Multiple Phases and Meromorphic Deformations of Unitary Matrix Models

MULTIPLE PHASES AND MEROMORPHIC
DEFORMATIONS OF UNITARY MATRIX MODELS
LEONARDO SANTILLI∗ AND MIGUEL TIERZ†,‡
Abstract. We study a unitary matrix model with Gross–Witten–Wadia weight function and
determinant insertions. After some exact evaluations, we characterize the intricate phase diagram.
There are five possible phases: an ungapped phase, two different one-cut gapped phases and two
other two-cut gapped phases. The transition from the ungapped phase to any gapped phase
is third order, but the transition between any one-cut and any two-cut phase is second order.
The physics of tunneling from a metastable vacuum to a stable one and of different releases of
instantons is discussed. Wilson loops, β-functions and aspects of chiral symmetry breaking are
arXiv:2102.11305v1 [hep-th] 22 Feb 2021
investigated as well. Furthermore, we study in detail the meromorphic deformation of a general

class of unitary matrix models, in which the integration contour is not anchored to the unit circle.
The ensuing phase diagram is characterized by symplectic singularities and captured by a Hasse
diagram.
Contents
1. Introduction 2
2. The model 2
3. Phase structure 6
4. Wilson loops and instantons 14
5. Meromorphic deformation of unitary matrix models 20
6. Outlook 28
Appendix A. Technical details of the solution 28
Appendix B. Filling fraction fluctuations 31
References 32
1
2
1. Introduction
The study of the spectral and critical properties of models of random matrices has become a
widely popular and interdisciplinary subject in this century, in great part due to the vast scope
of fields where such models appear naturally and play a prominent role [1–3]. One among the
possible many ways to highlight this relevance and versatility of random matrix theory is to simply
point out the fact that the same model oftentimes appears in remarkably different contexts and,
in addition, in a significant manner.
A paradigmatic example of this phenomenon could very well be the so-called Gross–Witten–
Wadia (GWW) model [4–6]. Originally proposed in the study of gauge theory, it is ubiquitous
and pivotal in many other areas, such as combinatorics, representation theory and spectral theory
[1–3].
With this fact in mind, in this work we will study unitary matrix models, starting, precisely,
with the specific case of a generalized form of the Gross–Witten–Wadia model. A possible inter-
pretation of the model is as a one-plaquette model of two-dimensional lattice QCD with fermionic
or bosonic quarks, which equivalently corresponds to a massive deformation of a model introduced
by Minahan [7, 8].
We will show that, while quite simple, this model retains several features of a sensible quantum
field theory in the continuum. In turn, its simplicity allows us to exploit standard techniques
from random matrix theory to characterize the theory at large N and suggests more general and
deeper problems to consider. Some of them, we will already tackle here, by discussing at length
the case of meromorphic deformations of unitary matrix model, as we explain below.
Before that, a rich phase diagram will be obtained and analyzed in detail. Phase transitions
such as the ones we obtain in our analysis are relevant to the study of deconfinement transitions
in QCD models in four dimensions [9, 10] and in black holes physics [11–13]. More recently, this
type of phase transitions has been argued to describe the critical behaviour of models exhibiting
partial deconfinement [14–18].
We introduce the model in what follows, in Section 2, which includes a discussion on interpre-
tations of the model, notation, relationship with other systems and an introductory discussion of
its mathematical properties, including exact evaluations without scaling limits.
Then, the main results of the paper are presented and organized as follows: there are three
main contributions, as far as new results are concerned. In Section 3, we fully characterize the
rich phase structure of the unitary matrix model. In Section 4, we study Wilson loops in the
same setting of Section 3 and discuss at length the physical interpretation of the phase transitions,
including the role of instanton contributions.
Finally, in Section 5 we study, in a general framework and going beyond the specific model
studied in the previous sections, the case where the integration contour is deformed in C∗ away
from the unit circle. In spite of the vast body of results on random matrix ensembles, holomorphic
matrix models [19] are arguably understudied.
The aim of Section 5 is to adapt the results on holomorphic matrix models to unitary matrix
models. We do so for a very general set of unitary matrix models and, only as an illustrative
example, we discuss the particular case of the holomorphic GWW matrix model. Non-traditional
tools in this area, such as symplectic singularities and Hasse diagrams, are introduced here to
fully understand the meromorphic models.
We conclude with possible avenues for further research in Section 6.
2. The model
In this section we present the model and study some of its exact features at finite N .
2.1. The model and its interpretations. Consider a one-plaquette model of two-dimensional
lattice gauge theory [20, 21] with gauge group U (N ) and K pairs of real fields, that can be either
3
bosonic or fermionic. Each pair is called a flavour. We encode the choice of matter fields in the
binary variable (
+1 fermions,
=
−1 bosons.
We must impose (anti-)periodic boundary conditions, so the discrete space-time is effectively
reduced to a point with a loop attached to it, along which the gauge connection travels. Let
mf > 0 be the mass of the f th flavour, and introduce the notation
() √
µf = 1 + mf .
The partition function of this theory has the matrix model representation [7]
Z K † N †
e λ (TrU +TrU )
,K () ()
Y
ZU (N ) (λ) = dU det µf − U det µf − U
U (N ) f =1
where λ ≡ N gYM 2 is the ’t Hooft coupling for the bare gauge coupling gYM , U ∈ U (N ) is the
plaquette gauge variable and dU is the normalized Haar measure on U (N ). Particularizing to
the degenerate case with all equal masses, µf = µ ∀f = 1, . . . , K, the partition function becomes:
Z h iK N †
ZU,K
(N ) (λ) = dU det (µ − U ) det (µ − U ) †
e λ (TrU +TrU )
U (N )
I N h iK N −1 dzj
e λ (zj +zj )
Y 2
Y
(2.1) = |zj − zk | (µ − zj ) µ̄ − zj−1 .
TN 1≤j<k≤N 2πizj
j=1
The second line is written in terms of the eigenvalues zj ∈ T of U ∈ U (N ).

The matrix model (2.1) is a massive deformation of the model studied in [22, 23]. Besides, (2.1)
is a generalization of the celebrated Gross–Witten–Wadia (GWW) model [4, 5] by determinant
insertions, and reduces to it for K = 0 or µ → ∞. For λ−1 = 0 the model (2.1) is a particular
case of a correlator of characteristic polynomials in a Circular Unitary ensemble (CUE) [24], a
fundamental object in random matrix theory with many applications [25, 26]. Likewise, for this
value of λ−1 = 0, it generalizes a matrix model that describes non-intersecting random walks [27],
and has also appeared in gauge theory, for instance in [28] and later on in somewhat disguised
forms. Still for this particular value, the model describes the probability that a random Young
diagram has a value less than or equal to N [29].
Indeed, it is well-known that there is a rich and intricate connection between problems such as
vicious walkers and the study of random partitions [1, 2]. Interestingly, the formalism of certain
low dimensional gauge theories, certainly 2d Yang-Mills theory but also 3d Chern-Simons theory
for example, share many common features with such, at first, seemingly different areas [30, 31].
One simple reason for this to be the case, is the well-known representation of these theories in
terms of heat-kernels on group manifolds [30–33].
Integrable systems interpretation. In the two different limiting cases K = 0 (pure GWW)
and λ−1 = 0 with µ = 1, the matrix model is known to be the integral representation of a
τ -function of a Painlevé system, in particular Painlevé III0 and V, respectively [34, 29, 35, 36, 1].
In fact, the more general form of the τ -function of PV is very close in form to (2.1), differing
N
though in having (2.1) with e λ z , instead of (2.1) with the full GWW weight.1 If one inspects
the integral representation of the τ -function of PVI [29, 35, 1], it is seemingly unrelated to (2.1)
and that may partially explain why our model here is essentially unstudied, whereas its two
limiting cases appear in many works, often analyzed simultaneously or in a comparative fashion
[34, 36, 29, 35, 1].
1Such an expression also appears in the study of spacing distributions [1, Eq. (8.119)].
4
It is also worth mentioning another possible interpretation of (2.1), from the point of view of
integrable systems [37]. The study of the so-called Schur flow [38, 39], analogous to Toda flows
but on the unit circle, precisely entails the generalization of a given weight function of the matrix
model by multiplication by the GWW weight function.
This induces a flow that has many implications. For example, the recurrence coefficients of
the polynomials, orthogonal with regards to the weight function of the matrix model, satisfy the
non-linear Ablowitz–Ladik equation, with the parameter g21 = Nλ interpreted as time.
YM
Because of this and since our analysis is in a planar limit and centered around the matrix model,
the results obtained are not obviously transferable into this integrable systems and spectral theory
language [40]. We will further discuss about this at the end, in the outlook Section 6.
Gauge theory interpretation. Rather, we consider (2.1) instead as a toy model for lattice
two-dimensional QCD, although we will comment on some of the many other interpretations of
the model. For example, (2.1) can be regarded as an effective description of two-dimensional
QCD on a small spatial circle [28]. In fact, compactifying the spatial direction generates a mass
gap for all but the zero-modes. Taking the small circumference limit, we are left with an effective
theory with integration only over the gauge and matter zero-modes.
It was observed in [22] that the massless theory with fermions, that is = +1 and µ = 1, shows
a Fisher–Hartwig (FH) singularity. The theory with bosons, on the contrary, yields a singular
matrix model in the massless case.2 Here we recognize the singularities encountered in [22] as the
remnants of the IR singularities due to massless fields, and resolve them via mass deformation.
On the parameter µ. Notice that there is a slight difference in the definition of µ between the
fermionic and the bosonic theory in (2.1). In the first case µ > 1 is real, while in the second case
µ ∈ C with |µ| > 1. It is easy to see that the phase of µ ∈ C can be reabsorbed in a rotation of
the integration contour T, and we henceforth restrict our attention to a real µ > 1 in both cases,
with the understanding that for bosons µ really means |µ|.
Besides, treating µ as a real variable with this caveat in mind, the integrand in (2.1) is analytic
in µ > 1, and therefore the results for any other µ ∈ C with |µ| > 1 are obtained from analytic
continuation of our results. In particular, we could not attain negative values of µ moving along
the real line, because we would cross the FH singularity. Nevertheless, it is possible to take a
path from µ > 1 to any µ0 < −1 that runs in the complex plane outside the unit disk.
Remark 2.1. The independence of the partition function on arg µ is the U (1) freedom to choose
the origin of T, and is the incarnation of the residual diagonal U (1) ⊂ U (N ) gauge symmetry.
Our choice arg µ ∈ 2πZ fixes this residual gauge freedom.
Notation. We introduce the notation
1 K
Y = , τ =
λ N
for, respectively, the inverse of the ’t Hooft coupling and a real Veneziano parameter, whose sign
carries information on the type of fields we consider.
The parameter space of the theory is
M = (µ, τ, Y ) ∈ (1, ∞) × R2 .

Remark 2.2. For the role of the mass as a regulator (for the FH singularities on the mathematical
side, for the IR singularities on the field theory side), we do not expect the continuation from M
to the sheet {µ = 1} × R2 , studied in [22, 23], to be analytic.
2According to the discussion in the previous paragraph, these well-known facts in random matrix theory can be
reinterpreted as stemming from Coleman’s no-go theorem [41] for massless bosons in QCD2 .
5
2.2. Exact finite N evaluations. We now present various analytical results for the matrix
model (2.1).
It is possible to evaluate the partition function of any unitary matrix model at finite N via the
Heine–Szegő identity, that for (2.1) gives
(2.2) ZU,K
(N ) (λ, µ) = N ! det [Zjk ] ,
1≤j,k≤N
with
K
X K! dp
Zjk = (−µ)p (1 + µ2 )K−p p Ik−j (2x)|x= N
p!(K − p)! dx λ
p=0
where Ik (x) is the modified Bessel function. We have simplified the expression assuming = +1,
although a similar determinant expression exists for = −1 as well. Formula (2.2) allows to
efficiently compute ZU,K
(N ) exactly for fixed N and K. This is done in Table 1 in Appendix A.1.
The partition function for generic masses, encoded in the parameters (µ1 , . . . , µK ), can be also
related to the expectation value of Wilson loops in arbitrary representations in the pure GWW
model, thanks to the Cauchy identity (see [42] for a similar procedure). For the fermionic theory
we write
   
K I N K
zj z̄j  N (zj +z̄j ) dzj
ZU+1,K
Y Y Y Y
(N ) =
 µ2f  |zj − zk |2  1− 1− eλ
TN µf µf 2πizj
f =1 1≤j<k≤N j=1 f =1
 
YK XX
=  µ2f  sR01 (−µ−1 −1 −1 −1
1 , . . . , −µK )sR02 (−µ1 , . . . , −µK )
f =1 R1 R2
I N
Y Y N dzj
× |zj − zk |2 sR1 (z1 , . . . , zN )sR1 (z̄1 , . . . , z̄N ) e λ (zj +z̄j )
TN 1≤j<k≤N 2πizj
j=1
where the sum runs over Young diagrams R of length at most N and the first row at most K, R0
is the conjugate diagram, and sR is the corresponding Schur polynomial, that is, the character
of the irreducible representation associated to the diagram R. In the last line we identify the
correlator of two Wilson loops in the pure GWW model,
(2.3)  
ZU+1,K
(N )
YK XX
(GWW)
=  µ 2
f sR01 (−µ−1 −1 −1 −1
1 , . . . , −µK )sR02 (−µ1 , . . . , −µK ) hWR1 W R2 i
(GWW)
,
ZU (N ) f =1 R1 R2
with W meaning that the Wilson loop involves conjugated variables. We can in principle further
expand the product of the two Schur polynomials with the Littlewood–Richardson rule, but this
would entail inverting the variables z̄1 , . . . , z̄N in the second Schur polynomial, as in [42].
Expression (2.3) is suggestive but not very useful as it is, because the vacuum expectation value
(vev) of a Wilson loop in a generic representation is not known in closed form. Nevertheless, we
can go deeper in the character expansion thanks to the formula [43]
 
