =  Rχ
, (A.3)
θ q z ρG uρF
2
ρG ,ρF
Y  Rψ +1 
Iψ (⃗z, ⃗u) = θ q 2 z ρG uρF ,
ρG ,ρF

Q∞
with θ(x) = (x; q)∞ (qx−1 ; q)∞ and (x; q)∞ = j=0 (1 − xq j ).

– 52 –
 The index of an SU(N ) gauge theory with F fundamentals Q, F̃ antifundamentals
Q̃, H fundamental (or antifundamental) Fermi Λ, K antisymmetrics A and K̃ conjugate
antisymmetrics à can be expressed as
2(N −1)
(F,F̃ ;H;K;K̃) (q; q)∞
ISU(N ) ⃗ ⃗n; ⃗h; ⃗r, ⃗s) =
(m;
N!
QN QH RΛ +1
N Q ±1
i<j θ ((zi /zj ) ) a=1 θ(q zi ha )
I Y
dzi i=1
2
×  
JK i=1 2πizi
QN QF
θ (q RQ z m ) · F̃ θ(q RQ̃ z −1 n )
Q
i=1 a=1 i a a=1 i a

QN
δ(1 − i=1 zi )
×Q Q
 . (A.4)
K
· K̃ q RÃ sa zi−1 zj−1
Q
i<j a=1 θ(q RA ra zi zj ) a=1 θ

Throughout the paper, we have conventionally omitted the fugacities by leaving a · in

the argument, when the associated field is absent.
The index of an USp(2N ) gauge theory with F fundamentals Q, H fundamental
Fermi multiplets Λ and one traceless antisymmetric chiral A is denoted as

QN
N
θ zi±1 zj±1 ±2
Q
i=1 θ(zi )
2N I Y
(F ;H;1) (q; q) dzi i<j
⃗ ⃗h; r) =
IUSp(2N ) (m; ∞
rA ±1 ±1
2N N ! θ (q rA r)N −1 JK i=1 2πizi
Q

i<j θ q zi zj r
QH  R +1 
Λ ±1
Y a=1 θ q
N 2 zi ha
× QF . (A.5)
RQ z ±1 m )
i=1 a=1 θ(q i a

– 53 –
References
[1] A. Gadde, S. S. Razamat and B. Willett, On the reduction of 4d N = 1 theories on S2 ,
JHEP 11 (2015) 163 [1506.08795].
[2] F. Benini and A. Zaffaroni, A topologically twisted index for three-dimensional
supersymmetric theories, JHEP 07 (2015) 127 [1504.03698].
[3] F. Benini and N. Bobev, Exact two-dimensional superconformal R-symmetry and
c-extremization, Phys. Rev. Lett. 110 (2013) 061601 [1211.4030].
[4] F. Benini and N. Bobev, Two-dimensional SCFTs from wrapped branes and
c-extremization, JHEP 06 (2013) 005 [1302.4451].
[5] N. Seiberg, Electric - magnetic duality in supersymmetric nonAbelian gauge theories,
Nucl. Phys. B 435 (1995) 129 [hep-th/9411149].
[6] K. A. Intriligator and P. Pouliot, Exact superpotentials, quantum vacua and duality in
supersymmetric SP(N(c)) gauge theories, Phys. Lett. B 353 (1995) 471
[hep-th/9505006].
[7] M. Sacchi, New 2d N = (0, 2) dualities from four dimensions, JHEP 12 (2020) 009
[2004.13672].
[8] J. Jiang, S. Nawata and J. Zheng, 2d dualities from 4d, 2407.17350.
[9] A. Amariti, P. Glorioso, F. Mantegazza, D. Morgante and A. Zanetti, Dualities from
dualities in 2d N = (0, 2), 2410.12453.
[10] M. Berkooz, The Dual of supersymmetric SU(2k) with an antisymmetric tensor and
composite dualities, Nucl. Phys. B 452 (1995) 513 [hep-th/9505067].
[11] M. A. Luty, M. Schmaltz and J. Terning, A Sequence of duals for Sp(2N)
supersymmetric gauge theories with adjoint matter, Phys. Rev. D 54 (1996) 7815
[hep-th/9603034].
[12] S. Pasquetti and M. Sacchi, From 3d dualities to 2d free field correlators and back,
JHEP 11 (2019) 081 [1903.10817].
[13] S. Benvenuti, I. Garozzo and G. Lo Monaco, Monopoles and dualities in 3dN = 2
quivers, JHEP 10 (2021) 191 [2012.08556].
[14] I. G. Etxebarria, B. Heidenreich, M. Lotito and A. K. Sorout, Deconfining N = 2
SCFTs or the art of brane bending, JHEP 03 (2022) 140 [2111.08022].
[15] S. Benvenuti and G. Lo Monaco, A toolkit for ortho-symplectic dualities, JHEP 09
(2023) 002 [2112.12154].
[16] L. E. Bottini, C. Hwang, S. Pasquetti and M. Sacchi, Dualities from dualities: the
sequential deconfinement technique, JHEP 05 (2022) 069 [2201.11090].

