Linked and Knotted Gravitational Radiation

Amy Thompson1,2 , Joe Swearngin 2,3

, and Dirk Bouwmeester1,2
1
Dept. of Physics, University of California, Santa Barbara, CA 93106
2
Huygens Laboratory, Leiden University, PO Box 9504, 2300 RA Leiden, The Netherlands
3
Dept. of Physics, University of California, Los Angeles, CA 90095
arXiv:1402.3806v2 [gr-qc] 15 Aug 2014

Abstract
We show that the torus knot topology is inherent in electromagnetic and gravitational radi-
ation by constructing spin-N fields based on this topology from the elementary states of twistor
theory. The twistor functions corresponding to the elementary states admit a parameteriza-
tion in terms of the poloidal and toroidal winding numbers of the torus knots, allowing one to
choose the degree of linking or knotting of the associated field configuration. Using the gravito-
electromagnetic formalism, we show that the torus knot structure is exhibited in the tendex and
vortex lines for the analogous linearized gravitational solutions. We describe the topology of the
gravitational fields and its physical interpretation in terms of the tidal and frame drag forces of
the gravitational field.

1 Introduction
Knots and links are quite remarkable given that they are as old and ubiquitous as ropes and thread
and yet have only relatively recently seen a rigorous formulation within mathematics. The study of
knots and links has enjoyed a close relationship with physics since its inception by Gauss [1]. Today
the application of these topological structures in theoretical physics is more widespread than it has
ever been, from fault resistant quantum computing [2], hadron models [3, 4], topological MHD and
fluid mechanics [5,6], classical field theories [79], quantum field theory [10,11], DNA topology [12],
to nematic liquid crystals [13] just to name a few. In this article we shall focus on the application
of an important class of knots, torus knots, to classical electromagnetic and gravitational radiation.
A hopfion is a field configuration based on a topology derived from the Hopf fibration. The
electromagnetic hopfion (EM hopfion) is a null solution to the source free Maxwell equations such
that any two field lines associated to either the electric, magnetic, or Poynting vector fields (EBS
fields) are closed and linked exactly once [7]. When an EM hopfion is decomposed onto hyperplanes
of constant time there always exists a hyperplane wherein the EBS fields are tangent to the fibers
of three orthogonal Hopf fibrations. If one extends the Poynting vector to be a future pointing
light-like 4-vector its integral curves comprise a space filling shear-free null geodesic congruence
dubbed the Robinson congruence by Roger Penrose.
Ranada [14] rediscovered the EM hopfion solution and noted that its topology was invariant
under time evolution. The search for generalizations of the hopfion solution led to the introduction
of a set of non-null EM solutions based on torus knots [15], but the topology was not preserved

Both authors contributed equally to this work.

email: amy@physics.ucsb.edu

1
during time evolution. Around the same time, the Kerr-Robinson theorem was used to derive the
hopfion from the Robinson congruence itself using 2-spinor methods [16]. Inspired by the role of
the Robinson congruence in the evolution of the hopfion, the conservation of field line topology was
tied to the shear-free property of the Robinson congruence [17].

2 Complex Analytic and Twistor Methods

The fundamental role of the Robinson congruence and its connection to the topology of physical
systems leads one to consider its relationship to twistor theory. It has long been known that
complex contour integral transforms can be used to find the solution to real PDEs. In 1903,
Whittaker used this technique to construct the general solution to Laplaces equation [18]. In 1915,
Bateman extended this method to give solutions to the vacuum Maxwell equations [19]. Twistor
theory was developed by Roger Penrose in the late 1960s as an extension of the sl(2, C) spinor
algebra. From this perspective, the complex analytic structure of Bateman can be related to the
geometry of spinor fields on space-time [20], which encode linear and angular momentum and are
represented by two SL(2, C) spinors A0 and A . The linear momentum is the flagpole of A0 so
0
that pa = A A and the angular momentum bivector is related to A by
0 0
M ab = i (A B) A B + c.c.

