You are on page 1of 27

Twistor-Hopfions and Gravitational Linking

Alexander Wickes

August 27, 2013

University of California, Santa Barbara


College of Creative Studies
Submitted in fulfillment of the requirements for the
degree of Bachelor of Science in Physics

Advisor: Professor Dirk Bouwmeester Date

Student: Alexander Wickes Date


Abstract
The electromagnetic hopfion is a topologically nontrivial radiative solution to the
Maxwell equations with field line structure topologically equivalent to the Hopf fibra-
tion at all times. In Penrose’s twistor formalism, an elementary state is a spin-(n/2)
field resulting from the Penrose transform of a twistor wavefunction of simplest non-
trivial form. We derive expressions for a few classes of elementary states. Furthermore,
we identify the electromagnetic hopfion as an elementary state and develop general-
izations called twistor-hopfions. Finally, motivated by the Hopf structure seen in the
general twistor-hopfion, we develop the concept of gravitational linking and propose a
definition for the gravitational helicity in analogy with the classical magnetic helicity.

1
Contents
1 Introduction 3

2 Twistors and Their Relevance 4


2.1 Spinors and 2-Spinor Calculus . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Twistors Encode Total Momentum in a Holomorphic Structure . . . . . . . . 6
2.3 Solving the Massless Field Equation: The Penrose Transform . . . . . . . . . 8
2.4 Evaluating the Penrose Transform on Elementary States . . . . . . . . . . . 9

3 Twistor-Hopfions: Electromagnetic Case 11


3.1 The Null EM Hopfion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2 The Non-null EM Hopfion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

4 Gravitational Hopfions and Gravitational Linking 14


4.1 A Method for Plotting Gravitational Field Lines . . . . . . . . . . . . . . . . 15
4.2 The Type N Gravitational Hopfion . . . . . . . . . . . . . . . . . . . . . . . 17
4.3 The Type III Gravitational Hopfion . . . . . . . . . . . . . . . . . . . . . . . 19
4.4 Gravitational Helicity and Linking . . . . . . . . . . . . . . . . . . . . . . . 21

5 Conclusion 23

6 Acknowledgements 23

7 References 23

2
Figure 1 The Robinson congruence on a constant-
time hypersurface. The time-slice of the congruence
lies tangent to the Hopf fibration stereographically
projected onto R3 . In other words, each of the curves
in the image is a Villarceau circle on a torus where
the tori are nested and fill all of space, degenerating
into the infinitely-thin torus circling the central struc-
ture and the infinitely-thick torus (straight line) going
through the central structure. [Image from [5].]

1 Introduction
Following Dirac’s discovery in 1928 of the relativistic wave equation for the electron, physi-
cists were intrigued with the notion of a relativistic wave equation for particles of arbitrary
spin. This idea was explored in some depth by Dirac, Pauli, and Wigner, among others, in
the next few decades [1–3]. The equations they discovered made extensive use of spinors; in
particular for massless fields, they exhibited a great deal of symmetry, being invariant under
the conformal symmetry group.
The conformal symmetry exhibited by the massless field equations, along with an intu-
ition for analytic continuation, led Penrose to develop his twistor formalism, a reparametriza-
tion of total momentum in terms of spinors [4]. The name ‘twistor’ in fact came from the
so-called Robinson congruence, a spacetime-filling set of twisting shear-free null rays with
the structure of the Hopf fibration stereographically projected to R3 (at each constant-time
hypersurface of Minkowski space M) and which moves at the speed of light opposite the
direction of the one straight line through the center (see Figure 1).
This congruence was first brought to Penrose’s attention by Ivor Robinson, who tried
using it in search of a nowhere-singular vacuum electromagnetic field of finite total energy.
Although Robinson was apparently unsuccessful, Penrose found such a field using early
twistor methods in 1965 [6]. This field, known as an electromagnetic hopfion, has appeared
time and again in the research literature in various forms: in Volovik and Mineev’s 1977
analysis of particle-like solitons in superfluid phases [7]; in Kamchatnov’s topological MHD
solitons of 1982 [8]; in Rañada’s work on a topological theory of electromagnetism starting
in 1989 [9, 10]; in Faddeev and Niemi’s exploration of knot-like structures in solitons in
1997 [11]; and in Irvine and Bouwmeester’s work on visualizing and generalizing Rañada’s
work to knotted field configurations [12].
More recently, there has been growing interest in generating this type of field via twistor
methods à la Penrose [13] and in visualization and generalization to arbitrary spin [14].
The primary purpose of this thesis is to present recent work1 on generating fields based
on the Hopf fibration using twistor methods, hereafter referred to as twistor-hopfions or
1
Presented [15] and to be presented in a future publication [16].

3
often just hopfions. Furthermore, we present an analysis of the gravitational solutions,
including a method of visualization based on the technique of Nichols et al using the gravito-
electromagnetic decomposition of the Weyl tensor [17].
Finally, a discussion is presented on the topic of linking, specifically in the context of the
gravitational field. The related notion of (singular) rings has appeared in the rotating black
holes of Kerr and Newman [18], and more recently in the 5-dimensional rotating black ring
of Emparan and Real [19]. However, there has been little, if any, work done on the linking
of the gravitational field itself. This can be attributed to the fact that until the recent work
of Nichols et al there was no concept of gravitational field lines. Therefore, an elementary
development of gravitational linking in terms of gravito-electromagnetism is presented.

2 Twistors and Their Relevance


Ever since Witten’s seminal casting in 2003 of perturbative gauge theory as a string theory
in twistor space [20], twistors have played a prominent role in the study of pure Yang-Mills
theory, especially of scattering amplitudes in N = 4 supersymmetric Yang-Mills theory
[21–24]. Although the present work does not directly relate to proper Yang-Mills theory,
this still provides some motivation to more fully explore twistors and their application in
various areas of physics. The twistor formalism relies heavily upon the use of spinors. Thus,
we begin by giving an introductory review of the spinor calculus.

2.1 Spinors and 2-Spinor Calculus


The essential idea behind spinors is that vectors are inadequate if we wish to describe not
only the orientation of an object in spacetime but also the entanglement of that object
with its environment, as is often the case in quantum mechanics. This is the so-called
orientation-entanglement relation [25]: an object entangled to its environment is fundamen-
tally distinguishable after one full rotation, but is indistinguishable after two full rotations
(see Figure 2).
Mathematically speaking, the special orthogonal group SO(p, q) of rotations/boosts in
a Minkowskian spacetime of metric signature (p−, q+) has a double cover called Spin(p, q).
The objects that this group acts on (generalizing the notion of tensors) are called the spinors
of SO(p, q); they pick up a minus sign under rotation by angle 2π and thus return to their
original sign under rotation by angle 4π. An elucidating description is attributed to Michael
Atiyah [26]:

No one fully understands spinors. Their algebra is formally understood but their
general significance is mysterious. In some sense they describe the “square root”
of geometry and, just as understanding the square root of −1 took centuries, the
same might be true of spinors.

