Ring wormholes and time machines

Valeri P. Frolov,1, ∗ Pavel Krtouš,2, † and Andrei Zelnikov1, ‡

1
Theoretical Physics Institute, University of Alberta, Edmonton, Alberta, Canada T6G 2E1
2
Institute of Theoretical Physics, Faculty of Mathematics and Physics,
Charles University, V Holešovičkách 2, Prague, Czech Republic
In the present paper we discuss properties of a model of a ring wormhole, recently proposed by
Gibbons and Volkov [1–3]. Such a wormhole connects two flat spacetimes which are glued through
discs of the radius a bounded by the string with negative angle deficit −2π. The presence of the
string’s matter violating null energy condition makes the wormhole static and traversable. We
study gravitational field of static sources in such a spacetime in the weak field approximation. In
particular, we discuss how a field of an oblate thin massive shell surrounding one of the wormhole’s
mouth is modified by its presence. We also obtain a solution of a similar problem when both mouths
of the wormhole are located in the same space. This approximate solution if found for the case when
arXiv:2305.03887v1 [hep-th] 6 May 2023

the distance L between these mouths is much larger than the radius a of the ring. We demonstrate
that the corresponding locally static gravitational field in such a multiply connected space is non-
potential. As a result of this, the proper time gap for the clock’s synchronization linearly grows with
time and closed timelike curves are formed. This process inevitably transforms such a traversable
ring wormhole into a time machine. We estimate the time scale of this process.
Alberta Thy 2-23

I. INTRODUCTION rounded by some mass distribution [16]. In the presence

of closed timelike curves a problem of self-consistency of
Even in the absence of matter the space topology in the the standard physics becomes highly non-trivial [15, 17–
general relativity can be very non-trivial. A well known 19]. For example, infinite amplification of zero-point vac-
example is a famous Einstein-Rosen bridge connecting uum fluctuation during the creation of the time machine
two asymptotically flat spaces. Examples of solutions of may result in (formally) infinite growth of the value of the
the initial value problem describing wormholes in a three renormalized vacuum stress-energy tensor, that indicates
dimensional space were constructed by Misner [4] and that its backreaction would become important [20, 21].
Brill and Lindquist [5]. Wormholes and their properties Hawking formulated this problem as a chronology protec-
were described in detail by Wheeler in his famous book on tion conjecture [22]. For general discussion of wormholes
the geometrodynamics [6]. A generic property of vacuum and a time machine problem see e.g. books [23–25].
wormhole solutions of the Einstein equations is that as The simplest model of a traversable wormhole con-
a result of their time evolution they shrink and form a necting two asymptotically flat spaces has a spherically-
singularity so fast that one can not use them to ‘travel in symmetric geometry [12–14]. For such a wormhole there
space’. Gannon [7] proved that any asymptotically flat exists a two-sphere of the minimal area in the wormhole
spacetime with a nonsimply connected Cauchy surface throat and the null-energy condition should be violated in
has singular time evolution if it satisfies the weak energy its vicinity (see e.g. [13]). Recently, an interesting model
condition (see also [8–10]). of a traversable wormhole was proposed by Gibbons and
A special class of wormholes, called traversable, at- Volkov [1–3]. In this model two flat spacetimes are con-
tracted in the past a lot of attention. Their characteris- nected through a disk such that its boundary is a circu-
tic property is that a region with the non-trivial topol- lar ring where there exists a conical curvature singularity
ogy is located inside a compact spatial domain and the with the angle deficit −2π. This model of a traversable
particles and light can penetrate through a topological wormhole is a particular implementation of a loop-based
handle and return to the exterior region without meet- wormhole discussed by Visser [23]. One can identify this
ing a singularity [11–14]. As it was demonstrated by ring with a cosmic string with the corresponding negative
Thorne and collaborators [12, 15] a relative motion of energy distribution. Namely the presence of this matter
the traversable wormhole’s mouths can generate closed violates the null energy condition and makes it possible
timelike curves, so that such a wormhole becomes a ‘time an existence of a static traversable wormhole. The ge-
machine’. A similar effect of ‘time machine’ creation from ometry of the spacetime everywhere outside the ring (in-
a traversable wormhole can be achieved when the worm- cluding the disk where two flat spacetime geometries are
hole’s mouths are not moving, but one of them is sur- glued) is flat. The radius of the ring a can be arbitrary
large and a particle propagating through the disc from
one flat space to the other meets neither negative energy
density regions, nor a strong gravitational field. We refer
∗ [email protected] to this type of geometries as to the ring wormhole. A
† [email protected] main purpose of this paper is to demonstrate how such a
‡ [email protected] ring wormhole can be transformed into a time machine.
 2

For this goal we discuss a gravitational field created X− R− X+ R+

by static sources in the presence of a ring wormhole. We
2 2
consider two cases: (i) The ring wormhole connects two
different flat spaces, and (ii) such a wormhole connects
two spatially separated regions of the same spacetime.
In the latter case such a space is multiply connected. Y− Z− Y+ Z+
We assume that the gravitational field is weak and use
1
the linearized gravity equations. We also assume that
1
source of the gravitational field is an oblate massive thin
shell spheroid surrounding a mouth of the ring wormhole
confocal to the string ring. An exact solution for the
gravitational field of such a shell for the ring wormhole FIG. 1. A ring wormhole connecting two flat spaces R+
connecting two flat space is obtained. and R− .
For the second problem (the wormhole in a single
space) an approximate solution is found for the case when
the distance L between the mouths is much larger than identify surfaces of the discs D− and D+ . This identifi-
> <
the ring’s size a. We demonstrate that in the second cation is done as follows. We denote by D± and D± the
case the locally static gravitational field is non-potential ‘right’ and ‘left’ faces of the discs. Namely, if the disc
and as a result of this closed timelike curves are created D± is reached by a point with the positive value of Z±
>
some time after the massive source surrounding one of we say that it belongs to D± . In the opposite case when
the mouths is ‘switched on’. We also estimate the corre- a point approaches the disc with Z± < 0 we say that it
< < > >
sponding time required for this as a function of mass of belongs to D± . One identifies D+ with D− and D+ with
<
the shell, its size and the distance between the mouths. D− , without any rotation around z-axis.
The paper is organized as follows. In the next section This means that a free particle 1 moving in space R+ ,
< >
we remind the reader properties of the ring wormholes. which meets the disc D+ , enters space R− at D− with
In the section III we explain the meaning of an approxi- identical coordinates X− = X+ , Y− = Y+ , Z− = Z+ = 0,
mation of a weak gravitational field in application to the and continues its motion with increasing Z− coordinate.
> <
wormhole spacetime. Gravitational field in the presence Similarly if a particle 2 enters D+ , it appears at D− part
of a ring wormhole connecting two flat spaces is discussed of the disc D− as illustrated in Fig. 1.
in section IV. A similar problem for a string wormhole In what follows we denote R e the space obtained by uni-
connecting two separated regions of a single space is con- fying two spaces R+ and R− with identification described
sidered in section V. Closed timelike paths formation, i.e., above. We call the discs D− and D+ in R− and R+ , re-
appearance of a time machine, in the space of a worm- spectively, the wormhole mouths, and their identification
hole with a non-potential locally static gravitational field D the wormhole throat.
is discussed in section VI. In particular, it contains esti- The radius a of the disc has the dimension of length.
mation of the time required for a closed timelike curves It is convenient to use it as a natural scale and write the
formation. In Section VII we discuss the obtained results metric in a flat space in the form
and their possible consequences. Two appendices collect
information concerning the static gravitational field in dL2 = δij dX i dX j = a2 dl2 , (2.1)
2 i j 2 2 2
the weak field approximation and general properties of dl = δij dx dx = dx + dy + dz . (2.2)
locally static non-potential gravitational fields.
In this paper we use sign conventions adopted in [26]. In what follows we shall use the dimensionless metric dl2
and dimensionless coordinates xj = X j /a. We denote by
(z, ρ, ϕ) dimensionless cylindrical coordinates
II. GEOMETRY OF A RING WORMHOLE
ρ2 = x2 + y 2 , x = ρ cos ϕ , y = ρ sin ϕ , (2.3)
A. A ring wormhole connecting two asymptotically with the metric taking the form
flat spaces
dl2 = dρ2 + dz 2 + ρ2 dϕ2 . (2.4)
Let us discuss first a spatial configuration of the worm-
hole spacetimes. We begin with a case of a ring worm- The ring equations are
hole connecting two flat spaces. Such a wormhole can ρ = 1, z = 0. (2.5)
be obtained as follows. Consider two copies of a three-
dimensional flat space. Denote by X i = (X± , Y± , Z± ) It will be very useful to introduce also the oblate
standard Cartesian coordinates in these spaces. Let D± spheroidal coordinates (χ, ϑ, ϕ) related to the cylindri-
be discs in R± of radius a in the planes Z± = 0 with the cal coordinates as follows:
center at the origin of these spaces (see Fig. 1).
To join the spaces R− and R+ with a wormhole, we z = sinh χ cos ϑ , ρ = cosh χ sin ϑ . (2.6)
 3

3 − =
 3 +  3   3 
=
4 4
=
4
=
4 − = − = + = + =
4 4 4 4

R− R+
− = o  + = const
2 2
1 1

 =  =0 z−  =  =0 z+ D− D− D+ D+ z

 0  0 + = 0
− =  − = 0 + = 
− 
2 2
FIG. 2. Ring wormhole in space R e with two asymptotic re-
gions R− and R+ . Each of the domains R± is covered by
cylindrical coordinates ρ± , ϕ± , z± . Directions related to an- FIG. 3. Ring wormhole space Rwh with one asymptotic re-
gular coordinate ϕ± are not shown. They can be obtained gion. We have global cylindrical coordinates ρ, ϕ, z and two
by rotations around axes ρ± = 0. The whole space R e is copies of the oblate spheroidal coordinates χ± , ϑ± , ϕ± . The
covered by one copy of the oblate spheroidal coordinates coordinate ϕ ≡ ϕ± is not shown. The discs D± representing
χ ∈ (−∞, ∞), ϑ ∈ (0, π) and ϕ ≡ ϕ± ∈ (−π, π). Dashed lines the wormhole’s mouths are indicated by double lines located
corresponds to χ = const, dotted lines to ϑ = const. Double at z = ±`/2. The values of the coordinate lines indicated
lines indicate the discs which are identified as indicated by in the diagram corresponds to the domains V̄± described in
arrows. They represent the mouths of the wormhole. the text and Fig. 4. The coordinates in V̄± are related to the
opposite coordinates in the leaked domains V̂∓ by χ+ = −χ−
and ϑ+ = π − ϑ− . Compare also Fig. 5.
A two-dimensional surface χ = const is an oblate
spheroid, confocal with the ring, with cosh χ being its
larger semi-axis and sinh χ being the smaller one. A sur- B. A ring wormhole in a single space
face ϑ = const is a hyperboloid confocal also with the
ring. Let us note that the relations (2.6) can be written It is easy to use a simple procedure to construct a space
in the following complex form that contains both mouths of a single wormhole. One
just places two discs representing the wormhole’s mouths
z + iρ = sinh(χ + iϑ) . (2.7) into one flat space. Obviously, there is an ambiguity in
The flat metric in the oblate spheroidal coordinates is the choice of positions and orientations of the discs. We
consider the simplest case when the discs are orthogonal
dl2 = (sinh2 χ+ cos2 ϑ)(dχ2 + dϑ2 ) + cosh2 χ sin2 ϑ dϕ2 . to a common axis and separated by a distance L.
(2.8) If we introduce global dimensionless flat and cylindri-
For a single flat space R, the oblate spheroidal co- cal coordinates (x, y, z) and (z, ρ, ϕ), the centers of two
ordinates have a two-fold degeneracy: (χ, ϑ, ϕ) and discs D± of radius 1 are located along z-axis at z = ±`/2,
(−χ, π − ϑ, ϕ), correspond to the same point. For this and the discs are orthogonal to the z-axis. Here, we de-
reason, it would be sufficient to consider the ranges of noted by ` = L/a the dimensionless version of the discs’
the oblate spheroidal coordinates χ ∈ (0, ∞), ϑ ∈ (0, π) distance.
and ϕ ∈ (0, 2π). To form the wormhole, we identify the discs D− and
In the wormhole space R, e we introduce separate copies D+ in a similar manner as we described above: Each
>
of flat (x± , y± , z± ) and cylindrical (z± , ρ± , ϕ± ) coordi- of the discs has two faces which we denoted by D± and
< < > > <
nates in spaces R− and R+ . However, we take advantage D± . One identifies D− with D+ and D− with D+ , with-
of the degeneracy of the oblate spheroidal coordinates out any rotation around the z-axis. This identification is
and define just single coordinates (χ, ϑ, ϕ) covering the illustrated in Fig. 3.
whole wormhole space, where χ ∈ (−∞, ∞), simply as- We call the resulting space Rwh . The two-faces discs
suming that χ > 0 in R+ and χ < 0 in R− . D− and D+ are called left and right mouths of the worm-
Fig. 2 shows space R e in (χ, ϑ) coordinates. To specify hole. Identified, they form the throat of the wormhole.
a point, one needs to add an azimuthal angle ϕ. The Strictly speaking, the standard coordinates (x, y, z) and
coordinates (χ, ϑ, ϕ) cover the complete space and they (ρ, ϕ, z), when understood as coordinate maps on Rwh ,
are continuous at the discs representing the mouths of are well defined only outside the throat.
the ring wormhole. The left figure corresponds to R− Clearly, the wormhole spacetime Rwh has a differ-
domain where χ < 0, while the right one corresponds to ent topology than the original empty flat space. Its
R+ where χ > 0. first homology group is non-trivial, since we have non-
Let us note that it is possible to generalize the de- contractible loops spanned through the wormhole.
scribed ring wormhole space to the case when there exist In Rwh , we introduce two sets of oblate spheroidal co-
not only one but several discs in space R+ , each ‘con- ordinates. One of them, (χ+ , ϑ+ , ϕ+ ), is centered on the
nected’ with its own version of flat space. We do not con- ring at z = +`/2, and the other one, (χ− , ϑ− , ϕ− ), is
sider such multi-wormhole configurations in the present centered on the ring at z = −`/2.
paper. For χ± > 0, we restrict the coordinates (χ+ , ϑ+ , ϕ+ )
 4

