Vortices∗

F.A. Schaposnik†
arXiv:hep-th/0611028v1 2 Nov 2006

Departamento de Fı́sica, Universidad Nacional de La Plata

C.C. 67, 1900 La Plata, Argentina

Abstract
I discuss in these lectures vortex-like classical solutions to the equa-
tions of motion of gauge theories with spontaneous symmetry break-
ing. Starting from the Nielsen-Olesen ansatz for the Abelian Higgs
model, extensions to the case in which gauge dynamics is governed
by Yang-Mills and Chern-Simons actions are presented. The case of
semilocal vortices and also the coupling to axions is analyzed. Finally,
the connection between supersymmetry and the existence of first order
BPS equations in such models is described.

∗
Lectures given at the “4th Chilean School of Astrophysics, Cosmology and Gravita-
tion” Valparaiso, October 2006.
†
Associated with CICPBA

1
 Turning and turning in the widening gyre
The falcon cannot hear the falconer
W.B.Yeats, The Second Coming[1]
These notes cover the topics of 4 lectures on vortex solutions in
spontaneously broken gauge theories and their connection with super-
symmetry, given at the 4th Chilean School of Astrophysics, Cosmology
and Gravitation held in Valparaiso last October. The bibliography at
the end of the notes is not complete: it only refers to works containing
results discussed in the lectures. A more complete list of references
can be found in some of the excellent books and reviews on this topic
[2]-[7].

1 Nielsen Olesen vortices

Vortices are very ubiquitous objets. Starting from the 1957 work of
Abrikosov [8] on superconductivity, belonging to the area of condensed
matter, they were then rediscovered by high energy physicists in a
field theory context, after the well-honnored paper by Nielsen and
Olesen [9]. In the present century a revival took place due to their
cosmological interest in connection with cosmic superstrings (see [10]
and references therein).
Nielsen and Olesen [9] introduced vortices in high-energy physics
in an attempt to describe classical string-like structures with the prop-
erty of having equations of motion identical with those of the Nambu
dual string.
In their work, Nielsen and Olesen started from a 3 + 1 Lagrangian
for an Abelian U (1) gauge theory coupled to a charged scalar with
a symmetry breaking potential and first found a classical string-like
(vortex) solution. Then they obtained the remarkable result the action
is proportional to the area of the surface swept out in space and time
by such vortex. This result is the basis of their identification of the
Abelian Higgs Lagrangian with the Nambu Lagrangian for the dual
string [11].
By making a static axially symmetric ansatz they reduced the
Euler-Lagrange equations to a pair of second order non-linear coupled
differential equations in the radial variable. Although they didn’t find
exact solutions (in fact, analytical vortex solutions are not known)
they gave arguments to show that nontrivial solutions exist. They

2
correspond to vortex-like object carrying a magnetic field in its core,
where the scalar field vanishes. The magnetic flux is quantized (al-
ready at the classical level).
Such solutions look like type-II superconductivity vortices and this
is no coincidence: it is not difficult to see the connection between the
Abelian-Higgs model Lagrangian and the Ginsburg-Landau fenomeno-
logical model for superconductivity, provided the Higgs scalar in the
relativistic field theory is identified with the order parameter in the
Ginzburg-Landau model. In order to have type-II superconductivity
the relation between the gauge coupling constant e2 and the symmetry
breaking coupling constant λ should be e2 < 8λ (with the conventions
adopted below); for e2 > 8λ a complete Meisner effect takes place
(type-I superconductivity). The limiting point

e2 = 8λ (1)

plays also an important role in the field theory/string theory context.

In fact, a surprising connection was already notice in 1976, when it
was observed [12] that relation (1) is precisely the same needed for the
supersymmetric extension of the model [13]-[14].
The Abelian Higgs model dynamics in (3+1)-Minkowski space is
governed by the action:
 
1 1
Z
4 µν ∗ µ
S = d x − Fµν F + (Dµ φ) (D φ) − V (|φ|) (2)
4 2

Here Fµν is the field strength,

Fµν = ∂µ Aν − ∂ν Aµ , (3)

φ is a complex scalar,
φ = φ1 + iφ2 , (4)
the covariant derivative is defined as

Dµ = ∂µ + ieAµ , (5)

and the symmetry breaking potential taken in the form

V (|φ|) = λ |φ|2 − φ20

(6)

Here φ0 is a real positive constant with [mass] dimensions.

3
 The resulting equations of motion are

∂ µ Fµν = jν
δV
D µ Dµ φ = − (7)
δφ∗
with the Higgs field current given by
ie
jν = (φ∂ν φ∗ − φ∗ ∂ν φ) + e2 |φ|2 Aν (8)
2
The Nielsen-Olesen strategy to construct nontrivial regular solu-
tions to the equations of motion with finite energy (per unit length)
starts from a trivial (constant) solution and implies the following steps:

1. Trivial zero-energy solution |φ| = φ0 , Ai = 0

2. non-trivial but singular solution (fluxon with N units of magnetic

flux) which in polar coordinates reads

φ = φ0 exp(iN ϕ) , Ai = N ∂i ϕ , εij Fij = 4πN δ(2) (~x)

1
I
εij Fij = N ∈ Z (9)
4π
This configuration is a gauge transformed of the trivial one. The
appropriate gauge group element to consider is

gN (ϕ) = exp(iN ϕ) (10)

which is ill-defined at the origin.

3. Regular Nielsen-Olesen vortex ansatz

1 −1
φ = f (r)gN (ϕ) , Ai = a(r) gN (ϕ)∂i gN (ϕ) (11)
i
with
f (0) = a(0) = 0 , f (∞) = φ0 , a(∞) = 1 (12)

This ansatz leads to the same magnetic flux as that in (9) but there
is no singularity in view of the boundary conditions that one imposes
at the origin on f and a. Conditions at infinity guarantee that each
term in the Lagrangian goes to zero at infinity.

4
 Being axially symmetric, ansatz (11) can be interpreted as a mag-
netic tube with quantized flux (a vortex). Being z-independent, such
a solution has infinite energy. However, if we define an energy E per
unit length, conditions (12) ensure its finiteness,
 
1 2 1
Z
2 2 2 2 2
E= d x F + |Di φ| + λ(|φ| − φ0 ) (13)
4 ij 2

Being z independent, such a static vortex solution can be also inter-

preted as a soliton solution in 2 + 1 dimensions. Because it depends
only on x, y, it can also be thought as an instanton solution in d=2
Euclidean dimensions. In any two of this cases, space at infinity can be
1 . But also the gauge group, U (1), can be identified
seen as a circle S∞
1
with a circle S so that one can see gN as establishing a map
1
gN : S∞ → S1 (14)

Such maps are classified into homotopy classes Π1 (where the subindex
refers to the dimension n of the sphere at infinity, in this case n = 1).
Now, one has
Π1 (S 1 ) = Z (15)
and in this way one connects the vortex magnetic flux (9)with a topo-
logical charge.
Unlike monopoles or instantons, no analytic solution to the equa-
tions of motion is known. However, numerical solutions can be easily
constructed and there are also proofs that solutions exist (see [3] and
references therein). The asymptotic behavior of the vortex solutions
is
1
a(r) = − + mrK1 (−mV r)
e
φ(r) = φ0 + O(exp(−mφ r)) (16)

where mV and mφ are the masses of the vector and scalar particles
respectively,

mV = eφ0 (17)
√
mφ = 2 2λφ0 (18)

Note that when condition (1) is satisfied both masses coincide, this
being the first indication that the model can be viewed as describing
the bosonic sector of an N = 2 supersymmetric theory, with the scalar

5
and the gauge field in the same supermultiplet. In fact when mV ∼ mφ
we have a well defined vortex line or a well defined string of extension
1/mV ∼ 1/mφ . Inside the string one basically has magnetic field while
the scalar field vanishes. Outside the string it is the magnetic field
which is nearly zero while the Higgs scalar is practically in its vacuum
value |φ| = φ0 .
Explicit solutions in the whole r range can be more easily found
observing that, under magic condition (1) the second order Euler-
Lagrange equations can be reduced to first-order ones. There are
different ways of arriving to this result. One was discovered in [12]
and starts by demanding that the space-space energy-momentum com-
ponents (Trr and Tϕϕ ) vanish. The other one establishes a bound
(Bogomol’nyi bound) for the vortex energy (per unit length) which is
saturated whenever (1) holds [15]. Being configurations static, such
a bound also corresponds to a bound for the action and hence they
satisfy the equations of motion.

The Bogomol’nyi bound

Let us see how Bogomol’nyi found in ref.[15] a bound for the energy
starting from formula (1). To this end, let us write the scalar field in
terms of its real and imaginary components

φ = (φ1 , φ2 ) = (φa ) , a = 1, 2 (19)

and scale fields and coordinates according to

1
φa → φ0 φa , Ai → φ0 Ai , x → y (20)
eφ0
so that the energy reads now
 
1 2 1 λ
Z
2 2 a a 2 2
E = φ0 d y F + Di φ Di φ + 2 (|φ| − 1) (21)
4 ij 2 e

where the covariant derivative Diab is defined as

Diab = δab ∂i + εab Ai (22)

One easily sees that the energy can be rewritten as

6
  2
11
Z
E = φ20
d y 2 a a
Fij ± εij (φ φ − 1) 2
42
1 2  λ 1

a ab
+ Di φ ∓ ε εij Dj φ b
+ − (φa φa − 1)2
4 e2 8
 
1 1 ab a b
± εij Fij ∓ ε εij ∂i (φ Dj φ ) (23)
4 2

The term
1 ab
ε εij ∂i (φa Dj φb ) (24)
2
gives no contribution to the energy, as can be seen by using the Gauss-
Stokes theorem and the fact that Di φb = 0 on S∞ 1 . Then, using (9)

we have
 2
1 1
Z
2 2 a a 2
E = φ0 d y Fij ± εij (φ φ − 1)
4 2
1  2  λ 1
 
a ab
+ Di φ ∓ ε εij Dj φ b
+ − a a 2
(φ φ − 1) ± φ20 πN
4 e2 8
(25)

Were the third term absent, the energy would just be the sum of
to positive definite terms plus a term proportional to the topological
charge. This is precisely what happens for the Bogomol’nyi point (1)
for which one then has
E ≥ φ20 π|N | (26)
The bound is attained when the two perfect squares vanish, and this
leads to the so-called Bogomol’nyi or BPS equations for vortices,

Fij = ∓εij φ2 − 1

Di φa = ±εab εij Dj φb (27)

The ”PS” refers to M. K. Prasad and C. M. Sommerfield who

found exact monopole solutions for the non-Abelian version of the
model for a particular (Bogomol’nyi) value of the coupling constant
[16] (λ = 0). That their solution corresponds to a Bogomol’nyi bound
was first noted in [17].
As mentioned above, eqs.(27) were independently obtained by ask-
ing the stress tensor components Tij to vanish. This is consistent

7
with the eqs. of motion only if λ = 8e2 . When matter interaction
exceeds that of the electromagnetic one (λ > 8e2 ) vortices attract
whereas in the opposite case they repel. In the former case a system
of many superimposed vortices decays into separated vortices. In the
case λ = 8e2 vortices do not interact and this is consistent with thew
fact that the energy is proportional to the topological charge. An ex-
act (numerical) solution to the system (27) was presented in [12]. The
existence theorem for vortex (and monopole) solutions can be found
in A. Jaffe and C. H. Taubes book on Vortices and Monopoles [3].

Chern-Simons vortices
As explained above, if one defines the Maxwell-Higgs system in d =
2+ 1 dimensions, vortices can be seen as static finite energy (solitonic)
solutions of the equations of motion. Now, as it is well-known, there
is in three dimensional space time a second Lorentz invariant term
which can be added to the Maxwell term, namely the Chern-Simons
term which, for a U (1) gauge theory reads
µ ναβ
LCS = ε Fνα Aβ (28)
4
Here µ is a parameter with mass dimensions usually called the topo-
logical mass. This name is due to the following important fact related
to the Chern-Simons term [18]-[20]. Consider the U(1) gauge theory
with Lagrangian
1 µ
LM +CS = − Fµν F µν + εναβ Fνα Aβ (29)
4 4
Now, from the eqs. of motion for this Lagrangian,

∂µ F µν + µ∗F ν = 0 (30)

with the dual field strength ∗F ν given by

∗ ν 1
F = εναβ Fαβ (31)
2
one can find
∗F ν + µ∗F ν = 0 (32)
showing that the gauge field is massive.

8
 Clearly the field equations are gauge invariant but the Lagrangian
is not. Under a change

Aµ → Aµ + ∂µ Λ (33)

the Chern-Simons Lagrangian changes by a total derivative

µ
LCS → LCS + ∂α (∗F α Λ) (34)
2
The action is however gauge-invariant since the field strength fall off
at large distances and one accepts appropriate Λ′ s.
Let us then consider the Maxwell-Higgs model with the addition
of a CS term, with Lagrangian
 
1 µ ναβ 1
Z
3 µν ∗ µ
S = d x − µν F F + ε F A
να β + (D µ φ) (D φ) − V (|φ|)
4 4 2
(35)
The equations of motion now read

∂ µ Fµν + µ∗Fν = jν (36)

δV
D µ Dν φ = (37)
δφ∗

Now, integrating the temporal component of eqs.(36) (the Gauss law)

in the plane one has
Z Z
d2 x ∂ i Fi0 + µεij F ij = d2 xj0


(38)

The first term in the l.h.s. is a surface term and gives no contribution.
The second one is just the magnetic flux and that in the r.h.s. is
identified with the electric charge of the configuration. One then has

µΦ = Q (39)

with
1
Z Z
Φ= d2 xεij F ij , Q = d2 xj0 (40)
2
If one looks for static configurations, the current component j0 is just

j0 = e2 |φ|2 A0 (41)

9
We then see that either A0 6= 0 or we cannot have vortex-like solutions
like those discussed for the pure Maxwell-Higgs system. Interestingly
enough, if vortex solutions exist, magnetic flux and electric charge
are related according to (39). Being the magnetic flux related to the
topological charge, the electric charge is then also quantized already
at the classical level.
In fact, it is very easy to extend the ansatz (11) in order to include
an A0 [21],

φ = f (r) exp(inϕ) , Ai = na(r)∂i ϕ , A0 = a0 (r) (42)

Apart from conditions (12), one should ask

a0 (0) = a0 (∞) = 0 (43)

As it is to be expected in view of our previous discussion, the vector

mass is modified due to the presence of the Chern-Simons term (while
the scalar field mass remains unchanged)
r
µ2 µ
mA± = + e2 φ20 ± (44)
√4 2
mφ = 2 2λφ0 (45)

Using the axially symmetric ansatz, numerical solutions can be found.

