Introducing Physical Warp Drives
IIntroducing Physical Warp Drives
Alexey Bobrick, Gianni Martire
Advanced Propulsion Laboratory at Applied Physics, 477 Madison Avenue, NewYork, 10022, U.S.E-mail: [email protected]
September 2020
Abstract.
The Alcubierre warp drive is an exotic solution in general relativity.It allows for superluminal travel at the cost of enormous amounts of matterwith negative mass density. For this reason, the Alcubierre warp drive has beenwidely considered unphysical. In this study, we develop a model of a generalwarp drive spacetime in classical relativity that encloses all existing warp drivedefinitions and allows for new metrics without the most serious issues presentin the Alcubierre solution. We present the first general model for subluminalpositive-energy, spherically symmetric warp drives; construct superluminal warp-drive solutions which satisfy quantum inequalities; provide optimizations for theAlcubierre metric that decrease the negative energy requirements by two orders ofmagnitude; and introduce a warp drive spacetime in which space capacity and therate of time can be chosen in a controlled manner. Conceptually, we demonstratethat any warp drive, including the Alcubierre drive, is a shell of regular or exoticmaterial moving inertially with a certain velocity. Therefore, any warp driverequires propulsion. We show that a class of subluminal, spherically symmetric warp drive spacetimes, at least in principle, can be constructed based on thephysical principles known to humanity today.
Submitted to:
Class. Quantum Grav.
1. Introduction
The classical-relativistic Alcubierre drive solution allows timelike observers to travelsuperluminally, although at the expense of using material with negative rest-massenergy (Alcubierre 1994); for recent reviews see also (Lobo 2007, Alcubierre &Lobo 2017). This solution is given by the following asymptotically-flat metric:d s = − c d t + (d x − f ( r s ) v s d t ) + d y + d z , (1)where r s = (cid:112) ( x − v s t ) + y + z . The metric describes a spherical warp bubble(a region deviating from the flat metric) moving along the x -axis with an arbitraryvelocity v s , which may be larger or smaller than the speed of light c .The shape function f ( r s ), present in the metric, defines the size and profile of thewarp bubble. For large distances r s from the bubble, f ( r s ) = 0 and the spacetime isflat. For small distances, r s ≈
0, the shape function f ( r s ) = 1, and the metric describesthe flat internal region of the bubble. In a coordinate system with x → x (cid:48) = x − v s t , thisinternal region is described explicitly by the Minkowski flat metric. The intermediate a r X i v : . [ g r- q c ] F e b ntroducing Physical Warp Drives f ( r s ) (cid:38)
0, corresponds to the spherical boundary of the warp. Inthe original study, (Alcubierre 1994) chose function f ( r s ), somewhat arbitrarily, as: f Alc ( r s ) = tanh( σ ( r s + R )) − tanh( σ ( r s − R ))2 tanh( σR ) , (2)where parameters R and σ − define the radius and the thickness of the transition fromthe internal to the external region, correspondingly.In the case of superluminal motion, the metric possesses a black hole-like eventhorizon behind the bubble and a white hole-like event horizon in front of it (Finazziet al. 2009). These event horizons arise because timelike observers cannot exit thesuperluminal ship in the direction ahead of it, and cannot enter it from behind. Inboth cases, the timelike observers would have to move superluminally when outsideof the ship.The energy density for the Alcubierre drive, as measured by Eulerian observers( u µ = (1 , , , T = − π ρ v s r s (cid:18) d f d r s (cid:19) , (3)where ρ ≡ y + z is the cylindrical coordinate.Despite its interesting properties which allow timelike observers to travel atarbitrary velocities, the Alcubierre drive solution possesses several drawbacks. Asnoted earlier, it requires negative energy densities, Equation (3), and thus violates theweak energy condition. Although negative energy densities are a general propertyof any superluminal drive (Olum 1998, Visser et al. 2000), the energy density isalso negative at subluminal speeds for the Alcubierre drive, even in the weak-fieldapproximation (Lobo & Visser 2004). Additionally, superluminal motion allows forclosed timelike loops, e.g. leading to grandfather paradox, and violates the null energycondition and causality, e.g. (Everett 1996), although the latter may be recovered atthe expense of Lorentz invariance (Liberati et al. 2002). When moving superluminally,the drive has an additional problem. It leads to quantum instabilities related to pairproduction near the horizon behind the warp as well as accumulation of particles atthe horizon in the front part of the warp (Finazzi et al. 2009).The Alcubierre drive is also problematic at sub-relativistic speeds. Firstly, itrequires unphysically large amounts of (negative) energy. For instance, it would requirean amount of negative energy comparable to the mass of the Sun to produce relativisticbubbles of ≈ meter sizes (Alcubierre 1994). Furthermore, such high negative energydensities do not appear even theoretically feasible. There are no known materialswhich would allow for gathering large amounts of negative energy in a controlledway. While zero-point vacuum fluctuations may produce negative energies in curvedspacetimes, for Alcubierre drives this situation is only possible if the walls of thebubble had thicknesses comparable to Planck scales. Such thin walls, however, requireextreme amounts of energy – comparable to the rest-mass energy of the Universe –as may be seen from Equation (3). Therefore, there is no physical way to create anAlcubierre drive (Pfenning & Ford 1997, Ford & Roman 1997).Finally, there is no proposed way of creating an Alcubierre drive, even if negativeenergy were available. In the original study, (Alcubierre 1994) suggested that thevelocity may be time-dependent, v s = v s ( t ). Indeed, Equation (3) retains its form evenfor time-dependent velocities. And, since v s = 0 corresponds to flat spacetime, it wasassumed that the Alcubierre drive might be generated through acceleration. However, ntroducing Physical Warp Drives v s = v s ( t ) corresponds to a time-variablestress-energy tensor which does not satisfy continuity equations. Alternatively, suchsolutions may be said to require an implicit dynamical field to effectively providepropulsion for the object, e.g. (Bassett et al. 2000). Generally, there are no self-consistent warp drive solutions proposed in the literature which can self-accelerate atall from zero velocities, not to mention gain superluminal speeds.Despite the rather extensive work on the properties of the Alcubierre drivesolution, it remains unclear which of the above issues are features of the Alcubierresolution specifically or more fundamental properties of warp drives as such. New warpdrive solutions have been introduced only in very few studies. (Van Den Broeck 1999)reduced the energy requirements of the Alcubierre drive to about the mass of theSun while satisfying the vacuum energy inequalities. The reduction was realizedby decreasing the externally measured size of the warp bubble down to 10 − mwhile keeping the internal volume constant. This solution satisfies the weak energyconditions, although it requires that classical gravity remains applicable down to suchsmall scales, at which it was never tested. However, as we show in Appendix A througha coordinate transformation, this solution is equivalent to the Alcubierre solution.