to the domain where z > −δ̃ and similarly, we restrict

the coordinates (χ− , ϑ− , ϕ− ) to the domain where z < δ̃.
We assume that these coordinates are valid not only D- D+

for positive values of χ± but are extended to some neg- V- V-
z
ative value −δ < χ± . This provide us with a ‘leakage’
of these coordinates through wormhole mouths. Namely, W W
coordinates (χ+ , ϑ+ , ϕ+ ) cover not only the space outside
discs D+ but also some domain −δ < χ+ in the vicinity
of the disc D− . Similarly, the coordinates (χ− , ϑ− , ϕ− ) Domain V− = V̄− ∪ V̂− .
cover exterior of the disc D− and a domain −δ < χ− in
the vicinity of D+ .
We denote the corresponding domains of definitions of W W
coordinates (χ± , ϑ± , ϕ± ) as V± , respectively, see Fig. 4. z 
These two domains intersect in a narrow strip V̄ where V+ V+
−δ̃ < z < δ̃, and in a spheroidal neighborhood V̂ of the D- D+
throat, given by −δ < χ± < δ, see Fig. 4. We assume
δ and δ̃ to be sufficiently small so that V̄ and V̂ do not
overlap. Domain V+ = V̄+ ∪ V̂+ .
We denote the part of V+ with χ+ > 0 as V̄+ and the
V-
‘leaked’ part of V+ , i.e., the part with −δ < χ+ ≤ 0, as
V̂+ . We define V̄− and V̂− analogously. The domains are W D+
related  
V V V z
V− = V̄− ∪ V̂− , V+ = V̄+ ∪ V̂+ , D- W
(2.9)
V̄ = V̄− ∩ V̄+ , V̂ = V̂+ ∪ V̂− . V+
Domains V̄ and V̂ .
In domains V̄+ and V̄− , the oblate spheroidal coordi-
nates are related to the global cylindrical coordinates as
FIG. 4. Domains V− (top) and V+ (middle) are covering to-
gether the whole space. They are composed by the ‘primary’
`
z = ± + sinh χ± cos ϑ± , parts V̄± and ‘leaked’ parts V̂± . The bottom diagram shows
2 two intersections V̄ and V̂ of these two domains. Horizontal
ρ = cosh χ± sin ϑ± , (2.10)
direction corresponds to the z axis, vertical to ρ direction.
χ± > 0 , ϑ± ∈ (0, π) . The mouths D± of the wormhole are represented by the thick
vertical lines. The distance between mouths is `.
These formulas also establish relations between coordi-
nates (χ+ , ϑ+ ) and (χ− , ϑ− ) in V̄ , C. Static wormhole spacetimes

cosh χ+ sin ϑ+ = cosh χ− sin ϑ− ,

(2.11) Let us note that after fixing three-dimensional spatial
−`/2 + sinh χ− cos ϑ− = `/2 + sinh χ+ cos ϑ+ . geometry R, we have to fix also full spacetime geometry
M . In this paper we will assume that the time direction
In the ‘leaked’ domains V̂− and V̂+ , the oblate coor- is added to the given space in a trivial way, namely that
dinates (χ+ , ϑ+ ) and (χ− , ϑ− ) are related to the global the resulting spacetime will be locally static with time-
cylindrical coordinates (ρ, z) as slices locally equivalent to the chosen spatial geometry.
The simplest example is a globally static spacetime
` M = R × R, with a global dimensionless coordinate t
z = ∓ + sinh χ± cos ϑ± ,
2 independent of the spatial geometry, and with spacetime
ρ = cosh χ± sin ϑ± , (2.12) metric
χ± < 0 , ϑ± ∈ (0, π) . ds2 = −dt2 + dl2 . (2.14)

As a consequence, we have in V̂ We can generalize this case in several ways. First, we can
admit a redshift factor
χ+ = −χ− , ϑ + = π − ϑ− . (2.13) ds2 = −e2U dt2 + dl2 , (2.15)
The relation of coordinates ϕ± is trivial. We can con- where U is globally well-defined spatial function.
sistently set ϕ = ϕ− = ϕ+ on all overlaps of the domains When the spatial manifold is topologically non-trivial,
of definition. as the space Rwh discussed above, we can have more elab-
 5

the very existence of a static traversable wormhole. This

χ- , ϑ- V- distribution of matter can be identified with an infinitely
thin cosmic string (strut) with a negative angle deficit
−2π. In order to exclude the δ−like singularity at the
ring one can smear the distribution of the matter of the
corresponding cosmic string. This makes the spacetime
 
V V V smooth, but the equation of state of the corresponding
matter is not very realistic. If λ is the radius of the
smeared string, we assume that the parameter d = λ/a
is small. This always can be achieved since the radius
of the ring a can be arbitrary large. In what follows
V+ χ+ , ϑ+ we always assume that, if necessary, the corresponding
smearing is done.

FIG. 5. Coordinate maps (χ− , ϑ− ) in the domain V− (top

half) and (χ+ , ϑ+ ) in V+ (bottom half). III. WEAK GRAVITATIONAL FIELD

A. Gravitational potential
orate generalizations. First, we can identify the mouths
of the wormhole at different times. In the weak-field approximation, a solution of the Ein-
Namely, suppose we introduce time t synchronized stein equation for a static matter distribution can be
with clocks at infinity, defined everywhere except at the written in the form
wormhole throat. It complements the global flat coordi-
nates (x, y, z) or cylindrical coordinates (ρ, z, ϕ). ds2 = −e2U dt2 + e−2U dl2 . (3.1)
Now, we do not identify the discs D± at the same
values of time t, but we identify D− at t = to with D+ at where the gravitational potential U obeys the Poisson
t = to + ∆t, with ∆t being a constant. We require that equation
the spacetime geometry is still described by the metric
4U = 4πµ . (3.2)
ds2 = −dt2 + dl2 . (2.16)
Here µ is the mass density defined with respect to the
It is straightforward to see that the metric is continu- flat metric dl2 . The metric (3.1) is a perturbation of the
ous through the wormhole, since the constant shift ∆t flat background spacetime
disappears in the differential dt.
Consider a particle which enters the disc D− at time ds2o = −dt2 + dl2 . (3.3)
t = to . It appears from the disc D+ at time t = to + ∆t.
We thus say that there exists a time gap ∆t associated For details and further discussion of static metrics in the
with the wormhole. Since the proper time at both sides weak field approximation see Appendix A.
of the identified discs must be the same (the geometry We denote ξ µ a time-like Killing vector and uµ a nor-
is continuous through the wormhole), the time gap must malized four-velocity of static observers moving along
remain the same. This means that if a second particle this Killing vector,
is sent through D− shortly after the first one at time
uα = e−U ξ α . (3.4)
t = to + δt it will appear at D+ at time t = to + δt + ∆t.
However, if we put some matter around one of the Obviously,
mouths of the wormhole, then its gravitational field af-
fects a proper time of static observers and the static uα = −eU t,α . (3.5)
spacetime metric needs further modification. In general,
the time shift of coordinate t will not remain constant. The four-acceleration of these static observers,
Our spacetime becomes only locally static as we will de-
scribe in the following sections. wµ = uµ;ν uν , (3.6)

satisfies the relation w[α,β] = 0 and hence, at least locally,

D. A nature of the ring one has

wα = U,α . (3.7)
Before going further let us make a following remark.
Gibbons and Volkov [1–3] demonstrated that in fact there We call wα the acceleration field, and −U,α the grav-
exists a δ−like singularity of the curvature at the location itational field strength. Since wα is orthogonal to the
of the ring. The corresponding distribution of matter time direction uα , it can be restricted to its spatial com-
violates the null energy condition which makes it possible ponents. The field strength is a fictitious force that one
 6

assumes in the non-inertial static frame to explain a ten- If we have two such potentials U and Û on domains
dency of free observers to move with respect to the frame. V and V̂ , respectively, on their intersection V ∩ V̂ the
Non-moving static observers have to ‘compensate’ this potentials may differ only by a constant,
force by a real force equal to wj per unit mass. Of
course, in the full spacetime description, the static ob- wα = U,α = Û,α ⇒ ∆U = U − Û = const . (3.11)
servers move along non-geodesic trajectories with four-
acceleration wα caused by the real force. We can cover the whole manifold by simply connected
In static spacetimes we can effectively ‘ignore’ the time domains with associated with them potentials. Next, we
coordinate. The weak-field equation (3.2) can be under- may shift the potentials by a constant on some of these
stood as the equation on spatial section t = const, or, domains to fit potentials in the neighborhood domains.
better, on a factorized three-dimensional space, points of However, in topologically non-trivial spacetime it may
which are orbits of the Killing vector ξ α . We will see in not be possible to do it consistently everywhere. It may
a moment, that the latter approach is more general. happen that we find two domains V and V̂ with poten-
tials which are already continuously extended along some
path in the manifold. However, these potentials do not
B. Locally static spacetimes fit; they differ on the intersections of these domains by a
constant ∆U .
The above description of the static spacetime, as well A maximal extension of the potential U can be speci-
as of the weak field, was local. Let us make some re- fied by choosing a point po of vanishing potential (typi-
marks about global aspects. We start with a general cally in an asymptotical region) and paths γp from this
static spacetime without any approximation. point to each point p in the spacetime. The potential is
First, we observe that metric (3.1) is invariant under given by
a scaling transformation Z
U= wα dxα . (3.12)
U = Û + ∆U , t = e−∆U t̂ , ˆ2 ,
dl2 = e2∆U dl (3.8) γp

parametrized by a constant1 ∆U . This rescaling tells us, For a continuous change of the path γp the potential U
that two potentials on the background spacetime which changes smoothly. However, in a topologically non-trivial
differ only by a constant are in some sense equivalent. space some pathes heading to two close points in the
This opens a very important possibility for globally intersection V ∩ V̂ cannot be smoothly deformed to each
non-trivial spacetimes. Consider spacetime with a non- other. Then, such a potential may not be continuous.
trivial first homology group, i.e., spacetime containing Impossibility to extend the potential to a single-valued
non-contractible loops. From the duality between ho- global smooth function implies that also Killing vector
mology and de Rham cohomology we know that in such ξ α , given by (3.4), cannot be defined globally. It can be
a spacetime closed 1-forms do not have to be exact,2 continuously extended to domains V and V̂ , similarly to
potential U . But on the intersection V ∩ V̂ , it will differ
σ[µ,ν] = 0 ; σµ = α,µ . (3.9) by rescaling

In particular, we can have spacetime with global static ξ α = e∆U ξˆα . (3.13)
observers given by globally well-defined four-velocity uα .
It gives a globally well-defined four-acceleration wα . Al- The same applies to the time coordinate t related to the
though it is closed, four-velocity by (3.5). On the intersection of domains V
and V̂ we find
w[µ,ν] = 0 , (3.10)
t = e−∆U t̂ . (3.14)
it may be non-potential. Such wα corresponds to a non-
trivial element of the first cohomology group. Spacetime with such a structure is called a locally
Of course, locally, on a topologically trivial domain V , static [16], in contrast to the globally static spacetimes.
we can always find a potential U such that (3.7) holds. In globally static spacetimes all quantities ds2 , dl2 , uα ,
But in a general case such a potential does not have a wα , ξ α , U , and t are defined globally. In locally static
global smooth extension to the whole spacetime. spacetimes, ξ α , U , dl2 , and t have only local meaning.
See Appendix B for more details.

