fiber [31], Peregrine soliton in nonlinear fiber optics system,
water tank and plasma [32–35], dark soliton on the surface of
water [36]. However, a number of phenomena arise in physics
can not be described by the continuous NLS equation. In the
past two decades, the discrete nonlinear systems have
attracted much research attention because of their applications
in optical waveguide arrays, bio-physics, nonlinear optical
couplers, Bose–Einstein condensates and so on [37–41].In
this paper, we consider the following discrete nonlinear
Schrödinger (DNLS) equation [42] in the form of
Q
t
QQQQQQi
d
d
2,2
n
nnnnnn11
2
11
=-++ +
+-+-
∣∣( )()
where Q
n
(t) is the complex field amplitude at the lattice site
n=0,±1,±2, K and t is the continuous evolution variable.
Equation (2 ) is also called Ablowitz–Ladik equation which is
an integrable discrete version of NLS equation with a non-
canonical Poisson structure [43]. If the nonlinear part in
equation (2) is replaced by a Kerr nonlinearity (i.e.,
QQ
nn
2
∣
),
it can model the propagation of discrete self-trapped beams in
an array of weakly coupled nonlinear optical waveguides, but
it is non-integrable [44 ]. Reference [45] have shown that the
analytical study in the framework of equation (2) is valuable
not only for such non-integrable DNLS equation but also for
other discrete systems. Besides, equation (2) with discrete
self-trapping nonlinearity can describe the interactions of
partial transverse electric modes in an array of focusing
waveguides [43, 46].
Several methods have been proposed to derive the loca-
lized-wave solutions for the continuous NLS equation, such as
the DT [47–53], Hirota bilinear method [54–56], Hyperbolic
function method [57]. Those methods can also be used to
derive the traveling-wave solutions of many nonlinear evol-
ution equations [58]. However, the discretization of the NLS
equation results in the impossibility of the derivation about the
space variable x. Thus the above methods can not be directly
applied to the DNLS equation and need to be modified (for
example, the lax pairs should be discretized to derive the DT).
Reference [59] has studied the dynamical properties of solitons
for equation (2) via a variational technique, which is based on
an effective Lagrangian. In [60], analytical and numerical
results for on-site and intersite collisions between solitons have
been reported for equation (2). On the other hand, it was found
that MI existed not only in continuous media but also could
develop in discrete equations with applications to optical and
electrical fields [46]. For equation (2), [46]
has investigated the
MI and showed in detail that the breather-type solutions are a
direct continuation of MI at larger amplitudes. Although the
single breather and rogue wave have been obtained [46],the
dynamics of interactions between two different waves need to
be analyzed and discussed. Therefore, in this paper, we would
like to answer the following questions about the DNLS
equation: (1) what are the interaction behaviors of the breathers
on the nonvanishing background; (2) whether the breather and
RW can interact on the same background; (3) whether there
exists the equal-eigenvalue degenerate second-order breathers
and what’s the difference between such breathers and the ones
given by the elementary DT.
This article is organized as following: in section 2,wegive
the exact first-order breather solution of equation (2) by the
elementary DT on the nonvanishing background, and figure
out the impact of parameters on the velocity and period.
Furthermore, via the second-order breather solution, we ana-
lyze the dynamic behaviors of breather–breather and breather-
rogue wave interactions. In section 3, we use the generalized
DT to obtain equal-eigenvalue degenerate second-order
breather solution and analyze the characteristics of interactions
between two breathers. The final section is our conclusion.
2. Elementary DT for equation ( 2) and localized-wave
solutions
The Lax pair of equation (2) is given as [61]
UU
zQ
Rz
a,,3
nnnn
n
n
1
1
F=F =
+
-
⎛
⎝
⎜
⎞
⎠
⎟
()
t
VV
RQ z zQ zQ
zR z R Q R z
b
d
d
,
ii1 ii
ii i i1
3
n
nn n
nn n n
nnnn
1
21
1
1
1
1
2
F
=F
=
+- - +
-+ - --
-
-
-
-
-
-
-
⎛
⎝
⎜
⎞
⎠
⎟
()
()
()
with
Q
n
n
*
=-
,where
,
nnn1, 2,
T
F=F F()
(the superscript T
signifies the vector transpose) is the vector eigenfunction, z is a
complex spectral parameter. It can be verified that the compat-
ibility condition
U
t
UV V U
n
nn n n1
¶
¶
+=
+
leads to equation (2).
As a special gauge transformation, the DT comprises
of the eigenfunction and potential function transformation
[61]. The elementary N-time successively-iterated DT of
equation (2) can be given based on Lax pair(3a) and(3b).In
doing so, the N-time iterated eigenfunction transformation is
assumed to be of the form
TT
A ntz B ntz
C ntz D ntz
,
,, ,,
,, ,,
4
n
N
n
N
n
n
N
NN
NN
F= F =
⎛
⎝
⎜
⎞
⎠
⎟
()()
()()
()
[] [] []
with
5
Antz z a tz
Bntz b tz
,, ,
,, ,
N
N
j
N
n
j
Nj
N
j
N
n
j
Nj
1
2
1
12
å
å
=+
=
=
-
=
+-
()
() ()
() ()
()
()
6
Cntz c tz
Dntz z d tz
,, ,
,, ,
N
j
N
n
j
Nj
N
N
j
N
n
j
Nj
1
12
1
22
å
å
=
=+
=
+-
-
=
+-
()
() ()
() ()
()
()
where
tb tc t,,
n
j
n
j
n
j
() () ()
() () ()
and
tjN1
n
j
()( )
()
can be
uniquely determined by requiring
T
n
N
zz
k
=
∣
[]
Φ
k,n
= 0 and
T 0
n
N
zkn,
z
k
1
*
F=
=
∣
[]
kN1
)
. The functions
kn,
F=
fg,
kn kn
T
,,
)
and
gf,
kn
kn kn
T
,
,,
**
F=()
respectively denote the
solution of Lax pair(3) with z=z
k
and
z
1
k
*
=
.
2
Phys. Scr. 93 (2018) 115203 MLiet al