arXiv:1404.

7329v2 [hep-th] 7 Jan 2015

Lax pair formulation in the simultaneous presence of

boundaries and defects

Anastasia Doikou
Department of Mathematics, Heriot-Watt University,
EH14 4AS, Edinburgh, United Kingdom
and
Department of Computer Engineering & Informatics,
University of Patras, GR-Patras 26500, Greece
E-mail:

A.Doikou@hw.ac.uk

Abstract
Inspired by recent results on the effect of integrable boundary conditions on the bulk
behavior of an integrable system, and in particular on the behavior of an existing defect we
systematically formulate the Lax pairs in the simultaneous presence of integrable boundaries
and defects. The respective sewing conditions as well as the relevant equations of motion on
the defect point are accordingly extracted. We consider a specific prototype i.e. the vector
non-linear Schrodinger (NLS) model to exemplify our construction. This model displays
a highly non-trivial behavior and allows the existence of two distinct types of boundary
conditions based on the reflection algebra or the twisted Yangian.

Contents
1 Introduction

2 The Lax pair construction

3 The vector NLS theory

3.1

The reflection algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.2

The twisted Yangian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4 Discussion

15

Introduction

The main objective of the present investigation is the construction of Lax pairs in the simultaneous presence of integrable boundaries and defects. This study is mainly inspired
by recent results on the study of integrable impurities at both clasical and quantum level
[1][23] as well as the findings of [24], where the presence of non trivial boundary conditions yielded highly non-trivial behavior associated to the defect. We shall provide here the
generic framework for the defect in the presence of two distinct types of boundary conditions, and shall exemplify our construction using the vector non-linear Schrodinger (NLS)
model. The underlying high rank algebra glN that rules the model allows the existence of a
rich variety of boundary conditions that preserve integrability, which renders the behavior
of the system even more intricate. It is worth noting that in addition to the derivation of
the local integrals of motion presented in [24] the Lax pair construction is equally important
providing the sewing conditions compatible with integrability. They are entailed as analyticity requirements imposed on the time components of the Lax pair around the defect point
(see also e.g. [19]). The derivation of Lax pairs, which is one of the main results in this
context in the simultaneous presence of bounaries and defcts and is presented in section
2, is the starting point for making contact with the Backlund transformation on the defect
point (see e.g. [6]). The present investigation may be seen as the continuation of the work
in [24], where the local integrals of motion were extracted. Here explicit derivation and
computation of Lax pairs, derivation of the sewing conditions and the equations of motion
associated to the defect point is presented.
Let us introduce below the general setting of our formulation. We shall consider here the
1

situation where both non-trivial boundary conditions and point-like defects are present in
the 1+1 dimensional classical field theory. Let us first introduce the modified monodromy
matrix to include the point-like defect at x = x0 (see e.g. [25, 4, 19] and references therein):
T (; L, 0) = T + (L, x0 ; ) L(x0 ; ) T (x0 , 0; )

(1.1)

where L is the algebraic object associated to the local defect and is required to be a
representation of a suitable quadratic classical algebra as will be discussed later in the text,
we also define the bulk monodromy matrices as (see e.g. [25]):
Z x

T (x, y; ) = P exp{
dx U (x ; )}
(1.2)
y

and they are solutions of the differential equation

T (x, y)
= U (x, t, )T (x, y).
x

(1.3)

The fact that T is a solution of equation (1.3) will be extensively used subsequently for
obtaining the relevant integrals of motion. U is part of the Lax pair U , V that satisfy
the linear auxiliary problem. Let be a solution of the following set of equations

= U (x, t, )
x

= V (x, t, )
t

(1.4)
(1.5)

U , V are in general n n matrices with entries functions of complex valued fields, their
derivatives, and the spectral parameter . Compatibility conditions of the two differential
equations (1.4), (1.5) lead to the zero curvature condition
i
h

(1.6)
U V + U , V = 0,
giving rise to the corresponding classical equations of motion of the system under consideration. Special care should be taken on the defect point; the zero curvature condition on
the point reads as [19, 20]
dL()
+ () L() L() V
()
=V
dt

(1.7)

are the time components of the Lax pairs on the defect point form the left (right).
where V
This is an intricate equation, which arises naturally when studying the continuum limit of
discrete integrable theories (see e.g. [19]).
2

Let us also briefly review the algebraic setting regarding the system in the presence of
point like defects and integrable boundaries, i.e. the Hamiltonian formalism. The existence
of the classical r-matrix [26], satisfying the classical Yang-Baxter equation
h
i h
i
r12 (1 2 ), r13 (1 ) + r23 (2 ) + r13 (1 ), r23 (2 ) = 0,
(1.8)
guarantees the integrability of the classical system. The monodromy matrices T as well
as the defect matrix L and consequently the modified monodromy matrix T (L, 0; ) satisfy
the classical quadratic algebra [25]:
n
o h
i
T1 (x, y, t; 1), T2 (x, y, t; 2) = r12 (1 2 ), T1 (x, y, t; 1) T2 (x, y, t; 2) . (1.9)
also the main requirement [19] for the L-matrix in (1.1) so that integrability is ensured is:
n
o h
i
L1 (1 ), L2 (2 ) = r12 (1 2 ), L1 (1 ) L2 (2 ) .
(1.10)
and
{T1+ (1 ), T2 (2 )} = 0.

