
2
Ref. [23] derived a GMP algebra for all topological flat-
bands of C 6= 0. However, the assumption of constant
Berry curvature oversimplified the flatband problem.
In this work, we relax the above assumption by fo-
cusing on ideal flatbands satisfying the following condi-
tions [24]:
gab
k=1
2ωabΩk,Ωk>0 for ∀k.(2)
The ωab is a constant uni-modular matrix. The Berry
curvature is assumed to be positive definite but not nec-
essarily uniform. Recently, it is known that Eq. (2) is
necessarily [25] and sufficiently [26–28] equivalent to the
momentum-space holomorphicity, which is a key geo-
metric property of the momentum-space (= boundary-
condition space). For all C= 1 ideal flatbands such
as those realized in the chiral model of twisted bilayer
graphene [29,30] and others [31–33], it has been shown
that momentum-space holomorphicity highly constrains
the wavefunction to admit a universal form descend-
ing from the LLL wavefunction [24]. Such simplicity
enables exact construction of many-body model wave-
functions [30] as well as exact projective pseudopotential
Hamiltonians [24], hence the ideal condition plays an im-
portance role in experimental realization and identifica-
tion of FCIs [34,35]. Recently the ideal condition Eq. (2)
has been generalized to a larger family called “vortexabil-
ity” [36] by allowing nonlinear real-space embedding [37].
While much progress are made for C= 1, less is known
for generic cases of C>1. Recently, solvable models
based on twisted multilayer graphene sheets were pro-
posed which realize ideal flatbands of arbitrary Chern
number [38,39]. Remarkably, exact Halperin type FCIs
were found by numerical diagonalization [38]. Moti-
vated by the C= 1 case, it was conjectured that the
momentum-space holomorphicity is also the fundamen-
tal reason for their emergence [38], however, the nature
of the Bloch wavefunction and why FCIs are stable are
still far from thorough understanding.
In this work, we prove that the ideal quantum geo-
metric condition, without assuming the flatness of Ωk
or gkthemselves, is sufficient to guarantee exact GMP
algebra and model FCI states occurring as exact zero-
energy ground states of proper short-ranged interactions
in all C ≥ 1 ideal flatbands. We achieve these results
by pointing out the importance of an emergent Hilbert
space in which the wavefunctions’ normalization factors
are adjusted to flatten Berry curvature in an exact man-
ner. Importantly, this leads to a simple algebra for
the projected coordinates identical to the guiding cen-
ter algebra in LLs, and enables an exact derivation for
the single-particle wavefunction in ideal flatbands. The
general form Bloch wavefunction is found to be a non-
linear superposition of LLL wavefunctions of Cdistinct
boundary conditions, generalizing the previously pro-
posed color-entangled wavefunctions with uniform Berry
curvature [40–42] to the case of fluctuating Berry cur-
vature. The density algebra is proved to be closed, ex-
tending the GMP algebra initially derived for C= 1 LLs
to generic C ≥ 1 ideal flatbands. The closeness of the
density algebra directly leads to the exact FCIs stabi-
lized by generic M−body repulsive interactions which
are Halperin type when C>1 and typically non-Abelian
when M > 2. Thus the ideal condition Eq. (2) provides a
general and exact statement for stabilizing FCIs regard-
less of the nonuniformness of Berry curvature.
The paper is structured as follows. In Sec. II, we pose
the problem by showing unusual numerical observations.
In Sec. III, we review key analytical results for ideal
flatbands, emphasizing their real and momentum-space
boundary conditions and define the emergent Hilbert
space. We show in Sec. IV a well defined guiding center in
the ideal flatband problems which leads to the derivation
of single-particle wavefunctions in Sec. V. In Sec. VI, the
density algebra is derived and the origin of model FCIs is
explained. We discuss application of the theory to FCIs
and symmetry breaking phases [34,43] in moir´e flatband
systems in Sec. VII.
II. MOTIVATION FROM UNUSUAL
NUMERICAL OBSERVATIONS
The ideal flatbands are not abstract concepts, rather
they can be realized in concrete models including Dirac
fermion based models [29,33,38,39] and Kapit Mueller
models [31,32,44,45]. In particular, in the chiral twisted
multilayer graphene models [38,39] ideal flatbands of ar-
bitrary C 6= 0 are realized exactly. In this section, we use
this model as a concrete platform to numerically study
the interacting spectra, which point out a couple of im-
portant and unusual aspects of the projected density op-
erator that motivates the theory to be discussed in the
following sections.
For self-consistency, we first briefly review the chiral
twisted multilayer graphene models [38,39], although the
model details are not crucial for the current discussion.
Details about the chiral twisted graphene model and be-
yond can be found in the literature [29,30,46–51]. The
chiral twisted multilayer model consists of two sheets of
graphene multilayers each of which has n−layers and is
A/B stacked without twist [52–54]. The top sheets are
twisted relative to the bottom sheets as a whole by a twist
angle θ. See Fig. 1(a) for illustration of the setup. At
magic twist angles, the model exhibits two exactly dis-
persionless degenerate bands showing in Fig. 1(b), which
have Chern number C=±n. Such degeneracy can be
easily lifted by sublattice contrasting potential, and we
focus on the C=nband. The flatband wavefunction is
derived in Refs. [38,39], and its components on the out-
most layers are found to be responsible for the high Chern
number and zero-mode FCIs. We hence project the wave-
function to the outmost layers to get an effective single-
component wavefunction, denoted as φBloch
k(r). It can
be numerically verified that φBloch
k(r) is smooth in both
kand r. The Berry curvature associated to φBloch
k(r),