SU (N) Skyrmions From Harmonic Maps

SU (N) skyrmions from harmonic maps S 2 Gr (2,N)
H. J. Wospakrik and W. J. Zakrzewski

Citation: Journal of Mathematical Physics 43, 1856 (2002); doi: 10.1063/1.1452303
View online: http://dx.doi.org/10.1063/1.1452303
View Table of Contents: http://scitation.aip.org/content/aip/journal/jmp/43/4?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Skyrmions, rational maps, and scaling identities
J. Math. Phys. 52, 033509 (2011); 10.1063/1.3566530
Half-Skyrmions and spike-vortex solutions of two-component nonlinear Schrdinger systems
J. Math. Phys. 48, 053518 (2007); 10.1063/1.2722559
SU(N) skyrmions and harmonic maps
J. Math. Phys. 40, 6353 (1999); 10.1063/1.533097
Spherically symmetric solutions of the SU (N) Skyrme models
J. Math. Phys. 40, 6223 (1999); 10.1063/1.533088
Dyon-Skyrmion lumps
J. Math. Phys. 40, 1136 (1999); 10.1063/1.532793
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
JOURNAL OF MATHEMATICAL PHYSICS
VOLUME 43, NUMBER 4
APRIL 2002
SU N skyrmions from harmonic maps S 2 \Gr2,N

H. J. Wospakrika) and W. J. Zakrzewskib)
Department of Mathematical Sciences, University of Durham,
Durham DH1 3LE, United Kingdom
Received 18 June 2001; accepted for publication 3 January 2002

We construct multiskyrmion fields of the SU(N) Skyrme models by using harmonic maps of S 2 to the Grassmannian Gr(2,N), which we express in terms of
rank-2 projectors. Within this construction we derive some approximate spherically
symmetric solutions of SU(N) Skyrme models and show that their energies are
marginally higher than those for the rank-one cases. We also discuss the possibility
of generating exact spherically symmetric solutions using this construction. In particular, we present arguments which suggest that the only solutions obtained in this
way are embeddings. 2002 American Institute of Physics.
DOI: 10.1063/1.1452303
I. INTRODUCTION
The SU(N) Skyrme models are minimal generalizations of the SU(N) chiral models in (3
1)-dimensional spacetime which possess static finite energy solutions called multiskyrmions.
They are examples of topological solitons in three spatial dimensions. Recently, they have received a great deal of attention, especially for the N2 case, since, it has been argued1,2 that they
describe, at a classical level, low energy states of nuclei.
The SU(N) Skyrme models are described by SU(N) group valued functions U of spacetime
coordinates x (x ,t). Their dynamics is determined by the Lagrangian density
L
F2
1
F2 2

Tr L L
M Tr U 1 U2I ,
2 Tr L ,L L ,L
16
32a
16
where
L U 1 U
are the SU(N) algebra valued currents, F189 MeV is the pion decay constant, and a is a
dimensional constant. The last term in 1 describes the mass term where M is the pion mass.
Multiskyrmions are stationary points minima or saddle points of the corresponding static
energy functional. The requirement of the finiteness of the multiskyrmions energy imposes the
condition that the field U(x ,t) goes to a constant matrix U 0 at spatial infinity. As by a global
SU(N) transformation, this U 0 can be brought to the identity matrix I, so without the loss of
generality we can impose the following boundary condition on U:UI as x .
As this boundary condition is effectively a mapping from S 3 SU(N) we see that each
multiskyrmion configuration can be classified by an integer valued winding number
1
24 2
R3
3
.
i jk Tr L i L j L k d x
Electronic mail: H.J.Wospakrik@durham.ac.uk

Electronic mail: W.J.Zakrzewski@durham.ac.uk
0022-2488/2002/43(4)/1856/19/$19.00
1856
2002 American Institute of Physics
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
SU( N ) skyrmions from harmonic maps
J. Math. Phys., Vol. 43, No. 4, April 2002
1857
This number is a topological invariant, i.e., an element of the third homotopy group
3 (SU(N))Z, which classifies the solitonic sectors of the model. Following Skyrme1 and
Witten,2,3 B is identified with the baryon number of the multiskyrmion configuration.
In what follows we express the energy in the same units as the baryon number B. This is
achieved by rescaling the spatial coordinates4,5 x 2x /aF and by taking F/4a1/12 2 . In these
units the energy functional is given by
E
1
12 2
R3
m 2
1
1
Tr L 2i Tr L i ,L j 2
Tr U 1 U2I d3 x ,
2
16
2
where m 2M /aF. From 4 we find the equation for the stationary points:
m 2
1
i L i L j , L i ,L j
UU 1 0.
4
2
The first exact solution of 5, in the SU2 case, was found by Skyrme.1 This solution, also
known as a hedgehog solution, describes a spherically symmetric energy lump and has the baryon
number B1. Since Wittens suggestion that the Skyrme model arises in the large N c limit of
QCD,2 most of the studies involving SU(N) for N2 considered configurations which were
embeddings of the SU2 fields.
The first nonembedding solution, for N3, was presented by Balachandran et al.,6 who found
an SO3 subgroup soliton which has baryon number B2. Another configuration, which has a
large strangeness content was then found by Kopeliovich et al.7
A more systematic study, using the harmonic map ansatz method, for any N, was then carried
out by Ioannidou et al.8,9 This method enabled the construction of new static spherically symmetric solutions. The method was, in fact, a generalization of the rational map approach of Houghton
et al.10 In their seminal work,10 Houghton et al. have shown how to use rational maps of S 2
S 2 to construct field configurations for the SU2 Skyrme model which have arbitrary baryon
number B and, for low values of B, are close to the exact solutions of the model. In Ref. 9,
Ioannidou et al. took the SU2 ansatz of Houghton et al. and rewrote it in terms of a projector of
S 2 C P 1 and then generalized it to more general projectors of S 2 C P N1 of rank-one.
In this paper we generalize the method of Ioannidou et al. further by considering projectors of
S 2 into the Grassmannian Gr(2,N), i.e., using projectors of higher rank we concentrate our
attention on the rank-2 projectors. We find that, in contradistinction to the rank-1 case, in which
exact spherically symmetric solutions can be found numerically by using the Veronese sequence
of N mutually orthogonal vector fields in C P N1 , such a construction is now more involved.
In Sec. II, we discuss the harmonic maps of R 2 into the Grassmannian Gr(n,N) in terms of the
(NN) projectors of rank-n, with 1nN and then present our generalization of the harmonic
map ansatz. In Sec. III we rewrite various quantities of the SU(N) Skyrme models in spherical
polar coordinates and then, using the harmonic map ansatz, reexpress them in terms of rank-n
projectors. Next, in Sec. IV, we derive the stationary equations of the approximate energy functional which we then use to study various field configurations for the N3, 4, and 5 cases using
one and two rank-2 projectors in Secs. V and VI, respectively. Then, in Sec. VII, we formulate a
condition on the sequence of N mutually orthogonal (Nn) matrix fields in Gr(n,N) which
would give exact spherically symmetric solutions of the SU(N) Skyrme models. In Secs. VIII and
IX, we analyze this condition further, consider some special configurations, and present arguments
suggesting that its only solutions are embeddings.
II. HARMONIC MAPS INTO GRASSMANN MANIFOLDS
Let C be a complex variable and M k M k ( , ), k0,1,..., where (N1), be a set

