Theoretical Exploration of Competing Phases of Lattice Bose Gases in a Cavity

Renyuan Liao, Huang-Jie Chen, Dong-Chen Zheng, and Zhi-Gao Huang

Fujian Provincial Key Laboratory for Quantum Manipulation and New Energy Materials,
College of Physics and Energy, Fujian Normal University, Fuzhou 350117, China and
Fujian Provincial Collaborative Innovation Center for Optoelectronic
Semiconductors and Efficient Devices, Xiamen, 361005, China
(Dated: January 17, 2020)
arXiv:1701.03676v3 [cond-mat.quant-gas] 22 Jan 2018

We consider bosonic atoms loaded into optical lattices with cavity-mediated infinite-range inter-
actions. Competing short- and global-range interactions cultivates a rich phase diagram. With a
systematic field-theoretical perspective, we present an analytical construction of global ground-state
phase diagram. We find that the infinite-range interaction enhances the fluctuation of the number
density. In the strong coupling regime, we find four branches of elementary excitations with two
being “partilce-like” and two being “hole-like”, and that the excitation gap becomes soft at the
phase boundary between compressible phases and incompressible phases. We derive an effective
theory describing compressible superfluid and supersolid states. To complement this perturbative
study, we construct a self-consistent mean-field theory and find numerical results consistent with
our theoretical analysis. We map out the phase diagram and find that a charge density wave may
undergo a structure phase transition to a different charge density wave before it finally enters into
the supersolid phase driven by increasing the hopping amplitude.

PACS numbers: 67.85.Hj, 05.30.JP, 03.75.Hh, 03.75.Lm

I. INTRODUCTION folds: On the one hand, in the atomic limit, we obtain

analytically the ground state energy density. With this,
we construct the phase diagram consistent with previ-
Ultracold gases in optical lattices are one of the most
ous numerical approach [16, 18]. Based on this, we carry
intriguing systems in which the power of atomic and laser
out a field-theoretical analysis [19–25], by which physics
physics can be exploited to explore generic phenomena
close to phase boundaries between compressible phases
of solid-sate physics [1]. They have been proven to be
and non-compressible phases can be qualitatively exam-
impressively successful in simulating strongly correlated
ined. On the other hand, we construct a a self-consistent
models like the Bose-Hubbard models, which features a
local mean-field theory which is numerically cheap. With
quantum phase transition from a superfluid to a Mott
this, we find interesting structural phase transitions be-
insulating phase [2, 3]. Recent years have seen major ad-
tween different charge density wave driven by hopping
vances in the exploration of many-body systems in which
amplitudes before it enters the supersolid phase.
matter is strongly coupled to light [4]. In particular,
The paper is structured as follows: In Sec. II the model
recent experimental realization of competing short- and
is introduced. We present a functional integral formula-
infinite-range interactions [5] for bosonic atoms in opti-
tion of this problem. In the atomic limit, we construct
cal lattices has opened a new avenue for exploring new
the ground state phase diagram analytically. Then we
phases of matter. This is achieved by trapping quantum
proceed to study physics close to compressible and in-
gases in an optical lattice inside a high finesse optical
compressible boundary by carrying out perturbative ex-
cavity. The infinity-range interactions is mediated by a
pansion on the hopping parameter which is assumed to be
vacuum mode of the cavity and can be independently
small. In Sec. III, we formulate a self-consistent mean-
controlled by tuning the cavity resonance [6–8].
field theory, by which properties of compressible phases
Essential physics of this system falls into the category are investigated. Finally, in Sec. IV, the conclusions are
of the extended Hubbard model [9, 10]. In the pres- drawn.
ence of the global-range interactions, novel phases such
as charge density wave (CDW) and supersolid (SS) phase
emerge, in addition to the conventional superfluid (SF) II. MODEL AND FIELD-THEORETICAL
phase and Mott insulating (MI) phase. Understanding TREATMENT
the phase diagram and related phase transition has been
the focus of recent studies [11–18]. These studies mainly We consider the system described by the following
concentrated on global phase diagrams and drew heavily canonical Hamiltonian realized very recently [5]
on sophisticated numerical methods, making the study of X †  U X
the properties of these phases and analytical methods to Ĥ = − tij b̂i b̂j + h.c. + n̂i (n̂i − 1)
understand the underlying physics a useful complement. <ij>
2 i
In this study, we shall explore the relevant physics " #2
transparently by employing both analytical approach and K X X X
− n̂i − n̂i − µn̂i . (1)
numerical approach. The findings of our study are two M i∈e i∈o i
 2

