# Cosmology and Brane Worlds: A Review

###### Abstract

Cosmological consequences of the brane world scenario are reviewed in a pedagogical manner. According to the brane world idea, the standard model particles are confined on a hyper–surface (a so–called brane), which is embedded in a higher–dimensional spacetime (the so–called bulk). We begin our review with the simplest consistent brane world model: a single brane embedded in a five–dimensional Anti-de Sitter space–time. Then we include a scalar field in the bulk and discuss in detail the difference with the Anti-de Sitter case. The geometry of the bulk space–time is also analysed in some depth. Finally, we investigate the cosmology of a system with two branes and a bulk scalar field. We comment on brane collisions and summarize some open problems of brane world cosmology.

## 1 Introduction

The idea of extra dimensions was proposed in the early twentieth century by Nordstrom and a few years later by Kaluza and Klein [1]. It has reemerged over the years in theories combining the principles of quantum mechanics and relativity. In particular theories based on supersymmetry, especially superstring theories, are naturally expressed in more than four dimensions [2]. Four dimensional physics is retrieved by Kaluza–Klein reduction, i.e compactifying on a manifold of small size, typically much smaller than the size of an atomic nucleus.

Recent developments in string theory and its extension M–theory have suggested another approach to compactify extra spatial dimensions. According to these developments, the standard model particles are confined on a hypersurface (called brane) embedded in a higher dimensional space (called bulk). Only gravity and other exotic matter such as the dilaton can propagate in the bulk. Our universe may be such a brane–like object. This idea was originally motivated phenomenologically (see [3]–[7]) and later revived in string theory. Within the brane world scenario, constraints on the size of extra dimensions become weaker, because the standard model particles propagate only in three spatial dimensions. Newton’s law of gravity, however, is sensitive to the presence of extra–dimensions. Gravity is being tested only on scales larger than a tenth of a millimeter and possible deviations below that scale can be envisaged.

From the string theory point of view, brane worlds of the kind discussed in this review spring from a model suggested by Horava and Witten [8]. The strong coupling limit of the heterotic string theory at low energy is described by eleven dimensional supergravity with the eleventh dimension compactified on an orbifold with symmetry, i.e. an interval. The two boundaries of spacetime (i.e. the orbifold fixed points) are 10–dimensional planes, on which gauge theories (with the gauge groups) are confined. Later Witten argued that 6 of the 11 dimensions can be consistently compactified on a Calabi–Yau threefold and that the size of the Calabi-Yau manifold can be substantially smaller than the space between the two boundary branes [9]. Thus, in that limit space–time looks five–dimensonal with four dimensional boundary branes [10]. This provides the underlying picture for many brane world models proposed so far.

Another important ingredient was put forward by Arkani-Hamed, Dimopoulos and Dvali (ADD), [11] and [12], following an earlier idea by Antoniadis [13], who suggested that by confining the standard model particle on a brane the extra dimensions can be larger than previously anticipated. They considered a flat bulk geometry in ()–dimensions, in which dimensions are compact with radius (toroidal topology). The four–dimensional Planck mass and the ()–dimensional Planck mass , the gravitational scale of the extra dimensional theory, are related by

(1) |

Gravity deviates from Newton’s law only on scales smaller than . Since gravity is tested only down to sizes of around a millimeter, could be as large as a fraction of a millimeter.

ADD assumed that the bulk geometry is flat. Considerable progress was made by Randall and Sundrum, who considered non–flat, i.e. warped bulk geometries [14], [15]. In their models, the bulk spacetime is a slice of Anti–de Sitter spacetime, i.e. a space-time with a negative cosmological constant. Their discovery was that, due to the curvature of the bulk space time, Newton’s law of gravity can be obtained on the brane of positive tension embedded in an infinite extra–dimension. Small corrections to Newton’s law are generated and constrain the possible scales in the model to be smaller than a millimetre.

They also proposed a two–brane model in which the hierarchy problem, i.e. the large discrepancy between the Planck scale at GeV and the electroweak scale at GeV, can be addressed. The large hierarchy is due to the highly curved AdS background which implies a large gravitational red-shift between energy scale on the two branes. In this scenario, the standard model particles are confined on a brane with negative tension sitting at , whereas a positive tension brane is located at . The large hierarchy is generated by the appropriate inter-brane distance, i.e. the radion. It can be shown that the Planck mass measured on the negative tension brane is given by (),

(2) |

where is the five–dimensional Planck mass and the (negative) cosmological constant in the bulk. Thus, we see that, if is not very far from the electroweak scale TeV, we need , in order to generate a large Planck mass on our brane. Hence, by tuning the radius of the extra dimension to a reasonable value, one can obtain a very large hierarchy between the weak and the Planck scale. Of course, a complete realization of this mechanism requires an explanation for such a value of the radion. In other words, the radion needs to be stabilized at a certain value. The stabilization mechanism is not thoroughly understood, though models with a bulk scalar field have been proposed and have the required properties [16].