(Nat´ario 2002) constructed a warp drive solution without space contraction orexpansion, contrary to the earlier assumption that it facilitated the movement ofwarp drives. (Nat´ario 2006) constructed a new subluminal warp drive solution in theweak-field regime, which required negative energies. (Loup et al. 2001) had previouslyintroduced a modified version of the Alcubierre drive intended to alter the rate oftime for the observers inside the bubble. However, their modification reduces to theoriginal Alcubierre metric, as we also show in Appendix A. Finally, (Lentz 2020)has recently proposed a warp drive metric claiming to have purely positive energyeverywhere in both subluminal and superluminal regimes, although without providingmeans to reproduce the study.The works above, to our knowledge, summarize all the modifications of theAlcubierre drive available in the literature. Superluminal travel had also been studiedby (Krasnikov 1998) and (Everett & Roman 1997). In these studies, the authorsintroduced Krasnikov tubes. Krasnikov tubes are ‘spacetime tunnels’ which allow forsuperluminal travel without violating causality, but only for round trips and withmuch larger energy requirements than the Alcubierre drive. Superluminal travel hasalso been discussed in the context of wormholes, e.g. (Garattini & Lobo 2007), andtime-machine metrics, e.g. (Fermi & Pizzocchero 2018), in all cases requiring negativeenergies. Finally, modified gravity theories may provide some desirable propertiesfor the Alcubierre drive. For instance, conformal gravity allows for construction ofAlcubierre solutions with positive energy only (Varieschi & Burstein 2013), whileextra-dimensional theories of gravity may reduce the energy requirements of the drive(White 2013).In this study, we show that the properties of the Alcubierre metric – in particular,its negative energy density and the accompanying immense energy requirements – arenot a necessary feature of warp drive spacetimes. In Section 2, we discuss that anygeneral warp drive, including the Alcubierre metric, may be thought of as a shell ofpositive- or negative-energy density material which modifies the state of spacetime inthe flat vacuum region inside it. In Section 3, we introduce, for the first time, the mostgeneral spherically symmetric warp drives. We show that the reason for the negativeenergy requirements of the Alcubierre metric and all the warp drives introduced inthe literature is, likely, the truncation of the gravitational field outside of the metric, ntroducing Physical Warp Drives
2. General class of warp drives
We define a general warp drive spacetime, as shown in Figure 1, as an asymptotically-flat vacuum region D out (background) which encloses a compact arbitrarily curvedregion D warp with a spherical topology (the warping region); D warp , in turn, enclosesa flat extended ‡ compact region D in with a trivial topology (the ‘passenger’ area).This definition covers both all the existing warp drive spacetimes, such as the one by(Alcubierre 1994), as well as trivial solutions, such as weak-field perturbations of theMinkowski spacetime. To formalize the difference between extreme and mild caseslike these, we shall consider the warped region D warp by using comoving referenceframes located inside and outside it, which will allow us to quantify how strong itsgravitational influence is.We shall focus on warp drive spacetimes which, intuitively, are stationary or non-changing from the ‘passenger’ point of view. More formally, we consider warp drivespacetimes which, by definition, admit a global Killing vector field, ξ , which is alignedwith the four-velocity of the boundary of the region ∂ D in (the inner boundary of theshaded region in Figure 1). Such a field establishes the global frame of rest withrespect to the warped region. Subsequently, whenever we discuss a physical motionof an observer relative to the warp drive, we consider, locally, the motion relative tothe global reference frame defined by the field ξ . Furthermore, using this vector fieldwill allow us to apply the well-known techniques suitable for spacetimes possessing atimelike Killing vector to analyse and classify warp drive spacetimes.Since the internal region D in is flat, vector field ξ is constant in that region.Therefore, in the internal region, this vector field ξ defines a reference frame (tetrad)for the internal observer § O in . Similarly, in the asymptotic infinity, vector field ξ defines the frame of the remote comoving observer O out , co . This comoving referenceframe O out , co , as we further argue, is the most natural frame against which one cancompare the frame O in inside the warped region. Finally, this remote comoving ‡ To avoid effectively including general asymptotically-flat spacetimes in the defintion, we requirethat the extent of the region D in should be non-vanishing and physically interesting. In other words,the inner region D in should be large enough to allow observers to conduct physical experiments ofdesired scales. § While, in relativity, physical observers are timelike by definition, we shall occasionally bementioning timelike and spacelike observers for convenience. By these, we shall understand physicalobservers and spacelike frames, correspondingly. ntroducing Physical Warp Drives Figure 1.
A schematic illustration of a warp-drive spacetime. Thespacetime consists of three regions: Asymptotically-flat vacuum background D out (background), general stationary curved region D warp with a spherical topology(the warping region) and a flat inner region D in (‘passenger’ space). Any suchspacetime, including the Alcubierre drive metric, is realised through a shell ofordinary or exotic negative energy density material filling the warping region D warp . Axis x shows the direction of motion, while ρ is the cylindrical radius.As we discuss in Section 4.1, flattened disk-shaped metrics minimise energyrequirements of the particular Alcubierre, but not necessarily other, warp drivespacetimes. As we also discuss in Section 5, warp drive spacetimes require someform of propulsion in order to accelerate. For this reason, in physical realisationsof such spacetimes, the front and rear parts are likely asymmetric. observer O out , co may be moving with constant three-velocity v s with respect to atimelike observer O out , the latter of which we shall consider to be at rest. Everywherein this study, apart from the general discussion in this section, we shall consider warpdrives spacetimes which are axisymmetric along the direction of motion.The three-velocity of the comoving observer v s , which represents the velocity ofthe warped region relative to the remote observer O out , may be slower or faster thanthe speed of light with respect to the observer. This corresponds to four-velocity ofthe comoving observer O out , co being timelike, null or spacelike. Both observers O in and O out , co , therefore, may in general, at least formally, be timelike, null, or spacelike.If region D warp contained vanishing stress-energy tensor, the whole spacetime wouldbe close to Minkowski spacetime and the two observers O in and O out would both be ntroducing Physical Warp Drives D warp to be sufficiently curved such that, as we show further, the norms of the four-velocities of observers O in and O out , co may be different from each other. Therefore,all stationary warp drive spacetimes may be split into four distinct classes, based onwhether the remote comoving observer O out , co moves subluminally (subluminal drives)or superluminally (superluminal drives), and whether the internal observer O in hasthe same norm as the comoving observer O out , co (mild warp drives) or different normfrom the comoving observer (extreme warp drives): Class I: Mild subluminal warp drives:
These spacetimes are defined by thevector field ξ being timelike everywhere. Consequently, three-velocities of suchdrives are subluminal, i.e. v s < c . Spacetimes of this class approach the flatMinkowski spacetime in the trivial limit, and observers O in , O out , co reduce toa pair of co-moving timelike Lorentz observers. Non-trivial members of thisclass contain spacetimes with region D warp sufficiently curved, so that tetradsof observers O in and O out , co differ significantly from each other, i.e. the observersread off different rates of clocks and lengths of rulers. At the same time, suchspacetimes also contain weak-field solutions corresponding to classical shell-likeobjects moving with subluminal velocities and weakly modifying the state of thespacetime inside them. Such solutions are possible because the D warp region maybe set arbitrarily close to being flat, rendering the whole spacetime arbitrarilyclose to Minkowski spacetime. Class II: Mild superluminal warp drives:
These spacetimes are characterized bythe vector field ξ being spacelike or null everywhere. Consequently, such warpdrives have luminal or superluminal velocities, i.e., v s ≥ c . These spacetimesalso admit a trivial limit, wherein they reduce to flat Minkowski spacetime, withthe tetrads O in , O out , co corresponding to a pair of comoving null or spacelike(superluminal) ‘observers’. Weak-field members of the class correspond to smallamounts of ‘superluminal matter’ in the region D warp introducing small differencesin the measurements of frames O in , O out , co . By ‘superluminal matter’ weunderstand the matter at rest with respect to a space-like reference frame. Inthe case of the stress-energy tensor for a perfect fluid, such matter violates thedominant energy condition. A general spacetime of this class introduces non-trivial differences between frames O in and O out , co . Since superluminal mattercannot be produced from physical matter, and since null or spacelike tetradscannot be associated with physical observers, the spacetimes of this class havelimited interest. Class III: Extreme superluminal warp drives:
These spacetimes are defined bythe vector field ξ being timelike in the inner region D in , but null or spacelike in theasymptotic infinity of the outer region D out . The remote comoving observers insuch spacetimes move luminally or superluminally relative to the resting timelikeobserver O out , i.e., v s ≥ c . This class of spacetimes does not contain trivialsolutions and the warped region D warp is sufficiently curved to allow timelikeobservers O in to be moving superluminally relative to the timelike observer O out (and as a consequence, the timelike observer O in will be travelling back in timefrom the point of view of yet another remote timelike observer O (cid:48) out ). At thesame time, the comoving observer O out , co is formally superluminal; in otherwords, remote timelike observers cannot be comoving with a warp drive of thisclass. For such spacetimes, we can define a Killing horizon ∂ D K as the minimal ntroducing Physical Warp Drives ξ becomes null. A necessary, butnot sufficient, condition for warp drives of this class to be physical is that theKilling horizon ∂ D K does not intersect with region D warp . If the Killing horizondid intersect with region D warp , some part of the matter in that region wouldbe at rest with respect to spacelike tetrads aligned with vector field ξ . In otherwords, a fraction of matter in the region D warp would have to be superluminal (inthe same sense, as for Class II above) and violate the dominant energy condition. Class IV: ‘Extreme’ subluminal warp drives:
Spacetimes of this class aredefined by the vector field ξ being null or spacelike in the inner region D in , buttimelike at the asymptotic infinity of the outer region D out . Since the comovingobserver O out , co is timelike, such spacetimes are subluminal, i.e. v s < c . Since theKilling vector is spacelike in the inner region D in , no timelike internal observerscan be at rest relative to the inner boundary of the drive ∂ D in . This propertybears similarity to black hole spacetimes, except for that the inner region D in forthis class of spacetimes is flat everywhere. A necessary condition for spacetimesof this class to be physical is that the Killing horizon coincides with the boundaryof the inner region, ∂ D in . Otherwise, some fraction of matter in the region D warp will be superluminal, as in the previous class.In practical terms, as follows from our definition, any warp drive spacetime maybe seen as a shell of ordinary or exotic material which fills region D warp and moves withsome constant velocity relative to an external timelike observer O out . The presenceof the shell inevitably modifies the space and time in the inner region D in . Thismodification leads (in addition to other effects) to a difference in the measurementsof times and lengths between an observer in the inner region D in and a comovingremote observer O out , co . General warp drive spacetimes, therefore, form a continuousfamily which includes both trivial (flat or nearly-flat) and non-trivial (strongly curved)spacetimes. The Alcubierre drive is a truncated warp drive spacetime defined by meeting the threerequirements. First, that the external spacetime D out is Minkowskian. In other words,the matter in the warped region D warp does not exhibit gravitational influence outsideof it (Requirement 1). Second, observers O in and O out are always timelike, and thetetrads of these observers are equal to each other. In other words, the clock rates andthe lengths of the rulers of observers O in and O out are synchronized and put equal toeach other (Requirement 2). Finally, the warped region D warp is functionally limitedto Equations (1) and (2) in Section 1 (Requirement 3). For subluminal velocities, theAlcubierre drive belongs to Class I, mild subluminal warp drives. For superluminalvelocities it belongs to Class III, extreme superluminal warp drives.Compared to general warp drives, this set of constraints may seem artificial.For example, requiring that the material of the Alcubierre drive does not gravitatein region D out , which is non-typical for massive objects, as we discuss further inSection 3, unnecessarily puts strong constraints on the stress-energy tensor in region D warp . Similarly, requiring that the time of observers O in is synchronized with the timeof observers O out means that, at least in the subluminal regime, the time of internalobservers O in is accelerated compared to the time of remote comoving observers O out , co . In comparison, general warp drive spacetimes do not need to be truncated ntroducing Physical Warp Drives O in and O out , co .A more general class of warp drive metric was proposed by (Nat´ario 2002).However, their most general warp drive, similarly to the Alcubierre solution, stillimposes Requirements 1 and 2 from the Alcubierre metric. That is, is also makesregion D out strictly flat outside of the drive and synchronizes the time of the internaland the resting (not comoving) observers. Similarly to the Alcubierre solution, the(Nat´ario 2002) class of spacetimes also belongs to Class I in the subluminal regimeand Class III in the superluminal regime.Further, in Sections 4 and 3, we show that general warp drive spacetimes arediverse and can have a variety of properties that are often more appealing comparedto either the Alcubierre spacetime or to the (Nat´ario 2002) class of spacetimes. In this section, we summarise the standard tools applicable for analyzing general warpdrives, in order to apply them further in Sections 4 and 3.The internal physical volume of a warp drive measured by observer O in , inside thewarped region, is given by V in ≡ (cid:82) D in d x i in . For truncated warp drive spacetimes, i.e.spacetimes where the outside vacuum region D out is flat, the external volume may bedefined as V out ≡ (cid:82) D in ∪ warp d x i out . The integral is calculated over the volume enclosingthe warped region and assuming, during the integration, that the volume’s interior isflat. Such a construction provides a measure of the size of the warp bubble as observedfrom outside. It may be formally done with the help of the Cartesian coordinate gridof the remote observer O out , analytically extended to cover the warped region. Thecomoving external volume V out , co may be similarly defined for the comoving observer.For general asymptotically-flat metrics, we do not consider external volumes as theyare coordinate-dependent.We calculate the energy density, the momentum flux, and their volume integralsfor warp drive spacetimes, and the mechanical stress distributions, in the coordinatesystem adapted to the remote comoving observer O out , co . The coordinate systemadapted to the comoving observer is aligned with the Killing vector field ξ , whichallows for a 3+1 decomposition of the spacetime and thus provides coordinate-invariantmeasures for the total energy, mass and momentum of the drive. We also considera system of Eulerian observers defined through four-velocities u µ = √ − g (1 , , , w in each of the coordinate systems is given by w =( − g ) − T . The momentum flux j i and the pressure P in the same reference systemare given by j i = u µ T µi = √ − g T i and P = T µν ( δ µν − u µ u ν ) = T + ( g ) − T ,where T ≡ T µν g µν . Notably, the total energy and momentum contain the contributiononly from the warped region D warp : E = (cid:82) D warp ( − g ) − T √− g d x i out , J i = (cid:82) D warp √ − g T i √− g d x i out .Warp drive solutions may be compared based on how they satisfy different energyconditions. Among these are the strong and weak energy conditions. Furthermore, wealso consider quantum inequalities, e.g. (Pfenning & Ford 1997). These inequalities,at least in the weak-field regime, potentially allow the spacetime curvature to modify ntroducing Physical Warp Drives τ π (cid:90) ∞−∞ (cid:104) T µν u µ u ν (cid:105) τ + τ d τ ≥ − π τ , (4)where τ is the proper time of the observer and τ is an arbitrary constant smaller thanthe local radius of curvature.