These spinors are combined into a single object Z = A , A0 called a twistor. In this formalism,

massless linear relativistic fields are expressed in the form of symmetric spinor fields. In 1969, the
general solution to the massless spin-N field equations was given as a complex contour integral
transform now called the Penrose transform [21].
Within the twistor framework the solutions of the massless spin-N equations are represented
by the Penrose transform of homogeneous twistor functions (see Appendix). The elementary states
are a canonical example of such functions whose singularities define Robinson congruences on
Minkowski space M. The space-time fields corresponding to the elementary states are finite-energy,
and in the null case are everywhere non-singular [22]. For integer spin fields, the expansion of a
solution over the elementary states in twistor space T is related to the expansion over spherical
harmonics in M through the Penrose transform [23]. These properties have made the elementary
states the topic of many studies [2426], and for many problems it is assumed that considering the
elementary states is sufficient to describe any solution [27].
While investigating these twistor functions and their connection to field topology in M, we have
previously shown that the EM hopfion and the analogous gravitational hopfion are elementary
states of twistor theory [9]. Using the earlier construction for EM fields by Bateman [19], Kedia, et
al. have shown that the EM hopfion is the simplest case in a set of null EM fields based on torus
knots [28]. Here we show that field configurations based on all the torus knots are contained within
the elementary states of twistor theory. The Hopf fibration appears as the degenerate case whereby
the linked and knotted toroidal structure degenerates down to the linked hopfion configuration.
This generalization leads to a construction for spin-N fields based on torus knots. We will focus
our analysis on the spin-1 and spin-2 fields, where the topology is physically manifest in the field
lines.1
The concept of tendex and vortex lines, gravitational lines of force for a particular observer, was
developed by Nichols, et al. [29], who were motivated by the desire to understand the non-linear
dynamics of curved space-time in a more intuitive, directly physical way than previous approaches.
1
For the Weyl fields, the linked and knotted topology appears in the current.

2
The physical understanding of the electromagnetic field is based upon the decomposition of the
Faraday field strength tensor onto hyperplanes of constant time yielding two spacial vector fields
interpreted as the electric and magnetic fields. Analogously, the Weyl curvature tensor admits
a decomposition onto constant time hyperplanes yielding two spacial tensors called the gravito-
electric (GE) and gravito-magnetic (GM) tensors. The integral curves of the eigenvector fields of
these tensors are called tendex and vortex lines respectively and represent the gravitational analog
of electromagnetic field lines. This method was elucidated through a series of papers where it was
applied to I) weak field solutions [29], II) stationary black holes [30], and III) weak perturbations
of stationary black holes [31]. This method of GEM decomposition is well-suited to studying linked
and knotted fields because, as we have shown previously [9], the field topology is manifest in the
lines of force for both the electromagnetic and analogous gravitational solutions.

3 Parameterization of the Elementary States

We will now relate the torus knot topology to the twistor elementary states to obtain solutions to
the EM and gravitational spinor field equations.
Torus knots are closed curves on the surface of a torus which wind an integer number of times
about the toroidal direction nt and poloidal direction np as in Fig. 1, where nt and np are coprime
and both greater than one. If nt and np are not coprime, then there are ng = gcd(nt , np ) linked
curves, each corresponding to a (nt , np ) mod ng torus knot. If either nt or np is equal to one, then
the knot is trivial with ng linked curves.

Figure 1 Torus knots (green) wind (nt , np ) times around a torus (purple) in the toroidal and poloidal directions,
respectively. Shown here are the cases of a) trefoil (2,3) knot, b) cinquefoil (2,5) knot, c) septafoil (2,7) knot, and d)
nonafoil (2,9) knot.

