Darboux Transformation and Exact Solutions of the Continuous Heisenberg Spin Chain Equation
Ai-Hua Chenaand Fan-Fan Wangb
aCollege of Science, University of Shanghai for Science and Technology, Shanghai, 200093, People’s Republic of China
bDepartment of Mathematics, East China University of Science and Technology, Shanghai, 200237, People’s Republic of China
Reprint requests to F.-F. W.; E-mail:ffwang@ecust.edu.cn Z. Naturforsch.69a,9 – 16 (2014) / DOI: 10.5560/ZNA.2013-0067
Received March 11, 2013 / revised September 10, 2013 / published online October 30, 2013 In this paper, we give theN-fold Darboux transformation (DT) for the continuous Heisenberg spin chain which describes the motion of the isotropic ferromagnets in the complex case. By using this DT, we getN-soliton solutions and a new exact solution of the spin chain from a trivial seed solution and a plane wave seed solution, respectively.
Key words:Darboux Transformation; Heisenberg Spin Chain; Breather Solution; Rational Solution.
PACS numbers:02.30.Ik; 02.30.Jr
1. Introduction
In the 1970s, the continuous Heisenberg spin chain which describes the motion of the magnetization vec- tor of the isotropic ferromagnets has attracted much attention [1–4]. In [3], the explicit single-soliton so- lutions in the isotropic case were presented. In [4], the inverse scattering method was applied to the continu- ous Heisenberg spin chain and its Lax representation was obtained. In [5], the N-soliton solution was ob- tained by applying the Wadati gauge transformation to the inverse scattering problem of a nonlinear evolu- tion equation. In [6], theN-soltion solution was given explicitly according to the bilinear equations obtained by Hirota. Since the 1970s, the continuous Heisenberg spin chain also received much attention. In [7], the higher-order Heisenberg spin chain equations were de- duced by Chen and Li and they proved that these equa- tions are equivalent to the evolution equations of the Ablowitz–Kaup–Newell–Segur (AKNS) By using the nonlinearization method, Qiao [8] presented a finite- dimensional integrable system and the involutive so- lutions of the higher-order Heisenberg spin chain. In [9], the relation among the different systems associated with the Heisenberg magnetic equation was studied by the reduction procedure.
The Darboux transformation (DT) as a useful method to get explicit solutions of nonlinear partial differential equations has been utilized to obtain so- lutions of the continuous Heisenberg spin chain. In [10,11], for the spectral problem related to the con- tinuous Heisenberg spin chain, explicit one-fold Dar- boux transformations were constructed to obtain its soliton solutions. In [12], Cie´sli´nski and Czarnecka constructed the Darboux–Bäcklund transformation for the two-dimensional Heisenberg chain. In [13], Saleem and Hassan constructed the Darboux transformation for the generalized Heisenberg magnet model and ob- tained multi-soliton solutions in terms of quasidetermi- nants.
In [11], the soliton solutions were obtained for the continuous Heisenberg spin chain equation by the one- fold DT, and the iterative relationship between theNth solution and the (N+1)st solution can be derived by applying the one-fold DTNtimes. However, the rela- tionships between the new solutions and the seed solu- tions are not given. In this paper, according to the de- terminant representation of DT [14–17], we directly construct the N-fold Darboux transformation for the continuous Heisenberg spin chain by using a similar method as in [18]. By using of theN-fold DT, we can obtain the relationships between the new solutions and
© 2014 Verlag der Zeitschrift für Naturforschung, Tübingen·http://znaturforsch.com
the seed solutions without tedious and complicated it- erations.
In this paper, we study the continuous Heisenberg spin chain introduced in [4], where Takhtajan consid- ered an infinite linear chain with spin-densitysj(x,t) for j=1,2,3 and length of the spins equal to 1, i.e.
s21(x,t) +s22(x,t) +s23(x,t) =1. The motion equation of the continuous Heisenberg spin chain is
St= 1
2i(SSxx−SxxS), (1)
whereS=s1σ1+s2σ2+s3σ3, andσj(j=1,2,3)are the Pauli matrices
σ1=
0 1 1 0
, σ2=
0 −i
i 0
,
σ3=
1 0 0 −1
.