The behavior of the Higgs scalar and the spacial components of the
gauge field is qualitatively the same as in the pure Maxwell-Higgs
case. Concerning the electric field, it vanishes at the origin and has
its maximum value at finite r.
The energy E of the vortex can be seen
to be E± ∼ m2A± log e2 /m2A± and hence the mA− solution has lower
energy.
No Bogomol’nyi bound for the energy can be constructed when the
CS term is included. Hence, there are in this case (and with this field
content) no BPS first order equations for the Maxwell-Higgs-Chern-
Simons model.
There is however a way out to this problem; one is to modify the
field content. The other one has been discovered in two simultaneous
works [22]-[23]. Basically, it consist in not including the Maxwell
term -that is, the gauge field dynamics is solely governed just the
Chern-Simons action- and changing the symmetry breaking potential
from the usual fourth order one to a sixth order potential which is
acceptable from the point of view of renormalizability since the model

10
is defined in 2+1 space-time. The model can be seen as the truncation
at large distances and low energies where the lower derivative Chern-
Simons term dominates the higher derivative Maxwell term.
The Lagrangian of the model is
 
µ ναβ 1
Z
3 ∗ µ (6)
S= d x ε Fνα Aβ + (Dµ φ) (D φ) − V (|φ|) (46)
4 2
Although one can include in the potential an arbitrary sum of fourth-
order and sixth order potential, Bogomol’nyi equations will arise for
a particular combination that can be written in the form
e4
2
V (6) = 2
|φ|2 |φ|2 − φ20 (47)
8µ
There is no independent coupling constant in this potential because it
has been already defined at the Bogomol’nyi point. Note that being
defined in a d = 3 dimensional theory, it corresponds to a renormaliz-
able potential.
Variation of the action yield to the field equations
µ αβγ
ε Fβγ = jα (48)
2
δV (6)
D µ Dµ φ = − (49)
δφ∗
where j α is still given by (8). The time component of eq.(48) is the
Gauss law,
µF12 = j0 (50)
Noting that there is no metric in the Chern-Simons action (in this
sense it is a topological Scwartz like action [24]) one concludes that it
does not contribute to the energy momentum tensor, which is defined
as
δS
Tµν = 2 µν (51)
δg
Thus, the energy functional for static field configurations reads
e2 2 2
 
1
Z
2 2 2
E= d x |Di φ| + A0 |φ| + V (|φ| ) (52)
2 2

In the static case j0 = e2 |φ|2 A0 so that we can use the Gauss law to
solve for A0 giving
µ F12
A0 = 2 2 (53)
2e |φ|

11
Inserting this in (52) one can easily accommodate the energy à la
Bogomol’nyi, a sum of squares plus the a term proportional to the
magnetic flux,
 2
2 !
 µ −1 e
Z
2 2 ∗ 2 2

E = d x |(D1 ± iD2 )φ| +  φ F12 ± φ (φ0 − |φ| 
2e µ
eφ20
∓ Φ (54)
2
where we have discarded a surface term which can be converted to
a line integral vanishing for finite energy configurations. Then, for a
fixed value of the flux, Φ = 2πN/e we have a bound for the energy,

E ≥ πeφ20 |N | (55)

The bound is saturated when the following first-order equations hold,

D1 φ = ∓iD2 φ
m2 |φ|2 |φ|2
 
eF12 = ∓ 1− 2 (56)
2 φ20 φ0
where m = 2e2 φ20 /|µ|. This mass coincides with the vector and the
scalar field masses m = mA = mφ , this again signaling some relation
with N = 2 supersymmetry.
Being a minimum for the energy, a solution of these equations also
solve are also static solutions of the eqs. of motion.
Axially symmetric solutions have been studied using an ansatz
of the form (42). Concerning the Higgs, the field profile is qualita-
tively analogous tho those for the Maxwell-Higgs system. As for the
magnetic and electric fields, as it happens for the Maxwell-CS-Higgs
model, they are concentrated in a ring with its maximum occurring
when |φ|2 = φ20 .

Semilocal vortices
Suppose that one replaces the complex scalar field in the Maxwell-
Higgs model by an SU (2) doublet φ = (φI , φII ) [25]. The action is

1 1 1
Z
4
S = d x − Fµν F µν + (Dµ φI )∗ (D µ φI ) + (Dµ φII )∗ (Dµ φII )−
4 2 2

V |φI |2 + |φII |2


(57)

12
This action is not only invariant under a local U (1)Local gauge group
but also under a global SU (2)Global group. The total symmetry group
is then
G = SU (2)global × U (1)local (58)
The potential has an O(4) symmetry and when it corresponds to a
symmetry breaking one the vacuum manifold is a three sphere defined
by
2 2 2 2
φI1 + φI2 + φII
1 + φII
2 = φ20 (59)
which is simple connected so that there are no topological string so-
lutions (Π1 (S 3 ) = 0). Now SU (2) global rotations in the direction of
the constant doublet (φI0 , φII
0 ) leave invariant the vacuum so that the
symmetry breaks to a global U (1)G ,

SU (2)global × U (1)local → U (1)global (60)

Flux-tube solutions to the equations of motion of such a model

have been presented in [25] where, in view of the symmetry scheme,
they have been called “semilocal strings”. In order to understand the
nature of such solutions, let us note that the equations of motion for
the model are formally the same as those for the Abelian Higgs model
with just one scalar, except that now such scalar has been replaced
by an SU(2) doublet. Then any (Nielsen-Olesen) vortex solution

φ = f (r) exp(iN ϕ) , Ai = a(r)∂i φ (61)

of the ordinary model can be extended to one of the present one by

introducing an SU (2) doublet Φ0 of unit norm,

a exp(iϕI )
 
Φ0 = √ (62)
1 − a2 exp(iϕII )

and writing the semilocal solution in the form

Φ = f (r)Φ0 , Ai = a(r)∂i φ (63)

Here 0 ≤ a ≤ 1.
Now, in order to have finite energy, covariant derivatives of ΦI and
ΦII should vanish at infinity, this implying

lim Di φI = 0 → (i∂i ϕI − iN ∂i ϕ) = 0
r→∞
lim Di φII = 0 → (i∂i ϕII − iN ∂i ϕ) = 0 (64)
r→∞

13
so that the two phases can differ just in one constant,
ϕI = Nϕ + C
II
ϕ = Nϕ (65)
and we have that asymptotically, finite energy configurations should
take the form
 
a exp(iC)
Φ(∞) = φ0 exp(iN ϕ) √ (66)
1 − a2
Although there are still a U (1)Global rotations of this configuration
that pass to another one which is still a minimum of the potential, the
energy of the transformed configuration should be infinity because
of the change in the covariant derivatives. That is, the space that
label finite energy configurations does not correspond to the vacuum
manifold V (the zeroes of the potential) but to the gauge orbit from
any point Φ0 ∈ V. In this sense the relevant homotopy group is
Π1 (U (1)) = Z. Thus configurations with different winding numbers
are separated by infinite energy barriers, an information that is not
contained solely in the vacuum manifold V but also in the kinetic
energy term. Now, since Π(V) = 0 it is not topologically guaranteed
that a configuration characterized by some number N is stable.
In summary, although the vacuum manifold is simply connected
and a field configuration that winds at infinity can be unwind without
any cost of potential energy, this does not guarantee that there is no
cost of the gradient part of the energy and in fact, this is what happens
due to the non-triviality of Π(GLocal /HLocal ) = Π1 (U (1)) = Z.
Stability of the configurations have been studied in [26]-[6]. Again,
the Bogomol’nyi point plays a central role;for e2 = 8λ one can repeat
the Bogomol’nyi argument and write a bound for the energy, which is
just that given in eqs.(25)-(26) but where the single complex scalar is
replaced by an SU(2) doublet. Being a local minimum, the solution
written in (63) with f (r) and a(r) the radial functions for the Nielsen-
Olesen solution is classically stable. One should mention the existence
of zero-modes [56] which will be described below.

Non Abelian vortices

The first reference to (local) vortex solutions in non-Abelian gauge
theories can be found in the Appendix of Nielsen and Olesen pio-
neer paper [9]. However, it was rapidly recognized that the boundary

14
conditions imposed in [9] have solutions that in some cases could be
deformed continuously to topologically trivial ones [27]. It was then
understood that in order to have topologically stable solutions, sym-
metry breaking has to be complete and hence more than one Higgs
scalar has to be introduced. This fact was discussed in detail in [28]-
[30] where it was observed that one needs Φ1 (GL /H) to be nontrivial.
For example, for SU (2) gauge theory, just one Higgs field in the ad-
joint representation is not enough: one needs a second one and a choice
of symmetry breaking potential so that the only residual symmetry
group element is just the unit matrix in the adjoint so that H = Z2
and Φ1 (SU (2)/Z2 ) = Z2 (ZN when the gauge group is SU (N )).
Let us discuss in some detail the necessary conditions on the vac-
uum manifold in order to have stable vortices when the gauge group
is G = SU (N ). To begin with, the election of a representation for
the Higgs field is crucial. Indeed, if we choose the fundamental rep-
resentation of SU (N ) or any other faithful representation (i.e., iso-
morphic to the group) the fundamental homotopy group Π1 (G/H) =
Π1 (SU (N )/I) = 0 = Π1 (SU (N )/I) since SU (N ) is simply connected.
The situation is different if the Higgs fields are in the adjoint, which
does not correspond to a faithful representation of SU (N ).
Noting that elements Ω0m such that

Ω0m = exp(2πim/N )I , m = 0, 1, . . . N − 1 (67)

can be taken as the elements of the Abelian group ZN , one can define
representatives of each homotopy class in the form

Ωm (ϕ + 2π) = exp(2πim/N )Ωm (ϕ) (68)

In brief, although we say that the symmetry is completely broken,

there is the residual symmetry H = ZN and then

Π1 (G/H) = Π1 (SU (N )/ZN ) = ZN (69)

It should be noticed that, taken as a gauge transformation group ele-

ment, Ωm (ϕ) is singular, are the non-Abelian counterpart of exp(imϕ),
for (Abelian) Nielsen-Olesen vortices.
Let us now discuss the number of Higgs scalars (in the adjoint) that
one has to introduce to break the symmetry so as to have topologically
non trivial solutions. We start by recalling the standard Cartan-Weyl
basis,

[Hi , Hj ] = 0 , j = 1, 2, . . . , N − 1

15
 [Hi , Eα ] = riα Eα
[Eα , Eα ] = riα Hi

Nαβ Eγ if rγ = rα + rβ
[Eα , Eβ ] = (70)
otherwise

Here rα is a root vector associated with Eα , riα one of its components.

It has been proved that in order to completely break the symmetry
and have topologically non-trivial solution it is enough to have one
Higgs field Ψ in the Cartan subalgebra (the one generated by the
Hi ’s) and N − 1 fields ΦA in the orthogonal complement [31]-[32].
With this in mind, let us take as the gauge Higgs action
N −1
!
1
Z X
S = Tr d4 x − Fµν F µν + Dµ ΦA D µ ΦA + D µ ΨDµ Ψ − V (Φ, Ψ)
2
A=1
(71)
where

Aµ = Aaµ ta , Dµ = ∂µ + e[Aµ , ]
Fµν = ∂µ Aν − ∂ν Aµ + e[Aµ , Aν ]
1
[ta , tb ] = if abc tc , Trta tb = δab , a, b = 1, 2, . . . , N 2 − 1(72)
2
The potential should be chosen so as to ensure complete symmetry
breaking.
A nonabelian extension of the Nielsen-Olesen ansatz can be then
proposed in the form

ΦA = f a (r)Ω−1
n (ϕ)(EaA + EaA )Ωn (ϕ) , A = 1, 2, . . . , N − 1
N
X −1
Ψ = Cj Hj
j=1
n
Aϕ = a(r)M , Ar = 0
e
M = diag(1/N, 1/N, . . . , 1/N, (1 − N )/N ) (73)

where M belongs to the Cartan subalgebra. Note that Ψ is taken as

a constant everywhere. One can see that with this ansatz all fields
ΦA satisfy the same equation of motion which, as it happens with
the gauge field one, coincides with those for the Abelian Higgs model
so that their properties are the same as those already discussed for
Nielsen-Olesen vortices.

16
 An “electromagnetic” tensor Fµν can be introduced in order to
characterize the vortices. Following [31]-[32] we choose

Tr(MFµν )
Fµν = (74)
Tr(M2 )

Then, the flux associated to the magnetic field F12 reads, for ansatz
(73),
1 1
Z
Θ= d2 xεij Fij = √ 2πn (75)
2 Ne
It is important to stress that ansatz (73) in fact works for any
n = 2k + m, with m = 1, 2, . . . , N − 1 and k ∈ Z. What is happening
is that the magnetic flux is not a topological number as in the Abelian
case. Indeed, as explained above, the topological charge is an element
of ZN label in this case by m.