Here tij is the hopping matrix element between site i eigenvalue corresponding to Ĥ0 for per “supercell” (with
and site j, b̂†i and b̂j are the bosonic operators satisfying one oven and one odd sites) after taking account of the
commutation relation [b̂†i , b̂j ] = δij , n̂i = b̂†i b̂i is the self-consistency conditions for θ is given by
associated number operator which counts the particle  2
number on site i, and µ is the chemical potential. U 2µ
E(ne , no ) = (ne + no ) − (1 + )
The subscript e (o) refers to even (odd) lattice sites 4 U
i = (ix , iy ) of square lattice potential defined as ix + iy ∈
 
U 2K 2µ 2
even (odd), and <ij> denotes pair of site i and j. The + (1 − )(ne − no )2 − (1 + ) . (5)
4 U U
on-site repulsive interaction is characterized by U , while
the infinite-range attractive interaction is denoted by K, Here ne (no ) represents the occupation number of one
which favors overall particle number imbalance between even (odd) site. The ground state is achieved by mini-
even and odd sites. Interplay of three energy scales is mizing the eigenvalue E(no , ne ) with respect to quantum
expected to give rise to a multitude of ground-state numbers ne and no . Since the system enjoys an Ising-
manifolds. type Z2 symmetry corresponding to exchange of even and
odd sites, we may choose ne ≥ no from now on. To fa-
Within the framework of Euclidean functional inte- cilitate the analysis, we define 1 + 2µ/U = n + x, with
gral, theR partition function of the system may be cast n = int [1 + 2µ/U ] being the integer closest to 1 + 2µ/U ,
as Z = D[b∗i , bi ]e−S with the action given by [26, 27] and x ∈ (−1/2, 1/2). Firstly, let’s consider the case
Rβ P
S = 0 dτ i [b∗i ∂τ bi + H(b∗i , bi )], here β = 1/kB T is the K/U ∈ (0, 1/2), the system is in a Mott insulating (MI)
inverse temperature. To decouple the off-site terms in the phase with ne = no = n/2 if n is even. If n is odd, then
action, we introduce a real field θ(τ ) and complex bosonic the system is in the MI phase when K/U < |x|, and in
fields Ψi (τ ) by performing Hubbard-Stratonovich trans- a partially polarized charge density wave (CDW) phase
formations, resulting in an equivalent representation of with ne = no + 1 = (n + 1)/2, vice versa. Secondly, if
the partition function K/U ∈ (1/2, 1), the systemhenters into i a fully polarized
1+2µ/U
CDW phase with ne = int 2(1−K/U) and no = 0. Fi-
Z Z
∗
Z = D[Ψi , Ψi ] D[θ, b∗i , bi ]e−SR , (2)
nally, if K/U > 1, the ground energy is unstable toward
collapse since it corresponds to an infinite filling. The
where the resultant action is given by SR =
Rβ P above discussions for the ground-state phase diagram are
0
dτ ij Ψ∗i t−1
ij Ψj + S0 + SI with summarized in Fig. 1.
Z β The low temperature properties of the system may
be captured by only considering the particle and hole
X
Kθ2 − 2Kθ(−1)ix +iy b∗i bi

