Exact solutions of completely integrable

systems and linear ODEs having elliptic

function coecients
N. Kostov
1
, Z. T. Kostova
2
1
Institute of Electronics, Bulgarian Academy of Sciences,
72 Tsarigradsko chaussee, 1784 Soa, Bulgaria
2
National High School Soa, blvd.
Montevideo 21, Soa Bulgaria
July 11, 2007
Abstract
We present an algorithm for nding closed form solutions in el-
liptic functions of completely integrable systems. First we solve the
linear dierential equations in spectral parameter of Hermite-Halphen
type. The integrability condition of the pair of equations of Hermite-
Halphen type gives the large family of completely integrable systems
of Lax-Novikov type. This algorithm is implemented on the basis
of the computer algebra system MAPLE. Many examples, such as
vector nonlinear Schodinger equation, optical cascaded equations and
restricted three wave system are considered. New solutions for opti-
cal cascaded equations are presented. The algorithm for linear ODEs
with elliptic functions coecients is generalized to 2 2 matrix equa-
tions with elliptic coecients.
1
1 Intorduction
We consider a linear dierential equation in spectral parameter
L =
_
p
0
d
m
dx
m
+
m1

j=1
p
mj
(x)
d
mj
dx
mj
_
= , (1)
where p
j
are expressed in terms of doubly periodic functions having the same
periods. We will also require that our coecients are in fact elliptic functions.
New important development of algorithms for ODEs with elliptic coecients
is paper [BLH]. They are implemented in Maple 9 as DEsolve function. The
geometry of ODEs with elliptic function is considered in [GK] using Hermite-
Halphen algorithm. The aim of present paper is to present applications of
algorithms for ODEs with elliptic coecients to obtain periodic solutions of
completely integrable and near to completely integrable systems.
Let z(x) be a solution of
y

(x) 4r(x)y

(x) 2r

(x)y(x) = 0, (2)
and set
y
1
(x) =
_
z(x) exp
_

C
2
_
dx

z(x

)
_
, (3)
and
y
1
(x) =
_
z(x) exp
_
C
2
_
dx

z(x

)
_
, (4)
where C is a constant given by
C
2
= z

(x)
2
2z(x)z

(x) + 4r(x)z(x)
2
, (5)
if C = 0, the y
1
(x), y
2
(x) are linearly independent and form a basis of
y

(x) r(x)y(x) = 0, (6)

If C = 0, the the basis for solution space of (6) is given by y
1
(x) and y
2
(x) =
_
z(x)
_
1
z(x)
dx.
For proof see for example [BLH].
2
Introduce Hermite polynomial in spectral parameter as solution of the
following nonlinear dierential equation
1
2
FF
xx

1
4
F
2
x
(u(x) + )F
2
+
1
4
R() = 0, (7)
where r(x) = (u(x) +), C
2
= R(), F(x) = z(x). In modern literature (see
for example references in [CEEK]) functions (3) and (4) are called Baker-
Akhiezer (BA) functions.
1.1 Vector nonlinear Schr odinger equation
We consider the system of coupled nonlinear Schr odinger equations
i

t
Q
j
+ s

2
x
2
Q
j
+
_
n

k=1
|Q
k
|
2
_
Q
j
= 0, j = 1, . . . , n, (8)
where s = 1, = 1. We seek solution of (8) in the following form [EEK])
Q
j
= q
j
(z) e
i
j
, j = 1, . . . n, (9)
where z = x ct,
j
=
j
(z, t), with q
j
,
j
real. Substituting (9) into
(8) and separating real and imaginary parts by supposing that the functions

j
, j = 1, . . . n behave as

j
=
1
2
scx + (a
j

1
4
sc
2
)t s C
j
_
z
0
dz

q
j
(z

)
2
+
j0
,
we obtain the system ( = s = 1)
d
2
dz
2
q
j
+
_
n

k=1

s
q
2
k

a
j
s
_
q
j

C
2
j
q
3
j
= 0, k, j = 1, . . . n, (10)
where C
j
, j = 1, . . . n are free parameters and
j0
are constants. These
equations describe the integrable case of motion of a particle in a quartic
potential perturbed with inverse squared potential, which is separable in
ellipsoidal coordinates. The solutions of the system (10) are then given as
q
2
i
(z) = 2
F(z, a
i
)

n
k=i
(a
i
a
k
)
, i = 1, . . . , n, (11)
3
where F(z, ) is Hermite polynomial associated with Lame potential and is
dened as solution of (7). The nal formula for the solutions of the system
(8) then reads
Q
i
(x, t) =