(2)
If we letw=s3andu=s1−is2, then from (1), we get ut=i(uwxx−wuxx), wt= i
2(u∗uxx−uu∗xx), (3) where w is a real-valued function, u is a complex- valued function, andu∗denotes the complex conjugate ofu. According to the length constraint condition, we have
|u|2+w2=1. (4)
In Section2, according to the spectral problems in [7–11], we give the Lax pair of (3) and (4). Then we construct the N-fold Darboux transformation for the Heisenberg spin chain (1). In Section3, using different seed solutions, we get exact solutions of (1). From the trivial seed solution, we get breather solutions for the first two components and soliton solution for the third component of the spin chain. These solutions are sim- ilar with those in [10,11]. From the plane wave seed solution, we get a new exact solution of the spin chain.
In Section4, we make our conclusion.
2. Darboux Transformation
The spectral problem of the system (3) and (4) (cf.
[7–11]) is given by φx=Uφ=iλ
w u u∗ −w
φ with |u|2+w2=1,
(5)
and the corresponding auxiliary spectral problem is given by
φt=Vφ (6)
= 2iwλ2+12(uu∗x−u∗ux)λ 2iuλ2+ (wux−uwx)λ 2iu∗λ2+ (u∗wx−wu∗x)λ −2iwλ2−12(uu∗x−u∗ux)λ
! φ,
whereφ= (φ1,φ2)T.λis a constant spectral parameter, uandware functions ofxandt.
We want to find a transformation ¯φ=Tφ, such that the Lax pair
φx=Uφ, φt=Vφ (7) is transformed into
φ¯x=U¯φ¯, φ¯t=V¯φ¯, (8) where ¯U and ¯V have the same forms asU andV re- spectively, which are obtained by replacingu,u∗,w, ux,u∗x,wx with ¯u, ¯u∗, ¯w, ¯ux, ¯u∗x, ¯wx, respectively, inU andV. Then the matrixT satisfies
Tx+TU=U T¯ , Tt+TV=V T¯ . (9) According to the zero curvature equation, if(u,w)sat- isfies (3),(u,¯ w)¯ also satisfies (3). The transformation φ¯=Tφ is a Darboux transformation for the Lax pair (5) and (6).
In the following, we first construct the Darboux transformation for the Lax pair (5) and (6). Then we give the relationship between(u,w)and(u,¯w).¯
Firstly, the Lax pair (5) and (6) satisfies the loop group conditions
U(λ∗) =σ2U∗(λ)σ2, V(λ∗) =σ2V∗(λ)σ2, (10) whereσ2 is the Pauli matrix defined in (2). Equation (10) implies thatT satisfiesT(λ∗) =σ2T∗(λ)σ2(con- straints imposed by reduction groups, see [19–21]).
Then from (9), we have to demandT(0) =const. in order to preserve the conditionsU(0) =V(0) =0 and U(0) =¯ V¯(0) =0. For simplicity, we supposeT(0) =I.
So we let
T =
1+
N
∑
k=1
Akλk
N
∑
k=1
Bkλk
−
N
∑
k=1
B∗kλk 1+
N
∑
k=1
A∗kλk
(11)
whereAkandBk(1≤k≤N)are functions ofxandt. Let Φ(j) = (ϕ1(λj),ϕ2(λj))T and Ψ(j) = (ψ1(λj),ψ2(λj))T be two basic solutions of (5) and (6) withλ =λj (j=1,2, . . . ,N). ThenΦ∗(j) = (−ϕ2∗(λj),ϕ1∗(λj))T and Ψ∗j = (−ψ2∗(λj),ψ1∗(λj))T be two basic solutions of (5) and (6) with λ =λ∗j (j=1,2, . . . ,N). Ak and Bk (1≤k≤N) satisfy the following two algebraic systems:
N
∑
k=1
(Ak+ηjBk)λjk=−1, (12)
N k=1
∑
(−Bk+η∗jAk)λ∗jk=−η∗j, (13)
where
ηj=ϕ2(λj)−rjψ2(λj)
ϕ1(λj)−rjψ1(λj), 1≤ j≤N. (14) Note that complex λj and real rj should be suitably chosen such that the determinants of the coefficient matrices of systems (12) and (13) are nonzero and thus AkandBk(1≤k≤N)can be uniquely determined.
