UUITP-03/07

hep-th/yymmnnn

The Grassmannian Sigma Model in

SU (2) Yang-Mills Theory

David Marsh∗

Department of Theoretical Physics, Uppsala University

arXiv:hep-th/0702134v3 31 Jul 2007

SE-751 08, Uppsala, Sweden

Abstract
Spin-charge separation in pure SU (2) Yang-Mills theory was recently found to involve the
dynamics of an O(3) non-linear sigma model and, seemingly, a Grassmannian non-linear
sigma model. In this article we explicitly construct the Grassmannian sigma model of
the form appearing in the the spin-charge separated SU (2) theory through a quaternionic
decomposition of the manifold, thus verifying its relevance in this context. The coupling
between this model and the O(3) non-linear sigma model is further commented upon.

∗ dama3517@[Link]
1 Introduction

In a recent article by Faddeev and Niemi [1], spin-charge separation in pure SU (2) Yang-
Mills theory was investigated. Through a non-linear change of variables, the gluon field was
split into two charged scalar fields on the one hand, and an uncharged spin–carrying field
on the other. Similar approaches are well-known from condensed matter physics, where the
charged excitations are called holons, while the spin–carrying particles are called spinons.
It was found that the Yang-Mills theory counterparts of the spinon and the holon were
subject to a mutually attractive compact U (1) force, which is known to become strong at
short distances. This provides a rationale for the physical relevance of the phenomenon. It is
reasonable to expect that in the confining phase the gluon would seem pointlike, while under
very particular circumstances, involving the presence of a non-vanishing holon condensate,
the gluon might be split.
The possibility of this condensate holding the clues for understanding confinement in the
low-energy regime of the theory was discussed in [1], [2]. Its prospects for being the missing
mechanism for explaining high-temperature superconductivity is discussed in [2],[3]. Its
role as the scale factor of a conformally flat space-time and its implications for theories of
gravity was discussed in [1], [4].
Interpretations apart, the dynamics of the splitted gluons, i.e. the pure SU (2) Yang-Mills
Lagrangian in spin-charge separated variables, contains a form of the O(3) non-linear sigma
model, coupled to the real Grassmannian non-linear sigma model. The Grassmannian
present is the G2,4 , the manifold of two-dimensional subspaces of a four dimensional vector
space. This latter manifold is important in the interpretation of the internal U (1) force,
since it gives significant contributions to the field strength and seems on the whole to be a
key to understanding spin-charge separation in gauge theories.
Yet the form of the Grassmannian sigma model appearing in [1] is surprising, and, to our
knowledge, is not previously discussed in the litterature. It is therefore of value to show by
geometric construction that the Grassmannian sigma model may indeed be written on the
form suggested by Faddeev and Niemi, as we do in this note.
We also deepen the study of the coupling between this model and the O(3) non-linear sigma
model, which in [1] only was interpreted in certain limiting cases. As will be discussed, the
interaction terms may in general be written as a function of local sections of the holomorphic
cotangent bundle of the spheres factorizing the oriented Grassmannian.

2 Spin-Charge Decomposition

Following [1], we will be considering SU (2) Yang-Mills theory over Euclidean space-time
(R4 , δ) with indices a, b.
The SU (2) Yang-Mills gauge field A can be written in terms of the linear combinations

σi σ3 σ− σ+
Aa = Aia = Aa + Xa+ + Xa− .
2 2 2 2

1
The σ i are the Pauli matrices while the fields and off-diagonal projectors are denoted

Aa = A3a
Xa± = A1a ± iA2a
1
σ ± = (σ 1 ± iσ 2 ).
2

Under an infinitesimal gauge transformation U = eiθa σ

a /2
, with the choice of covariant
derivative D + = ∂ + iA, the gauge fields transform as

δθ Aia = ∂a θ i + ǫijk Aja θ k . (1)

Consider gauge transformations along the σ 3 direction only. These transformations form an
U (1) subgroup of SU (2), denoted UC (1) , where C stands for colour. For an infinitesimal
h ∈ UC (1) ,
3
h = eiωσ /2 .
According to (1), the gauge fields now transform as