Following the twistor program, we represent solutions to the massless spin-N equations in M as
the Penrose transform of functions f (Z) in twistor space [32]. The details of the Penrose transform
calculation for these fields are given in the Appendix.
Consider the twistor functions corresponding to the elementary states [24]

(C Z )c (D Z )d
f (Z) = (1)
(A Z )a (B Z )b

where (A Z ) is the SU (2, 2) twistor inner product. Choosing a = 1 yields null/Type N solutions
and we must have b = 2h + 1 + c + d to give the correct homogeneity h = 2h 2 for a solution

3
with helicity h. We will show that the class of generating functions of the form2

(C Z )h(np 1) (D Z )h(nt 1)
f (Z) = , (2)
(A Z )(B Z )h(np +nt )+1

lead to field configurations with a torus knot topology where np and nt correspond to the poloidal
and toroidal winding numbers.
We choose the dual twistors

A = (0, 2, 0, 1)

B = ( 2, 0, 1, 0)

C = (0, 2, 0, 1)

D = ( 2, 0, 1, 0). (3)

A and C correspond to Robinson congruences with opposite twist, both with central axes aligned
along the +z-direction. B and D correspond to Robinson congruences with opposite twist, but
in the z-direction. This choice leads to spin-N fields which propagate +z-direction with field line
configurations that are based on a torus knot structure.

4 Electromagnetic Torus Knots

After applying the spin-1 Penrose transform to Eqn. (2), the resulting spinor field is
0 0
(AC 0 C C )np 1 (AD0 DD )nt 1
A0 B 0 (x) = AA 0 AB 0 (4)
(AE 0 B E 0 )np +nt +1

Note that the Latin script spinor variables are the spinors associated to the Latin twistor variable.
0
Ergo, AA0 is defined implicitly by A Z = AA0 A (see Appendix). The solution in Eqn. (4)
satisfies the source-free spinor field equation by construction and yields the field strength spinor
0
AA A0 B 0 = 0,
FA0 B 0 AB = A0 B 0 AB + c.c.

The spin-1 fields are null torus knots with a Poynting vector that is everywhere tangent to
a Hopf fibration and propagates in the z-direction without deformation. The solutions have the
same topology3 as the electromagnetic fields in Ref. [28]. The electric and magnetic vector fields
each have the following topological structure as shown in Fig. 2. There are 2ng core field lines,
where ng = gcd(nt , np ), which are linked (and knotted if nt , np > 1). Each core line has the same
configuration as the corresponding torus knot with (nt , np ) shown in Fig. 1. With the choice for C
and D given in Eqn. (3), the poloidal and toroidal winding numbers for the EM case are related
to the exponents in Eqn. (1) by c = np 1 and d = nt 1. A single core field line is surrounded
by nested, toroidal surfaces, each filled by one field line. A second core field line, also surrounded
by nested surfaces, is linked with the first so that there are 2ng sets of linked nested surfaces which
fill all of space. A complete solution to Maxwells equations consists of an electric and a magnetic
2
We use the conventions given in Eqn. (18) of the Appendix.
3
There is an overall constant factor of 4nt np in Ref. [28] that does not appear in our construction, but it does not
affect the topology.

4
field orthogonal to each other, both with this field line structure. The (1,1) case corresponds to the
electromagnetic hopfion.
At t = 0 the electric and magnetic fields are tangent to orthogonal torus knots, as shown in
Fig. 3 (first row). The fields will deform under time evolution, but the topology will be conserved
since E~ B
~ = 0 [17, 33].

Figure 2 The field line structure based on (2,3) trefoil knot. a) The core field line is a torus knot (green). b) Each
field line except the core lies on the surface of a nested, deformed torus. c) One field line fills a complete surface
(red). d) Another field line fills a second surface (blue) linked with the first. The two linked core field lines and the
nested surfaces around them fill all of space.

