Itzhak Bars and Moises Picon - Twistor Transform in D Dimensions and A Unifying Role For Twistors

a
r
X
i
v
:
h
e
p
-
t
h
/
0
5
1
2
3
4
8
v
5

2
1

M
a
r

2
0
0
6
USC-05/HEP-B3 hep-th/0512348
Twistor Transform in d Dimensions
and a Unifying Role for Twistors
Itzhak Bars
a
and Moises Picon
a,b
a)
Department of Physics and Astronomy,
University of Southern California, Los Angeles, CA 90089-0484, USA
b)
Departamento de Fsica Teorica,
University of Valencia and IFIC (CSIC-UVEG), 46100-Burjasot (Valencia), Spain
Abstract
Twistors in four dimensions d = 4 have provided a convenient description of massless
particles with any spin, and this led to remarkable computational techniques in Yang-Mills
eld theory. Recently it was shown that the same d = 4 twistor provides also a unied
description of an assortment of other particle dynamical systems, including special examples
of massless or massive particles, relativistic or non-relativistic, interacting or non-interacting,
in at space or curved spaces. In this paper, using 2T-physics as the primary theory, we
derive the general twistor transform in d-dimensions that applies to all cases, and show
that these more general twistor transforms provide d dimensional holographic images of an
underlying phase space in at spacetime in d + 2 dimensions. Certain parameters, such as
mass, parameters of spacetime metric, and some coupling constants appear as moduli in the
holographic image while projecting from d + 2 dimensions to (d 1) + 1 dimensions or to
twistors. We also extend the concept of twistors to include the phase space of D-branes,
and give the corresponding twistor transform. The unifying role for the same twistor that
describes an assortment of dynamical systems persists in general, including D-branes. Except
for a few special cases in low dimensions that exist in the literature, our twistors are new.
1
I. BOTTOMUP APPROACH TO SPACETIME
In this section we will discuss mainly spinless particles to establish some concepts in
a familiar setting. A similar analysis applies to particles with spin
1
, therefore when we
discuss twistors in four dimensions, d = 4 in section-I B, we include the spin in some of
the discussion. Generalizations of the twistors to higher dimensions, and D-branes, will
be presented in the following sections. Other generalizations, including spinning particles,
supersymmetry, compactied internal spaces, will be given elsewhere.
A. Phase space
A massless and spinless relativistic particle in (d 1) space dimensions is described by
its position-momentum phase space coordinates (x
i
(t) , p
i
(t)) , i = 1, 2, , (d 1) while its
time development is governed by the Hamiltonian H = [p[ =
_
p
2
. Hence its action is
S =
_
dt (
t
x p [ p[) . (1.1)
For a spinning particle we include spin degrees of freedom in an enlarged phase space. Of
course, a massless particle is a relativistically invariant system, and this is veried by the
fact that this action is invariant under Lorentz transformations SO(d 1, 1). However, this
symmetry is only partially manifest in this action: rotation symmetry SO(d 1) is evident
while the boost symmetry is hidden. To make the Lorentz symmetry fully manifest one
must introduce a gauge symmetry together with extra matter gauge degrees of freedom and
a gauge eld. This process is a rst step in the bottomup approach that helps us discover
a deeper point of view of symmetries and their connection to spacetime.
The well known bottom-up approach in this example is to introduce the worldline
reparametrization gauge symmetry and then use the larger phase space (x
() , p
()) with
= 0, 1, 2, , (d 1) . The action in the rst order formalism is
S (x, p) =
_
d
_
1
2
ep
_
, (1.2)
where
is the Minkowski metric and e () is the gauge eld coupled to the generator of
gauge transformations p
2
/2. The gauge transformations
e =
() ,
= () p
= 0, (1.3)
1
A detailed discussion of the spinning particle and the related generalized twistors, including higher dimen-
sions and supersymmetry, will be given in a future paper [1].
2
transform the Lagrangian into an ignorable total derivative
S =
_
d
(p
2
/2) = 0. The
action (1.2) has an evident global Lorentz symmetry due to the fact that all terms are Lorentz
dot products. Noethers theorem gives the conserved charges L
= x
. These are
gauge invariant
= 0, and hence they are physical observables. So the L
remain
conserved, and act as the generators of symmetry of the gauge invariant action S (x, p) ,
even if some arbitrary gauge is xed.
The relation between Eqs.(1.1,1.2) is obtained in a xed gauge. This is the reverse process,
namely it is part of the topbottom approach that will be discussed in the next section. The
equation of motion with respect to the gauge eld e requires that the gauge generator vanishes
p
2
=0, implying that the physical sector (massless particle on mass shell) must be gauge
invariant. The gauge symmetry can be xed by taking x
0
() = t () = , and the constraint
can be solved for the canonical conjugate to x
0
= , namely p
0
= [ p[ . The remaining phase
space (x, p) provides a parametrization of the gauge invariant sector. Taking the positive
root p
0
= +[p[, we derive the non-covariant action Eq.(1.1) for the massless particle as the
gauge xed form of the gauge invariant Eq.(1.2). Similarly, the gauge xed form of the
gauge invariant L
, given by L
ij
= x
i
p
j
x
j
p
i
, L
0i
= p
i
x
i
[ p[ , are the generators of the
non-linearly realized hidden global SO(d 1, 1) symmetry of the gauge xed Lagrangian in
Eq.(1.1). These generate the Lorentz transformations of phase space (x() , p ()) at any
through the Poisson brackets
x() =
1
2
, x () , and
p () =
1
2
, p ()
where is treated like a parameter. It can be checked that under these transformations the
non-covariant looking action in Eq.(1.1) is Lorentz invariant (dropping an ignorable total
derivative S =
_
d
(f ()) = 0).
The covariant formulation in Eq.(1.2) provides a greater exibility to analyze the system
from a broader and more fundamental perspective. For example, one may choose other
gauges besides the timelike gauge x
0
() = that relates Eqs.(1.1) and (1.2). In particular
the lightcone gauge x
+
() = , in which the constraint p
2
= 0 is solved for the canonical
conjugate p
= p
2
/2p
+
, has certain advantages in computation.
One may also analyze the system covariantly. For example in covariant quantization one
may apply the constraints on the physical states to derive the Klein-Gordon equation, and
from it the Klein-Gordon free eld theory
p
2
[) = 0 gauge invariant physical states. (1.4)
x[p
2
[) = 0 =
(x) S
KG
=
_
d
4
x
. (1.5)
The covariant formulation in Eq.(1.2) is one of the stations in the bottomup approach
toward a deeper point of view of symmetries and spacetime.
We are not done yet with the hidden symmetries of the non-covariant action in Eq.(1.1).
This system has the larger symmetry SO(d, 2) , namely conformal symmetry which is a
3
general symmetry of massless systems. This symmetry persists as a hidden symmetry in
the covariant action in Eq.(1.2) and in the Klein-Gordon action in Eq.(1.5). It has been
known that the SO(d, 2) symmetry can be made manifest in two ways: one is twistors [2][3]
(in d = 4), and the other is two time physics (2T-physics) [4][5] in any d. Actually these
are related to each other by gauge transformations in the 2T-physics formalism [6][7] in the
twistor gauge as we will discuss in the rest of the paper in more detail. Either way, the route
to making the SO(d, 2) symmetry manifest involves introducing a gauge symmetry, extra
matter gauge degrees of freedom, and gauge elds.
B. Twistor space in d=4
The twistor formalism in d = 4 [2][3] starts from a dierent description of the mass-
less particle. Instead of phase space degrees of freedom x
, p
that are SO(3, 1) vectors,

it introduces SO(3, 1)=SL(2, C) spinor degrees of freedom Z
A
=
_
_
, A = 1, 2, 3, 4, con-
structed from SL(2, C) doublet spinors

,
, each described by two complex degrees of

freedom , = 1, 2. The quartet Z
A
is the spinor representation 4 of the conformal group
SO(4, 2) =SU(2, 2), while its conjugate

Z
A
=
_
Z
C
_
A
=
_
_
,with the SU(2, 2) metric
C =
1
1, (1.6)
is the anti-quartet

4 that corresponds to the second spinor representation of SO(4, 2) . An
over-bar such as

means complex conjugate of
. The spinor is subject to a SU(2, 2)

invariant helicity constraint Z
A

Z
A
=

= 2h, where h is the helicity of the

particle. The helicity constraint is the generator of a U(1) gauge symmetry that acts on the
twistor through the local phase transformation Z
A
() Z
A
() = e
i()
Z
A
() . The gauge
invariant action that describes the dynamics of twistors in four dimensions is
S (Z) =
_
d
_
i
Z
A
DZ
A
2hV
_
, DZ
A

Z
A
iV Z
A
. (1.7)
Here the 1-form V d is a U(1) gauge eld on the worldline, DZ
A
is the gauge covariant
derivative that satises
(DZ
A
) = i() (DZ
A
) for
V = / and
Z
A
= i() Z
A
.
Note that the term 2hV (absent in previous literature) is gauge invariant since it transforms
as a total derivative under the gauge transformation. The reason for requiring the U(1)
gauge symmetry is the fact that the overall phase of the Z
A
is unphysical and drops out in
the relation between phase space and twistors, as in Eq.(1.8). Furthermore, the equation
of motion with respect to V imposes the constraint Z
A

Z
A
2h = 0, which is interpreted
as the helicity constraint. Taking into account that
_
Z
A

Z
A
2h
_
is the generator of the
U(1) gauge transformations, the meaning of the vanishing generator (or helicity constraint)
is that only the U(1) gauge invariant sector of twistor space is physical.
4
To establish the equivalence between the massless spinless particle in Eqs.(1.1,1.2) in
d = 4, and twistors with vanishing helicity h = 0, we must choose a U(1) gauge for Z
A
and
solve the constraint Z
A

Z
A
=

= 0. First we count degrees of freedom. The

twistor Z
A
has 4 complex , or 8 real, degrees of freedom. Gauge xing the U(1) symmetry
and solving the helicity constraint removes 2 real degrees of freedom, leaving behind 6 real
degrees of freedom, which is the same number as the phase space degrees of freedom (x, p) .
More explicitly, Penrose has provided the transformation between twistors and the phase
space of spinless massless particles as follows

= ix

= p
, (1.8)
where the 2 2 Hermitian matrices x

, p
are expanded in terms of the Pauli matrices

x

2
x
)

, p
2
p
(1, ) ,
(1, ) . (1.9)
Here
can be gauge xed by choosing a phase, and the helicity constraint is explicitly
solved since
Z
A
Z
A
=
_
_
_
_
=

= i
+ i
= 0. (1.10)
Furthermore, since
is a 22 matrix of rank one, its determinant vanishes. Then

the parametrization p
of the momentum p
, with
2p
0
= Tr (p) =

, insures
automatically that p
0
> 0 and that the mass shell condition is satised det
_
_
= det (p) =
p
= 0. Furthermore, by inserting the twistor transform in Eq.(1.8) into the twistor action
S (Z) we recover the massless particle action Eq.(1.2) on which p
2
= 0 is already imposed
by the form of p
S (Z) =
_
d
_
_
x

+ 0
_
=
_
d
x

(1.11)
=
1
2
_
d
x
Tr (
) =
_
d
x
, (with p
2
= 0) (1.12)
Finally, xing the reparametrization symmetry x
0
= shows that S (Z) gives the original
massless particle action Eq.(1.1). From this we conclude that the canonical structure of
twistors determined by S (Z) , namely
_
Z
A
,

Z
B
=
B
A
, is equivalent to the canonical struc-
ture in phase space determined by S (x, p) , namely [x
, p
] = i
, in the gauge invariant

sectors determined respectively by the constraints

ZZ = 0 and p
2
= 0.
The analog of covariant quantization described in Eq.(1.5) can also be done in twistor
space. At the quantum level the Hermitian ordered product is applied on physical states
1
2
(Z
A