δh Aa = ∂a ω
∂h X ± = ∓iωXa± . (2)

So Aa transforms as the UC (1) gauge field and the off-diagonal components as UC (1) charged
fields. The third component of the SU (2) field strength is
3
Fab =∂a A3b − ∂b A3a + ǫ3jk Aja Akb =
i
=∂a Ab − ∂b Aa + (Xa+ Xb− − Xb+ Xa− ) = Fab + Pab , (3)
2
where Fab is the U (1) field strength and Pab is given by

i
Pab = (Xa+ Xb− − Xb+ Xa− ) = A1a A2b − A1b A2a . (4)
2
This second rank tensor has a natural interpretation in terms of the Grassmannian G2,4 of
unoriented two dimensional planes through the origin of a four dimensional vector space.
Since SO(4) (with the standard action) acts transitively on G2,4 , and the stabilizer at
p ∈ G2,4 is given by rotations in the plane p as well as in the plane orthogonal to p, the
representation of the manifold as a homogeneous space is

SO(4)
G2,4 ≃ .
SO(2) × SO(2)

Analogous to the homogeneous coordinates of real projective space, p ∈ G2,4 can also be
represented by a 2 × 4 matrix M whose row-vectors span p. Let A1 , A2 ∈ R4 span p. A
matrix representative of p is given by
 1
A1 A12 A13 A14

M= .
A21 A22 A23 A24

2
The representation is not unique since, if equivalence is taken to mean representing the
same element of the Grassmannian, then for all g ∈ GL(2, R)

M ∼ g M. (5)

Thus, the manifold can be represented as the coset space

M2,4
G2,4 ≃ ,
GL(2, R)

where M2,4 denotes the set of 2 × 4 matrices of maximal rank.

Standard local coordinates (equivalents of the inhomogeneous coordinates for the projective
space) are obtained in six charts for the pairs (i, j) i, j = 1, . . . , 4. Since M is of maximal
rank, at least one of its minors is non-vanishing, and, equivalently, at least one of its six
2 × 2 column matrices, mij , is invertible. To be explicit, consider the chart (1, 2), and
suppose that the inverse of the matrix formed out of the first and second column of M
exists. Denote it m−1
12 .
!
P32 P42
1 0
M ′ = m−1
12 M =
P12 P12 ,
0 1 PP13 12
P14
P12

where equation (4) has been used. It follows that

P12 P34 + P13 P42 + P14 P23 = 0 . (6)

This is the Plücker equation. It describes a quadratic hyper-surface in R6 called the Klein
Quadric. The anti-symmetric tensor Pab with six independent entries is called the Plücker
Compound and its entries the Plücker coordinates. The equivalence relation, (5), turns into

Pab ∼ det g Pab .

So, supplying the Grassmannian with Plücker coordinates embeds the manifold as the Klein
Quadric in RP 5 .
Define the complexified four-component vector va = A1a + iA2a . Note that equation (5)
implies that

v a ∼ w va , w ∈ C. (7)

That is, va is an element of the three dimensional complex projective space, CP 3 . Represent
p ∈ G2,4 by an orthonormal frame, e1a , e2a and introduce the spinon field ea = √12 (e1a + ie2a )
that satisfies

ea ea = 0 (8)
ēa ea = 1. (9)

Note that by these restrictions w in (7) turns into a phase.

3
Spin-charge separation1 is obtained through the change of variables

Xa+ =ψ1 ea + ψ2 ēa , (10)

where ψ1,2 are complex valued scalars called holons. We rewrite the Plücker compound in
these coordinates to obtain
i
Pab = (|ψ1 |2 − |ψ2 |2 )(ea ēb − eb ēa ) =
2
i
= ρ cos θ (ea ēb − eb ēa ) = ρ2 t3 Hab ,
2
(11)
2
where the condensate ρ and the real vector ~t are defined as