2
F(z, a
i
)

n
k=i
(a
i
a
k
)
exp(
i
), (12)
where

j
=
_
_
_
1
2
icx + i(a
j

1
4
c
2
)t
1
2
(a
j
)
z
_
0
dz

F(z

, a
j
)
_
_
_
,
and i = 1, . . . , n and we have made use of (11) and (9). To obtain the special
class of periodic solution of (10) we introduce the following ansatses
q
i
() =
_
A
i
( +

) + B
i
, i = 1, 2, 3, or i = 1, . . . , 4. (13)
As a result we obtain:
m

k=1
A
k
= 2, a
i
=
m

k=1
B
k

B
i
A
i
, m = 3 or 4, (14)

4C
2
i
A
2
i
= (4
3
g
2
g
3
)
|
=B
i
/A
i
, i = 1, . . . , 3 or 4 (15)
and using the well known relations
_
z
0
dz

(z

) ( a
j
)
=
1

( a
j
)
_
2z( a
j
) + ln
(z a
j
)
(z + a
j
)
_
, (16)
and
(z +

) ( a
j
) =
(z +

+ a
j
)(z +

a
j
)
(z +

)
2
( a
j
)
2
. (17)
We derive the following result
Q
j
=
_
A
j
(z +

+ a
j
)
(z +

)( a
j
)

exp
_
i
2
cx + i(a
j

1
4
c
2
)t (z +

)( a
j
)
_
, (18)
4
where

j=1
A
j
= 2, a
j
=

k=1
B
k

B
j
A
j
,
C
j
A
j
=
i
2
_
4
3
g
2
g
3
|=
B
j
A
j
( a
j
) =
B
j
A
j
= a
j
, j = 1 . . .
1
,
1
= 3, 4 (19)
To obtain the class of periodic solutions of system (10) for n = 3, 4 we
introduce the following two ansatses in terms of the Weierstrass function
( +

)
q
i
() =
_
A
i
( +

)
3
+ B
i
( +

)
2
+ C
i
( +

) + D
i
, (20)
where i = 1, . . . 3. Next for conciseness we denote = ( +

), then the
second ansatz have the form
q
i
() =
_
A
i

4
+B
i

3
+ C
i

2
+ D
i
+ E
i
,
i = 1, . . . 4 (21)
with the constants A
i
, B
i
, C
i
, D
i
, E
i
dened from the compatibility condition
of the ansatz with the equations of motion (10). Inserting (20) and (21) into
Eqs. (10), using the basic equations for Weierstrass function [WW]
_
d
d
()
_
2
= 4()
3
g
2
() g
3
,
d
2
d
2
() = 6()
g
2
2
, (22)
and equating to zero the coecients at dierent powers of we obtain the fol-
lowing algebraic equations for the parameters of the solutions A
i
, B
i
, C
i
, D
i
, i =
1, 2, 3 for n = 3
A
1
+ A
2
+ A
3
= 0, B
1
+ B
2
+ B
3
= 0, (23)
C
1
+ C
2
+C
3
= 12, C
i
=
2
3
B
2
i
A
i

1
4
A
i
g
2
, (24)
a
i
=
3

i=1
D
i
5
B
i
A
i
, D
i
=
5
9
B
3
i
A
2
i

1
3
B
i
g
2

1
4
A
i
g
3
. (25)
5
The analogical algebraic system for n = 4 is as follows
A
1
+ A
2
+ A
3
+ A
4
= 0, B
1
+ B
2
+ B
3
+ B
4
= 0, (26)
C
1
+ C
2
+ C
3
+C
4
= 0, D
1
+ D
2
+ D
3
+ D
4
= 20, (27)
C
i
=
3
5
B
2
i
A
i

3
10
A
i
g
2
, D
i
=
14
45
B
3
i
A
2
i

53
180
B
i
g
2

2
9
A
1
g
3
(28)
E
i
=
49
225
B
4
i
A
3
i

113
450
B
2
i
A
i
g
2

11
36
B
i
g
3
+
9
400
A
i
g
2
2
. (29)
a
i
=
4

i=1
E
i
7
B
i
A
i
,
Another result from the algebraic systems is the expression for constants C
i
which parametrise our solutions. For them we obtain
C
2
i
=
(a
i
)
2

k=i
(a
i
a
k
)
,
where i, k = 3 or 4 and parameters are dened by (for n = 3)