X N |R| N
N Y (N − j)! 
exp TrU = dim R  sR (z1 , . . . , zN ).
λ λ (N − j + Rj )!
R j=1
6
Using this equation and its conjugate and applying twice the Littlewood–Richardson rule we get
 
K
ZU+1,K
Y X
(N ) =  µ 2
f sR01 (−µ−1 −1 −1 −1
1 , . . . , −µK )sR02 (−µ1 , . . . , −µK )
f =1 R1 ,R2 ,R3 ,R4
 
|R3 |+|R4 | N 2
N Y ((N − j)!)
× dim R3 dim R4   c13;24
λ (N − j + R3,j )!(N − j + R4,j )!
j=1
R R with cR
P
where c13;24 ≡ R cR1 R3 cR2 R4 , Rj Rk the Littlewood–Richardson coefficients, and we have
used the orthogonality of the Schur polynomials. The bosonic model ZU−1,K (N ) admits a closely
related expression, with the partitions R1 and R2 instead of their conjugate in the first line and
dropping the restriction on the first rows.
Finally, there is an additional, simpler although more formal closed form expression for ZU,K
(N )
[37]
X 2
,K K K K
ZU (N ) = sR N Y − , − 2 , − 3 , . . .
µ µ µ
R
where now the Schur functions must be interpreted as characters of U (∞), and the sum runs over
Young diagrams R with at most N rows.
3. Phase structure
This section is dedicated to the large N analysis of the matrix model (2.1) and the determina-
tion of its phase diagram.
Write the partition function (2.1) as
I N
−N 2 Seff (z1 ,...,zN )
Y dzj
(3.1) ZU (N ) = e
TN 2πiz jj=1
where the effective action Seff is the sum of a potential and the Coulomb interaction between
eigenvalues:
N N
1 X 1 XX
Seff (z1 , . . . , zN ) = Veff (zj ) + 2 Vint (zj , zk )
N N
j=1 j=1 k6=j
2
Veff eiθ = − cos θ − τ log 1 + µ2 − 2µ cos θ

λ

iθ iϕ
θ−ϕ
Vint e , e = − log 2 sin .
2
writing z ∈ T as z = eiθ , −π < θ ≤ π.
The potential Veff eiθ admits isolated minima at each point in M. Besides, there exists
a surface {λ = λ∗ (µ, λ)} ⊂ M at which it passes from a single-well to a double-well profile.
Explicitly, these two regimes are separated by
(µ − 1)2
λ∗ (µ, τ ) = ,
τµ
and the potential develops stationary points at θ = ±θ∗ with
p
µ2 (2 − λ2 τ 2 ) + 2λµ3 τ + 2λµτ − µ4 − 1
tan θ∗ = ± .
−λµτ + µ2 + 1
We stress that λ∗ is not a critical value of the model (2.1).
7
The potential is plotted in Figure 1 (τ > 0) and in Figure 2 (τ < 0). Clearly, the two figures
have the same shape but upside down. Nonetheless, it is important to distinguish the role of
minima and maxima to understand the phase structure.
Figure 1. Veff (eiθ ) at µ = 3 and τ = 2. Left: λ = 12 . Right: λ = 34 .
Figure 2. Veff (eiθ ) at µ = 3 and τ = −2. Left: λ = − 21 . Right: λ = − 43 .
Now that we have set the ground, we are ready to discuss the large N limit of the model (2.1).
3.1. Large N . We now take the large N ’t Hooft and Veneziano limit of the partition function
(3.1). This means that we consider the planar limit with both λ and τ fixed. The leading
contributions to the integral at large N come from the saddle points of the effective action:
∂Seff
(3.2) =0 j = 1, . . . , N.
∂θj
Introducing the eigenvalue density
N
2π X iθ
ρ(θ) = δ e − eiθj ,
N
j=1
with normalization chosen so that
Z π
dθ
(3.3) ρ(θ) = 1,
−π 2π
we can collect the system (3.2) of N coupled equations in a single singular integral equation at
large N . The saddle point equation then reads
Z π
dϕ θ−ϕ 2µτ sin θ
(3.4) P ρ(ϕ) cot = 2Y sin θ − .
−π 2π 2 1 + µ2 − 2µ cos θ
The solution to (3.4) must satisfy the non-negativity constraint
(3.5) ρ(θ) ≥ 0, −π < θ ≤ π,
8
that follows from the compactness of the integration domain.

Ungapped solution: Phase 0. We begin assuming ρ(θ) is supported on the whole circle,
−π < θ ≤ π. We exploit µ > 1 to obtain the solution [44]
µ cos θ − 1
(3.6) ρ0 (θ) = 1 + 2Y cos θ − 2τ .
1 + µ2 − 2µ cos θ
The derivation is standard [4, 45], thus we omit it. We call Phase 0 the region of M for which
the solution (3.6) is valid.
Critical loci. In those regions of M for which the solution (3.6) violates the constraint (3.5), we
should drop the assumption suppρ = (−π, π] and look for a new solution, whose support has one
or more gaps on the unit circle. The arcs on which ρ is supported are called cuts.
We find a phase transition with a gap opening at θ = ±π at the critical surface
1 τ
(3.7) Ycr,a = + .
2 µ+1
Another phase transition, with a gap opening at θ = 0, takes place at the critical surface
1 τ
(3.8) Ycr,b = − + .
2 µ−1
Besides, there exists a multi-critical point at the value τ = τcr,+ (µ) at which ρ0 (±π) = 0 = ρ0 (0),
determined as the unique point at which Ycr,a and Ycr,b meet:
µ2 − 1
(3.9) τcr,+ (µ) = .
2
Examples of the limiting cases of ρ0 (θ) are shown in Figure 3.
Figure 3. ρ0 (θ) at the transition point. Left: µ = 3, τ = − 12 and λ = λcr,a .

Centre: µ = 3, τ = 1 and λ = λcr,b . Right: µ = 3, τ = 4 and λ = 23 .
Besides the two critical surfaces just described, looking at ρ0 (θ) for negative Y and τ we also
find values at which it attains zero value at two distinct, symmetric points in the interior of
(−π, π), as in Figure 4. We expect a new phase transition into a two-cut solution.
One-cut solution: Phase Ia. We now solve Equation (3.4) dropping the assumption that ρ(θ)
is supported on the whole T, and replace it by the assumption that the support is an arc Γ ⊂ T.
The derivation is standard and we relegate it to Appendix A.2.
Introduce the trace of the resolvent in the large N limit,
Z
dw z+w
(3.10) ω(z) = %(w) , z ∈ C \ Γ.
Γ 2πiw z −w
We adopt the standard notation
ω± (eiθ ) ≡ lim ω(z = (1 ± )eiθ ).
→0+
9
Figure 4. ρ0 (θ) at the transition point. µ = 2, τ = −1 and λ = − 25 24 . The

support of ρ will break in two disjoint cuts beyond this critical value.
Then
ω+ (eiθ ) − ω− (eiθ ) = 2%(eiθ ), eiθ ∈ Γ.
We find (see Appendix A.2 for the details)
" #
1 τ µ 1
q
ωIa (z) = −iW (z)+ (eiθ0 − z) (e−iθ0 − z) Y 1 + −p − .
z 1 + µ2 − 2µ cos θ0 z − µ z − µ−1
The first term is regular and, taking the discontinuity at z = eiθ ∈ Γ, we arrive at
" #
θ p τ µ(µ − 1)
(3.11) ρIa (θ) = 2 cos · 2 cos θ − 2 cos θ0 · Y − p
2 1 + µ2 − 2µ cos θ0 (1 + µ2 − 2µ cos θ)
The angle θ0 is fixed by normalization:
!
µ−1
(3.12) Y (1 − y0 ) + τ p −1 = 1.
1 + µ2 − 2µy0
where y0 := cos θ0 . Equation (3.12) admits a unique real solution, thus the problem is completely
determined.
One-cut solution: Phase Ib. The solution above has been derived assuming that Γ is an arc
along T joining e−iθ0 to eiθ0 running counter-clockwise, thus the gap has opened around θ = π.
For the gap opening at θ = 0, the procedure is identical, but now Γ is an arc from θ̃0 > 0 to
2π − θ̃0 . The procedure of Appendix A.2 leads us to
 
q
θ τ µ(µ + 1)
ρIb (θ) = 2 sin 2 cos θ̃0 − 2 cos θ −Y + q 
2
1 + µ2 − 2µ cos θ̃0 (µ2 + 1 − 2µ cos θ)
which is non-negative definite. There is, however, a more direct route to get the correct answer.
Looking back at the matrix model (2.1) we can chose a different parametrization 0 ≤ θ < 2π,
and the solution with the gap opening at θ = 0 is recovered from the solution (3.11) in Phase Ia
upon replacement Y 7→ −Y , µ 7→ −µ and eventually θ + π 7→ θ.
In conclusion, we have two different phases with a one-cut solution, as expected: one for
Y > Ycr,a (τ, µ), that we have called Phase Ia, and one for Y < Ycr,b (τ, µ), that we have called
Phase Ib.
10
Two-cut solution: Phase II. We have seen that at τ = τcr,+ = (µ2 − 1)/2 the critical surfaces
Y = Ycr,a and Y = Ycr,b meet. Thus, we expect a new phase characterized by a two-cut solution
in the region
µ2 − 1

(µ, τ, Y ) : µ > 1, τ > , Ycr,a < Y < Ycr,b ⊂ M.
2
with gaps around θ = 0 and θ = ±π, and eigenvalue density supported on
n o n o
suppρII = Γ ∼
= Γu t Γd := eiϕ ∈ T : θ̃0 ≤ θ ≤ θ0 t eiϕ ∈ T : −θ0 ≤ θ ≤ −θ̃0 .
That is, Γ is the union of two disjoint arcs, Γu and Γd , as in Figure 5.
Figure 5. The two-cut support Γ in Phase II.
To determine ρII (θ) it is simpler to adopt a different strategy, detailed in Section 3.4 below.
Two-cut solution: Phase III. The fact that the potential Veff (eiθ ) develops a double well for
negative Y and τ in a given range hints at the existence of a two-cut solution in that region
of M, with the eigenvalues sitting around the two minima. This observation is corroborated
looking at the shape of ρIa (θ) and ρIb (θ) in the negative quadrant, where they become negative
in Ycr,b < Y < Ycr,a for τ below a certain threshold.
We find a transition from Phase 0 to a two-cut phase in
Ycr,c+ < Y < Ycr,a and Ycr,b < Y < Ycr,c−
where the critical surfaces Y = Ycr,c± are given by
µ h
2
p i
2 − 1) + (µ4 − 1)] .
Ycr,c± (τ, µ) = 2 µ (τ − 1) − 3τ − 1 ± 2 −τ [2τ (µ
(µ + 1)2
The two curves Ycr,± form an ellipse in each (τ, Y )-leaf of M at fixed µ, with the physical critical
curve being the first branch of the ellipse encountered when decreasing Y from 0.
In this phase, that we call Phase III, the eigenvalues distribute along a contour Γ which consists
of two cuts, with gaps opening around ±θ∗ , see Figure 6.
The eigenvalue density is
p
ρIII (θ) = 2 [cos (θ∗ − δθ) − cos θ] [cos (θ∗ + δθ) − cos θ]
" #
τ µ(µ + 1)(µ − 1)
(3.13) × −Y + p .
(µ2 + 1 − 2µ cos (θ∗ − δθ)) (µ2 + 1 − 2µ cos (θ∗ + δθ)) [µ2 + 1 − 2µ cos θ]
Note that the argument of the outer square root is non-negative definite. The value of θ∗ is
known explicitly, as obtained from Phase 0, and the dependence of δθ on the parameters is fixed
11
Figure 6. The two-cut support Γ in Phase III.
by normalization. Equivalently, we can fix cos (θ∗ + δθ) and cos (θ∗ − δθ) comparing the large z
behaviour of ω(z) computed in this phase with its definition.
For multi-cut solutions, the dependence on the number of eigenvalues filling each cut should be
taken into account when computing physical observables [46]. We analyze the role of the filling
fractions in Appendix B: the upshot is that our conclusions are unaltered, both in phase II and
III, although for different reasons.
3.2. Phase diagram. Putting all the information together, the following phase diagram emerges.
0) When both Y and τ are small, Phase 0 holds, with the eigenvalues spread on the whole
circle.
Ia) When Y > Ycr,a the system is in a new phase, Phase Ia, with a one-cut solution gapped
around θ = ±π.
Ib) Likewise when Y < Ycr,b the system is in Phase Ib, with a one-cut solution gapped around
θ = 0. 2
II) At τ > µ 2−1 the two critical surfaces cross each other. In the region Ycr,a < Y < Ycr,b the
system is in Phase II, a two-cut solution with density of eigenvalues gapped both around
θ = 0 and θ = π.
III) The system develops a new two-cut phase, Phase III, in the region Ycr,b < Y < Ycr,a
and also bounded by an arc of ellipse determined by Ycr,c± . The density of eigenvalues is
gapped around θ = ±θ∗ , with θ∗ → π as Y → Ycr,a and θ∗ → 0 as Y → Ycr,b .
See Figure 7 for a slice of M at fixed µ.
Taking the massless limit µ → 1+ , the critical surface Ycr,b is rotated onto the vertical axis.
Using the analytic dependence on µ, we can also reach µ → −1− by first going to the negative
real axis walking through C outside of the unit disk and then taking the limit |µ| → 1+ . In that
case, it is Ycr,a that is rotated onto the vertical axis.
3.3. Free energy and massless theory. Before delving in the analysis of Wilson loop vevs in
the next section, we comment on the free energy of the model, defined as
1
F= log Z.
N2
12
μ=3 slice of parameter space