From the form ofT in (11), we find that detT(λ) = (|AN|2+|BN|2)
N
∏
j=1(λ−λj)(λ−λj∗), (15) which means that λj andλ∗j are roots of detT(λ) = 0. At the same time, from T(0) = I, we know that detT(0) =1= (|AN|2+B2N)|λ1|2|λ2|2· · · |λN|2, i.e.
|AN|2+|BN|2= 1
|λ1|2|λ2|2· · · |λN|2. (16) In the following theorem, we give the relationship between(u,w)and(u,¯ w).¯
Theorem 1. If(u,w)is a solution of (3) and (4),(u,¯ w)¯ with
¯
u=uA2N−u∗B2N−2wANBN
|AN|2+|BN|2 ,u[N], w¯=w(|AN|2− |BN|2) +u∗A∗NBN+uANB∗N
|AN|2+|BN|2 ,w[N]
(17)
is a new solution of (3) and (4), where AN and BN are determined by (12) and (13).
Proof. From (17), it can be verified by a straightfor- ward calculation that|u|¯2+w¯2=1. In order to prove that(u,¯ w)¯ is a new solution of (3), we need to verify the validity of (9).
Step 1: We verify the validity of the first part of (9), i.e.Tx+TU=U T¯ .
LetT−1=adj(T)/detT and (Tx+TU)·adj(T) =
f11(λ) f12(λ) f21(λ) f22(λ)
,F(λ). (18) A direct calculation shows that λ−1f11(λ), λ−1f12(λ),λ−1f21(λ), andλ−1f22(λ)are all 2Nth- order polynomials inλ, andF(λ∗) =σ2F∗(λ)σ2. By using (5), (12), (13), and (14), we get
ηjx=iu∗λj−2iwλjηj−iuλjη2j, 1≤j≤N. (19) We can verify thatλjandλ∗j (1≤j≤N)are roots of
fkl(λ) =0 fork,l=1,2. Therefore, we have
(Tx+TU)·adj(T) =detT·P(λ) (20) withP(λ∗) =σ2P∗(λ)σ2, i.e.
P(λ) = p(1)11λ p(1)12λ
−p(1)12∗λ p(1)11
∗
λ
!
, (21)
wherep(1)11 andp(1)12 are independent ofλ. We can fur- ther rewrite (20) as
Tx+TU=P(λ)T. (22)
Comparing the coefficients ofλN+1in (22), we find iwAN+iu∗BN=p(1)11AN−p(1)12B∗N,
iuAN−iwBN=p(1)11BN+p(1)12A∗N.
(23) From (17) and (23), we have
p(1)11 =i ¯w, p(1)12 =i ¯u. (24) Then we obtainTx+TU=U T¯ .
Step 2: As in Step 1, we verify the validity of the second part of (9), i.e.Tt+TV=V T¯ similarly.
LetT−1=adj(T)/detT and (Tt+TV)·adj(T) =
g11(λ) g12(λ) g21(λ) g22(λ)
,G(λ). (25) A direct calculation shows that λ−1g11(λ), λ−1g12(λ), λ−1g21(λ), and λ−1g22(λ) are all (2N + 1)st-order polynomials in λ and G(λ∗) = σ2G∗(λ)σ2. By using (6), (12), (13), and (14), we result in
ηjt=2iu∗λj2+ (u∗wx−wu∗x)λj
−
4iwλ2j + (uu∗x−u∗ux) ηj
−
2iuλ2j+ (wux−uwx)λj
η2j, 1≤j≤N. (26)
We can verify thatλjandλj∗(1≤j≤N)are roots of gkl(λ) =0 fork,l=1,2. Therefore, we have
(Tt+TV)·adj(T) =detT·Q(λ) (27) withQ(λ∗) =σ2Q∗(λ)σ2, i.e.