of (1) mutually orthogonal (Nn) matrices (nN), i.e.,
M k M l M k 2 kl ,
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1858
where kl is the Kroneckers delta, and where

M k 2 M k M k
are (nn) nonsingular matrices. Then the corresponding projector P k (n) onto each matrix M k is
given by
P k n M k M k 2 M k .
Clearly, Tr P k (n)Tr I n n, where I n is an (nn) unit matrix which means that each projector
P k (n) has rank-n. From 8, we see that the projectors P k (n) are mutually orthogonal, i.e.,
P k (n) P l (n) kl P l (n), and are Hermitian, i.e., P k (n) P k (n), as by definition M k are mutually
orthogonal and, by construction 7, M k 2 are Hermitian.
In the following we want to present a generalized harmonic map ansatz. To do this we use a
sequence of mutually orthogonal matrices (M 0 ,M 1 ,...,M ) obtained from a sequence of holomorphic analytic matrices (M , M ,..., M ), M 0, via the GramSchmidt orthogonalization
process.
We can do this using the operator P Ref. 11 defined by its action on any matrix M
C Nn as
P M M M M 2 M M .
Then we have
M 0 M ,
M 1 P M ,
...,
k
k1
M k P
M P P
M ,
...,
M P
M.
10
With this construction the following properties of the matrices M k hold when M 0 is
holomorphic:11
M k M k1 M k1 2 M k 2 ,
11
M k M k 2 M k1 M k 2 .
12
Geometrically, the matrices M k define harmonic maps R 2 Gr(n,N), i.e., in terms of the
projectors P k (n), where Gr(n,N) is the Grassmann manifold of n-dimensional planes in C N . 11
For rank-2 projectors which we are considering in this paper, the matrix M k is given by
M k M k1 ,M k2 ,
13
where M k1 and M k2 are two N-component column vector fields and M k 2 is a (22) matrix.
Using this column vector notation the entries of the projector P k (2) are given by
P k 2 ab
1
2
2
2
k1 b M k 11
k2 b M k 12
k2 b
M k 22
M k1 a M
M k2 a M
M k1 a M
Dk
2
k1 b ,
M k 21
M k2 a M
14
k j ) a , j1,2 is the complex conjugate of (M k j ) a , and where

where a,b1,2,...,N, (M
D k Det M k 2 .
15
Clearly, Tr P k (2)2. Furthermore, if we let

P k j 1 M k j M k j 2 M k j ,
16
then from 14 it follows that
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1859
P k 2 P k1 1 P k2 1 ,
17
P k2 1 M
k2 M
k2 2 M
k2
,
18
k2 I P k1 1 M k2 ,
M
19
where
with

which is orthogonal to M k1 , i.e., M k1
M k2 0. Thus, we see from 17 that each projector of
Gr(2,N) is really a sum of two mutually orthogonal rank-1 projectors.
Note that for some cases this construction does not work. To see this take the case where the
initial (N2) matrix M 0 is chosen to be given by
M 0 M 01 , M 01 ,
20
i.e., M 02 is a derivative of M 01 . Then it follows from 10 that

M 1 0,M 12 .
21
Thus, in this special case, M 1 2 is singular, and so the projector P 1 (2) does not exist.
Let now be related to the , , which locally parameterize S 2 , via tan(/2)e i . Then the
matrices M k define harmonic maps S 2 Gr(n,N). Taking our sequence of projectors P k (n)
P k in 8 we can now formulate our generalized harmonic map ansatz for the SUN Skyrme
fields in three dimensions. Namely, we take the matrix USU(N) of the form
Uexp i
k0
g k P k
nI
N
exp ni
k0
g k /N
l0
AlPl ,
22
where g k g k (r) for k0,...,1 are the profile functions and

