coefﬁcient is given by
c
ml
R
R
2
’
m
r;
l
r;
;0
. Since
’
m
and
l
belong to representations of
C
n
and
O
2
,respectively, they both properly transform under a discreterotation of order
n
:
’
m
r;
2
=n
e
im
2
=n
’
m
r;
and
l
r;
2
=n
e
il
2
=n
l
r;
. Thus, by performing the change of variable
!
2
=n
in the deﬁnition of
c
ml
one arrives at the symmetry relation
c
ml
exp
i
l
ÿ
m
2
=n
c
ml
. The
c
ml
coefﬁcient is thenzero unless the following condition is fulﬁlled:
l
ÿ
m
kn
k
2
Z
;
where
j
m
j
n
2
:
(1)The
m
representation of
C
n
is thus excited by initial ﬁeldshaving angular momenta
l
m;m
n;m
2
n;
...
. In an
O
2

O
2
interface each representation of angular momentum
m
in the second medium is excited by one, andonly one, angular momentum component
l
arising from theﬁrst medium and verifying
l
m
. Contrarily, in an
O
2

C
n
interface the symmetry restriction (1) impliesthat, due to the cutoff
j
m
j
n=
2
in the
C
n
medium, thereare inﬁnite angular momenta
l
that can excite a givenrepresentation of index
m
in the second medium.Thus, when the incident ﬁeld carries an angular momentum
l
that overcomes the limiting value for pseudoangularmomentum inthe secondmedium, it will excite awave thatwill propagate with
different
constant pseudoangular momentum
m
given by the ‘‘pass rule’’ (1). This result is validfor waves verifying an equation of the type
L
j
j
ÿ
i@=@z
, linear or nonlinear, stationary or evolving. Aparticularly interesting situation occurs when the incidentﬁeld is a vortex ﬁeld of the
O
2
nonlinear medium. Thisvortex ﬁeld
v
l
is a stationary solution of the evolutionequation with welldeﬁned angular momentum
l
Þ
0
:
L
j
v
l
j
v
l
ÿ
v
l
. We consider here individual ‘‘canonical’’ vortices with a single phase singularity (i.e.,with only one point in which
v
l
0
):
v
l
r;;z
e
il
f
v
l
r
e
ÿ
iz
[13]. The vorticity or topological charge of such solutions will be given by the circulation of its phasegradient around the singularity
v
1
=
2
H
r
arg
v
l
d
r
, which equals the angular momentum for canonical vortices
v
l
. On the other hand, the propagatingwave
m
with pseudoangular momentum
m
excited by
v
l
will evolve in the
C
n
medium. There are different optionsfor the asymptotic states of
m
when
z
!1
. One possibility is that this wave asymptotically tends to a stationarysolution
m
!
z
!1
v
m
r;;z
e
im
g
m
r;
e
ÿ
i
0
z
in therepresentation of
C
n
given by the conserved pseudoangularmomentum
m
. If
v
m
has a single phase singularity then itwill have the structure of an individual canonical discretesymmetry vortex, its vorticity or topological charge
v
0
being directly given by
m
:
v
0
m
[12]. The formation inthe
C
n
medium of an asymptotic stationary state in theform of a discretesymmetry vortex is a dynamical issuethat depends on the structural parameters of the secondmedium as well as on the characteristics of the input vortexﬁeld
v
l
. If dynamics allows the stabilization of thediscretesymmetry vortex solution, the
O
2

C
n
interfacewill realize the mapping of an
O
2
vortex with charge
v
l
(exceeding the limiting value for pseudoangular momentum) into a
C
n
vortex with charge
v
0
m
Þ
0
, such that
v
0
<v
. The pass rule for pseudoangular momentum (1)becomes a pass rule relating input and output vorticities:
v
ÿ
v
0
kn
k
2
Z
;
(2)where
v
0
presents a cutoff in terms of
n
given by:
j
v
0
j
<n=
2
(even
n
) and
j
v
0
j
n
ÿ
1
=
2
(odd
n
) [12]. Note that
m
n=
2
solutions are not vortices but nodal or dipolemode solitons [12]. Wewill refer to the process of mappingan individual vortex into another with different topologicalcharge as ‘‘vortex transmutation.’’We will provide now a physical example of a system inwhich the phenomenon of vortex transmutation takesplace. It is an optical interface separating two 2D dielectricmedia with Kerr nonlinearity, these two media being ahomogeneous medium and a 2D square optical lattice.This system is equivalent to a 2D BEC in which a periodicpotential is abruptly switched on. They constitute an
O
2

C
4
interface given by the following equation:
r
2
t
ÿ
V
x;y;z
z
j
j
2
ÿ
i@@z;
(3)in which
r
t
is the 2D gradient operator and
V
x;y;z
V
0
z
V
1
x;y
ÿ
V
0
where
z
is the step functionand
V
0
and
V
1
x;y
V
1
cos
2
2
x
cos
2
2
y
(
V
1
isthe potential strength and
is the lattice spatial period)deﬁne the refractive index proﬁle of the homogeneousmedium and of the 2D optical lattice, respectively:
V
0
ÿ
n
2
ÿ
n
20
and
V
1
x
ÿ
n
2
x
ÿ
n
20
,
n
0
being a reference refractive index introduced by the slowly varyingenvelope approximation. The nonlinear function
z
1
ÿ
z
permits the nonlinear response of thesystem to be different in the two media. All distancesappearing in Eq. (3) are normalized and dimensionless(
x
k
0
x
0
,
z
k
0
z
0
). In order to solve the evolution problem in this system we solve ﬁrst Eq. (3) for
z <
0
, whichbecomes an ordinary nonlinear Schro¨dinger equation(NLSE) for a homogeneous medium. Since our aim is toevidence the phenomenon of vortex transmutation we areinterested in ﬁnding canonical vortex solitons of differentcharges in the homogeneous
O
2
medium:
v
l
x
;z
e
il
f
v
l
r
e
ÿ
iz
. This can be done by standard methods. Ata given value of
l
, a family of
O
2
vortices are foundcharacterized by their power
P
l
R
R
2
j
v
l
j
2
and theirpropagation constant
, which are related through therelation
P
l
. In the case of a Kerr nonlinearity,
behaves as a scaling parameter and
P
l
is
independent [1].Oncethevortexsolution
v
l
is found,it istaken asan initialsolution for propagating it in the 2D optical lattice (
z >
0
):
x
;
0
v
l
x
;
0
e
il
f
v
l
r
. Thus we solve Eq. (3) for
z>
0
, which becomes a NLSE with the periodic potentialPRL
95,
123901 (2005)PHYSICAL REVIEW LETTERS
week ending16 SEPTEMBER 2005
1239012