1 Please, notice a difference between character for Laplace opera- C. Locally static weak gravitational field
tor, 4U , and the letter Delta in the constant ∆U .
2 Here, one should understand tensor indices as abstract indices,
not coordinate ones. On a simple local coordinate map, any Our goal now is to build a locally static spacetime de-
closed form is exact. But we speak here about tensor fields de- scribing the gravitational field in the presence of a ring
fined on the whole spacetime. wormhole in the weak field approximation.
 7

We start with a globally static spacetime, in fact, the procedure (3.12). However, similarly to the potential,
with a flat spacetime which differs from the Minkowski the extended inertial coordinates will not be continuous
spacetime just by a wormhole identifications described through the wormhole and we will have to discuss, how
in Sec. II. On this background we find a gravitational they are related. We will demonstrate this procedure in
potential U which satisfies locally Poisson equation (3.2) Sec. V.
with some physically reasonable sources. But we are in- Non-existence of a global static time coordinate t is
terested in situations when the potential cannot be ex- also a reason why we cannot use a global spatial sec-
tended globally. tion t = const. However, we can still introduce the three-
In such a case, the globally well-defined quantities are dimensional factorized space of orbits of the Killing vec-
acceleration field wα , four-velocity uα , and, of course, the tor ξ α and solve the weak-field equations (3.15) and
full perturbed metric ds2 . (3.16) on it.
The weak-field equation (3.2) can be rewritten using Finally, let us note that rescaling (3.17) can be used
globally defined quantity wj . The potential relation (3.7) also for identification of locally inertial coordinates near
is superseded by (3.10). We get an arbitrary point p. By a suitable choice of ∆U one
can always put Û (p) = 0. For this gauge the coordi-
∇[i wj] = 0 , (3.15) nate t̂ at p coincides with the proper time at this point
i
∇i w = 4πµ . (3.16) and locally Cartesian coordinates x̂j near p coincide with
proper lengths along the axes.
These equation can be understood as the gravito-static
equations on the three-dimensional factorized space men-
tioned above. D. Gravity field flux
Taking into account (3.1), we conclude that inertial
time t and flat metric dl2 , as well as the implicitly de- Let S be a closed two-dimensional surface embedded
fined cartesian coordinates xi , depend on the choice of in the spatial section t = const. We define a gravity field
the potential. Their perturbed versions are not thus de- flux over this surface as
fined globally. I
1
Similarly to the discussion above, if we find domains V ΠS = wa na d2S . (3.18)
and V̂ on which the potentials differ by a constant ∆U , 4π S
expression (3.1) for the metric on these domains define Here na is an outward-pointing unit normal to S and d2S
(t, xj ) and (t̂, ĵ) which will be related by is a two-dimensional surface area element on S.
Let S1 and S2 be two homological closed surfaces and
U = Û + ∆U , V is a three-dimensional volume restricted by them
−∆U
t=e xj = e∆U x̂j ,
t̂ , (3.17)
∂V = S2 − S1 . (3.19)
ˆ2 .
dl2 = e2∆U dl
Then the Gauss theorem applied to volume V and the
Identification of the resulting spacetime with the original weak-field equation (3.16) implies that
background with metric ds2o thus cannot be done globally Z Z
in the coordinates (t, xj ). 1
ΠS2 − ΠS1 = ∇i wi d3V = µ d3V . (3.20)
However, the inertial coordinates are very simple and 4π
offer a straightforward local identification of the per- It has important consequences:
turbed spacetime with the background. We want to take
advantage of them. Therefore, we modify our perturba- (i) If there is no matter in the space between S1 and
tion procedure slightly. S2 , the fluxes through these surfaces coincide.
Namely, we do not require that our perturbed metric Applying (3.20) on a boundary ∂V of a compact volume
ds2 is a global perturbation of one unperturbed back- V , we have:
ground ds2o . To justify the validity of the weak-field
equations, it is enough that the resulting spacetime is (ii) Consider adiabatic change of the distribution of
a perturbation of a background on local domains. The mass. It does not change the flux through a two-
domains must cover whole resulting spacetime. But the dimensional surface S = ∂V enclosing a three-
background can be assumed for each domain separately. dimensional volume V until the matter crosses S.
In this approach we can easily identify background
In the case of space with a non-trivial topology or with
spacetimes with the resulting perturbed spacetime using
several asymptotic regions, one can formulate a more gen-
locally defined inertial coordinates. However, we have to
eral proposition:
face a problem of the unequality of these coordinates on
overlaps of the domains. (iii) The flux through any closed two-dimensional sur-
In the specific cases of the wormhole spacetimes, we face S does not change under an adiabatic change
will be able to extend the potential and related quantities of a mass distribution until the matter crosses the
to the full spacetime, except the wormhole throat, using surface S.
 8

This last proposition has a direct counterpart in a well wormhole in R+ is inside the cavity. In this case the cor-
known result of the Maxwell electrodynamics, connect- responding solution inside the shell and out of it remains
ing the change of the flux over a closed surface with the the same as in a case without the ring wormhole, while
current crossing it. The result can be easily proved by in the cavity of the shell and everywhere in space R− the
using the Stokes theorem. In the case of a weak gravi- potential U is constant equal to Uo .
tational field, the proposition does not follow just from In a special case when all the axes of the ellipsoidal
properties of the Poisson equation. One would need to shell are equal, it takes the form of the sphere. For such
specify laws governing the dynamics. We will not discuss a case, this conclusion remains valid even if the gravita-
these laws here. Let us just mention, that there exists a tional field of the shell is not week. Namely, one can use
well known analogy between the weak gravitational field a massive thin shell approach by Israel [31] to glue the
and the Maxwell equations, see, e.g., [27, 28]. Schwarzschild metric outside the shell with the inner flat
solution. By imposing the ring wormhole inside the shell
does not change the gravitational field around.
IV. GRAVITATIONAL FIELD NEAR THE Let us emphasize that the presence of a ring worm-
WORMHOLE CONNECTING TWO hole has non-trivial consequences. We can point out two
ASYMPTOTIC DOMAINS aspects:
(i) Even in the case of a ring wormhole connecting two
In this and the following sections, we work in a three- flat spaces, the above described solution of the Pois-
dimensional flat space with a non-trivial identification son equation (3.2) with the described source is not
given by the wormhole. We will discuss some solutions of unique.
the field equation (3.2), or (3.15) and (3.16), respectively.
As earlier, in Sec. II we use a radius a of the ring to define (ii) In the presence of a ring wormhole in a single space,
dimensional metric and coordinates. We will return to the statement that the gravitating force vanishes in
the spacetime description in Sec. VI. the cavity of the shell is not correct.
We will discuss these issues below.
A. Gravitational field of a massive shell
B. Zero modes
It is instructive to discuss first a gravitational field of a
thick homogeneous ellipsoidal shell in a simple flat space The space Re with the wormhole connecting two asymp-
which does not contain any wormhole. Let us consider totic regions has the first cohomology group trivial and
a shell bounded by two similar ellipsoids filled by the thus, we do not have to face problems with non-existence
matter of constant density. It will be sufficient for us of a global potential. However, it has two asymptotic re-
to observe that according to famous Newton’s theorem, gion and it allows ambiguity in the potential, even for
cf. [29], the gravitational force −∇U inside the shell (in- trivial sources. Let us illustrate this point.
side the inner ellipsoid) vanishes.3 The potential in the Consider the function4
cavity is thus constant U = Uo . Here, we normalize the
potential U so that it decreases at the infinity outside the Z(χ) = − arccot(sinh χ) , (4.1)
outer ellipsoid.
with arccot s taking values in the interval (0, π). It has
The field both outside and in the shell’s interior (be-
the following properties
tween both ellipsoids) can also be written explicitly
[29, 30]. However, these expressions are rather compli- 1
cated. Instead of this, we will consider below a much Z 0 (χ) =,
cosh χ (4.2)
simpler model in which the matter surrounding a throat π
is taken in the form of a thin spheroidal mass shell. This Z(−∞) = −π, Z(0) = − , Z(+∞) = 0 .
2
will allow us to obtain simple analytical expressions for
the gravitational field. Let m∞ and C be two constants, then it is easy to check
Using the described field in the shell’s cavity, it is that
easy to construct a solution of the equation (3.2) in Uzm (χ) = m∞ Z(χ) + C (4.3)
the space R e where the ring wormhole connects two flat
spaces. We assume that the disc associated with the ring is a solution of the homogeneous Laplace equation
4Uzm = 0 . (4.4)

3 Let us stress that Newton’s theorem requires the bounding ellip-

soids being similar and, therefore, they are not confocal. Thus, 4 The function Z is related to Legendre function Q0 evaluated
even when the ellipsoids are oblate spheroids, they cannot be on the imaginary axis with a branch cut chosen such that it is
specified just by two values of spheroidal coordinate χ. analytical on the whole imaginary axis.
 9

a non-zero force acting on a unit-mass particle equal to

m∞
− ∇Uzm = − ∇χ . (4.7)
cosh χ
For positive m∞ this force is directed towards the de-
creasing χ. In R+ this force describes an attraction of
the particle by the wormhole, while in R− it is repulsive
from the wormhole.
R- R+
It is easy to calculate a flux ΠD through the worm-
hole associated with the zero mode. Consider a closed
wzm ρ- wzm ρ+
two-dimensional surface S surrounding the throat D of
the wormhole, with normal oriented towards R+ asymp-
totic. Since the zero mode has vanishing sources, the flux
does not depends on exact shape and position of the sur-
z- z+ face, cf. argument (3.20). The calculation for spheroid
R- R+ χ = const gives
ΠS = m∞ , (4.8)
j
FIG. 6. Potential Uzm (top) and field strength wzm (bottom) where we used
of the zero mode. The left diagram shows the field in space
1
R− , and the right one in R+ . Thick segments on the ρ± axes nα = p 2
δχα ,
represent the mouths of the wormhole. Angular coordinate ϕ sinh χ + cos2 ϑ (4.9)
is ignored and can be recovered by a rotation around the axes q
2
ρ± = 0. dS = sinh χ + cos2 ϑ cosh χ sin ϑ dϑdϕ .
For surface D at χ = 0, representing the identified discs,
χ we obtain the flux ΠD through the wormhole throat.
A zero-mode field can be created in the following pro-
cess. Let us start with empty space with vanishing po-
π
-
2
tential. Consider a particle of mass m at the infinity R+
and move it adiabatically to bring it close to the throat.
The flux ΠD through the surface D at the throat is not
 (χ)
-π changed in such a process until the mass crosses D. After
crossing D the flux becomes ΠD = m and it remains such
FIG. 7. Functions Z(χ).
if the mass stays in R− . When the mass m is brought to
infinity in the space R− , one is left with a space without
any matter, but with a non-vanishing flux ΠD through
the throat. It will be exactly the zero mode described
We call such a solution a zero mode. It is convenient to
above.
use the standard ambiguity in the choice of the potential
and to put C = 0, then
C. A field of a massive thin shell around the
Uzm |χ=+∞ = 0, Uzm |χ=−∞ = −m∞ π . (4.5) wormhole