2
=
7

63
2
g
2

5
+
297
2
g
3

4
+
4185
16
g
2
2

18225
8
g
2
g
3

2
+
91125
16
g
2
3

3375
16
g
2
2
, (30)
and (for n = 4)

2
=
9

231
2

7
g
2
+
2145
2
g
3

6
+
63129
16

5
g
2
2

518505
8
g
2
g
3

4
+
_

563227
16
g
3
2
+
4549125
16
g
2
3
_

3
+
991515
2
g
3
g
2
2

2
+
_
361179
4
g
4
2

5273625
4
g
2
g
2
3
_

972405g
3
g
3
2
1500625g
3
3
. (31)
Using the general formulae, we will consider below the physically important
cases of n = 3, 4 [EEK] which are associated with the three-gap 12( +

),
and four-gap elliptic potentials 20( +

).
6
The Hermite polynomial F((x), ) associated to the Lame potential
12() has the form
F((), ) =
3
6( +

)
2
3 5(3( +

)
2
+g
2
)

3
2
5
2
4
(4( +

)
3
g
2
( +

) g
3
). (32)
The solution is real under the choice of the arbitrary constants a
i
, i = 1, . . . , n
in such way, that the constants a
i
, i = 1, . . . , n lie in dierent lacunae.
Comparing (20) and (32) and using (11) the solutions of polynomial equations
(23),(24),(25) can be given by
A
i
=
2 5
2
3
2

3
k=i
(a
i
a
k
)
, (33)
B
i
=
2 3
2
5(a
i
)

n
k=i
(a
i
a
k
)
, (34)
=
2
5
3

i=1
a
i
. (35)
The Hermite polynomial F((), ) associated to the Lame potential 20()
can be written as
F((), ) = 11025( +

)
4
1575( +

)
3
+
(135
2

6615
2
g
2
)( +

)
2
+
(10
3
+
1855
4
g
2
2450g
3
)( +

) +

113
2

2
g
2
+
3969
16
g
2
2
+
195
4
g
3
. (36)
Comparing (21) and (36) and using (11) the solutions of polynomial equations
(26-29) can be given by
A
i
=
11025 2

k=i
(a
i
a
k
)
,
B
i
=
1575 2(a
i
)

k=i
(a
i
a
k
)
(37)
=
2
7
4

i=1
a
i
.
7
Next solution of system (10,n = 3) we obtain using the following ansatz
q
i
() =
_
A
i
( +

)
2
+ B
i
( +

) + C
i
, i = 1, 2, 3, (38)
then we have
3

i=1
A
i
= 0,
3

i=1
B
i
= 6, (39)
a
i
=
3

k=1
C
k
3
B
i
A
i
, C
i
=
B
2
i
A
i

1
4
A
i
g
2
, (40)
C
2
i
3
3
4
A
2
i
= (4
5
+ 27
2
g
3
+ 27g
2
2
21
3
g
2
81g
2
g
3
), (41)
where = 3B
i
/A
i
.
1.2 Optical cascading equations
Let us consider the system of two ordinary dierential equations,
q
1
+ A
0
q
1
+ B
0
q
1
q
2
= 0, (42)
q
2
+ C
0
q
2
+ D
0
q
2
2
= 0, (43)
where we have A
0
, B
0
, C
0
, D
0
are constants.
Introducing new variable
q
2
1
=
4F
B
0
D
0
, F =
2
3 + 9
2

9
4
g
2
(44)
where F is Hermite polynomial [WW], g
2
, g
3
are elliptic invariants dened
in [WW]. = ( +

) is Weierstrass function shifted by half period

is
related to sn Jacobian elliptic function with modulus k
( +

; g
2
, g
3
) =
2
k
2
sn
2
(, k) (1 + k
2
), (45)
where =

e
1
e
3
and e
i
, i = 1, 2, 3, e
3
e
2
e
1
are the real roots of the
cubic equation
4
3
g
2
g
3
= 0. (46)
8
Using wave height and modulus k =
_
(e
2
e
3
)/(e
1
e
2
) we have the
following relations
e
1
=
1
3
(2 k
2
)
2
, e
2
=
1
3
(2k
2
1)
2
, e
3
=
1
3
(1 + k
2
)
2
,
g
2
= 4(e
1
e
2
+ e
1
e
3
+ e
2
e
3
) =
4
3