3. General spherically symmetric warp drives
One may gain considerable intuition into the nature of warp drive spacetimes byconsidering spacetimes which are spherically symmetric in the comoving referenceframe. Since spherically symmetric spacetimes are fully solvable in relativity, we mayachieve full characterization of such warp drive spacetimes. More formally, in thissection, we limit ourselves to spacetimes which contain an SO(3) group whose orbitsare orthogonal to the Killing vector field ξ introduced in Section 2. The Killing vectorfield ξ defines the local frame of rest in the comoving reference frame. As we commentfurther, the presence of such a group is possible only for class I and IV spacetimes,i.e., subluminal warp drives.Class I of subluminal warp drives – mild subluminal warp drives – containstrivial solutions and may be associated with ordinary objects, such as a thin shellof non-exotic material. However, the well-known Alcubierre and Nat´ario drives, in thesubluminal regime, also belong to the same class. By considering general sphericallysymmetric subluminal warp drives, we shall gain considerable intuition about thesespacetimes and provide a possible explanation for their negative energy requirement.The most general metric for a spherically symmetric stationary spacetime can bewritten in Schwarzschild coordinates as:d s = − N ( r ) c d t + Λ( r )d r + r (d θ + sin θ d ϕ ) , (5)where functions N ( r ) and Λ( r ) are general. Therefore, energy density of a sphericallysymmetric warp drive spacetime, in the comoving reference frame, is given by: w ( r ) = 18 πr (cid:18) − ( r Λ( r ) ) (cid:48) (cid:19) (6)A necessary condition for the spherically symmetric metric in Equation (5) (above) todescribe a warp drive spacetime is that the energy density w in both the inner region D in and outer region D out be equal to zero. In particular, as follows from Equation (6),for any spherically symmetric warp drive, Λ( r ) = 1 in region D in . Furthermore,integrating Equation (6), we can express the metric function Λ( r ) directly throughthe energy density distribution:Λ( r ) = 11 − r (cid:82) r πw ( r (cid:48) ) r (cid:48) d r (cid:48) (7)Equations (6) and (7) allow us to analyze the energy density distribution ofspherically symmetric solutions and to construct such spacetimes from any desiredenergy distribution. Thus, we may see that for purely positive energy spacetimes,Λ( r ) > D out . This result is in agreement with the Birkhoff’stheorem, and states that the metric in the region D out is described by the Schwarzschildsolution. In comparison, for truncated warp drive spacetimes, such as the Alcubierre ntroducing Physical Warp Drives (cid:82) r out πw ( r (cid:48) ) r (cid:48) d r (cid:48) = 0. Asa result, any non-trivial truncated spherically symmetric warp drive must containregions where the energy density is negative. Therefore, the fact that Alcubierreand Nat´ario solutions are truncated may partly explain why they require negativeenergy even in subluminal regimes. Finally, for any spherically symmetric warp drivespacetime, since Λ(0) = Λ( r → ∞ ) = 1, coordinate r describes the same physicallength scale both inside the drive and at the asymptotic infinity. In other words,unlike the drives discussed in (Van Den Broeck 1999), spherically symmetric warpdrives cannot contain objects which significantly exceed the sizes of warp drives, asmeasured by external observers.We may further calculate the spatial component of the stress energy tensor.To do so, we assume that the material in region D warp is an isotropic fluid, i.e. T rr = P ( r )Λ( r ). This way, we obtain the pressure of the material which constitutesthe warp drive: P ( r )Λ( r ) = 1 − Λ( r ) + rN (cid:48) ( r ) /N ( r ) r (8)By requiring that P = 0 in the inner vacuum region D in , and by remembering thatΛ( r ) = 1 in that region, we can see that N ( r ) is constant in D in . In the externalregion D out , from Equation (8), N ( r ) = (Λ( r )) − = 1 − r (cid:82) r πw ( r (cid:48) ) r (cid:48) d r (cid:48) . Thisresult also follows from Birkhoff’s theorem, according to which spherically symmetricspacetimes must be static and asymptotically flat. In the vacuum region, such setup,by definition, corresponds to the Schwarzschild solution (assuming the central objectis electrically neutral), which matches by the form the equation for N ( r ). In this case, (cid:82) r πw ( r (cid:48) ) r (cid:48) d r (cid:48) will have the effective meaning of enclosed mass.We can now obtain N ( r ) in region D warp , and this way determine N ( r ) in theinner region D in . We do this by writing down the equation for continuity of the stress-energy tensor, T µr ; µ = 0, similarly to how it is done for Tolman–Oppenheimer–Volkoff(TOV) equation. As a result, we obtain: N (cid:48) ( r ) N ( r ) = − (cid:18) P + ρ (cid:19) − P (cid:48) ( r ) (9)Firstly, we make a simplifying assumption that density ρ is a function of only pressure P , as, for example, is the case for the polytropic equation of state. Then, by usingthe condition that P = 0 at the inner and outer boundary of the warp region D warp ,we find that the integral of the right-hand side over D warp vanishes. As a result, weconclude that N ( r in ) = N ( r out ), where r in and r out are the locations of the inner andouter boundary of the region D warp .From the solution for N ( r ) in the outside region D out , we see that the time-timecoefficient of the metric N ( r ) at the asymptotic infinity can only be larger than insidethe warped region. In other words, for a subluminal warp drive based on non-exoticmatter, the time inside the drive can only pass more slowly than it does for a remoteMinkowski observer comoving with the drive. For scale, an Earth-mass shell of 10meter radius will slow down the rate of time by a small fraction of 4 · − . If werequire that the time in the inner region goes faster than in the reference frame ofthe comoving observer, the material in the warp drive would have to contain negativeenergy. This is the case for the Alcubierre and Nat´ario drives. In these drives, thetime is passing more quickly in the inside region than it is in the reference frame of theremote comoving observer (due to the clocks in the inner region being synchronized ntroducing Physical Warp Drives The method from the previous section could potentially have been useful for analysingsuperluminal warp drives. However, superluminal classes of warp drives (Classes II andIII) cannot be spherically symmetric. This fact is closely related to the impossibility ofsuperluminal spherical objects in special relativity, e.g. (Fayngold 2002). In Section 2,we have introduced a Killing vector field ξ , which defines a local frame of rest withrespect to the drive. In superluminal warp drives, vector field ξ becomes spacelikewhether inside or outside of the drive. In order both to maintain spherical symmetryand for the field ξ to remain orthogonal to the orbits of the SO(3) group, the spatialcomponent of the field ξ can only be radial with respect to the drive. Therefore, atasymptotic infinity, comoving observers for such a spacetime would have to be movingradially towards or away from a spherically symmetric drive. In other words, warpdrives cannot be both spherically symmetric and move superluminally.Similarly, Class IV drives cannot be spherically symmetric. As follows fromEquation (7), in order for the radial basis vector inside the drive to be timelike,one would need to have non-zero energy density present in the vacuum inner region D in . Therefore, all spherically symmetric positive-energy warp drive solutions belongto Class I. These solutions are always subluminal, satisfy the energy conditions,and are devoid of causality paradoxes present in superluminal metrics. Overall, thesubluminal spherically symmetric solution presented in this section is the first exampleof manifestly positive energy warp drive spacetimes.
4. Axisymmetric warp drives with a general internal region
In this section, we explore the diversity of axisymmetric warp drive solutions. Thereare some limitations on possible axisymmetric solutions. Explicitly expressing ametric through a desired stress-energy tensor in a closed form is not possible for ageneral axisymmetric spacetime. Therefore, we leave generalizing the positive energyspherically symmetric warp drive solutions to the axisymmetric case to future studies.However, by constructing sufficiently broad classes of metrics, we show that interestingsolutions are possible even within these less general classes.