Spinors play a central role in the twistor formalism, in which we associate with each

4
Figure 2 The orientation-entanglement relation in
three dimensions: a cube tied to its environment by
strings fixed to the corners of a room becomes entan-
gled after one full rotation. After another full rotation,
the cube can be disentangled. [Image from [25].]

0
4-vector xa an SO(3, 1) spinor xAA under the correspondence2
 0 
a AA0 a AA0 1 x + x3 x1 + ix2
x ↔x ≡ x σa = √ 1 2 0 3 , (1)
2 x − ix x − x
It is useful and typical to equate the two for simplicity of notation. When doing so, lower
case Latin letters (world indices) will correspond to the associated unprimed and primed
0 0 0 0 0
capital Latin letters (spinor indices), e.g. xa = xAA , F ab = F AA BB = F ABA B , and so on.
Note that due to the distinguishable nature of primed vs unprimed indices we are free to
move primed indices past unprimed indices and vice versa without ambiguity.
0
The spinor representation xAA of a world vector xa is also known as the SL(2, C) repre-
0
sentation. This is because, denoting the matrix with components xAA by X, we have
1
det X = x2 , (2)
2
which is invariant under X → LXL† for L ∈ SL(2, C). In other words, the Lorentz group
SO(3, 1) is represented here by SL(2, C)/Z2 (notice that L and −L represent the same
transformation). Furthermore, this determines the manner in which spinor indices are raised
0 0
and lowered: by contracting with the antisymmetric tensor εAB (or εA B ), so that for example
AB 0
xAA0 = x εB 0 A0 . Because of the antisymmetry of ε, it is important to keep track of which
indices are raised and which are lowered; this is not the case with world indices.
Finally, a spinor with a single index is referred to as a spin-vector. Given a spin-vector
κA0 we define a null vector called its flagpole by ka = κ̄A κA0 .3 Notice that if a vector pa in
2
In the present work we use the following index conventions: early and latter lower case Latin letters
for world indices (a, b, c, . . . and p, q, r, s, . . . ), middle lower case Latin letters for spatial indices (i, j, k, . . . ),
primed and unprimed upper case Latin letters for spinor indices (A, A0 , B, B 0 , . . . ), and Greek letters for
twistor indices (α, β, γ, . . . ).
3
Incidentally, there is another object associated to a spinor called its flag which represents the overall
phase. Under a continuous rotation κA0 7→ eiθ κA0 with θ varying from 0 to π, the flag of κ returns to its
original position for a total rotation of 2π, while κ itself picks up a minus sign. Thus in order to return a
spinor to its original state its flag must undergo a total rotation of 4π.

5
complexified Minkowski space CM is null then the determinant of pAA0 is zero. In that case
the matrix pAA0 is of rank 1 and can thus be written as the outer product of two spin-vectors,
pa = πeA πA0 , where π
e and π are not necessarily related. If further pa is purely real then pAA0
is Hermitian and so π e is just the complex conjugate π̄ of π so that p is the flagpole of π. We
will use the convention that the “complex conjugate” or “c.c.” of a spinor, e.g. κA0 , gives
the conjugate spinor, e.g. κ̄A .
Finally, note that we make extensive use of standard index symmetrization and anti-
symmetrization, for example
1
ψ (ab) ≡ (ψ ab + ψ ba ), (3)
2!
1
φ[abc] ≡ (φabc − φbac + φbca − φacb + φcab − φcba ). (4)
3!
We will also make use of standard notation for inner products between spinors,
0 0
hπκi ≡ εAB πA κB , [πκ] ≡ εA B πA0 κB 0 . (5)

2.2 Twistors Encode Total Momentum in a Holomorphic Struc-


ture
Twistors have been defined in a variety of ways, and thus it is somewhat difficult to give a
concrete definition. For example, twistors are sometimes thought of as the reduced spinors of
SO(2, 4) [5, p. 440], although this definition is not particularly useful in most applications.
The most relevant definition of twistors for this application is that they combine the 4-
momentum pa and the 6-angular momentum M ab in a simple form, unique up to phase, with
a holomorphic structure that will be described in the following paragraphs [27].4
More precisely, a twistor Z α is defined as a pair of constant spinors

Z α = (ω A , πA0 ) = (ω 0 , ω 1 , π00 , π10 ) (7)

representing a null complex line z a in CM, called its α-plane, via the incidence relation
0
ω A = iz AA πA0 . For null twistors, i.e. twistors whose product (12) defined below is zero, the
α-plane intersects real Minkowski space M on a null ray. Furthermore, for arbitrary points
x in CM we define the associated spinor field
0
ΩA (x) ≡ ω A − ixAA πA0 (8)
4
A more utilitarian (and equivalent) definition used in high energy theory is as follows, see e.g. [28, p.
494]. Decompose pµ = λα λ eα̇ and define the twistor Z = (λ, µ) and dual twistor W = (e µ, λ)
e where µ and µ e
are the Fourier conjugates of λ
e and λ respectively, so that scattering amplitudes in twistor space and dual
twistor space are written
Z Z
2 e iµα̇ λ α
M (Z) = d λe M (λ, λ), M (W ) = d2 λeieµ λα M (λ, λ). (6)
eα̇ e e

The relationship between the Fourier structure of twistors seen here and the holomorphic structure of twistors
described herein is strikingly beautiful: it turns out that writing scattering amplitudes in terms of twistors
simplifies many hitherto extremely complicated expressions immensely [24].

