Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

1

1. LECTURE I: SOLITONS

Consider a real scalar field theory in 1 space and 1 time dimensions with the action
integral given by,

!
1
I[ϕ] = d2 x{ ∂µ φ(x)∂ µ φ(x) − V (φ)} (1.1)
2
where

1
V (φ) = λ(φ2 − v 2 )2
8
We have added a constant 18 λv 4 to the potential in order to make its minima to be
at V = 0. V = 0 takes its minima at the points φ = ±v.
The potential is symmetric under φ → −φ and φ = 0 is a local maximum. The
Eq. of motion are,

1
∂ 2 φ(x) + λφ(x)(φ(x)2 − v 2 ) = 0 (1.2)
2
The energy functional is given by,

!
1 1
E[φ] = dx[{ (∂t φ(x))2 + (∂x φ(x))2 + V (φ)}] (1.3)
2 2
The question of interest for us is to find finite energy solutions of field equations.
One obvious solution is φ(x) = 0. The energy per volume of this solution is 18 λv 2 .
The other solutions are φ = ±v which have zero energy. Are there any other finite
energy solutions? We shall mainly be interested in time independent solutions. For
such solutions equations of motion become,

1
−∂x2 φ(x) + λφ(x)(φ(x)2 − v 2 ) = 0
2
2
Multiply by φ′ = ∂x φ to obtain,

1 d ′ d
(φ (x))2 = V
2 dx dx
The requirement of finite energy implies the boundary conditions that both sides
vanish for large values of x2 . We then obtain

lim φ(x) → ±v (1.4)


x→±∞

(φ′ (x))2 = 2V (φ(x)) (1.5)

The energy functional can be factorised,

! " "
1
E[φ] = dx[{ (φ′ − 2
2V (x)) + φ ′
2V (φ)}] (1.6)
2
The first term vanishes by virtue of Eq. (1.5) and we thus obtain,

! ∞ "

E[φ] = dxφ 2V (φ)
!−∞
v "
= dφ 2V (φ)
!−vv
1√
= dφ λ(φ2 − v 2 )
−v 2

the integral can be performed easily to give,

2√ 3
E[φ] = λv (1.7)
3
3
Note that in 2-space time dimensions the field φ is dimensionless. hence λ has the
dimension of mass2 . In fact we express the solution in terms of the mass parameter
m2 by writing φ(x) = σ(x) − v.

1
V = λ[(σ − v)2 − v 2 ]2
8
1
= λ(σ 2 − 2vσ)2
8
1
= λ(4v 2 σ 2 − 2vσ 3 + σ 4 ) (1.8)
8
Then the mass of the σ particle is

m2 = λv 2 (1.9)

Then Eq. (1.7) can be written as,

2 m2
E[φ] = ( )m (1.10)
3 λ
λ
For week coupling ( i.e. λ << m2 , or g ≡ m2 ≪ 1 ) this minimum energy is much
larger that m.
Note that the first term in Eq. (1.6) must vanish in order to minimise the energy
( which is also dictated by field equations.) This gives us,

"

φ = 2V (x)

Thus,


# = dx
2V (x)
This can easily be integrated to give,
4

1
φ(x) = vtanh[ m(x − x0 )]
2
#
where x0 is an integration constant. Since φ′ = 2V (x) the energy density be-
comes &(x) = φ′2 (x). So it is given by,

v 2 m2 1
&(x) = (1.11)
4 (cosh 12 m(x − x0 ))2

1
As |x − x0 | >> m this density behaves as

v 2 m2 −m(x−x0 )
&(x) = e (1.12)
4

We see that the energy density is localised in a small region around x0 and it
1
goes to zero exponentially fast outside a region of radius |x − x0 > m. We have a
localised lump of energy which can be identified with a particle of mass M given by
the energy of the solution,

2 m2
M = E[φ] = ( )m (1.13)
3 λ
λ 2m
If we define the dimensionless coupling g ≡ m2 the mass then will be M = 3 g.

This shows that for week coupling i.e. for g ≪ 1 the mass of the soliton is much

bigger than the mass of the quantum particles 2m. also the Compton wavelength
1 1
M of the soliton will be much smaller than its classical extension m.