Z
A
+

Z
A
Z
A
)[) = 2h[). Wavefunctions in twistor space are obtained as (Z) = Z[)
where the operators Z
A
are diagonalized on the states labelled as Z[. Using the fact that the
5
canonical conjugate acts as

Z
A
(Z) = Z[
Z
A
[) =

Z
A
(Z) , the physical state condition
in Z space
1
2
Z[(Z
A

Z
A
+

Z
A
Z
A
)[) = 2hZ[) produces Penroses homogeneity constraint,
Z
A
Z
A
(Z) = (2h 2) (Z) , known to correctly describe the quantum wavefunction
(Z) of a particle of helicity h.
As the analog of the Klein-Gordon eld theory of Eq.(1.5), we propose an action in twistor
space eld theory that takes into account at once both the positive and negative helicities
S
h
() =
_
d
4
Z
_
Z
A
Z
A
+ 2h + 2
_
. (1.13)
The minimal action principle yields the homogeneity constraints. Indeed, the equations of
motion derived from S
h
() for ,
show that (Z) is the helicity +h wavefunction, while
(Z) is the helicity h wavefunction, so that together they describe a CPT invariant free
eld theory. Therefore S
h
() is the twistor equivalent of the Klein-Gordon, Dirac, Maxwell
and higher spinning particle free eld actions in four at dimensions. The free eld theory
S
h
() is evidently invariant under conformal transformations SU(2, 2) by taking (Z) to
transform like a scalar while Z
A
transforms like the fundamental representation of SU(2, 2) .
Our eld theory proposal for an action principle S
h
() for any spinning particle seems to
be new in the literature.
What do we learn from the twistor approach? Perhaps, as Penrose would suggest, twistors
may be more basic than spacetime? Without a conclusive answer to that question so far, it is
nevertheless evident through the twistor program [2],[3], and the recent twistor superstring
[8]-[13] that led to computational advances in Super Yang Mills theory [14][15], that twistor
space is a useful space, as an alternative to space-time, to discuss the physics of massless
systems.
But there is more to be said about twistors and spacetime. In addition to what has
traditionally been known about twistors, it has recently been shown [7] that the same twistor
Z
A
in Eq.(1.7) which is known to describe the on-shell massless particle, also describes a
variety of other on-shell dynamical systems. In particular the twistor transform of Eq.(1.8)
has been generalized so that the same twistor also gives the d=4 particle worldline actions for
the massive relativistic particle, the particle on AdS
4
or AdS
3
S
1
or AdS
2
S
2
, the particle
on RS
3
, the nonrelativistic free particle in 3 space dimensions, the nonrelativistic hydrogen
atom in 3 space dimensions, and a related family of other particle systems. For example the
twistor transform for the massive relativistic particle is [7]

= ix

2a
1 +a
,
=
1 +a
2a
p
+
m
2
2(x p)a
x
, (1.14)
where a
_
1 +
m
2
x
2
(xp)
2
. Using this transform instead of Eq.(1.8) we nd that the action S (Z)
in Eq.(1.7) reduces to the action for the massive particle S =
_
dt
_
t
x p
_
p
2
+ m
2
_
6
instead of the massless particle of Eq.(1.1). The mass parameter emerges as a modulus in
relating the twistor components
_
_
to phase space (x
, p
) in a dierent way than

Eq.(1.8). As seen in [7] the mass parameter can also be thought of as the value of an extra
momentum component in 4+2 dimensions. Similarly in the case of the twistor transform for
the H-atom, a combination of mass and the Coulomb coupling constant is a modulus, and
so on for other moduli in more general cases. We see that certain mass parameters, certain
curvature or other spacetime metric parameters, and certain coupling constants emerge as
moduli in the generalized twistor transform.
The results in [7] imply that twistor space is a unifying space for various dynamics, with
dierent Hamiltonians, which must be related to one another through a web of dualities.
This raises deeper questions on the meaning of space and time, and accentuates the feeling
that twistor space may be even more fundamental than was thought of before: namely,
from the point of view of twistor space, spacetime and dynamics are emergent concepts, as
explicitly shown in the examples in ref.[7]. There seems to be a deeper meaning for twistors
in the context of unication that goes beyond the originally envisaged role for twistors
2
.
Actually ref.[7] provides a connection between the more general twistor properties just out-
lined and the concept of 2T-physics. This relation will be explained through the topdown
approach in the next section. Here we will briey describe the relevant properties of 2T-
physics that unify various particle dynamics in 1T-physics, and thus promote the notion of
spacetime to a higher level.
C. 2T-physics
2T-physics can be viewed as a unication approach for one-time physics (1T-physics) sys-
tems through higher dimensions. It is distinctly dierent than Kaluza-Klein theory because
there are no Kaluza-Klein towers of states, but instead there is a family of 1T systems with
duality type relationships among them.
A particle interacting with various backgrounds in (d 1) + 1 dimensions (e.g. electro-
magnetism, gravity, high spin elds, any potential, etc.), usually described in a worldline
formalism in 1T-physics, can be equivalently described in 2T-physics. The 2T theory is in
d + 2 dimensions, but has enough gauge symmetry to compensate for the extra 1 + 1 di-
mensions, so that the physical (gauge invariant) degrees of freedom are equivalent to those
2
Promoting these results to the quantum level we must expect another remarkable result that the wave
equations that follow from our proposed action S
h
() in Eq.(1.13) must also correctly describe all the
other cases unied by the same twistor as given in [7]. This point will not be further discussed in this
paper and will be taken up in a future publication.
7
encountered in 1T-physics.
One of the strikingly surprising aspects of 2T-physics is that a given d + 2 dimensional
2T theory descends, through gauge xing, down to a family of holographic 1T images in
(d 1) + 1 dimensions. Each image fully captures the gauge invariant physical content of
a unique parent 2T theory, but from the point of view of 1T-physics each image appears
as a dierent 1T-dynamical system. The members of such a family naturally must obey
duality-type relationships among them and share many common properties. In particular
they share the same overall global symmetry in d + 2 dimensions that becomes hidden and
non-linear when acting on the fewer (d 1) + 1 dimensions in 1T-physics. Thus 2T-physics
unies many 1T systems into a family that corresponds to a given 2T-physics parent in d+2
dimensions.
The essential ingredient in 2T-physics is the basic gauge symmetry Sp(2,R) acting on
phase space X
M
, P
M
in d +2 dimensions. The two timelike directions is not an input, but is
one of the outputs of the Sp(2, R) gauge symmetry. A consequence of this gauge symmetry
is that position and momentum become indistinguishable at any instant, so the symmetry
is of fundamental signicance. The transformation of X
M
, P
M
is generally a nonlinear map
that can be explicitly given in the presence of background elds [16], but in the absence of
backgrounds the transformation reduces to a linear doublet action of Sp(2, R) on
_
X
M
, P
M
_
for each M [4]. The physical phase space is the subspace that is gauge invariant under
Sp(2, R) . Since Sp(2, R) has 3 generators, to reach the physical space we must choose 3
gauges and solve 3 constraints. So, the gauge invariant subspace of d +2 dimensional phase
space X
M
, P
M
is a phase space with six fewer degrees of freedom in (d 1) space dimensions
(x
i
, p
i
) , i = 1, 2, (d 1) .
In some cases it is more convenient not to fully use the three Sp(2, R) gauge symmetry
parameters and work with an intermediate space in (d 1) + 1 dimensions (x
, p
) , that
includes time. This space can be further reduced to d 1 space dimensions (x
i
, p
i
) by a
remaining one-parameter gauge symmetry.
There are many possible ways to embed the (d 1) + 1 or (d 1) phase space in d + 2
phase space, and this is done by making Sp(2,R) gauge choices. In the resulting gauge xed
1T system, time, Hamiltonian, and in general curved spacetime, are emergent concepts. The
Hamiltonian, and therefore the dynamics as tracked by the emergent time, may look quite
dierent in one gauge versus another gauge in terms of the remaining gauge xed degrees of
freedom. In this way, a unique 2T-physics action gives rise to many 1T-physics systems.
The general 2T theory for a particle moving in any background eld has been constructed
[16]. For a spinless particle it takes the form
S =
_
d
_

X
M
P
M

1
2
A
ij
Q
ij
(X, P)
_
, (1.15)
8
where the symmetric A
ij
() , i, j = 1, 2, is the Sp(2, R) gauge eld, and the three Sp(2, R)
generators Q
ij
(X () , P ()) , which generally depend on background elds that are functions
of (X () , P ()), are required to form an Sp(2, R) algebra. The background elds must
satisfy certain conditions to comply with the Sp(2, R) requirement. An innite number of
solutions to the requirement can be constructed [16]. So any 1T particle worldline theory,
with any backgrounds, can be obtained as a gauge xed version of some 2T particle worldline
theory.
The 1T systems discussed in [7], and alluded to in connection with twistors above, are
obtained by considering the simplest version of 2T-physics without any background elds.
The 2T action for a free 2T particle is [4]
S
2T
(X, P) =
1
2
_
d D
X
M
i
X
N
j

MN
ij
=
_
d
_

X
M
P
N
1
2
A
ij
X
M
i
X
N
j
_
MN
. (1.16)
Here X
M
i
=
_
X
M
P
M
_
, i = 1, 2, is a doublet under Sp(2, R) for every M, the structure
D
X
M
i
=
X
M
i
A
j
i
X
M
j
is the Sp(2,R) gauge covariant derivative, Sp(2,R) indices are
raised and lowered with the antisymmetric Sp(2, R) metric
ij
, and in the last expression
an irrelevant total derivative (1/2)
(X P) is dropped from the action. This action

describes a particle that obeys the Sp(2, R) gauge symmetry, so its momentum and position
are locally indistinguishable due to the gauge symmetry. The
_
X
M
, P
M
_
satisfy the Sp(2, R)
constraints
Q
ij
= X
i
X
j
= 0 : X X = P P = X P = 0, (1.17)
that follow from the equations of motion for A
ij
. The vanishing of the gauge symmetry
generators Q
ij
= 0 implies that the physical phase space is the subspace that is Sp(2, R)
gauge invariant. These constraints have non-trivial solutions only if the metric
MN
has two
timelike dimensions. So when position and momentum are locally indistinguishable, to have
a non-trivial system, two timelike dimensions are necessary as a consequence of the Sp(2, R)
gauge symmetry.
Thus the
_
X
M
, P
M
_
in Eq.(1.16) are SO(d, 2) vectors, labelled by M = 0
, 1
, or M =
, , and = 0, 1, , (d 1) or = , 1, , (d 2) , with lightcone type denitions of

X
=
1
2
_
X
0
X
1
_
and X
=
1
2
(X
0
X
3
) . The SO(d, 2) metric
MN
is given by
ds
2
= dX
M
dX
N
MN
= 2dX
+
dX
+ dX
dX
(1.18)
=
_
dX
0
_
2
+
_
dX
1
_
2
_
dX
0
_
2
+
_
dX
1
_
2
+ (dX
)
2
(1.19)
= 2dX
+
dX
2dX
+
dX
+ (dX
)
2
. (1.20)
where the notation X
indicates SO(d 2) vectors. So the target phase space X

M
, P
M
is
at in d +2 dimension, and hence the system in Eq.(1.16) has an SO(d, 2) global symmetry.
9
The conserved generators of SO(d, 2)
L
MN
= X
M
P
N
X
N
P
M
,
L
MN
= 0, (1.21)
commute with the SO(d, 2) invariant Sp(2, R) generators X X, P P, X P.
The Sp(2, R) local symmetry can be gauge xed by choosing three gauges and solving
three constraints, but to keep some of the subgroups of SO(d, 2) as evident symmetries it is
more convenient to choose two gauges and solve two constraints.
The SO(d 1, 1) covariant massless particle emerges if we choose the two gauges,
X
+
() = 1 and P
+
() = 0, and solve the two constraints X