(1.11)

Here, we are interested as already mentioned in the situation when both local defects
and integrable boundary conditions are present. As in [24] we shall distinguish two cases
of boundary conditions the soliton preserving (SP) associated to the reflection algebra and
the soliton non-preserving (SNP) associated to the twisted Yangian:
Reflection algebra: The classical reflection algebra [27] describes the so called solitonpreserving (SP) boundary conditions:
n
o
T1 (), T2 () = r12 ( ) T1 () T2 () T1 () T2 () r21 ( )
+ T1 () r21 ( + ) T2 () T2 () r12 ( + ) T1 ()
(1.12)
The corresponding representation of the algebra is given as [27]:
Ta () = Ta (L, 0; ) Ka () Ta1 (L, 0; )

(1.13)

where the monodromy matrix includes also the defects and is expressed as in (1.1). Recall
that T are defined in (1.2). Also K is a c-number representation
{Ka (), Kb ()} = 0.

(1.14)

of the classical reflection algebra (1.12). The local integrals of motion are obtained via the
generating function
a ()Ta ()),
G() = ln tra (K
(1.15)
3

 is also a c-number (non-dynamical) (1.14) solution of the classical reflection equation

K
Twisted Yangian: The classical twisted Yangian [27, 28] describes the soliton nonpreserving boundary conditions (SNP) and is defined as:
n
o
T1 (), T2 () = r12 ( ) T1 () T2 () T1 () T2 () r21 ( )
t1
t2
+ T1 () r21
( + ) T2 () T2 () r12
( + ) T1 ()

(1.16)

ti

denotes transposition on the auxiliary space i. The generic representation of the classical
twisted Yangian is given by [27, 29]
Ta () = Ta (L, 0; ) Ka () Tata (L, 0; ),

(1.17)

where K is a non-dynamical solution of the reflection equation (1.14). The local integrals of
motion are obtained via the generating function (1.15). Having set the algebraic background
we may now proceed with the construction of the Lax pairs.

The Lax pair construction

We shall focus on the SP case mainly because in this case non-trivial bulk behavior is
manifest. Nevertheless, we shall briefly describe the SNP case as well in the subsequent
section. Note that the basic expressions derived here are to our knowledge novel. We shall
first focus on the construction of the time component of the Lax pairs for the left and right
bulk theories. To derive the time component of the Lax pair in the simultaneous presence
of boundaries and defects we shall need the following fundamental algebraic relations (see
also [25, 19, 30], for more details):
o M (, ) h
i
n

ab
Ta (), U
()
=
+
M
(,
),
U
()
,
(2.1)
b
ab
b
x
where
+
Mab
(, ) = Ta+ (L, x; ) rab ( ) Ta+ (x, x0 ; ) La (x0 ; ) Ta (x0 , 0; ),

Mab
(, ) = Ta+ (L, x0 ; ) La (x0 ; ) Ta (x0 , x; ) rab ( ) Ta (x0 , 0; )

x > x0
x < x0 .
(2.2)

Recall that here we are interested in the SP (reflection algebra) boundary conditions. In
order to formulate the suitable Poisson bracket for our purposes we shall also need the
following
i
n
o M (, ) h

ab
+
M
(,
),
U
()
(2.3)
Ta1 (), U
()
=
ab
b
b
x
4

where we define:
+
+
1
Mab
(, ) = Ta (0, x; )1 L1
Ta+ (L, x; )1 ,
a (x0 ; ) rab ( + ) Ta (x, x0 ; )

+
1
Mab
(, ) = Ta (x, 0; )1 rab ( + ) Ta (x0 , x; )1 L1
a (x0 ; ) Ta (L, x0 ; ) ,

x > x0
x < x0 .
(2.4)

Using the fundamental relations (2.1), (2.3) we can formulate the necessary for our purposes
Poisson bracket
o
 h
i
n


Ka ()Ta (), Ub () =
Mab (, ) + Mab (, ) + Mab (, ) + Mab (, ), Ub ()
x
(2.5)

and M , M are defined as:

M
ab (, ) = Ka () Mab (, ) Ka () Ta ()

M
ab (, ) = Ka () Ta () Ka () Mab (, ).