The Nonabelian Bogomol’nyi bound

A Bogomol’nyi bound for the nonabelian vortices discussed above was
found in [33]. We here sketch the SU (2) case which can be easily ex-
tended to SU (N ). As explained above, the potential should be chosen
so as to completely break the symmetry. One can take a potential of
the form
1
V (Φ, Ψ) = Tr g(Φ2 − Φ20 )2 + g′ (Ψ2 − Ψ20 )2 + g′′ (ΦΨ)2


(76)
4
minimum for the energy will take place when

Tr(ΦΨ) = 0 (77)

The Higgs field Ψ in (73) does not play any dynamical role since the
ansatz corresponds to the condition

Di Ψ = 0 (78)

everywhere. Then, the energy E (per unit length) of a general config-

uration is always bounded by that satisfying the ansatz (73) according
to
 
1 2 g
Z
2 2 2 2
E ≥ Φ0 Tr d x F + Di ΦDi Φ + + 2 λ(Φ − 1) (79)
2 ij 4e

17
Now, as usual, the r.h.s. in (79) can be written as a sum of perfect
squares plus additional surface terms/. After some work one finds
 2
1~ α
Z
E ≥ Φ0 d x2 2
Fij + 2 ~
εij (Φ − 1)Ψ +
4 1 + α2
1 
~ + bεij Ψ
~ ∧ D~j Φ −
2 1 ~ij .Ψ
~ +
2
D i Φ εij F
2(1 + α ) 2(1 + α2 )
α2
  
2 2
λ− (Φ − 1) (80)
2(1 + α2 )
with α, an arbitrary constant, should be chosen so that the term in
the last line is semipositive definite. This means
1
|α| = 1 if λ ≥
8
α2 1
= λ if λ ≤ (81)
2(1 + α2 ) 8
Being the last term in the second line of (80) related to the magnetic
flux (75) the energy bounded can be written as
Φ20 1
E ≥ ∓ πn if λ ≥
2 8
√ 2 1
E ≥ ∓ 2λΦ0 2πn if λ ≤ (82)
8
and the bond saturates whenever the following BPS first order eqs.
hold
α ~ = 0
F~ij + εij (1 − Φ2 )ψ
1 + α2
~ + αεij Ψ
Di Φ ~ ∧ Dj Ψ
~ = 0 (83)
Now, one can easily see that these two eqs. are compatible if and only
if α = ±1 this implying
e2
λ= =1 (84)
8π
and we again find the Bogomol’nyi point as a necessary condition for
bounding the energy.
In summary, the complete set of BPS equations can be written as
1 ~ = 0
F~ij ± εij (1 − Φ2 )ψ
2
~ ± εij Ψ
Di Φ ~ ∧ Dj Ψ
~ = 0
Di Ψ = 0 (85)

18
with the double sign related to the n sign. At the bound, the energy
is again proportional to the magnetic flux

E = Φ20 π|n| (86)

but, as explained above, |n| is not the topological charge. Indeed, the
topological charge is, in this case, an integer modulo 2 and hence the
more stable solutions other than the trivial vacuum are those with
n = ±1.

2 Nonabelian Chern-Simons vortices

Non-Abelian Chern-Simons vortices have been first discussed in [31]-
[32]. Let us start by writing the Chern-Simons action for the case of
an SU (N ) gauge theory in d = 3 space-time dimensions,
 
µ αβγ 2
Z
3
SCS [A] = ε Tr d x Fαβ Aγ − Aα Aβ Aγ (87)
2 3
This is not a gauge invariant action. Indeed, under a gauge transfor-
mation
i
Agµ = g−1 Aµ g + g−1 ∂µ g (88)
e
the action changes according to

SCS [Ag ] = SCS [A] + 8π 2 µω[g] (89)

with ω[g] the winding number associated to the gauge transformation,

1
Z
ω[g] = Trεαβγ
g−1 ∂α gg−1 ∂β gg−1 ∂γ g = m ∈ Z (90)
24π 2
Since for quantizing the theory one needs that

exp(iSCS [Ag ]) = exp(iSCS [A]) (91)

(with ~ = 1) one needs the coefficient µ to be quantized

e2 m
µ= , m∈Z (92)
4π
Note that this condition is valid both in Minkowski and in Euclidean
space since the CS term does not depend on the metric and is then
unaffected by a Wick rotation.

19
 We can then add the Chern-Simons term to the Yang-Mills Higgs
system (71) (considered in 2 + 1 dimensions)
N −1
1
Z X
3 µν
S = Tr d x − Fµν F + Dµ ΦA D µ ΦA + D µ ΨDµ Ψ
2
A=1
!
− V (Φ, Ψ) + SCS [A] (93)

and look for vortex like solutions which, as in the Abelian case, should
now carry not only magnetic but also electric flux.
We shall again propose an axially symmetric ansatz as (73) but
this time including also the A0 field,

ΦA = f a (r)Ω−1
n (ϕ)(EaA + EaA )Ωn (ϕ) , A = 1, 2, . . . , N − 1
N
X −1
Ψ = Cj Hj
j=1
n
Aϕ = a(r)M , Ar = 0
e
n
A0 = a0 (r)M
e
M = diag(1/N, 1/N, . . . , 1/N, (1 − N )/N ) (94)

Given the “electromagnetic tensor” defined as in (74)

Tr(MFµν )
Fµν = (95)
Tr(M2 )

with a magnetic flux defined as in (75),

1 2π
Z
Θ= d2 xεij Fij = √ n (96)
2 Ne
we have now an electric charge Q which can be defined as follows. We
first define the electric field from (95) as

Ei = Foi (97)

which can be seen to reduce, with our axially symmetric ansatz to

1 da0
Er = F0r = − √ (98)
N dr

20
Then, from the Gauss law,
dEr
+ σ = µB (99)
dr
where
1
B = εij Fij , σ = e2 f 2 a0 (100)
2
since limr→∞ Er = 0 according to the boundary conditions, one gets,
from (99) a relation between the charge Q,
Z
Q = d2 xσ (101)

and the flux,

Q = µΦ (102)
Now, due to the quantization condition (92), eq.(102) becomes

Q = mnQ0 , m, n ∈Z
e
Q0 = √ (103)
2 N
Charge quantization can be connected with the angular momentum J
of the vortex, Z
J= d2 xεij xi Toj (104)

with Tµν the energy momentum tensor. One can easily find that

2Q 1
J = −√ =− nm (105)
Ne 2N

Starting from ansatz (94) solutions can be constructed exactly as

in the previous cases. Both the magnetic field and the Higgs scalar Φ
have the same qualitative behavior as for neutral vortices. Concerning
the electric field, it can be related with that of the abelian Chern-
Simons model already discussed.
Also as in the Abelian case, one can find Bogomol’nyi equations
and a bound starting from a Lagrangian where the Yang-Mills term
is absent and choosing a sixth-order potential for the Φ field [34].
restricting to the SU (2) case one starts from

Z
S = SCS [A] + Tr d4 x (Dµ Φa D µ Φa + Dµ ΨDµ Ψ − V (Φ, Ψ)) (106)

21
with
V (Φ, Ψ) = V1 [Φa Φa ] + V2 [Ψa Ψa ] + g(Ψa Φa )2 (107)
Again the Ψ field will be taken as an spectator just present to com-
pletely break the symmetry. It will be chosen so as to make vanish V2 .
The third term precisely ensures symmetry breaking. Concerning V1 ,
as in the Abelian case, if one wants to obtain first order Bogomol’nyi
equations one needs a sixth order potential with a coupling constant
which is related to the gauge coupling constant

e4 a a a a 2 2


V1 [Φ] = Φ Φ Φ Φ − Φ 0 (108)
8µ2
The energy for a static configuration is given by
 2 
e ~ 1
Z Z
2
E = d xT00 = d x 2 ~ 2 ~ 2 ~
(A0 ∧ Φ) + (Di Φ) + V1 [|Φ|) (109)
2 2
Now the Gauss law,

~ ∧ D 0 Φ)
~ = µ ~ ~
e(Φ εij Fij ≡ µB (110)
2
allows to write
~0 ∧ Φ
~ = µ ~ ~
A B∧Φ (111)
e2 |Φ|2
so that the energy becomes

µ2
 
1
Z
E= d x 2 ~ 2 ~ 2 ~
|B| + (Di Φ) + V1 [|Φ|) (112)
2e2 |Φ|2 2
which can be written a la Bogomol’nyi in the form

1
Z
E = 2
d x ~ ± εij Ψ̌ ∧ Dj Φ)
(Di Φ ~ 2+
4
µ2 e3

2 2 2 2 e 2 ~ (113)
(B ∓ |Φ| (|Φ| − Φ 0 )Ψ̌) ∓ Φ (Ψ̌.Φ)
2e2 |Φ|2 2µ2 2 0
We recognize in the last term the magnetic flux
1 ~ = 2π n
Z
Θ= d2 xΨ̌.B (114)
2 e
so that the energy is again bounded by the flux,

E ≥ Φ20 πn (115)

22
and the bound is attained whenever the first order BPS equations
hold,
~ ± εij Ψ̌ ∧ Dj Φ
Di Φ ~ = 0
3
~ ∓ e |Φ|2 (|Φ|2 − Φ20 )Ψ̌ = 0
B
2µ2
Di Ψ~ = 0 (116)
Qualitatively, the vortex solutions to this system are the same as those
solving the second-order Euler-Lagrange equations described in the
precedent section.

More on non-Abelian vortices and

semilocal strings
We have already constructed semilocal vortices for the case in which
the gauge group was U (1)local but one considers an SU (2)global doublet
of complex scalars. One can extend the construction to a non-Abelian
gauge group U (N )local provided the global group SU (Nf )global is such
that Nf > N .
Before explaining this, we shall start buy describing a class of non-
Abelian vortices different from those discussed in sections 5 and 6.
They were originally discussed in refs.[35]-[37]. Let us consider gauge
fields taking values in an U (Nc )gauge theory and scalars φai carrying
not only a color index a = 1, 2, . . . , N but also a flavor index i =
1, 2, . . . , Nf . We shall first consider N = Nf and write these scalars
as an N × N matrix,
Φ = (φai ) , a, i = 1, 2, . . . , N (117)
The action is
e2

1
Z
4 † µ
S = d x Tr(F µν
Fµν ) + Tr(Dµ Φ D Φ) − Tr(Φ† T A Φ)2
4e2 8
e2

† 2 2
− (Tr(Φ Φ) − Φ0 ) (118)
8
Here T A , A = 1, 2, . . . , N are the SU (N ) generators normalized so
that Tr(T A T B ) = 1 and
i i
Dµ Φ = (∂µ − A0µ − AA T A )Φ (119)
2 2 µ

23
The theory has a U (N )local × SU (N )global symmetry. On the scalars
this acts according to

Φ → U ΦV , U ∈ U (N )local V ∈ SU (N )global (120)

Concerning symmetry breaking, the last term in (118) forces Φ to

develop a vacuum expectation value while the last but one forces it to
diagonal,
Φvac = Φ0 δia (121)
This vev is preserved only for transformations in which U = V −1
that is, one performs a global gauge transformation and a related
global flavor transformation simultaneously. The pattern of sponta-
neous symmetry breaking is then

U (N )local × SU (N )global → SU (N )diag (122)

This mechanism was known as color-flavor locking in a QCD context,

[39]-[40].
Let us now discuss the vortex solutions of this model. For simplic-
ity, we shall consider the case N = 2 [36] but the generalization to
SU (N ) will be evident.
Since there is a U (1) gauge symmetry which is spontaneously bro-
ken, there are of course Abelian Nielsen-Olesen vortex solutions which
we have already studied. But there are other novel solutions, related
to the non-Abelian character of the theory which are obtained as fol-
lows.
Let us take the vacuum (121) and make just one of its components,
to wind. This, in the particular N = 2 case means
 
exp(iϕ) 0
Φv = Φ0 (123)
0 1

Once one has such a vacuum, it is easy to write an ansatz for the
Higgs scalar everywhere,
 
f1 (r) exp(iϕ) 0
Φ = Φ0 (124)
0 f2 (r)

such that

lim f1 (r) = lim f2 (r) = Φ0

r→∞ r→∞

f1 (0) = f2 (0) = 0 (125)

24
Concerning the ansatz for the gauge field,

 
SU (N ) 1 0
Ai = (1 − a3 (r))∂i ϕ
0 −1
U (1)
Ai = (1 − a(r))∂i ϕ (126)

with
lim a(r) = lim aN (r) = 0
r→∞ r→∞

a(0) = aN (0) = 1 (127)

This ansatz, in which Φv 11 is the scalar components that winds is
usually called a (1, 0) string. If one chooses Φv 22 to be the sole topo-
logically nontrivial component will have a (0, 1) string,
 
f1 (r) 0
Φv = Φ0 (128)
0 f2 (r) exp(iϕ)

Let us note that if we define

1
Ωi [Φ] = (∂i Φ) Φ−1 (129)
i
then, for the (1, 0) string one has
 
1 0
Ωi [Φv ] = ∂i ϕ (130)
0 0

Then, if we associate Ωi with a topological charge T through the

formula I
T = Tr Ωi dxi (131)

so that
T(1,0) = T(0,1) = 2π (132)
Let us note that if instead of vacuum (123) or (128) one proposes an
“Abelianized” one
 
exp(iϕ) 0
Φv = Φ0 = exp(iϕ)I (133)
0 exp(iϕ)

one would obtain for T

T(1,1) = 4π = T(1,0) + T(0,1) (134)