2
= X P = 0 to obtain
the (d 1) + 1 dimensional phase space (x
, p
) embedded in (d + 2) dimensions
X
M
=
_
+
1,
x
2
/2 ,
_
, (1.22)
P
M
= ( 0 , x p , p
) . (1.23)
The remaining constraint, P
2
= 2P
+
+ P
= p
2
= 0, which is the third Sp(2, R)
generator, remains to be imposed on the physical sector. In this gauge the 2T-physics action
in Eq.(1.16) reduces to the covariant massless particle action in Eq.(1.2). Furthermore, the
Sp(2, R) gauge invariant L
MN
= X
M
P
N
X
N
P
M
take the following nonlinear form
L
= x
, L
+
= x p, L
+
= p
, L
=
x
2
2
p
x p. (1.24)
These are recognized as the generators of SO(d, 2) conformal transformations of the (d 1)+
1 dimensional phase space at the classical level. Thus the conformal symmetry of the massless
system is now understood as the Lorentz symmetry in d + 2 dimensions.
Having established the higher symmetrical version of the theory for the massless particle
as in Eq.(1.16) we reach a deeper level of understanding of the symmetries as well as the
presence of the d + 2 nature of the underlying spacetime. Furthermore we learn that the
higher symmetrical parent theory can be gauge xed in many ways that produce not only the
massless particle system Eq.(1.1) we started from, but also an assortment of other particle
dynamical systems, as discussed before [4][5][7].
To emphasize this point we give also the massive relativistic particle gauge by xing two
gauges and solving the constraints X
2
= X P = 0 explicitly as follows
X
M
=
_
_
_
+
1 +a
2a
,
x
2
a
1 +a
,
_
_
_
, a
1 +
m
2
x
2
(x p)
2
(1.25)
P
M
=
_
m
2
2(x p)a
, (x p) a , p
_
, P
2
= p
2
+ m
2
= 0. (1.26)
10
In this gauge the 2T action reduces to the relativistic massive particle action
S =
_
d
_

X
M
P
N
1
2
A
ij
X
M
i
X
N
j
_
MN
=
_
d
_
x
1
2
A
22
_
p
2
+ m
2
_
_
. (1.27)
A little recognized fact is that this action is invariant under SO(d, 2). This SO(d, 2) does
not have the form of conformal transformations of Eq.(1.24), but is a deformed version of it,
including the mass parameter. Its generators are obtained by inserting the massive particle
gauge into the gauge invariant L
MN
= X
M
P
N
X
N
P
M
L
= x
, L
+
= (x p) a, (1.28)
L
+
=
1 +a
2a
p
+
m
2
2 (x p) a
x
(1.29)
L
=
x
2
a
1 +a
p
(x p) ax
(1.30)
It can be checked explicitly that the massive particle action above is invariant under the
SO(d, 2) transformations generated by the Poisson brackets x
=
1
2
MN
_
L
MN
, x
_
and
p
=
1
2
MN
_
L
MN
, p
_
, up to a reparametrization of A
22
by a scale and an irrelevant total
derivative.
Since both the massive and massless particles give bases for the same representation of
SO(d, 2), we must expect a duality transformation between them. Of course this trans-
formation must be an Sp(2, R) =SL(2, R) local gauge transformation
_
_
() with unit
determinant = 1, that transforms the doublets
_
X
M
P
M
_
() from Eqs.(1.25,1.26) to
Eqs.(1.22,1.23). The , , , are xed by focussing on the doublets labelled by M = +
_ _
1+a
2a
_
_
m
2
2(xp)a
_
_
=
_ _
1+a
2a
_
0
_
m
2
2(xp)a
_
_
2a
1+a
_
__
1
0
_
. (1.31)
Applying the inverse of this transformation on the doublets labelled by M = gives the
massless particle phase space (re-labelled by ( x
, p
) below) in terms of the massive particle

phase space (labelled by (x
, p
))
_ _
2a
1+a
_
0
_
m
2
2(xp)a
_
_
1+a
2a
_
__
x
_
=
_
2a
1+a
x
1+a
2a
p
+
m
2
2(xp)a
x
_
x
_
(1.32)
This duality transformation is a canonical transformation x
, p
= x
, p
. Also
note that the time coordinate x
0
is dierent than the time coordinate x
0
, and so are the
corresponding Hamiltonians for the massless particle p
0
=
_
p
i
p
i
versus the massive particle
p
0
=
_
p
i
p
i
+ m
2
.
The same reasoning applies among all gauge choices of the 2T theory in Eq.(1.16). All
resulting 1T dynamical systems are holographic images of the same parent theory. The global
11
symmetry SO(d, 2) of the 2T-physics action is shared in the same singleton
3
representation by
all the emergent lower dimensional theories obtained by dierent forms of gauge xing. These
include special cases of particles that are massive or massless, relativistic and nonrelativistic,
in at or curved spaces, free or interacting. This is an established fact in previous work on
2T-physics [4][5], and it came into new focus by displaying the explicit twistor/phase space
transforms given in [7].
As seen above, the descendants of the d+2 dimensional 2T-physics action are 1T-physics
dynamical systems that are dual to each other. Therefore we must expect that they all have
the same twistor representation modulo twistor gauge transformations. This will be derived
through the topdown approach in the next section.
It must be emphasized that as a by product of the topdown approach certain physical
parameters, such as mass, parameters of spacetime metric, and some coupling constants
appear as moduli in the holographic image while descending from d + 2 dimensional phase
space to (d 1) + 1 dimensions or to twistors. Explicit examples of these have appeared in
[7].
II. TOPDOWN APPROACH
The 2T-physics action (1.16) and the twistor action (1.7) in four dimensions are related
to one another and can both be obtained as gauge choices from the same theory in the 2T-
physics formalism. To demonstrate this fact and setup a general formalism for deriving the
twistor transform in any dimension, with or without supersymmetry, we discuss a unied
theory that denes the topdown approach. This formalism was introduced in [6] and
developed further in the context of the twistor superstring [12][13]. In this section we begin
without supersymmetry or compactied dimensions. These will be introduced later.
In the case of d = 4 the generalized twistor transform was applied explicitly to specic
cases in [7], but the derivation of the general formula was relegated to the present paper.
In this section we will derive the general twistor transform between twistor space Z in d-
dimensions and the d + 2 dimensional phase space X
M
, P
M
or d dimensional phase space
x
, p
. We will show how it works explicitly in d = 3, 4, 5, 6 and higher dimensions.

3
At the classical level all Casimir eigenvalues vanish, but at the quantum level, due to ordering of factors
the Casimir eigenvalues are non-zero (see Eq.(2.9)) and correspond to the singleton representation.
12
A. SO(d, 2) local and SO(d, 2) global symmetry
In addition to the phase space SO(d, 2) vectors
_
X
M
, P
M
_
, we introduce a group element
g () SO(d, 2) in the spinor representation. It is given by
g () = exp
_
i
2
S
MN
MN
()
_
= exp
_
1
4
MN
MN
()
_
(2.1)
The 2T particle action with Sp(2, R) & SO(d, 2) local and SO(d, 2) global symmetry is
S
2T
(X, P, g) =
_
d
_
1
2
ij
X
i
X
j
1
2
A
ij
X
i
X
j
+
4
s
d
Tr
_
ig
1
gL
_
_
, (2.2)
where the trace is in spinor space
4
and the matrix L is given by
L
1
4i
MN
L
MN
=
1
4i
_
X

P P

X
_
. (2.3)
The rst two terms of the action S
2T
(X, P, g) are the same as Eq.(1.16), hence these terms
are invariant under Sp(2, R) which acts on X
M
i
=
_
X
M
, P
M
_
as a doublet for every M,
and on A
ij
as the gauge eld. Furthermore, by taking g () as a singlet while noting that
L
MN
=
ij
X
M
i
X
N
j
= X
M
P
N
X
N
P
M
is Sp(2, R) gauge invariant, we see that the full action
is gauge invariant under Sp(2, R) . The action can be rewritten in the form
S
2T
(X, P, g) =
_
d
_
1
2s
d
ij
Tr
_
_
gX
i
g
1
_ _
gX
j
g
1
_
1
2
A
ij
X
i
X
j
_
. (2.4)
When both X
M
i
and g () are transformed under local Lorentz transformations as
R
X
M
i
=
MN
R
() X
iN
and
R
g =
1
4
(g
MN
)
MN
R
() , the structures (gX
j
g
1
) and X
i
X
j
are gauge
invariant under
R
. Therefore the Lagrangian has a gauge symmetry with local SO(d, 2)
R
pa-
rameters
MN
R
() when g is transformed on the right side. In addition, there is a global sym-
metry under SO(d, 2)
L
when g () is transformed from the left side as
L
g =
1
4
MN
L
(
MN
g),
with independent parameters
MN
L
.
4
The trace in spinor space gives the dimension of the spinor Tr (1) = s
d
and Tr
_
N
_
= s
d
MN
. For even
dimensions s
d
= 2
d/2
for the Weyl spinor of SO(d, 2) , and the

M
,
M
are the gamma matrices in the bases
of the two dierent spinor representations.The correctly normalized generators of SO(d, 2) in the spinor
representation are S
MN
=
1
2i
MN
, where the even-dimension gamma matrices satisfy
M
N
+
N
M
=
2
MN
, while
MN
=
1
2
_
M
_
,
MNK
=
1
3!
_
K
permutations
_
, etc. There exists a
metric C of SO(d, 2) in the spinor representation such that when combined with hermitian conjugation it
gives C
1
_
M
_
C =
M
and C
1
_
MN
_
C =
MN
. So the inverse g
1
is obtained by combining
hermitian and C-conjugation g
1
= C
1
(g)
C g. In odd number of dimensions the even-dimension

gamma matrices above are combined to a larger matrix

M
=
_
0
M
0
_
for M = 0
, 1
, 0, 1, , (d 2)
and add one more matrix for the additional last dimension

d1
=
_
1
0
0
1
_
. The text is written as if d is
even; for odd dimensions we replace everywhere