(2.6)

M , M are defined in (2.2), (2.4).

After considering the generating function of the local integrals of motion G() we conclude that (see also [25]):
i
V (, ) h
+ V (, ), U ()
U () =
x

(2.7)

where the bulk quantities V are expressed as (they are defined up to an overall normalization factor):



Vb (, ; x) t () tra Mab (, ) + Mab (, )

x 6= x0
(2.8)
where we define
a () Ta ())
t() = tra (K

(2.9)

T is defined in (1.13) in the reflection algebra case, and in (1.17) in the twisted Yangian
case. Expansion of the V in powers of 1 will provide the time components of the Lax
pairs associated to each local integral of motion
V (, ) =

X V(n) ()
n

(2.10)

As already mentioned earlier in the text special care is taken on the defect point. Taking
into account the L-matrix satisfies the quadratic algebra (1.10) we conclude:
n
o
+ (, ) Lb () Lb () M
(, )
Ta (), Lb () = M
(2.11)
ab
ab
5

and
n

Ta1 (),

+ (, ) Lb () Lb () M
(, ).
Lb () = M
ab
ab

(2.12)

where we define
+ (, ) = T + (L, x0 ; ) rab ( ) La () T (x0 , 0; )
M
a
a
ab
(, ) = T + (L, x0 ; ) La () rab ( ) T (x0 , 0; )
M
a
a
ab
+

1
1

Mab (, ) = Ta (x0 , 0; ) La () rab ( + ) Ta+ (L, 0; )1

(, ) = T (x0 , 0; )1 rab ( + ) L1 () T + (L, 0; )1 .
M
ab

(2.13)

Then it is straightforward to show that

o 



n
+
+

Ka () Ta (), Lb () = Mab (, )+ Mab (, ) Lb ()Lb () Mab (, )+ Mab (, )

(2.14)
where
(, ) = K
a () M
(, ) Ka () T 1 ()
M
a
ab
ab

Mab (, ) = Ka () Ta () Ka () Mab (, ).

(2.15)

(x0 ), i.e. the Lax pairs on the defect point from the left
The aim now is to identify V
and from the right. Indeed, bearing in mind the zero curvature condition on the defect point
operators (defined up to an overall normalization
we may directly identify the relevant V
factor)


(, ; x0) t1 () tra M
(, ) + M
(, ) .
V
(2.16)
b
ab
ab
(, ) = P
As in the bulk case V
n

(n) ()
V
.
n

Then analyticity conditions imposed on the time components of the Lax pairs
(n) (x ) V(n) (x0 )
V
0

(2.17)

will provide the wanted sewing conditions as explained in [19].

Note that in the case of SNP boundary conditions the same expressions for the time
components are valid, but now we define

ta

Mab
(, ) = Mab
(, )

ta
(, ) = M
(, ) .
(2.18)
M
ab
ab
In the next section we shall consider the vector NLS model in the presence of defects
and boundaries associated to the reflection equation and the twisted Yangian.
6

The vector NLS theory

We shall focus hereafter on the glN vector NLS model. Let us first review the necessary
information regarding the model. The Lax pair for the generalized NLS model is given by
the following expressions [25]:
U = U0 + U1 ,

V = V0 + V1 + 2 V2 ,

(3.1)

where
N 1
1 X
U1 = (
eii eN N ),
2i i=1

V0 = i

N
1
X

1
X
N
U0 =
(i eiN + i eN i )
i=1

1
X
N
2

(i eiN i eN i ),
(i j eij |i | eN N ) i
i=1

i, j=1

V1 = U0 ,

V2 = U1 .

(3.2)

eij are N N matrices such that: (eij )kl = ik jl . Consider also the corresponding classical
r-matrix [31, 32]
N
X
P
r() =
where P =
eij eji
(3.3)

i,j=1
P is the glN permutation operator.
The fields i , j satisfy1 :
n
o n
o

i (x), j (y) = i (x), j (y) = 0,

n
o

i (x), j (y) = iij (x y).