25
 Now, since the covariant derivative (119) has to vanish at infinity,
 
i 0 i A A
Di Φ|∞ = ∂i − Ai − Ai T Φv | ∞ = 0 (135)
2 2

one easily finds that

1 1
(∂i Φv )Φ−1 A0i + AA A 


v = i T ∞
(136)
i 2
1
I I
T = Tr (Ai + Ai T )dx = Ai dxi
A A i
(137)
2
Note that only the U (1) component of the gauge field contributes to
the topological charge We then conclude that the magnetic flux Φ of
these vortex solutions is precisely connected to T,
1
Ω= T (138)
e
and in accordance, one can see that the Bogomol’nyi bound will turn
to be
1
E ≥ Φ0 Ω = Φ0 T (139)
e

3 Bogomol’nyi equations and Super-

symmetry algebra
We already mentioned at the beginning that the Bogomol’nyi point
(1),
e2 = 8λ (140)
at which solutions to a first-order set of equations (Bogomol’nyi eqs.)
exist and an exact vortex solution is known, not only corresponds to
the limiting point separating type-I from type-II superconductivity,
but, as noticed in [12], coincides with that which has to be chosen in
order to have a supersymmetric extension of the model
This connection between Bogomol’nyi equations and bounds and
supersymmetry began to be clarified after the work of Witten and
Olive [41]. Let us then review the results in that paper which starts
with a two 1 + 1 dimensional supersymmetric model with Lagrangian
1
(∂µ φ)2 + ψ̄ 6 ∂ψ + V 2 [φ] + V ′ [φ]ψ̄ψ


L=− (141)
2

26
where ψ is a Majorana fermion and V [φ] an arbitrary function. Our
metric is diag gµν = (−1, 1), (µ = 0, 1). The gamma matrices are
   
0 2 0 −1 1 3 1 0
γ = −iσ = , γ =σ = (142)
1 0 0 −1

The charge conjugation matrix C is

C = σ 2 , (γ µ )t C = −Cγ µ (143)

Here σ i are the standard Pauli matrices. We write the Majorana

spinor in the form  +
ψ
ψ= (144)
ψ−
and then
ψ̄ = ψ t C = ψ − −ψ +


(145)
The equations of motion are,
1
φ = V V ′ + V ′′ ψ̄ψ
2
6 ∂ψ − V ′ ψ = 0 (146)

Under the following supersymmetry transformations

δφ = ǭψ
δψ = (6 ∂φ − V [φ]) ǫ
δψ̄ = −ǭ (6 ∂φ + V [φ]) (147)

the Lagrangian changes as a total divergence and then one finds, using
Noether theorem that the conserved supersymmetry current is

J µ = − (6 ∂φ + V [φ]) γ µ ψ (148)

so that the conserved supercharge is

Z
Q = − dx (6 ∂φ + V [φ]) γ 0 ψ (149)

Writing
Q+
 
Q= (150)
Q−

27
one finds from (149)
Z
Q+ = dx ∂0 φ ψ + + (∂1 φ + V [φ])ψ −

Z
Q− = dx ∂0 φ ψ − − (∂1 φ − V [φ])ψ +


(151)

Calling H the Hamiltonian and Z the quantity

∂φ
Z
Z = dxV [φ] (152)
∂x
one finds
{Q± , Q± } = 2(H ± Z) (153)
where we have used

{ψ ± (x, t), ψ ± (cy, t)} = δ(x − y) (154)

Note that Z can be written in the form

∂φ ∂F [φ]
Z Z
Z = dxV [φ] = dx (155)
∂x ∂x

with (∂F [φ])/(∂φ) = V [φ]. Then, taking as an example the usual

symmetry breaking potential with two ground states φ = ±a,

V = −(φ2 − a2 ) (156)

one has
1
F = λ(a2 φ − φ3 ) (157)
3
and so ∞  
∂ 1
Z
Z=λ dx a φ − φ3
2
(158)
−∞ ∂x 3
Consider static configurations. This integral vanishes for topologically
trivial states, φ(x = −∞) = φ(x = ∞) = ±a. If instead one takes
φ(x = −∞) = ∓a and φ(x = ∞) = ±a one has Z = ±4λa3 /3. One
associates the positive value for a kink state and the negative value
for anti-kink state. Appropriately normalized, the value of T can be
related to the usual kink topological charge defined as
Z ∞
1 ∂φ
T = dx = φ(x = ∞) − φ(x = −∞) (159)
2a −∞ ∂x

28
 Eq.(153) implies, for a particle with rest mass M = H that

M ≥ |Z| (160)

and the bound is saturated for those states |physi such that

Q+ |physi = 0 or Q− |physi = 0 (161)

so that physical soliton or anti-soliton states such that (162) holds are
annihilated by Q+ or Q− . Now, eq.(161) implies, according to (151)
(in the static case),

(∂1 φ + V [φ])ψ− = 0 or (∂1 φ − V [φ])ψ+ = 0 (162)

and then one has the following Bogomol’nyi eqs.

∂1 φ = −V [φ] or ∂1 φ = V [φ] (163)

Let us note at this point that the energy for static configurations (the
rest mass M) in the bosonic sector takes the form
1
Z
M = dx ∂1 φ)2 + V 2 [φ]


(164)
2
and can be trivially written a la Bogomolnyi in the form
1
Z
M = dx ((∂1 φ) ± V [φ])2 ∓ Z (165)
2
from which the bound (160) can be equally obtained.
Of course, the solution to these BPS equations also solve the second
order Euler-Lagrange eqs. Moreover, one can look at eq.(163) as the
square root of the equation of motion for static bosonic configurations.
One can see this by integrating the equation for configurations such
that ∂φ/∂x and V [φ] vanish at infinity

∂12 φ = V V ′ → (∂1 φ)2 = V [φ]2 (166)

Note that since the fermionic supersymmetry transformation law

(147) can be written, for static configurations, as
  +
∂1 φ − V 0 ǫ
δψ = (167)
0 −∂1 φ − V ǫ−
putting this variation to zero is an alternative way to obtain the Bo-
gomol’nyi eqs without necessity of constructing the supercurrent.

29
 One can easily see that there is a zero mode for the Dirac equation
associated with each Bogomol’nyi equation. These zero modes are
 +  
χ0 0
χ0 = (6 ∂φ − V ) or χ0 = (6 ∂φ + V ) (168)
0 χ−
0

where χ± 0 are constant components. One can understand this by not-

ing that the kink (antikink) solution of the Bogomol’nyi eqs.(163)
solves, together with ψ = 0, the coupled Euler-Lagrange eqs. of the
model. But a supersymmetric transformation of such solution will still
be a solution and this means that δψ (the SUSY transform of ψ = 0)
will solve the Dirac equation. Such δψ is precisely (167).
Let us end by stressing that eqs.(153) correspond to an N = 2
extended supersymmetry algebra with central charge Z; the equality
M = Z when Bogomol’nyi bound is saturated eliminates half of the
SUSY generators (either Q− or Q+ ) which in turn is associated with
a Dirac equation zero-mode.

Supersymmetry and ANO vortices

We are now ready to see how the vortex BPS equations for the Abelian
Higgs model can be derived by SUSY considerations. We shall follow
here refs.[42]-[43]. One has here two possibilities; either one studies
the N = 2 supersymmetric extension of the d = 2 + 1-dimensional
model looking at vortices as static solutions of eqs. of motion or one
analyzes the 2-dimensional (Euclidean) model in which vortices can
be seen as instanton solutions. Let us start by discussing the d = 2 + 1
approach.

The 2 + 1 N = 2 supersymmetric model

We take as metric diaggµν = (+ − −). A two component Majorana
spinor will be written as
 +
α ψ
(ψ ) = (169)
ψ−

Spinorial indices are raised and lowered using Cαβ = iεαβ with ε−+ =
1,
ψ α = C αβ ψβ (170)

30
Dirac matrices are chosen as

γ 1 = σ1 , γ 2 = σ2 , γ 0 = σ3 , (171)

with σ i the standard Pauli matrices. Superspace coordinates are writ-

ten as (xµ ; θ α ) with θ α two anticommuting spinor coordinates.
The Higgs field φ, together with a higgsino ψ and an auxiliary field
F are accommodated in a complex scalar superfield,
1
Φ(x; θ) = φ(x) + θ̄ψ(x) − θ̄θF (x) (172)
2
The gauge field Aµ together with a photino ρ are accommodated in a
vector superfield

Γα (x; θ) = iAµ (x)(γ µ θ)α − θ̄θρα (x) (173)

Finally a real scalar superfield will be necessary in order to implement

the Fayet-Iliopoulos gauge symmetry breaking mechanism,
1
S(x; θ) = N (x) + θ̄χ(x) − θ̄θD(x) (174)
2
where N is a scalar, χ is a Majorana fermion and D an auxiliary field.
Calling Dα the (super) covariant derivative,

∂
Dα = + i(γµ θ)α ∂µ (175)
∂θ α
the spinorial electromagnetic superfield Wα is given by
1 i i
Wα = Dβ Dα Γβ = ρα (x)+ εµνσ F νσ (x)(γ µ θ)α − θ̄θ(6 ∂ρ)α (x) (176)
2 2 2
With all this the N = 1 supersymmetric version of the Abelian Higgs
model Lagrangian takes, in terms of superfields, the form

1 α 1
Z
LN =1 = d2 θ W Wα − (D α + ieΓα ) Φ∗ (Dα − ieΓα ) Φ
2 4
√

1 α ∗
− D SDα S + 2λSΦ Φ + ηS (177)
4

where η is the real Fayet-Iliopoulos coefficient. Integration out over θ

one gets the N = 1 supersymmetric Lagrangian in components which

31
takes the form
1 1 1 1
LN =1 = − F µν Fµν + ∂ µ N ∂µ N + (D µ φ)∗ Dµ φ + D 2
4 2 2 2
√ 1 √
+ 2λD|φ|2 + ηD + |F |2 + 2λN (F ∗ φ + F φ∗ )
2
i i i √
ρ̄ 6 ∂ρ + χ̄6 ∂χ + ψ̄ D
6 ψ − 2λN ψ̄ψ
2 2 2
ie √
+ (ψ̄ρφ − ρ̄ψφ∗ ) − 2λ(ψ̄χφ + χ̄ψφ∗ ) (178)
2
Solving the equations of motion for the auxiliary fields one gets
√ √
D = 2λ|φ|2 + ξ , F = −2 2λN φ (179)

Here ξ is the Fayet-Ilopoulos parameter which will be written as

√
ξ = − 2λφ20 (180)

so that symmetry breaking takes place for

|φ| = φ0 > 0 (181)

With this, Lagrangian (178) takes the form

1 1 1
LN =1 = − F µν Fµν + ∂ µ N ∂µ N + (Dµ φ)∗ Dµ φ − 4λN 2 |φ|2
4 2 2
2 2 i i i √
−λ(|φ| − φ0 ) + ρ̄ 6 ∂ρ + χ̄ 6 ∂χ + ψ̄ D
6 ψ − 2λN ψ̄ψ
2 2 2
ie √
+ (ψ̄ρφ − ρ̄ψφ∗ ) − 2λ(ψ̄χφ + χ̄ψφ∗ ) (182)
2
This Lagrangian is invariant under the following infinitesimal super-
symmetry transformations
i
δρ = − εµνα F να γ µ ǫ , δAµ = −iǭγµ ρ (183)
2
√
δN = ǭχ , δχ = − 2λ(|φ|2 − φ20 )ǫ − iγ µ ǫ∂µ N (184)
√
δφ = ǭψ , δψ = −iγ µ ǫDµ φ − 8λN φǫ (185)
with ǫ an anticommuting real (i.e. Majorana spinor) parameter.
It is possible to extend this N = 1 supersymmetry to an N = 2
supersymmetry. This can be achieved by considering SUSY transfor-
mations (183)-(185) where the parameter ǫ is replaced by a complex

32
one, ǫc and real fermions χ and ρ are combined into a Dirac fermion
Σ,
Σ = χ − iρ (186)
In terms of this new field the Lagrangian (182) becomes
1 1 1
LN =2 = − F µν Fµν + ∂ µ N ∂µ N + (D µ φ)∗ Dµ φ − 4λN 2 |φ|2
4 2 2
2 2 i i √
−λ(|φ| − φ0 ) + Σ̄ 6 ∂Σ + ψ̄ D 6 ψ − 2λN ψ̄ψ
√ 2 2 √
e + 8λ e − 8λ c
− (Σ̄φ + h.c.) + (Σ̄ φ + h.c.) (187)
4 4
where Σc is the charge conjugate (the complex conjugate) of Σ.
Concerning the SUSY transformations (183)-(185), they take the
form, in terms of Σ,
√
 
1 2 2
δ̂Σ = − να µ
εµνα F γ + 2λ(|φ| − φ0 ) + i 6 ∂N ǫc (188)
2
1 c 1
δ̂Aµ = (ǭ γµ Σ + h.c.) , δ̂N = (ǭc Σ + h.c.) ; , δ̂φ = ǭc ψ (189)
2 2
 √ 
δ̂ψ = − iγ µ Dµ φ + 8λN φ ǫc (190)

Here ǫc is a complex parameter which can be written as

ǫc = ǫ exp(iα) (191)

and so (190) can be interpreted as transformations (183)-(185) with

real ǫ followed by a phase transformation for fermions,

Σ → exp(iα)Σ , ψ → exp(iα)ψ (192)

However the last term in (187) is not invariant under this phase rota-
tion,
ψ̄Σc φ + h.c. → exp(2iα)ψ̄Σc φ + h.c. (193)
so one needs
e2
λ= (194)
8
as necessary and sufficient condition for the N = 2 SUSY. This was
a well-known result holding in general when, starting from an N = 1
supersymmetric gauge model, one attempts to impose a second SUSY
(see for example [44]): conditions on coupling constants have to be

33
imposed so as to accommodate different N = 1 multiplets into an N =
2 multiplet. Eq.(194) is an example of such conditions. These kind
of conditions for the Abelian Higgs model were obtained, following
different routes, in [13]-[14],[45].
We shall now study the supersymmetry algebra in order to reobtain
the results above in a way closer to the one used by Olive and Witten
for their 1+1 model.
The Noether current associated with supersymmetry invariance
can be compactly written in the form
X  δL δL