Y
4
Ia II
2
0 τ
-4 -2 2 4 6 8
III Ib
-2
-4
Figure 7. Phase diagram of the model in the (τ, Y ) plane, at µ = 3. The blue
straight lines are Y = Ycr,a and Y = Ycr,b , the black curve is Y = Ycr,c± , the
2
red dot is the multi-critical point at τ = µ 2−1 . The gray shaded region in the
ungapped phase, Phase 0. The other light shaded regions are the two-cut phases,
Phase II and III.
The free energy in Phase 0 is easily obtained, and corresponds to the analytic continuation of
Szegő’s strong limit theorem in the bulk of the ’t Hooft parameter space [47]. It takes the value

2 τ 2 1
(3.14) F0 = Y − 2Y − τ log 1 − 2 .
µ µ
It is clearly separated into three contributions: pure gauge (Y 2 ), matter only (∝ τ 2 ) and the
interaction. At strong coupling λ → ∞ (Y → 0) we are left with a matter contribution which
counts gauge singlets: indeed, the integral over the gauge group projects onto gauge invariant
states.
Massless theory. As we have stressed, a core assumption of our analysis is |µ| > 1, and the
massless limit |µ| → 1+ can only be taken at the end. Due to the non-analyticity for µ ∈ T, the
resulting model will differ from a model with massless matter [22, 23].
A main consequence of this non-analyticity is the spontaneous chiral symmetry breaking, that
we will discuss in Section 4.4. On the other hand, it is well known that the large N limit and the
massless limit do not commute.
In the naı̈ve |µ| → 1 limit, the free energy in (3.14) has a logarithmic divergence in the matter
contribution. Setting instead |µ| = 1 from the beginning, and arg µ = θ̃, 0 < θ̃ ≤ 2π, the
partition function acquires a FH singularity and the large N limit cannot be understood by
standard methods. We use known results on Toeplitz determinants to derive the free energy in
Phase 0 in the massless theory [48]:

θ̃
F0 µ = eiθ̃ = Y 2 − 2Y τ cos θ̃ − 2τ 2 log 2 sin + τ 2 log N.

2
That is, the contribution from matter fields has an additional factor of log N and dominates
at large N . Remarkably, this matches the logarithmic divergence of the naı̈ve massless limit of
(3.14). The result is in fact much more general [48] and directly extends to the case of various
13
Veneziano parameters τ1 , . . . , τn associated to different µ1 , . . . , µn that approach the unit circle

from outside at different angles θ̃1 , . . . , θ̃n .
3.4. Stereographic projection. To better understand Phase II and the transition from a one-
cut to a two-cut phase, we map the model onto the real line and study the resulting Hermitian
matrix model at large N . It can be interpreted as a massive deformation of the model in [49].
We conformally map the unit circle on the real line through the stereographic projection, see
Figure 8. The drawback of the stereographic map is that it introduces a puncture on the circle
at θ = ±π: this has no effect at finite N , but the Hermitian matrix model will fail to reproduce
Phase 0 of the unitary matrix model because of this change in topology [45]. Phase 0 and its
associated transitions are well understood from the unitary matrix model side, and we use the
conformally mapped model as yet another way to gain further insight into the one-cut to two-cut
transition.
Figure 8. The stereographic projection. The red cross is the puncture on the
circle, the ticker line is the cut Γ.
Our choice of coordinates is consistent with Phase Ia on the circle, but Phase Ib is easily
retrieved rotating T by eiπ , so that the puncture is placed at θ = 0. The Hermitian matrix model
is
!
1−x2
N j
(1 + η 2 x2j )K 2N Y
Z
1+x2 dxj
ZUp.(N ) = (µ − 1)2K
Y Y
(3.15) (xj − xk )2 e j
RN 1≤j<k≤N j=1
(1 + x2j )K+N 2π
with the superscript p. as notation to remind that it comes from the projection of the model
(2.1). We have adopted the shorthand notation
µ+1
η := , 1 < η < ∞.
µ−1
Remark 3.1. Thanks to the mapping of the Vandermonde determinant form the unit circle to
the real line that shifts τ 7→ τ + 1 in the denominator of the integrand in (3.15), the stability
issues pointed out in [49] do not arise here. We can thus safely allow τ < 0 without spoiling the
convergence of the matrix model.
Phase I. The saddle point equation for the Hermitian matrix model (3.15) is
dy ρp. (y) η2τ
Z
4Y τ +1
P =x + − .
2π x − y (x2 + 1)2 1 + x2 1 + η 2 x2
The solution is found by standard large N techniques [50]. Using a one-cut ansatz for the density
ρp. (x) supported on [−A, A] ⊂ R we find
(3.16) " #
2 x2 − 1 − 2

p τ + 1 τ η 2 A
ρp. 2
I (x) = 2 A − x
2 √ −p − 2Y 3 .
1 + A2 (1 + x2 ) 1 + η 2 A2 (1 + η 2 x2 ) (1 + A2 ) 2 (1 + x2 )2
14
The value of A is fixed by normalization:

Z A
dx p. τ +1 τ A2
ρI (x) = 1 =⇒ √ −p + 2Y 3 = 0.
−A 2π 1 + A2 1 + η 2 A2 (1 + A2 ) 2
As a cross-check, turning off the mass deformation, µ → 1, sends η → ∞ and we recover the
eigenvalue density found in [49]. Besides, sending A2 → ∞ and expanding at leading order in A1
the normalization becomes the consistency condition
1 τ
Y = + ,
2 µ+1
correctly reproducing the critical surface Ycr,a in the limit in which A is back-projected to eiπ . We
stress that, requiring that ρp. (x)dx descends from a measure on T, the non-negativity constraint
ρ(x) ≥ 0 must be imposed.
Looking at ρp.
I (0), we find that the critical point is fixed by the condition
s
2 + A2 1 + A2

2
2Y + τ + 1 − τ η = 0.
1 + A2 1 + η 2 A2
Phase II. From the result above as well as from the analysis of the unitary matrix model, we
find a phase transition to a two-cut solution, with a gap opening at x = 0. The new phase is the
conformal image of Phase II of the unitary matrix model.
We look for a new eigenvalue density, supported on [−A, −B] ∪ [B, A]. The result is
"
p.
p
2 2 2 2
x2 (A2 + B 2 + 2) + 3(A2 + B 2 ) + 2A2 B 2 + 4
ρII (x) = 2 (A − x )(x − B ) |x| −2Y 3
[(1 + A2 )(1 + B 2 )] 2 (1 + x2 )2
#
τ +1 τ η4
(3.17) −p +p .
(1 + A2 )(1 + B 2 )(1 + x2 ) (1 + η 2 A2 )(1 + η 2 B 2 )(1 + η 2 x2 )
The parameters A and B are fixed by normalization,
! !
(A2 − B 2 )2 A2 + B 2 + 2 η 2 A2 + η 2 B 2 + 2
−Y 3 −(τ +1) p − 1 +τ p −1 = 1,
[(1 + A2 )(1 + B 2 )] 2 (1 + A2 )(1 + B 2 ) (1 + η 2 A2 )(1 + η 2 B 2 )
and by an additional self-consistency condition on ω(z),

s
A2 + B 2 + 2 2 (1 + A2 )(1 + B 2 )
2Y + τ + 1 − τ η = 0,
(1 + A2 )(1 + B 2 ) (1 + η 2 A2 )(1 + η 2 B 2 )
which reproduces the criticality condition for B → 0.

This is not the end of the story for Phase II. Indeed, fluctuations in the number of eigenvalues
in each cut may contribute at leading order in the evaluation of observables [46]. However, for
the symmetric two-cut solution we find out that this is not the case, as proved in Appendix B.
4. Wilson loops and instantons

We continue the investigation of the features of the phase transitions and establish their order
by evaluating the vacuum expectation value (vev) of the Wilson loop in the fundamental repre-
sentation. Moreover, we further discuss the different physics of the various transitions by looking
at the different contributions by instantons.
15
4.1. Wilson loops. Wilson loops are order operators in gauge theories that, for simple connected
gauge group, describe the holonomy of the gauge connection around a closed path. For our one-
plaquette model, we consider the Wilson loop in the fundamental representation wrapping the
plaquette, and compute its vev. It is given by
* X N
+
1 1 † 1
hWi = TrU + TrU = cos θj
2N 2N N
j=1
with the average taken in the unitary ensemble (2.1). We use the eigenvalue density at large N
found in each phase to evaluate the Wilson loop.
Wilson loops: Generalities. From the matrix model (2.1) we immediately get the relation
1 1 ∂ 1 ∂F
hWi = Z= .
2N Z ∂(N Y ) 2 ∂Y
Therefore, all the information about the order of the transition can be extracted from the Wilson
loop vev. This is precisely what we expect from an order parameter, and follows from the Wilson
loop belonging to the class of order operators of QCD2 .
Being ρ(θ) continuous on the whole M, the Wilson loop vevs are continuous as well, implying
that every phase transition we find must be at least second order.
Wilson loops: Evaluation. We focus now on the Wilson loop vev at large N . In the ungapped
phase we find
Z π
dθ τ
(4.1) hWi0 = ρ0 (θ) eiθ = Y − .
−π 2π µ
This reproduces the GWW result as τ → 0, but also as µ → ∞, as expected when the matter
becomes non-dynamical. For a Wilson loop winding k > 1 times around the plaquette, either in
clockwise or anti-clockwise direction, we get
τ
hW k i0 = − k .
µ
In Phase Ia the Wilson loop vev is
Z π " #
2 1
r
y − y0 τ µ(µ − 1)
Z
dθ iθ
hWiIa = ρIa (θ) e = dy y Y −p
−π 2π π y0 1−y 1 + µ2 − 2µy0 (1 + µ2 − 2µy)
" #
(1 − y0 )(3 + y0 ) τ 1 + µ(µ − 1)y − µ 3
0
(4.2) =Y − µ2 + 1 + p
4 2µ 1 + µ2 − 2µy0
where we have used the change of variables y = cos θ, with y0 = cos θ0 . The value of y0 as a
function of the gauge theory parameters is known from (3.12).
The Wilson loop vev in Phase Ib is obtained likewise,
" #
(1 + ỹ0 )(3 − ỹ0 ) τ µ(µ + 1)ỹ − µ 3−1
0
(4.3) hWiIb = Y − µ2 + 1 + p ,
4 2µ 1 + µ2 − 2µỹ0
where ỹ0 = cos θ̃0 .
To study the derivative of hWiIa and establish the order of the phase transition, it suffices to
notice that " #
d Y τ (µ − 1)µ(y0 + 1) ∂y0
hWiIa = 1 + (−2y0 − 2) + · .
dY 4 2 (1 + µ2 − 2µy0 )3/2 ∂Y
This implies
d
lim hWiIa = 1
y0 →−1 dY
16
which matches the derivative of hWi0 . The computations are identical for the transition between
d2
Phase 0 and Phase Ib. Taking a further derivative, dY 2 hWi vanishes identically in Phase 0, but
does not vanish at the critical loci when computed in Phases Ia and Ib.
We conclude that the Wilson loop vev is an order parameter of class C 1 at the critical surfaces
Y = Ycr,a (τ, µ) and Y = Ycr,b (τ, µ), thus the system shows a pair of third order phase transitions.
In particular, both the GWW transition [4, 5] and the transition in [27] are special points on the
critical locus of the present model.
Crossing from a one-cut to a two-cut phase, the first derivative of the Wilson loop is not
protected. Indeed, in Phase II the derivative of the Wilson loop vev has the schematic form
(4.4)
Z ỹ0
∂y0 ỹ0 ∂ ∂ ỹ0 ỹ0 ∂
Z Z
d ∂
hWiII = y f (y, y0 , ỹ0 )dy + y f (y, y0 , ỹ0 )dy + y f (y, y0 , ỹ0 )dy,
dY y0 ∂Y ∂Y y0 ∂y0 ∂Y y0 ∂ ỹ0
with the first term coming from the derivative of the explicit dependence on Y , and the other two
from the dependence on Y through y0 and ỹ0 . The integrand evaluated at the endpoint vanishes,
hence those contributions do not appear.
In (4.4), f (y, y0 , y˜0 ) is known explicitly from Section 3.1,
s " #
2 (y − y0 )(ỹ0 − y) τ µ(µ + 1)(µ − 1)
f (y, y0 , y˜0 ) = −Y + p ,
π (1 + y)(1 − y) (1 + µ2 − 2µy0 )(1 + µ2 − 2µỹ0 )(1 + µ2 − 2µy)
but the difficulty comes from the only implicit knowledge of the dependence of y, ỹ0 on Y .
Passing from Phase II to Phase Ia, the first term in (4.4) matches continuously with the
corresponding expression in Phase Ia, as ỹ0 → 1. The integral in the second summand in (4.4)
also agrees with the corresponding contribution in Phase Ia at ỹ0 → 1. Both facts follow from
lim y0 |II = y0 |Ia .
ỹ0 →1
Moreover, the symmetries of the integrand allow to combine the third term in (4.4) with the
second term, in a simpler expression. Moreover, the symmetric form of the equations fixing y0 , ỹ0
can be used to show that
∂ ỹ0 ∂y0
= .
∂Y ∂Y ỹ0 ↔y0
By this we mean that the expressions on the two sides agree upon exchanging all ỹ0 with y0 .
Due to the complicated dependence on the parameters, the derivatives of the boundaries y0 , ỹ0
are not continuous at the transition point. The differentiability of hWi above followed by the
vanishing of the term multiplying such derivatives. This does not happen for the transition from
a two-cut to a one-cut phase. Therefore, the sum of the second and third terms in (4.4) gives an
obstruction to the differentiability of hWi, so we expect a second order transition. In a sense, the
obstruction arises from taking a limit that breaks explicitly the y0 ↔ ỹ0 symmetry of Phase II.
The proof is very similar for the transition from Phase II to Phase Ib or from Phase III to
either Phase Ia or Ib.
The argument fails at the critical surface Ycr,c and at the multi-critical point at which Ycr,a =
Ycr,b . Indeed, when passing directly from Phase 0 to a two-cut phase, the simplifications that arise
from closing both gaps simultaneously imply that the Wilson loop vev is C 1 . This is consistent
with the observation of the previous paragraph, as these transitions preserve the Z2 -symmetry of
the two-cut phase.
4.2. Phase structure and remarks. Summing up the results extracted from the analysis of
Wilson loop vev, we find that
• the transition from Phase 0 to any other phase is third order, but
• the transition from a one-cut to a two-cut phase is second order.
17
In the rest of this subsection we gather comments on various aspects of the phase structure we
uncovered, insisting on the role of the second order phase transitions.
Remark 4.1. As obtained in the previous subsection, the second order discontinuities are finite
jumps, not divergences. The correlation lengths remain finite at each transition. These finite
discontinuities vanish in the limit |µ| → 1.
Metastability. While the third order transitions we find are a continuation of the GWW tran-
sition in M, it is worth to further comment on the second order transitions we obtain. The
phase transition to a two-cut solution happens slightly beyond the values of τ where the potential
develops a double-well structure. The proposal in [51] states that a second order transition can
be associated with tunneling from a metastable vacuum to a stable one. Our analysis confirms
that picture in the one-plaquette model we consider. In Section 4.5 we study instanton effects,
expanding this discussion leading to a further refined distinction between second and third order
phase transitions in this model, from the instantonic point of view.
Critical behaviour. It is worthwhile to notice that the phase diagram in Figure 7 resembles
that in [52], where a unitary matrix model with potential Y1 cos(θ) + Y2 cos(2θ) was analyzed.
The critical behaviour close to a transition to a two-cut phase in our model differs from
that
PK found in similar models in the literature, for matrix models with potentials of the form
n=1 Yn cos(nθ). This is so because the potential in (2.1) includes both a polynomial and a log-
arithmic part, requiring different scaling approaching the critical regime from the two-cut phase.
This distinction, however, fades away approaching the multicritical point.
Double-scaling limit. The statements above can be refined exploiting the double-scaling limit.
In particular, we can zoom in the critical regime, tuning Y towards a transition to the ungapped
phase. In the double-scaling limit, the dynamics is governed by Painlevé II equation. The proof
follows from [53, 27] with minimal variations. An alternative proof can be given using orthogonal
polynomials [54]. We have checked explicitly that, in the double-scaling limit, the problem reduces
to the analogous one for the pure GWW model.
For the transition from a one-cut to a two-cut phase, however, there is no double scaling that
gives Painlevé II.
Other gauge groups. Throughout the work, we focus on gauge theories with gauge group U (N )
or SU (N ). Nevertheless, by direct computation or by universality arguments, it can be shown
that the phase diagram of Figure 7 and the associated phase transitions carry over to theories
with gauge group G, that is to say, matrix models like (2.1) with the integration over U (N )
replaced by integration over G, for any
G ∈ SO(2N ), SO(2N + 1), Sp(N ), O− (2N ), O− (2N + 1) .