(3.5)

From the zero curvature condition (1.6) the familiar classical equations of motion for the
generalized NLS model are entailed [25].
To obtain the local integrals of motion of the NLS model one has to expand the monodromy matrix in powers of 1 [25]. Let us consider the following ansatz for T as ||
T (x, y, ) = (I + W (x, )) exp[Z(x, y, )] (I + W (y, ))1
(3.6)
P
where W is off diagonal matrix i.e. W = i6=j Wij eij , and Z is purely diagonal Z =
PN
i=1 Zii eii . Also
(n)

(n)
X
X
Wij
Zii
, Wij =
.
(3.7)
Zii () =
n
n

n=1
n=1
1

The Poisson structure for the generalized NLS model is defined as:
n
o
X Z L  A
A
B
B 

A, B = i
dx
i (x) i (x) i (x) i (x)
L
i

(3.4)

(n)

Inserting the latter expressions (3.7) in (1.4) one may identify the coefficients Wij and
(n)
Zii . Notice that as i the only non negligible contribution from Z (n) comes from
(n)
the ZN N term, and is given by:
(n)
ZN N (L, L)

1 Z
X
N
= iLn,1 +
i=1

(n)

dx i (x) WiN (x).

(3.8)

(n)

It is thus sufficient to determine the coefficients WiN in order to extract the relevant local
integrals of motion. Indeed solving (1.4) one may easily obtain (see also [25] for more details
on the relevant computations):

(1)
(2)
WiN (x) = i i (x), WiN (x) = i (x),
also
X

(1)
(1)
(1)
(2)
(1)
WN j (x)Wji (x). (3.9)
WN i (x) = i i (x), WN i (x) = iWN i (x) +
i6=j, i,j6=N

Note that we choose to consider as a defect matrix the following:

L(; x0 ) = + P(x0 ),

P=

N
X

(3.10)

Pkl ekl

k,l=1

where Pkl satisfy the ultra-local glN classical algebra [25]:

n
o
Pij (x0 ), Pkl (x0 ) = Pil (x0 )jm Pkj (x0 )il .

3.1

(3.11)

The reflection algebra

To identify the time components of the Lax pairs associated to each local integral of motion
in powers of 1 :
we expand the expressions for V , V

V (, ) =

X
V(n) ()
n=1

(, ) =
V

(n)
X
V
()
n=1

(3.12)

Recall, that in the SP case only the odd integrals of motions are the conserved charges
(see e.g. [27]) . Computing the bulk V operators, associated to the first two non-trivial
integrals of motion i.e. number of particles and Hamiltonian, for the right and left theories
we recover the familiar bulk components:
V(1) = eN N ,

(3.13)

whereas V(3) relevant to the Hamiltonian are given in (3.1), (3.2) (, , ),

(2) , associated to the momentum is zero, given that the momentum is not a conserved
and V
quantity anymore.
8

To extract the desired sewing condition in the SP case we focus on the computation of
the operators on the defect point and impose the suitable analyticity condition from the left
and the right. Moreover, having this information at our disposal we will be able to obtain
operators associated to the
the associated equations of motion on the defect point. The V
first three orders of the expansion are given below:
(1) () = eN N
V
(2) () = 1 (A + A )
V
2

1
(3)

V () = 2 eN N + (A A ) + (B + B )
2
2

(3.14)

(we have divided all the results coming from (2.16) with an overall 21 factor); we also define:
X
X (1)
+(1)
A =
(PN j WN j )eN j +
WjN ejN
j6=N

A =
+

A =
+

A =

X

+(1)
iWN j

j6=N

j6=N

(PjN

+(1)
WjN )ejN

(PjN +

(1)
WjN )ejN

(PN j +

(1)
WN j )eN j

j6=N

+(1)
WN i Pij

+(1)
Pjl WlN

X

WN j eN j

WN j eN j

WjN ejN .

(1)

+(1)

j6=N

+(1)

(3.15)

j6=N

+(1)
PN N WN j

eN j + i

(1)

WjN ejN

j6=N

+(1)
iWjN


X (1)
+(1)
+ Pjl PlN + WjN PN N ejN + i
WN j eN j
2

l6=N
j6=N


X
X 
(1)
(1)
+(1)
(1)
+(1)
(1)
WN j PjN + WN j WjN eN N .
PjN WN i WjN WN i eji +

j6=N

PN N PN j +

i,j6=N

j6=N

j6=N

j6=N
i6=N


X
X 
(1)
+(1)
(1)
(1)
(1)
+(1)
PN j WjN WN j WjN eN N
WjN PN i WjN WN i eji

j6=N

j6=N

i,j6=N

(3.16)
Pij are associated to defect and are defined in (3.10), (3.11).
Similarly, we define B+ , B+ as:

X
X +(1)
X  (1)
(1)
(1)
Pjl WlN 2 PjN PN N WjN PN N ejN + i
WN j eN j
iWjN
B+ =
l6=N
j6=N


X 
X
+(1)
(1)
+(1)
+(1)
+(1)
(1)
PjN WN i + WjN WN i eji +
WN j PjN + WN j WjN eN N

j6=N

B+ =

i,j6=N

j6=N


X +(1)
X
X (1)
X  (1)
+(1)
WjN ejN
PN l Plj + PN N WN j eN j + i
WN l Plj + 2
iWN j +

j6=N

l6=N

l6=j

j6=N



X
X 
+(1)
(1)
+(1)
+(1)
+(1)
(1)
PN j WjN + WN j WjN eN N .
WjN PN i + WjN WN i eji +
j6=N

i,j6=N

(3.17)
Due to continuity requirements around the defect point (2.17) we obtain the following
sewing conditions:
+(1)

(1)

PN j = WN j WN j
+(1)
(1)
PjN = WjN WjN .