Another puzzle which might be addressed with brane models is the cosmological constant problem. One may invoke an extra dimensional origin for the apparent (almost) vanishing of the cosmological constant. The self-tuning idea [17] advocates that the energy density on our brane does not lead to a large curvature of our universe. On the contrary, the extra dimension becomes highly curved, preserving a flat Minkowski brane with apparent vanishing cosmological constant. Unfortunately, the simplest realization of this mechanism with a bulk scalar field fails due to the presence of a naked singularity in the bulk. This singularity can be shielded by a second brane whose tension has to be fine-tuned with the original brane tension [18]. In a sense, the fine tuning problem of the cosmological constant reappears through the extra dimensional back-door.

Finally, we will later discuss in some detail another spectacular consequence of brane cosmology, namely the possible modification to the Friedmann equation at very high energy [19]. This effect was first recognised in [20] in the context of inflatonary solutions. As we will see, Friedmann’s equation has, for the Randall–Sundrum model, the form ([21] and [22])

(3) |

relating the expansion rate of the brane to the (brane) matter density and the (effective) cosmological constant . The cosmological constant can be tuned to zero by an appropriate choice of the brane tension and bulk cosmological constant, as in the Randall-Sundrum case. Notice that at high energies, for which

(4) |

where is the five dimensional gravitational constant, the Hubble rate becomes

(5) |

while in ordinary cosmology . The latter case is retrieved at low energy, i.e.

(6) |

Of course modifications to the Hubble rate can only be significant before nucleosynthesis. They may have drastic consequences on early universe phenomena such as inflation.

In this article we will review these and other aspects of the brane world idea in the context of cosmology. In order to give a pedagogical introduction to the subject, we will follow a phenomenological approach and start with the simplest model, i.e. the Randall-Sundrum model, with a brane embedded in a five–dimensional vacuum bulk spacetime (section 2). Later we will include a bulk scalar field (section 3). In section 4 we will discuss the geometry of the bulk–spacetime in some detail. In the last part we will discuss more realistic models with two branes and bulk scalar fields (section 5). In section 6 we will discuss brane collisions. Open questions are summarized in section 7. We would like to mention other review articles on brane worlds and cosmology, taking different approaches from the one taken here [23]-[28]. We will mostly be concerned with the case, in which the bulk space–time is five–dimensional.

## 2 The Randall–Sundrum Brane World

Originally, Randall and Sundrum suggested a two–brane scenario in five dimensions with a highly curved bulk geometry as an explanation for the large hierarchy between the Planck scale and the electroweak energy–scale [14]. In this scenario, the standard model particles live on a brane with (constant) negative tension, whereas the bulk is a slice of Anti–de Sitter (AdS) spacetime , i.e. a space-time with a negative cosmological constant. In the bulk there is another brane with positive tension. This is the so–called Randall–Sundrum I (RSI) model. Analysing the solution of Einstein’s equation on the positive tension brane and sending the negative tension brane to infinity, an observer confined to the positive tension brane recovers Newton’s law if the curvature scale of the AdS is smaller than a millimeter [15]. The higher–dimensional space is non–compact, which must be contrasted with the Kaluza–Klein mechanism, where all extra–dimensional degrees of freedom are compact. This one–brane model, on which we will concentrate in this section, is the so–called Randall–Sundrum II (RSII) model. It was shown, there is a continuum of Kaluza–Klein modes for the gravitational field, contrasting with the discrete spectrum if the extra dimension is periodic. This leads to a correction to the force between two static masses on the brane. To be specific, it was shown that the potential energy between two point masses confined on the brane is given by

(7) |

In this equation, is related to the five–dimensional bulk cosmological constant by and is therefore a measure of the curvature scale of the bulk spacetime. Gravitational experiments show no deviation from Newton’s law of gravity on length scales larger than a millimeter [29]. Thus, has to be smaller than that length scale.

The static solution of the Randall and Sundrum model can be obtained as follows: The total action consists of the Einstein-Hilbert action and the brane action, which in the Randall–Sundrum model have the form

(8) | |||||

(9) |

The parameter (the bulk cosmological constant) and (the brane tension) are constant and is the five–dimensional gravitational coupling constant. The brane is located at and we assume a symmetry, i.e. we identify with . The ansatz for the metric is

(10) |

Einstein’s equations, derived from the action above, give two independent equations:

The first equation can be easily solved:

(11) |

which tells us that must be negative. If we integrate the second equation from to , take the limit and make use of the –symmetry, we get

(12) |

Together with eq. (11) this tells us that

(13) |

Thus, there must be a fine–tuning between the brane tension and the bulk cosmological constant for static solutions to exist. In this section we will discuss the cosmology of this model in detail.