Note that the momentum of the solution is found by integrating the T01 = φ′ ∂0 φ
component of the energy momentum tensor with respect to space. Obviously this
gives us zero. Thus we have a particle at rest. To obtain a moving solution we can
apply a Lorentz boost which gives us a time dependent solution,
5

1
φ(t, x) = vtanh[ γm(x − x0 )]
2
1
where γ = √ is the Lorentz boost factor. The energy of this solution will
(1−β 2 )
be E = γM and its momentum will be p = γβM . It is easy to see that the usual
#
energy momentum realtionship for a massive particle is satisfied, E = p2 + M 2 .
Let us summarise the salient features of this solution which is called a soliton:
1. The soliton is a localised lump of energy which maintains its shape as it moves.
Its energy does not dissipate
2. Its existence and stability against the decay to vacuum is intimately related to
the fact that the ground state has a ”non-trivial topology”. In this case it consists
of two points +v and −v. The soliton configuration goes to +v at x = +∞ and
to −v at x = −∞. Compared with the vacuum solution the entire space including
the boundaries at x = ±∞ is mapped to a single point in the field space, namely,
to the point φ = v or φ = −v. The two solutions can not be continuously deformed
into each other by time evolution. This cal also be confirmed by the existence of a
topologically conserved current, a current whose conservation law has nothing to
do with the Noether theorem and is conserved irrespective of the field equations.
In fact any 2 dimensional scalar field theory has such a conserved current which is
given by,

1
jµ (x) ≡ εµν ∂ ν φ(x), ∂ µ jµ (x) = 0 (1.14)
2v
The conserved charge of this current is

!
1 ∞
n[φ] = dxφ′ (x)
2v −∞
1
= [φ(+∞) − φ(−∞)] (1.15)
2v
For the vacuum solution φ(+∞) = φ(−∞) and thus n[φ] = 0. For the soliton, on
6
the other hand, φ(+∞) = v = −φ(−∞). Thus n[φ] = 1. Note that we could also
have an anti-soliton for which φ(+∞) = −v = −φ(−∞). In this case we obtain
n[φ] = −1. This solution is obviously given by,

1
φ(x) = −vtanh( m(x − x0 )
2

2. SOLITONS IN HIGHER DIMENSIONS

2.A. Domain Walls

Consider the same scalar field theory in 1 + 3 dimensions. The boundary of space
is now S 2 but the vacuum manifold is still the set of two points ±v. There is no
non trivial continuous map from the S 2 to the set ±v. So we can not expect to find
a higher dimensional generalisation of the above solution which would depend on
all three space coordinates x, y, z. However, we can still find a solution with a finite
energy density, namely,

1
φ(t, x, y, z) = ±vtanh[ m(x − x0 )]
2
The energy density will be given by the same formulae as before,

v 2 m2 1
&(t, x, y, z) = (2.1)
4 (cosh 12 m(x − x0 ))2
1
As |x − x0 >> m this density behaves as

v 2 m2 −m(x−x0 )
&(t, x, y, z) = e (2.2)
4
Note that in 4- dimensional space time v has the dimension of a mass. We notice
1
that there is an energy concentration in an interval of the order of m about x0 and
independent of the y, z coordinate. So wherever we move in the z, y plane we shall
7
find the same energy density. Thus we have a wall of energy extended in the y, z
plane sitting at the point x0 . This is called a domain wall. The energy density along
the wall is constant.
Terminology :
The word Soliton will refer to any localised solution of classical field equations
which do not dissipate in time.

2.B. sine- Gordon Equation

This is 1 + 1 dimensional field theory with the potential,

V (φ) = α[1 − cos(aφ)] (2.3)

The minima of the potential are located at

aφ = 2νπ, ν = 0, ±1, ±2, .... (2.4)

A finite energy solution must approach one of these minima at x = ±∞. Assume
aφ(+∞) = 2πν+ and aφ(−∞) = 2πν− . Define the topological current by,

a
jµ (x) ≡ εµν ∂ ν φ(x), ∂ µ jµ (x) = 0 (2.5)

Then the topological charge will be given by,

! ∞
a
n[φ] = dxφ′ (x)
2π −∞
a
= [φ(+∞) − φ(−∞)]