Below we introduce a method, by which one can construct new metrics for a warpdrive spacetime. For constructing new solutions, following the discussion in Section 3,we focus on choosing how the spacetime properties of the observers inside the innerregion O in relate to those of remote observers O out . This choice may put significantconstraints on the energy content of the spacetime, and may potentially be exploitedto find spacetimes with lower energy requirements. Additionally, having an explicitway to define the properties of spacetime inside the inner region D in allows us toexplore and demonstrate the diversity of possible warp drive solutions.As discussed in Section 2, any stationary warp drive spacetime may be associatedwith a coordinate system x µ co , adapted to the Killing vector ξ which defines the restframe of the craft. We also adopt a global coordinate system x µ which asymptotically, ntroducing Physical Warp Drives O out in region D out . Since the two charts x µ co and x µ cover the whole spacetime and overlap, it is, inprinciple, possible to introduce a mapping from one to another in the regions D in and D warp . The procedure for constructing an axisymmetric warp drive spacetime fromthis subclass is as follows:1. Choose a one-to-one mapping x µ co ( x ν ) between the coordinate system x µ co adoptedto the comoving observer inside the inner region O in and the coordinate system x ν adopted to the external observer at rest O out .2. Choose functions f x ( x i co ), f y ( x i co ), f z ( x i co ), f t ( x i co ) which are equal to 1 in region D in and are equal to 0 in region D out . These functions define the shape and sizeof the warp region D warp from the point of view of observer O in .3. Formulate the metric of the spacetime as:d s = − c (d t (1 − f t )+ f t d t co ) +(d x (1 − f x )+ f x d x co ) + (10)+(d y (1 − f y ) + f y d y co ) + (d z (1 − f z ) + f z d z co ) , Express the metric in terms of a common coordinate system x µ (cid:48)(cid:48) , by expressing x µ co and x ν explicitly through x µ (cid:48)(cid:48) .4. Analyze the metric with the methods introduced in Section 2.3The procedure above is based on the idea of explicitly comparing themeasurements of observers O in and O out . Indeed, in region D in all the functions f η = 1, and the metric corresponds to the coordinate system of observer O in (d s = − c d t + d x + d y + d z ). Similarly, asymptotically, in region D out allthe functions f η = 0 and the metric corresponds to the coordinate system of theremote observer at rest O out (d s = − c d t + d x + d y + d z ). Several studies inthe literature, e.g. (Loup et al. 2001), have proposed modifications of the Alcubierremetric without explicitly considering the measurements of inner and outer observers.Such modifications may be erroneous and may reduce to coordinate transformations,as happens in (Loup et al. 2001) case. We discuss these studies further in AppendixA. The procedure developed here assumes that the observers O in and O out aretimelike, and therefore leads to warp drive spacetimes of Class I in the subluminalregime, and of Class III in the superluminal regime. While these classes of spacetimesare arguably most interesting, the procedure may be easily generalized to cover thespacetimes of Class II and Class IV. We also comment that while the procedureensures a relationship for the measurements of observers O in and O out , one mayalso use it to obtain different coordinate representations and, potentially, differentspacetimes satisfying the same relationships between the observers. Subsequently, onemay choose, for example, the most interesting spacetimes of the class.As a demonstration, we show how the above procedure may be used to constructthe Alcubierre metric given by Equation (1).1. We choose the one-to-one mapping as follows: d t co = d t d x co = d x − v s d t d y co = d y d z co = d z (11) ntroducing Physical Warp Drives f x = f , where f is given by Equation (2), whereas functions f y , f z and f t cancel out in Equation (10) and will not contribute to the metric.3. We arrive at the metric d s = − c d t + (d x − f ( r s ) v s d t ) + d y + d z , as givenby Equation (1).This example once again highlights the fact that the Alcubierre metric is basedon a rather artificial identification of spacetime properties between the two observers O in and O out , in particular increasing the rate of the time passing for the observer O in . In the following sections, we explore other possible axisymmetric spacetimes andshow that they have more appealing properties than the Alcubierre metric. E tot by flattening the Alcubierre metrics We start by considering the flattened Alcubierre metrics. The longitudinal extent isa simple property of the Alcubierre metric. However, it has not been studied in theliterature, despite having several interesting properties.The longitudinal extent has several peculiarities for the Alcubierre metric. Indeed,in Equation (1), for sufficiently large distances r s from the center of the warped region,function f ( r s ) vanishes in the Alcubierre solution. Therefore, the external observer O perceives the boundaries of the warped region D warp as a sphere of radius r s movingwith velocity v s . If the velocity v s is subluminal due to Lorentz contraction, theboundary of the warped region D warp must be elongated for the comoving observers.Moreover, when the Alcubierre warp drive approaches the speed of light while at thesame time preserving its energy content, given by Equation (3), the warped regionmust appear infinitely elongated. The divergent elongation of the solution in thecomoving frame puts into question the possibility of accelerating Alcubierre metricsbeyond the speed of light.To analyze the solutions deformed along the axis of motion, we switch to moreconvenient cylindrical coordinates ( y, z ) → ( ρ, θ ). For the transition regions describedby Equation (1), a more general non-spherically symmetric metric may be obtained byreplacing f ( r s ) ≡ f ( (cid:112) ( x − x s ) + ρ ) −→ f ( x − x s , ρ ). Notably (cid:107) , for the Alcubierredrive, recalculating the energy density with f = f ( x − x s , ρ ) simplifies the expression,compared to Equation (2): w = − π v s (cid:18) ∂f∂ρ (cid:19) (12)This form leads to three useful implications. The first implication is that,for a given velocity of the warp drive v s , the most optimal way of reducing thetotal energy, as measured by Eulerian observers, is by flattening the shape of thewarp drive. Indeed, as √− g = ρ for the Alcubierre metric, the total energy E = − v s (cid:82) ∞−∞ d x (cid:82) ∞ ρ d ρ (cid:16) ∂f∂ρ (cid:17) . And indeed, flattening the warp drive by a factorof α X ( α X >
1) – i.e., by replacing f ( x − x s , ρ ) → f ( α X ( x − x s ) , ρ ), as may beshown through variable change α X ( x − x s ) → x (cid:48) – leads to the energy reduction by E → E/α X . Similarly, elongating the drive (choosing α X <
1) increases its energyrequirements. Therefore, putting aside the fundamental issues related to negativeenergy of Alcubierre drives, an optimal implementation of such spacetimes wouldlikely be flattened in shape. (cid:107)
We believe, this form has not been found in the literature so far ntroducing Physical Warp Drives α X = 1 + v leads to E → E/ (1 + v ), asymptotically removing thedependency of the total energy on velocity. As discussed earlier in Section 1, the factthat the total energy of the Alcubierre drive depends on velocity is problematic. Thisis because the energy and momentum conservation, applicable to asymptotically-flatspacetimes, implies that the Alcubierre drive must be changing its already very largeenergy (and mass) as it accelerates. Removing or softening the dependence of thetotal energy on velocity may, in principle, lead to more efficient ways of acceleratingthe drive to large velocities.Finally, the third implication of Equation (2) is that it allows one to constructsuperluminal solutions which satisfy the quantum inequalities given by Equation (4).Indeed, selecting a sufficiently large α X allows one to reduce the thickness of thewarp in x -direction down to nearly-Planck scale size (correspondingly, allowing onlyfor extremely thin physical observers inside the warp bubble). As a result, thelocal curvature radius in Equation (4) may be arbitrarily small, thus satisfying thequantum inequalities given by Equation (4). At the same time, such superluminaldrives still maintain a macroscopic size in ρ -direction perpendicular to the directionof motion and, more importantly, do not increase their densities due to contraction.Such solutions, may offer an exciting possibility of superluminal physical