A k e ig k 1 .
23
The profile functions g k (r) are required to satisfy the boundary conditions: g k ()0 and
g k (0)2 , where 0 or 1.
III. SUN MODEL-FOR THE GENERALIZED HARMONIC MAP ANSATZ
In this section, we first rewrite the energy 4 and the topological charge 3 in the spherical
polar coordinates (r, , ). Later we introduce the harmonic map ansatzso it is convenient to
replace the angular coordinates by the complex or holomorphic coordinates ( , ).
A. Energy
The energy 4, when written in the spherical holomorphic coordinates (r, , ), takes the
form
E
i
12 2
d d dr r 2 Tr
1
1
1
2
L ,L L r ,L
2 2 L r 2 L L
r
4r 2 r
1
1 2 2
L ,L 2 ,
16r 4
24
where, for simplicity, we have put the pion meson mass m 0.
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1860
With the matrix USU(N) given by the harmonic map ansatz 22, the currents L a s in 2
take the following form:
1
L r i
g k
k0
P k
nI
,
N
25
where g k dg k /dr and
L I
k0
ig k
1 P k
l0
e 1 P l
ig l
k0
e i g k g k1 1 M k1 M k 2 M k ,
26
where by definition g 0, and that L (L ) .

Using the expressions for L r and L in 25 and 26, respectively, we find that the traces in
the energy functional 24 become
Tr L r2
n2
N

1
k0
g k
k0
g 2k ,
27
Tr L L 2
k0
S k Tr M k1 2 M k 2 ,
28
g k g k1 2 S k Tr M k1 2 M k 2 ,
29
Tr L r ,L L r ,L 2
k0
Tr L ,L 2 8
k0
S 2k Tr M k1 2 M k 2 2 S k S k1 Tr M k1 2 M k1 2 ,
30
where by definition S 1 0 and for k0,...,1:

S k 1cos g k g k1 .
31
B. Topological charge
The topological charge 3 in the spherical holomorphic coordinate system takes the form
B
1
82
dr d d Tr L r L ,L .
32
Using the expression for L in 26 we find that

1
L ,L 2
k0
S k M k M k 2 M k1 2 M k 2 M k M k1 M k 2 M k1
,
33
and so, using this commutator and L r as given in 25, the topological charge 32 becomes
1
B
2
dr
k0
g k g k1 1cos g k g k1 Nk ,
34
where
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
Nk
i
2
d d Tr M k1 2 M k 2 .
1861
35
As g k ()0, we finally obtain

1
1
N g 0 g k1 0 ,
B
2 k0 k k
36
and so we see that the only contributions to the topological charge comes from g k (0).
In Appendix A we show that if D k Det M k 2 is nonsingular in the whole complex plane,
then Nk obey the recurrence relations
Nk Nk1 k ,
37
where k is the highest degree of 2 in D k , i.e.,

lim D k 2 k ,
38
and where by definition N1 0.

Thus, if we know k then we can determine Nk . In fact, if
H 0 M 01M 02M 0n
39
is the exterior product of the column vectors of M 0 which form the n-dimensional subspace of C N ,
then
0 deg H 0 ,
40
i.e., is the highest degree of in H 0 .
IV. APPROXIMATE FORMULATIONS
In this section, we derive field equations for the profile functions g k from the energy 24 into
which we have inserted the expressions of the generalized harmonic map ansatz.
To do this we, first of all, take a holomorphic matrix M 0 M ( ) and then evaluate the
sequence (M 1 ,M 2 ,...,M ) using the formulation of Sec. II. Then we compute the angular integrals Nk in 35,
Ik
i
2
d d 1 2 2 Tr M k1 2 M k 2 2 ,
41
and
Hk
i
2
d d 1 2 2 Tr M k1 2 M k1 2
42
for k0,...,1, where by definition H0 0.

In terms of Nk , Ik , and Hk , the energy 24, for our ansatz 22, reduces to
E
1
6
dr r 2
1
n2
N
1
k0
g k
1
I S 2 Hk S k S k1 .
2r 2 k0 k k
k0
g 2k 2
Nk 1 4 g k g k1 2 S k
k0
43
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1862
Introducing
F k g k g k1 ,
44
with g 0 we find that, in terms of F k , the energy integral 43 becomes

1
E
6
n2
dr r
N
k0
k1 F k
lk
k0
F l
N 4F 2k 1cos F k
2 k0 k
1
2
I 1cos F k 2 Hk 1cos F k 1cos F k1 ,
2r k0 k
45
and the topological charge 36 becomes

1
1
N F 0 .
2 k0 k k
46
To derive the equations for the profile functions g k from 45 we note that
E
F l
r
2
2n 2 l1
i0
i1 F i 2n

ji
i0
F j
Nl F l 1cos F l ,
47
where E denotes the integrand of E.

Thus our field equations for the functions F i and so also for g i are given by
2n 2 l1
N
i0
i1 F i 2n
i0 ji
F j
l
2
2n 2 l1
r
N
Hl 1cos F l1 Hl1 1cos F l1
r2
r2
i0
i1 F i 2n
0.
1
N F 1cos F l
r2 l l
i0 ji
F j
sin F l
2Il 1cos F l
Nl F 2l 4
2
2r
r2
48
Now a question arises: what is the best choice of the initial matrix M 0 that would yield low
energy field configurations which, hopefully, are close to the exact solutions of the full equations
of the model, i.e., Eq. 5? To answer this question we note that each Nk is, in fact, the energy of
the Grassmannian Gr(n,N) models. Thus in order to have minimal Nk , and so also energy to be
close to the exact multiskyrmion energy, the entries of the matrix M 0 must be polynomials in
.11,12
In the following sections we make this choice for M 0 and consider various fields for the
SU3, SU4, and SU5 cases with one and two rank-2 projector approximations. Our choices are
dictated by simplicity and they lead to energy density distributions which are spherically symmetric. Moreover, looking at the general case we see that they also correspond to setting some g i
functions in 22 equal to zero.
V. ONE PROJECTOR APPROXIMATIONS
In this case, we take only one profile function g 0 F 0 F, i.e., 1, and so the approximate
energy 45 with the mass term reduces to
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1
6

