a

r
X
i
v
:
g
r
-
q
c
/
0
1
1
0
0
8
6
v
3


1
3

M
a
r

2
0
0
2
WARP DRIVE WITH ZERO EXPANSION
JOS

E NAT

ARIO
Abstract. It is commonly believed that Alcubierres warp drive works by
contracting space in front of the warp bubble and expanding space behind it.
We show that this contraction/expansion is but a marginal consequence of the
choice made by Alcubierre, and explicitly construct a similar spacetime where
no contraction/expansion occurs. Global and optical properties of warp drive
spacetimes are also discussed.
Introduction
Alcubierres warp drive spacetime ([Alc94]) allows eective superluminal travel
within General Relativity. As remarked by Alcubierre himself, it requires nega-
tive energy densities, and there are reasons to believe that all spacetimes allowing
superluminal travel must violate energy conditions ([VBL00], [GW00]). Another
feature is the existence of horizons ([CHL99]), which has been argued on general
grounds to be unavoidable ([Low99]).
The traditional heuristic explanation of how the warp drive spacetime works is
that space in front of a given region (the warp bubble) is contracted, whereas
space behind the same region is expanded. In response, the warp bubble moves
forwards with a speed determined by the contraction/expansion rate.
We will show in this paper that this contraction/expansion is not necessary at
all, and that in particular it is possible to construct a similar spacetime where no
contraction/expansion occurs. Heuristically, one could best describe the warp drive
spacetime as sliding the warp bubble region through space; space in front of the
bubble may get contracted (and space behind it expanded), or not, depending on
the details of the construction.
The paper is divided in three sections.
In the rst section we dene the warp drive spacetime generated by any (time-
dependent) vector eld X in Euclidean 3-space. This is a globally hyperbolic space-
time foliated by Euclidean 3-spaces such that the integral lines of X represent time-
like geodesics (corresponding to the so-called Eulerian observers). We show that
such spacetimes always violate some energy condition, and compute the expan-
sion of the Eulerian observers volume element to be X. Finally, we show how
Alcubierres warp drive can be obtained from a particular choice of X.
In the second section, we explicitly construct a volume-preserving warp drive by
selecting an appropriate divergenceless vector eld X.
In the third section, we take advantage of the formalism developed thus far
to analyse some simple examples, and nd them to possess horizons and in-
nite blueshifts. These results are similar to those obtained (some numerically)
This work was partially supported by FCT (Portugal) through programs PRAXIS XXI and
PROCTI.
1
2 JOS

E NAT

ARIO
in [CHL99], but independent of the particular choice of warp drive, and perhaps
easier to visualize.
1. Warp Drive Spacetimes
Denition 1.1. A warp drive spacetime is a globally hyperbolic spacetime (M, g),
where M = R
4
with the usual Cartesian coordinates (t, x, y, z) (t, x
i
) and g is
given by the line element
1
ds
2
= dt
2
+
3

i=1
(dx
i
X
i
dt)
2
for three unspecied bounded smooth functions (X
i
) (X, Y, Z).
Notice that the Riemannian metric h induced in the Cauchy surfaces {dt = 0}
is just the ordinary Euclidean at metric. A warp drive spacetime is completely
dened by the vector eld
X = X
i

x
i
= X

x
+Y

y
+Z

z
,
which we can think of as a (time-dependent) vector eld dened in Euclidean 3-
space. The future-pointing unit normal covector to the Cauchy surfaces is
n
a
= dt n
a
=

t
+X
i

x
i
=

t
+X.
Denition 1.2. The observers whose 4-velocity is n
a
are said to be the Eulerian
observers.
Proposition 1.3. Eulerian observers are free-fall observers.
Proof. Using the geodesic Lagrangian,
L =
1
2
_

t
2
+
3

i=1
_
x
i
X
i

t
_
2
_
,
it is a simple matter to write down the geodesic equations:
d
d
_
L

t
_

L
t
= 0
d
d
_

t X
i
_
x
i
X
i

t
__
+

t
X
i
t
_
x
i
X
i

t
_
= 0;
d
d
_
L
x
i
_

L
x
i
= 0
d
d
_
x
i
X
i

t
_
+

t
X
j
x
i
_
x
j
X
j

t
_
= 0.
It is immediate to check that any curve satisfying

t = 1, x
i
= X
i
is a solution of
these equations.
Proposition 1.4. The extrinsic curvature tensor associated to the foliation of a
warp drive spacetime by the Cauchy surfaces {dt = 0} is
K =
1
2
_