In the absence of the ring wormhole, one should put

m∞ = 0 as well, in order to avoid a discontinuity in the Let us assume now that there exists a static mass dis-
first derivative at χ = 0. For nonvanishing m∞ , a jump in tribution surrounding the wormhole. In particular, we
the field strength across the disc χ = 0 would correspond consider a thin massive shell which has a form of an
to a mass distribution on the disc, and the potential (4.3) oblate spheroid So in R+ surrounding the wormhole and
would not be a solution of source-free (4.4). confocal to its ring. In what follows we assume that the
In the presence of the ring wormhole connecting spaces this massive shell is formed by massive particles that are
R+ and R− , the situation is quite different. Even if one brought together from R+ infinity. During this process
imposes a condition that the potential vanishes at infin- no zero-mode field would be created and they should be
ity of R+ by putting C = 0, there exists a one-parameter excluded in our solution.
family of homogeneous solutions parameterized by a con- In oblate spherodial coordinates (χ, ϑ, ϕ) the equation
stant m∞ . The difference of potentials at the infinities of shell is χ = χo > 0. Let M be the mass of the massive
of the spaces R+ and R− is shell, and m = M/a be its dimensionless variant. The
mass is not distributed homogeneously over the shell sur-
∆U = Uzm |χ=+∞ − Uzm |χ=−∞ = m∞ π . (4.6) face and its surface density is
m
σ= p
2
. (4.10)
In the presence of a zero-mode potential, there exists 4π sinh χ + cos2 ϑ cosh χ
 10

of the thick shell discussed in Sec. IV A is

(
m Z(χ) for χ > χo ,
Ush (χ) = (4.12)
Uo = m Z(χo ) for χ < χo .

The potential inside the shell, as well as in R− , is con-

stant. Outside the shell, it is the same as the zero mode,
and it vanishes at the infinity of R+ . The jump in the
R- R+ field strength at the shell corresponds to the surface mass
distribution (4.10).
wsh ρ- wsh ρ+ Of course, we can superpose this solution with the zero
mode solution

U = Ush + Uzm . (4.13)

z- z+
At the infinity of R+ the potential U again vanishes,
R- R+ while at the infinity of R− its value is

j U |χ=−∞ = m Z(χo ) − m∞ π . (4.14)

FIG. 8. Potential Ush (top) and field strength wsh (bottom)
of a thin massive spheroidal shell located around the worm- We thus found that a general potential (4.13) of the
hole in R+ . The left diagram shows the field in space R− ,
thin shell vanishing in the infinity of R+ contains two con-
and the right one in R+ . Thick segments on the ρ± axes rep-
resent mouths of the wormhole. The shell (depicted by the
stants, m and m∞ . Their meaning can be understood us-
thick ellipse) is confocal with the wormhole mouth. Angular ing the gravity field flux. As we have seen, cf. (3.20), the
coordinate ϕ is ignored and can be recovered by a rotation flux may be used to define the mass function mS = ΠS
around the axes ρ± = 0. characterizing mass inside a surface S. For the equipo-
tential surface χ = const the same calculation as in (4.8)
gives
The distributional volume density is5
mS (χ) = mΘ(χ − χo ) + m∞ . (4.15)
m
µ = σ δSo = δ(χ − χo ) . It should be emphasized that both in R+ and R− the nor-
4π(sinh2 χ + cos2 ϑ) cosh χ
(4.11) mal to the spheroid χ = const is chosen in the direction
Surprisingly, this mass distribution can be obtained [29] of growing χ.
as a thin shell limit of the thick homogeneous spheroidal In the exterior of the shell, the mass function is mS =
shell discussed in Sec. IV A. Nonhomogeneity of the sur- m + m∞ . Inside the shell (including space R− , where
face mass density arises from the fact, that the thick shell χ < 0), the mass function is just mS = m∞ . It correctly
is formed by two similar oblate spheroids and its thick- jumps by the mass of the shell when crossing the shell.
ness vary with position on the spheroid. We will see that Since mS does not change when we further decrease χ, we
for the thin shell the potential Ush depends only on the can interpret m∞ as a mass ‘far behind’ the wormhole,
χ coordinate and hence it is constant on the shell.6 located under arbitrary negative χ, i.e., located in the
In fact, to obtain a solution of the Poisson equation for infinity of space R− .
the thin shell we can use solutions Z(χ) of the Laplace However, this mass had to be brought into the infin-
equation given by (4.1). The field Ush analogous to that ity of R− through the wormhole from the infinity of R+ .
Because by fixing vanishing potential U at the infinity of
R+ , we have chosen that all matter forming the gravita-
tional field has to be brought from this infinity. The mass
5 Here, δS is a covariant δ-function localized on a surface S, m∞ at the infinity of R− thus had to be moved from the
Z Z infinity of R+ , and the potential difference (4.6) corre-
f δS dV = f dS , sponds to the work needed for that.
S
We interpreted the potential assuming that it vanishes
and δ(χ − χo ) is its coordinate representation. They are related
at infinity of R+ . We selected one of the asymptotics
δS dl = δ(χ − χo ) dχ , as the origin of all matter. We thus have broken the
with the proper length dl orthogonal to S symmetry of the system. Of course, we can also allow
q the matter to originate at the infinity of R− . In such a
dl = sinh2 χ + cos2 ϑ dχ . case, we have to recover the constant C in (4.3).
Then, one can interpret the zero mode as the gravita-
6 This solution is similar to the well known solution for a static tional field trapped by the wormhole, which holds mem-
electric potential of a charged conducting oblate spheroid which ory about a previous activity how the matter has been
is discussed, e.g., in [30] moved through the wormhole.
 11

We require arctan s ∈ (− π2 , π2 ). Clearly

Z+ (χ) = Z− (−χ) = − sinh χ + Z− (χ) . (4.20)

Function Z+ (χ) asymptotically behaves as

Z+ |χ→−∞ = 0 , Z+ |χ→+∞ ≈ − sinh χ , (4.21)

and function Z− (χ) as

Z− |χ→−∞ ≈ sinh χ , Z− |χ→+∞ = 0 , (4.22)

R- R+
and
whm ρ- whm ρ+
1
Z± 0 (0) = ∓ . (4.23)
2

z- z+
χ

R- R+
-1

j
FIG. 9. Potential Uhom (top) and field strength whom (bot- -2

tom) of asymptotically homogeneous field in R+ . The left -3

diagram shows the induced field in space R− , and the right

-4
one the modified homogeneous field in R+ . Thick segments
- (χ) + (χ)
on the ρ± axes represent mouths of the wormhole. Angular -5

coordinate ϕ is ignored and can be recovered by a rotation

around the axes ρ± = 0.
FIG. 10. Functions Z− (χ) and Z+ (χ).

D. Wormhole in a homogeneous gravitational field To check the validity of the boundary condition of the
field Uhom at z → ∞ in R+ , we use relation (4.20)
Let us consider a homogeneous gravitational field in a
flat space. Its potential is Uhom = wo sinh χ cos ϑ − wo Z− (χ) cos ϑ . (4.24)

U = wo z . (4.16) In R+ , where z = sinh χ cos ϑ, the first term correctly

reproduces the asymptotic behavior (4.16), since the sec-
The acceleration field has constant value wo and is di- ond term falls down at χ → +∞.
rected along z-axis One can interpret this result as follows. The ring
wi = ∇i U = wo ∇i z . (4.17) wormhole changes the homogeneous gravitational field
(4.16), which was originally introduced only in R+ .
Let us assume now that a ring wormhole leading to When we consider it in the full wormhole space R,e we call
another space is immersed in this gravitational field. For it truncated field Ut . In the presence of the wormhole,
simplicity, we assume the wormhole is in the origin z = 0, it will be modified by an induced field Ui . The contribu-
with its mouth orthogonal to z-axis. To find how such tion of the induced field in the region R+ is described by
a wormhole modifies the field, one needs to solve the term −wo Z− (χ) cos ϑ, while in the region R− it is given
Laplace equation in R e space which has an asymptotic by field (4.18) penetrating through the wormhole, i.e., by
(4.16) at R+ -infinity. The corresponding solution is −wo Z+ (χ) cos ϑ. Together,
Uhom = −wo Z+ (χ) cos ϑ . (4.18) Uhom = Ut + Ui , (4.25)
Here, we introduce functions7 Z± (χ)
where the truncated and the induced fields are
1 1

Z± (χ) = ∓ sinh χ − 1 + sinh χ arctan sinh χ . Ut = wo Θ(χ) sinh χ cos ϑ , (4.26)
2 π
(4.19) Ui = −wo Z− (|χ|) cos ϑ , (4.27)

Θ(χ) being the Heaviside step function.

7 Functions Z± are related to Legendre function Q1 evaluated on These fields have a natural physical interpretation.
the imaginary axis, with properly chosen asymptotic, and with a The truncated potential Ut vanishes in R− and describes
branch cut chosen such that it is analytical on the whole imagi- the homogeneous field in R+ . It is continuous every-
nary axis. where, including the wormhole mouth χ = 0, but it is
 12