dr
1863
1 2
I 1cos F 2
A N n r 2 N0 1cos F 2N0 1cos F 0
F
2
2r 2
m 2 Nn 1cos
nF
N
n 1cos
Nn F
N
49
where
A N n
2n Nn
,
N
50
and the field equation for the approximate function F becomes
F A N n
sin F
N0 1cos F
2A N n
2I0 1cos F
2
F
2
2 N0 F 4
r
r
2r
r2
m 2 A N n sin
nF
Nn F
sin
N
N
0.
51
In the following we restrict our attention to the rank-2 case only, i.e., n2. To solve 51 we
impose the boundary conditions: F(0)2 and F()0. Thus, the baryon number of these
configurations is BN0 . Finally, we compare the approximate energies of each subcase with the
corresponding energies of the one rank-1 projector approximations.8
As a first attempt to solve 51, let us take the initial matrix M 0 (M 01 ,M 02), where both
column vectors are given by the following Veronese type form:
M 01 1,C N1
,..., C N1
k ,..., N1 T ,
1
k
M 02 M 01 ,
52
denotes the binomial coefficients, and the superscript T denotes transposition.

where C N1
k
This special form of M 0 enables us to express the determinant D 0 of M 0 2 in the following
closed form:
D 0 N1 1 2 2 N2 .
53
Then from 38 we conclude that N0 2(N2), which is consistent with 40. This result can be
verified explicitly using definition 35 with the help of the recurrence relations A4, which yields
Tr M 1 2 M 0 2
2 N2
.
1 2 2
54
For this case, according to 21, D 1 0, and so according to formula B3, I0 N20 4(N2) 2 .
In the SU3 case, we find that A 3 (2) 34 , N0 2, and so I0 4, which all coincide with the
values of the corresponding quantities in the rank-1 projector approximation.8 Thus, we conclude
that their energies also coincide, i.e., E2.444 04 (m 0). This equality holds for m 0 as
well. This is to be expected as in this case our rank-2 projector is really a sum of two rank-1
projectors constructed from the first two vectors of the Veronese sequence.
It is clear that, for SU(N) with N3 we have: A N (2)A N (1), N0 (N1), and I0 (N
1) 2 . Thus, this Veronese type configuration for N3 will lead to energies higher in comparison
with those for the rank-1 projector approximations.8
So, in the following, we look only at M 0 with N0 (N1) and I0 (N1) 2 in order to
compensate A N (2)A N (1) in the energy integral 49. More specifically, we look at the following
two subcases for which the determinant of M 0 2 is of the form,
D 0 c 1 2 N1 ,
55
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1864
or
D 0 c 1 2 N3 1 2 2 ,
56
where c and are some constants. For the subcase 56 we choose the column vector M 01 to have
M 01 .
the Veronese type form 52 while M 02 N2
A. SU3
As here we have N3, so subcase 56 is the same with the Veronese type subcase that we
have discussed previously. Thus in this section we consider only subcase 55.
In this subcase, we can choose the initial matrix M 0 to be given by
M 0
1
0
a b 2
c
1
57
If we now require that D 0 (1 2 ) 2 , then c2 while there is an infinite number of solutions

for a and b as here we have only one equation for a and b. In the following we restrict our
attention to the solution: a0 and b1.
Starting from the corresponding initial matrix M 0 we find that
M 1
T
& & 2 & 1
,
D0
2 & 1
58
which is orthogonal to M 0 , i.e., M 0 M 1 0. Note that M 1 2 is singular, i.e., D 1 0.

Using these two basis matrices, M 0 and M 1 , we find from 35 that N0 2, and from 41
I0 4. These results coincide with the corresponding numbers in the one rank-1 projector approximation of the SU3 case described by the initial vector field8
f 0 1,& , 2 T .
59
As A 3 (2) 32 A 3 (1) both rank-1 and rank-2 projector approximations have equal energy, i.e.,
E2.444 04 (m 0). This equality holds for m 0 as well.
The result D 1 0 that we have encountered previously is in fact a general property for SU3
case with rank-2 projectors ansatz, which can be seen as follows. Using the splitting relation 17
we see that
M 1 I P 01 1 P 02 1 M 01 , I P 01 1 P 02 1 M 02 .
60
As the case M 020 is trivial so we will only consider the nonzero case. We observe that
V 0 M 01 ,
V 1 I P 01 1 M 02 ,
61
V 2 I P 01 1 P 02 1 M 02
are mutually orthogonal and so they span C 3 . Thus,
M 01 V 0 V 1 V 2 ,
62
where , , and are expansion coefficients which depend on and . As

I P 01 1 P 02 1 V 0 V 1 V 2 V 2 ,
63
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1865
TABLE I.
m
E(1)
E(2)
0.0
0.2
1.0
2.23
7.0
30.0
3.644 10
3.682 91
4.172 41
5.001 86
7.471 87
14.339 3
3.813 87
3.861 58
4.420 30
5.334 05
8.023 36
15.428 7
so
M 1 V 2 ,V 2 ,
64
and it clearly follows that M 1 2 has a vanishing determinant.