ρ2 =|ψ1 |2 + |ψ2 |2 (12)

 
  cos φ sin θ
~t = 1 (ψ1∗ , ψ2∗ ) ~σ ψ1
=  sin φ sin θ  . (13)
ρ2 ψ2
cos θ
The factorization of the Plücker compound into a pre-factor and the tensor Hab corresponds
to restricting the representation of the Grassmannian to orthonormal pairs of spanning
vectors. The tensor Hab is invariant under the changes of matrix representative, M , in (5).
In analogy with the electromagnetic field strength, the ”electric” and ”magnetic” fields are
defined as
~ )i =H0i
(E (14)
~ )i = 1 ǫijk Hjk ,
(B (15)
2
satisfying

E~ ·B~ =0 (16)
E~ 2+B~ 2=1 (17)
~ ,B
(E ~ ) ∼ −(E~ ,B
~ ). (18)

In this notation, (16) is the Plücker equation, while the constraints (17) and (18) reduce
the space to RP 5 by identifying anti-podal points on S 5 . Thus we can write
~

E = cos ϑ k̂

~ = sin ϑ ˆl
B (19)
~

~ ×B ~ .
S =E

For future reference we will also introduce a parametrization of the holons

θ
ψ1 = ρeiξ cos e−iφ/2 (20)
2
θ
ψ2 = ρeiξ sin eiφ/2 . (21)
2
1
Using a different separation between the spin and the charge variables, it has been found that SU (2)
Yang-Mills theory possesses features similar to those of liquid crystals, see [5]

4
Although Pab is invariant, the spinon ea picks up a phase under rotation of the orthonormal
frame e1 , e2 in (7), so the off-diagonal gauge fields are invariant under UI (1) ⊂ SL(2, R)
acting as

Aa → Aa , X + → X +
ψ1 →eiλ ψ1 (22)
−iλ
ψ2 →e ψ2 (23)
e →e−iλ e. (24)

Clearly ρ2 and ~t are invariant. A suitable UI (1) gauge field is

Ca = iēb ∂a eb
Ca → Ca + ∂a λ. (25)

Also note that

t± := t1 ± it2 →e∓iλ t± . (26)

This motivates the introduction of an UC (1) × UI (1) invariant vector ~n. Take

e0 = eiη |e0 | (27)

iη
ea = e êa , (28)

so that η → η − λ in (24). Then ~n, which will describe the O(3) non-linear sigma model,
can be defined as

n± =e2iη t±
n3 =t3 .

In terms of these new coordinates, the UC (1) transformation, (2) is

Aa → Aa + ∂a ω, X ± → e∓iω X ± (29)
−iω
ψ1 →e ψ1 (30)
−iω
ψ2 →e ψ2 (31)
ea →ea . (32)

For UC (1) the gluon field degrees of freedom are described by the multiplet

Ai ∼ (Aa , ψ1 , ψ2 , ea ).

The first field is the gauge field, while the second two are charged but spin-less complex
fields that transform with the same charge under UC (1) . These holons are oppositely
charged with respect to the internal force, giving rise to a mutual attraction between the
holons and the spinon, in the decomposition (10). The spinons on the other hand do not
carry colour charge but transform projectively under Lorentz (SO(4)) transformations [1].

5
Considering the UI (1) transformations the multiplet structure is

Ai ∼ (Ca , ψ1 , ψ2 , Aa ).

The dynamics of the gluon fields in this decomposition are stipulated by the classical SU (2)
Yang-Mills Lagrangian with the off-diagonal gauge fields fixed (Maximum Abelian Gauge)
1 i 2 1 + +2
LY M = (Fab ) + |Da Xa | . (33)
4 2
Re-expressing it in spin-charge separated, UC (1) × UI (1) invariant variables leads to [1]

1 2 1 1 ρ2 ~ )2 + (∂ B
~ )2 }
LY M = Fab + ρ2 Ja2 + ρ2 (DaĈ ~n)2 + {(∂ E
4 2 8 4
1 3 3
+ ρ2 {n+ (∂a eˆ¯b )2 + n− (∂a êb )2 } + (1 − n3 )ρ4 − ρ4 , (34)
4 8 8
where
1
Fab = ∂a Jb − ∂b Ja + ~n · ∂a~n × ∂b~n − {∂a (n3 Ĉa − ∂b (n3 Ĉa )} − 2ρ2 n3 Hab (35)
2
i ∗
Ja = 2 {ψ1 Da ψ1 − ψ1 D̄a ψ1∗ + ψ2∗ Da ψ2 − ψ2 D̄a ψ2∗ }. (36)
2ρ
The covariant derivatives and the connection are defined as