not smooth here. Its field strength is discontinuous at ditions. Besides natural conditions of decreasing of the
χ = 0. In fact, it is discontinuous at both faces D< and gravitational potential U at the infinity and jump condi-
D> of the ring wormhole. The jump of the normal com- tions for it on the surface of the shell one needs to impose
ponent of the field strength is +wo and −wo at D< and additionally a condition of the regularity and absence of
D> , respectively, since the normal is pointing in a neg- matter at the discs which are identified in such a space.
ative z direction at D< and in a positive z direction at The latter condition means that the field’s strength ∇U
D> . Hence, Ut does not satisfy the homogeneous Laplace should be continuous on the discs. This problem belongs
equation at the wormhole. The jump corresponds to a to a class of so called mixed boundary value problem. A
massive thin shell at the wormhole mouth χ = 0 with a general discussion of such problems and references to the
surface density ∓ w 4π .
o
papers where they have been studied can be found in the
The potential induced by the wormhole, Ui , is also con- nice book [32].
tinuous everywhere and not smooth at the wormhole. Us- We will solve this problem only approximately, for the
ing properties of function Z+ one can find that it satisfies wormhole mouths sufficiently distant, `  1.8
the Laplace equation everywhere except the wormhole
throat χ = 0. The jump of the normal component of the
field strength at χ = 0 is in this case −wo and +wo at A. Large distance approximation – 0th order
D< and D> , respectively. So, it also corresponds to a
massive shell at the throat of the wormhole with a sur- Let’s start with the zeroth-order approximation. We
face density ± w 4π . Potential Ui has dipole character both
o
return to the first-order corrections below. In the ze-
in R− and R+ . roth order, we assume that the field of the shell near
Clearly, if one adds both potentials Ut and Ui , then the the second mouth D+ is negligible, and we can thus ig-
sum (4.24) is a solution of the source-free Laplace equa- nore any interaction with the wormhole. The potential is
tion everywhere, including the wormhole throat χ = 0. thus given by field (4.12) positioned around D− , which
Let us note that the potential Uhom defined by (4.18) is achieved by employing (χ− , ϑ− , ϕ− ) coordinates,
is finite and its value is proportional to wo (see Fig. 9). (
Hence, for small wo the field strength outside the ring m Z(χ− ) for χ− > χo ,
U= (5.1)
is uniformly small. However, the corresponding field Uo = m Z(χo ) for 0 < χ− < χo .
strength diverges at the ring. This divergency is a conse-
quence of the assumption that the corresponding cosmic However, even in the zeroth approximation, we have
string located at the ring is infinitely thin. As it was to extend the potential through the wormhole. We per-
discussed in Sec.II, the smearing of the distribution of form that separately in the domain V− around D− and
the matter of the string makes the background space- in V+ around D+ . We thus obtain two potentials, U−
time of the wormhole regular. One can expect that the and U+ , each of which satisfies the Laplace equation. On
gradient of the potential Ũ on such a smooth background the intersection V̄ , both potentials agree. In the zeroth-
would be regular in the vicinity of smeared string as well. order approximation, the potential U− is constant inside
The adopted weak-field approximation remains valid un- the shell around the mouth D− , and it thus leaks as con-
til a narrow ‘tube’ surrounding the smeared string. The stant Uo ‘behind’ the wormhole, into the domain V̂− . On
behavior of the gravitational field inside such a ‘tube’ the other hand, the potential U+ is negligible around D+
depends on details of the string’s matter distribution. in the zeroth order, and thus nothing leaks ‘behind’ the
wormhole into the domain V̂+ . See Fig. 12.
Clearly, the potentials differ around the mouths as
V. GRAVITATIONAL FIELD NEAR THE U+ = U− + ∆U , (5.2)
WORMHOLE IN SPACETIME WITH ONE
ASYMPTOTIC and in this order of approximation we obtain
∆U = −Uo = −m Z(χo ) in V̂ . (5.3)
In this section, we will study a wormhole connecting
two places in one asymptotically Minkowski spacetime, It indicates that the global potential is not well defined
and we will discuss a gravitational perturbation of such a and we deal with a locally static spacetime. The globally
space. We described the background wormhole spacetime well defined quantity is the geometry.
Rwh in Sec. II.
Now we consider the gravitational potentials U of the
8
massive thin shell located around the wormhole mouth We must distinguish two approximate schemes employed here.
D− . In oblate spheroidal coordinates (χ− , ϑ− , ϕ− ) con- First, we use the perturbative expansion of the gravitational in-
teraction, where we consider only the first-order approximation
structed around the mouth D− , the shell is given by
in which the gravity description reduces to the Newtonian poten-
χ− = χo > 0. It is highly nontrivial to find the full, con- tial satisfying the Poisson equation. Second, for the particular
sistent solution of the Poisson equation in the wormhole wormhole spacetime, we look for the gravitational potential of
spacetime Rwh . Namely, one needs to obtain a solution of the massive shell using the expansion for large mouths’ distance
the Laplace equation with specially chosen boundary con- `. Here, we discuss the later expansion.
 13

The perturbed metrics in the domains V− and V+ are

D-
D+
ds2± = −e2U± dt2± + e−2U± dl±
2
. (5.4)

V- V- We want that these two metrics describe one globally
well-defined geometry. Therefore, they have to coincide
shell
as tensors on both intersections V̄ and V̂ . Their iden-
tification will be simple on the intersection V̄ , since the
shell potentials U− and U+ coincide here. We can express
the both metrics just in global cylindrical coordinates

V+ V+ (ρ, z, ϕ) on V̄ , cf. (2.3).
D-
However, the potentials differ on other intersection V̂ .
D+
Therefore, we have to adjust the coordinate identification
in this domain to match both metrics. By identifying the
V-
proper time of the static observers, we get

eU− dt− = eU+ dt+ . (5.5)

 
V V V
Identifying spatial distance, we find
D-
D+
shell e−U− dl− = e−U+ dl+ . (5.6)
V+
Here, the background proper lengths dl− and dl+ should
FIG. 11. Coordinate domains around the wormhole in the be expressed using (2.8) in appropriate spheroidal coor-
presence of gravitational field of a thin massive shell near the dinates (χ− , ϑ− , ϕ− ) and (χ+ , ϑ+ , ϕ+ ), respectively. Re-
left wormhole mouth. Domains V− (top) and V+ (middle) are lation (5.6) then gives the identification of coordinates
covering together the whole space. The division into main (χ± , ϑ± , ϕ± ) in the domain V̂ , which guarantees the
part V̄± and ‘leaked’ part V̂± is indicated. The bottom dia- matching geometry. The resulting relation is, however,
gram shows two intersections V̄ and V̂ of these two domains. rather complicated.
The wormhole mouths D± are represented by the thick verti- The identification of coordinates is simpler if expressed
cal lines. The real size of both mouths is the same. However,
in cylindrical coordinates. For that, however, we have
since the gravitational field of the shell affects the geometry,
the global cartesian coordinates used for mapping to the dia-
to introduce gravitationally modified cylindrical coordi-
gram do not represent the geometry properly. The coordinate nates ‘leaked’ behind the wormhole. Let’s consider first
size of the both mouths is thus different, as can be observed coordinates (χ− , ϑ− , ϕ− ) around the wormhole mouth
in the diagram. The distance between mouths should be large D− . In the domain V̄− (above the mouth), they are
– diagrams do not reflect this feature accurately. related to the original global cylindrical coordinates
(ρ, z, ϕ) as (2.10). However, behind the mouth, in the do-
main V̂− , they are related by analogous relations (2.12)
V+ to new ‘leaked’ cylindrical coordinates (ρ− , z− , ϕ− ). Of
2
V- course, the flat background metric dl− is given on both
shell sides of the wormhole throat by (2.3)
U+ U+
2
dl− V̄−
= dρ2 + dz 2 + ρ2 dϕ2 ,
2
(5.7)
dl− V̂−
= dρ2− + dz−
2
+ ρ2− dϕ2− .

U- U- 2
Similarly, dl+ is related by (2.10) to (ρ, z, ϕ) in V̄+ and by
(2.12) to new ‘leaked’ coordinates (ρ+ , z+ , ϕ+ ) in domain
D- D+
V̂+ , and
FIG. 12. Diagram shows the zeroth order approximation of 2
dl+ = dρ2 + dz 2 + ρ2 dϕ2 ,
V̄+
potentials U± drawn along the axis z. The potentials U− and (5.8)
2
U+ are continuous on the domains V− and V+ , respectively. dl+ V̂+
= dρ2+ + dz+
2
+ ρ2+ dϕ2− .
They coincide on the intersection domain V̄ but differ by a
constant ∆U on the intersection domain V̂ . The fall-off of Comparing (5.7) and (5.8) using relation (5.6), we find
the potential is exaggerated to compensate a small distance
between mouths. Top part of the diagram indicates the do- e−U− dρ− = e−U+ dρ , e−U− dz− = e−U+ dz (5.9)
mains V− and V+ and the position of the wormhole and of
the massive shell. in domain V̂− , and

e−U− dρ = e−U+ dρ+ , e−U− dz = e−U+ dz+ (5.10)

 14

for any ∆t∗ , ρ∗ and ϕ∗ . The term ±0 distinguishes the

z- , ρ - V-
faces of the mouths.
It means that the identification of points through the
wormhole throat changes under the gravitational pertur-
bation. It causes a complication with the size of the
 
V V V wormhole. To keep the proper size of the wormhole
mouths a the same from both sides, one has to choose
a different coordinate radius for each mouth, a− , and
shell a+ , such that
V+ z+ , ρ +
a = e−U− a− = e−U+ a+ , i.e., a+ = e∆U a− . (5.15)
Notice that the change in the coordinate size of the
FIG. 13. Finite wormhole with distant mouths. Diagram
wormhole mouths is of the first order in the gravity per-
shows coordinates xj− and xj+ defined on the domains U− and
U+ , respectively. Since the potential modifies the geometry, turbations and it produces only next-order perturbations
the coordinates continuously extended through the wormhole to the gravitational potential. Thus, we can continue
does not match the original coordinates. to use the potential calculated on the original wormhole
background.
Let us summarize. We have perturbed the background
in V̂+ . with metric (3.3) and found the perturbed metrics (5.4).
Assuming, that the mouths of the wormhole are iden- Since the potentials U− and U+ cannot be extended to
tified at times t− = to− and t+ = to+ , the differential a global function on the background spacetime, the per-
relation (5.5) gives turbation has to be performed separately on domains V̄−
and V̄+ . In these domains we can use global cylindrical
t+ − to+ = e−∆U t− − to− ,



(5.11) coordinates (t, ρ, z, ϕ) to identify the background space-
time with the perturbed spacetimes.
valid in V̂ . The perturbed metrics ds2− and ds2+ can be easily
For spatial coordinates, we get that in domain V̂+ the extended to domain V− and V+ through the worm-
‘leaked’ cylindrical coordinates (ρ+ , z+ , ϕ+ ) are related hole throat, using, for example, spherodial coordi-
to the global cylindrical coordinates (ρ, z, ϕ) as nates (t− , χ− , z− , ϕ− ) and (t+ , χ+ , z+ , ϕ+ ), respectively.
These metrics match trivially on the intersection V̄ , as
`  ` can be seen in global coordinates (t, ρ, z, ϕ), since the
ρ+ = e∆U ρ , z+ + = e∆U z + . (5.12)
2 2 potentials are equal here.
However, since the potentials differ on the other in-
Similarly, in domain V̂− , the ‘leaked’ cylindrical coordi-
tersection V̂ , the metrics cannot be matched without a
nates (ρ− , z− , ϕ− ) are related to the global cylindrical
proper identifications of the coordinates. Because the po-
coordinates (ρ, z, ϕ) as
tentials U− and U+ differ only by a constant, cf. (5.2),
`  ` the metrics ds2− and ds2+ on V̄ differs only by a rescaling
ρ = e∆U ρ− , z− = e∆U z− − . (5.13) of time and spatial directions. It can be demonstrated
2 2
using cylindrical coordinates (t± , ρ± , z± , ϕ± ) introduced
See Fig. 13. ‘behind’ the wormhole. Namely, the global cylindrical
We can reformulate these relations in the rule, how the coordinates are related to ‘leaked’ cylindrical coordinates
global cylindrical coordinates defined on both side of the by a rescaling (5.12) and (5.13).
wormhole throat should be identified through the throat. However, this shows that we need to perform pertur-
We observe that t+ = t in V̄+ and t− = t in V̄− . Relation bation of the background in two domains independently,
(5.11) thus gives the identification of time coordinate t = and only after that we can glue together the global ge-
to− + ∆t∗ in V̄− with t = to+ + e−∆U ∆t∗ in V̄+ for any ometry representing a weak gravitational field. As the
∆t∗ . Similarly, we observe that the coordinates (ρ, ϕ) result, the spacetime is not globally static but only lo-
approaching the mouth D− from V̄− , are matched by cally static spacetime.
‘leaked’ coordinates (ρ− , ϕ− ) approaching D+ from V̂− .
The result for the time coordinate and relations (5.13)
thus give the identification B. Large distance approximation – 1st order

  h ` i
Let us now improve our approximation of the shell field
t, ρ, z, ϕ V̄− = to− + ∆t∗ , ρ∗ , − ± 0, ϕ∗
2 in the wormhole spacetime. We now include an interac-
l , tion of the field with the wormhole in the first order of 1` .
h ` i We start with some geometrical preliminaries.
t, ρ, z, ϕ V̄+ = to+ + e−∆U ∆t∗ , e∆φ ρ∗ , ∓ 0, ϕ∗
 