S0 = dτ
0 i excitations on one supercell [28], since tunneling be-
X U ∗ ∗
 tween sites are completely neglected and hence the sys-
+ b∗i (∂τ − µ)bi + bi bi bi bi , (3) tem consists of P isolate−βEpair sites. For Pbrevity, let us
2 −βEsh
i define Cp = s=e,o e sp
and C h = s=e,o e ,
Z β X where Esp and Esh is the particle and hole excitation
SI = − dτ (Ψ∗i bi + b∗i Ψi ) . (4) on s = (e/o) site, respectively. To put it explicitly,
0 i Eep = E(ne + 1, no ) − E(ne , no ) and Eeh = E(ne −
Before embarking on detailed analysis with field- 1, no ) − E(ne , no ), and similarly for Eop and Eoh . The
theoretical machinery, we make some comments. The partition function on one supercell is therefore approx-
free part S0 is readilyP solvable since the corresponding imated as z0 = e−βE(ne ,no ) (1 + Cp + Ch ). The varia-
Hamiltonian Ĥ0 (θ) = i Ĥ0i (θ) can be diagonalized in tion of density fluctuations on a supercell is given by
the occupation number representation. With the inter- δn = (Cp − Ch )/(1 + Cp + Ch ). Equivalently interest-
2
acting part SI present, the physics could not be solved in ing is the square density fluctuations δn2 ≡ n2 − hni ,
a close form, however, we can gain physical insights by which is related to the isothermal compressibility via
2
seeking perturbative expansion on top of S0 in terms of thermodynamic relation δn2 = hni κT v0 /β with v0 be-
fields Ψi , which serves as the superfluid order parameter. ing the volume of one supercell. We find that δn2 =
Now we subject the action SR to a saddle point anal- (Cp + Ch + 4Cp Ch )/(1 + Cp + Ch )2 . The tempera-
ysis. Extremum of P variation of the action with respect ture dependence of δn and κT is shown in Fig. 2. At
to θ(τ ) yields θ = i (−1)ix +iy < n̂i > /M . The phys- zero temperature, the particle number fluctuation and
ical meaning is clear: M θ counts the particle number thermal compressibility is frozen out, indicating its non-
difference between even sites and odd sites, and θ could compressible nature. It clearly indicates that the larger
be regarded as an order parameter representing charge K/U is, the larger thermal fluctuation it induces. We at-
degrees of freedom. tribute this fluctuation-enhancing behavior to the effects
Let’s consider the atomic limit where the hopping am- of the infinite-range interactions.
plitude between sites is negligible (tij /U = 0), and the We proceed to take into account the effects of a finite
resultant action reduces to a free one: SR = S0 . The hopping amplitude. We can evaluate the partition func-
 3

0.2 (a) 0.4 (b)

K/U =0
K/U = 0.1
0.15 K/U = 0.2 0.3
K/U = 0.3

v0 κT
δn
0.1 0.2

0.05 0.1

0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15
kB T /U kB T /U

FIG. 2. (color online). Temperature dependence of (a) the

variation of particle number δn on one supercell (with one
even site and one odd site) and (b) isothermal compressibil-
FIG. 1. (color online). Global ground-state phase dia- ity κT for M I(1,1) phase at different infinite-range interaction
gram spanned by µ/U and K/U at atomic limit (tij /U = strength K/U = 0, 0.1, 0.2 and 0.3. Here µ/U = 0.60, and v0
0). The phase diagram can be loosely divided into three is the volume of one supercell.
regimes depending on the strength of infinite-range interac-
tion: (1)K/U ∈ (0, 0.5), the system is either in a Mott in-
sulating (MI) phase with ne = no or in a partially polarized Pd
1/[2t j=1 cos (kj λ/2)], with d being the dimension of
charge density wave (CDW) phase with ne − no = 1, where the system and λ being the wavelength of the laser field
we always assume ne ≥ no as the system enjoys an Ising-type
forming the optical lattices.
Z2 symmetry; (2)K/U ∈ (0.5, 1), the system is in a fully po-
larized CDW phase with no = 0; (3)K/U > 1, the system is The excitation spectrum of the system corresponds
unstable toward collpase. to the poles of the Green’s function. It can be readily
found by seeking solutions ω for the secular equations
det G −1 (k, ω) = 0. It features four branches of excitation
tion by performing Taylor expansion in the exponent spectrum ωi (i=1..4), as shown in Fig. 3 at k = (~0, 0)
R
* + in terms of the tuning parameter zt/U with z = 4 being
Z P X Sl
Z
∗
= D[Ψi , Ψi ]e − dτ ij Ψ∗
i t−1
ij Ψj I
, (6) the coordination number of square lattices. In absence
Z0 l! of hopping (zt = 0), the incompressible M I(1,1) phase
l=0 0
possesses only one type of particle excitations and one
( e−S0 O)/( e−S0 ).
R R
where hOi0 = To the type of hole excitations, while incompressible CDW(2,1)
quadratic order in the fields Ψi , by transforming phase carrying charge order possesses two types of par-
to momentum-frequency representation, we obtain ticle excitations and two types of hole excitations. At a
Z/Z0 = D[Ψ∗ (k), Ψ(k)]e−Sg with the gaussian action
R
finite hopping, these two phases both accommodate two
given by branches of particle excitations and two branches of hole
M X excitations. The minimal energy difference between one
Sg = Ψ† (k)G −1 (k)Ψ(k), (7) particle excitation and one hole excitation corresponds
2
k=(k,iwn ) to the energy gap for density fluctuations. This excita-
tion gap becomes soft at the phase boundary where phase
where we have defined Ψ(k) = (Ψe (k), Ψo (k))T , and
transition from a non-compressible phase to a compress-
used a shorthand notation k = (k, iwn ), with wn being
ible phase occurs.
the bosonic Matsubara frequencies. The inverse Green’s
The phase boundary separating the superfluid phase
function assumes the form of 2 × 2 matrix
and the non-compressible phase occurs [21, 22, 29] at
det G −1 (0, 0) = 0, which yields
!
ne +1 ne
−1 Eep −iwn + Eeh +iwn −t̃−1 (q)
−G = no +1 no (8)
.
−t̃−1 (q) Eop −iwn + Eoh +iwn