Da± = ∂a ± iAa (37)

DaC ψ1 = (∂a + iAa − iCa )ψ1 (38)
DaC ψ2 = (∂a + iAa + iCa )ψ2 (39)
DaC eb = (∂a + iCa )eb (40)
Ĉa = iēˆ · ∂a êa = Ca + ∂a η. (41)

The terms of present interest include

ρ2  ~ 2 
~ )2 ,
(∂ E ) + (∂ B (42)
4
which is the ostensible G2,4 Grassmannian sigma model in [1]. The coupling between this
model and the O(3) non-linear sigma model described by ~n is given by

ρ2
{n+ (∂a ê∗b )2 + n− (∂a êb )2 }. (43)
4

3 Quaternionic Decomposition

If (42) actually corresponds to the G2,4 sigma model, the metric tensor of the six-dimensional
space of Plücker coordinates (E ~ and B ~ space) should be taken to be flat, at least in
the neighbourhood of the embedded Grassmannian. Consistently joining this fact with

6
(A1a , A2a ) ∈ (R4 × R4 , δ) may formally be stated as finding a solution to a set of 15 coupled
partial differential equations. We will not pursue this direct approach but instead induce
the metric tensor on G2,4 by using the spherical factorization of the Grassmannian of ori-
ented two dimensional subspaces of a four dimensional vector space, G̃2,4 . The oriented
Grassmannian is the two-fold covering of G2,4 and satisfies equations (16), (17), but not
(18). It is embedded as the Klein Quadric in S 5 .
There are several well-known ways of doing this decomposition, the arguably simplest is
by representing each p ∈G̃2,4 by a decomposable element of Λ2 (R4 ). The Hodge star dual
operator maps each oriented plane to its oriented orthogonal complement. Expressing p in
the eigenbasis of the Hodge star exhibits the product sphere nature of G̃2,4 .
In this context however, it is fruitful to take p ∈ G̃2,4 spanned by orthonormal basis (e1 , e2 )
and identify these vectors with the quaternions (H, δ) in an obvious way

ea = e0 + ie1 + je3 + ke3 = e0 + ~e,

with i2 = j 2 = k2 = ijk = −1. The quaternionic multiplication rules may be expressed as,
for p, q ∈ (H, δ)

p = a + ~u q = b + ~v
pq = ab − ~u · ~v + a~v + b~u + ~u × ~v . (44)

Two independent divisions of the spanning quaternions can be made

 2
e ~ −B~ = ξ~−
= e2 ē1 = E (45)
e1 R
 2
e ~ +B~ = ξ~+ .
= ē1 e2 = E (46)
e1 L

Since ξ~ ± = 1, these are maps from G̃2,4 to S 2 × S 2 ⊂ (H × H, δ). The spheres are
standardly interpreted as being the sphere of complex structure of R4 and a trivial CP 1
bundle.
To see this, associate with p an almost complex structure consisting of positive π2 rotations
of the vectors in p and its orthogonal complement, and extend this rotation linearly to the
entire space H. Denote this almost complex structure Jp . Then (Jp )2 = −1.
Under the action of Jp , the real unit quaternion u = 1 + ~0 will be rotated into some unit
quaternion orthogonal to it.

Jp (u) = v(p) ∈ S 2 ,

where the sphere is the sphere of imaginary unit quaternions. The pair (u, v(p)) can be
used to label the complex structures on R4 , and the sphere is the set of complex structure
of R4 . Since Jp : (e1 , e2 ) → (e2 , −e1 ), we can identify v(p) ∈ S 2 explicitly
 2
e ~ −B~ = ξ~− .
v(p) = = e2 ē1 = E (47)
e1 R

7
The elements p̃ of the Grassmannian sharing complex structure with p satisfy

V ∈ p̃ Jp (V ) ∈ p̃.