2 The oblate spheroidal coordinates (χ− , ϑ− , ϕ− ) and
(5.14) (χ+ , ϑ+ , ϕ+ ) introduced around the mouths D− and D+
 15

are related by (2.11). We evaluate it at the mouth D+ The corrected field of the shell thus reads
where χ+ = 0
U = Ush t + Ush i (5.19)
sinh χ− cos ϑ− = ` , cosh χ− sin ϑ− = sin ϑ+ . (5.16)
with Ush t given by (5.1) and Ush i given by (5.18).
Solving these equations one finds The constant wo in the induced field has to be fixed
s by the condition, that the field strength should by con-
`2 tinuous through the wormhole. Namely, we require
sin ϑ+ = + 1 sin ϑ− ≥ ` sin ϑ− . (5.17)
cos2 ϑ−
dU dU
The value of sin ϑ+ is restricted by 1, thus the angle ϑ− = ,
dz <
D+ dz >
D−
evaluated at D+ is small, ϑ− D+  1/`. Substituting (5.20)
this result back to (5.16) we obtain that up to the quan- dU dU
= .
tity of order O(`−2 ) the value of sinh χ− on the mouth dz >
D+ dz <
D−
D+ is constant and equal to `.
From the discussion of the first order approximation, In our approximation, we assume that the field is ho-
we expect that the potential U of the shell around the mogeneous and thus, it is sufficient to check these condi-
wormhole mouth D− is not globally defined. As we dis- tions just along the z-axis. Relation (2.12) gives
cussed in section III, we can define the potential U using
integration (3.12) along a path from infinity not cross- z ∓ `/2 = sinh χ± cos ϑ± , (5.21)
ing the wormhole throat. It will not be continuous at the
throat of the wormhole, but it is smooth everywhere else. where along the axis, values of cos ϑ± are just −1 or +1
And although the potential is not continuous through the depending on a position with respect of the wormhole
wormhole, the field strength −∇U is globally defined and mouths. Hence
smooth. We want to find an approximation of this po- dχ± cos ϑ±
tential. = . (5.22)
dz cosh χ±
In the zeroth approximation of the potential we as-
sumed that the field (5.1) is negligible at the mouth D+ . Using these relations, we can calculate z-derivatives of
Now, we take into account, that it is not negligible here. the potential U at the mouths of the wormhole. Actually,
We denote this potential as truncated, Ush t . Equipo- it turns out that it is the same on both faces D< and D>
tentials of Ush t are surfaces of χ− = const. Therefore, of the mouth. Using relations (5.1), (5.18), (4.2) and
it follows from the discussion above, that the field (5.1) (4.19) one finds
can be treated as homogeneous near the mouth D+ in
the next order of approximation. dU m wo

= − 1 + V` ,
We can thus use our discussion from Sec. IV D. Here dz D+ `2 + 1 2
we learnt that the truncated homogeneous field around (5.23)
dU wo

the mouth of the wormhole has to be compensated by = 1 + V` .
the field (4.27) adjusted to the mouth at z = 2` . dz D− 2
This field is nonvanishing on both sides of the worm-
hole, it thus changes the field not only in domain V̄+ Here V` is coming from the derivative of Z− 0 (`), i.e.,
‘above’ the mouth D+ , but also in domain V̂+ ‘below’ the form contribution of the induced field at distant worm-
mouth. Above the mouth D+ we use expression (4.27) hole mouth
employing spheroidal coordinates (χ+ , ϑ+ , ϕ+ ). Below 2

π `

4
the mouth, we transfer the induced field into spheroidal V` = arctan ` − + 2 =− + O(`−5 ).
π 2 ` +1 3π`3
coordinates (χ− , ϑ− , ϕ− ), using relations9 (2.11). In
(5.24)
these domains, values of relevant spheroidal coordinate
Now, it is easy to check that the continuity conditions
χ± are positive, and both sets of coordinates can be un-
(5.20) are satisfied provided
derstood as functions of the global coordinates (ρ, z, ϕ),
cf. (2.10). m m
wo = = 2 + O(`−4 ). (5.25)
We can thus write the induced field Ui outside the (`2 + 1)(1 + V` ) `
wormhole throat as

We see that the contribution of the induced field is rather
Ush i = wo Z− (χ− ) cos ϑ− − Z− (χ+ ) cos ϑ+ . (5.18) small.
The potential (5.19) is shown in figure 14. As we em-
phasized, it is not globally smooth, it has discontinuity
9 We do not employ here the corrections of the coordinate match- across the wormhole throat.
ing discussed after (5.6) in the previous section (which we also Similarly as we did in the zeroth order approxima-
expressed using cylindrical coordinates in (5.14)) since it would tion, it can be extended as a solution of the homoge-
produce corrections of higher order. neous Laplace equation through each wormhole mouth,
 16

V+
V-
shell

U+ U+

U- U-

ρ+ D- D+

FIG. 15. Diagram shows the first order approximation of

potentials U± drawn along the axis z. The potentials U−
and U+ are continuous on the domains V− and V+ , respec-
tively. The potentials include the induced field correcting the
zeroth-order contribution on the wormhole throat. The both
D- D+ potentials coincide on the intersection domain V̄ but differ by
z+ a constant ∆U on the intersection domain V̂ . Top part of the
diagram indicates the domains V− and V+ and the position
FIG. 14. Potential U [top] and field strength [bottom] of the of the wormhole and of the massive shell.
thin massive shell localized around the wormhole mouth D− .
The potential depicted here is smooth outside the wormhole
throat, but it is discontinuous at the throat. The zeroth order The both potentials U− and U+ along z-axis are de-
value Ush t of the potential corresponds to the shell in space picted in Fig. 15.
without a wormhole. The correction given by induced field We can estimate the difference between both poten-
Ush i makes the field strength smooth through the wormhole tials. Analogically to (5.2) we obtain the correction to
throat. (5.3)
m
obtaining thus two potentials U− and U+ defined in do- ∆U = Ush t z= ` − Uo = −Uo − + O(`−2 )
2 `
mains V− and V+ , respectively. However, in this case we  1  (5.29)
have to adjust the induced field Ush i to guarantee also = m arccot sinh χo − + O(`−2 ) .
`
the continuity of the potential. It can be achieved by
adding a suitable constant to the extended field ‘behind’ Since the jump ∆U plays a role in the identifications of
the mouth. We obtain the geometry through the wormhole (5.14), namely in the
( identification of the time coordinate (5.11), it is worth to
Ush t + Ush i in V̄+ , investigate it more from a different point of view.
U+ = (5.26)
Ush t z= ` + Ush i in V̂+ ,
2

and C. Invariant IC calculation

(
Ush t + Ush i in V̄− ,
U− = (5.27) In a multiply connected space with a non-potential lo-
Uo + Ush i in V̂− . cally static gravitational field one can define the following
integral
Here Uo = Ush t |z=− ` = mZ(χo ) is the value of the
2 I
truncated potential inside the shell ‘leaking’ behind the
mouth D− – the effect which we encountered already in IC = wi dxi . (5.30)
C
the zeroth order approximation..
The constant Here C is a closed contour, and wi is a three-dimensional
m co-vector of the acceleration. This integral is a topologi-
Ush t z= ` = −m arccot ` = − + O(`−3 ) (5.28) cal invariant in the following sense. The integral (5.30) is
2 `
the same for any two closed paths, which can be obtained
‘leaking’ through the mouth D+ into V̂+ appears only in one from another by a continuous transformation.
the first order of approximation. Comparing with (5.25), Clearly, for a globally potential field wi = U,i , this in-
we observe that it is actually more significant than the tegral vanishes. For a locally potential field which is
induced field Ush i itself. not potential globally, it estimates the non-potentiality of
 17

the field wi . If evaluated along a path between different

points which belongs to a domain, where the potentiality
holds, it gives a potential difference. Therefore, we can 6
conclude, that for a closed path it estimates a potential
difference which is an obstacle for a global definition of 9
the potential, 5
8
∆U = IC . (5.31)
Signal to
Let us calculate the invariant IC for situation discussed 4 7 the past
in this section: the wormhole space Rwh with one of its
6
mouths surrounded by a massive thin shell. We choose
a closed C which passes through the wormhole. Since 5
its value does not depend on a special choice of such 3
a path we use for C the contour shown at Fig. 16. It 4
consists of three parts C− , C+ and C∞ . We choose C−
and C+ to p be intervals along z− axes and C∞ to be a 2 3
circle R ≡ ρ2 + z 2 =const connecting the end points 2
of these intervals. In the limit R → ∞ the contribution Time moments
identified via
of the part C∞ to IC vanishes. 1 wormhole 1


C FIG. 17. Red dashed lines mark the connection of events via
wormhole. Before the massive shell was turned on around
D− the events are globally synchronized. After an adiabatic
 = o
formation of a massive shell the pace of time at D− slows
down as seen by an observer at infinity and the mouths of the
wormhole become desynchronised from his/her point of view.
z When the time difference accedes `/c the chronology horizon
C− − C+ forms. After that the time machine appears.
2 2

FIG. 16. Contour C VI. TIME MACHINE FORMATION

For the other parts of the contour C one has

Z ∞ Let us show that in the spacetime with a ring worm-
χ =∞ hole, where one of its mouthes is surrounded by mat-
I− = wχ− dχ− = U |D−<
0 − ter, closed timelike curves are created. Assume that
= −Ush t |χ− =χo − Ush i |D−
<,
D− mouth of the wormhole is inside the massive shell
Z ∞ (5.32) while D+ mouth is located far away outside the shell
I+ = −
χ =∞
wχ+ dχ+ = −U |D+> (see Fig. 17) where its gravitational field is small. Be-
0 + cause these mouthes are identified through the wormhole
=Ush t | >
D+ + Ush i |>
D+ , the proper times at D− and D+ are the same. Hence,
the coordinate times t− at D− and t+ at D+ are related
The induced potential Ush i (5.18) is a symmetric func- as (5.11). In this setup ∆U = IC > 0. It means that for
tions with respect to reflection χ−  χ+ , ϑ−  π − ϑ+ the same proper time interval the coordinate time at the
and, hence, satisfies the condition disk D− which is inside the massive shell spans more than
Ush i |D−
< = Ush i | > , (5.33) that at D+ . Let us assume that clocks at the disks are
D+
synchronized at the moment t = 0. If a particle enters
therefore it cancels in the sum of I− and I+ . Then we D− at time t1 it appears from D+ at slightly different
obtain time t2 . Denote the time gap ∆t = t2 − t1 . Then in such
IC = − m[Z(χo ) + arccot `] . (5.34) a process the time gap is negative

∆t = − eIC − 1 t2 < 0 .
 
For `  1 we finally get (6.1)
h 1i Let us consider the following experiment. Suppose a
IC =m arccot sinh χo − + O(`−2 ) . (5.35)
` null ray emitted at some time t0 at the point with coordi-
One can check that when sinh χo  ` the integral IC nate z = 0 propagates along z−axis to the left, enters the
is positive. The main contribution comes from the first disc D− at time t1 and after it appears from D+ at the
term. The work along the closed path passing through time t2 returns to the point of its emission at the time t3 .
the wormhole does not vanish and the gravitational field Let us find how the corresponding time t3 (as measured
is non-potential. by an observer at infinity) depends on t0 . For the metric
 18

