Godel Universe from String Theory

G\"odel universe is a direct product of a line and a three-dimensional spacetime we call G$_\alpha$. In this paper, we show that the G\"odel metrics can arise as exact solutions in Einstein-Maxwell-Axion, Einstein-Proca-Axion, or Freedman-Schwarz gauged supergravity theories. The last allows us to embed G\"odel universe in string theory. The ten-dimensional spacetime is a direct product of a line and the nine-dimensional one of an $S^3\times S^3$ bundle over G$_\alpha$, and it can be interpreted as some decoupling limit of the rotating D1/D5/D5 intersection. For some appropriate parameter choice, the nine-dimensional metric becomes an AdS$_3\times S^3$ bundle over squashed 3-sphere. We also study the properties of the G\"odel black holes that are constructed from the double Wick rotations of the G\"odel metrics.

between the cosmological constant and the matter density in Gödel universe makes it an unrealistic model to challenge the chronology protection. Nevertheless it is a good toy model to study the effects of naked CTCs in both classical and quantum gravities.
The study of CTCs has enjoyed considerable attention with the development of the string theory and the AdS/CFT correspondence. Large number of supersymmetric or nonsupersymmetric Gödel-like solutions, including also black holes and time machines, with naked CTCs were constructed in gauged supergravities in higher dimensions, see e.g. [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17]. These works raise important issues whether the problems of CTCs may be resolved by stringy or quantum considerations, or whether naked CTCs in the bulk implies the breakdown of unitarity of the field theory.
Electric and magnetic fields periodic in z as a replacement for such matter were used to construct Gödel metric in [21], where z is a circular coordinate. It is worth mentioning that the three-dimensional metric with dz 2 removed can arise as an exact solution of Einstein-Maxwell theory with a topological term A∧F [13]. The embedding of the three-dimensional metric in heterotic string theory was first given in [11].
The main purpose of this paper is to construct more Lagrangians that admit Gödel metrics as exact solutions. We present three Lagrangians that admit the Gödel metrics as solutions, all involving only the fundamental matter fields. These are Einstein-Maxwell-Axion (EMA), Einstein-Proca-Axion (EPA) and Freedman-Schwarz [25] SU (2) × SU (2) gauged supergravity theories. The Freedman-Schwarz model can be obtained from the effective actions of string theories via the Kaluza-Klein reduction on S 3 × S 3 . Thus the four-dimensional Gödel universe can be embedded in string theory.
The paper is organized as follows. In section 2, we give a review of Gödel metrics and their properties. We also show that for appropriate choice of parameter, such a metric can also describe a direct product of time and a squashed 3-sphere. In section 3, we construct EMA and EPA theories that admit Gödel metrics, and also solutions involving squashed 3-sphere. We generalize the theories to higher dimensions and also give the effective threedimensional theories by Kaluza-Klein reduction. In section 4, we consider double Wick rotations and obtain two types of black hole solutions. In section 5, we show that Gödel metrics are exact solutions of the Freedman-Schwarz model and hence obtain the corresponding ten-dimensional solutions of string theories. We conclude the paper in section 6.
2 Gödel universe 2.1 The metrics of G α × R In this paper, we consider a class of metrics in the following form where ℓ and α are constants. The metric is a direct product of R associated with the coordinate z and the three-dimensional metric of (t, φ, r), which we shall call G α . The original Gödel metric [1] is recovered when we take α = 1 2 , corresponding to G 1/2 × R. To avoid pedantry, we shall refer the metric (2.1) with generic α also as the Gödel metric.
In addition to the constant shifting symmetry along the (t, φ, z) directions, the Gödel metric (2.1) is invariant under the constant scaling Note that imposing this scaling symmetry also implies that φ describes a real line, rather than a circle. This scaling property indicates that the metric is homogeneous. Since the metric G α is three dimensional, its curvature is completely determined by the Ricci tensor, whose non-vanishing components are given by Here we present the curvature in tangent space with the vielbein It is clear that when α = 1, the metric is locally AdS 3 (3-dimensional anti-de Sitter spacetime), i.e. G 1 = AdS 3 .

Energy condition and α value
It is convenient to define the energy-momentum tensor in the vielbein basis (2.4) We find T ab = diag{ρ, p, p,p}, with The original α = 1 2 case gives rise to matter with uniform pressure [1]. The null-energy condition requires that which is satisfied by 0 < α ≤ 1. As we shall see presently this implies that Gödel metrics in general have naked CTCs in the framework of Einstein gravity.