6
which vanishes on the α-plane defined by Z.5 From here we define the quantities
0 0
pa = π̄A πA0 , M ab = iΩ(A π̄ B) εA B + c.c. (9)
(A0 B0)
where c.c. denotes the complex conjugate of the previous term, here −iΩ̄ π εAB . Obvi-
ously p is constant in spacetime, and it can easily be seen that M ab has x dependence
M ab (x) = M ab (0) − 2x[a pb] . (10)
Thus, these quantities obey the transformation laws associated with the 4-momentum and
the 6-angular momentum respectively, and as such are precisely those quantities.
Next, we define the dual twistor (dual to Z) by
0 0 0
Z̄α = (π̄A , ω̄ A ) = (π̄0 , π̄1 , ω̄ 0 , ω̄ 1 ), (11)
so that the twistor product is defined by
0
Z α Z̄α = ω A π̄A + πA0 ω̄ A . (12)
This definition is such that for a massless particle the helicity is h = 12 Z α Z̄α . The helicity of
a massless particle is defined as the constant of proportionality between the Pauli-Lubanski
spin vector Sa and the 4-momentum pa , i.e.
hpa = Sa ≡ ∗Mab pb (13)
where ∗ is the Hodge dual operator, ∗Mab = 12 εabcd M cd . Thus the relation between the
helicity and the twistor product can be seen immediately from
0
Sa = ∗Mab pb = (ω(A π̄B) εA0 B 0 + c.c.)π̄ B π B (14)
1
= ( Z α Z̄α )pa . (15)
2
This will become important later when we find solutions to the massless free field equation.
From the above definitions, it is straightforward to check that the commutation relations
for the Poincaré algebra of p and M on M,
1 1 [a b]
[pa , pb ] = 0, [M ab , pc ] = η [a c pb] , [M ab , Mcd ] = η [c M d] , (16)
2i 4i
are completely equivalent to a set of commutation relations for Z and Z̄ given by
[Z α , Z β ] = 0, [Z̄α , Z̄β ] = 0, [Z α , Z̄β ] = δβα . (17)
These are in fact the canonical commutation relations for general Fourier conjugates. Thus,
upon first quantization of the twistor variables, a general twistor function f (Z) can also
be represented as a function f (Z̄) on dual twistor space. This is completely analogous to
ordinary quantum mechanics, where the canonical commutation relations (i[xa , pb ] = ηba and
so on) imply that a wavefunction ψ on spacetime can be represented either in position space
by ψ(x), or in momentum space by ψ(p). In the case of twistor wavefunctions, however,
the choice of representation implies that the wavefunction is either holomorphic, with Z̄α
represented by −∂/∂Z α , or anti-holomorphic, with Z α represented by ∂/∂ Z̄α . It is in this
sense that twistors encode total momentum in a holomorphic structure.
5
The flagpole of the spinor field Ω(x) is in fact tangent to the Robinson congruence [5, pp. 60-62], twisting
around itself and hence giving rise to the name ‘twistor’ as previously mentioned.

7
2.3 Solving the Massless Field Equation: The Penrose Transform
As mentioned in section 1, Penrose was partly motivated to develop his twistor formalism
in search of general solutions to the massless free field equations. He was indeed successful,
which will be made clear shortly. We will focus on fields with non-negative helicity; the
associated treatment of negative-helicity fields is then straightforward but not relevant to
the topics discussed herein.
Consider the field ϕA01 ···A02h (x) of a massless free particle of helicity h = 0, 12 , 1, 23 , 2, . . . in
an arbitrary spacetime. The general form of the massless free field equation is given by
0 1
∇AA1 ϕA01 ···A02h = 0 or (∇a ∇a + R)ϕ = 0, (18)
6
where the second form is the scalar case h = 0 with R the scalar curvature of the spacetime,
although we will be working in flat spacetime (R = 0) for the remainder of the present
work. It is important to note that a massless free field is symmetric in its spinor indices, i.e.
0
ϕA01 ···A02h = ϕ(A01 ···A02h ) , so that the covariant derivative ∇AA1 could just as well be replaced by
0
any other ∇AAr , r = 1, . . . , 2h. This set of equations is quite remarkable in that it contains
the massless Dirac equation (spin 1/2), the vacuum Maxwell equations (spin 1), the massless
Rarita-Schwinger equation (spin 3/2), linearized Einstein gravity (spin 2), and so on.
The essence of Penrose’s solution, the so-called Penrose transform, is that there should
be some integral transformation mapping a twistor wavefunction f to a massless free field.
Additionally, as the field must be an eigenfunction of the helicity operator h (helicity is the
only quantum number for massless free particles), so must the twistor wavefunction.
To find the general form of f , recall from section 2.2 that a general twistor wavefunction
ends up being represented either holomorphically as a function of the twistor Z or anti-
holomorphically as a function of the dual twistor Z̄. Furthermore, from our expression for
the helicity h = (1/2)Z α Z̄α we find the symmetrized form of the helicity operator h =
(1/4){Z α , Z̄α }. Thus, choosing the holomorphic representation f (Z) in which Z̄ → −∂/∂Z,
we obtain the helicity operator in twistor space
1 ∂
h = − (Z α α + 2). (19)
2 ∂Z
The operator X a ∂a seen here is known as the homogeneity operator. An eigenfunction
F (X) of this operator with eigenvalue n, called its degree of homogeneity, satisfies F (tX) =
tn F (X). Thus an eigenfunction of the helicity operator with eigenvalue h must be homoge-
neous of degree n = −2h − 2.
We are now in a position to give the integral form of the Penrose transform for positive
helicity h. Starting with a twistor wavefunction f (Z) of homogeneity −2h − 2, we define an
auxiliary function
0
f ∗ (x, π) ≡ f (ixAA πA0 , πA0 ). (20)
Then the Penrose transform of f is given by the projective contour integral
I
1 0
ϕA01 ···A02h (x) = πA01 · · · πA02h f ∗ (x, π)πB 0 dπ B . (21)
2πi

8
This expression as written is a bit ambiguous, so some explanation is needed. First of all, the
function f (Z) clearly cannot actually be holomorphic, otherwise the integral would vanish
by Cauchy’s theorem. More precisely, f will be meromorphic, that is, it will be holomorphic
except on a set of isolated singularities. Furthermore, f must have at least two poles in
order for the integral to be nonzero, otherwise whatever contour under consideration could be
shrunk down to nothing on the Riemann sphere opposite the pole. If it has exactly two poles,
they will be separated by a contour Γ, but then the sign is ambiguous. However, note that the
overall sign does not have any effect on the field equations (18), so either choice of orientation
for Γ is acceptable. Finally, for more than two poles, the Penrose transform becomes a
“branched” contour integral over the overlaps of neighborhoods containing poles [5, pp. 155-
160]. This is stepping into the realm of sheaf cohomology and as such is beyond the scope
of the present work6 ; as such we will only be focusing on the simpler twistor wavefunctions.
Now that the transformation has been made precise, it is important to see exactly why
it works. This is rather simple: notice that
0 0 ∂f
∇AA1 f ∗ = iπ A1 , (22)
∂ωA
so taking the derivative of the field produced by the transformation, we find
I
AA01 i 0 ∂f 0
∇ ϕA01 ···A02h = π A1 πA01 πA02 · · · πA02h πB 0 dπ B (23)
2πi ∂ωA
0
which trivially vanishes because of the inner product π A1 πA01 of π with itself.
Finally, as mentioned above there is a class of simplest nontrivial twistor wavefunctions,
with exactly two poles, which can be transformed using only basic contour integrals. These
take the general form
(C · Z)c (D · Z)d
f (Z) = (24)
(A · Z)1+a (B · Z)1+b
with a, b, c, d ≥ 0 and a + b − c − d = 2h, so that f (Z) is meromorphic and homogeneous of
degree −2h − 2, as required. Massless fields generated with a twistor wavefunction of this
form are called elementary states [5, 29], and will be the focus of the present work.

2.4 Evaluating the Penrose Transform on Elementary States


In this section, we will obtain an expression for the general elementary state produced by
the twistor wavefunction (24). We begin with the results of [14], where the expression was
derived and analyzed for the case a = c = d = 0, b = 2h, i.e.
1
f (Z) = . (25)
(A · Z)(B · Z)2h+1
There it was found using basic methods of contour integration that the field in this case
takes the form
AA01 · · · AA02h
ϕA01 ···A02h (x) = , (26)
[AB]2h+1
6
Note, however, that a twistor wavefunction is defined only by its pole structure—adding to it any proper
holomorphic function results in the same physical field.