(3.1) the equation of the null ray moving along z−axis is Let us emphasize, that the main conclusion concern-
ing a closed time-like curves formation in the wormhole
dz spacetime is a rather robust property. Such curves arise
= −e2U (z) , (6.2)
dt as a result of the asymmetry of the mass distribution
where U (z) is a value of the potential U on z-axis. A sign surrounding its two mouths.
minus in the right-hand side of this equation indicates For more details concerning non-potential gravita-
that at both stages (motion to D− and motion from D+ ) tional field and time machine formation in such fields
the ray moves in the direction opposite to the direction of see [16, 23–25].
z-axis. Using (6.2) one finds that the time of the arrival
of the ray to D− is
Z 0 VII. DISCUSSION
t1 = t0 + e−2U dz . (6.3)
−`/2 In the present paper we study the gravitational field
Similarly, time t3 of arrival to the original point z = 0 in a spacetime with a traversable ring wormhole. We
after the ray appears from D− at time t2 is considered two types of such wormholes: (i) a wormhole
connecting to flat spaces, and (ii) a wormhole connecting
Z `/2 two distant domains in the same space. We focused on
t3 = t2 + e−2U dz . (6.4) study solutions of the gravity equations in the weak field
0
approximation in the presence of matter in these spaces.
Using equations (6.1), (6.3) and (6.4) one finds For the wormhole connecting two flat spaces we demon-
strated that there exist zero mode solutions describing
t3 − t0 = − eIC − 1 t2 + B ,
 
the gravitational field trapped by the wormhole. Such a
Z `/2
(6.5) field obeys the homogeneous Laplace equation. One can
B= e−2U dz > 0 . relate them to a proper choice of the matter distribution
−`/2
located at the infinity either of R+ , or/and R− spaces.
For a chosen direction of null ray motion IC > 0 . Hence, We also obtained a solution for the gravitational field of
t3 −t0 is a decreasing function of the time t2 at D+ which the massive thin oblate spheroidal shell confocal to the
for large ` is close to the time of an observer at infinity. disc representing the ring wormhole. A main important
This equation shows that t3 becomes lesser than t0 when feature of such a solution is that one cannot impose a
condition that the corresponding gravitational potential
B U vanishes at both infinities in R± . There always exists a
t2 ≥ T, T = . (6.6)
e IC − 1 non-vanishing difference of these potentials which is pro-
Because B > 0 and IC > 0 this means that for sufficiently portional to the mass m of the shell and depending on its
late time t2 the null ray arrives at the initial point z = 0 size. We also found an exact solution for the case when
earlier than it was emitted and a closed time curve is a ring wormhole is immersed in the gravitational field
formed. which is homogeneous at R+ infinity. Such a solution
In the weak field approximation adopted in this paper contains the dipole type component which describes the
IC is small and B ≈ `. Then formation of the time modification of the homogeneous field by the wormhole.
machine happens after time In section V we study the gravitational field of a mas-
sive thin shell surrounding one of the mouths of the ring
` wormhole connecting two distant domains in a single
T ≈ . (6.7)
IC space. We solved the gravity equation in the approxima-
tion when the distance L between the mouths is much
Let r = sinh χo be a characteristic size of the massive thin larger that the ring radius a. A main property of this
shell surrounding the D− mouth of the ring wormhole. solution that the gravitational potential U becomes a
Then the characteristic time of the transformation of the multi-valued function. This happens because the space is
wormhole into the time machine is multi-connected. While the strength of the gravitational
T ≈ r`/m . (6.8) force, w~ is well defined and unique, a solution of the
equation ∇U = −w ~ does not posses this property. One
Restoring the dimensions one can write this expression can integrate this equation and find the function U along
in the form any chosen curve connecting two points. But the value of
RLc U may be different for two different paths which cannot
T ≈ . (6.9) be transformed one to the other by means of a contin-
GM uous transformation. As a result the gravitational field
Here T is the characteristic time of closed timelike curves is non-potential. In such a case there exists a growing
formation in the initially traversable ring wormhole sur- time gap for clocks synchronized along specially chosen
rounded by a massive thin shell of radius R and mass M . non-contractible closed paths. This implies that a closed
L is the distance between mouths. timelike curves formation occurs in such a space. This
 19

is a mechanism which transforms an original traversable The four-acceleration of these static observers,
wormhole without a time-gap into a time machine.
Let us emphasize that above described properties of wµ = uµ;ν uν , (A4)
the ring wormholes are similar to well known properties
of the ‘standard’ wormholes with a spherical topology of is orthogonal to uα . Using the orthogonality, the Killing
their throat. A main difference is that for the ring worm- equation ξ(µ;ν) = 0, and staticity uα U,α = 0, one finds
hole an observer passing through it moves in a flat (or that U is a local potential of wµ
practically flat spacetime), while in the case of ‘standard’
wormholes he/she should pass a domain filled with the wα = U,α . (A5)
matter violating the null energy condition. In this paper
we focused on the classical aspects of the ring wormhole We call U the gravitational potential, wα the accel-
model. It is well known that for the ‘standard’ wormholes eration field, and −U,α the gravitational field strength.
the quantum effects are important and may dramatically Since wα is orthogonal to the time direction uα , it can
change their property in the regime of the time-machine be restricted to spatial components. The field strength is
formation. Namely, it was shown [20, 21] that at the mo- a fictitious force that one assumes in the static frame to
ment close to the time when closed timelike curves are explain a tendency of free observers to move with respect
formed, the renormalized quantum average of the stress- to the frame. Non-moving static observers have to ‘com-
energy tensor of a quantum field infinitely grows. Hawk- pensate’ this force by a real force equal to wj per unit
ing formulated a chronology protection conjecture [22], mass, cf. (3.7). Of course, in the spacetime description,
according to which the back-reaction of quantum effects the static observers move along non-geodesic trajectories
should always forbid formation of closed timelike curves. with four-acceleration wα caused by the real force.
There exists quite a lot of papers where different aspects
of the role of quantum effects in wormholes and time
Weak static gravitational field
machines are discussed (see e.g. [33–40]). It would be in-
teresting to study quantum effects in the ring wormholes
and time-machines. These models are rather simple from Now we want to study a weak gravitational field on a
the mathematical point of view and they may provide one flat background. We assume that the resulting spacetime
with a simple analytical tools for study such a quantum is static, as described above, and we want to formulate
mechanical problem. the field equation in an approximation of a weak field.
Since the metric (A1) is invariant under time reflection
t → −t, the extrinsic curvature of a surface t = const
ACKNOWLEDGMENTS vanishes, and Gauss-Codazzi equations imply that
1
The authors V.F. and A.Z. thank the Natural Sciences Gtt = − R . (A6)
and Engineering Research Council of Canada and the Kil- 2
lam Trust for their financial support. P.K. was supported Here, Gµν is Einstein tensor of the spacetime metric ds2
by Czech Science Foundation Grant GAČR 22-14791S. and R is a scalar curvature of the spatial metric dq 2 (see,
e.g., (B.2.6) in [25]).
We choose a stress-energy tensor for a static distribu-
Appendix A: Static spacetime
tion of matter in the form
Metric T µν = ε uµ uν . (A7)

Let us consider a static spacetime. It is locally de- Here ε is the mass density. In the metric (A1) one has
scribed by metric √
dm = ε q d3 x . (A8)
ds2 = gµν dxµ dxν = −e2U dt2 + dq 2 ,
Using Einstein equations and relation (A6) one gets
dq 2 = qij dxi dxj , (A1)
∂t U = 0 , ∂t qij = 0 . R = 16πε . (A9)
We denote ξ µ = δtµ a time-like Killing vector and uµ a Let us consider a special case when the spatial geom-
four-velocity of static observers moving along this Killing etry is conformally flat with a factor given by e−2U
vector,
dq 2 = e−2U dl2 , (A10)
uα = e−U ξ α , uµ uν gµν = −1 . (A2)
Obviously, dl2 = δij dxi dxj being a flat spatial metric. Then one has

uα = −eU t,α . R = 2e2U 24U − (∇U )2 .

(A3) (A11)
 20

Here, ∇, 4 and (. . . )2 are the covariant derivative, the Then, equation (A13) for the potential takes the form
Laplace operator, and the square with respect to the flat
spatial metric dl2 . 4U = 4πeU µ . (A19)
We say that the gravitational field is weak if
Let us note that metric (A1), (A10) and the field equa-
|∇U ∇U |  |∇∇U | . (A12) tion (A19) are invariant under the following scaling trans-
formation
In such a case the second term in the brackets of (A11) U = Û + ∆U ,
can be omitted and one has R = 4e2U 4U . Equation
(A9) implies that in the weak field approximation the t=e −∆U
t̂ , dl2 = e2∆U dˆl2 , (A20)
mass density ε is of the same first-order as R. µ=e −3∆U
µ̂ .
The second term in the bracket can be neglected in our
approximation, but the prefactor is not negligible. From
(A9) we obtain the equation for potential U Small value of potential U case

4U = 4πe−2U ε . (A13)
Let us emphasize, that in the above discussion we as-
In fact, in the full theory, for a static distribution of the sume that the potential U is slowly changing in space,
matter the stress-energy tensor should be slightly mod- but we do not assume that its value is small.
ified by adding a contribution responsible for repulsive To simplify the presentation we make an additional
forces keeping the matter at rest. Indeed, for the stress- assumption that the potential U is uniformly small, that
energy tensor (A7) one has is its value satisfies the relation

|U |  1 . (A21)
T µν;ν = ε wµ , (A14)
everywhere in space.
This means that a static distribution of matter is pos- In this approximation equation (A19) takes the form
sible only when some repulsive forces are present that
compensate the gravity attraction. This can be achieved 4U = 4πµ . (A22)
by including the pressure p and writing Tµν in the form
In our perturbative approach it is the same as ε in the
T µν = (ε + p)uµ uν + pg µν . (A15) leading order

For this tensor µ=ε, (A23)

T µν;ν = (ε + p)wµ + p;µ . (A16) but for U that are not small, they differ by the factor
e−3U , cf. (A18)..
In the weak field approximation, both ε and wα are
small. We say that each of this quantities is of the first
order in the weak field approximation. To satisfy the Appendix B: Locally static spacetimes
conservation law T µν;ν = 0, the pressure should be of the
second order. Such pressure does not contribute to the In this appendix we demonstrate how to describe a
Einstein equations in the leading first-order approxima- locally static spacetime without introducing initially a
tion. static time coordinate t and a gravitational potential U
Let us discuss now the other Einstein equations (be- as we did in Eq. (A1) in the previous Appendix.
sides (A11)) in the same leading order approximation. It Let us consider a spacetime M with metric g and as-
is easy to check that all time-spatial components of the sume that it admits two vector fields, a future directed
Einstein tensor Gtj vanish identically, and the spatial unit vector u and a vector w which obey the equations
components Gij are of the second order in ∇U (see, e.g.,
(B.2.6) in [25]). This means that in the leading order of uµ uµ = −1 ,
the weak field approximation the spatial components of uµ;ν = −wµ uν , (B1)
the Einstein equations are also satisfied. w[µ;ν] = 0 .
When working in the flat space, it is natural to use
a mass density in (A8) normalized on the flat volume We call such a spacetime locally static. The relations
element dV = d3 x. The relation to ε follows from (B1) imply that

wµ = uν uµ;ν , wµ uµ = 0 . (B2)
dm = µ dV . (A17)
Denote by γ integral lines of u
Thanks to rescaling (A10) we obtain
dxµ
µ = e−3U ε . (A18) = uµ . (B3)
dτ
 21

Normalization condition u2 = −1 implies that τ is a

proper time along a worldline γ. p
The last of the relations in (B1) implies that the 1-
form wµ dxµ is closed. Let p0 be a point in M and Ωp0
C
be a simply connected region which contains this point. p
Then there exist such a function U in Ωp0 that  p
p
wµ = U,µ . (B4)

Let p be a point in Ωp0 and Cp0 p be a path in Ωp0 con-

necting points p0 and p. Then
Z  ( ps )
ps C
U (p) = U (p0 ) + wµ dxµ . (B5)
Cp0 p

The function U called the gravitational potential is de- FIG. 18. Projection of M onto the space of Killing trajectories
fined in Ωp0 up to a constant U0 ≡ U (p0 ). Let us denote S.