### 2.1 Einstein’s equations on the brane

There are two ways of deriving the cosmological equations and we will describe both of them below. The first one is rather simple and makes use of the bulk equations only. The second method uses the geometrical relationship between four–dimensoinal and five–dimensional quantities. We begin with the simpler method.

#### 2.1.1 Friedmann’s equation from five–dimensional Einstein equations

In the following subsection we will set . We write the bulk metric as follows:

(14) |

This metric is consistent with homogeneity and isotropy on the brane located at . The functions and are functions of and only. Furthermore, we have assumed flat spatial sections, it is straightforward to include a spatial curvature. For this metric, Einstein equations in the bulk read:

(15) | |||||

(16) | |||||

(17) | |||||

(18) | |||||

where the bulk energy–momentum tensor has been kept general here. For the Randall–Sundrum model we will now take and . Later, in the next section, we will use these equations to derive Friedmann’s equation with a bulk scalar field. In the equations above, a dot represents the derivative with respect to and a prime a derivative with respect to .

Let us integrate the 00–component over from to and use the fact that , , and (i.e. -symmetry). Then, taking the limit we get

(19) |

Integrating the –component in the same way and using the last equation gives

(20) |

These two conditions are called the junction conditions. The other components of the Einstein equations should be compatible with these conditions. It is not difficult to show that the restriction of the component to leads to

(21) |

where we have made use of the junction conditions (19) and (20). This is nothing but matter conservation on the brane.

Proceeding in the same way with the 55–component gives

(22) |

Changing to cosmic time , writing and using the energy conservation gives ([30], [31])

(23) |

In this equation . This equation can easily be integrated to give

(24) |

The final step is to split the total energy–density and pressure into parts coming from matter and brane tension, i.e. to write and . Then we find Friedmann’s equation

(25) |

where we have made the identification

(26) | |||||

(27) |

Comparing the last equation with the fine–tuning relation (13) in the static Randall–Sundrum solution, we see that in this case. If there is a small mismatch between the brane tension and the five–dimensional cosmological constant, then an effective four–dimensional cosmological constant is generated. Another important point is that the four–dimensional Newton constant is directly related to the brane tension. The constant appears in the derivation above as an integration constant. The term including is called the dark radiation term (see e.g. [32]-[34]). The parameter can be obtained from a full analysis of the bulk equations [35]-[37] (we will discuss this in section 4). An extended version of Birkhoff’s theorem tells us that if the bulk spacetime is AdS, this constant is zero [38]. If the bulk is AdS–Schwarzschild instead, is non–zero but a measure of the mass of the bulk black hole. In the following we will assume that and .

The most important change in Friedmann’s equation compared to the usual four–dimensional form is the appearance of a term proportional to . It tells us that if the matter energy density is much larger than the brane tension, i.e. , the expansion rate is proportional , instead of . The expansion rate is, in this regime, larger in this brane world scenario. Only in the limit where the brane tension is much larger than the matter energy density, the usual behaviour is recovered. This is the most important change in brane world scenarios. It is quite generic and not restricted to the Randall–Sundrum brane world model. From Friedmann’s equation and from the energy–conservation equation we can derive Raychaudhuri’s equation:

(28) |

We will use these equations later in order to investigate inflation driven by a scalar field confined on the brane.

Notice that at the time of nucleosythesis the brane world corrections in Friedmann’s equation must be negligible, otherwise the expansion rate is modified and the results for the abundances of the light elements are unacceptably changed. This implies that . Note, however, that a much stronger constraint arises from current tests for deviation from Newton’s law [39] (assuming the Randall–Sundrum fine–tuning relation (13)): TeV and . Similarily, cosmology constrains the amount of dark radiation. It has been shown that the energy density in dark radiation can at most be 10 percent of the energy density in photons [33].

#### 2.1.2 Another derivation of Einstein’s equation

There is a more powerful way of deriving Einstein’s equation on the brane [40]. Consider an arbitrary (3+1) dimensional hypersurface with unit normal vector embedded in a 5 dimensional spacetime. The induced metric and the extrinsic curvature of the hypersurface are defined as

(29) | |||||

(30) |

For the derivation we need three equations, two of them relate four–dimensional quantities constructed from to full five–dimensional quantities constructed from . We just state these equations here and refer to [41] for the derivation of these equations. The first equation is the Gauss equation, which reads

(31) |

This equation relates the four–dimensional curvature tensor , constructed from , to the five–dimensional one and . The next equation is the Codazzi equation, which relates , and the five–dimensional Ricci tensor:

(32) |

One decomposes the five–dimensional curvature tensor into the Weyl–tensor and the Ricci tensor:

(33) |

If one substitutes the last equation into the Gauss equation and constructs the four–dimensional Einstein tensor, one obtains

(34) | |||||

where

(35) |

We would like to emphasize that this equation holds for any hypersurface. If one considers a hypersurface with energy momentum tensor , then there exists a relationship between and ( is the trace of ) [42]:

(36) |

where denotes the jump:

(37) |

These equations are called junction conditions and are equivalent in the cosmological context to the junction conditions (19) and (20). Splitting and inserting the junction condition into equation (34), we obtain Einstein’s equation on the brane:

(38) |

The tensor is defined as

(39) |

whereas

(40) | |||||

(41) |

Note that in the Randall–Sundrum case we have due to the fine–tuning between the brane tension and the bulk cosmological constant. Moreover as the Weyl–tensor vanishes for an AdS spacetime. In general, the energy conservation and the Bianchi identities imply that

(42) |

on the brane.

Clearly, this method is powerful, as it does not assume homogeneity and isotropy nor does it assume the bulk to be AdS. In the case of an AdS bulk and a Friedmann–Robertson walker brane, the previous equations reduce to the Friedmann equation and Raychaudhuri equation derived earlier. However, the set of equations on the brane are not closed in general [43], as we will see in the next section.

### 2.2 Slow–roll inflation on the brane

We have seen that the Friedmann equation on a brane is drastically modified at high energy where the terms dominate. As a result the early universe cosmology on branes tends to be different from standard 4d cosmology. In that vein it seems natural to look for brane effects on early universe phenomena such as inflation (see in particular [44] and [45]) and on phase–transitions [46].

The energy density and the pressure of a scalar field are given by

(43) | |||||

(44) |

where is the potential energy of the scalar field. The full evolution of the scalar field is described by the (modified) Friedmann equation, the Klein–Gordon equation and the Raychaudhuri equation.

We will assume that the field is in a slow–roll regime, the evolution of the fields is governed by (from now on a dot stands for a derivative with respect to cosmic time)

(45) | |||||

(46) |

It is not difficult to show that these equations imply that the slow–roll parameter are given by

(47) | |||||

(48) |

The modifications to General Relativity are contained in the square brackets of these expressions. They imply that for a given potential and given initial conditions for the scalar field the slow–roll parameters are suppressed compared to the predictions made in General Relativity. In other words, brane world effects ease slow–roll inflation [44]. In the limit the parameter are heavily suppressed. It implies that steeper potentials can be used to drive slow–roll inflation [45]. Let us discuss the implications for cosmological perturbations.

According to Einstein’s equation (38), perturbations in the metric are sourced not only by matter perturbations but also by perturbations of the bulk geometry, encoded in the perturbation of . This term can be seen as an external source for perturbations, absent in General Relativity. If one regards as an energy–momentum tensor of an additional fluid (called the Weyl-fluid), its evolution is connected to the energy density of matter on the brane, as one can see from eq. (42). If one neglects the anisotropic stress of the Weyl-fluid, then at low energy and superhorizon scales, it decays as radiation, i.e. . However, the bulk gravitational field exerts an anisotropic stress onto the brane, whose time-evolution cannot be obtained by considering the projected equations on the brane alone [43]. Rather, the full five–dimensional equations have to be solved, together with the junction conditions. The full evolution of in the different cosmological eras is currently not understood. However, as we will discuss below, partial results have been obtained for the case of a de Sitter brane, which suggest that does not change the spectrum of scalar perturbations. It should be noted however, that the issue is not settled and that it is also not clear if the subsequent cosmological evolution during radiation and matter era leaves an imprint of the bulk gravitational field in the anisotropies of the microwave background radiation [47]. With this in mind, we will, for scalar perturbations, neglect the gravitational backreaction described by the projected Weyl tensor.

Considering scalar perturbations for the moment, the perturbed line element on the brane has the form

(49) |

where the functions , , and depend on and .

An elegant way of discussing scalar perturbations is to make use of of the gauge invariant quantity [48]

(50) |

In General Relativity, the evolution equation for can be obtained from the energy–conservation equation [49]. It reads, on large scales,

(51) |

where is the non-adiabatic pressure perturbation. The energy conservation equation, however, holds for the Randall–Sundrum model as well. Therefore, eq. (51) is still valid for the brane world model we consider. For inflation driven by a single scalar field vanishes and therefore is constant on superhorizon scales during inflation. Its amplitude is given in terms of the fluctuations in the scalar field on spatially flat hypersurfaces:

(52) |

The quantum fluctuation in the (slow–rolling) scalar field obey , as the Klein–Gordon equation is not modified in the brane world model we consider. The amplitude of scalar perturbations is [50] . Using the slow–roll equations and eq. (52) one obtains [44]

(53) |

Again, the corrections are contained in the terms in the square brackets. For a given potential the amplitude of scalar perturbations is enhanced compared to the prediction of General Relativity.