M
for both
M
and

M
.
13
Using Noethers theorem we construct the conserved charge J
MN
() of the global left
side symmetry SO(d, 2)
L
. We nd J
MN
iTr
_
MN
gLg
1
_
, but we prefer to write it in
spinor space in the form
J
B
A
=
_
gLg
1
_
B
A
= J
MN
()
_
1
4i
MN
_
B
A
. (2.5)
Note that the matrix J
B
A
must have the same form as the matrix L
B
A
of Eq.(2.3), i.e.
J =
_
1
4i
MN
_
J
MN
() , since gLg
1
=
1
4i
(g
MN
g
1
) L
MN
is a Lorentz transformation of the
gamma matrices that mixes them among themselves. By using the equations of motion for
(X, P, g) one can show that these charges are conserved
J
B
A
() = 0.
As seen from the form of J in Eq.(2.5), it is gauge invariant under the local SO(d, 2)
R
as
well as the local Sp(2, R) transformations. Therefore the SO(d, 2)
L
charges J
B
A
are physical
observables that classify the physical states under SO(d, 2)
L
representations. In particular
the Casimir operators of these representations are given by C
n
=
1
s
d
tr ((2J)
n
) . With this in
mind we study the properties of J. In particular the square of the matrix J, given by (J
2
)
B
A
=
(gLg
1
gLg
1
)
B
A
= (gL
2
g
1
)
B
A
, contains important information about the physical states
as we will see below. To proceed from here we will outline the rest of the computation of J
2
at the classical and quantum levels.
If the square of the matrix L
2
is computed at the classical level, i.e. not caring about the
orders of generators L
MN
, then one nds that (L
2
)
B
A
is proportional to the identity matrix
B
A
, (L
2
) =
_
1
4i
MN
L
MN
_
2
=
1
8
L
MN
L
MN
1. Furthermore by computing, still at the classical
level
1
2
L
MN
L
MN
= X
2
P
2
(X P)
2
, and imposing the classical constraints X
2
= P
2
=
(X P) = 0, one nds that L
2
= 0 in the space of gauge invariants of the classical theory.
Then this implies also J
2
= 0 in the space of gauge invariants of the classical theory. By
taking higher powers of J, we nd J
n
= 0 for all positive integers n 2. Therefore all Casimir
eigenvalues are zero C
n
= 0 for all the classical physical congurations of phase space. This
is a special non-trivial representation of the non-compact group SO(d, 2)
L
, and all classical
gauge invariants, which are functions of L
MN
, can be classied as irreducible multiplets of
SO(d, 2)
L
.
We now consider the quantum theory. All the physical (gauge invariant) states must fall
into irreducible representations of the global symmetry SO(d, 2)
L
. In the quantum theory
the L
MN
form the Lie algebra of SO(d, 2) , therefore if the square of the matrix L is computed
at the quantum level, by taking into account the orders of the operators L
MN
, one nds
L
2
=
_
1
4i
MN
L
MN
_
2
=
d
2
_
1
4i
MN
L
MN
_
+
1
8
L
MN
L
MN
1. (2.6)
In this computation we used the properties of gamma matrices
MN
RS
=
MNRS
+ (
NR
MS
MR
NS
NS
MR
+
MS
NR
) + (
NR
MS
MR
NS
) .
14
The term
MNRS
L
MN
L
RS
vanishes for L
MN
= X
[M
P
N]
due to a clash between symme-
try/antisymmetry. The term
NR
MS
turns into a commutator, and after using
the SO(d, 2) Lie algebra for
_
L
MN
, L
RS
it produces the linear term proportional to d/2

in Eq.(2.6). The term
NR
MS
produces the last term in Eq.(2.6). Furthermore
the Casimir
1
2
L
MN
L
MN
does not vanish at the quantum level. As shown in [4], in the
Sp(2, R) gauge invariant physical sector of phase space one nds that it has the xed value
1
2
L
MN
L
MN
= 1d
2
/4 rather than zero. Hence, in the physical sector of the quantum theory
the matrix J
B
A
satises the following algebra
_
J
2
_
B
A
=
d
2
J
B
A
+
1
8
_
1
d
2
4
_

B
A
, on physical states. (2.7)
We can compute the higher powers J
n
on physical states by repeatedly using this relation,
(J
n
)
B
A
=
n
J
B
A
+
n

B
A
, (2.8)
and then compute the Casimir eigenvalues
5
C
n
=
1
s
d
Tr ((2J)
n
) = 2
n
n
. Evidently the C
n
will end up having xed values determined by the dimension d of SO(d, 2)
R
. In particular,
C
2
= 1
d
2
4
, C
3
= d
_
1
d
2
4
_
, C
4
=
_
1
d
2
4
__
1 +
3d
2
4
_
, etc. (2.9)
Therefore, at the quantum level we have identied a special unitary representation that
classies all physical states of the theory. This is the singleton representation of SO(d, 2) for
any d. Our approach shows that the singleton is more fully characterized by the constraints
satised by the charges in Eq.(2.7). We will see in the next section that these constraints
will be satised explicitly at the quantum level by constructing J
B
A
in terms of twistors.
B. Twistor gauge and the general twistor transform
There are dierent ways of choosing gauges to express the theory given by S
2T
(X, P, g)
in terms of the physical sector. One extreme in gauge space is to eliminate g completely,
while another extreme is to eliminate (X, P) completely. When g is eliminated we obtain the
phase space description, and when (X, P) is eliminated we obtain the twistor description.
Since SO(d, 2)
R
is a local symmetry that acts on g () from the right, g g
= gg
R
, and
g
R
has exactly the same number of degrees of freedom as g, one can gauge x the extended
classical theory by choosing the gauge g () = 1. In that case the theory described by Eq.(2.3)
in terms of (X, P, g) reduces to the theory described by only (X, P) in Eq.(1.16)
S
2T
(X, P, g)
g=1
= S
2T
(X, P) , J
B
A
= L
B
A
. (2.10)
5
Note that in the literature one may nd that the denition of the cubic and higher Casimir eigenvalues
are given as a linear combination of our C
n
.
15
In this gauge the conserved charge becomes J = L, so that the global symmetry SO(d, 2)
L
becomes the SO(d, 2) global symmetry of the (X, P) theory. This reects the fact that to
maintain the gauge g = 1, a transformation of g from the left must be compensated by a
transformation from the right, therefore SO(d, 2)
L
and SO(d, 2)
R
become identied.
In the g = 1 gauge there still remains the Sp(2, R) gauge symmetry. If one xes this gauge
as in Eqs.(1.22,1.23) then we see that the original SO(d, 2)
L
is interpreted, in this gauge, as
the conformal symmetry of the relativistic massless particle given in Eqs.(1.24). But if one
xes Sp(2, R) as in Eqs.(1.25-1.26) then the original SO(d, 2)
L
is interpreted as the hidden
SO(d, 2) of the massive particle given in Eqs.(1.28-1.30). So, the original SO(d, 2)
L
applied
on g can take on many possible physical interpretations as the hidden symmetry of various
dynamical particle phase spaces that arise from Sp(2, R) gauge choices. Recall that for all
cases the conserved SO(d, 2) charges are just the physical charges J
B
A
= (g
1
Lg)
B
A
whose
classical and quantum properties were already computed in a gauge invariant way in the
previous section.
To obtain the twistor description of the system we eliminate
_
X
M
, P
M
_
completely and
keep only g as discussed in [6]. This is done by using the Sp(2, R) and the SO(d, 2)
R
local
symmetries to completely x X
M
, P
M
to the convenient form X
+
= 1 and P
+
= 1, while
all other components vanish
X
M
= (
+
1,
0,
+
0,
0,
i
0), P
M
= (
+
0,
0,
+
1,
0,
i
0), i = 1, , (d 2) . (2.11)
These X
M
, P
M
already satisfy the constraints X
2
= P
2
= X P = 0. In this gauge the only
non-vanishing component of L
MN
is L
+
+
= 1, so that
L
fixed
=
2
4i

L
+
+
=
i
2
. (2.12)
Hence the physical content of the theory is now described only in terms of g and the xed
matrix embedded in the Lie algebra of SO(d, 2) .
The matrix has very few non-zero entries as seen by choosing a convenient form of
gamma matrices
6
for SO(d, 2). Then, up to similarity transformations, can be brought to
6
An explicit form of SO(d, 2) gamma matrices that we nd convenient in even dimensions, is given by
0
= 1 1,
i
=
3

i
,
= i
1 (note
0
= i
1
1 and
1
=
2
1), where
i
are
the SO(d 1) gamma matrices. The

M
are the same as the
M
for M =
, i, but for M = 0 we
have

0
=
0
= 1 1. From these we construct the traceless
+
=
_
1
0
0
1
_
,
+
= i
2
_
0
0

0
_
,
= i
2
_
0
0
0
_
,
=
_

0
0
_
, with
=
_
1,
i
_
and
=
_
1,
i
_
. Then
1
2
MN
J
MN
=
J
+
+
1
2
J
J
+
takes the matrix form given in Eq.(2.19). We can further

write
1
=
1
1,
2
=
2
1 and
r
=
3
r
, where the
r
are the gamma matrices for SO(d 3). These
gamma matrices are consistent with the metric C =
1
1 c of Eq.(1.6), and footnote (4), provided
16
the form
7
=
i
2
=
1
2
_
0
0
0
_
=
_
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
_
. (2.13)
The identity matrix 1, and the 0s, are
s
d
4

s
d
4
square block matrices embedded in the spinor
representation of SO(d, 2) . Then the gauge invariant 2T action in Eq.(2.2), and the gauge
invariant SO(d, 2)
L
charges in Eq.(2.5), take the twistor form similar to Eq.(1.7)
S
2T
(X, P, g) =
4
s
d
_
d Tr
_
i
gg
1
_
=
4
s
d
_
d i
Z
a
A
Z
A
a
S
twistor
, (2.14)
(J)
B
A
=
_
gg
1
_
B
A
=
_
Z
a
A
Z
B
a

1
s
d
tr
_
Z

Z
_
B
A
_
, (2.15)
The Z
a
A
,

Z
B
a
are the twistors that already obey the constraints
_
ZZ
_
b
a
= 0 in Eq.(2.16)
below, so
4
s
d
_
d i
Z
a
A
Z
A
a
is the full twistor action (for an equivalent gauge invariant
action that also produces the constraints, see Eq.(2.40)). Due to the form of it is useful to
think of g as written in the form of
s
d
4

s
d
4
square blocks. Then Z
a
A
with A = 1, 2, , s
d
and a = 1, 2, ,
s
d
4
emerges as the rectangular matrix that corresponds to the last block of
columns of the matrix g, and similarly

Z
A
a
corresponds to the second block of rows of g
1
.
Since g
1
= C
1
g
C, we nd that

Z = c
1
Z
C, where C =
1
1c is given in footnote (6).
Furthermore, as part of g, g
1
, the Z
a
A
,

Z
B
a
must satisfy the constraint

Z
A
a
Z
b
A
= 0 since the
product

Z
A
a
Z
b
A
contributes to an o-diagonal block of the matrix 1 in g
1
g = 1,
g
1
g = 1

Z
A
a
Z
b
A
= 0. (2.16)
A constraint such as this one must be viewed as the generator of a gauge symmetry that
operates on the a index (the columns) of the twistor Z
a
A
.
Let us do some counting of degrees of freedom. To describe the particle in d dimensions
we only need 2 (d 1) physical degrees of freedom corresponding to phase space (x, p) . This
counting applies no matter if the particle is massless or massive, relativistic or not relativistic,
in at space or curved space, interacting or not interacting. Our twistors are expected to
apply to all these cases, so we must have the same number of physical parameters in the
twistors given above. Any extra parameters in Z
a
A
beyond 2 (d 1) must be either gauge
degrees of freedom of the twistor, or there must be additional relations among the Z
a
A
. The
s
d
s
d
matrix g (with s
d
= 2
d/2
for even d) is constructed from
1
2
(d + 2) (d + 1) group
c
1
(
r
)
c =
r
. It is possible to choose hermitian
r
with c = 1 for SO(d 3) . If one works in a basis
with c ,= 1, then hermitian conjugation of of SO(d 3) spinors (which occur e.g. in

of Eq.(2.18)) must
be supplemented by multiplying with c, as in

(1 c) .
7
The gamma matrices
M
of footnote (6) can be redened dierently for the left or right sides of g up
to similarity transformations. Thus, for the right side of g we apply a similarity transformation so that
1
=
3
1, etc., to obtain
=
_
1
_
/
2 in the form given in Eq.(2.13).