(3.18)

The conditions associated to the 3d order are given by



X
+(1)
+(1)
(1)
+(1)
Plj WN l PN N PN j
WN j WN j = i WN j PN N

(3.19)

l6=N

+(1)

WjN



X
X
+(1)
+(1)
Pjl PlN ,
Pjl WlN +
= i WjN PN N

(3.20)

(1)



X (1)
(1)
= i PN N WjN
WlN Pjl + PjN PN N

(3.21)

(1)



X +(1)
+(1)
WN l Plj PN N PN j .
= i PN N WN j

(3.22)

(1)

WjN

l6=N

l6=N

and
+(1)

WjN

+(1)

WN j

WjN

WN j

l6=N

l6=N

As expected the odd quantities are non-trivial, associated to the corresponding conserved
quantities i.e. number of particles, Hamiltonian etc, whereas the even ones are naturally
zero. Notice than (3.20), (3.21) and (3.19), (3.22) are compatible as one would expect
due to the gluing conditions (3.18). Moreover, the sewing conditions derived above are
compatible with the Hamiltonian action in accordance to the relevant generic proposition
proved in [19] for systems associated to the Yangian classical r-matrix.
The next natural step is to identify the equations of motion for the system. It is clear
that in the bulk the equations of motion associated to the Hamiltonian are given by the
familiar expressions for the vector NLS model:
N 1

X
i (x, t)
2 i (x, t)
i
=
+
2
|j (x, t)|2 i (x, t),
2
t
x
j=1

i {1, . . . , N 1},

x 6= x0 .
(3.23)

Recall that the Hamiltonian of the system was explicitly derived in [24] as:
I3 = 2i

x0

dx
0

N
1 
X
i=1

|i (x)|2

X
k

|k (x)|2
10

i (x)i (x)

+ 2i

X
i

i (x0 )i (x0 )

N
1 
X


X

|k+ (x)|2 + i+ (x)i+ (x) 2i

i+ (x0 )i+ (x0 )
x
i=1
k
X
X i (1)
X0
X
+(1)
2

PN i Pij PjN
WN i Pij PjN + 3
PN i Pij WjN +
2i
(j (0)(0)) + D3 + D3 +

2i

dx

|i+ (x)|2

i6=N,j

i6=N,j

j6=N,i

(3.24)
where D3 is defined as follows
D 3 = D3 D1 D2 +

D13
3

(3.25)

D1 = PN N
X
X +(1)
X
X +(1) (1)
(1)
+
D2 =
WN+j WjN

WN j PjN +
PN j WjN
WN j WjN
j6=N

D3 =

j6=N

+(1)
+(2)
WN j WjN

Xj6=N
(2)
PN j WjN +
+
j6=N

j6=N

+(2)
WN j PjN

j6=N

+(2)
+(1)
WN j WjN

j6=N
+(1)
+(1)
+(1)
WN i Wij WjN

i6=N i,j6=N
X +(2) (1)
Xj6=N +(1) +(1) (1)
WN j WjN
WN i Wij WjN
j6=N

i6=N,j6=N

+(1)
+(1)
WN i Wij PjN

+(1)

(2)

WN j WjN

Xj6=N +(1)
(1)
WN j Pij WjN .

i6=N,j6=N

(3.26)

i6=N,j

and
3 (W (n) ) = D3 ((1)n W (n) ).
D
ij
ij

(3.27)

However, special care is taken on the defect point (1.7), then the following time evolution
for the defect degrees of freedom is obtained:
1
1
P ij = (Bil+ + Bil+ )Plj Pil (Blj + Blj ).
2
2

(3.28)

It is worth noting that in the periodic case studied in [24] the sewing condtions were used
in order to eliminate the dependent terms in the zero curvature condition on the defect
point. In the present case however these terms disappear automatically due to the various
contributing terms emerging from the double monodromy matrix T . Comparing also the
equations of motion above with the ones extracted in the periodic case [24] we conclude
that are somehow double, due to the presence of the reflecting boundary conditions.