The arguments presented so far suggest that, at least for scalar perturbations, perturbations in the bulk spacetime are not important during inflation. This, however, might not be true for tensor perturbations, as gravitational waves can propagate into the bulk. For tensor perturbations, a wave equations for a single variable can be derived [51]. The wave equation can be separated into a four–dimensional and a five–dimensonal part, so that the solution has the form , where is a (constant) polarization tensor. One finds that the amplitude for the zero mode of tensor perturbation is given by [51]

(54) |

with

(55) |

where we have defined

(56) |

It can be shown that modes with are generated but they decay during inflation. Thus, one expects in this scenario only the massless modes to survive until the end of inflation [51], [52].

From eqns. (54) and (53) one sees that the amplitudes of scalar and tensor perturbations are enhanced at high energies. However, scalar perturbations are more enhanced than tensors. Thus, the relative contribution of tensor perturbations will be suppressed, if inflation is driven at high energies.

Finally, we would like to mention that there are also differences between General Relativity and the brane world model we consider for the prediction of two–field brane inflation. Usually correlations are separated in adiabatic and isocurvature modes for two–field inflation [53]. In the Randall–Sundrum model, this correlation is suppressed if inflation is driven at high energies [54]. This implies that isocurvature and adiabatic perturbations are uncorrelated, if inflation is driven at energies much larger than the brane tension.

Coming back to cosmological perturbations, the biggest problem is that the evaluation of the projected Weyl tensor is only possible for the background cosmology. As soon as one tries to analyse the brane cosmological perturbations, one faces the possibility that the terms might not vanish. In particular this means that the equation for the density contrast , which is given by (, is the wavenumber)

(57) |

cannot be solved as involves and can therefore not be deduced solely from the brane dynamics [43].

### 2.3 Final Remarks on the Randall–Sundrum Scenario

The Randall–Sundrum model discussed in this section is the simplest brane world model. We have not discussed other important conclusions one can draw from the modifications of Friedmann’s equation, such as the evolution of primordial black holes [55], its connection to the AdS/CFT correspondence (see e.g. [56]-[61]) and inflation driven by the trace anomaly of the conformal field theory living on the brane (see e.g. [62]-[65]). These developments are important in many respects, because they give not only insights about the early universe but gravity itself. They will not be reviewed here.

## 3 Including a Bulk Scalar Field

In this section we are going to generalize the previous results obtained with an empty bulk. To be specific, we will consider the inclusion of a scalar field in the bulk. As we will see, one can extend the projective approach wherein one focuses on the dynamics of the brane, i.e. one studies the projected Einstein and the Klein-Gordon equation [66], [67]. As in the Randall-Sundrum setting, the dynamics do not closed, as bulk effects do not decouple. We will see that there are now two objects representing the bulk back-reaction: the projected Weyl tensor and the loss parameter . In the case of homogeneous and isotropic cosmology on the brane, the projected Weyl tensor is determined entirely up to a dark radiation term. Unfortunately, no information on the loss parameter is available. This prevents a rigorous treatment of brane cosmology in the projective approach.

Another route amounts to studying the motion of a brane in a bulk space-time. This approach is successful in the Randall-Sundrum case thanks to Birkhoff’s theorem which dictates a unique form for the metric in the bulk [38]. In the case of a bulk scalar field, no such theorem is available. One has to resort to various ansatze for particular classes of bulk and brane scalar potentials (see e.g. [68]–[75]). We will come back to this in section 4.

### 3.1 BPS Backgrounds

#### 3.1.1 Properties of BPS Backgrounds

As the physics of branes with bulk scalar fields is pretty complicated, we will start with a particular example where both the bulk and the brane dynamics are fully under control [77] (see also [78] and [79]). We specify the bulk Lagrangian as

(58) |

where is the bulk potential. The boundary action depends on a brane potential

(59) |

where is evaluated on the brane. The BPS backgrounds arise as particular case of this general setting with a particular relationship between the bulk and brane potentials. This relation appears in the study of supergravity with vector multiplets in the bulk. The bulk potential is given by

(60) |

where is the superpotential. The brane potential is simply given by the superpotential

(61) |

We would like to mention, that the last two relations have been also used in order to generate bulk solutions without necessarily imposing supersymmetry [70],[76]. Notice that the Randall-Sundrum case can be retrieved by putting . Supergravity puts further constraints on the superpotential which turns out to be of the exponential type [77]

(62) |

with . In the following we will often choose this exponential potential with an arbitrary as an example. The actual value of does not play any role and will be considered generic.