In other words, the set of p̃ sharing a complex structure is the set of one-dimensional complex
subspaces with respect to Jp , i.e. CP 1 ≃ S 2 . The independent coordinate on this sphere is
given by (46).
In conclusion we can write (ξ~+ , ξ~− ) ∈ S 2 ×S 2 ⊂ (H ×H, δ) and induce the Euclidean metric
on the spheres to get the standard round metric in spherical local coordinates (α1 , β1 , α2 , β2 )

ds2 = dα21 + sin2 α1 dβ12 + dα22 + sin2 α2 dβ22 .

Consistently we can take the metric of the six-dimensional space of independent Plücker
coordinates to be flat, at least in the neighbourhood of the Grassmannian. This results in
the non-linear sigma model Lagrangian
~ )2 + (∂ B
LG2,4 = gab ∂a pi ∂b pi = (∂ E ~ )2 , (48)

with pi , i = 1, . . . , 6 being the Plücker coordinates in agreement with [1]. The subsidiary
constraints are given by (16)-(18), such that G2,4 ≃ RP 2 × RP 2 .

4 The Interaction

The O(3) non-linear sigma model appearing in the decomposition is given by

(∂~n)2 + (~n · (∂a~n × ∂b~n))2 ,

for the unit vector ~n. This model has since long been thought to be related to the low-energy
regime of SU(2) Yang-Mills, and it is known to support stable knotted solitons, as discussed
in [1]. The coupling between the O(3) non-linear sigma model and the Grassmannian model
is quite intricate, and was only analysed in [1] in the purely electric (ϑ → 0 in (19)) and
purely magnetic (ϑ → π/2) limits. We will analyse the situation in general and will be able
to assert that the coupling is related to local sections of the holomorphic cotangent bundle
(T ∗(1,0) S 2 ) of the spheres factorizing the Grassmannian.
In the two sets of spherical coordinates (α1 , β1 , α2 , β2 ) of the two spheres of G̃2,4 , these
forms are given by

dzi = dαi + i sin αi dβi , i = 1, 2. (49)

The coupling (34), can be elaborated by plugging in the expression for e in terms of the
electric and magnetic fields,

ρ2 n +  ~ +B
 
~ ) · E~ −B~ − 2iS
~

~ −B
 
~ ) · E~ +B~ − 2iS

~ +
∂a (E ∂a (E
128S~ 2
2
ρ n−       
+ ~ +B
∂a (E ~ ) · E~ −B~ + 2iS
~ ~ −B
∂a (E ~ ) · E~ +B~ + 2iS
~ . (50)
128S~ 2

8
Expressed in the spherical unit vectors,

ρ2 n+      
~+ ) · ξ~− + i(ξ~+ × ξ~− ) ∂a (ξ~− ) · ξ~+ + i(ξ~+ × ξ~− ) +

∂a (ξ
128(sin ϑ cos ϑ)2
ρ2 n− 
~+
 
~− ~+ ~ −

~−
 
~+ ~+ ~ −

+ ∂ a (ξ ) · ξ − i(ξ × ξ ) ∂a (ξ ) · ξ − i(ξ × ξ ) .
128(sin ϑ cos ϑ)2
(51)

For the moment, identify

 
cos β1 sin α1
ξ~+ = r̂1 =  sin β1 sin α1  ,
cos α1

where the components are taken with respect to the standard Cartesian basis. Expressing
ξ~− in the orthonormal triplet (r̂1 , β̂1 , α̂1 ), of the S+
2 gives

ξ~− = cos 2ϑ r̂1 + sin 2ϑ(cos τ β̂1 + sin τ α̂1 ), (52)

where the degrees of freedom previously described by the pair (α2 , β2 ) now are replaced by
the angles (ϑ, τ ). We have for a = 1, . . . , 4

∂a (r̂ )dxa = ∂i (r̂ )dαi1 = sin α1 dβ1 β̂1 − dα1 α̂1 . (53)