Metrics asymptotic to G α
We now consider deformations of G α by introducing a function of two constants and deform the metric (2.1) to become It is straightforward to verify that the curvature tensors (2.3) remain unchanged. This implies that the metric (2.9) is locally the same as (2.3). However, globally, the deformed metric (2.9) is different from (2.1). An important difference is that in the deformed metric (2.9), the coordinate φ is periodic, namely This ensures that the metric is absent from a conical singularity at r = r 0 . After imposing this condition, the coordinate transformation that relates (2.1) to (2.9) breaks the scaling symmetry (2.2) and hence the two metrics are not equivalent globally.
To demonstrate this explicitly, we note that the parameter a is trivial in that it can be eliminated by the coordinate transformation r → r − 1 2 a, without altering the global structure. The positive parameter b can be set to 1 without loss of generality, using the scaling r → √ b r, together with appropriate scalings of the rest coordinates. Now let r = coshr, φ =φ/ √ α, t =t −φ/ √ α, the metric (2.9) with a = 0 and b = 1 becomes On the other hand, we find that under the coordinate transformation r = coshr + cosφ sinhr , rφ = 1 √ α sinφ sinhr , and hence the two metrics are not globally equivalent. An important consequence is that the metric (2.11) has CTCs for r > r c with Here r = r c is the velocity of light surface (VLS) for which gφφ = 0. We shall call the metric (2.9) as the deformed Gödel metric that is asymptotic to the Gödel metric. For general parameters (a, b), there can be two VLS's between which g φφ > 0. The global structure of the metric (2.9), written in somewhat different parametrization, was analysed in [13].
The emerging of the CTCs in Gödel metrics is a consequence of that α ≤ 1. If one allows α > 1, equation (2.13) has no real solution for r c , and the metrics do not have naked CTCs. However, as we saw earlier that in the framework of Einstein gravity, α > 1 violates the null energy condition. In higher-derivative gravity due to the α ′ -correction of string theory, solutions with α > 1 were constructed in [2]. However, the theory, when treated on its own, involves inevitable ghost modes.

Mass and angular momentum
The general Gödel metric has two Killing vectors (2.14) (The Killing symmetry in z direction can be broken by the matter sector in some solutions.) Following the Wald formalism [29], which computes the variation of the on-shell Hamiltonian δH associated with a Killing vector with respect to the integration constants of the solutions, we read off the associated conversed quantities by evaluating δH at asymptotic infinity, and where ∆φ is given by (2.10). Note that if one chooses a fixed period ∆φ = 2π instead, rather than given by (2.10), the solution will have naked singularity at r = r 0 , for generic α. We have also set the convention dz = 1. For periodic z, this means period ∆z = 1; for real line z, this implies that the extensive quantities such as M and J are in fact uniform densities over the line z.

Squashed 3-sphere S 3 α × T
In the Gödel metrics, if we let ℓ 2 = −l 2 < 0, the metric (2.9) becomes where we have swapped the role of (t, z), and set, without loss of generality, h = 1 − r 2 . Let r = cos θ, we have This metric describes a direct product of time with a squashed 3-sphere, which we call S 3 α . When the squashing parameter α = 1, S 3 α becomes the round S 3 , written as a U (1) bundle over S 2 . The regularity of S 3 α requires that The period ∆z can be divided by a natural number n without introducing any singularity to the manifold, giving rise to S 3 α /Z n . (Such compact Gödel universe was also considered in [26].)

Gödel solutions from Lagrangian formalism
As mentioned in the introduction, theories in literature associated with Gödel universe (2.1) or (2.9) typically involve a matter energy-momentum tensor with unknown Lagrangian origin. The known example in four dimensions is the Einstein-Maxwell theory with an axion and a negative cosmological constant [21] where F = dA is the field strength. For the metric (2.1), the solutions for matter fields are In this case, the continuous shifting symmetry along the z-direction is broken to a discrete symmetry by the Maxwell potential A that is periodic in z. (The axion χ field does not break this symmetry since only dχ appears in the theory.) The solution is best described

Einstein-Maxwell-Axion theory with a topological term
In addition to (3.1), we introduce an additional topological term: where F = dA is the field strength, ε µνρσ is the density of Levi-Civita tensor whose components are ±1, 0. We choose the convention ε 0123 = 1. The axion, Maxwell and Einstein equations of motion are given by For the general Gödel metric (2.9) with (2.8), we consider the following ansatz for the axion and Maxwell field We find that the equations of motion are all satisfied provided that The general solution contains three integration constants, (a, b, q). The reality condition requires that |q| < √ 2 ℓ. It follows that we have 0 < α ≤ 1 and k ≥ 1. The original α = 1 2 Gödel metric corresponding to q = ℓ. The AdS 3 factor arises when α = 1, corresponding to turning off the Maxwell field. In section 3.3, we consider the case with α < 0, for which the metric describes S 3 α × T. It is worth pointing out that in four dimensions, the axion χ is Hodge dual to a 2-form potential B (2) with The Lagrangian (3.3) is equivalent to For the Gödel solutions we have B (2) = rdt ∧ dφ. The 3-form field strength G (3) is suggestive of string theory, which we shall discuss in section 5.