5 Gravito-electromagnetic Torus Knots

The spin-2 solutions will be analyzed in terms of the gravito-electromagnetic tidal tensors. The
Weyl tensor Cabcd can be decomposed into an even-parity electric part Eij corresponding to the
tidal field and an odd-parity magnetic part Bij for the frame-drag field, in direct analogy with
the decomposition of the electromagnetic field strength tensor into an electric field and a magnetic
field. For an observer at rest, this gives
Eij = Ci0j0 (5)
Bij = Ci0j0 . (6)
These tensors are symmetric and traceless, and are thus characterized entirely by their eigenvalues
and eigenvectors. One may then study the eigensystem associated with these matrix-valued fields
instead of the Weyl tensor itself, allowing for a more intuitive approach to understanding the
gravitational field. The integral curves of the eigenvectors of the tidal tensor are called tendex
lines and their eigenvalues define the tidal acceleration along these lines. The integral curves of
the eigenvectors of the frame-drag tensor are called vortex lines and their eigenvalues define the
gyroscope precession about the vortex lines. Together, the tendex and vortex lines are the analog
of electromagnetic field lines. [29]
After applying the spin-2 Penrose transform to Eqn. (2), the resulting spinor field is
0 0
(AF 0 C F )2(np 1) (AG0 DG )2(nt 1)
A0 B 0 C 0 D0 (x) = AA0 AB 0 AC 0 AD0 . (7)
(AE 0 B E 0 )2(np +nt )+1
The source-free field equation and Weyl field strength spinor are
0
AA A0 B 0 C 0 D0 = 0,
CA0 B 0 C 0 D0 ABCD = A0 B 0 C 0 D0 AB CD + c.c.

5
 Figure 3 A comparison of the spin-1 (EM) and spin-2 (gravity) trefoil knots at t = 0. The first row is the EM trefoil
knot: a the electric field, b the magnetic field, and c the Poynting vector field. The second row is the gravito-electric
trefoil knot: d the negative eigenvalue field ~e , e the positive eigenvalue field ~e+ , and f the zero eigenvalue field
~e0 .The third row is the gravito-magnetic trefoil knot: g the negative eigenvalue field ~b , h the positive eigenvalue
field ~b+ , and i the zero eigenvalue field ~b0 . The color scale indicates magnitude of the eigenvalue, with lighter colors
indicating a higher magnitude.

The Weyl tensor can then be decomposed into the GEM components. For Type N, the eigen-
values for both the GE and GM tensors take the form {, 0, +}, with (x) 0 +(x) for
all points x in space-time The magnitude of the eigenvalues is
22np 3 (1 + r2 + t2 2tz)2 (r2 z 2 )np 1 (r4 2r2 (1 + t2 ) + (1 + t2 )2 + 4z 2 )nt 1
= 5 (8)
(r4 2r2 (1 + t2 ) + (1 + t2 )2 ) 2 +nt +np
We label the eigenvectors {~e , ~e0 , ~e+ } and {~b , ~b0 , ~b+ } corresponding to the eigenvalues for the
tidal and frame-drag fields respectively. For the zero eigenvalue, the eigenvectors ~e0 and ~b0 are both
aligned with the Poynting vector of the null EM torus knots. For the remaining eigenvectors, we
can construct Riemann-Silberstein (RS) vectors f~e = ~e + i~e+ and f~b = ~b + i~b+ which are related
to each other by
f~e = ei/4 f~b . (9)

6
At t = 0, the eigenvectors of the GE fields have precisely the same structure as the EM fields, and
the GM eigenvector fields have the same structure but rotated by 45 . For the spin-2 case, the
poloidal and toroidal winding numbers are related to the exponents in Eqn. (1) by c = 2(np 1)
and d = 2(nt 1). The surfaces of the ~e eigenvector, color-scaled according to the magnitude of
the eigenvalue, for different values of (nt , np ) are shown in Fig. 4. The other GEM fields can be
constructed by rotating ~e according to Eqn. (9): ~e+ is found by rotating ~e by 90 about the
Poynting vector. ~b and ~b+ are found by rotating ~e and ~e+ by 45 , respectively. The eigenvalues
of the GEM fields for a given (nt , np ) have the same magnitude (color-scaling) given by |(x)| in
Eqn. (8).

Figure 4 The eigenvector field ~e for the gravitational field based on the a) trefoil (2,3) knot, b) cinquefoil (2,5)
knot, c) septafoil (2,7) knot, and d) nonafoil (2,9) knot. The color scaling is the same as in Fig. 3.