Using that

∂a (ξ~+ ) · (ξ~− + iξ~+ × ξ~− )dxa =

 
= ∂i (r̂)dαi1 · sin 2ϑ(cos τ βˆ1 + sin τ αˆ1 ) + i(r̂ × sin 2ϑ(cos τ βˆ1 + sin τ αˆ1 )) =
= sin 2ϑ (−(sin τ + i cos τ )dα1 + sin α1 (cos τ − i sin τ )dβ1 ) =
= sin 2ϑ e−i(τ + 2 ) (dα1 + i sin α1 dβ1 ) ,
π

the interaction terms can be reformulated. Write the terms with the derivative acting on the
S+2 vector in terms of the angles (β , α , ϑ, τ ) as above, while the terms with the derivative
1 1
acting on the S−2 vector are expressed in (α , β , ϑ, τ ′ ). The inner product in the cotangent
2 2
space is Euclidean, so the first term of (51) turns into
1 2 ′
ρ n+ ei(τ −τ ) (dz1 , dz̄2 ). (54)
32
The full interaction between the Grassmannian sigma model and the O(3) sigma model is
given by,
1 2 ′ 1 2 ′
ρ ℜ{n+ ei(τ −τ ) (dz1 , dz̄2 )} = ρ sin θ ℜ{ei(φ−2η+τ −τ ) (dz1 , dz̄2 )} (55)
16 16
where the angles (θ, φ, η) correspond to the degrees of freedom of the O(3) sigma model, as
defined in equations (20) and (27). This establishes how the Grassmannian sigma model of
G2,4 , as it appears in the spin-charge separated SU (2) Yang-Mills Lagrangian, relates to
the O(3) non-linear sigma model.

9
5 Discussion

We have shown that the Grassmannian sigma model G2,4 appears in the spin-charge sep-
arated SU(2) Yang-Mills theory, and that the interaction can in general be expressed as
involving the inner product of the local sections of the holomorphic and anti-holomorphic
cotangent bundles of the constituent spheres of G̃2,4 .
In [5], through a different spin-charge separation in a different gauge, the Yang-Mills theory
was cast in a form very similar to the condensed matter system of nematic liquid crystals.
Interestingly, also the current decomposition mimics this property.
The product manifold decomposition of the unoriented Grassmannian, G2,4 ≃ RP 2 × RP 2 ,
is naturally interpreted as the product of two coupled nematic crystals. Particularly, in the
manifestly Lorentz invariant vacuum of the theory (θ → 0) [1], the crystals decouple and the
O(3) non-linear sigma model no longer contributes to the dynamics. Possible topological
defects are typically classified by the homotopy of the vacuum manifold, and the existence
of non-trivial vortices in nematic crystals in three dimensional space corresponds to the
non-trivial element of π1 (RP 2 ) = Z2 . This suggests that also the Grassmannian part of the
decomposed Yang-Mills theory may support non-trivial vortices.
These results are parts of a larger programme on understanding and interpreting the struc-
ture emerging from the Yang-Mills Lagrangian in spin-charge separated variables.

6 Acknowledgements

I would like to thank prof. Antti Niemi for presenting me this problem and for discussions.
I also owe thanks to Sergey Slizovskiy for discussions and translations and in particular to
Maxim Zabzine for discussions and clarifications. I would like to thank Marit Strömberg
and Olof Ohlsson-Sax for proof-reading.

References

[1] Faddeev, L.D. and Niemi, A.J. arXiv:hep-th/0608111, Nuclear Physics B Volume 776,
Issues 1-2, 30 July 2007, Pages 38-65.
[2] Niemi, A.J. AIP [Link].806: 114 -123, 2006. arXiv:hep-ph/0510288.
[3] Andersson, P.W. Science 235 (1987) 1196 and Faddeev, L.D. and Takhtajan,
L. [Link]. A85 (1981) 375 and Lee, P.A., Nagosa, N. and Wen, X.-G.
arXiv:cond-mat/0410445.
[4] Slizowskiy, S. arXiv:hep-th/0612055.
[5] Chernodub, M.N. [Link].B637:128-132,2006, arXiv:hep-th0506107 and Chernodub,
M.N. JETP Lett.83:268-272,2006, arXiv:hep-th/0507221.

10