9
where A, B are spinor fields defined by
0 0 0
AA = iAA xAA + AA (27)
0 0
with AA and AA the spinor components of the dual twistor Aα = (AA , AA ), and similarly
0
for B. These “calligraphic” spinor fields are defined such that Aα Z α = AA πA0 with Z =
0
(ixAA πA0 , πA0 ). The term in the denominator of this result can be further written out
1
[AB] = hABi(x − q)2 (28)
2
where 0 0
a AA B A − AA B A
q = . (29)
ihABi
Although writing out these products like this makes things more concrete, it will be best to
stick with the calligraphic notation from now on for simplicity.
Next, we carry out the transformation for a slightly different elementary state, with
a = b = 1 and c = d = 0 (so h = 1), i.e.
1
f (Z) = . (30)
(A · Z)2 (B · Z)2

In this case the Penrose transform is given by


0
πA01 πA02 πB 0 dπ B
I
1
ϕA01 A02 (x) = . (31)
2πi [Aπ]2 [Bπ]2
0 0 0 0
Denoting ρA = A0 /A1 , and similarly for ρB , and ζ = π 1 /π 0 , this becomes

(oA01 + ζιA01 )(oA02 + ζιA02 )


I
1
ϕA1 A2 (x) =
0 0
0 0 dζ. (32)
2πi(A1 B 1 )2 (ζ + ρA )2 (ζ + ρB )2

From here it is a simple matter of applying Cauchy’s theorem, but we must take care in
evaluating the residue as the pole is now of order 2. In general, the formula for the residue
of a function F (z) at a pole z0 of order n is given by

1 dn−1
Res F (z) = lim n−1 ((z − z0 )n F (z)) . (33)
z=z0 (n − 1)! z→z0 dz

Thus, evaluating the residue at −ρA , the field becomes

∂ (oA01 + ζιA01 )(oA02 + ζιA02 )


 
1
(34)
(A10 B 10 )2 ∂ζ (ζ + ρB )2

ζ=−ρA

which, after some work, can be written, up to constant factors, as


A(A01 BA02 )
ϕA01 A02 (x) = . (35)
[AB]3

10
By a similar approach as that used here and in [14], it is straightforward to show that for

(C · Z)c (D · Z)d
f (Z) = , (36)
(A · Z)(B · Z)1+b

the resulting field is, up to constant factors,

[AC]c [AD]d
ϕA01 ···A02h (x) = A(A01 · · · AA02h ) , (37)
[AB]1+b

where we define C and D in the same way as A and B. Furthermore, we find that for the
state
1
f (Z) = (38)
(A · Z) (B · Z)1+b
1+a

with c = d = 0, the resulting field is, up to constant factors, given by


1
ϕA01 ...A02h = A(A01 · · · AA0b BA0b+1 · · · BA02h ) . (39)
[AB]2h+1

Finally, for the most general elementary state (24) the field can be written in the form
0 0 2h
1 (C 1 )c (D1 )d ∂ a (ρC + ζ)c (ρD + ζ)d Y
ϕA01 ...A02h (x) = (oA0i + ζιA0i ). (40)
a! (A10 )1+a (B 10 )1+b ∂ζ a ζ=−ρA (ρB + ζ)1+b

i=1

With a = 1 this evaluates to the explicit expression


" 0 0 0 
[CA]c [DA]d C1 D1 B1
ϕA01 ...A02h (x) = 10 c +d − (1 + b) AA01 · · · AA02h (41)
A [BA]1+b [CA] [DA] [BA]
2h
#
X
+ AA01 · · · AA0i−1 ιA0i AA0i+1 · · · AA02h , (42)
i=1

which can be shown to reproduce (39) in the case c = d = 0. With a = 2 we would obtain a
more complicated expression, and for the most general elementary state (24) the explicitly
evaluated expression should become quite complicated.

3 Twistor-Hopfions: Electromagnetic Case


Next we come to the topic of twistor-hopfions. We define a twistor-hopfion as any elementary
state which results from some choice of the dual twistors A, B, C, D such that in the spin-1
case the usual hopfions seen in the literature are reproduced, e.g. the “electromagnetic knot”
of Rañada [10, 13, 14]. We will begin by studying this hopfion.

11
3.1 The Null EM Hopfion
As mentioned in section 2.3, the vacuum Maxwell equations are contained in the spin-1 form
0
of the massless free field equation (18), given by ∇AA ϕA0 B 0 = 0. More explicitly, the field
ϕA0 B 0 corresponds to the electromagnetic field strength tensor via the relation

Fab = ϕA0 B 0 εAB + c.c. (43)

The standard electric and magnetic fields are obtained by projecting the field strength tensor
and its dual, respectively, onto spatial foliations orthogonal to a timelike 4-velocity ua ,

Eb = ua Fab , Bb = −ua ∗ Fab . (44)

We will make the standard choice choice of coordinates defined by ua = (1, 0, 0, 0).
In section 2.4 we began by giving the elementary state for a = c = d = 0, b = 2h, with
twistor wavefunction for h = 1 given by
1
f (Z) = . (45)
(A · Z)(B · Z)3

Since the product [AB] is just a scalar, the field goes as ϕA0 B 0 ∼ AA0 AB 0 , so that Fab has a
doubly degenerate principle null direction defined by the flagpole of A, i.e. the electromag-
netic field is classified as null. By making the choice
√ √
Aα = (0, 2, 0, 1), Bα = ( 2, 0, 1, 0), (46)

we in fact recover Rañada’s “electromagnetic knot” hopfion, up to a constant overall factor.


The full electric and magnetic fields themselves in this configuration are quite complicated,
but they are neatly combined in the form of a so-called Riemann-Silberstein (RS) vector,
 
(t − i − z)2 − (x − iy)2
1
F~ ≡ E~ + iB
~ = −i(t − i − z)2 − i(x − iy)2  . (47)
(−(t − i)2 + r2 )3
2(x − iy)(t − i − z)

~ for this field are given by


The energy density u and Poynting vector S
(1 + x2 + y 2 + (t + z)2 )2
u= , (48)
(1 + 2(t2 + r2 ) + (t2 − r2 )2 )3
 
2(x(t − z) + y)
~= u  2(y(t − z) − x)  .
S (49)
(1 + x + y 2 + (t − z)2 )
2
x2 + y 2 − (t − z)2 − 1
These fields are visualized for t = 0 and t = 1 in Figure 3.
Note that this field configuration has topologically non-trivial field line structure: any
two field lines belonging to either E,~ B,
~ or S ~ at any given time are topologically linked
circles. Rañada verified this by computing the magnetic helicity integral, which measures
linking (and not to be confused with the helicity of massless particles), with the result
~ ~ ~ ·B~ = 0 and so the time
R
A · Bd3 x = 1 in some units at t = 0. Since the field is null, E

12
Figure 3 Field line structure of the null EM hopfion at t = 0 and t = 1. In column (a) the electric field, (b)
the magnetic field, and (c) the Poynting vector. The field line structures for both E ~ and B~ are topologically
~
equivalent to the Hopf fibration at all times, and for S is just the Hopf fibration moving along the ẑ direction
at the speed of light.