i
X
j
+
j
X
i
_
dx
i
dx
j
.
1
We shall use the notation and sign conventions in [Wal84], except for the use the latin indices
i, j, k as numerical indices running from 1 to 3 and referring either to the space coordinates (x, y, z)
or to space components in an orthonormal frame including
_

x
,

y
,

z
_
.
WARP DRIVE WITH ZERO EXPANSION 3
Proof. As is well known, the extrinsic curvature tensor is given by
K =
1
2

n
ah =
1
2

t
+X
h.
Now

t
h =
t

ij
dx
i
dx
j
=

ij
t
dx
i
dx
j
= 0.
On the other hand, since X is tangent to the Cauchy surfaces, we can use the usual
formula for the Lie derivative of the metric,

X
h =
_

kj
D
i
X
k
+
ik
D
j
X
k
_
dx
i
dx
j
=
_
D
i
X
j
+D
j
X
i
_
dx
i
dx
j
,
where D stands for the Levi-Civita connection determined by h. Since h is just the
at Euclidean metric, D = and we get the formula above.
Corollary 1.5. The expansion of the volume element associated with the Eulerian
observers is = X.
Proof. We just have to notice that
= K
i
i
=
i
X
i
.
Corollary 1.6. A warp drive spacetime is at wherever X is a Killing vector eld
for the Euclidean metric (irrespective of time dependence). In particular, a warp
drive spacetime is at wherever X is spatially constant.
Proof. Since the Cauchy surfaces are at, all curvature comes from the extrinsic
curvature. Thus the spacetime will be at wherever the extrinsic curvature is zero,
i.e., wherever
X
h = 0.
Theorem 1.7. Non at warp drive spacetimes violate either the weak or the strong
energy condition.
Proof. If the strong energy condition holds and = 0 at some event, then blows
up in nite time (see [LL97]). Since X is nite, the strong energy condition
can only hold if 0. However, it follows from Einsteins eld equations that the
energy density measured by Eulerian observers is (see [Wal84])
= T
ab
n
a
n
b
=
1
16
_
(3)
R + (K
i
i
)
2
K
ij
K
ij
_
=
1
16
_

2
K
ij
K
ij
_
where
(3)
R = 0 is the scalar curvature of the at Cauchy surfaces dt = 0. Thus if
= 0 we have 0, and = 0 i K
ij
0. Consequently if the spacetime does
not violate neither the strong nor the weak energy conditions it must be at.
Example 1.8. Alcubierres warp drive (see [Alc94]) is obtained by choosing
X = v
s
f(r
s
);
Y = Z = 0
with
v
s
(t) =
dx
s
(t)
dt
;
r
s
=
_
(x x
s
)
2
+y
2
+z
2
1
2
,
4 JOS

E NAT

ARIO
for a smooth function f : [0, +) [0, 1] approximating a step function equal to 1
in a neighbourhood of the origin and equal to 0 in a neighbourhood of innity, and
an arbitrary function x
s
(t). If we call the region f(r
s
) = 1 the interior of the warp
bubble and the region f(r
s
) = 0 the exterior of the warp bubble, we see that both
these regions are at; nevertheless, Eulerian observers inside the warp bubble move
with arbitrary speed v
s
with respect to Eulerian observers outside the warp bubble.
It is a simple matter to obtain the expansion of the volume element associated with
the Eulerian observers and the energy density measured by Eulerian observers in
this example:
=
x
X = v
s
f

(r
s
)
x x
s
r
s
;
=
1
16
_
(
x
X)
2
(
x
X)
2
2
_
1
2

y
X
_
2
2
_
1
2

z
X
_
2
_
=
1
32
v
2
s
[f

(r
s
)]
2
y
2
+ z
2
r
2
s
.
It is convenient to replace the x coordinate with
= x x
s
(t).
This eectively corresponds to replacing X with X v
s
, so that the Eulerian
observers inside the bubble stand still whereas the Eulerian observers outside the
bubble move with speed v
s
in the negative -direction. Obviously and retain
their values, but now
r
s
=
_

2
+y
2
+z
2
_1
2
does not depend on the coordinate t.
Denition 1.9. The vector eld X is said to generate a warp bubble with velocity
v
s
(t) if X = 0 for small x (the interior of the warp bubble) and X = v
s
(t) for
large x (the exterior of the warp bubble), where
x = x
i

x
i
= x

x
+y

y
+z

z
.
2. Warp Drive With Zero Expansion
We introduce spherical coordinates (r, , ) in the Euclidean 3-space with the x
axis as the polar axis. Using the usual isomorphism R
3