2
(1 k
2
+k
4
),
g
3
= 4e
1
e
2
e
3
=
4
27

6
(k
2
+ 1)(2 k
2
)(1 2k
2
). (47)
Inserting this expression in (42) we have the following nonlinear dierential
equation with spectral parameter = C
0
/2
1
2
FF

1
4
F
2

(u() + )F
2
+
1
4
R() = 0, (48)
with eigenvalue equations
R() = 4
5
21
3
g
2
+ 27g
2
2
+ 27
2
g
3
81g
2
g
3
= 0,
u() = (B
0
q
2
+ + A
0
) = 6( +

), (49)
or in factorized form
R() = 4

(
i
) = 0,
1
=
_
3g
2
,
2
= 3e
3

3
= 3e
2
,
4
= 3e
1
,
5
=
_
3g
2
. (50)
It is well known that equation (48) is reduced to linear periodic spectral
problem of one dimensional Schr odinger equation with two gap potential
u(x) = 6( +

) and with ve normalized eigenfunctions q

(i)
1
, (i) = 1, . . . 5:
d
2
q
(i)
1
d
2

2
u()q
(i)
1
=
i
q
(i)
1
, (i) = 1, . . . , 5. (51)
Under these conditions the second equation (43) is automatically satised.
Second equation can be considered as self-consistent equation for potential
u(). Finally the ve spectral families of periodic solutions can be written in
the following Table 1
9
Table 1: Five spectral families of periodic solutions
(I). q
1
=
6

B
0
D
0

2
k
2
E
(u)
2
q
2
=
1
B
0
(u() +
3g
2

1
2
1
) (i)=1
(II) q
1
=
6

B
0
D
0

2
k E
(cd)
2
q
2
=
1
B
0
(u() +
3g
2

2
2
2
) (i)=2
(III) q
1
=
6

B
0
D
0

2
k E
(sd)
2
q
2
=
1
B
0
(u() +
3g
2

3
2
3
) (i)=3
(IV) q
1
=
6

B
0
D
0

2
k
2
E
(sc)
2
q
2
=
1
B
0
(u() +
3g
2

4
2
4
) (i)=4
(V) q
1
=
6

B
0
D
0

2
k
2
E
(u+)
2
q
2
=
1
B
0
(u() +
3g
2

5
2
5
) (i)=5
where
E
(sc)
2
= sn(, k)cn(, k),
E
(sd)
2
= sn(, k)dn(, k),
E
(cd)
2
= cn(, k)dn(, k),
E
(u)
2
= sn
2
(, k)
1 + k
2

1 k
2
+ k
4
3k
2
,
(52)
are normalized two-gap Lame functions [WW], cn, dn are Jacobian elliptic
functions and potential u() have the form
u() = 6
2
k
2
sn
2
(, k) 2(1 + k
2
)
2
. (53)
2 2 2 matrix spectral problems and inte-
grable systems
2.1 Baker-Akhiezer function
Let us start with two linear systems
d
1j
dx
+ F
1j
+ G
2j
= 0,
d
1j
dt
+

A
1j
+

B
2j
= 0, (54)
d
2j
dx
+ H
1j
F
2j
= 0,
d
2j
dt
+

C
1j


A
2j
= 0, (55)
which constitute [AKNS] scheme in particular case F = i, G = iu(x, t), H =
iu(x, t), where coecients depend on an arbitrary spectral parameter .
10
The compatibility conditions
j,xt
=
j,tx
, j = 1, 2 yield to the following
nonlinear system of equations:
F
t

A
x
+

CG

BH = 0,
G
t

B
x
+ 2(

BF

AG) = 0, (56)
H
t

C
x
+ 2(

AH

CF) = 0.
The general system (56) is equivalent also to zero curvature representation
U
t
V
x
+ [U, V ] = 0, (57)
where
U =
_
F G
H F
_
, V =
_
A B
C A
_
. (58)
The periodic solutions in elliptic functions are generated through special
matrices L whose representations are polynomials in the spectral parameter
and L obey the following set of equations:
L
x
= [U, L], L
t
= [V, L], L = , (59)
where
V =
N