The bulk equations of motion comprise the Einstein equations and the Klein-Gordon equation. In the BPS case and using the following ansatz for the metric

(63) |

these second order differential equations reduce to a system of two first order differential equations

(64) | |||||

Notice that when one easily retrieves the exponential profile of the Randall-Sundrum model.

An interesting property of BPS systems can be deduced from the study of the boundary conditions. The Israel junction conditions reduce to

(65) |

and for the scalar field

(66) |

This is the main property of BPS systems: the boundary conditions coincide with the bulk equations, i.e. as soon as the bulk equations are solved one can locate the BPS branes anywhere in this background, there is no obstruction due to the boundary conditions. In particular two-brane systems with two boundary BPS branes admit moduli corresponding to massless deformations of the background. They are identified with the positions of the branes in the BPS background. We will come back to this later in section 5.

Let us treat the example of the exponential superpotential. The solution for the scale factor reads

(67) |

and the scalar field is given by

(68) |

For , the bulk scalar field decouples and these expressions reduce to the Randall-Sundrum case. Notice a new feature here, namely the existence of singularities in the bulk, corresponding to

(69) |

In order to analyse singularities it is convenient to use conformal coordinates

(70) |

In these coordinates light follows straight lines . It is easy to see that the singularities fall in two categories depending on . For the singularity is at infinity . This singularity is null and absorbs incoming gravitons. For the singularity is at finite distance. It is time-like and not wave-regular, i.e. the propagation of wave packets is not uniquely defined in the vicinity of the singularity. For all these reasons these naked singularities in the bulk are a major drawback of brane models with bulk scalar fields [80]. In the two-brane case the second brane has to sit in front of the naked singularity.

#### 3.1.2 de Sitter and anti de Sitter Branes

Let us modify slightly the BPS setting by detuning the tension of the BPS brane. This corresponds to adding or substracting some tension compared to the BPS case

(71) |

where is real number. Notice that this modification only affects the boundary conditions, the bulk geometry and scalar field are still solutions of the BPS equations of motion. In this sort of situation, one can show that the brane does not stay static. For the detuned case, the result is either a boosted brane or a rotated brane. We will soon generalize these results so we postpone the detailed explanation to later. Defining by the position of the brane in conformal coordinates, one obtains

(72) |

The brane velocity vector is of constant norm. For , the brane velocity vector is time-like and the brane moves at constant speed. For the brane velocity vector is space-like and the brane is rotated. Going back to a static brane, we see that the bulk geometry and scalar field become dependent. In the next section we will find many more cases where branes move in a static bulk or equivalently, a static brane borders a non-static bulk.

Let us now conclude this section by studying the brane geometry when . In particular one can study the Friedmann equation for the induced bulk factor

(73) |

where is evaluated on the brane. Of course we obtain the fact that cosmological solutions are only valid for . Now in the Randall-Sundrum case leading to

(74) |

In the case the brane geometry is driven by a positive cosmological constant. This is a de Sitter brane. When the cosmological constant is negative, corresponding to an AdS brane. We are going to generalize these results by considering the projective approach to the brane dynamics.

### 3.2 Bulk Scalar Fields and the Projective Approach

#### 3.2.1 The Friedmann Equation

We will first follow the traditional coordinate dependent path. This will allow us to derive the matter conservation equation, the Klein-Gordon and the Friedmann equations on the brane. Then we will concentrate on the more geometric formulation where the role of the projected Weyl tensor will become transparent [86],[87]. Again, in this subsection we will put .

We consider a static brane that we choose to put at the origin . and impose . This guarantees that the brane and bulk expansion rates

(75) |

coincide. We have identified the brane cosmic time . We will denote by prime the normal derivative . Moreover we now allow for some matter to be present on the brane

(76) |

The bulk energy-momentum tensor reads

(77) |

The total matter density and pressure on the brane are given by

(78) |

The boundary condition for the scalar field is unchanged by the presence of matter on the brane.

The Einstein equation leads to matter conservation

(79) |

By restricting the component of the Einstein equations we obtain

(80) |

in units of . The last term is the dark radiation, whose origin is similar to the Randall-Sundrum case. The quantity and satisfy the differential equations [31]

These equations can be easily integrated to give

(81) |

up to a dark radiation term and we have identified

(82) |

This is the Friedmann equation for a brane coupled to a bulk scalar field. Notice that retarded effects springing from the whole history of the brane and scalar field dynamics are present. In the following section we will see that these retarded effects come from the projected Weyl tensor. They result from the exchange between the brane and the bulk. Notice, that Newton’s constant depends on the value of the bulk scalar field evaluated on the brane ():

(83) |

On cosmological scale, time variation of the scalar field induce a time variations of Newton’s constant. This is highly constrained experimentally [88],[89], leading to tight restrictions on the time dependence of the scalar field.