~ · Bd
~ 3 x of the helicity is zero, thus linking is conserved [10]. This notion of
R
derivative E
linking will become important later in our generalization to gravitation. It is also worth
noting that for a null EM field, since the helicity is conserved the field lines preserve their
topological structure and therefore may be thought of as stretchable filaments which flow
along the direction of energy transport, i.e. the Poynting vector [30]. In the case of the null
EM hopfion, the Poynting vector lies tangent to the Robinson congruence, which we recall
consists entirely of null lines, and so the field lines are simply propagated along straight lines
at the speed of light.

3.2 The Non-null EM Hopfion


Now that we have obtained one twistor-hopfion, any other elementary state constructed from
the same choice of the dual twistors A and B will also be a twistor-hopfion with its own
interesting properties. As the next example, we look at another case considered in section
2.4, equation (30), reproduced here for convenience:
1
f (Z) = . (30)
(A · Z)2 (B · Z)2

Once again the factor [AB] is just a scalar, so the field goes as ϕA0 B 0 ∼ A(A0 BB 0 ) . Thus, the
EM field strength tensor has two distinct principle null directions given by the flagpoles of
A and B, and so this EM field is classified as non-null.

13
Figure 4 Field line structure of the non-null EM hopfion at t = 0 and t = 1. In column (a) the electric
field at t = 0, and at t = 1 (b) the electric field, (c) the magnetic field, and (d) the Poynting vector. The
magnetic field and Poynting vector are omitted at t = 0 since they are identically zero at that time.

The Riemann-Silberstein vector for this field configuration is given by


 
−2(xz − iy(t − i))
2
F~ =  −2(yz + ix(t − i))  . (50)
(−(t − i)2 + r2 )3
(t − i)2 + x2 + y 2 − z 2
Unfortunately this does not seem to easily relate to the RS vector for the null EM hopfion,
as will be the case for the type N gravitational hopfion. However, it is easily verified that at
t = 0 the E ~ field has the structure of the Hopf fibration, while B
~ and hence S~ are identically
zero since the RS vector is purely real at that time. These results are reflected in the field
line structures shown in Figure 4. Furthermore, for at least the E ~ field some notion of linking
appears to be conserved, not in that the linking number is conserved but in that the field
lines still lie on tori. This could be further analyzed using the helicity integral given in the
previous section, but for the purpose of the present work we will leave it at that.
The energy density and Poynting vector for this configuration are given by
t4 + 2t2 (1 + 3x2 + 3y 2 − z 2 ) + (1 + r2 )2
u=2 , (51)
(1 + 2(t2 + r2 ) + (t2 − r2 )2 )3
 
2yz + x(1 + t2 + x2 + y 2 − z 2 )
~= 8t 2xz + y(1 + t2 + x2 + y 2 − z 2 ) .
S (52)
2(1 + 2(t2 + r2 ) + (t2 − r2 )2 )3
2z(x2 + y 2 )
These expressions are somewhat messy and not particularly useful in the present context so
we will not go into further analysis.

4 Gravitational Hopfions and Gravitational Linking


In this section we present a detailed method for plotting gravitational field lines in Math-
ematica based on the technique developed recently by Nichols et al [17]. Furthermore, we
discuss the gravitational twistor-hopfions, elementary states with h = 2, and end with a
discussion of gravitational linking.

14
4.1 A Method for Plotting Gravitational Field Lines
In the gravito-electromagnetism formalism, the Weyl tensor Cabcd is decomposed into an
even-parity electric part Eab and an odd-parity magnetic part Bab .7 These are obtained,
in analogy with electromagnetism, by projecting the Weyl tensor and its dual onto spatial
foliations orthogonal to some timelike 4-velocity ua ,

Eab = γar γb s Crcsd uc ud , Bab = −γar γb s ∗ Crcsd uc ud , (53)

where the spatial projection operator γab is defined by γab = gab + ua ub and again ∗ is the
Hodge dual operator, ∗Cabcd = 21 εabrs C rscd . The gravito-electric and gravito-magnetic tensors
are then symmetric and trace-free, with spatial parts given by Eij = Ci0̂j 0̂ and Bij = ∗Ci0̂j 0̂
where 0̂ denotes the component in the timelike direction û. The gravito-electric tensor is
sometimes referred to as the tidal field, since two freely falling particles with separation vector
ξ k experience a relative acceleration ∆ai = −E ij ξ j . Similarly, the gravito-magnetic tensor
is sometimes called the frame-drag field, because a gyroscope separated from an inertial
observer by ξ k is seen to precess with angular velocity ∆Ωi = B i j ξ j .
As symmetric and trace-free tensors, the tidal and frame-drag fields are completely char-
acterized by their eigensystems. Nichols et al concluded that spacetime curvature may be
visualized as the integral curves of the eigenvector fields associated with the gravito-electric
and gravito-magnetic fields [17]. In this scheme, the tidal field is to be thought of as a field
that stretches and compresses along its field lines, called tendex lines, and the frame-drag
field is to be thought of as a field which exhibits differential precession along its field lines,
called vortex lines. The relative strength and orientation of these effects is determined by the
eigenvalue corresponding to the particular field line, as can be seen from the expressions for
the relative acceleration ∆ai and angular velocity of precession ∆Ωi given above. This data
can be visually conveyed by coloring of the field lines according to the relative magnitude
of the eigenvalue at each point, as is done in [14, 17]. We provide a prototypical example
in Figure 5, using the type N gravitational hopfions at t = 0, to be discussed in the next
section.
For gravitational fields with simple expressions, analytic expressions for the eigenvector
fields may be obtained, and thus plotting the field lines as described above is a relatively
straightforward exercise in Mathematica or any other capable program. However, for more
complicated fields it is not always possible to obtain analytic expressions, and so one must rely
on numerical results for the eigenvector field. In this case, while the numerically computed
vector fields modulo direction and magnitude may be “continuous,” the actual fields used in
plotting are not and so one typically gets useless results when attempting to plot field lines.
The task of plotting arbitrary eigenvector fields is therefore somewhat nontrivial, since most
programs do not have programming built in for this kind of functionality.
This problem is not intractable, and can be resolved as follows. In each step of a typical
method/algorithm for solving a differential equation ~y˙ = f~(t, y), given the current data ~yn
and tn , there are various intermediate increments schematically of the form
~ki = f (tn + εi (h), ~yn + ~δi (h, ~kj )) (54)
7
In the present work we consider only vacuum spacetimes, in which the Riemann tensor Rabcd is equal to
the Weyl tensor Cabcd .