Q(λ) = (28)
q(2)11λ2+q(1)11λ q(2)12λ2+q(1)12λ
−q(2)12∗λ2−q(1)12∗λ q(2)11∗λ2+q(1)11∗λ
! ,
whereq(k)11 andq(k)12 (k=1,2)are independent ofλ. We can further rewrite (27) as
Tt+TV=Q(λ)T. (29)
Comparing the coefficients ofλN+2in (29), we get 2iwAN+2iu∗BN =q(2)11AN−q(2)12B∗N, 2iuAN−2iwBN=q(2)11BN+q(2)12A∗N.
(30) From (17) and (30), we obtain
q(2)11 =2i ¯w, q(2)12 =2i ¯u. (31) If we further compare the coefficients ofλN+1of (29), by using (31), we have
1
2(uu∗x−uxu∗)AN+2iwAN−1+ (u∗wx−wu∗x)BN (32) +2iu∗BN−1=q(1)11AN+2i ¯wAN−1−q(1)12B∗N−2i ¯uB∗N−1,
−1
2(uu∗x−uxu∗)BN−2iwBN−1+ (wux−wxu)AN +2iuAN−1=q(1)11BN+2i ¯wBN−1+q(1)12A∗N+2i ¯uA∗N−1.
According to step 1, by comparing the coefficients of λNinTx+TU=U T¯ , we find
ANx+iwAN−1+iu∗BN−1=i ¯wAN−1−i ¯uB∗N−1, BNx+iuAN−1−iwBN−1=i ¯wBN−1+i ¯uA∗N−1. (33)
By using (32), (33), and the expressions of ¯u and
¯
w, through tedious calculation (which can also be done easily by using the Mathematica software), we achieve
q(1)11 =1
2(u¯u¯∗x−u¯∗u¯x), q(1)12 =1
2(w¯u¯x−u¯w¯x). (34) Then we have Tt+TV =V T¯ . The proof is com- plete.
From Theorem1and the transformationw=s3,u= s1+is2, we find that
s¯1= (u¯+u¯∗)/2, s¯2=i(u¯−u¯∗)/2, s¯3=w¯ (35) is a new real-valued solution of the Heisenberg equa- tion (1).
In the following section, we will obtain exact so- lutions of (3) and (4) by the use of Theorem1. Then through (35), we will obtain real-valued exact solutions of the Heisenberg equation (1).
3. Exact Solutions
In this section, we will take a trivial solution and a plane wave solution as the seed solution, respectively, to obtain exact solutions of the Heisenberg equation (1).
3.1. Trivial Seed Solution
In this subsection, by using the N-fold Darboux transformation and taking a trivial solution as seed so- lution, we give the breather and soliton solutions of (1).
It is easy to see that(u,w) = (0,1)is a trivial solution of (3) and (4), and we take it as the seed solution. From (16) and (17), the solution of (3) and (4) is given by
u[N] =−2|λ1|2|λ2|2· · · |λN|2ANBN,
w[N] =1−2|λ1|2|λ2|2· · · |λN|2|BN|2. (36)
From (35), we can obtain a series of solutions of the Heisenberg equation (1) which are given by
s1[N] = u[N] +u[N]∗
2 =−|λ1|2|λ2|2· · · |λN|2
·(ANBN+A∗NB∗N), s2[N] = i(u[N]−u[N]∗)
2 =i|λ1|2|λ2|2· · · |λN|2
·(A∗NB∗N−ANBN),
s3[N] =w[N] =1−2|λ1|2|λ2|2· · · |λN|2|BN|2. (37) We choose two basic solutions of (5) and (6) as
ϕ(λj) = (ei(λjx+2λ2jt),0)T, ψ(λj) = (0,e−i(λjx+2λ2jt))T.