ne + 1 ne

no + 1 no

1
+ + = . (9)
Eep Eeh Eop Eoh (zt)2
In the above, t̃−1 (q) is the Fourier transform of t−1
ij .
For convenience we shall consider the nearest neigh- We show the phase boundary in Fig. 4. Evidently the
bor hopping only with amplitude t, then t̃−1 (q) = regime of M I phase diminishes as K/U increases. In
 4

0.8 0.5
(a) (b) 0.15 (a) 0.15 (b)
0.6 ω1 Compressible Phase Compressible Phase
(SF or SS) (SF or SS)
ω2
0.3
ω1 ω3
ω2 ω4
ω3 K/U = 0.10
0.1 0.1 K/U = 0.25
ω4 0.1
0.2

zt/U
zt/U
ω
ω

M I(1,1) CDW(2,1)
−0.1
0.05 0.05
−0.2

−0.3
M I(1,1)
M I(2,2) M I(1,1) M I(2,2)
−0.6 −0.5 0 0
0 0.04 0.08 0.12 0 0.02 0.04 0.06 0 1 2 0 1 2
zt/U zt/U µ/U µ/U

FIG. 3. (color online). Four branches of excitation spectrum FIG. 4. (color online). The phase boundary separating com-
ωi (i=1..4) at momentum k = (~0, 0) as a function of tunneling pressible (SF or SS) and non-compressible phases (MI or
parameter zt/U for (a) the M I(1,1) phase with µ/U = 0.50 CDW) for two typical infinite-range interaction strengths: (a)
and K/U = 0.25; (b) the CDW(2,1) phase with µ/U = 1.0 and K/U = 0.10 and (b) K/U = 0.25. Increasing K/U leads to
K/U = 0.40. The upper two branches are particle excitations the broadening of the region of CDW phases and the shrink-
and the lower two branches are hole excitations. For both in- ing of the region of M I phases. Here a SF phase stands for a
sulating phases, there exists an energy gap for a particle-hole superfluid phase which has off-diagonal long-range order, and
excitation, manifesting their non-compressible nature. Here a SS phase stands for a supersolid phase which has both diag-
z = 4 is the coordination number for square lattices. onal and off-diagonal long-range orders. Here the unlabeled
small lobes are CDW phases generated by the infinite-range
interaction.
stark contrast, the regime of CDW phase gets broadened
as K/U increases. This suggests that the infinite-range 2)/Es2p − Bs (ns − 1)/Es2h , and
interaction favors the formation of density modulation in
the form of a checkerboard pattern with alternating site ueo Eep + Eop Eeh + Eoh
occupation. = Ae Ao + Be Bo
K Eep + Eop + K Eeh + Eoh + K
To explore the physics of compressible superfluid X Esp + E−sh
and supersolid phases, we proceed even further by − As B−s . (11)
evaluating the action to the quartic order in or- s=e,o
Esp + E−sh − K
der
P parameter Ψi , S = S0 + Sg + S4 with S4 =
P ∗ ∗ Here Ee2p = E(ne + 2, no ) − E(ne , no ) is the “double
k+l=m+n uss Ψs (k)Ψs′ (l)Ψs (m)Ψs (n), where we
′ ′
ss′