– 54 –
[17] S. Bajeot and S. Benvenuti, Sequential deconfinement and self-dualities in 4dN = 1
gauge theories, JHEP 10 (2022) 007 [2206.11364].
[18] S. Bajeot and S. Benvenuti, 4dN = 1 dualities from 5d dualities, JHEP 08 (2024) 197
[2212.11217].
[19] A. Amariti and S. Rota, 3d N=2 SO/USp adjoint SQCD: s-confinement and exact
identities, Nucl. Phys. B 987 (2023) 116068 [2202.06885].
[20] A. Amariti, F. Mantegazza and D. Morgante, Sporadic dualities from tensor
deconfinement, JHEP 05 (2024) 188 [2307.14146].
[21] A. Amariti and F. Mantegazza, A new 4d N = 1 duality from the superconformal
index, JHEP 06 (2024) 206 [2402.00609].
[22] A. Amariti and F. Mantegazza, Confinement for 3d N = 2SU (N ) with a Symmetric
tensor, 2405.11972.
[23] S. Benvenuti, R. Comi, S. Pasquetti and M. Sacchi, Deconfinements,
Kutasov-Schwimmer dualities and Dp [SU (N )] theories, 2407.11134.
[24] C. Hwang and S. Kim, S-confinement of 3d Argyres-Douglas theories and the
Seiberg-like duality with an adjoint matter, 2407.11129.
[25] S. Bajeot and S. Benvenuti, S-confinements from deconfinements, JHEP 11 (2022) 071
[2201.11049].
[26] C. Csaki, M. Schmaltz and W. Skiba, Confinement in N=1 SUSY gauge theories and
model building tools, Phys. Rev. D 55 (1997) 7840 [hep-th/9612207].
[27] K. Nii, On s-confinement in 3d N = 2 gauge theories with anti-symmetric tensors,
1906.03908.
[28] A. Amariti, F. Mantegazza and S. Rota, Rank-two tensors and deconfinement in su(n)
gauge theories with 4 supercharges, To Appear (2025) .
[29] T. Dimofte, D. Gaiotto and N. M. Paquette, Dual boundary conditions in 3d SCFT’s,
JHEP 05 (2018) 060 [1712.07654].
[30] V. P. Spiridonov and G. S. Vartanov, Elliptic Hypergeometry of Supersymmetric
Dualities, Commun. Math. Phys. 304 (2011) 797 [0910.5944].
[31] O. Aharony, S. S. Razamat, N. Seiberg and B. Willett, 3d dualities from 4d dualities,
JHEP 07 (2013) 149 [1305.3924].
[32] M. Dedushenko and S. Gukov, IR duality in 2D N = (0, 2) gauge theory with
noncompact dynamics, Phys. Rev. D 99 (2019) 066005 [1712.07659].
[33] S. S. Razamat and G. Zafrir, E8 orbits of IR dualities, JHEP 11 (2017) 115
[1709.06106].

– 55 –
[34] V. P. Spiridonov and S. O. Warnaar, Inversions of integral operators and elliptic beta
integrals on root systems, Advances in Mathematics 207 (2006) 91.
[35] B. Nazzal, A. Nedelin and S. S. Razamat, Minimal (D, D) conformal matter and
generalizations of the van Diejen model, SciPost Phys. 12 (2022) 140 [2106.08335].
[36] F. J. van de Bult, An elliptic hypergeometric beta integral transformation, arXiv
preprint arXiv:0912.3812 (2009) .
[37] C. Closset and I. Shamir, The N = 1 Chiral Multiplet on T 2 × S 2 and Supersymmetric
Localization, JHEP 03 (2014) 040 [1311.2430].
[38] M. Honda and Y. Yoshida, Supersymmetric index on T 2 × S 2 and elliptic genus,
1504.04355.
[39] T. Okazaki and D. J. Smith, Boundary confining dualities and Askey-Wilson type
q-beta integrals, JHEP 08 (2023) 048 [2305.00247].
[40] S. S. Razamat, O. Sela and G. Zafrir, Curious patterns of IR symmetry enhancement,
JHEP 10 (2018) 163 [1809.00541].
[41] A. Gadde and S. Gukov, 2d Index and Surface operators, JHEP 03 (2014) 080
[1305.0266].
[42] A. Gadde, S. Gukov and P. Putrov, Walls, Lines, and Spectral Dualities in 3d Gauge
Theories, JHEP 05 (2014) 047 [1302.0015].
[43] F. Benini, R. Eager, K. Hori and Y. Tachikawa, Elliptic genera of two-dimensional
N=2 gauge theories with rank-one gauge groups, Lett. Math. Phys. 104 (2014) 465
[1305.0533].
[44] F. Benini, R. Eager, K. Hori and Y. Tachikawa, Elliptic Genera of 2d N = 2 Gauge
Theories, Commun. Math. Phys. 333 (2015) 1241 [1308.4896].

– 56 –