solutions.Perhaps, they may help probe the physics of superluminal motion and the problemsassociated with it, e.g. the violation of causality (Krasnikov 1998), (Everett &Roman 1997). However, they more likely probe the limits of applicability of thequantum inequalities (4), which are derived in semiclassical gravity approximation.Indeed, superluminal motion violates the averaged null energy condition (Visseret al. 2000), and the latter does not have the same dependency on the dimensionsof the bubble as the quantum inequalities, e.g. (Graham & Olum 2007). As a result,even the extremely flattened version of the Alcubierre drive, as discussed here, doesnot satisfy the averaged null energy conditions. The Alcubierre metric is artificial in the sense that it forces the clocks and rulers ofobservers O in and O out to be synchronized. An arguably more natural choice wouldbe to require that the observer O in should experience the same time dilation and spacecontraction as a Lorentz observer would experience when moving with velocity v s < c .Since Lorentz transformations are defined for subluminal speeds, in this section, weconsider subluminal warp drives of Class I. To construct the spacetime, we choose:d t co = γ (d t − v s d xc ) (13)d x co = γ (d x − v s d t ) , where γ is the Lorentz gamma-factor. Using these definitions to construct the warpdrive metric with Equation (10) leads to a diagonal metric:d s = − c F d t + F d x + d ρ + ρ d θ , (14)where F ≡ f (1 − f )( γ − γ is replaced by anyconstant). ntroducing Physical Warp Drives w = − π ( ρF (cid:48) ρ ) (cid:48) ρ ρF (15)For this metric, independent of the choice of function f ( x − x s , ρ ), the region D warp contains areas of both positive and negative energy density. This result is likelyrelated to the fact that the spacetime is effectively flat beyond a certain radius, ratherthan asymptotically approaching the flat spacetime, similarly to the Schwarzschildsolution. In the case of spherically symmetric solutions, as discussed in Section 3,such truncation of the gravitational field also requires the warp region D warp to containregions of positive and negative energy.The Lorentz drive metric may be generalized to porduce a continuous limitto flat spacetime. Indeed, if region D warp were replaced by Minkowski spacetime,the whole spacetime would be Minkowski space. Therefore, one may introduce acontinuum of solutions parametrized by a dimensionless parameter λ ∈ [0 , F ≡ λf (1 − f )( γ − λ = 0 the whole spacetime reduces to a flat spacetime,and for λ = 1 the Lorentz warp drive solution is recovered. The intermediate values of λ allow for solutions with smaller energy requirements. To our knowledge, this is thefirst example of a family of warp drive solutions containing near-Minkowski metrics. The warp drive solution by (Van Den Broeck 1999) is intended to reduce the energyrequirements of warp drives by significantly expanding the volume of the Alcubierredrive inside radius r s while significantly decreasing the externally measured size ofthe craft. In this section, we realize the idea of (Van Den Broeck 1999) in a simplerand, importantly, coordinate-independent form. We select the relation between clocksof the internal and the external observers to be the same as in the Alcubierremetric, d t co = d t , but choose the internal spacetime to be expanded in x -direction:d x co = A (d x − v s d t ), d ρ co = d ρ , d θ co = d θ , where A >
1. Substituting these relationsinto Equation (10), we obtain the resulting spacetime:d s = − c d t + (d x + f ( A − x − f Av s d t ) + d ρ + ρ d θ (16)This spacetime leads to the energy density distribution of: w = − π (cid:34) v s A f (cid:48) ρ (1 + ( A − f ) + ( A −
1) ( ρf (cid:48) ρ ) (cid:48) ρ (1 + ( A − f ) (cid:35) (17)The expression contains a velocity-independent component, which also leads to regionsof positive and negative energy density. We see here that while stretching thespace inside the warp drive does lead to higher energy density, the metric may beoptimized to achieve the highest internal volume per unit energy needed to constructthe spacetime. In this section, we consider spacetimes, in which the clocks in the region D in run ata different rate compared to the clocks in the region D out . This is done by lettingd t co = A − d t , where A > D in . ntroducing Physical Warp Drives s = − c ((1 − f )d t + A − f d t ) + (d x − f v s d t ) + d ρ + ρ d θ (18)The resulting energy density for the Eulerian observers is: w = − π (1 − ( A − − f ) (1 + ( A − − f ) · v s (cid:18) ∂f∂ρ (cid:19) (19)This demonstrates that the rate of clocks used in the original Alcubierre solution( A = 1) in fact leads to the simplest possible expression for the total energy of thewarp drive, other parameters being constant. From Equation (19) it follows thatslowing down the clocks inside the Alcubierre drive leads to higher amounts of energydensity from the point of view of Eulerian observers. Finally, our method can be used to construct relatively complex relations betweenthe inner and outer regions. For example, the following metric corresponds to thetwo Minkowski regions D in and D out spinning with respect to each other. In otherwords, the observers at rest in region D in will be rotating with respect to observers inregion D out without experiencing any centrifugal or Coriolis forces typical to rotatingsystems. Such settings are impossible to realize in the absence of the transition region D warp . The settings may be achieved by relating d θ co = d θ − ω s d t , in addition to thedefinition used for the Alcubierre drive, where ω s is a constant parameter defining theangular velocity. Substituting these relations into Equation (10) leads to the followingmetric: d s = − c d t + (d x − v s f d t ) + d ρ + ( ρ d θ − f ρω s d t ) , (20)The energy density for Eulerian observers then is: w = − π (cid:34) v s (cid:18) ∂f∂ρ (cid:19) + ρ ω s (cid:32)(cid:18) ∂f∂ρ (cid:19) + (cid:18) ∂f∂x (cid:19) (cid:33)(cid:35) (21)We see that the energy density is negatively defined independent of the spin or velocityof the inner region. Additionally, the contributions from the velocity in x -directionand the velocity of angular rotation have a similar form. The main interest for suchspacetimes may be that they offer a stationary, dissipationless way of storing energyor angular momentum.
5. Discussion
One of the main conclusions of our study is that warp drives are simpler andmuch less mysterious objects than the broader literature has suggested when citing(Alcubierre 1994). Warp drives are inertially moving shells of positive or negativeenergy material which enclose a ‘passenger’ region with a flat metric. The mainfeature distinguishing warp drives from trivial inertially moving low-mass shells isthat the large amount of energy contained in the warp shell allows one to modify thestate of spacetime inside it. In particular, as shown in Sections 3 and 4, the time inthe inner region may go faster or slower than it would go without the shell. Similarly, ntroducing Physical Warp Drives
Warp drives, being inertially moving shells of normal or exotic material, do not haveany natural way of changing their velocities. They are just like any other types ofinertially moving objects. Similarly, just like for any other massive objects, achievinga certain velocity for a warp drive requires an externally applied force or, morepractically, some form of propulsion. Propulsion may be realized, for example, byan interaction with a bosonic field, or regular gaseous or plasma material.Whatever is the acceleration mechanism, it must obey the conservation of 4-momentum. This is because all warp drive spacetimes are asymptotically-flat. Anunfortunate error, introduced in (Alcubierre 1994), was to postulate the velocity inEquation (1) to be time-variable. An Alcubierre spacetime with time-variable velocityalso changes its energy and momentum with time, and, this way, such a constructionviolates energy conservation. More technically, the metric given by Equation (1) doesnot satisfy the continuity equations, unless additional dynamical fields are implicitlyintroduced to compensate for that. In view of this, no metric which describes anaccelerating warp drive solution has so far been presented in the literature. ntroducing Physical Warp Drives