3.2

The twisted Yangian

Let us now briefly describe in this subsection the SNP case. This case is not particularly
inspiring in the sense that it does not provide any unexpected bulk behavior due to the
presence of the integrable boundaries. Nevertheless, it is worth studying it for completeness
11

as well as for showing that the presence of boundaries does not always leads to drastic effects
in the bulk. Moreover, this study further confirms the results presented in [24].
It is technically more tractable to consider the discrete vector NLS and then take the
suitable continuum limit. Admittedly, working directly on the continuum case it is much
more complicated mainly due to various contributions from the diagonal Z terms. The
discrete monodromy matrix in the presence of a single defect on the n-th cite:
0n () . . . L01 ()
T0 (L, 1; ) = L0L () . . . L

(3.29)

In general, we introduce the notation:

T0 (i, j; ) = L0i . . . L0j ()

(3.30)

i > j.

matrix acts on the nth site of the one dimensional

it is clear that in the case i > n > j the L
discrete system.
and consequently T satisfy the classical quadratic algebra (1.10). Where
Both L and L,
the L matrix of the discrete vector NLS model is given as
L() = (i + N) eN N +

N
1
X

ell +

l=1

Due to the quadratic algebra we obtain:

n

N
1
X

(l elN + l eN l ).

(3.31)

l=1

k , l = ikl
o
N, k = ik
n
o
N, k = ik .

(3.32)

The defect Lax operator is

n () = + P(n) .
L

(3.33)

the entries of the P matrix Pij satisfy the classical glN algebra:
o
n
(n)
(n)
(n)
(n)
Pij , Pkl = Pil jk Pjk il .

(3.34)

As already mentioned we focus on the SNP case that is we consider the representation
of the classical twisted Yangian (1.16). Expansion of trT (), T defined in (1.17), but now
T is the discrete monodromy matrix will lead to the local integrals of motion (see also [33]
for relevant results)
(2)
Id

N
1
X

l=1 j6=n,n1

(j+1) (j)
l
l

N
1
1 

X
X
N
(n+1) (n)
(n) (n1)
2

(N(j) )2
l
PlN + PN l l
+ i
l=1

12

j=1

N
1
X

(n+1) (n1)
l
l

l=1

N
1 
X

(L) (L)

l l

(1)

(1)

+ l l

l=1


.
(3.35)

By taking the continuum limit as (see also [19] and references therein)




(j)
(j)
l+ (x), l+ (x) ,
j > n, x > x0
l , l




(j)
(j)
l (x), l (x) ,
j < n, x < x0
l , l

(3.36)

we obtain the continuum vector NLS momentum [24]

I2 =

dx

x0

N
1 
X
i=1


X

i (x)i (x) i (x)i (x) +

i (x0 )i (x0 )
i

N
1 
X

i+ (x)i+ (x) i+ (x)i + (x)

x
i=1
X0
X

+
i (0)i (0) +
i (0)i (0) + 2D2

dx

i+ (x0 )i+ (x0 )

(3.37)
D2

D2 is defined as D2 = D2 21 , and D1 , D2 are given in (3.26). The charge I2 provides

indeed the momentum of the system:
P =

I2
.
2i

(3.38)

In general, as discussed in [24] in this case only the even charges survive.
Let us now formulate the corresponding discrete Lax pairs. In the discrete case the zero
curvature condition is expressed as (see also [25]):
dLj ()
= Aj+1(, ) Lj () Lj () Aj (, ),
dt
n ()
dL
n () L
n () An (, ).
= An+1 (, ) L
dt

j 6= n,
(3.39)

As in the continuum case the time components of the Lax pair can be identified:


(j)
(j)
Aj (, ) = t1 () tra Mab (, ) + Mab (, )
and
Aj (, ) =

X Aj(m) ()
m

(3.40)

(3.41)

I):
We also define the discrete quantities (we have considered for simplicity K = K
(j)

Mab (, ) = Ta (N, j; ) rab ( ) Ta (j 1, 1; ) Tata (N, 1; ),

13

j 6= n, n + 1. (3.42)

Around the defect point we define:

(n)
an () rab ( ) Ta (n 1, 1; ) T ta (N, 1; )
Mab (, ) = Ta (L, n + 1; ) L
a
(n+1)

Mab (, ) = Ta (L, n + 1; ) rab ( ) Lan () Ta (n 1, 1; ) Tata (N, 1; )

(3.43)

Mab (, ) = Mtaba (, ).