17
parameters
MN
as in Eq.(2.1), but only 2 (d 1) of those parameters, corresponding to a
coset, contribute in Eqs.(2.14,2.15) due to the form of as will be explained in section (II C).
So for suciently large d, we expect to nd many relations among the s
2
d
/4 = 2
2d2
entries
in the rectangular matrix Z
a
A
(g) .
In d = 3, with s
3
/4 = 1 columns, the Z
A
form the fundamental representation of
Sp(4, R) =SO(3, 2) . The single Z
A
has just 4 real components that automatically satisfy
the constraint

Z
A
Z
A
= 0. This number of degrees of freedom precisely matches the ex-
pected number 4 of physical degrees of freedom, 2 (d 1) = 2 (3 1) = 4, for d = 3. So
there are no extra relations among the 4 real twistor entries in Z
A
.
In d = 4 dimensions, with s
4
A
form the fundamental representation
of SU(2, 2) =SO(4, 2) . The single Z
A
has 4 complex components or 8 real parameters that
must satisfy the U(1) gauge constraint

Z
A
Z
A
= 0 of Eq.(2.16). The U(1) gauge symmetry
together with the constraint remove 2 real parameters. So the Z
A
contains 82 = 6 physical
degrees of freedom, which is just the correct number 2 (4 1) = 6 in d = 4, as discussed in
section (I B). So there are no extra relations among the 4 complex twistor entries in Z
A
.
For d = 5, 6, with s
5
/4 = s
6
a
A
is a doublet under an SU(2) gauge
symmetry for d = 6, and SU(2) U(1) for d = 5. Beyond the SU(2) or SU(2) U(1) gauge
freedom there seems to be further relations, but these amount to a simple pseudo-reality
condition on Z
a
A
, consistent with the transformation rules of Z
a
A
under Spin(6, 2) SU(2).
The pseudo-reality condition emerges from the pseudo-reality of the spinor representation
of SO(6, 2) or SO(5, 2) . So, again there are no complicated relations among the entries of
Z
a
A
for d = 5, 6 as seen by the following simple counting. The number of real entries in the
pseudo-real Z
a
A
is s
2
6
/8 = s
2
5
/8 = 8
2
/4 = 16. The 3 SU(2) gauge conditions together with
the 3 constraints remove 6 parameters, leaving 16 6 = 10, which is precisely the correct
number of physical degrees of freedom for d = 6, i.e. 2 (6 1) = 10. Similarly, for d = 5
the extra U(1) and its constraint removes two more real parameters and this matches the
correct number 2 (5 1) = 8.
For higher dimensions there are gauge symmetries among the columns but there also are
further complicated relations among the Z
a
A
. It turns out that in all cases d 3 just the
rst column of Z
a
A
(i.e. a = 1) already contains all the parameters that describe the physical
degrees of freedom, but it is still useful to deal with the full Z
a
A
since all components can be
conveniently given explicitly in terms of gamma matrices, as seen in Eqs.(2.37,2.38) below.
We can construct explicitly the Z
a
A
,

Z
A
a
that satisfy all of the relations discussed above
at the classical level. This will give the twistor transform we are after. The key is the gauge
invariant J
B
A
. We identify its two dierent forms in the two dierent gauges, one in terms of
18
phase space and the other in terms of twistors, as given in Eqs.(2.10,2.15)
1
4i
MN
L
MN
g=1
= J
X,P0
= Z

Z, with
_
ZZ
_
b
a
= 0, Tr
_
Z

Z
_
= 0. (2.17)
Of course, the (X, P, g = 1) on the left side of the equation are gauge transformations of the
(X
M
=
M
+
, P
M
=
s
d
4

M
+
, and Z (g)) on the right side. So this equation must contain the
twistor transform. More explicitly we write Z in terms of its components
Z
a
A
=
_
_
,

Z
A
a
=
_

_
,

Z
A
a
Z
b
A
=
_
+
_
b
a
= 0, (2.18)
where , are
s
d
2

s
d
4
rectangular matrices. Then we use the gamma matrices, and the
denition of

, in footnote (6), to write the basic relation (2.17) between phase space and
twistors more explicitly as
1
2i
_
L
+
+
1
2
L
2L
2L
+
L
+
+
1
2
L
_
= J =
_

_
. (2.19)
Comparing the lower o diagonal blocks we learn part of the twistor transform
=
1
2
L
+
=
1
2
_
X
+
P
+
. (2.20)
We should also note the twistor relations that follow from the other three blocks. In three or
four dimensions a single doublet satises this equation automatically. In higher dimensions
a single column cannot do it automatically, instead we have
s
d
4
columns in with certain
relations among them. Thanks to the relations among columns, that can be expressed in
terms of gamma matrices as in Eqs.(2.38) below, the equation above will be satised.
Next we consider
_
X
_
J and use the dierent gauge xed forms of the gauge invariant
J to show that it vanishes as follows
_
X
_
J =
_
X
_
L =
1
4i
_
X
_ _
X

P P

X
_
=
1
2i
X X
_
P
_
1
4i
X P
_
X
_
= 0. (2.21)
The last zero is because X P = P P = 0 in the Sp(2, R) gauge invariant physical sector.
Similarly we show also
_
P
_
J = 0. Therefore
_
P
_
J = 0,
_
X
_
J = 0. (2.22)
Hence, every column of J is a simultaneous null eigenstate of the matrices
_
X
_
and
_
P
_
.
19
Furthermore, because J can be written in the form J = Z

Z, it must be that the s
d
s
d
4
matrix Z is a collection of these null eigenstates, so it is possible to write Z as a linear
combination of
s
d
4
columns of L = J as follows
Z
a
A
=
_
_
= L
_
4
_
, (2.23)
where
i
are
s
d
4

s
d
4
matrices that will be determined below by requiring J = L = Z

Z. So
Z must also satisfy the null conditions
_
P
_
Z = 0,
_
X
_
Z = 0 and LZ = 0. (2.24)
The last null relation LZ = 0 is satised automatically if the rst two are satised. This
property is sucient to determine Z
a
A
up to gauge transformations as follows. The explicit
matrices are obtained by using the gamma matrices in footnote (6)
_
X
_
=
_
X
2X
2X
+
_
,
_
P
_
=
_
P
2P
2P
+
_
(2.25)
Then the zero eigenvalue conditions
_
X
_
Z = 0 =
_
P
_
Z are solved by
= i
X
2X
+
= i
P
2P
+
, = i
X
2X
= i
P
2P
. (2.26)
To show that these expressions are consistent with each other, note that the second set is
obtained by inverting the rst set as long as X
2
= P
2
= X P = 0 are satised in the
physical sector. For example, multiply both sides of the equation = i
X
2X
+
by i
X
2X
,
then use
= 2
and X
2
= X
2X
+
= 0, to obtain = i
X
2X
. So
we can concentrate on the consistency of the rst set only. The dierence between the two
expressions for must vanish, this implies that should satisfy the Dirac-like equation
_
X
+
P
+
= L
+
= 0. (2.27)
This is a consistent equation provided the vector L
+
is null
L
+
L
+
=
_
X
+
P
+
_
2
= 0. (2.28)
This is indeed correct in the physical sector that satises X
2
= P
2
= X P = 0. Then we
can solve for by writing
=
_
X
+
P
+
_
. (2.29)
This satises the equation (2.27) automatically for any
s
d
4

s
d
4
matrices , . These in
turn are determined, up to
s
d
4

s
d
4
gauge transformations, by satisfying the
s
d
4

s
d
4
matrix
equation (2.20).
20
In this way we have seen that all of the forms given in Eq.(2.26) are consistent with
each other and determine Z
a
A
up to a gauge transformation. Thus the null conditions in
Eq.(2.24) is all that is needed to determine Z
a
A
up to a gauge transformation, but in turn
these followed from the basic relation in Eq.(2.19).
The following subset of our relations resemble the twistor transform in four dimensions,
but , are
s
d
2

s
d
4
matrices that must obey all the relations above
= i
X
2X
+
, and
=
1
2
_
X
+
P
+
. (2.30)
Indeed we can check directly that by inserting only these relations into the right hand
side of Eqs.(2.19,2.17), we derive the SO(d, 2) generators in terms of phase space L
MN
=
X
M
P
N
X
N
P
N
that appear on the left side of those equations. Furthermore by inserting
only these relations into the twistor action we derive the phase space action that determines
the canonical structure
4
s
d
_
d i
Z
a
A
Z
A
a
= i
4
s
d
_
d Tr
_

_
=
4
s
d
_
d Tr
_
_
X
2X
+
_
(2.31)
=
4
s
d
_
d
1
2
_
X
+
P
+
_
Tr
_
_
X
2X
+
_
(2.32)
=
_
d
_
X
+
P
+
_
X
X
+
_
(2.33)
=
_
d
_
P
P
+
X
+
__
X
+
X
+
_
(2.34)
=
_
d
_
X
+
P
+
_
=
_
d
X
M
P
M
.
(2.35)
The last line follows when the constraints X
2
= P
2
= X P = 0 are satised in the physical
sector. This shows the consistency of our twistor transform of Eq.(2.30) for spinless particles
in all dimensions.
It is also interesting to give an explicit formula for both and in terms of the Sp(2, R)
gauge invariant L
MN
. This is already obtained through the relation between Z and L given
in Eq.(2.23). It turns out that it is sucient to take only one of the
i
to be nonzero. So
we will take
2
=
3
=
4
= 0 and determine
1
,= 0 from the relation J = L = Z

Z. The
other possibilities are gauge equivalent. The equivalence is guaranteed by the Sp(2, R) gauge
invariance conditions X
2
= P
2
= X P = 0. This gives
Z
a
A
=
_
_
= L
_
1
0
0
0
_
(2.36)
21
We write out the L = J in Eq.(2.19) more explicitly in terms of
s
d
4

s
d
4
blocks by using the
gamma matrices in footnote (6), and we obtain
=
_
L
+
+ L
r0
r
+ iL
12
+
1
2
L
rs
rs
L
01
iL
02
(L
r1
+ iL
r2
)
r
_
hu, (2.37)
=
_
L
+
0
+ L
+
r
L
+
1
+ iL
+
2
_
i
2hu. (2.38)
Here
1
= hu is written as a product of a unitary matrix u and a Hermitian matrix h. The
matrix u is an arbitrary
s
d
4

s
d
4
unitary matrix that belongs to the gauge group that acts on
the twistors. The generator of this gauge group is the constraint
_
ZZ
_
b
a
= 0. The matrix h
is determined by insuring
=
1
2
L
+
that was established in Eq.(2.20) and is given by

h = 2
3/4
_
L
+
0
+ L
+
r
_
1/2
. (2.39)
To summarize, in Eqs.(2.37-2.39) we have given a more fundamental form of the twistor
transform between Z
a
A
and the (d + 2) dimensional phase space
_
X
M
, P
M
_
. The formulas
are Sp(2, R) gauge invariant since (X, P) appear only in the form L
MN
= X
M
P
N
X
N
P
M
.
These transforms are consistent with Eqs.(2.30,2.26,2.24) which somewhat resemble the more
traditional form of the twistor transform.
Note that Z (L) written in terms of L
MN
depends only on 2 (d 1) independent com-
binations of the L
MN
, since the L
MN
obey the constraints L
MN
L
NK
= 0 that follow from
X
2
= P
2
= X P = 0 in the physical sector. From these we derive the explicit relations
_
L
+
_
2
= L
+
, L
=
L
+
L
+
L
+
, L
+
L
+
= 0 = L
.
So, all L
MN
are then written only in terms of the 2d vector components L
, but those are

lightlike vectors and therefore contain only 2 (d 1) independent degrees of freedom.
Both the 2T particle and the corresponding twistors are SO(d, 2) covariant descriptions,
the rst is in terms of vectors X
M
i
=
_
X
M
, P
M
_
and the second is in terms of spinors Z
a
A
of
SO(d, 2). This covariance is achieved by having gauge symmetries in both versions, in the
rst case the gauge symmetry is applied on the i index of X
M
i
and in the second case the
gauge symmetry is applied on the a index of Z
a
A
.
To account for the gauge symmetry and the constraint (2.16) we may derive them from a
twistor action principle. This is done by rewriting the twistor action in Eq.(2.14) in a form
similar to Eq.(1.7)
S
twistor
=
4
s
d
_
d Tr
__
i
ZDZ
_
h
d
V
, (DZ
A
)
a
Z
a
A
iZ
b
A
V
a
b
. (2.40)
22
The equation of motion of the
s
d
4
s
d
4
matrix gauge eld V
a
b
generates the constraint
_
ZZ
_
b
a