Jµ = δΦ + δΨ − Λµ (195)
δ∇µ Φ δ∇µ Ψ
Φ,Ψ

where Φ and Ψ represent generic bosonic and fermionic fields and is

included to take into account the possible variation of the Lagrangian
through a divergence term,

δL = ∂µ Λµ (196)

From this, the conserved supersymmetric charge Q can be defined as

Z
Q = d2 xJ 0 (197)

Writing
Q = ǭc Q + Q̄ǫc (198)
in order to take rid of the infinitesimal Grassman parameter we find
from Lagrangian (187)
 
1 µνλ e
Z
2
Q = d x − ε Fµν γλ + i 6 ∂N − (|φ| − v ) γ 0 Σ+
2 2
2 2

∗ e ∗

0

i(6D φ) − N φ γ ψ (199)
2
Concerning the conjugate charge Q̄ one has
 
1 µνλ e
Z
2 0 2 2
Q̄ = d xΣ̄γ − ε Fµν γλ − i 6 ∂N − (|φ| − v ) +
2 2
0
 e 
ψ̄γ −i D 6 φ − Nφ (200)
2
If we restrict the model to the case N = 0 and, after computing the
Poisson brackets we put all fermions to zero we end with the original

34
Abelian Higgs model. Since we are interested in static configurations
with A0 = 0 we impose these conditions finding the following anti-
commutation relation among spinor supercharges,

{Qα , Q̄β } = 2 (γ0 )αβ P 0 + δαβ T (201)

where
e2
 
1 2 1
Z
0 2
P =E= d x Fij + |Di φ|2 + (|φ|2 − φ20 )2 (202)
4 2 8

wile the central charge T is given by

Z e 
T = d2 x εij Fij (|φ|2 − φ20 ) + iεij Di φ(Dj φ)∗ (203)
2
with i, j = 1, 2. One can easily see that T can be rewritten in the form
Z
T = d2 x∂i Ωi (204)

with
Ωi = εij eφ20 Aj + iφ∗ Dj φ


(205)
so that using Stokes theorem (and taking into account that Di φ → 0
at infinity) one ends with
 
2πN
I
T = eφ20 Ai dxi = eφ20 , N ∈Z (206)
e

where the integer n characterizes the homotopy class to which Ai

belongs. To find the Bogomolnyi bound from the supersymmetry al-
gebra, let us observe that, being the anticommutators (201) Hermitian
one has
{Qα , Q̄β }{Qα , Q̄β } ≥ 0 (207)
Then, using (201) one has

4E 2 − T 2 ≥ 0 (208)

or
|T |
E≥ (209)
2
which is nothing but the bound (26) originally obtained by Bogomolnyi
by completing squares. The bound is attained for states |physi which

35
are annihilated by the charges. To extract from this condition the
Bogomolnyi eqs., it will be convenient to write
 
Q+
Q= , Q̄ = (Q̄+ Q̄− ) (210)
Q−
after a little work one finds
1 e 
Q+ = − εij F ij Σ+ − (|φ|2 − v 2 ) Σ− − i(D1 φ)∗ ψ+ + (D2 φ)∗ ψ−
2 2
1 e 
Q− = − εij F ij Σ− − (|φ|2 − v 2 ) Σ+ − i(D1 φ)∗ ψ− + (D2 φ)∗ ψ+
2 2
(211)
Hence, if one combines spinor components so that
QI = (Q+ + iQ− ) , QII = (Q+ − iQ− ) (212)
one has
 
1 e
QI = − εij F ij + (|φ|2 − v 2 ) ΣI − (i (D1 φ)∗ − (D2 φ)∗ ) ψI
2 2
 
1 e
QII = − εij F − (|φ| − v ) ΣII − (i (D1 φ)∗ + (D2 φ)∗ ) ψII
ij 2 2
2 2
(213)
We see that if the bosonic fields satisfy
1 e
εij F ij + (|φ|2 − v 2 ) = 0
2 2
i (D1 φ)∗ − (D2 φ)∗ = 0 (214)
QI vanishes (while QII not). But if we put ǫcII = 0 we then have
Q|phys >= 0 for any configuration satisfying (214). This in turn will
suffice to ensure that the bound (207) is attained.
Analogously, the supersymmetric charge will annihilate physical
states if
1 e
εij F ij − (|φ|2 − v 2 ) = 0
2 2
i (D1 φ)∗ + (D2 φ)∗ = 0 (215)
together with ǫcI = 0. But (214) and (215) are nothing but Bogo-
mol’nyi equations so that we see that when the Bogomol’nyi bound
is attained half of the supersymmetry is lost. Again, a fermionic zero
mode arises associated with this supersymmetry breaking and its form
can be just inferred by SUSY transforming a trivial (zero) solution as
in the Witten-Olive example we discussed.

36
Uniqueness of Bogomol’nyi equations
I will discuss here how the BPS structure of general gauge theories
(depending on the Maxwell invariant F µν Fµν and F̃ µν Fµν ) coupled
to Higgs scalars is insensitive to the particular form of the gauge La-
grangian. Indeed, analyzing their supersymmetric extension, one can
explicitly understand why this happens. Hence Maxwell, Born-Infeld
or more complicated non-polynomial Lagrangians all satisfy the same
Bogomol’nyi equations and bounds which are dictated by the under-
lying supersymmetry algebra ([46]).
Due to the interest aroused by Born-Infeld theories in the context
of the dynamics of D-branes (see [48] and references there) their Bogo-
mol’nyi equations for different cases were investigated [49]- [51] (early
constructions were reported in [52]-[53]). Bogomol’nyi equations were
found to coincide with those arising in ordinary gauge theories. Con-
cerning SUSY Born-Infeld theories, they were originally studied in
[54]-[55] (see also [56]).
One can quickly understand why BPS equations are not sensitive
to the dynamics that one chooses for the gauge field by noting that
they can be derived, in a SUSY framework, by imposing the vanishing
of (half of) the supersymmetry variations of the gaugino and higgsino
fields and these variations are formally the same for very different
Lagrangians. The dynamics associated with the Lagrangian enters
however through the equation of motion for the auxiliary field D (of
the gauge field supermultiplet) which appears in the supersymmetric
transformation law for the gaugino. It is then through D that the
form of the Lagrangian may in principle determine the form of the
BPS relations.
To see this in detail, let us take as an example the Abelian-Higgs
model in d = 3 dimensions, for which Bogomol’nyi equations were first
derived [15],[12] and which we have already discussed in section 2. The
arguments should hold, however, for other models like for example the
SO(3) gauge theory and in other dimensionalities of space-time.
We shall first consider d = 4 dimensional Minkowski space (with
signature (+, −, −, −)) so that one could also use the results to analyze
other d = 4 models and then proceed to dimensional reduction to
d = 3. The gauge vector superfield V is written, in the Wess-Zumino
gauge,
1
V = −θσ µ θ̄Aµ + iθθ θ̄λ̄ − iθ̄ θ̄θλ + θθ θ̄θ̄D (216)
2
Here Aµ is a vector field, λ = (λα ) and λ̄ = (λ̄α̇ ) are two-component

37
spinors (α, α̇ = 1, 2) which can be combined to give a four-component
Majorana fermion and D is an auxiliary field.
From V the chiral superfield Wα can be constructed,
i
Wα y, θ, θ̄ = −iλα + θα D − (σ µ σ̄ ν θ)α Fµν + θθ σ µ ∂µ λ̄ α



(217)
2
Here λ, λ̄, D and Fµν = ∂µ Aν − ∂ν Aµ are functions of the variable
y µ = xµ +iθσ µθ̄ where xµ is the usual 4-vector position. The SUSY ex-
tension of (standard) gauge-invariant (Maxwell, Yang-Mills) theories
are precisely constructed from W by considering W 2 and its hermitian
conjugate W̄ 2 .
Now, as stressed in [54], another superfield combination enters
into play if one wishes to construct general gauge invariant SUSY
Lagrangians. In particular, one needs to consider two superfields X
and Y defined as
1
X = (D α Dα W 2 + D̄α̇ D̄α̇ W̄ 2 ) (218)
8
i
Y = − (D α Dα W 2 − D̄α̇ D̄ α̇ W̄ 2 ) (219)
16
with covariant derivatives given by
∂
∂ ∂
Dα = α
+ 2i σ µ θ̄ α µ , D̄α̇ = − (220)
∂θ ∂y ∂ θ̄ α̇
when acting on functions of (y, θ, θ̄) and
∂ ∂ ∂
Dα = , D̄α̇ = − − 2i (θσ µ )α̇ †µ (221)
∂θ α ∂ θ̄ α̇ ∂y
on functions of (y † , θ, θ̄). The only components of these superfields
having purely bosonic terms are
1 1
W 2 θθ = D 2 − F µν Fµν − i F µν F̃µν

(222)
2 2
1 1
W 2 θθ = D 2 − F µν Fµν + i F µν F̃µν

(223)
2 2

1
X|0 = −(D 2 − F µν Fµν − iλ∂/ λ̄ − iλ̄∂λ) /̄
2
1
X|θθ̄ = iθσ p θ̄∂p (D2 − F µν Fµν − iλ∂/ λ̄ − iλ̄∂λ) /̄
2
1 1
X|θθ̄θθ̄ = θ θ̄θ θ̄  (D 2 − F µν Fµν − iλ∂/ λ̄ − iλ̄∂λ)
/̄ (224)
4 2

38
and
1 1 µν
Y |0 = ( F F̃µν + λ∂/λ̄ − λ̄∂λ) /̄
2 2
i 1
Y |θθ̄ = − θσ p θ̄∂p ( F µν F̃µν + λ∂/λ̄ − λ̄∂λ) /̄
2 2
1 1
Y |θθ̄θθ̄ = θ θ̄θ θ̄  ( F µν F̃µν + λ∂/λ̄ − λ̄∂λ)
/̄ (225)
8 2
with F̃µν = (1/2)εµναβ F αβ .
A third superfield combination is necessary for constructing general
gauge invariant SUSY Lagrangian. This combination is W 2 W̄ 2 with
its highest component taking the form
 
1 1
W 2 W̄ 2 |θθθ̄θ̄ = θθ θ̄θ̄ (D 2 − Fµν F µν )2 + ( F̃µν F µν )2 (226)
2 2
Remark that in all three cases (224)-(226), all dependence on the
curvature Fµν and the auxiliary field D is through the combination
 
1 1
t = 2 D 2 − F µν Fµν (227)
β 2
and this fact will have important consequences in our discussion. Here,
in order to define a dimensionless variable t we have introduced a
parameter β with the same dimensions as Fµν (i.e. dimensions of a
mass in d = 4). It corresponds to the absolute field in the Born-Infeld
theory [57]-[58] as will become clear below.
As it happens for the last component of W 2 W̄ 2 , also the term W 2
2
(W̄ ) depends on Fµν and D through the combination (227). Indeed,
the last component in W 2 (W̄ 2 ) contains the term D 2 − 12 F µν Fµν +
iFµν F̃ µν (D 2 − 12 F µν Fµν − iFµν F̃ µν ) so that the sum of θ (θ̄ )integrals
leads to the well-known SUSY extension of the Maxwell theory.
We are ready to write a general N = 1 supersymmetric Lagrangian
endowed with gauge-invariance in terms of X, Y and W 2 W̄ 2
∞
1 1 X
Z Z
2 2 2 2

r
d=4
arst d4 θ W 2 W̄ 2 X s Y t


L = 2 W d θ + W̄ d θ̄ + 2
4e e
r,s,t=0
(228)
with e the fundamental gauge coupling constant, which has been fac-
torized in both terms for later convenience.
The second term accounts for the non-polynomial features of the
general bosonic theory to be supersymmetrized. As explained in [54],

39
supersymmetry imposes two constraints on coefficients arst . Their ex-
plicit form will not be relevant for our discussion. What one should
retain is that expression (228) gives then the most general Lagrangian
corresponding to the supersymmetric extension of a general bosonic
Lagrangian depending on the two algebraic Maxwell invariants F µν Fµν
and F̃ µν Fµν .
We shall focus now on a d = 3, N = 2 supersymmetric theory
which can be obtained from Lagrangian (228) by dimensional reduc-
tion. The standard procedure for dimensional reduction, say in the
x3 spatial coordinate, implies identifying A3 with a scalar field N .
It can be shown that without including a Chern-Simons term, the
bosonic part of the Lagrangian (228) can only yield electrically neu-
tral configurations, so that as long as one looks for self-dual equations
associated with (static) vortices, the A0 field (as well as the N field)
can be put to zero and so we will do from here on (the case N 6= 0 can
be equally treated without additional complications). So far, without
the addition of a Chern-Simons term, no electrically charged vortices
exist and then the most general gauge field configurations are pure
magnetic Nielsen-Olesen type soliton solutions. This implies that no
d = 3 version of the F̃µν F µν functional are available and that we can
simply identify the field strength with the magnetic field B by
1
Fµν F µν = B 2 (229)
2
with
1
B = εjk F jk i, j = 1, 2 (230)
2
Once the dimensional reduction is carried on, one ends with the d = 3
version of the SUSY Lagrangian given in eq.(228). As it is well known,
supersymmetry can be extended from N = 1 to N = 2 in this process.
From what we have seen, the gauge field dependent terms in the
bosonic part of this N = 2 supersymmetric Lagrangian can be com-
pactly written in the form
∞
1 X n
LA [Aµ , D] = cn t (231)
e2
n=0

where t (defined in (227)) now reads

D2 − B 2
t= (232)
β2

40
and cn are some coefficients which can be computed in terms of the
arst ’s.
Concerning the Higgs field sector, in d = 4 dimensions the coupling
between the scalar Higgs field φ and the gauge field Aµ arises from
the superfield interaction term
∗
Ld=4
A−φ = Φ exp(V )Φ (233)

where Φ is a chiral scalar superfield containing a Higgs field φ, a

higgsino ψ and an auxiliary field F . One can easily see that the part
of LA−φ containing the auxiliary field D is [43]

