
Caution on Gross-Neveu criticality with a single Dirac cone:
Violation of locality and its consequence of unexpected finite-temperature transition
Yuan Da Liao,1, 2 Xiao Yan Xu,3Zi Yang Meng,4, ∗and Yang Qi1, 2, 5, †
1State Key Laboratory of Surface Physics, Fudan University, Shanghai 200438, China
2Center for Field Theory and Particle Physics, Department of Physics, Fudan University, Shanghai 200433, China
3Key Laboratory of Artificial Structures and Quantum Control (Ministry of Education),
School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai 200240, China
4Department of Physics and HKU-UCAS Joint Institute of Theoretical and Computational Physics,
The University of Hong Kong, Pokfulam Road, Hong Kong SAR, China
5Collaborative Innovation Center of Advanced Microstructures, Nanjing 210093, China
(Dated: November 9, 2023)
Lately there are many SLAC fermion investigations on the (2+1)D Gross-Neveu criticality of a single Dirac
cone. While the SLAC fermion construction indeed gives rise to the linear energy-momentum relation for all
lattice momenta at the non-interacting limit, the long-range hopping and its consequent violation of locality on the
Gross-Neveu quantum critical point (GN-QCP) – which a priori requires short-range interaction – has not been
verified. Here we show, by means of large-scale quantum Monte Carlo simulations, that the interaction-driven
antiferromagnetic insulator in this case is fundamentally different from that on a purely local 𝜋-flux Hubbard
model on the square lattice. In particular, the antiferromagnetic long-range order has a finite temperature
continuous phase transition, which appears to violate the Mermin-Wagner theorem, and smoothly connects to
the previously determined GN-QCP. The magnetic excitations inside the antiferromagnetic insulator are gapped
without Goldstone mode, even though the state spontaneously breaks continuous 𝑆𝑈 (2)symmetry. These
unusual results point out the fundamental difference between the QCP in SLAC fermion and that of GN-QCP
with short-range interaction.
I. INTRODUCTION
Massless Dirac fermions are ubiquitously present as the
low-energy description of many condensed matter systems in-
cluding graphene [1], twisted bilayer graphene [2–5], d-wave
superconductors [6–9], algebraic spin liquid [6,7,10–17] and
the deconfined quantum criticality [18–30]; in high-energy
physics, the dynamical massless Dirac fermions in quantum
chromodynamics and the existence of a deconfined phase in
compact quantum electrodynamics have attracted great atten-
tions and remains unsolved [10,15,31–36]. Nonetheless, it is
generally believed that strong local interactions can generate
a finite mass for the Dirac fermions and spontaneously result
in a quantum phase transition [37–42]. The corresponding
quantum critical point (QCP) are typically described by the
Gross-Neveu (GN) university classes [43,44]. In particular,
a single Dirac cone, realized in the the SLAC fermion model
with long-range hopping in (2+1)D [45,46], was found to
give rise to an Ising-type ferromagnetic order that generates a
𝑍2symmetry-breaking mass gap [47], or an antiferromagnetic
Mott insulator that breaks the 𝑆𝑈 (2)spin rotational symme-
try [48]. The associated QCPs from Dirac semimetal (DSM)
to insulators are believed to belong to the (2+1)D chiral Ising
or Heisenberg GN universality classes.
The SLAC fermion construction gives rise to a linear
energy-momentum relation for all lattice momenta at the non-
interacting limit (shown in Fig. 1(a)), therefore reduces the
finite-size effect suffered by other local cousins such as the
∗zymeng@hku.hk
†qiyang@fudan.edu.cn
honeycomb and 𝜋-flux models where only a small region of
the Brillouin zone (BZ) displays the relativistic behavior at
low-energy. The fundamental difference of the SLAC fermion
model compared with its local cousins, i.e., the necessity of
avoiding the Nielsen-Ninomiya theorem [49–51] by violat-
ing locality on finite size lattices and the assumption that the
locality of the Dirac operator is recovered in the thermody-
namic limit (TDL), has not be investigated. This means, with
the long-range interactions in the SLAC fermion models (the
bare interaction is on-site but the long-range hopping medi-
ates long-range interaction), whether the GN transition and
the symmetry-breaking phases obtained thereafter can be dis-
cussed as if they were from a purely local model in the origin
sense of GN-QCP [43,44], are questionable.
This is the problem solved in this article. Here we show, by
means of large-scale QMC simulations, that the phase diagram
of the SLAC fermion model is fundamentally different from
that of a purely local 𝜋-flux Hubbard model on the square
lattice. In particular, we find the antiferromagnetic insula-
tor (AFMI) phase in the SLAC fermion model exists at finite
temperatures, which appears to violate the Mermin-Wagner
theorem [52–54]. The AFMI phase emerges from the high-
temperature paramagnetic (PM) phase via a finite-temperature
continuous phase transition, and this continuous transition line
smoothly connects to the previously determined GN-QCP at
the ground state [48]. Contrary to the picture of the Mermin-
Wagner theorem, where the low-energy fluctuation of the gap-
less Goldstone mode destroys the long-range order at any finite
temperature, we find that the magnetic excitations inside the
AFMI are gapped without Goldstone mode, although the state
spontaneously breaks continuous 𝑆𝑈(2)symmetry.
Our results suggest that the long-range interaction in the
SLAC fermion model has altered the low-energy effective
arXiv:2210.04272v2 [cond-mat.str-el] 7 Nov 2023