h
d
b
a
= 0 where h
d
is chosen to make the matrix traceless or with trace depending on the
number of dimensions d (see e.g. the counting of degrees of freedom for d = 5, 6 following
Eq.(2.16)). Equivalently, the matrix V itself can be taken as traceless or with trace depending
on the dimension, and this will result in the same constraint. Once this constraint is satised
this action reduces to the previous one in Eq.(2.14). For d = 3, 4, 5, 6 this is the full action
principle in terms of twistors since there are no other conditions (other than reality or
pseudo-reality in some dimensions) as discussed following Eq.(2.16). However, for d 7
some more conditions on a general complex or (pseudo)real Z are required to make it satisfy
also the basic relation Eq.(2.17) or its equivalent null conditions in Eq.(2.24). Although we
have given the full twistor transform for any dimension, we have so far given the full action
principle only for d 6.
Up to now we have described the twistor transform for the free 2T particle in d + 2
dimensions. But from here it is an easy step to obtain the twistor transform for an assortment
of non-trivial particle dynamics in 1T-physics. Having established the transform between
twistors and the Sp(2, R) doublets
_
X
M
, P
M
_
in (d + 2) dimensions, we can now make
Sp(2, R) gauge choices to produce various dynamical systems in (d 1) + 1 dimensions,
in the physical sector that satises X
2
= P
2
= X P = 0. Examples that occur in this
paper are the massless particle in d dimensions of Eq.(1.22,1.23), or the massive particle in
d dimensions of Eq.(1.25,1.26). Other Sp(2, R) gauge choices that include interacting and
curved background cases are found in [4][5]. By inserting the gauge choice for
_
X
M
, P
M
_
into
Eqs.(2.30,2.26) we obtain the corresponding twistors in d dimensions, such as the twistors for
the massless particle of Eq.(1.8), or the massive particle in of Eq.(1.14). For more examples
see [7] where the computations for the twistor transform were done explicitly in d = 4, but
the same explicit formulas also apply in d dimensions by inserting the corresponding gamma
matrices in d dimensions, as given in the expressions above.
C. Geometry: twistors as the coset SO(d, 2) /H
A geometric view of twistors in d dimensions can also be given in the form of a coset
as follows. The starting point for twistors was the twistor gauge of 2T-physics in Eq.(2.14)
which involved the group element g () in the spinor representation of SO(d, 2) and the
special matrix in Eq.(2.13). The action and its SO(d, 2) global symmetry charge (on the
left side of g) have the form
S (g) =
4
s
d
_
d Tr
_
ig
1
g
_
, (J)
B
A
=
_
gg
1
_
B
A
. (2.41)
23
The action is like a sigma model, but it is linear instead of being quadratic in the Cartan
connection ig
1
g, and has the special insertion on the right side of g. The insertion
determines important properties of this action. The equation of motion for g is [, g
1
g] =
0. Using this one can show that the global current is conserved
(gg
1
) = 0, as expected
from Noethers theorem.
We recall that the current J
B
A
is gauge invariant and contains all the physical information
of the theory as seen in the previous sections. In particular the current satises J
2
=
(gg
1
) (gg
1
) = (g
2
g
1
) = 0 at the classical level, which is consistent with the covariant
approach in section (II A). This property of the current captures all the essential aspects of
the physical sector at the classical level.
Now, let us determine the independent degrees of freedom that contribute to the current.
We will nd that there are precisely 2(d 1) degrees of freedom, precisely equal to the
number of physical degrees of freedom. Since g
1
g is a SO(d, 2) transformation applied
on a generator in the algebra of SO(d, 2) , we can eliminate from g () the subgroup H
that leaves invariant, and keep only the coset degrees of freedom in SO(d, 2) /H
. To do
this, we can decompose g () = T
() H
() and write gg
1
= T
T
1
since by denition
H
H
1
= . Here H
= exp (h
) and T
= exp (t
) , where h
(t
) is a linear combination
of all the SO(d, 2) generators
MN
that commute (do not commute) with , i.e. [h
, ] = 0
and [t
, ] ,= 0.
To characterize the sets of generators (h
, t
) consider the decomposition of SO(d, 2)

with respect to the SO(d 2) SO(2, 2) subgroup. We can write J
MN
= J
ij
J
J
i
where i = 1, 2, , (d 2) spans the SO(d 2) basis and = +
, +, (or 0
, 0, 1
, 1)
spans the SO(2, 2) basis. Furthermore we decompose SO(2, 2) =SL(2, R)
+
SL(2, R)
and note that each index is in the

_
1
2
,
1
2
_
representation of SL(2, R)
+
SL(2, R)
. The
SL(2, R)
+
SL(2, R)
generators can be identied explicitly as

SL(2, R)
+
:
1
2
_
J
+
+ J
+
_
, J
+
+
, J
(2.42)
SL(2, R)
:
1
2
_
J
+
J
+
_
, J
+
, J
+
(2.43)
From the general SO(d, 2) commutation rules
_
J
MN
, J
KL
= i
__
J
ML
NK
(M N)
_
(K L)
, (2.44)
with
+
=
+
= 1 and
ij
=
ij
, it is easy to verify that these indeed form the
SL(2, R)
+
SL(2, R)
algebra. Under commutation with the generator

1
2
_
J
+
+ J
+
_
each
SO(d, 2) generator has a denite charge 0,
1
2
, 1. We list the generators according to those
24
charges as follows
1
1
2
0 +
1
2
+1
J
_
J
i
J
i
_
J
ij
,
J
+
+J
+
2
_
J
+
,
J
+
J
+
2
, J
+
_
_
J
+
i
J
+i
_
J
+
+
(2.45)
These charges are conserved additively in the general commutation rules
_
J
MN
, J
KL
=
given above. Furthermore the charge
1
2
generators form doublets under SL(2, R)
and
vectors under SO(d 2) as indicated, while the charge 1 generators are singlets under
both. From this structure of the commutation rules
_
J
MN
, J
KL
= we easily see that

the generators that commute or do not commute with
(or J
) are
h
: J
,
_
J
i
J
i
_
,
J
ij
_
J
+
,
J
+
J
+
2
, J
+
_
(2.46)
t
:
1
2
_
J
+
+ J
+
_
,
_
J
+
i
J
+i
_
, J
+
+
(2.47)
We note that each set forms a subalgebra [h
, h
] h
, [t
, t
] t
while [h
, t
] h
+t
.
In particular, within t
the generator
1
2
_
J
+
+ J
+
_
forms a U(1) subgroup and classies
the others according to their charges above, while the remaining coset T
/U (1) forms an
algebra similar to the Heisenberg algebra,
_
J
+
i
, J
+j
= i
ij
J
+
+
, since J
+
+
commutes with
both J
+
i
, J
+j
. From this we conclude that the general T
= exp (t
) can be parametrized
as follows
T
= exp (t
) = exp
_
1
2
_
+
+
_
()
_
exp
_
+
z ()
_
exp
_
+i
k
i
() +
+
i
q
i
()
_
(2.48)
The number of parameters in this coset is precisely 2 (d 1) , which is the same as the
number of physical degrees of freedom. If we insert the explicit set of gamma matrices
MN
given in footnote (6) into this expression, we can write T
in the form of
s
d
4

s
d
4
similar to
. Then, as seen from Eq.(2.14), the rst block of columns that forms an s
d
s
d
4
rectangular
matrix is the twistor Z
a
A
(t
) now written in terms of only the 2 (d 1) parameters of the

coset t
SO(d, 2) /H
.
This result is of course in agreement with the form of the Sp(2, R) invariant Z
a
A
(L)
of Eqs.(2.37,2.38), which can also be written only in terms of 2 (d 1) parameters as de-
tailed in Sec.(??). We now understand that there is a close relationship with the geometric
interpretation as a coset.
25
III. D-BRANES AND TWISTORS
In the discussion following the twistor action in Eq.(2.40) we explained that more condi-
tions are needed on Z in order to obtain the proper twistor that is equivalent to the phase
space of a particle when d 7. What if those conditions are never imposed? What would
then be the content of Z
a
A
? We nd that the extra degrees of freedom can be interpreted
as collective coordinates of D-branes.
To see this let us consider the properties of the rectangular matrix Z
a
A
that follow from the
action in Eq.(2.40). The gauge group acts on the right side and there is a global symmetry
with conserved charges J that acts on the left side. These properties are summarized by the
equations
ZZ = 0, J = Z

Z trace. (3.1)
Thus, the global current J is a s
d
s
d
matrix in the fundamental representation of the global
group G that can be expanded in a complete set of SO(d, 2) gamma matrices as follows
J = Z

Z = J
0
+
M
J
M
+
1
2
MN
J
MN
+
1
3!
MNK
J
MNK
+ (3.2)
There are as many terms
M
1
Mn
as necessary to span all the generators of some group
G whose fundamental representation has the same dimension s
d
of the SO(d, 2) spinor.
If the group is constrained to be SO(d, 2) there is only the term proportional to
MN
in
the expansion (3.2), and then the group element g is constructed by exponentiating the
generators as in Eq.(2.1). But if the group is more general, then the exponent in Eq.(2.1)
contains all the terms that appear in Eq.(3.2).
Thus, with a more general g represented as a s
d
s
d
matrix, a more general twistor Z
would emerge, with more degrees of freedom than the particle phase space. This is achieved
with an action that is written in the form of Eq.(2.41), which in turn is obtained by gauge
xing from the 2T-physics action in (2.2), but by taking g to be a group element not just
in SO(d, 2) , but an element in the smallest group G that contains SO(d, 2) in the spinor
representation. The parent 2T-physics theory in Eq.(2.2) has an interaction term of the
form of
4
s
d
Tr (ig
1
gL) which is unchanged. But we emphasize that now L is proportional

to only
MN
, while ig
1
g has all the terms in Eq.(3.2), so L couples to only the SO(d, 2)

subgroup of G. Then the action in Eq.(2.2) still has local Sp(2, R) and SO(d, 2) symmetries,
but now it has global G symmetry instead of only global SO(d, 2) on the left side of g.
This generalization of the group element g allows Z to contain the extra degrees of free-
dom. We emphasize that the spinor representation of SO(d, 2) , whose dimension is s
d
, must
correspond to the fundamental representation of G. This requirement determines G as we
see in Table 1 below.
26
We already know that for d = 3, 4, 6 the groups G =Sp(4, R), SU(2, 2), Spin(6, 2) re-
spectively are exactly equal to SO(d, 2) in the spinor representation. Therefore for these
cases there are no other terms in Eq.(3.2) other than
MN
, provided some (pseudo)reality
conditions are imposed as given following Table 1 below. Extra terms usually appear for
d 7.
As an example of what terms appear, consider SO(7, 2) for d = 7. The spinor represen-
tation has dimension 16. The smallest group with a 16 dimensional fundamental represen-
tation is SO
(16), where the

indicates the appropriate analytic continuation that contains
SO(7, 2) as a subgroup. The number of generators of SO
(16) is
1615
2
= 120. The number
of generators represented by the 1616 gamma matrices is
M
9,
MN
98
2
= 36,
MNK
987
123
= 84,
MNKL
9876
1234
= 126. We see that the 120 generators of SO
(16) are
represented by
1
2
MN
L
MN
+
1
3!
MNK
L
MNK
. Therefore the more general twistor in d = 7 has
the expansion
d = 7 : J = Z

Z =
1
2
L
MN
MN
+
1
3!
L
MNK
MNK
. (3.3)
If we impose additional conditions on Z as in the previous sections, then we eliminate the
term L
MNK
= 0, and the remaining L
MN
necessarily satises all the conditions of section
(II B) since they all followed from

ZZ = 0 that was imposed by the gauge symmetry H.
However, if we do not impose the conditions of section (II B) then we can interpret the
degrees of freedom L
MNK
as D-brane degrees of freedom. To see this, consider the smallest
extended super-conformal algebra that contains spin(7, 2) SO
(16). This is OSp(16[2) .