1
Ld=4
A−φ |D = D|φ|2 (234)
2
On the other hand, gauge symmetry breaking can be achieved à la
Fayet-Iliopoulos so that the complete D dependence of the supersym-
metric Lagrangian arising from the Higgs coupling to Aµ and D is
given by
1
Ld=4 d=4 d=4
D [A, φ, D] ≡ LA−φ |D + LF I = D(|φ|2 − ξ 2 ) (235)
2
where ξ is a real constant. This Lagrangian remains unchanged after
dimensional reduction so that we can write the D dependent terms of
the d = 3 bosonic part of the Lagrangian as
∞  
1 X 1 1
Ltotal
D [A, φ, D] = 2 cn (D − B ) n + D(|φ|2 − ξ 2 ) (236)
2 2
e β2 2
n=0

In d = 3 space-time, dimensions of parameters and fields are [β] = m2 ,

1 1 1
[e] = m 2 , [ξ] = m 2 , [Aµ ] = m, [D] = m2 and [φ] = m 2 . Then, for
dimensional reasons, one can infer that coefficients c′n s can be written
in the form
cn = β 2 λn (237)
where λn are dimensionless coefficients.
We can now obtain the equation of motion for D so as to eliminate
the auxiliary field from the physical spectrum
∞  
X 2n 1 1
λn (D − B ) n−1 D + (|φ|2 − ξ 2 ) = 0
2 2
(238)
e2 β 2 2
n=0

41
One can easily see that the only nontrivial solution to eq.(238) takes
the form
e2
D = − (|φ|2 − ξ 2 )
4λ1
B = ±D (239)

These two equations can be readily combined into one which is nothing
but the well-honored Bogomol’nyi equation for the magnetic field of
the Nielsen-Olesen vortices
e2
B=∓ (|φ|2 − ξ 2 ) (240)
4λ1
This shows that the Bogomol’nyi gauge field equation for vortex con-
figurations is independent of the particular form of the gauge field
Lagrangian one chooses since we have proven formula (240) for the
general supersymmetric Lagrangian (228)+(233). It should be noted
that the choice of λ1 = −1/2, λn = 0 for n 6= 1 corresponds to the
usual value of the Maxwell term while the choice λ1 = −1/2, λ2 = 1/8,
λ3 = 1/32, . . ., gives a Dirac-Born-Infeld Lagrangian for the gauge
field, which, once the auxiliary fields are put on shell takes the form
r 
2 1
LDBI = −β 1 + 2 Fµν F µν − 1 (241)
2β

Let us now analyze the N = 2 supersymmetry transformations

leaving invariant the three dimensional Born-Infeld SUSY theory. We
shall not write the complete set of transformations but just those
which are relevant for the discussion of Bogomol’nyi equations, namely
those for the higgsino and gaugino (which we call ψ and Σ):
  
0 D1 + iD2 ǫ+
δǫ ψ = −i D
6 φǫ = (242)
D1 − iD2 0 ǫ−
1 ij
  
1 2 εij F +D 0
ǫ+
δǫ Σ = ( εµνα F µν γ α +D)ǫ = 1 ij
2 0 2 εij F − D
ǫ−
(243)
where we call ǫ the Dirac fermion transformation parameter (we have
already made N = A0 = 0 and considered the static case).
As it is well-known by now, making zero half of the SUSY vari-
ations associated with the higgsino and gaugino fields, one gets the

42
 Bogomol’nyi equations. For instance, by demanding that those gener-
ated by ε+ be zero, one gets the following self-dual equation from the
higgsino’s variation

δǫ+ ψ = 0 → D1 φ = iD2 φ (244)

One should note that this transformation law just depends on the way
the parallel displacement is defined in terms of the gauge connection
and not on the explicit form of the gauge field action. One can then
understand why eq.(244) is completely independent of the particular
form the gauge field action is chosen, at least for minimally coupled
gauge theories 1 . Regarding the equation derived from the gaugino
transformation,
1
δǫ+ Σ = εij F ij + D = 0 (245)
2
it could, in principle, depend on the particular Lagrangian chosen
through the D term. However, as we have seen (eq.(239)), the solution
to the equation of motion for D takes the same simple form for any
gauge field Lagrangian since D always enters through the combination
D2 − B 2 .
This feature can be also checked by analyzing the two supercharges
which can be obtained following the usual Noether construction. As
it has been shown in [51] for the Born-Infeld case, supercharges Q and
Q̄ can be always put in the form
i
Z Z
Q̄ = i d2 x Σ† H[B, D] (γ 0 B + D) + d2 x ψ † D6 φ
2
i
Z Z
Q = −i d2 x (B + γ 0 D) H[B, D] Σ − d2 x γ 0 (6D φ)† ψ(246)
2
with H some real functional of D and B which can be computed order
by order in 1/β 2 . Furthermore, eqs.(246) also hold when one considers
not just SUSY Born-Infeld theory but the general Lagrangian, viz.
eq.(228). Only the actual form of H will change, depending on the
different sets of possible a′ s coefficients. What one can easily see is
that the following formula holds
∞
X 1
H = HM axwell + H [B, D]
2n n
(247)
n=1
β

1
For an analysis of Bogomol’nyi equations in non-minimally coupled gauge theories,
see ref.[47].

43
with
HM axwell = 1 (248)
Hn [B, D]|B 2 =D2 = 0 (249)
It is clear that condition Q̄|physi = 0 is satisfied whenever (B +
γ 0 D)ǫ = 0 and D 6 φǫ = 0, independently of the precise form the func-
tional H takes. Choosing just the upper component of the transfor-
mation parameter, ǫ+ , yields again the two Bogomol’nyi equations
(240),(244). Of course, this is to be expected since both (B + γ 0 D)ǫ
and D 6 φǫ, appearing in (246), provide the transformation laws of gaug-
ino and higgsino respectively.
Concerning the supercharge algebra, when the Bogomol’nyi equa-
tion B = ±D is used, only the Maxwell part of H survives, this show-
ing again why the BPS structure is not sensitive to the particular form
of the gauge field Lagrangian.

In conclusion, we have analyzed the most general Lagrangian cor-

responding to the supersymmetric extension of a general bosonic La-
grangian depending on the two algebraic Maxwell invariants F µν Fµν
and F̃ µν Fµν . This general Lagrangian includes, for a particular choice
of coefficients, the Born-Infeld supersymmetric Lagrangian, and also
an infinite class of Lagrangians having causal propagation [54]. We
have shown why the Bogomol’nyi relations associated with the bosonic
sector remain unchanged in spite of the actual form of the gauge
field Lagrangian: Maxwell, Born-Infeld or more complicated non-
polynomial Lagrangians all have the same BPS structure.

4 Including axions
Cosmic strings appearing as topological defects in Grand Unified The-
ory aroused a lot of interest about twenty years ago as a source of pri-
mordial density perturbations from which galaxies eventually grew.
In this scenario, since Abrikosov-Nielsen-Olesen strings arise in any
gauge theory with a broken U (1) symmetry, such strings could appear
whenever a U (1) symmetry becomes broken as the Universe cools.
However, after the microwave background results from COBE,
BOOMERanG and WMAP it was accepted, at the end of the 90’s,
that cosmic strings or other topological defects arising at GUT scales
could not provide an explanation for the bulk of the density pertur-
bations.

44
 The possibility of microscopic fundamental strings in superstring
theories acting as seeds for galaxy formation was also excluded in
the context of perturbative string theory [59]: among other problems,
their tension µ , close to Planck scale, would produce inhomogeneities
in the cosmic microwave background far larger than observed. For
example in perturbative heterotic string theory Gµ = αGU T /16π ≥
10−3 while the isotropy of the cosmic microwave background implied
Gµ = αGU T /16π ≤ 10−3 [60]
The question was be recently reconsidered, after the relevance of
branes and new kind of extended objects was understood (see [61]
and references therein) opening the possibility to the string tension
to be much lower, anything between the Planck scale and the weak
scale. In order to understand the nature and structure of such stringy
cosmic strings, the embedding of BPS objects in supersymmetry and
supergravity models has become an active area of research so that the
properties of BPS solitons, their connections with the supersymmetry
algebra and their cosmological applications have been discussed by
many authors [43],[62]-[73].
Having in mind the study of BPS solitons in a string theory con-
text, where the axion is almost unavoidable, it is natural to consider
gauge-Higgs systems in which the axion field is included. In particu-
lar, an N = 1 globally supersymmetric model in d = 3 + 1 dimensions
consisting of an axion superfield S coupled to Wα W α , with Wα the
chiral superfield strength, was analyzed in [69] and finite energy cos-
mic string solutions for the resulting bosonic Maxwell-Higgs action
coupled to an axion field were constructed. Also, the impact of axions
on dynamics of a d = 3 + 1 Yang-Mills theory supporting non-Abelian
strings has been analyzed in [73].
We shall consider here this issue taking as an example a d = 2 + 1
space-time dimensions for which a rich variety of flux tube solutions
exists already when the axion field is absent. Indeed, as we have
seen, when gauge fields with dynamics governed by a Chern-Simons
action are coupled to charged scalars with an appropriate sixth order
symmetry breaking potential, the model admits BPS equations with
vortex-like solutions carrying both magnetic flux and electric charge
[22]-[23]. It should be stressed that in the absence of the Chern Simons
term, electrically charged vortices with finite energy (per unit length)
do not exist [74]. Hence, the model we are interested in could show
novel aspects of charged string like configurations when an axion is
present, in particular with respect to their application to cosmological

45
problems. Of course, this in the perspective that at high tempera-
tures, a relativistic four dimensional quantum field theory becomes
effectively three dimensional.

The model
The coupling of an axion to a gauge field with dynamics governed
by a Chern-Simons action poses some problems [75]-[76]. To discuss
how can they can be overcome, let us consider the following (2 + 1)-
dimensional bosonic action,

Z nκ o
S= d3 x f (s)ImDµ S F̃ µ + |Dµ φ|2 + K ′′ (s)|Dµ S|2 − W (φ, S)
8δ
(250)
where φ and S = s + ia are complex fields, Aµ is a U (1) gauge field,
and F̃ µ is defined as,
F̃ µ ≡ ǫµνσ ∂ν Aσ (251)
W (φ, φ∗ , S, S ∗ ) is a potential term, f and K are arbitrary functions of
s, the real part of S, primes stand for ∂/∂s, κ is a constant and δ is
a dimensionless parameter (which in the 4 dimensional case is related
to the Fayet-Iliopoulos term and the Planck mass). Finally, Dµ is the
covariant derivative acting on the fields φ and S according to

Dµ φ = ∂µ φ − ieAµ φ
Dµ S = ∂µ S + 2iδAµ (252)

As done in [69] for the 3 + 1 model, we shall identify a in (250) with

the axion field and s with a dilaton field. Note that because of the
definition of the axion covariant derivative, the Chern-Simons term
appears in action (250) multiplied by the factor f (s),
κ
Z
SCS [A, s] = d3 xf (s)ǫµνσ Aµ ∂ν Aσ (253)
4
The action (250) is invariant under the local transformation

S −→ S − 2iδΛ(x)
∗
S −→ S ∗ + 2iδΛ(x)
φ −→ eieΛ(x) φ (254)
∗ ∗ −ieΛ(x)
φ −→ φ e
Aµ −→ Aµ + ∂µ Λ(x)

46
 The time component of the gauge field equation of motion is the
Chern-Simons version of the Gauss law and can be used to solve for
A0 giving
κB
A0 = (255)
2(e |φ| + 4δ2 K ′′ )
2 2

where
1
B ≡= f (s)Fxy − ǫij (Ai + ∂i a)∂j f (s) (256)
2δ
The energy can be found from the energy-momentum tensor obtained
by varying the curved space form of the action with respect to the
metric,
1 √
Z
δS = d3 x g T µν δgµν (257)
2
Integration of the time-time component T 00 gives
κ2 B 2
Z  
2 2 ′′ 2
E = d x |Di φ| + K |Di S| + W (φ, S) +
4(e2 |φ|2 + 4δ2 K ′′ )
(258)
After some work, this expression can be written in the form

κ2
Z
E = d2 x|Dx φ ± iDy φ|2 + K ′′ |Dx S ± iDy S|2 + ×
4(e |φ| + 4δ2 K ′′ )
2 2

  ′′
 2
e2 |φ|2 + 4δ2 K
e(|φ|2 − |φ0 |2 ) − 4δK ′ 


B ±
κ2 f (s)/2
1 
2 2 2 ′′

2
+W − 2
e |φ| + 4δ K e(|φ|2 − |φ0 |2 ) − 4δK ′
(κ f (s))
 
1
± e(|φ|2 − |φ0 |2 ) − 4δK ′ ǫij Ai + ∂i a ∂j logf (s)


2δ


± e|φ0 |2 Fxy  (259)

We thus see that the first two lines in (259) have been accommodated
as perfect squares. This, together with an appropriate choice of the
potential so as to cancel the third line, would lead to a Bogomol’nyi
bound for the energy given by the magnetic flux Φ appearing in the
last line, Z
Φ= d2 xFxy (260)

47
There is however the term in the forth line in (259) with no definite
sign preventing the obtention of a bound. Only if we put f (s) = 1,
which corresponds to a normal Chern-Simons action for the gauge
field (see eq.(253)), this term vanishes. In that case one does have a
bound,
E ≥ ±e|φ0 |2 Φ = 2πe|φ0 |2 |n| (261)
whenever the potential is chosen as
1 2 2 2 ′′

2 2 ′ 2


W = e |φ| + 4δ K e(|φ| − |φ0 | ) − 4δK (262)
κ2
The bound is saturated by fields obeying the self-duality equations

Dx φ = ∓iDy φ
Dx S = ∓iDy S
κ2 Fxy = ∓2 e2 |φ|2 + 4δ2 K ′′ e(|φ|2 − |φ0 |2 ) − 4δK ′ (263)

where the upper (lower) sign corresponds to positive (negative) values

of Φ and n.
As it is well-known, the presence of a Chern-Simons term forces a
relation between magnetic flux and electric charge [20]. This makes
the Chern-Simons vortices both magnetically and electrically charged.
To see this phenomenon in the present case let us write the gauge field
equation of motion, for the case f (s) = 1 in the form
κ µαβ
ε Fαβ = J µ (264)
2
with J µ = (ρ, J~) the conserved matter current and ρ the electric charge
density,

J µ = −2 e2 |φ|2 + 4K ′′ δ2 Aµ − ie(φ∂ µ φ∗ + φ∗ ∂ µ φ) − 4δ∂ µ a (265)

We then see that eq.(255) can be rewritten in the form

ρ = −κB (266)

d2 xρ and
R
so that the usual relation between electric charge Q =
magnetic flux in Chern-Simons theories holds,

Q = −κΦ (267)

Note that both the Higgs scalar and the axion contribute to the electric
charge.