(38) For simplicity we letrj=−1(j=1,2, . . . ,N)in (14).
Then we have
ηj=e−2i(λjx+2λ2jt), j=1,2, . . . ,N. (39) From the linear system (12), (13) and the expression of (39), we find that ifλj(j=1,2, . . . ,N)are real, the obtained solutions are trivial. To get non-trivial solu- tions, we chooseλj(j=1,2, . . . ,N)as complex.
WhenN=1, takingλ1=τ1+iτ2, whereτ1andτ2
are real, the solution of (3) and (4) is u[1] =−e2i(τ1x+2(τ12−τ22)t)
τ12+τ22
·h
2τ22tanh(2τ2(x+4τ1t))−2iτ1τ2 i
·sech(2τ2(x+4τ1t)), w[1] =1− 2τ22
τ12+τ22
sech2(2τ2(x+4τ1t)).
(40)
-2 -1
0 1
2 x
-0.4 -0.2
0 0.2
0.4
t -11-20
2
s1 1 -2
-1 0 x 1
(a)
-3 -2
-1 0 x
0 0.2
0.4 0.6
0.81 t -11-20
2
s2 1 -3
-2 x -1
(b)
-0.5 0 x 0.5
-0.2 -0.1
0 0.1
0.2
t -11-20
s3 1 -0.5
0 x 0.5
(c)
Fig. 1 (colour online). Plots of solution (41) of Heisenberg equation (1) whenτ1=1 andτ2=2.
Then a solution of the Heisenberg equation (1) is s1[1] = −1
τ12+τ22 h
2τ22tanh(2τ2(x+4τ1t))
·cos(2τ1x+4(τ12−τ22)t) +2τ1τ2sin(2τ1x+4(τ12−τ22)t)i
·sech(2τ2(x+4τ1t)), s2[1] = 1
τ12+τ22 h
2τ22tanh(2τ2(x+4τ1t)) (41)
·sin(2τ1x+4(τ12−τ22)t)
−2τ1τ2cos(2τ1x+4(τ12−τ22)t)i
·sech(2τ2(x+4τ1t)), s3[1] =1− 2τ22
τ12+τ22
sech2(2τ2(x+4τ1t)).
For given values of τ1=1 and τ2=2, we give the plots of s1[1], s2[1], and s3[1] in Figure 1. We find that s3[1] is a bell-shaped single-soliton solu- tion, and s1[1] and s2[1] are single-soliton solutions with exchange interactions, i.e. they are breather so- lutions.
WhenN=2, for simplicity, taking λ1=τ1+iτ2, λ2=−τ1+iτ2, whereτ1andτ2are real, the solution of (3) and (4) is
u[2] =−4τ1τ2
θ12 h
θ2+2iτ1τ2(e2α1−e2α2)i
(θ3+iθ4),
w[2] =1−8τ12τ22 θ12
·
(eα1+eα1+2α2)2+ (eα2+e2α1+α2)2−2θ5 (42)
-2 0
x 2 -0.5
0 0.5 t -2
0 2 s1 2
-2 0 x 2
(a)
-2 0
x 2 -0.5
0 0.5 t -2
0 2 s2 2
-2 0 x 2
(b)
-1 0
x 1 -0.2
0 0.2 t -11-20
2 s3 2
-1 0 x 1
(c)
Fig. 2 (colour online). Plots of solution (43) of Heisenberg equation (1) whenτ1=1 andτ2=2.