6 Conclusion
Here we have shown that the null EM torus knot solutions correspond to a class of elementary
states characterized the poloidal and toroidal winding numbers of the associated field configuration.
Using the relationship between fields of different spin in the twistor formalism, we constructed the
analogous gravitational radiation configuration that possesses tendex and vortex lines based on a
torus knot structure. Since the topology is manifest in the tendex and vortex lines, the gravito-
electromagnetic tidal tensor decomposition is a straightforward method for characterizing these
field configurations.
The elementary states were known as early as the 1970s [24], however the explicit forms of their
associated spinor and tensor representations on M were never published.4 The modern rediscovery
of these solutions has raised interest in obtaining a more complete physical understanding of the
topological properties of these fields. The parameterization of the twistor functions corresponding
to the elementary states in terms of the poloidal and toroidal winding indicates that the torus knot
structure is indeed inherent in the elementary states. For both electromagnetism and gravity, the
topology appears in the configuration of the lines of force.
4
from private discussions with Roger Penrose

7
Acknowledgments
The authors would like to thank J.W. Dalhuisen for discussions and A. Wickes for help with the
figures. This work is supported by NWO VICI 680-47-604 and NSF Award PHY-1206118.

Appendix: The Penrose Transform for Spin-N Torus Knots

To obtain a spinor field A01 A02h (x) with helicity h which satisfies the spin-N massless field equation
0
AA1 A01 A02h (x) = 0

we will calculate the Penrose transform

I
1 0
A01 A02h (x) = A01 A02h f (Z)B 0 d B (10)
2i

where is a contour on the Celestial sphere of x that separates the poles of f (Z). Consider the
twistor function given by Eqn. (2)

(C Z )h(np 1) (D Z )h(nt 1)
f (Z) = .
(A Z )(B Z )h(np +nt )+1

(For further review of the background material on twistors and the Penrose transform see Ref. [9]).
0
Let A = (A , A ) be a dual twistor such that
0 0
A Z = iA xAA A0 + A A0
0
AA A0 (11)
0 0
where Z = (ixAA A0 , A0 ). Similar relations hold for the other dual twistors B Z B B B 0 ,
0 0
C Z C C C 0 , and D Z DD D0 . We want to write the Penrose transform as an integral over
the CP1 coordinate = 10 /00 , so we have
0 0 0
C 0 d C = C 0 dD0 D C
= 00 d10 10 d00
10
= (00 )2 d( ). (12)
00

Adopting the canonical spin bases {oA0 , A0 } we have that

A0 = 00 oA0 + 10 A0
10
= 00 (oA0 + ( )A0 ). (13)
00
Observing that
1 A0 0 0 10
A A 0 = A0 + A1 ( ) (14)
00 00

8
and similarly for B, C, and D, we see that the Penrose transform becomes an integral manifestly
over CP1 . Thus
I
1 0
A01 A02h (x) = f (Z)A01 A02h B 0 d B (15)
2i
I (C 00 + C 10 ( 10 ))h(np 1) (D00 + D10 ( 10 ))h(nt 1)
1 00 00 10 10 10
= 0 0 1 0 10 00 10 10 h(n +n )+1
(oA1 + ( )A01 ) (oA2h + ( )A02h )d( )
2i (A + A ( 0 ))(B + B ( 0 )) p t 00 00 00
0 0
0 0
(C 1 )h(np 1) (D1 )h(nt 1) ( + )h(np 1) ( + )h(nt 1)
I
= 0 0 (oA1 + A01 ) (oA2h + A02h )d
2iA1 (B 1 )h(np +nt )+1 ( + )( + )h(np +nt )+1
0 0 0 0 0 0 0 0
where = A0 /A1 , = B 0 /B 1 , = C 0 /C 1 , and = D0 /D1 .
The above change of variables leaves the contour integral in a form that is straightforward to
calculate. The contour is taken to enclose the pole giving the result
0 0
(C 1 )h(np 1) (D1 )h(nt 1) ( + )h(np 1) ( + )h(nt 1)
A01 A02h (x) = 0 0 Res (oA1 + A01 ) (oA2h + A02h )
A1 (B 1 )h(np +nt )+1 = ( + )( + )h(np +nt )+1
0 0
(C 1 )h(np 1) (D1 )h(nt 1) ( )h(np 1) ( )h(nt 1)
= (oA01 A01 ) (oA02h A02h )
A10 (B 10 )h(np +nt )+1 ( )h(np +nt )+1
0 0 0 0 0 0 0 0
(A1 C 0 A0 C 1 )h(np 1) (A1 D0 A0 D1 )h(nt 1) 10 0 0 0
= 10 00 00 10 h(n +n )+1 (A oA01 A0 A01 ) (A1 oA02h A0 A02h )
(A B A B ) p t