15
Figure 5 A prototypical example (the type N gravitational hopfion at t = 0) of gravito-electromagnetic field
visualization, demonstrating the physical interpretation of the field lines. The gravito-electric eigenvector
field with (a) negative eigenvalue stretches objects along its field lines, and with (b) positive eigenvalue
compresses along its field lines. The gravito-magnetic eigenvector field with (c) negative eigenvalue rotates
gyroscopes about its field lines in a counter-clockwise fashion, and with (d) positive eigenvalue rotates
gyroscopes about its field lines in a clockwise fashion. We show (e) a shuttle for reference, used in the field
line pictures to demonstrate the physical effects. The color scale in each picture indicates the strength of
the associated effect, with lighter colors indicating greater strength. [Shuttle model courtesy of NASA.]

used to determine a final approximation for ~yn+1 with tn+1 = tn + h. The resolution is that
for each increment ~ki computed, we define
~k 0 = sgn(~ki · f~(tn , yn ))~ki (55)
i

and use this increment instead for all later computations involving ~ki .
In Mathematica for example, we define a modified 4th order Runge-Kutta method by
eigCRK4[]["Step"[rhs_, t_, h_, y_, yp_]] :=
Module[{k0, k1, k2, k3, ypNew},
k0 = yp;
k1 = rhs[t + h/2, y + h k0/2]; If[k1.yp < 0, k1 = -k1];
k2 = rhs[t + h/2, y + h k1/2]; If[k2.yp < 0, k2 = -k2];
k3 = rhs[t + h, y + h k2]; If[k3.yp < 0, k3 = -k3];
ypNew = Normalize[(k0 + 2 k1 + 2 k2 + k3)/6];
{h, h ypNew, ypNew}];
eigCRK4[___]["DifferenceOrder"] := 3;
eigCRK4[___]["StepMode"] := Fixed;

16
Then we may plot arbitrary eigenvector fields using NDSolve simply by passing the argument
Method -> eigCRK4. This is extremely useful, as will be seen in our plots of the gravitational
twistor-hopfions in the following sections.

4.2 The Type N Gravitational Hopfion


Once again, as mentioned in section 2.3, linearized Einstein gravity is contained in the spin-2
0
form of the massless free field equation (18), given by ∇AA ϕA0 B 0 C 0 D0 = 0. In this case the
field ϕA0 B 0 C 0 D0 corresponds to the Weyl tensor via the relation

Cabcd = ϕA0 B 0 C 0 D0 εAB εCD + c.c. (56)

The decomposition into gravito-electric and gravito-magnetic parts proceeds as in the pre-
vious section.
We begin again with the elementary state (24) with a = c = d = 0, b = 2h, except in
this case h = 2 so that the twistor wavefunction is given by
1
f (Z) = . (57)
(A · Z)(B · Z)5

Again, the product [AB] is just a scalar so the field goes as ϕA0 B 0 C 0 D0 ∼ AA0 AB 0 AC 0 AD0 .
Thus the Weyl tensor has a quadruple principle null direction defined by the flagpole of A,
making this field Petrov type N. With choice of dual twistors A and B as in section 3.1 we
obtain the type N gravitational hopfion studied in [14]. We make the standard choice of
timelike direction ua = (1, 0, 0, 0) for which the gravito-electric and gravito-magnetic field
line structures are displayed in Figures 5, 6.
In this case it will be edifying to work out the spectral structure. It turns out that
both the tidal and frame-drag fields possess the same eigenvalues labelled by {λ+ , λ0 , λ− } =
{+Λ, 0, −Λ} corresponding respectively to eigenvectors {~e+ , ~e0 , ~e− } for the tidal field and
{~b+ , ~b0 , ~b− } for the frame-drag field, where Λ is given by

(1 + x2 + y 2 + (t − z)2 )2
Λ(x) = > 0. (58)
2(1 + 2(t2 + r2 ) + (t2 − r2 )2 )5/2

As explained in the previous section, this determines the overall strength of the tidal and
frame-drag effect.
Next we will give the expressions for the eigenvector fields along with some discussion.
Before we begin however, it is important to note that the overall magnitude and sign of the
eigenvectors is completely arbitrary; the eigenvector field is determined only by direction
modulo sign. Thus we will often refer to “the eigenvector” by giving a vector field, when we
really mean that the eigenvector field simply lies tangent to the given vector field.
Consider first the eigenvector fields corresponding to the zero eigenvalues. These are
found to take the form  
2(x(t − z) + y)
~e0 = ~b0 =  2(y(t − z) − x)  , (59)
x2 + y 2 − (t − z)2 − 1

17
Figure 6 Gravito-electromagnetic field line structure of the type N gravitational hopfion at t = 0 and t = 1:
the tidal fields (a) e− , (b) e+ , and (c) e0 ; and the frame-drag fields (d) b− , (e) b+ , and (f ) b0 . The field
lines are colored by the relative magnitude of their eigenvalues at each point, with lighter colors indicating
greater magnitude.

which is, comparing with (49), equivalent to the Poynting vector S ~ of the null EM hopfion.
Next, consider the remaining eigenvectors with nonzero eigenvalues. While the individual
fields take on quite complicated forms, it turns out that we can again construct Riemann-
Silberstein structures f~e = ~e− +i~e+ and f~b = ~b− +i~b+ which greatly simplify their expressions.
We find that
f~e = eiπ/4 f~b = ei Arg θ F~ ,
p
θ(x) = −(t − i)2 + r2 (60)
where F~ is the Riemann-Silberstein vector (47) for the null EM hopfion. Thus the tidal and
frame-drag fields are directly related to the null EM hopfion, the RS vectors being related
simply by a local duality rotation. Notice that Arg θ = 0 at t = 0 so that initially the tendex
and vortex structures are just the same as the field line structure of the null EM hopfion,
with the vortex structure only differing by a global rotation by 45◦ . For t 6= 0, we see that
Arg θ 6= 0 and so the tendex and vortex structures differ from the field line structure of the
EM hopfion. The field line structures at t = 0 and t = 1 are shown in Figure 6.