18A more subtle point is that the Alcubierre and Nat´ario drives, as well as thespacetimes constructed in Section 4, represent classes of different objects parametrizedby velocity v rather than the same object changing its velocity v . In particular,different warp drive solutions with different values of v have different mass, differentenergy content, and often different shapes in their reference frame of rest. Anyrealistic object should at least conserve its ADM-mass in the subluminal regime and itsanalogue in the superluminal regime. A natural way of constructing such spacetimesis by defining them explicitly in their frame of rest, as discussed in Section 2 andimplemented for spherically symmetric drives in Section 3. Implementing metricsfor such accelerating objects and more general axisymmetric objects which preservetheir shape and mass in the comoving frame as they change their velocity remains asubject for future studies. As a simple compromise, one may also adjust the shapeof the warp drive with velocity so as to conserve the mass of the drive, as suggestedin Section 4.1. Finally, among all classes of subluminal warp drive solutions, theparticularly interesting ones in the practical sense are those classes which contain acontinuous set of solutions ranging from trivial to highly curved.At least in the subluminal case, warp drive spacetimes may be constructed byusing purely positive energy density, as presented in Section 3 for the sphericallysymmetric case. They can likewise be constructed using purely negative energydensity, as is the case for the Alcubierre solution, or constructed using both positiveand negative energy density. In Section 3 we showed, for the first time, that the onlytype of modification to the internal spacetime that is achievable with purely positiveenergy for spherically symmetric warp drives is slowing down the rate of time insidethe craft.In Section 4 we demonstrated that, by using both positive and negativeenergy density, one may achieve a variety of modifications for the spacetime insidemore general axisymmetric subluminal warp drives. The range of all the possiblemodifications achievable with purely positive energy in the general axisymmetric case,and whether the class of axisymmetric spacetimes in Section 4 may lead to purelypositive energy metrics, remain important open questions in the field. Similarly, inSection 2 we have provided a new argument why superluminal warp drive solutionsmay always violate weak energy conditions resulting in their negative energy densityrequirements. While this is an established fact (Olum 1998, Visser et al. 2000), astrict independent proof of this based on our argument is another important avenuefor future studies. Our conclusions do not support the recent claim in (Lentz 2020)of superluminal purely positive energy warp drive solutions, which merits furtherinvestigation. An interesting question remaining is: assuming that a practical realization of a warpdrive spacetime is possible, what would the optimized versions look like? As weshowed in Section 4.2, a more optimal implementation of the Alcubierre warp driveswould be flattened in shape, since such shapes are more efficient in terms of energyrequirements. In particular, flattening the shape by a factor of 10 would lead toproportionally smaller energy requirements. This conclusion likely holds for the,more physical, purely positive energy subluminal warp drives as well. Curiously, aswe discussed in Section 4.1, extreme flattening of Alcubierre drives may allow forsuperluminal solutions which satisfy quantum inequalities, without reaching extreme ntroducing Physical Warp Drives f ( r s ). The originally proposed function givenby Equation (2), is not optimized and was originally chosen solely for demonstration.By applying the variational method to the expression for the total energy of theAlcubierre drive, we find that the shape function optimizing the energy is given by¯ f ( r s ) = min( r r s , r is a free parameter determining the inner size of theregion D warp . Using this slower-decreasing shape reduces the energy requirement fora similarly sized Alcubierre drive by about a factor of three. The physical reasonfor it, as we discuss in Section 3, is perhaps related to the fact that truncating thegravitational fields of a warp drive, as done in the Alcubierre solution, may increase the(absolute) amount of necessary negative energy compared to the more slowly falling offsolutions. We provide the details of this derivation in Appendix A.3, and also remarkthat optimizing the shape or matter distribution in region D warp can be equally wellperformed for all the other warp drive solutions.Given the wide range of possible states of spacetime achievable inside a warpdrive, it is also possible to imagine more complex and instrumental optimizations.For example, one may speculate that it is possible, at least in principle, to form aregion inside a subluminal warp drive which is similar to ergospheres of spinning blackholes. In this case, such a region would be used as an efficient energy storage. Theenergy could then potentially be extracted through a Penrose process applied to thepropellant of the craft, when passing through the ergoregion. Similarly to the Penroseprocess for spinning black holes, the extracted energy would likely be coming from therotation of some regions of the spacetime.Finally, since all warp drive objects require propulsion in order to accelerate, anypractical implementation of such objects would have to be asymmetric in shape, sincethe back part would have to accommodate a propellant exhaust system. One mayfurther hypothesise on setups, wherein black hole-like regions of the spacetime maybe used to produce accretion power. Accretion of material onto black holes is knownto be a few tens of times more efficient at extracting rest-mass energy in the form ofelectromagnetic radiation from the material than nuclear burning (Frank et al. 2002).Such a process could potentially provide both a source of energy and a source ofpropulsion. In Section 3, we demonstrated that it is possible to construct non-trivial warp drivesolutions with purely positive energy. In other words, at least in principle, one canconstruct objects of progressively larger masses and with progressively more salientmodifications to the internal spacetime. While the mass requirements needed for suchmodifications are still enormous at present, our work suggests a method of constructingsuch objects based on fully understood laws of physics.The most promising way of practically probing such spacetimes is throughlaboratory experiments – most importantly – through analogue gravity experiments,e.g. (Barcel´o et al. 2005). Another important avenue of exploring such spacetimes,especially the accelerating solutions, is through numerical relativity. Such experimentsmay bring a better understanding of purely positive energy drives, and negative energysolutions, as well as the possibility of accelerating objects superluminally.Since the introduction of (Alcubierre 1994), much theoretical effort has been put ntroducing Physical Warp Drives
Acknowledgements
We would like to thank Parsa Ghorbani, Lorenzo Pieri, Adam Lewis, Philip Chang,Hrant Gharibyan and Stefano Liberati for their insightful comments and helpfuldiscussions at different stages of this work. ntroducing Physical Warp Drives Appendix A. Existing warp drive solutions
The previously mentioned Alcubierre and Nat´ario metrics are distinct from each otherand satisfy the definition of warp drive metrics. In this section, we list the metricspresent in the literature that claim or intend to describe new warp drive metrics.We further show that these metrics reduce to the Alcubierre and Nat´ario solutions.We also comment that the (Lentz 2020) study likely forms a new class of warp drivespacetimes, though it does not provide means for reproducing itself.
Appendix A.1. Loup metric
The non-refereed study by (Loup et al. 2001), also discussed in the (Alcubierre &Lobo 2017) review, aims to reduce the energy requirements of the (Alcubierre 1994)drive by introducing a lapse function, which modifies the time-components of themetric as a function of spatial coordinates. However, their metric is equivalent to theAlcubierre metric.Indeed, their defining Equations (7)–(9), expressed in our notation, read:d s = − A ( r s ) c d t + (d x − f ( r s ) v s d t ) + d y + d z , (A.1)where A ( r s ) ≥ r s = 0 and,asymptotically, at r s → ∞ . However, by transforming to a new time coordinate A ( r s )d t → d¯ t , and checking that the coordinate transformation has a finite non-vanishing Jacobian and is, therefore, well-defined, we arrive at:d s = − c d¯ t + (d x − ¯ f ( r s ) v s d¯ t ) + d y + d z , (A.2)where ¯ f ( r s ) ≡ f ( r s ) A ( r s ) . Therefore, the Loup metric is that of the Alcubierre drive.The shape function, ¯ f ( r s ), as in the Alcubierre solution, is equal to 1 at r s = 0and asymptotically approaches 0 at infinity. Unlike in the Alcubierre solution, theshape function now decreases non-monotonically, which corresponds to redistributingthe energy density, according to Equation (3). Assuming the volume of the inner flatregion is preserved, such a shape function only increases the total energy required bythe solution. Appendix A.2. Van Den Broeck metric