B. SU4
1. D 0 c 1 2 3
In this subcase we can choose the initial matrix M 0 to be given by

M 0
a b 2
c d 2
65
If we require D 0 (1 2 ) 3 , then there is an infinite number of solutions for a, b, c and d as here

we have only three equations for the four parameters a, b, c, and d. In the following we consider
the solution: a2, b1, c1, d2.
Then, starting from M 0 in 65 we have computed the corresponding mutually orthogonal
matrix M 1 and by using these two basis matrices, M 0 and M 1 , we have computed explicitly the
integrals N0 and I0 in expressions 35 and 41, respectively and we have found that N0 3 and
I0 233 . We note that, in the one projector rank-1 case with initial vector field
f 0 1,) ,) 2 , 3 T ,
66
I0 9 which is larger than the above-given result. However, as A 4 (2)2 while A 4 (1) 43 it is not
clear which energy is larger. To assess this we have solved numerically Eq. 51 for F. In Table I
we present our results for the energies E(2) and compare them with the results using one rank-1
projector, E(1), 8 for different values of the mass m .
From Table I we see that for all the masses at least to m 30.0 we always have E(2)
E(1). When we have solved 51 for the approximate profile function Fg 0 , we found that it
is very close to that found in Ref. 8 using rank-1 projector ansatz.
2. D 0 c 1 2 1 2 2
In this subcase, we choose the initial matrix M 0 to be given by the following Veronese type
form:
M 0
)2
2)
67
which has D 0 12(14 2 )(1 2 ) 2 . For this configuration, we have found that it has N0
3 but I0 7.143 57 and its energy is E3.769 29 (m 0), which is a little lower than the
energy of Sec. V B 1, but it is still higher than the energy of the one rank-1 projector approximation, i.e., E3.644 10.
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1866
C. SU5
1. D 0 c 1 2 4
In this subcase we can choose the initial matrix M 0 to be

M 0
a b 2 c 3 d 4
e
0
0
1
1
0
68
If we require D 0 (1 2 ) 4 , then there is an infinite number of solutions for a, b, c, d, and e as

here we have only four equations for these five parameters. Here we consider only the solution:
abce& and d1.
For this configuration, we have found that N0 4 but I0 12.2667, which is much lower than
16. However, as A 5 (2) 125 , we see that its energy is E5.105 80 (m 0), which is still higher
than E4.837 92 in the rank-1 projector approximation.13 When we have solved the equations for
the approximate profile function Fg 0 for each rank 1 and 2, we found that their difference is
very small.
2. D 0 c 1 2 2 1 2 2
In this subcase we choose the initial matrix M 0 to have the following Veronese type form:
M 0
6 2
23
12
24
69
which has D 0 144(13 2 ) 2 (1 2 ) 2 .

We have found that this configuration has N0 4 but I0 12.4444 and its energy is E
5.118 75 (m 0) which is higher than the energy of the case in Sec. V C 1.
VI. TWO PROJECTOR APPROXIMATIONS
Now we consider the case of two projectors. Here we have two profile functions: F 0 and F 1 ,
i.e., 2, and so the energy integral 45 becomes
E
1
12
dr r 2 A N n F 20 A N 2n F 0 F 1 A N 2n F 21 N0 4F 20 1cos F 0 N1 4F 21
1cos F 1
1
I 1cos F 0 2 H1 1cos F 0 1cos F 1 I1 1cos F 1 2 , 70
r2 0
where now, for simplicity, we have set m 0.

The field equations for F 0 and F 1 are
F 0 A N n
N0 1cos F 0
1
2
1
A N 2n F 1 A N n F 0 A N 2n F 1
2
r
2
r
2
sin F 0
2I0 1cos F 0 H1 1cos F 1
N0 F 20 4
0,
2
2r
r2
r2
N1 1cos F 1
2 1
1
A N 2n F 0 F 1 A N 2n
A 2n F 0 A N 2n F 1
2
2
r
r 2 N
sin F 1
2I1 1cos F 1 H1 1cos F 0
N1 F 21 4
0.
2r 2
r2
r2
71
72
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1867
In the following, we take n2 and solve these equations numerically by imposing the
boundary conditions: F 0 (0)2 , F 1 (0)0, and F 0 ()F 1 ()0. Thus, the baryon number of
these configurations is BN0 . We then compare the approximate energies of each of these cases
with the corresponding energies of the one rank-1 projector approximations.8
For the SU3 configurations, we have shown that M 1 2 is singular. Thus for this case the
projector P 1 does not exist. As for the SU4 and SU5 cases that we have considered previously,
M 1 2 is nonsingular, so in this section we consider these two cases only.
A. SU4
Starting from the initial matrix M 0 in 65, we find that

M 2 0,
73
so from 35, 41, and 42 we have N1 0, I1 0, and H1 0. As for this case, N4 and n
2, the energy integral 70 reduces to the energy integral 49 for the corresponding one projector of rank-2 projector approximation.
We note that M 2 0 is in fact a general property for SU4 with rank-2 projectors ansatz. To
k
M 0 is given by
prove this it is convenient to note that M k P
M k I P 0 P 1 P k1 k M 0 .
74
Then using the splitting relation 17 in 74 for M 2 gives