= ν+ − ν− (2.6)

The energy of the static solutions will again be given by


8

! ∞ "

E[φ] = dxφ 2V (φ)
!−∞
v "
= dφ 2V (φ)
−v
! 2πν+
"
a
= 2πν−
dφ 2b[1 − cosaφ]
a

Define aφ = 2θ and use cos2θ = (cosθ)2 − (sinθ)2 . Then

! 2πν+
"
a
E[φ] = 2πν−
dφ 2b[1 − cosaφ]
a
! 2πν+
"
a
= dφ 4bsin2 θ]
2πν−

√ 2 ! πν+
a

=2 b dθsinθ
a πν−
√ 4
= b [cos(πν+ ) − cos(πν− )]
a
√ 4
= b [(−1)ν+ − (−1)ν− ]
a
This is non zero only if n = ν+ − ν− is odd. Otherwise it vanishes. However, as
long as the topological charge is non zero the solution is in a different topological
class than the vacuum which corresponds to φ = 0 and n = 0.
It is not hard to see that a static solution can extend only between two neighbour-
ing minima of V (φ). To see this we recall again that, independent of the details of
the potential, the static field equations can be written as,


±# = dx
2V (φ(x))
or,
9

! φ(x) ! x

± # = dx′ = x
φ(0) 2V (φ) 0

This shows that as x → ±∞, φ must approach a global minimum of V (φ) so that
we obtain ∞ on both sides of the equation. Thus a static solution can interpolate
only between two successive minima of V (φ) at φ(−∞) and φ(∞).
10
2.C. Solitons in more than 1 Space Dimensions

The finite (total) energy solutions we discussed so far are localised in 1 space
dimension. Are there any finite energy solutions in a pure scalar field theory model
which are localised in more than 1 space dimensions?
Assume the space has d and the space-time has 1 + d- dimensions. It is rather
easy to show that there can not be any finite energy static solutions with only scalar
fields present for any space dimensions d > 1. Let us consider any scalar field model
in d space dimensions and assume that φa (x) is a finite minimum energy solution
of this model with any potential V (φ). The energy functional will be

E[φ] = T + U

where,

!
1 .
T [φ] = dd x{ (∂i φ(x)) 2
}
! 2
U [φ] = dd xV (φ(x))

Now define a new function φ′ (x) by


x
φ′a (x) ≡ φa ( ) (2.7)
s
where s is any positive number. Evaluate E for φ′ . This will turn E into a function
of s of the form,

E(s) = sd−2 T [φ] + sd U [φ]

This function is supposed to have a minimum at s = 1, ie. ,

dE(s)
0= |s=1 = (d − 2)T [φ] + dU [φ]
ds
11
Since T [φ] and U [φ] are both manifestly positive the r.h.s can not vanish for d
greater or equal to 2. In other words if d > 2 the only possibility is T [φ] = U [φ] = 0
which implies that φ(x) = v = constant.
For d = 1 we discover,

T [φ] = U [φ]

In fact the solutions we discovered do satisfy this equality.


For d=2 we find that

U [φ] = 0

Since we have assumed that the minimum value of the potential is V = 0 this
implies that the solution φ(x) is everywhere in ground state. It is a constant equal
to v.
This argument can be generalised. Consider an action of the form,

I[A, φ] = IY M [A] + Iφ [φ, A] + U [φ]

where,

!
1
IY M [A] = dd+1 x{− T rFµν F µν }
! 2
1
Iφ [φ, A] = dd+1 x{ gab (φ)Dµ φa (x)Dµ φb (x)}
! 2
U [φ] = dd+1 xV (φ(x))

where,

Dµ φa (x) = ∂µ φa (x) − ieAsµ (x)(T s )ab φb (x) (2.8)


12
T s are the generators of the gauge group in the representation to which the scalar
fields φa belong. Note that the kinetic energy terms of the φ fields can have a
non-trivial metric gab which is a function of φ’s but not their derivatives.
Under what conditions can such a model admit a finite energy static solutions?
The energy of this solution will be given by,

