
2
potential. In contrast to the true disorder, quasiperiodic
potential can cause a delocalization-localization transi-
tion in 1D even in the absence of interaction, and this
Hamiltonian is known as Aubry-Andr´e (AA) model66.
While the nature of the localized phase observed due to
quasiperiodic potentials and true disorders are the same
(in both cases, eigenstates are exponentially localized),
there are differences in critical properties associated with
the localization-delocalization transition. In both cases,
the localization length ξdiverges at the transition fol-
lowing as ξ∼δ−ν, where δis the distance to the critical
point in the parameter space and νis the localization
length exponent. In true disorder-driven localization,
there is a rigorous bound on the localization length ex-
ponent ν, i.e., it must satisfy ν≥2/d criteria to ensure
the stability of the transition67. However, in quasiperi-
odic models, such criteria do not apply. For most of the
one-dimensional quasi-periodic models, νis close to 168.
In recent days, there have been numerous studies, both
theoretically and experimentally, involving a variant of
the AA model that shows the coexistence of both local-
ized and delocalized states separated by a mobility edge
69–83.
In this work, one of the main aims is to address the
question of what happens to the fate of Anderson lo-
calization in the context of SFQM in the presence of
quasiperiodic potential, we call it L´evy quasicrystal. In-
terestingly, a similar question has been asked in a recent
study26, but a detailed theoretical understanding of the
phase diagram of the different phases was lacking there.
Also, the L´evy-flight models have been used extensively
in understanding real-life financial markets, the traveling
behavior of humans, and even biological systems. If one
wants to model such a system and take into account cor-
related random events, our L´evy quasicrystal model can
be an extremely suitable candidate for such cases. We
demonstrate the phase diagram as a function of αand the
strength of the quasiperiodic potential ∆, which is shown
in Fig. 1. While we find that like the Aubry-Andr´e model
(which is the discretized version of the usual Schr¨odinger
equation), there exists a completely delocalized (DL) and
Anderson localized (AL) phase, but on top of that there
is a parameter region where these two types of states co-
exist, and they are separated by the mobility edge (ME).
Those ME phases can also be characterized as different
psphases (where s > 0 can be any positive integer) based
on the fraction of delocalized states in the spectrum. We
also compare our model with a potential experimentally
realizable AA model with power-law hopping.
The paper is organized as follows. In Sec. II, we intro-
duce the model of L´evy quasicrystal. Next, we discuss
the analytical prediction of the mobility edge in Sec. III.
In Sec. IV, we show our numerical results. Sec. V shows
the comparison between L´evy quasicrystal and the effec-
tive power-law hopping model and finally, we summarize
our results in Sec. VI.
II. MODEL
The one-dimentional space-fractional Schr¨odinger
equation is given by,
H|ψ⟩=DαPα|ψ⟩+V|ψ⟩=E|ψ⟩,(1)
where Dα∈Ris a constant (note that D2is equivalent
of inverse of mass term in usual Schr¨odinger equation),
Pαis the α-th power of the momentum operator, Vis
the potential energy operator and |ψ⟩is the eigenstate
with eigenvalue E2,3,5.
The position space representation of α-th power of
the momentum operator is ⟨x|Pα|ψ⟩=−ℏαDα
|x|ψ(x)4,5,
where Dα
|x|is the Riesz Fractional Derivative of order α.
While usually to get the finite first moment of the L´evy
process, αis taken within the limit α∈(1,2] 4, but given
there exist, biological models, where the L´evy parameter
is taken to be 0 < α < 115, also in a very recent optical
experiment, the parameter range α < 1 of the fractional
Schr¨odinger equation has been realized55 in the temporal
domain, that motivates us to even explore the extended
parameter regime, and we use the same position space
representation of α-th power of the momentum opera-
tor for α∈(0,2] to get the discretized version. Riesz
Fractional Derivative of order α(note that we consider
α= 1 point separately in the appendix), can be written
as5,84(note that the expression of the Riesz Fractional
Derivative is taken from Ref.84),
Dα
|x|=−1
2 cosαπ
2(I−α
++I−α
−),(2)
where I−α
±are given by (approximating Gr¨unwald-
Letnikov operators)5,84,
I−α
±ψ(x) = lim
a→0
1
aα
∞
X
n=0
(−1)nα
nψ[x∓an],(3)
for 0 < α ≤1.
I−α
±ψ(x) = lim
a→0
1
aα
∞
X
n=0
(−1)nα
nψ[x∓(n−1)a],(4)
for 1 < α ≤2. Where astands for a small positive step
length. Here, α
n=Γ(α+1)
Γ(n+1)Γ(α−n+1) , where Γ(.) is the
usual Gamma function. It is straightforward to check
that for α= 2 Riesz Fractional Derivative becomes the
standard second-order derivative.
In this work, we consider the potential to be space-
dependent (quasiperiodic potential), hence the model can
be represented by the equation,
⟨xl|H|ψ⟩=⟨xl|DαPα|ψ⟩+ ∆ cos(2πβl +ϕ)ψ(xl).(5)
xlis equally spaced grid points i.e xl=la,a=lattice
constant, a > 0 and l∈Z.βis an irrational number.