M 2 I P 0 2 P 1 2 2 M 0 ,
I P 01 1 P 02 1 P 11 1 P 12 1 2 M 0 .
75
Now, M 2 0 follows from the completeness relation for the rank-1 projectors in C 4 , i.e.,
P 01 1 P 02 1 P 11 1 P 12 1 I.
76
B. SU5
As Sec. V C 1 in the one projector approximation has lower energy, we restrict our attention to
this subcase. Starting from the matrix M 0 in 69, we have computed the corresponding matrices
M 1 and M 2 , and we have found that N1 2.0, I1 4.289 89, and H1 3.573 57.
Solving Eq. 72 with the correct values for Nk , Ik , and Hk , k0 and 1, we have found that,
as A 5 (2) 125 and A 5 (4) 85 , this configuration has energy: E5.024 69. This is higher than the
energy of the one rank-1 projector approximation case, i.e., E4.837 92; however, they are
marginally higher than the exact energy of the SU2 case with B4, i.e., E4.464.5
VII. CONDITION FOR AN EXACT SPHERICALLY SYMMETRIC SOLUTION
In this section we consider whether one could also construct exact spherically symmetric
solutions of the SU(N) Skyrme field equations 5 using rank-2 projectors. When written in the
spherical holomorphic coordinates, Eq. 5 without the mass term becomes
r r 2L r
1 2 2
1 2 2
L , L r ,L L , L r ,L
L L
8
2
1 2 2
1 2 2
L r , L ,L r L r , L ,L r
1
8
16r 2
2 2 L , L ,L 1 2 2 L , L ,L 0.
77
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1868
Now, let us look in detail at all the terms in these equations in our generalized harmonic map
ansatz 22 case. We find that
1
L L 2i
sin F l M l1 M l 2 M l1
M l M l 2 M l1 2 M l 2 M l ,
l0
78
L , L r ,L 2i
l0
S l F l M l1 M l 2 M l1
M l M l 2 M l1 2 M l 2 M l ,
79
L r , L ,L r
l0
F 2l b l M l1 M l 2 M l ,
80
L , L ,L 2
l0
2S l b l M l1 M l 2 M l1 2 M l 2 M l S l b l1 M l2 M l 2 M l1
S l1 b l M l1 M l1 2 M l2 2 M l 2 M l ,
81
where S l is given by 31 and where

b l e iF l 1 .
82
We have checked that none of the configurations that we have considered so far in this paper
is an exact solution of 77. This can be seen as follows. First, we multiply 77 by M l from the
right, which results in a set of equations for M l . As the terms r (r 2 L r )M l are proportional to M l
while the others are not, the contracted equations have the following general structure:
1
a lM l
k0
M k A kl 0,
83
where a l depend only on N, n, g i , g i , and r while the (nn) matrices A kl depend on g i , , and
as well. Clearly, these equations are inconsistent, unless each matrix A kl is proportional to the
(nn) unit matrix I n , i.e., A kl b kl I n , where each b kl is independent of and .
Armed with this observation we can now ask the following question: for which forms of the
matrices M k do the SU(N) Skyrme field equations 77 have exact solutions? In Appendix C, we
have looked at the SU(N) chiral models and we found that, if we are able to find the matrices M k
which satisfy the condition
M k1 2 M k 2 Kk 1 2 2 I n ,
84
where Kk are some constants which depend on N, n, and k, then this configuration could possibly
give exact solutions for the profile functions g k .
To see how this may work in our case we have put the condition 84 into 7881, which
has turned the field equations 77 into the following reduced set
1
k0
P k
nI
P k1 P k k P k2 P k1 k 0,
N k
85
where k , k , and k are some functions of the profile functions g i and their derivatives, r and
Kk . In the appendix of Ref. 13 we have analyzed this type of equation for the case of the rank-1
projectors for the alternative SU(N) Skyrme model in Ref. 13, k 0 and we have found that
these equations do, indeed, yield consistent and compatible equations for the corresponding profile
functions g l .
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1869
To show whether this is also true for the case of rank-n projectors, we put the condition 84
into 35, 41, and 42, which yields the following relations:
Nk nKk ,
86
Ik nK2k ,
87
Hk nKk Kk1 .
88
Using these relations in 85 and following the procedure of the appendix of Ref. 13, we have
found that the resulting equations for the profile functions g k do, indeed, coincide with 48. This
also means that the energy integral 24, in this case, is exact i.e., not an approximation.
VIII. FURTHER ANALYSIS OF THE CONDITION 84
As the matrices M k for k0 are generated from the initial matrix M 0 , so in this section, we
derive the conditions that M 0 should satisfy in order for condition 84 to hold. To do so, we put
84 into A3 and A4 with k0, and obtain
M 0 2 M 0 2 K0 1 2 2 I n ,
89
log D 0 nK0 1 2 2 ,
90
and
respectively.
The general solution of Eq. 89 is
M 0 2 1 2 K0 G , ,
91
where G( , ) is an (nn) Hermitian matrix satisfying
G G 1 0.
92
Equation 92 has the general solution14

H ,
G , H
93
H are arbitrary (nn) matrices of one variable.

where H and H
Thus, with G( , ) given by 93, solution 91 becomes
H ,
M 0 2 1 2 K0 H
94
and from condition 84 it follows that for l1

H .
M l 2 K0 K1 Kl1 1 2 K0 2l H
95
Furthermore, from 91 it follows that

Det H
D 0 1 2 nK0 Det H
96
is a solution of 90. By putting D 0 into A6 we have found that N0 nK0 , as we have assumed
( ) is antiholomorphic with the only singularity at
that Det H( ) is holomorphic and Det H
, which is consistent with 86. Thus, we can choose Det G( , ) 1, and so from 95 it
follows that
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1870
D l K0 K1 Kl1 n 1 2 n K0 2l ,
97
from which, according to 38,
l n K0 2l .
98
Using the recurrence relations 37, with l given by 98, we derive

Nl l1 N0 nl .
99
Now, with M 0 2 given by 94, the projector P 0 (n)M 0 M 0 2 M 0 becomes