The study by (Van Den Broeck 1999) provided a metric, which intended to significantlyreduce the energy requirements compared to the Alcubierre solution. Such a reductionwas made by reducing the outer surface area of the warped region D warp and throughexpanding the volume in the interior region D in .The (Van Den Broeck 1999) metric is given as:d s = − c d t + B ( r s ) (cid:0) (d x − f ( r s ) v s d t ) + d y + d z (cid:1) , (A.3)where B ( r s ) ≥ B (cid:29) r s = 0 and asymptotically decreasing to unity. By applying a coordinatetransformation defined by d¯ x = d xB ( r s ), d¯ y = d yB ( r s ), d¯ z = d zB ( r s ) and ¯ t = t and ensuring that the transformation is well-defined, by checking that the Jacobianof the transformation is finite and non-vanishing, the metric transforms to:d s = − c d¯ t + (cid:18) d¯ x − f ( r s ) B ( r s ) B (0) ( v s B (0))d¯ t (cid:19) + d¯ y + d¯ z (A.4) ntroducing Physical Warp Drives r s = 0 or, equivalently, x = v s t . In the newcoordinates this corresponds to ¯ x = B (0) v s t . Therefore, in the new coordinates, theobject moves with velocity ¯ v s ≡ B (0) v s . Finally, introducing a new shape function¯ f (¯ r s ) = f ( r s ) B ( r s ) B (0) , we arrive again at the Alcubierre metric:d s = − c d¯ t + (d¯ x − ¯ f (¯ r s )¯ v s d¯ t ) + d¯ y + d¯ z (A.5)The modified shape function ¯ f (¯ r s ) satisfies the condition ¯ f (0) = 1 and decreases tozero asymptotically at large values of ¯ r s . As for the Alcubierre solution, d¯ x − ¯ v s d¯ t ,d¯ y , d¯ z , d¯ t correspond to the coordinates adapted to a resting observer inside theinner region D in . Therefore, the physical size of the inner region of the metric, asmeasured by the inner observer, is given by the region where function ¯ f (¯ r s ) is closeto unity. In the Van Den Broeck example, function B ( r s ) is sharp-peaked at thecenter, B (0) = B ( ˜ R ) ≈ , and B ( ˜ R + ∆) = 1, where ˜ R = ∆ = 10 − m. In thecoordinates of the internal observer, O in , the inner region is limited to r s = ˜ R , or¯ r s = B (0) r s = 100 m, which corresponds to the physical size of the inner region. Thelocation, where function B ( r s ) decreases to 1, corresponds to ¯ r s ≈
200 m, at whichpoint ¯ f (¯ r s ) = 10 − , i.e. is nearly vanishing. In the Van Den Broeck example, function¯ f (¯ r s ) should subsequently decrease further to zero, between distances correspondingto r s = R and r s = R + 10 v s L P , i.e. in a thin region 100 v s Planck scales thick. Since,at these distances, B (¯ r s ) = 1, the interval of ¯ r s , at which the function ¯ f (¯ r s ) decreasesfrom 10 − to 0, is also 100 v s Planck scales thick.Therefore, in summary, the Van Den Broeck solution is equivalent to theAlcubierre solution. The shape function f (¯ r s ) in their study is chosen to decreaseto nearly zero within a volume comparable to the inner volume of the drive (Region1). Subsequently, further out, the function f (¯ r s ) decreases to exactly zero (Region 2).Since in the second outer region (Region 2), function f (¯ r s ) decreases from a very smallinitial value to zero, it is expected that Region 2 should correspond to a small totalenergy and satisfy the quantum inequalities due to its near-zero thickness. As followsfrom our derivation, the total energy in the inner Region 1, and of the Van Den Broeckmetric as a whole, is comparable to that of the standard Alcubierre solution of similardimensions. Our derivation suggests that the total energy should be proportional to B (0) , through the ¯ v s term. The absence in the van den Broeck expression for Region1 of dependence on B (0) or on the velocity at all, potentially explains why the energiesthey obtain for that region are small. Appendix A.3. Optimizing the energy requirements of the Alcubierre solution
This section provides details on how one may optimize a warp drive solution in termsof its energy requirements. As an example, we choose the well-known Alcubierresolution. The total energy measured by Eulerian observers on a hypersurfaceof constant t for this spacetime is given, as follows from Section 2.3, by E = (cid:82) D warp ( − g ) − T √− g d x i out . For the Alcubierre solution (1), one may verifythat the contravariant time-time component of the metric tensor and the metricdeterminant are equal to minus unity, i.e. g = g = −
1. In this case, by usingEquation (3) for the value of energy density T , the total energy is obtained from asimple expression: E = − (cid:90) D warp π ρ v s r s (cid:18) d f d r s (cid:19) d x (A.6) ntroducing Physical Warp Drives f ( r s ) in this equation defines the location of the wall of the warp bubbleand is given by Equation (2). As we discuss in the main text, the specific form ofthe function was chosen in (Alcubierre 1994) rather arbitrarily in order to satisfy therequirement that f ( r s ) = 1 for r s → f ( r s ) = 0 for r s → ∞ . Therefore, one maysearch for other functions satisfying the same constraints and leading to some furtherdesired properties, for example, an optimised energy.To formulate a variational problem, we switch to spherical coordinates ( r s , θ, ϕ )centered at x = x s ( t ), y = z = 0, with the pole aligned with the direction of motion.In these coordinates, ρ = r s sin θ , and we get: E = − v s (cid:90) ∞ r d r s (cid:90) π d θr s sin θ (cid:18) d f d r s (cid:19) = − v s (cid:90) ∞ r d r s r s (cid:18) d f d r s (cid:19) (A.7)The Lagrangian for this system is L = r s f (cid:48) r s , and therefore the Euler-Lagrangeequation for the function f ( r s ) which optimizes the energy reads: ∂∂r s ∂ L ∂f (cid:48) r s = 2 ∂∂r s ( r s f (cid:48) r s ) = ∂ L ∂f = 0 (A.8)The solution of this equation is ¯ f ( r s ) = C + Dr s . The requirement that f ( r s ) = 0 at r s → ∞ can be satisfied by setting C = 0, while the requirement that f ( r s ) = 1 for r s → D = r . Since the variational problem was solvedfor r > r , the optimal solution in the whole space is given by ¯ f ( r s ) = min( r r , f decreases the needed (absolutevalue of) negative energy of the Alcubierre solution by about a factor of 3.Intuitively, the decrease may be understood because a smooth fall-off is more naturalfor gravitating bodies than a sharp-exponential cut in the metric introduced byEquation (2). Repeating the derivation but assuming that f is a general axisymmetricfunction of both ρ and x − x s shows that in this case the energy is optimized by aninfinitely thin shape, similar to the conclusions we obtained in Section 4. We concludethis section by mentioning that one can apply a similar method to other warp drivespacetimes (or even classes of warp drive spacetimes) to optimize their properties suchas the total energy. Bibliography
Alcubierre M 1994
Classical and Quantum Gravity (5), L73–L77.Alcubierre M & Lobo F S N 2017 Fundamental Theories of Physics , 257.Barcel´o C, Liberati S & Visser M 2005
Living Reviews in Relativity (1), 12.Bassett B A, Liberati S, Molina-Par´ıs C & Visser M 2000 Phys. Rev. D (10), 103518.Everett A E 1996 Phys. Rev. D , 7365–7368.Everett A E & Roman T A 1997 Phys. Rev. D , 2100–2108.Fayngold M 2002 Special relativity and motions faster than light .Fermi D & Pizzocchero L 2018
Classical and Quantum Gravity (16), 165003.Finazzi S, Liberati S & Barcel´o C 2009 Phys. Rev. D (12), 124017.Ford L H & Roman T A 1997 Phys. Rev. D , 2082–2089.Frank J, King A & Raine D J 2002 Accretion Power in Astrophysics: Third Edition .Garattini R & Lobo F S N 2007
Classical and Quantum Gravity (9), 2401–2413.Graham N & Olum K D 2007 Phys. Rev. D (6), 064001.Krasnikov S V 1998 Phys. Rev. D , 4760–4766.Lentz E W 2020 arXiv e-prints p. arXiv:2006.07125.Liberati S, Sonego S & Visser M 2002 Annals of Physics (1), 167–185.Lobo F S N 2007 arXiv e-prints p. arXiv:0710.4474.Lobo F S N & Visser M 2004
Classical and Quantum Gravity , 5871–5892.Loup F, Waite D & Halerewicz, E. J 2001 arXiv e-prints pp. gr–qc/0107097. ntroducing Physical Warp Drives Nat´ario J 2002
Classical and Quantum Gravity (6), 1157–1165.Nat´ario J 2006 General Relativity and Gravitation , 475–484.Olum K D 1998 Phys. Rev. Lett. (17), 3567–3570.Pfenning M J & Ford L H 1997 Classical and Quantum Gravity , 1743–1751.Van Den Broeck C 1999 Classical and Quantum Gravity (12), 3973–3979.Varieschi G U & Burstein Z 2013 ISRN Astronomy and Astrophysics , 482734.Visser M, Bassett B A & Liberati S 2000
Nuclear Physics B Proceedings Supplements (1), 267–270.White H 2013 Journal of the British Interplanetary Society66