may evaluate the coefficients uss′ at zero momentum particles” excitation energy at even sites and Ee2h =
and zero frequency[23]. By performing derivative expan- E(ne − 2, no ) − E(ne , no ) is the “double holes” excita-
sion, we keep only the most relevant terms in a long- tion energy at even sites, and similar expressions for Eo2p
wavelength approximation, and obtain an effective action and Eo2h . The universality class and associated quantum
of Ginzburg-Landau-Wilson type [30, 31] criticality is intimately related to the relevant parameters
given above.
S − S0 = At zero temperature, we assume that field configura-
Z
d2 r X tions for Ψe and Ψo are spatially and temporally homoge-
rs |Ψs |2 + as Ψ∗s ∂τ Ψs + bs |∂τ Ψs |2 + uss |Ψs |4 nous. The grand potential Ω = − ln Z/β of the system


dτ
2 s
reduces to a simple form as follows
d2 r 2 2
 
λ ∇
Z
+ dτ reo Ψ∗e (1 − )Ψo + c.c + ueo |Ψe |2 |Ψo |2 .
X
2 16 Ω = Ω0 + rs |Ψs |2 + reo (Ψ∗e Ψo + c.c)
s=e,o
(10) X
+ uss |Ψs |4 + ueo |Ψo |2 |Ψe |2 . (12)
To present the coefficients above in a succinct fash- s=e,o
2
ion, we define As = (ns + 1)/Esp and Bs = ns /Esh .
Then the relevant coefficients are given as follows: rs = Quite generally, the realization of the phase is deter-
−(As Esp +Bs Esh ), as = As −Bs , bs = As /Esp +Bs /Esh , mined by seeking the global minimum of Ω. The saddle
reo = 1/(zt), uss = (As + Bs )(As Esp + Bs Esh ) − As (ns + point condition ∂Ω/∂Ψs = 0 yields 2rs Ψs + 2reo Ψ−s +
 5

1.5 2
(a)
1 (a) ψe ψe
ψ

ψo ψo

ψ
1
0.5
0 0
0 0.1 0.2 0.3 0 0.5 zt/U 1 1.5

3 6
(b) ne (b)
4
no ne
n

n
2 no
1 0
0 0.1 zt/U 0.2 0.3 0 0.5 zt/U 1 1.5

FIG. 5. (color online). Self-consistent mean-field calculation FIG. 6. (color online). Self-consistent mean-field calculation
at zero temperature for K/U = 0.4 and µ/U = 1.0: (a) The at zero temperature for K/U = 0.7 and µ/U = 1.0: (a) The
magnitude of the order parameters |ψe | and |ψo | in the super- magnitude of the order parameters |ψe | and |ψo | in the super-
solid phase as a function of zt/U ; (b) The density at even site solid phase as a function of zt/U ; (b) The density at even site
ne and odd site no as a function of zt/U . ne and odd site no as a function of zt/U .