0 0 0 0
(C 0 D0 AC C D )h(np 1) (E 0 F 0 AE DF )h(nt 1)
= AA01 AA02h
(A0 B 0 AA0 B B 0 )h(np +nt )+1
0 0
(AC 0 C C )h(np 1) (AD0 DD )h(nt 1)
= AA01 AA02h . (16)
(AB 0 B B 0 )h(np +nt )+1
In the case of spin-1 and spin-2, the classical field strength spinors are given by

FA01 A02 A1 A2 = A01 A02 A1 A2 + A1 A2 A01 A02

CA01 A04 A1 A4 = A01 A04 A1 A2 A3 A4 + A1 A4 A01 A02 A03 A04 . (17)

Choosing the standard basis to be the extended Pauli matrices we define the following symbols,
referred to as the Infeld-van der Waerden symbols

       
0
1
1 0 0
1
0 1 0
1
0 i 0
1
1 0
0AA 1AA 2AA 3AA . (18)
2 0 1 2 1 0 2 i 0 2 0 1

The spinor fields are then related to the world tensor description by
A1 A01 A2 A02
Fab = FA01 A02 A1 A2 a b
A1 A01 A A04
Cabcd = CA01 A04 A1 A4 a d 4 . (19)

References
[1] J. H. Przytycki. Knots: From combinatorics of knot diagrams to combinatorial topology based
on knots. 2007, arXiv:math/0703096.

9
 [2] C. Nayak, S. H. Simon, A. Stern, M. Freedman, and S. Das Sarma. Non-abelian anyons and
topological quantum computation. Reviews of Modern Physics, 80(3):1083-1159, 2008.

[3] T. H. R. Skyrme. A unified field theory of mesons and baryons. Nuclear Physics, 31:556569,
1962.

[4] L. Faddeev and Antti J. Niemi. Stable knot-like structures in classical field theory. Nature,
387:5861, 1997.

[5] A. M. Kamchatnov. Topological solitons in magnetohydrodynamics. Sov. Phys. JETP, 82:117

124, 1982.

[6] Amy Thompson, Joe Swearngin, Alexander Wickes, and Dirk Bouwmeester. Constructing a
class of topological solitons in magnetohydrodynamics. Physical Review E, 89(4):04310415,
2014.

[7] Antonio F. Ranada and Jose L. Trueba. Topological electromagnetism with hidden nonlinear-
ity. Modern Nonlinear Optics, Part III, 119:197253, 2002.

[8] William T. M. Irvine and Dirk Bouwmeester. Linked and knotted beams of light. Nature
Physics, 4:716720, 2008.

[9] Joe Swearngin, Amy Thompson, Alexander Wickes, Jan Willem Dalhuisen, and Dirk
Bouwmeester. Gravitational hopfions. 2014, arXiv:gr-qc/1302.1431.

[10] Edward Witten. Quantum field theory and the Jones polynomial. Communications in Math-
ematical Physics, 121(3):351399, 1989.

[11] G. D. Robertson. Torus knots are rigid string instantons. Communications in Mathematical
Physics, 226(3-4):244250, 1989.

[12] Javier Arsuaga, Mariel Vazquez, Paul McGuirk, Sonia Trigueros, De Witt Sumners, and
Joaquim Roca. DNA knots reveal a chiral organization of DNA in phage capsids. Proceedings
of the National Acadamy of Sciences, 102(26):9165-9169, 2005.

[13] T. Machon and G. Alexander. Knots and nonorientable surfaces in chiral nematics. Proceedings
of the National Acadamy of Sciences, 110(35):14174-14179, 2013.

[14] Antonio F. Ranada. A topological theory of the electromagnetic field. letters in mathematical
physics, 18(2):97106, 1989.