(3.44)

P A(m)
Expansion in powers of 1 of Aj = m jm will provide the time component of the Lax
pairs associated to each integral of motion. The first non trivial is the one associated to
the momentum:

X  (j)
(2)
(j)
Aj () = i eN N +
l eN l + l elN , j 6= n, n + 1
(3.45)
l6=N

and around the defect point we derive:


X (n1)
X  (n+1)
(n)
e
+

+
iP
l
elN

A(2)
()
=
i
e
+
Nl
NN
n
l
Nl
l6=N

(2)
An+1 ()

= i eN N

l6=N


X (n+1)
X  (n1)
(n)
+
l
+ iPlN elN .
l
eN l +
l6=N

(3.46)

l6=N

We are now taking the continuum limit exploiting (3.36) and

(2)

Aj V+(2) (x),
(2)
Aj V(2) (x),
(2)
+(2) (x0 ),
An+1 V
(2) (x0 ),
A(2)
n V

j > n + 1, x > x0
j < n, x < x0
(3.47)

we end up with the following expressions on the continuum time components of the Lax
pair associated to the momentum:

X
V(2) (; x) = i eN N +
l (x) eN l + l (x) elN
(3.48)
l6=N

and on the defect point:

(2) (; x0 ) = i eN N +
V


X +
l (x0 ) + iPN l (x0 ) eN l + l (x0 ) elN

l6=N

+(2) (; x0 ) = i eN N
V



X +

l (x0 )eN l + l (x0 ) + iPlN (x0 ) elN . (3.49)

+
l6=N

Then continuity conditions imposed around the defect point (2.17) lead to the sewing
conditions extracted in the periodic case [24] i.e.

PN l = i(l+ l )
14

PN l = i(l+ l ).

(3.50)

That is the presence of non trivial integrable boundary conditions in the SP case does not
really affect the behavior of the defect as expected. With this we conclude the derivation
of Lax pairs in the simultaneous presence of boundaries and defects for vector NLS model.

Discussion

The Lax pairs in the simultaneous presence of point like defects and integrable boundaries
were constructed. Based on this construction the vector NLS model was investigated. We
have used the vector NLS model as a prototype in order to exemplify our results as well as
to make contact with recent relevant results found in [24]. The vector NLS model displays
a highly non-trivial behavior compared to the single NLS due to the fact that it is ruled by
the high rank glN algebra, and this allows the existence of two distinct types of boundary
conditions associated to the reflection algebra and the twisted Yangian as opposed to the
self-dual sl2 case, where basically the two algebras coincide. It is worth noting that we
have basically considered K I for simplicity but without loss of generality. Another
choice of K matrices would have no effect whatsoever on the bulk Lax pair or the Lax pairs
associated to the defect point and this is one of the key points in the present derivation,
but it would only add some extra boundary terms in the relevant local integrals of motion
(see also [24]). Thus what is important in the present description is in fact the choice of the
underlying classical algebra whether this is the reflection algebra or the twisted Yangian.
Although this construction is the first significant step towards understanding the behavior of such systems there are still various points in this context to further explore. For
instance the solution of the relevant equations of motion exploiting the integrable sewing
conditions across the defect point is a pertinent issue to address. It would be also very
interesting to identify the classical scattering data of the solitonic solutions with the defect
derived via the inverse scattering method and compare with the corresponding quantum
results derived in numerous works (see e.g. [3, 7, 10, 22, 23]). The relevant solitonic solutions in this context can be extracted via the associated Backlund transformations, which
in turn can be derived through the solution of the inverse scattering [35]; this is a particulary interesting issue especially with regard to the defect point, and its connection with the
integrable sewing conditions. Another significant question worth exploring would be the
identification of the explicit form of the Jost functions in the presence of defects and the
formulation of the associated Gelfand-Levitan-Marchenko equations at both classical and
quantum level [25, 36].
15

This line of attack can be implemented into various classical systems and connection with
relevant results from the quantum point of view identified via the Bethe ansatz formulation
(1)
(see e.g. [22, 23]) can be made. Finally, a similar description in the context of An affine
Toda field theories would also yield highly non-trivial findings on the behavior of the system.
The boundary affine Toda field theories associated to the twisted Yangian and the reflection
algebras was studied in [34] and highly non trivial boundary behavior was exhibited in the
case of reflecting boundaries. Similarly, intricate phenomena associated to the defects are
expected in the case of the SP boundary conditions. Hopefully, all the aforementioned
issues will be addressed in future investigations.