0M
0 ,
P 0 n 1 2 K0 M
100
0 M 0 H 1 . Using 94 then it follows that the matrix M

0 satisfies
where M
0 2 H
1 M 0 2 H 1 1 2 K0 I n .
M
101
0 (M
01 ,M
02 ,...,M
0n ) are
Equation 101 implies that the column vectors of the matrix M
mutually orthogonal. Thus, using the SU(N) global symmetry, we can bring the column vectors
0 j , ( j1,2,...,n) to live in n-disjoint subspaces. In this case, the projector P 0 (n) has a block
M
diagonal form. For example, in the rank-2 projector ansatz, i.e., n2, for Ne v en, we have
P 0 2
P 01 1
P 02 1
102
where
P 0s 1
0M
0
M
0 2
M
103
s0, 1, are rank-1 projectors. For Nodd, still in the rank-2 projector ansatz, Eq. 101 requires
0 j should be zero, so if we choose (M
0 j ) N 0, then the ((N1)
that one of the entries of M
(N1)) submatrix of P 0 (2) has the block form 102 while P 0 (2) NN 0.
Thus, as far as condition 84 is concerned, it seems that the only exact spherically symmetric
solutions of the SU(N) Skyrme models using projectors of rank-2 are embeddings of a pair of
SU( N /2) solutions of rank-1 projector ansatz, where N N, or (N1) for N even or odd,
respectively.
IX. SOME SPECIFIC CONFIGURATIONS
In this final section we consider only the rank-2 projector ansatz, i.e., n2. Thus, here, the
SU3 case should be excluded, as from our analysis in Sec. V A, M 1 2 is singular so it is
automatically not proportional to the nonsingular matrix M 0 2 as required by 84. For the same
reason, we also exclude the case M 02 M 01 , i.e., Eq. 21. Next, we have a look at some
specific forms of the initial matrix M 0 , for the N4 and 6 cases.
A. SU4
In this case, we can choose the initial matrix M 0 to be given by

M 0
104
where f ( ) is an arbitrary polynomial function of only . Then
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
M 0 2 1 2
1 f
0
1871
105
which is of the form 94.

For the corresponding projector P 0 (2), we have found by direct evaluation that it has the
block form 102 with
P 0s 1
1
1
2
1
106
s0, 1, which clearly are the rank-1 projectors of a one skyrmion SU2 solution. As D 0 (1
2 ) 2 , this configuration has N0 2, or K0 1, and so according to 87, I0 2. Then from 99
we obtained N1 0 or K1 0 according to 86 so from 95 we conclude that M l 0, for l
2, which is consistent with our general result in Sec. VI A. Thus, this configuration has only one
projector, i.e., P 0 (2).
Substituting N0 I0 A 4 (2)2 in the 1-projector energy integral 49 we found that it has
energy: E2E SU(2) , where E SU(2) is the energy of the SU2 Skyrme model,9 as we would have
been expected.
B. SU6
In this case, we can choose the initial matrix M 0 to be given by

M 0
&
& f
2 f
&
107
from which it follows that M 0 2 is of the form 94 as well with K0 2.

We have also found that the corresponding projector, P 0 (2), has the block form 102 with
P 0s (1), s0, 1, are the rank-1 projectors of the SU3 solution.9 As D 0 (1 2 ) 4 , this configuration has N0 4, from which we derived that I0 8, N1 4, I1 8, H1 8, but N2 0 or
K2 0 and so M l 0 for l3. Thus, this configuration has only two projectors, i.e., P 0 (2) and
P 1 (2).
Substituting the correct values for Nk , Ik , and Hk , k0, 1, into the 2 projectors energy
integral 70 we have found that, as A 6 (2)A 6 (4) 38 , this configuration has energy: E
2E SU(3) , where E SU(3) is the energy of the SU3 Skyrme model.9
X. CONCLUSIONS
In this paper we have studied the SU(N) Skyrme models by constructing SU(N) multiskyrmions fields using the harmonic mappings method where we have generalized the method of
Ioannidou et al.9 by considering projectors of S 2 into the Grassmannian Gr(n,N), i.e., using
projectors of rank n1. In this approach a rank-n projector P k (n) is constructed from a member
of a set of mutually orthogonal (Nn) matrix fields M k . In particular, here we have concentrated
our attention only on the rank-2 projectors.
Using our construction we have studied some approximate spherically symmetric configurations of SU(N) Skyrme models. When we have solved the equations for the profile functions for
configurations with baryon number B(N1) we found that they are very close to those for the
rank-1 cases and that they have marginally higher energies. These results indicate that the rank-1
projector ansatz8 is the best way to approximate energy minima of the SU(N) Skyrme models.
We have also discussed the possibility of generating exact spherically symmetric solutions
using this construction. However, we have found that, in contradistinction to the rank-1 projector
ansatz in which exact spherically symmetric solutions can be found numerically by using the
Veronese sequence of N mutually orthogonal vector fields in C P N1 , such a construction is more
involved in our case. In particular, we have found that if the sequence of the (Nn) matrix fields
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1872
M k satisfy condition 84 then it seems that the only possible exact solutions are embeddings. For
example, for the rank-2 projector ansatz these are embeddings of a pair of SU( N /2) solutions of
rank-1 projector ansatz, where N N, or (N1) for N even or odd, respectively.
It is not clear whether there exist other conditions than 84 for the (Nn) matrix fields M k
that might lead to nonembedding solutions.
ACKNOWLEDGMENTS
The authors thank B. Piette and P. M. Sutcliffe for their interest. H.J.W. wants to thank the
QUE Project at the Department of Physics, Institute of Technology Bandung, Indonesia, for his
grant.
APPENDIX A: RECURRENCE RELATIONS FOR Tr M k 2 M k 1 2 AND Nk
In this appendix, we derive recurrence relations for Tr( M k 2 M k1 2 ) appearing in the

integral Nk of 35. To do this we rewrite the definition of M k1 in 10 as follows:
M k M k1 M k M k 2 M k M k .
A1
As M k is given by 11 so from the integrability condition
M k M k ,
A2
M k 2 M k 2 M k1 2 M k 2 M k 2 M k1 2 .
A3
we derive the recurrence relations
We note that for the n1 case, i.e., when M k are a sequence of N-component vector fields, Eq.
A3 gives the celebrated Toda equation.15 Thus, for n1 our Eq. A3 could be considered as its
generalization to the non-Abelian case.16
Furthermore, taking the trace of Eq. A3 we obtain
log Det M k 2 Tr M k1 2 M k 2 Tr M k 2 M k1 2 ,
A4
i.e., our recurrence relations for Tr( M k 2 M k1 2 ).