Here, O− (n) ⊂ O(n) consists of n × n orthogonal matrices with determinant −1, and Sp(N ) is
the compact symplectic group.
4.3. Continuum limit and β-function. The β-function of the theory, as a function of the ’t
Hooft coupling λ, can be computed using the chain rule through [4]
hWi loghWi
(4.5) β(λ) = 2λ2 ∂
.
∂Y hWi
This quantity can be used to test whether our model reproduces the expected features of QCD2
in the continuum limit. The fixed points of the RG flow, that capture the continuum physics, are
given by β(λ) = 0, which, from (4.5), can only happen at hWi = 0 or at hWi = 1.
Direct computations in Phases 0, Ia, Ib, show that only the solution to hWi = 0 is consistent,
while the solution to hWi = 1 always falls out of the phase in which it has been computed,
and thus should be discarded. It is a nice consistency check that the solution to be discarded
18
is precisely the one that would violate Elitzur’s theorem [55], and the one to be retained is in
agreement with the confining nature of QCD2 [4].
The continuum limit of a lattice theory consists in sending the lattice spacing to zero while
approaching a critical curve [56]. In particular, this requires |µ| → 1.
Taking the continuum limit from Phase 0, approaching either Ycr,a or Ycr,b , we find that the
unique consistent solution is Y = 0 (i.e. λ = ∞). This is physically meaningful for a toy model
of QCD2 : the theory flows to a strongly interacting theory in the deep infrared.
Taking the continuum limit from Phase Ia close to the transition to Phase II, we find a trivial
solution with λ = 0 = τ , describing a theory of free gauge bosons without matter. The continuum
limit approaching the critical surface between Phase Ib and Phase II, instead, yields a non-trivial
fixed point at
1
λ= ≈ 121.4.
Y
Remark 4.2. The existence of a continuum theory is not established by our analysis, because
correlation lengths remain finite. While this has no effect in our model, which consists of a single
plaquette, it may (and most likely shall) wash away the fixed point in the continuum limit of a
more realistic lattice model.
4.4. Chiral symmetry breaking. Let us focus now on the model with fermionic matter. The
fermion two-point function is by definition
1 ∂ ∂ 1
hψ̄f ψf i = − Z=− log Z.
N Z ∂µf ∂µf N
Due to our degenerate choice of masses, we can only compute the average over flavours of such
quantity:
K
* +
1 X 1 ∂F
ψ̄f ψf = .
K τ ∂µ
f =1
In Phase 0 we find
1 ∂F0 2 τµ
= 2 Y + .
τ ∂µ µ µ2 − 1
This quantity diverges as µ → 1, therefore we expect the chiral symmetry to be spontaneously
broken in the continuum, consistently with the analysis of the β-function in Phase P
0.
In Phases Ia and Ib, we can move along Y = 0 and study the behaviour of h K1 f ψ̄f ψf i on
that subspace of M. The result is read off directly from [45]:

1 ∂FIa µ + 1 + 4τ
=−
τ ∂µ Y =0 2τ µ(µ − 1)
in Phase Ia, which is non-vanishing and continuous at the transition point.
In Phase Ib we get

1 ∂FIb 1 − µ + 4τ
=− ,
τ ∂µ Y =0
2τ µ(µ + 1)
again non-vanishing and continuous at the transition point, and goes to 1 in the µ → 1+ limit.
This latter result, in turn, hints at a transition to a free theory:
P the free energy of a theory of
K free flavours of mass m goes as Ffree ∝ Km, whence h K1 f ψ̄f ψf i|free = 1. Note that this
computation has been done at infinite gauge ’t Hooft coupling, which has the physical meaning
of governing the theory in the deep infrared.
To sum up, we have observed that the phase transition from Phase Ib to Phase 0 is accompanied
with spontaneous chiral symmetry breaking.
19
4.5. Instantons. We discuss non-perturbative effects in the unitary matrix model, coming from
unstable saddle point configurations [57].
P An instanton configuration is characterized by a collection of integers {N0 , N1 , . . . } with
k Nk = N . For example, the d-instanton configuration is associated with the symmetry break-
ing pattern
U (N ) → U (N0 ) × U (N1 ) × · · · × U (Nd ),
with the eigenvalues zj ∈ T of U ∈ U (N ) grouped in d different sets, sitting at d different extrema
of the potential. For the one-instanton configuration,
N
X
ZU (N ) (ν) = e−ν` Z`
`=0
with Z` the partition function of a U (N − `) × U (`) model, and we have turned on a chemical
potential ν > 0 for the instanton number.
The `-sector leads to non-perturbative corrections to the free energy of the matrix model, of
the form
~
e−N `Sinst (λ) f` (~λ)
where ~λ generically denotes the couplings of the theory, and the functions {f` }` admit themselves
a N1 expansion.
Instanton effects and third order transitions. Let us consider our model (2.1) at large N and
focus on the one-cut phase, in which the interpretation of instanton effects is more transparent.
We discuss them in Phase Ia, being the corresponding analysis in Phase Ib completely analogous.
Most of the details are just an extension of the thorough analysis in [57].
The contribution of an instanton excitation, obtained moving one eigenvalue from the minimum
of Veff to the maximum at θ = ±π is found to be
q
2 2 −1 1
πSinst = 2Y 1 − x0 − x0 cosh
x0
" s ! !#
−1 1 − x20 µ−1 x0
+ τ tanh (µ − 1) +p log
(µ − 1)2 + 4µx20
p
(µ − 1)2 + 4µx20 1 + 1 − x20
where cosh−1 and tanh−1 are the inverse of the hyperbolic functions, and x0 = sin θ20 . One of the
results in [57] (already conjectured in [58]) is that the GWW transition is triggered by instantons.
We see that the result carries over to the present model, as
lim Sinst = 0
x0 →1
and the instanton excitations cease to be suppressed at the critical point when the gap closes.
Analogous conclusions are found if we go to Phase III, in which the effective potential has
developed a double well, and consider the instanton configuration with a few eigenvalues taken
to the local maximum at θ∗ . Approximating close to the transition to Phase 0, we find

2 τ µ(µ + 1)
3

πSinst = (δθ) sin θ∗ −Y + + O (δθ) .
(µ − 1) (1 + µ2 − 2µ cos θ∗ )
At the critical surface, δθ → 0 and we find again that the third order transition is triggered by
instantons.
20
Figure 9. Instantons in Phase III. A single eigenvalue (blue dot) is moved on

top of the maximum of Veff , while all the others (gray sea) fill the minima.
Instanton effects and second order transitions. We now turn our attention to the analysis
of instanton effects in the two-cut phase, starting from Phase III. We consider a single eigenvalue
placed on the maximum of the potential, as sketched in Figure 9.
We find that the instanton action is the sum of two pieces,
Z y∗ s " #
dy (y − yL )(yR − y) τ µ(µ2 − 1)
Sinst,L = −Y + p ,
yL π 1 − y2 (1 + µ2 − 2µyL )(1 + µ2 − 2µyR )[1 + µ2 − 2µy]
Z yR s " #
dy (y − yL )(yR − y) τ µ(µ2 − 1)
Sinst,R = −Y + p ,
y∗ π 1 − y2 (1 + µ2 − 2µyL )(1 + µ2 − 2µyR )[1 + µ2 − 2µy]
where y∗ = cos θ∗ and yL,R = cos(θ∗ ± δθ). The two are associated with the eigenvalue escaping
from the left and right cut, respectively. There exists a third relevant quantity, namely the
tunneling from one cut to the other,
Z yR s " #
dy (y − yL )(yR − y) τ µ(µ2 − 1)
Stunnel = −Y + p .
yL π 1 − y2 (1 + µ2 − 2µyL )(1 + µ2 − 2µyR )[1 + µ2 − 2µy]
All the three effects are non-perturbatively suppressed by a factor e−N Sinst , with Sinst the cor-
responding action. The three contributions are still suppressed at the critical loci, although the
tunneling term will coalesce with one of the other two.
The situation is slightly different in Phase II, where the two wells have equal depth, see Figure
10. In this case, Stunnel is simply twice Sinst , and both go to zero as a gap closes. The phase
transition takes place when the tunneling between the two cuts ceases to be suppressed in one
direction (e.g. passing through θ = 0 in Figure 10) but remains non-perturbative in the other
direction (e.g. passing through θ = π in Figure 10).
The picture we infer is that the third order phase transitions are associated with releasing
non-perturbative instabilities, while the second order transitions correspond to release only those
instabilities in one direction. This is also in agreement with the proposal in [51, 18] relating
second order phase transitions in GWW-type models to partial deconfinement.
5. Meromorphic deformation of unitary matrix models

We now depart from the model (2.1) with the aim of setting the stage for the study of mero-
morphic deformations of unitary matrix models, in which the integration contour is deformed in
C∗ and not bound to be the unit circle. This consists of an adaptation of the theory of holo-
morphic matrix models [19] to unitary matrix models and, as we shall show, is instrumental in
21
Figure 10. Instantons in Phase II. A single eigenvalue (blue dot) is moved on
top of the local maximum of Veff at θ = 0, while all the others (gray sea) fill the
minima.
understanding their phase diagram from new angles. This section can be read independently of
the rest of the paper.
N
A unitary matrix model is characterized by a weight function e− λ V (z) , with V (z) admitting
the expansion
X tn t−n −n