Einstein-Proca-Axion theory
In this subsection, we replace the previous Maxwell field by a Proca field of mass µ, with the Lagrangian The equations of motion are (There should be no confusion between the Proca mass µ and the spacetime indices.) The metric ansatz is given in (2.9). We consider the following ansatz for A and χ: Substituting these into the equations of motion, we find Gödel metric corresponds to taking q = ℓ/ √ 2.

The embedding of squashed 3-sphere
In the embedding of the Gödel metric in both EMA and EPA theories discussed above, the cosmological constant Λ is negative. When it is positive, the metric G α × R becomes S 3 α × T, as in (2.17). For the EMA theory, we have For the EPA theory, we have Thus we see that the embedding of the S 3 α in EPA theory requires a tachyonic vector with µ 2 < 0, while in the EMA theory, no exotic matter is required.

Generalizing to higher dimensions
In this section, we generalize the Gödel universe to higher dimensions by considering G α × R n , namely The α = 1 2 solution can be still solved by (1.2), but with u µ = (ℓ −2 , 0, . . . , 0). In order for the metrics to be solutions of some Lagrangians, we can replace the axion χ in the previous subsections by a (n − 1)-form potential B (n−1) with the field strength The Lagrangian (3.9) is now replaced by The corresponding ansatz for G (n) is given by (3.17) with k = 1.

Effective three-dimensional theories
The non-trivial part of the Gödel universe is the three-dimensional metric G α . For the solutions in subsections 3.1 and 3.2, the Killing symmetry in z direction is maintained by the matter fields. We can thus perform dimensional reduction on the coordinate z. The EMA theory becomes In this case, the three-dimensional G α metric is now supported by (3.20) The G α metric of this theory was constructed in [13], where the global structure of G α was discussed. Under the Kaluza-Klein reduction, the EPA theory becomes In this case, the G α metric is supported by .
The general metric (2.9) now becomes with the functionf now given byf For the EMA theory, we find that the matter fields are given by For the EPA theory, we have Thermodynamics of three-dimensional black holes was studied in [13]. Here we would like to derive the first law in our context and notations. New subtlety emerges in the EMA theory, where the parameter q is an integration constant. For simplicity, we shall set ℓ = 1 for the following discussions. We also assume that the coordinate φ is periodic with ∆φ = 2π. Thus the solution describes a rotating metric. The null-Killing vector on the horizon r = r 0 with f (r 0 ) = 0 is given by It is straightforward to verify that the surface gravity and hence the temperature are given The mass and angular momentum can be read off from the Wald formalism, given by Two situations emerge at this stage. For black holes of the EPA theory or the effective theories in three dimensions, the parameter q and hence α are fixed constants. In these cases, the first law of black hole thermodynamics reads In the EMA theory; on the other hand, the parameter q is an integration constant, and hence it can be varied and should be involved in the first law. To complete the first law involving the parameter q, we first note that the electric charge of the Maxwell field vanishes, namely * F + dχ ∧ A = 0 . (4.10) (In [13], an electric charge associated with pure gauge transformation of A was introduced. We shall not consider this here.) The linear charge density of the axion field on the other hand is non-vanishing The corresponding thermodynamical potential can be read off from the 2-form potential B (2) that is Hodge dual to the axion, as in (3.7). It is given by We find the first law reads It is puzzling that an extra factor α is needed for the completion of the first law above.

Type II
In this interpretation, we switch t and φ in (4.2) and write the metric as (4.14) Note that we also made a coordinate transformation so that the null Killing vector at the degenerate surface r = r 0 withf (r 0 ) = 0 is ξ = ∂ t . In other words, the metric is non-rotating on the horizon. The temperature is given by The solution has no CTCs since g φφ = ℓ 2 and further more g tt > 0 for r > r 0 , and hence t is globally defined outside the horizon. Note that in this case, the entropy is a constant since the radius of the φ circle is constant. We can also show, using the Wald formalism that the mass and angular momentum both vanish. The solution can be viewed as thermalized vacuum. In the EMA theory, q is an integration constant, which leads to non-zero electric charge and potential, give by The axion charge and its thermodynamical potential are given by (4.11) and (4.12). This leads to the first law of black hole "thermodynamics" provided that ∆φ = π.