To get a feeling of the physics involved in the Friedmann equation, it is convenient to assume that the scalar field is evolving slowly on the scale of the variation of the scale factor. Neglecting the evolution of Newton’s constant, the Friedmann equation reduces to

(84) |

Several comments are in order. First of all we have neglected the contribution due to the term as we are considering energy scales below the brane tension. Then the main effect of the scalar field dynamics is to involve the potential energy and the kinetic energy . Although the potential energy appears with a positive sign we find that the kinetic energy has a negative sign. For an observer on the brane this looks like a violation of unitarity. We will reanalyse this issue in section 5, when considering the low energy effective action in four dimensions and we will see that there is no unitarity problem in this theory. The minus sign for the kinetic energy is due to the fact that one does not work in the Einstein frame where Newton’s constant does not vary, a similar minus sign appears also in the effective four–dimensional theory when working in the brane frame.

The time dependence of the scalar field is determined by the Klein-Gordon equation. The dynamics is completely specified by

(85) |

where . We have identified

(86) |

This cannot be set to zero and requires the knowledge of the scalar field in the vicinity of the brane. When we discuss cosmological solutions below, we will assume that this term is negligible.

The evolution of the scalar field is driven by two effects. First of all, the scalar field couples to the trace of the energy momentum tensor via the gradient of . Secondly, the field is driven by the gradient of the potential , which might not necessarily vanish.

#### 3.2.2 The Friedmann equation vs the projected Weyl tensor

We are now coming back to the origin of the non-trivial Friedmann equation. Using the Gauss-Codazzi equation one can obtain the Einstein equation on the brane [66],[67]

(87) |

Now the projected Weyl tensor can be determined in the homogeneous and isotropic cosmology case. Indeed only the component is independent. Using the Bianchi identity where is the brane covariant derivative, one obtains that

(88) |

leading to

(89) |

Upon using

(90) |

one obtains the Friedmann equation. It is remarkable that the retarded effects in the Friedmann equation all spring from the projected Weyl tensor. Hence the projected Weyl tensor proves to be much richer in the case of a bulk scalar field than in the empty bulk case.

#### 3.2.3 Self-Tuning and Accelerated Cosmology

The dynamics of the brane is not closed, it is an open system continuously exchanging energy with the bulk. This exchange is characterized by the dark radiation term and the loss parameter. Both require a detailed knowledge of the bulk dynamics. This is of course beyond the projective approach where only quantities on the brane are evaluated. In the following we will assume that the dark radiation term is absent and that the loss parameter is negligible. Furthermore, we will be interested in the effects of a bulk scalar field for late–time cosmology (i.e. well after nucleosynthesis) and not in the case for inflation driven by a bulk scalar field (see e.g. [81]-[85]).

Let us consider the self-tuned scenario as a solution to the cosmological constant problem. It corresponds to the BPS superpotential with . In that case the potential for any value of the brane tension. The potential can be interpreted as a vanishing of the brane cosmological constant. The physical interpretation of the vanishing of the cosmological constant is that the brane tension curves the fifth dimensional space-time leaving a flat brane intact. Unfortunately, the description of the bulk geometry in that case has shown that there was a bulk singularity which needs to be hidden by a second brane whose tension is fine-tuned with the first brane tension. This reintroduces a fine-tuning in the putative solution to the cosmological constant problem [18].

Let us generalize the selftuned case to , i.e. and is the exponential superpotential. The resulting induced metric on the brane is of the FRW type with a scale factor

(91) |

leading to an acceleration parameter

(92) |

For the supergravity value this leads to . This is in coincidental agreement with the supernovae results. This model can serve as a brane quintessence model [77],[86]. We will comment on the drawbacks of this model later. See also [91] and [92] for similar ideas.

#### 3.2.4 The brane cosmological eras

Let us now consider the possible cosmological scenarios with a bulk scalar field [86],[87]. We assume that the potential energy of the scalar field is negligible throughout the radiation and matter eras before serving as quintessence in the recent past.

At very high energy above the tension of the brane the non-conventional cosmology driven by the term in the Friedmann equation is obtained. Assuming radiation domination, the scale factor behaves like

(93) |

and the scalar field

(94) |

In the radiation dominated era, no modification is present, provided

(95) |

which is a solution of the Klein-Gordon equation as the trace of the energy-momentum of radiation vanishes (together with a decaying solution, which we have neglected). In the matter dominated era the scalar field evolves due to the coupling to the trace of the energy-momentum tensor. This has two consequences. Firstly, the kinetic energy of the scalar field starts contributing in the Friedmann equation. Secondly, the effective Newton constant does not remain constant. The cosmological evolution of Newton’s constant is severely constrained since nucleosynthesis [88],[89]. This restricts the possible time variation of .