(5.1) V (z) = zn + z .
n n
n≥1
−N
The function e V (z) is singular at z ∈ {0, ∞} ⊂ P1 and possibly has other zeros and poles
λ
in C∗ ∼
= P1 \ {0, ∞}. The Vandermonde determinant appearing in a unitary matrix model is
conveniently rewritten in meromorphic form
 
N
Y 1 1 Y 1 Y
(zj − zk ) − = N −1
 (zj − zk )2 .
zj zk z
1≤j<k≤N j=1 j 1≤j<k≤N
To deform a unitary matrix model, the integration cycle TN is replaced by any half-dimensional
cycle CN in (C∗ )N .
Definition. Let N ∈ N, λ ∈ C∗ and V (z) as in (5.1). A meromorphic matrix model Z is the
integral
I N
Y Y N dzj
(5.2) Z= (zj − zk ) 2
e− λ V (zj ) ,
CN 1≤j<k≤N j=1
2πizjN
with integration contour

X X
(5.3) CN = N ` C` , N` = N, [C` ] = [T] ∈ H1 (C∗ ).
` `
Condition (5.3) means that each C` is homotopic to the unit circle in the holed plane C∗ .
Dropping it, we may stretch C` along any direction along which, asymptotically, < λ1 V (z) > 0.
Eventually the one-cycle C` pinches at z = ∞ ∈ P1 . To get an honest deformation of a unitary
matrix model we do not allow this situation, otherwise we would fall back in the holomorphic
deformation of a Hermitian matrix model.
The couplings {tn } in (5.1) are usually subject to reality conditions, such as
t−n = t̄n , ∀n ≥ 1,
22
and possibly other relations. We write the constraints collectively as Φ ~ ({tn }) = 0. Besides, a
rescaling of all {tn } together can be reabsorbed in a redefinition of λ, hence the couplings {tn }
are homogeneous coordinates on a projective space.
The matrix model (5.2) sets a natural stage to complexify the couplings. We denote by T
the physical space of couplings, namely the collection of independent {tn } after imposing the
constraints and modulo scaling. More formally,
n o
(5.4) T = ({tn ∈ C, ∀n 6= 0} / C∗ ) ∩ Φ~ ({tn }) = 0 ,
with the C∗ -action being multiplication of all couplings by a non-vanishing constant. Whenever
~ can be rewritten in homogeneous form, T is a projective variety.
the constraints Φ
5.1. Large N limit. We are interested in the large N limit of (5.2). Define the effective potential
1
W (z) := V (z) + log z.
λ
At large N , the eigenvalues will be gathered around the saddle points of W (z) in C∗ ,
W 0 (zsp,` ) = 0, ` = 1, 2, . . . , g + 1.
Here we are assuming there is a finite number g + 1 of saddle points zsp . The integration contour
CN in (5.3) can be chosen in such a way that each C` passes through zsp,` . The integers N` in
(5.3) then count the number of eigenvalues around the `th saddle point zsp,` . At large N , the
density of eigenvalues will vanish on C` away from a compact interval Γ` ⊂ C` , called a cut, with
zsp,` ∈ Γ` . Therefore
g+1
[
suppρ = Γ` =: Γ,
`=1
and ρ(z) is normalized.
Remark 5.1. The requirement that the integration cycle passes through all the g + 1 saddle
points does not fix it uniquely. The shape of each C` , and thus of the cuts Γ` at large N , can be
homotopically deformed in an open neighbourhood of zsp,` , meaning that the matrix model (5.2)
depends on (up to) g additional parameters.
In the large N limit, the eigenvalue density solves the saddle point equation
Z
dw ρ(w) 1
(5.5) = W 0 (z),
Γ 2π z − w 2
where 0 means holomorphic derivative ∂z ∂
. Here, dw is a holomorphic differential on Γ and, given
any parametrization w : s 7→ w(s) ∈ Γ, dw = ẇ(s)ds is understood, with ds the line element and
ẇ the derivative of the map w : s 7→ w(s).
Recall the definition of the trace of the resolvent ω(z) at large N :
Z
ρ(w)
ω(z) := dw , z ∈ P1 .
z−w
Equation (5.5) implies that ω(z) solves [50]
(5.6) ω(z)2 − W 0 (z)ω(z) + f (z) = 0,
where
W 0 (z) − W 0 (w)
Z
dw
f (z) = ρ(w) .
Γ z−w 2π
Defining
1
(5.7) y(z) = ω(z) − W 0 (z),
2
23
(5.6) becomes
2
2 1 0
(5.8) y − W (z) + f (z) = 0.
2
This equation goes under the name of spectral curve. The steps from (5.5) to (5.8) are standard
and have been applied to holomorphic matrix models since their early days [59] to establish a
bridge between matrix models and geometric problems. The novel aspect of (5.8) compared to
the literature is hidden in the form of W 0 and f , which in the present case are not ordinary
polynomials but Laurent polynomials, or meromorphic functions.
We assume for now that V (z) is a Laurent polynomial on P1 , with singularities at {0, ∞}. The
extension to a meromorphic weight function on C∗ is worked out below. Write
d+ +1 d+
X tn n 1 X 1
(5.9) V (z) = z =⇒ W 0 (z) = sgn(n)tn+1 z n + ,
|n| λ z
n=−d− +1 n=−d−
where we assume d− ≥ 2 (otherwise we get back the known setting).

We need to introduce some notation. Define
(5.10) g = d+ + d− − 1,
which agrees with the counting of saddle points above, and also t0 = −λ for later convenience.
Besides, denote ρk the moments of the eigenvalue density,
Z
ρk = wk ρ(w)dw, k ∈ Z.
Γ
After some rewriting we get
g+1
1 X n
(5.11) W 0 (z) = z tn+1−d− sgn(n − d− ),
λz d−
n=0
 !
d− −1 n−1 g g
!
1 X n X X X
(5.12) f (z) = d− z tk+2−d− ρk−n + zn tk+2−d− ρk−n  ,
λz
n=0 k=−1 n=d− k=n
where sgn(0) = +1 by convention. The spectral curve takes the schematic form
P (z)
(5.13) y2 =
4λ2 z 2d−
where P (z) is a polynomial in z of degree deg(P ) = 2g + 2, with coefficients read off from
(5.11)-(5.12) and that depend on the parameters {tn }, on λ and on the moments {ρk }.
A major difference with respect to the standard unitary matrix models is that {ρk } are free
complex parameters of the theory: they can be tuned deforming Γ, as discussed in Remark 5.1.
Recalling that both z, λ ∈ C∗ , it is possible to recast (5.13) in a more standard form ŷ = P (z),
describing an hyperelliptic complex curve of genus g [59].
Γ is the union of g + 1 branch cuts stretched between pairs of roots of P (z). The roots of
P (z) move inside C∗ as the parameters are varied.3 The coalescence of two roots produces a
singularity of the curve (5.13) and corresponds, on the matrix model side, to a phase transitions
from a (g + 1)-cut to a g-cut phase, with either
• two cuts joining, or
• one cut collapsing.
3Without loss of generality, {0, ∞} ⊂ P1 are not roots of P (z), because they would correspond to a “non-
minimal” choice of d± in (5.9). They can be avoided simply defining ŷ = 2yλz d− ±m with m the multiplicity of the
root, and minus (resp. plus) sign if the root is z = 0 (resp. z = ∞).
24
The hyperelliptic curve (5.13) is fibered over the moduli space M of the model (5.2), defined
as
M = C∗ ×T × Cg ,
with C∗ parametrized by λ, T defined in (5.4), and the last factor parametrized by the moments
{ρk }. Note that one of the moments is fixed comparing (5.6) with the definition of ω(z) at
|z| → ∞.
Definition. A critical locus C is an irreducible component of the locus in M at which two roots
of P (z) coalesce.
The critical loci C ⊂ M necessarily have positive complex codimension, and the hyperel-
liptic fibration is singular along them. Singularities in higher codimension, placed at the (self-
)intersection of critical loci, correspond to multicritical points of the matrix model.
The theory of Abelian differentials provides a suitable framework to analyze the genus g hy-
perelliptic curve (5.13) [60]. At this stage, the analysis of the spectral curve works exactly as in
the holomorphic deformation of Hermitian matrix models, thus we omit the details and refer to
[60, 61].
Remark 5.2. We are now in the
position to elaborate more on Remark 5.1, from a point of view
advocated in [62]. Let A` , B` be a basis of one-cycles in the hyperelliptic complex curve. The
A-cycles are chosen to go around the cuts Γ` . Therefore
I I
N`
y(z)dz = ω(z)dz = =: ξ` .
A ` A ` N
The first equality follows from the definition (5.7) noting that y(z) and ω(z) only differ by a
regular term. Introducing chemical potentials for the filling fractions ξ` and extremizing the
action with respect to these quantities gives their saddle point value as a function on M . More
precisely, one gets a set of equations analytic in the ratios s` := ξλ` [62]. At this point, it is
possible to invert the relations and express the moments {ρk } in terms of the complex variables
s` , keeping the latter as free parameters.
Note that only g out of the g + 1 of both quantities are free.
The study of the phases of the model (5.2) leads to a stratification of the parameter space M .
We postpone the analysis to Section 5.4, discussing explicit models first.
Genus 0. The unique way to obtain a genus 0 spectral curve is from the holomorphic deformation
of the CUE. In that case, the model has no couplings and, as opposed to g ≥ 1, the additional
condition derived from the definition of ω(z) is automatically fulfilled, leaving ρ−1 as unique,
unconstrained parameter. Then, (5.13) describes a P1 fibered over C. If we try to get a less
trivial model by considering the insertion of (det U )τ N , the consistency condition, which in g ≥ 1
fixes one of the {ρk }, imposes τ = 0.
5.2. Holomorphic GWW. We now put the machinery at work and revisit the phase structure
of the holomorphic GWW model. The phase diagram of this model has been obtained in [63]
for λ ∈ R, while the behaviour at complex coupling has been partially analyzed in [64], although
without fully exploiting the holomorphic deformation.
The GWW model has t−1 = t1 = 1, and tn6=±1 = 0, whence d+ = 0, d− = 2, g = 1 and only
the moments ρ−2 , ρ−1 appear in (5.12). Fixing ρ−2 as a function of λ and ρ−1 , the spectral curve
of the holomorphic GWW model is [63]
ŷ 2 = z 4 + 2λz 3 + (ρ−1 + 1)λ2 − 2 z 2 + 2λz + 1.

(5.14)
It is an elliptic curve. Following the strategy outlined above, we think of (5.14) as an elliptic
fibration over C∗ × C, with coordinates on the base λ and ρ−1 , and identify the phase transitions
with singularities of the fibration.
25
The discriminant of (5.14) is

λ2 2 2
(5.15) ∆= λ ρ−1 − 4 (λ (ρ−1 + 1) − 4) (λ (ρ−1 + 1) + 4) ,
4
from which the critical loci are
{∆ = 0} = C1 ∪ C10 ∪ C2 ,

4 ∗ 4 ∗
C1 := ρ−1 = −1 + , λ ∈ C , C10 := ρ−1 = −1 − , λ ∈ C ,
λ λ

4
C2 := ρ−1 = 2 , λ ∈ C∗ .
λ
In Kodaira’s classification [65], C1 and C10 are singularities of type I1 and C2 is of type I2 . The
GWW critical points (λ, ρ−1 ) = (±2, 1) are singled out as the codimension-two singularities at
which C2 intersects one of the other two critical curves. Besides, we recognize the elliptic curve
(5.14) as the Seiberg–Witten curve of N = 2 supersymmetric four-dimensional SU (2) gauge
theory with two flavours [66].
The original GWW transition is thus, from the perspective of the holomorphic deformation, one
of the possible ways to approach the codimension-two singularity from a generic direction. The
singularity at the multicritical points (λ, ρ−1 ) = (±2, 1) is of Kodaira type III. The corresponding
symmetry is A1 , which is precisely the symmetry of Painlevé II, that is known to control the
GWW phase transition [54, 53]. If, instead, we approach the multicritical point not from a
generic direction but moving along a critical locus, the singularity type is enhanced to I∗0 .
It is possible to allow t−1 6= t1 . This corresponds to introduce a θ-term in the GWW lattice
action, 2πθ
= t1 −t
2
−1
. The procedure goes through with only minor modifications, the unique
difference being that the singularities C1 , C10 are placed at ρ−1 = − t−1
1
± λ√4t−1 , so in particular
the Z2 -symmetry is preserved.
5.3. Meromorphic deformations. The formulation can be extended to include weight func-
tions with zeros and poles in C∗ . For concreteness, we consider the illustrative example of our
original model (2.1) at λ−1 = 0. In this case

0 1 1 1+τ
W (z) = −τ + −1
+ ,
z−µ z−µ z
ρ̃−1 (µ) ρ̃−1 (µ−1 )