z µ = eU uµ ,  = ±1 . (B6)
The integral lines γ provide a foliation of the region
Using relations (B1) one gets Ωp0 . Each of the Killing trajectory can be specified by
giving 3 numbers y i , i = 1, 2, 3, and a point p on a given
zµ;ν = eU (−uν wµ + uµ wν ) . (B7) trajectory can be specified by giving its Killing time value
t. This defines coordinates xµ = (t, xi ) in Ωp0 . In these
For  = −1 one has z[µ;ν] = 0, while for  = 1 one has
coordinates the metric takes the form
z(µ;ν) = 0. We denote
ds2 = −e−2U dt2 + hij dy i dy j , ∂t U = ∂t hij = 0 .
ξ µ = eU uµ , η µ = −e−U uµ . (B8) (B16)
In what follows we assume that ξ 2 does not vanish
Thus one has
and remains negative everywhere in M . This property
ξ(µ;ν) = 0, η[µ;ν] = 0, ξ 2 = −e2U . (B9) excludes a case of black holes. But in the spacetime con-
sidered in the present paper this property is valid.
The first of these equations shows that ξ is a Killing vec- Following [41] we denote by S a collection of all tra-
tor, while the second equations implies that there exists jectories of ξ in M . There exists a natural map Ψ from
such a scalar function t in Ωp0 that M to S which is defined as follows: For any point in M
there exists a Killing trajectory passing through it which
ηµ = t,µ . (B10) determines a point in S (see Fig. 18). Three-dimensional
space S has the induced metric
It is also easy to check that
dh2 = hij dy i dy j . (B17)
ξ µ ∂µ U = 0 . (B11)
Since Lξ w = 0 and ξ · w = 0, the 4-vector of accel-
Consider displacement dxµ along a worldline γ. Then eration w has a natural projection w ~ on S. In (t, y i )
one has coordinates one has w = (0, w).~ Let us consider a path
dt = ηµ dxµ = −e−U uµ dxµ CpS0 p connecting two points pS0 and pS in S. We define
(B12) a 3D potential U (pS ) as follows
= −e−U uµ uµ dτ = e−U dτ . Z
One also has U (pS ) = U0 + wi dy i , (B18)
CpS
0p
µ µ
dx dx
= eU = eU uµ = ξ µ . (B13) where U0 is a constant which is chosen as a value of the
dt dτ
potential at pS0 . The quantity U (pS ) − U0 can be in-
Hence terpreted as a work done by the gravitational field on a
particle of unit mass for its motion along the path C S .
ξ µ ∂µ = ∂t , (B14) Consider a closed path C S which starts at some point
p0 and returns to it again and denote
and t is the Killing time parameter. Let us emphasize I
that the gravitational potential U in Ωp0 is defined up to
a constant. A shift U → U + c results in the rescaling IC S = wi dy i . (B19)
CS
both the Killing vector and Killing time
Since 3D 1-form w is closed, the Stokes’ theorem implies
ξ µ → e−c ξ µ , t → ec t . (B15) that IC S depends only on the cohomology class of the
 22

path C S . If S is a simply connected manifold the invari- that the proper time gap is constant for closed paths for
ant IC S vanishes. If S is not simply connected there may which the invariant IC vanishes.
exist paths for which IC S 6= 0. For such a closed path If a locally static spacetime M is simply connected so
work done by the gravitational field does not vanish. We that IC = 0 for any closed path in it, then
call such a field non-potential. For this field the potential
U is a multi-valued function. (i) gravitational potential is globally uniquely defined,
up to a constant;
A path C S connecting pS and p0S in S can be lifted
p
to M as follows. Consider a Killing trajectory corre-
sponding to pS and choose a point p on it. If in local p
coordinates y i the equation of C S is y i = y i (λ), then d 
the equation of the path S in (t, y i ) coordinates in M is p
xµ = (tp , y i (λ)), where tp is the time coordinate of point
p. We require that the path C has a property that its  p
tangent vector is orthogonal to uµ . For a given initial
point p and chosen path C S in S this property uniquely  p
define C. It is easy to see that p

Z Z
d
p
wi dy i = wµ dxµ . (B20)
CS C

FIG. 19. Time gaps.

Let C S be a closed path in multiply-connected S pass-
ing through a point pS and C̃ be its lift to M passing
through p. The end point p0 of C̃ lies on the same Killing
(ii) gravitational field is potential;
trajectory as the initial point p but in a general case it
does not coincide with it. Denote by ∆C a path from (iii) Killing vector field is globally uniquely defined after
p0 to p along their common Killing trajectory. The path choice of its norm in some point.
C = C̃ + ∆C R is closed. Since w is orthogonal to a Killing
trajectory, ∆C wµ dxµ = 0, and one has In other words, this spacetime is (globally) static.
Let us now repeat the above consideration, but do not
I I assume that the invariant IC for a chosen closed path
IC = wi dy i = wµ dxµ . (B21) vanishes. Once again the Killing time interval dt for two
CS C close lifts of C S with starting points at p and p̃ remains
constant along the path. Denote by dτ proper time in-
The value of this integral does not depend on a choice terval between initial points p and p̃, and by dτ 0 proper
of the initial point p on the path, and relation (B21) is time interval between final points p0 and p̃0 (see figure 19).
valid not only for a chosen lift of the path C S but for Then using (B12) one gets
any closed path C in M such that its projection on S 0

coincides with C S . dτ = eU (p) dt, dτ 0 = eU (p ) dt . (B22)

Let us consider a closed path C S with a starting point Thus
pS for which IC = 0 and chose two close points p and p̃ 0
on the Killing trajectory determined by pS . They have dτ 0 = eU (p )−U (p) dτ = exp (IC )dτ . . (B23)
the same coordinates y i . Let us denote by dt time differ-
ence between p and p̃ and assume that dt is small. By Denote by ∆τp the proper time gap for the point p and
construction, dt remains the same for the paths C and by ∆τp̃ the proper time gap for p̃. Then one has (see
C̃ with the initial points p̃ and p, respectively. In par- figure 19)
ticular this is true for the end points p0 and p̃0 of these
curves. Since IC = 0, the value of the potential U de- d∆τ ≡ ∆τp̃ − ∆τp = dτ 0 − dτ = (exp (IC ) − 1)dτ . (B24)
fined along the paths by equation (B18) at the end point
coincides with its value at the initial point, one concludes Hence, the proper time gap d∆τ linearly grows with time
that the initial proper time interval dτ between p and p̃ τ so that
is the same as the final proper time interval dτ 0 between d∆τ
p0 and p̃0 . We call the proper time interval ∆τp between = exp (IC ) − 1 . (B25)
dτ
the initial point p of C and its final point p0 the proper
time gap (see figure 18). The above discussion implies For more details see [16, 24, 25].
 23

[1] G. W. Gibbons and M. S. Volkov, “Zero mass limit of Rev. D 43, 3929–3947 (1991).
Kerr spacetime is a wormhole,” Phys. Rev. D 96, 024053 [21] V. P. Frolov, “Vacuum polarization in a locally static
(2017). multiply connected space-time and a time machine prob-
[2] G. W. Gibbons and M. S. Volkov, “Weyl metrics and lem,” Phys. Rev. D 43, 3878–3894 (1991).
wormholes,” JCAP 05, 039 (2017). [22] S. W. Hawking, “The Chronology protection conjecture,”
[3] G. W. Gibbons and M. S. Volkov, “Ring wormholes via Phys. Rev. D 46, 603–611 (1992).
duality rotations,” Phys. Lett. B 760, 324–328 (2016). [23] M. Visser, Lorentzian wormholes: From Einstein to
[4] C. W. Misner, “Wormhole Initial Conditions,” Phys. Rev. Hawking (1995).
118, 1110–1111 (1960). [24] V. P. Frolov and I. D. Novikov, eds., Black hole physics:
[5] D. R. Brill and R. W. Lindquist, “Interaction energy in Basic concepts and new developments (1998).
geometrostatics,” Phys. Rev. 131, 471–476 (1963). [25] V. P. Frolov and A. Zelnikov, Introduction to Black Hole
[6] J. A. Wheeler, ed., Geometrodynamics (New York, Aca- Physics (Oxford University Press, 2011).
demic Press, 1962). [26] C. W. Misner, K. S. Thorne, and J. A. Wheeler, Gravi-
[7] D. Gannon, “Singularities in nonsimply connected space– tation (W. H. Freeman, San Francisco, 1973).
times,” Journal of Mathematical Physics 16, 2364–2367 [27] B. Mashhoon, “Gravitoelectromagnetism: A Brief re-
(1975). view,” (2003), arXiv:gr-qc/0311030.
[8] C. W. Lee, “A restriction on the topology of cauchy sur- [28] S. J. Clark and R. W. Tucker, “Gauge symmetry and
faces in general relativity,” Communications in Mathe- gravitoelectromagnetism,” Class. Quant. Grav. 17, 4125–
matical Physics 51, 157–162 (1976). 4158 (2000).
[9] G. J. Galloway, “Minimal surfaces, spatial topology and [29] S. Chandrasekhar, Ellipsoidal figures of equilibrium (Yale
singularities in space-time,” Journal of Physics A: Math- University Press, 1969).
ematical and General 16, 1435 (1983). [30] L. D. Landau and E. M. Lifshitz, Electrodynamics of Con-
[10] J. L. Friedman, K. Schleich, and D. M. Witt, “Topologi- tinuous Media (Pergamon, New York, 1984).
cal censorship,” Physical Review Letters 71, 1486 (1993). [31] W. Israel, “Singular hypersurfaces and thin shells in gen-
[11] M. S. Morris and K. S. Thorne, “Wormholes in space- eral relativity,” Nuovo Cim. B 44S10, 1 (1966), [Erra-
time and their use for interstellar travel: A tool for teach- tum: Nuovo Cim.B 48, 463 (1967)].
ing general relativity,” Am. J. Phys. 56, 395–412 (1988). [32] I. N. Sneddon, Mixed Boundary Value Problems in Poten-
[12] M. S. Morris, K. S. Thorne, and U. Yurtsever, “Worm- tial Theory (North-Holland Publishing Company, 1966).
holes, Time Machines, and the Weak Energy Condition,” [33] M. Visser, “Hawking’s chronology protection conjecture:
Phys. Rev. Lett. 61, 1446–1449 (1988). Singularity structure of the quantum stress energy ten-
[13] M. Visser, “Traversable wormholes: Some simple exam- sor,” Nucl. Phys. B 416, 895–906 (1994).
ples,” Physical Review D 39, 3182–3184 (1989). [34] H. D. Politzer, “Simple quantum systems in space-times
[14] D. Hochberg and M. Visser, “Geometric structure of the with closed timelike curves,” Phys. Rev. D 46, 4470–4476
generic static traversable wormhole throat,” Phys. Rev. (1992).
D 56, 4745–4755 (1997). [35] M. Visser, “From wormhole to time machine: Comments
[15] J. Friedman, M. S. Morris, I. D. Novikov, F. Echev- on Hawking’s chronology protection conjecture,” Phys.
erria, G. Klinkhammer, K. S. Thorne, and U. Yurt- Rev. D 47, 554–565 (1993).
sever, “Cauchy problem in space-times with closed time- [36] S. V. Krasnikov, “On the quantum stability of the time
like curves,” Phys. Rev. D 42, 1915–1930 (1990). machine,” Phys. Rev. D 54, 7322–7327 (1996).
[16] V. P. Frolov and I. D. Novikov, “Physical Effects in [37] B. S. Kay, M. J. Radzikowski, and R. M. Wald, “Quan-
Wormholes and Time Machine,” Phys. Rev. D 42, 1057– tum field theory on space-times with a compactly gener-
1065 (1990). ated Cauchy horizon,” Commun. Math. Phys. 183, 533–
[17] F. Echeverria, G. Klinkhammer, and K. S. Thorne, “Bil- 556 (1997).
liard balls in wormhole space-times with closed timelike [38] P. F. Gonzalez-Diaz, “Ring holes and closed timelike
curves: Classical theory,” Phys. Rev. D 44, 1077–1099 curves,” Phys. Rev. D 54, 6122–6131 (1996).
(1991). [39] M. Visser, “The Quantum physics of chronology protec-
[18] A. Carlini, V. P. Frolov, M. B. Mensky, I. D. Novikov, tion,” in Workshop on Conference on the Future of Theo-
and H. H. Soleng, “Time machines: The Principle of self- retical Physics and Cosmology in Honor of Steven Hawk-
consistency as a consequence of the principle of minimal ing’s 60th Birthday (2002) pp. 161–176.
action,” Int. J. Mod. Phys. D 4, 557–580 (1995), [Erra- [40] R. Emparan and M. Tomašević, “Quantum backreaction
tum: Int.J.Mod.Phys.D 5, 99 (1996)]. on chronology horizons,” JHEP 02, 182 (2022).
[19] J. Dolanský and P. Krtouš, “Billiard in the space with a [41] R. P. Geroch, “A Method for generating solutions of Ein-
time machine,” Phys. Rev. D 82, 124056 (2010). stein’s equations,” J. Math. Phys. 12, 918–924 (1971).
[20] S. W. Kim and K. S. Thorne, “Do vacuum fluctuations
prevent the creation of closed timelike curves?” Phys.