Next, if D k Det M k 2 0 in the whole complex plane C then by applying Stokes theorem in
the plane to A4, i.e.,
d d
1
2
d d ,
D k
Nk Nk1 .
Dk
A5
we obtain
i
4
D k
Dk
A6
Thus, if k is the highest degree of 2 in D k , i.e., lim D k 2 k , then A6 reduces to the

following simple recurrence relations for Nk , i.e.:
k Nk Nk1 .
A7
However, if D k 0 at some points, these recurrence relations cease to hold.

APPENDIX B: REDUCED FORMULA FOR EVALUATING Tr M k 1 2 M k 2 2
In this appendix, we derive a formula for simplifying the calculation of the trace in I for the
case n2. Using the formula,
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
Tr H 2 TrH 2 2 Det H,
1873
B1
which is true for any (22) matrix H we note that for D k Det M k 2 0, we have
Tr M k1 2 M k 2 2 Tr M k1 2 M k 2 2 2
D k1
.
Dk
B2
Thus, if D k1 0, then from B2 we have

Tr M k1 2 M k 2 2 Tr M k1 2 M k 2 2 .
B3
APPENDIX C: CONDITION 84 FROM THE SU N CHIRAL MODELS
To simplify our search for finding a condition for exact solution of the SU(N) Skyrme model
Eqs. 77, in this appendix, we look at the corresponding SU(N) chiral model equations,
r r 2L r
1 2 2
L L 0,
2
C1
i.e., 77 without the Skyrme terms. In terms of the rank-n projectors P k , Eq. C1, after we have
put in 25 and 78, becomes
1
k0
P k
nI
r 2 g k 1 2 2 sin F k M l1 M l 2 M k l1
M l M l 2 M l1 2 M l 2 M l
N r
0,
C2
where F k g k g k1 , g 0.
Next we multiply Eq. C2 from the right by M s , and noting that M s are independent matrix
fields, we obtain
r r 2 g s
n
g
N k0 k
I n 1 2 2 M s1 2 M s 2 sin F s1 M s 2 M s1 2 sin F s 0.
C3
Finally, summing over s from 0 to l, and noting that M 1 0 by definition, we find that C3
gives us
r r
p0 qp
F q
n l1
p1 F p
N
p0
I n 1 2 2 M l 2 M l1 2 sin F l 0.
C4
Thus, in order to have a compatible and a consistent set of equations for the functions F l in
C4, the matrices M l must satisfy
M l 2 M l1 2 Kl 1 2 2 I n ,
C5
i.e., condition 84.

T. H. R. Skyrme, Nucl. Phys. 31, 556 1962.
E. Witten, Nucl. Phys. B 223, 422 1983.
3
E. Witten, Nucl. Phys. B 223, 433 1983.
4
N. S. Manton, Phys. Lett. B 192, 177 1987.
5
R. A. Battye and P. M. Sutcliffe, Phys. Rev. Lett. 79, 363 1997.
6
A. P. Balachandran, A. Barducci, F. Lizzi, V. G. J. Rodgers, and A. Stern, Phys. Rev. Lett. 52, 887 1984.
7
V. B. Kopeliovich, B. E. Schwesinger, and B. E. Stern, JETP Lett. 62, 185 1995.
1
2
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36
1874
T. Ioannidou, B. Piette, and W. J. Zakrzewski, J. Math. Phys. 40, 6353 1999.

T. Ioannidou, B. Piette, and W. J. Zakrzewski, J. Math. Phys. 40, 6223 1999.
10
C. J. Houghton, N. S. Manton, and P. M. Sutcliffe, Nucl. Phys. B 510, 507 1998.
11
W. J. Zakrzewski, Low Dimensional Sigma Models IOP, Bristol, 1989.
12
A. Din and W. J. Zakrzewski, Nucl. Phys. B 174, 397 1980.
13
H. J. Wospakrik and W. J. Zakrzewski, J. Math. Phys. 42, 1066 2001.
14
E. Witten, Commun. Math. Phys. 92, 455 1984.
15
J. Bolton and L. M. Woodward, in Harmonic Maps and Integrable Systems, edited by A. P. Fordy and J. C. Wood,
Aspects of Mathematics Vol. 23 Vieweg, Braunschweig/Wiesbaden, 1994, pp. 60 81.
16
M. Bruschi, S. V. Manakov, O. Ragnisco, and D. Levi, J. Math. Phys. 21, 2749 1980. See also V. B. Matveev and M.
A. Salle, Darboux Transformations and Solitons Springer, Berlin, 1991.
8
9
175.111.89.14 On: Fri, 02 Oct 2015 08:31:36

SU (N) Skyrmions From Harmonic Maps

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

SU (N) Skyrmions From Harmonic Maps

Caricato da

Copyright:

Formati disponibili

SU (N) skyrmions from harmonic maps S 2 Gr (2,N)

H. J. Wospakrik and W. J. Zakrzewski

JOURNAL OF MATHEMATICAL PHYSICS

VOLUME 43, NUMBER 4