18
In electromagnetism, there are two quantities invariant under local duality rotations:
the energy density and the Poynting vector. These are given for the null and non-null EM
hopfions in sections 3.1 and 3.2 respectively. In direct analogy, two covariant quantities
arise naturally in the study of gravito-electromagnetism that are invariant under duality
transformation of the tidal and frame-drag fields, namely the super-energy density U and
the super-Poynting vector P~ given by
1
U = (Eij E ij + Bij B ij ), Pi = εijk E jl B kl . (61)
2
Continuing the analogy with electromagnetism, these quantities obey their own conservation
equation [31], although it is somewhat more complicated and beyond the scope of the present
work so we will not go into the details. In the case of the type N twistor-hopfion, we find
that the super-energy density and super-Poynting vector are given by

(1 + x2 + y 2 + (t − z)2 )4
U= , (62)
2(1 + 2(t2 + r2 ) + (t2 − r2 )2 )5
 
2(x(t − z) + y)
U
P~ =  2(y(t − z) − x) . (63)
(1 + x + y 2 + (t − z)2 )
2
2 2 2
−1 + x + y − (t − z)
Comparing with the energy density (48) and Poynting vector (49) of the null EM hopfion, we
see that the two sets of local duality invariants bear a striking resemblance to one another.
The primary difference is in the larger power in the denominator of each quantity in the
gravitational case, indicating that the energy falls off more rapidly for higher spin fields.

4.3 The Type III Gravitational Hopfion


As our final example, we consider the gravitational analog of the non-null EM hopfion of
section 3.2, the elementary state c = d = 0, a = 1, b = 2h − 1 = 3 with twistor wavefunction
1
f (Z) = . (64)
(A · Z)2 (B · Z)4

In this case the field goes as ϕA0 B 0 C 0 D0 ∼ A(A0 AB 0 AC 0 BD0 ) , so that the Weyl tensor has one
triple and one single principle null direction, given by the flagpoles of A and B respectively.
Thus, this is a gravitational field of Petrov type III.
Unfortunately the eigenvalues and eigenvectors for this twistor-hopfion have proven dif-
ficult to obtain analytically, and so we can not give them explicitly for all times. However,
we have verified that the eigenvalue structure differs slightly from that of the type N fields,
taking the form {λ+ , λ0 , λ− } = {Λ − λ, λ, −Λ} where λ = 0 at t = 0 and λ+ ≥ λ0 ≥ λ− for
all points in spacetime. Thus, we will continue to label the eigenvectors by {−, 0, +} based
on their relative ordering. The tendex and vortex structures have been plotted numerically
in Figure 7, using the methods of section 4.1. At t = 0 the fields bear the field line structure
of the Hopf fibration, as can be seen from a quick glance at the figure.

19
Figure 7 Gravito-electromagnetic field line structure of the type III gravitational hopfion at t = 0 and t = 1:
the tidal fields (a) e− , (b) e+ , and (c) e0 ; and the frame-drag fields (d) b− , (e) b+ , and (f ) b0 . The field
lines are colored by the relative magnitude of their eigenvalues at each point, with lighter colors indicating
greater magnitude.

The local duality invariants, on the other hand, can be easily computed for this field.
The super-energy density and super-Poynting vector for this configuration are given by

(1 + x2 + y 2 + (t − z)2 )2 (t4 + 2t2 (1 + 7x2 + 7y 2 − z 2 ) + (1 + r2 )2 )


U= , (65)
8(1 + 2(t2 + r2 ) + (t2 − r2 )2 )5

 
2(x(t − z) + y)
U
P~ =  2(y(t − z) − x)  (66)
2(1 + x2 + y 2 + (t − z)2 ) 2 2 2
−1 + x + y − (t − z)
 3 
x + 2yz + x(1 + t2 + y 2 − z 2 )
4tU y 3 − 2xz + y(1 + t2 + x2 − z 2 ) .
+ 4
t + 2t (1 + 7x + 7y 2 − z 2 ) + (1 + r2 )2
2 2
2z(x2 + y 2 )

20
Comparing with the energy density (51) and Poynting vector (52) of the non-null EM hopfion,
we once again see that the two sets of local duality invariants are curiously similar. In fact,
the two vector terms seen here in the super-Poynting vector are exactly the same as the
vector terms in the Poynting vectors (49) and (52) of the null and non-null EM hopfions
respectively, even up to the prefactor of t in the latter term.

4.4 Gravitational Helicity and Linking


Finally, we come to the topic of gravitational helicity or linking. The notion of helicity here
is not to be confused with the helicity of massless particles referred to in our discussions
of the twistor formalism and elementary states, which is an entirely different concept that
just happens to have the same name. The helicity of a vector field is a quantity which,
approximately speaking, sums the linking of every pair of field lines within a given volume.
Thus if a volume V is filled with curves tangent to a unit vector field where any two curves are
linked once, the helicity in this volume would be V . While this definition is not completely
clear, it can be made precise [32].
First consider two closed curves 1 and 2 parameterized by ~x(σ) and ~y (τ ) respectively.
The Gauss linking number L12 counts the (oriented) linking of the two curves and is given
by I I  
1 ~r d~x d~y
L12 = · × dσdτ (67)
4π 1 2 r3 dσ dτ
~ which can be partitioned into N flux tubes
where ~r = ~x − ~y . Now consider a vector field B
each carrying total flux Φi . Then the helicity K is defined by
N X
X N
K= Lij Φi Φj . (68)
i=1 j=1

If we let N → ∞ this becomes the general integral expression for the helicity,
ZZ
1 ~r h ~ ~
i
K= · B(x) × B(y) d3 xd3 y. (69)
4π r3

In the case of electromagnetism, with B ~ being the magnetic field (hence the notation), this
expression is further simplified using the integral for the vector potential in the Coulomb
gauge, Z
~ 1 ~r ~ 3
A(x) = × B(y)d y. (70)
4π r3
This results in the common definition of the magnetic helicity used in plasma physics and
elsewhere, Z
K= A ~ · Bd
~ 3 x, (71)

which is easily seen to be gauge invariant.


We now return to the topic of twistor-hopfions and gravitation. As mentioned in section
3.1, Rañada computed the magnetic (and electric) helicity of the null EM hopfion to be 1 in

21
some units, at any time, supporting the fact that its field lines remain topologically linked
circles for all times. Considering that each of the twistor-hopfions studied above exhibits
field line structure matching the null EM hopfion at t = 0, it is natural to ask whether the
same is true for the other twistor-hopfions. More specifically, is there an integral describing
the helicity of the tendex and vortex lines of the tidal and frame-drag fields?
To answer this question, we propose a definition for the helicity of a tendex or vortex
structure. Consider an eigenvalue λ(x) with corresponding choice of smooth, oriented unit
eigenvector field v̂(x) for a tidal field Eab or frame-drag field Bab . We define the helicity
associated with this eigensystem as the integral
Z Z
1 ~r 3 3