References
[1] G. Delfino, G. Mussardo and P. Simonetti, Phys. Lett. B328 (1994) 123,
hep-th/9403049.
[2] G. Delfino, G. Mussardo and P. Simonetti, Nucl. Phys. B432 (1994) 518,
hep-th/9409076.
[3] R. Konik and A. LeClair, Nucl. Phys. B538 (1999) 587; hep-th/9793985.
[4] P. Bowcock, E. Corrigan and C. Zambon, JHEP 01 (2004) 056, hep-th/0401020
[5] P. Bowcock, E. Corrigan and C. Zambon, JHEP 08 (2005) 023, hep-th/0506169.
[6] E. Corrigan, C. Zambon, Nonlinearity 19 (2006) 1447, nlin/0512038.
[7] E. Corrigan and C. Zambon, JHEP 07 (2007) 001, arXiv:0705.1066 [hep-th].
[8] E. Corrigan and C. Zambon, J. Phys. A42 (2009) 475203, arXiv:0908.3126
[hep-th].
[9] E. Corrigan and C. Zambon, Nonlinearity 19 (2006) 1447, nlin/0512038.
[10] E. Corrigan and C. Zambon, J. Phys. A43 (2010) 345201, arXiv:1006.0939
[hep-th].
[11] I. Habibullin and A. Kundu, Nucl. Phys. B795 (2008) 549, arXiv:0709.4611
[hep-th].
[12] V. Caudrelier, IJGMMP vol.5, No. 7 (2008) 1085, arXiv:0704.2326 [math-ph].

16

[13] F. Nemes, Int. J. Mod. Phys. A25 (2010) 4493; arXiv:0909.3268 [hep-th].
[14] R. Weston, arXiv:1006.1555 [math-ph].
[15] M. Mintchev, E. Ragoucy and P. Sorba, Phys. Lett. B547 (2002) 313,
hep-th/0209052;
M. Mintchev, E. Ragoucy and P. Sorba, J. Phys. A36 (2003) 10407, hep-th/0303187.
[16] V. Caudrelier, M. Mintchev and E. Ragoucy, J. Phys. A37 (2004) L367,
hep-th/0404144.
[17] A.R. Aguirre, T.R. Araujo, J.F. Gomes and A.H. Zimerman, JHEP 12 (2011) 56,
arXiv:1110.1589 [hep-th].
[18] A.R. Aguirre, J. Phys. A 45 (2012) 205205, arXiv:1111.5249 [math-ph].
[19] A. Doikou, Nucl. Phys. B854 (2012) 153, arXiv:1106.1602, [hep-th];
J. Avan and A. Doikou, JHEP 01 (2012) 040, arXiv:1110.4728 [hep-th].
[20] J. Avan and A. Doikou, JHEP 11 (2012) 008, arXiv:1205.1661 [hep-th].
[21] A. Doikou and N. Karaiskos, Nucl. Phys. B867 (2013) 872, arXiv:1207.5503
[hep-th].
[22] A. Doikou and N. Karaiskos, JHEP 02 (2013) 142, arXiv:1212.0195 [math-ph].
[23] A. Doikou, JHEP 08 (2013) 103, arXiv:1304.5901 [hep-th];
A. Doikou, Nucl. Phys. B877 (2013) 885, arXiv:1307.2752 [math-ph];
A. Doikou, arXiv:1308.1790 [math-ph].
[24] A. Doikou, Nucl. Phys. B884 (2014) 142, arXiv:1312.4786.
[25] L.D. Faddeev and L.A. Takhtakajan, Hamiltonian Methods in the Theory of Solitons,
(1987) Springer-Verlag.
[26] M.A. Semenov-Tjan-Shanskii, Funct. Anal. Appl. 17 (1983), 259.
[27] E.K. Sklyanin, Funct. Anal. Appl. 21 (1987) 164;
E.K. Sklyanin, J. Phys. A21 (1988) 2375.
[28] G.I. Olshanski, Twisted Yangians and innite-dimensional classical Lie algebras in
Quantum Groups (P.P. Kulish, Ed.), Lecture notes in Math. 1510, Springer (1992)
103.
17

[29] A. Doikou, J. Phys. A33 (2000) 8797, hep-th/0006197.

[30] J. Avan and A. Doikou, Nucl. Phys. B800 (2008) 591, arXiv:0710.1538 [hep-th];
J. Avan and A. Doikou, Nucl. Phys. B821 (2009) 481, arXiv:0809.2734 [hep-th].
[31] E. Sklyanin, J. Sov. Math. 19 (1982) 1564.
[32] C.N. Yang, Phys. Rev. Lett. 19 (1967) 1312.
[33] A. Doikou, D. Fioravanti and F. Ravanini, Nucl. Phys. B790 (2008) 465,
arXiv:0706.1515 [hep-th].
[34] A. Doikou, JHEP 0805 (2008) 091, arXiv:0803.0943 [hep-th];
J. Avan and A. Doikou, Nucl. Phys. B821 (2009) 481, arXiv:0809.2734 [hep-th].
[35] K. Konno and M. Wadati, Progr. Theor. Phys. 53 (1976) 1652.
[36] K. Sogo and M. Wadati, Progr. Theor. Phys. 69 (1983) 431.

18