4uss Ψ3s + 2ueo Ψs Ψ2−s . Clearly, if Ψs = 0, then from 0.3

the above equation we immediately obtain Ψ−s = 0, SF
namely Ψe and Ψo vanishes identically at the transition SS
2
point. The phase boundary is determined by ro re = reo , 0.2 CDW
which reproduces Eq. (9). Typically close to the phase zt/U
boundary, the order parameterpfield satisfies a simple MI
scaling Ψs /Ψ−s = reo /rs = r−s /rs . At this level
the phase transition is of a continuous one. However, 0.1
when the system is deep into a superfluid phase with a
crystalline order, there may induce a structural transi-
tion (where quantum numbers ne and no change) from 0
a CDW phase to another CDW phase. We expect it 0 0.5 1 µ/U 1.5 2
to be a first-order one, since it involves a discontin-
uous change of the free energy. In broken-symmetry FIG. 7. (color online). Phase diagram spanned by zt/U and
phases, the order parameters are determined by the co- µ/U from self-consistent mean-field calculation for K/U =
efficients. The grand potential is fully determined as 0.4.
Ω = Ω0 (ne , no , µ) + δΩ(ne , no , µ, zt). Given µ and zt,
the global minimum of the grand potential is achieved by
minimizing over non-negative integer of ne and no . How- WeNmay diagonalize H MF in the basis spanned by
ever, it should be noted that such perturbative treatment |ne > |no > by simultaneously imposing self-consistency
only give qualitatively sensible physics for the regime conditions for the charge order parameter θ =< n̂e − n̂o >
deep into the compressible phases. /2 and for the superfluid order parameters ψe =< b̂e >
and ψo =< b̂o >. It is clear that for self-consistent equa-
tions there always exists a trivial solution with ψe = ψo
III. SELF-CONSISTENT MEAN-FIELD and θ = 0 which corresponds to the SF phase. For a
THEORY. SS phase to be a true ground state, we require that its
ground energy is lower than that of a SF one.
The perturbative treatment given in the previous sec- The numerical results from this self-consistent theory
tion is valid only for a small hopping parameter. To ex- are shown in Fig. 5, Fig. 6 and Fig. 7. For K/U = 0.4,
plore physics deep into the compressible phases, we resort µ/U = 1.0 and zt/U = 0, the system is evidently in the
to a self-consistent mean-field approximation formulated phase of CDW(2,1) , as could be read from Fig. 1. Now
below. The mean-field Hamiltonian for a supercell can the evolution of the order parameters with respect to the
be constructed as follows: tuning parameter zt/U is shown in Fig. 5. Across the
X U  transition point zt/U = 0.123, both ψe and ψo acquires a
Ĥ MF
= n̂s (n̂s − 1) − µn̂s − 2Kθ(n̂e − n̂o ) nonzero value, signaling that the system enters into a SS
s=e,o
2 phase. When zt/U is further increased to zt/U = 0.192,
h  i the system enters into the SF phase with ψe = ψo and
−zt ψo b̂†e + ψe∗ b̂o − ψo ψe∗ + h.c. + 2Kθ2 . (13) φ = 0. This observation is consistent with our general ar-
 6

guments made above based on Landau-type free energy. where the hopping parameter is sufficient small. We find
The behavior of number density clearly follows the steps that the long-range interaction greatly enhances the ther-
of the superfluid order parameters. Now let us turn to mal fluctuation of the particle number. In the strong
Fig. 6, where at zt/U = 0 the system is in the phase of coupling regime, we identify four branches of elementary
CDW(4,0) since we choose K/U = 0.7 and µ/U = 1.0 for excitation, which corresponds to two types of hole exci-
illustration. As clearly seen in Fig. 6b, with the increase tation and two types of particle excitation. We derive
of hopping parameter zt/U , the system first undergoes low-energy effective energy functional for the system in
a structure transition from CDW(4,0) to CDW(5,0) , and the regime of a small hopping parameter. Finally we
further increment of zt/U drives the system to enter a SS construct a self-consistent mean-field theory, by which
phase with nonzero superfluid order parameter ψe and we find there exists structural phase transition between
ψo . With even larger zt/U , the system finally favors a different CDW phases driven by the hopping parameter.
SF phase over a SS phase with same superfluid order Currently, new research interests [32–35] including super-
parameter and particle density at even and odd sites. solidity breaking translational invariance and relaxation
To appreciate how zt/U affects the phase diagram dynamics are being cultivated along the lines of global
shown in Fig. 1, we show a phase diagram spanned by collective light-matter interaction.
zt/U and µ/U at K/U = 0.4 in Fig. 7. Evidently, for
a MI phase, increasing of zt/U to some finite value, the
system enters into the SF phase; while for a CDW phase,
increasing of zt/U first drives the system into a SS phase,
and finally into a SF phase with a sufficiently large zt/U .

IV. CONCLUSION

In summary, we have carried out field-theoretical per- We acknowledge funding from the NSFC under Grants
turbative study on physics in the strong coupling regime No. 11674058, 11274064 and NCET-13-0734.