48
Supersymmetric extension
The SUSY extension of the Chern-Simons-Higgs system with a sixth
order symmetry breaking potential was analyzed in [42]. Let us study
now the case in which the axion field is also present.
To do so, we will consider the following d = 3 action, written in
terms of superfields as
1 κ
Z  
SSUSY = − d3 x d2 θ F (Σ + Σ† )Im ∇˜ aΣ W a
2 4δ
+ H(Φ , Φ)(∇ Φ)† ∇a Φ + Kss̄ (Σ, Σ† )|∇
† a ˜ a Σ|2

+ 2V (Φ† , Φ, Σ, Σ† ) (268)

Here µ = 0, 1, 2 and a = 1, 2. Γa , Φ and Σ are spinor, complex scalar

and axionic superfields and

∇a Φ = (Da − ieΓa )Φ
˜ a Σ = Da Σ + 2iδΓa
∇
δ
Da = + i(γ µ )ab θ b ∂µ
δθ a
The lowest component in Γa is the gauge field, that in Φ corresponds
to the Higgs field and that in Σ is S = s + ia with a the axion field.
Concerning F, H, K and V , they are functionals of superfields to be
fixed later. Subindexes in these functionals mean derivatives, thus
Kss̄ = ∂S ∂S ∗ K = ∂Σ ∂Σ̄ K|θ=0 and so on. They should be chosen so
that the supersymmetric action (268) is invariant under the super-
gauge transformations,

Φ −→ eieΛ Φ
Γa −→ Γa − Da Λ
Σ −→ Σa − 2δΛ (269)

for any real scalar superfield Λ. Written in the Wess-Zumino gauge,

i.e.,
Γa |θ=0 = D a Γa |θ=0 = 0,
the spinor superfield Γa is given by

Γa (x, θ) = iθ b (γ µ )ba Aµ (x) − 2θ 2 λa (x)

49
where λ(x) is a Majorana spinor, the photino. Then, the spinor field
strength, defined as
1
Wa = D b Da Γb
2
takes, in terms of component fields, the form
1
Wa (x, θ) = λa (x) − θ b (γ µ γ ν )ba Fµν − iθ 2 (γ µ )ba ∂µ λb (x)
2
and satisfies the Bianchi identity,

Da Wa = 0.

The complex scalar superfield is defined as

Φ(x, θ) = φ(x) + θ a ψa (x) − θ 2 F (x)

where φ stands for the Higgs complex scalar field, ψa is a Dirac

bispinor, the higgsino and F is a complex auxiliary field. Finally,
the superfield Σ which contains the axion S as its lowest component,
can be reduced to a complex scalar field Θ by exponentiation,

Θ = eΣ

The supersymmetric transformations (269) take Γa out from the Wess-

Zumino gauge. One can however implement a composition of SUSY
and gauge transformations such that the Wess-Zumino gauge remains
valid. To do that, the new SUSY transformation for scalar and spino-
rial superfields are, respectively,

δηWZ Γa = iη b Qb Γa + Da K̃
δηWZ Φ = iη b Qb Φ + iK̃Φ (270)
δηWZ Σ = iη b Qb Σ + iK̃

where Qa = i∂a + θ b (γ µ )ba ∂µ , η a a Majorana spinor and the real scalar

superfield K̃ is defined as,

K̃ = iθ a (γ µ )ab η b Aµ + θ 2 λa ηa .

Let us restrict the action (268) to the case in which

F (Σ + Σ† ) = H(Φ† , Φ) = 1 (271)

50
Then, written in components, the action takes the simple form

SSU SY = SB + SF (272)

with the pure bosonic action given by

 
κ µνσ 1
Z
3 2 2 2 2
SB = d x ǫ Aµ Fνσ + |Dµ φ| + R|D̃µ S| − |Vφ | − |Vs |
4 R
(273)
while the fermionic one is given by

1
Z
↔ ↔
SF = d3 x iψ̄γ µ Dµ ψ + iRχ̄γ µ ∂ µ χ − Rss̄ (χ̄χ)2
2
1 h i
+ iχ̄γ µ χ(Rs D̃µ S − Rs̄ D̃µ S ∗ )
2 
e2 2
  2 
4δ 2
+ ψ̄ψ |φ| − Vφφ̄ + χ̄χ R − Vss̄
κ κ
|Rs |2
   
2δe ∗ ∗ 2δe ∗
− ψ̄χ φR + Vφ̄s − ψ̄ χ φ R + Vφs̄ − |χ̄χ∗ |2
κ κ 4R
e2 ∗2 e2 2
   
1 ∗ 1 ∗
+ ψ̄ ψ −Vφφ − φ + ψ̄ψ −Vφ̄φ̄ − φ
2 κ 2 κ
2 4δ2 2
   
1 ∗ Rs 4δ 2 1 ∗ Rs̄
+ χ̄ χ Vs − Vss − R + χ̄χ Vs̄ − Vs̄s̄ − R
2 R κ 2 R κ
   
2δe ∗ 2δe
+ ψ̄ ∗ χ φ R − Vφs + ψ̄χ∗ φR − Vφ̄s̄ , (274)
κ κ

where we have redefined R ≡ Kss̄ . Here we have taken into account

that Z
I = d3 xd2 θ (D a Σ) Wa

is a surface term which does not modify the equations of motion

Z Z
I = d3 xd2 θ D a (ΣWa ) = d3 x D2 D a (ΣWa )|θ=0
Z
= = i d3 x ∂ba Db (ΣWa )|θ=0 ,

and can hence be neglected.

In order to extend the N = 1 supersymmetry to an N = 2 super-
symmetry, we shall follow [42] and allow the transformation parameter
η to be complex, i.e., η ∗ 6= η. Now, this is the same as making an

51
N = 1 SUSY transformation followed by a U (1) fermion phase rota-
tion. Thus, the new transformation for fermions will be ψa → eiα ψa
and ψa∗ → e−iα ψa∗ and the same for χa and χ∗a . The new SUSY trans-
formations act then as rotations on the fermions and one can then see
that the only terms which do not respect the extended SUSY invari-
ance are those on the last three lines in (274). Hence, in order to get
an N = 2 supersymmetric model we need that,

e2 ∗2 2δe ∗
− Vφφ − φ =0 , φ R − Vφs = 0
κ κ
4δ2 2 4δ2 2 2δe
R − Vss̄ − R =0 , φR − Vφ̄s̄ = 0
κ κ κ
where V = V (u, v) ≡ V (φ∗ φ, S + S ∗ ). These equations imply that,
e
Vu = − (eh(u) − 2δr(v)) H(u)
κ
2δ
Vv = − (2δr(v) − eh(u)) R(v)
κ
where
d d
r(v) = R(v) , h(u) = 1. (275)
dv du
We obviously have h = u − u0 = |φ|2 − |φ0 |2 , and since S + S ∗ = 2s,
then R = K ′′ and r = 2K ′ where primes stand for derivatives with
respect to s. From (273) the potential is,
1 1 2
W = |Vφ |2 + |Vs |2 = 2 (e2 |φ|2 + 4δ2 K ′′ ) e(|φ|2 − |φ0 |2 ) − 4δK ′

R κ
which is exactly what we obtained in (262).
In order to get the Bogomol’nyi bound and the self-dual equations
one can analyze the supercharge algebra as in [62]. Alternatively,
one can directly consider the component field SUSY transformations
(δη X = η a δa X),

δa Aµ = −i(γµ )ba λb , = 21 (γ µ γ ν )ab Fµν

δa λb
δa φ = −ψa , δa S
= −χa ,
δa ψb = ǫab F − i(γ µ )ab Dµ φ, = ǫab J − i(γ µ )ab D̃µ S,
δa χb
δa F = i(γ µ )ba Dµ ψb + 2λa φ, = i(γ µ )ba ∂µ χb + 2λa ,
δa J
(276)
and their complex conjugated (δX † = η ∗a δa X † ) and reobtain Bogo-
mol’nyi equations (263) just by putting all fermion fields to zero in

52
(276) and then ask the SUSY transformations for ψa and χa to vanish
once the auxiliary fields have been written in terms of dynamical fields
using their equations of motion. The first condition corresponds to a
restriction to the bosonic sector, the second one implies that physical
states are supersymmetry invariant.

String-like solutions
We present in this section some vortex solutions to the BPS equations
of motion. We choose for the Khäler potential the form
K = −M 2 log (S + S ∗ ) (277)
As in [69], we shall analyze separately two cases: first, we con-
sider the case in which the vortex is supported by the Higgs field
(“φ-strings” solutions) , in the sense that at infinity it behaves as in
the ordinary Nielsen-Olesen vortex, with its winding number linked
to the magnetic flux. Then, we shall consider the case in which the
vortex is supported by the axion field, a solution that we shall call an
“s-string”. In this case it is the axion winding number which is related
to the magnetic flux.
In the first case, in order to obtain φ-string solutions we make the
ansatz [69]
v(r)
φ1 = f (r)einθ S = s(r) − 2iδmθ Aθ = n (278)
r
where n is the topological charge of the Higgs and m is the topological
charge of the axion.
It is convenient to work with dimensionless variables by defining
e2 φ20
τ = αr, α=
κ
x(τ ) = v(τ /α)
y(τ ) = eδ−1 s(τ /α)
z(τ ) = φ−1
0 f (τ /α) (279)
With this convention the equations read,
  
′ 2τ 2 4β 2 4β
x = − z + 2 z −1+ (280)
|n| y y
2
y ′ = − (|m| − |n| x) (281)
τ
′ z
z = (1 − x) |n| (282)
τ

53
where β = M 2 /φ20 . From the first two equations we can integrate y(τ )
in terms of z(τ ), obtaining

y(τ ) = 2(|n| − |m|) log τ − 2 log z(τ ) + k (283)

with k an arbitrary integration constant. Thus, we end with a system

of two first-order coupled differential equations for x′ (τ ) and z ′ (τ ).
The boundary conditions for equations (280) and (282) can be
determined as follows. The function x must vanish at the origin, so
eq.(282)implies that z also vanishes at the origin as τ |n| . For large
τ , the function x tends to 1, thus z also tends to 1 unless |n| = |m|.
In this last case, z → z0 , where z0 is the solution of the algebraic
equation
4β
z02 + =1 (284)
k − 2 log z0
To solve the differential equations we employ a relaxation method
for boundary values problem. Such a method determines the solution
by starting with an initial guess and improving it iteratively.
Already from ansatz (278) one can see that the magnetic flux and
electric charge ar quantized according to
2π 2πκ
Φ= n, Q=− n (285)
e e
There is an interesting property, typical of Chern Simons vortices that
also holds in our model: both the magnetic field and the (radial)
electric field are concentrated in rings surrounding the zeroes of the
Higgs field.
We also solved the BPS equations when the axion field s tends
asymptotically to a constants (“s-string”, [69]). In this case the mag-
netic charge is equal to the topological charge of the axion field, so we
have as an appropriate ansatz,
v(r)
φ = f (r)einθ S = s(r) − 2iδmθ Aθ = m (286)
r
With this ansatz the BPS equations take the form
  
′ 2τ 2 4β 2 4β
x = − z + 2 z −1+ (287)
|m| y y
2
y ′ = − |m| (1 − x) (288)
τ
′ z
z = (|n| − |m| x) (289)
τ

54
Again, we can integrate y(τ ) in terms of z(τ ), obtaining

y(τ ) = 2(|n| − |m|) log τ − 2 log z(τ ) + k (290)

We see from this equation that for consistency, z(τ ) ∼ τ n−m for τ →
∞ in contrast with what happens for the φ-string. Concerning the
gauge field boundary condition, one has limτ →∞ x(τ ) = 1.
As a summary of the numerical analysis of the BPS solutions, one
should note that axionless string solutions found in ([22])-([23]) are
not much modified by the axion which, however, contributes to the
electric charge of the string configuration. These explicit solutions
could be of interest in the context of cosmic strings and, due to the
coupling to the axion and their electric charge, their dynamics could
be very different of that of ANO vortices.

Acknowledgment: I would like to thank the organizers, lecturers and

participants, and in particular Prof. Pedro Labraña, for the splendid
days I spent during the Valparaiso school. This work was partially sup-
ported by PIP-CONICT, PICT-ANPCYT, UNLP and CICBA grants.