and the solution of (1) is s1[2] =4τ1τ2
θ12
−θ2θ3+2τ1τ2(e2α1−e2α2)θ4 , s2[2] =4τ1τ2
θ12
θ2θ4+2τ1τ2(e2α1−e2α2)θ3 ,
s3[2] =1−8τ12τ22 θ12
(eα1+eα1+2α2)2 + (eα2+e2α1+α2)2−2θ5
,
(43) where
θ1= (e2α1+1)(e2α2+1)τ12−(2 eα1+α2cos(β1−β2)
−e2α1−e2α2)τ22,
θ2= (e2α1+1)(e2α2+1)τ12+ (2 eα1+α2cos(β1−β2)
−e2α1−e2α2)τ22,
θ3= (eα2+e2α1+α2)sinβ2−(eα1+eα1+2α2)sinβ1, θ4= (eα2+e2α1+α2)cosβ2−(eα1+eα1+2α2)cosβ1, θ5= (eα1+eα1+2α2)(eα2+e2α1+α2)cos(β1−β2) with
α1=2τ2(x+4τ1t), β1=−2(τ1x+2 τ12−τ22)t , α2=2τ2(x−4τ1t), β2=2(τ1x−2(τ12−τ22)t).
For given values ofτ1=1 andτ2=2, we give the plots of (43) in Figure2. We find thats1[2]ands2[2]
are two head-on breather solutions, ands3[2]is a bell- shaped two-soliton solution.
3.2. Plane Wave Seed Solution
In this subsection, by the modified Darboux trans- formation method [22–24], we can get rational solu- tions of (3) and (4) by the one-fold Darboux transfor- mation. Then, new exact solutions of (1) can be ob- tained.
In order to get rational solutions, we take a plane wave solution as the seed solution. It is easy to see that(u,w) = (bei(2x+4at),a)satisfies (3), wherea and b(b6=0)are real. To satisfy (4), we havea2+b2=1.
Taking(u,w)as the seed solution, we suppose that the basic solutions of (5) and (6) have the following forms:
φ1(λ),φ1(x,t)
= (a1x+b1t+c1xt+d1)ei(x+2at), φ2(λ),φ2(x,t)
= (a2x+b2t+c2xt+d2)e−i(x+2at),
(44)
whereai,bi,ci, anddi(i=1,2)are complex, andλ= a+ib. Substituting (44) into the Lax pair (5) and (6) and comparing all the coefficients, we have
a1=ia2,b1= (−2b+4ia)a2, b2= (4a+2ib)a2,
c1=c2=0,d2=a−ib
b a2−id1.
(45)
In order to preserve the positivity of|φ1|2+|φ2|2and get some solutions with simple forms, we choosea2= i,d1=a−ib
2b . Then, φ1(x,t) =h
−x−(4a+2ib)t+a−ib 2b
i
ei(x+2at), φ2(x,t) =h
ix−(2b−4ia)t+b+ia 2b
i
e−i(x+2at). (46)
For simplicity, we letr1=0 in (14), and we get η1(λ),η1(x,t) =φ2(x,t)
φ1(x,t)
=
ix−(2b−4ia)t+b+ia 2b
−x−(4a+2ib)t+a−ib 2b
e−2i(x+2at). (47)
-2 -1
0 1
2
x -5
-2.5 0 52.5
t -1-2210 s1 1
-2 -1
0 1
2
x -1-210
(a)
-2 0
2 x -5
-2.5 0 52.5
t -1-2210 s2 1
-2 0
2 x -1-210
(b)
-1 0
1 x -1
-0.5 0 0.5 1
t -1
0 1 s3 1
-1 0
1 x -1
0 1
(c)
Fig. 3 (colour online). Plots of solution (51) of Heisenberg equation (1).
Then from the linear systems (12) and (13), we ob- tain
A1=−a−ib+ (a+ib)|η1|2
1+|η1|2 , B1= 2ibη1∗
1+|η1|2. (48) From (16) and (17), the solution of (3) and (4) is given by
u[1] =uA21−u∗B21−2wA1B1,
w[1] =w(|A1|2− |B1|2) +u∗A∗1B1+uA1B∗1. (49)
Now from (35), we obtain the solution of the Heisen- berg equation (1) which is given by
s1[1] =1 2 h
u(A21−B∗12) +u∗(A∗12−B21)
−2w(A1B1+A∗1B∗1)i , s2[1] =i
2 h
u(A21+B∗12)−u∗(A∗12+B21)
−2w(A1B1−A∗1B∗1)i ,
s3[1] =w(|A1|2− |B1|2) +u∗A∗1B1+uA1B∗1.