E[A, φ] = EY M [A] + Eφ [φ, A] + U [φ] (2.9)

where,

!
1 .2 .2
EY M [A] = dd x{ (E i + Bi }
! 2
1
Eφ [φ, A] = dd x{ gab (φ)Di φa (x)Di φb (x)}
! 2
U [φ] = dd x{V (φ(x))} (2.10)

Each term is separately positive. Now we define new functions,

1 x
A′µ (x) ≡ Aµ ( )
s s
x
φ′a (x) ≡ φa ( )
s
Evaluating the energy functional E for A′ and φ′ we obtain,

E[A′ , φ′ ] = sd−4 EY M [A] + sd−2 Eφ [φ, A] + sd U [φ]

Now evaluate s derivative at s = 1,

0 = (d − 4)EY M [A] + (d − 2)Eφ [φ, A] + dU [φ]

Clearly for d > 4, apart from the vacuum configuration, there can be no static
13
finite energy solution. For d = 4 we can satisfy this constraint only for pure Yang-
Mills configuration. It will be a static solution of the Yang-Mills equations in a 5
dimensional spacetime or in 4-dimensional Euclidean space. This latter solution is
called the instanton solution. Note that in this case the energy functional of the 5
dimensional theory or 4-dimensional Euclidean action will be scale invariant. For
d = 3 we can have solution in the presence of the scalar fieds. Magnetic monopoles
are the examples of such solutions. For d = 2 we obtain the (Nielsen-Olsen) gauge
string or the Abrikosov vortex solution.
This kind of demonstration of the conditions for the existence of finite energy
static solutions based on scaling arguments is called Derricks theorem.

2.D. Vortices and Strings

As we saw in the previous section the energetic arguments do in principle allow for
finite static solution in the presence of a vector field. So let us consider the scalar
QED in 1 + 2 dimensions. The action will be,

!
1
I[ϕ] = d3 x{∇µ φ∗ (x)∇µ φ(x) − V (φ) − Fµν F µν }
4
where

1
V (φ) = λ(φ∗ φ − v 2 )2
4
We have added a constant 18 λv 4 to the potential in order to make its minima to be
at V = 0, at the points where | φ |2 = v 2 .
We shall use the polar coordinates (r, ϕ) in the plane. For any fixed r we have a
circle S 1 in the (r, ϕ) space and as we go to large values of r, i.e. to the boundary of
1
space we obtain the boundary circle S∞ . As we said the potential is non-negative
and it assumes its minimum at
14

φ∗ φ = v 2

This defines the vacuum manifold which is also a circle, Sv1 , of radius v. An
arbitrary φ satisfying this equation can be written in the form of,

φ(t, r, ϕ) = veiα(t,r,ϕ) , α(t, r, ϕ) = real (2.11)

where, for any t and r, u(t, r, ϕ) ≡ eiα(t,r,ϕ) is a map from Sr1 to the group manifold
U (1). We write an Ansatz for a static solution of the form

φ(r, ϕ) = vf (r)eiα(ϕ)
= vf (r)u(ϕ) (2.12)

The phase u(ϕ) ≡ eiα(ϕ) defines a map u(ϕ) from the spatial S1 to the group U (1).
This mapp has to be single valuesd, viz,

α(r, ϕ + 2π) = α(r, ϕ) + 2nπ, n = 0, ±1, ±2, ... (2.13)

For later generalisation we write the integer n as,


! 2π
1
n=− dϕu(ϕ)∂ϕ u(ϕ)†
2πi 0
! 2π
1
= dϕ∂ϕ α(ϕ)
2π 0
1
= [α(ϕ + 2π) − α(ϕ)]

1
= 2nπ = n

Note that the function α(ϕ) can change in any continuous way. As long as the
condition of the single valuedness of u(ϕ) is maintained the integer n will not change.
15
It is the winding number of the map u : S 1 → U (1). It serves to classify all such
maps into equivalence classes. Two maps u and u′ are defined to be equivalent iff
they have the same winding number n. The maps with the same winding numbers
can be continuously deformed into each other. The set of all equivalence classes
is denoted by Π1 (S 1 ). Obviously it is isomorphic to the group of integers Z. A
representative of the class with the winding number n is

u(ϕ) = einϕ (2.14)