References
[1] W.B.Yeats, in Michael Robartes and the Dancer, Wordsworth
Editions, Hertfordshire, 1994.
[2] P. Goddard and D. I. Olive, “New Developments In The Theory
Of Magnetic Monopoles,” Rept. Prog. Phys. 41 (1978) 1357.
[3] A. M. Jaffe and C. Taubes, “Vortices And Monopoles. Structure
Of Static Gauge Theories,” Birkhiuser, Boston, 1980.
[4] R. Rajaraman, “Solitons And Instantons. An Introduction To
Solitons And Instantons In Quantum Field Theory,” North-
Holland, Amsterdam, 1982
[5] J. Preskill, “Vortices and Monopoles,” Lectures presented at the
1985 Les Houches Summer School, Les Houches, France
[6] A. Achucarro and T. Vachaspati, “Semilocal and electroweak
strings,” Phys. Rept. 327 (2000) 347 [Phys. Rept. 327 (2000)
427] [arXiv:hep-ph/9904229].
[7] D. Tong, “TASI lectures on solitons,” arXiv:hep-th/0509216.

55
 [8] A. Abrikosov, “On the magnetic properties of superconductors of
the second group” Sov. Phys. JETP 32 1442 (1957) [Reprinted
in Solitons and Particles, Eds. C. Rebbi and G. Soliani (World
Scientific, Singapore, 1984), p. 356].
[9] H. Nielsen and P. Olesen, “Vortex-line models for dual strings,”
Nucl. Phys. B61 45 (1973) [Reprinted in Solitons and Particles,
Eds. C. Rebbi and G. Soliani (World Scientific, Singapore, 1984),
p. 365].
[10] J. Polchinski, “Cosmic Superstrings Revisited,” AIP Conf. Proc.
743, 331 (2005) [Int. J. Mod. Phys. A 20, 3413 (2005)]
[arXiv:hep-th/0410082].
[11] Y. Nambu, Lectures prepared for the Copenhagen Summer Sym-
posium (1970), unpublished; T. Goto, Prog. Theor. Phys. 46,
1560 (1971).
[12] H. de Vega and F .A. Schaposnik, “Classical vortex solution of
the Abelian Higgs model,” Phys. Rev. D14(1976)1100.
[13] P.Fayet, Il Nuovo Cimento, “Higgs Model And Supersymmetry,”
A31 (1976)626.
[14] A.Salam and J.Strathdee, “Supersymmetry, Parity And Fermion
- Number Conservation,” Nucl.Phys.B97(1975)293.
[15] E.B.Bogomol’nyi, “The stability of classical solutions,” Sov. Jour.
Nucl. Phys.24(1976)449.
[16] M. K. Prasad and C. M. Sommerfield, “An Exact Classical Solu-
tion For The ’T Hooft Monopole And The Julia-Zee Phys. Rev.
Lett. 35, 760 (1975).
[17] S. R. Coleman, S. J. Parke, A. Neveu and C. M. Sommerfield,
“Can One Dent A Dyon?,” Phys. Rev. D 15, 544 (1977).
[18] R. Jackiw and S. Templeton, “How Superrenormalizable Interac-
tions Cure Their Infrared Divergences,” Phys. Rev. D 23, 2291
(1981).
[19] J. F. Schonfeld, “A Mass Term For Three-Dimensional Gauge
Fields,” Nucl. Phys. B 185 (1981) 157.
[20] S. Deser, R. Jackiw and S. Templeton, “Three-Dimensional Mas-
sive Gauge Theories, ” Phys. Rev. Lett. 48 (1982) 975; “Topo-
logically Massive Gauge Theories,” Annals Phys. 140 (1982) 372
[Erratum-ibid. 185 (1988 APNYA,281,409-449.2000) 406.1988
APNYA,281,409].

56
[21] S. K. Paul and A. Khare, “Charged Vortices In Abelian Higgs
Model With Chern-Simons Term,” Phys. Lett. B 174, 420 (1986)
[Erratum-ibid. 177B, 453 (1986)].
[22] J. Hong, Y. Kim and P. Y. Pac, “On The Multivortex Solutions
Of The Abelian Chern-Simons-Higgs Theory,” Phys. Rev. Lett.
64 (1990) 2230.
[23] R. Jackiw and E. J. Weinberg, “Selfdual Chern-Simons Vortices,”
Phys. Rev. Lett. 64 (1990) 2234.
[24] D. Birmingham, M. Blau, M. Rakowski and G. Thompson, Phys.
Rept. 209 (1991) 129.
[25] T. Vachaspati and A. Achucarro, “Semilocal cosmic strings,”
Phys. Rev. D 44 (1991) 3067.
[26] M. Hindmarsh, “Existence and stability of semilocal strings,”
Phys. Rev. Lett. 68 (1992) 1263.
[27] S. Mandelstam, “Vortices and Quark Confinement in Nonabelian
Gauge Theories,” Phys. Lett. B 53, 476 (1975).
[28] H. J. de Vega, “Fermions And Vortex Solutions In Abelian And
Nonabelian Gauge Theories,” Phys. Rev. D 18 (1978) 2932.
[29] P. Hasenfratz, “A Puzzling Combination: Disorder X Order,”
Phys. Lett. B 85 (1979) 338.
[30] A. S. Schwarz and Y. S. Tyupkin, “Vortices In Unified Theories
Of Weak And Electromagnetic Interactions,” Phys. Lett. B 90
(1980) 135.
[31] H. J. de Vega and F. A. Schaposnik, “Electrically Charged Vor-
tices in Nonabelian gauge theories with Cher-Simons Term,”
Phys. Rev. Lett. 56, 2564 (1986).
[32] H. J. de Vega and F. A. Schaposnik, “Vortices and electrically
charged vortices in nonAbelian gauge theories,” Phys. Rev. D
34, 3206 (1986).
[33] L. F. Cugliandolo, G. Lozano and F. A. Schaposnik, “Bogomolny
equations for non-Abelian gauge theories,” Phys. Rev. D 40, 3440
(1989).
[34] L. F. Cugliandolo, G. Lozano, M. V. Manias and F. A. Schapos-
nik, “Bogomolny equations for non-Abelian Chern-Simons Higgs
theories,” Mod. Phys. Lett. A 6 (1991) 479.

57
[35] R. Auzzi, S. Bolognesi, J. Evslin, K. Konishi and A. Yung, “Non-
abelian superconductors: Vortices and confinement in N = 2
SQCD,” Nucl. Phys. B 673, 187 (2003) [arXiv:hep-th/0307287].
[36] M. Shifman and A. Yung, “Non-Abelian string junctions
as confined monopoles,” Phys. Rev. D 70, 045004 (2004)
[arXiv:hep-th/0403149].
[37] V. Markov, A. Marshakov and A. Yung, “Non-Abelian vor-
tices in N = 1* gauge theory,” Nucl. Phys. B 709, 267 (2005)
[arXiv:hep-th/0408235].
[38] M. Shifman and A. Yung, “Non-Abelian semilocal strings in N
= 2 supersymmetric QCD,” Phys. Rev. D 73 (2006) 125012
[arXiv:hep-th/0603134].
[39] K. Bardakci and M. B. Halpern, “Spontaneous breakdown and
hadronic symmetries,” Phys. Rev. D 6, 696 (1972).
[40] K. Rajagopal and F. Wilczek, “The condensed matter physics of
QCD,” arXiv:hep-ph/0011333.
[41] E. Witten and D. I. Olive, “Supersymmetry Algebras That In-
clude Topological Charges,” Phys. Lett. B 78, 97 (1978).
[42] C. k. Lee, K. M. Lee and E. J. Weinberg, “Supersymmetry And
Selfdual Chern-Simons Systems,” Phys. Lett. B 243 (1990) 105.
[43] J. D. Edelstein, C. Núñez and F. Schaposnik, “Supersymmetry
and Bogomolny equations in the Abelian Higgs model,” Phys.
Lett. B 329 (1994) 39 [arXiv:hep-th/9311055].
[44] M. F. Sohnius, “Introducing Supersymmetry,” Phys. Rept. 128,
39 (1985).
[45] P. Di Vecchia and S. Ferrara, “Classical Solutions In Two-
Dimensional Supersymmetric Field Theories,” Nucl. Phys. B
130, 93 (1977).
[46] H. R. Christiansen, C. Núñez and F. A. Schaposnik, “Uniqueness
of Bogomolnyi equations and Born-Infeld like supersymmetric
theories,” Phys. Lett. B 441, 185 (1998) [arXiv:hep-th/9807197].
[47] H. R. Christiansen, M. S. Cunha, J. A. Helayel-Neto,
L. R. U. Manssur and A. L. M. Nogueira, “Self-dual vortices in
a Maxwell-Chern-Simons model with non-minimal Int. J. Mod.
Phys. A 14 (1999) 1721 [arXiv:hep-th/9805128].

58
[48] A. A. Tseytlin, “Born-Infeld action, supersymmetry and string
theory,” in the Yuri Golfand memorial volume, ed. M. Shifman,
World Scientific, 2000. In *Shifman, M.A. (ed.): The many faces
of the superworld* 417-452. arXiv:hep-th/9908105.
[49] G. W. Gibbons, “Born-Infeld particles and Dirichlet p-branes,”
Nucl. Phys. B 514, 603 (1998) [arXiv:hep-th/9709027].
[50] J. P. Gauntlett, J. Gomis and P. K. Townsend, “BPS
bounds for worldvolume branes,” JHEP 9801, 003 (1998)
[arXiv:hep-th/9711205].
[51] S. Gonorazky, C. Núñez, F. A. Schaposnik and G. A. Silva, “Bo-
gomol’nyi bounds and the supersymmetric Born-Infeld theory,”
Nucl. Phys. B 531, 168 (1998) [arXiv:hep-th/9805054].
[52] K. Shiraishi and S. Hirenzaki, Bogomol’nyi Equations for Vortices
in Born-Infeld Higgs Systems, International Journal of Modern
Physics A6 (1991) 2635.
[53] A. Nakamura and K. Shiraishi, “Born-Infeld monopoles and in-
stantons,” Hadronic J. 14, 369 (1991).
[54] S. Deser and R. Puzalowski, “Supersymmetric Nonpolynomial
Vector Multiplets And Causal Propagation,”J. Phys. A13 (1980)
2501.
[55] S. Ceccotti and S. Ferrara, “Supersymmetric Born-Infeld Lagran-
gians,”Phys. Lett. B187 (1987) 335.
[56] T. Hagiwara, “An Effective Lagrangian For Multi - Photon Pro-
cesses And A Nonlinear Born-Infeld Lagrangian,” Nucl. Phys. B
189 (1981) 135
[57] M. Born, “On the Quantum Theory of the Electromagnetic
Field,” Proc. R. Soc. London A143 (1934) 410.
[58] M. Born and M. Infeld, “Foundations of the New Field Theory,”
Proc. R. Soc. London A144 (1934) 425.
[59] E. Witten, “Cosmic Superstrings,” Phys. Lett. B 153 (1985) 243.
[60] M. B. Hindmarsh and T. W. B. Kibble, “Cosmic strings,” Rept.
Prog. Phys. 58, 477 (1995) [arXiv:hep-ph/9411342].
[61] J. Polchinski, “Introduction to cosmic F- and D-strings,”
arXiv:hep-th/0412244.
[62] J. D. Edelstein, C. Núñez and F. A. Schaposnik, “Supergravity
and a Bogomolny bound in three-dimensions,” Nucl. Phys. B 458,
165 (1996) [arXiv:hep-th/9506147].

59
[63] J. D. Edelstein, C. Núñez and F. A. Schaposnik, “Bogomol’nyi
Bounds and Killing Spinors in d=3 Supergravity,” Phys. Lett. B
375, 163 (1996) [arXiv:hep-th/9512117].
[64] G. Dvali, R. Kallosh and A. Van Proeyen, “D-term strings,”
JHEP 0401 (2004) 035.
[65] P. Binetruy, G. Dvali, R. Kallosh and A. Van Proeyen, “Fayet-
Iliopoulos terms in supergravity and cosmology,” Class. Quant.
Grav. 21 (2004) 3137.
[66] A. Achucarro and J. Urrestilla, “F-term strings in the Bogomolnyi
limit are also BPS states,” JHEP 0408 (2004) 050.
[67] R. Jeannerot and M. Postma, “Chiral cosmic strings in super-
gravity,” JHEP 0412 (2004) 043.
[68] S. S. Gubser, C. P. Herzog and I. R. Klebanov, “Variations on the
warped deformed conifold,” Comptes Rendus Physique 5 (2004)
1031.
[69] J. J. Blanco-Pillado, G. Dvali and M. Redi, “Cosmic D-strings as
axionic D-term strings,” Phys. Rev. D 72 (2005) 105002.
[70] S. L. Parameswaran, G. Tasinato and I. Zavala, “The 6D Super-
Swirl,” arXiv:hep-th/0509061.
[71] A. Achucarro, A. Celi, M. Esole, J. Van den Bergh and A. Van
Proeyen, “D-term cosmic strings from N = 2 supergravity,”
arXiv:hep-th/0511001.
[72] R. Auzzi, M. Shifman and A. Yung, “Composite non-Abelian flux
tubes in N = 2 SQCD,” arXiv:hep-th/0511150.
[73] A. Gorsky, M. Shifman and A. Yung, “Nonabelian strings and
axion,”, Phys. Rev. D 73, 105012 (2006), arXiv:hep-th/0601131.
[74] B. Julia and A. Zee, “Poles With Both Magnetic And Electric
Charges In Nonabelian Gauge Theory,” Phys. Rev. D 11, 2227
(1975).
[75] M. Burgess, “Chern-Simons vortices in an open system,” Phys.
Rev. D 52 (1995) 1165
[76] W. Garcia Fuertes and J. Mateos Guilarte, “Self-dual solitons in
N = 2 supersymmetric Chern-Simons gauge theory,” J. Math.
Phys. 38 (1997) 6214.

60