(50)
If we further leta=0, b=1, the solution of the Heisenberg equation (1) is
s1[1] =− h
(4x2+16t2+1)2−32x2i
cos 2x−8x(4x2+16t2−1)sin 2x (4x2+16t2+1)2 , s2[1] =
8x(4x2+16t2−1)cos 2x+h
(4x2+16t2+1)2−32x2 i
sin 2x (4x2+16t2+1)2 , s3[1] =− 64xt
(4x2+16t2+1)2.
(51)
The plots of (51) are given in Figure3. Note that s3[1]is a rational solution. To the best of our knowledge, (51) is a new solution of the Heisenberg equation (1).
Since the seed solution is related toλ, we cannot ap- ply theN-fold DT whenN=2,3, . . .. Although we can iterate the one-fold DT to get new solutions, it would be much more complicated.
4. Conclusions
In this paper, we construct the N-fold DT of the Heisenberg equation (1). By the use of the obtained DT, from the trivial seed solution (u,w) = (0,1), we get breather solutions for the first two components and
soliton solution for the third component of the spin chain (s1,s2,s3). From the plane wave seed solution (u,w) = (bei(2x+4at),a)witha2+b2=1, we get a new exact solution of the spin chain(s1,s2,s3). Particularly, the third component of the spin chain is a rational so- lution.
Acknowledgements
We are most grateful to the anonymous referee for pointing out much relevant literature and crucial help in improving the original manuscript. The work de- scribed in this paper was supported by the Fundamen- tal Research Funds for the Central Universities.
[1] K. Nakamura and T. Sadada, Phys. Lett. A 48, 321 (1974).
[2] M. Lakshmanan, Phys. Lett. A61, 53 (1977).
[3] J. Tjon and J. Wright, Phys. Rev. B15, 3470 (1977).
[4] L. A. Takhtajan, Phys. Lett. A64, 235 (1977).
[5] H. Shimizu, J. Phys. Soc. Jpn.52, 507 (1984).
[6] Y. Fukumoto and T. Miyazaki, J. Phys. Soc. Jpn. 55, 4152 (1986).
[7] D. Y. Chen and Y. S. Li, Acta. Math. Sin. 2, 343 (1986).
[8] Z. J. Qiao, Phys. Lett. A186, 97 (1994).
[9] D. L. Du, Phys. A303, 439 (2002).
[10] Y. S. Li, Chin. Ann. Math. Ser. A8, 139 (1987).
[11] J. Wang, J. Phys. A: Math. Gen.38, 5217 (2005).
[12] J. L. Cie´sli´nski and J. Czarnecka, J. Phys. A: Math.
Gen.39, 11003 (2006).
[13] U. Saleem and M. Hassan, J. Phys. A: Math. Theor.43, 045204 (2010).
[14] G. Neugebauer and R. Meinel, Phys. Lett. A100, 467 (1984).
[15] J. L. Cie´sli´nski, J. Math. Phys.32, 2395 (1991).
[16] A. L. Sakhnovich, Inv. Prob.10, 699 (1994).
[17] J. S. He, L. Zhang, Y. Cheng, and Y. S. Li, Sci. China Series A: Math.49, 1867 (2006).
[18] F. F. Wang and A. H. Chen, J. Math. Phys.53, 083502 (2012).
[19] A. V. Mikhailov, Physica D3, 73 (1981).
[20] J. L. Cie´sli´nski, J. Phys. A: Math. Theor.42, 404003 (2009).
[21] J. Cie´sli´nski, arxiv:1303.5472 [nlin.SI].
[22] A. Ankiewicz, J. M. Soto-Crespo, and N. Akhmediev, Phys. Rev. E81, 046602 (2010).
[23] X. L. Wang, W. G. Zhang, B. G. Zhai, and H. Q. Zhang, Commun. Theor. Phys.58, 531 (2012).
[24] B. G. Zhai, W. G. Zhang, X. L. Wang, and H. Q. Zhang, Nonlin. Anal.: Real World Appl.14, 14 (2013).