1
R
3

2
R
3
provided
by the Euclidean metric and the Hodge star, given in spherical coordinates by
e
r


r
dr (rd) (r sind);
e

1
r

rd (r sind) dr;
e

1
r sin

r sin d dr (rd),
WARP DRIVE WITH ZERO EXPANSION 5
we have

x
dx = d(r cos ) = cos dr r sin d
r
2
sin cos d d +r sin
2
dr d = d
_
1
2
r
2
sin
2
d
_
.
Consequently if we set
X v
s
(t)d
_
f(r)r
2
sin
2
d

2v
s
f cos e
r
+ v
s
(2f +rf

) sin e

with f(r) =
1
2
for large r and f(r) = 0 for small r, we are guaranteed to have a
divergenceless eld generating a warp bubble with velocity v
s
(t)

x
. Interestingly,
formulae for the extrinsic curvature tensor
K
ij
=
1
2
_

i
X
j
+
j
X
i
_
in spherical coordinates are available, as this corresponds to the so-called rate-of-
strain tensor of uid mechanics (see for instance [Ach98]). If we set
X = X
r
e
r
+X

+X

we have
K
rr
=
X
r
r
= 2v
s
f

cos ;
K

=
1
r
X

+
X
r
r
= v
s
f

cos ;
K

=
1
r sin
X

+
X
r
r
+
X

cot
r
= v
s
f

cos ;
K
r
=
1
2
_
r

r
_
X

r
_
+
1
r
X
r

_
= v
s
sin
_
f

+
r
2
f

_
;
K
r
=
1
2
_
r

r
_
X

r
_
+
1
r sin
X
r

_
= 0;
K

=
1
2
_
sin
r

_
X

sin
_
+
1
r sin
X

_
= 0,
yielding the energy density (as measured by Eulerian observers)
=
1
16
K
ij
K
ij
=
v
2
s
8
_
3(f

)
2
cos
2
+
_
f

+
r
2
f

_
2
sin
2

_
.
Notice that
= K
rr
+K

+ K

= 0,
as it should. This gives some insight into the geometry of this spacetime: for
example for cos > 0 (i.e., in front of the bubble), K
rr
< 0 where f

> 0 (i.e., in
the bubbles wall), indicating a compression in the radial direction; this is however
exactly balanced by K

+ K

= K
rr
, corresponding to an expansion in the
perpendicular direction.
6 JOS

E NAT

ARIO
3. Null geodesics and horizons
Besides violating energy conditions, warp drive spacetimes may have other patholo-
gies, such as horizons and innite blueshifts. To see evidence of this, let us consider
the case of a vector eld X generating a warp bubble with velocity v
s

x
satisfying
X
t
= 0
_

dv
s
dt
= 0
_
.
Notice that
<

t
,

t
>= 1 +X
2
and hence such spacetime can only be stationary if |v
s
| < 1.
Let us take v
s
> 1. Since null geodesics must satisfy
ds
2
= 0 dt
2
=
3

i=1
(dx
i
X
i
dt)
2

_
_
_
_
dx
dt
X
_
_
_
_
= 1
we see that a ash of light outside the warp bubble can be pictured in the Euclidean
3-space as a spherical wavefront which is simultaneously expanding with speed 1
and moving in the direction of X with speed X = v
s
. Thus it is clear that events
inside the warp bubble cannot causally inuence events outside the warp bubble at
large positive values of x. Assuming cylindrical symmetry about the x-axis, there
will be a point on the positive x-axis where X = 1; the cylindrically symmetric
surface through this point whose angle with X is given by
sin =
1
X
is a horizon, in the sense that events inside the warp bubble cannot causally inu-
ence events on the other side of this surface (see gure 1). Notice that away from
the warp bubble we have
sin =
1
v
s
which is the familiar expression for the Mach cone angle. Notice also that the
interior of the warp bubble is causally disconnected from part of the bubbles wall,
as is unavoidable (see [Low99]).
Inside and outside the warp bubble, the warp drive spacetime is at, and hence
the null geodesics are straight lines. If we assume that the warp bubbles wall is
very thin, to study light rays crossing it all we have to decide is how much these
light rays are refracted. According to the Fermat principle of General Relativity
(see [SEF92]), and since the proper time of an Eulerian observer inside the bubble
is just t, we must minimize the t coordinate of the event at which the observer sees
the light ray. If we choose the observer sitting at the centre of the bubble, light from
every point in the bubbles wall will reach him simultaneously; thus this observer
will see the rst light ray emitted at a given event to reach the warp bubbles wall.
If x(t) is the position of the light ray at time coordinate t, the unit vector
n =
dx
dt
X
is orthogonal to the wavefront (and in fact represents the direction of the corre-
sponding light ray from the point of view of a Eulerian observer outside the warp
bubble). The rst light ray to reach the warp bubble will do so at a tangency
WARP DRIVE WITH ZERO EXPANSION 7
X
1