Its two supercharges satisfy
_
Q
i
A
, Q
j
B
_
=
ij
_
1
2
_
MN
_
AB
L
MN
+
1
3!
_
MNK
_
AB
L
MNK
+
q
ij
C
AB
, with i = 1, 2 labelling the Sp(2) , and C
AB
, q
ij
both symmetric. The usual
d = 7 Poincare super-algebra is a subalgebra obtained from the above by decomposing
the SO(7,2)SO(6,1)SO(1,1) for spinors (16 =8
+
+8
) as A =
+

and vectors as
M =
, , and keeping only the operators Q

i
+
and all the L
MN
, L
MNK
with a single +
as
follows
_
Q
1
+
, Q
2
+
_
= L
+
(
+
+
1
2
L
+
(
+
. (3.4)
In the massless particle gauge, L
+
is the momentum p
and then L
+
are the commuting

D2-brane charges in d = 7 dimensions (like generalized momenta). The other components
of L
MN
, and L
MNK
are functions of phase space, including the particle as well as D-brane
canonical degrees of freedom, and do not generally commute among themselves
8
.
As another example consider SO(8, 2) for d = 8. The two spinor representations are 16,16.
The smallest groups with 16 dimensional fundamental representations are SO
(16), Sp
(16),
8
The commutation rules of the L
M1Mn
are isomorphic to the commutation rules of the
M1Mn
. From
this we see that the D-brane charges L
+
commute among themselves as well as with the momenta

p
= L
+
.
27
SU
(16) . To decide which is the smallest one that contains SO(8, 2) as a subgroup we analyze
the number of generators represented by the gamma matrices. The number of generators
of SO
(16) is
1615
2
= 120, for Sp
(16) is
1617
2
= 136, and for SU
(16) is (16)
2
1 = 255.
The number of generators represented by the gamma matrices is
M
10,
MN
109
2
=
45,
MNK
1098
123
= 120,
MNKL
10987
1234
= 210,
MNKLR
+

1
2
109876
12345
= 126, where the
last one is self dual (hence the extra factor of
1
2
). The 45 generators of SO(8, 2) represented
by
MN
must be included as one of the criteria in choosing the smallest G. Then we see
that there is no combination of gamma matrices that can be used to construct SO
(16) and
Sp
(16) and therefore we must take SU
(16) as the smallest group that contains spin(8, 2) .

The smallest superconformal algebra is SU(16[1) . The 255 generators of SU
(16) that appear

in
_
Q
A
,

Q
B
_
are then represented by
1
2
MN
L
MN
+
1
4!
MNKL
L
MNKL
. The extra L
MNKL
lead
to the D-brane degrees of freedom. To see the content of D-brane commuting
8
charges we
must decompose SO(8, 2) to SO(7, 1) by M =
, and identify the D-brane commuting

charges as the D3-brane Z
+
in 8 dimensions.
d Spin(d, 2) spinor G G
super
(N) generators of G in Spin(d, 2)basis
contained
in product
3 Spin(3, 2) 4 Sp(4, R) OSp(N[4)
MN
10
(4 4)
s
4 Spin(4, 2) 4,
4 SU(2, 2) SU(2, 2[N)

MN
15
4
4
5 Spin(5, 2) 8
+
spin
(7)
SO
(8)
F(4)
OSp(8|2N)
MN
21
MN
21

M
7
(8 8)
a
6 Spin(6, 2) 8
+
SO
(8) OSp(8[2N)
MN
28
(8 8)
a
7 Spin(7, 2) 16 SO
(16) OSp(16[2N)
MN
36

MNK
84
(16 16)
a
8 Spin(8, 2) 16, 16 SU
(16) SU(16[N)
MN
45

MNKL
210
16 16
9 Spin(9, 2) 32 Sp
(32) OSp(N[32)
MN
55

M
11

M
1
M
5
462
(32 32)
s
10 Spin(10, 2) 32
+
Sp
(32) OSp(N[32)
MN
66

M
1
M
6
+ 462
(32 32)
s
11 Spin(11, 2) 64 Sp
(64) OSp(N[64)
MN
78

MNK
286

M
1
M
6
1716
(64 64)
s
12 Spin(12, 2) 64,64 SU
(64) SU(64[N)
MN
91

MNKL
1001

M
1
M
6
3003
64 64
Table 1: Smallest groups G and supergroups G
super
that contain Spin (d, 2) , and D-branes.
In Table-1 we give a list of the smallest groups G that contain spin(d, 2) for 3 d 12. We
also include the smallest supergroup G
super
that contains G. We list the gamma matrix repre-
sentation of the generators of G, and their numbers as subscripts, as represented by antisym-
metrized products of gamma matrices
M
1
Mn
1
n!
_
M
1
M
2
M
3

Mn
permutations
_
in
dimension d + 2 labelled by M. The last column gives information on whether the gamma
matrices occur in the symmetric or antisymmetric products of the spinors of SO(d, 2), when
both spinor indices A, B are lowered or raised in the form
_
M
1
Mn
_
AB
by using the metric
C in spinor space. The phase space of the D-branes correspond to the extra generators
beyond
MN
as explained in the examples above. In the case of d = 5, one option is to
keep the D0-brane associated with
M
, another option is to remove it with the extra

28
U(1) gauge symmetry as discussed in the counting done in section (II B). The D-brane does
not occur for the supergroup F(4) that can be used for the d = 5 superparticle as described
below.
Groups that are larger than the listed G may be considered in our scheme in every
dimension (e.g. SU(8) instead of SO(8) in d = 6, etc.). In that case the number of generators
M
1
Mn
increases compared to the ones listed in the table for each d. Furthermore the
corresponding D-brane degrees of freedom also get included in the model.
When Z
a
A
is obtained from the group element g through the relation Z

Z = gg
1
, with
the group G listed in the table above, then Z is real or pseudo-real when the group is SO or
Sp and it is complex when the group is SU. Given those properties, in general the quadratic
_
Z

Z
_
B
A
contains just the gamma matrices listed above which correspond to the generators
of the group G. If the (pseudo)reality properties associated with G are not obeyed by Z then
more D-brane terms will appear generally in the expansion of Z

Z as in Eq.(3.2) as compared
to those on the table.
The case of d = 11 is particularly interesting since it relates to M-theory as follows. The
corresponding twistors are spinors of Spin(11, 2) that are 64 dimensional. The smallest group
is Sp
(64) whose generators are represented by

MN
(78) +
MNK
(286) +
M
1
M
6
(1716) .
To identify the commuting charges of D-branes we decompose M =
, and keep all the

generators with a single +
, as follows L
+
L
+
L
+
5
. Here L
+
is the momentum
in 11 dimensions and L
+
, L
+
5
are the D2-brane and D5-brane commuting
8
charges
respectively.
Let us mention that the discussion above with the group G can be directly generalized
to the supergroup G
super
(N) listed in Table 1 (with some limits on N as discussed below)
by following [6]. The 2T-physics action is still of the same form as S (X, P, g) of Eq.(2.2),
but now we have a supergroup element g and a supertrace coupling
4
s
d
Tr (ig
1
gL) , and
the matrix L is of the form L =
1
4i
_
MN
0
0
0
_
L
MN
where the gamma matrices couple to
the bosonic subgroup G as above. This 2T superparticle action reduces to the standard
massless superparticle action in the particle gauge for dimensions d = 3, 4, 5, 6 [6] with N
supersymmetries. It can also be gauge xed to the twistor gauge to give supertwistors that
are equivalent to the super phase space in those dimensions [12][13]. One can go beyond those
gauge choices and obtain a twistor description of many other dual superparticle theories that
give the super generalizations of the ones studied recently in [7].
The supergroup can be enlarged to have more fermionic generators, but keeping G as a
bosonic subgroup. For example, for d = 4 we may take SU(2, 2[N) instead of the smallest
N = 1 shown in the table. For physical purposes the total number of real fermionic generators
cannot exceed 64 (32 ordinary supercharges and 32 conformal supercharges). For example,
for d = 4 we can go as far as N = 8, or G
super
=SU(2, 2[8) which has 64 real fermionic
29
parameters. Similarly, for d = 10 we may take OSp(1[32) or OSp(2[32) . In the more general
cases the coupling L can be of the form of the previous paragraph L =
1
4i
_
MN
0
0
0
_
L
MN
.
The model can also be generalized by adding d
more dimensions
_
X
I
, P
I
_
in addition to
the d + 2 dimensions
_
X
M
, P
M
_
, but keeping the same g G
super
. The generalized action
has the form [18][12][13][17]
S
2T
_
X,

P, g
_
=
_
d
_
1
2
ij

X
i

X
j
1
2
A
ij

X
i

X
j
+
4
s
d
Str
_
ig
1
L
_
_
(3.5)
where

X

M
=
_
X
M
, X
I
_
,

P

M
=
_
P
M
, P
I
_
, and we now take the more general coupling
L =
1
4i
_
MN
0
0
0
_
L
MN
+

4i
_
0
0
0
IJ
_
L
IJ
. The xed parameter =
s
d
s
d
is determined by local
bosonic and fermionic symmetries in this action, to be the ratio of the spinor dimensions
of SO(d + 2) and SO(d
) . In this latter scheme we obtain interesting cases, such as super-