K= · [λ(x)v̂(x) × λ(y)v̂(y)] d xd y (72)
4π r3
Z Z
1 ~r 3 3

= · [v̂(x) × v̂(y)] λ(x)d xλ(y)d y . (73)
4π r3

In other words, we associate the eigensystem with the vector field defined by λ(x)v̂(x), and
take the absolute value to reflect the arbitrary choice of orientation of v̂. This definition
could in fact suffice as a definition for the helicity of an eigenvector field in general. As
it applies to the gravitational twistor-hopfions of sections 4.2 and 4.3, we are specifically
interested in the eigensystems e± , b± (the integral vanishes for e0 , b0 , reflecting the fact that
these fields do not contribute to the tidal or frame-drag effects).
Alas, while this might be a nice definition in theory, it may be completely impractical
for actual computations. Certainly we would expect the helicity for e± , b± for the type N
gravitational hopfion to be 1 in some units, at t = 0, since the same is true of the null
EM hopfion. Whether this remains true for all times is unclear; the fact that the field
lines appear to lie on nested tori (see Figure 6) seems to indicate that the answer might
be ‘yes.’ Furthermore, it is straightforward to see that local Lorentz transformations induce
local duality rotations (consider classical electromagnetism, for example), which supports the
claim that linking is conserved if it is to be thought of as a physically real, Lorentz-covariant
quality.
In future work, it is hoped that this definition could be simplified, modified, or replaced
using more natural objects like the tidal or frame-drag fields themselves, or any of the various
“potentials” for the gravito-electromagnetic fields (the 4-acceleration, the shear, the vortic-
ity, the Lanczos potential, etc. [31]) in analogy with the electric/magnetic helicity. One
possibility is that instead of considering as fundamental the linking of the field lines them-
selves, we consider instead the linking/nesting of the surfaces they lie on. This alternative
definition of gravito-electromagnetic helicity may then avoid direct reference to eigensystems.
For the type N hopfion,
R for example, it can be verified that Eij B ij = 0 at all times, so that
a possible formula Eij B ij d3 x for the time derivative of the ‘helicity,’ in analogy with the
electric/magnetic helicity discussed in section 3.1, would vanish, thus supporting the claim
that the field lines lie on linked/nested tori at all times (see Figure 6). Another possibility
is that the correct approach to gravitational helicity lies in Chern-Simons theory. For now,
however, the original definition of the gravitational helicity given in this section will suffice.

22
5 Conclusion
Arkani-Hamed has said that, “Due to the effects of both gravity and quantum mechan-
ics, spacetime is necessarily an approximate notion that must emerge from more primitive
building blocks” [33], and according to Penrose one of those building blocks is the twistor
formalism [4]. In light of such outlooks, it is not surprising that we may extract a great
deal of structure from highly complicated electromagnetic, gravitational, and more general
fields. Here we derived expressions for a few classes of the elementary states of the Pen-
rose transform for solutions to the massless free field equation. Furthermore, we analyzed
the null electromagnetic hopfion in the context of elementary states and extended this field
configuration to more general twistor-hopfions, producing topologically nontrivial electro-
magnetic and gravitational fields. Finally, we proposed a new notion of gravitational linking
and discussed its significance in the context of the gravitational twistor-hopfions.

6 Acknowledgements
The author would like to express his deepest gratitude to Dirk Bouwmeester, Joe Swearngin,
Amy Thompson, and Jan Willem Dalhuisen for a great deal of support, guidance, and useful
discussions. This work is supported in part by the 2011 UCSB Worster Fellowship and the
2012 College of Creative Studies SURF grant. New material presented here is to be published
in a future work [16].

7 References
[1] P. A. M. Dirac, Proceedings of the Royal Society of London. Series A-Mathematical
and Physical Sciences 155, 447 (1936).

[2] M. Fierz and W. Pauli, Proceedings of the Royal Society of London. Series A. Mathe-
matical and Physical Sciences 173, 211 (1939).

[3] V. Bargmann and E. P. Wigner, Proceedings of the National Academy of Sciences of


the United States of America 34, 211 (1948).

[4] R. Penrose, Gravitation and geometry , 341 (1987).

[5] R. Penrose and W. Rindler, Spinors and Space-Time: Volume 2, Spinor and Twistor
Methods in Space-Time GeometryCambridge monographs on mathematical physics
(Cambridge University Press, Cambridge, 1988).

[6] R. Penrose, Proceedings of the Royal Society of London. Series A. Mathematical and
physical sciences 284, 159 (1965).

[7] G. E. Volovik and V. P. Mineev, Sov. Phys. JETP 46, 401 (1977).

[8] A. M. Kamchatnov, Sov. Phys. JETP 82, 117 (1982).

23
[9] A. F. Rañada, letters in mathematical physics 18, 97 (1989).

[10] A. F. Rañada and J. L. Trueba, Modern Nonlinear Optics, Part III 119, 197 (2002).

[11] L. Faddeev and A. J. Niemi, Nature 387, 58 (1997).

[12] W. T. M. Irvine and D. Bouwmeester, Nature Physics 4, 716 (2008).

[13] J. W. Dalhuisen and D. Bouwmeester, Journal of Physics A: Mathematical and Theo-


retical 45, 135201 (2012).

[14] J. Swearngin, A. Thompson, A. Wickes, J. W. Dalhuisen, and D. Bouwmeester, (2013),


arxiv:1302.1431.

[15] A. Thompson, J. Swearngin, A. Wickes, J. W. Dalhuisen, and D. Bouwmeester, Bulletin


of the American Physical Society 58 (2013).

[16] A. Thompson, A. Wickes, J. Swearngin, and D. Bouwmeester, (work in progress).

[17] D. A. Nichols et al., Phys. Rev. D 84, 124014 (2011).

[18] E. T. Newman et al., Journal of mathematical physics 6, 918 (1965).

[19] R. Emparan and H. S. Reall, Physical Review Letters 88, 101101 (2002).

[20] E. Witten, Communications in Mathematical Physics 252, 189 (2004).

[21] F. Cachazo, P. Svrcek, and E. Witten, Journal of High Energy Physics 2004, 006
(2004).

[22] I. Bena, Z. Bern, and D. A. Kosower, arXiv preprint hep-th/0406133 (2004).

[23] N. Arkani-Hamed, J. Bourjaily, F. Cachazo, A. Hodges, and J. Trnka, Journal of High


Energy Physics 2012, 1 (2012).

[24] N. Arkani-Hamed, F. Cachazo, C. Cheung, and J. Kaplan, Journal of High Energy


Physics 2010, 1 (2010).

[25] C. W. Misner, K. S. Thorne, and J. A. Wheeler, Gravitation (WH Freeman, 1973).

[26] F. G. Staff, The strangest man: the hidden life of Paul Dirac, Mystic of the Atom
(Faber & Faber, Limited, 2009).

[27] R. Penrose and M. MacCallum, Physics Reports 6, 241 (1972).

[28] A. Zee, Quantum field theory in a nutshell (Princeton University Press, 2010).

[29] R. S. Ward and R. O. Wells Jr, Twistor geometry and field theory (Cambridge University
Press, 1991).

[30] W. T. Irvine, Journal of Physics A: Mathematical and Theoretical 43, 385203 (2010).

24
[31] R. Maartens and B. A. Bassett, Classical and Quantum Gravity 15, 705 (1998).

[32] M. A. Berger, Plasma Physics and Controlled Fusion 41, B167 (1999).

[33] N. Arkani-Hamed, Daedalus 141, 53 (2012).

25

You might also like