[1] I. Bloch, J. Dalibard, and W. Zwerger, Rev. Mod. Phys. [16] B. Sundar and E. J. Mueller, Phys. Rev. A 94, 033631
80, 885 (2008) (2016)
[2] D. Jaksch, C. Bruder, J. I. Cirac, C. W. Gardiner, and [17] J. Panas, A. Kauch, and K. Byczuk, Phys. Rev. B 95,
P. Zoller, Phys. Rev. Lett. 81, 3108 (1998) 115105 (2017)
[3] M. Greiner, O. Mandel, T. Esslinger, T. W. Hänsch, and [18] T. Flottat, L.deForges deParny, F. Hebert, V. G.
I. Bloch, Phys. Rev. Lett. 95, 260401 (2005) Rousseau, and G. G. Batrouni, Phys. Rev. B 95, 144501
[4] I. Carusotto and C. Ciuti, Rev. Mod. Phys. 85, 299 (2017)
(2013) [19] M. P. A. Fisher, P. B. Weichman, G. Grinstein, and D. S.
[5] R. Landig, L. Hruby, N. Dogra, M. Landini, R. Mottl, Fisher, Phys. Rev. B 40, 546 (1989)
T. Donner, and T. Esslinger, Nature 532, 476 (2016) [20] J. K. Freericks and H. Monien, Phys. Rev. B 53, 2691
[6] K. Baumann, C. Guerlin, F. Brennecke, and T. Esslinger, (1996)
Nature 464, 1301 (2010) [21] R. A. Lehrer and D. R. Nelson, Phys. Rev. B 58, 12385
[7] R. Mottl, F. Brennecke, K. Baumann, R. Landig, T. Don- (1998)
ner, and T. Esslinger, Science 336, 1570 (2012) [22] D. vanOosten, P. vanderStraten, and H. T. C. Stoof,
[8] H. Ritsch, P. Domokos, F. Brennecke, and T. Esslinger, Phys. Rev. A 63, 053601 (2001)
Rev. Mod. Phys. 85, 553 (2013) [23] K. Sengupta and N. Dupuis, Phys. Rev. A 71, 033629
[9] O. Dutta, M. Gajda, P. Hauke, M. Lewenstein, (2005)
D. Lühmann, B. A. Malomed, T. Sowiński, and J. Za- [24] F. E. A. dos Santos and A. Pelster, Phys. Rev. A 79,
krzewski, Rep. Prog. Phys. 78, 066001 (2015) 013614
[10] S. F. Caballero-Benitez, G. Mazzucchi, and I. B. Mekhov, [25] B. Bradlyn, FranciscoEdnilsonA. dosSantos, and A. Pel-
Phys. Rev. A 93, 063632 (2016) ster, Phys. Rev. A 79, 013615 (2009)
[11] Y. Li, L. He, and W. Hofstetter, Phys. Rev. A 87, 051604 [26] H. T. C. Stoof, K. B. Gubbels, and D. B. M. Dickercheid,
(2013) Ultracold Quantum Fields (Springer, Bristol, UK, 2009)
[12] H. Habibian, A. Winter, S. Paganelli, H. Rieger, and [27] A. Altland and B. Simons, Condensed Matter Field The-
G. Morigi, Phys. Rev. Lett. 110, 075304 (2013) ory, 2nd ed. (CUP, Cambridge, UK, 2010)
[13] Y. Chen, Z. Yu, and H. Zhai, Phys. Rev. A 93, 041601 [28] F. Gerbier, Phys. Rev. Lett. 99, 120405 (2007)
(2016) [29] D. B. M. Dickerscheid, D. vanOosten, P. J. H. Denteneer,
[14] N. Dogra, F. Brennecke, S. D. Huber, and T. Donner, and H. T. C. Stoof, Phys. Rev. A 68, 043623 (2003)
Phys. Rev. A 94, 023632 (2016) [30] I. Herbut, A Modern Approach to Critical Phenomena
[15] A. E. Niederle, G. Morigi, and H. Rieger, Phys. Rev. A (Cambridge University Press, New York, 2007)
94 (033607) [31] S. Sachdev, Quantum Phase Transitions, 2nd ed. (Cam-
 7

bridge University Press, New York, 2011) 012128 (2017)

[32] J. Leonard, A. Morales, P. Zupancic, T. Esslinger, and [34] E. I. R. Chiacchio and A. Nunnenkamp
T. Donner, Nature 543, 87 (2017) [35] A. Morales, P. Zupancic, J. Leonard, T. Esslinger, and
[33] W. Casteels, R. Fazio, and C. Ciuti, Phys. Rev. A 95, T. Donner arxiv: 1711.07988