In order to be more quantitative let us come back to the exponential superpotential case with a detuning parameter . The time dependence of the scalar field and scale factor become

where and are the time and scale factor at matter-radiation equality. Notice the slight discrepancy of the scale factor exponent with the standard model value of . The redshift dependence of the Newton constant is

(96) |

For the supergravity model with and this leads to a decrease by (roughly) 37% since nucleosynthesis. This is marginally compatible with experiments [88],[89].

Finally let us analyse the possibility of using the brane potential energy of the scalar field as the source of acceleration now. We have seen that when matter is negligible on the brane, one can build brane quintessence models. We now require that this occurs only in the recent past. As can be expected, this leads to a fine-tuning problem as

(97) |

where is the amount of detuned tension on the brane. Of course this is nothing but a reformulation of the usual cosmological constant problem. Provided one accepts this fine-tuning, as in most quintessence models, the exponential model with is a cosmological consistent quintessence model with a five dimensional origin.

### 3.3 Brief summary

The main difference between a brane world model with a bulk scalar field and the Randall–Sundrum model is that the gravitational constant becomes time–dependent. As such it has much in common with scalar–tensor theories [90], but there are important differences due to the projected Weyl tensor and its time–evolution. The bulk scalar field can play the role of the quintessence field, as discussed above, but it could also play a role in an inflationary era in the very early universe (see e.g. [81]-[85]). In any case, the cosmology of such a system is much richer and, because of the variation of the gravitational constant, more constrained. It remains to be seen if the bulk scalar field can leave a trace in the CMB anisotropies and Large Scale Structures (for first results see [87]).

## 4 Moving Branes in a Static Bulk

So far, we were mostly concerned with the evolution of the brane, without referring to the bulk itself. In fact, the coordinates introduced in eq. (14) are a convenient choice for studying the brane itself, but when it comes to analysing the bulk dynamics and its geometry, these coordinates are not the best choice. We have already mentioned the extended Birkhoff theorem in section 2. It states that for the case of a vacuum bulk spacetime, the bulk is necessarily static, in certain coordinates. A cosmological evolving brane is then moving in that spacetime, whereas for an observer confined on the brane the motion of the brane will be seen as an expanding (or contracting) universe. In the case of a scalar field in the bulk, a similar theorem is unfortunately not available, which makes the study of such systems much more complicated. We will now discuss these issues in some detail, following in particular [73] and [74].

### 4.1 Motion in AdS-Schwarzschild Bulk

We have already discussed the static background associated with BPS configurations (including the Randall–Sundrum case) in the last section. Here we will focus on other backgrounds for which one can integrate the bulk equations of motion. Let us write the following ansatz for the metric

(98) |

where is the metric on the 3d symmetric space of curvature . In general, the function and depend on the type of scalar field potential. This is to be contrasted with the case of a negative bulk cosmological constant where Birkhoff’s theorem states that the most general solution of the (bulk) Einstein equations is given by , and where

(99) |

We have denoted by the AdS scale and the black hole mass (see section 2). This solution is the so–called AdS-Schwarzschild solution.

Let us now study the motion of a brane of tension in such a background. The equation of motion is determined by the junction conditions. The method will be reviewed later when a scalar field is present in the bulk. The resulting equation of motion for a boundary brane with a symmetry is

(100) |

for a brane located at [36]. Here is the velocity of the brane measured with the proper time on the brane. This leads to the following Friedmann equation

(101) |

So the brane tension leads to an effective cosmological constant . The curvature gives the usual term familiar from standard cosmology while the last term is the dark radiation term whose origin springs from the presence of a black-hole in the bulk. At late time the dark radiation term is negligible for an expanding universe, we retrieve the cosmology of a FRW universe with a non-vanishing cosmological constant. The case corresponds of course to the Randall-Sundrum case.

### 4.2 Moving branes

Let us now describe the general formalism, which covers the case of the AdS–Schwarzschild spacetime mentioned above.

Consider a brane embedded in a static background. It is parametrized by the coordinates where and the are world volume coordinates. Locally the brane is characterized by the local frame

(102) |

which are tangent to the brane. The induced metric is given by

(103) |

and the extrinsic curvature

(104) |

where is the unit vector normal to the brane defined by (up to a sign ambiguity)

(105) |

For a homogeneous brane embedded in the spacetime described by the metric (98), we have , where is the proper time on the brane. The induced metric is

(106) |

The local frame becomes

(107) |

while the normal vector reads

(108) |

The components of the extrinsic curvature tensor can found to be

(109) | |||||

(110) |

The junction conditions are given by

(111) |

This implies that the brane dynamics are specified by the equations of motion

(112) |

and