(1 + τ )
f (z) = τ + −1
− ρ−1 .
z−µ z−µ z
In the second line, we have defined
Z
ρ(w)
ρ̃−1 (µ) := dw,
Γ w−µ
with, in particular, ρ̃−1 (0) = ρ−1 . Comparing the definition of y(z) with the spectral curve at
large |z|, we find a pair of consistency conditions, fixing ρ̃−1 as a function of the other parameters,
4ρ−1 (τ + 1) + µτ (τ + 6) + µ
ρ̃−1 (µ) = − .
4 (µ2 − 1) τ
Note that the two conditions fix ρ̃−1 (µ±1 ) independently, and the solutions are consistently
mapped into each other under µ ↔ µ−1 . We get
P (z)
(5.16) y2 = ,
4z 2 (z − µ)2 (z − µ−1 )2
26
with P (z) a polynomial of degree 4. The spectral curve

thus describes again en elliptic fibration
over the moduli space C∗ × µ ∈ P1 : 1 < |µ| < ∞ × C, parametrized by (τ, µ, ρ−1 ). The dis-

criminant takes the form ∆ = µ5 (µ + 1)2 (µ − 1)2 (τ + 1)4 ∆,e the last term being a cumbersome
polynomial of degree 6 in ρ−1 , degree 8 in τ and degree 10 in µ. The critical points of the unde-
formed model become higher-codimensional singularities, at which two roots of P (z) collide. The
collection of all critical loci in this model is
6
[
C4 ∪ C2 ∪ C20 C1j ,
j=1
with the subscript indicating the order of vanishing of ∆ along the component C . Taking what
we have called the continuum limit in Section 4.3, that is, sending τ → τcr (µ) and then µ → ±1,
with ρ−1 = − µτ set to its undeformed value, yields a non-minimal singularity ∆ ∝ (µ ± 1)12 .
5.4. Stratification of the moduli space. The critical loci and their intersections endow the
parameter space M with additional structure.
The stratification of an algebraic variety V is a collection of open sets {VI }, with V0 a point
and V max = V , with a partial order given by the inclusion of the closures of {VI }. The parameter
space M of the model (5.2) is the union of
reg
M reg , CIreg , CIJ , ...
where the superscript means the regular part, and CIJ = CI ∩ CJ , and so on. The inclusion
reg
relations CIJ = CI ∩ CJ ⊂ CI are obvious.
The partial order can be represented with the aid of a Hasse diagram:
M reg
CIreg CJreg ··· CKreg
reg reg
CIJ ··· CIK
.. .. ..
. . .
In general, this does not define a full-fledged stratification of M because multiple final points
may exist. Nonetheless, whenever the potential (5.1) has a Z2 -symmetry, the Hasse diagram
inherits it. This Z2 -symmetry acts as an automorphism of the Hasse diagram, which is mapped
into itself under reflection along the vertical axis. By construction, the Hasse diagram resulting
from folding the initial diagram via this Z2 -symmetry determines a stratification of M /Z2 .
We draw the Hasse diagram of the holomorphic GWW model of Section 5.2:
M reg
C1reg C2reg C1reg

0
(5.17) λ=2 λ = −2
27
The diagram of the meromorphic model of Section 5.3 is schematically
M reg
C1reg
1
C1reg
2
C1reg
3
C2reg C4reg C2reg
0 C1reg
4
C1reg
5
C1reg
6
.. ..
. .
µ=1 µ = −1
The vertical red, dashed line is there to emphasize the Z2 reflection symmetry. Folding the
diagram along that line yields the stratification of M /Z2 .
Remark 5.3. The results of Section 5.2 with t−1 6= t1 show that, even for models in which a Z2
reflection symmetry is not manifest from the potential but emerges at large N , the Z2 -folding
yields a stratified moduli space.
Symplectic singularities. Recall that H1 (Σg , C), the first homology group of a hyperelliptic
curve Σg of genus g, is a symplectic space. The A- and B-cycles that we have implicitly used
in the study of the spectral curve (5.8) can be chosen to be Darboux coordinates in H1 (Σg , C).
Moving along M reg corresponds to vary the symplectic structure without changing the topology
of Σg . At the critical loci CI , however, either
• a B-cycle collapses, or
• an A-cycle collapses.
Both situations correspond to a singularity of the symplectic form. Therefore, the analysis of
the phase structure of the meromorphic matrix models can be rephrased in terms of symplectic
singularities in the sense of Kaledin [67].
The appearance of symplectic singularities is not entirely unexpected. The consideration of
holomorphic matrix models in their large N limit lead to the Seiberg–Witten curves [66] of certain
N = 2 gauge theories [59]. The so-called Coulomb branch of these theories is a symplectic
singularity and is stratified [68]. In fact, the use of Hasse diagrams in the present work was
inspired by [69, 70].
As a final remark, we emphasize that the structure uncovered in this section is not specific of
the meromorphic matrix models. The parameter spaces of unitary or Hermitian matrix models
inherit it, as they can be realized as slices inside the parameter space of our meromorphic models.
As an example, the phase diagram of the GWW model is
M reg |GWW = λ ∈ R \ {0} , λ2 6= 4

•
C1 |GWW = {λ = 2} • • C10 |GWW = {λ = −2}

(5.18)
It is found by fixing ρ−1 = 1, taking the slice λ ∈ R \ {0} and identifying the intersection of such
subspace with the strata in (5.17).4
4Accidentally, this is precisely the Hasse diagram of the reduction to three dimensions of the SU (2) theory with
two flavours, captured by the holomorphic GWW of Section 5.2, cf. [70, Eq.(4.2)]. It should be stressed, however,
that the strata in (5.18) are real, not hyperKähler.
28
6. Outlook
We conclude by commenting, in a qualitatively and non-exhaustive fashion, on avenues that
we have considered at some point but not pursued here.
It would be interesting to know if the results obtained have a meaning from the point of view
of integrable systems such as the Schur flow. While both the matrix models considered and the
study of such flows have in common an associated system of orthogonal polynomials, the way
this association actually works is quite different. For example, the recurrence coefficients of the
polynomials, central in the integrable systems description, are not directly relevant in the type
of matrix model analysis presented.
On the other hand, for what concerns matrix models on the real line, the spectral properties
of Jacobi matrices can be more directly related to matrix models, since it is known that, under
rather general conditions, a suitably normalized counting measure of the zeroes of the orthogonal
polynomials converges weakly, in the large N limit, to the density of states of the matrix model
[71]. With this in mind, a question would be whether our results have any implication in the
study of the spectral properties of CMV matrices [40], for example. The large N planar limits
taken make this possibility not obvious.
Another reason to further study any eventual implications of the planar limit and the ensuing
phase structure, from the point of view of integrable systems, would be the connection obtained,
presented in Section 5.4, between the phase diagram of meromorphic matrix models and the
symplectic foliation of singular varieties [67]. Again, the ’t Hooft scaling of the couplings involved
at large N obscures the relation between the symplectic structures we naturally find and those
in the integrability literature [72].
Also, from a mathematical point of view, it would be interesting if proofs of the order of
the phase transition, in particular for the second order phase transitions, can be obtained in
alternative or more rigorous ways.
Regarding more physical considerations, when discussing the model with fermionic matter and
its interpretation in terms of chiral symmetry breaking, it is worth mentioning that recently [73],
the chiral symmetry breaking phase transition in four-dimensional QCD has been studied from
the point of view of thermodynamic geometry [74, 75]. The argument is based on the observation
that the grand canonical partition function
∞
X
Z(v)
b = e−N v ZU (N ) , v>0
N =0
determines a metric g thermo on a two-dimensional parameter space with coordinates (F, v) [74],
where F is the free energy and v the grand canonical chemical potential. Then, a second order
phase transition is triggered by the instability at det g thermo = 0.
Any eventual use of this observation or other ideas from information geometry to further
understanding phases in matrix models would be of interest.
Acknowledgements. We thank Jorge Russo for a careful reading and valuable commentaries.
The work of LS is supported by the Fundação para a Ciência e a Tecnologia (FCT) through the
doctoral grant SFRH/BD/129405/2017. The work is also supported by FCT Project PTDC/MAT-
PUR/30234/2017.
Appendix A. Technical details of the solution

A.1. Exact expressions at finite N via Toeplitz determinants. Table 1 collects the ex-
plicit expressions for ZU=+1,K
(N ) for the first few values of N and K, computed using the Toeplitz
determinant formulation detailed in Section 2.2.
29
N K ZU+1,K
(N )
1 1 (µ2 +1)I0 (2Y )−2µI1 (2Y )
1 2 (µ4 +6µ2 +1)I0 (2Y )− 2µ Y (
2Y (µ2 +1)+µ)I1 (2Y )
1
Y2
[Y (Y (µ +15µ +15µ +1)+4µ3 )I0 (2Y )
6 4 2
1 3
−2µ(Y 2 (3µ4 +10µ2 +3)+3Y (µ3 +µ)+2µ2 )I1 (2Y )]
2
2(µ +µ2 +1)I0 (4Y )2 −4µI0 (4Y )((µ2 +1)I1 (4Y )+µI2 (4Y ))−2(µ2 −1) I1 (4Y )2
4
2 1
−2µ2 I2 (4Y )2 +4µ(µ2 +1)I1 (4Y )I2 (4Y )
1
(2Y )4
[8Y 2 (4Y 2 (µ8 +8µ6 +30µ4 +8µ2 +1)−8Y (µ5 +µ3 )−µ4 )I0 (4Y )2
−8Y µ(4Y (µ2 +1)+µ)I0 (4Y )((4Y 2 (µ4 +6µ2 +1)−µ2 )I1 (4Y )+2Y µ(4Y (µ2 +1)+µ)I2 (4Y ))
4

2 2 −2 16Y 4 (µ2 −1) −128Y 3 (µ5 +µ3 )+8Y 2 (µ6 +4µ4 +µ2 )+µ4 I1 (4Y )2
2
−8Y 2 µ2 (4Y (µ2 +1)+µ) I2 (4Y )2
+8Y µ(16Y 3 (µ6 +7µ4 +7µ2 +1)+4Y 2 (µ5 +6µ3 +µ)+4Y (µ4 +µ2 )+µ3 )I1 (4Y )I2 (4Y )]
1
(2Y )6
[8Y 2 (−4Y 2 (27µ4 +62µ2 +27)µ4 −16Y 3 (5µ6 −21µ4 −21µ2 +5)µ3
+16Y 4 (µ12 +21µ10 +195µ8 +334µ6 +195µ4 +21µ2 +1)−36Y (µ7 +µ5 )−9µ6 )I0 (4Y )2
2
−8Y µI0 (4Y ) 2Y µ(4Y 2 (3µ4 +10µ2 +3)+6Y (µ3 +µ)+3µ2 ) I2 (4Y )
+(32Y 5 (3µ10 +55µ8 +198µ6 +198µ4 +55µ2 +3)+16Y 4 µ(3µ8 +60µ6 +130µ4 +60µ2 +3)
−32Y 3 µ2 (2µ6 +3µ4 +3µ2 +2)−8Y 2 µ3 (9µ4 +22µ2 +9)−36Y (µ6 +µ4 )−9µ5 )I1 (4Y ))
2 3 2 6 3
−2 4Y 2 (9µ4 +26µ2 +9)µ4 +96Y 4 (µ2 −1) (µ4 +4µ2 +1)µ2 +64Y 6 (µ2 −1) −2048Y 5 (µ3 +µ)
+48Y 3 (µ9 +11µ7 +11µ5 +µ3 )+36Y (µ7 +µ5 )+9µ6 )I1 (4Y )2
2
−8Y 2 µ2 (4Y 2 (3µ4 +10µ2 +3)+6Y (µ3 +µ)+3µ2 ) I2 (4Y )2
+8Y µ(32Y 5 (3µ10 +55µ8 +198µ6 +198µ4 +55µ2 +3)+16Y 4 µ(3µ8 +60µ6 +130µ4 +60µ2 +3)
+96Y 3 (µ8 +8µ6 +8µ4 +µ2 )+24Y 2 µ3 (3µ4 +10µ2 +3)+36Y (µ6 +µ4 )+9µ5 )I1 (4Y ) I2 (4Y )]
2
6 −(µ2 (−I2 (6Y ))+(µ2 +1)I0 (6Y )−µI1 (6Y )+µI3 (6Y )−I2 (6Y ))((µ2 +1)I1 (6Y )−µ(I0 (6Y )+I2 (6Y )))
2 2

+((µ2 +1)I0 (6Y )−2µI1 (6Y )) ((µ2 +1)I0 (6Y )−2µI1 (6Y )) −((µ2 +1)I1 (6Y )−µ(I0 (6Y )+I2 (6Y )))
3 1 2
+((µ2 +1)I2 (6Y )−µ(I1 (6Y )+I3 (6Y ))) ((µ2 +1)I1 (6Y )−µ(I0 (6Y )+I2 (6Y )))
−((µ2 +1)I0 (6Y )−2µI1 (6Y ))((µ2 +1)I2 (6Y )−µ(I1 (6Y )+I3 (6Y )))))
Table 1. Analytic evaluation of ZU+1,K

(N ) in terms of modified Bessel functions for
various N and K.
A.2. Large N limit: Gapped solutions. In this appendix we sketch the computation of ω(z),
defined in (3.10), which allows to extract the eigenvalue density in the phases with one or more
gaps. The procedure is standard and we follow closely [76, 47], glossing over many details. We
work in Phase Ia, since all other phases are analyzed in similar fashion.
Introduce the function %(z) of complex variable z ∈ C such that %(eiθ ) = ρ(θ) for eiθ ∈ Γ. The
saddle point equation (3.4) is rewritten as
Z
dw z+w
(A.1) P %(w) = W 0 (z)
Γ 2πw z − w
where
µ−1