k=0
_
V
11
k
V
12
k
V
21
k
V
22
k
_

Nk
=
_

V
11

V
12

V
21

V
22
_
. (60)
or in explicit form
dA
dx
= HB + GC, A(x, ) =
n+1

j=0
A
n+1j
(x, t)
j
, (61)
dB
dx
= 2FB 2GA, B(x, ) =
n

j=0
B
nj
(x, t)
j
, (62)
dC
dx
= 2FC + 2HA, C(x, ) =
n

j=0
C
nj
(x, t)
j
, (63)
and for N = 1
dA
dt
=

V
12
C

V
12
B, A(, ) =
n+1

j=0
A
n+1j
(x, t)
j
, (64)
11
dB
dt
= 2

V
11
B 2

V
12
A, B(, ) =
n

j=0
B
nj
(x, t)
j
, (65)
dC
dt
= 2

V
11
C + 2

V
21
A, C(, ) =
n

j=0
C
nj
(x, t)
j
. (66)
The equations (61),(62),(63) yield that
(A(x, )
2
B(x, )C(x, ))
x
= 0 (67)
and hence

2
= A(x, )
2
B(x, )C(x, ) = R
2n+2
(), (68)
where the integration constant R
2n+2
is a polynomial in of degree 2n + 2.
After a chain of simple transformations we obtain

1,x
=
1
2B
(B
x
2R
2n+2
G)
1
, (69)

2,x
=
1
2C
(C
x
2R
2n+2
H)
2
. (70)
After integration as result we have

1
=

Bexp
_

_
R
2n+2
_
x
G
B
dx
_
, (71)

2
=

C exp
_

_
R
2n+2
_
x
H
C
dx
_
. (72)
2.2 Restricted multiple three wave interaction system
Let us consider coupled quadratic nonlinear oscillators

db
j
d
+ uc
j

1
2

j
b
j
= 0, (73)

dc
j
d
+ u

b
j
+
1
2

j
c
j
= 0, (74)

du
d
+
n

j=1
b
j
c

j
= 0, (75)
12
where is the evolution coordinate and
j
are constants. The equations
(73-75) can be written as Lax representation
dL
d
= [M, L], (76)
of the following linear system:
d
d
= M(, )(, ) L(, )(, ) = 0, (77)
where L, M are 2 2 matrices and have the form
L(, ) =
_
A(, ) B(, )
C(, ) D(, )
_
, (78)
M(, ) =
_
/2 iu
u

/2
_
. (79)
where
A(, ) = a()
_

2
+

2
n

j=1
_
c
j
c

j
b
j
b

j
_

j
_
, (80)
B(, ) = a()
_
u
n

j=1
b
j
c

j

j
_
, (81)
C(, ) = a()
_
u

j=1
c
j
b

j

j
_
, (82)
where D(, ) = A(, ) and a() =

n
i=1
(
i
). The Lax representation
yields the hyperelliptic curve K = (, )
det(L()
1
2
1
2
) = 0, (83)
where 1
2
is the 22 unit matrix. The curve (83) can be written in canonical
form as

2
= 4
2n+2

j=1
(
j
) = R(), (84)
13
where
j
=
k
are branching points. Next we develop a method which allows
to construct periodic solutions of system (73-75). The method is based on the
application of spectral theory for self-adjoint one dimensional Dirac equation
with periodic nite gap potential U = u cf. Eqs. (73,74)

d
1j
d
U
2j
i
j

1j
= 0, (85)

d
2j
d
U

1j
+ i
j

1j
= 0, (86)
with spectral parameter and eigenvalues
j
= i
j
/2. The equation (76) is
equivalently written as
dA
d
= iuC iu

B, A(, ) =
n+1

j=0
A
n+1j
()
j
, (87)
dB
d
= iB 2iuA, B(, ) =
n

j=0
B
nj
()
j
, (88)
dC
d
= iC + 2iu

A, C(, ) =
n

j=0
C
nj
()
j
, (89)
or in dierent form we have
A
j+1,
= iuC
j
iu

B
j
, A
0
= 1, A
1
= c
1
, (90)
iB
j+1
= B
j,
2iuA
j+1
, B
0
= 2u, (91)
iC
j+1
= C
j,
2iu

A
j+1
C
0
= 2u

, (92)
where c
1
is the constant of integration. Dierenciating Eq. (87) and using
(83) we can obtain
BB

u
BB

1
2
B
2

+
_

2
2
i
u

u
+|u|
2
_
B
2
= 2u
2
. (93)
Using (69) the eigenfunction
1
for nite-gap potential U have the form