twistors with some compactied subspaces, without D-branes. For example supertwistors for
AdS
4
S
7
, AdS
5
S
5
, AdS
7
S
4
with a of total 10 or 11 dimensions emerge [12][13][17] by us-
ing supergroups with only 32 real fermions, namely G
super
=OSp(8[4
), SU(2, 2[4), OSp(8
[4)
respectively.
This analysis taken to the maximum allowed number of supersymmetries and the maxi-
mum number of dimensions leads to d = 11 with OSp(1[64) as the hidden global supersym-
metry for M-theory, and suggests that the extended supertwistors may well play a role in
a nice description of M-theory. This can be studied through the toy M-model [19] that has
the action S (X, P, g) of Eq.(2.2) with the group OSp(1[64) , and includes the D2 and D5-
branes. The supergroup OSp(1[64) is motivated by other considerations as well [20][21][22].
In other approaches to twistors in D=11 [23][24], twistors in the fundamental representation
of OSp(1[64) were used in [24] for the formulation of a superstring action in an extended
D=11 superspace.
The twistor action (2.40) is a gauge xed form of the 2T-physics action S (X, P, g) of
Eq.(2.2) for a general g. We can play the game of gauge xing the local symmetries Sp(2, R)
and SO(d, 2) of S (X, P, g) in many possible ways and derive a multitude of 1T-physics
systems with a rich web of dualities among them. D-branes are included in this web of
dualities. Then twistors can be shown to unify many dual theories including D-branes. It
would be interesting to pursue this line of reasoning in more detail.
IV. DISCUSSION
In this paper we gave the general twistor transform that maps twistor space to phase
space in d dimensions with one time. The general transform can be specialized to a variety
of special dynamical particle systems that include particles with or without mass, relativistic
30
or nonrelativistic, in at or curved spaces, interacting or non-interacting. Thus, the scope
of our formulas is much larger than the traditional twistor transform. The special cases of
phase space described by the same twistor are those that can be derived by gauge xing
the parent unifying theory in 2T-physics. Thus, either the twistor description or the vector
SO(d, 2) description in 2T-physics provide a unication of those 1T-physics systems and
establishes a duality relationship between them.
To our knowledge this is the rst time that twistors have been successfully dened gen-
erally in d dimensions. We have insured that our twistor transform is fully equivalent to
particle phase space for all dimensions. If we specialize to the phase space of massless par-
ticles only, then our result agrees with twistors that were previously dened for d 6 in
another approach [25]. Even for d 6 our twistors for the phase spaces other than the mass-
less particle are all new structures. For a rather dierent approach to twistors for massive
particles, which uses double the number of twistors compared to our formulas and only in
d = 4, see [3],[26]-[30].
Beyond twistors for particles, we have also dened twistors for a phase space that includes
also D-brane degrees of freedom. Including the D-branes may lead to some interesting
applications of twistors, in particular for M-theory. The twistor action principle in Eq.(2.14)
applies generally to twistors including D-branes for a generally complex Z in every dimension.
If (pseudo)reality conditions are imposed on Z as mentioned after the table in the previous
section, then for d 6 we obtain only the particle phase space out of the twistor. In d 7
there are automatically D-branes even with the (pseudo) reality conditions. However, if
extra constraints are applied on Z, as detailed in section (II B), then again the degrees of
freedom in Z are thinned down to only the particle phase space without D-branes in every
dimension.
Quantization of twistors for any spin in four dimensions was discussed in section (I B).
Here we suggested a free eld theory in twistor space that describes any spinning parti-
cle. This could also lead to some interesting applications of twistors in eld theory with
interactions in four dimensions.
Generalizations of our results in many directions are possible. Some of these are already
briey described in recent papers [12][13], such as twistors for spinning particles, super-
twistors in various dimensions, including compactied dimensions, and supertwistors for
supersymmetric AdS
5
S
5
, AdS
4
S
7
, AdS
7
S
4
. We plan to give details of those structures
in future publications.
It must be emphasized that in all cases the underlying theory is anchored in 2T-physics,
and therefore by gauge xing the Sp(2, R) gauge symmetry these twistors describe not only
massless systems, but much more, as discussed in this paper and [7] with examples. Hence
the twistors play a role in some kind of unication of 1T-physics systems via dualities, or via
31
higher dimensions with 2T, but in a way that is distinctly dierent than the Kaluza-Klein
scheme, since there are no Kaluza-Klein excitations, but instead there is a web of dualities.
In this context it is also interesting that some parameters such as mass, moduli of some
metrics, and some coupling constants for interactions, emerge from the higher dimensions
as moduli while holographically projecting from d + 2 dimensions down to d dimensions.
Furthermore concepts such as time and Hamiltonian in 1T-physics are derived concepts
that emerge either from 2T-physics and its gauge choices, or from the details of the twistor
transform to 1T-physics systems.
Acknowledgments
I. Bars was supported by the US Department of Energy under grant No. DE-FG03-
84ER40168; M. Picon was supported by the Spanish Ministerio de Educacion y Ciencia
through the grant FIS2005-02761 and EU FEDER funds, the Generalitat Valenciana and
by the EU network MRTN-CT-2004-005104 Constituents, Fundamental Forces and Sym-
metries of the Universe. M. Picon wishes to thank the Spanish Ministerio de Educacion y
Ciencia for his FPU research grant, and the USC Department of Physics and Astronomy for
kind hospitality.
[1] I. Bars, B. Orcal, and M. Picon, in preparation, I. Bars and B. Orcal, in preparation.
[2] R. Penrose, Twistor Algebra, J. Math. Phys. 8 (1967) 345; Twistor theory, its aims and
achievements, in Quantum Gravity, C.J. Isham et. al. (Eds.), Clarendon, Oxford 1975, p.
268-407; The Nonlinear Graviton, Gen. Rel. Grav. 7 (1976) 171; The Twistor Program,
Rept. Math. Phys. 12 (1977) 65.
[3] R. Penrose and M.A. MacCallum, An approach to the quantization of elds and space-time,
Phys. Rept. C6 (1972) 241; R. Penrose and W. Rindler, Spinors and space-time II, Cambridge
Univ. Press (1986).
[4] I. Bars, C. Deliduman and O. Andreev, Gauged Duality, Conformal Symmetry and Space-
time with Two Times , Phys. Rev. D58 (1998) 066004 [arXiv:hep-th/9803188]. For reviews
of subsequent work see: I. Bars, Two-Time Physics , in the Proc. of the 22nd Intl. Col-
loq. on Group Theoretical Methods in Physics, Eds. S. Corney at. al., World Scientic 1999,
[arXiv:hep-th/9809034]; Survey of two-time physics, Class. Quant. Grav. 18, 3113 (2001)
[arXiv:hep-th/0008164]; 2T-physics 2001, AIP Conf. Proc. 589 (2001), pp.18-30; AIP Conf.
Proc. 607 (2001), pp.17-29 [arXiv:hep-th/0106021].
[5] I. Bars, Conformal symmetry and duality between free particle, H-atom and harmonic oscilla-
tor, Phys. Rev. D58 (1998) 066006 [arXiv:hep-th/9804028]; Hidden Symmetries, AdS
d
S
n
,
32
and the lifting of one-time physics to two-time physics, Phys. Rev. D59 (1999) 045019
[arXiv:hep-th/9810025].
[6] I. Bars, 2T physics formulation of superconformal dynamics relating to twistors and super-
twistors, Phys. Lett. B 483, 248 (2000) [arXiv:hep-th/0004090].
[7] I. Bars and M. Picon, Single twistor description of massless, massive, AdS, and other inter-
acting particles, arXiv:hep-th/0512091.
[8] E. Witten, Perturbative gauge theory as a string theory in twistor space, Commun. Math.
Phys. 252 (2004) 189 [arXiv:hep-th/0312171]; Parity invariance for strings in twistor space,
hep-th/0403199.
[9] N. Berkovits, An Alternative string theory in twistor space for N=4 Super Yang-Mills, Phys.
Rev. Lett. 93 (2004) 011601 [arXiv:hep-th/0402045].
[10] N. Berkovits and L. Motl, Cubic twistorial string eld theory, JHEP 0404 (2004) 56,
[11] N. Berkovits and E. Witten, Conformal supergravity in twistor-string theory, JHEP 0408
(2204) 009, [arXiv:hep-th/0406051].
[12] I. Bars, Twistor superstring in 2T-physics, Phys. Rev. D70 (2004) 104022,
[13] I. Bars, Twistors and 2T-physics, AIP Conf. Proc. 767 (2005) 3 , [arXiv:hep-th/0502065].
[14] F. Cachazo, P. Svrcek and E. Witten, MHV vertices and tree amplitudes in gauge the-
ory, JHEP 0409 (2004) 006 [arXiv:hep-th/0403047]; Twistor space structure of one-
loop amplitudes in gauge theory, JHEP 0410 (2004) 074 [arXiv:hep-th/0406177]; Gauge
theory amplitudes in twistor space and holomorphic anomaly, JHEP 0410 (2004) 077
[15] For a review of Super Yang-Mills computations and a complete set of references see:
F.Cachazo and P.Svrcek, Lectures on twistor strings and perturbative Yang-Mills theory,
PoS RTN2005 (2005) 004, [arXiv:hep-th/0504194].
[16] I. Bars, Two time physics with gravitational and gauge eld backgrounds, Phys. Rev. D62,
085015 (2000) [arXiv:hep-th/0002140]; I. Bars and C. Deliduman, High spin gauge elds and
two time physics, Phys. Rev. D64, 045004 (2001) [arXiv:hep-th/0103042].
[17] I. Bars, Lectures on Twistors, hep-th/0601091.
[18] I. Bars, Hidden 12-dimensional structures in AdS
5
x S
5
and M
4
x R
6
supergravities, Phys.
Rev. D 66, 105024 (2002) [arXiv:hep-th/0208012]; A mysterious zero in AdS
(
5) x S
5
super-
gravity, Phys. Rev. D 66, 105023 (2002) [arXiv:hep-th/0205194].
[19] I. Bars, Toy M-model, to be published. For a brief discussion see [21] and the second review
paper in [4].
[20] I. Bars, S Theory , Phys. Rev. D55 (1997) 2373 [arXiv:hep-th/9607112].
[21] I. Bars, C. Deliduman and D. Minic, Lifting M-theory to Two-Time Physics , Phys. Lett.
33
B457 (1999) 275 [arXiv:hep-th/9904063].
[22] P. West, Hidden superconformal symmetry in M theory, JHEP 0008 (2000) 007
[23] I.A. Bandos, J.A. de Azcarraga, J.M. Izquierdo, J. Lukierski, Phys. Rev. Lett. 86 (2001) 4451
[24] I.A. Bandos, J.A. de Azcarraga, M. Pic on and O. Varela, Phys. Rev. D69 (2004), 085007
[25] I. Bengtsson and M. Cederwall, Particles, Twistors And The Division Algebras, Nucl. Phys.
B302 (1988) 81. M. Cederwall, AdS twistors for higher spin theory, AIP Conf. Proc. 767
(2005) 96 , [arXiv:hep-th/0412222].
[26] Z. Perjes, Rep. Math. Phys. 12, 193 (1977).
[27] L.P. Hughston, Twistors and Particles, Lecture Notes in Physics 97, Springer Verlag, Berlin
(1979).
[28] P.A. Tod, Rep. Math. Phys. 11, 339 (1977).
[29] S. Fedoruk and V.G. Zima, Bitwistor formulation of massive spinning particle,
arXiv:hep-th/0308154.
[30] A. Bette, J.A. de Azcarraga, J.Lukierski and C. Miquel-Espanya, Massive relativistic par-
ticle model with spin and electric charge from two-twistor dynamics, Phys. lett.B595, 491
(2004), [arXiv:hep-th/0405166]; J.A.de Azcarraga, A.Frydryszak, J.Lukierski and C.Miquel-
Espanya, Massive relativistic particle model with spin from free two-twistor dynamics and
its quantization, arXiv:hep-th/0510161.
34

Itzhak Bars and Moises Picon - Twistor Transform in D Dimensions and A Unifying Role For Twistors

Uploaded by

Copyright:

Available Formats

Itzhak Bars and Moises Picon - Twistor Transform in D Dimensions and A Unifying Role For Twistors

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Itzhak Bars and Moises Picon - Twistor Transform in D Dimensions and A Unifying Role For Twistors

Uploaded by

Copyright:

Available Formats

a

= 0, and hence they are physical observables. So the L

that are SO(3, 1) vectors,

, each described by two complex degrees of

. The spinor is subject to a SU(2, 2)

= 2h, where h is the helicity of the

= 0. First we count degrees of freedom. The

are expanded in terms of the Pauli matrices

is a 22 matrix of rank one, its determinant vanishes. Then

, in the gauge invariant

show that (Z) is the helicity +h wavefunction, while

) in a dierent way than

(X P) is dropped from the action. This action

, , and = 0, 1, , (d 1) or = , 1, , (d 2) , with lightcone type denitions of

indicates SO(d 2) vectors. So the target phase space X

() = 0, and solve the two constraints X

) below) in terms of the massive particle

. We will show how it works explicitly in d = 3, 4, 5, 6 and higher dimensions.

C g. In odd number of dimensions the even-dimension

it produces the linear term proportional to d/2

takes the matrix form given in Eq.(2.19). We can further

2 in the form given in Eq.(2.13).

that was established in Eq.(2.20) and is given by

, but those are

) consider the decomposition of SO(d, 2)

and note that each index is in the

generators can be identied explicitly as

algebra. Under commutation with the generator

= we easily see that

) now written in terms of only the 2 (d 1) parameters of the

gL) which is unchanged. But we emphasize that now L is proportional

g has all the terms in Eq.(3.2), so L couples to only the SO(d, 2)

(16), where the

(16). This is OSp(16[2) .

, , and keeping only the operators Q

are the commuting

commute among themselves as well as with the momenta

(16) and therefore we must take SU

(16) as the smallest group that contains spin(8, 2) .

(16) that appear

, and identify the D-brane commuting

4 SU(2, 2) SU(2, 2[N)

, another option is to remove it with the extra

(64) whose generators are represented by

, and keep all the

) . In this latter scheme we obtain interesting cases, such as super-

), SU(2, 2[4), OSp(8

You might also like