0 1 µ
W (z) = −i Y z − −τ 1+ + .
z z − µ z − µ−1
Equation (A.1) is valid for z ∈ Γ, and is complemented by the normalization condition
Z
dw
(A.2) %(w) = 1.
Γ 2πiw
30
Recall that we have started with a Z2 -symmetric system, invariant under z 7→ z −1 for z ∈ T. We
will thus find an eigenvalue density with symmetric support, and in particular ∂Γ = e−iθ0 , eiθ0

in a one-cut phase. Then, depending on whether the gap opens at θ = π or θ = 0, Γ will be the
arc on the unit circle connecting −θ0 to θ0 or θ0 to −θ0 , respectively, with orientation always
taken counter-clockwise.
Recall from the definition (3.10) that
ω+ (eiθ ) − ω− (eiθ ) = 2%(eiθ ), eiθ ∈ Γ.
In turn, from the definition of Cauchy principal value and (A.1) we immediately get
(A.3) ω+ (eiθ ) + ω− (eiθ ) = −2iW 0 (eiθ ).
The normalization (A.2) and the definition (3.10) imply that ω(z) → 1 as |z| → ∞. We have
then reduced the problem of finding the eigenvalue density to the problem of determining the
discontinuity of ω(z) along Γ, from the knowledge of its regular part and the boundary condition
ω(z → ∞) = 1. It is standard procedure to reduce the problem (A.3) to a discontinuity equation
for a new, auxiliary function Ω(z) related to ω(z) via
q
(A.4) ω(z) = (eiθ0 − z) (e−iθ0 − z)Ω(z).
We take the square root with positive value, but any potential ambiguity in the intermediate
steps and definitions from now on, would drop out from the final answer.
Writing
q h√ p i p
iθ −iθ
(e 0 − z) (e 0 − z) = z · 2 cos θ − 2 cos θ0 = ∓eiθ/2 2 cos θ − 2 cos θ0
± ±
for z = eiθ ∈ Γ, we obtain from (A.3) the discontinuity equation for Ω(z):
iW 0 (eiθ )
(A.5) Ω+ (eiθ ) − Ω− (eiθ ) = 2e−iθ/2 √ .
2 cos θ − 2 cos θ0
For a multi-cut phase, with
Γ∼= {θ0,− ≤ θ ≤ θ0,+ } ∪ {θ1,− ≤ θ ≤ θ1,+ } ∪ · · · ∪ {θk,− ≤ θ ≤ θk,+ }
the procedure is the same, but with Ω(z) defined via
v
u k
uY
ω(z) = t eiθj,+ − z eiθj,− − z Ω(z).
j=0
Let us now introduce a closed contour CΓ which is a Jordan curve enclosing Γ but not z, and
oriented counter-clockwise. See Figure 11 for the contour CΓ in Phase Ia.
From the definitions (3.10) and (A.4) it follows that Ω(z) falls off (at least) as 1/z at infinity.
Then, for z lying in the exterior of CΓ , Cauchy’s theorem together with (A.5) implies
iW 0 (w)
I
dw
Ω(z) = p
CΓ 2πi (z − w) (eiθ0 − w) (e−iθ0 − w)
On the other hand, because W 0 (w) is meromorphic we can deform the contour CΓ into an infinitely
large circle, picking the poles of the integrand. We find
iW 0 (z) iW 0 (w)
I
dw
Ω(z) = − p − p
(eiθ0 − z) (e−iθ0 − z) C∞ 2πi (z − w) (eiθ0 − w) (e−iθ0 − w)
X iW (w)
(A.6) + Res p
w=zp (z − w) (eiθ0 − w) (e−iθ0 − w)
z ∈{0,µ,µ−1 }
p
31
Figure 11. Contour CΓ encircling the cut Γ.
where the first term is the residue at w = z, the second term is the remaining contour integral
along a circle at infinity, which in our case simply contributes Y , and the last term includes the
residues at the poles zp of W 0 (w).
In Phase Ia, explicit computation of each term leads to
" #
1 τ µ 1
q
0 iθ −iθ
ωIa (z) = −iW (z)+ (e 0 − z) (e 0 − z) Y 1 + −p − .
z 1 + µ2 − 2µ cos θ0 z − µ z − µ−1
The solution in the other phases is found likewise.
Appendix B. Filling fraction fluctuations

This appendix contains the analysis of the effect of taking into account fluctuations of the
filling fractions around the equilibrium configuration.
For a generic matrix model in a two-cut phase, the dependence of the filling fractions on the
parameters of the theory should be taken into account when computing physical observables [46].5
Below we briefly review how this effect comes about, and study it for the model at hand. We
start with Phase III, and look at Phase II projected onto the real line, as in Section 3.4.
B.1. Phase III. For the two-cut solution in Phase III, let NL be the number of eigenvalues in
the left arc around θ = π, 0 ≤ NL ≤ N , and NR = N − NL the number of eigenvalues in the right
arc around θ = 0. Let also ξ = NNL and 1 − ξ = NNR denote the corresponding filling fractions.
The values of yL = cos (θ∗ + δθ) and yR = cos (θ∗ − δθ) can be fixed, as functions of ξ and of
the other parameters, through the equations
s " #
2 yL (yR − y)(yL − y) τ µ(µ2 − 1)
Z
dy −Y + p = ξ,
π −1 1 − y2 (1 + µ2 − 2µyL )(1 + µ2 − 2µyR )(1 + µ2 − 2µy)
s " #
2 1 (y − yR )(y − yL ) τ µ(µ2 − 1)
Z
dy −Y + p = 1 − ξ,
π yR 1 − y2 (1 + µ2 − 2µyL )(1 + µ2 − 2µyR )(1 + µ2 − 2µy)
that come from the definition of ξ after changing variables y = cos θ. Then, the value of ξ is fixed
by the equilibrium condition

dSeff
(B.1) = 0.
dξ ξ=ξsp
5The original work [46] dealt with Hermitian matrix models, but the argument extends to the present setting.
32
For example, approximating close to the critical surface diving Phase III from Phase Ib, we find
√ !
∂ξsp 2 − 2yL 1 τ τ µ(µ + 1)
= − + p
∂yR yR =1 π 2 µ − 1 (µ − 1)3 1 + µ2 − 2µyL
where we have also substituted Y = Ycr,b . It has been shown in [46] that the quantum fluctuations
around the saddle point ξsp contribute to the free energy a term of the form − N12 log ϑ(N ξsp ),
where ϑ(z) is the Jacobi theta function. See Appendix B.2 below for more details and a very
short review of the derivation. This is a sub-leading contribution to the free energy but, due to
the dependence on N ξsp , each derivative generates a factor of N . Therefore, the ξsp -dependent
part becomes of the same order as the leading order term when differentiating the Wilson loop
vev, and must be taken into account. The relevant part of the derivative is
2
X d ∂y ∂ξsp 2

log ϑ(z)|z=N ξsp ,
dz ∂Y ∂y
y∈{yL ,yR }
which yields a non-trivial contribution to the derivative of the Wilson loop vev in Phase III.
However, when approaching the critical loci, ξ → 0 if Y → Ycr,a or ξ → 1 if Y → Ycr,b , and the
derivative of the theta function evaluated at an integer vanishes.
This shows that the effect of the filling fractions does not play any role in determining the
order of the phase transition, despite being non-trivial in the bulk of Phase III.
B.2. Phase II. We now discuss the same effect in Phase II. It is more convenient and akin to
the work [46] to do this in the alternative, Hermitian matrix model presentation of Section 3.4.
The argument can be succinctly summarized as follows.
Consider a two-cut solution with support suppρp. NL
II = ΓL ∪ ΓR , and denote by ξ = N and
NR
1 − ξ = N the corresponding filling fractions, as above. The saddle point value ξsp of ξ is fixed
by (B.1). Then, in the large N approximation, the partition function takes the form [46]
N 2
X −N 2 Fpert − N2 (ξ−ξsp )2 ∂ξ2 Seff | +O((ξ−ξsp )3 )
Z= e ξ=ξsp
NL =0
1
where Fpert is the perturbative free energy to all orders in theexpansion. This yields [46, 77]
N2
1 1 1 1
− 2 log Z = F + 2 F nlo − 2 log ϑ(N ξsp ) + 2
+ O(N −4 ),

log 2π ∂ ξ S eff

N N N 2N 2 ξ=ξsp
where F is the leading order or planar free energy, F nlo the next-to-leading order correction,
and (after an implicit resummation) we have recognized the Jacobi theta function ϑ(N ξsp ). The
modular parameter of the theta function is i2π/(∂ξ2 Seff (ξsp )), and the dependence on it is kept
implicit in the notation.
For the case at hand, however, the effective action is an even function, the two wells have
identical depth, and all the physical observables we consider preserve this property. We thus have
ξsp = 21 , independent of the parameters of the theory, and the effect we have just described will
remain sub-leading [77]. This would not be the case for other type of physical observables that
are not protected by the parity symmetry. See [46] for discussion and examples.
References
[1] P. J. Forrester, Log-gases and random matrices, vol. 34 of London Mathematical Society
Monographs Series. Princeton University Press, Princeton, NJ, 2010,
10.1515/9781400835416.
[2] J. Baik, P. Deift and T. Suidan, Combinatorics and random matrix theory, vol. 172 of
Graduate Studies in Mathematics. American Mathematical Society, Providence, RI, 2016.
33
[3] G. Akemann, J. Baik and P. Di Francesco, The Oxford Handbook of Random Matrix
Theory, Oxford Handbooks in Mathematics. Oxford University Press, 2011,
10.1093/oxfordhb/9780198744191.001.0001.
[4] D. Gross and E. Witten, Possible Third Order Phase Transition in the Large N Lattice
Gauge Theory, Phys. Rev. D 21 (1980) 446.
[5] S. R. Wadia, A Study of U(N) Lattice Gauge Theory in 2-dimensions, [1212.2906].
[6] S. R. Wadia, N = Infinity Phase Transition in a Class of Exactly Soluble Model Lattice
Gauge Theories, Phys. Lett. B 93 (1980) 403.
[7] J. A. Minahan, Matrix models with boundary terms and the generalized Painleve II
equation, Phys. Lett. B 268 (1991) 29.
[8] J. A. Minahan, Flows and solitary waves in unitary matrix models with logarithmic
potentials, Nucl. Phys. B 378 (1992) 501 [hep-th/9111012].
[9] B. Sundborg, The Hagedorn transition, deconfinement and N=4 SYM theory, Nucl. Phys.
B 573 (2000) 349 [hep-th/9908001].
[10] O. Aharony, J. Marsano, S. Minwalla, K. Papadodimas and M. Van Raamsdonk, The
Hagedorn - deconfinement phase transition in weakly coupled large N gauge theories, Adv.
Theor. Math. Phys. 8 (2004) 603 [hep-th/0310285].
[11] L. Alvarez-Gaume, C. Gomez, H. Liu and S. Wadia, Finite temperature effective action,
AdS(5) black holes, and 1/N expansion, Phys. Rev. D 71 (2005) 124023 [hep-th/0502227].
[12] L. Alvarez-Gaume, P. Basu, M. Marino and S. R. Wadia, Blackhole/String Transition for
the Small Schwarzschild Blackhole of AdS(5)x S**5 and Critical Unitary Matrix Models,
Eur. Phys. J. C 48 (2006) 647 [hep-th/0605041].
[13] T. Azuma, P. Basu and S. R. Wadia, Monte Carlo Studies of the GWW Phase Transition
in Large-N Gauge Theories, Phys. Lett. B 659 (2008) 676 [0710.5873].
[14] M. Hanada and J. Maltz, A proposal of the gauge theory description of the small
Schwarzschild black hole in AdS5 ×S5 , JHEP 02 (2017) 012 [1608.03276].
[15] D. Berenstein, Submatrix deconfinement and small black holes in AdS, JHEP 09 (2018) 054
[1806.05729].
[16] M. Hanada, G. Ishiki and H. Watanabe, Partial Deconfinement, JHEP 03 (2019) 145
[1812.05494]. [Erratum: JHEP 10, 029 (2019)].
[17] M. Hanada, A. Jevicki, C. Peng and N. Wintergerst, Anatomy of Deconfinement, JHEP 12
(2019) 167 [1909.09118].
[18] M. Hanada, G. Ishiki and H. Watanabe, Partial deconfinement in gauge theories, PoS
LATTICE2019 (2019) 055 [1911.11465].
[19] C. Lazaroiu, Holomorphic matrix models, JHEP 05 (2003) 044 [hep-th/0303008].
[20] P. Rossi, M. Campostrini and E. Vicari, The Large N expansion of unitary matrix models,
Phys. Rept. 302 (1998) 143 [hep-lat/9609003].
[21] M. Billò, M. Caselle, A. D’Adda and S. Panzeri, Finite temperature lattice QCD in the
large N limit, Int. J. Mod. Phys. A 12 (1997) 1783 [hep-th/9610144].
[22] J. G. Russo and M. Tierz, Multiple phases in a generalized Gross-Witten-Wadia matrix
model, JHEP 09 (2020) 081 [2007.08515].
[23] J. G. Russo, Phases of unitary matrix models and lattice QCD2, Phys. Rev. D 102 (2020)
105019 [2010.02950].
[24] F. Haake, M. Kus, H.-J. Sommers, H. Schomerus and K. Zyczkowski, Secular determinants
of random unitary matrices, J. Phys. A 29 (1996) 3641 [chao-dyn/9603006].
[25] Y. V. Fyodorov, G. A. Hiary and J. P. Keating, Freezing Transition, Characteristic
Polynomials of Random Matrices, and the Riemann Zeta Function, Phys. Rev. Lett. 108
(2012) 170601 [1202.4713].
[26] Y. V. Fyodorov and J. P. Keating, Freezing transitions and extreme values: random matrix
theory, and disordered landscapes, Phil. Trans. R. Soc. A 372 (2014) 20120503 [1211.6063].
34
[27] J. Baik, Random vicious walks and random matrices, Comm. Pure Appl. Math. 53 (2000)
1385 [math/0001022].
[28] J. Hallin and D. Persson, Thermal phase transition in weakly interacting, large N(C) QCD,
Phys. Lett. B 429 (1998) 232 [hep-ph/9803234].
[29] P. J. Forrester and N. S. Witte, Application of the τ -function theory of Painlevé equations
to random matrices: PVI, the JUE, CyUE, cJUE and scaled limits, Nagoya Math. J. 174
(2004) 29 [math-ph/0204008].
[30] S. de Haro and M. Tierz, Brownian motion, Chern-Simons theory, and 2-D Yang-Mills,
Phys. Lett. B 601 (2004) 201 [hep-th/0406093].
[31] P. J. Forrester, S. N. Majumdar and G. Schehr, Non-intersecting Brownian walkers and
Yang-Mills theory on the sphere, Nucl. Phys. B 844 (2011) 500 [1009.2362]. [Erratum:
Nucl.Phys.B 857, 424–427 (2012)].
[32] M. Romo and M. Tierz, Unitary Chern-Simons matrix model and the Villain lattice action,
Phys. Rev. D 86 (2012) 045027 [1103.2421].
[33] A. Gorsky, A. Milekhin and S. Nechaev, Two faces of Douglas-Kazakov transition: from
Yang-Mills theory to random walks and beyond, Nucl. Phys. B 950 (2020) 114849
[1604.06381].
[34] P. J. Forrester and N. S. Witte, Application of the τ -function theory of Painlevé equations
to random matrices: PV, PIII, the LUE, JUE, and CUE, Commun. Pure Appl. Math. 55
(2002) 679 [math-ph/0201051].
[35] P. J. Forrester and N. S. Witte, Discrete Painlevé equations, orthogonal polynomials on the
unit circle, and N-recurrences for averages over U(N)—PIII0 and PV τ -functions, Int.
Math. Res. Not. 2004 (2004) 159 [math-ph/0308036].
[36] M. Adler and P. Van Moerbeke, Recursion relations for unitary integrals, combinatorics
and the Toeplitz lattice, Commun. Math. Phys. 237 (2003) 397 [math-ph/0201063].
[37] M. Adler and P. van Moerbeke, Integrals over classical groups, random permutations, Toda
and Toeplitz lattices, Commun. Pure Appl. Math. 54 (2001) 153 [math/9912143].
[38] L. Faybusovich and M. Gekhtman, On Schur flows, J. Phys. A 32 (1999) 4671.
[39] A. Mukaihira and Y. Nakamura, Schur flow for orthogonal polynomials on the unit circle
and its integrable discretization, J. Comput. Appl. Math. 139 (2002) 75.
[40] I. Nenciu, CMV matrices in random matrix theory and integrable systems: a survey, J.
Phys. A 39 (2006) 8811 [math-ph/0510045].
[41] S. R. Coleman, There are no Goldstone bosons in two-dimensions, Commun. Math. Phys.
31 (1973) 259.
[42] L. Santilli and M. Tierz, Exact results and Schur expansions in quiver
Chern-Simons-matter theories, JHEP 10 (2020) 022 [2008.00465].
[43] I. Bars, U(N ) Integral for Generating Functional in Lattice Gauge Theory, J. Math. Phys.
21 (1980) 2678.
[44] G. Mandal, Phase Structure of Unitary Matrix Models, Mod. Phys. Lett. A 5 (1990) 1147.
[45] L. Santilli and M. Tierz, Exact equivalences and phase discrepancies between random
matrix ensembles, J. Stat. Mech. 2008 (2020) 083107 [2003.10475].
[46] G. Bonnet, F. David and B. Eynard, Breakdown of universality in multicut matrix models,
J. Phys. A 33 (2000) 6739 [cond-mat/0003324].
[47] L. Santilli and M. Tierz, Phase transition in complex-time Loschmidt echo of short and long
range spin chain, J. Stat. Mech. 2006 (2020) 063102 [1902.06649].
[48] H. Widom, Toeplitz determinants with singular generating functions, Amer. J. Math. 95
(1973) 333.
[49] J. G. Russo, Deformed Cauchy random matrix ensembles and large N phase transitions,
JHEP 11 (2020) 014 [2006.00672].
[50] P. Di Francesco, P. H. Ginsparg and J. Zinn-Justin, 2-D Gravity and random matrices,
Phys. Rept. 254 (1995) 1 [hep-th/9306153].
35
[51] M. Hanada and B. Robinson, Partial-Symmetry-Breaking Phase Transitions, Phys. Rev. D