1
(, ) =
_
U()
U(0)
B(, )
B(0, )
_
1/2
exp
_
i
_

0
_
R()
B(

, )
d

_
.
14
Analogously we can write expression for
2
(, ) and nally elliptic solu-
tions of initial system of restricted three interaction system take the form
b
j
() = b
0
j

1
(,
j
), c
j
() = c
0
j

2
(,
j
), j = 1 . . . n, (94)
where b
0
j
, c
0
j
are constants xed by initial conditions.
3 Implementation
In [GK] H-H (Hermite-Halphen) algorithm is presented and implemented in
computer algebra REDUCE. Geometric interpretation of solutions found in
[GK] is discussed in [EK]. Recently [BLH] Maple 9 [CGGMW] implementa-
tion of algorithm for solving linear ODEs having elliptic function coecients
is reported. New algorithm is found. This implementation is very important
for deriving new solutions of integrable and nonintegrable dynamical systems
with elliptic solutions [EK, GR, GH, B]. Important problem in deriving el-
liptic solutions is factorization of algebraic curves [CGHKW]. Algorithm
for deriving elliptic solutions presented above is implemented on computer
algebra Maple 10. The source code is available under request. Using this
implementation new solutions of Manakov system in external potential are
derived in [KEGKS]
References
[BLH] Burger, R., Labahn, G., van Hoeji, M: Closed form solutions of lin-
ear odes having elliptic function coecients. ISSAC04 Proceedings
(2004) 5864
[GK] Gerdt, V., Kostov, N: Computer algebra in the theory of ordinary
dierential equations of Halphen type, Computers and mathemat-
ics,Proc.of Int. Conf. Cambridge, USA, Springer, New-York, (1989)
178188.
[WW] Whittaker, E., Watson, G.: A Course of Modern Analysis, (Cam-
bridge University Press, Cambridge, 1986).
[EK] Enolskii, V., Kostov N.: On the geometry of elliptic solitons, Acta
Applicandae Math., 36 (1994) 5786
15
[CEEK] Christiansen, P., Eilbeck, J., Enolskii, V., Kostov, N.: Quasi-
periodic and periodic solutions for Manakov type systems of coupled
nonlinear Schrodinger equations, Proc. Royal Soc. London A. 456
(2000), 22632281
[EEK] Eilbeck, J., Enolskii, V., Kostov, N.: Quasi-periodic and periodic
solutions for vector nonlinear Schrodinger equations, Journ. Math.
Phys. 41 (2000) 82368248
[K] Kostov, N.: Quasi-periodic and periodic solutions for dynamical sys-
tems related to Korteweg-de Vries equation, The European Physical
Journal B. 29 (2002), 255260
[AKNS] Ablowitz, M., Kaup, D., Newell, A., Segur, H.: The inverse scat-
tering transformFourier analysis for nonlinear problems, Studies in
Appl. Math. 53 (1974) 249315
[KEGKS] Kostov, N., Enolskii, V., Gerdjikov, V., Konotop, V.,Salerno,
M., Two-component Bose-Einstein condensates in periodic poten-
tial, Phys. Rev E. 70 (2004), 056617
[S] Smirnov, A.: Elliptic soliton and Heuns equation, The Kowalevski
property, CRM Proc. Lecture Notes. 32 (2002) 287305
[CGGMW] Maple Reference Manual, B.W. Char, K.O. Geddes, G.H.
Gonnet, M.B. Monagan and S.M. Watt, Watcom Publications, Wa-
terloo, (1988)
[CGHKW] Corless, R., Giesbrecht, M., van Hoeij, M., Kotsireas, I., Watt,
S.:Towards factoring bivariate approximate polynomials, Interna-
tional Conference on Symbolic and Algebraic Computation , Pro-
ceedings of ISSAC01, ACM Press New York, NY, USA, (2001)
[GH] Gesztesy, F., Holden, H.: Soliton Equations and Their Algebro-
Geometric Solutions. Vol. I: (1+1)-Dimensional Continuous Models,
Cambridge University Press, Cambridge, (2003).
[GR] Gesztesy, F.,Ratneseelan, R.: An alternative approach to algebro-
geometric solutions of AKNS hierarchy, Rev. Math. Phys., 10 (1998)
345391
16
[B] Brezhnev, Yu.: Elliptic solitons and Grobner bases, Journal of Math-
ematical Physics, 45 (2004) 696712
17