[15] Manuel Arrayas and Jose L. Trueba. Electromagnetic torus knots. 2011, arXiv:hep-
th/1106.1122.

[16] J. W. Dalhuisen and D. Bouwmeester. Twistors and electromagnetic knots. Journal of Physics
A: Mathematical and Theoretical, 45(13):135201, 2012.

[17] William T. M. Irvine. Linked and knotted beams of light, conservation of helicity and the flow of
null electromagnetic fields. Journal of Physics A: Mathematical and Theoretical, 43(38):385203,
2010.

[18] E.T. Whittaker. On the partial differential equations of the mathematical physics. Mathema-
tische Annalen, 57:333355, 1903.

10
[19] Harry Bateman. The Mathematical Analysis of Electrical and Optical Wave-motion on the
Basis of Maxwells Equations. Cambridge University Press, Cambridge, 1915.

[20] Roger Penrose. Twistor theory: its aims and achievements. In C. J. Isham, R. Penrose, and
D. W. Sciama, editors, Quantum Gravity, an Oxford Symposium. Oxford University Press,
Oxford, 1975.

[21] Roger Penrose. Solutions of the zero-rest-mass equations. Journal of Mathematical Physics,
10(38):3839, 1969.

[22] Roger Penrose. On the origins of twistor theory. In W. Rindler and A. Trautman, editors,
Gravitation and geometry, a Volume in Honour of I. Robinson, pages 341361. Bibliopolis,
Naples, 1987.

[23] Emil Grgin. A Global Technique for the Study of Spinor Fields. Ph.D. Thesis, Syracuse
University, 1966.

[24] Roger Penrose and M. A. H. MacCallum. Twistor theory: An approach to the quantisation of
fields and space-time. Physics Reports, 6(4):241316, 1972.

[25] Lane P. Hughston and Richard S. Ward. An introduction to twistor theory. In Advances in
twistor theory. Pitman Publishing Limited, London, 1979.

[26] Andrew Hodges. Twistor diagrams. Physica A: Statistical Mechanics and its Applications,
114(1):157175, 1982.

[27] M. G. Eastwood and A. M. Pilato. On the density of twistor elementary states. Pacific Journal
of Mathematics, 151(2):201215, 1991.

[28] Hridesh Kedia, Iwo Bialynicki-Birula, Daniel Peralta-Salas, and William M. T. Irvine. Tying
knots in light fields. Physical Review Letters, 111(15):150404-14, 2013.

[29] David A. Nichols, Robert Owen, Fan Zhang, Aaron Zimmerman, Jeandrew Brink, Yanbei
Chen, Jeffrey D. Kaplan, Geoffrey Lovelace, Keith D. Matthews, Mark A. Scheel, and Kip S.
Thorne. Visualizing spacetime curvature via frame-drag vortexes and tidal tendexes: General
theory and weak-gravity applications. Phys. Rev. D, 84:124014, Dec 2011.

[30] Fan Zhang, Aaron Zimmerman, David A. Nichols, Yanbei Chen, Geoffrey Lovelace, Keith D.
Matthews, Robert Owen, and Kip S. Thorne. Visualizing spacetime curvature via frame-drag
vortexes and tidal tendexes II. Stationary black holes. Phys. Rev. D, 86:084049, 2012.

[31] David A. Nichols, Aaron Zimmerman, Yanbei Chen, Geoffrey Lovelace, Keith D. Matthews,
Robert Owen, Fan Zhang, and Kip S. Thorne. Visualizing spacetime curvature via frame-drag
vortexes and tidal tendexes III. Quasinormal pulsations of schwarzschild and Kerr black holes.
Phys. Rev. D, 86:104028, 2012.

[32] R. Penrose and W. Rindler. Spinors and Space-Time: Volume 2, Spinor and Twistor Meth-
ods in Space-Time Geometry. Cambridge monographs on mathematical physics. Cambridge
University Press, Cambridge, 1988.

[33] Manuel Arrayas and Jose L. Trueba. Exchange of helicity in a knotted electromagnetic field.
2011, arXiv:hep-th/1105.6285.

11