102 (2020) 096013 [1911.06223].
[52] K. Demeterfi and C.-I. Tan, Periodic regularization, multiband structure and orthogonal
polynomials, Phys. Rev. D 43 (1991) 2622.
[53] J. Baik, P. Deift and K. Johansson, On the distribution of the length of the longest
increasing subsequence of random permutations, J. Amer. Math. Soc. 12 (1999) 1119
[math/9810105].
[54] V. Periwal and D. Shevitz, Unitary matrix models as exactly solvable string theories, Phys.
Rev. Lett. 64 (1990) 1326.
[55] S. Elitzur, Impossibility of Spontaneously Breaking Local Symmetries, Phys. Rev. D 12
(1975) 3978.
[56] P. M. Hernandez, Lattice field theory fundamentals, in Modern perspectives in lattice QCD:
Quantum field theory and high performance computing. Proceedings, International School,
93rd Session, Les Houches, France, August 3-28, 2009, (Oxford, UK), pp. 1–91, Oxford
University Press, 8, 2009.
[57] M. Marino, Nonperturbative effects and nonperturbative definitions in matrix models and
topological strings, JHEP 12 (2008) 114 [0805.3033].
[58] H. Neuberger, Nonperturbative Contributions in Models With a Nonanalytic Behavior at
Infinite N , Nucl. Phys. B 179 (1981) 253.
[59] R. Dijkgraaf and C. Vafa, Matrix models, topological strings, and supersymmetric gauge
theories, Nucl. Phys. B 644 (2002) 3 [hep-th/0206255].
[60] L. Chekhov, A. Marshakov, A. Mironov and D. Vasiliev, Complex geometry of matrix
models, Proc. Steklov Inst. Math. 251 (2005) 254 [hep-th/0506075].
[61] G. Álvarez, L. M. Alonso and E. Medina, Determination of S-curves with applications to
the theory of non-Hermitian orthogonal polynomials, J. Stat. Mech. 2013 (2013) P06006
[1305.3028].
[62] A. Bilal and S. Metzger, Special geometry of local Calabi-Yau manifolds and superpotentials
from holomorphic matrix models, JHEP 08 (2005) 097 [hep-th/0503173].
[63] G. Álvarez, L. Martı́nez-Alonso and E. Medina, Complex saddles in the
Gross-Witten-Wadia matrix model, Phys. Rev. D 94 (2016) 105010 [1610.09948].
[64] C. Copetti, A. Grassi, Z. Komargodski and L. Tizzano, Delayed Deconfinement and the
Hawking-Page Transition, [2008.04950].
[65] K. Kodaira, On compact analytic surfaces II, Ann. Math. 77 (1963) 563.
[66] N. Seiberg and E. Witten, Monopoles, duality and chiral symmetry breaking in N=2
supersymmetric QCD, Nucl. Phys. B 431 (1994) 484 [hep-th/9408099].
[67] D. Kaledin, Symplectic singularities from the Poisson point of view, J. Reine Angew. Math.
2006 (2006) 135 [math/0310186].
[68] P. C. Argyres and M. Martone, Towards a classification of rank r N = 2 SCFTs. Part II.
Special Kahler stratification of the Coulomb branch, JHEP 12 (2020) 022 [2007.00012].
[69] A. Bourget, S. Cabrera, J. F. Grimminger, A. Hanany, M. Sperling, A. Zajac and Z. Zhong,
The Higgs mechanism – Hasse diagrams for symplectic singularities, JHEP 01 (2020) 157
[1908.04245].
[70] J. F. Grimminger and A. Hanany, Hasse diagrams for 3d N = 4 quiver gauge theories –
Inversion and the full moduli space, JHEP 09 (2020) 159 [2004.01675].
[71] L. Pastur and M. Shcherbina, Eigenvalue distribution of large random matrices, vol. 171 of
Mathematical Surveys and Monographs. American Mathematical Society, Providence, RI,
2011, 10.1090/surv/171.
[72] R. Killip and I. Nenciu, CMV: The unitary analogue of Jacobi matrices, Commun. Pure
Appl. Math. 60 (2007) 1148 [math/0508113].
[73] P. Castorina, D. Lanteri and M. Ruggieri, Fluctuations and thermodynamic geometry of the
chiral phase transition, Phys. Rev. D 102 (2021) 116022 [2010.03310].
36
[74] G. Ruppeiner, Thermodynamics: A Riemannian geometric model, Phys. Rev. A 20 (1979)

1608.
[75] G. Ruppeiner, Riemannian geometry in thermodynamic fluctuation theory, Rev. Mod. Phys.
67 (1995) 605.
[76] S. Jain, S. Minwalla, T. Sharma, T. Takimi, S. R. Wadia and S. Yokoyama, Phases of large
N vector Chern-Simons theories on S 2 × S 1 , JHEP 09 (2013) 009 [1301.6169].
[77] T. Claeys, T. Grava and K. D. T.-R. McLaughlin, Asymptotics for the partition function in
two-cut random matrix models, Commun. Math. Phys. 339 (2015) 513 [1410.7001].
∗
Grupo de Fı́sica Matemática, Departamento de Matemática, Faculdade de Ciências, Universi-
dade de Lisboa, Campo Grande, Edifı́cio C6, 1749-016 Lisboa, Portugal.
Email address: lsantilli@fc.ul.pt
†
Departamento de Matemática, ISCTE - Instituto Universitário de Lisboa, Avenida das Forças
Armadas, 1649-026 Lisboa, Portugal.
Email address: mtpaz@iscte-iul.pt
‡
Grupo de Fı́sica Matemática, Departamento de Matemática, Faculdade de Ciências, Universi-
dade de Lisboa, Campo Grande, Edifı́cio C6, 1749-016 Lisboa, Portugal.
Email address: tierz@fc.ul.pt

Multiple Phases and Meromorphic Deformations of Unitary Matrix Models

Uploaded by

Copyright:

Available Formats

Multiple Phases and Meromorphic Deformations of Unitary Matrix Models

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Multiple Phases and Meromorphic Deformations of Unitary Matrix Models

Uploaded by

Copyright:

Available Formats

MULTIPLE PHASES AND MEROMORPHIC

DEFORMATIONS OF UNITARY MATRIX MODELS

LEONARDO SANTILLI∗ AND MIGUEL TIERZ†,‡

investigated as well. Furthermore, we study in detail the meromorphic deformation of a general

The second line is written in terms of the eigenvalues zj ∈ T of U ∈ U (N ).

Figure 1. Veff (eiθ ) at µ = 3 and τ = 2. Left: λ = 12 . Right: λ = 34 .

Figure 2. Veff (eiθ ) at µ = 3 and τ = −2. Left: λ = − 21 . Right: λ = − 43 .

that follows from the compactness of the integration domain.

Figure 3. ρ0 (θ) at the transition point. Left: µ = 3, τ = − 12 and λ = λcr,a .

Figure 4. ρ0 (θ) at the transition point. µ = 2, τ = −1 and λ = − 25 24 . The

Figure 5. The two-cut support Γ in Phase II.

Figure 6. The two-cut support Γ in Phase III.

μ=3 slice of parameter space

Veneziano parameters τ1 , . . . , τn associated to different µ1 , . . . , µn that approach the unit circle

The value of A is fixed by normalization:

and by an additional self-consistency condition on ω(z),

which reproduces the criticality condition for B → 0.

4. Wilson loops and instantons

Figure 9. Instantons in Phase III. A single eigenvalue (blue dot) is moved on

5. Meromorphic deformation of unitary matrix models

with integration contour

where we assume d− ≥ 2 (otherwise we get back the known setting).

The discriminant of (5.14) is

with P (z) a polynomial of degree 4. The spectral curve

CIreg CJreg ··· CKreg

C1reg C2reg C1reg

The diagram of the meromorphic model of Section 5.3 is schematically

M reg |GWW = λ ∈ R \ {0} , λ2 6= 4

C1 |GWW = {λ = 2} • • C10 |GWW = {λ = −2}

Appendix A. Technical details of the solution

Table 1. Analytic evaluation of ZU+1,K

Figure 11. Contour CΓ encircling the cut Γ.

Appendix B. Filling fraction fluctuations

[51] M. Hanada and B. Robinson, Partial-Symmetry-Breaking Phase Transitions, Phys. Rev. D

[74] G. Ruppeiner, Thermodynamics: A Riemannian geometric model, Phys. Rev. A 20 (1979)

You might also like