exterior of the
warp bubble
interior of the
warp bubble
horizon
x
Figure 1. Computing the horizon.
between the wavefront and the warp bubble. Therefore n must be orthogonal to
the warp bubble, and will therefore coincide with the direction of the light ray in-
side the warp bubble (see gure 2). Incidentally, there will be no aberration: the
observer in the centre of the bubble will see light coming from the same direction
as seen by an Eulerian observer outside the bubble in the light rays path
2
. It is
possible to show that this is the limit of the results obtained in [CHL99] as the
thickness of bubbles wall tends to zero. If the thickness is nite, however, there
will always be some aberration.
In this approximation, it is also very easy to understand the origin of the horizon:
it marks the boundary of the set of points accessible to light rays emerging from the
warp bubble. Also, it is clear that the only some points which are visible from inside
the warp bubble; they are delimited by a surface similar to the horizon, which we
call the visibility horizon (see gure 3). Notice that this observer will see nothing
at all from directions beyond the direction of the visibility horizon.
We now compute redshifts. Recall that the 4-velocity of Eulerian observers
satises n
a
= dt; consequently, the energy of a particle with 4-velocity
U
a
=

t

t
+ x

x
+ y

y
+ z

z
is just n
a
U
a
=

t. From the geodesic equations, we see that
X
t
= 0
d
d
_

t X
_
x

tX
__
= 0
2
This however is slightly misleading: it is not hard to show that the expression for the refraction
angle r (measured by an Eulerian observer inside the bubble) in terms of the incidence angle i
(measured by an Eulerian observer outside the bubble) at a point in the bubbles wall making an
angle with the x-axis is
sin r =
sin i
1 + vs cos(i )
.
Thus by making vs large enough there will be light coming from arbitrary directions arbitrarily
close to the centre of the bubble.
8 JOS

E NAT

ARIO
dx
dt
n =
dx
dt
( = ) X 0
n
X
light ray
refracted light ray
warp bubble
interior of the
warp bubble
exterior of the
Figure 2. Refraction at the warp bubbles wall.
and hence
E (1 +X n) = E
0
where E is the energy measured by the Eulerian observers and E
0
is a constant
(corresponding to the energy measured by Eulerian observers when X = 0, i.e.,
inside the bubble). Thus the Eulerian observer at the centre of the warp bubble
will see a blueshift of 1+v
s
for light directly in front of him, decreasing to 1 for light
coming at 90

, and to zero for light at the edge of the visibility horizon. Conversely,
light emitted by the Eulerian observer at the centre of the warp bubble towards
his back will suer a redshift of 1 +v
s
; light rays emitted in directions approaching
the horizon will become increasingly blueshifted, reaching innite blueshift at the
horizon.
References
[Ach98] D. Acheson, Elementary uid mechanics, Oxford University Press, 1998.
[Alc94] M. Alcubierre, The warp drive: hyper-fast travel within general relativity, Class. Quan-
tum Grav. 11 (1994), L73L77.
[CHL99] C. Clark, A. Hiscock, and L. Larson, Null geodesics in the alcubierre spacetime: the
view from the bridge, Class. Quantum Grav. 16 (1999), 39653972.
[GW00] S. Gao and R. Wald, Theorems on gravitational time delay and related issues, Class.
Quantum Grav. 17 (2000), 49995008.
[LL97] L. Landau and E. Lifshitz, The classical theory of elds, Butterworth-Heinemann, 1997.
[Low99] R. Low, Speed limits in general relativity, Class. Quantum Grav. 16 (1999), 543549.
[SEF92] P. Schneider, J. Ehlers, and E. Falco, Gravitational lenses, Springer-Verlag, 1992.
[VBL00] M. Visser, B. Bassett, and S. Liberati, Superluminal censorship, Nucl. Phys. Proc. Suppl.
88 (2000), 267270.
[Wal84] R. Wald, General relativity, University of Chicago Press, 1984.
WARP DRIVE WITH ZERO EXPANSION 9
warp bubble
horizon
visibility horizon
X
n
Figure 3. Visibility horizon.
Department of Mathematics, Instituto Superior Tecnico, Portugal