Freedman-Schwarz model
In section 3 we constructed some ad hoc theories that admit Gödel metrics as exact solutions.
The Maxwell and axion fields are common occurrence in supergravities, indicating that there may exist an exact embedding of Gödel universe in supergravity and hence in string theory.
In this section, we consider Freedman-Schwarz SU (2) × SU (2) gauged supergravity whose bosonic sector consists of the metric, a dilaton ϕ, an axion and two SU (2) Yang-Mills fields.
After truncating to the U (1) 2 subsector, the corresponding Lagrangian is where (g 1 , g 2 ) are the gauge coupling constants of the two SU (2) Yang-Mills fields. The theory admits the general deformed Gödel metric (2.9) with the matter fields given by with the parameters

Gödel universe from string theory
Freedman-Schwarz model can be obtained from the Kaluza-Klein reduction on S 3 × S 3 [27,28]. The relevant part of the effective Lagrangian of strings in ten dimensions is where F (3) can be either NS-NS or R-R fields. Following the reduction ansatz given in [28], we find that the ten-dimensional solution is given by Φ = 0, together with (dψ 2 + cos θ 2 dφ 2 + g 2 q 2 rdφ) 2 + dθ 2 2 + sin 2 θ 2 dφ 2 2 , (5.5) Here α is again given by (5.3). The solution involves both electric string and magnetic fivebrane/fivebrane charges, given by electric : (5.7) These can be either all NS-NS charges or R-R charges, and the latter corresponds to the D1/D5/D5 configuration. When q 1 = 0 = q 2 , the metric becomes AdS 3 ×S 3 ×S 3 ×R, which is the decoupling limit of the string/fivebrane/fivebrane configuration [30]. The rotations associated with parameters (q 1 , q 2 ) turn the AdS 3 into the G α . We thus expect that there should be a rotating string/fivebrane/fivebrane configuration whose decoupling limit gives rise to our ten-dimensional solution (5.6). If we set either g 1 = 0 or g 2 = 0, but not both, the associated S 3 is flatten to become R 3 . The metric configuration becomes G α × S 3 × R 4 .
The heterotic string solution of G α × S 3 × K 3 was first constructed in [11].
To study the global structure, we first denote r 0 as the largest root of f (r). Shifting the coordinates as we find that the metric is singular at r = r 0 , where the degenerate Killing vector is purely spatial ξ = ∂ φ . The absence of a conical singularity requires that . (5.9) Thus in this system, there are three periodic coordinates φ, and (ψ 1 , ψ 2 ), with ∆ψ 1 = 4π = ∆ψ 2 . The coordinate t on the other hand is not required to be periodic. The analysis of CTCs in ten dimensions becomes more subtle. Note that we have for the region r ≥ r 0 ; however, naked CTCs still exist. One way to see this is to consider the general periodic Killing vector The absence of naked CTCs requires that ξ 2 ≥ 0 for all real (β, γ 1 , γ 2 ) in the r ≥ r 0 region.
This can be easily established not true. Negative modes arise for large enough r.

Conclusions
Four-dimensional Gödel metrics of G α × R are perhaps the simplest solutions that exhibit naked CTCs with no globally spatial-like Cauchy horizon. In this paper, we showed that the Gödel metrics could arise as exact solutions in Lagrangian formalism. We constructed EMA and EPA theories that admit Gödel solutions. We also showed that Gödel universe could emerge from Freedman-Schwarz SU (2) × SU (2) gauge supergravity. This allows us to give exact embeddings of the Gödel metrics in string theories. The ten-dimensional solution describes a direct product of a line and an S 3 × S 3 bundle over G α . Classically, we find that naked CTCs persist in higher dimensions. (In [11], string quantization was performed on G α × S 3 × K 3 and it was demonstrated that CTCs can resolved by the quantum effects.) For some appropriate choice of parameters, the nine-dimensional metric can describe an AdS 3 × S 3 bundle over a squashed 3-sphere S 3 α , in which case, there is no CTC. In the suitable limit, the solution becomes the supersymmetric AdS 3 × S 3 × R 4 vacuum.
The scaling symmetry of the metric (2.1) resembles that of the anti-de Sitter spacetimes. This is suggestive that there may exist a boundary field theory at the r → ∞ boundary of Gödel universe. The exact embedding of the Gödel metrics in string theory, as the decoupling limit of the rotating D1/D5/D5 intersection, provides a tool of investigating the boundary field theory in the context of string theory.