The covariant derivative is defined by,

∇µ φ = (∂µ − ieAµ )φ

This derivative transforms covariantly under,

φ′ (x) = u(x)φ(x)
i
A′µ (x) = Aµ (x) + u∂µ u−1 (2.15)
e

We would like to find finite energy static solutions of the coupled set of equations.
The most obvious solution is the vacuum solution,

φ(x) = v
Aµ (x) = 0

This solution has zero energy. However, it is not the unique zero energy solution.
Since the energy functional is gauge invariant any gauge transform of it will also
have zero energy. Thus the set of all zero energy solutions is given by,
16

φ(x) = vu(ϕ) = veinϕ


i n
Ai (x) = u∂i u−1 = ∂i ϕ
e e
Of course this still corresponds to the vacuum configuration, i.e.

∇µ φ = (∂µ − ieAµ )φ = 0, Fµν = 0, r ∕= 0

However, this configuration has a δ- function singularity at r = 0 for any n ∕= 0.


To see this we can evaluate the magnetic flux through any area,

!
Φ= .B
dS. .
!
= lim d.l.A.
r→∞
2πn
= (2.16)
e

Exercise

Evaluate the field strength F12 = B and show that it has a δ function singularity
at the origin.

We will search for a solution of the coupled system which approaches this con-
figuration as r → ∞ but will be regular everywhere. The Ansatz will be of the
form,

φ(r, ϕ) = vf (r)u(ϕ)
. ϕ) = 1 a(r)u∂u
A(r, . −1 (2.17)
ie
We shall impose boundary conditions which will smooth out the singularity at the
17
origin but will maintain the non zero flux. With our Ansatz the energy functional
becomes,

! ∞
n2 2 1 2 2 n2 ′2
E[f, g] = 2πv 2
rdr{f ′2 + (a − 1)f 2
+ λv (f − 1) 2
+ a } (2.18)
0 r2 4 e2 v 2 r 2
The boundary conditions in r should be chosen such that this integral converges
both as → 0 as well as → ∞. We impose the boundary conditions,

f (∞) = 1
a(∞) = 1 (2.19)

The polar coordinate system is not well defined at r = 0. Hence we demand


f (0) = 0. We also demand a(0) = 0 and ensure that there are no singularities at
the origin.
The equations which determine the functions f and g are obtained from,

δE[f, a]
=0
δf (r)
δE[f, a]
=0
δa(r)
It is convenient to introduce dimensionless coordinate, ρ = evr = mv r. We also
λ m2s
introduce β 2 ≡ e2 = m2v . In terms of these variables we obtain,

f ′ n2 f 1
f ′′ + − 2 (1 − a2 ) + β 2 (1 − f 2 )f = 0
ρ ρ 2

a
a′′ − + (1 − a2 )f 2 = 0 (2.20)
ρ
where now prime means derivative with respect to ρ.
It has been shown that these equations do have solutions subject to the given
boundary conditions, although no exact solutions are known. Numerical solutions
18
are also known. It is easy to obtain solutions for the limiting case of ρ ≪ 1 and
slightly more difficult for ρ ≫ 1. In the first case we obtain,

f (ρ) = ρ|n| + ...


a(ρ) = ρ2 + ... , ρ≪1

f (ρ) = 1 − e−cρ + ..., c = min(β, 2)


a(ρ) = 1 − e−ρ + ..., ρ≫1

In terms of the original radial variable r these functions exponentially approach


1 1
unity over a distance of mv and mφ . The magnetic field, however, does not vanish
at the origin. The picture which we get is that the Higgs field vanishes at the
1
origin and approaches its vacuum value a distance mφ far away from the origin.
The vanishing of φ at the origin means the maximum potential energy for this field
1
which is concentrated in an area of radius mφ around the origin in the x − y plane.
1
The magnetic flux is concentrated in the area of radius mv . Looking at the same
solution from the 1 + 3 dimensional point of view we shall have a concentrated tube
of energy and magnetic flux along the z- axis. For a heavy vector boson and in
1 1
case of mφ ∼ mv the tube will be a string like object with a finite constant energy
per length. This is called the gauge string. The embedding of this solution in more
complicated models has interesting applications in cosmology.

You might also like